Í f f I I I F b Annals of the Missour Volume 75, ‘Number 4 ^». Annals of the Missouri Botanical Garden tics TER . | arrangements for sss i inside back cover of the last issue of anh ¿plis E Sones Aa ae Marshall R. Crosby ` 2: | a Missouri. Botanical Garden Volume 75 Annals Number 1 of the 1988 Missouri Botanical Garden NS CHANGES IN PLANT Alwyn H. Gentry? COMMUNITY DIVERSITY AND FLORISTIC COMPOSITION ON ENVIRONMENTAL AND GEOGRAPHICAL GRADIENTS'? ! This and dade de papers comprise the proceedings of the Missouri Botanical Garden's 33rd Annual Es Symposium — Species Diversity. The symposium took place in St. Louis, Missouri on October 10 and 11, ? [tha i the National Geographic Society for a series leg that supported much of the research summarized here. Collection of the Madagascar data set was funded by the World Wildlife Fund. The coastal Colombian and Ecuadorian data sets and some of the mazonian Peru data were gathered incidental to floristic projects funded by the National Science Foundati tion. Additional — ania data sets were funded by USAID (DAN-5542-G-SS- the Smithsonian Institution to T. Erwin. The Po sets from eastern Brazil and Paraguay were gathered as part and colleagues who collaborated in gathering the data summarized here were R. Neumann, R. Palacios, C. Cristóbal, and A. Schinini (Argentina); K. Kubitzki, M. Fallen, H. Popppendieck, and W. Lippert (Germany); J. Miller, D. Faber-Langendoen, E. Zardini, and C. Burnett (U.S.A.); C. Ramírez (Chile); E. Lott (Mexico); D. Stevens, P. Moreno, and A. Grijalva (Nicaragua); H. Cuadros, E. Renteria, Mo , A. Juncos C. Restrepo, J. Ramos, P. Silverstone, and O. de Benavides (Colombia); C. Dodson (Ecuador); B. Stein, R. G. Troth-Ovrebo, and P. Berry (Venezuela); F. Ayala, C. Díaz, R. Vasquez, N. Jaramillo, D. Smith, R. Tredwell, K. Young, and D. Alfaro (Peru); A. Peixoto and O Peixoto (Brazil); V. Vera, J. Dávalos, and S. Keel (Paraguay); G. Pilz renun D. Thomas (Cameroon); L. Dorr, L. Barnett, and A. Rakotozafy (Madagascar); J. Connell and J. Tracy (Queensland); J. Tagai (Sarawak: G McPherson (New Caledonia); and V. Kapos (Jamaica). Additional original data using the same or comparable techniques were made available 5 E uu A d 1 š -S. A.). rrozzi lto and especially J. Miller f" FALE pli and technical expertise, E. Zardini for help in t eni and with ihe illustrations, and J. Hall and D. Thomas for providing African data. I thank S. Hubbell, T. Givnish, L. Emmons, Ashton, P. Raven, and d D. Thomas for review comments. * Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166, and Washington University, St. Louis, U.S.A. ANN. MISSOURI Bor. Garb. 75: 1-34. 1988. 2 Annals of the Missouri Botanical Garden ABSTRACT Tr ind in community composition and diversity of neotropical forests as measured by a series of samples of 1) plants = 2.5 cm dbh in 0.1 ha, (2) plants over 10 cm dbh in 1-ha plots, and (3) complete local Gn are aes as a function of various environmen reaches an asympiote (community satura represented in adjac ent forest types on pee substrates may c tal parameters. "These t found in similar data sets from other continents. Altogether the basic 0. 1- A ands. New data from ten I-ha tree si sees in upper Ama . (2) The nearly linear increase of lowland neotropical plant species richness with pre n n?) at about 4,000 mm P bee rainfall. (3) Although n s ^S rends are also compared with those ha data sets are reported for 87 sites zonia are oteworthy trends include: (1) ange dramatically, diversity tends to chan relatively little in upper Amazonia. (4) The species present at different sites are very different but the families represented and their diversities are highly predictable from environmental parameters. (5) On an gradient in the tropical Andes there is a near the upper limit of forest above 3, 000 m. at least not in the sampled habit groups dry forests. (9) Central African forests a am precipitation, but forests in tropical West Africa are relatively oe (10) Tropical Australasian pied diverse than equivalent neotropical forests; the wo are similar in plant species richness and (with a sharp, essentially linear decrease in diversity from about 1,50 6) There is no indication of a ” . (7) Even near timberline, montane tropical fores . (8) “Moist subtropical forests are marked very few notable exceptions) floristic comp . The predictability of the floristic compositions and diversities of tropical p plant communities seems de d albeit circumstantial, evidence that altitudinal O m to mid-elevation 2 in diversity, are as diverse as y less e than their inner- osition these communities are at ecological anc and erhaps evolutionary equilibrium, despite indications that certain aspects of their diversity are generated a kaa sa: stochastically. Comparisons of the species richness (or other facets) of different forests or different vegetation types are often difficult because of the dissimilarity of the available data. In trop- ical Asia there is a wealth of data for trees in large sample areas (Ashton, 1964, in press; Whitmore, 1984; Proctor et al., 1983; Kar- tawinata et al., 1981) but few published data on nontrees. In the Neotropics there are several local florulas (Croat, 1978; Dodson & Gentry, 1978; Janzen & Liesner, 1980; Dodson et al., 1985), but until recently there have been no tree-plot data from high diver- sity regions based on reliable identifications. Africa has far more extensive coverage by regional and country-wide floras but no local florulas nor large-plot data from high-diver- sity regions. ecently, a series of 0.1-ha samples of many of the world's most diverse extra-trop- ical plant communities has been accumulating (e.g., Naveh & Whittaker, 1979; Cowling, 1983; Peet & Christensen, 1980; Rice & Westoby, 1983; Eiten, 1978). Elsewhere we have reported the first comparable data set for tropical forests (Gentry & Dodson, 1987a, b). A standardized ca technique that "L. only plants > 2.5 cm in diameter in a has also been jp and applied to a series of tropical forests (Gentry, 1982b, 1986b; Lott et al., 1987; Stallings et al., i ress; Lorence & e 1988); the methodology for obtaining these 0.1-ha sam- ples, each the sum of ten 2 x 50 m belt transects, is discussed in detail elsewhere (Gentry, 1982b, in prep.). The primary data set on which this paper is based are these 0.1-ha samples, which are now available for 38 lowland neotropical sites, 11 montane neo- tropical sites, and 13 subtropical and 9 tem- perate-zone sites in the Americas. Similar data sets are available from 6 sites in tropical Af- rica, 3 sites in tropical Australasia, 2 sites in Europe, and from several tropical islands: New Caledonia, Madagascar, Mauritius, Jamaica Tables 1, 2; Fig. 1). Supplementary data are taken from local florulas in the Neotropics (Dodson & Gentry, 1978, 1988; Croat, 1978; — Volume 75, Number 1 1988 entry Plant Community Diversity FiGURE Locations of study sites. “Pos = 0.1-ha samples (see Tables 1, 2). Arrows = local florulas. For location of 1-ha tree plots see Gentry, 1 Janzen & Liesner, 1980; Dodson et al., 1985; Hammel, pers. comm. (La Selva, Costa Rica)) and from the Makokou region of northwestern Gabon (Halle, 1964, 1965; Halle & Le Thomas, 1967, 1970; Florence & Hladik, 1980; Hladik & Halle, 1973; Hladik & Gen- try, in prep.). A supplemental data set is pro- vided by a series of 1-ha tree plots in various parts of the Neotropics (Gentry, 1988; Prance et al., 1976; Campbell et al., 1986; see also Gentry, 1982b) and Paleotropics (e.g., Ash- ton, 1964, 1977, in press; Gartlan et al., 1986). Here I first review how the species richness of plant communities changes on five different environmental gradients: latitudinal, precipi- tational, edaphic, altitudinal, and interconti- nental. Observations on a few noteworthy trends in forest structure are also included. Second, I analyze some patterns of floristic change along the same environmental gra- dients. Finally, I use these analyses to ex- amine briefly the question of why some plant communities have so many more species than others. In all of these analyses I will use number of species as the simplest and most appro- priate measure of diversity, as suggested by Whittaker (1977). Shannon-Wiener H' val- ues are reported in Tables | and 2, but are so tightly correlated (R? = 0.93) with the absolute number of species that their use would add little to the analysis. Moreover, the wet- forest H' values of 7 to 8 are far above the levels at which H' has been statistically ana- lyzed (cf. May, 1975). TEMPERATE-TROPICAL PATTERNS Figure 2 summarizes the latitudinal trends in species richness, based on the 74 lowland (= « 1,000 m) 0.1-ha sites for which com- parable samples are available. It is well known that tropical forests are generally far richer than temperate forests in species (e.g., see Richards, 1952; MacArthur, 1972). Figure 2 indicates that for vascular plants species- rich tropical forests are typically an order of magnitude more diverse. Also apparent in Figure 2 are several much less well-known corollaries to the general latitudinal diversity gradient. 1) The difference in species richness between different tropical forests is far greater than the difference between temperate zone and species-poor tropical forests. Whereas the temperate forest samples have 15-25 species and tropical dry forest ones mostly 50-60 species, the samples of moist and wet tropical forests average about 150 species and pluvial forests over 250 species (Gentry, 105°03'W 4 Annals of the Missouri Botanical Garden TABLE 1. Site characteristics for 0.1-ha samples. Nu Alti- Precipi- ber Grid tude tation Fami Number of Site Coordinates (m) lies Species H' Reference Temperate North America Burling Tract, Virgin- 38°55'N 30 1,053 12 21 3.54 Givnish et al., ia 77°10'W unpubl Northwest Branch, 39?02'N 20 1,060 14 20 3.22 — Maryland 77%02'W Tyson Reserve, Mis- 38?30'N 150 932 12 23 3.260 Zimmerman & souri (oak woods) 90°3 1'W Wagner, 1979 Tyson Reserve, Mis- 38°30'N 150 932 11 25 3.68 Zimmerman & souri (chert glade) 90931'W Wagner, 1979 Babler State Park, 38°32'N 150 930 13 21 3.61 — Missouri 90°4.0'W Cuivre River State 3901'N 140 930 15 26 3.46 — Park, Missouri 91°00' W Valley View Glades, 38°15'N 225 930 14 22 3.68 — Missouri 90°37'W Indian Cave State 40°30'N 320 900 12 23 3.74 Tate, 1969 Park, Nebraska 95°43'W Great Smoky Moun- 21-30 White, 1983 tains National (upper 5%) Park, Tennessee / N.C. Europe Süderhackstedt, West 549N 20 695 10 15 2.19 Walter & Lieth, ermany 11°E Allacher Lohe, West 48%04'N 530 866 11 20 3.41 Walter & Lieth, Germany 11?30'E 960 Temperate and Subtropical South America Rio Jejui-mi, Para- 24?4 150 1,800 31 85 5.40 S. Keel & V. guay 55?*30'W Vera, pers. comm. Parque El Rey, Ar- 24?45'S 1,000 1,500 27 40 4.18 Brown et al., gentina 64°40'W 1985 Salta, Argentina 24°40'S 1,300 712 16 25 3.41 Walter & Lieth, 65°30'W Arroyo Riachuelo, 27°30'S 60 1,200 2T 4T 4.46 Walter & Lieth, Corrientes, Argen- 58°50'W tina Alto de Mirador, Chile 40?14'S 800 4,000 13 16 3.45 Ramirez & Ri- 73°18'W veros, 1975 Bosque de San Mar- 39°30'S 30 2,316 14 18 3.25 Riveros & Ra- tin, Chile 73°10'W mirez, 1978 Puyehue National 40%43'S 500 3,000 13 16 2.41 Muñoz, 1980 ark, Chile 72°18'W “Subtropical” Central America Chamela, Mexico 19°30'N 50 733 37 92 5.76 Lott et al., 1987 Volume 75, Number 1 Gentry 5 1988 Plant Community Diversity TABLE l. Continued. Num Alti Precipi- ber of Grid tude tation Fami- Number of Site Coordinates (m) (mm) lies Species H' Reference Chamela, Mexico 19°30'N 50 733 34 83 5.42 Lott et al., 1987 105?03'W Chamela, Mexico 19%30'N 50 733 46 105 5.9 Lott et al., 1987 105%03'W Los Tuxtlas, Mexico 18?35'N 200 4,953 40 108-109 4.52 Lot-Helgueras, 95?08'W 1976 Cerro Olumo, Nicara- 12°18'N 750 2,000 36 97-98 5.8 — gua 85°24' W Cerro El Picacho, 13%00'N 1,400 2,000 39 65 5.22 — Nicaragua 85?55'W Lowland Neotropics (12°N to 12°S, < 1,000 m) Corcovado, Costa Rica 8*30'N 30 3,800 46 132 6.56 Hartshorn, 83?35'W 1983 Guanacaste (upland) 10930'N 100 1,600 21+ 53 Hartshorn, Costa Rica (700 = 85°10'W 1983 Guanacaste (gallery), 10930'N 50 1,600 33+ 68: Hartshorn, Costa Rica (800 = 85°10'W 1983 Curundu, Panama 8°59'N 20 1,830 42 90 5.78 Gentry, 1982b 79°33'W Madden Forest, Pana- 9%66'N 50 2,433 45 126 6.34 Gentry, 1982b ma 79°36'W Pipeline Road, Pana- 9°10'N 300 3,000 58 167 6.77 Gentry & Em- ma 79°45' W mons, 1987 Galerazamba, Colom- 10°48'N 10 500 21 55 5.05 — bia 75%15'W Tayrona, Colombia 11?20'N 50 1,500 3l 65 536 — 74°02' W Bosque de la Cueva, 11°05'N 360 2,000 36 93 5.5 — Colombia 73°28' W Tutunendo, Colombia 5°46'N 90 9.000 53 258 7.57 Gentry, 1986b 76°35'W Bajo Calima, Colombia 3°55'N 100 7,470 58 265 7.74 Gentry, 1986b 77°02' W Boca de Uchire, Ven- 10°09'N 150 1,200 20 66 5.16 Gentry, 1982b ezuela 65°25' W Blohm Ranch, Vene- 8°34'N 100 1,400 31 68 5.38 Troth, 1979 zuela 67°35'W Estacion Biologico de 8°56'N 100 1,312 21+ 59 Gentry, 1982b los Llanos, Vene- 67°25'W zuela (5 2 Cerro Neblina, Vene- 0°50'N 140 3,000 31 97 5.33 zuela (No. 1) 66?11'W Cerro Neblina, Vene- 0°50'N 140 3,000 26 83 4.95 zuela (No. 2) 66?11'W Rio Palenque, Ecua- 0°34'S 200 2,980 50 119 6.15 Dodson $ Gen- dor (No. 1) 79°20'W try, 1978 6 Annals of the Missouri Botanical Garden TABLE 1. Continued. Num- Alti- Precipi- ber of tation Grid tude Fami- Number of Site Coordinates (m) (mm) lies Species H' Reference Rio Palenque, Ecua- 0°34'S 200 2,980 43 121 6.18 Dodson & Gen- dor (No. 2) 19?20'W try, 1978 Centinela, Ecuador 0°34'S 550 3,000 55 140 4.78 Gentry, 1986b 79°18'W Jauneche, Ecuador 1916'S 60 1,855 38 96 5.39 Dodson et al., 79°42' W Capeira, Ecuador 2°00'S 50 804 26 60 5.41 Dodson & Gen- 79°58'W try, 1988 INPA, Manaus, Brazil 3°5 75 1,995 34 101 Gentry, 1978 60°W Mocambo, Belem, 1°30'S 30 2,760 39 131 6.42 Pires & Prance, azil 47°59'W Linhares, Espirito 19*18'S 50 1,403 53+ ca. 212 7.4 Peixoto & Gen- Santo, Brazil 40°04’ W try, in prep. Jacarepagua, Rio de 23?05'S 200 1,500 45+ ca. 160 Janeiro, Brazil 43?25'W Tarapoto, Peru 6*40'S 500 1,400 38 97 5.96 — 76°20'W Sucursari, Peru 3°15'S 140 3,500 46+ ca. 240° 7.46 — 72°55' W Yanamono, Peru (up- 3°28'S 140 3,500 50 212 7.49 Gentry & Em- land) (No. 1 72°50'W mons, 1987 Yanamono, Peru (up- 3°28'S 140 3,500 50 225 7.59 Gentry & Em- land) (No. 2) 72°50'W mons, 1987 Yanamono, Peru (ta- 3°28'S 130 3,500 51 163 6.67 huampa) 72°50'W Mishana, Peru (flood- 3°47'S 130 3,500 58 249 7.03 Gentry & Em- plain) 73°30'W mons, 1987 Mishana, Peru (ta- 3°47'S 130 3,500 40 168 6.44 huampa) 73°30'W Mishana, Peru (upland 3°47'S 140 3,500 46 196 7.21 Gentry & Em- white sand) 73°30'W mons, 1987 Bosque von Humboldt, 8°50'S 270 2,500 44 154 6.37 Peru 75°00’ W Cabeza de Mono, 10°20'S 320 3,500(+) 42 147 6.82 Gentry, 1988 Peru 75°18’ W Cocha Cashu, Peru 11%51'S 400 2,000 49 162 6.78 Gentry & Ter- 7119 W borgh, in press Tambopata, Peru (lat- 12°50'S 260 2,000 48 149 6.7 Erwin, 1985 eritic terra firme) 69°17' W Tambopata, Peru 12%50'S 260 2,000 43 130 6.44 Erwin, 1985 (sandy terra firme) 69°17'W Africa Makokou, Gabon (No. 0°34'N 500 1,755 39 135 6.44 Hladik, 1978 1) 12°52’E Makokou, Gabon (No. 0°34’N _ 500 1,755 32 116 6.25 Hladik, 1978 2) 12°52'E Volume 75, Number 1 1988 Gentry 7 Plant Community Diversity TABLE l. Continued. Num- Alti- ^ Precip- ber o Grid tude tation Fami Number of Site Coordinates (m) lies Species H’ Reference Omo Forest, Nigeria 7°N 50 1,800 29 73 4.42 Richards, 1939 5*E Oban Forest, Nigeria 5°10'N 50 4,000 ? (53+ +) (200 m?) 8*28'E Mt. Cameroon, Cam- 4°N 230 8,000 37 129 6.31 Richards, 1963 eroon 9E Korup National Park, 5°N 50 5,460 43 139 6.34 Gartlan et al., Cameroon 8°33 l'E Belinga, Gabon (500 1°N 750 1,800 26(+) 115 Aubreville, 1967 14°F Perinet, Madagascar 18°55'S 950 1,200 52+ ca. 199 48?25'E Australia Davies River State 17%05'S 800 2,300 41 115 6.29 Connell et al., Park, Queensland 145°34'E 1984 Asia Semengoh Forest, Sa- 1°50'N 20 4,000 47 243 1.39 Walter & Lieth, rawak 110?05'E 1960 Bako National Park, 1?52'N 30 4,000 39 143 6.5 Ashton, in press Sarawak 110%06'E Tropical Islands Riviére des Pirogues, 22°10'S 360 2,200 47 151 6.31 New Caledonia 166°50'E Round Hill, Jamaica 17°50'N 40 1,200 31 58 3.96 Kapos, 1982 77°15' W (4.47) Brise Fer, Mauritius 20°30'S 600 2,400 26 61 Lorence & Suss- 57?30'E man, 8 * Extrapolated from number of species in sample of < 1,000 m°. 1986b). 2) The latitudinal decrease in species richness seems to be asymmetrical about the equator; in the Southern Hemisphere it begins near the Tropic of Capricorn, but in the north it begins well inside the Tropic of Cancer, apparently near 12%N latitude. 3) Temperate zone forests are very similar in species rich- ness of woody plants, compared with the mas- sive differences in species richness between temperate and tropical forests or between dif- ferent tropical forests. Temperate zone forests are so massively depauperate that even if boreal forests with two or three species 2 2.5 cm dbh in 0.1 ha were included in Figure 2, they would not significantly change it, even though the reported values are for some of the reputedly richest temperate zone forests. 4) Species-poor tropical forests with single- species dominance are generally still much more diverse than any temperate-zone forest. 5) South temperate forests, at least in Chile, where data sets are available, have fewer species than temperate forests in North Amer- ica, contrary to the popular perception of the "rich" Valdivian forest; a major reason for this diffference is that Valdivian forests do not have sympatric congeners like the up to seven Quercus and four Carya species typical of 0.1-ha samples of eastern North American forests. 6) Subtropical dry forests can have more species than do full-tropical dry forests, even though wet or moist forests usually have fewer species in the subtropics than in the inner tropics. Annals of the Missouri Botanical Garden TABLE 2. Site characteristics for 0.1-ha samples from upland Neotropics (= 1,000 m, 12°N to 12"S). Parentheses indicate sites too incompletely sampled for a meaningful estimate of number of species in 0.1 ha. Median Grid Altitude Number of Number of Site Coordinates m Families Species H' (Monteverde, Costa Rica (200 m?)) 10%48'N 1,550 (33+) (61+) 84°50' W Cerro Kennedy, Colombia (500 m?) 11%05'N 2,600 26 50° 4.92 74°01'W (Cuchillo de San Antonio, Colombia (200 m?)) 10%58'N 1,710 (15+) (24+) 73°30'W Finca Zungara, Colombia (600 m?) 3*32'N 1,990 37+ 100 76°35'W Farallones de Cali, Colombia 3230'N 1,950 55 134-135 6.48 76°35'W Finca Mehrenberg, Colombia 2°16'N 2,290 40 107 4.46 76°12’ W La Planada, Colombia 1?10'N 1,800 38 116 5.14 77°58'W Pasochoa, Ecuador (400 m?) 0°28'S 3,010 21 28° 3.03 78°25'W Venceremos, Peru 5?45'S 1,850 46 159 6.65 "7040 W Incahuara, Bolivia 15%55'S 1,540 45 130 6.71 67°35'W Sacramento, Bolivia 16?18'S 2,450 32 93 4.89 67°48'W * Extrapolated from number of species in sample of < 1,000 m?. There are also latitudinal differences in forest structure. In general, tropical forests, far from being open and cathedral-like, are denser than temperate forests. This difference is almost entirely due to small-diameter plants, lianas, and trees less than 10 cm dbh (also see Gentry, 1982b). Biomass (as extrapolated from basal area) is roughly equivalent among different tropical forests (Y = 34.9 m?/ha, N = 36 (excluding Africa; X = 70.7 m?/ha, N = 6)) and north temperate deciduous forests (X = 29.6 m*/ha, N = 5) but markedly greater in the Valdivian forests (X = 155.7 m?/ha, N — 3) as well as in their north tem- perate equivalent, the northwestern conifer- ous forests (Waring & Franklin, 1979). DIVERSITY VS. PRECIPITATION In the Neotropics, plant species richness is strongly correlated with absolute annual precipitation (Gentry, 1982b). However, this relationship is more complex than originally suggested (Gentry, 1982b), and the correla- tion may not exist at all in the Paleotropics. In tropical Asia, high rainfall areas such as Mt. Cherrapunji, Assam, often have relatively low plant species richness (Ashton, in press). In tropical Africa, two high rainfall sites (> 5,000 mm per year) in southwestern Cam- eroon (Korup, Mt. Cameroon) have only mar- ginally more species in 0.1-ha samples than do samples from northeastern Gabon that re- ceive < 2,000 mm of annual rainfall. More- over, a more monsoonal climate site in Nigeria (Omo Forest) had many fewer species than the Gabon sites despite having similar pre- cipitation values. Thus, it seems likely that the generalization that diversity increases lin- early with precipitation (Gentry, 1982b) ap- plies only in the special case of the Neotropics, where total annual rainfall and strength of Volume 75, Number 1 1988 Gentry Plant Community Diversity 300 @Ə o o 200 phe o * ` ° ee ° ° . . 2 ^. i? £ ` `. z ` . ` ` so} 100 | = M ^ . T z ^ r ` Y ` e .. \ P ` ° . r `x r ` »- ` : ... ` `. Ë s i 40 30 20 10 0 10 20 30 40 50 °S Latitude °N FIGURE 2. Species richness of 1,000 m? samples of lowland (< 1,000 m) forest as a function of latitude. Closed line encloses continental African points; dotted line encloses Asian points; da and New Caledonian samples; other tropical and subtro diversity Madagascar point a t 19%, (circled). Dashed line separates dry forest (bo dot line encloses Australian ept mal sh- pical points all neotr a. exc ttom) forest (top) with three ai ales B sites (moist forest physiognomy despite relatively strong Fo sea de indicated by alternate lines. the dry season are strongly correlated. A po- tential test of the relative importance of dis- tribution and amount of precipitation comes from a single 0.1-ha site in coastal Brazil (Linhares), which has the unusual (for the Neotropics) condition of low, evenly distrib- uted annual rainfall. Although analysis of the Linhares diversity data is not completed (Pei- xoto & Gentry, in prep.) and the site is thus not included in Figure 3, it is obvious that its estimated 212 species in 0.1 ha are far more than would be expected from its 1,400 mm of annual precipitation. While the many additional 0.1-ha samples now available from the lowland Neotropics generally strengthen the previously reported relationship between neotropical plant species richness and precipitation (Gentry, 1982b), additional data sets at the upper end of the precipitation scale strongly indicate that the relationship becomes nonlinear, reaching a marked asymptote at around 4,000-4,500 mm of annual precipitation (Fig. 3). The re- lationship is significantly curvilinear (F 4.299, P « 0.05). From 4,000 mm to near the wettest place in the world (Tutunendo, Colombia) there is little or no change in the species richness of neotropical plant com- munities as measured by the 0.1-ha sampling protocol. The regularity of species richness patterns, and especially the apparent lid on community richness suggested by this as- ymptote, seem strong circumstantial evidence of the kind that zoologists (e.g., MacArthur, 1965, 1969) have construed as representing niche saturation and community equilibrium. It is also possible that part of the apparent lid on plant community richness merely re- flects the intrinsic limitations of the sampling technique. Figure 4 compares the accumu- lation of species with sample area for several representative sites. In low-diversity forests the species area curves level off below 500 m? indicating that most of the species present Annals of the Missouri Botanical Garden 300 250 o 200} o o a mn s 150? o a E 2 z 100: 50} 1 2 3 4 5 6 7 8 9 Annual precipitation (mm x 9 FIGURE 3. 0035 98x’. The curve is displaced slightly EE the data points, since questionable morphospecies and ios specimens are treated as distinct species uter while the data points represent d. by t mp with s sites exclude in a given community have been sampled, but in species-rich vegetations the species-area curves show little sign of leveling off. To what extent a larger sampling area might reveal significant diversity differences between the different high rainfall sites remains unknown. The strong relationship of species richness to precipitation in neotropical forests is fur- best estimates of species numbers. Data from Table 1 ther supported by preliminary data from 1-ha tree plots in upper Amazonia (Gentry, 1988). In these samples only trees and large lianas = 10 cm in diameter were censused (Fig. 5). The two most species-rich sites are from the everwet high rainfall (3,000-4,000 mm) Iquitos area of northern Amazonian Peru, where diversity reaches almost ridiculous ex- 300 | . Bajo Calima ° 250 | gran Tutunendo à p di e Yanamono 2 o p eee "d " 20 e UE o 5 0 L A y ae o iit ra E S "Lu Z p". ui "mm di l E — 1 š x . 00 | a í i ede A e. Tarapoto Ze P di a 50 = La e o- a e Blohm Ranch re . ee —_e— Sa — °. ° ud e= -. ë . . |. Northwest Branch a ñ L 1 L L 1 Ë 1 T 1 2 3 4 5 6 7 8 9 10 Transect number FIGURE 4. Species-area curve for 100 m? subsamples of representative high- and low-diversity 0. 1-ha samples. Volume 75, Number 1 Gentry 11 1988 Plant Community Diversity 800 Relatively Poor soil 700 fertile soil B. . 19.6/ha. söö m ianas (av. 19.6/ha.) B eses 30 cm dbh É - m trees? 30 c E = rich soil av.: aq. za = 500 Poor soil av.: 81/h E o o A 400 2 S 3 2 2 N - d = £ 300 s gj ig = > a| |si le o 3 = = É = a a o - G 3 2 z [7 a o o s o E 2115 % o| |° 2 200 o a a a c 2 o > E P 2 8 N oj |£ a 3 HIBE- FA s| Sis [S|[s| [5 [21 [3 S 2 =) i>] Je e| IF] lol |S] |z GQ SS SS GY te > zd FIGURE 5. Density of trees and large lianas in Amazonian plots. Black = lianas; hatched = trees > m dbh. 30 cm dbh; white = trees 10-30 c Aseasonal Dry Season Subtropical Number of Species a ° Extremely Poor Soil EIN Bre JU E Yanamono Mishana Manu Cabeza de Tambopata Tambopata Neblina Mono alluvial beso ra upland 2 FIGURE 6. Number pa in 1-ha Amazonian tree plots "m > 10 cm dbh). Black area = lianas > 10 cm greatest diameter Annals of the Missouri Botanical Garden tremes (Fig. 6). At Yanamono there are 300 species > 10 cm in diameter out of the 606 individual plants in a hectare plot! The other 1-ha plots in Amazonian Peru are in areas with generally greater dry season stress and less overall precipitation. Two sites between 10? and 12°S latitude have about 200 species > 10 cm dbh, while several 1-ha plots in different habitat types at Tambopata Reserve in southeastern Madre de Dios (12?50'S) have between 153 and 181 species. Thus tree species richness also appears to be greatest in aseasonal high rainfall areas, at least within Amazonia. Epiphyte diversity likewise increases in wetter areas. While epiphytes can be well represented in areas with high atmospheric humidity but relatively low rainfall, our data indicate that absolute precipitation is gener- ally a remarkably good predictor of epiphyte diversity (Gentry & Dodson, 1987b). We have data sets from a series of local florulas in western Ecuador and southern Central Amer- ica; epiphytes vary from 9-24 species (2- 4% of the total flora) in dry-forest sites to 72-216 species (12-16% of the total flora) in moist-forest sites to 238-368 species (23- 24% of the total flora) in wet-forest sites (Gentry & Dodson, 1987a, b). For a series of 1,000 m? samples in which all plant species were identified and tabulated in three western Ecuadorian forests, 3 epiphytes constituted 2% of the species in a dry forest, 13 epiphytes constituted 8% of the species in a moist forest, and 127 epiphytes constituted 35% of the species in a wet forest (Gentry & Dodson, 1987b). The wet forest at Rio Palenque is so diverse in plant species that, even excluding tree species, it has more species of herbs (including herbaceous epiphytes) or of shrubs in 0.1 ha than any nontropical plant com- munity in the world (Gentry & Dodson, 19872). DIVERSITY VS. SOIL. NUTRIENTS There has been much recent interest in the relationships between tropical soil nutrient levels and plant community richness (Ashton, 1977, in press; Gartlan et al., 1986). These authors suggest that phosphorus, magnesium, and potassium are among the nutrients whose levels are most strongly correlated with trop- ical plant community diversity. Nevertheless, at least in the Neotropics, soil nutrients are far less important than biogeographic factors or precipitation in determining plant species richness (Gentry, 1982b; Stark et al., sub- mitted ms.). Multiple regression of a series of 31 lowland neotropical sites for which we have both soil and species richness data for O.1- ha samples produced the equation: Species Richness = 84.48 0.025(mean annual precipitation) — 0.100(extractable soil K). R .76, N = 31 (Stark et al., submitted ms.). Thus our data indicate that the nutrient most closely correlated with neotropical species richness is K. The importance of K agrees with what Ashton (1977, in press) found for a large series of tree plots in Borneo, Gartlan et al. (1986) also found available K to be highly and significantly correlated with floristic diversity in a series of sites in Cam- eroon. Our data contrast with those of Ashton (1977, in press) and Gartlan et al. (1986) in that we do not find phosphorus to be strongly correlated with diversity. This may be due in part to different techniques of nutrient ex- traction (ammonium acetate vs. HCl). It is also related to the fact that the most species- rich 0.1-ha sample (Bajo Calima, Colombia) comes from a peculiar white clay soil with O phosphorus as measured by our technique. Whereas Ashton's (in press) data sets in- dicate greatest diversity at intermediate nu- trient values, a “humped” nutrient/diversity curve that fits the model proposed by Tilman (1982, 1984), I see no indication in my data of a general decrease in diversity on richer soils in the Neotropics. Quite the contrary, the most species-rich tree plot in the world at Yanamono, Peru, is on relatively rich soil (Gentry, 1988; Stark et al., submitted ms.); further south, in an area with a strong dry season, the 0.1-ha Cocha Cashu sample on unusually rich alluvial soil is farther above the precipitation-diversity regression line than Volume 75, Number 1 1988 Gentry - $19 Plant Community Diversity number of Species FicURE 7. Number of individuals/ speci any other site (Gentry, 1985a). My data do fit well with Ashton's along the low nutrient end of the diversity-soil nutrient gradient, where there is a general increase in species richness from nutrient-poor to intermediate sites, contrary to the suggestions of Huston (1979, 1980). Another way of comparing the effects of soil fertility on diversity is by comparing oth- erwise approximately matched site pairs on fertile and poor soils. The series of six tree plots in Amazonian Peru fall into three natural groups based on latitude and strength of the dry season (Gentry, 1988). Of the two plots in the everwet Iquitos area, one on rich soil has (marginally) more species than a nearby site on white sand; on a species per individual basis the difference would be much stronger (Figs. 7, 8). Of two sites from central Peru, the one on rich alluvial soil (Manü Park) has more species than one on poor soil (Iscozacin). Several plots at Tambopata south of the Hold- co white sand (Mishana 4 Neblina) Ez 'sub'tropical (Tambopata) WEA tropical terra firme (Yanamono, Manu, & Cabeza de Mono) number of individuals in 1-ha Amazonian tree plots (plants 2 10 cm diam.). ridge system tropical-subtropical demarca- tion have fewer species than the full-tropical ones on either rich or poor soils. Moreover, the site with the most nutrient-poor soil of all, Cerro Neblina, on pure white sand, has many fewer species than do any of the other sites. Thus the Amazonian tree plot data gen- erally support the idea that relatively rich soil correlates with relative richness in tree species. Especially noteworthy in the context of the relative importance of soil nutrients and pre- cipitation as determinants of species richness is the series of 0.1-ha samples from different substrates in the Iquitos area (Table 3). All of the sites have the high species richness (168-212 species) that would be expected (Gentry, 1982b) in a region with high rainfall and no dry season. While samples from the forests subjectively judged likely to be sub- jected to greater stress (i.e., seasonally in- undated tahuampa or white-sand campina- rana) have slightly lower species richness, all Annals of th Missouri Sone Garden % of species with number of individuals 0 ES o ç 3 o -a c cc c 9 oo = S ç o C c 3 E <> z 5 5 S o ç 2 = aaz o c s 2 "mm 2= ç < 29 NS G G ° Q o ua a O o o 20 o 29 O E € € gd o g = == SURE 8. Percent of species with different numbers of individuals in und tree plots (plants = 10 cm diam.). Lower white bar — p Ves riiv dolied bar — e individual hatched bar — bar — r more individuals. | across plots of 50% of the An are represented by single individuals. sites are very diverse compared with mois- ture-stressed sites with a strong dry season or low annual precipitation. I conclude that the species richness of neo- tropical plant communities generally in- creases with soil fertility and with precipita- tion, when such broader-scale biogeographic factors as latitude and altitude are controlled. This relationship would predict that the high- est neotropical a-diversities should be found in upper Amazonia, where the soils are rel- atively rich, compared with those of compar- ably high rainfall areas of the Guayana Shield. My data for 0.1-ha samples and for 1-ha tree plots both appear to fit this prediction. More- over, many other kinds of organisms, includ- ing birds, reptiles and amphibians, butterflies, and bats, appear to show excactly the same pattern of greatest diversity in areas with relatively fertile soils near the base of the Andes, suggesting that this relationship is a general biogeographic trend (Gentry, 1988). It is possible that increased productivity on the generally richer soils of this region makes possible finer niche partitioning and special- ization in otherwise marginal habitats (cf. Em- mons, 1984; Gentry & Emmons, 1987). Even though the effect of soil nutrients on a-diversity may be relatively minor, soil nu- trients undoubtedly do play a major role in contributing to the high overall diversity of Amazonian forests through their effect on B-diversity (e.g., Gentry, 1981, 1986a, c). Much of upper Amazonia, probably more than any other part of the lowland Neotropics, constitutes a conspicuous habitat mosaic, with very different sets of plant species occurring in adjacent communities on different sub- strates (Salo et al., 1986; Gentry, 1986a, c). Table 3 shows how little overlap in species there is between different, more or less equally diverse plant communities on different sub- strates in the Iquitos area. Only 3-24 species TABLE 3. Number of species shared by 1,000 m? samples of Iquitos area forest types. Mishana Yanamono Yanamono Yanamono Mishana Campi- Mishana No. 1 No. 2 Tahuampa Lowland narana Tahuampa Yanamono Terra firme No. 1 212 91 20 24 12 14 Terra firme No. 2 230 20-21 19 9 8 White-water tahuampa 163 9 5 ca. 19 Mishana owland noninundated 249 55 17 Campinarana (white sand) 196 3 168 ack-water tahuampa Volume 75, Number 1 1988 Gentry Plant Community Diversity Laguna Cocococha km5 © 18 spp. shared by plots 3, 4, and 6 + 172 s ° % pp a : o poor soil terra firme forest t on old sandy river terrace S 4 = ° 32 spp. shared c 181 < 3 spp 3 terra firme 83 spp. shared P nina LC 1 í : ° AU bid ^ VA š i P y? { y San, Q $ CA 15 l pe 2 2)/d LD L--parghbosr Sop Y a.158 spp L - P m i 2 “° et / F. (oi flog st m youn Pla, o secondar tn 27 spp. shared _ fores Es URE 9. Location A I-ha tree plots in the Tambopata Wildlife pao Madre de Dios, Peru. Indicated m ins numbers for plots 2 on field identifications of Gary Hartshorn (pers. Hi ice Plot 1 data in part base er of species will = be higher as well. Shared species indicated only for plots (3, 4, 6) la: Mb and identified by m out of the ca. 200 species sampled for any habitat are shared by a different adjacent habitat. The one exception is the Mishana white-sand and floodplain samples (55 species overlap), but these two vegetation types have similar substrates and are not very well dif- ferentiated. While some of this lack of overlap might be due to inadequacy of the sampling technique in such diverse plant communities, a repeat sample of the same forest at Yan- amono gave a much greater, almost 50% overlap in species; in other species-rich moist and wet forests similar repeat samples of the same vegetation always give the same ca. 50% overlap in sampled species (Gentry, 1982b), contrasting strongly with the < 20% overlaps between different communities. Sim- ilarly, for two 1-ha tree plots on terra firme forest on poor sandy soil at Tambopata, 83 species (46% of the 181 species in plot | and 48% of the 172 species in plot 2) were shared with the other plot, for a coefficient of as- sociation of 44%. Only 16-18% of the species of either poor soil plot were shared with a nearby tree plot on rich alluvial soil (coefh- cients of correlation of 10-11%) (Fig. 9). Incompletely analyzed data for additional plots in other forest types at Tambopata indicate that they, too, will show little overlap in species with sandy soil or alluvial forests. The unique- ly high species richness of the Tambopata reserve for such well-known groups as birds Donahue et al., in press) and butterflies (La- mas, 1985) has been suggested as largely due to the reserve’s habitat diversity, a conclusion that clearly accords with the botanical evi- ence. Thus the high species richness of woody plants in Amazonia as compared with the rest of the Neotropics (Gentry, 1982a) is largely B-diversity due to habitat specialization. Typ- ically, related species may fill similar niches in forests on different upper Amazonian sub- strates (Gentry, 1981, 1986c). Dramatic dif- ferences in specific composition, though not intracommunity diversity, accompany spe- Annals sib Eod Garden Decrease in diversity with altitude 200 180 lowland Amazonian average 160 eVenceremos moist & wet forest i h average elncahuara 140 e . Centinela : e Farallones x ` La Planada 120 ` eFinca Zíngara * Los Tuxtlas E Merenber 100 e Cerro Olumo E š id ES - ^ z 1 w e Sacramento o ———Chamela average ` ` o u e o ` 2 . 5 ^ 80 ` T - ` . ° z: ° : eCerro El Piopaco S 60 I———— dry forest average z M. pt e Cerro Kennedy E 40 x Parque El Rey 1. Most diverse ` * temperate site Salta ` £ ` i ? Eastern North America average ` Pasoc 20 Temperate average X ? naa a ivian average k Suderhackfested 0 1 2 FIGURE 10. regression are for Andean Altitude in kms. Vas richness of 0. I-ha samples vs. altitude. Points to right of dashed line and the calculated sites. Compa individual temp parative data from other selected sites to left of dashed lin erate and subtropical sites: o M is in northwest ps rge e. Stars are for ntina; Los Tuxtlas is in Veracruz, l a and Pace El Rey are in ; Cerro Olumo and Cerro El Pichaco are in Nica a; Centinela is an isolated ridge west of the Andean Cordillera Occidental in Ecuador. Average species richness for other site-series indicated by lines Spanning appropriate altitudinal range; Chamela is western Mexican dr forest. Several of the Andean values e preliminary, being based only on field cog w ations with herbarium comparison of vouchers still ded or on samples of less than 1,000 m? (see Table 2). cializations for different edaphic conditions, often related to different soil-nutrient avail incomplete, the trend of decreasing diversity ability in different Amazonian habitats with increasing altitude is clear. At least within the Andes, this inverse correlation is linear (Fig. 10), but the relatively low diversity of i two Central American lower montane sites ALTITUDINAL TRENDS . suggests that the extra-Andean decrease in Eleven sites in tropical forests between diversity with altitude may not follow the same 1,500 and 3,100 m altitude, mostly in the rules; certainly Central American montane Andes, are included in Table 2. Although the forests have very different floristic composi- available data set for upland sites is very ti tions as well. Although there has been much Volume 75, Number 1 Gentry 17 1988 Plant Community Diversity € ~ ~ E ma ~~ ~ ~ o E E o e E [e] - A + o o N re - G ° ° o - o o 0 ~ S aq» i 0 + L9 or o + I 5 9 iios ~ D o rv N > 5>5 Ç 2 2 o E at ° o C q ki r e g Ee “v D S =< < 9 3 oro z oa % ov" aS 2 8 Í” o 2 2,3301 ot eee sS 5 S s Š S Š sS zesz ee § < c m A = - S 2 < ° Š s2 52 225322; S s SES ECE SH EES Ps ee e S £ S EST š s š ES Š S Š s Š ° Zo Ë Š aE EG a 30 9 9 o 6 @ o 9 § a $ 256 9 9o s 5 5 9 5 s 6c 525 Poor * m > 2 2 à > > o 3 o & E oO m - Ó EGS » E Z o o Lowland Moist and wet Forests Chocó Region jas Upland Forests and Pluvial Fore FiGURE 11. Percent of hemiepiphytes don portion of bar) in nd climbers for 0.1-ha samples of yee (in altitudinal sequence) compared with lowland Chocó area and non-Chocé area samples. Note apparent peak in hemiepiphytes at 1,800 m speculation in the literature about a “‘mid- most striking is the increase in sampled hemi- altitude bulge” in diversity (Janzen, 1973; epiphytes around 1,800 m (Fig. 11). How- Janzen et al., 1976; Scott, 1976; Gentry & ever, since increased numbers of hemi- Dodson, 1987b), there is no hint of such a epiphytic species (and individuals) are phenomenon in the data of Figure 10. In- concomitant with decreased numbers of free- stead, there seems to be a constant rate of climbing liana species and individuals, there decreasing species richness in moist Andean is no net change in community diversity. Also forests from the lowland tropics to near tree noteworthy is the relative abundance of hemi- line. Unfortunately, no sites have been sam- epiphytic climbers in wet lowland sites in the pled from the Andean foothill region between Chocó area, a typical example of the tendency 600 and 1,500 m, making it difficult to judge of the forests of this region to have features at what altitude the decrease in diversity be- and taxa more characteristic of upland forests gins. Clearly there is no altitudinal effect up (Fig. 11; Gentry, 1986b). At higher altitudes to at least 500 m (Cocha Cashu, Peru; see free- albis lianas take over again, so that Gentry, 1985a). Since samples from sites at at 2,500 m and above, hemiepiphytes have , 100 m would be near the average value for completely dropped out. — lowland wet- and moist-forest sites (Fig. 10), Even near the tree line above 3,000 m, we can assume that there is little or no de- Andean forests are more species rich than crease in diversity up to that altitude. are temperate forests. Our highest-altitude Although there is no increase in diversity sample, from 3,010 m at Pasochoa in the at middle elevations, there are some note- Ecuadorian Andes, has 25 species compared worthy physiognomic changes. One of the with only 21-30 species in the richest 5% of 18 Annals of the Missouri Botanical Garden TABLE 4. Representation of different habits in local florulas (from Gentry & Dodson, 1987b). B Capeira Santa Rosa Jauneche Colorado Habit Number % Number % Number % Number % Epiphyte (including stranglers) 8 2 19 3 72 12 216 16 Parasites + saprophytes 4 l 6 l 4 l 12 l Climbers 112 24 115 18 136 22 258 20 Trees > 10 cm dbh 69 15 142 21 112 19 290 22 Terrestrial herbs, shrubs, treelets 270 58 381 58 280 47 540 41 Total species 463 667 604 1,316 * Data from B. Hammel (pers. comm.). some 312 Great Smokies Mountains samples (White, pers. comm.) and 15-26 (X = 20.5 for the 13 other temperate-zone forests listed in Table 1 — SOME INTERCONTINENTAL DIVERSITY TRENDS At a continental level, the Neotropics have many more species of plants than do either the Asian or Australasian tropics (Raven, 1976; Prance, 1977; Gentry, 1982a). Else- where, I have suggested that the "excess" neotropical species are mostly in herbaceous, epiphytic, and shrub taxa that have speciated explosively along the lower slope of the Andes and in southern Central America. To what extent, if any, does higher a-diversity of neo- tropical forests contribute to the continental pattern’ While I have relatively few comparable paleotropical data sets, a few general trends seem evident. One surprising indication from the available African data is that Central Af- rican forests (X = 127 spp., N = 5) may be as diverse in species = 2.5 cm dbh as their neotropical equivalents (X = 105 spp., N = 9) for sites with 1,600-2,000 mm of precip- itation. Even though the two high-rainfall sites in Cameroon do not show the increases in species richness that might be expected in the Neotropics, they are still very diverse, and the drier Gabon samples actually have more species than would be expected for similar rainfall values in the Neotropics. Moreover, one of the high rainfall sites with anomalously low diversity (Mt. Cameroon) is on the slopes of an active volcano, and the other (Korup) is on an unusually poor, highly leached skel- etal soil (Thomas, pers. comm.). est African forests, including Nigeria’s Omo Forest in my data set and the Ghana forests studied by Hall & Swaine (1981), may be poorer in species for historical reasons since there are suggestions that most West African forests may have been extensively altered by Bantu populations prior to the first European colonization (Keay, 1953; Jones, 1956). Even though my anomalously low di- versity Omo Forest site was in a plot of pro- tected forest considered to be climax (though surrounded by a mosaic of other plots sub- jected to varying degrees of degradation his- torically) (G. Pilz, pers. comm.), a number of its constituent species, such as Pausinystalia macroceras, Spathodea campanulata, Markhamia lutea, and Musanga cecro- pioides, seem more characteristic of late sec- ondary than of primary forest. Nor is the high diversity of Central African forests restricted to woody plants. Data com- parable to a complete local florula are avail- able for one African forest site at Makokou, Gabon (Hladik & Halle, 1973; Florence & Hladik, 1980; Hladik & Gentry, in prep.). Comparison of these data with local florulas from the Neotropics indicates that Makokou is not only as species rich as equivalent neo- tropical local florulas, but it also has a similar habit composition (Table 4; Gentry & Dod- son, 1987b). Similarly, data from 1-ha tree plots indicate that African forests may be almost as rich in tree species as comparable neotropical and Southeast Asian forests (Gart- lan et al., 1986; Thomas, pers. comm.: 138 Volume 75, Number 1 1988 Gentry Plant Community Diversity TABLE 4. Continued. Rio Palenque La Selva? Makokou Number % Number % Number % 238 23 368 25 66+ 6+ 6 1 8 1 9 1 171 16 182 12 259 23 165 16 310 21 389 34 475 45 622 42 418 37 1,055 1,490 1,140 spp. in 0.64 ha on transect S, Korup National Park, Cameroon; Gentry, in press). On the other hand, it is noteworthy that my single Madagascar site is richer in species than any of the continental African sites, which might be anticipated from the now widely accepted hypothesis that Africa’s low conti- nent-wide plant (and bird) species richness stems largely from extinctions associated with climatic deterioration during the Pleistocene or late Tertiary, whereas Madagascar was protected by being an island (Raven & Ax- elrod, 1974; Axelrod & Raven, 1978). Quite the opposite of Africa, Asian forests have been widely thought to have more tree species than neotropical forests (e.g., Ashton, 1977; Whitmore, 1984). This conclusion was based on comparison of extant neotropical data for l-ha tree plots with similar Asian data sets. However, the previously available neotropical tree plots were all from areas that would be anticipated on biogeographical or ecological grounds to have species-poor for- ests (Gentry, 1988). Hectare plots in upper Amazonia consistently have more tree species than in most Asian forests (Gentry, 1988), and the most species-rich l-ha plots are in upper Amazonia. Indeed, these plots are so diverse—up to 300 species out of 606 in- dividuals = 10 cm diameter at Yanamono, eru—that it is hard to imagine how a forest could be much more diverse. I conclude that plant community diversity, at least of woody plants in plots of 1 ha or 15 n g o 9 9 o š š š < = xx 2 o E z z c 3 % o š " 3 3 3 2 E LÀ d w m o = Q c ° > 3 “a 2 E 5 > ° S x 2 = = = 5 a 9 = s = c 2 3 o —— t S a 9 * 3 23 6$ È $ : — o s. is] o - 5I > z 2 z Ficunk 12. Basal forest type; bar = + 1 areas for 0.1-ha samples of some different forest types. Line = average basal area for s.d. 20 Annals of the Missouri Botanical Garden % of species 10 20 30 4 50 60 70 80 90 100 Yanamono 2 | $ | i | š | ° | | 3 | $ | Ë H E 46 other families | Tambopata upl. 1 8 | 9 | ° | H ri | š | š | š E B | 30 other familie | Tambopata upl. 2 3 | 2 | ° | 3 | i $ | P| £| HH H ë| 34 other families == Tambopata alluv. Š | $ | 3 | H | H | s | š | š | H | M | 31 other families | Cocha Cashu ° | $ | 3 | ° | 2 | $ HHBH 37 other families Ni Cabeza de Mono ° | s | ° | š s | z] ° ° | š HB 5| 28 other families Mishana 2 | 2 | 5 | 4 | 3 $ | 3 š H HH 31 other families Neblina Š | 3 | 9 | ° | 5 | 2 | š | š HH 21 other families | — ae EEBESHHHHEHEHHEESHSSHEF"TM s | š [| 3| £J£2]21|$2|$|8[E8s | Korup Q © 5 o LI w o e 12other families Makokou (200 m?) Š | 5 | Š 5 3 | 5 E °| ° | š | š | ES | v ho other families | Kuala Belalong i | E | H | Ej Š š | i H ABR 37 other families J Andulau $ | š | Š | š | š p | 3 | 3 | E HEB 40 other families | FIGURE 13. Familial compositions of some E tree plots (plants = 10 cm dbh) in Amazonia (Yanamono to Xingu), Central Africa (Korup and Makokou and Southeast Asia (Kuala Belalong, Andulao). plots except as otherwise noted. Family codes are the first letters of the familial n names, self-evident except me — Melasto pi mel — Myristicaceae, m tree plots, is just as laceae, sap — Sapotaceae (the second, if present, — Sapind = Myrtaceae. Note that Leguminosae, the dominant family in all neotropical and yr une speciose as is Dipterocarpaceae in Southeast Asia. Also note that, except for dena d aceae), myr — the forests on all three continents are mostly composed of species belonging to the same few woody familie less, has a similar range of variation according to local environmental conditions in all three of the world's main tropical regions; what happens at larger spatial scales remains an open question. Although tropical forest a-diversity may be similar on different continents, its structure is not. For example, lowland neotropical for- ests have fewer lianas than African forests and more lianas than Asian forests (Emmons & Gentry, 1983). Large palms as a major and characteristic canopy element of lowland terra firme forest seem largely restricted to the Neotropics (Gentry & Emmons, 19 Madagascar (20 palms = 10 cm dbh/ha a Perinet), and a few other islands (e.g., New Caledonia: 20 palms = 10 cm dbh/ha at Riviére des Pirogues). While stem densities of trees = 10 cm dbh may be similar from continent to continent (Dawkins, 1959), trop- ical African forests tend to have more large trees and higher basal areas (and presumably biomasses) (70.7 m?/ha vs. 34.9 m?/ha) than do neotropical or Australasian forests (Fig. 2). On the other hand, Asian dipterocarp forests may have uniquely high densities of small polelike trees. Such structural differ- ences, only beginning to be discovered, may be critical to forest organisms. For example, the intercontinental difference in liana density Volume 75, Number 1 1988 Gentry Plant Community Diversity . FOR B AND WET SITES L š [31515 TS š]: | LL 1 Lš |šlšl El [—— —— — Eee JE Av. Lead EP feme; with H | | 3 | s Av. for 27 fams. with AV. OF 2 BORNEO SITES «2 spp. av. . | DAVIES RIVER ST. PK., AUSTRALIA RIVIER DES PIROGUES, NEW CALEDONIA 3) average for 2 pluvial-forest sites in Chocó; 4) continuation of 3; 5) rata E Dav umns represent: entral ae sites (i.e. Ree n k, Queensland. Australia; 9) Rivière des Pirogues, New Caledonia. Shortest column Pudet are two species da may have been the critical factor selecting for different locomotor adaptations among canopy vertebrates on the three continents (Emmons & Gentry, 1983). FLORISTICS Neotropical plant communities are put to- gether in decidedly nonrandom ways. Thus community-level frequency of different seed dispersal and pollination syndromes is gen- erally predictable from environmental param- eters (Gentry, 1982b, 1983). Similarly, the floristic composition of different plant com- munities is remarkably consistent, at least at the familial level. Legumes are virtually al- ways the dominant family in neotropical and African lowland primary forests. The only neotropical exceptions are on extremely rich soils where Moraceae become very diverse and are occasionally as species-rich as Le- Am in 0.1-ha plots (Gentry, 1986b, c). Of the 43 continental neotropical lowland 0.1-ha samples between 23.5°N and S lati- tudes, 39 had Leguminosae as the most species-rich family. The dominance of le- gumes in the Neotropics and Africa is equally apparent when only trees > 1 are considered (Fig. 13). Indeed, legumes con- tribute almost exactly as much to the diversity of neotropical and African forests as dipter- ocarps do in Southeast Asia. Similarly, in Af- rica, on the rich volcanic soil of the Mt. Cam- eroon plot, several families, especially Rubiaceae, Apocynaceae, and Euphorbiaceae have more species than legumes, but this for- est, on the lower slopes of an active volcano, may not be strictly primary. The other families that contribute most to species richness of different plant communi- ties are also predictable. In the Neotropics the same 11 families— Leguminosae, Lau- raceae, Annonaceae, Rubiaceae, Moraceae, Myristicaceae, Sapotaceae, Meliaceae, Pal- mae, Euphorbiaceae, and Bignoniaceae— contribute about half (3895-7395; X = 52%) of the species richness to 0.1-ha samples of any lowland forest. At least eight of these families are always among the ten most species-rich families in any lowland neotrop- ical moist or wet forest (Fig. 14; Gentry, 1987b). Similarly, in 0.1-ha samples of low- land neotropical dry forests, Bignoniaceae, the preeminent liana family, is always second only to Leguminosae in its contribution to species richness (Fig. 15). Somewhat surprisingly, the dominant fam- ilies in neotropical forests also tend to be the most speciose on other continents. Rubiaceae, Annonaceae, and Euphorbiaceae are always among the ten most species-rich families in Africa and Asia, just as they are in the Neo- tropics. The rest of the 11 most species-rich neotropical families (Lauraceae, Moraceae, 22 Annals of the Missouri Botanical Garden na E ETE EREEREER Capeia E EPHE ténis M Boca de Uche] E |: Je ER TE Jane uaos| š | š LE d L. NL Blohm Ranch | E HE BUY = other families | GuanacasteupD| E E I amare Guanacastaggan]| c | = ElK EEE iS ane | eoo) E d dE Jaek E TRE EEEE E E REEE Sis E FicURE 15. Number of species per family for 0.1-ha samples of lowland neotropical dry forests. For three IUe dd MO Rt o Oe ple are, respecto o pedi ea specte Sapotaceae, Palmae, Myristicaceae, Meli- aceae, and Bignoniaceae) are all represented in at least some samples from both Africa and Asia and, except for Bignoniaceae and Pal- mae, are among the ten most species-rich families in at least one African or Asian sam- ple. Thus, with the exception of the substi- tution of Dipterocarpaceae for Leguminosae as the most species-rich woody family in Southeast Asian forests, pantropical familial composition of lowland forests is remarkably similar. Other minor differences include Ebenaceae (almost always present in Africa and Asia and among the ten most species-rich families in about half the samples from those continents but only occasionally represented in the neo- tropical samples, never by more than a single species), Olacaceae (usually represented on all continents but generally among the ten most species-rich families in Africa, never so in Asia or the Neotropics), and Sterculiaceae (always among the ten most species-rich fam- ilies in Africa; represented by 1-3 species in almost all neotropical and Asian samples, al- though among the ten most species-rich fam- ilies only in Cocha Cashu, Peru). Dichapeta- laceae are almost always among the ten most species-rich families in African samples but e al . number uides: in 1,000 m? indicated by the dotted outline. Shortest column segments are two tall. are only occasionally represented by one or two species in the Neotropics and are absent from my Asian samples. Apocynaceae and Sapindaceae almost always turn up in samples from any continent but are generally among the ten most species-rich families in Africa (always in the case of Apocynaceae) but only rarely elsewhere. Disproportionately repre- sented in Asia, besides Dipterocarpaceae, are Myrtaceae (always among the most species- rich families vs. almost always present but only rarely among the most species-rich fam- ilies in the Neotropics and represented by a single species in a single sample on continental Africa). Other noteworthy anomalies include 9 species of Proteaceae, 7 of Elaeocarpaceae, and 6 of Monimiaceae in the Queensland sam- ple (these three families ranking 3rd, 5th, and 6th in diversity after Lauraceae, Myrtaceae, and Rubiaceae), 7 Araliaceae species and 5 of Cunoniaceae in the New Caledonia sample (ranking 5th and 8th, respectively, in familial diversity), and 8 and 3 species, respectively, of Xanthophyllum (Polygalaceae) at Semen- goh and Bako, Borneo. Put another way, all of the paleotropical forests sampled were constituted almost en- tirely of the same plant families encountered in equivalent samples of neotropical forests. Volume 75, Number 1 1988 Gentry Plant Community Diversity Although 13 families not represented in the Neotropics were included in the paleotropical samples, and although each African and Asian sample included 1-3 families not represented in the Neotropics, with two exceptions, the sum contribution of all of these to species richness of the Asian and African forests is negligible. The two exceptions are Diptero- carpaceae in tropical Asia and Pandanaceae in Madagascar (3 spp.), Queensland (2 spp.), and New Caledonia (4 spp.). Excluding these two families, an average of 2 species (and ca. 3 individuals) per sample was contributed to paleotropical community diversity by families not included in the equivalent neotropical samples. At this level New Caledonia was the most distinctive, with one species each of Bal- anopaceae, Epacridaceae, Oncothecaceae, and Pittosporaceae, plus 4 of Pandanaceae. The Madagascar sample included, besides 3 Pandanaceae, a species of Sarcolaenaceae and two of Pittosporaceae, the Queensland sample a species of Balanopaceae and 2 of Pandana- ceae (plus one of the sometimes Cunoniaceae segregate Davidsoniaceae). In Africa, Ancis- trocladaceae was represented by one individ- ual at one site, Medusandraceae by one in- dividual at one site, and Scytopetalaceae by two species at one site. Only in the latter case did an endemic family contribute significantly to a site’s diversity, with Ouabangia alata the 5th most common species (13 individuals) at Korup and Rhaptopetalum cf. coriaceum represented by three individuals at the same site. It is perhaps worth noting that several of the endemic families included in the above total are somewhat dubious segregates— Pan- daceae (from Euphorbiaceae), Irvingiaceae (from Simaroubaceae), and Ixonanthaceae made up of the same plant families, with the exception of the Dipterocarpaceae for Le- guminosae substitution in Southeast Asia. Even at the generic level, there are striking floristic similarities between the compositions of lowland tropical forests on different con- tinents. The generic similarity is especially marked between Africa and South America. An average of 30% (with extremes of 25% at Korup to 34% at Belinga) of the genera at the six continental African sites are neo- tropical genera, nearly all also included in the neotropical samples. When complete local flo- ras are compared, generic concordance be- tween tropical Africa and the Neotropics re- mains equally high. Thus 30% of the genera represented at Makokou Gabon also occur in the Neotropics. Both sets of figures would be much higher if such tenuously differentiated genera as Pycnanthus and Virola (Myristi- caceae) or Macrolobium and its segregates (Leguminosae) were considered to be conge- neric. Generic overlap between tropical Asia and the Neotropics is less, averaging 23%, and between Australasia and the Neotropics in- termediate (25% neotropical genera in the Queensland sample, 26% in the New Cale- donia one). These relationships might be pre- dictable from Cretaceous and Tertiary plate tectonic history and the timetable of Gondwa- nan breakup. In this light, it is especially interesting that about 36% of the genera sam- pled at Perinet, Madagascar, are shared with the Neotropics, the highest value for any pa- leotropical site. ere are also consistent and predictable floristic changes along environmental gra- dients, at least in the Neotropics. On poorer soils families like Burseraceae, Lauraceae, and Sapotaceae become more prevalent, whereas on the richest soils palms and Moraceae are disproportionately speciose. In neotropical areas with a strong dry sea- son, floristic composition is likewise predict- able. Leguminosae are always the most species-rich family, with Bignoniaceae, rep- resented mostly by wind-dispersed lianas, al- ways second (Fig. 15). On an altitudinal gradient in the Andes, Lauraceae consistently replace Leguminosae as the most species-rich family at intermediate elevations (Fig. 16). Other families that con- tribute to the diversity of middle elevation forests are Rubiaceae, Melastomataceae, Eu- horbiaceae, Moraceae, Guttiferae, tree ferns, (hemiepiphytic) Araceae, and Palmae. Fam- Annals of the Missouri Botanical Garden o uj = o i | = z de ° | z o « n É Le v o |c] o 85/8 z z q . P o E lelo HR 213] oj |] Ë a |s|5 JE > uj <| e > fun] 2 mar ur < S š T Š o N o nm E MYRI 5 Y < ° u ° ° EE) > ° < o ==. x m m z £ Š z x ; ° É cc [s z u x - o o lt Q E 3 er O >a W Cw « o I O o o « a ANNO LEGU Formar m: Puedo a) species pe son i for 0.1- eot ropics From lefi to ii ps Je are 1) average are 20 lowland erage for 4 sites at 2,000—3,000 m (Sacramento, Finca Mehrenberg, Cerro Kennedy, Finca Zungara); 4) Cerro Kennedy, Colombia (2,600 m, 500 m? of sample area); ilies like Bignoniaceae, Sapotaceae, Myristi- caceae, Meliaceae, Sapindaceae, Bursera- ceae, and Chrysobalanaceae are especially noteworthy as absent or much more poorly represented than in lowland forests. At higher elevations (> 2,000 m), Melastomataceae, Compositae, Rubiaceae, and tree ferns be- come more prevalent, although of these only Compositae increase in absolute number of species. At even higher altitudes, Aquifolia- ceae, Myrtaceae, and Theaceae become rel- atively more important, while near timberline Compositae and Ericaceae predominate. Curiously, the site at 1,000 m altitude at Perinet, Madagascar, had ecd an iden- tical familial composition to t le- elevation neotropical site; in ‘adie to Lau- raceae being the most speciose family, Rubiaceae, Euphorbiaceae, Moraceae, and Guttiferae followed in species richness; the only substantial differences are a transposition of the roles of Melastomataceae (more species in the Andes) and Myrtaceae (more species at Perinet), the presence of several species of Monimiaceae and Oleaceae in Madagascar, and the frequency of hemiepiphytic Araceae in the Neotropics (Fig. 17). A Queensland, Australia, site from 850 m was also rather similar in familial composition to the Andean middle-elevation sites, again with Lauraceae dominating, closely followed by Rubiaceae, though with greater prevalence of such south- ern families as Proteaceae, Elaeocarpaceae, and Myrtaceae. Such strikingly repeated pat- terns in parts of the world so widely separated today can hardly be due to chance. any of the major latitudinal changes in floristic composition are well known, with fam- ilies such as Fagaceae and Juglandaceae re- placing the tropical taxa in North America (Fig. 18). Perhaps less emphasized are how remarkably similar in familial composition dif- ferent eastern North American forests are. While species, and to some extent genera, do change from place to place, from a world 5) Pasochoa, Ecuador (3,010 m, 200 m? of sample area). Site data from Table 2. Shortest column segments are two species tall. Volume 75, Number 1 1988 Gentry Plant Community Diversity SUAN uars =|18 fams. with 314 species each | PERINET, MADAGASCAR z E 8 TE Ë MERE | E | š | š |š |š | |š|š|š|š [ë [š [S]EISIEE] | Av. for 62 fams. with<2 spp. av. | AV. FOR 4 NEOTROPICAL SITES 1000-2000 m. FiGURE 17. Number of species per family for 0.1-ha sample at Perinet, Madagascar (950 m) (top two columns) compared with average for four mid-elevation neotropical sites (1,000—2,000 m) (bottom two columns). Note the remarkable similarity of familial composition. Perinet data based only on field identifications pending herbarium comparison of vouchers. Shortest column segments are two species tall. perspective the overall floristic composition of most of these forests is as similar as is their diversity. The contrastingly austral compo- sition of the Valdivian flora is also well known. There are also floristic similarities between the austral and north temperate ones. For example, gymnosperms and Fagaceae be- come more prevalent in both north temperate and south temperate areas. One interesting and previously unremarked floristic difference between the Valdivian forests and their north- ern equivalents is that the former lack sym- patric congeners. The difference in diversity between eastern North American and Valdi- vian forests (as well as between the North American forests and my two European sam- ples) is almost entirely accounted for by this lack of sympatric species in genera like Quer- cus and Carya. Why Chilean Nothofagus species, unlike their northern cousins, should be almost entirely allopatric is unclear, but the effect of this pattern on the diversity of the south temperate forests is obvious. DISCUSSION To this point I have attempted to present a series of observations of changes in diversity and floristic composition on various gradients. I now focus briefly on some theoretical gen- eralizations that would seem to derive from these data. The overall message is that plant com- munities are put together in decidedly non- random ways. Diversity and floristic com- position are highly predictable from environmental and geographical factors, with maximum plant community diversity occur- ring in full-tropical lowland areas with rich to intermediately infertile soils and high annual precipitation and/or little dry-season stress. Such patterns are often taken as evidence of niche saturation and community equilibrium (MacArthur, 1965, 1969; Cody, 1975; see also Whittaker, 1977). Much of the controversy about equilibrium vs. nonequilibrium communities has focused on the role of niche specificity vs. stochastic generation or maintenance of diversity (e.g., Hubbell, 1984; Hubbell & Foster, 1986; Ashton, 1969; Connell, 1978). My data sug- gest that even though tropical forests contain many different plant species, they are far from random assemblages. Can these data and conclusions be reconciled with the very dif- ferent ones of Hubbell (1979; Hubbell & Foster, 1986, 1987)? Below I will focus on several points that may be relevant to this debate. From a somewhat different perspective, some authors (e.g., Federov, 1966) have ar- gued that the exceedingly high diversity of tropical forests is too great to be accounted for by niche specificity; therefore, some kind of nonselective or stochastic mechanism must be invoked. However, it seems to me that it is stochastically most unlikely that the ex- treme species richness of forests like that at Yanamono, Peru, with 300 species out of 606 individuals in a hectare, would result 26 Annals of the Missouri Botanical Garden Av. of 20 lowl. Neotr. moist and wet sites z a "d "d Los Tuxtlas, Mexico TILI — POLY w per] — D MIRI 1122 CELA EUPH| SAPO PALM | SOLA] MYRT SAPO peas — — oO ARAC £ RUBI [on t ° RUBI D ` == < LAUR Š SAPI ° x — PIPE = BIGN : S e 2 E] o ° E = s E o [9] g < (0) . c x = z as o MORA 5 : o 2 Oo a 2 e ` E 1 — « o x c a = o S Sg S > ANNO = £ E 7 s c 4 E] 2 E i BIGN LEGU LEGU FACA FIGURE 18. Number of species per deii E "s 0. I-ha a at different latitudes. From lefi to average [p seven temperate North merita sites (28?15'N-39?2'N); 6) Suderhackstedt, Germany from random processes, unless there is a po- tential sample universe of many thousands of tree species. Forty-eight species are repre- sented in the first 50 individuals sampled at Yanamono, and the 65 individuals in the first Yanamono 0.1-ha subplot constitute 58 species. Such high levels of diversity, far from indicating stochasticity, would seem to indi- cate very strong ecological pressures resulting in phenomenally low densities of the individual species (and high community diversity). he striking regularities in the patterns discussed above clearly indicate that at some levels both community composition and di- versity are highly predictable. How this re- lates to community equilibrium remains clouded, however, in part because of defini- tional problems. Hubbell & Foster (1986) efined an equilibrium community as one in which a particular combination of species maintains itself against outside perturbations, whereas the predictable diversities of different tropical forests with similar environments but different asema of species Is more akin to the ` theory of island bio- geography (MacArthur & Wilson, 1967), considered by Hubbell as a nonequilibrium theory because of the taxonomic randomness Involved. Many Amazonian forests are clearly richer in tree species than equivalent Central American forests (Gentry, 1988). They also have much greater habitat differentiation and B-diversity (Gentry, 19863). Thus some of the higher diversity of the Amazonian forests may be due to the **mass effect" phenomenon of Shmida & Wilson (1985), with accidental immigrants adapted to other environments contributing significantly to the a-diversity of an individual Amazonian forest. Arguing along similar lines from the nonequilibrium view- point, Hubbell & Foster (1986) suggested that biogeographical pattern plays a major role in tropical forest a-diversity: if the re- gional diversity is greater, as it certainly is in Amazonia, more species, on the average, 4°N); 7) average for three Valdivian, Chile, sites ont 40?43'S). Volume 75, Number 1 1988 Gentry Plant Community Diversity should occur in individual forests due purely to phenomena associated with patch dynamics and local immigrations. But there are also problems with such interpretations. That the families and genera represented in these dif- ferent samples are so predictable strongly sug- gests that at least some kind of familial-spe- cific niches may be involved. Moreover, the apparent partitioning of the species of each family into different sets of species specialized for different substrates in Amazonia seems strong circumstantial support for selectionist interpretations (Gentry, 1985b). ata from several l-ha tree plots in the Tambopata Reserve, Madre de Dios, Peru, can be used to document the effect of sub- strate specificity on species composition. Data are available from two completely identified 1-ha samples from similar poor-soil terra firme forest separated by about 1.5 km, a com- pletely identified plot in mature forest on rich alluvial soil, and from as yet incompletely identified plots in young riverside secondary forest, in swamp forest, and in forest in a transitional area between the rich floodplain and poor-soil uplands (see Fig. 8). As is usually the case in species-rich tropical forests, most of the species sampled were represented by one or two individuals on a single plot and are inadequately sampled to draw any con- clusions about habitat specificity. Table 5 lists the species that occur in all three completely sampled plots plus all species that are common (i.e., > 10 individuals) in at least one of the plots plus a few other selected species. At one extreme are 13 species that occur both in the two poor-soil plots and in the fertile-soil alluvial plot. These might be classed as ecologically insensitive; none of them oc- curs in the secondary forest, six of them (and possibly more) in the swamp plot, and most (perhaps all) of them on the intermediate plot. All of these are essentially uniformly dispersed with similar numbers of individuals in each hectare. Bertholettia excelsa, which has 1- 2 large emergent trees per hectare through- out the Tambopata Reserve (except in sec- ondary forest), is a good example of this pat- tern. Other good examples include the subcanopy tree Leonia glycicarpa, which has 6-19 individuals per hectare, and the canopy tree Symphonia globulifera with 1-4 indi- viduals per hectare, again excluding the sec- ond growth plot. At the opposite extreme are the 21 habitat specialists listed in Table 5, locally common, but occurring in only a single habitat: good soil, poor soil, swamp, or second growth. The extreme case is Lueheopsis hoehnei, the ab- solute dominant in the swamp plot with 265 trees, but completely absent elsewhere; that this is only the second record of such a locally common species from Peru is instructive as to the state of Amazonian Peruvian floristic knowledge. Another example worth mention- ing is Sparrea schippii, previously unre- ported from Peru (or indeed from Amazonia), which is the fourth most common species in the alluvial plot. The large number of species that are completely faithful to a single habitat (presumably also including many additional less common species not listed in Table 5) is a good example of the importance of niche specificity in maintaining overall Amazonian species diversity. Each community is rich in large part because it has many species unique- ly adapted to a specific substrate. Perhaps more interesting from the view- point of ecological theory are the other two distributional categories indicated in Table 5. The first are species that are common in one habitat but also have a few individuals in one or more of the other habitats. Some of these may represent cases of “mass effect” (Shmida & Wilson, 1985), with an occasional individ- ual surviving but not reproducing outside its normal ecological range. The second pattern is species of the poor soil forest that are com- mon in one of the two sample plots but absent from the other. These may be examples of “ecological equivalents" (Shmida & Wilson, 1985), where due to some accident of dis- persal or establishment, a given species occurs at one site but not at another where it would be equally well adapted. The ecological equiv- alent hypothesis seems especially germane to Cordia, where Cordia mexiana and C. pan- amensis occur in one upland plot while C. 28 Annals of the Missouri Botanical Garden TABLE 5. Differences in occurrence and abundance in different 1-ha tree plots of some common Tambopata species (all species occurring in all three complete plots or with 10 or more individuals in any one plot plus a few others). Plot 1 is relatively fertile terra firme; plot 2 is swamp uen plots 3 and 6 are on poor sandy, upland terra firme; plot 4 is on rich alluvial soil; and plot 5 is in young riverside secondary forest. Plot Number l 2 3 4 5 6 Ecologically insensitive, + uniformly dispersed Bertholettia excelsa 2 (1) Eschweilera coriacea 6? Glycidendron amazonica 1 p — — Nr ww nA De kK =— 2 25 N> — @ > —1 S — WON AN ( Leonia glycicarpa 19 ( Lindackeria paludosa ( Minquartia guianensis Ocotea rubrineruis Oenocarpus mapora Symphonia globulifera Tapirira guianensis N S AWWA RAAN > O on Clarisia racemosa Ecologically sensitive but widespread Amaioua corymbosa l Euterpe precatoria 17 Iriartea deltoides 106 Iryanthera juruensis 5 Iryanthera laevis 17 Mabea — Pourouma minor 18 — — + — X — — N — — Ao — — — —— =— Awanrnwuea == _ ° N Ww — Pseudolmedia laevis 14 Siparuna decipiens 13 Socratea exorrhiza 10 TO + Tetragastris altissima l Ecologically restricted Cecropia membranacea — — — — 46+ — Ficus insipida — — — — 13 = Citharexylum poeppigti = E = — 94- = Sparrea schippii — — -— 17 — — Rinorea viridifolia X — — 27 — — Astrocaryum murumuru : — = 16 = E Myroxylon balsamum 1 ? — 6 = == Cordia lomatoloba — — — 2 — — Cordia nodosa X — — 1 — — Mauritia flexuosa — 41 = = — — Lueheopsis hoehnei — 265 Sc x — Pithecellobium latifolium = 12 = x 1 = Rouchera punctata 1 (2) 19 — = 15 Virola sebifera l Ouratea — — 14 — — 14 Euceraea nitida -— (1) 6 — — 8 Cedrelinga cateniformis -— — 3 — — 5 Cordia mexiana — — 10 == — — Cordia a panamensis = ? 4 — ies P Cordia ucayaliensis == - _ A _ 7 Volume 75, Number 1 1988 Gentry Plant Community Diversity TABLE 5. Continued. Major density differences not explainable by ecology Bixa arborea Hevea guianensis Pseudolmedia laevigata Plot Number 1 3 4 5 6 1 15 — — — 1 23 — 4 ? Ocotea (domatia) Arrabidaea tuberculata — X = present in Hartshorn plot but number of individuals not known. ? = presence or absence not known due to incomplete identifications (plot 1) or sampling not yet completed lot 2). (plot ( ) = species occurring in plot 2 only in corner on higher ground. toqueve and C. ucayaliensis occur in the other. Both of the latter situations represent the pattern thought to be prevalent on Barro Colorado Island (Hubbell & Foster, 1986), with high diversity of a given community due in large part to nonequilibrium fluctuations in its species. At a different level, my data on plant com- munity composition also seem much less pre- dictable and much more in accordance with the nondeterministic, nonequilibrium view- point. In the nine 1-ha tree plots that have been analyzed, there was not a single repe- tition of a most-dominant species. Although one (or a few) species is always much more common, there is a different “dominant” species in each plot. Even if the several most- dominant species in each plot are compared, there is little overlap. Considering only those species with 10 or more individuals per hect- are in at least one tree plot generates a list of 54 species documented to be relatively common locally somewhere in upper zonia. But of these, only five species are abun- dant on two different plots. The shared abun- dant species include Astrocaryum murumura, common on all three rich-soil plots (Yana- mono, Cocha Cashu, Tambopata alluvial), and Hevea guianensis, common on three poor- soil plots. The other shared abundant species are Otoba parviflora on two of the three rich- soil sites (Yanamono and Cocha Cashu) and Iriartea deltoidea on a different pair of rich- soil sites (Cocha Cashu and Tambopata al- luvial). But even though the same species are ma- generally not abundant at different sites with similar ecology, they are usually present. If we take the three rich-soil tree plots as an example, all the abundant Tambopata allu- vial-plot species are present in the Yanamono plot, and all the abundant Yanamono species, except Otoba glycicarpa and Carapa gui- anensis, are present at Tambopata. All of the abundant Cocha Cashu species are at Yana- mono, and all but four of the abundant Yana- mono species are at Cocha Cashu. If we com- pare the sandy-soil plots from Mishana and Cabeza de Mono, there is only one species abundant at both sites (Hevea guianensis), but all the common Cabeza de Mono species are present at Mishana, and most of the com- mon Mishana species are at Cabeza de Mono. In contrast, only four species that are abun- dant at any poor-soil site are present at all in any rich-soil site, all at Yanamono, which has an intermediate level of soil fertility. The relevant point is that although the species present at a site may be predictable, the fre- quency of a particular species in different forests seems entirely unpredictable and is likely determined stochastically. This is sim- ilar to the concept that Shmida & Wilson (1985) have termed “ecological equivalen- cy," i.e., the coexistence of species with ef- fectively identical niche and habitat require- ments for largely stochastic reasons. It is also the pattern that would be predicted by Hub- bell's (1979; Hubbell & Foster, 1986) com- munity drift theory. Indeed, Grubb (1986) generalized that the relatively sparse or rare 30 Annals of the Missouri Botanical Garden species that of necessity constitute the bulk of the species of species-rich communities should interact so infrequently with each oth- er that niche differentiation becomes largely irrelevant. he same conclusion arises from the 0.1- ha samples. There are almost always a few very common species in any sample. One of these is usually much more common than all the others; at 11 sites the most common species was between two and seven times more common than the second most common species. Yet the only repeat of a “most dom- inant" species among 25 moist- and wet-for- est sites is Catoblastus velutinus, shared be- tween Rio Palenque and immediately adjacent Centinela in western Ecuador. For 12 dry- forest sites there was not a single repeat of a “most dominant” species. Even if all 213 species that are dominant or subdominant in any of these samples (i.e., among the most common 5-10 species) are considered, only 38 are repeated in two or more different samples; ten of these repeated common species (13 if Rio Palenque and Cen- tinela are considered part of the same site) are in repeat samples of the same forest. Thus, only 24 species are abundant at more than one site. One species, Socratea exor- rhiza, is abundant at four sites, and three species—Jessenia bataua, Arrabidaea ox- ycarpa, and Arrabidaea pubescens—are abundant at three sites. Ten of the 24 species abundant at more than one site are shared between different dry forests, twelve are shared between different moist forests (typi- cally between Central America and Amazon- ia), and one (Mansoa verrucifera) is abundant in one dry-forest and in one moist-forest site. A major part of the debate on whether tropical-forest ecosystems are at equilibrium or nonequilibrium may be a by-product of the scale of a particular study or the focus of a particular author. The “rare” species that "random walk" through a 50-ha plot on Bar- ro Colorado Island are mostly common under- story or second growth species that would be regarded as permanent and continuous mem- bers of the more comprehensive moist-forest plant community that a biogeographer might define. The numerous microhabitat specialists suggested by a casual glance at a series of the Hubbell-Foster Barro Colorado Island dis- tribution maps become nonspecialists if the relatively few individuals that occur away from a favored habitat are emphasized. I suspect that differences in taxonomic focus may also relate to the interpretational differences. Hub- bell focused entirely on a particular combi- nation of species in addressing the question of community equilibrium. My data suggest that while the species that make up different communities may be very inconstant from place to place, at the same time the different families (and perhaps genera) that contribute to community floristic diversity are very con- sistent. Perhaps the family is the basic unit on which selection for low population densities (and thus indirectly for high species richness) occurs. For example, many seed predators and leaf-eating insects are host-specific at the generic or familial, as well as the specific, level (Janzen, 1975, 1980, 1984). If family- specific predators and/or family-specific com- petition are added to the scenario of dynamic forests with some niche differentiation, an ex- planation pleasing to selectionists and non- selectionists alike could begin to take shape. As indicated in Table 1, the familial diversity of tropical moist and wet forests, unlike the species richness, is both high and remarkably constant. It is certainly within the realm of possibility that this is due to ca. 50 family- specific niches in a given forest, whereas the differing species richness of different forests could be largely stochastically generated by factors relating to higher turnover in the more species-rich sites on better soils and with higher productivities. A different type of reconciliation, espe- cially of the differences between upper Am- azonian and Central American species rich- ness, their causes, and the equilibrium status of the forests involved, might come from a different approach to the data. Hubbell & Foster (1986) emphasized that niche differ- entiation among Barro Colorado Island species seems to consist mostly of separation into Volume 75, Number 1 1988 Gentry 31 Plant Community Diversity about a dozen generalized multi-species guilds based on degree of shade tolerance and pref- erence for edaphic or topographic microsites. Indeed they suggest that lack of niche dif- ferentiation might make possible the co-oc- currence of many potential competitors which are forced to share the same generalized niche. But by and large the lowland forests of Central America are composed of the same species that in Amazonia would be regarded as the most widespread and weediest species of their respective families or genera (Gentry, 1982a, 198 A relatively depauperate Central American forest made up of species adapted for weedy generalized strategies might have little in common with an upper Amazonian forest composed mostly of narrow-habitat spe- cialists. Given a few million more years of evolution, perhaps Central American forests might seem much closer to ecological equi- librium than they do today. t is no accident that many tropical biol- ogists who have considered the question of why tropical forests are so rich in plant species have greatly modified their original views, no matter on which side of the equilibrium/non- equilibrium question they began (Ashton, 1969 vs. 1984; Hubbell, 1979 vs. Hubbell & Foster, in press; Gentry, 1982b vs. 1982a and Gen- try & Dodson, 1987b). We still know so little about tropical forests that generalizations elude us. Almost certainly there are elements of truth on both sides of the question. Quite possibly different forests and the different taxa that make them up will often prove to be doing things quite differently. Although we have not yet reached anything like a consen- sus on how different factors, or even which factors, interact to determine diversity, it seems abundantly clear from the data pre- sented here that there are discernible and surely deterministic patterns in the species richness of different plant communities. I con- clude that which families, how many species, and possibly what individual species make up a tropical plant community are to a large extent deterministic and predictable from sim- ple environmental parameters; how common the species are and how they are put together into different communities may be completely random. LITERATURE CITED 1964. Ecological studies in the mixed dip- f, + ASHTON, P. t f Brunei State. Oxford Forest. Mem. F 25: 1-75. . 1969. Speciation among tropical forest trees: some deductions in the light of recent evidence. Biol. J. Linn. Soc. 1: 1 A sat ban s; of rainforest research to evolutionary theory. Ann. Missouri Bot. Gard. 64 94-705. 1984. Biosystematics of tropical qe plants: roblem of rare species. Pp. 497-518 i Grant a Plant Biosystematics. ieden Press, cological studies in the mixed dipterocarp forest of northwest Borneo III. Patterns of species richness (in press). AUBREVILLE, A. 1967. La forét po des Montagnes e Belinga. Biol. Gabon. 3: 96-112. E D. & P. Raven. 1978. Late Cretaceous and Tertiary vegetation history of Africa. Pp.77-130 in M. Werger ined Biogeography and Ecology of Southern Africa. W. Junk, The Hague. Brown, A. D., S. C. botes IKIAN & L. M. MALMIERCA. 1985. Estudio floristico-estructural de un sector de selva semidecidua del noreste Argentina 1. Compo sición floristica, densidad, y diversidad. Darwiniana 6: 27-41. CaAMPBELL, D. G., D. C. DALY, G. T. PRANCE & U. MACIEL. 1986. Guam ais sal prd of terra firme and varzea tropical forest on the Rio Copy, M. L ory of continental species diversities: bird ec, over Mediter- abitat gradients. Pp. ies in M. Cody è i y and Evolution of M elknap, sinc] Cambridge, Mas- setts Const LL, i H. 19 78. Diversity 1 ai live rain forests coral Lig 310. J. 8. c 199: y & L.J. Wess. 1984. Compen- satory P DE growth, and mortality as factors maintaining rain forest tree diversity. Ecol. Monogr 54: 141-164. Diversity. relations in pape th Croat, T. 1978. Flora of Barro Colorado bind. Stan- ford Univ. Press, Stanford, Californ Dawkins, H. C. The volume m of natural tropical high-forest and limitations on its improve- ments. Empire Forest. Rev. 38: 175-180. Dopson, C. & A. H. Gentry. 1978. Flora of the Rio alenque Science Center. Selbyana 4: 1-628. 19 Flora de Capeira. Banco Central del Ecuador, Quito. & F. M. VALVERDE. 1985. La Flora de Jaüneche. Banco Central del Ecuador, Quito. DONAHUE, P. K., T. PARKER III, B. SORRIE & D. SCOTT 32 Annals of the Missouri Botanical Garden Birds of the wo se Nature Reserve, Dept. Madre e Dios, Peru (in EITEN, G. 1978. Delimitation of the cerrado concept. Vegetatio 36: 169-178. Emmons, L. H. 1984. Geographic variation in a and diversities of non-flying mammals in Amazo Biotropica 16: 210-222. . H. GENTRY. 1983. Tropical forest struc- ture and the distribution of gliding ae rl tailed vertebrates. Amer. Natur. 121: 524. Erwin, T. L. Tambopata Reserved DR Madre d m Peru: history and deseripiion of the reserve. ista Peruana Entomol. 27: bano A. A. 19 The airibbire of the tropical rain forest ag Jas š in the humid tropics. J. Ecol. 54: 1- FLORENCE, J. & h A ADIK. 1980. Catalogue des pha- nérogames et des pteridophytes du Nord-est du Ga- bon (Sixième liste). Adansonia, sér. 2, 20: 235-253. GARTLAN, J. S., D. M. NEWBERY, D. W. Thomas & P. 1986. The influence of topography Bs soil phosphorus on the vegetation of Korup For- t Reserve, Cameroun. Vegetatio 65: 131-148. NA A. H. 1978. Diversidade e regenaragáo da capoeira do d com referencia especial as Big- noniaceae. Acta zonica 8: 67-70. Dira patterns and an tional species of t siflora vitifolia complex: Amazonian species o due to e e l dif- fanaa communities. Pl. Syst. Evol. 105 ————, 1982a. Dieu uad Bonen Taras: phy- nd South America, Pleistocene climatic icta, or ana cident of t e É orogeny? Ann. Missouri Bot. -593. Gard. vh connections hetwen D Y neotropical plant species diversity. Tai Bio -84. 1983. une E and diversity in neo- tropical forest up. . Sonderbd. Naturwiss. 303- -314 Verh. Hamburg 7 ——. 1985. dius vesulisdos preliminares de estudios botánicos en e nü Pp. 2/1-2/24 in a. Pi Datos para la Conservación, La Molina, Peru. 1985b. Plant begin and diversity pat- terns in Amazonian Peru. Natl. Geogr. Soc. Res. Rep. 10: 245-252. . Endemism in tropical vs. temperate plant communities. Pp. 153-181 in M. Soulé (edi- tor), Conservation Biology. Sinauer Press, Sunder- land, Massachusetts Species richness and floristic com- position of Chocó region plant communities. Caldasia 15: 71-91. 6c. Summario de patrones fitogeográficos neotropicales y sus implicaciones para el desarrollo en la Amazonia. Revista Acad. Colomb. Ci. Exact. 16: 101-116 Tree species richness of upper Ama- sonisri fopbátz: Proc. Natl. Acad. U.S.A. 85: 156- 159 e transect method. /n: A. Peixoto Mops Manual for Phytosociology and Floristics of the Mat Atlantica. CNPq, Brasilia (in prep. Trop. 35 (Suppl. 1): 7 —— & — ——. & C. Dopson. 1987a. Contribution of non- trees to species richness of tropical rain forest. Bio- tropica 19: 149-156. ——— & 1987b. Diversity and phytogeog- s. Ànn. Mis- raphy of neotropical vascular epiphyte souri Bot. Gard. 74: 205-233. & L. H. Emmons. 1987. Geographical vari- ation in fertility and composition of p: understory of neotropical forests. Biotropica 19: 216-227. & J. TERBORGH. 1988. Comparison and dy namics of the Cocha Cashu mature floodplain forent. In: À. ded (editor), Four Neotropical Forests. Yale Univ. Press, New Haven, Connecticut (in press). GIVNISH, d ~ J. W. TERBORGH & D. M. WaLLER. Plant mporal community structure, and species ra in forest herbs of the Virginia piedmont (unpubl.). Gruss, P. J. 1986. Problems posed by sparse and patchily distributed species in species-rich plant com munities. Pp. 207-225 in J. Diamond & T. Caso (editors), Community Ecology. Harper & Row, New York ork. HarL, J. B. & M. D. Swaine. 1981. Distribution and Ecology of Vascular Plants in a Tropical Rain Forest, e boma et Belinga. Biol. Ga 1 Seconde liste des Phanérogames et Pteridophytes du N.-E. Gabon n Belinga et Mekambo). Biol. Gabon. 1: 337-34 LE THomas. 1967. role liste des oo du N.-E. Gabon. Biol. Gabon. 3(2): 3-120. Quatriéme liste des Pha- nérogames et ae du N.-E. Gabon (Bassin de l'Ivindo). Biol. Gabon. 6: 131-138. HaRTSHORN, G. S. 1983. Plants (Introduction). Pp. 118-183 in D. Janzen (editor), Costa Rican Natural History. Univ. Chicago Press, Chicago HLADIK, A. 1978. Phenology of leaf producing in rain Ecology of Arboreal Folivores. Smithosonian Inst. ress, Washington, Ss N. HALLE 1973. Catalogue des Phanéro- u N.-E. F a oo liste). Adan- 97 9. ie so E abundance, and diversity ina tropical dry forest. Science 203: 1299- 1309. . 1984. Methodologies for the study of the origin and maintenance of tree aa in tropical rain forest. Pp. 8-13 in G. Mau echon, M. Had- ley & T. Younes (editors), The Sicha of Species Diversity in Tropical Forest Ecosystems. Biology In- ternational (IUBS). o Issue 6. & FOSTER. iology, chance and history and the E of tropical rain forest tree communities. Pp. 314-329 in J. Diamond & T. Case (editors), Community Ecology. Harper & Row, New York. & 1987. La estructura especial en gran escala de un eps neotropical. Revista Biol. E Structure, dynamics, and equilib- Volume 75, Number 1 1988 Gentry 33 Plant Community Diversity rium status of old-growth forest on Barro Colorado Island. /n: A. Gentry (editor), Four Neotropical For- ests. Yale Univ. Press, New Haven, Connecticut (in press). Huston, M. 79. A general v oppo of species diversity. Amer. Naturalist 113: 81- 80. Soil nutrients and tree Pop rr n Costa Rican forests. J. Biogeogr. 7: 147- jubes D. H. 73. Sweep samples of tropical bus insects: effects of seasons, vegetation types, eleva- tion, time of day, and insularity. Ecology 54: 687- 708. 97 Interactions of seeds and their insect predators/parasitoids on a tropical deciduous forest. 154-186 in P. Price (editor), Evolutionary Strat- egies of Parasitic Insects and Mites. Plenum Press, New York . Specificity of seed-attacking beetles in a Costa Rican deciduous forest. J. Ecol. 68: 929- 52. 1984. Two ways to be a tropical big moth: Santa Rosa saturniids and sphingids. Oxford Surv. Evol. Biol. 1: 85-140. & R. LIESNER. 1980. Annotated checklist of lowland Guanacaste Province, Costa Rica, exclusive of rue and nonvascular cryptograms. Brenesia 18: 15-90. , D. M. ATAROFF, M. Farinas, S. REYES, N. Rincon, A. SOLER, P. SORIANO & M. VERA. 1976. Changes in the arthropod community along an ele- vational transect in the Venezuelan Andes. Biotropica 8: 193-203. Jones, E. W. 1956. Ecological studies on the rain forest of southern Nigeria IV (Part 2). The plateau forest of the Okomu Forest Reserve. J. Ecol. 44:83-117 Kapos, V. 1982. An Ecological Investigation of Scler phylly in Two Tropical Forests. Ph.D. Dissertation Washington University, St. Louis, Missour KARTAWINATA, K., R. ABDULHADI & T. Pa uA Composition and structure of a lowland dip- terocarp forest at Wanariset, East Kalimantan. Ma- layan Forester 44: 397-406 Keay, R. W. J. 1953. An Outline of Nigerian Vege- tation, 2nd edition. Ministry of Information, Lagos. Lamas, G. 1985. Los Papilionoidea lef nd de la Zona Reservada de Tambopata, e de Dios, Peru. I. Papilionidae, Pieridae y Nymphalidae (En Parte). l. 27: 59-7 ntomo. Diei n t. Mod 1985. oe Syst. Bot. Missouri Bot. Gard. 25: 187-2 Lor- Herce AS, A. 76. La estación de biología trop- al Los Tuxtlas: pasado, presente y futuro. Pp. 31- in A. Gómez-Pompa, C. Vazquez-Yanes, S. del Amo Rodriguez & A. Butanda "i De esie: Rege- neración q Selvas. CECSA, Mexi Lorr, E. J., S ILLOCK & J. A. c 1987. CIUS diversity and structure of a tropical deciduous forest of coastal Tales: Biotropica 19: -235. MACARTHUR, R. 1965. Patterns of species diversity. Biol. Rev. 40: 510-533. —— ———. 1969. Patterns of communities in the tropics. Biol. J. Linn. Soc. 1: 19-30. . Geographical Ecology: patterns in the distribution of species. Harper ow, New Yor & . Witson. 1967. The Theory of [dki Higese noh: Princeton Univ. Press, Princeton, New Jersey. May, R. 1975. Patterns of species abundance and J Belknap, Cambridge, Massachusetts. 980. Flora del Parque Nacional Pu- yehue. Editorial Universitaria, Santiago, Chile. NaveH, Z. & R. H. WHITTAKER. 1979. Structural and floristic diversity of shrublands and woodlands in northern Israel and other Mediterranean areas. Veg- N. L. CHRISTENSEN. 1980. Hardwood forest vegetation of the North Carolina Piedmont. ids Geobot. Inst. ETH Stiftung Rubel Zürich 69: PEIXOTO, A. 8: A. H. GENTRY. Diversidade e composição da mata de tabuleiro da Reserva Florestal de Linhares (Espirito Santo, Brasil). Rev. Bras. Bot. (in prep.). Pires, J. M. € G. T. Prance. 1977. The Amazon ii a natural heritage to be pr reserved. Pp. 158- 193 in G. Prance & T. Elias (editors), Extinction Is ae New York Botanical Garden k PRANCE, C. T. 1977. Floristic inventory P the tropics: where do we stand? Ann. Missouri Bot. Gard. 64: 659-684 , W. A. RODRIGUEZ & M. F. pa SiLva. 1976. Inventario forestal de um hectare de mata de terra . 30 da estrada Manaus-—Itacoatiara. Acta Si e ds structure and floristics. J. Ecol. 71: -260 RAMÍREZ a M. RivEROS G. 1975. Los alerzales de Cordillera Pelada: flora y fitosociologia. Medio Ambiente 1: 3-13 Raven, P. H. 1976. "Ethics and attitudes. Pp. 179 in J. Simmons et al. (editors), Conservation of Threatened Plants. Plenum, New York, on. AXELROD. 1974. — dap biogeog- raphy and past continental movemen . Mis- 9-67 souri Bot. Ga 1: Rice, B. & M. Westosy. 1983. Plant species richness at the 0.1 hectare scale in Australian vegetation : re compared to other continents. Vegetatio 52: 129- RICHARDS, P. W. 1939. Ecological studies on the rain forest of southern Nigeria. I. The structure and flo- ristic composition of the primary forest. J. Ecol. 27: 1-61. 52. The Tropical Rain Forest. Cambridge Üniv. Press, Cambridge. 963. Ecological notes on West African veg- isin III. The upland forests of Cameroons moun- tain. J. Ecol. 51: 529-554. Riveros G., M. & C. Ramirez G. bus Fitocenosis epifitas de de asociación Lapagerio-Aextoxiconetum en el Fundo San Martin (Valdivia. Chile). Act. Ci. Venez. 29: 163-169. Annals of the Missouri Botanical Garden SaLo, J., R. KALLIOLA, I. HAKKINEN, is MAKINEN, P. NIEMELA, M. PUHAKKa & P. CoLE dynamics and the diversity of pee lowland forest. Nature 322: 254 Scorr, N. J., JR. of the h tof. 1976. The abundance and diversity of tropical forest litter. Biotropica r SuMIDA, A. & M. Wilson. 1985. Biological determi- nants of p E diversity. J. a Q 12: 1-2 STALLINGS, J. L. West, W. Hann & I. GAMARRA. Primates = their pd to habitat i in the Par- aguayan Chaco. /n: K. Redford & J. Eisenberg (ed- iud Mrs of the reae Essays in Honor of Ralph M. Wetzel. In STARK, N. A. H. GENTRY & H. on URING. Prediction of tropical plant species idu from soil and precip- ate V eder and other birds 3 Indian Cave State. Park. Nebraska Bird Rev. 37(4): Tusa, G. D. 82. Resource Competition and Com unity Structure. Princeton Univ. Press, Princeton, New erse rimental 1984. PI td nutrient gradient. Ecology 65: 1445- 5 1453. 9 Trot, R. G. Eisenberg (editor), Vertebrate Feology i in the North- ern Neotropics. Smithsonian Inst. Press, Washing- ton, D.C. WALTER, H. & H. LierH. tatlas. Fischer-Verlag, Je Warinc, R. H. & J. F. een Evergreen 1979. pl ec forests of the Pacific Northwest. Science Lr Klimadiagramm-Wel- Ea stern olga eastern North Amer- e plant road level. . 70: 734-141. 1984. "Tropical Rain Forests of the t. Clarendon, Oxford. Wanwa R. H. 1977. Evolution E ue diversity communities. Evol. Biol. 10: 1-87. ZEE M.& . WAGNER. 1913. A description woody vegetation of oak-hickory forest in the northern ers Mos Bull. Torrey Bot. Club 106: PATTERNS OF VASCULAR PLANT DIVERSIFICATION IN THE FOSSIL RECORD: PROOF AND CONJECTURE Karl J. Niklas! ABSTRACT n analysis of historical trends in diversification can suffer from a variety of defects and limitations. Amon events since peaks bove the species level. However, genus n ee surrogate database for UE patterns of species origination. Computer simulations rev ealing patterns of family extinctions may be in species and family extinctions rarely, volum coincide. A pluralistic approach to evaluating nadequate for evaluating mass species extinction diversification is advocated involving the examination Er biotic changes within assemblages and trends in morphological, anatomical, and reproductive evolution The notion of diversity in paleontology differs little from its use in ecological studies (Raup & Stanley, 1971; Whittaker, 1977). Diversity can be defined as the number of taxa in a community or as the synthetic char- acteristic of taxonomic richness and equita- bility, i.e., the relative evenness of the im- portance values of taxa within a sample (Lloyd & Ghelardi, 1964). However, the nature of the fossil record precludes a direct compari- son of diversity between past and present biotas (Krassilov, 1975; Padian & Clemens, 1985). In ecology. taxonomic richness re- flects the number of species in a sample of standard size, while measures of equitability involve some gauge of the abundance or pro- ductivity of one species divided by the im- portance values of all other species within the sample (Pielou, 1977). By contrast, fossil as- semblages usually reflect time-averaged sam- ples of a geographically ill-defined area. Fur- ther, terrestrial organisms are preserved where they were buried, not generally where they lived, and most often after significant periods of decomposition (cf. Niklas et al., 1980; Pa- dian & Clemens, 1985). Frequently, trans- port to sites of burial results in the disarticula- tion of land plants and animals, and in deposition of parts in sedimentologically dif- ferent microenvironments (cf. Kidwell, 1986). Consequently, there is great difficulty (1) in recognizing a species from its parts, (2) reas- sembling the components of a community, and thereby (3) arriving at a measure of di- versity that is comparable to contemporary ecological studies (Raup, 1976, 1979; Niklas al., 1980; Knoll, 1984; Benton, 1985). Despite its limitations, the fossil record pro- vides a potentially valuable perspective on a number of evolutionary issues. The paleon- tologist can compare patterns of diversifica- tion over billions of years of Earth's history and can track the origin, radiation, and even- tual taxonomic diminution of organisms that no longer exist or are rare in present biotas. Although the nature of paleontological data defines the temporal and taxonomic resolution with which evolutionary issues can be ad- 14853, ANN. ! Section of Plant Biology and Section of Ecology and Systematics, Cornell University, Ithaca, New York U.S. A. Missouni Bor. GARD. 75: 35-54. 1988. 36 Annals of the Missouri Botanical Garden dressed, recognition of large-scale patterns that develop, change, and disappear over long intervals of time confers a perspective that augments neontological studies. This paper is concerned with changes in tracheophyte diversity over the last 400 mil- lion years of the Phanerozoic. The fossil re- cord of terrestrial vascular plants is reviewed and discussed within the context of large-scale patterns in composition and taxonomic rich- ness of assemblages through time. The bulk of the data to be discussed has been presented elsewhere in a variety of forms (Niklas et al., 1980, 1985; Tiffney, 1981; Knoll, 1984, 1986; Knoll et al., 1984). Consequently, many facets of tracheophyte diversification will be referenced but not discussed in detail. The principal focus of this paper will be on the limitations of the data and on the quali- tative statements about plant evolution that can be made from an inherently imperfect fossil record. LARGE-SCALE CHANGES AND TAXONOMIC LEVEL The species is the taxonomic level of choice for determining changes in diversity through geologic time. However, species-level com- pilations present a variety of procedural dif- ficulties, the most significant of which involve sampling errors (Signor & Lipps, 1982; see also Pease, 1985). Comparisons of global pat- terns in species diversification with higher taxonomic levels are desirable, because they provide insights into the information to be gained or lost by each type of compilation. The upper panel of Figure 1 shows a plot of the total species-richness of vascular plants through the Phanerozoic. The data are taken from primary paleobotanical citations tabu- lated by Niklas et al. (1985) and are segre- gated into three major categories of tracheo- phytes: pteridophytes, gymnosperms, and angiosperms. The plot shows a Silurian to mid- Devonian radiation of primitive pteridophytes followed by a Carboniferous proliferation of more advanced or derived pteridophytes and early gymnosperms. A Permo-Triassic de- crease in overall diversity precedes a more or less constant level of species number until the middle of the Cretaceous when angio- sperms begin to radiate. Diversity of pteri- dophytes decreased during the Permo-Trias- sic and reached a relatively constant level throughout much of the Mesozoic and Ce- nozoic. With the advent and proliferation of angiosperms, the number of gymnosperm species decreased significantly through much of the late Cenozoic. The lower panel in Figure | shows large- scale temporal patterns in diversity at the family level. These data (from Knoll, 1984) were compiled independently from the data on species from Niklas et al. (1980). Data for pteridophytes and gymnosperms are not segregated, thus providing for direct com- parisons between species- and family-level di- versity for nonangiosperms and for angio- sperms. Differences in the diversity between Devonian and Carboniferous families are less pronounced than those seen in the top panel. As Knoll (1984) noted, the number of species in late Paleozoic families is higher than in early Paleozoic families, presumably due to the appearance of more complex or numerous morphological and anatomical features upon which species can be based. Ordinal changes in plant diversity compiled by Knoll (1984) conform in large measure to the trends seen in the family diversity plot. Despite the differences between species and family diversity, it is evident that the number of vascular plant taxa increased throughout much of the Paleozoic and has accelerated during the Cenozoic with the advent of an- glospermy. The value of comparing diversity among different taxonomic levels can be further il- lustrated by examining data for a specific group rather than the entire domain of tra- cheophytes. The selected group ought to have a fossil history that spans the Phanerozoic and thus obviates any idiosyncratic feature of a geologic period. Three major groups of vascular plants have a virtually continuous fossil record since the Upper Devonian: ly- copods, sphenopsids, and ferns (Stewart, 1983). Of these three groups, the ferns are the most abundant through time. Hence, they Volume 75, Number 1 1988 Niklas 37 Vascular Plant Diversification in Fossil Record 421408 360 286 248 213 144 65 2 700 o 600 /. o / u ° o 500[r FM w o” a of 9 400l Y | ANGIO- u o SPERMS Oo 7 Le—e—e—ele., a 300r ed Ao ` 3 ul e N o ° m Ny s Y A z e IA GYMNO- e. = 200+ Py t SPERMS ° z "A N a 1 Iool- d, AA eo” PTERIDO- | "aT 20-800. Lo “ep” — PHYTES Bo ole | I&O loo O / o 140} o - I20F / 7 O m 3 Ioor - z < u BOF ° J u / : | 7 « 60 ^| ANGIO- a / SPERMS = L- 4 = 40 Vi z o-e*. —° e A N l|-e—e-—e*T Pe e-e-e—e0-e 20F 7 R-*-e e... h pt NON-ANGIOSPERMS (0) e T T T T T T T T TN T T | T T T T Y P: g ul. mujt v N w s|. MuLMU| L M u| L a-a u Mia Eo Ot Mi PI S C P R J K T noo T 421408 360 286 248 213 144 65 e GURE l. Large-scale changes in diversity of tracheophytes at the species- and family-level (upper and Westphalian, S — Sp kanian jn the Cretaceous (K) Cenomanian-Campanian, and M = Maastrichtian. are convenient for comparing differences in diversity based on compilations of species, genera, and families. The most complete treatment of fossil ferns is that of Boureau (1970). Unfortunately, this does not reflect recent advances in taxonomy nor does it pre- sent a critical evaluation of stratigraphic oc- Phanerozoic. Species no diversity data were 1984). The wankaq = is , T = n, N = Nam , W= L = Berriasian-Barremian, p ie Aptian- rela) U= currences (see for example Phillips, 1974; Stewart, 1983). Nonetheless, the data con- tained within this volume are useful because they are easily accessible and reflect the state of paleobotanical knowledge as compiled in a single reference. Accordingly, the data for fossil ferns are dealt with here solely for the 38 Annals of the Missouri Botanical Garden 421 rá 360 286 248 213 144 65 2 esL ^ SPECIES A _ o GENERA / | € FAMILIES — J| A x A < - 7 N 4 P a" Pme u 3Or O © | El ° š N "d Nol Sono” | ac ui 20} /N NM U - m o o e = ij J \ as o-¢ a > vy z!Or ao tom DO e^ d z A nd ^ + 4 O S| D C P R J K T 421408 360 286 248 213 144 65 2 FIGURE 2. Taxonomic diversification of ferns based on data pious by sage eatin Stratigraphic resolution of s iU genus, and family occurrences from the 1. Consequently, the time-scale of this figure is ns precise than ruris plotted in Figure purpose of comparing the qualitative diversity measurements made at different taxonomic levels within a major plant group. igure 2 shows the total diversity of fossil ferns based on species, genus, and family compilations. Since the citations given in the data sets are rarely more specific with regard to stratigraphy than lower or upper, middle, and lower divisions of each period, the res- olution of the geologic scale is necessarily more crude than desirable and not directly comparable to those given in Figure 1. Peaks in the species diversity of ferns occur in the Lower and Upper Carboniferous, the Middle Jurassic, and in the Eocene. The max- imum number of fossil fern species is recorded in the Eocene. Species numbers significantly decrease in the Middle Carboniferous, Lower and Middle Triassic, and in the Oligocene. Generic tabulations reveal similar peaks and valleys in diversity as those seen in the species data; however, the overall diversity maximum in the Lower Eocene is less pronounced in the generic data set than in the species com- pilation. Family diversification is muted throughout much of the Phanerozoic. None- theless, the Triassic decrease in diversity (from ten families in the Upper Carboniferous to six source is coarser than that of species diversity families in the Middle Triassic) occurs in both the species and generic plots. at then is the appropriate taxonomic level for gauging long-term patterns in di- versification? Large-scale changes in tracheo- phyte diversification at the species and family levels are poorly correlated (Fig. 1). This is not surprising given the diversity of lineages that comprise the tracheophytes. The diver- sification of tracheophyte species may show episodes of increase due to the radiation of a few families even if the majority of the re- maining families dwindle in species numbers or go to extinction. Even for a particular group of plants, such as the ferns, there ap- pears to be only a loose correlation between changes over geologic time in species or genus and family numbers, presumably for much the same reason— “taxonomic stacking." The family may not be the appropriate taxonomic level for analysis for many types of evolutionary questions, as, for example, the identification of mass-extinction events (see next section). A family is eliminated from the fossil record with the extinction of its last surviving species. This terminal extinction event may occur well after the major period of species-attrition within a family. Thus, there Volume 75, Number 1 1988 Niklas 39 Vascular Plant Diversification in Fossil Record is a double-edged effect in the use of higher taxa for determining unusual perturbations in overall diversity. A ““mass-extinction event" may be due to the geologically synchronous elimination of the relic species of numerous, species-depauperate families. On the other hand, many species-rich families could undergo a geologically sudden and severe de- pauperation in species without the total dis- appearance of these families from the fossil record. A dramatic reduction in taxonomic diversity below the family level would reflect a significant event in the history of life that could go undetected in family tabulations of diversity. Similar arguments could be made concerning the effectiveness of genus tabu- lations in detecting intense episodes of species extinction. Species compilations comprise the most bi- ologically relevant database for paleontology. However, a paleospecies is not operationally defined or identified at the same level of taxo- nomic scrutiny or with the same body of in- formation and experimental format as a species of living organism. This severely limits the application of conventional species concepts to fossil material. Recognition of a fossil plant species is per- haps even more difficult than that of a fossil animal species. Plants are phenotypically more plastic than many major groups of animals (Cahn & Harper, 1976; Harper & Bell, 1979; Harper, 1985; White, 1979). The recogni- tion of ecotypes or taxonomic varieties of living plants is notoriously difficult and re- quires detailed transplant experiments, cy- tological and electrophoretic examination, and field observations of population dynamics across environmental gradients (Harper, 1977, and references therein). Fossil plants are most frequently found as disarticulated organs which, for many plants, have been previously assigned to separate taxa (Knoll & Rothwell, 1981; Stewart, 1983). Only when found in organic connection can organ genera be placed in synonymy. Consequently, the recognition of genus as an “organ” can only e done on an ad hoc basis. Since whole plant reconstructions are still rare in the paleobo- tanical literature, the inflation of species rich- ness due to the effects of dealing with frag- mented parts is a serious limitation to assessing the true taxonomic diversity in a fossil assem- blage. Perhaps the only reasonable solution to the choice of taxonomic level is to deal with species and genus tabulations for each family. Cor- relations in the patterns of diversification among all three taxonomic levels for partic- ular groups of plants could then be used to estimate long-term trends or geologically sud- den changes in taxonomic richness. As yet this suggested type of multiple-level analysis has not been undertaken for any group of organisms. Mass EXTINCTIONS: AN ARTIFACT or HIERARCHY? Measurements of diversity based on taxa above the species level have been used to determine large-scale evolutionary patterns. Perhaps the best example of this approach comes from the use of family-level data on marine animals to determine major diversi- fication and extinction patterns (Sepkoski, 1980, 1981a, b, 1984; Raup & Sepkoski, 1982, 1984). Sepkoski (1981a, b, 1984, and elsewhere) has argued that family data rep- resent a good proxy for estimating diversity of species (cf. Sepkoski et al., 1981). In ad- dition, comprehensive data with good strati- graphic resolution do not exist for marine animal species or even genera (e.g., Sepkoski, 1984: 247-248). However, as will be shown in this section, the behavior of families as evolutionary “units,” and in particular the pattern of family extinction rates, is depen- dent upon the frequency distribution of species within families. Raup & Sepkoski (1982, 1984) identified five statistically significant **mass extinction" events in the Phanerozoic record of marine animal families: late Ordovician (Ashgillian), late Devonian (Frasnian), late Permian (Gua- dalupian-Dzhulfian), late Triassic (Norian), and late Cretaceous (Maastrichtian). Although the magnitudes of species-extinctions are hard to 40 Annals of th Missouri ECCE Garden 30 r ITERATION 0) IT 15 FAMILIES 57 F 41 20 SPECIES 48275 B S 1698 o uJ 3 = 2.0 Q qe > $ = ° W = L5 N do u a E ` > l o e-? e—e— e — e —e — h o dL o 1 1 1 L 1 | O IO 20 30 40 50 NUMBER OF ITERATIONS FIGURE 9. Patterns of the total extinction rate com- puted for a species frequency distribution with man species-poor families and a few species-rich families (N — 111). A total of 43,070 species are represented in this artificially created distribution. of family extinction would not coincide with a mass species-extinction event: by the 15th iteration, over 96% of all species are elimi- nated, despite the survival of all families. Similar incongruities between the extent to which family extinctions reflect or coincide with species extinctions are seen in the other two simulations. An even distribution in the number of families containing the various cat- egories of species numbers produces an as- cending but spiked pattern of family-extinc- tion rates (Fig. distribution skewed toward many species-poor families yields a broad, irregularly plateaued pattern of family- extinction rates, which is somewhat similar to the pattern generated from the dicot data set ig. ese two frequency distributions are dearly artificial since neither is encoun- tered in real species, genus, or family tabu- lations. Nonetheless, they are of interest be- cause they indicate that no frequency distribution produces a correlation between species- and family-extinction patterns. omputer simulations do not indicate that family-level compilations are necessarily in- adequate to gauge “‘mass extinction” events. learly, there is no reason to assume that E Volume 75, Number 1 1988 Niklas 45 Vascular Plant Diversification in Fossil Record species extinctions are randomly mieu among all families. However, (1) "spikes" in the regression of family-extinction rates against geologic time can reflect the syn- chronous demise of many species-poor fam- ilies belonging to a once robust taxonomic (or grade-level) cohort; (2) patterns in family- extinction rates are dependent upon the fre- quency distribution of species among the fam- ilies considered; (3) mass species-extinction and mass family-extinction events may not coincide; indeed, the vast majority of species within a clade may disappear well before the event is noticeable at the family level (as in the case of Ginkgo); and (4) analyses of mass extinctions require information on the chang- ing patterns of species distributions within suprageneric taxa. Similar arguments can be made for the use of generic tabulations to detect mass-extinction events (cf. Raup Sepkoski, 1986; see also Hoffman, 1986). Similarly, diversity measured solely on the basis of species numbers ignores the relative abundance of individuals within taxa— an es- sential feature to understanding the ecological significance of an extinction event. The “mass extinction” of many species that contribute only a fraction of the biomass to a biota has a decidedly different effect than one involving species contributing many individuals to a biota. NONTAXONOMIC MEASUREMENTS or DIVERSITY A treatment of diversity strictly from the perspective of changes in species numbers ignores many significant aspects of plant evo- lution. Qualitative changes in reproductive and vegetative morphology and quantitative changes in the numbers of individuals within a taxon are not reflected in patterns of chang- ing species numbers. For example, the num- ber of gymnosperm species during the Juras- sic and Cretaceous remained relatively constant despite considerable taxonomic turn- over within various lineages and significant alterations in morphology and anatomy of representative species. Ginkgo biloba, as a species, occurs from the Mesozoic to the pres- ent, yet the number of individuals significantly declined during the Cenozoic. Clearly, eval- uations of diversity based on something other than species numbers are important in eval- uating adaptive evolution and changes in com- munity structure (Spicer & Hill, 1979; Phil- lips & DiMichele, 1981; DiMichele et al., 1985; Fisher, 1985). Another reason for evaluating nontaxo- nomic measurements of diversity is that in- ferences on genomic (hence speciation) rates of evolution based on rates of morphological evolution may be faulty. Schopf et al. (1975) argued that differences in the number of species among lineages can be the conse- quence of dealing with taxa that differ in their degree of morphological complexity. If fossil species truly represent genomically distinct entities (sensu the biological species concept), then rates of morphological and genomic evo- lution would be highly correlated. These au- thors, however, assert that this direct cor- respondence can never be proven. Therefore, it is advisable to look at rates of morphological evolution and rates of paleospecies origination separately. Fortunately, changes in species numbers and within-assemblage species composition, and morphological patterns of long-term evo- lution can be dealt with separately to visualize tracheophyte diversification. As in most cases with the use of paleontological data, however, quantitative analyses can only be used to con- struct qualitative comparisons or generaliza- tions. Among these guarded generalizations are: (1) The taxonomic composition of fossil ag assemblages is rarely, if ever, stable. is altered most significantly by the Le. stitution of taxa within related lineages during periods of relatively constant over- all species numbers. These “intrataxo- nomic" alterations occur over extended geological time-scales and are generally oe among lineages that share a mmon mode of sexual reproduction (Knoll, 1986). 46 Annals of the Missouri Botanical Garden TABLE 1. Mean species numbers, X, of vascular plants in fossil assemblages for representative intervals. Niklas et al., 1980 Knoll, 1986 X (number X (number f floras) of floras) Early Devonian 5 (7) 4.0 (8) Late Devonian 8.2 (9) 10.7 (7) Early Mississippian 10 (7) 11.9 (15) Late Mississippian 22.5 (2) 23.3 (8) Late Jurassic 27.5 (15) 30.8 (6) Early Cretaceous 21.5 (15) 30.2 (22) Late Cretaceous 43 (7) 54.4 (17) (2) Global species numbers generally in- crease most significantly with the radia- tion of plants sharing a novel mode of sexual reproduction (Niklas et al., 1980, 1985). (3) At least within the temporal resolution permitted by most paleontological studies, changes in taxonomic composition and large-scale species numbers are tightly linked to changes in vegetative morphol- ogy and the appearance of diverse growth habits (Niklas, 1987) (4) The appearance of new reproductive modes and diverse morphologies within a clade or cohort of clades may not nec- essarily result in the ecological displace- ment of previously existing species. Throughout much of the Paleozoic and Mesozoic, taxonomic radiations are as- sociated with the exploitation of environ- ments not previously occupied (Tiffney, 1981; Knoll, 1986) (5) The competitive advantages conferred upon a taxon by novel reproductive or vegetative capabilities are most pro- nounced early during its taxonomic ra- diation (Knoll et al., 1984; Knoll & Nik- las, Each of these generalities is not without exception or debatable inference, since each is based on a limited number of studies. It is more instructive to review the nature of the data upon which these statements are based and to examine their deficiencies. To date, only two studies have examined changes in the mean species numbers of plants throughout the Phanerozoic (Niklas et al., 1980; Knoll, 1986). In both cases, fossil as- semblages (= “floras””) were selected from similar depositional environments so as to minimize the differential effects of sedimen- tological factors on preservation. The floras were selected from what were inferred to be warm climates. Comparison between the mean species numbers per flora for representative time periods is given in Table 1. Despite dif- ferences in the sources of data, both studies reported remarkably similar trends. Mean species numbers within floras have increased significantly at least twice during the last 420 million years—doubling between Late De- vonian and Late Mississippian floras, and once again between Early and Late Cretaceous flo- ras. (The lack of an objective baseline com- parison for the Late Devonian floras precludes evaluation of the data from Early and Late Devonian floras. Each of the two increases in mean species number coincides with an increase in overall species diversity (Niklas et al., 1980, 1985). This is to be expected, since the latter incor- porates data used to compile within-floras species numbers. However, each of the two increases in mean species numbers correlates with major transitions in the taxonomic com- position of floras. This is shown elegantly in a study by Knoll (1986), who presented a unique analysis of the taxonomic composition of floras throughout the Phanerozoic. Figure 10 (redrawn from Knoll’s study) shows two significant resortings of suprageneric groups within floras. One occurs with the advent of seed plants in the Late Devonian and Missis- sippian; another reflects the radiation of flow- ering plants in the Cretaceous. If the various plant lineages within Knoll’s data set are grouped according to their principal modes of reproduction, then the changes in taxo- nomic composition are seen more clearly (Fig. 11). As in the plots of large-scale species numbers (Fig. 1), the early Paleozoic floras dominated by pteridophytes are replaced by gymnosperm-dominated Mesozoic floras, Volume 75, Number 1 1988 Niklas Vascular Plant Diversification in Fossil Record which in turn are replaced by floras dominated by angiosperms. However, Figure 11 masks the taxonomic restructuring in floras that occurs within each category of reproductive mode. The expan- sion of gymnosperm species in Mesozoic floras occurs at the expense of pteridophytes made up of lineages that did not fall off at equal rates. [n general, the numbers of fern species are much less affected than those of lycopods and sphenopsids. Similarly, gymnosperm lin- eages undergo a more or less continuous in- trataxonomic restructuring. e evolutionary appearance of novel modes of reproduction, for example, seeds and flowers, is often associated with alter- ations in growth habit (Tiffney, 1981; Tiffney & Niklas, 1985). The number of plant fam- ilies characterized by cryptogamic (pterido- phytes) and phanerogamic reproduction (gymnosperms and angiosperms), as well as the number of families with principally non- arborescent and arborescent growth forms, are plotted through time in Figure 12. In this figure a number of “cross-overs”” are seen which can be related to changes in taxonomic composition. For example, during the Car- boniferous and Permian, arborescence rises even though cryptogamic reproduction re- mains more common. Although seed plant species dominate Mesozoic floras, nonarbo- rescence is more common than arborescence during the Jurassic. During the Cretaceous, the number of phanerogamic plant families gradually increases, until by the Late Cre- taceous, arborescent/phanerogamic-domi- nated floras which persist through the Ter- tiary are established. The Permo-Triassic is a period of considerable restructuring in both the principal modes of reproduction and growth habits. Before the Triassic, arbores- cent cryptogamic families are dominant, while during the Triassic nonarborescent phanero- gamic families are common. From a much more comprehensive anal- ysis, Tiffney & Niklas (1985) concluded that the history of clonality in land plants can be segregated into three stages: (1) Silurian to lower Carboniferous during which clonal lin- eages dominated, but in which arborescence appeared as a vegetative correlate with het- erospory or the seed habit; (2) Permo-Triassic to Cretaceous, in which families of arbores- cent gymnosperms gradually increased and gained numerical dominance over families of arborescent and nonarborescent pterido- phytes; and (3) Cretaceous to present, which marks the combination of rhizomatous growth and the seed habit (herbaceous angiosperms) which became increasingly more important in the later Tertiary and Quaternary. Unfortunately these analyses are based on familial data (whole-plant reconstructions are too few to determine large-scale patterns in the evolution of tracheophyte growth habits). Accordingly they provide few insights into species patterns of growth habit or values of relative abundance. Although nonarbores- cent families dominate much of the early and middle Mesozoic, the abundance of arbores- cent gymnosperm species in these floras is much higher than that of pteridophytes (Fig. 10). Clearly the family-diversity plots shown in Figure 12 are not reflective of community structure. As Schopf et al. (1975) pointed out, it is possible that paleontologists fail to recognize the true diversity of morphologically simple organisms and overestimate the diversity of morphologically complex organisms. Thus, periods of rapid taxonomic diversification may be inherently related to (and possibly the product of) episodes of rapid morphological diversification. New vegetative and reproduc- tive features provide the potential to discrim- inate new phenetic taxa (= paleospecies). As the number of potential taxonomic characters increases, the number of possible permuta- tions of characters increases exponentially. There is little agreement among specialists as to which feature(s) (anatomy, morphology, reproduction, or even geologic age) contrib- ute(s) to identification of a new species. Cer- tainly among disparate taxonomic groups of plants, species are recognized on often very divergent categories of features: more derived taxa, such as the angiosperms, have more numerous and potentially more complex fea- 48 Annals of the Missouri Botanical Garden 421 408 360 286 248 213 144 65 2 1007 | I l + CAYTONIALES PROBLE- SOFMATICA / 7 n CYCADEOID < 80 = 5 ANGIOSPERMS -J u. 70 p = _ CONIFERS z 60 a o E 50 ln 4 FP = u 40 - o FERNS a a 30 4 uj = PTERIDO- = 20r SPERMS u ee o E a lor re a Kupu wag aaa rim C T T qT Y T y T T i T Li T Y T U L M UIT V N W S|L M UILM U L M 'u L A-A U M|Pa Eo Ol MiP S D C P J T I 421 408 360 286 248 213 144 65 2 FicurE 10. Percent taxonomic composi ure redrawn from Knoll, 1986; courtesy of A. H. Kno Mice by Niklas et al. (1985) for overall patterns in Ea Early Paleozoic floras were dominated by ar trimerophytes) which radiated into various D^ more : den the perce Tertiary. tures with which to identify species (e.g., floral structure) than more archaic taxa, such as pteridophytes (e.g., stelar anatomy, spore-wall characters). he Devonian flora provides a convenient illustration of the potential relationship be- tween taxonomic and morphological rates of evolution. There is considerable agreement among specialists as to the features that dis- tinguish Devonian genera and higher taxa. Additionally, there are authoritative treat- ments of the stratigraphic occurrences of taxa and the first and last occurrences of various vegetative and reproductive features upon which they are based. For example, Chaloner & Sheerin (1979) provided a comprehensive stratigraphic treatment of Devonian genera as well as the first and last occurrences of tion of tracheophytes in iid assemblages. through the Phanerozoic oll). es show simil presente vascular land plants. i nt taxonomic representation of nonflowering groups declines abruptly during the Late Cretaceous and various reproductive and vegetative features. Their data are plotted in Figures 13 and 14. (Nonvascular genera, such as Sporogonites, Parka, Pachytheca, or those having dubious status as tracheophytes, such as Taeniocra- da, are excluded from these plots.) Anatom- ical (tracheid, stelar, and stomatal type; Chal- oner & Sheerin, 1979, figs. 2, 3) features are plotted separately from reproductive fea- tures, such as the position, shape, and type of dehiscence of sporangia (Chaloner & Sheerin, 1979, fig. 4). The data indicate that the number of Devonian genera increases from the Pridolian to the Givetian and then under- goes a modest decline in the Frasnian and Famennian (Fig. 13). By contrast, the number of vegetative and reproductive features upon which Devonian taxa are based increases Volume 75, Number 1 1988 Niklas 49 Vascular Plant Diversification in Fossil Record 42| 408 360 286 248 213 144 65 2 100 - —— m pat | UNKNOWN |/ 90 o BIOTICALLY < L POLLINATED 4 = 89 WIND POLLINATED 3 ANGIOSPERMS u GYMNOSPERMS 70 " = 2 60r 5 2 a 50L IW 4 = = W 40h ER - ul Peso a 0 30r "FREE-SPORING" PL m a PTERIDOPHYTES E 20r " 2 LJ o ac unb DN S uJ a (0) I I T I U 1 | T TT Li T U T T T T Y U| L MU|TV N W S| L MUILMU L M U L A-A U MjPa Eo Ot Mi PL S D C P R J K T 42| 408 360 286 248 213 144 65 2 FIGURE 11. Percent taxonomic composition of tracheophytes (shown in Fig. 10) converted to represent major categories of sexual reproduction. Free-sporing pteridophytes dominate most of the early Paleozoic floras; Oren predominate through most of the Mesozoic; and angiosperms dominate the Late Cretaceous and Ter throughout this interval. Linear regression analysis of the total number of taxonomic characters versus the number of genera yields r = 0.96 (N = 8), which is significant at the 1% level. This correlation, however, reveals very little with regard to the quantitative de- crease in Late Devonian genera despite an increase in the potential number of taxonomic characters in the Frasnian-Famennian. Clear- ly, as the number of genera increases it is reasonable to expect an increase in the num- ber of taxonomically distinguishable features. Analyses of the appearance of new genera and of new vegetative/reproductive features yield poor correlations (Fig. 14). For example, a regression of the number of new genera against that of the number of reproductive features yields r= 0.283. Regressions of either the number of vegetative or the total number of features against the number of Devonian genera yield lower coefficients of correlation. Consequently, the taxonomic rec- ognition of new Devonian genera does not appear to be correlated necessarily with the evolutionary appearance of novel reproduc- tive or vegetative features. In addition, a re- view of the generic descriptions for Devonian vascular plants compiled by Gensel & An- drews (1984) suggests that paleobotanical treatments of early Paleozoic floras are tax- onomically conservative. Therefore, at least at the generic level, it does not appear that estimates of taxonomic diversification in the Devonian are artifacts of rapid morphological evolution. Since most Devonian genera have few species, this conclusion appears warrant- ed at the species level as well (cf. Knoll et al., 1984). Although the perceived taxonomic diver- sification of early vascular land plants appears not to be biased by rapid morphological evo- lution, other episodes of large-scale increases 50 Annals of the Missouri Botanical Garden 421408 360 286 248 215 144 65 e | L[ 1 1 "op e CRYPTOGAMIC REPRODUCTION / 4 PHANEROGAMIC REP. d di40 o NON- ARBORESCENT / 4 30L | 4 ARBORESCENT Í teo o = 25r A00 2 o A z E aA < I 20b " 4 80 '- = HN o. "y O Le < : S PTs Ga P E iuf NM LY eS [des u P ie AAA A—A—A FAAEA / a e00 o-Ofo.. A07 Bu c lO-|* o-ol, Axa Og. "nr d ET m / "020 ,A—A Aa > = O az ? l m —A O-O 2 5 IZ woo i ee 20 A O i A T T < T T T ess: | T T 1 T q U T y T T 1 O U[L MUJ|TV N W S|L MULMU L M U L A-Al| U MjPaEo OL Mi P. S| D C P R J K 421408 360 286 248 213 144 65 2 12. Absolute numbers of spore- and seed-bearing, nonarborescent and arborescent tracheophyte families through the Phanerozoic (re edrawn from Tiffney & Niklas, 1985). The scale of the verti ical axis changes from increments of five families to increments of 20 between the Lower and Upper Cretaceous. in numbers of species have not been rigor- ously examined. For philosophical and prac- tical reasons, therefore, it is reasonable to view conservatively the overall patterns of tracheophyte species diversification in terms of anatomical, morphological, and reproduc- tive evolutionary trends, rather than strictly in the context of patterns of species origi- nation. NULL OR BIOLOGICAL HYPOTHESES? Historically, evolutionary theory has come almost exclusively from observations made on living organisms. Charles Darwin was able to use the fossil record as evidence for evolution but derived his notion of natural selection from insights gained from animal and plant breeding, biogeography, and natural history. Indeed, he found the fossil record singularly intractable in supporting many features of his theory (Rudwick, 1976). The Modern Syn- thesis incorporated paleontology, but even the work of George Gaylord Simpson may be viewed as an ad hoc rationalization of patterns seen in the fossil record based on neontolog- ically derived theory (cf. Gould, 1980). Re- cently, however, paleontology has generated evolutionary hypotheses based on patterns seen in the fossil record. This significant shift in the source of evolutionary speculation has had many effects, not the least of which is a re-evaluation of the biases, artifacts, and lim- itations that are inherent to paleontological data. Clearly, the fossil record can be used to generate evolutionary hypotheses, but only provided it reasonably reflects biological phe- nomena. For a long time paleontologists have recognized the numerous geological factors that contrive to filter and distort biological processes preserved in the record. Much of the recent literature focuses on attempts to Volume 75, Number 1 1988 Niklas 51 Vascular Plant Diversification in Fossil Record [ [ [ | I I 35 ==- NUMBER OF GENERA a - O NO. OF REPRODUCTIVE FEATURES m A NO. OF VEGETATIVE FT et SOr mz0+4 IN 7 Y L Ya B, n A á Ya 25 p B Pad ~ e ^ A n pa s——a~ w20F / >. 4 m Y a” ES a-t lor Ke N " A Lo 0 Y ` ` a y. » LT 5r m7 AU - a8 s PRI GED SIEG EMS EIF GIV FRAS FAM FIGURE 13. Changes in the total numb ge Poata, Sporogonites, Spongiophyton, Protosalvinia): P P ms = Emsian, Eif = Eifelian, Giv = Giv deal with these factors and to reconstruct information lost from the fossil record (Nich- ols & Pollock, 1983). The fossil record is the principal source of information on long-term patterns of evolu- tion. Retained within it are imperfect records of taxonomic diversification, major episodes of adaptive radiations, and major extinctions. This paper has focused on the quality of these patterns for vascular land plants. Quantitative analyses of the paleobotanical literature pro- vide a basis for reconstructing the broad pat- terns of floristic and vegetational change oc- curring over the last 400 million years. In particular, large-scale patterns in numbers of species have been used to reconstruct and identify two major floristic changes, one at the end of the Paleozoic and another toward the end of the Mesozoic; within-assemblage taxonomic compositions have been used to treat broad patterns in vascular plant ecology; and morphological/anatomical data have been used to reconstruct trends in organography and adaptation. As has been seen, however, er of Devonian vascular plant gen a biously vascular plants a and the number of d m Cha loner & Sheerin (1979, figs. 1-4). D (e.g., Parka, Pachytheca, Leia ng = Pridolian, Ged = Gedinnian, Sieg = Siegenian, etian, Fras = Pa Fam = Famennian. potentially serious limitations and distortions exist in the data. In almost all cases, quan- titative analyses can be used comfortably only to draw qualitative conclusions. The most serious difficulty with the fossil record comes from attempts to infer mech- anisms from patterns. Ecologists are currently debating the use of patterns as data, as the recent furor over the use of null models attests (Harvey et al., 1983). Is there a nul pothesis for the pattern of species diversifi- cation seen in the fossil record? The answer is an equivocal yes. The fossil record of land plants exists because of processes of non- marine clastic and pyroclastic deposition. The vast majority of fossil plants are preserved in lowland flood plain or lakeside environments or are entombed in volcanic ashfalls and mud- flows. Consequently, it is conceivable and even probable that much of the patterns seen in tracheophyte species diversification can be explained in terms of factors that influence patterns of nonmarine sedimentation. Prin- cipal among these is tectonics, which controls 52 Annals Mi of the issouri Botanical Garden 15 PS Sm 10} ihe, f Bi m : , ` ` a = a e uj UJ 7 * a o 4 z z d . 2 y 5r z FIRST APPEARANCE OF: 7 = = * =-=- DEVONIAN GENERA , / ^ VEGETATIVE FEATURES aco 17 O REPRODUCTIVE FTS. o iol — TOTAL OF 4+0 u. uj or er u < e iii EA TE 5 ra N ~N 7| = > o A — — = 6 ° NARA "d O o I PRI GED SIEG EMS EIF GIV FRAS. FAM FIGURE 14. Data fam Chaloner & Sheerin (1979, rates of uplift, erosion, subsidence of sedi- ment-accumulating basins, and volcanism (Blatt et al., 1972 For example, variation in the type of non- marine sediments deposited during a geologic period could contribute to the apparent pat- tern of species diversification. Are periods of high numbers of species also those in which geological factors favored the deposition of sediments in which fossils are easily re- covered? Fossil plants are preserved most often and with high morphological fidelity in fine-grained carbonaceous detrital sediments and volcanic ash deposits. Niklas et al. (1980) attempted a limited analysis of data on coal resources to estimate coarsely the variation in nonmarine carbonaceous sediments through the Phanerozoic. We concluded that the Car- boniferous, Cretaceous, and Tertiary were qualitatively different from other geologic pe- riods. These three “coal eras" have large coal tonnages per unit of outcrop area. High num- bers of species for these three coal-rich eras could reflect a combination of extensive sedi- mentological **sampling" of terrestrial floras, excellent preservation (e.g., coal ball petrifac- tions), and the consequence of intensive eco- Numbers of first appearances 2 Devonian genera, and vegetative and reproductive features. figs. 2-4 nomic exploration of coal resources. These factors probably account for the unusually high numbers of species reported for the Car- boniferous and Cretaceous-Tertiary. Therefore, it is safe to assume that tec- tonics and erosional patterns of deposition have contributed to the fluctuation in numbers of species in substantial ways. Nonetheless, these nonbiological or null hypotheses are inadequate to explain the patterns of taxo- nomic turnover within fossil assemblages or broad patterns in plant organographic /repro- ductive evolution. Regardless of the quantity of sediment deposited or the preservational status of plant parts recorded for a geologic period, there exists no necessary and sufh- cient correlation between the magnitude and direction of physical factors operating in the fossil record and patterns in taxonomic turn- over and morphological evolution. Patterns in plant fossil record can be viewed either from the perspective of “objects” (taxa and their origination, persistence, and extinction) or "properties" (morphological and repro- ductive innovations and elaborations). A strictly exclusive treatment of either per- spective is unlikely to lead to any insights into Volume 75, Number 1 1988 Niklas 53 Vascular Plant Diversification in Fossil Record the relationship between evolutionary pat- terns and mechanisms (Sober, 1985). Null hypotheses are an essential component to this type of inquiry, but the assessment of alter- native hypotheses and multiple-causation in paleontology requires both geological and bi- ological insights. e salient conclusions that emerge from a review of the plant fossil record can be briefly summarized under five points. (1) The concepts of a paleospecies and an extant species are significantly different (see Gin- gerich, 1985). (2) Consequently, the fossil record of species diversification is best viewed as a document of trends in morphological/ reproductive diversification. Quantitative es- timates of “diversity” can be used to draw primarily qualitative conclusions. (3) The taxonomic richness recorded for any geologic period does not provide information on species equitability, which is an essential component to considerations of paleoecology and the ramifications of phenomena such as extinc- tion. (4) Although the properties of taxa above the level of species are potentially interesting, they do not necessarily reflect those of species, which are the primary focus of evolutionary mechanisms. Finally, (5) much of the pattern of taxonomic diversification could be ex- plained by factors operating in a strictly geo- logical context; however, identification of long-term patterns in biological phenomena requires the examination of trends in mor- phology, anatomy, and reproductive systems, in conjunction with patterns of species diver- sification (Raup, 1983). LITERATURE CITED Benton, M. J. 1985. Patterns in the diversification of esozoic non-marine tetrapods and problems in his- torical diversity analysis. Special Pap. Palaeontol. 33: 185-202. BLarr, H., G. MIDDLETON & R. Murray. 1972. Origin of d cepa Rocks. Prentice-Hall, Englewood Cliffs, New Jer Bov REAU, E. (edito n 1970. Traite de desig ai Y V (authored by H. N. Andrews, C. sold, E. e Doubinger ^s S. uius T son et Cr, Cann, M. & J. L. [M 1976. The biology of the leaf mark polymorphism in Trifolium repens Distribution of phenotypes on a local scale. Heredity 37: 309-325. CHALONER, W. G. & A. SHEERIN. 1979. Devonian mac- rofloras. r d Palaeontol. 23: 145-161. CLAYTON, W. D. Some aspects of the genus SE Kew nu 27: 281-287. —— 19 The logarithmic distribution of angio- sperm families. Kew Bull. 29: 271-279. CRoNQUIST, A. 1968. The Evolution and a of Flowering Plants. Houghton Mifflin Co. 81. An Integrated System of Classification of Flowering Plants. Columbia Univ. Press, New York. DiMicHELE, W. A., T. L. PhiLLiPS & R. A. PEPPERS. 198 The ¡nens of climate and depositional B. H. Tiffney (editor), Geological Factors and the Evolution of Plants. Yale Univ. Press, New Haven, Connecticut. FisHer, D. C. 1985. Evolutionary morphology: beyond the analogous, the anecdotal, and the ad hoc. Pa- leobiology 11: 120-138. GENSEL, P. G. & H. N. ANDREWS. 1984. Plant Life in the Devonian. Praeger Press, New Yor GINGERICH, P. D. 19 Species in the fossil record: concepts, trends, and transitions. Paleobiology 11: 27-41. GivNisH, T. J. 1986. Biomechanical constraints on crown eometry in forest herbs. Pp. 525-584 in T. J. Givnish (editor), On the Economy of Plant Form and Function. Cambridge Univ. Press, Cambridge. Goutb, S. J. C. G. Simpson, paleontology and the modern synthesis. Pp. 152-172 in E. Mayr & W. = Provins (editors), The Evolutionary Synthe- on the Unification of Biology. Har- e vut Press, Cambridge, Massachusetts. HarLanD, W. B., A. V. ind P. G. LL G. Pickton, A. G. SM A p Time Scale. Cambridge Univ. Press, Cam- brid HARPER, + L. 977. "i Biology of Plants. Ac- a a TN ien —. Modules, branches, sons 2 €— = resources, o. Po. -3 . C. . W. & R. E. Cook a laos Biology and Dre lution of Clonal Organisms. Yale Univ. Press, New Haven, Connecticut. A. D. BELL. 1979. The population dynamics of d form in organisms with modular construc- 29-32 in R. M. Anderson, B. D. Turner tion. Pp. & L .H. COLWELL, J. W. SILVERTOWN & 1983. Null -w in ecology. Ann. . Ecol. Syst. 14: 189-21 Horus. A. 1986. Neutral Í of Phanerozoic di- versification: implications s ee . N. Jb. Geol. Palaont., Abh. 172 44. KIDWELL, S. M. 86. Able ^is fossil concentrations: paleobiologic implications. Paleoecology 12: 6-24 KITCHE LL, J. A. & D. PENA. 19 Periodicity of ex- 4. Patterns of extinction in the fossil vascular plants. Pp. 21-68 ¿n M. Nitecki eiie enema Univ. Chicago Fren, Chicago, Illino 54 Annals of the Missouri Botanical Garden nities through geological time. Pp. 126-141 in J. Diamond & T. Case (editors), Community Ecology. Harper and Row, New York. J. NikLas. 1987. Adaptation, plant evo- lution, re a fossil record. Rev. Palaeobot. Palynol. 50: 127 W. Roruw ELL. 1981. spectives in 1980. dope d T: , K. KLAS, NSEL & 5 "Hu TIFFNEY. 1984. Character b ae and patterns of evo- lution in early vascular plants. Paleobiology 10: 34 47 aaa per- KRASSILOV, V. A. 1975. Paleoecology of Terrestrial Plants. John Wiley and Sons, New York. LLovp, M. & R. J. GHELARDI. 1964. A table for cal- culating the * “equitability” component of species di- 1983. Estimating taxonomic diversity, extinction rates, and speciation rates from fossil data w capture-recapture models. bi 9:1 3. NikLas, K. J. 1987. en scale changes in Aoa and plant terrestrial communities. Pp. 383-404 in D. Jablon . KNoLL. 1980. Ap- Pe iren in ae diversity of fossil plants: a preliminary assessment. Pp. 1-84 in M. K. Hecht, W. C. Steere & B. Wallace (editors), Evolutionary Biology, Volume 12. Plenum Press, New York. & 85 : Pio in vas- cular land plant diversification: an analysis at the species level. Pp. n J. W. Valentine (ed- itor), Phanerozoic Diversity Pu. Princeton Univ. Press, Princeton, New Jersey. Panian, K. € W. A. CLEMENS. 1985. Terrestrial ver- tebrate PRU episodes and insights. Pp. 41-96 in J. W. Valentine (editor), Phanerozoic Dun da Princeton Univ. Press, Princeton, New Jer- EN YG. M. 1985. Biases in the durations and di- versities of fossil taxa. Paleobiology 11: 272-292. PuiLüPs, T. L. 1 volution and vegetative mor- phology in coenopterid ferns. Garden ah 461. Ann. Missouri Bot. . DIMICHELE. 1981. Paleoecology of Middle RA sasana age coal swamps in southern Illinois — Herrin Coal Member of Sahara Mine No. 6. Pp. 231-284 in K. J. Niklas (editor), Paleobotany, Paleoecology, and Evolution. Praeger Press, New or iiec E. C. 1977. Mathematical Ecology, 2nd edi- John Wiley and Sons, ork. 76. Species diversity in d. Phanerozoic: an interpretation. Paleobio 9-2 lases in the fossil record of P cu and genera. Bull. Carnegie Mus. Nat. Hist. 13: 85- 91. . On the pu pns of major biological m d 5. J. J. SEPKOSKI, jx. 82. Mass extinctions in the marine fossil record. Science 215: 1501- 1503 1986. Patterns of change in plant commu- Periodicity of extinctions — Y ——. 1 in the geological past. Proc. Natl. Acad. U.S.A. 81: 801-805. & Periodic extinction of fam- ilies and gone. Suena; 231: 833-836. B. NLEY. 1971. Principles of Pa- leontology. W. H. Freeman and Company, San Fran- cisco, California. Raven, J. A. 1986. Evolution of plant life forms. Pp. 421-492 in T. J. Givnish (editor), On the Economy of Plant Form and Function. Cambridge Univ. Press, Se RUDWICK J. S. The Meaning of Fossils. "dr History Publications, New York. Schorr, T. J. M. , S. J. Gourp & D. S. ; possis versus morphological : influence of morphologic com- plex ol 1: 63-70. a y J., Jn. 1980. The three great evolutionary faunas a E deed d Puer oe Geol. Soc. Amer. Abs. with Pro "M A fe ir de dua of the Pisses marine fossil record. Paleobiology 7: 3. : rr The uniqueness of the Cambrian fau- na. Pp. 203-207 in M. E. Taylor (editor), Short Papers nu d Second International Symposium on the porn System. U.S. Geol. Surv. Open-File iia 81- ^ kinetic model of Phanerozoic taxo- omic diversity: III. Post Paleozoic families and mass extinctions. Paleobiology 10: 246-267. , R. BomBacn, D. M. Raur & J. W. VAL- ENTINE. 1981. Phanerozoic marine diversity and the fossil record. Nature (London) 293: 435-437. Sicvon, P. W. & J. W. Lipps. 1982. Sampling bias, gradual extinction patterns and catastrophes in the fossil record. Geol. Soc. Amer., Special Pap. 190: 1-296. SoBER, E. 1985. The Nature of Selection. MIT Press, ambridge, Massachusetts. Shops: R. A. € C. R. Hitt. 1979. Principal compo- nents and correspondence analyses of quantitative data from a Jurassic plant bed. Rev. Palaeobot. Paly- nol. 28: 273-299, STEWART, W. N. 1983. Paleobotany and the EUN of Plants. um Univ. Press, Cambridg TirrNEv, B. H. Diversity and major RUN in the evolution E land plants. Pp. 193-280 in K. J. Niklas (editor), Paleobotany, Paleoecology, and cid lution, Volume II. Praeger Publishers, New K. J. NiKLAs. 1985. Clonal growth in m plants: a paleobotanical p p. 35-66 in J. B. ackson, . Co is pe Population ise and Evolusiori of Clonal Organ- w Hav isms. Yale Univ. Press, aven, Connecticut. us a 1979. "The plant as a metapopulation. Ann. col. m 10: 109-145. War TAKER, R. 977. uus of species E in land communities. Pp. . K. Hecht, W C. Steere, B. Wa llace (editors) “Evolutionary. Biology, Volume 10. Plenum Press, New York. EFFECTS OF ARIDITY ON Mary T. Kalin Arroyo," PLANT DIVERSITY IN THE irc Sides Just. Amen” and Carolina Villagrán NORTHERN CHILEAN ANDES: RESULTS OF A NATURAL EXPERIMENT" ABSTRACT Hyperarid climates in western South America from 15°S to 29°S, extending up to 3,000 m in the northern Chilean Andes, result primarily from the Andes intercepting precipitation from the Intertropical Convergence during which i cie is thought to have increased from east to west, as opposed to west to east, as occurs north of 25% today. For t les remaining intact vegetation belts (desert scrub, Andean, and high Andean) by hriday on à d scale. Although diversity (measured as a synthetic characteristic e areas of highest rainfall, annual herbs gain greatest prominence in areas d intermediate aridity, while woody species were propottionately a "d strongly represented under extreme dry The wo ood y habit is generally more common in the northern Andes eee in some eh North. pete plant also seen in many species-rich, climatically benign tropical plant communities. The hypo othesis is developed that harsh arid environments of the northern Chilean Andes and in species-rich tropical communities, respectively. We predict that additional life-history trait similarities (e.g., in reeding systems) will emerge for the plant species of abiotically and biotically “harsh” environments, respectively. 1 Research supported by FONDECYT, Chile, and DIB, Universidad de Chile, Santiago, and a John Simon a help with plant identifications, as are many students who helped in fieldwork. We are especially grateful to the Ministerio de Obras Publicas (MOP), Santiago, for supplying weather data and to colleagues at the Missouri Bota espa €— for attending to Ya i-a ic reque * Laboratorio de Sistemática y Ecología Vegetal, eee de Biologia, Facultad de Ciencias, Universidad de Chile, Casilla 653, Santiago, Chile 3 Research Associate, Missou "i Dainiai Garden, St. Louis, Missouri 63166, U.S.A. * Departamento de Biologia, Facultad de Ciencias, Universidad de La Serena, La Serena, Chile. ANN. MISSOURI Bor. GARD. 75: 55-78. 1988. Annals of the Missouri Botanical Garden Areas of the earth's surface where local mountain ranges exert strong influence over precipitation and temperature patterns ex- perience especially radical environmental al- terations during major global climatic changes. How floras adjust to rapidly emerging harsh environmental conditions, and the patterns of species richness and community diversity re- sulting from such restructuring should be just as relevant for the development of compre- hensive diversity theory as phenomena seen in highly productive, abiotically benign en- vironments. Diversity trends in harsh habitats should also have direct bearing for conser- vation in that changes induced in ecosystems by human activities are frequently similar in nature and magnitude to those engendered by natural climatic change. Yet, as perusal of the literature on species diversity and com- munity structure (e.g., Tilman, 1986) will show, harsh habitats have received relatively little attention. Apriorily, reductions in species richness are expected in harsh habitats because produc- tivity is limited by abiotic factors. Such losses, moreover, might be exponential due to the compound direct (physiological) and indirect (historical and, on other organisms such as pollinators, seed dispersal agents) effects of habitat harshness. Patterns of species loss in harsh habitats, nevertheless, are likely to be far more complex than this. Expected losses could be mitigated by a number of factors related to community dynamics. During their evolutionary histories, floras accumulate many life forms varying in ecophysiological and de- mographic properties. At the onset of harsh conditions, loss of richness could be sup- pressed by life-form shuffling. In many warm deserts, for example, the annual habit is se- lectively favored (Raven & Axelrod, 1978; Pavlik, 1985), and species richness levels may be relatively high. Much present knowledge of life form shifts derives from comparative studies of distinct plant communities with flo- ras of heterogeneous phytogeographic origin. Because such comparisons could be con founded by phylogenetic constraints in certain taxa, a critical assessment as to whether life form shifts stall losses in diversity will be best obtained by comparing life forms in floristi- cally homogeneous communities subject in- ternally to different degrees of habitat harsh- ness. To facilitate the interpretation of results, moreover, such gradients in habitat harshness must be well documented as to their abiotic characteristics. aintenance of diversity in harsh environ- ments should also be affected by changes in the relative balance of biotic and abiotic se- lective factors. In early successional com- munities, because of low dominance levels, local diversity may be high despite total low numbers of species present (Houssard et al., 1980). In that resources are severely limited, and competition for light is reduced due to abiotically induced low productivity, harsh habitats may be likened to the early stages of primary succession maintained on a long- term basis. For harsh habitats with low species richness, then, diversity as reflected in the relative abundance of species might be rela- tively high. Another way of viewing this postu- lated analogy is that loss of diversity with greater harshness should be partially, and in- creasingly, counteracted as harshness in- creases. This hypothesis may be tested by comparing species numbers (species richness) with measurements of diversity as a synthetic characteristic combining richness and abun- dance, following Whittaker (1972). e objectives of this paper are, first, to assess how plant species richness, life forms, and community diversity patterns are affected by severe aridity, and, second, to discuss some implications of diversity patterns in harsh hab- itats for community structure in general. The northern Chilean Andes located at 18°-28°S in western South America provide a unique setting for this. Close to 10 degrees latitude of absolute desert at low elevations gives way to a sequence of three high-elevation vege- tation belts (desert scrub, Andean, and high Andean), these exhibiting varying degrees of aridity latitudinally and altitudinally. Because aridity gradients are overlain by temperature variation, the Andean system is also ideal for assessing the relative effects of temperature versus aridity on plant diversity. As Diamond (1986) pointed out, to inter- Volume 75, Number 1 1988 Arroyo et al. 57 Aridity/Plant Diversity in Northern Chilean Andes pret the results of any “natural experiment," as the above would be classed, clear under- standing of the timing of historical events that led to the ““observed results" is essential. As we mentioned above, the identification of sa- lient present-day abiotic parameters respon- sible for maintaining the “‘particular species mix in the test tube" is equally important if unequivocal answers to the kinds of theoret- ical questions outlined earlier are sought. Con- sequently, before examining plant diversity patterns in the northern Andes (Section III), we will devote considerable space to outlining the present climatic characteristics of the northern Andes (Section I), and the historical development of arid climates at subtropical latitudes in western South America in general will be reviewed in detail (Section II). DaTA BASE AND METHODS For species richness patterns and estimates of diversity we compiled data over a number of years in six altitudinal transects (18°S, 19%, 21°S, 24°S, 26°S, 28°S) running from the upper margin of the Atacama Desert (1,500- 3,000 m elevation) to the upper limit for vascular plants (4,500-5,000 m, depending on latitude) (Fig. 1). Records of species pres- ent every 50 or 100 m of elevation were obtained by walking the transects and by climbing a number of high summits in each area. The transects followed the main Andean penetration routes and for this reason had no fixed courses. In each case routes taken tra- versed approximately one-fourth of a degree of latitude. For these transects plant cover was measured in altogether 1,620 minimum area quadrats. Replicate quadrats were sam- pled at 50-100 m elevational intervals along each transect. Cover of shrub and perennial herb species was estimated from the surface area projected by each individual of a species within a quadrat. The largest and smallest diameter of the individual's crown was mea- sured for calculation of an elliptic to circular area. Annual herb cover was initially mea- sured on a phytosociological scale (transects 18°S and 199?S). In later work annual herb cover was estimated from the percentage of points 10 cm apart on line transects inter- cepted. The data for all 50-100 m elevational sampling intervals was subsequently pooled for 500-1,000 m elevational intervals. Species richness for equivalent 500 m el- evational intervals at different latitudes was compared with mean annual precipitation us- ing regression analysis, employing linear, semilogarithmic and log-log models, and with mean annual precipitation and mean annual temperature using multiple regression anal- ysis. Precipitation and temperature for the series of 500 m elevational levels on each transect were estimated from curves con- structed from weather station data available for the particular area under consideration. Weather data were obtained from di Castri & Hajek (1976) and from records supplied by the Ministerio de Obras Publicas (MOP) in Santiago. For life form analyses, species were classed as annual herbs, perennial herbs, and woody plants (primarily shrubs). For the small num- ber of Cactaceae present, columnar species and large cushion species were included in the woody category. The smaller cacti species were categorized as perennial herbs. To fa- cilitate statistical analysis (G-tests) the life form data for pairs of adjacent transects were combined. The indices exp H' where H' = - pin p, and 1/A where A = p? (Peet, 1974), were employed to measure community diversity. In these indices p, = the proportional abun- dance of the ith species. Relative cover was used as the proportional abundance of a species. Details on some of these transects appear in Arroyo et al. (1982, 1984) and in Villagran et al. (1982, 1983). I: PRESENT-DAY CLIMATES IN THE NORTHERN CHILEAN ANDES Intensely arid climates in western South America extend from 15°S in southern Peru to around 29°S in Chile. True absolute desert (without vegetation except along main water courses as the Rio Lluta, Rio Azapa, and Rio Annals of the Missouri Botanical Garden 68? T ü ] $ O Ç Cariquim D , l M ES 9 =: V ° Ç 3 2100-4700m P © cl 7 IQUIQUE : a XS o / 9 2 Ollagüe 2. 3 \ e 7 AES \ Pn c yj "m E c» > 2900 4800m SY E j > S.Pedro Á At ` o A m E ar 4 D I J p I o 9 : / ANTOFAGASTA O / — | nn m é | 8; d ANTARTICO Ñ p i n o 9 Socompa o o Y ^ L - x ,2800-4600m i sl L. = ⁄ A E — TALTAL = I n O / 9^ . m 26*L < / 8 Y - E cHANARALI bg 2 NP M J Maricunga | 4 ü 3 © 2000- 4800m ~ = COPIAPO q ( z / 28°F i r LJ Pd á VALLENAR i O e Conay L ) ° È O 50 100 km bé 4300m i ) 30°C 1 L1 1 70° 68° Volume 75, Number 1 1988 Arroyo et al. 59 Aridity/Plant Diversity in Northern Chilean Andes Loa, or in coastal fog pockets such as Paposo and Taltal) occurs south of 17°S to 25°S (Fig. 1). The desert rises abruptly from a narrow coastal strip to 1,500 m at the Arica deflec- tion (18°S) in northern Chile and reaches a maximum elevation of 3,000 m at 24?-25*S, in from the coastal cities of Antofagasta and Taltal. South of 25°S aridity decreases again. The Chilean—Peruvian arid diagonal is a "rain shadow” and a “cold air” desert (Rauh, 1983). The present climate (Figs. 2, 3) is determined primarily by the annual behavior of the Intertropical Convergence situated over equatorial latitudes (Gomez & Little, 1981) which brings moisture from the northeast, and by a Polar front bringing precipitation from the southwest, together with the interplay of these precipitation sources and the major ocean currents. South of 24°-25°S, most pre- cipitation is received during the winter months (May-August; "invierno chileno") from a northward extension of the Polar front. Con- sequently, the climate is essentially an arid version of the true Mediterranean climate further south at 30°-38°S. Here there is no east-west reduction in rainfall (Fig. 2). Most precipitation above 3,000 m in the Chilean Andes for these latitudes is in the form of snow. North of 24°S, where a southward exten- sion of the Intertropical Convergence during the summer months comes into play and the Polar front is ' negligible, precipitation Is re- ceivedd March; ij leno oline! Vade rainfall below 4,000 m or as transient snow and hail above 4,000 m. The winters are cold and dry. At these latitudes the Andes generate a rain shadow by forcing moisture-laden air from the northeast to rise and cool on their eastern slopes. The greatly diminished saturated air masses ascending onto the western side of the Andes undergo adiabatic heating, further re- ducing potential precipitation. As t air masses reach lower elevations E the Pacific, they are further dried by cold surface waters from the Circumantarctic Current car- ried northward by the equatorward-running Humboldt Current (Peru Current) (Zinsmeis- ter, 1978) and by cold bottom water upwelled from the Pacific by the Humboldt Current as it is deflected away from the coast by the Coriolis effect in subequatorial latitudes. As a result of these features and southward weak- ening of the Intertropical Convergence, the Chilean Andes north of 24°-25°S are char- acterized by steep east-west and north-south reductions in precipitation (Fig. 2). Typical of the east-west gradient, recorded annual precipitation for Parinacota (18°S), situated at 4,395 m, is 372 mm. Murmuntane, situated at 3,280 m and less than 100 km to the west, receives only an average of 156 mm of rain- fall annually, while mean annual rainfall for Arica on the coast (29 m) is less than 1 mm Southwards at 21°S, recorded annual precip- itation for Cebollar at 3,730 m is 53 mm, while at 24°S (at Imilac, 3,232 m) recorded annual rainfall is 2.4 mm. In the northern Chilean Andes, as a result of reduced cloudiness due to higher than av- erage adiabatic heating, the normal latitudinal decrease in temperature is essentially absent to at least around 24°S (Fig. 3) (from 25? to 28*S there are too few temperature records for the Andean highlands). This fortuitous situation avails easy temperature control of comparisons of diversity with increasing arid- ity. The average lapse rate above 1,000 m is about 6.5°C per 1,000 m increase in ele- vation. Below 1,000 m, where coastal fog results in abnormal cooling, the lapse rate is lower. Mean annual temperature at Mur- muntane (3,280 m) is 9.7?C. Mean annual temperature at Parinacota (4,395 m) is 2.5°C. — FIGURE 1. denotes upper limit of absolute Hatched areas: freshwater lakes. Location and altitudinal ranges of transects in the northern Chilean Andes. Heavy dashed line desert in the Andean highlands. Stippled areas: major present-day salt lakes. 60 Annals of the Missouri Botanical Garden PRECIPITATION au rs n xs - "" ABSOLUTE DESERT ° L e, LA? L 22 24 26 LATITUDE (*S) FIGURE 2. de Obras Publicas (MOP), Santiago, a Andean highlands between 26? and 28 Il: HISTORICAL DEVELOPMENT OF ARIDITY IN THE NORTHERN CHILEAN ÁNDES Proposed timetables for development of arid climate in western South America span Mio- cene (e.g., Muizon & DeVries, 1985) to Qua- ternary initiation dates (e.g., Ochsenius, 1983a). Ochsenius & Santana (1974), Och- senius (1983b), and Axelrod (1979b) agree that maximum aridity was reached very re- cently. EMERGENCE OF THE ANDES That many of the high Andean summits rose rapidly and only very recently is critical to understanding the development of hars arid climates at subtropical latitudes in west- ern South America. From the late Cretaceous into early Pa- leocene times, volcanic rocks and associated sediments, deposited close to the present con- tinental margin in Mesozoic times as a result of the closure of the Nazca and American plates (Rutland, 1971; Mortimer et al., 1974), were uplifted to form a proto-Andean divide west of the present Cordillera Occidental Mean annual precipitation related to elevation and latitude in northern Chile. Data from Ministerio nd di Castri & Hajek (1976). Precipitation data unavailable for the ç (Mortimer, 1980). This primitive axis, how- ever, was almost entirely eroded away, leav- ing the Altos de Camilica Formation in south- ern Peru (Tosdal et al., 1984) and the Putani Formation east of Arica (Mortimer & Saric, 1972). The modern landscape differentiated 20? rie o eP a 23 dl “~ (59 H23 20 X: 22 O 2204 ra L 22 23 ul = 10° + fe a 22 2! E aie: < sof 247 "e 2! 20 aw 29 A — '? ul e——20 a 21 “18 = ° L W 2 a e —20 E -5° F e— 23 1 1 1 l J o 2000 4000 6000 ELEVATION (m) FIGURE 3. Mean temperature plotted against ele- vation for areas between latitude 18? and 24°S in the northern Chilean Andes. Data sources as for Figure 2. Data point for 5,300 m at 23°S is from Corrida de Cori ie 1977), Argentina, close to the Chilean bor- der. sociated with each datum point indicate the corresponding latitude. Volume 75, Number 1 1988 Arroyo et al. 61 Aridity/Plant Diversity in Northern Chilean Andes initially in the late Paleocene-early Miocene interval. It comprises (a) the nonvolcanic Cor- dillera de la Costa, (b) a low-lying depositional basin referred to as the Pampa de Tamarugal in northern Chile or Llanuras Costaneras in southern Peru, (c) the precordillera, a loosely defined transition zone at 2,000-4,000 m, and (d) the parallel cordilleras Occidental and Oriental with summits close to 7,000 m. These cordilleras are separated by (e) a broad in- tervening high-altitude plain, the Altiplano, of some 500,000 km? (Allmendinger, 1986) at 3,700-4,000 m (Fig. 4). The Cordillera Occidental and the Cordi- llera Oriental emerged relatively late after two major focal points of rhyolitic volcanic activ- ity developed in the late Oligocene and into the early Miocene (Tosdal et al., 1984; Na- ranjo & Paskoff, 1985) following extensive north-south block faulting and differential uplifting (Mortimer & Saric, 1972). Such ac- tivity persisted well into the late Miocene and early Pliocene (Megard et al., 1985), by which time deposition of sediments derived from the surrounding eroding volcanic axes produced the Altiplano (Fig. 4). During the same epoch, lava flows moving westward and sediments from the Cordillera Occidental backed up against the Cordillera de la Costa (Naranjo & Paskoff, 1985), elevating the northern part of the Pampa de Tamarugal in Chile and the Llanuras Costaneras in southern Peru. The major increase in height in the Andes, nevertheless, occurred only as of the middle Miocene and onward (Mortimer et al., 1974) in north Chile and the Pliocene in southern Peru (Tosdal et al., 1984) as the result of andesitic volcanism (Fig. 4). Andesitic activity continued across the Cordillera Occidental and Cordillera Oriental throughout the Pliocene and Pleistocene into the Holocene, giving rise to the some 800 volcanoes present in north- ern Chile. Over 30 of these exceed 6,000 m elevation. Tosdal et al. (1984) estimated that the southern Peruvian Andes were uplifted 0.06-0.19 mm per year throughout the Neo- gene. The Cordillera Occidental at its south- ern edge is thought to have increased in height by 0.5 mm per year as of Holocene times (Rutland et al., 1965). MIOCENE CLIMATES Although significant uplifting had occurred by the Miocene, there is no evidence at this stage of the strong east Andean rain shadow seen in the Andes today. Axelrod (1979b) reviewed the limited paleobotanical evidence for the Miocene. Berry (1919) described leaf remains from the Tumbez area in northern Peru. Included is material identified as An- nona, Banisteriopsis, Ficus, Persea, and Styrax. Although Berry's identifications re- quire verification, the leaf types present are indicative of a fairly closed tropical viney forest and of productive environments. A second flora studied by Berry (1917, 1939) from Potosi, Bolivia, now above 4,000 m and immediately to the east of the Cordi- llera Occidental, was considered by Ahlfeld (1956) to be of Miocene age and would have thus been deposited during the early phases of the uplifting of the Altiplano. It reportedly contains Calliandra, Cassia, Copaifera, Dalbergia, Escallonia, Passiflora, Termi- nalia, Inga, and Weinmannia. Today similar floras strongly dominated by woody legumi- nous taxa typically occur in neotropical for- mations under high rainfall regimes but usu- ally with a distinct dry season in the Venezuelan llanos and adjacent Orinocan for- ests and areas transitional between Amazonia and the Brazilian Planalto. Elements remi- niscent of Weinmannia, Escallonia, and Inga suggest that a semiseasonal forest gave way to a middle-altitude montane forest similar to that seen today further north in the Andes in Colombia where rainfall is very high. A third flora, from Psillypampa, Bolivia, still further to the east of the main Andean axis, today at 2,600 m, was regarded as Plio- cene age by Berry (1922) but suspected by Axelrod (19792) as possibly of Miocene age. It contains material identified as Heliconia, Myrica, Pisonia, and Pithecellobium and many other genera with fairly small leaves. For the Miocene then, there appears to have been a transition from viney forest on the extreme Pacific border, into semiseasonal vegetation types at mid elevations on the Al- tiplano, and finally into more xeric, small. 62 Annals of the Missouri Botanical Garden w EXTREME ARIDITY HOLOCENE UPPER PLEISTOCENE > ANDEAN AXIS E WARM HUMID GLACIAL COLDER ERY | INTERGLACIAL ü irre GLACIAL| ARID INTERGLACIAL LOWER PLEISTOCENE GLACIAL E EE IE: wane INTERGLACIAL d WARM, PROBABLY SEASONAL LOWER MIOCENE COOLER GLACIAL | DRY INTERGLACIAL WARM, PROBABLY MOSTLY ASEASONAL FIGURE 4. Miocene and climatic tendencies west and east of the northern Andes during the er VERY WARM MAINLY ASEASONAL Stages in the evolution of the Cordillera Occidental (A) and Cordillera Oriental (B) as Ln ernary. Hatched areas rhyolitic volcanic activity. Stippled areas: Atiplano. Black areas: andesitic volcanoes leaved forms at lower elevations on the east- ern side of the Altiplano. This suggests that, in contrast to today, the western side of the Miocene Andes may have been wetter than the eastern side. The expected positions and water temper- atures of currents in the Pacific Ocean sup- port the above interpretation. Prior to the consolidation of the Antarctic Icesheet, the west wind drift is believed to have lain further south than today; moreover, as Zinsmeister (1978) suggested, cold surface waters would have been directed between East and West Antarctica rather than up the South Ameri- can coast. With a less active and considerably warmer Humboldt Current in the Miocene, the counter-running warm current bathing the coasts of Ecuador and Peru today most likely extended south of its present position, very possibly engendering precipitation pat- terns similar to that in recent El Nino events (Cane, 1983), which brought torrential rains to coastal and lowland areas of southern Ec- uador and northern Peru (Rasmusson & Wal- lace, 1983). A warm, wet Miocene for the western flanks of the subtropical Andes finds good support from marine fossils deposited in high coastal cliffs in northern and central Chile during marine transgressions (Mortimer, 1972) and in the related Pisco Formation in the Llanuras Costaneras of southern Peru (Muizon & DeVries, 1985). At 30°S on the coast of Chile, dendrophyllid corals accompanied a warm- water ostracod fauna, of which some sub- tropical elements extended as far south as 47°S (Herm & Paskoff, 1967; Herm, 1969). The Pisco Formation contains turtles, sloths, terrestrial carnivores, and other large-bodied animals that could only have existed under fairly productive environments. Vertical in- cisions up to 1,000 m deep traverse the cen- tral depression in northern Chile (e.g., que- bradas Vitor, Azapa, Lluta, Camarones). These, which geologists agree are indicative of high pluviosity, were initially cut down in the Miocene (see Mortimer, 1973; Paskoff & Naranjo, 1979; Naranjo & Paskoff, 1980a). Finally, the Miocene was the time of maxi- mum copper enrichment in the Chilean Andes Volume 75, Number 1 1988 Arroyo et al. 63 Aridity/Plant Diversity in Northern Chilean Andes (Clark et al., 1967). The high water table that this process requires and an active period of erosion seen in canyon development seem to be indisputable evidence against dry climates in the western deserts at this stage. PLIOCENE CLIMATES The Pliocene is poorly known for arid sub- tropical latitudes in western South America. Marine faunas on the Chilean coast at 30°S (Herm, 1969) and in the Pisco Formation in southern Peru (Muizon & DeVries, 1985) show declines in species richness and an influx of elements from cooler waters at the Mio- cene—Pliocene boundary. Zinsmeister (1978) related such changes to increased incorpo- ration of cold water into the Humboldt Cur- rent due to northward displacement of the west wind drift and reduced flow through Drake Passage. Certain periods in the Plio- cene in the Colombian Andes, situated away from the influence of the Humboldt Current, saw lower tree lines than at present and the first appearance of a high-elevation flora (Hooghiemstra, 1984). Thus changes in ma- rine faunas along the Pacific coast must have been due at least partially to a general global trend toward climatic cooling. Climates were evidently drier than in the Miocene— canyon cutting in the Atacama ceased abruptly at this stage (Mortimer, 1973). Vallea, Borrer- ia, Niphogeton, and Eryngium, genera con- sidered indicative of open conditions, became abundant occasionally in high-elevation Co- lombian forests (Hooghiemstra, 1984). Dur- ing this period large mammals (Equus, Me- gatherium) appeared in the present area of the Atacama desert. This scant information for the Pliocene suggests a gradual transition from the closed Miocene forests into more open, savannalike vegetation at low eleva- tions, with small, evergreen treelets devel- oping at mid elevations. The presently dis- junct montane genus Kageneckia (Rosaceae), which occurs in central Chile and again in eastern Bolivia and southern Peru, could have been present in these Pliocene montane for- ests. Prosopis, the only surviving tree genus at low elevations in the Atacama today, prob- ably dates to lowland Pliocene vegetation. PLEISTOCENE-HOLOCENE CLIMATES The fairly uneventful, drier and cooler, but far from hyperarid Pliocene of the western margin of subtropical South America gave way to a Pleistocene characterized by marked alternating wet and dry periods. For tropical and subtropical lowland areas east of the Andes in South America the gla- cials were cold-dry times of forest contrac- tion, while the interglacials were wet-warm times of forest expansion (Damuth & Fair- bridge, 1970; Colinvaux, 1979; Ab'Saber, 1982; Prance, 1982). It has not been sufh- ciently appreciated that the wind systems pro- posed by Damuth & Fairbridge (1970) to account for dry glacial periods in the Amazon Basin predict precisely the opposite climatic trends for corresponding periods on the west- ern side of the Andes, i.e., warm-dry periods alternating with wet-cold periods (Fig. 5). During the glacials, the Damuth & Fair- bridge model sees a low-pressure focus over Antarctica moving northward, bringing in- creased moisture from a southwestern source to mid- and subtropical latitudes in South America (Fig. 5). Because the cordilleras were now strongly elevated, the destination of much of this precipitation would have been the west- ern side of the central Andes. Geological evi- dence suggests that the Atacaman area indeed experienced very wet climates in the Pleis- tocene. Many of the salt lakes (salares) pres- ently occupying 2,800 km’ bear extensive deposits of Pleistocene lacustrine and diato- maceous earth (Stoertz & Ericksen, 1974; Naranjo & Paskoff, 1980b). Analyses of old shoreline lines (Tricart, 1969) indicate that approximately one-half of the present salares in northern Chile (e.g., salares de San Martin, Ollague, Uyuni, Coipasa) formerly constituted extensive, deep, perennial lake systems (Stoertz & Ericksen, 1974). Wet glacial pe- riods on the western side of the central Andes are also suggested by the fact that the Pleis- tocene snow line was depressed to a greater Annals of the Missouri Botanical Garden TROPICAL s. VA x I WET AND DRY.- VA n FA. | EN l; U 1 7! ./ HUMIO H | SUB- TRIER H TP 3 An Me o 1 (ee Oo H » [a j `x O š q a Ei on be E CURRENTS = 8 L 2 WINDS S JULY —- x - in a u. INTERGLACIAL (WARM) FiGuRE 5. TROPIC CAPRICORN OF GLACIAL ( COLD) Probable glacial and interglacial climates for the present Atacama desert region (area under large circles) in relation to the rest of South America. WD: warm-dry. CH: cold-humid. Modified from Damuth & Fairbridge (1970). extent on the western side of the Andes (Has- tenrath, 1967). The pollen record for the Andes, although still sparse, is also consistent with strong east- west climatic differentiation in the northern Andes during the Pleistocene. Heusser (1983) provided evidence of northward migrations of Nothofagus and Podocarpus in central Chile during glacial phases. This evidence indicates wet-cold climates for the western side of the Andes. On the eastern side of the subtropical Andes, in contrast, drought-tolerant taxa ap- peared on the border of the puna in the Junin area in Peru (Hansen et al., 1984) and close to the Bolivian- Peruvian border (Graf, 1981) during cold periods of the Pleistocene and of the Holocene, respectively. Late glacial as- semblages from eastern Patagonia at 41°S are also consistent with colder and drier condi- tions than today (Markgraf, 1983). Eastern high elevations at Mediterranean latitudes, however, and the extreme edge of the puna in Jujuy (unlike the puna further north in Bolivia and Peru, and the Patagonia further south) were wetter than today during cold periods (Markgraf, in press). Such simulta- neously wetter climates on the western (cf. Heusser, 1983) and eastern sides of the An- des at Mediterranean latitudes (out of the range of the Intertropical Convergence and hence very dry during the interglacials) are not unexpected; the greatly increased wester- lies there relative to interglacial periods would have increased precipitation on the Pacific side of the Andes and augmented that making its way across the Andes. As each interglacial ensued and climates warmed, the southwesterlies would have re- sumed their present position (Fig. 5) with the Intertropical Convergence coming back into play in the northern Andes as seen today. Because the Andes rose significantly during the several wet-cold and warm-dry cycles of the Pleistocene, east-west climatic differen- tiation must have increased as the Pleistocene progressed. As a result, aridity should have intensified at each new interglacial. Throughout the climatically turbulent Pleistocene, nevertheless, the Atacaman re- gion continued to support large mammals, including Mastodon and Macrauchenia (Mares, 1985). That such animals did not Volume 75, Number 1 1988 Arroyo et al. 65 pia, Diversity in Northern Chilean Andes TABLE 1. for zonal and aa (bog) components of each de considered separat mber of gene Figure 1 for exact locations. pun 18"S receives the most precipitation; latitude 24°S Chilean Andes. at each harm See receives the least precipitation. ber of species per genus and n Comparison of the number of species present (species richness) for the total vascular flora and ely at different latitudes in the northern era per family are also given for the total flora Summer Precipitation Winter Precipitation 18°S 19°S 21*S 24°S 26% 28% Total flora Number of species 391 219 164 ran 144 270 Number of genera 195 138 110 55 90 162 Species/genus 2 1.6 1.5 1.4 1.6 1.7 Number of families 64 53 7 30 42 58 Genera/family 3 26 3 1.8 2.1 2.8 Zonal flora Number of species 333 190 141 59 110 200 % of total flora 85.2 86.7 86 76.6 76.4 74.1 Azonal flora (bogs) Number of species 58 29 23 18 34 70 % of total flora 14.8 13.2 14 23.4 23.6 25.9 ' Species/genus and genus/family ratios exclude some Cactaceae of doubtful generic affinity. become extinct until the end of the Pleisto- cene (Ochsenius, 1983a) agrees entirely with Axelrod's (1979b) suggestion that the climate on the western slopes of the northern Chilean Andes reached its present intensely arid state only very recently in Holocene times. III: PATTERN OF DIVERSITY IN THE NORTHERN CHILEAN ANDES SPECIES RICHNESS Excluding a small number of systematic problems that still require attention, the succession of vegetation belts (desert scrub, Andean, high Andean) up from the desert edge to the upper vegetation limit on the six transects (Fig. 1) yielded 769 species in 290 genera of vascular plants. Five hundred twen- ty-one species have been collected on the four northernmost summer-rainfall gradients (Ta- e 1). Here species richness peaks at mid elevation, where conditions are intermediately arid and cold (Fig. 6). The resultant curves tend to be steeper towards the north where the east-west precipitation gradient is most severe. Three hundred twenty-seven species occur on the winter-rainfall transects at 26°S and 28°S; here maximum species richness, in con- trast with the northern transects, occurs at lower elevations. These winter-rainfall tran- sects nevertheless tend to show a mid-ele- vation bulge of their own which reflects the lowermost elevation of permanent winter snow in these areas. Latitudinally, species richness drops off by 80% from 18°S (maximum rainfall from the 140 p o uj o oor uj a: (09) - u O 60} ac ui D L 2 2 = 20r 2000- 3000- 4000- 2500 3500 4500 ELE VATION (M) FIGURE 6. Variation in number of species (species richness) with elevation at siflerent -— in the northern Chilean Andes. ainfall zone summ (solid line); O = winter- iael zone poids line). Annals of the 66 Missouri Botanical Garden 'jugogiuZis jou = SN *S00'0 > d xxx ‘TO'O > d xx ‘SO'O > d 1199172) 10] spars] eouroytusig , 'e[qeordde jou 1591-2) - "uonejidioo1d 10] S6 | pue gI USam]aq oj?rpoutri9jUI st Bare SIY} 1841 jsa33ns (ogeuueg *'seorqng se1q() ap oueijsrurjA) suoneaoprsuoo [eorgo[oipAu “S.8Z-0Z 10} a]qe|reaeun eJep uonejidioadq | (‘dds +£2) (‘dds 96) (-dds gg) (‘dds 51) SN ‘ZS'T x09 P — x««08 11. SN :F8'I WEL LI HOPS €Z %811 OL %C ZI squey [enuuy xx«9 LOI SN ‘ZOO — «««08 PI — «««IU£Ul — 208 P4 SLI %0'0S BF %90S € 956g bl LOI -$q19q [pruua1aq sx«89'CT SN ‘£87 SN'OZ — «««8VIL — 25621 ct %0'9% SZ MOLE Z€ ?5VLl SZ səən x sqniuç wu Qee — uu /6 urut 99c uonejtidioa1d pajeunsy ur 000'S-000'p Š (dds 9pz) (‘dds [c I) (‘dds 6g) (‘dds 917) xxx [6 OI m SN ‘ZSO SNI SN 29€ %9Z1 IE %8'EZ OF %U8Z SZ AT8I 6€ squay [enuuy «666 E SN 220 SN ‘00°€ K£ L `Y M%EES IEI 2⁄0 19 ?5079€ Z€ %S6P LOT squay [eruua1ag SN :90'0 A SN *00'0 SN'.8*'0 SN ‘0€°0 AUPE v8 MBSE PS %O9E ZE BPE OL səən X sqniuç £ WU 6tc — uu og uni ¢GT uoyendioeid pojeurms7] e š ur 000'P-000'€ bD a (‘dds oer) (‘dds 9p 1) (‘dds 91) (‘dds 69) VN = VN SN '0S'0 VN 2591 61 AYOZ €F ?5)0 0T I %9 2 LI squey [enuuy VN VN SN *00'1 VN %9 PP 8S o,U986 IF %00Z ¿Z %%28*€ vs sqiəu [eruua1ag VN VN «SN 58070 ¿YN %80? €S WSZ? c9 %0'0L L 2,9'0t 87 səən X sqnius urur 00 — uu 9 urut go uonejdioa1d pojeurns ur 000'£-000'Z g' ^q Yad gosg D “SA V g “Sa y (d) SovZ-27 (9) So8Z-9Z (A) SohZ-1Z — (V)S6I-8I (suostedulo”) 189 AA ISe) 1891-1) (suosueduro^) usajsa AN) 1891-9 sapuy 413158 sapuy U491sƏ M (2261) zmsymy ay) iof sojpunisg VE 2]qnj ui suoissa14824 əy} u1 posn sjunoum fo sa#psaav 240 spjojf uiəlsəm 40f wsof ofi] usa13 n fo o8p1u2242d ay) 24D0duio2 $1891-) IYJ, ^ (2261) zmswqiny uio4f 040 uoi824 spsoy ay) 07 puodsa110» uo19ə1 UDINYJ ay) 10f vq 'sopup ui24140u 211 fo apis (auquasip) ui21$02 pup (uva py) usajsam ay] uo suuof afr] ui pa110da4 DIDP uols 421102m fo SIFDIIAÐ 240 SDIO/ ULIISDI ay? 40f uoipdioo4d fo saypunjsy ‘paipoipur spsoy fo aid ay} aunuadíp ay) iof vq "| 24n2i] ui SI99SUDA] xis əy} uo Juasasd ACUSAN Volume 75, Number 1 1988 Arroyo et al. 67 Aridity/Plant Diversity in Northern Chilean Andes Intertropical Convergence) to 24°S (minimum rainfall from the Intertropical Convergence) (Table 1); however, recuperation is rapid around 26°S and 28°S in the influence of the winter-rainfall pattern. Species/genus ratios are low in all cases (1.4-2; Table 1), although Senecio (18°S: 26 species), Adesmia (28°S: 13 species), Nototriche and Werneria (18°S: 12 species each), Stipa (18°S: 11 species), (18°S; 7 species each), and Chaetanthera (28°S: 7 species) are speciose. Under both precipitation regimes, as aridity increases, trends for fewer species per genus and fewer genera per family are seen (Table 1). Thus fewer genera and fewer species have survived in the most arid regions, and/or these appear to have undergone less local speciation. Pre- cise records of precipitation for the highlands at are unavailable. Hydrological esti- mates (Ministerio de Obras Publicas, Santia- go) suggest levels roughly similar to those midway between 18? and 19°S. Species rich- ness is lower at 28°S than midway between 18° and 19°S, indicating that the decline in species richness from 18° to 24°S is probably slightly inflated by a small latitudinal effect independent of precipitation. By comparing floras on the east side and west side of the Andes for areas with equiv- alent precipitation, the degree to which his- torical and biogeographical factors contribute to present richness patterns may be assessed. Similar amounts of precipitation are received at 18°-19°S on the Chilean side of the Andes and at 21°-24°S on the Argentinian side at 4,000-5,000 m (Table 2). Nevertheless, around a third more species occur on the eastern side, which experienced less severe fluctuations during the Pleistocene. Such low- er richness levels on the western side of the Andes undoubtedly also reflect reduced north- south migration possibilities there on account of the desert extending above 3,000 m at 24°S (Arroyo et al., 1982; Villagran et al., 1983). Thus, in addition to the direct effect of lowered precipitation on species richness, there seems to be a substantial indirect effect of the evolution of aridity. TABLE 3. (species richness) on mean northern Chilean Andes. Each regression is based on species numbers at 18%, 19%, 21*S, and 24°S. Mean annual temperature for these latitudes is relatively con- stant with elevation (see Fig. 3). y = number of species; x = mm precipitation. Degrees of freedom for all regres- Best fit regressions of number of species annual precipitation in the sions = 2 Elevation Regression F-ratio! 2,000-2,500 m y= -0.17e%%= 23,05*** 2,500-3,000 m y = —0.60x^* 3.78 NS 3,000-3,500m y = 3.09e°!* 4.61* 3,500-4,000 m y = 0.42x + 32.67 18.09*** 4,000-4,500 m y = 2.29x° La 4,500-5,000 m y = 0.12x + 5.03 6.03* ' Significance levels for regressions: * P « 0.05; ** P 0.01; *** P < 0.005; NS = not significant. A more precise appreciation of the com- bined indirect and direct effects of aridity on species richness in the northern Chilean An- des may be obtained by relating species rich- ness to precipitation for areas with similar mean annual temperatures. Most regressions (best fits varied from a linear, through semilog to log-log model) (Table 3) were significant. The degree of curvature where fits were cur- vilinear, however, was always very shallow. The overall regression of species richness on precipitation, combining all elevations, was also highly significant (F = 14.087; d.f. = 22; P — 0.002), emphasizing very high pene- trance of precipitation on species richness on a biogeographical scale. Aridity, moreover, completely obliterates the effects of temper- ature regionally. In the multiple regression of species richness on mean annual precipitation and mean annual temperature as independent variables (y = 0.216x, + 0.017x,; d.f. = 21; P = 0.005, where x, is mean annual precipitation and x, is mean annual temper- ature), the partial regression coefficient for precipitation was highly significant (P = 0.007), whereas that for temperature (nor- mally a strong determinant of species-richness patterns along elevational gradients) was non- significant (P — 0.991). ere are also clear reductions in total plant cover with aridity (Fig. 7). In the most benign of the wettest areas in the northern Andes, total cover does not exceed 50%. 68 Annals of the Missouri Botanical Garden G Š sor a ul 1995 Š S 1895 219s IO r26*s,^ `a p E a 24*S L 1 l l ` ~~ 2000- 3000- 4000 - 2500 3500 4500 ELEVATION (M) FIGURE 7. Variation in plant cover (% of ground vegetated) with elevation at different latitudes in the northern Chilean Andes. @ = summer-rainfall zone (solid line); O = winter-rainfall zone (broken line). However, at 24°S, less than 25% of total sur- face is covered. The regression of plant cover on plant species richness (y = 0.22x + 8.21; F = 12.31; d.f. = 17; P = 0.003) was highly significant. However, as indicated by the regression coefficient (also compare Figs. 6, 7), reduction in plant cover and species rich- ness reduction are not strictly parallel. This suggests that although relatively few species survive on the most arid sectors of the north- ern Andes, some (e.g., Adesmia polyphylla, Stipa frigida, Philippiamra fastigiata) thrive. Species that are rare at latitudes 18°S and 19°S (e.g., Portulaca philippii, Sola- num phyllanthum, Dunalia spinosa, many species of Senecio), on the other hand, are conspicuously absent on the drier 21°S and 24°S transects. Stochastic effects bearing on the smaller populations of such species per- haps have prevented survival in the most arid habitats. This last factor probably contributes to the lower species/genus ratios seen with severe aridity. HIGH ANDEAN BOGS VERSUS ZONAL VEGETATION In the wettest area of the northern Chilean Andes (18°S) above 4,000 m, previous fresh- water lakes dating to the pluvial cycles of the Pleistocene support azonal cushion bogs (bo- fedales). In the driest areas (24?-26*S) such lakes have been reduced to salares with small marginal patches of bog vegetation. These important grazing resources for the Altiplano economy (Castro et al., 1982) contain such typical species as Gentiana sedifolia, Azo- rella trifoliolata, Lachemilla spp., Werneria spp., and Colobanthus quitensis. They are unique to the central and drier sectors of the southern Andes and are unknown in the far northern Andes, where azonal and zonal vege- tation pin tend to intergrade imperceptibly (Cleef, 1980). Despite reduction in area suitable for de- velopment of bog vegetation towards 24°S in the northern Andes and the greater depen- dence of bog elements on fresh water, bog habitats have lost proportionately fewer species than zonal habitats (Table 1). For example, bog species constitute only 14.83% of the total flora at 18°S in comparison with 23.38% at 2458. Relatively lighter losses in species richness in the bog floras possibly result from new introductions repeatedly offsetting losses due to reduced habitat size. Figure 8 plots floristic divergence for zonal and azonal (bog) ele- ments for all possible pairs of the six transects against the geographical distance separating each pair of transects. For vegetation types, floristic divergence increases with dis- tance; however, the bog floras are notably less disparate than the zonal floras for equiv- alent separation distances. Floristic affinity of the bog floras is higher because of lack of local endemic speciation and because the lat- itudinal ranges of bog species along the Andes are wider in comparison with zonal elements (Arroyo et al., 1982). The broader latitudinal ranges of bog species, in turn, may be related to aspects of reproductive biology. Many bog elements are strongly autogamous (e.g., Co- lobanthus quitensis, Gentiana prostrata, Cardamine glacialis), greater reliance on wind-pollination is also evident (Arroyo et al., 1983), and some dominant bog species (e.g., Distichia muscoides, Oxychloe andina) are clearly adapted for bird dispersal. These three features should facilitate long-distance dis- Volume 75, Number 1 1988 Arroyo et al. 69 Aridity/Plant Diversity in Northern Chilean Andes persal between the islandlike high-Andean bogs, thereby maintaining their floras rela- tively homogeneous. They should also pro- mote reintroductions of species lost due to reduced habitat size. LIFE FORMS—INTERANDEAN TRENDS Table 2 compares life-form spectra for the wettest (18°-19°S) and driest (21°-24°S) ex- tremes of the summer-rainfall area (compar- ison A vs. B) and for the summer-rainfall areas vs. the winter-rainfall area (26?-28?S) (com- parisons A vs. C & B vs. C) at three eleva- tional levels. Comparing the wet and arid ex- tremes of the summer-rainfall area, perennial herbs are statistically underrepresented at the arid extreme for 3,000-4,000 m and 4,000- 5,000 m. Because of small numbers of species, the data for 2,000-3,000 m could not be tested statistically; however, a similar trend is evident with relatively fewer perennial herbs occurring at 21°-24°S. The difference is most pronounced in the upper Andean belt (4,000— 5,000 m) and along the desert edge (2,000— 3,000 m). Woody species, in contrast, tend to be more strongly represented as aridity in- creases; the trend is clear in the driest (2,000— 3,000 m) of the three vegetation belts and where aridity is overlain with cold stress (4,000-5,000 m). Contrary to expectations, annual herbs did not increase proportionately with greater aridity. For the mid-elevation belt (3,000-4,000 m) there is a weak, nonsignifi- cant trend for more annual species at 21°- 24°S; however, there were proportionately fewer annuals on the desert edge at 21°-24°S and no significant increase again in the upper Andean belt. The difference was exceedingly dramatic along the desert edge at 21°-24°S, where the flora is comprised of seven species of shrubs, two perennial herbs, and one an- nual (Philippiamra fastigiata). In the most stressful climate (aridity confounded by cold stress; 4,000-5,000 m; 21°-24°S) the rela- tive loss of perennial herbs and annuals has been to the extent that the shrub life form becomes dominant for species number. Clear- 1.0 s | z u; - oO a > a 08 F e E L Y a ° e Zo 3 sel 9 ° Fas n (Bogs) o 2 4 6 8 IO DISTANCE ( LATITUDE) FiGURE8. Floristic dde all pairs of zonal and azonal (bog) floras n 18? and 28°S in the northern Chilean Andes p uh against the latitudinal distance in degrees separating the pairs of floras com- pared. Florist ~~ ah ael a "inn (A, B) = Ny / areas A and B. ly therefore, in the summer-rainfall area in the northern Andes, the perennial herb life form has been most affected by aridity, fol- lowed by the annual herb life form; woody elements have been least affected. For 26*-28%S (winter rainfall), intermedi- ate proportions of annuals relative to 18°- 19° and 21°-24°S were predicted. For the lowermost desert belt (2,000-3,000 m) and mid-elevation belt (3,000-4,000 m), there were no significant differences for any life form when 26°—28°S was compared with 18°- 19°S (highest rainfall). For these latitudinal comparisons, however, annuals are more common and perennial herbs less common in the high-elevation (4,000-5,000 m) belt in the summer-rainfall area. When 26?-28"S is compared with 21°-24°S (lowest rainfall), contrary to expectation, annuals did not prove to be in excess in the 21°-24°S area. For the mid-elevation belt (3,000-4,000 m), as is indicated by the nonsignificant difference, the winter-rainfall area has proportionately just as many annual species as the much drier 21?-24*S summer-rainfall region. For the low- and high-elevation belts there are more an- nuals at 26°—28°S than in the very low rainfall area at 21°-24°S. The higher proportion of annuals at 26°-28°S, where there is winter Annals of the Missouri Botanical Garden snow (and hence much local moisture) in place of rainfall received gradually over the summer months, as at 21?-24*5, again suggests that the annual life form is only adaptive in arid climates up to a certain limit. To test the robustness of these trends (de- crease in perennial life form, increase in an- nual life form up to a certain level of aridity, dominance of shrub habit in areas of extreme aridity) over a wider sector of the Andes, we compared our data with higher rainfall An- dean communities on the eastern side of the Andes (Ruthsatz, 1977) situated immediately opposite the most arid area of the Chilean Andes considered by us (Table 2). Er (1977) gave the life forms of over 90% o the species she considered. For 22°-24°S es and 18°-19°S (west), most similar in precip- itation amounts, there were no significant dif- ferences in life forms for any of the three altitudinal levels (comparison D vs. A; Table 2). From east (22?-24*S) to west (21°-24°S over a very severe rainfall gradient (compar- ison D vs. B) the life form trends were in very good agreement with those seen previously in north-south comparisons along the western side of the Andes: i.e., perennial herbs are more strongly represented on the wetter east- ern side of the Andes and shrubs more strong- ly represented on the drier western side in the most extreme environments, with annuals increasing only in the more benign mud-ele- vation belt on the western side. Thus the life- form tendencies revealed on the western side of the northern Andes are also valid on a regional scale. — LIFE FORMS—INTERCONTINENTAI COMPARISONS Collins et al. (1983) provided life-form data for an altitudinal sequence of floras in the Great Basin in Utah, North America, sharing several floristic elements with the Andes (e.g., Atriplex, Ephedra) but where rainfall is gen- erally higher. Comparisons for the Andes and Great Basin vegetation belts with similar mean annual temperature (Table 4) reveal the same life-form trends seen across aridity gradients within the Andes, now on an intercontinental scale. For the Andes, with lower precipitation, the woody life form is twice as common and the perennial herb life form only one-half to one-third as common as in the Great Basin. Annual herbs are in excess in the mid-ele- vation belt on the western side of the Andes compared with Mt. Nebo. However, despite the much lower rainfall along the desert mar- gin and in the high-elevation belt compared with Arches and Bald Mountain, respectively, there is no proportional increase for annuals in these harshest climates in the Andes. This once again emphasizes that annuals increase with aridity but that there is a critical point at which the woody habit becomes relatively more appropriate for arid climates. Most surprisingly, as the Arches vs. 2,000- 3,000 m (22°-24°S, eastern Andes) shows, woody species continue to be in excess and perennials in deficit in the Andes when annual precipitation is relatively similar. This indi- cates that the northern Andes exhibit a gen- eralized excess in woody species over the Great Basin beyond local trends in the latter for increased frequency of woody species with habitat harshness. COMMUNITY DIVERSITY PATTERNS In harsh arid habitats, as was argued ear- lier, relatively higher community diversity levels, as seen at the early stages of primary succession, are to be expected as a permanent community feature in spite of overall low species richness. In the Andes, regressions of exp H’ and 1 /X on species richness and cover show (Figs. 9, 10) that harsh environmental conditions, on a biogeographical scale, lead to an overall trend for loss of diversity. Thus the primary succession analogy proposed is not entirely supported. Over each gradient (Fig. 11) di- versity peaks at mid elevations, where con- ditions are neither the coldest nor the most arid, as was seen for species richness and cover; however, decreases in diversity are not commensurate with reductions in species rich- ness. This phenomenon is more easily appre- Arroyo et al. 71 Aridity/Plant Diversity in Northern Chilean Andes Volume 75, Number 1 1988 LE 4. Statistical Pii (G-tests) for life forms in the flora of the northern Andes (west and east "s and the Great Basi Comparisons are made for elevational levels with the closest mean annual temperatures. Data for ie ym areas are those in Table 2. Data for the Great Basin taken from Collins et 1983). al. ( G-test (Utah vs. Eastern ) G-test (Utah vs. Western Side of Andes) Side of Andes Great Basin, Utah 8- (2,000-3,000 m) 21-24°S (2,000-3,000 m) 22-245 (2,000-3,000 m) Arches (1,200 m) (13.5°C; 217 mm)! 68 mm 6 mm 200 mm Shrubs & trees 18.32% 12.59*** (Andes?) NA‘ 17.66*** (Andes) Perennial herbs 62.73% 11.80*** (Utah) NA 12.34*** (Utah) Annual herbs 18.94% 3.09; NS? NA 1.20; NS (322 spp.) Mt. Nebo (2,500 m) (9.4°C; 641 mm) 153 mm 30 mm 249 mm Shrubs & trees 12.98% 37.88*** (Andes) 25.48*** (Andes) 48.12*** (Andes) Perennial herbs 76.51% 53.04*** (Utah) 56.58*** (Utah) 44.08*** (Utah) Annual herbs 10.51% 5.90* (Andes) 17.82*** (Andes) 0.76; NS (647 spp.) Bald Mt. (3,200 m) (—2.4°C; 1,028 mm) 266 mm 97 mm 330 mm Shrubs & trees 8.96% 6.38* (Andes) 26.48*** (Andes) 5.85* (Andes) Perennial herbs 85.82% 5.86* (Utah) 32.81*** (Utah) 6.50* (Utah) Annual herbs 5.22% 1.10; NS 3.19; NS 0.29; NS (134 spp.) ' Mean annual temperature and mean annual precipitation. ? Area for which life form is proportionately better represented. ? Significance levels for G-test: + G-test not applicable ciated in Figure 12, in which diversity and species richness are compared for equivalent elevations from the wetter 18?5 to the drier 24°S. At 3,000-3,500 m and 3,500-4,000 m, where conditions are more favorable on each transect, there are significant reductions in diversity as aridity increases; however, they are shallower than for species richness. As higher elevations doubly stressed by aridity and cold temperature are reached, there is no significant difference in diversity from 18? to 24°S in spite of considerable reduction in species richness. Thus, in the sense that com- munity diversity is maintained relatively high in the face of large losses in species richness, the diversity trends seen in the most arid areas of the northern Andes are indeed analogous to a primary succession situation. The more gradual loss of diversity relative * P < 0.05; ** P < 0.01 ; *** p < 0.005; NS = not significant. to species richness may be related to the relative contribution of each life form and the presence of far more rare species in the rel- atively benign areas. In the most productive environments (18°S), one life form typically stands out as strongly dominant. Moreover, within that life form a few fairly abundant species tend to be accompanied by large num- bers of relatively rare species. At mid ele- vations at 18°S, for example, in spite of the fact that perennial herbs are well represented in numbers of species (see section on life forms), the woody life form is more abundant than the herbaceous life forms (which can persist in low densities because of the rela- tively benign conditions). Within the shrub guild, close coexistence, moreover, seems to have precipitated fairly strong dominance hierarchies as seen in great abundance of such Annals of the Missouri Botanical Garden | xe) L L L 1 1 L d md [e] 40 80 120 160 MEAN e" PER 500M ELEVATION PER 500M ELEVATION SPECIES RICHNESS FIGURE 9. Linear regressions y puit on num- ber of species dera richness).—A. exp H' vs. species richness (y — + 1.96; F = 11.511; d.f. = 17; P = 0.003, where y = exp H' and x = species rich- ness). — B. 1 /X vs. species richness (y = 0.019 + 0.007x + 78; F = 2.811; df = 17; P = 0.01, where y = 1/X and x = species richness). Points on graphs cor- respond to 500-m elevational intervals on the four sum- mer-rainfall transects. species as Fabiana densa, Baccharis boli- viensis, Ephedra breana, and Diphloste- phium meyenii. From 4,500 to 5,000 m at 18°S, bunch grasses become strongly domi- nant over shrubs, and again there are few strongly dominant species (Festuca ortho- phylla, Poa sp.). In the less productive en- vironments at 21%-24%S, in contrast, for the high elevations in particular, abundance levels for shrubs and herbaceous species are prob- ably more equitable, in spite of the fact that there are relatively fewer species of perennial herbs. These last considerations are based on 2 ó 40 E S a ° WwW ar LJ > g a uJ a = o T z e < š | [6] SS oss 1 L 1 Ë l 1 J 5 15 25 35 45 PLANT COVER (%) 2 O 3.0 [ " Fr < > ul ur ul = Q Q 19) a d F B -|< = ° u l [e] 1 1 1 L 1 L 1 J z 5 15 25 35 45 PLANT COVER (%) FIGURE 10. Linear regressions of diversity on plant cover (% ground m . exp H' vs. plant cover = 0.032x + 1.904; F = 12.06; d.f. = 17; P = 0.005, where y = exp H' and x = plant cover).—B. 1/A vs. plant cover y = 0.019x + 1.762; F = 7.06; = 17; P = 0.05, where y = 1/X and x = plant Cun Points on graphs correspond to 500-m eleva- tional intervals on the four summer-rainfall transects. direct observation; much analytical work is still required to characterize dominance re- lations in the Andean flora. DISCUSSION Patterns of plant diversity have been de- scribed for an area that underwent dramatic environmental upheavals in the Pleistocene and which acquired its present extreme arid character over a relatively short period. The effects of aridity on species richness in the northern Andes clearly have been se- vere. The 769 species for all transects are believed to include about 75% of all species Volume 75, Number 1 1988 Arroyo et al. 73 Aridity/Plant Diversity in Northern Chilean Andes 4 4 160 4.0} 4 + 4 4 43 120 3.0 F J] ü J oe 4 480 ~ o 20L Ë ° = |] | a o -|< " . oF - 1" 2 s = T 4 ° _ = [e] SX af al Kaya; Y L 1 1 1 1 1 4 [e] = x o o u =~ o e 21°S 7 24°S 199 u — 30F | al bii 4 + 80 2.0+ H Pe. B / N | — y 1.0 L = 4 40 r é 1 a `. ia 3 ` 4 1 L L L 1 1 1 L 2000 3000 4000 2000 3000 4000 ELEVATION (m) Mean exp H', exp H'/ (1/X) FIGURE northern Chicas Andes Lack of variation in exp H' a satis index are similar to those obtained using exp H'. Vertical bars are 95% yi Pd of 1/ onfidence SCR pe the m from 1,500 m to the upper vegetation limit from 17°S to 28°S. This leads to an estimate of under 1,000 species per 10? latitude or the equivalent of less than one-fifth of the total Chilean flora (Marticorena & Quezada, 1985). This is only one-sixth more species present over 1? of latitude from 3,500 to 5,000 m in Parque Nacional Huascaran (8*5) in the northern Peruvian Andes (estimated by David Smith, pers. comm., to have 660 species). It is only just over three times the number above tree line (309 species) at 33°S (Arroyo et al., 1983) for an area similar to that sampled in the individual northern tran- sects and in the paramos of Colombia (4?N; Rangel et al., 1983), where 321 species occur above tree line on an area covering much less than 1? of latitude. The Andes of Jujuy in northwestern Argentina (22°-24°S) immedi- ately across the main divide, where it was seen that rainfall is much higher, support 622 species (Ruthsatz, 1977), in contrast to only 199 at 21? and 2458 on the Chilean side. , and species richness vs. elevation at different latitudes in the 1/X) indicates that diversity trends resulting from Areas least affected by aridity and cold temperature proved to be richest in species. Such areas also exhibit high. community. di- versity. Highly stressed areas with low species richness, nevertheless, are relatively robust for community diversity. Any seasonal vari- ation in annual growth, as along the desert edge, could greatly affect community diver- sity estimates. Revisits to the Andean high- lands in from Arica (18°S) in a very wet year subsequent to vegetation sampling leave little doubt that annual cover at lower elevations there fluctuates widely. However, the trends above 4,000 m, where there are very few species of annuals, are unambiguous. Maintenance of relatively high local levels of community diversity in the most arid areas will probably turn out to be the result of lowered dominance levels in the drier areas. Thus aridity, the very feature driving lowered species richness in the Andes from the outset, probably eventually reduces the rate at which local diversity declines. In this sense the not 74 Annals of the Missouri Botanical Garden 3000- 3500 m 20r ~ LI 7 : S NN w e A4 1.0 F ` -|< ° po --- | ~ L T [6] +; 1 1 1 1 1 T " 4000-4500 m e 4 w 3.0 F 4 2.0 F 1.0 7 [e] k l 1 + 3500 - 4000 m 4500- 5000 m o SPECIES RICHNESS (+) ] 100 ] 60 ` L grio a 4 \ "c a --7 ~~ [ T 20 ~~ 1 4 | 7" 18 20 22 24 LATITUDE (°S) FIGURE uo to 24*S in confidence intervals for the means. strongly exponential drops in species richness with aridity should be recalled. They seem to indicate that loss of species richness has been counteracted to some extent. The life-form changes we have demonstrated must also be important here; however, it is far too early to sort out the relative contributions of dif- ferent factors against species richness loss. Stress-tolerant (e.g., annual life form) and stress-resistant (e.g., shrub life form) strate- gies are often considered as hierachically equivalent alternatives for harsh environ- ments. Our results suggest that woody species gain prominence over annuals on the harshest of the Andean habitats. Significantly, more- over, some of the most successful “perennial herbs" from the driest areas of the Andes (e.g., Sisymbrium philippianum, S. lana- tum, Tarasa operculata) have semiwoody stems. Thus the trend towards woodiness goes beyond simple reshuffling of the taxa present Mean exp H', exp H'/ (1/X), and species richness vs. latitude from 18°S (maximum precipitation) Pp exp P preci imum precipitation). Lack of variation in exp from Be eatin of 1/X as a diversity index are similar to those obtained using exp H'. H'/(1/X) indicates that diversity trends resulting Vertical bars are e 95% and is probably being actively selected for in some individual taxa. Woody species are probably best adapted to extreme aridity through a combination of ecophysiological and demographic features. Root/shoot ratios for warm-desert species are usually around one, and there is relatively little vertical root growth (Barbour, 1981). In contrast, root/shoot ratios in excess of four have been reported for shrubs in cold North American and Eurasian deserts (Caldwell, 1985). This suggests that for cold arid areas, where growth is relatively slow, large long- lived species might be favored by being able to produce larger root systems that would reach deep into the soil over a prolonged growth season. Soils in the northern Chilean Andes are largely volcanic, hence surface water is probably always limited. Further, in the driest areas of the summer-precipitation zone, rain comes in light morning showers Volume 75, Number 1 1988 Arroyo et al. 75 Aridity/Plant Diversity in Northern Chilean Andes and is never abundant at any one time. We suspect that this factor favors shrubs in the driest areas and, as was mentioned earlier, explains why annuals are generally less fre- quent in the summer-rainfall zone. Because they tend to be long-lived, woody species require less frequent establishment events (Schaffer & Gadgil, 1975). This should be highly advantageous in habitats where seed germination and seedling establishment are precarious, as is the case with strongly arid climates. Seed production is not only a func- tion of pollination success and of resources allocated to reproduction, but also of the prob- ability of an adult reaching reproductive ma- turity and its physiological state at that stage. Woody-stemmed shrubs are more likely to reach reproductive maturity than soft- stemmed herbs because of their greater re- sistance to drought. In spite of the funda- mental nature of the question, as far as we know, to date there have been no attempts to appraise the relative importance of the physiological and demographic features of woody species in harsh environments. The greater representation of woody species in the northern Andes might reflect the Andes being located some ten degrees of latitude closer to the equator than is the Great Basin. The floristic matrix out of which the Andean flora evolved, as a result, probably possessed a higher percentage of woody elements ini- tially. Fossil floras for the Great Basin (Ax- elrod, 1979a, 1983) and as far south as the Chihuahuan desert (Wells & Woodcock, 1985) show North American desert floras emerging out of open woodland with many herbaceous elements. The radical climatic changes at each glacial/interglacial interface in the Andes should have further impover- ished the herbaceous flora. These historical factors, apart from effects of present climatic characteristics, should have produced a grad- ual accumulation of woody elements in the northern Andes. Heavy grazing in the Andes (alpaca, llama) has possibly influenced the broad interconti- nental differences in life form. Interestingly, areas of the northern Andes in which woody species are most strongly represented today are least affected by grazing (e.g., 21°-24°S), while those with well-developed herbaceous floras can be heavily grazed (north of 19°S). Very plausibly, climatic and biotic factors have acted in concert to produce the interconti- nental differences in life form. Emphasis on the long-lived, woody life form, as seen in harsh environments in the northern Chilean Andes, is also a feature of tropical and other forests developed in abiotically be- nign conditions. For the tropics in particular, longevity is usually seen as a correlate of large body size resulting from selection for com- petition for light (Grime, 1979). Much wood- iness in the tropics could equally bear relation to the demographic advantages of being long- lived, as was suggested for abiotically harsh habitats. In the tropics, seed predation (Ra- mirez & Arroyo, 1987) and fungal infections (Ramirez & Arroyo, 1984) can significantly lower successful seed germination. Addition- ally, juvenile mortality can be high due to strong intra- and interspecific competition for light and nutrients (Connell et al., 1984) and leaf predation (Clark & Clark, 1985). These. features have direct parallels in the extremely arid Andean ecosystem, the difference being that in the tropics they are mediated through the biotic environment rather than the abiotic environment. Grime (1979) recognized that "stress" in productive environments arises mainly through competitive depletion of re- sources (the biotic environment). Yet he per- haps placed undue emphasis on longevity in tropical ecosystems as being a result of se- lection for direct competitive ability without paying much attention to the demographic advantages. As in arid environments, the role of longevity for tropical forests needs more critical assessment. Diversity patterns in the kinds of “‘abioti- cally" harsh environments we studied and in "biotically harsh" tropical forest communities might also show convergent trends. Many tropical forests lack clear dominants—domi- nance is probably less well developed in the Annals of the Missouri Botanical Garden most arid environments we studied. Lack of dominance in tropical forests has been seen as an effect of reduced opportunities for co- evolution resulting from low interspecific con- nectance sensu Pimm (1984) (Connell, 1980), or in Buckley's (1983) words, “in diverse, well-mixed communities individuals cannot predict their neighbours." Hubbell & Foster (1986) proposed a similar but not identical hypothesis: individual species are seen to re- flect the temporal and spatial average of biotic selective conditions created by ever changing and diffuse competitors, leading to guilds of functionally equivalent generalists. Sustained iotic interactions are also seen to be limited in very arid environments. Wetter deserts like the Mohave and Sonoran provide some evi- dence of root competition between widely spaced individuals (Yeaton & Cody, 1976; Phillips & MacMahon, 1981). Gulmon et al. (1979), by contrast, concluded that limited opportunities for establishment lead to little competition between individuals of the long- lived Copiapoa (Cactaceae) in the Chilean coastal desert at 25?S, where ca. 25 mm of rain falls annually. This is tantamount to low connectance, the abiotic environment assum- ing the role of the biotic environment in a tropical forest. We suspect that a suite of common fea- tures will be revealed for these two kinds of "harsh" environments for the organisms that inhabit them. Convergence could be expected in breeding system (e.g., levels of sexual di- morphism), seed size, and sexual selection. Some of these possibilities are presently being studied by us in cold alpine habitats in the Chilean Patagonia (e.g., Arroyo & Squeo, 1987) LITERATURE CITED AB'sABER, A. N. 1982. The paleoclimate and paleo- ecology of Brazilian Amazonia. -59 in G. das (editor), Biological Diversification in the Trop- olumbia York. Reni. F. 1956. . In: W. F. Jenks Meu E ndbook of South Pd. Geology. Mem. Geo c. Amer. 65: 167- ob R. W. 1986. Tectonic aia MR southeastern border of the Puna plateau, northwest ern Argentine Andes. Bull. Geol. Soc. Amer. 97: 1070-1082. ARROYO, M. T. K. & F. Squeo. 1987. Experimental detection of anemophily in Pernettya mucronata (Ericaceae) in western Patagonia. Bot. Jahrb. Syst. 108 p OP , J. ARMESTO & R. PRiMACK. 1983. Tendencias ETE He us y Dr nadiialea en mecanismos de polini- ación en los Andes templados de ATA Re- vista vie Hist. Nat. 56: 159- RTICORENA & C. VUE 1984. La flora de la Cordillera de los Andes en el áre Laguna Grande M EA Chica, III Región, Chile. Gayana, jd 6. , C. VILLAGRÁN, C. MARTICORENA & J. ARMESTO. 1982. Flora y lociones biogeográficas en una tran- secta Ad Ww en los Andes del norte de Chile (18- : A. Veloso & E. e A El Am- biente Natural y las Poblacion s de los Andes cel Norte Grande de Chile ee p 18°- 28°S) 1: 71-92. AXELROD, D. I. . Age and origin of the Sonoran Desert vegetation. Occas. Pap. Calif. Acad. Sci. 132: -14. ——— —. 1979b. ta vegetation, its age and origin. Pp. 1-72 in J. R. Goodin & D. K. Northington (editors), Arid o Plant Resources, Proc. Int. Arid Lands Conf. Pl. Resources. Texas Tech University, HE Texas : Paleobotanical history of the western E. Pp. 113-129 in S. G. Wells & D. R. Har- Rn (editors), Origin and Evolution of Deserts. Univ. exico Press, Albuquerque, New Mexico. E. e C. nD Arid. a Ecosystems: structure, m an management. Cambridge Univ. Press, Cambr Berry, E. W. 1917. Fossil plants from Bolivia nd cm . upon the age of uplift of qu eastern Andes Proc. U.S. Natl. Mus. 53: 103-164. 1919. Miocene fossil plants from northern Proc. U.S. Natl. Mus. 55: 279-294. . 1922. Pliocene fossil plants from eastern Bo- i m Hopkins Univ., Stud. Geol. 4: 145-202. 939. The fossil flora of Potosi, Bolivia. Johns . 13: 9-72. E . A possible mechanism for main- taining diversit ity in species-rich communities: an ad- dendum to Connell’s hypothesis. Oikos 40: 312. CALDWELL, M. 85. Cold desert. Pp. 198-221 in B. Chabot & H. A. Mooney (editors), Physiological Ecology of North American Communities. Chapman & Hall, New CaNE, M. A. 1983. Oceanographic events during El Nino. Science 222: 1189-1195. CasTRI, F. pt & E. R. Hajek. 1976. Bioclimatologia de Chile. Vicerectoria AME Universidad Ca- tólica de Chile, Santia Castro, M., C. VILLAGRÁN "A M. T. K. Arroyo. 1982. Estudio etnobotánico en la pre-Cordillera y Altiplano de los Andes de Arica (18-195). In: A. Veloso & - Bustos TUE El Ambiente Natural y las Pobla- nes Humanas de los Andes m due Grande de Chile une pn 18°-28°S) 2: 203 Volume 75, Number 1 1988 Arroyo et al. 77 Aridity/Plant Diversity in Northern Chilean Andes CLARK, A. H., A. E. S. Mayer, C. Mortimer, R. H. SILLITOE, "à v. Cooke & N. J. SNELLING. 1967. Implications of the isotopic age of ignimibrite flows, aa Atacama Desert, Chile. Nature 215: 723- ke D. B. & D. A. CLARK. 1985. Seedling damage of a tropical tree: impacts of herbivory and meristem T n del paramo neo- tropical y sus lazos australo- persos Colombia Geo- grà NE Counvat, P. pon The ice-age Amazon. Nature 278: 399- COLLINS, P. D., K. R & B. PENDLETON. 1983. Comparative life po and floral charac- unes of desert and mountain floras in Utah. Great sepa 43: 385-393. Godin. J.H. 1980. Diversity and the coevolution of competitors, or the ghost of competition past. Oikos 35: 131-138. ——, J. G. Tracey & L. is Wess. 1984. Compen- satory recruitment, gro and mortality as factors maintain ning rain Coen inn din: Ecol. Monogr Damuth, J. E. € R. W. m 1970. Equatorial Atlantic deep-sea arkosic sands and ice-age aridity al South inen “Bull Geol. Soc. Amer. in tropica DIAMOND, J. 1986. pci laboratory experiments, 2 experiments, and natural experiments. Pp. 3- n J. Diamond & T. J. cus (editors), Community colo ogy. Harper & Row, New York. . & A. V. Geoecology of the Palynological investigations of two o pos glacial peat bogs near the 2 y of Bolivia d u. J. Biogeogr. 8: 353 19 de F P. : Plant st oe and vegetation pa E Wiley & Sons, New York. GULMON, S. P. W. RUNDEL, . EHLERINGER & H. iud 1979. Spatial relationships and com- er in a Chilean desert cactus. Oecologia 43: 40-43. Hansen, B. C. S., H. E. Wnicur, Jr. & J. P. BRADBURY. 1984. Pollen studies in the Junin area, central Pe- ruvian Andes. Bull. Geol. Soc. Amer. 95: 1454- 65. Observations on the snow HASTENR ATH, S. L. der er Entwicklung der Mollusken-Faunen. Zitteliana 2: 59. & R. Paskorr. 1967. Vorschlag zur Gliede- rung des marinen Quartárs in nord- und mittel-Chile. J. Jahrb. Geol. Paláontol. 10: 577-588. Heusser, C. J. 1983. Quaternary calles record from num de Tagua-Tagua, Chile. Science 219: 1429- 2. Hio HIEMSTRA, H. 1984. Vegetational and x History of the High Plain of Bogotá, Colombia: continuous record of the last 3.5 million years. Dis. precia Botanicae, Bd. 79. J. Cramer, Vaduz. Houssanp, C., J. ESCARRE & F. Romane. 1980. velopment of species diversity in some i ead plant communities. Vegetatio 43: 59- HunBELL, P. S. & R. Foster. 1986. Biology, chance and history and the structure of tropical rain forest tree communities. Pp. 314-329 in J. Diamond & T. J. Case (editors), Community Ecology. Harper & Row, New York. Mares, M. A. 1985. Mammal faunas of xeric habitats and the Great American Biotic Interchange. Pp. 489- 520 in F. G. Stehli & S. D. Webb (editors), The a American Biotic Interchange. Plenum Press, or Miras, V. . Late and postglacial vegetational and paleoclimatic changes in subantarctic, temper- ate, and arid environments in Argentina. Palynology 7:4 2 and |o, chan paleoclimates of the Ar- ntine dm oe 4 (in press). MARTICORENA, C. $ M QUEZA 1985. Catálogo de 85. Tectonic significance of silicic dikes contem- poraneous with latest Miocene Quechua 3 tectonism in Ki Rimac Valley, western Cordillera of Central Peru. J. Geol. 93: 373-376. Morata, C. The evolution of the continental margin of northern Chile. Int. Geol. Congr. 24: 48- l. The Cenozoic history of the southern Atacama Desert, Chile. J. Geol. Soc. 129: 505-526. 975. Cenozoic studies in northernmost Chile. eal, Rundschau 64: 395-420. Drainage evolution in the Atacama Desena E E Chile. Revista Geol. Chile : 3-28 1972. Landform evolution in Rev. Geo- N. R SARIC. lids region of Tarapacá ag Chile. rphol. Dynam. 4: 162-170. TEF ARRAR & N. "at 1974. K-Ar ages from Tectary lavas of the northernmost Chilean Andes. Geol. Rundschau 63: 484-491. Muizon, C. DE & T. DeVries. 1985. Geology and paleontology of late Cenozoic marine deposits in the Sacaco area d^ Geol. Rundschau 74: 547-563. NARANJO, J.A & R. PASKOFF. aru sur; revisión cronológica. Revista Geol. Chile 10: 85-89. . 1980 i» Estratigrafi a de los de- pósitos Cenozoicos de la region de Chiuchiu-Calama, Desierto de Atacama. Roce Geol. Chile 13-14: 79-85 1985. Evolución cenozoica del piedemonte andino en la pampa del Tamarugal, norte de Chile (18°-21°S). IV Congr. Geol. Chile, Univ. del Norte de Chile, Antofagasta 4: 149-165. OcusENIUS, C. 1983a. Aridity and biogeography in northernmost South America during the late Pleis- tocene (Peri-Caribbean x belt, 62°-74°W). Zen tralbl. ys Paláontol. 1: 264-278. : paste hologenesis of the Pacific ¿qna ia in the context of the tropical South America pa In: T. Smiley (editor), The Geo- 78 Annals of the Missouri Botanical Garden logical Story of the World's Deserts. Striae 17: 112- 13 & R. SANTANA. 1974, VANS D n geográficas y aer v d entre los ambientes lén- ticos de la e Atacama y Alti ies Boliviano, Tropico de ees Bol. Prehist. Chile 7-8: 99- 138. Paskorr, R. & J. NARANJO. 1979. Les grandes étapes de Tale geomorphologique des Andes pendant le Cenozoique dans le sud du désert d'Ata- cama n Compt. Rend. Hebd. Séances Acad. Sci. 16: 1203 6. PAVLIK, B. x 1957; Sand dune flora of the Great Basin Sri ts of California. Madroño 32: 197- peer, A K. 1974. The measurement of a diver- Annual Rev. Ecol. Syst. 5: 285-30 ee D. L. & J. A. MacManow. 1981. (e and spacing patterns in desert shrubs. J. Ecol. 69 97-115. HMM S. L. 1984. The complexity and stability of osystems. Nature 307: 321-326. Prance, G. T. 1982. A review of the ap eae Apos evidences for Pleistocene climatic chang Neotropics. Ann. Missouri Bot. Gard. 69: 594- 624. diria N. € . K. Arroyo, 1984. Infección de millas por r hon ngos en Copaifera pubiflora Benth. Caesalpinioideae] en los Altos roa eiae de Venezuela. Bol. ie . Venez. Ci. Nat. 165-173. & Variación d y tem- poral de la pubes de semillas en Copaifera pubiflora Benth. o sa esl en Venezuela. eiae 19: a Dia JARAM MILLO & Š. SALAMANCA evados Mid eae Spermato- phyta n der Hammen, A. P. Preciado & P. ae elton). Studies on J Andean Eco- syst . Cramer, Vaduz 1 205. Re E. M. & J. M. WaLLace. 1983. Meteo- rological ispeci of the El Nino/Southern Oscillation. Science : 1195-1 1983. The Peruvian- dia Ris In: N. E. West So Temperate Desert mi-des- erts. Ecosystems of the World. Elsevier, pene yasa 5: 239-267. Raven, P. H. & D. I. AXELROD. 1978. Origin and pei d 2 the California flora. Univ. Calif. Publ. Bo RurHsaTz, B. 1977. Pflanzengesellschaften und ihre Lebensbedingungen in d n Halbwüsten Nordwest- Bg CU Diss. Bot. 39: 1-168. RurLAND, R. W 71. Andean orogeny and ocean floor spreading, Nature 233: 252-255. r & R. L. Grasty. 1965. Isotopic ages and dn uplift. Nature 208: 677-678. SCHAFFER, W. M. & M. D. Gapcit. 1975. Selection for optimal life histories in plants. Pp. 142-157 in . Cody & J. M. Diamond (editors), Ecology and ib olas of Communities. Belknap Press, Harvard . E. ERICKSEN. 1974. Geology of salars in ninotthem Chile. Geol. Surv. Prof. Pap. 811: 1-65. TILMAN, D. Evolution and differentiation in & Row, New York. H. CLARK & E. Farrar. 1984. Cenozoic polyphase landscape and tectonic evolution of the Cordillera outhernmost Peru. Bull. eol. Soc. Amer. 95: 1318-1332. Tricart, J. 1969. Le Salar de Huasco. Rev. Geomorph. Dynam. 2: 1-84. VILLAGRÁN, C., M. T. K. ARROYO & J. ARMESTO. 1982. Vegetación de una transecta altitudinal en los Andes del Norte de Chile (18-19°S). In: A. Veloso & E. Bustos (editors), El Ambiente Natural y las Pobla- ciones Humanas de los Andes e Norte Grande de Chile (Arica, Lat. 18 285) 1: 13-70. & C. MARTICORENA. 1983. Efectos de M desertización en la distribución de la flora andina de Chile. Revista Chilena Hist. Nat. 56: 137-157. WELLS, . Woopcock. 1985. Full-glacial vegetation of Death Valley, California: juniper wood- land opening to yucca semi-desert. Madroño 32: 11- 23 WHITTAKER, R. H. 1972. o and measurement of species diversity. Taxon 21: 213-251. YEATON, R. I. & M. L. Copy. 1976. Competition and spacing in plant PELES the northern Mojave Desert. J. Ecol. 64 6. ZINSMEISTER, W. J. Yu Effect of formation of the west Antarctic ice sheet on vei E marine -26 faunas of Chile. Antarctic J. U.S, PATTERNS OF SPECIES William E. Duellman? DIVERSITY IN ANURAN AMPHIBIANS IN THE AMERICAN TROPICS! ABSTRACT The Neotropical Region has a greater species richness of anuran amphibians than any other region in the world. Approximately 44% of the total number of species of anurans of the world (3,533) occur in the American tropics, and many ig species are discovered every year. Patterns of species diversity were determined by analyzing data from sites—32 in the lowland tropics (11 Middle American, 21 South American), nine in montane cloud forests oe Middle American, five South American), and seven in supra-treeline regions of the Andes. Taxa also were noted as to their reproductive mode (site of egg deposition, site of larval development, and associated parental care, if any). As expected, for the entire anuran fauna, there are gradients from lower diversity in dry regions to higher diversity in wet regions, and from — diversity at high elevations to higher diversity at low elevations. The greatest species diversity is in the equatorial region of the upper Amazon Basin. However, different patterns emerge when taxonomic groups (families, subfamilion and large genera) are examined independently. Two factors contribute significantly to the different patterns observed: (1) the historical bio- t opment of eggs and tadpoles. Sites’ haoin the highest species diversity also have the greatest diversit : a the À I uran is dependent on rainfall, patterns of diversity of component groups results partly from their geographic histories and partly from the consequences of the nature of their reproductive modes Despite the fact that most amphibians have (salamanders), Gymnophiona (caecilians), and rather narrow physiological tolerances and Anura (frogs and toads). The last is by far ecological distributions, in their successful at- the largest, with 3,533 currently recognized tainment of independence from water and the species. colonization of the land they have undergone Anyone who has spent a rainy night in a a remarkable adaptive radiation. The living neotropical rainforest is aware of the noctur- amphibians exhibit a greater diversity of life nal cacophony raised by calling males of many histories than any other vertebrates. e species of frogs. The American tropics are nearly 4,000 living species of amphibians are especially rich in species of anurans; at one grouped into three distinct orders—Caudata — site in the upper Amazon Basin in Ecuador ! The compilation of data from so many sites has been possible through the generosity ee colleagues who provided unpublished data on n their own fieldwork; for their kind help I am grateful to Patricia A. Burrowes, nald I. Crom i iarmi Zimmerman. My own ipai on neotropical anurans have been supported by the Museum of Natural History at The University of Kansas and by grants from the National Science Foundation and the National Geographic Society. Many persons hel par me in the fieldwork, but I am especially indebted to John E. Simmons and Linda Trueb. 2 Museum of Natural History and Department of Systematics and Ecology, The University of Kansas, Lawrence, Kansas 66045, U.S.A. ANN. MISSOURI Bor. Ganp. 75: 79-104. 1988. 80 Annals of the Missouri Botanical Garden 104 96 88 80 Y Ë T I NE A AO 4 E NX J Ts J ZV 1 ad d Ww 1 £5 7 Lone k Y) $ Bud Sx ^ 2 7 ie = 03 1 ARM Y Tos a "iod Y | j -118 4 4s ta SS rr re M hae 6° ` bi i D ° i H | i c 7 T 82 x L " y hala P SCALE Hil us (0) 200 800 kms 3. ——r ; : O 100 500 miles 14% 156" l l l a V— rs 104 96 88 80 FIGURE 1. Study sites in Mexico and Central America. Dots are pou sites; squares are montane cloud forest sites. Numbers i to the jd of sites in the te 4 TEHU, *, , 14 = RINC, 15 = TUIR there are more species of anurans than occur in all of the United States and Canada (Duell- man, 1978). The diversity is much higher than on other Gondwanan continents— 56.0 species/million km? in South America, as compared with 11.7 in Africa and 20.1 in Australia (Duellman, 1979a). Therefore, the American tropics represent an ideal region for investigation of patterns of species diver- sity in anuran amphibians. Salamanders and caecilians have been omitted because the for- mer are poorly represented in South America and because the latter are too poorly known to be analyzed. The purposes of this paper are to (1) sum- marize data on species diversity from many sites in the American tropics, (2) examine the patterns of species diversity, and (3) interpret these patterns in relation to present climatic patterns, historical biogeography, and repro- ductive modes. Í consider the term species = CHIN, 9 = PURU, 10 = LSEL, e xt: 1 = MAZA, 2 = PIST, 3 = APAT, 4 = VHER, 5 = TABO, 12 = BCIS, 13 = CRUC, diversity to be synonymous with species den- sity and species richness. MATERIALS AND METHODS For purposes of defining the anuran fauna of tropical America, I did not include any species confined to Argentina, Chile, or Uru- guay, and Í included no North American species that do not range into the tropical regions of Mexico. Lists of species of anurans were obtained for 48 sites in tropical South America, including the Andes (Figs. 1, 2); no sites were incorporated from the West Indies. Each site is a relatively small area, usually encompassing fewer than 10,000 hectares. I have worked at of the sites. In the following list of sites, each is iden- tified by an abbreviation (used in tables), name, geographical coordinates, elevation, vegeta- tion formation (Holdridge, 1964), and ref- Volume 75, Number 1 1988 Duellman 81 Anuran Amphibian Diversity in American Tropics 200 Scale of Miles \ 80 FIGURE 2. Study sites in South America. Dots are lowland sites; de mala are cloud forest sites, and triangles = RGR NT, 4 are high-Andean I Numbers eu to the list ear in the tex AMPA, 5 = = RUSI, 7 SALA, 14 = OZAP. 15 = PALE, S ALI, 8 = O — PURA, 10 - PLAN, 11 = ANGE, 12 = SCEL, 13 = = TROM, 2 = BELE, 18 = MANA, 19 = AZAM, 20 = CENE, 21 = TAPA, 22 = AHUA, 23 = PANG, 24 = BALT, 25 = CAMA, 26 = TAMB, 27 = COSN, 28 = ACAN, 29 = SROS, 30 = FILA, 31 = CCOR, 32 = BORA, 33 = YBYC erences to the source(s) of data. Sites are referenced by numbers in brackets to the maps in Figures 1 and LOWLAND SITES AMPA.—Rio Amparrado, Antioquia, Co- lombia [4] 06°50'N, 76?25'W, 800 m. Very wet tropical forest. J. D. Lynch (pers. comm.). APAT.— Apatzingán, Michoacán, Mexico [3] 19%06'N, 102?22'W, 335 m. Very dry tropical forest. Duellman (1961, 1965b). BALT.— Balta, Loreto, Peru [24] 10%08'S, 71°13'W, 300 m. Wet tropical forest. Duell- man (unpubl. data). BCIS.— Barro Colorado Island, Panamá, Panama [12] 09°10'N, 79*50'W, 150 m. Wet tropical forest. Myers & Rand (1969). BELE.—Belém, Para, Brazil [17 ] 01?21'S, 48°30'W, 12 m. Wet tropical forest. Crump (1971). BORA.—Boracéia, Sao Paulo, Brazil [32 ] 23°38'S, 45°50'W, 900 m. Wet subtropical forest. R. W. Heyer (pers. comm.). CALI.—Rio Calima, Valle, Colombia [7 ] 03°57'N, 76%44'W, 200-460 m. Wet pre- montane forest. J. D. Lynch (pers. comm.). CAMA.—Cuzco Amazónico, Madre de Dios, Peru [25] 12*33'S, 69*03'W, 200 m. Wet tropical forest. Duellman (unpubl. data). CCOR.—Cerro Cora (Parque Nacional), Amambay, Paraguay [31 | 22?30'S, 56%05'W, 200-300 m. Dry tropical forest (and savan- na). N. J. Scott (pers. comm.). Annals Missouri Botanical Garden of the TABLE l. Distribution of species of anuran amphibians by family groups. Total Trop. Amer. % Trop. Amer. Family Group Genera Species Genera Species Genera Species Allophrynidae l l l l 100 100 Arthroleptid Arthroleptina 3 47 = = 0 0 stylosterninae 5 25 — — 0 Brachycephalidae 2 3 2 3 100 100 ufonidae 27 340 7 138 26 41 Centrolenidae 2 71 2 71 100 100 Dendrobatidae 3 117 3 117 100 100 Discoglossidae 5 16 — 0 0 Heleophrynidae 1 3 — 0 0 Hemisotinae l 8 — = 0 0 Hylidae Hemiphractinae 6 62 6 59 100 95 Hylinae 23 394 19 356 83 90 Pelodryadinae 3 147 — — 0 0 Phyllomedusinae 3 42 3 42 100 100 Hyperoliidae Hyperoliinae 7 133 — m 0 0 Kassininae 5 41 — — 0 0 Leptopelinae 2 42 = = 0 0 Leiopelmatidae 2 4 — = 0 0 Leptodactylidae Ceratophryinae 2 9 2 7 100 78 Hylodinae 3 22 3 21 100 95 Leptodactylinae 11 117 10 104 91 89 Telmatobiinae 35 592 25 554 74 94 Microhylidae Asterophryinae 7 40 PEE = 0 0 Brevicipitina 4 17 EE = 0 0 Cophylinae 9 35 — = 0 0 yscophinae 2 8 — — 0 0 Genyophryninae 6 76 — ee 0 0 Melanobatrachinae 3 4 — — 0 0 Microhylinae 27 95 17 38 63 40 Phrynomerinae 1 4 0 0 Scaphiophryninae 2 7 — == 0 0 Myobatrachidae Limnodynastinae 10 2 — — 0 0 yobatrachinae 10 T — == 0 0 Pelobatidae gophryinae 7 73 = — 0 Pelobatinae 2 10 1 2 50 20 Pelodytidae 1 2 m — 0 pidae Pipinae 1 7 1 7 100 100 Xenopodinae 3 19 = = 0 0 Pseudidae 2 4 2 3 100 75 Ranidae Mantellinae 3 60 — zm 0 0 Petropedetinae 11 86 = = 0 0 Raninae 24 459 1 21 4 3 Rhacophoridae Philautinae l 03 — = 0 0 Rhacophorinae 9 123 n — 0 0 Volume 75, Number 1 1988 Duellman 83 Anuran Amphibian Diversity in American Tropics TABLE l. Continued. Total Trop. Amer. % Trop. Amer. Family Group Genera Species Genera Species Genera Species Rhinodermatidae 1 2 — — 0 0 Rhinophrynidae l l l l 100 100 Sooglossidae 2 3 = = 0 0 Totals 301 3,533 107 1,545 36 44 CENE.—Rio Cenepa, Amazonas, Peru [20] 04°28'S, 78°10'W, 210 m. Wet tropical forest. R. W. McDiarmid (pers. comm.). CHIN.—Chinaja, Alta Verapaz, Guate- mala [8] 16%02'N, 90°13’W, 140 m. Wet tropical forest. Duellman (1963). CUYU.—Rio Cuyuni (13 km S of El Do- rado-Santa Elena de Uairén road), Bolivar, Venezuela [5] 06%37'N, 61°32'W, 140 m. Wet tropical forest. Duellman (unpubl. data). FILA.—Filadelfia, Boquerón, Paraguay [30] 22*15'S, 60%05'W, 200-300 m. Very dry tropical forest. N. J. Scott (pers. comm.). LSEL.—La Selva, Heredia, Costa Rica [10] 10%25'N, 83°57'W, 90 m. Wet tropical forest. Scott et al. (1983). MANA.—Manaus (INPA-W WF reserve), Amazonas, Brazil [18] 03?13'S, 60%02'W, 50 m. Wet tropical forest. Hodl (1977) and B. L. Zimmerman (pers. comm.). MANT.—El Manteco, Bolivar, Venezuela [3] 07?25'N, 62°21'W, 305 m. Dry tropical forest (savanna). Hoogmoed & Gorzula (1979). MASA.—Hato Masaguaral, Guarico, Ven- ezuela [2] 08°33'N, 67°35'W, 75 m. Dry tropical forest (savanna). Staton & Dixon (1977). MAZA.—Mazatlán, Sinaloa, Mexico [1 ] 23?13'N, 106?25'W, 10 m. Dry tropical for- est. Hardy & McDiarmid (1969). OYAP.—Upper Riviére Oyapock (Pina, Zidok, Trois-Sauts), French Guiana [8] 02?16'N, 52°52’W. Wet tropical forest. Les- cure (1976). PALE.—Rio Palenque, Pichincha, Ecua- dor [15] 00?18'S, 79?11'W, 200 m. Wet tropical forest. R. W. McDiarmid (pers. comm.) and Duellman (unpubl. data). PANG.—Panguana, Rio Lullapichis, Huánuco, Peru [23] 09535'S, 74?48'W, 200 m. Wet tropical forest. Toft & Duellman (1979) and Schlüter (1984). PIST.—Pisté, Yucatan, Mexico [2] 20?42'N, 88?28'W, 10 m. Very dry tropical forest. Duellman (19652). RINC.— Rincón de Osa, Puntarenas, Cos- ta Rica [14] 08?42'N, 83°29'W, 10 m. Wet tropical forest. Scott et al. (1983). SARA.— Rio Sarabia, Oaxaca, Mexico [6] 17?05'N, 95?02'W, 80 m. Wet tropical for- est. Duellman (1960). SCEL.—Santa Cecilia, Napo, Ecuador [12] 00°03'N, 76?59' W, 340 m. Very wet tropical forest. Duellman (1978). TABO.— Finca Taboga, Guanacaste, Cos- ta Rica [11] 10?20'N, 85?12'W, 40 m. Dry tropical forest. Scott et al. (1983). TAMB.—Tambopata, Madre de Dios, Peru [26] 12°50’S, 69°17'W, 290 m. Wet tropical forest. R. W. McDiarmid (pers. comm. ). TAPA.—Rio Tapajos, Para, Brazil [21] 04°36'S, 56?14'W, 85 m. Wet tropical for- est. R. I. Crombie (pers. comm.). TEHU.—Tehuantepec, Oaxaca, Mexico [7] 16?20'N, 95°14’W, 35 m. Dry tropical forest. Duellman (1960). TROM.—Rio Trombetas, Lago Jacaré, Pará, Brazil [16] 01?17'S, 56?46'W, 110 m. Wet tropical forest. R. I. Crombie (pers. comm.). TUIR.—Rio Tuira at Rio Mono, Darién, Panama [15] 07?42'N, 77?35'W, 130 m. Wet tropical forest. Duellman (unpubl. data). YBYC.—Ybycui (Parque Nacional), Pa- raguari, Paraguay [33] 25%50'S, 56°50'W, 300 m. Dry subtropical forest. N. J. Scott (pers. comm.). 84 Annals of the Missouri Botanical Garden TABLE 2. Tropical American genera of anurans and numbers of species inhabiting different regions—South America (SAMER), Middle America (MAMER), West Indies (W IND) in lowlands (LOW) and highlands (HIGH) .' MAMER SAMER SAMER MAMER MAMER W Genus LOW HIGH LOW HIGH SAMER IND Total Allophrynidae Allophryne 1 ees — = = = l Brachycephalidae Brachycephalus 2 E — — == — 2 Psyllophryne 1 — — = — — l Bufonidae Andinophryne — 4 — — — — 4 Atelopus 8 31 — 1 — 44 Bufo 23 16 10 10 2 — 61 Crepidophryne = — — l m — l Dendrophryniscus 3 — — E — — 3 Frostius l — — — — -— l Melanophryniscus 3 l — — — — 4 Oreophrynella — 2 -— — — — 2 Osornophryne — 3 — == = en 3 Peltophryne — = = -— — 9 9 Rhamphophryne l 5 = — — = 6 Centrolenidae Centrolene — 1 — — — — 1 Centrolenella T 50 2 3 8 — 70 Dendrobatidae Colostethus 15 43 — — 5 2 65 Dendrobates 32 8 2 3 2 — 47 Phyllobates 3 — 2 — — — 5 Hylidae Hemiphractinae LLL — 3 — — — — 3 Flectonotu — 1 — — — l 2 Fritziana — 3 — — — -- 3 Gastrotheca 5 32 — — 2 — 39 emiphractus 4 — — — l — 5 Stefania = T — — = — 7 Hylinae Anotheca — — — 1 — — 1 Aparasphenodon 2 ees — — — — 2 Aplastodiscus — 1 — _ — — l Calyptahyla = — x = ++ 1 1 Corythomantis l - — — — l Hyla 99 51 8 63 9 5 235 Nyctimantis l — — — E — 1 Ololygon 49 — 3 — 2 — 54 Osteocephalus 5 l — — — 6 Osteopilus — — _ — — 3 3 Phrynohyas 4 e — — l e 5 Phyllodytes 3 = = ies — 1 4 Plectrohyla — = — 13 — 13 Pternohyla = 1 l — m 2 Ptychohyla =- — 6 — — 6 Smilisca — — — 2 = 6 Sphaenorhynchus 10 -— — — — — 10 Trachycephalus 3 — — = === — 3 Triprion E — 2 — -— — 2 Volume 75, Number 1 1988 Duellma I Anuran Ampnibinh Diversity in American Tropics TABLE 2. Continued.' Genus SAMER SAMER HIGH MAMER MAMER HIGH & SAMER W IND Phyllomedusinae r Ceratophrys Lepidobatrachus Hylodinae Crossodactylus ylodes Megaelosia A iie i Aden Edalorhina H ydrolaetare Leptodactylus Lithodytes Paratelmatobius Pseudopaludicola Vanzolinius Telmatobiinae Adelophryne Atopophrynus Barycholos Batrachophrynus Crossodact ylodes Cycloramphus Dischidodactylus Eleutherodactylus Euparkerella Geobatrachus oloaden Hylactophryne Odonto Scythrophrys Syrrhophus Telmatobius orapa Tomodactylus Zachaenus ji dio gia Arco ^ leas rm | ç | wv | d lele bo — [e] — = = 00-0 | |= a we | "m—— — J] -—— — -J a o RK APONK DK ND N m w — NOW - Qo = O — nar — Q Q NM o — — O 00 — Ñ — tS — m T. Annals of the Missouri Botanical Garden TABLE 2. Continued.' SAMER LOW SAMER HIGH Genus MAMER LOW MAMER MAMER HIGH SAMER Total Ctenophryne Dasypops Dermatonotus Elachistocleis Gastrophryne MATES Glossostoma Hamptophryne Hyophryne Hypopachus Myersiella Otophryne Relictovomer Stereocyclops Sy e s Synco Pelobatiduc Scaphiopus Pipidae Pipa Pseudidae Lysapsus 2 — Pseudis 1 Ranidae ana Rhinophrynidae Rhinophrynus Totals = l 537 537 78 N YU å- DN” -NU BRK — eK | N 64 146 1,545 ' Lowland species are those distributed primarily below 1, above 1,000 m CLOUD FOREST SITES COSN.— Rio Cosnipata, Cuzco, Peru [27 | 13%05'S, 71?18'W, 1,700 m. Wet subtrop- ical forest. Duellman (unpubl. data). UC.—Las Cruces, Puntarenas, Costa Rica [13] 08?48'N, 83%00'W, 1,500 m. Pre- montane rainforest. Scott et al. (1983). PLAN.—La Planada, Narino, Colombia [10] 01?03'N, 77°55'W, 1,700 m. Very wet premontane forest. P. A. Burrowes (pers. comm.). PURU.—Purulha, Baja Verapaz, Guate- mala [9] 15?26'N, 90°20'W, 1,600 m. Very wet subtropical forest. Campbell (1982). ZAP.—Quebrada Zapadores, Pichincha, Ecuador [14] 00?17'S, 78%47'W, 2,010 m. Wet subtropical forest. Duellman (unpubl. data). RGRA.—Rancho Grande, Aragua, Vene- 000 m, whereas highland species are distributed primarily zuela [1] 10?22'N, 67°42'W, 1,100 m. Very wet montane forest. Duellman (unpubl. data). RMES.—Rayón Mescalapa, Chiapas, Mexico [5] 17?12'N, 93%02'W, 1,700 m. Very wet subtropical forest. Campbell (1982). SALA.—Rio Salado, Napo, Ecuador [13] 00°13'S, 77°44’W, 1,410 m. Subtropical wet forest. Duellman (unpubl. data). .—Vista Hermosa, Oaxaca, Mexico [4] 17°51'N, 96°20'W, 1,500 m. Very wet subtropical forest. Campbell (unpubl. data). HIGH ANDEAN SITES ACAN.—Abra Acanacu, Cuzco, Peru [28 | 13?12'S, 71?42'W, 3,250 m. Subalpine plu- vial páramo. Péfaur & Duellman (1980). .—Abra Huancabamba, Piura, Peru AHU [22] 05?22'S, 79?*32'W, 2,800-3,100 m. Volume 75, Number 1 1988 Duellm 87 Fein Amphibien Diversity in American Tropics Very humid montane forest. Duellman (un- publ. data). ANGE.—Paramo El Angel, Carchi, Ec- uador [11] 00?43'N, 77?49'W, 3,350 m. Very wet montane forest. Duellman (unpubl. data). ZAM.—Abra de Zamora, Loja, Ecuador [19] 03°59'S, 79°07'W, 2,700 m. Very wet montane forest. Duellman (unpubl. data). PURA.—Paramo de Puracé, Cauca, Co- lombia [9] 02?19'N, 76?15'W, 3,400 m. Plu- vial montane forest. Duellman (unpubl. data) and J. D. Lynch (pers. comm.). RUSI.— Páramo de la Rusia, Boyacá, Co- lombia [6 ] 05%54'N, 73?12'W, 3,340 m. Very wet montane forest. Duellman (unpubl. data). SROS.— Santa Rosa (4 km W), Puno, Peru [29] 14?36'S, 70°50'W, 4,010 m. Humid montane forest. Péfaur & Duellman (1980). THE TROPICAL AMERICAN ANURAN FAUNA Seventeen of the 46 family groups (families and subfamilies) of anurans occur in the American tropics; nine of these are endemic to the region and three others are extratrop- ical only in southern South America. Of the 301 genera and 3,533 species of anurans known worldwide (figures updated from Frost, 1985), 107 genera (35.5%) and 1,545 species (43.7%) occur in the American tropics (Table 1). Ninety-six of the 107 genera are endemic to the American tropics. Of the 1,545 species known from the American tropics, the majority (1,138) occur in South America; 64 of these also occur with an additional 261 species in Middle America (Mexico and Central America), and 146 species are known from the West Indies (Ta- ble 2). Thus, of the number of anurans known worldwide, 32.2% occur in tropical South America, 9.2% occur in tropical Middle America, and 4.1% occur in the West Indies. Two families make the largest contribution to the anuran fauna of the American tropics. The Leptodactylidae (41 genera, 562 species) and the Hylidae (28 genera, 443 species) account for 72% of the 1,399 species of anurans on the mainland. Six endemic fam- ilies (Allophrynidae, Brachycephalidae, Cen- trolenidae, Dendrobatidae, Pseudidae, and Rhinophrynidae) account for 14% of the mainland species. Four widespread families (Bufonidae, Microhylidae, Pelobatidae, and Ranidae) make up the remaining 14% of the fauna, with the Bufonidae (10 genera, 129 species) making the largest contribution. Two large genera—Hyla (Hylidae, 230 species) and Eleutherodactylus (Leptodactylidae, 308 species)—make up 38% of the anuran fauna of the mainland tropics. PATTERNS OF SPECIES DIVERSITY Patterns of anuran species diversity were determined from compilations of species from each of 48 sites in tropical America (Tables 3-7). For purposes of discussion these are divided into three categories: (1) lowland sites (N — 32) at elevations of less than 1,000 m, (2) montane cloud forest sites (N — 9) at elevations of 1,100-2,010 m, and (3) high Andean sites (N = 7) at elevations of 2,700- 4,010 m. The patterns of species diversity are examined with respect to (1) climate and habitat, (2) taxonomic composition, (3) his- torical components, and (4) reproductive modes. The comparisons reflect existing. knowl- edge of the fauna at each of the sites. Dis- parity exists in the sampling effort at the sites. Thus, for sites that have been subjected to intensive study (e.g., Barro Colorado Island, Santa Cecilia), the species lists can be viewed as nearly complete. However, at other sites (e.g., Rio Cuyuni, Rio Tapajós) sampling has been limited to short periods of time, and therefore additional species are expected. CLIMATE AND HABITAT In the lowlands of the American tropics, the greatest number of species of anurans is found in the equatorial part of the upper Amazon Basin. The greatest number of species known from one locality is 84 at Santa Cecilia in Amazonian Ecuador. Throughout the up- per Amazon Basin in areas receiving more than 3,000 mm of rain per year and having only a limited dry season species diversity is high (55-84, X — 62.8, N — 6). Eastward Annals o Missouri Botanical Garden TABLE 3. Anuran species diversity at 11 lowland sites in Middle America. Wet Forest Genus LSEL RINC TUIR C entrolenella — = 1 Eleutherodactylus l 2 > 1 Tomodactylus = Gastrophryne — -— Glossostoma — — Hypopachus == l Scaphiopus — — 2 Rhinophrynus — = Totals 13 14 in the middle and lower parts of the Amazon Basin, species diversity is lower (32-59, X = = n those parts of the basin rainfall is less than 3,000 mm annually, and there is a prolonged dry season of four to six months. The same climatic pattern exists in the Guianan Region, where anuran species diversity is even lower (29-35, X = 32.0, N= e correlation of decreased anuran species diversity with lower rainfall and more prolonged dry seasons is reflected in the num- ber of species of anurans in the Venezuelan llanos and the Chacoan scrub forests of Par- aguay, where rainfall is less than 2,000 mm annually and the dry season is more than six months long. In the Chaco, species diversity is 22-29 (X = 25.3, N = in the llanos, 16-26 (X = 21.0, N = 2). This pattern seems to hold for the Atlantic coastal forests of southeastern Brazil where at one site with rainfall exceeding 3,600 mm annually there are 65 species of anurans. ...ky Although rainfall is high (over 4,000 mm) and aseasonal in the Trans-Andean rainforest, however, species diversity (35-49, X = 42.6, N = 3) is slightly lower than that in the middle and lower Amazon Basin characterized by less rainfall and a distinct dry season. The same general pattern holds in the low- lands of Middle America. In the wet forests, species diversity is higher (13-42, X = 27.5, N = 6) than in the dry forests (6-22, X = 14.4, N = 5). However, both figures are much lower than for comparable habitats in South America. Although montane rainforests may not re- ceive as much precipitation as many lowland rainforests, the montane forests are shrouded in fog almost daily. Moreover, because of lower temperatures and insolation, evapora- tion is lower. Thus, cloud forests provide a moist environment for anurans. Average an- uran species diversity is higher in some South American cloud forests (20-39, X = 25.4, Volume 75, Number 1 Duellman 89 Anuran Amphibian Diversity in American Tropics 1988 TABLE 3. Continued. Dry Forest MAZA APAT PIST TEHU TABO 5 3 2 4 3 1 l — — l = — — 1 2 1 — — 1 1 1 _ _ = LI 1 1 1 1 2 1 - 1 1 -- 1 1 — 1 — 1 2 2 5 = — — 1 1 ss = — 1 3 T. _ = l — = 1 = _ " 2 — — l — l — — = 1 I a = e: _ 1 1 1 1 2 = — l l 1 17 10 6 17 22 N = 5)than in Middle American cloud forests (16-30, X = 20.5, N = 4). Near the northern extent of both lowland rainforest and cloud forest in Mexico, the numbers of species of anurans are higher in the cloud forests than in the lowland rainforests. On the other hand, species diversity in South American cloud for- ests is considerably less than in lowland rain- forests. The high Andes provide a distinctive en- vironment of supra-treeline habitats (páramo and puno). In these high-elevation sites an- uran species diversity is low (5-15, X — 8.1, N = 7). One of the sites (Abra Zamora, 2,700 m) is subpáramo; it has 15 species, whereas the other higher sites have 5-9 (X — 7.0). Generally in the supra-treeline habitats in the Andes, the number of species of anurans di- minishes from north to south (Duellman, 1979b; Pefaur & Duellman, 1980). This seems to be correlated at least in part with a decrease in the amount of moisture and an increase in the seasonality of precipitation. The elevational gradient in anuran species diversity in the Andes has been discussed by Duellman (1979b, 1983) and by Lynch & Duellman (1980). This gradient is demon- strated well by an equatorial transect span- ning the Andes and the lowlands on either side of the mountains (Fig. 3). Many species in cloud forests have rather narrow altitudinal ranges. Thus, although the number of species (17) at 2,600 m on the Amazonian slopes is not much fewer than that at 1,100 m (22 species), no species occur at both locations. It is obvious that diversity generally de- creases from wetter to drier regions and from lower to higher elevations. However, this ex- planation is simplistic and does not take into account some important aspects of the biology of anurans, such as reproductive mode. TAXONOMIC COMPOSITION Analysis of species diversity by taxonomic groups reveals a variety of patterns. Within the tropical lowlands, three family groups ac- count for most of the taxa at any one site. Of these, the Hylinae and Leptodactylinae are present at all sites. Within the Amazon Basin, hylines are most numerous in the upper part of the basin, but they constitute a higher percentage of the anuran fauna in the middle and lower parts of the basin (Fig. 4). They are numerous in the Atlantic coastal forest of southeastern Brazil, but the hyline fauna is depauperate in the Chocoan and Central American rainforests. Likewise, the number of species of hylines diminishes in the dry forests of Middle America, the Chaco, and the Venezuelan llanos. The leptodactylines present a different pic- ture (Fig. 5). They are well represented throughout the cis-Andean lowlands and form a large part of the anuran fauna in the Ven- ezuelan llanos (38-50%) and in the Chaco (36-44%). Leptodactylines constitute a much smaller part (3-15%) of the anuran fauna in the trans-Andean and Central American low- lands. Telmatobiines contrast strongly with lep- TABLE 4. Anuran species diversity at 16 sites in lowland wet forest in South America. Genus Amazonian PANG TAMB CAMA BALT MANA Allophryne Brachycephalus Atelopus Bufo Dendrophryniscus Rhamphophryne Centrolenella Gastrotheca Hemiphractus l a Nyctimantis Ololygon Osteocephalus Phrynohyas Smilisca Sphaenorhynchus Trachycephalus Agalychnis Phyllomedusa Ceratophrys Crossodactylus nomera Edalorhina Leptodactylus Lithodytes Paratelmatobius Eleutherodactylus oloaden Ischnocnema Phyzelaphryne Proceratophrys Thoropa Chiasmocleis Ctenophryne Elachistocleis Glossostoma Hamptophryne Myersiella Otophryne Synapturanus yncope ““microhylid”” Pipa Lysapsus Rana Totals TABLE 4. Continued. Atlantic Amazonian Guianan Trans-Andean Forest TROM TAPA BELE CUYU OYAP AMPA CALI PALE BORA — — — 1 1 E == um ann a B — — — — ms — l l l 5 — l — = 1 = 3 5 2 2 3 2 4 3 2 l l — l — — 1 = = = = == 1 = = = — l = — 2 8 5 5 = 3 — — — 3 3 2 — l 3 2 l 2 2 1 3 B — = B a — = 2 E = = = = = 1 1 = = = = _ = 1 1 = = 9 14 11 6 7 7 2 4 15 5 6 6 3 l 1 = 3 7 l 2 l 2 5 = -— l l 2 2 l 1 = = — — - = = = l 1 1 ~ = — 1 l = = = B l = = = = — — 1 = - — — z 1 1 1 = l 3 3 4 2 — l — 2 l l — — l — — l = - = — — — — — l - = = - - — — — 2 = = = — E — — = l l 2 l l 1 — — € l 5 3 4 2 1 4 3 - = SR zn - — — — I 2 l = - l 4 -— RR > == a, — "— 1 — — = — — = — — = 3 2 5 l 1 13 13 16 9 = "i = m = = Es = l S 1 ES PS A edi -— PES -— = — - = - - = — 2 = z — — — = = "n 1 > == == 1 em aa paa — == a2 a 1 ETIN i a" ane -— sasa oe eee 1 —— eee — i mack ES = zs e = 1 1 1 = — u = — — — - — l -— "-— — 1 E = ET 2 l 1 — = $e = - a n 2 CE — ns i m" a zd Annals of th e Missouri Botanical Garden TABLE 5. Anuran species diversity at five non-rainforest lowland sites in South America. Genus Llanos Chacoan MANT MASA FILA Bufo Melanophryniscus Hyla Ololygon Phrynohyas Sphaenorhynchus Phyllomedusa Lepidobatrachus Leptodactylus Physalaemus Pleurodema Pseudopaludicola Odontophrynus ;hiasmoclei: Dermatonotus Elachistocleis Pseudis Rana Totals la al TABLE 6. Anuran species diversity at nine montane cloud forest sites in Middle and South America. Genus Middle America South America VHER RMES PURU CRUC RGRA PLAN QZAP SALA COSN Atelopus Buf ufo Centrolene Centrolenella Dendrobates Flectonotus Gastrotheca Hemiphractus Anotheca Hyla Osteocephalus Agalychnis Phyllomedusa Leptodactylus Eleutherodactylus Telmatobius Hypopachus Rana Totals 2 — | Volume 75, Number 1 1988 Duellman 93 Anuran Amphibian Diversity in American Tropics TABLE 7. Anuran species diversity at seven high montane sites in the Andes. Genus RUSI PRUA ANGE AZAM AHUA ACAN SROS ^ iM l 1 1 1 = — Bu — — — — l ^ dM — 1 l — — — Centrolenella == 1 1 1 == = = Colostethus l = 1 1 l === — Gastrotheca — = 1 1 1 2 1 Hyla 1 == + e = — = Pleurodema = = z = s= l 2 Eleutherodactylus 1 5 1 9 3 = = Phrynopus 1 — = 1 2 2 = Telmatobius = = = 1 l 2 l Totals 8 9 15 8 7 5 todactylines (Fig. 6). Telmatobiines (com- posed mostly of species of Eleutherodactylus) make up a significant part (21-35%) of the anuran fauna in the lowlands of the Chocoan region and lower Central America; this is about the same as their contribution (25%) to the anuran fauna of the Atlantic coastal forest in southeastern Brazil. In the Amazon Basin the number of species (and percentage contri- bution) diminishes rapidly from the equatorial part of the upper basin to the southern, mid- dle, and lower parts of the basin. Telmato- biines are poorly represented in the Chaco and are absent from the Venezuelan llanos. Although dendrobatids are represented by far fewer species, their pattern of diversity parallels that of the telmatobiines (Fig. 7). However, dendrobatids are absent from the Atlantic coastal forest in southeastern Brazil, the Chaco, and the llanos. FIGURE 3. Andes showing certain aspects of the anuran fauna with forest mode/percent with stream-dependent mode. T79 T78 "TT Santa Cecilia (H) e " E + 6| Paso de Guamani (E) e , Quito (D) "d e Río Salado (G) Queb. Zapadores (C) e ° ePaopallacta (F) e Río Faisanes (B) e Río Palenque (A) l 1 l Elev. (m) Eu 9/3/| - 4000 7 E 6/2 | 1099 20/16/84 | E2000 24/20/10% € ] B 000 49/29/7 - A Above, map of central Ecuador showing location of sites (A-H); below, equatorial transect of the at each site. Numbers are total number of species/ percent 94 Annals of the Missouri Botanical Garden / : PUT uL E 80 70 60 50 40 T T l T T -]10 HYLINAE 15-911/312 Z D A 21/55% 24/41% ( + + 10} 20|- EN — ANTL TER Kilometers "4 ` “Y EM 3000 Meters | $ 1000 Meters \ 3or +30 \ x y L 80 70 60 50 40 FIGURE 4. of species/ percentage of total at each localit Bufonids are about equally abundant throughout the lowlands, but they tend to account for a higher percentage of the anuran fauna in the llanos (11-13%) and Chaco (3- 9%) than elsewhere. Centrolenids make a large contribution to the anuran fauna in the Cen- tral American (10-17%) and trans-Andean (10-18%) lowlands; they are few in number or absent in the cis-Andean lowlands. Ranids are represented in South America by only one currently recognized species of Rana, which is absent from the southern and eastern low- lands. Pseudids are represented by only three Hyline frogs in lowland sites in South America and lower Central America. Numbers are number uly. species; these are restricted to the middle and lower Amazon Basin, the llanos, and the Cha- co. Pipids are widespread in the cis-Andean lowlands and represented in eastern Panama by a single species. Three family groups have restricted distributions in the American trop- ics. The Allophrynidae occurs only in the Guianan forests. Brachycephalids and hylo- dines are restricted to southeastern Brazil, where they account for 1.5% and 6.2% of the anuran fauna, respectively. In the Middle American lowlands, rhino- phrynids and pelobatids make a minor con- Volume 75, Number 1 1988 uellm 95 Anuran Amphibien Diversity in American Tropics LEPTODACTYLINAE -æ 6/17$ Á 8/213 ]o DA + f, u 320 ` 299 `< -8/36%9 Le @ 12/413 Kilometers a ] 3000 Meters 11/4430 - 1000 Meters ao . ie i h 5 1 | 80 70 60 50 40 FIGURE 5. Leptodactyline frogs in lowland sites in South America and lower Central America. Numbers are number of species/ percentage of total at each locality. tribution to the anuran fauna in dry forests, and hylines and telmatobiines (mostly Eleu- therodactylus) contribute significantly to the anuran fauna of the wet forests. Taxonomic composition changes radically in the montane cloud forest and high Andean sites. A significant part of the anuran fauna in cloud forests is made up of telmatobiines (mostly Eleutherodactylus); these account for 28-56% (X = 38.8, N = 5) of the anuran fauna in South American cloud forests and 12-27% (X = 21.0, N = 4) in Middle Amer- ican cloud forests. The most striking differ- ence between cloud forests in Middle and South America is the number of species of ghe in Middle America these make up X = 44.0, ), as contrasted ea 0-209; (X = 8.2, N = 5) in South America. Centrolenids are a major component of the anuran fauna in South American cloud forests (10-25%, X = 19.0, N = 5), whereas in Central America they account for no more than 10% of the fauna at any one site. Hemi- phractines and dendrobatids each account for an average of 9% of the anuran fauna in South American cloud forests, whereas hemi- 96 Annals of the Missouri Botanical Garden _ 400 — 890 Kilometers 3 3000 Meters 1000 Meters TELMATCBIINAE i ; i 80 70 L 40 FIGURE 6. phractines are absent at all of the Middle American sites, and only one dendrobatid oc- curs at one of the sites. On the other hand, Rana makes up an average of 6.5% of the anuran fauna at Middle American sites, but the genus is absent in South American cloud forests. In humid high Andean sites, telmatobiines (Eleutherodactylus and Phrynopus) account for the majority of the anuran fauna—45- 74% (X = 62.0, N = 5). Bufonids are the other major component, accounting for as much as 22% of the fauna. The rest is com- Telmatobiine frogs in lowland sites in South America and lower Central America. Numbers are number of species/ percentage of total at each locality. posed mainly of dendrobatids (Colostethus) and hemiphractines (Gastrotheca). In the southern part of the Andes, aquatic telma- tobiines (Telmatobius) and leptodactylines (Pleurodema) make significant contributions to the small anuran faunas. From these data it is apparent that the patterns of diversity of various taxonomic components are not reflected in the overall diversity of anurans. This is especially no- ticeable in different patterns of diversity among taxonomic groups in Middle America and South America and suggests that the histories Duellman 97 Anuran Amphibian Diversity in American Tropics Volume 75, Number 1 1988 400 Kilometers [===] 3000 Meters 1000 Meters i DENDROBATIDAE +10 80 ¿Ë 40 FIGURE 7. of the family groups may play an important role in the geographic differences in species diversity. HISTORICAL COMPONENTS An overwhelming amount of data on plate tectonics and biogeography shows that (1) South America had a land connection with Middle America at the time of the Creta- ceous- Tertiary boundary; (2) South America was isolated from other land masses through- out most of the Tertiary; and (3) the isthmian Dendrobatid frogs in lowland sites in South America and lower Central America. Numbers are number of species/ percentage of total at each locality. link providing a continuous land connection between Middle America and South America has persisted since the late Pliocene (Stehli & Webb, 5). The early unification of the land masses provided an opportunity for in- terchange of Laurasian and Gondwanan bio- tas at the end of the Cretaceous. The inter- change, followed by separation of the land masses, resulted in isolation of taxa in South America and Middle America and subsequent vicariance of taxa. Reunification of the lan masses provided the opportunity for more recent dispersal. 98 Annals of the Missouri Botanical Garden TABLE 8. Biogeographical components of the tropical American anuran fauna. NORTH AMERICAN Pelobat anidae Rhinophrynidae MESOAMERICAN Bufonidae (pt) Bufo (pt) Crepidophryne Hylinae (pt) Anotheca Plectrohyla Pternohyla Ptychohyla Smilisca Triprion Phyllomedusinae (pt) Agalychnis Pachymedusa Telmatobiinae (pt) Eleutherodactylus (pt) Hylactophryne Syrrhophus Tomodactylus Microhylidae (pt) Gastrophryne Hypopachus SOUTH AMERICAN Bufonidae (pt Bufo (pt) All other South American genera Centrolenidae Dendrobatidae He All other South American genera Phyllomedusinae (pt) Phyllomedusa Ceratophryinae ylodinae Leptodactylinae Telmatobiinae (pt Eleutherodactylus (pt) All other South American genera a es (pt) All South American genera Pipidae Pseudidae = The history of anurans with respect to the inter-American biotic exchange has been re- viewed by Savage (1982) and by Vanzolini & Heyer (1985). Savage championed the idea of a Mesoamerican herpetofauna that evolved ABLE 9. Biogeographic a leas of anuran faunas at 48 sites in the American tropics. in isolation in Central America. Thus, the tropical American frog fauna is composed of three historical elements—North American, Mesoamerican, and South American (Table pare ontheses indic "ate number of si . Numbe "TS Inc olumns are perc entages Numbers in Sites South American Meso-american North American Middle American lowlands Dry forests (5) Wet forests (6) Middle American cloud forests (4 — South American lowlands Trans-Andean Chocó (3) Upper Amazon Basin (6) Middle and lower Amazon Basin (4) Atlantic coastal forest (1) Guianan forests (2) Venezuelan llanos (2) Chacoan forests (3) — South American cloud forests (5 High Andean sites (7) 20.3 63.6 33.6 59.0 13.6 79.9 100.0 — 100.0 — Volume 75, Number 1 1988 Duellm Anuran Amphibian Diversity in American Tropics TABLE 10. Anuran reproductive modes at 48 sites in tropical America. Numbers in parentheses in the first column indicate the number of sites; in the other columns the numbers are the range and mean percents Sites Forest Stream Equatorial upper Amazon Basin (3) 46.4-47.7 (47.1) 5.9-12.7 (8.3) Southern Amazon Basin (3) 22.5-36.4 (29.2) 1.6-1.8 (1.7) Middle and lower Amazon Basin (4) 18.8-28.8 (24.1) 0.0-8.5 (4.7) Brazilian coastal forest (1) 30.8 23.1 Guianan forest (2) 24.1-25.7 (24.9) 0.0-8.6 (4.3) Venezuelan llanos (2) 0.0-3.8 (1.9) — Chacoan forest (3) 3.4-9.1 (5.5) — Trans-Andean forest (3) 59.2- 74.3 (67.2) 12.2-29.5 (23.4) Lower Middle American wet forest (4) 55.6-60.7 (57.9) 14.3-26.2 (17.5) Upper Middle American wet forest (2) 23.1-35.7 (29.4) — ddle American dry forest (5) 0.0-20.0 (11.6) — Middle American cloud forest (4) 23.5-50.0 (35.9) 33.3-52.9 (45.9) South American cloud forest (5) 61.0-92.0 (76.4) 28.0-43.5 (38.1) Wet high-Andean sites (6) Dry high-Andean sites (1) 57.1-100.0 (81.2) 12.5-33.3 (22.8) 20.0 The North American anurans make only a minor contribution to the tropical anuran fauna. These include pelobatids (two species of Scaphiopus), rhinophrynids (one species of Rhinophrynus), and ranids (21 species of Rana) in Middle America; of these, only one species of Rana occurs in South America. There is no evidence that any North Amer- ican anurans entered South America during the connection with South America at the Cretaceous- Tertiary boundary. On the other hand, during the Cretaceous- early Tertiary connection, members of five family groups dispersed into Middle America from South America. These include some groups of bufonids (Bufo), hylines (Hyla), phyllomedusines (ancestor of Agalychnis and Pachymedusa), telmatobiines (Eleuthero- dactylus), and microhylids (ancestor of Gas- trophryne and Hypopachus). During their isolation in the Tertiary, these groups differ- entiated into many species. The “Hyla” vi- cariant presumably was ancestral to the Mid- dle American genera Anotheca, Plectrohyla, Pternohyla, Ptychohyla, Smilisca, and Tri- prion, in addition to the numerous Middle American and Holarctic frogs currently placed in the genus Hyla. Likewise, the “Eleuthero- dactylus" vicariant presumably was ances- tral to the genera Hylactophryne, Syrrho- phus, and Tomodactylus, as well as the many 66 species of Middle American Eleuthero- dactylus placed in the subgenus Craugastor by Lynch (1986). With the closure of the Panamanian Portal in the late Pliocene, a few groups of Me- soamerican taxa dispersed into South Amer- ica (principally the Chocoan region), and many South American taxa dispersed into Central America. Mesoamerican dispersalists into outh America include only Agalychnis, Smilisca, Eleutherodactylus (members of subgenus Craugastor), and Rana. Only Rana palmipes is widely distributed in the cis-An- dean lowlands, where Agalychnis is repre- sented by one species in the upper Amazon Basin, as contrasted with three species in the trans-Andean lowlands. Twenty-one genera included in ten family groups have dispersed from South America into Central America since the late Pliocene. Two genera of bufonids (some groups of Bufo and Atelopus) have reached Costa Rica, and Atelopus has differentiated in Central Amer- ica. Centrolenella has dispersed to southern Mexico and has undergone extensive specia- tion in Central America. All three genera of dendrobatids have dispersed to Costa Rica Dendrobates to Nicaragua) and have differ- entiated in Central America. Among hemi- phractines, one South American species each of Gastrotheca and Hemiphractus have — 100 Annals of the Missouri Botanical Garden 80 70 60 50 40 T T T 39/15 61/18 TOS {10 N | e 56/15 4 ne 57/26 99/30 j OF 59/12» 5 da Fi Š Nu E Col s E 36/2 /- 10+ "id PU a 29/2 SQ 20} Kilometers EXE 3000 Meters 1000 Meters 30+ 1 i L 80 70 40 Reproductive modes of anurans at lowland sites in South America and lower Central America. iban: are e percentages of species with forest modes/stream-dependent modes. reached Panama, and another Gastrotheca has dispersed to Costa Rica. The hyline gen- era Hyla and Ololygon have dispersed to Mexico and differentiated in Middle America, whereas the South American Phrynohyas venulosa extends throughout the tropical low- lands of Middle America. Phyllomedusa has reached Costa Rica, where a distinct species occurs; another South American species ex- tends into eastern Panama. Although lepto- dactylines are widely dispersed in Middle America, they have not differentiated from outh American populations; one Leptodac- tylus reaches Texas, and four other species have dispersed shorter distances in Middle America. Physalaemus reaches southern Mexico and Pleurodema extends only to cen- tral Panama. South American Eleutherodac- tylus extends northward to Nicaragua. Three outh American species of microhylids have dispersed to Panama and one to Costa Rica. Pipids are represented by a distinct species in eastern Panama. When the anuran diversity at the 48 sites is examined with respect to the histories of the component taxa, we see that slight dif- ferences exist among the South American sites but that these are notably different from the Middle American sites (Table 9). Moreover, North American and Mesoamerican elements Volume 75, Number 1 1988 Duellman 101 Anuran Amphibian Diversity in American Tropics play a minor role in the South American fau- na, whereas at Middle American sites, a sig- nificant contribution is made by South Amer- ican elements. The recent dispersal of South American taxa into Central America accounts for the greater species diversity in lower Cen- tral American lowland forests than in those in Mexico. REPRODUCTIVE MODES Mode of reproduction is a combination of ovipositional and developmental factors, in- cluding oviposition site, ovum and clutch characteristics, rate and duration of devel- opment, stage and size of hatchling, and type of parental care, if any. Duellman & Trueb (1986) recognized 29 reproductive modes in anurans worldwide. At the 48 tropical Amer- ican sites analyzed here, 17 modes occur. These modes and the taxa exhibiting them are: 1. Eggs and tadpoles in lentic water.— All allophrynids, ceratophryines, ranids, and pseudids; most bufonids, hylines, and micro- hylids; telmatobiines (Proceratophrys), and some leptodactylines (Paratelmatobius, Pseudopaludicola, Pleurodema marmora- ta). 2. Eggs and tadpoles in lotic water.— Bu- fonids (Atelopus, Bufo haematiticus group), hylines (Hyla bogotensis, H. callipleura, H. larinopygion, H. circundata groups and most Middle American cloud forest groups, Plec- trohyla, Ptychohyla), hylodines, and tel- matobiines (Paratelmatobius, Telmatobius). 3. Eggs and tadpoles in water in basins constructed by males.—Hylines (Hyla boans group). 4. Eggs and tadpoles in tree holes or bro- meliads.—Some hylines (Anotheca, Nycti- mantis, Phrynohyas resinifictrix, Hyla as- tartea, H. bromeliacia). 5. Eggs in streams; tadpoles on rocks in spray zone.—Telmatobiines (Cycloramphus and Thoropa). 6. Eggs in aquatic foam nest; tadpoles in ponds.— Most leptodactylines. 7. Eggs in terrestrial foam nest; direct de- velopment into froglets.— Leptodactylines (Adenomera). 8. Eggs embedded in dorsum of aquatic female; tadpoles in ponds.—Pipids (Pipa ar- rabali). 9. Eggs embedded in dorsum of aquatic female; direct development into froglets.— Pipids (Pipa pipa). 10. Eggs terrestrial; tadpoles in streams.— Centrolenids (Centrolene). 11. Eggs terrestrial; tadpoles carried to water by adults.—All dendrobatids. 12. Eggs terrestrial; direct development into froglets.—Brachycephalids, bufonids (Osornophryne, Rhamphophryne), telma- tobiines (Barycholos, Eleutherodactylus, Holoaden, Hylactophryne, Ischnocnema, Phrynopus, Phyzelaphryne, Syrrhophus, Tomodactylus), microhylids (Myersiella, Synapturanus, Syncope). 13. Eggs arboreal; tadpoles in ponds.— Hylines (Hyla leucophyllata and H. parvi- ceps groups) and all phyllomedusines except Phyllomedusa cochranae group. 14. Eggs arboreal; tadpoles in streams.— Centrolenids (Centrolenella), hylines (Hyla albosignata and H. lancasteri groups), and phyllomedusines (Phyllomedusa cochranae group). 15. Eggs in dorsal pouch of female; feed- ing tadpoles in ponds. —Hemiphractines (some Gastrotheca). 16. Eggs on dorsum or in dorsal pouch of female; nonfeeding tadpoles in tree holes or bromeliads. —Hemiphractines (Flectonotus, Fritziana). 17. Eggs on dorsum or in dorsal pouch of female; direct development into froglets.— Hemiphractines (Hemiphractus, some Gas- trotheca). Lynch (1979) and Duellman (1982) noted the fidelity of certain reproductive modes to tropical forest environments and designated forest and nonforest reproductive modes. In the foregoing list, numbers 10-17 are con- sidered to be forest modes. In these modes the eggs (and in some cases the tadpoles) are exposed to the air as they develop on the ground, on vegetation, or on the back of the 102 Annals of the Missouri Botanical Garden female. Thus, egg development is dependent upon high humidity. Modes 2, 5, 10, 14, and 11 (in part, Co- lostethus) are dependent on the existence of high-gradient streams for the survival of the tadpoles. Of these, modes 10, 11, and 14 also are forest modes, for the eggs are de- posited on the ground or on vegetation. Analysis of the reproductive modes at the 48 tropical American sites reveals great dis- parities in the proportions of nonforest, forest, and stream-dependent modes (Fig. 8, Table 0). The highest proportions of forest modes are in the equatorial part of the upper Amazon Basin, the trans-Andean forests, the lower Central American wet forests, and the humid montane sites. In Middle American dry for- ests, the Venezuelan llanos, and the Chacoan dry forests, the proportion of forest modes is less than 12%. Reproductive modes depen- dent on streams are most common in the humid montane forests (28-53%), the trans- Andean forests (12-30%), the lower Central American wet forests (14-26%), and the coastal forest of southeastern Brazil (23%). Stream modes are uncommon in the Amazon Basin (6-8%) and absent in the Venezuelan llanos, Chacoan dry forests, and Middle American dry forests. Thus, we find that there is a trend from a high proportion of forest modes in aseasonally wet lowland forests to low proportions (or absence) of these modes in dry forests and llanos. Stream-dependent modes are uncom- mon or absent in most lowland sites, and these make up a larger proportion of the anuran faunas in montane sites (Fig. 3). Likewise, most montane sites have a high proportion of species having forest modes of reproduction Fig. 3 DISCUSSION When all of the factors (environmental, historical, and reproductive) are viewed as a whole in attempting to provide an explanation for the patterns of species diversity in neo- tropical anurans, we find that each factor influences species diversity to varying degrees in different areas. The combined effects of these factors result in the existing patterns of diversity. In the cis-Andean tropical lowlands of South America, Mesoamerican and North American components play a minor role. No taxa of Mesoamerican origin are found there, and one northern group (Ranidae) is represented by a single species, Rana palmipes, at sites pri- marily north of the Equator. Some South American family groups have restricted dis- tributions; thus, Allophryne ruthveni is re- stricted to the Guianan forests, and Brachy- cephalus and hylodines are restricted to the coastal forests of southeastern Brazil. Within the cis-Andean lowlands there is a decrease in annual rainfall and an increase in the seasonality of rainfall away from the equa- torial part of the upper Amazon Basin. This trend is reflected in the diminution in overall anuran species diversity from the equatorial upper Amazon Basin to the lower basin in the east, the Chaco in the south, and the llanos in the north, and a concomitant trend in the reduction in forest modes of reproduction. Thus, in the cis-Andean lowlands, climatic factors seem to have an important effect on overall anuran species diversity. In the cis-Andean lowlands stream-depen- dent modes of reproduction are poorly rep- resented, but in southeastern Brazil, where high-gradient streams are numerous, 23. of the anurans have stream-adapted repro- ductive modes. Compared with other lowland sites, those in the trans-Andean Choco and lower Central America (Costa Rica and Panama) have a high percentage (61.9%) of forest modes of reproduction and also a high percentage (20.0%) of stream-dependent modes. The presence of high-gradient streams provides the necessary habitat for the stream-adapted reproductive modes. Overall diversity at these sites is less than in the upper Amazon Basin, and the high percentage of forest modes of reproduction is correlated with the few species of pond-breeding hylids and leptodactylids as compared with the upper Amazon Basin. The principal differences between the lower Cen- tral American and Chocoan sites are the his- torical components of their anuran faunas. Volume 75, Number 1 1988 Duellm 103 Mrd Amphibian Diversity in American Tropics At the Chocoan sites more than 90% of the taxa are South American, whereas at the Cen- tral American sites 55% of the species are Mesoamerican. One North American species (Rana palmipes) occurs in the Chocó; the same species occurs with two other species of Rana in lower Central America. There is a gradient in decreased annual precipitation and increased seasonality from the lower Central American rainforests to those in northern Central America and Mexico. This is reflected in a decrease in overall anuran species diversity. Furthermore, in the north- ern forests there is an absence of stream- dependent reproductive modes, a decrease to only 7.1% of South American taxa, and an increase to 14.276 of North American taxa. These trends are continued in contrasting Me- soamerican dry forests with the northern rain- forests. In the dry forests there are no stream- dependent reproductive modes, and only 11.6% of the species (all Mesoamerican in origin) have forest modes of reproduction. In the dry forests, 16.176 of the species are North American in origin Comparison of Middle and South American cloud forests with one another and with ad- jacent wet lowland forests reveals that overall diversity is much less in the cloud forests in lower Central America and in South America than in the adjacent lowlands; however, in northern Middle American cloud forests the overall diversity is only slightly less than that in the lowlands. The scarcity of ponds in the montane cloud forests is reflected in the few pond-breeding species of anurans at these sites; this is the single largest group of anurans having low diversity in the cloud forests. In contrast, all cloud forests have a high per- centage of stream-breeding anurans (45.9% in Middle America and 35.9% in South Amer- ica) and species with forest modes of repro- duction (35.9% in Middle America and 76.3% in South America). The major genera having forest modes are Colostethus, Centrolenella, and Eleutherodactylus, and the first two also are stream-breeding taxa. On the other hand, in Middle American cloud forests, the major stream-breeding genera are Hyla, Plectro- hyla, and Ptychohyla. All species in South American cloud forests are members of groups having a South American origin, but in Middle American cloud forests 33.3% of the stream- breeders and 28.3% of the forest modes are species of Mesoamerican origin. No North American groups are represented in South American cloud forests, and North American groups (species of Rana) constitute only 6.5% of the anurans at sites in Middle American cloud forest. All of the taxa at the high Andean sites are South American in origin. The major re- productive modes at wet paramo sites are forest modes (76.3%) or stream modes (38.2%), some of which are included in the forest modes. The anuran fauna of the wet paramos has been derived from that of the adjacent cloud forests, whereas that of the drier puna habitats in the central and southern Andes, which is less diverse and contains about 60% pond-breeding species, has been derived from the Patagonian region (Duellman, Our knowledge of the anuran fauna of trop- ical America is fragmentary. Of the sites in- cluded herein, only those at La Selva, La Planada, Barro Colorado Island, Palenque, Santa Cecilia, and Belém can be considered to have been studied thoroughly enough to provide a reasonably complete list of species and knowledge of their reproductive modes. In order to refine the patterns of anuran species diversity presented here, more data need to be accumulated from many other sites. I have emphasized the role of reproductive mode in understanding anuran species diver- sity. The diversity of modes seems to be cor- related with humidity and with stream gra- dients. Accurate microclimatic measurements and data on the physiology of anuran eggs and larvae are needed in order to quantify and refine these presumed correlations. Much of the anuran fauna in tropical America is disappearing at a rapid rate. The wet lowland rainforest at Santa Cecilia, Ec- uador, that provided data for the largest as- semblage of anurans in the world has been destroyed. Other tracts of rainforest and es- pecially cloud forest are being eliminated be- 104 Annals of the Missouri Botanical Garden fore their biotas are inventoried. Fortunately, many tracts of forest are being set aside as preserves, and data from some of them have been incorporated into this review. However, the establishment of reserves usually is based on the known or presumed existence there of high diversity of birds or primates, with little or no attention given to more poorly known groups such as anurans. LITERATURE CITED CAMPBELL, J. A. 1982. The Biogeography of the Cloud Forest Herpetofauna of Middle America, with Spe- Sierra de Las Minas of Gua- a. Ph.D. Dissertation. Univ. Kansas, Lawrence, as Crump, M. L. Quantitative analysis of the eco- logical distribution of a tropical herpetofauna. Occas. Pap. Mus. Nat. Hist. Univ. Kansas 3: 1-62. DuELLMAN, W. E. 1960. A distributional study of the amphibians of the Isthmus of Tehua uw rom niv. Kansas Publ. Mus. Nat. Hist. 13: The amphibians and ee at Mi- choacán, México. Univ. Kansas Publ. Mus. Nat. Hist. 15: 1-148 Amphibians and reptiles of the rain- forests of southern El Petén, Guatemala. Univ. Kan- sas Publ. Mus. Nat. Hist. 15: 205-249. 1965a. Amphibians and reptiles of the Yu- catan Peninsula, México. Univ. Kansas Publ. Mus Nat. E 15: 577-614. . A biogeographic account of the her- RE Dus of Michoacán, dern . Univ. Kansas Publ. Mus. n Hist. 15: 627- The biology pa an equatorial herpeto- nai in Arnszontan Ecuador. Misc. Publ. Mus. Nat. Hist. Univ. Kansas 65: 1-352. 79a. The South American herpetofauna: a panoramic view. Pp. 1-28 in W. E. Duellman (ed- itor), The South American ala Its Origin, Evolution, and Dispersal. Monogr. Mus. Nat. Hist. Univ. Kansas 7: 1-485. 19 he e herpetofauna of the Andes: pat- terns of distribution, origin, differentiation and com- (editor), The South American Herpetofauna: Its Or- igin, Evolution, and Dispersal. Monogr. Mus. Nat Hist. Un S Kansas 7: 1-485. - Quaternary a ecological fluctua- sification in the Tropics. Columbia Univ. Press, New York. 1983. estes climática cuaternaria en los Andes efectos sobre la especiación. Pp. 177- 20 P. J. Salinas (editor), Zoologico Neotropical, Actas VIII Congreso Latinoamericano de Zoologia, Mérida, Venezue o^ . TRuEB. 1986. Biology of Amphibians. McGraw- Hill Book Co., New York. Frost, D. R. 1985. Amphibian Species of the World. Assoc. Systematics Collections, Lawrence, Kansas. Harpy, L. M. & R. W. McDiarmip. 1969. The am- phibians and ais of Sinaloa, México. Univ. Kan- sas Publ. Mus. Nat. Hist. 18: 39-252. HóbL, W. 1977. Call differences and calling site seg- scenes in anuran species from central Amazonian oating meadows. Oecologia 28: 351-363. HoLDRIDGE, L. R. Life Zone s: Tropical Science vue San Jose, Costa HoocMoED, M. S. & S. J. GORZULA. 1979, Checklist of the savanna a inhabiting frogs of the El Manteco region with notes on their ecology and the asc etd of a new dies of treefrog (Hylidae, Anura). Zool. ded., Leiden 54: 183-216. 976. Contribution à l'étude des amphi- biens de Guyane Frangaise. 6. Liste préliminaire des anoures. Bull. Mus. Natl. Hist. Nat. Paris (Zool.) 265: 475-524 Lynch, J. D. 19 79, The amphibians of the lowland tropical forests. Pp. 189-215 in W. ellman (editor), The South American Herpetofauna: Its Or- igin, Evolution, and Dispersal. Monogr. Mus. Nat. Hist. Univ. Kansas 7: 1-485. . The definition of the Middle American clade of Eleutherodactylus based on jaw muscula- ture (Amphibia: Leptodactylidae). l E EA 42: 258. E. DuELLMAN. 1980. The Eleuthero- dactylus of the Amazonian slopes of the Ecuadorian Andes (Anura: Leptodactylidae). Misc. Publ. Mus. Nat. Hist. pr Kansas 1-86 MYERS, & S. RAN 1969. Checklist of amphibians us reptiles n Barro Colorado Island, Panama, with comments on faunal change and sam pling. lidia Contr. Zool. 10: 1-11. PÉFAUR, J. E. & W. E. DUELLMAN. 1980. Community structure in high eru IÓ Trans. Kan- Acad. Sci. 83: re AGE, AL M. 1982. m enigma of the Central Amer- ican herpetofauna: Lo or vicariance? Ann Missouri Bot. Gard. 69: 464-547. SCHLÜTER, A. : Okologische Untersuchungen an inem Stillgewasser im tropischen Peru i hibien. E Dissertation. Scorr, N. J., . SAVAGE & D. C. Ronson. Checklist of a and amphibians. Pp. 3 in D. H. Janzen (editor), Costa Rican Natural oe Univ. Chicago Press, Chicago. STATON, M. A. & J. The delo uela: noteworthy re ords, a re POE and ecological notes. J. Herpetol. 11: STEHLI, G. & S. » Ns (editors). 1985. The Great American Biotic Interchange. Plenum Press, New or Torr, C. A. & W. E. DUELLMAN. 1979. Anurans of the lower Río Llullapichis, Amazonian Peru: a pre- pei analysis of community structure. Herpeto- ica 35: 71-77. PUE P. E. € W. R. Hever. 1985. The American herpetofauna and the interchange. Pp. 475-487 in G. Stehli & S. D. Webb (editors), The Great Amer- ican Biotic Interchange. Plenum Press, New York MANAGEMENT OF HABITAT FRAGMENTS IN A TROPICAL DRY FOREST: GROWTH' Daniel H. Janzen? ABSTRACT Tropical conservation biology is inescapably the biology of habitat fragments and has been focused on habitat deca are two principal kinds of forest initiation (assuming that there are nearby seed sources). a wind of a relatively intact forest, n initial invasion is primarily by individuals of large wind-dispersed ndreds of years. However, these tree species are a minorit are down trees that will persist and characterize the s ite for hun . Habitat restoration is primarily the initiation, growth, and coalescence of habitat fragments. Management of the total flora. Such forests of plis -dis spersed trees are relatively inhospitable to animals, highly deciduous, n e ara forest patch. Su ch pat a wind- -generated forest. arm an there is any kind o | ed crossing the open a s may grow and coalesce to form a forest type as artificial as is a forests contain more food items of interest to animals, are more vergreen ar of attraction for animals in an abandoned open area, rea. This results in accumulation of an entirely with the difficult decision of just which of the above, or other, ps types is to be promoted. The same will apply to rainforest when its restoration becomes a focus of concern Tropical conservation biology is inescap- ably the biology of habitat fragments. There are two kinds of fragments. First, much of what is worthy of conservation has already been broken into decomposing habitat frag- ments that are refugia and remnants. Even a large national park that is a solid block of pristine forest is a fragment. The biology of the decomposition process of these fragments is of intense contemporary interest to con- servation planners and managers (e.g., Love- joy et al., 1986; Diamond, 1986; Janzen, 1986a, c; Wileove et al., 1986; Uhl & Busch- bacher, 1985; Newmark, 1987). Second, habitat restoration is primarily the initiation and coalescence of growing habitat frag- ments. Management of a tropical wildland therefore becomes the art and science of ar- resting decomposition of habitat fragments and promoting their growth and coalescence. In such an arena, today’s management ac- tions will determine the nature of wildland habitats for centuries to come; forces that determine accumulation of structure and species are significantly within human control. Here I examine the biology of habitat ini- tiation and growth in a Costa Rican tropical dry forest. Dry forest is the most threatened of all the major lowland tropical forest habi- tats. It once covered more than half of the world’s tropics (e.g., Brown & Lugo, 1982; Murphy & Lugo, 1986) but now supports a diverse array of breadbaskets, cotton fields, and pastures. In Pacific Mesoamerica, for example, less than 0.1% of the original trop- ical dry forest, which once covered an area the size of France (equal to five Guatemalas in area), has conservation status, and there ' This and by the Servicio de Parques Nacionales de Costa R W. Hallwa study was supported by NSF BSR 83-07887, ica. The manuscript has been constructively reviewed by chs. ? Department of Biology, University of Pennsylvania, BSR 84-03531, BSR 83-08388, and DEB 80-11558, Philadelphia, Pennsylvania 19104, U.S.A. ANN. MISSOURI Bor. GARD. 75: 105-116. 1988. 106 Annals of the Missouri Botanical Garden TABLE 1. Monthly prec o (rounded to the nearest mm) in the administration area of Santa Rosa National Park, Guanacaste Pro Meteorology Institute in San José). ince, Costa Rica (data collected by park rangers and extracted from the Year Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Total 1980 l 0 5 0 184 175 139 159 331 417 240 9 1,660 1981 0 1 1 11 353 582 172 478 195 268 153 27 2,241 1982 16 2 0 41 919 129 117 34 328 197 37 1 1,820 1983 2 0 22 4 21 180 106 107 188 201 79 7 917 1984 6 8 0 0 118 218 278 162 613 261 52 7 1,723 1985 0 0 0 3 99 211 154 169 214 436 119 26 1,431 are no remaining large areas of even relatively pristine dry forest (Janzen, 1986a). This dry forest area exists almost entirely as habitat fragments and degraded patches, all of which are still shrinking if not in conserved areas. f there is ever to be a Pacific mesoamerican dry forest national park large enough to truly maintain the animals, plants, and habitats that greeted the Spaniards in the early 1500s, it will have to be grown rather than decreed (Janzen, 1986a). e study site is the 108-km? Santa Rosa National Park and its immediate surround- ings, the site of the proposed 800-km? Guana- caste National Park (GNP) (Janzen, 1986a). GNP is an approximate rectangle of variably damaged dry forest in northwestern Costa Rica, lying between Volcán Orosi and Volcán Cacao on the east and the Pacific Ocean to the west, with the Interamerican Highway cutting through the middle. From the ocean to the tops of the 1,500-m volcanoes is a moisture and elevational gradient that ranges from the driest part of Costa Rica (the sea level tip of the Santa Elena Peninsula, with less than 1 m of rainfall during its 5-6-mont rainy season, Table 1) to seasonal rainforest on the volcano sides to ever-wet cloud forest on the volcano tops. The dry forest discussed here grows primarily within 10 km of the Administration Area of Santa Rosa National Park and is a mosaic of pristine forest, 0— 400-year-old woody succession, and aban- doned pastures of an introduced African pas- ture grass (jaragua, Hyparrhenia rufa). The anta Rosa dry forest contains about 700 species of plants (Janzen & Liesner, 1980), 115 species of mammals (Wilson, 1983), 170 species of resident birds (Stiles, 1983), and 13,000 species of insects (including about 3,000 species of moths and butterflies, Jan- zen, 1988a, b). The essay presented here is based on pro- cesses observed within the area of the pro- posed Guanacaste National Park. Space does not permit review of the pertinent literature from other habitats, but this will be done in a later publication. WHAT ARE THE TRAITS OF THE Two PRINCIPAL KINDS OF HABITAT FRAGMENTS? DECOMPOSING HABITAT FRAGMENTS As logging, burning, and clearing for pas- tures and agriculture have whittled away at Guanacaste Province’s dry forest in north- western Costa Rica (e.g., Uhl & Buschbacher, 1985; Janzen, 1986a), the large and multi- habitat dry forest expanses have been reduced to fragments ranging in size from a few hundred hectares to areas containing only a single remaining tree. Species richness in these fragments is decaying to substantially lower equilibrium densities than once existed, but many decades will pass before a stable equi- librium level is reached (if ever). The rate and depth of decay depends on at least three major variables: . How directly prejudicial is the forest fragmentation process to the organisms in the remaining fragments? This aspect of frag- mentation cannot be separated from the sim- ple effects of reducing the size of the habitat (and thus the sizes of the constituent popu- lations). Except in experiments (e.g., Lovejoy Volume 75, Number 1 1988 Janzen 107 Habitat Fragments in Tropical Dry Forest et al., 1986) and massive logging or resettle- ment schemes, habitat is almost never re- duced in area as if a giant cookie-cutter had cut out a small relict piece in one fell swoop. For example, the fragmentation and decom- position of dry forest by repeated fires has a quite different effect on its species composi- tion than if some of the forest is left standing and some is bulldozed for cotton fields that are then allowed to change to brushy pastures. 2. Are there other areas that continually generate incoming individuals? In other words, is the population of any given species in a habitat fragment maintained by internal re- cruitment, colonists, or both? Incoming in- dividuals may originate in the same or dif- ferent kinds of habitat as the recipient habitat (e.g.. Janzen, 1986e). 3. What proportion of the species have what population decay rates and individual life spans? A tree species whose individuals have a 400-year life span may persist in a habitat fragment long after that fragment can support a breeding population of the species, while an insect population might well disap- pear within weeks of the time when its pop- ulation no longer achieves recruitment. Whether the observer is left with a feeling of pending doom as a dry forest is increasingly fragmented depends in great part on which kinds of organisms are censused as indicators. Agroecosystems based on what once was trop- ical dry forest are especially rich in the living dead—individuals living out a physiological life span but no longer occupying a habitat in which recruitment occurs (Janzen, 1986b). GROWING HABITAT FRAGMENTS The second kind of habitat fragment is very different from the shrinking relict mentioned above. In the long view of tropical conser- vation, growing habitat fragments are ex- tremely important and complex; while shrink- ing habitats may even be accepted as they stand, growing habitats require management technology and choices. As more tropical res- toration projects take root, there will be more efforts at initiating habitats, reconstructing habitats from fragments, and even reinstalling species from distant habitats and geographies. Restoration will often initiate growing habitat fragments that accumulate species richness to equilibrium levels as yet unknown. As they grow and coalesce, the three variables men- tioned above will generally apply, but with different relative importance. . The selection of ecological processes e.g., fire, hunting, grazing) to be stopped will determine the species richness, life forms, and habitat structure. Stopping hunting but allow- ing fire to continue yields a very different kind of habitat fragment than does allowing hunting but stopping fires. Stopping anthro- pogenic fires has a very different effect than does stopping natural fires. While the site may initially have had an equilibrium array of species, for all practical purposes and for many centuries it can sustain many different kinds of habitats during the restoration phase. Which habitat occurs will depend on the initiation and management processes. 2. Proximity of source areas for incoming species, as measured by the ability of indi- viduals of species to arrive, establish, and recruit once present, will be very important in determining whether a growing fragment quickly or slowly rises to an equilibrium value. Since different groups of organisms have ex- tremely different abilities to colonize, and since the chance of establishment in a habitat de- pends at least on what species have colonized earlier and the environmental traits of the habitat, I expect the rate of species buildup to be very irregular, with long periods of relative stasis. 3. Species will appear in a fragment long before they have attained a density at which their population can survive through recruit- ment and/or persistence within the habitat. That is to say, the species richness of a grow- ing habitat fragment can easily be greater than the equilibrium density for a fragment of that size and set of conditions. Study of the decay of species richness with- in a fragment requires either actual or infer- ential knowledge of reference or benchmark habitat. This is becoming extremely difficult in dry forest because this forest type has already been thoroughly fragmented for de- es 108 Annals of the Missouri Botanical Garden cades or even centuries. On the other hand, there are numerous opportunities to study species accumulation in growing habitat frag- ments, even if there are not enough species in the region to completely restore the habitat. [INITIATION AND GROWTH OF HABITAT FRAGMENTS The manner of initiation and growth of habitat fragments in Santa Rosa's dry forest is a novel process in the evolutionary and ecological history for these habitats and for this fauna and flora (just as it would be for other tropical restoration projects). though much of Santa Rosa has been cleared to bare ground by volcanic activity in the past, habitat restoration (primary succession) on such a site is different in many ways from the secondary succession that occurs when habitats are initiated and grow in old fields, abandoned pastures, or repeatedly burned forest. However, be they novelties or other- wise, understanding the biology of growing fragments is critical to survival and mainte- nance of Santa Rosa's tropical dry forest. Growing habitat fragments in Santa Rosa are most commonly initiated in. abandoned pastures. Even if land is lumbered or cleared for fields, it is usual for such areas to be converted to pastures before the restoration process is allowed. Most Santa Rosa pastures are almost pure jaragua (Hyparrhenia rufa) with a sprinkling of native herbs, thoug few pastures of native grasses remain on the poorest, driest, and/or rockiest soils (e.g., Santa Elena Peninsula and Cerro El Hacha), and pastures of other introduced grasses oc- cur on the moister volcanic slopes. The aban- doned or lightly used pastures in Santa Rosa range from grass patches of less than a hect- are to several km? in area. Abandoned pastures initially range from virtually lacking trees and treelets to quite rich in sucker shoots from old and large root systems. Santa Rosa pastures are also dotted with the occasional seedling or sucker shoot from a sapling tree or treelet. The numbers, sizes, and kinds of large woody plants in a £5 newly abandoned pasture depend on its age, frequency and timing of burning, timing of pasture “cleaning,” proximity to forest (e.g., Purata, 1986), kind of livestock, stocking density, and numerous other variables (just as is the case in extratropical invasion of abandoned fields and pastures by forest, e.g., Olsson, 1984, and included references). All of the initially present woody plants are po- tential starting points for dry forest habitat initiation (see “nuclear trees" discussed below) and add to the composition of a growing hab- itat fragment without having to arrive by dis- persal. Expanding and filing habitat fragments range from being totally isolated to being ex- pansions of an existing forest into an adjacent pasture. Placement of these fragments is not haphazard and often has clear relationships to fire patterns, wind direction, animal avail- ability, soil moisture and contour, kinds of nearby forest, and other factors. TWO PRINCIPAL KINDS OF HABITAT INITIATION Almost all fires were stopped in abandoned pastures in Santa Rosa's dry forest through an active fire control program initiated in the 1984 dry season. All modern fires in the Santa Rosa area are set by humans or arrive as free-running wildfires that were set by hu- mans. In some of the park pastures, fires even stopped at the time of park establishment in 1972 (or later, 1972-1983) because they were sufficiently isolated that only deliberately set fires got to them. There are even some old pastures or pasture remnants in the park that have not been burned for many decades because woody succession isolated them from the pastures that traditionally burned. As a general rule in Santa Rosa, when woody succession has proceeded to the point where the overstory canopy shades out the grasses and herbs, the forest does not carry a fire though there are special circumstances, to be described elsewhere, in which enough dry forest can burn to initiate a return to grass- land). — Volume 75, Number 1 1988 Janzen 109 Habitat Fragments in Tropical Dry Forest When livestock were finally removed from Santa Rosa in 1978-1979, an important eco- logical process was altered. Livestock had maintained the jaragua at a low level through heavy grazing. Grass fires had relatively little fuel, and a conspicuous equilibrium was main- tained between forest/pasture edges; in dry years or when the fire was late in the dry season, it nibbled away the margins of the forest. In wet years or when there was an early burn, the forest margins moved slightly into the pasture. When the cattle were re- moved, the jaragua formed dense stands l- 2 m in height within a single rainy season. These stands shade woody plant seedlings and provide enough fuel to support very hot and thorough fires at any time in the dry season (such stands even support fires during droughts in the rainy season). This initiated a rapid decline in the area and quality of dry forest fragments surviving in the abandoned pas- tures and on their margins. This decline stim- ulated the fire control program initiated in the 1984 dry season. The two processes of forest initiation de- scribed below are occurring in fire-free and cattle-free contemporary Santa Rosa. Were the park to be returned to either cattle-rich, fire-rich, or cattle- and fire-rich status, the details of these processes would undoubtedly be different. Habitat establishment through seed dispersal by wind. In many of Santa Rosa's abandoned pas- tures free of cattle and fire, a distinctive set of woody species has appeared as the forest moves into the pasture. At least 9075 of the trees and treelets have wind-dispersed seeds. The vertebrate-dispersed initial colonizers, such as Cecropia obtusifolia, Piper spp., Tre- ma micrantha, Muntingia calabura, Spon- dias mombin, and Bursera simaruba, are almost entirely absent. For example, the old- est explicit regeneration plot (CT Regenera- tion Plot, on soils derived from a flat, volcanic welded ashflow) in the park was last swept by fire in the 1979 dry season (it was a jaragua pasture that had been frequently burned pas- ture of some kind for several hundred years). Today (end of the 1986 rainy season), the following 12 species constitute at least 90% of the 5,000-plus individuals and biomass of trees and treelets in the 3-ha plot (all trees of all sizes in the plot are registered and mapped): Ateleia herbert-smithii, Acosmium panamensis, Lysiloma auritum, Dalbergia retusa (Leguminosae); Rehdera trinervis (Verbenaceae); Cordia alliodora (Boragina- ceae); Luehea speciosa (Tiliaceae); Swieten- ia macrophylla (Meliaceae); Tabebuia rosea and T. ochracea (Bignoniaceae); Cochlo- spermum vitifolium (Cochlospermaceae); and Hemiangium | excelsum (Hippocrateaceae). All plants in this plot arrived by their own means, and the vegetation was not manipu- lated. All these species have wind-dispersed seeds. However, Santa Rosa has a native tree and treelet flora of 215 species (Janzen & Liesner, 1980) of which 25% are wind dispersed; 64% are vertebrate dispersed. More than 100 species of vertebrate-dispersed trees and tree- lets maintain breeding populations within 200 m of the CT Regeneration Plot. It is clear that habitat initiation in this plot is based on an ecologically distinctive subset of the species pool (cf. Beyer, 1975, for a curiously similar extratropical example). There is one major reason why vertebrate- dispersed species are scarce in this restoration site. Among the many seed-dispersing ver- tebrates in the park, only white-tailed deer (Odocoileus virginianus) and coyotes (Canis latrans) encounter resources in a pure stand of jaragua. A deer sometimes uses it for cover when resting and spits out Spondias mombin (Anacardiaceae) and Byrsonima crassifolia (Malpighiaceae) nuts when chewing its cud (Janzen, 1985). Coyotes hunt there for cotton rats, Sigmodon hispidus, and defecate tree seeds such as Manilkara chicle (Sapotaceae) and B. crassifolia. Other vertebrates that disperse trees and treelets (e.g., collared pec- caries (Tayassu tajacu), coatimundis (Vasua narica), magpie jays (Calocitta formosa), and tapirs (Tapirus bairdii)) only defecate in the pasture when they happen to cross it on their 110 Annals of the Missouri Botanical Garden way from one forest to another; in addition to the two plants listed above, these animals have dispersed a few individuals of Entero- lobium cyclocarpum and Acacia collinsii (Leguminosae), Genipa americana and Ali- bertia edulis (Rubiaceae), and Annona re- ticulata (Annonaceae) into the CT Regen- eration Plot. This particular abandoned pasture did not have a tree that served as a nuclear tree (see below) though the single old Acrocomia vinifera palm could have served that role. Also, animal-rich forest only occurs along one side of the plot and so there is little reason for animals to cross it. There are many places in Santa Rosa where patches of wind-generated forest have re- cently been initiated as major protrusions into abandoned pastures. Examination of these forest fragments (or peninsulas) over the past few years allows a number of generalizations: 1. Not only is an abandoned jaragua pas- ture initially unattractive to seed-dispersing animals because the grass usually does not represent food or other resources, but the growing forest made up predominantly of wind- dispersed trees offers no food for frugivory (by monkeys, bats, coatis, tayras, tapirs, pec- caries, deer, numerous birds) compared with a forest fragment containing a more balanced mix of seed dispersal types. Since the wind- dispersed trees are first to colonize the site, they physically and numerically dominate it for many decades as they live out their life spans. While the site does offer shade and some foliage and seeds, it still lacks a major class of food. . The wind-dispersed initial colonizers are large trees (attaining heights of 15-25 m) and live O to several hundred years. There is no wave of early successional, short- lived, and comparatively small species. It is even tempting to suggest that a wind-dis- persed colonizing tree has to be tall in dry forest, or it won't be tall enough to have its seeds blow over the canopy and into tree falls and other gaps in the canopy. Put another way, in a climatic regime that grows a tall forest, only canopy-level or emergent trees (and high-climbing vines) will successfully col- onize that forest via wind-dispersal or evolve wind-dispersal. 3. Not only does a wind-dispersed dry for- est lack fruit, but all its seeds are well pro- tected against vertebrate postdispersal seed predators. The species with big seeds in con- spicuous fruits or other dispersal units (e.g., Lonchocarpus spp., Acosmium, Swietenia, Hemiangium, Ateleia) have very toxic seeds (e.g., Janzen, 19864). Those with edible seeds (e.g., Cochlospermum, Rehdera, Luehea) have very small seeds that easily escape by self-burial in the litter. 4. As the first wave of wind-dispersed trees reaches maturity and begins bearing seeds, the dispersal process in the fragment does not change. By contrast, in a site that is invaded by vertebrate-dispersed tree species, the pat- tern of the dispersal process will change as the trees come into reproductive maturity and begin to attract animals. Some of these ani- mals (e.g., agoutis (Dasyprocta punctata)) play no part in initiating the forest fragment. This attraction will increase the rate of ac- cumulation of plant species in the habitat fragment. 5. The majority of the large trees in Santa Rosa (many with fruits eaten by vertebrate dispersal agents) require at least as much sunlight as is made available when a tree falls to grow into saplings (the evergreen Manil- kara chicle is the only obvious exception). Once the habitat fragment is occupied by large wind-dispersed trees, the species com- position of the canopy will not significantly change its composition until these wind-dis- persed trees begin to die of senescence (and create tree falls) many decades later, even if there is a very heavy flow of vertebrate-dis- persed seeds into the site during intervening years, 6. A wind-dispersed forest does not ex- pand into a pasture from all compass direc- tions. A pasture on the downwind side of a forest of mixed dispersal types receives a heavy inoculum of wind-dispersed seeds, but a pas- ture upwind of forest receives none. This pro- cess is especially critical to restoration efforts if the upwind margin of the pasture borders Volume 75, Number 1 1988 Janzen Habitat Fragments in Tropical Dry Forest on a park boundary with agricultural land, ocean, or other nonforest habitat. Verte- brate-dispersed seeds in a pasture may have come from any direction. In Santa Rosa, the dry season winds blow almost invariably from northeast to southwest; rainy season winds blow in many different directions, but there are no wind-dispersed trees or treelets that mature their seeds at this time (see 10 below). 7. The only parents of wind-generated fragments in pastures are trees near the for- est-pasture interface. Even a few tens of me- ters of forest thoroughly filter out wind-borne tree seeds (e.g., Augspurger, 1986). On the other hand, the vertebrate-dispersed seeds falling into a growing habitat fragment in an abandoned pasture may have passed through forest expanses hundreds to thousands of me- ters wide. 8. Wind-dispersed seeds at Santa Rosa move up to about 200 m into pastures, but a more common distance is 0-100 m. The seeds move greater distances if the parents are maximally tall individuals that have grown up in competition with other forest trees (i.e., near forest margins) rather than shorter iso- lated trees in open pastures. This means that wind-generated growing habitat fragments will always be near another forest with adult trees in it. 9. Wind-generated seed shadows are often dense and very uniform compared with ver- tebrate-generated seed shadows (which have conspicuous heterogeneity associated with perches, waterholes, trails, shade, steep con- tour, tourist presence, and other factors). If there is one member of a wind-dispersed species of tree maturing in a growing habitat fragment, there are likely to be numerous others (unique individuals of vertebrate-dis- persed trees are commonplace in growing habitat fragments, see below). 10. Forests of wind-dispersed species bear no ripe fruits or seeds throughout the rainy season (no Santa Rosa species of wind-dis- persed tree or treelet ripens its fruits in the rainy season). Associated with this, the wind does not blow with equal intensity and pattern throughout the year. The highest average wind velocity occurs during the first half of the dry season (December through early March). The days with the strongest wind occur in late December and January. During the dry sea- son the most intense winds occur during the central part of the day (when it is driest and fruits are most likely to be shed or broken off, and fruits are most likely to dehisce). During the rainy season, there are many near- ly calm days, and winds often occur during rainstorms (when a wind-dispersed fruit or seed would have little or no chance of dis- persal). Another barrier to wind-dispersal dur- ing the rainy season is that wind-dispersed units of dispersal lose weight at maturity by drying out, which does not occur readily in the rainy season. Along the same lines, almost all wind-dis- persed seeds at Santa Rosa appear to ger- minate within a few weeks of being wetted by the first rains. There is a major pulse of seed- ling appearance with the first rains (from seeds accumulated during the dry season) followed by no further seedling input or site coloni- zation during the six-month rainy season. Ver- tebrate-dispersed seeds in Santa Rosa display many kinds of dormancy and new seedlings of many species appear at different times during the rainy season. Since many of these seeds are dispersed during the rainy season, a habitat fragment can be initiated at this time of year from dispersing seeds or from seeds in the soil. Habitat establishment through seed dispersal by vertebrates. Vertebrates play a much more complex role than does wind in initiation and growth of habitat fragments. This is because: a) There are nearly three times as many species of vertebrate-dispersed than wind-dis- persed trees and treelets in Santa Rosa, and they bring with them more life forms and kinds of ways to occupy a site than are found among the wind-dispersed species. b) There are many more kinds of verte- brate-generated seed shadows than wind-dis- persed seed shadows in Santa Rosa. c) The pattern of forest initiation depends 112 Annals of the Missouri Botanical Garden on the biology of the animals as well as on the biology of the plants. The wind has not evolved and coevolved with plants, and the plants can only evolve, not coevolve with the Here, I focus on one particular kind of vertebrate-generated forest initiation in Santa Rosa pastures, that which is associated with nuclear trees. A nuclear tree is a single tree that appears in a pasture and is then attractive to animals that defecate or otherwise drop seeds in its vicinity (see McDonnell & Stiles, 1983; Beck- with, 1954; Debussche et al., 1982, for or- nithological extratropical examples). The an- imals go to the tree as a perch, for shade, to eat its fruits, to mark ranges, and other rea- sons. Whether a nuclear tree and its asso- ciated seed rain actually becomes a growing fragment of forest habitat depends on nu- merous environmental variables ranging from the rate and kind of seed rain to the depth of the grass, soil conditions, species of nuclear tree, fire regime, and other factors. For ex- ample, many of the best contemporary ex- amples of vertebrate-generated growing hab- itat islands in Santa Rosa are centered on guanacaste trees (Enterolobium cyclocar- pum) on small rocky outcrops that protected the sapling from low grass fires (in the heavily grazed pastures). Within a few years after cattle were removed from Santa Rosa, the high grass fires swept across these small frag- ments and obliterated many of them and their nuclear guanacaste trees. Representative nuclear trees in Santa Rosa are guanacaste, cenizaro (Pithecellobium sa- man, Leguminosae), guacimo (Guazuma ul. mifolia, Sterculiaceae), jobo (Spondias mom- bin, Anacardiaceae) and figs (Ficus spp., Moraceae), but many other species can serve this function (some are remnant wind-dis- persed trees standing as the sole survivors of a dwindled forest fragment). A growing dry forest fragment that has been generated by vertebrates around a nu- clear tree in the center of an abandoned pas- ture has a number of distinctive traits: Wind-dispersed species of trees and treelets are absent (unless they were present as persistent suckers at the time of fragment initiation). That is to say, the species pool rom which the fragment is potentially drawn is some subset of the 138 species of verte- brate-dispersed trees and treelets in Santa osa. 2. Certain vertebrate-dispersed species are missing because they are usually dispersed by vertebrates that either never cross pastures or are very unlikely to carry one of these seeds when they do. For example, agoutis and monkeys are very unlikely to carry the large seeds of guapinol (Hymenaea courbaril, Le- guminosae), panama (Sterculia apetala, Sterculiaceae), and tempisque (Masticho- dendron capiri, Sapotaceae) across a pas- ture. 3. The species and age of the nuclear tree will influence the species composition of ar- riving seeds. If it bears edible fruits (which is very likely since wind-dispersed nuclear trees are almost nonexistent in large pastures), the species composition of the seed rain during fruiting will be strongly influenced by the as- sortment of animals attracted to its fruit. Ad- ditionally, different species of trees offer dif- erent kinds of perches and shade, which likewise will influence animal visitors and their seed cargoes. 4. The species and age of the nuclear tree will influence the survival of the seedlings below it through differential shade effects on grasses and woody seedlings, protection of seedlings from fire through shading out grass- es and keeping the soil moist, and the duration of its deciduousness during the dry season. 9. The trees accumulating around a nu- clear tree are usually smaller and shorter than the nuclear tree because they are younger than it and because they compete with it. The canopy is therefore dome-shaped with the nu- clear tree near the center rather than flat- topped (as is the case in wind-generated hab- itat fragments). 6. The understory of a vertebrate-gener- ated habitat fragment often contains repro- ducing treelets and shrubs, e.g., Chomelia spinosa, Alibertia edulis, Psychotria spp. Volume 75, Number 1 1988 Janzen 113 Habitat Fragments in Tropical Dry Forest (Rubiaceae); Piper spp. (Piperaceae); Ery- throxylon havanense (Erythroxylaceae); Malvaviscus arboreus (Malvaceae), Hirtella racemosa (Chrysobalanaceae); Ximenia americana (Olacaceae); Casearia spp. (Fla- courtiaceae); Ocotea veraguensis (Laura- ceae); and Annona reticulata (Annonaceae). Reproducing small plants are generally miss- ing from the understory of a wind-generated forest fragment. 7. Owing to the larger species pool and area from which the species are drawn, species richness of trees and treelets is substantially greater in growing vertebrate-generated hab- itat fragments than in wind-generated ones. However, if there is frequent animal move- ment through a wind-generated habitat frag- ment, the sum of the wind-dispersed species and the animal-dispersed species gives the most species-rich habitats of all. 8. Except for the very large seeds that are not carried out into a pasture (see 2 above), all sizes of seeds arrive at a vertebrate- generated habitat fragment. However, there should be some sorting with distance as frag- ments get so far out into a pasture that small frugivorous birds are less likely to go to them. In contrast, a wind-generated growing forest fragment should also display sorting whereby the species with lighter seeds move further downwind in the initial invasion. 9. A vertebrate-generated habitat frag- ment can grow in any direction, rather than downwind as does a wind-generated habitat fragment; and a vertebrate-generated frag- ment can range from a distant island to a peninsula, while a wind-generated fragment is usually a peninsula from an established orest. 10. The seeds in a vertebrate-generated habitat fragment may come from as far as many thousands of meters from the fragment, and from any compass direction. The tree species in the fragment are thus likely to represent a much greater portion of their respective populations than is the case with a wind-generated patch (which is derived from the immediately adjacent forest). Associated with this process is the fact that vertebrate- generated habitat fragments often contain iso- lated representatives of a species; whether these individuals can reproduce (through long- distance pollination or by being self-compat- ible) varies with the species. However, it is quite likely that such species occasionally ex- ceed the species carrying capacity of the frag- ment, since if it does not maintain itself in the fragment, it is not present in some senses. 11. Since there are some vertebrate-dis- persed tree species in fruit throughout the year in Santa Rosa, there are likewise trees and treelets in fruit throughout the year in the vertebrate-generated growing habitat fragments. As a habitat fragment grows by ver- tebrate input of seeds (and by internal re- cruitment), its margins become less likely sites of seed deposition. The rate of expansion of a habitat fragment noticeably declines as it becomes more than about 0.5-1 ha in area. DISCUSSION I have briefly described two prominent kinds of forest habitat initiation and growth in aban- doned pastures in Santa Rosa's dry forests. There are also many other kinds of habitat initiation occurring in this park at present. It is evident that the initial processes in habitat initiation. will determine the species compo- sition and structure of the forest for many centuries on these sites. It is equally evident that the actions of park managers will deter- mine the kinds of habitat initiation that occur. If the park is left completely to itself, one kind of mosaic will occur. If livestock are used to depress the grass in certain areas, a different kind of mosaic will occur. If nuclear trees are planted far out into pastures or seeded there in the dung of cattle (as occurs with guanacaste and cenizaro trees if the cat- tle have access to fruit crops), a third kind of mosaic will occur. If native timber trees are seeded in by air or hand, and then natural succession is allowed to continue (as an ex- ample for those who want to know how to return worthless dry forest pastures to timber production), the resultant forest will depend 114 Annals of the Missouri Botanical Garden in great part on the particular species of tim- ber trees used (fleshy fruit-bearing versus wind-dispersed, evergreen versus deciduous, fast- versus slow-growing). Pastures near for- ests rich in seed-dispersing mammals and birds will initiate quite different kinds of forest than will those in areas relatively free of verte- brates due to poaching, heavy tourist use, or experimental removal. It is clear that the managed fate of the entire area of tens of km? of pasture in Santa Rosa or several hundred km” of pasture in Guanacaste National Park should not be the same from area to area. But whatever kinds of management of habitat fragment initiation and growth are to be applied, it will require basic research into the multiple pathways that secondary succession can follow in this species- rich situation. Whether these pathways will finally converge on a single kind of vegetation for any given site is almost irrelevant, since this will take thousands of years to occur and we have to deal with the situation at hand. How broadly applicable is the philosophy espoused in this essay? Rainforest restoration has not been explicitly tried, but it is an in- tegral part of traditional shifting agriculture in some areas and has occurred throughout the wet tropics where farms and homesteads have been abandoned when farmers have been relocated out of an area slated to become a rainforest national park. The details of rain- forest restoration will obviously be different from those in dry forest, but there will be many similarities. ne difference is apparent even at this early stage of understanding. When a clean pasture is abandoned next to a pristine rain- forest, the rate of rainforest tree and treelet movement into it appears to be much slower than in seemingly equivalent cases in dry forest. This is the case even in the wetter eastern end of Guanacaste National Park as compared with the central drier part. There are at least three possible causes. First, rain- forest vertebrates may be even more reluctant to enter rainforest pastures and use nuclear trees, for example, than are dry forest ver- tebrates. Second, the general lack of strong directional winds in rainforest may impede movement of wind-dispersed species into pas- tures. Third, and I suspect the most likely case, a seedling from a dry-forest seed dis- persed into a dry-forest pasture finds itself in a much more foreign environment than does a seedling from a rainforest seed dispersed into a rainforest pasture. The heat, sun, wind, dryness, and fluctuations of an open pasture are much more similar to the rigors of dry forest than they are to the understory of a rainforest (even in a tree fall). Additionally, dry-forest pasture soils are rich in spores o endomycorrhizal fungi while rainforest pas- ture soils are generally very poor in them (D. Janos, pers. comm.). Certainly at Santa Rosa there is no indication that the physical con- ditions of open pasture habitat are lethal to tree seedlings, even when they are species commonly associated with relatively moist habitats (e.g., Hymenaea courbaril, Manil- kara chicle). Blydenstein (1967) described the mosaic of Venezuelan grazed pasture and forest patches as the result of a dynamic interaction between dryness and fire, just as occurred in Santa Rosa before the fire and cattle were eliminated. In some ways, this dynamic is a fine-scale version of the global balance be- tween the dwindling tropical wildlands and restoration systems such as that occurring in Santa Rosa National Park and the proposed Guanacaste National Park area. At present we are in a continuous drought and fire phase, with the forest patches shrinking very, very rapidly. If the dynamic is to have any sig- nificant balance, forest restoration is going to have to start up throughout the tropics. It must start not only in places like Guanacaste National Park, where remnants of the plants, animals, and habitats are still present, but also in the large expanses of the tropics where there is virtually nothing left. For example, in countries like Costa Rica, virtually all rain- forest parks are on hilly to mountainous soils, with the true flatlands either cleared or almost cleared. The real challenge is whether with fencerow populations, woodlot populations, and living dead (Janzen, 1986f) one can put Volume 75, Number 1 1988 Janzen 115 Habitat Fragments in Tropical Dry Forest back together even a first approximation of the flatland rainforests that once stood on the enormous acreages of low-yield rainforest fields and pastures. The question has been raised as to why Guanacaste National Park needs to be as large s 800 km?, especially since about 90% of the habitat has been partly homogenized through partial destruction. The traditional answer is that a large area is needed to main- tain even minimal populations of the wide- ranging large animals and the big trees. A less traditional, but nonetheless compelling, answer is that the area must be large to main- tain intact wildlands and be extremely ac- cessible to tourists, researchers, managers, and school groups; these are the people whose votes and long-term use will keep GNP alive. But there is also a less conventional biological reason that is especially pertinent to dry for- est. Intact dry forest is a very complex mosaic of many small fragments of habitat types nes- tled in amongst each other, generated by soil type, drainage, wind exposure, slope, and rainfall. This mosaic of conditions and species is largely responsible for the many pathways that may be taken by secondary succession in an abandoned pasture; two pastures with identical conditions may easily be positioned next to two quite different source areas for the species that will colonize it. A given animal or plant species exists as a population sup- ported by one or more of these habitat types, but the support base is highly fragmented. A dry forest wildland must be big so that the total area contains enough pieces of a given habitat type (e.g., dry ridges) to support healthy populations of the species that are supported by that habitat type. Yes, there are 800 km? of dry forest in GNP, but, for ex- ample, less than 100 km? supports the en- demic population of the tree Ateleia herbert- smithii, the only legume in the world that is known to be wind-pollinated (Janzen, 1988c). This worry applies to rainforest as well. Yes, it all appears to be just a sheet of green, but as soon as it is necessary to find any tree species, or any constellation of species, the biologist learns to search for the subtle dif- ferences in drainage, soil, and other factors that lead to the highly localized target. When the professional conservationists tell us that X% of the rainforest still remains, and that Y% of X is disappearing daily, it is imperative to remember that 0% of many of the habitat types of tropical rainforest remain. Further- more, that Y% is not spread proportionately over the remaining types, giving us an equal amount of time to crusade for what remains. It is critical to identify those rainforest habitat types that are as threatened as is the tropical dry forest and focus restoration projects on them. The battle for mesoamerican tropical dry forest conservation should have been fought in the year 1800; don’t wait until the year 2000 to begin to fumble with the rain- forest pieces worth restoration. LITERATURE CITED AUGSPURGER, C. 1986. Morphology and dispersal po- tential of wind-dispersed diaspores of neotropical trees. Amer. J. Bot. 73: 353-363. BECKWITH, S. L. 1954. Ecological succession on aban- doned farm lands and its relationship to wildlife man- agement. Ecol. Monogr. 24: 349-376. Bryer, W. N. 19 Types of seed dispersal: their effects on species diversity of trees. Amer. Nat. 109: 103-104. BLYDENSTEIN, J. 1967. Tropical savanna vegetation of the llanos of Colombia. Ecology 48: 1-15. Brown, S. & A 1982. The storage and production of organic matter in tropical forests and their role in the global carbon cycle. Biotropica 14: -187. DEBUsscHE, M., J. EscanRE & J. Lepart. 1982. Or- sag was: and plant succession in Mediterranean abandoned orchards. Vegetatio 48: 255-266. DIAMOND, 1986. The aea of a nature reserve system for Indonesian New Guine . 485-503 in M. J. Soulé (editor), pte: Biology. Si- nauer, a Massachusetts. JANZEN, D. 1983. No park is an island: increase in nrference from outside as park size decreases. Oi- kos 4 410. E Sp ondias mombin is culturally T prived in MUN: Ue free forest. J. Trop. Ecol. 1: 131-155. 8 Guanacaste National Park: Tropical Ecological and Cultural Restoration. Editorial Uni- versidad Estatal a Distancia, San José, Costa Rica. 1986b. The future of dim ecology. Annual Rev. Ecol. Syst. 17: 305-3 986c. The eternal Must threat. Pp. 286- 303 in M. E. Soulé (editor), E Biology. Sinauer, Sunderland, Massachusetts. 1986d. Mice, big ppc and seeds: it matters who defecates what where. Pp. 251-271 in 116 Annals of the Missouri Botanical Garden A. Estrada & T. H. Fleming (editors), Frugivores and e Dispersal. W. Junk Publishers, Dordrecht. — . Lost plants. Oikos 46: 129-131 : Er The future of e ecology. Annual Rev. Ecol. Syst. 17: 305-324 1988a. Ecological characterization of a Costa Rican dry forest caterpillar fauna. Biotropica (in press). 1988b. Biogeography of an unexceptional place: what determines the saturniid and sphingid moth fauna of Santa Rosa National Park, Costa Rica, and what does it mean to conservation biology. Bre- nesia press). 988c. Natural history of a wind-pollinated e. American dry forest legume tree (Ateleia herbert-smithii Pittier). Legume on St. Louis, Missouri, June 1986 (in pre . LIESNER. 1980. rennet check-list of plants of lowland Guanacaste Province, Costa Rica, exclusive of grasses and non-vascular cryptogams. Brenesia 18: 15-90. . S. Martin. 1982. Neotropical anach- ronisms: the fruits the gomphotheres ate. Science viens: 1984. A seasonal census of phenolics, fibre and alkaloids in foliage of forest gidae and Saturniidae. Biol. J. Linn. Soc. 21: 439- 454. Lovejoy, T. E., R. O. BIERREGAARD, JR., A. B. RYLANDS, J. MALCOLM, C. E. QUINTELA, L. H. HARPER, K S. Brown, JR., A. PowELL, G. V. N. POWELL, H. O. R. ScuuBART & M. B. Hays. 1986. Edge and other effects of isolation on Amazon forest frag- ments. Pp. 257-285 in M. J. Soulé (editor), Con- servation Biology. Sinauer, Sunderland, Massachu- setts. McDonneLL, M. J. & E. W. Stites. 1983. The struc- tural complexity of oldfield vegetation and the re- cruitment of bird-dispersed plant species. Oecologia 56: 109-116. Murpny, P. G. & A. E. Luco. 1986. Se SURE dry forest. Annual Rev. Ecol. Syst. 17 NEW MARK, W. SA 1987. A land- bridge a ieee malian extinctions in western North American parks. Nature 325: 430-4 Olsson, G. 1984. Old Field Forest Succession in the Swedish West Coast SM Ph.D. Dissertation. University of Lund, Sweden. Purata, S. E. 1986. Floristic and . changes during old-field succession in the Mexican tropics in relation to site history and species maliy J. Trop. Ecol. 2: 257-270. 1983. Checklist of birds. Pp. 530-544 D. H. Janzen (editor), V Rican Natural History. 1985. A disturbing syn- ergism between cattle ranch burning practices and selective tree harvesting in the eastern Amazon. Bio- tropica 17: 265-268. McLELLAN & A. E DoDsoN. 1986. Habitat fragmentation in the te on . J. Soulé (editor), p suu aruwa 83. Checklist of mammals. Pp. 44 7 in D. H. Janzen (editor), vu Rican Natural History. Univ. Chicago Press, Chicago. FACTORS CONTROLLING Jared Diamond' SPECIES DIVERSITY: OVERVIEW AND SYNTHESIS ABSTRACT Factors controlling — diversity are usually presented as a laundry list without organization—for instance, cies than the temperate zones be ecause hcl stability and reduced seasonality and high aper instead proposes a fourfold grouping of factors, termed the QQID iaie Q = resource quality, consisting of the habitat and resource factors that determine the * ‘number of niches." Q = resource and consumer antity, consisting of factors determining the er of consuming individuals exti immigration, and speciation), which affect species diversity in both equilibrial pe “mansana situations. l and latitudinal gradients. A major problem for the future involves seeking generalizations as to which factors h circumstances. Another problem is to convert this a grouping into a natural hierarchy of factors, possibly based on a hierarchy of processes in space an INTRODUCTION: THE QQID APPROACH This article aims to provide an overview of the many factors controlling species di- versity. My minimum goal is to construct a simple, four-step checklist for analyzing the determinants of species diversity in any given case (Table 1). I shall also briefly suggest an approach by which this checklist, designed for practical purposes, might be converted into a natural hierarchy of the determinants of species diversity. My examples will be drawn mostly from the other papers of this sym- posium. At the outset, let us be clear why the prob- lem of understanding species diversity is com- plicated. In the first place, species diversity is surely not determined in all cases by the same single factor but is the outcome of many contributing factors. Secondly, while one can formulate “rules” about species diversity, each rule has many exceptions. For example, island species diversity usually increases with island area, but there are more frogs on little Barro Colorado Island than on the much larger Cuba (Duellman, this volume). Island species di- versity generally decreases with distance from the mainland, but one of the world’s most remote archipelagoes, Hawaii, has more species of Drosophila than do continents. Species diversity as one ascends a mountain generally decreases with altitude, but along the western slopes of the Andes above Chile’s Atacama Desert it increases from middle to high elevations (Arroyo et al., this volume). Small-bodied species are generally more di- verse than large-bodied species, but whale diversity exceeds insect diversity in the open ocean. These exceptions to rules based on single factors arise in part for the obvious reason that species diversity is the outcome of many factors, so that the effect of one factor may be overridden by others. Partly, too, the reason is that some of the determi- nants of species diversity, such as predation, herbivory, disturbance, seasonality, and en- vironmental predictability, control diversity in a nonmonotonic way, so that an increase in those factors may yield either an increase or a decrease in diversity. Do the factors controlling species diversity just constitute a laundry list, a catalog without organization? Discussions of the latitudinal ' Department of Physiology, University of California Medical School, Los Angeles, California 90024, U.S.A. ANN. MISSOURI Bor. GARD. 75: 117- 129. 1988. 118 Annals of the Missouri Botanical Garden TABLE l. The QQID formulation of factors controlling species diversity. z = Quality . Diversity of niches or resourc abi tat structural vidcre" habitat diversity, resource species diversity, temporal variability, and dics of consumer strategies. Q — Quantity 2. Number of consumer individuals, N where R= available quantity of resourc R, — resource requirements per didus (increases with body size) A = area P = productivity per unit area I = Interactions 3. Species interactions Effects on Increase of N Predation or herbivory on competing species Decrease of N Resource competition Predation or herbivory on the consumer itself Effects on individual fitness Increase of fitness Mutualism Decrease of fitness Interference competition Parasitism Disease D = Dynamics 4. Species dynamics Effects of dynamics at equilibrium E ffects of extinction rates (decrease with N, etc.) Effects of immigration rates (increase with proximity, vagility, larval settling rates, etc. Effec Nonequilibrium situations Pulse disturbance Transiently depressed species diversity Recent pulse decrease in area Transiently elevated species diversity = ts of speciation rates (decrease with vagility, etc. gradient in species diversity often read that way: “The tropics have more species than the temperate zones because of greater sta- bility and reduced seasonality and higher productivity and more diverse resources n fact, life is full of multi-deter- ed: phënamena (e.g., whether to declare war, whom to marry, how many children to bear) that at first appear to be influenced by innumerable factors, but for which the factors actually prove to fall into Just a few groupings. In some cases, the groupings may even define a matural hierarchy or decision path of fac- tors. For example, suppose you want to know what determines the number of children in any given candy store. The number obviously depends on the quality or variety of candy to be seen in the window. It is also heavily in- fluenced by the quantity of candy or the size of the store. A little reflection also shows that the number of children in the store depends on interactions, such as those with attractive fairy godmothers beckoning at the door, or with bullies or hungry lions inside. Finally, there is also some role of dynamic or non- equilibrium factors, such as how long ago the doors opened, or how large is the pool of Volume 75, Number 1 1988 iamond 119 Factors Controlling Species Diversity children available for colonizing the store, and where the store is (particularly relative to a school). Thus, the numerous determinants of the number of children in the store sort out into four sets of considerations: quality of resources, quantity of resources, interactions, and dynamics. It seems to me that species diversity as well is determined by the same four sets of factors—quality, quantity, interactions, and dynamics (Table 1). As a simple mnemonic, I shall abbreviate this accounting as “QQID Q. Discussions of species diversity dub: start out with an analog of the quality or variety of candy: namely, the diversity of niches or of resources. Hence my first, sim- plest approach will consider just the diversity of niches without taking into account the quantity of resources or the number of con- sumer individuals or species interactions. We shall pretend that the environment is constant and without temporal variation, that the world is at equilibrium, and that species dynamics are nonexistent or irrelevant. Q. Next, I shall take account of the quan- tity of resources, which partly determines the number of consumer individuals. I. Thirdly, I shall add consideration of species interactions. D. Finally, I shall incorporate consider- ation of dynamics and shall allow for the pos- sibility of nonequilibrium. Q: RESOURCE QUALITY The ecological equivalent of the variety of candy is the number of niches or of resources. One of the basic findings of ecology is that species diversity increases with niche or re- source diversity, as expressed in the state- ment that each species must occupy a distinct niche. This ubiquitous generalization is the outcome of two facts. First, any given ge- notype can do certain things (e.g., harvest certain resources by a certain method) better than other things. Second, a single gene pool (i.e., one species) can comprise only a certain diversity of genotypes because of the con- straint that the individuals carrying those ge- notypes must be reproductively inter-com- patible in order to remain members of the same gene pool. The combination of those two facts has the consequence that each species occupies a certain “niche,” however defined, and that species diversity increases with niche or resource diversity. Many familiar predictors of species diver- sity fall under this heading of niche diversity. Among the many examples of predictors that could be cited, I shall discuss five that are well illustrated by the papers of this sympo- sium. Habitat structure. It is commonly found that habitats with a more complex or variegated structure contain more species than do simpler habitats. Thus, within any given group of taxa there are gen- erally fewer species in a rock desert than in an adjacent grassland, fewer in the grassland than in an adjacent savanna, and fewer in the savanna than in an adjacent tropical rain- forest. The symposium papers by Gentry, Er- win, and Duellman emphasized that peak di- versities of plants, beetles, and frogs are achieved in tropical rainforest. For birds a rough quantitative measure of habitat struc- tural complexity that serves to predict species diversity is the habitat's foliage height diver- sity: that is, a diversity measure of how foliage is distributed among different vertical layers of the habitat (MacArthur et al., 1966). Habitat diversity. Another expression of niche specialization is that a particular species tends to occur only in certain habitats and not others. Thus, as one proceeds along a habitat gradient, one accumulates more and more species, and the accumulated number of species increases with the diversity of habitats encountered. Famil- iar examples are that species accumulate as one goes along an elevational gradient on a mountain, a depth gradient in the sea, or a horizontal sequence of habitats on land or in the intertidal zone. One example from this symposium is that most neotropical wet forest tree species are confined to a single forest 120 Annals of the Missouri Botanical Garden type or soil type (Gentry, this volume). Another example is that 83% of the beetle species at Manaus, Brazil, are similarly confined to a single forest type (Erwin, this symposium). Species themselves as niches or resources. Species may constitute niches or resources for exploitation by other species. Thus, di- versity of consumer species tends to increase with diversity of resource species. For in- stance, there are more stenophagous herbiv- orous beetle species in tropical rainforest than in a temperate woodland because the rain- forest has many more tree species. An ex- ample is that many neotropical rainforest bee- tle species are confined to a single tree species; some may even be confined to the interface of a particular pair of tree species (Erwin, this symposium). Temporal variability. Time also serves as a niche dimension that can be partitioned, so that temporal variability provides opportunities for differentiation ab- sent in an environment that is constant with time. For example, the 24-hour solar cycle permits differentiation between nocturnal, crepuscular, and diurnal species exploiting similar resources in the same habitat. The annual cycle permits species to specialize by adopting various seasonal strategies, as ex- emplified by the coexisting insect species with different overwintering strategies, or by bird species that coexist in the breeding season but segregate in winter as a result of some being migratory, others resident. Still longer cycles in a variable environment permit the differentiation of K strategists from r strat- egists. A striking example from this sympo- sium is the 97% turnover of beetle species in a single tree species of Barro Colorado Island between the wet season and dry season (Erwin, this symposium). Consumer strategies. Species that harvest similar resources in the same habitat may coexist by employing different foraging techniques, or by adopting differing life-history strategies. Here, too, this symposium has provided a striking example: the 28 alternative reproductive modes by which frogs solve the common problem of producing offspring while protecting them from desiccation (Duellman, this volume). These five sets of examples do not exhaust the axes along which coexisting species may segregate. Readers will undoubtedly be able to think of further biologically significant ways in which differences in resources or in other niche parameters are exploited by different species. Q: QUANTITY OF CONSUMER INDIVIDUALS OR OF RESOURCES In the preceding discussion we have ig- nored consideration of the quantity of re- sources. We have pretended that a resource is either present in sufficient quantity to sup- port a species, or else the resource is absent. However, resource quantity is obviously im- portant because species are packaged in dis- crete units (1.e., individuals), and a population consisting of too few individuals cannot sur- vive. There is no hard rule as to how many constitutes “too few," but a population con- sisting of one individual of a species practicing sexual reproduction is clearly doomed to ex- tinction within one generation, a population size of two (one male, one female) is extremely precarious, and only populations with an ef- fective size above 500 are considered rea- sonably safe even in the short run (Frankel & Soule, 1981; Soulé, 1986). Thus, the species diversity of any given group of taxa generally increases with the group’s total pop- ulation size. However, population size, the number of consumer individuals, does not depend only on quantity of resources. More generally, population size equals the total quantity of resources available, divided by the quantity of resources required to sustain one individ- ual. Resource quantity in turn equals the eras of area times productivity per unit area, while quantity of resources required per individual increases with body size. Thus, the second “Q” in our *QQID" formulation Volume 75, Number 1 Diamond 121 Factors Controlling Species Diversity groups three factors: the increase in species diversity with area, productivity, and decreas- ing body size. Area. The most familiar generalization of island biogeography is that species diversity on is- lands, mainland habitat patches, or arbitrarily defined mainland census plots increases with area. This species/area relation arises partly from the increase in habitat diversity with increasing area sampled, but also from the direct proportion between area on the one and and resource quantity and thus consum- er population size on the other hand. An ex- ample from this symposium is Janzen's (this volume) comment that conservation areas in Costa Rican dry forest should be at least 500- 1,000 km? in extent, because smaller areas would contain too few individuals of important species to sustain their populations. Productivity. Productivity increases with increasing rainfall or temperature, hence with decreas- ing latitude or altitude. Since number of con- sumer individuals increases with productivity, species diversity also increases with produc- tivity. This symposium has provided three clear examples. First, plant species diversity on the western slope of the Andes rising out of the Atacama Desert of Chile increases with rainfall (Arroyo et al., this volume). Second, neotropical plant species diversity, collective- ly or else of each life form considered indi- vidually, increases with rainfall up to an asymptotic value (Gentry, this volume). Fi- nally, neotropical rainforest beetle species di- versity is much higher in the forest canopy than in lower vertical strata, because the can- opy intercepts most of the solar energy and is the most productive stratum (Erwin, this symposium). (Note that these arguments im- plicitly assume a more or less uniform in- crease in productivity across the resource spectrum. If production of only certain re- sources is increased, as in a eutrophic pond, the outcome may be reduction rather than increase in species diversity, because those consumer species specializing on the in- creased resource type will thrive and may eliminate other species by preempting their resources.) Body size. The product of area times productivity equals the total quantity of available re- sources, but the body sizes of the consumers determine among how many individuals those resources may be apportioned. Thus, for a given resource quantity and hence given con- sumer biomass (ignoring second-order effects from the variation in metabolic rate per gram of tissue with body size), consumer population size decreases with body size. Hence, all other things being equal (which they often are not), there tend to be more species of small-bodied animals than of large-bodied animals. In the canopy, Erwin (this symposium) encountered thousands of species of beetles but not of elephants. The whole neotropical region con- tains only 1,545 species of frogs (Duellman, this volume), a number exceeded by the beetle species in a single tree canopy (Erwin, this symposium). Éven among beetles, species di- versity is highest in Erwin's smallest size class of beetles. All these examples illustrate that species diversity increases with the quantity of re- sources and, more generally, with consumer population size. QQI: SPECIES INTERACTIONS In our discussion of resource quality and quantity so far, we have ignored species in- teractions and have implicitly lumped all re- source species together simply as “food.” However, species interactions may boost or lower species diversity in comparison with the value that one would predict by ignoring species interactions. Some of these effects of species interactions on species diversity are mediated by effects of species Interactions on population numbers (the sec of “QQID”), while other effects require instead consideration of individual fitnesses. 122 Annals of the Missouri Botanical Garden Effects on number of individuals. A familiar example of how species inter- actions may lower species diversity is that certain consumer species may competitively lower diversity of other consumer species at the same trophic level by preempting re- sources and hence lowering the population sizes of their competitors. An example of re- source competition on a gigantic temporal and spatial scale is that the evolutionary history of vascular plants has involved a parade of successive dominants, starting with the rhy- niophytes and proceeding through pterido- phytes and gymnosperms to the angiosperms. The rhyniophytes disappeared completely, but the pteridophytes and gymnosperms continue to survive today, albeit represented by many fewer species than formerly. The most likely explanation is that each new evolving group of vascular plants preempted resources that would otherwise have been utilized by pre- viously evolved groups, thereby reducing their numbers of individuals and consequently of species, or even driving them to extinction (Knoll, 1986; Niklas, this volume). Species interactions may have the opposite effect —boosting species diversity by boosting the quantity of available resources, these sup- porting greater numbers of consumer indi- viduals. This situation arises when predators or herbivores reduce the numbers of individ- uals of their prey or plant species, thereby making more resources available for other consumer species at the same trophic level and thus increasing the numbers of individuals and hence species diversity of those other consumers (Paine, 1966). However, preda- tors or herbivores can also reduce species diversity of consumer species by greatly re- ducing numbers of individuals. Thus, preda- tion and herbivory can either increase or de- crease species diversity, depending on circumstances such as the intensity of pre- dation or herbivory. By analogy with the in- termediate disturbance hypothesis, one might speculate that a community's species diversity initially increases with increased community- wide intensity of predation or herbivory, then decreases with further increase in intensity. (The intermediate disturbance hypothesis pro- poses that low levels of physical disturbance also increase species diversity by removing some consumer individuals, thus reducing competition for resources, while an increase in disturbance that more severely reduces consumer population sizes decreases species diversity. (See Yodzis (1986) for further dis- cussion.) Effects on individual fitness. The species interactions of interference competition (e.g., physical aggression), par- asitism, and disease tend to decrease species diversity by decreasing individual fitness. Conversely, the species interaction of mu- tualism tends to increase species diversity by increasing individual fitness. Effects on fitness merge into effects on population size as the effects on fitness become strong enough to kill individuals or else to permit them to sur- vive where they otherwise could not. QQID: Dynamics Our discussion so far has ignored dynamic considerations. The final step in our QQID analysis takes species dynamics into account. rst retain the implicit assumption of equilibrium that we have made up to this point. Even at equilibrium, consideration of dynamics predicts trends in species diversity that one could not interpret without consid- ering dynamics. We shall then relax our as- sumption of equilibrium and thereby encoun- ter still further trends in species diversity. There are three dynamic processes underly- ing species diversity: extinction, which tends to decrease species diversity; immigration, which tends to increase it; and speciation, which also tends to increase it. Effects of dynamics at equilibrium. Effects of extinction dynamics. All else being equal, the probability (per unit time) of extinction increases with decreasing popula- tion size, hence with decreasing area. That inverse dependence of extinction rates on area provides the major reason why species num- ber on islands increases with area. In this instance, dynamic considerations do not pre- Volume 75, Number 1 Diamond 123 Factors Controlling Species Diversity dict a new trend in species diversity that we have not already considered; instead, they provide the basis of a trend that we had al- ready noted. Effects of immigration dynamics. The second rule of island biogeography, after the species/area relation, is that species diversity tends to be higher on islands close to a col- onization source than on distant islands, even if the close and distant islands are identical in area and in resources. In this case the phenomenon cannot be discussed at all with- out reference to dynamic considerations: im- migration rates from the mainland source to a nearby island are higher than to a distant island, with the result that equilibrium species diversity is higher on the nearby island. Another consequence of immigration dynam- ics is that species groups with high immigra- tion rates (high dispersal ability) are repre- sented on islands by a higher fraction of the mainland species pool than are species groups with low immigration rates. Thus, compared with mainlands, oceanic islands have more species of birds and bats than of flightless mammals. As a final example of the effects of immigration dynamics on species diversity, high larval settling rates increase the diversity of barnacles, coral reef fish, and other marine organisms with planktonic dispersal (Rough- garden, 6) Effects of speciation dynamics. Any- thing that increases speciation rates will tend to increase species diversity. Speciation rates depend on numerous factors, such as fre- quency of chromosomal rearrangements, ease of developing reproductive isolation, and dis- persal rates. I shall provide a few examples involving dispersal, which is important in spe- ciation since reproductive isolation is more likely, even over shorter distances, for taxa with poor dispersal ability than for others with great dispersal ability. In part for that reason, there are many more species of flightless bee- tles and land snails than of tardigrades, which are readily wafted in the aerial plankton and are virtually panmictic and cosmopolitan. Be- cause tropical species of birds and possibly of other taxa tend to be more sedentary than temperate species, considerations of dispersal and speciation rates also contribute to the higher species diversities in the tropics. Note that dispersal has opposing effects on species diversity: with increasing dispersal, the frac- tion of the regional species pool that reaches a given site increases (increasing the species diversity at that site), but the frequency of speciation and thus the size of the regional species pool itself decreases. This symposium provided several examples of the sensitivity of species diversity to dis- persal through its effects on immigration and speciation rates. As an example of the effect on immigration rates, the plant diversity and composition of Costa Rican dry forests depend on the relative opportunities for seed dispersal by wind and by animals (Janzen, this volume). As an example of the effect of dispersal on speciation rates, the roles of biotic vectors both for pollination and for seed dispersal were a decisive factor in the diversification of an- giosperms (Niklas, this volume). Biotic vectors can carry out pollination between conspecific individuals separated by a much greater dis- tance than can be effectively bridged by wind, thus permitting angiosperms to live at much lower population densities than other plants and hence to evolve high diversities of rela- tively rare species. Biotic seed dispersal per- mits angiosperms to reach sites accessible only with greater difficulty to other plants. Nonequilibrium situations. All out discussions so far have referred to species diversity at equilibrium. However, it is a debated question whether it is frequent or exceptional for species communities to be at equilibrium. There is no doubt that many communities have species diversities below equilibrium values, while other communities have species diversities above equilibrium values (Janzen, this volume). Subequilibrial diversities are a transient result of pulse dis- turbances, while supraequilibrial species di- versities are a transient result of pulse de- creases in area. Pulse disturbances leading to subequi- librial species diversity. If populations or resources are decreased or wiped out by dis- 124 Annals of the Missouri Botanical Garden turbance at a site, species diversity at the site will transiently be below the equilibrium value until the resources are restored, or until the consumer populations are restored by immi- gration or by speciation. The time constants or relaxation times for species communities to recover from disturbance vary enormously. It may be a few months or years after a storm batters a coastline until resources have been replenished and consumer species have re- turned. When a volcanic explosion destroys the biota of an island, as happened on Kraka- tau, it may be decades or centuries before immigration has restored the original species diversity. As a result of Pleistocene glaciations that backed populations of many northern European tree species against the Alps and exterminated them, tree species diversity in Europe today, 10,000 years after the end of the Pleistocene, is still below North American levels. It will presumably take much longer than 10,000 years for European tree species diversity to be restored by a combination of immigration and speciation. Finally, for many millions of years after an asteroid collision caused mass extinctions at the Cretaceous/ Tertiary boundary —if indeed there was such a mass extinction, and if it was caused by an asteroid — species diversity of large terrestrial vertebrates remained low until it was even- tually restored and surpassed by speciation of mammals. Pulse decrease in area, leading to supra- equilibrial species diversity. equilibrial species diversity tends to increase with area, a decrease in area eventually leads to a decrease in species diversity. However, the greater the area after the pulse decrease, the slower the “‘relaxation time" required for species diversity to decay to the new lower equilibrial value (because "relaxation time" depends on extinction rates which are in- versely proportional to area according to Dia- mond, 1972). Immediately after the pulse decrease in area, species diversity equals that prevailing at the site immediately before the pulse decrease in area. If one looked at the site immediately after the pulse area decrease and did not know that there had been such Because an area decrease, one would be puzzled to find species diversity higher than the site would support if it had not just suffered such a pulse area decrease. This “supersaturation effect” lasts only for a century or so in the case of birds in a forest fragment of one km?, but lasts for many millenia for birds or mammals on large land-bridge islands such as Java and Borneo, or on a large mountaintop with Pleis- tocene habitat connections to other moun- taintops, such as the mountains rising out of the Great Basin (Brown, 1971 iamond, 1984). Islands that lie today in shallow water near continents were connected to those con- tinents at Pleistocene times of lower sea level by land bridges and were finally severed from those continents 10,000 years ago by rising sea levels. The larger land-bridge islands, which include Trinidad, Sri Lanka, Fernando Po, and Formosa, as well as Java and Borneo, are still supersaturated with bird and mammal species (Terborgh, 1974). It will presumably require many tens of thousands of years be- fore their species diversities have declined back to the equilibrial values appropriate to their modern areas. These considerations of how relaxation times of supersaturated habitat fragments in- crease with area have interesting implications for understanding continental biotas. Relax- ation times for birds and mammals on islands of a few thousand square kilometers, and for insects, lizards, plants, and other species living at higher population densities than birds and mammals on still smaller islands, are in excess of 10,000 years. We must therefore expect that relaxation times for the world's conti- nents are far longer, perhaps hundreds of thousands of years. As the continental tropical rainforests expanded in concert with Pleis- tocene climatic fluctuations, rainforest species diversity must also have tended to expand and contract. However, the expanses of rain- forest in South America, Africa, or Asia are so large that species relaxation times for the rainforest biota may be longer than the in- terval between Pleistocene climatic optima. Thus, when the continental rainforests con- tracted during dry periods of the Pleistocene, Volume 75, Number 1 1988 Diamond 125 Factors Controlling Species Diversity the rainforest biotas may still have been su- persaturated at the time when the next wet phase arrived. Species diversity on the major continents may never have a chance to de- cline to *equilibrium values" and may be chronically supersaturated. e decreases in species number after hab- itat fragmentation, and their relaxation times, are of great significance in the worldwide ex- tinction spasm that is now under way. This accelerating extinction wave is due partly to the habitat fragmentation and reduction in habitat area that humans are producing by destroying natural habitats. The habitat frag- ments thus created start off with their pre- fragmentation species diversity and are grad- ually losing populations at rates that depend on their area. We have already launched a process that, if it is not miraculously reversed, must result inevitably in a massive extinction wave, even though the wave itself has not yet reached massive proportions. Some econo- mists ignorant of biology question those ex- tinctions that have already occurred, note that massive extinction has not yet occurred, and on this basis belittle predictions of an impending extinction spasm. This reasoning reminds me of the story of the man who fell off the top of the Empire State Building and who had a friend working on the 20th floor. The worker on the 20th floor looked out the window, saw his friend plunging past, and shouted out in concern, “° od, what is happening?" to which the falling man shouted back as he plunged past, “Nothing much is happening, everything is okay so far." As stewards of the world's biota, we have already pushed most of the world's species off the top of the Empire State Building. Those who deny the impending extinction crisis demand to see bodies smeared on the pavement before they will discuss erecting a safety net. SPECIES DIVERSITY GRADIENTS ALONG ENVIRONMENTAL GRADIENTS Discussions of species diversity often focus on the famous changes of diversity over en- vironmental gradients, such as habitat gra- dients, altitudinal gradients, and especially latitudinal gradients. All too often, ecologists seek to identify “the cause" of such a gra- dient. We should be suspicious of any such attempt. Since species diversity depends on many factors, diversity changes over such gradients are also likely to arise from gradi- ents in multiple controlling variables. What we should seek instead is to provide a quan- titative partitioning or accounting to tell us how the various factors that control species diversity vary along the environmental gra- dient and to tell how much each of the factors contributes to the species diversity gradient. Even along a given environmental gradient, the accounting will surely differ for different groups of species. For example, the form of the latitudinal gradient for birds is very dif- ferent from that for salamanders, and these two gradients must be explained by different mixes of contributing factors. To illustrate how the “QQID” approach provides a checklist of factors that may con- tribute to species diversity gradients, let us consider two of these famous gradients: the altitudinal gradient and the latitudinal gradi- ent. Altitudinal gradients in species diversity. s one ascends from sea level towards the summit of a high mountain, species diversity tends to decrease with elevation, as exempli- fied in this symposium by the decrease in Andean tree species diversity (Gentry, this volume) and frog species diversity (Duellman, this volume) with altitude. However, this pat- tern is by no means universal. For example, the diversity of plant species along the western Andean slopes of northern Chile is extremely low at sea level, increases from middle to high elevations, and decreases only from high el- evations onwards (Arroyo et al., this volume). In the Mediterranean zone of California the species diversity of birds reaches a maximum at middle elevations (Cody, 1975). How can we account for any one of these gradients, and why does the form of the gradient differ from case to case? Consideration of the QQID checklist suggests at least three important 126 Annals of the Missouri Botanical Garden contributing factors that vary along the al- titudinal gradient, one of them involving re- source quality, the other two involving re- source quantity. Changes in habitat structural diversi- ty. Habitat physiognomy varies dramati- cally along an altitudinal gradient. For ex- ample, in the moist tropics habitat structural complexity decreases monotonically along the altitudinal gradient, from tropical rainforest at the base through montane forest and then alpine elfin scrub at higher elevations, to al- pine grassland and eventually rocky slopes and glaciers at the highest elevations. This continuous decrease in habitat structural complexity and consequently in “number of niches" contributes to the continuous de- crease in plant and frog species diversity with elevation in the wet tropics. In the Mediter- ranean zone, however, scrub formations such as chaparral at sea level yield to forest at higher elevations before finally yielding to al- pine habitats on the highest summits, and this intermediate maximum in habitat structural complexity contributes to the intermediate maximum in species diversity. Productivity gradient. Temperature generally decreases with increasing altitude, while the altitudinal gradient of rainfall (and hence of productivity, which depends both on rainfall and temperature) varies from site to site. In the moist tropics, productivity de- creases with elevation, or there may be a slight increase in productivity from sea level up to a gentle maximum at medium-low el- evations, followed by a decrease in produc- tivity thereafter. This productivity gradient reinforces the effect of the gradient in habitat structural complexity and also contributes to the decrease in species diversity with altitude in the moist tropics. However, in the Medi- terranean zone of California and in the Ata- cama Desert there is a marked maximum in productivity at middle elevations (owing to the marked maximum in rainfall there), and this contributes to the species diversity max- imum at middle elevations. - Area gradient. The distribution of avail- able area with altitude depends on the form of the mountain. On conical mountains, area decreases continuously with altitude, tending to cause a monotonic decrease in species di- versity with altitude. However, Tibet and the Peruvian/Ecuadorean Andes have a more trapezoidal shape, with a broad plateau at high elevations, so that the maximum area may actually be at high elevations rather than at sea level. These area considerations may con- tribute to the fact that species diversity at high elevations on the Tibetan Plateau and on the Andean Altiplano is much higher than in structurally similar habitats of New Guinea, whose mountains more nearly approximate steep narrow ridges with only tiny areas at high elevations. Quantitative analysis of the altitudinal distribution of area contributes to understanding the relative numbers of mon- tane and lowland bird species on various is- lands of the Solomon Archipelago (Mayr & Diamond, 1976). hus, to account for the altitudinal gra- dient of species diversity in any particular case, one should at minimum consider that site’s altitudinal gradient of habitat structural complexity, productivity, and area. Terborg 1977) has shown that the quantitative ac- counting falls out differently for different trophic groups of birds (insectivores, frugi- vores, and nectarivores) along the altitudinal gradient of the Peruvian Andes. The same three variables—habitat structural complex- ity, productivity, and area—are also likely to be major contributors to species diversity gra- dients along horizontal habitat gradients such as the gradients of desert, grassland, scrub, and woodland in the Mediterranean zone, as illustrated by Cody’s (1975) analysis. — Latitudinal gradients in species diversity. No discussion of species diversity would be complete without consideration of the latitu- dinal gradient. Species diversity of most broadly defined groups of plants and animals is maximal in the tropics and decreases to- wards the poles. Examples considered in this symposium are the-high tropical diversity of plants (Gentry, this volume), beetles (Erwin, this symposium), and frogs (Duellman, this Volume 75, Number 1 1988 Diamond 127 Factors Controlling Species Diversity volume). In the analysis of plant species di- versity in northern Chile by Arroyo et al. (this volume), a steep species diversity gradient arising from the rainfall gradient is super- imposed on a gentler species diversity gra- dient Menem with latitude itself. However, some plant and animal groups, such as sand- pipers and Old World salamanders, do not exhibit a diversity peak in the tropics. Again using the QQID checklist, we can identify at least five factors with major contributions to the latitudinal gradient. Two of these factors involve resource quality, one involves re- source quantity, and two involve dynamics. abitat structural diversity. Habitat structural diversity tends to decrease from the equator to the poles, the extreme ends of the gradient being equatorial tropical rain- forest as contrasted with the polar ice caps. This environmental gradient contributes to the polewards decline in species diversity. radient of resource types. The variety of resources, or of resources available year- round, tends to decrease with latitude. For example, the proportion of insect species with very large bodies decreases polewards, with the result that bird species (e.g., coucals) spe- cializing on very large insects are mainly trop- ical. Nectar and fruit are available year-round in the tropics but not in the Arctic, contrib- uting to the decrease in diversity of obligately frugivorous and nectarivorous bird species with latitude. Productivity gradient. Productivity on land tends to decrease with latitude, rein- forcing the polewards decline in species di- versity. This latitudinal gradient in produc- tivity is less regular in marine environments, because latitudinal effects of temperature changes in productivity are overridden by effects of nutrient upwelling in some high- latitude marine areas. Disturbance gradient. One reason often proposed for the latitudinal gradient in species diversity is that disturbances on a geological time scale are supposedly more violent and produce more extinctions at high latitudes than at low latitudes. In particular, glaciations have periodically wiped out species diversity at high latitudes. This argument, if valid, would involve a contribution of species dynamics to the latitudinal gradient of species diversity. In recent years there has been increased ap- preciation of the historical importance of en- vironmental disturbances in the supposedly stable tropics. The Pleistocene involved al- ternate wet and dry periods that caused large- scale habitat changes in the tropics. It is nevertheless probably still true that environ- mental changes over geological times have been more devastating of habitats and more destructive of species diversity at high than at low latitudes. radient in dispersal and speciation rates. Tropical species of birds, and possi- bly of some other taxa, tend to be much more sedentary than temperate species. Practically all bird species of North America and Europe are known to have crossed water gaps of at least several miles in modern times, while most species of the continental tropics ap- parently do not cross water gaps (Diamond, 1976; Diamond & Gilpin, 1983). These low dispersal rates in the tropics may have con- tributed to tropical species diversity by mak- ing it possible for formerly conspecific pop- ulations to achieve reproductive isolation over shorter distances, and thus by enhancing spe- ciation rates. Thus, the latitudinal gradient in species diversity involves multiple factors, but these actors are not infinite in number. at is now required is to attempt to partition the contributions of these various factors to lat- itudinal gradients of species diversity in par- ticular cases. SUMMARY AND OUTLOOK We have seen that determinants of species diversity can be grouped into four sets of factors that may be remembered by the mne- monic “QQID”: (resource) quality, (resource) md (species) interactions, and dynam- his is not to say that all four sets of rite are equally important in explaining species diversity of different taxa, or at dif- ferent sites. For example, an interesting in- 128 Annals of the Missouri Botanical Garden terpretation of neotropical tree diversity with- in tree guilds dispenses almost entirely with considerations of niche differentiation and segregation by resource utilization, and in- stead stresses the dynamics of speciation, im- migration, and extinction (Hubbell & Foster, 1986). A general explanatory theory of species diversity must ultimately tell us under what sorts of circumstances each factor is likely to be important and what factors contribute to species diversity gradients along various en- vironmental gradients. At present, I doubt there is a single case where we have an ad- equate accounting that considers all four pos- sible sets of factors for a given group of taxa at a given site. Thus, we shall have to obtain such analyses for many individual cases be- fore we can begin to compare those cases and arrive at generalizations about species diver- sity. I view the gathering of such accountings as one of the two major tasks for future studies of species diversity. The other major task is to attempt to con- vert empirical groupings of factors controlling species diversity, such as the one that I have proposed, to natural hierarchical groups. The QQID grouping is offered just as a convenient empirical checklist; it does not necessarily correspond to any scheme in nature. Must we always content ourselves with such an arbitrary laundry list, or is there any natural organization to the laundry list? I suggest that it may be possible eventually to account for species diversity by a hierarchy of processes in space and in time. The spatial hierarchy would begin or end with an understanding of species diversity at a single point in space, then within a single type of habitat (so-called alpha diversity), then diversity from end to end of a habitat gradient (species turnover along this gradient being termed beta diver- sity), and finally species diversity over areas large enough to permit geographic replace- ment (gamma diversity), or over whole bio- geographic regions, or over the whole world. A hierarchy in time might begin with the rapid increase in species diversity during recovery from a storm, then the much slower increase following a glacial period with its attendant fluctuations in sea level, and finally the slow generation of species over geological/evolu- tionary time scales. [n this way, it may even- tually be possible to obtain not just a con- venient checklist, but a natural explanation for the number of biological children in the world's candy store. LITERATURE CITED Brown, J. H. 1971. Mammals on mountaintops: no eub insular biogeography. Amer. anuales 105: -478 Copy, M. L Towards a theory of continental species diversities: bird distributions over Mediter- ranean habitat gradients. Pp. 214-257 ¿n M. Cody & J. M. Diamond (editors), Ecology and Evo- in of ori Harvard Univ. Press, Cam- , Mass setts. Sioa i M. 1912. E kinetics: esti- mation of relaxation times for avifaunas of Southwest Pacific islands. Proc. Natl. Pues Sci. U.S.A. 69: 3199-3202 197 Relaxation and differential extinction on land: bridge islands: applications to natural pre- idi Proc. 16th Int. Ornith. Congr.: 616-628. — 84. “Normal” extinctions of isolated pop a Pp. 191-246 in M Nitecki (editor), Extinctions. Univ. Chicago os Chicago, Illinois. —— T & M. E. Girin. 1983. ilici and the evolution of the Philippine avifauna. Oikos -321. 41: 307 m O. H. € M. E. SouLé. 1981. Conservation Evolution. Cambridge Univ. Press, Cambridge HE s. P. & R. B. Fo: tree communities. Pp. 314-329 in ecce Community Ecology. Harper and Ro: ow, New KNoLL, A. H. 1986 poe of bi i in plant com- munities through geological times. Pp. 126-141 in . M. Case (editors), Community New York. & M. L. Copy. 1966. On the Wake isuwa habitat selection and bird er. Naturalist 100: 319-332. n of the montane avifauna of Northern um Proc. Natl. Acad. Sci. U.S.A. 73: 1765- 169. Pus a T. 1966. Food web aia and species rsity. Amer. Naturalist 100: 75. ROUGHGARDEN, J. A o of food-limited d Ringe animal competition communities. Pp. 4 92-516 i i . J. Case (ed- itors), Comuni Ecology. Harper iud Row, New or — Volume 75, Number 1 Diamond 129 1988 Factors Controlling Species Diversity Sout, M. E. 1986. Viable Populations. Cambridge Yobzis, P. 1986 rouen api and com- Univ. Press, Cambridge TERBORGH, J. R. 1974. Preseryaiion of natural diver- sity: the d of —— prone species. Bio- Science 24: ui 77. "Bid s d cies diversity on an Andean elevalional poet Ecology 58: 1007-1019. munity structure. Pp. 480-491 in J. M. Diamond & T vili Pun Ecology. Harper and Row, New Yor A CONTRIBUTION TO THE POLLEN MORPHOLOGY OF NEOTROPICAL LAURACEAE!' Bhoj Raj? and Henk van der Werff* ABSTRACT The present study is a p investigation of 80 species a to 17 of the 22 known ra of Laura Obse neotropical gene ade with light and s ilions were ma nning electron mic roscopes. An ultrastructural study A the ole wall of 11 genera was also d. na b transmission electron microscopy. ‘th st of the genera are easily identifiable by their sizes an 1 with pointed or blunt, monomorphic or dimorphic spinu The spi numerous tightly woven strands; their bases are encompassed by a thick, . cushionlike number E hid dpi ia , spher roidal, and inapertu . The exine haped bodies. "The surface of inules are made u form. The be was fou und to be the dominant layer of the pollen wall and to show varied composition and structural details. It is remarkable that a similar type of intine has been encountered in the inaperturate ra as Canna, Heliconia, Hernandia, and Palmeria. The taxonomic ues ioa of the paly- such disparate gene Wr] observations are discussed. pollen grains belonging to TAXONOMIC REVIEW AND CLASSIFICATION Lauraceae form a large, predominantly tropical family of trees and shrubs. The sole exception is the genus Cassytha, which con- sists of leafless, twining parasites much like Cuscuta (Convolvulaceae). Cassytha is some- times treated as a separate family, but since it differs only in habit from the woody Lau- raceae, we maintain it in the Lauraceae. Cen- ters of species richness are tropical Asia and tropical America. Lauraceae are rather poor- ly represented in Africa, are rather diverse in Madagascar, and occur in Australian rain forests, the Pacific Islands, and in New Zea- land (two species). The family is easily recognized by its flow- ers. The perianth consists of two cycles of three tepals each, which are usually equal. By far the most common flower colors vary from white to green or yellowish; very rarely does one encounter reddish flowers (Kubitz- kia). The stamens are also arranged in cycles of three each, and in principle there are four staminal cycles. The fourth or innermost cycle is nearly always reduced to staminodes or is entirely lacking. Three cycles are usually present but can be reduced to two or one fertile cycle in some genera. The three sta- mens of cycle three have nearly always two glands attached at or near their base. In the genus Pleurothyrium these glands become greatly enlarged and surround all nine fertile ' We wish to thank the Directors of the Missouri Botanical Garden, St. Louis, and the Royal Botanic Gardens, w, for generously providing polliniferous material for the present study. We would like to extend our special wa l to Miss Pia Bro C calculating the number of spinules of all the species investigated, and to Dr. Gamal El-Gha Qatar, Qatar, for helping with light microphotography. ensen, Botanical Institute, Aarhus, Denmark, for undertaking the onerous task o zaly, University of Our warm thanks are due to Mrs. Elisabeth Grafstrom, Palynological Laboratory, Stockholm, for her kind help in embedding and sectioning material for transmission We appreciate the critical comments and suggestions by Dr. John Rowley, Botanical Ferguson, Royal Botanic Gardens, Kew, and Keating, s hern Illinois University, Edw arista, lilipais: Finally, we wish to express our sincere thanks to Dr. J. Praglowski, d Laboratory, Stockholm, for his constant help and valuable comments during the course of this study. uthors gratefully disi ledge the financial support from Missouri Botanical Garden, which enabled us to obs TEM studie : Palynolo 3 Missouri Botanic al Garden, P.O. Box 75: 130-167. 1988. ical Laboratory, Swedish uy An o History, S-104 05 Stockholm 50, Sweden. uis, Missouri 63166-0299, U.S.A. ANN. MISSOURI Bor. GARD. Volume 75, Number 1 1988 Raj & van der Werff 131 Pollen Morphology of Neotropical Lauraceae stamens. The ovary is superior (except in the monotypic African Hypodaphnis), has one locule, and develops into a drupe, which may be subtended by the enlarged floral tube. The flowers can be perfect or unisexual. Although the Hernandiaceae and Moni- miaceae are frequently mentioned as close relatives of the Lauraceae, flowers of Lau- raceae are rarely confused with other families. The Lauraceae are of considerable eco- nomic importance. Persea americana is widely cultivated for its edible fruit, the av- ocado. Most species of Lauraceae possess varying amounts of aromatic oils, some of which are economically important. Examples are camphor oil from Cinnamomum cam- phora; rosewood oil from Aniba rosaeodora; and Brazilian sassafras oil from Ocotea odo- rifera. The bark of Cinnamomum verum yields cinnamon. Leaves of Laurus nobilis and Um- bellularia californica are used as spices (bay leaves); in Central America, Litsea glauces- cens is used for the same purpose. Several other species are used on a local basis, for instance Sassafras albidum for the prepa- ration of root beer and Dicypellium cary- ophyllatum (now supposedly very rare due to excessive exploitation) for its spicy fruits. The wood of many species of Lauraceae is valuable. Some species have unusually hard or decay-resistant wood used for construction, and the wood of other species is excellent for making furniture (Richter, 1981). CLASSIFICATION A workable generic classification of the Lauraceae does not yet exist. Several reasons lie behind the poor understanding at generic level. The most important reason is that, with very few exceptions (Cassytha, Hypodaph- nis), the genera do not have exclusive char- acters, that is, characters restricted to one particular genus. As a consequence, the gen- era are separated by combinations of char- acters. Very often both floral and fruit char- acters are used, and, because specimens only rarely have flowers and fruits (fruits usually require one or two months to mature), keying specimens to genus is frequently impossible. A second difficulty is that in some cases the characters separating genera are weak or dif- ficult to interpret. Thus, even perfect speci- mens cannot always be unambiguously as- signed to a genus. Results of these difficulties are that many generic identifications are suspect and that species are frequently transferred between genera or described in incorrect genera. It seems that the Lauraceae are the only family in which every botanist who seriously worked on the family has redescribed an already pub- lished species in a genus different from the one in which it was previously described. Tables 1, 2, and 3 present three recent generic classifications of Lauraceae. Tables 1 and 2 are abbreviated versions of classifica- tions by Kostermans (1957) and Hutchinson (1964) based on flower and fruit characters. Table 3 shows Richter’s (1981) generic groupings and is based on studies of secondary xylem and bark. Cassytha, not being woody, is excluded from Richter’s groupings. Kostermans (1957) recognized five tribes in the woody Lauraceae, which are separated on inflorescence and cupule characters. One tribe consists solely of the African Hypo- They are the Perseae (inflorescence exinvol- ucrate, cupule lacking), subdivided into two subtribes; the Cinnamomeae (inflorescence exinvolucrate, cupule present), likewise sub- divided into two subtribes; the Litseae (inflo- rescence with decussate bracts, cupule pres- ent), subdivided into two subtribes; and the Cryptocaryeae (inflorescence exinvolucrate, drupe fully enclosed in floral tube), also sub- divided. The subdivisions in each of these four tribes are based on numbers of anther cells. Kostermans (1957) postulated an evolu- tionary trend in the Lauraceae from genera with a very shallow floral tube, where the fruit is not subtended by a cupule (tribe Perseae), through genera with a deeper floral tube, where the mature fruit is subtended by a cupule up to one-third the size of the fruit 132 Annals of the Missouri Botanical Garden TaBLE l. Classification. (after Kostermans, 1957) of the neotropical genera of Lauraceae. Subfamily Lauroideae ibe Perseeae. Inflorescence paniculate, exinvolucrate. Cupule absent. Subtribe Perseineae. Anthers four-celled. ;enera Persea, Phoebe, Caryodaphnopsis Subtribe Beilschmiediineae. Anthers two-celled. Genera Beilschmiedia, Mezilaurus Tribe Cinnamomeae. Inflorescence paniculate, exinvolucrate. Cupule present. Subtribe Cinnamomineae. Anthers four-celled. Genera Ocotea, Nectandra, Pleurothyrium, Sassafras, Umbellularia, Dicypellium Subtribe Anibineae. Anthers two-celled. Genera Aiouea, Aniba, Endlicheria, Licaria, Urbanodendron, Kubitzkia, Phyllostemonodaphne Tribe Litseae. Inflorescence involucrate with decussate bracts. Cupule present. Subtribe Litseineae. Anthers four-celled. Genus „itsea Subtribe Lauriineae. Anthers two-celled. Genus Lindera Tribe Cryptocaryeae. Inflorescence paniculate, exinvolucrate. Fruit fully enclosed in flower tube. Subtribe Eusideroxylineae. Anthers four-celled. Genus Eusideroxylon (not neotropical) Subtribe Cryptocaryineae. Anthers two-celled. Genus Cryptocarya Tribe Hypodaphneae. Ovary inferior. Genus Hypodaphnis (African) Subfamily Cassythoideae Genus Cassytha (tribe Cinnamomeae), and through genera with century in Ocotea. This tribe is characterized a deep floral tube, where the fruit is fully by having all anthers with introrse cells. The enclosed in the floral tube (tribe Cryptocar- Cinnamomeae (exinvolucrate inflorescence, yeae), to the tribe Hypodaphneae, where the at least one anther row extrorse, all anthers fruit is fully enclosed in and becomes fused four-celled) differs from the Sassafrideae only with the floral tube. This is a stimulating the- in having some anthers extrorse. Hypodaph- ory, although it is not certain whether it is nis is placed in this tribe. The fifth tribe, the phylogenetically accurate, that is, whether Litseae, is also separated based on its invol- the Perseae are indeed the most primitive — ucrate inflorescence. tribe. The position of the Litseae, recognized The third recent classification of Lauraceae by their inflorescence type, is not clear in this (Richter, 1981) is not based on floral or fruit theory. characters, but on wood and bark. It should Hutchinson (1964) also recognized five therefore not be surprising that it differs in tribes of woody Lauraceae but defined his several features from Kostermans's an tribes quite differently. His Apollonieae are — Hutchinson's classifications. Richter recog- characterized by exinvolucrate inflorescence, nized three main groupings of genera without two-celled anthers, and fruit not fully enclosed giving them taxonomic status. Each group by floral tube. His Cryptocaryeae are as his includes primitive and more advanced genera Apollonieae except for the fruit being enclosed (as defined by wood and bark characters). by the floral tube. Tribe III, the Sassafrideae, Ancestral genera from which these groups is of doubtful value; its single New World were derived are not yet known. The first species, Sassafridium veraguense, has been group includes mostly palaeotropical gen- placed by nearly all botanists during the last era—from this group only Beilschmiedia, Volume 75, Number 1 Raj & van der Werff 133 Pollen Morphology of Neotropical Lauraceae TABLE 2. Classification of Lauraceae— Hutchinson (1964), neotropical genera. Tribe eee bise Fertile stamens 9; calyx usually persistent. DE Inflorescence exinvolucrate; anthers two-celled; fruit not or partly enclosed by cupule. ual. Beilschmiedia, Kubitzkia, Aniba, Nobeliodendron, Urbanodendron Fertile stamens 6-3. , Licaria, IIS Misanteca, Mezilaurus Ai Flowers picea or polygamo-dioeciou Aniba, Endlicheria Tribe 2. Cryptocaryeae. Inflorescence exinvolucrate; anthers two-celled; fruit enclosed in floral tube. Cryptocarya Tribe 3. Sassafridea My drin (now include Tribe 4 Flowers e l. ae. Inflorescence exinvolucrate; anthers four-celled; all anther cells introrse. n Ocotea) namomeae. As tribe 3, but at least one row of anthers with extrorse cells. Nectandra, Persea, Phoebe, Pleurothyrium, Caryodaphnopsis, Synandrodaphne (= Rhodostemonodaphne) Flowers bisexual or polygamous Ocotea, Dicypellium Tribe 5. Litseae. Inflorescence involucrate; flowers umbellate or solitary within the involucre. l Anthers four-celle itsea, Umbellularia Anthers two-ce Lindera, Lau Tribe 6. Cassy iue "Bons leafless herbs. Cassy p Caryodaphnopsis, and Cryptocarya are represented in the New World, albeit with few species. The second group includes most of the genera occurring in the subtropics of the Northern Hemisphere (Persea is the ex- ception), but several of the genera are best represented in tropical areas. This group is represented in the New World by Litsea, Lindera, Sassafras, and Umbellularia. The third group includes all other (15 or 16) neo- tropical genera plus a few paleotropical seg- regates of Persea and Phoebe/Cinnamomum (Alseodaphne, Dehaasia, Neocinnamomum, and Nothaphoebe). Within this group, Per- sea, Cinnamomum, Phoebe, and allies can be separated from the remaining endemic neotropical genera (Ocotea is also present in Africa and Madagascar), but the remaining genera do not fall in distinct groupings based on wood or bark characters. The classifications by Kostermans and Hutchinson show that our knowledge of the relationships among the genera of Lauraceae is still limited. Essentially, these classifications are only keys to genera, often based on readily observable but artificial characters such as the number of fertile stamens and number of anther cells on the stamens. The genera, de- fined with the help of these characters, often include species that do not possess the re- quired characters but are placed in a certain genus due to general resemblance to other TABLE Genera groups according to Richter (1981), neotropical genera. Group III Group I Beilschmiedia Persea Crypto Phoebe PALMAE SN Pleurothyrium ë Endlicheria »roup II . eg iouea indera Nectandra Litsea o Ocotea E, AULA TIU Urbanodendron Sassafras Kubitzkia Dicypellium Licaria Aniba Anaueria Mezilaurus [Rhodostemonodaphne] 134 Annals of the Missouri Botanical Garden species in that genus. Needless to say, this frustrates the efforts of the botanists who are not familiar with the group but try to key specimens to genus. A good example of these problems is found in the genus Persea. One of its characters is the presence of nine four- celled anthers and three staminodes. How- ever, Kopp (1966), who monographed the genus, included in Persea also species with six four-celled anthers and six staminodes, and Kopp included species with some or all anthers two-celled, largely because of simi- larities in fruit and vegetative characters. In Persea the staminal configurations seem to be constant for a given species, but there are a few cases where the number of anther cells varies between flowers of the same inflores- cence (Aiouea lundelliana, Kubitzkia ma- crantha). Normally, when flowers have both two- and four-celled anthers, the outer six anthers are always of one type and the inner three anthers the other; for instance, in End- licheria anomala the outer six anthers are two-celled, the inner four-celled, while in Phoebe subg. Heteranthera the outer six are four-celled and the inner three are two-celled. Mez (1889) realized this and restricted the couplet in his generic key “anthers 4-celle or anthers 2-celled" to the outer six anthers. This greatly reduced the number of species that did not fit in any of the genera and made identifications easier. It is regrettable that this has been overlooked by later workers. he differences in treatment of the genera with two-celled anthers show the lack of un- derstanding of the generic relationships. In Hutchinson’s classification, two-celled an- thers characterize two tribes, the Apollonieae and Cryptocaryeae; the Cinnamomeae consist of four-celled genera, and the Litseae include both two- and four-celled genera. Thus, Hutchinson regarded the difference in num- ber of anther cells as very important. Kos- termans considered this difference as second- ary, as all of his tribes include both two- and four-celled genera (except the monotypic Hy- podaphneae), and accepted that the transition from four-celled to two-celled genera oc- curred several times. Richter found that wood and bark characters do not support the idea that genera with two- and four-celled anthers form distinct groups, and it seems likely that the two-celled genera have been indepen- dently derived from four-celled genera. This theory is strengthened by the observation that in some two-celled neotropical genera the up- per two cells have been lost, while in other genera the lower pair has disappeared. There are also indications that some two-celled gen- era (Aiouea, Endlicheria) are not monophy- letic, but contain groups of species that were independently derived from four-celled ances- tors; on the other hand, other two-celled gen- era of comparable size (Aniba, Licaria) seem to be monophyletic. In addition to the differences in treatment of the two-celled genera, the three classifi- cation systems differ in many other details. Hutchinson and Richter recognized some neo- tropical genera that Kostermans did not ac- cept. Examples are (from Hutchinson): Sas- safridium, now accepted as belonging in Ocotea; Nobeliodendron and Misanteca, both now included in Licaria (Kurz, 1983); and Synandrodaphne, included in Ocotea by Kostermans (1957), also frequently included in Nectandra but recently recognized as Rho- dostemonodaphne by Rohwer & Kubitzki (1985). Richter also recognized several gen- era that Kostermans had treated as syn- onyms: Caryodaphnopsis, Nectandra, Pleu- rothyrium, and Anaueria. More differences exist in the placement of the genera, largely because Kostermans and Hutchinson used dif- ferent characters to define their tribes. For instance, Hutchinson placed most two-celled, exinvolucrate genera in the tribe Apollonieae, whereas in Kostermans's system these genera are divided between the Perseae and Cinna- momeae. Kostermans's Perseae are divided between Hutchinson's Apollonieae and Cin- namomeae. Although both classifications need to be modified, incorporating results from Richter's work, it seems that Kostermans's system needs fewer changes and has the ma- jor groups better defined than Hutchinson's system. Both Kostermans's and Hutchinson's clas- sifications are essentially keys and aimed at easing the identification of specimens to ge- Volume 75, Number 1 1988 Raj & van der Werff Pollen Morphology of Neotropical Lauraceae 135 neric level. Information needed for an un- derstanding of the phylogeny was (and is) not available. As has been mentioned earlier, workable keys to the genera are still not avail- able and a large obstacle is that both floral and fruit characters are important for the definition of the genera although rarely spec- imens include flowers and fruits. This diff- culty is not likely to be solved in the near future. It is possible to write a generic key based on floral characters with which most specimens can be identified. Fruiting material can be recognized either by cupule shape or leaf venation; cupules are likely to offer im- portant characters for defining phylogenet- ically sound genera. ifficulties in resolving the phylogenetic relationships should be viewed as distinct from difficulties involving identifications. Analyses of flower and fruit types have shown an ex- tensive reticulate pattern of variation within the Lauraceae, and it is not known which taxa are primitive and which are derived, nor where the family originated. Raven & Axelrod (1974) postulated that, based on present pat- terns of distribution and endemism and on the fossil record, in the New World some taxa migrated from South America northward, while others, of Laurasian origin, were present in North America and moved southward. Gen- try (1982) supported the hypothesis that the Lauraceae are a tropical Gondwanaland fam- ily. Richter's (1981) work on wood and bark anatomy has shown that the mostly small genera found in temperate and subtropical parts of the Northern Hemisphere (Apollo- nias, Laurus, Lindera, Litsea, Sassafras, and Umbellularia) have a similar wood struc- ture. Because these genera occur in different continents, and because two other genera, now largely represented in the tropics, also occur in northern subtropical/ temperate areas (Ocotea in the Canary Islands; Persea in Nort America and the Canary Islands), it is tempt- ing to regard the Lauraceae as a Laurasian family that has very successfully spread into the tropics. The fact that Ocotea is much better represented in the tropics than in the temperate zones does not imply a tropical origin. Accepting the dioecious species as de- rived from species with hermaphrodite flow- ers, one should note that most Ocotea species in the South American lowlands are dioecious, whereas nearly all Central American Ocotea species and most Andean species have perfect flowers. Further, in Central America Ocotea is much more weakly separated from its clos- est allies (Vectandra, Phoebe). These obser- vations do not support the view that in the New World Ocotea and its allies originated in South America and moved north, but rather the reverse (Rohwer, 1980). Much more information is needed for mak- ing secure claims about the phylogeny of the Lauraceae. First of all, our knowledge about the distribution of the genera is incomplete. Recently Caryodaphnopsis, previously known as an Asian genus of seven species, was found in South America (van der Werff & Richter, 1985); recent collections show that this genus includes six to eight neotrop- ical species and occurs in Costa Rica, Pan- ama, Colombia, Ecuador, Peru, and Brazil. Also during the last few years, four genera previously not known from Central America have been discovered there (Aniba, Cary- odaphnopsis, Mezilaurus, and Pleurothy- rium; pers. obs.). Characters beyond flowers and fruits should be investigated. An excellent example is the work by Richter (1981) on wood and bark anatomy; studies of pollen, such as here re- ported, are very promising, and a study of leaf venation (Klucking, 1987) will appear soon. Other areas worthy of investigation are the chemical constituents and the structure of the cuticles. Results of such studies will be of much help in assessing the importance of the various morphological characters and in determining which characters should be used to define monophyletic genera and which are important on the species level. PALYNOLOGICAL REVIEW Extensive palynological investigations have been few. These were based mostly on ob- servations with light microscopes. Pollen-mor- phological studies have been made by classical and by recent workers. The former generally 136 Annals of the Missouri Botanical Garden included information on pollen as a matter of secondary importance. The history of pollen descriptions probably began with Kolreuter (1811), followed by Purkinje (1830), Fritzsche (1832), Mohl (1834), Schnizlein (1843-1870), Berg & Schmidt (1858), Griffith & Henfrey (1875), Edgeworth (1877), Mez (1888), Knell (1914), Armbruster & Oenike (1929), Wodehouse (1932, 1965), Zander (1935, 1937, 1941), Cranwell (1942, 1953), Selling (1947), Ka- sapligil (1951), Erdtman (1952), Veloso & Barth (1962), and Pal (1976). The general descriptions given by the said authors is that the pollen grains are devoid of furrows and pores. Likewise, the sculpture spoken of by Wodehouse, as well as Mez, Knell, Zander, and Selling, to mention a few, are matters of controversy. Short pollen descriptions and illustrations have also found a place in many regional floras: see Cranwell (1942, 1953), Ikuse (1956), Zinderen Bakker (1956), F. H. Wang (1960), J. L. Wang (1962, 1969), Rao & Lee (1970), Heusser (1971), Huang (1970, 1972), Markgraf & D'Antoni (1978), Lieux (1980), and Lewis et al. (1983). Casual reference to pollen morphology mainly from an embryological standpoint has also been made by Schroeder (1952), Sastri (1958, 1962), Mitroiu (1970), and Pal (1975). Recent noteworthy palynotaxonomic and phylogenetic studies are those of Agababian (1969, 1973), Ghosh (1977), Suryanarayana & Deodikar (1978), and Datta & Chanda (1980). Of late the specialized wall structure of pollen grains has attracted the attention of many workers such as Takeoka (1965), Walker (1976), Rowley & Vasanthy (1980), Kubitzki (1981), Hesse & Kubitzki (1983), Hesse & Waha (1983), and Sohma (1985). The pollen wall is characteristic because of the extreme thinness of exine and the massive nature of the intine. A similar type of spo- roderm stratification has been encountered in Cannaceae, Heliconiaceae, Hernandiaceae, Strelitziaceae, and Monimiaceae (Erdtman, 1952; Kress et al., 1978; Kress & Stone, 1982, 1983a, b; Stone et al., 1979, 1981; Stone, 1987; Rowley & Skvarla, 1974, 1975, 1986; Foreman & Sampson, 1987). There is a scanty pollen record of fossil lauraceous genera. The few recorded are those of Macko (1959) and Macphail (1980). It is interesting that Selling (1947) stated that it is not “as if it [lauraceous pollen] had been overlooked, nor do I think that the possibility of the pollens being easily destroyed and therefore rarely preserved in peat (as was suggested by Rudolph 1936, p. 297, with regard to the pollens of this family) is of importance for the interpretation of their ab- sence. The present study is a broadly based pol- len-morphological investigation of the appar- ently eurypalynous family. The aim is to pres- ent detailed morphological descriptions based on light microscopy (LM) and scanning elec- tron microscopy (SEM), as the previous works are limited to very brief, sometimes incorrect accounts, depending mostly on light micro- scope observations. The complicated exine structure, sculpture, and ultrastructural de- tails can be adequately studied with scanning electron microscopy and transmission elec- tron microscopy (TEM) only. The present investigation is aimed not only at giving a better understanding of pollen morphology but also at determining the cor- rectness of the grouping of different genera under different tribes, subtribes, and their subfamilies respectively. The deflections of the palynological compass needle, at times oversimplified, have been taken into account to see whether it points towards better group- ing, rearrangement, affinities, or evolutionary tendencies among the genera investigated. MATERIALS AND METHODS Polliniferous material of all the species in- vestigated (except one sample each of Ku- bitzkia mezii and Mezilaurus navalium, ob- tained from Kew) was collected from the herbarium of the Missouri Botanical Garden Volume 75, Number 1 1988 Raj & van der Werff 137 Pollen Morphology of Neotropical Lauraceae (MO), St. Louis, Missouri, U.S.A. Permanent slides were made from unacetolyzed, fuch- sine-stained pollen grains. Acetolysis easily disintegrates the pollen grains because of the extreme thinness of exine (Erdtman, 1943). A total of 80 species belonging to 17 out of the known 22 neotropical genera have been investigated. The LM studies were made using a Leitz Wetzlar Dialux microscope with apo- chromatic oil immersion objective (X100, N.A. 1.32) and periplan eye pieces (GF x 10). Measurements were made under oil immer- sion and are based on 30 or more pollen grains per species. Exine thickness was measured in the center of mesocolpia of pollen grains lying in equatorial view and includes spinules and other exinous excrescences. The maximum and minimum numbers of spinules per species were calculated using a programmable com- puter (Texas Instruments II Programmable 59; for details see Christensen, 1986) and and are presented in Table 1. Pollen descriptions are based on LM and SEM observations. A general description is given for each genus, followed by a description of the taxa sectioned. Pollen mor- phological data of the species are compiled in Table 1. The classification and order of gen- era follows that of Kostermans (1957), and the species are arranged in alphabetical order. The arrangement of photographs is according to the classification. For scanning electron microscopy, unace- tolyzed pollen grains were suspended in a drop of absolute alcohol and transferred to brass stubs. The pollen grains were coated with gold palladium for seven minutes using a Fine Coat ION Sputter JFC-IIOO. Scanning micro- graphs were taken with a Jeol JSM 255-II microscope using Ilford FP4 film. For transmission electron microscopy, un- acetolyzed anthers were fixed in TAG solution (Rodewald & Karnovsky, 1974), buffered in sodium cacodylate buffer, postfixed in osmium tetroxide, and embedded in Spurr. Ultrathin sections were cut with a diamond knife using an LKB ultratome and poststained with uranyl acetate and lead citrate. The terminology fol- lows that of Erdtman (1969). GENERIC DESCRIPTIONS The pollen grains of all the species inves- tigated are inaperturate and spheroidal. The exine consists of an extremely thin, contin- uous or interrupted layer with an ornamen- tation consisting of pointed or blunt spinules made up of numerous twisted strands or of minute processes of varying shape (except Cryptocarya, where the exine is thin and smooth). The thinness of the exine makes it exceedingly difficult to distinguish the sexi- nous layer from the nexinous layer. Ultra- structural study of the pollen wall leads to the conclusion that the exine is totally ectexinous. The dominant massive layer is the intine. It is stratified, and its composition can be brillar, lacunar, lamellated, or granular. The intinous nature of this layer has been verified by acetolysis. It is uniformly thick throughout and does not show special thickenings that indicate possible germinal exits or germinative zones at any place. Instead, the entire wall may be equally suited for pollen tube egress. Persea. Figure 1. Diameter of pollen grains 29-44 um. Ex- ine 0.5-1.5 um thick, scattered with spinules 1-3 um apart of faintly discernible twisted strands. Spinules < 0.5-ca. 1 um long, point- ed or blunt, monomorphic or dimorphic with a basal cushionlike form of irregular shape; spinule surface devoid of processes but sur- rounded by a ring of densely spaced processes of varying shape and size. Intine 3-4 um thick, appearing homogeneous. In P. americana and P. mutisii the spi- nules are dimorphic; blunt and pointed spi- nules are intermingled. In P. caerulea and P. schiedeana the pollen grains are dimor- phic; one type of pollen grains has mono- morphic spinules and the other dimorphic. In P. veraguasensis neither the pollen grains nor the spinules are dimorphic. Persea fulva. Figure 1C. TEM obser- vation: pollen wall 4 um thick, its outer, very thin electron-dense layer consisting of glob- ular elements or in places baculalike elements 138 Annals of the Missouri Botanical Garden FIGURE 1. A, B. Persea americana. — A. Entire MA grain; mS s appear as white dots in the peripheral region; at a lower focus in the middle they appear as dark dots. LM x 1,000. — B. Part o and the cushionlike bases studded with minute processes. SEM x 11,000. "gil P. fulva. Part of pollen wall showing the outer thin, electron-dense layer made up of globular or in places baculalike elements (arrow) m a discontinuous tectum (T). The nda elos intinous layer shows ae (c) and vacuoles (v). 000. TEM x Volume 75, Number 1 Raj & van der Werff 139 1988 Pollen Morphology of Neotropical Lauraceae FIGURE 2. A-C. Phoebe porphyria. — A. Entire pollen grain, muss view Poen densely spaced spinules as Sue dots. —B. The same in a cross section. LM x z 00 . Part of exine showing es ules and their circular cushion, its surface scattered with a few minute gd Pede SEM x1 helicterifolia. Part of pollen wall Wis details of s TEM x20,000. Note the Hs ¿uen pisq commissural line (arrow) demarcating the outer and inner intine layers. 140 Annals of the Missouri Botanical Garden FIGURE 3. SEM x 1 supporting a discontinuous tectum and mi- nute, pointed, solid spinules 3 um apart, with a prominent, circular basal cushion intruding into the intinous layer, surrounded by densely spaced globular elements. The spinules and exinous surface covered with a “fuzzy coat- ing" of granular composition. The underlying massive intinous layer distinguished into an outer, thick fibrillar layer with irregularly ori- ented narrow channels and vacuoles filled with osmiophilic material and into an inner, thin, electron-dense granular layer with a slightly undulating inner margin. Phoebe. Figure 2. Diameter of pollen grains 21-33 um. Ex- ine 1-2 um thick, scattered with spinules 1— 2.5 um apart of easily discernible twisted strands. Spinules < 1-2 um long, pointed, with a prominent basal, circular, cushionlike form. These cushionlike forms densely spaced, their surface and the general surface of exine scattered with sparsely spaced, minute, gran- ular processes. Phoebe helicterifolia. Figure 2D. TEM observation: pollen wall 5 um thick, differ- Caryodaphnopsis fosteri.— A. Entire pollen grain; spinules appear as white dots in the peripheral region; at a lower focus in the middle they appear as dar x 1,000 ea. Part of exine showing spinules and their basal cushion, the surface covered with granular processes. 000. ots.—B. The same in optical cross section. LM entiated into an outer, thin, electron-dense ornamented exinous layer composed of closely packed granular elements and minute, point- ed, spinules 2 um apart, with a distinct, cir- cular basal cushion intruding deeply into the intinous layer and making its distal surface distinctly undulating. Spinules and exinous surface covered with a “fuzzy coating." The underlying massive intinous layer distinctly demarcated into an outer, relatively thick, distally undulating layer beset with usually radially oriented cavities, these partly filled with osmiophilic material, and into an inner, unevenly thickened, electron-dense fibrillar layer, its inner margin uneven. The outer and inner layers of intine clearly separated by a thin, discontinuous electron-dense commis- sural line. Caryodaphnopsis. Figure 3. Diameter of pollen grains 27-33 um, rare- ly 41 um. Exine 2 um thick, scattered with simple spinules 2-3 um apart (rarely the spi- nules show faintly discernible twisted strands). Spinules about 1.5 um long, pointed, with a basal cushionlike form, this circular to irreg- Volume 75, Number 1 1988 Raj & van der Werff 141 Pollen Morphology of Neotropical Lauraceae A-C. Beilschmiedia pendula. —.4. Entire pollen grain showing spinules as white dots. —B. The L C | same in optical cross section. LM x 1,000.— devoid of granular processes. SEM x 11,000. D, E spinules magnification. SEM x 11,000. ular in shape, its surface devoid of processes. The general surface of exine sparsely covered with minute processes. Intine 2 um thick, appearing homogeneous. Beilschmiedia. Figure 4A-C. Diameter of pollen grains 23-38 um. Ex- ine 1-2 um thick, scattered with spinules 2- 3 um apart of easily discernible twisted strands. . Part of exine beset with spinules, their basal cushionlike form . Mezilau rus navalium.— D. Entire pollen grain showing , their basal cushions beset with minute processes. SEM x 3,300.—E. Part of the same at a higher Spinules 1-1.5 um long, pointed, with a basal cushionlike form of irregular shape; its sur- face and the general surface of exine devoid of processes. Intine 2.5-5 um thick, appear- ing homogeneous. Mezilaurus. Figure 4D, E. Diameter of pollen grains 22-26 um. Ex- ine > 1 um thick, scattered with spinules 2 142 Annals of the Missouri Botanical Garden um apart of easily discernible twisted strands. Spinules < 0.5 um long, pointed, with an ill- defined basal cushionlike form; its surface and the general surface of exine beset with densely spaced processes. Intine 2 um thick, appear- ing homogeneous. Ocotea. Figure 5. Diameter of pollen grains 17-39 um. Ex- ine 1-2.5 um thick, scattered with spinules 1—4 um apart of distinctly discernible twisted strands. Spinules < 0.5-2 um long, pointed or blunt with well-defined, thick, basal, cir- cular, cushionlike form, its surface uneven. The cushionlike forms densely spaced and each surrounded by a ring of partially fused elements of varying size and shape. The gen- eral surface of exine beset with small pro- cesses of varying size and shape. In a few species, such as O. cernua and O. ira, transparent porelike or colpuslike areas devoid of spinules occur. In O. cuprea the pollen grains are dimorphic; one type of pollen grain has pointed spinules made up of twisted strands, with a prominent, circular, basal cushionlike form. These cushionlike forms are densely spaced and appear as small islands surrounded by partially fused, small, globular processes. [n places five or six spinules are grouped on a single cushion. In the second type of pollen grain, the spinules are pointed and broad basally but without a basal cush- ionlike form. The intine is 2-5 um thick and homogeneous to many layered. Ocotea puberula. Figure 5C. TEM ob- servation: pollen wall 4 um thick, differen- tiated into an outer, very thin, electron-dense ornamented exinous layer composed of mi- nute, closely packed globular elements and minute, pointed, solid spinules, 3 um apart with a distinct, circular basal cushion slightly intruding into the intinous layer and surround- ed by densely spaced globular elements. The spinules and general exinous surface covered with a “fuzzy coating." The underlying mas- sive intinous layer clearly distinguished into an outer, relatively thick layer beset with narrow, radially oriented channels in contact with the surface, these partly filled with bead- shaped osmiophilic material, and into a rel- atively thin, electron-dense, rather homoge- neous layer of varying electron density, its inner margin irregular. Nectandra. Figure 6. Diameter of pollen grains 18-33 um. Ex- ine 0.5-1.5 um thick, scattered with spinules 1-4 um apart of faintly discernible twisted strands. Spinules 0.5-1.5 um long, pointed, with a distinct, basal, circular to irregular cushionlike form, its surface and the general surface of exine devoid of processes. In N. ambigens the spinules are crowded in places to form a rosettelike pattern or are sometimes found in pairs on a common basal cushion. In N. falcifolia the spinules are fre- quently crowded in places or are dimorphic. In N. grandis they are blunt and vestigial. The intine 1-4 um thick and homogeneous or apparently bizonal or lamellated. Nectandra gentlei. Figure 6E. TEM ob- servation: pollen wall 5 um thick, its outer, very thin, electron-dense, ornamented exi- nous layer composed of closely packed gran- ular to globular elements and minute, pointed, solid, spinules 1 um apart with a distinct, circular, basal cushion intruding into the un- derlying layer. The underlying massive intin- ous layer made up of an upper, relatively thick layer beset with ill-defined narrow chan- nels partly filled with osmiophilic material and of an inner, relatively thin, many-layered granular stratum of varying electron density, its inner margin very irregular. Pleurothyrium. Figure 7. Diameter of pollen grains 21-30 um. Ex- ine 0.5 um thick, scattered with spinules 1— 2 um apart of easily discernible twisted strands. Spinules < 0.5 um long, pointed, witha prom- inent, basal, circular cushionlike form, totally or partially surrounded by a ring of densely spaced, partially fused minute processes. The general surface of exine beset with sparsely Volume 75, Number 1 Raj & van der Werff 143 1988 Pollen Morphology of Neotropical Lauraceae FIGURE 5. A, B. Ocotea cuprea.— A. Part of exine showing spinules, circular, cushionlike smooth form, urrounded by densely spaced minute processes. SEM x 11,000.— B. Part v Rae same showing a few spinules and details of their basal part. Note the twisted strands of the spinules. SEM x 22,000. — C. O. puberula. Part of pollen wall differentiated into an outer ly electron- s layer, composed of globular elements (long arrow). Vote the spinules and exine covered with a "fuzzy co (thick, short Bein radially oriented channels 000. filled with bead-shaped osmiophilic dol (small, thin pde TEM x 144 Annals of the Missouri Botanical Garden ^om A-C. Nectandra reticulata. — A. Entire pollen n showing de as e dots in the peripheral region and a as dark dots in the middle. — B. The same in optic Hes section. LM x 1,000.—C. Part of exine ipt spinules and their smooth, basal cushionlike form. SE. M x 11,000. =D Rioters grandis. of exine showing a surface pattern dissimilar to N. reticulata. SE M X 11,000.—E. Nectan entlei. Part a pollen wall showing the outer thin, electron-dense layer made up of granular to globular us l'union. Volume 75, Number 1 1988 Raj & van der Werff 145 Pollen Morphology of Neotropical Lauraceae spaced, minute processes. Intine 1.5 um thick, appearing homogeneous. Pleurothyrium zulianense. Figure 7D. TEM observation: pollen wall 2 um thick, its outer thin, electron-dense ornamented exi- nous layer made up of granular to globular elements and minute, pointed, solid spinules um apart with a distinct, circular, basal cushion intruding into the intinous layer. The underlying massive intinous layer distin- guished into an outer thick, structurally loose, granular stratum bearing closely packed, nar- row, radially oriented channels filled with os- miophilic material and into an inner, thin, granular layer, its margin irregular. Umbellularia. Figure 8. Diameter of pollen grains 31-36 um. Ex- ine 1 um thick, scattered with spinules 3 um apart of faintly discernible twisted strands. Spinules < 1 um long, pointed, with a prom- inent, basal, circular cushionlike form, par- tially surrounded by a ring of densely spaced, partially fused minute processes. The general surface of exine studded with densely spaced, minute processes of varying size and shape. Intine 2.5 um thick, appearing homogeneous. Umbellularia californica. Figure 8D. TEM observation: pollen wall 6 um thick and consisting of an outer, thin, electron-dense ornamented exinous layer composed of dense- ly spaced clavate to globular elements of vary- ing size and of minute, solid, spinules 3 um apart with a distinct, circular basal cushion intruding deeply into the intinous layer, and surrounded by densely spaced globular ele- ments of varying size. The spinules and gen- eral exinous surface covered with loose "fuzzy coating." The underlying massive intinous re- gion clearly distinguished into an outer, rel- atively thick stratum with scattered, radially oriented, ill-defined channels partly filled with osmiophilic material and into an inner, rela- tively thin, granular layer of varying electron density, its inner margin smoot Aiouea. Figure 9. Diameter of pollen grains 19-29 um. Ex- ine 0.5 um thick, scattered with spinules l- 3 um apart of very faintly discernible twisted strands. Spinules 0.5 um long, pointed, with a basal, thin, circular to irregularly shaped cushionlike form, frequently surrounded by a ring of densely spaced, partially fused, minute processes. Cushionlike forms usually fused, their surface uneven, devoid of processes. General surface of exine beset with sparsely spaced, minute processes. Intine 1.5-3 um thick and appearing homogeneous. In A. costaricensis the pollen grains ex- hibit in places porelike transparent areas de- void of spinules. Aiouea trinervis. Figure 9D. TEM ob- servation: pollen wall about 4 um thick, dif- ferentiated into an outer, very thin, electron- dense ornamented exinous layer made up o compact granular elements and spinules. The spinules minute, pointed, solid, 1-3 um apart with a distinct, circular basal cushion intrud- ing into the subtending layer, and surrounded by densely spaced globular elements of vary- ing size. The underlying massive intinous stra- tum consisting of an outer, relatively thick, structurally loose fibrillar layer made up o vacuoles and channels filled with osmiophilic material and of a denser inner, 2—3-layered granular zone of varying electron density, its inner margin slightly irregular. Aniba. Figure 10A-C. Diameter of pollen grains 16-27 um. Ex- ine 0.5 um thick, scattered with spinules 1 um apart of faintly discernible strands. Spi- nules < 0.5 um long, pointed, without dis- — The upper part of the intinous layer is beset with ill-defined channels partly filled with osmiophilic material (thin arrow); in the lower part undulating layers of varying electron density are seen (thick arrow). TEM 22,000 146 Annals of the Missouri Botanical Garden FIGURE 7. A-C. de rium densiflorum. — "s Entire pollen ein dnd spinas, Ri and e dots.— da The same in optical cross section. LM x 1,000.— C. Part of exine sh surface Bon . P. zulianense. Part of pollen wall showing the outer thin, electron- Pudet exinous E er up Peur to UL. elements (arrow) and showing the outer thick, structurally loose, gr ranular str n of intine (S) and an inner thin, granular layer (L). Note the radially oriented channels in the outer intine filled with osmiophilic material (c). TEM x 20,000. tinct, basal cushionlike forms. The general Kia Meme 10D. surface of exine densely spaced with granular processes. Intine 1.5-2 um thick and ap- Diameter of pollen grains 15-27 um. Ex- pearing homogeneous. In pollen grains of 4. ine 0.5-1 um thick, scattered with spinules firmula 2-3 porelike or irregular openings 0.5-2 um apart of faintly discernible strands. encountered. Spinules « 0.5-1 um long, densely or sparse- Volume 75, Number 1 Raj & van der Werff 147 1988 Pollen Morphology of Neotropical Lauraceae FIGURE 8. Umbellularia californica. — A. Entire pollen grain, yee spinules as white dots in the peripheral region and as dark dots in the middle.—B. The same in optical cross section. LM x 1,000.— C. Part of exine ps 2 their basal, circular, cushionlike form and the = pee spaced, = processes. SEM x11,000.—D. Part of pollen wall showing the outer, thin, electron-dense layer composed of clavate to globular elements (arrow) and spinules (s), their bases deeply rud ni into t pen intine and surrounded by globular elements (G); spinules and exine surface covered with loose “fuzzy co ting. ” The underlying thick, outer intine scattered with ill-defined na pe with osmiophilic cial inner intine thin, granular, and of varying electron density. TEM x 20, 148 Annals of the Missouri Botanical Garden . B. Aiouea costaricensis. — A. Entire pollen grain d pa as es dots in the peripheral region and as dark dots in the middle. —B. The same in optical cross section. LM x 0. C, D. rinervis. — .. Part of exine showing surface details. SEM x 11,000. PD. oe of polen wall ae thin, electron-dense exinous layer made up of granular elements (thin arrow) and spinules, their ases r into the intine and nd by globular elements (thick arrow). Outer ie s iatis Abr rillar in composition and beset with radially oriented channels filled with osmiophilic material; inner layers denser, of varying lor density. TEM x ¿ 0 ly spaced, usually pointed, frequently vesti- varying size and shape. General surface of gial. The basal cushionlike form of the spi- exine beset with densely spaced, minute pro- nules not easily discernible due to the cesses. Intine 1.5-2.5 um thick, appearing surrounding densely spaced processes of stratified. Volume 75, Number 1 1988 Raj & van der Werff 149 Pollen Morphology of Neotropical Lauraceae FIGURE 10. region and as dark dots in the middle. —B. The sa A, B. Aniba burchellii.— A. Entire polle grain ranas spinules as white dots in the perm me in optic al cross section. LM x 1,000.—C. A. riparia X A kappleri. [n of exine showing surface details. SEM x 11,000. D—F. Endlicheria endlicheriopsis. — D. Entire pollen grain showing spinules as white do gain in optica aced processes. SEM x 11,000. Endlicheria serica. Figure 11A. TEM observation: pollen wall about 3 um thick, differentiated into an outer, very thin, elec- tron-dense ornamented exinous layer com- posed of globular to granular elements, these appearing to fuse to form a thin tectum, and of minute, pointed, solid spinules 2 um apar with a distinct basal cushion intruding deeply into the underlying layer. The subtending ts in the peripheral region and a cross section. LM x 1,000.— F. Part of exine showing spinules and the surrounding densely s dark dots in the middle. —E. Two pollen assive intinous layer loosely granular throughout, densely spaced with radially ori- ented channels partly filled with osmiophilic material, its inner margin more or less smooth. Licaria. Figure 11 B-D. Diameter of pollen grains 17-33 um. Ex- ine 0.5-1 um thick, scattered with spinules 150 Annals of the Missouri Botanical Garden FIGURE 11 b ea of granular to globular elements, which a — A. Endlicheria serica. Part of pollen wall showing an outer thin, — sa dense exinous layer to form a thin tectum. Note traces o zy covering" e of spinules and exine (arrow). Intine loosely yea Pis dri o ii radially oriented n the surfac als partly aes ek osmiophilic material. TEM grain snot VINE spin white and dark dots. —C. = 000 Licaria armeniaca. — B. Entire pollen in spica d section. LM x1,000.—D. L 000. capitata. Part of exine diosa spinules and granular pistes SEM x11 1-2 um apart of faintly discernible twisted strands. Spinules < 0.5 um long, pointed, with a basal, circular, cushionlike form, sur- rounded by a ring of densely spaced, partially fused processes of varying size and shape. The general surface of exine beset with mi- nute processes. Intine 1-3.5 um thick, ap- pearing homogeneous. Licaria triandra. Figure 12A. TEM ob- servation: pollen wall 2 um thick, consisting of an outer, very thin, electron-dense, orna- mented exinous layer composed of scattered or irregularly clustered granular elements and minute, pointed, solid spinules 1 um apart with a distinct basal cushion intruding into the underlying layer. The subtending massive Pollen Morphology of Neotropical Lauraceae Volume 75, Number 1 Raj & van der Werff 151 8 FIGURE 12.—A. Licaria Miedo Part of pollen wall showing an outer thin, electron-dense exinous layer (arrow). The underlying massive intinous layer differentiated into an outermost thick iini this fibrillar in composition with nba A narrow channels and with vacuoles filled with osmiophilic material (a); the second layer structurally loose and fibrillar, sprinkled with osmiophilic material (b); the third layer densely granular and sprinkled with osmiophilic material (c); the innermost layer denser than the other layers, extremely irregular in outline, structurally compact and granular (d). TEM x 20,000. B-D. Kubitzkia mezii. — B. Entire pollen grain, spinules appear as dark dots (lower focus).—C. The same in optical es section. LM x 1,000.— D. Part o exine showing spinules and the surrounding granular processes. FEM x 11,00 152 Annals of the Missouri Botanical Garden intinous stratum made up of 4 distinct layers of varying composition. The outermost rela- tively thick, transversely organized, compact fibrillar layer with scattered, narrow channels and vacuoles partly filled with osmiophilic ma- terial. The second layer equally thick, less electron dense, structurally loose and fibrillar, sprinkled with osmiophilic material. The third layer is thinner, densely granular in compo- sition, much more electron dense, and sprin- kled in places with osmiophilic material. The layer next to the cytoplasmic boundary ex- tremely irregular in outline, in places very thin or absent, and also much denser than the other layers, structurally compact and granular. Kubitzkia 5, Figure 12B- Diameter of pollen grains 24-27 um. Ex- ine 1 um thick, scattered with spinules 2 um apart of easily discernible twisted strands. Spi- nules 0.5 um long, pointed, with a prominent basal, circular cushionlike form surrounded by a ring of densely spaced, partially fused processes of varying size and shape. The gen- eral surface of exine beset with densely spaced minute processes. Intine 3 um thick, ap- pearing homogeneous. Litsea. Figure 13. Diameter of pollen grains 27-35 um. Ex- ine 1 um thick, scattered with spinules 3 um apart of not easily discernible strands. Spi- nules < 1 um long, pointed with a basal cushionlike form, circular to irregular in shape and devoid of processes, partially surrounded by a ring of densely spaced, minute processes. Intine 4 um thick and appearing homoge- neous. Litsea glaucescens. Figure 13D. TEM observation: pollen wall 5 um thick, its outer, very thin, electron-dense, ornamented exin- ous layer composed of scattered or irregularly clustered granular to globular elements and of minute, pointed, solid spinules 3 um apart with a distinct basal cushion intruding into the intinous layer. The underlying massive intinous layer distinctly distinguished into a thicker outer layer inlaid with channels and vacuoles filled with osmiophilic material and into a thinner, denser, homogeneous layer with a slightly uneven inner margin. Cryptocarya. Figure 14. Diameter of pollen grains 30-33 um. Ex- ine 0.5-1 wm thick, appears disrupted in optical cross section (in LM), its outer surface much wrinkled. Intine 2.5-4 um thick, ap- pearing stratified. Cryptocarya aschersoniana. Figure 14D. TEM observation: pollen wall about 4.5 um thick, its outer very thin, electron- dense stratum consisting of an uneven tectum supported by densely spaced clavatelike to globular elements and of a supratectal thin coating of compact material. The subtending massive intinous layer distinctly distinguished into an outer thick, less electron-dense, struc- turally loose, transversely organized fibrillar layer inlaid with scattered vacuoles, these partly filled with osmiophilic material, and into an inner electron-dense, structurally com- pact, transversely organized fibrillar layer in- terrupted by long, narrow channels filled with osmiophilic material. These channels travers- ing the entire thickness of the layer, conical in shape, and having an outlet into the layer above. The intinous layer next to the cyto- plasmic boundary denser than the other layers and in places beset with narrow, short chan- nels filled with osmiophilic material. Cassytha. Figure 15. Diameter of pollen grains 22-28 um. Ex- ine 0.5 um thick, its surface scattered with minute, spinuloid excrescences. Intine 4 um thick, appears homogeneous. DISCUSSION GENERAL POLLEN-MORPHOLOGICAL FEATURES The pollen grains of Lauraceae are inaper- turate and more or less spheroidal. Kasapligil (1951), however, found monocolpate pollen Raj & van der Werff 153 Pollen Morphology of Volume 75, Number 1 19 g Neotropical Lauraceae water — A. Entire pollen grain showing spinules as white and dark dots. —B. The same in optical cross section. LM x 1,000.—C. Part of exine showing spinules and their basal cu ne ns. SEM x 11,000.— D. Part of pollen wall showing an outer very thin, electron-dense exinous layer composed of scattered or irregularly clustered granular to globular elements (arrow). Intine distinctly differentiated into an outer, thick stratum inlaid with channels and vacuoles filled with osmiophilic material (a) and an inner thin, denser, homogeneous layer (b). TEM x 20,000. FIGURE 13. Litsea glaucescens. 154 Annals of the Missouri Botanical Garden FIGURE 14. Cryptocarya UE a E Entire qa grain showing the smooth exinous surface The ded in M ae al cross sect LM x1 —C. Part of exine showing the smooth and wrinkled D SEM x 0.— D. Part of nea wall idus a thin, electron-dense layer Ao of tectum (T), supported by e to globular elements (e), and covered a thin coatin compact material (arrow). The subtending intinous layer distinguished into an outer thick, fibrillar layer inlaid with vacuoles (v), followed by Volume 75, Number 1 1988 Raj & van der Werff Pollen Morphology of Neotropical Lauraceae 155 F ;15. Cassytha filiformis. — 4. Entire pollen grain showing uneven exinous bye beset with spinuloid processes. SEM x5,200.— B. Part of the same at a higher magnification. SEM x1 grains in Umbellularia californica (not con- firmed by the present investigation) and acol- pate grains in Laurus nobilis (not included in the present study). Also Markgraf & D'An- toni (1978) described the pollen grains of Nectandra angustifolia as tricolpate, prolate spheroidal (not included in the present study). e pollen grains of this family are con- sidered to be “delicate”” since they disinte- grate readily or come out in a more or less shriveled and wrinkled condition after ace- tolysis. They are tenui-exinous and, as TEM observation shows, entirely ectexinous. The exine is provided with spinules or spinuloid projections, or rarely the exine is smooth as in Cryptocarya aschersoniana (Heusser, 1971, described the exine of Cryptocarya alba as foveolate). The intine represents the dominant, massive layer of the pollen wall. Out of the 80 taxa investigated, the largest pollen grains are those of Beilschmiedia miersii, Cryptocarya aschersoniana, Oco- tea calophylla, O. spixiana, Persea amer- icana, P. caerulea, P. fulva, and others (over 30 um in diameter); while the smallest belong to Aniba burchellii, Endlicheria glomerata, Nectandra purpurea, Ocotea cernua, and Persea veraguasensis (under 20 um in di- ameter). In the remaining taxa the diameters are 20-30 um. The grains are usually mono- morphic; however, dimorphic grains occur in Persea caerulea and P. schiedeana. ome species of Persea have dimorphic spinules. Because Persea shows various stages of reduction of the number of fertile stamens and/or anther cells (from the normal pattern of nine four-celled stamens to six four-celled and three two-celled or six four-celled stamens and three staminodia or nine two-celled sta- mens; Kopp, 1966), it would be worthwhile to investigate whether in species with dimor- phic pollen grains each stamen has only one kind of pollen grain and inner and outer sta- mens have different pollen grains or whether each stamen possesses both kinds of pollen grains. e an electron-dense, structurally compact, fibrillar stratum inlaid with n arrow, conical channels ipee the innermost layer denser than the other layers and likewise inlaid with channels (white arrow). TEM x 20,000. Missouri Botanical Garden Annals of the 156 yede un g so[nuids wede un c'() sə5e|d ur *j1ede wr g so[nuids wede um e-z so[nuids yede wn e-z se[nuids sase¡d ut pared Aq[euorse2oo ede um c'z-7 sapnutds yede um [ so[nuids yede un g so[nuids yede un c'() sajnurds yede um | 'pe[duruugjur jun[q pue pajutod ‘orydiounp so[nu -ids (q) 2tqdaourouour sə[nutds (e) otydiounp sute18 uojpod yede uri *pə[Burur:iəjur *jun[q pue payutod ‘orydsounp se[nutds yede un | “papmolo seoe|d ur ede um ç ‘orydiounp saoe[d ut so[nutds yede um ç ‘pajutod pue uy} ‘pews 10 junjq *xorg ‘ews sap[nuids :sepnutds S2rudiourp utu (q) sa[nutds orqduourouour qua (e) orydiounp sures3 uə[|od yede un Z ‘pa[sunusajut *jun[q pue pajurod ‘orydiounp sə[nuids 199 186/1€9 Cd Cd c» CGN (e c V v 9£-T£ IV- 08 vr-6c 6t-t£ 9£-0€ umpp»anu [y snanppizo[y pjnpuod `g ns4aiu cg Dipolrut y2s]iog aeaunporurgosieg, equiqng 119I50f 7) sisdouydvporsv’) ny jus piudydiod ` DUDIIXIU ` D1]0f119191]ə1/ ` pupyəununy ` LALLA DUD2140]S02 ` 2904 J SISUISVDNIDLIA `d DUDIPALYIS `d nsimui cq vanf qd DINLIDI `d DUDILIWD `] DISLIg aeautasiaq equiqng Əəpəəsiəd equ], aeaprouneT Á[rurejqng SxX1euray UMNUIXEIA (um) 1ələu mg 'sajuids fo 12qumu umunuiu pup umunxpu (c) pub :saanuids fo q18u3]. (p) ‘ssauyory? ou (£g) ‘ssəauyoryi 2u1xa. (7) ‘suivi? uəyod fo 13j12umip. (I) :pasoduos asp s121204012 uajJod uu 301, 2D22D4n»'] 10f vyp jpoidojoydíour-uo]o,] `p AAV, 157 Raj & van der Werff Pollen Morphology of Neotropical Lauraceae wede um [ sapnuids SS'I EOZ €'0 Y I LE-ZE 12]1u22 "NI nede ur g sop[nuids O12 986 I € I ££-87 M2up1D ^N ¿Siudiounp :papmozo saoe[d ut quede um 5-g səmuıds ¿G1/127 061/8€€ S'o ez S'O — 62-9c pijofto]of N yede um z so[nuids CIO Z9L ço > Z I 6Z-9Z DN wede um e-c'[ so[nuids bbl ISZ €'0 c I SZ-61 DIIDILOS `N nede ur [ so[nuids 60€'T 60€£'I so> £ SO 03-81 DIO)fISS12 CN yede wun "d se[nutds IZS c9L I > CS I 66-e SUDIIDA N nede wr g se[nutds Lg 012 S'I Z SO — ££-8c suəs1qum Nj wede wn ç sapnuids 8£v IZS s0> SI SO 2-2 DijofimoD `N DIPUDIIIN/ mede ur | so[nutds POZE 6v6'€ €0 € I ££-0€ anya () wede un g so[nutds 999 c9L Z S SZ 62-17 psoja Y) mede um | sapnuids — 000'T < co > z I 07-81 sisuan3v19a () wede um 1 sə|jnuids 9977 Evo I z I LZ-S7 nznpuo| Y) rede um p sapnurds 19 bre ë € c 6£-v£ vunixids () i1ede um g so[nuids IZS 199 so> + I L&-Vc mpoimys “O wede um g'g sep[nuids CCP LSS I ST c leig pjjayo]nd `O nede um £ so[nutds IEZ 267 I ec S'I LZ-vG pynsaqnd ) nede un g sopnutds ZOE IZS «902 z I v2-07 n5uo]qo () nede wn g so[nutds IZS 319 I z çI vc nuupupny () sure1d Maj e ut uaos seoie oaxmo1od Jo ayysndjoos yeam “juaredsuen ‘yede um [ so[nuids 816'1 NA €02 € I 9gz-ec Dn O ede wr z-[ se[nutds z9g/0Gt'| Z19/1SPZ I ra ST 9-02 sIiSUJUDINA `Q wede um g-z sends 127/719 c 19/826 SI Š € 87-97 pyofisua Y wede um g se[nuids 719 Z9L I sz €T 67-97 auudppo4puap `Q ae¡nsur uo papMoso Apuanbax] ede un z so[nuids Ice 099 I € I Lc-vG paidn2 `Q 1ede un [ sapnuids TA | 99z'c I c I Gee Dsoqui&402 `Q irede um [ se[nuids LYO'I 60€£'1 902 B I 61-21 Dnu () nede wr £ so[nuids 98€ Z19 I € I 6£-T£ pjj&ydojpo () DION) SBIUMLIOUIBUUTO) equiqngo 9E£9UIOUIEUUT)) aqu] sx1euroy ununu UINUITXB JA] (un) (urm) (urm) (ur) səmuıdg əuyu; — eutxq 19]9ul “eq Volume 75, Number 1 1988 penuguo) “y 318v] Annals of the 158 Botanical Garden issouri M 1ede um g'o se[nuids PILE LESS 902? G $0 61-91 19d doy x muvdu y uəəs sguruado re[ngouut 10 aytesod €-% wunsip sure18 uə[|od may e ur ede wr | sapnurds OST'I 816'I S0> CI NU £6-0c opman `F nede um [ se[nuids SSZ'I £v9'c 90» CI S'O L6-66 OHNE E irede um q sapnuids vVLUI OST'I 90^ çI S'O 03-81 oi ia y Dqiup yede um z sapnuids 19€ 6Lt ço > € I €Z-0Z s1049U14] `F nede wr | so[nuids 60€'I 66€'I £0 - El S'O Ic-6l is p e[qiu1e2sip so[nutds jo ploasp seae juaredsuen ayyasod sede um g-z səmuids +61/8€£p BEE/ZOL ço > € I 62-Ze $15U3214018502 `F DINO IRIUIQUY aquiqng yede um ç so[nuids oge IZS I CZ ce 9e-Ig£ DIVULOfIDI ` ) D14D]m]]əqu ) wede um z-[ se[nutds 66£/66S'1 135/8803 | S0» I I vo- TG ia els y jetisysea x suede wn [ so[nuids 66C'1 8I6'I co» cz co té sipupn4g `y auydppouowa|sopoy y 1ede wn ç so[nutds O12 S18 I sal €'0 0£-87 deg wede ur q so[nutds SS'I 880°C Ske. ot NU Y6-66 aene d xede wn [ sapnurds 66€'I 880°C s0> SI S'o v2-12 unzojisuap ‘q WUMAIXYI04NI] J wede um g so[nutds 62Y 619 S'0 > € I 9c-tc STAN wede um g sapnurds 8tt 099 I c I e DIJOJIDI]DS `v yede um g se[nutds 099 :18 I c I Of-12 D]D]n21]24 `N yede wun [ so[nuids PLU 60€'1 ço > z I 61-81 paindind ^w ede wn g'g sapnurds 18S 876 I c €'0 +Z-61 piqnpsad `v yede ur 1 so[nuids OSE 99z'z ç:0 > rA I CZ-0Z D32DUDAQUI2UI ^N yede ur z so[nuids OZ€ ger I Z I 22-61 SISUDDIUTPADUL ^N mede um [ səluids 60t'I 816'I S'O G I t£6-61 peus N nede wn g sapnuids 319 128 I sE S'I TE-93 Beene A 1ede um ç so[nuids ISZ E67 I ce I LZ-SZ DIOYIPUDIZ ^x mede wn g se[nutds 66€ IZS I sz S'o Z-Z D$0Q0]2 `N sx1euray ununu UIDUIIX?]A (urm) (urm) (un) (un) sajnuidg eunu[ Əumg PyIIDəƏuI “eg pennae). `+ AL 159 Raj & van der Werff Pollen Morphology of Volume 75, Number 1 1988 Neotropical Lauraceae so2uo -3sə19xə ojnutur *pooeds Á[osuap UMA 19səq JINS eutxo — -— co v I 82-22 snuaofinf 77) DYIASSDI) IBIPIOYIÁSSB) Aqruregjqng paidni -191ut Suureadde 398]MS sy ‘pagum JO uloours əurxə == == juasqe t c'o ££-0€ DUDIUOSLIYISD `) nXan201d&17) oeaut&re20]d&1) equiqng oeoÁ1e201dÁg)) aqu] yede ur ç sapnuids £6Z €6b i p I CE-!Z suaosaonp]2 `] DIS] Əəeəutəs1r] equiqng areas] equ] mede um g sands 179 199 S'o £« I- 46-6 nzaw `y PRESTA nede ur q sapnuids LVO'I 816'1 c0» SI S'O £2-11 pipun `T uede wr z-[ se[nutds — 674/996 +vSL'1/9903%Z S0> SE S0 Sc-t£6 nad 7T wede un z-[ se[nuids sos 000'T < c0» SZ I Sc -0cC pipndv2 7] j1ede um I sa¡nuids CEO'S 6v6'€ c0 € I ££-16 DIDIUIWAD “J DIADIV] wede um [ se[nuids OST'I 9977 902? I I S6-0c6 Bee Y yede un q sapnuids 826 OST'I 6G: el c'o 0Z-91 nuunuissa] ` i1ede um g so[nuids [cc 099 I ç I L6-vc SIUE | yede um g'o sa[nuids c6lUt Z20'L c0 > c c'o ZZ-LI nsiddaod :dsqns pipjnoiivd q ede um [ so[nuids SI8 60€'T S'O eG S'O 61-SI DIDIDUO]S ^ ede wn q sapnuids LVO'I 66€'I co > z c'o -L sisdo119Y9puo `q Di42121]pu3] sx1euroy umutu UINULEX® J] (vr) (ur) (um) (un) sgnuidg — ounu[ Əumwq IU AE Jenuuo) “y AVE, 160 Annals of the Missouri Botanical Garden The exine ornamentation in all the taxa (except Cryptocarya) consists of solid, ap- parently simple spinules or as in /Vectandra reticulata, Ocotea cuprea, Persea ameri- cana, Phoebe porphyria, and others, “wick- like" i.e., made up of twisted strands. They are pointed or blunt, uniformly spread over the surface or rarely, in some portions of a grain, in groups; the spinules are sometimes crowded as in Nectandra ambigens to give a rosettelike appearance, or they occur in pairs. Similar wicklike spinules have been re- ported in many groups, e.g., Euphorbiaceae (Thanikaimoni et al., 1984), Linaceae (Punt & den Breejen, 1981), Monimiaceae (Walker & Doyle, 1975) and Plumbaginaceae (No- wicke & Skvarla, 1977). The density and morphology of spinules differs in different species. Thus the highest number of spinules per grain is met with in Persea and Licaria, and the lowest in Beilschmiedia, Caryodaphnopsis, and Lit- sea. Further, the size of the pollen grains does not seem related to the number of spi- Spinules are usually monomorphic; however, dimorphic spinules have been found in Nectandra falcifolia, Persea caerulea, and P. fulva. Spinules are usually broad at base. Their common characteristic feature is the presence of a conspicuous, circular, cushion- like base which invariably protrudes proxi- mally; its surface is either smooth or studded with granular to globular elements. In Ocotea cuprea these cushionlike structures are rel- atively big and compactly arranged, thereby giving a more or less rounded insulaelike ap- pearance to the exine. The stratification of the sporoderm pre- sents interesting features that at times are difficult to interpret. A strikingly similar type of statified sporoderm has been reported in the Zingiberales and in certain monimiaceous genera. The fascinating aspect of the pollen wall is the extreme thinness of the exine and the thick, structurally complex intine. The fine structure of the lauraceous genera investigated shows an electron-dense, tremely thin, coherent exine made up of gran- ular, globular, or irregularly shaped elements nules. ex- and of spinules. Only in Persea fulva (Fig. 1C), Endlicheria serica (Fig. 11A), and Cryptocarya aschersoniana (Fig. 14D) is there a clear indication of parts of well-defined tectum supported by a well-defined bacular layer. In Umbellularia californica the exine surface and spinules are covered by a distinct, superficial coating referred to as "fuzzy sur- face coating" by Rowley & Skvarla (1986); traces of this material are also discernible in Aiouea trinervis, Ocotea puberula, Persea fulva, and Phoebe helicterifolia. The coherent nature of the thin exine, which is emphasized in the present study, has often been compared with exines of some members of the Zingiberales (Kress et al., 1978; Stone i al., 1979) and reported as incoherent or *exine-less." The incoherent nature of any exine, however thin, or the conception of an “exine-less” pollen wall, both from a morphological and functional point of view, has been convincingly argued and re- jected by Hesse & Waha (1983). The dominant stratified layer of the spo- roderm is the intine. This thick layer is com- pletely destroyed by acetolysis and, in keeping with common wisdom, has to be interpreted as intine. In this layer an outer channeled zone can be easily recognized and has been designated variably: "lacunes" after Le Thomas, 1980; “tubules” according to Sedg- ley, 1979; “onciform zone" Rowley & Vasanthy, 1980; **Zwischenkórper" in Kress & Stone, 1982; or simply outer intine or intine 1. These channels, which are long and narrow and of varying length, are predomi- nantly radially aligned and partly or com- pletely filled with osmiophilic material. Thin partition walls as encountered in Strelitzia reginae (Musaceae) (Hesse & Waha, 1983) are absent. It is postulated that these channels may act as deposits for material involved in pollen germinating or incompatibility pro- cesses (Stone et al., 1979; Hesse & Kubitzki, 1983). It is remarkable that the characteristic spo- roderm stratification encountered in laura- ceous genera, namely extremely thin exine and a thick, stratified, channeled intine, is Volume 75, Number 1 1988 Raj & van der Werff 161 Pollen Morphology of Neotropical Lauraceae apparently limited to inaperturate pollen grains belonging to such disparate genera as Canna, Heliconia, Hernandia, and Palmeria. Further, the massive channeled intinous zone mentioned above has been found also to resemble to a lesser extent that of the conifers (viz., Larix europea, Cupressineae, and Tax- odineae) that Wodehouse (1932) noted and remarked as “one of the most remarkable examples of convergence I have yet encoun- tered among pollen-grain forms." The functional significance of the thick- ened intinous layers or onci frequently ob- served beneath apertures appears to be to act as storage areas for gametophytic incompat- ibility proteins (Heslop-Harrison, 1976). TAXONOMY There is uniformity of opinion among tax- onomists that the family Lauraceae is rather old and probably derived from primitive mem- bers of the Monimiaceae of the Hortonia- type. In Kostermans's (1957) classification, only a sequence of floral characters are adopted; not a chronological family tree, nor a pa- leontological succession, nor a phylogenetic trend has been envisaged, but similar genera are grouped. Palynologically most of the relationships are justified, e.g., the genera Persea and Phoebe of the tribe Perseeae are satisfactorily included in the subtribe Perseineae, and thus separated from Beilschmiedia and Mezilau- rus of the subtribe Beilschmiedimeae. The pollen grains of Persea and Phoebe have the highest number of spinules per pollen grain, and their anthers are four-celled compared with Beilschmiedia and Mezilaurus, which have the lowest number of spinules and have two-celled anthers. The former two genera are easily separated from each other because the pollen grains of Persea are either dimor- phic or if monomorphic, the spinules are di- morphic, whereas the pollen grains of Phoebe are wholly monomorphic. Among the Central American species of Phoebe, P. mexicana stands clearly apart due to its high number of spinules. This pollen feature supports the recognition of P. costaricana as distinct from . mexicana. The genera Beilschmiedia and Mezilau- rus can be separated by the difference in size of spinules and in the surface of the basal cushions and exine. Pollen-morphologically, Mezilaurus does not seem to be close to Li- caria. Caryodaphnopsis, although similar to Persea in floral and fruit characters, has pol- len grains quite unlike those of Persea and Phoebe. On the other hand, the pollen grains are comparable to those of Beilschmiedia, and recognition of Caryodaphnopsis is sup- ported by pollen morphology. Its position in Richter's (1981) classification seems to be more appropriate than in those of Kostermans (1957) and Hutchinson (1964). Kostermans (1957) divided Ocotea, of the tribe Cinnamomeae, subtribe Cinnamomi- neae, into three subgenera: Ocotea, Nectan- dra and Pleurothyrium. His contention that macromorphological differences, such as in the position of the anther cells, size and shape of the staminal glands, and cupule shape, are not big enough to maintain them as distinct genera does not improve the classification. Pollen-morphologically there are enough dif- ferences to recognize them as distinct genera. Pollen grains of Ocotea and Pleurothyrium are apparently very similar, including the or- ganization of spinules, which suggest close relationship. However, the characters that distinguish them are pollen size and ultra- structural details of sporoderm. Pollen grains of Nectandra are quite distinct with respect to exine surface and to sporoderm details. There seems enough micro- and macromor- phological characters to maintain these as three distinct genera. The transfer of the two species Nectandra grandis (Fig. 6D) and and N. kunthiana to the new genus Rhodoste- monodaphne (Rohwer, 1986) seems justified, as the pollen grains are quite different from those of the other species investigated. The spinules in both species appear vestigial with- out the usual twisted strands easily discernible in other species. 162 Annals of the Missouri Botanical Garden The monotypic genus Umbellularia is con- sidered to be related to Litsea. Pollen-mor- phologically such a relationship seems to exist as the pollen grains of the two genera are more or less of the same size, the number of spinules per pollen grain is more or less the same, and the ultrastructural details of the pollen wall are very similar. Subtribe Anibineae of tribe Cinnamomeae is represented by seven genera, of which the pollen grains of five, Aiouea, Aniba, Endli- cheria, Licaria and Kubitzkia (Systemono- daphne sensu Mez), have been investigated. These five genera are rather closely related, and taxonomic problems still remain unsolved in this group. Aiouea, recently revised by Renner (1982), appears to include species independently derived from four-celled ances- tors. Some Central American Aiouea species are morphologically very similar to sympatric Ocotea species (van der Werff, 1987) and are much less similar to the Guyanan- Bra- zilian species group which includes the type species. The observation that Ocotea ira and Aiouea costaricensis both possess porelike, transparent areas devoid of spinules on the pollen wall (otherwise a very rare character in Lauraceae) strengthens the theory that Aiouea costaricensis is more closely related to sympatric Ocotea species of the O. insu- laris group (as defined by Rohwer, 1986) than to the S. American Aiouea species, and that Aiouea is a polyphyletic genus. Kubitzki (1982) considered Aniba and Licaria closely related but had no hesitation in maintaining them as separate genera, a point of view we share. MacBride's (1938) suggestion to treat Licaria, Kubitzkia, and Endlicheria (the only dioecious genus in this group) as subgenera of Aniba has never gained acceptance. Ku- bitzkia is probably closely related to Licaria, but is easily separated by the number of fertile stamens. Pollen morphology also supports the inter- relationship of the above genera, and at the same time distinguishes them from one another. All genera have pollen grains of more or less the same size and exhibit the highest number of spinules (excluding Kubitzkia) per pollen grain. The highest number of spinules is found in Licaria. There is a gradual de- crease from Endlicheria to Aniba to Aiouea. Kubitzkia, with a much lower number of spinules per pollen grain (661/521), seems to be a misfit in this subtribe, and pollen morphology fails to support its taxonomic re- lationship with Licaria. Better placement would be in the vicinity of Beilschmiedia as classified by Hutchinson (1964). In Aiouea the basal cushion of spinules is more or less smooth or in places surrounded by a ring of partially fused processes. In An- iba the basal cushions are not very pro- nounced, and the general surface of the exine is studded with granular processes. In End- licheria densely spaced processes surround and hide the basal cushion of spinules. In Licaria the basal cushions are smooth and surrounded by a ring of densely spaced, par- tially fused processes of varying size and shape, and the general surface of exine is studded with densely spaced granular processes. In Kubitzkia the basal cushion of spinules is prominent and surrounded by a ring of dense- ly spaced, partially fused processes of varying size and shape: the general surface of exine is studded with densely spaced, minute pro- cesses. Another group of pollen morphological features that distinguish these genera (except - Kubitzkia, not investigated by TEM) are the ultrastructural details of the sporoderm. In the classification proposed by Koster- mans (1957), Litsea is included in subtribe Litseineae of tribe Litseeae. It is closely re- lated to Umbellularia, and pollen morphology justifies this relationship. Pollen morphology would thus support the transfer of Umbel- lularia to subtribe Litseineae, corresponding with the classification of Hutchinson (1964) and Richter (1981). Cryptocarya of tribe Cryptocaryeae, sub- tribe Cryptocaryineae, is considered isolated among the neotropical genera. Its isolated position is very well reflected in its pollen grains. Its pollen grains are quite different from those of the other genera investigated. The exine surface is devoid of spinules and spinuloid excrescences, instead it appears smooth and wrinkled. The inclusion of this genus by Richter (1981) is his Group I, along Volume 75, Number 1 1988 Raj & van der Werff Pollen Morphology of Neotropical Lauraceae with Beilschmiedia and Caryodaphnopsis, is not supported pollen-morphologically. Cassytha, of subfamily Cassythoideae, is a parasitic or partially autotrophic twiner. Because of its aberrant habit and ecology, it has often been treated as a separate family, Cassythaceae, but in floral characters it re- sembles Lauraceae and approaches Crypto- carya. The pollen grains of Cassytha are characteristic and unlike those of the other genera except Cryptocarya, which it resem- bles in the wrinkled exine surface but from which it differs by the presence of minute, scattered, irregular projections or low warts. From the above account, pollen characters have been found useful in elucidating rela- tionships of many genera but inadequate to clarify the positions of others. However, in the absence of any clear characters, it is important to assess critically each line of evi- dence and this we have attempted to do for the palynological data. The treatment of Laurales in both classical and modern systems of classification is as a considerably old order. Hutchinson (1964) placed Laurales in his Lignoseae, this con- sisting of woody families. Engler (1936), how- ever, did not consider the order as so prim- itive—in his view the amentiferous plants were the most ancient dicotyledons. Cronquist (1968) regarded the members of Laurales as rather primitive and placed them in the most primitive order Magnoliales. Takhtajan (1969), while formulating his system, took into account the importance of pollen mor- phology and considered the Magnoliales as the most primitive order; he accepted Laura- les as near Magnoliales but slightly more ad- vanced. The inaperturate type of pollen grains of Lauraceae are also encountered in some fam- ilies of Laurales, such as Amborellaceae, Go- mortegaceae, Gyrocarpaceae, and Hernan- diaceae and in Sarcandra (Chloranthaceae), Hortonia, Levieria, Peumus, and Tambou- rissa (Monimiaceae). The rest of lauralean families exhibit apertures that are monosul- cate or disulculate or diporate. It is believed that the inaperturate condition is palynolog- ically more advanced than a monosulcate con- dition from which the former has been de- rived; Doyle (1969) postulated such derivation on the basis of the Cretaceous pollen record. Kostermans (1957) postulated a develop- ment within the Lauraceae from genera with a very shallow floral tube, where the fruit is not subtended by a cupule (tribe Perseeae), through genera with a deeper floral tube, where the fruit at maturity is subtended by a cupule up to one-third the size of the fruit (tribe Cinnamomeae) and genera with a deep floral tube that fully encloses the fruit (tribe Cryptocaryeae). If the pollen-morphological data are interpreted in relation to this view, Persea could be regarded as the most prim- itive genus because of the large number of spinules; and if the reduction of the number of spinules is interpreted as an advanced trait, Cryptocarya, because of the total absence of spinules, would stand out as advanced. Thus these two genera would represent the two extremes, and the remaining genera, be- cause of the intergrading characteristics (size of pollen grains and number of spinules), would fall in between but not necessarily in a phy- logenetic sequence. Cassytha is the only her- baceous genus that is distinct from the rest of the genera because of its habit. The pollen grains exhibit spinules reduced into minute excrescences, and this suggests a major evo- lutionary trend. Pollen morphology might provide additional evidence to support sepa- rating Cassytha into a subfamily of its own. LITERATURE CITED AGABABIAN, V. SH. 1969. Pollen morphology of some ;. IV. Biol. Zhurn. Armen. 22(3): 45-58. [In Russian x . 1973. Pollen grains of primitive angiosperms. zv. Akad. Nauk, Arm. SSR. 169 pp. [In Russian.] ARMBRUSTER, L. & C. OENIKE. 1929. Die Pollenformen Mittel zur Honingherkunftsbestimmung. Karl T. 1858-1863. Darstellung und Beschreibung der offizinellen Pflanzen. Leipzig. CHRISTENSEN, P.B. 1986. Pollen pou studies e Malvaceae. Grana 25: 95-1 CRUEL. L. M. . New Zealand oe studies. I. Key to the pollen grains of families and pee in the native flora. Rec. Auckland Inst. Mus. 2: 280- 53. New Zealand pollen studies. The mo cotyledons. A comparative account. Bull. renner Inst. Mus. 3: 1-91. 164 Annals of the Missouri Botanical Garden CRONQUIST, A. 1968. The Evolution and Classification of Flowering Plants. Thomas Nelson, London & Edin- burgh urgh. Datta, K. & S. CHANDA. 1980. Pollen morphology of a few members of the order Laurales (sensu Takh- DOYLE, 1969. Cretaceous angiosperm pollen i the Atlantic coastal plain and its evolutionary signif- icance. J. Arnold Arbor. 50: 1-35. EDGEWORTH, M. P. 1877. Pollen. Hardwicke & Bogue, Londor ENGLER, "a ` 1936. Syllabus der Pflanzenfamilien, 11th edition. Revised by L. Diels. Gebr. Borntraeger, Ber- in. ErDTMAN, G. 1943. An sei to Pollen Analysis. Ronald Press, New ; —— 1952. Pollen Ms A and Plant Taxon- omy. Angiosperms. Almqvist & Wiksell, Stockholm. 1969. Handbook of Palynology. Munksgaard, Copenhagen. FoREMAN, D. B. & F. B. Sampson. 1987. Pollen mor- Bird of Palmeria wed and Wilkiea hue- ana (Monimiaceae). Grana 26: 127-133 Filled HE, J. 2. Beiträge zi zur Kenntniss des Pollen. . Stettin and ns erlin. GENTRY, A. H. 1982. Neo dd floristic diversity: phyortrphical connections between Central and South Am , Pleistocene climatic luco n or an aie of the Andean orogeny? Ann. Missouri Bot. Gard. 69: 557-593. GHOSH, K. 1977. On the Pollen Esc of Some ymnosperms, Monocotyledons and Primitive Di- cotyledons in Relation to DD and Phylogeny. Ph.D. m University of Calcutta, Calcutta. GRIFFITH, J. A. HENFREY. 1875. The Micro- graphic a a E to the nm = investigation of the structure and nature of mic scopic objects, 3rd oea I (text), " ua) PA don HesLorP-HARRISON, J. 1976. The adaptive significance of the exine. Pp. 27-37 in I. erguson & J. Muller (editors), The Evolutionary Significance of the Exine. Academic Press, London & New Hesse, M. & K. The -poroderm ul- trastructure in Persea, Neoni dra, Hernandia, Go- mortega and some other lauralean genera. Pl. Syst. Evol. 141: 299-311. 1983. The fine structure of the pollen wall in Strelitz zia reginae (Musaceae). Pl. Syst. 98. Huanc, 970. Pollen grains of Formosan plants (6). Taiwania 15: 73-179. 972. Pollen Flora s Taiwan. wan Univ., Bot. Dept. Pre HUTCHINSON, J. 1964. The Cnm = ee Plants, olume E Clarendon Press , M. 195 Pollen Grains of dus Hirokawa ubl. Co o., Tokyo. KasaPLiGIL, B. 51. Morphological and ontogenetic studies of Umbellularia californica Nutt. and Lau- rus nobilis L. Univ. Calif. Publ. Bot. 25: 115- 240 sas E. Leaf Venation Patterns. Lau: e. J. Cie Verlag, Berlin National Tai- IKUSE KneLL, A. K. 1914. Die Pollenkórner als Diagnostikum in Drogenpulvern (Blüten, Kráutern und Bláttern). Wurzburg. er 1 KOLREUTER, J. G 181 1. Dissertationis de antherarum ulvere contin uatio. Sectio quarta figura an therarum pulveris. Mém. Acad. Imp. Sci. Saint Pé- 1966. A taxonomic revision of the genus diei in the ra Hemisphere. Mem. New York . Gard. 14: 1-120. Ee A. J. i H. 1957. Forest Res. Inst. 57: 1-64. Kress, W. J. & D. E. Stone. 1982. Nature of the sporoderm in monocotyledons, with special reference to the pollen grains of c anna and Heliconia. Grana -148. 21: Lauraceae. Commun. —— ——. 1983a. Morphology and phylo- genetic significance of exine-less pollen of Heliconia (Heliconiaceae). uo ot. 8: 149-167. & —— ———. 1983b. Pollen intine structure, cytochemistry and a in monocots. Pp. 159- Ottaviano (editors), 3 in D. L. Mulcahy & E. Pollen: biology and implications for plant breeding. Elsevier, New Yor .C. SELLERS. 1978. Ultrastruc- pu of exine- ps pollen: Heliconia (Heliconiaceae). Amer. J. Bot : 1064-1076. KuniTzki, K. Y The tubular exine of Lauraceae and Hernandiaceae, a novel type of exine structure in id plants. Pl. Syst. Evol. 138: 139-146. 982. Aniba. Flora Neotropica Monogr. 31: Kurz, Fa 1983. Fortpflanzungsbiologie = Gattun- gen neotropischer Lauraceen un ision der Gat- tung v aria. Ph.D..Thesis. Ds of Hamburg, Hambur Le Thomas, A. 1980. Characters of the pollen din of African Annonaceae and their significance for phylogeny of runde angiosperms. Pollen & Spores 22: 267-342 LEwis, W. H., P VINAY & V. E. ZENGER. 1983. Air- borne and allergenic pollen of North Pide Johns Hopkins Univ. Press, Baltimore & Londor Lieux, M. H. 1980. An atlas of pollen of me shrubs, and woody vines of Louisiana and other southeastern states, dd I. dd to Lauraceae. Pollen & Spores 22: 17-5 Mac Brune, E 1938. p In: Flora of Peru. Field at. Hist., Bot. Ser. 13(2): 819-931 ollen p ar spores from Miocene rown coals in Lower Siles i il Wroclawsk. Towarz. Nauk., Series B. 96. MACPHAIL, LM. K. 1980. Fossil a ieee Beilschmie- dia (Lauraceae) pa in New Zealand. New Zealand ~ MARKG RAF, y, & H. L. Dist. of Argentina. Univ. Mrz, C. 1978. Pollen Flora a Press, Tucson, Arizona. 1888. Morphing Sa! über die Familie Lauraceen. Verh. Bot r. Brandenburg 30. HS Berlin.] —— ——. 1889. vea Americanae. Jahrb. Konigl. Bot. a t. Berlin 5: Mirror, N. 1970. Etudes morphopoliniques et des a ho dl sur les “Polyca E elo ave onsidérations pbylogénéiques Bus 1969: % > Lucr. Crád. Bot. Volume 75, Number 1 Raj & van der Werff 165 1988 Pollen Morphology of Neotropical Lauraceae Mont, H. 1834. Beiträge zur Anatomie und Physiologie and E peers avocado stigma and style. J. Cell der Gewüchse. Erstes Heft. Uber den Bau und die Sci. 38: 49-60. Formen der Pollenkorner. aur er und Co., Ber SELLING, O. H. 1947. Studies in Hawaiian pollen sta- NowIcKE, J. W. & J. J. SKVAR 1977. Pollen mor tistics. Part II. The pollen of the Hawaiian phanero- phology and the relationship of the Plumbaginaceae, 1975. Studies in Lauraceae. II. Some aspects of embryology of Cinnamomum cecidodaphne Meissn. Geobios 2: 83-84. Pollen grains of some Lauraceae. J. : 99- P. DEN BREEJEN. 1981. The Northwest European pollen flora. 27. Linaceae. Rev. Palaeobot. Palynol. 33: 75-115. PURKINJE, J. E. 1830. De cellulis antherarum fibrosis nec non de granorum pollinarium formis. C en- tatio Phytotomica. Pressburg (Bratislava). Rao, A. N. & Y. K. LEE. 1970. Studies on Singapore pollen. Pacific Sci. 24: 255-268. Raven, P. & D. AXELROD. 1974. Angiosperm biogeog- raphy and past continental movements. Ann. Mis- souri Bot. Gard. 61: 539-673. RENNER, a 1982. Aiouea. Fl. Neotrop. Monogr. 31: 85-116. ii T G. 1981. Anatomie des sekundären Xy- lems und der Rinde der oe Sonderbd. Na- turwiss. Verh. Hamburg 5: 1-148. RopEWaLD, R. R. bu e. 1974. Porous substructure of the glomerular iet oe in the rat and ae J. Cell Biol. 60: iocbein, J. “sm sapa einer p der Gating a Aubl. (Lauraceae), sensu lato. Mitt. Inst. . Bot. Hamburg 20: 1-278. K. la ase 1985. Entwicklungslinien i in de "ii Rd MNA Bot. Jahrb. Syst. 107: ROWLEY, 7 R & J. J. SkvaRLa. 1974. Origin of the inner intine in pollen of Canna. In: C. J. Arceneaux aisi, 32nd Ann. Proc. Electron Microscopy Soc. . Claitor's, Baton Rouge. 1975. The glycocalyx and initi- ation of exine spinules n ee of Canna Amer. J. Bot. 62: ee . Developmen of the pollen grain wall in a Nordic J. Bot. 6: 39-65. & G. VasaNTHY. 1980. Sporopollinin accu- mulation on the surface coating at Cinnamomum pollen and its influence on Abstracts of the Third International Palynolagical Cae Cambridge. lp. on Abstrac RuboLPH, K. 1936. Mikrofloristische d) tertiarer ie une im nórdlichen Bohmen. Bei Bot. Centralbl. 54: 244 SasTRI, R. L. N. 1958. Studies? in Lauraceae. II. Em- aryology of Cinnamomum and Litsea. J. Indian Bot. Soc. 37 278. 162. Studies in Lauraceae. III. ape a of Cassytha. Bot. Gaz. (Crasifordeville) 123: 206. SCHNIZLEIN, A. -1870. aru ga EE aturalium Regni Vegetabilis, I-IV. SCHROEDER, C. A. 1952. Floral development, sporo- genesis and émbryolog gy in the avocado, Persea amer- icana. Bot. Gaz. (Crewfordsville) 113: 270-278. SEpcLEY, M. 1979. Structural changes in the pollinated gams. Special Publ. Bernice Pauahi Bishop Mus. 38: SKvARLA, J. J. & J. R. RowLEY. 1970. The pollen wall of Canna and its similarity to re germinal apertures of other pollen. Amer. J. Bot. -529. SouMa, K. 1985. Ultrastructure > pollen wall of Lin- dera umbellata Thunb. var. membranacea (Maxim.) “ana Era Sci. Rep. Tohoku Imp. Univ., 987. Romei basis for the con- vergence of Sassafras (Laurales) and Heliconia (Zin- giberales) pollen. Grana 26: 179-192. , S. C. SELLERS & W. J. Kress. 1979. Ontog- n uh exineless ME in Heliconia, a banana rel- ative. Ann. Missouri Bot. Gard. 66: 701-730. ; 81. Ontogenetic and evolutionary implications of a neotenous exine in Tapeinochilos (Zingiberales: Costaceae) pollen. Amer. J. Bot. 68: 49-63 SURYANARAYANA, M. C. & G. B. DEODIKAR. 1978. Co e to the palynotaxonomy of Indian Laura- ae. II. Indian Palynol. Conf. Bangalore. 31. [Ab stract 1980. Contributions to the palyno- morphology and palyno- -taxonomy of Lauraceae from India. Pp. 79-90 in P. K. K. Nair (editor), Advances in Pollen-Spore Research, Volumes 5-7. Today & Tomorrow's Publishers, New Delhi, India Ber M. 1965. pon Peli pisan bein mbrane of trees. Sci j adm Univ. MES 17: :43- 69. ñq A. 1969. Flowering Plants: origin and dis- persal. Oliver and Boyd, Edinburgh. [Translated from G., NI, S. Nitsson & E. RÖM. 1984. Omniaperturate Euphorbi- aceae hi with striate spines. Bull. Jard. Bot. Natl. Belgique 54: 105-125. VELOSO, H. P. & O. M. BARTH. TH RAND, GRAF 1962. bn siste- o Bra e Myristicaceae. Mem. Inst. (wal Cruz 60 9-90. WALKER, J. W. 1976. Evolutionary significance of the exine in the pollen of primitive angiosperms. Pp. 251-308 in I. K. Ferguson & J. Muller (editors), The Evolutionary Significance of the Exine. Academ ic Press, London & New York. & J. A. DoviE. 1975. The bases of angiosperm phylogeny. Ann. Missouri Bot. Gard. 62: 664-723. Wanc, F. H. 1960. Pollen Grains of China. [In Chinese.] 69. i ege study on the pollen grains of Lauraceae of Tai Taiwan Forest Res Inst. 175: 1-36. [In eee WERFF, H. VAN DER. 1987. Six new species of neo- ee Lauraceae. Ann. Missouri Bot. Gard. 7 401-412. D H. G. RICHTER. 1985. Caryodaphnopsis 166 Annals of the Missouri Botanical Garden id Shaw (Lauraceae), a genus new to the Neo- ropics. Syst. es 10: 166-173. Sis ISE, R. P. 1932. Tertiary ao I. Pollen of the living ao UN. of the n River flora. Bull. Torrey Bot. Club 59: 313-340. 19 Pollen Grains. Their Structure, Iden- ar and Significance in Science and Medicine. , New York & London. ZDE, "E. 1935. Pollen Gestaltung und Herkunfts- bestimmung bei Blütenhonig, I. Verlag der Reichs- eu Imker, Berlin. — Pollengestaltung und Herkunftsbestim- mung bei Blütenhonig, II. Verlag Liedloff, Loth & Michaelis, Leipzig. 19 ollengestaltung und Herkunftsbestim- mung bei Blütenhoni nig, III. Verlag Liedloff, Loth & Michaelis, Leipzig. ZINDEREN BAKKER, E. M. van. 1956. South African Pollen Grains ‘and “pores, II. A. A. Balkema, Am- sterdam, Cape Tow LIST OF SPECIES INVESTIGATED Aiouea Aubl. . costaricensis (Mez) Kosterm. — Costa Rica, Harts- horn 1121. A. laevis (Mart.) Kosterm. — Venezuela, 7937. A. trinervis Meissn. — Brazil, Aniba Aubl. A. UN Kosterm. — Brazil, Heringer & Eiten 15 Bernardi Mori 16711. A. canela (Kunth) Mez— Brazil, Heringer et al. 1602 A. emule (Nees & Mart.) Mez — Brazil, Cult. Bot. Gardn. Rio de Janeiro 11030 A. riparia X A. ka appleri— Venezuela, L. Berti Bulle kanqaqa Nees B. miersii (Gay) Kosterm. — Chile, Zöllner 11607. B. pendula (Sw.) Benth. — Panama, Croat 12928. or lal Airy Shaw erff — Peru, Foster 9585. Marcano Cassytha C. filiformis L. — Venezuela, Liesner & Gonzalez 5783. Cryptocarya R. Br. C. asc ^ une Mez— Brazil, Hoehne s.n. Endlicheria Nees E. reia Tos (Mez) Kosterm. — Maguire 24898a. E. glomerata Mez — Brazil, Heringer & Eiten 15154. E. pui (Spreng. Pegs subsp. uma term.) Koch, ined.— Peru, Schunke 2924. E. sericea Nees — Brazil, Raa et al. 1278. E. tessmannii O. C. Schmidt— Peru, Croat 20771. E. verticillata Mez— Peru, Woytkowski 6304 Kubitzkia v. d. Werff . mezii (Kosterm.) v. d. Werff — Brazil, Pires et al. 50919 (NY); Surinam, Irwin 55719 Licaria Aubl. PA armeniaca (Nees) Kosterm.— Peru, Revilla et al. L. cuui pp & Cham.) Kosterm. — Guate- mala, Lundell 15769. L. peckü (I. M. al Kosterm. — Belize, Proctor 36035. = L. triandra (Sw.) Kosterm. — Venezuela, Steyermark et al. 122775. Litsea Lam. L. glaucescens Kunth — Mexico, Breedlove 33 753. Mezilaurus Taubert M. navalium (Allemao) Taubert — Brazil, Glaziou 1212. di inei Roland. ex Rottb. N. VE (Ruiz & Pavon) Mez— Ecuador, Mexia 846 N. “ambigens (Blake) Allen—Mexico, Wendt et al. N. > (Meissn.) Mez— Venezuela, Aristeguieta 7281. N.? cissiflora Nees— Bolivia, Krukoff 10890. N. coriacea (Sw.) Griseb. — Mexico, Breedlove 24564. N. cuspidata Nees— Brazil, Irwin et al. 21038. N. falcifolia ( a ee — Argentina, Krapovickas & lrigoyen N. gardneri TER al Irwin et al. 166 N. gentlei Lundell — Panama, Mori & Kallunki Pm N. globosa (Aubl.) Mez— Panama, Croat 7343. N. grandiflora Nees— Brazil, Hatschbach 32569. N. lanceolata Nees— Argentina, Schwarz 5103. N. loeseneri Mez — Mexico, Brigada Vazquez 1575. N. martinicensis Mez— osta n Hartshorn 1226; Venez uela, Liesner et al. N. ne (Sw.) cd un Rica, Harts- horn 1753. N. a Lundell — Mexico, Hinton 13918. N. purpurea (Ruiz & Pavon) Mez— Panama, Croat 7634. N. reticulata (Ruiz & Pavon) Mez — Mexico, Dorantes 4 N. salicifolia | (Kunth) Nees— Guatemala, R. Tun Ortiz N. put Allen — Costa Rica, Koptur SK-103. Ocotea Au O. calophy lla Mez — Colombia, García- -a 20740. O. cernua (Nees) Mez— Belize, Proctor ¿ d O. corymbosa (Meissn.) Mez — Brazil. Smith et al. 14596. O. cuprea Mez— Peru, Rimachi 4 O. dendrodaphne Mez— Costa Ries Burch 4589. O. ensifolia TEM ds Eiten 10 O. pra Aubl. — ezuela, Sie ermark 117621. ra Mez & Pitt SE ans Lao & Gentry 548. prie Vattimo — Brazil, Handro si Meissn.) Mez— Panama, Croat 16515. ees— Panama, Nee & Gentry 8683. pulchella Mart. — Brazil, Hatschbach 20405. skutchii Allen— Panama, Croat 97 spixiana (Nees) Mez— Brazil, Chagas - 32. tonduzii Standl.— Costa Rica, Haber . veraguensis (Meissn.) Mez — Mexico, 25 Do- rantes 2881. villosa Kosterm. — Brazil, RB208640 O. whitei V duni Mori & Kallunki 5625. Persea Mill. . americana Miller — Mexico, Q. Valdivia 176. P. caerulea (Ruiz & Pavon) Mez— Venezuela, Stey- ermark 104771. — qux "c e o~ Š = = Z $7 Ramalho 1863- . fulva Kopp— Brazil, Irwin et al. 303 P. mutisii Kunth — Venezuela, Luteyn Lebron Lu- teyn 6051 Volume 75, Number 1 1988 Raj & van der Werff 167 Pollen Morphology of Neotropical Lauraceae P. schiedeana Nees — Panama, Hammel et al. 6966. : Lei penai Seem.— Panama, Tyson 6689. Phoebe Nee P. costaricana Mez & Pittier — Panama, v. d. Werff 7315. P. hammeliana Burger, ined.—Costa Rica, Hammel 4091. P. es (Meissn.) Mez— Mexico, Breedlove b; mexicana Meissn. — Panama, Croat 146 P. porphyria (Griseb.) Mez— Argentina, Venturi 7554. P. smithii Allen— Costa Rica, Hartshorn 2130. Rhodostemonodaphne Rohwer & Kubitzki R. grandis (Mez) Rohwer — Venezuela, Blanci 278. R. kunthiana (Nees) Rohwer — Peru, Croat 19790. e td Nees oe A. C. Smith—Peru, Gentry & Revilla 662 P. s n NUN de Bruyn 1422. — Peru, Kayap 14 Labs hoq Nutt. U. californica (Hook. — iy —U.S.A., Cult. California, Berkeley, Axelrod s THE TAXONOMIC SIGNIFICANCE OF POLLEN MORPHOLOGY IN THE COLUMNEA ALLIANCE (GESNERIACEAE: GESNERIOIDEAE)' Karen J. Fritz and Norris H. Williams** ABSTRACT Pollen of 67 species of the five genera of the Columnea alliance A aM ado eae: Gesnerioideae) was examined a sensu stricto has a peculiar pollen type supporting its treatment as distinct ia iu other four genera of the alliance. Bucinellina is also distinct from the other genera, and its status as a us is likei vise supported. Pollen structure does not confirm the distinctiveness of Trichantha, Pentadenia, and Dalbergaria and may be used better at the sectional rather than the generic level. The mainly tropical family Gesneriaceae contains over 2,800 species. A large influx of collections of newly discovered species from the American tropics has cast doubts on some traditional generic limits in the exclusively neotropical subfamily Gesnerioideae. Parallel adaptations to certain classes of pollinators have become apparent and complicate the efforts of taxonomists. Taking this into con- sideration, Wiehler (1983) proposed new ge- neric and tribal limits for the Gesnerioideae and a new subfamily (Coronantheroideae). His work parallels Burtt’s (1963) for subfamily Cyrtandroideae, and it is in the context of Wiehler’s classification that this study was one. The Columnea alliance (tribe Episcieae) is a natural group of closely related genera that share ornithophily as the pollinator syndrome. The alliance comprises over 240 species in five genera: Pentadenia Hanst. (ca. species), Dalbergaria Tussac (ca. 90 species), Trichantha Hook. (ca. 70 species), Bucinel- lina Wiehler (2 species), and Columnea L. (ca. 75 species, Wiehler, 1983). One of the main taxonomic problems in the Episcieae is the disposition of these taxa. Should species distributed variously in several groups for- merly regarded as sections of Columnea in the broad sense be given generic status, or should they be treated as subgeneric groups of Columnea? Morley (1976), who based his conclusions mainly on cytological and hy- bridization results of Sherk & Lee (1967) and Moore & Lee (1967), disputed treating them as distinct genera. Morton, who worked with Trichantha and Columnea for a number of years, first lent support to the concept of the We are grateful to Hans Wiehler for providing plant material and to the Marie Se elby Botanical Garden Sarasota, ru grac tously allowing use of the , with the SEM. Gratitude is due William L. Stern part on a thesis ned to Florida Science degree by K.J.F. Grant-in-Aid-of- Research from Sigma Xt (to ? Department of Zoology, University o int We comments on the State U hiversity in partial fulfillment of the requi r of The research was funded in part by NSF grant DEB 791 1555 o N.H. W.) d by a K.J.F. thank Willia m Miller for technical dens manuscript. This i J "California, Davis, California 95616, U.S. A. 3 Department of Natural Sciences, Florida State Museum, University of Florida, Gainesville, Florida 32611, 4 * To whom all correspondence and reprint requests should be sent. ANN. MISSOURI Bor. Garp. 75: 168-191. 1988. Volume 75, Number 1 1988 Fritze & Williams 169 Pollen Morphology in Columnea genus Trichantha (Morton, 1963) but later reversed his opinion (Morton, 1971). He found that the appendages in the sinuses between corolla lobes that distinguished his concept of Trichantha from other taxa were rudimen- tary throughout the complex. However, Wiehler (1983) maintained that on the basis of several important characters, Trichantha and the other four genera are as distinct from one another as from any other genera in the tribe and therefore should be given generic status. Among the characters Wiehler (1973) considered most important in arriving at ge- neric limits within the Columnea alliance and in other Episcieae are fruit type, number of nectary glands present and their degree of dorsal connation, plant habit, and corolla shape. The importance of corolla shape is a source of debate in the classification of the Columnea alliance. Within this group, ge- neric limits based on corolla morphology are vague. Although he used it as a convenient means for separating certain genera, Wiehler de-emphasized corolla shape in his overall classification of the subfamily, since it appears to be an adaptation to a pollinator class and is not necessarily indicative of close relation- ships among taxa. In other words, it is of taxonomic use in separating species but con- tributes little to an evaluation of systematic relationships. Morley (1973, 1974, 1976), however, believed that degree of corolla zygo- morphy is correlated with degree of crossa- bility and is therefore important to classifi- cation in the Columnea alliance. Five separate nectary glands distinguish Pentadenia from the rest of the Episcieae. All other genera in the Episcieae possess glands which vary in degree of reduction and dorsal connation even within species (Wilson, 1974). In the Columnea alliance (except Pentade- nia) the nectary consists solely of a large double dorsal gland. The “fern-frond” habit (or extreme aniso- phylly) makes Dalbergaria distinct in the alliance (Wiehler, 1971). Plant habit varies within other genera of the Columnea alliance. Fruit type in the Episcieae is a valuable suprageneric character. It separates the Dry- monia complex from the Episcia complex, the two having different types of capsules, and separates both from the Columnea alli- ance, which has a berry. Within the Colum- nea alliance genera differ with respect to fruit shape and color. Subtle differences in shape create some skepticism regarding its reliabil- ity (Morley, 1976). Published studies of gesneriaceous pollen are mainly morphological and make little mention of the taxonomic implications. Woods (1964) studied 180 species in over 50 genera and made a distinction between Burtt's subfamilies based on pollen-wall sculpturing and grain size. Uniform or homobrochate ex- ines are more prevalent in the Cyrtandroi- deae; heterobrochate exines are more prev- alent in the Gesnerioideae. Grains tend to be larger in the Gesnerioideae than in the Cyr- tandroideae. Erdtman (1966) surveyed 20 species from 17 genera of the Gesneriaceae with the light microscope and supplied a diagram of Colum- nea microphylla pollen. Melhem & Mauro (1973) obtained data on eight local species from three genera in Sao Paulo, Brazil. Based on their observations and those of Erdtman, they concluded that there was great variation of pollen types in the family. They remarked only on morphological distinctions among pol- len of these species and included no taxonomic interpretations. Skog (1976), in his taxonomic treatment of the tribe Gesnerieae, presented a brief dis- cussion of his observations on the grains of 27 examples. He made use of the scanning electron microscope, but little information could be obtained from his micrographs other than sculpturing patterns, since the pollen grains were collapsed. The absence of pollen data from Wiehler's (1973) study prompted a preliminary SEM survey by Williams (1978) of 30 species rep- resenting 19 genera to ascertain whether or not pollen might be of use in further clarifying systematic relationships in the Gesnerioideae. Williams (1978) found a wide variety of forms in the Gesneriaceae, particularly in tribe Epi- 170 Annals of the Missouri Botanical Garden TABLE 1. Species examined. Greenhou Accession Number Species Origin Columnea arguta C. Morton G-979 SEL Panama C. bilabiata Seemann W-2233 SEL Colombia C. billbergiana Beurling W-1166 SEL Panama C. cobana F. D. Smith W-2056 SEL Guatemala C. dodsonii Wiehler W-1500 SEL cuado C. dressleri Wiehler W-2176 SEL Panama C. erythrophaea Decne. G-1057 SEL exi C. flaccida Seemann G-327 SEL Panama C. gallicauda Wiehler W-2179 SEL Panama C. gloriosa Sprague W-2131 SEL nam C. guatemalensis Sprague W-2055 SEL Guatemala C. hirsutissima C. Morton W-2132 SEL Pana C. kienastiana Regel W-1694 SEL Colombia C. kucyniakii Raym. W-2020 SEL Ecuador C. linearis Oersted G-325 SEL Costa Rica C. maculata C. Morton W-2262 SEL anam C. nicaraguensis Oersted W-2641 SEL Panama C. oerstediana Klotzsch ex Oersted W-2270 SEL Panama C. purpusii Standley G-1209 SEL Mexic C. querceti Oersted W-2641 SEL Costa Rica C. repens (Hook.) Hanst. G-920 SEL Jamaica C. rubra C. Morton W-2236 SEL Panama C. rubricaulis Standley W-2328 SEL Nicaragua C. rutilans Sw. G-843 SEL Jamaica C. schiedeana Schldl. 6-725 SEL Mexico C. verecunda C. Morton G-87 SEL Costa Rica C. zebranella Wiehler W-1595 SEL Panama Dalbergaria asteroloma Wiehler W-2247 SEL Ecuador D. aureonitens (Hook.) Wiehler W-1818 SEL Venezuela D. cruenta (Morley) Wiehler W-1146 SEL anama D. eburnea Wiehler W-1704 SEL Colombia D. ericae (Mansf.) Wiehler W-1630 SEL Ecuador D. florida (C. Morton) Wiehler W-1622 SEL Pana D. inaequilatera (Poeppig) Wiehler W-2036 SEL Ecuador D. kahlbreyeriana (Masters) Wiehler W-1590 SEL Colombia D. perpulchra (C. Morton) Wiehler W-1572 SEL Panama D. picta (Karsten) Wiehler W-1794 SEL Colombia D. polyantha Wiehler W-1152 SEL Panama D. puyana Wiehler W-2040 SEL Ecuador D. sanguinea (Pers.) Steudel W-1709 SEL Panama D. sanguinea (Pers.) Steudel G-85 SEL Hispaniola . sanguinea (Pers.) Steudel W-1628 SEL na D. silvarum (C. Morton) Wiehler W-2450 SEL Panama D. vittata Wiehler W-2265 SEL Panama Pentadena augustata Wiehler W-2185 SEL Costa Rica P. byrsina Wiehler W-2138 SEL Ecuador P. ecuadorana Wiehler W-1894 SEL Ecuad P. microsepala (C. Morton) Wiehler W-1837 SEL Venezuela P. nervosa Kl. ex Oersted W-1948 SEL nama P. orientandina Wiehler W-2273 SEL Ecuador P. spathulata iere bane W-1955 SEL Ecuador P. strigosa (Benth.) c W-4128 SEL Colombia P. zapotaliana Wie W-2167 SEL Ecuador Trichantha dus a Wiehler G-804 SEL Puerto Rico Volume 75, Number 1 1988 Fritze & Williams Pollen Morphology in Columnea 171 TABLE 1. Continued. reenhou Species Accession Number Origin T. brenneri Wiehler W-2275 SEL Ecuador T. calotrica (F. D. Smith) Wiehler W-2181 SEL Panama T. citrina Wie W-2451 SEL Panama T. dissimilis (C. Morton) Wiehler W-1177 SEL T. filifera Wiehler W-1631 SEL Colombia T. herthae (Mansf.) Wiehler W-1573 SEL Ecuador T. minor Hook. W-1685 SEL Colombia T. mira (Morley) Wiehler W-1586 SEL Panama T. moorei (C. Morton) C. Morton W-2193 SEL Panama T. parviflora (C. Morton) Wiehler W-1993 SEL Panama T. pulchra Wiehler W-2368 SEL Panama T. purpureovittata Wiehler W-1721 SEL ru T. tenensis Wiehler W-1585 SEL Ecuador Bucinellina nariniana (Wiehler) Wiehler W-1642 SEL Colombia B. paramicola (Wiehler) Wiehler W-1634 SEL Colombia scieae. Fritze (1979) found the diversity of pollen to be of potential taxonomic use in the Columnea alliance, and Williams (1978) in- troduced a successful technique for preparing gesneriaceous pollen that shows expanded, clean grains and allows full observation of all external features. MATERIALS AND METHODS Pollen was collected from mature anthers of living plants grown in the greenhouses at the Marie Selby Botanical Gardens, Sarasota, Florida (Table 1). The pollen was acetolyzed following a modified method of Erdtman (1966) and was stored in 70% ethyl alcohol (EtOH). Slides for vouchers were made from which sizes and shapes of pollen grains could be determined. Measurements of the lengths of the polar and equatorial axes of 50 grains per species of representative species were made at 400 x using an ocular micrometer. Pollen for the scanning electron microscope (SEM) was dehydrated through an alcohol series followed by an amyl acetate/ EtOH se- ries (1:3, 1:1, 3:1, 100% amyl acetate). A drop of suspended pollen was then placed on a round glass coverslip and attached to an SEM stub with double-sided tape. The pollen was air dried under a cover to prevent con- tamination. Dried pollen was coated with 5 nm of gold palladium and scanned with a Cambridge Stereoscan model 54-10 instru- ment at 5, 10, or 20 kV. Fractured pollen was also viewed with the SEM. Pollen was transferred from the original SEM stub to a second stub fitted with a piece of double-sided tape covering the entire surface. Transfer was accomplished by touching the two stubs to- gether until the pollen adhered to the tape. The process was repeated using a third stub with double-sided tape on it, which was pressed against the second stub. The adhesion of the pollen grains to both stubs pulled the grains apart. This procedure requires no elaborate microtomy and yields replicates with no loss of material. The fractured grains were re- coated and scanned. Photographs were taken using Type 665 Positive/Negative Land film with a Polaroid camera mounted on the mi- croscope, and negatives were later contact printed. NOTE ON TERMINOLOGY Terminology used for the apertures, shape, and areas of the surface of the pollen grain follows Erdtman (1966). Since Erdtman’s terms do not extend to the detail of exine sculpturing revealed by the SEM, it is nec- essary to make a precise distinction between the punctate and reticulate patterns found in 172 Annals of the Missouri Botanical Garden the Columnea alliance. An exine is punctate if the tectum is perforate and the majority of the perforations are much smaller in diameter than the width of the muri (remnants of the tectum). The exine is reticulate if the punctae are enlarged (lumina) and at least as wide as the muri. A long aperture extends beyond half the distance from the equatorial margin of the grain to the polar axis when seen in polar view (Fig. 2E). An aperture of inter- mediate length reaches about one-half this distance (Fig. 8C), and a short aperture ex- tends less than half this distance (Fig. 12G, H). Terms describing exine structure follow Walker (1974a) RESULTS AND OBSERVATIONS Several pollen types appeared with some overlap among genera. Types are distin- guished mainly by shape, exine pattern, and aperture length and shape. All grains are monads, isopolar, and tricolp(or)ate, and either punctate or reticulate. Their sizes, based on the length of the longest axis, range from approximately 28 to 51 um. The features are summarized for each species in Table By far the greatest homogeneity turned up in Columnea, in which 26 species were ex- amined (Figs. 1, 2, 3, 4, 5A-C). With the exception of three species, all have subprolate to spheroidal grains (see P/E, Table 2) with uniformly punctate exine, long, elliptic ap- ertures, and circular ambs. There are slight differences in size and shape of punctae and apertures between species. Columnea repens and C. rutilans (Fig. 5B, C) stand apart by having suboblate or oblate grains, exine re- ticulate grading to punctate adjacent to the colpi and at the polar regions, tapered ap- ertures, and triangular ambs. Columnea ku- cyniakii has suboblate grains, exine reticulate grading to punctate adjacent to the colpi and poles, long, sharply tapered apertures, and triangular ambs. The majority of the 15 Dalbergaria species examined (Figs. 5D-H, 6, 7, 8) have grains distinct from those of Columnea. Those of the former are suboblate and have the exine reticulate grading to punctate adjacent to the colpi and at the poles, long, sharply tapered apertures, and angularaperturate, circular or triangular ambs. Dalbergaria aureonitens, D. sanguinea, and D. florida (Fig. 8) do not fit the general pattern; they have pollen sub- oblate to spheroidal, reticulate (as above, but => Fic URE 1.— 4. Columnea schiedeana, equatorial view; X 1,550.— B. C. schiedeana, mesocolpial region; x 7,333.— —E. .. C. verecunda, equatorial view; X 1,550.— D. S. view; 387 H. C. purpusii, mesocolpial region; X7, B. C. verecunda, mesoc olpial region; x 7,333. uerceti, mesocolpial region; x 7,333. —G. C. 333. Ms purpusii, polar view; xL5 FIGURE 2.— 4. Columnea arguta, ee view; X 1,300.— B. C. arguta, at end of colpus; x 5,717.— guatemalensis, kb ipd view; X 1,33. view; X 1,383. —F. C. kienastiana, me meis region; x 6,0 FicuRE 3.— A. Columnea linearis, equatorial view; X 1,567 D. C. C. eae polar view; X 1,433.— region; X 1,333.—F. C. dressleri, mesocolpial region; -. guatemalensis, ur olpial region; x 6,833. bana, mesoc "ipia region; X 7,000 hirsutissima, mesocolpial region; X 6,867.—EF. C. d 6,667.— G. C. dodsonii, equatorial view; x 7 — E. C. cobana, polar —H. C. .—G. C. kienastiana, polar view; x 1,500. —B. C. linearis, mesocolpial pon x /,333.— ssleri, Equatorial 33.—H. C. dressleri, exine fracture through mesocolpial region; x 10,000. FiGURE 4.— A. Columnea gallicauda, im view; X 1,383. flaccida, a view; X 1,48. view; X 1,307.— polar view; x 1,50 ae bilabiata, pues view; X 1,46 FIGURE 5.— A. p dix de b din view; X 1,316 rutilans, polar view; x 1,3 -o a silvarum polyantha, qu. um x 1,4 0G through colpus; x 8,0 rubricaulis, equatorial view; 7.—G. C. rubra, polar view; x 1,43. 1, polar view; x 1,4 D. asteroloma, polar view; 350.—H. D. — B. C. zebranella, ah view; X 1,687.—C. C. x 1,383.— EF. C. oerstediana, equatorial um C. erythrophaea, — B. "> repens, polar view; x 1,233.—C. C. —E. D. puyana, polar view; x1,733.— ericae, exine fracture > a£ (UT ENS ` Pollen Morphology in Columnea o E S z o5 o HH = LL Volume 75, Number 1 1988 Annals of the 174 Missouri Botanical Garden j ` yr Š ; Ay X » e TI E y Es ë i Volume 75, Number 1 988 Fritze & Williams Pollen Morphology in Columnea v o) r V. Vig. 0° OP y DW Wy ^V e y * A M. P Y! Botanical Garden issouri Annals of the M 176 178 Annals of the Missouri Botanical Garden the reticulation not reduced to punctae at the poles), and with the colpi more elliptic, these of intermediate length. Dalbergaria eburnea (Fig. 7E, F) and D. asteroloma (Fig. 5G) grains are suboblate to spheroidal with the tectum nearly punctate, tectal perforations near the colpi and at the poles reduced in size or absent, and the apertures narrow. Three distinct types of pollen are found within Trichantha (15 species examined). One type (Figs. 9A-G, 10E) is suboblate with a reticulate exine becoming punctate adjacent to the colpi and at the poles and has long apertures widest at the equator and a trian- gular amb. These grains resemble pollen of Dalbergaria and Pentadenia orientandina. Tricantha ambigua, a tetraploid, has pollen of this type but with four apertures (Fig. 10C). A second type of pollen with short colpi (Fig. 10A, B, G, H) is oblate to spheroidal, retic- ulate, and punctate around the colpi with intermediate elliptic apertures; this type has a somewhat triangular amb. The third type (Fig. 11A-D) is suboblate and punctate, and has very short oval apertures (almost like ) its amb is circular to slightly trian- Most species of Pentadenia (9 species ex- amined) possess suboblate to oblate grains that are reticulate, becoming punctate adja- cent to the colpi but remain reticulate at the poles; they have either short or intermediate apertures and somewhat triangular ambs (Figs. 12A-H, 13C, D). Pentadenia ecuadorana (Fig. 12C, D), P. microsepala (Fig. 12E, F), and P. angustata (Fig. 13C) grains have nar- row apertures of intermediate length, but the apertures are widest at the equator. Grains of Pentadenia spathulata (Fig. 12G) and P. zapotalana (Fig. 12H) have short elliptic ap- ertures. Pentadenia orientandina pollen (Fig. 13A, B) is suboblate and reticulate, grading to punctate at the poles and adjacent to the colpi; it has long luminal baculae, long ap- ertures tapered at the poles, and circular ambs. It stands apart from other pentadenias and resembles some Trichantha pollen types (e.g., Fig. 9). Pentadenia strigosa (Fig. 13E, F) pollen shows a combination of features found nowhere else in the Columnea alliance. The pollen is suboblate and reticulate with long luminal baculae, and the reticulum is neither reduced at the poles nor around the colpi. It has long, elliptic, slightly tapered apertures and a circular amb. Bucinellina, with only two species (Fig. FIGURE 6.— A. pun cruenta, polar view; X 1,516 D. kahlbreyeriana, polar view pad 83. — D. D. a view; x 1,500 ,833 x1 —H. D. inaequilatera, nnd region; X 6,000 FIGURE 7.— A. Dalbergaria vittata, polar view; x 1,500.— B. D. vittata, mesocolpial region; x 7,333. 400.— D. D. ericae, mesocolpial region; x 7,000. —E. ericae, iur ud view; erica x 1,500.— Pi mesocolpial region; x 7,500.— to aperture; X 6, 83. E] Fic sanguinea, (W-1628) polar view; ey sanguinea, (G-85) dud view; X 1,5 D. san polar view; x 1,510.—H. D. aureonitens, exine fracture FIGURE 9.— A. Trichantha tenensis, polar view; pureovittata, polar view; x 1,233.— F. T. moorei, equatorial view; x 1,2 through mesocolpial region; x 4,833. FIGURE 10.— A. a P n view; X 1,117. polar /equatorial view; x 1,2 T. x 1,483.—F. C. ku y Se ent. e x 7,333. mesocolpial region; x 6,000. ka hibreyeriana, mesocolpial region; X 6,667. ulchra, mesocolpial ee x7 E 8.— A. Pide florida, n view; X 1,367 x 1,333. T. citrina, d pue view; X 1,383.— —G. T. mbigua, mesocolpial region; —B. D. cruenta, mesocolpial region; X 7,333.— —E. D. perpulchra, —G. D. inaequilatera, equatorial view; —C. D. D. eburnea, equatorial view; G. D. picta, polar view; x 1,367.— 4H. D. picta, adjacent — B. D. florida, mesocolpial region; X6,833.—C. D. —D. D. sanguinea, (W-1628) mesocolpial region; X 7,500.— uinea, (1.170 9) polar view; X 1,443.—G. D. aureonitens, through mesocolpial region; x 5,000. —B. T. brenneri, polar view; x 1,250.—C. T. pur- . mira, polar view; X 1,383 r, polar view; X 1,200.—H. T. calotricha, exine fracture —C. T. SERI . kucyniakii, polar vie —H. T. calotr icha — B. T. dissimilis, Pits view; X 1,100. x6 —E. —G. T. calotricha, Fide view; x 1, 150. Volume 75, Number 1 1988 Fritze & Williams Pollen Morphology in Columnea 179 eae. 180 Annals of the Missouri Botanical Garden Volume 75, Number 1 Fritze & Williams 1988 Pollen Morphology in Columnea Annals of the Missouri Botanical Garden 184 Annals of the Missouri Botanical Garden FIGURE 11.— A. Trichantha pulchra, epi view; X 1, x 5,267.—C. T. parviflora, polar view; 423.— D. 13G, H), has pollen that is readily recognizable by its extreme oblate shape, short oval ap- ertures, and triangular ambs. The two species, however, show a sharp difference in exine patterns: Bucinellina nariniana has reticu- late sculpturing (Fig. 13G) with wide muri, and B. paramicola (Fig. 13H) is nearly tec- tate-imperforate. One fracture was made of each type of grain in the Columnea alliance. Although they do not reveal layers visible with TEM, frac- tures provide useful data. One can see a di- 100.—B. T. pulchra, i Hr ee adjacent to colpus; T. herthae, polar view; x 1,1 versity of exines, each with a different nexine- to-sexine ratio and nexine-to-tectum and col- umellae ratios. The exine of Columnea dress- leri pollen (Fig. 3H) is very uniform with closely packed columellae and a nexine-to- sexine ratio of approximately 1:6. The tec- tum and nexine are of equal thickness. Dal- bergaria ericae pollen (Fig. 5H) by contrast, has a nexine-to-tectum ratio of about 1:3. The nexine-to-sexine ratio approximates 1 : 8. The nexine becomes thick next to the colpus, and the columellae are evenly spaced but FIGURE 12. P. ecuadorana, polar view; view; X 1,533. zapotalana, polar view; A. Pentadenia nervosa, ae view; 300. — x 1,20 uadorana, ME bis x 6,000.—E. P. uir ru polar LOB. microsepala, me olas region; X /,667.—G. P. s 353.—H. P. 33 —B. P. vosa, mesocolpial region; x 7,500.—C. pathulata, polar view; x 1,35 Volume 75, Number 1 Fritze & Williams Pollen Morphology in Columnea Volume 75, Number 1 1988 Fritze & Williams Pollen Morphology in Columnea 187 absent in the lumina. Dalbergaria aureoni- tens (Fig. 8H) has a similar exine, but the columellae are much more irregularly spaced, since they do not occur in the lumina where there is no tectum. Trichantha calotricha pollen (Fig. 9H) has a tectum over three times thicker than the nexine. The columellae are short and vary in girth. Trichantha pulchra pollen (Fig. 11B) is at variance with all others examined in that its nexine is about five times thicker than the tectum and increases in thickness at the col- pus. The columellae are unequally spaced and relatively short. e nexine and tectum of Pentadenia spathulata (Fig. 13D) are of equal thickness, and the columellae are well spaced but absent where there is no tectum. Pentadenia stri- gosa (Fig. 13F) shows a thick tectum relative to the nexine, and baculae fill the lumina. DISCUSSION Pollen characters are constant, at least within certain genera and subgeneric groups, and appear to be of taxonomic use. The char- acters found to be most useful are overall shape, sculpturing, and aperture length. Size did not appear significant, although pollen of Tricantha tends to be slightly larger com- pared with other genera. Some pollen forms are associated with co- rolla form and sometimes with general mor- phology of the parent plant. The more vari- ation there is in corolla form in a genus, the more variation there is in pollen form. The cohesiveness of macromorphological charac- ters in Columnea is supported by remarkable pollen uniformity; the uniformly punctate ex- ine is characteristic of the genus. All species with this pollen type possess the characteristic galea of the flower. However, species outside the genus Columnea that exhibit a similar "columneoid" corolla do not have the pollen characteristics of Columnea (Dalbergaria picta, D. ericae, D. kalbreyeriana). Pollen of Columnea repens and C. ruti- lans, the only two columneas examined that occur outside the mainland of Central and outh America, is different from that of main- land species (Figs. 5B, C). These two species are endemic to Jamaica, and different selec- tion pressures there could be a factor in their divergence from the general pattern. Colum- nea kucyniakii does not fit very well with the remainder of the genus Columnea. Its pollen is more similar to the pollen of Dal- bergaria or Trichantha, but the species does not easily fit into the concept of either of the genera in terms of floral features and general habit (Wiehler, pers. comm.). e two species of Bucinellina share the same grain shape and aperture size, but they have different exine patterns and a slight dif- ference in grain size. They are from the same general locality, and their flowers and habit are fairly similar; the reason for the pollen difference is not evident. Trichantha purpureovittata, T. tenensis, and T. brenneri are closely related and have similar corollas as well as similar general plant habit (Wiehler, 1975). The similarity of their pollen correlates well with these characters. Trichantha pulchra, T. herthae, and T. par- viflora make up another group of morpho- logically similar species (Wiehler, 1977, pers. comm.). They have small yellow corollas with tough, incurved lobes and similar pollen. Trichantha calotricha has a corolla similar to those of the rest of this group, but its pollen is very different. Trichantha citrina and T. mira are also closely related to each other (Wiehler, 1978), and their pollen reflects this relationship. Dalbergaria species have similar pollen and a moderate degree of uniformity in their FIGURE 13.— A. Fentadenia Wide polar view; X 1,400. 1,4 athulata, exine Z— D. P.s angustata, polar vie —B. P. pesi mesocolpial region; x. 233. == E. P strigosa polar view; x 1, 467.—F. T strigosa, exine fracture at polar end of colpus; X5, 067. — G. Bucinellina 33. nariniana, polar view; x 1,467.— . B. paramicola, polar view; x 1, 188 Annals of the Missouri Botanical Garden TABLE 2. MEA of ponen characters in the Columnea alliance. p — punctate, r — reticulate, s — short, | = long, i = intermediate, c = circular, a = angular, e = elliptic, o = oval, n = narrow, t = tapered at poles; P/E = ratio of a axis length to equatorial axis length. Sculpture Apocol- Mesocol- Colpus Colpus Species P/E pium pium Border Length Shape Amb Columnea arguta 1.20 p p p l e c C. bilabiata p p p l et c C. billbergiana p p p l e [o C. cobana 0.92 p p p l et c C. dodsonii p p p l e c C. dressle 1.12 p p p l e c C. Pae ee — p p p l e c C. flac = P p p l e c c “allic on — p p p l e c C. gloriosa - p p p l e c C. guatemalensis 1.10 p p p l e c C. hirsutissima 1.04 p p p l e c C. kienastiana 1.02 p p p l et c C. kucyniakti p r p l t a C. lineari 1.09 p p p l e c C. maculata 1.16 p p P l e c C. nicaraguensis - p p p l e c C. oerstediana p p p l e c a purpusit 0.99 P p p l e c `. querceti 0.99 P p p ] e c e repens 0.72 p r p l t a C. rubra p p p l e c C. rubricaulis p p p l e c C. rutilans 0.89 p r p l t a C. schiedeana 0.94. p p p l e c C. verecunda 0.90 p p p l e c C. zebranella 1.10 p p p l et c Dalbergaria asteroloma p r p l n c D. aureonitens 0.89 r r p 1 e c D. cruenta 0.7 p r p l t a D. eburnea 0.89 p p p l nt a D. ericae 0.76 p r p l t a D. florida 0.84 r r p l e a D. Wb ames 0.78 p r p l t a D. kahlbreyeriana — p r p l t a D. perpulchra 0.78 p r p l t a D. picta 0.85 p r p l t a D. polyantha 0.76 p r p l t c D. puyana 0.86 p r p l t a D. sanguinea! -— r r p l e c D. sanguinea? r r p i e c D. sanguinea? 0.91 r r p i et c D. silvar 0.70 p r p l t c D. vittata 0.80 p r p l t a Pentadenia angustata r Ë p l n a P. byrsina r r p i t a P. ecuadorana 0.75 r r P 1 n a P. microsepala - r r p 1 n a P. nervosa r r p 1 t a P. orientandina — p r p l t a Volume 75, Number 1 1988 Fritze & Williams 189 Pollen Morphology in Columnea TABLE 2. Continued. Sculpture Apocol- Mesocol- Colpus Colpus Species P/E piu pium Border Length Shape Amb P. spathulata — r r p s e a P. strigosa 0.82 r r r l t c P. zapotalana — p r p s e a Trichantha ambigua = p r p i n a T. brenneri — p r p l t a T. calotricha 0.79 r r p i e a T. citr 0.80 P r p l t a T. pirum 0.66 r r p i e a T. filifera — r r p i e a T. herthae 0.86 p p p s o c T. minor 0.86 p r p l t a T. mi 0.75 p r p l t c T. moorei 0.85 p r p l t a T. parviflora — p p p s o € T. pulchra == p p p s o c 1: uo 0.94 p r p l t a T. tenensis 0.79 p r p l t a Bucinellina paramicola — p p p s o a B. nariniana — r r p s o a ! Accession number W-1 709. 2 Accession number G-85. > Accession number W-1628. corollas. Most pollen grains show slight vari- ations on a theme, e.g., long apertures and a tectum punctate around the colpi and at the poles. Pollen of Dalbergaria sanguinea, a tetraploid, is no larger than that of the rest of the genus but is distinct by showing no reduction of the reticulum at the poles and by having apertures of intermediate length. Dalbergaria aureonitens is closely related to D. sanguinea (Wiehler, 1973), which i reflected by pollen similarities between them. Dalbergaria florida is somewhat atypical in the genus although its pollen exhibits feat >s that closely resemble Dalbergaria sangu. a and D. aureonitens. In Pentadenia there is a relatively high degree of uniformity in corolla forms (Wieh- ler, pers. comm.) and in the pollen. Most corollas are tubular, small, and generally non- descript. Pentadenia strigosa, however, 1 strikingly different from the rest of the species in the alliance in its much larger corolla and peculiar pollen characteristics, probable ad- aptations to pollination by bats. un un The Gesneriaceae are placed in the rela- tively advanced order Scrophulariales, and the reticulate, tricolpate pollen of Gesneri- aceae is advanced among dicotyledons (Walk- er & Doyle, 1975). This family has a spe- cialized shape in that it departs from the spherical. The exine pattern is specialized as it differs from the primitive tectate grain. A reduction in aperture size can be seen, par- ticularly in the Episcieae. Some similarities to pollen of other families in the order Scrophulariales can be seen in pollen of the Gesneriaceae. In the Scrophu- lariaceae and Myoporaceae one finds tricol- pate pollen (Niezgoda & Tomb, 1975) and finely reticulate exines reminiscent of Colum- nea. The colpi are long, but the apertures are diorate, a rare type not found in the Gesneriaceae. Erdtman (1966) found tri-colp(or)ate, oblate-spheroidal, subprolate, or prolate-spheroidal pollen in the Scrophu- lariaceae. These attributes are within the range of pollen features of the Gesneriaceae. Boj (1961) made an extensive survey of 190 Annals of the Missouri Botanical Garden the Acanthaceae and found few features that can be compared to the Gesneriaceae other than the reticulate exine. Erdtman (1966) examined 55 species in 35 genera of Acan- thaceae and found it to have uniform pollen. One tricolporoid type is subprolate to prolate and has fine reticulation in which the brochi decrease in size toward the colpi. The grains are also within the size range found in the Gesneriaceae. Buurman (1977) found a number of fea- tures in Bignoniaceae similar to those of ges- neriaceous pollen. The number of apertures varies, but the tricolpate type exhibits a pro- late shape and a uniformly punctate tectum, the lumina of which decrease (almost imper- ceptibly) in size adjacent to the colpi. These features are predominant in Gesneriaceae tribe Gloxineae (see Williams, 1978). Buurman (1977) outlined within the tricolpate group some evolutionary trends that may have some application to the Gesneriaceae (see below). Erdtman (1966) examined pollen of 25 species in 20 bignoniaceous genera and found a size range that exceeds that of the Gesneriaceae. This study has answered several questions regarding pollen use in taxonomic consider- ations of the Columnea alliance. Pollen char- acters separate Columnea and Bucinellina from the rest of the alliance, but there is much variation within the other genera. Pol. len appears to be useful, however, in several ways. Subgeneric groups that exhibit similar pollen could be classified as sections, since they are found to be closely related by other criteria. Differences between species in some pollen features are more apparent than dif- ferences between genera, so pollen characters seem to be good species indicators. Pollen does appear to be useful for separating the Columnea alliance from other alliances and other alliances from each other. Evolutionary trends based on the pollen information gathered here can be related to Buurman's (1977) scheme. Tectal features in the Columnea alliance exhibit a trend from perforate to uniformly reticulate to punctate near colpi and at poles. Caution must be ex- ercised when this scheme is applied to the Columnea alliance, however, since next to Bucinellina, Columnea is considered most advanced, but it has a uniformly punctate tectum. The shape and size of the grains in the Columnea alliance exhibit a trend from spherical and small to oblate or prolate and large. We postulate that the ancestral pollen was tricolpate, spherical, and uniformly re- ticulate with small lumina, much like pollen that occurs in the tribe Gloxineae (Williams, 1978) LITERATURE CITED Bos, R. 1961. Pollen idee, der Le in the Acan- thaceae. Grana l. 3: Burtt, B. L. 1 Studies in s DM of the Old World. XXIV: tentative keys to the tribes and genera. Notes Roy. Bot. Gard. Edinburgh 24: 205- 220. BUURMAN, J. 1977. Contributions to the pollen mor- hology of the Bignoniaceae with special reference to the tricolpate type. Pollen & Spores 19: 447- 519 ERDTMAN, G. 1966. Pollen Morphology and Plant Tax- onomy. Hafner Publishing Co., New Yor FRITZE, K. 1979. Pollen Rd nd the Taxon- omy of the oa Alliance Ca : Ges nerioideae). M.S. Thesis. Florida State Deino, buco Florida. MELHEM, T. C. Mauro. 1973. Pollen morpho- logical Maus in Gesneriaceae. Hoehnea 3: 13-27. 67. Moore, H. E. & R. broadening basis of classification in the Gesneriaceae. Baileya 15: 97-108. Morey, B. D. 1971. A hybrid swarm between two hummingbird-pollinated species of Columnea (Ges- neriaceae) in Jamaica. Bot. J. Linn. Soc. 6 96. 1972a. The distribution and variation of some gesneriada on Caribbean islands. Pp. 239-257 in D. H. Valentine (editor), en Phytogeography, and Evolution. Academic Press, New York. 1972b. Some us type diversity in Colum- nea L. sensu lato (Gesneriaceae). Bot. J. Linn. Soc. 65: 25-36. 973. Ecological DE of sig Sa to Columnea taxonomy. Pp. 265-281 in V. H. Hey- wood (editor), ME x | Academic Press, Ps Noe on some critical characters in Columnea cios Ann. Missouri Bot. Gar 514-5 = 197 A key, typification and synonymy of the sections in the genus Columnea L aceae). € Morton, C. V. A revision of Trichantha (Ges- iig ro Contr. U.S. Natl. Herb. 38: 1-27. 971. A reduction of Trichantha to Colum- ea [o Phytologia 22: 223-224. NiEZGODA, C. T. . S. Toms. 1975. Systematic Volume 75, Number 1 1988 Fritze & Williams 191 Pollen Morphology in Columnea palynology of the tribe Leucophylleae (Scrophulari- aceae) and selected Myoporaceae. Pollen & Spores 17: 495-516. SHERK, L. C. & R. E. LEE. 1967. Interspecific hybrid- ization in the genus Columnea (Gesneriaceae). Bai- leya 15: SKOG, L F. 1976. n study of the tribe Gesnerieae, with a revision of Gesneria (Gesneriaceae: Gesnerioideae). Smithsonian Contr. Bot. 29: 1-182. WALKER, J. W. 1974a. Evolution of exine structure in the pollen of primitive angiosperms. Amer. J. Bot. 1: 891-902. Aperture AeA in the polen of of . J. primitive angiosperms. Ame Bot. 61: 1112 1136. & J. A. DoYLE. 1975. The bases of angiosperm phylogeny: palynology. Ann. Missouri Bot. Gard. 62: 664-72 . S. Kemp. 1972. Preliminary studies of exine stratification in the pollen of pa ive angio- sperms. Brittonia 24: 129-130. iss WIEHLER, H. 971 e cha some abes Gesneriaceae in cultivation. Baileya 18: 133-138. 73. One hundred transfers from Alloplec- tus and Columnea (Gesneriaceae). Phytologia 27: 309-329. 75. Three new species of Trichantha from 19 Peuador and Peru (Gesneriaeae). Selbyana 1: 36- . New genera and species of iras aceae from the Neotropics. Selbyana 2: 67- 78. Miscellaneous transfers and new eA of neta yan sage Selbyana 5: 61-93. 1 A synopsis of the neotropical Gesne- riaceae. Selbyana s 1-219. WiLLIAMS, N. H. Pollen structure and the sys- tematics of the neotropical Gesneriaceae. Selbyana 2: 310-322 Wilson, C. L. Floral anatomy in Gesneriaceae. II. Gesnerioideae. Bot. Gaz. (Crawfordsville) 135: 256- ne P. 1964. Pollen morphology in Gesneriaceae. Oth International Botanical Congress, Edinburgh, i 97. [Abstract.] OBSERVATIONS ON THE CHROMOSOME CYTOLOGY OF VELLOZIACEAE Peter Goldblatt! and Muriel E. Poston? ABSTRACT mosome numbers of 15 species in four of the six genera of Velloziaceae were counted from root tip C ae The family was dde known cytologi ically from have been reported elsewhere. The American . The African Xerophyta has count pea an earlier record of n = 24—26 Barbacenia appears to be nri paleohexaploids on the derived base of n m a single count, although n= = 24 in two m examined, and the oO a has n = we A base number of x = 9 i aleotetraploid genus. Nera a hypodiploid, and Ru and Talbotiopsis ı are roposed for Velloziac Velloziaceae are a small family of petaloid monocots comprising six genera, Barbacenia (102), Barbaceniopsis (3), Nanuza (1), and Vellozia (122) in South America, and Xe- rophyta (50) and Talbotiopsis (1) in sub- Saharan Africa, south Arabia, and Madagas- car. Until recently, Velloziaceae were barely known cytologically, the only chromosome record being n = 24-26 for Talbotiopsis elegans (Stenar, 1925, as Vellozia). Atten- tion was first drawn to the lack of cytological information for Velloziaceae by Ayensu (1973) in his extensive study of the family. In a review of cytological evolution in the angio- sperms, Raven (1975) again focused atten- tion on the scant cytological data for the family, stimulating this investigation. Prelim- inary findings, unfortunately inexact, were included in Raven's review. These and ad- ditional counts are presented here with cor- rections where necessary. MATERIALS AND METHODS Plants for study were obtained from a live collection maintained at the Smithsonian In- stitution by Drs. L. B. Smith and E. S. Ayensu for their anatomical and taxonomic studies of the family. The material of Xerophyta reti- nervis was gathered in the wild by Goldblatt specifically for cytological study. Species ex- amined are listed in Table 1 with collection data and chromosome numbers. All counts were made from root tip mitoses. Roots were harvested from actively growing plants and pretreated in 0.003 M hydroxy- quinoline for six hours at refrigerator tem- peratures. They were then fixed in 3:1 eth- anol-acetic acid, hydrolyzed in 10% HCI for six minutes, and then squashed in lacto-pro- pionic orcein. OBSERVATIONS Barbacenia A diploid number of 2n = 34 was found in each of four species of Barbacenia ex- amined (Table 1). A preliminary count for this genus, n — 16 (Goldblatt in Raven, 1975), is incorrect. Chromosomes are all of similar size, 1-2.5 um long, and are metacentric to submetacentric. ' B. ` "aad of African Botany, 63116- 0299, U.S Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri ` Department al om Howard University, Washington, D.C. 20059, U.S.A. ANN. Missouni Bor. Garb. 75: 192-195. 1988. Volume 75, Number 1 1988 Goldblatt & Poston 193 Chromosome Cytology of Velloziaceae Chromosome numbers in Velloziaceae. a reference is given All counts are original except one for Talbotiopsis, for which Haploid Species o. n Collection Data Barbacenia B. aff. albiflora L. B. Smith 17 Brazil. Minas Gerais: Biri-Biri, Mun. Diamantina, Hatsch- bach 30183 (US B. coronata P. Ravenna 17 Brazil. Minas Gerais: Pico Itambe, Hatschbach 30095 (US). B. globata Goethart & Henrard 17 Brazil. Minas Gerais: Cons. Mata, Hatschbach 30212 (US). B. paranaensis L. B. Smith 17 Brazil. Parana: Fda. Morungaia, rio do Furil Mun. Senges, Hatschbach 29212 (US). Vellozia sect. Radia V. hirsuta Goethart & Henrard 7 Brazil. bey Gerais: Diamantina, Serra do Espinhaco, & Ayensu 15999 (US). V. riedeliana Goethart & Henrard 7 Brazil. Minas Gerais: 27 km W of Serro, Smith & Ayensu 15983 (NY). V. tubiflora (A. Rich.) Kunth 7 es bs Gerais: Rio Itacambirucu, Grao-Mogol, Hatschbach 41274 (US). sect. Vellozia V. alata L. B. Smith 7 Brazil. Minas Gerais: 3 km N of Chapeado Sol. Serra do Ciop, Smith & Ayensu 15951 (US). V. bahiana L. B. Smith & Ayensu 8 Brazil. Bahia: exact locality unknown, Maia s.n. (US). V. caruncularis Mart. ex Seubert 7 Brazil. Minas Gerais: 9 km W of Serro Cerrado, Smith & Ayensu 15977 (US). V. compacta Mart. ex Schultes f. 8 "i LO Gerais: 27 km W of Serro, Smith & Ayensu 15986 (US). V. pterocarpa L. B. Smith & Ayensu 8 vd Mims Gerais: Diamantina, Hatschbach & Ahuma- da 31705 (US). Xerophyta X. humilis (Baker) Dur. & Schinz 24 South Africa, without precise locality, Gaff s.n., no vouch- X. retinervis Baker 24 ue Africa: Pretoria, hills at Bot. Res. Inst., Goldblatt ., NO VO Talbotiopsis T. elegans (Hook. f.) L. B. Smith 24 South Africa: exact locality unknown, Meyer s.n. (NA). 24-26 Stenar, 1925. Vellozia The preliminary count (Goldblatt in Raven, fuse, lightly staining areas stand out. The significance of these shadowy chromatic bod- ies is unclear, but they do not seem to be 1975) for the genus, n = 9, was not confirmed by further examination. Species of Vellozia sect. Radia have 2n = 14, while 2n = 16 and 14 were found in species of sect. Vel- lozia. A third section, Xerophytoides, is so far uncounted. The chromosomes are com- parable in size and appearance to those of Barbacenia. In some preparations two dif- chromosomes. Xerophyta A diploid number of 2n — 48 was found in the two species of this Afro-Madagascan genus counted, Xerophyta humilis and X. retinervis. Chromosomes are generally simi- 194 Annals of the Missouri Botanical Garden lar in size and appearance to those of Bar- bacenia, and 1-2 um long. Talbotiopsis One collection of this monotypic genus, renamed Talbotiopsis (= Talbotia) by Smith (1985), was examined. Our count, 2n — 48, substantiates Stenar's (1925) report of n = 24-26 for T. elegans (published under the synonym Vellozia elegans). Chromosomes of Talbotiopsis are similar to those of Xero- phyta. DISCUSSION The chromosome data presented can only be regarded as preliminary for Velloziaceae, given that we now have counts for just 15 species in four genera out of a total of 250 species in six genera. Nevertheless, the avail- able counts are fairly consistent within genera and so suggest that they comprise a repre- sentative sample of the chromosome variation in Velloziaceae. e counts suggest the following hypoth- esis of cytological evolution. First, base num- ber may be x — 9 for Velloziaceae. It follows that the number n = 17 in Barbacenia would represent aneuploid reduction from a paleo- oo base of n = 18. The numbers n = 8 and 7 in Vellozia are then interpreted as aneuploid on the family base of n = 9. X rophyta and Talbotiopis appear to be pa- leohexaploids derived from the secondary base of x = 8. The shared number in Xerophyta and Talbotiopsis supports the current belief (Ayensu, 19773) that these two African genera are more closely allied to one another than to the other genera of the family, all South American. Other scenarios can be construct- ed, but the one outlined seems to us the most parsimonious, and thus recommended at least in the light of current knowledge of Vellozia- ceae and the patterns of numerical chromo- some change that occur in plants (Raven, 1975; Goldblatt, 1980). A base number of x — 8 for Velloziaceae appears at first to be more parsimonious interpretation, with polyploid doubling (to n = 16) and subsequent aneuploid increase to achieve n = 17 in Bar- bacenia. However, given that ascending aneuploidy is at least four times less common in the flowering plants than descending aneu- ploidy, we think the latter possibility is less likely, although not implausible. e chromosome numbers in Velloziaceae tell us little about possible relationships of the family. The most critical current phylogenetic opinion (Dahlgren et al., 1985) treats Vel- loziaceae as the sole family of Velloziales, one of six single family orders comprising Bro- meliiflorae. The reasons for removing Vello- ziaceae from Liliiflorae and Liliales, to which the family is traditionally assigned, include the Strelitzia-like epicuticular waxes, copious starchy endosperm, and stomata with subsid- iary cells (Dahlgren et al., 1985). All of these apparently fundamental features correspond with other families of Bromeliiflorae and con- flict with Liliiflorae. Base number in the Bromeliaceae, the fam- ily and order possibly closest to Velloziaceae, is x = 25 (Raven, 1975), which contrasts sharply with the suggested x = 9 in Vellozia- ceae. Cytology thus appears to contribute lit- tle to our understanding of relationships of the families of Bromeliiflorae. However, there seems reason to suppose that if Bromeliiflorae sensu Dahlgren et al. do constitute a natural alliance, then x = 9 or 8 are important base numbers. Other orders of Bromeliiflorae in- clude Philydrales (Philydraceae, x — 9 or 8), Haemodorales (Haemodoraceae, possibly x — 8), and Pontederiales (Pontederiaceae, x — 8) (base numbers from Goldblatt, 1980). The last order included in Promelüflorae by Dahl- gren et al., Typhales, with x 5, is very different and may even be placa! here. LITERATURE CITED AYENSU, E. S. 197 Mes e and evolution of the Velloziaceae. Pp. 1 5-119 in B. J. Meggers, E. S. Ayensu & W. D. Dh (editors), Tropical ed Ecosystems in Africa and South America: a comparative review. Smithsonian Institution Press, Washington DAHLGREN, R. M. T . T. CLIFFORD & P. P. Yero. e ie The Families of the Monocotyledons. Sprin- erlag, Berlin. GoL A ATT, P. 1980. Polyploidy in angiosperms: mono- Volume 75, Number 1 Goldblatt & Poston 195 1988 Chromosome Cytology of Velloziaceae cotyledons. Pp. 219-239 in W. H. Lewis (editor), ^ ———— Mer A new genus of Velloziaceae. Phyto- Polyploidy: oe "d Plenum, New York. dogia 132, Raven, P. H. 1975. The bases of angiosperm phylog- STENAR, "à A S. 1925. Embryologische Studien .... eny: ii Ann. Missouri Bot. Gard. 62: 724- idonee Abhandlung, Uppsala. SMITH, L. B. 1962. A synopsis of the American Vel- loziaceae. Contr. U.S. Natl. Herb. 35: 251-292. CHROMOSOME COUNTS AND KARYOMORPHOLOGY OF SOME WEST TROPICAL AFRICAN SCILLEAE (LILIACEAE) S. O. Oyewole! ABSTRACT Karyomorphological analyses of Urginea ensifolia, U. pauciflora, Drimiopsis barteri, Dipcadi tacazzeanum, and D. longifolium are presented. is 2n = 24, m and D. longifolium are All the na E. tigated were obtained from the he five species. Both Urginea species are 2n — 2n = 12 wild in Nigeria. The karyotypes 20, Dadian barteri 2n = 24, respectively. Other members of and Urginea and all the known species of Albuca, which have previously been investigated and reported, are summarized. The tribe Scilleae Bak. (Liliaceae) consists of six genera in West Tropical Africa (Albuca L., Dipeadi Medic., Drimia Jacq., Drimiop- sis Lindl., Scilla L., and Urginea Stein). The latest treatment of the Liliaceae in West Trop- ical Africa (Hepper, 1968) shows that these genera are represented by three (five, Gledhill & Oyewole, 1972), two, one, one, one, and four (six, Oyewole, 1975a) species, respec- tively. Most of the representatives of each genus show striking morphological similarities as well as population variations within each species, which make their taxonomic treat- ment difficult. There is evidence that many natural populations of the representatives of the tribe are not yet in herbarium collections, so it is likely that there are more taxa in the tribe than are now known. In this paper, new reports on chromosome number and morphology are given for five representatives of the Scilleae. MATERIALS AND METHODS Populations of each species were sampled during several field trips to different parts of Nigeria (Table 1). The species were identified using specimens at the Herbarium of the Fed- eral Institute of Forest Research, Ibadan (FHI). Plants of each species were cultivated at the University of Ilorin, Nigeria. Voucher specimens are deposited at FHI, Ahmadu Bel- lo University Herbarium (ABUH), and the Herbarium of the University of Ilorin (IUH). Each plant was investigated separately, but plants of the same species were treated to- gether. Cytological studies were carried out on squash preparations of young root tips following conventional methods as earlier re- ported (Oyewole, 1972). Chromosome index, r (ratio of long chromosome arm to the short arm), was determined according to Levan et al. (1964) as modified by Oyewole (1972), and the values were employed in analyzing the karyomorphology of each taxon. RESULTS AND DISCUSSION Table 2 summarizes earlier work on the tribe Scilleae while Table 3 summarizes karyotype data on the new reports. URGINEA The basic chromosome numbers of this genus are x — 5 and x — 7 (Darlington & Wylie, 1955; De Wet, 1957; Jones & Smith, 1967). The four species listed in the Flora of West Tropical Africa (Hepper, 1968) are ' Department of Biological Sciences, University of Ilorin, P.M.B. 1515, Ilorin, Nigeria. ANN. Missouni Bor. Garb. 75: 196-202. 1988. Volume 75, Number 1 Oyewole 197 1988 Chromosome Counts of West African Scilleae TABLE l. Sources of materials investigated (Scilleae) . Taxa Collection Site Herbarium Voucher Habitat Fashola Rocks 12 km north of along Oyo-Iseyin Road also 6 km to Oshogbo along S00/0660, in ABUH, IUH, FHI S00/0752 in ABUH, IUH, FHI /rginea marshy foot of rocky hills in de- ensifolia ciduous woodland n marshy roadside, deciduous Oshogbo-Gbongan Road woodland a Hills, Minna Urginea 800/0764 in n gravelly soil on hillcrest on the pauciflora ABUH, IUH, FHI outskirt of Minna, in low rassland savan Dipcadi Igbetti about 150 km north S00/1001 in IUH, in dark, humus soil in shaded de- tacazzeanum Oyo I ciduous woodlan 12 km north of Ilorin along S00/1091, S00/ in dark, humus soil in open shal- Lagos-Kaduna Road 1092 in IUH, low soil on rock outcrops in Shao, 22 km northwest of Ilo- FHI grassland savanna rin Dipcadi Babanloma, about 72 km north 800/1008, in IUH, in dark, humus soil in rock insel- longifolium of Ilorin along Lagos- Ka- FHI bergs; deciduous woodland duna Road about 20 km to Kabba along 800/1104-1110, in in dark gravelly soil at foot of Kabba-Okene Road IUH rock hills; deciduous woodland Nasarawa Village along Mok- 800/2174, in IUH in dark brown, clay-loamy soil; wa-New Bussa Road open, seasonally marshy graz- ing land Drimiopsis Tegina, a small junction village JMC/79, in IUH in open, seasonally marshy rock barteri along Lagos- Kaduna Road, soil in grassland savanna about 144 km to Kaduna Kaduna Airport environment 800/1221, in IUH gravelly brown soil on rock out- rops in grassland savanna FHI— Herbarium of the Federal Institute of Forest Research, Iband ABUH — Ahmadu Bello University Herbarium (Department of Biological Sciences, Ahmadu Bello University, Zaria, Nigeria). IUH — University of Ilorin Herbarium (in Department of Biological Sciences). TABLE 2. Previous chromosome counts in Scilleae. Somatic Taxa Number Haploid Karyotype! 1. Albuca L. x = 9 (Oyewole, 1972) A. abyssinica Murray 18 3L 68 A. fibrounica D. Gledhill & S. O. Oyewole 36 6L 128 abromarginata De Wild. 36 6L 125 A. de A. Chev. 36 6L 12S A. nigritana (Baker) Troupin cytotype I 54 9L 18 S cytotype H 54 12 L 155 2. Urginea Stein x = 5 (Oyewole, 1975b, 1987a, c) U. altissima Baker sensu stricto (2n = 20 + 2ff) 22 4L TS U. gigantea Vedi ) Oyewole 22 4L 7S U. viridula Bak 20 4L 6S U. indica (Roxb.) Kunth 20 Variable ' L = long chromosome (> 4.5 um in length); S = short chromosome (< 4.5 um in length). 198 Annals of the Missouri Botanical Garden TABLE 3. Summary of karyotype data (Scilleae) (chromosome length in um). Homologues Taxa NE 2 3 4 5 Urginea ensifolia Chromosome length 7.0 6.88 6.8 6.0 5.63 r-value 6.0 16.0 23.0 5.6 36.0 Centromeric location subterminal terminal terminal subterminal terminal Urginea pauciflora Chromosome length 9.13 8.13 5.31 3.94 3.75 r-value 8.6 15.25 4.5 2.94 6.5 Centromeric location terminal terminal subterminal submedian subterminal Dipcadi tacazzeanum Chromosome length 7.3 6.3 5.1 3.5 2.6 r-value 9.2 6.0 3.0 Centromeric location terminal terminal terminal subterminal submedian Dipcadi longifolium Chromosome length 8.88 8.13 7.13 6.38 6.0 r-value 10.85 20.4 M» 11.0 Centromeric location terminal terminal terminal terminal terminal Drimiopsis barteri Chromosome length 9.13 9.13 7.88 6.75 6.25 r-value ; 7.11 3.5 2.38 Centromeric location subterminal terminal subterminal submedian submedian U. altissima Baker, U. indica (R. & B.) Kunth, U. ensifolia (Thonn.) Hepper, and U. pauciflora Baker. More recent work, how- ever, has shown that U. altissima is a com- plex of three distinct species (Oyewole, 1975a), all of which have been investigated karyotypically (see Table 2). Urginea indica has been treated separately on account of its variable nature (Oyewole, 1987b, c) and is included in Urginea ensifolia has a somatic comple- ment of 20 chromosomes. The karyotype is represented by twelve long and eight short chromosomes (Figs. 1A, 2A). Chromosome lengths vary between 1.0 um and 7.0 um. The six long pairs have terminal to subter- minal centromeres. The first two short pairs have submedian to median centromeres, while the last two pairs are dotlike and without observable second arms: they are telocentric. The third long pair has an inconspicuous sec- ond arm and a secondary constriction on the long arm. Urginea pauciflora also has a somatic complement of 20 chromosomes. The karyo- type is represented by three long and seven short pairs. The chromosome lengths vary from 2.5 um to 9.13 um. All the long chro- mosomes have subterminal to terminal cen- tromeres. Three short pairs have subterminal to terminal centromeres; three others have submedian to median centromeres, while the seventh pair has a very inconspicuous second arm (Figs. 1B, 2B). All Urginea species so far investigated have 2n — 20 except U. volubilis, a Madagascan species (2n = 14, Jones & Smith, 1967). ith a basic number of x = 5 or x = 7, the West African species of Urginea are poly- ploids. However, from karyotype studies, these species have somatic complements that are resolvable into homologous pairs. Urginea al- tissima sensu stricto, with 2n = 20 + 2ff, has been shown to have normal meiosis with ten bivalents (Oyewole, 1987a). The basic number o — 5 therefore applies to the West African Urginea species which are thus tetraploids. Volume 75, Number 1 1988 Oyewole 199 Chromosome Counts of West African Scilleae TABLE 3. Continued. Homologues 6 7 8 9 10 11 12 5.0 3.0 2.0 1.5 1.0 i 2.0 1.0 0 0 terminal submedian metacentric telocentric telocentric 3.50 3.13 3.0 3.0 2.5 1.83 24.0 ] š median terminal telocentric submedian terminal 2.2 subterminal 5.75 5.0 4.88 3.0 2.15 2.75 2.25 10.5 15.0 9.52 59.0 2.0 : 0 terminal terminal terminal terminal submedian terminal telocentric 6.25 5.38 4.75 4.31 4.0 3.75 3.75 2.90 1.86 2.17 1.16 4.0 1.50 1.14 submedian submedian submedian median subterminal median median DIPCADI Two basic numbers, x = 4 and x = 9, are already reported for this genus (Darlington & Wylie, 1955); Hepper (1968) recognized two species, D. longifolium Lindl. and cazzeanum (Hochst. ex A. Chev.) Baker, into which he merged Morton's (1961) D. fila- mentosa Medic. as a morphological variant. Several natural populations of individuals identifiable as D. filamentosa have recently been encountered during field trips in Nigeria, and the cytogenetic relationship of this group with the other species of the genus is still being investigated at Ilorin, Nigeria. Dipcadi tacazzeanum (excluding all ma- terials identifiable as D. filamentosa) has a somatic chromosome complement of 2n = 12. Chromosome lengths range between 2.2 um and 7.3 um, and the karyotype consists of three long and three short pairs. The fifth pair has a submedian centromere, while all the others have terminal to subterminal centro- meres (Figs. 1C, 2C). The third pair has a secondary constriction in the long arm. Dipcadi longifolium has a somatic com- plement of 2n = he complement consists of 16 long and eight short chromosomes, with chromosome lengths ranging from 2.2 um to m. One pair of short chromosomes is telocentric, another has submedian centro- mere, while the remaining two short and all the eight long pairs have terminal to subter- minal centromeres (Figs. 1D, 2D). One of the short pairs with the centromere in the ter- minal region varies morphologically in differ- ent individuals—one or both members have an extended centromeric region. Four pairs (1st, 3rd, 4th, and 6th) have a secondary constriction each in the long arm. With a somatic pain of 24 chromosomes, this w-— is a polyploi es & Smith um reported a somatic T hash number of 2n — 12 for a diploid species suspected to be D. gracillium. The records of 2n — 8, 18, and 34 for three different species, by which the basic numbers of x = 4, 9 were determined, were from southern African materials (see Darlington & 200 Annals of the Missouri Botanical Garden E l. Somatic metaphase complements. tacazzeanum. — D. Dipcadi longifolium. — E. Drimiopsis barteri. Wylie, 1955). The two somatic numbers re- ported here show two ploidy levels and favor a new basic number of x — 6, which is sup- ported by chromosome morphology. Thus, one of the two species is a diploid, 2n — 12 (D. tacazzeanum) and the other a tetraploid, 2n = 24 (D. longifolium). DRIMIOPSIS This genus is represented by D. barteri Baker, as Hepper (1968) recorded. This species has a somatic number of 2n — 24. The chromosomes fall into twelve morpho- logical pairs, with members of the pairs gen- erally unequal. Chromosome lengths vary be- tween 3.0 um and 10.0 um. The complement does not show bimodal categorization into long and short chromosomes. However, seven pairs are longer than 5.0 um while the other five are shorter than 5.0 um. Four pairs (1st, 2nd, 3rd, and 10th) have terminal to subterminal —A. Urginea ensifolia. — B. Urginea pauciflora.— C. Dipcadi centromeres, while the others have subme- dian to median centromeres (Figs. 1E, 2E). Plants of this species are known to be sexually sterile; meiotic behavior and cause of sexual sterility have been reported (Oyewole, 1984a, b). Darlington & Wylie (1955) reported a ba- sic chromosome number o = r the genus from South African materials. A so- matic chromosome count of 2n — 24 indicates triploidy. However, Oyewole (1984a, showed that a basic number of x — 6 rather than x — 8 is more consistent with the somatic complement of the West African species of Drimiopsis. Two wild and morphologically distinct populations recently sampled in Ni- geria have a somatic complement of 2n — 24 each, as in D. barteri, and are sexually re- productive. Their cytogenetic relationship with D. barteri and with each other, as well as their taxonomic positions, are being investigated at Ilorin, Nigeria. Oyewole Chromosome Counts of West African Scilleae Volume 75, Number 1 A Minus... B li lu FHETTITTIITITTRL C li ji bbe: 3344 10 pm 0 ilh ilan... E ) MM ipaa 2o ea ABORARX2IA RE 2. Drawings of the somatic karyotypes. a am ensifolia. — B. Urginea pauciflora. — C. Dipcadi be — D. Dipcadi longifolium. — E. Daop barteri Chromosome length bimodality in certain members of the tribe Scilleae has been re- ported (Jones & Smith, 1967; Oyewole, 1972, 1975b). This has held true in the present work for the two species of Urginea and the diploid D. tacazzeanum but not with D. longifolium and D. barteri. Also, while the West African members of Albuca, Urginea, and Dipcadi can be said to have a prepon- derance of chromosomes with terminal to sub- terminal centromeres, Drimiopsis contains a higher number of submetacentrics. It is not possible, therefore, to formulate a common pattern of karyotype evolution in the tribe from mere chromosome morphology. The morphological similarity among the different genera in this tribe is not correlated with similarity in karyotype morphology. If the morphological similarity is a result of com- mon ancestry for the members of the tribe Scilleae, then karyotypes have evolved along various lines. Alternatively, morphological similarity in the tribe may be a result of convergent evo- lution, in which case the tribe would be poly- phyletic. LITERATURE CITED DARLINGTON, C. D. € A. P. WYLIE. 1955. Chromosome Atlas of Flowering Plants. George Allen & Unwin, pe De Wet, J. M. J. 1957. Chromosome numbers in the Scilleae. ora 22: 145-159. GLEDHILL, D. & S. O. OYEWOLE. 1972. The ursi P. of Albuca in West Africa. Bol. Soc. Brot., Ser. 46: 93-106. HEPPER, F. N. 1968. PETA In: Flora of West Trop- ical Africa 3(1): 104-110 Jones, K. € J. B. ae 1967. The chromosomes of the Liliaceae. I— The uM of twenty-five trop- ical species. Kew Bull. 21: 31-38 Levan, A., K. FREDGAR & A. A. SANDBERG. 1964. No- menelature for the centromeric position on chro- mosomes. Hereditas 52: 201-220. Morton, J. K. 1961. West African lilies and orchids. po -9 in A West African Nature Handbook. Long- I , London RE Ss. O. ies in the genus Albuca L. in West Brot., Ser. 2, 46: 149-170. 197 Taxonomic treatment of the genus 1 972. Cytological and cytogenetic stud- Africa. Bol. Soc 202 Annals of the Missouri Botanical Garden Urginea yal a (L.) Baker complex in West Af- rica. Bol. rot., Ser. 2, 46: 163-172. 1 ` Cytotaxonomic studies in the genus n d in West A n U. altissi Em U. viridula Bak. (emend.). Bol. Soc. Brot., Ser 2, 46: 213-223 984a. Pachytene a and karyotype of Mii barteri. Cytologia 49: 81-86. 1984b. M ne and sexual sterility in Drimiopsis barteri. Cytologia 49: 87-93. 1987a. Cytotaxonomic studies in the genus U Irginea Stein in West Africa. II: karyotype evolu- tion in a ips altissima (L.) Baker. Annals Missouri Bot. 26-130. 198 cA studies in his genus Urginea Stein in West Africa. III: the case of Ur- ginea indica (Roxb.) ep in Nigeria. Pss Mis- souri Bot. Gard. 74: -136. Cyiasanami studies in the g Urg inea Stein in West Africa enus .) Kunth. Annals Missouri Bot. Gard. 74: PRESTOEA (PALMAE) IN CENTRAL AMERICA! Andrew Henderson? and Greg de Nevers? ABSTRACT Differences among the morphologically similar Prestoea, Euterpe, and Neonicholsonia are discussed. All species ral America are treated chyspatha, E. williamsii, and E. simiarum are placed in synonymy under P. longipetiolata. Euterpe simplicifrons is transferred to Prestoea. A key and illustrations are provide The neotropical Prestoea Hook f. has re- mained problematic. Moore (1963) argued in favor of keeping Prestoea separate from the morphologically similar Euterpe. Wessels Boer (1965) argued for uniting the two. Doubts about the characters used by Moore have been expressed by Galeano-Garcés (1986) and Henderson (1986). A third genus, Neo- nicholsonia, is also similar to Prestoea but had formerly been distinguished by its spicate inflorescence. Our discovery of a Prestoea having usually spicate inflorescences has raised doubts about the distinctness of Neonichol- sonia from Prestoea. Differences between the three genera as understood by us are given in Table 1, which shows that three groups exist. However, any change in ranking should await a study of all species throughout their neotropical ranges. The Central American species of Prestoea are poorly known; too many names are in use; and the regional floras (Standley, 1937, for Costa Rica and Bailey, 1943, for Panama) are outdated. Here we treat all Central Amer- ican species, based on extensive fieldwork and study of herbarium specimens, including all relevant types. Eight species are recognized. Although Prestoea carderi (W. Bull) Hook f. was reported by Hooker (1890) to have come from Guatemala, the description clearly states that the type material came from Co- Although the species are relatively easy to distinguish in the field, this is not so in the herbarium, where the most useful character is the hairs, or their absence, on the rachillae. Sections of rachilla are illustrated for each species, as is the habit. The flowers and fruits of Central American Prestoea provide few distinguishing characters. There is substantial variation within species. KEY TO THE Vin OF PRESTOEA IN CENTRAL AMERIC la. Leaves entire lb. Leaves regularly pinnate 2a. Sheaths closed, forming a green, maroon, or purple-black crownshaft; rachillae at an- thesis with sessile, trustose, mostly un- ses ched h y brown tomen- 1) P. allenü E Sheathe open, not forming a crownshaft; rachillae at anthesis free of crustose hairs (except P. darienensis) (4) P. integrifolia 2 N o ' Fieldwork in Panama by Henderson was supported by National Science Foundation Dissertation Improvement by a Lawren ial Award. A ; pe Barneby. We acknowledge the help of s Dransfield, Gloria Galeano, Michael Grayum, Heraclio om ead, an Poklon McPherson, Ghillean Prance, Rober 2 The atalie New York Botanical Garden, D New York 10458- 5126, U.S. A. , P.O. Box 3 The Missouri Botanical Garden 299, St. Lou uis, Missouri 63166, U.S.A. Present address: California Academy of Sciences, Golden Gate Park, San yee California 94118, U.S.A ANN. Missour! Bor. GARD. 75: 203-217. 1988. 204 Annals of the Missouri Botanical Garden TABLE 1. Comparison of Neonicholsonia, Prestoea, and Euterpe. Neonicholsonia Prestoea Euterpe Leaf sheaths Pinnae Inflorescences Prophylls Rachillae Sepals of staminate owers Filaments of staminate flowers Staminodes Pistillate flowers — residue posi- Raphe eds open and not forming a crownshaft always spreading always spicate much shorter than the pe- El e but with gro ups of hairs rachillae surface visible united into a cupule with long-acuminate lobes inflexed at apex absent superficial on rachillae; bracteoles not promi- nent subapical few, large, deeply sunken either semiopen and form- ing an asymmetric crownshaft or open* spreading to vertical usually branched, rarely spicate much shorter than the pe- uncular bract occasionally glabrous, but often with various hairs, rachillae surface visible free and imbricate or brief- ly connate at base inflexed at apex 6, dentate superficial on rachillae; bracteoles obscure, rarely prominent lateral numerous, anastomosing to form a network always closed and forming a symmetric tubular crownshaft* pendulous, occasionally spreading always branched more or less equal to the peduncular bract never glabrous, densely white- to brown-ap- pressed tomentose or velutinous, rachillae surface not visible free and broadly imbricate not inflexed at apex absen sunken in rachillae; brac- teoles prominent lateral numerous, anastomosing to form a network * Prestoea inflorescences are more or = terete in gea ae i a with a crownshaft the bud expands before the subtending leaf falls, thereby giving a swollen Euterpe inflorescences are dorsiventrally and o nshaft. compressed and mostly expand after the uice “leaf falls thus abe crownshaft is generally tubular and closed. 3a. Inflorescence spicate or with 2-4 rachillae; pin- on e ET Rachillae at anthesis glabrous, or with p of nae elliptic and abruptly tapered at apex; inflo- rescence horizontal; endosperm homogenous to slightly ruminate P. semispicata . Inflorescence usually with many branches; pin- nae linear and gradually tapered; inflorescence erect or horizontal; endosperm ruminate .............. 4a. Rachillae at anthesis with short, stiff, sim- ple to branched, pen hairs; stam p mooth, shiny, with p iin cleanly ins and nodes prom P. decurrens : Rachillae at anthesis glabrous, or with se - e EN c petals gi ta Rachillae at anthesis with sessile, crustose, branching hairs; inflorescence erect, 2-2.5 m l — LJ ong 2) P. darienensis atches long, loosely intertwined hairs, or densely reddish brown tomentose; inflorescence erect or Borzental ens m 1.6 m à jon ng 6a. kV stem usually less than 2 m a crei erect |... ( . roseospadix 6b. Rachillae at anthesis with patches of long, posa intertwined hairs, or densely red- dish brown tomentose; stem u sually greater than 2 m El inflorescence arching to hor izontal . Rachillae at anthesis densely reddish brown to- mentose; stem thin, often procumbent, less than 2.5 m tall; inflorescence arching; rachis 1-8 cm long with 2-15 rachillae CANA, iia . Rachillae at anthesis with patches of long, lo all; inflorescence arching to horizon- = rachis 33-50 cm long with (18-)42- 48 achillae (7) P. sejuncta Prestoea allenii H. Moore, Principes 9: 72. 1965. TYPE: Panama. Chiriqui: vicinity of Cerro Punta, 2,000 m, 24 Volume 75, Number 1 1988 Henderson & de Nevers 205 Central American Prestoea May 1946, P. Allen 3531 (holotype, BH; isotype, MO). Figures 1, 2. Stems solitary or cespitose, erect, to 12 m tall, 9-18 cm diam. Leaves 6-8; sheaths deciduous, forming a distinct green, maroon, or purple-black crownshaft 68-75 cm long; petiole 20-60 cm long; rachis 1.8-2.5 m long; pinnae 33-51 per side; middle pinnae 118 cm long, 6 cm wide; apical pinna not wider than others. Inflorescence infrafoliar, horizontal; peduncle ca. 20 cm long, 1.8-2.3 cm diam.; prophyll 40-51 cm long, 10.5 cm wide; peduncular bract ca. 1.1 cm long, 4- 6 cm wide, inserted 5-6 cm above base of peduncle; rachis 41-85 cm long; rachillae 23-100, to 73 cm long, at anthesis with sessile, crustose, mostly branched hairs (oc- casionally brown tomentose); flowers gla- brous; fruit 10-11 mm diam.; seeds with ru- minate endosperm. Common name. “Maquenque” (Pana- ma). Distribution. | Eastern Nicaragua to western Panama, in cloud forest 1,000- 3,000 m Additional specimens examined. NICARAGUA. RIVAS: Isla de Ometepe, NW slopes of Volcán Maderas, 11°26- 27'N, 85°30-31'W, 1,000-1,350 m, 24 Feb. 1978, Stevens 6510 (MO). CosTA RICA. ALAJUELA: Reserva Bio- 1986, de Nevers et al. 7779 (MO). CARTAGO: above Finca La Florita on road from Cartago to El General, 2, 450 m, 8 Apr. 1953, ui 6677 (BH). HEREDIA: CHAT Dom ingo de Vara Blanca, 2,200 m, 22 Feb. 1937, Valerio 1597 (F). LIMÓN: Cordillera de Talamanca, Atlantic slope, Valle de Silencio, area just N of Cerro Hoffmann, 4.5 airline m W of Rica/ Panama Eu PA 82°58'W, 2,350-2,450 m, Davidse et al. 2 O (MO). PUNTA- o , Panama, Aug. 1983, Musci, epipetric” (sic), Gómez et al. 21835 (CAS, , 0 m, 7 June 2 (NY); 1, 800 n m, 16 June 1986, Me between Sabalito and Finca López bove Beneficio de Wa Chong, 1 Feb. 1967, Moore & Parthasarathy 9440 (BH); Las Cruces ridge, San Vito de Java, 1,200 m, 2 Feb. 1967, Moore & Parthasarathy PANAMA. CHIRIQUÍ: 2.2 km SW of Cerro e IDAAN water tank, along ridge trail SW of eg de elena above vegetable gardens, 2,100- 2,250 Aug. 1974, Croat 26316 (MO); Cerro "n Macho, i side, 2,150 m, 8?49'N, 82%24'W, Dec. 1985, de Nevers & Charnley 6685 (MO, NY: road to Cerro Punta from Alto Quiel, above Boquete, 3.5 mi. up Cerro Punta road, 1,850 m, 8°51'N, 82°29'W, 16 Jan. 1986, de Nevers & McPherson 6800 (MO, NY); path above Cerro Punta to Boquete, 8*50'N, 82°30'W, ,500 m, moist forest, 16 Mar. 1983, Hamilton & Stock- well 3392 (CAS, MO, NY). Z Prestoea allenii varies considerably with altitude. At lower elevations mature plants are sometimes solitary and have green crown- shafts, and immature plants can lack crown- shafts. At higher elevations the stems are usually cespitose, and the crownshafts ma- roon or purple-black. In Nicaragua and Costa Rica, lower-elevation populations with green crownshafts may represent a distinct taxon (e.g., de Nevers et al. 7779). However, the rachilla hairs are similar to those of P. allenii, and such collections are tentatively referred to that species. Robert Read (pers. comm.) reports that color variability in crownshafts of the same species is not uncommon in cer- tain palms. Prestoea allenii is the largest species of the genus in Central America and occurs at the highest altitude. This is not an uncommon correlation in neotropical palms, occurring, for example, in Geonoma and Chamaedorea. 2. Prestoea darienensis A. J. Henderson, Brittonia 38: 266. 1986. TYPE: Panama. Darién: Serranía de Pirre, on the ridge, 1,130 m, 18 Jan. 1985, A. Henderson & J. Contraires 97 (holotype, NY; is- otype, PMA). Figures 3, 4. Stem solitary, erect, 2.5 m tal, 10 cm diam. Leaves 6; sheaths persistent, not form- ing a crownshaft; petiole 80 cm; rachis 165 cm; pinnae 29 per side; middle pinnae 75 cm long, 5 cm wide; apical pinna not wider than others. Inflorescence inter- or infrafoliar, erect; peduncle 75 cm long, ca. 1.5 cm diam.; prophyll 60 cm long, 4 cm wide; peduncular bract 2.23 m long, 2.5 cm wide, inserted 23 cm above base of peduncle; rachis 80-135 cm long; rachillae ca. 60, 45-72 cm long, at anthesis scabrid with crustose, branching hairs; flowers glabrous; fruit 8 mm diam.; seeds with ruminate endosperm. Distribution. locality. Known only from the type 206 Annals of the Missouri Botanical Garden N A LUC, > FIGURES 1, 2. Prestoea dariensis can be distinguished from the similar P. sejuncta by the rachillae with short, crustose, branching hairs. The in- florescence bud of the type is 2.5 m long, the longest of any Prestoea seen, and only ap- proached by some specimens of P. sejuncta, which occasionally reach 2.1 m. 3. Prestoea decurrens (H. A. Wendl. ex Burret) H. Moore, Gentes Herb. 9: 286. 1963. Euterpe decurrens H. A. Wendl. ex Burret, Bot. Jahrb. Syst. 63: 63. 1929. TYPE: Costa Rica. Alajuela: San Carlos, 24 Mar. 1901, Koschny s.n. (B, destroyed). Neotype (here designated): Costa Rica. Heredia: Finca La Selva on Rio Puerto Viejo just E of its junction with Rio Sarapiqui, 12 Dec. 1984, A. Henderson 50 (NY). Figures 5, 6. Prestoea allenii.— 7. Habit, showing crownshaft and infrafoliar inflorescence (de Nev McPherson 6800) .—2. Part of rachilla, showing sessile, crustose, mostly unbranched hairs (Allen 3531). r O um. & ers Scale Stems usually cespitose, (1-)2-7, erect, 3—7(-10) m tall, 3.2-10 cm diam., smooth and shiny, green or yellow. Leaves 7-9; sheaths 26-34 cm long, semipersistent, not forming a crownshaft; petiole 50-95 cm long; rachis 1.6-2.3 m long; pinnae 36-50 per side; middle pinnae 52-66 cm long, 3-4 cm wide; apical pinna not wider than the others. Inflorescence infrafoliar, white at anthesis, erect or diagonal; peduncle 15-25(-42) cm long, (0.7-)1-3 cm diam.; prophyll 17.5- 19.5 cm long, 3 em wide; peduncular bract 42-81 cm long, 2.5 cm wide, inserted 2.5- 6.5 cm above base of peduncle; rachis 8-27 cm long; rachillae 7-50, 20-56 cm long, at anthesis with short, stiff, simple to branched, white, persistent hairs; staminate petals pilose near apex; fruit 6.5-8 mm diam.; seeds with ruminate endosperm; seedling leaves pinnate. Volume 75, Number 1 1988 Henderson & de Nevers 207 Central American Prestoea 3 Ay y Je ul x ux BT f = + e * FIGURES 3, 4. restoea darienensis. — 3. 97). Habit, showing long, erect, interfoliar pu bud oo —4. Part ee hilla, showing sessile, crustose branching hairs (Henderson 97). Scale bar = 250 Common name. ‘‘Palmitillo’’ (Costa Rica). Distribution. Throughout Nicaragua, Costa Rica, and Panama, usually between sea level and 1,500 m. This species is also known from northern Colombia (Galeano-Garcés, 1986 Additional specimens examined. NICARAGUA. MA- TAGALPA: Comarca Wanawás, beside Río Bilampi, 12%3'N, 85?13'W, 180-200 m, 14 May 1980, Moreno & Ara- quistain 2389 (MO, US). RÍO SAN JUAN: near Cano Chon taleno, 20 km NE of El Castillo (Rio Indio watershed), 200. 7-9 Mar. 1978, Neill. Hind: 18 Apr. 1978, Neill & Vincelli 3495 (MO, US). zELAYA: “Kurinwacito,” 13°8'N, 84°57'W, 80 m, 24 Mar. 1084 Moreno 23880 (US); Mun. Siuna, Caño El Léon, road to Hormiguero, 2 Feb. 1983, Ortiz 730 (MO): Mina Nueva rV ca. 11.3 km N of main road leading W from 14 El Empalme to Rosita, 22 Apr. 1979, Pipoly 5322 (US); Caño between Cerro La Pimienta and El Hormiguero, ca. 13%45'N, 85%59'W, 800-1,000 m, 15 Mar. 1980, Pi- poly 6018 (MO, US); ca. 6.3 km S of bridge at Colonía Yolonia and ca. 0.8 km S of ridge of Serranias de Yolonia on road to Colonia Manantiales (Colonia Somoza), 11*36'N, 84°22'W, 200-300 m, 29-31 Oct. 1977, Stevens 4823 (BH, MO, US); 13-14 Feb. 1978, Stevens 6387 (MO); Cano Costa Riquita, ca. 1.8 km SW of Colonia Naciones Unidas, above road between Colonia Nueva Leon and Colonia Naciones aoe ca. 11?43'N, 84°18'W, 150- 180 m, 6-7 Nov. 1977, Stevens 5034 (BH, MO, US); S slope of Cerro El iin down to near Cano Majagua, ca. 13?45 85°0'W, 800-1,000 m, 9 Mar. 1978, Stev 'ens 6814 (BH, MO); trail from Cerro ol to San Sucio, ca. 13?45'N, : 1978, Stevens 6838 (BH, MO); 6.3 km S of bridge of Colonia Yolonia on road to Colonia Manantiales of Nueva Guinea, 200-300 m, 13 Feb. 1978, Vincelli 250 (MO). Costa RICA. ALAJUELA: E of San Rafael, S of hot springs, W of La Marina, 10?23'N, 84?23'W, 500 m, 19 May 1968, Burger & Stolze 5021 (F, NY); plains of San Carlos, 100 m, 3 Apr. 1903, Cook & Doyle 54 US) Reserva Biológica de San R m to Colonia Palmareña, 10?4'W, 84*32'N, 850- 100 m, 30 May 1986, de Nevers et al. 7780 (MO, NY. slopes of Miravalles, above Bijagua, lower montane rainforest, ca. 1,500 m, Nov. 1982, Gómez et al. 19185 (CAS, CR, MO); vicinity of Guatuso de San Rafael on Rio Frio, 10?43'N, 84?48'W, 80-100 m, 4 Aug. 1949, Holm & Iltis 996 (BH, MO); Río Cuarto, Sarapiqui valley, 1945, Langlois 12 (BH); beside Laguna María Aguilar, 780 m, 28 Mar. 1969, Lent 1531 (NY); 2 km N of 208 Annals of the Missouri Botanical Garden Santa Rosa, 15 km N of Boca Arenal on Quesada- Muelle San Carlos- Los Chiles road, 100 m, 10%38'N, 84°31'W, 28 Apr. 1983, Liesner et al. 15045 (MO, WIS); Rio María Aguilar between Cariblanco and San Miguel, valley of Río Sarapiqui, ca. 700 m, 23 Mar. 1953, Moore 6560 (BH); between Corazón de Jesüs and La Virgen, Rio 5 iquí 1953, Moore 6576 (BH); . n ‘oad from Arenal, 750 m, 1974, Read & Daniels 74-26 (US). CARTAGO: between Rio Pacuare and Grano de Oro, 7 km below Hacienda Moravia, ca. 900 m, 13 Apr. 1953, Moore 6699 (BH). HEREDIA: Finca La Selva, on Rio Puerto Viejo above junction with Rio Sarapiqui, 20 Feb. 1981, Folsom 9056 (DUKE); 27 June 1979, Holdridge 5107 (BH); 17 Oct. 1980, Hammel 10189 (DUKE); 5 May 1982, Hammel 12036 (DUKE); 10 May 1982, Hammel 12168 (DUKE); — & ter 9407 (BH); 18 Apr. 1972, Opler 723 (F); 13 May 1984, Wilbur & Jacobs 34374 (DUKE); 13 May 1984, Wilbur & Jacobs 34393 (DUKE); 1 June 1985, Wilbur 37722 (DUKE). : Hacienda Ta- 6737 (MO). PUNTARENAS: along short cut road to Golfito from Villa Briceno on Interamerican Highway, W side of 59925 (CAS, MO); road to Rincón de Osa, 16.5 km W 8*45'N, 25 May 1986, de Nevers of Rincón de Osa, Osa Peninsula, 30 m, 7 Aug. 1967, Raven 21593 (DS, F, NY). PANAMA. CHIRIQUÍ: above Chiriqui Grande on road to Fortuna Dam, 20 Jan. 1985, Read et al. 85-20b (US). cocLé: along river leading up mountain to Alto Calvario and trout stream from La Junta near Limon, 800- 1,000 m, 12 Oct. 1977, Folsom 5904 (BH, MO); forest at base of Cerro Pilon above El Valle, n 1978, Hammel 1711 (BH, MO). COLON: Rio Guan ide: 3 km upstream of the road, 27 Oct. 1985, de Nevers & Charnley 6107 (MO, NY); 18 Jan. 1980, Moore et al. 10515 (BH) 14 Dec. 1974, Mori & Kallunki 3716 Ariel 15 Mar. 1986, Hammel & Trainer 14775 (MO); . 1983, Nee 7253 (CAS, MO); ridge top Vui. N b. Rio Escandaloso toward Cerro Bruja, 450 m, 27 Apr. 1978, Hammel 2707 (MO). COMARCA DE SAN BLAS: El Llano-Carti road, km 27.6, Rio Pingandi, downstream of road, 9?19'N, 78°55'W, 150 m, 9 Mar. 1985, de Nevers et al. 5065 (CAS, MO); El Llano-Carti road, km 26.5, along Rio Carti Chico, 9°19'N, 78%55'"W, 200 y . 5346 (MO, NY); 9 (MO, NY); Yar ee continental divide between ea and San José, 9°20'N, 79°08'W, 400-500 m, 5 Feb. 1986, de Nevers et al. 6900 (MO, NY); Rio Cangandi at confluence of Quebrada Titamibe, 9°24'N, 79°7'W, 60 m, 8 Feb. 1986, de Nevers & die 7017 (MO, NY); Rio Taindi (Taimdi of maps), ove confluence with Rio Mandinga, 9?25'N, 79°11" Ww. 3 Apr. 1986, de Nevers et al. 7626 (MO, NY) 3 Apr. 1986, de Nevers & Herrera 7629 (MO); trail to Cerro Obu (Habu of maps) from Río Urgandi (Río Sidra), 9*23'W, 78?48'N, 100-300 m, 24 June 1986, de Nevers & Herrera 7988 (MO); Cerro Mali, near Colombian border, 1,400 m, 23 Jan. 1975, Gentry & Mori 13823 (BH, MO, NY). PANAMÁ: pipeline road near Gamboa, 9?10'N, 79°45'W, 100 m, 24 Feb. 1985, de Nevers & Charnley 4942 (MO). The type is no longer extant at Berlin. Burret cited a paratype (Wendland 63) con- sisting only of fruit. This is not at Gottingen among Wendland’s other specimens, and is apparently lost. We therefore designate Hen- derson 50, from the same general area as the paratype, as neotype. Burret (1929) con- sidered P. decurrens to be closely related to Euterpe macrospadix Oerst. and placed them in the same subsection (see Henderson, 1986), but they are unrelated. The confusion prob- ably arose because the type of E. macro- spadix at Copenhagen appears to be a mix- ture of Prestoea leaves (probably P. longipetiolata) and a Euterpe inflorescence. pecimens from the Osa Peninsula in Costa Rica (Raven 21593, Croat & Grayum 59925) and Rio Guanche in Panama have the typical tomentum of P. decurrens, but the inflorescence is less stout, the peduncle is longer and thinner, and there are 7-20 (vs. 50) rachillae. The strongly cespitose (vs. erect) stems are thinner and weaker than usual for the species. 4. Prestoea integrifolia de Nevers & A. J. Henderson, sp. nov. TYPE: Panama. Colón: Santa Rita Ridge, km 21.2, 9°20'N, 79°45'W, 350 m, 24 Feb. 1986, G. de Nevers 7212 (holotype, MO; iso- types, CAS, COL, K, PMA, NY). Figures 8 ^ b aliis speciebus integrifolius inflorescentia erecta, rachillis ium pilis “2 Aa obsitis necnon seminum endospermate ruminato divers Stems cespitose, one well-developed, erect, 2.8-5.8 m tall, 3-4 cm diam.; internodes 3— 8 cm long; adventitious roots forming a prom- inent cone at base of stem, 18-30 cm long, 6—8 mm diam., red, covered with small round projections. Leaves 8-12, spreading; sheaths not forming a crownshaft, 30-38 cm long, closed basally for 15-18 cm, open apically, covered sparsely with closely appressed whit- Volume 75, Number 1 1988 Henderson & de Nevers 209 Central American Prestoea FIGURES 5, 6. Prestoea decurrens. —5. Habit, showing smooth stem, persistent leaf sheaths not forming a crownshaft, erect, interfoliar inflorescence bud, and infrafoliar inflorescence (Langlois 12). Photograph courtesy BH.—6. Part of rachilla, showing short, stiff, simple to branched hairs (Henderson 50). Scale bar = 500 um. ish-brown scales; petiole channeled adaxially, rounded abaxially, 30-50 cm long, covered with closely appressed brown scales; rachis 40-65 cm long, ridged adaxially, rounded abaxially and with scales similar to those of petiole; blade entire, 103-113 cm long, 30- 35 cm wide, deeply bifid at apex for 60-65 cm; veins prominent adaxially, 14-15 per side, with brown scales proximally. Inflores- cence interfoliar, erect at anthesis; peduncle 40-75 cm long, sparsely covered with brown scales; prophyll 21-35 cm long, covered with scales similar to those of peduncle; peduncular bract inserted 5-10 cm above base of pe- duncle, 80-130 cm long, with scales similar to those of peduncle, at anthesis brown on outside, whitish on inside; rachis 18-32 cm long; rachillae 23-28, 35-54 cm long, 1 mm diam. at middle at anthesis; rachis and rachil- lae white at anthesis, becoming reddish in fruit, with sparse, hyaline, simple or branched hairs; triads subtended by a low bract; sta- minate flowers 4 mm long, sessile; sepals 3, free, imbricate, triangular, keeled, membra- naceous, hyaline margined, apiculate, 1.5 mm long; petals 3, free, valvate, lanceolate, 4 mm high; stamens 6; filaments unequal, 0.9-1.5 mm long; anthers dorsifixed at center of the- cae, introrse; thecae unequal, 0.8-1.2 cm long; pistillode 1 mm long, deeply trifid; pis- tillate flowers 2 mm long, surrounded by 2 low bracteoles; sepals 3, free, imbricate, gla- brous, broadly ovate; petals similar to sepals but slightly smaller; gynoecium ovoid, pseu- domonomerous, 1.5 mm long; stigmas sessile; staminodes minute, dentate; immature fruits with ruminate endosperm. Distribution. Only known from the type locality. 210 Annals of the Missouri Botanical Garden tegrifolia. — 7. Habit, Nevers 7212). — 8. Part of rachllay emg hyaline, simple and branched hairs (de Nevers 7212). Scale bar = FIGURES 7, 8. Prestoea Additional specimens examined. PANAMA. COLON: same locality as type, 13 May 1986, de Nevers et al. 7738 (MO, NY); km 22, 500 m, 17 Feb. 1986, Hammel et al. 14473 (MO, NY). Prestoea integrifolia differs from all other Central American members of the genus by its entire leaves. Three other extra-Central American species have entire leaves: P. sim- plicifrons,* P. simplicifolia Galeano, and P. cuatrecasasii H. Moore. The holotype of P. simplicifrons is no longer extant at B, and no isotypes are known, but there is another collection from at or near the type locality (Henderson & Bernal 156). Thus represented, P. simplicifrons has rel- atively short rachillae with a moderate to dense covering of reddish brown hairs, whereas P. integrifolia has long rachillae sparsely cov- ered with hyaline hairs. Prestoea simplicifolia (represented at NY by an isotype and by Henderson & Bernal 140) has a stout, pendulous inflorescence with thick rachillae, in contrast with P. integri- folia, which has a thin, erect inflorescence with thin rachillae. + Prestoea peri (Burret) A. J. Henderson & de Nevers, com v. Euterpe simplicifrons Burret, Engler Bot. Jahrb. 63. 53. 1929. showing entire leaves and interfoliar, erect inflorescence (de Prestoea cuatrecasasii (represented by the original description) has seeds with homoge- nous endosperm, as opposed to the ruminate endosperm of P. integrifolia. Prestoea pubigera (Griseb. & H. A. Wendl.) Hook. f. is sometimes reported to have entire leaves (e.g., Galeano-Garcés, 1986), but all specimens examined, including the type at Góttingen, have the lower part of the leaf with separate but unequal pinnae, and these are joined in the upper part. 5. Prestoea longipetiolata (Oersted) H. Moore, Gentes Herb. 9: 286. 1963. Eu- terpe longipetiolata Oerst., Vidensk. Meddelel. Kjoebenh. 1858: 32. 1859. TYPE: Costa Rica. Cartago: Turrialba, May 1847, A. S. Oersted 6562 (holo- type, C). Figures 9, 10. Euterpe Agi he aio Burret, Bot. Jahrb. Syst. 63: 57. YPE: Costa Rica. Puntarenas: Canas Gordas, 1, 100 n m, Feb. 1897, H. Pittier 11124 (holotype, B destroyed; isotypes, M, US) did Map n Laon: Fieldiana, Bot. 31: 5. 1964 E: Nicaragua. Matagalpa: Cordillera Dite de ere along road to ae diriger cloud forest area, 1,300-1,400 m, 23 Feb. 1963, L. O. Wil- liams, A. Molina & T. P. Williams 24922 (ho- & L. O. Williams, Ceiba & L. lotype, Malortiea simiarum Standle 3: 102. 1952. Euterpe simiarum (Standley Volume 75, Number 1 1988 Henderson & de Nevers Central American Prestoea O. Williams) H. Moore, Principes 1: 145. 1957. TYPE: — RN vicinity of Finca San Roque, sierra f Jinotega, 1,300-1,500 m, 5 July 1947, P. [dud 10923 (holotype, F). Stems solitary or cespitose, often procum- bent, 0.5-3 m tall, ca. 5 cm diam. Leaves 4—8; sheaths persistent, not forming a crown- shaft; petiole 80-240 cm long; rachis 116- 209 cm long; pinnae 21-33 per side; middle pinnae 45-56 cm long, 1.5-3 cm wide; apical pinna often markedly wider than others. In- florescence interfoliar or infrafoliar, arching; peduncle 12-100 cm long, 3-6(-11) mm diam.; prophyll (9.5-)15-30 cm long, 1.5- 2 cm wide; peduncular bract (32-)56-114 cm long, 2-4 cm wide; rachis 2-9 cm long; rachillae (2-)3-8(-20), (8-)16-35 cm long, at anthesis densely reddish brown tomentose; flowers glabrous; fruit 6-11 mm diam.; seeds with ruminate endosperm. Distribution. From Nicaragua to west- ern Panama (Chiriqui), 1,000-1,800 m. Con- trary to a report by Wessels Boer (1971), this species does not occur in Venezuela (Hen- derson & Steyermark, 1986). Additional specimens examined. NICA JINOTEGA: Ocotillo near Sta. Lastenia, Cordillera Central de Nicaragua, 1,550 m, 17 Jan. 1965, Williams et al. 27806 (F, NY). MATAGALPA: Cerro Carlota, 12%58'N, 85%52'W, 1,250-1,300 m, 23 Oct. 1982, Moreno Ded (MO); Cerro El Picacho, N of Selva Negra, 1 85°55'W, 1,500 m, 7 July 1983, Moreno eon (MO, US); 7 July 1983, Moreno 21671 (MO); Cerro Carlota, 12%58'N, 85%52'W, 1,250-1,300 m, Moreno 18149 June 1983, Stevens 9168 (MO, US); along Highway 3, ca. 1 km NW of La Fundadora entrance, unnamed peak ca. 500 m W of per ca. 13%1'N, 85%56'"W, 1,450- 1,520 m, 24 May 1981, Stevens & Henrich 20456 (MO); W sl f erro El oN, ca. 13°0'N, 85°55'W, 1.350- 1,590 m, ur June 1983, Stevens & Moreno 22168 (MO, US); Cordillera Central A irn ragua, along road to La Fundadora, 1,300-1, p2 Feb. 1963, het et al. 24918 (F); Corde Conca de Nicaragua, Xelaju, 13%02'N, 85°55 13 Feb. 1965, Williams et al. 29266 (F, RU [id d ALAJUELA: Buena Vista, road to San Carlos valley, 600 m, 11 Apr. 1903, Cook & Doyle 38 (US); Juan Vinas, Reventazón valley, near Juan Vinas River, 1,00 Apr. 1903, Cook & Doyle 173 (US); along Camino sn de Hule, SE of Planatillo i ego 1,200-1,400 m, July 1976, Croat 36792 (MO); Reserva Biologia de San Ramon, road from Las Lagunas to Colonia Palmarena, 10%4'N, 84°22’W, 850-1,100 m, 30 May 1986, de Nevers et al. 7769 (MO, NY); near Rio Naranjo, 2 km W of Orosi, 1,400 m, 16 Jan. 1977, Lent 4066 (F); slopes of ridge separating Rio Paz Grande and Rio Paz Chiquita, about halfway between Vara Blanca and Cari- ped valley of Rio Sarapiqui, 1,340-1,500 m, Moore 9 (BH); Paraiso, ye of Muñeco, 9-10 Mar. 1974, ne & puro 74-80 (US); Guadalupe de Zarcero, 1,200 m, 1973, ape A567 (F); La Pena de Zarcero, een Alfar Ruiz, 1,375 m, 23 Jan. 1939, Smith 1544 (NY); El Muñeco. S of Navaro, 1,400 m, 8 Feb. 1924, Standley 33600 (F, rim CARTAGO: Canon 2s 19 el Rio Grande de Orosi y Aluvión, 23 3, Chacon et al. 1489 (MO); Rio Tambor, 3 k SE p Cachi, 1,420 m, 22 Aug. 1971, Lent 2060 (F); about 5 km beyond Hacienda Moravia, 1,000-1,200 m, 13 Apr. 1953, Moore 6686 (BH); fords of 2 Vueltas, Tucurrique 5 m, Dec. 1898, Tonduz 12924 (F, US). PUNT ARENAS: viris lona Linda, 1 mi. of Canas Gordas, 1, 7 Feb. 1973, Croat 22232 (CAS, K, MO); foothill of i Cordillera de Talamanca, around Tres Colinas, 9°07" 83°04 4'W, 1,800-1,850 m, 20 Mar. 19 of Rio San Luis just S of Monteverde, 10°16'N 1,000-1,200 m, 18 June 1985, Hammel & Haber 13924 (MO); 1,800 m, 16 June 1986, Hammel 14961 (NY); Finca Las Cruces, on trail to Rio Java, ca. 1,000 m, 31 Jan. 1967, Moore & Parthasarathy 9426 (BH); Mon- teverde, along foot trail in forest reserve, 3 Nov. 1974, Moore et al. 10170 (BH); Las Cruces, Finca Kilpauk, 15 Dec. 1961, Read 655 (BH); San Vito de Coto Brus, Las Cruces Botanical Garden, Jan. 1985, Wilson s.n. (BH). san josé: El General, 1,490 m, Feb. 1939, Skutch 4184 (US); between San Isidro and La Georgina, 17 Nov. 1973, Moore & McAlpin 10150 (BH); 17 Nov. 1973, Moore & McAlpin 101504 (BH). PANAMA. BOCAS DEL TORO: La Fortuna Dam Area, N of dam along continental divide trail W of oleoducto road, 8°47'N, 82°15'W, 1,200- O m, 11 Feb. 1986, Hammel & McPherson 14458; near continental divide in vicinity of Cerro Colorado, 9.4 road miles from Chami camp, ca. 8°35'N, 81?45'W, ca. 1,700 m, 15 Apr. 1986, McPherson 8917 (MO, NY). CHIRIQUÍ: 9 mi. from Rio Chiriquí Viejo bridge near Nueva California on road to Rio Sereno, 7 Apr. 1979, Hammel et al. 6829 (MO Malortiea simiarum was originally distin- guished by its pinnate leaves and little- branched inflorescence when being compared with Reinhardtia (Malortiea). The size, de- gree of branching, and pubescence of the type fall within the range of variation observed in Costa Rican material of P. longipetiolata. Euterpe brachyspatha, as judged from the original description, was named after a mis- interpretation of the inflorescence. Burret (1929) described the spadix (inflorescence) as 98 cm long and the spathe (peduncular bract) as 17 cm long, which is impossible. The ho- 212 Annals of the Missouri Botanical Garden FicunEs 9, 10. Prestoea longipetiolata. — 9. Habit, showing procumbent stem and arching, interfoliar inflo- rescence (Moore 9426) . Photograph courtesy of BH.— 10. Part of rachilla, showing dense tomentum (McPherson 8917). Scale bar = 250 um. lotype is lost, and the isotypes are incomplete. However, the specimens described by Burret represent P. longipetiolata. The Munich iso- type has the broad apical pinna typical of most but not all. specimens of P. longi- petiolata. À topotype, Croat 22232, is typ- ical P. longipetiolata. Euterpe williamsii was originally contrast- ed with E. brachyspatha, presumably rep- resented by the original description. In Bur- ret's (1929) key, E. brachyspatha and F. longipetiolata are contrasted in the same couplet. [n its protologue, Glassman did not contrast E. williamsii with E. longipetiolata. Euterpe williamsii agrees in the diagnostic characters of size, branching, and pubescence of the inflorescence with E. longipetiolata. 6. Prestoea roseospadix (L. Bailey) H. Moore, Principes 9: 73. 1965. Euterpe roseospadix L. Bailey, Gentes Herb. 6: 201. 1943. TYPE: Panama. Chiriqui: vi- cinity of Bajo Chorro, 1,900 m, 20-22 July 1940, R. E. Woodson & R. W. Schery 623 (holotype, MO; isotype, BH). Figure 11. Stems solitary, erect, 0.3-3 m tall, 8-10 cm diam. Leaves 4-6; sheaths persistent, not forming a crownshaft; petiole 61-76 cm long; rachis 120-125 cm long; pinnae 21-27 per side; middle pinnae 38-50 cm long, 2-2.5 cm wide; apical pinna not wider than others. Inflorescence infrafoliar, erect; peduncle 16- 38 cm long, 0.5 cm wide; prophyll 20-23 cm long, 2 cm wide; peduncular bract 70- 80 cm long, 2 cm wide, inserted ca. 14 cm above base of peduncle; rachis 16-40 cm long; rachillae 9-16, 20-40 cm long, gla- brous; flowers glabrous; fruit 9-10 mm diam.; seeds with ruminate endosperm. Volume 75, Number 1 1988 Henderson & de Nevers 213 Central American Prestoea Distribution. Western Panama (Chiri- qui and Veraguas) at altitudes around 1,500 m Additional specimens examined PANAMA. CHIRIQUÍ: Cerro Horqueta, 2,100 m, 24 July 1966, Blum & Dwyer 2665 (MO); 24 July 1966, Blum & Dwyer 2671 (MO); 8 Aug. 1967, Kirkbride 162 (MO); Bajo Chorro, Bo- quete, ca. 2,000 m, 11 Jan. 1938, Davidson 100 (F); lower slopes of Cerro Pate Macho, 8?49'N, 82?24'W, 1,600 m, 31 Dec. 1985, de Nevers & Charnley 6697 (MO, NY); 17 Jan. 1986, de Nevers & McPherson 6829 (MO, NY); La Fortuna hydroelectric dam project, behind camp, 1,300-1,400 m, 23 . 1978, Hammel 2255 (BH, MO). vERAGUAS: valley of Río Dos Bocas on d between Alto Piedra (above Santa Fé) and Calovébora, 350-400 m, 29 Aug. 1974, Croat 27440 (MO) "i Prestoea sejuncta L. Bailey, Gentes Herb. 6: 201. 1943. TYPE: Panama. Ca- nal Area: Madden Lake area, upper Rio Pequeni, 100 m, 29 July 1941, A. G. B. Fairchild & D. Jobbins 2635 (ho- lotype, BH; isotype, MO). Figures 12- 14 Stem solitary or cespitose, erect, 5-9 m tall, (4.5-)9-13 cm diam. Leaves 5-8; leaf sheaths persistent or deciduous, not forming a crownshaft; middle pinnae 61-86 cm long, 3-4.5(-6) cm wide; apical pinna not wider than others. Inflorescence interfoliar, erect, or arching, or horizontal, straight or curved; prophyll 13-45(-75) cm long, (1.2-)3-5 cm wide; peduncular bract 69-156(-215) cm long, (2.5-)3.6-5.5 cm wide, inserted (2-)4.5-13 cm above prophyll; peduncle 23- 71 cm long, 0.6-1.4 cm wide, narrow, cy- lindric, not flaring at base; rachis (15-)33- 50 cm long; rachillae (18-)42-48, 30-70 cm long, essentially glabrous but with patches of long, loosely intertwined hairs; flowers gla- brous; fruits 7-10 mm diam.; seeds with rum- inate endosperm. Distribution. Known from central Pan- ama (Chiriquí, Coclé, Comarca de San Blas, Colón), 100-1,100 m. This species is also reported from coastal Ecuador by Dodson & Gentry (1978). Other specimens from Ec- uador (e.g., Balslev & Henderson 62107) are clearly referable to this species, and it presumably occurs in intervening Colombia. Additional specimens examined. PANAMA. CHIRIQUÍ: o continental divide trail west of road, 1,150 m, 8 82?15'W, 18 Jan. 1986, de Nevers & MePhearson 6849 MO, NY). COLÓN: trail from Alto Pacora to Cerro Brew- ster, 9218'N, 79?16'W, 700 m, 18 Nov. 1985, de Nevers et al. 6223 (MO, NY). cocré: El Valle de Antón, La Mesa, ca. 1,000 m, 2 Sep. 1941, Allen 2740 peri vi BH, MO). COMARCA DE SAN BLAS: Cerro Brewster, 9?18'N (eap W, 850 m, 25 Apr. 1985, de Nevers et al. 5541 NY). PANAMÁ: 3 mi. N of Cerro Azül, 26 July 1970, iid 11589 (MO); Cerro Jefe, ca. 700 m, 20 Jan. 1980, Moore et al. 10522 (BH); Rio Pequeni, slopes of Cerro San Francisco, 150-300 m, 9?22'N, 79°31'W, 29 Nov 1985, Henderson & Brako 505 (MO, NY); Gorgas Me- morial Lab's yellow fever research camp, ca. 25 km of Cerro Azul on Rio Piedras, 550 m, 20-22 Nov. 1974, Mori & Kallunki 3454 (BH, MO) Among the specimens examined, there ap- pear to be two ecotypes, which may turn out to represent distinct taxa. At higher altitudes (900-1,200 m) in premontane rain forest (Holdridge et al., 1971), the inflorescence is straight and erect in bud (Fig. 13) and rela- tively long and thick (1.5-2.1 m x 3-5 cm). At lower altitudes (10-200 m) in tropical moist forest and tropical wet forest (Holdridge et al., 1971), the inflorescence is curved and horizontal in bud (Fig. 14) and is relatively short and thin (75-80 x 1.2-2.5 cm). When Bailey described P. sejuncta, he cited two specimens, one of each ecotype. The holotype is the short-inflorescence ecotype, and the paratype, Allen 2740, is the long-inflores- cence ecotype. Although letters accompa- nying Allen 2740 from Paul Allen to Bailey clearly outlined this variation, Bailey included only the dimensions of the smaller Madden Lake plant. 8. Prestoea semispicata de Nevers & A. . Henderson, sp. nov. TYPE: Panama. Comarca de San Blas: Cerro Brewster, 9°18'N, 79?*16'W, 800 m, premontane rain forest, 19 Nov. 1985, G. de Nevers, A. Henderson, H. Herrera, G. Mc- Pherson & L. Brako 6290 (holotype, MO; isotypes, AAU, BH, CAS, COL, FTG, K, NY, PMA). Figures 15, 16. omnibus congeneribus inflorescentia simplice vel pa auciramosa necnon seminum endospermate subruminato diver 214 Annals of the Missouri Botanical Garden Wt!) Prestoea. — 11. P. roseospadix, part of rachilla, showing absence of hairs (Woodson 623). FicunEs 11, 12. Scale bar = 400 um.— 12. P. sejuncta, part of rachilla, showing long, loosely intertwined hairs (de Nevers € Henderson 6411). Scale bar = 500 um Stems cespitose, only one well-developed, to 145 cm long, 3.5-9 cm diam., often pro- cumbent and partly subterranean; roots vis- ible above ground, spiny and occasionally swollen. Leaves 4-10, arching to erect; sheaths not forming a crownshaft, 17-20 cm long, brown, persistent; petiole 29-100 cm long, densely covered with closely appressed, brown hairs; rachis 52-180 cm long; pinnae 12-20 per side, elliptic, abruptly and asym- metrically long-apiculate, glossy green adax- ially, lighter green abaxially; middle pinnae 19-51 x 2.5-6 cm. Inflorescence infrafo- liar, protandrous, arching, borne at or near ground level; peduncle 6-50 cm long, 1.5- 3 mm diam., terete, with scattered brown scales; prophyll erect and persistent in leaf axil, inserted at base of peduncle, (1.2-)4.5- 19 cm long, 1.1-1.8 cm wide, dorsiventrally compressed, ancipitous, splitting apically; pe- duncular bract (9-)29-90 cm long, 1-2 cm wide at middle, inserted (0.7—)1.5—7 cm above insertion of prophyll, terete in bud, apically pointed, brown at anthesis, soon dropping; rachillae 1(-4), 8-30 cm long, glabrous; pe- duncle and rachis white at anthesis and be- coming red in fruit; triads densely crowded and borne to apex of rachillae, slightly sunk- en, subtended by a low bract; staminate flow- ers 4 mm long, sessile or borne on a short, flattened pedicel, white; sepals 3, triangular, gibbous, imbricate below, 1 mm long, ciliate; petals 3, free, lanceolate, valvate, 4 mm long; stamens 6; filaments slightly flattened, with long reflexed apex in bud, 3 mm long; anthers dorsifixed, latrorse, 2 mm long; pistillode prominent, as long as stamens in bud, briefly trifid at apex; pistillate flower 3 mm long, surrounded by 2 low bracteoles; sepals 3, free, imbricate, minutely ciliate; petals 3, free, val- Volume 75, Number 1 1988 Henderson & de Nevers 215 Central American Prestoea a.— 12. Str (Henderson & Brako rescence with short peduncle and few rachillae (de Nevers & l 641 FIGURES 13, 14. peduncle and numerous rachillae Prestoea sejuncta vate above, imbricate below; gynoecium ovoid, pseudomonomerous, 2 mm long, the ovule attached laterally; stigmas sessile, elongate, not recurved at anthesis; staminodes denti- form; fruit spherical, 8 mm diam., with lateral stigmatic residue, black; epicarp smooth; me- socarp fleshy; seed spherical, 7 mm diam., the endosperm homogenous to slightly ru- minate; embryo basal. Common name. “Siler burwi" (Kuna, Panama). Distribution. Known only from the low mountains of Central Panama, from the west- ern end of the Serranía de San Blas to El Cope in Coclé, where it is uncommon between 350 and 850 m. It grows on steep slopes and ridge tops in premontane rainforest and trop- ical wet forest. traight inflorescence bud and long inflorescence with elongate — 14. Curved inflorescence bud and small inflo- 1). Additional specimens examined. PANAMA. COCLE: El Valle de Antón, Cerro Gaital, 8°37'N, 80°6’W, 1,000 m, 26 Nov. 1985, de Nevers et al. 6351 (CAS, MO, NY); continental divide above El Copé, 900 m, 19 Jan. 1978, Hammel 967 (MO); 27 Nov. 1985, de Nevers et al. 6381 (MO, NY, PMA, other duplicates to be distrib- uted). COMARCA DE SAN BLAS: same locality as type, 16 Oct. 1984, de Nevers et al. 4027 (MO, NY); 25 Apr. 1985, de Nevers et al. 5550 (MO, NY); 19 Nov. 1985, de Nevers et al. 6242 (MO, NY); 21 Dec. 1985, Hammel km 16.5, 9?10'N, 78%55'"W, 350 m, 13 Mar. 19 Nevers & Herrera 5153 (MO); 12 Mar. 1986, de Nevers et al. 7371 (MO); 22 Nov. 1985, de Nevers & Henderson 6312 (MO, NY); 8 Mar. 1986, de Nevers & Herrera 7260 (MO, NY); 18 June 1986, de Nevers & Herrera 7945 (CAS, MO); trail from Cerro bud s e Río Titamibe, 9%24'N, 79*8'W, 60-100 m, 28 Jan. 1985, e Nevers et al. 4719 (CAS, MO, NY); trail de Rio Esadi to Cerro Banega, 300-530 m, 9?23'N, 78*51'W, 21 Dec. 1985, de Nevers & Herrera 6671 (CAS, MO, NY); Yar Bired (Cerro San José), continental divide be- tween Cangandi and San José, 9°20'N, 79°8’W, 400- 500 m, 7 Feb. 1986, de Nevers & Herrera 6961 (MO, NY); trail to Cerro Obu (Habu of maps) from Rio Urgandi (Rio Sidra), 9°25'N, 79?11'W, 100-300 m, 3 Apr. 1986, 216 Annals of th Missouri Botanical Garden FIGURES 15, 16. Prestoea semispicata. — 15. Habit, cence, and abruptly tapering me (de Nevers et al. 6290) .— 16. Part of rachilla, showing absence of hairs 250 u (de Nevers 5550). Scale bar de Nevers & Herrera 8029 (CAS, MO); Cerro Obu, 18*48'W, 9°23'N, 25 June 1986, de Nevers & Herrera 8055 (CAS, MO). Prestoea semispicata is unusual in the genus by the usually spicate inflorescence, seeds with homogenous to slightly ruminate endosperm, and shape of the pinnae. In pop- ulations where individuals with branched in- florescences occur, spicate inflorescences are also found. In fact, branched and spicate in- florescences form on the same plants. Pres- toea semispicata appears morphologically similar to that group of Prestoea character- ized by a weakly developed stem, absence of crownshaft, markedly unequal prophyll and peduncular bract, short rachis with few ra- chillae, and seeds with either homogenous or ruminate endosperm (Henderson, 1986). this group, P. semispicata shares with P. cuatrecasasii and P. schultzeana (Burret) showing procumbent stem, arching, infrafoliar inflores- H. Moore seeds with homogenous endosperm but differs in the shape of the pinnae and in having usually spicate inflorescences. Some specimens of P. longipetiolata from Nica- ragua and P. pubens H. Moore from Colom- bia have two or three rachillae on the inflo- rescence, but these are densely tomentose and not glabrous as in P. semispicata. LITERATURE CITED de L. H. 1943. Palmaceae. In Flora of Panama. n. Missouri Bot. Gard. 30: 327 BURRET, “M. 1929. Jahrb. Syst. 63: Dopsow, C. H. & A. 1 GENTRY. 1978. Flora of ns Rio Pa 2e Science Research Center. Selbyana 7-396. ie Gattung Euterpe Gaertn. Bot. -76. GALEANO- eus És, G. . Two new species of Palmae from Colombia. Brittonia 38: 60-64 gie cien A. 1986. A new Prestoea Pima from nama, with notes on the genus. Brittonia 38: 266- 268 Volume 75, Number 1 1988 Henderson & de Nevers Central American Prestoea J. STEYERMARK. 1986. New palms from Venezuela. Brittonia 38: 309-313 HoLDRIDGE, L. R., W. C. GRENKE, W. H. HATHEWAY, T. Lianc & J. Tosi, Jk. 1971. Forest Environments in Tropical Life Zones. Pergamon Press, New York. ia y a 90. Prestoea carderi. Bot. Mag. 116: t. be H. E 63. The types and lectotypes of some palm genera. Gentes Herb. 9: 245-274. STANDLEY, P. C. 1937. Palmae. In Flora of Costa Rica. Publ. Field Mus. Nat. Hist., Bot. Ser. 18: 107-128. WesseLs Boer, J. 1965. Palmae. Flora of Surinam. E. J. Brill, Leiden. l 971. Clave [sg las palmas Venezolanas. Acta Bot. Venez. 6: 299-362. KARYOTYPE VARIATION IN PANCRATIUM HIRTUM A. ; (AMARYLLIDACEAE)! S. O. Oyewole? ABSTRACT Natural populations of Pancratium hirtum A. Chev. from different ecological niches show definable morpho- logical variation. Samples were grown phological variants (momo type differentiation involves at least chromatin material per nucleus. The chromosome basic number o otype is reported for the first time in the genus. Population presence of accessory chromosomes in one morph in an experimental garden and investigated karyotypically. Five mor- med as showing aryotype di divergence is less pronounced than karyotypic divergence Pancratium L. is represented in West Tropical Africa by two species, P. hirtum A. Chev. and P. trianthum Herb. Their tax- onomy was well documented by Morton (1965), who reported the somatic chromo- some number of 2n — 22 for them. Morton did not include analysis of the karyotypes. Pancratium hirtum, with large chromo- somes, is especially suitable for study of chro- mosome morphology. MATERIALS AND METHODS Pancratium hirtum grows in a variety of niches in savanna vegetation, where it exhibits minor but definable differences in leaf size, leaf color, extent of pubescence, length and form of the peduncle, and texture of the out- ermost tunic of the bulb (Table 1). Five morphological groups (morphotypes, Fig. 1) were recognized during field study and samples were collected. Not fewer than 40 bulbs of each morphotype were grown sepa- rately on adjacent beds in the experimental garden. Their habitats are described in Ta- ble 2. Each bulb in each morphotype was ex- amined cytologically using root tip squashes as outlined in Darlington & LaCour (1969). Chromosome counts were made from several metaphase plates in each preparation. Mea- surement of chromosomes using calibrated micrometer eyepiece graticule was impracti- cable due to their unusual lengths. Photo- graphs of metaphase plates were taken at x 7.5 ocular and x 40 objective of the Olym- pus (Vanox model) Research Microscope. Chromosomes were measured from the pho- tomicrographs. The measurements were pooled from 15-20 complements for each morphotype, and average lengths were de- termined. Chromosome morphological deter- minations were according to Levan et al. (1964) as modified by Adhikary (1974). Evi- dence of chromosomal changes was mani- fested in unequalness of members of homol- ogous chromosome pairs (such changes usually affect one arm of a member). In such in- stances, the unaffected arm length was em- ployed to identify the members, and the chro- mosome index was based solely on the length of the longer member of the pair. Idiograms were constructed from enlarged photomicro- graphs. ! The University of llorin Senate Research Grant No. 8.184.34 is gratefully di oie iaa ? Department of Biological Sciences, University of Ilorin, P.M.B. 1515, Ilorin, Nigeria ANN. Missouni Bor. GARD. 75: 218-225. 1988. Oyewole Karyotype Variation in Pancratium hirtum Volume 75, Number 1 1988 219 RESULTS Karyotype data are summarized in Table z E E ° E 3. Figure 2 shows metaphase plates of the = Ë bs £ Ë somatic complements, and Figure 3 presents HE À “Á the idiograms. All morphotypes have a so- ae "E Š matic chromosome number of 2n = 22, ex- = eg 4 5 = cept E, in which the complement is 2n = EY Š S 22 + 4 B-chromosomes. e 3 X 53: + e Morphotype A. The chromosomes vary in length from 5.5 um to 19 um, with a total T "T E length of chromatin material of 216.5 + 6.5 E m + um. The complement (Figs. 4A, 5A) consists > S = je of four pairs with median to submedian cen- Ps >» tromeres (1st, 2nd, 3rd, and 11th), one tel- š E E = 5 i. ocentric pair (8th), and six pairs with terminal > ™ BE $ to subterminal centromeres (4th-7th, 9th and TẸ n ES: 10th). The longest two pairs have unequal = = members, one member of each having lost a portion of its long arm. E " 2 š Morphotype B. Chromosomes vary in = al * length from 11.38 um to 41.75 um, withan , AMETE average total chromatin length of 439.5 + & |9] 2 g = : Š 18.5 um. The complement (Figs. 4B, 5B) 2 E E 5-5 pl consists of three pairs (lst, 2nd, and 11th) S ME ii BS with median to submedian centromeres, and Í -= = eight (3rd—10th) with terminal to subterminal š centromeres. The second- and third-longest E se g P2 pairs have unequal members: a member of 2 a8 n " the former having lost a portion of the long € El E ii s E arm, while one member of the latter lost a E ES sg E portion of its short arm. > E - £ t E es M: B3. Morphotype C. Chromosomes vary in E 23 3 - E 3 length from 6.3 um to 19.14 um, and the < ei ci total chromatin length averages 237.3 + 23.3 £ um. The complement (Figs. 4C, 5C) consists E o À of three pairs (Ist, 2nd, and 11th) with me- E Ë T y 3 dian to submedian centromeres and eight pairs — * E só ° 3 (3rd- 10th) with terminal to subterminal cen- S lus 2 E g tromeres. The third-longest pair has unequal la 33 B ° = members—the shorter member has a shorter 3 E E eie E So second arm. One member of the shortest pair š COR 2 also shows loss of a portion of one arm. i: n à Morphotype D. Chromosome length P 2 varies between 8.6 um and 27.3 um, with an E z Pe average total chromatin length of 340.4 + E E 49.8 um. The complement (Figs. 4D, 5D) high, trigonal in trans- high, obtusely trigonal, high, trigonal in trans- gonal in transverse sec- apex, trigonal in trans- verse section, green, verse section, green, tion, green, glabrous, receptacle 1-3 mm green, glabrous, recepta- cle 1-3 mm high receptacle 1- > glabrous glabrous, receptacle 1- 3 mm high light green, 22-31 cm glabrous, receptacle 1- 3 mm high light green, 26-33 cm dark green, 20-26 cm light green, 24-35 cm dark green, 21-30 cm Leaf long, 5-9 mm wide long, 5-9 mm wide single, deciduous, half-en- long, 4-9 mm wide long, 9-15 mm wide single, deciduous deciduous, half-en- circling, single lipped, green , single about > Bract single, deciduous, half-encir- single, persistent until cling, single-lipped, dark green fruit ripening, fully en- circling, 2-lipped, green circling, single lipped, light green half-encircling, single lipped, yellowish green 220 Annals of the Missouri Botanical Garden FIGURES 1, 2.— of eru hirtum morphotypes, consists of five pairs (1st-3rd, 10th, and 11th) with median to submedian centromeres and six pairs (4th—9th) with terminal centromeres. Five of the eleven pairs (1st, 2nd, 3rd, 8th, and 10th) show evidence of loss of portions from one member of each pair. First and 10th pairs show loss in the long arm; the second and third pairs show loss in the short arm, 1. Pancratium hirtum Vili dn with mature fruits, x 0.2. —2. Mature fruits, with peduncles, x 0.4 and the eighth shows loss in one of the two equal arms. Morphotype E. The autosomes vary in length from 5.6 um to 17.9 um, and the B-chromosomes vary between | um and 1.5 um. The average total chromatin length is 219.14 + 2.5 um. The complement (Figs Volume 75, Number 1 1988 Oyewole 221 Karyotype Variation in Pancratium hirtum | | i 5 FIGURE 3. Map 10 of Nigeria showing locations where Pancratium hirtum populations were het di in Oyo and Kwara States. Broken line = state boundary; circles = state capital city; solid lines = 4E, 5E) consists of five pairs (1st, 2nd, 4th, 8th, and 11th) with median to submedian centromeres, and six pairs (3rd, 5th, 6th, 7th, Oth, and 10th) with terminal to subterminal centromeres. All the B-chromosomes are telo- centric. One member of the second-longest autosomal pair has chromosome loss in the long arm, while one member of each of the eighth and ninth pairs shows loss in the short arm. DISCUSSION Previous chromosome counts in Pancra- tium show a chromosome number of 2n = 22 (Ponnamma, 1978; Lakshmi, 1980). This number is confirmed here except that the presence of accessory chromosomes had not been reported previously in the genus. e five karyotypes share a basic plan: the largest two and the smallest pairs of chro- mosomes are metacentric while all the others are acrocentric. Deviations from this plan consist of increase in the number of meta- centrics (A, D, and E with an increase of one, two, and two pairs, respectively). Apart from these, there is evidence of structural changes in the chromosomes as a result of loss or gain of segments. This is common to all the karyo- 222 Annals of the Missouri Botanical Garden TABLE 2. Sources of material of Pancratium hirtum. Herbarium Voucher Habitat Mor- pho- types Collection Site A Iseyin-Igbetti about 150 km northwest of Iba- dan B Affon, 25 km southeast of Ilorin C Shao, 22 km northwest of Ilorin D Oke-Oyi, about 20 km north of Ilori E Okene- uvis Road, bout 50 km from b. 800/2106 in Uni- versity of Ilorin Herbarium (IUH) S00/2199 in IUH S00/2200 in IUH 500/2201 in IUH 500/2202 in IUH deciduous woodland, in dark humus, under the shade of trees such as Butyrospermum paradox- um, Lophira lanceolata, etc. disturbed woodland, in dark humus soil on shallow inselbergs and foot of rocky hill under the shade of Parkia biglobosa open and exposed brown soil of old mats of Afrotri- lepis pilosa on rock outcrops in savanna grass- land open brown humus soil among rock boulders in sa- vanna grasslan dark brown gravelly clay-loam in floodplains in sa- vanna woodland under stands of Khaya senegal- nsis FIGURE 4. Somatic metaphase complements of Pancratium hirtum morphotypes. Oyewole 223 Volume 75, Number 1 1988 ion in Karyotype Variat Pancratium hirtum S6 + vYU6Ic 8' 6v + VOTE EEG + E LES S'8l + COEP G9 + €9Ic6 y1dua] uneuro1qo [8101 OIOI — — — — səurosoutoiuo OI CI — — — — Á1088999 y uprpəu: uerpəu: uerpoaur ueipour ueipoul UOH?20| 919W014U99 Gil Ol O'I cUI cl Supp L9'S L9'8 ££ 9 8t TI as qiguo| eurosouroqo Il [Buu Ia1qns uerpour ppuru} [guru] peura} uonpoo| Ə1əutonuəo OT LOIT 0'Tc S 9 097 SMBAT Sc 9 001 cel sl El EPA qi2ue[ aurosouroJqo OI jeuluLiayqns purua} peura} [eurturiə1 Ieurur:ə1qns uOne2o| 919uro.1uəo 0 € eel Q'TI O'STI 0'9 me £9'9 SL'OI 0'8 O' TI OL qigue| aurosoutoqo 6 u?ipaurqns peura} peua} peutua} puua} uon?oo| 219U101}U39 voz IL LI 8€ 076 0 — 8t 66 0T 8€'8 Q'SI OL u18uə[ surosouro1y9 8 peua} peura} ]gurur1o1 peua} [euturiə1 uon?o2o| a1aurodqua2 Sc vel 0'8 ori OFT eng: S68 041 0'6 SSI ye u18uə[ aurosouro.1y9 E ppuru} peua} peutua} purua} peua} UON BIO] a1auro1jua2 asl OTT 6 Ic OST OTI SUBA cZ'8 0'€l SZ'6 0'OI Ju) q13ue[ eurosouro1qo 9 [eururiə1 peua} peura} Ieuturiə1 [eururiə1 UONRIO] 919W1014U39 0"S€ Sc OI £9'€l O'vI 0'8 meme 0'6 Q'SI $26 S/'81 06 q1gue| aeurosouroq5 G ueipauiqns [eururiə1 [euro] [euro [euruLia] uonpoo|[ a1euio1ju92 LOTI e801 66 01 LUEI S76 an[pa-4 88 [I S¿ ¿I 66 II Sc T6 SL'OI qi2ue| eurosouro1qo T peua} u?ipaulqns Ieururiə1qns Ieuturiə1qns uerpoaurqns UOLO] a1ouro1]ua32 Sc TI 957 TOE 63€ ec sic. xim €£9'cI 90' T6 c6 El SL SG Sç ¿l yp3uəj əurosotuoiuo t ue?ipaul ueipaur ueipoul uetpaur ueIpeul UON BIO] a1euroljua2 Lil 8vV'I 9 I LE'I 01 dl S YI 08'Sc GP SI SG L6 091 u18uə[ əurosouro1uo G ueIpoul upipəu: uepəu:r uenpəur u?ipaul U01800] Ə919uronuəo LE'I oll UNI Gl UT an[pA-7 887I S6 L6 PUOI SL 0'61 i2ue[ aurosouro1qo I iC a ) g V sengojouro ‘(wn ui yi2ua] awosowosys) umiynu umnezoueg 240/ pop addjoAuDy `ç 3l8V 224 Annals of the Missouri Botanical Garden ATE A Uu | II [ul Thy (^u (til ms D Hoan IINE: RE 5. Idiograms of the somatic chromosome Pub nds of Pancratium hirtum morphotypes. types. Further deviation is the presence of B-chromosomes in E One way of generating genetic variation is by changes in chromosomal morphology, which are reflected in the karyomorphology of the population or the species (Coates, 1979; Coates & James, 1979). Another way is by genic changes that may not be immediately detectable (see Linhart et al., 1981). Thus, while genic changes may not be as immedi- ately detectable as changes in the karyotype, karyotype variation may not be immediately accompanied by morphological divergence. Karyotype variation in P. hirtum is as- sociated with population differentiation and ecological preference. Dickinson & Antono- vics (1973) maintained that karyotypic dif- ferentiation is a direct response of various biotypes to differing habitat pressures. White (1973) opined that chromosome rearrange- ments underlie reproductive isolation and, hence, speciation. It is probable that the an- cestral population of P. hirtum in West Africa was at some time afflicted by some drastic environmental events that left survivors whose genetic systems suffered some changes. Such survivors occupied different ecological niches to which each had adapted for continued sur- vival. Each has thus become ecologically iso- lated. Such a situation would be reinforced further by environmental barriers to long- distance pollen dispersal. The smallness of the population that would initially inhabit each ecological niche would enhance both chro- ines evolution and speciation, as asserted by Wright (1940), Bush et al. (1977), and Bengtsson (1980). These events would lead to reproductive isolation and thereby lay the foundation for further divergence of the pop- ulation, culminating, in time, in the formation of several species. Again, the effects of the environment, especially edaphic factors, may e the major driving force in the karyotypic differentiation in this species. For instance, Morton's (1965) pl. 14, fig. 1 is similar to the karyotypes reported here, but it is not identical with any of them. All of Morton's materials were collected from Ghana. It is therefore possible that other karyotypes may still be encountered within the tropical West African region. In conclusion, karyotype differentiation in P. hirtum has involved: (i) changes in chro- mosomal morphology resulting from loss or gain of chromosomal segments, which might have been accompanied by changes involving rearrangements of genes and/or gene blocks in inversions and translocations; and (ii) vari- ations in the length of total chromatin material per somatic nucleus. Therefore, it seems like- ly that ecotype differentiation at the mor- phological level appears genetically fixed, and this is accompanied by varying degrees of karyotypic change, which presumably origi- nated once the ecotypes had become estab- lished. LITERATURE CITED ADHIKARY, A. K. Precise a baie of cen- tromere location. Cytologia 39: 11-16. Volume 75, Number 1 Oyewole 225 1988 Karyotype Variation in Pancratium hirtum BENGTSSON, B. O. Levan, A. K., K. FREDGAR & A. A. SANDBERG. 1964. 1980. Rates of karyotype evolution in placental mammals. Hereditas 92: 37- E Bush, G. L., S. M. Case, A. C. WirsoN & J. L. PATTON. 1977. Rapid piece and chromosomal pes in mammals. Proc. Natl. Acad. U.S.A. 74: 3942- 46. Coates, D. J. 1979. Karyotype analysis in nie aba A d ga Chromosoma (Berl.) 72: 347-356. & S. H. James. pis saqma variation in Sty lidi um crossocephalum an ynamic co- apace of its lethal system. endo (Berl.) 72: 357- DARLINGTON, C, D. & S. F. LaCour. 1969. The Han- dling of Chromosomes, 6th edition. George Allen & nwin, London. Dickinson, H. € J. ANToNOVICs. 1973. Theoretical considerations of sympatric divergence. Amer. Nat uralist 107: 256-274 LAKSHMI, N. tatami studies in eight genera of Amaryllidaceae. Cytologia 45: 663-673 dao cu for centromeric position on chromo es. Hereditas veh ma gy Lissa, i : N, K. B. STURGEON & M. L. Dav 981. Gen etic a n in space and time ina of ponderosa pine. Heredity 46: 407- Morton, J. K. 1965. The experimental taxonomy of the West African species of a i um L. (Ama ryllidaceae). Kew Bull. 19: 337- PoNNAMMA, M. G. 1978. Studies on ^ SN ornamen- tals I. Karyomorphology of diploid and triploid taxa í uem triflorum Roxb. Cytologia 43: 717- TA d J.D. 1973. The Chromosomes, 6th edition. Chapman & Hall, London. WRIGHT, S. 1940. Breeding structure of populations in relation to speciation. Amer. Naturalist 74: 232- 248 THE ARCHITECTURE OF INFLORESCENCES IN THE MYRTALES! Focko Weberling? In Memory of Dorothea Weberling, April 25, 1928- February 2, 1988 ABSTRACT In the Myrtales a hae saa d fores enois es are found. In some families such as Oliniaceae or Alzateaceae h (formerly included in in majority of the genera polytelic inflorescences are jJoruy 8 Poty type eas found. These ofien show a botrytic ramification, although the basic type of ramification is a A abs Among the in the genus Gin nore proliferation is not rar r branches are frequently occurring an the o of Myrtaceae in which th iwed, Aud gno The ae sos of the pd olas S Donee can be delayed for such a long i e that the MS ón shoot may even form branches above oe o se rare and only a single case is reported for Medinilla magnifica. On the peri ae FUNDAMENTAL FEATURES OF INFLORESCENCE MORPHOLOGY Since differences and conformities in the arrangement of flowers are characteristic for smaller or larger taxonomic groups, these cri- teria have been used in many ways since the very beginning of systematic botany. For this purpose an elaborate descriptive terminology is in use designating the different modes of foliation and especially of ramification by well- known terms such as raceme, spike, umbel, panicula, etc. Many efforts have been made to establish a natural system for the immense diversity of inflorescences. The results, how- ever, remained insufficient, primarily because the empirical basis was too small. Above all, it does not follow that the flower-bearing parts HM in part by grants from the U.S. National Science Foundation to P. H. Raven, most recently BSR- 85 18 FDR (West German "arce di Ulm, Abt. für Spezielle Botanik (Biologie V), Postfach 4066, Oberer Eselsberg, D-7900 Ulm, y). ANN. Missour! Bor. GARD. 75: 226-310. 1988. Volume 75, Number 1 1988 Weberling Inflorescences in the Myrtales ° 66 to which a term such as “spike,” “umbel,” or “dichasium” is applied are necessarily ho- mologous. For that reason, one cannot achieve correct interpretation of the morphology of inflorescences by focusing exclusively on those flower-bearing ramification systems that by some conspicuous quality appear to be “units.” One must also consider the position of these entities within the structural plan of the whole plant. Only in this way it is possible to as- certain which flower-bearing elements may legitimately be compared as identical struc- tures. This, however, is connected with the elucidation of the structural plans of flowering plants and therefore needs a broad empirical basis. According to Troll (in Troll & Weber, 1955; Troll, 1964, 1969), the great diversity of inflorescences is due to the variation of two types only: the polytelic and the monotelic types. In the monotelic inflorescence (Fig. 1 I), the apex of the inflorescence axis commonly ends with a terminal flower. This also applies to all the floral branches below the terminal flower. All of these branches, whether branched or not, are homologous elements, and they are all referred to by the term par- acladia (pc; singular: paracladium), because these branches repeat the structure of the main axis of the flowering system. Accord- ingly, their ramifications are called paracladia of the second to n" order. The whole area that produces flowering paracladia (Bereiche- rungstriebe, **enriching branches") may be designated as the enrichment zone (Bereiche- rungszone, “paracladial zone”). In the lower part of the flowering shoot this zone is com- monly preceded by an inhibition zone, within which the development of paracladia is in- hibited more or less abruptly (Figs. 1, 3 I). The same zonation can be recognized in the individual paracladia if these are not reduced in any way. In perennials the axillary buds at the base of the whole stem do not develop within the same season but will give rise to innovation shoots at the beginning of the fol- lowing season. Therefore, this area has to be distinguished as an innovation zone. In the polytelic type of inflorescence (Fig. 1 II), which is no less frequent in angiosperms, there is no terminal flower at the summit of the primary axis. The shoot apex remains indefinite after developing a smaller or great- er number of lateral flowers, the last of which often do not complete their development but atrophy in the same way as the end of the axis. This apical flowering system, which is composed of lateral flowers only (or cymes, see below), is a characteristic feature of this type of inflorescence and is now referred to by the special term florescence. The term "florescence" should not be confused with the term “inflorescence,” which has no spec- ified morphological signification and may be used to designate any flower-bearing ramifi- cation system. (The same applies to the term "partial inflorescence," which can designate without any morphological relevance any part of a flowering system, while the term “‘partial florescence" means a distinct part of a flo- rescence, namely a cymose branch.) Instead of ending in a single flower, as in the monotelic inflorescence, the floral axis here terminates in a multiflowered so-called polytelic flores- cence If the lateral flowers composing the flo- rescences are provided with prophylls (Fig. 2 I), these may produce secondary flowers or dichasial or monochasial flowering systems from their axils (Fig. 2 II). This mode of ramification, in which the production of branches is restricted to the axils of prophylls of consecutive order, is called cymose (see Schimper, 1835; Wydler, 1851a: 305 ff.; Eichler, 1875: 34 ff., but not in the strict sense; Troll, 1957: 234 ff.; 1964: 63). The diverse sympodial ramification systems re- sulting from this mode of branching (Fig. 2 III- VI) are often briefly called cymes (cymae, see p. 231). In such cases the florescence consists of cymose partial florescences (pf) as, for example, in the inflorescences of most Scrophulariaceae and Labiatae (Fig. 3 III). Within both families, the derivation given here is verified by many transitional forms. Below the florescence terminating the main axis there may be some branches that repeat the structure of the main stem by producing 228 Annals of the Missouri Botanical Garden FIGURE racladia of sec ond and thir “Gru ^». pe’, pe”, bi, basal ed ( I Diagrams ofa (1) and (Il) polytelic inflorescence. T, terminal flower; pc, paracladium; 0 c ix re zone”); inh.z. inhibition zone Y. “Hemmungszone”). From Troll (in Troll & Weber, 1955), slightly modified florescences themselves and that therefore are also called paracladia. Their florescences are termed co-florescences in order to distin- guish them from the main florescence of the main axis. The apices of the co-florescences remain indefinite like those of the main flo- rescence. Thus the whole flower-bearing ram- ification system appears as a system of flo- rescences: a (polytelic) synflorescence. In the case of a monotelic inflorescence, the synflo- rescence consists of a terminal flower and (monotelic) paracladia. ithin the polytelic synflorescences, the same zonation can be observed as in monotelic flowering systems (Fig. 3 II, III): an enricA- ment zone (Bereicherungszone: Troll), which precedes the main florescence, an inhibition zone (Hemmungszone: Troll), and in peren- nials, an innovation zone. The three zones together form the so-called hypotagma. The florescence is separated from the paracladial zone by a basal internode (Grundinterno- dium: Troll), which may be of remarkable length. It also proved to be useful to designate the ultimate internode preceding the terminal flower of a monotelic system by a special term: final internode (Grundinternodium, Fig. 1 I). The polytelic type probably has been de- rived repeatedly from the monotelic during the evolution of angiosperms by reduction of the terminal flower and specialization of the paracladia of the monotelic system into either single lateral flowers or lateral cymes, which then constitute elements of the florescences, whereas the others are differentiated as par- Volume 75, Number 1 Weberling 229 1988 Inflorescences in the Myrtales e (gà | | A 1 ` — V — VI Ficure 2. I, II. Vertical diagrams of polytelic inflorescences. —I. Heterothetic np sap —II. Diplo-thyrse; this inflorescence can be de rimary flower; S, seco III-VI from Troll (1959), modified. acladia (of the polytelic type), which them- selves form florescences (Weberling, 1961, 1965, 1983a, b; Troll & Weberling, 1966). In both monotelic and polytelic types, the different elements may vary in many different quantitative respects according to the prin- ciple of variable proportions: in polytelic i florescences, the main florescence may extended and the number of its flowers may be increased in many ways, or it may be reduced or even be missing altogether (trun- cate polytelic synflorescences). In both cases the paracladia may be well developed or re- duced, their number may be increased, or they may be missing or modified in different ways. The development of paracladia may be basitonic or acrotonic (see below). Variation also exists in the diversity of phyllotaxis and duced from the diplobotrys in I by are fertile. II-VI. Vertical (III) and transverse (IV) diagrams illustrating d v d Ill, IV. Dichasium.—V, VI. Monochasia.—V. Scorpioid cyme (cin ndary flower; ph, pherophyll (“Tragblatt,” “Deckblatt” de pr. posu hyp, pa, the assumption that the ae a order cinnus). elicoid bine (bos foliation, shortening or lengthening of the in- ternodes in different parts of the plant, dif- ferent intensity and different modes of ram- ification, and so on. Especially in woody plants the zonation of flowering systems can be al- tered extremely by, for example, the complete reduction of the inhibition zone. An essential difference between the types seems to be that in inflorescences of the poly- telic type the shoot apex of the inflorescence axis remains indeterminate. This also occurs, however, in some monotelic inflorescences in which the terminal flower aborts. In such truncate monotelic synflorescences, how- ever, the paracladia usually end in terminal flowers, thus demonstrating the monotelic character of the whole system. n many lianas, the growth of the main 230 Annals of the Missouri Botanical Garden pz inh.z 1 Iinn.z E 3. I Diagram of a monotelic inflorescence in form of a panicle. T, terminal flower; Pc, Pc’, Pc", IGU Pc", paracladia E. to fourth order; EJ, fi (enrichment zone); jaa inh.z., inhibition zone; ed E the thyrses (III). GJ, bi, Sur internode; PF, partial wee the rest as in axis is indefinite. Therefore the main axis, though it may produce lateral flower-bearing branches, never ends in a terminal flower. The sample applies to rosette geophytes with indefinite main axes, as in Plantago or Phyl- lactis (Valerianaceae). In the latter, the apex of the rosette changes periodically from the formation of absolutely sterile zones to the formation of fertile regions, in which thyrsoid "partial inflorescences” originate from the axils of the rosette leaves. (The same applies to the so-called proliferating inflorescences of many Myrtaceae.) In cases like these, the monotelic or polytelic character of the inflo- rescence is revealed by the monotelic or poly- telic character of the paracladia. us the question of whether the terms monotelic and polytelic correspond to the old classification of inflorescences into two groups called “indeterminate” (indefinite, racemose) and ““determinate” (definite, cymose) must be answered in the negative. Apart from this answer, the statement given by Rickett (1955: 419), that “current usage" of this classifi- cation, “at least in English, is both confused nal mn rta BZ, pz, paracladial zone .z., innovation zone.—ll, III. Diagrams of polytelic ain florescence (HF, mf) ide. the ia (cof) in form of botrya (H) or of pe, pe”, paracladium of first and second order; HZ”, inhibition zone of pc; nl 9° and inaccurate," still refers to this classifi- cation in general. To a great extent this is due to the fact that the “cymose type" is often equated with an “overtopping” of the (somewhat sympodial) branching system (Rickett, 1955: 426; Goebel, 1931: 81 ff.). However, it is neither necessary nor pos- sible to renounce all the classical descriptive terms used for the description and distinction of flower-bearing ramification systems. Many of these terms can be used without any change, and some need to be clarified and specified in their application. This also has been done by Troll (1964: 33 f.), who partly referred to Eichler (1875: 34 f.) in his classification of descriptive terms, which may be repeated here: — . Simple inflorescences: botrys (raceme), spike (stachys), spadix, umbella, capitu- II. Complex (compound) inflorescences a. with racemose partial inflorescences: diplobotrys (and related forms: bispica, bi-umbella, etc.), panicula. Volume 75, Number 1 1988 Weberling Inflorescences in the Myrtales b. with cymose partial inflorescences: thyrse (including thyrsoid), cymoid. Among the compound inflorescences the panicle is distinguished by the main axis end- ing in a terminal flower (as do all of the branches). Thus the panicle is a determinate inflorescence, and this term never refers to any indeterminate inflorescence, as it often is used by English language authors. If all branches of the panicle are reduced to single flowers (uniflorous paracladia), a bot- ryslike system results that, however, still ends in a terminal flower and therefore is termed a “botryoid”; if the flowers are sessile, a spi- coid. Complete reduction of all lateral flowers (all paracladia) leads to a uniflorous system (single terminal flower). The term thyrse, which was often confused with the term panicle (Celakovsky, 1893: 45), needs some further comments too. (There is, of course, some connection between panicles and thyrsoids, as seen, e.g., in the so-called thyrso-paniculate systems of Sambucus and Viburnum: Troll & Weberling, 1966.) The term, used by Linnaeus for a *'coarctate pan- icle of ovate form" (Rickett, 1955: 443), was specified by De Candolle (1827: 417) as "compound of small cymes along an axis of indefinite growth” (Rickett, 1955: 443; Bra- vais & Bravais, 1837: 197: "groupe de cimes disposées d'aprés l'évolution centripéte comme les fleurs le sont dans l'épi"). It was applied, however, to determinate thyrsic branching systems as well (see Troll, 1964: 63 f.). Briggs & Johnson (1979: 177, 247) re- stricted the term thyrse to a “‘blastotelic in- florescence with a multinodate main axis that bears lateral cymes"; thus a determinate thyrselike ramification system should be termed a thyrsoid (which appears consequent, as this term is comparable with terms such as botryoid, spicoid, cymoid in the terminol- ogy of Troll). Although we fear that a change of terminology used hitherto in so many pub- lications might cause new confusion, we hes- itantly follow this suggestion of Briggs & Johnson. Consequently, and in accordance with these authors, the adjective “‘thyrsoid”’ must be replaced by “‘thyrsoidal’’ (thyrsusar- tig), while the adjective thyrsic (thyrsisch) should be used now for a thyrse in the re- stricted sense. The terms used here are illustrated by Fig- ure 2 II-VI. The vertical diagram in Figure 2 II shows a diplo-thyrsic inflorescence. The distal part of it, comprising three pairs of lateral cymes, forms a simple thyrse (mon- othyrsus, haplothyrsus), whereas the branch- es originating from the two proximal nodes are thyrses themselves (named “Spezialthyr- sen” by Troll, “special thyrses,”’ infrathyrse). In this diagram, the lateral cymes are triadic, i.e., comprising three flowers only. They may, however, produce more flowers by continuous branching from the consecutive axes (Fig. 2 III). The characteristic trait of this mode of ramification is that each axis before ending in a terminal flower produces two and only two leaves: the prophylls, which are mostly transversal-opposite and identical with the prophylls (the first two leaves) of other branches. In monocotyledons there is fre- quently only one prophyll, often in adaxial position. The internode preceding the pro- phyllar node is the hypopodium, the inter- node following the prophyllar node and ter- minating in the flower is the epipodium. If the two prophylls are separated by another internode, this is called the mesopodium. From the axils of the prophylls of the primary flower arise branches that end in the secondary flow- ers and bear two prophylls of second order. From the axils of the prophylls of second order the ramification can continue in the same way. This mode of cymose ramification can result in a complete symmetrical dichasium as shown in Figure 2 IV. In many cases, however, the continuation of the ramification is limited at a certain stage to one of the two prophyllar axils, thus resulting in the for- mation of a monochasium, either a scorpioid cyme (cincinnus, Fig. 2 V) or a helicoid cyme (bostryx, Fig. 2 VI). In the monocotyledons a rhipidium or a drepanium can be formed. In thyrsoidal inflorescences, as displayed in a great variety of forms within the Cary- ophyllaceae for example, a high degree of 232 Annals of the Missouri Botanical Garden diversity can result from differences between paracladia in vigor of development. In man cases the paracladia are more extensively de- veloped toward the base of the plant (basi- tonic ramification) or, if the paracladial zone is preceded by an inhibition zone, toward the middle of the plant (basi-mesotonic; in me- sotonic ramification, vigor of the paracladia increases from both ends to the middle part of the flowering system), whereas in others the distal paracladia are the most vigorous (acrotonic ramification). Examples for these models can be found within the genus Silene. In thyrsoids with decussate foliation and ramification (as in Silene), extreme acrotonic branching results in the exclusive develop- ment of the uppermost pair of cymose par- acladia, which far overtop the main axis by vigorous development and copious branching. Such dichasia as represented, e.g., by Silene vulgaris (Moench) Garcke often have been regarded as ideal “cymose inflorescences.” They are, however, connected by many tran- sitional forms with basi-mesotonic thyrsoids as represented, e.g., by Silene nutans L. In other plants transitions can even appear in the same individual, if the subdistal pair of paracladia develops at a later stage. Thus these "dichasial" inflorescences do not represent an inflorescence type of their own but must be regarded as thyrsoids with acrotonic ram- ification. The complete inflorescence only simulates a cymose ramification, which ac- tually takes place in the cymose paracladia. This induced Troll (1959a: 115; 1964: 102 to name this extreme form of an acrotonic thyrsoid a cymoid. In similar fashion, mono- chasial or pleiochasial overtopping of the main axis can be included in this term. As is well known, foliation and phyllotaxis often change in the transition from the veg- etative to the flower-bearing parts of a plant. The foliage leaves may be diminished from leaflike frondose to smaller frondulose organs or even convert by change of diverse pro- portions into bracteose organs (bracts), which are often more or less scalelike. The sub- tending leaves of the flower-bearing branches, the pherophylls (Briggs & Johnson, 1979: A 179, 246; “Tragblatt,”” “Deckblatt””), may thus be frondose, frondulose, or bracteose. The mode of efflorescence can, but need not be characteristic for the diverse forms of inflorescences. One should expect that an- thesis of flowers follows the way of their ini- tiation, advancing from, for example, the old- est flowers at the base of a botryoid to the top of this flowering system. In monotelic synflorescences, however, the terminal flower usually blooms before the neighboring later- als. To some extent this dominant position results from the fact that the organs of the terminal flower arise from the inflorescence apex immediately, whereas the lateral flowers are formed by lateral apices. Thus the ter- minal flower is somewhat in advance in re- lation to the ultimate lateral flowers. It is not rare for the flowers or partial inflorescences in the axils of the ultimate leaves preceding the terminal flower to be retarded or even to abort. Thus the terminal flower can be pre- ceded by sterile leaves that are called “Zwi- schenblatter” (Nordhagen, 1937: 12; Troll, 1964: 15) or metaxyphylls (Briggs & John- son, 1979: 179, 244). Beyond that, cases of a complete basipetal efflorescence—advanc- ing from the top to the base of a flowering system—are not rare (Meconopsis). In the florescences of polytelic inflorescences, the efflorescence usually advances from base to apex (acropetal). For thyrses this means, of course, that an acropetal sequence of primary flowers that open first is followed and over- lapped by a secondary sequence of the sec- ondary flowers of the cymose partial flores- cences (Troll, 1957: 380 f.). There are also florescences with divergent effloration, the best known example may be Dipsacus. However, the sequence of paracladia in polytelic sys- tems mostly unfolds in a basipetal order. Starting with the paracladia immediately be- low the main florescence, the progression of unfolding usually depends on the vigor of the individual and the length of the flowering pe- riod, thus determining the extension of the paracladial zone and the inhibition zone (Troll, 1950). Nevertheless, in annual or hapax- anthic plants this progression can reach the Volume 75, Number 1 1988 Weberling 233 Inflorescences in the Myrtales base of the plant, and the most basal para- cladia can, if they are fully developed, be the most vigorous. ONAGRACEAE Polytelic structure of inflorescences is manifested throughout the Onagraceae. The florescences are botrya or spikes. Even when the pedicels remain undeveloped, the inflo- rescences frequently (e.g., Oenothera biennis L., Fig. 4) look like botrya, because pedicels are simulated by the long gynoecium of the epigynous flowers. In some taxa, e.g., Epi- lobium alsinoides Cunn. subsp. tenuipes (Hook. f.) Raven & Engelhorn, the pedicels lengthen after anthesis. Plants in the flowering state, as shown for the case of Epilobium angustifolium (Fig. 3 II), usually show a clear zonation: below the main florescences there is a more or less extended enrichment zone (paracladial zone, pz) which stretches down the primary axis until the development of paracladia is inhibited. Transition between this "inhibition zone" (inh.z., Hemmungszone) and the paracladial zone may be more or less abrupt. In perennials the axillary buds at the base of the whole stem do not develop within the same season but will give rise to innovation shoots at the beginning of the following season (innovation zone," inn.z., Innovationszone). Diversity in appearance of the different genera and species depends highly on quan- titative variations of these zones, different lengths of the internodia, and different folia- tion of the florescences. The phyllotaxis can be alternate or decus- sate, sometimes even verticillate mostly with three leaves at the same node. It is not rare to find that in the transition from the hypo- tagma to the main florescence the position of the leaves changes from decussate to alter- nate (e.g., Epilobium montanum L.). he subtending leaves of the flowers can be quite different within the same genus. They can be bracteose as in Fuchsia arborescens Sims (Fig. 15 II) or foliaceous as in Fuchsia magellanica Lam. (Fig. 14), giving the in- florescence a “naked” or “leafy” shape. In many species of Epilobium or, for example, Onagraceae. Oenothera biennis. — Top, center. The same plant with the iii cut Ke show the frondo-bracteose foliation. — Top, With developing paracladia. Bottom, left. Plant with fully developed paracladia right. Proliferation apex of the main florescence. (Original photographs from Troll.) FIGUR left. Flowering plant.—Top, c in Oenothera biennis (Fig. 4), they are leafy in the basal part of the florescences, dimin- ishing distally, and finally becoming bracteose (frondo-bracteose foliation). Only in Circaea are the subtending leaves of the flowers miss- ing. With the exception of the genus Lud- wigia, the flowers do not bear prophylls. Many representatives of the family are half- rosette plants, which means that the primary 234 Annals of the Missouri Botanical Garden FIGURE 5. O. stricta. —//I. O. fruticosa. Onagraceae.—I. Oenothera rosea. —/I. axis starts with the formation of a leafy rosette and continues its development with elongated internodes. In Oenothera biennis, O. muri- cata. L., and some other species, the rosette leaves that develop during the first year of the biennial life cycle do not persist until the plant is flowering. This plant in full devel. opment shows a large main florescence, a preceding paracladial zone, and a long zone of inhibition. This basi-mesotonic support of ramification is shown for O. biennis in Figure 4. In vigorous plants of O. biennis the par- acladia can develop strongly and in great number (Fig. 4). In species forming typical half-rosette plants, such as O. indecora Cam- bess., O. rosea L'Hér., and O. stricta Ledeb., the development of vigorous “rosette shoots" (Rosettentriebe) can be observed (Fig. 5). They form a second series of paracladia separated from the upper paracladia by the zone of inhibition. sually the extension and the stoutness of the p.z. depends on the vigor and age of the plants. In the annual Camissonia scapioidea (Torrey & A. Gray) Raven and C. walkeri FIGURE 6. Onagroceae. Oenothera laciniata, show ing a small main florescence, a short inhibition zone, and vigorous rosette shoots. (USA. Texas: Brazos Co. N. Bean 331 (M, as O. sinuata).) (A. Nelson) Raven subsp. tortilis (Jepson) Ra- ven, the paracladial zone can be extended over the whole elongated part of the main shoot, leaving no residuum of the zone of inhibition between the upper paracladia and the rosette shoots (““simple or branching from the base . . . the stems quite simple above," Abrams, 1951: 206). These plants show a pronounced basitonic support in the devel- opment of the paracladia and a scapelike pro- longed internode that precedes the main flo- rescence. Especially within the Oenothera alliance (in tribe Onagreae), the proportions of the main florescence, the paracladial zone, and the inhibition zone and the development of the internodes can be modified in different ways and can be very characteristic for some taxa. Sometimes, in Camissonia cheiranthi- folia (Hornem. ex Sprengel) Raven, for ex- ample, the development of paracladia can be restricted to the basal rosette. The reason is often that this rosette comprises the whole hypotagma while the main florescence is the only part of the axis with elongated inter- nodes. The early development of these basal paracladia is shown for O. fruticosa L. in Figure 5. In some species, for example, Oenothera laciniata (L.) Hill (Fig. 6), Camissonia hir- tella (Greene) Raven, and Calylophus hart- wegii (Benth.) Raven, even the main flores- cence often or always remains relatively short Volume 75, Number 1 1988 Weberlin g 235 Inflorescences in the Myrtales and few flowered. In O. caespitosa Nutt., Camissonia palmeri (S. Watson) Raven, and many other species, the internodes of the primary axis, including the main florescence, are undeveloped throughout (**Caespitose pe- rennial, acaulescent or nearly so," Abrams, 1951: 195). Among these species, O. triloba Nutt. (O. rhizocarpa Sprengel, Lavauxia tri- loba Spach), a hapaxanthic taprooted plant, is of special interest because of its basicarpous hygrochastic capsular fruits (Sernander, 1927: 73 ff.). In such cases the paracladia, however, can be well developed, forming pla- giotropic (Figs. 7, 8) or ascendent branches with frondose (co-)florescences. In other rosette plants, for example, Ca- missonia graciliflora (Hook. & Arn.) Raven (Fig. 15 III), even the paracladia, if devel- oped, remain short, the whole plant forming a single rosette (also Oenothera ovata Nutt., O. xylocarpa Cav., O. nana Griseb., Figure 9, C. tanacetifolia (Torrey & A. Gray) Ra- ven, Figure 10, C. subacaulis (Pursh) Raven, and O. formosa Brandegee), or a cespitose tuft. In contrast to Oenothera triloba, which always retains its rosette stage, O. macro- carpa Nutt. subsp. macrocarpa (O. mis- souriensis Sims) is a perennial tap-rooted geo- phyte. After the juvenile rosette stage, it grows with slightly prolonged internodes and can develop innovation shoots, which can inno- vate secondarily. The same could possibly apply to O. californica (S. Watson) S. Wat- son, which Abrams (1951: 195) described as “perennial from underground rootstalks, rather coarse stemmed, rarely simple." In the fruiting plant of Oenothera triloba (Fig. 8), the top of the primary axis is crowned by a bunch of foliage leaves as a result of the proliferating growth of the florescence axis. This proliferation may occur repeatedly in the same primary axis, thus forming zones with buds of paracladia alternating with single flowers in the axils of more or less foliaceous leaves. Proliferation can be observed in many additional species of Oenothera, for instance, O. macrocarpa and O. biennis (Fig. 4), and in many other members of the family, since FiGURE 7. Texas: Tarrant Co., Fort Worth, A. Ruth 70 (M).) Onagraceae. Oenothera triloba. (USA. the predisposition if frequently given by a strong florescence axis with more or less leafy pherophylls. In O. triloba, O. macrocarpa, and O. bien- nis, it is the apex of the main florescence that returns to vegetative growth (Spatprolifika- tion, Troll, 1960: 116, prolification of the florescence, retarded proliferation), whereas O. multicaulis Cav. (Figs. 11, 12) does not form a main florescence at all. Rather, it develops long paracladia from the axils of rosette leaves (Frühprolifikation), which can grow straight upright (Fig. 11) or more or less ascendent (Fig. 12). In most herbaceous genera of Onagraceae the same architecture and zonation as de- scribed for the basi-mesotonic ramified Epi- lobium angustifolium (Fig. 3 II) and for Oe- nothera biennis (Fig. 4) can be observed. This applies to many species of Epilobium and all or most species of Boisduvalia, Clarkia (ex- cept species with decumbent or prostrate stems), Gaura and allied genera, Lopezia, and Circaea, although some species of Bois- duvalia, e.g., B. densiflora (Lindley) S. Wat- son (Troll, posthumous notices), can be branched from the base. Plagiotropic growth, often connected with proliferation, is also characteristic for Clarkia 236 Annals of the Missouri Botanical Garden FIGURE 8. Onagraceae. Oenothera triloba.— Left. Fruiting plant with proliferating main axis. — Top, right. Fruiting plant with terminal vegetative rosette and stolonlike fruiting and proliferating paracladia. — Bottom, / right. Detail. (Original photographs from Troll. prostrata Lewis & Lewis, C. davyi (Jeps.) Lewis & Lewis (Lewis & Lewis, 1955, fig. llb, c), Camissonia strigulosa (Fischer & Meyer) Raven, and many species of Ludwig- ia. Plants growing under severe environmen- tal stresses may develop reduced inflores- cences only. This is especially common in annuals. Often only the main florescence is formed, and in extreme but not uncommon cases only the lowermost flower of the main florescence comes to anthesis. Clarkia bottae (Spach) Lewis & Lewis (Fig. 13 I) is an ex- ample. In such cases this lowermost flower can be erected in pseudoterminal position, which also is characteristic for Epilobium al- pinum L. f. pusillum Hausskn. The same occurs facultatively in E. montanum (Fig. 13 ID. For the genus Epilobium two features must be mentioned, each of them characteristic for a great number of species. In many species the buds of the innovation zone develop more or less proleptically into epigeal or subterra- nean stolons. If these stolons creep at the surface of the soil, their leaves are more or less foliaceous, as in E. alpinum L. The sub- terranean stolons bear scalelike cataphylls. Volume 75, Number 1 1988 Weberlin ling 237 Inflorescences in the Myrtales FIGURE 9. Onagraceae. Oenothera nana. All inter- nodes of the primary axis, including the main flores- cence, remain short; this also applies to the paracladia (pc). (Peru. Puno: above Chimu, K. A. Santarius 2056/ 364 (M).) 10. Onagraceae. Camissonia tanacetifolia. (USA. Washington. Spokane Co., July 1884, Suksdorf (M).) s.n. FIGUR Onagraceae. Oenothera multicaulis. rel wth o ERN leaved proliferating main shoot an flowe acladia. (Peru: between Huancayo and n Miller 2345 /75 (M).) Their internodes may be very slender (£. lin- eare Muhlenb., E. obscurum Schreber) and often form small bulbs at their ends, especially well developed in E. palustre L. (Irmisch, 1847; Warming, 1918: 348; Troll, 1937: 811, fig. 639). In others the stolons are some- what thickened, forming storage organs in this way (E. hirsutum L., Fig. 14). There are also many transitional forms between subterra- nean and epigeal stolons, which then develop foliaceous leaves in place of scales (E. hir- sutum), and between stolon-forming innova- tion buds and those that remain short (the latter in E. montanum, E. collinum Gmelin, E. roseum Schreber, transitional in E. par- viflorum Schreber, and E. lanceolatum Seb. & Mauri). In E. palustre there occur also transitional ascendent shoots with foliaceous leaves coming to flower in the same year as the main shoot and thus forming proleptical- 238 Annals of the Missouri Botanical Garden SURE 12. Onagraceae. Oenothera multicaulis. Plant with proliferating main shoot with rosetted leaves and ascendent paracladia with co-florescences. (Peru: Huancayo. Santarius 2204/717 (M).) FIGURE 14. Onagraceae. —Lefi. Epilobium iip tum; plant with vigorous subterranean stolons. —R i Fuchsia magellanica; inflorescence. (Original ue graphs from Troll.) anthetical (see Müller-Doblies & Weberling, 1984) innovation shoots, while normally it takes until the next season for the stolonlike or short remaining innovation shoots to de- velop into a new flowering system. (For dis- cussion of different modes of perennation in Epilobium, see Keating et al., 1982.) A second peculiarity is especially charac- teristic for a number of species belonging to the series Similes, Microphyllae, and Sparsi- florae Hausskn., distributed in Australia, Tas- mania, and New Zealand. These species show a more or less plagiotropic growth correlated with proliferation of the florescence apex, which seems to be a constant feature within the mee series. "All of the leaves are op- posite” (Raven & Raven, 1976: 15), and "the stem creeps and roots at the nodes, and continues growing vegetatively beyond the area where the flowers are produced." Thus the apex of the florescence regularly returns to a vegetative phase in which paracladia are developed and returns again to the production of a certain number of flowers. In Epilobium I nummulariifolium R. Cunn., a node bearing E Onagraceae. One-flowered nscale a pair of flowers may be followed by two nodes of Clarkia bottae (I) and Epilobium montanum L. (11). : . bi, basal internode; Co, scars of the cotyledons. (Orig- with paracladia, but more frequently one of inal drawings from Troll.) the axils of a pair of leaves produces a flower, Volume 75, Number 1 1988 Weberling Inflorescences in the Myrtales 239 Onagraceae. —1. Fuchsia coccinea; node of the florescence with regular and accessorial flowers FIGURE 15. II. F. arborescens; inflorescence with main florescence (mf) and paracladia (pc).—Ill. Camissonia Zracilifora; oll.) habit (after J. W. Hook.). —IV. Epilobium nummulariifolium. (1, H, IV original drawings from Ti while the other one bears a paracladium (Fig. 15 IV). Troll (unpubl. notes) found the pri- mary axis passing over to plagiotropic growth after the unfolding of the cotyledons. The axils of the cotyledons produced paracladia. According to Raven & Raven (1976: 15) this "habit has evolved independently on at least seven occasions among the Australasian species, as indicated by the morphological, genetic, and geographical relationships of the species involved: 1) E. willisii; 2) E. mac- ropus; 3) E. pedunculare; 4) E. crassum; 5) E. margaretiae; 6) E. brunnescens, E. per- nitens, E. nerteroides, and E. nummulari- ifolium; and 7) E. komarovianum and F. angustum. There is presumably an ecological solution as to why this peculiar plant form, unknown elsewhere in the world, should have evolved repeatedly in Australasia. ithin the genus Ludwigia (including Jus- sieua) one can find species that show the typical zonation as described for polytelic 240 Annals of the Missouri Botanical Garden FIGURE 16. cosa. — Center. L. alata. — Right. itata). (Original photographs from Troll. L. octovalvis ( ) plants with botrytic florescences. Ludwigia suffruticosa (L.) Gomez serves as an example (Fig. 16). The main inflorescence is a frondose botrys. The same applies to the paracladia, which are preceded by an inhibition zone. Axoscopic accessorial buds can be observed accompanying the paracladia as well as the single flowers of the florescences. The apex of the inflorescence axis proliferates repeat- edly: after having produced single lateral flow- ers forming the main florescence, it develops a paracladial zone and then returns to produce single flowers again. This zonation found in L. octovalvis is also found in L. alata Ell. (Fig. 16) and L. perennis L. In the latter the pherophylls within the florescences are some- what smaller than in the preceding paracladial zone. Here the paracladia mostly comprise a co-florescence only, since the prophylls bear flowers in their axils already. The axoscopic accessorial buds within the florescence form small flower-bearing branches, while those that accompany the paracladia remain undevel- oped. In Ludwigia virgata Michaux, develop- ment of the paracladia usually is suppressed. The plants are extremely heterophyllous, bearing lanceolate leaves in the proximal part of the hypotagma and linear leaves in the distal part, whereas the foliation of the main florescence consists of small bractlike leaves. Ludwigia octovalvis (Jacq.) Raven resembles Onagraceae.— Left. Ludwigia suffruti- L. cap- FIGURE 17. sette. (Deutsch-Mühlenweiher, Ruppert s.n. (M).) Trapaceae. Trapa natans; flowering ro- Saarbrücken, 1904, L. virgata in the suppression of the paracla- dia, but the flowers of the main florescence of L. octovalis (= L. capitata Michaux) usu- ally are condensed into a head by suppression of internode development (Fig. 16). At the base of the plants, Troll (unpubl. notes) some- times observed stolonlike innovation shoots. On the other hand, we could not find a main florescence in herbarium specimens of Ludwigia abyssinica A. Rich. The speci- mens investigated only had paracladia ending in a botrytic florescence, their foliation being bracteose in the upper paracladia and frondo- bracteose in the lower paracladia. If this pre- liminary result proves to be true, L. abyssin- ica has a truncate polytelic synflorescence. Ludwigia palustris Ell. differs from the preceding species by having plagiotropic growth. The plant, which grows in stagnant or slow-flowing water or on muddy soil, de- velops roots from the nodes and different forms of leaves depending on ecological conditions. In these features L. palustris resembles species such as L. helminthorrhiza (Martius) Hara (Jussieua natans Humb. & Bonpl.) and L. adscendens (L.) Hara (J. repens L.), which are well known for peculiar respiratory roots that develop at the nodes in addition to normal roots. They also show a more or less pro- nounced heterophylly. Here again the inflo- Volume 75, Number 1 1988 Weberling Inflorescences in the Myrtales rescence apex regularly proliferates and re- turns to flower formation. In L. peduncularis (Wright ex Griseb.) Gomez (L. clavellina) a pair of flowers often is followed by a pair of paracladia. In the picture of L. helminthor- rhiza given by Humboldt & Bonpland (1805, t. 3, under Jussieua natans), each flower is accompanied by a vegetative accessorial branch. The most interesting species of Ludwigia is L. sedoides (Kunth) Hara, an aquatic herb in which the branches end in floating rosettes. The rosettes are formed by alternate leaves that gradually differ in the length of their petioles, thus forming a configuration similar to a rosette window. Usually the flowers are described as soli- tary in the axils of the foliage leaves. Actually they all form a botrytic main florescence which proliferates, however, after the formation of only a few flowers, forming a vegetative flow- erless zone and then returns to flower pro- duction. Thus the flower-bearing zones are interrupted by vegetative zones. Paracladia that repeat the architecture of the main shoot seem to develop rarely only from the more remote parts of the stem. In all of these respects Ludwigia sedoides is completely congruent with 7rapa natans L., Trapaceae (Fig. 17). The submerged leaves of Trapa, however, are pinnatifid and thus differ from the floating leaves that have a rhombic lamina and somewhat inflated peti- oles, which keep the rosette floating. In the ligneous (Fuchsia, Hauya) or suf- fruticose Onagraceae, the architecture of the inflorescences does not differ fundamentally from that of the herbaceous genera. In Fuchsia the diversity of inflorescences mainly depends on differences of foliation, development of the internodes, and degree of ramification within the flower-bearing parts. As was shown already, the difference between the bracteose and the leafy character of the pherophylls (Figs. 14, 15 II) has a great effect on the appearance of the plants. In F. ma- gellanica the leafy botrytic florescences pro- liferate frequently and often repeatedly, giv- ing the impression that there are “flowers solitary in the axils of leaves." From the veg- etative zones of the "interrupted flores- cences" paracladia can develop. Accessorial flowers can occur in a phylloscopic sequence (Fig. 15 I). Quite a different impression is given by the botrytic florescences of F. tri- phylla L., F. coccinea Sol., and F. fulgens Mocino & Sessé (see Raimann, 1893b, fig. 94A), the foliation of which is bracteose or frondo-bracteose. In F. boliviana Carr. the pendant botrytic frondo-bracteose main flo- rescence and co-florescences are hanging by the inclination of their prolonged basal inter- nodes, while the flower-bearing zone is more or less condensed (Fig. 18). The bracteose inflorescence terminating the foliaceous shoots of F. arborescens Sims (Fig. 18) is often called a panicle but comprises a botrytic main flo- rescence and a smaller or greater number of paracladia, which are restricted to their co- florescences. The inflorescences of Hauya elegans DC. subsp. cornuta (Hemsley) Breed- love & Raven is a leafy proliferating few- flowered spike. Although the structure of the inflorescences within the woody Onagraceae seems to be clear, the position of the flower- bearing branches within the whole ramifica- tion system should be examined in compar- ative investigations regarding the different growth forms of the ligneous plants. LYTHRACEAE The diverse forms of ramification found in the inflorescences of Lythraceae have been carefully investigated and described by Koehne (1883). Thus we only need to treat some new fundamental and comparative as- pects here. The phyllotaxis is usually decus- sate, rarely alternate; sometimes verticils with three or many (e.g., species of Rotala) leaves are formed. With some exceptions (Cuphea sect. Lythrocuphea and some other species) prophylls are nearly always present within the inflorescences. Eichler (1878b: 478) named as the only exceptions Cuphea subgen. Cu- phea (subgen. Lythrocuphea Koehne) and the secondary flowers within the cymes of Lythrum species. Annals of the Missouri Botanical Garden Ne - i v ALD ° =. `V w y xb - E \ Ñ \ E 18. Onagraceae. Inflorescences of Fuchsia Pena from Troll.) Within the family only the genera Gal- pinia, Lagerstroemia, Lawsonia, Rhyncho- calyx (now treated as Rhynchocalycaceae), and Woodfordia are characterized by mon- otelic inflorescences. The inflorescence of the ligneous Lager- stroemia indica L. (Figs. 19 I, 20 I) is a diplothyrsoid (a diplothyrse with terminal flower) with three- to seven-flowered cymes. A remarkable feature of this inflorescence is that effloration takes place from base to top, though the terminal flower of the whole sys- tem precedes the neighboring lateral ones. The same applies to the terminal flowers of the thyrsoid paracladia in the proximal part of the inflorescence. Correlated with this mode of effloration is the delay in the formation of the ultimate lateral flowers below the terminal flower, which can be more or less impeded (Fig. 20 Ia). In case of complete reduction of the uppermost paracladia their subtending bracts appear as metaxyphylls. arborescens (lefi) and F. boliviana (right). (Original Generally Lagerstroemia tomentosa Presl shows the same mode of ramification as L. indica. In L. calyculata Kurz, L. hypoleuca Kurz, and L. speciosa Pers., however, the distal part of the inflorescence bears a con- siderable number of uniflorous paracladia forming a botryoidlike zone. This also occurs in the thyrsoid paracladia at the base of the inflorescence, which apparently are more nu- merous and vigorously developed in these species. In any case, a careful comparative examination of the inflorescences of the nu- merous species of this genus is needed in order to find out if the peculiarities mentioned here are useful for diagnostic and systematic aims. The mode of effloration found in Lager- stroemia indica can also be observed in Law- sonia inermis L. (Fig. 19 II), which differs from L. indica in that the ramification of the cymes does not exceed the three-flowered stage (Fig. 20 II). On the other hand, the number of flowers of the inflorescence fre- Volume 75, Number 1 1988 Weberling Inflorescences in the Myrtales 243 SN Ny. Ws N NN "J EN x = Tl / NS I II FicuRE 19. Lythraceae. —1. Lagerstroemia indica. —//. Lawsonia inermis; flowering branches. (After Koehne, 893.) quently is increased by accessory phylloscopic branches of first (as,) and second (as,) order. They can form a single flower provided with two prophylls, as in pc, to pc,, or can develop into triads (pc;) and even into thyrsoids (pce). The highest degree of ramification and most vigorous development of accessory branches takes place at the base of the whole inflores- cence. The formation of accessory branches can be repeated within the paracladia (pc,, pc.) and even within the accessory branches themselves (as, in pc,). The diagrams in Fig- ure 20 II-V give additional information about the serial position of the accessory branches. Accessory branches or flowers are very com- mon within the whole family. Lagerstroemia parviflora Presl (Fig. 21 I) has advanced to the differentiation of a 244 Annals of the Missouri Botanical Garden e e TT cw e a s (© 5 (e e @ ee) (e e e) (Se (sees) es) ° TI ° “ie, ° dodo i ndi (e! lad (e) aso pco — — pcs Il IV V Lythraceae. Lagerstroemia indica (1, la) and E: inermis (II-V). Vertical Du up of flowering brand hes (1, II) and horizontal pr ojections zm (HI (V) with ui ud branches as, and as,, the distal part (pc,-pc,) of the inflorescence in Il (IV). la is ul flower of another ene of Lagerstroemia indica with two preceding pairs of nur with undeveloped llas buds. (From posthumous notices of W. Troll, except la.) long shoot-short shoot system in which flower formation is restricted to the short shoots until the next vegetation period. At the mo- ment of their unfolding and flowering the long (brachyblasts), while the long shoots (macro- blasts) continue the vegetative growth. In this case the buds of the short shoots are protected by a series of bud scales and kept undeveloped shoot has lost its leaves already. The flowering short shoot proliferates. Thus the inflores- cence consists of paracladia only, which form botryoids rising from the axils of normally Volume 75, Number 1 1988 Weberlin g 245 Inflorescences in the Myrtales (|, ^ uim: UTA D FIGURE 21. last year's vegetative long shoot with a flowering short shoot. —1l. Lafoensia replicata. (From posthumous notices of W. Troll.) developed or, at the base of the short shoots, very small foliaceous leaves. The basis of the brachyblast that bears bud scales may be in- terpreted as a zone of inhibition or perhaps also as an innovation zone. If Punica (formerly Punicaceae) ought to be included in Lythraceae, forming subfamily Punicoideae, it could be placed close to La- gerstroemia for its monotelic inflorescences. These inflorescences, however, are highly re- duced and frequently limited to their terminal flower. They may develop one or two decus- sate pairs of uniflorous paracladia originating from the axils of bracteose leaves. The uni- florous paracladia bear prophylls that are in- serted immediately below the flower. Most of these lateral flowers degenerate sooner or lat- er. Sonneratia (formerly Sonneratiaceae), which is regarded now to form subfamily Son- « wa m N NN NS" IN NS LO NN , NS » LOS Lythraceae. Vertical projections of flowering branches.—I. Lagerstroemia parviflora; piece of a neratioideae within the Lythraceae, also has monotelic inflorescences: terminal botryoids with two or three decussate pairs of uniflorous paracladia that bear large prophylls imme- diately below the flowers. At least the low- ermost paracladia arise from the axils of fo- liage leaves. Sometimes only the terminal flower develops (S. acida L. f.). The genus Duabanga, which was formerly included in Sonneratiaceae and now is re- garded to represent subfamily Duabangoideae of the Lythraceae, has terminal few-flowered panicles in which ramification only somtimes goes beyond the formation of Pc’. Crypteroniaceae, formerly sometimes joined with Sonneratiaceae, differ fundamen- tally from Sonneratia by polytelic di- to pleio- botrytic inflorescences of paniclelike appear- ance with frondo-bracteose (Axinandra Thwaites, Dactylocladus Oliv.) or mainly 246 Annals of the Missouri Botanical Garden bracteose (Crypteronia Bl.) foliation and many-flowered botrya. In the axils of the low- ermost and foliaceous leaves of the flowering system we could observe well-developed bot- rytic accessory branches. Sometimes we found the terminal flower of the inflorescence of Lagerstroemia spe- ciosa missing, though terminal flowers still were present in all paracladia. One might overestimate this fact, however, if one inter- prets this occurrence of truncate monotelic inflorescences as first indication of a gradual transition to the polytelic type within the fam- ily. In any case polytelic inflorescences occur in most Lythraceae. The inflorescence of Galpinia transvaal- ica N. E. Br. resembles those of Lagerstroe- mia in the terminating leafy branches and in its paniclelike structure: the inflorescence axis and all floriferous branches end in terminal flowers. Its foliation, however, is frondulose to bracteose. In contrast to Galpinia, Lawsonia, and most species of Lagerstroemia, the flower- bearing systems of the shrubby genus Wood- fordia (W. floribunda Salisb., W. uniflora (A. Rich.) Koehne) are brachyblastlike branches originating in the axils of foliaceous leaves that are still present during anthesis. They are thyrses or diplo-thyrses (in the nar- row sense, cf. Fig. 2 II) with frondulose leaves on their main axes. Since they often bear a series of cataphylls at their bases, it seems likely that they do not unfold immediately after initiation but remain as buds for a while, perhaps to endure an unfavorable season. Frequently these brachyblasts are accom- panied by phylloscopic accessory branches of similar shape. n the flowering systems of Lafoensia, another ligneous genus comprising large trees or shrubs, all axes bearing more than one pair of leaves proved to be auxotelic or anauxotelic (see p. 290). The inflorescence of L. replicata Pohl, which may serve here as an example (Fig. 21 II), shows a complete reduction of cymose branching. Thus a terminal and sev- eral pairs of lateral botryslike structures with acropetal effloration are formed. They can be interpreted as the terminal main florescence and the co-florescences of a polytelic system. The same applies to L. acuminata DC., L. pacari St. Hil., L. punicifolia DC., and L. densiflora Pohl. The deciduous prophylls of the flowers are situated immediately at the base of the flowers. In other species, as in L. nummulariifolia St. Hil., the number of flow- ers is diminished. While in the Lafoensia species treated here the subtending leaves within the inflores- cences are more or less foliaceous, Physo- calymma scaberrimum Pohl is characterized by bracteous foliation throughout the whole flower-bearing system that terminates vege- tative branches. This system is a diplo- or pleiobotryum with a large paracladial zone (Fig. 22 I). The axes of all botrya end in a vegetative bud. According to Koehne (1893), flowering of this tree takes place when the plants have lost their leaves. In the shrubby Pemphis madagascariensis (Baker) Koehne, the production of flowers seems to be limited to brachyblastlike branches which, however, proliferate like the macroblastic main axis. The botrytic flower-bearing zones comprise a few flowers only (Fig. 22 II). Pemphis aci- dula Forster, the second species of the genus, is similar. A gradually progressing differentiation of the ramification system into flower-bearing short shoots and vegetative long shoots can be seen within the genus Ginoria Jacq. As reported by Koehne (1883: 115) in G. amer- icana Jacq., the flower-bearing branches arise from the axils of foliaceous leaves. After the appearance of a pair of bud scales, these twigs develop several pairs of foliaceous leaves that bear single flowers in their axils (Fig. 23 I). At a later stage the twigs pass over to veg- etative growth. Ginoria spinosa and G. gla- bra Griseb. are reported to be similar, but the flowering branches are shorter with fewer flowers and do not proliferate regularly. In Ginoria curvispina Koehne, the flower- bearing brachyblasts unfold in the season af- ter their initiation, when their pherophylls are already lost (Fig. 23 IIa). Their foliation be- gins with one to three pairs of cataphylls Volume 75, Number 1 88 Weberling 247 Inflorescences in the Myrtales FiGURE 22. Lythraceae. Vertical projections of flowering branches.—I. Physocalymma scaberrimum.— //. Pemphis madagascariensis. (From posthumous works of W. Troll.) followed by foliaceous leaves. The. whole brachyblast can bear one to four pairs of single flowers that can originate in the axils of the cataphylls as well as in the axils of the foliaceous leaves. Finally the brachyblasts proliferate. According to Koehne (1883: 116), the fo- liaceous leaves of the brachyblasts resemble those of the macroblasts but measure only one-fourth of their size. Thus the brachyblasts with their botrytic main florescences contrast conspicuously with the vegetative macroblasts, which at their dis- tal end can continue growth with the devel- opment of a new annual shoot. According to the observations of Koehne, these new shoots can also bear single flowers in the axils of their lowermost foliaceous leaves (Fig. 23 IIb). Koehne also reported that two similar long shoots can originate at the very base of the annual shoot and also bear up to four pairs of flowers (Fig. 23 IIc). Thus the limitation of flower production to the short shoots seems not to be complete yet. In Ginoria diplusodon Koehne (Fig. 23 III), the brachyblasts remain shorter, their foliation consists of scales, one to three pairs of which can bear flowers in their axils, thus forming a short botrys. (Sometimes the pro- phylls of these flowers bear secondary flow- ers.) Even here the annual shoots continuing the macroblasts (Fig. 23 V) develop one to three pairs of single flowers originating in the axils of cataphylls at the base of the annual shoots (Koehne, 1883). Ginoria rohrii (Vahl) Koehne is similar, but the brachyblasts are shorter and the two- to eight-flowered flo- rescence is umbel-shaped (Figs. 23 IV, 24); 248 Annals of the Missouri Botanical Garden FIGURE 23. Lythraceae. Vertical diagrams of flowering shoots. —1l. Ginoria americana; proliferating brach- st year's macroblast with flowering and proliferating brachyblasts (a), single flowers ateral macrobla ( o Prophylls of the flowers omitted. (Construction according to report by Koehne (1883) and to our results. the prophylls are sterile. The brachyblasts unfold when the macroblast is already defo- liated. The uppermost brachyblasts are re- ported to proliferate occasionally. According to Troll (unpubl.), the terminal bud of the macroblast often fails. Then the macroblast is continued by the development of the up- permost lateral buds. Their foliation begins with two or three pairs of bud scales. The distal pair can be frondulose; according to Koehne (1883: 117), the flowers originating in the axils of the bud scales may bear a second flower in the axil of one of their pro- ylls. Although the florescences of the ligneous Lythraceae such as Ginoria, Lafoensia, Pem- phis, and Physocalymma present themselves as botrya or spikes, the basic form of the florescence in the polytelic Lythraceae is a thyrse (in the narrow sense, i.e., without ter- minal flower). Among the ligneous taxa this is demonstrated by Adenaria: in A. flori- bunda Kunth, the flower-bearing twig is a foliaceous, proliferating thyrse with man flowered cymes, which by abbreviation of all internodes except the epipodia become umbel- shaped or headlike (Fig. 25). Pehria (= Gris- lea), as seen in P. compacta (Rusby) Sprague, is said to be similar in this respect (Koehne, 1893); the same applies to Decodon verti- cillatus (L.) Ell. Within the suffruticose and herbaceous taxa Volume 75, Number 1 1988 Weberlin g 249 Inflorescences in the Myrtales RE 24. Lythraceae. Ginoria rohrii; macroblast GU m two brachyblasts. (June, 1907, Eichinger s.n. ( M).) there are also many groups with copiously branched thyrses. As an example, /Vesaea crassicaulis (Guill. & Perr.) Koehne (Fig. 26 I) may be mentioned. Its partial florescences are loosely branched with only the hypopodia remaining relatively short, whereas in others, for example, N. pedicellata Hiern, even the hypopodia are prolonged and form long pe- duncles. In N. aspera (Guill. & Perr.) Koehne all internodes remain short, and the partial florescences form many-flowered glomerules (see also N. sarcophylla Koehne in Koehne, 1903: 227, fig. 43E) Frequently the partial florescences are re- duced to uniflory, thus converting the thyrse into a botrys or spike. This applies to some species of Nesaea (e.g., N. linifolia Hiern (Fig. 28 III) and N. baumii Koehne), most species of Lythrum, and all species of Pleu- rophora, Cuphea, Heimia, Peplis, and Di- plusodon. In Rotala, only one species, R. serpiculoides Hiern, is reported to have 3- 12 flowered dichasia (Koehne, 1883: 124), while in Ammannia reduction of the cymes to uniflorous elements occurs facultatively in many species A remarkable arrangement of flowers with- in the botrytic florescences can be observed in Rotala wallichii Koehne. It was reported FIGURE 25. Lythraceae. Adenaria floribunda; qe ering branch. (Bolivia: Vic. Sorata, M. Bang 1311 (M).) by Koehne (1883: 118) for the verticillate species of the genus that in each whorl the number of flowers equals the number of leaves, i.e., the axil of each leaf bears a single flower. In R. wallichii (Fig. 26 II), however, the proximal verticils of a florescence comprising six or seven leaves bear only two or three flowers. The number of flowers increases dis- tally, until each leaf axil of a verticil bears a ower. The disposition of the flowers in the lower whorls can be quite different, without any obvious rule. During the development of a shoot the number of leaves in the consec- 250 Annals of the Missouri Botanical Garden Os. Ñ AQ TA ` Z. S > QN WT V IS A 27 77, \ I SS II FIGURE 26. Lythraceae. —1. Nesaea crassicaulis; diagram of a flowering shoot. —1I. Rotala wallichii; flowering branch. (For each verticil of the main axis the number of leaves and of axillary flowers is added.) utive verticils increases gradually from two to four and six, then perhaps up to seven and decreasing to six again. Among the taxa with copiously branched cymose partial florescences, the species of Nesaea sect. Nesaea (= sect. Typonesaea Koehne) subsect. Tolypeuma (E. Meyer) Koehne are remarkable. By shortening of the internodes of all consecutive ramifications the cymes are contracted to small headlike ag- gregations of flowers enveloped by an invo- lucrum that is formed by the large cordate- lanceolate prophylls. The corollae of the flow- ers rise slightly above the border of these prophylls, which often are somewhat whitish and violet-nerved basally. The hypopodia of the capitulate partial florescences are pro- longed. Since the main axis of the thyrsic florescences—the main florescence and the co-florescences—is anauxotelic, the flores- Volume 75, Number 1 1988 Weberling 251 Inflorescences in the Myrtales FIGURE 27. cences of these species, i.e., Nesaea erecta Guill. & Perr. (Fig. 27) or N. cordata Hiern, appear as botrytic systems composed of ca- pitula. In contrast to these species, the inflores- cence of Nesaea linearis Hiern (sect. Ty- Lythraceae. Nesaea erecta.—1. Flowering plant.—II. Cymose partial florescence. (Rhodesia: Mapanza, E. A. Robinson 2839 (M).) ponesaea subsect. Syntolypaea Koehne) is described to form a terminal capitulum com- posed of ““dichasia singula confertissima, ses- silia, bracteolis superne dilatatis . . .” (Koehne, 1903: 223, 230, fig. 45C). Likewise the lateral capitula of /Vesaea 252 Annals of the Missouri Botanical Garden radicans Guill. & Perr. (also subsect. Toly- peuma), which are of similar appearance to those in N. erecta and N. cordata, are de- scribed to have four or six enlarged “‘pro- phylls," each of them subtending a small *'di- chasium" (Koehne, 1903: 8, 231). Indeed, these capitula proved to have a thyrselike ramification and to be provided with a ter- minal flower as Koehne had affirmed already. He did not investigate N. linearis in this re- spect. At least for /V. radicans, however, it seems to be evident that the inflorescences are monotelic, as Koehne (1903: 8) affirmed with the statement that the inflorescences of N. radicans seem to form a transition to those of Woodfordia, Lagerstroemia, and Law- sonia. À remarkable condensation of florescences is typical for Pleurophora polyandra Hook. & Arn. also; here, however, the florescences are spikes (Fig. 28 I, IT). As the treatment of inflorescences in the Lythraceae has shown already, the disposition for proliferation of inflorescences is wide- spread within this family. Proliferation takes place especially in frondose florescences, for example in Heimia, Nesaea crassicaulis (Fig. 26 I), N. sagittifolia Koehne, or Adenaria floribunda (Fig. 25). Another factor responsible for the high de- gree of growth-form diversity of inflores- cences, especially in herbaceous and suffru- ticose taxa, is the variation of the proportions between the different zones of the flowering plant, i.e., in polytelic taxa the main flores- cence, the enrichment zone, the inhibition zone, and, in perennials, the innovation zone. Thus, Troll (1970: 92) reported that in Diplusodon thymifolius DC. the main flo- rescence remains very small in comparison with the paracladial zone. The same seems to apply to D. virgatus Pohl, whereas in other species, e.g., D. villosissimus Pohl and D. villosus Pohl, we found the main florescence to be predominant, although the paracladial zone can be well developed too. The distal paracladia of these species had no hypotagma. Of course, to a certain degree such differ- ences as the unfolding of a smaller or greater number of paracladia depend on the devel- opmental stage of a plant and on environ- mental circumstances. The latter especially applies to annuals. Under severe environ- mental stresses such plants can be highly re- duced and develop reduced inflorescences only, as Troll (1964: 363-364) described for mmannia coccinea Rottb. The original plant depicted by Troll is presented here in a pho- tograph (Fig. 28 IV). Probably the plant in- vestigated and figured by Troll is not Am- mannia coccinea but A. latifolia L.; the same then applies to Figure 28 IV. The hypotagma is reduced to two nodes. Only the uppermost of them bears paracladia, while the paracladia in the axils of the cotyledons remain unde- veloped. The largest part of the plant is formed by the frondose proliferating florescence with many triadic partial florescences. In reduced plants of Pleurophora polyandra the inflo- rescence frequently only consists of a short headlike spike that is preceded by the coty- ledons and two pairs of foliaceous leaves with paracladia absent from their axils (Fig. 28 I). In other individuals, more or less vigorous paracladia ending in a florescence can develop from the axils of the cotyledons and/or from the axils of the following leaf pairs (Fig. 28 II). In P. pusilla Hook. & Arn. the reduction of ramification seems to be a constituted char- acter. By dense sowing of Lythrum hyssopifolia L., plants without any paracladia were ob- tained by Troll (posthumous notices), whereas in normal plants the paracladial zone is well developed. The frequent occurrence of accessory buds that develop as single flowers or vegetative or flower-bearing branches has been men- tioned already. With the exception of Cuphea (Eichler, 1878b: 479), there is scarcely a genus of Lythraceae in which they are com- pletely missing. Since Koehne (1873: 112) has given a survey of the various forms and their distribution within the family, we can confine ourselves to the treatment of the ge- nus Lythrum, in which the occurrence of accessory flowers is very important for the appearance of the inflorescences. Volume 75, Number 1 1988 Weberling 253 Inflorescences in the Myrtales In Lythrum the florescences of subgen. Lythrum (= Salicaria Koehne) and of sect. Mesolythrum Koehne of subgen. Hyssopi- folia are spikes or botrya. A flowering individual of Lythrum sali- caria L. (Fig. 29 II, III) shows a clear zo- nation: the frondulose main florescence, a paracladial zone (enrichment zone) in which the paracladia are still unfolding, and an in- hibition zone below the paracladial zone. Since Lythrum salicaria is a perennial, the inhi- bition zone is preceded by an innovation zone at the base of the plant. According to Troll (posthumous notices), L. salicaria is a sub- shrub rather than a perennial herb since the primary axis as well as the primary root shows intense lignification, as in Hypericum per- foratum. In young plants (Fig. 29 I) a favored development of the axillary buds of the cot- yledons (cotyledonary shoots) and of the next basal leaf pair can be seen. The cotyledonary shoots can develop branches from the axils of their prophylls at early stages, and a phyl- loscopic accessory bud can arise in the axils of the cotyledons. In vigorous plants the par- acladial zone can be considerably expanded, and the paracladia can develop second-order paracladia. In the basal part of the flowering zone the phyllotaxis changes from decussate to alternate (Wydler, 1860: 238; for the phyllotaxis of L. virgatum and L. hyssopi- folia see Wydler, 1872: 254). The flores- cences are thyrses, however exceedingly mod- ified by the formation of phylloscopic accessory cymes or flowers (Roeper, 1826: 109). Eichler (1878b: 479) observed that below the cymose partial florescence there are most- ly two accessory three-flowered cymes (Fig. 30 IIa), the lower of which is often rudimen- tary. There is an additional accessory flower below each of the lateral (secondary) flowers of the primary partial florescence. These ad- ditional flowers bear prophylls as do the reg- ular lateral (secondary) flowers of the cyme. Thus a transversal series of five flowers re- sults, below which there is a three-flowered cyme still provided with prophylls of first or- der, and below this another one of rudimen- tary flowers without prophylls. Lythrum pur- FIGURE 28. dra; reduced form (I) a ant with w basal paracladia (II). E pen linifolia; riea x ri ln individual. — nnia coccinea; reduced individ- Amm ual. (Photographs 111 ud IV from Troll.) Lyt pog " II. digi lis e ell-d shianum Steudel is reported to be similar in this respect (Fig. 30 IIb). Troll (posthumous notices), however, found triadic flower groups in place of the singular accessory flowers that accompany the sec- ondary flowers of the regular partial flores- cence (Fig. 30 Ia). He also noted that there may be one (as,) or two (as,, as;) accessory cymes. In the latter case the as, bears single accessory flowers in the axils of its prophylls, and the as, likewise is provided with prophylls (Fig. 30 Ic). In all partial florescences, anthesis of the primary flower takes place first, followed by the secondary flowers, then the primary flow- 254 Annals of the Missouri Botanical Garden FIGURE 29. shoots (cos) and a second pair of basal branches.—lI. Flowering individual with lar enrichment zone (paracladial zone, pz), an expanded paracladial zone. (Photographs from Troll. ers of the as,, and so forth. As a consequence, the efflorescence of the whole florescence is progressing from base to top repeatedly “as in Verbascum” (W ydler, 1843: 184, 1851a: 370-371). According to W ydler (1860: 240), who referred to Koch, it is not rare to have a vegetative accessory bud below the single flowers of the spicate florescence. Koehne (1903: 6) reported that in Ly- thrum tribracteatum Salzm. and some other species the accessory branch forms a short, dense botrytic florescence. By recaulescent shifting, the place of at- tachment of the subtending leaves (phero- phylls) can be relocated onto the pedicels of the flowers, which in reality originate from the axils of the pherophylls. Thus these can be situated just below the flower. Examples can be found in Rotala and Decodon. nd inhibition zone (iz). — ) ~ i 4 AL YW UY y w pz TI Lythraceae. Lythrum salicaria L. —1. Young plant with favored oar of the cotyledonary arge main florescence (m III. Plant with ola leaf whorls and In most species of Cuphea, the attachment of the flowers within the botrytic florescences is modified by the coalescence of their pedicels with the main axis up to the next node (Hochs- tetter, 1850: 182; Wydler, 185la: 371, 1861; Eichler, 1878b: 478 ff.). This sort of concrescence is well known as concaules- cence. In C. nitidula Kunth, C. appendic- ulata Benth., and some other species (cf. Koehne, 1883: 119), the flowers are mostly still in normal axillary position. In the case that both axils of the decussate leaves are fertile (“Cupheae oppositiflorae’’), the pedi- cels of both flowers are attached to the main axis up to the next node (Fig. 31). Thus the flowers and prophylls attached between the insertion of the leaves (or somewhat below) really originate from the leaf axils of the pre- ceding node. In **C. oppositiflorae" even the Volume 75, Number 1 1988 Weberling 255 Inflorescences in the Myrtales ° o No 0000) a o ası Ia asp e i 0008; (OOO) AA - P " 9000 ç (00000) (OOO) Ib LS a LS FIGURE 30. Lythraceae. —l, Il. Partial florescences and accessory flowering branches (as, as,) of Lythrum salicaria in diagrams according to Troll (la—c) and Eichler, and of L. purshianum according to Eichler (IIb). — HI. Horizontal (Illa) and vertical (Illb) projections of ramification in alternatiflorous Cuphea latifolia after Eichler (slightly modified). initially vegetative branches show a slight con- caulescence (Koehne, 1873: 111). In the case that only one leaf axil of a node is fertile (C. alterniflorae), there is only one flower at- tached at the flanks of the next node. The bud situated in the second axil of the opposite leaves remains within the axil. Since the fertile axil of the third node always is placed above 256 Annals of the Missouri Botanical Garden FIGURE 31. Lythraceae. e eft. Cuphea latifolia; plant at the beginning of flowering.—Top, right. C. procumbens; florescence.— Bottom. C. micropetala; florescence (left), and detail (right). (Photographs from Troll.) Volume 75, Number 1 1988 Weberling 257 Inflorescences in the Myrtales - SE A > FIGURE 32. (1I). d A from Tro t LA E a "be X A 4 ph ` g 407 Ë dau « the fertile axil of the first, and the fourth always is superposed to the fertile axil of the second node (Fig. 30 IIIb), the result is two vertical rows of flowers diverging by 90? (just as two vertical rows of axillary buds, Fig. 30 a). The ligneous genus Rhynchocalyx Oliv., formerly included in Lythraceae and now re- garded to represent a separate family, has monotelic inflorescences. In R. lawsonioides iv., the inflorescence forms a bracteose thyrso-paniculate system at the end of leafy shoots. At the bases of the paracladia uniflo- rous or triflorous accessory branches can be found. The monotypic genus Alzatea, which is now regarded to form a separate family, has diplothyrsoidal inflorescences, probably of monotelic character. On the latter point some uncertainty remains because of the fragmen- tary state of the examined herbarium material of A. verticillata Ruiz & Pavón. The low- ermost thyrsoidal paracladia are accompanied bretaceae. Combretum coccineum; flowering branches in anthesis (I) and before anthesis ll.) by accessory branches that form thyrsoids too. In the monotelic character of their conical dithyrsoidal inflorescences, Rhynchocalyx and Alzatea at least do not join the progressed members of the Lythraceae. COMBRETACEAE The inflorescences of Combretaceae are polytelic throughout. Although there are pro- phylls present in Laguncularia, Lumnitzera, and Macropteranthes, the florescences are nearly always spikes or rarely botrya. The central type of inflorescence architec- ture within the family may be represented by Combretum coccineum (Sonn.) Lam. (Fig. 32 I). The leafy shoots terminate in a synflo- rescence composed of a botrytic main flo- rescence at the top of the whole ramification system, preceded by several pairs of paracla- dia. Phyllotaxis is often not strictly decussate 258 Annals of the Missouri Botanical Garden Il FIGURE 33. Combretaceae. Combretum cocci- neum.—4. Mode of effloration.—Il. Detail. (Photo- graphs from Troll.) within the family. Besides Terminalia, seven genera are reported to have alternate leaves, though in many cases they are subdecussate. In many climbers only the distal branches show decussate leaves. In Terminalia, Bu- cida, Buchenavia, Ramatuela, Anogeissus, Finetia, Conocarpus, and Lumnitzera, the leaves are spiral. The four distal pairs of par- acladia are confined to their (co-)florescences, whereas the two proximal pairs bear two or four pairs of second-order paracladia, which are likewise limited to their (co-)florescences. In descriptive terms the ramification can be classified as a heterothetic triplobotryum. The foliation in the proximal part of the main axis is frondose but quickly changes over to brac- teose; the flower-bearing twigs only develop small bracts. As shown by Figure 33 I, II, the efflorescence progresses from base to top in the sequence of the paracladia (I) as well as in the paracladia themselves (II); in this course the main florescence gains a slight lead in comparison with the preceding paracladia. Figure 33 II also shows that all flowers of the florescences, even those of plagiotropic ori- entation, are turned upwards, a typical fea- ture of many species, which has led to the popular name “cepillo de mono" (monkey's brush). This name especially applies to C. fruticosum (Loefl.) Stuntz (C. secundum Jacq., Fig. 34), in which the paracladia are poorly developed in favor of the vigorous develop- ment of the dense-flowered main florescence. In Figure 32 II, each of the first-order par- acladia is accompanied by an accessory branch bearing a botrytic florescence. This is typical of many Combretaceae. Synflorescences similar to those described here for Combretum coccineum can be found in various taxa of the family, such as La- guncularia racemosa (L.) Gaertner, Thiloa glaucocarpa (Martius) Eichler, Calycopteris floribunda (Roxb.) Lam., Meiostemon tet- randrus (Exell) Exell & Stace (cf. table I in Exell & Stace, 1966), Campylogyne Welw. ex Hemsley (the florescences with showy bracts), species of Buchenavia and Termi- nalia (T. tomentosa Bedd.; T. paniculata Roth, see also Sell, 1982, fig. 10a; T. citrina (Gaertner) Roxb. ex Fleming; 7. chebula Retz., cf. Brandis, 1893, fig. 55), and in Quisqualis. In Q. indica L., the inflorescence can consist of a spicate main florescence ter- minating leafy twigs, contrasting with them by its bracteous foliation. The main flores- cence, however, can be accompanied by one or two pairs of paracladia originating from the axils of the foliaceous leaf pairs preceding the main florescence. These paracladia usu- ally bear one or two more or less foliaceous leaf pairs at their bases. Likewise, the low- ermost leaf pair of the main florescence may be somewhat foliaceous. Cacoucia coccinea Aublet (Combretum cacoucia (Ball.) Exell) seems to be characterized by developing only a long voluminous main florescence (Fig. 35). The same applies to some other species for- merly included in this genus (Cacoucia splen- dens Hemsl. = Combretum bracteatum (Laws. pro parte) Engl. & Diels, Cacoucia paniculata Laws.). Lumnitzera coccinea and L. racemosa dif- fer considerably in the position of their spicate florescences (see also Brandis, 1893). In L. Volume 75, Number 1 1988 Weberlin 259 g Inflorescences in the Myrtales 1 34. Combretaceae. Combretum fruticosum GURE a (II). (El Salvador: Weberling 164, 175.) littorea (Jack) Voigt, normally a bracteose spicate main florescence terminates a leafy shoot, and paracladia are missing. In L. r cemosa Willd., however, the synflorescence consists of paracladia only, which form 10- 20-flowered, long-peduncled spikes originat- ing in the axils of foliaceous leaves of a leafy shoot with indeterminate growth. In L. coc- cinea, however, the foliaceous shoots that end in a spicate main florescence are not very large and represent branches of a shoot with indeterminate vegetative growth, too. ithin Terminalia, a great number, per- Q Cru E we AT ES ` ba a 26928 e w^ eornm» (= C. secundum); flowering branch (I) and botrytic haps the majority, of the species are char- acterized by leafy proliferating inflorescences. In some of these species the flowering branch- es grow with well-developed internodes, and after the production of several spicate para- cladia which originate in the axils of foliaceous leaves, these branches return to vegetative growth. Terminalia oblonga (Ruiz & Pavon) Steudel serves as an example (see also T. brownii Fresen., Sell, 1982, fig. 10d). In other species the internodes of the main axis remain short within the flowering zone. Thus the leaves are “crowded at the ends of the 260 Annals of the Missouri Botanical Garden > " 4 T xe—— | “py | Je. | (Caceeta Cv FIGURE 35. Combretaceae. A coccinea uon Aublet, J. B., Hist. Pl. Guiane Franc. III, pl. 1775). branches," and the synflorescence has a ro- settelike shape. This applies to 7. catappa L. (Fig. 36; for the crown form and branching pattern of T. catappa and T. latifolia see Fisher & Hibbs, 1982), T. bellirica (Gaert- ner) Roxb., 7. zollingeri Exell, T. sumatrana Miq., T. trivialis Slooten, and many other species, also to Bucida L. In T. gigantea Slooten, which shows a similar architecture, the spicate paracladia are very long, with many loosely arranged flowers. In T. australis Camb., on the other hand, the paracladia bear a smaller number of densely aggregated flow- ers that form a long-peduncled head. The flowering branches of this species are brachy- blastlike proliferating twigs with well-devel- oped internodes. They develop from axillary buds of an older shoot but also arise from its terminal bud and later continue its growth. In T. triflora (Griseb.) Lillo, the brachyblasts, which bear several bud scales at their bases, remain very short, at least during the anthesis of their four to six paracladia, which likewise form few-flowered, long-peduncled heads. Formation of flowers is limited to brachy- blasts in many species of other genera. Thus in the species of Pteleopsis Engl., the flow- ering branches are brachyblasts originating from older macroblasts. They bear a few bud scales at their bases, some pairs of foliaceous leaves, and an umbel-shaped to botrytic main florescence, which in its proximal part can bear small foliaceous pherophylls. We could, however, also observe heterothetic diplobot- rytic inflorescences with somewhat umbellate florescences in some specimens of P. myrti- folia (Lawson) Engl. & Diels. In Combretum salicifolium E. Meyer, and in C. apiculatum Sonder (Fig. 37 I) brachy- blastlike branches situated on macroblasts of the previous year bear simple paracladia in the axils of the two or three lowermost pairs of foliaceous leaves; these brachyblasts pro- liferate at the end of the anthesis of the par- acladia (see also Brandis, 1893, fig. 59). Com- bretum bracteosum (Hochst.) Brandis is similar. The florescences of Combretum salicifo- lium are contracted to 15-20-flowered head- like aggregations (see Brandis, 1893: 122, fig. 59A). The same applies to C. erythro- phyllum (Burch.) Sonder (Fig. 37 II), in which the proliferating brachyblasts bear several pairs of long-peduncled globose umbels arising from the axils of foliaceous leaves. Capituli- form florescences are also reported, for ex- ample in C. punctatum Bl. (van Slooten, 1922), whereas in C. apiculatum Sonder the axillary florescences are spicate (Fig. 37 I). Capitulate and very densely flowered, long- peduncled florescences can also be found in Guiera senegalensis Lam. and in species of Buchenavia, Finetia, and Anogeissus. In Buchenavia capitata (Vahl) Eichl. and B. ochroprumna Eichl., the paracladia bearing globose, ca. 20-flowered heads were observed on terminal rosette-shaped proliferating brachyblasts. They were inserted here at the base of the brachyblasts, thus preceding the leaf rosette. In contrast to these species, the floral zone of the brachyblasts in B. kleinii Exell succeeds the leaf rosette, and the inflo- rescences are ovoid. In some other species the paracladia bear loosely flowered spikes and are inserted at the top of the brachyblasts Volume 75, Number 1 1988 Weberling Inflorescences in the Myrtales A IN A v C FIGURE 36. acladium, the leaves partly removed; apex (a) of the main axis.—Il. Distal part of a paracladium.— lll. Conocarpus erecta; capitulate paracladium. more or less following the leafy zone (B. ma- crophylla Eichl., B. suaveolens Eichl.) or at the base of the brachyblasts, preceding the rosette (B. oxycarpa (Martius) Eichl.). In Anogeissus acuminata (Roxb. ex DC.) Wall. ex Bedd. and A. latifolia (Roxb. ex .) Wall. ex Bedd., the synflorescences con- sist of a globose main florescence and an enrichment zone comprising several paracla- dia, each originating in the axil of a foliaceous leaf and bearing a globose co-florescence on a long peduncle. The paracladia can be ac- companied by similar-shaped accessory branches. In Conocarpus erecta L. (Fig. 38) the inflorescence terminating a leafy shoot is com- N K NES, š WKS a): SA Se WSS us SN E Sy on Ç ; | UU hay ‘Vi i Yi A Z Uy ⁄ Y) Uy ^, Y / EM I UDB Combretaceae. I, Il. Terminalia catappa.—41. Proliferating inflorescence with one spikelike par- posed of one terminal and many lateral, short- peduncled, ovoid capitula, which together form a heterothetic botrytic or diplobotrytic sys- tem. The basal leaves within this system are foliaceous, diminishing distally, and ultimately convert to small bracts. At least in the distal part of the inflorescence the paracladia are accompanied by accessory branches forming particular peduncled capitula. In general the tendency to form condensed inflorescences is not very strong within the Combretaceae. Exell (1962), however, saw some evolutionary consequences resulting from a conflict between pollination strategies and the development of winged fruits of "massed flowers" that might cause "space 262 Annals of the Missouri Botanical Garden FIGURE 37. Combretaceae. Brachyblasts. —1. Combretum apiculatum. (SW Africa, Windhoek, R. Seydel 3666 ( M). m C. erythrophyllum. (South Africa, Cape Prov.: King Williams Town, D. M. Comens 1715 (M).) problems." The problem, however, seems to be more complex and needs empirical treat- ment with regard to flower biology as well as to carpology and chorology. PENAEACEAE The inflorescences of Penaeaceae were carefully investigated in the monographic work of Dahlgren (1967a-c, 1968, 1971). The basic type of ramification is a thyrsoid (a thyrse with a terminal flower) as in Sonder- othamnus petraeus (Barker) Dahlgren (Fig. 39 II) and S. speciosus (Sonder) Dahlgren. In the majority of taxa the cymose (triadic) paracladia are reduced to their primary flow- er. The thyrsoid thus is converted to a stachy- oid as presented by Saltera sarcocolla (L. Bullock (Figs. 39 L 41 II, 42 I). A peculiarity of the Saltera inflorescence is that the terminal flower is preceded by two pairs of pale membranaceous bracts. The low- S er bracts are narrow but broaden distally to an emarginate end; the next pair is much narrower, nearly linear, but also broadens to a truncate or slightly emarginate end. Two or three of the preceding leaf pairs bear single flowers in their axils, each of them with a pair of linear-spathulate prophylls. Since these prophylls are similar in shape to the upper pair of bracteoles preceding the flower that terminates the whole inflorescence, one could wonder if this flower is terminal or lateral. Dahlgren (1968), however, reported that the inflorescence can be reduced so far that the terminal flower is the only one remaining (Fig. 43 II). This, together with the comparison with the structure of inflorescences in some related taxa, particularly that of Stylapterus fruticulosus (L. f.) A. Juss. (Figs. 40 II, 41 IV), confirms that there is a true terminal flower. In some other taxa the terminal flower is Volume 75, Number 1 1988 Weberling 263 Inflorescences in the Myrtales sometimes facultatively lacking or missing in all specimens. Thus in Glischrocolla formosa (Thunb.) R. Dahlgren the “inflorescence apex generally bears a terminal flower, but this may drop in an early stage" (Dahlgren, 1967b). The inflorescences of Brachysiphon ru- pestris Sonder and B. mundii Sonder still have a terminal flower, whereas B. fucatus (L.) Gilg and B. acutus (Thunb.) A. Juss. have indeterminate spikes. In such cases the "apex ends in some scalelike leaves" (B. fucatus) or *as a dry tip" (Dahlgren, 1968). In B. fucatus (Fig. 39 IV), however, all the para- cladia end in a terminal flower, and this refers not only to paracladia that are provided with the pair of prophylls only (and therefore oth- erwise could be interpreted as partial flores- cences of a polytelic thyrse), but also to the lower paracladia with three or more leaf pairs. With some alterations the same applies to Brachysiphon acutus. Consequently we have to interpret these inflorescences as truncate monotelic systems. In Stylapterus the inflorescence is “gen- erally indeterminate, the apical tip being de- generated and dry, but in S. fruticulosus and S. ericifolius it often ends with a terminal flower" (Dahlgren, 1967a). The semidi- agrammatic figures of “selected rich-flowered inflorescences" of both species (Fig. 40) show that there is a strong tendency to reduce the number of flowers in the paracladia to their terminal flower. Particularly in S. fruticulo- sus the result of this reduction is that the terminal flower of the whole ramification sys- tem as well as the terminal flower of the paracladia are preceded by two or three pairs of sterile leaves (see also Fig. 41 IV). This is in good concordance with the occurrence of two pairs of sterile bracts below the terminal flower of the Saltera inflorescence (Fig. 39 I). We also should note that the flowering system of Stylapterus fruticulosus has a sec- ondary inhibition zone below the terminal flower in addition to the primary inhibition zone that commonly precedes the paracladial zone. Dahlgren (1971) described the inflores- cence of Penaea acutifolia A. Juss. as “gen- FIGURE 38. Combretaceae. Conocarpus erecta; dia- gram of a flowering branch. erally with a terminal flower," whereas P. mucronata L. was reported as “with or with- out terminal flower," and P. cneorum Meerb. as “without or sometimes with a terminal flower, when well developed with 6-14 flow- ers." [n our own investigations based on very restricted material only, we could only once find a terminal flower. An indeterminate (an- auxotelic) inflorescence of P. mucronata is shown in Figure 42 II, a terminal flower in Figure 41 III. Dahlgren (1971: 8) stated that in general the “flowering sequence is acro- petal, but when a terminal flower is present, this develops in head of the upper neighboring flowers." This feature is very typical for ter- minal flowers. In all genera, innovation, i.e., the produc- tion of vegetative shoots continuing the ram- ification system and perhaps later producing Annals of the Missouri Botanical Garden Volume 75, Number 1 1988 Weberlin 265 g Inflorescences in the Myrtales Penaeaceae. Semidiagrammatic representation of selected rich-flowered Donum —I. Sty- lapterus ericoides (shed bracteoles with broken lines). —1II. S. fruticulosus. (All from Dahlgren terminal inflorescences, takes place from the axils of foliage leaves somewhat below the flower-bearing zone. Inasmuch as the spikelike inflorescences actually no longer develop a terminal flower, they could be interpreted as a botrytic poly- telic main florescence. We hesitate, however, to classify them in this way, not only because terminal flowers are reported for some cases, but particularly because of the lack of par- acladia terminating in co-florescences. If such paracladia could be found in well-developed inflorescences, this would prove that the strong evolutionary tendency from the formation of monotelic inflorescences to the development of polytelic systems that appears indepen- dently in various branches of the family has really arrived at the polytelic stage. Without this proof we prefer to classify the indeter- minate inflorescences as truncate monotelic systems. A special case within the family is the in- florescence of Endonema. In both species, E. lateriflora (L. f.) Gilg and E. retzioides Son- der (Fig. 43 I), two or more single flowers are situated in the axils of leafy pherophylls FIGURE 39. sarcocolla. (South Africa. Cape Prov.: Forstmeier s.n. IV. B. fucatus. (II-IV from Dahlgren.) Penaeaceae. Semidiagrammatic representations of richly developed inflorescences.—I. Saltera —II. Sonderothamnus petraeus.—///I. Brachysiphon acutus. — 266 Annals of the Missouri Botanical Garden ii Y, y A Z š Z W cn 8 e NY FIGURE 41. Penaeaceae. I, Il. Flowering branches. —1l. Penaea mucronata. (South Africa. Cape Prov.: near Elim, Weberling 7819.) —//. Saltera sarcocolla. (South Africa. Cape Prov.: Fernkloof, Hermanus, Weberling 7862a.) —IH, IV. Penaea mucronata and Stylapterus fruticulosus; terminal flowers subtended by three decussate pairs of bracteolate leaves or bracts. In IV the lowest pair is shed. (HI, IV from Dahlgren.) Volume 75, Number 1 1988 Weberling Inflorescences in the Myrtales 267 a 4 4%, Z e S SS I FIGURE 42 Penaeaceae.— l. Saltera sarcocolla; inflorescence. (South Africa. Cape Prov.: Weberling 7862a.) — II. Penaea mucronata; inflorescence apex ending in a sterile bud. (South Africa. Cape Prov.: Weberling 7819.) on a young branch, which continues vege- tative growth. Each of these lateral flowers bears two (E. retzioides) or three (E. later- iflora) pairs of bracts. Dahlgren (1967c) re- ported that suppressed buds are sometimes found in the axils of the lowest pair of ““brac- teoles.” Thus the whole branchlet in Figure 43 I may be explained as a proliferating monotelic inflorescence with highly reduced paracladia. MELASTOMATACEAE Variety of inflorescences in Melastomata- ceae corresponds to the size of the family but shows less diversity than Myrtaceae. The inflorescences are monotelic through- out. Cases of complete transition to polytelic structures were not found. Even truncation seems to be rare. A single case noted by Troll (posthumous manuscript) is that of Med- inilla magnifica Lindley, in which the inflo- rescence sometimes ceases development be- fore the ultimate lateral cymes and the terminal flower of the distal thyrsic zone have completed their formation. Thus the whole end of the inflorescence atrophies (Fig. 64 III). Cremers (1983/1986) reported that the thyrse of Desmoscelis villosa (Aublet) Naudin remains indefinite. This, however, may onl happen facultatively; in several plants inves- tigated terminal flowers of the thyrsoids were well developed (Fig. 45). On the other hand, proliferation is not infrequent. The basic form of ramification is a pleio- thyrsoid terminating in a leafy shoot, as in Miconia argentea (Sw.) DC. (Fig. 44). Such bracteose or frondo-bracteose pleiothyrsoids, diplothyrsoids, or even monothyrsoids (hap- lothyrsoids) can be found in a terminal posi- tion in nearly all the 1,000 species of the genus Miconia as well as in many other lig- neous genera such as Conostegia, most species of Tococa, and the large genera Leandra and Clidemia. Elongate conical thyrsoids or dip- lothyrsoids are also typical of many species 268 Annals of the Missouri Botanical Garden I of Tibouchina. In some species of Tibouchina and of Heterocentron, the thyrsoids get a somewhat botrytic appearance since the hy- popodia of the paracladia of first order are long compared with the other internodes (“thyrsus racemiformis"). Many examples have been analyzed and figured by Cremers (1983/1986), who published a thorough study on inflorescence structures of Guianese Me- lastomataceae. His investigations also include interesting studies on growth forms. While the manuscript of the present article was in press, W. S. Judd (1986) published the results of studies on variation in inflorescence posi- tion in Miconieae. The cymose ramification of the partial in- florescences usually remains limited, and the mode of ramification is different. Eichler "HA un FIGURE 43. Penaeaceae. —1. Endonema retzioides; part of a branchlet with two uniflorous inflorescences. — II. Saltera sarcocolla; branchlet with terminal uniflorous inflorescence. (Both after Dahlgren, 1967c, 1968: 1. Hohenacker s.n.; I. South Africa: S. Cape Peninsula, Dahlgren & Peterson 652.) (1878c: 483) observed a formation of a heli- coid cyme (bostryx) by preferential ramifi- cation from the axils of the 8-prophylls in Tibouchina (cited by Eichler as Lasiandra). A bostryx also is reported for Centradenia floribunda Planchon by W ydler (1878: 349), who supposed that ramification continued from the axils of the 8-prophylls as in Tibouchina. Ziegler (1925: 410), however, observed a preferential ramification from the axils of the larger a-prophylls. The statement of Eichler (1878c: 483) that in the helicoids of Centra- denia only the fertile prophylls are developed proved to be wrong (Troll, posthumous manu- script). Krasser (1893) reported helicoid cymes for Miconia secundiflora Cogn. Sim- ple or double helicoids also occur in M. hook- eriana Triana (Fig. 46 I) and Fordiophyton. Volume 75, Number 1 1988 Weberli ng 269 Inflorescences in the Myrtales FIGURE 44. Melastomataceae. Miconia argentea; monotelic pleiothyrsoid inflorescence. (El Salvador: Weberling 2024.) The formation of scorpioid cymes (cincinni) by preference of the ramification from the axils of the a-prophylls was reported by Eich- ler (1878c: 483) for Salpinga and Clidemia. Scorpioid ramification of the cymose partial inflorescences is also known for Rhynchanth- era (Wydler, 1851a: 370), the Bertolonieae and Sonerileae (see Krasser, 1893, and p 278), Arthrostema, Centradenia inaequila- teralis (Schldl. & Cham.) G. Don (see p. 282), Amphiblemma cymosum (Schrader & Wendl.) Naudin (see p. 272), Appendicu- laria thymifolia (Bonpl.) DC., and Aciotis acuminifolia (DC.) Triana, 4. acutiflora (Martius) Triana, 4. longifolia Triana, and others. The cymose paracladia in the mono-thyr- soid inflorescence of Miconia hookeriana Triana are simple helicoid cymes in the distal part, and double helicoid cymes in the prox- imal part (Fig. 46 I). The inflorescence, which has only caducuous bracts, is preceded by several pairs of foliage leaves bearing vege- tative buds in their axils. From the axils of older leaves sylleptic branches with several T % 2 á t PA, } w du uidi FiGURE 45. Melastomataceae. Desmoscelis villosa. (N. Brazil: Ph. v. Luetzelburg 22181 (M).) —I. Distal part of Il. T, terminal flower. foliaceous leaves and a terminal thyrsoid can develop (Fig. 46 II). In most of the ligneous Melastomataceae, the axillary buds of the leaf pairs preceding the terminal inflorescence develop innovation shoots some time after anthesis, as is shown here for Miconia guatemalensis Cogn. (Fig. 47 II). The order of precedence in this de- velopment can be indicated already by the size of the buds (Fig. 47 I) On the other hand, there are ligneous members of the family with proliferating in- florescences. A favorable disposition for pro- liferation seems to be given with the occur- rence of leafy pherophylls, as in Huilea macrocarpa Uribe (Fig. 47 III). After the production of long-peduncled axillary thyr- soids or cymes the apex of the inflorescence axis returns to vegetative growth, producing axillary buds that develop vegetative branch- es instead of flower-bearing paracladia (Mora Osejo, 1966). Among the numerous species of Clidemia, which mostly have terminal bracteose thyrsoid or diplothyrsoid inflores- cences, there is C. rubra (Aublet) Martius 270 Annals of the Missouri Botanical Garden Miconia hookerian soid. —1l. Vertical projection of a flowering branch oa (From FIGURE 46. Mela stomataceae. with leafy proliferating inflorescences (Fig. 48 I). In contrast to Huilea macrocarpa, how- ever, the paracladia in C. rubra are reduced to sessile densely flowered cymes. Proliferation also can be observed within the genus Comolia, which also comprises some herbaceous species with proliferating inflo- rescences (Cremers, 1983/1986). Prolifer- ating inflorescences are characteristic of many or all species of many additional genera, in- cluding Amplectrum (homoeandrum Stapf), Blastus, Bellucia, Cambessedesia, Ernestia, Maieta, Marumia, Mecranium, Meriania, Microlicia (M. pseudo-scoparia Cogn.), Mi- crophysca, Myriaspora (M. egensis DC.), Myrmidone, Ochthocharis, Opisthocentra, Ossaea, Plethiandra, Tococa, Trembleya, and others. In Acanthella, the paracladia of the proliferating inflorescence are reduced to single flowers; the same applies to species of Blakea (B. spruceana Cogn., see fig. 80 in Krasser, 1893), Kibessia, Pyxidanthus, and Topobea. Topobea and Blakea are placed in I —1. Horizontal projection of the ramification of a thyr- Troll.) the Blakeae, which is characterized by the flowers bearing an involucrum of two or more pairs of bracts. This may be interpreted as the remnant of a formerly richer ramification. In addition, in many species of Blakea and Topobea, we found one to three accessory single flowers below the original one. Delayed anthesis of the paracladia of pro- liferating inflorescences in such ligneous plants results in a more or less pronounced cauliflo- ry. Thus in contrast to Clidemia rubra, which was mentioned above for its proliferating i in- florescences, C. septuplinervia Cogn. is cau- liflorous (Cremers, 1983/1986). The same applies to many Memecyloideae. Nearly all species of this subfamily are characterized by auxotelic (or sometimes anauxotelic) inflores- cences or by more or less cauliflorous partial inflorescences. Transitional forms may be found even in the same plant. Only in Me- mecylon cumingianum Presl we found axil- lary and terminal thyrsoids. The latter, how- ever, were terminating an older ligneous stem Volume 75, Number 1 Weberling 271 1988 Inflorescences in the Myrtales aso as{ aso as, aso as3 IV FicuRE 47. Melastomataceae. Vertical diagrams of flowering shoots and inflorescences.—I. Miconia squa- mulosa (from Mora).—1l. M. guatemalensis. (El Salvador: Weberling 2299.) — III. Huilea macrocarpa; with a proliferating inflorescence axis (from Mora.) —IV. Tococa symphyandra; synflorescence with accessory flower- bearing branches of first to third order (as,-as,). (All from Troll.) 272 Annals of the Missouri Botanical Garden ^4 I etd 48. ling 732.) — Marre branch gon Troll.) that was still bearing its foliage leaves. Cau- liflory is also well known for the genera Lor- eya, Henriettea, Henriettella (e.g., H. mac- fad yenii (Triana) Alain (Fig. 48 III), and A. glazioviana Cogn., see fig. 79B in Krasser, 1893), Bellucia (e.g., B. imperialis Sald. & Cogn.), and Myriaspora. All of these genera belong to the Miconieae, which also includes Clidemia. The contours and forms of terminal inflo- rescences can be modified by different fea- tures. Thus the conical thyrsoid, as mentioned already for Conostegia or Miconia (Figs. 44, 47 I), can be transformed into a corymboid by more or less effective suppression of the internodal growth of the inflorescence axis, combined with pronounced basitonic devel- opment of the paracladia and especially pro- longation of their hypopodia. Two shrubs, Vici eund —1. Clidemia rubra; proliferating flowering shoot. Tococa spadiciflora; spadixlike inflorescence (from Troll). (El Salvador: La Palm —Ill. Henriettella macfadyenii; Fordiophyton fordii (Oliv.) Krasser and Di- chaetanthera cormybosa (Cogn.) Jacques- Félix (see Krasser, 1893: 156), may be men- tioned as examples. In Amphiblemma cymosum (Schrad. & Wendl.) Naudin (Fig. 49), a corymboid of pleiochasial structure is formed. The inter- nodes in the distal part of the inflorescence axis, with the exception of the final internode, remain very short. As a result the insertions of the paracladia are crowded together (Fig. 49 I). The arrangement of these paracladia is not decussate as might be expected, but alternate, since phyllotaxis changes above the uppermost pair of foliaceous leaves, which is usually somewhat smaller. The bracteose pherophylls of the paracladia are shifted on their axillary paracladia by recaulescence. Usually there are five paracladia (maximum Volume 75, Number 1 Weberling 273 1988 Inflorescences in the Myrtales < Pc 7 u ZI “ay C» P. / K QS FIGURE 49. Melastomataceae. Amphiblemma cymosum.—4. Inflorescence apex, the paracladia have been cut off. —Il. Vertical projection at the beginning of anthesis of a flowering shoot.—Ill. Pleiochasial corymboid at the end of anthesis from above. Pc,-Pc,, paracladia of first to fifth order; E, terminal flower. (All from Troll.) 274 Annals of the Missouri Botanical Garden eight) forming helicoid cymes. If there are more than five paracladia, the lower ones can form “paired”’ or “double helicoids," because in these more vigorous paracladia both first- order prophylls are fertile. Since the paracla- dia elongate considerably by the formation of many flowers, the position of the terminal flower finally is at the deepest point of the whole pleiochasium (Fig. 49 III). In older and vigorous plants more paracladia, mostly dou- ble-helicoid, can develop from the axils of the foliaceous leaf pairs in the lower part of the stem (Fig. 49 II, pc,-pc,). The monotelic (pleio-)thyrsoid as the basic form of inflorescences in Melastomataceae can also be modified by reduction of 1) the ramification of the paracladia and 2) the num- ber of paracladia. 1) Reduction of cymose ramification of the paracladia ultimately results in the formation of dibotryoids, as shown for Leandra sylva- tica Cogn. or Miconia sarmentosa Cogn. by Cremers (1983/1986, figs. 25/3, 27/2; see also Tibouchina frigidula (DC.) Cogn., fig. 30/2), or in the formation of simple botryoids. A transitional stage on the way to formation of botryoids is exemplified by 7ococa gui- anensis Aublet (Fig. 50 II). Compared with other elongate thyrsoids like that of Tococa formicaria Martius (Fig. 50 III) or Allomor- phia magnifica Guill. (Fig. 50 I), its rami- fication appears to be much reduced, although the proximal paracladia are still three-flow- ered. If these paracladia are reduced to uni- florous elements, the inflorescence forms a simple botryoid, as occurs facultatively in Ti- bouchina canescens (D. Don) Cogn. (Cre- mers, 1983/1986, fig. 30/5), Clidemia capi- tellata (Bonpl.) DC. (Fig. 51 HI), and C. minutiflora (Triana) Cogn. (Cremers, 1983/ 1986, fig. 32/7), or generally in Adelobotrys spruceana Cogn. (Cremers, 1983/1986, fig. 19/9) or Castratella piloselloides (Bonpl.) Naudin (Cremers, 1983/1986, fig. 19/9). In another way the dense spikelike or near- ly spadix-formed inflorescence of Tococa spadiciflora Triana (Fig. 48 II) can be de- rived from an elongate monothyrsoid: as in- dicated in Figure 50 III, the hypopodia of first and second orders (dotted lines) remain undeveloped. If this is combined with a short- ening of the main axis, a capitulate inflores- cence like that of T. capitata Cogn. (= Sa- graea capitata Triana) results. According to Troll (posthumous manuscript), the inflores- cence apex of T. spadiciflora probably re- mains indefinite. If in contrast to the previous examples all internodes of the inflorescence are prolonged, a loosely branched thyrsoid results, as in Nep- sera aquatica (Aublet) Naudin (Fig. 57 I). As already mentioned for ligneous mem- bers of the family, foliaceous proliferating bo- trytic inflorescences are not rare (see also Tibouchina axillaris Cogn., Fig. 51 I). Com- olia purpurea Miq. (Cremers, 1983/1986, fig. 22/1) and Tibouchina petroniana Cogn. (Cremers, 1983/1986, fig. 30/8), with its diplobotrytic inflorescences, are herbaceous examples. 2) This reduction of cymose ramification of the paracladia resulting in the formation of botryoids can be combined with diminution of the number of paracladia. Thus the bo- tryoids of Castratella piloselloides mentioned above (Fig. 50 IV) bear two or three pairs of paracladia only. Reduction can go further, as in Clidemia involucrata DC., with sometimes only two uniflorous paracladia below the ter- minal flower (Cremers, 1983/1986, fig. 21/ 11). Ultimately a solitary terminal flower re- mains, as in the fruticose Tibouchina sello- wiana (Cham.) Cogn. and T. petroniana Cogn. (Fig. 51 II; Cremers, 1983/1986, figs. 30/ 6, 7). Cogniaux (1888: 598) described the flowering system of the latter as “floribus ad apices ramulorum solitariis; bracteis saepius 6." Indeed, the main axis and a series of FiGURE 50. Melastomataceae. magnifica. — 1I. Tococa guianensis. — ll. n T. spadiciflora .— IV. Castratella piloselloides. V-X. Rich- — Vertical diagrams of inflorescences and flowering plants. —L dci Lio . fo ormicaria; the ee (dotted lines) remain veloped owered plants of Pterolepis ane (V, poor- VI); P. repanda (VII, VIII); and P. pauciflora (IX, X). (All from Troll. Volume 75, Number 1 1988 Weberling 275 Inflorescences in the Myrtales € x she sl he AE LG AE AR .Qo. Yo > 276 Annals of the Missouri Botanical Garden FicunE 51. Mela: niana.—//]. stomataceae. foliaceous paracladia originating from the ax- ils of foliaceous leaf pairs in the proximal part of the main axis bear a single terminal flower only. Each of these flowers is preceded by three pairs of sterile bracts—a remnant of the thyrsoid ramification system In the prevalently herbaceous genus Sal- pinga were found two-flowered inflorescences in S. pusillum (Gleason) Wurd. (= Macro- centrum pusillum) and uniflorous inflores- cences in S. glandulosum (Gleason) Wurd., whereas within the related genus Macrocen- Vertical diagrams of inflorescences. Clidemia ide 2 Il diagrams according to Martius, Flora Brazil. XIV/3, pl. 92, XIV /4, pl. 128; lll after Cremers, modified O e € S £ — I. Tibouchina axillaris.—7//. T. petro- Y trum, M. cristatum (L. C. Rich.) Triana ap- parently has cymoids with scorpioid paracla- dia (Krasser, 1893, fig. 75C); the same applies to M. latifolium Wurd. (Cremers, 1983/ 1986: 69). According to Cremers (1983/ 1986, fig. 25/5, 5'), M. vestitum Sandw. normally develops the terminal flower only, but in some exceptional cases this can be accompanied by two lateral flowers. n such cases uniflory is obviously facul- tative. This especially applies to individuals of herbaceous annual species grown under Volume 75, Number 1 1988 Weberling Inflorescences in the Myrtales unfavorable conditions. In such plants the ramification is more or less reduced, some- times with the result that only the terminal flower is developed. According to Troll (post- humous manuscript), examples can be found within the genus Pterolepis, which is closely related to Tibouchina. Figure 50 shows vig- orous (V) and reduced (VI) plants of P. tri- chotoma Cogn., which differ in the number and the vigor of their cymose paracladia. In Figure 50 V the lowest pair of paracladia has the longest hypopodia of first and second or- ders and is the most copiously branched. The uppermost pair of paracladia, however, is also vigorous and bears more flowers than the preceding one. This indicates a certain ten- dency of an acrotonic support of ramification. Therefore it is not surprising that in the re- duced plant (Fig. 50 VI) the distal pair of paracladia is the only one fully developed. In P. repanda (DC.) Triana (Fig. 50 VII, VIII) the acrotonic support of ramification becomes so effective that only the distal pair of par- acladia is developed and a more (VIII) or less (VII) copiously branched cymoid is formed. In P. pauciflora (Naud.) Triana pauperization of the inflorescence finally progresses to the formation of a three-flowered (Fig. 50 IX) or uniflorous (X) inflorescence. Some other ex- amples have been given by Cremers (1983/ 1986), who analyzed species of Acisanthera and Appendicularia and found uniflorous in- dividuals in Acisanthera bivalvis (Aublet) Cogn. (Cremers, 1983/1986, fig. 18/8); a similar diversity of inflorescence forms occurs in Appendicularia thymifolia (Bonpl.) DC. Altogether the inflorescences of the mostly or exclusively annual species of Acisanthera, Appendicularia, Aciotis (e.g., A. amazoni- ca, A. aequilaterialis, A. dichotoma Cogn.), and probably also Pterogastra are very vari- able: plants with extremely acrotonic ramifi- cation forming “dichotomic” inflorescences (cymoids) can be found beside plants with extremely reduced, sometimes even uniflo- rous inflorescences. The herbaceous Cata- coryne linnaeoides Hook. f. probably also can develop uniflorous inflorescences. Obligatorily uniflorous inflorescences are known for some fruticose taxa, especially the genus Chaetostoma and some species of the genera Marcetia, Lavoisiera, and Microli- cia. A flower-bearing shoot system of Marcetia sertularia DC. (Fig. 52 IIT) shows the branch- es more vigorously developed in the distal zone, i.e., an acrotonic ramification system. Most of the densely foliate branches as well as their relative main axes end in a terminal flower (Fig. 52 II). Lateral flowers or even vestiges of lateral flowers inserted below the terminal flowers cannot be found, in contra- distinction to some other species of the genus. The whole system, however, cannot be re- garded as an entire inflorescence but must be interpreted as a ramification system com- prising numerous uniflorous inflorescences. Accessory buds or branches are very com- mon within the family (see also Wagner, 1907). They form single flowers or dyads in the inflorescence of Pterolepis trichotoma (Rottb.) Cogn. (Fig. 50 V) and occur in the leaf axils of the main axis as well as of the paracladia and can even form triadic or many- flowered cymes, as in Adelobotrys ciliata (Naudin) Triana, A. permixta Wurd., Lean- dra polyadena Ule, and L. rufescens (DC.) Cogn. (Cremers, 1983/1986, figs. 19/2, 6, 24/1, 4), or even thyrsoid flowering systems, as in Miconia kappleri Naudin and M. til- lettii Wurd. (Cremers, 1983/1986, figs. 26/ 1, 2). If there are two or more buds, they are always arranged in a phylloscopic se- quence, as in Topobea guianensis Aublet (Cremers, 1983/1986, fig. 31/6) or Creo- chiton Bl. (see Wagner, 1907). Sometimes differentiation between the buds of the same axillary series can take place. In Ernestia confertiflora Wurd. the upper of two buds develops into a vegetative branch and the lower forms a flowering system, whereas in Maieta guianensis Aublet the reverse situ- ation occurs (see Cremers, 1983/1986, figs. 23/5, 25/6). Troll (posthumous manuscript) found in an inflorescence of Tococa sym- phyandra (Triana) Cogn. that 92 of a total 220 flowers belonged to accessory branches (Fig. 47 IV). The formation of cymoids has been men- tioned already in the context of the reduction Annals of the Missouri Botanical Garden 3 2. Melastomataceae. —1. Sonerila rotundifolia; flowering plant (after Beddome). E, terminal flower. II, oe Marcetia sertularia; —1I. Single branch with terminal flower (after De Candolle).—Ill. Flowering branching syst of inflorescences in reduced individuals of Pterolepis species (p. 277). Cymoids occur especially within the mostly herbaceous So- nerileae and Bertolonieae. Only a few ex- amples are needed here. In Calvoa sessiliflora Cogn. (Sonerileae) the cymoids are very conspicuous, especially in fruiting plants (Fig. 53 II). In this plant the-leaves of the main axis are already lost. The foliation of the main stem is leafy and only the distal pair of leaves is bracteose (Fig. 54 I). These bracts are inconspicuous and by recaulescence united with their axillary branches, the uppermost paracladia, which form simple scorpioids. Between these two scorpioids the main axis ends with the ter- minal flower that is sessile because the final internode remains undeveloped. The same ap- plies to the flowers of the scorpioid paracladia. The internode preceding the insertion of the Volume 75, Number 1 279 Weberling 1988 Inflorescences in the Myrtales scorpioid paracladia is very thin in proportion to the other internodes of the main stem. The same difference appears between the scor- pioids and the preceding paracladia, which in thickness attain about double the size. These paracladia, which originate from the axils of foliage leaves, repeat the architecture of the main stem: they end in a cymoid and their foliation is leafy, with the exception of the pherophylls bearing the scorpioids. However, there is one important difference: the leaf pairs, especially that in the median position are anisophyllous, and the leaf pointing out- wards by far exceeds the size and differen- tiation of its partner (Fig. 54 II). The trans- I | HIC > FIGURE 53. Melastomataceae.—I. Sonerila marga- ritacea.—//. Calvoa sessiliflora; flowering plant. (From Troll.) T fly » (d CB 072 mm pr A MN CO M < I IJJ C e 2 Min unm SS SS = zm WTA Pc 2 — | Pc3 y ET AQ W j 9, 7 A y WAWAN W. 777 TITA An ANNA Il I AY | 22 FiG 54. Melastomataceae. I, II. Calvoa sessiliflora.—1. Vertical diagram of a flowering individual; E, terminal flower; CoT, cotyledonary branches. —II. Paracladium (Pc, from I), view from the ventral (upper) side; vb, prophylls; +, the larger leaf of the anisophyllous second leaf pair.—Ill. Centradenia grandifolia; analytical vertical diagram of a flowering branch. (Il, Ill from Troll.) 280 Annals of the Missouri Botanical Garden UN ( 2) < C27 ES i d A Y A. Q2 7 < pil ) cnn DY, Ro HD x “I (I MD, Pea. DP iD Í AD ° «mmm C10 UU IES CTD SUM NA FIGURE 55. marmorata (11). (From Troll.) terminal flower; a, B = prophylls versely arranged prophylls are asymmetric, the side pointing outwards (downwards) being larger. The whole paracladial zone comprises two or three pairs of paracladia and is pre- ceded by an inhibition zone, within which only the cotyledonary branches show a slight ten- dency to unfold. In some way the species resembles Amphiblemma cymosum (Fig. 49), which belongs to the same tribe. In A. cy- mosum, however, the flower-bearing system that terminates the shoot comprises about five alternately arranged paracladia. The flower- ing system of Sonerila margaritacea Lindley (Fig. 53 I) is also similar to that of Calvoa sessiliflora inasmuch as there are two scor- pioid paracladia originating from the axils of a pair of bracts below the terminal flower. The whole *'double-scorpioid" is separated from the basal part of the plant by a long and comparatively thin internode. This scape is preceded by three pairs of frondulose to brac- teose sterile leaves that form a transition be- tween the preceding foliaceous leaf pairs and the uppermost pair of minute bracts. Since "y My VA E (im. E y A ALLN AID NY < S N > » QDD E A y E ES Ed | | tm oa x === => => == == => —— — | E=3 === E jm Melastomataceae. Vertical inflorescence diagrams of Sonerila margaritacea (I) and Bertolonia Pc,-Pc,, first through tenth order paracladia; E, terminal flower; E”, secondary the preceding pairs of foliage leaves bear par- acladia in their axils, the three pairs of sterile leaves form an intercalary inhibition. zone within the flowering system. Each of the par- acladia pc,-pc, (pc,-pc, are missing) in Fig- ure 55 Í nearly gives a complete copy of that part of the main stem that follows the point of its insertion, apart from the fact that all of them bear simple scorpioids. In 5. pilosula Thwaites this also applies to the main axis. Unlike previous species, Sonerila rotun- difolia Bedd. (Fig. 52 I) is a rosette plant. The basal part of the main axis bearing the foliaceous rosette leaves is short and vigorous; only the distal part tapers and forms a flower- bearing scape. Since the plant is hapaxanthic, however, the vigorous basal part cannot be termed a rhizome. The umbellike inflores- cence that is elevated above the rosette leaves by a long internode consists of two scorpioid paracladia that are inserted below the ter- minal flower. In 5. scapigera Dalz. there are more long-peduncled umbellike flowering sys- tems arising from the rosette. The morphol- Volume 75, Number 1 1988 Weberling Inflorescences in the Myrtales FiGURE 56. leaves removed. (From Troll. ogy of this plant has not been clearly eluci- dated, but we only can confirm that each of the flowering shoots bears its own basal rosette of foliage leaves. A transitional form that connects Calvoa sessiliflora and Sonerila margaritacea on the one hand with Sonerila rotundifolia and similar species on the other hand is Berto- lonia marmorata Naudin (Bertolonieae). In this semirosulate plant (Figs. 55 II, 56) the basal leaves are decussate, whereas the cau- line leaves change over to alternate arrange- ment. In the distal part of the main axis there is only one scorpioid paracladium inserted below the terminal flower. Some more para- cladia arise from the axils of the frondulose leaves in the lower part of the stem. These paracladia, however, can bear more than one scorpioid. The axillary buds of the upper ro- sette leaves can also produce paracladia, but most of them remain vegetative. The lower- most axillary buds ordinarily do not develop, nor do they function as innovation buds. Thus the plant is hapaxanthic. Since the scorpioid is formed gradually during anthesis and turns to the direction of the main axis, it can appear as a spike later on. Eriocnema acaulis Triana, also in the Ber- tolonieae, seems to resemble Bertolonia in its Melastomataceae. Bertolonia marmorata.—I. Flowering plant. —1l. Same, with most of the rosette l growth form, but has long-peduncled umbel- like cymoids similar to those in Sonerila ro- tundifolia. On the other hand, Cincinnobot- rys oreophila Gilg (1898, pl. VI), which belongs to the Sonerileae (Jacques-Félix, 1976), bears only one scorpioid paracladium below the terminal flower, which is elevated above the basal rosette by a long internode of the main axis (Fig. 57 II). Finally Monolena primulaeflora Hook. f. may be mentioned here for its very peculiar architecture, which Troll investigated (post- humous manuscript). With regard to the ar- chitecture of the primary axis, this plant is similar to Sonerila rotundifolia. The main axis is short and vigorous in its basal part, which bears the rosulate foliage leaves. In its distal part the axis tapers and forms a long scape that bears the terminal inflorescence consisting of the terminal flower and one or two scorpioid paracladia. The latter originate in the axils of two broad bracts that fit very well to protect the young flowers. Below these broader bracts there is a pair of smaller bracts that remains sterile. The significant feature of this plant is that it develops plagiotropic paracladia from the axils of the basal foliage leaves. These paracladia are equal to the main axis especially in their vigor. Together with 282 Annals of the Missouri Botanical Garden FIGURE 57. Prance et al. 11645 (M). Ulugura Mts., Schlieben 2824 (M).) Melastomataceae. the bulbous base of the primary axis they form a knobby lump. They remain short, and after the development of foliage leaves, finally taper into a scape bearing an inflorescence. The vigorous basal region of the paracladia can ramify repeatedly and form monochasial or dichasial sympodia. Thus the older plant gives the impression of a multiplicity of foliage leaves and inflorescences that extend and un- fold in a centrifugal manner. The character of the plant becomes still more complex be- cause of the anisophylly of the leaf pairs. One may suppose that the vigorous basal paracla- dia ought to be explained as proleptical in- novation shoots. A difference in size and/or form of paired leaves—commonly called anisophylly (Fig. 58)—already has been mentioned for some taxa. It is typical for a considerable number of species. It ensues from the dorsiventrality Nepsera aquatica; inflorescence. (Brazil. —Il. Cincinnobotrys oreophila; flowering plant. (Tanganjika Territory, distr. Mana L I { Amazonas: vicinity of Manau of the plagiotropic axis and is consequently often connected with the the development of the axillary shoots (see Troll, 1937: 608). This usually means that the leaves inserted at the underside of the axis are the larger ones (+ leaves), as shown here for Centra- denia grandifolia (Schldl.) Endl. (Fig. 59 I and the diagram in Fig. 59 II). In C. inae- quilateralis (Schldl. & Cham.) G. Don and C. grandifolia, which have very pronounced anisophylly, only the axils of the (+) leaves subtend axillary shoots. Thus there appear two rows of branches inserted on the under- side of the mother axis. This is shown for C. inaequilateralis by Figure 59 III, IV. The flowering shoot ends in a terminal flower, which is accompanied by a scorpioid or rarely he- licoid paracladium. This arises from the axil of the larger leaf of the small and scarcely anisophyllous leaf pair below the terminal Volume 75, Number 1 1988 Weberli 283 ng Inflorescences in the Myrtales flower (Fig. 59 V). The paracladia of the preceding leaf pairs differ from this upper- most paracladium by their leafy and more extended hypotagma. They repeat the archi- tecture of their mother shoot by ending in a terminal flower (E') and bearing one scorpioid paracladium (pc') in one of the axils of the distal leaf pair. The effloration of the para- cladia proceeds from base to top (Fig. 59 IV). In contradistinction to C. inaequilateralis, the paracladia (pc) of C. grandifolia (Fig. 54 III) develop flowering second-order paracla- dia (pc') from the axils of all leaf pairs, and even third-order paracladia (pc"), which are all helicoids. Generally in anisophyllous species the ax- illary shoot may be favored or inhibited in the same way as the subtending leaf, or rarely the favored (+) leaf bears a small (—) branch in its axil and the small (—) leaf a vigorous (+) branch. The latter case was described for Dissotis rotundifolia (Sm.) Triana by Troll (posthumous manuscript; see Fig. 64 I for explanation). For the Blakeae (Topobea and Blakea in- cluding Pyxidanthus), it was mentioned al- ready that their single flowers are enclosed by an involucrum formed by several pairs of bracts. The same applies to some species of Dissotis (e.g., D. rotundifolia) and many species of Osbeckia. In other species of this latter genus the involucrum is composed of large pherophylls of some paracladia that are crowded at the ends of the shoots, forming small heads. Often, as in O. brachystemon Naudin, O. chinensis L., and O. capitata Benth. ex Walp., we found the involucrum consisting of two leaf pairs which sometimes (in O. capitata) included only the terminal flower and one pair of uniflorous paracladia or even only the terminal flower. Many-flow- ered heads were observed in O. chinensis. The dense heads of Dissotis capitata (Vahl) Hook. f. (Fig. 60) consist of the terminal flower and one many-flowered monochasial- helicoid paracladium (Troll, posthumous manuscript). In the annual Nerophila gen- tianoides Naudin the involucrum is formed by foliaceous leaves. FIGURE 58. Aubl.; anisophyllous branch. (Brazil: vicinity of Pará, Baker 91 (M).) Melastomataceae. Maieta guianensis Well known for its large and showy pink bracts is Medinilla magnifica Lindley. This evergreen plant must be regarded as a shrub since it has a basitonic mode of ramification that is conspicuous even when young (Fig. 61 I). The inflorescences contrast sharply with the vegetative parts by the abrupt change of the leaf character and its pendulous posi- tion (Fig. 61 II). This position is produced primarily by an active incurvation of the low- ermost pedunclelike internode (segregation internode); later it is the weight of the many- flowered inflorescence that keeps it in a hang- ing position, since the pink-colored segrega- tion internode is very thin. The proximal part of the diplothyrsoid inflorescence bears mostly two pairs of thyrsoid paracladia; in vigorous inflorescences there may be three or even four or five pairs. Sometimes the lowermost pair of bracts remains sterile. The bracts of the lowermost pairs commonly form false tet- ramerous verticils by an abbreviation of the internode between two dimerous verticils. In the distal and thyrsoid part of the inflores- 284 Annals of the Missouri Botanical Garden FIGURE 59. Melastomataceae. I, Il. Centradenia grandifolia. —/. Anisophyllous branch.—ll. Horizontal na gram; L (as in of main shoots with e flower (E) and the distal paracladium terminal flower); the small (—) and large (+) leaves (E) and scorpioid pc. cence the phyllotaxis changes to the forma- tion of proper tetramerous verticils. Some- times trimerous or hexamerous verticils occur; this also applies to M. sieboldiana Planchon. In the distal region of the thyrsoid part of the inflorescence the bracts become smaller and inconspicuous. e ramification of the paracladia within the thyrsoid region decreases gradually from the formation of maximally seven-flowered cymes in the proximal part to uniflorous par- n I), the larger, fertile leaves; l, the smaller, sterile leaves. —III—V. C. dL nea mi at the beginning of efflora n (E it it s.—V. Distal paracladium (pc,) ith its ol pue ) a — sequence of effloration. (From Troll. acladia in the distal part. The inflorescence can be closed by a terminal flower but often remains indefinite, and the ultimate lateral flowers atrophy (Fig. 64 III). In this case the result is a truncate monotelic synflorescence. If the terminal flower is present, its devel- opment is slightly precurrent in relation to the neighboring lateral flowers. Generally the effloration of the whole synflorescence is ac- ropetal. The thyrsoid paracladia at the base of the Volume 75, Number 1 1988 Weberlin 285 Inflorescences in the Myrtales FicunE 60. Melastomataceae. Dissotis capitata.— I. rescence pus above. (From Troll.) inflorescence more or less repeat the structure of the distal part with some alterations due to their lateral position and consequently their dorsiventrality. Their foliation starts with di- merous verticils of large bracts, especially the prophylls, that in their axils can bear thyrsoid paracladia of second order. The whole inflo- rescence is then a triple thyrsoid. In the distal part of the thyrsoid paracladia the bracts become smaller. The conformity of the paracladia with the distal thyrsoid part of the inflorescence also refers to the possi- bility that the apex can atrophy without the formation of a terminal flower. The inflorescences of Medinilla sieboldi- ana are also pendulous (Fig. 62 I) but differ from those of M. magnifica by their very inconspicuous bracts and by the lack of thyr- soid paracladia. The inflorescence thus is a monothyrsoid. It resembles M. magnifica in having tetramerous verticils of bracts. In the proximal part the cymose paracladia can II Shoot with terminal inflorescence.—Il. Headlike inflo- comprise 15 flowers, but distally they are gradually reduced to triads. The development of the terminal flower is conspicuously pre- current to the neighboring lateral flowers. Medinilla pendula Merr. is another species with long-peduncled hanging diplothyrsoid in- florescences. There are, however, also species of Medinilla with upright inflorescences, as in M. javanensis Bl. (Fig. 62 II) and M. venosa Bl. (Fig. 63 I). In the latter species and in M. magnifica there sometimes occur cauliflorous inflorescences from the axils of leaves that have fallen off already (Fig. 63 II). Species with proliferating inflorescences are not rare, and some of them show a more or less pronounced tendency to cauliflory. Med- inilla myrtiformis Triana bears few-flowered botryoids and triadic accessory branches (per- haps even botryoids) in the axils of foliage leaves of proliferating axes. Medinilla par- viflora Baker and M. papillosa Baker are 286 Annals of the Missouri Botanical Garden FIGURE 61. rescence, thyrsoid paracladia. (From Troll. similar, the latter showing a slight tendency to cauliflory. In M. monantha Merr. the ax- illary flowering systems are reduced to their terminal flower, which, however, is preceded by two pairs of small bracts. On the other hand, the axillary partial inflorescences of M. ericarum Jum. & Perrier are long-peduncled diplothyrsoids. In Medinilla ramiflora Merr., the axillary glomerate thyrsoid flowering systems efflo- resce when their foliaceous pherophylls have fallen off, and M. tawaensis Merr. with like- wise thyrsoid partial inflorescences is evi- dently cauliflorous. The same probably applies to M. clarkei King. Medinilla sedifolia Jum. & Perrier, an epiphytic plant with succulent foliage leaves, is another example with proliferating inflo- rescences. A zone with a few single flowers in the axils of foliaceous leaves is followed by a zone with vegetative axillary shoots (Troll, 1973: 105, who partly refers to W. Rauh). Melastomataceae. Medinilla magnifica.—7/. Basitonic ramification of a young plant.—Il. Inflo- ) Probably after a while the main axis can re- turn to the production of lateral flowers again. The foliation of the uniflorous paracladia nor- mally consists only of the scalelike prophylls. Sometimes, however, the prophylls are folia- ceous and are followed by two pairs of scales, which in all cases investigated remain alto- gether sterile. There is also one case of “epiphyllous in- florescences" reported for the family: Phyl- lagathis scortechinii King, which was inves- tigated by Weber (1982). Among the ca. 35 species of Phyllagathis Bl. with essentially terminal **umbel-shaped" or “headlike” in- florescences, P. scortechinii is an exception. Its foliage leaves “are posed terminal on woody, axis-like structures (‘carriers’), from which they fall off after their life span. Ad- ditionally the ‘carrier’ bears several inflores- cences and/or vegetative buds along its upper side" (Fig. 64 II). Weber showed “‘that this structure is no axis, but a basal, anatomically Volume 75, Number 1 1988 Weberling 287 Inflorescences in the Myrtales I "T FIGURE 62. Melastomataceae. javanensis; terminal inflorescence. (From distinctly differentiated part of the leaf, onto which the axillary shoots (multiplied by ac- cessory shoot formation) are displaced." PSILOX YLACEAE Psiloxylon mauritianum Baillon, now rec- ognized to represent a separate family Psi- loxylaceae, “has small axillary racemiform inflorescences; these are perhaps anthotelic (botryoids?), but the limited available material (all dried) is insufficient to determine whether the apparently terminal flower is indeed truly so. Disperse phyllotaxy in the inflorescence (as well as in vegetative regions) and general recaulescence of the bracts increase the dif- ficulty of interpretation." We cannot add much to this statement given by Briggs & Johnson (1979: 181) for the same reason: scantiness of material. It does appear, how- ever, that there is no terminal flower. In this case the inflorescence could be termed a bot- — I. Medinilla sieboldiana; shoot with pendulous terminal inflorescence.—Il. M. Troll.) rytic florescence. According to our observa- tions, the flowers are subtended by bracteous pherophylls and bear bracteous prophylls. In the material investigated (Fig. 65) the flower- bearing systems are brachyblasts reduced to their florescence and inserted on older axes. Thus the plant might be called cauliflorous. MYRTACEAE In Myrtaceae, the “central type" and per- haps the phylogenetically primitive form of inflorescence is a monotelic thyrsoid (i.e., a thyrsic inflorescence with terminal flower) or panicle terminating a leafy shoot. Inflores- cences of this type can be found among the Myrtoideae (species of Syzygium and Fuge- nia) and the Leptospermoideae (species of Metrosideros, Eucalyptus, and Angophora), and they are characteristic of the Hetero- pyxidoideae (Heteropyxis natalensis Har- vey, Fig. 66; see also Weberling, 1963). A 288 Annals of the Missouri Botanical Garden 63. Melastomataceae. FIGURE (From Troll.) typical monotelic inflorescence is exemplified oy Syzygium aromaticum (L.) Merr. (Figs. 67, 68 I). The inflorescence axis ends with a terminal flower. This also applies to all floral branches below the terminal flower. All these branches, whether branched or not, are ho- mologous and all are referred to by the term paracladia. Accordingly the ramifications of these paracladia are called paracladia of sec- ond to n^ order (pc', pc", As is well known, the complexity of such an inflorescence, that is, the degree of ram- ification of the paracladia and the extension of the enrichment zone, may be modified to a certain extent in the same species. On the other hand, the differences in complexity may be a distinguishing character between differ- ent taxa. Thus in contrast to Syzygium aro- maticum, in S. paniculatum Banks & Gaert- ner (Fig. 67) the paracladia of first order are Medinilla venosa. — /. i Terminal inflorescence.—II. Cauliflorous inflorescences. uniflorous throughout (Troll, 1969: 258). This results in the formation of a botryoid. In this species, reduction may even go further until only the terminal flower remains. In Syzygium aromaticum, frequently in S. paniculatum, as well as in many other Myrtaceae, a pair of sterile bracts (metaxy- phylls, Zwischenblatter) Atc the ter- up flower can be observed. In S. panic- ulatum they may be eed by more foliaceous leaves that bear single-flowered (monadic) paracladia in their axils (Troll, 1969: 258- In Eugenia lanceolaria Roxb. (now Sy- macrocarpa zygium lanceolarium) and F. Roxb. (Syzygium macrocarpum) the usual form of the inflorescence seems also to be a botryoid, whereas in S. thumra (Roxb.) Merr. & Perry the ramification of the floral branch- es is increased (Fig. 69). To a certain extent Volume 75, Number 1 1988 Weberlin Inflorescences in the Myrtales 289 E: AP Sa SK FIGURE 64. Melastomataceae. ^ E D DE E om H » L 1] A : L II. Phyllagathis scortechinii; * (IF) and al (N).— from Weber, 1982.) this is due to a high degree of cymose branch- ing, which means a consecutive ramification from the axils of the prophylls. An inflores- cence with the main axis bearing lateral cymes is defined as a thyrsoid; in this sense the inflorescence of S. thumra as well as of Het- eropyxis (Fig. 66) is a pleiothyrsoid. As already shown by the few species men- tioned here, the foliation of the flower-bearing system may consist of bracts only (bracteose), or there may be a transition from foliage leaves at the base to distal bracts (frondo- bracteose), or the foliation may be leafy throughout (foliose, frondose). The latter ap- plies to the inflorescence of Syzygium acu- minatum (Roxb.) Miq. (Fig. 70). This inflo- rescence also presents another feature that is significant for many Myrtaceae: the main axis of the inflorescence is not closed by a terminal flower but ends in a bud (blastotelic in the sense of Briggs & Johnson, 1979: 176). Nevertheless, all of the paracladia, including those with more than one pair of flower-bear- ‘carrier’ mE leafand some (partly ese) fa tovestences inuit III. Medinilla magnifica; inflorescence apex in state of truncation. (1, III from Troll; II ing branches, are provided with terminal flow- ers, thus demonstrating the monotelic char- acter of the inflorescence (Troll, 1969: 255; Radlkofer, 1890: 184). The close morphological relations between these different forms of monotelic inflores- cences becomes evident by the comparison of closely related taxa, as among the various species of Metrosideros sensu lato that Daw- son (1968) investigated. Among these, M. albiflora Sol. ex Gaertner (Fig. 71 I) has a bracteose (thyrso-paniculate) inflorescence, and M. carminea W. Oliver (Fig. 71 II) a frondose diplobotryoidal inflorescence; both still terminate in an apical flower. As a result of further reduction, the inflorescence of M. diffusa Sm. (Fig. 71 III) consists of several densely contracted pairs of triadic paracladia only but still ends in a terminal flower. In the similar inflorescence of M. perforata A. Rich. (7 scandens Sol. ex Gaertner?), the terminal flower is replaced by a vegetative bud (Fig. 71 IV). Moreover, in the loose inflorescences 290 Annals of the Missouri Botanical Garden FIGURE 65. Psiloxylon mauritani- ; Fl. Maurit. Il, No. Psiloxylaceae. cum. (Mauritius 1825, Sieber s.n.; of M. kermadecensis W. Oliver (= M. poly- morpha Hook.?), not only the terminal flower closing the inflorescence axis, but also those of the paracladia of first order are replaced by buds (Fig. 71 V). The same applies to M. umbellata Cav. (= M. lucida A. Rich.?) with more contracted (Fig. 71 VI) or reduced in- florescences (Fig. 71 VII). Similar differences can be found among species of Angophora. While in A. hispida (Sm.) Blaxell, 4. floribunda (Sm.) Sweet, A. costata (Gaertner) Britt., A. melanoxylon Bak., and A. subvelutina F. Muell. the main axis of the (thyrso-)paniculate inflorescences is closed by a terminal flower, the thyrso- panicula of other species end in a bud. These terminal buds, which appear ^to consist of vegetative rather than floral or- gans," become abortive in many cases (the inflorescence being anauxotelic in the sense of Briggs & Johnson, 1979: 176). In many FIGURE 66. eropyxidaceae. Heteropxyis natal- ensis; diagram of a ton ering shoot. At the interruption si Africa: koppie in the area of the University of . Transvaal, Weberling 7644. other cases sooner or later the terminal bud continues growth beyond the flowering region, producing a vegetative shoot (for inflores- cences of this kind the term auxotelic is used by Briggs & Johnson, 1979). In Angophora costata, Briggs & Johnson (1979, fig. 6b, c) observed inflorescences closed by a terminal flower (“‘anthotelic”’ inflorescences) as well as those ending in a terminal bud, which some- times aborted and sometimes continued growth. The inflorescence of A. subvelutina was found with a terminal flower (Briggs & Johnson, 1979), while Troll (unpubl. data) found a terminal bud continuing growth in the same species. Thus especially within An- Volume 75, Number 1 1988 Weberling Inflorescences in the Myrtales f FIGURE 67. culatum. (From Troll.) gophora (and in some other taxa of the Eu- calyptus alliance) a high degree of flexibility in the formation of terminal flowers is re- vealed. Dawson (1968: 48) pointed out that in those species of Metrosideros that bear ter- minal buds, the *'bud is inactive during flow- ering, but may later develop into a leafy branch.” As will be shown later, however, the moment in which the apical vegetative bud turns to continue growth may be sooner or later. The reversion of the inflorescence apex to vegetative growth, commonly called prolif- eration (Troll, 1959a: 116), is characteristic for many Myrtaceae, perhaps even for the majority of species. Parkin (1914: 556) regarded proliferating inflorescences as a separate type of inflores- cence, which he called ““intercalary inflores- cences," because “the flower-bearing part of T | ) Myrtaceae. Inflorescences.—Left. Syzygium aromaticum (from Kohler, 1923). —Right. S. pani- the axis is ... intercalated between two fo- liage-bearing portions.” In using this term he especially referred to the Australian (or prev- alent Australian) genera Callistemon, Mela- leuca, and Metrosideros. Indeed the appear- ance of the flowering shoots of Melaleuca (Fig. 72) and Callistemon (Fig. 72) suggest this term, all the more as the process of the formation of a terminal inflorescence and pro- liferation can recur in regular intervals of development, mostly in connection with cli- matic factors. We must emphasize, however, that there are many taxa (e.g., l'eronica, Lysimachia, several ligneous Melastomataceae or Rubi- aceae) that include species with terminal brac- teose and terminal foliose inflorescences as well as species with proliferating leafy inflo- rescences. Often these forms are connected by continuous series of intermediate forms. Thus the so-called *'intercalary inflores- Annals of the Missouri Botanical Garden E 68. Myrtaceae.—1. Syzygium aromaticum; few-flowered inflorescence, the bracts fallen off. II, I. ie. T Inflorescence, the ter minal flower aborted (truncate monotelic synflorescence). —1lII. Distal part of the inflorescence in detail. Asterisk indicates place of the aborted terminal flower. (From Troll.) cences" are regarded as derived from the terminal ones. The fact that the development of flower- bearing systems and vegetative zones occurs successively in the same shoot, which contin- ues growing over a long time, may prove to be favorable under certain conditions, espe- cially for plants with enduring leaves. In the examples mentioned above, the pro- liferation takes place after or during the ef- floration of the inflorescence. The effloration, however, may be delayed somewhat more while the shoot apex reverts to vegetative growth and may even form branches, as in Beaufortia decussata R. Br. (Fig. 73 II). In such cases the normal zonation in the flowering systems of monotelic as well as of polytelic character (Fig. 3) seems to be re- versed, since the continuation of the vege- tative ramification system takes place above the flower-bearing zone. On the other hand, the extreme retardation of formation and an- thesis of the flowers may result in different manifestations of cauliflory. Thus in Calo- thamnus rupestris Schau. and C. villosus Ait., anthesis can be delayed so long that the resting flower buds are occluded by perider- mal tissue. The degree of ramification of the flower- bearing systems can be very different, and this applies to proliferating inflorescences as well as to nonproliferating inflorescences. Thus in the proliferating inflorescence of Pimenta dioica (L.) Merr. (Fig. 74), Gomidesia hook- eriana Berg, Krugia ferruginea (DC.) Ur- ban, or Calyptranthes chytraculia (L.) Swartz, the partial inflorescences that origi- nate from the axils of foliage leaves are highly ramified and present themselves as thyrsoid- paniculate systems. In contrast to the leafy Volume 75, Number 1 1988 Weberling 293 Inflorescences in the Myrtales K ra Magn í ^tt añ wo E Me COR DAVIS L3 ek eS 205 Hy š m~ > T Z, D» x Y FIGURE 69. main axis, they end in a terminal flower, thus revealing the monotelic character of the whole inflorescence. The same applies to some species of Marlierea and Myrcia, which pro- duce more or less ample ramified thyrsoid- paniculate paracladia, whereas other species of these genera, as well as Mitranthes eg- gersii and Myrrhinium loranthoides (Ca- brera, 1978: 78), only bear simple thyrsoids in the axils of foliage leaves. A high degree of flexibility in the alter- native formation of terminal flowers or ter- minal auxotelic or anauxotelic buds is rep- resented by Decaspermum paniculatum Lindley. As reported by Briggs & Johnson (1979: 187) “the second-order and higher- order axes may produce: a) both terminal and lateral paniculate shoots . . . or b) lateral pa- niculate shoots only, the apical bud retaining Myrtaceae. Syzygium thumra; flowering branch. (From Wight, 1843.) the capacity for continued growth.” Figure 75 I, which illustrates the results of Briggs & Johnson, resembles Figure 75 II, reporting our own results. The somewhat contradictory examples of inflorescence structure, however, can be elucidated in an acceptable way as being derived from a monotelic thyrsoid-pa- niculate base type. In D. parviflorum Kurz, we observed an indeterminate main axis bear- ing axillary loosely branched monothyrsoids with slightly disperse phyllotaxy. Figure 75 I and II also represents examples of the development of accessory branches. Within the Myrtaceae these accessory branches are mostly phylloscopic. They occur as single accessory flowers (Fig. 75 I, II), triadic branches (Fig. 75 I), or even as bo- tryoids (Fig. 75 I) and perhaps also as thyr- soids. Sometimes several accessory axes in a 294 Annals of the Missouri Botanical Garden Me yild. ;. ft YUCCA COMET ORLAKR EL = — ` | E Pumpliy Luth 7 - / A 4 , ( ////# JHA / gl ue FALE Artt FIGURE 70. Myrtaceae. Syzygium acuminatum; flowering branch. (From Wight, 1843.) vertical series are developed in the same axils. In this case they may be different in their degree of ramification or not, both demon- strated in Myrceugenia exsucca (DC.) Berg (Fig. 80 IX). Within the large genus Eucalyptus, ter- minal thyrsoid-paniculate inflorescences end- ing in a terminal flower as well as proliferating inflorescences can be observed. Although some sections of the genus may exclusively follow Volume 75, Number 1 1988 Weberling Inflorescences in the Myrtales « STN S ü AN FIGURE 71. fusa.—IV. M. perforata.— ermadecensis.— VT, Myrtaceae. Inflorescence diagrams.—I. Metrosideros albiflora. VII M. umbellata. — Ce TS eS pe S —Al. M. carminea.—//l. M. dif- VIII. Rhodomyrtus tomentosa.—/X. Acca sellowiana. (I-VI from Dawson (1968) with some graphical alterations; bracts were ommited by Dawson.) one or the other mode, even closely related species can be different in this respect. Thus in many species of sect. ddnataria of subg. Symphyomyrtus (Pryor & Johnson, 1971), the main axes of the inflorescences were found to have terminal flowers (E. melliodora A. E. paniculata Sm. (Fig. 76 1), E. polycarpa F. Muell., E. polyanthemos Schau., E. populnea F. Muell., E. Viii Schau. (Fig. 76 II), while others (e.g., sideroxylon A. Cunn.) have a A main axis. In still others (e.g., E. intertexta Cunn., R. T. Baker) inflorescences with terminal flowers and proliferating inflorescences were even found on the same herbarium sheet (Penfold & Willis, 1961, pl. 38, fig. 6). Ac- cording to Johnson (1972: 23, fig. 5) this form “has adaptive significance" and is de- rived from forms without a terminal flower (the reverse seems more probable). The partial inflorescences are condensed cymes (Fig. 75 III, IV), which often are de- scribed as umbels but more correctly should be called umbellasters (Johnson, 1972; Briggs 296 Annals of the Missouri Botanical Garden FIGURE 72. Right. Callistemon rigidus; proliferating shoot after anthesis, with fruits of preceding flowering phase. (From Troll.) — Johnson, 1979) or sciadioids (Troll. 1964: 53). The number of flowers in such umbellasters can be seven or more, but often their number is reduced, and they are not rarely uniflorous (Fucalyptus globulus Labill., E. tetraptera Turcz.). They can show significant charac- ters, such as concrescence of their gynoecia in E. lehmannii (Schau.) Benth. or broad- ening of the hypopods, especially in E. pla- typus Hook. (Fig. 77). In E. cinerea F. Muell. (Fig. 73), E. macrorhyncha F. Muell., and other species with proliferating inflorescences, a strong delay of anthesis as already noted for Beaufortia decussata takes place. As a consequence the top of the inflorescence axis may produce vegetative branches while the partial inflorescences at its base are still in flower. The connection of these conditions with the formation of cataleptic (Müller-Dob- lies & Weberling, 1984) flower-bearing short Myrtaceae. I, Il. Melaleuca decora; inflorescence before (left) and after (center) proliferation. — shoots and with cauliflory seems to be obvious here. Frequently the flower-bearing branches are reduced to botryoids, as in Myrciaria flori- bunda (Willd.) Berg (Cremers, 1983/1986, fig. 42/7) or Eugenia guatemalensis Cogn. In species of Blepharocalyx (B. tweedii Hook. & Arn., B. gigantea Lillo, cf. Cabrera, 1978, fig. 143; Digilio & Legname, 1966, fig. 82) or Backhousia myrtifolia Hook. & Harvey, the paracladia are long-stalked and loosely branched, seven- to five-flowered cymes. This reduction of the paracladia goes even further to form long-stalked, loosely branched triads, as in Myrcianthes cispla- tensis (Camb.) Berg, Myrceugenella apicu- lata (DC.) Kausel, or Eugenia pseudo-mato Legr. (Digilio & Legname, 1966, figs. 84, 87) or to long-stalked, biflorous or uniflorous (monadic or dyadic) paracladia still with pro- phylls (Eugenia pungens Berg; Digilio & 297 Weberling Volume 75, Number 1 1988 Inflorescences in the Myrtales A el eus > &Z/) an? ii Ms UNA AG | ARS ET KATA S UMED R N A 77 ZN ANN 4 77 LEAN NE UNA SOL MPa. | Zz, NIG Z 7 (Up, ; RK É Ha INR Ne PE INTA AS. UWA Sd Àj NA 7 SSO eue 1 SEA COO, DAN QNS E is =>" LAM WI AS ERES ^D jo: S A e ESOS VS y » rin Z yrtaceae.—I. Eucalyptus cinerea; diagram of a flowering shoot. The proliferating inflorescence Y M FIGURE 73. axis already had developed branches when anthesis had started (Weberling 7413). — II. Beaufortia decussata; flowering shoot with fruits of preceding flowering phases. 298 Annals of the Missouri Botanical Garden FIGURE The inflorescence axis still ends in a bud. (From Troll. 74. Myrtaceae. Pimenta dioica; inflorescences.— Left. Proliferation has taken place already. — Right. Troll.) Volume 75, Number 1 1988 Weberlin g 299 Inflorescences in the Myrtales FiGURE 76. Legname, 1966, fig. 85). Condensed and ses- sile triadic or uniflorous paracladia compose the spikelike proliferating inflorescences so characteristic of many species of Calotham- nus, Callistemon (triadic pc in C. suberosum only; Dawson, 1978a, fig. 3), Melaleuca (Fig. 72), and Beaufortia (Fig. 73 II). Leafy proliferating botrya are character- istic for all or many species of the myrtoid genera Ugni, Myrtus, Fenzlia, Psidium, Myrceugenia, Luma, Acca (Fig. 71 IX), and others and of the leptospermoid genera Er- emaea, Baeckea, Balaustion, Calythrix, Wehlia, Pileanthus, Thryptomene, Micro- myrtus (Fig. 78 III, IV), Verticordia (Fig. Myrtaceae. —1. Eucalyptus paniculata.—//. E. pruinosa; flowering branches. (From F. Mueller.) 78 I, II), Regelia (also triadic pc?, cf. Briggs & Johnson, 1979), and others. In many species of Kunzea (e.g., K. cap- itata Reichb. and K. ericifolia Reichb., Fig. 79 I), the botrytic or spikelike inflorescences are capitate, as in species of Agonis (Fig. 80 II) and Sinoga. The agglomeration of flowers into showy heads decidedly supports the attraction of pol- linators. This is especially true for the pseu- danthial inflorescences of Darwinia and Ac- tinodium. In both, the dense heads are botrya that can proliferate after flowering (especially in Darwinia, Fig. 81). In Darwinia the sub- tending leaves of the flowers are inconspic- — FIGURE 75. Eucalyptus intertexta.— //]. In Oct. 1903, J. L. Boorman s.n. (M)). proliferating. —lV. Myrtaceae. I, Il. Decaspermum paniculatum.—/. Flexibility elie ord ir od (from Briggs & Johnson; showing the actual branching of a herbarium specimen: Cas h Diagram of a partial inflorescence (“umbellaster”). a, apex of blastotelic 1 main axis; a.tr, accessory triadic branch; ab, accessory botryoid; af, accessory flower. (IV From Troll. SIAATS HERBA. UN MUNCHEN EUR S hens am VY Myrtaceae. Eucalyptus platypus.— Left. Herbarium specimen. — Right. Detail. (Algeria: Oran à Santa Cruz, 9.5.1918, A. Faure s.n. (M).) dX ^n w » ] ier e AE e 2 > A, Volume 75, Number 1 1988 Weberling Inflorescences in the Myrtales MA FIGURE 79. branch. (ll From Troll.) uous, but the adjacent leaves are often greatly enlarged, petaloid, and colored, giving the capitate inflorescence a flowerlike shape (Fig. 82). For the western Australian species, Briggs & Johnson (1979: 201) found “a sequence from (for example) the little-modified D. thy- moides Benth., through D. virescens (Meissn.) Benth., where the petaloid leaves are about as long as the perigynia (‘floral tubes’), to the pendulous ‘bells’... of D. meeboldii C. A. ardn. (‘Mondurup Bell’) or D. speciosa (Meissn.) Benth., in which individual flowers are hidden by several series of long petaloid and the inflorescence axis "is an- ” (Figs. 80 IV, 82). In Actinodium leaves," auxotelic' \ II Myrtaceae.—1. Kunzea ericifolia (Meebold 11881).—J//. Chamelaucium uncinatum; flowering cunninghamii Schau. ex Lindley, the “daisy- like" flower heads have several rows of “rays” formed by modified sterile flowers (Figs. 80 V, 82). The axis of the capitulum “is some- what swollen but not flat, and variation in the length of the peduncles of the monads brings the flowers to approximately the same level, although the outermost arise c. 2-5 mm be- low the innermost” (Briggs € Johnson, 1979: 202). It is not rare that the development of triad- ic to uniflorous paracladia and the arrange- ment in leafy thyrsic, botrytic or spikelike proliferating inflorescences is limited to branches of more or less long main shoots FIGURE 78. inflorescence axis has returned to vegetative grow pa ralia, F. Mueller s.n. DJ II, IV. Tien e microphylla. —Ill. Flowering branching system.—IV. ne eae.—1. Verticordia pholidophylla. —Il. V. spicata; flowering branches. The apex of the (L Watheroo, Meebold ipo (M); II. W. Headlike Australia: condensed inflorescence with beginning proliferation (L. Boorman in 9.1908, New South Wales, Australia; M). 302 Annals of the Missouri Botanical Garden FIGURE 80. Myrtaceae. Inflorescence diagram Albany, A. Meebold 10140, (M) ). boldii.—V. Actinodium cunninghamii.— remain undeveloped.— VIII II-V from Briggs «€ PRAE VII — IIl. Darwinia with pet 1 Ch JH

i=] a [md E I mm; pyrenis trigonis 4 mm longis dorsaliter laevibus. Shrub 1.5 m tall. Branchlets glabrous. Leaves coriaceous, shining above, obovate- oblong or oblong-ovate, rounded, obtuse or subobtuse at apex, obtuse to subacute at base, 5-9.5 cm long, 3.5-5 cm wide, entire, gla- brous, punctate below, slightly recurved on margins; midrib subelevated, subimpressed above, 1 mm wide; lateral nerves obsolete or impressed above, 6-11 each side, forking before reaching margins, not manifestly anas- tomosing. Petioles 9-13 mm long, glabrous. Pistillate inflorescence axillary and lateral, solitary or fasciculate, fruiting peduncle 1 mm ong, puberulent; fruiting pedicels 1-2 on a peduncle, 2-5 mm long, puberulent. Calyx minutely puberulent, the lobes in fruit subor- bicular, rounded, 1-1.5 mm long, 2 mm wide, ciliolate. Fruit globose, 6 X 6 mm. Pyrenes trigonous, 4 mm long, dorsally smooth above, without ridges. This taxon resembles /lex tepuiana Edwin in leaf shape but differs in the dorsally smooth pyrenes and much longer petioles. From /. solida Edwin it is differentiated by the much shorter fruiting pedicels and much longer pet- ioles, while from /. fanshawei Edwin it may be distinguished by the dorsally smooth achenes and rounded leaf apex. Ilex guaiquinimae Steyermark, sp. nov. TYPE: Venezuela. Bolivar: Cerro Guai- quinima, southeast-central part of the summit, 5%44'4”N, 63?41'8"W, 730- 900 m, 24 May 1978, Julian A. Stey- ermark, Paul Berry, G. C. K. & E. Dunsterville 117276 (holotype, VEN; isotype, MO). 326 Annals of the Missouri Botanical Garden Arbor, ramulis glabris; foliis late oblongis apice late nervis lateralibus utroque latere 6-7 nal 30- 45? i longis glabris; inflorescentia foeminea solitaria axillari la- teralique pedunculo sub fructu 3-5 mm longo glabro; pedicellis 3 mm longis glabris; calyce glabro, lobis (ii suborbicularibus rotundatis; bacca globosa 5 mm lon 5.5 mm diam.; pyrenis 5 mm longis dorsaliter late uni- sulca Tree with glabrous branchlets. Leaves broadly oblong, broadly rounded at apex, acute to subacute at base, 10-16 cm long, 6.5-8 cm wide, entire, glabrous, punctate below; midrib 2 mm wide, shallowly and widely sul- cate above, conspicuously elevated below; lat- eral nerves slender, side, ascending at an angle of 30—45?, 1.5-1.7 cm apart, impressed or slightly elevated above, slightly elevated below, branching before reaching margin; tertiary venation slightly evident above, obscure below. Petioles 10-18 mm long, glabrous. Pistillate inflorescence soli- tary, axillary, lateral. Fruiting peduncle 3-5 mm long, glabrous; pedicels 3 mm long, gla- brous. Calyx glabrous, the lobes shallowly sub- orbicular, rounded. Fruit globose, 5 mm long, 5.5 mm diam. Pyrenes 5 mm long, 3 mm wide, dorsally broadly 1-sulcate. Closely related to [lex jenmanii Loes., from which it differs in the fewer lateral nerves more distantly separated (15-17 mm vs. 4- 10 mm in /. jenmanii), broadly oblong leaves rounded at the apex, punctate lower leaf sur- face, shallowly and widely sulcate upper mid- rib, and solitary, axillary, and lateral inflo- rescence. From /. marginata Edwin it is distinguished by the ascending lateral nerves of the leaf blades. Ilex holstii Steyermark, sp. nov. TYPE: Ven- ezuela. Territorio Federal Amazonas: Dept. Atabapo: Cerro Marahuaca, “Sima Camp," south-central portion of forested slopes along east branch of Cano Negro, 3°43'N, 65?*31'W, 1,140 m, 21-22, 24 Feb. 1985, Julian A. Steyermark & Bruce Holst 130481 (holotype, MO; iso- type, VEN) Arbor 4-metralis, ramulis ue foliis ile apice acuminatis basi acutis vel subacutis 5.5-8 cm longis 1.3- 2.5 cm latis integerrimis a tris. subtus ut “ma pens s lateralibus angulo 5-20*; petiolis 6-10 mm longis glabris; inflorescentia foeminea solitaria axillari la- teralique, pedunculo 2-7 mm longo gla abro vel absenti; pedicellis buds solitariis vel duo 0. mm latis; ca aa gla manifestis truncatis; bac x 4-6 mm; E poeci 4, trigonis 4 mm long DER e 1-sulcat Tree 4 m tall with glabrous branchlets. Leaves subcoriaceous, lanceolate, acuminate at apex, with an attenuate mucro 0.5 mm long, acute or subacute at base, 5.5-8 cm long, 1.3-2.5 cm wide, 2.7-4 times longer than broad, entire, glabrous, the midnerve shallowly sulcate above, slightly elevated be- low; lateral nerves obsolete above, manifestly impressed below, subhorizontally spreading at an angle of 5-20", tertiary venation faintly reticulate below. Petioles 6-10 mm long, gla- brous. Pistillate inflorescence solitary, axil- lary, lateral. Fruiting peduncle when present -7 mm long, glabrous; fruiting pedicels sol- itary or 2 together, slender, 4-7 mm long, glabrous. Calyx in fruit glabrous, 2.5 mm diam., the lobes scarcely evident, truncate. Fruit globose, 4-6 x 4-6 mm. Pyrenes 4, trigonous, 4 mm long, 2-2.5 mm wide, broad- ly unisulcate dorsally. This species is similar to Ilex oliveriana Loes. in unisulcate pyrenes, leaf shape and mucronate apex, and the size of the fruit; but it differs in the longer petioles, more promi- nently impressed, subhorizontal lateral nerves, and longer mucros of the leaf blades. From I. macarenensis it differs principally in the dorsally unisulcate achenes. Ilex ignicola Steyermark, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: Dept. Rio Negro, Cerro Aratitiyope, 70 km south-southwest of Ocamo, 2?10'N, 65%34'W, 990-1,100 m, 24-28 Feb. 1984, Julian A. Steyermark, Paul Ber- ry & Francisco Delascio 130112 (ho- lotype, MO; isotype, VEN). etralis, ramulis novellis minute puberulis vel Arbor 6-m i gine foliis 'elliptico- oblongis apice anguste obtusis vel cutis basi acutis vel subacutis 5.5-8 cm longis 2.5- Volume 75, Number 1 1988 Steyermark 327 Flora of Venezuelan Guayana—lV 3.5(-4) cm latis integerrimis glabris; costa superne sub- elevata vel leviter sulcata plerumque puberula subtus im- pressa vel elevata plerumque puberula; nervis lateralibus utroque latere 7-9; petiolis 5-9 mm longis minute pu- berulis glabrescentibus; inflorescentia foeminea solitaria axillari lateralique, pedicellis 3-4 mm longis minute puberulis; calyce sub fructu puberulo, lobis leviter 0; pyrenis quattuor trigonis 4-5 mm longis dorsaliter Le 1 -sulcatis. Tree 6 m tall; young stems minutely pu- berulent, glabrescent. Leaves subcoriaceous, elliptic-oblong, narrowly obtuse or subacute at apex, acute to subacute at base, 5.5-8 cm long, 2.5-3.5(-4) cm wide, entire, glabrous, subrevolute; midnerve subelevated or shal- lowly sulcate above, puberulous, impressed or elevated below, mainly puberulous; lateral nerves 7—9 each side, inconspicuous, scarcely evident below, impressed above, ascending at a 45? angle; tertiary venation obsolete above, scarcely evident beneath. Petioles 5-9 mm long, minutely puberulous or glabrescent. Pis- tillate inflorescence solitary, axillary, lateral. Pedicels 3-4 mm long, minutely puberulent. Calyx puberulous in fruit, the lobes shallowly suborbicular, rounded, 1 mm wide, puberu- lent. Fruit subobovoid, 7-8 mm long, 5 mm broad. Pyrenes 4, trigonous, 4-5 mm long, 2 mm wide, dorsally broadly unisulcate, the sulcation sharply angled. This species is marked by the usually mi- nutely puberulent upper and lower midnerves, the puberulent pedicels and calyx, and the slightly longer than broad subobovoid fruits with unisulcate pyrenes. Ilex jauaensis Steyermark, sp. nov. TYPE: Venezuela. Bolivar: Meseta de Jaua, Cer- ro Jaua, southwest summit, 4?48'50"N, 64?34'10"W, gallery forest along trib- utary of Rio Marajano, 1,750-1,800 m 22-28 Feb. 1974, J. A. Steyermark, V. C. Espinoza & C. Brewer-Carías 109294 (holotype, VEN; isotype, MQ). Arbor 4- sites depu m foliis oblongo- obovatis apice rotundatis retusis basi acutis vel obtusis 4-5.5 cm longis 2-3 cm latis potes bns subtus punctatis; solitaria axillari lateralique; pedicellis sub fructu 13-14 mm longis glabris; n er fructu glabro, lobis leviter suborbicularibus 2 mm latis; bacca in sicco 7 mm longo, mm diam., in vivo 9 mm a vn 10 mm diam.; pyrenis trigonis 4.5- 6m m longis dorsaliter late unisulentis cristis marginalibus prominentibus Tree 4 m tall with glabrous branchlets. Leaves coriaceous, oblong-obovate, mainly rounded at the retuse apex, acute to obtuse at base, 4-5.5 cm long, 2-3 cm wide, entire, glabrous, punctate below; lateral nerves 5 each side, narrowly sulcate above, slightly manifest and subelevated below; tertiary ve- nation not evident. Petioles 5 mm long, gla- brous. Pistillate inflorescence solitary, 1-flow- ered, axillary, lateral. Fruiting pedicels 13- 14 mm long, glabrous. Calyx glabrous in fruit, the lobes shallowly suborbicular, 2 mm wide. Fruit 7 mm long, 8 mm wide (dried state), 9 mm long, 10 mm wide (living state). Pyrenes trigonous, 4.5-6 mm long, 3-4 mm wide, with prominent marginal ridges. Paratype. VENEZUELA. BOLÍVAR: Cerro Marutani, 1,420 m, Steyermark et al. 124032 (VEN). The retuse, mainly rounded leaf apex dis- tinguishes this taxon from /lex sulcata Edwin and /. tiricae Edwin; the longer petioles, shorter leaves, and oblong-obovate leaf shape differentiate it from /. lasseri Edwin; the lon- ger petioles, shorter pistillate pedicels, and obtuse to acute leaf base distinguish it from I. solida; and the longer and less thickened petioles differentiate the new species from /. tepuiana Edwin. Ilex longipilosa Steyermark, sp. nov. TYPE: Venezuela. Bolivar: 3 km south of El diu non-forested south slope and sum- of El Abismo, 4?30'N, 61?35'W, oo osc m, 11 Nov. 1985, Ronald Liesner 19878 (holotype, MO; isotype, VEN). Figure 6A. nin 1.5-metralis, ramulis dense hirtellis pilis usque m longis instructis; folis cuneiformi-obovatis vel Rb dom ovatis apice rotundatis abrupte apiculatis basi cuneatis 4-7 cm longis 2.5-5.5 cm latis plerumque in- ui vel raro 1- e denticulatis subtus costa valde pilosa pilis 0.3-0.9 mm munitis; nervis lateralibus subtus pilo- sulis pilis e nds alibi subtus sparse pilosulis vel glabratis, subtus minute punctatis; nervis lateralibus utroque latere 6-8 angulo 10-20? divaricatis; venulis Annals of the Missouri Botanical Garden tertiariis subtus grosse reticulatis; petiolis 3-7 mm longis fl dense hirtellis pilis patentibus; inflorescentia foeminea s litaria axillari lateralique 10-20 mm longo ramosa 5-6 1 ngis praedita; pedu o 4-9 mm longo puberulo; pedicellis bracteis deltoideo-lanceolatis acutis vel acuminatis l. 2 mm longis 0.5 mm latis glabris vel extus basi floribus 4-meris, calycis us vel paullo subacutis 1.3 m erosis; petalis ios E. 3 mm longis 2 mm latis Shrub 1.5 m tall, the branchlets densely hirtellous with spreading hairs up to 0.5 mm long. Leaves cuneiform-obovate or suborbic- ular-ovate, abruptly apiculate at the rounded apex, cuneate at base, 4-7 cm long, 2.5-5.5 cm wide, mainly entire or rarely with 1-2 minute toothlike projections, subrevolute, the lower midrib strongly pilose with spreading hairs 0.3-0.9 mm long, the lateral nerves beneath with spreading hairs, the lower sur- face pilosulous to glabrate, the upper surface glabrous except for the minutely puberulent sulcate midrib, minutely punctate on lower surface; midnerve sulcate above, elevated be- low; lateral nerves 6-8 each side, divaricately spreading at a 10—20? angle, slightly manifest below, inconspicuous and faintly sulcate above. Petioles 3-7 mm long, densely hirtellous with spreading hairs. Pistillate inflorescence soli- tary, axillary, lateral, 10-20 mm long, cy- mosely or paniculately branched with 5-6 lateral axes 4-10 mm long; peduncle 4-9 mm long, shortly puberulous; pedicels 1.5-2 mm long, shortly puberulous. Bracts deltoid- lanceolate, acute to acuminate, 1.2 mm long, 0.5 mm wide, mainly glabrous or with a few hairs at the outer base. Flowers 4-merous; calyx lobes suborbicular, rounded or slightly subacute, m wide, the margins slightly irregularly erose. Petals obovate-oblong, rounded, 3 mm long, 2 mm wide. From the taxa placed by Loesener (1901) in his section 7 Megalae, subsection Pedi- cellata Loes., especially /lex villosula Mar- tius, Í. velutina C. Martius, I. brasiliensis (Sprengel) Loes., and /. theezans var. riedelii Loes., the new species differs in its solitary, axillary, and lateral (vs. fasciculate) inflores- cences, punctate lower leaf surface, and dif- ferently shaped leaves. The longer pubes- cence of the new species is somewhat reminiscent of /. maguirei Wurdack of Cerro a Neblina, which differs in its rounded leaf bases, differently shaped leaves, shorter petioles, pubescent, elongated fruit, and only -3-flowered inflorescences. Ilex magnifructa Edwin, var. minor Stey- ermark, var. nov. TYPE: Venezuela. Ter- ritorio Federal Amazonas: Cerro Uali- pano, orillas del Rio Parucito, 65%43"W, 1,760 m, Feb. 1962, DM Cardona 2943 (holotype, US; isotype, MO) var. magnifructa pedicellis fructiferis 5-10 mm longis recedit. Leaf blades broadly oblong-obovate, round- ed at apex, broadly rounded or obtuse at base, 7.5-10.5 cm long, 4-8 cm wide. Petioles 2- 3 mm long. Calyx lobes 4, deltoid, acute. Fruiting pedicels 5-10 mm long. This variety differs from var. magnifructa in the shorter fruiting pedicels. Ilex marahuacae Steyermark, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: Cerro Marahuaca, “Sima amp,” south-central portion, forested slopes along E branch of Caño Negro, 3°43'N. 65°31'W, 1,140 m, 21-22, 24 Feb. 1985, J. A. Steyermark & B. K. Holst 130425 (holotype, MO; isotype, VEN) Arbor 4-metralis, ramulis glabris; foliorum laminis ova- erosis a tag pyrenis 4 trigonis dorsaliter laevibus vel sublaevi Tree 4 m tall, the stems glabrous. Leaf blades subcoriaceous, ovate or oblong-ovate, obtusely to acutely acuminate at apex, the acumen 8-10 mm long, rounded to subobtuse Volume 75, Number 1 Steyermark 329 1988 Flora of Venezuelan Guayana—IV FIGURE 6.— A. Ilex longipilosa, portion of flowering branch, staminate. — B. Ilex acutidenticulata. — b. Branched fruiting peduncles. — C. Ilex spathulata. — D. Ilex brevipedicellata. — E. Ilex summa. — F. Ilex wurdackiana, staminate branch.—G. Ilex wurdackiana, portion of fruiting branch. at base, 6.5-11 cm long, 2.5-4.5 cm wide, pressed above, slightly elevated below; ter- punctate, glabrous; lateral nerves 4-5 each tiary venation reticulate and rather conspic- side with less conspicuous intermediate nerves, | uously impressed above, subobsolete below. subhorizontally spreading at 10-20?, im- Petioles 6-8 mm long, glabrous. Infructes- 330 Annals of the Missouri Botanical Garden cence solitary, axillary. Peduncle 5-6 mm long, 2-3-flowered, glabrous. Pedicels 3 mm long, glabrous. Flowers 4-merous. Calyx lobes shallowly suborbicular, rounded, 0.3 mm long, 1 mm wide, minutely erose, sparsely ciliolate. Calyx tube glabrous. Fruit globose, 5 mm diam. Pyrenes 4, trigonous, 5 mm long, 3 mm wide, dorsally smooth or essentially so. This taxon is characterized by the elon- gated acumen of the obscurely crenulate-ser- rulate leaves, the solitary axillary inflores- cence, the dorsally smooth pyrenes, and the epunctate leaves. Ilex paruensis Steyermark, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: Dept. Atabapo, Serrania del Parü (Aro- ko), sector centro-sur, sabanas herbáceas en el borde sur de la tercera meseta central, 4%27'N, 65%32'W, 1,100 m, 7 Oct. 1979, O. Huber 4437 (holotype, VEN) x 2-metralis, ramulis pubescentibus pilis arcte ver- CN seriatis cupo foliorum laminis late obo- did arcte negro- punctatis scrobiculatisque, subtus costa a pubescenti pilis brevibus patentibus munita supra sulcata minute puberulenti; petiolis 2 mm longis crebiter puberulis; infructescentia foeminea solitaria quam longiori; fructu subgloboso 3 mm diam. (immaturo). Shrub 2 m tall; stems minutely pubescent in vertical lines. Leaf blades coriaceous, broadly obovate, rounded at the minutely mu- cronulate apex, acute at the base, 2.5-3.5 cm long, 1.5-2.3 cm wide, epunctate below, scrobiculate and densely black-puncticulate above, the midrib puberulent below with spreading hairs, minutely puberulent in the sulcation above, elsewhere glabrous on both surfaces or sparsely puberulent near base of upper surface, revolute; lateral nerves 4-5 each side, inconspicuous both sides, faintly impressed below. Petioles 2 mm long, densely and minutely puberulent. Infructescence sol- itary or fasciculate, axillary and lateral. Pe- duncles 1-3-flowered, up to 3 mm long, densely puberulent. Pedicels 1-3 mm long, densely puberulent. Calyx lobes scarcely de- veloped, broadly shallowly repand-rounded, slightly apiculate at the apex, glabrous. Stig- ma broadly flattened and depressed at the slightly umbonate summit. Immature fruit subglobose, 3 mm diam This species is related to /. vacciniifolia Klotzsch, from which it differs in the epunc- tate lower leaf surface, strongly black punc- ticulate upper surface, shorter fruiting pe- duncles and pedicels, mucronulate leaf apex, and glabrous, subobsolete calyx lobes. Ilex paujiensis Steyermark, sp. nov. TYPE: Venezuela. Estado Bolivar: 3 km south of El Pauji, 4?30'N, 61?35'W, 900 m, 19 Oct. 1985, Ronald Liesner & Bruce Holst 18796 (holotype, MO; isotype, Arbor 4- metralis, a glabris; m ee ines sis basi cuneatis 6.5— ongis 2.5-4.5 cm latis integerrimis ram subtus va Y» e negro punctatis; costa ^y qudd angust rape nervis Berai -9; ru tertiariis is 5-10 mm dion s 1.5-2 mm Fm glab ntia foeminea solaria e lateralique; pedicellis fructiferis 2-3 m s ute puberulentibus; calyce minute A lobis iE us rotundatis d liolatis vel obsolete erosis 1 mm latis; petalis 3-3.5 n longis 1.5 mm latis ds p aio 5i bis ciliolatis; bacca globoso Tree 4 m tall with glabrous branchlets. Leaves coriaceous, obovate, rounded an sometimes retuse at apex, cuneately narrowed at base, mainly 6.5-11 cm long, 2.5-4.5 cm wide, 2-21 times longer than broad, entire, glabrous, strongly black-punctate beneath, the upper midnerve narrowly sulcate; lateral nerves 7-9 each side, slightly elevated or impressed above, finely impressed below; ter- tiary venation faintly manifest beneath. Pet- ioles 5-10 mm long, 1.5-2 mm wide, gla- brous. Pistillate inflorescence solitary, axillary, ateral. Fruiting pedicels 2-3 mm long, mi- nutely puberulent. Calyx tube minutely pu- berulent; calyx lobes suborbicular, rounded, some of them minutely ciliolate, others ob- soletely erose, 1 mm wide. Petals 3-3.5 mm Volume 75, Number 1 1988 Steyermark 331 Flora of Venezuelan Guayana—lV long, 1.5 mm wide, minutely papillate on the lower half of the upper surface, the margins papillate-ciliolate. Fruit globose, 6 x 6 mm. Paratype. VENEZUELA. BOLIVAR: region of rios Ica- bara and Hacha, 450-850 m, Bernardi 2622 (VEN). The elongate petioles and short fruiting pedicels ally this taxon to [lex gransabanen- sis Steyerm., but /. paujiensis differs in hav- ing the punctations on the lower leaf surface larger and more conspicuous, less conspicu- ous lateral nerves, more cuneately acute leaf bases, and narrowly sulcate upper midnerves. The collection of Bernardi 2622 at VEN, presently treated as a paratype of /lex pau- jiensis, was identified by Edwin as /. andar- ensis Loes. and forms the basis for his inclu- sion of that species in his treatment for the Venezuelan Guayana (1965). [lex andaren- sis, originally described from Andara Moun- tain, Peru, was based on a Spruce s.n. col- lection and is completely different from the Bernardi specimen, having, amon other differences, ovate, oval- bone or ellip- tic leaf blades which are obtuse, rounded, or cuneately obtuse at the base and acute or obtusely acuminate at the apex with an acu- men 7-12 mm long. Ilex polita Steyermark, sp. nov. TYPE: Ven- ezuela. Bolivar: Guayaraca, southern base, Auyán-tepui, between escarpment and Rio Guayaraca, 5?44'N, 62%32'W, 950 m, 25-27 Nov. 1982, G. Davidse & O. Huber 22679 (holotype, MO; iso- type, VEN). Frutex 2-metralis, ramulis glabris; foliis superne nitidis ovatis = ovalibus Le he dis e E pud piens vel rotundat longis i sical suborbicularibus apice rotundatis vel subacutis 1.2-1.5 mm latis marginibus ciliolatis. Shrub 2 m tall with glabrous branchlets. Leaves coriaceous, shining and mainly ener- vate above, ovate, oval, oblong, or elliptic- oblong, obtuse or rounded at apex, obtuse or rounded at base, 4-7.5 cm long, 1.8-4.5 cm wide, entire, glabrous, epunctate below, 1.3- 2.6 times longer than broad, revolute; mid- nerve narrowly and shallowly sulcate above, somewhat elevated below; lateral nerves 5- 9 each side, ascending at an angle of 15- 20°, obsolete above, faintly impressed or slightly elevated below; tertiary venation slightly manifest below. Petioles 4-10 mm long, glabrous. Pistillate inflorescence soli- tary, axillary, lateral. Peduncle 7-10 mm long in fruit. Pedicels, when solitary, 3-5 mm long in fruit, 1-2 mm long in fruit in cymose inflorescences, sparsely puberulent. Flowers 4-merous. Calyx tube sparsely and minutely puberulent, the lobes suborbicular, rounded or slightly subacute at apex, 0.5 mm long, l mm wide, + densely ciliolate. Paratypes. VENEZUELA. BOLÍVAR: Auyán-tepui, Guayaraca, 950 m, Davidse & Huber 22733 (MO, VEN); m S of SW corner of Amaruay-tepui, 5°54'N, 62?15' W, 500 m, Liesner & Holst 20114 (MO, VEN). This species is characterized by the com- bination of its densely ciliolate calyx lobes, shallowly sulcate upper midnerve, and oblong, elliptic-oblong, ovate, or oval leaf blades lus- trous on the upper surface and obtuse or rounded at the apex. Ilex spathulata Steyermark, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: Dept. Atabapo, Cerro Marahuaca, slopes upstream from Río Yameduaka, 3?38'N, 65°28'W, 1,225 m, 17-18 Feb. 1985, Ronald Liesner 17599 (holotype, MO; isotype, VEN). Figure 6C. ex 2-metralis; ramulis glabris; foliis anguste spa- glabris subtus pu lteralibus principe s de. y. vel fere; petioli m longis glabris; inflorescentia fosa Md ide, 1-2- vides pedicellis su ub anthesi 4-4.5 mm longis sub fructu mm lo puberulentibus; calyce sub anthesi ato glabro, lobis akar odaia rotundatis ralique cymosa 2-5-flora, pedunculo EE. pe 2 mm longis; bacca diia) globosa 332 Annals of the Missouri Botanical Garden Shrub 2 m tall, the branchlets glabrous. Leaves subcoriaceous, narrowly spathulate, rounded and emarginate at apex with a thick- ened blunt tip in the sinus, conspicuously nar- rowed at the base and decurrent, 2.5-4.5 cm long, 0.5-1.2 cm wide, (214-)414-5 times longer than broad, entire, glabrous, punctate beneath, revolute; midnerve sulcate above; principal lateral nerves 5-6 each side, im- pressed above, slightly elevated below; ter- tiary venation obsolete or inconspicuous above, slightly more conspicuous below. Petioles 5- 8 mm long, glabrous. Pistillate inflorescence solitary, axillary, lateral, 1-2-flowered. Ped- icels in flower 4-4.5 mm long, in fruiting specimens 10-12 mm long, minutely puber- ulent. Calyx 0.7-1 mm long, 1.5 mm wide in anthesis, glabrous; lobes suborbicular, rounded, 0.5 mm long, 0.8 mm wide. Petals 2-2.5 mm long, 1 mm wide. Staminate in- florescence fasciculate or solitary, axillary and lateral, cymosely 2—5-flowered, pedunculate; peduncle 1.5 mm long, puberulent; pedicels 2 mm long. Immature fruit globose, 6 mm x 5.5 mm. aratype. Nei A. TERRITORIO FEDERAL AMA- ZONAS: Cerro Marahuaca, same locality as type, 19 Feb. 1985, Liesner N 7688 (MO, VEN). This taxon is characterized by the narrowly spathulate leaves, mainly 44-5 times longer than broad and the completely glabrous calyx. From /lex huachamacariana Edwin it may be distinguished by having shorter fruiting pedicels, more conspicuous lateral nerves, and more conspicuous tertiary venation on the lower leaf surface. From /. gleasoniana Stey- erm. it differs in the much narrower and thin- ner leaf blades and in the glabrous calyx. Ilex summa Steyermark, sp. nov. TYPE: Venezuela. Bolivar: Dist. marcaibarai-tepui, | summit, 61°59'W, 2,400 m, 26 Mar. Bruce K. Holst 3617 (holotype, MO: isotype, VEN). Figure 6E. Frutex 2-metralis, ramulis glabris; foliis ovatis apic l apiculatis vel subacutis basi rotundatis vel lat 1 2.5-4.3 cm longis 1.5-3 cm latis, marginibus be tertiis dentibus 3-5 acutis vel a praeditis glabris ubique punctatis valde revolutis; nervis lateralibus utroque late catis subtus elevatis pite venulis tertiariis supra obsoletis subtus grosse reticulatis; petiolis 3-6 mm longis glabris; inflorescentia inea solitaria axillari latera- fruc nm longis glabris; bacca globoso diam.; pyrenis 4 trigonis 4-4.5 mm longis dorsaliter un- isulcatis Shrub 2 m tall with glabrous branchlets. Leaves coriaceous, ovate, apiculate or sub- acute at apex, rounded or broadly obtuse at base, 2.5-4.3 cm long, 1.5-3 cm wide, 1.4- 1.6 times longer than broad, the upper !4 of the margins with 3—5 acute to setulose teeth, glabrous, black-punctate on upper and lower surfaces, the margins strongly revolute; mid- nerve sulcate above with minute puberulence in sulcation, elevated below; lateral nerves 4— 5 each side, prominently sulcate above, ele- vated below, spreading divaricately, dichot- omously branched 2-4 mm from the margins; tertiary venation obsolete above, grossly re- ticulate below. Petioles 3-6 mm long, gla- brous. Pistillate inflorescence solitary, axil- lary, lateral. Peduncles in fruit 4-6 mm long, glabrous. Pedicels in fruit 4-9 mm long, gla- brous. Stigma depressed in fruit. Fruit glo- bose, 6-7 mm diam. Pyrenes 4, trigonous, 4—4.5 mm long, dorsally unisulcate. This species differs from [lex acutidenti- culata Steyerm. in having smaller, broadly ovate leaves rounded or broadly obtuse at base and shortly apiculate at the apex; more numerous marginal teeth; shorter, glabrous petioles; and more divaricately spreading lat- eral nerves prominently sulcate above and elevated below. Ilex wurdackiana Steyermark, sp. nov. TYPE: Venezuela. Bolivar: Cerro Vena- mo, northwest slopes, SE of road camp km 125, to line of NE-facing, sandstone bluffs, 1,100-1,140 m, 21 Apr. 1960, J. A. Steyermark & S. Nilsson 449 (ho- "m VEN). Figure 6F, ru mulis superioribus puberulentibus; foliis rigidi -coriaceis dine bus late ovatis vel oblongo lanceolatis apice acutis vel acuminatis basi late rotundatis Volume 75, Number 1 1988 Steyermark 333 Flora of Venezuelan Guayana—lV vel subcordatis 4-10.5 cm longis 2-6 cm latis integerrimis glabris subtus valde punctatis; costa superne elevata nervis lateralibus utroque latere 4-6 supra subsulcatis vel valde sulcatis subtus elevatis; venulis tertiariis supra obsoletis subtus grosse reticulatis paullo elevatis; petiolis 1-3 mm longis glabris; inflorescentia masculina fasciculata axillari lateralique, pedunculis 7-8 filiformibus cymosis 4-6 mm longis glabris; pedicellis filiformibus 1-2 mm longis glabris; floribus 4-meris; calycis lobis suborbicularibus rotundatis 0.3-0.5 mm longis 0.7 i foeminea ciliaris axillari lateralique; pedicellis sub fructu 2.5-4 mm longis; bacca immatura ovoideo-subgloboso 2.5-3.5 mm longo 3 mm Woody i upper portions of stems puberulent. Leaves stiff-coriaceous, subses- sile, broadly ovate or oblong-lanceolate, acute to acuminate at apex, broadly rounded to subcordate at base, 4-10.5 cm long, 2-6 cm wide, entire, glabrous, strongly punctate be- neath; upper midrib prominently elevated; lat- eral nerves 4-6 side, subsulcate to prominently sulcate above, elevated below; tertiary venation obsolete above, grossly re- ticulate below. Petioles 1-3 mm long, gla- brous. Staminate inflorescence fasciculate with 7—8 filiform, cymosely branched, glabrous pe- duncles 4-6 mm long; pedicels filiform, 1-2 mm long, glabrous. Flower 4-merous; calyx lobes suborbicular, rounded, 0.3-0.5 mm long, 0.7 mm wide, glabrous. Petals broadly oblong, rounded at apex, 0.7-1 mm long, 0.8 mm wide. Anthers suborbicular-oblong. Pis- tillate inflorescence solitary, axillary, lateral. Fruiting pedicels 2.5-4 mm long, glabrous. Stigma depressed-flattened. Immature fruit ovoid-subglobose, 2.5-3.5 mm long, 3 mm wide. Paratype. | VENEZUELA. BOLÍVAR: laderas del Cerro Uei, ES T brazos del Rio Uei (afluente del Rio RM y Cuyuni), selva nublada, cg 1,050 m, 27-28 1970, J. A. Steyermark, G. . & E. o te 104565 (VEN). This species is noteworthy in its epiphytic shrubby habit. Other distinguishing charac- ters are the conspicuously elevated upper midrib, subsessile leaves, these entire or with a few minute excrescences, broadly rounded to subcordate at base, and the fasciculate staminate inflorescence with filiform pedun- cles and pedicels. The species is endemic to the region of the Cerro Venamo and tribu- taries of the Rio Venamo and Rio Cuyuni. It is a great pleasure to dedicate this dis- tinct species to Dr. n J. Wurdack, who has identified much material belonging to the genus /lex and who has rendered invaluable help to the author on many occasions. LITERATURE CITED Epwin, G. 1965. Aquifoliaceae. In: B. Maguire & Col- laborators, The botany of the Guayana Highland — part VI. Mem. New York Bot. Gard. 12(3): 124- 150 LOESENER, TH. 1901. Nova Acta Acad. i-r E Cur. 78: 1-567, pls. Monographia Aquifoliacearum. -Carol. German. Nat. SAPINDACEAE ALLOPHYLUS Allophylus parimensis Steyermark, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: Sierra Parima, vecindades Si- marawochi, Rio Matacuni, 3?49'N, 64*36'W, 6-7 km west of the Brazil- Venezuela frontier, 795-830 m, 18 Apr.-23 May 1973, Julian A. Steyer- mark 107033 (holotype, MO; isotype, VEN). Figure 7. Arbor 10-metralis, ramulis novellis modice pubesce tibus pilis subadpressis "praediis foliis lesa peo. latis, laminis lanceolato-ellipticis apice acuminatis basi acu tis, foliolis lateralibus sessilibus 4-5 cm longis 1- 3 3 cm latis (immaturis), terminali (intermedia) 5.5-6 cm longis .5-1.8 cm latis subintegris vel marginibus superioribus dbi date 1-5 dentibus tenuibus adpressis setulosis in- costa sparsim tenuibus 8-9; petiolis (novellis) 1.5-2.5 cm adpresso-pubescentibus; thyrsis anguste rac 3.5-5 cm longis 6-8 mm latis densifloris multifloris; flo- ribus solitariis vel IDA in cincinnis 2-3-floris dis- positis vel interdum axibus infimis elongatis racemifor- mibus 5-6-floris; pedunculo pedicellisque atque rhachidi adpresso-pubescentibus; PON Y ipei bois exterioribus ubique glabris interioribus extu e pulverulentibus intus glabris; petalis intus deine villosulis flamentis gla- bris; disco modice hispidulo. Tree 10 m tall; young branchlets moder- ately subappressed pubescent. Leaves 3-fo- liolate; leaflets lance-elliptic (immature) acu- minate at apex, acute at base, the lateral sessile 4-5 cm long, 1-1.3 cm wide; terminal (intermediate) leaf subsessile, 5.5-6 cm long 1.5-1.8 cm wide, entire or the upper margins with 1-5 appressed, slender, setulose teeth, 334 Annals of the Missouri Botanical Garden glabrous above, sparsely appressed-pubescent below on surface, midnerve, and lateral nerves; lateral nerves slender, 8-9 on both sides. Young petioles 1.5-2.5 cm long, densely ap- pressed-pubescent. Inflorescence a narrow ra- cemiform thyrse, axillary in the upper leaf axils, with 20-30 short, lateral, sessile or pedunculate cincinni, the upper axes mainly 1-flowered, the others usually 2—3-flowered, sometimes the lowest racemiform, elongated, 5-6-flowered and up to 12 mm long; rachis of inflorescence densely subappressed-pubes- cent, the hairs antrorsely suberect-ascending; pedicels 1-2 mm long, sparsely puberulous; peduncle 1-2 cm long, densely appressed- pubescent. Outer sepals ovate-oblong, round- ed, cucullate, 1.5 mm long, | mm wide, strongly ciliate, elsewhere glabrous both sides; inner sepals with incurved margins, suborbic- ular from a short claw, broader than long, 1.3 mm long, 1.6 mm wide, strongly ciliate, glabrous within, minutely puberulent without. Petals broadly obovate or suborbicular, ob- ovate, rounded at summit, narrowed to the base, 1.1 mm long, 0.7 mm wide above the middle, glabrous without, densely ciliate, with- in densely barbate-villosulous on the 2-lobed scale and lamina. Disk fleshy, shallowly lobed, moderately hispidulous. Filaments glabrous, 0.8 mm long. Ovary bilobed, suborbicular, densely hispidulous. This species is allied to Allophylus strictus Radlk. and 4. laevigatus (Turcz.) Radlk., from both of which it differs in the hispidulous disk, glabrous filaments, and sparsely pubes- cent style. It further differs in the sparingly 1 -5-denticulate upper leaf margins with slen- der appressed teeth, the sparsely puberulous pedicels, more densely pubescent interior of the petals, more densely and closely flowered inflorescence with a densely pubescent rachis, and the sometimes racemosely elongated low- est axes bearing up to 6 flowers. RUBIACEAE COUSSAREA Coussarea evoluta Steyermark, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: Cerro de la Neblina, on hills 1.5 km south of Base Camp on Rio Ma- warinuma, 0?50'N, 66?10'W, 140-340 m, 4 Dec. 1984, R. Liesner & D. Bell 17493 (holotype, MO; isotype, VEN). Figure 8. Stipulae suborbiculares late rotundatae 2-5 mm lon gae; inflorescentia magna longipedunculata 15-17 cm longa 7-10 cm lata pedunculo incluso, axibus lateralibus elongatis infimis 2-3.5 cm longis; corollae lobis ad di- midium vel magis longitudinem tubi corollae attingentibus recurvatis 7-10 mm longis; floribus sessilibus dense fas- ciculatis. Tree, 8 m tall, the stem glabrous; stipule suborbicular, shallow, broadly rounded, 2-5 mm long, 5-8 mm broad, glabrous. Petiole 1.5-2.3 em long, 3 mm thick, glabrous. Leaf blades large, subcoriaceous, oblong-elliptic, abruptly obtusely acuminate at apex, acumen 1-2 cm long, acute at base, 18-29 cm long, 5.0-13 cm wide, glabrous both sides, domatia absent; midrib prominent and elevated below, impressed above; lateral nerves 10-12 each side of the midrib, ascending from a 25-35? angle; tertiary venation grossly reticulate be- neath with large areoles, subelevated on lower surface. Inflorescence pedunculate, 15-17 cm long including the peduncle, 7-10 cm wide at the base, cymose-paniculate, the lat- eral axes subverticillate; peduncle 6-7.5 cm long, microscopically sparsely puberulent; lowest axes verticillate or subverticillate, usu- ally 3-4, widely spreading, 2-3.5 cm long, the other axes shorter, the lowest tier sepa- rated by 4-5 cm from the middle tier, the middle tier separated by 2-3.5 cm from the upper tier; lateral axes usually branched at the apex into 2-3 short, mostly unequal axes 7-15 mm long or unbranched, or with short axes 1-2 mm long, each ending in sessile groups of 2-7 flowers. Flowers sessile; calyx and hypanthium 2.5-3.5 mm long, micro- scopically sparsely puberulent; hypanthium obconic-urceolate or oblong-obconic, 1.2 mm long. 1 mm wide, obtusely ribbed, minutely puberulent; calyx tube deeply campanulate- urceolate, 2 mm long, 1.5 mm wide, minutely sparsely puberulent; calyx teeth 5, linear or subsubulate, recurved, 0.5 mm long, minute- Volume 75, Number 1 1988 Steyermark 335 Flora of Venezuelan Guayana—IV llmm FiGURE 7. Allophylus visi a mA, sak m dorsal side, outer s ved.—F. Pistil. —G. Di sk. ur Sta ly hirtellous. Corolla salverform, tube 12.5- 13 mm long, 1 mm wide; 4 corolla lobes linear, recurved, 7-10 mm long, 1 mm wide, glabrous within, microscopically puberulent without and at apex, the tube densely papil- late-puberulent without. Stamens 4 in the up- per 5 mm of corolla tube, included; anthers linear, 4.5 mm long, obtuse at apex with rounded connective, rounded at base. Disk short-columnar, truncate, 0.8 mm long. This species is characterized by the large inflorescences with elongate lateral axes, com- pact clusters of sessile flowers, elongated pe- duncles, corolla lobes at least half the length of the corolla tube, and shallow, rounded sti- pules. it of flowering branch. —B. Con —C. Parton ul upper margin of leaflet. —E. Flower with one of inner sepals uu G'| nvex dorsal side, inner sepal. I. Petal, abaxial view.—J. Petal, adaxial view. FARAMEA Faramea boomii Steyermark, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: Cerro de la Neblina, Base Camp, Rio Mawarinuma, terra firma for- est, 0*50'N, 66?10'W, 4 Dec. 1984, 140 m, B. Boom & A. Weitzman 5278 (holotype, MO; isotypes, NY, VEN, CH, PORT, K, BR, COL, U, INPA, P, MG, B, AAU, US) Stipulae aristatae, aristis 2-3 mm longis; foliorum lam- inis elliptico-oblongis apice abrupte obtuseque caudatis 1 0-15.5 cm longis 3-6 cm latis, ad marginem valde incrassatis; i ntia latiore quam latiore, floribus m longis basi 1.5 mm latis; ntheris 5 mm longis apice breviter cuspidatis. 336 Annals of the Missouri Botanical Garden ii nun acra T ` NS | | "p n FIGURE 8. — Coussarea evoluta. — 4. Habit of flowering branch.—B. Detail of portion of inflorescence, corollas absent. —C. Calyx and hypanthium.— D. Corolla. —E. Distal portion of corolla lobe. Tree 12 m tall, the young stems glabrous. Stipules persistent, laterally connate, the sheath 2-2.5 mm long, 3 mm wide, abruptly aristate, the arista indurated, 2-3 mm long, arising from just below the summit of the sheath, the part protruding above the summit 2 mm long. Leaf blades elliptic-oblong, abruptly obtusely caudate, acute at base, 10— 15.5 em long, 3-6 cm wide, the obtuse acu- men 12-13 mm long, 1.5-2 mm wide, gla- brous throughout, with a strong marginal nerve; principal lateral nerves widely spaced 8-13 mm distant, widely spreading-ascending at 10-25? angle, impressed above, slightly elevated below, the intermediate nerves finer and less conspicuous below; petiole 10-13 mm long, glabrous; tertiary venation scarcely evident above, grossly and irregularly retic- ulate below, slightly manifest. Inflorescence terminal, pedunculate, 3.5 cm long, 6 cm broad, cymosely branched in 3 main divari- cately spreading, glabrous axes 10-13 mm long, 0.5 mm diam., each axis branched into 2 lateral axes 5 mm long, mainly 3-flowered with a central short axis bearing a solitary flower; central flower 1 mm long pedicellate, other flowers 1-1.5 mm long pedicellate; ped- icels glabrous. Bracts obsolete, not manifest. Calyx and hypanthium 2.5 mm long, gla- brous; hypanthium columnar-urceolate, 1.5 Volume 75, Number 1 1988 Steyermark 337 Flora of Venezuelan Guayana—IV mm long, 1.2 mm wide; calyx broadly cu- pulate, truncate or essentially so at apex, 0.8- l mm long, 1.2-1.3 mm wide, eglandular within. Disk shortly columnar, 0.8 mm long. Corolla white, salverform, 25 mm long, the tube 17 mm long, 2 mm wide above, nar- rowing to 0.8 mm wide at base; 4 lobes linear- ligulate, narrowed to a rounded apex, 8-9 mm long, 1.5 mm wide at base, 0.5 mm wide at apex. Stamens 4, included; anthers linear, in the upper third of the corolla tube, cus- pidate at apex, 5 mm long, 0.5 mm wide, the cusp 0.3 mm long. From Faramea neblinae Steyerm. this species differs in the aristate stipules, mar- ginally thickened nerves of the leaves, shorter pedicels, and shorter calyx and hypanthium. rom F. crassifolia Benth. it differs in the longer corolla and anthers, differently shaped leaves, more evident lateral and tertiary nerves on the lower leaf surface, and absence of inflorescence bracts. From F. angustifolia Benth. & Hook. it differs in the truncate calyx and longer corolla and anthers. It differs from F. panurensis Muell. Arg. by having apicu- late anthers, shorter peduncles, thickened marginal nerves, longer anthers, and vaginate stipules. Faramea morilloi Steyermark, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: Parcela Fénologica, 16 km de Cruce con la carretera a Puerto Ay- acucho-Sanariapo, via a Gavilan, 100 m, 17 Nov. 1977, G. Morillo 67 16 (ho- lotype, VEN). x 1.5-2-metralis glaber; stipulis aristatis, artistis 4- 5n mm . foliis lanceolato-ellipticis apice acuminatis basi acutis 8-10.5 cm longis 1.5-3.5 cm latis; nervis lateralibus principalibus utroque latere 9-11; petiolis 2- longis; inflorescentia nn solitaria; pedicellis sub anthesi 3-5 mm longis sub fructu 5-6 mm longis; bracteis duobus apice pedicelli insidentibus subulatis 10-12 mm longis; hypanthio calyceque 5 mm longis 1.1 mm latis, hypanthio 1 mm lo ongo; e sub fructu deciduo; calycis inaequalibus subulatis 1-2 mm longis; corolla bate tubo 17 .2 m i 7 mm lon antheris linearibus 3 mm longis; stylo 15 mm amin fructu subgloboso rugoso 9 mm longo 11 mm lato; seminibus orbicularibus 7 mm diam. dorsaliter 8-costatis Shrub 1.5-2 m tall, glabrous. Sioda sheath 2 x 2 mm, aristae acicular, 4-5 mm long. Leaves papyraceous, lance-elliptic, acu- minate at apex, acute at base, 8-10.5 cm long, 1.5-3.5 cm wide; principal lateral nerves 9-11 each side, with fainter intermediate nerves, widely spreading, conspicuously ele- vated and anastomosing below, inconspic- uously impressed above; tertiary venation faintly reticulate above. Petioles 2-3 mm long. nflorescence solitary, axillary, shortly pedi- cellate. Pedicels in anthesis 3-5 mm long, in fruit 5-6 mm long. Two bracts at apex of pedicel subulate, 10-12 mm long. Calyx and hypanthium 5 mm long, 1-1.15 mm wide; hypanthium short-columnar, 1 mm long; ca- lyx deciduous in fruit; calyx tube short-cylin- dric, 2 mm long, 1.5 mm wide, twice longer than the hypanthium, the lobes unequal, 2 elongated teeth with a lateral subulate ap- pendage, one of the teeth with 2 shorter sub- ulate projections 0.6 mm long on one side and another short subulate projection on the side of a longer tooth. Corolla hypocrateri- form, the tube very slender, 17 mm long, 1- 1.2 mm wide; 4 lobes narrowly lanceolate- subulate, recurved at tip, 7 mm long 1 mm wide above base, 1.5 mm wide at base. Sta- mens 4, included, inserted at the summit of the corolla tube. Anthers linear, 3 mm long; filaments 1 mm long, attached 4 mm below orifice of tube. Stigmas subulate, 3 mm long. Fruit depressed-globose, rugose, broader than high, 9 mm high, 11 mm broad. Seed 1, orbicular, 7 mm diam., with 8 longitudinal ridges radiating from dorsal side. With its solitary axillary flowers, this taxon is related to Faramea egregia Sanders and F. spathacea Muell. Arg. ex Standley. From both of these it is distinguished by the smaller calyx and hypanthium, smaller corolla tube and lobes, and shorter petioles. The rugose fruits and ridged seeds are noteworthy. From F. brevipes Steyerm. it differs in the decid- uous fruiting calyx, acutely acuminate leaves, and setaceous (vs. subfoliaceous) bracts sub- tending the calyx and hypanthium. 338 Annals of the Missouri Botanical Garden Faramea paludicola Steyermark & Boom, sp. nov. TYPE: Venezuela. Territorio Fed- eral Amazonas: Depto. Rio Negro, upper Cano Baria, **swampy" area between Rio Mawarinuma and headwaters of Rio Ba- ria, ca. 0%52'N, 66?15'W, 130 m, 26 Mar. 1984 (fl. buds), R. Liesner 16960 (holotype, MO; isotypes, NY, ). A Faramea torquata Muell. Arg. folis nervis margi- nalibus valde prominentibus differt. Shrub 2 m tall. Stipule with shallow cori- aceous sheath broader than long, 7 mm wide, 4 mm high, glabrous, with a rigid awn 3 mm long arising from the center in a depression below the summit. Leaves coriaceous, elliptic- oblong, abruptly caudate at apex, acute at base, 23-30 cm long, 7-11 cm wide, con- spicuously marginally nerved, glabrous on both sides, acumen obtuse, 13-18 mm long, 1.5- 2 mm wide; lateral nerves strongly elevated below, impressed above, anastomosing 7-11 mm from margin, principal elevated nerves 15-18 each side; midrib stout, elevated above, impressed below; tertiary venation finely re- ticulate, subelevated both sides. Petiole gla- brous, stout, vaginate in upper half, open adaxially, 15-17 mm long, 4 mm wide. In- florescence slenderly pedunculate, terminal, 2-3 pedicels arising together, purple, 3-3.5 cm long, 1 mm wide, with 3-4 axes 8-12 mm long arising umbellately and divaricately, these separated 8- 10 mm distance by a sec- ond higher tier of 3 axes 6-7 mm long, ter- minating in a final cluster of closely positioned several axes 2-3 mm long; axes of inflores- cence purple, microscopically papillate but appearing glabrous; peduncle appearing gla- rous, the base subtended by 2 bracteoles united by a suborbicular sheath 1 mm long terminating abruptly in a 2-2.5 mm glabrous awn. Flowers 3-5-umbellate at the ends of the lowest primary axes on pedicels 1-2 mm long. Flowers on purple pedicels, 2-3 mm long, 1 mm wide, ebracteolate. Calyx and hypanthium microscopically papillate-puber- ulent, 2-2.5 mm long; hypanthium obconic, 1.5-2 mm long, 2 mm wide at summit, grad- ually narrowed to a base 1 mm long; calyx 2-2.5 mm wide at summit, 0.8-1 mm high, subtruncate or slightly undulate; corolla green in bud, 4 mm long, 1 mm wide at base, 0.8 mm wide at the rounded tip, MONET papillate-puberulent; corolla lobes mm long, ligulate, obtuse; ee 1.5 mm long, filaments 1 mm long, attached at base of corolla; gynoecium immature. Fruit un- known. Faramea yavitensis Steyermark, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: Yavita, 28 Jan. 1942, Llew- elyn Williams 14026 (holotype, VEN; isotype, rutex 2-metralis glaber; stipulis insu, aristis 4 mm longis; foliis elliptico- -oblongis api ice abru — acu- mine 7-12 mm longe, basi cuneatim acutis 12-16 c longis 4-6.5 cm latis, nervis saraq oo eee 7 10, venatio tertiaria inconspicua; petiolis 1.2-1.5 cm longis; infructescentia epedunculata, axibus secundariis tribus primariis umbellatis 2.7-3.2 cm longis; axibus se- cundariis tribus 8 persistentibus 2.5 mm longis; bracteis sub axibus primariis persistentibus subulatis 5.5 mm longis, sub axibus secun 3.5 mm longis; calycis tubo sub fructo vel quattuor, setaceis persistentibus sub fructu 2-3.5 mm longis; ind subgloboso 4-5 mm longo 6-7 mm lato. Glabrous shrub 2 m tall. Stipular teeth linear, 4 mm long, persistent on upper nodes, deciduous or not evident on lower nodes. Leaves elliptic-oblong, abruptly caudate at apex with acumen 7-12 mm long, cuneately acute at base, 12-16 cm long, 4-6.5 cm wide; lateral nerves 7-10 each side, divari- cately spreading at an angle of 5-25?, ele- vated below, slightly elevated or impressed above; tertiary venation inconspicuous with large, areolate, AR A reticulation on both sides. Petioles 1.2-1.5 cm long. Infructes- cence epedunculate, the 31 primary rays um- bellate, 2.7—3.2 cm long, 1.5 mm wide, each ray terminating in 3 secondary axes 8-10 mm long and dilated apically; secondary axes ending in branches with 2 lateral fruiting ped- icels 4-5 mm long on either side of a central sessile fruit. Bracts at base of axes persistent, subulate, those subtending pedicels 2.5 mm long, those subtending secondary axes 3.5 Volume 75, Number 1 1988 Steyermark 339 Flora of Venezuelan Guayana—IV mm long, those subtending primary axes 5.5 mm long. Calyx teeth persistent in fruit, se- taceous, 2-3.5 mm long. Fruit depressed- globose, 4-5 mm high, 6-7 mm broad. This taxon is related to Faramea multi- flora A. Rich. and its varieties (for discussion of these, see Mem. New York Bot. Gard. 17: 390-395. 1967) but differs from it and re- lated species in having the persistent fruiting calyx with elongated teeth. Faramea yutajensis Steyermark, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: below summit of east slope of unnamed peak, 8 km NW of Yutajé settlement, 4 km west of Rio Coro Coro, west of Serrania de Yutajé, 5°41'N, 66°10'W, 1,500-1,760 m, 4 Mar 1987, R. Liesner & B. Holst 21640 (holotype, MO; isotype, VEN). ex 2-metralis glaber; stipulis aristatis, aristis 2.2 In axillarique fere sessili vel pedunculata, pedunculo solitario 10-15 mm longo glabro da apicem 2 mm lato basi 1.5 mm lato; bracteis duobus sub fructu ovatis acutis 13- 23 mm longis 6-8 mm latis dba fructu globoso 9-10 mm longo 10-11 mm lato calyce persistente 2 mm longo 2.5 mm lato coronato. Glabrous shrub, the stem 1-angled or ridged on each side. Stipular sheath triangular, car- inate, 1 mm long, 1.2 mm wide at base, tapering into an acicular awn 2.2 mm long. Leaves oblanceolate or lance-elliptic, abruptly and obtusely acuminate at apex, the acumen 2-12 mm long, 3 mm wide, conspicuously acutely attenuate at base, 8-14 cm long, 2- m wide, slightly decurrent on petiole; principal lateral nerves 7-11 each side, sub- horizontally spreading, anastomosing near margin, slightly elevated below, slightly sul- cate above, intermediate nerves shorter and less conspicuous; tertiary venation reticulate and slightly manifest below, faintly reticulate above. Petioles 3-9 mm long. Inflorescence terminal and axillary, nearly sessile or on a peduncle 10-15 mm long. Peduncles slightly dilated to 2 mm toward summit, 1.5 mm wide at base, solitary. Subtending two bracts at base of fruit ovate, acute, 13-23 mm long, 6-8 mm wide. Fruit globose, 9-10 mm high, 10-11 mm wide, crowned by the dr calyx tube, this 2 mm long and 2.5 mm wide. This species is distinguished by having sol- itary primary inflorescence axes with only one or few flowers which terminate the peduncle and by the subtending enlarged bracts. Far- amea yutajensis differs from Faramea ani- socalyx Poeppig & Endl. by having smaller green bracts and by having the solitary pri- mary ray bearing one to few flowers. From F. parvibractea Steyerm. and F. cardonae Steyerm. it differs in the solitary peduncle with one to few flowers. FERDINANDUSA Ferdinandusa boomii Steyermark, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: Cerro de la Neblina, Base Rio Mawarinuma, | 0950'N, 66*10 W, 8 Dec. 1984, 140 m, B. Boom & A. Weitzman 5274 (holotype, MO; isotypes, NY, VEN, GH, INPA, PORT, U, COL K schultesii Steyerm. corollis longioribus, pedicellis longioribus, foliis majoribus plerumque pan te rotundatis vel subcordatis recedit; corollis 40-41 mm es s, tubo 35 mm longo; pedicellis 2-4 mm ies foliis m longis 5-9 cm latis basi rotundatis vel paullo A enn Tree 10 m tall; stems densely puberulent; petioles glabrous both sides, 8-15 mm long. Leaf blades coriaceous, ovate to oblong-ovate, abruptly shortly and obtusely acuminate at apex, rounded or slightly peg at the base, 7.3-14 cm long, 5-9 cm wide, the acumen 5-6 mm long, 2-3 mm adde gla- brous on both sides; lateral nerves 7-8 each side of midrib, arcuate-ascending to the mar- gins; tertiary venation iue oai 8 reticu- late above, the smaller areolae 2-3 mm diam. Inflorescence terminal with 3-5 Deli cymes, 4-4.5 cm long, 4-4.5 cm wide, each cyme 4-12-flowered; peduncle of each cyme 1-2 cm long, minutely puberulent-hirtellous Annals of the Missouri Botanical Garden with divaricate hairs; axes of cyme similarly pubescent. Calyx and hypanthium 2-2.5 mm long, the hypanthium columnar-obconic; ca- lyx lobes 4, deltoid, obtusely acute, 0.2-0.5 mm long, 0.7 mm wide at base, sparsely short ciliolate on margins and in sinuses, apparently eglandular within. Disk shorter than to equal- ing calyx lobes. Corolla 40-41 mm long, gla- brous without and within, the tube 35 mm long, 2.5 mm wide throughout except at ori- fice where 5 mm wide; 4 lobes 4 mm long, 4.5-5 mm wide, lobed above the middle, gla- brous within and at sinuses; 4 stamens un- equally inserted above style, in the upper 76 of the tube; anthers broadly oblong. 1.2 mm long, rounded at both ends; filaments 5 mm long. Style ending 7 mm below orifice of co- rolla. From Ferdinandusa schultesii Steyerm. to which this species is related, it is separated by the longer corolla and corolla tube, longer pedicels, and larger leaves generally broadly rounded to subcordate at the base. MORINDA Morinda longipedunculata Steyermark, sp. nov. TYPE: Venezuela. Territorio Fe- deral Amazonas: Dept. Rio Negro, Cerro de la Neblina, Canyon Grande, along Rio Mawarinuma between the mouth of the canyon and the first major fork of the river, 7 airline km ENE of Puerto Chimo, 0°50'N, 66%02'W, 350-400 m, 9-14 July 1984, G. Davidse & J. S. Miller 27327 (holotype, MO; isotype, VEN). Figure 9 } Frutex 2-metralis; foliis late tis apice abrupte caudatis basi cuneatis 17-30 cm longis 6-8 cm latis supra glabris subtus modice hirtellis pilis patentibus instructis; pedunculis axillaribus 20 cm longis breviter puberulis; floribus 4-meris; calyce dart campanulato 1.5 x .9 mm subtruncato extus hispidulo; corolla cylindrica extremitatibus angustatis alabastro 8 mm longa extus his- pidula. Shrub 2 m tall; upper part of stems densel y pubescent with short, appressed-ascending hairs. Stipule deeply bifid, short-pubescent without, the subulate teeth 2.5 mm long. Pet- ioles 1.5-2 cm long, moderately hirtellous. Leaf blades membranous, broadly oblanceo- late, abruptly caudate at apex, long tapering to a cuneate decurrent base, 17-30 cm long, 6-8 cm wide, upper surface with scattered, pale raphides, glabrous, the lower surface moderately hirtellous with spreading hairs; midrib and lateral nerves densely pubescent with spreading hairs 0.5-0.6 mm long; lateral nerves 9-10 each side; tertiary venation in- conspicuous below, scarcely manifest above. Peduncles axillary, greatly elongated, equal- ing or slightly shorter than leaves, 20 cm ong, 2 mm wide, moderately shortly puber- ulous, apically branched into 3-4 primary axes 9 mm long in anthesis, the primary axes again branched into 3-4 secondary axes 3- 4 mm long (in anthesis), 10 mm long (in fruit). Bracts at base of secondary axes shallowly 3-4-lobed, 2 mm long (fruiting). Flowers 4-merous, at summit of secondary axes, con- gested, sessile, several. Calyx cupuliform- campanulate, 1.5 x 1.5 mm, subtruncate with shallowly raised border, hispidulous with- out. Corolla cylindric-tubular, 8 mm long (late bud), 1.5 mm wide at the middle, slightly narrower at apex and base, outer surface hispidulous except glabrous in basal 1.2 mm; tube glabrous within; lobes suborbicular, rounded, 0.5 x 0.5 mm. Stamens inserted in the lower third of the corolla; anthers lin- ear-oblong, 2.2 mm long. Disk cupuliform, 0.5 mm high. Ovary 4-celled. oe dark red, globose, 7-8 mm long, 8-8.5 m iam., umbonate, 4-seeded. — This taxon is noteworthy for the excep- tionally elongated peduncle PSYCHOTRIA Psychotria anartiothrix Steyermark, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: Dept. Rio Negro, lower part of Rio Baria, inundated forest along riv- er, 1°10'N, 66?25'W, 22-23 July 1984, 80 m, Gerrit Davidse 27651 (holotype, MO; isotype, VEN). Frutex 2-metralis, stipularum aristis acicularibus 5-6 mm longis minute puberulentibus glabrescentibus; foliis elliptico-lanceolatis extremitatibus acutis 4-8.5 cm longis 1-2.5 cm latis; inflorescentia dense cymoso-corymbosa, Volume 75, Number 1 1988 Steyermark Flora of Venezuelan Guayana—IV FIGURE 9. thium.—C. Corolla, late bud stage. — D. axibus principalibus tribus angulis superioribus seriebus Ji oe ap ada ee orb d oe y I O r corollae tubo hypanthioque similiter pubescentibus. Shrub 2 m tall with glabrous stems. Stipular sheath 1 mm long, 2 mm wide, truncate, bearing on each side 2 acicular teeth 5-6 Morinda longipedunculata. — A. Habit of flowering and fruiting es e Calyx and hypan- C S orolla, interior view.—E. tamen, ventral v mm long arising 0.3 mm below summit of sheath, minutely puberulent becoming gla- brous. Leaf blades subcoriaceous, elliptic-lan- ceolate, acute at base and apex, 4-8.5 cm long, 1-2.5 cm wide, glabrous throughout except for minutely hispidulous puberulous 342 Annals of the Missouri Botanical Garden midrib on lower side, or completely glabrous; lateral nerves 8-9 on each side of midrib, arcuately ascending at 15—30°, subelevated below, slightly anastomosing near (1-2 mm) the margin, inconspicuous above; tertiary nerves slightly evident both sides. Inflores- cence densely cymose-corymbose, 8 mm high, 1.6 cm broad, pedunculate; peduncle 1.4- 1.8 cm long, 1 mm diam., microscopically puberulent on the angles with spreading hairs. Three main axes of inflorescence 2-3 mm long, 0.5-1 densely flowered, shortly branched, microscopically puberulent mm wide, with spreading, stiff, puberulent-hispidulous hairs in vertical lines on the angles. Flowers sessile; calyx and hypanthium 1.2 mm long in anthesis; hypanthium short-columnar, 0.5 mm long, mm wide, minutely hispidulous on angles, glabrous elsewhere; calyx cupulate, 5-lobed, 0.7 mm long, 1.1 mm broad; teeth broadly shallowly triangular, subacute, 0.5 mm long, 0.8 mm wide, ciliolate on margins, dorsally puberulent, eglandular within. Co- rolla white, cylindric, broadened at orifice, 5 mm long; tube 3.5 mm long, 0.7 mm wide except ] mm wide at orifice and 0.9 mm wide at base, sparsely puberulent-hispidulous in lines without, pilose in upper 1.5 mm within; 5 lobes fleshy and thickened with a cornic- ulate apex 1.5 mm long, 0.6 mm wide, pu- berulent in lines. Stamens 5, included, at- taining the orifice, the upper half of corolla; anthers linear, obtuse at apex, 0.8 mm long. Stigmas 2, rhomboid-ligulate, truncate at apex, papillate-puberulent, 0.3 mm long; style fili- form, papillate, 3 mm long. Disk depressed cupulate, shorter than calyx lobes, 0.4 mm high. Fruit orange, globose, 5 X 5 mm; py- rene dorsally acutely 3-costate, ventrally flat with a narrow sulcate depression along the middle. The minute puberulence of short, stiffish, divaricate hairs in more or less irregular ver- tical lines along the upper angles of the in- florescence axes, exterior of corolla tube, and the hypanthium is characteristic of this taxon. Psychotria edaphothrix Steyermark, sp. nov. TYPE: Venezuela. Bolivar: 0-3 km west of El Polo (8.6 km west of El Pauji), 4?30'N, 61?40'W, 650-800 m, 5 Nov. 1985, R. Liesner 19595 (holotype, MO; isotype, VEN). Figure 10. x 3-metralis; stipulae vagina 2 mm longa 5.5 mm Fru lata glabra i in dentes duos triangulari-lanceolatos longiat- tenuatos atos glabros E Heredes praedita, pedunculo 1.2-6 em longo 1 tomentoso- hirsutulo; calycis lobis 0.8 mm longis ciliatis; corolla alba, sub anthesi 14 mm longa basi 2 mm lata omnino sy NR uud intus basi pilis brevibus vestito. Shrub 3 m tall, the stem glabrous. Stipule sheath 2 mm .9 mm wide, glabrous with 2 hinaia: icedi long-attenuate teeth on each side, 4 mm long, 1.5 mm wide at base, interior of sheath with numerous acu- leiform processes. Leaves oblong-elliptic to broadly oblanceolate, acuminate at apex, acute at base, (9.5-)17-25 cm long, (3-)6-8.5 cm wide, glabrous both sides; lateral nerves 17- 2] each side, ascending at an angle of 25- 35°, elevated below, impressed above; tertiary veinlets conspicuously transversely connect- ing the secondary nerves, conspicuously el. evated below with reticulate pattern. Inflo- rescence 2.5-4 cm high, 2.5-6 cm wide, divided into 3 densely-flowered, subhemi- spherical heads, these 1.7-2.5 cm high, 1- 3 cm broad; peduncle 1.2-6 cm long, 1.5- 2.5 mm wide, densely brown tomentose with spreading hairs. Main axes of inflorescence 0.8-2 cm long, ascending to divaricate, densely brown tomentose. Bracts of inflores- cence lanceolate, caudate-acuminate, arising at the sides and apices of the axes, but not at their bases. Flowers 15-21 in one branched axis, 42—63 in one inflorescence, sessile, each flower subtended laterally by 2 lanceolate, caudate bracts 6 mm long, 1 mm wide, pu- berulous on both sides, the 2 bracts subtended by a larger middle bract 9 mm long, 2.5 mm wide, oblanceolate, acuminate-caudate, pu- berulous on both sides, with a larger outer bract, this lanceolate, caudate-acuminate, 13 mm long, 3 mm wide, puberulous on both sides. Calyx and hypanthium 2 mm long, the hypanthium 0.5 mm high, 1.5 mm wide, Volume 75, Number 1 1988 Steyermark 343 Flora of Venezuelan Guayana—IV dere cnt TATI CS ANTAS Senan ESC FIGURE 10. Psychotria edaphothrix.— A. Habit of flowering branch.—B. Portion of inflorescence, corollas absent. —C. Corolla, interior view.— JD. Stipule with bases of petioles. densely brown tomentose-hirsutulous; calyx lobes 5, deltoid, obtusely acute, 0.8 mm long, l mm wide at base, ciliate, glandular within in sinuses. Corolla white, cylindric, symmet- ric, 14 mm long, 3.5 mm wide below orifice, 2 mm wide at base, densely pilosulous without with spreading hairs, glabrous without at the very base, within shortly pubescent at base with a zone of hairs 1 mm above base; corolla lobes oblong-ovate, 2 mm long, 1.2-1.5 mm wide. Stamens 5, included in the upper half; anthers linear, obtuse at apex, 3 mm long, 344 Annals of th Missouri Econ Garden the filaments inserted half way up corolla tube. Disk subcupulate, about the height of the calyx tube. Psychotria edaphothrix 1$ somewhat in- termediate between Palicourea and Psycho- tria, in some respects resembling Palicourea longistipulata (Muell. Arg.) Standley in the densely pubescent zone confined to the in- terior base of the corolla tube, aculeiform structures within the stipular sheath, pubes- cent exterior of the corolla, and large leaves with numerous lateral nerves. It differs, how- ever, in the large subtending bracts of the compact inflorescence, densely tomentose ca- lyx, glabrous stem and leaves, and white, sym- metrical corolla and white bracts. In other respects the new taxon resembles Psychotria maturacensis Steyerm. of the Neblina area, but the stipular sheath is conspicuously bi- dentate on each side, the hypanthium is densely tomentose, the calyx lobes are con- spicuous and densely ciliate, and the leaves have more lateral nerves on each side. Since the corolla is symmetrical and shows no gib- bosity at base, I am including it in Psychotria as a borderline species. Psychotria pectinata Steyermark, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: Cerro de la Neblina, white water tributary of Rio Mawarinuma, ca. 3 km upstream (SE) of Base Camp, 0?49'N, 66°08’W, 150 m, 17 July 1984, Ronald Liesner 15955 (holotype, MO; isotype, VEN). Figure 11. Frutex 0.75-2 metralis, caulibus glabris; stipulae den- tibus setaceis 4-5 mm longis 0.2-0.3 mm latis dense pectinato-ciliatis pilis brunneo-hirtellis vestitis; foliorum laminis anguste oblanceolatis vel lanceolato-ellipticis apice cephala vel in capitula tria vel quattuor ramosa 5 longo 10-11 mm lato densiflora; pedunculo 3-13 mm longo 1 mm lato glabro; bracteis sub. inflorescentia duas, fructu elliptico-oblongo 9 mm longo 5 mm lato obtuse 10-costato. Shrub 0.75-2 m tall, the stems glabrous. Stipular sheath 2.5-3 mm long, 4 mm wide with 2 elongated, setaceous or lance-linear teeth 4-5 mm long, mm wide, dense- ly pectinate-ciliate with brown, hirtellous hairs. Petiole 5-15 mm long, glabrous. Leaves lan- ceolate-elliptic or narrowly oblanceolate, acute to acuminate at apex, cuneately acute at base, partly decurrent on the petiole, 10-15 cm long 1.5-2.5 cm wide, glabrous on both sides; lateral nerves 8-10 each side of the midrib, 1.5-3 mm distant, slightly impressed on both sides, the midrib slightly raised above. Inflo- rescence capitate, 5-6 mm high, 10-11 mm road, monocephalous or branched into 3-4 heads, 15-40-flowered, axillary or terminal, with 2 spreading bracts at the base at the summit of the peduncle. Peduncle erect, 3- 13 mm long, 1 mm wide, glabrous. Each axis of the inflorescence bearing ca. 5 closely crowded, pedicellate flowers and subtended by 3-4 lanceolate, acute, brown-ciliate bracts 1.5 mm long. Pedicels 0.5-1 mm long, gla- brous. Calyx tube and hypanthium 1.5-2 m long, glabrous; hypanthium obconic, 1 mm long, 1-1.5 mm above; calyx lobes 5, con- spicuous, spreading and slightly squarrose, lanceolate, acute, 0.8-1.5 mm long 0.6 mm wide, lili Disk higher than calyx tube in anthesis, fleshy, oblong-ovoid. Corolla white, small, the tube broadly cylindric, 2 mm long, 1.5 mm wide at summit, 1.3 mm wide at base, glabrous without; lobes 5, re- curved, 1.8 mm long, 0.8 mm wide, con- spicuously corniculate, the tube densely pu- bescent within at orifice. Stamens at orifice, slightly protruding above the tube; anthers 0.5 mm long; filaments 0.4 mm long. Style filiform, glabrous, 2 mm long, exserted. Fruit elliptic-oblong, 9 mm long, 5 mm wide, gla- brous, obtusely 10-costate, crowned by the persistent calyx. Paratyp VENEZUELA. TERRITORIO FEDERAL AMAZONAS: Cerro de la Neblina, Rio Mawarinuma, up- Ls ua Base Camp, 0%50'N, 66?10'W, 140 m, 2 ay 1984, Stein, Gentry & Thomas 1715 (MO, Cerro de i: Neblina, Caño Blanco, white water tributary of Rio Mawarinuma, m upstream (SE) of Base Camp, 0?49'N, 66?8' w. 150 A Miller 17 ARE VEN); same locality, Boom et al. 5 0,G A, PORT, U, NY, VEN); same di. jos 59612 (MO, VEN). This species is readily distinguished by the densely hirtellous, pectinate-ciliate stipular Volume 75, Number 1 1988 Steyermark 345 a Flora of Venezuelan Guayana—lV F A. Habit of A FIGURE 11. Psychotria PER — with a subtending bract. — interior view.—G. S teeth, densely flowered heads of small flowers, brown-ciliate, elongate calyx lobes, and strongly corniculate corolla lobes. It was not- ed that within a densely flowered head of an inflorescence only one or a few flowers pro- duce seed, and the majority do not possess a fertile ovary. There may be a tendency here for monoecism. Psychotria steinii Steyermark, sp. nov. TY enezuela. Territorio Federal Amazonas: Cerro de la Neblina, summit, Camp 2, 2-8 km NE of Pico Phelps, , D. Showing corniculate soralla. lobe tipule. Cu Disk and portion of calyx and hypanthium. nch.— B. Flower from outer portion of inflorescence ol .—E. Calyx and hypanthium.—F. Corolla, 0?49'40"N, 65?59'W, 15 Apr. 1984, 2,100 m, B. A. Stein & A. Gentry 1553 (holotype, MO; isotype, VEN). Figure 12 Suffrutex vel Subherhaces 0. 35- 1 -metralis, caulibus ongis; inflorescentia terminali plerumque tribus simul, omnibus conspicue nrecteats bracteis e 4- 7 ovatis acutis vel acuminatis mm lon. -o mm latis glabris vel basim versus sparsim hirtellis, margiribus ciliatis; pedunculo pubes inflorescentiae axibus pu- bescentibus; ralla .9 mm longa, tubo intus minute papillato sed admodum glabro extus glabro. 346 Annals of the Missouri Botanical Garden FIGURE 12. Psychotria steinii. — A. Habit of flowering branch. —B. Portion of infloresc "enc ene. — D. Stipule with bases of petioles Subherbaceous or suffruticose plant 0.35- 1 m tall, the stems glabrous except at upper youngest nodes where hirtellous with spread- ing hairs. Stipule with a shallow sheath 0.2- 0.5 mm high, 1-4 mm wide, glabrous, in- durated with 2 teeth on each side, linear- lanceolate, acute, 0.5-0.8 mm long. Leaves shortly petiolate, leaf blades subcoriaceous, ovate, acuminate at apex, acute to obtuse at base, 2.5-4.5 cm long, 0.9- cm wide, glabrous on both surfaces, the margins strong- ly ciliate, the lower side with conspicuous cystoliths; midrib below glabrous to hirsutu- lous-ciliolate along margins; lateral nerves 6— 8 on each side of the midrib, sulcate above, slightly elevated below, ascending at a 45? —C. Pyr .—E. Calyx, showing portion of interior with calycine yes and disk. angle, ending near or at margins without anas- tomosing; tertiary venation inconspicuous; petioles 2-3 mm long. Inflorescence terminal, usually 3 together; each inflorescence with 4-7 conspicuous, foliaceous bracts arising at the summit of 2 short lateral axes 1-2 mm long and 1 central subsessile axis scarcely 1 mm long; bracts green, similar to leaves in shape but smaller, ovate to elliptic-oblanceo- late, acute to acuminate, 8-13 mm long, 3- 5 mm wide, glabrous or sparsely hirtellous toward base, the margins ciliolate, the nar- rower lateral bracts narrowed to a subpetiol- ate base, the broader bract subtending the inflorescence at a slightly lower level and not narrowed at base. Peduncle 3-10 mm long, Volume 75, Number 1 1988 Steyermark 347 Flora of Venezuelan Guayana—IV 0.6-0.8 mm wide, densely hirtellous with spreading hairs 0.1 mm long. Flowers sessile, 2, or 2 on each axis. Calyx and hypanthium 1.5-1.8 mm long; hypanthium short colum- nar, 1 mm long, 1 mm wide at summit, gla- brous; calyx 0.8-1 mm long, ine 5-lobed, the tube 0.5-0.6 mm long, 1.5 m wide, glabrous without; lobes broadly trian- gular, narrowed to a subacute apex, 0.5-0.7 mm long, ciliolate on margins, glabrous else- where, irregularly erose, glandular within at base of each of the 5 sinuses. Corolla white, tubular-infundibuliform, 1 mm long, the tube gradually enlarged upward, 7-8 mm long, 2 mm wide at base, 4 mm wide at orifice, glabrous without, within microscopically pa- pillate but essentially glabrous; lobes 5, 4- 4.5 mm long, 2 mm wide, ligulate-oblong, obtuse, glabrous without, microscopically papillate within but essentially glabrous. Sta- mens protruding slightly above orifice; an- thers linear, 2.5 mm lon mm wide at the orifice; flaments attached in upper half of tube, 2 mm long; disk shorter than calyx tube. Fruit blue, fleshy, subglobose, 11 mm long, 12 mm wide, 2-seeded; pyrenes broadly ovoid, dorsally 3-costate with 1 additional cos- ta on each side, the inner face subconcave, 5 mm long, 3.5 mm wide. Par pes VENEZUELA. TERRITORIO FEDERAL quibu jm de la Neblina, Camp 12, Venezuelan- Brazilian frontier, 1,950 m, 26-27 Feb. 1985, Boom et al. 5982 (MO, VEN, GH, INPA, NY) This taxon is related to Psychotria dui- dana Standley of Cerro Duida and P. oblata of Mount Roraima. It is characterized by the essentially glabrous, although minutely papil- late, interior of the corolla tube, much small- er, ciliolate leaves, densely pubescent pedun- cle, smaller corolla, and pubescent axes of the inflorescence above the peduncle. Psychotria thesceloantha Steyermark, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: upper Cano Baria, swampy area between Rio Mawarinuma and headwaters of Rio Baria, 0%52'N, 66?15'W, 130 m, 4 Dec. 1984, R. Lies- ner 16963 (holotype, MO; isotype, VEN). Figure 13 Suffrutex 1-metralis, caulibus glabris; stipulae vagina truncata 5 mm longa 9 mm lata glabra; foliorum laminis intus glabris; floribus omnibus a bracteolis duabus circum- cinctis bracteolis lanceolato-linearibus acuminatis 13 mm longis 1.5 mm latis setuloso- belio FEMME lobis aris- tatis 2.5-4 mm longis ciliato-plumosi Suffruticose dn 1 m tall, the stem gla- brous, 4-5 mm diam. except 8- 10 mm diam. at nodes. ha large, submembranous, leaf blades elliptic-ovate, the apex acute to ob- tusely subacuminate, the base cuneately acute, 21-24 cm long, 10-11 cm wide, glabrous below except minutely sparsely puberulent on midrib and nerves; lateral nerves 10-11 eac side; tertiary venation not evident; petiole 4.5— 5 cm long, glabrous to sparsely puberulent. Inflorescence axillary, subhemispheric, 2 cm long, 4 cm wide, appearing monocephalous on a short glabrous peduncle 0.4 mm long, not enveloped by the stipular sheath, con- sisting of 3-4 separate monocephalous heads in a compact mass appearing as monoceph- alous, each portion of the head hemispheric, 1.8-2 cm long, 2—2.8 cm wide, the complete head surrounded by 4 main outermost bracts separate to the base, elliptic-ovate, abruptly acuminate, rounded at base, carinate, 2 cm long, 1.3-1.5 cm wide, puberulent without, glabrous within; large bracts within the out- ermost bract broadly ovate, acute, ecarinate, 1.5 em long, 1 cm wide, puberulent without, glabrous within; each of the 3-4 smaller heads consisting of multibracteate flowers, each sur- rounded by an obovate, abruptly acuminate bractlet 1.5 cm long, 0.5 cm wide, puberulent on both sides; each bractlet subtending 2 ob- lanceolate bracteoles rounded at apex, 1.2- 1.3 cm long, 0.4 cm wide, puberulent with- out, sparsely puberulent within and subtend- ing a group of 3 flowers; each pair of brac- teoles subtending an ultimate pair of floral bracteoles, these lanceolate-linear, acumin- 348 Annals of the Missouri Botanical Garden ; 13. "wm thesceloantha.— 4. Habit of flowering branch.—B. Individual flower without corolla, L filed by two bracteoles and a bract inflorescence ate, 1.3 cm long, 1.5 mm wide, setulose- serrulate in the upper half of the margins. Hypanthium obconic-columnar, 2 mm long, ] mm wide, glabrous; calyx lobes 5, seta- ceous, unequal, plumose-ciliate. This unusual species has large, axillary, compound inflorescences that appear to monocephalous but are composed of 3-4 con- densed heads having the outermost and major bracts separate to the base. The ultimate three flowers are surrounded by narrowly oblan- ceolate, serrulate-ciliolate bractlets, with the large inner bracts acuminate and puberulent on both sides, while the calyx lobes are elon- .—C. Inflorescence with bracts.—D. One of the inner bracts of the gate, awnlike, and plumose-ciliate. In the present taxon the inflorescence is not envel- oped by the stipular sheath as in Psychotria celiae Steyerm. of Cerro de la Neblina. Psychotria yutajensis Steyermark, sp. nov. PE: Venezuela. Territorio Federal Amazonas: Dept. Atures, summit of east slope of unnamed peak, 8 km NW of Yutajé settlement, 4 km west of Rio Coro Coro, west of Serrania Yutajé, 5?41'N, 66°10'W, 1,500-1,760 m, 4 Mar. 1987, R. Liesner & B. Holst 21649 (holotype, MO; isotype, VEN). Volume 75, Number 1 1988 Steyermark 349 Flora of Venezuelan Guayana—IV Frutex 1.5-metralis glaber; ~~ leviter dentatis, va- gina 1.5 mm alta 4-6 mm lata, utroque latere bidentatis, dentibus late ovato-lanceolatis acutis 1 -2 mm longis intus sericeis; petiolis 9-20 mm longis; poet pasate terminali o-subhemi- triangularibus apice Porr, floribus 5-meris pedicel- latis, pedicellis articulatis 0.5-1.5 mm longis 0.5-0.8 mm latis; calyce hypanthioque 2.3 mm Pus a do- liiformi vel breviter oo 1.2 x minute puberulenti; calyce 1.2 mm longo 1.5 mm lato minute puberulenti, tubo | mm longo intus ad medium uui sinus 1-glandulifero, caly anceo tis 0.2-0.3 mm longis; corolla doliiformi > mm p medium m lata, tubo 2.5 mm longo extus dense papillato- puberuleni lobis lineari-oblongis obtusis 0.7 mm ongis 0.4 mm | p » corniculis Incrassatis: st ib basi ll bacca Ru abii 4-5 mm lo onga 6 mm lata calyce persistenti nata; pyrenis eem 3.5 x 3.5 mm dorso con vexis ES conca Glabrous shrub 1.5 m tall. Stipular sheath shallow, 1.5 mm high, 4-6 mm wide, laterally 2-toothed on each side with the teeth broadly ovate-lanceolate, acute, 1-2 mm long, seri- ceous within. Leaves chartaceous, lanceolate- elliptic, broadest at the middle, acute at base and apex, 8-13 cm long, 2.5-5 cm wide. Petioles 9-20 mm long. Inflorescence ter- minal, many-flowered, 2-2. high ex- cluding the peduncle, 3-4 cm wide in anthe- sis, 2.5-3.5 cm high, 4-7 cm wide in fruit, the 5-8 axes irregularly and + trichoto- mously branched at the apices, in anthesis 3- 15 mm long, in fruit 10-23 mm long, 1-1.5 mm wide, the upper ones the shortest. Pe- duncle green, becoming purple, 10-12 mm long, 2 mm wide, glabrous. Bracts at the base of the axes lance-triangular with subulate tips, 0.6-1 mm long, glabrous. Flowers numerous, 5-merous, pedicellate, pedicels green becom- ing gray-purple, 0.5-1.5 mm long, articulate. Calyx and hypanthium 2.3 mm long, the hy- panthium barrel-shaped or shortly cylindric, 1.2 x 1.2 mm, minutely puberulent; calyx 1.2 mm high, 1.5 mm wide, minutely pu- berulent, the tube longer than the teeth, these triangular-lanceolate, acute, 0.2-0.3 mm long; interior of tube bearing 1 gland at the middle beneath each sinus and alternating with each calyx tooth. Corolla white, barrel- shaped, 3 mm long, 1 mm wide at base, 1.6 mm wide at the middle, 1.2 mm wide at the summit, tube 2.5 mm long, densely papillate- puberulent without, glabrous within; 5 lobes linear-oblong, obtuse, 0.7 mm long, 0.4 mm wide, conspicuously dorsally corniculate with a thickened appendage. Stamens 5, inserted at the base of the corolla; anthers 2 mm long, 0.3 mm wide. Style 0.8-1 mm long; stigmas linear-lanceolate, acute. Disk cupular, 0.3 mm high. Fruit green to purplish black, + bilobed, 4—5 mm high, 6 mm broad, shallowly crowned by the persistent calyx. Pyrenes 2, subglo- bose, 3.5 X 3.5 mm, convex dorsally, con- cave ventrally. This species is related to Psychotria cer- atantha Standley & Steyerm. of the Vene- zuelan Guayana, from which it differs in the smaller corollas with internally glabrous tubes with the stamens attached at the base. RUDGEA Rudgea corocoroensis Steyermark, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: 5-8 km of Yutaje set- tlement, 3 km west of Rio Coro Coro, along stream on south slope below pla- teau, E side of unnamed peak, o Serranía Yutajé, 5°40'N, 66°9'W, 700- 1,000 m, 10 Mar. 1987, R. Liesner & B. Holst 21827 (holotype, MO; isotype, VEN). Arbor 4-metralis; stipularum vaginis 2-3 mm longis pm Ra vel infra apicem 4 aculeis 5-7 rigidis 2-4 mm ongis munitis; foliis lanceolatis apice acu tis vel ieri sae basi obtuli vel Pei FAN glabris; nervis en us utroque latere 10-11 valde adscendentibus; enatio tertiaria obsoleta vel inconspicua; petiolis 3-7 mm un glabris. Inflorescentia terminali bid arc m: lata pedunculo excluso 2-3 cm longa 1.5-2 cm lata, seriebus verticalibus tribus Won ordinatis, axibus nfimis 4—5-verticillatis 4-8 mm longis dense pilosulis, aliis brevioribus; pedunculo 2-2.5 cm longo pilosulo; brac- teis quattuor sub verticillo infimo suborbicularibus 3 x 3 m marginibus ciliolatis; quoque axe in flores 6-7 ag- gregatos sessiles desinenti; bractea involucrali ciliolata pa- tenti sub basi florium; pedunculo 2-2.5 cm longo pilosulo; hypanthio brevicampanulato 1 mm longo 1.5 m calycis tubo 2 mm longo 1 mm lato glabro, lobis 4-5 inaequalibus lanceolatis subacutis vel obtusis majoribus -1.8 mm longis 0.7 mm latis, brevioribus 1 mm longis = mæ 350 Annals of the Missouri Botanical Garden dense ciliolatis; corolla infundibuliformi, tubo 2 mm longo orificio deorsumque e pilosulo extus glabro; in quinque lanceolatis- a acutis 4 mm longis 1.2 m latis extus glabris intus dense minuteque ea pe paullo exsertis, a: 1.2 mm lon Tree 4 m tall. Branchlets glabrous. Stipular sheath 2-3 mm long, glabrous, with 5-7 rigid aculeae arising at or just below the sheath summit. Leaf blade subcoriaceous, narrowly oblong-lanceolate, acute or obtuse at apex, obtuse to subobtuse at base, 7-10 cm long, 2—4 cm wide, glabrous on both sides, slightly revolute; lateral nerves 10-11 on each side, strongly ascending, scarcely anastomosing; tertiary venation obsolete or inconpsicuous. Petiole 3-7 mm long, glabrous. Inflorescence terminal, pedunculate, thyrsoid-paniculate, 2— 3 cm long excluding the peduncle, 1.5- wide, in 3 main verticillate tiers, the lowest ter 12-17 mm distant below the middle tier; lowest axes 4—5-verticillate, 4-8 mm long, densely pilosulous, the other verticels with shorter axes; each axis subtended by spread- ing, glabrous, ciliolate involucral bract, ter- minating in a congested group of 6—7 sessile flowers. Peduncle 2-2.5 cm long, pilosulous, 4 suborbicular bracts 3 x 3 mm subtending base of main lowest verticels. Hypanthium short-campanulate, 1 mm long, 1.5 mm wide, glabrous. Calyx tube 2 mm long, 1 mm wide, glabrous. Calyx lobes 4-5, unequal, lanceo- late, subacute to obtuse, the longer lobes 1.2- m long, 0.7 mm wide, the shorter 1 mm long, densely ciliolate. Corolla white, sub- infundibuliform, the tube 2 mm long, 1.5 mm wide at base, 2 mm wide above, densely pilose at orifice and 1 mm downward; lobes 5, lan- ceolate-oblong, acute with involute tip, 4 mm long, 1.2 mm wide, glabrous without, densely minutely gray papillate within. Stamens 5, slightly exserted, anthers 1.2 mm long; fila- ments 0.8 mm long, glabrous. Style 8 mm long, scabridulous-papillate, stigmas 2, 0.8 mm long. This taxon is related to Rudgea mori- chensis Steyerm. from which it differs in the glabrous hypanthium, conspicuously lobed ca- lyx, much shorter corolla tube, and shorter anthers. From R. bolivarensis Steyerm. it is distinguished by the lanceolate, longer calyx lobes and by the narrower, oblong-lanceolate leaf blades with shorter petioles. SABICEA Sabicea bariensis Steyermark, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: upper Rio Baria, mostly non- inundated area along riverside, 0?55'N, 66°16'W, 140 m, 9 May 1984, A. Gen- try & B. Stein 47314 (holotype, MO; isotype, VEN). Planta scandens, caulibus _strigosis; foliorum laminis atis basi 5 cm lata; floribus pedicellatis, pedicellis > 5 mm longis; calycis Ba inaequalibus, m acuminatis 3-4.5 mm longis 1.5-2.5 m adpresso- pubescentibus intus glabris infra sinum squa- mellis duabus muniti Vine with strigose pubescent stems. Leaves firmly membranous, elliptic-ovate, subacu- minate at apex, acute at base, the larger leaf blades 11-13 cm long, 5.5-7 cm wide, sparsely puberulous above, the midrib and ateral nerves strigose, the lower surface mainly glabrous except for strigillose midrib, the margins appressed-ciliolate; lateral nerves 10 each side, arcuate-ascending at 50-60? angle, impressed above, subelevated below; tertiary venation slightly evident, transversely connecting the secondary nerves; petioles 2— 3.5 em long, strigillose. Stipule reflexed, ovate, acute to acuminate, 10-12 mm long, 6-6.5 mm wide, strigose without, glabrous within. Inflorescence pedunculate, bracteate, 20—30- flowered, cymosely trichotomously branched, 2.3 cm long, 3.5 cm wide. Peduncle 4-5 mm long, densely antrorsely strigose. Lower axes of inflorescence 1.7 cm long with the flowers ca. 10-flowered. Lowest bracts subtending lowest axes of inflorescence 7-9 mm long, 3 mm wide, connate at their bases, divaricate, sparsely appressed-pubescent without, gla- brous within; bracts subtending other axes paired, 7 mm long, 2 mm wide, appressed- pubescent without. Flowers on pedicels 3-5 Volume 75, Number 1 1988 Steyermark 351 Flora of Venezuelan Guayana—lV mm long, these densely appressed pubescent. Calyx and hypanthium 5-7(-9.5) mm long, appressed-pubescent; hypanthium longer than the calyx lobes, 3.5-5.5 mm long, glabrous within; calyx lobes unequal, 4 larger and 1 smaller, the large ones erect-spreading, ovate, acuminate, 3-4.5 mm long, 1.5-2.5 mm to- ward the base, appressed-pubescent without, glabrous within, the smallest one 1.5 mm long, 1 mm wide, 2 unequal, elongate, dark squamellae situated below sinus of calyx lobes. Corolla narrowly cylindric, salverform, 14.5 mm long, the tube 10.5 mm long, 1.8 mm wide, densely antrorsely appressed pubescent without, glabrous within; 5 lobes lanceolate- ligulate, acute, 4 mm long, 1 mm wide at base, densely sericeous without, glabrous within. Anthers 2.8 mm long. YENEIUELA TERRITORIO FEDERAL Paratypes. AMAZONAS: 3-4 pstream from Neblina Base Camp, gravel banks of main Pie dine l of Rio Mawari 50'N, 66°10'W, 180 m, 4 Dec. 1984, Kral 71981 (MO, VDB, VEN); upper Cano Baria, swampy area between Rio Ma- warinuma and headwaters of Rio Baria, 0°52'N, 66°15'W, 130 m, 26 Mar. 1984, Liesner 16967 (MO, VEN). From other pedunculate species this differs from Sabicea grandiflora Steyerm. in the larger calyx lobes, nonarachnoid pubescence of the leaves, and divaricately spreading calyx lobes. From morillorum Steyerm. it is mainly distinguished by its nonarachnoid pu- bescence. R. Kral! and Maria das Graças de Lapa Wanderley: TEN NOVELTIES IN XYRIS (X YRIDACEAE) FROM THE PLANALTO OF BRAZIL ABSTRACT Ten taxa of Xyris (nine species and one variety), sent to the senior author for definitive treatment during the early 1980s, are herein described, dian di proposed for addition to the flora of Brazil. All are from the Brazilian Planalto: five (X. dia phanobracteata, X. jataian Goids, four (X. seubertii var. expla cae, peo . villosicarinata) are from Gerais, and one (X. Fuse is from Paraná. These new Xyris are presented alphabetically and each is contrasted with related specie Among the undetermined Xyridaceae from Brazil sent to the senior author for definitive treatment during the years 1980, 1981, and 1982 are ten that have long awaited publi- cation as new taxa. They are arranged al- phabetically. 1. Xyris diaphanobracteata Kral & Wanderley, sp. nov. TYPE: Brazil. Goiás: Rod. GO-12; km 5-10 ao sul de Alto Paraiso, campo arenoso, flor amarela, 24 May 1975, G. Hatschbach 36815 (ho- lotype, MBM; isotypes, US, VDB). Fig- ure Planta perennis, caules breves, radices graciles. Folia ensiformi-linearia, 15-20 cm longa, disticha, flabellate p aci pallide villoso- ciliatis vaginae carinatae, carinibus ciliatis, lateribus valde multicostatis, glabris, ur scatis, basin versus castaneis, rugulosis, marginibus in laminas gra- datim convergentibus, infimis irat expansis, tum abrupte ad basim dilatatis, aciebus ad basim anguste scariosis, antrorse longiciliatis, pilis albidis. Vaginae sca- porum basim versus ancipitae, icon glabrae, mul- ticostatae, ferrugineae, nitidae, a medio ad apicem acute carinatae, ; carimibus rufis, ciliatis, laminis revilue, obtusis. i graciles, recti, 4.5-6 dm sus ancipiti, costis longiciliatis, pilis albidis. Spicae late ovoideae, 1-1.5 cm longae, obtusae, multibracteatae, di laxis, fere arai spiraliter imbricatis, sub- e squarrosis, laceratis; area dorsalis Med pout bracteae steriles plures. Sepala latera- lia libera, valde curvata, leviter exserta, valde inaequi- latera, obla E. tenuia, 6-7 mm longa; ala carinalis lata, ciliato- fimbriolata. Laminae petalorum suborbiculatae avidulae. a numerosa, ellipsoidea, lampro-brunnea, prominente placenta basalis; semin longa, pallide vel me ua 12-14-c Plants short-stemmed perennials with fine roots. Leaves ensiform-linear, 15-20 cm long, distichous, flabellately spreading, strictly bas- al, longer than the scape sheaths; blades flat to slightly twisted, 3-5 mm wide, flattened, ! Robert od Herbarium, Department of Biology, Vanderbilt University, Box 1705, Station B, Nashville, LS. A. xin see 272 Dia. x das Graças de Lapa Wanderley, Instituto de Botánica, Caixa Postal 4005, 01000 São Paulo. “SP, Brasil. ANN. Missour! Bor. GARD. 75: 352-372. 1988. Volume 75, Number 1 1988 Kral 8 das Graças 353 Brazilian Xyris yellow-green, inconspicuously multinerved, papillose, thick-nerved toward margin, red- dish brown; apices abruptly narrowed, in- curved-acute, thickened; margins with edges pale villous-ciliate; sheaths ciliate-carinate, the sides strongly multicostate, smooth, yellowish- uscous, castaneous toward the base, rugu- lose, with margins gradually narrowing into the blades, gradually expanding below, abruptly dilated at base, there the edges nar- rowly scarious, antrorsely long-ciliate with white hairs. Scape sheaths ancipital toward base, pilose-ciliate, smooth, multicostate, red- dish brown, shining, acutely carinate from middle to apex, the carinae reddish, ciliate, the blades short, obtuse. Scapes slender, straight, 4.5—6 dm high, slightly twisted, nar- rowly 2-3-costate, elliptic in cross section, ancipital toward the apex, the costae long- ciliate with white hairs. Spikes broadly ovoid, 1-1.5 cm long, obtuse, the bracts numerous, loose, nearly uniform, spirally imbricate, sub- scarious, convex, ecarinate, broadly ovate, 5- 7 mm long, rounded, brownish yellow, mar- ginally squarrose, lacerate; dorsal area indis- tinct; sterile bracts many. Lateral sepals free, strongly curvate, slightly exserted, strongly inequilateral, broadly oblanceolate, thin, 6-7 mm long; keel wide, ciliate-fimbriolate. Petal blades suborbicular to reniform, 7-8 mm long, strongly erose, yellow. Staminodia bibra- chiate, the branches flattened, long-penicil- late. Anthers lanceolate-oblong, sagittate, ca. 2 mm long. Capsule turgidly obovoid, ca. 3 mm long, the placentation basal; seeds nu- merous, ellipsoid, 0.5 mm long, palely or deeply lustrous brown, prominently longitu- dinally 12-1 4-costate. This novelty most resembles X. pterygo- blephara Steudel, particularly by having long, pale ciliation on scape and leaf edges; disti- chous, flabellately spreading, flattened leaves; and concolorous bracts. Xyris diaphano- bracteata differs conspicuously in its broader spikes and uniform bracts (sterile ones more numerous) with thin, translucent texture and squarrose edges. The lateral sepals are broad- er, more strongly curvate, blunter, exserted, very inequilateral (rather than equilateral), and more prominently ciliate. 2. Xyris dissitifolia Kral & Wanderley, sp. nov. TYPE: Brazil. Paraná: Mun. Cam- piná Gde. do Sul, campo umido encosta de morro, flor amarella, alt. 1,500-1,700 m s/mar, 15 Jan. 1969, Hatschbach & Ckoczicki 20752 (holotype, MBM; iso- types, US, VDB). Figure 2 Planta perennis, caespitosa, radicibus gracilibus. Caules breves vel elongati, per bases laxas foliorum obtecti. Folia laxa, anguste ope 2-5 dm longa, iun) erecta ve l aut leviter fasi i p ndi val nibus latis integris nitidis, dorsis rotundato- plicatis, valde unicostatis, pe basin papillosis aut rugulosis. Vaginae scaporum laxa u versus conduplicatae tubulosae multicostae, nitidae, t api gra- j ci, ca oblongae vel late ellipsoideae aut obovoideae, pauciflorae, ca. l cm longae; bracteae laxe spiraliter imbricatae, convexae, pallide ferrugineae, marbinibus integris, tum erosis, ad apicem persaepe vil- pem ciliatis, EP cann sordidis; bracteae steriles 4-6, ae, ca. 4-7 mm lo - m late ova- na, ca. 3 mm lineari-fusiform longitudine minute striata. m longa, bicaudata, brunneola, Plants perennial, cespitose, slender-rooted. Stems short or elongated, covered by the lax bases of leaves. Leaves loose, narrowly linear, 2-5 dm long, distichous, erect or slightly fla- bellately spreading, longer than the scape eaths; es 2-5 times longer than the sheaths, subterete to angulate or slightly com- Annals of the 354 Missouri Botanical Garden TP. "Y p Posie Z, VT iif 75 V Volume 75, Number 1 1988 Kral 8 das Graças 355 Brazilian Xyris pressed, twisted, medially strongly sulcate and few-ribbed, olivaceous, smooth from middle to apex, rugulose-papillose and subterete to- ward base; apices gradually contracted, nar- rowly obtuse at tip, erect, usually thickened and triangulate; sheaths lax, wide, ecarinate or carinate, pale lustrous brown, gradually narrowing from base to tip, producing near the apex a scarious, oblong, obtuse ligule 3— 4 mm long, the margins broad, entire, shining, the backs rounded-plicate, strongly unicos- tate, papillose or rugulose except for the base. Scape sheaths lax, conduplicate toward base, tubular, multicostate, shining, twisted, open toward apex, with blades similar to leaf blades but shorter. Scapes slender, 3-6 dm high, subterete, or elliptic in cross section, ca. 1 mm thick, olivaceous, toward apex 1-2-cos- tate and finely striate, the costae smooth, toward base often obtusely triangulate or sul- cate. Spikes oblong-cylindric or broadly ellip- soid to obovoid, few-flowered, ca. 1 cm long; bracts loosely spirally imbricate, convex, pale reddish brown, papillose, rounded, with mar- gins entire, then erose, at apexes often villous- ciliate with sordid trichomes; sterile bracts 4— , obovate, ca. 4-7 mm long, the lowest shortest, prominently unicostate; fertile bracts few, oblong to obovate, up to 9 mm long, the inner ones navicular, carinate; dorsal area inconspicuous, short, narrowly elliptic, pale brown. Petal blades broadly ovate, ca. 1 cm long, subacute, slightly erose, yellow. Stami- nodes bibrachiate, the branches oblong, flat, long-penicillate at apex. Anthers broadly lan- ceolate, ca. 2 mm long, sagittate; filaments flat, ca. 3 mm long. Capsule oblong to ob- ovoid, ca. 5 mm long, apiculate, the placenta basal. Immature seeds linear-fusiform, 1.5 mm long, brownish, minutely longitudinally striate. Mature seeds not seen. Additional specimens examined. BRAZIL. PARANA: Serra Ibitiraquire, Abrigo 1 (Mun. Campina Gde. do Sul), 7 : campo umido encosta de morro, alt. 1,700 m, 23 Jan 1970, Hatschbach 23404 (MBM, US, VDB); Serra Ca- pivari Grande (mun. Campiná Gde. do Sul), encostas gra- minosas umidas de morro, alt. 1,500 m, 8 Feb. 1971, Hatschbach 26322 (MBM, US, VDB). This species, with its slender, long leaf blades, loose and distichous leaf sheaths along definite stems, papillose-backed bracts with small, pale brown dorsal areas, and villous sepal tips, shows some characters of Xyris filifolia Nilsson but has more stem, hairier sepal keels, and less evident dorsal areas. On the other hand, the vestite sepals liken it to X. rigida Kunth, or X. regnellii Nilsson, but leaf characters do not agree, nor do char- acters of fertile bract or scape. 3. Xyris itambensis Kral & Wanderley, sp. nov. TYPE: Brazil. Minas Gerais: Serra do Ispinhaco, sandy soil, eastern slopes of Pico do Itambe; elev. ca. 1,550 m; sandstone outcrops with shrubby vege- tation, adjacent grassy slopes, and valley of small stream; mostly sandy soil but with overlying black humus in many places, 12 Feb. 1972, W. R. Anderson, M. Stieber, J. H. Kirkbride, Jr. 35879 (holotype, UB; isotypes, NY, US). Fig- ure 4 Planta perennis, ree a ata Liege glabra; caules breves; radices graciles. Folia erecta vel leviter expansa, 1.5-3 m basalia, vaginis scaporum viande torta laminae vaginis 3-6-plo lon teretes vel ‘subteretes, filiformes, 0. 5-0. 6 mm crassae, fimbriatis, in medio longitudine multicostatae, ad apicem ligulam erectam podes acutam 4-6 mm longam fas- cientes. nae scaporum ad basim teretes, laxe con- duplicatae, mulli tortae et leviter flexuosae, pur- pureae vel pallide ferrugineae, apicem versus apertae subteretes, laeves, ecostati vel unicostati, costibus glabris. Spicae late ellipsoideae vel anguste obovoideae, 6-8 mm longae, 4-5 mm crassae, pauciflorae, pluribracteatae, E FIGUR blade. a ‘on at leaf sheath—blad e junc Xyris op eae intei 36815).—a. Habit sketch.—b. Leaf tip. —c. iaces of mid e. Leaf base. —f. Spike. —g. Sector of scape m below apex.—h. Fertile bract. —i. Lateral sepal. lu “Peral and stamen, stylar apex, staminode, Papi ‘beard h hair.— k. Seed. 356 Annals of the Missouri Botanical Garden Z om Ü ÉS < thi UA LA ye M ad LX TREE SHEET Bini i Prae EHE Kis, Mo, £t CE ` > > Volume 75, Number 1 1988 Kral 8 das Graças 357 Brazilian Xyris rain aie pluribus, convexis, inferioribus subor- bicular vel late ovatis, ongis, superioribus a eae usque ad 5 mm irse bracteae fertiles late oblongae vel obovatae, ad 5-6 mm longae, marginibus tenuibus, laceratis s, ad apicem parce llosa. intimis m teolae. Antherae lanceolato- hid 1.5 ittatae, filamentis longiores. Staminodia hachi. brachiis. eras, longipenicillatis. Pla- centa centralis. Semina matura non visa sed immatura cylindrico-fusiformia, 1.2 mm longa Plants perennial, densely cespitose, smooth; stems short; roots slender. Leaves erect to slightly spreading 1.5-3 dm long, strictly bas- al, longer than the scape sheaths, twisted and flexuous, lustrous; blades 3-6 times longer than sheaths, terete or subterete, filiform, 0.5— 0.6 mm thick, green, longitudinally finely striate, minutely rugulose toward base; tips gradually narrowing, narrowly conic; sheaths ecarinate, abruptly dilated at base, shining red-brown, fimbrio-ciliate, prominently lon- gitudinally costate, gradually narrowing above, lustrous brown, with margins broad, thin but firm, entire to sparsely villous-fimbriate, at middle longitudinally multicostate, at apex producing an erect, oblong, acute ligule 4-6 mm long. Scape sheaths terete at base, laxly conduplicate, multicostate, twisted and some- what flexuous, purplish or pale red-brown, open toward apex, with margins broad, entire, purplish, and at apex cuspidate. Scapes very slender, 3.5-5.5 dm high, twisted and flex- uous, olivaceous, subterete, smooth, ecostate to unicostate, with costae smooth. Spikes broadly ellipsoid to narrowly obovoid, 6-8 mm long, 4-5 mm thick, few-flowered, sev- eral-bracted, the sterile bracts several, con- vex, the lower ones suborbicular to broadly ovate, 2-3.5 mm long, the upper sterile bracts navicular, up to 5 mm long; fertile bracts broadly oblong to obovate, to 5-6 mm long, with thin, lacerate margins, sparsely red-vil- losulous at apex, the inner ones strongly pli- cate; dorsal area elliptic, ca. 2 mm long. Lat- eral sepals free, curvate, oblong, ca. 5-6 mm long, acute, slightly inequilateral, the keel narrow, red-fimbriolate from middle to tip. Petal blades oblong, 6-6.5 mm long, broadly acute, entire, yellow. Anthers lanceolate-ob- long, 1.5 mm long, emarginate and sagittate, longer than the filaments. Staminodia bibra- chiate, the branches subterete, long-penicil- late. Placenta central. Immature seeds cylin- dric-fusiform, 1.2 mm long. Mature seeds not seen. In habit the new species resembles X. fili- folia. Nilsson, a slender species which also has small but prominent dorsal areas and strongly ligulate leaves with very slender and terete blades. The scapes, unlike those of X. filifolia, are terete, the tips of the fertile bracts are villosulous with red hairs, and the lateral sepal keels are densely villosulous above the middle. Xyris itambensis relates also to the complex involving X. rigida Kunth, but is more slen- der, has fewer flowers, and has subequilateral (rather than strongly inequilateral) sepals. 4. Xyris jataiana Kral & Wanderley, sp. nov. TYPE: Brazil. Goiás: Mun. Jatai Rod. Jatai-Caiapenia, km 50 (mun. Jatai), Goiás; flor amarela, do brejo, 24 Jul 1977, G. Hatschbach 40060 (holotype, MBM; isotypes, US, VDB). Figure 5. nta perennis, glabra, densicaespitosa; radices gra- O % £ = s = Uu et c ” e [md = = a ° = ° "E o E £ 4 qs 0 n D c talico similes, luteoli vel brunneoli, persaepe nitidi, quam in centro laminae crassiores; vaginae ecarinatae, anguste laminis distincte latiores, infime gra ginae scaporum laxae, plerumque apertae, tortae, basin E 2. steafat mid blade. sepals.—i. Petal blade, stamen. i e—sheath juncti Xyris dissitifolia dap end & Ckoczicki B —a. Habit sketch.—b. Leaf apex.—c. Secto —d. Lea on.—e. taminode and Re Der hair.—k. Stylar apex. —l. — f. Spike. — g. Fertile bract. —h. Two lateral eed. 358 Annals of the Missouri Botanical Garden CAE pana Zam Xyris seubertii var. espinhacae (Irwin et al. — —a. Habit sketch.—b. Leaf apex.—c. Sector of mid blade.—d. Sheath-blade junction.—e. Leaf base. —f. Spike. —g. Sector of scape.—h. Fertile bract. —i. Lateral sepals.—j. Petal and stamen, staminode with a view of beard hair, stylar apex. Volume 75, Number 1 1988 Kral & das Gragas 359 Brazilian Xyris versus paucicostatae, castaneae, apicem versus carinatae, pressis vel wae ob- r torti, teretes, 5-6 dm alti, in eas gradatim transientes; bracteae fertiles late obovatae, ca. 4.5-5 mm longae, apice rotundatae; area dorsalis distincta, ovata ses elliptica, ca. 1.5-2 mm longa, olivacea vel atroferruginea. Sepala lateralia libera, leviter inae- m longa, acuta; angusta, apicem versus aminae petalorum late ellipticae, 5-5.5 mm longae, luteolae, apice anguste rotundatae, erosae. Stami- nodia bibrachiata, brachiis longipenicillatis. Antherae ob- longae, ca. 2 mm longae. Capsula matura non visa. Fe] E: S = % = m ° = aa £ a e c ° = = ° = ° O ° = em et m e Plants perennial, glabrous, densely cespi- tose from slender roots. Rhizomes ascending, their bases covered by persistent old leaves. Stems short or elongated. Leaves of rhizomes and stems subdistichous, rigid, twisted and flexuous, 2-4 dm long, longer than the sheaths of the scapes; blades 5-8 times longer than sheaths, twisted, narrowly linear, 0.8-1.5 mm wide, flattened, yellowish green, longitudinal- ly 4-5-nerved, sulcate below; apices some- what contracted, asymmetrically narrowly rounded, callused; margins wirelike, yellowish to brown, often shining, thicker than the cen- tral part of the blade; sheaths ecarinate, nar- rowly folded, deeply castaneous, lustrous, with thin, entire margins, these gradually con- verging, at the apex forming an erect, acute ligule 10 mm long, this distinctly wider than the blade, gradually dilating below. Sheaths of scapes lax, largely open, twisted, few-ribbed toward the base, castaneous, carinate toward the apex, with blades short, erect, compressed or triangulate, obtuse. Scapes erect, slightly twisted, 5-6 dm long, brown below, com- pressed, terete toward the apex, ecostate, yel- low-green. Spikes many-flowered, many-brac- teate, narrowly obovoid, 1-1.5 cm long, attenuate; bracts convex, ecarinate, reddish brown, spirally imbricate, rounded, entire to minutely erose; sterile bracts several, narrow- ly obovate, 2-4 mm long, shorter than the m long, apically rounded; dorsal area dune ovate to elliptic, ca. 1.5-2 mm long, olivaceous to ark reddish brown. Lateral sepals free, slightly inequilateral, narrowly oblanceolate, ca. 5 mm long, acute; carinal keel ciliate, toward the base narrow, toward the apex wide, firm. Petal blades broadly elliptic, 5-5.5 mm long, yellow, narrowly rounded, erose. Sta- minodia bibrachiate, the branches long-hairy. Anthers oblong, ca. 2 mm long. Mature cap- sule not seen; placentation basal. Seeds not seen. This species, with its distinct, elongate lig- ule broader than the leaf blade base, multi- flowered spikes with (often) reddish brown dorsal areas, and free lateral sepals with ciliate keels, shows affinities to the rare owns- iana Smith, a species thus far known only from Para. However, X. jataiana has a lon- ger ligule; its leaf blade margins are more broadly incrassate; the leaf apecies are bluntly callus tipped; the flowers are larger; and the lateral sepals are more nearly equal. Mature fruit and seeds of either of these species are unavailable. Ə. Xyris lutescens Kral & Wanderley, sp. nov. TYPE: Brazil. Goiás: Mun. Posse, Nova Vista, brejo da borda de chapada, 8 Oct. 1976, G. Hatschbach 39021 (holotype, MBM; isotypes, US, VDB). Figure 6 Plan nta perennis, solitaria vel cespitosa, basibus bulbosis, atrocastaneis. graciles. Caules breves. Folia en- siformi-linearia, 10-20 cm longa, torta, erecta vel leviter expansa, ear bus longi e planae vel leviter torta bie vaginis longiores, 1.2-2.2 mm eo irent contrac- acuti, incras en mar gines angusti, leviter incrassati, nitidi luteoli; vaginae ecarinatae vel asym- metrice carinatae, multicostatae, lateribus stramineis, transversim paiesi sis, marginibus pallide longiciliatis, in laminas gradatim convergentibus, infime gradatim expan- sae tum abrupte ad basim dilatatae, acies scariosae. Va- ginae scaporum laxae, luteo- -carinatae, tortae, acute 2- multi-costatae laminis d brevi- par dm alti, era 8 mm crassa, attenuata, bracteis spiraliter brit: tenuibus, pallide he ecarinatis, marginibus tenuissimis 360 Annals of the Missouri Botanical Garden idm FIGURE 4. Xyris itambensis (Anderson et al. 35879). —a. Habit sketch. —b. Leaf apex.—c. Dorsal (left) and side duh views of leaf-sheath junction.—d. Leaf base.—e. Spike. —f. Fertile bract.—g. Lateral sepal. —h. Petal blade and stamen, staminode, enlarged version of beard hair, stylar apex. Volume 75, Number 1 1988 Kral 4 das Graças Brazilian Xyris et profunde iius bracteae steriles ovatae vel subor- biculatae, 2.5-3.5 mm longae, fertilibus breviores et in eas gradatim Mobi suae bracteae fertiles late obovatae vel suborbiculatae, usque 5 mm longae; area dorsalis indistincta. Sepala lateralia libera, valde Ten oblanceolata, ca. 5 mm longa, acuta, tenuissima; ala cari- medium irregulariter dates. Laminae petalorum late obovatae, ca. 5 mm longae, late uteolae. Staminodia bibra- chiatis, brachiis longipenicillatis, erectis, pilis clavatis. An- therae late oblongae, mm longae, retusae et sagit tatae, ens ca. 2- o breviores. Capsula per leviter compressa, c m longa, acuta; placen basalis. Semina axlipdrico: -ellipsoidea a, 0.8-0.9 mm Eus translucida, pallide ferruginea, longitudinaliter et promi- ente 8-costata. Plants perennial, solitary, with bulbous, deep chestnut bases. Roots slender. Stems contracted. Leaves ensiform-linear, 10-2 cm long, twisted, erect to slightly spreading, longer than the scape sheaths; blades flat or slightly twisted, 3-4 times longer than the sheaths, 1.2-2.2 mm wide, strongly flattened, yellow-green, punctate, inconspicuously lon- gitudinally multinerved; apices rather abrupt- ly contracted, incurved-acute, thickened; margins narrow, slightly incrassate, shining, yellowish; sheaths ecarinate or asymmetri- cally carinate, multicostate, with sides stra- mineous, transversely rugulose, the margins long-ciliate with pale hairs, narrowing grad- ually into the blades, gradually spreading be- low, then abruptly dilating at the base, the edges scarious. Sheaths of scapes loose, yel- lowish-carinate, twisted, sharply 2-many- ribbed; blades similar to blades of leaves but short. Scapes slender, flexuous, twisted, sub- terete, yellow with tints of brown, 4-6 dm high, strongly 2-many-costate and angulate, ca. ] mm thick toward the apex, the costae papillose to tuberculate or glabrous, the in- tervals punctate. Spikes many-flowered, cy- lindro-ovoid to narrowly ellipsoid, 1-2 cm long, 5-8 mm thick, attenuate, the bracts spirally imbricate, thin, pale yellow-brown, ecarinate, the margins very thin and deeply lacerate; sterile bracts ovate to suborbicular, 2.5-3.5 mm long, shorter than the fertile bracts and grading into them; fertile bracts broadly obovate to suborbicular, to 5 mm long; dorsal areas indistinct. Lateral sepals free, strongly inequilateral, oblanceolate, ca. 5 mm long, acute, very thin; keel wide, ir- regularly ciliate-lacerate from the middle to the apex. Blades of petals broadly obovate, ca. 5 mm long, broadly rounded, slightly erose, yellowish. Staminodia bibrachiate, the branches pubescent with long, clavate tri- chomes. Anthers broadly oblong, ca. 1 mm long, emarginate or sagittate, half as long as the filaments. Capsule ellipsoid, slightly com- pressed, ca. 2.5 mm long, acute; placentation basal. Seeds cylindro-ellipsoid, 0.8-0.9 mm long, translucent, pale red-brown, longitudi- nally and strongly 16-18-ribbed. This species, with sheath edges gradually converging to blade bases, and lacking dorsal areas on its thin, pale-margined, broad, rag- ged bracts, appears to blend characteristics of the species complex around X. ciliata Thunb. The yellow tints of both spike and complanate foliage are distinctive, together with the strongly and sharply multicostate scape. In most xyrids the anthers are longer than the free portions of the filaments, or at least these are equal; in X. lutescens the anthers are considerably shorter than the fil- aments. 6. Xyris obcordata Kral & Wanderley, sp. nov. TYPE: Brazil. Minas Gerais: Serra do Espinhaco, wet sand among outcrops, cut-over gallery forest and adjacent out- crops, Rio Jequiti, ca. 25 km E of Dia- mantina, elev. 790 m, 17 Mar. 1970, H. S. Irwin, S. F. da Fonseca, R. Souza, R. Reis dos Santos & J. Ramos 27763 (holotype, UB; isotypes, NY, US). Fig- ure 7 Planta perennis, densicaespitosa, basibus infirmis. Caules breves. Radices graciles. Folia laxa, 1-2 dm longa, in- distincte disticha, glabra, vaginis scaporum longiora; lami- nae planae, valde compressae, vaginis 1-2-plo longiores, 2-3.5 mm latae, longitudine indistincte multinervosae, bus scariosis, integris, In laminas gr: vel anguste elliptici, longitudine ea acute striato- nervosae, a medio ad apicem valde stati et vulgo ancipiti, costis compressis, latis, nw cia antrorse albo- Annals of the Missouri Botanical Garden FIGURE 5. leaf at mid blaa ciliatis. Spicae multiflorae, late ovoideae, 7-9 mm longae, om m latae; atae, laxe spiraliter imbricatae, pallide fulvae, marginibus amplis tenuissimis, ás Vis jataiana (Hatschhach 40060).—a. Habit sketch.—b. Bract.—c. Leaf ap e.—e. Upper part of leaf sheath at blade base.—f. Leaf base.—g. Spike. dd Tue sepal.— i. Petal blade, pe enlarged cells of staminode beard, stylar apex. x.—d. Sector of ad apicem ferrugineis; bracteae steriles 5-6, .9 mm longae; bracteae fertiles usque ad 4.5 mm ae area dorsalis late vel anguste elliptica, ca. 34 bracteis breviori, viridis. Sepala lateralia libera, spathulata, ca. 4 mm longa, tenuia, ala carinali tenuissima, subintegra. Kral & das Gragas 363 Volume 75, Number 1 1988 Brazilian Xyris FIGURE 6. Xyris lutescens (Hatschbach 39021).—a. Habit sketch. — b. Leaf tip. —c. Sector of mid blade. — d. Leaf sheath —e. Spike.—f. Sector of mid scape.—g. Fertile bract.—h. Lateral sepal.—i. Floral parts.—j. See ovoidea, c mm longa, oe nitida, promi- tata. Laminae petalorum obovatae, ca. 3 mm longae, rotun- datae, luteolae. Staminodia bibrachiata, brachiis com- nente longitudinaliter 14-16-cos pressis, longipenicillatis. Antherae oblongo-lanceolatae, ca. ongae, valde emarginatae et sagittatae, filamentis Pl ‘al. d l : h aliados, Capsula obovoidea, ca. 2 mm lo onga, placenta ants perennia ensely cespitose, the centralis. Semina numerosa, turgide et late asymmetrice bases soft. Stems short. Roots slender. Leaves 364 Annals of the Missouri Botanical Garden lax, 1-2 dm long, indistinctly distichous, smooth, longer than the scape sheaths; blades flat, strongly compressed, 1-2 times longer than the sheaths, 2-3.5 mm wide, longitu- dinally indistinctly multinerved, green; apices abruptly narrowed, narrowly acute, slightly thickened, incurved; margins thin, uniformly finely white ciliate; sheaths carinate (carinae entire), stramineous, the margins scarious and entire, narrowing gradually into the blades, gradually dilating toward base. Scape sheaths lax, twisted, low-costate, like leaves in color and with similar tips. Scapes slender, 2.5-4 dm high, ca. 1 mm broad, greenish, in cross section elliptic or narrowly elliptic, longitu- dinally finely and sharply striate-nerved, from middle to apex strongly bicostate and com- monly ancipital, the costae compressed, wide, finely antrorsely white ciliate. Spikes many- flowered, broadly ovoid, 7-9 mm long, soft, if pressed then flattened, 6-7 mm wide; bracts ecarinate, ovate to broadly obovate or sub- orbicular, obcordate, loosely spirally imbri- cate, thin, pale tan except ferruginous apex, the ample margins very thin, friable; sterile bracts 5-6, 2-2.5 mm long; fertile bracts up to 4.5 mm long; dorsal area broadly to nar- rowly elliptic, about 25 as long as the bracts, green. Lateral sepals free, spathulate, acute, ca. 4 mm long, thin, the keel very thin, sub- entire. Petal blades obovate, ca. 3 mm long, rounded, yellowish. Staminodia bibrachiate, with branches flattened, long-penicillate. An- thers oblong-lanceolate, ca. 1.5 mm long, strongly emarginate and sagittate, longer than the filaments. Capsule obovoid, ca. 2 mm long, placentation central. Seeds numerous, irl and broadly asymmetrically ovoid, ca. 0.3 mm long, deep reddish ing, prominently longitudinally 14—16-ribbed. rown, shin- This soft-based, lax-leaved plant appears to be in the same complex as X. mima Smith & Downs and X. moraesii Smith & Downs; it is most similar in habit to the latter, differing mainly in the broader leaf blades, slightly narrower spikes, coloration of bract apex, and subentire sepal keels. Xyris obcordata is tall- er than the former and has broader spikes and entire (vs. ciliate) sepal keels. It is unlike either in its ancipital, broadly bicostate scape and reddish coloration of the thin, friable apex of the obcordate bracts. 7. Xyris pranceana Kral & Wanderley, sp. nov. TYPE: Brazil. Goiás: Chapada dos Veadeiros, 2 km from ror Cer- rado, common, 18 July 1964, Prance & N. T. Silva 581 = (holotype, UB; isotypes, NY, US, VDB). Figure 8. Planta perennis, soa pase glabra, basibus bul- bosis, castaneis, rhizomate nullo vel bre vi horizontali, per bases persistentes dE fud obtecto. ciles. Folia rigida, 3 dm longa, erecta, torta et leviter flexuosa, vagini is Reet longiora; laminae teretes vel 5-plo longiores, 0.3-0.5 mm cras- eas "n atim transientes; bra oblongae, ca. 5 mm lon infernis tenuiores, ad a brevilinearis, ferrugin Ee usd Sepala lateralia libera, valde inaequilatera, lineari- -elliptica, curvata, 5-5.5 mm longa, acuta; ala carinali firma, a basi integra, a medio ad apicem ciliata, ad apicem alis et pilis ferrugineis. Lami- mm longae, eatae. Staminodia bi- n ca. 1.5m Capsula Tesi mm longa, atroferruginea, minute sed acute multicostata. Plants perennial, densely cespitose, smooth, the bases bulbous, castaneous, the rhizome none or short and horizontal, covered by chaffy bases of old leaves. Roots slender. Leaves stiff, 1.5-3 dm long, erect, twisted and somewhat flexuous, longer than the scape sheaths; blades terete or subterete, 4-5 times longer than the sheaths, 0.3-0.5 mm thick, shallowly sev- eral-ribbed, often 1-2-sulcate, reddish brown, or brownish, or greenish-brown; apices Volume 75, Number 1 Kral 4 das Gragas 365 1988 Brazilian Xyris FIGURE 7. Xyris obcordata (Irwin et al. 27763).—a. Habit sketch. —b. Leaf apex.—c. Sector of leaf at mid blade.—d. Sector of leaf at blade—sheath junction.—e. Leaf base. —f. Very enlarged small sector of leaf blade edge.—g. Spike. —h. Fertile bract. —i. Lateral sepal.—j. Petal blade and stamen, staminode, stylar apex.—k. Seed. 366 Annals of the Missouri Botanical Garden strongly narrowed, conic-subulate; sheaths shining, ecarinate at base, carinate toward apex, reddish brown to brown, much wider than the blades, strongly ligulate, the ligule firm, flat, oblong, 4-5 mm long, apically ob- tuse, the margins thin, entire, gradually ex- panding downward, then abruptly dilating at base. Scape sheaths twisted, low-multicostate, lustrous reddish brown, with short, subulate lades. Scapes commonly 2.5-4.5 dm long, erect, somewhat twisted, terete, 0.5 mm thick, inconspicuously longitudinally striate, oliva- ceous. Spikes ellipsoid, 8-10 mm long, at- tenuate, many-flowered, the bracts convex, ecarinate, pale reddish brown or brownish, shining, thin, entire, loosely spirally imbricate; sterile bracts 5-7, elliptic to ovate, the lowest elliptic, much shorter than the fertile bracts and grading gradually into them; fertile bracts ovate to broadly oblong, ca. 5 mm long, the upper ones more folded and thinner than the lower ones, all short-carinate at apex; dorsal area short-linear, reddish brown, apical. Lat- eral sepals free, strongly inequilateral, linear- elliptic, curvate, 5-5.5 mm long, acute; keel rm, entire at base, sparsely (deciduously) ciliate from middle to apex, the keel and hairs at the sepal apex reddish brown. Petal blades narrowly obovate, ca. 5.5 mm long, yellow, acute, sparsely erose, the base cuneate. Staminodia bibrachiate, the branches long- penicillate. Anthers oblong-lanceolate, ca. 1.5 mm long, sagittate. Capsule ca. 4 mm long, lanceoloid, the tip subulate; placenta central. Seeds ovoid to ellipsoid, ca. mm long, reddish brown, minutely but sharply multi- costate. While the dorsal areas of the bracts of this distinctive species are small, they are dark in contrast to the rest of the bract and are thus evident. The slender, terete leaf blades are much narrower than the dilated, strongly lig- ulate sheaths. The only hairs on the plant are those found scattered along the keel apex of the lateral sepals. 8. Xyris sceptrifera Kral & Wanderley, sp. nov. TYPE: Brazil. Goiás: Serra dos Cristais, 17°S, 48'W, creek margin ca. 5 km S. of Cristalina, Goiás, elev. 1,200 m, 3 Mar. 1966, H. S. Irwin, J. W. Grear, Jr., R. Souza € R. Reis dos Santos 13410 (holotype, UB; isotypes, NY, US). Figure 9. Planta perennis, caespitosa, basibus bulbosis castaneis; radices graciles. Folia rigida, glabra, solum basalia, valde lo e n glabrae; atim contracti, i, pagina superiore concava; margines (si evidentes) incrassati, glabri; vaginae basin versus rotundatae, longitudine multicostatae, atro- castaneae, nitidae ae, laminis subteretibus, meee acutis. Scapi lineares, recti, vade Ule. ipd t flexuosi, ca. 7- d alti, teretes, gs dilute virides. ate mind m vel tru tica, pallide viridis. Sepala lateralia libera, valde inaequi- latera, elliptica, ca. 5.5 mm longa, valde incurvata, leviter exserta; ala carinalis ipii a basi ad medium integra, a medio retrorse fimbriata, vel antrorse dense fimbriata, o ferrugineis. aa ae pe- talorum obovatae, ca. 4 mm longae, ad apicem anguste rotundatae, dentatae, luteolae. Staminodia bibrachiata, brachiis elongatis Vic ees AO oblongae, ca. 2 mm longae, profunde sagittata longiores. AR Ae DUE ca. 2. 5 mm lo centralis. Semina longa, ferruginea, longitudine valde multicostata. Plants perennial, cespitose, with casta- neous bases, bulbous; roots slender. Leaves rigid, smooth, strictly basal, strongly twisted and flexuous, 5-6 dm long, erect or slightly spreading or excurved, longer than the scape sheaths; blades subterete or angulate, few- nerved and sulcate, m wide, yellow- green, 3-4 times longer than the sheaths; apices erect, gradually narrowed, aristate, with upper surface concave; margins (if evident) thickened, smooth; sheaths rounded toward base, longitudinally multicostate, dark cas- taneous, lustrous, entire-margined, narrowing gradually into the blade, conduplicate at apex, toward base gradually widening then abruptly dilating. Scape sheaths loosely conduplicate Volume 75, Number 1 Kral 8 das Graças 367 1988 Brazilian Xyris S mm V CHA FIGURE 8. Xyris e (Prance & Silva 58195).—a. Habit sketch.—b. Leaf apex.—c. Sector of mid blade.—d. Leaf base.—e. Spike. —f. Fertile bract.—g. Lateral sepal.—h. Flower parts. —i. Seed with attached funicle. 368 Annals of the Missouri Botanical Garden or open, slightly carinate, the blades subter- ete, erect, acute, short. Scapes linear, shal- lowly multicostate, strongly twisted and flex- uous, ca. 7-10 dm high, terete, 1-2 mm thick, smooth, pale green. Spikes ovoid, el- lipsoid or cylindric, 1-2.5 cm long, 5-7 m thick, with many spirally imbricate bs convex-backed, oblong to obovate, brown, shallowly erose or entire; sterile bracts few, oblong, slightly shorter than the fertile bracts and grading into them; fertile bracts 5-5.5 mm long, convexly rounded, broadly rounded to truncate apically, the dorsal area elliptic, pale green. Lateral sepals free, strongly in- equilateral, elliptic, ca. 5.5 mm long, strongly curvate; carinal keel narrow, entire from base to middle, retrorsely fimbriate at middle, spreadingly or antrorsely densely fimbriate at apex, the trichomes ferrugineous. Petal blades obovate, ca. 4 mm long, yellow, at apex nar- rowly rounded, toothed. Staminodia bibra- chiate, the branches elongate, long-penicil- late. Anthers oblong, ca. 2 mm long, deeply sagittate and retuse, longer than the fila- ments. Capsule cylindric, ca. 2.5 mm long; placentation central. Seeds narrowly oblong- fusiform, 1. . mm long, reddish-brown, longitudinally strongly multiribbed. Additional specimens examined. BRAZIL. Golás: k bank, ca. 6 mi. S of Cristalina, el. 1,175 m Nov. 1965, H. S. Irwin, R. Souza & R. Reis dos Santos 9947 (NY, US). MINAS GERAIS: Morro das P a da Nees elev. da Fonseca, i aes R. Reis dos FUROR & 7 Ram 25625 (NY, L This tall, bulbous-based species with twist- ed and flexuous leaves and scapes plainly relates to Xyris goyazensis Malme, and a full study may later reveal a varietal rela- tionship. However, the fertile bracts, in ad- dition to being more numerous in a larger spike, lack an apical tuft of villous hairs, and the lateral sepals, while fimbriate, have nar- rower keels which are entire toward the base. The apex of the leaf, unlike that of X. go- yazensis, is spinulose. Xyris sceptrifera dif- fers from the closely related X. veruina Malme by having fertile bracts with more obtuse tips, smaller and narrower dorsal areas, and nar- rower sepal keels which are less copiously fimbriate. 9. Xyris seubertii Nilsson var. espinha- cae Kral & Wanderley, var. nov. TYPE: Brazil. Minas Gerais: Serra do Espinha- co, wet sand, sandstone precipices and adjacent cerrado, ca. 18 km west of Grào Mogol, elev. 950 m, 21 Feb. 1969, H. S. Irwin, R. Reis dos Santos, R. Souza & S. F. da Fonseca 23667 (holotype, UB; isotypes, NY, US). Figure 3. Planta Pile caespitosa, drea subbulbosis; canles breves; radices graciles. Folia flal contracti, tenues, ; vaginae praete r bases carinatae, carinis cari iiis purpureis vel ferrugineis, glabris, valde multic basim versus papillosis, margini nibus edis in pee gradatim convergentibus, infime gra datim expansis, tum dos pte ad basim dilatatis castaneis vel doses Scari. gr tort 3 dm alti, su te ae Scum minati, ariosis y latip, a dare xe apicem ibe olatis; bracteae steriles infimo oblongo vel ste uos ato e mm bus carinato, areis dorsa libus linearibus, [ers excurrentibus, pari intimo lat ores, obovato a ongo; bracteae tiles *, Era a ca. 4. 5-50 mm eee serosa? plicatae, area m longa, lobis cae, ca. 5 mm longae, ac minodia bibrachiata, brocha. linearibus compressis api- cibus et lateribus longipenicillatis. Antherae lineari-oblon- gae, ca. 2 mm longae, refusae et sagittatae, filamentis 2-plo longiores. Capsula anguste ellipsoidea, ca. 3.5 mm longa; placenta centralis. Semina matura non visa. Plants perennial, cespitose, with subbul- bous base; stems short, the roots slender. Leaves flabellately spreading, narrowly linear, commonly 5-7 cm long, mostly longer than scape sheaths; blades 1-3 times longer than sheaths, flat, 0.8-1.2 wide, smooth, strongly flattened, olivaceous, longitudinally indistinctly nerved; apices gradually, then abruptly, narrowed, narrowly acuminate, sca- brous, erect; margins thin, antrorsely scabrid; sheaths except for bases smooth-carinate, the Volume 75, Number 1 1988 Kral 8 das Graças Brazilian Xyris 369 — Z = < — m Eas, ç C = = — —— 22 — n = — —[[x e — RE FIGURE 9. Xyris sceptrifera (Irwin et al. 13410).—a. Habit sketch. —b. Leaf apex.— c. Sector of leaf at mid blade.— d. Sector of leaf showing sheath and blade junction. —e. Leaf base. —f. Spike. —g. Fertile bract showing exserted lateral sepal tips. —h. Petal, stamen, stylar apex, staminode.—i. Seed.—j. Lateral sepal. 370 Annals of the Missouri Botanical Garden sides purplish or ferruginous, smooth, strong- ly multicostate, papillose toward base, the margins gradually narrowing into the blades, below gradually widening then dilating abruptly at base, castaneous or deep red- brown. Sheaths of scapes lax, twisted toward base, multicostate, brown, toward apex open, green, with margins broadly scarious, with blades as on foliage leaves but short. Scapes slender, straight or somewhat twisted, 2-3 dm high, subterete, olivaceous, ecostate or finely 1—3-costate, the costae smooth or finely scabrid. Spikes few-flowered, ovoid or ellip- soid or obovoid, 5-7 mm long, the bracts convex, ecarinate or carinate, subdecussate, pale brown, the margins broad, entire, scar- ious, rusty fimbriolate at apex; sterile bracts 4, the lower pair oblong or narrowly obovate, ca. 3 mm long, carinate, with linear, short- excurrent dorsal areas; inner pair wider, ob- ovate, ca. 3.5 mm long; fertile bracts 4, ob- long, ca. 4.5-5 mm long, rounded-plicate, the dorsal area lanceolate, greenish, 3-4 mm long. Lateral sepals ca. 1⁄4 connate, inequi- lateral, curvate, ca. 5 mm long with lobes oblong, acute, the keel narrow but strong, densely rusty fimbriolate from middle to apex. Petal blades elliptic, ca. 5 mm long, acute, subentire, yellow. Staminodia bibrachiate, the branches linear, flattened, apically and lat- erally long-penicillate. Anthers linear-oblong, ca. 2 mm long, retuse and sagittate, 2 times as long as filaments. Capsule narrowly ellip- soid, ca. 3.5 mm long, the placentas central. Mature seeds not seen. The new variety appears to be a small variant of X. seubertii, agreeing with that species by having greenish, excurrent dorsal areas and connate lateral sepals with densely rusty-pubescent keels. The bracts are sparse- ly but definitely villosulous (vs. glabrous) api- cally; the placentation is definitely free-cen- tral rather than basal as in the type variety. 10. Xyris villosicarinata Kral & Wan- erley, sp. nov. TYPE: Brazil. Minas Ge- rais: wet depression near creek, grazed campo and cerrado, upland valley, Serro do Itabirito, Minas Gerais, ca. 45 km SE of Belo Horizonte, ca. 1,500 m elev., 8 Feb. 1968, H. S. Irwin, H. Maxwell & D. C. Wasshausen 19570. (holotype, UB; isotypes, NY, US). Figure 10. Planta perennis, caespitosa, basibus leviter bulbosis, fuscis. Caules breves, per bases persistentes veternas fo- liorum obtecti. Radices graciles. Folia rigida, erecta, 0.8- 1.5 dm longa, solum basalia, vaginis scaporum longiora; laminae vaginis 4—5-plo longiora, tortae, 1-1.3 mm latae, compressae, flavovirentes vel ferrugineofuscae, transverse , leviter T n costales, pallide gals d aM integrae vel leviter laceratae, dorsis papillosis; bracteae steriles 4, subdecussatae, pari infimo oblongo, ca. 3 mm longo, pari intimo obovato, ca. 5 mm longo; bracteae fertiles obovatae, rotundatae, pli- catae, usque ad 7 mm longae, tenues, area dorsali indis- tincta. Sepala lateralia libera, “wag. ui ues lineari- oblonga, ca mm longa, sa; ala carinali apicem versus irregulariter ciliata vel fimbriolata, lateribus villo- sulis. Laminae petalorum elliptic ae, 6-7 mm longae, acu- tae, integrae, flavae Staminodia bibrachiata, brachiis compressis, i dun ian Antherae lanceolatae, ca. 2 mm longae, = ae, filamentis longiores. Placenta bas- alis. Semina | non visa. Plants perennial, cespitose, with bases slightly bulbous, brown. Stems short, covered by bases of old foliage leaves. Roots slender. Leaves rigid, erect, 0.8-1.5 dm long, strictly basal, longer than the scape sheaths; blades 4—5 times longer than sheaths, twisted, 1- 1.3 mm wide, flattened, yellow-green to red- dish brown, transversely rugulose, distinctly longitudinally 3-4-nerved; apices narrowly acute, sometimes aristulate; edges thickened; sheaths slightly dilated toward base, ecari- nate, gradually narrowing above into blades, transversely rugulose, carinate toward apex, the sides brownish, the edges ciliate with long pale hairs. Scape sheaths brownish to reddish brown, transversely rugulose, multicostate, carinate toward apex, with short, incurved blades. Scapes terete or subterete, + spirally twisted, flexuous, 2-3 dm high, ca. 1 mm Kral & das Graças 371 Volume 75, Number 1 1988 Brazilian Xyris tam n mM KEES one + ATA * ^, ut £53 iu A HAC fe ; 5 ^ x`. 1 e E ER t£ ir i" ste (à Ys Y v © FIGURE 10. Xyris villosicarinata (Irwin et al. 19570).—a. Habit sketch. puc ondes x.—c. saw yd id mid blade.—d. Sector at leaf blade-sheath junction.—e. Leaf base. —f. Spike.—g f eid h. Two views of fertile bract.—i. Lateral sepal.—j. Petal and stamen, SML pore ets, Sole. iu de 2 apex. 372 Annals of the Missouri Botanical Garden thick, transversely rugulose, green, bicostate to ecostate. Spikes narrowly ovoid or ellipsoid, 0.7-1 cm long, few-flowered, the bracts con- vex, slightly carinate, with midrib, pale shin- ing reddish brown, entire or slightly lacerate, the backs papillose; sterile bracts 4, subde- cussate, the lowest pair oblong, ca. 3 mm long, the inner pair obovate, ca. 5 mm long; fertile bracts obovate, rounded, plicate, up to 7 mm long, thin, the dorsal area indistinct. Lateral sepals free, subequilateral, linear-ob- long, ca. 6 mm long, obtuse; keel irregularly ciliate or fimbriolate toward apex, the sides villosulous. Petal blades elliptic, 6-7 mm long, acute, entire, yellow. Staminodia bibrachiate, the branches flattened, long-penicillate. An- thers lanceolate, ca. 2 mm long, sagittate, longer than the filaments. Placenta basal. Seeds not seen. This bears several of the characters of Xyris tortula Martius, but the foliage is ru- gulose throughout (not smooth), the margins of the narrow, twisted leaf blades are not cartilaginous as in X. tortula, and the sides of the lateral sepal keel toward its apex are densely villosulous with pale hairs, rather than confined to the keel only as in X. tortula. NEW AND NOTEWORTHY TAXA FROM PANAMA! Gordon McPherson? ABSTRACT Three new species, Prunus fortunensis (Rosaceae), Symplocos panamensis de EE and Angostura T bad (Rutaceae), as well as four new generic records for Panama are described. Botanical exploration of Panama has been going on for decades and yet many species still remain undiscovered. Collecting trips made in 1986 to such western Panamanian sites as the Fortuna Dam area, the Bocas del Toro slope above Chiriqui Grande, and Cerro Colorado could be counted on to yield new species as well as new records, some of them generic, for the Panamanian flora. Moreover, other rich but long-accessible areas such as the El Llano to Carti strip, Santa Rita Ridge, and Cerro Jefe continue to yield novelties. For the most part these are not fully appre- ciated until they reach the hands of appro- priate specialists and then, typically, are pub- lished only after some delay. In some cases, however, it is possible to readily evaluate the collections of special interest, and the follow- ing three species are here described as new. Prunus fortunensis McPherson, sp. nov. (Rosaceae). TYPE: Panama. Bocas del Toro: Fortuna Dam region, 1,250-1,300 m, 11 Feb. 1986, McPherson 8404 (ho- lotype, PMA; isotypes, CAS, CR, DUKE, K, MEXU, MICH, MO, US, UTD). Fig- ure 1. Species combinatione foliorum parvorum integrorum glabrorum cum inflorescentia racemosa et floribus parvis circa 10-stamineis a ordin regionis diversa. Tree 14 m; branchlets glabrous, somewhat longitudinally striate in older portions, the slightly raised lenticels 0.5-1 mm long; nodes marked by raised leaf scars, these pubescent along their adaxial margin when newly ex- posed, glabrescent. Leaf blades elliptic to slightly ovate, 3.5-6 cm long, 1.4-3 cm wide; base obtuse; apex acuminate, ultimately nar- rowly obtuse; margin entire; both surfaces glabrous, the laminar glands usually 1 pair, intramarginal, basal; secondary veins 5-6(-7) on each side of the midrib, all of the venation slightly impressed on the upper surface, the midrib alone raised on the lower surface. Pet- oles 5-8 mm long, caniculate, glabrous. Sti- ned triangular, 1.5-2 mm long, 1 mm wide at base, caducous. Inflorescences in the axils of leaves of the current season, racemose, 3- 5.5 cm long, glabrous; peduncle 6-10 mm long, bearing 14-20 flowers; pedicels 2-4 mm long; bracts small, absent at anthesis. Hypanthium campanulate at anthesis, gla- brous, 1-1.5 mm long from base to rim; lobes ank R. Liesner, J. Zarucchi, and A. Gentry for help with the generic records, and J. Myers for the incide dee p was supported by grants from the National Science Foundation and the National eographic Soci ? Missouri Nel Garden, P.O. Box 299, St. Louis, Missouri 63166, U.S.A. ANN. MISSOURI Bor. Garp. 75: 373-378. 1988. 374 Annals of the Missouri Botanical Garden FIGURE 1. Prunus fortunensis (McPherson 8404). — A. Flowering branchlet. —B. Flower, lateral and top lews. ca. 0.6 mm long, | mm wide at the base, thers ca. 0.5 mm long. Ovary ca. 1 mm diam., obtuse, somewhat cucullate. Petals white, glabrous; style 1.5-2 mm long, glabrous; stig- roughly circular, 1-1.5 mm diam., entire. ma flattened, 0.5-0.8 mm wide. Fruit un- Stamens 9-11(-15), 1.8-2 mm long, the an- known. Volume 75, Number 1 1988 McPherson 375 New Panamanian Taxa This species is distinguished from its con- geners in the region (P. annularis Koehne, P. brachybotrya Zucc.) by its combination of small, entire, glabrous leaves, racemose inflorescence, and small flowers with ca. 10 stamens. Symplocos panamensis McPherson, sp. nov. (Symplocaceae). TYPE: Panama. Co- lón: Santa Rita Ridge, ca. 500 m, 16 Feb. 1986, McPherson 8447 (holotype, PMA; isotypes, BM, CAS, COL, CR, D , K, L, MEXU, MICH, MO, NYBG, US, UTD). Figure 2. Species pilis appressis, es chartaceis fere glabris den- s brevibus, sepalis partim pu- m fere glabris, floris 40— 45-stamineis, stylis ane pubescentibus omnino a congeneribus regionis diversa. Tree 10-12 m; twigs pale in color, some- what flattened, the youngest often channeled (at least on drying); young stems and buds white pubescent, the buds densely so, the hairs strongly appressed, those of the stem not dense, + persistent but becoming sparse on older stems. Leaf blades obovate-elliptic to elliptic, 7-21 cm long, 2.7-7.2 cm wide, thinly chartaceous; base usually somewhat at- tenuate and narrowly obtuse, sometimes acute; apex acuminate; margin entire proximally, denticulate to dentate distally, often pubes- cent; upper surface glabrous, the lower sur- face pubescent while immature but glabrous or nearly so at maturity, a few hairs some- times remaining near the base; midrib adax- ially sunken over most of its length, raised abaxially, pubescent with white, appressed hairs; secondaries 5—6(—8) on each side of the midrib, somewhat arcuate, puberulent. Petiole 2-5(-7) mm long, channeled and gla- brous adaxially, appressed-pubescent abaxi- ally. Inflorescences axillary, the subtending leaf often fallen; axis 1-5 mm long, pubes- cent, bearing (1—)3-7 flowers; bracts several, the peduncular bracts densely pubescent abaxially, the pedicellar bracts less so, all glabrous adaxially. Hypanthium 1 mm long, glabrous or sparsely pubescent. Calyx lobes white, 5, 1.5-2 mm long, 1.5-2 mm wide at the base, obtuse, pubescent abaxially along the midline with white, appressed, somewhat flexuous hairs; margins pubescent. Corolla white, of 4-6 partly fused petals 11-16 mm in length and connate near the base, adnate to androecium about halfway up, the spread- ing portion of the lobes typically ca. 4 mm long, 2-5 mm wide, glabrous or sparsely pu- berulent with appressed hairs, occasionally bearing a misshapen anther apically. Stamens —45, exserted, the filaments straplike, the free portions of the outer ones up to 5 mm long, 0.7 mm wide, the inner stamens shorter, narrower, and somewhat inflexed. Ovary 3-locular; ovules 4 per locule, the superior portion of the ovary ca. 1 mm long, pubes- cent; style 9-11 mm long, at least sparsely pubescent over its entire length; stigma cap- itate, ca. 1 mm diam. Fruit unknown. This species differs from other Symplocos species of the region (S. austin-smithii Stand- ley, Š. chiriquensis Pittier, S. serrulata Humb. & Bonpl., inter alia) in its appressed hairs, thin, nearly glabrous, distally dentate leaf blades with few secondaries, short peti- oles, only partially pubescent sepals, nearly or quite glabrous, rather long corolla, 40—45 stamens, and styles pubescent their entire length. Additional specimen examined. PANAMA. oe ma Cuasi-Cana trail on Cerro Campamineto, east o Bocas, headwaters of Rio Cuasi, 29 Apr. 1968, Kirkbride & Duke 1246 (MO). Angostura kunorum McPherson, sp. nov. (Rutaceae). TYPE: Panama. San Blas: along El Llano-Carti road, ca. 14.5 mi. from Interamerican Highway, ca. 350 m, 17 June 1986, McPherson 9525 (ho- lotype, PMA). Figure 3. Species 5-7- Meri foliolis grandibus latis, inflores- centia anguste paniculat androeciis omnino fertilibus a congeneribus diversa. valvatis, Slender tree 4 m tall; young, leaf-bearing stems ca. 1 cm diam., roughened by many small ridges (at least on drying), conspicuously lenticellate, brown puberulent with a mixture of erect, curved, whitish hairs and minute, 376 Annals of the Missouri Botanical Garden FIGURE 2. Symplocos panamensis (McPherson 8447).— A. Flowering branchlet. —B. Inflorescence.—C. Opened S flower. —D. Stamen.—E. Detail of leaf margin. brown, spherical, sessile granulations as well as with intermediates; leaf buds with a similar but denser indument; leaf scars ca. 1 cm long, ca. 1 cm wide. Leaves alternate; blades pal- mately compound, 5- or 7-foliolate, the leaf- lets diverging through 1-9 of a circle, ellip- tic, the central and largest leaflet ca. 37 cm long, 12-13 cm wide, the outermost and Volume 75, Number 1 McPherson 377 1988 New Panamanian Taxa 4 B fe ce FIGURE 3. Angostura kunorum (McPherson 9525).—A. Flowering branchlet. —B. Flowers. smallest 19-29 cm long, 7-11.5 cm wide, on each side of the largest leaflets and ca. 18 attenuate at base, shortly and sharply acu- on each side of the smallest leaflets; petiolules minate at apex, chartaceous, the margins en- canaliculate, puberulent like the stem but gla- tire, the surfaces densely punctate, glabrous — brescent, pulvinate basally, the central one at least at maturity (except for the midrib, ca. 3.5 cm long, the outermost 0.8-1.5 cm this puberulent like the stem and reddish brown long, all ca. 3 mm diam.; petiole ca. 28-31.5 when dry), the secondaries slightly impressed cm long, ca. 5 mm diam. near mid length, on the adaxial surface, ca. 27-30 in number broadly caniculate, conspicuously lenticellate, 378 Annals of the Missouri Botanical Garden pubescent like the stem but partially glabres- cent, pulvinate basally, the apex of the petiole flattened, ca. 1 cm diam. Inflorescence ap- pearing pseudoterminal but perhaps truly ter- minal, very narrowly paniculate (the lateral dichasia subsessile), the axis of the single known inflorescence 8 cm long (further col- lections may be significantly longer), ca. 4 mm diam. near mid length, puberulent like the stem; stalks of the lateral dichasia 1-2 mm long; bracts triangular, ca. 2 mm long, 2 mm wide at base, stout, acute, puberulent; pedicels 3-5 mm long, puberulent. Buds straight. Calyx cupular, white, 5—6-lobed, 2 mm long, 4-5 mm diam.; lobes shallow, acute, densely puberulent, densely beset with swollen surficial glands ca. 0.2 mm diam. Corolla of 4-6 petals, white, valvate (or very nd ie in bud, up to 2 cm long in bud, long when open (the recurved Sorten not included); petals linear, connate by marginal tomentum, separating and recurving at an- thesis and eventually % to completely unat- tached to one another, tomentose on both surfaces, the hairs longer and curlier than those of the calyx and unaccompanied by brown granulations, the swollen surficial glands obvious, clear to brownish (at least on drying). Stamens all fertile, 5 or 6; filaments adnate to petal bases, 8-9 mm long, ca. 1.5 mm wide near base, somewhat flattened, tomen- tose; anthers 11-13 mm long, ca. 0.8 mm wide, introrse, pubescent with long, straight, + erect hairs, abaxially densely beset with swollen surficial glands. Disk cupular, some- times bearing impressions of the petal bases and hence ridged, 1 mm high, 2-3 mm diam. Ovary 5-carpellate, 1 mm long, glabrous; style ca. 2 mm long, glabrous; stigma ca. 2 mm long, cylindric, slightly thicker than style, glandular. Fruit unknown. Angostura kunorum differs from other species of this varied genus in its 5-7 large, broad leaflets, narrowly paniculate inflores- cence, straight buds with essentially valvately arranged petals, and entirely fertile androe- cium. Among the genera recently added to the known flora of Panama are Sparattanthe- lium of the Hernandiaceae (S. amazonum Mart., McPherson 9643), Etaballia of the Leguminosae (E. cf. guianensis Benth., McPherson 9513), Metteniusa of the Ica- cinaceae (M. tessmanniana (Sleumer) Sleu- mer, Gentry 138294, McPherson 7364), and Plinia of the Myrtaceae (P. spp. McPherson 7341, 9089). These four genera are primarily South American in their distri- butions. NOTES TWO NEW RUSHES (JUNCUS, JUNCACEAE) FROM CHIAPAS, MEXICO This note describes a new species and a new variety of Juncus from the state of Chia- pas in Mexico. Juncus chiapasensis Balslev, sp. nov. TYPE: Mexico. Chiapas: Lagunas de Montebello near Guatemala border, 30 km (air) E of La Trinitaria, 1,200 m, Roe, Roe & Mori 966 (holotype, DS-584358; iso- types, F, WIS). Figure 1A, B. Planta perennis, dna folia linearia, teretia, sep- tata; inflorescentia anthela, 10-20 x 5-12 cm, 8-17- capitulata; capitula 10—35-flora; tepala 2.5-3 mm longa, castanea; stamina 3, 1.8-2.5 mm longa; capsula 4-4.5 x 2 mm, obclavata, castanea, tepalis longior. Perennial, caespitose herbs 60-90 cm tall. Rhizome 3 mm diam. Culms erect, 2-3 mm diam., terete, smooth or finely striate. Basal sheathing bladeless leaves absent or one to each culm, up to 7 cm long. Foliar leaves 1-2 basal and 1-2 cauline to each culm, 10- 45 cm long; sheaths 4-8 cm long with mem- branaceous margins, terminally biauriculate, the auricles 2-3 mm long and rounded; blades 1.5-3 mm diam., round in cross section, con- spicuously cross septate. Inflorescence a de- compound anthela, 10-20 x 5-12 cm, of 8-17 flower heads, these globose, 10-13 mm diam., 10-35 flowered, castaneous, the ul- timate head-bearing branches ca. 0.5 mm thick. Lower inflorescence bract 3-7 cm long, much shorter than the inflorescence, resem- bling basal and cauline leaves; upper bracts progressively shorter, the floral bracts acu- minate, ca. 0.2 mm long, membranaceous. Tepals subequal, 2.5-3 mm long, lanceolate, acute, the outer ones V-shaped in cross sec- tion, the inner ones flat. Stamens three, 1.8- 2.5 mm long; filaments linear; anthers oblong, ANN. Missouni BOT. GARD. 0.6-0.8 mm long, about half as long as the filaments. Capsule 4-4.5 x 1-1.2 mm, con- spicuously longer than the tepals, obclavate, trigonous, acute, gradually tapering towards the apex, castaneous, unilocular. Seeds 1 x 0.2 mm, ellipsoid, apiculate, reticulate, yel- low-brown, with a thick, hyaline outer seed coat. This new species is known only from a few localities in the state of Chiapas, Mexico, near the Guatemalan border. It has been collected at elevations of 1,200-2,200 meters in marshes along lake shores. It belongs to Jun- cus subg. Septati Buchenau (1875), as evi- denced by its terete, hollow, and cross-septate leaf blades. This subgenus, with some 80 species distributed worldwide, is the largest in the genus. In the neotropical region, some 15 species belong to subg. Septati. The in- florescence of this new species is also typical of that subgenus: the flowers are congested into heads arranged in an anthela of which the proximal branches overtop the distal branches. Juncus chiapasensis differs from the closely related J. guadeloupensis Bu- chenau by having globose flower heads and obclavate, castaneous capsules much longer than the tepals. Juncus guadeloupensis has golden brown capsules and is endemic to the Caribbean island of Guadeloupe. Juncus de- bilis A. Gray likewise occurs in Chiapas and has longer capsules than tepals, but it is much smaller than J. chiapasensis in all dimen- sions. Additional specimens examined. MEXICO. CHIAPAS: La Trinitaria, Lagunas de Montebello, 1,300 m, Breed- love & Thorne 21247 (DS); San Cristóbal valley, 2,200 m, Breedlove & Thorne 21285 (DS, mixed with J. ebrac- teatus E. Meyer; MICH). 79: 379-382. 1988. 380 Annals of the Missouri Botanical Garden 3mm NS ° Y di Nt ÙÙ | * AV Ñ WS LAN | 3 A * ok Tind 10cm C "d URE l. A, B. Juncus chiapasensis (Roe, Roe & Mori 966). — A. Habit. —B. Flower with protruding capsule. , D. Juncus liebmannii var. polycephalus (Breedlove 15065). =, Habit. —D. Flower with capsule and floral TE Volume 75, Number 1 1988 Notes Juncus liebmannii Macbride, Field Mus. Nat. Hist., Bot. Ser. 11(1): 9. 1931 (as J. liebmannii). Replaced synonym: Jun- cus brevifolius Liebm. (1850), non Hoffsgg. & Link in Rostkov (1801). This species has one variety in the Andes of Ecuador and Colombia (var. quitensis (Buchenau) Balslev, 1979), and two varieties in Mexico and Central America, one of which is described here as new. Juncus liebmannii was first collected in Puebla, Mexico by Lieb- mann in 1841 and described as J. brevifolius (Liebmann, 1850). Liebmann was aware that Hoffmannsegg & Link in Rostkov (1801) had used the name J. brevifolius before, but since the entity bearing the first application of the name was a synonym of a species of Luzula, he I free ü use the name again for a species of Jun his was accepted by Buchenau (1873, E 1890, 1906), who called the typical variety J. breufültus var. mexicanus. Macbride (1931) renamed the species ac- cording to current nomenclatural rules. Juncus liebmannii Macbride var. poly- cephalus Balslev, var. nov. TYPE: Mex- ico. Chiapas: southern city limits of Te- pisca, 1,800 m, Breedlove 15065 (holotype, DS-609015; isotypes, MICH, NY). Figure 1C, D. nta perennis; rhizoma repens, internodiis 1 cm lon- Perennial herbs 60-70 cm tall. Rhizome horizontally creeping, 2-3 mm diam. Culms erect, 1-2 mm diam., terete, finely striate. Cataphylls 1-2 to each culm, up to 12 cm long. Foliar leaves 2-3 to each culm, inserted along the whole culm, 10-25 cm long, the basal and upper foliar leaves with shorter blades than the middle ones; sheaths 5-15 cm long, with a distinct membranaceous mar- gin terminating in two involute auricles; blades 1-1.5 mm diam., round or slightly flattened in cross section, hollow and cross septate. —— ——. 18 10: 145-1 ^w Inflorescence a decompound anthela, 3-8 x 2-4 cm, of 10-30 flower heads, these semi- globose or conical, 4-6 mm diam., 10-15 flowered, castaneous, the ultimate head-bear- ing branches 0.2-0.3 mm thick. Lower in- florescence bract up to 8 cm long, with a foliar blade, this round in cross section and with transverse septa, or only 3-4 cm long and with the blade reduced to a small acicular appendage; upper bracts much smaller; floral bracts 1-2 mm long, membranaceous. Tepals subequal, 2-2.5 mm long, lanceolate, acute, the outer ones V-shaped in cross section, in- ner ones flat. Stamens six, 1-1.5 mm long; filaments linear; anthers oblong, ca. 0.3 mm long, about !4 as long as the filaments. Cap- sules 2-2.5 x 1-1.5 mm, obovoid, acumi- nate to short-beaked, about as long as the tepals, keeled along the sutures, castaneous, unilocular. Seeds 0.4 x 0.2 mm, ellipsoid, apiculate, reticulate, yellow-brown. This new variety agrees with J. liebmannii var. liebmannii in all characters except for the inflorescence, which is divided into 10- 30 distinct flower heads. In the typical variety the inflorescence is divided in two subequal parts with the flowers arranged diffusely in 2-3-flowered glomerules but not in distinct heads I thank Kirsten Tind for the beautiful and illustrative plate and Rupert Barneby for lin- guistic assistance. LITERATURE CITED BarstEv, H. 1979. Juncaceae. In: C. Harling & B. Sparre e Flora of Ecuador 11: 1-45. BUCHENAU, 1873. Über einige von Liebm exico DNA Pflanzen. Abh. Nine. "Ve. reine justa en 3: 339-350. 875. Monographie der Juncaceen vom Cap. Abh. Naturais Ve Bremen 4: -5 86 ereine ie Juncaceen aus Mittelamerika. Flora . 1890. Monographia Juncacearum. Bot. Jahrb. Syst. 12: 1-495. 6. Juncaceae. In: A. Engler, Das Pflan- zen. n IV. 36: 1- rH id 25). LIEBMANN, F. 1850. s Juncaceer. Vidensk. Meddel. Dansk alii Foren. Kaaba vn 2: 36- 8. 382 Annals of the Missouri Botanical Garden MACBRIDE, F. 1931. Spermatophytes, mostly Peruvian ; va dp ag o . 11(1): 1-36. RosTKOv, G. T. Dissertatio Botanica Inaguralis de lunco, etc. i A. Grunnerti, Hala Sa — —Henrik Balslev, Botanical Institute, Uni- versity of Aarhus, Nordlandsvej 68, DK- 6240 Risskov, Denmark. A NEW SPECIES OF STRYCHNOS (LOGANIACEAE) FROM NICARAGUA While preparing treatments of Logani- aceae for the forthcoming Flora de Nicara- gua and Flora Mesoamericana, | encoun- tered a new species of Strychnos from the Caribbean lowlands of Nicaragua. Strychnos is a pantropical genus of some 200 species. f the approximately 90 species of the Amer- ican tropics, only nine, including the one de- scribed here, occur in Mexico or Central America. [n the most recent treatments of the American species of Strychnos (Krukoff & Barneby, 1969; Krukoff, 1972), S. ni- caraguensis would be assigned to sect. Brev- iflorae Progel subsect. Eriospermae Krukoff & Barneby on the basis of its recurved spines, terminal inflorescences, short styles, and seeds with testa composed of soft fibers. Unfortu- nately, the corolla of S. nicaraguensis re- mains unknown, so that the length of the corolla tube in relation to the calyx, one of the important sectional characters in Strych- sections), cannot be positively ascertained. Nevertheless, other characters, principally the recurved spines and fibrous testa, assure its placement in subsect. Eriospermae of sect. Breviflorae. Strychnos nicaraguensis Huft, sp. nov. TYPE: Nicaragua. Zelaya: N of Talolinga in gallery forest, 19 Aug. 1983, J. Sandino 4509 (holotype, F-1988116, F neg. no. 62188; isotype, MO, not seen). Frutex volubilis; caules non conspicue lenticellis prae- diti, spinis recurvatis. Folia glabra rhombeo-ovata, cdi lo peo. basi acuta vel acuminat su non impressa, nervis late ralibus obscuris. Flores 4-7, cy mis crainalibus om ceolata ciliata al pdas, corolla staminaque non visa. Fructus laevis, ca. 1.5 mm diametro, luteus, caos tenui; semina 1 vel 2, depresso-globosa, testa fibrosa. Liana; stems not conspicuously lenticel- ANN. late, with recurved spines to 2 cm long; branchlets wandlike, flexuous, densely pu- berulous, soon glabrate. Leaves with petioles 1-2 mm long, puberulous; blades rhombic- ovate, 3-5(-5.5) cm long, 1.4-2.2 cm wide, 1.7-)2-2.9(-3.2) times as long as wide, gla- brous, dull green above, pale below, tripli- nerved, the midvein scarcely or not at all impressed above, the secondary veins ob- Scure; apex long-acuminate; base acute to acuminate; margin entire. Flowers 4-7 in compact, terminal, slender-pedunculate cymes 5-7 cm diam., the peduncles 1-1.5 cm long; bracts lanceolate, 1.5-3 mm long, ciliate; pedicels 1-2 mm long; calyx segments 5, free, slightly unequal, deltate-lanceolate, na- vicular, long-acuminate, ciliate, otherwise gla- brous, 1.2-1.8(-2) mm long; corolla not seen; stamens not seen; style ca. O. ong. Fruits globose, smooth, ca. 1.5 cm diam., the shell ca. 0.5 mm thick, yellow; seeds 1 or 2, depressed globose, ca. 5 mm long, ca. 7 mm diam., the testa fibrous. — Subsection Ériospermae is distinguished from subsect. Breviflorae Krukoff & Bar- neby, which comprises the rest of sect. Brev- iflorae, solely by the characteristic soft fi- brous testa of the seeds, which breaks away from the dry, shrunken endosperm of old seeds and encloses the endosperm like a sac. Strychnos nicaraguensis belongs to a group of species of subsect. Eriospermae that are characterized by conspicuously lenticellate branchlets, recurved spines, midveins pressed above, faint or obscure foliar retic- ulations, distinctly pedicellate flowers, and sparingly ciliate calyx lobes. It is distinguished from all species in this group by the absence of conspicuous lenticels and by its distinctive small rhombic-ovate leaves without an im- pressed midvein. The only other uus of this group in Central America, S. bra 75: 383-384. 1988. im- Missouni Bor. GARD. 384 Annals of the Missouri Botanical Garden stantha Standley, further differs by its much larger fruits (to 9 cm in diameter) with thick shells, as does S. nigricans Progel of south- eastern Brazil. The species most closely re- lated to 5. nicaraguensis appears to be S. mattogrossensis S. Moore, a species widely distributed in the Amazon basin. Both species have small fruits with thin shells ca. 0.5 mm thick, but S. mattogrossensis differs in having ovate, obovate, or elliptic leaves that are not at all rhombic-ovate and larger, more highly branched cymes 7-25 mm in diameter. Other species in this group include S. cerradoensis Krukoff & Barneby, native to the state of Minas Gerais, Brazil, which differs from S. nicaraguensis by its thick-shelled fruit, and S. alvimiana Krukoff & Barneby (Phytologia 27: 105. 1973), of the state of Bahia, Brazil, which has larger leaves that are dull on both surfaces and rounded or obtuse at the base. Additional specimens examined. NICARAGUA. 1983, Ortiz 586 (F); Municipio de Siuna, * ‘Calera, " 13°46'N, 84°46'W, 300- 45 m, 12 Mar. 1984, Ortiz 1790 (F). LITERATURE CITED KRukorr, B. A. 1972. een species of Strychnos. Lloydia 35: 193-2 — & R. dead "usn Supplementary notes on the American species of Strychnos. VIII. Mem New York Bot. Bard. 20(1): 1-93. —Michael J. Huft, Missouri Botanical Gar- den. Mailing address: Department of Bot- any, Field Museum of Natural History, Chi- cago, Illinois 60605, U.S.A. TWO NEW SPECIES OF /NGA (LEGUMINOSAE) FROM PANAMA This paper describes two species of /nga from Panama, bringing the total number of Inga species known from that country to 29 (D'Arcy, 1987). The two species are unre- lated within the genus, apparently belonging to different sections. Both are from areas where endemism is reported in other groups. Inga jefensis Liesner & D'Arcy, sp nov. TYPE: Panama. Panamá: 1 mi. upstream from Frizzel's Finca Indio, slopes of Cer- ro Jefe, flower & fruit, 9 Sep., Foster & Kennedy 1828 (holotype, PMA; isotype, MO). Figure 1 Arbor parva, differt Ingis aliis panamensibus foliolis glabris concoloribus, rhachibus non alatis, inflorescentia umbellata, pedicellis gracilibus longioribus, fructibus lon- gioribus. Tree 15 m tall, the branchlets copiously lenticellate, nearly glabrous but with occa- sional minute trichomes. Leaves with petiole 5-10 cm long, terete, slender, unwinged, drying olive green; rachis ca. 15 cm long, resembling the petiole, the glands ca. 2 mm across; petiolules (Gentry 8854) ca. 6 mm long, thick, drying olive green; leaflets 4-5 pairs, elliptic to obovate, acuminate, basally obtuse, 6.3-19 cm long, drying concolorous, olive green (emerging growth drying reddish brown), glabrous, the lateral veins 9-10 on each side; stipules oblong or obovate, ca. 15 mm long, 6 mm wide, drying enervate, mem- branaceous. Inflorescence umbellate; pedun- cle ca. 7 cm long, slender; pedicels slender, 7-12 mm long. Flowers ca. 23; calyx tubular, ca. 2 mm long, glabrate, the lobes short, rounded; corolla tubular, 13-14 mm long, lobed 7-4 its length, the lobes ca. 3 mm long, ca. 3 mm wide at the base; stamens ca. 3 cm long, exserted ca. 1.5 cm, the exserted por- tion longer than the corolla. Fruit linear, com- pressed around the seeds, the margins slightly sinuate between the seeds, ca. 30 cm long, 18-25 mm wide, 8 mm thick (dried). The fruit of this species is similar to but longer than that of /nga portobellensis Beur- ling (sect. Leptinga fide León, 1966) from the Caribbean coast north of the localities for the new species. Inga jefensis differs in its smaller calyx (2 mm vs. 20-25 mm long) and notably in having an unwinged leaf rachis. This species is also somewhat similar to Inga paterno Harms (also in sect. Leptinga), which ranges from Costa Rica to Mexico, but /. paterno has sessile florets and a short, mas- sive fruit 4-7 mm wide and 2-3 cm thick. Inga jefensis is known only from Central Panama. The two locations, Cerro Jefe, a mountain ridge behind Panama City with el- evations up to 900 m, and Santa Rita Ridge, another mountain ridge paralleling the Carib- bean coast, are both areas known to have many endemic species. The collection from Cerro Jefe was taken in flower and fruit in September, and the collection from Santa Rita Ridge was taken in December in flower. Paratypes. | PANAMA. COLÓN: Santa Rita Ridge road 4 mi. from Transisthmian Highway to Agua Clara weather station, 500 m, flower, 11 Dec., Gentry et al. 8854 MO). Inga spiralis Liesner & D'Arcy. TYPE: Pan- ama. Panamá: 23.4 km from Panamer- ican Highway, fruiting, 13 Apr., Mori & Kallunki 5577 (holotype, PMA; iso- type, MO). Figure 2. Arbor parva, differt Ingis aliis panamensibus foliis gran- dibus, foliolis grandibus discoloribus rigidis, rhachibus non alatis, calyce lato, fructu solido spiraliter 1Y4-plo super- posito puberulenti. Tree 15 m tall (Dressler 4325), the branchlets drying reddish with scattered mi- nute simple hairs, prominently lenticellulate when young, glabrescent, becoming grayish. ANN. MISSOURI Bor. GARD. 75: 385-388. 1988. 386 Annals of the Missouri Botanical Garden RE l. Eo (x 1).—C. Fruit (X0.5). Leaves with the petiole 10-21 cm long, sub- terete, slightly angled basally, unwinged, drying reddish, with sparse minute simple hairs; rachis 15-30 cm long, nonwinged, re- sembling the petiole, the glands subcupulate, Inga jefensis (Foster & Kennedy 1828 (MO)).—A. Leaf and twig with stipule (x0.5).— ca. 3 mm across; petiolules ca. 4 mm long, stout; leaflets elliptic to obovate, apically rounded and short apiculate, basally obtuse, 20-30 cm long, 16 cm broad, drying dis- colorous, slate-colored above, reddish be- Volume 75, Number 1 1988 Notes 387 Vm FIGURE 2. Inga spiralis. — A. Flower (x 44).—B. Fruit and pedicel borne on branch (x 0.5). —C. Leaf (x 4).— D. Stipule and old inflorescence (x 0.5). (A, C, from Dressler 4325 (MO); B, from Mori & Kallunki 5577 (MO); D, from Liesner 1314 (MO).) neath, glabrous above, softly puberulent be- neath with short hairs; lateral veins 16-18 on each side; stipules conspicuous, subfolia- ceous, subcircular, ca. 5 cm long, drying prominently nerved, persistent. Inflorescence (Dressler 4325) umbellate, the peduncle 7 cm long, straight, ca. 4 mm thick, borne ‘‘on branches ca. 3 cm diam.," pedicels 3-4 mm long, broadening upwards. Flowers 40-50; calyx 1 cm long, lobed YY, way down, the 388 Annals of the Missouri Botanical Garden lobes oblong, 5-6 mm long, 4-5 mm wide at the base; corolla 2 cm long, deeply lobed, the lobes ca. 10 mm long, 4 mm wide, apically acute; filaments 3-4 cm long, exserted 2-3 cm. Fruit flattened, circinnate, coiled 17, times, 3-4 cm across, conspicuously rugose nervate, forming a massive spiral 7-9 cm diam., softly puberulent with dense, short, whitish hairs. This species is distinct in its massive, tightly coiled fruit. It is known only from the El Llano-Carti road in Panama Province, not far south of Nusigandi. The fruit somewhat resembles that of /nga davidsoniae Standl. (sect. Inga ser. Densiflorae according to León, 1966) of Chiriqui Province, which is also short and thick, but that fruit is not coiled, and the leaflets of Inga davidsoniae are small in con- trast to the unusually large leaves of the new species. Specimens in flower were collected in March and April and fruiting specimens in April. Paratypes. PANAMA. PANAMA: El Llano-Carti high- as ca. 20 km N of El Llano, 300-350 m, flowering, 6 pr. Dressler 4325 (MO). SAN BLAS: El Llano-Carti road, continent divide to 1 d N of divide, flowering, 30 , Liesner 1314 (M This research was —m by National Science Foundation Grant BSR-8305425, W. G. D'Arcy, principal investigator. LITERAUTRE CITED D'arcy, W. G. 1987. Flora of Panama: checklist & index. Monogr. Syst. Bot. Missouri Bot. Gard. 17, 1 LEÓN, J. 1966. Central American and West Indian species of Inga poe Ann. Missouri Bot. 359. Gard. 53: 265- — Ronald L. Liesner and William G. D'Arcy, Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166, U.S.A. NOTES ON THE FIJIAN ENDEMIC MER YTA TENUIFOLIA (ARALIACEAE) Meryta J. R. & G. Forst. comprises about 30 species of dioecious, simple-leaved arali- aceous trees and shrubs, nearly all of which are endemic to one or a few Pacific islands. The genus is centered in New Caledonia, where 11 species occur (Lowry, unpubl. data), and four species are recognized in Samoa and Tonga (Cox, 1985), one of which also appears to have been collected recently on Alofi (Mo- rat & Veillon, 1985). Meryta reaches its western limit in Micronesia on Yap and has one species each in Vanuatu (formerly New Hebrides), Norfolk Island, New Zealand, and Rarotonga (Cook Islands). In Polynesia, per- haps three to five species occur on Tahiti, where they form a polymorphic and taxo- nomically difficult complex; two species have been described from Raiatea, and one species, M. brachypoda Harms, occurs on Raiavavae and on Tubuai. One endemic species of Mer- yta is found on Rapa to the south, and an undescribed species has been collected in the Marquesas. Until recently, Meryta was thought to be lacking from Fiji (Smith & Stone, 1968). In December 1968, however, a single pistillate collection with mature fruit was made in the mountainous interior of Viti Levu and was described as M. tenuifolia by Smith (1971). This material is somewhat fragmentary and the descriptive notes that accompany it are sketchy. Nevertheless, it is clear that M. ten- uifolia is very distinctive within the genus and is remarkable in being a large, highly branched canopy tree to nearly 25 m tall. During a visit to Fiji in November 1985, I was able to re-collect Meryta tenuifolia near the village of Vanualevu, at the edge of the Rairaimatuku Plateau. A small population comprising several individuals of this tree, known locally as *'lutulutu," was found on a gentle slope in dense, undisturbed forest at 980 m elevation, less than one km to the east of Vanualevu, perhaps two km to the north- northeast of the type locality. We located staminate plants in bud and flower as well as pistillate individuals with flowers and nearly mature fruits. Pressed specimens supple- mented with FAA-preserved material and col- or photographs permit the following amended description. Meryta tenuifolia A. C. Smith, Pacific Sci. 25: 499. 1971, emend. Lowry. TYPE: Fiji. Viti Levu: Nandronga & Navosa (now Navosa) Prov., rocky bank of Nggalivava Creek, a northward flowing stream joining Lumunda Creek (Singa- toka River tributary), ca. 1.5 km S of Vanualevu, 750-800 m, 4 Dec. 1968 (fr), M. J. Berry (coll. E. Damanu) 97 (holotype, BISH; isotype, K). Figure 1. Dioecious, branched, glabrous trees to ca. 25 m tall; trunk to ca. 70 cm dbh, fluted to ca. 1 m; stems robust, covered with numerous leaf scars. Leaves simple, alternate, clustered at the ends of branches; blades medium green above, lighter below, chartaceous, broadly el- liptic, (9-)11-22 x (5-)7-12(-13) cm, the venation light yellow green, the midvein raised but without prominent bulges below, the sec- ondary veins 8-11, diverging from the mid- vein at 50-60? angles, arcuate at the ends, the higher-order veins evidently raised above and below, forming a dense network, the apex rounded to obtuse or broadly acute, the mar- gin entire, minutely thickened and revolute, the base obtuse to rounded and often shortly decurrent; petiole slender, 1.5-2 mm diam., with an expanded, clasping, brownish base with scarious margins. Inflorescence a panicle of racemules (or sometimes umbellules), ter- minal, erect, light green, occasionally tinged ANN. Missouni Bor. Garp. 75: 389-391. 1988. 390 Annals of the Missouri Botanical Garden š cC te” airs x " tap x \ A 8 t E LR veo A E € CU DN We EN a po We > #5 m al P di Ficu Meryta tenuifolia.— A. Branch of pistillate plant with infructescence.—B. Nearly mature frui C asa of staminate oo with inflorescence.—D. Tertiary axis with staminate flowers.—E. Staminate pm at anthesis (top vie Volume 75, Number 1 1988 Notes orangish; primary axis slender, in staminate plants to 20 cm long, in pistillate plants to 10 cm long; the secondary axes 6-8, scat- tered, ascending, on staminate plants 6-9 cm long, on pistillate plants 3-4 cm long, each subtended by an early-caducous, ovate, cu- pulate, scarious, strongly adaxially concave cataphyll 12-15 mm long; tertiary axes sub- tended by small, scarious, caducous bractlets, the axes on staminate plants 6-12, ascending, 12-25 mm long, each with 3-6 racemules (sometimes reduced to umbellules) of (3-) 4—6 sessile flowers, the axes on pistillate plants 4—6, ascending to spreading, 4-12 mm long, each with 4-6(-7) racemosely arranged ses- sile flowers. Staminate flowers ovate in bud; sepals wanting; petals 4, ovate, spreading at anthesis, ca. 1.5 mm long; stamens 4, inflexed in bud, ascending at anthesis, cream white, the filaments slender, 1.5-2 mm long, the anthers with 4 thecae, dorsifixed; the rounded nectar disk yellowish, weakly 4-sided. Pistil- late flowers ovate-pyriform in bud; sepals wanting; petals 4 or 5, narrowly deltoid, sub- acute, recurved after anthesis, ca. 0.5 mm long, expanding to 0.6-0.9 x 0.9-1.3 mm in fruit; stamens vestigial, 0.5-0.8 mm long, caducous, the anthers sterile; ovary inferior, (5-)6-10-carpellate, ca. 1.5-2 mm high at anthesis, the dns nectar disk 1-1.5 mm diam., expanding to ca. 2 mm in fruit; styles (5-)6-10, nud ball the free arms erect to ascending at anthesis, 0.3-0.6 mm long, in fruit expanding to 0.5-1 mm long, becom- ing divergent. Fruit a drupe, olive green when nearly mature, subglobose-oblate, 2- 3.5(-4.5) x (3.5-)4-5.5 mm, smooth and plump when fresh, turning strongly and acutely (5-)6-10-costate when dry. Additional specimens examined. FIJI. VITI LEVU: Navosa Prov., edge of Rairaimatuku Plateau, E and above ) P. P. Lowry 3838 3 sheets), P, US); (pistillate fl, fr), 3839 (BISH, x (3 sheets), P, US). As Smith (1985) indicated, Meryta tenui- folia does not appear to be closely related to the very distinctive species occurring in Va- nuatu and Samoa. Furthermore, it does not show strong affinities with species of Meryta on Rarotonga, Norfolk Island, and New Zea- land. However, until my upcoming detailed revision of the genus has been completed, it will not be possible to determine whether M. tenuifolia is closest to Polynesian species such as M. choristantha Harms of Rapa, as sug- gested by Smith (1985), or to New Caledonian species such as M. balansae Baill. I am grateful to A. C. Smith for providing locality information and a map of the original collection, and for valuable suggestions on the manuscript. Thanks are also due to J. M Miller and S. Vodonaivalu for assistance in the field, to C. A. Todzia and B. C. Stone for additional comments, and to J. K. Myers for preparing the illustration. This work was sup- ported in part by NSF Doctoral Dissertation Improvement Grant BSR83-14691, the Mis- souri Botanical Garden, the W. Alton Jones Foundation, and the Division of Biology and Biomedical Sciences of Washington Univer- sity, St. Louis, Missouri. LITERATURE CITED P. A. The genus Ere An in 3- Cox, 1985. moa. J. Arnold Arbor. 66 . & J.-M. L jon! — , Paris, Sér. 59-329. s of Pacific Island plants, I. New lowering cies from Fiji. Pacific Sci. 25: 491- i S lora Vitiensis Nova, bie 3. Pacific Tropical Botanical Garden, Lawai, Kauai, Hawaii. & . STONE. 1968. Studies of Pacific Island plants, XIX. The Araliaceae of the New Hebrides, Fiji, Samoa, and Tonga. J. Arnold Arbor. 49: 431- 493. —Porter P. Lowry II, Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166-0299, U.S.A.; and Laboratoire de Phanérogamie, Muséum National d'His- toire Naturelle, 16, rue Buffon, 75005 Paris, France. PASSIFLORA PUSILLA (PASSIFLORACEAE), A NEW SPECIES FROM CENTRAL AMERICA During the course of an ongoing study of Central American Passifloraceae, the follow- ing distinctive new species has been found in collections generated by the Flora de Nica- ragua project and from older Costa Rican collections. Passiflora pusilla MacDougal, sp. nov. TYPE: Nicaragua. Chontales: Hacienda Corpus, W of Juigalpa, ca. 100 m, 12°07'N, 85?28'W, 14 June 1984, Ste- vens 22968 (holotype, MO; isotype, HMNH). Figure 1. Passiflora hirsuta nana, scandens vel decumbens; cau- lis 12-55 cm longus, triqueter vel subtriqueter; petioli eglandulosi; folia trilobata, basi cordati, lobis obtusis vel truncatis, marginibus eg pedunculi ber uni- flori; bracteae nulli; biseriata, filamentis interioribus submicroscopicis, 0.2-1.5 mm longis; oper- culum plicatum; ovarium dense puberulum; fructus 3.5- 4 cm longus, 0.5-0.7 cm latus, anguste fusiformis, sex- ii uqa, semina Pas seriebus 2 longitudinalis dis- positu Diminutive, weakly climbing or often de- cumbent herb, 12-55 cm long, often fertile within 10-15 cm of its base. Roots perennial, the primary root 2.5-4 mm diam.; 1-5 an- nual stems arising from axillary buds at base of stem. Plant hirsutellous throughout with cylindrical, straight or slightly antrorsely bent, unicellular but often many-septate pelucid trichomes (0.1-)0.4-1.4(-1.8) mm long, these generally intermixed with microscopic, appressed trichomes 0.06-0.08(-0.1) mm long. Stem triangular or subtriangular (drying sulcate), 3-carinate, nearly glabrous below, becoming puberulent above with microscopic, appressed trichomes, and conspicuously hir- sute on the carinae with trichomes (0.1—)0.4— 0.6(-0.8) mm long. Stipules (2.5-)3-4(-5) x 0.4-0.5 mm, narrowly lanceolate to linear- triangular, subfalcate, hirsute, greenish, the apices not necrescent. Petioles 1-2.5(-3.6) cm, eglandular, canaliculate, sometimes tinged purplish. Laminae 1-2.4 x 2-4.5 cm at fer- tile nodes, depressed obovate in general out- line, cordate at base, very shallowly 3-lobed (and in one collection, very rarely the lateral lobes shallowly and obscurely lobed at base), the lateral lobes obtuse, + rounded, the cen- tral lobe broadly obtuse or nearly truncate, the angle between lateral lobes (80-)85-100 (—108)°, ratio of lateral to central lobe lengths 1.1-1.4, ratio of laminar width to length 1.6- 2.2, the margins entire and minutely setose to strigose; laminar nectaries absent. Tendrils absent, or present at distal nodes and then capillary and nonlignified; posture of devel. oping tendrils and shoot apex unknown. Pro- phyll of vegetative ramifying bud 1, narrowly ovate, caudate. Peduncles solitary at nodes, 4.5-15(-21) mm, uniflorous, ebracteate. Flower ca. 1 cm diam., with 1-1.7 mm stipe (elongating to 1.7-3 mm in fruit), the hy- panthium 3.5-4 mm diam., hirsute, the lon- ger trichomes often borne 0.1-0.2 mm above the surface of the epidermis on cylindrical, slightly raised bases; sepals 7 x 1.8-2.2 mm, lanceolate, pale yellowish green; petals 2.7— 3.5 X 1-1.3 mm, narrowly ovate or oblong, pale yellowish green; filamentous corona in 2 series, the outer 5-6 mm long, yellow toward apex, yellowish green to greenish below and sometimes with 1-3 narrow purplish bands at base or in lower half; inner series rudi- mentary, the members few, borne at base of operculum, 0.2-1.5 mm, capillary, slightly clavate; operculum 0.7-1 mm, membranous, the margin erose, plicate; nectary with an- nulus or nectar ring adjacent to limen; limen (disk) cupuliform, closely surrounding base of androgynophore, its edge + erect and 0.7- 0.8 mm high; staminal filaments connate ANN. Missouni Bor. GARD. 75: 392-395. 1988. Volume 75, Number 1 1988 Notes 393 4.2-6 mm along androgynophore, the free portions ca. 2 mm long; anthers 1.8-2.1 mm long, ovary 1.4-1.8 x 0.7-1 mm, narrowly obovoid or ellipsoid, 6-ridged or hexagonal in cross section, densely puberulent with ap- pressed trichomes 0.05-0.1 mm long; styles ca. 3 mm?; stigmas capitellate. Fruit 3.5- 4 x 0.5-0.7 cm, narrowly fusiform or fusi- form, the stipe often indistinct, distally cau- date, hexagonal and 6-carinate, sparsely and minutely puberulent, dehiscent; arils whitish, scanty, shorter than seed; seeds 3.1-3.4 x 1.7-1.8 mm, obovate, obliquely beaked at chalazal end, the short beak sharply angled toward the raphe, with 2 longitudinal rows o teeth (or 5-6 transversely sulcate, the ridges traversed by a broad longitudinal furrow). Phenology. This is a species of strongly seasonally dry habitats, and it apparently dies back to the ground each dry season. The small size of the plant suggests that it may flower and set seed within only a few months of germination; nevertheless, nearly all of the specimens studied were collected with roots intact and show remnants of dead shoots from previous seasons. The short herbaceous shoots may be expected to emerge after the spring rains in April. Flowering from May through the summer and into the end of the rainy season, Passiflora pusilla has been found in fruit from late June to November. Habitat and distribution. In the low tropical dry and gallery forests in the general vicinity of Lake Nicaragua, this passionflower is associated with the distinctive soil type called "sonsocuite" in Nahuatl. This sticky black soil is alkaline, poorly drained, and is season- ally inundated. It supports a forest of low stature, with Crescentia, Cordia, and mi- mosoids like Acacia, but has been mostly converted to pasture or intensive cultivation of cotton, rice, or sesame. This small pas- sionflower has been found mainly below 300 m elevation in partial shade at the edges of the *sonsocuite," on banks at the edges of roads and pastures, or in the shade of asso- ciated gallery forest. There are two records from a disturbed area at 800 m on the Meseta Central of Costa Rica. The specific epithet refers to the very small size of this plant. Paratypes. Costa RICA. GUANACASTE: Santa Rosa National Park, 30 km W of Liberia, 0-320 m, 10%50'N, 85°35'W, 18 Aug. 1984 (fr), Janzen 12412 (MO); 5 9 E j © ° Ww Q z [oe] bo en E — S = É 9 8 Lond — Ne) ~J] -J pe" 3 o a “a 3 c = Re Rockwood 2516 (MO); 23 km SW of Liberia, 10?24'N, 85?34'W, 1-200 m, 23 July 1964 (fl, 2 og 1424 (WIS). SAN JOSE: hem Ana [9°56'N, 84° uL 25 Nov. 1963 (fl, fr), Jiménez ug e Santa Ana [9°56'N, prod W], 8 odriguez C. 464 (UC). haan CHONTALES: Ha- cienda Corpus, W of Juigalpa, ca. 100 m, 12?07'N, 85?28'W, 20 May 1984 (fl), Stevens 22898 (MO, HMNH). ° ¿O B Passiflora pusilla is referred to subg. Plectostemma Masters sect. Xerogona (Ra- finesque) Killip on account of the small flower with a plicate operculum, subtriangular car- inate stem, absence of either floral bracts or extrafloral nectaries, dehiscent fusiform fruit, and testal architecture of a chalazal beak sharply angled towards the raphe with a fun- damentally transversely grooved sclerotesta. The dehiscent, hexagonal, fusiform fruit. of P. pusilla, characteristic of sect. Xerogona, is known in its fully mature state only from Jiménez M. 2136 Passiflora pusilla is perhaps most similar vegetatively to P. tenella Killip. Passiflora tenella is endemic to the Pacific tropical de- ciduous forests of Ecuador and Peru (Holm- Nielsen et al., in press). The two species share a very reduced size, similar eglandular leaves, and solitary peduncles. The laminae of P. tenella differ, however, in being less pubes- cent (especially abaxially), with the apices of the lateral lobes acute and the angle between them broader. The ovary of P. tenella is nearly glabrous. Its fruit, although also fusi- form, is shorter and relatively broader, lack- ing the caudate apex seen in P. pusilla; whether it dehisces is unknown. The seed of the South American species differs greatly by being transversely sulcate with five rugulose ridges. Killip (1938) placed P. tenella in sect. 394 Annals of the Missouri Botanical Garden FIGURE l. — Passiflora pusilla. — A. Habit. —B. Detail of leaf, node, stipules, and peduncle. —C. Nearly seed mature fruit.—D, E. Seed. Decaloba (DC.) Mast., rather than in sect. relationship of their respective habitats and Xerogona, based on the rugulose testal ridges vegetation types (Gentry, 1982), cannot be and the presence of floral bracts. A close confirmed without further study of the poorly kinship between P. pusilla and P. tenella, understood South American species. phytogeographically plausible due to the close Passiflora pusilla is also similar to both Volume 75, Number 1 1988 Notes 395 P. conzattiana Killip and P. goniosperma Killip of Mexico. Passiflora conzattiana is a small, low-growing, often prostrate vine of wet montane or Liquidambar cloud forests in eastern Mexico. It has similar leaves, small flowers, and very similar fruits, but is a larger plant having apically truncate to lunately bi- lobed laminae with acute to acuminate lateral lobes, generally less pubescence, and seeds with five or six smooth transverse ridges. Pas- siflora goniosperma is likewise very similar and is from a more similar seasonally dry habitat in southern Mexico. This poorly known species is also a larger plant, having more deeply bilobed leaves, larger fruits, and un- usually sculptured seeds that have a single toothed ridge down the length of the face of the testa. Both the more glabrous race of the Pacific tropical deciduous forests and the densely pubescent typical race from central Oaxaca share this form of seed (MacDougal, unpubl.). In contrast, the seed of P. pusilla has a furrow in the comparable position with two rows of teeth on either side. Both Mexican species possess only a single series of coronal filaments that are conspicuously and nearly uniformly reddish purple below their yellowish tips. I thank Warren Douglas Stevens for hab- itat information and for reading and criticizing the manuscript. The illustration was pre- pared by John Myers with support from the Missouri Botanical Garden. This research was supported in part by a fellowship from the Noyes Foundation. LITERATURE CITED Gentry, A. H. 1982. gig cis e patterns as evi- dence for a Chocó refuge. Pp. 6 inG. Prance (editor), Biological rut data in the Tropics. Co- lumbia Univ. Pre " New York. HorM-NIELSEN, L. B., P. JoRGENSEN, & J. E. LAWESSON. Passifloraceae. he: G. Harling & B. Sparre (editors), Flora of Ecuador. University of Góteborg & Riks museum, Stockholm (in press The American species of Passi- floraceae. Publ. Field Mus. Nat. Hist., Bot. Ser. 19: 1-613 —John M. MacDougal, Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166-0299, U.S.A. NOTES ON RHODOGNAPHALOPSIS AND BOMBACOPSIS (BOMBACACEAE) IN THE GUAYANAS The genus Rhodognaphalopsis was es- tablished by A. Robyns in his revision of Bom- bax in the broad sense (1963). In his generic key, he separated Rhodognaphalopsis and Rhodognaphalon from Bombacopsis, Er- iotheca, and Pachira using only pollen characters: “Pollina colpata vel colporata; sexinium reticulatum ad interdum fragmen- timuratum”” for characterizing the last three genera and “Pollina porata vel cop(or)ata [ sic ]; sexinium structura uniformi, tegellatum ad punctatitegillatum tegilloque processibus spinulatis vel baculatis praedito" for the first two. By comparing Robyns's generic descrip- tions, one can also find that Rhodogna- phalopsis is often lepidote on several organs while Bombacopsis is not; apparently it was overlooked that at least Bombacopsis qui- nata (Jacq.) Dugand (= Pochota quinata (Jacq.) W. D. Stevens) has prominently lep- idote flowers. The palynological differences between Rhodognaphalopsis and Bombacopsis are not impressive (e.g., see fig. 25 in Robyns, 1967), but perhaps more importantly, Bom- bacopsis is palynologically diverse (see Nils- son & Robyns, 1986), enough so as to ac- commodate Rhodognaphalopsis easily. Since this questionable palynological difference is not correlated with any gross morphological characters, it appears more appropriate to merge the two genera. It should be noted, however, that while Rhodognaphalopsis is relatively homogeneous, Bombacopsis is not. It could well be that some species of Bom- bacopsis, as circumscribed by Robyns, will eventually be found to be better placed in Pachira. Rhodognaphalopsis most resem- bles the type element of Bombacopsis and there is little danger of Rhodognaphalopsis falling into the synonymy of Pachira. It can also be noted that Hutchinson (1967) judged the palynological difference to be of little importance and considered Rhodognapha- lopsis, as well as the African genus Rhodo- gnaphalon, to be indistinguishable from Bombacopsis. Pittier erected the genus Bombacopsis for two Central American species that he consid- ered intermediate between Pachira and Bom- bax. These, Bombacopsis sessilis (Benth.) Pittier and B. fendleri (Seem.) Pittier (= Bombacopsis quinata (Jacq.) Dugand), have the seeds embedded in wool, character- istic of the genus Bombax, as contrasted with Pachira, in which the seeds are embedded in the fleshy dissepiments of the capsule. They further share with Bombax the manner in which staminal fascicles divide at once into single filaments. Pittier distinguished Bom- bacopsis from Bombax by the fewer stamens and more slender flowers of the former. Later, Robyns (1963) also recognized the genus Bombacopsis. He distinguished it from Bom- bax by its persistent or even accrescent calyx, while from Pachira it was separated by hav- ing abundant, wooly, elongated investiture of the seed; by possession of smaller flowers; by differences in the pollen and cotyledons; and usually by smaller and more numerous seeds. Many of the species were transferred from Bombax, while some others had previously been recognized in Pachira. Recently, it has been reiterated that Po- chota Ramirez Goyena has nomenclatural priority over Bombacopsis (Stevens, 1987). Since then, Bombacopsis has been proposed for conservation (Nicolson & Robyns, 1987). ANN. Missouri Bor. GARD. 75: 396-398. 1988. Volume 75, Number 1 1988 Notes Given that we cannot accept Rhodogna- phalopsis and Bombacopsis as distinct gen- era and that several of these species will be treated in the upcoming Flora of the Vene- zuelan Guayana, we are left with the dilem- ma of how to treat them. One could make the new combinations of Bombacopsis in an- ticipation of conservation, but it must be tak- en into consideration that the two previous proposals to conserve Bombacopsis against Pochota have been rejected and that there is no assurance that the current proposal will be successful, a process which in any case will require years to complete. If, on the other hand, the more strictly correct approach of making the combinations in Pochota is fol- lowed, the distinct possibility of the names being overturned will remain. Although the latter choice will require the larger number of transfers, to account for Bombacopsis as well as for Rhodognaphalopsis, we have re- luctantly concluded that it is preferable. Ac- cordingly, the following new combinations are proposed: Pochota amazonica (Robyns) Steyerm. & W. D. Stevens, comb. nov. Bombacop- sis amazonica Robyns, Bull. Jard. Bot. État 33: 186. 1963. Pochota coriacea (Martius & Zucc.) Stey- erm. & W. D. Stevens, comb. nov. Born- bax coriaceum Martius & Zucc. in Mar- tius, Nov. Gen. Sp. Pl. 1: 93. 1826. Rhodognaphalopsis coriacea (Martius & Zucc.) Robyns, Bull. Jard. Bot. État 33: 289. 1963. Pochota coriacea subsp. orinocensis (Ro- byns) Steyerm. & W. D. Stevens, comb. nov. Rhodognaphalopsis coriacea subsp. orinocensis Robyns, Mem. New York Bot. Gard. 17: 197. 1967. Pochota cowanii (Robyns) Steyerm. & W. tevens, comb. nov. Bombacopsis cowanii Robyns, Mem. New York Bot. Gard. 17: 190. 1967. Pochota flaviflora (Pulle) Steyerm. & W. . Stevens, comb. nov. Bombax flavi- O Pulle, Recueil Trav. Bot. Néerl. 9: 150. 1912. Rhodognaphalopsis flaviflora (Pulle) Robyns, Bull. Jard. Bot. État 33: 285. 1963. Pochota gracilis (Robyns) Steyerm. & W. tevens, comb. nov. Rhodogna- phalopsis gracilis Robyns, Mem. New York Bot. Gard. 17: 199. 1967. Pochota humilis (Spruce ex Decne.) Stey- erm. . D. Stevens, comb. nov. Pa- chira humilis Spruce ex Decne., Fl. Serres Jard. Eur. 23: 52. 1880. Rho- dognaphalopsis humilis (Spruce ex Decne.) Robyns, Bull. Jard. Bot. État 33: 294. 1963. Pochota maguirei (Robyns) Steyerm. & tevens, comb. nov. Rhodo- gnaphalopsis maguirei Robyns, Mem. New York Bot. Gard. 17: 200. 1967. Pochota minor (Sims) Steyerm. & W. D. Stevens, comb. nov. Carolinea minor Sims, Bot. Mag. 34: pl. 1412. 1811. Pachira minor (Sims) Hemsley, Biol. Cent.-Amer., Bot. 1: 124. 1879. Bom- bax minus (Sims) Ducke, Arch. Jard. Bot. Rio de Janeiro 6: 65. 1933. Rho- dognaphalopsis minor (Sims) Robyns, Bull. Jard. Bot. État 33: 278. 1963. din jenmani [i] Oliver, Hooker's Icon. Pl. 18: pl. 720. 1 7. 1898. Bombacop-^ sis jenmani (Oliver) Lasser in H. Pittier et al., Catalogo de la Flora Venezolana 2: 133. 1947. Pachira cardonae Cuatrec., Trop. Woods 101: 15. 1955. Pochota nitida (Kunth) Steyerm. & W. D. tevens, comb. nov. Pachira nitida Kunth, Nov. Gen. Sp. 5: 302. 1821. Rhodognaphalopsis nitida (Kunth) Ro- byns, Bull. Jard. Bot. Etat 33: 282. 1963. ombax poissonianum Schumann in Martius, Fl. Bras. 12(3): 225. 1886. Pachira obtusa Spruce ex Schumann in oe Fl. ies 12(3): 232. 1886. Bombax obtusum (Spruce ex Schumann) Bakh., Bull. Jard. Bot. Suse sér. Pochota obovata (Robyns) Steyerm. & W. D. Stevens, comb. nov. Bombacopsis 398 Annals of the Missouri Botanical Garden obovata Robyns, Mem. New York Bot. Gard. 17: 192. 1967. Pochota orinocensis (Robyns) Steyerm. & W tevens, comb. nov. Bombacop- sis orinocensis Robyns, Mem. New York Bot. Gard. 17: 193. 1967. Pochota pseudamazonica (Robyns) Stey- erm. & W. D. Stevens, comb. nov. Bom- bacopsis pseudamazonica Robyns, Mem. New York Bot. Gard. 17: 193. 1967. Pochota pseudofaroensis (Robyns) Stey- erm. . D. Stevens, comb. nov. Rho- dognaphalopsis pseudofaroensis Ro- byns, Mem. New York Bot. Gard. 17: 201. 1967 Pochota sordida (R. Schultes) Steyerm. & W. D. Stevens, comb. nov. Bombax sor- didum R. Schultes, Bot. Mus. Leafl. 16: 75. 1953. Rhodognaphalopsis coria- cea var. sordida (R. Schultes) Robyns, Bull. Jard. Bot. État 33: 292. 1963 Rhodognaphalopsis discolor Robyns, Mem. New York Bot. Gard. 17: 198. 1967. Pochota trinitensis (Urban) Steyerm. & W. D. Stevens, comb. nov. Pachira trin- itensis Urban, Notizbl. Bot. Gart. Berlin- Dahlem 8: 28. 1921. Bombacopsis trin- itensis (Urban) Robyns, Bull. Jard. Bot. Etat 33: 191. 1963. Bombacopsis mucronulata Pittier, Arb. Arbust. Venez. /3: 34. 1923. Bombacopsis pachiroides Pittier, Arb. Arbust. Venez. 2/3: 35. 1923. Pochota wurdackii (Robyns) Steyerm. & W. D. Stevens, comb. nov. Bombacop- sis wurdackii Robyns, Mem. New York Bot. Gard. 17: 194. 1967. LITERATURE CITED 1967. xfor Mio: D. e & A. ROBYNS. 1987. (883) Proposal to conserve 5024a Bombacopsis against Pochota (Bombacaceae). Taxon 36: 655-656. S S. & A. Bombacaceae. In: World Pollen and Spore Flora 14. Stockholm RoBYNs, A. 19 ssai de monographie du genre Bom- bax s.l. (Bombacaceae). Bull. Jard. Bot. État. 33: Hurc HINSON, 4 The Genera of Flowering Plants, 967. Bombacaceae. /n: B. Maguire et al., The Botany of the es nm VII. Mem New York Bot. Gard. 17: STEVENS, W. D 87. On de uon and recognition of the genus Pochota Ramirez Goyena (Bombaca- ceae). Taxon 36: 458-464 — Julian A. Steyermark and Warren Doug- las Stevens, Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166, U.S.A. Chromosome numbers of plants have ‘become i increas- ingly the y dex evolution. Nearly all modern taxonomic monographs and revisions include chromosomal data and use this informa- ti on in analyzing species or generic relationships or evolu- ) plant ie ets L 4 P & nan Ch 4 1 " i romosome a year, scattered in the literature. They are reported in brief ` articles, as major compilations ip: families or genera, and in revisions and monographs in numerous botanical and 4t umbers | other journals and in books. Index to Plant Chromosome Numbers is compiled by an international committee and collated and edited by Peter dblatt, Missouri Botanical Garden. It is indispensible to those engaged in a variety of botanical studies ranging from atics and evolution to plant breeding, forestry, and Apnea Counts for all plant groups — algae, fungi, bryophytes, - fidophytes, and spermato phytes—are included and arranged alphabetically by family, genus and species. Each count es the name of the taxon as used in the original report, the number reported, and reference to the mp pert ences are contained i in a bibliography which contains citations at the level of about ay per year. Ís published as separate volumes in the series MONOGRAPHS in Systematic Botany from the Missouri Botanical en. simplest way to order is to use the order form below or a photocopy of it. Please include aiian requested + you order on separate paper. Orders should be prepaid: a $1.00 fee will be sod to orders requiring invoicing, &n £ ; d © shipments are e until payment is received. Please add $1.50 for o Postage and handlin 1975-1978. Vol. 5. 1981... $15.00 copy(ies) of IPCN 1975-1978 @ $15.00 $ — — | 1979-1981. Vol. 8.1983.__ $15.00 copy(ies) of IPCN 1979- 1981 a $15.00 TU Y CEPR 1982-1983. Vol. 13. 1985. $1500 copy(ies) of IPCN 1982-1983 @ $15.00 — 1984-1985. Vol. 23. 1988. $15.00 copy(ies) s) of IPCN 1984-1985 @ $15.00 — — “Total for books — — — -Add hanc oa a “Payment enclosed. , ink wil be added 0 total). CONTENTS les in Plant Community Diversity and Floristic Composition on Environmental and Geographical Gradients Alwyn H. Gentry Patterns of Vascular Plant Diversification in the Fossil Record: Proof and Conjecture ae . Niklas ^ Effects of Aridity on Plant Diversity in the Northern Chilean Andes: Results of a Natural š . Experiment Mary T. Kalin posse Francisco A. Squeo, Juan J. Armesto € —— Carolina Villagrán (085 Patterns of Species Diversity in Anuran MM in the American Tropics William E Duellman > Management of Habitat Fragments in a Tropical Dry Forest: Growth Daniel H. . Janzen .... | Factors Controlling Species SEN Overview and Synthesis Jared Diamond ... 117 A Contribution to the Pollen Morphology of Neotropical Lauraceae Bhoj Raj € Hab ~ van der Werf. > The Taxonomic Significance of Pollen Morphology in the Columnea Alliance (Coser ES Gesnerioideae) Karen J. Fritze & Norris H. Williams aama - Observations on the Chromosome Cytology of Velloziaceae Peter Goldblatt & Muriel E. Poston 22 | Chromosome Counts and Karyomorphology of Some West Tropical African Scilleae (Lili- , actas): 5.0. Oyewole —— S Breii (Palmae) i in Central America ~ Andrew Henderson & Greg de Nevers —— Rai Variation in n Penco hirtum À A. Chev. dose p PNU Oyewole .... Bde ` The Architecture of TORES in the Myra Becks Weberling = Flora of the Venezuelan Guayana—IV Julian A. Steyermark nimme Ten Novelties i in Xyris Otyridaceae £ from the Planalto of Brazil - R. Kral & Maria . Graças , de Lapa Wanderley .... : Ne and d Noteworthy Taxa from Panama E ~=- > Gordon McPherson Annals of the — Missouri - Bot anical a Volume 75, Number 2 Annals of the Summer 1988 Missouri Botanical Garden The Annals, published Bem contains papers, primarily in systematic botany, con tributed from the Missouri Botanical Garden, St. Louis. Papers originating outside de Garden will also be accepted. Authors should write the Editor for information concerning arrangements for publishing in the Annars. Instructions to Authors are printed on the PN back cover rot the last issue e of each volume. Editorial Cocaine E Rog ers eoim Marshall R. Crosby Editor, Missouri Botanical Garden Missouri Botanical Garden | Janice | Wi Soa onan | .— Gerrit Davidse e Editorial l. Assistant, Missouri i Botanical — Missouri Polang Garden John D. a Missouri enter Garden: id ` Saint Louis Universtiy: 3 aie Goldblatt D > Missouri Botanical Garden | E Henk van P Wer i Botanical Garden Volume 75 Number 2 1988 Annals of the Missouri Botanical Garden W POLLEN MORPHOLOGY OF PILLANSIA L. BOLUS (IRIDACEAE)! Peter Goldblatt? and Bruce A. Stein? ABSTRACT The monotypic southwestern Cape genus Pillansia is a nO isolated member of Iridaceae subfamily Ixioideae. It shares and inflorescence structure other Iridaceae. some iis apomorphies with Ixioideae, although it nicle rather than a differs in its unspecialized leaf anatomy Pillansia L. Bolus is a monotypic genus of Iridaceae—Ixioideae, the largest of the four subfamilies currently recognized. It is a relatively rare, narrow endemic of the south- western Cape, South Africa (Goldblatt, 1977) and is restricted to rocky sandstone sites in the Caledon district. It has a basic chromo- some number of x = 20 and thus appears to be paleopolyploid. Most other genera of Ixioi- deae have base numbers in the x — 11-9 range or are neopolyploid with n = 16-13. Although Pillansia diverges in some impor- tant respects from other members of the subfamily, it is widely accepted as belonging to Ixioideae. Nevertheless, it is taxonomically isolated and appears to have no identifiable close relatives. It accords with Ixioideae in several specialized features (synapomorphies) (Goldblatt, in prep.) that characterize the subfamily. These include a long-lasting peri- anth; sessile flowers subtended by a pair of opposed bracts; a well-developed, though short, perianth tube; and a basal rooting corm. The pollen morphology of Pillansia has ' Supported by U.S. National Science Foundation grants DEB 81-19292 and BSR 85-00148 to P.G. We thank Mr. Mike Veith, Washington University, for technical assistance. ? B. A. Krukoff Curator of African Botany, Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166, U.S.A 3 Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166, U.S.A. Current address: International Program, The Nature Conservancy, 1785 Massachusetts Avenue NW, Washington, D.C. 20036, U.S.A. ANN. Missouni Bor. Garb. 75: 399-401. 1988. 400 Annals of the Missouri Botanical Garden Ficu Scanning electron micrographs of Pillansia sip tid — 1. 5 um. Whole grain in equatorial view x], 200. E Deal of the exine surface X 6,000. Scale bars — until now been unknown, and it is important to establish whether it has the basic reticulate (to retipilate) exine of most Iridaceae (Schulze, 1971) or the specialized micropunctate and micropapillate exine characteristic of Ixioi- deae (Schulze, 1970, 1971) METHODS Pollen of Pillansia templemanii (Baker) L. Bolus was extracted from flowers from herbarium specimens (voucher: Bolus s.n., MO 2080184) and subjected to standard ac- etolysis treatment (Erdtman, 1960). Aceto- lyzed pollen was mounted on aluminum stubs, gold coated, and viewed in a Hitachi S-450 scanning electron microscope. OBSERVATIONS Pollen of Pillansia is 65-75 um long (po- lar axis as measured from scanning electron micrographs), ellipsoid to widely ellipsoid (Fig. 1), and monosulcate with the sulcus running the length of the grain. The exine is densely micropunctate (punctitegillate of some au- thors) and micropapillate (Fig. 2). The pores in the tectum are round to irregular in shape and 0.15-1 um in diameter. There are ca. 50 pores/100 um’. The papillae are small, mostly rounded excrescences, and up to 0.4 um in diameter and height. They are scattered fairly regularly with a density of ca. 52/100 um? between the pores. DISCUSSION Pillansia differs from all other Ixioideae in its paniculate inflorescence and leaves lack- ing a distinct central vein or central cluster of veins, both unspecialized conditions. Other Ixioideae have a spicate inflorescence or one believed to be derived from a spike (flowers solitary on branches or flowers sessile in a corymbose panicle) and have distinct (or more or less distinct) central veins. Pillansia differs further from other Ixioideae in leaf anatomy (P. Rudall, pers. comm.) by having isodia- metric to longitudinally elongated mesophyll cells and epidermal cells with nearly straight walls and one or no papillae. Other Ixioideae have transversely elongated mesophyll cells and epidermal cells with sinuous walls and two or more papillae The pollen of Bilans clearly matches that of other Ixioideae, which lends further Volume 75, Number 2 1988 Goldblatt & Stein Pillansia Pollen Morphology 401 support for its retention in that subfamily. In a survey of pollen morphology in 21 of the 45 genera of Ixioideae, Schulze (1970, 1971) found that 20 have micropunctate exines. Only Micranthus is reported to deviate, hav- ing the retipilate exine characteristic of the other subfamilies. Schulze's observations have been confirmed by SEM studies by de Vos (1974a, b, 1982) for Syringodea, Duthieas- trum, and Tritonia, and by Straka & Fried- rich (1984) for a Malagasy species of Gladi- olus (as Geissorhiza). Pollen morphology provides no additional evidence for the rela- tionships of Pillansia within Ixioideae —It re- mains a puzzling isolated and apparently re- lictual genus. Its unusual combination of features suggests that it may represent a link between Ixioideae and the remaining Irida- ceae and could, it fact, be very close to the ancestral type of Ixioideae. t now seems all but certain that micro- punctate exine is basic for Ixioideae, and the presence of retipilate exine in Micranthus, an apparently typical member of the subfam- ily, is surprising, particularly as the closely allied Thereianthus is reported to have mi- cropunctate exine. This may represent an example of a reversal to an ancestral condi- tion, but pollen of all three species of Mi- cranthus should be critically reexamined. LITERATURE CITED 1974a. Duthiella, 'n nuwe genus van die Iridaceae. J. S. African Bot. 40: 301-309. 74b. Die Suid-Afrikaanse genus Syringo- des. t 9. I Bot. 40: 201-254. . The African genus Tritonia Ker-Gawler T ds ‘part 1. J. S. African Bot. 48: 105-163. ERDTMAN, C. 60. The acetolysis method — a revised description. Svensk Bot. Tidskr. 54: 561-564. GoLDBLATT, P. 1977. Chromosome number in Pillansia Gar (Iridaceae). Ann. Missouri Bot. . 64: 136-138. ScHULZE, W. 1970. Beitrage zur Pollenmorphologie der Í ae Ixioideae. Wiss. Z. Friedrich-Schiller- v. Jena, Math.-Naturw. Reihe 19: 437-445. Beitráge zur Pollenmorphologie der Iridaceae und ihre Bedeuting für die Taxonomie. Feddes Rep. 82: 101-124. STRAKA, H. & B. FRIEDRICH. 1984. Palynologia Mad- agassica et Mascarenica. Gymnospermae und Mono- cotyledones. Trop. & Subtrop. Pflanzenwelt 49: 1- 89. EIGHT NEW SPECIES AND ONE NEW COMBINATION OF NEOTROPICAL LAURACEAE?’ Henk van der Werf? ABSTRACT k in progress on the systematics of neotropical Lauraceae had yielded iiid undescribed species. In Wor this contribution, eight species (Aiouea obscura van der Werff, Aiouea vexatrix van der Werff, Pho ebe elegans van der W. Werff, Caryodaphnopsis illustrated, and discussed. A new amb aioi: Ocotea erectifolia (Allen) van der Werff is made. Lauraceae are a large tropical family of trees and shrubs with the number of species in the New World estimated at 700-800. The taxonomy of the neotropical Lauraceae is poorly understood, and the entire family needs much work. The facts that many Lau- raceae have small, inconspicuous flowers and are not frequently collected, that the genera are poorly defined, and that many species are known only from a few poor specimens have rendered the family almost inaccessible for the nonspecialist. Recent collections have shown that quite a few very distinct species await descriptions. In this contribution, eight species, mostly belonging to the smaller and relatively better-known genera, are described and discussed. A new combination is made and its synonymy given. Aiouea obscura van der Werff, sp. nov. TYPE: Costa Rica. Puntarenas: along highway from Palmar Norte to Chaca- rita, ca. 2 km N of Chacarita. Tree, 10 m. Flowers pale green. B. Hammel, M. Grayum & G. de Nevers 15197 (holo- type, MO; isotypes, BM, CR, F, MEXU, NY, PMA, U). Figure 1. Arbor, 10 m. Ramuli graciles, bier glabri. Gemma terminada glabra. Folia alterna, uste elliptica, basi apiceque acuta, 15-18 x 3-4 cm, membranacea, sub- triplinervia, brochidodroma, in sicco olivacea; nervi la- terales 2-3. Venatio super parve elevata, subtus magis elevata. Domatia plerumque in axillis nervorum lateralium basalium. Petioli teretes, glabri, 1-1.5 cm longi. Inflo- rescentiae axillis bractearum deciduarum super partem foliiferam, 15 cm longae, graciles. Pedicelli 8-10 mm longi. Tepala 6, aequalia, glabra, ca. 2 mm longa, late jin Stamina 9, 2-locellata; 6 exteriora introrsa, ca. m longa, anthera parum latiore quam filamento, apice cathe rae i M filamento pubescenti; 3 in- teriora extrorsa, ca onga, apice locellos exce- i ° magnae a. 0.4 mm diametro, parum e. Staminodia non visa. Ovarium gla- rum, ovatu "lm m longum, sensim in stylo brevi attenuatum. [ees ignotus. Tree, 10 m tall. Twigs slender, terete, gla- brous. Terminal bud glabrous. Leaves alter- nate, narrowly elliptic, the base and tip acute, 15-18 x 3-4 cm, membranaceous, subtrip- liveined, the basal veins reaching + % to the apex, the other 2-3 pairs of lateral veins in the upper half of the lamina, the lateral veins all curving toward the apex and loop-con- nected, drying olive green. Venation and re- ticulation slightly raised on upper surface, slightly more so on the lower surface. Domatia often present in the axils of the large lateral veins. Petioles terete, glabrous, 1-1.5 cm ! [ thank the various botanists working in Mexico, Panama, and Colombia for their o excellent collections, sent to me for pel gir pe Their work is essential for a better understanding of the ts of js A oai Dr. J. Dwyer kindly checked the Latin Pisa John Myers made m BR, and NY are gratefully acknowled, commented on p the Donors es , G, 2 Herbar uraceae. Dr. W. Burger ged. rium, Missouri "Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166-0299, U.S.A. ANN. Missouni Bor. GARD. 75: 402-419. 1988. Volume 75, Number 2 1988 n der Werff 403 New Neotropical Lauraceae FIGURE 1. Outer stamen.—E. long. Inflorescences above the leaf-bearing part of the twigs, in the axils of deciduous bracts; terminal buds inconspicuous but al- ways present above the lateral inflorescences. Inflorescences glabrous, to 15 cm long, slen- der, paniculate. Flowers pale green. Pedicels 8-10 mm long. Tepals 6, equal, glabrous, 2 mm long, broadly elliptic. Stamens 9, 2-celled; the outer 6 introrse, ca. 1 mm long, the anther slightly wider than the filament, the tip of the anther protruding beyond the anther cells, the filaments pubescent; inner 3 stamens ex- trorse, ca. 1.2 mm long, the glands large, ca. 0.4 mm diam., attached slightly above the base of the filaments, reaching the upper part of the filaments, leaving the anthers exposed Aiouea obscura. — A. Flowering branch.—B. Flower.—C. Inner stamen with basal glands.—D. Ovary. but shielding the ovary; connective tissue also protruding beyond anther cells. Staminodia not seen. Ovary glabrous, ovate, ca. 1 mm long, gradually narrowed into the short style. ruit unknown. Aiouea obscura is only known from the type collection. Characteristics are laxly flow- ered inflorescences, relatively long pedicels, and dark-drying, subtripliveined leaves with loop-connected lateral veins. It resembles closely several other dark-drying Lauraceae with membranaceous leaves and lax inflores- cences, such as Ocotea tenera Mez & J. Smith (known from Costa Rica) and Phoebe glabra van der Werff (southern Mexico). 404 Annals of the Missouri Botanical Garden Aiouea obscura differs from both species in having two-celled anthers; from O. tenera it differs further in having subtripliveined leaves and from P. glabra in having narrower leaves with less prominently raised reticulation. None of the other Central American species are similar to A. obscura. These observations re- emphasize that our current generic concepts place seemingly closely related species in dif- ferent genera and that these concepts ur- gently need re-examination. Aiouea vexatrix van der Werff, sp. nov. TYPE: Panama. Panamá: Cerro Cam- ana, above Capira, elev. ca. 900 m, 8?40'N, 79%50'W. Slender treelet, 3 m tall. Stem with small red ants in center. Perianth green. McPherson 9226 (ho- lotype, MO). Figure 2A-G. Frutex vel arbor parva, ad 7 m. Ramuli juvenales angulati, minute fusco-puberuli, vetustiores teretes glab- rique. Ramuli fistulosi frequenter formicis habitati. Petioli ad 1.5 cm longi, glabri vel minute puberuli. Folia alterna, in sicco atro- pr glabra, chartacea, elliptica, basi apiceque acuta, 13-27 x 5-9 cm, super costa nervisque immersis, retieulatione parve elevat ta, subtus costa ner- visque elev vi laterales utroque costae latere 6- 8. Inflo usss al glabrae vel minute puberulae versus basim, a cm longae; ramuli inflorescentiarum complanati in sicco. Flores in vivo virides, ad 3 mm longi. Tepala 6, aequalia, glabra, erecta per anthesin, late ovata, 2 mm longa, 1.7 mm lata. Stamina 9, 2-locellata, glabra, inclusa, 6 exteriora i ntrors sa, 3 interiora extrorsa filamentis oribus. Staminodia nulla. Ovarium tepalis in bear cupulae e persistentibus, pedicello p latim in cupula dilatato. Fructus ellipsoideus, ad 2 c longus, fere omnino exsertus. Shrub or small tree, 5(-7) m tall. Leafy twigs angular, minutely brownish puberulous, becoming round and glabrous on older parts; twigs consistently hollow and with pores giving access to the hollow center; often ants present in the hollow twigs (fide collectors). Terminal bud small, with very fine, copper-colored pu- bescence. Petioles to 1.5 cm long, minutely puberulous or glabrous, the lamina decurrent as narrow ridges. Leaves drying dark olive green, rarely gray-green, glabrous, charta- ceous, elliptic, rarely slightly obovate, the base and apex both acute, 13-27 x 5-9 cm, the upper surface with immersed midrib and lat- eral veins, the final reticulation slightly raised; lower surface with midrib and lateral veins elevated, the final reticulation less elevated. Lateral veins 6-8 pairs. Inflorescences axil- lary, often seemingly terminal, glabrous or with some minute puberulence especially near the base, 6(-9) cm long, the branchlets not terete, these flattened after drying. Flower glabrous, ca. 3 mm long, on pedicels to 3 mm long; tepals 6, equal, glabrous, erect at an- thesis, broadly ovate, ca. 2 mm long, 1.7 mm wide. Stamens 9, all 2-celled, the outer 6 introrse, the inner 3 extrorse. Outer stamens 1.5 mm long, the filament glabrous and ca. 0.6 mm long, the anther narrowly triangular, wider than the filament, ca. 0.9 mm long wit a sterile apical section; anther cells large, ca. 0.3 mm long. Inner stamens 1.5 mm long, the anther 0.8 mm long, with sterile tip; fil- ament slender with 2 large glands attached near the base. Glands collar-shaped, the basal part spreading horizontally, then abruptly curved inward and downward (toward the ovary) the tip of the gland flattened and resting on the upper part of the ovary. Stam- inodia lacking. Ovary glabrous, ellipsoid, ca. 0.8 mm long, largely sunken in the flower tube, at the tip gradually narrowed into the style, this 0.8 mm long. Infructescences to 7 cm long. Cupule ca. 8 mm diam., the tepals persistent on the rim, the pedicels gradually widened into the cupule. Fruit ellipsoid, ca. 2 cm long, almost completely exserted. Oc- casional stamens remaining attached to the cupule in late fruiting stage. Paratypes (all MO). | PANAMA. PANAMÁ: Cerro Cam pana, Correa 295, 1026; same locality, Croat 12153, 14689, 17203, 25 120, 35960; same locality, Garner 13; same locality, Gentry 1832, 5776; same locality, Hamilton 4056, 4061; same locality, Hammel 3776; same locality, Kirkbride 245; same locality, Luteyn 1812; same locality, McPherson 7461, 7921; same locality, Miller 975; same locality, Mori 1917, 2457, 7701; same 11611; same locality, Sytsma 1150, 2942, Rita Ridge Road, Correa 1056; same locality, Dressler 3705; same locality, Foster 1735; same locality, Sytsma 4238, 4252; Cerro Brewster, de Nevers 5573. COMARCA DE SAN BLAS: El Llano-Carti Road, de Nevers 4240. Aiouea vexatrix has a limited distribution Volume 75, Number 2 1988 van der Werff New Neotropical Lauraceae 405 TABLE 1 Diagnostic characters for Aiouea vexatrix, Ocotea paulii, O. atirrensis, and O. nicaraguensis. The numbers in parentheses given with the flowering and fruiting periods represent the number of collections examined. cotea Aiouea vexatrix Ocotea paulii Ocotea atirrensis nicaraguensis Anthers 2-celled 4-celled 4-celled 4-celled upule tepals persistent tepals deciduous tepals deciduous tepals persistent Leaf color when dark olive green green dark olive green reen dry Leaf shape elliptic elliptic obovate obovate to oblan- Leaf texture Length of inflores- cence Flowering period Fruiting period thinly chartaceous 1⁄4 leaf length or less April-July (15) July- Mi (20) stiffly chartaceous + equal to leaves October-January (11) January-May thickly chartaceous X equal to leaves thinly chartaceous X equal to leaves January- April (22) April-September (13) in central Panama, where the abundance of collections indicates that it is common. e placement of this new species in the genus Aiouea is provisional. The Central American species of Aiouea are quite differ- ent morphologically from the South American species, which include the type of the genus (van der Werff, 1984, 1987). Aiouea vex- atrix is another example of such an aberrant Aiouea species. It has the following charac- ters unusual for Aiouea: lack of staminodia, short inflorescences with flattened axes, dark olive green leaves, and persistent tepals on the cupule. It is included in Aiouea solely because of its hermaphrodite flowers with nine two-celled stamens and because it does not agree with the other, much better defined genera with nine two-celled stamens (Aniba, Beilschmiedia, Cryptocarya, Kubitzkia, Phyllostemonodaphne, Urbanodendron). As discussed below, A. vexatrix is probably re- lated to Ocotea. The current circumscriptions of the neotropical genera of Lauraceae attach much importance to the number of cells of the anthers. This is an artificial character that obscures true relationships. However, dis- carding the present imperfect generic clas- sification implies its replacement with a better classification, which I cannot offer at this mo- ment. Therefore, I place this new species in Aiouea, a genus consisting of a group of closely related species in the lowlands of Venezuela, the Guianas, Brazil, and Paraguay, plus sev- eral unrelated species in the Andes and Cen- tral America. It is likely that the Andean and Central American species have been inde- pendently derived from Ocotea or Nectandra ancestors that lost two of their four anther cells. Aiouea vexatrix is related to a group of Ocotea species that grow as shrubs or treelets, have angular (or almost winged), hollow twigs frequently inhabited by ants, flattened inflo- rescence branchlets, and glabrous flowers with erect tepals. The following names have been applied to these species: Ocotea nicaraguen- sis Mez, O. paulii Allen (Fig. 2H), O. pe- dalifolia Mez, O. pentagona Mez, O. atir- rensis Mez & J. D. Smith and O. wedeliana Allen. Their distributions are in Panama, Cos- ta Rica and Nicaragua. A survey of specimens at the Missouri Botanical Garden suggests that four species are involved, A. vexatrix, O. paulii (isotype, MO!), O. atirrensis (iso- type, US!), and O. nicaraguensis (type W, probably destroyed; type photo, MO!). Th main differences between these species are presented in Table 1. Ocotea wedeliana is known to me only from three isotypes (MAD, US, GH), all rather poor specimens from which one cannot draw firm conclusions. The flowers have four-celled anthers; the leaves are char- taceous and dry dark. A fruiting collection identified by Allen as O. wedeliana has the cupule of O. paulii, but thinner leaves. The distributions of the four recognized species follow: Aiouea vexatrix is only known from areas rather close to Panama City (Pa- 406 Annals of the Missouri Botanical Garden "wa 2. A-G. Aiouea vexatrix. — A. Flowering branch.—B. Flowers.—C. Inner stamen with glands.—D. —E, F. Outer stamens seen ventrally and dorsally.—G. Fruits. —H. Fruit of Ocotea pauli. namá, Colón, Comarca de San Blas). Ocotea paulii has a wider distribution in Panama (from Darién to Veraguas) and is uncommon in Costa Rica. It occurs with Æ. vexatrix on Cerro Campana, but no intermediates have been found there. Three collections, inter- mediate between 4. vexatrix and O. paulii, are known from the western edge of the dis- tribution of 4. vexatrix. Both Hammel 3557 O) and Sytsma 4407 (MO) have leaf tex- ture and color of 4. vexatrix, flowers with four-celled anthers, and inflorescences longer than typical for 4. vexatrix, but shorter than for O. paulii. Allen 3439 (AA), identified as O. wedeliana Allen by Allen, has the thin leaves of A. vexatrix and cupule shape of O. paulii. These specimens, as well as the type of O. wedeliana Allen (US!, GH!, MAD!) are probably hybrids between 4. vexatrix and O. pauli. Ocotea atirrensis has been collected fre- quently in Costa Rica, but I have seen only Volume 75, Number 2 1988 van der Werff 407 New Neotropical Lauraceae two collections from Panama, both from Bo- cas del Toro. No intermediates between O. atirrensis and O. paulii are known to me. Ocotea atirrensis is characterized by large, obovate, chartaceous leaves that dry dark green and have acuminate tips; its tepals are not persistent on the cupule. cotea nicaraguensis includes O. penta- gona (syntypes: Biolley 7106, Tonduz 7613, 8362, all BR!) and a syntype of O. pedalifolia (Pittier 9172, BR!). The other syntype of O pedalifolia (Pittier 9179, BR!) is O. atir- rensis. Not as well represented as the other species, O. nicaraguensis awaits more spec- imens for a better understanding. Diagnostic characters are the strongly angled stems and stifly chartaceous, green-drying leaves with acute tips; the cupule is crowned with per- sistent tepals. The few collections do not in- dicate a well-defined flowering period. Its ob- ovate to oblanceolate leaves are distinct from the other species. Ocotea nicaraguensis is known from Costa Rica and Nicaragua. Old collections of A. vexatrix have been distributed as O. subsericea Standley and as O. atirrensis and may be present in other herbaria under these names. i guna cogolloi van der Werff, . nov. TYPE: Colombia. Antioquia: Municipio de San Luis, left bank of Rio Claro, 325-500 m, 5°53'N, 74°39' W. Tree, 15-18 m, flowers yellow. A. Co- gollo & R. Borjo 2019 (holotype, JAUM, n.v.; isotype, MO). Figure 3. Arbor, 30 m. Ramuli teretes, juniores minute ferru- gineo- -pubescentes, vetustiores glabrescentes. Folia op- nnata, nervis utroque cos- Laminae men 15-20 x 5-8 cm, basi acuta, apice acuminata; super glabrae, venatione immersa; subtus glaucae, costa nervisque elevatis et mi- nute ferrugineo-pubescentibus pilis ferrugineis praeditis. Petioli teretes, 1.5- minute ferrugineo- pubescentes. Inflores cenciae. axillares, 5 mm longa, apicibus saepe recurvatis, minute pubescentia; 3 interiora anguste ovata, apicibus acutis m lo onga, minute pubescentia; stam- ora ca. 2 mm longa, glabra, filamentis ca. 0.6 mm longis, antheris ca. 1.3 mm longis, locellis introrsis; : interiora ca. 2 mm longa, glabra, lamen m longis, basibus 2 glandulis dbi auctis, persan ca. né, 6 mm longis, locellis extrorsis. Sta- minodia 3, glabra, apicibus dilatata. Ovarium globosum, .9 mm diametro. Stylum gracile, ca. 1.2 mm longum. Fructus pyriformis, in sicco ca. 4 cm longus. Tree, to 30 m tall. Twigs terete, the youn- ger ones with minute, ferruginous pubes- cence, becoming glabrous with age. Leaves opposite, decussate, pinnately veined, lateral veins 8-12 pairs. Laminae elliptic, 15-20 x 5-8 cm, the base acute, the apex acuminate, the upper surface glabrous with immersed veins, the lower surface gray-glaucous, waxy, the midrib and lateral veins elevated and with minute, ferruginous pubescence, the smaller veins with few ferruginous hairs. Petioles 1.5- 2 cm long, with minute ferruginous pubes- cence, terete. Inflorescences axillary, to 8 cm long, much shorter than the leaves, branched from the base, pyramidal-paniculate, minute- ly and densely ferruginous pubescent. Flowers yellow; pedicels 2-3 mm long. Tepals 6, un- equal; the outer 3 narrowly triangular, ca. 1.5 mm long, the tip often recurved, minutely brown pubescent; inner three ca. 4 mm long, narrowly ovate, the tip acute and recurved, minutely brown pubescent. Stamens 9, all 4-celled, the outer 6 ca. 2 mm long, glabrous, filaments ca. 0.6 mm long; anthers ca. 1. mm long, the cells introrse; the inner 3 slen- der, glabrous, ca. 2 mm long, filaments ca. 1.2 mm long, with anthers ca. 0.6 mm long, the cells extrorse; the filaments with 2 rather small, globose, basal glands. Staminodia 3, glabrous, ca. 1 mm long, the tip widened. Ovary globose, ca. 0.5 mm diam., the upper part with brown hairs; style slender, ca. 1.2 mm long. Fruit avocado-shaped, ca. 4 cm long when dry. Paratype. COLOMBIA. ANTIOQUIA: Municipio de San Luis, Rio Claro, 350 m (fr), 4. Cogollo et al. 2195 (JAUM, MO). Caryodaphnopsis cogolloi is known only from a small (+ 2 km?) forest remnant in the Magdalena Valley in Colombia. This forest patch is home to, based on collections by Cogollo, two other undescribed species of Car- yodaphnopsis (known from fruiting material) 408 Annals of the Missouri Botanical Garden D FIGURE 3. Caryodaphnopsis cogolloi. — A. Flowering branch.— B. Detail of lower leaf surface. —C. Flowers.— D. Flower with several tepals removed, showing outer and inner i ee stam m and ovary with Senda style. —E. Outer stamen.— F. os stamen with basal glands.—G. Ovary.—H. pe Volume 75, Number 2 1988 van der Werff 409 e New Neotropical Lauraceae and two undescribed species of Licaria. This shows how poorly collected neotropical Lau- raceae are. Caryodaphnopsis cogolloi is closely re- lated to C. inaequalis . Smith) van der Werff & Richter, a species known from the Peruvian- Brazilian border area. Both species have pinnately veined leaves, a rare character in Caryodaphnopsis (van der Werff & Richter, 1985). The two species differ as fol- lows: C. cogolloi has four-celled anthers, out- er tepals 1.5 mm long having acute, recurved tips, denser tomentum on the flowers, and a glaucous undersurface of the leaves. Cary- odaphnopsis inaequalis has two-celled an- thers, outer tepals ca. 0.5 mm long with blunt tips and not recurved, rather scarce tomen- tum on the flowers, and leaves green below. Recent collections of Caryodaphnopsis show that the neotropical species fall into two groups. One group includes the species with pinnately veined (or subtripliveined) leaves and an avocado-shaped fruit (C. cogolloi, C. inaequalis, C. theobromifolia); the other group includes the species with strongly three- veined leaves (the basal lateral veins reach the leaf apex) and small, round fruits (C. fosteri and three or four undescribed species). The new species is named after its collec- tor, Alvaro Cogollo, who collected several un- described species of Lauraceae in the Mag- dalena Valley in Colombia. Licaria velutina van der Werff, sp. nov. TYPE: Mexico. Veracruz: Mpio. San Andres Tuxtlas, Cerro Vigia near Esta- ción de Biologia Tropical Las Tuxtlas, 300 m, tree 18 m with yellowish flowers, G. Ibarra M. & S. Sinaca C. 100 (ho- lotype, MEXU; isotypes, CHAPA, HBG, MO). Figure 4. Arbor, 8-20 metralis. Ramuli obtuse angulati, lenti- cellati, juveniles albido-vel bubalino-velutini, veteres fusco- pubescentes. Folia alterna, anguste elliptica vel anguste ice basique acuta, 12-30 x 2.5—6.5 cm; ge brescentia, nervis lateralibus 10-14, venatione super im mersa vel perobscure elevata, sibus costa Vrai de vata, nervis et venatione elevata. Petioli 1-2 cm Inflorescentiae folis perbreviores, paniculatae, in axillis bractearum deciduarum, ad 13 cm longae, immaturae equi veteres pubescentes sparsiore. Flores pedicellis m longis, glabri, me E late i inis ca. 2mm ra parum maj lon Mgr 1.5 mm lati; ion xterio joria ipie incurvata, mm lo ca. 0.5 mm lata; stam a 3, 2- loce a, p ca. 1 mm longa, pe Ovarium glabrum, ellip- enuatum, ca. 1.3 mm lon gum. Infructescentia ad 7 cm longa, plerumque solo fruc- tu. Cupula cylindrica, ad 2 cm longa, 2.5 cm lata, 1.5 cm profunda pa simplici; fructus (in sicco) cupula ca. 1 cm longio Tree, 8-20 m tall. Twigs obtusely angled, with gray lenticels, the inflorescence-bearing part covered with white or yellowish veluti- nous pubescence, this changing to very short, brown pubescence on fruiting twigs. Leaves alternate, narrowly elliptic or narrowly ovate, the tip gradually acute or narrowly rounded, the base acute, 12-30 x 2.5-6.5 cm, when young with appressed pubescence, but soon glabrescent, with 10-14 pairs of lateral veins, the venation immersed or faintly elevated above, the midrib prominently raised below, the lateral veins and tertiary venation raised below. Petioles 1-2 cm long. Inflorescences much shorter than leaves, paniculate, in the axils of deciduous bracts, to 13 cm long; immature inflorescences velutinous; older in- florescences with sparser pubescence; bracts of the inflorescence with white pubescence on the outside, glabrous inside, ovate, ca. mm long, deciduous at anthesis. Flowers on glabrous pedicels, these 2-4 mm long; flowers glabrous, globose or broadly elliptic, ca. 2 mm long and 1.5 mm wide; tepals 6, incurved, the outer 3 broader than the inner 3, ca. 0. mm long, 0.5 mm wide; fertile stamens 3, their tips just exposed; anther cells almost apical, small, opening toward the tip and ex- trorse. Stamens fully connate, forming a dome ca. 1 mm high and ca. 1.3 mm wide at the base; ovary globose, ellipsoid, gradually nar- rowed into style, ca. 1.3 mm long. Staminal glands 6, reduced to small flaps, ca. 0.3 mm tall, visible at the base of the anthers. In- fructescences to 7 cm long, usually with only one fruit. Cupule deeply cup-shaped, 2 cm long, 2.5 cm wide, the cup ca. 1.5 cm deep with gray lenticels, rather thin, ca. 1 mm thick at the margin, not double-rimmed, often 410 Annals of the Missouri Botanical Garden with dried stamens attached to the margin. Fruit to 1 cm longer than the cupule when dry, ovoid, ca. 2.5 cm long. Common names. Laurel baboso, Laurel pimienta. Paratypes. MEXICO. CHIAPAS: Mpio. Tecpatán, Co- Andres Tuxtlas, rd de Biología Tropical Las Tuxtlas (fr), Calzada 178 (F, MO); same locality (fr), Calzada 695 (F, MO); same locality, May 1981 (fr), Gentry & Lott 32260 (MO); same locality, June 1981 (st), Gentry & Lott 3252 1 (MO); same locality, RA 1984 (fr), Ibarra M. 1462 (CHAPA, HBG, MEXU, MO); same locality, lote 67, July 1984 (fr), Ibarra M. & Sinac 19 (CHAPA, HBG, MEXU, MO); Laguna Escondida, 3 km NW of Estación Las Tuxtlas, 200 m, June 1985 (fl), Sinaca C. 107, 110, 111 (CHAPA, HBG, MEXU, MO); Camino a Cárdenas, 4.5 km de la Estación Las Tuxtlas, June 1985 (fl), Sinaca C. 114 (MO); lote 71, Estación Las Tuxtlas, 350 m, Aug. 1985 (fr), Sinaca C. 207 (CHAPA, HBG, MEXU, MO); Estación Las Tuxtlas, June 1981 (fr), Wendt et al. 3418 (CHAPA, CAS, LL, MEXU, MO). Licaria velutina is closely related to L. excelsa, known from southern Mexico and Panama. Licaria velutina differs by having narrow leaves 4-6 times longer than wide, densely pubescent young twigs, less pubescent inflorescences, the pubescence contrasting with the glabrous flowers, and large cupules with simple margins (our fruiting material of L. excelsa shows always double-rimmed cu- pules). All collections of L. excelsa are from above 1,000 m elevation, whereas £L. velutina is only known from elevations of 200-300 m Measurements and illustrations of stamens, staminal glands, and ovary are based on boiled parts, which shrink and change shape while drying. Nectandra mirafloris van der Werff, sp. nov. TYPE: Nicaragua. Jinotega: Laguna de Miraflores, small tree at edge of swamp, 12 May 1976, 1,200 m, Neill 329 (distributed by Seymour as no. 7204), (holotype, MO). Figure 5. Arbor parva, 2-8(-15) m. Ramuli modice appresse strigosi, Ape vetustiores lenticellati. Folia alter- a, firme chartacea, 15- x 7-10 cm, apice , basi acuta vel obtusa; petioli 1-1.5 cm longi; laminae super nitidae, reticulatione elevata sed costa nervisque immersis, glabra; subtus opacae, leviter ad- re cen trales adpresse pubescentes, ramuli pubescentia densiore, saepe albo-pubescentes. Flores pedicellati, extus dense albo-pubescentes; tepala 6, aequalia, basi connata, intus dense papillosa, ca. 3 mm | 6 exteriora ca. 0.9 m theris ee ug UK 3 interiora ca. m longa, filamentis ongis, 2 Sand. on munitis, locellis Vise ae 3, ca. 0.8 mm longa, Aca mia. Ovarium globosum, glabrum; sty- lus ovario perbrevior. Fructus late ellipticus, 2 x 1.5 cm, cupula parva, debaldea pedicellus frugifer inflatus. Small tree, 2-8(-15) m tall. Twigs gray, with small, appressed hairs, becoming gla- brous, often developing lenticels after the first year. Leaves alternate, elliptic, firmly char- taceous, 15-20 -10 cm, the tip bluntly acute, the base acute to obtuse, petioles 1— 1.5 cm long, laminae shiny above, glabrous with raised reticulation, but with immersed midvein and lateral veins, these in 6-9 pairs, opaque below, these with some appressed short hairs (especially near the base) and frequently tufts of axillary hairs, the midvein promi- nently raised, the lateral veins and final re- ticulation raised. Inflorescences in the axils of persistent leaves or deciduous bracts, 6- 12 cm long, paniculate, the main axis with some appressed pubescence, the branchlets with much denser pubescence, sometimes ap- pearing white pubescent. Flowers pedicellate, the pedicels ca. 2 mm long, densely white- pubescent; tepals 6, all equal, united at their bases, densely white pubescent outside, dense- y papillose inside, ca. 3 mm long; stamens, 9, 4-celled, the outer 6 with a very short filament, appearing sessile, quadrangular, ca. 0.9 mm long, the anther cells introrse, oc- cupying almost the entire anther and ar- ranged + in an arc, papillose; the inner 3 anthers ca. 1.2 mm long, with lateral anther cells, the filament ca. 0.3 mm long, each with 2 large glands near the base; staminodia pres- ent, club-shaped, ca. 0.8 mm long. Ovary globose, glabrous; style much shorter than ovary, glabrous. Fruit broadly ellipsoid, 2 x 1.5 cm, seated on a small, platelike cu- pule; pedicel swollen in fruit. Volume 75, Number 2 1988 van der Werff New Neotropical Lauraceae FicunE 4. Licaria velutina. — A. Flow tepals removed, show of twig, showing pubescence.—G. Det wering branc Paratypes. NICARAGUA. ESTELI: Laguna Miraflor, Laguna 336 (MO); same location, Moreno 19434, 21118, 8227 (MO); Cerro Quiabü, Stevens 16918, 16247 (MO); same locality, Grijalva & Araquistain 641 (MO); same locality, Moreno 1309, 19266, 21185A (MO); Mesas Plan Helado, 2 km from Laguna Miraflor, Moreno 15846 (MO); El Chaparral, 1 km W of Laguna Miraflor, Moreno 22382 (MO); Mesas Plan Helado, 21.5 km E of Esteli, h.—B. Fru wing three two- pues stamens with —C. Flowers. —D. Ovary.—E. Flower with said ipm fused Viam and lobe-shaped basal glands. —F. Detail il of inflorescence branch, showing pubescence. Moreno 15410 (MO). JINOTEGA: km 150 de la carretera atagalpa-Jinotega, Moreno 472 (MO); Laguna de Mir- aflores, Neill 339 (MO) (distributed by Seymour as Neill 1968). Nectandra mirafloris is only known from an area with cloud forest on the border of the 412 Annals of the Missouri Botanical Garden f "wh d fae? wA r peavey Sig IGURE 5. Nectandra mirafloris.— A. Flowering branch.—B. Flowers.—C. Dehisced tepals with attached stamens.—D. Young fruit, showing abscission line of tepals.—E. Fruit. Volume 75, Number 2 1988 van der Werff 413 New Neotropical Lauraceae departamentos of Esteli and Jinotega. This is one of the best-developed montane areas in Nicaragua and is inhabited by several species otherwise unknown in Nicaragua, such as Os- munda regalis (W . D. Stevens, pers. comm.). Specimens with buds or flowers have been collected from late December to May; the fruiting collection (Stevens 16347) was made ovember. Nectandra and Ocotea are the two largest genera of neotropical Lauraceae. The differ- ences between the two genera are not always easy to see and some authors have proposed to merge them under Ocotea (Kostermans, 1957; Howard, 1981; Liogier, 1982; the last-mentioned two transferred several West Indian Nectandra species to Ocotea). The character most frequently mentioned in the literature as separating the two genera is the position of the anther cells; they are arranged in an arc in /Vectandra and in two rows in Ocotea. This character separates most species quite readily but is intermediate in some. Two additional characters help separate Nectan- dra from Ocotea. In Nectandra the inner faces of the tepals and the anthers have pa- pillose pubescence; in Ocotea these surfaces are either glabrous or strigose. Also, in Nec- tandra the tepals are usually basally conni- vent, and in older flowers an abscission line forms underneath the tepals, which fall off as a unit together with all anthers. In Ocotea the tepals are free and fall off individually, often leaving stamens attached to the floral tube. Thus, on the young cupules of Nectan- dra species, one never finds stamens, but in Ocotea very frequently a few stamens can be found on young cupules. In Nectandra mirafloris the position of the anther cells is intermediate between Nec- tandra and Ocotea, but the papillose indu- mentum on the tepals and anthers and the dehiscence of the tepals as a unit lead me to place this species in Nectandra. I regard a montane species from Panama and Costa Rica, Nectandra cufodontisii (O. Schmidt) Allen (basionym: Ocotea cufodontisii O. Schmidt; heterotypic synonym: Ocotea seibertii Allen according to W. Burger in litt.), as its closest relative. This species is very similar in leaf shape, venation type, and fruit shape but dif- fers by lacking white pubescence on flowers and buds; its flowers are slightly larger, as are the leaves; and the outer stamens have anthers on filaments ca. 0.3 mm long, con- trasted with nearly sessile anthers in V. mir- afloris. Several collections of N. mirafloris were annotated as Persea, one as Nectandra sanguinea and one (the type) as Anacardium occidentale. The epithet mirafloris refers to the type locality and is a reminder to collectors to look out for flowering Lauraceae. Persea pajonalis van der Werff, sp. nov. TYPE: Peru. Boundary of provinces Oxa- pampa and Pasco: San Gotardo; 3 m shrub in pajonal, 2,500-3,000 m, 29 Dec. 1983, Foster, Chanco & Alban 7647 (holotype, MO). Figure 6A-E. Frutex vel arbor parva. Ramuli c 5-8 mm diametro, angulares, hornotini sparsim rumes cinereo- ubescentes, vetustiores glabri; pag conspicuis ag- gregatis. Folia alterna, coriacea, obovata, p os rotundata vel subacuta, basi obtusa vel RE T -17 x 4-8 cm; super glabra costa ie (10-12 jugis) i = subtus sparsim adpresse angen praecipue secus co tam iia a costa, nervis venationeque elevata. Pei crassi, 5-8 m ongi. Inflorescentiae axillares, = 1€ ongae, puberuli gen bescentia evanescente versus basim M ias nici. Te- pala 6, 08646 ualia, 3 exteriora ovata, 3 x 3 mm, 3 in- 4.5 x berul tamina mm, 9, 4- locellata, 6 exteriorad liona, pot longa; 3 interiora xtrorsa, 3.5 mm longa, basi ee 2 glandulis sessilibus praedita. Staminodia 3, 1.5 mm longa, pubes- centia. Ovarium glabrum Ea wauas 2 mm longum. Fructus immaturus viridis, magnopere tepalis persisten- tibus obtect c Shrub or smali tree, 2-6 m tall. Twigs thick, 5-8 mm diam. during first year, an- gled, hollow, when young with gray appressed pubescence, glabrescent, 2-year-old twigs gla- rous. Terminal buds with appressed gray pubescence, usually hidden by the leaves. Twigs with conspicuous clusters of scars from bracts of old terminal buds. Leaves alternate, evenly distributed along twigs, coriaceous, ob- ovate, the tip rounded or slightly Eo the base obtuse or subcordate, 7 4 cm; laminae glabrous above, Dou ap- Annals of the Missouri Botanical Garden Ro — > = A <” e A yt ‘fe, A, ⁄ N - y T (j A Jija FIGURE 6. A-E. Persea pajonalis.— A. Fi Flowers.—D. Flo Leaf of Persea sessilis. pressed pubescent below, especially along ma- jor veins; midrib and lateral veins (10-12 pairs) immersed above, prominently raised on lower surface, tertiary venation also raised on lower surface. Petioles thick, 5-8 mm long. Inflorescences in axils of persistent leaves, 5- 18 cm long, shorter than or about as long as the subtending leaves; paniculately branched, the peduncle to 12 cm long; flowers and ped- icels with dense, bronze-colored puberulence, this becoming much sparser toward the base lowering branch.—B. Detail of twig with scars from bracts. —C. wer with tepals removed, showing stamens, two basal glands, and staminode.—E. Fruit. —F. of the inflorescence. Tepals 6, unequal, the outer 3 ovate, ca. 3 mm, the inner 3 ellipsoid, ca. 4.5 x 2.5 mm, bronze-puber- ulent outside. Stamens 9, all 4-celled, the outer 6 introrse, ca. 3 mm long, the filaments ca. 1.5 mm long, narrower than the anther; the anther cells arranged in 2 rows; inner 3 stamens extrorse, ca. 3.5 mm long, with 2 globose glands attached near the base of the filaments. Staminodia 3, ca. 1.5 mm long, pubescent, with a broad triangular head. Ovary Volume 75, Number 2 1988 van der Werff 415 New Neotropical Lauraceae glabrous, round or slightly obovate, ca. 2 mm long; style distinct, ca. ong. Im- mature fruits green, largely hidden by per- sistent tepals. Paratypes. PERU. BOUNDARY OF OXAPAMPA AND PASCO: San errr in dwarf forest, 2,650-2,800 m, Gentry et al. 39998 (MO); same locality, van der Werff et al. 8578 (MO). Persea pajonalis is only known from a few collections from the San Gotardo area west of Oxapampa at rather high elevations. It is restricted to a vegetation type called pajonal, a name used for open, nonforest vegetation. Some of the pajonales seem to have an edaph- ic origin (only found on nutrient-poor sand- stone), but in the San Gotardo area the pajonal represents a subparamo scrub rich in Erica- ceae and Myrsinaceae; other taxa frequently found in high-elevation vegetation (Araliace- ae, Jamesonia, Weinmannia) were present as well. However, groups indicative of nu- trient-poor soil, such as Eriocaulaceae, ter- restrial Utricularia, and Pinguicula, were found in the area too. Persea pajonalis belongs to subg. Erio- daphne sect. Eriodaphne, using Kopp’s (1966) classification. Among the South Amer- ican species of this subgenus, Persea pajona- lis can be immediately recognized by its near- ly sessile leaves with rounded or subcordate base. Persea sessilis Standley & Steyerm., a Guatemalan species only known from the fruiting type collection, has similar leaves (Fig. 6F) The holotype (F!) consists of good veg- etative material and remnants of infructes- cences. The following differences are evident between P. pajonalis and P. sessilis: in P. pajonalis the leaves have scarce scattered hairs on the lower surface, 10-12 pairs of lateral veins, and rounded or slightly acute tips, and the twigs are hollow; in P. sessilis the leaves have glabrous lower surfaces, acute to acuminate tips, and 15 or more pairs of lateral veins, and the twigs are solid. I expect that when flowers of P. sessilis become avail- able, additional differences will be found and that the striking leaf shape is more a habitat adaptation (both are shrubs occurring on high- elevation mountain ridges) than an indication of close relationship. Phoebe elegans van der Werff, sp. nov. TYPE: Mexico. Oaxaca: Mpio. San Miguel Chimalapa, Cima del Cerro Salomón, 16?46'15"N, 94211" 45"W, 1,770 m, 3 m tree, flowers green with red margin, pedicels reddish. Abundant. 11 Apr. 1986 (fl, fr) M. Ishiki 1501 (holotype, MO; isotypes, CHAPA, HBG, LL, MEXU, n.v.). Figure 7. Arbor parva, ad 6 m. Ramuli tenues, teretes, glabri; emma terminalis s vel aliquot pilis adpressis aucta. Petioli glabri, 1-2 cm longi, leviter canaliculati. Folia alterna, glabra, Mont 5-9 x 2-3 cm, basi obtusa vel rotundata, apice valde acuminata; acumine ad 2 cm lo ves, ceteris nervis lateralibus (3-4 jugis) waq nqa 6, aequalia, per an nthesin erecta, ca. 1. m la ata, apice pese y Stamina 9, ongo. Fructus ellipsoideus, ca. ] cm longus, 0.7 cm latus; cupula parva, non profunda, sensim in pedicello attenuata tepalis persistentibus. Small tree, 3-6 m tall, the main stem thin, ca. 1 cm diam., branches smooth, delicate, horizontal. Twigs slender, terete, glabrous, the terminal bud with few appressed hairs or glabrous. Petioles glabrous, 1-2 cm long, the margins of the laminae decurrent as narrow ridges and the petioles thus slightly canalicu- late. Leaves alternate, glabrous, ovate, mostly 5-9 x 2-3 cm, the base obtuse or rounded, the apex strongly acuminate, the acumen to 2 cm long, often falcate; laminae tripliveined, the basal laterals leaving the midvein at or very near the base of the laminae; other lat- erals (3-4 pairs) weakly developed; upper leaf surface with immersed venation, the lower surface with slightly raised venation; numer- ous small oil cells present on lower surface; margins of laminae smooth and slightly thick- ened. Inflorescences in the axils of normal leaves, glabrous, 2.5—4 cm long, shorter than the leaves, once or twice cymosely branched, 416 Annals of the Missouri Botanical Garden FIGURE 7. cence.—C. basal er ig b o elegans. —JD. Flower. —E. Ovar but on each inflorescence usually 2 or 3 flow- ers; bracts present in young inflorescences, linear, ca. 1.5 mm long, deciduous with age; pedicels slender, to 1 cm long. Flowers gla- brous, ca. 2 mm long, 3.5 mm wide. Tepals 6, equal, erect at anthesis, rounded at tip, ca. 1.5 mm long an mm wide. Fertile stamens 9, all 4-celled, glabrous, the outer 6 ca. 1.2 mm long, the filament slightly wider than the anther, the locelli arranged in 2 rows, —A. “watas branch.—B. Detail of twig showing petiole and base of inflores- —F. Outer stamens.—G. Staminode. —H. Inn ner stamens with small introrse; the inner 3 stamens with extrorse cells, the filaments wider than the anthers, ca. 1.5 mm long; staminal glands small and visible as two small bulges at the base of the inner anthers. Staminodia 3, ca. 1 mm long, with a triangular head. Ovary globose, gla- brous, ca. 1 mm diam. Style distinct, ca. 0.8 mm long with a large stigma. Base of the floral tube and base of the stamens covered with stiff, translucent hairs. Fruit an ellipsoid Volume 75, Number 2 1988 van der Werff 417 New Neotropical Lauraceae berry, ca. 1 cm long, 0.7 cm wide; cupule small plate, gradually narrowed into the ped- icel and crowned with persistent tepals. Paratypes. MEXICO. OAXACA: Mpio. San Miguel Chi malapa, Cima del Cerro Salomón, (buds) /shiki 1454, (fr) 1529 (MO, CHAPA), (fl) Ishiki 1616 (CHAPA, MEXU, MO). Phoebe elegans is a very delicate and at- tractive species, so far only known from one mountain in Oaxaca, Mexico, near the border with Chiapas; it is said to be abundant in elfin forest and transition into cloud forest. The relatively long, spreading petioles, and the ovate, long-acuminate, triveined leaves sep- arate it at once from the other neotropical Phoebe species, although it possesses all di- agnostic characters for the group: tepals erect at anthesis, tepals persistent in fruit, flowers with long pedicels, staminodia present, and tripliveined leaves. Phoebe, as accepted here, is a large (at least 180 binomials) genus occurring in the Asian and American tropics. Agreement on the generic boundaries of Phoebe has not yet been reached, and it is not clear how and if Phoebe can be separated from Cinnamomum and Ocotea (Kostermans, 1961; van der Werff, 1987). Until a careful study of these three groups has been made, I will continue the traditional usage of Phoebe for a heter- ogenous group of species in the Neotropics. I realize that the discordant species have to be transferred to other genera, and earlier (van der Werff, 1987) I discussed the char- acters I consider diagnostic for neotropical Phoebe. Altogether, the neotropical Phoebe species form a large group to which more than 70 species have been attributed. Pleurothyrium hexaglandulosum van er Werff, sp. nov. TYPE: Panama. Co- lon: Rio Guanche, ca. 5 km upstream from Portobelo, 50 m, tree, 5 m, inflo- rescence pendent, flowers green, becom- ing yellow. Hammel & Trainer 14781 (holotype, MO; isotype, BR, others to be distributed). Figure 8. Arbor parva, 5 m. Folia alterna, chartacea; ge dense tomentelli; laminae anguste obovatae, 30-45 10-15 cm, versus basim sensim attenuatae, basi eae rotundatae vel subcordatae, apice acuminatae; super ve- illares, puberulae, paniculatae, 40-65 cm lo celli cinereo-pubescentes, 1.5-2 cm longi. Flores virides, 8-9 mm E ata, Ó exteriora bedi lateralibus, 3 in- teriora eee extrorso-lateralibus. Filamenta staminum interiorum andulis magnis, stamina exteriora cingen- tibus aucta. P esum late globosum, pubescens. Fructus us. Small tree, 5 m tall. Twigs terete, densely brown tomentellous, 5 cm below the tip 4-5 m diam. Leaves alternate, chartaceous; the petioles ca. 5 mm long, densely tomentellous; laminae narrowly obovate, 30-45 x 10-15 cm, gradually narrowed toward the base, there abruptly rounded to subcordate, the tip acu- minate; glabrous above, with appressed hairs on main veins below, especially near the base, otherwise glabrous; venation immersed above, midvein prominently raised below, secondary veins (14-18 pairs) raised and the final re- ticulation slightly less raised; secondary veins arching upward near the margin and promi- nently loop-connected in the upper 25 of the lamina. Inflorescences axillary, 40-65 cm long, brown puberulous, paniculate, laxly branched, the basal lateral branches 20-25 m long, the upper ones gradually shorter. Pedic els 1.5-2 cm long, densely gray-pubes- cent. Flowers ca. 8-9 mm diam., greenish becoming yellow (fide collector). Tepals 6, equal, 4 mm long, glabrous inside except where in bud the glands and stamens not pressed against tepals, thus showing a narrow line of hairs central in the lower part of the tepal (the 6 glands not completely fused in this species), this line expanded in a diamond- shaped outline marking the space between anther and glands and with lines to the margin and tip of the tepal. Glands of the inner 3 stamens prominent, surrounding the outer stamens, but not becoming fused. Stamens 9, raised above the glands, the filament with some hairs on the back, 0.8-1 mm long; anthers 4-celled, the outer 6 anthers curved 418 Annals of the Missouri Botanical Garden IGURE 8. G Flower, showing pubescence pattern n tepals, : A Y — UN MESES, Ed 1.5mm Nun. Pleurothyrium Ba quis und — A. Flowering branch.—B. Detail of leaf base and petiole. —C. large staminal grands, and stamens.—D. Old flower, glands and stamens fallen of:—E. Outer (left) and inner (right) stamen inward, the anther cells lateral; inner anthers shorter, the anther cells lateral. No stami- nodia seen. Ovary broadly ovate, 1 mm long, mm wide, with short, gray pubescence. Style short, 0.2-0.3 mm long, gray pubes- cent. Stigma platelike. Glands and stamens deciduous in older flowers and the tepals be- coming reflexed, thus fully exposing the ovary. Fruit unknown. Pleurothyrium hexaglandulosum is only known from two collections. It is closely re- lated to P. maximum O. Schmidt from Am- azonian Peru. Shared characters are large Volume 75, Number 2 van der Werff 419 1988 New Neotropical Lauraceae inflorescences, oblanceolate-obovate leaves Allen, Mem. New York Bot. Garden 23: with thick, short petioles, and loop-connected secondary veins, forming a submarginal vein. It differs in having gray or brown pubescence (rufous in P. maximum), much smaller flow- ers with dense gray pubescence (rufous in P. maximum), and more widely branched inflo- rescence (lateral branches ca. 2 cm long in P. maximum, to cm long in P. hexa- glandulosum). Schmidt (1933) noted a re- lationship between P. maximum and P. wil- liamsii O. Schmidt. I have not seen material of the latter species; it differs from P. hex- aglandulosum, according to its description, in the shorter inflorescences (to 12 cm long), shorter pedicels (4-6 mm long), and brown- tomentose flowers. Pleurothyrium hexaglandulosum is the first record of Pleurothyrium in Panama. Its specific epithet refers to the six glands of the inner three stamens. In nearly all Pleurothy- rium species these are fused and cannot be recognized individually, and for a long time it was assumed that in Pleurothyrium all nine stamens had two glands. In Pleurothyrium hexaglandulosum (and to a lesser degree in P. maximum), the glands remain separated and show clearly that in Pleurothyrium only the inner stamens have glands, as Rohwer & Kubitzki (1985) stated. Croat & Grayum 59792 (F, MO), col- lected in Costa Rica, Puntarenas, along road between Rincón de Osa and Rancho Que- mado, is provisionally placed here. It differs from the type collection in being less pubes- cent. Ocotea erectifolia (Allen) van der Werff, comb. nov. BASIONYM: Phoebe erectifolia 860. 1972. TYPE: Venezuela. Bolivar: Meseta del Jaua, Steyermark 97926 (holotype, NY). Ocotea budowskiana Bernardi, bu eda 30: 256. 1975. : Venezuela. Bolivar: Meseta de Jaua, Stey- ermark 109330 (bolotype, G!; patea F!). Ocotea erectifolia is very distinct with co- riaceous, few-veined, ascending leaves. It was first published as a Phoebe species. When Bernardi later recognized it as an undescribed Ocotea species, he overlooked the earlier de- scription of Phoebe erectifolia. However, there is no doubt that the two species are the same, and hence the new combination in Oco- tea is made. LITERATURE CITED Howarp, R. A. . Nomenclatural notes on the Lauraceae of the Lesser Antilles. J. Arnold Arbor. -61. KoPP, L. E. 1966. A taxonomic revision of the genus Persea in the Western Hemisphere. Mem. New York Bot. Gard. 14: 1-120. KOSTERMANS, A. J. G. H. 1957. Lauraceae. Reinwardtia 193-256. 1961. The New World FE af oo mum Trew (Lauraceae). Reinwardtia 6: Liocier, A. H. 1982. M nip. a IX. Phyto logia 50: 161-170. ROHWwER, J. & K. KuBITZKI. 1985. p bae grues im ped Komplex (Lauraceae). Bot. Jahrb. Sys 107: 35. SCHMIDT, O C. dinen südamerikanischen Lauraceen I. Repert. 9. 1933. Beitrage zur Kenntnis der an Sock: 4. Notes on neotropical Lau- . Ann. Missouri Bot. Gard. 71: 1180-1183. Et Six new species of neotropical Laura- ceae. Ann. sg = Gard. 74: 401-412 H. G. Ric 1985. Caryodaphnopsis Airy-Shaw aaa a ea new to the Neo tropics. Syst. Bot. 10: 166- A REVISION OF PANICUM SUBGENUS PHANOPYRUM SECTION STOLONIFER A (POACEAE: PANICEAE)! Fernando O. Zuloaga? and Tatiana Sendulsky? ABSTRACT A revision of Panicum subgenus Phanopyrum section Stolonifera e presented. Panicum soderstromii is described as ne ndreanum, P. biglandulare, P. crater which stipitate at its base a and iferum, P. ir P. ae P. latissimum, P. pulchellum, and the REE P. venezuelae are included in this section, can be characterized by the inflorescences with unilateral racemose branches, the upper anthecium short and glabrous and smooth, and the presence of non-Kranz leaf anatomy. The presence of one egulare, P. chapadense, P. rude, P. piauiense, or two (rarely a) pairs of crateriform, ocellate glands is a singular feature of this section; these glands may be constantly present, present or absent i me specimens, or completely absent in others. Keys to all 13 species and SEM micrographs of the upper lia and glands on the lower lemma are provided. Each species is ‘lustra ted. Section Stolonifera is one of the most at- tractive and interesting sections of the genus Panicum. The name was given as an informal group by Hitchcock & Chase (1910), who included in it P. stoloniferum Poiret, P. fron- descens G. Meyer, P. pulchellum Raddi, and P. biglandulare Scribner & Smith. These species were mainly distinguished in the view of Hitchcock & Chase as summarized in the following diagnosis: Decumbent or creeping perennials, rooting at the lower nodes, with branching culms; ligule short, membranous; leaf blades lanceo- late or ovate-lanceolate; panicles of racemose, secund, spikelike, ascending branches, + di- vergent from the axis, with spikelets in pairs along one side of the branches; upper glume and lower lemma exceeding the anthecium in engt In 1940, Pilger gave the rank of section to species of the Stolonifera group. Hsu (1965), in his worldwide treatment of Pani- cum, characterized this section basically as having a papery ligule, panicles with spikelike racemes, upper glume and lower lemma lon- ger than the upper anthecium, the latter smooth, lodicules very thin and weakly three- -nerved. Hsu placed this section in his sub- genus Sarmentosum, along with, among oth- ers, sections Sarmentosa Pilger, Parviglumia (A. Hitchc. & Chase) Pilger, and Parvifolia (A. Hitchc. & Chase) Pilger. In the present work, Stolonifera is treated, according to the infrageneric classification of ! The first author E the Consejo Nacional de Investigaciones Científicas y Técnicas de la Repüblica Argentina (CONICE Institution, Washington T) for a grant that allowed initiation of the present studies (1982-1983) at the Smithsonian iunior author thanks the Fundação de Amparo a Pesquisa do Estado de São Paulo and the Conic Nacional de Desenvolvimento Tecnologico e Cientifico for the grants that have made part of this work possible. Both authors express gratitude to Dr. Lyman B. Smith for provision of facilities during their several stays in ashington and for critical review of the manuscript; to the late Dr. Thomas R. Soderstrom for the opportunity to study in the Herbarium and Laboratory of the Grass Section of the Smithsonian Institution, and for attention, help, and suggestions: and to Alasdair G. Burman and Emmet Judziewicz for comments on the manuscript. This pap based on material examined from the following herbaria: B, BAA, CEPEC, COL, CTES, F, GH, HRB, LIL, MO. MSC, P, R, RB, SL, SP, UB, US, and VEN. 2 Instituto de Botánica Darwinion, Casilla de Correo 22, San Isidro, Argentina. 3 Instituto de Botánica, Caixa Postal 4005, Sáo Paulo, Brazil. ANN. MISSOURI Bor. GARD. 75: 420-455. 1988. Volume 75, Number 2 1988 Zuloaga & Sendulsky Revision of Panicum sect. Stolonifera Zuloaga (1987) within subg. Phanopyrum (Raf.) Pilger, being most closely related to sect. Phanopyrum and Laxa. Section Phan- opyrum Raf. was raised to generic rank by Nash (1903), taking into consideration as di- agnostic characters spikelet compression, habitat, length of the upper anthecium, and presence of a stipe in the base of the upper anthecium. This idea was accepted in recent years by various authors (Brown, 1977; Gould & Shaw, 1983; Lazarides & Webster, 1984). However, we believe that none of these char- acters are strong enough to maintain Phan- opyrum apart from Panicum. As Hitchcock & Chase (1910) stated: “this species [P. gymnocarpon ] departs somewhat from the usual characters of the genus Panicum, but the divergence does not seem sufficient to justify segregating the single species as the type of a separate genus." Subgenus Phanopyrumincludesnon-Kranz species, these being anatomically distin- guished by having two bundle sheaths around the vascular bundles, the inner mestome sheath with thick-walled cells surrounded by a parenchymatous sheath with thick-walled and completely empty cells (with these par- enchymatous sheath cells bigger than the me- sophyll cells). The number of mesophyll cells between the vascular bundles ranges from 5 to 12, with a distance of (150)230-270 (-380) um. Exomorphologically, the spikelet is characterized by having a lower glume l- 3 (occasionally —5)-nerved and upper glume and lower lemma 5(occasionally —7)-nerved. The species are usually found in shady and humid places, with some exceptions in species of sect. Lorea, Parvifolia, and Stolonifera, some of which grow in open and more or less drier habitats. Within this subgenus, sect. Stolonifera is defined by the following diagnostic features: Inflorescence type. All species are char- acterized by having the spikelets congested on short, unequal pedicels along racemose, unilateral branches (occasionally with short tertiary branchlets in P. latissimum Mikan ex Trin., P. rude Nees, and P. venezuelae Hackel), with the branches usually alternate, remote, and diverging from the axis. This character is also present in sect. Phanopy- rum and sect. Laxa (A. Hitchc. & Chase) Pilger. In the former, the spikelets are usually borne on tertiary branchlets appressed uni- laterally along secondary branches. Within sect. Laxa there is a variation from short- pedicelled spikelets in unilateral racemose branches (as for example in P. pilosum Sw. and P. leptachne Doell) to spikelets disposed along either side of short tertiary branchlets (e.g., P. boliviense Hackel, P. bresolinii L. B. Smith & Wassh.). Sections Parvifolia, Monticola Stapf, and Verrucosa (A. Hitchc. & Chase) C. C. Hsu, included in subg. Phan- opyrum, are characterized by open and lax panicles with long-pedicelled spikelets; in sect. Lorea Zuloaga some species have open and lax panicles, whereas others have spikelike inflorescences. In sect. Megista Pilger, all the branches of the panicle are whorled, with the spikelets short-pedicelled along the branches. In sect. Parviglumia, there is a gradation from species with spikelets arranged in open and diffuse panicles to others with short-ped- icelled spikelets along racemose, not unilat- eral branches. Surface and ornamentation of the upper anthecium. The epidermis of the upper an- thecium is completely glabrous, indurate, smooth, and shiny, with scattered stomata toward the upper margins of the palea (Fig. 2c, d). A similar pattern is present within the subgenus in sect. Phanopyrum and sect. Lo- rea (in the latter it is possible to find occasional prickle hairs, stomata, and bicellular micro- hairs toward the apex of lemma and palea). On the other hand, sect. Laxa differs by having conspicuous prickle hairs toward the apex as well as simple papillae regularly dis- tributed in longitudinal rows. Membranous anthecia are present in some species of this section. Stipe presence, type of spikelet, and rel- ative length of the upper anthecium. In all of the species of the section, the spikelets are biconvex with the upper anthecium short- 422 Annals of the Missouri Botanical Garden ly stipitate at the base; additionally, the an- thecium is reduced in relation to the length of the upper glume and lower lemma (usually 1⁄4—3⁄4 the length of the upper glume and lower lemma). KEY TO THE SECTIONS OF Sunc. PHANOPYRUM la. Upper glume and lower lemma 7-9-nerved; panicles with all the branches whorled Sections of subg. Phanopyrum can be sep- arated by the following key: ect. Megista lb. Upper glume and lower lemma 3-5(occasionally -7)nerved; with the branches alternate to opposite, not sect. Lorea whorled. 2a. Leaf branches stiff, pungen 2b. Leaf blades not ungent, not s long, cylindric hairs (occasionally glabrous, when a lower elet) t. Parviglumia 3b. Upper anthecium without long, cylindric hairs all over its surface (if glabrous lower saba present, and lower glume more than 4 the length of the spikelet). 4a. Panicles with racemose, unilateral branches, the spikelets disposed in pairs on short pedicels along one side of the branches, sometimes with the spikelets short-pedicelled in short tertiary branchlets on both a of the branches; upper anthecium smooth, glabrous or with short prickle hairs at the summit. 9a. Upper indem glabrous, smooth and shiny, indurate, shortly stipitate, and shorter than the upper glume and lower lemma. 6a. Grasses aquatic, culms succulent; lower glume Y the length of bn de upper anthecium Y3 the length of the spikelet. Southeastern United Sta sect. Phanopyrum 6b. Grasses ee forests or edges of forests or in open habitats, culms not succulent; e 14-34 the length of the spikelet; upper anthecium 14-34 e bg i of the iie Mesoamerica to South Am erica t. Stolonifera Sb. Upper anthecium papillose, with simple papillae regularly distributed in dinil rows and prickle hairs toward the apex x of lemma and indurate to membranous, almost aching the same length of the upper glume and ber lemma, not stipitate ..... sect. Laxa > = Panicles lax and diffuse, the spikelets long- to short-pedicelle d, not in unilateral, racemose branches; a anthecium pilose, with bottlelike bicellular microhairs all over the surface, h rugose to Ta. Upper aa smooth and shining sect. Parvifolia 7b. Upper anthecium rugose, with transverse or longitudinal and transverse wrinkles. 8a. Upper anthecium with transverse wrinkles. Lower glume Y, the length of the spikelet; upper glume and lower le mma verrucose. Eastern United S sect. oo 8b. Upper anthecium with longitudinal and transverse wrinkles. Lower glume 4-4 t length of = spikelet; upper glume and lower lemma not verrucose. Mesoamerica to Argentin ect. Monticola A singular and isolated, although not con- stant, feature of species of sect. Stolonifera is the presence of one or two (rarely three) pairs of crateriform and ocellate glands on either side of the midnerve on the outer sur- face of the upper lemma. These glands are always present in P. pulchellum, P. biglan- dulare (Fig. la, b), P. soderstromii Zuloaga & Sendulsky (Fig. 1c-f), and P. crateriferum Sohns, while in P. rude, P. piauiense Swallen, . chapadense Swallen, and P. venezuelae they are sporadic; they are completely absent in P. stoloniferum, P. andreanum Mez, P. brachystachyum Trin., P. latissimum, and P. irregulare Swallen. Occasionally, one pair of glands is present on the upper glume of spikelets of P. venezuelae; in this species the glands differ from those of the other species by being slightly depressed rather than cra- teriform (Fig. 2a, anicum venezuelae has cleistogamous flowers in most of its spike- lets, this character showing up elsewhere in Panicum only in subg. Dichanthelium. Pan- icum irregulare is the only species within the genus with the lower flower hermaphrodite. The pubescence of the spikelet may vary from pilose to hispid (as in P. chapadense, P. andreanum), papillose-pilose (in P. bra- chystachyum), or entirely glabrous (in P. ir- regulare and P. stoloniferum). Volume 75, Number 2 1988 Zuloaga & Sendulsk 423 y Revision of Panicum sect. Stolonifera dons l. lemma.— b. De ig of glands on lower lemma. c—f. P. soder rs of glands, lateral view.—f. Det c-f based on Pereira 2138 (US). Meenifcations: a, X75; b, x 175; c, X35; d, —e. Spikelet with two ss on m r lem baled o on Matuda A 4258 x 200; e, X35; f, x Swallen (1966) included eight species in the group Latissima of Panicum, but he di not delimit it or indicate its links with other species of Panicum. The species accepted as valid from the Latissima group (P. rude, P. latissimum, and P. piauiense, together with chapadense and P. soderstromii) are treated in the present work in sect. Stolonif- era, since we regard the habit and size of the plants as insufficiently strong characters to keep them in a different section. It should be pointed out that further anatomical studies would be useful to establish the relationship between species of the Latissima group and the rest of species of sect. Stolonifera. Species of sect. Stolonifera occur from Mexico to Árgentina. Some are widely dis- Scanning electron ied of spikelets of Panicum species. a, b. P. biglandulare.—a. Lower c. Spikelet, ventral view.—d. Detail of gland ail of gland on lower lemma. a, b stromii. — tributed while others are found only in re- stricted areas. To the former group belong P. stoloniferum and P. pulchellum, the first ranging from Mexico, the Lesser Antilles, and South America to Argentina; P. pulchellum ranges from Mexico to Brazil and Bolivia. Two species are confined to Mesoamerica; P. bi- glandulare in Mexico and Guatemala and P. crateriferum in Mexico. Panicum irregulare grows from Costa Rica to Colombia and P. andreanum in Colombia and Venezuela, while P. venezuelae is found from Mexico to Brazil. The other species are all endemic to Brazil: P. brachystachyum from Minas Gerais, P. rude from Espirito Santo to Rio Grande do Sul, P. latissimum from Espirito Santo and Rio de Janeiro, P. chapadense from Goias, Annals of the Missouri Botanical Garden 1 N D iMt t M N ; ` LU à N A » ES TM y "li Illy 4 ee e a - er — — d FIGURE 2. Scanning electron micrographs of Panicum species. a, b. P. venezuelae.— a. side.—b. Detail of gland on lower lemma. c, d. P. pulchellum.— c. Apex of the upper anthecium, ventral side. — d. Base of the upper anthecium showing stipe. a, B g on Pinto 307 (US); c, d based on Hitchcock 20536 (US). Magnifications: a, x 35; b, x 100; c, x 100; , Spikelet, ventral P. piauiense from Piaui and Bahia, and P. were made using the scanning electron mi- soderstromii from Bahia. croscope. Anthecia were removed from dried Classical taxonomic methods have been ap- herbarium specimens, secured on stubs, car- plied utilizing a Wild M5 dissecting micro- bon coated in a vacuum evaporator, coated scope. Observations at higher magnification with a gold—palladium alloy, and examined in Volume 75, Number 2 1988 Zuloaga & Sendulsky 425 Revision of Panicum sect. Stolonifera a Cambridge S4-10 or Cambridge Stereoscan 250 Mk2 scanning electron microscope op- erating at 10-20 kV. TAXONOMIC TREATMENT Panicum section Stolonifera A. Hitchc. & Chase ex Pilger, in Engler, Nat. Pflan- zenfam. ed. 2, 14e: 16. 1940. TYPE: P. stoloniferum Poiret. Group ignia A. Hitchc. & Chase, Contr. U.S. Natl. erb. 120. 1910 us "ue Contr. U.S. Natl. m 17: 461, 500. Perennial, plants small to robust, stolonif- erous or decumbent, rooting and branching at the lower nodes to erect, leaning in the vegetation or not leaning; culms hollow, rare- y solid, simple or branching. Ligule mem- branous-ciliate to membranous. Leaf blades ovate-lanceolate to long-lanceolate, flat, densely pilose to glabrous, shortly pseudo- petiolate. Panicles pyramidal, sometimes ob- long, composed of few to numerous, unilat- eral, racemose branches arranged along either side of the axis and bearing secund, paired, short-pedicellate spikelets. Spikelets ellipsoid KEY TO THE SPECIES OF SECTION STOLONIFERA to lanceolate, pilose to glabrous. Lower glume 3(rarely -5)-nerved, !4—?4 the length of the spikelet, pilose toward the apex or glabrous. Upper glume and lower lemma subequal (or upper glume shorter than the lemma), acute to acuminate, 5(occasionally —7)-nerved; low- er lemma with or without 1 or 2 (occasionally 3) pairs of crateriform, ocellate glands on the middle portion. Lower palea hyaline, gla- brous; male flower present or absent (her- maphrodite flower present in P. irregulare). Upper anthecium ellipsoid to lanceolate, gla- brous, smooth and shiny, indurate, shortly stipitate at the base; upper lemma with the margins inrolled over the palea. Lodicules 3-nerved. Caryopsis with punctiform or ovate hilum, the embryo less than half the length of the caryopsis. An American section, including 13 species distributed from Mexico to Argentina, com- monly found in forests or at the margins of forests, occasionally on “campos rupestres” P. brachystachyum) and on “cerrados” ( chapadense, P. soderstromii, P. piauiense). Collections come from 0-2,800 m elevation. — la. Lower — hermaphrodite; caryopsis free from the lemma and palea, similar to the caryopsis in the upper 6. P anthec . irregulare lb. Lower puis male or neuter, not hermaphrodite. 2a. Spikelets 4.9-5.2 mm long 3. P. brachystachyum 2b. gr 1.8-3.8(-4) mm long. Leaf blades amplexicaul, 7-12 cm me 7. P. latissimum ds Leaf blades not amplexicaul, 0.3-5 c 4a. Spikelets mostly with don ad Lb bs lemma gibbous with the upper margins inrolled; upper glume with or without p ers; anthers of these flowers 0.2-0.3 mm long; lower . P. venezuelae a oo a "10 45 c J c ernodes . Spikelets with chasmogamous flowers only; anthers 1-2 mm long; lower a not gibbous, the upper margins not inrolled; upper glume always without glands i land 12. P. stoloniferum ands. e 1⁄4—3⁄4 the length of the ia can culms robust and erect; leaf blades m long; Brazil, from Bahia to 7a. Lower glume with long, papillose-pilose hairs toward the iae? pedicels an branches with long hairs exceeding the s e . Lower glume shortly pilose to nearly glabrous, without long, Pario -pilose hairs toward the apex; pedice 8a. bo with long, creeping ee each erect culm orange and cor 8b. Plants short-rhizomatous, MUT the lowest internodes orange xr Rio Grande . soderstromii length of the spikelet .. s and branches short-pilose to scabrou with na lowest a Qa. Plants with thickened, fusiform roots; culms branched; panicles 7-22 ml 8. m long; Brazil, Bahia, Piaui P. piauiense 426 Annals of the Missouri Botanical Garden Ob. Plants with thin roots; culms simple or rarely branched; panicles 20- m long; Brazil, Espirito Santo to Rio Grande do Sul ....... 10 P. rude 6b. Lower glume 4-1 the length of the spikelet; culms extensively creeping and rooting at the lower nodes; leaf blades ovate-lanceolate to lanceolate, 2.5-10 Colombia and Mower (P. pulchellum in Brazil species with m long, spikelet Mesoamerica to leaves 2.5-5.5 c 1 cm long; m lon g). 0a. ras jemma without dnd pire length : width ratio 5-6: 1. Leaf blades lanc ndreanum P.a 10b. bows pem with glands; odes length: width ratio 3-4: 1. Leaf blades ovate-lanceolate to lanceola lla. Spikelets i )3.2-3. mM 4) mm long; nodes P e 2. P. biglandulare llb. S moe .8-3.1 m a. Spikelets 1.8-2.3 mm long; low glume by a conspicuous internode; leaf blades length : width ratio P.p m long; nodes densely pubesce er glume P from the upper iad 1 N > :1 Spikelets 2.5-3.1 m ower and upper glume; leaf blades length: width ratio 5. mm long; internode inconspicuous between t P. a 1. Panicum andreanum Mez, Bot. Jahrb. 125: 5. 1921. TYPE: Escuque, Moritz 1538 (lectotype, B, fragments at US (80458, 1108611)). Figure 3. Slender, creeping perennials with the culms decumbent and rooting at the lower nodes, then becoming erect and climbing up to 2.5 m into the shrubs, freely and densely branch- ing at the lower and upper nodes; internodes cylindric, hollow, glabrous to sparsely pilose, 1.5-8 cm long; nodes brown, glabrous to sparsely pilose. Leaf sheaths 2-7 cm long, commonly longer than the internodes at basal nodes or shorter at the upper ones, glabrous, striate and auriculate, the auricles small, densely pilose, one of the margins densely ciliate, the other glabrous. Ligule membra- nous-ciliate, 0.3-0.5 mm long; external ligule conspicuous, formed by a row of dense whitish hairs. Leaf blades lanceolate, 3-9 cm long, 0.3-0.6(-1.1) cm wide, flat, tapering into a finely attenuate apex, slightly narrowed to subcordate basally, hirsute, with long, thick papillose hairs to glabrous on both surfaces, the margins scabrous and cartilaginous, ciliate or glabrous basally; pseudopetiole glabrous, ca. 2 mm long. Panicles terminal, oblong, 6-16 cm long, 1-3 cm wide, with racemose primary branches alternate and divergent from the axis, the spikelets borne in pairs on short, unequal scabrous or pilose pedicels (the lower subsessile), these arranged along lower sides of branches; axis longitudinally ridged, smooth, scabrous, the axis of the branches triquetrous, scabrous, the axils of the branches pilose with dark hairs. Spikelets lanceolate, somewhat compressed laterally, 2.5-2.8 mm long, 0.4- 0.6 mm wide, acuminate, greenish to stra- mineous, pilose, with the upper glume and lower lemma subequal (or the upper glume a little shorter than the lemma). Lower glume ovate, acute, 1.1-1.4 mm long, Vs(-!4) the length of the spikelet, 3-nerved, the midnerve scabrous toward the apex, shortly pilose on the inner surface, sparsely pilose on the outer surface, bearing a few long hairs at the base. Upper glume 2.2-2.4 mm long, 5(-7)- nerved, the midnerve scabrous toward the apex, with long hairs in the hyaline margins. Lower lemma glumiform, 2.4-2.6 mm long, 5-nerved, hispid toward the margins. Lower palea lanceolate, 1.8-2.1 mm long, 0.3-0.4 mm wide, hyaline, short-pubescent at the apex, otherwise glabrous, the margins scabrous; male flower present. Upper anthecium lanceolate, 1.7-1.8 mm long, 0.4 mm wide, stramineous. Caryopsis ovoid, 1.2 mm long, 0.3 mm wide; hilum oblong. In flower November to May. Distribution. |. Venezuela and Colombia. In wet forests on sandy soils at 400—1,800 m elevation. Additional specimens examined. COLOMBIA. CUNDINAMARCA: near Quetame, Rio Negro valley, between Quetame and Piperal, Killip 34220 (COL, US); entre Quetame y Sasumuco, Triana 263 (US); Quetame, Triana Volume 75, Number 2 1988 Zuloaga & Sendulsky Revision of Panicum sect. Stolonifera 427 FIGURE 3. Panicum andreanum.— a ST SESSA SENS NS NS ~ NS SNS ws S f^ yr Habit.—b. Ligule.—c. Branch of a panicle showing pedicels.—d. w.—f. Spikelet, ventral view.— g. Upper anthecium, ventral view.— Racemose branch.—e. Spikelet, dorsal vie . Upper anthecium, dorsal view.—i. Caryopsis, hilum side. —j. Caryopsis, embryo side. Based on Muller s.n. US-1762338). 13 (COL). META: Cordillera de La Macarena, mesa del Rio Sansa, Idrobo & Schultes 1291 (US); Las Lagartijas, plateau between Rio Papamene and Rio Duda, Co a ibe trail, of Uribe, Fosberg 19508 (US); Ur carretera a Villavicencio, entre Puente Quetame y Buena Vista, García Barriga et al. 18936 (COL); 14 km NW of Villavicencio along the road to Bogotá, Davidse & Llanos 5516 (COL); Villavicencio, André 871 (paratype, . NORTE DE SANTANDER: región del Sarare, hoya del Rio Marguá entre Junin y Córdoba, Cuatrecasas 13373 (COL, 428 Annals of the Missouri Botanical Garden FIGURE 4. e. Spikelet, ventral view. —f. -—g. Upp M. hilum side. —i. Caryopsis, embryo side. a, b based on Matuda 316 (US); c-i based on Matuda 2006 (US). US). Without department and locality, Karsten s.n. (para- type, US). VENEZUELA. MÉRIDA: between Mucuchachi and Canagua, Steyermark 56345 (US). TACHIRA: Cordero, Muller s.n., 14 Nov. 1939 (US-1762338); Cerro Las Minas, 18 km SE of Santa Ana, Steyermark et al. 120043 (US); Cerro La Camiri, just south of the town of Rio jew Panicum biglandulare.— a. Habit.—b. Ligule.—c. Racemose branch.—d. Spikelet, lateral view. — pper anthecium, dorsal vi ] er anthecium, ventral view aryopsis, Negro, Davidse & González 21555 (US); 2 km E of El Variante, Davidse & González 21523 (US). When describing this species, Mez cited three syntypes for it, of which the specimen Volume 75, Number 2 Zuloaga & Sendulsky 429 Revision of Panicum sect. Stolonifera Moritz 1538 is selected here as lectotype of . andreanum, taking into consideration its protologue. 2. Panicum biglandulare Scribner & Smith, U.S.D.A. Bull. (1895-1901) 4 13, pl. 4. 1897. TYPE: Mexico. Chiapas: near Pinabete, 8 Feb. 1896, at an alti- tude of 6,500 to 8,000 feet, Nelson 3781 (holotype, US; isotype, GH). Figure 4. Perennials 40-80(-100) cm tall. Culms decumbent and rooting at the lower nodes or ascending, branching, with many nodes; in- ternodes cylindric, hollow, sparsely pilose to glabrous, with thin, whitish hairs; nodes greenish to purplish, constricted, glabrous. Leaf sheaths 2-4 cm long, usually shorter than the internodes, striate, densely ciliate at the margins, otherwise glabrous. Ligule mem- branous, shortly ciliate or glabrous at the apex, 0.3-0.5 mm long, external ligule a conspicuous ring of white hairs or absent; collar shortly pilose. Leaf blades ovate-lan- ceolate to lanceolate, 3-10.5 cm long, 1.1- 1.9 cm wide, acuminate, with strigose to pa- pillose hairs on both surfaces to glabrous, asymmetrical and subcordate basally, the margins ciliate to scabrous, the lateral nerves anastomosing; pseudopetiole short, glabrous to short-pilose. Panicles exserted, oblong, 8- 18 cm long, 1-4 cm wide, with 5-10 short, sparse, alternate, and racemose branches, these slightly divergent from the axis, the spikelets borne in pairs (the lower one occa- sionally abortive) on the branches, the axis of the branches triquetrous (rather flat), hir- sute, more so toward the base; the axils of the branches densely pilose; pedicels short, hispid. Spikelets lanceolate, acuminate, his- pid, (3-)3.2-3.7(-4) mm long, 1-1.1 mm wide, greenish to purplish. Lower glume ovate, acute, 1.3-1.8 mm long, ! the length of the spikelet, hirsute on the outer surface, sparsely and shortly pilose basally, with long, papillose- pilose hairs toward the apex, 3-nerved, the midnerve finely scabrous at the upper part. Upper glume shorter than the lower lemma, sometimes leaving the summit of the anthe- cium exposed, acute, 5(—7)-nerved, glabrous on the inner surface, with stiff, papillose hairs on the outer surface, these becoming more abundant toward the hyaline margins. Lower lemma acute, 5(-7)-nerved, scabrous, with long, papillose hairs toward the hyaline mar- gins; 2 conspicuous crateriform glands pres- ent between the midnerve and the 2 imme- diate lateral nerves. Lower palea lanceolate, acute, 3-3.2 mm long, 0.8 mm wide, mem- branous, scabrous at the margins, otherwise glabrous; male flower present, anthers ca. 1.3 mm long. Upper anthecium lanceolate, acute, 2-2.5 mm long, 0.7-0.8 mm wide, stramin- eous. Caryopsis ellipsoid, 1.4 mm long, 0.6 mm wide, brown; hilum oblong. In flower No- vember to August. Distribution. Occurring in Mexico and Guatemala in forests between 1,300 and 2,800 m elevation. arar i specimens examin ned. GUATEMALA. ALTA ERAPÁ an, von Tuerckheim II 1342 (GH, NY, US), il 1 956 (NY. Pis hills on Coban and Tres Cruces, Standley pa US). SAN MARCOS; near Aldea Fra- ternidad, u rdo, umulco, northwestern slopes of Volcán bare Steyermark 36678 (F, US). MEXICO. CHIAPAS: 2 mi. NE of Pueblo Nuevo Solist, Lathrop 5820 (US); Laguna Montebello, Montebello National Park, Breedlove & Dressler 29530 (F, NY); SE of Cerro Baúl ng a logging road to Colonia Gigaroa, road to Bochil, Breedlove 23323 (NY); Montebello, Carl- 6 (RB, US); in the paraje of Kulak’ tik, Ton 1713 (F); 25 mi. E of La Trinitaria, Lago of Monte Bello, Breedlove 9680 (US); Clinica Yerba Buena, 2 km NW of Pueblo Nuevo Solis- huacan, Raven & Breedlove 19846 (US); Sierra Madre, Tateoka 1009 (US). 3. Panicum brachystachyum Trin., Gram. Pan. 138. 1826. TYPE: Brazil. Minas Gerais: Serra do Cipo, Jan. 1825, Langsdorff s.n. (holotype, LE, not seen, fragment at US; isotype, P). Figure 5. Perennials ca. 18-35 cm tall. Culms erect, branching at the base; internodes cylindric, hollow, sparsely pilose; nodes stramineous, 430 Annals of the Missouri Botanical Garden AM d y r y x $1 Í 3 NING $ Ny 1⁄2 uz à P A VU FIGURE 5. Panicum Brachystachyum, —a, b. Habit.—c. Ligule.—d. Spikelet, lateral view.—e. Upper cium, dorsal view. —f. F (1829). glabrous to short-pilose. Leaf sheaths stra- mineous, striate, sparsely pilose, with small auricles, densely pilose, the margins short- ciliate. Ligule membranous-ciliate, arcuate, .4 mm long. Leaf blades lanceolate, 3- 4.5 cm long, 0.2-0.3 cm wide, acuminate, nthe- er.—g. Lodicules. a, c-g based on Langsdorff s.n. (US) ; b based on ae of Trinius flat or subinvolute, glabrous, subcordate to narrowed basally, the margins scabrous and with sparse, papillose hairs; the midnerve in- conspicuous. Panicles exserted, formed of 2- 4 alternate, densely flowered, distant and spreading racemose branches 1-4 cm long; Volume 75, Number 2 1988 Zuloaga & Sendulsky 431 Revision of Panicum sect. Stolonifera peduncle hispid; axis flattened, scabrous and long-hispid, the axils of the branches densely papillose; axis of the branches hispid to sca- brous, the pedicels arranged in pairs on one side of the branch, the pedicels triquetrous, hispid and with long papillose-pilose hairs. Spikelets broadly ellipsoid, 4.9-5.2 mm long, ca. 3.5 mm wide, grayish, gaping; glumes and lower lemma with long papillose-pilose hairs, the upper glume and lower lemma sub- equal, acute to acuminate, exceeding the up- per anthecium in length. Lower glume ovate, acuminate, 2.8-3 mm long, 44-14 the length of the spikelet, covered with thick, papillose hairs, 1-3(-5)-nerved, the midnerve sca- brous. Upper glume broadly ovate, 5-nerved, with thick papillose hairs near the margins, the rest of the surface with fine whitish hairs. Lower lemma glumiform, 3—5-nerved, with sparse, irregularly scattered, papillose hairs, these more dense toward the apex, the mar- gins hyaline. Lower palea lanceolate, 3.9- 4.5 mm long, 1.2 mm wide, membranous, the margins shortly ciliate; male flower pres- ent, the anthers dark purple, 3.2 mm long. Upper anthecium lanceolate, 4—4.4 mm long, 1.2 mm wide, membranous, acute to acu- minate, whitish to stramineous; lemma slight- ly carinate, 3-5-nerved; anthers ca. 3.3 mm long, lodicules ca. 0.5 mm long, cuneate, with raised distal margins; ovary ovoid; anthers dark purple. Caryopsis not seen. This species has been collected only once, in 1825, byLangsdorff “in saxosis montis alti da Lapa," now the Serra do Cipo, Minas Gerais, Brazil. The collection presumably con- sisted of a single individual, which was divided in three parts. One part is the holotype in LE, from which the plate in Trinius (1829) was probably drawn. The second part is the P isotype. The third part is a fragment of the type (taken from the LE specimen) in US and consists of a single branch without base. The plant from the Trinius plate and the US spec- imen are illustrated in the present paper to give a more complete view of this rare and probably extinct grass. Many collecting trips have been made to the Serra of Cipó by T. Sendulsky and by other botanists, but this species has never been collected again. 4. Panicum chapadense Swallen, Con- trib. Science 22: 8, fig. 4. 1958. TYPE: Brazil. Goiás: collected on sandstone out- crop, 7 km south of Veadeiros, region of the Chapada dos Veadeiros, 24 Apr. 1956, Dawson 14602 (holotype, R; iso- type, US). Figure 6. P. Rn Swallen, Phytologia 14: 78. 1966. TYPE: . Go as collected between rocks, at Pirineus, 18 Oct: 956, Macedo 4805 (holotype, US; iso- types, 6 SP, US) Rather robust perennials with long, creep- ing, horizontal rootstocks, the culms erect, 50-140 cm tall, with a cormlike base, 2 or 3 basal internodes orange-colored, 1-4 cm diam., glabrous, shining, lightly covered with aphyllous, velutinous old scales; new inno- vations appearing between those thickened internodes and covered by small, hard, yellow, pilose scales; upper internodes cylindric, solid, shortly pilose or glabrous, striate; nodes dark, constricted and shortly pilose, the first node swollen, yellow, glabrous, shining. Leaf sheaths stramineous, 5-11 cm long, longer (basal) or shorter than the internodes, densely villous to papillose-pilose all over the surface or pilose toward the apex only; the upper margins ciliate, the lower ones membranous; auricles small, rounded, sometimes densely pilose. Ligule membranous, ciliate, ca. 0.5 mm long, with or without long hairs behind the membrane at the base of the blade; ex- ternal ligule present or absent. Leaf blades lanceolate, 12-23 cm long, 1.3-2.5 cm wide, with ciliate to scabrous margins, subcordate basally and velutinous to glabrous on both surfaces, the midnerve not prominent; pseu- dopetiole small. Panicles terminal, lax, ob- long to pyramidal, with many flowers, 13-26 cm long, 2-6 cm wide, the primary branches racemose, dense and alternate, appressed or slightly divergent from the axis (the lower branches shortly branching at the very base), the spikelets secund and arranged in pairs on 432 Annals of the Missouri Botanical Garden A = C$ Fay = 2 ATL bs AA A NE ES rd Cc OF. f Se 372 AJÍ 23) s, STA ERP wee. > MW P. : F = M Y ARA GRES LES DS >= AED ye p- Pog: 5 SA > poe y pe Xx 4⁄2 = FIGURE 6. Panicum chapadense.—a. Leafy stems.—b. Culm showing cormlike base.—c. Ligule.—d. Portion of a racemose branch.—e. Spikelet, ventral view, lower lemma with glands.—f. Spikelet, lateral view.— g. Spikelet, dorsal view.—h. Spikelet, ventral view, lower lemma without glands.—i. Upper anthecium, ventral view.—j. Upper anthecium, dorsal view. Based on Burman €: Filgueiras 450 (SP). short, unequal pedicels, on the lower side of axils of the branches densely pilose, brownish, the branches; axis longitudinally ridged, near- sometimes with 1 or 2 long hairs; pedicels ly glabrous or finely hispid, the axis of the hispid and slightly pubescent. Spikelets nar- branches triquetrous, hispid to scabrous, the rowly ellipsoid, 2.5-3(-3.3) mm long, 0.6- Volume 75, Number 2 1988 Zuloaga & Sendulsk 433 y Revision of Panicum sect. Stolonifera 0.8 mm wide, stramineous, with purplish traces, the upper glume and lower lemma subequal and exceeding the upper anthecium in length. Lower glume ovate, acute, 1.6- 1.9 mm long, 14-34 as long as the spikelet, 3(-5)-nerved, shortly pubescent on both sur- faces (more so on the upper surface). Upper glume acute, 2.2-2.8 mm long, 5-nerved, the midnerve scabrous, densely hispid on the outer surface, with long, papillose, whitish and fringed hairs toward the margins, the inner surface pilose toward the apex. Lower lemma glumiform, 2.2-2.8 mm long, 5-nerved, with a pubescence similar to that of the upper glume, with or without 2-4 crateriform glands on the middle portion. Lower palea lanceo- late, 1.8-2.2 mm long, 0.5-0.7 mm wide, stramineous, shortly pubescent, the margins ciliate; male flower present. Upper anthe- cium narrowly ovoid, 1.5-1.9 mm long, 0.5- 0.7 mm wide, acute, stramineous. Caryopsis not seen. In flower March to October. Distribution. Endemic to Goias, Brazil. Occurring in “campos rupestres” and “cer- P rados" between stones in rocky habitats at 1,000-1,600 m elevation. Additional specimens examined. BRAZIL. GOIÁS: 5- 15 km S of Veadeiros, road to Sao Joao d’Alianga, Prance & Silva 58824 (MO, NY, US); 22 km N of Alto do Paraiso, Irwin et al. 32492 (F, NY); 20 km by road N of Alto Paraiso, Anderson 6760 (NY, UB, US); ca. 15 km S of Goiás Velho, Anderson 9976 (UB); ca. 15 km N of Corumbá do Goiás, Anderson 10305 (UB); serra do Pirineus, Burman & Filgueiras 410, 450 (SP); 12 k NW of Veadeiros, road to Cavalcante, /rwin et al. (US); ca. 15 km S of Veadeiros, Irwin et al. 12782 (MO, NY); Corumbá, Montes Pirineus, Onishi et al. 98 (R). Without locality, Macedo 4380 (US). A remarkable feature of this species is the presence of cormlike structures at the base of each culm, a character occasionally pres- ent in other species of Panicum, for example, P. bulbosum Kunth and P. paucifolium Swal- len. When describing P. chapadense, Swallen included it in sect. Laxa. This species has no affinity with sect. Laxa, being distinct by the type of spikelet, presence of glands on the lower lemma, and smooth and glabrous upper anthecium. 5. Panicum crateriferum Sohns, J. Wash. Acad. Sci. 46: 378, figs. 10-22. 1956. TYPE: Mexico. Guerrero: on steep grassy slopes and narrow ravine with open pine woods and scattered oaks on granitic soil at km 339-340 between Acahuizotla and Agua de Obispo, on highway to Aca- pulco, ca. 3,000 ft., 1 Oct. 1949, Moore Jr. 5148 (holotype, US; isotype, GH). Figure 7. Perennials, the culms decumbent, creeping and rooting at the lower nodes, becoming erect, 20-60 cm tall, branching; internodes hollow, 2.5-6 cm long, glabrous to sparsely pilose with whitish long hairs; nodes dark, densely p with whitish hairs. Leaf sheaths 0. m long, shorter than the internodes, ous to hispid, more densely so toward the upper portion, the margins long-ciliate toward the apex, otherwise glabrous; collar a nitid, wide rim of dense, whitish, antrorse hairs. Ligule membranous-ciliate, ca. 0.4 mm long. Leaf blades ovate-lanceolate to lanceo- late, flat, 4-6 cm long, 0.6-1.3 cm wide, asymmetrical basally, with the adaxial surface sparsely papillose-strigose, the abaxial surface glabrous to sparsely papillose-strigose, the margins ciliate basally, otherwise glabrous; midnerve inconspicuous, the lateral nerves anastomosing; pseudopetiole small, pilose. Panicles lax, 5-10 cm long, 1.5-6 cm wide, with 4-6 racemose and alternate branches, distant and divergent from the rachis, the uppermost branch consisting of a long pedicel and a single spikelet only; the spikelets borne in pairs, one subsessile (occasionally abortive), the other shortly pedicellate, arranged along the lower side of the branches; axis longitu- dinally ridged, glabrous, the axis of the branches triquetrous (one side flattened), sca- brous, with or without scarce, long, papillose hairs, the axils of the branches densely pilose with stiff and papillose hairs; pedicels short, scabrous. Spikelets narrowly ovoid to lan- ceolate, 2.5-3.1 mm long, 0.8-1.1 mm wide, sparsely to densely papillose-pilose or hirsute, the glumes and lower lemma subequal or the upper glume a little shorter than the lemma, Annals of the Missouri Botanical Garden FIGURE 7. Panicum crateriferum.—a. Habit. —b. Ligule.—c. Portion of a racemose branch.—d. Portion of a branch showing pedicels.—e. Sp ventral view. Based on Hinton et al. 10801 (U acute, the margins hyaline. Lower glume ovate-lanceolate, 1.3-2 mm long, 4-% the length of the spikelet, acuminate, with long papillose-pilose hairs toward the apex and margins, the rest of the surface shortly pilose, 3(-5)-nerved, the midnerve scabrous. Upper tkelet, lateral view.—f. Upper anthecium, dorsal view.— g. Upper anthecium, glume 2.1-2.8 mm long, 5(-7)-nerved, pa- pillose-hirsute toward the apex. Lower lemma 2.4-3 mm long, glumiform, 5(-7)-nerved, sparsely pilose, long-pilose or glabrous toward the margins, with 2 crateriform glands toward the upper part. Lower palea lanceolate, 2.2— Volume 75, Number 2 1988 Zuloaga & Sendulsk 435 y Revision of Panicum sect. Stolonifera 2.6 mm long, 0.5-0.6 mm wide, hyaline, ciliate at the margins, scabrous at the apex; male flower present. Upper anthecium ellip- soid, 1.5-2 mm long, 0.5-0.6 mm wide, stra- mineous. Caryopsis not seen. In flower Oc- tober to November. Distribution. Endemic to Guerrero and Oaxaca, Mexico; up to 1,000 m elevation. Additional specimens examined. MEXICO. GUERRERO: Domingo, Hinton et al. 14725 (paratype, i, San oe de Petlapa, Vera Santos 3437 (US); La Soledad, Ernst 2593 (US). A few three-flowered spikelets were found on the collection Hinton et al. 10801; here the spikelets had two lower flowers, one neuter and the other with stamens, and one her- maphrodite flower in the upper anthecium. These spikelets have two lemmas, both with crateriform glands, one neuter (the lower without a palea) and the other, in an inter- mediate position, with its corresponding palea and male flower. This characteristic three-flowered spikelet has been found previously and consistently in P. quadriglume (Doell) A. Hitchc. 6. Panicum poer Swallen, J. Wash. Acad. Sci. 30: 216. 1940. TYPE: Costa Rica. San José: vicinity of El General, 760 m, Feb. 1939, Skutch 4115 (ho- lotype, US; isotypes, GH, MO, NY). Fig- 8 ure Probably perennials. Culms decumbent, rooting and branching at the lower nodes, becoming erect, 60-80 cm tall; internodes cylindric, 4.3-11 cm long, glabrous, hollow; nodes glabrous. Leaf sheaths 3.2-4.8 cm long, shorter than the internodes, auriculate, the auricles pilose, the margins glabrous, membranous. Ligule membranous-ciliate, 0.3-0.5 mm long, with long hairs toward the base of the blade; adaxial surface of the collar shortly and densely pilose. Leaf blades ovate- lanceolate, 7-13 cm long, 1.5-3.3 cm wide, narrowed and somewhat asymmetrical basal- ly, finely scabrid and with strigose hairs to nearly glabrous on the adaxial surface, the abaxial surface finely scabrid to nearly gla- brous, the midnerve conspicuous, the lateral nerves anastomosing; pseudopetiole pilose, stramineous, ca. 1 mm long. Panicles ter- minal, oblong, exserted, 14-28 cm long, 3- 7 cm wide, with 15-30 racemose branches + divergent from the axis, alternate, and distant, usually drooping, the spikelets borne in pairs, one subsessile (occasionally abortive), the other shortly pedicellate arranged along one side of the branch; axis cylindric, finely scabrid, the axis of the branches somewhat flattened, scabrous, sparsely hirsute, the axils of the branches pilose; pedicels triquetrous, scabrous, pilose toward the base. Spikelets long-ellipsoid, biconvex, 1.8-2.3 mm long, m wide, greenish, scabrous to short pilose, the upper glume and lower lemma subequal (or the upper glume occasionally shorter), both with long hairs toward the mar- gins to glabrous. Lower glume ovate, acute, 0.7-1.1 mm long, 44-14 the length of the spikelet, 3-nerved, the midnerve scabrous. Upper glume 5-nerved, acute, pilose to finely scabrid toward the apex. Lower lemma glu- miform, 5-nerved, acute, scabrous toward the apex. Lower palea elliptic, 1.4-1.5 mm long, 0.5 mm wide, hyaline, membranous, scabrous on the wings. Lower flower hermaphrodite; anthers ca. 1 mm long; stigmas 2, plumose Caryopsis of the lower anthecium a little smaller than that present in the upper an- thecium, 0.8-0.9 mm long, 0.4—0.5 mm wide, free from its lemma and palea. Upper an- thecium ovoid, 1.3-1.5 mm long, 0.5-0.6 mm wide, stramineous, brown at maturity. Caryopsis ovoid, 0.8-1.2 mm long, 0.4—0.6 mm wide, the hilum punctiform to oblong. In flower November to April. Distribution. Costa Rica to Colombia and Venezuela at 650-2,000 m elevation. Additional specimens examined. Costa Rica. ALAJUELA: Rio Grande, cerca de San Ramon, Brenes 19683 (NY). SAN JOSE: Basin of El General, Skutch 4816 (GH, NY). CoLomBIA. Without locality: Smith 257 1 (GH). VENEZUELA. ARAGUA: slopes of mountainside near stream, between Choroni and Maracay, Soderstrom 978 (US); 436 Annals of the Missouri Botanical Garden FIGURE 8. Panicum irregulare.— a. Leafy branch.—b. Ligule.—c. Racemose branch.—d. Spikelet, lateral view.—e. Lower anthecium showing hermaphrodite flower.—f. Caryopsis of the lower anthecium, embryo side g. Caryopsis of the lower anthecium, hilum side | 1 side. Based on pper anthecium, dors ° al view.—i. Upper ant , embryo side. —k. Caryopsis of the upper anthecium, Volume 75, Number 2 1988 Zuloaga & Sendulsk 437 y Revision of Panicum sect. Stolonifera Parque Nacional Henry Pittier, NW of Maracay, Rancho Grande, Davidse 3017 (US). Although having the diagnostic characters of sect. Stolonifera, P. irregulare differs by the presence of an hermaphrodite flower in the lower floret, a character unknown else- where in the genus. This lower flower develops a normal caryopsis similar to the one present in the upper floret, only a little smaller. Unlike the upper anthecium, in the lower floret the lemma and palea are membranous, and the caryopsis is completely free from these bracts. Pohl (1980) suggested that this species may be a hybrid between genus Panicum and Pseudechinolaena, but we could find no evi- dence to support this. As previously stated, the species matches the characters that dif- ferentiate sect. Stolonifera from the other sections in subg. Phanopyrum. 7. Panicum latissimum Mikan ex Trin., in Sprengel, Neue Entdeck. Pflanzenk. 2: 87. 1821. TYPE: “Panicum latissi- mum Mikan detexit in Brasil et comm. an Mikan, sub cujus nom specium des- cripsi in Spr. gl. n. Entdx” (holotype, LE, not seen, fragment at US (974701)). Figure 9. P. n Raddi, Agrost. Bras. 46. 1823. TYPE Without locality: Raddi s.n. (holotype, PI. sb seen, fragment at US (80732)). Robust perennials up to 2.5 m tall. Culms decumbent and rooting at the lower nodes to erect, branching at the upper nodes; inter- nodes cylindric, hollow, glabrous, up to 1.5 cm diam.; nodes brown, constricted, glabrous. Leaf sheaths 12-14 cm long, striate, stra- mineous, densely pilose, with long, whitish, caducous hairs or glabrous. Ligule membra- nous-ciliate, small, 0.2-0.3 cm long, external ligule absent. Leaf blades 30-35 cm long, 6-12 cm wide, flat, acuminate, glabrous, cor- date and amplexicaul basally, the margins conspicuously ciliate to glabrous, the mid- nerve prominent or not prominent, the lateral nerves anastomosing; pseudopetiole brown- ish, glabrous, 0.6-1.3 cm long. Panicles py- ramidal, lax and diffuse, many-flowered, 30- 45 cm long, 10-18 cm wide, with alternate, distant, hirsute to scabrous branches diverg- ing from the axis, sometimes with secondary and tertiary branchlets; spikelets short- to long-pedicelled, along the lower side of the branches; axis longitudinally ridged, minutely scabrous, the branches hirsute to scabrous, triquetrous; axils of the branches short- to long-pilose; pedicels pilose. Spikelets ellip- soid, acute, 2.7-3.1 mm long, 0.9-1.1 mm wide, stramineous to brownish or purplish, the glumes and lower lemma shortly pilose and scabrous, the upper glume and lower lemma subequal, acuminate, both with long hairs to- ward the margins. Lower glume narrowly ovate, acute, 2.5-2.7 mm long, 12-34 the length of the spikelet, acuminate, shortly pi- lose to scabrous on the outer surface, densely pilose toward the apex of the inner surface, 5-nerved, the midnerve scabrous. Upper glume 2.5-2.9 mm long, 5(-7)-nerved, the midnerve scabrous. Lower lemma 2.7 mm long, glumiform, 5-nerved. Lower palea el- liptic, 1.8-2.3 mm long, 0.6-0.8 mm wide, stramineous, shortly pilose, the margins cil- iate; male flower present, the anthers ca. 1.7 mm long. Upper anthecium ellipsoid, acute, 2-2.2 mm long, 0.8 mm wide, stramineous. Caryopsis not seen. In flower October to April. Distribution. Brazil. In mountains, hu- mid and rocky habitats at 500-1,000 m el- evation. Additional specimens examined. BRAZIL. ESPÍRITO SANTO: Municipio de Alfredo Chaves, Vila Sao Bento de Uránio, mata higrofila, Zuloaga et al. 2410 (RB, SI, US). RIO DE JANEIRO: between Alto Boa Vista and Silvestre, Chase 8377 (F, MO, NY, US); vicinity of Paineiras, Corcovado, L. Smith 1205 (F, GH, US); Co rcovado, Riedel 329(US); without ry ct. 1836 (R are, Pabst et al. dir: Petrópolis, Gok & Dionisio 451 (RB), Parco n. (R); Serra dos Orgàos, perto do Veu das Noiv pra 712 (F); Serra dos Orgaos, Vidal 11-5580 TF). Pereira 187 (RB); ss Nacional da Tijuca, Bom Retiro, Soderstrom et al. 1855 (US); Bico do Papagaio, Landrum 2201 (RB), Ule 4158 (R, US); Tijuca, Chase 12159 (US), Oct. 1883, Schwacke & Saldanha s.n. (R); Pico da Tijuca, Chase 8486 (US); Estrada da Guanabara, Mata do Sumare, Sucre 1748 (RB); Guanabara, Alto da Boa Vista, Sucre 2091 (R); Sumaré, Sucre 407 1 (RB); Alto da Pedra da Gavea, Sucre 4297 (RB); Teresópolis, Vidal 18, 374 (R); estrada Tere- sópolis-Friburgo, Canoas, Braga 1532 (RB); Rio de Ja- 438 Annals of the Missouri Botanical Garden neiro, Riedel 464 (R). Without locality: Burchell 1110, 1381, 2158 (US); Gardner 210 (GH, US); Glaziou 504, 6973 (US), 17928 (P, US); Riedel s.n. (P; Gaudichaud s.n. (P) Panicum latissimum is clearly distin- guished from other species of sect. Stolonif- era and from the rest of the genus by having leaves up to 12 cm wide. 8. Panicum piauiense Swallen, Sellowia 18: 110. 1966. Based on P. blepharo- phorum Mez, Bot. Jahrb. Syst. 56, Beibl. 125: 4. 1921. Not Panicum blepharo- phorum J. S. Presl. TYPE: Brazil. Piauí: without locality, July-Sep. 1839, Gard- ner 2016 (holotype, BM, not seen, frag- ment at US; isotypes, GH, NY, P, US). Figure 10. eS ae Mez, Feddes a. Spec. Nov Beih. 15: 132. 1918. TYPE: Brazil. Goiás: bc de locality, 1841, Tm 3512 (ho- lotype, B, not seen; isotypes, BR, P, fragments at US). Cespitose, moderately robust, short-rhizo- matous perennials, 40-80 cm tall, with con- spicuous, fusiform, long root tubers up to 3- 5 mm thick, the cataphylls lanate. Culms erect, branching, many-noded, the internodes cylindric, densely to sparsely pilose or gla- brous, hollow; nodes brown, constricted, shortly pilose. Leaf sheaths 4-6 cm long, longer than the internodes, stramineous, gla- brous or scarcely pilose, one of the margins densely ciliate, with short, whitish hairs, the other glabrous. Ligule membranous-ciliate, 0.5 mm long, sometimes with long hairs to- ward the back at the base of the blade; ex- ternal ligule a row of antrorse whitish hairs, the collar stramineous, pilose. Leaf blades lanceolate, 9-13 cm long, 1-1.5 cm wide, acuminate, flat, cordate to subcordate basally, scabrous to densely villous on both surfaces, the adaxial surface shortly pilose at the base, the margins white and cartilaginous, long- ciliate or glabrous basally, otherwise minutely scabrous; pseudopetiole small, shortly pilose. Panicles lax, oblong, 7-22 cm long, 2-4.5 cm wide, with the branches alternate and diverging from the axis, rarely with short, appressed secondary branchlets; axis longi- tudinally ridged, sparsely hispid, scabrous, the branches triquetrous, scabrous and sparsely hispid, bearing spikelets in pairs: one subses- sile, the other shortly pedicellate, the lower one distant; axils of the branches long-pilose to villous; pedicels pilose or scabrous. Spike- lets narrowly ellipsoid, 2.4-2.8 mm long, 0.8 mm wide, stramineous to purplish, sparsely pilose or with long hairs at the margins of the glumes and lower lemma. Lower glume ovate, acuminate, 1.8-2.5 mm long, 1⁄¿—3⁄4 the length of the spikelet, scabrous to sparsely pilose on the outer surface, densely pubescent toward the apex on the inner surface, 3-nerved, the midnerve scabrous. Upper glume acute to acuminate, 2.3-2.7 mm long, pilose to sca- brous, with long, stiff and whitish hairs toward the margins on the outer surface, the inner surface shortly pilose, 5-nerved, the midnerve scabrous. Lower lemma glumiform, acute, 2.3-2.6 mm long, 5-nerved, with pubescence similar to that of the upper glume, with or without 2-4 prominent, ocellate and crater- iform glands on the middle portion. Lower palea elliptic, 1.7-2 mm long, 0.6 mm wide, stramineous, shortly pilose, the margins cil- iate; male flower absent. Upper anthecium narrowly ovoid, 1.7-2 mm long, 0.6 mm wide, stramineous. Caryopsis ellipsoid, 1.3 mm long; hilum oblong. In flower January to April. Distribution. Brazil, in cerrados of Ba- hia and Piaui at 800 m elevation. Additional specimens examined. BRAZIL. BAHIA: Chapadào do Panair, Serra do Mimo, Black 55-17982 (IAN); Espigào Mestre, ca. 100 km WSW of Barreiras, Anderson et al. 36751 (F, MO, R, US); Serra de Teririco, Gruta do Pequerio, Zehntner 67 (R); Serra do Teririco, Zehntner 3746 (RB, US); Serra do Sincorá, 15-20 km from Andarai, along the road to Itaete, Harley et al. 18652 (MO, P). This species can be mistaken for poorly developed plants of P. rude, but the latter normally reaches greater size (2 or 3 m high). Panicum piauiense differs further from P. rude by having food-storing thickened roots and by having culms that branch from the base up to the upper part of the plant. The Volume 75, Number 2 Zuloaga & Sendulsky 439 1988 Revision of Panicum sect. Stolonifera DI RD Neches SSS d ISR, p i ü Lp es M NA \ \ \ \ NN \ N N À MA FIGURE 9. Panicum latissimum. — a. Blades and portion of the panicle.—b. Ligule.—c. Portion of the panicle showing racemose branches.—d. Spikelet, ventral view.—e. Spikelet, lateral view.—f. Upper anthecium, dorsal view.—g. Upper anthecium, ventral view. Based on Chase 8486 (US). two species also have different distributions, P. piauiense are unique within Panicum. P. piauiense occurring only in Bahia and Soderstrom (1981) reported tubers in the Piauí, P. rude from Espirito Santo to Rio nonpanicoid grasses Puelia ciliata Franch., Grande do Sul. The fusiform root tubers of Lophatherum gracile Brongn., Molinia cae- 440 Annals of the Missouri Botanical Garden h FicuRE 10. Panicum piauiense. — a. Habit. —b. Ligule.—c. Detail of a racemose branch.—d. Spikelet, ventral view.—e. Spikelet, lateral view.—f. Spikelet, dorsal view.—g. Spikelet, ventral view, Sind — ada h. Spikelet, lateral view, lower nnd with glands.—i. Upper anthecium, dorsal vie ahaa ventral view. — ‘yopsis, side.—l. Caryopsis, embryo side. a-f, i-l based on sym m. 36751 (US); g, h based on Ps 3746 (US rulea (L.) Moench, and Sucrea sampaiana (Raddi) Kunth, Rev. Gram. 1: 30. 1830. (A. Hitchc.) Soderstrom. TYPE: Brazil. Rio de Janeiro: in sylvaticis prope Catumby, non procul ad urbe Rio 9. Panicum pulchellum Raddi, Agrost. de Janeiro, Raddi s.n. (holotype, PI, not Bras. 42. 1823. Eriochloa pulchella seen, fragments at BAA, US). Figure 11. Volume 75, Number 2 1988 Zuloaga & Sendulsk 441 y Revision of Panicum sect. Stolonifera P. leptostachyum J. S. Presl, Rel. Haenk. 1: 311. 1830. ymenachne leptostachya (J. S. Presl) hus. 6. 1886. ex. P. bipustulatum Schldl., Linnaea 26: 135. 1853. Probably perennial. Culms decumbent, ex- tensively creeping and rooting, geniculate at the lower nodes, then becoming erect, freely branching, 10-65 cm tall; internodes long, compressed, pilose to glabrous; nodes ob- scure, densely villous with whitish hairs. Leaf sheaths 0.7-2.5 cm long, shorter than the internodes, striate, membranous, densely pi- lose, with long, whitish hairs to glabrous, the margins ciliate. Ligule membranous-ciliate, ca. 0.4 mm long, the collar densely pilose. Leaf blades ovate-lanceolate, acuminate, 2.5— 5.5 cm long, 1-2 cm wide, asymmetrical and cordate basally, strigose to glabrous on both surfaces, the basal margins long-ciliate with thick, caducous hairs, otherwise scabrous to ciliate, the abaxial surface often purplish; midnerve prominent, the lateral nerves anas- tomosing; pseudopetiole densely pilose, with long, thick hairs. Panicles terminal, 4-18 cm long, 1.5-4 cm wide, short- to long- exserted, the peduncle hispid, formed by 5- 20 secund, alternate or occasionally opposite and racemose branches, these distant, as- cending or reflexed, divergent from the axis; axis longitudinally ridged, hispid, the branch- es triquetrous, flattened on one side, densely hispid toward the base, hispid to scabrous on the rest of the surface, with spikelets borne in pairs, one subsessile, the other short-ped- icellate (the subsessile spikelet frequently abortive), the axils of the branches pilose. Spikelets narrowly ellipsoid, 1.8-2.3 mm long, 0.6-0.7 mm wide, greenish, the glumes and lower lemma hirsute, the hairs papillose, rigid, short; upper glume and lower lemma subequal (or the upper glume shorter), acu- minate. Lower glume ovate, acute, 0.8-1.1 mm long, 1⁄4—-1⁄ the length of the spikelet, short-pilose on the middle portion, hirsute to- ward the margins, separated from the upper glume by an internode, 3-nerved, the mid- nerve scabrous. Upper glume 1.8-2 mm long, 5-nerved, hirsute. Lower lemma 1.7-2.1 mm long, 5-nerved, short-pilose on the middle por- tion and long-pilose toward the margins, bear- ing 2 crateriform ocellate glands between the midnerve and the 2 immediate lateral nerves, or the glands occasionally absent. Lower pa- lea 1.4-1.6 mm long, 0.4-0.6 mm wide, lanceolate, hyaline, short-ciliate at the mar- gins, glabrous in the rest of the surface; male flower usually absent. Upper anthecium el- lipsoid, acute, 1.2-1.5 mm long, 0.7-0.9 mm wide. Caryopsis ellipsoid, 1-1.2 mm long, 0.5-0.6 mm wide; hilum oblong. In flower all year. Distribution. This species occurs from Mexico, Guatemala, Belize, Honduras, Nic- aragua, Costa Rica, and Panama to Colombia, Venezuela, Ecuador, Peru, Boliva, and Brazil. It grows in humid and shaded places from sea level to 2,000 m elevation. Chromosome number. = 10 (Davidse & Pohl, 1974); 2n = 20 (Gould & Soder- strom, 1970; Pohl & Davidse, 1971) Selected specimens examined. BELIZE. EL CAYO: Chalillo crossing, Lundell 6513 (F, US); Norris Woods, MO). STANN CREEK: Big Creek, Schipp cerca + Buzinio Hills, Gentle 5095 (F, US); Branch, Monkey River, Gentle 3962 (F, GH, MO, NY, US) Edwards Road, near Columbia, Gentle 6447 (F). Without district: Gracie Rock, Sibun River, Gentle 1539 (MO); Pine Ridge, near Manatee Lagoon; Peck 279 (GH). BOLIVIA. LA PAZ: Prov. Larecaja, ruta entre Caranavi Guanay, puente sobre el Rio Coroico, Croat 51685 (MO); Guanay, Rusby 217 (NY, US). BRAZIL. MATO GROSSO: Santa Anna da Chapada, Malme 3396 (US). MINAS GERAIS: Vicosa, Chase 9445 (GH, MO, NY, US); Pico do Itabira, Mattos s.n. (R-38680). RIO DE JANEIRO: Angra dos Reis, Castellanos 801 (F); Teresópolis, Serra dos Orgãos, paio 2426A (US) matas do Andarahy e "Trapichdirs, Ku Almann s.n., Apr. 1917 (R); Parque p Itatiatia, ago » ¢ Ey et al. 2367 (RB, SI, US). Eoo ANTIOQUI de “sl Kang Denslow 1 (MO); WAR el Seok Providencia, 26 km S y 1 al W air of Zaragoza, in valley of Rio Anori, Denslow 2728 (COL), Denslow 265 1 (COL); Granja de las Mercedes en Venecia, Barkley & Gutierrez 637 (COL , US); Tirana . CH S de 3 población, Forero et al. 3436 (COL, MO). CUNDIN- CA: Sasaima, vereda San Bernardo, La Maria, Bar- rios "125 73 (COL); Estación Central de Investigación La Esperanza, Obregón 19 ( L); en las cercanias del Bo- G of Pueblo Bello, Angel 733 (US). km SE of Villavicencio, Haught 2531 (COL, F US) margen derecha del Rio Guayabero, caudal de la Annals of the side. Based on Chase 12421 (US). Macarena, Pinto y Bischler 344 (P, COL); Villavicencio, Cuatrecasas y García Barriga 4496 (COL, F, US), Triana 14 (COL); Cordillera la Macarena, mesa del Rio Sansa, Vids & Schultes 1282 (COL); Reserva Nacional de La carena, junction of Rio Sansa and Rio Guejar, Thomas et al 1426 (COL); valley of Rio Tigre, Fosberg 19043 442 Missouri Botanical Garden FIGURE 11. Panicum pulchellum. — a. Habit. —b. Ligule.—c. Portion ofa racemose branch showing pedicels. — d. Racemose Mis e —e. oe ikelet, dorsal view.—f. Spikelet, lateral view.—g. Spikelet, ventral view.—h. Upper anthecium, dorsal v pper anthecium, ventral view.—j. Caryopsis, embryo side.—k. Caryopsis, hilum (US). NARINO: Ricaurte, von Sneidern A-539 (GH). SANTANDER: Cordillera Este, Mesa de Los Santos, Killip & Smith 15346 (MO, NY, US). NORTE DE SANTANDER: Region del Sarare, Hoya del Rio Margua, bosques en la n ada del Rio Negro, Cuatrecasas 12910 (COL, US). LLE DEL CAUCA: Costa del Pacifico, Rio Cajambre, Cua- Volume 75, Number 2 1988 Zuloaga & Sendulsky 443 Revision of Panicum sect. Stolonifera trecasas 17090 (F). COSTA M ALAJUELA: San Ramón, Brenes 2138 al NE in , NY); cercanias de Pejivalle, 616 (F, GH, MO, US). H A: banks of Rio María da Godfrey 66533 (US) alle del Río Sarapi- qui, E de La Virgen, Pohl 12827 (MO, NY). PUNTARENAS: 1 km al N de la Carretera Interamericana, Poh vidse 11605 (F) SÉ: environs de Buenos Aires, NY, US), 3874 (GH, MO, NY, US); El General, Skutch 3890 (GH, MO, NY, US); San Antonio de Desamparados, Pohl & Lucas 12992 (MO). ECUADOR. AZUAY: entre e Río Gamolotay y el Río Norcay, Steyermark 52886 (F). BO : Sibambe, Acosta Solís 5338 (F, US), 5339 m W of Bucay, Hitchcock Y, US). IMBABURA: entre El Pajón y Cachaco, Acosta Solís 12709 (US). EL ORO: between La Cholita and Portobello, Hitchcock 21212 (NY, US). NAPO Pas- TAZA: Tena, Asplund 8871 (US). PICHINCHA: Santo Do- mingo de los Colorados, Asplund 16416 (NY, R, US), Acosta Solis 10883 (F, US). Junction of the provinces of Guayas, Canar, Chimborazo, & Bolivar: near the village of Bucay, Camp 3819 (GH, NY, US); Bucay, Rose 22446 (NY, US). Spars ALTA VERAPAZ: near Secanquim, Mason 3153 (US of Quirigua, Weatherwax 92 (US); near Quiriguá, Stand- ley 23709 (F, GH, MO, US), 24246 E MO, NY, US). PETÉN: Lancandón, Contreras 3340 (GH , US); La Lib- ertad, Aquilar 185 (MO, US), Lundell 2120 (US), 2545 (GH, US); Rio Pasión, 4 km NE of Puste, Lundell 18099 (GH). QUEZALTENANGO: between Finca Pirineos and Pat- zulín, Standley 86608 (F, US); El Palmar, Kellerman 6246 (F, US). RETALHULEU: Rio Coyote, Standley 87445 (F). SANTA ROSA: La Joya de Limón, E of Cuilapa, Standley 78309 (US); near =: Molino, epe 78429 (F, US). TEXACAPA: 13 km , Harmond & Fuentes 1854 (MO). o ATLÁNTIDA: Cutiapa, m de Las Ceibas, Velson et al. 3412 (MO); Lancetilla, Pohl Agua Fria, Molina 7642 (US). MORAZÁN: Rio Yeguare, Molina 1374 (MO). OLANCHO: Jutiapa forest camp, near Salamá, Pohl & Gabel 13746 (F). vicini of Juticalpa, Standley 17863 (F); between Catacamas and La Presa, Standley 18307 (F). YORO: Ciudad de Yoro y alas Nelson 1989 (MO). MEXICO. CHIAPAS: Campo Experi- mental de Quina, Vera Santos 2731 (US); near the junc- i i e an Quintin, near Sohns 1652 (US), Breedlove 33374 cordia, Morton o, Rovirosa 323 (NY). VERACRUZ: Córdova, Hitchcock 6444 (US); Valle de Córdova, d 1455 (GH, US). NICARAGUA. CHONTALES: Santo Tomás, Sey- mour 2753 i. GH, NY); Santo Domingo, Š 27411 (F). MATAGALPA: ‘Cordillera Central de r 9 km N of Matagalpa, Williams et al. 23740 (F, PP US). ZELAYA: a lo largo del Rio Grande, us na (F, GH, US); near El Recreo, Standley 19570 (F). p AMA. CHIRIQUÍ. San Bartolo, 19 km W de Pto. Armuellos, Busay 610 (F, MO). coLÓN: Canal Zone, Mount Hope Cemetery, Standley 28824 (US); Canal Zone, cerca de Culebra, Pittier 2226 (US); near Fort Sherman, Pond 31045 (US); Barro Colorado Island, trail at rear 8, Croa 7436, 13150 (MO); Canal Zone, Quebrada Bonita, Sty ermark & Allen 17157 (US); Canal Zone, between Fran field and Catival, Standley 30176 (US). PANAMA: E p the Rio Tacumén, Standley 26554 (US); Rio Tapia, Hitchcock 22945 (F, R. US); Altos de Campana, Méndez 179 (MO); sabana near Chepo, Hunter & Allen 54 (GH, US); Archipiélago Perlas, San José Island, Erlandson 170 d bes 2 Johnston 1136 (GH, bid 1274 L (GH). PERU. 11315 (US); without UN 5 x ay 1930, (F-659976, US). V on [sla Carestia, saltos allo, m o Sanariapo, Maguire et al. 36162 (NY, US). ANZOÁTEGUI: Fila El Guácharo, Davidse cafetales al NW de Buenos Aires, kms aéreos al N de Bergantin, Davidse & González 19632 (MO). ARAGUA: 12 km S of Alto de Choroni, Davidse 3078 (MO); Cor- dillera Interior, entre el Pauji y el Socorro, Steyermark 118086 (MO). DISTRITO n Cotiza, Chase 12421 acional Yacambu, V186 (US). MÉRIDA: 2 km del Rio er & González de (MO). MIRANDA: Guinand Estate, iter 5975 (NY, US) Los Teques, Chase 12303 (NY , US), Archer 3055 (US). PORTUGUESA: b arigua, dei 12086 doi uc E 2 (MO); cerro Las Minas, Steyermark et al. 119937a (VEN). YARACUY: Cerro La Ee 7 km N of Nirguá. Davide et al. 20809 ( 10. Panicum rude Nees, Agrost. Bras. 158. 1829. TYPE: Brazil. Minas Gerais: habitat in marginibus sylvarum Districtus Adamantum prope Milho verde, Martius n. (holotype, M, not seen, fragments at BAA, US). Figure 12. P. bambusaefolium Desv., Opusc. 83. 1831. TYPE: Brazil. Without state and locality: Desvaux s.n. (holotype, P; fragments at BAA, US). P. secundum Trin., Spec. Gram. 3: pl. 324. 1836. P. secundum var. secundum (as 'subaequiglume") Doell, in C. Martius, Fl. Bras. 2(2): 194. 1877. TYPE: Brazil. Without locality: Riedel s.n. (holotype, LE, not seen, fragment at US; isotypes, P, US). P. secundum var. ri, rr Doell, in C. Martius, Fl. Bras. 2(2): 1 P. semitectum Elles Sellowia 18: 112. 20 Dec. 1966. ot Panicum semitectum Swallen, Phytologia 14: 68.5 Dec. ind TYPE: Brazil. Paraná: Jaguariaiba, in woods, overhanging on bank, 25 Feb. 1946, Satan 8675 (holotype, US). 444 Annals of the Missouri Botanical Garden ` Q a) SAW e W S 4 MES WW O NN YE QW. AY IY Y Y, FIGURE 12. Panicum rude. — a. Portion of a culm and panicle.—b. Ligule.—c. Detail of a racemose branch.— d. Spikelet, ventral view.—e. Spikelet, lateral view.—f. Spikelet, ventral view, lower lemma with glands.— g. Upper anthecium, dorsal view. Ipper anthecium, ventral view.—i. Caryopsis, embryo side. —j. Caryopsis, —h. Upp ew.—i. C hilum side. f based on Riedel s.n. (US-974743) ; a-e, g-j based on Sendulsky 1039 (US). P. pompale Swallen, Sellowia 18: 110. 1966. TYPE: Bra- ler-Urussanga, 23 Aug. 1958, Reitz & Klein 7043 zil. Rio de Janeiro: top of sheer face of Corcovado, I (holotype, : Rio de Janeiro, 11 Jan. 1925, Chase 8165 I (ho- P. albospiculatum Swallen, Sellowia 18: 110. 1966. TYPE: lotype, US). Brazil. Santa Catarina: Rio Cagador, 22 Jan. 1946, P. kleinii Swallen, Sellowia 18: 111. 1966. TYPE: Brazil. Swallen 8291 (holotype, US). Santa Catarina: Pinhal da Companhia, Lauro Mul- P. apricum Swallen, Sellowia 18: 112. 1966. TYPE: Bra- Volume 75, Number 2 1988 Zuloaga & Sendulsk 445 y Revision of Panicum sect. Stolonifera zil. Santa Catarina: Campo dos Padres, ov. 1956, Smith, Reitz & Klein 7643 (holotype, US; isotype, NY). Robust, cespitose, short-rhizomatous pe- rennials. Culms decumbent and rooting at the lower nodes to erect or leaning among branch- es of trees, 60-200(-300) cm tall, usually simple, the internodes cylindric, glabrous, hol- low; nodes obscure, compressed, glabrous. Leaf sheaths 7-20 cm long, longer than the internodes, stramineous, auriculate, striate, papillose-pilose with caducous hairs to densely or sparsely hispid with whitish and appressed hairs, otherwise completely glabrous, the mar- gins pilose to glabrous. Ligules membra- nous-ciliate, 0.5-2.5 mm long; external ligule conspicuous, similar to the inner ligule but smaller; collar brown, pilose to glabrous. Leaf blades lanceolate to long-lanceolate, acumi- nate, flat, 15-45 cm long, 1.6-5 cm wide, cordate to subcordate basally, hispid on both surfaces with appressed short hairs to gla- brous, the margins scabrous, ciliate or gla- brous; pseudopetiole brownish, 0.5—1 cm long and with ciliate margins. Panicles lax, py- ramidal to oblong, many-flowered, 20-65 cm long, 5-25 cm wide, with secund branches alternate to subopposite, diverging toward the base of the panicles and contracted toward the apex, the spikelets short-pedicelled and disposed in pairs in the branches; axis lon- gitudinally ridged, scabrous to densely hispid; branches and branchlets longitudinally ridged, scabrous and long- to short-pilose, the axils of the branches brownish, shortly pilose; ped- icels scabrous, sometimes with long hairs. Spikelets ellipsoid, 2.3-3 mm long, 0.8- 1.1 mm wide, stramineous or nearly purplish, the glumes and lower lemma sparsely to densely pilose, upper glume and lower lemma sub- equal and longer than the upper anthecium. Lower glume shortly pilose on both surfaces, acute, 1.4-2.5 mm long, 1⁄¿—3⁄4 the length of the spikelet, 3(-5)-nerved, the nerves anas- tomosed apically, the midnerve markedly sca- brous. Upper glume 5(-7)-nerved, acute to acuminate, scabrous, pilose toward the mar- gins and the base or with whitish hairs on the entire surface. Lower lemma 5(-7)-nerved, acute to acuminate, with pubescence similar to that of the upper glume, occasionally with one pair of glands on the middle portion. Lower palea lanceolate to elliptic, 1.8-2. mm long, 0.6-0.7 mm wide, hyaline, with long-ciliate margins; male flower present or absent. Upper anthecium ellipsoid, 1.7-2.3 mm long, 0.6-0.8 mm wide, stramineous to brownish at maturity. Caryopsis ovoid, light brown, 1.4-1.7 mm long, 0.8-1 mm wide; hilum oblong. In flower September to April. Distribution. Brazil Common in interior or edges of forests at 0—1,900 m elevation. Common names. Papanduva, capim-pa- panduva, papua, capim papua, cana-de-ma- caco (Smith et al., 1982); capim de anta (Pereira 2281). Additional specimens examined. BRAZIL. DISTRITO FEDERAL: 10 km NW of Planaltinha, Irwin et al. 13196 cco, d 4898 (US); Serra do agers mW of Barào de Cocais, Irwin et al. 29321 (MO, m Serra de Ouro Preto, Magalhaes Gomez 282 1 (US); Serra do Espinhaço, 35 km E of Belo Horizonte, Irwin et al. 30390 (MO, NY); Serra do Espinhago, Pico do Itambé, Anderson et al. 35723, 35894 (MO); Ouro Pre- to, Pires & Black 3375 (US); P peur L. Bailey 1094 US); Itacolumy, Chase 9420 (F, N PARANÁ: Cu- ritiba, Swallen 8594 (US); Banhado- he roqua ra, Swallen 8644 (US); Paredào da Santa, Hatschbach 35468 (MO): Fazenda Monte Alegre, Rio Harmonia, Hatschbach 3027 (SI); Serra Capivari Grande, Hatschbach 22959 (NY, US); Jaguariaiva, Dusén 13233 (NY, US), 15920 (MO); Rio Iguacu, Salto Grande, Hatschbach 14934 (US); Ba- rigny, Dusén 15765 (F, MSC); Jacarei, Dusén 17012 (F). RIO DE JANEIRO: Corcovado, Chase 5 II (NY, US); Pao de Acucar, Chase 10043 (US). RIO GRANDE DO SUL: Cambara do Sul, Jan — un — 4. > 12 > = > 2 2 Zz > NN = 3 : of Campo Alegre, Smith & Klein 7344 (US); Pinhal da Companhia, Lauro Muller- ssanga, Reitz & Klein 7043 ~ beirao, Klein 6924 (US), Klein & (US); between Fazenda Santo Antonio and the falls of Rio Canoas, Campo dos Padres, Smith & Klein 787 1 (NY, US); Itajaí, Morro da Ressacada, Klein 1775 (NY, US); Porto União, Orth 2510 (US); Pilões, Reitz & Klein 2456, 2762 (US), 3630 (NY, US); Brus- d m Bresolin 7647 9220 (NY): base of Morro do Funil, Smith & Klein 154 70 (R, US); Sabiá, Klein 2252 (NY, US), 2271 (US); Pin- 446 Annals of the Missouri Botanical Garden heiral, Smith & Reitz 8750 (US); Morro do Cambirela, Klein & Bresolin 9714 (US); Morro da Bateia, Reitz 1907 (US); Biturina, L. Emygdio 693 (R). SAO PAULO: Sao Paulo, Parque do Estado, grounds of the Instituto de Botánica, Davidse 10510 (MO), Hoehne 27202 (F, NY, US), Sendulsky 278, 417, 1039 (SP, US), Skvortzov 157 (SP, US); Igaratá, 1 Mar. 1939, Gehrt s.n. (US); Parque Estadual das Fontes do Ipiranga, da Silva 258 (MO). The polymorphous nature of Panicum rude lies behinds its numerous descriptions under different names. Its pilosity is variable on the vegetative and floral parts, with leaf sheaths, leaf blades, and inflorescences varying from densely pilose with different types of pubes- cence to glabro The j. can be densely pilose on the glumes and lower lemma to glabrescent (only minutely scabrous on the glumes). The lower glume varies from Y to about % the length of the spikelet, even in the same specimen (as for example in /rwin et al. 29321). The crateriform glands on the middle por- tion of the lower lemma may be present or absent on the same specimen, but they are most often absent. These glands are present in the type specimens of “P. bambusaefoli- um” and “P. secundum," and they occur in Irwin et al. 13196. In the voluminous re- cently collected material in the Instituto de Botánica of Sao Paulo studied by T. Sendul- sky, glands were not detected. The illustration of P. secundum in Trinius (1836, pl. 324) draws attention to the pe- culiar one-sided position of the leaves. Pan- icum rude grows mostly at the borders of forests; when the culms develop in more or less open and uniformly lighted areas, they bear leaves distichously or alternately ar- ranged. On the other hand, when the culms grow at forest margins and lean against dense vegetation, they receive light only from one side, which promotes unilateral arrangement of the leaves due to the twisting of the culm (Fig. 12). Swallen (1966) treated five species closely related to P. rude within the Latissima group. The characters used by Swallen to separate these species were mainly pubescence of the leaf sheaths and blades and the sizes of plants and spikelets. We conclude that these char- acters do not justify separation of species. Swallen cited Chase 8165 as type of “P. pompale" and Reitz & Klein 7043 as type of “P. kleinii.” It should be noted that these specimens are divided into two sheets each: Chase 8165 I, Chase 8165 II, Reitz & Klein 7043 I, and Reitz & Klein 7043 II. In both cases, Chase 8165 land Reitz & Klein 7043 I contain the upper portion of a culm (in- cluding the panicle), and Chase 8165 II and Reitz & Klein 7043 Il contain the vegetative part of the plant only. Consequently, Chase 8165 I and Reitz & Klein 7043 I should be considered as holotypes of the two names. 11. Panicum soderstromii Zuloaga & Sendulsky, sp. nov. TYPE: Brazil. Bahia: Municipio de Mucujé, 3 km ao S de Mucujé, na estrada para Jussiape, 1,000 m de alt., 13%00'S, 41?24'W, campo us 26 July 1979, S. A. Mori, R. ing, T. S. dos Santos & J. L. Hage 12652 (holotype, CEPEC; isotype, MO). Figure 13. Gramen probabiliter perenne (basis non visa), 4 cm altum, culmis erectis, caespitosis, cylindricis, pilosis. Foliorum vaginae internodis superantie, uo et longe pilosae, basilitex pilis papillosis instructae. ula bre- breviter cata. Folio- rum laminae anguste lanceolatae, acutae, 2-1 m lo gae, utrinque breviter pilosae, asidus angustis, dic marginibus rotundatis. Paniculae e dag ues pyramidales, effusae, 8-18 cm longae, 2-3(-4.5) c latae; ramis adscentibus. Spiculae late da nen que hiantae, 2-2.8 mm longae, mm latae, stra- mineae vel violaceae; gluma inferior ovata, Y2-% lon- gitudinis spiculae aequans, 3-n nervis, pilis longis, albis, ad u spiculae subae- quans, cium inferum: lemma ovatum, spiculam aequans, 5-nerve, marginibus subtiliter ciliatis, 2-4 ped crateriformi- bus, o cellatis, infra apicem sitis; palea acuta, ad apicem pilosa. Anthoec cium superum clibsoideum, stramineum oe hie Neg gs s Caryopsis ellipsoi- dea succinea; hilum ovatum sub- n ae rds. ca. 1⁄4 caryopsis aequans. Cespitose, rather robust probable peren- nial, 45-70(-95) cm tall (base not seen), the lower nodes covered with small, pubescent sheaths. Culms erect, branching at the me- dian and upper nodes; internodes cylindric, hard, solid or hollow, pilose, striate, the nodes Volume 75, Number 2 Zuloaga & Sendulsky 447 1988 Revision of Panicum sect. Stolonifera < = " y ESTOS 22 2227 d $2 Ss gule.—c. Portion of a branch showing pedicels.— w.—g. Spikelet, dorsal view.—h. Spikelet, h two pairs of glands.—i. Spikelet, lateral view, lower lemma with two pairs of glands.—j. Upper anthecium, dorsal view.—k. Upper anthecium, ventral view.—l. Caryopsis, embryo side.— m. Caryopsis, hilum side. Based on Mori et al. 12652 (MO). dark, constricted, pilose. Leaf sheaths longer than the internodes, slightly auriculate, tight- ly embracing the culms, with long, papillose, pilose hairs; the lower sheaths approximate, short. Ligule membranous-ciliate, ca. 0.5 mm ong; collar area densely pubescent, dark. Leaf blades long-lanceolate, 2-13 cm long, 0.7- 1.3 cm wide, stiff, attenuate toward the apex, 448 Annals of the Missouri Botanical Garden a little narrowed and truncate basally, with rounded margins, velutinous on both surfaces; the midnerve not prominent; the basal leaves ca. 2-3 cm long, smaller than the uppermost leaves; pseudopetiole short. Panicles ter- minal, pyramidal, 8-18 cm long, 2-3(- cm wide, with numerous, many-flowered, up- wardly appressed or spreading racemose branches bearing spikelets in pairs on unequal pedicels; axis glabrous, longitudinally ridged, finely hispid along the ridges; axils of the branches slightly swollen, dark brown, pilose; pedicels with 2 or 3 papillose-pilose whitish and thickened hairs, these longer than the spikelet. Spikelets broadly ellipsoid, 2-2.8 mm long, 1- m wide, laterally com- pressed, usually gaping, stramineous or pur- plish. Lower glume ovate, acute, 3-nerved, 14-% the length of the spikelet, with a tuft of long, white, papillose-pilose hairs at the apex, otherwise scaberulous. Upper glume ovate, acute, 3—5-nerved, a little shorter than the lower lemma, finely ciliate at the margins, otherwise scaberulous. Lower lemma ca. 2 mm long, ca. ] mm wide, 5-nerved, finely ciliate at the margins, otherwise scaberulous, with 2-4 crateriform, ocellate glands; the in- ner surface pilose. Lower palea acute, finely pilose at the apex and scabrid at the keels; male flower present, anthers 0.8-1.1 mm long. Upper anthecium ovoid, 1.1-1.4 mm long, 0.4-0.7 mm wide, stramineous or dark brown at maturity, smooth, shining. Caryop- sis broadly ovoid, ca. 1.5 mm long, 1 mm wide, amber; hilum ovate, sub-basal; embryo ca. Y the length of the caryopsis. In flower July to September. Distribution. Brazil. Bahia, on rocks, in open and sunny habitats on “campo rupestre” at 1,000 m elevation. aratypes. BRAZIL. BAHIA: Serra da Jacobina, An- drade-Lima 70-6 159(IPA, SP); Morro do Chapeu, 1,000 m, E. Pereira 2138 (RB, US); Jacobinas, Serra do Brite, 11-095, 40°01'W, H. P. Bautista & R. P. Orlandi 1000 (HRB, US). The number of glands is generally constant for the same individual, two or four. Some- times there is a third pair of rudimentary glands. This species is related to P. chapa- dense Swallen, from which it differs mainly by having hairs on the pedicels, pilose lower glumes (with long hairs toward the apex), and smaller spikelets. None of the specimens was collected with its base, so it is not possible to know if cormlike bases as found in P. chapadense are present in the new species. We have the pleasure of naming this species in honor of the late Dr. Thomas R. Soder- strom, our friend, colleague, and renowned North American agrostologist. 12. Panicum stoloniferum Poiret, En- cycl. Meth. Suppl. 4: 272. 1816. TYPE: French Guiana. Cayenne: Cayenne, without collector (isotype, P, fragment at US). Figure 14. P. pu G. dus Prim. Fl. Esseq. 56. 1818. : Guiana: in graminosis umbrosis insulae Ar- euch Merer s.n. upra not located; frag- nt of an isotype at US (79732)). P. olyracfolium addi, Agrost. Bras. 43, pl. 1, TYPE: Brazil. Rio de Janeiro: in viciniis Rio panim Raddi s.n. (holotype, PI, not seen, frag- ment a B P. ctenodes Trin., Spec. Gram. 2: tab. 171. 1829. P. ctenodes var. major Trin., Spec. Gram. 2: tab. 171a. 1829. P. stoloniferum var. major (Trin.) Kunth, Distr. s Gra 389. 1831. Not Rev. Gram. l. tab. 18 P. Nds ha c ex Kunth, Enum. Pl. 1. 89. 1833. nom. nud. P. brachyclados C. Reichb. ex Trin., Mem. Acad. Imp. na. Cayenne: n 1835, Leprieur s. pe ea , P, fragment at US). Surinam. With- out locality: Kappler 1500 Wwe MO, US). P. . is Syn. Pl. Glum. 1: 6 54 P. umb alzm. ex Steudel, Syn. Pl. Ren 1: 65. 1854. nom. nud. Stoloniferous perennials. Culms genicu- late, long, decumbent, rooting and branching at the lower nodes, then becoming erect; erect portion of the culms 10-60 cm tall Lu 1 m tall according to herbarium labels); in- ternodes cylindric to compressed, hollow. branching at the middle and upper nodes, hispid in a longitudinal line to glabrescent, stramineous, sometimes purplish; nodes ob- scure, constricted, sparsely pilose to glabrous. Leaf sheaths splitting, striate, stramineous, Volume 75, Number 2 1988 Zuloaga & Sendulsky 449 Revision of Panicum sect. Stolonifera FIGURE 14. cemose branch.—e. S Panicum stoloniferum.— a. Habit.—b. Ligule.—c. Racemose branch showing pedicels.—d. Ra- »pikelet, ventral view.—f. Spikelet, dorsal view.—g. U, d pper anthecium, dorsal view.—h. Upper anthecium, ventral view.—i. Caryopsis, embryo side.—j. Caryopsis, hilum side. Based on Burkart et al. 26833 (US). shorter than the internodes, sparsely pilose to glabrous, the upper margins ciliate, the lower margins membranous. Ligule membra- nous, short, 0.2-0.4 mm long, laciniate at the apex; external ligule conspicuous, formed by a row of dense, whitish hairs. Leaf blades ovate-lanceolate to lanceolate, flat, 1.5-13 cm long, 0.3-3.5 cm wide, acuminate, con- tracted and asymmetrical basally, shortly pi- lose to minutely scabrid or glabrous on both surfaces (with long hairs toward the base or glabrous), the abaxial surface mostly purplish, 450 Annals of the Missouri Botanical Garden the midnerve prominent, the lateral nerves usually anastomosing; pseudopetiole dark, shortly pubescent. Panicles exserted, (1.5-) -9(-22) cm long, (0.8—)1.5—3(—6) cm wide, formed by numerous, dense, spikelike, alter- nate to subopposed racemose branches, + divergent from the axis, the spikelets borne in pairs, densely congested along the lower side of the branches; peduncles hispid to gla- brous; axis longitudinally ridged, hispid to sca- brous or glabrous; axis of the branches tri- quetrous (one side flat), scabrous to glabrous, the axils shortly and densely pilose; pedicels short, 0.5-1 mm long, pilose to glabrous. Spikelets lanceolate, 2.2-3.2 mm long, 0.5- 0.8 mm wide, glabrous, dark green, the upper lume shorter than the lower lemma, occa- sionally subequal. Lower glume ovate, 0.7— 1.3 mm long, glabrous, 1⁄4 the length of the spikelet, 3-nerved, the midnerve scabrous. Upper glume gibbous, acute, 1.9-2.8 mm long, 5(- 7)-nerved, the midnerve scabrous or glabrous apically. Lower lemma acuminate, 2.4-3 mm long, 5(-7)-nerved, the midnerve scabrous. Lower palea elliptic, 1.4-1.9 mm long, 0.3- m wide, brownish, glabrous, the margins finely ciliate to glabrous; male flower absent. Upper anthecium ellipsoid, 1.3-1.9 mm long, 0.4- m wide, stra- mineous, brownish at maturity; lemma 5- nerved; anthers brown, 0.5 mm long. Car- yopsis ellipsoid, 1-1.5 mm long, 0.5-0.8 mm wide; hilum oblong. In flower all year. Distribution. A widely distributed species found in Mesoamerica, Lesser Antilles, and South America, from Colombia to Argentina. The plants form dense and weedy ground vegetation in the moist and shaded forests at 150-1,400 m elevation. Common names. Capim-do-Mato, ca- pim-do-brejo (Smith et al., 1982). Representative specimens examined. ARGENTINA. CHACO: Isla Soto, Burkart et al. 30688 (RB, SI, US Schinini 16130 (MO, SI); Resistencia, Meyer 366 (SI). ENTRE RIOS: Isla Curuzu-Chalí, Burkart et al. 26833 (SI, US), 26851 (SI). CORRIENTES: 42 km E de Ituzaingó, Puesto de Prefectura, Zuloaga et al. 620 (SI); Isla Apipé rande, Puerto San Antonio, Krapovickas et al. 23850 (CTES, US). FORMOSA. Colonia Clorinda, Venturi 9164 — (US). MISIONES: Posadas, Ekman 623 (US); Eldorado, selva a orillas del Parana, Burkart 14528 (SI, US), brera et al. 28875 (SI, US); Puerto Rico, Cabrera et al. re 03 (SI); Campo Grande, Montes 10780 (SI); San tonio, iin 7063 (SI); San Juan, Montes 15308 (SD: Santa Ana, Montes 15278 (MO, SI); Arroyo Piray- Guazú, Cabrera et al. 28875 (SI); Santa Ana, Rodriguez 676 (F, SI, US); entre Pto. Aguirre y Pto. Iguazü, Votfhuzel & Van de Venne 37 (SI). SANTA FE: Puerto Piracuacito, Lewis 946 (SI, US). BELIZE. EL CAYO: Retiro, Lundell 63 14 (F, NY, US). TOLEDO: Upper Jacinto Creek, Gentle 5276 (US); beyond San Antonio, Gentle 7552 US). BOLIVIA. BENI: vicinity of Chacobo village Alto Ivón, Boom 4086 (US). COCHABAMBA: Antahuacana, Buchtien 2502 (MO, US). LA PAZ: Tipuani, Hacienda Casana, Buch- tien 7120 (MO, NY, US); between Coroico and Caranavi, Davidson 4788 (MO); Polo-Polo bei Coroico, Buchtien s.n. (MO, SI, US-711096), 264 (F); San Carlos, Buchtien 3 (US); Mapiri, Buchtien 1172, 1173 (US). SANTA CRUZ: Buena Vista, Steinbach 5130 (F, NY, US), 6855 (F, MO, US); Montero a Puerto Grether, Renvoize & Cope 3962 (MO). BRAZIL. — R Irwin et al. 6 (RB, US), erat (NY); Riozinho, km NW u Rd Gr xr ri & Souza 17604 (US). AMAZONAS: Cucuhi, Ri o, Baldwin 3252 (US); Esperanga, Pires & Blac k € east bank of Rio Madeira, 1 km N of Humaitá i US); vicinity of Tototovi, Pronet etal. US), 10282 (R, US); Rio Solimões por encima de la boca, Prance et tal. 2 pies Fróes 28028 (US); basin of the u di r Juruá, Fróes 5 (US). BAHIA: 22 km de la ro Fade via a -[tabuna, Moni 12844 (MO); we rda no km 13 da rodovia Valéncia-Guaibim, Carvalho & i o (CEPEC, MO); próximo ponte sobre Rio Mucuri, na rodovia BR- 101, Mori et al. 10537 (CEPEC, MO, RB); km 22 da antiga rod. Camaca/Itaimbé, Hage & Mattos Silva 304 (CEPEC); Mun. Ilheus, area do CEPEC, Hage & Brito 1395 (CEPEC), Santos 3399, 3787 (CEPEC). ESPÍRITO SANTO: Santa Bárbara e Caparao, Mexia 4099 (NY, Prance & Silva 59692. (F, MO Alto Turiagu, Nueva Esperanga, Jan (RB), 266 (MO, NY). MATO GROSSO: cine do Roncador, 55 km N de Barra do Garças, Prance & Silva 59442 (MO, NY). MATO GROSSO DO SUL: Dourados, Colónia Agrí- cola Federal, Swallen 9410 (US). PARÁ: Varadouro de Periquito a Pimental, Tapajó, Kuhlmann 1915 (RB, US); Rio San Manuel, entre Igarapé Prata a Igarapé Preto, Pires 3810 (US); a Black 47-2116 — = trada da Cachoeira Porteira, km 72, Cid et al. s July 1980 (MO); Belém, Pires & Black 599 (US), pus 8103 (F, US), Silva 24 (F); Moju River, Rubber Estate Fábrica, Goeldi 18 (F, US); Boa Vista, Rio Tapajós, Swallen 3198 (R, RB, US). PARANÁ: Garuva, Hatschbach 53 (BAA, SI); Ilha dos Bandeirantes, Rio Paraná, Lindeman & Haas 4368 (US). nio DE JANEIRO: Serra do Andarai, Rosa 95 (R); Parque Nacional da Tijuca, Serra dos Pretos Forros, Martinelli 3118 (R); Cachoeira de cau-Nova Fri riburgo, & Soderstrom 9061 (R). RIO GRANDE DO SUL: Finca Peixoto, Malme 1332 (GH); Sao Leopoldo, Dutra s.n. (US-1388850); Este, pr. Porto Alegre, Rambo 38269 (US). RORAIMA: vicinity of Aguaris, Volume 75, Number 2 1988 Zuloaga & Sendulsk y 451 Revision of Panicum sect. Stolonifera Prance et al. 9651 (F, US); Rio Jarani, Pires 14420 (US); Eee Maita and Paramiteri Indian village Prance et al. 10560 (US). SANTA CATARINA Pedro, Klein 11770 (US); Brago Joaquim, Luis — Reitz & Klein 2062 (US). PAULO: Cainu 7818 (R); Morro das peel Brade 7846 (R), ae je ipe d s.n. (US-1761194). COLOMBIA. AMAZONAS: Pue rino and vicinity, yon lower Rio Loretoyacu, Zarucchi 1067 (COL); Fleuve Amazona, 5 km en amont Gillett 16529 (COL, US); edge of Río Agua Branca about m W of Leticia, Trapecio, Schultes & Black 46- 337 (US). ANTIOQUIA: 15 km W of Chiborodó, Feddema 1968 (NY, US). chocó: bank of Rio San Juan, near Andagoya, Killip 35389 (COL); Hoya del Rio San Juan, Quebrada Cunperro, abajo de Noanamá, Forero et al. 4860 (COL); hoya del Río San Juan, arriba de Palestina, Forero et al. 4169 (COL, MO); Muqui, alto de Buey, Kjall von Sneidern a-26 (COL, MO); Unguia, Forero et al. 1988 (COL, MO); hoya del Río San Juan, Quebrada La Sierpe, Forero et al. 3973 (MO). MAGDALENa: Santa Marta, H. H. Smith 2126 (MO, NY). META: about 20 km SE of Villavicencio, Killip 34256 (COL, F, US); reserva de La Macarena, margen izquierda del Rio Duida, 20 km de su desembocadura, Pinto et al. 727, 773 (COL); serranía de La Macarena, orilla del Rio Sansa, F. ernandez El Coco, Cuatrecasas 21251 (F, US); Bio Calima, entre La Herradura de Or- doriez y Peria de Campo Triste, Cuatrecasas 16673 (F, US). vauPÉs: Rio Vaupés, above raudal Yurupari, Schultes & Cabrera 18723 (GH, US); Cano Curuyari, afluente izquierdo del Vaupés, selva y matorral en Zurubi, Cua- al S de Cahuita, Pohl & Pinette 13188 (F); La Bomba, Pohl & Davidse 11105 (F). PUNTARENAS: Golfo Dulce Area, vicinity of Esquinas Experimental Station, Allen 5299 (F, MO, US); 5 km SE of Rincón, Osa Peninsula, Pohl & Davidse 10711 (F, MO, US). SAN JOSÉ: mee is El General, Skutch 4816 (US). ECUADOR. GUAYA Domingo, Dodson 5817 : Tena and Arquidona, Asplund 9171 (R, US); Tena, pn plund 10306 (NY, US). PICHINCHA: Par Santo Mn dA eina ne Solís 13926 (US); 20 km W of San Dom e Los Colorados, Cazalet 5140 (NY, US) TUNGURAHUA: valle of Rio Pastaza 10058 (US). Gu ae ALTA VERAPAZ: Rio Santa Isabel, Steyermark 458 US). PETEN: s Tikal National Park, Tikal, Lundell p (US). di ice um near Piqui- (F, ) GUYANA: Yarikita, Soderstrom 2026 (NY 0 (US); oo ange Gleason 290 (GH, NY, US); i, Potaro River, Hitchcock 17410 (GH, NY, US); Kaituku mountains, in drainage of Takutu : Mato Sao River, A. C. Smith 3349 (GH, NY, US), 3423 (US); Barima River, Jenmann 7115 (US), Archer 2513 (US); vicinity of Issorora on Aruka River, Hitchcock 17568 (MO, US); Mazaruni Station, Archer 2431 (NY, US), Tutin 139 (GH, US). FRENCH GUIANA: C i Mori 8932 (NY); rece pene a Itany, Hoock 111 (US). HONDURAS. ATLANTIDA: Lance tilla, 10 km al SW de Tela, Nelson 5205 (MO); vichiky of Tela, Standley 55116 (US). GRACIAS A DIOS: alrede- dores del Río Plátano, T 4019 (MO). MEXICO. CHIA- PAS: Libertad, Matuda 18138 (F, US); Escuintla, Matuda Dwyer 2108 (MO); cercanias de la pu Chiriqui, von Weddell 2575 (GH, MO). CHIRIQUÍ: near Puerto Ar- muellos, Woodson Jr. 858 (MO). COLON: Trinidad basin, near Cirri River, Pittier 4027 (NY, pu DARIÉN: vicinity of Paya, Río Paya, Stearn et al. 440 (MO); vicinity of Campamento Buenavista, Río Chucunaque above conflu- ence with Río Tuquesa, Stearn et al. 827 (MO, US), 957 (MO). PANAMÁ: east of the Rio irem Standley 26682 (US). PARAGUAY: entre el Rio y el Rio Aquidabán, Fiebrig 4706 (F); near Tobati, pura 4843 (US); Alto u 3540 (MO, US). Without rias d Weddell 3152 (F, NY). PERU. AMAZONAS: left ban o Mararion, above Cas- cadas de Mayasi, Wurdack 1976 (NY, US). HUANUCO: Tingo Maria, Asplund 13210 (NY, US), Allard 21662 (US); 6 km S of Tingo Maria, Seibert 2258 (MO, US), Storp 9479 (F). JUNÍN: Colonia Perené, Hitchcock 22058 (US); Chanchamayo Valley, Schuncke 129 (F, US); bajo Rio Nanay, Williams 189 (F); above Pongo de Manseri- che, right bank of mouth of Rio Santiago, Mexia 6151 F, US); La Merced, Hacienda Schuncke, McBride 5646 F, US). LORETO: Río Hueppi, Gentry et al. 21852 (MO); Rio Itaya, Diaz et al. 653 (MO); Cano Iricahua, abajo de Jenaro Herrera, Encarnación 25080, 25087 (US); Caño Supai, Encarnación 25056 (US); lower Rio Nanay, Williams 589 (US). SAN MARTÍN: Quebrada de Almendras, Schunke Vigo 4461 (F, US); A E CE Williams 5554 F); Juan Jui, Klug 3813 (F, GH, MO, US); San Roque, Williams 7521 (F, US). SURINAM: Kayselberg airstrip, Cramer 2980 (NY); Oelemari, Wessels Boer 926 (GH); aa 25 (U opposite Gansee, va — — ^ A 4 (US); El Tucuche Mountain, Soder- strom 1048 (US). VENEZUELA. T. F. AMAZONAS: Isla Se- bastián, Río Casiquiare, Liesner & Clark 8945 (MO); Sierra Parima, Steyermark 107022 (MO, NY), Cardona 1346, 1478 (US); 5 km E of San Fernando de Atabapo, Davidse et al. 17164 (M uayapo, Bajo Caura, Williams 11999 (F, US); Rio PURA. between Guaiquinima and Rio To rono, Killip 37425 (NY, US), 37527 (US), 37480 (NY, US); Caño Pablo, Liesner & Morillo 13943 (MO); El Dorado, Curet 213 (US); Salto de Chalimano en el Rio Paramichi, Steyermark 90706 (US); selva al lado del Río Nichare, Steyermark 95673 (MO, NY, US) DELTA Annals of the Missouri Botanical Garden AMACURA: 33 km al E de El Palmar, Steyermark 93096 (US) a lo largo del Cano Araguao, Steyermark et al. 114792 (MO); 73 km al SE de Piacoa, Davidse & Gon- zález 16466 (MO). LARA: near Barquimiseto, Saer 284 (NY, US). MIRANDA: along Rio Grande del Tuy, above Paparo, Pittier 6328 (US). MoNAGas: 1.5 km N of La Hormiga, Wurdack & Monachino 39527 (RB, NY, US); 2 km N of Santa Ines, Pursell et al. 9167 (US). sucRE: Península de Paria, entre Los Pocitos de Santa Isabel y Roma, Dumond et al. 7659 (NY). ZULIA: 3 km E of Rio de Oro, Davidse et al. 18784 (MO); alrededores de Ca- sigua El Cubo, Bunting 7815 (MO); intersección del Rio Catatumbo y la ruta entre Maracaibo y La Fría, Davidse et al. 18838 (MO); Quebrada Tayaya, Davidse et al. 18493 (MO). Hitchcock & Chase (1910, 1915) distin- guished P. stoloniferum from P. frondescens by the sizes of the plants and the panicles and by the length of the upper glume compared with the lower lemma. Abundant material showed a complete gradation in these char- acters. Therefore, we are treating P. fron- descens as a synonym of P. stoloniferum. 13. Panicum venezuelae Hackel, Oes- terr. Bot. Z. 51: 368. 1901. Brachiaria venezuelae (Hackel) Henrard, Blumea 3: 435. 1940. TYPE: Venezuela. Without locality: Eggers 13471 (holotype, W, not seen, fragment at US). Figure 15. P. ineptum A. Pa & Chase, Contr. U.S. Natl. Herb. 7: 509 P. berteronianum M Bot. Jahrb. Syst. 2 Beibl. 125: . 1921. Not P. berteronianum ultes, 1854. Venezuela. Federal District: li Guayra, Zoll- .n. (holotype, B, not seen, fragment at US) TYPE: ner s Stoloniferous, densely and freely branching perennials, with very long, slender, prostrate stolons. Culms decumbent to ascending, densely branching mostly at the lower nodes, 40-80 cm tall; internodes 3-8 cm long, cy- lindric to compressed, hollow, sparsely pilose to glabrous; nodes densely pilose, with long whitish, usually retrorse hairs. Leaf sheaths 1-4 cm long, usually shorter than the inter- nodes, stramineous, striate, sparsely to dense- ly pilose with long, whitish hairs, the margins ciliate. Ligule membranous-ciliate, ca. 0. mm long, with long hairs toward the back at the base of the blade; collar stramineous, pi- lose. Leaf blades ovate-lanceolate to lanceo- late, acuminate, flat, 3-9 cm long, 0.5-1 cm wide, cordate basally, densely to sparsely hir- sute; the margins long-ciliate toward the base, otherwise ciliate to scabrous, cartilaginous; midnerve inconspicuous; pseudopetiole short, ca. 0.2 cm long. Panicles lax, long-exserted, 2-11 cm long, 1-3 cm wide, formed by 4- 10 distant, alternate, short branches, these divergent from the axis and racemose, with cleistogamous spikelets in pairs arranged along either side of a ventral septum; chasmoga- mous spikelets occasionally present; axis lon- gitudinally ridged, long-hirsute, the axils of the branches densely pilose, axis of the branches triquetrous, densely hirsute; pedi- cels short. Axillary panicles similar to the terminal panicle, but short-exserted and few- flowered. Spikelets ellipsoid, 2.5-3 mm long, 1.1-1.3 mm wide, stramineous to greenish, the glumes and lower lemma sparsely to densely hirsute. Lower glume ovate, 1.5-1.7 mm long, 2 or more the length of the spikelet, 3-nerved, the midnerve scabrous. Upper glume 2.6-2.9 mm long, gibbous, 5(-7)- nerved, with thick, papillose hairs over the entire surface or only toward the apex, 2 glands occasionally present on the middle por- tion of the outer surface, the inner surface scabrous. Lower lemma larger than the upper glume, the apex + inflated, with a few thick hairs, the middle portion flattened and gla- brous, the margins inrolled toward the apex, 5-nerved, the lateral nerves remote from the midnerve, 2 glands sometimes present on the middle portion of the outer surface. Lower palea elliptic to obovate, 1.6-1.8 mm long, 0.7 mm wide, hyaline, shortly pilose toward the upper margins, glabrous over the rest of the surface; male flower absent. Upper an- thecium ellipsoid to obovoid, obtuse, 1.6-1.8 mm long, 0.8-1.1 mm wide, stramineous; lemma 5-nerved, strongly convex; anthers 3, those of the cleistogamous spikelets small, 0.2-0.3 mm long; anthers of the chasmog- amous spikelets 0.9 mm long. Caryopsis 1.3 mm long, 1 mm wide; hilum punctiform. In flower all year. Distribution. This species occurs in Guatemala, Honduras, Cuba, Haiti, the Do- minican Republic, and Venezuela to northern Volume 75, Number 2 Zuloaga & Sendulsky 453 1988 Revision of Panicum sect. Stolonifera FI > m —c. Racemose branch.—d. Spikelet, ventral view. — FIGURE 15. Panicum venezuelae. — a. Habit. —b. Ligule ) —h. Spikelet, e. Spikelet, dorsal view.—f. Lower lenma.—g. Spikelet, ventral view, lower lemma with glands Brazil in humid and shady places at 100- Bamia: Alagoinhas, Chase 8135 (US); camino de Santa : Iné 1,200 m elevation. 6 km SŠ of Cocos, Anderson et al. 37036 (F, MO); Serra do Itiuba, 6 km E of Itiuba, Harley et al. 16203 (MO, Representative specimens examined. BRAZIL. ALA- US); Santa Terezinha, Bondar 2609 (SP, US); Paragua- Goas: Tapera, Pickel s.n. (US-1645543), 2469 (US). cú-valley, Muritiba, Pinto 307 (US); Feira de Santana, 454 Annals of the Missouri Botanical Garden Chase 8066 (F, RB, US), 8070 (US); Cachoeira, vale dos rios Paraguaqü Harley 21497 (CEPEC); rodovia Sta. Inés a Rio aos 10 km, Pinheiro 1855 (CEPEC, US). cEARÁ: Serra e Ceu, Eugenio 278 (RB). P 8 (RB); Areia, Coelho de Moraes TE va Cru Montanhas, Swallen 4813 (RB, US). CUBA. ORIENTE: Barbi, Loma del Gato, Sierra Maes- tra, Ekman 15661 (US), Loma del Gato, S of Loma San Juán, Leon et al. 10190 (US). GUATEMALA. GUATEMALA: near Fiscal, Standley 59580, 80411, 80463 (F, US), 80630 (F). HONDURAS. EL PARAÍSO: Quebrada de El Muro, between Las Mesas and Yuscarán, Standley 29263 (US); road to Yuscarán, Swallen 11329, 11333 (US). MORAZÁN: near El Jicarito, Standley 20874, 21640 (F), 27498 (US), Swallen 11377 (US); trail from La Quince, El Zamorano, Standley 14567, 21272 (US); campus of Escuela Agricola El Zamorano, Pohl & Davidse 12458 (MO); 8 km S of La Venta by road, Davidse & Pohl 2155a(MO).S DOMINGO. MONTE CRISTI: near Arroyo Seco, Ekman 12608, 13085 (US); Puerto Plata, Baja- bonico, Ekman 14499 (US). SANTIAGO: San José de las Matas, Ekman 14602 (US), Jiménez 950 (US). VENEZUELA. DISTRITO FEDERAL: Colinas de Yaguara mayo 1449 (F); Antimano, Pittier 12581 (US); entre Caracas y La Guayra, Burkart 17013 (SI). FALCÓN: Ser- rania de San Luis, Fila Las Playitas, Ruiz 2543 (MO) LARA: en cerros arriba del caserio Simara, Burandt Jr. v0192 (MO); Loma de León, Iribarre, Tamayo 3743 (MO); Hacienda Sosa, Badillo s.n. (US-1760677). Henrard (1940) transferred P. venezuelae to Brachiaria without explanation. Previ- ously, when describing P. ineptum, Hitch- cock & Chase (1915) did not refer it to any of the groups of Panicum. Brown (1977) pointed out that this species, being a C, plant, is not actually a Brachiaria, since the genus Brachiaria is totally C, or Kranz. Sendulsky (1978), in her treatment of Brachiaria for Brazil, called attention to the size of the an- thers and to the unusual form, for Brachiar- ia, of the long-winged lodicules. She suggested retention of this species in Panicum. he panicle, habit, and spikelets (which bear glands on the lower lemma and occa- sionally on the upper glume also) suggest in- clusion of P. venezuelae in sect. Stolonifera. However, this is the only species in the section with cleistogamous spikelets, a feature oth- erwise only found in Panicum in species of subg. Dichanthelium A. Hitchc. & Chase. Also, the glands differ from those in other species of sect. Stolonifera; in P. venezuelae the glands are bigger and depressed, not cra- teriform. LITERATURE CITED Brown, W. V. 1977. The Kranz syndrome and its Mun in grass systematics. Mem. Torrey Bot. Club 23: 1-97, DAVIDSE, 7 & R. W. Pour. 1974. Chromosome num- bers, meiotic behavior, and notes on tropical Amer- ican grasses (Gramineae). Canad. J. Bot. 52: 317- 328 DoELL, J. C. 1877. Tribe 3, Paniceae. In: C. F. P. von Martius (editor), F Brasiliensis 2(2): 33-343, pl. 12-49 (fascicle 72). GouLp, F. W. € R. a 1983. Grass Pi mg T 2nd edition. Texas A&M Univ. Press, College Sta tion, Texas. & T. R. SODERSTROM. 1970. Chromosome of some corer a Colombian grasses. . J. Bot. 48: 1633- HERIARD. J. TH. 1940. up on ds nomenclature of some grasses. Blumea 3: - HITCHCOCK, A. S. & A. CHASE. ican species of Panicum. Contr. U.S. Na 15: 1-396. num ue 411-480. 1910. The North Amer- tl. Herb. — — —. 1915. Tropical North American species of Panicum. Contr. U.S. Natl. Herb. 17: 5. The classification of Panicum (Gra- its allies, with a reference to the Sg ei of lodicule, style base and lemma ac Sci okyo, Sect. 3, "nl 9: 43-150 Lazanipes, M. & R. D. WE akirra (Pan- - D oaceae), a new genus for Australia. Brunonia 6. NS ds V. 1903. Gramineae, Pp. 48-160 in J. K. Small, Flora of the Southeastern D States. C. ash, New York. PILCER, R. Gramineae. /n: A. Engler, Die Na- türlichen Pflanzenfamilien, 2nd edition, 14e: 8-25. Pour, R. W. 1 ramineae. /n: W. Burger (editor), Flora Costaricensis. F LN Bot. 4: 1-6 & G. DavipsE. 197 romosome e Costa Rican grasses. 5 23: 293-324 SENDULSKY, T. 1978. Brachiaria: taxonomy of culti- ed and native species in Brazil. Hoehnea 7: 99- 39. SMITH, L. B., D. C. WASSHAUSEN & R. M. KLEIN. 1982. Canines Pp. 633-756 in R. Reitz (editor) Flora Ilustrada Catarinense, Part I. Itajai, Brazil SODERSTROM, T. R. . Sucrea (Poaceae M S soideae), a new genus from Brazil. Brittonia 83: 198- 10. 2 SWALLEN, J.R. 1966. The Latissima Group of Panicum. Species Graminum Iconibus et Descriptionibus Ilustravit, Volume 1. Impensis Aca- demiae Imperialis Scientarium, Petropoli. . 18 i i criptionibus lustravit, Volume 2. Impensis Acade- miae Scientarium, Petropoli. . 1835. Panicearum genera retractavit spe- Volume 75, Number 2 1988 Zuloaga & Sendulsky Revision of Panicum sect. Stolonifera 455 ciesbusque ‘ais sadhana U ustrawit, Mem. Acad. Imp. Sci. Saint Petersbourg VI, 9-355. 1836 inis raminum Iconib criptionibus ‘Tlustravit, Volume 3. Impensis ade e Imperialis Scientarium, Petro 1987 poli. Hane, A, F. O. Systematics of the New World atics and Evolution. Smithsonian Institution Press, Washington, D APPENDIX I List of taxa and informal groups. Species listed in italics are accepted. Those listed in roman are not accepted rius = (Hackel) Henrard Dichantheliu Eriochloa iani ada) Kunth ymenachne leptostachya (J. S. Presl) Fourn. : A nasus pia gracile Brongn. Lorea, sect. egista, Molinia caerulea (L.) Moench ect. d e Enosseulitun Swallen e bertero E Seribner & Smith bipustulatum L iepharophorom M blepharophorum J; s. Presl brachystachyum Trin bresolinii L. B. Smith & Wassh. b ulbosum Kunt TUTTTITTTTTTET o chapadense Swallen crateriferum Sohns ctenodes Trin ctenodes var. major Trin. frondescens G. Meyer paucifolium Swallen piauiense len pilosum Swartz pirineosense Swallen quadriglume (Doell) Henrard rude secundum Trin secundum var. inaequiglume Doell secundum var. subaequiglume Doell . semitectum Swallen . soderstromii Zuloaga & Sendulsky . stoloniferum Poiret . umbrosum Salzm. ex Steudel e SU Are a BS foviv iv iid ui IU Sv IU DUI IU rB tw uy a be] Em RSS = Ë [^7] e a Lnd Phanopyrum, subg. Stolonifera, sect. Sucrea sampaiana (A. Hitchc.) Soderstrom Verrucosa, sect. A REVISED TREATMENT OF James S. Miller: BORAGINACEAE FOR PANAMA! ABSTRACT The Boraginaceae known from Panama now number 52 species compared with the 33 that were recognized the last time the family was treated. Eight of these are described as new species, five in the genus Cordia and three in Tournefortia. 4mong the new records for Panama is Moritzia lindenii, the genus being previously unknown in the country. Most genera of Panamanian Boraginaceae are South American in origin. More than half of the species are widespread in the Neotropics, but the remainder show stronger affinities with the rest of Central America than with South America or with the West Indies. The Panamanian species of Boraginaceae American countries have altered taxonomic have been treated numerous times since the concepts of some Panamanian species. Ex- turn of the century. Standley’s floras of the amination of collections while preparing a re- Canal Zone (1928) and Barro Colorado Island vision of Cordia for Mexico and Central (1933) predate Ivan Johnston’s studies of the America (Miller, 1985) and of floristic treat- family, and many of the names Standley used ments for Nicaragua (Miller, in press) and were later placed in synonymy by Johnston, Mesoamerica (Miller, in prep.) revealed nine who published extensively on the family from previously unrecorded species as well as eight the 1920s to the 1950s. Johnston's book The undescribed species for Panama, a significant Botany of San José Island (1949b) treated increase from 33 to 52 species. another island flora, but unfortunately all of Five subfamilies are currently recognized the works up to this time essentially describe within the Boraginaceae (Johnston, 1951; the flora of the Canal Area. Thomas B. Croat’s Cronquist, 1981): Cordioideae, Ehretioideae, Flora of Barro Colorado Island (1978) pro- Heliotropioideae, and Boraginoideae, all of vides an excellent treatment of the species which have representatives in Panama, and on the island. the Wellstedioideae, which consists of two The Boraginaceae of Panama were treated African species. The Cordioideae and Ehre- for the entire country by Nowicke (1969) in tioideae have sometimes been treated togeth- the Flora of Panama. Subsequently, the er as a separate family, Ehretiaceae (Lindley, number of collections from Panama in- 1830; Airy Shaw, 1973; Hutchinson, 1973). creased, especially from regions then poorly They are, however. clearly related to the oth- known. Floristic studies in other Central er three subfamilies and are tied to them by ' IV. G. D'Arcy suggested this project and provided encouragement throughout. C. W. Hamilton and K. Krager helped during my visit to Panama. J. K. Myers provided the illustrations, J. D. Dwyer helped with the Latin descriptions, M. J. Hufi sent information regarding collections at the Field Museum, $t . G. D'Arcy, P. P. Lowry Il, and W. D. Stevens read portions of the manuscript. G. D. McPherson and J. W. Nowicke provided valuable review comments and l thank them for their help. The staff of the Missouri Botanical Garden assisted in many ways and their help is greatly appreciated. | also thank the curators of the following herbaria for lending specimens and for hospitality: A, AAU, ARIZ, ASU, B, BM, BR, C, CAS, CFMR, CHAPA, CR, DS, DUKE, ENCB, F, FHO, G, GB, GH, HAL, HNMN, IND, INIF, K, L, LE, LL, MEXU, MICH, NMC, NY, P, PMA, POM, RSA, SCZ, SD, SLPM, TEX, U, UC, d USJ, WIS. My wife, Leslie, provided encouragement and accompanied me during field studies in Pana ? Missouri Botanie al Garden, P.O. Box 299. Se, Louis, Missouri 63166, U.S. A. ANN. Missouni Bor. GARD. 75: 456-521. 1988. Volume 75, Number 2 1988 Miller 457 Revision of Panamanian Boraginaceae intermediate genera. Recent authors (e.g., Cronquist, 1981) have generally accepted their inclusion in a broadly defined Boragi- naceae. Recognition of one family with five subfamilies also provides more information about relationships. PHYTOGEOGRAPHIC RELATIONSHIPS OF PANAMANIAN BORAGINACEAE The Boraginaceae of Panama make up an assemblage of species derived from different regions. In order to assess their phytogeo- graphic relationships, the distributions of the constituent genera and species were deter- mined. The genera Borago and Cynoglossum were excluded from the generic analysis, as both are not indigenous. The remaining seven genera are listed in Table 1 with their pre- sumed centers of origin. Four of the seven genera are clearly South American in origin. All three species of Mo- ritzia occur in South America, with only one ranging north as far as Costa Rica. Although Cordia, Heliotropium, and Tournefortia are pantropical in distribution, they are strongly centered in South America. Bourreria and Hackelia clearly originated in North America. The majority of the species of Bourreria occur in Mexico, with an ad- ditional group centered in the Greater An- tilles; only a few species range south to south- ern Central America and South America. Hackelia is distributed widely in north tem- perate regions with the greatest concentration of species in the western United States (Gen- try & Carr, 1976) Ehretia 1s more problematic but certainly seems to be of Old World origin. Most of its species occur in Africa but it is well repre- sented in tropical and subtropical Asia. Whether or not this strong African represen- tation indicates that Ehretia originated there or radiated there as the climate became drier at the end of the Oligocene (Raven & Axelrod, 1974) is not clear. Regardless, Ehretia is poorly represented in the New World, with only three species, and the southern limit of distribution is in Chiriqui Province of western Panama. The presence of Ehretia in Panama, then, probably results from southward migra- tion. From this it seems clear that the generic affinities of Panamanian Boraginaceae are predominantly South American. Four of the seven genera that account for all but four of the native Panamanian Boraginaceae are cen- tered in South America. Hackelia and Bour- reria are North American in origin and, at least in the New World, so is Ehretia. The strong South American tie agrees with other analyses at familial or generic levels. Gentry (1985) reported that 84% of the plant species in Panama were members of Gondwanaland families, and Karr (1985) reported that at least 50% of the bird species in Panama were members of families that were South Amer- ican in origin. Distributions of Panamanian species of Bo- raginaceae are summarized in Table 2 follow- ing Davidse's (1985) format. Distributional data were derived from herbarium specimens and literature (Gentry & Carr, 1976; Gibson, 1970; Johnston, 1924, 1927, 1928, 1930, 1935, 1940, 1949a, c, 1950; Miller, 1985, in press; Nash & Moreno, 1981; Nowicke, 1969). Each species was scored for presence in North America, Central America, South America, and the West Indies. In addition, presence in the Old World or cultivated status was noted. INTRODUCED SPECIES Two introduced species of Boraginaceae are known from Panama. Borago officinalis L. is widely cultivated throughout the world and occasionally is found naturalized, al- though these populations do not appear to persist for long. Cynoglossum amabile Stapf & J rumm. is a native of China originally imported as an ornamental and has become established in numerous localities at high el- evations in the Neotropics. COSMOPOLITAN SPECIES Two species, Heliotropium curassavicum L. and H. indicum L., are widespread in the 458 Annals of the Missouri Botanical Garden TABLE l. Native genera of Panamanian Boragi- vón ex A. DC. and C. megalantha S. F. Blake naceae and presumed centers of origin. Num f % of Genus Species Pana- (Number of i manian Center of Species) Panama Species Origin Bourreria (50) 2 4 North America Cordia (300+) 27 54 South America Ehretia (50) 1 2 Paleotropics Hackelia (45) 1 2 orth Ámerica ra (150) 4 8 South America Mori 1 2 South America o (150) 14 28 South America New World and have become widespread in the Old World as introduced weeds (Nowicke & Miller, in press). NORTHERN SPECIES Seven species of borages reach their south- ern limits in Panama. A single species each of Bourreria and Ehretia, both genera prob- ably northern in origin, are included in this group. Five Central American species of Cor- dia make up the remainder of this group, of which four are Central American species of lowland wet forests. Cordia diversifolia Pa- range north to Veracruz in southern Mexico; C. lucidula I. M. Johnston and C. porcata Nowicke range only as far north as Nicara- gua. These four belong to Central American species ipe The fifth, Cordia inermis (Miller) I. M. Johnston, reaches from Panama north to Sinaloa, Mexico in dry disturbed areas. It is a member of a group of about eight species that is widespread in the Neo- tropics of which Cordia foliosa Martens & Galeotti is the only other species restricted to Central America. NORTHERN- SOUTHERN SPECIES Twenty-seven species, comprising slightly more than half of the Boraginaceae known from Panama, range widely in the Neotropics. With two exceptions, all of these are truly widespread species that indicate no particular geographic affinity. The genus Moritzia con- sists of three species and is entirely restricted to South America except for M. lindenii (A. DC.) Gürke ex Benth., which ranges north to Costa Rica. Hackelia is a northern genus, but H. mexicana (Schldl. & Cham.) I. M. Johnston is widespread in Central America TABLE 2. Distributions of Panamanian Boraginaceae. Number of Number of % of Category! Species Combined Categories Species Flora 1. Cultivated 1 2. Alien Weeds l | nini A b 3. Cosmopolitan2 2 Cosmopolitan 2 4 4. CA & WI & NA 0 5. CA & WI 0 6. CA & NA 0 Northern 7 13 7. WI & NA 0 8. CA 7 e Al & WI o | Southern 0 0 11. CA & SA & WI & NA 3 = EN x SA 7 l : Southern- Northern Combination 27 52 14. CA & SA 13 15. Panama 7 16. Panama & Costa Rica 4 Endemic Element 14 27 17. Panama & Colombia 3 Abbreviations: CA = Subtropical Mexico); SA Ce ntral America (Costa Rica-Tropical Mexico); NA — = South America; WI = West Indi North America (United States to les. ? Includes all native aia species distributed on at least one Old World continent. Volume 75, Number 2 1988 Miller 459 Revision of Panamanian Boraginaceae TABLE 3. Endemic elements among the Boraginaceae of Panama and distributions. Species Distribution! Elevation (m) Cordia anisophylla CN, DA, PA, SB 0-1,000 Cordia correae CC, PA, VE 800-1,000 Cordia croatii CC, VE, Costa Rica 800-1,200 Cordia lasiocalyx BT, CA, CC, DA, PA 0-800 Cordia leslieae PA 800- 1,000 Cordia protracta SB, Colombia sea level Cordia tacarcunensis DA, Colombia 100 Tournefortia brenesii VE, Costa Rica 800-1,350 Tournefortia johnstonii CH, VE, Costa Rica 1,000-1,300 Tournefortia longispica BT, CH, CC, VE 600-1,500 Tournefortia multiflora CN, VE 400-900 Tournefortia ramonensis BT, CH, Costa Rica 2,000-3,000 Tournefortia tacarcunensis DA, Colombia 1,500 Tournefortia urceolata CH, CN, SB 400-2,300 ! Abbreviations for provinces: BT = Bocas del Toro; puta d Area; CH = Chiriqui; CC = Coclé; CN = Colón; Pa DA = Darién; PA = Panama; SB = San Blas; VE = Ver and the Andes. Cordia and Tournefortia con- tain 22 of the widespread species, all of which are found in most of Central America, but only about one-third of them range as far south as southern Brazil and Argentina. The majority have southern limits of distribution in northern South America or extend south only in the Andes. ENDEMICS Fourteen species of Panamanian borages, all species of Cordia and Tournefortia, are known only from Panama and adjacent Co- lombia or Costa Rica. Seven are known only from Panama (Table 3); three range slightly into Colombia, and four others extend into Costa Rica. All are relatively rarely collected. Eight are described as new in this paper. Al of the species of Tournefortia and three of Cordia occur at mid to high elevations, while the widespread taxa generally inhabit low- ands. More than half of the Panamanian species of Boraginaceae are widespread or introduced and are not helpful in indicating geographic affinity. The pattern of the lowland Pana- manian flora being composed primarily of South American elements has been reported for other groups (Davidse, 1985; Raven & Axelrod, 1974, 1975). Hammel (1986) found a similar pattern for a subset of the flora of — La Selva in lowland Costa Rica. The fourteen endemic elements indicate that the Pana- manian flora is old enough to have become distinct from that of surrounding areas. The relationships of the endemics are poorly understood but the majority are probably with species of Colombia and Ecuador. Seven species have distributions extending to the north, and many are members of species com- plexes restricted to Mexico and Central Amer- ica. These species, and the absence of south- ern elements, indicate that Panamanian borages have a stronger relationship at the species level with Central America than with South America. Although most of the species belong to originally South American genera, a significant portion of these seem to have reached Panama from the north, perhaps as a result of secondary radiations of Cordia and Tournefortia in Mexico and northern Central America. SYSTEMATIC TREATMENT Boraginaceae Juss., Gen. Pl. 128. 1789. Trees, shrubs, lianas, vines, or herbs, often conspicuously pubescent, the hairs often with a basal cystolith. Leaves estipulate, simple, alternate or rarely opposite. Inflorescence cy- mose to paniculate, the branches often scor- pioid, helicoid, or reduced and capitate to 460 Annals of the Missouri Botanical Garden glomerate. Flowers perfect or imperfect, usu- ally 5-merous; calyx usually persistent, tu- bular to campanulate, usually 5-lobed; corolla gamopetalous, usually 5-lobed; stamens usu- ally as many as the corolla lobes and alternate with them; ovary superior, 2-carpellate but often becoming falsely 4-locular; ovules usu- ally 4, anatropous; style 1, terminal or gy- nobasic, simple or branched; stigmas 1-4. Fruits drupaceous and l-4-seeded, some- times dry at maturity, or of 4 nutlets. KEY TO THE GENERA OF BORAGINACEAE IN PANAMA The Boraginaceae are worldwide in distri- bution and comprise about 100 genera with approximately 2,000 species. Nine genera are known from Panama containing 52 species. Although no collections of the genus Borago are known from Panama, the Eu- ropean species Borago officinalis L. is often cultivated in gardens in other parts of Tropical America and is included here as it undoubt- edly occurs in Panama. la. Plants trees or shrubs; stigmas 2 or x fruits fleshy at least when young. 2a. Style twice divided into 4 cia ... Cordia 2b. Style once divided into 2 s 3a. Er lobes valvate; nace Heshy, longer than 8 mm x lobes imbricate or open in bud; corolla thin, shorter Bourreria Ehretia than 5 lb. Plants E ane vines, or clambering shrubs, rarely small trees; stigma p fruits dry except in Tour- nefortia. 4a. Fruits entire to shallowly 4-lobed; style terminal; corolla white, green, or yellow-green, rarely purple 5a Plants woody; fruits fleshy To urnefortia 5b. Plants herbaceous Heliotropium EN c , diss deeply 4- e gynobasic; corolla blue. E Calyx campanulate ; fruits dry obed, consisting of 4 separate nutlets or the nutlets solitary by abortion in Moritzia; 7a. Nutlets with plochidiate spines; calyx lobes less than 4 mm lon 8a. Cauline leaves clasping at base; nutlets spreading, the spines ca. 0.5 mm long „n ynoglossum 8b. e jun cuneate to decurrent but not clasping at base; nutlets erect, the spines 1-4 mm lon Hackelia 7b. Nutlets Tae bcn shallowly ribbed; calyx lobes more than 10 mm long coco. Borago 6b. Calyx cylindrica Moritzia Borago L., Sp. Pl. 137. 1753; Gen. Pl. ed. 5. 67. 1754. TYPE: Borago officinalis La Sp. PL. 137. 1753. Annual or perennial herbs, the stems hir- sute. Leaves alternate, the basal ones petio- late, the cauline ones sessile. Inflorescence a corymbose group of racemes, bracteate. Flowers bisexual; calyx with 5 lobes, these free to nearly the base; corolla broadly cam- panulate to rotate, the 5 lobes imbricate, the tube short, appendaged in the throat; stamens 5, exserted, the filaments broad, the anthers linear; ovary 4-lobed, the ovules 4, the style gynobasic, filiform, the stigma emarginate. Nutlets 4, obovoid or oblong, the gynobase flat or nearly so. Borago comprises three species from southern Europe and the Mediterranean re- gion, one of which, Borago officinalis, is widely cultivated. Borago officinalis L., Sp. Pl. 137. 1753. TYPE: without locality or collector (ho- lotype, LINN (Savage Catalog number 188.1), not seen; microfiche, MO). Annual herb, 30-60 cm tall, the stems coarsely hirsute. Basal leaves petiolate, ob- ovate to oblong, 6-12 cm long, 2-6 cm wide, the apex acute to obtuse, the base cuneate and decurrent along the petiole, the margin entire to irregularly undulate, the adaxial sur- face hirsute to scabrous, the veins prominent, the lower surface pubescent with most of the hairs restricted to the veins, the uppermost leaves sessile, lanceolate. Inflorescence loose- ly racemose, bracteate, the rachis hirsute. Flowers borne on pedicels 1-5 cm long; calyx rotate, with 5 lanceolate lobes to 10-18 mm Volume 75, Number 2 1988 Miller 461 Revision of Panamanian Boraginaceae long, 2-3 mm wide, hirsute; corolla blue, yellow in the throat, rotate, 18-22 mm long, the 5 lobes ovate to lanceolate, 8-11 mm long, the tube to 3 mm long; stamens 5, the filaments to 2 mm long, broad, the anthers lanceoloid, 5-7 mm long, with an appendage to 3 mm long at the base; ovary ca. 2 mm broad, the 4 lobes globose, the style 5.5-7.5 mm long. Fruits with the calyx and style per- sistent, the 4 nutlets obovoid, 4-6 mm long, 2-3 mm broad, finely ribbed, tuberculate at the apex. Distribution. Borago officinalis is na- tive in Europe, north Africa, and adjacent Asia but is widely cultivated and occasionally becomes naturalized. Although Borago officinalis has not been collected in Panama, it seems almost certain that it is present in Panamanian gardens. It is often cultivated as a culinary or medicinal herb and is known from most neotropical countries, often as an adventive. Bourreria P. Browne, Civ. Nat. Hist. Ja- maica. 168. 1756; nom. cons. TYPE: Bourreria succulenta Jacq., Enum. Syst. Pl. 14. 1760; Select. Stirp. Amer. Hist. 44. 1763. Trees or shrubs. Leaves alternate, petio- late, the margin usually entire. Inflorescences terminal cymes. Flowers bisexual, actino- morphic; calyx campanulate, 2-5-merous, the lobes valvate in bud; corolla relatively large, salverform, white to yellow in the Central American species, usually 5-merous, the lobes imbricate in bud; stamens 5, the filaments adnate to the base of the tube, the anthers ovate to oblong; ovary 4-locular, the style terminal, bifid; stigmas 2, flattened. Fruits drupaceous, enclosing 4 bony nutlets, the en- dosperm carnose, the cotyledons flat. Bourreria has been considered to comprise about 50 species (Airy Shaw, 1973); how- ever, Gibson's (1970) estimate of 15-20 species is probably more realistic. This poorly understood genus needs revision. Numerous species have been published based upon minor variations in leaf shape and indument—res- olution of the problems this has created will necessitate field study of these characters. Bourreria ranges from Mexico and southern Florida through the West Indies and Central America to northern South America, with the majority of the species occurring in Mexico and the West Indies. About seven species are known from Central America, although only two have been found in Panama. Despite con- siderable confusion about delimitation of species, the two known from Panama are among the most distinct and easily recognized members of the genus. De Candolle (1845) placed the species cur- rently recognized as belonging to Bourreria in Ehretia sect. Bourreria (P. Browne) DC. All recent authors have, however, accepted Bourreria as distinct generically on the basis of its valvate calyx lobes and corollas that are larger and more fleshy than those in Ehretia. Miers (1869) pointed out that Bourreria has fruits that dry at maturity and separate into four single-seeded pyrenes with an apical at- tachment, whereas species of Ehretia have drupaceous fruits that usually remain entire at maturity, probably until they are dispersed. They later divide into two 2-seeded pyrenes. KEY TO THE SPECIES OF BOURRERIA IN PANAMA la. Corollas 28-48 mm er most leaf blades greater than 12 cm costaricensis lb. Corollas 8-12 mm ac most leaf Hades les than 12 cm lon b. an Bourreria costaricensis (Standley) A. Gentry, Phytologia 26: 67. 1973. Schle- gelia costaricensis Standley, ae Field Mus. Nat. Hist., Bot. Ser 128. 1938. TYPE: Costa Rica. Aue ca- taratas de San Ramon, Mar. 1931, M. Brenes 13570 (holotype, F PEN 16; isotype, NY). Bourreria — var. ers Schery, Ann. Missouri ot. . 29: 366. 1942. Bourreria panamensis I. M. bacs J. Arnold Arbor. 29: 229. 1948. , H. von Wedel 2 oen MO, not E Tree 10-15 m tall, the twigs glabrous. Leaves persistent; petioles 10-25 mm long, glabrous; leaf blade elliptic to obovate, 6.5- 462 Annals of the Missouri Botanical Garden 23.5 cm long, 3.7-10.5 cm wide, the apex obtuse to rounded and sometimes mucronate, the base acute to cuneate, the margin entire, the adaxial and abaxial surfaces glabrous. In- florescence a small terminal cyme. Flowers sessile, bisexual; calyx campanulate, 13-18 mm long, 10-18 mm wide at the mouth, glabrous, the 5 lobes triangular; corolla white, broadly funnelform, 2.8-4.8 cm long, 5-mer- ous, the lobes depressed ovate; stamens 5, the filaments 18-20 mm long, the upper 14- 18 mm free, slightly puberulent at the point of insertion, the anthers narrowly ellipsoid, 4 mm long. Fruits green, globose, 2-2.5 cm iam. Distribution. This species occurs in wet to moist forests from sea level to 1,700 from Nicaragua to Panama. In Panama it is known from Bocas del Toro and Colón. This is one of the most distinctive species of Bourreria with its large, funnelform co- rollas more than 2.5 cm long. It is quite similar to Bourreria superba 1. M. Johnston of western Mexico in general appearance but is widely separated geographically and grows in a very different habitat. Bourreria costari- censis differs further from B. superba by having glabrous twigs, leaves, and staminal filaments. Additional specimens examined. PANAMA. BOCAS DEL TORO: Chiriqui Lagoon, Isla Colón, Wedel 2974 (MO). CHIRIQUÍ: Fortuna Dam area, 1,200 m, McPherson 6782 MO); Fortuna Dam Region, near forestry experimental station, S of lake, 1,150 m, McPherson 7873 (MO). COLÓN: Santa Rita Ridge Road, near junction of Trans- isthmian Highway, D'Arcy et al. 15554 (MO). ~ Bourreria oxyphylla Standley, Trop. Woods 16: 40. 1928. TYPE: Belize. El Cayo: San José, Nov. 1927, J. B. Aitken 4 (holotype, F 572622). Beureria wagneri Standley in Yuncker, Publ. Field Mus. t. Hist., Bot. Ser. 9: 328. 1940. TYPE: Honduras. Atlántida: foothills back of La Ceiba, 23 July 1938, >. Yuncker, J. M. Koepper & K. A. Wagner 8608 (holotype, F 941533). Shrub or small tree to 5(-15) m tall, and ca. 1 m diam., the twigs glabrous to puber- ulent or sparsely strigillose. Leaves persistent, the petioles 7-17 mm long, glabrous to pu- berulent, the blades elliptic to elliptic-oblong, 4.5-10.5 cm long, 2-5 cm wide, the apex acute to obtuse and often abruptly short acu- minate, the base acute to obtuse, the margin entire, the adaxial surface glabrous, the abax- ial surface essentially glabrous but sparsely puberulent along the major veins. Inflores- cence terminal, cymose, to 9 cm broad, the branches sparsely to evenly strigillose to pu- berulent. Flowers sessile, bisexual; calyx nar- rowly campanulate, 5-6 mm long, strigillose to puberulent, the 3-5 lobes deltate and often bifid at the apex; corolla white to pale green, tubular with spreading lobes, 8-12 mm long, 5-merous, the lobes oblong-obovate, 5-6 mm long, the tube 5-6 mm long; stamens 5, exserted, the filaments 3-4 mm long, gla- brous, the anthers oblong, 2-3 mm long; ovary ovoid, 1.5-2 mm long, the disc annular, the style 6-8 mm long, the stigmas discoid. Fruits yellow to orange and later turning black, ovoid to subglobose, 6-12 mm long, 5-11 mm broad. Distribution. Bourreria oxyphylla is known from southern Mexico through Belize to Nicaragua with a single collection from San Blas in Panama and a few collections from Colombia. It occurs in wet forests, where it ranges from sea level to nearly 800 m in elevation. Bourreria oxyphylla is distinctive and is one of only two members of the genus in Central America with glabrous, elliptic leaves; the other, Bourreria costaricensis, is readily separated by its larger leaves, some to 10.5 cm wide, and corollas more than 2.8 cm long. Bourreria wagneri was described from a pop- ulation in Honduras with slightly more pu- berulent twigs but is otherwise identical to other populations and is not recognized as distinct. The southern populations from Pan- ama and Colombia have slightly larger, less lustrous leaves than the populations from northern Central America, but the differences Volume 75, Number 2 1988 Miller Revision of Panamanian Boraginaceae are slight and they do not seem to warrant taxonomic recognition. itional specimens examined. PANAMA. SAN BLAS: ub y Soskatupu, west end, Kirkbride 195 (MO). Cordia L., Sp. Pl. 190. 1753; Gen. Pl. ed. 5. 87. 1754. TYPE: Cordia sebestena L., Sp. Pl. 190. 1753 Trees or shrubs. Leaves alternate, decid- uous or persistent, petiolate, the petioles usually adaxially canaliculate. Inflorescence cymose, paniculate, spicate, capitate, or glomerate. Flowers perfect, or unisexual by abortion, with the plants then dioecious; calyx 3-5(-10)-lobed or rarely circumscissile; co- rolla funnelform, campanulate, or tubular with reflexed or spreading lobes, (4—)5(—18)-lobed, or sometimes the lobes nearly lacking and the corolla apically undulate to frilled or nearly truncate; stamens as many as corolla lobes, the lower part of the filaments adnate to co- rolla tube, often with hairs at or near insertion, the anthers oblong to ellipsoid; ovary entire, falsely 4-locular; disc annular to crateriform; style terminal, twice bifid, the 4 stigma lobes clavate, filiform, or discoid. Fruits borne with the calyx persistent, variable, dry with a fibrous wall and capped by the persistent, cartilaginous base of the style (sect. Geras- canthus), dry and bony-walled (sect. Rhab- docalyx), or with a thin exocarp, juicy to mucilaginous mesocarp, and bony endocarp (sects. Varronia, Myxa, and Cordia), usually KEY TO THE SPECIES OF CORDIA IN PANAMA la. sans bright red-orange; fruit drupaceous and totally enclosed in the accrescent calyx at maturity (sect. E Cordi 1-locular and 1-seeded, the endosperm lack- ing, the cotyledons plicate. The pantropical genus Cordia is the largest in the family: there are about 300 species, mostly neotropical. Species of Cordia are found in a wide variety of habitats, but, al- though many species can be found in wet forests, those found in dry, disturbed areas are a more important component of the vege- tation. Cordia is the only Panamanian genus of the Cordioideae. The South American genera, Auxemma Miers and Patagonula L., differ from Cordia in the unusual form of their fruiting calyces and by having short styles and monomorphic flowers. Although clearly related to Cordia, they seem to represent a distinct lineage. Cordia, however, is a very diverse assemblage, and a number of authors have suggested dividing it into 3-12 segre- gate genera (Mez, 0; Friesen, 1933; Nowicke & Ridgway, 1973). Nevertheless, Cordia sensu lato appears to be a distinct monophyletic group and its division into nar- rowly defined genera seems unwarranted. Johnston (1930, 1940, 1949a, b, 1950, 1951) treated the genus in a broad sense and recognized five to seven sections in his various works. Recent authors (Nowicke, 1969; Gib- son, 1970; Nowicke & Ridgway, 1973; Opler et al., 1975; Miller, 1985) have recognized five sections. Twenty-seven species in four of the sections are known from Panama. sebestena ~ c 2a. Co are marcescent; 3 oblon : Corolla white to yellow; zu not n aiios by the calyx at maturity or if enclosed, then not drupaceous. a fibrous wall (sect. Gerascanthus omatia sim at the base of ila inflorescences; leaves and twigs with stellate hairs; corolla lobes C. alliodora N c . Corolla deciduous; fruit with a Trees or shrubs with iw s stems; leaf m paniculate, branching mo argin entire or dentic e than twice; fruits generally as symmetrical (sect. 5a. Corolla yellow to iens white; calyx si aa and str 5b. Corolla white; calyx opening with valvate lobes a E 3b. Ant domatia absent; leaves M twigs glabrous; corolla lobes deltate or ovate 0. C. megalantha wall. ulate; inflorescences cymose to Myxa) C. dentata iate r if circumscissile than not striate. alyx costate; leaf margin with short, filiform teeth toward the apex or entire U. C. diversifolia 464 Annals of the Missouri Botanical Garden 6b. Calyx smooth or striate, not costate; leaf margin entire. 7a. Ovary and fruit n C. bicolor Tb. Ovary and fruit glabro 8a. Adax d d o evenly strigillose or scabrous to strigose; plants usually dioecio 9a. Corolla tube 1.5-3.9 mm long; fruits ovoid, 6-10 mm lon E and siger with simple hairs; the larger leaves generally C. panamensis 10b. Sem: and peduncles with echinate hairs; the larger leaves generally 1.2-20.4 cm wide . cymosa Ob. Corolla tube ca. 10-12 mm long; fruits ellipsoid, 8.5-16.5 mm ong "M nisophylla . Adaxial leaf surface glabrous or nearly so; plants usually with bisexual baut these usually distylous, dioecious only in C. collococca and C. tacarcunensis. lla. Leaf margin distinctly revolute. 2a. Stems sub-alate; abaxial leaf surface glabrous; leaves not bullate eo > C. leslieae 12b. Stems not winged or ridged; abaxial leaf surface velutinous; leaves ate C. dwyeri llb. Leaf margin not or only slightly revolute. 13a. Corolla campanulate C. eriostigma 13b. Corolla tubular, with reflexed or spreading lobes. 14a. Calyx distinctly 5-lobed. nflorescences axillary, numerous on a single stem; plants dioecious . tacarcunensis 15b. Inflorescences terminal or subterminal, Du per stem; plants with bisexual flowers. 16a. Fruits white; leaf blades falcate ............... C. protracta 16b. Fruits red or orange; leaf blades flat, not falcate C. correae 14b. Calyx (2-)3(-4)-lobed, circumscissile or dehiscing irregularly. Plants no leaves persistent; flowers borne on current seaso 18a. Apex of m leaf caudate, the caudex 1-3.5 c long C. one 18b. fic of the leaf acumin 9a. Le 2 blade elliptic P elliptic-ovate, less m wide, usually less than 8 cm . croatii 19b. Leaf blade ovate to pid be or lance- e usually m an 5 cm wide, usually more than ong. 20a. Fruit rel the stone not rostrate; ant ad .1-1.2 mm long; calyx 3- mm C. lucidula 20b. Fruit E the stone not rostrate; anthers 1.9-2.3 mm long; calyx Ps 2d 1 a tn el . porcata 17b. Plants dioeci leav ; flowers usually borne current season's gro wth eei ei EE below the 4b. Shrubs, usually multistemmed; leaf margin serrate i minutely denticulate; Dc co ensed, spicate, capitate, or if cymose, then less than 2.3 cm broad, dichotomous, and webcam only twice; fruits ipi im or nearly so (sect. Varronia). 21a. Inflorescences s e. 22a. Leaf blade lb inflorescences terminal C. curassavica 22b. Leaf blade ovate; inflorescences axillary, the peduncle adnate to the petiole .................. . spinescens 21b. Inflorescences cymose or capita 23a. Inflorescence branched, b ons C. bifurcata 23b. Inflorescence capitate. alyx lobes with prolonged filiform tips. 25a. Peduncles 0.5-2 cm long; corolla 5-10 mm long U... C. globosa 25b. Peduncles 3-7 cm long; corolla 3-6 mm long C. bullata 24b. Calyx lobes acute to acuminate but lacking ROMPE tips. 26a. Inflorescences terminal or internodal C. inermis 26b. Inflorescences axillary C. linnaei Volume 75, Number 2 1988 Miller 465 Revision of Panamanian Boraginaceae Cordia alliodora (Ruiz Lopez & Pavon) Oken, Allg. Naturgesch. 2(2) 1098. 1841. Cerdana alliodora Ruiz Lopez & Pavon, Fl. Peruv. 2: 47, pl. 184. 1799. TYPE: Peru. Huanuco: Pozuzo, Hipólito Ruíz & José Pavón (holotype, B, not seen; photo, MO) Tree to 20(-25) m tall, the twigs stellate- pubescent when young, ending in obovoid ant domatia. Leaves deciduous; petioles (5-)8- 28(-35) mm long, stellate-pubescent; blade elliptic to narrowly elliptic or slightly obovate, (3.5-)5-17(-20.5) cm long, (1.4-)2-7(-8.5) cm wide, the apex acuminate or acute, the base acute to obtuse, the adaxial surface gla- brous to sparsely stellate-pubescent, the abax- ial surface sparsely to densely stellate-pubes- cent. Inflorescence terminal, usually arising from an obovoid ant domatium, paniculate, to 25(-30) cm broad, the branches usually densely stellate-pubescent. Flowers borne on short spurs to 1.5 mm long, bisexual, mono- morphic; calyx tubular, (4-)4.5-5.5(-6.5) mm long, 10(-12)-ribbed, stellate-pubescent, with (4-)5(-6) small teeth; corolla marces- cent, white, (8.5—)9.5-12(-14) mm long, (4-) 5(-6)-merous, the lobes oblong, (4.5-)5-7 (-8.5) mm long, the tube (3.5-)4.5-6(-8.5) mm long; stamens (4-)5(-6), the filaments 9-12 mm long, the upper (3.5-)5.5- 7.5 (-9) mm free, sparsely pubescent at insertion, the anthers oblong, 1.5-2(-2.5) mm long, borne at the same height or above the stigmas; ovary ovoid to very broadly ovoid, (0.7—)1-2(-2.5) mm long; disc depressed obovoid to very broadly obovoid, 0.5-1 mm tall; style 4.5- 6.5 mm long, the stigma lobes clavate. Fruits enclosed by the persistent corolla and calyx, ellipsoid, (4.5-)5- 7(-8) mm long, (1-) 2- mm broad, the wall thin, fibrous. Distribution. Cordia alliodora occurs in dry to wet forests from sea level to 1,100 m in elevation and ranges from Mexico to South America and to the West Indies. This species is known from all provinces in Pan- ama. Cordia alliodora is the widest-ranging species of the genus. Its ant domatia and stellate indument are distinctive. It is also one of only two homostylous species in sect. Ger- ascanthus, the other being Cordia tricho- toma (Vell. Arráb. ex Steudel of southern South America (Gibbs & Taroda, 1983). This condition is derived in the section, and these two species form a monophyletic group fur- ther characterized by stellate hairs on the stems and leaves and by flowers considerably smaller than those of their relatives. The flow- ers of Cordia alliodora vary from short-style to forms with the stigmas and anthers borne at about the same height, but individual plants appear to be constant in the ratio of anther and stigma height. No plants have been found with styles exceeding the stamens consider- y. Cordia alliodora is valued as a timber tree and, for this reason, there has been a great deal of interest recently in establishing plan- tations of it throughout Central America (Stead, 1980). Flowering occurs at the onset of the dry season with fruits maturing and being dispersed during mid dry season. Com- mon names include Laurel, Laurel Blanco, and Laurel Negro. Additional specimens examined. PANAMA. BOCAS DEL TORO: Almirante region, Cooper & Slater 22 (CFMR, 2 US); Changuinola Valley, Dunlap 291 (F, GH, US); vi cinity of Chiriqui Lagoon, Old Bank Island, Wedel 1877 MO, US). CANAL AREA: Summit Area, Avilla 302 (MQ); U.S. Army Tropic Test Center, Fort Clayton, vicinity Gorgas Hospital, Blum 2210 (MO, SCA) between Gatún and Bohio, Christopherson 118 ( eet ~ 4799 (F, SCZ); 7694 (MO); 8104 (F (3), MO, SCZ); fe a 8401 (F, MO); Gaillard Hia 2 mi. ; 14048 MO); Barro Colorado E D'Arc: 4 TEX) Albrook, U.S. Army Tropic Test Enn Site, Dwyer & Robyns 56 (BR. ~ = 719 (F, ); Curundu, Harvey 5249 (F); in govern- ment fondi sions Las Cruces trail, 75 m, Hunter & Allen 449 (BR, F, G, MO, NY, P, UC, US); near Gamboa, along edge 4 area, along shore side, Kennedy & Steiner 2453 (C, CAS, ENCB, F, L, MO, N vicinity of Culebra, Maxon 4901 (C, F, NY, Gamboa, Piper s.n. (GH (2); Chiva-Chiva Trail, Piper 5755 (GH, NY, US); vicinity of Ancon, Piper 6007 (US); Barro Colorado Island, Shattuck 780 (GH, MO), 789 (MO (2)); along the old Las Cruces trail, between Fort Clayton and Corozal, Standley 29229 (US); between France Field, Canal Zone, and Catival, province of Colon, jeg d 30200 (US) Fort Clayton rea, Tyson 3473 (MO, SCZ); inte of Fort Clayton, 466 Annals of the Missouri Botanical Garden Walker 1 (UC); Ancon, Wheeler $ se sn. (GH (2); ceolate, 10.5-21(-26) cm long, (3-)5-12.5 Barro Colorado Island, Wilson 80 (F, MO); Woodworth ). CHIRIQUÍ: 1 la carretera a Los e PO s— < 7 326 (NY (2), US), 327 (NY). COLÓN: Juan. Min a, Rio Chagres, 25 m, Allen 4198 (G, MO); entres ne purse. Icacal EA is in between Salud y e Rio Indio, Howell 2 7 (MO). DARIEN: Battelle Memorial liar sea level, Canal Bioenviron- m, - Allen 1607 (F, MO, NY , US); Island of Taboga, Barclay 981 (F, US); a orillas del Río Aguacate cerca de Nuevo Arraijan, Cedeno 13 (F, MO); en El Jobo, San Carlos, Gonzalez 19 (F, MO); Isla Taboga, Hjerting & Rahn 617 (C, US); San José Island, Johnston 570 (GH); near Chepo, Kluge 3 (CFMR, US); Isla Tabogo, Macbride 2822 (F, GH, US); Chepo, Paul 326 (US). SAN BLAS: Permé, Cooper 649 (CFMR, DS, F, US); mainland op- posite Ailigandi, from mouth of Ailigandi River to 2.5 mi. inland, Lewis et al. 154 (MO). VERAGUAS: border of Ve- raguas, Coclé, and Herrera provinces along the Rio Santa María near bridge of Panamerican Highway, 16 km SW of Aguadulce, 0-50 m, Knapp et al. 3348 (MO) — anisophylla James $ Miller, sp. TYPE: Panama. Colón: Santa Rita Ridge Road, 9 km from Boyd-Rooseveli Highway, 350 m, Premontane Wet For- est, 15 Mar. 1975, S. Mori & J. Kal- lunki 5076 (holotype, MO 2664952; isotypes, MO, US). Figure 1. Arbor ud ks m alta. Folia persistentia, petiolis (4- j6- 12 mm long T o ovatis, 10.5- 21(- -26)« cm longis, (3-)5-12.5 m latis, minoribus ovatis ad cordatis, (3.5-)7.5 C 13.5) em longis, 4.5- 7.3(71 4) utis, basi obtusis, superficie strigillosa, pagina inferiore strigillosa ad pilosa. Inflorescentiae terminales in axillis ramorum dispositae, cymosae. Flores heterostyli; pet tubuliformis, 7-9 mm longus; corolla alba, tubuli- formis, 10-12 mm longa, 5-lobata; stamina nth oblongis. Fructus drupaceus, putamine inaequilateraliter m lat ovoideo, 8.5-16.5 mm longo, 6-7.5 mm i -_. [^ Small erect tree or large shrub to 5(-8) m tall, the bark brown, smooth, lateral branch- ing dichotomous in a horizontal plane, the twigs strigillose to pilose. Leaves persistent; petioles (4-)6-12 mm long, canaliculate adaxially, strigillose to pilose; blades aniso- phyllous, the larger ones ovate to ovate-lan- cm wide, the smaller ones orbicular, (3.5—) 7.5-8.5-13.5) cm long, 4.5-7.3(-14) cm wide, the apex acuminate, the base subobtuse to obtuse, the margin entire, the adaxial sur- face evenly strigillose, the abaxial surface strigillose to unevenly pilose. Inflorescences terminal or borne in the axils of branches, rarely internodal, loosely branched cymes, the peduncle 5-10(-13.5) cm long, strigillose to nearly pilose. Flowers sessile, distylous; calyx tubular, 7-9 mm long, 3 mm wide at mouth, the 5 lobes + deltate, 1.8 mm long, ribs absent, strigillose, densely pilose on in- terior surface; corolla white, tubular with re- flexed lobes, ca. 10-12 mm long, 5-merous, the lobes 2.7 mm long, 1.8 mm wide, the tube 9.3 mm long; stamens 5, the filaments 10.5 mm long, the upper 2.6 mm free, villose at and above insertion, the anthers oblong, 2 mm long; ovary oblong, 1.8 mm long, 1.1 mm broad, glabrous; disc small, not evidently distinct from the base of the ovary; style 5.3 mm long, the stylar branches 0.9 mm long, the stigma lobes clavate. Fruits seated in the cupulate calyx, drupaceous, glabrous, the stone slightly inequilaterally ovoid, 8.5-16.5 mm long, 6-7.5 mm broad, the endocarp bony. Distribution. Cordia anisophylla oc- curs in wet forests from sea level to 1,000 m in elevation and is known only from Pan- ama in the provinces of Colón, Darién, Pan- ama, and San Blas. Cordia anisophylla is closely related to C. panamensis and to C. cymosa but differs from them by having corollas more than 10 mm long and fruits more than 8.5 mm long. This species is also one of only two distylous Central American members of sect. Myxa, the other being C. dentata. While the Cordia panamensis species complex is one of the most taxonomically confusing within sect. Myxa, Cordia anisophylla is probably the most distinctive species of the group in its elongate, distylous flowers and ellipsoid fruits. Additional specimens examined. NAMA. COLÓN: Santa Rita Ridge, Croat 13895 (MO); Santa Rita lumber Volume 75, Number 2 1988 Miller Revision of Panamanian Boraginaceae 467 ` S z ^X P pb Z a x P DE - un —À Ne S< < b T \ Ls A. Wee. y Cordia anisophylla. — A. Flowering branch.—B. Flower with corolla opened.—C. Fruit. A, b n. FIGURE Mori & Kallunki 5076 (MO), Colón, Panama; C from Mori & Kallunki 6373 (MO), Panamá, Panam road, ca. 15 km E of Colón, Dressler 3799 (DUKE, F (2), MO (2)); Santa Rita Ridge Road, ca. 8 mi. E of the Transisthmian Highway, along trail N of road, 350-440 m, McPherson & Merello 8244 (MO). DARIÉN: along Rio Chucanaque between El Real and Rio Canalones, Duke 4979 (MO). PANAMÁ: region of Cerro Jefe, 1,000 m, Correa et al. 1589 (MO); El Llano-Carti Road, 20.7 km from Interamerican Highway, 350 m, Mori & Kallunki 5114 (MO); 5-10 km NE of Altos de Pacora on trail at end of road, 700-800 m, Mori & Kallunki 6057 (DUKE); El Llano-Carti Road, 12.2 km from Interamerican High- way, Mori & Kallunki 6373 (MO, SCZ, US); El Llano- Carti Road, ca. 9 mi. from Pan-American Highway along newly cut by-pass, 300-400 m, Sytsma 4127 (MO). SAN BLAS: near Nusigandi on Llano-Carti Road, 300-350 m, McPherson 10796 (MO) 468 Annals of the Missouri Botanical Garden ids EDU A. DC. in DC., Prodr. 9: 845. TYPE: Surinam: Hostmann ds UE G-DC, not seen; micro- fiche, MO; isotype, P). Cordia dieci a Pittier, Contr. U.S. Natl. Herb. 18(6): 252. 1917. TYPE: Guatemala. Alta Verapaz: vicinity of Sedang: 550 m, 30 Apr. 1905, H. Pittier 189 (holotype, US 472845). Cordia belizensis Lundell, Amer. Midl. Naturalist 29: 48 43. TYPE: Belize. Toledo: Monkey Ch in high ridge between Swacey Branch and a leaf Creek, 5 July 1942, P. H. Cail 4045 (ho. lotype, MICH; isotypes, GH, NY). Tree to 20 m tall, the young twigs velu- tinous to puberulent. Leaves persistent; pet- ioles (2-)3-8(-16) mm long; blades elliptic to ovate or narrowly elliptic, (8-)10-19(-22) cm long, (3.3-)4-9(-11.5) cm wide, the apex acuminate, the base obtuse to rounded, the margin entire, the adaxial surface sparsely strigillose to scabrous, the abaxial surface pale, puberulent to strigillose. Inflorescence ter- minal or borne in the axils of branches, cy- mose, to 13(-20) cm long, 14(-28) cm broad, the branches velutinous. Flowers sessile, bi- sexual, monomorphic with the stamens as long or longer than the stigmas; calyx tubular- campanulate, (3.3-)3.6-4.3(-4.9) mm long, ribs absent, densely strigillose, the 5 lobes deltate to attenuate; corolla white, tubular with reflexed lobes, 6.2-7 mm long, 5-mer- ous, the lobes oblong, (2.3-)2.7-3(-3.8) mm long, the tube (2.5-)3.6-4.6(-5) mm long, pubescent in the mouth; stamens 5, the fil- aments (3.8-)4.8-6(-6.4) mm long, the up- per (1-)1.8-3 mm free, pubescent at the point of insertion, the anthers oblong, 1-1.6 mm long; ovary ellipsoid, 0.8-1.6 mm long, strigillose; disc crateriform, 0.3-1 mm tall; style 3-3.7 mm long, the stigma lobes discoid to broadly clavate. Fruits seated in the cu- pulate calyx, white, drupaceous, the stone inequilaterally ovoid, (7.3-)10.5-13 mm long, (4.5-)7.5-9.5 mm broad, the exocarp dense- ly strigillose, the endocarp bony. Distribution. | Cordia bicolor occurs in wet forests from sea level to 500 m in ele- vation and ranges from Mexico to South America. In Panama this species is known from the Canal Area, Chiriqui, Colon, Darien, Panamá, and Veraguas. Cordia bicolor is perhaps most closely re- lated to the C. panamensis complex in that it also has dichotomous lateral branching, an- isophyllous leaves, and similar indument. However, it differs from the other Central American species in its strigillose fruits, a relatively rare character found in several un- related South American species. Cordia to- queve Aubl. of South America is the only other species of this complex that shares this trait. Additional specimens examined. PANAMA. CANAL AREA: Barro Colorado Island, Aviles 21 (F); Croat 5630 (DUKE, F, MO, NY, SCZ), 5823 (DUKE, F, NY, SCZ (2), 7705 (DUKE, F, MO, NY), 8004 (F, MO, NY, SCZ), 8804 (DUKE, MO), 8809 (F (2), DUKE, MO, NY), 9447 (MO), 14855 (MO, SCZ, UC); Duke 8379 (NY); without definite locality, end s.n. (F (2) Barro Colorado Island, Foster 1778 (DU " Pipeline Road between mile marker 0 mi. N of Gamboa, vind et al. 5446 (F, MO, NY); near Fort Randolph, Maxon & Mi 6520 (US); Barro Colorado Island, (p zenheimer 25 2 (MO); near old Fort Lorenzo, mouth of Rio Chagres, Piper 5964 (US). CHIRIQUÍ: at Monte Rey above Boquete, Croat 15770 (NY). c ar radio tower at the end of turnoff to Santa Rita Ridge Road 200-300 m, Miller & Miller 910 (MO). DARIEN: . 1 mi. NE of Nura, 200 m, Duke 10081 (MO, US). PANAMA: San José Island, Erlanson 256 (G, NY, US); Johnston 417 (GH), 545 (GH, MO, U, US), 618 (DUKE, ; 620 (GH, U, US), 621 (DUKE, GH, LL, U), 785 (GH, b e (GH), 1076 (GH (2)); Altos del Rio , Lewis et al. 2267 (MO, UC). VERA- on slopes yh Cerro Tute below Agricultural School, Gentry 6204 (MO); NW of Santa e near entrance to school, Mori & Kallunki 4888 (AAU, MO, US). £ uc o < (c Cordia bifurcata Roemer & Schultes, Syst. 466. 1819. Varronia dichot- oma. Rik Lopez & Pavón, Fl. Peruv. 2: 23, t. 146. 1799, not Cordia dichotoma G. Forster, Fl. Ins. Austr. 18 n. 110. 1786. TYPE: Peru. Huánuco: Chaca- huasi, Hipólito Ruíz & José Pavón s.n. (holotype, B-W, not seen; microfiche, MO) Shrub to 3 m tall, the twigs sparsely strigil- lose, the hairs shorter than 0.5 mm, ap- pressed, white translucent. Leaves deciduous, Volume 75, Number 2 1988 Miller 469 Revision of Panamanian Boraginaceae on short spurs 1.5-2 mm long; petioles 2-8 mm long, sparsely strigillose; blade ovate to lanceolate, (2.5-)3.7-9(-12) cm long, (0.8-) 1.2-4(-4.7) cm wide, the apex acuminate, the base acute, the margin slightly serrate to entire, the adaxial surface sparsely puberulent with short, erect hairs, the abaxial surface sparsely pubescent. Inflorescence terminal or lateral, a small forking cyme with 4 or more branches, 0.8-2.3 cm broad, the peduncle 1.4-6 cm long, strigillose. Flowers sessile, distylous; calyx short-tubular, 2-2.7 mm long, strigillose, the 5 lobes deltate to shallowly deltate; corolla white, tubular, (2.7-)3.2-3.7 mm long, truncate at the apex, canescent in the middle of the tube; stamens 5, the fila- ments (2.2-)2.7 -3.5 mm long, the upper 0.6- 1(-1.3) mm free, canescent at insertion, the anthers broadly ellipsoid, 0.5-0.6 mm long; ovary very broadly obovoid to broadly de- pressed obovoid, 0.6-1 mm long; disc thin, cuplike, nearly completely adnate to the ovary; style (2.2-)3-3.5 mm long, the stigma lobes flattened. Fruits drupaceous, 43-14 enclosed in the slightly accrescent calyx, bright red, the stone ovoid, 4-4.5 mm long, 2.2-3 mm broad, the endocarp bony. Distribution. Cordia bifurcata occurs in wet forests from sea level to 600 m in elevation and ranges from southern Nicara- gua south to Peru. In Panama this species is known from the provinces of Bocas del Toro, Colón, and Darién. Cordia bifurcata is distinctive in its small, shrubby habit and small, cymose inflores- cences and has no close relatives in Panama. It is a member of a taxonomically difficult group of mostly South American species that are all similar in general aspect. The only other member of this group found in Central America is C. foliosa Mart. & Gal. of south- ern Mexico and Guatemala, which differs from C. bifurcata in its elliptic, firm-textured leaves with a scabrous upper surface, apiculate flow- er buds, and acuminate calyx lobes. Two other species of Central America, C. inermis and C. linnaei, are also closely related to this group, but neither of these has a branched inflorescence. Additional specimens examined. PANAMA. BOCAS DEL TORO: region of Almirante, Cooper 83 (F, NY); Chan- guinola Valley, Dunlap 90 (GH). COLÓN: vicinity of San Juan near Caman Plant Lake, Blum & Tyson 537 (MO). DARIÉ i Punta Guayabo Grando to Rio Jaqué, 50-200 m, Antonio & Hahn 4415 ut vicinity of Paya, Rio Paya, Stern et al. 268 (G, MO, US). Cordia bullata (L. Roemer & Schultes, Syst. Veg. 4: 462. 1819. Varronia bul- lata L., Syst. ed. 10: 916. 1759. TYPE: Jamaica, P. Browne s.n. (holo- type, LINN, not seen (Savage Catalog number 255.2); microfiche, MO). a — var. angustata DC., Prodr. 9: 496. 1845. E: Guadalupe, 1818, Krause s.n. (holotype, “DC. not seen; microfiche, ). Cordia m ., Prodr. 9: 498. 1845. Varronia rrima bet Friesen, Bull. Soc. Bot. Genéve Ser. d 24: t. 1, f. 5. 1933. TYPE: 1822, ig ex herb. Balbis (holotype, G-DC, fiche, MO). not seen; micro Shrub to 1(-3) m tall, the twigs hirsute to hirtellous, the hairs erect to spreading. Leaves deciduous, on short spurs to 1 mm long; pet- ioles 2-7(-10) mm long, shallowly canali- culate to flattened adaxially, hirsute to hir- tellous, the hairs erect to spreading; blade ovate to narrowly ovate, (1.5-)2-8(-9.2) cm long, (0.8-)1.2-4.7(-6.3) cm wide, the apex acute to slightly attenuate, the base subobtuse to obtuse or acute and abruptly decurrent along the petiole, the margin serrate, usually unevenly so, the teeth usually sharp and often short-apiculate, the adaxial surface usually bullate, strigose, the lower surface coarsely pubescent, the hairs + restricted to the veins. Inflorescence internodal or less commonly subterminal or terminal, a dense, often slight- ly ellipsoid head, 8-12(-15) mm broad, the peduncle (1.5-)3-6.5(-13.5) cm long, hir- sute, the hairs erect to spreading. Flowers EV. sessile; calyx campanulate, (2-)2.5- -3.8) mm long, (2-)2.5-3.5 mm wide at d mouth, ribs absent, strigillose, the 5 lobes deltate to triangular, 0.5-1 mm long, with a prolonged filiform tip 1-3(-3.5) mm long; corolla white, tubular, 3-5.3(-6) mm long, 470 Annals of the Missouri Botanical Garden undulate, the lobes scarcely distinct, the tube 1.2-2.8 (-3.2) mm long; stamens 5, the fil- aments (3.3-) 3.8-4.3(-5) mm long, the up- per 1-1.7(-2) mm free, puberulent in a ring in the mouth of the corolla tube, the anthers ellipsoid, 0.6-1 mm long; ovary ovoid, 1- 1.8 mm long, glabrous; disc usually indistinct from the base of the ovary; style 0.6-4.3 mm long, the stylar branches 0.4-1 mm long, the stigma lobes clavate to discoid. Fruits drupaceous, red, the stone ovoid, 3.7-5.3 mm long, 2.5-3.2 mm broad, the mesocarp thin, the endocarp bony. Distribution. Cordia bullata occurs in dry forests from sea level to 1,400 m in elevation. It is known from Nicaragua to southern Mexico, northern South America, the Greater Antilles, and a single collection from the province of Veraguas in Panama. Cordia bullata is closely related to and often confused with C. globosa (Jacq.) Kunth but differs by having peduncles greater than 3 cm long and corollas less than 6 mm long. Both are common in northern Central Amer- ica, and populations with interspecific hybrids are known from Nicaragua and Honduras. Additional specimens examined. PANAMA. VER- AGUAS: El Cuchillo, near Cerro Tute, up from Santa Fe, 1,300 m, Hamilton et al. 1203 (MO). Cordia collococca L., Sp. Pl. ed. 2. 274. 1762; excluding Cordia glabra L. — Bourreria succulenta Jacq.; 1. M. John- ston, J. Arnold Arbor. 21: 345. 1940. TYPE: without definite locality, collector unknown (holotype, LINN, not seen (Savage Catalog number 253.8); mi- crofiche, MO). Cordia micrantha Sw., Prodr. 47. 1788. TYPE: "habitat in Jamaica," not seen. Small tree to 8(-15) m tall, the twigs sparsely to evenly strigillose, later waxy. Leaves deciduous; petioles 5-12(-15) mm long, sparsely strigillose; blade oblong-obovate to elliptic or obovate, (4.7-)5.5-14(-15.2) cm long, (2.5-)3-6.5(-7) cm wide, the apex acute, often with short-acuminate tip, rarely acute, the base cuneate to acute, the margin entire, the adaxial surface glabrous or nearly so but with numerous small papillae, the abax- ial surface evenly strigillose to hirtellous. In- florescence terminal, borne on old wood just before the new vegetative shoots appear, or axillary, cymose, (6-)7-14(-18) cm broad, the branches sparsely strigillose. Flowers ses- sile, unisexual by abortion, the plants dioe- cious; female flowers with small, nonfunc- tional anthers; male flowers with shortened, reduced styles; calyx cupulate, (1.7-)2-2.6 (-3) mm long, ribs absent, evenly strigillose, circumscissile or unevenly 3-lobed; corolla white, tubular with reflexed lobes, (4.5—)4.8— 6.3(-6.9) mm long, 5-merous, the lobes ob- long-ovate to ovate, (2.3-)2.8-3.7(-4) mm long, the tube 1.7-2.9(-3.3) mm long; sta- mens 5, the filaments 2.5-5 mm long, the upper (0.8-)1.7-2.2 mm free, puberulent to pubescent below insertion, the anthers oblong to ellipsoid, 0.5-1.6 mm long; ovary ovoid to oblong, 0.8-1.2 mm long, glabrous; style 0.4-2 mm long, the stigma lobes clavate to filiform. Fruit borne with the small calyx per- sisting at the base, bright red, drupaceous, glabrous, the stone inequilaterally ovoid, 7.5- 9.3 mm long, 5.5-7.3 mm broad, the en- docarp bony. Distribution. Cordia collococca occurs in dry forests from sea level to 200(-900) m in elevation from Mexico south to northern South America and the West Indies. In Pan- ama this species is known from Bocas del Toro, Canal Area, Chiriqui, Colón, Herrera, Los Santos, Panamá, and San Blas. Cordia collococca is a relatively common species throughout much of its range, al- though it has not been collected frequently in Panama. It is easily confused with Cordia eriostigma but can be distinguished by being a smaller tree with an even indument of short appressed hairs on its abaxial leaf surface, deciduous leaves, and flowers that are uni- sexual by abortion. Cordia collococca is gen- erally found at elevations below 200 m in dry forests, while C. eriostigma usually grows at 600- 1,400 m in moist forests. Cordia micrantha is clearly a synonym of C. collococca, but Swartz specified no type Volume 75, Number 2 1988 Miller 471 Revision of Panamanian Boraginaceae other than “habitat in Jamaica," and the choice of a lectotype will require study of material in European herbaria. The name Cordia glabra has been improperly applied to this species; Johnston (1940) showed that this name should be considered a synonym of Bourreria succulenta. Additional specimens examined. PANAMA. BOCAS DEL TORO: Almirante, Cooper 406 (CFMR, F, NY, US); on lower Changuinola River, Stork 273 (UC, US). cANAL AREA: between Farfan beach and Vera Cruz, Duke 11733 (MO). CHIRIQUÍ: Progresso, Cooper & Slater 300 (CFMR, US); without definite locality, Cooper & Slater 307 8 (MO). SAN BLAS: Permé, Cooper 235 (NY, Cordia correae James S. Miller, sp. nov. TYPE: Panama. Coclé: La Mesa, 4 km north of El Valle, disturbed tropical wet forest and roadside, 875 m, 3 Jan. 1974, M. Nee & J. D. Dwyer 9164 (holotype, MO 2414635; isotype, DUKE). Fig- ure 2. Arbor vel frutex ad 8 m alta, ramunculis | aide ad strigillosis. Folia persistentia, petiolis 5-10(-1 m lon- gis, strigillosis; laminae aniso phyllae, coriaceae, is maioribus ovatis ad a anguste ovatis, 10.3-17.6 cm longis, 4.8-8.8 cm latis, apice acuminatis, basi ee ad obtusis, superficie sparsim jb pagina inferiore mi- nute strigillosa inr niae dos patra vel a. cym ad 5.5 cm lata ores sessiles; calyx urceo hn 5 p mm longus, G 1 Fructus ee. "i omi putamine Wa eiim ovoideo, 8.4-13 cm longo, 6.2-9 mm lato, ruminato Tree or shrub 4(-8) m tall, the twigs nearly glabrous to strigillose. Leaves persistent; pet- ioles 5-10(-14) mm long, canaliculate adax- ially, unevenly and often sparsely strigillose; blades anisophyllous, coriaceous, the larger ones ovate to narrowly ovate, 10.3-17 long, 4.8-8.8 cm wide, the smaller ones ovate, 7-8 cm long, 4-5.5 cm wide, the apex acu- minate, the base rounded to obtuse or rarely acute, the margin entire, the adaxial surface with widely scattered appressed hairs, the abaxial surface minutely strigillose. Inflores- cences subterminal, internodal or axillary, few per stem, cymose, to 5.5 cm broad, expanding somewhat in fruit, peduncle 1.8-5.3 cm long, strigillose, the hairs brown. Flowers sessile; calyx urceolate, 5.6 mm long, 3 mm wide at the mouth, the 5 lobes deltate, 0.7-1.4 mm long, ribs absent, rufous-strigillose; corolla white, tubular with reflexed lobes, m long, 5-merous, the lobes oblong, 3.6 mm long, 1.9 mm wide, the tube 7.8 mm long; stamens 5, the filaments 10.5 mm long, the upper 6 mm free, glabrous, the anthers ob- long, ca. 1 mm long; ovary ovoid, glabrous; style ca. 4 mm long, the stylar branches 2.3 mm long, the stigma lobes clavate. Fruits borne in the slightly expanded, saucer-shaped calyx, orange at maturity, drupaceous, gla- rous, the stone inequilaterally ovoi A- 13 mm long, 6.2-9 mm broad, ii. the endocarp bony. Distribution. Cordia correae occurs in wet forests from 800 to 1,000 m in elevation and is known only from Panama in the prov- inces of Coclé, Panamá, and Veraguas. Cordia correae is known from only a few collections from Panama in the region of El alle and from Cerro Jefe. Its closest relative is probably C. protracta I. M. Johnston, a species of low elevations along the Atlantic coast of San Blas and adjacent Colombia. The two species share similar habits of growth, branching patterns, anisophyllous leaves, and distinctly five-lobed calyces. The fruits of C. correae, however, are orange, subglobose, and have a ruminate surface, while those of C. protracta are white, elongate, and ridged lon- gitudinally. Cordia correae, which is endemic to Panama, is named in honor of Profesora Mireya D. Correa A., who has done much to advance the study of Panamanian plants. Additional specimens examined. PANAMA. COCLÉ: La Mesa above El Valle, P: road which ends in pasture, 10 m, Croat 25310 (MO, NY); vicinity of La Mesa, N of El Valle, 1,000 m, Gentry 6813 (AAU, MO, NY); Cerro Pilón, El Valle Site Area of WEPCOR, Kirkbride 1071 (NY); La Mesa, 4 km N of El Valle, 850-875 m, Nee & Dwyer 9214 (MO (2), NY). PANAMÁ: along road, 18.9 km N of Cerro Azul, Mori & Kallunki 4998 (AAU, DUKE, NY, (2)). vERAGUAS: vicinity of Cerro Tute, for- ested slopes along trail to summit, 850-1,000 m, Mc- Pherson 10684 (MO). Cordia croatii James S. Miller, sp. nov. TYPE: Panama. Verguas: 5 mi. west of fruit. A, B from Nee & Dwyer 9164 (MO), Panama. Santa Fe on road past Escuela Agricola Alto Piedra on Pacific side of divide, 800- 1,200 m, 7. B. Croat 23059 (holotype, MO 2198065; isotypes, AAU, BR, C, CAS, CR, DUKE, F, L, LL, MEXU, NY, RSA, US, WIS). Figure 3. ad 20 m alta, ramunculis glabris ad strigillosis. Folia SUE petiolis 6- ticae ad elliptico-ovatae, (4.5-)5.7-8.2(- .8) cm latae, apice a acuminato, bas acuta ad obtusa. Inflorescentiae termi- 472 Annals of the Missouri Botanical Garden FIGURE 2. Cordia correae. — A. Flowering branch.—B. Flower with calyx and corolla opened.—C. Drie Coclé, Panira C from Mori & Kallunki 4998 (NY), Panamá, nales, cymosae d uuo s cm latae. Flores bisexu- ales; calyx campanulatus, 3-4. mM glaber, 3(-4)- lobatus; ig alba, tubularis, 5- 7. 2 mm longa, 5-lobata, lobis reflexis, ovatis; stamina 5, filis 4-6.5 mm longis, villosis, Paene oblongis, longis. Fructus dru- paceus, putamine oie DAD ovoideo, 8-11 mm longo, 6-8(-11) m Tree to 20 m tall, the twigs nearly glabrous to sparsely strigillose, often with considerable waxy deposits. Leaves persistent; petioles 6— 12 mm long, deeply canaliculate adaxially, nearly glabrous to sparsely strigillose; blades Volume 75, Number 2 1988 Miller 473 Revision of Panamanian Boraginaceae FIGURE 3. Cordia croatii. — A. Flowering branch. fruit. — A-C from Croat 23059 (MO), Veraguas, Panama; D from Tonduz 12520 (US), Alajuela, Costa Rica. elliptic to elliptic-ovate, (4.5-)5.7-8.2(-11.2) cm long, (2-)3-4.3(-5) cm wide, the apex acute to slightly acuminate, the base acute or less commonly obtuse and slightly decur- rent, the margin entire, the adaxial surface glabrous to papillose with widely scattered appressed hairs, the abaxial surface glabrous. Inflorescence terminal, cymose (3-)8-12 (-15) cm broad, the branches sparsely brown- strigillose to ferruginous-puberulent. Flowers sessile, monomorphic, the stamens longer than style; calyx campanulate, 3-4.3 mm long, 3.4-4 mm wide at mouth, the 3(-4) lobes ovate, rounded at apex, 1-1.8 mm long, ribs absent, glabrous; corolla white, tubular with —B. Flower with corolla opened. —C. Calyx.—D. Dried reflexed lobes, 5-7.2 mm long, 5-merous, the lobes ovate, 2-4.3 mm long, 1.5-3 mm wide, the tube 2.2-3 mm long, glabrous; stamens 5, the filaments 4-6.5 mm long, the upper 2-2.5 mm free, villous along the lower free portion, the anthers oblong, 1.1 mm long; ovary ovoid to conical, 1-1.6 mm long, 1- 1.4 mm broad, glabrous; disc crateriform, 0.4-0.5 mm tall, 1-1.1 mm broad, glabrous; style 2-2.3 mm long, the stylar branches 1.7-2.5 mm long, the stigma lobes discoid. Fruits borne in the saucer-shaped calyx, dru- paceous, glabrous, the stone inequilaterally broadly ovoid, 8-11 mm long, 6-8(-11) mm broad, the endocarp bony. 474 Annals of the Missouri Botanical Garden Distribution. Cordia croatii occurs in Cordia brevispicata var. hypomalaca Greenman, Publ. cloud forests from 800 to 1,200 m in ele- vation and is known from the San Ramón region of Costa Rica and the provinces of Coclé and Veraguas in Panama. Cordia croatii is distinct in its small elliptic leaves. It is known only from a few Pana- manian collections and several from the San Ramón region of Costa Rica, although further collecting efforts may reveal it in cloud forests in between. This species is somewhat unusual in sect. Myxa in that it occurs at relatively high elevations, unlike the majority of its rel- atives, which are usually found in lowland wet forests. Cordia croatii shares a three-lobed calyx with Cordia lasiocalyx Pittier, C. lu- cidula I. M. Johnston, and C. porcata No- wicke and is probably closely related to these species. Cordia croatii is named in honor of Dr. Thomas B. Croat who collected the type material and has contributed greatly to the study of Panamanian botany ns examined. PANAMA. COCLÉ: *Lallathin 1F (MO); 1-1 (MO). itional specim Cerro Pilón, 2,900 ft., Cordia curassavica (Jacq. Roemer & Schultes, Syst. Veg. 4: 460. 1819. Var- ronia curassavica Jacq., Enum. Syst. Pl. 14. 1760. TYPE: Curassao, Jacquin (not seen). Cordia d Kunth in Humb., Bonpl. & Kunth, Nov. Gen E j icana DC., Prodr. es eche: collector unknown (holotype, P, not n; microfiche, Cordia. brevispicata Mart ens & Galeotti aa Acad. Roy. Sci. Bru s 11(2): 18 Mexico. Pu 2 pond acán, js do. Cun 7192 (ho- lotype, BR; isotypes, BR, G, K). Cordia ideni DC., Prodr. 9: 493. 1845. TYPE: Mexico: Herb. Pavón (holotype, G-DC, not seen; microfiche, M SE hispida Benth., Bot. Voy. Sulphur 139. 1845. : Honduras: Gulf of Fonseca, Sinclair s.n. (ho- ie Cordia dece S. Watson, Proc. Amer. Acad. Arts 24: 62. 1889. TYPE: Mexico. Sonora: Guaymas, 1887, E. Palmer 281 (holotype, GH; isotypes, C, K, NY, , US (3)). Cordia socorrensis Brandegee, Erythea 7: 5. 1899. TYPE: M olima: Socorro Island, Mar.-June 1897, Anthony 384 (holotype, UC 78381; isotypes, DS, F, GH, K, MEXU, MO, POM, SD, US) Columbian Mus., Bot. Ser. 2: 338. 1912. TYPE: Mexico. Oaxaca: Cerro San Filipe, 1,700 m, 30 June 1907, Conzatti 1831 (lectotype, here des- rd F 225986; ; isolectotypes, F, GH). In de- Ing this varie gnat d Conzatti's rice at the Fi eld Museum as the type and listed two accession numbers (225986 and 246873). Neither sheet had been clearly marked as holotype by Greenman, and the better of the two specimens is — as a lectotype here to rectify this situ- ati Cordia Scusa J. F. Macbr., Contr. Gray Herb. 49: 16.1 TYPE: Mexico. Michoacán or Guerrero: 1898, E. oo 265 (holotype, GH; 1 Aug. isotypes, G (2), K Cordia chepensis a us U.S. Natl. Herb. 18: 253. 7. TYPE: Pan ama. Panamá: Chepo, d 1911, H. Pittier 4511 (holotype, US 679672; type, US). nds faa Pittier, Contr. U.S. Natl. Herb. 18: 253. TYPE: Costa Rica. Limon: Porto Limon, 27 cd 1911, H. Pittier 3641 (holotype, US 678699; type, GH). Cordia mollis Pittier, Contr. U.S. Natl. Herb. B 294. 9] . 19 Apr. 134 (holotype, US 472788). "H. Pittier Shrub to 2(-4) m tall, the twigs glabrous to strigillose or puberulent or rarely hirsute but always with small, globose wax particles. Leaves deciduous, on short spurs to 1 mm long; petioles 1-8(-21) mm long, strigillose or puberulent to hirsute; blades lanceolate to narrowly elliptic or elliptic-ovate, (1-)2-9.4 (716) cm long, 0.5-4(-7.3) cm wide, the apex acute, the base cuneate to acute and sometimes decurrent, the margin serrate, oc- casionally merely undulate, the adaxial sur- face scabrous to papillose, the abaxial surface strigillose with most hairs restricted to the major veins, or tomentulose. Inflorescence terminal, spicate, 1.5-8.8(-15) cm long, the peduncle 1.8 cm long, puberulent or strigil- lose to nearly glabrous. Flowers sessile, disty- lous; calyx campanulate, 2-3.2(-3.8) mm long, the 5(-6) lobes deltate; corolla white, tubular with reflexed to spreading lobes, (3.8-)4.8-6.8 (-8.2) mm long, 5(-6)-mer- ous, the lobes ovate to depressed ovate, 1.2- 1.8(-2.8) mm long, the tube 2.4-3.4 mm long; stamens 5, the filaments 3.2-5(-6) mm long, the upper 0.8-2 mm free, the free portion glabrous, puberulent to pubescent be- neath the point of insertion, the anthers el- Volume 75, Number 2 1988 Miller 475 Revision of Panamanian Boraginaceae lipsoid, (0.3-)0.7—1 mm long; ovary ovoid to broadly ovoid, (0.8-)1-1.2(-1.6) mm long; disc crateriform, 0.4—0.6(-0.8) mm tall; style (1.4-)2-4(-5.7) mm long, the stigma lobes clavate. Fruits drupaceous, red, 4%-% en- closed in the slightly accrescent calyx, the stone ovoid, (3.7—)4—4.5 (-6) mm long, 2.2- mm broad, the endocarp bony. Distribution. Cordia curassavica is common in a wide variety of habitats but is found most often in disturbed or dry areas from sea level to 2,000 m in elevation. This species ranges from Sonora and Baja Cali- fornia in northern Mexico south to northern South America and east to the West Indies. It is known from all of the provinces in Pan- ama. Cordia curassavica is extremely variable, and many of its variants have been recognized as taxonomically distinct by previous authors. Much of the variation throughout the range of C. curassavica is in overall size of the plants, size of leaves, and size of inflores- cences. Individuals from populations in Baja California and Socorro Island are quite small in stature with leaves much reduced in size; the synonym C. socorrensis Brandegee is based on a collection of this sort. The most diminutive plants occur in populations from the Tehuacan region of Puebla, Mexico, the area from which the type of another synonym, Cordia brevispicata Martens & Galeotti, was collected. The other extreme variant of Cordia cu- rassavica occurs along the Atlantic coast of Central America, and only in Panama can it be found on the Pacific slope. These plants are more robust and differ from other pop- ulations in being larger in all aspects as well as in having broader leaves. Also, the pubes- cence of the upper leaf surface differs from what is seen in other populations; the hairs are represented only by the persistent bases, lacking shafts. Cordia chepensis, based on a type collection from Chepo, and Cordia lit- toralis, based on a population from the At- lantic coast of Costa Rica, are synonyms of this sort. These were referred to as “‘typical Cordia curassavica” by Johnston (19494), who felt that with further study, several seg- regate species would be recognized. Numer- ous collections from all portions of the range indicate that there are no clear morphological discontinuities. Although populations of Cordia curassa- vica vary over its geographic range, a much greater component of this variation appears to be due to phenotypic response to local climate rather than genetic differences be- tween the populations. While collections made in the field from different regions often vary widely in appearance, most of these differ- ences are not evident in the plants that have been raised in the greenhouse from seed col- lected in Mexico, Nicaragua, and Panama. Adult plants raised from seed under uniform conditions from morphologically and geo- graphically diverse populations are often vir- tually indistinguishable. Cordia curassavica hybridizes with C. spi- nescens L. and with C. bullata (L.) Roemer & Schultes and probably hybridizes with sev- eral additional species (Miller, in prep.). Ob- servations made on numerous hybrids show that they vary in pollen stainability according to the parentage, but even if sterile, they are capable of persisting by spreading rhizoma- tously. As a result, these hybrids are repre- sented in herbarium collections, leading to confusion and the long list of synonyms as- sociated with this species. Although interspe- cific hybridization appears to be relatively un- common, it may be adding to the variability of populations through rare backcrossing to parent plants. Data indicate that Cordia curassavica is best treated in a broad sense, as much of the variability between populations is phenotypic. While there are considerable differences be- tween the extremes, none of the intermediates exhibit any significant reduction in pollen stainability. Despite this variability, Cordia curassavica is a well-marked species easily distinguished from C. spinescens by having lanceolate leaves and elongate, terminal, spi- cate inflorescences. Cordia curassavica dif- fers from Cordia guanacastensis Standley, 476 Annals of the Missouri Botanical Garden the other Central American species with which it could be confused, by having much more elongate spikes generally less than 8 mm broad and by having the peduncles nearly glabrous or puberulent to strigillose rather than hispid as in C. guanacastensis. Additional specimens examined. PANAMA. BOCAS DEL TORO: Santa Cat eee River bank and beach, Blackwell et al. 2710 (MO, SCZ, UC); vicinity of Almirante, Chan- guinola Canal, vdd 1389 (MO, SCZ); Chiriqui x Isla Colón, 0-12 Wedel 562 (GH, MO, U), 2478 (MO, NY, US); Chiriqui Lagoon, Columbus Island, W a 2608 (MO, US); Chiriqui Lagoon, Isla Colon, Wedel 2923 MO, NY, US). CANAL AREA: artlett & Lasser 16312 (GH, MIC erman area, Blum et al. 386 (MO, SCZ): Ce Navy Pipe Line along main dirt road, Correa & Haines 541 Agee road : near beach ~ Weather Station, Duke 4284 (MO); Cm Due. Dwyer 7190 (MO); Fort San Lorenzo, Ebinger 457 (F, MEXU, MO); Ancon Hill, 100-200 m, Killip 12059 (GH, NY, US), 12106 (US); near Fort Randolph, Maxon & Harvey 6506 (US); low woods E of Bella Vista, a suburb of Panama C ity, Maxon & Valentine 6945 (US); McDaniel 4996 (MO); thagres River, McDaniel 5179 (MO); Curundu, 30-40 m, Miller 1033, 1034, 1035, 1036, 1037, 1038, 1039, 1040, 1041, 1042, 1043, 1044 (MO); along Fort Sherman road (S2 or 82), Mori & Kallunki 2712 (MO); Ancon Hill, Standley 25207 (US); Balboa, Standley 25552, 27152 (US); near Fort Randolph, Standley 28604 (US); Mount Hope Cem- etery, Standley 28786 (US); vicinity of Fort icd Standley 31220 (US); Fort Sherman near road to Ga Tyson & Dw yer 1206 (MO, SCZ), Howard Air Force evil drop zone, Tyson 1865 (MO); Fort Clayton, Tyson et al. 2304 (NY); vicinity of Miraflores Lake, just outside Naval reserve, White 244 ( : : CHIRIQUÍ: along trail north of Cerro Punta, Croat 10477 (MO, NY). cocLÉ: vicinity of Valle, 800-1,000 m, Allen 100, 753 (F, MO); hills S of El Valle de Antón, 700 m, Allen 2510 (US (2); Rio Hato Airstrip, Blum & Dwyer 2474 (AAU, SCZ); Burch et al. 1148 (K, MO, UC, US); 10 km W of Agua Dulce on the Interamerican Highway, Correa 87 (MO, SCZ); 3 mi. NE of Antón, Croat 9618 (MO, SCZ); road from Pan-American High- way to El Valle, 100-1,000 m, D'Arcy & Sytsma 14647 (MO); savannas near El Valle, Duke & Mussell 6615 (MO); Penonomé, Du ver 2000 (MO (2); along El Valle de Antón, 1 km up road toward La Mesa near first waterfall, s. & Kauke 2753 (MO); W of Rio Guias, Gentry 5845 (MO); E seek slopes of the crater as El Valle de gees E 2 mi. W of town, 1,000 1 Luteyn 1245 (F, GH, MO); along the road to El Valle between the Pan- American Highway and El Valle, 400 m, Miller et al. 773 (MO); Agua Dulce, Pittier 4800 (NY, US); Rio Grande en Coclé, Rosario 20 (F); El Valle de Antón and vicinity, 500-700 m, Seibert 439 (F, MO); 3-6 km SE of El Valle de Antón, Wilbur & Luteyn 11765 (DS, GH, LL, MICH, MO, NY, RSA, US); between Agua Dulce and Antón, 15-50 m, Woodson et al. 1207 (F, MO, NY); between Las Margaritas and El Valle, Woodson et al. 1293 (F, MO, NY). COLÓN: d of Camp Pina, 25 m, Allen 3586 (F, G G, NY, ) 1 E from Puerto Pilón n road to ía Chiquita, Correa & Haines 235 (F, 0 (2); Mice) of San Miguel de la Borda, Croat 9867 Mo en Bellas and Salud, near sea level, Croat 36870 (Moy. pee ocean trail between Rio Indio and Miquel de la Borda, sea level, Croat 36913 (MO); Maria Chiquita, E of Rio Piedras towards Portobelo, Dwyer & Kirkbride 7775, 7788 (MO, ); road to Portobelo between Rio Piedras and Portobelo, roadside near ocean, Elias & Kirkbride 1648 (MO, UC); without definite locality, Kuntz s.n. (NY (2)); Salud, Lao & Hold- eridge 246 (MO); Nuevo i de beach and adjacent roadside, Lewis et al. 1856 (MO, US); mouth p Rio Piedras, beach and adjacent areas, Lewis et al. 3166 (MO, SCZ, UC); coastal thickets between Rio Gua ee and Rio Buenaventura, 7.5 mi. SW of Portobelo, Webster 16778 (MO); along pada between 5-7 mi. SW of Portobelo towards Maria Chiquita, Wilbur & Weaver 11177 (F, MICH, MO, NY, US); 6 mi. SW of Portobelo on the very edge of the Caribbean, Wilbur & Luteyn 11665 (DS, F, GH, LL, MICH, MO, NY, US). DARIÉN: coastal thicket near Jaque, Duke 10668 (MO); near Yav- iza, 50 m, Gentry & Mori 13500 (MO). HERRERA: alrede- dores de Ocú, Diaz 48 (F, MO); a between Las Minas and Pesé, (U. of Fla. site #4), c , Duke 12318 MO (2)); Ocú, Ebinger 1050 (F. e US); 10 mi. S of Oct on Las Minas Road, 300 m, Graham 251 (GH, MICH); 12.5 mi. S of Ocú, 1,200 ft., Lewis et al. 1633 ENCB, MO); 1 mi. N of Chupampa on the road to Ocú, Wilbur et al. ls (DS, F, GH, LL, MICH, MO). Los SANTOS: 1-2 mi. W of Candelaria, Duke 12435 (MO (2); N of Rio Caldera near Punta Mala, Stimson 5291 JC); Los Asientos, Wendehake 37 — — P wa & Lasser 16476 (GH, MICH, MO); 1.5 mi. above In- teramerican Highway on road to Cerro Campana, Croat 12040 (MO); ag area near the sea, Taboga, D'A > Dp’ d 6806 (MO (2), US); Cerro Campana, D ye 960 O); near beach at Nueva p orgona, Duke 4514 MOOD )); Cerro Campana, 2,400- 2,700 ft., Duke 8674, 10721 (MO); along road from Pn Highway to Coronado Beach, Duke 11805 (MO); Bar Mouth, Changni valley, Dunlap 132 (F); Rio near beach, Dwyer 1798 (MO); s Dwyer 1860 (MO); Cerro Campana, % of the w summit from Sree Highway, Dwyer et al. 4549 (MO, SCZ) between Rio Pacora and Chepo, poto savanna, Dwyer et al 5125 (MO, SCZ, UC); La Cam- pana, Cerro Campar Na, Pay 370 (F, MO); pes near Río Mar q: 502 (F, MO); Chagres, Fendler 130 , US); of Chorrera city limits, Folsom 3463 o Taboga Island, hill behind beach on main island, near sea level, Gentry 5732 (F, i y Og: Island, Killip. 3168 (US); next to bridge over small s stream 10.6 mi. W I Luteyn & Foster 1401 ages , MO); alon é Miller e d 42 (MO); Ponte Paitilla, Piper 5398 (US). n near Old Fort Lorenzo, mouth of Rio Chagres, Piper 5925, 5932 (US); Las Sabanas, Standley 25857 (US); near Punta Paitilla, Standley 26282 (US); vicinity of Juan Franco Race Track near Panama, Stand- Volume 75, Number 2 1988 Miller 477 Revision of Panamanian Boraginaceae ley 27795 (US); is a Standley 28021 (US); umba Muerto Road, near Panama, Standley 29785 (US); Nuevo San Francisco, “mansa 30738 (US); dirt road to Ojos de Agua where it branches off the carretera usage wa (between Panama City and Colon) about 5 of Panama City, Stimson et al. 5055 (GH, SCZ, UC) on lower Changuinola River, Stork 132 (UC, US); Goofy Lake, SW facing slope, 500 m, Sullivan 78 (MO); W slope of Cerro Campana, 2,300 ft., Tyson 4038 (MO, SCZ) slopes of Cerro Jefe beyond Cerro Azul between 4-8 mi. in mostly heavily wooded slopes, Wilbur & Weav- er 11345 (DS, GH, MICH, MO); weedy ao within i po, 7 (DS, GH, MICH, MO); Isla Taboga, ca. 0-186 m 1485 (F, MO, NY). SAN BLAS: vicinity " Puerto Obaldia, Croat 16880 (MO); Isla Soskatupu, Duke 8945, 10191 (MO); Ailigandi, San Blas Islands, Dwyer 6809 (MO, TEX); Island of Soskatupu, on the only hill on the island, 150 ft., Kirkbride 187 (MO, NY); on trail to inland village of Armila, 3-8 km SW of Puerto Obaldia, Mori et al. 6804 (MO, US). VERAGUAS: ca. 5 mi. N of Sole vicinity of Santa María River, Blum & Tyson 624 (MO, SCZ); irs ig ntiago, a 10734 (MO); roadside savanna 2-4 mi. E p pulp Je ca. 30 m, Duke 12366 (MO (2 >) Rue 12 mi. from Santiago toward divisa on po Highway, Dae & Kirkbride 7449 (MO); Santiago, 2 mi. W of Santiago on Transisthmian Highway, y, Dwyer et al. 7550 (MO, UC); Rio Gatu at intersectio h highway from Santiago to Santa F ( ; cliffs, and adjacent swamp, Lewis et al. 2849 (MO, NY, SCZ) road between Laguna La Yeguada and Calobre, pes 1472 (MO); Projecto Agro-forestal Alto Guarumo, N of Santiago, S of San Sebastian, 300-400 m, Meijer & Lao 362 (MO); road to Santa Fe, 15 km from Santiago, 150 m, Sullivan 409 (MO). Cordia eymosa (J. D. Smith) Standley, Publ. Field Mus. Nat. Hist., n ne 18: 981. 1938. Cornutia cym . D. Smith, Bot. Gaz. (Crawfordsville) 1 10. 1905. TYPE: Costa Rica. Alajuela: Paturages de la Palma, 1.460 m, 19 Nov. 1898, A. Tonduz 12555 (holotype, US 1323323; K, US (2)) Tree to 15(-30) m tall, the twigs ferru- ginous-tomentulose with scattered echinate hairs. Leaves persistent; petioles 15-41 mm long, shallowly canaliculate adaxially, ferru- ginous-tomentulose with scattered echinate hairs; blades dimorphic, the larger ones ellip- tic or elliptic-ovate, (15-)18-32(-37.5) cm long, (7.5-)11-20.5(-26) cm wide, the smaller ones orbicular to circular, 9.5-16 cm long, 8.5-17 cm wide, the apex obtuse or less commonly rounded or acute, the base obtuse or less commonly acute, the margin isotypes, entire, the adaxial surface strigillose, the abaxial surface soft-pubescent, with most of the the hairs restricted to the veins. Inflores- cence terminal, cymose, (14.5-)17-29 cm long, 18.5-31 cm broad, the peduncle (1-) 3.5-8.2(-11.2) cm long, ferruginous-tomen- tulose with scattered echinate hairs. Flowers sessile, unisexual by abortion, the plants dioe- cious, the male flowers with reduced styles, the female flowers with small, nonfunctional anthers. Female flowers with a tubular calyx ca. 3.7 mm long; corolla white, tubular with reflexed lobes, 4.6-5.3 mm long, 5-merous, the lobes oblong-ovate, 2-2.2 mm long, l- 1.2 mm wide, the tube 2.4-3.3 mm long; stamens 5, nonfunctional, the filaments ca. 3.5 mm long, the upper 0.5-1.8 mm free, glabrous or nearly so, the anthers ellipsoid, shriveled, 0.2-0.4 mm long; ovary ellipsoid to ovoid, 1.5 mm long, 0.8-0. road, glabrous; disc small, crateriform or indistinct from the base of the ovary; style 2.5-3.5 mm long, the stylar branches 1.2-2 mm long, the stigma lobes clavate to discoid. Male flow- ers with a campanulate calyx, 2.2-3 mm long, 2-3 mm wide at mouth, strigillose to puberulent, the 3-5 lobes depressed-ovate to deltate, ca. 0.8 mm long; corolla white, tu- bular-campanulate with reflexed lobes, 3.8- 4.7 mm long, the 5 lobes oblong-ovate, 1.9— 2.] mm long, mm wide, the tube 2- 2.7(-3.8) mm long; stamens 5, the filaments (2.5-)4-5 mm long, the upper 1.6-2.4 mm free, pubescent at and just above the point of insertion, the anthers ellipsoid to oblong, -] mm long; ovary ovoid, 0.6-1 mm long, 0.4-0.6 mm broad, glabrous; disc crateri- form, 0.3-1 mm tall, 0.6-1.2 mm broad, or occasionally not distinct from the base of the ovary, glabrous; style 0.7-1(-1.8) mm long, the stylar branches 0.3-0.5(-1) mm long, the stigma lobes filiform to clavate. Fruits seated in the cupulate calyx, white, drupa- ceous, glabrous, the stone inequilaterally ovoid, 7.3-10 mm long, 4.8-9 mm broad, endocarp bony, 1-locular. Distribution. Cordia cymosa ranges from Costa Rica south through Panama and 478 Annals of the Missouri Botanical Garden Colombia to Ecuador, mostly in cloud forests and rarely at low elevations in wet forests. In Panama it is known from Bocas del Toro, Canal Area, Coclé, and Panama. Cordia cymosa is a member of the C. panamensis group, one of the taxonomically most complex species groups of sect. Myxa. It is probably most closely related to C. pan- amensis but differs by having larger stature. Cordia cymosa is easily distinguished from the other members of the complex by its scattered echinate hairs on the stems, peti- oles, and peduncles. Additional specimens examined. PANAMA. CANAL AREA: Mt. Lerio, Va UE 152 (US EN BOCAS DEL TORO: region of Cer olorado, 4.3 mi. Chami, 1,500 m McPherson 9595 (MO). ee El Valle de Antón, diston 8809 (US). PANAMÁ: Cerro Campana, Folsom et al. 2312 (MO); Cerro Campana, trails just inside entrance to Parque Nacional, 850 m, Vue x Miller 998 (MO); along the Panamerican Highwa mi. E of highway checkpoint at turnoff to Chepo, Mais et al. 1018 (MO). Cordia dentata Poiret, Encycl. 7: 48. 1806. TYPE: Curasao: Von Rohr 1799 (holo- type, P in herb. Jussieu, not seen; mi- crofiche, MO) Cordia a Bertero ex Sprengel, Syst. Veg. 1 649 Veracruz: Linden 284 (holotype, BR; isotypes, "K, MICH). Cordia oe Bertol., Rendiconto Sess. Ordinarie Ac- cad. Sci. Ist. Bologna 1860-1861: 63. 1860; Mem. Reale Accad. Sci. Ist. Bologna, 11: 199, t. 11. 1861. TYPE: Guatemala: In Volcano d’acqua a Val- lasquezio (not seen). In leptopoda: K. Krause, Bot. Jahrb. Syst. 37: 628. . TYPE: Colombia: in planitiebus ad flumen Magdalena, prope Purificacion, 200-500 m, Leh- nn 7347 (not seen). Cordia. MoS asd Univ. bun Publ. Bot. 10: 187. 922. TYPE: Mex z: Remudadero, C. . isotypes, GH, MO, Purpus 8937 bip “UC: S). Tall shrub or tree to 7(-10) m tall, the twigs puberulent to nearly glabrous. Leaves semideciduous; petioles 8-20 mm long; blades elliptic to widely elliptic or ovate, occasionally obovate, 4.6-10 cm long, 3-7 cm wide, the apex obtuse, or less commonly acute or rounded, the base obtuse to rounded, rarel acute, the margin entire or slightly denticu- late, the adaxial surface strigillose to sca- brous, the hairs arising from a distinct cys- tolith, the abaxial surface nearly glabrous to puberulent with dense tufts of curly hairs in the axils of the major veins. Inflorescence terminal, cymose-paniculate, 15-20 cm broad, the branches puberulent to sparsely strigillose. Flowers sessile, distylous; calyx campanulate, 3- m long, circumscissile and tearing open somewhat unevenly, faintly 10(-12)-ribbed; corolla yellow to almost white, campanulate, 9-12 mm long, 5(-6)-merous, the lobes depresssed ovate, 1.8-4 mm long; stamens 5(-6), the filaments 4.7-9.4 mm long, puberulent at the point of insertion, the anthers ellipsoid to oblong, 1.2-2 mm long; ovary ellipsoid to globose, 1-1.5 mm long; style 3.5-5.3 mm long, the stigma lobes cla- vate. Fruits borne in the saucer-shaped calyx, translucent white, drupaceous, the stone el- lipsoid, symmetrical or nearly so, 9-11 mm long, 5.5-7.2 mm broad, the endocarp bony. Distribution. Cordia dentata occurs in dry forests from sea level to 400(-1,400) m in elevation from Mexico to northern South America and the West Indies. In Panama this species is known from the Canal Area and the provinces of Chiriqui, Herrera, Los San- tos, Panamá, and Veraguas. Cordia dentata is the most distinctive member of sect. Myxa in Central America. It is the only species with a yellow corolla that is relatively large, campanulate, and rather showy, and it is often cultivated for this reason, as well as for its sweet, edible fruits. It differs further from most other mem- bers of the section by having a circumscissile calyx. The unusual calyx and corolla of C. dentata suggest that it is not closely related to any of the other Panamanian members of sect. Myxa, and its relationships with species from other geographic regions are obscure. Johnston (1940) stated that the type of C. dentata was collected by Von Rohr, but the only sheet in the Jussieu herbarium is labeled as a Vahl collection. It was apparently col- lected by Von Rohr, and the sheet at Paris is a duplicate that Poiret received from Vahl. Cordia dentata is common throughout Volume 75, Number 2 1988 Miller 479 Revision of Panamanian Boraginaceae much of Mexico, Central America, the West Indies, and northern South America but has been only rarely collected in Panama. While most species of sect. Myxa occur in wet for- ests, C. dentata occurs in drier regions, often in second-growth and disturbed areas. It is frequently found growing along roadside ditches and fencerows from Costa Rica north- ward. Additional specimens examined. PANAMA. CANAL AREA: bur Lirio, Christopherson 125 (US); between Gorg and Mamei, 10-30 m, Pittier 2236 (NY, US). iai unt pes of David, Pittier 2820 (GH, NY, US); vicibit of San Felix, 0-120 m, Pittier 5448 (GH, NY), 5458 (US). HERRERA: Pesé, ca. 20 m, Allen 805 (MO, NY (2), US); Chitre, my i 314 (F, MO); ca. 2 Croat 4190 (MO, SCZ); lighway, Tyson et al. 3156 (MO, SCZ). PANAMÁ: near (no Kluge 52 (CFMR, F, US). vERAGUAS: 5 mi. E of Santiago, Tyson et al. 4289 (MO, SCZ). Cordia diversifolia Pavón ex A. DC. in DC., Prodr. 9: 474. 1845. TYPE: Nueva Espana, Pavón s.n. (holotype, G-DC, not seen; microfiche, MO; isotype, C). Cordia johnstonii Cuf., Arch. Bot. Sist. 10: 41. 1934. TYPE: Costa Rica. Atlantida: 28 mi. from Puerto món, Cufodontis 365 (not seen). Cordia pedes Lundell, ise gya H 49. 1968. TYPE: Guatemala. Petén: Remate, 26 Apr. 1960, E. Con- treras 894 (holotype, LL. isotypes, F, K). Shrub to slender tree to 5(-10) m tall, the twigs hispid to hirsute. Leaves persistent; pet- ioles 5-19 mm long, hispid to hirsute; blades narrowly elliptic or sometimes lanceolate or oblanceolate, 10.5-19 cm long, 3.2-6.8 cm wide, the apex acuminate to acute, the base acute, the margin with few, short, filiform teeth toward the apex or entire, the adaxial surface strigose, the abaxial surface soft stri- gose, the hairs mostly restricted to the veins. Inflorescence terminal, paniculate, 5.5-11 cm broad, the branches hispid to villous. Flowers sessile or on short spurs to 2(-4) mm long, unisexual by abortion, the plants dioecious, the male flowers lacking styles, the female flowers with small, nonfunctional anthers; ca- lyx tubular, 4.5-5.5 mm long, 10-ribbed, 3- 4(—5)-lobed; corolla white, tubular with re- flexed lobes, 8-9.5 mm long, 5-merous, the lobes oblong to obovate; stamens 5, in male flowers the filaments 7.7-9(-10.7) mm long, the upper (3-)4.5-5.3 mm free, glabrous or the lower free portion pubescent, the anthers oblong, 0.4-1.8 mm long, in female flowers the stamens much reduced, the anthers 0.4— 0.7 mm long; ovary ovoid, 0.4—1.2 mm long; disc widely depressed obovoid, 0.5-1 mm tall, style 7-8.5 mm long, absent in male flowers, the stigma lobes filiform. Fruits borne in the expanded, saucer-shaped calyx, white, dru- paceous, the stone ovoid to ellipsoid, 7-9 mm long, 4-6 mm broad, the endocarp bony. Distribution. Cordia diversifolia oc- curs in wet forests of the Atlantic coast from sea level to 600 m in elevation from Mexico to Panama. In Panama this species is known from Bocas del Toro and the Canal Area. Cordia diversifolia is an uncommon species of Atlantic lowland forests in Central America but is apparently quite common in lowland Bocas del Toro. It is distinctive in its narrowly elliptic to lanceolate leaves, leaf margins with short, filiform teeth, costate calyx, and fili- form stigma lobes. This is the most widespread member of a small complex of closely related Central American species that is defined by the presence of denticulate leaf margins, cos- tate calyces, and consistent dioecy. The group also includes C. cordiformis I. M. Johnston, C. salvadorensis Standley, and C. skutchii I. M. Johnston, all of which occur from Nic- aragua northward. Additional specimens examine a PANAMA. BOCAS DEL TORO: Almirante, out along the road to the “Bomba,” Blum 1314 (MO, SCZ); Lincoln Creek, Carleton 40 (NY, US); Changuinola hei Cooper & Slater 61 (CFMR, F), 112 (CFMR, F, US); region of Almirante, Cooper 342 (CFMR, F (2), NY, US); Changuinola, Croat 16309 F, , NY); Changuinola Valley, Dunlop 175 (F, US); Changuinola to 5 mi. S at junction of rios Changuinola and Terebe, 100-200 ft., Lewis et al. 795 (MO, UC, US), 937 (K, MO, US), 938A (MO), 946 (F, MO, NY, UC, US); Chiriqui to 5 mi. S along Rio Guarumo, Lewis et al. 1994 (F, MO (2), US); Almirante, near road to Chiriqui, McDaniel 5077 (MO); Shepherd Island, McDaniel 5077, 5161 (MO); on lower Changuinola Riv- er, Stork 284 (UC, US); Almirante region, Taylor & Slater 61 (US); Chiriqui Lagoon, Water Valley, Wedel 632 (MO); Chiriqui Lagoon, Wedel 1070 (MO); Chiriqui Lagoon, Old Bank Island, Wedel 1894 (GH, MO, US), 2051 (MO, US); Chiriqui Lagoon, Pumpkin River, Wedel — 480 Annals of the Missouri Botanical Garden 2580 (MO, US); Chiriqui Lagoon, Wedel 2595 (MO, US); vicinity of Nievecita, ca. 0-50 m, Woodson et al. 24 (F, MO, NY); Rio O between Finca St. Louis and Konkintoe, ca. 10-50 m, Woodson et al. 1895 (MO, NY). CANAL AREA: Gamboa Naval Reservation, Ebin- ger 482 (BR, ENCB, F, GH, MO). Cordia dwyeri Nowicke, Phytologia 18: 419. 1969. TYPE: Panama. Colón: Santa Rita Ridge, 19 km from the Transisth- mian Highway, 28 Jan. 1968, D. Dwyer 8857 (holotype, MO 2518567; isotypes, F, GH, MO). Sparsely branched tree to 10(-20) m tall, the twigs reddish-brown villous. Leaves per- sistent; petioles (2-)4—8(- 1 4) mm long, thick, villous; blades coriaceous, bullate, ovate to elliptic, (9-)19-35(-60) cm long, (5-)8-16 (-25) cm wide, the apex acuminate to acute, the base rounded to obtuse or rarely subcor- date, the margin entire, revolute, the adaxial surface distinctly bullate, often drying with a silver cast, nearly glabrous with a few scat- tered, appressed hairs, the abaxial surface brown pubescent to densely brown pilose. In- florescence axillary or terminal, cymose, 4- 15(-23) cm broad, the branches densely ve- lutinous to tomentose, the hairs reddish brown. Flowers sessile, bisexual, monomorphic; calyx cupulate, 6-7 mm long, ribs absent, densely strigillose, with 2-3 unevenly deltate lobes; corolla white, cupulate with reflexed lobes, 8— 9.5(-10.6) mm long, 5-merous, the lobes ovate to widely elliptic, 3-4 mm long, the tube 4.5-7 mm long; stamens 5, the filaments (5-)8-11(-13) mm long, the upper (2-)4.5- 6.5 mm free, densely pubescent just above the point of insertion, the anthers ellipsoid to oblong, 1.4-1.8 mm long; ovary ovoid to conical, 1-1.7 mm long, glabrous; disc trans- versely oblong, 0.4-0.8 mm long; style 5- 7.4 mm long, the stigma lobes clavate. Fruits half enclosed in the slightly accrescent calyx, white, drupaceous, inequilaterally ovoid, the stone ca. 1.7 cm long, ca. 1.3 cm broad, the endocarp bony, pebbled on the surface. Distribution. | Cordia dwyeri occurs in wet forests from sea level to 450 m in ele- vation and ranges from southern Nicaragua to northern South America. In Panama this species is known from the provinces of Bocas del Toro, Colón, and Panamá. Cordia dwyeri appears to have no close relatives in Central America and is very dis- tinctive in its habit of growth and in its ex- tremely large, bullate leaves. Its closest rel- ative is probably C. trichoclada DC. of South America, a species that has large bullate leaves but differs in having a costate calyx. Cordia dwyeri is apparently common throughout lowland wet forests of Costa Rica and Pan- ama, although it is not often collected. This may be due to relatively rare flowering or may relate to the difficulty of preparing spec- imens from such bulky plants. Additional specimens examined. PANAMA. BOCAS DEL TORO: Palo Blanco, Gordon 95c (MO). coLÓN: East Santa Rita Ridge, Correa & Dressler 663 (MEXU, MO); Santa Rita Ridge, Croat 13856 (MO). PANAMA: El Llano-Carti Highway, about 8 km N of El Llano, Dressler 4573 (ENCB, F, MO); forest and roadside between 6-12 km N of El Llano on Carti Road, 1,200 ft., Hammel 866 (MO); El Llano-Carti Road, 7.9 km from Interamerican Highway, 400 m, Miller et al. 872 (MO); El Llano-Carti Road, 7.9 km from Interamerican Highway, 350 m, Mori & Kallunki 5608 (AAU, MO, NY); along El Llano- ala Road, 7 km N of Pan-American Highway at El Llan 450 m, Nee 10402 (MEXU, MO, US); El Llano- Cart Road, ca. 9 mi. from Pan-American Highway along newly cut bypass, 350-400 m, Sytsma 4145 (MO). Cordia eriostigma Pittier, Contr. U.S. Natl. Herb. 18: 251, fig. 101 17. TYPE: Colombia. Cauca: El Paso de la Balsa, on the Cauca River, near Jamundi, 480 m, 10 Feb. 1906, H. Pittier 1489 (ho- lotype, US 531695). Tree to 15(-30) m tall, the twigs minutely brown strigillose, later Rye tent; petioles (7-)10-24(-30) sparsely strigillose; blade elliptic or slightly ovate or obovate, occasionally narrowly ellip- tic, (5-)6-16(-22.5) cm long, (2.4-)4.3- 7.5(-11) cm wide, the apex usually obtuse and acuminate at the very tip, occasionally acute or rounded, the base obtuse to acute, the margin entire, the adaxial surface gla- brous or nearly so, usually with small papillae, the abaxial surface with few, small, scattered hairs, some attached medianly. Inflorescence Volume 75, Number 2 1988 Miller 481 Revision of Panamanian Boraginaceae terminal, cymose, 8-13(-17) cm broad, the branches brown strigillose. Flowers sessile, bi- sexual, monomorphic; calyx cupulate to cam- panulate, 2.8-3.5(-4.7) mm long, ribs ab- sent, evenly strigillose, opening without distinct lobes or unevenly 3-5-lobed; corolla white, campanulate, 5.7—6.3(- 7.3) mm long, 5(-6)- merous, the lobes deltate to shallowly deltate, (2.5-)3-4.3 mm long, the tube 2.4-3.4 mm long; stamens 5(-6), the filaments (4—)5-5.5 mm long, the upper (1.4-)2-2.5 (-3) mm free, puberulent below insertion, the hairs often spreading onto corolla tube; anthers oblong to ellipsoid, 1-1.7 mm long; ovary ovoid to broadly ovoid, 1.3-2 mm long, glabrous or with short bristles on the upper portion; disc crateriform, small; style 1.5-2.6 mm long, the stigma lobes spatulate to discoid. Fruit borne in the saucer-shaped calyx, red to or- ange at maturity, drupaceous, glabrous, the stone inequilaterally ovoid, 6-7 mm long, 3.5- 5.5(-6) mm broad, the endocarp bony. Distribution. Cordia eriostigma occurs in moist to wet forests at 200-1,200 m in elevation and ranges from Mexico to Colom- bia. In Panama this species is known only from the province of Cocle. Cordia eriostigma is uncommon and is often confused with C. collococca, from which it can be distinguished by having persistent leaves with nearly glabrous abaxial surfaces, bisexual flowers, campanulate corollas, val- vate calyx lobes, and ovaries with distinct bristles on the upper portion. Cordia eriostig- ma differs further from C. collococca in usu- ally being found above 600 m in elevation. Additional specimens examined. El Valle site area of WEP 1071 (MO, US); El Valle de Anton, Lao 281 (F, Cordia globosa (Jacq.) Kunth in Humb., Bonpl. & Kunth, Nov. Gen. Sp. 3: 76. 1818; Varronia globosa Jacq., Enum. Syst. Pl. 14. 1760. TYPE: not seen. ropra humilus Jacq., Enum. Syst. Pl. 14. 1760. Cor- ı humilus (Jacq.) D. Don, Gen. Hist. 4: 383. tes Cordia globosa var. humilus (Jacq.) L M. PANAMA. COCLÉ: COR, Cerro Pilón, Kirkbride cea J. Arnold Arbor. 30: 98. 1949. TYPE: not drda jocmeliana K. Krause, Bot. Centralbl. 32: 344. 1914. Varronia jacmelia na (K. Krause) Friesen, Bull. Se Bot. Genéve, Ser. 2, 24: 177. 1933. : Haiti. Ouest: near Jacmel, Krause 11808 Eum phe Varronia alae sa s var. mexicana Friesen, Bull. Soc. Bot. Genéve, Ser. 2, 24: 162, t. 1, f. 4. 1933. Varronia mexicana Friesen, Bull. So 24: 162. 1933. TYPE: navaca, 5,000 ft., 6346 (holotype, G; isotypes K, L, LE, MEXU (2), MO, NY, US (3 ). Shrub to 3(-4) m tall, the twigs strigillose. Leaves deciduous, on short spurs to 1 mm long; petioles (2-)3-12(-25) mm long, strigil- lose, the hairs appressed to spreading; blades ovate to lance-ovate, (1.3-)2-5.7(-9) cm long, (0.7-)1-2.7(-5) em wide, the apex acute, the base subobtuse to acute and de- current along the petiole for a short distance, the margin serrate, usually unevenly so, the teeth usually blunt, occasionally short-apic- ulate, the adaxial surface usually + smooth, occasionally slightly bullate, strigose to strigil- lose, or scabrous, the abaxial surface strigil- lose to strigose, most of the hairs restricted to the veins, rarely nearly tomentose. Inflo- rescence subterminal, a dense, globose head, 8-14(-16) mm broad, the peduncle 0.5-1.5 (-3.6) cm long, strigillose. Flowers sessile, distylous; calyx campanulate, (2.3-)3-4(-4.2) mm long, ribs absent, the 5 lobes deltate to triangular with prolonged filiform tips 2-4 mm long; corolla white, tubular, (5-)6-8 (- 10) mm long, undulate to shallowly lobed, the lobes scarcely separate to transversely ellip- tic-oblong, the tube (2.4-)3-4(-4.5) mm long; stamens 5, the filaments (4.5-)5-8(-9) mm long, the upper (1.3-)2-3(-4) mm free, pu- berulent in a ring in the mouth of the corolla tube, the anthers ellipsoid, 0.8-1.1 mm long; ovary ovoid, 1-1.5(-1.8) mm long; disc cra- teriform, 0.5-1 mm tall; style (3-)4.5-7 (-7.5) mm long, the stigma lobes filiform to clavate. Fruits drupaceous, red, the stone ovoid, 3.5-4.8 mm long, (1.5-)2-3.8 mm broad, the endocarp bony. Distribution. Cordia globosa occurs in dry to moist forests from sea level to 600 m 482 Annals + pw pl Garden in elevation and ranges from southern Florida and Mexico to northern South America and the West Indies. This species is known in Panama from one collection. Cordia globosa is a common weedy shrub throughout Central America, the West Indies, and parts of northern South America, but apparently rare in Panama. Cordia globosa is most closely related to C. bullata, which ranges from Costa Rica to Mexico and the West Indies, but the former differs in having corollas more than 5 mm long and peduncles less than 3.6 cm long. The leaves of C. glo- bosa are also generally smaller and have a less prominent indument. Nevertheless, these two species are often difficult to separate, and interspecific hybridization appears to occur in some populations in Nicaragua, Honduras, and Yucatan (Miller, in prep.). Jacquin did not specify a type when he published the names Varronia globosa and V. humilus. None of his collections of this species seem to be present in the Willdenow or Linnaean herbaria. There may be a spec- imen at BM (Stafleu, 1967), but Johnston (1949c) stated that no type was preserved. Proper lectotypification will have to await ex- amination of specimens in European herbaria. An odd situation exists with Varronia hu- milus var. mexicana and Varronia mexi- cana, two names published by Friesen based on the same type for either a new species or a new variety. Friesen did not indicate at which rank he felt it should be treated. dditional specimens examined. PANAMA. PANAMA: Punta Paitilla, Standley 26268 (US) Cordia inermis (Miller) I. M. Johnston, J. Arnold Arbor. 30: 95. 1949. Lantana inerma Miller, Gard. Dict. ed. 8, 1768. TYPE: Mexico. Veracruz: Houston s.n. (holotype, BM, not seen). Cordia cana Martens & Galeotti, as Acad. Roy. Sci. Bruxelles 11: 331. 1844. TYPE: Mexico. Oaxaca: Bois de la late Pacifique, 1840, Galeotti 7140 (holotype, BR). Cordia insularis Greenman, Proc. Amer. Acad. Arts 33: 482. 1898. TYPE: Mexico. Nayarit: Tres Maria Islands, 3-25 May 1897, E. W. Nelson 4296 (ho- lotype, GH; isotype, US). Erect shrub to 2(—3) m tall, the twigs strig- illose to puberulent when young, later gla- brous and lenticellate. Leaves deciduous; pet- ioles (2-)4-12(-22) mm long, strigillose to puberulent; blades elliptic-ovate to narrowly elliptic or elliptic-lanceolate, (2-)3.5-10(-14) cm long, (0.7-)1-3.5(-6.8) cm wide, the apex acuminate to acute, the base attenuate, the margin serrate, the adaxial surface strigillose to strigose, the abaxial surface strigillose. In- florescence internodal or terminal, a small globose head, (3-)4-7(-10) mm broad, the peduncle (0.7-)1-5(-9) cm long, strigillose to puberulent. Flowers sessile, appearing uni- sexual, the plants subdioecious, the male flow- ers with reduced styles, the female flowers smaller than the males, with small nonfunc- tional anthers; calyx campanulate, 1.8-3 mm long, ribs absent, strigillose, the (4—)5 lobes deltate, 0.5-1 mm long; corolla white to greenish white, 2.5-3.5(-3.8) mm long, trun- cate to undulate but lacking distinct lobes, puberulent to pubescent in the middle of the tube; stamens 4—5, in male flowers the fila- ments (3.3-)3.5-4 mm long, the upper l- 1.6 mm free, puberulent at insertion, the anthers oblong, 0.7-1 mm long, in female flowers the filaments 2-2.5 mm long, the upper 0.3-0.5 mm free, puberulent at inser- tion, the anthers ellipsoid, ca. 0.3 mm long; ovary ovoid to nearly spherical, 0.5-1 mm long; disc crateriform, 0.2-0.6 mm tall; style in male flowers 0.7-1 mm long, the stigma lobes much reduced, filiform to slightly flat- tened, the style in female flowers 2-3.5 mm long. Fruits drupaceous, (3-)4-5.5(-6.5) mm long, 2-3(-3.2) mm broad, V4-24 enclosed in the slightly accrescent calyx, the calyx and fruit bright red at maturity, the stone ovoid, the endocarp bony. Distribution. Cordia inermis occurs in disturbed areas from sea level to 1,100 m in elevation and ranges from Mexico to Panama. In Panama this species is known from the Canal Area, Chiriqui, and Panamá. Cordia inermis is one of the most common pi species of the genus in Central Amer- ca. It reaches its southern limit in Panama Volume 75, Number 2 1988 Miller 483 Revision of Panamanian Boraginaceae where it is known from relatively few collec- tions. In Panama, C. inermis can be confused with C. linnaei, from which it differs in having terminal or internodal rather than axillary inflorescences. Cordia inermis has been described as dioe- cious (Opler et al., 1975). The female flowers have small nonfunctional anthers, and the male flowers have reduced styles and stigmas. The pollen from anthers of female flowers is small, deformed, and completely nonstaina- ble. The female flowers appear to be 100% male sterile, but there is no indication of com- plete female sterility in male flowers, even though the gynoecium is reduced. In fact, some collections of male plants have both flowers and fruits on the same branch, and this species is probably subdioecious. Additional specimens examined. PANAMA. CANAL AREA: Quarry Heights, hilltop, Dwyer 2610 (MO (2), NY, US). CHIRIQUÍ: 1 mi. W of airport at Puerto Armuelles, near sea level, Croat 22531 (F, MO); Rio San ipit. 2 mi. W of David, 150 ft., Tyson 912 (MO, RSA). PANAMÁ: Taboga Island, Standley 27039 (US); Woodses et al. 1478 (F, MO, NY). Cordia lasiocalyx Pittier, Contr. U.S. Natl. Herb. 18: 251. 1917. TYPE: Panama. Darién: open field around Garachine, sea level, 12 Feb. 1912, H. Pittier 5694 (holotype, US 715984; isotype, F). Shrub or small tree to 5(-10) m tall, the branching pattern divaricate, the twigs gla- brous. Leaves persistent; petioles (3-)4-7 (710) mm long, canaliculate adaxially, gla- brous; leaf blades elliptic-oblong to slightly obovate, (6-)9-13(-16) cm long, (2-)2.6- 5(-5.4) cm wide, the apex abruptly caudate, the caudex (1-)2-3(-3.5) cm long, the base acute and slightly decurrent, the margin en- tire, the adaxial surface glabrous, the abaxial surface sparsely and minutely strigillose. In- florescence terminal, cymose, 2.5-6.5(-10) cm broad, the peduncle 2.5-5 cm long, gla- brous or with a few widely scattered hairs. Flowers sessile, monomorphic, the stamens exceeding the style; calyx cupulate to cam- panulate, 3.8-5 mm long, 3.6-4 mm wide at the mouth, ribs absent, glabrous or nearly so, the 3 lobes deltate, 1-2 mm long; corolla white, tubular, 7.6-10 mm long, 5-merous, the lobes oblong, 3.3-4.6 mm long, 2-2.5 mm wide, the tube 4.3-6 mm long; stamens 5, the filaments 6-9.5 mm long, the upper 2.6-3 mm free, pubescent at insertion, the anthers oblong, ca. 1.3 mm long; ovary ovoid to broadly ovoid, 1.3-1.8 mm broad, gla- brous; style 3.5-5 mm long, the stylar branches 1.6-2.2 mm long, the stigma lobes clavate. Fruits borne in the cup-shaped calyx, white, drupaceous, glabrous, the stone strong- ly inequilaterally ovoid, 10-11 mm long, 6.4- 9 mm broad, ruminate, the mesocarp muci- laginous, endocarp bony, 1-locular. Distribution. Cordia lasiocalyx occurs in moist forests from sea level to 800 m in elevation and is known only from Panama, where it occurs in Bocas del Toro, the Canal Area, Coclé, Darién, and Panama. Cordia lasiocalyx is a member of a tax- onomically difficult species complex that is characterized by glabrous leaves, a three-lobed calyx, and white fruit. Its caudate leaf apex readily separates it from its closest Central American relatives, C. croatii, C. lucidula, and C. porcata, all of which have acuminate leaf apices. Cordia lasiocalyx is probably most closely related to C. lomatiloba I. Johnston, a species of Amazonian South America. Additional specimens examined. PANAMA. BOCAS DE ToRO: Island Potrero, se mi Valley. Dunlap 329 (F, S). CANAL AREA: Barro Colorado rie "dad 4601 (F, NY, SCZ (2)), 8568, 8572 (DUKE, F, MO, NY, SCZy Foster 833 A Knight 69-20 (MO (2); Oppenheimer 837 (DUKE, MEXU, MO, TEX); Shat- tuck 813 (F, MO (3)), 853 (F, MO (2), US). COCLE: El Valle de Antón, Allen 2302 (MO); Cerro Pilon, 2,500- 3,000 ft., Dwyer 8315 (DUKE); along ridge of Cerro Gaital, N slope of mountains near La Mesa, N of El Valle, 800-900 m, Knapp & Dressler 4879 (MO); La Mesa, 2 km W of Cerro Pilon, 860 m, Sullivan 453, 498 (MO). DARIEN: Rio Pirre, Duke & Bristan 8282, 8300 (MO); Puerto Indio, less than 50 m, Hammel 1080 (MO); Trail from Canglon- Yaviza Road to Rio Chucanaque, 7.7 mi. E of Canglon, 50 m, Knapp & Mallet 3956 (MO); forests around Pinogana, Pittier 6560 (US). PANAMÁ: Rio Maje, along river from waterfalls near Bayano Lake to finca of Choco Indian Eduardo Maycha, ca. 2 mi. upstream, 30- 60 m, Croat 34581 (MO); vicinity of El Llano, Duke 5837 (MO); del G.M.I. Isla Bayano, Garibaldi sí " MEXU, MO); Cerro Campana, 900 m, Knapp & 2308 (MO); Cerro Campana, above Su Lin Motel, hope et al. 4218 (MO). 484 Annals of the Missouri Botanical Garden Cordia leslieae James S. Miller, sp. nov. TYPE: Panama. Panamá: Cerro Jefe 5.8 mi. above Lago Cerro Azul, 840 m, 30 July 1983, James S. Miller & Leslie A. Miller 886 (holotype, MO 3386970). Figure 4. ad 7 m alta, ramunculis glabris, subalatis. Folia persistentia, petiolis 0.8-2 cm longis, applan ind uncia continuis secus jc lamina coriacea, obovata, 9.2- 2 cm longa, 5.3- m. cm lata, apicibus te tuse ad acuminatis, basibus bisexualis; calyx tubulo-campanulatus, 5.5-6 mm longus; 8-10 mm longa, poen lobis ovatis, in- crassatis ad apices, stamina 5. Fructus drupaceus, au- rantiacus, putamine inaequilateraliter punto 1.4 c longo. Tree to 7 m tall, the twigs glabrous, sub- alate. Leaves persistent; petioles 0.8-2 cm long, flattened in cross section and canalicu- late adaxially, the bases continuous along the stem for a short distance; blades coriaceous, obovate, 9.2- 16.2 cm long, 5.3-8.4 cm wide, the apex variable, retuse or acuminate, the base cuneate, the margin entire, revolute, the adaxial surface glabrous, slightly papillose, the abaxial surface glabrous. Inflorescence terminal, cymose-paniculate, 5.5-7.5 cm long, 5-9.5 cm broad, with 50-120 or more flowers, the branches sparsely strigillose. Flowers sessile or nearly so, bisexual; calyx tubular-campanulate, 5.5-6 mm long, 3-3.4 mm wide at the mouth, ribs absent, d strigillose, the 5 lobes deltate, 1 - 1.4 mm long; corolla white, tubular with reflexed lobes, 8- 10 mm long, 5-merous, the lobes ovate to oblong with a triangular thickening at the apex, 4 mm long, 2-3 mm wide, the tube 4— 6 mm long; stamens 5, the filaments 9.3 mm long, the upper 4.3 mm free, pubescent at insertion, the anthers oblong, 2 mm long; ovary obloid, 1.8 mm long, 0.7 mm broad; style 5.5 mm long, the stylar branches 2 mm long, the stigma lobes clavate to discoid. Fruits seated in the cupulate, slightly accrescent ca- lyx, orange, drupaceous, the stone inequilat- erally ovoid, 1.4 cm long, mesocarp muci- laginous, endocarp bony, 1-locular. Distribution. Cordia leslieae occurs in cloud forests at 800-1,000 m in elevation and is known only from Cerro Jefe, in the province of Panamá. Cordia leslieae is apparently not closely related to any other species of Cordia in Central America. This species is distinct in having sub-alate stems and petioles and in having thickened areas at the ends of the corolla lobes. It is named in honor of my wife, Leslie Miller, who assisted me with field stud- ies in Panama and discovered the tree from which the type collection was made. Additional specimens examined. | PANAMA. PANAMA: Cerro Jefe, Dressler 3489 (MO, PMA); Cerro Jefe region roadside pA Be 200-800 m, Hammel 4850 (MO); Cerro Jef Panama City, forested slopes below summit, 850- id m, MePherson 9735 (MO); Cerro Jefe, near summit, along road to east about a quarter mile below tower, 750-850 m, McPherson 11191 (MO); Cer- ro Jefe, 850-900 m, Sytsma 2009 (MO Cordia linnaei Stearn, J. Arnold Arbor. 52: 627. 1971. TYPE: Jamaica. St. Andrews: pastures behind Hope Gardens, 600-700 ft., 22 Oct. 1956, Proctor 15789 (ho- lotype, BM, not seen; isotype, IJ). Shrub to 4 m tall, the twigs coarsely pu- bescent, the hairs erect, brown. Leaves de- ciduous, on short spurs 1-1.5 mm long; pet- ioles (1-)1.5-3(-7) mm long, pubescent, the hairs erect, brown; blades ovate to lanceo- late, (2-)3.4-9(-10.6) cm long, (0.9-)1.2- 2.9(-5.3) cm wide, the apex acute to acu- minate, the base cuneate, the margin sharply serrate, the adaxial surface coarsely puber- ulent, the hairs short, slightly swollen at the base, appressed, the abaxial surface softly pubescent, the hairs wavy, erect. Inflores- cences numerous, axillary, capitate, 7-11 mm broad, the peduncle (0.5-)0.8-2.8(-6.4) cm long, pubescent, the hairs erect, brown. Flow- ers sessile, distylous; calyx cup-shaped, 2- .9(-2.7) mm long, ribs absent, strigillose, the 5(-6) lobes deltate to shallowly triangular, sometimes with a short apiculate tip; corolla white, tubular, 3-3.8(-4.2) mm long, trun- cate at the apex; stamens 5, the filaments (2.5-)3-3.5(-3.8) mm long, the upper 0.3- 1(-1.2) mm free, glabrous to canescent at insertion, the anthers ellipsoid, 0.4-0.6 mm Volume 75, Number 2 1988 Miller 485 Revision of Panamanian Boraginaceae FIGURE 4. & Miller 886 (MO), Panamá, Panama. long; ovary ellipsoid, 0.6-1 mm long; disc crateriform to nearly flat, thin, ca. 0.2 mm tall; style (2.1-) 2.5-3(-3.5) mm long, the stigma lobes filiform or nearly so. Fruits dru- paceous, 3.5-4(-5) mm long, (2.6-)3- 3.5(-5) mm broad, 25 to nearly totally en- closed in the calyx, red, the stone ovoid to broadly ovoid, the endocarp bony. Distribution. | Cordia linnaei occurs in moist to wet forests from sea level to 900 m in elevation from Mexico to Colombia and in the West Indies. In Panama, C. linnaei is known from the Canal Area, Colón, Darién, Panamá, San Blas, and Veraguas. Cordia leslieae. — A. Fruiting branch. —B. Sub-alate twig with petiolar attachment. From Miller This species was treated as Cordia lineata (L.) Roem. & Schultes in the Flora of Pan- ama (Nowicke, 1969), as it has been in most recent works, but Stearn (1971) showed that this name is based on a long string of illegit- imate names. It is easily recognized by its numerous, small, axillary inflorescences, which never branch cymosely as they do in C. bi- urcata. The most likely Panamanian species with which C. linnaei could be confused is C. inermis, which has terminal and internod- al, rather than axillary, inflorescences. Additional specimens examined. ANAMA. CANAL P AREA: near beach, at Ft. Kobbe, Duke 4195 (MEXU, MO); along abandoned road C29 just N of Las Cruces 486 Annals of the Missouri Botanical Garden Trail, 6 km E of Gamboa, 150 m, /Vee 9042 (MEXU, MO, US); Balboa, Standley 25474, 26068, 29245 (US). COLÓN: 20 km from Transisthmian Highway on Santa Rita Ridge, SE facing slope, 400 m, Knapp & Schmalzel 1765 (MO); Santa Rita Ridge, ca. 4-5.5 mi. E of Trans- isthmian Highway, Lewis et al. 5267 (MO); lumber road at about 8 km NE of Santa Rita Ridge along ridge, 650 ft., Wilbur & Weaver 108. 30 (DUKE); wooded slopes on Chepigana, Duke & Briston 264 (DUKE, MO); "His Boca Grande, Duke 8846 (MO). PANAMÁ: bon of Cerro Jefe, Altas de Pacora, 2,400 ft., Antonio 3218 (MO); vicinity anam Highway, Bartlett & Lasser 16954 (MO), Interamerican Highway on road to Cerro UE ‘Croat 12057 (F, MO, NY); Cerro Jefe, D'Arcy 9748 (MO); Panamerican d at Rio Mamonica, 4 mi. beyond H, MO); grasslands on Cerro Cam- , Duke 8677 (DUKE, MO); end of road near Rio enirn road is 2 km N of cement plant on Colón Highway, ca. m E of turnoff to end of road, 500 ft., Hammel 915 (MO); summit of Cerro Jefe and forests along road beyond a Hayden 1025 (MO); in woods near Panama, Hayes 559 (K); Punta Paitilla, d E of Panama City, "Heriberto 212 (GH, US); Cerro e, 4.8 mi. TE reus Panamerican Hig (MO); El Llano-Carti Road, anamerican Highway at El Llano, 300 m, 'Nee 7 7902 (MO) Rio — Standley 28153 (US); Cerro Jefe, 850-900 m, Syt 1959 (MO); slopes of Cerro Jefe beyond bn re I be. tween 4-8 mi. in mostly heavily odd slopes, Wilbur & Weaver 11348 Cowie =x ween Gane and e. ca. 25 m, Woodson et al. 1 (MO). s AS: Perm Cooper 276 (F, GH, NY, Ds. hills SE M Puerto Obaldia Croat 16737 (MO (2), SCZ); F ND on main land in front of Ustupo, D'Arcy 9481 (MO), Maletappi, i , Duke 8487 (MO); along of Rio Mulatupo, Elias 1742 (MO); mainland opposite Playón Chico, 0-3 mi. from Caribbean, 0-200 m , Gentry fade (MO); mainland opposite Playon Chico, 0-3 mi. Caribbean, 0-200 m, Gentry 6415 (MO); marea. op SES Ailigandi, from a of Ailigandi River to and, Lewis et al. 204 (MO, NY, UC, US). vERAGUAS: Isla i Coiba, Dwyer HH (MO x US); 1-2 mi. above Santa Fe, Gentry 3049 (F, MO, NY). Cordia lucidula I. M. Johnston, J. Arnold Arbor. 21: 352. 1940. TYPE: Panama. Bocas del Toro: Potrero, Changuinola Valley, 20 Oct. 1923, V. C. Dunlap 284 (holotype, US; isotypes, F, NY). Tree to 5(-10) m tall, the twigs glabrous to sparsely strigillose. Leaves persistent; pet- ioles (4-)7-12(-16) mm long, glabrous to virus blades ovate to narrowly ovate, (9-) 11-22(-28) cm long, 5-9(-12) cm wide, the apex acuminate, the base obtuse and slightly decurrent or rarely rounded or acute, the margin entire, the adaxial surface gla- brous or rarely strigillose, the abaxial surface minutely strigillose rarely approaching gla- rous. Inflorescence terminal or rarely inter- nodal or axillary, cymose, 4.5-8.7 cm broad, the peduncle 1.8-3.7(-7.8) cm long, strigil- lose. Flowers sessile, monomorphic; calyx cu- pulate to campanulate, 3-4 mm long, ribs absent, glabrous to strigillose, the 3 lobes + deltate, 0.8-1 mm long; corolla white, tubular with reflexed lobes, 5-7 mm long, 5-merous, the lobes oblong, 2.3-3 mm long, the tube 2.8-4.2 mm long; stamens 5, exserted, the filaments 4.5-7.5 mm long, the upper 2.4- 3 mm free, pubescent at the point of insertion, the anthers ellipsoid to oblong, ca. 1.2 mm long; ovary ovoid, 1-1.3 mm long, glabrous; disc crateriform, ca. 0.7 mm tall; style 1.9— 4.1 mm long, the stigma lobes clavate. Fruits 10-13 mm long, 9.5-12 mm broad, borne in the slightly expanded, cup-shaped calyx, drupaceous, red, the stone broadly inequila- terally ovoid, the surface ruminate and slight- ly ridged, the endocarp bony. Distribution. |. Cordia lucidula occurs in wet forests from sea level to 1,500 m in elevation and ranges from Nicaragua to Pan- ama. [n Panama C. lucidula is known from the provinces of Bocas del Toro, Chiriqui, Darién, and Panamá. Cordia lucidula 1s extremely variable and is the most commonly collected member of a difficult group of closely related species, in- cluding C. lasiocalyx and C. porcata, found in Panama and Costa Rica. Cordia lucidula can be distinguished by its three-lobed calyx, more or less ovate leaves, and bright red fruits without apical prolongation. Leaf shape and texture often vary on a single plant and with time of year and locality. Cordia lucidula differs from C. lasiocalyx by having an acu- minate, rather than caudate leaf apex. Cordia lucidula is most likely to be confused with C. porcata, which differs in having white fruits and anthers more than 1.9 mm long Additional specimens examined. PANAMA. BOCAS DEL TORO: region of Almirante, Cooper 372 (C, CFMR, F, Volume 75, Number 2 1988 Miller 487 Revision of Panamanian Boraginaceae GH, US); aero Valley, Dunlap 356 (CFMR, F, US) Rio Terebe j Puerto Palenque, 350 ft., Kirkbride & Duke. 549 (F, MO, NY, SCZ); cloud fun of Cerro Bonyic, above Quebrada Huron, 500- 1,200 ft., Kirkbride & Duke 596 (F, MO, NY, SCZ); Pero pen to 5 mi. S along Rio Guarumo, Lewis et al. 208 UC); Chiriqui Lagoon, Fish Creek Hills, Wedel 2394 (GH (2), MO), 2425 (GH, MO, US). CHIRIQUÍ: San Bartolo of Puerto Ael. 2 COCLÉ: 7 km N of El Cope, Folsom & Collins 64 70 (NY). Hry Pa Brao along headwater of Rio Tu uqueza, air distance from continental divide, in the vicinity E opper go d mining camp of Tyler Kittredge, Croat 27126 (MO); trail from Pucuro to Cerro Mali, vicinity d mouth of Tapaliza ee: ca. 00 m, ‘Gan & Mori 13546 (MO, SCZ, US); 0-2 mi. E of Tres Bocas along the shortest headwater of the Rio io re Kirkbride & Duke 1184 (MO, NY). PANAMÁ: El Llano-Carti n oe m from Inter- American Highway, 300-400 m, Mori 7709 (MO, U). Cordia megalantha S. F. Blake, Proc. Biol. Soc. Wash. 36: 200. 1923; nom. nov. for Cordia macrantha S. F. Blake, Contr. U.S. Natl. Herb. 24: 19. 1922, non Cho- dat, 1921. TYPE: Guatemala. Izabal: Quebrada, 18 May 1919, S. F. Blake 7498 (holotype, US 989592). Tree to 30(-60) m tall, the twigs glabrous. Leaves deciduous; petioles (8—)11-33(-55) mm long, glabrous; blades elliptic to obovate, (4.6-)6-19(-21) cm long, 2.9-8(-12.6) cm wide, the apex acute to acuminate or rarely obtuse, the base acute or rarely obtuse to rounded, the margin entire, the adaxial and abaxial surfaces glabrous. Inflorescence ter- minal, paniculate, to 30 cm broad, the branches glabrous except for puberulent tips. Flowers on short spurs 2-5 mm long, disty- lous; calyx tubular, (8.5-)9-10(-11) mm long, striate to 10-20 ribbed, glabrous to puberulent, the hairs dark brown, unevenly lobed, tearing upon dehiscence or dehiscing circumscissilly; corolla marcescent, white, funnelform, 28-43(-50) mm long, the 5(-6) lobes deltate to ovate, 11-13 mm long; sta- mens 5(-6), the filaments 14.5-19 mm long, the upper 5-10(-13) mm free, pubescent at insertion and frequently over the entire free portion, the anthers oblong, 2-4 mm long; ovary ovoid to conical, 1.3-2.5(-4) mm long; disc depressed ovoid, 0.5-1 mm tall; style 16-19 mm long, the stigma lobes clavate. Fruits enclosed by the persistent calyx and corolla, ellipsoid to narrowly ellipsoid, 8-12 mm long, 4-6 mm broad, the wall thin, fi- brous. Distribution. Cordia megalantha ranges along the Atlantic coast of Central America in wet forests from sea level to 400 m in elevation. Disjunct populations occur on the Pacific side of Central America on the Osa Peninsula of Costa Rica and the Burica Pen- insula in Chiriqui where the only known col- lection from Panama was made. Often exceeding 30 m in height, Cordia megalantha is the tallest Central American species of Cordia. Although there are no re- ports of C. megalantha being cultivated as a timber tree, as is C. alliodora, it may have potential in wet regions. It apparently does not flower each year (T. Wendt, pers. comm.) and this, combined with its height, may be responsible for the paucity of collections. Cor- dia megalantha is very distinctive among the Panamanian members of the genus in its large size, marcescent corolla, and deltate corolla lobes. Additional specimens examined. PANAMA. CHIRIQUÍ: west of San Bartolo Limite near Costa Rican border, Croat 22175 (MO) Cordia panamensis Riley, Kew Bull. 1927: 135. 1927. TYPE: Panama. Panama: secondary growth at sea level near Old Panama, L. A. M. Riley 143 (holotype, K; isotypes, MO, US) Tree to 10(-15) m tall, the young twigs hirsute. Leaves persistent; petioles (5-)7-12 (7-17) mm long, villous to hirsute or occa- sionally strigillose and with scattered, erect, longer hairs; blades dimorphic, the larger ones ovate to ovate-elliptic, (12-)17-28(-35) cm long, (5-)6.5-13.5(-15) cm wide, the apex 488 Annals of the Missouri Botanical Garden acuminate, the base obtuse to rounded or occasionally somewhat cordate, often some- what asymmetrical, the smaller ones orbicu- lar, (5-) 6.5-9(-10) cm long, (3.8-)5- 8.4(-10) cm wide, the margin entire, the adaxial surface scabrous, the abaxial surface + pubescent, the hairs stiff but not rough to the touch, nearly erect to spreading. Inflo- rescence terminal or in the axils of the branches, cymose, (5-)7-13.5(-16) c broad, the lower portion of the branches vil lous to hirsute, the tips of the branches dense- ly strigillose. Flowers sessile, unisexual by abortion, the plants dioecious, the male flow- ers with reduced styles, the female flowers with small, nonfunctional anthers. Female flowers with a tubular calyx 2-3.6 mm long, ribs absent, strigillose, the 3-5 lobes unevenly deltate to triangular, 0.4-0.6(-1) mm long; corolla white, tubular with reflexed lobes, 3.8- 4.5(-5.5) mm long, (4-) 5(-6)-merous, the lobes oblong-ovoid, 1-1.8 mm long, the tube (2-)2.5-3(-3.7) mm long; stamens (4—)5(—6), the filaments 3-3.6(-4.3) mm long, the upper 0.4- mm free, glabrous, the anthers el- lipsoid, shriveled and containing only aborted pollen, ca. 0.3 mm long; ovary ellipsoid, 1- 1.3(-1.8) mm long, glabrous; disc scarcely distinct from the base of the ovary; style 2- 3.4(-4) mm long, the stigma lobes discoid. Male flowers with a campanulate calyx (2.5-)3-4 mm long, ribs absent, strigillose, the 3-5 lobes uneven, deltate to triangular, 0.7-1.1 mm long; corolla white, tubular-cam- panulate with reflexed lobes, 4.5—6(—6.5) mm long, 5(-6)-merous, the lobes oblong-ovoid, (1.5-)2-2.3(-2.7) mm long, the tube (1.5-)2.5-3.9 mm long; stamens 5(-6), the filaments 4.8-7.2 mm long, the upper 2- 3.5(-4.3) mm free, villous at the point of insertion, the anthers ellipsoid, 0.8-1. long; ovary ovoid, 0.6-1.2 mm long, gla- brous; disc crateriform, 0.5-0.7 mm long; style abortive, 1-2(-3.3) mm long, the stigma lobes filiform. Fruits seated in the cupulate calyx, white, drupaceous, glabrous, the stone inequilaterally ovoid, 6-7.5 mm long, (4.3-) 5.5-6 mm broad, the endocarp bony. Distribution. Cordia panamensis oc- curs in dry to wet forests and ranges from southern Mexico to northern South America from sea level to 1,000 m in elevation. In Panama it is known from all regions except Bocas del Toro. The Cordia panamensis complex is one of the most taxonomically difficult groups in the genus. Its Central American members, which include C. anisophylla and C. cymosa, are similar in appearance and are sometimes sympatric. The group reaches its greatest di- versity in Panama and northern South Amer- ica. Cordia panamensis differs from C. cy- mosa by being smaller in all aspects and by having only simple (vs. echinate) hairs. It differs from C. anisophylla in having a short- er corolla and fruits that are nearly as broad as long. Cordia panamensis is extremely variable in Panama with individual popula- tions differing considerably in leaf size, shape, and indument. North of Costa Rica, it is more constant morphologically. Cordia panamen- sis is most closely related to Cordia hebecla- da 1. M. Johnston of Colombia, Ecuador, and Peru, which differs only in its evenly velutin- ous twigs and lower leaf surfaces. Additional specimens examined. PANAMA. CANAL AREA: Juan Mina, Bartlett & Lasser 16523 (DUKE, MO); beside railroad tracks along road to Tropic Test Center Miraflores Annex, Correa 261 (DUKE, MO); Guil- lard Highway near Paraiso, Croat 10141 (MO, SCZ), Barro d Island, Croat 10250 (MO, UC), 10390 DUKE, F, MO, NY, SCZ, UC); Ancon Hill near gates to Ts Heights Military Preserve rear gate, Croat 11950 (F, MO, SCZ); Barro Colorado Island, Croat 11967 DUKE, F, MO, NY, SCZ); road S-11, NW of Escobal, Croat 12460 ) (MEXU, MO (2)); Road K-2G, Croat 15142 (MO); and Fort Sherman, Croat 15400 (MO); Pipeline Road 2 mi. from Gamboa gate, Croat 16612 (MO); vicinity of Madden Lake along Boy Scout Road, less than 100 m, Croat 38322 (MO); near Pedro Miguel railway station, D'Arcy & D’Arcy 6021 (CHAPA, MO (2), RSA, US (2); near beach at Ft. Kobbe, towards Vera Cruz, Duke & Mussell 6559 (F, MO); Farfan Beach, Dwyer 6766 (MO); Barro Colorado Island, Foster 972 (DUKE, F), 1916 (DUKE, F, MO); Curundu rel nq Survival School, Gentry 1464 (MO); , Gervais 133 (F, US); without definite locality, Johansen 3 (US); Fort jar Lorenzo, area Limón Bay, Gatún Locks and Gatún Lake, Johnston (A (2)); Barro Colorado Island, Kenoyer 632 (MO); Ancon Hill, 100-200 m, Killip 12113 (US); 1 mi. W of the ~ ~ e e Volume 75, Number 2 1988 Miller 489 Revision of Panamanian Boraginaceae junction of the Cocosolo Hospital and the lookout to Gatün Lake, Lazor & Blum 5377 (MO); Fort Kobbe Military Reservation, Luteyn 1087 (DUKE, F, MICH, MO); Bal- boa, Maxon 6926 (US); Pipeline Road, 10-15 mi. from Gamboa, 100 m, Miller 1028b (MO); Summit Gardens, 0-100 m, Mori & Kallunki 1777 (MO); along road to golf course along Chagres River, Gamboa, 30 m, Nee 7378 (ENCB, L, MEXU, MO, NY, RSA, US); along road to Radar Station on Semaphor Hill, 1 km N of Summit Garden, 100-150 m, Nee 7493 (DUKE, MO, POM, US); Gamboa, Pittier 3708 (NY, US); Egr of Rio Cocoli, Road K9, Stern et al. e bon MO, UC, US); observatory at Miraflores Locks, m, Suet 608 (MO); rots Air Force Base, Tyson 1114 (MO, SCZ) Fort Clayton, Farfan Beach area, Tyson 1823 (MO, SCZ), Curundu near Survival School, Tyson 4174 (MO, SCZ); Miraflores on road to water iig Tyson 6596 (DUKE, ie along Pipeline Road, 3 NW of Gamboa, 100- ES e ster 16 6768 (M 0; ares Wheeler s.n. i 4 (F, MO); near MO, US). coa f: y Portobelo, 5 m, Holderidge 6440B (DUKE, MO); Tres Brazos sawmill, Icacal, which is in between Salud y Boca de Río Indio, Howell 48 (MO); Salud, Lao & Holderidge 222 (MO); Nuevo Chagres, beach and adjacent roadside, Lewis et al. 1858 (ENCB, MO); along the beach between Fato and Playa de Damas, Pittier 3932 (US), 3982 (NY). DARIEN: about 10 mi. S of El Real on Rio Pirre (House no. 22), Duke 5482 (MO, NY); Rio Pirre, Duke 8248 (MO); Río Pucro, below village of Pucro, Duke 13125 (F, MO, NY); 4.5 km S of El Real, along dry stream bed of Río Uruseco, Mori & Kallunki 5372 (AAU, MO, NY); 3 mi. E of Santa Fe, Tyson et al. 4676 (DUKE, MO, SCZ). HERRERA: entre Tres Puntas y Chepo (Las Minas), Carrasquilla & Lao 340 (MO); Llano de las Minas, 350 m, Lao 47 (MO). Los santos: Las Tablas, id 1159 MO). PANAMÁ: S south of El Valle de Antón, 700 m Allen 2481 (MO, US); Trapiche, Perlas Islands, Allen 2613 (MO); San ‘tou Island, Anderson s.n. (NY); out C-15 road, just outside the zone, B Dwyer 2677 (MO, SCZ); roadside on way to Cerro Campana, 1⁄4 mi r from highway, Croat 12030 (MEXU, MO, NY); Cerro Campana, along road above FSU cabin, Croat 14207 (MO (2), SCZy between Cerro Azul and Cerro Jefe, Dress- ler 3865 (DUKE, F, MO); El Llano-Carti Road, 12-14 km N of El Llano, Dressler 4359 (DUKE, ENCB, F, MO); vicinity of El Llano, Duke 5813 (MO); Rio San Tomas, Duke & Mussell 6650 (AAU, MO); grassland on Cerro Campana, 2,400-2,700 ft., Duke 8667 (MO, US); Isla de Pedro Gonzales, Dwyer 1705 (MO); Tocumen, thicket near airport, Dwyer 4044 (US); San Jose Island, Erlanson 226 Sei US), 276 (G (2), NY, US), 446 (US); Cer na, near tower, Folsom et al. 2312 (MO); eae to Ft. Clayton, 50 m, Haines 564 (DUKE (2), MO); San José Island, near Punta del Cabo, Harlow 17 (GH, US); in San José Islan (GH); Bella Vista; sea level, Killip 12014 (GH, NY, US); low woods E of Bella Vista, a suburb of Panama City, Maxon & Valentino 6962 (US); S of Farfan beach along shore and adjoining road, sea level, Miller & Hamilton 730 (MO); along the road to Cerro Campana, 600 m, beso et al. 738 (MO); along the road to Cerro Campana, mi. from the Pan-American Highway, Miller & Mil- e 959, 961 (MO); along the road to Cerro Campana, ca. 5 mi. from the Pan-American Highway, Miller $ Miller 1000 (MO); Morro Island, just off N shore of Taboga Island, Mori et al. 4074 (AAU, MO, NY, WIS); Bella Vista, Standley 25347 (MO, US) near the big swamp E of Río Tocumen, Standley 26647 (US); near Matias Hernández, Standley 28877 (US); Punta Paitilla, Standley 30816 (US); between Matias Hernández and Juan Diaz, Standley 32039 (US); Archipielago de Las Perlas, San José Island, coast beside La Bodega, Stimson 5323 (DUKE, SCZ (2)); Taboga Island near village, Per- las Islands, Tyson 5593 (DUKE, SCZ); of Pacora, 25 m, Woodson et al. 757 (F, MO, NY); Isla gandi, de 682 7 (MO) vi VERAGUAS: Isla de Coiba, [he F, MO), 2343 (BR, MO (2), SCZ); Cerro Tute, Pun ft., al Sd 4293 (US), 4335 (MO); Santiago, 4 mi. from Transisthmian Highway toward Atalaya, & Kirkbride 7412 (MO (2), NY, UC); southern shore of Ensenada Santa Cruz, northern tip of Coiba Island, Foster F). 1626 (DUKE (2), Cordia porcata Nowicke, Phytologia 18: 397. 1969. TYPE: Panama. Colón: Santa Rita Ridge lumber road, 3 Oct. 1968, M. Correa A. & R. Dressler 1076 (ho- lotype, MO 2062961; isotype, MO). Shrub or small tree 2-6 m tall, the twigs glabrous. Leaves persistent; petioles 5-11 mm long, broadly canaliculate adaxially, glabrous; blades narrowly elliptic to lance-elliptic, (5-)8.3-21(-24) cm long, 3-7.5(-10) cm wide, the apex acuminate, the base acute or rarely obtuse, slightly decurrent, the margin entire, sometimes slightly revolute, the adax- ial surface glabrous, often drying with a sil- very sheen, the abaxial surface glabrous to sparsely and minutely strigillose. Inflores- cence terminal or internodal, cymose, 2.5- .5(-10) em broad, usually with 50-100 flowers, the peduncle 1.8-4.4 cm long, the branches canescent to strigillose, the hairs rown. Flowers monomorphic, sessile; calyx campanulate, 5.2-6.5 mm long, 2.6-5 mm 490 Annals of the Missouri Botanical Garden wide at the mouth, ribs absent, nearly gla- brous to minutely strigillose, the (2-)3(-4) lobes + deltate, 1.3-1.8 mm long; corolla white, tubular with reflexed lobes, 6.5-10 mm long, (4-)5-merous, the lobes oblong, 3.1-4.5 mm long, 1.5-2 mm wide, the tube 3.7-6.4 mm long; stamens (4—)5, the fila- ments (4.5—)7—8.4 mm long, the upper (1.3-) 2.5-4 mm free, the lower free portion pu- bescent, the anthers narrowly oblong, 1.9— 2.3 mm long; ovary ovoid to broadly ovoid, 0.8-1 mm broad, glabrous; style (4-)5.2- 5.8 mm long, the stylar branches 2-2.5 mm long, the stigma lobes clavate. Fruits borne in the slightly expanded, saucer-shaped calyx, white at maturity, drupaceous, glabrous, the stone inequilaterally ovoid and sharply apic- ulate at the apex, 8.7-11 mm long, 4.8-6.5 mm broad, essentially smooth, the endocarp ony. Distribution. Cordia porcata occurs in wet forests and ranges from southern Nica- ragua to Panama from sea level to 1,500 m in elevation. In Panama it is known from the provinces of Coclé, Colón, Los Santos, and Panamá. Cordia porcata differs from C. lucidula, the species with which it is most likely to be confused, in having lance-elliptic leaves, an- thers more than 1.9 mm long, and smooth, rostrate fruits. Although C. porcata is quite common in some areas of Panama, it is known from only a small number of collections in northern Costa Rica and a single collection from Nicaragua. Addit ional specimens examined. PANAMA. COCLE: Valle de Anton, Croat oS above El Valle, road which ends s , 81 0 m , Croat 25310 (MO. NY); ed Pilon, El Valle, 3, 000 ft., Duke & Panahin 15015 (MO); ld of La Mesa, N of El Valle, 1,000 3 (AAU, MO, NY). CoLÓN: Cerro Santa mi. from the Transisthmian Highway, 800- 900 ft., Antonio 1805 (MO); Santa Rita Ridge Road, ca. 1 hour walk from end of road, Antonio 4490 (MO); East Santa Rita Ridge, Correa & Dressler 635 (MO, SCZ); Santa Rita Ridge, Correa & Dressler 1076 (MO), Croat 13837 (MO), D'Arcy & D'Arcy 6169 (C, F, MO); Santa Rita lumber road, ca. m E of Colon, Dressler & Lewis 3702 (F, MO, US); a Rita Ridge, Dwyer 8581 (F, MO); Santa Rita Ridge, highway to 8 mi. east, 800 ft., Dwyer et al. 9027 (MO (2), NY); Dee Rita Ridge, Dwyer & Gentry 9366 (MO, NY), 9373 (MO); near Agua Clara rainfall station, Santa Rita Rideo, Foster 1730 (DUKE); Santa Rita Ridge, 2-3 mi. from Transisthmian Highway, Gentry 1865 (MO); Santa Rita Ridge, E ridge, Gentry & Dwyer 4816 (DUKE, F, MO, RSA); Santa Rita Ridge east of Transisthmian Highway, 300-500 m, Gentry 6546 (F, MO); Santa Rita Ridge, 400-500 m, Knapp et al. 1680 (MO); end of Santa Rita Ridge Road, 21 km from Transisthmian Highway, 400-500 m, Knapp & Schmalzel 5243 (MO); Santa Rita Ridge Road, 21- 26 km from Transisthmian Highway, 500-550 m, E 5867 (MO); Santa de ed 2 mi. ay Transisthm al. 7 (F, MO (2)); Santa Rita LE oí arca Highway, Lewis (F, MO, NY); Santa (Panamanian Highway R 20 D), 17-35 km from Boyd- Roosevelt Highway, 400-800 m, Mori & Crosby 6308 (MO, USy; Santa Rita Ridge Road, 7.8 km from the Boyd- Roosevelt Highway, ca. 25 km W of Colón, Mori & Dressler 7906 (AAU, MO, NY); 2.7 mi. by gravel road, NE of carretera Transisthmica, on the Santa Rita Ridge, Nee & Mori 3676 (MO, WIS (2)); Santa Rita Ridge, end of road from Transisthmian Highway, ca. 10 mi. from highway, Porter et al. 4762 (MO); Santa Rita, Suere & Dressler 4818 (MO); Santa Rita Ridge road, 20 km from Transisthmian Highway, 100-1,200 ft., Sytsma 1094 (MO); Santa Rita Ridge, 20 km from Transisthmian High- way, Sytsma 1108 (MO); Santa Rita Ridge Road, 20- 22 km from Transismithian Highway, 1,000-1,200 ft., Sytsma 1311, 1312 (MO); Santa Rita Ridge, Sytsma 1548 (MO); upper Río Piedras lm along trail from end of Santa Rita Ridge Road, ca. 11 km SW of Cerro Bruja, 600-700 m, Sytsma et al. i (MO). Los SANTOS: Cerro Pilón, 2,700 ft., Dwyer & Lallathin 8586 (MO). PANAMÁ: El Llano-Carti Road, 13.7 km N of Pan- American Highway, Folsom 3589 (MO) Cordia protracta I. M. Johnston, J. Arnold . : 349. 1948. TYPE: Panama. San Blas: Permé, G. P. Cooper 244 (ho- lotype, GH; isotypes, F, MICH, NY, US (2)) Tall shrub, the twigs sparsely strigillose. Leaves persistent; petioles 4-9 mm long, can- aliculate adaxially, sparsely strigillose; blades anisophyllous, falcate, the larger oblong-ovate, 15.6-24.6 cm long, 5.4-10.3 cm wide, the apex long acuminate, the acumen to 3 cm long, the base asymmetrical, rounded to ob- tuse, the margin entire to slightly undulate, the adaxial surface glabrous, the abaxial sur- face glabrous but sparsely strigillose along the veins. Inflorescences terminal or borne in the branch axils, few per stem, cymose, to 6.5 cm long, 4.5 cm broad, with 20 or more flowers, the branches strigillose. Flowers ses- Volume 75, Number 2 1988 Miller 491 Revision of Panamanian Boraginaceae sile or nearly so, bisexual, monomorphic; ca- lyx tubular, 7-7.5 mm long, ca. 4 mm wide at mouth, ribs absent, sparsely strigillose, the 5 lobes deltate, ca. 1 mm long; corolla white, tubular with reflexed or spreading lobes, ca. 12 mm long, 5-merous, the lobes oblong to ovate, 4-4.5 mm long, ca. 2.5 mm wide, the tube 7-7.5 mm long; stamens 5, exserted, the filaments ca. 11 mm long, the upper 4— 4.5 mm free, slightly puberulent at insertion, the anthers oblong, ca. 2.2 mm long; ovary narrowly ovoid, ca. 2.0 mm long, ca. 1.1 mm broad, glabrous; style ca. 7.5 mm long, the stylar branches ca. 2.5 mm long, the stigma lobes clavate. Fruits borne in the slightly ac- crescent saucer-shaped calyx, white, drupa- ceous, glabrous, the stone inequilaterally ovoid, 10-11 mm long, 6-7 mm broad, ru- minate, the endocarp bony, 1-locular. Distribution. Cordia protracta is known only from wet forests at low elevations along the coast of San Blas and Chocó provinces in Colombia. Cordia protracta is known from only a few localities in Panama. A distinctive species, it is perhaps most closely related to C. correae but differs by having falcate, shiny leaves, and white fruits. Further, C. correae occurs at high elevations in cloud forests in contrast with the lowland wet forests inhabited by C. protracta. Additional specimens examined. PANAMA. SAN BLAS: vicinity of Puerto Obaldia, Croat 16873 (MO (2)); Mu- latuppu, Rio Ibedi, Duke 8483 (MO); 3-4 hours up Rio Mulatupo by foot, Kirkbride 233, 234 (MO). Cordia sebestena L., Sp. Pl. 190. 1753. TYPE: without locality or collector's name (holotype, LINN (Savage Catalog num- ber 253.2), not seen; microfiche, MO). Cordia speciosa Salisb., pr Stirp. Chap. Allerton 111. . TYPE: not see Small tree or shrub to 8 m tall, the twigs glabrescent. Leaves persistent; petioles (1.0—) 1.5-3.8(-4.5) cm long, pubescent, the hairs simple, appressed; blades ovate, (7-)9- 20(-22) cm long, (4.5-)6-12(-14) cm wide, the apex acute, the base rounded to obtuse, rarely somewhat cordate, often slightly un- even, the margin entire or occasionally slight- ly undulate, the adaxial surface scabrous, the hairs from a basal cystolith, the abaxial sur- face nearly glabrous with hairs sparse and restricted to the veins. Inflorescence subter- minal, cymose, 6.5-12 cm broad, with 12- 45 flowers, the branches strigillose. Flowers on pedicels 4-6 mm long, distylous; calyx tubular-campanulate, 11-24 mm long, ribs absent, glabrous or with an indument of 2 types of hairs, the first type simple and straight, 0.4-0.6 mm long, appressed, white to translucent, the second type simple, curly, 0.2 mm long or shorter, brown, usually 2-lobed but sometimes with up to 5 irregular and uneven lobes; corolla bright reddish orange, funnelform, 30-58 mm long, 5-7-merous, the lobes ovate to very widely ovate, 8-10 mm long; stamens 5-7, the filaments 22-33 mm long, the upper 2-6 mm free, glabrous, the anthers oblong, 2.8-3.8 mm long; ovary conical, 1.5-3 mm long; style 13-35 mm long, the stigma lobes clavate. Fruits com- pletely enclosed in the accrescent calyx and often extending in a thin tip up to 12 mm beyond the fruit, drupaceous, white, the stone ovoid, 2-4 mm long, 1.5-2.3 cm wide, the endocarp bony. Distribution. Cordia sebestena is basi- cally pan-Caribbean in distribution, occurring from southern Florida through the West In- dies, and to the Atlantic coast of southern Mexico, Central America, and northern South America. It grows along coastal strands and is particularly common on the offshore islands of Central America. In Panama it appears to be native only in the Comarca de San Blas. Cordia sebestena is grown ornamentally throughout warm areas of the world for its bright orange-red flowers. It is the only Pan- amanian species of sect. Cordia and is dis- tinctive in its large, funnelform corolla and large, drupaceous fruits completely enclosed by the accrescent calyx. The fruits are edible and are very sweet, although quite mucilag- inous. 492 Annals of the Missouri Botanical Garden Additional specimens examined. PANAMA. PANAMA: SAN BLAS: Isla Sos- katupu, Duke 8963 (MO, US), 1547 6 (MO); Guadia Tupo, Dwyer 6864 (MO (3); 50 mi. W of Ailigandi, on Ailigandi, Hammel & D’Arey 5051 (MO), Playon Chico and vicinity, Pinkanti hillside near bay, Stier 186 (MO), Playón Chico and vicinity, Yantuppu, Stier 192 (MO). d aa L., Mant. Pl. 2: 206. . M. Johnston, J. Arnold Arbor. 2 o 1949. TYPE: based on a col- lection from “India orientali” (cf. John- ston, 1949a) (holotype, LINN (Savage Catalog number 253.2), not seen; mi- crofiche, MO). Varronia ferruginea d Tab. Encyc. 1: 418. n oir., Encyc. 4: 263. 1797; Desv., J. Bot. (Des vaux) 1: 266, t. 9. 1809; MA Hn (Lam. Roemer & Schultes, Syst. Veg . 1819. TYPE: is on plants cultivated in ua rua P-JU, mber 6525a, not seen; microfiche, MO). Cordia | riparia Kunth in Humb., Tus & Kunth, Nov. Gen. Sp. 3: 71, t. 207. 1818; I. M. Johnston, J. i Arbor. 30: 103. DR TYPE: Colombia: ox, Magdalena Valley, no collector named on specimen (holotype, P in herb. Humboldt, not seen; — che, MO). Cordia laxiflora ed in s do & Kunth, Nov. 1 Sp. 3: 72. 1818 . Johnston, J. Arnold Arbor. 30: ve ce TYPE: Colombia: between Mompox and Morales, Magdalena Valley, no col- lector named on d (holotype, P in her umboldt, not seen; microfiche, MO). Cordia sc aur ibas DC., Pr ade 9: 490. 1845. TYPE: ana. Witho t definite locality: 1838, Schom- icis 406 (holotype, G-DC, not seen; microfiche M Cordia thibaudiana DC., Prodr. 9: 489. 1845. TYPE: wit cality n collector (holotype, G-DC, not en; phedi ). Cordia pe ae E Mem. Torrey Bot. Club 6: 83. : Bolivia: near Cochabamba, Bang 1291 not =- Cordia eie. . M. ed J. Arnold Arbor. . 1949. TYPE: Co an José: vicinity of El e nerál, 1,190 m, pod 1936. . Skutch 2828 (holotype, GH; isotypes, K, MICH, MO, NY, US). Shrub 1-3(-6) m tall, the branches arch- ing to sprawling, the twigs puberulent to hir- sute. Leaves deciduous, on slightly recurved spurs 1.5-3(-4) mm long; petioles (1-)3- 11(—15) mm long, puberulent El nos blade ovate to elliptic-ovate, (3-)4-11.5(-14) cm long. (1.4-)2-6.5(- 7.8) cm ic the apex acute to attenuate or slightly acuminate, the base obtuse to rounded, the margin coarsely serrate to minutely denticulate, the adaxial surface scabrous to scabrid, rarely merely papillose, the abaxial surface puberulent to tomentulose, sometimes with most of the hairs restricted to the veins. Inflorescences axillary, spicate, (1.3-)2.5-8.5(-11.5) cm long, 4- 7(-9) mm broad, the peduncle adnate to the petiole at the base, (1-)2-4.5(-6.5) cm long, puberulent to hirsute. Flowers sessile, disty- lous; calyx campanulate, (1.8-)2.3-3.5(-4.2) mm long, ribs absent, strigillose to puberulent and usually with small, globose wax particles, the 5 lobes deltate to shallowly triangular; corolla white, tubular, (3-)3.8-5.4(-6.2) mm long, truncate to frilled at the apex, the tube 1.7-)2-2.7(-3.1) mm long; stamens 5(-6), the filaments (2.7-)3.3-5(-6.2) mm long, the upper (0.5-)1-2(-2.4) mm free, puberulent just beneath the point of insertion, the fre portion glabrous, the anthers ellipsoid, 0.6— 0.8(-1) mm long; ovary ovoid to broadly ovoid, 0.7-1.3 mm long; disc crateriform to annular (0.2-)0.4-0.6(— 1.1) mm tall or enclosing the entire surface of the ovary; style (1.5-)2- 4(-4.8) mm long, the stigma lobes clavate. — Fruits drupaceous, 12 to nearly completely enclosed in the slightly accrescent calyx, red, 3.7-4.1 mm long, (2-)2.7-3.3 mm broad, the stone ovoid to broadly ovoid, the endocarp bony. Distribution. | Cordia spinescens is a very widespread, weedy species that occurs from Central Mexico into South America, although it is apparently absent in the West Indies. It grows from sea level to 2,000 m in elevation in a wide variety of habitats. In Panama it is known from all regions except Los Santos. Cordia spinescens is extremely variable but is easily recognized by its axillary spicate inflorescences with the base of the peduncle adnate to the petiole of the subtending leaf. This is the most commonly collected species of Cordia in Mexico and Central America. Cordia spinescens is often found m open disturbed areas and is very common in moist ditches along roadsides. Unlike the other Volume 75, Number 2 1988 Miller 493 Revision of Panamanian Boraginaceae shrubby species of sect. Varronia that usually have a rather erect form of growth, C. spi- nescens often has long arching branches. The species of sect. Varronia character- ized by spicate inflorescences, three of which occur in Central America, make up the most complex group in the entire genus (discussion under C. curassavica). This assemblage is centered in the Andes, and numerous species have been described from this region, al- though some should certainly be placed in synonymy. As a group, they are phenotypi- cally plastic, and apparently all of the species involved are interfertile. Since most species of sect. Varronia with spicate inflorescences are widespread, natural hybridization is com- mon, which has contributed to the confusion surrounding this group. As defined here, Cor- dia spinescens has its closest relatives in South America in Cordia multispicata Cham., and perhaps in the West Indies in Cordia brownei (Friesen) I. M. Johnston. There is extreme morphological variability between populations of Cordia spinescens. Plants from low to middle elevations have ovate leaves with acute to slightly acuminate apices and elongate spikes in the axils of fully expanded leaves with the peduncle adnate to the petiole. At higher elevations in Panama and Costa Rica, the plants have more atten- uate leaves, and the spikes appear well before the leaves are expanded, often giving the ap- pearance of a paniculate rather than spicate inflorescence. These are not true panicles, however, since as flowering proceeds, the leaves expand and the inflorescence structure thus becomes the same as in plants from lower elevations. This variant from the uplands of anama and Costa Rica was originally de- scribed by Johnston (1949a) as Cordia cos- taricensis, and the type (Skutch 2828) as well as several collections made at a similar stage of development differ from the more typical lowland plants in having shorter, roader spikes with more crowded flowers, more acuminate calyx lobes, more attenuate leaf apices, more prominently serrate leaf margins, and evenly hirsute or velutinous stems. However, intergradation between these upland populations and typical C. spinescens is so extensive that more intermediates are observed throughout their combined range than are individuals exhibiting characters of the extremes. These intermediates do not ex- hibit any reduction in pollen stainability and there is no evidence of habitat differentiation between them and either of the extremes. With a species as variable as Cordia spi- nescens, it is not surprising that there has been considerable nomenclatural confusion. The name Cordia ferruginea (Lam.) Roemer Schultes was widely applied by earlier authors, although Johnston (19492) correctly pointed out that the Linnaean name C. spi- nescens has priority and must be accepted. The types of these two names are clearly conspecific, despite the fact that Linnaeus mistakenly described C. spinescens as East Indian; no species of sect. Varronia are na- tive in the Old World. This confusion probably resulted from the fact that C. spinescens had been cultivated in Europe at least as early as the late eighteenth century, as indicated by Johnston (1949a), suggesting that Linnaeus probably based his description on a cultivated specimen. Lamarck clearly based his descrip- tion of Varronia ferruginea on plants that were cultivated in Paris in 1791. Additional specimens examined. PANAMA. BOCAS DEL TORO: along railroad track near station at Milla 5, Croat 16485 (F, MO); Changuinola Valley, Dunlap 280 (US), 281 (GH); Changuinola Valley, Isla Potrero, aed bein (CFMR, F); upper Rio Changuinola, few miles upst from Changuinola, Dwyer 4385A (MO); trail leading 1o to ); San Pedro, Gordon 4 tinental divide on carretera del Oleoducto ca. 1 km N of pe sei Arena ortuna Hydro Electric dd d 0 m, Knapp 5085 (MO); Changuinola to 5 mi. S at sede of ríos Changuinola and Terebe, 100- 200 ft., Lewis et al. 845 (K, MO, UC, US); Almirante, near edad to Chiriqui, ca. 200 ft., McDaniel 5075 (MO); Chiriqui Lagoon, Water Valley, Wedel 1633 (MO, US), 1672 (MO), 1801 (GH, MO); Chiriqui Lagoon, Old Bank Island, Wedel 2169 (MO, US); Quebra Nigua, Wedel 2741 (MO, US); Chiriqui — Isla Colón, Wedel 2839 (MICH, , NY, NAL AREA: Barro Colorado Island, Aviles 103 (MO); sus 495 (F, GH); Escandente, alrede- dores de la repressa Miraflores, Correa 1181 (MO); Barro Colorado Island, Croat 4400 (F, MO, SCZ); Pipeline Road at Río Agua Salud, Croat 4731 (MO, SCZ (2)); Frijoles, near railroad station, Croat 6265.4 (MO); Barro Colorado Island, Croat 7470 (F, MO, NY, SCZy near Frijoles 494 Annals ie Hi Garden station, Croat 8922 (MO, NY, SCZ), 8922 (MO, SCZ); Barro Colorado Island, Croat 11763 (F (2), MO), 12811 (F, MO); Pipeline Road N of Gamboa ca. 24 km beyond gate, less than 300 ft., Croat 38264 (MO); Barro Col. orado Island, D'Arcy 3983 (MO); Pan-American High- way near La Chorrera, D’Arcy 9429 (MO); road C-21, Duke 5791 (MO (2), NY); Albrook, ed 6714 (MO); Frijoles, Ebinger 307 (F, MEXU, MO, US); near Farfan beach, Gentry 1399 (F, MO, NY, SCZ); Pipeline Road, 2-4 mi. N of Gamboa, ca. 100 m, Gentry 6541 (C, CAS, F, L, MEXU, MO, NY, RSA); Corozal, Greenman & cu 5207 (GH, MO (2); Gatun Station, Hayes i. (G, MO), s.n. (GH (2), MO, US), 608 (MO, NY (2), US): Heriberto pr (NY, US) Barro Colorado Island, Hladik 65 (MO); Pipeline Road near Rio Agua Salud, emsky-Young 1809 (C, F, MO, NY, o Island, Kenoyer 410, 641 (US); Gamboa, McDaniel 5045 (MO) Curundu, 30-40 m, Miller 1046 (MO); RR e 81 (C); near old Fort Lorenzo, mouth of gres, Piper 5938, 5967, 5988 (US); vicinity of fusi Piper 6030 (US); between Corozal and Ancon, 10-30 m, Pittier 2171 (NY, US); grounds of Fort San Lorenzo, Porter et al. 5006 (MO, SCZ, UC); W slope of Ancon Hill, vicinity of Balboa, 20- 75 m, Seibert 403 (MO, NY, US); Barro Colorado Island, Shattuck 42 (F, GH, MO), 480 (F, MO (2), 1073 (F, MO (2)); Balboa, Standley 25570, 25591 (US); Summit, Standley 26954 (US); Balboa, Standley 27008 (US); Hills W of the Canal, near Gatun, Standley 27 183 (US); Gamboa, Standley 28534 (US); along the old Las Cruces Trail, between Fort Clayton and Corozal, Standley 291 14 (US); vicinity of Summit, Standley 29998 (US); vicinity of Fort Sherman, Standley 31060 (US); Obispo, Stand- ley 31785 (US); Miraflores Locks area, Tyson 1134 (MO, SCZ); 1 mi. N of summit on road to FAA radar tower, Tyson et al. 2773 (MO, SCZ); Fort San Lorenzo, Tyson & Blum 3792 (MO); vicinity of Miraflores Lake, White 166 (MO, NY); north side of canal beyond bridge, White 1 (F, MO); vicinity of Miraflores Lake, White 245 (MO, NY); 1 mi. SW of Cocoli in the Rodman Naval Ammu- nition Depot, Wilbur et al. 12914 (F, LL, MEXU, MICH, NY); Barro Colorado Island, Woodworth & Vestal 441 (F, GH), 444 (MO), 519 (F, MO). CHIRIQUÍ: Bajo Mono- obalo Trail, western slopes o 1,000 ft., Allen 4781 (MO); Ce M 1417 (MO); Boquete- Palo Alto- Arco Iris, Beliz MO); Finca Collins, vicinity of Boquete, Blum & nor 25254 (MO, SCZ (2)) Dist. San Felix, corregi- miento of Hato Culantro, Hamlet of Cerro Otoe, 3,000 ft., Bort 2 (MO); Burica as San Bartolo Limite, 9 km W of Puerto Armuelles, 5 Chorcha, Castillo 3 (F, M Landian, NE del ud de Fortuna (Hornito), 1,100 m, Correa et al. 23: MO); Nor-oeste del campa- mento Fortuna, 1.000-1,200 m, Correa et al. 2590 (MO), 2619 (F, MO); NE del campamento de Fortuna (Hornito), 1,000-1,200 m, Correa et al. 2882 (MO); trail north of Cerro Punta, Croat 10499 (MO); between Bambito and Cerro Punta, Croat 10672 (MO); along the j uadelupe, Croat 16045 8 (F, L, LL, MO, NY, Peninsula, a Bari. along ridge above Brazo Seco near Costa Rican border, 100-200 m, Croat 22566 seta Methodist youth camp between Nueva Swissa and ro Punta, Croat 26256 (MO, NY, US); in and along Cerro Horqueta, 1,650 m, Croat 26999 M i continental divide on Cerro Colorado, on upper mining road 20-28 mi. from San Felix, 1,200- 1,500 m, Croat 33379 (MO); Cerro Colorado, along road to copper mine, 34.1 km beyond bridge over Rio San ix, 13.1 km beyond turnoff to Felix near town of San Feli Duke et al. 13628 (MO, SCZ); NW Horqueta, 5,000-5,800 ft., Dwyer et al. 537 (MO, UC, US); Tole vicinity of Santa Ana Well, ca. 1,000 ft., Dwyer & Kirkbride 7453 (MO, UC); Boquete, Cerro Horqueta, 5,000-6,000 ft., Dwyer € Hayden 7692 ( Boquete, 7,000 ft., i Volcán to Rio Sarana. road that turns eastward 7.2 km Hagen & Hagen 207: end of road along Rio Palo Alto, Hammel 5753 (M Palo Alto, 4.5 mi. NE of Boquete, 6,000 ft., Eto! 7478 (MO); NW of Boquete, 1,350-1,680 m, Huft 1812 (MO); Burica Peninsula, forest along ARF and ad- jacent n ee rada Merida, 4 mi. S of Puerto Armuelles, 0 0 m, Liesner 397 (F, MO); upland in 5.2 mi. NW 3 d Hato del Volcán on the road to Cos Rica, 5,500 ft., Luteyn 832 (MO); Guadelupe, 1.5 m N of Cerro Punta, Mori & Kallunki 5718 (MO); Dos Lagunas 4 s W of El Hato del ud 1,300 m, Mori -1,136 m, Pittier .2 km abo Boquete, 5,500 ft., Proctor 31842 (LL); Boquete, Palo Alto, just E x Boquete, 5,000 ft., Stern et al. 1088 (GH, MO, US); SE slopes of Cerro Pate Macho, trail from Rio Palo Alto, 4 km NE of B o ft., Tyson 566 1 (MO, SCZ); Dist. Boquete, above Jarimillo Arriba, along N slopes of Cerro Palo Alto, Webster 16687 (MO); wooded slopes and thicketed Mies along the trail between Cerro Punta and the Quebra ajo Grande, 2,000-2,100 m, Wilbur et al. 11908 (DS, F, LL, MICH, ur et al. 130 N o Pena Blanca, 1,750-2,000 m Woodson & Sc e 314 (MO); vicinity of Bajo Chun. ,900 m, Woodson & Schery 613 (MO); vicinity of Casita Alta, Volcán de Chiriqui, 1,500-2,000 m, Woodson et al. 912 (MO, NY (2)). cocLé: El Valle de Anton, trails near Finca Tomas Arias, 600 m, Allen 4232 (F, MO); El Valle, back of Club Campestre, Dwyer 10511 (MO (2)); W of Rio Guias, Gentry 5844 (MO); 46 km N of Penonome on road to Coclesito, 100 ft., Hammel 1702 (MO); El Valle de Antón, 1,000-2,000 ft., Lewis et al. 2575 (MO, UC); Boca del Toabre at confluence of Rio Toabre and Rio Coclé del Norte, Lewis et al. 5498 (MO, SCZ, UC); foot of Cerro Pilon, above El Valle de Anton, 2,000 ft., Porter et al. 4364 (MO); foot of Cerro Pilon, above El Valle de Antón, 2,000 ft., Porter et al. 4620 (MO, SCZ), 4656 (MO, UC); El Valle de Antón, narrow valley behind hotel Pan Americana, Wilbur & Luteyn 11714 (F, LL, MICH, MO, NY, RSA, US). COLON: vi- cinity of Portobelo, Croat 335 73 (MO); along Rio Iguanita Volume 75, Number 2 88 Miller 495 Revision of Panamanian Boraginaceae near bridge along Portobelo road, ipi 49776 (MO) near Colon, Lehmann 996 (US); between Standley 30376 (US); along epis a 5-7 mi. S z : i Wilbur Weaver 11175 (F, GH (2), MICH, MO). DARIÉN: on hills above west end of airstrip at Caña near Rio Caria, Croat 38073 (MO); along Rio Pirre, Duke 4974 (MO, NY); wooded ridge just S of El Real, Duke 5051 (MO); Rio Balsa between N. Q. Chusomocatre and Rio Areti, Duke 8706 (MO); Isla Casaya, Duke 10385 (MO); 0.5-1.5 mi. E of Manene, Hartman 12102 (MO); El Real, trail to Rio Pirre, Kennedy 2817 (MO); Manene to mouth of Río Cuasi, Kirkbride & Bristan 1417 (MO, NY). HERRERA: 11 mi. S of Ocú on Las Minas Road, Graham 240 (GH, MICH). PANAMA: vicinity of Pacora, 35 m, Allen 1010 (F, MO, US); San José Island, Anderson s.n. (GH); weedy aren a of Mii Airport, D'Arcy 9643 (MO); Rio onfluence with > Corso, Duke 12009 men, Dwyer MO), 4225 (MO); San José 47 (US), 59, NY), 191 (US); Harlow US); Johnston 120 (GH, US), 592, 907, 980 (GH), H (2)); Perlas blanda, S tip of Isla Del Rey, 0-20 m, Knapp & Mallet 2911 (MO); x 2 (MO M 0) "loc key x Erlanson Standley 26602 (US); vicinity of Juan Franco Race Track, Standley 27721 (US); Rio Tocumen, Standley 29483 zul t ., Tyson 6173 (F); La Caren. ps Mendozas, quebrada cerca del campo de Juegos, Vergara & Torres 81 (MEXU); pea ou Ph ee 34 (MO); thickets and forests near Arr son et al. 1357 (F, MO, NY). SAN BLAS: 5 hill a ol Puerto Obaldia, Croat 16703 (MO, NY); Mulatupu, "e Ibedi, Duke 847 MO); Sasardi, 20 m, Duke 4 (MO); along canal Mula o Obaldia, Gentry 14 479 (MO); mainland op- posite Playón Chico, 0-3 mi. from Caribbean, 0-200 m, Gentry 6400 (MO); caper opposite Ailigandi, from mouth of peas Riv 2.5 mi. inlan 165 (MO, US); along Rio ‘Ailigandi, 0-100 ft., 182 (MO). VERAGUAS: hills W of Sona, 500 m, Allen 1044 (MO, NY, US); Isla de Coiba, near Maria River, across bay from Colonia Penal, Antonio 2327 (MO); S of Santa Fe, Nee 8014 (MEXU, MO, RSA, US); 2 km NW of Atalaya, 100 m, Nee 8200 (MO); La Mesa, Tyson 6070 (MO, SCZ). Cordia tacarcunensis James 5. Miller, sp. nov. TYPE: Panama. Darién: trail from Pucuro to Cerro Mali, vicinity of Ta- paliza River, 100 m, tropical moist for- est, 13 Jan. 1975, Alwyn H. Gentry & Scott Mori 13546 (holotype, MO 2288082). Figure 5. Arbor vel frutex ad 3 m alta, ramunculo glabro. Folia persistentia, petiolis 6-10 mm longis; laminae anguste ovato-ellipticae, 8.1-14.5 cm longae, 4.5-6.5 cm latae, apice acuminatis, base acutis, superficie papillosa. Inflo- rescentiae axillares, parvae dichotome cymosae, 4-6 cm latae. Flores unisexuales, plantis dioeciis; calyx campanu latus, 2.7-3 mm longus, 5-lobatus, strigillosus; pdas + alba, E 4.8 mm longa, 5-lobata, lobo reflexa, oblonga, 2.2 mm longa; stamina 5, filis villosis, antheris ellipticis. Fructus F putamine inaequilateraliter ovoideo, 5 mm lon Small tree or shrub 3 m tall, the twigs glabrous. Leaves persistent; petioles 6-10 mm long, canaliculate adaxially, minutely strigil- lose; blades narrowly elliptic-ovate, 8.1-14.5 cm long, 4.5-6.5 cm wide, the apex acu- minate, the base acute and sometimes slightly decurrent, the margin entire, the adaxial sur- face lacking hairs but densely covered with small scaly papillae, the abaxial surface nearly glabrous, with small scaly papillae and a few widely scattered appressed hairs. Inflores- cences axillary, numerous per stem, dichot- omous cymes, 4—6 cm broad, the axes dense- ly brown strigillose. Flowers unisexual, the plants dioecious. Female flowers with small, nonfunctional anthers, sessile; calyx campan- ulate, 2.7-3 mm long, 3-3.5 mm wide at mouth, ribs absent, dark-brown strigillose, the 5 lobes shallowly triangular, 0.5 mm long; corolla white, tubular with reflexed lobes, 4.8 mm long, 5-merous, the lobes oblong, 2.2 mm long, 1 mm wide, the tube 2 mm long; stamens 5, nonfunctional, the filaments 2.8 mm long, the upper 2 mm free, villous toward the middle of the free portion, the anthers ellipsoid, 0.5 mm long; ovary ovoid, 2.3 mm long, 2 mm broad; disc crateriform, 0.4 mm tall, 1.2 mm broad, glabrous; style 2.5 mm long, the stylar branches 1.9 mm long, the stigma lobes fan-shaped. Male flowers un- known. Fruits seated in the slightly accrescent saucer-shaped calyx, drupaceous, glabrous, the stone inequilaterally broadly ovoid, 5 mm long, 5 mm broad, endocarp bony, 1-seeded. Distribution. | Cordia tacarcunensis is known only from the type collection made 496 Annals of the Missouri Botanical Garden FIGURE 5. Gentry & Mori 13546 (MO), Darién, Panam Cordia tacarcunensis. near the base of Cerro Tacarcuna on the Colombia—Darién border. Cordia tacarcunensis is probably most closely related to C. protracta and C. cor- reae, with which it shares a similar growth — A. NM branch.—B. flower with corolla opened. —C. Calyx. From habit and a three-parted calyx. It is distinc- tive, however, in its small, axillary inflores- cences and its fan-shaped stigma lobes. In addition to the C. panamensis and C. diver- sifolia species groups, C. tacarcunensis is the only member of sect. Myxa in Central Volume 75, Number 2 1988 Miller Revision of Panamanian Boraginaceae 497 America known to be dioecious; the breeding systems of its two presumed closest relatives are not known. Cynoglossum L., Sp. Pl. 134. 1753; Gen. Pl. ed. 5. 65. 1754. TYPE: Cynoglossum officinale L., vide Britton & A. Brown, Ill. Fl. N. U.S. ed. 2, 3: 75. 1913. Perennial (rarely annual or biennial) herbs from a thickened rootstock, usually branched, pubescent or rarely glabrous. Leaves alter- nate, simple, entire, the basal leaves on dis- tinct petioles, the cauline leaves usually ses- sile. Inflorescences racemes or panicles, the branches scorpioid, usually ebracteate. Flow- ers bisexual, usually pedicellate; sepals 5, nearly distinct to the base, accrescent in fruit; corolla blue, purple, or rarely white, salver- form to campanulate, 5-lobed, with 5 appar- ent protuberances in the mouth; stamens 5, the anthers on short filaments or nearly ses- sile, oblong to ellipsoid; ovary 4-lobed, the style gynobasic, the stigma 1, capitate. Fruits of 4 spreading nutlets, attached apically to the gynobase, the scar restricted to the apical half of the ventral surface, the dorsal surface with short glochidiate spines. Cynoglossum contains about 80 species found throughout much of the world although generally absent from lowland tropical forest. Many species are cultivated ornamentals and this, combined with the epizoochorously dis- persed, glochidiate spiny fruits, has allowed many members of the genus to become nat- uralized far from their natural ranges. Cynoglossum amabile Stapf & J. R. Drumm., Kew Bull. 1906: 202. 1906. SYNTYPES: China. Yunnan: Mengtze, W. Hancock 133 (K, not seen); Szemao, 350 m, A. Henry 9365 (MO). Sichuan: Tatsienlu, Soulié 861 (K, not seen); without locality, 2,700—4,050 m, A. E. Pratt 887 (K, not seen); without locality, M. Leichtlin s.n. (K, not seen). Erect biennial or perennial herb from a thick rootstock, to 0.5(-1) m tall, the stems densely strigose. Basal leaves on petioles to 4.5(-11) cm long, the blade narrowly elliptic, 6-11(-17) cm long, (1-)1.5-2.5(-3) cm wide, the apex acute, the base attenuate to cuneate, the margin entire, the adaxial sur- face strigillose to strigose, the abaxial surface pilose to pubescent; cauline leaves sessile, lan- ceolate to narrowly elliptic, 3-6(-8) cm long, 0.7-2 cm wide, the apex acute, the base clasping and usually somewhat lobed, the margin entire, the adaxial surface strigillose to strigose, the abaxial surface pubescent to hirtellous. Inflorescence terminal, a panicle of small cymes, to 16(-30) cm long, the branches cymose, densely strigillose. Flowers bisexual, on pedicels to 5 mm long; sepals 5(- 1), lanceolate to lance-ovate, 2-3(-4) mm long, strigose; corolla blue, rotate, 5(—7)-mer- ous, the lobes widely obovate, 2-3(-4) mm long, the tube 1.3-2.5 mm long, with 5 hood- ed protuberances in the mouth, these alter- nate with the stamens and often puberulent; stamens 5(-7), the anthers ellipsoid, (0.5-) 0.8-1.3 mm long, nearly sessile or on short filaments to 0.5 mm long, inserted just below the mouth of the corolla tube; ovary 4-lobed, the surface smooth, the disc basal, the style gynobasic, 0.8-2 mm long, the stigma cap- itate. Fruits of 4 spreading nutlets, 4-6 mm broad, the nutlets ovate, flat to convex on the dorsal surface, with glochidiate spines to 0.5 mm long. Distribution. Cynoglossum amabile is native to China but has become commonly naturalized in the Neotropics, being found in open areas at elevations above 1,500 m in elevation. In Panama, it is known only from Chiriqui; it is undoubtedly also in adjacent parts of Bocas del Toro. Cynoglossum amabile can be easily dis- tinguished from Hackelia mexicana (Schldl. & Cham.) I. M. Johnston, the other common small blue-flowered species of Boraginaceae in Chiriqui, by the shorter spines on its fruits and the cauline leaves usually clasping at the base, rather than cuneate as in H. mexicana. The only one of the syntypes that I have seen 498 Annals of the Missouri Botanical Garden is the collection made by A. Henry, which is labeled as having pink flowers. As blue flowers are characteristic of the species and pink- flowered forms are unusual, the selection of a lectotype will have to be from one of the remaining specimens. Additional specimens examined. PANAMA. CHIRIQUÍ: Dist. of Boquete, E of Cerro Punta, area called Bajo Chorro, 2,600 m, Antonio 1032 (MO); Monte Azul, 1. 4 2723 (MO); Cerro Punta-David, 1,000-2,500 m, Beliz 218 (MO); Volcán Barú, 3,474 m, Beliz 355 (MO); Finca Collins, vicinity of Boquete, Blum & Dwyer 2580 (MO); 5.4 km del Hato de Volcán en el camino a Las Lagunas, Correa & Lazor 1472 (MO); 2 mi. N of El Hato del Volcán, Croat 10466 (MO Bambito and Cerro Punta, Croat 1 05 92 (MO); roadsides between Cerro Punta and Bajo Grande, Croat & Porter 16003 (MO); along the Rio Chiriqui Viejo just above Guadelupe, Croat & Porter 16053 (MO); E of Boquete along forested slopes and pastures on Cerro Azul near Quebrada Jara- millo, 1,500-1,620 m, Croat 26779 (MO); 10 mi. above Boquete on road to Volcán Barü, 2,600 m, Croat 34828, 34829 (MO); across river from town of Cerro Punta, D'Arcy & D'Arcy 6528 (MO); Alto Respinga, 2,750 m D'Arcy 12158 (MO); E slope of Volcán de Chiriqui (Barú) WNW of Boquete, 2,200-2,3 avidse 10173 (MO); ji d 1,500 m, Duke la a 13608 (MO); N f Boquete, ea "ei ai 5,000- 5,800 ft., is et el 443 (GH, MO); Boquete, Finca Collins, 5, 000 ft., Dwyer & Hayden 7650 (MO); Cerro Horqueta, 4,500-5,000 ft., veria & Tallallum 8748 (MO); above Cerro Punta, 6,500 ft., Folsom et al. 2039 (MO); along A trail, Ge Respinga, E of town of Cerro Punta, 2 2,500 m, Gentry 5931 (GH, MO); above ein oa nta to Boquete, 5500 m & Stockwell 3345, 3420 (M j O (2)); vicinity of Las of Rio Chiriqui Viejo W of Cerro Punta, 2,200 m, Liesner 290 (MO); N end of town of Cerro lunki 5626 (MO); Bajo Grande, ca. own of Cerro Punta, 2,200 m, /Vee 9952 o o of Cerro Punta, 6,800 ft., Ridgway & Solis 2396 (MO); Volcán Barú, E slope along road to Boquete, 8 km W of e ete 2,200 m, Stein 1282 (MO); vicinity of at , Finca Collins, 5,500 ft., Stern et al. 1097 H. MO. vicinity of Boquete, Finca Collins, “El Velo,” 6, 150 ft., Stern al. 1962 (MO); 3.7 km E of bridge NE of so Punta on road through Bajo Grande, 2,250-2,400 m, Stevens 18150 (MO); 3.7 km along road through Bajo Grande from bridge NE of Cerro Punta, 2,250-2,400 m, Sytsma & Stevens 2155 (MO); Bambito, 1 mi. SW of Cerro - Tyson 5623 (MO); above Cerro Punta 9 (MO); vicinity of Bajo Mono and Quebrada Chiquero, 1,500 m, Woodson & Schery 523 (GH (2), MO). Ehretia P. Browne, Civ. Nat. Hist. Jamaica 168. 1756. TvPE: Ehretia tinifolia L., Syst. Nat., ed. 10. 936. 1759. Trees or shrubs, pubescent or glabrous. Leaves alternate, petiolate, entire or serrate. Inflorescences terminal, cymose to panicu- late. Flowers bisexual; sepals 5, imbricate or open in bud; corolla white, tubular with 5 spreading lobes; stamens 5, usually exserted, the lower portion of the filaments adnate to the corolla tube, the anthers oblong to ellip- soid; ovary ovoid, 2- or 4-locular, the style terminal, bifid, the stigmas 2, clavate or cap- itate. Fruits drupaceous, ovoid to nearly spherical, the stone separating into 2, 2-seed- d or 4, 1-seeded pyrenes. The pantropical genus Ehretia comprises about 50 species with most occurring in Af- rica and tropical Asia. Only three species are known from the New World, one of which is found in Panama. Ehretia latifolia DC., Prodr. 9: 503. 1845. TYPE: Herb. Amat. (holotype, G-DC, not seen; microfiche, Ehretia mexicana S. Watson, Proc. Amer. Acad; Arts 26: 144 , GH, MO, NY (2), UC, US (2)). Ehretia luxiana J. D. Smith, Bot. Gaz. Sie cbe 18: 5. 1893 (corrected ddr TYPE: Gua Quiche: ce Miguel Uspantan, 6,100 ft. ps ig nd & Lux 3065 (holotype, F 575900; isotypes, Ehre cola Robinson, Proc. Amer. Acad. Arts 29: 31 e Mexico. Jalisco: valley, Zapotlán, 19 May 18 93, C. G. Pringle 4382 (holotype, F 106011; Ne A, BM, GH, MO, NY, UC, US ( Ehretia viscosa Fern. in Sarg., Trees € Shrubs 1: 25, 1902. TYPE: Mexico: Morelos, near Cuer- navaca, 29 May 1899, C. G. Pringle 7777 (ho- lotype, F 120287; isotypes, BH, GH, MEXU (2), UC). , NY, Ehretia dehuacana Greenman, Publ. Field Columbian s., Bot. Ser. : 339. 1912. TYPE: Mexico. Pue- bla: a: Las Mohoneras, Tehuacán, 2,200 m, C Con- zatti 2220 (ine: F 235156; isotype, GH). Ehretia austin-smithii Bonum Publ. Field Mus. Nat. Hist., Bot. Ser. 18: 9 i Alainela: Zarcero, in pastur m, Mar. 1938, Austin Smith H528 DE F 919653; isotype, MO). Volume 75, Number 2 1988 Miller 499 Revision of Panamanian Boraginaceae Tree to 10 m tall, the twigs glabrous or nearly so. Leaves persistent; petioles 7-18 mm long, glabrous or nearly so; blades ovate, 4-13 mm long, 2-6.6 cm wide, the apex acute to slightly acuminate, the base obtuse to rounded, the margin serrate, the adaxial surface glabrous to sparsely strigillose, the abaxial surface glabrous. Inflorescence ter- minal, paniculate, to 8 cm long and 7 cm broad, the peduncle glabrous to sparsely pu- berulent. Flowers sessile, bisexual; sepals 5, ovate to narrowly triangular, 1.5-2 mm long, ciliate along the margin but otherwise gla- brous; corolla white, 5-merous, the lobes ovate, 1.5-2.8 mm long; stamens 5, the filaments 4.5-5 mm long, the upper 3.4-4.4 mm free, glabrous, the anthers ellipsoid, 1-1.5 mm long; ovary broadly ovoid, 1-1.5 mm long, the style bifid, 1.4-3.3 mm long, the stigma lobes truncate. Fruits drupaceous, white, el- lipsoid to ovoid, 10-15 mm long. Distribution. Ehretia latifolia is known from Mexico south to Chiriqui Province in Panama, where it is found at 1,000-2,500 m in elevation. Ehretia latifolia, as here defined, is a widespread, variable species. The numerous synonyms are based on minor variations in leaf shape and indument, characters that vary considerably on an individual tree as well as between individuals of a single population (Miller, unpubl.). As in most species of Eh- retia, the plants are quite attractive while in flower, but flowering occurs only for a short period. Additional specimens examined. PANAMA. CHIRIQUÍ: a lo largo del camino que va d lava, 6,000 ft., Correa x up 1410 (F to canyon to Bambi to, 5,000 ft., Rio iniu, Viejo Valley, near Bambito, White 220 (F, GH, MO). Hackelia Opiz in Bercht., Oekon.-techn. Fl. Bohm. 2(2): 146. 1838. TYPE: Hackelia deflexa (Wahlenb.) Opiz in Bercht., Oe- kon-techn. Fl. Bohm. 2, pt. 2, 147. 1839; I. M. Johnston, Contr. Gray Herb. 68: 45. 1923. Erect, perennial or biennial herbs, pubes- cent or less commonly glabrous. Leaves al- ternate, entire, the basal leaves usually long- petiolate, the cauline leaves short-petiolate to sessile. Inflorescence a raceme or panicle, the branches scorpioid, ebracteate or with incon- spicuous bracts. Flowers bisexual; sepals 5, free to the base or nearly so, slightly accres- cent in fruit; corolla blue, often with a yellow center, or rarely white to pale yellow, sal- verform, 5-lobed, with 5 well-developed pro- tuberances in the mouth; stamens 5, included in the corolla tube, the anthers elliptic to oblong, on short filaments; ovary 4-lobed, the style gynobasic, the stigma capitate. Fruits of 4 nutlets, the attachment to the pyramidal gynobase medial, the scar conspicuous, the dorsal surface with elongate glochidiate spines, these longer along the margins. ackelia is a genus of about 40 species of the New World, Europe, and Asia; it is clearly centered in western North America. Only a single species occurs in Panama. Hackelia mexicana (Schldl. & Cham.) I. M. Johnston, Contr. Gray Herb. 68: 46. 1923. Cynoglossum mexicanum Schldl. & Cham., Linnaea 5: 114. 1830. Echi- nospermum mexicanum (Schldl. & Cham.) Hemsley, Biol. Cent.-Amer., Bot. 2: 377. 1882. Lappula mexicanum (Schldl. & Cham.) E. Greene, Pittonia 2: 1882. 1891. TYPE: Mexico. Veracruz: in Monte Macuiltepetl, near Jalapa, Schiede 208 (not seen). Lappula costaricensis Brand in Fedde, Repert. Spec. No gni Veg. 18: 310. 1922. Hackelia cos- taricensis (Brand) I. M. Johnston, Contr. Gray Herb. 923. TYPE: Costa Rica: San José, Hoffman 2 (not seen Lappula guatemalensis Ll in Fedde, Repert. Spec. Nov 1922. Hackelia gua- sig Pflanz zenr. 4, 252: 120. & Lux 3043 (not seen). Guatemala: Huehuetenango, Seler & Seler 3144 (not seen). Erect perennial herb from a thick root- stock, to 1.5 m tall, the stems pubescent. Basal leaves on petioles 9-17 cm long, lan- ceolate, 6-15 cm long, 2.5-4.5 cm wide, the 500 Annals of the Missouri Botanical Garden apex acuminate to attenuate, the base cu- neate to decurrent, the margin entire, cauline leaves lanceolate to elliptic, 8-23 cm long, 2-7.5 cm wide, the apex acuminate, the base cuneate to decurrent, the margin entire, the uppermost usually sessile, those below on pet- ioles to 16 cm long, the adaxial surface stri- gose, the abaxial surface strigose to sparsely pilose, densely strigose along the main veins. Inflorescences terminal or from the upper leaf axils, a simple or once-branched raceme, 12- 25 cm long, the rachis strigose to hirsute. Flowers bisexual, on pedicels to 5 mm long, these elongating to 15 mm in fruit; sepals 5, lanceolate, 1.5-2 mm long, strigose; corolla blue, yellow in the mouth, rotate, 5-merous, the lobes very widely ovate or obovate to widely oblong, 1.5-2.5 mm long, the tube 1.4-2 mm long, with 5 protuberances in the mouth; stamens 5, the anthers ellipsoid, 0.5— 0.7 mm long, nearly sessile, inserted in the middle of the corolla tube; ovary 4-lobed, the lobes tuberculate, enclosed at the base by the disc, the style gynobasic, 0.5-0.7 mm long, the stigma capitate. Fruits of 4 nutlets, 2.5- 3.2 mm long, densely glochidiate, the spines 1-4 mm long. Distribution. | Hackelia mexicana oc- curs from Mexico, south through Central America, to Venezuela, Colombia, Ecuador, and Peru, where it can be found in open disturbed areas at 1,200-3,500 m in ele- vation. In Panama, it is known only from upland Chiriqui, but it is certainly expected in adjacent areas of Bocas del Toro. Hackelia mexicana is a common weed of upland Chiriqui, where it is highly visible and distinct with its bright blue flowers. In this region H. mexicana is most easily confused with Cynoglossum amabile, although, as in- dicated under the latter species, H. mexicana differs in having cauline leaves that are not clasping at the base, nutlets with a medial attachment to the gynobase, and much longer glochidiate spines on the nutlets. Additional specimens examined. PANAMA. CHIRIQUÍ: Cerro Punta, area called Bajo Chorro, 2, m, Antonio 1023 (MO); Cerro Punta, 7,000 ft., Blum et al. 2422 (MO); slopes of Las Cumbres near Cerro Punta, Croat 13679 (MO), Las Cumbres, hogback ridge N of ro Punta, Croat & Volcán Barü, 2,900-2,95 slope of Volcán de Chiriqui (Bard), above Boquete, Da- vidse & D'Arcy 10284 (MO); Potrero Muleto, Volcán de Chiriqui, 10,400 ft., Davidson 1018 (GH, MO); just below last climb in Alto Respinga, 2,700 m, D'Arcy 12135 (MO); Alto Respinga and above, 2, 800 m, D'Arcy MO); along Boquete trail, Cerro Aopo 2 000- 2,500 m, E of town of Cerro Punta, Gentry 6014 (MO); path above Cerro Punta to Boquete, 2,500 m, dr iia Stockwell 3421 (MO); quete, m Chorro, Hladik 195 (MO); near Paso de Musa in pasture and disturbed oak forest, ca. 2,300 m, Mori & Kallunki 5737 (MO); around El Potrero camp, 2,800- of bridge NE of Cerro Punta on road through Bajo Grande, 2,250-2,400 m, Stevens 18217 (MO); 3.7 km along road through Ba ridge NE of Cerro Punta, 2,250-2,400 i. , Sytsma & Stevens 2117 (MO); along the trail between Cerro Punta and the Quebrada Bajo Grande, 2,000-2,100 m, Wilbur et al. 11897 (GH); Finca Lerida to Peña Blanca, 1,750-2,000 m, Woodson & Schery 333 (GH (2), MO); vicinity of Casita Alta, Volcán de Chiriquí, 1,500-2,000 m, Woodson et al. 890 (GH (2), MO). Heliotropium L., Sp. Pl. 130. 1753; Gen. Pl. ed. 5, 130. 1754. TYPE: Heliotro- pium europeaum L., Sp. Pl. 130. 1753. Annual or perennial herbs or rarely low shrubs. Leaves alternate, rarely opposite or whorled. Inflorescences bracteate or ebrac- teate, helicoid cymes borne singly or in groups of 2-4, or the flowers borne individually along leafy stems. Flowers bisexual; sepals 5, im- ricate, free or nearly so to the base, often unequal in size, occasionally accrescent; co- rolla salverform, funnelform, or tubular, white, or white with a yellow center, or occasionally blue to purple, 5-lobed; stamens 5, inserted in the throat of the corolla tube, the anthers free or apically connate; ovary 4-locular, often 4-lobed, the style terminal or absent, the stig- ma 1, conical. Fruits dry, breaking into 2 or 4 nutlets at maturity. Heliotropium comprises about 200 species and is essentially cosmopolitan, with the greatest number of species occurring in dry, tropical regions. Despite the relative abun- Volume 75, Number 2 1988 Miller 501 Revision of Panamanian Boraginaceae dance of species in most neotropical countries, only four species are known from Panama. These are all widespread, weedy species found throughout the Neotropics, and two have be- come widespread in the Old World. Heli- otropium arborescens L., a South American species, is often cultivated for its attractive purple flowers and might be found in gardens in Panama, although no collections exist. The pantropical genera Heliotropium and Tournefortia L., and the monotypic Argen- tine genus /xorhea Fenzl make up the subfamily Heliotropioideae. Heliotropium is a morphologically diverse genus, and John- ston (1928) recognized 11 sections in South America, three of which are known from Pan- ama. Heliotropium is a genus of herbs with dry fruits, in contrast with the woody habit and fleshy fruits that characterize Tournefor- tia. KEY TO THE SPECIES OF HELIOTROPIUM IN PANAMA la. Plants glabrous ............. Helio — curassavicum lb. Plants with pubescent stems and lea 2a. Inflorescence terminal or dl spi- cate or a helicoid cyme, with numerous flowers. 3a. Corolla lavender; plants erect; leaves wider than 2 cm .. Heliotropium indicum 3b. Corolla white; plants procumbent; leaves narrower than 2 cm .. ESTARIA Heliotropium procumbens 2b. Inflorescence axillary, 1-flowered Heliotropium lagoense Heliotropium curassavicum L., Sp. Pl. 130. 1753. TYPE: Curaçao: P. Browne s.n. (holotype, LINN (Savage Catalog number 179.11), not seen; microfiche, MO Low herb, often somewhat succulent, gla- brous, often glaucous, the stems procumbent to ascending. Leaves lacking a distinct petiole; blades oblanceolate, 10-35 mm long, l- 5(-10) mm wide, the apex acute to rounded, the base cuneate, the margin entire, glabrous and often glaucous on both surfaces. Inflo- rescence internodal, a once- or twice-branched helicoid cyme, rarely simple, the peduncle 11-20(-32) mm long, the branches (1.8-)3- 6(-8) cm long. Flowers bisexual; sepals 5, lanceolate to ovate or oblong, ca. 1.5 mm long, glabrous; corolla white, 2-2.5 mm long, the 5 lobes 1 mm long, the tube 1 mm long, glabrous; stamens 5, the anthers nearly ses- sile, inserted near the middle of the corolla tube, ellipsoid, 0.5-0.8 mm long; ovary ovoid, the disc well developed, the stigma sessile, broadly conical. Fruits ovoid, 1-2 mm long, glabrous, 4-lobed, separating into 4 nutlets at maturity. Distribution. Heliotropium curassavi- cum is usually found growing along the edges of lakes, streams, or tidal flats from sea level to 600 m in elevation. It occurs from the United States through Central America and the West Indies and South America; it has also apparently become introduced and wide- spread in the Old World (Nowicke & Miller, in press). This species is known in Panama only from the province of Los Santos. Heliotropium curassavicum is one of the most distinctive species of the genus and is the only member of sect. Halmyrophila 1. . Johnston. It is easily recognized by its glabrous, succulent nature. Several varieties of this species have been recognized (John- ston, 1928; Frohlich, 1981); the populations in Panama are all of the typical variety. Additional specimens exam ANAMA. LOS SANTOS: Salinas de Chitré, D ye & Croat 4199 (MO) args Beach, Dwyer 4177 (MO (2)); Monagre Beach, mi. SE of Chitre, [oe et al. 3023 ). Heliotropium indicum L., Sp. Pl. 130. 1753. rvPE: P. Browne s.n. (holotype, LINN (Savage Catalog number 179.2), ). not seen; microfiche, Annual herb to 50 cm tall, the stems pu- bescent to pilose, the hairs simple. Leaves on petioles (7-)10-25(-40) mm long, pubes- cent, often pilose at the base; leaf blade ovate, (2.7-)5-10(-12) cm long, (2-)3-5(-7) cm wide, the apex acute to obtuse, the base ob- tuse to truncate and usually decurrent along the petiole, the margin unevenly serrate to undulate, the adaxial surface with widely scat- tered appressed hairs, the abaxial surface nearly glabrous with only a few hairs scattered 502 Annals of the Missouri Botanical Garden along the veins, to nearly villous. Inflores- cence internodal, an unbranched or very rare- ly dichotomous helicoid cyme, the peduncle (1-)2-3(-6) cm long, pubescent, the fertile portion (6-)9-16(-20) cm long. Flowers bi- sexual; sepals 5, lanceolate, 2-3 mm long, pubescent; corolla purple to occasionally white, salverform, 5-merous, the lobes ovate, 1-1.5 mm long, the tube 3-4 mm long, pubescent outside; stamens 5, the anthers sessile or nearly so, inserted just below the middle of the corolla tube, ellipsoid, 0.6-0.8 mm long; ovary globose, 0.5-1 mm long, the disc well- developed, the style 0.5-1 mm long, the stig- ma capitate. Fruits angular ovoid, with an apical beak, 2-3 mm long, glabrous, the 2 lobes spread apart and ultimately separating into 2 nutlets at maturity. Distribution. Heliotropium indicum is a weed of disturbed habitats from sea level to 1,000 m in elevation nearly throughout the world with the exception of cold regions. In Panama, it is known from all regions except the Comarca de San Blas, but is probably there as well. Heliotropium indicum is a coarse, annual weed and one of the most commonly en- countered species of the genus. Although it is essentially worldwide in distribution, John- ston (1928) suggested that it was probably South American, possibly Brazilian, in origin. Its closest relative is H. elongatum Hoffm. ex Roemer & Schultes, a species of south- eastern South America, and the two make up sect. Tiaridium (Lehm.) Griseb., which is characterized by a weedy, annual habit, sal- verform corollas, and ribbed, glabrous fruits. Heliotropium indicum is easily recognized by its purple (rarely white) corollas an strongly angular fruits with prolonged apices. Additional specimens examined. PANAMA. BOCAS DEL TORO: Bocas del Toro, Carleton 195 (GH); Old Bank Island, Wedel 1986 (GH, MO). CANAL AREA: Pipeline Road to 18 km N of Gamboa, D’Arcy | (MO); flooded pasture sae ate River, about 2 mi. N of Gamboa, Lazor MO); Barro ere Island, Starry 249 (MO). 21 Burica Peninsula, 1 mi. of Puerto Armuelles, 50 m, Croat 22029 (MO); Burica Pu. dist. Guanabano, disturbed areas along Que- brada Quanabano, 0-100 m, Croat 22532 (MO); vicinity of San Bartolome, Peninsula de Burica, 0-50 m & Schery 925 (GH). cocLÉ: El Valle, Aguilar 48 (MO); Rio Coclé, W of Penonome, Folsom 2917 (MO); 12 mi. NE of Penonomé, 1,200 ft., Lewis et al. 1524 (GH, MO); Boca del Toabre at as of Río Toabre and Rio Coclé del Norte, Lewis et al. 5512 (MO). COLÓN: Portobelo, 5-100 m, Pittier 2470 (GH). DARIÉN: El Real, Correa & Lazor 1534 (MO); El Real, Rio Tuira, Stern et val hes (GH, Tuna HERRERA: Roadside between El Potrero and Las Minas, eda 9650 (MO); just S of Ocu, D' pus 4127 (MO (2)); 5 km W of turnoff from highway 105 to Potuga, Hammel E 4 ( ; 3 km from Pesé on road to Ocú, Huft 1734 (MO). Los SAN TOS: along road between bas and Jobero, 50-80 m, Croat 34446 (MO); 5 mi. S of Pocri, D'Arcy & Croat 4209B (MO); Rio Tonosi vicinity of Tonosi, Lewis et al. 1558 (GH, MO). PANAMÁ: Isla Taboga, 0-350 m, Allen 1297 (GH, MO); beneath bridge on Interamerican Highway near end of Tocumen Airport runway, Croat 977 1 (MO); between Chepo and wharf, Dodge 10721 (MO); Isla San Miguel, Duke 10937 (MO); Bayano Guipo forest, disturbed area round lake near Bayano Bridge, Folsom 3551 (MO), Taboga Island, Macbride 2792 (MO); between Chepo and Rio Bayano, Porter et al. 5173 (MO); Taboga Island near village, Perlas Islands, Tyson 5591 (MO); Rio Ta- tare, Woodson & Schery 995 (GH). VERAGUAS: 1.3 km E of the intersection of the Panamerican Highway and road P38 to Atlaya, Folsom 2932 (MO); 2 mi. S of Canazas, Tyson 3725 (MO) Heliotropium lagoense (Warm.) Girke in Engl. & Prantl, Nat. Pflanzenfam. 4 Abt. 3a, 97. 1893. Schleidenia lagoensis Warm., Vidensk. Meddel. k Na- turhist. Forren. Kjobenhavn 1867: 15. 1868. TYPE: Brazil. Minas Gerais: Lagoa Santa, 1863-1866, Warming 21971 (C, holotype, not seen; photo, MO). Procumbent herb, the stems with a few scattered, appressed hairs. Leaves on short petioles to 1.5 mm long; blade narrowly el- liptic to oblanceolate, 3-6 mm long, 1.2 mm wide, the apex acute, the base acute, the margin entire, the adaxial surface glabrous, the abaxial surface glabrous or with a few scattered, appressed hairs. Flowers borne in- dividually in the leaf axils, on pedicels 2-4 (-6) mm long; sepals 5, ovoid, ca. 2 mm long, glabrous; corolla pale blue to white and yellow in the throat, the 5 lobes widely ovate, ca. 1 mm long, the tube ca. 2 mm long, glabrous; stamens 5, the anthers ovoid, ca. 0.4 mm long, sessile, inserted near the middle of the corolla tube; ovary ovoid, ca. 0.5 mm long, Volume 75, Number 2 1988 Miller 503 Revision of Panamanian Boraginaceae the disc scarcely evident, the stigma sessile, capitate. Fruits ovoid, 1.2 mm long, glabrous, 4-lobed, separating into 4 nutlets at maturity. Distribution. Heliotropium lagoense occurs from Mexico through South America and the West Indies, and is found at eleva- tions below 200 m (rarely to above 1,000 m usually in open savannas. In Panama, it is known from a single collection from the prov- ince of Cocle. — Heliotropium sect. Orthostachys R. Br. is the largest section of the genus and cer- tainly the most complex taxonomically. It is found throughout tropical regions of the world and is particularly well represented in the Neotropics, where Johnston (1928) suggested that approximately 50 species occur. Heli- otropium lagoense and H. procumbens are the only members of the section known to occur in Panama. Heliotropium lagoense is a member of subsect. Axillaria and is easily recognized by its diminutive habit of growth and axillary flowers that are not aggregated into the helicoid cymes that characterize all of the other Panamanian species. Additional specimens examined. PANAMA. COCLÉ: mountains beyond Pintada, 400-600 m, Hunter & Allen 525 (MO) d pope procumbens I Gard. . 8, no. 10. 1768; I. M. John- Sion NS Gray Herb. m 52. 1928. TYPE: Colombia. Bolivar: Cartagena, Houston s.n. (holotype, BM, not seen). Heliotropium americanum Miller, Gard. Dict. ed. 8, no. 10. 1768. TYPE: Mexico. Veracruz: Houston s.n. (ho lot type, BM, not seen). Heliotropium inundatum Sw. Prodr. 40. 1788. TYPE: J . Actorum Acad. re: ee Humboldt 57 holotype, P, not seen; microfic Heliotropium simplex Meyen, Reise 1: 43 36. 1834. TYPE: " acna: Arica, Meyen s.n. (holotype, B, not en). bo inundatum var. cubense wes, Le i 540. 1845. TYPE: Cuba. La Habana bie Nd sd de la Sagre 239 a G. DC. n; microfiche, ). Heliotropium bridgesii Rusby, Mem. Torrey Bot. Club 4: 224. 1895. TYPE: Bolivia. Cochabamba: Cocha- amba, M. Bang 950 (holotype, NY; microfiche, MO; isotype, US). Heliotropium inundatum . chacoense R. E. Fries, Ark. Bot. 6(11): 22. 1906. TYPE: “Bolivia. Tarija: along Río Pilcomayo near Ft. Crevaux, Fries 1614 (isotype, US) Herb to 30 cm tall, the stems procumbent to ascending, strigose to pubescent. Leaves on petioles (3-)5-10(-15) mm long, strigose to pubescent; leaf blade elliptic to narrowly elliptic, 11-20(-35) mm long, 6-1 1(-17) m wide, the apex acute to rounded, the base acute to cuneate, the margin entire, the adax- ial surface strigose, the abaxial surface stri- gose to sericeous. Inflorescence internodal or terminal, a once- or twice-branched helicoid cyme, the peduncle (3-)8-20(-26) mm long, strigose to sericeous, the fertile portion 20- 45(- 75) mm long. Flowers bisexual; sepals 5, lanceolate, 1-1.2 mm long, one often ex- ceeding the others in length, strigose; corolla white, the 5 lobes lanceolate to lance-ovate, 0.5-0.6 mm long, the tube 0.9-1.2 mm long, villous in mouth, strigose to strigillose outside; stamens 5, the anthers ellipsoid but acuminate at the apex, 0.2-0. m long, sessile or nearly so, inserted from near the base to just beneath the middle of the corolla tube; ovary globose, 0.2-0.3 mm long, the disc scarcely evident, the stigma sessile, capitate. Fruits globose, ca. 1 mm long, strigillose, faintly 4-lobed, separating into 4 nutlets at maturity. Distribution. Heliotropium procum- bens is widespread from the southern United States south throughout all of the Neotropics, at elevations of 01,500 m in a wide variety of habitats. In Panama, it is known from the provinces of Coclé, Herrera, Los Santos, and A common weed, Heliotropium procum- bens is extremely variable in shape, size, and indument of its leaves. Its wide geographic distribution and morphological variability have spawned considerable taxonomic problems, and numerous segregates have been pro- posed. Despite this, the species has been in- terpreted broadly by most recent authors. Annals of the Missouri Botanical Garden Additional specimens examine PANAMA. COCLÉ: 20 mi. S of Nata, D'Arcy & Croat 4120 (MO); Río Cocle, Tonosi, vicinity of Tonosi, Lewis et al. 1575 (MO). PANAMA: , 50- 80 m, Huft 1775 (MO); region, Mni & Harvey and Rio Bayano, Porter 5170 (M Tyson 5501 (MO); iiem roadsides within 1 mi. of Chepo, Wilbur & Luteyn 11799 (GH, MO). Torn De in Meisner, Pl. Vasc. Gen. 1: 88. 1840. rvPE: Moritzia cil- iata (Cham.) DC. in Meisner, Pl. Vasc. Gen. 2: 188. 1840 Erect perennial herbs. Basal leaves often forming a spreading, open rosette, the cauline leaves alternate and usually considerably smaller than the basal ones. Inflorescence ter- minal, ebracteate, a sparsely branched cyme of spikes or racemes. Flowers bisexual; calyx tubular to narrowly campanulate; corolla tu- bular with spreading lobes, 5-merous, the lobes ovate to deltate, the tube with protuberances or tufts of hairs in the mouth; stamens 5, on short filaments, the anthers oblong; inserted above the middle of the corolla tube; ovary 4-lobed, the style gynobasic, the stigma ob- scurely bilobed. Nutlets solitary by abortion, erect, smooth to muricate but lacking spines. Moritzia is an essentially South American genus of five species, only one of which, M. lindenii, extends into Central America. It is closely related to Thaumatocaryon (John- ston, 1924, 1927), from which it differs in having all of the leaves alternate and in lack- ing protuberances in the mouth of the corolla tube. Moritzia lindenii (A. DC.) Gürke ex Benth. in Engl. & Prantl, Nat. Pflanzenfam. 4(3): 121. 1894. Meratia lindenii A. DC. in DC., Prodr. 10: 104. 1846. TYPE: Ven- ezuela. Distrito Federal: Caracas, Linden 944. (not seen). Erect perennial herb to 50 cm tall, the stems strigose. Basal leaves sessile or on broad petioles to 5 cm long, narrowly elliptic to lanceolate or oblanceolate, 8-16 cm long, 1- 3.5 cm wide, the apex acute, the base atten- uate, the margin entire, the adaxial surface strigose, the abaxial surface strigose, the cau- line leaves sessile, lanceolate to lance-ovate, 2.5-7 cm long, 0.4-1.4 cm wide, the apex attenuate to acuminate, the base acute, the margin entire, strigose on both surfaces. In- florescences terminal, cymose, 3-10 cm long, the branches strigose. Flowers bisexual; calyx cylindrical, the 5 lobes lanceolate, 1.5-2.2 mm long, strigose; corolla blue, tubular with spreading lobes, 5-merous, the lobes widely ovate to depressed ovate, 0.9-1.3 mm long, the tube 2-2.4 mm long, pubescent in the mouth, strigillose outside; stamens 5, the an- thers ellipsoid, 0.6-1 mm long, sessile, in- serted just below the mouth of the corolla tube; ovary 4-lobed, the style gynobasic, 0.9- 1.1 mm long, the stigma capitate. Nutlet ovoid, 2-2.5 mm long, muricate. Distribution. Moritzia lindenii ranges from Venezuela, Colombia, and Ecuador north to Panama and adjacent Costa Rica. It occurs above 3,000 m in elevation. In Panama, it is known only from the province of Bocas del Toro but may be in upland Chiriqui as well. Moritzia lindenii has been collected only once in Panama, although several collections are known from adjacent areas of Costa Rica. It is expected at high elevations in the same general region as Cynoglossum and Hacke- lia, the other two blue-flowered, herbaceous Boraginaceae known from Panama. lt differs from members of these genera by lacking spines on its nutlets, only one of which de- velops to maturity, in contrast with the glo- chidiate spines on the four nutlets of Cyno- glossum and Hackelia. Additional specimens examined. PANAMA. BOCAS DEL TORO: Cerro Fabrega and vicinity near Costa Rican fron- tier, south of summit, 3,150-3,335 m, Weston 10162 (MO). Tournefortia L., Sp. Pl. 140. 1753. TYPE: Tournefortia hirsutissima L., Sp. Pl. 140. 1753; I. M. e Conti: Gray Herb. 92: 66. 19 Small trees, shrubs, or woody vines. Leaves alternate or rarely opposite, petiolate or rarely Volume 75, Number 2 1988 Miller 505 Revision of Panamanian Boraginaceae sessile, entire. Inflorescence terminal or inter- nodal, dense to lax, a sparsely to profusely branched cyme. Flowers bisexual; sepals 5, one often exceeding the others in length, per- sistent; corolla white to green or yellow-green, tubular, with 5 spreading lobes; stamens 5, the anthers usually sessile or nearly so, borne within the corolla tube; ovary ovoid to glo- bose, 4-locular, the style terminal or absent, the stigma conical. Fruit drupaceous, often white at maturity, later drying and separating into 2 or 4 bony nutlets containing l or 2 seeds. Tournefortia includes about 150 species and has representatives in most warm areas of the world, although most species occur in KEY TO THE SPECIES OF TOURNEFORTIA IN PANAMA la. 2a. Leaves densely white puberulent to tomentose 2b. Leaves evenly short-strigillose below; corolla tu = a. DA oppos Stems te puberulent Plants vining; fruits distinctly 4-lobed; the anthers apically connate n Cyphocyema I. M. Johnston). below; ges m long be 3.3-5 m de Plants various in = fruit not deeply 4-lobed; the anthers free he Tournefortia). 3 the Neotropics. Mallatonia (Griseb.) Britton, Argusia Amman, and Messerschmidia L. ex Hebenstreit have been treated as distinct by several authors, but Nowicke & Skvarla (1974) showed that pollen morphology does not support their continued separation. The three species that have been placed in these segregate genera differ from other members of Tournefortia in being strand plants with a pronounced corky exocarp and a similar se- riceous indument. Species of Tournefortia vary in habit. Most Panamanian species are lianas, sprawling shrubs, or sparsely branched erect shrubs, only a few becoming small trees. There often appears to be considerable variation in habit within a single species. e 2-2.3 mm long _ | volubilis culata T. ramonensis Ha Stems densely and unevenly pubescent to hirsute Qu c . Leaves alternate. T. johnstonü T. angustiflora 5a. Corolla tube 10-13 mm E 5b. Corolla tube up to 10 m 6a. Stems shaggy- lae. tn hairs 3-4 mm long; sepals 5.5-7.5 mm lon, 6b. Stems glabrous or with hairs ri than 1.5 mm long; sepals less than 5 mm long. 7a. Corolla lobes 3-4.5 mm 8a. Corolla tube white sii green stripes, 3.5-5 mm "s n 1.7-3.2 mm long; anthers bilobed, pendent in the mo as of the corolla t : 8b. Corolla green to yellow-green anthers lanceolate, sessile and hen below the mouth of the corolla tube T. cuspidata multiflora 3.5-5 mm long; , the tube 7-10 m a inde epals T. urceolata andi c . Corolla lobes 1-3 mm (odd 9a. Sepals up to 2 mm lon š dan tube 5. 5. 6.5 mm long, the lobes 2.3-3 mm long; tertiary veins T. bicolor 10b. Corolla 3.5-4 mm long, the lobes 1.5-2 mm long; tertiary veins evident .. glabra 9b. T longer than 2 m Corolla tube 9-9. 5 mm long; tertiary veins obscure T. tacarcunensis ib, Corolla tube 3-6 mm long; tertiary veins evident. 12a. Sepals 4.3-4.5 mm 12b. Sepals 2.5-4 m 13a. Stems ares ka hirsute; leaves strigose; corolla lobes 1-1.6 m T. ps ong T. brenesii Anu i ong 13b. Stems glabrous or sparsely strigillose; leaves essentially glabrous; corolla lobes 2-2.5 mm long T. Tournefortia angustiflora Ruiz Lopez & Pavón, Fl. Peruv. 2: 25, pl. 151. 1799. TYPE: Peru. Huánuco: Chicoplaya and Pueblo Nuevo, Hipólito Ruíz & José Pavón s.n. (not seen). longispica Scandent shrub to 1 m tall, occasionally a liana or tree to 5 m tall, the twigs glabrous to puberulent. Leaves alternate; petioles 5— 12(-16) mm long, sparsely strigillose to pu- berulent; blade lance-ovate to lanceolate, 6- 506 Annals of the Missouri Botanical Garden 15(-17) cm long, 2-6(-8) cm wide, the apex acuminate to attenuate, the base acute to cuneate, the margin entire, the adaxial sur- face sparsely strigillose to nearly glabrous, the abaxial surface with short, appressed hairs along the veins. Inflorescence terminal or internodal, a sparsely branched cyme, the peduncle (1-)2-4(-8) cm long, strigillose or puberulent to nearly glabrous, the fertile branches recurved, 2-9(-14) cm long. Flow- ers sessile, borne 2-4 mm apart; sepals 5, triangular, 1.2-1.6 mm long, sparsely stri- gillose; corolla white, 5-merous, the lob ovate, 2-2.6 mm long, the tube 10-13 mm long, strigillose outside, puberulent on the in- ner surface of the lobes; stamens 5, the an- thers lanceoloid, 2.5-3 mm long, sessile, in- serted in the lower half of the corolla tube; ovary globose, ca. 1 mm long, the style to 0.8 mm long or lacking, the stigma conical. Fruits white, ovoid, often slightly inequilater- al, 3-5 mm long, glabrous. Distribution. Tournefortia angustiflora is wide-ranging and common in wet forests from Mexico south through northwestern South America to Peru, from sea level to 600 m in elevation. It is known from most prov- inces in Panama and is probably in all of them. Tournefortia angustiflora is distinctive within the genus in having narrow, tubular corollas more than 1 cm long and somewhat asymmetrical fruits. It is vegetatively similar to T. bicolor, but the two are quite different when fertile. Additional specimens examined. PANAMA. BOCAS DEL TORO: Daytonia Farm, region of Almirante, Cooper 166 (GH); 10-15 mi. inland from mouth of Changuinola River, Lewis et al. 876 (GH, MO); Chiriqui Lagoon, Water Valley, Wedel 1600, 1785 (GH, MO), 1802 (MO), 1832 (GH, MO); Chiriqui Lagoon, Old Bank Island, Wedel 1880, 2004 (GH, MO); A Lagoon. Isla Colón, Wedel 2867, 2960 (GH, MO). « AREA: end of Gatin Lake Dam, Blum & Tyson. 2000 (MO); Barro io Island, Croat 7018, 8677, 9100 (MO); Foster 1681 (GH); alluvial bottom near Bohio, 10-20 m, Maxon 4777 (GH); between Gorgona and Mamei, 10-30 m Pittier 2259 (CH); valley A Masambi on the road to Las — bea 20-10 d Pittier 2592 (GH); Fort San Lor , Porter et al. 5009 (MO); Barro Colorado Island, Shatiuck 495 (MO); Woodworth & Vestal 378 (A, MO). cocré: El Valle de Antón, 600 m, Allen 2059 GH, MO); summit of Cerro Pilon, above El Valle de Antón, 2, dd re esc et al. 4501 (MO); 46 km N from Pe oad to Coclesito, 100 ft., Hammel 1695 (MO); "hills NE of El Valle de Antón, 2. 000 ft., Lewis et al. 1799 (MO); Boca del Toabre at tantis of Rio Toabre and Rio Coclé del Norte, "gis et al. 5508 om Portobelo 1465 (MO RRERA: roadside between El Potrero and Las Minas, Croat 9655 (MO); 12 Ocü on as Minas road, Graham 235 (GH); Punta res Tyson 2714 (M 1 Š of Ocú, Tyson et al. 2 0) ( hepo, Hunter & Allen 9 Alahajuela Chagres Valley, 30-100 m, Pittier 2371 (GH). SAN BLAS: mainland opposite Ailigandi, from mouth of Ailigandi River to 2.5 mi. inland, Lewis et al. 167 (MO). Tournefortia bicolor Sw., Prodr. 40. 1788 . TYPE: Jamaica (not seen). Tournefortia laevigata Lam., Encycl. 1: 416. 1791. TYPE: Guadeloupe, Badier s.n. i bip “ d nitida Kunth in Humb., . & Kunth, G 4. 1819. TYPE: ; Colombia. Bo- near Cartagena (not seen). Tournefortia bicolor Sw. var. calycosa J. D. Smith, Bot. (Crawfordsville) 14: 27. 1889. TYPE: Guate- mala. Alta Verapaz: Pansamala, 3,800 ft., H. von Tuerckheim 980 (holotype, US 944708). Woody vine, shrub, or small tree to 3(-7) m tall, the twigs glabrous or sparsely short- strigillose. Leaves alternate; petioles (0.8-)1— 2 cm long, glabrous or very sparsely short- strigillose; leaf blade elliptic or e to nar- rowly elliptic or lance-ovate, (8-)11-14(-19) cm long, (3.5-)5-8 cm wide, the apex acu- minate to acute, the base obtuse to rounded or less commonly acute, the margin entire, the adaxial and abaxial surfaces glabrous or with a few widely scattered, appressed hairs, the tertiary veins obscure. Inflorescence ter- minal, a dense cyme, the peduncle to 3 mm long, the branches glabrous to sparsely strigil- lose, the fertile branches 2-5 cm long. Flow- ers sessile, crowded, usually borne less than mm apart; sepals 5, lanceolate, 1.5-2 m long, sparsely to evenly strigillose; corolla Volume 75, Number 2 1988 Miller Revision of Panamanian Boraginaceae 507 white, sometimes with a greenish tint, 5-mer- ous, the lobes ovate and often apiculate, 2.3- 3 mm long, the tube 5.5-6.5 mm long, strig- illose outside; stamens 5, the anthers lanceo- loid, 1.3-2 mm long, sessile, inserted below the middle of the corolla tube; ovary globose, 0.4—0.6 mm long, the stigma sessile, conical. Fruits ovoid, white, 3-5 mm long, glabrous. Distribution. Tournefortia bicolor is common in wet forest from sea level to 1,800 m throughout the Neotropics, ranging from Mexico through Central America and the West Indies to northern and western South Amer- ica. In Panama, it is known from all provinces except Herrera. Tournefortia bicolor is closely related to T. hirsutissima (Johnston, 1935) but has gen- erally been considered to be distinct (No- wicke, 1969; Gibson, 1970). Nash & Moreno (1981), however, treated 7. bicolor as a form of T. hirsutissima. Both species are wide- spread in the Neotropics, and although T. bicolor generally occurs in wetter habitats, the two can be found together at numerous localities such as along the shore of Barro Colorado Island (Croat, 1978). Tournefortia bicolor differs from T. hirsutissima in being essentially glabrous and in having sepals 1.5- 2 mm long, a corolla 5.5-6.4 mm long, and corolla lobes 2.3-3 mm long; 7. hirsutissima is generally pubescent, the sepals are 2.5-4 mm long, and the corollas are 3.5-5.3 mm long with lobes 1-1.6 mm long. Morpholog- ical and ecological data strongly support rec- ognition of 7. bicolor. Additional specimens examined. PANAMA. BOCAS DEL TORO: Río San Pedro, Gordon 78C (MO); along runway at Bocas, Lazor et al. 2347 (MO); Chiriqui Lagoon, Water Valley, Wedel 986, 1838 (GH, MO); Chiriqui Lagoon, Old Bank Island, Wedel 1952, 1992, 2090 (GH, MO); Chiriqui Lagoon, Fish Creek Hills, Wedel 2427 (GH, MO); Chiriqui Lagoon, Cocoa Cay, Wedel 2877 (GH, MO). CANAL AREA: Victoria Fill, near Miraflores Locks, Allen 1713 (GH, MO); east slope of Cerro Jefe, 2,700 ft., Blum & Duke 2190 (MO); Barro Colorado Island, Croat 7686, 7822, 8373 (MO); hills S of Pedro Locks, Croat 9176 (MO); Barro Colorado Island, Croat 9560 (MO); road from Gatün Locks to old French Canal and vicinity, Duke & Mussell 6654 (GH, MO); near Gatun Station, Panama railroad, Hayes s.n. (GH); Chagres, Fendler 232 (MO); 1 mi. from Gaillard Highway on small dirt track off Chiva Chiva, 0-25 m, Knapp & Schmalzel 4850 (MO); 2 mi. W of Canal Zone-Ferry Thatcher bridge, Lazor 2189 (MO); Bella Vista, Piper 5380 (GH); ap Colorado Island, Shattuck 807 (MO (2)), 1064, n. (MO); Fort bob near old hospital building #519, Tyson & Blum 3932 (MO); near Gorgas Memorial Lab, White 83 (GH, MO); Barro Colorado Island, Woodworth & Vestal 345 (GH, MO), 482 (GH), 514, 615 (GH, MO). CHIRIQUÍ: disturbed cloud forest at Monte Rey about Boquete, Croat 15770 (GH, MO); windswept ridge 8 km N of Los Planes de Horta. LR.H.E. Fortuna Hydro- iare Project, Knapp 4982 (MO); Quebrada Melliza, uerto Armuelles, 0-150 m, Liesner 507 E MO). iiber El Valle, Aguilar 47 (MO); N rim of El Valle de Anton, 600-1,000 m, Allen 1738 (GH, MO); roadside S of El Valle, D’Arcy et al. 13321 (MO); forest behind Club Campestre, Duke 13259 (MO (2)); hills NE of El Valle de Antón, 2,000 f o Grande on road to Cascajal, 200 yards past continental divide, 450-500 m, Sytsma 3929 (MO); 4 mi. past Llano Grande on road to Cascajal, rocky faced hill ca. 2 km W along continental divide, 600 m, Sytsma 396 1 (MO); 4 mi. past Llano Grande on road to Cascajal, rocky faced hill ca. 2 km W along continental divide, 0 m, Sytsma 3980 (MO). COLON: Portobelo, Billberg 270 (GH); Santa Rita Ridge lumber road, Correa & Dressler 737 (MO); Santa Rita, camino a la zona madere- ra a 15 km NE de la Transisthmica, Correa & Dressler 1137 (MO); Miguel de la Borda, Croat 9853 (MO); Santa Rita Ridge, Croat 13864 (MO); vicinity of Rio Indio on road from Portobelo to Nombre de Dios, Croat 33585 (MO); along Rio Iguanita near bridge along Portobelo road, less than 50 m, Croat 49777 (MO); Santa Rita lumber road, near Agua Clara weather station, Dressler 3851 (MO); Santa Rita East Ridge, Dwyer & Correa 8401 (MO (2); along Río Guanche, 1-4 km S of the Portobelo highway, 0-50 m, Knapp et al. 4610 (MO); Santa Rita Ridge, ca. . E of Transisthmian Highway, Lewis et al. 5252 (MO); Santa Rita Ridge, ca. 4-5.5 mi. E of Transisthmian Highway, Lewis. et ha 5285 (GH, MO); 10 mi. SW of Portobelo, 2-4 m coast, 10-200 m, Liesner 1079 (GH, MO); las ad weedy roadside along Portobelo road at bridge over Rio Viejo, 9 m, Nee & Tyson 10896 (MO (2)). DARIEN: Punta Guayabo Grande, 0-50 m, Antonio & Hahn 4225 (MO); Chucanaque, a lo largo del ultimo hasta Rio Tupisa, Cor- rea & Lazor 1543 (MO); Rio Sambu, 0-5 mi. e us Venado, Duke 9270 (MO); Isla e Duke (MO); p Ag pue Duke 10405 (MO); Rio [oi Duke 8 (MO); coastal thicket near E. uke L. M 40) Hydro “Cam mp Pico Pendejo in oon forest on Rio Sabana, 50 ft., Duke 15450 (MO); Neate to the mouth of the Rio Cosa. Kirkbride & Bristan 1500 (MO); trail between Cana and Boca de Cupe, peru of El Real, along road to Pirre River, Stern et al. 619 (GH, MO); vicinity of Caña, 1,750 ft., Stern et al. 692 (GH, MO); Tucute, Terry & Terry 1397 (GH, MO). Los SANTOS: Punta Mala, Croat 9759 (MO); D'Arcy & Croat 4218 (MO); Loma Prieta, 800-900 m, Duke 11848 (MO); road between Tonosi and Guanica, Stern et al. 33684 508 Annals of the Missouri Botanical Garden (MO); Guayabo, several miles W of Tonosi, Stern 33704 (MO). PANAMA: Santa Lucia, Rio San daos near jen ro Campana, Croat 14676 (MO); El Llano- S Carti Road 8.2 mi. N of Interamerican Highway, 300-3 m, rp 33701 (MO); middle slopes of Cerro Campana from y dime ao n 150 m, Croat 35946 (MO) Cerro Campan O m, D'Arcy 11143 (MO), alrede- dores de Chagres, pun 37 (MO); Cerro Jefe, Duke 9391, 9430 (MO); Río Pacora, just below confluence with Rio Corso, Duke 12029 (MO); San Jose Island, along ie between Bodega Bay and Rio Mata Puerco, Duke 12056 (MO (2); Dema on 7238 (MO); Cerro Jefe, 2, 0 ft., Dwyer et al. 4 (GH, MO); San José Island, Er- lanson 3, 87, oP oa Chagres, Fendler 232 (GH); Taboga Island, behind rocky beach, near hotel, sea level, Gentry 5745 (MO); Llano-Carti Road, 200 m, Hahn 345 (MO); forest and roadside between 6-12 km north of El Llano on Carti road, 1,200 ft., Hammel 854 (MO); 10 mi. from Pan-American Highway on the road from El Llano to Carti, 350 m, Huft et al. 1875 (MO); San Jose Island, Johnston 629, 758, 1355 (GH); 12-16 km above Pan-Am Highway on road from El Llano to Carti-Tupile, 150-400 m, Kennedy et al. 3109 (GH, MO); along road up to Cerro Campana, along edge of montane forest, Lazor 2219 (MO); Panama City, Macbride 2730 (MO); El Llano-Carti Road, 8.5 km from highway, 1,200 ft., Mori 4556 (MO); El Llano-Carti Road, 8.5 km from Inter-American Highway, 350 m, Mori et al. 4556 (MO); El Llano-Carti Road, 8.5 km from Inter-American High- way, 350 m, Mori & Kallunki 5173 (MO), 6.5 km by road N of Lago Cerro Azul, 650-730 m, Nee 9311 (MO); Cerro Jefe, 4.7 mi. above Goofy Lake, 800 m, Sytsma et al. 2812 (MO); El Llano-Carti Road, 9 km above Pan- American Highway, 900- 1,000 ft., Sytsma 3096 (MO), 6 km S of El Valle on highway 71, 2,400 ft., Sytsma & D'Arcy 3558 (MO); El Llano-Carti Road, 6 mi. from Pan-American Highway, 300-400 m, Sytsma 4001 (MO), Isla a 0-186 m, Woodson et al. 1538 (MO). SAN BLAS: Permé, Cooper 218 (GH); Rio Ailigandi, Duke 10834 (MO); mainland i eue Ailigandi, from mouth of Ailigandi river to 2.5 and, Lewis et ee 5 (MO); near Puerto Obaldia, w pepe rilla o La m, McPherson 6963 (MO). VERAGUAS: o w 3 a e] un Š Fe, 400-800 m, Liesner 838 (GH, MO (2)) ku ras brenesii ur Publ. Field s. Nat. Hist., Bot. Ser 989. 1938 TYPE: Costa Rica. e El Silencio (Los Angeles) de San Ramón, Feb. 1933, Brenes 17103 (holotype, F 859855). Shrub to 3 m tall, the branches brown- strigose to pubescent, later glabrous. Leaves alternate but often densely clustered near the ends of branches; petioles 3-10(-27) mm long, glabrous to brown pubescent; leaf blades lanceolate, 9.5-17(-19) cm long, (2-)2.5- 5.5(-6.3) cm wide, the apex acuminate to attenuate, the base acute, the margin entire to unevenly undulate or dentate, the adaxial surface glabrous, the abaxial surface glabrous except for scattered short appressed hairs along the secondary veins, the tertiary and quarternary veins clearly visible. Inflores- cence terminal or subterminal, cymose, branching dichotomously 2-4 times, the pe- duncle to 11 cm long, brown puberulent to strigillose. Flowers on short pedicels to 5 mm long, borne (2-)4-10 mm apart; sepals 5, ovate to lanceolate, 4.3-4.5 mm long, ca. 1.7 mm wide, glabrous to sparsely strigillose; corolla white, tubular with spreading ovate to lanceolate lobes, these acute at the apex, 2.6— 2.7 mm long, the tube 4.8-5 mm long, strigil- lose outside; stamens 5, the anthers oblong to lanceoloid, sessile or nearly so, inserted in the middle of the corolla tube; ovary globose, ca. 1 mm long, the disc crateriform, the style ca. 1 mm long, the stigma pyramidal. Fruit ovoid, 6-8 mm long, 5-6 mm wide, white at maturity, often capped by the persistent stig- ma. Distribution. Tournefortia brenesii is restricted to cloud forests from 800-1,350 m in elevation in Panama and Costa Rica. In Panama it is known only from the province of Veraguas. Tournefortia brenesii is distinct in its com- pact habit of growth, with the leaves clustered densely near the branch tips, and broad, ac- crescent sepals. ae and specimens examined. MA. VER- i of Escuela Agricultura Alto Piedra near a Fe, along i trail to top of Cerro Tute, 3,500 ft., ana 4957 (MO); Cerro Tute, ride up fro rne. Escuela ecc Santa Fe, 1,000-1,300 m, Hamilton & Dressler 3080 (MO); ridge of ua de Tute, trail to Cerro Tute, above vedi gricola Alto de Piedra, just W of Santa Fe, 800-1,350 m, Knapp 2409 d Corio Tue, ca. 10 m above 1, 000 m, Mori 6760 (MO). Tournefortia cuspidata Kunth in Humb., Bonpl. & Kunth, Nov. Gen. Sp. 3: 83. Volume 75, Number 2 1988 Miller 509 Revision of Panamanian Boraginaceae 1818. TYPE: Ecuador. Guayas: Guaya- quil, Humboldt & Bonpland s.n. (ho- lotype, P, not seen). ddr iiw obscura A. DC. in DC., Prodr. 9: 517 845. TYPE: Guyan. a: Schomburgk 571 (holotype, -DC, not seen; barr he, MO). Scandent woody vine or sprawling shrub to 2 m tall, the twigs villose with hairs to 4 mm long. Leaves alternate; petioles 3-8 mm long, stout, shaggy-villose; blade lance- vin elliptic, ovate, or rarely obovate, 8-13(-1 cm long, 3-6(-9) cm wide, the apex acu- minate, the base acute to obtuse, the margin entire, the adaxial surface strigose, the abax- ial surface strigose to nearly villous, especially along the veins. Inflorescence terminal, a few- to much-branched, dense cyme, the sterile portions of branches shaggy villous. Flowers sessile, Dad closely spaced; sepals lanceolate, 9. m long, strigose to hirsute; corolla white to green, 5-merous, the lobes ovate, 1— 1.8 mm long, the tube 6.5-8 mm long, stri- gose to sericeous on the outside; stamens 5, the anthers lanceoloid, 1.3-1.5 mm long, ses- sile, inserted beneath the middle of the corolla tube; ovary globose to ovoid, ca. 1 mm long, the style 0.4-0.5 mm long, the stigma con- ical. Fruits ovoid, 3-4.5 mm long, white at maturity, breaking into (2-)4 nutlets at ma- turity, glabrous. Distribution. Tournefortia cuspidata is common in disturbed areas of moist to wet forests at elevations below 400 m from Mex- ico to South America, and it occurs on Trin- idad and Tobago. In Panama it is known from Bocas del Toro, Canal Area, Coclé, Colon, Darién, Panamá, and San Blas. Tournefortia cuspidata is a distinctive species readily recognized by its shaggy, vil- lous twigs with the hairs up to 4 mm long and sepals greater than 5 mm long. This is one of the most commonly collected species in southern Central America and northern South America. It is perhaps most closely related to T. hirsutissima but differs in its much longer hairs and sepals. Although the long hairs on the stems are usually diagnostic, these are greatly reduced in plants from ad- jacent Chocó, Colombia. Additional specimens examin ed. PANAMA. BOCAS DEL TORO: along railroad tracks near station at Milla 5, Croat & Porter 16493 (MO); Changuinola to 5 mi. S at junction 100-200 ft., Lewis et Escobal, Croat 12445, 13114 (MO); Gaillard Highway between Paraiso and the Continental Divide, Croat 14833 (MO); Pipeline Road to 18 km amboa, D’Arcy s.n. (MO); Barro Colorado Island, Dwyer et al. 8447 (MO); Gamboa, Naval Pil ay Ebinger 892 (GH, MO), Pipeline Road, 5-6 mi. amboa, 100-200 m, Gen- try 667 1 (MO); Barro hd Island, ge pa 41 (MO); NW shore of Gatún Lake, ca. 4 mi io Chagres, Leuis 1823 (MO); Pipeline ies eid mile markers O and 11.1, ca. 16 mi. a, Lewis et al. 5452 (MO); T ue Road, 14.4 ps from Chagres Airport Road, 00 m, Mori & Kallunki 2062 (MO); Barro ique TH Shattuck 100 (GH, MO (2)), 676 (GH, MO (3)), 957 (MO); Starry 228 (MO); Whetmore & Abbe 201 (GH (2), MO); Whetmore & Woodworth 854 (GH); along the margin of Pipeline Road N o ilbur & We m 5696 € — Rio San i ae. n (MO). near pes along river, Croat bid y; Peluca, ca. 2.7 km fr om Tran pe ridi High- way on road to No nie de pad sie 2638 Pia along Rio Bo oquerón above Min n #1 (ma ganese mine), main valley of ne Rio sali, 100- 200 m, Knapp & Sytsma 2465 (MO); along Rio Bo- querón near No.l (manganese mine), E of Salamanca, A (MO); roadside and woods R Allen 4638 (MO); vicinity of Canglon, trail to south, 1 km W of bridge in n, 50 ft., Antonio 4584 (MO); camino de El Real a & Lazor 1555 (MO); vicinity of El Real along road to airport, Croat & Porter 15445 (MO), El Real, D'Arcy 5512 (MO); 2-3 mi. SE of El Real, Duke 4853 (MO); road from El Real to Pinogana, Duke 5001 (GH (2), MO); Isla Boca Grande, Duke 8843 (MO); Río Sambu, 0-5 mi. above Rio Venado, Duke 9268 (MO); Enseriada del Guayabo, 16-19 km SE of Jaqué, Garwood MO); Manene to mouth of Rio Cuasi, Kirkbride & Bristan 1415 (MO); Rio Tuquesa, at lower Tuquesa Min- ing Company camp called Charco Chiva, 100 m, Mori 510 Annals of the Missouri Botanical Garden 5 (MO); vicinity of Piu along Rio Chucunaque, E T bridge, trail between Paya and Payita, MO); Gold n at Cana, 480 m, Sullivan Ps (MO), 3 mi. E of Santa Fe, Tyson et al. 4652 (MO); 2 mi. E of Santa Fe, Tyson 4818 (MO). PANAMÁ: Rio Pita, 0-2 mi. above tidal limit, Duke 4793 (MO (2)); forests along headwaters of Rio Corso, off Rio Pacora, 500 m, 11929 (MO); N of El Llano, 500-800 m, Gentry 5560 (MO); Pipeline Road, 50 m, Hamilton et d 3266 (MO); 10 mi. from Pan-American Highway on the road from El Llano to Carti, 350 m, Hufi et al. 1874 (MO); Alcalde the Serranía de Maje, 50-100 m, Knapp & Mallet a (MO); El Llano-Carti Road, 6-10 mi. from Pan-American Highway, 400 m, Miller et al. 857 (MO); along » Pan- American Highway 6.5 mi. E of highway checkpoint at turnoff to Chepo, Miller et al. 1019 (MO); Gorgas Me- morial Labs yellow fever research camp, ca. 25 NE of Cerro Azul on Rio Piedras, 550 m, Mori & during 3464 (MO); Arenoso, lower Rio Trinidad, 26-50 Seibert 625 (GH (2), MO); 2 mi. E of El Llano, Tyso son 1737 (MOy 5 mi. E of Canita or 23 mi. E of Chepo, Tyson & Smith 4152 (MO); 1 mi. E of El Llano, Tyson 6883 (MO); La Rana de Oro, Pedregal, Villamil 33 (MO), wet savanna E of Pacora, 25 m, Woodson et al. 721 Sa 722 (GH (2), MO). SAN BLAS: El Llano-Carti Road, 3 , de Nevers & Herrera 4222 (MO); trail from eee Obaldia to La Bonga, tributary of the Rio Armila, a. 2 hours walk from Puerto Obaldia, 0-50 m, Knapp & Mallet 4679 (MO); along newly cut road from El Llano to Carti-Tupile, continental divide to 1 mi. from divide, 300-500 m, Liesner 1297 (GH, MO). Tournefortia glabra L., Sp. Pl. 141. 1753. YPE: without locality or collector (ho- lotype, LINN (Savage Catalog number 193.5), not seen; microfiche, ). Small tree or shrub to 5(-10) m tall, the twigs glabrous to sparsely strigillose. Leaves alternate; petioles 1—4(-5) cm long, sparsely strigillose to nearly glabrous; leaf blade nar- rowly elliptic to lance-obovate, (8-)12-18 (7-25) cm long, (2.5-)5-8(-12) cm wide, the apex acuminate, the base attenuate and usu- ally strongly decurrent, the margin entire, the adaxial surface sparsely strigillose to nearly glabrous, the abaxial surface with a few hairs scattered along the veins, the tertiary veins evident. Inflorescence internodal or terminal, a loosely branched cyme, the peduncle 3-8 (714) cm long, sparsely strigillose to nearly glabrous, the branches 3-8 cm long. Flowers sessile, borne (1-)2-3(-4) mm apart; sepals lanceolate, 1-1.2 mm long, sparsely strigillose; corolla white or greenish white, 5-merous, the lobes lanceolate, 1.5-2 m long, the tube 3.5-4 mm long, strigillose out- side; stamens 5, the anthers ellipsoid, 0.6-1 mm long, sessile or nearly so, inserted above the middle of the corolla tube; ovary ovoid, 0.7-1.5 mm long, the disc crateriform to scarcely evident, the style 1-1.2 mm long, the stigma capitate. Fruits white at maturity, very broadly ovoid, 3-4 mm long, glabrous. Distribution. Tournefortia glabra oc- curs in diverse moist to wet habitats from sea level to 2,700 m in elevation. It is found from Mexico through Central and South America and in the West Indies. In Panama, it is known from Bocas del Toro, Chiriqui, Cocle, olón, Panamá, San Blas, and Veraguas. Tournefortia glabra is one of the few small understory trees in the genus found in Central America. This species is vegetatively quite similar to 7. acutiflora Martens & Galeotti of northern Central America and southern Mexico, but differs in having calyx lobes shorter than 1.5 mm long. Additional specimens examined. PANAMA. BOCAS DEL TORO: region of Almirante, Daytonia far GH, y along road to Chiriquí Grar Mc ir 7391 (MO); Chiriqui b Water Valley, IV. 1573, 1746 Qs MO). CHIRIQUÍ: 12 mi. fro ITO ipid l, pe m, ns ~ Gualaca on road t 1751 (MO); Cerro Colorado i. from Lower Elevation Camp, 4,20 Antonio 490 a Palo Alto-Boquete, 1,300-1, Tam n 4 km del Hato nde Volcán en el camino a a Las e ado rre & Lazor 1461 iia of Boquete along jd niis and pastures Cerro Azul near Quebrada Jaramillo, 1,500-1,620 m, a 26781 (GH, MO); Chiquero, Bo- quete, Davidson 552 (GH, MO); TTC-BMI Cloud Forest Litter Study Cerro Horqueta, 1,500 m, Duke et al. 13637 MO); NW of Boquete, Cerro Horqueta, 5,000-5,800 ft., Dwyer et al. 478 (MO); Boquete, Finca Collins, 5,000 ft., Dwyer & Hayden 7644 (MO); Boquete, Fred Collins Finca. Ebinger 667 (MO); road to Rio Serano, Folsom et al. 2107 (MO); path from Linares farm ca. 1,400 m to top of Cerro Hornito at 1,750 m, Folsom et al. 7227 (MO); Las Lagunas area of Hato del Volcán, 1,400 m, Hamilton & Stockwell 3574 (MO); rado, 1,200-1,4 road along Rio Palo Alto ca. me ME 1858 (MO); Collins Finca, 2,000 m, Kirkbride MO); N of San Felix, 36.2 km by road from the bind iu Highway, Mori & Kallunki 6017 (MO); Dos Lagunas, 4 km W of El Hato del Volcán, 1,300 m, Volume 75, Number 2 1988 Miller 511 Revision of Panamanian Boraginaceae Mori & Bolten 7400 (MO); Las Lagunas, 2 mi. SW of El Volcán, 4,200 ft., Tyson 850 (MO); Rio Chiriqui Viejo mbil: White 46 (GH, MO); Finca Lerida to Peña Blanca, 1,750-2,000 m, Woodson & Schery , MO); vicinity of Casita Alta. Volcán de Chiriqui, 1,500-2,000 m, Woodson et al. 986 (GH, MO). cocLÉ: El Valle, 600-1,000 m, Allen 1200 (GH), 1794 (GH, MO); La Mesa above El Valle, 810 m, Croat 25298 (GH, MO); Cerro Pilon near El Valle, Duke 12060 (MO); El Valle de Antón at the foot of Cerro Pilon, 2,000 ft., Dwyer & Correa 7964 (MO); El Valle de Antón at the foot of Cerro Pilon, 2,000 ft., Dwyer & Correa 7965 (MO); Cerro Pilon, 2,700 ft., Dwyer & Lallathin 8610 (MO); vicinity of La Mesa above El Valle, 900 m, Gentry 7408 (GH, MO); hills above El Valle, 1,000 m, Gentry 6895 (MO); La Mesa region N of Cerro Gaital vicinity of El Valle, 2,400 ft., Hammel 3899 (MO); E Valle de El Valle Wepcor site, on tra en site, Kirkbride 1086 (MO); foothills and summit of Cerro Caracoral, near La Mesa N of El Valle de Antón, 800- 1,100 m, Knapp 1111 (MO); Cerro Pilon, 900-1,173 m, Liesner 768 (MO); along road above El Valle toward upper ridges above town, Miller et al. 775 (MO); foot of Cerro Pilón, above El Valle de Antón, 2,000 ft., Porter et al. 4640 (MO); NE slopes and summit of Cerro Caraco- ral, north rim of El Valle, 2,700-3, 200 ft., Sytsma 4051 rom 600 ft., Hammel 2708 (MO). DARIÉN: between Quebrada Venado and Peje swamp on the headwaters of Río Tu- uesa, Bristan 1006 (MO); Rio Ucurganti, Bristan 1139 (MO); trail SE of Manene to Rio Et gh pad 12197 (MO); premontane rainforest 0-2 m Tres Bocas along the shortest headwater of Rio as Kirkbride & Duke 1144 (MO (2); Rio Tuquesa, at middle Tuquesa Mining Company camp called Charco Peje, 250 m, Mori 7002, 7093 (MO); Paya, Rio Paya, Stern et al. 396 FSU building, PA 14809 (MO); Cerro Campana, 800 m, D’Arcy 11141 (MO); Cerro Campana, 2,500 ft., p: Arcy & eee 14931 (MO); Be Bayano, near crossing of Pan-Am Highway, above confluence with Rio Chepo, Duke 3974 (MO (2)), 3998 (MO): Piria-Canasas Trail near Piria, 100 m, Duke 14330 (MO); area around Torti Arriba, Folsom 5129 (MO); near top of Cerro Cam- pana, above cabin, Gentry 0); 3.8 mi. E of Río Ipeti, S reri i river and into lower slopes of Serrania e Maje, 200 Huft & Jacobs 2002 (MO); near Cerro Sonia. on pue radiating from end of road which Pan-American Highway, foothills of the Serrania de Maje, 50-100 m, Knapp & Mallet 5167 (MO); Cerro Campana, trails just inside entrance to Parque Nacional, 850 m, Miller & Miller 995 (MO); dy Nacional Cerro Campana, 2 km N of highway 707, 850 m, Stein E ý roma 1140 (MO); base of Serranía de Canazas, a. 15 km SW of Canaza near Rio Torti, 150 m, Stein 1387 (MO); steep slopes S of Canita, 200 m, Webster et al. 16885 (MO); lower slopes and trail to Cerro Cam- pana, pad 8400 (MO). SAN BLAS: El Llano-Carti Road, km rail id to bes 50 m, de Neve s Her gt ead- waters of Río Mulat a "lias 1751 (MO). v US UAS: El Cuchillo, near Cerro Tute, up from Santa Fe, 1,300 m Hamilton et al 1240 ( Tournefortia hirsutissima L., Sp. Pl. 140. 1753. Messerschmidtia hirsutissima (L.) Roemer & Schultes, Syst. Veg. 4: 451. 1819. TYPE: without locality or col- lector (holotype, LINN (Savage Catalog number 193.1), not seen; microfiche, Tournefortia Mida os DC., Prodr. 9: 517. 1845. : 1837, Schomburgk 70 (holotype, G-DC, dede seen; microfiche, MO; isotype, K). Woody vine, sprawling shrub, or small tree, the twigs strigose to hirsute. Leaves alternate; petioles (3-)8-15(-20) mm long, strigose to hirsute; leaf blade lance-ovate to narrowly elliptic, (7-)11-18(-20) cm long, (2.5-)3.5—- 6(-8) cm wide, the apex acuminate to acute, the base acute to obtuse, the margin entire, the adaxial surface strigose, the abaxial sur- face strigose to pubescent, the tertiary veins evident. Inflorescence terminal or rarely internodal or axillary, a much-branched, dense cyme, the peduncle (1-)2-4(-5) cm long, strigose to hirsute, the fertile branches 1.5- 4(-6) cm long. Flowers sessile, borne 1-2 (-3) mm apart; sepals lanceolate, 2.5-4 mm long, strigillose; corolla white, 5-merous, the lobes ovate, 1-1.6 mm long, the tube 3.5- 5.3 mm long, strigillose outside; stamens 5, the anthers lanceoloid, 1.1—1.3 mm long, ses- sile, inserted beneath the middle of the corolla tube; ovary globose, 0.8-1 mm long, the disc scarcely evident, the stigma nearly sessile, capitate. Fruits white, ovoid, 3-4 mm long, strigillose. Distribution. Tournefortia hirsutissima is found in diverse habitats from sea level to 2,000 m in elevation, and is ubiquitous throughout the Neotropics. In Panama, it is nown from the Canal Area, Chiriqui, Darien, Panamá, and San Blas. Tournefortia hirsutissima is one of the 512 Annals of the Missouri Botanical Garden most commonly collected members of the ge- nus and appears to be most closely related to T. bicolor (discussion under that species). The two can be distinguished most easily by the usually strigose upper leaf surface in T. hir- sutissima, while it is glabrous or nearly so in T. bicolor. One collection from Darién (Stern 501) is unique in having a white abaxial leaf surface, but it otherwise fits well within 7. hirsutissima, agreeing in all other aspects. Additional specimens examined. PANAMA. CANAL Vit forest reserve, near crossing of Cruces trail and Madden Dam road, Bartlett & eom 16348 (GH, MO); Juan Mina, Bartlett & Lasser 16529 (GH, MO); Barro Colorado Island, Croat 4797, 6246, 10750A, 11718 (MO); de a of Gatün Locks, Croat 12432 (MO); vicinity of Fort San Lorenzo, Croat 12521 (MO); Barro Colorado Island, aye 15054 (MO); along road between Gatún Locks and Ft. Sherman, Croat 15380 (MO); hill above Thatcher-Ferry Bridge, Croat 17014 (MO); Fort ~ O = Elias 7513 (MO); Gatün Sidon. Hayes 6 1(GH); vicinity of Rio Chagres Bridge, road leading to abandoned weather post, ca. 2 km by roa of Gamboa along highway to Balboa, e 427 1 (MO); west side of Ferry Thatch- er Bridge, a mangrove margin, Lazor 2881 v ni Pipeline Road, Meyer 160 (MO); Pipeline Road, mi. from 100 m, Miller 1027 (MO); long T from Fo Sh an to Fort San ee orte ss Are U oed E aa Dam, near Rio eiber ; Barro Colorado Island: near mouth o Chagres iver, Tyso For ur uae Tyson 1571, Blum 3 8 (GH, MO); Barro Colorado Island, Wilson 44 WEN bae Llanos on slopes of Volcán de Chiriqui Viejo and along Río Chiriqui Viejo, 1,200 m, Allen 968 (MO); Boquete, lumber road into the hills east of the Rio Caldera, 4,500-6,500 ft., Allen 4654 (GH, MO); Burica Peninsula, Quebrada Manzanillo, 9 km SSW of Puerto Armuelles, Busey 746 (MO); 1 mi. W of airport at Puerto Armuelles, Croat 21905 (MO); Puerto Limónes, along coast near the beach, Croat 22119 (MO); Burica Pen- insula, Distrito Guanabano, disturbed areas along Que brada Quanabano, 0-100 m, Croat 22534 (MO); San Bartolo Arriba W of Puerto Armuelles, Croat 26700 (MO) Las Lagunas area W of Hato del Volcán, 1,400 m, Hamilton & Stockwell 3562 (MO); Burica Peninsula, 2.5 km W of Puerto Armuelles, 80 m, Liesner 16 (MO); Burica Peninsula, Quebrada Merica, 4 mi. S of Puerto idee 0-100 m, Liesner 384 (MO); along road 3- mi. NW of El Hato del Volcán towards Costa Rica across the Rio Chiriquí Viejo, 3,000-4,000 ft., Luteyn 859 (MO); valley of the upper Rio Gariche, 1,050-1,100 m, Seibert 357 (GH, MO). COLON: 4 mi. ortobelo, near sea level, Antonio 1764 ( 269 (GH); Isla Grande, D'Arcy 4033B (MO); Portobelo, Dwyer 5146 (GH); La Guayra, E of Portobelo, Knapp & Mallet 5725 (MO). DARIEN: headwaters of Rio Chico, o MO); Portobelo, Billberg 500-750 ft., Allen 4638 (GH); vicinity of Cana, 1,750 ft., Stern et al. 501 (GH). PANAMA: a orillas de la carretera Pacora 1 km despues del Autodromo Panamá, Car- rasquilla 191 (MO); a 4 km del corregimiento de Pacora, Carrera 19 (MO); vicinity of Macambo, Croat 14906A O); Cerro Azul, Dwyer 1876 (MO); Tocumen, Dwyer 4056 (MO (2); Las Guacas, población a orillas del Río Chagres, Kant 39 (MO); El Llano-Carti Road, 18 km from [nteramerican H L (MO); El L al. 45 ] Llano-Carti Road 17.5 km from Interamerican Highway, 350 m, Mori et al. 4623 (MO); road to Cerro Campana, 10 km from Interamerican Highway, m, Sullivan 438 (MO); between Chepo and La Capitana, Tyson 6758 (MO (2); wet savanna E of Pacora, Woodson et al. 721 (GH (2)). SAN BLAS: through cultivation on mainland in front of Ustupo, D’Arcy 9507 (M Tournefortia johnstonii Standley, Publ. Field Mus. Nat. Hist., Bot. Ser. 18: 991. 1938. TYPE: Costa Rica. Heredia: Cerro de las Caricias, north of San Isidro, 2,000-2,400 m, P. C. Standley & J Valerio 52087 (holotype, US 1306982). Shrub to 3 m tall, or rarely a small tree to 6 m tall, the twigs velutinous. Leaves op- posite, the lateral buds usually with small ex- anding leaves with appearance of stipules; petioles 9-30(-40) mm long, velutinous; leaf blade lanceolate to ovate, (7.8-)10-18(-23) cm long, (3-)4-8(-10) cm wide, the apex acuminate, the base cuneate to acute, the margin entire to unevenly undulate, the adax- ial surface velutinous, the abaxial surface ve- lutinous, the secondary veins 10-12, the ter- tiary and quaternary veins clearly evident. Inflorescence terminal, a once- or rarely twice- branched cyme, the branches (2.5-)4— long, the peduncle (3-)5-12 cm long, velu- tinous. Flowers bisexual, sessile, borne 1-5 (-7) mm apart; sepals 5, lanceolate, 6-10.5 mm long, strigillose to strigose; corolla tubular with spreading lobes, white to green, 5-mer- ous, the lobes lance-ovate to ovate, the apex acuminate, 2.6-5 mm long, the tube 8-10 mm long, strigillose, at least near the apex; stamens 5, the anthers lanceoloid, 1.5-2 mm long, sessile, inserted just below the mouth of the corolla tube; ovary ovoid, 0.8-1 mm long, the disc scarcely evident, the style 3-3.5 mm long, the stigma conical. Fruits white, ovoid, 4.5-7 mm long, separating into 2-4 nutlets at maturity. Volume 75, Number 2 1988 Miller 513 Revision of Panamanian Boraginaceae Distribution. Tournefortia johnstonii occurs only in Panama and Costa Rica in cloud forests at elevations of 1,000-3,000 m. In Panama it is known only from the provinces of Chiriqui and Veraguas. Tournefortia johnstonii is very similar in general aspect to T. ramonensis but differs in being a smaller plant with longer, coarser indument on the twigs. Also, the lower leaf surfaces of T. johnstonii are usually darker in color after drying. Additional specimens examined. PANAMA. CHIRIQUÍ: 2,000 m, Croat 26456 (MO); E of Boquete along steep forested slopes and in wooded pastures on Cerro Azul near Quebrada Jaramillo, 1,620-1,700 m, Croat 26867 (MO); Alto Respinga and above, 2,800 m, D'Arcy 999 ] (MO); mountain directly S of Cerro Respinga, D'Arcy 10811 (MO); E slope of Volcan de Chiriqui (Baru), above Boquete, along road in oak forest, 2,600 m, Davidse & D'Arcy 10264 (MO); Volcan de Chiriquí, 9,000 ft., Da- vidson 976 (GH, MO); along Boquete trail, Cerro Res- pinga, 2,000-2,500 m, Gentry 5978, 6013 (MO); path above Cerro Punta to Boquete, 2,500 m, Hamilton & Stockwell 3417 (MO); Hamilton & Krager 3741 (MO); km NW of Cerro Punta, 7,200 ft., Hammel 1472 (MO); Cerro Hen 5,600 ft., Hammel 3043 (MO); 2r Mir Alemán, 8 mi. N of Los Planes de Hornito I.R.H.E. “sasan Hxdroelectrie Project, 1,200 m, et al. 4155 (MO); trail from Cerro Punta to Boquete, 2,160-2,260 m, McDaniels 10255 (MO); N of San Felix at Chiriquí- Bocas del Toro border, on Cerro Colorado copper mine road along continental divide, 5,000-5,500 ft., Mori € Kallunki 5939 (MO); 3.7 km E of bridge NE of Cerro Punta on road through Bajo Grande, 2,250- 2,400 m, Stevens 18199 (MO); 3.7 km along road through Bajo Grande from bridge NE of Cerro Punta, 2,250- 2,400 m, Sytsma & Stevens 2143 (MO); along the n between Cerro Punta and the Quebrado Bajo Gran 2,000-2,100 m, Wilbur et al. 11903 (MO); rini Chiriqui, 1,500- VERAGUAS: Cerro Tute, "E slopes, 1 km beyond Escuela Agricola Alto Piedra above Santa Fe, 1,200-1,450 m, a & Anderson 4593 (MO Tournefortia longispica James 5. Miller, sp. nov. TYPE: Panama. Bocas del Toro: road from Fortuna Dam to Chiriqui Grande, 3.1 mi. N of continental divide, 700 m, disturbed primary forest, G. McPherson 6778 MO 3386969). Figure 6. (holotype, Suffrutex ad 2 m altus, ramulis glabris. Folia alterna; petioli h 5- )2.5-4(-6.5) cm longi; lamina elliptica ad oblonga, (14-)1 7-25 cm EC (5.5-)6.5-9(-10.5) cm minata, basi acuta ad dipsa glabra. i ramorum duorum, ad 30 nthesin composita, bE 10-22 cm longo, glabro vel fere gla ro. Flores sessiles, 1-5 mm ane uum lanceolata, 2.5-3 mm longa; corolla alba ad p o-viridis, lcd lobis patulis, ovatis ad late an 2-2. 5 mm longis, t tubo extus strigilloso; an- theris 1.5-2 mm longis, sessilibu stylus 0.6-1 mm longus, stigmate conoidi. Fructus globos ad valde lato-ovoidei, 6-9 mm longi, 9-12 mm lati, glabri, laeves Shrub to 2 m tall, the twigs glabrous. Leaves alternate; the petioles (1.5-)2.5-4(-6.5) cm long, glabrous or nearly so; leaf blade elliptic to oblong, (14-)17-25 cm long, (5.5-)6.5- 9(-10.5) cm wide, the apex acuminate, the base acute to attenuate, the margin entire, the adaxial surface glabrous, the abaxial sur- face glabrous but with very small appressed hairs along the major veins, the secondary veins 8-11, the tertiary veins evident, the quaternary veins often obscure. Inflorescence terminal or subterminal, usually a twice- branched cyme, the branches elongate, to 30 cm long at fruiting, the peduncle 10-22 cm long, glabrous to shortly brown puberulent or strigillose. Flowers sessile, bisexual, borne l- 5 mm apart; sepals 5, lanceolate, 2.5-3 mm long, sparsely to evenly strigillose; corolla tu- bular with spreading lobes, white to pale green, 5-merous, the lobes ovate to widely ovate, 2- 2.5 mm long, the tube 4—5.5 mm long, strigil- lose outside; stamens 5, the anthers lanceo- loid, 1.5-2 mm long, sessile, inserted in the middle of the corolla tube; ovary ovoid, 0.8- 1 mm long, the disc crateriform or barely evident, the style 0.6-1 mm long, the stigma conical. Fruits white, globose to very broadly ovoid, 6-9 mm long, 9-12 mm broad, gla- brous, smooth. Distribution. Tournefortia longispica is known only from Panama from the provinces of Bocas del Toro, Chiriqui, Coclé, and Vera- guas, where it occurs in cloud forests at el- evations of 600-1,500 m. 514 Annals of the Missouri Botanical Garden FIGURE 6. D. Fruits. A— SOR de l Toro, Panam Tournefortia longispica is distinct within the genus in having an elongate, dichoto- mously branched inflorescence and a nearly glabrous leaves. It has no known clos relatives in Panama but could be M with T. glabra, which differs in having a much-branched inflorescence. Several of the Tournefortia longispica. — 4. Flowering branch.—B. Flow er.—C. Flo wer with Candle opened.— McPherson 6778 (MO), Bocas del Toro, Panama; D from McPherson 7247 (MO), Bocas collections have large, swollen fruits appar- ently parasitized by insects, a common con- dition in the genus. Additional specimens examined. PANAMA. BOCAS DEL TORO: E slope 24 Finca Serrano, 4,600 ft., Hammel 6166 (MO); along trail on Volume 75, Number 2 1988 Miller 515 Revision of Panamanian Boraginaceae divide separating Chiriqui and Bocas del Toro, 1,150 m, McPherson 7247 (MO). CHIRIQUÍ: Fortuna Dam area to N of reservoir near Quebrada Bonito, 1,100 m, Churchill 5811 (MO); W side of Rio Hornito, 1,100-1,300 m, D'Arcy 16009 (MO). cocL£: El Copé on Pacific side V4 hour walk from sawmill, Antonio 2109 (MO); Cerro Pilón, El Va lle, 3,000 ft., Duke & Lallathin 14962 (MO); fe, wyer & Lallathin 8687 (MO); 2658 (MO); New Works at Rivera o Calvario, is 800 m, Folsom 3164 (MO); Cerro rn . 000- 1,17 pine 786 (MO). vERAGUAS: Ridge of Cordillens fi Tute, rail to Cerro Tut Escuela Agricola Alto de Piedra, just W of "e Fe, 800-1,350 m, Knapp & Dressler 5425 (MO). Tournefortia maculata Jacq., Enum. Syst. Pl. 14. 1760; Select. Stirp. Amer. Hist. 47. 1763. TYPE: Colombia, Bolivar: Car- tagena (not seen). Woody climbing vine, the twigs sparsely strigillose, later glabrous. Leaves alternate; petioles (3-)5-12 mm long, sparsely short- strigillose; leaf blade (3-)5-9 cm long, (1.5-) 2.5-4 cm wide, the apex acuminate to at- tenuate, the base acute to obtuse and often slightly decurrent, the margin entire, the adaxial surface very sparsely short- seca lice. the abaxial surface evenly short-strigillose. Inflorescence usually terminal, a loose, much- branched cyme, the branches to 5 cm long, the peduncles 18-24 mm long, puberulent to strigillose. Flowers sessile, borne 1-10 mm apart, bisexual; sepals lanceolate to triangu- lar, 0.8-2 mm long, strigillose; corolla pale yellow-green, 5-merous, the iform, 1.5-3 mm long, the tube 3.3-5 mm long, densely strigillose; stamens 5, the anthers lan- ceoloid, 0.8-1 mm long, connate apically, nearly sessile, inserted in the mouth of the corolla tube; ovary ovoid, 0.8-1 mm long, the disc crateriform to scarcely evident, the style 0.8-4 mm long, the stigma capitate. Fruits 4-lobed, 2.5-4 mm long, glabrous, white with black markings. Distribution. Tournefortia maculata ranges from Mexico to South America and the West Indies and is found in a wide variety of habitats from sea level to 1,500 m in elevation. In Panama it is known from most provinces and probably occurs in all of them. Tournefortia maculata is distinctive, with its four-lobed fruits and glabrous leaves. The only other Central American Tournefortia with four-lobed fruit is T. volubilis, which differs in having sericeous leaves. These two belong to sect. Cyphocyema I. M. Johnston, a com- plex group of species centered in South Amer- ica and characterized by four-lobed fruits and apically connate anthers. Additional specimens examin ned. PANAMA. CANAL Maxon & Valentine 6942 (GH); Fort San Lorenzo, Fort Sherman Military Reservation, Maxon & Valentin 2,000 m, Woodson 1007 (GH). garitas and El Valles, Woodson et al. 1279 (GH, MO). COLON: Santa Rita Ridge Road, 4 mi. from Transisthmian Highway to Agua Clara weather station, 500 m, uud et al. 8835 (MO (2); Santa Rita Ridge, ca. 5.5-6 m E of Transisthmian Highway, Lewis et al. 5387 (MO); along Rio Viejo, between de Portobelo Road and Que- brada Ruiz, 4 km NE of Puerto Pilon, 5 m, Nee 7171 (MO). DARIEN: Rio Tuqueza below Quebrada Venado, Bristan 1076 (MO); Cerro Pirre, 2,500-4,500 ft., Duke & Elias 13665 (MO). Los SANTOS: Loma Prieta, 800- 900 m, Duke 11879 (MO); Pocri, Dwyer 1193 (MO); Guarare, Dwyer 2469 (MO); Loma Prieta, Cerro Grande, 2,400-2,800 ft., Lewis et al. 2241 (MO). PANAMÁ: Cerro Azul, Goofy Lake, Dwyer 2412 (MO); San José Island, Erlanson 114, 150, 241, 249, 379, 388, 501 (GH); Taboga Island, 0-25 m, Pittier 3536 (GH); Cerro Jefe, to Udirbi, El gll Road, 200-350 m, D'Arcy et al. 16037 (MO). vERAGUAS: along road between Santa Fe and Escuela Agricola Alto Piedra, 600-800 m, Croat 26007 (CH, MO); Rio de Jesus, Dwyer 339 (MO); Isla de Coiba (Penal Colony), Dwyer 2330 (MO); 5 mi. E of Santiago, Tyson et al. 4284 (MO). Tournefortia multiflora James S. Miller, sp. nov. TYPE: Panama. Veraguas: NW of Santa Fe, 8.8 km from Escuela Agri- cola Alto de Piedra, on road to Calove- bora, elev. 1,900 ft., S. Mori 6659 (ho- lotype, MO 3386967). Figure 7 Frutex vel arbor parva ad 5 m alta, ramulis glabris. Folia alterna sed saepe prope apices caulis conferta, paene sessilia vel in petiolis ad 2(- 7) cm = s. lamina elliptica, 16-50 cm longa, m 18 cm apice obtuso ad rotundato et saepe abrupta brevi- elogia lasa ase at- tenuata quasi glabra. ia ia terminalis, paniculata, 516 Annals of the Missouri Botanical Garden FicuRE 7. Tournefortia multiflora.— A. Flowering branch.—B. Flowers.—C. Flower with corolla opened showing dea anthers. From Mori 6659 (MO), Veraguas, Panama. 20-40 cm longa, 16-30 cm lata, pedunculo 10-19 cm longo, fusco-puberulo. Flores sessiles, seorsum 2-5 mm dispositi; sepala 5, lanceolata, 1.7-3.2 mm longa; corolla viridi-alba, striis atroviridibus ornata, urceolata lobis pa- tulis, pentameris, lobis linearibus, 3-4.5 mm ee tubo 3 ongo, extus puberulo; stamina 5, antheris conspicua bilobatis et in ore be tube nidos .5-0.8 mm longis; stylus 0.5-1.6 mm longus, stigmate darse, puberulo. Fructus albi, ovoidei, 4.5-5.5 mm lon- , 3-4 mm lati, glabro Shrub or small tree to 5 m tall, the twigs glabrous. Leaves alternate but often crowded near the stem apices, nearly sessile or on petioles to 2(-7) cm long; blade elliptic, 16— Volume 75, Number 2 1988 Miller Revision of Panamanian Boraginaceae 517 50 cm long, 8-18 cm wide, the apex obtuse to rounded and often abruptly short-acumi- nate, the base attenuate, the margin entire, the adaxial surface essentially glabrous with widely scattered, short, appressed hairs, the abaxial surface glabrous, strigillose along the midrib and the 15-22 arcuate secondary veins, the tertiary and quaternary veins clear- ly evident. Inflorescence terminal, paniculate, 20-40 cm long, 16-30 cm broad, the pe- duncle 10-19 cm long, the peduncle and branches brown-puberulent. Flowers sessile, borne 2-5 mm apart; sepals lanceolate, 1.7— 3.2 mm long, puberulent or strigillose to gla- brous; corolla urceolate with spreading lobes, greenish white with darker green stripes, the lobes linear, 3-4.5 mm long, the tube 3.5- 5 mm long, puberulent outside; stamens 5, the anthers distinctly bilobed and pendent in the mouth of the corolla tube, 0.5-0.8 mm long; ovary globose, 0.5-0.9 mm long, the disc crateriform, the style 0.5-1.6 mm long, the stigma conical, puberulent. Fruits ovoid, white, 4.5-5.5 mm long, 3-4 mm broad, glabrous. Distribution. Tournefortia multiflora is known only from Panama, where it occurs in wet forest in the provinces of Veraguas and Colón from 400-900 m in elevation. Tournefortia multiflora is distinctive with its large leaves to 50 cm long and unique in the genus in its large, many-flowered inflo- rescence, unusual striped flowers with linear corolla lobes, pendent anthers, and puberu- lent stigma. Despite these distinctions, it does ave a gynoecium with a conical stigma and the two-lobed fruits that characterize the ge- nus. It also has pollen grains that are similar to the majority of other species of Tourne- fortia (Nowicke & Miller, unpubl.). Johnston (1954) noted that pollen morphology seems to be a valuable generic character in the Boraginaceae, and, although unique in some characters, T. multiflora clearly falls within the morphological and palynological bound- aries of the genus. Additional specimens examined. PANAMA. COLON: Santa Rita Ridge trail, beyond end of Santa Rita Ridge Road (Panamanian Highway (R20D)), 17- 35 km from Boyd- Roosevelt Highwa 25976 (MO); N of Santa Fe, 6.5 km fi cola Alto de Piedra, Mori & Kallunki 3067 (MO) Tournefortia ramonensis Standley, Publ. Field Mus. Nat. Hist., Bot. Ser. 18: 992 1938. TYPE: Costa Rica. Alajuela: P Angeles de San Ramón, Apr. 1928, Brenes 6118 (holotype, F 851176; iso- type, NY). Shrub to 3 m tall, the twigs densely pu- berulent. Leaves opposite; petioles 13-28 mm long, densely puberulent; leaf blades widely ovate to lance-ovate, 8.9-15 cm long, 3.5- 9 cm wide, the apex acuminate, the base acute and decurrent, the margin entire to unevenly dentate, the adaxial surface sparsely strigillose, the abaxial surface sparsely pu- bescent, lighter in color than above, the ter- tiary and quaternary veins clearly visible. In- florescence terminal or axillary, a once- or twice-branched cyme, the branches 3-7.5 mm long, the peduncle 6-9 cm long, brown pu- berulent. Flowers sessile, borne 2-4 mm apart, bisexual; sepals lanceolate, ca. 6 mm long, strigillose; corolla tubular with spreading lobes, greenish white, the lobes ovate, 2 mm long, the tube 9.5 mm long, puberulent outside; ovary ovoid, 1 mm long, the disc crateriform, the style 4 mm long, the stigma pyramidal. Fruits broadly ovoid, white, 4-5 mm long, mm broad, glabrous, the style often persisting. Distribution. Tournefortia ramonensis is known only from high elevations in cloud forests in Costa Rica and the provinces of Bocas del Toro and Chiriqui in Panama. Tournefortia ramonensis is distinctive in being one of only two Panamanian members of the genus with opposite leaves. The other, T. johnstonii, is a low shrub rather than a tree and has coarse velutinous twigs com- pared with the puberulent twigs of T. ramo- nensis. Additional specimens examined. PANAMA. BOCAS DEL TORO: Robalo Trail, northern slopes of Cerro Horqueta, 518 Annals of the Missouri Botanical Garden 6,000- 7,000 ft., Allen 4919 (GH, MO). CHIRIQUÍ: vi- cinity of Pros Chorro, 1,900 m, Woodson & Schery 609 (GH (2), M Tournefortia tacarcunensis A. Gentry & Nowicke, Ann. Missouri Bot. Gard. 64: 134. 1977. TYPE: Panama. Darien: Cer- ro Tacarcuna, west ridge, trail from sum- mit camp to waterfall east of camp, 1,550-1,700 m, A. H. Gentry & S. Mori 14114 (holotype, MO 2280271). Low subshrub to 0.5 m tall, the twigs gla- brous or nearly so. Leaves alternate; petioles 2-3 mm long, glabrous or nearly so; leaf blade elliptic to elliptic-ovate, 8.3-23 cm long, 3.2- 8 cm wide, the apex acuminate, the base cuneate, the margin entire to unevenly un- dulate, the adaxial surface glabrous, the abax- ial surface glabrous, much lighter than above, tertiary and quaternary veins obscure. Inflo- rescence terminal, a once- or twice-branched dichotomous cyme, the peduncle 2-6 cm long, brown-strigillose. Flowers borne 2-3 mm apart, bisexual; sepals lanceolate, 3.5-4.5 mm long, sparsely brown-puberulent; corolla tu- ular, white or greenish white, the lobes spreading, very widely ovate, rounded at the apex, 2-2. m long, the tube 9-9.5 mm long; stamens 5, the anthers ovoid to lanceo- loid, 1.3-1.5 mm long, sessile, inserted just below the mouth of the corolla tube; ovary globose to ovoid, 1-1.7 mm long, the disc scarcely evident, the style 6-7 mm long, the stigma conical. Fruits broadly ovoid, white, 4-6 mm long, 4-5.5 mm broad, glabrous. Distribution. Tournefortia tacarcunen- sis is known only from Darién in Panama and Valle in Colombia in cloud forests of the Ser- rania del Darién above 1,500 m in elevation. Tournefortia tacarcunensis is poorly understood, being known from only the type and a collection from adjacent Colombia, both from Cerro Tacarcuna. While it appears to be endemic in this region, this is one of the most poorly known areas of both Panama and olombia Tournefortia urceolata James S. Miller, sp. nov. TYPE: Panama. Bocas del To- ro: along continental divide from road branching N off main Fortuna-Chiriqui Grande Highway : near continental divide, adoc MO 3386968). Figure 8. utex ad 2 m altus, ramulis glabris vel fere glabris. Folia alterna, petiolis 8-32 mm longis; lamina lanceolata vel angusto-elliptica ad elliptica, (10.7-)13.5-31(-38) cm longa, (3.2-)4-14(- m lata, apice acuminata, basi attenuata, glabris vel fere glabra. Inflorescentia terminalis, pyramidalis cymis parvis praedita, 13-28 cm longa, pe dunculo (5-)9-21 cm longo. Flores sessiles 2e in pedun- m dispositis; viridi-flava, tubularis ad urceolata lobis patulis, lanceolatis, 3-4 mm longis; antheris 1-2 mm longis, sessilibus, infra faucem fere ad medium corollae tubi insertis; stylus 1.7- .9 mm longus, stigmate spinulo ad puberulo. Fructus late ovoidei, albi, 4-6 mm longi, 4-6 mm lati, glabri. Shrub to 2 m tall, the twigs glabrous to brown puberulent. Leaves alternate; petioles 8-32 mm long, glabrous to brown-puberu- lent, the blade lanceolate or narrowly elliptic to elliptic, (10.7-)13.5-31(-38) cm long, 3.2-)4-14(-17.5) cm wide, the apex acu- minate, the base attenuate, the margin entire, the adaxial surface glabrous, the abaxial sur- face essentially glabrous, strigillose along the major veins and papillose between them, the secondary veins 9-15, the tertiary and qua- ternary veins clearly evident. Inflorescence — terminal, pyramidal, a panicle of small cymes, 13-28 cm long, the peduncle (5-)9-21 cm long, brown strigillose to puberulent. Flowers sessile or on short petioles to 1 mm long, borne 1-3 mm apart; sepals 5, lanceolate, 3.5-5 mm long, sparsely to evenly strigillose; corolla green to greenish yellow, tubular to urceolate with spreading lobes, 5-merous, the lobes lanceolate, 3-4 mm long, the tube 7- 10 mm long, strigillose outside; stamens 5, the anthers lanceoloid, 1-2 mm long, sessile, inserted from just below the mouth to just above the middle of the corolla tube; ovary ovoid, 0.5-0.8 mm long, the disc scarcely evident to crateriform, the style 1.7-2.5 mm long, the stigma conical, spinulose to puber- ulent. Fruits broadly ovoid, white, 4-6 mm long, 4-6 mm broad, glabrous. Distribution. Tournefortia urceolata is Volume 75, Number 2 1988 Miller 519 Revision of Panamanian Boraginaceae FIGURE 8. known only from Panamá in Chiriqui, Colón, and San Blas, where it occurs in cloud forests at elevations of 400-2,300 m Tournefortia urceolata is distinct in its large glabrous leaves with relatively obscure venation and urceolate corollas. Vegetatively it somewhat resembles T. glabra but can be e Tournefortia urceolata. — 4. Flowering branc Fruit. A-C from Croat € Grayum 60301 —B. Flower.—C. Flower with corolla opened.— D. h. MO) , Bocas del Toro, Paes D from Croat 37301 (MO). distinguished by its less prominent tertiary veins, sepals longer than 3 mm, and corolla onger than Additional — examined. PANAMA. CHIRIQUÍ: Fortuna Dam area, on Kaolin hill, just N of reservoir, 1,100-1,400 m, D'Arcy et al. 15919 (MO); forested eh below divide ridge near Cerro Pate Macho, 6,200 ft., Hammel 5817 (MO); Palo Alto, 4.5 mi. NE of Bo- — Annals of the Missouri Botanical Garden quete, forest along western branch of headwaters of Rio Palo Alto, 6,300 ft., Hammel 7414 (MO); S slopes of Cerro Pate Macho, along Rio Palo Alto, 1,300-1,800 m, Knapp et al. 2077 (MO); ca. 0.5 km E of Cerro Pate Macho, headwaters of Rio Palo Alto, 1,800-2,100 m, Knapp et al. 2121 (MO); SE slopes and summit of Cerro Pate Macho, trail from Hio Palo Alto, 4 km NE of Bo- quete, 1,700-2,100 m, Sytsma et al. 4845 (MO). COLÓN: on Santa Rita Ridge Road, 17-35 km from Boyd-Roo- sevelt Highway, 400-800 m, Mori & Crosby 6413 (MO). SAN BLAS: Cerro Brewster, headwaters of Rio Cangandi, 2,300 ft., De Nevers et Hm 5500 (MO). Tournefortia volubilis L., Sp. Pl. 140. TYPE: without locality or collector (holotype, LINN (Savage Catalog num- ber 193.3), not seen; microfiche, MO). icd lees Kunth in Humb., Bonpl. & Kunth, i 1818. TYPE: Venezuela. Sucre: prope TNR bra B-WILLD, not seen; mi- MO). fiche, dinde idi d ur Kunth in Humb., Bonpl. & Kunth, n. Sp. 3: 79, t. 201. 1818. TYPE: Mexico. ires Acapulco, iudi Mir d 3863 (ho- lotype, P, not seen; micro d dr potosina en Gota U. S. Natl. Herb. : 1 PE: e San Luis Potosi: Tam masopo Canyon , C. G. Pringle 3518 (holotype, US 316801; o. CAS). Woody vine or occasionally a small shrub, the twigs densely puberulent. Leaves alter- nate; petioles 4—8(—12) mm long, densely pu- berulent; leaf blade lanceolate to lance-ovate, (2.7-)5—7(—10.4) cm long, (1-)2-3.5(-5) cm wide, the apex acuminate, the base obtuse to rounded, the margin entire, the adaxial sur- face strigillose, the abaxial surface densely puberulent to nearly tomentose. Inflorescence terminal or internodal, a loosely branched cyme, the peduncle 3-16 mm long, the fertile branches 3-5(-10) cm long. Flowers sessile, borne 1-5 mm apart; sepals linear-lanceolate, 1.2-1.8 mm long, densely strigillose; corolla dull yellow or white to pale green, the lobes linear-lanceolate, ca. 1.5 mm long, the tube 2-2.3 mm long, densely strigillose outside; stamens 5, the anthers ovoid, 0.5-0.7 mm long, connate apically, sessile, inserted near mouth of corolla tube; ovary ovoid, 0.6-0.8 mm long, the stigma conical, exceeding the calyx lobes, elongate. Fruits distinctly 4-lobed, 2-3 mm long, glabrous, white with dark brown to black spots at the apex. Distribution. Tournefortia volubilis is widely distributed in dry, disturbed areas from sea level to 600 m in elevation from Mexico throughout Central America to Panama, and is found in the West Indies. In Panama it is known only from the provinces of Los Santos and Panamá. Tournefortia volubilis is a member of the complex sect. Cyphocyema I. M. Johnston, a group centered in South America. It differs from the only other Central American species of the section, 7. maculata, in having pu- bescent leaves (discussion under 7. macula- ta). Additional specimens examined. ANAMA. LOS SANTOS: Monagre Beach, Dwyer 4182 (MO (2)). PANAMÁ: Coronado Beach, 6 mi. E of San Carlos, Croat 14261 MOT near Playa Rio Mar, 10-100 ft., Duke 11761 MO). = LITERATURE CITED AIRY SHAW, H. K. 73. A Dictionary of the Flowering Plants and Ferns, 8th edition. Cambridge Univ. Press, bridge. CANDOLLE, A. P. DE. 1845. Boraginaceae. In: Prodro- mus Systematis Naturalis Regni Vegetabilis 9: 466- 501 Croat, T. B. 1978. Flora of Barro aid Island. Stanford Univ. Press, Stanford, Califor CRoNQUIST, A. 1981. An Integrated uan ‘of Classi- fication of Flowering Plants. Columbia Univ. Press, New York. idis G. 1985. The phytogeographic relationships the Panamanian grasses -2 Bn and Natural History of Panama, W. G. D'Arcy M. D. Correa A. (editors). Missouri Botanical Gar- den FRIESEN, C. V 33. Les caractéres essentiels de la anamanian pla P. atural History of Panama, W. G. D'Arcy & M. D. Correa A. (editors). Miss n. . L. A revision of the genus Hackelia (Boraginaceae) i in North America, north of Mexico. Mem. New York Bot. Gard. 26: 121-227. Gibbs, P. E. & N. Taropa. 1983. Heterostyly in the Cordia alliodora—C. trichotoma complex in Brazil. Rev. Braz. Bot. 6: 1-10. Boraginaceae. /n: Flora of Gua- tamala, Pedana. Bot. 24(9): 111-167. Volume 75, Number 2 1988 Miller Revision of Panamanian Boraginaceae 521 HAMMEL, B. E. 1986. puna and phytogeo- the flora of La Selva 9: 14 9 95. he Families of Flowering Plants. arendon Press, Oxfor 19 JOHNSTON, I. 24. Studies į in = P 2. A supe a of the erican native and immigrant borages of the subfamily Boraginoideae. Contr. Gray erb. 70: 3-61. 1927. Studies in the Boraginaceae, 6. A re- vision of the South ru Boraginoideae. Contr. Gray Herb. 78: 3- —— 1928. Ae in the Boraginaceae, 7. The South American species of Heliotropium. Contr. Gray Herb. 81: 3-73. 1930. Studies in the Boraginaceae, 8. Ob- servations on the species of Cordia and iiy jain known from Brazil, eic Uruguay, and Arge tina. Contr. Gray Her : 3-89. 1935. Studies in the Boraginaceae, 10. The Why een al y aa South America. J. Ar- nold Arbor. -64. 940. pecus in the Boraginaceae, 15. Notes on some Mexican and Central American species of Cordia. J. Arnold Arbor. 21: 336-355. 949a. Studies in the Boraginaceae, 17. Cor- dia section Varronia in Mexico and Central America. J. Arnold Arbor. 30: 85-104. 1949b. Botany of San José Island. Sargentia 7: 1-306. . 1949c. Studies in the Boraginaceae, 18. Bo- raginaceae of the southern West Indies. J. Arnold Arbor. 30: 111-138. 1950. Studies in the Boraginaceae, 19. B) Cordia section Gerascanthus in Mexico and Central America. J. Arnold Arbor. 30: 179-187. ——. 1951. Studies in the Boraginaceae, 20. Re resentatives of three subfamilies in eastern Asia. J. Arnold Arbor. 32: 1-26, 99-122. ———. 1954. Studies in the Boraginaceae, 26. Fur- ther revaluation of the genera of the Lithospermae. J. Arnold Arbor. 35: 1-81. Karr, J. R. 1985. Birds of — fae ia ia and ecological dynamics. Pp. 77 n The Botany and aura History of Pan y "M. D. Cor s sms 1 . An Introduction to the Natural Sys- of Botany. London Mrz, C. 1890. Morphologische und anatomische Stu- dien über die Gruppe der Cordieae. Bot. Jahrb. Syst. 12: 526-588. Miers, J. 1869. On the yh aol carpical structure of the Ehretiaceae and Cordiaceae. Ann. Mag. Nat Hist. Ser. IV, 3: 383-388. . Systematics of the genus Cordia (Boraginaceae) i in Mexico and Central age Ph. D. Thesis. St. Louis University, St. Loui i . Boraginaceae. In: W. D. Een (editor), Flora de bip (in pres Nasu, D. m & N. P. Mores: 1981. In: Flora de Veracruz, Fasciculo 18. Novi, E; W. 1969. Boraginaceae. In: Flora of Pan- nn. Missouri Bot. Gard. 56: 33-69. 8 ILLER. Boraginaceae. In: Flora of Boraginaceae. Ceylon (in press). & J. E. Ripcway. 1973. Pollen studies in the genus Cordia (Boraginaceae). Amer. J. Bot. 60: 584- 91. es La J. BRYA, 1974. A palynological in- genus Tournefortia (Boraginaceae). EL E P deg 1036. T "P. A., H. G. BAK E ba W. FRANKIE. 1975. Reproductive ky a e Costa Rican Cordia species (Boragin dd p 7: 234-247. Raven, P. H. & D. I. AXELR 1974. Angiosperm biogeography and ids oda movements. Ann Missouri Bot. Gard. 61: 539-673. 975 ind of the flora and na of Latin America. Am . 63: 420-429. I Introduction of Jacquin's Carib- : acquin, Enumeratio Sys- < P nter Documen- 1933. Arnold Arbor. STEAD, J. W. 19 "4 Commonwealth Forestry Institute International Provenance Trials of Cordia alliodora (R. & P.) Oken. ind Commonwealth Forestry Conference. 1980 STEARN, W. T. 1971. Taxonomic and nomenclatural notes on goer nel plants. J. Arnold Arbor. 52: 614-6 THE GENUS XYRIS (XYRIDACEAE) IN VENEZUELA AND CONTIGUOUS NORTHERN SOUTH AMERICA! Robert Kral? ABSTRACT This work is a taxonomic treatment of those species of Xyris (Xyridaceae) now known to occur in Venezuela and contiguous northern South America, including parts of Brazilian Amazonia. A a peat discussion of. the the A ern: accepted sectio sections (placentation type) breaks down when species (14 newly described) , y of Xyris is given, particularly as to the construction of the inflorescence and flower. ns of Xyris are stated; it appears that the major distnguishing feature for these Guayanan material is studied. including 4 subspecies (one newly described) a r. Problems with e taxonomy is done for 87 d 7 varieties. Synonymy and full descriptions are presented, along with diagnostic keys. Nearly all taxa are uod with full plates. The first comprehensive treatment of Xy- ris of the Guayana Highland was authored y Dr. Bassett Maguire of the New York Botanical Garden and Dr. Lyman B. Smith of the Smithsonian Institution (Botany of the Guyana Highland V, Memoirs of the New York Botanical Garden 10: 8-72. 1963). That excellent work has provided a basis for this supplemental study, which includes new information on species already described and descriptions of several new Xyris. Also, the coverage is extended to include Andean species from western Venezuela and Colombia as well as those from the Guayanas to the east and the contiguous territories of Brazil. Useful sourceworks for this have been: J. M. Idrobo, "X yridaceas de Colombia,” Caldasia V1(29): 185-245. 1954; J. Lanjouw, Xyridaceae in Flora of Suriname I(l): 225-244. 1966; L. B. Smith & R. J. Downs, Xyridaceae in Flora Brasilica IX(11): 1-211 + index. 1968; L. B. Smith € R. J. Downs, “Las especies Pe- ruanas de la Familia Xyridaceae," Publica- ciones del Museo de Historia Natural “Ja- vier Prado," Serie D. Botanica (15)1-13, + figs. 1963. As was stated by Maguire & Smith (l.c.), Brazilia, to the south of the Amazon, and Guayana, to the north, comprise the two larg- est centers for Xyris in the world; each major area has large numbers of endemics, with new ones being discovered on any major expedi- tion into these regions. The authors continue: It is Poggi that Ayris 1 in eid America, with its pre at one time an essertall identical pun continuous area, and that in came divided, and that since ` then a parallel evo- 1 11 ! Many hundreds of specimens have been Pao in the E of the work and | acknowledge the . GH, K, L, assistance of curators and staffs Y, P, U, o kindly made loans and facilities available. I could not have done the work were it not oe Dr. buo. B. Sah, who provided me the opportunity ake the project in the first place, who reviewed desc riptions of most of the new species, and who is therefore coauthor of them; Dr. Otto Huber, ecologist and authority on the botany of tropical American savanna, who has provided specimens for and who has provi e study, and is ther eding as much ennance study, who led me to a g m deeply grateful. Special thanks are due to Dr. George Rogers, Editor, and to Ms. Wilson, Editorial puces d their patient, thorough, adi da by aes of my w work. Their constructive gratefully ackno refore wledged. ? Herbarium, Box 1705, ion B, Vanderbilt QURE Nashville, Tennessee 37235, U.S. A. ANN. Missouni Bor. GARD. 75: 522-722. 1988. Volume 75, Number 2 1988 Kral 523 Xyris My observations in no way conflict with the above. I can only add that, since the addition of the monotypic Aratitiyopea by Steyermark (Ann. Missouri Bot. Gard. 71: 298-300. 1984), there are now five genera known for the Xyridaceae, all occurring in, if not solely confined to, the Venezuelan Guayana, with the greatest diversity being in the high sandstone tepuis of the region. Xyris is by far the largest of the five. At the rate new species are now being found it will prob- ably well exceed 300 species worldwide. Prior workers have organized these into three sub- genera, namely Pomatoxyris Endl. with ca. 20 species, Australian, distinguished by axile placentation; Xyris (Fuxyris Endl.) with ca. 100 species, pantropic and North American, distinguished by marginal or parietal placen- tation; and Nematopus Seub., with probably more than 200 species, mainly South Amer- ican and distinguished by basal or free-central placentation. In dissecting ovularies of Gua- yanan xyrids preparatory to describing them, I discovered that many of these species, placed by prior workers in Vematopus, actually have axile placentation or are transitional from ax- ile, this evidenced by the presence of complete or partial septa in the ovary and fruit. Out of slightly more than 80 taxa studied, 33 show evidence of some septation; some show completely axile placentation. I suggest that sectionalization of Xyris based on mode of placentation perhaps does not reflect the real situation and that there appear to be no other solid characters by which to place species within sections or sub- genera consistently. Certainly the sections are arbitrary. Alternatives would be to put the more than 30 Guayanan species with partial or complete septation into Pomatoxyris, a perhaps more realistic alternative, even though it eliminates the geographic integrity of that section. Another would be to dispense with Pomatoxyris, since subgenus Xyris, even as represented in the southeastern United States, may show transition from axile to parietal placentation (X. elliottii Chapman, X. bald- winiana Schultes, X. brevifolia Michaux); therefore, the limits of section Xyris could be amplified to accommodate the situation. A third alternative, namely to eliminate the sec- tions altogether, would in my view be best, in that mode of placentation has been nearly the entire basis. In any event, it would appear that in regard to a phylogeny the Guayana Highland show the greatest diversity in xyrids, with placen- tation in all genera except Xyris wholly axile. It seems that distichous-leaved xyrids with axile placentation have evolved from polys- tichous-leaved ancestors with axile placenta- tion, at least if Abolbodaceae are to be con- sidered as part of Xyridaceae, a generally accepted classification based on the sound anatomical analyses of Carlquist and others. That being so, Xyris may have had its origin in the Guayana Highland. No further introduction is needed here save to relate what appears to be the basic mor- phology in Xyris. Species of Xyris are mostly high hydro- period, rosulate plants of acid boggy sites, their alternate, equitant, distichous, lineal leaves approximate and arising from con- tracted or elongated stems or from scaly rhi- zomes; some rise from bulbous axillary off- shoots; the roots are simple and fibrous (Fig. Leaves are highly variable in structure but are similar to those of /ris. The leaf base is open-sheathing, the sheath edges broad and thin, often scarious or hyaline, and entire or variously ciliate; the sheath base is frequently abruptly orbicular-dilated, this most obvious in species with bulbous bases. Above the clasp- ing sheath base, the sides narrow abruptly or gradually to the blade, there converging and merging with it as the ventral margins close. At this point there may be a simple transition to the inner margin of the blade (most species), or the transition may be abrupt, with a ligule of various proportion being produced (Fig. IIb, c). Often in ligulate species the inner edge of the leaf blade may show a strong sulcus or groove for some distance above this junc- tion (Fig. IIc). The leaf blade then shows a convergence of leaf sheath edges to form either a set of ventral edges (as in sulcate-bladed 524 Annals of the Missouri Botanical Garden Q E 2 C ( FIGURE I. Sketches of some common habits in Xyris.—a. Base sofi, leaves flattened, sheaths keeled. Base bulbous, sheaths fleshy-based, blades curvate, flattened. —c. Ba rhizomatous, leafy shoots frondlike. — nual, leaves spreading flabellately; scape sheaths longer than principal inue flexuous and twisted. —d. Plant caulescent and tous. —f. Ann species), or the ventral edges have joined into a single, compound-traced, ventral edge (Fig. IIa). Much Xyris taxonomy centers around the characters of the leaf sheath and blade, which tend to be more consistent than most, outside of flower and fruit features. For ex- —b. ase elongate- apar sheaths firm, blades Plant stout-rhizoma- ample, Xyris leaf blades may be terete or angulate or flattened; their tips may be ex- tremely varied, with everything from emar- ginate to conic-subulate, even setiform; their edges and borders are incrassate to thin and variously pubescent or totally smooth; the FIGUR o base, pom ulat Leaves, spikes, and upper scapes. — a. te, blade flattened. —b. Side and ventral views of leaf with sheath abruptly dilated > Side and ventral views of leaf with sheath gradually dilated base (a at is type), ligulate, blade flattened. — c. Sector of leaf at junction of ligule and blade, blade subterete, with Volume 75, Number 2 Kral 525 1988 Xyris ventral sulcus. —d. Sector of leaf blade, this subterete and bicostate. —e. Thin, involute, multicostate blade. — f. Sector of flattened, ciliate leaf blade. —g. Sector of flattened, scabrid leaf blade. — h. Sector of flattened leaf blade with incrassate edges. — i. Sector of terete leaf blade. —j. Acute, incrassate-edged leaf apex. —k. Acute, ciliate leaf apex. —l. Acuminate, incrassate-edged, incurved leaf apex. — m. Acuminate, spinulose-tipped leaf apex.— n. Asymmetrically spinulose-tipped, acute leaf apex.—o. Trigonous-acute leaf apex. — p. Leaf apex narrowly and bluntly conic, trichomiferous. —q. Ovoid spike with spirally imbricate bracts (commonest type with little bract gradation) ; scape subterete, bicostate. —r. Broadly ovoid spike, sterile bracts involucrate, scap ancipital. —s. Spike biflorous, decussate, scape flattened, incrassate-bordered, twisted. —t. Spike biflorous, de- cussate, lowest bracts exceeding spike, incurvate. —u. Spike ellipsoid, bracts decussate, 5-ranked, little graded. — v. Spike ovoid, bracts spirally imbricate, lowest pair of bracts distinctly longer than the fertile bracts. —w. Spike obovoid, sterile bracts many, grading gradually into the fertile bracts. 526 Annals of the Missouri Botanical Garden surfaces are variously pigmented, smooth to variously rugose, with raised or sunken or flush stomates, or even quite pubescent (Fig. Id—p). Most of such external characters have long been noted in the literature, particularly by G. O. A. Malme, A. Nilsson, Seubert, and others in the late 19th and early 20th centuries. ris plants have scapes arising in axillary fashion enfulded in closed-based, distally open and bladed scape sheaths. The scapes are also quite diverse, usually terete and solid basally and variously multicostate. Distally they vary the most, from terete and ecostate to multi- costate, variously flattened, even ancipital; they range from smooth to variously scabrid or ciliate, particularly on the costae (Fig. Hq- w). Usually the scape terminates in a single spike (most exceptions being either the Gua- yanan X. bicephala Gleason or anomalous examples). The spike is mostly conelike with either spirally or decussately imbricate bracts. Subsections have been developed on the char- acter of the bracts. One direction of evolution has been toward species with the lower bracts elongate, even leaflike (X. cipoensis, X. hys- trix, X. involucrata, X. uleana, etc.) with the fertile bracts much shorter and abruptly reduced. Another is toward sterile bracts few at the spike base, somewhat larger than the fertile bracts and transitional to them. A third trend has been for sterile bracts to be many, the lowermost very small, gradually passing into the larger fertile ones. The commonest condition is for there to be a few sterile bracts somewhat smaller than the spirally imbricate fertile ones. Significant also is a bract zone that may run the length of the midbract or be variously apical, the dorsal area. Doubt- lessly homologous to the leaf blade in Xyris, this is frequently photosynthetic and is mostly distinctly different from the usually chaffy bract matrix. As mentioned above, some bracts in Xyris are elongated and may have blades, and when this is so, such are mostly extensions of dorsal area tissue, or at least of midrib tissue (Fig. IVa-h). Bracts on yet other species may lack a dorsal area, and a few species may be versatile in this character. Some ex- amples of X. tenella Kunth have bracts with and others without dorsal areas on the same plant, or may have them on the sterile bracts but lack them on the fertile ones. Backs o bracts also vary considerably, from rounded- convex and ecostate (carinate) to strongly folded, navicular, or costate, while the usually thin edges vary from strongly bordered and variously ciliate, lacerate, or fimbriate to en- tire. In the axils of the fertile bracts the flowers are solitary, the axis of the indeterminate inflorescence being contracted and headlike or conelike, or quite elongate (as in X. steno- stachya Steyerm., the most extreme). These flowers are conspicuous at anthesis, usually but a few hours for a single bloom, a riven spike producing usually only one or tvo flow- ers simultaneously (in the species with con- tracted spikes several flowers may be open at once, e.g., X. involucrata Nees). A dissection of the flower bud just prior to its expansion is the best approach to under- standing the Xyris flower (Fig. IIIa-f). The usual Xyris bud is narrowly obovoid and plan- oconvex. There are three sepals. The lateral sepals are boatlike, their keels directed out- ward, their concave sides clasping the edges of the compressed corolla bud as well as the edges of the inner (dorsal) sepal, which forms a cap over the corolla with its two edges overlapping on the ventral side. The lateral sepals (Fig. IVl, m) are the bonanza for xyr- idologists in that they vary little from biotype to biotype in a species. In most cases they are free; in fewer instances they are connate; Flower and fruit structure. —a. Ve => ntral and cross- sectional view of sepals in bud, lateral sepals peta (claws fore pde ned) with adnate stamens; 4. staminodia, POMA foreshortened, Nomina flattened, bifurcate Volume 75, Number 2 Kral 527 1988 Xyris blades and terminal rufis of trichomes; 5. cross section of flower toward base as parts are oriented in bud, placentation parietal; 6. stylar apex, showing horseshoelike stigma pattern at about level of staminode tip. — d. At left petal blade, claw, adnate stamen (note 3 traces in claw); at right an upper sector of petal claw, oblique view showing branching of median claw trace: one branch into filament base, the other forward into mid-base of Lars blade. —e. St —€— showing yokelike dong bifurcated trace, and ipi in stamin odial nt ; away from axis. Middle rank: abov e— parietal pasa n in ovary; below—cross section of dehiscing fruit. Right rank: above—central placentation in ovary; below—cross section of dehiscing pm valves lacking septa. 528 Annals of the Missouri Botanical Garden very infrequently they may be free toward a spike base, increasingly connate toward that spike apex (e.g., the X. thysanolepis com- plex). The sides of a lateral sepal are mostly very thin; in some cases they are equal or approximately so, and the sepal is equilateral; in fewer instances they are inequilateral (Fig. IVi, j). In all cases the sepal has a midrib (costa or carinal keel) that usually conforms with the angle of the sepal fold; in some cases this keel is thick and firm, made up of many strands of cells oriented longitudinally and breaking outward. The crest of such keels may range from entire to variously scabrid or ciliate (Fig. IVI). In other cases the keel may be produced into a sheet of cells one layer thick, its border made up of simple or compound strands of cells, these forming a lacerate or fimbriate outline (Fig. IVm). The best stage for seeing these characters comes after mature fruit has just formed and the sepals are hardened to produce a typical bor- der. At anthesis the outer (dorsal) sepal ab- scisses, falling away calyptralike as the corolla expands (Fig. IIIa). In the bud, the anterior (inner) petal is enfolded by the right (observ- ing the bud from the adaxial side) edge of the left lateral petal and is enfolded by the left edge of the other lateral petal, the right edge of which overlaps the right side of the ventral (inner) petal (Fig. Ib). All petals are roughly equal in size and usually separate to the base. Each has a long claw and a broad, usually yellow (in Africa there are blue-flowered species) spreading blade, this of a distinctive outline and apical border. There are three functional stamens adnate along the length of the claw. What appear to be three traces run the length of the claw, but the median trace is compound, branching at the blade base, the adaxial branch departing into the divergent filament base and going up into the anther connective (Fig. IId), the dorsal branch supplying the middle of the petal blade. Xyris anthers are mostly tetrasporangiate and bilocular, variously separated by a flat- tened connective, usually bifid distally, and sagittate proximally; their dehiscence is lon- gitudinal, either lateral or extrorse. Alternat- ing with the petal claws and slightly inside them is a whorl of three staminodes in most Xyris flowers (in a few these are reduced to nubs of tissue at the floral base or are totally absent, e.g., X. savanensis Miq.). The stam- inode is usually made up of an elongate slen- der claw about as long as the petal claw, and a yokelike flattened blade wherein the single trace bifurcates, tailing out toward the spread- ing staminodial branch tip, where there are usually borders of slender, moniliform hairs. The character of the cells making up the beard hairs deserves more attention than it has gotten heretofore and may well be reliable to determine complexes of species (Fig. IIc- 4, e). The ovary is superior, the style tubular, branching midway into three subequal branches whose involute edges are pollen re- ceptive distally. Placentation varies; in most Nematopus it is either plainly basal with a brush of long funiculi or free-central with shorter, ascending funiculi. In other Xyris the placentation is plainly parietal or appears marginal. But in a significant number of Gua- yanan species it may be axile and the ovary distinctly trilocular, or it may be axile toward the base and parietal upward in the ovary, or the septa may pull away so that the axis is left with the ovules (Fig. Ig). The fruit is usually thin-valved, though there are some definite exceptions in which the walls are hard. Dehiscence is loculicidal along the dorsal side of the carpel. The seeds may be few or very numerous, diverse in shape and size, 0.3-5 mm long. The outer integument is usually raised into a variously longitudinally ribbed, sometimes also cross-lined surface which is specific in char- acter; distally it may form an empty ribbed beak or it may separate into a crown of narrow scales (e.g., X. teinosperma Idrobo & Smith). The contents of the seed are a starchy and proteinaceous, translucent or farinose endo- sperm and a small basal-lateral embryo. The type for the genus is Xyris indica L. This was based by Linnaeus on an east Indian element and a North American element. J. E. Smith (Rees’ Cycl. 39. 1818) designated Volume 75, Number 2 Kral 529 1988 Xyris / FIGURE IV. Bract and lateral sepal VAS — a. Ovate bract with ovate, subapical dorsal area. — b. Lanciform bract, keeled, with elliptic-linear dorsal area. —c. Lanciform bract, keeled, bladed, the blade comprising a cusp and dorsal area. —d. Ovate bract, mostly Pur area, ls convex bract medially-apically carinate. —e. Broadly ovate bract with narrow, subapical dorsal area. —f. Broadly obovate bract, border thin, rij jas dorsal area lacking.—g. Broadly elliptic ie border thin, iis, friable; dorsal area lacking. — h. Lanceolate bract, keeled; dorsal area lacking. —i. Dorsal and lateral views of lateral sepal, this oblong- elliptic, ipie slightly curvate, with narrow, thick, entire keel. —j. Dorsal and lateral views of lateral sepal, this oblanceolate, ip lateral, with narrow, thick, entire keel. x Connate lateral sepals. —l. Midsector of lateral sepal showin thickened, scabro-ciliate keel. — m. Midsector of lateral sepal showing thin (made up of a single layer of calli]. ciliate-fimbriate kee 530 Annals of the Missouri Botanical Garden the Indian element as X. indica while using the North American element for X. torta In the key below, the characters are based on principal leaves from healthy plants, spikes at seeding time, lateral sepals at seeding time, and on mature fruit and seed. Any attempt to determine sterile or depauperate material, or specimens with immature flowers, fruit, or seed is virtually impossible in a group of this SO rt. The key below applies primarily to the Xyris of Venezuela, Colombia, and the Guia- nas. Some taxa discovered in contiguous Bra- zil, if in habitat likely to be found in Venezuela and the Guianas, are likewise included. The key applies to healthy material with normally developed, seeding spikes. Principal, not juvenile, leaves are to be used. Overlaps in variation—particularly in trichome char- acters, lateral sepal and dorsal area color or accrescence, and plant pigmentation—often have necessitated the same species coming out in two or more leads. KEY TO XYRIS OF GUAYANA AND AMAZONIAN NORTHERN SOUTH AMERICA I. Placentation evidently marginal or parietal, the capsule valves on dehiscence retaining placentae and seeds Section Xyris iia II. Placentation free-central or basal or apparently axile, capsule valves on dehiscence with or without septa, ds but not retaining see other Xyris I. Section Xyris (conventional). la. Keel of lateral sepals at least in ee e or irregular. m long; dorsal a 2a. Seeds fully 1 m (us nose and/or e conspicuous; scapes terete inia) with cilia a ve midne ually) 3 or more stron ein occasionaly a ie sheath sometimes long- 3a. Principal foliage leaves longer all w ith tips lacking tufts of long trichom 3b. ica foliage leaves poorly dev ea e ex the low es with apical tufts 2b. See ds 04 4-0.8(- 0.9) m bicostate; leaf sheath never ciliat e ad spike bracts when dry inrolling gr = X. fa allax eeded by scape “siw spike bracts when e of ends trichom 6. X. brackyfolia m long; dad area not venose or costate as in above; scapes seldom more than . Plant base strongly tinged with red or purple; scapes ecostate distally or with costae sont seeds fus iform or narrowly ellipsoid, mostly opaque, dark-ribbed, farinous, 0.7- O costae, these usu nslucent, 0.4-0 Keel E lateral oak either ciliate o 5a lb. h, stramineous, dull brown or tan; scapes always with 1-2 (ra = a ually papillose or scaberulous; seeds ovoid or broadly ellipsoid, amber, mm Fee plants annual 3. X. Jupicat ral sepals with keel n dp daii dull brown, at maturity strongly spreading, the dorsal are . navicularis eeled, paler but indist 5b. TAR sepals wit strongly keele th keel dec entire; bracts pale greenish tan, lustrous, the indistinct oras] are 5; X. a D: II. Other Xyris. la. Margins of leaf sheaths entire 2a. Plants mostly densely cespiose, evidently perennial; leaf sheaths hard, lustrous, castaneous to reddish b t to leaf blades ed, for rown, usually in strong contr 3a. Leaf blades not flattened or s color and texture. both in most of their length either terete or very thickened and blades deeply ribbed and sulcate; ligules prominent, abruptly narrowed to l 4a. Dorsal area distinguishable 7. X. neblinae en cA Dor n, », pale e brown, or reddish bro 6a. Mature spikes sas over 1.5 cm long, e inde or - distinctly broadest above middle; leaf blades at least 1 mm thic . Mature spikes aud pus 1.5 cm long, ovoid to cylindrical or subglobose: leaf blades less than 1 m Ta. Bract edges ts aue villose; plants lacking stout, scaly rhizomes . Juncifolia X. lanulobractea 7b. Bract edges not densely pale-villose; plants with short, stout, ascending, sca 10. rhizomes y X. terrestris Kral Volume 75, Number 2 1988 Xyris 4b. Dorsal area imperceptible. 8a. Leaf blades and upper sheaths coarsely ribbed and sulcate; sheath bases brown (bases of blades and apices of sheaths mostly rose-purple) 11. X. scabridula 8b. Leaf blades not deeply ribbed and sulcate; sheath bases (usually tips also) nearly blac X. atriceps 3b. Leaf blades flattened, for most of their length 2-edged, the edges either thickened or sharp; ligules evident or not. 9a. Dorsal area e 10a. ge Henn flattened, at least 2-2.5 mm broad distally, ancipital. lla. Several bracts leaflike, forming an involucre 13. X. involucrata b. No bracts leafli a. Scapes bispicate; seeds n ^ mm long 15. X. bicephala capes unispicate; seeds over 2 mm long _ 16. X. teinosperma Se eeds m long 10b. Scapes distally round or if strongly Pee! distally not as wide as 2 mm there, n ancipital. 1 3a. Lowermost sterile ia distinctly shorter than the fertile ones and grading gradually into the 14a . Leaf blade us cartilaginous-thickened, entire or ciliolate; bract border pale villosulous (low-elevation savanna o E 14b. Leaf blade edge not cartilaginous-bordered, poor bract borders not pale villosulous, sometimes sparsely short-brown ciliolate ................. . X. contracta 13b. Lowermost sterile bracts as long as or longer than the fert le o 9a. Lateral sepals connate, crested with a red villous keel apex; plants glaucous; orsal area of lowest 1-2 bracts often excurrent as a gre X. seubertii 15b. Lateral sepals free, not crested as above; plants not glaucous; dorsal areas bladeless 16a. Dorsal area large, triangular, punctate, occupying most of bract above middle; low-elevation savanna or granite outcrops, T. F. Amazonas, Venezuela. 17a. Bract borders reddish, long-ciliate; keel of lateral ew and leaf blade edges ciliate; seeds 1.3-1.4 mm long ... X. huberi l7b. Bract borders not reddish, eciliate; keel of lateral iama . scabrid and leaf blade edges scabrociliate only tow 21. -~J ard base; see X. graniticola 16b. Dorsal area narrower, less conspicuous, not evidently pau high savanna of tepuis 22. 5 mm lon X. frondosa 9b. Dorsal area not evident. ngest leaves rarely with sheaths over 14 as long as blades, mostly 1⁄4 as long or ile er. 19a. Leaf blades under 3 mm 20a. a lateral Bat toward apex red villous-ciliate; edges of young bracts apex red villos 23. X. chimantae 20b. Keel ot cif Sepals ZI or lacerate but not as above; bract tips entire or brown cilio 21a. i of leaf blades strongly callused, UM blunt; leaf blades coarsely few ; high tepuis of Amazona 2a. Foliage smooth; leaf she sheath edge ú ath wt tan to stramineous, lustrous; ming a pne broad a apically; keel of lateral sepals suben .24. X. s scm 22b. Foliage harsh, dba eaf blades al base; leaf shea apex and blade bases strongly tinged with red or rose; shea caida uminate, meine “spinulose; leaf blades not coarsely ribbed; high elevations. 23a. Flowers/spike 5-14; lateral sepals inequilateral, the keel long- ciliate above middle 25. X. columbiana 23b. Flowers/spike 4 or fewer; lateral sepals arii the keel either not evident or entire, papillate, or distantly scabrociliate. 24a. Lateral sepals ca. 5 mm long; bracts pa to orbicular, ecarinate, nearly black; seeds ca. 1 mm lon X subulata ee 24b. Lateral sepals ca. 7-8 mm acia bracts narrow inate, brown; seeds ca. 2 mm long ... 27. X. au 19b. Leaf blades mostly over 3 mm w 532 Annals of the Missouri Botanical Garden 25a. Scapes ancipital (flattened and sharp plan ah mostly 2.5 mm or more wide (variously scabrid, ciliate, or 26a. Spikes obovoid or hemi sphe ric, ca. as mne e as long or broader; scape edges scabrid; bracts dark lustrous brown (Duida tepui and Pw highlands) d tatei 26b. o cylindric, distinctly lo dg to pale ascending-ciliate; bracts dull, dark dinde: black NER i > Ss 25b. 20 not sharp-edged distally, mostly less than 2.5 m Le af blades with a rusty border, ciliate (often with reddish vein 30. vals . culmenicola 27b. Leaf blades without a rusty border, eciliate ......................... 31. X. lugubris 18b. Mind leaves commonly with sheaths over !4 as long as blades, frequently % as long or lon 28a. Basal bracts forming a leafy ipea iil than the spike ............ 14. X. pallidula 28b. Basal bracts not forming a leafy in 29a. Bra cts toward apex with a scarious, i pale or reddish border; bases E leaf sheaths o orbicular-dilated. a. a carious bract borders pale to deep red or red-brown; seeds ca. 1 m long . 32. X. th steed oe ee 30b. Searious bract borders almost always pale, off-white; seeds 0.5-0.7 m long. 31a. Apices of most bracts acute and distinctly folded; stem bases cloaked with very numerous, closely overlapping leaf sheath aas dioses a lustrous brown; endemic to summit elevations, Roraima and adjacent tepuis 33. X. concinna 31b. ju. p most bracts broader, often broadly rounded; stem “saa usually not as elongate, not as above; widely distributed medium to high elevations .......................... 34. X. hymenachne Bracts not bordered as above; bases of leaf sheaths not abruptly orbicular- dilated. 32a. Scapes ancipital and strongly ciliate distally. 33a. Scapes bispic iu 15. X. bicephala 33b. Scapes unispic 34a. Hairs of leaf and scape edges is forming a dense order; apices o leaves erect; seeds ca. 1.5 mm long. 5a. Main stems rebranching, n branches elongate, forming frondlike plates of leaves; tips of leaves arrowly acute; spikes ellipsoid or obovoid ............ 38. X. ptariana (extreme) 35b. Main stems not rebranching; tips of leaves round- ed; spikes hemispheric or subglobose ....................... 35. X. decussata 34b. Hairs of leaf and sape po srl white or blonde; apices of leaves various; s va 36a. Seeds at least 2 m ipei ake hemispheric or ubglobose; bracts very lustrous with subcucullate tips; tips of most leaves incurved-blunt .................... 36. X. albescens 36b. Seeds ca. 1.5 mm long; spikes obovoid or oblong; bracts dull, sooty brown, with flat tips; dps x most E erect and acute „u X. fuliginea 32b. PN ancipital o never ciliate, at most in ds along stems EN "frondlike plates of leaves (like giant Fissidens). 372. "Sp ikes 1.5 cm long or shorter, ovoid or broadly apie d; lateral pal under 1 cm long, strongly curved, blun 38a. Ligular apex often excurved; scapes s Aa flattened distally, mostly 2.5 mm broad or wider; leaf blades commonly over 4 mm wide U. . ptariana 38b. B apex commonly erect or ascending; sc apes nar- rower, usually 2 m paee or narrower distally; lea des usually under 4 mm wide ............ . X. witsenioides 37b. Spikes mostly 2-2.5 cm long, ellipsoid. cylindric or narrowly obovoid; lateral sepals over 1 cm long, narrow z acute „n . X. xiphophylla Volume 75, Number 2 Kral 533 1988 Xyris 2b. Plants ipud IM perennials or short-lived perennials or annuals of low- or moderate-elevation savanna, mostly low of habit and slender-scaped, and with small or narrow spikes; leaf sheaths mostly e. ur capes ancipital, edges at least in combination broader than the scape body; dorsal area and midrib of at least one is the lowest sterile bracts excurrent as a strong cusp or blade (body of lower bracts with a stro ). 40a. Basal bracts foliacecus, slightly to much longer than the fertile bracts, sharply alate- pn spreading or ascending and often forming an involucre; sepals acute . 4l. X. spruceana 40b. Basal bracts tending to be incurved, cucullate, the lowest uva to much longer o. fs fertile bracts, thus spikes not as noticeably involucrate and spreading-foliaceous; sepals se 42. X. uleana (complex) . Scapes flattened or terete, but not winged as above; lowest bracts shorter in relation to fertile racts. 4la. Florets 4 or fewer 42a. Lateral sepals connate. Fertile florets 1 per s 44a. Leaf blades with UM cartilaginous-thickened; scapes boue — . Xe = 3 c c with costae making edges meraldae 44b. iud blades filiform, the edges not noticeably pu dens d; capes filiform-terete, ecostate 45. X. subuniflora Fe ji florets more than 1 per spike 46. X. connosepala 43b. 42b. Lateral sepals 45a. Leaf blades strongly flattened, with narrow, pale, incrassate borders |... Te X. pco 45b. Leaf blades terete or somewhat flattened, without pale, incrassate border 46 eaf sheath apically with a broad ligule, this narrowing abruptly ts a terete blade; spikes aae Protierous aquatic or emergent plants with foliage often maroon an 48. X. spathacea 46b. Leaf sheath with oo Sea more «Pad to a more flattened, often strongly ribbed blade; spikes not proliferou 7a. Mature scapes equalled or exceeded by Es leaves; bracts at maturity strongly excurved, thus spikes broadly turbinate .............. X. oe 47b. Mature scapes definitely longer than leaves; bracts and spikes no as a 48a. Seeds ca. 1 mm long, including an apical pale coma of short squamellae; scapes and leaf blades rugose or rugulose . X. mima 48b. Seeds 0.4-0.7 mm long, —— x squamellae; scapes and leaf blades smooth o 492. Lateral sepals narrow, p entire; bract tips acute, with narrow dorsal areas „u 90. X. toronoana 49b. Lateral nek broad, ciliate or ciliolate; bract tips broader, with broader dorsal areas (see no. 47 4lb. Florets 4-ma 50a. Leaf blades filiform and flaccid; sheath bases broadly scarious-bordered, the scarious borders distally producing a strong ligule much broader than the leaf blade base; rhizomatous soft aquatics or marsh emergents 5la. E 2: blade ‘aie scapes arising from long, stout, scaly rhizomes; seeds c 51. m lon : E 51b. us blade "immu strongly nerved; scapes arising from slender rhizomes; seeds ca. 1 mm long 52. X. apureana 90b. Leaf blades wider and/or ee sheath at apex not as above; erhizomatous plants of drier sites. 52a. Spike elongate, more than 4 times as long as wide. 53a. Bracts distichous, spikes flattened 53. X. stenostachya 53b. Bracts polystichous, spikes terete. 54a. Leaf blades at least toward base with thickened, lustrous cartila- ginous margins; leaf sheath keels distally with cartilaginous dark costa. 55a. Lara a acute, fusiform or ellipsoid; seeds 1.2-1 54. X. stenocephala 55b. Spikes ma the mature ones ipid a 0.6-0.7 m lon e X. sin hoa ya 534 Annals of the Missouri Botanical Garden 54b. Leaf blades lacking thickened, lustrous, cartilaginous margins; leaf keels lacking cartilaginous dark costa. 56a. Spike lanceoloid to fusiform, acuminate; fertile bracts at apex with lanceolate dorsal area forming a sharp subapical keel; leaf blades terete most of their i above base .... X. sni a 56b. Spike ovoid to ellipsoid, lanceoloid, or Mur blunt o acute; fertile bracts with broad dorsal areas not forming a sharp subapical keel; leaf blades usually terete only at or toward tips (f at ^ NOS 57. X. paraensis (complex) 52b. aS » a broader, shorter outline. 57 eral sepals subequilateral or den prins al. capes sca abrid, t ongly bicostate distally; seeds apically with coma of mh narrow scales... 58. X. mima 58b. Scapes smooth or at most papillate, usually ecostate or at most striate; seeds without coma. 59a. Spikes broadest at or above middle; leaf blades with strong re Dodd. lateral sepals strongly curved, the keel evenly 9. X. rubrolimbata 59b. ard ovoid, acute; leaf blades lacking strong red n D se pals not strongly curved, the narrow keel e subentire . X. pane 57b. Lateral ri either connate or very inequilateral. 60a. The lateral sepals connate; fertile bracts pectinate- -bordered ....... X. pectinata 60b. 2 lateral sepals free; fertile bracts not pectinate- bordered. aminodia lacking beard; foliage prevalently (one var. ex- cepted) papillose or rugulose, the leaf blades strongly nerved; dorsal areas of fertile bracts mostly narrowly elliptic or linear; seed tips truncate, with a central low apiculus ......... X. savanensis 61b. Staminodia bearded; cd prevalently smooth the leaf blades either less strongly nerved or smooth; dorsal areas of fertile bracts broader or more raised; ot tips not trun- cate. 62a. Backs of bracts bois ges d bract edges at least apically villous or pilose-ciliate 63a. Lowest iur “of bracts incurved- — at tips not much if at all exceeding spike ........................ 42. X. pose ae 63b. Lowest pair of bracts with spreading, trigonou subulate tips many times longer than spike ..... ei . calderonii 62b. Backs and edges of bracts smooth and e 64a. Leaf blades with cartilaginous- thickened bor- ders; seeds 1.2-1.7 mm long „uuu X. stenocephala 64b. Leaf blades without cartilaginous thickened borders; seeds less than m lon X paraensis (complex) lb. Margins of leaf sheaths at some or all levels je aea ciliate with various sorts of hairs. t all spike brac 65a. Dorsal area eviden on some or eaves at junction of sheath and blade ia flattened. 67a. Scapes either strongly flattened distally or with 2 costae making 2 strong edge 68a. Leaf blade edges and/or scape Dies usually with long, straight cilia, Be smooth, if so then thickened. 69a. Spikes broad; dorsal area broadly lins sena occupying most of upper par act; scapes mostly 2 mm wide or more „u 63. X. Ji NDS eaf blade and scape edges scabrociliate with shorter, stiffer hairs .. 65. X. bicostata 67b. Scapes distally terete or at least thickened in cross section, 0-many-costate, costae in bicostate types not ciliate or scabrociliate, not making 2 strong edges. Oa. Scapes smooth, without strong costae or merely fluted la. Flowers usually more than 4. Bracts, or some of them, with long hairs on margins or backs or on oth. Volume 75, Number 2 Kral 535 1988 Xyris 73a. Bracts rounded, edges villous-ciliate 73b. Bracts acute, some bract backs with white t . X. arachnoidea 72b. Bracts heey pubescent margins or backs. 7 of bracts (at le = = zn areas) pale brown or pale red. rni thin, lacer 75a. Leaf sheath bases eee dilated; plant base thus bulbous. 76a. ins bracts apically carinate, acute; leaf blades arsely few-nerved „u 69. X. araracuare 76b. Fertile (pee more rounded, less carinate at apex leaf blades more flattened, less coarsely nerve X. lacerata 75b. Leaf sheath bases not abruptly siio plant base not bulbous (spike oblate, often proliferous) .............. X. oblata 74b. Matrix of bracts dark brown, spikes dark; cae —À aea deep brown or olivaceous or reddish (leaf tip excentrically spi- nulose) . setigera To 71b. Flowers almost always 4 or fewer (spikes mostly narrow; dorsal area narrow often streaklike, absent on some bracts). 77a. Sheath cilia long and spreading but firm; seeds ovoid, ca. 0.5 mm long; leaf blades linear 72. X. nea (complex) 77b. Sheath cilia arachnoid; seeds ellipsoid or cylindric, ca. 1 mm long; leaf blades filiform 73. X. byssacea 70b. ao with strong costae a. Scapes sharply 3-or-more-costate; plant base sub-bulbous; spike broadly ovoid or ellipsoid 68. X. malmeana 78b. Scapes with costae fewer if multicostate; papillate-scabrid plant base not bulbous (the plants annual and low); spike ellipsoid to lanceoloid ...................... 72. X. tenella (complex) 66b. Leaves at junction of sheath and = iml not flattened, at most elliptic in cross section, mostly terete or oval, often wit - a tral sulcu pikes few-flowered and n epis rowly ins to linear-lanceoloid). Oa. One pair of lower (sterile) bracts jdn than the rest and connivent over spike top X. cryptantha 80b. uid bracts not as above, the lowest pair slightly to considerably shorter than the 75. X. oxylepis spi 79b. a dera many-flowered and broader. ract tips m or bluntly acute. 82a. Outer bracts densely hirsute-tomentose or hirsute-ciliate with white hairs; one tips straight, erect or ascending, bract matrix dark-castaneous .............. 76. X. wurdackii 82b. Outer bracts not hirsute-tomentose or male ciliate, mostly — bract tips becoming excurved, bract matrix bro . X. frequens 81b. Bract tips broadly angled to rounded. Dorsal area broad, comprising most of the bract above middle; leaves strongl angulately ribbed; mostly plants of low-altitude savanna ...... 78. X. subglabrata 83b. Dorsal area narrower; leaves not strongly ribbed; mostly plants of medium- to riage altitude tepuis Scapes smooth, ‘usually lustrous, at most punctate 5a. Bracts dark Lodi or dark red-brown, dorsal areas pale and inconspicuou . X. setigera Lig ns x) 85b. Bracts pale b with darker dorsal areas ........... 79. X. lithophila Scapes papillate or tuberculate, usually rugulose or rugose, at a above middle 80. X. carinata 84b. c 65b. Dorsal area not evident. 86 vidis Á— a= waq at junction of sheath and blade, with the ventral edges about as narr s the 87a. poses terete NE ecostate, smooth and also punctate; leaf apex usually excentrically spinulose-tipped. 88a. Leaves 0.5-3 mm wide; lateral sepals 5-6 mm long; seeds e E 0.6 mm X. setigera (complex) 88b. Leaves 2-3 mm wide; lateral sepals 6.5-7 mm long; seeds b $- 1 mm long - 82 x riparia 87b. Scapes terete or somewhat flattened distally, there costate, usually tuberculate-scabrid, ciliate, PE: and/or rugulose at least on the costae; leaf apex not excentrically spinulose-subulat 536 Annals of the Missouri Botanical Garden 89a. from main bra 90a. Bracts with scarious and lacerate borders different abruptly in texture and color Borders of b red or red-bro 91a. Spikes ellipsoid to broad- cylindric; foliage above dilated base yellow- ally green, scabrid-papillose-rugulose; scapes ancipital dist . X. roraimae 91b. Spikes broadly obovoid or turbinate to subglobose; foliage except for leaf edges and costae not yellow-green, e ; scapes not an 90b. iig of bracts A l . X. ia (aa) pale. Apicos of bracts, particularly basal and inner ones, acute, cus spikes . X. concinna 92b. Apis of bracts poa obtuse, not folded; spikes aay ‘ellipsoid, ovoid, or su contrasting in color 9 . hymenachne h glo . Bracts very pd usually ie die: brown, the borders entire aní not muc 3a. Surfaces of leaves and scapes completely rugulose-papillose-tuberculate. 94a. Lateral sepals all free; capsule valves with strong septa 4. X. pila 94b. Lateral sepals connate; sl valves lacking septa ....... 85 93b. Surfaces of leaves and scapes nsolida oth or at most papillate only toward “leaf sheath base, but edges of leaf blades and of ancipital scape white-ciliolate; 86. X. lateral sepals free, capsule valves with septa ............................ kukenaniana 86b. Leaves definitely thickened at junction of sheath and blade, not 2-edged. Tips of leaf blades blunt; scapes rugose-tuber culate or rugulose-papillate ___. 87. X. delicatula 95b. Tips of leaf blades subulate-spinulose; scapes totally smooth, or smooth and puncticulate 81. X. setigera (complex) 1. Xyris fallax Malme, Bih. Svensk. Vet.- Akad. Handl. 22, Afd. 3, no. 2: 12, pl. 1, f. 5. 1896. TYPE: Brazil. Mato Grosso: Sta. Anna da Chapada, G. Malme 1432 (lectotype, S). Figure 1. X. Lanjouw, Rec. Trav. Bot. Neerl. 34: , f. 5. 1937. TYPE: Suriname: in savannis hu- ni prope Zanderij I, Suriname, Pulle 39 (holo- U). type, X. erythema a den & Lyman B. Smith, Mem. New York . Gard. 10: 12, fig. 1A-F. 1963. TYPE: Tum ome in wet places, scrub and low forest (8-10 m) on shoulder of east flank above Thompson Camp, 1,418-1,525 m, 10 Aug. 1960, Upper Mazaruni River Basin, Mt. Ayanganna, Brit- ish Guiana, S. S. & C. L. Tillett & R. Boyan 45074 (holotype, N; isotype, US). Low and slender to robust, solitary or ces- pitose, glabrous or papillose-rugulose peren- nial to 1 m high, the stems contracted, some- times perennating by stout (-4 mm thick horizontal rhizome. Leaves erect to spreading flabellately, (5-)10-30(-40) cm long, often suffused with maroon pigment; sheaths 1⁄2 to under !Á as long as blades, soft, pale lustrous brown, the dilated base entire, multicostate, gradually narrowing and carinate upward, the carina often papillose or ciliate-scabrid, the Y margins frequently sparsely to densely spreading-pilose-ciliate, gradually narrowing to blade or there producing a broadly trian- gular, incurved ligule to 2 mm long; blades ensiform-linear, 1-5(-7) mm wide, much flat- tened, sometimes slightly twisted, the apex narrowly incurved-acute, the margins thin or (usually) lustrous-incrassate, often papillate or minutely scabrociliate, more often smooth. Scape sheaths shorter than main foliage leaves, loosely tubular, multicostate, deep reddish brown proximally, distally opening and broad- ening, apically with short, erect blades. Scapes straight or somewhat flexuous, twisted, terete and multicostate distally, 1-2 mm thick, the costae smooth or papillate. Spikes lance-cy- lindric to broadly ellipsoid or ovoid, mostly 1-2 cm long, acute, the base short-attenuate, multiflorous; sterile bracts several, the lowest much smaller than the fertile bracts, narrowly triangular-ovate, carinate, grading gradually upward into the fertile bracts, these mostly broadly ovate or obovate, 5-7 mm long, sub- entire, apically narrowly rounded, the back with distinct elliptic dorsal areas, convex and ecarinate, the dorsal area bisected by a nar- row, low but distinct midnerve. Lateral sepals Volume 75, Number 2 1988 537 " PEA — oa N Z = : == UTI —a = ri = = LIII] 5 > a E = TIL g T š K FIGURE 1. Xyris fallax (Davidse et al. 1785) .—a. Habit sketch. — b. d. Leaf base-blade junction. —e. Leaf base. —f. Spi enlarged part A diem ha placentation. — Leaf tip h. Petal blade, sta minode, air.—j. Stylar apex. Wara Capsule with one valve removed, sind n two ene of / .—c. Sector off (e joe — ke.—g. Lateral sepal. — 538 Annals of the Missouri Botanical Garden free, subequilateral, linear-oblanceolate, 4.5- 5 mm long (often much reduced in length even in fruit), acute, red-brown, the firm broad keel lacerociliate from ca. middle to apex, sometimes villous-ciliate at apex. Petal blades unfolding in morning, broadly obovate, 4-5 mm long, yellow, the broadly rounded apex lacerodentate. Staminodia bibrachiate, the broad branches long-penicillate toward tips. Anthers oblong, ca. 1.5 mm long, deeply bifid and sagittate, on filaments ca. 1 mm long. Capsule ellipsoid or narrowly obovoid, 3-5 mm long, the placentae parietal and extending 14-23 up the ovulary. Seeds numerous, l- 1.5 mm long, asymmetrically cylindrical or fusiform, deep red-brown and translucent, longitudinally distinctly and closely multi- ribbed, the ribs often crossed or overlain by an occasional broader, deeper-brown rib. Distribution. Trinidad South to Mato Grosso (Norte), westward into the Andean foothills; Africa. This xyrid shows weedy tendencies, is often abundant in recently disturbed wetlands, if acid, and along fluctuating shorelines and banks. While commonest in low-elevation sa- vanna, it may be found in wetlands at ele- vations of nearly 1,500 meters. In ciliation of sheath it varies considerably from totally entire to densely long-brown-ciliate. 2. Xyris laxifolia C. Martius, Fl. Bras. 24(2): 58. 1841. TYPE: Brazil. Without definite locality: Mart. Herb. no. 540 (holotype, M). Figure 2. ? X. macrocephala M. Vahl, Enum. Pl. 2: 204. 1805. : French Guiana, **e Cayenne Vahl" í : this a mixture of X. ambigua Beyr. and X jupicoi | Rich.). X. macrocephala var. major (C. ea Nilsson, Kgl. Sv. Vet. Akad. € 24(14): 30. 1892. X. caroliniana Walt. var. major (C. Mart Idrobo & Smith, Caldasia 6(29): 199, fig. 4. X. jupicai Rich. var. major (C. Martius) ad & Downs in Reitz, Fl. Ill. Catarinensis, Pt. 1, fasc. xii: 9. 1965. Robust, solitary or cespitose perennial 0.5— 1.5 m tall, the bases usually suffused with red or purple, all surfaces smooth. Principal leaves erect or spreading flabellately, 4-7 dm long, the sheaths over 2 as long to longer than blades, at very base broad with scarious entire margins, red to purple or charcoal, tapering gradually, keeled into junction with blades, these broadly linear, mostly 1-2 cm wide, strongly flattened, straight, the apex abruptly incurved-acute or erect-acute, the margins thin and hyaline or slightly incrassate, the surfaces deep lustrous green. Scape sheaths shorter than leaves, proximally loosely tu- bular, multiribbed, deeply tinted with red, purple, or lustrous brown, distally opening and keeled, producing a short, flat, green blade. Scapes straight, stiffly linear, distally terete to oval or elliptic in cross section, sometimes 2-edged but usually ecostate, smooth, green. Spikes ovoid to cylindrical, 3-3.5 cm long, blunt to acute, green-brown, of many tightly and spirally imbricate firm bracts, the sterile ones much smaller than and grading into the fertile ones, keeled, all with distinct and usu- ally greenish dorsal areas; fertile bracts ob- ovate to ovate or suborbicular, convex-backed and ecarinate, 7-10 mm long, apically nar- rowly rounded, entire, the matrix deep to pale reddish brown, lustrous, in sharp contrast to the paler and dull dorsal areas. Lateral sepals free, oblong-curvate, 5-6.5 mm long, acute, the pale brown, thin sides subequilateral, the dark reddish brown keel wide but thin, lac- erate or lacerofimbriate from ca. middle to tip. Petal blades broadly obovate, ca. 5 mm long, yellow, the broadly rounded apex erose, the base broadly cuneate. Staminodia bibra- chiate, the broad, flat branches apically re- branched and long-penicillate, the cells con- gested with dark material, the terminal few often double. Anthers ca. 2.5 mm long, lance- oblong, shallowly bifid, deeply sagittate, on filaments 0.5-0.6 mm long. Capsule plano- convex, broadly to narrowly obovoid, 5.5- 6.5 mm long, often longer than the sepals; placentation parietal. Seeds ellipsoid-fusiform, 0.7-0.9 mm long, slightly to conspicuously farinose or translucent, with 6-8 conspicu- ous, minutely pebbled, dark, longitudinal ribs and many slightly less prominent cross-lines. Distribution. Southern Mexico south- ward through Central America and at the Volume 75, Number 2 Kral 539 1988 | | | i ii lh d ih FIGURE 2. Xyris laxifolia Tm & Gonzales 15946).—a. Habit sketch.—b. Leaf tip.—c. Leaf blade- sheath junction . Leaf base.—e. Spike.—f. Fertile bract.—g. Lateral sepal.—h. Petal blade, stamen.—i. Staminode.— j. Sp iylar apex. a? Seed. 540 Annals of the Missouri Botanical Garden lower elevations south into Argentina. Coastal Plain of the southeastern United States (var. iridifolia). The specimen sent from C in the type folder representing X. macrocephala Vahl and bearing Vahl's identification as X. mac- rocephala (Vahl script) should bear the oldest name for this species. However, the sheet has two elements on it: one is a spike of X. am- bigua Beyr., a Mesoamerican, Caribbean, and North American species but definitely not from the type area (Cayenne, French Guiana); the other is X. jupicai L. C. Richard, a previously named species. This situation appears to be best handled by selecting the earliest available indisputably typified name. Material at Flor- ence of the type collection of X. macroceph- ala (which I have seen only as o appears likewise to be mixed. The main dif- ficulty here is one of deciding the peer is whereabouts of the lectotype and the varied interpretations of Vahl's original, and cryptic comments add much to the problem. Xyris laxifolia, like X. jupicai, is a com- mon weed of disturbed, mildly to very acid open areas and riverine forest borders. In North America it ranges from southeastern Virginia southward into northern Florida, thence west in the Coastal Plain into south- central Texas, with an outlier in Tabasco, Mexico. These northern populations are des- ignated var. iridifolia (Chapm.), but even varietal distinction is difficult. In South Amer- ica the species is general in wetlands, partic- ularly at lower elevations from the Andean foothills to the Atlantic. Its most common associate xyrid is X. jupicai from which X. laxifolia differs primarily by having taller habit, broader and purple-based leaves, wider and smooth-edged scapes, larger and darker spikes with narrower bract apices.. The seeds are mostly farinose. Much confusion in iden- tification comes from larger specimens of the former and smaller specimens of the latter, also from the fact that the red pigments that so definitely mark X. laxifolia in the field do not persist for very many years on herbarium material or are entirely lost when specimens are treated with alcohol or formalin (as they so often are in the tropics). 3. Xyris jupicai Rich., Act. Soc. Hist. Nat. Paris I: 100. 1792. TvPE: French Guiana: “Cayenne,” LeBlond (lectotype, P). Fig- ure 3. X. us Pers., Syn. Pl. I: 56. 1805, non Lamarck, 191. X. Ps Michaux, Fl. Bor. Am. I: 23. 1803. X. communis Kunth, Enum. Pl. 4: 12. 1843. TYPE: Brazil: without precise locality, “Amazonas, Para, Poep- in 5. X. gymnoptera Griseb., Cat. PL. Cub. 223. 1866, in part (and in part X. ambigua Beyr.). TYPE: Cuba: without definite locality, 1865, C. Wright 3228 (isotypes, NY, US X. acuminata Miq. ex. Steud., Syn. Pl. Glum. 2: 284. 1855. X. jupicai var. brachylepis Malme, Sv. Bot. Tidskr. 21: 394. 1927. X. macroc icy f. minor (C. Martius) M. Kuhlmann & Kuhn, Fl. Dist. Ibiti. 34. 1947. Annual or short-lived, perennial, solitary or tufted, 1-10 dm high, the stems contract- ed, mostly dying by end of season, rarely perennating by bulbous overwintering lateral uds. Leaves mostly erect or ascending, 1- 6 dm long; sheaths entire, often as long as the blades, tapering gradually from a dilated, pale green, dull brown or stramineous, keeled, ribbed base to the blade, there with edges convergent and merging with blade or with a short, erect triangular ligule; blades linear, strongly flattened, straight, the apex acumi- nate, erect or incurved, the margins thin or narrowly incrassate, smooth or papillate, the surface yellow-green with short streaks of ma- roon, finely nerved. Scape sheaths much shorter than leaves, the tubular bases multi- costate and twisted, stramineous, pale green or pale brown, upwardly dilating and open, keeled, then narrowing to a slightly divergent, cusplike flat blade. Scapes straight, erect, rarely somewhat twisted, proximally terete, multistriate and 1-2-costate, tan or stramin- eous, distally slightly compressed and 1-2- costate, green, the costae narrow but strong, usually papillose-tuberculate, rarely smooth. Mature spikes ovoid, ellipsoid or oblong, 0.5- Volume 75, Number 2 Kral 541 1988 Xyris Lam | MN / N) ACN FIGURE 3. Xyris jupicai (Kral 25970) .—a. Habit sketch.—b. Leafapex.—c. Leaf at blade—sheath juncti i. Staminode, eine d ertile bract.—g. Lateral sepal.—h. Petal blade, stamen.— . Seed. d. Leaf base. —e. Spike hair.—j. Stylar apex. — One valve of capsule.— 542 Annals of the Missouri Botanical Garden 1.5(-2.5) cm long, blunt or rarely acute, of many (several in depauperate individuals) rather loosely and spirally imbricate bracts, several of the lower ones sterile, narrower and shorter than the fertile bracts and grading into them; fertile bracts obovate to ovate, 5— 7 mm long, apically rounded to broadly acute, the margin entire, aging erose, the backs strongly rounded, ecarinate, pale to dark red- brown, lustrous, the dorsal areas rectangular to elliptic, green, aging brown. Lateral sepals linear-oblanceolate, slightly curvate, from 3 mm long to equaling bracts, equilateral, very thin, usually pale green-tan, the thin but broad brownish keel lacerate from ca. middle to acute apex. Petal blades broadly obovate, ca. 3-4(-5) mm long, yellow, the broadly round- ed to subtruncate apex denticulate-erose, the base cuneate. Staminodia bibrachiate with 2 broad flat branches terminally penicillate, most of the cells congested with dark material. An- thers lance-oblong, ca. 1.5 mm long, deeply bifid and auriculate on filaments ca. 1 mm long. Capsule thin, narrowly obovoid or ellip- soid, plano-convex, ca. 3-4 mm long, the placentae parietal, continuous to locule sum- mit. Seeds broadly ellipsoid, ca. 0.4-0.5 mm long, pale amber, longitudinally with several papillate straight or anastomosing ribs and with several weaker cross-lines. Distribution. “Throughout the southeast- ern United States southward through the Ca- ribbean and Central America, mostly at lower elevations, south into Argentina. This morning-bloomer is a common wet- lands weed over much of its range, generally in Central and South America, sharing its habitat with X. laxifolia, with which larger specimens are confused. Original material of “X. macrocephala" Vahl is largely X. ju- picai (see discussion of species 2). 4. Xyris navicularis Griseb., Cat. Pl. Cub. 223. 1866. TYPE: Cuba: “savannas Da- yanigua, C. Wright 3229" (presumed location of lectotype, HAC; isotypes, NY, US). Figure 4. X. subni culas Malme, Ark. Bot. 138: 15. 1913. TYPE: elize: Stann Creek, Honduras, Rev. J. Robertson a BM). Perennial or annual, usually cespitose, to 4.5 dm tall. Leaves flabellately spreading, (4-)10-15(-20) cm long; sheaths entire or, rarely, papillose or scabridulous-edged, the broad, clasping base maroon, brown or red- brown, often papillose, narrowing gradually into the blade, the ligule lacking or incon- spicuous; blades linear-gladiate, flattened, sometimes slightly twisted, sometimes cur- vate, 2-5 mm wide, yellow-green or maroon- tinted, the apex incurved-acute, incrassate, the edges thin, entire or more often papillate or tuberculate-scabrid, the surface smooth or with short lines of papillae or tubercles proxi- mally. Scape sheaths shorter than leaves, ter- ete and multicostate, lustrous brown proxi- mally, the blade often conspicuous, leaflike. Scapes linear, sometimes twisted, distally flat- tened, bicarinate, 0.8-1.2 mm broad, the costae often papillose-tuberculate. Spikes nar- rowly ovate to oblong (0.7-)1-2(-2.5) cm long, of several to many loosely spirally im- bricated, pale brown or dark brown bracts; lowest sterile bracts distinctly smaller than the fertile bracts, narrower, mostly lanceo- late, keeled, acute, grading gradually into the fertile bracts, these ovate to broadly oblong or suborbicular, ecarinate or imperceptibly so, ca. 4-5 mm long, broadly rounded, the rounded backs papillate, the edges thin, at first entire, the dorsal areas subapical, lan- ceolate to ovate, paler, often inconspicuous (particularly in age). Lateral sepals free, equi- lateral, elliptic, 3.5-4.5 mm long, strongly curvate, lustrous red-brown, the keel firm but narrow, ciliate or ciliolate from near base to acute apex. Petal blades broadly obovate, ca. 5 mm long, yellow, the rounded apex crenate- erose. Staminodia bibrachiate, the flat branches at tips long-penicillate. Anthers broadly oblong, ca. 1.5-2 mm long, deeply bifid and sagittate, on filaments ca. 1.5 mm long. Capsule obovoid, 4 mm long; placen- tation 3-parietal. Seeds ellipsoid or ovoid, 0.5— 0.6 mm long, short-caudate, with 9-10 raised and smooth longitudinal ribs per side and sev- 543 Volume 75, Number 2 Kral 988 Xyris LEAN ro A i Xyris ie Bs Proctor 38799) .—a. Habit sketch.—b. Leaf apex.—c. Leaf blade-sheath — f. Fertile bract.—g. Lateral sepal.—h. Petal blade, stamen.—i. Stami- deii (outer be showing line of dehiscence along dorsal bundle, this revealing FIGURE 4. junction.—d. Leaf base.— e. —j. Stylar apex. — k. ta, node. tips of two placental lines of seeds) .—l. Annals of the Missouri Botanical Garden eral indistinct crosslines, deep, lustrous, trans- lucently amber. Distribution. | Low-altitude sandy palm or pine savanna from western Cuba through Belize and Honduras southward into Nica- ragua (Zelaya); Colombia. Xyris subnavicularis Malme, supposedly distinguished by having longer leaf sheaths associated with ciliate-scabrid blades and by its indistinct dorsal areas, falls in regard to these characters well within the total range of variation of X. navicularis. The species definitely centers in Cuba as to abundance. 5. o Lam., Illustr. I: 132. 1791. "Guyane francaise," Herb. Hor- nemann (P, photo at F). Figure 5. X. platycaulis Poiret, Encycl. 8: 820. x Dietr., Sp. Pl. 2: o € X. . C. Mariu; FL Bras. 24, Beibl. 2: 57. 1841. X. i Ku 1843. ? X. bahiensis Steudel, Syn. Pl. Con 2: 287. 1855. Low to moderately robust, solitary or ces- pitose, a des soft-based annual (1-)1.5- . Leaves erect to spreading flabel- lately, he principal ones 5-40 cm long; sheaths up to 1⁄2 as long as blades, pale brown to stramineous or pink, soft, entire, multi- costate, keeled just above base, the mid costa often papillose, the edges scarious, entire, gradually narrowing to the blade, there with a ligule up to 1 mm long; blades flattened, straight, ensiform-linear, pale or deep green, bluntly incurved-acute, the edges thin, or slightly incrassate and papillose-tuberculate. cape sheaths shorter than principal leaves, loosely tubular, multicostate, proximally lus- trous brown or pale brown, distally with short, erect blades. Scapes straight or flexuous, twisted, mostly bicostate distally, even ancip- ital, 1-3 mm wide, the costae entire or pa- pillose. Spikes broadly ovoid, subglobose or hemispheric, 0.4-1.5 cm long, blunt, of sev- eral thin, scarious, pale to red-brown, lus- trous, spirally imbricate bracts, these with short, subapical, gray -green, lanceolate dorsal areas, or the lowest bracts with elongate-lan- ceolate dorsal areas. Sterile bracts few, strongly keeled, about as long as the fertile bracts, these broadly obovate to suborbicular, mm long, apically rounded, entire, the backs rounded, keeled strongly toward apex. Lateral sepals free, subequilateral, linear-ob- lanceolate, 4-5 mm long, acuminate or nar- rowly acute, the keel firm, smooth, slightly broadened distally. Petal blades obovate, ca. 3 mm long, coarsely dentate at broadly round- ed apex. Staminodia broadly bibrachiate, the branches distally penicillate. Anthers oblong, ca. 1 mm long, deeply bifid and sagittate, on filaments ca. 0.5 mm long. Capsule oblong- obovoid or ellipsoid, ca. 3.5 mm long, um- bilicate, the valves firm, lustrous brown, the placentation parietal from base to tip of fruit. Seeds broadly ovoid or ellipsoid, 3.5-4.5 mm long, translucent, pale to deep brown, biapi- culate, strongly longitudinally ribbed, sparsely and finely cross-lined. Distribution. Widespread and weedy in open, wet, acid areas and often littoral; Africa (including Madagascar); South America from the Guianas south to southern Brazil. 6. Xyris brachyfolia Kral & Wanderley, sp. nov. TYPE: Brazil. Amazonas: plateau of northern massif of Serra Araca, 0°51- 57'N, 63?21-22'W, 1,200 m; southern extremity of northern plateau of Serra Araca, shrub forest. Growing on floor of moist shrubby forest, 15 Feb. 1984, G. T. Prance, I. L. do Amaral, J. J. Pipoly, A. S. Tavares, M. G. da Silva, C. D. A. da Mota & A. Cress 29079 (holo- type, INPA; isotypes, NY, VDB). Figure 6 Planta perennis, caespitosa, glabra, 4-5 dm alta. Caules breves. Radices a de ic curta, suberec- ta, brunneola, nitida, vaginis scaporum breviora, persaepe stricte vagina, vel usque "y 5 cm pes nga, vulgo vagina; vaginae albovillosiciliatae, ad basin gradatim dilatatae, in laminas gradatim a er oes leviter compressae, excurvatae . Vaginae sca- porum elongatae, usque m ER cg T aniaya ad n tortae, nitidae. Scapi graciles, erecti, torti, apicem versus teretes, multistriati vel valde costati, 0.7-0.9 mm crassi. Spicae pauciflorae, ellipsoideae, tum anguste obovoideae, Volume 75, Number 2 Kral 1988 Xyris 545 FIGURE 5. d. Leaf base.—e. Upper sc Staminode.—k. Stylar apex. EE Fertile 5-7 mm ongae. Bracteae laxae spiraliter imbricatae, area dorsali conspicue; bracteae steriles 5—6, infimis late ovatis vel suborbiculatis, 1-2 mm longis, margine scariosis et doler albovillosis, Men NP x ovatis, 3.5-4.5 m Xyris anceps dimid et al. 6950) .—a. s sketch.—b. Leaf tip. —C. Mi sep yapu sector.— al blade, stamen.—J. m. Seed. act.—h. Lateral sepal.—i. Pet f: e.—g. Two views of capsule, at t lefi one valve, at right intact capsule.— 4.5- longis, rotundatis, integris; bracteae fertiles ovatae 5 mm longae, rotundatae, i Sepala lateralia libera, subaequilateralia, lanceolata, ca. 546 Annals of the Missouri Botanical Garden FIGURE 6. Xyris brachyfolia (Prance et al. 29079).—a. Habit sket ch.—b. Principal leaf.—c. Spike Lateral sepal.—e. Fertile bract. —f. Peta uals stamen.—g. Staminode and beard hair. —h. Stylar r apex.— Valve of capsule with attached placenta, seeds. —j. Seed. Volume 75, Number 2 1988 Kral Xyris 547 m longa, acuta, delicatula, leviter curvata, ala carinali distaliter distante scabrolacera vel scabrida. La talorum obovata, luteola, ca. 5 mm longa, anguste ro- integra. Staminodia bibrachiata, brachiis longi- pebicilatis: Antherae d tae, 1-1.2 mm longae; filiis ca. ] mm longis. Capsula ca. 3 mm longa, ovoidea, pla- centa basalis-parietalis. m mina fusiformes, ca. 1.5 mm longa, translucentia, atroferruginea, valde longitudine cos- talis. Cespitose perennial, smooth, 4-5 dm high. Stems short. Roots slender and fibrous. Prin- cipal leaves short, suberect, brownish, lus- trous, shorter than the scape sheath, often strictly sheath, up to 5 cm long, then mostly sheath; sheaths white-villous-ciliate, gradually broadening to base, gradually narrowing to apex, eligulate; blades lightly flattened, ex- curvate, acute, entire. Scape sheaths elon- gate, to 1 cm long, multicostate, twisted, shin- ing, short-bladed or cuspidate. Scapes slender, erect, twisted, terete toward apex, multistriate to shallowly costate, 0.7-0.9 mm thick. Spikes few-flowered, ellipsoid, becoming narrowly obovoid, 5-7 mm long. Bracts spirally and loosely imbricate, the dorsal area conspicu- ous; sterile bracts 5-6, the lowest broadly ovate to suborbicular, 1-2 mm long, margin- ally scarious and dorsally white-villose, the uppermost ovate, 3.5-4.5 mm long, rounded, entire; fertile bracts ovate, 4.5-5 mm long, rounded, entire, the matrix brown, lustrous, the dorsal area ovate, green, medially shal- lowly costate. Lateral sepals free, subequilat- eral, lanceolate, ca. 5 mm long, acute, deli- cate, slightly curvate, the keel distally distantly scabrolacerate or scabrid. Petal blades ob- ovate, yellow, ca. 5 mm long, narrowly round- ed, entire. Staminodes bibrachiate, the branches long-penicillate. Anthers lanceolate, 1-1.2 mm long, on filaments ca. 1 mm long. Capsule ca. 3 mm long, ovoid, the placentae basal-parietal. Seeds fusiform, ca. 1.5 mm long, translucent, deep red-brown, strongly ribbed longitudinally. This species is morphologically closest to extremes of X. fallax Malme and is partic- ularly noteworthy because of its transitional placentation type: basal-parietal. This is a fur- ther indication of the weakness of sections in Xyris based primarily on placentation. The long and strongly ribbed seeds with strong, irregularly raised outer seed coat are hardly distinguishable from those of X. fallax; pig- mentation of leaves and scapes is likewise within the range of that species. However, the distinctly pale-villous patches on the spike bracts are notable and distinct. The sheath borders are consistently long-ciliate. 4. Xyris neblinae Maguire & Lyman B. Smith, Mem. New York Bot. Gard. 10: 26, fig. 11A—E. 1963. TYPE: Venezuela. T. F. Amazonas: summit savanna near west escarpment 2 km north of Cumbre Camp, occasional, 1,800 m, Cerro de La Neblina, Rio Yatüa, 12 Jan. 1954, B. Maguire, J. J. Wurdack & G. S. Bunt- ing 37243 (holotype, NY; isotypes, US, VEN). Figure 7. Tall, rushlike, cespitose, brittle perennial to 1 m tall, the stems contracted. Leaves erect, to 7 dm long; sheaths less than V4 as long as blades, firm, eciliate, dull-castaneous, nearly black, narrowing gradually from an abruptly dilated, deepset base to blade, at blade level with an erect, truncated ligule broader than leaf blade base; blades narrowly linear, terete except at the often narrow and dorsiventrally flattened base, ca. 1-1.5 mm thick, multiribbed; conic-subulate apically, the surface smooth except toward papillate base. Scape sheaths shorter than principal leaves, fluted, apically short-bladed. Scapes flexuous, twisted, terete distally, ca. 1 mm thick, ecos- tate, striate, sometimes papillate. Spikes el- lipsoid to lance-ovoid, 1-2 cm long, acute and attenuate, reddish brown, of many, loose- ly and spirally imbricated papillose bracts with distinct, paler, red-brown to green dorsal areas, these with a low midnerve; sterile bracts sev- eral, oblong or narrowly ovate, slightly to much shorter than the fertile bracts, the low- est smallest and keeled, grading gradually into the fertile bracts, these ca. 8 mm long, broad- ly oblong, ciliolate, the backs convex and ecarinate, the apex broadly to narrowly rounded. Lateral sepals free, subequilateral, lineal, ca. 8-10 mm long, sometimes exsert- 548 Annals of the Missouri Botanical Garden FIGURE 7. Xyris neblinae (Maguire et al. 37243) .—a. Habit sketch.—b. Leaf apex.— c. Sector of leaf blade a few cm above sheath apex.—d. Leaf sheath-blade junction. —e. Leaf base.—f. Spike. —g. Lateral sepal. —h. Volume 75, Number 2 1988 Kral Xyris 549 ed, the narrow, red-brown keel red-ciliolate or villosulous-ciliate from near base to the acute but blunt apex. Petal blades broadly obovate, 7-8 mm long, yellow, the broadly rounded apex lacerodentate. Staminodia bi- brachiate, the short, broad, flat branches without hairs. Anthers lance-oblong, 2-2.5 mm long, emarginate, shallowly auriculate, on filaments ca. 1 mm long. Capsules obovoid, 4-4.5 mm long, the placentation basal, the funiculi stubby, the valves firm, dehiscing only 24 down, lacking septa. Seeds few, mostly 3- 4 per capsule, cylindric, 2.5-3 mm long in- cluding a pale apiculus 0.5 mm long, the surfaces deep amber, longitudinally finely ribbed, the ribs (particularly toward base and tip of seed) muriculate, papillate or tuber- culate. Distribution. Known only from the type area, in wet rocky savanna. This rare species, so far known only from Cerro de La Neblina, most resembles the brit- tle, rushlike X. juncifolia Maguire & Smith of Cerro Guaiquinima of neighboring Estado Bolivar in Venezuela. However, that species has leaf bases brown (rather than castaneous or near black), strongly inequilateral (versus equilateral) sepals, bearded (versus beardless) staminodia, and shorter and smoother seeds. Nonetheless, the superficial resemblance is striking. 8. Xyris juncifolia Maguire & Lyman B. Smith, Mem. New York Bot. Gard. 10: 26, fig. 10A-E. 1963. TYPE: Venezuela. Bolivar: common in Cumbre Camp, 2,000 m, Cerro Guaiquinima, Rio Pa- ragua, 25 Dec. 1951, B. Maguire 32750 (holotype, NY; isotype, US). Figure 8. Slender, rushlike, cespitose, brittle peren- nial to 1 m high, the stems contracted or short-ascending and short. Leaves erect or ascending, 4-6 dm long; sheaths dull brown or red-brown, less than 1⁄2 as long as blades, eciliate, narrowing gradually from a slightly dilated base to leaf blade, there producing an erect, firm, broad, rounded ligule to ca. 2 mm long; blades terete, fluted, narrowly lineal, to 1.5 mm thick, toward base above ligule usu- ally deeply sulcate, toward apex narrowing to a blunt, callused tip. Scape sheaths much shorter than leaves, tubular and multiribbed proximally, opening distally, slightly dilated, with a cusplike, blunt-tipped blade. Scapes slightly flexuous and twisted, about the width of leaf blades or slightly wider, wandlike, dis- tally terete, sometimes shallowly grooved. Spikes ellipsoid to obovoid, 1.5-2 cm long, reddish brown, acute, basally attenuate, of many tightly spirally imbricate, ciliolate bracts with distinct, deeper brown dorsal areas, the sterile bracts numerous, broadly ovate, broad- ly rounded, much smaller than and grading into the fertile bracts; these obovate, ca. 6 mm long, rounded-folded, ecarinate but with a pale, low midrib, the apex broadly rounded, aging lacerate. Lateral sepals free, strongly inequilateral, oblong-curvate, ca. 5 mm long, acute or blunt, the dark red-brown, firm keel ciliolate, toward apex reddish fimbriolate. Pet- al blades broadly elliptic or broadly obovate, ca. 5 mm long, the broadly rounded apex lacerate. Staminodia bibrachiate, the narrow flat branches distally long-penicillate. Anthers ca. 1.5 mm long, oblong, deeply bifid and sagittate, on filaments ca. 1.5 mm long. Cap- sule broadly obovoid, slightly compressed, ca. 3 mm long, the placentation basal, the valves producing low septa. Seeds few, ellipsoid-cy- lindrical, ca. 1.5 mm long, apiculate, deep amber, longitudinally rather coarsely 20-24- ri Distribution. Locally frequent, summits of tepuis Guaiquinima and Jaua, Bolivar, and tepuis Duida and Paru, Amazonas, Venezuela. Petal blade, stamen.—i. Stylar apex.—j. Staminode.— k. Capsule; below with seeds on funicles, valves separating naturally, above with two valves removed, showing seedless funicles. —l. Seed showing ribbing and muriculation (seeds much darker than shown). 550 Annals of the Missouri Botanical Garden iam FIGURE 8. Xyris juncifolia ciui 32150) .—a. Habit sketch. —b. Leaf tip. —c. Leaf blade-sheath m inner view, side view.—d. Leaf base. —e. Spike. —f. Fertile bract. —g. Lateral sepal. —h. Petal blade, s Volume 75, Number 2 1988 Kra Xyris This species, with its fusiform-ellipsoid spikes; terete, rushlike, brittle foliage; and distinct dorsal areas, most resembles X. ne- blinae of Cerro Neblina but differs in its in- equilateral (rather than equilateral) lateral se- pals and brownish (rather than castaneous) base. These two species and the Colombian X. terrestris form a complex of brittle-fo- liaged, rushlike xyrids closest to X. atriceps Malme subsp. neblinensis. Known until 1951 only from the type locality, X. juncifolia has now been collected from Cerro Jaua (Stey- ermark et al. 109427, NY, VEN), from Cer- ro Duida (Steyermark 124560, NY, VEN), and from Paru (Cowan & Wurdack 31185, GH, NY, US), as well as several times more from the type region. 9. Xyris lanulobractea Steyerm., Field- iana, Bot. 28(1): 109, fig. 16D-H. 1951. TYPE: Venezuela. Bolivar: swampy ground, 1,200 m, Kavanayen, Gran Sa- bana, 26 Oct. 1944, J. A. Steyermark 59336 (holotype, F; isotypes, US, VEN). Figure 9 Slender, rushlike, cespitose, hard-based, glabrous perennial 4-7 dm high, the stems contracted. Leaves erect, 2-4 ong; sheaths entire, elongate but less than 1⁄4 of blade length, deep glossy red-brown, the backs ecarinate, the sides tapering gradually into blade, there with an erect ligule to 3 mm long; blades terete or oval in cross section, 0.5- 0.7 mm thick, fluted, slightly flattened and deeply sulcate at base above ligule, apex ta- pering-subulate-conic or with very tip dilated clavately. Scape sheaths much shorter than leaves, tubular at base, multicostate, twisted, open at apex, keeled, stubby-bladed. Scapes slightly twisted, straight or flexuous, terete m thick. Spikes broadly ovoid to obovoid, 0. T- c 1.5) cm long, blunt, dull brown, attenuate, the many spirally imbricate bracts with conspicuous dorsal areas; sterile bracts several, ovate, narrowly to broadly rounded, ecarinate, the lowest evidently much smaller than the fertile bracts, grading into them; fertile bracts broadly elliptic to obovate, ca. 6 mm long, with broadly rounded, white- villosulous borders, convex and ecarinate backs, and reddish brown, obovate or broadly elliptic dorsal areas. Lateral sepals strongly curvate, elliptic, ca. 4 mm long, thin, the broad firm keel ciliolate below middle, in- creasingly pale-villous-fimbriate above mid- dle. Petal blades broadly obovate to subor- bicular, yellow, ca. 5 mm long, the rounded apex serrulate-dentate. Staminodia bibra- chiate, the slender recurved branches densely penicillate- pilose. Anthers oblong, 1 mm long, ifid and sagittate, on filaments ca. 1 mm long. Capsule obovoid, ca. 2 mm long, pla- centation basal, the valves without septa. Seeds few on short, stubby funicles, cylindrical or lance-ovoid, amber, 1.1-1.3 mm long, often angulate, finely ribbed longitudinally. Distribution. Low to medium-elevation savanna, southeast Venezuela (Bolivar), Guayana, and contiguous northern Brazil (Amazonas). Additional specimens examined. BRAZIL. AMAZONAS: rd. to Igarapé Preto ca. 60 km SE of Transamazon Hw 2 July 1979, Calderon et al. 2743; Mun. Humaita, estrada da Humaita—Jacarecanga, km 62, 17 June 1982, Teixeira et al. 104.938 (INPA, NY, US, VDB). GUAYANA. UPPER MAZARUNI DIST.: Makwaima savanna near Ma- yoripai, at Kako River, 8 Feb. 1985, J. Renz 14145 (U). VENEZUELA. BOLIVAR: meseta norte de Serrania Car- aruban, 19 Feb. 1964, G. Agostini 403 (NY, U, VEN); Auyantepui, Sept. 1937, F. Cardona 262 (US, VEN); ca. 17 km al NE de Haber: Huber et al. 6732 (MYF, VDB, VEN) hacia Icabarü, 27 July 1983, A : Alarcon 7891 (MYF, NY, VEN); 20 km NE de Urim ~ et al. 9650 m VD ); inferior, 18 Nov. 1984, Huber eta VEN); ca. 35 km A W de Caserio de Chiguao, 23 Mar. 1985, Huber 10355 (MYF, VDB, VEN); 46 km N of Sta. Elena, 28 July 1983, Kral 70562; Río Yuruani just above falls, 17 Dec. 1984, Kral 72163; N of Rio Yuruani Ferry, Kral 72194 (Kral numbers to be distributed, pres- ently MYF, VDB); between Urarupata and Enemasic, 6 Feb. 1952, Maguire 33234 (GH, US, VEN); 13 km NE E i. pu ra Stylar apex.—k. Dehiscing capsule base) .— (note that dehiscence in this species is not to capsule 552 Annals of the Missouri Botanical Garden FIGURE 9. Xyris lanulobractea (Maguire 33234, Kral 70562) .—a. Habit sketch.— b. wu rum —c. Leaf at sheath-blade junction.— . Leaf base.—e. Spike.—f. Fertile bract.—g. Lateral sepal. —h. Petal, stamen.—i. Staminode.—j. Stylar apex.—k. Capsule, one valve removed, showing placentation.—l. Seed. Volume 75, Number 2 1988 Kral 553 Xyris Sta. Elena, 2 Dec. 1982, Steyermark & Liesner 127486 (VDB, VEN). This slender morning bloomer is most sim- lar superficially to X. globosa Nilsson, dif- fering in its somewhat more slender habit, more tapering (rather than bulbous) base, more consistently terete leaf blades, ligulate sheath apex, plumose (rather than beardless) stam- inodes, and broader petal blades. Brazilian collections of X. globosa may well turn out to be X. lanulobractea. 10. Xyris terrestris Idrobo & Lyman B. Smith, Caldasia 6(29): 208, fig. 10. 1954. TYPE: Colombia. Vaupés: “Cerro de Caenda” (sabanas), 380-670 m, Rio Kubiyu, 4 Nov. 1952, H. García-Bar- riga 15090 (holotype, COL; isotypes, GH, NY, US). Figure 10 Solitary or small-clumped, slender, smooth perennial 3-5 dm high, the stems short, erect from a short, stout, horizontal or ascending rhizome. Leaves mostly erect or ascending, twisted, 1-2.5 dm long; sheaths much less than 4 as long as blades, entire, proximally rich, lustrous red-brown or castaneous, strongly dilated at very base, thence upward gradually narrowing into an erect, narrowly triangular, chaffy ligule 3-6 mm long; blades filiform-linear, flexuous, terete or elliptic or rounded-angulate in cross section, 0.5-0.6 mm thick, yellow-green, blunt-conic at apex, longitudinally with a single, spiral, shallow to deep, papillate, usually rusty-colored sulcus. Scape sheaths shorter than leaves, with short- er blades similar to those of leaves. Scapes linear, twisted and flexuous, distally terete, ca. 0.5-0.6 mm thick, ecostate but multistri- ate. Spikes ellipsoid, or lance-ovoid, 0.7- 2 cm long, subacute, attenuate-based, of nu- merous, spirally imbricate, stiff, brownish bracts, the sterile bracts numerous, the lowest lance-triangular, keeled, much smaller than, and grading into, the fertile bracts, these mostly broadly obovate, broadly to narrowly rounded apically, 4-5 mm long, the margins entire to erose in age, the back rounded, ecarinate, with strong but small, red-brown to yellow-brown, elliptic dorsal areas. Lateral sepals strongly curvate, oblong, ca. 4 mm long, free, very inequilateral, the broad, deep, reddish brown keel scabrociliate from near base to blunt apex. Petal blades obovate, ca. 4 mm long, yellow, the broadly rounded apex erose-denticulate. Staminodia bibrachiate, the flattened branches long-penicillate. Anthers oblong, ca. 2 mm long, shallowly bifid and auriculate, on filaments ca. 0.5 mm long. Capsule narrowly obovoid-apiculate, brown, 2-2.5 mm long, dehiscing only % way to base, the placenta massive and basal, the fun- iculi short, broadly clavate. Seeds several, cylindric-fusiform, 1.3-1.5 mm long, red- amber, irregularly anastomosing-ribbed lon- gitudinally. Distribution. Sandy savannas, Vaupés, southeastern Colombia, rare. Additional specimens examined. COLOMBIA. VAUPÉS: Mesa de Yambi savanna, 15-16 Apr. 1953, Schultes & Cabrera 14235A (COL, GH, NY, U, US); Araracuara savannas, Rio Caquéta, 6 Sept. 1959, Maguire & Fer- nandez 44163 (NY) There are so many species shared by bor- der states in Colombia and Venezuela that it is reasonable to expect this plant to be found in T. F. Amazonas in Venezuela, where there is much savanna suitable for it. 11. Xyris scabridula Steyerm., Fieldiana, Bot. 28(1): 111. 1951. TYPE: Venezuela. T. F. Amazonas: around rills on rocky dry ridgetop, Brocchinia Hills, 1,700- 1,900 m, Cerro Duida, 1 Sep. 1944, Steyermark 58168 (holotype, F; iso- types, NY, VEN). Figure 11. Slender, cespitose perennial 1.8—4 dm high, the stout stems mostly contracted. Leaves shorter than scape sheaths, strongly flexuous and twisted, mostly erect; sheaths eciliate, 1⁄4 or less of blade length, firm, at very base deep red-brown or castaneous, strongly ribbed, lus- trous, becoming roseate or purple above, sca- bridulous, narrowing gradually to blade; blades variously elongate, strongly rib-angled in cross section, also deeply sulcate, narrowly linear, ca. 1 mm thick, reddish or purplish and pale- Annals of the Missouri Botanical Garden FIGURE 10. Xyris terrestris (Schultes & Cabrera 19179).— a. Habit sketch. —b. Leaf apex.—c. Leaf blade a few cm below apex.—d. Further down leaf blade than c (and more reduced) .—e. Leaf sheath—blade junction. — . Leaf base. —g. Spike.—h. Lateral sepal and greatly enlarged sector of keel hairs. —i. Petal blade, stamen.— j. Staminode.—k. Stylar apex. —1. Outline of capsule.—m. Dehisced capsule showing basal placentation.—n. Volume 75, Number 2 1988 Kral 555 Xyris scabridulous toward base, smooth, green to- ward tip, narrowed abruptly to a blunt, cal- lused apex. Scape sheath much shorter than blade, mostly tubular, twisted, strongly mul- ticostate, reddish brown or pink at base, dis- tally with a short, blunt cusp. Scapes twisted and flexuous, subterete toward apex, ca. 0.8 mm thick, with 1-3(-4) strong, smooth cos- tae and some less distinct ribs, or nearly ecos- tate, proximally strongly multicostate. Spikes ovoid, 7-8 mm long, of several erect, spirally imbricate, ecarinate, brown, definitely papil- late bracts without dorsal areas, the sterile bracts several, the lowest much smaller than the fertile bracts, orbicular or reniform, grad- ually grading into fertile bracts, these 4—5 mm long, broadly obovoid, broadly rounded apically, appearing entire but minutely pap- illate-ciliate. Lateral sepals free, equilateral, elliptic and strongly curvate, ca. 4 mm long, obtuse, the keel rusty-ciliolate. Distribution. Cerro Duida, Territorio Federal Amazonas, and Chimanta Massif, Es- tado Bolivar, Venezuela. Additional specimens examined. VENEZUELA. BOLIVAR: savanna summit of Macizo Chimanta, Huber & Colella 9001 (NY, VDB, VEN); Huber & Steyermark DI (NY, VDB, VEN); Huber gt al. 9070 (NY, VDB, EN); Huber 9576 (NY, VDB, VEN); Steyermark & n ie 1010 (F); Steyermark et al. 115922 (F); Stey- ermark 128429 (VDB, VEN). This taxon has some affinity to X. sub- glabrata Malme of lower elevations in T. F. Amazonas, Venezuela, but has smoother spikes with no evident dorsal area. It also resembles X. stenophylloides Malme, an equally rare plant from the same area, whose leaf blades, though very narrow, are flattened. None of these species are much collected, so that com- parisons of flowers and seeds are not yet made. 12. Xyris atriceps Malme, Bull. Torrey Bot. Club 58: 325. 1931. TYPE: Vene- zuela. T. F. Amazonas: forming tussocks, 6,700 ft., Cerro Duida, Ridge 15, Aug. 1928-Apr. 1929, G. H. H. Tate 688 (lectotype, NY; isolectotype, US). Densely tufted, slender, low to tall peren- nials 2-6 dm high, the stems short to elon- gate, up to 5 cm long. Leaves elongate, linear to filiform, erect or ascending, 1.5-3 dm long; sheaths deep reddish brown to nearly black, lustrous, smooth to papillose-rugulose apical- ly, less than 1⁄2 as long as blades, entire, tapering evenly from broad, ecarinate base to distinct, firm or thin, erect, broad ligule 2-10 mm long, there broader than the usually terete blade; blades deep green, linear to fil- iform, 0.5-1 mm thick, sometimes fluted, usually with a median ventral sulcus, the apex blunt, broadly rounded or truncate, rarely conic, smooth, the surfaces smooth except toward the often rugulose-papillose base. Scape sheath shorter than leaves, loose, often pur- plish, rarely pink, the blade short, erect, blunt. Scapes flexuous or straight, twisted, terete, 0.5-1 mm thick, smooth or white-puncticu- late from sunken stomata. Spike ovoid to ob- long or obovate-turbinate, 0.5-1.5 cm long, the base attenuate, of several pale to deep brown or castaneous, spirally loosely imbri- cate bracts without distinct dorsal areas. Ster- ile bracts up to 6, erect or squarrose-tipped, triangular to obovate, entire to lacerate or ciliolate-villosulous-bordered, smaller than and grading into the fertile bracts, these oblong to obovate, 3.5-8 mm long, entire to villo- sulous-ciliate, ciliolate or pectinate-lacerate, the backs broadly rounded, smooth to papil- lose. Lateral sepals free, subequilateral, straight or curvate, linear or oblong.linear, 3.5-7 mm long, acute, sides pale brown, keels deep reddish brown, ciliolate to densely vil- lous-ciliate. Petal blades broadly obovate to suborbicular, 6-7 mm long, the broadly rounded apex coarsely erose. Staminodia bi- brachiate, the branches long-penicillate, or staminodia absent (subsp. marahuacae). An- thers oblong to lance-oblong, 2-2.5 mm long, shallowly bifid apically, deeply sagittate at base, on filaments ca. 1 mm long. Capsule cylindric to ovoid or ellipsoid, ca. 3-4 mm long, placentation appearing central but cap- sule valves with septa from base to near apex. Seeds several, deep reddish brown, cylindric or ellipsoid, 1.5-2 mm long, including a short- 556 Annals of the Missouri Botanical Garden it l f if jm f) P: DE ETEN pce Diar diei e r AhixxIiIIlTIMemsc S Ry ^^ Ni U US “N FiGURE 11. Xyris scabridula (Steyermark 58168).—a. Habit sketch.—b. Leaf apex.—c. Leaf sector at midblade.—d. Leaf blade-sheath junction.—e. Leaf base.—f. Spike, scape apex.—g. Midscape.—h. Fertile bract.—i. Lateral sepal. Volume 75, Number 2 1988 Kral 557 Xyris conic, pale appendage, the seed body finely but distinctly longitudinally ribbed. Distribution. This species appears to be confined to higher elevations in the higher tepuis of Bolivar and T. F. Amazonas, Ven- ezuela (and probably contiguous Brazil), and sorts to four fairly distinct morphologies, here given rank of subspecies, keyed as follows: KEY TO SUBSPECIES OF XYRIS ATRICEPS la. Spikes narrowly obovoid; leaf € lacking sulcus; bracts subentire, with a costa dis- tally, the bract margins ciliolate ilh bros h hairs subsp. neblinensis . Spikes narrowly to broadly obovoid; leaf blades sulcate at least ventrally; bracts becoming lac- erate, ecostate distally, the bract margins var- ious. 2a. Spikes under 1 cm long; leaf blades under 1 mm thick; bearded staminodia present. 3a. a broadly obovoid, base strongly tenuate; bract edges dist sa white ciliate and ere sp. atriceps . Spike narrowly obovoid or elipsoid or ovoid, base short-attenuate if at all so; bract Pene ird dly villos ub usu- dh co MM. very lacerate, sub- squa subsp. ió 2b. a l- T 5c m long; leaf blades ca. m thick; nins absent subsp. marahuacae — E w c 12A. Xyris atriceps Malme subsp. atri- ceps. Figure 12A (in part). Spikes broadly obovoid, attenuate-based, ca. 7 mm long, the bracts erect, the margins white ciliate; staminodial beard present; seeds ca. 1.5 mm long. Distribution. This subspecies is still known only from southern Cerro Duida, the type area. Additional material from the type area is Steyermark 58185 (F, US). 12B. Xyris atriceps subsp. chimanten- sis Maguire & Lyman B. Smith, Mem. New York Bot. Gard. 10: 19-20. 1963. TYPE: Venezuela. Bolivar: locally fre- quent, Rio Tirica, 1,925 m, 5 Feb. 1955; Chamanta Massif, J. A. Steyermark & J. J. Wurdack 485 (holotype, NY; iso- type, US). Figure 12A, a-f, g (left)-i (left) Spikes ellipsoid or narrowly obovoid, ca. 6-7 mm long, acute at base, the bracts loosely imbricate, narrower than in type, the margins at first sordidly villosulous, later becoming much lacerate, spreading; staminodial beard present; seeds ca. 1.5 mm long. Distribution. Abundant locally in high savanna of Chimantá Massif and taxonomi- cally closest to subsp. atriceps. Now fre- quently collected from the summit elevations of the Massif as follows. Additional specimens rein ined. | VENEZUELA. BOLÍVAR: "Huber et al. 9066 (NY, VDB, VEN); Steyer- mark et al. 115852 (VEN), 128008(VEN, VDB), 128958 (VEN, VDB), 128810(VEN), 129908 (MYF, VDB, VEN); Steyermark 128167 (VEN), 128854 (VEN). 12C. Xyris atriceps subsp. marahua- cae Kral & Lyman B. Smith, subsp. nov. TYPE: Venezuela. T. F. Amazonas: Dept. Atabapo, Cerro Marahuaca, cumbre, parte central de la meseta Sur-Este, al lado de una grieta, a lo largo de la Que- brada Yekuana, afluente del Rio Negro, 3?40'30"N, 65°26'20’W, 2,560 m, 10- 12 Oct. 1983, Steyermark 129579 (ho- lotype, VEN; isotype, VDB). Figure 12B. Planta fragilis, perennis, caespitosa, 4-5 dm alta, gla- breves aut elongatis, usque m, ente apicem b gris ligula rigida erecta linearo- Ni ta, usque ad 1 cm longa. Vaginae scaporum prope asin castaneae, apicem versus apertae, laminis elongat atis Scapi leviter ind teretes, ca. 1 mm crassi, multistriati, olivacei u brunneoli. Spicae multiflorae, anguste vel lat obo- oideae spiraliter imbricata, firmae, ecarinatae, rigida ae, a dorsali, ad apicem villosiciliatae, tum valde Rosi pipe bracteae steriles ovatae, plures, fertilibus breviores, in fertiles gradatim transientes; brac- teae fertiles late obovatae, 7-8 mm longae, ad apicem late rotundatae. Sepala lateralia libera, subaequilatera, oblanceolata, atrobrunneola, 6.5-7 mm longa, leviter cur- vata, obtusa; ala carinali a medio ad apicem aut solum 558 Annals of the Missouri Botanical Garden e Ficu 2A. Xyris atriceps.—a. Habit sketch (subsp. dir ar q —b. Leaf apex (general).—c. Leaf blade, midblade (general) .—d. Leaf blade-sheath cit (ge ral). —e. Leaf base (general) .—f. Spike (subsp. chimantensis) .—g. Lateral sepal a left subsp. chim at right subsp. neblinensis).—h. Petal blade, stamen, stylar apex (general) .—i. Seeds (at left frog neblincnsis, at right subsp. chimantensis, subsp. atriceps). FIGURE 12B. Xyris atriceps subsp. marahuacae (from the type) .—a. Habit sketch. —b. Leaf apex.—c. pu blade, MT sector. —d. Leaf sheath-blade junction (at left); leaf base (at right) .—e. Spike.—f. Lateral Volume 75, Number 2 Kral 559 1988 Xyris sepal.—g. Petal blade, stamen.—h. Stylar apex.—i. Capsule, idealized section showing two septa (stippled) , placentation, vascular supply to funiculi) .—j. Valve removed, showing ovular attachment to inner edge of valve septum.—k. Seed (from quasimature fruit) . 560 Annals of th Missouri Eod Garden ad apicem villosifimbriolata. Laminae petalorum late obo- vatae, ca. 6-6.5 mm longae, apice late po Stam- inodia nulla. Antherae oblongae, ca. 2 mm longae, leviter emarginatae et auriculatae; filiis ca. 1. Sm mm longis, latis. gitudine multilineata. Brittle, tufted, glabrous perennial 4-5 dm high. Roots slender. Stems short or elongate, up to 5 cm long. Main leaves erect, to 3 dm long, twisted, flexuous, longer than the scape sheaths; blades 6-10 times longer than the sheaths, narrowly linear, terete, 1-3-sulcate, ca. 0.8-1.2 mm thick, olive-green, deeply sulcate ventrally at base; tips narrowly conic; sheaths ecarinate, shining, deep castaneous, entire, dilated at base, narrowing gradually, then abruptly, to blades, ligulate at apex, the ligule rigid, erect, linear-triangular, to 1 cm long. Scape sheaths castaneous toward base, opening toward apex, elongate-bladed, with blades similar to those of leaves but narrower. Scapes slightly twisted, terete, ca. 1 mm thick, multistriate, olivaceous to brown. Spikes mul- tiflorous, narrowly to broadly obovoid, ca. 1 cm long, obtuse. Bracts erect, loosely spirally imbricate, firm, ecarinate, rigid, sooty brown, without dorsal area, villous-ciliate at apex, aging strongly lacerate, eciliate; sterile bracts ovate, several, shorter than the fertile bracts and passing gradually into them; fertile bracts broadly obovate, 7-8 mm long, broadly rounded at apex. Lateral sepals free, sub- equilateral, oblanceolate, deep brown, 6.5-7 mm long, slightly curvate, obtuse; keel villous- fimbriolate from middle to tip or solely at tip. Petal blades broadly obovate, ca. 6-6.5 mm long, broadly rounded apically. Staminodia none. Anthers oblong, ca. 2 mm long, slightly emarginate and auriculate; filaments broad, ca. 1.5 mm long. Quasimature capsules ca. 4 mm long, ovoid, with valves deeply septate from middle to base; seeds numerous, ellip- soid, ca. 1.5 mm long, translucid, longitudi- nally prominently multilined. Paratypes. All from the same massif as the type: 2- . 1975, S. S. Tillett et de 752-333 (NYF, US, VEN); 16 Feb. 1981, Steyermark et al. 124371 (NY, VDB, VEN); 1-2 Feb. 1982, Simo: et al. 126038 (VDB, VEN); 2 Feb. 1982, Steyermark et al. 126038 (VDB, VEN); 1-2 Feb. 1982, Steyermark et al. 125992 (VDB, VEN); 9-10 Feb. 1982, Steyermark et al. 126293 (VDB, VEN); 26 Mar. 1982, Steyermark & Delascio 129201, 129224 (VDB, VEN); 12-13 Oct. 1983, Stey- ermark 129476 (MO, VDB, VEN). This subspecies, abundant on summits of Cerro Marahuaca, is robust as is subsp. ne- blinensis and has the thickest spikes and leaves, lacks staminodia, and has very dis- tinctively villose sepal keels and tips, with tips of young bracts also densely villous-ciliate. 12D. Xyris atriceps subsp. neblinensis Maguire & Lyman B. Smith, Mem. New York Bot. Gard. 10: 19. 1963. TYPE: Venezuela. T. F. Amazonas: Cerro de La Neblina, Rio Yatua, locally abundant, east escarpment of Upper Cano Grande basin, at 1,900 m, summit, 1,200- 2,200 m, B. Maguire, J. J. Wurdack & C. K. Maguire 42416 (holotype, NY; iso- types, GH, K, US). Figure 12A, g (right), i (left), Leaves mostly esulcate; spikes narrowly obovoid, the bracts light brown, entire, with a low costa toward apex, the margin ciliate with short, yellowish hairs. Seeds ca. 2 mm long, the longest in the complex. Distribution and remarks. This subspe- cies appears to be relatively common in the high, wet open paramolike summit elevations along the Neblina Massif and has been found by several collectors. Additional specimens examined. VENEZUELA. T. F. AMAZONAS: Neblina Massif: Liesner 16000 (MO, NY, VDB, VEN); Vires. et al. 37244, 42342 (NY, US); Steyermark 103966 (NY, US, VEN); Thomas & Plow- man 3080 (MO, NY, VDB, VEN). BOLÍVAR: Chimanta Massif, Huber et al. 10151 (MYF, VDB, VEN) (inter- mediate between this and the subsp. chimantensis). 13. Xyris involucrata Nees in J. Bot. (Hooker) 2: 397. 1840. TYPE: “British Guiana, Schomburgk 1054” (lectotype, K; isolectotypes, K, L). Figure 13. X. asterocephala Seub. in C. Martius, Fl. Bras. 3(1): 219. 1855. Solitary or tufted, rather soft-based, stiff, short-lived perennial 2-6 dm high, the stems Volume 75, Number 2 Kral 561 1988 Xyris ALAS » C atii ZZ ty FIGURE 13. Xyris involucrata (Koyama & Agostini 7261) .—a. Habit sketch. —b. Leaf tip.—c. Leaf blade- sheath junction.—d. Leaf base.—e. Spike.—f. Fertile bract.—g. Lateral sepal.—h. Petal blade, stamen.—i. Staminode.—j. Stylar apex.—k. Open capsule.—l. Seed. 562 Annals of the Missouri Botanical Garden contracted. Leaves spreading flabellately 1- 3 dm long, the sheaths entire, equal in length to blade or longer, sharply keeled, often cas- taneous at very base, abruptly dilated, green or stramineous above, narrowing gradually to blade, eligulate; blades broadly linear, flat, 3— mm wide, at apex abruptly rounded or broadly incurved-acute, bright pale green with submarginal reddish borders, edges white-cil- late. Scape sheaths shorter than leaves, sharp- ly keeled, short-bladed. Scapes rigid, straight or slightly twisted, ancipital, strongly flattened distally, 3-4 mm wide, edges pale-ciliate, sur- face on either side of the strong and subterete scape center low-ribbed, smooth. Spikes hemi- spherical to broadly ovoid or subglobose, 0.5- 1 cm long (sometimes longer in fruit), the bracts in tight flat spirals, the lowest several bracts sterile, foliaceous, spreading leaflike, the bases strongly ciliate-keeled, the blades ciliate as in leaves, the inner sterile bracts progressively shorter, grading into the fertile bracts, the fertile bracts ca. 4(-5) mm long, firm, erect or spreading, rhombic or oblong- obovate, acute or short-acuminate, often vil- losulous-ciliate at apex, the shallowly convex- rounded backs with a large, medially papillate, rhombic, reddish brown dorsal area, bisected medially by a narrow but evident midrib. Lat- eral sepals free, subequilateral, oblong-cur- vate, ca. 3 mm long, blunt, thin, the darker, firm keel reddish ciliate or ciliolate from below middle to tip. Petal blades elliptic, 4-5 mm long, acute, entire, yellow. Anthers oblong, ca. 1.5 mm long, deeply bifid and sagittate, on filaments ca. 1 mm long. Staminodia bi- brachiate, the flat triangular branches api- cally plumose with penicillate hairs. Capsule broadly obovoid, planoconvex, ca. 2 mm long, the placentation basal, the valves not pro- ducing septa. Seeds ellipsoid-cylindric, 0.5- .6 mm long, apiculate, dark amber, lustrous, longitudinally with a few narrow but distinct ribs and very finely cross-lined. Distribution. Low- and high-altitude wet savanna (up to 1,500 meters), southeastern Colombia eastward across the Guayanas into Surinam, southward into the Amazon Basin of northern Brazil. As in the capitate-spiked, involucrate xy- rids, this one, which blooms from late morning into the afternoon, has several lovely pale yellow flowers simultaneously. This effect, along with that of the bright pale green, red- ordered leaves and bracts and the gaudy, eryngiumlike inflorescences, makes it one of the more handsome plants of the Guayana boglands. Its affinities are with the recently described X. egleri Smith & Downs of Para, Brazil, a smoother plant with shorter invo- lucral bracts, and with the following species, here described as X. pallidula, likewise from Para. 14. Xyris pallidula Kral & Wanderley, sp. nov. TYPE: Brazil. Amazonas: Mun. Humaita, estrada Humaita—Jacarecan- ga, km 150, a 60 km ao Sul. Campo natural, solo arenoso. Erva de 40 cm de altura; flores amarelas, 21 June 1982, L. O. A. Teixeira, A. J. Fife, K. Mc- Farland, C. D. A. Mota, J. L. dos San- tos, S. P. Gomes & B. W. Nelson 1263 (holotype, INPA; isotypes, NY, VDB). Figure 14. rba perennis, caespitosa, 6-7 dm alta, glabra. Caules He breves. Radices hea Folia principalia leviter flabellate amps: 10- tin m longa, vaginis scaporum parum datim LU Ron eligulatae; laminae valde compr pansae, subuliformes, 7-15 mm longae, abrupte in bracteas spicae transientes; bracteae spicae e rhombo ongae, firmae, ad apicem acuminato- subulatae; Fa sine area dorsali. Sepala lateralia libera, aequi- lateralia, valde curvata, oblonga, 4-5 mm longa, obtusa, ad apicem albovillosiciliata, ala carinali ciliati, apicem versus persaepe rufofimbriolati. Laminae petalorum ob- ovatae, luteolae, ca. 5 mm longae, acuda, integras. Stam- bin bibrachiata longipenicillatis. A , Va- dosae bifidae et sagittatae, ca. 1 mm lo ongae, filus ca. 0.5 mm longis. Capsula et semina non visa, sed placenta basalis. Cespitose, smooth perennial 6-7 mm high. Stems short. Roots slender. Principal leaves Volume 75, Number 2 1988 Kral 563 Xyris FIGURE 14. Xyris pallidula (Teixeira et al. 1263).—a. Habit sketch.—b. Leaf tip.—c. Leaf blade-sheath junction.—d. Leaf base.—e. Spike.—f Upper scape.—g. Fertile bract.—h. Lateral sepal.—i. Petal blade, stamen.—j. Staminode and beard hair. slightly spreading flabellately, 10-17 cm long, somewhat shorter than the scape sheaths; sheaths elongate, entire, 1-2 times longer than the blades, pale reddish brown, gradually narrowing from dilated base to apex, eligulate; blades strongly flattened, narrowly linear-gla- diate, 2-3 mm wide, pale olive, the margins pale red-brown-bordered, slightly thickened, papillose, the apex incurved-acute. Scape sheaths multicostate toward base, reddish, shining, strongly carinate toward apex, with blade as in leaves but shorter. Scapes rigid, 564 Annals of the Missouri Botanical Garden linear, slightly twisted, pale olive, mostly el- liptic in cross section toward apex, strongly and acutely bicostate laterally. Spikes mul- tiflorous, broadly ovoid to turbinate or hem- ispheric, ca. 10 mm long, involucrate, pale brown; involucral bracts several, spreading, subulate, 7-15 mm long, abruptly grading into bracts of spike; spike bracts rhombic- obovate, 5-7 mm long, firm, apically acu- minate-subulate, entire, without dorsal area. Lateral sepals free, equilateral, strongly cur- vate, oblong, 4-5 mm long, obtuse, white- villous-ciliate at apex, the keel ciliate, fre- quently red-fimbriolate toward apex. Petal blades obovate, yellowish, ca. 5 mm long, acute, entire. Staminodia bibrachiate, long- penicillate. Anthers oblong, shallowly bifid and sagittate, ca. 1 mm long, the filaments ca. 0.5 mm long. Fruit and seed not seen, pla- centation basal. This species, yet known only from the type material, is clearly related to the widespread X. involucrata but is easily distinguished by the lack of a castaneous “patch” on the leaf sheath base, by the eciliate leaf blades and bracts, and by the definitely subulate-tipped spike bracts. 15. eds bicephala Gleason, Brittonia 3: . 1939. TYPE: Venezuela. Bolivar: Is 220 m, Gran Sabana, Dec. 1937-Jan. 1938, G. H. H. Tate 1114 (holotype, NY). Figure 15. Robust, cespitose, thick- and hard-based perennial 2.5-10 dm high, the stems stout, short or elongate to 2 dm. Leaves spreading flabellately, 2-5 dm long; sheaths with entire margins, castaneous or near black, 25 or more as long as blades, tapering evenly to blade, there imperceptibly short-ligulate; blades broadly linear, flat, 5- 10 mm wide, the apex abruptly narrowed and rounded, incurved- acute, or narrowly rounded, the edges densely pale-pilose, densely ciliate or ciliolate, sur- faces deep green, low-nerved. Scape sheath loose, shorter than leaves, distally ciliate-car- inate, short-bladed. Scape flattened-ancipital, densely albociliate, 2.5-4 mm wide, bispicate rarely monospicate). Spike broadly ovoid, obovoid or subglobose, 0.8-1.5 mm long, deep brown or castaneous, the bracts rigid, with or without a small, paler, elliptic, subapical dorsal area; sterile bracts many, ovate-tri- angular, smaller than and grading into the numerous fertile bracts, these broadly elliptic- ovate to oblong, 5-8 mm long, narrowly to broadly rounded, entire to erose or finely lac- erate (rarely also red-ciliolate), lustrous to- ward base, dull toward tip, backs slightly rounded, ecarinate. Lateral sepals free, equi- lateral, linear-oblanceolate and often excur- vate, 5.5-6.6 mm long, pale brown with firm, dark keel, this red-ciliolate or red-fimbriolate from middle to apex. Petal blades broadly obovate, 5.5-6 mm long, yellow, apically ob- tuse and erose. Staminodia bibrachiate, the slender branches long-villous-penicillate from base to tip. Anthers oblong, emarginate and auriculate, ca. 2 mm long, on filaments ca. l mm long. Capsule narrowly ellipsoid, 4- 4.5 mm long, the placentation basal-axile (septa detaching from central axis and falling with valves). Seeds few, cylindric-fusiform, often curvate, 2.5-3 mm long, including a pale, narrowly conic appendage (separated outer integument), and with numerous pale, flattened, longitudinal ribs. — Distribution. Common in boggy rapa- teaceous savanna at medium to high eleva- tions, the Guayana Highland of Estado Boli- var, Venezuela, eastward into the Pakaraima Mountains of Guayana. This, the only known species of bispicate Xyris, unfolds its pale yellow blooms in the morning. Its dark bracts usually have incon- spicuous but often detectable dorsal areas. It may intergrade with X. decussata Gl. and with X. albescens Steyerm., both of which it strongly resembles in bract and seed char- acters. It and several other species of the Guayana Highlands hitherto considered part of a well-defined section, “Nematopus,”” are showing septate ovary and fruit, and axile placentation. Volume 75, Number 2 1988 Kral Xyris 565 ° a t C STU < 4mm | FIGURE 15. Xyris bicephala (from the type).—a. Habit sketch.—b. Leaf apex.—c. Leaf blade-sheath junction.—d. Leaf base.—e. Spikes.—f. Lateral sepal.—g. Petal blade, stamen.—h. Staminode, enlarged sector of hair.—i. Opening (lefi) and closed (right) capsule.—j. Capsule, showing one valve, placenta.—k. Style apex.—l. Seed. 566 Annals of the Missouri Botanical Garden 16. Xyris teinosperma Idrobo & Lyman B. Smith, Caldasia 6: 224. 1954. TYPE: Colombia. Vaupés: Yapoboda, 10 Dec. 1943, P. M. Allen 3195 (holotype, COL; isotype, MO). Figure 16. Sturdy solitary to densely cespitose peren- nial 5-8 dm high, the stems short and stocky. Leaves mostly spreading flabellately, 4-6 dm long; sheaths entire, in longer leaves less than V$ as long as blades, the dilated bases cas- taneous and lustrous, upwardly green or stra- mineous, keeled, tapering evenly to blades, there eligulate or with a narrow thin ligule to 3 mm long, the blades linear, flattened, straight and stiff, 4-10 mm wide, tapering slightly above middle then abruptly narrowed at apex to an incurved or erect, narrowly rounded or broadly acute, thickened tip; margins carti- laginous-thickened, pale, smooth; surfaces deep yellow-green, finely nerved, smooth. Scape sheaths much shorter than leaves, closed below, multicostate, distally keeled, open and with a stubby blade. Scapes distally ancipital, 3-4 mm wide, with 2 smooth costae making edges, surfaces yellow-green, striate, smooth. Spikes ovoid, 2-3 cm long, acute, of very many firm, tightly spirally imbricate bracts, these brown with conspicuous paler dorsal areas; sterile bracts many, the lowest much smaller than the fertile bracts, trian- gular-ovate, keeled, grading into fertile bracts, these oblong to obovate, ecarinate, 10-15 mm long, apically narrowly rounded and sub- entire. Lateral sepals free, subequilateral, lin- ear-oblanceolate, 10-11 mm long, acute, the wide keel above middle finely lacerate and/ or villosulous, aging subentire. Petal blades broadly oblong-elliptic, 1-1.2 cm long, yel- low, the broadly rounded apex erose. Anthers oblong-linear, ca. 3.5 mm long, deeply bifid and sagittate, on filaments ca. 1.5 mm long. Staminodia multibrachiate, the slender branches densely plumose with long, penicil- late hairs. Capsule narrowly ellipsoid, 8-10 mm long, the massive placenta basal, the valves dehiscing to reveal deep septa at base. Seeds several on short, bulbous funicles, lin- ear, ca. 3 mm long, including an apical coma of pale, narrow, erect squamellae ca. 1 mm long, the narrow, pale brown seed body with a few strongly raised, pale, short-squamellate ribs. Distribution. Wet, low savanna in SE Colombia, SW Venezuela, and contiguous Amazonas, Brazil. This is one of the most distinctive species of Xyris, particularly noticeable in the low savannas along the upper Rio Orinoco. Its slender, long-comose seeds are the longest known for the genus; its lovely pale yellow petal blades, unfolding in midday, form the largest known flower in Xyris. 17. Xyris lomatophylla C. Martius, Flora 24(2): 57. 1841. TYPE: Colombia. Ama- zonas: "In campis, Arara Coara, Mar- tius" (lectotype, M; phototype, GH). Fig- ure 17. Robust, stiff, hard-based, solitary to ces- pitose perennial 5-7 dm high, the stems con- tracted. Leaves spreading flabellately, 0.5-3 dm long; sheaths eciliate, cartilaginous-keeled, fully the length of the blades, abruptly dilated at very base, dark brown or castaneous, shad- ing distally to brown, narrowing gradually up- ward into blade, there with a short, narrowly triangular, erect ligule; blades flattened, twist- ed, ensiform, mostly 3-5 mm wide, narrowly acute or abruptly narrowly rounded, with a pale-cartilaginous-thickened, smooth or (fre- quently) ciliolate border, the surface smooth, finely nerved, dark green. Scape sheaths slightly shorter than leaves, twisted, proxi- mally dark red-brown, keeled, with a cartila- ginous costa, distally with a strong, short blade like that of leaves. Scape twisted, straight or flexuous, distally terete or oval in cross sec- tion, 1-1.5 mm wide, ecostate, smooth, lon- gitudinally striate. Spikes dull brown, ovoid to cylindrical, 1-3 cm long, blunt, base at- tenuate or broadly rounded, with many spi- rally imbricate bracts with large, deep brown, apically broadened dorsal areas and broad, wooly-villous margins; sterile bracts many, the lowest much smaller than the fertile bracts, Volume 75, Number 2 Kral 567 1988 Xyris FicURE 16. Xyris teinosperma (Kral € Huber 70708).— a. Habit sketch.—b. Leaf tip.—c. Leaf blade- sheath junction.—d. Leaf base.—e. Spike.—f. Lateral sepal.—g. Petal blade, stamen.—h. Stylar apex.—i. Staminode, enlarged tip of beard hair. —j. Capsule, one valve removed to show placentation. —k. Seed. 568 Annals of the Missouri Botanical Garden FIGURE 17. Xyris lomatophylla i reru & Tillett 3058) .—a. Habit sketch. —b. Leaf apex.—c. ee blade junction. —d. Leaf bas e.—e. pu — f. Fertile bract.—g. Lateral sepal.—h. Petal blade, stamen.— Staminode.—j. Stylar apex.—k. S Volume 75, Number 2 1988 Kral 569 Xyris ovate, slightly keeled, grading gradually larg- er to fertile bracts, these obovate or broadly elliptic, 7-7.5 mm long, ecarinate. Lateral sepals free, very inequilateral, oblong-cur- vate, ca. 6 mm long, blunt, lustrous-brown, thin, the wide, firm keel increasingly densely brown-ciliate from near base to tip, there fre- quently villous-fimbriate. Petal blades broadly obovate, yellow, to nearly suborbicular, ca. 5 mm long, the broadly rounded apex lacerate. Staminodia bibrachiate, the flat, narrowly tri- angular branches tipped with a dense tuft of penicillate hairs. Anthers lance-oblong, ca. 2.5 mm long, deeply bifid and sagittate, on filaments 0.5 mm long. Capsule ellipsoid, planoconvex, ca. 3-4 mm long, the placen- tation basal, the valves without septa. Seeds few, cylindrical, 2-2.5 mm long including a white, conic apiculus ca. 0.5 mm long, pale amber, longitudinally finely lined, coarsely overlain by a few prominent dark ribs. Distribution. Low, wet, sandy savanna, Amazonian southeastern Colombia through the savannas along the upper Rio Orinoco and the Rio Negro, T. F. Amazonas, Venezuela, southeastward into Pará, Brazil. Additional speci imens examined. BRAZIL. PARA: attributed to Brazil Bt Ks perd to Dr. L. B. Smith, actually Colombian. In spike character this species most resem- bles X. globosa but differs in its (usually) more elongate spikes; broader, flatter, cartilagi- nous-bordered leaves; and longer, narrower seeds. In the Venezuelan savanna along the upper Rio Orinoco it may be the dominant species. 18. Xyris contracta Maguire & Lyman B. Smith, Mem. New York Bot. Gard. 10: 33. 1963. TYPE: Venezuela. T. F. Ama- zonas: Cerro de La Neblina, Rio Yatua, summit, 1,200-1,300 m, flowers yellow; infrequent in stream bed, upper Canon Grande, 1,900 m, 11 Dec. 1957, B. Maguire, J. J. Wurdack & C. K. Ma- guire 42359 (holotype, NY; isotypes, US, VEN). Figure 18. Cespitose, thick-based, smooth perennial 2-3(-5.5) dm high, the stem short, to ca. 3 cm long. Leaves ascending, 1-2(-3) dm long; sheaths entire, less than Y2 as long as blades, lustrous at base, deep red-brown, dorsally rounded-convex, firm, narrowing gradually and keeled to blade and with a prominent, erect, broadly oblong, blunt ligule 2.5-3 mm long; blade narrowly linear, ca. 1(-3) mm wide, flat, acute at apex, the margins without border, entire, surfaces finely nerved, green. Scape sheath slightly to much shorter than leaves, proximally maroon or castaneous, opening distally, green, producing a strong blade. Scapes linear, straight, distally terete, ca. 1 mm thick, ecostate or with 1-2 very low, smooth costae, or flattened and 2-edged, to 2 mm wide. Spikes ellipsoid, ca. 1.5 cm long, acute, dull red-brown with numerous bracts in several subvertical ranks. Sterile bracts several, lowest keeled, slightly smaller than the fertile bracts and grading into them; fertile bracts oblong, 7-8 mm long, strongly rounded-folded, ecarinate, subentire, apically rounded-emarginate, when young frequently villose-ciliate apically, all with pale, subapical, somewhat indistinct, elliptic dorsal areas. Lat- eral sepals free, subequilateral, oblanceolate- linear, 7-8 mm long, straight, pale red-brown, the wide, thin keel ciliolate or papillate from middle to apex, or at apex also sparsely red- villosulous. Petal blades broadly obovate or reniform, ca. 7-8 mm long, yellow, the broad- ly rounded apex lacerate. Staminodia bibra- chiate, the branches long-penicillate. Anthers ca. 2.5 mm long, oblong, on filaments ca. 1 mm long. Capsule ellipsoid, 4-4.5 mm long, septa lacking, the placentation basal. Seeds numerous, ellipsoid-fusiform, ca. 2 mm long, including a narrowly conic, pale apiculus ca. 0.4 mm long, the body dark, translucent, longitudinally prominently several-ribbed, with occasional cross-ribs. Distribution. So far known only from the type area, re-collected there by Steyer- mark (104018, US, VEN) and on the Bra- 570 Annals of the Missouri Botanical Garden FIGURE 18. Xyris contracta. (from the type) .—a. Habit sketch. —b. Leaf tip.—c. Adaxial (inside) view of leaf-sheath junction and side view of same.—d. Leaf base.—e. Spike.—f. Lateral sepal.—g. Petal blade, stamen.—h. Staminode, enlarged sector of beard hair.—i. Stylar apex.—j. Capsule, one valve removed. — k. Seed. Volume 75, Number 2 1988 Kral 571 Xyris zilian side of the Neblina crest (Rio Cauaburi, Brazil, Maguire et al. 60483, NY). This rare plant is very distinctive in its combination of narrow, smooth leaves, dark, lustrous, ribless sheath bases, contracted leaf blade base, large ellipsoid spikes, and distinc- tive seeds. The only exceptional material is that collected by Steyermark which, while definitely this species, is a longer- and wider- leaved extreme with scapes distally flattened and 2-edged, much wider than in the type. I have not seen any further collections of this among many recent collections made by the Neblina workers. 19. Xyris seubertii Nilsson, Sv. Vet. Akad. Handl. 24(14): 51, pl. 4. 1892. TYPE: Guyana: “British Guiana, Rich. Schom- burgk n. 897” (phototype, US). Figure 19. X. pateram Heimerl, Ann. Naturhist. Hofmus. Wien 8, pl. 4, f. 1-3. 1906. TYPE: Brazil: Tamberlik s.n . (W— lost; phototype, F). Slender, solitary to cespitose, hard-based, smooth and glaucous perennial 2-7 dm high, the stems contracted. Leaves spreading fla- bellately, 0.5-2.5 dm long; sheaths eciliate, at very base abruptly dilated, castaneous to dull, dark brown, above pink or pale purple, sharply keeled, narrowing gradually to blades, often 2 or more as long as blades, at apex producing a narrowly triangular, erect or spreading ligule or tapering directly into blade; blades linear-ensiform, flattened, straight or slightly twisted, 1.5-3.5 mm wide, narrowing above middle gradually into an incurved-acute or acuminate tip, the margins thin, entire to papillose, tuberculate, or scabridulous-cilio- late, the surfaces green, smooth, finely mul- tinerved. Scape sheaths often as long as leaves, tubular and multicostate toward base, often rose or purple, ribbed, above open, keeled, with short, flat blades similar to those of leaves. Scapes straight or flexuous, twisted, subterete toward apex, 0.5-0.7 mm thick, ecostate or with 1 or more costae and striate, mostly smooth. Spikes ovoid, broadly ellipsoid or sub- orbicular, 0.8-1.2 cm long, pale brown, of many spirally imbricate bracts with distinct though narrow dorsal areas, the sterile bracts few, broadly rounded, slightly shorter and narrower than the fertile bracts but with 1 or 2 of the lowest often with dorsal areas prolonged as cusps or short blades equaling or exceeding spike; fertile bracts strongly con- vex-backed, obovate, 5-6 mm long, broadly rounded at apex, sometimes with apical, red- villosulous tuft. Lateral sepals ca. YY, con- nate, inequilateral, lanceolate, 6-6.5 mm long, blunt, slightly exserted, thin, the tan, firm keel above middle to apex increasingly dense- ly reddish villous. Petal blades broadly ob- ovate to reniform, ca. 6-7 mm long, yellow, the very broadly rounded apex wavy-erose. Staminodia bibrachiate, the branches long- penicillate. Anthers narrowly oblong, ca. 2 mm long, deeply bifid, deeply sagittate, on filaments ca. 1 mm long. Capsule obovoid, ca. 3 mm long, placentation free-central, the valves without septa. Seeds numerous, broad- ly ellipsoid to ovoid, ca. 0.3-0.4 mm long, apiculate, amber, finely striate-reticulate. Distribution. Abundant and widespread from the Gran Sabana of Estado Bolivar, Ven- ezuela, eastward into Guyana, increasingly abundant south of the Amazon Basin in the Brazilian planalto southward to Sào Paulo. This species is distinctly weedy, often a pioneer in mechanically disturbed, eroded, or burned savanna. Its glaucous foliage and rath- er pale yellow flowers, which unfold in late morning, make it particularly handsome. 20. Xyris huberi Kral & Lyman B. Smith, nom. nov. TYPE: Venezuela. T. F. Ama- zonas: Cerro Yapacana, en la sabana grande entre el Cano Cotua y el pie del cerro, 3?45'N, 66%45'W, 125 m, 7 May 1970, J. A. Steyermark & G. Bunting 103241 (holotype, US; isotype, VEN). Figure 20. X. foveolata Kral & Lyman B. Smith, Phytologia 53: 35-436. 1983, non Irmscher. X. yapacanensis Steyerm. & Lyman B. Smith, nom. nud. Short-lived, cespitose, lustrous perennial from a short, thick, subvertical rhizome, the 572 Annals of the Missouri Botanical Garden Lam <= ARS ARSS W y U a SN Z SN SS SS eG === SS FIGURE 19. Xyris seubertii (Kral 70579).—a. Habit sketch.—b. Leaf apex.—c. Leaf at blade—sheath junction.—d. Leaf base.—e. Spike. —f. Lateral sepal (pair, connate) .—g. Petal blade, stamen.—h. Staminode, beard hair.—i. Stylar apex.—j. Capsule, placentation, one valve removed, one valve outline.—k. Seed. Volume 75, Number 2 1988 Kral 573 Xyris roots slender. Leaves ensiform-linear, (7-)8- 15(-16) cm long, spreading flabellately, lon- ger than the scape sheaths; sheaths entire, carinate, the carinae ciliate-scabrid, the sides deep brown, narrowing from the dilated, cas- taneous base gradually to the blade, there the edges converging to form a triangular, linear- acute, slightly spreading ligule 1.5-2 mm long; blades equal to or twice as long as sheaths, at or somewhat twisted, strongly com- pressed, 1.5-2 mm wide, the surfaces green, tinged with brown or rusty brown, punctate (stomata depressed), longitudinally multi- nerved; apices narrowly acute, erect or in- curved, slightly thickened; margins thick- ened, yellowish, densely pale ciliate with antrorse hairs. Sheaths of scapes multicostate, twisted, carinate, the carinae ciliate, the blades short, similar to leaf blades. Scapes 1.5-2.5 dm high, straight or somewhat flexuous, slightly twisted, distally 1-1.5 mm wide, punctate, bicostate, brownish, ancipital, the edges antrorsely long-ciliate. Spikes obovoid, 6-7 mm long, obtuse, few-flowered, bracts loosely imbricate, subdecussate, with scari- ous, reddish brown, revolute, ciliate borders and large, pale brown dorsal areas; sterile bracts 2-4, the lower pair oblong-triangular, 4-5 mm long, strongly carinate, the inner pair absent or ovate-triangular, ecarinate, the backs low-rounded, l-nerved; fertile bracts narrowly ovate, narrowly rounded-folded, ca. 5 mm long, toward apex subcuculate, strong- ly papillose. Lateral sepals free, equilateral, somewhat curved toward base, lance-linear, ca. 4 mm long, narrowly rounded, acute or bidentate at apex; keel firm, entire toward base, ciliate from middle to apex. Petal blades obovate, ca. 5 mm long, pale yellow, the broadly rounded apex lacerate. Staminodia bibrachiate, the branches long-penicillate api- cally. Anthers linear-lanceolate, ca. 1.5 mm long, deeply bifid and sagittate on filaments ca. 1 mm long. Capsule narrowly obovoid, 2.5 mm long, the valves eseptate, the pla- centation basal. Seeds cylindric-fusiform, am- ber, 1.3-1.4 mm long, longitudinally spirally anastomosing-lined, with a few stronger, dark- er ribs. Distribution. Confined to low-elevation savanna along the upper Rio Orinoco and tributaries, T. F. Amazonas, Venezuela. Additional specimens examined. VENEZUELA. T. F. AMAZONAS: Dept. Atabapo, Cerro Yapacana, 3 June 1978, Buby 2030 (VEN); ca. 1 km a E del Caserio de Guar- inuma, 23 Feb. 1979, Huber 3356 (VEN); Cerro Ya- pacana, 28 Feb. 1980, Huber 4815, 4829 (US, VEN); Dept. Casiquiare, 2-3 km al SE del bajo Guasacavi, 10 Mar. 1980, Huber 5114 (US, VEN). This species is distinguished by a combi- nation of flattened, prominently thick-edged and ciliate leaves; ancipital and ciliate scapes; and hooded-tipped and reddish-margined, cil- iate bracts. The margins of the leaf sheath, while thin, are firm and terminate in a long, narrow, sharp ligule. The large pale dorsal areas and surfaces of leaf blades and scapes are all strongly punctate. 21. Xyris graniticola Kral, sp. nov. TYPE: Venezuela: Amazonas. Dept. Atures, vegetación de laja (VL) sobre aflora- miento granítico en raudal “‘pereza” en el Rio Autana, 9 Nov. 1984, F. Guán- chez & E. Melgueiro 3425 (holotype, VEN; isotypes, TFAV, VDB). Figure 21. Planta perennis, densicaespitosa, Lunas: radices gra- ongis. Folia prin- cm aminas gradatim decrescentes, ad diis re, ici rigida, erecta, i otriangulata, ad | laminae anguste lin p "o eie s tae, 0. $- pansae, integrae, decussatae, infimae 4 steriles, par in feriora lanceolata, ca. 4 mm longa, carinata, area dorsali lineari, par intima late triangulata, ca. 3.5 mm longa, area dorsali triangulata, leviter nervata; bracteae fertiles oblongae, 4-4.5 mm longae, anguste obtusae, subcon- duplicatae, anguste rotundatae, i inconspicue unicostatae, dorsum ad basin castaneo, ad apicem area in 2 ovata, viridi. Sepala lateralia libera, aequilateralia, mm longa, obtusa, x id curvata, ala carinali angusta, integra . Laminae pet longae, apice subtruncata regulariter dentatae. Staminodia bibrachiata, brachiis lon- 574 Annals of the Missouri Botanical Garden kh FIGURE 20. Xyris huberi (Huber 4815).—a. Habit sketch.—b. Leaf apex.—c. Sector of leaf midblade.—d. Leaf-sheath junction.—e. Trichomes of leaf blade margin, enlarged.—f. Leaf base.—g. Spike. —h. Fertile bracts (left and right of spike) .—i. Lateral sepals (extremes at left and right) .—j. Petal blade and stamen; stylar apex on lefi, staminode on right. —k. Enlarged tip of beard hair. —l. Capsule, one valve removed, showing basal- central placentation. —m. Seed Volume 75, Number 2 Kral 575 1988 Xyris "i fei Antherae ip Jo profunde bifidae et sa- Distribution. A “laja” plant known only mm s ca. 0.5 mm longis. Capsula 3 mm Pid cbovoidea: placenta basalis. Se- c , translucentia, atrofer- ruginea vel fusca, b aa multicostata. The plant perennial, densely cespitose, smooth; the roots slender-fibrous. Stems short or up to 3 cm long. Principal leaves erect to slightly flabellately spreading, 10-25 cm long, about as long as the scapes, longer than the scape sheaths; sheaths entire, carinate, shin- ing, pale brown, Y2 or more the length of the blades, gradually dilated toward base, grad- ually narrowing into the blades, ligulate at apex, the ligule erect, rigid, linear-triangular, up to 1.5 mm long; blades narrowly linear, twisted, somewhat flattened, 0.5-0.7 mm wide, inconspicuously few-nerved, the margin toward the base beveled-incrassate, reddish, hispidulous; tips narrowly conic, dorsally sca- brid at apex. Scape sheaths brownish toward base, shining, open, short-bladed toward apex. Scapes somewhat twisted, terete or elliptic in cross section, ca. 0.5 mm wide, ecostate, oli- vaceous. Spikes few-flowered, narrowly to broadly turbinate, ca. 4-5 mm long; bracts slightly spreading, entire, decussate, with con- spicuous dorsal areas, the lowest 4 sterile, the lowermost pair lanceolate, ca. 4 mm long, carinate, the dorsal area linear, the inner pair broadly triangular, ca. 3.5 mm long, the dor- sal area triangular, lightly nervose; fertile bracts oblong, 4-4.5 mm long, narrowly ob- tuse, subconduplicate, narrowly rounded, in- conspicuously unicostate, the back casta- neous toward base, with an ovate green dorsal area toward apex. Lateral sepals free, equi- lateral, ca. 4 mm long, oblong, obtuse, slightly curvate, the keel narrow, entire to remotely papillate. Petal blades obtriangular, yellow, ca. 2-2.5 mm long, the apex subtruncate, irregularly dentate. Staminodes bibrachiate, the branches long-penicillate. Anthers oblong, deeply bifid and sagittate, ca. 1 mm long, on filaments ca. 0.5 mm long. Mature capsule obovoid, ca. 3 mm long, the placenta basal. Seeds ovoid or ellipsoid, ca. 0.5 mm long, translucent, deep reddish brown or brown, finely longitudinally many-lined. from the type locality. This species comes, in treatments of Ven- ezuelan Xyris, nearest those perennials with discernably leafy stems, distinct dorsal areas, and comparatively low stature, such as X. frondosa Mag. & Sm. However, the few- flowered spikes, in their turbinate outline and broad dorsal areas, are strikingly similar to those of X. guianensis Steud., here differing primarily in the nearly entire-keeled, less cur- vate lateral sepal. The latter plant is also shorter-stemmed, its leaves wiry-bordered. 22. Xyris frondosa Maguire & Lyman B. Smith, Mem. New York Bot. Gard. 10: 33, fig. 18A-E. 1963. TYPE: Venezuela. Bolivar: frequent in scrub forest near Summit Camp, 1,925 m, Central Sec- tion, Chimantá Massif, 2 Feb. 1955, J. A. Steyermark & J. J. Wurdack 346 (holotype, NY; isotypes, F, US, VEN). Figure 22. Moderately low, slender, densely cespitose perennial 2-3 dm high, the stems ascending, elongate, and forming frondlike plates of leaves. Leaves flabellately spreading, mostly 15-20 cm long; sheaths firm, entire, up to 1% as long as blades, deep reddish brown, papillate, tapering evenly to blade, often cil- iate-carinate, at blade junction with an erect, narrowly triangular ligule to 1 mm long or eligulate; blades flat, often twisted, linear, 1.5— 2.5 mm wide, tapering evenly to an acute- incurved, densely pale-ciliate apex, the mar- gins ascending-ciliate with pale narrow hairs, submarginally with a deep reddish brown bor- der, the surfaces green or maroon, finely mul- tinerved, mostly smooth. Scape sheaths short- er than leaves, loosely carinate, keel ciliate, open at apex, short-bladed. Scapes slenderly linear, straight or slightly flexuous, slightly twisted, distally terete or slightly compressed, ecostate or low-bicostate, smooth. Spikes el- lipsoid, 6-7 mm long, reddish brown, several- flowered, the bracts tightly spirally imbricate and with distinct reddish brown dorsal areas, 576 Annals of th Missouri y m Garden FIGURE 21. Xyris graniticola (from the isotype) .—a. Habit sketch. —b. Leaf apex.—c. Sector of midblade.— d. Leaf sheath-blade junction (at left) ; E (at right) .—e. Spike. —f. Fertile bract.—g. Lateral os — L See h. Petal blade, stamen.—i. Staminode. apically often white-villosulous-ciliate; sterile bracts smallest, grading into the fertile bracts, triangular or ovate, acute, keeled, the fertile bracts obovate, ecarinate, 4.5-5 mm long, style branches. —k. Capsule, one valve removed. — subentire, broadly rounded apically, backs convexly rounded, the dorsal areas bisected by a low but distinct rib. Lateral sepals oblong, curvate, free, inequilateral, ca. 4.5 mm long, Volume 75, Number 2 Kral 577 1988 Xyris E era 22. Xyris ie p the type) .—a. Habit sketch.—b. Leaf apex.—c. Leaf blade-sheath junc- —d. Leaf base.—e. Spike.—f. Lateral sepal.—g. Petal, stamen.—h. Staminode.—i. Stylar apex.—j. e one valve removed, 2 basal placentation.—k. Seed. blunt, pale reddish brown with a dark, firm, the apical margin lacerate-dentate. Stami- irregularly tuberculate-ciliate or ciliolate keel nodia bibrachiate, the broad, flat branches from base to apex. Petal blades broadly ob- densely penicillate at tip. Anthers oblong, ca. ovate to suborbicular, yellow, ca. 5 mm long, 1 mm long, bifid, auriculate, on flattened fil- 578 Annals of the Missouri Botanical Garden aments ca. 1 mm long. Capsule ellipsoid or narrowly obovoid, ca. 3 mm long, the pla- centation basal, the valves without evident septa. Seeds several, on long funiculi, ellipsoid to oblong-cylindrical, ca. 1 mm long, trans- lucently deep amber, pale apiculate, with fine, longitudinal, anastomosing ribs. Distribution. Abundant in wet, high sa- vanna, the Chimantá Massif and associated systems, usually above 1,500 meters, Estado Bolivar, Venezuela. Additional collections examined (since those lished by Maguire & Smith (1963)). seii og BOLÍVAR: Macizo del Chimanta, d b Steyer mark 6907 (VDB, VEN); Chimanta, Apacara-tepui, Huber & Steyermark 6961 (VDB, VEN); seccion cna del Chi- mantá-te epui, Huber & Steyermark 7161 (VDB, VEN); Auyan- -tepui, Huber & Medina 8538 (NY, VEN); Apa- cara-tepui, Huber & Colella 8727 (NY), 8736, 8774 (NY, VDB, VEN); Churi-tepui, Huber & Colella 9000 11471 (MYF, VDB, V NY, US, VEN), 93757 (NY, US, VDB, VEN), 93992 (L, NY, US, VEN); Cerro Jaua, Steyermark 98055 (F, VEN), 109387 (NY, VEN); Cerro Guanacoco, Steyer- mark et al. 109731 (US, VEN), 109734 (US, VEN); Cerro Jaua, Steyermark 109426 (US, VDB, VEN), 109460 (K, MO, US, VEN), 109610 (US, VDB, VEN); Cerro Guaiquinima, Steyermark & Dunsterville 113173 (F, US); Murey (Eruoda) tepui, Steyermark et al. 115839 (F, MO, US, VEN). This low-growing species with its long- stemmed fronds of rusty-bordered, ciliate leaves and small spikes with distinct dorsal areas, has no near morphological neighbor within its narrow range. 23. Xyris chimantae Kral & Lyman B. Smith, Phytologia 53: 432-433, fig. la- h. 1983. TYPE: Venezuela. Bolivar: Chi- mantà Massif, central section, swampy depression in wet savanna along east branch of headwaters of Rio Tirica, 2,120 m, 12 Feb. 1955, J. A. Steyermark & J. J. Wurdack 768 (holotype, VEN; iso- types, F, NY). Figure 23. Cespitose perennial to 6 dm high, the stems either short or ascending through deep sub- strate and to 1 dm long, the bases mostly covered by scalelike old leaf bases. Roots slender, arising from lowermost nodes. Prin- cipal leaves stiff, spreading distichously, (1.8-)2-4(-4.8) dm long, longer than scape sheaths; blades narrowly linear, 3-4 times longer than the sheaths, slightly twisted, flat- tened, somewhat thickened and with thick margins, 2-2.5 mm wide, olivaceous to yel- low-brown, finely papillose-rugulose; apices abruptly incurved-acute, thickened, entire to scabrociliate; margins slightly papillose to mi- nutely scabro-ciliate; sheaths ecarinate, the broad bases firm, lustrous, castaneous to yel- low-brown, entire, narrowing gradually into blades, eligulate. Scape sheaths proximally tubular, dark, multicostate, papillose, distally open, short-bladed. Scapes slenderly lineal, straight or slightly flexuous, slightly twisted, distally subterete to oval or elliptic in cross section, ecostate to narrowly bicostate, the costae papillose to scabrid. Spikes obovoid to obconic, ca. 1 cm long, many-flowered, the bracts loosely imbricate, subdecussate, ecar- inate, without dorsal area, smooth, light to deep brown, strongly lacerate; sterile bracts several, ovate, shorter than the fertile bracts and grading into them; fertile bracts several, ovate, 7 mm long, narrowly rounded, reddish ciliate at apex when young, with the median nerve low but manifest. Lateral sepals free, subequilateral, oblanceolate, ca. 5.5 mm long, included to exserted at apex of spike, slightly curvate, acute; keel reddish fimbriociliate from middle to tip. Petal blades obovate-rhombic, ca. 6 mm long, yellow, apically narrowly rounded, erose. Staminodia bibrachiate, the branches long-penicillate. Anthers narrowly oblong, ca. 2 mm long, shallowly bifid, shal- lowly auriculate, on filaments ca. 1 mm long. Capsule cylindric-ellipsoid, ca. 5 mm long, the valves with septa from base to tip. Seeds narrowly ellipsoid-fusiform, ca. 2 mm long, including a terminal white, conic scale 0.6- .7 mm long, the seed body reddish brown, translucent, conspicuously longitudinally mul- tiribbed. Distribution. Locally abundant in wet savanna at elevations over 1,900 meters, the Volume 75, Number 2 Kral 579 1988 Xyris FIGURE 23. Xyris chimantae (from the type) .—a. Habit sketch. —b. Two leaf tips.—c. Leaf blade, sector at midblade.—d. Two views of leaf sheath—blade junction.—e. Leaf base. —f. Spike and upper scape.—g. Lateral sepal.—h. Petal blade, stamen, staminode, stylar apex. —i. Capsule, showing side views of two septa (stippled) .— j. Seed. 580 Annals of the Missouri Botanical Garden Chimantá Massif in Estado Bolivar, Venezue- la. Additional specimens examined. VENEZUELA. BOLÍVAR: Huber & Dezzeo 8592 (NY); Huber & Colella 8735 (NY, VDB, VEN); Huber et al. 10181 (MYF, VDB); Steyermark & Wurdack 486 (F, NY, US, VEN); Steyermark & Wurdack 1010 (F, NY, VEN) Steyer- mark 75870 (F, NY, VEN); Steyermark et al. 115835 (US, VEN); Steyermark et al. 115922 (MO, US, V DB); Steyermark et al. 128375 & 128380 (VDB, VEN). This species is most similar to X. lugubris Malme but is overall more slender and has spikes of narrower outline and longer seeds. 24. Xyris stenophylloides Malme, Bull. Torrey Bot. Club 58: 323. 1931. TYPE: Venezuela. T. F. Amazonas: gorge of Cano Negro, Savanna Hills, growing in tussocks, Cerro Duida, 4,000 ft., G. H. H. Tate 808 (lectotype, NY; paratype, *Savanna Hills, summit of Cerro Duida, 2 Sept. 1944, Steyermark 58211," F, NY, US, VEN). Figure 24. Slender, densely cespitose, hard-based, smooth perennial 3-4 dm high, the stems contracted. Leaves erect or ascending, to 2.5 dm long; sheaths entire, much less than Y as long as blades, the bases lustrous pale brown shading upward through yellow-brown to green, tapering from a slightly dilated base gradually to blade, there abruptly narrowed and producing an oblong, rounded or acute ligule to 1 mm long; blades narrowly linear, pale green, proximally and medially some- what flattened, ca. 1 mm wide, narrowing gradually above middle to a narrowed, bluntly conic or flattened, thickened apex, the mar- gins thick and rounded, smooth, the surfaces green, strongly 1-2-nerved, sometimes sul- cate. Scape sheaths shorter than leaves, proximally tubular, striate, lustrous brown at base, open toward apex, there producing a short, erect, cusplike blade. Scapes straight or flexuous, twisted, distally subterete, 0.5- 0.7 mm thick, with broad, low costae and shallowly sulcate, striate. Spikes broadly tur- binate, 7-8 mm high, red-brown, of several loosely subdecussate bracts without dorsal areas, the sterile ones several, the lowest much smaller, ovate, narrowly rounded, slightly convex-carinate, grading gradually to fertile bracts, these oblong, 5.5-6 mm long, apically truncate or broadly rounded, erose or lac- erate, ecarinate, slightly folded, low-papillose. Lateral sepals free, equilateral, elliptic-linear, roadly acute, ca. 5 mm long, tan, the firm narrow keel entire. Petal blades broadly ob- ovate, ca. 4 mm long, the broadly rounded apex erose. Staminodia bibrachiate, the slen- der branches long-penicillate-pubescent. An- thers lance-oblong, ca. 1.5 mm long, deeply bifid and sagittate, on filaments ca. 1 mm long. Capsule obovoid, ca. 3 mm long, the central axis extending % the fruit length, the placentation appearing axile at least at fruit base, the valves with broad, thin septa. Ma- ture seeds fusiform, ca. 1.5 mm long, pale brown, translucent, longitudinally multi- ribbed. Distribution. Known only from the Cer- ro Duida area in rocky “open” seeps above 1,000 meters, very seldom collected, the only records other than types being the following. Additional specimens examined. VENEZUELA. T. F. AMAZONAS: Jan.-Feb. 1969, M. Farinas, J. Velasquez & E. Medina 406 (NY, VEN); vertiente norte del Tepuy Duida, 16 Nov. 1982, F. Guanchez 2342 (TFAV, VDB, VEN); between Brocchinia Hills and Savanna Hills, 1,050- 1,600 m, 2 Sep. 1944, Steyermark 58211 (F, NY, US, VEN); topotype, S. S. Tillett et al. 751-67 (MYF, VEN). This species is in several characters closest to X. scabridula Steyerm. from the same massif; however, X. scabridula is harsher in foliage, has the leaves more thickened, and the lateral sepals broader and ciliate. 25. Xyris columbiana Malme, Ark. Bot. 13(3): 40. 1913. TYPE: Venezuela. Meri- da: “Jaji? (fide L. B. Smith, Caldesia 6(29): 22. 1954). The original label in- formation is “Xyris tortilis Kl. Colum- bian. Moritz," the location given as “Taji,” and was therefore interpreted by Malme as in Colombia (Moritz 1202, Herb. Berol. € 415, BM; isolectotype, L). I have seen only the isolectotype and must also assume that the original ma- terial at B is lost, while that at BM re- mains to be designated lectotype. Figure Volume 75, Number 2 Kral 581 1988 Xyris Zam FIGURE 24. Xyris stenophylloides (Tate 808, Steyermark 58211).—a. Habit sketch.—b. Lea Sector from midblade.—d. Leaf at junction idi gii and blade.—e. Leaf base, inside view, showing orientation of blade base.—f. Spike and upper portion of scape.—g. Lateral sepal.—h. Petal blade.—i. Staminode.—j Stylar apex.—k. Opened capsule showing two ce, placentation.—l. Seed. 582 Annals of the Missouri Botanical Garden Cespitose, soft-based perennial to 6 dm high, the stems short. Leaves erect or as- cending, mostly 2-3 dm long; sheaths entire, less than Y as long as blades, narrowing grad- ually from the somewhat dilated, reddish green base to the blade, there producing a broadly oblong ligule 1-2 mm long; blades narrowly ensiform-linear, 0.5-1.5 mm wide, slightly twisted, flattened but round-edged, the apex narrowly acute, conic, the surfaces variously papillose-rugulose, dull green. Scape sheaths shorter than foliage leaves, the bases tubular, multicostate, the open apex producing a short, erect blade. Scapes straight or somewhat flex- uous, narrowly lineal, terete or subterete dis- tally, ca. 1 mm thick, unicostate, the costa low but strong, smooth or papillate. Spikes broadly obovoid to broadly ellipsoid or sub- globose, 7-9 mm long, of several spirally imbricate, thin, castaneous bracts without dorsal areas, the sterile ones mostly 4-6, the lowest pair strongly keeled, ovate-triangular, 3.5-4 mm long, the inner slightly longer and broader, grading into the fertile bracts, these broadly obovate, ca. 5 mm long, broadly rounded apically, entire-margined, the backs slightly convex, ecarinate. Lateral sepals free, inequilateral, 5-6 mm long, strongly curvate, oblanceolate, pale brown with deep brown, firm keels, these ciliolate to villosulous-ciliate from ca. middle to apex. Petal blades ca. 6 mm long, elliptic, the broadly acute apex erose. Staminodia bibrachiate, the branches densely long-penicillate. Anthers oblong, ca. 2 mm long, the apex deeply bifid, the base sagittate, on filaments ca. 1 mm long. Capsule ellipsoid, 4-5 mm long; placentation free-central, the axis tending to separate into 3 branches dis- tally, the ovules very numerous. Seeds ellip- soid, ca. 0.7 mm long, translucent, finely ribbed longitudinally. Distribution. Andean paramos, appar- ently rare, Colombia and Venezuela. Additional specimens examined. COLOMBIA. BOYACA: o 25631 (US); C Lm ee al N divioria entre Boy Santander del Sur May 1972, H. García- Bs & R. Jaramillo M. 20267 E T a l km al sur de Hoyo Rico, ca. 2,600 Sep. 1948, F. A. Barkley 18A. 150 (US). Í MERIDA: Hoya del Río Capaz, bosque de San Eusebio: La Pinuela, bosque y pantamo, 2,600 m, 22 Oct. 1969, J. Cuatrecasas et al. 28152 (US). This has definite affinities with X. subulata Ruiz Lopez & Pavón but differs in having somewhat larger habit, larger, broader spikes which produce more florets, and more con- spicuous indumentum of the lateral sepal keels. 26. Xyris subulata Ruiz Lopez & Pavón, Fl. Peruv. I: 46, pl. 71. 1798. TYPE: Peru. Huanuco: Pachitea, Pillao, 1787, Ruíz Lopez & Pavón (lectotype, MA; isolectotype, presumably at F Schizmaxon distichioides Steud., Bot. Zeitung (Berlin) 14: 391. 1856. Densely cespitose, dwarf or to 5 dm high, the stems abbreviated to quite elongate and frondlike, to 1 dm long, the leaves erect to spreading flabellately, shorter than scapes or surpassing them, the sheaths entire or ciliate, mostly firm and lustrous, equaling blades in length or less than 4 as long, the blades mostly narrowly linear, the tips incurved-acute to blunt, the margins thin or incrassate, entire to papillate-scabrid or ciliolate, the surfaces usually dark green, smooth to (toward base) papillose-rugulose. Scape sheaths tubular proximally, open and carinate distally, the blades short to much elongated as in leaves. Scapes straight or flexuous, twisted, narrowly lineal to filiform, terete to somewhat com- pressed toward apex, ecostate to bicostate, smooth or papillate or ciliolate-scabrid along costae. Spikes mostly narrowly oblong to ovoid, mostly blunt, 0.5-1 cm long, the bracts de- cussate, thin, carinate to ecarinate, entire, deep olive brown, reddish brown or (mostly) near black, the sterile bracts usually 4, the lowest pair keeled, ca. 1⁄4 length of spike or less, the inner pair broader, slightly longer and ecostate, the fertile bracts mostly 2-4, broadly ovate, apically narrowly or broadly rounded, the backs papillose to smooth, fold- ed-rounded. Lateral sepals mostly oblong-lin- ar, free, subequilateral, mostly navicular, 3.5-6 mm long, narrowly to broadly acute, Volume 75, Number 2 583 1988 FIGURE 25. Xyris columbiana (Cuatrecasas et al. 28152) .—a. Habit sketch. —b. Leaf apex.— c. Leaf blade at junction with sheath. —d. Spike.—e. Lateral sepal. —f. Petal blade, stamen.—g. Staminode, enlarged sector of beard hair. —h. Stylar apex.—i. Capsule, one valve removed. 584 Annals of the Missouri Botanical Garden with or without keel, entire to ciliolate or papillate along the crest or back medially. Petal blades broadly ovate to nearly orbicular, yellow, 3.5-5 mm long the broadly rounded apex erose or entire. Staminodia bibrachiate, branches long-penicillate. Anthers oblong, .6-2 mm long, the connective broad, on filaments 1-2 mm long. Capsule firm-valved, ellipsoid, 2-4 mm long, the placentation bas- al-central, the funicles elongate, the valves without septa or producing these at very base. Seeds ovoid to fusiform-cylindric, 0.5-1 m long, prominently longitudinally ribbed or al veolate-reticulate. This species has at least five marked va- rieties, the complex ranging widely along the "young" western cordilleras, from Costa Rica south to Chile, always in the high, cool to cold, “páramo.” Only two of the varieties are known definitely to occur in the area of this treatment, but the key below includes the type variety which may yet be found there. KEY TO SUBSPECIES OF XYR/S SUBULATA la. Edges of leaf sheaths variously ciliate; stems usually elongate. The plants with leaves much shorter than scapes; seeds ca. 1 mm long; lateral sepals ca. 9 mm long E. 6 subulata var. subulata . The plants dwarf, with leaves ur equa ing or exceeding scapes; . long; lateral vi 3.5-4 m 26B. X. pias var. C lb. pus of leaf eati. entire; stems mostly con- cted 26C. X. subulata var. acutifolia to c 26A. Xyris subulata Ruiz Lopez & Pavón var. subulata. Figure 26A. Plants 10-30 cm high, the stems often elongate, with overlapping leaves forming fronds. Leaves flabellately ascending, loosely imbricate, 1⁄4—2⁄4 as long as scapes, the lus- trous, pale red-brown sheaths 42 as long as the darker blades or longer, spreading arach- noid-ciliate, tapering gradually from the broad, clasping bases into blades, eligulate, the blades narrowly linear to filiform, mostly 0.5- wide, pale to deep green or red-green, com- pressed, flat to somewhat thickened, often terete toward apex, the tips terete, narrowly conic but blunt, the ventral edge a broad, papillose-rugulose band, the dorsal edge smooth or papillose, the surfaces papillose or smooth. Scape sheaths shorter than or slightly overtopping principal leaves, with strong, in- curved-tipped blades similar to leaves. Scapes filiform, straight or flexuous, slightly twisted, terete below, subterete toward apex, or slight- ly compressed and bicostate, 0.3- wide, the surfaces mostly olive, smooth or more often) tuberculate-papillose at least on costae. Spikes mostly ovoid to lance-ovoid, 2—4-flowered, 5-7 mm long, pale brown to near black, the bracts thin, loosely imbricate, decussate, entire or erose, without dorsal areas, the sterile pairs 2, the lowest pair short- est, ovate, carinate, the inner pair mostly broader, slightly longer, less carinate, the low- est pair of fertile bracts ecarinate but folded, much longer than the sterile bracts, ca. 4-5 mm long, entire, aging lacerate, dull or lus- trous. Lateral sepals free, very thin, sube- quilateral, ca. 5 mm long, navicular, the mid- zone nearly without keel, entire. Petal blades broadly obovate to nearly suborbicular, ca. 5 mm long, yellow, broadly rounded, subentire. Anthers ca. 1 mm long, oblong, deeply bifid at apex, the filaments ca. 1.5 mm long. Stam- inodia bibrachiate, the flattened, narrow branches elongate, long-penicillate. Capsule firm-valved, brown, ellipsoid, ca. 3 mm long, placentation basal, funiculi elongate. Seeds fusiform-cylindric, ca. 1 mm long, caudate, red-brown, translucent, longitudinally irreg- ularly multiribbed. — Distribution. South America, Andean paramos, rare in Colombia, increasingly fre- quent, Ecuador south through Peru. For Co- lombia I have only one certain record. ional specimen examined. COLOMBIA. Net. Páramo Frontino Cerro de Campanas, Es- peletia paramo, 3,650-4,290 m, clumps of ca. 20 plants, 58 Oct. 1976, J. D. Boake & J. B. McElroy 287 (U). The long “stems” of this and related taxa may in part be a result of the deep humus deposits in which dense clumps of this sort of Volume 75, Number 2 Kral 585 1988 Xyris imm FIGURE 26A. Xyris subulata var. diga (drawn from several Peruvian onn —a. Habit sketch. — b. Leaf apex.—c. Short sector of leaf blade.—d. Leaf sheath-blade junction. — —f. Spike.—g. Back (abaxial) view, two fertile bracts.—h. ¡pe sepal.—i. Petal blade, stamen. L: LM —k. Stylar apex.— l. Capsule.—m. Seed. plant are rooted, this “burying” resulting in 26B. Xyris subulata var. breviscapa Id- significant elongation of internodes compa- robo & Lyman B. Smith, Caldasia 6: rable to that in X. witsenioides Oliv. and 220, fig. 16. 1954. TYPE: Colombia. Pu- related species. tumayo: alta cuenca del Rio Putumayo, 586 Annals of the Missouri Botanical Garden filo de la Cordillera entre El Encano y Sibundoy, paramo de San d de Bordencillo, 3,250 m, 4 Jan. 1941, J. Cuatrecasas 11744 (holotype, 1 iso- type, COL). Figure 26B. Densely cespitose perennial, tufted as in moss, to 1 dm high, usually much lower, the stems ascending from fascicled rhizomes, forming frondlike plates of leaves and leaf bases. Leaves 2.5-10 cm long, loosely im- bricate, ascending-flabellate, the sheaths as long as blades or slightly longer, lustrous pale brown, smooth-keeled, low-costate, tapering gradually from broad, clasping bases into the blades, eligulate, the blades filiform to nar- rowly linear, flattened, ca. 0.5 mm wide, the apex subterete, slender-incurved, smooth, the ventral margin slightly thickened, pale, pa- pillose-rugulose, the surface olive, with a few low nerves, usually smooth. Scape sheaths slightly longer than or slightly shorter than principal leaves, with elongate, incurved blades as in leaves. Scapes to 1 dm high, overtopping leaves or overtopped by leaves, filiform, 0.3— 0.4 mm thick distally, there terete or slightly flattened, usually bicostate, the costae sca- bridulous, the surfaces olivaceous, usually papillate. Spikes narrowly to broadly ovoid, pale brown to dark brown, 4-7 mm long, the bracts decussately arranged, loosely imbri- cate, thin, the sterile of 2 pairs, the lowest pair broadly ovate-triangular, keeled, ca. 2 mm long, the inner pair ovate, carinate only toward the apex, ca. 3 mm long, the fertile bracts 2(-4), oblong-ovate, 3.5-4 mm long, the apex entire or erose, scarious, folded or carinate. Lateral sepals free, subequilateral, oblong, 3.5-4 mm long, the apex obtuse, nearly without keel, this a darker thickening medially, entire. Petal blades broadly elliptic to broadly obovate, yellow, 3.5-4 mm long, subentire, apiculate, erose. Anthers oblong, 0.6-0.7 mm long, deeply bifid, on filaments ca. 1 mm long. Staminodia bibrachiate, long- penicillate. Capsule broadly ellipsoid or ob- ovoid, brown, ca. 2 mm long, the placentation basal, the funiculi elongate. Seeds ca. 0.5 mm long, ovoid, short-apiculate, rather coarsely ridged and cross-ridged, this irregularly re- ticulate. Distribution. High grassy paramos of the Andes, Colombia south into Peru and Bolivia. Additional specimens examined. BOLIVIA: ide eal, Cocopunco, 10,000 ft., 24-29 Mar. H. H. Tate 321 (NY); Tolapampa, 25 Sep. E 10, 000 ft, R. S. Williams 1638 (NY —a putative isotype of X. cryptocarpum Rusby, which appears to be a nom. nud.). COLOMBIA. ANTIOQUIA: Páramo Frontino, near Llano Grande, Espeletia páramo, 3,450 m, Boeke & McElroy ae (NY, U). azuay. Cordillera Du alrededores del Páramo de Patococha entre Gualaceo y Limón, 6-7 Au 1969, 3,400-3,450 m, m 8626 (NY). CAUCA: Macizo Colombiano, Valle de Las Papas, alrede- dores de Valencia, ca. 3 km from Colombiano, Páram abierto y bosque Filo de La Cuchilla que ir por el sur La Laguna de La ip rens 3,450 m, 5 Sep. 1958, Idrobo et al. 2967 (VEN). S Millstraek Amaluza- Balslev 9680, 9707 (NY). PERU. CUSCO: Cruces, Cerro de Cusilluyoc, 3,800-3,90 1925, Pennell 13886 (F, NY) 0 m, 3 May This variety is distinguished from the oth- ers primarily by its low habit, seeding spikes often overtopped by leaves, and distinctly shorter, coarsely reticulate seeds. 26C. Xyris subulata var. acutifolia Hei- merl, Ann. k. k. Naturh. Hofmus. Wien 21: 63. 1906. X. acutifolia (Heimerl) Malme, Ark. Bot. 13(3): 40. 1913. TYPE: Colombia: Depto. Cundimarca, “Bogotá, Goudot” (presumably at W). Figure 26C. Cespitose glabrous perennial, the stems contracted to slightly elongated. Leaves erect or slightly spreading, to 2 dm long; sheaths mostly 1⁄2 as long as blades or less, the bases shining, pale brown to brown or tan, mostly narrowing gradually into blades, eligulate or with narrow, scarious ligules to 2 mm long, entire; blades filiform to narrowly linear, 0.5- mm wide, slightly to very flattened, often twisted, narrowly acute apically, incurved, the margins smooth to papillate-scaberulous or Volume 75, Number 2 Kral 587 FIGURE 26B. Xyris — var. breviscapa (Barclay & PAS 5915) .—a. Habit sketch.—b. Leaf apex. — c. Midsector of leaf blade.—d. Leaf blade-sheath junction.—e. Leaf.—f. Spike.—g. Fertile bract.—h. Lateral sepal.—i. Petal Sande stamen.—j. Staminode.— k. RAIN two valves sasa to show basal placentation.— l. Seed.—m. Stylar a 588 Annals of the Missouri Botanical Garden pale-ciliolate, the surfaces deep green, strong- ly or weakly nerved, smooth to papillose-ru- gulose, the latter particularly toward base. Scapes to 3 dm high, straight or flexuous, twisted, narrowly lineal, 0.5-1 mm thick, dis- tally terete or oval, ecostate to commonly bicostate, the surfaces smooth to papillose- rugulose, the costae smooth to scabrid. Scape sheaths shorter than leaves, with short, erect blades. Spikes narrowly oblong to ovoid, most- ly blunt, 0.5-1 cm long, the bracts decussate, thin, mostly ecostate, entire, deep olive brown, reddish brown or (mostly) near black, the sterile bracts usually 4, the lowest pair strong- ly keeled, ca. % spike length or less, the inner pair broader, slightly longer and ecostate, the fertile bracts 2-4, broadly ovate, narrowly or broadly rounded apically, the backs papillose to smooth, folded-rounded. Lateral sepals ob- long-linear to elliptic-linear, free, subequilat- eral, mostly navicular, 4-6 mm long, nar- rowly to broadly acute, with or without keel, entire to ciliolate or papillate along the crest or back medially. Petal blades mostly ovate, 4—5 mm long, pale yellow, narrowly rounded apically, coarsely erose. Staminodia bibra- chiate, with broad, flat branches apically mul- tipenicillate. Anthers oblong, ca. 2 mm long, blunt, widely separated by broad connective on filaments ca. 2 mm long. Fruit ellipsoid, ca. 3-4 mm long, the placentation basal- central, the funicles elongate; capsule valves without septa or with septa only at base. Seeds ellipsoid-fusiform, ca. 0.7-1 mm long, deep amber, strongly longitudinally multiribbed. Distribution. Páramos, high mountain bogs, western cordilleras, mostly over 2,000 meters high, Costa Rica south through the Andean chains to Peru, locally abundant. Selected additional igi a examined. COLOMBIA. : Páramo Frontino Campanas, Boeke 287 (NY). BOYAC Á: Sierra Nevada del Cocuy, Páramo (ned Cleef 10004 (COL, U, US, VDB). CAUCA: Valle de Las Papas, near Valencia, Core 999 (GH, NY); Cundimarca, páramo entre Cógua y San Cayetano, Laguna Verde y alrededores, 500 m al N M A La Laguna Verde, Cleef 6240 (COL, U, VDB). META: Páramo de Sumapaz, Hoya de La Quebrada Sitiales, alles Pan- tanoso 0.5 km SW de La Laguna La Primavera, Clee 1050 (U, US, VDB). HUILA: -Balsillas" on Rio Balsillas, Rusby & Pennell 775 (GH, NY); Norte de Santander, Paramo del Hatico, en route from Toledo to Pamplona, Killip & Smith 20620 (GH, PH, NY). eus RICA. CARTAGO: bog in cloud forest along Pan Am. Highway near Cerro Las Vueltas, Holm & Iltis 444 jx GH, MO, , U, US). LIMÓN: Cordillera de n Valle de Silencio, along Rio bug 1.5 iar km W of Costa Rican- anamanian bor Davidse et a; 2857 (MO, VDB, VEN). s SAN JOSÉ; cloud forest area N of Cerro e La Muerte, Cordillera de Talamanca, et al. 24146 (F, NY, US). PERU. 7714 (MO, VDB). VENEZUELA. ANZOATEGUI: Cerro Peonia above Sta. Cruz, headwaters of Rio Manantiales, E of Bergantin, Steyermark 61697 (F, NY, US, VEN). LARa: trail from Hunocaro to Buenas Aires (Caserio) below Pár- amo Las Rosas, Liesner et al. 8118 (MO, VDB). MERIDA: Páramo de Sto. Domingo, A. Jahn 1102 (GH, MO, NY, US, VEN); San Raphael, E. Reed 854 (US). sucRE: Cerro Turumuguire, Steyermark 62624 (F, MO, NY, US, VEN TACHIRA: Páramo de Tama, cerca de la frontera Colombo- Venezolana, Steyermark & Dunsterville 98625 (F, NY, VEN). TRUJILLO: Su-Oaramo y páramo de Guara- macal, F. Ortega 2654 (MO, PORT, NY, VDB). c This is the most abundant variety, there- fore the most collected, and is usually distin- guished from the others by its generally short- er stems, paler brown sheaths, and particularly its entire (vs. long-ciliate) leaf-sheath margins. 27. Xyris valdeapiculata Kral, sp. nov. TYPE: Venezuela. T. F. Amazonas: Dpto. Rio Negro, Valle de Titirico N of Pico Phelps in Cerro Neblina, ca. 0%56'N, 65?58'W, ca. 2,200 m; peat bog inter- spersed with shrub and low, rocky but wet ridges, 1 Dec. 1984, fls. open in A.M., R. Kral 71919 (holotype, VEN; isotypes, F, MO, NY, US, VDB). Figure 27 nta perennis, caespitosa, glabra, gracilis. Radice Eum Caules breves. Drs aaa leviter scan vel flabellate Sa Ls 8- onga, vaginis scaporum tegrae, .. ersus jai ati pallide leviter dilatatae, multicostatae, in m een tes et carinatae, a apicem eae a laminae anguste linearia, 2-3-plo vagines longiores, 1-2.5 mm latae, compressae, olivaceae, lon- gitudine leviter multinervosae; apices incurvato-acuta, le- viter incrassata; margine tenuis, retrorse scabrociliatae, natae) ca. 10 mm longae; bracteae sine area dorsali, laxe Volume 75, Number 2 Kra 1988 Xyris C AN © O SO QQ SSS o m rs - OS - FIGURE 26C. Xyris subulata var. A (Steyermark & 3 paci veld —a. Habit sketch.—b. Leaf apex.—c. Lea T ae a x juncti —d. Leaf base.—e. Spi upper — f. Lateral sepal.—g. Petal Sta i. St tylar apex.—j. Capsule, 7. Prae aver placentation.—k. Seed. blade, s Sed —h. minode.— 590 Annals of the Missouri Botanical Garden ———cဠJi n + ue n qup Mlinar IMN m I ". FIGURE 27. Xyris valdeapiculata d. d GET ira junction.—e. Leaf bas - Spik ke.—g. Staminode. stamen. —J. .—c. Leaf blade, midsector. — ferte act.—h. Lateral sepal.—i. Petal blade, —k. Capsule hind free- eum placenta Mu oblique view of valve. —l. Stylar SEHE, Pier Volume 75, Number 2 1988 Kral Xyris 591 Fi een imbricatae, leviter expansae, integrae, infimae 5-6 steriles, par inferior anguste triangulata, ca. 4 mm longa, pcm carinata, scabrociliata, 2-3-plo “ha s bre- viores, intimae dai dcm lanceolatae, carinatae, sque ad 5 mm longae, in fertiles abrupte transientes; bracteae Puit late eer ba usque ad 8 mm longae medio ad apicem valde carinatae, integrae demum laceratae). Sepala lateralia libera, subaequilateralia, 7-8 mm longa, acuminata, leviter curvata, ala carinali an- gusta, incrassata, integra vel leviter scabrida. Laminae petalorum obtriangulatae, luteolae, ca. 5 mm longae, apice laceratae. Staminodia bibrachiata, brachiis disparibus, ob- longis, penicillatis. Antherae oblongae, profunde bifidae pallide brunneola, opaca, longitudine spiraliter multicos- tata, apiculo albosquamoso, ca. 0.5 mm longo Plants perennial, cespitose, smooth, slen- der, the roots slender-fibrous. Stems short. Leaves slightly to flabellately spreading, 8- 14 cm long, longer than the scape sheaths; sheaths entire, ecarinate toward base, lustrous pale brown, slightly dilated, multicostate, gradually narrowing, carinate into the lea blades, at apex short-ligulate; blades narrowly linear, 2-3 times longer than the sheaths, 1- 2.5 mm wide, flat, olivaceous, finely multi- nerved; tips incurved-acute, slightly thick- ened; edges thin, retrorsely scabrociliolate, pale. Scapes slightly twisted or straight, 20- 35 cm high, ca. 1 mm wide, subterete toward apex but prominently bicostate, the costae densely pale scabrociliate. Spikes few-flow- ered, brownish, narrowly turbinate, becoming broadly turbinate, ca. 10 mm long; bracts without dorsal area, laxly spirally imbricate, slightly spreading, entire, the lowest 5-6 ster- ile, the lowest pair narrowly triangular, ca. 4 mm long, strongly carinate, scabrociliate, 2— 3 times shorter than the fertile bracts, the inner bracts slightly longer, lanceolate, cari- nate, up to 5 mm long, grading abruptly into the fertile bracts; these broadly lanceolate, up to 8 mm long, strongly carinate from mid- dle to tip, entire (aging lacerate). Lateral se- pals free, subequilateral, 7-8 mm long, acu- minate, slightly curvate, the keel narrow, thick, entire to slightly scabrid. Petal blades obtriangular, yellow, ca. 5 mm long, apically lacerate. Staminodia bibranched, the branch- es unequal, oblong, penicillate at tips. Anthers oblong, deeply bifid and sagittate, ca. 1 mm long, on filaments ca. 1 mm long. Mature fruit narrowly ellipsoid, 4.5 mm long, acute, the valves with septa toward base; placenta free-central, with thickened funicles. Seed cy- indric, 2 mm long including apiculus, pale brown, opaque, longitudinally spirally ribbed, the apiculus a pale scale ca. 0.5 mm long. Distribution. locality. Known only from the type In general dimensions and in strongly two- edged, ciliate scape, the new species resem- bles X. bicostata Maguire & Lyman B. Smith but lacks a dorsal area. Also, the lustrous brown leaf sheaths are entire rather than long- ciliate. In the locale, it is associated with an abundance of X. xiphophylla Maguire & Lyman B. Smith, X. bicostata Maguire & Lyman B. Smith, and X. atriceps Malme. Its petals unfold in late morning. As in several of the high-tepui endemics of the Guayana Highland, X. valdeapiculata has an elon- gated (2 mm) seed, the name for the species chosen to reflect the long, pale, thin apiculus. 28. Xyris tatei Malme, Bull. Torrey Bot. Club 58: 324, pl. 24, fig. 1A. 1931. TYPE: Foncea. T. F. Amazonas: moist slopes of the Savanna Hills, summit of Mount Duida, 4,400 ft., G. H. H. Tate 835 (lectotype, NY; phototypes, F, NY). igure 2 Robust, cespitose, hard-based perennial 6- 9 dm high, the stem stout, contracted, or up to 9 cm long. Leaves spreading flabellately or ascending, 2.4—6 dm long; sheaths eciliate, usually 1⁄4 as long as blades, or shorter, broad at very base, lustrous deep red-brown or cas- taneous, thence upward shading to pale green, narrowing gradually to blade, there with an erect, pale, triangular ligule 1-2 mm long; blades linear, flattened, 3-7 mm wide, nar- rowing gradually above middle, abruptly nar- rowed at apex, incurved-acute, the tip some- what incrassate; margins slightly to very thickened, pale, smooth or papillose; surfaces green, multinerved, smooth. Scape sheaths 592 Annals of the Missouri Botanical Garden ETE. d z > ARIS AAA AA exacte rt + D LE c£) =: == FIGURE 28. d. Leaf base.—e. Spike.—f. Sector of upper part of scape.—g. Fertile bract. —h. Lateral sepal.—i. Petal blade, stamen.—j. Staminode, enlarged tip of beard hair.—k. Stylar apex.—l. Capsule, spread at maturity to show basal-central orientation of placenta, septa of valve bases.—m. Seed. Xyris tatei (Tate 835) .—a. Habit sketch. —b. Leaf apex.—c. Leaf at blade-sheath junction. — Volume 75, Number 2 1988 Kral 593 Xyris shorter than leaves, tubular and costate at base, keeled, open toward apex, producing a short, erect blade. Scapes flattened distally, 2-3 mm wide, the edges comprising 2 densely papillate pale costae, the sides sometimes with 1-2 more lower costae. Spikes globose to hemispherical or broadly turbinate, 1-1.5 cm high, the base attenuate, with many firm, lustrous brown, imbricate bracts in nearly ver- tical rows and without dorsal areas; fertile bracts numerous, the lowest bracts much smaller than the fertile bracts, ovate-trian- gular, carinate, grading into the fertile bracts, these broadly oblong to obovate, 6-8 mm long, broadly rounded at apex, with reddish scarious (rarely also reddish ciliolate) borders or subentire, the backs low-convex with a low, pale median nerve toward apex. Lateral sepals free, equilateral, narrowly oblong-elliptic, ca. -6.5 mm long, acute, reddish brown, the firm keel red-ciliolate from middle to apex. Petal blades obovate, ca. 5-5.5 mm long, the narrowly rounded apex and margins erose. Anthers oblong-lineal, 2-2.2 mm long, deeply bifid and shallowly auriculate, on erect fila- ments ca. 1 mm long. Staminodia with branches rebranched, the flattened ultimate branchlets long-penicillate. Capsule narrowly ellipsoid, acuminate, ca. 4-5 mm long, the placentation appearing basal-central, but each valve with a strong septum at base. Seeds few, cylindrical-fusiform, ca. 1.5-2 mm long, dark amber with apex conic and pale, the body finely and prominently longitudinally multiribbed. Distribution. Known only from high, wet, rocky savanna, Cerro Duida, at eleva- tions over 1,000 meters. Additional specimens examined. VENEZUELA. T. F. AMAZONAS: Cerro Duida (all specimens): Rio Cunucunuma, m high, occasional along Culebra Creek at 1,100 m, Maguire, Cowan & Wurdack 29511 (NY); moist slopes of Savanna Hills, summit of Mt. Duida, 4,400 ft., Tate 778 (NY) This taxon resembles X. albescens Stey- erm. superficially, but that species has its leaf- blade edges and its scape edges prominently ciliate with pale hairs, while its seeds are longer and much narrower. 29. Xyris melanovaginata Kral & Ly- man B. Smith, sp. nov. TYPE: Venezuela. Bolivar: Dist. Piar, Macizo del Chimanta. Sección oriental del Chimanta-tepui, ca- beceras del afluente derecho superior del Rio Tirica (“Caño del Grillo"), 5°18’N 62°3'W, ca. 2,450 m, 7-9 Feb. 1983, O. Huber & J. A. Steyermark 7136 (holotype, VEN; isotype, VDB). Figure 29 Planta robusta, perennis, caespitosa, 5-10 dm alta, basibus firmis, per bases persistentes veternas foliorum obtectis. Caules incrassati, varie elongati (basibus in sub- stratio profunde elongati). Folia principalia rigida, disticha, ria, ca. 2 mm E laminae vel late s 7 mm longae, = rotundatae integrae vel ero ecar- Inatae, leviter con area dorsali. Sepala isterdin libera, subaequilatera, pri linearia, ca. 5 mm longa, obtusa, vel acuta, pallide brunneola; ala carinali firma, bibrachiata, hrachiis longipenicillatis. Antherae latae ob- longae, ca. 1 mm longae, filiis ca m longis. Capsula anguste ellipsoidea vel obovoidea, 4- 5 mm longa; pla- centae axiales. Semina pauca, ellipsoidea, apiculata, 1.3- mm longa, porphyrea, translucida, longitudine leviter multicostata. Robust, cespitose perennial 5-10 dm high, the stems stout, contracted or elongated to 2 dm, producing frondlike plates of leaves. Leaves spreading flabellately, 2-3 dm long, sheaths entire, mostly 42 as long as blades or shorter, firm, castaneous, lustrous, tapering gradually from broad bases into blades, there 594 Annals of the Missouri Botanical Garden FiGUR i š. D oa ë; the type) .—a. Habit sketch. —b. Leaf apex.—c. Leaf at ca. mid- blade. — Leaf bas - Two spikes.—f. Fertile bract.—g. Lateral sepal.—h. E dent pea blade and on, stylar apex, en phos Pol sket a of beard hair, staminode.—i. Capsule: above, cross section showing dat. of. septa; below, longitudinal view showing placentation, septa shaded.—j. Seed. Volume 75, Number 2 1988 Kral 595 Xyris producing a scarious, narrowly triangular lig- ule ca. 2 mm long; blades flat, gladiate-linear, 4-6 mm wide, the apex incurved-acute, in- crassate, the margins usually with sub- marginal narrow red-brown bands, the edges cartilaginous, smooth or ciliolate or pale-pi- lose-ciliate, the surfaces deep green, finely multinerved, smooth. Scape sheaths shorter than leaves, proximally castaneous and tu- bular, keeled and open above, producing a short, flat blade. Scapes strongly flattened, distally ancipital, 2-4 mm wide, the edges smooth to variously pale-ciliate. Spikes ovoid to ellipsoid or cylindric, 1-3 cm long, blunt, dark red-brown or olivaceous-castaneous, at- tenuate, of many spirally imbricate bracts, the sterile bracts many, broadly ovate to ob- ovate, ecarinate or distally carinate, grading gradually larger into fertile bracts, these ob- long to broadly obovate, 5.5-7 mm long, broadly rounded, entire or erose (lacerate in age), ecarinate and but slightly convex-backed, without dorsal area. Lateral sepals free, sub- equilateral, oblong-linear, blunt or acute, pale brown with firm ciliolate or pilose-ciliate keel. Petals with blades obovate, ca. 6 mm long, yellow, the broadly rounded apex erose. Stam- inodia bibrachiate, the branches at apex long- penicillate. Anthers broadly oblong, ca. | mm long, the connective broad, the filaments ca. 0.5 mm long. Capsule narrowly ellipsoid to obovoid, dorsiventrally compressed, 4-5 mm long, the placentae axile, the valves with prominent septa. Seeds ellipsoid, apiculate, 1.3-1.5 mm long, reddish brown, translu- cent, longitudinally finely anastomosing- ribbed. Distribution. So far known only from the Chimanta Massif, there much collected in recent years. Additional specimens examined. VENEZUELA. BOLÍVAR: sector centro-noreste, 26-29 Jan. 1983, Huber & Steyermark 6950 (VDB, VEN); en ladera del Rio t Steyermark & Wurdack 347 (NY, US, VEN); island in Rio Tirica above Middle Fall below Summit Camp, Stey- ermark & Wurdack 487 (NY, US); upper shoulder of O SE facing upper shoulder, Steyermark 759 (F, NY, VEN); W part of Abacapá-tepui, 13 Apr. Lota, Steyermark 74851 (F, NY, VEN); sector septen- trional, Murey-tepui, 24 Feb. 1978, Steyermark et al. 115769 (US, VEN), 115836 (MO, US, VEN); cumbre del Cerro Apacará, 8 June 1946, F. Cardona 1586 (VEN); centro noreste sector, 26-29 Jan. t ermark et al. 128024, 128083, 128131, 128224(VDB, VEN); Amuri-tepui, 2-5 Feb. 1983, Steyermark et al. 128459; same locality, 6 Feb. 1983, a 128775 (VDB, VEN); seccion A 9 Feb. 1 , Steyermark et al. 128883 (VDB, V Ze sltplanicie ororena del copan-tepui, 14-1 , Steyermark et al 129866 (MYF, VEN, VDB): Churi-tepui, 3 Feb. 1953, Wurdack 34307 (NY, US). This species, obviously abundant on the summits of the Chimanta Massif, and which has been in collections for many years, has been identified variously as X. decussata Gl., X. albescens Steyerm., X. tatei Malme, and otherwise. It is often very long-stemmed and is frequently a “plate” former like X. wit- seniodes, X. ptariana, X. frondosa, and oth- er species. While it has ancipital scapes that are often ciliate as in X. decussata, it lacks the dense, continuous bands of reddish hairs of that species. Its elongate spikes are dis- tinctive. While its scape and leaves are often pale-ciliate, its leaf blades are much longer than its leaf sheaths, its spikes are longer than in X. albescens, and its lateral sepals and seeds are shorter. Thus, more by a combi- nation of characters held in part by different species, it is unique. 30. Xyris culmenicola Steyerm., Field- iana, Bot. 28(1): 1951. TYPE: Venezuela. T. F. Amazonas: Brocchinia Hills, 1,700- 1,980 m, Cerro Duida, 1 Sep. 1944, J. Steyermark 58198 (holotype, F; iso- types, GH, NY, US). Figure 30. Cespitose, robust, hard-based perennial 5— 7 dm high, the stems contracted. Leaves erect to somewhat spreading, 2-3.5 dm long; sheaths entire, less than 42 as long as blades, deep, lustrous, reddish brown or castaneous at base, tapering gradually and keeled from broad base to blade, there essentially eligu- late, the blades flat, linear, 2.4-5 mm wide, tapering very gradually from ca. midblade or 596 Annals eae BONN Garden S “< fnm = AAC ¿RPP iA swe z. £ = < a EZ. qaz a no Y FicuRE 30. Xyris culmenicola (from holotype).—a. Habit sketch.—b. Leaf apex.—c. Sector of leaf, mid- blade.—d. Leaf base.—e. Spike. —f. Lateral sepal.—g. Petal, stamen.—h. Staminode, enlarged sector of beard hair.—i. Stylar apex.—j. Capsule, one valve removed; valve, showing septum.—k. Seed. Volume 75, Number 2 1988 Kral 597 Xyris below to a narrowly acute, erect or curvate apex, the tip slightly callused, the margins pilosulous-ciliate, the surface prominently multinerved, smooth, often with alternating broad, red-brown bands, 2 of these making submarginal borders. Scape sheaths much shorter than leaves, loosely tubular and cas- taneous below, open above, green, carinate and ciliate-keeled, with a short, erect, ciliate, lade. Scapes straight, slightly twisted, dis- tally slightly compressed but in cross section oval or oblong, smooth, low-ribbed. Spikes dark brown, ovoid to broadly ellipsoid or broadly obovoid, 1.5-2 cm long, blunt, at- tenuate-based, of many firm, spirally imbri- cate, entire bracts without dorsal areas. Ster- ile bracts many, the lowest by far the smallest, keeled, ovate-triangular, grading gradually to fertile bracts, the fertile bracts obovate to oblong, 7-8 mm long, entire (becoming some- what lacerate), often emarginate, the apex broadly to narrowly rounded, the backs ecar- inate, slightly convex, not folded. Lateral se- pals free, subequilateral, linear-oblanceolate, ca. 7-7.5 mm long, broadly acute, pale brown, curvate, the strong, dark keel rusty-ciliate from middle to apex. Petal blades broadly obovate, ca. 6 mm long, yellow, the broadly rounded apex erose. Staminodia quadribra- chiate, the branches long-penicillate. Anthers oblong, ca. l. m long, short-bifid, auric- ulate, on filaments ca. 1 mm long. Capsule cylindric, ca. 4 mm long, the placentation apparently basal but with 3 strong septa in- truding except at ovary apex. Seeds few on long funiculi, narrowly cylindric-fusiform, ca. 2 mm long, pale brown, translucent, including a translucent, pale, conic appendage ca. 0.5 mm long (outer integument), the body lon- gitudinally multiribbed with a few coarser ribs produced by the outer integument. Distribution. Grassy, rocky, wet savan- na, summit elevations along Cerro Duida and Cerro Marahuaca, T. F. Amazonas, Vene- zuela. Additional specimens examined. VENEZUELA. T. F. AMAZONAS: gallery forest Et open area on Plateau of Huachamacari, 1 Mar. 1985, Liesner 18121A (MO, VDB, VEN); Cerro siquid Río Cunucunuma, near E escarpment at 1,900 m, Maguire et al. 30123 (NY, US); summit of Cerro Duida, Brocchinia Hills, above Vegas Falls, 1 Sep. 1944, sleyernare 58141 (F, , US); summit of Cerro Duida, Savanna Hills l, 025- " 200 m, 2 Sep. 1944, Steyermark Sono NY, US— cotype); Cerro Marahuaca, ección norocci- nal 2,500 m, 16 Feb. 1981, stir et al. 124426 (MO, VEN); Cerro Marahuaca, cumbre altiplanicie no arbolada, 2,580 m, 31 Jan. 1982, Steyermark et al. 125892 (VEN, VDB); Cerro Marahuaca, at: aislada al Sur-Oeste del Cerro, 2,480 m, 9-10 Feb. 1982, Stey- ermark et al. 126290 (VDB, Mo Cerro Huachamacari, cumbre, 1,800 m, 10 Feb. 1982, Steyermark et al. 126451; Cerro Marahuaca, across parte central de la Meseta Sur-Este, 2,560 m, 10-12 Oct. 1983, Steyer- mark 129444 (VDB, VEN); Cerro Duida, 29 Jan.-11 Feb., 1975, S. S. Tillett et al. 751-74 (topotype, M YF, NY, VEN This falls in treatments next to X. lugubris, X. tatei, and X. albescens but is distinct in its longer spike, duller bracts, and more uni- formly tapered, distinctively pigmented leaf blades, these sharper at apex. 31. Xyris lugubris Malme, Bull. Torrey Bot. Club 58: 324, pl. 24, fig. LA. 1931. TYPE: Venezuela. T. F. Amazonas: sum- mit of Mount Duida, 7,100 ft., Peak No. 7, G. H. H. Tate 639 (Tyler-Duida Ex- pedition Aug. 1928-Apr. 1929) (holo- type, NY; phototype, NY). Figure 31. Cespitose, hard-based perennial 3-10 dm high, the stems mostly contracted. Leaves erect or ascending, 2—5.5 dm long, the sheaths entire, less than Y? as long as blades, deep lustrous red-brown, smooth or papillate, keeled, gradually narrowed to blade, there eligulate or with an erect, narrowly triangular ligule to 4 mm long, the blades flattened, ensiform, 3-5 mm wide, gradually narrowed above middle to a narrowly acute, straight or incurved, thickened tip, the margins blunt or sharp-edged, smooth to papillose or tuber- culate-scabrid, the surfaces dull green or green with long bands of brown or red-brown, mul- tiribbed, smooth or papillate. Scape sheaths shorter than leaves, strongly costate, sharply keeled, with short, cusplike blades. Scapes straight or flexuous, distally subterete or slightly compressed, elliptic or oval in cross section, ca. 2 mm wide, smooth or sometimes striate, often punctate. Spikes broadly ob- 598 Annals of the Missouri Botanical Garden FIGURE 31. Xyris lugubris (from the holotype).—a. Habit sketch. —b. Leaf i —c. Leaf sheath—blade ~a —d. Leaf base.—e. Spike.—f. Lateral sepal.—g. Petal blade, stamen.—h. Staminode.—i. Stylar x.—j. Seed.—k. Fruit, two valves removed to show central-basal placentation. Volume 75, Number 2 1988 Kral 599 Xyris ovoid to subglobose or hemispherical, ca. 1 cm long, blunt, of many dull, dark brown bracts in nearly vertical ranks of 5 or more; sterile bracts several, the lowest much smaller than fertile bracts, grading into them, the fertile bracts broadly oblong, 6-8 mm long, rounded and slightly folded, the apex narrow- ly rounded, the margins entire or lacerate, the backs folded-convex, often low-carinate toward apex. Lateral sepals equaling bracts, .5-8 mm long, linear-elliptic, dark red- brown, apically acute or narrow but blunt, the keel ciliolate from near base to apex. Petal blades obovate, 5-7 mm long, yellow, the apex narrowly rounded, the margins suben- tire. Staminodia bibrachiate, the broad, flat- tened branches sparsely penicillate. Anthers lance-oblong, ca. 2 mm long, deeply bifid and sagittate, on filaments ca. 0.5 mm long. Cap- sule ellipsoid, 4-6 mm long, the placentation central, the valves septate only near base. Seeds rather few, cylindrical, ca. 2 mm long, including a pale, squamiform apex ca. 0.5 mm long, coarsely longitudinally few-ribbed and finely lined. Distribution. Wet, rocky savanna at or near summits, cerros Sipapo, Duida, and Ne- blina, T. F. Amazonas, Venezuela, infrequent. Additional specimens examined. VENEZUELA, T. F. AMAZONAS: Camp Savanna, Campo Grande, 1, Cerro Sipapo, 10 Dec. 1948, Maguire & Politi 27582 (GH, NY, US, VEN); Duida near summit of Culebra Peak, e 29155 (GH); Neblina, e m, u Steyermark 9453 (NY, VDB, VEN); Steyermark & Lu- teyn 129818 (MO, NY, VDB, VEN). 32. bed thysanolepis Maguire & Ly- . Smith, Mem. New York Bot. Card. 10: 17, fig. 3A- F. 1963. TYPE: Venezuela. Bolivar: rare around moist depressions and swales bordering river, scrub forest near Summit Camp, 1,925 m, central section Chimanta Massif, 2 Feb. 1955, J. A. Steyermark & J. J. Wurdack 356 (holotype, NY; isotypes, MO, NY, US, VEN). Figure 32A-C. A study of Venezuelan Xyris done in a way to reveal complexes has brought Xyris thy- sanolepis, X. jauana, and X. sipapoa into close alignment. In preparing descriptions and illustrations from type material of the three, and in studying materials collected later by myself and other collectors, I was forced to reassess taxonomic rank. Xyris thysanolepis and X. sipapoa were published in the same work and at that time were considered spa- tially isolated. Xyris jawana was considered well marked primarily on account of its con- nate lateral sepals, otherwise overlapping morphologically with X. thysanolepis. Ordi- narily sepal connation, particularly high con- nation, is a very good character within com- plexes, but in this complex the character breaks down. A check of the structure shows that while lateral sepals of X. thysanolepis may be connate, those of X. jauana may be free. A check of type material of X. sipapoa shows that sepals range even in a single spike from connate to free. If this usually significant character difference is removed, a complex of three former species becomes a single species of two varieties, as below: KEY TO THE VARIETIES OF XYRIS THYSANOLEPIS la. Leaf sheath bases bright to dull brown, eciliate to sparsely ciliate; apices of fertile bracts uio nded, the lacerate-scarious borders dis- eaf blades sqa or in- conspicuously ciliolate Bü ipu 5 ME thysanolepis cluding X. jauana) lb. Leaf sheath bases lustrous des inns or pale rown, mostly ev Hepa ciliolate; apices of fer- tile bracts narrowed, somewhat keeled, the scar- ious bo e red-brown; edges of rders pale leaf blades End id with fine, white hairs . thysanolepis var. sipapoa What then seems to emerge is a complex of medium- to high-elevation bog plants with scabrid, mostly unicostate scapes and lacer- ate-bordered, scarious-edged bracts, habitally similar, and showing a morphological affinity to X. confusa Smith & Downs of the Andes, possibly also to the Andean X. andina Malme. 600 Annals of the Missouri Botanical Garden Variety thysanolepis includes what was treat- ed as X. jauana, which varies in regard to sepal connation and which (contrary to the type description) bears no evident dorsal area, only a low, short, apical carina. Variety thy- sanolepis is mostly in high areas of the Gran Sabana, Estado Bolivar, with one known out- lier in Territorio Federal Amazonas (Cerro Yavi). The more slender var. sipapoa is ap- parently not rare on Cerro Sipapo, and plants answering to the type description have also been collected in Estado Bolivar (bog by rd. to Salto Aponguayo in rocky, sandy seeps, ca. 11 km SE of jct. with rd. to Kavanayen, 27 July 1983, Kral & Gonzales 70524, VDB, VEN). 32A. Xyris thysanolepis var. thysano- lepis. Figure 32A, X. jauana Lyman B. Smith & Steyerm., " Soc. Venez. Ci. Nat. 132-133: 277, fig. 2a-g. 1976. TYPE: Nenenuels: Bolivar: Meseta del A Cerro Jaua, selva de galeria al borde del tributario del Rio Ma- rajano, cumbre, 4?48'50"N, 64?34'10"W, 1,750- 1,800 m, 22-28 Feb. 1974, J. A. Steyermark, V. C. Espinosa & C. Brewer-Carias 109390 (holo- type, VEN; isotypes, NY, US). Figure 32B. Slender, solitary to cespitose, soft to firm- based perennial 2-10 dm high, the stem short or elongated to 6 cm, thick, the roots slender- fibrous. Leaves erect to spreading flabellately, 1-3 dm long; sheaths mostly 1⁄2 or more as long as blades, the slightly to very (orbicular) dilated base ciliate or not, tan to castaneous, shading above to pink, pale brown or stra- mineous, narrowing gradually into blade, keeled, the keel ciliolate or entire, the apex eligulate; blades linear, 2-5 mm wide, strong- ly flattened, the apex narrowly incurved-acute or acuminate, the margins usually thin, entire or ciliolate; surfaces pale green or yellow- green flecked with red, finely nerved, other- wise smooth. Scape sheaths shorter than to nearly as long as leaves, below terete and multicostate, above open, producing a short, erect, incurved-tipped, flat blade. Scapes straight or slightly flexuous, slightly twisted, subterete or oval in cross section toward apex, 1-2 mm wide, with 1, 2, (or more) costae, with 1 costa strong often making 1 edge, ciliate to densely ciliolate-scabridulous, the surfaces otherwise smooth, pale green, striate. Spikes pale red-brown, broadly ellipsoid, drying broadly obovoid, turbinate or globose, 0.7-1.3 cm long, of several spirally imbri- cate, firm, papillate or rarely smooth bracts with broad, scarious-lacerate, reddish or pale E brown borders or at least scarious- tipped; sterile bracts few, the lowest pair much smaller and narrower than the fertile bracts, keeled; fertile bracts broadly oblong to ob- ovate, 6-8 mm long, the backs rounded or somewhat folded (inner ones increasingly fold- ed-carinate), the apices narrowly or broadly rounded, the thin, colored borders variously lacerate and erect to squarrose. Lateral sepals free or up to Y connate, subequilateral, thin, pale, lustrous reddish brown, 6-7 mm long, the apex acute, the narrow, curvate keel rusty- ciliolate to low-serrulate-lacerate from at least middle to apex. Petal blades elliptic to ob- ovate, 4-5 mm long, yellow, the broadly to narrowly rounded apex erose. Staminodia bi- brachiate, the narrow branches densely pen- icillate-ciliate. Anthers narrowly lance-oblong to linear-oblong, 1.5-2 mm long, deeply bifid and sagittate, on filaments ca. 1 mm long. Capsule planoconvex, broadly ellipsoid to nar- rowly obovoid, mm long, the placen- tation basal, tbe valves without septa. Seeds numerous on elongate funicles, cylindric-fu- siform, ca. 1 mm long, deep to pale amber, coarsely longitudinally anastomosing-ribbed. Distribution. Locally abundant along streams through wet, rocky, medium- to high- elevation savanna, in Territorio Federal Ama- zonas and (more often) Estado Bolivar, Ven- ezuela. Additional specimens examined. VENEZUELA. T. F. a depression on summit, Cerro Yavi, 2,200 m 3 Mar. 1947, Phelps & Hitchcock 38 (NY, VEN). BOL ee norte de la Cumbre del Cerro Roraima, 2,810 m, 27 Mar. 1982, Aymard & Luteyn 2479 (NY, PORT); hacia Salto Aponwao (north part of Gran Sabana), 1,200 m, 7 Mar. 1983, Huber & Entralgo 7405 (MYF); cumbre del Sororopan-tepui al N de Kavanayen, 2,040 m, 28 June 1983, Huber & Alarcon 7741 (MYF); small stream Volume 75, Number 2 Kral 601 1988 Xyris FIGURE 32A. Xyris thysanolepis (Kral 70446).—a. Habit sketch.—b. Leaf apex.—c. Leaf blade-sheath junction. —d. Leaf base. —e. Spike.—f. Fertile bract.—g. Lateral sepal.—h. Petal blade, stamen.—i. Staminode, enlarged beard hair apex. —j. Stylar apex.—k. Dehisced capsule, showing basal placentation.—l. Seed. 602 Annals of the Missouri Botanical Garden FIGURE 32B. Xyris thysanolepis (type of X. jauana) .—a. Habit sketch. pa Leaf tip.—c. Leaf blade—sheath junction. —d. Leaf base. —e. Spike, upper scape.—f. Lateral sepals.—g. Petal blade, stamen.—h. Staminode, enlarged beard hair apex.—i. Stylar apex.—j. Capsule outline, PIER —k. Seed. Volume 75, Number 2 Kral 603 1988 Xyris FIGURE E 32C. Xyris chine: var. sipapoa ue the type) .—a. ie sketch.—b. Leaf apex.—c. Leaf blade-sheath junction.—d. Leaf base.—e. Leaf base, side view.—f. e.—g. Lateral sepal.—h. Petal blade, stamen (just above, a fertile bract) .—i. Staminode, to lefi an enlarged ee = tip of beard hair.—j. Stylar apex.—k. Outlines of capsule with placenta, to right a valve outline. — 604 Annals of the Missouri Botanical Garden at ravine base in bog ca. 1 km E of Kavanayen, 26 July 1983, Kral 70446 (F, K, MO, NY, SP, U, US, VDB, VEN); 1.5 km E of Kavanayen, rocky seeps, S side of rd., 27 July 1983, Kral 70535 (F, K, MO, NY, SP, U, US, VDB, VEN); Macizo del Chimantá, Apacara-tepui, sector Norte del Macizo, ca. 2,200 m, 30 Jan.-1 Feb. 1983, Steyermark, Huber & Carreno E. 128407 (VDB, VEN). 32B. Xyris thysanolepis var. sipapoa (Maguire & Lyman B. Smith) Kral & Lyman B. Smith, stat. nov. X. sipapoa Maguire & Lyman B. Smith, Mem. New York Bot. Gard. 10: 18, fig. 4A—F. 1963. TYPE: Venezuela. T. F. Amazonas: fre- quent, banks of lower Cano Negro along open savannas, alt. 1,400 m, Cerro Si- papo, 25 Dec. 1948, B. Maguire & L. Politi 27911 (holotype, NY; isotypes, GH, NY, US) As in the rest of the species but usually somewhat lower in stature, the leaves strongly spreading flabellately, the bases abruptly or- bicular-dilated and lustrous, mostly ciliolate, the blades with a fine marginal dusting of white hairs. Spikes mostly ovoid or ellipsoid, drying obovoid, the bracts carinate toward apex, the apex narrowed and folded, the mar- gins pale-lacerate-scarious. Lateral sepals, sometimes in same spike, free to 3 connate. Distribution. Wet to rather dry, medi- um- to high-elevation rocky savanna, Cerro Sipapo and environs, Territorio Federal Ama- zonas, Venezuela, and Gran Sabana, Estado Bolivar, Venezuela. Additional specimens examined. VENEZUELA. T. F. AMAZONAS: Cerro Sipapo: frequent in bog-savanna, ter- races S of Cario Negro at 1,600 m, 6 Jan. 1949, ttle un & Politi 28194 (NY, US); a pet in savanna v po Grande, 1,500 m, 15 Dec. aguire & Polit 27684 (F, I NY, US); bog by in to Salto Aponguayo n rocky, sandy seeps, ca. 11 km of jct. with rd. to Kav van en 27 July 1983, Kral dis (VDB, VEN, and to be distributed); savanna, ca m, Torom-meru, NW of Parupa, 14 Dec. 1984, Kral E NE VDB, VEN, and to be distributed). 33. Xyris concinna N. E. Br., Trans. Linn. Soc. London, Bot. II(6): 68. 1901. TYPE: Venezuela. Bolivar: summit, Mt. Rorai- ma, 8,600 ft., McConnell & Quelch 496 (lectotype, K). Figure 33. Hard-based, densely cespitose perennial 0.8-2 dm high, the stem contracted, to 1.5 cm long. Leaves spreading flabellately, 5-10 cm long; sheaths about as long as blades, entire (rarely with a few brown cilia at base), cochleariform toward base, thence contracted abruptly, ciliate-keeled, gradually narrowing to blade, eligulate; blades flat, narrowly linear, 1.5-2.5 mm wide, gradually tapering distally to an attenuate, terete, or narrowly acute, flattened apex, the tip with a tuft or short fringe of stiff, pale hairs, the margins minutely scabrociliate or papillose, the surface finely nerved, green or maroon, smooth. Scape sheath slightly shorter than leaves, loose, sca- rid-keeled, short-bladed. Scapes slenderly linear, stiff, ca. 1 mm thick, oval or terete in cross section, unicostate, also finely fluted, the costa smooth, papillose or scabrociliolate. Spikes ovoid, 5-7 mm long, obovoid in fruit, reddish brown, of several spirally imbricate bracts without dorsal area and with broad, pale laceroscarious borders; sterile bracts 2- 4, ovate, keeled, grading larger into fertile bracts, these few, lance-ovate or oblong, acute to acuminate, slightly to very keeled, 6-7 mm long, the backs papillose. Lateral sepals free, narrowly oblanceolate, equilateral, ca. 6 mm long, narrowly acute, the sides dull, pale brown, the keel narrow, firm, dark brown, entire or (usually) ciliate from middle to apex. Petal blades ca. 5 mm long, yellow, elliptic, subacute, wavy-margined. Staminodia bibra- chiate, the branches flattened, the hairs pen- icillate but often multiseriate in narrow sheets. Anthers oblong, ca. 2 mm long, apically short- bifid, basally auriculate, on filaments ca. mm long. Capsule oblong-ellipsoid, ca. 3 mm long, the valves eseptate, the placentation basal, funicles long. Seeds fusiform, ca. 1 mm ong, red-brown, translucent, finely longitu- dinally ribbed. — Distribution. Infrequent in wet, high, sandstone savanna, the higher tepuis of the Gran Sabana, Estado Bolivar, Venezuela. Additional specimens examine d. VENEZUELA. BOLÍVAR: Matahui-tepui, 2,700-3,000 m, 22 Aug. 1982, A. Castillo 2282 (UCV, VDB); Kukenan-tepui, sector mas septentrional, algo separado del macizo principal, Volume 75, Number 2 Kral 1988 Xyris 605 5 Imm Q uo L EN FIGURE 33. aa concinna (Steyermark 112658).—a. Habit sketch.—b. Leaf apex.—c. Leaf blade-sheath junction.—d. base.—e. Small sector of leaf edge enlarged to show hairs. —f Spike.—g. Lateral sepal.— h. Petal blade, eae staminode, enlarged part of beard hair, stylar apex.—i. Capsule, two valves removed to show placentation. 2,500 m, 28 Apr. 1984, Huber 9449 (MYF, a & Alarcon 10602 (MYF, VDB); altiplanicie del Uei-tepui Cumbre del Ilú-tepui, sector centro-meridional, 2,630 m n del Sol), epos occidental por encima del valle del 29 Apr. 1984, Huber 9498 (MYF, VDB); cumbre sur- o Arabopo 50 m, 22 Jan. 1985, Huber 10022 occidental del Ilú-tepui, 2,700 m, 18 June 1985, Huber YE, VDB); re Roraima: cumbre, parte noreste de 606 Annals Mes po m Garden Venezuela inmediata al sur del hito que marca los limites con m Brasil y Venezuela, 2,750-2,800 m, 26 Aug.-2 1976, Steyermark et al. 112566 (MO, NY, US, VEN) 112648 (K, MO, NY, US, VEN). The affinities of this low plant of high- altitude tepuis are definitely with X. hyme- nachne C. Martius, which is frequent in the ola: lower-elevation savannas of the Gran Sabana. However, the leaves are harder, their bases more orbicularly dilated; the spike bracts are more folded and more acuminate; and the seeds are somewhat longer. 34. Xyris hymenachne C. Martius, Flora 24(2): 55. 1841. TYPE: Brazil: “Prov. Minarum, Mart. Hb. 872” (lectotype, BR; isolectotypes, L, M; phototypes, F, NY). Figure 34. X. arescens Kunth, Enum. Pl. 4: 3. 1843. X. Mri pm Gleason, Bull. Torrey Bot. Club 56: 393. 929. TYPE: Brazil. Rio Branco: Mt. Roraima, Phil- Sil Swamp, 5,100-5,200 ft., 11 Nov. 1927, G. H. H. Tate 334 (K). Solitary or cespitose, often bulbous-based, slender perennial mostly 2-6 dm high, the stems contracted. Principal leaves ascending or spreading flabellately, 0.5-2 dm long; sheaths entire or sparsely brown ciliate at dilated base, dull brown or stramineous, at very base sometimes castaneous and abruptly orbicular-dilated, V or more as long as blades, tapering above gradually to blade, usually eligulate, the blades flat, straight or twisted, strongly compressed, 1.5-4 mm wide, acute to long-acuminate, the margins with edges thin or slightly thickened, smooth, papillate or ciliolate, the surface smooth to papillose- rugulose, finely nerved, dull green. Scape sheaths slightly to much shorter than leaves, tubular and sharply costate below, above open, strongly keeled, short-bladed. Scapes narrow- ly lineal, twisted, sometimes flexuous, distally subterete, 0.6-0.7 mm thick, mostly unicos- tate (sometimes multicostate), the costae low but distinct, smooth to papillate or ciliolate. Spikes obovoid to subglobose, 5-8 mm long, attenuate, dull pale red-brown, of several loosely spirally imbricate bracts, these round- ed-convex, often medially low-ribbed, without distinct dorsal area, with erect, broad, pale, scarious-lacerate borders; sterile bracts slight- ly smaller than fertile bracts but lowest over 14 as long as spike, ecarinate, the inner sterile and lower fertile bracts obovate or broadly oblong, 7-8 mm long, the backs often pa- pillate. Lateral sepals free, very inequilateral, oblong-curvate, 4-6 mm long, obtuse, the darker keel fimbriociliate or ciliate at least from middle to apex. Petal blades obovate, ca. 5 mm long, the apex narrowly rounded, the margins wavy. Staminodia bibrachiate, the broad, thin branches each tipped by a few penicillate hairs. Anthers lanceolaate, ca. 1.5 mm long, deeply bifid and sagittate, on fila- ments ca. 0.5 mm long. Capsule ellipsoid to broadly obovoid, ca. 4 mm long, the valves without septa, the placentation basal. Seeds cylindric-fusiform or narrowly ellipsoid, on long funicles, 0.6-0.7 mm long, apiculate, amber, longitudinally conspicuously multi- ribbed. Distribution. Moist to wet, medium- to high-elevation savanna, South America east of the Andes, in northern South America, Colombia east into Guyana. Selected Venezuelan specimens. T. F. AMAZONAS: Dto. Río Negro, seeps along Rio Mawarinuma, just above Neb- lina expedition base camp, 26 Nov. 1984, Kral 71823 VEN, and to be distributed); same area, 2 km above Neblina base camp, 3 Dec. 1984, Kral 71937 (VEN, and to be Sayers Neblina, rocky places along Cano Grande at 1,100 m, 24 Nov. 1957, Maguire & Wurdack 42200A (NY, o BOLÍVAR: carretera El Dorado-La Gran Sabana, alrededores de km 132, ca. 1,200 m, 21 Feb. 1968, Bunting 3050, identified as X. submontana Gl. (US); hacia el Salto Aponwao, 1,200 m, 7 Mar. 1983, Huber & Entralgo 7408 (MYF, VDB, VEN); several numbers by Kral along El Dorado-Sta. Elena Highway (Ven. 10) through the Gran Sabana (numbers 70305, 70326, 70367, 70392, 70413, 70447, 70467, 70515, 70559, 70591, 70620, 70628, collected 22-29 July 1983, and with general distribution to include NY, SP, U, US, an Sinis die subsequently 72004, 72061, from the sa 13 & 21 os 1984 MYF, VEN, and to s distributed) 17k ji ~ as rs Lipari 1,900 m, b. 1948, Phelps 6 (NY); irn of Rio Cuyuni, NE of Luepa, 1 e m, 23 Apr. Nilsson 492 oy 1 y cens xd. at base of Ptari- -tepui, 4 4, Steyermark 59841 (F, US, VEN); Auyan- tepui, id occidental del cerro, vic. *Rio Lomita Camp," Stovermark Volume 75, Number 2 Kral 1988 Xyris E E 3 3 AN UN FIGURE 34. Xyris hymenachne (S sheath—blade junction.—d. Spik teyermark 93465, Kral 70326) .—a. Habit sketch.—b. Leaf tip.—c. Le —e. Fertile bract.—f. Lateral sepal.—g. Petal blade, stamen.—h. Stam node.—i. Stylar apex.—j. Leaf base.—k. Seed. af i- 608 Annals of the Missouri Botanical Garden 1,800 m, 7 May 1964, Steyermark 93465 (K, NY, VDB, US). This species, while very highly variable in stature and leaf, is distinguished primarily by the very broad, pale, scarious bract borders and the absence of a dorsal area. The leaf bases range from strongly dilated to indis- tinctly so, and the numbers of flowers and bracts in a spike vary widely. A perfect range of intermediates connects this to X. submon- tana Gleason, a not uncommon morphology in the Gran Sabana of Estado Bolivar, Ven- ezuela. This last was supposed to be distin- guishable on a basis of its comparative smoothness, the depressed-globose spikes, and the granular-papillose spike bracts. The species unfolds its blooms in morning. It is one of the weedier species, often coming in abundantly on moist to wet, nearly totally mineral sub- strates. 35. Xyris decussata Gleason, Bull. Torrey Bot. Club 56: 392. 1929. TYPE: Vene- zuela. Bolivar: Mt. Roraima, summit, 26 Nov. 1927, G. H. H. Tate 427 (lecto- type, NY). Figure 35. Robust but low, cespitose, thick-based pe- rennial 1.5-4.3 dm high, the stems stout and contracted or to 5(-15) cm long, ascending. Leaves rigid, spreading flabellately, 1-2 dm long; sheaths firm, entire, about as long as blades, castaneous, tapering gradually from broad base, apically with a short, often salient, small, narrowly triangular ligule; blades broadly linear, flat, 3-5 mm wide, abruptly broadly rounded at apex, the margin com- pletely and densely rusty ciliate with hairs arising from a dark brown, incrassate border, the surfaces green, finely nerved, smooth. Scape sheaths shorter than leaves, tubular- carinate, the keel ciliate, alate, terminating a very short, blunt, ciliate blade. Scapes vicis slightly twisted, distally flattened, 2- m wide, bicostate, the costae densely rusty ciliate. Spikes subglobose or depressed-glo- bose to obovoid, ca..1 cm long, dark brown, of many, firm but loosely appressed bracts without distinct dorsal areas, usually in nearly “(C); Cerro Roraima: c vertical ranks; sterile bracts several, the lower ones triangular, much shorter than the fertile bracts and grading gradually into them, the fertile bracts ovate or broadly oblong, ca. 6 mm long, apically broadly to narrowly round- ed and subentire, the backs ecarinate, slightly convex-rounded. Lateral sepals free, sub- equilateral, ca. 6 mm long, linear-oblanceo- late, obtuse or broadly acute, dark brown, the firm dark keel entire or ciliate or ciliato-lac- erate toward apex. Petal blades elliptic, ca. 5 mm long, yellow, the apex broadly and bluntly acute or narrowly rounded, the mar- gin wavy. Staminodia 4-brachiate, the branches long-penicillate. Anthers oblong, 1.5-1.7 mm long, cleft to below the middle, sagittate, on filaments nearly 2 mm long. Cap- sules ellipsoid, 4-5 mm long, placentation basal-central, the valves with low septa. Seeds numerous, narrowly cylindric-oblanciform, ca. 1.5-1.7 mm long, including a pale, narrowly conic apiculus ca. 0.3 mm long (loose outer integument), the body pale reddish brown, translucent, finely but distinctly longitudinally ribbed. Distribution. High-elevation tepuis, in boggy savanna, over 2,000 meters, south- eastern and southern Estado Bolivar, Vene- zuela. Additional specimens examined. VENEZUELA. BOLÍVAR: Cumbre del Ilu-(Uru-)tepui, sector centro-mer- idional, 2,630 m, 29 Apr. 1984, Huber 9509 (MYF, VDB); cumbre del Tramen-tepui, la porción mas noroc- cidental del Macizo del Ilu-(Uru-)tepui, 2,650 m, 23 Jan. 1985, Huber 10053 (MYF, VDB); Ilu-tepui, Gran Sa- bana, wet places in open low bush, Mesa Ridge, 6,000 13 r. 1952, Maguire 33408 (NY)— this is the u 6 1944, Stey- ermark 58806 (F, NY, EN) Chimantá dE upper mossy slopes, NW part summit of Abacapa-tepui, abov dia t line of sandstone bluffs, Steyermark 74998 cumbre parte NE de Venezuela im- ead al sur del hito que marca los limites con Guyana, Brazil, and Venezuela, 2,750-2,800 m, 26 Aug.-2 Sep. 1976, Steyermark et al. 112618 (VEN); Cumbre de Auyan-tepui, cerca de las orillas del sector oriental, al norte de la Misión de Camarata, 1,940 m, 27 Feb. 1978, Steyermark et al. 116115 (VEN). This endemic is vegetatively a stubby ver- sion of X. bicephala Gl., but even a first Volume 75, Number 2 Kral 609 1988 Xyris FIGURE 35. Xyris pip be (from the type, except for flower).—a. Habit sketch.—b. Leaf tip.—c. Two samples of leaf-sheath junction.—d. Enlarged sector of edge of leaf blade.—e. Leaf base.—f. Spike, upper scape.—g. Lateral sepal.—h. poer blade, on n.—i. Staminode and enlarged tip of beard hair. —j. Stylar apex.—k. Capsule, showing placentation; one n removed.—l. Seed. 610 Annals of the Missouri Botanical Garden 4 à I D^ ^ lcm i Í v NP, n V Na SS XL f | ie ASS SSS 4 =< SS Cc AL SS Drs e. | B E arar we FIGURE 36. Xyris albescens (Steyermark 59734) .—a. Habit sketch.—b. blade.—d. Bod. blade a at junction with sheath.— e. S a e.—f Ti Leaf tip.—c. Sector of leaf, m enlarged) .— ike.—h. Fe rtile b ract.—i. Later pal.—j. richomes from edee of leaf e (much al blade and stamen.—k. Sta al s Pet style branc "d —m. Seed.— n. Capsule, valve removed, prse septum (shaded). Volume 75, Number 2 1988 Kral Xyris glance reveals the distinctive dense fringe of rust-colored hairs marginally on the leaf blade. Taller material with more lustrous bracts and longer leaves but having such ciliation has been obtained from Chimantá, Ptari-tepui, and Ilu-tepui. This material appears to be from populations “influenced” by other species such as X. albescens Steyerm., or is identified as X. melanovaginata Kral & Smit 36. Xyris albescens Steyerm., Fieldiana Bot. 28(1): 105, fig. 16a, b. 1951. TYPE: Venezuela. Bolivar: Ptari-tepui, Bonne- tia roraimae forest on southwest-facing shoulder, 2,000-2,200 m, 2 Nov. 1944, A. Steyermark 59734 (holotype, F; isotypes, GH, NY, US). Figure 36. Robust, cespitose perennial (3.5-)4-10 cm high, the stems short, stout, ascending. Leaves elongate-linear, 2-6 dm long, mostly spread- ing or ascending; sheaths at least 1⁄2 as long as blades, keeled, firm, lustrous, deep red- brown or castaneous, the margins entire, ta- pering evenly from base to a small, indistinct ligule; blades evenly linear, flat and strongly flattened, pale green, (3-)5-6 mm wide, with apices asymmetrically narrowly acute or (in type) incurved and narrowly rounded, the edges densely white- (rarely pale red-brown-) ciliate, sometimes with a submarginal brown or red border. Scape sheaths much shorter than leaves, short- to elongate-bladed, the blades as in leaves. Scape pale green, distally broad and flattened (2-)2.5-3 mm wide, the edges densely pale ciliate as in leaf blades. Spike broadly ovoid, globose or hemispheric, (1-)1.5 cm long, blunt, rich, lustrous brown or castaneous, multiflowered; sterile bracts very many, triangular-ovate, the lowermost small, often squarrose, grading larger into the fertile bracts, these narrowly ovate to oblong, firm, 6-9 mm long, apically rounded, entire, sometimes somewhat cucullate, the dorsal area absent or inconspicuous, the backs rounded- convex, ecarinate. Lateral sepals equilateral, oblong-linear to linear-oblanceolate, mostly acute, 6-7 mm long, pale except for dark, scabro-ciliate to villosulous keel. Petal blades narrowly obovate, ca. 6-7 mm long, yellow, apically broadly rounded, erose. Staminodia bibrachiate, the slender branches long-peni- cillate from base to tip. Anthers lanceolate, ca. 2.5 mm long, apically deeply retuse, ba- sally sagittate, on filaments ca. 1 mm long. Capsule narrowly ellipsoid, ca. 5 mm long, the valves deeply septate at base, placentation appearing central but axile at base. Seeds narrowly fusiform-cylindric, 2-3 mm long, pale brown, including a white, triangular-lin- ear scale 0.5-0.7 mm long, the fruit body spirally coarsely parallel-ribbed. Distribution. Moist to rather dry rocky savanna, high elevations of tepuis, southern stado Bolivar, Venezuela. Additional specimens examined. VENEZUELA. BOLÍVAR: Cima del Roraima, Jan. 1977, Delascio & Brew- er-Carias 4802 (VEN); Sororopan-tepui, a e 1983, Huber = Alarcon 7725 (NY) ruani-tepui, 12 km al NNE del Kukenan-tepui, 29 Feb. 1984, Huber 9088 (MYF, VDB); Cumbre del Yu- ruani-tepui, a W del Cerro Kis 27 Apr. 1984, Huber 9414 (MYF, VDB); Kukenan-tepui, cumbre del sector mas vd via algo separado m, 28 Apr. ud 149 (K); vic. rd. campa- at valley of savanna of Rio Urarama below Urarama- “tepui, 'NE of Luepa, 1,220 m, 24-25 Apr. 1960, St & & Nilsson 641(NY, VEN); Prari- tepu i, dry sandy and rocky portion of southeast- bene slopes, 1, 600 m,l Nov. 1 944, Steyermark 59683 (F, NY); Chimanta Massif, elfin forest on plateau of southeast mr upper shoulder of Apacará- tepui, 2,000 m, 19 June 1953, Steyermark 75757 (F, NY, VEN); Meseta del Jaua, el Este del trib. del Rio specimens in Chimanta, altiplanicie en la base meridional de los fara- llones superiores del Apacara-tepui, sector norte del Ma- cizo, 2,200 m, Steyermark et al. 128393 (VEN, VDB). Material of this from Guyana, Upper Ma- zaruni District, north slope of Mt. Roraima, alt. 2,000-2,300 m, 16 Feb. 1985, J. Renz 14266, was lent by U. Other material with affinities, lent by NY, from T. F. Amazonas of Venezuela is Cerro Marahuaca, 2,500 m, 16 Feb. 1981, Steyermark 124439. Annals of the Missouri Botanical Garden 37. Xyris fuliginea Kral & Lyman B. Smith, sp. nov. TYPE: Venezuela. T. F. Amazonas: Dept. Atabapo, Cerro Mar- ahuaca—FHUIF Cumbre, zona boscosa en la falda este del riachuelo, 2,480- 2,500 m, 3?35'N, 65°20'W, 1-2 Feb. 1982, J. Steyermark, M. Guariglia, N. Holmgren, J. Luteyn & S. Mori 125978 (holotype, VEN; isotype, VDB). Figure 37 Planta perennis caespitosa, rigida, 3-5.6 dm alta. Ra- dices graciles. Caules breves aut usque ad 5 cm longi. Folia principalia vulgo flabellate expansa, 1.5-3 dm longa, vaginis scaporum parum longiora; laminae vaginas ae quantes vel eis leviter aut 1-2-plo breviores, linearotrian- 3-5 mm latae 3-plo spica —_ paribu m transientes; bracteae fertiles latae oblon late acuta; ala carinali lata, a me see apicem rufociliata. Laminae Seeman = obova 5.5-6 mm longae, undatae, erosae. Staminodia bibra- chiata, brachiis longipenicillati. Antherae lanceolatae, ca. emarginatae et i ee filiis latis, ca. 1 mm longis. Capsula ellipsoidea m longa, placentae aia sed valvis capsulae ad ds DELE longitudine prominente spiraliter multicostata, conicam pallidam squamellam ca. 0.5 mm longam includen Stiff tufted perennial 3-5.5 dm high. Roots slender, fibrous. Stems short or up to 5 cm long. Principal foliage leaves commonly spreading flabellately, 1.5-3 dm long, slightly longer than the scape sheaths; blades equal to sheaths or to 1⁄¿—2⁄4 as long, linear-trian- gular, level, flattened, 3-5 mm wide, oliva- ceous, smooth; apices gradually narrowed, acute, slightly thickened at tips; margins thin, nely densely white ciliolate or white ciliate; sheaths carinate, deep-castaneous, shining, entire, dilated at base, narrowing gradually into blades, eligulate. Sheaths of scapes cas- taneous toward base, terete, twisted, multi- costate, opening toward apex, carinate, with short blades. Scape ancipital toward apex, ca. 2-2.5 mm wide, white-ciliate, olivaceous. Spikes multiflorous, obovoid, 1-1.5 cm long, obtuse. Bracts erect, subdecussate to loosely spirally imbricate, firm, ecarinate, fuligineous (charcoal brown), without dorsal area, entire, lacerate; sterile bracts several, triangular- ovate, the lowest pair strongly carinate, ca- rinae ciliate, narrowly acute, ca. 3 times shorter than the spike, the inner pairs shorter than the fertile pairs, grading into them; fer- tile bracts broadly oblong to oblong-lanceo- late, 7-9 mm long, broadly acute to narrowly rounded. Lateral sepals free, subequilateral, ca. 7-8 mm long, deep red-brown, slightly curvate, broadly acute; keel broad, red ciliate from middle to apex. Petal blades broadly obovate, 5.5-6 mm long, yellow, broadly rounded apically, erose. Staminodia bibra- chiate, the branches long-penicillate. Anthers lanceolate, ca. 2 mm long, deeply emarginate and sagittate on broad filaments ca. 1 mm long. Capsule ellipsoid, 5 mm long, the pla- centae central but the capsule valves deeply septate toward base, thus the placentation subaxile. Seeds numerous, narrowly ellipsoid to narrowly claviform, ca. 1.5 mm long, lon- gitudinally spirally prominently ribbed, and including a pale, conical scale ca. 0.5 mm long. Distribution. Confined to the Cerro Marahuaca, wet high altitude, rocky savanna, elevation over 2,000 m, T. F. Amazonas, Venezuela. Paratypes. Summit, northeast, cn of ~ stream, 2,500 m, 16 Feb. 1981, Steyermark e 124439 (MO, NY, VDB, VEN); same jointe Hd has collectors as type, Steyermark et al. 125991, 125993 ls (VDB, ip dar noreste, 3?50'N, 2,580-2,600 m, 30 Mar.-1 Apr. 1983, Stey- erm rd Med 12924 (VDB, VEN); cumbre parte uud de la Meseta Sur-Este, a lo largo de la VY oum Yekuana, afluente del Rio Negro, 2,560 m, 10-12 Oct 1983, Steyermark 129470 (VDB, VEN) This species has been more of a problem than most in that it appears to be a meld of Volume 75, Number 2 Kral 613 Xyris / a ld ith i \ "7 «7 FIGURE 37. Xyris fuliginea (from the type) .—a. Habit sketch. —b. Leaf apex.—c. Leaf blade, Aa — d. Leaf blade-sheath junction.—e. Leaf base.—f. Spike.—g. Lateral sepal.—h. Petal blade, stamen.—i. inode, enlarged beard hair.—j. Stylar apex.—k. Capsule, ideal longisection showing two septa andj, placentae, two seeds.—l. Capsule valve showing septum (dark shading) .—m. Seed. 614 Annals of the Missouri Botanical Garden E 38A. Xyris oem a 105124, icd iid sepal and stylar apex from the typ . Spike. Habit Ma —b. Leaf a Leaf bla de—sheath junct —e. Lateral sepal.— f. oa se stamen.—g. Stylar apex. dia pomis —i. Seed. three other high-country species: X. albes- cens (ancipital scape, white-ciliate leaf and scape edges), X. culmenicola (obovoid, sooty brown spikes, ciliate, linear-triangular leaf ers and X. lugubris (similar bract, spike leaf characters). Yet the uniformity of "m now rather large set of samples from Cerro Marahuaca collected by Steyermark and oth- ! Volume 75, Number 2 1988 Kral Xyris 615 FIGURE 38B. Xyris ptariana (from the type).—a. Staminode.—b. Ovary and style branches; cross section elow.—c. Enlarged staminodial beard hair tip.—d. Petal blade, stamen. ers makes description of this cryptic mor- phology necessary. 38. Xyris ptariana Steyerm., Fieldiana, Bot. 28(1): 110, fig. 161, J. 1951. TYPE: Venezuela. Bolivar: forming dense tufts, wet bluffs, 2,000 m, Ptari-tepui, Gran Sabana, Steyermark 59919 (holotype, F; isotypes, F, NY, US, VEN). Figure 38A, B. Robust, densely cespitose, usually smooth, perennial to 1 m high, the stems elongate, branching from the base and rebranching, the ultimate branches forming frondlike plates of leaves. Leaves spreading-ascending, 1.5-3.5 dm long; sheaths mostly deep red-brown, sometimes pale brown, nearly as long as blades or longer, entire-edged, strongly keeled, nar- rowing evenly from the dilated base to the blades, there producing a triangular, scarious, spreading or recurved-tipped ligule to 3 mm long; blades flat, straight, linear-ensiform, (3-)3.5-12 mm wide, the apex incurved- acute, flat, the margins usually entire (on Abacapa-tepui densely red ciliate) or papil- lose-tuberculate, the base narrowed ca. 2-3 cm from ligule on ventral side, the surfaces deep green, strongly multinerved. Scape sheaths shorter than leaves, proximally tu- bular and multicostate, lustrous brown, dis- tally opening, strongly keeled, with a short blade. Scapes slender but stiff, straight or slightly flexuous and slightly twisted, distally flattened or in cross section narrowly ovate or elliptic, sometimes ancipital, 1.5-3 mm wide, the costa(e) 1-2, usually making up edges, smooth to tuberculate-scabrid or (on Abaca-tepui) red ciliate. Spikes ellipsoid to broadly obovoid, 1-1.5(-2) cm long, blunt, pale brown or olivaceous brown to reddish brown, of many spirally imbricate, thinnish bracts without dorsal areas, the lowest sterile bracts much smaller than the fertile bracts, ovate, obtuse-angled to acute, slightly keeled, grading into the larger, flatter fertile bracts, these obovate-oblong, 0.8-1 cm long, the apex broadly rounded, entire to erose, the backs but slightly convex, toward apex often with a low, lustrous costa. Lateral sepals free, sub- inequilateral, oblanceolate or spathulate, 6— 8 mm long, blunt, thin, with a broad, darker, entire to ciliolate keel. Petal blades broadly obovate to suborbicular, 6-7 mm long, yel- low, the broadly rounded apex erose. Stam- inodia clavate-trilobate when present (in the type), distally densely long-penicillate. An- thers lance-oblong, 1 -2 mm long, deeply bifid 616 Annals of the Missouri Botanical Garden and sagittate, on filaments 1-1.5 mm long. Capsule ellipsoid, ca. 6-7 mm long, the pla- centation appearing free-central but the 3 valves with strong, deep septa reaching the capsule axis, thus placentation actually axile. Seeds very numerous, asymmetrically ellip- soid, 1-1.5 mm long, pale amber with a pale, narrowly conic apical scale, longitudinally finely but distinctly ribbed. Distribution. Boggy, seepy, rocky areas of savanna, at elevations of 1,300 meters or more, Territorio Federal Amazonas and Es- tado Bolivar, Venezuela. Additional specimens < examined. VENEZUELA. T. F. in bogs about ño Negro Varna a l, 400 m, 15 Dec. 1948, & Politi 27703 (GH, K, US, VEN)— very large material as to plant height, leaf width, T etc.; summit, Cerro Marahuaca, 2,685 m, 15 Jan. 1981, Maguire 65634 (NY); Certo ACA "i cliffs below escarpment, 1,300 m, frequent, Dec 8, Maguire & Politi 27498 (F, NY, US); Cerro Paraque, ie 800 m, Feb. 1946, Phelps 47 (US); summit of Cerro Duida, on high moist ridgetop, -2 ft., 26 Nov.-16 Dec. 1927 G. H (NY)— this with scape ab oci wr norm. ie Cerro Apacara, 2, m, Rio Caroni, 11 Nov. 1946, F. Car- dona uf (US); Cumbre del Apra ada- pul s “et sur, ubicado ca. 30 km al E de Urim ,90 0 June 1984, Huber 9565 (MYF, VDB); Chimantá Massif, NW stone bluffs, 2,125-2,300 m, 14 Apr. 1953, 74998— this material with red-bordered scapes and leaf blades (F, NY, VEN). This species may be a complex of varieties held together by a combination of elongate and elongate-branched axes producing frond- like plates of leaves, by the curious ligule, and by the distinctive, indented ventral mar- gin of leaf blade just above the sheath. At one end of the variation is the type, narrowest example of X. ptariana in leaf and scape, and which differs the least from X. witse- nioides F. Oliver. At the other extreme, to the far west in Territorio Federal Amazonas, is X. xiphophylla Maguire & Lyman B. Smith. Several summit elevations in Estado Bolívar and some in Amazonas are now known to have this taxon, and populations on each are distinguishable, yet not sufficiently to al- low a specific treatment. 39. Xyris witsenioides F. Oliver, Thurn, Degn 207. 1886; Trans. Linn. Soc. II. 2: 285, pl. 50, figs. 9-15. TYPE: n Guiana, Roraima, ledge 7,300 ft.," Everard F. im Thurn, 14 Dec. 1884 (lectotype, K). Figure 39A, B. Densely cespitose, multibranched peren- nial (1.2-)2-6 dm tall, the primary branches elongate, ascending, rebranching to form frondlike plates of leaves, the whole producing large dome-shaped masses with bases buried in substrate. Leaves ascending, rarely spread- ing, 5-20 cm long; sheaths eciliate, the bases tightly imbricate, distichous on the elongate stems, castaneous or light brown, often per- sistent long after blades, fully as long as blades, strongly keeled, narrowing gradually to scar- ious, triangular, erect ligules 1-2 mm long; blades gladiate-linear, flattened, straight or falciform, 1-3(-4) mm wide, gradually nar- rowing from just below middle to apex, there incurved-acute or erect-acute, slightly thickened; margins thin, smooth or scaberu- lous-papillate, rarely ciliolate; surfaces yellow- green, smooth, finely but evidently multi- nerved. Scape sheaths shorter than leaves, closed at base, ciliate-keeled, multicostate, above with short, erect, ciliate blades. Scapes straight or flexuous, distally subterete to slightly compressed and oval or narrowly ob- long-elliptic in cross section, 0.7-1 mm wide, ecostate or 1(-2 or more) costate, the costae smooth, scaberulous or rarely ciliolate, the surfaces otherwise smooth. Spikes ellipsoid to obovoid, drying broader, 0.7-1.4 cm long, olive-brown or red-brown, of loosely spirally imbricate thin bracts without evident dorsal areas; sterile bracts several, the lowest much smaller and narrower than the fertile bracts, these oblong to obovate, shallowly convex, low-carinate toward apex, 5-7 mm long, rounded, entire, aging lacerate. Lateral sepals free, subequilateral, ca. 5-7 mm long, oblong- curvate, obtuse, the broad thin keel subentire to ciliate, aging lacerate. Petal blades broadly obovate to suborbicular, yellow, 5-6 mm long, the broad, shallowly rounded apex erose-den- ticulate to nearly entire. Staminodia distally above a short geniculation producing a single, Volume 75, Number 2 Kral 617 Xyris bs] x PAD junction. —d. Lea LT — sepal p broader and shorter than most) .—h. Petal blade, sta apex.—k. At lefi, a longisector through capsule, with a placental strand showin ng its proximal and distal attachment; at right, another septum with upper part removed showing neighboring locule, a valve removed showing septum and placental strand (dark shaded), 618 Annals of the Missouri Botanical Garden 415mm é mm Tr AN S ASS S [] Ü) S "n 4 // N `. b: J Al |] O s 29 1 N C O \) m f e O J e H S8 O 0 FicurE 39B. Xyris witsenioides (Maguire et al. 432334) .—a. Petal blade, stamen.—b. Stamen, removed.— rged to show geniculation, beard hairs.—d. Stylar apex, much enlarged over petal, . Two views of capsule parts: left, most of one locule shown, together with contiguous valves (septum shaded dark at lefi) ; right, a fruit section removed to show septum, two placental strands; at ase, a transverse section showing placental strands (center) , valve edges, and septa.—f. Seed. valves with complete septa, at fruiting ma- turity with placentae separating from each other except at base and apex and from the septa. Seeds fusiform, ca. 1-1.3 mm long, densely penicillate pubescent blade. Anthers oblong, 1.5-1.7 mm long, deeply bifid and sagittate, on filaments ca. 1 mm long. Cap- sules narrowly obovoid, 3.5-4 mm long, the Volume 75, Number 2 1988 Kral 619 Xyris deep to pale amber, finely longitudinally ribbed, the fine ribs overlain by a coarser, broader, subanastomosing ribbing. Distribution. Steep bluffs and rocky wet savanna, summits of tepuis, southern Terri- torio Federal Amazonas and Estado Bolivar in Venezuela, east to the boundary of Ven- ezuela with Brazil and Guyana at the type locality: Roraima. Additional specimens examined. BRAZIL: Roraima, Dec. 1909, ba Ule 8547 (L). Guyana: Mt. Roraima, Nov.-Dec. 1931, Abbensetts 20 (U), Summit Roraima, McConnell ^ Quelch 658 (K); Roraima escarpment, u Oct. 1973, Persaud 88 (NY, UC); N slope of Mt. raima, 2,300 m, 16 Feb. 1985, s 14270 (U); Summit Roraima, Autumn 1894, Quelch & McConnell 95 (K); Mt. Roraima, 2,300 m, 16 Feb. 1985, Renz . Roraima, ca. 6,500-7,000 ft. 17992 (MO, VDB); Neblina, upper escarpment slopes E of Camp III, 1,600-1,800 m, 27 Dec. 1953, Maguire et al. ipo S US); Neblina, spy Camp, 1,700 m, 5 Jan. 4, Maguire et al. 37064 (GH, MO, NY, US); Neblina, E of Camp III, 1 an, m, 24 Jan. 1954, Maguire et al. 37371 (NY, us Neblina, banks of Caro Grande, E of Cumbre Camp, 1,100 m, 24 Nov. 1957, Maguire et al. 42200 (GH, K, NY, VEN); Neblina, upper Canon Grande, 1,900 m, 11 Dec. 1957, Maguire et al. 42334 (F, K, NY, US, VEN); Cerro Manibus: cumbre, sección suroriental, 2,685 m, 15 Jan. 1981, Maguire et al. 65634 (NY, VEN); headwaters of Canon Grande, SE portion, 1,900 m, Steyermark 104007 (NY, US, VEN); Cerro Marahuaca-FHUIF, cumbre, altiplanicie de rocas expuestas, 2,330-2,470 m, 3-4 Feb. 1982, Steyermark et al. 126079 (NY, VDB); same area and collectors, 9- 10 Feb. 1982, Steyermark et al. 126313 (NY, VDB, VEN); Cerro Marahuaca, cumbre extremo noreste, 2,580- 2,700 m, 1 Apr. 1983, Steyermark & Delascio 129297 (VDB, VEN); Cerro Aratitiyope, ca. 70 km al SSW de camo con riachuelos afluente al Rio Manipitare, 1,550 m, 24-28 Feb. 1984, Steyermark et al. 130287 (VEN). BOLÍVAR: Alto Caroni, Guyana, del Cerro Yaipan, 7,800 m, F. Cardona 2421 (VEN); Valle d Lado derecho del Salto Angel, Auyantepuy, 15 A 1968, Foldats ae (VEN); Cumbre del tas fin ca. 12 — km al NNE del Kukenan-tepui, 2, 9 Feb. 1984, Huber 9078 (MYF, VDB); Macizo del EE sec- tor centro-noroccidental, 1,350 m, 2 Apr. 1984, Huber 9364 (MYF, leg Cumbre del Ilú rine -)tepui, sector idi 1984, Huber 9493 (MYF, VDB); sección mas Is septentrional del Brazo occidental del I SE de Canaima, 13 Nov. 1984, m, 19 Nov. 1984, Huber 9828 (MYF, VDB); Ea (Mataui-)tepui, c re meridional, 2,700 m, 15 June 1985, Huber & Alarcon 10525 (MYF, VDB); Serranía Guanay, sect. nororiental, cabeceras mas orientales del Rio Parguaza, ca. 1,700 m, 20-28 Oct. 1985, Huber 10968 (MYF, VDB); Altiplanicie del Auyan tepui, sector centro-oeste, 1,860 m . 1986, Huber 11237 (MYF, VDB); Ptari-tepui along trail from camp to SW shoulder of mt., vic. Cave Rock Camp, 1,600-2,000 m, 14-19 Aug. 1970, Moore et al. 9807 (VEN); Mt. Ro- raima, 9,000 ft., H. S. Irwin 441 (NY, US), Uaipan- tepui, summit of West Peak, 1,980 m, 4 Mar. 1967, Koyama & Agostini 7438 (NY, US); Cerro Guaiquinima, 1,500 m, 31 Dec. 1951, Maguire 32883 (NY, US); Cerro Guaiquinima, open savanna l of Cumbre Camp, 1,800 m, 29 Dec. 1951, Maguire 32799 (K, NY, US, VEN); ca. 6.5 km N of Pioneer Monument by Ven. 10, summit La uu ca. 1,200 m, Kral 70395 (F, K, L, MO, NY, US, VDB, VEN); Cerro Guaiquinima, North Valley 1 .600— 1,700 m, 4 Jan. 1952, Maguire 32957 (NY, US, VEN); Ilu- -tepui, Gran Sabana, 6,000-7,000 ft., 16 Mar. 195 NY, S); Mt. Roraima, 2, ermark 58719 (F, NY, US, VEN); Ptari-tepui, 2,000- 2,200 m, 2 Nov. 1944, Steyermark 59736 VEN); Sierra de Lema, base of uppermost waterfall of Rio Chicanan, Steyermark 89551 (NY, US, VEN); Au- pS H Os £ e. x a e w eo 1 (F, K, ro Guaiquinima, 26 Jan. 1977, Steyermark & Dunsterville 113518 (NY, VEN); Ptari-tepui, 2,360-2,420 m, 23 Feb. 1978, Steyermark et al. 115716 (F, MO, US, VEN); Aprada-tepui, 2,460- 2,500 m, 25 Feb. 1978, Steyermark et al. 115895 (F, , NY, U, US, VEN); Auyan-tepui, 1956, Vareschi & Foldats 4907 (VEN). The above citations, while far from com- plete, help illustrate the considerable variation in habit, leaf, and spike characters displayed sometimes even on the same tepui. This is perhaps one of the most distinctive morphol- ogies in the genus, individual plants forming enormous dome-shaped clumps of foliage from which the slender scapes stand out like long pins from a pincushion. Out of so many col- lections, some represented by many dupli- cates, few capsules with full seed can be found, thus this species appears to be one of the poorest seed setters in Xyris. The question then arises as to why it is so widespread and seemingly successful. 40. Xyris xiphophylla Maguire & Ly- man B. Smith, Mem. New York Bot. Gard. 10: 20-21, fig. 6A-E. 1963. TYPE: Venezuela. T. F. Amazonas: abundant in wet escarpment savanna, 4—8 km south 620 Annals of the Missouri Botanical Garden of Cumbre Camp, 1,850-1,900 m, Cer- ro de La Neblina, Rio Yatua, 15 Jan. 1954, B. Maguire, J. J. Wurdack & G. S. Bunting 37312 (holotype, NY; iso- types, US, VEN). Figure 40. Robust, densely cespitose, smooth peren- nial 5-8 dm high, the stems short and stout or up to 5 or 6 cm long; sheaths eciliate, as long as blades or longer, with broad, dark reddish brown or castaneous bases, gradually narrowing to blades and with a triangular erect or divaricate ligule to ca. 2-3 mm long; blades straight, strongly flattened, narrowly lance-linear, 5-10 mm wide, broadening just above ligule, then narrowing evenly to an acute, erect or slightly falcate apex, the tip slightly thickened; margins smooth, edges thin, producing a narrow, subcartilaginous border; surfaces deep green, smooth, with numerous low, wide nerves. Scape sheaths shorter than most leaves, deep brown, multicostate, cari- nate at base, opening and producing a short blade above. Scapes somewhat flattened dis- tally (narrowly elliptic or ovate in cross sec- tion), sometimes slightly ancipital, 2-3 mm wide, the edges blunt, the surfaces smooth, sometimes striate. Spikes narrowly cylindric- obovoid, to ca. 2 cm long, the base acute to attenuate, the apex blunt, the inflorescence of many, loosely imbricate but firm, dark brown or pale brown, dull bracts in nearly vertical rows and without dorsal areas, the sterile bracts much smaller, narrower than the fertile bracts and grading into them, the fertile bracts oblong, 10-15 mm long, api- cally narrowly rounded, with a short, sub- apical, glossy carina or apiculus, the margins entire, aging lacerate. Lateral sepals free, sub- equilateral, lineal, 10-12 mm long, acute, the narrow, dark keel entire or papillate to- ward apex. Petal blades broadly obovate, 6- 7 mm long, yellow, the shallowly rounded apex crenulate. Staminodia not evident. An- thers narrowly oblong, 2-2.5 mm long, deep- ly bifid and sagittate, on short filaments to ca. 0.5 mm long. Capsule cylindric, 6-8 mm long, strongly septate from tip to base, the placentation axile, the placental axis extend- ing 3⁄4 up the fruit. Seeds lance-ovoid, 1.5- 2 mm long, pale to deep amber, finely lon- gitudinally anastomosing-ribbed. Distribution. Wet, rocky savanna, sum- mit elevations, Cerro de La Neblina, Terri- torio Federal Amazonas, Venezuela, locally abundant. Additional specimens examined. VENEZUELA. T. F. AMAZONAS: low slopes near Cumbre Camp, 1,800 m, 4 Jan. 1954, Maguire et al. 37051 (paratype, NY, US); planicie de Zuloaga, Rio aa 2,300 m, 10-15 Oct. 1970, Steyermark 103825 (K, NY, US, VEN). More recent collections, not yet fully distributed, are: Valle de Titirico N of Pico Phelps, ca. 2,300 m, 1 Dec. 1984, Kral 71914; cant 2, 2.8 km NE of Pico Phelps, 2,100 m, open bog, 15 Apr. 1984, Stein & Gentry 1578 (MO, VDB, etc.); al norte del campamento base a lo largo del Rio Mawarinuma, afluente del Rio Baria, 1,880 m, 7-8 Feb. 1984, Steyermark & Luteyn 129819 (MO, NY, VDB) This species is another with the tendency to form frondlike plates of leaves and is su- perficially closest to forms of X. ptariana teyerm. In the savanna of the Valle de Ti- tirico it is the dominant xyrid, its pale yellow flowers unfolding in morning. 41. Xyris spruceana Malme, Bih. Svensk. Vet. Akad. Handl. 26(3)'*: 12, pl. 1, f. 2. 1901. TYPE: Venezuela: ad fluv. Guainia v. Rio Negro supra fl flu- vensis Casiquiari, in 1854, R. Spruce 2993 (isolectotypes, GH, NY). Figure 1. X. applanata Idrobo & Lyman B. m Caldas s 239 fig. 27a-h. YPE: Colom Vaupés: Rio Kananari, cer Tola sare de arenisca, 250 m, 28 1951, R. E. Schultes & I. Cabrera 14466 e COL; isotypes, F, GH, MG, US, VDB). Low, cespitose annual 0.8-2.5 dm tall, the stems contracted. Leaves ascending or spreading flabellately, mostly 5-15 cm long; sheaths from Y as long to longer than blades, eciliate, sharply keeled, brown at very base, above stramineous, pink or pale purple, shad- ing to pale green, progressively narrowing into blade, there either eligulate or with a narrowly triangular ligule less than 1 mm long; blades ensiform-linear, flat, 1.5-3 mm wide, nar- rowing gradually to an acute apex above mid- dle, the margins slightly incrassate, papillose Volume 75, Number 2 Kral 621 1988 Xyris E = = = — ee ===” iis == — FIGURE 40. Xyris xiphophylla (from the type) .—a. Habit sketch.—b. Leaf apex.—c. Sector of leaf blade at widest point.—d. Leaf sheath-blade junction.—e. Spike.—f. Fertile bract.—g. Lateral sepal.—h. Petal blade, stamen.—i. Stylar apex.—j. At left, median long view of capsule, showing placentae; at right, inside view of one valve, cross section of valve to show septum.—k. Seed. 622 Annals of the Missouri Botanical Garden FIGURE 41. Xyris prunis (Clark 6455).—a. Habit sketch.—b. Leaf apex.—c. Leaf sheath-blade junc- tion.—d. Leaf base.—e. —f Fertile bract.—g. Lateral sepal.—h. Petal blade, stamen.—i. Staminode, enlarged beard hair tip. iis p apex.—k. Capsule, showing basal placentation, valve outline. —l. Seed. Volume 75, Number 2 1988 Kral Xyris to scabridulous or minutely ciliate, the sur- faces green, often bordered by a pale brown band and a strong submarginal nerve, smooth, multinerved. Scape sheaths carinate at base, strongly bladed, similar to foliage leaves. Scapes appearing strongly flattened, distally 2-3 mm wide, scape body slightly flattened, usually with 1-2 ribs/side, the lateral costae alate, broader than scape body, ciliolate or papillate or entire. Heads turbinate to hemi- spheric, 0.7-1 cm high, with several spirally and loosely imbricate bracts, involucrate, with at least some sterile bracts leaflike with folia- ceous, spreading or ascending, green-keeled blades, the blades elongate, sharply and cil- iately keeled, spreading or overtopping, long- er than the spike to slightly shorter, and with bases broadly margined, scarious, reddish, fimbrio-ciliate; fertile bracts ovate or lance- ovate, 5.5-6.5 mm long, strongly carinate, the large green, venose dorsal area excurrent to form an acuminate-subulate apex, or mere- ly acute. Lateral sepals free, inequilateral, thin, broadly lanceolate, ca. 4 mm long, acute, the darker firmer keel ciliolate or villose with rusty hairs. Petal blades elliptic, 3.5-4 mm long, yellow, subacute, the margin undulate. Staminodia bibrachiate, the oblong, flat branches long-penicillate-ciliate. Anthers ob- long, ca. 1 mm long, deeply bifid and sagit- tate, on filaments ca. 1 mm long. Capsule obovoid, 2-3 mm long, placentation basal, the valves without septa. Seeds numerous, ovoid to ellipsoid, ca. 0.5 mm long, apiculate, lustrous but opaque, with ca. 14 strong, dark brown, longitudinal ribs. Distribution. Locally common in wet, sandy, savanna-forest transition or in white- sand savanna, along the Rio Negro and trib- utaries from Amazonian Brazil (Amazonas) northward into southern Colombia and south- ern Territorio Federal Amazonas, Venezuela. Additional specimens examined (only Colombian and Venezuelan records are cited here). COLOMBIA. VAUPES: Rio Apaporis, Hanon 250 m, 26 Nov. 1951, Garcia- oak y 13729 (NY); Puerto Colombia (opp. Maroa), , 12 Oct. 1957, Maguire et al. 41848 (F, NY); E Negro, El Castilo, San Filipe, 12 Dec. 1947, Schultes & López 9333A (NY); Cachivera de Jirijirimo, ca. 250 m, 13 June 1951, Schultes & Cabrera 12392 (GH, NY); Rio Apaporis, Cachivera de Jirijirimo, 13 June 1951, Schultes & Cabrera 12392 (GH); same locality, 12 Aug. 1951, Schultes & Cabrera 13512 (GH, NY); same lo- cality, 16 Sep. 1951, Schultes & Cabrera 14055 (F, GH); same locality, 21 Jan. 1952, Schultes & Cabrera 14949 (COL, F, GH, US, VDB); Rio Kananari, ca. 250 m, Cachivera duin 25 July 1951, Schultes & Cabrera 13187 (GH); Cerro Isibukuri on uro 6 Aug. 1951, Schultes & Cabrer 13401 (GH); Cerro Yapoboda, ca. 450 m, 5 1951, Schultes & oa 14263 (GH); Río Ka nde) ie eee Isibukuri, 29 Oct. 1951, Schultes 14535; Cerro Er Schultes & Cabrera rion F, GH, MG, NY, US, outh of Kananari, ca. ET Jan. 1949, Schultes < Cabrera 14949 (F, GH); r: Piraparana, 952, Schultes & Cabrera 17116 (GH); 18 Sep. 1959, Schultes E — 17504, 17553 (GH); Rio Pg de San Felipe and v Puerto 2 Nov. 1952, e ulis: et i 18189 (GH); Rio Piraparana, Cura -ree-ee-ko-mee-o-kee, 30 p. 1952, Schultes & Cabrera 17499 (F, GH, : i ribe, 2 Nov (GH); same locality, Nov. 1951, Schultes & Cabrera 19732 (GH); Mitu and vic. along Rio Vaupés at Sirius 14 Sep. 1976, Zarucchi 2049 (GH). VENEZU E. AMAZONAS: San Carlos de Río Negro, 28 Sep. 1978, Clark Clark 6455 (MO, NY); raudal “‘pereza”’ en el Rio 9 Nov. 1984, Guanchez & Melgueiro 3426 (TFAV, VDB); km 11 de la carretera San Carlos-Solano, 16 Sep. 1980, Huber et al. 5677 (US); carretera San Carlos- Solano, entre los km 4 y 20, 15 Sep. 1980, Huber & Medina 5639 (US); 4 km E San Carlos de Rio Negro, 11 Nov. 1977, Liesner 3365 (NY, MO); 1-2 km SE and E of San Carlos, 22 Apr. 1979, Liesner 6870 (MO, VDB); Rio Guainia, Yavita-Pimichin trail near Pimichin at 140 m, 22 Nov. 1957, Maguire et al. 36337 (GH, NY); Bana, 2 km N de San Carlos, 6 Feb. 1977, Morillo & Hasegawa 5155 (VEN); same locality, 6 Feb. 1977, Morillo & Villa 5355 (VEN). A number of specimens are cited above to illustrate the degree of variation in involucral bract and fertile bract in the species. This variation includes, as a continuum, X. ap- planata Idrobo & Lyman B. Smith. 42. Xyris uleana Malme, Repert. Spec. Nov. Regni Veg. 3: 113. 1906. TYPE: Auf nassem Sandbogen, BI. Gelb., Man- aos, Rio Negro, Mai 1902, Amazonas Expedition E. Ule, Herbarium Brasi- liense No. 6171 (lectotype, B; isolecto- types, GH, L, U; phototype, F). X. ro Malme, Bull. Ln Club 58: 325. 193 x it YPE: Venuezuela. T mazonas: summi 624 Annals of the Missouri Botanical Garden of Mount Duida, 4,000 ft., Gorge of Cano Negro, G. H. H. Tate 811 (lectotype, NY; ). X. — Idrobo & Lyman B. Smith, Caldasia 6(29): 1-238, fig. 26a-f. 1954. This complex of little plants varies so much as to leaf and scape, trichomal features, rel- ative length of spike and sterile bracts, and degree of sepal connation, that it is under- standable that several variants have been de- scribed as species. Certainly two varieties ap- pear, as follows: KEY TO VARIETIES OF XYR/S ULEANA — a. eee ii ancipital, the costae produced comprising the Pan leaf blades over 1 mm wide 42a. uleana var. EUM . Scapes at most low-costate, terete, rarely w l-few low costae; leaves 1 mm wide or Ae 42b. X. uleana var. angustifolia - c 42A. Xyris uleana Malme var. uleana. Figure 42 Cespitose, soft-based annual 0.6-1.5(-2.2) dm high, the stems contracted. Leaves erect to spreading flabellately, 3-10 cm long; sheaths eciliate, keeled, 12 as long as blades or more, brownish or tan, narrowing gradually to blades and usually eligulate; blades linear, strongly flattened, 1-3.5 mm wide, tapering to incurved-acute tips; margins thin or com- prising a yellowish-incrassate nerve, smooth or papillate toward leaf apex; surfaces green or tinged with pink or maroon, finely nerved or with some strong, yellowish nerves, gla- brous. Scape sheaths mostly shorter than leaves, keeled and multicostate proximally, distally with strong blades similar to leaves. Scapes straight, slightly twisted, appearing strongly flattened, 1-3 mm wide, with scape body actually often round and with 2 lateral costae, alate, in combined width broader than scape body, the wings smooth or scaberulous- ciliate, sometimes scape faces with | or 2 more low costae. Spikes broadly to narrowly ovoid, 3-5(-10) mm long, acute, of several tightly imbricate bracts, these with large dor- sal areas, otherwise thin, scarious, papillose, pilose-ciliate, the lowest pair sterile, often with backs alate-keeled, with excurrent-bladed dorsal areas, often longer than the other bracts, sometimes exceeding spike and with tips incurved-cucullate or erect; fertile bracts 2-3.5 mm long, broadly obovate or subor- bicular, cucullate, broadly rounded at apex, ecarinate, papillate, the margins usually pi- lose-ciliate. Lateral sepals free or connate at base, strongly inequilateral, 2.5-3 mm long, unt, the broad, curvate keel ciliolate and/ or papillate. Petal blades very broadly ob- ovate, ca. 2 mm long, yellow, the wide, shal- lowly rounded apex coarsely few-toothed. Staminodia bibrachiate, the narrow, flat branches long-penicillate-ciliate. Anthers lance-oblong, ca. 0.5 mm long, deeply bifid and auriculate, on flat filaments ca. 0.5 mm long. Capsule broadly obovoid, ca. 1.5 mm long, the placentation basal. Seeds narrowly ovoid or ellipsoid, 0.5-0.7 mm long, pale amber, longitudinally with many wavy, nar- row, sharp lines, sometimes also with a few, irregular, stronger, dark red ribs. Distribution. Sandy wet savanna at low to high elevations, northern South America west of the Andes, the Amazon system and tributaries, southern Colombia east to the Guianas, also including the upper Orinoco and tributaries; northern Brazil from Mato Grosso nd Amazonas east into Pará. Only some of the collections from Colombia, Venezuela, and the Guianas are cited below. Additional specimens examined. COLOMBIA. VAUPES: Rio Atabapo, 1 km W Cacagual, Maguire et a 36293 (NY, VEN); Rio Kubiyú, Cerro Kanenda, ca. 800-900 ft., 10 Nov. 1952, Schultes & Cabrera 15398 (id. as X. applanata Idr. & Sm.—COL, F, GH, MG, US, VDB), 18337 (GH), 18398 (COL, GH, VDB); Rio Kuduyari, Yapoboda, ca. 900-1,000 ft., 18 Nov. 1952, Schultes et al. 18531 (F, GH). GUYANA: Kaieteur Savanna, 1,200 ft., 7 Sep. 1937, Sandwith 1422 (K, U); Kaietuk savana, 1,100 ft., 20 Aug. 1933, Tutin 497 (K, U); Kaietur summit, 1,100 ft., 31 Aug. 1959, Whitton 217 (K SURINAM: wet-sand savanna in upper Sipaliwini area, 4 tapoe, savana-forest, 14 Aug. 1933, Lanjouw 433 (U); Lanjouw 433a (an older sara) VENEZUELA. T. F. AMAZONAS: laja en raudal **pereza" en el Rio Autana, 9 Nov. 1984, Guánchez & Melgeiro 3422 (TFAV, VDB); hasta el pie occidental del Cerro Yapacana, ca. 100 m, 14-28 Feb. 1978, Huber 1600 (US); al E del Cano Perro e Agua a unosa "n m al SE de la confluencia Orinoco- -1 Dec. 1978, Huber & Tillett 2814 otua y la base occidental del Cerro Yapacana, Huber & Tillett 3014 (US); Cerro Mo- riche, 120 m, 19 Feb. 1979, Huber 3198 (US); base of PN Volume 75, Number 2 Kral 625 1988 Xyris RO”, ^ "at V 05 mm Xyris uleana (Kral 70610) .—a. Habit sketch. —b. Leaf apex. —c. Leaf sheath— blade junction. — . Leaf base.—e. Spike. —f. Fertile bract.—g. Lateral sepal.—h. Petal blade, stamen.—i. Staminode, enlarged apical part of beard hair.—j. Stylar apex.—k. Dehisced capsule showing basal placentation.—l. Seed. FIGURE 42. 626 Annals of the Missouri Botanical Garden Cerro Yapacana, Kral & Huber 70680, 70701 (F, K, MO, NY, SP, US, VDB, VEN); 30 km N of Puerto Ayacucho, 80 m, 5 Nov. 1980, Maas & Huber 5149 (U, US, VDB); Salto Yutaje, 8 Nov. 1980, Maas & Steyermark 5174 (U, VDB, VEN); summit Cerro Gua- nay, 1,800 m, 2 Feb. 1951, Maguire et al. 31704 (id. as X. duidensis—GH, K, NY, US, VEN); near Pimichin, 140 m, 22 Nov. 1953, Maguire et al. 36338 (GH, NY, US); Cerro Sipapo, 1,400 m, 25 Dec. 1948, Maguire e Politi 27916 (NY); ibidem, savanna vic. base camp, 5 m, 30 Dec. 1948, Maguire & Politi 28041 (NY); p Falls, Rio Cuao, Rio Orinoco, 30 km below La Urbana, 14-15 May 1949, Maguire & Politi 29083 (F, US); Cerro Duida, Rio Cunucunuma, Cano Culebra, T m, 18 Nov. 1950, Maguire et al. 29519 (NY, US, VEN) summit Cerro Duida, Brocchinia Hills, 1,700- 1.980 m, 1 Sep. 1944, E 58145 (id. as X. duidensis —F, GH, , VEN); topotype, Fe DE 266 (VEN); top of Salto Aicha near di of Uaipan tepui, 1,100 m, 27-28 Nov. 1982, Dav & Huber 22857 dr VDB); Vallé VR a Guayay sector nor- Oct. 1985, jn 10967 (MYF, VDB); oot of erm of Uaipan- iR 1,200 m, 6 Mar. 1967, Koyam ma & Agostini 7513 (NY, US, VEN); along Rio Karuai, avanayen, 1, 220 , 30 Nov. 1944, Steyermark 60823 (F, US, VEN); Sierra Ichun, Salto aria Espuma del Rio Ichun, 28 Dec. 1961, Steyermark 90340 (F, NY, US, VEN); Sierra Pakaraima, 4-5 Ma 1973, Steyermark 107266 (F, MO, U, VEN, US); cumbre de Cerro Guaiquinima, Salto del Rio Szczerbanari, 750 m, 20-25 Jan. 1977, Steyermark et al. 113150 (MO, NY, US, VEN); Cerro Marutani, 1,200 m, 11 Jan. 1981, Steyermark 123881 (NY, VDN). Two Kral numbers from the Gran Sabana collected with A. Gonzalez in 1983 and with distributions to F, MO, NY, US, VDB are Kral 70389 and 70610; two more from the Gran Sabana in Dec. 1984 and awaiting full distribution are 72157 and 72192. This variety is one of the weediest of the annual Xyris species, coming in solidly in disturbed sandy moist savanna and sandy washes in the Guayana Highland. It can, in dense populations, range from dwarf plants a few centimeters tall to quite tall. Generally, well-spaced individuals on moist, peaty sites range between one and two decimeters in height. The pale yellow flowers open in the morning 42B. Xyris uleana var. angustifolia Lanj., Kew Bull. 1939: 562. 1939. TYPE: Guyana: Kaieteur Savanna, in damp sand, ca. 1,200 ft., petals deep orange- yellow, 7 Sep. 1937, N. Y. Sandwith 1421 (holotype, K; isotypes, NY, U). X. connectens Malme, Ark. Bot. 19(13): 2. 1925. TYPE: Brazil. Pará: cerca Belem, Ducke (type at BM). The plants usually more slender and lower than in the type variety, the leaf blades nar- rower, mostly 0.5-1 mm wide, the edges smooth or papillate, the lower bracts typically shorter than the larger fertile ones, all bracts with papillate-tuberculate dorsal areas as in the type variety but bracts themselves with acute tips and entire to lacerate, eciliate mar- gins. Lateral sepals mostly free. Distribution. Low- to medium-elevation savanna, from Territorio Federal Amazonas in Venezuela west through southern Estado Bolivar to Guyana, infrequent. Also in Ama- zonas and Para, Brazil. Addit tonal specimens examined. VENEZUELA. T. F. ple paloma 80 m, is Nov. 1980, Maas & Huber Mt. Aymatoi, 1,150 m, 15 Oct. 1981, Maas et al. 5687 (U); Kaietuk Savanna, 1,100 ft., Aug. 20, 1933, Tutin 197 (K). This low variety of X. uleana is often in mixed populations with the type variety but is less frequent and could be mistaken for smaller, shorter-spiked X. paraensis var. paraensis. However, the dorsal areas of the latter are not strongly papillose-tuberculate. The terete scapes have sharp but low costae, very unlike the usually winged costae of the type variety. 43. Xyris calderonii Kral, Lyman B. Smith & Wanderley, sp. nov. TYPE: Bra- zil. Amazonas: Transamazon Highway, 9 km W of Rio dos Pombos, ca. 1.5 km E of Igarapé dos Pombos, and ca. 64 km E of the Aripuanà. Common in white sand campina, flowers yellow, 18 June 1979, Cleofé E. Calderón, O. P. Mon- teiro & J. Guedes 2549 (holotype, INPA; isotypes, US, VDB). Figure 43. Planta humilis, annua, praeter inflorescentiam glabra. Radices filiformes. Folia linearia, solum basalia, 3-7 cm longa, flabellate expansa, vulgo vaginis scaporum (longiora, Laminae vaginis 2-4-plo longiores, 0.5-1 mm latae, pla- Volume 75, Number 2 1988 Kral 627 Xyris nae, rectae, longitudine paucicostatae, a basi ad apicem pansae, acies integrae. Vaginae scaporum laxae, plerum- que apertae, rectae, carinatae, laminis aut similis laminis foliorum aut bre vibus. Scapi tidy d filiformes plus iter torti, 5-10 cm .5 mm crassi, distaliter dh multicanal costis s hat ihe Spicae sub- globosae vel late a -5 mm longae, pluriflorae, . Bracteae steriles 2-4, subdecus- Tnimnnsve um, areis dorsalibus Hs c viridibus et x Momia similis ne pes hos triangulatis; per intimum ovatum, ario acutum vel acu minatum, profundo. ee trar a areis ais dorsalibus veli papillosis, vulgo sine laminis. Bracteae fertiles late ovatae vel suborbiculatae, 2.5-3 mm longae, valde rotundato- convexae, villosiciliati, areis dorsalibus ovatis, valde gra- nulato-papillosis. Sepala lateralia libera, oblonga vel ovata, m longa, valde inaequilatera; ala carin nali lata, a basi ad deco. integra, a medio di E ns F cillatis. Antherae oblongae, sagittatae, ca. 0.5 mm longae; "esa ëm ca. 0.8-1 mm longa. Capsula matura late ob- ata, planoconvexa, ca. 1 mm longa, placenta basalia. Semina late cllipsoiden, ca. 0.3 mm longa, pallide brun- neola, plus minusve reticulata. Low annual, smooth except for the inflo- rescence. Roots filiform. Leaves linear, strict- ly basal, 3-7 mm long, spreading flabellately, commonly longer than the scape sheaths. Lea blades 2-4 times longer than sheaths, 0.5- 1 mm wide, flat, straight, longitudinally few- costate, flattened from base to apex, brown to yellow-green, the tips abruptly narrowed, incurved-acute, the margins entire, not thick- ened; sheaths carinate, brown, many-ribbed, scarious except for the ribs, with the thin edges stramineous, gradually narrowing into the blades or apically producing a short, scar- ious, broad ligule, gradually dilating toward base, the edges entire. Sheaths of scape lax, mostly open, straight, carinate, with blades similar to those of foliage leaves or shorter. Scapes subterete, filiform, + spirally twisted, 5-10 cm high, 0.4-0.5 mm thick, distally with many low, smooth costae. Spikes sub- globose to broadly ovoid, 3-5 mm long, sev- eral-flowered, obtuse, involucrate. Sterile bracts 2—4, subdecussate, villous-ciliate, the lowermost pair foliaceous, rigid, 2-5 times longer than the spike, lanceolate to oblong, 2-3 mm long, carinate, the dorsal areas lin- ear, green, with blades similar to those of foliage leaves but triangulate; inner pair (if present) ovate, convex, scarious, acute, very villous-ciliate, the dorsal areas strongly pap- illate, without blades. Fertile bracts broadly ovate to suborbicular, 2.5-3 mm long, strong- ly rounded-convex, villous-ciliate, the dorsal areas ovate, strongly granular-papillose. Lat- eral sepals free, oblong to ovate, 2-2.5 mm long, very inequilateral, with keel broad, dis- tantly ciliate from base to middle, lacerate toward apex. Petal blades broadly obovate, ca. 3 mm long, yellowish, broadly rounded and strongly erose at apex. Staminodia bi- brachiate, the branches sparsely long-peni- cillate. Anthers oblong, sagittate, ca. 0.5 mm long; filaments ca. 0.8-1 mm long. Mature capsule broadly obovoid, plano-convex, ca. 1 mm long, the placenta basal. Seed broadly ellipsoid, ca. 0.3 mm long, pale brown, + reticulate. There is no question that the affinities of this little plant are with X. uleana Malme; in the production of narrow leaf blades it is most similar to var. angustifolia Lanj., which also sometimes produces long-tipped basal bracts. However, the lateral sepals are smaller with a somewhat different keel configuration, and the dorsal areas are consistently long-excur- rent to produce acicular blades several times longer than the subtended spike. The scapes are uniformly terete. 44. Xyris esmeraldae Steyerm., Field- iana, Bot. 28(1): 109, fig. 16c. 1951. TYPE: Venezuela. T. F. Amazonas: Es- meralda Savanna, 200 m, Cerro Duida, 22 Aug. 1944, J. Steyermark 57821 (holotype, F; isotype, NY). Figure 44A (44B—see synonym). Xyris V qaw Maguire & Lyman B. Smith, Mem. New York Bot. Gard. 10: 29, fig. 13A-E. 1963. TYPE: Venez ana azonas: Cerro Sipapo (Pa- raque), frequent along banks of small shaded stream savanna about pool, Cario Negro, 1,300 m, 15 Dec. 628 Annals of the ride Botanical Garden í cem FIGURE 43. Xyris calderonii (Calderon et al. 2549) .— aod uio —b. Leaf at midblade.— c. Leaf bla sheath e side (left) and ventral (right) views. d Mat spi ike.—e. Fertile bract. —f. Petal Ape ne stamen.—g. Stylar apex, staminode.—h. Lateral sepal, ene —i. Seed 1948, B. Maguire & L. Politi 27699 (holotype, NY; isotypes, US, VEN). Figure 44B Low, slender, cespitose, smooth to papil- lose-rugulose annual 0.5-3 dm high, the stems contracted, the roots filiform. Leaves ascend- ing or spreading flabellately, 1-5 cm long, often absent by full anthesis or represented only by scalelike sheaths; sheaths of longer Volume 75, Number 2 Kral 629 1988 Xyris foliage leaves less than 1⁄2 as long as blades, Distribution. Low (mostly under 500 m) pale lustrous brown, tapering gradually from savanna, southeastern Colombia eastward base to blade, there with a narrowly trian- gular, erect ligule to 2 mm long, at this level broader than blade is wide, the blades linear, flat or somewhat twisted, much compressed, ca. 1 mm wide, green or maroon, gradually narrowing toward apex, to a narrowly acute or acuminate tip, this dorsally scabrid, the margins mostly pale-incrassate, scaberulous or smooth, the surfaces strongly few-ribbed, smooth, rugose or rugulose. Scape sheaths shorter than most leaves, the base tubular, ribbed, keeled, the apex short-bladed or with a cusplike blade. Scapes much twisted, straight or flexuous, flattened distally, ca. 0.8-1 mm wide, ancipital with smooth or tuberculate- scabrid costae making edges, the surfaces ribless or with 1-2 ribs per side, smooth or rugulose-scabrid. Spikes elliptic-lanceolate, 3.5-5 mm long, acute, uniflorous, subtended by a narrow castaneous collar, of few, strong- ly convoluted, erect, reddish brown, decussate bracts with distinct dorsal areas; sterile bracts 2 pair, shorter than the fertile bracts, the lowest pair ovate to lance-triangular, acute, strongly keeled, the inner pair ovate-trian- gular, slightly keeled, distally with a strong, narrow midnerve, acute or narrowly rounded, entire or lacerate; fertile bract solitary, nar- rowly to broadly obovate, much inrolled, ca. 4-6 mm long, broadly acute or narrowly rounded, with broad dorsal area and strong midnerve. Lateral sepals ca. 1⁄4 connate, 4— 4.5 mm long, the lobes oblong, subequilateral but excentrically folded, pale red-brown, acute, the narrow firm keel scabrociliate at least from middle to apex. Petal blades ob- ovate or elliptic-oblong, 2-3 mm long, yellow, the rounded apex lacerate-dentate. Stami- nodes either not evident or with obreniform, beardless blades. Anthers oblong, ca. 1 mm long, deeply bifid and sagittate on filaments ca. 1 mm long. Capsules obovoid, 2-3 mm long, the valves eseptate, the placentation basal. Seeds several on long funiculi, ellipsoid, 0.7-1 mm long, including a short, silvery apiculus (outer integument), the body deep amber, finely longitudinally and spirally anas- tomosing-ribbed. along the upper Rio Orinoco and the Rio Negro and tributaries in Venezuela. Additional specimens examined. COLOMBIA. - An- aquetá, Araracuara, sabana de la ean 14165 (NY); Araracuara, Rio Caquetá, 5 Sep. 9, Maguire et al. 44131 (NY); Río Miritiparaná, 8 May 1952, Schultes & rape 16410 (GH, NY). VAUPÉS: Rio Vaupés, Caño Pact, 6 Mar. 1944, Schultes 5830 NY); Rio Kananari, Cerro Isibukuri, 29 Oct. 1951, Schultes & Cabrera 14536 (GH, VDB); Rio Piraparana, Cano Paca, 19 Sep. 1972, Schultes & Cabrera 17567 (GH, VDB); Rio Kananari, Cerro Isibukuri, 28 Oct. 1951, Schultes & Cabrera 14445 (GH); Rio Piraparana, Raudal Na-hoo- ae 30 Aug. 1952, Schultes & Cabrera 17110 (GH); S ilipes & vic., ca. 600 ft., 25 Oct. 1952, Scholes et al. 18026 (GH); Rio Parana Pichuna, ca , June 1953, Schultes & Cabrera 19933 (GH). VENEZUELA. T. F. AMAZONAS: Ri cuy, 2 Mar. 1944, Baldwin 3240 (VEN); of Cano Yagua at Cucurital de Yagua, 8 May 1979, Davidse et 2 17376 (MO, VDB); Cerro Duida, 1,000 m, Jan.-Feb. 1969, Farinas et al. 436 (NY, VEN); S de la Sierra de Untuyan en las cabeceras del Río Mavacá, 550 m, Guánchez 805 (TFAV, VDB); “Cerro la Trampa” al norte del medio Rio Autana, 12 Nov. 1984, Guánchez & Melgueiro 3531 (TFAV, VDB); IVIC Study area 4 km E of San Carlos de Rio Negro, 120 m, 12 Nov. 1980, R. Liesner 3381 (MO, NY, VDB); 10 km NE of San Carlos, white sand area, 7 Apr. 1979, Liesner 6286 (MO); Cerro Ed La Neblina, Puerto Chimo Camp on Rio Ma- warinum km E of Neblina Base — 150 m, 13 Feb. 1984, Lord 15884; Cerro Sipapo , savanna vic. Pe Camp, 3 1948, Maguire & Politi 7 es Gar Simp. hanks along lower Cano Negr 8, Ma e & Politi 27916 (NY); Rio Guai. nia, Hoe nc nhé near Pimichin, 22 Nov. ~ , 20 Apr. 1970, Steyermark & Bunting 102815 (U, US, VDB, VEN). It becomes necessary to consider this the same as X. rugulosa Maguire & Lyman B. Smith, which does not differ in any essential character but only in the tendency in the latter toward narrower bract tips and more leafiness (cf. Fig. 44A, B). The rugulous epi- dermal feature supposed to be critical for X. rugulosa appears also on leaves of X. es- meraldae. Some of the trouble in past inter- pretation must come from the fact that X. esmeraldae proper tends to lose most of its larger foliage leaves as the plants approach seeding. However, when one finds these leaves, they are remarkably the same as those of rugulosa 630 Annals of the Missouri Botanical Garden idm FIGURE 44A. — Xyri raldae (Liesner 7035) .—a. Habit sketch. —b. Leaf tip. —c. Leaf, midblade to base. — d. Spike.—e. Lateral iria two views.—f. Petal and stamen.—g. Staminode.—h. Stylar apex.—i. Capsule showing placentation.—j. See Volume 75, Number 2 Kral 631 Xyris À. -—— m Area ste Pots C E Mi ser AT Ano namo ufo eni sen FicuRE 44B. Xyris esmeraldae (from type of X. o uA —a. Habit sketch. — b. vit A Sed a xcd th— blade hacian — d. Leaf base.—e. Whole leaf.—f. Spike.—g. Lateral ee two view tal blade, stamen.—i. Stylar apex.—j. Capsule, eed. 632 Annals of the Missouri Botanical Garden 45. Xyris subuniflora Malme, Rec. Trav. Bot. Neerl. 9: 129. 1912. TYPE: Dutch Guayana: "In arenosis humidis, Suri- name," Splitgerber 990 (lectotype, L; isolectotype, U). Figure 45. Xyris d Fieldiana, Bot. 28: 107. 1951. nezuela. T. F. Amazonas: between Esme- lla Savanna and southeastern base of Cerro Dui- a, 200 m, 22 Aug. 1944, J. A. Steyermark 57835 (a DA F; isotypes, GH, NY, VEN). Densely cespitose, low and delicate, smooth, annual or short-lived perennial 3-10 cm high, the stems contracted. Leaves erect or spread- ing flabellately, 1-7 cm long; sheaths eciliate, soft, 1⁄2 as long as blades or less, the bases tan or pink, keeled, tapering evenly into blades or with a short, erect, triangular ligule less than 1 mm long; blades filiform but flat, 0.3— 0.5 mm wide, tapering slightly above middle, then abruptly acute-apiculate or incurved- acute, the margins entire, the surfaces very finely nerved, maroon with olive tints. Scape sheaths ca. as long as leaves, twisted, tubular proximally and multicostate, distally open with a strong flat blade as in leaves. Scapes filiform, straight or flexuous, twisted, ca. 0.3 mm thick, distally terete, ecostate, finely striate, oliva- ceous to maroon. Spikes narrowly ellipsoid or linear-ellipsoid, drying lance-ovoid, 4-5 mm long, few-bracted, uniflorous; bracts with dis- tinct lanceolate dorsal areas, the sterile bracts .5-3 mm long, lanceolate, navicular, acute or narrowed-retuse; fertile bracts oblong to ovate, conduplicately folded around floret, ca. 4 mm long. Lateral sepals ca. V$ connate, ca. 3-3.5 mm long, the narrowly triangular, erect lobes inequilateral, acute with very low smooth keels similar to dorsal areas of bracts but narrower. Petal blades broadly obovate, ca. 3 mm long, yellow, the broadly rounded apex denticulate-lacerate. Staminodia bibrachiate, arising from just below sinus between petal blades, the branches short-penicillate-pubes- cent. Anthers oblong, ca. 0.8 mm long, pp bifid and sagittate, on filaments ca. 0.5 m long. Capsule broadly ellipsoid, brown, ca. L5 mm long, the placentation basal. Seeds few, broadly ellipsoid, 0.5-0.6 mm long, apiculate, dark brown and opaque, rather coarsely lon- gitudinally ribbed and cross-ribbed. Distribution. | Low-elevation savanna, lo- cally abundant, southeastern Colombia (Vaupés) eastward into Surinam and in con- tiguous Pará, Brazil. Additional specimens examined. COLOMBIA. VAUPÉS: cano del Caribe and vic. ., 850-900 ft., 2 Nov. 1952 16 June 1957, ee 347, 781 (U); Zanderij savan- na, Jansma 1 (U); via secta ab Moengo tapoe ad Grote Zwiebelzwamp, savanna near km 10.9, 8 Oct. 1948, Lanjouw & Lindeman 720 (Uy; ibidem, ridge E of camp, 14.9 km, 20 Oct. 1948, Lanjouw & Lindeman 128a (U); Bronniveau Brinkheuvel, in kleine polletjes, Natur- reservaat Brinkheuvel, 11 Oct. 1967, Teunissen & Wild- schut LBB 11910 (U). VENEZUELA. T. F. AMAZONAS: entre Yavita y Maroa, ca. 204 km hacia Maroa (al sur) desde el empalme con la carretera Yavita-Pimichin, 125-140 m, 6- 19 July 1969, Bunting et al. 3925 (NY, U, VDB); carretera San Carlos-Solana, 120 m Huber et al. 5666 (US); ca. 20 km al SW de Maraca, Serranía del Atila, 760 m, Huber 6164 (US); rg en la mar el bajo Rio Pasimoni, 8 Feb. 1 Huber & "mas 5855 (VDB, VEN); 9 km NE < Carlos, 120 m, 26 Nov. 1977, Diener. 3913 (MO, NY, VDB); rios Pacimoni, Yatua, Casiquiare, 110 m, 28 Sep. 1957, Maguire et al. 41629 (GH, NY, VEN); Triana Savanna, Cerro Pitón, Maguire et al. 53595 (NY); Bruno Guaibana), laja de roca ignea, 100 m, 9 Apr. 1970, Steyermark A Bunting 102490 (F, US, VDB, nas terreno arenoso en el camino de Yavita, 128 m 1942, L. Williams 13996 (F, US, VEN). BOLÍVAR: savana bordering forest of Rio Karuai, between Kavanayen and base of Ptari-tepui, 1,220 m, 18 Nov. 1944, Steyermark 60603 (F, NY, VEN). ~ Perhaps the smallest, certainly the most slender, species of Xyris, very easily over- looked among the masses of other weedy Xy- ris of open savanna. 46. Xyris connosepala Lanj. & Linde- man, Bull. Torrey Bot. Club 75: 639. 1948. TYPE: Surinam: Tafelberg (Table Mountain). Frequent, wet sphagnum- lled cracks in rocks, Savanna No. IV, 16 Aug. 1944, B. Maguire 24395a (ho- lotype, NY; isotype, U). Figure 46. Delicate, cespitose, rosulate, smooth an- nual 1-2 dm high. Leaves spreading-ascend- BU Volume 75, Number 2 Kral 1988 Xyris 633 T Atii Toes SERA xy SE EO READE POPE T EOS y u — -— = = — —— — A, f 1 Ui FIGURE 45. Xyris subuniflora (Davidse et al. 16928) .—a. Habit sketch.—b. Leaf apex.—c. Leaf blad sheath junction.—d. Lea —e. Spi base. ke.—f. Fertile bract.—g. Lateral sepals (inside view) .—h. Petal blade, corolla showing attachment of two a —i. Staminode.—j. Stylar apex.—k. Dehisced capsule.—l. 634 Annals of th Missouri Bo Garden Zam Ə 0.5 mm FicURE 46. Xyris e (from the type) .—a. Habit sketch.—b. Leaf tip.—c. Leaf at sheath-blade dace da —d. Leaf base. — — f. Lateral sepals, front and back view.—g. Petal, stamen.—h. Staminode. — style apex.—yj. pane at ‘eft open to reveal placenta, at right a valve.—k. Seed Volume 75, Number 2 1988 Kral 635 Xyris ing, 2-5 cm long; sheaths reddish brown, ca. 1⁄2 as long as blades, tapering evenly from keeled base to blade, or with an erect, nar- rowly triangular ligule to 0.5 mm long; blade flat, linear, often twisted, 0.8-1.2 mm wide, the apex incurved-acute, slightly thickened, papillose-tipped, the margin a narrow, pale, cartilaginous band, the surface green or ma- roon, finely nerved. Scape sheaths ca. the same length as leaves, proximally tubular, multicostate, opening toward middle, at apex bearing a blade similar to that of leaves. Scapes filiform, maroon, terete, with 1 low but strong costa. Spikes ellipsoid, in fruit broadly ob- conic, ca. 0.5 mm long, red-brown, of 2-3 flowers. Sterile bracts ca. 4, decussate, tri- angular-ovate, keeled, smaller than and grad- ing into the fertile bracts, these 3-4 mm long, ovate, acute, entire, the backs ecarinate, with prominent lanceolate dorsal areas, excurved in fruit. Lateral sepals ca. 4 mm long, connate in basal 1⁄4, the lobes subequilateral, oblong, acute, pale brown, firm, keel in middle 4% papillate-tuberculate. Petal blades broadly ob- ovate, 3 mm long, yellow, the narrowly round- ed apex erose. Staminodia bibrachiate, the flattened branches with sparse penicillate hairs distally. Anthers oblong, ca. 1 mm long, deep- ly bifid apically, auriculate basally, on fila- ments ca. 0.5 mm long. Capsule ellipsoid, ca. 1.5 mm long, placentation central. Seeds nu- merous on elongate funicles, ellipsoid, ca. 0.5 mm long, biapiculate, deep brown, many- ribbed longitudinally. Distribution. Rocky moist savanna, known only from the type locality (additional material, Kramer & Hekking 2940, U). This species is so similar to the widespread X. guianensis as to make the observer doubt their distinctness. However, until more col- lections from the region show intermediacy in sepal connation, it is perhaps best to include it here as distinct. 47. Xyris guianensis Steudel, Syn. Pl. Glum. 2: 285. 1855. TYPE: Guayana: "Guiana anglica. Schomburghk No. 1038” (lectotype, K; isolectotypes, K, L). Figure 47. Xyris V Malme, Bih. Svensk. Vet. Akad. Handl. 26, ig. 8, pl. 1, f. 1. 1901 (lectotype, S; ne eb Y). Xyris yaran Malme, Repert. Spec. Nov. Regni Veg. 2. 1906. TYPE: ies Amazonas: “In arenosis as s, Amazonas, Brasil" Ule 6172 Cae S DM L, NY, US). Low, densely tufted, smooth annual 0.5- 3 dm high, the stems mostly contracted, sometimes 1-2 cm long. Principal leaves spreading flabellately, often maroon or red- brown, 2.5-7 cm long; sheaths entire, glossy red-brown, 4 the blade length or less, keeled, the keel often papillate-ciliolate, incrassate, tapering gradually to blade, there often with a scarious, narrowly triangular, erect, ligule to 2 mm long or eligulate, the blades flattened, linear, often twisted, 0.5-1 mm wide, the apex narrowly acute to acuminate, the tip sometimes with a tuft of scabrosity, the mar- gins narrow, incrassate, a pale or dark, smooth band, the surfaces finely multinerved and smooth, often with strong maroon tints. Scape sheaths longer than to slightly shorter than leaves, below terete and deep glossy red- brown, distally opening and keeled, producing a blade similar to leaf. Scapes filiform, twisted, sometimes flexuous, ca. 0.5 mm thick, distally terete or slightly compressed, ecostate to low- bicostate, the costae smooth. Spikes ellipsoid, drying obovate or turbinate, 4-7 mm long, pale red-brown with 2-3 (rarely a few more) florets; bracts few, decussate, with strong, green or maroon, lance-ovate dorsal areas bisected by a strong midnerve; sterile bracts mostly 4, the lower pair triangular, keeled, slightly shorter and narrower than the inner pair, the fertile pairs again slightly longer or equal to inner sterile pair, oblong, ca. 4.5-5 mm long, less keeled, more often convex-rounded or rounded-navicular, api- cally narrowly to broadly rounded, entire, scarious-bordered, the narrow tips often vil- losulous-ciliate. Lateral sepals pale red-brown, thin, subequilateral, oblanceolate, 4-5 mm long, acute, the narrow, firm keel papillate or ciliolate above middle, or the keel smooth. 636 Annals of the Missouri Botanical Garden FIGURE 47. Xyris ERE ii a E. —a. Habit sketch. —b. Leaf apex. —c. Leaf blade-sheath Poi —d. Leaf base.—e. Spike.—f. Lateral sepal.—g. Petal, stamen.—h. Staminode.—i. Stylar a Capsule spread to show ss > of a ‘three ie and placentation.—k. Se ed. Volume 75, Number 2 1988 Kral 637 Xyris Petal blades broadly obovate, ca. 3 mm long, yellow, the broadly rounded apex lacerate- dentate. Staminodia bibrachiate, the broad, thin branches penicillate-ciliate. Anthers ob- long, ca. 1 mm long, deeply bifid and sagit- tate, on filaments ca. 0.5 mm long. Capsule ellipsoid, ca. 2.5 mm long, the placentation basal-central, the valves with narrow septa below the middle. Seeds numerous, ovoid to ellipsoid, ca. 0.5 mm long, apiculate, dark amber, lustrous, finely ribbed longitudinally. Distribution. Locally abundant in low- to medium-altitude savanna, southeastern Co- lombia eastward into Surinam and in adjacent Brazil south into Goiás. Selected | specim COLOMBIA. AMAZONAS- VAUPÉS: Río A ap ca. 900 ft., 1 (GH). VAUPÉS: Cerro Yapoboda, u 1951, Schultes & Cabrera 14379-B (GH); i Cerro Kanen mi. up from mouth, , 10 Nov. 1952, Schultes lí Cabrera 18348 io Vaupés, ca. 800 ft., 20 Apr. 1953, & rar s.n. (GH); same dus 20 Apr. s & Cabrera 19200 (U). GUYANA: Saes- 197 73. Cooper 208 ied epe Plateau, 3 y 1944, Fanshawe (K, U). S savanna inter Zanderij I et Hannover, opn. 128. 8 “Oct 1958, J. & W. A. E. van Donselaar 363 (U); prope km 103, opn. 308, Gros-savanna, 30 Apr. 1959, J. van Donselaar 694 (U); Boven Coesewijne, savanne in lijn 5-6 van exploratie 860, 16 May 1956, Heyligers 14 (U); prope jodensa- vanne (Fluv. Suriname) tr. 30, p. 74, 30 Mar. cn Ede pe 802 (U); savanne tussen Paranam en Phed 9 Feb. 1961, Kramer & pee ge p Zander L 4 Sep. 1948, Lanjouw ph an ad Gro examine oris, Raudal d > Zion ridge E of w & Lindeman 928 upper Commewijne River, 14 July 1953, Lindeman 4232 (U); Iter secundum Surinamense, pr. Zanderij I, July- Sep. 1920, Pulle 58 (U); Zanderij, 28 June 1970, Teu nissen LBB 12761 (U); “Suriname” C. bis o hon Schweinitz Herb. (PH); Nat. Res. “Brinckheuvel, " 1967, Wilschut & Teunissen 11588 (U). a T. F. AMAZONAS: alrededores de Yavita (Rio Temi), 6-19 July 1969, Bunting et al. 3714 (MY, NY, U, VDB); Santa Cruz margen del Rio Atabapo, 4 Sep. 1960, awal 3689 (NY, VEN); del raudal **Moriche" en el Rio yapo, 10 May 1983, Guánchez 3100 (TFAV, VDB). campina near abandoned Petrobras airstrip, mun. Borba, 7 July 1983, Hill et al. 12975 (MO, NY, VDB); 5 km al N de la punta E del Cerro Yapacana, 28 June 1979, Huber 3903 (US); 2 km al W de San Antonio del Orinoco, 20 July 1980, Huber & Tillett 5423a (VEN); a unos 15 km al N del Cerro Yapacana, 27 July 1980, Huber & Tillett 5557 (MYF, VDB, VEN); granite outcrop 30 km below La Urgana, 100 m, 14-15 Mar. 1949, Maguire 29084 (NY, US); Cerro Duida, Río Cunucunuma, along Cario Culebra, 1,000-1,100 m, 18 Nov. 1950, Maguire et al. 29520 (NY, US); Santa Cruz, small village on Rio Atabapo, 17-18 Nov. 1979, Thomas & Rogers 2688 (NY); Cerro Duida immediatamente N de La Esmeralda, ca. 1,350 m, 29 Jan.-11 Feb. 1975, Tillett et al. 751- 93 (K, MYF, NY, US, VEN); Yavita, 128 m, 26 Jan. 1942, L. Williams 13996 (F); near Yavita, 10 June 1959, Wurdack & Adderley 42921 (GH, US, VEN). BOLIVAR: Salto Cama, ca. 1,000 m, 4 Dec. 1973, Davidse et al. 4864 (MO); 15 km al NW de Uiaren, Huber et al. 7602 (MYF, VDB); Macizo del Chimanta, 2,000 m 26-29 Jan. 1983, Huber & Steyermark 6874 (VEN): 15-20 km S del empalme Luepa-Kavanayen, Huber et al. 7252 (MYF); sector SSE altiplanicie in tal del 2 tepui, 13-16 Feb. 1984, Huber et al. 9012 (NY, VEN); Macizo del Guaiquinima, 1,350 m, 2 Apr. iu Huber 9365 (MYF, VDB); La Escalera, ca. 7 km N of Pioneer Monument, ca. 1,200 m, 24 July 1983, Kral 70316 (BM, F, K, L, MO, NY, SP, U, US, VDB, VEN, & others); above El Salto Yuruani, Rio Yuruani, Kral 70571 (MO, NY, US, VDB, VEN); 8 km N of San Rafael, Gran Sabana, 29 July 1983, Kral 70575 (MO, NY, US, VDB, VEN); ca. 2 km N of Luepa, 20 Dec. 1984, Kral 72206 (MYF, VDB, VEN) 17 km E of El dec 30 Oct. 1985, Liesner 19215 (MO, VDB, urs ro Guaiquinima, 1,600-1,700 m, 4 Jan. 1952, pers 32983 (GH, K, NY, US); Salto Acarima, Rio Uri. man, 9 Jan. 1955, ip de ag & í 44 (NY, US); Roraima, Glycon Swamp & vic., 1,830-1,920 m, 25 Sep. 1944, S E. 28630 T T Canes de uyan op sector oriental, al no e la Mision de Camarata, 2,140 m, 28 Feb 1978, poer et al. 116133 (MO, VDB, ' VEN). Examination of the type of X. filiscapa Malme reveals that this differs in no signifi- cant way; that is, a broad range of specimens of X. guianensis shows sepal keels varying from entire to ciliolate and leaf blade margins ranging from strongly to weakly incrassate bordered. Elements of the former described as having ciliate sheaths are either ciliate- keeled X. guianensis or ciliate-leaf-sheathed X. tenella Kunth (X. steyermarkii Maguire & Lyman B. Smith). Therefore Colombian material identified as X. filiscapa turns out to be X. guianensis. 48. Xyris spathacea Lanj., Pulle, Rec. av. Bot. Neerl. 34: 484, fig. 4. 1937 TYPE: Surinam: “Sanderij I, Sept. 1914, Leg. Essed" mS L; isotype, U). Figure 48A Xyris exserta mir & Lyman B. Smith, Caldasia 6: 229, f. 21. 1954. TYPE: Colombia. Vaupés: sands 638 Annals Mou DL NS Garden in rapids, Río Guaimia, Cario del Caribe near San José, 850-900 ft., 2 Nov. 1953, R. E. Schultes, E. D. Baker & Is. Cabrera 18276 (holotype, OL; isotypes, GH, Xyris ms Idrobo & Lym man B. Smith, la 6: 231-232, fig. 22. 1954. TYPE: Colombia. Vau , Ca. Í Dec. 1943, P. H. Allen 3287 (holotype, MO). Mostly low, slender, soft-based, solitary or cespitose, sometimes short-rhizomed ephem- erals (5-)10-30 cm high. Leaves erect or spreading, dimorphic, those of dry stages or of innovations 0.5-5 cm long, with sheaths over % as long to longer than blades, the broad, scarious margins abruptly converging distally to the short, terete, 0.2-2.5 mm thick, conic-tipped, stiff, often maroon blades, also often producing apically a scarious, triangular or oblong ligule 2 mm long; **wet" or sub- mersed-stage leaves lax, flaccid, filiform mostly 1-2 dm long, the thin sheaths much less than 1⁄2 as long as blades, pale brown or stramineous, sparsely costate, tapering grad- ually from base to blade, there with a scarious, narrowly triangular ligule to 3 mm long, the blades terete or at intervals somewhat flat- tened, ca. mm thick. Scape sheaths loosely tubular, much raised above the “‘dry’’- season foliage, shorter than the lax “wet”- phase leaves, twisted and fluted, multicostate, distally open, carinate, producing a short, cusplike blade. Scapes soft, filiform, straight or slightly flexuous, slightly twisted, terete, striate, 0.3-1 mm thick, ecostate. Spikes el- liptic to obovoid, 3-5 mm long, sometimes proliferous, acute, 2—4-flowered, of a few spi- rally arranged, scarious-bordered, loosely im- bricate bracts with distinct, elliptic-ovate, brown to green, papillose dorsal areas over 1⁄4 as long as the bract body; sterile bracts 3-4, narrowly ovate, slightly shorter than the fertile bracts and grading into them, the two lowest keeled, navicular, subacute; fertile bracts broadly elliptic to obovate, ca. 3 mm long, subacute, the margins sometimes white- villosulous-ciliate at apex, the backs rounded- convex, ecarinate, the dorsal areas bisected by a narrow but distinct midnerve. Lateral sepals free, subequilateral or somewhat in- equilateral, lanceolate to linear-oblanceolate, ca. 3-3.5 mm long, acute, the broad, low keel sparsely papillose or short-ciliate from middle to apex. Petal blades obovate, 3-3.5 mm long, yellow, the broadly rounded apex lacerate-dentate. Staminodia bibrachiate, the broad, flat branches copiously penicillate api- cally. Anthers oblong, ca. 1 mm long, bifid to below middle, sagittate, on filaments ca. 0.6-0.7 mm long. Capsule short-cylindric to narrowly obovate, ca. 2.5 mm long, the pla- centation basal, the valves without septa. Seeds numerous on long funicles, ellipsoid, 0.4—0.5 mm long, pale brown, translucent, with an irregular, coarse, partly anastomosing reti- naculum of a few strong, deep brown ribs. Distribution. Locally abundant in low, riverine and intermittently inundated savan- nas, southeastern Colombia, southern Vene- zuela, eastward to Surinam and contiguous Amazonian Brazil (Amazonas, Mato Grosso, Para). Additional specimens examined. d AMAZONAS: outh, 8 prolifera K. & S." do Roncador, Mun. o Garcas, G. T. Eiten 8572 (MO, VDB). een VAUPES: types of X. exserta and X. yapobodensis constituting all known records thus far. SURINAM: Zanderij, d.d. Aug. 1958, van Donselaar et al., s.n se) s via secta ab oengo tapoe ad Grote elzwamp nea , 29 Sep. 1948, Lanjouw & Lindeman 581 (U); Tibiti savanne near km 5.8, 15 Jan. 1949, Lanjouw & Lindeman 1854a (U); E of Kopie Peninica R, distr. Commewijne, 16 July 1953, Lindeman 4375 (U); Zanderij in savanna pool half . Lindeman 4483 (C); Zanderij, al- most dry along rd. to old radio station, 22 Nov 1953, Dun 5068 (NY, U, iu oe savanna on Braz. frontier, 305 m, 4 km S “4. mts., Oldenburger et al. 188 (U); Lander pool edge, 25 Jan. 1942, G. Stahel s.n. (GH, U). VENEZUELA. T. F. AMAZONAS: “Fundo Galletti”” Reserva forestal del Rio Si- papo, 2 Feb. 1983, Guánchez 2416 (TFAV, VDB); el medio Cano Yagua y al N del Cerro Cucuritu, 120 m, 18 Jan. 1979, Huber 3122 (US); a = orilla derecha (W) del Alto Cano Yagua, 18 Feb. Huber 3185 (NY); borde del alto Caio Yagua, 28 Feb. 1980, Huber 4818 (US, VDB, VEN); Savanna II, fls. open in late morning, base Cerro m, 10 Aug. 1983, Kral & Huber 70703 (VDB, VEN, d to be distributed). Volume 75, Number 2 1988 Kral 639 Xyris FIGURE 48A. Xyris spathacea (from the type).—a. Habit sketch.—b. Leaf apex.—c. Leaf DT. ike. ode. junction. —d. Leaf.—e. Leaf base. —f. Stylar apex.—k. Capsule, median.—l. Seed. f —g. Lateral sepal.—h. Petal blade, stamen.—i. Stamin 640 Annals of the Missouri Botanical Garden FIGURE 48B. Xyris spathacea (Huber 3185).— a. Habit sketch.—b. Leaf sheath—blade junction.—c. Spike, showing exserted sepals, proliferous habit. —d. Spike, flower, stylar apex.—e. Fertile bract. —f. Lateral sepal.— Volume 75, Number 2 1988 Kral 641 Xyris This species has a low stature and very short leaf blades when developing on drying sites left by retreating waters. On the other hand, such plants produce much longer, laxer leaves when water rises to submerse them or their bases. I find no way to distinguish ma- terial of X. spathacea (Fig. 48A) from what is being called X. exserta (Fig. 48B) from the upper Amazon Basin of western Brazil and from the low savanna of Venezuela and Co- lombia. The only difference that is of any real interest is the tendency toward proliferative spikes in the latter, doubtlessly this trait in- duced environmentally. Some such Venezue- lan and Brazilian extremes produce strong branches (scapes) that radiate umbel-like from the primary spike and terminate in floriferous spikes (Huber 3185, Prance et al. 16183). The type of X. exserta differs not at all from longer-leaved X. spathacea from Surinam. 49. Xyris cyperoides Gleason, Bull. Tor- rey Bot. Club 56: 17. 1929. TYPE: Guy- ana: Kaieteur Savanna, Potaro River, Sep.-Oct. 1881, G. S. Jenman 1056 (holotype, K). Figure 49. Xyris T Kral & Smith, ie bison idi 436-437, fig. 6. 1983. arpm TYPE: southea nt ea — Pian. 400 m, Cordillera Epicara. Rio Chic n, 5 Sep. 1 . Maguire et al. 53651 dolor VEN; isotypes, NY, US, VDB). Densely cespitose, the firm, pale reddish brown bases covered by persistent old leaf bases, the stems short to somewhat elongated, erect or ascending; roots slender. Principal leaves subdistichous, erect, twisted, flexuous, 0.8-1.8 dm long, longer than the scape sheaths (and often the scapes themselves); blades 5-6 times longer than sheaths, filiform, angulate to terete or somewhat compressed, 0.2-0.4 mm wide, longitudinally strongly nerved, shining, smooth with nerves reddish brown, wider than the greenish intervals; tips gradually narrowed, fimbriolate with clavate trichomes at apex; margins entire or miautely scabrid toward base; sheaths ecarinate, entire, strongly multicostate, smooth, gradually di- lating to base, producing an acute, scarious ligule to 1.5 mm long at apex. Scape sheaths similar to leaves but shorter. Scapes filiform, 1.5-2 dm long, 0.3-0.4 mm wide, slightly twisted and flexuous, terete, finely 1—several- costate, olivaceous, smooth. Spikes 2-flow- ered, ellipsoid or in mature state obconic, ca. 4 mm long; bracts loosely imbricate, decus- sate, triangular-ovate, mostly 6, smooth, pale lustrous reddish brown, sparsely ciliate toward apex, marginally scarious; sterile bracts 4, unicostate, the lowest pair lanceolate, ca. 2.5 mm long, cymbiform, the inner pair trian- gular-ovate, ca. 3 mm long, ecarinate; fertile bracts ovate, ca. 3 mm long, strongly convex or navicular, at length excurvate; dorsal area olive, then ferrugineous, conspicuous, large, nearly as long as bract. Lateral sepals oblong, ca. 3.5 mm long, slightly curvate, subequi- lateral, obtuse and slightly emarginate to acute; keel narrow, entire. Petal blades ob- triangular, ca. 3 mm long, yellow, subtruncate at apex, erose. Staminodia bibrachiate, the branches sparsely penicillate apically. An- thers oblong, 1.2 mm long, deeply bifid and sagittate, on filaments ca. 1 mm long. Capsule ellipsoid, ca. 2 mm long, the valves without septa, the placentation central. Seeds nu- merous, broadly ellipsoid, ca. 0.5 mm long, amber, finely longitudinally multiribbed. Distribution. Sandy, medium- to high- elevation savanna, apparently rare, from southern Estado Bolivar, Venezuela south- ward into contiguous Amazonas, Brazil and eastward to Surinam. Additional specimens examined. BRAZIL. AMAZONAS: anaus- p a ghway, forest at km 130, Steward et al. P20348 (specimen bearing nomen nudum cata (US). Guy YANA: NE Pr along Mire -mure Kappel savanna, Tafelberg, 300 m, Natte struik- "E in het Z. deel, Feb. 1961, Kramer & Hekking 3308 (U). VENEZUELA. BOLÍVAR: Cerro Pitón, 400 m, cordillera Epicara, Rio poe 3 Sep. 1 962. ermark 53563 (NY, us same locality, 9-1 aguire et al. 53713 ; de quinima, Salto del Rio San da 1-2 km rio arriba -25 Jan. 1977, Stey del Salto Szczerbanari, 750 m, 2 ermark et al. 113116-A (US, VDB, Auyan-tepui, sector oriental, al norte de la Misión de 642 Annals of the Missouri Botanical Garden FIGURE 49. Xyris cyperoides (from holotype of X. epicarae).—a. Habit sketch. —b. Leaf apex.—c. Leaf at midblade.—d. Lea eaf at t blade—sheath junction. —e. Leaf base.— f. Spike at anthesis.—g. Spike past anthesis. — h. Lateral sepal.—i. Petal and stamen.—j. Staminode.—k. Stylar apex.—l. Capsule, one valve removed. —m. Seed. Volume 75, Number 2 1988 Kral Xyris Camarata, 2,140 m, 28 Feb. 1978, Steyermark et al. 116133 (MO, VDB, VEN). In the spike this resembles X. guianensis Steud., but it is definitely longer-leaved in relation to scape, and the lateral sepal keels are entire. 50. Xyris toronoana Kral, sp. nov. TYPE: Venezuela. Bolivar: Distr. Piar, Macizo del Chimantá, sector centro-meridional. Amplio valle ubicado entre el borde nor- oriental del Torono-tepui y la sección central del Chimantá-tepui, drenando ha- cia el Sur, ca. 6?16'N, 62%09'W, + 2,100 m, 11-15 Feb. 1985, O. Huber, Teuvo Ahti & J. J. Pipoly 10,223 (ho- lotype, VEN; isotypes, K, MYF, NY, US, VDB). Figure 50 Herba humilis, perennis, glabris, caespitosa, rhizomate crassa, brevi, subverticali. Radices fibrosa. Folia linearis, DE! cm longa, erecta d d expansa, vaginis caporum parum longiora. La e foliorum principalium pe: He planae vel lios tortae, 0.9- 1.2 mm latae, ini itudine paucinervosae, ad margine rassati, gines leviter incrassati, minute ciliati vel scabriduli; va- ginae carinatae, integrae, Ne cag stramineae, infime abrupte dilatatae, marginibus in lam radatim con- tractis. Vagi scaporum Suse. laxae, tortae, laminis brevibus. Scapi recti vel aliquantum flexuosi, torti, 2-1.5 dm alti, olivacei, distaliter leviter m ca. ] mm lati, acute bicostati, costis iai is. Spicae anguste obovoideae vel ellipsoideae, ong e, ferrugineae, ciflorae, bracteis laxe ede deals minute pa- i valde laceratis, m longae in 5 mm longo, ns ° Carinato, areis dorsalibus anguste li nearibus, paro ovato, solum ad apicem carinato, areis dors Jus previlinearibus. Bracteae fertiles oblongae, ca. -7 mm ae, naviculares, distaliter carinatae, areis dorsalibus aa bak ca. 3 mm longis. Sepala late- ralia libera, inaequilatera, anguste hiec di 5.5-6.5 mm longa, acuta, leviter curvata; ala carinali integra. Laminae petalorum angustae obovatae, ca. 5 mm longae, luteolae, anguste rotundatae. Staminodia bibrachiata, he chiis longipenicillatis. Anthera oblongae, sagittatae, 1.5 mm longae; filamenta ca. 0.7-1 mm longa. Capsula an- guste ellipsoidea, 3 mm longa; placenta basalis. Semina merosa, anguste ovoidea, acuminata, 0.6-0.7 mm longa, translucida, ferruginea, longitudine subtiliter multicostata. Low, densely cespitose, smooth herb, the rhizome thick, short, subvertical; roots fi- brous. Leaves linear, (2-)4- 7 mm long, erect to slightly spreading, a little longer than the scape sheaths. Principal leaf blades flattened, sometimes slightly twisted, 0.9-1.2 mm wide, about twice as long as the sheaths, few-nerved, dark red-brown; tips gradually narrowed, in- curved-acute, slightly thickened, the margins thickened, papillose; blade margins slightly thickened, minutely ciliate to scabridulous; sheaths carinate, entire, eligulate, stramin- eous, abruptly dilated below, the margins gradually narrowed into the blades. Scape sheaths multicostate, loose, twisted, short- bladed. Scapes straight or somewhat flexuous, twisted, 1.2-1.5 dm high, olivaceous, slightly compressed distally, ca. 1 mm wide, acutely bicostate, the costae scabrid. Spikes narrowly obovoid to ellipsoid, 5-7 mm long, ferrugi- neous, few-flowered, the bracts loosely im- bricate, minutely papillose distally, the matrix thin, strongly lacerate at margin, ferrugineous or roseolate. Sterile bracts 2-4, subdecussate, ca. 5 mm long, the lower pair narrowly ovate, ca. 5 mm long, strongly carinate, with nar- rowly linear dorsal areas, the upper pair ovate, carinate only at apex, the dorsal areas short- linear. Fertile bracts oblong, ca. 4, 6-7 mm long, navicular, distally carinate, the dorsal areas linear, ca. 3 mm long. Lateral sepals free, inequilateral, narrowly lanceolate, 5.5- 6.5 mm long, acute, slightly curvate; carinal keel entire. Petal blades narrowly obovate, ca. 5 mm long, yellow, narrowly rounded. Staminodia bibrachiate, the branches long- penicillate. Anthers oblong, sagittate, 1.5 mm long; filaments 0.7-1 mm long. Capsule nar- rowly ellipsoid, 3 mm long; placenta basal. Seeds numerous, narrowly ovoid, acuminate, 0.6-0.7 mm long, translucent, red-brown, finely ribbed longitudinally. The low habit and the delicate, lacerate, thin bracts with narrow, dark dorsal areas would appear to put this species into the X. tenella complex. However, the sheaths are entire and long-ciliate, and the lateral sepals are entire and more inequilateral. The tufts of erect or ascending, thick rhizomes covered with leaf bases are distinctive. It is known from only the type locality. 644 Annals ME EL ER Garden Əl. Xyris aquatica Idrobo € Lyman B. Smith, Caldasia 6: 206, fig. 9. 1954. TYPE: Colombia. Amazonas: on rocks in swift brook, Rio Caquetá, vic. La drara, Apr. 1924, R. E. Schultes 5855 (holotype, COL; isotypes, GH, US). Fig- ure 51 Slender, lax, glabrous, soft-based, cespitose and profusely slenderly, scaly-rhizomatous perennial, 3-5 high but usually bent and trailing in rapid shoalwater. Leaves very soft, variously elongate, to 3 dm long, polystichous, the sheath less than 4 as long as the blade, the margins entire, the base lustrous pale brown or greenish brown, ecarinate, dilated, narrowing gradually upward to a prominent, narrow, scarious ligule to 6 mm long, there abruptly contracted to a terete, filiform, fluted blade up to ca. 0.5 mm thick. Scape sheath inflated-tubular, tan, opening distally to a bifid or acute apex, bladeless. Scapes straight or slightly flexuous, terete, 1-1.5 mm thick, ecostate. Spikes ovoid, narrowly ellipsoid or cylindric, 1-1.5 cm long, blunt, of many spi- rally imbricate, brown bracts, the sterile bracts several, ovate, slightly smaller than and grad- ing into the fertile bracts, these oblong to obovate, ca. 5 mm long, broadly rounded, entire, backs rounded-folded, ecarinate, with ovate-elliptic, greenish dorsal areas. Lateral sepals free, curvate, subequilateral, linear- oblanceolate, 4-5 mm long, acute, lustrous brown with a narrow, dark, firm keel, this scabrociliate from below middle to apex, or subentire. Petal blades obovate, ca. 5 mm long, the broadly rounded apex erose. Stam- inodia bibrachiate, the branches lineal, dense- ly penicillate-pubescent from base to tip. An- thers oblong, ca. 1.5 mm long, the upper Y bifid; filaments ca. 2 mm long. Capsule ob- ovoid, ca. 2 mm long or longer; placentation basal-central; valves eseptate. Seeds numer- ous, ovoid, ca. 0.5 mm long, apiculate, red- dish brown, translucent, finely irregularly longitudinally striate or subreticulate by anas- tomosing ribs. Distribution. A true aquatic, usually in shallow rapid shoalwaters of streams in the Amazon Basin of northern Brazil, southeast- ern Colombia, and T. F. Amazonas in Ven- ezuela. Mg ads specimens examined. BRAZIL. AMAZONAS: n. Humaitá, estrada Humaitá —km 150, a 65 km ao Su "m rio de nome Branco, L. O. A. Teixeira et al. 1370 (INPA, NY, VDB); Rio Putus. Rio Ituxi, Rio Curuquetë, Sao Paulo, 30 km above mouth of Rio Coti, 20 Jul 1971, G. T. Prance et al. 14460 (INPA, NY, VDB); Rio Mar- mellos, Aug. 1948, Schultes & López 10307 (US). COLOMBIA. VAUPÉS: Río Piraparaná (trib. of Río Apaporis), Cario Paca, 19 Sep. 1952, Schultes & Cabrera 17565 (GH, U); Rio Parana Pichuna, ca. 700 ft., June 1953, Schultes & Cabrera 19908 (NY). VENEZUELA. APURE: 9 km N of Cario Cochina de la Pica, along main e pim Rio Cinaruco and Rio Capanaparo, m, . 1979, O). T. F. AMAZONAS: entre B js Sipapo y E Venado, 25 Nov. 1977, A. Fernandez 7 (F, ajas con cano de morichal de los bans de E Sipapo, ca. 100 m, lugares abiertos en carretera hacia el Sipapo, 22 Mar. 1979, Trujillo & Pulido 15089 (MY); islas del Rio Cataniapo en el Raudal Rabipelado a unos 35 km al sur-este de Puerto Ayacucho, 6 Mar. 1981, Guánchez 904 (MYF, TFAV, VDB, VEN). One of the few truly aquatic Xyris, the whole plant is apparently often totally sub- mersed. Nearest it morphologically are X. apureana Kral & Lyman B. Smith and X. spathacea Lanjouw, but these usually have smaller spikes, often flatter leaf blades, and larger seeds. 22. Xyris apureana Kral & Lyman B. Smith, Ann. Missouri Bot. Gard. 69: 412-414, fig. la-i. 1982. TYPE: Ven- ezuela. Apure: Dist. Pedro Camejo, ca. , 2 km S of Cano la Cochina de La Pica along main road south of Paso de San Pablo to the Rio Cinaruco, 70 m, 2 Mar. 1979, G. Davidse & A. C. Gonzalez 15948 (holotype, US; isotypes, MO, VDB, VEN). Figure 52. Perennial, lax, densely cespitose, smooth herb. Rhizomes slender, short to elongate, ascending (relating to degree of depth in sub- strate). Roots slender. Leaves linear, 1.5-3 dm long, erect or slightly spreading, subdis- tichous, longer than the scape sheaths; blades 5-10 times longer than sheaths, 1-2 mm wide, flat, straight, longitudinally few-many- nerved and sulcate, flattened from base to middle, terete or subterete toward apex, the Volume 75, Number 2 1988 Kral 645 Xyris IMN 5mm FIGURE 50. Xyris toronoana (from the type).—a. Habit sketch.—b. Leaf tip.—c. Sector of midblade and sheath junction.—d. Leaf base.—e. Spike.—f. Fertile bract (somewhat flattened) .—g. Lateral sepal. —h. Petal blade and stamen.—i. Staminodium and enlarged beard hair. —]j. Stylar apex.—k. Capsule.—l. Seed. apices narrowed, narrowly rounded, thick- ened, often channelled; margins entire, not thickened; sheaths carinate, pale shining brown, several-nerved, scarious except for the ribs, the margins entire, gradually converging to blades, there producing a short, broad, scarious ligule. Scape sheaths lax, twisted, multicostate, opening toward apex, carinate, short-bladed. Scapes slender, terete, straight or somewhat flexuous, 2.5-3.5 dm high, 0.5- 646 Annals of th Missouri Botanical Garden emm — IGU 1. Xyris aquatica (from Fernandez p except pid noted) . Sm & Cabrera 17565, Fernandez 2887).— b. Leaf.—c. Spike.—d. Fertile bract.—e. Lateral sepal. Petal, stamen, staminode, style branches.—g. Capsule. —h. Se RA —a. Habit sketch (composite from —f Volume 75, Number 2 Kral 647 1988 Xyris | KY FIGURE 52. Xyris apureana (from the type).—a. Habit sketch. —b. Leaf apex.—c. Leaf blade sector, mid- blade.—d. Leaf sheath—blade junction.—e. Leaf base.—f. Spike.—g. Lateral sepal.—h. Petal blade, stamen, ed. staminode, stylar apex.—i. Se 648 Annals of th Missouri Eod Garden 0.6 mm thick, olive to reddish. Spikes several- flowered, elliptic, 5-6 mm long, acute, the bracts subdecussate, navicular or convex, ecarinate but medially 1-nerved, scarious, ferrugineous, erose; sterile bracts 4, the low- est pair at least 4 as long as the spike, oblong, the inner pair ovate, ca. 3 mm long; fertile bracts ovate, to 4 mm long, with reddish- scarious, entire to somewhat lacerate borders; dorsal area lanceolate, reddish brown or brown, 1⁄2 as long as or equal to the bract. Lateral sepals lanceolate, strongly inequilat- eral, ca. 4 mm long, acute; carinal keel nar- row but strong, entire. Petal blades broadly obovate, 2.5- m long, yellow, broadly rounded, strongly erose, cuneate. Staminodia bibrachiate, the branches long-penicillate. Anthers oblong, ca. 1.5 mm long, retuse and sagittate, slightly longer than the slender fil- aments. Capsule oblong, planoconvex, 2.5 mm long, amber, longitudinally finely multiribbed. Distribution. Known thus far only from the type locality in Apure, Venezuela. This subaquatic is superficially nearest X. aquatica on the one hand, and X. spathacea on the other. From both it differs in its more slender rhizomes, flattened leaves, and longer seeds. 53. Xyris stenostachya Steyerm., Field- iana, Bot. 28(1): fig. 16K, L. 1951. TYPE: Venezuela. T. F. Amazonas: among rock outcrops, 100 m, Sanariapo, Rio Ori- noco, J. Steyermark 56437 (holotype, F; isotypes, NY, US). Figure 53. Solitary or cespitose, soft-based annual mostly 1-4 dm high, the stems contracted. Leaves spreading flabellately, 0.3-1.5 dm long; sheaths Y2 as long as blades or shorter, tan or roseate, papillose or rugulose, narrow- ing evenly to blades, there producing a short, erect, scarious ligule, or ligule absent; blades linear-ensiform, flat, 1-3 mm wide, tapering above middle, then abruptly incurved-acute, the tip callused, the margins proximally tu- berculate-scabrid or papillose, distally mostly smooth, the edges thickened-rounded, the surfaces maroon to yellow-green, papillose- rugose at least proximally and finely nerved. Scape sheaths as long as or shorter than leaves, tubular, twisted, costate, with short, erect blades similar to leaves. Scapes straight or flexuous, twisted, ca. 0.5 mm wide, subterete or angulate distally, striate and/or with l- few costae and strongly scaberulous-rugose. Spikes lineal or lance-lineal, flattened, 1.5-5 cm long, the numerous distichous bracts na- vicular with strong dorsal areas medially uni- costate and with narrowly rounded backs; sterile bracts ca. 4, the lowermost sometimes with long-excurrent, green dorsal areas, more often smaller than the fertile bracts and grad- ing into them; fertile bracts lance-ovate to broadly oblong, 5-6 mm long, conduplicately rounded-folded around floret, apically broadly rounded, the margins scarious, a broad, pale entire to (in age) lacerate border around the large, lance-elliptic dorsal areas. Lateral se- pals free, subequilateral, elliptic-linear or lance-linear, ca. 4 mm long, apically subulate or narrowly acute, tan, the narrow, firm, brown keel ciliolate to entire. Corollas gamopetalous, the limb ca. m long, the lobes lance- oblong, ca. 4. 5 m mm long, yellow, the apex acute, the margin remotely lacerate-dentate. Staminodes subsessile in sinuses of corolla, bibrachiate, the fleshy narrow branches with numerous penicillate, clavate hairs. Anthers lance-oblong, 0.5 mm long, on fleshy fila- ments ca. 1 mm long. Capsules narrowly el- lipsoid, ca. 3 mm long; placentation basal. Seeds 3-6, some on funicles longer than themselves, fusiform, 1-1.5 mm long, amber, finely longitudinally striate, often with an ir- regular retinaculum of wider, reddish brown ridges. Distribution. Wet sandy savanna, pools and seeps on and around granite outcrops, southeastern Colombia (not on basis of spec- imens but on basis of similar geology imme- diately across the Rio Orinoco from an abun- dance of Venezuelan localities) eastward into the Orinoco of Territorio Federal Amazonas and Estado Bolivar, Venezuela. Selected specimens examined. VENEZUELA. T. F. AMAZONAS: 12.5 km S of Puerto Ayacucho, 50 m, 2 Nov. 1971, Davidse 28294A (MO); la margen derecha del Rio Volume 75, Number 2 Kral 649 1988 Xyris ES pu s FIGURE 53. Xyris stenostachya (Kral sh Aoi sel, .—a. Habit sketch.—b. Leaf tip.—c. Leaf at junction of blade and sheath.—d. Leaf base.—e. Spi - Sector of upper scape. —£8- Lateral sepal. of corolla limb.—i. Op ened corolla showing ea Kah and staminodes.—j. Stylar apex.—k. Capsule, open to ow one valve and ices A —l. Seed. —h. Oblique view 650 Annals of the Missouri Botanical Garden Guayapo, en el “Salto Moriche,” 9 Oct. 1983, Guánchez & Varadarajan 2573 (TFAV); 2 km al sur de Puerto Ayacucho, del Rio Autana en raudal ““seguera,” 8 Nov. 1984, Guánchez & Melgueiro 3324 (TFAV, VDB); aero- puerto de Puerto Ayacucho, 75 m, 13 Sep. 1977, Huber 1046 (MYF, NY, VDB, VEN); Cerro Yapacana, ca. 100 m, 14-28 Feb. 1978, Huber 1546 (US); 20 km SE de bs i Galipero, 3 Dec. 1979, Huber. 4759 (NY, US); na apacana savannas, 9 Aug. 1983, Kral & er 70681 (BM, F, K, L, MO, NY, SP, TFAV, U, US, VDB, VEN); fisheries lab, granite outcrops, Puerto Ayacucho, 11 Aug. 1983, Kral 70727 (BM, F, K, L, MO, NY, SP, TFAV, U, US, VDB, VEN, and others); Cano Sambolje near La Urbana, Maguire 28989 (NY, US); Yapacana Savan vy I, at Jan. 1951, Maguire et al. 30787 (C, NY); 35 km SE of Puerto Ayacucho, Steyermark et al. 122491 (MO, NY). Bo- LÍVAR: Cerro San Borja, occasional, Rio Orinoco at 100- 300 m, 12 Dec. 1955, Wurdack & Monachino 39845; Rio Villacoa, 1-4 km above Salto de Humito, 1956, Wurdack & Monachino 41164 (F, NY, ca This distinctive little annual is most abun- dant in the shallow pools or seeps in and around granitic lajas within the Venezuelan *piedmont" above the Great Bend of the Ori- noco. No other species remotely resembles it. The pale yellow flowers open in the morning. 54. Xyris stenocephala Malme, Bih. Kongl. Svenska Vetensk.-Akad. Handl. 22, Afd. 3(2): 18, tab. 1, fig. 1. 1896; Smith & Downs, Fl. Bras. 9(II): 105- 106, tab. 34, figs. 1020-1027. 1968. TYPE: Brazil. Mato Grosso: lugar aberto, pantanoso, cerca de Santa Ana da Cha- pada, 700-800 m, 28 Nov. 1894, Malme 1426 (holotype, S; isotype, MO; phototype, F). Figure 54 Perennial or annual, solitary or cespitose, glabrous plants (1-)2-5 dm high, the stem short. Leaves erect to spreading flabellately, 5-20 cm long; sheaths at least 1⁄2 as long as blades, entire, strongly nerved, carinate, pale brown to deep red-brown, sometimes papil- lose, often with a strong, deep brown, lustrous costa, tapering gradually from a slightly di- lated base to the blade, there eligulate or with a short, scarious ligule; blades narrowly linear, flattened, twisted, 1-4 mm wide, strongly nerved, tapering gradually to an incurved- acute, cartilaginous-bordered tip, the margins thickened, lustrous, pale to deep brown, rare- ly scabrociliate, the sides pale green to reddish green or ferrugineous. Scape sheaths shorter than principal leaves, tubular, ferrugineous, multicostate below, apically with short, flat blades. Scapes flexuous and twisted, distally terete, 1-1.5 mm thick, rarely unicostate with scabrid costa, mostly just striate, oliva- ceous to green-brown. Spikes mostly ellipsoid, 1-2 cm long, acute, smooth, of several tightly spirally imbricate bracts; sterile bracts usually 4, the lowest pair oblong, carinate, slightly shorter than the fertile bracts, these broadly ovate or obovate to suborbicular, 4.5-5.5 mm long, strongly convex and ecarinate, broadly to narrowly rounded apically, the matrix pale to deep lustrous brown with a pale, thin, entire to erose border, the dorsal areas pale to deep green or brown, ovate, large but subapical, with a strong but fine midrib. Lateral sepals free, strongly inequilateral, strongly curved, acute, ca. 4 mm long, the keel firm and broad, ciliolate from near base to tip; petal blades ca. 4 mm long, obovate, broadly rounded, lacerate, yellow. Staminodia bibrachiate, the branches densely penicillate. Anthers oblong, ca. 1.5 mm long, on filaments 0.5-0.7 mm long. Capsule planoconvex, obovoid, 2-2.5 mm long, the placenta basal. Seeds few, fu- siform-curvate, 1.2-1.7 mm long, dark red- brown, translucent, strongly beaked, longi- tudinally multicostate. Distribution. savanna, mostly in northern Brazil, in Ama- zonas, Pará, Mato Grosso, and with an outlier in Sào Paulo. Selected specimens examined. BRAZIL. AMAZONAS. "Estrada do Estanho," road to Igarapé Preto, 60 km SE of Transamazon Hwy., in white-sand savanna, 2 July 1979, Calderon et al. 2737 (INPA, US, VDB). PARÁ: Mun. Itaituba, 2 p pi Pee BR 163, km 794, Serra E Cachimbo, Base Aérea, campina, solo ar- . 1983, seated de en 934 (INPA, abe u, perto do barracao, Reg 7.19638 (UB); Mun. [taituba, arredores da Lagoas," Fazenda Campininbá, just N of Rio Moji-Guacú, 4 km NNW of Padua Sales, 14 Dec. 1962, Eiten 5104 (MO, US). Wet, rocky and/or sandy | p Volume 75, Number 2 Kral 651 1988 Xyris ES cm FIGURE 54. a stenocephala (Silva et al. 82).—a. Habit sketch.—b. Leaf tip.—c. Leaf blade-sheath soa le —d. ase.—e. Spike.—f. Fertile bract.—g. Lateral sepal.—h. Petal blade, stamen.—i. Stylar x.—]J. Cetus d» —k. Outline of capsule, showing placentation.—l. Seed. This taxon is most abundant in low-ele- parable areas north of the Amazon. It borders vation, white-sand savanna, particularly in on larger forms of X. paraensis, particularly eastern Mato Grosso (Norte) and southwest- var. longiceps, but differs markedly in its ern Para but should be looked for in com- glassy-bordered leaf blades and markedly 652 Annals of th Missouri End Garden larger seeds. Populations appear to be some- times annual, although they are usually pe- rennial. Perhaps this is a matter of degree of disturbance of habitat, particularly the very seasonal nature of moisture in white-sand sa- vanna. Or it could be that crowding produces a depauperate, annual habit. 55. Xyris cylindrostachya Kral € Wan- erley, sp. nov. TYPE: Brazil. Amazonas: Mun. Presidente Figueiredo, “Campina das Pedras," ubicada en el km 115, de la Rodovia BR-174 (Manaus-Caraca- rai), en el lado oriental del Igarapé das Lajes, 1958'S, 60%02'W, ca. 100 m, 29- 30 June 1985, O. Huber & Luiz O. Adao Teixeira 10663 (holotype, INPA; isotypes, NY, VDB). Figure 55. Herba perennis, glabra, densicaespitosa; radices gra- ciles. Caules brev tortae, 0.8-1 mm la gitudine indistincte nervosae, dilute virides; apices gra- datim contracti, subulati, subteretes; margines leviter in- crassati, integri, pallide brunneoli vel ad basin fuscobrunneoli, persaepe nitidi; vaginae carinatae, inte- ae, multicostatae, fer- ad apicem apertae, laminis brevibus. Scapi recti vel flexuosi, torti, 30-60 cm longi, ad apicem teretes, 1— 2 mm crassi, persaepe rubelli, sie striolati. epee ongae, mm Vnus bracteae fertiles late ovatae vel suborbicula- e, 4-5 mm longae, integrae, ecarinatae, valde convex areis dorsalibus a ovatis, fuscoviridibus. Sepala liotain libera, valde inaequilatera, curvata; ala carinali lata, a 4.5 ong eolae n personas bibrae bista, e E longipenicillatae. , 0.6- m longa, pallide ces trans- on Ba paee multicosta Smooth, densely cespitose perennial with slender roots. Stems short. Principal leaves narrowly linear, 10-30 cm long, erect to spreading flabellately, longer than the scape sheaths. Blades of leaves flattened, slightly twisted, 0.8-1 mm wide, 2-4 times as long as the sheaths, indistinctly nerved, pale green; tips gradually narrowed, subulate, subterete; margins slightly thickened, entire, pale brown to red-brown at base, often shining; sheaths carinate, entire, red-brown to ferrugineous, papillose, multicostate, gradually dilating be- low, then gradually narrowed into the blades, at apex producing a narrow, acute, scarious ligule to 2 mm long. Sheaths of scapes twisted, tubular, multicostate, ferrugineous, open at apex, with short blades. Scapes straight to flexuous, twisted, 30-60 cm long, terete to- ward apex, l-2 mm thick, usually reddish, finely striate. Spikes ovoid to (commonly) cy- lindric, 1-3 cm long, 5-7 mm thick, obtuse, multibracteate, brownish, the bracts tightly spirally imbricate. Sterile bracts 2-4, oblong, ca. 3 mm long, navicular, carinate, but slight- ly grading into the fertile bracts; fertile bracts broadly ovate to suborbicular, 4-5 mm long, entire, ecarinate, strongly convex, with broadly ovate, red-green dorsal areas. Lateral sepals free, strongly inequilateral, curvate; carinal keel broad, ciliolate from base to apex. Petal blades obovate, 4-4.5 mm long, yellow, broadly rounded, lacerate. Staminodes bibra- chiate, the branches densely long-penicillate. Anthers oblong, sagittate, 1.5 mm long; fil- aments ca. m long. Capsule planoconvex, obovoid, ca. 2 mm long; placenta basal. Seeds curvate-ellipsoid, apiculate, 0.6-0.7 mm long, pale red-brown, translucent, longitudinally finely multicostate. Paratypes. BRAZIL. AMAZONAS: Mun. Presidente A — ne Igarapé das Lajes, 01%58'S, 29-30 June 1985, O. Huber £ la Teixeira 10652 (INPA, ) P miná, Eds do Ariramba, campinas iundaveis da mar- m do Rio Jaramacarú; afloramento arenit 0 m, 8 June Tus G. Martinelli 6847 (INPA. NY, DB). This species most closely resembles X. stenocephala Malme, differing from it pri- marily in its longer, narrower leaf blades, these less strongly bordered and more incon- spicuously nerved. The spikes are very dif- ferent, being mostly narrowly cylindric and blunt (rather than ellipsoid and acute). The f. Volume 75, Number 2 1988 Kral Xyris 653 FIGURE 55. Xyris tind pur (from the type). —4. bon har m sector.—d. Two views of leaf Leaf apex.—c. Leaf midblade . Lateral e bra epal.—i. Petal blade and pt ax js Wen os and mos hna dai ui Sirlar apex. 1 Capsule outline, "eed. lacentation.— m. 654 Annals of the Missouri Botanical Garden lateral sepals are somewhat shorter and the seeds are significantly shorter (0.6-0.7 mm long vs. 1.2-1.7 mm long). These two species, along with the varieties of X. paraensis, X. savannensis, X. uleana, and X. mima, ap- pear to provide an abundant (and perhaps sometimes intergrading) display on the white- sand campinas of the Amazonian savannas. Dense stands are often depauperate and mixed, escaping standard measures and ag- gravating the taxonomic problem. 56. Xyris brachysepala Kral, sp. nov. TYPE: Brazil. Pará: Serra dos Carajás, 2 W of AMZA camp N-5, 6?04'5, 50°08' W, ca. 700 m, scrubby vegetation on ferric rock outcrops, moist low areas, 13 May 1 C. R. Sperling, R. S. Secco, M. Condon, A. L. Mesquita, B. G. S. Ribeiro & L. R. Marinho 5641 (holotype, INPA; isotypes, MG, NY, VDB). Figure 56. nta humilis, annua, glabra. Radices filiformes. Folia linear, solum basalia, (2-)8-12 cm I sat laminas gradatim convergentibus aut ad apicum ligulam oe curtam latam fascientibus, infime gradatim ex- pansae. Vag laxae, plerumque apertae, rec- ae, carinatae, sine lami nis aut laminis brevis. Scapi sub- teretes, subfiliformes, plus minusve spiraliter torti, ca. 0.5 mm un A p Spicae lanceoloideae vel anguste ellip- soideae, 1-2 cm longae, anguste acutae, pluriflorae, brac- teis arcte bla imbricatis, tenuibus, pallide rufobrun- a pice rinatae, acutae; area dorsalis anguste trian gulati, subapicales, virides, tum brunneolae, 2-3 lon gae. Sepala lateralia libera, curvata, aequilatera, lanceo lata, 2-2 m longa, acuta; = i sonis egra Laminae peri. obovatae, m longae, luteolae, ad apicem e . Antherae money dd aM bifidae et sagittatae, ca. 0. 5 mm longae; filamenta plana, ca. 0.5 mm longa . Staminodia parce penicillatis. Capsula leviter voidea, ca. 3 mm longa; ix ds basalis. Semina numerosa, ellipsoidea, 0.6-0.8 a, pallide rufobrunneola, longitudine subtiliter olla translucida. The plant low, annual, smooth. Roots fili- form. Leaves linear, strictly basal, (2-)8-12 cm long, erect to slightly bra mostly longer than the scape sheaths. of the principal leaves 3-4 times vide cee the sheaths, spongy, twisted, slightly compressed at base, otherwise subterete, ca. 0.5 mm thick, greenish, abruptly narrowed at apex, obtuse; sheaths subcarinate, roseate, entire, narrow- ing gradually to blades or producing a broad short scarious ligule at apex, gradually dilat- ing below. Scape sheaths lax, commonly open, straight, carinate, without blades or with short blades. Scapes subterete, subfiliform, + spi- rally twisted, ca. 0.5 mm thick, striate. Spikes lanceoloid to narrowly ellipsoid, 1-2 cm long, narrowly acute, several-flowered, with bracts tightly spirally imbricate, thin, pale red-brown, the margins very thin, broad, pale; sterile bracts 4, oblong to narrowly ovate, 2-4 mm long, acute to obtuse, obscurely carinate, slightly shorter than the fertile bracts; fertile bracts ovate, 5-7 mm long, carinate at apex, acute; dorsal areas narrowly triangular, sub- apical, green then (later) brown, 2-3 mm long. Lateral sepals free, curvate, subequi- lateral, lanceolate, 2-2.3 mm long, acute; keel narrow, entire. Petal blades obovate, ca. 2 mm long, yellow, erose at apex. Anthers oblong, deeply bifid and sagittate, ca. 0.5 mm long; filaments plane, ca. 0.5 mm long. Stam inodia sparsely penicillate. Capsule slightly compressed dorsiventrally, obovoid, ca. 3 m long; placentation basal. Seeds numerous, el lipsoid, 0.6-0.8 mm long, pale red-brown, longitudinally finely lined, translucent. Distribution. Known only from grassy, rocky, acid savannas in Para, Brazil. Additional specimens examined. BRAZIL. PARA: Maraba, N,, arredores do lago, canga, 14 May PAARE R. . Secco, C. Sperling, M. Condon, A. Mes , B. Gilberto R. & L. Marinho 155 (MG, NY, VDB) Muda Serra dos Carajás, N-4, proximo a transigáo para a mata campo rupestre, solo de canga e na mata de terra firme, 20 Mar. 1984, 4. S. L. da Silva, T c Bahia & M. R. Santos 1920 (MG, NY, VDB): Marabá, Mirum estrada p/N, transigao campo natural/veg. can 7 May 1982, R. S. Secco, C. Sperling, M. Condon: 4. ME B. Gilberto R. & Lucival Marinho 226. This little annual bears a strong resem- blance to longer-spiked forms of X. paraensis Poeppig ex Kunth, but its leaves differ by Volume 75, Number 2 Kral 655 1988 Xyris 08 mm FIGURE 56. Xyris brachysepala ia et al. 5641).—a. Habit sketch. —b. Leaf apex.—c. Leaf blade— Eg junction.—d. Spike.—e. Fertile b —f Lateral sepal.—g. Capsule, showing length relative to lateral pals.—h. Petal blade, stamen.—i. Stylar , apex. —j. Staminode.—k. Seed. 656 Annals of the Missouri Botanical Garden being distinctly terete apically; the fertile bracts are larger but distinctly thinner with distinctive broad, pale thin borders and with narrower dorsal areas, these forming a carina subapically on the bract. The lateral sepals are much more reduced than is typical in most Xyris. 97. Xyris paraensis Poeppig ex Kunth, Enum. Pl. 4: 9. 1843. TYPE: Brazil. Para: "Rio Para, Poeppig, 1832" (lectotype, B; phototype, US) Three varieties are delimited by Smith & Downs (Arq. Bot. Estado Sào Paulo, nov. ser. 4(2): 28. 1966) for this complex whose great- est diversity appears to be in Pará, Brazil. The varieties paraensis and longiceps are thus far the only two collected within the area of my concentration, so, while the key below is to three, I am putting down full descriptions only of those two. KEY TO VARIETIES OF V) R/S PAR AENSIS la. Lateral sepals as viewed from the side broadly oblong and blunt; T predominantly ovoid to nearly round, mostly 5 mm long or less ...... 57A. "i paraensis var. paraensis . Lateral sepals as viewed from the side lanceolate and acute; spikes ellipsoid to cylindric, at ma- turity predominantly 1-2 cm lon 2a. Spikes less than 5 mm thick, mostly cylindric, the bracts spiraled in 3-4 ranks 57B. araensis var. longiceps 2b. Spikes 5 mm thick, mostly ellipsoid, the bracts spiraled in 5-6 ranks 57C. X. paraensis var. polystachya p=. c 57A. Xyris paraensis var. paraensis. Figure 57A Cespitose or solitary, mostly low and slen- der, mostly annual plants 0.5-4 dm high, the roots fine, the stems short. Leaves erect to spreading flabellately, the principal leaves 2- 15(-20) cm long, the sheaths eciliate, stra- mineous to brown or red-brown, narrowing gradually from base to blade, with or without a narrowly triangular ligule to 1 mm long; blades flattened, slightly if at all twisted, 1— 2.5 mm wide, narrowly linear, tapering grad- ually to an acute, usually calloused apex, the margins thin, unbordered, usually smooth, the surface green to maroon, very finely nerved and smooth. Scape sheaths shorter than prin- cipal leaves, the base tubular, multicostate, keeled, distally open and with a prominent leaflike blade, or the blade short, erect, fleshy. Scapes straight or slightly flexuous, twisted, distally terete, 0.3-0.8 mm thick, mostly ecostate but finely striate, smooth. Spikes ovoid to subglobose, 0.3-0.5 mm long, mostly acute, the several to many bracts in a spiral, tightly imbricate, with strong, unicostate dor- sal areas, the few sterile bracts grading slight- ly larger into the fertile bracts, these obovate, low-auriculate, ca. 2-4 mm long, the apex broadly to narrowly rounded, entire or erose, the backs rounded, convex, ecarinate. Lateral sepals free, inequilateral, 3-4 mm long, strongly curvate, broadly acute to obtuse, the wide keel coarsely and irregularly ciliate to ciliolate. Petal blades broadly to narrowly ob- ovate, 2-4 mm long, yellow, the broadly rounded apex lacerate. Staminodia bibra- chiate, the flat, narrow branches sparsely pen- icillate terminally. Anthers broadly oblong, 0. mm long, the parallel sacs separated by vide connective, the filaments 0.5-0.7 mm long. Capsules broadly obovoid, some- what compressed dorsiventrally, ca. (1 —)2 mm long, the placentation basal, the valves with- out septa. Seeds ellipsoid or fusiform on long funicles 0.5-0.9 mm long, apiculate, amber, longitudinally distet but finely ribbed. Distribution. Sandy savannas of mostly white sand, locally abundant, rare in Belize; rare in southern Venezuela; frequent to com- mon eastward across Guyana to French Guiana and contiguous northern Brazil. Additional T eren savanna, Hattieville, 7 , Dwyer & Pippin 10981 (MO); wet aa Manatee Ta 5 Jan. 1906, Peck 269 (GH); All Pines, Schipp S-131 (BM, F, GH, MO, NY). BRAZIL. PARÁ: Mun. Itaituba, estrada Santa- rem-Cuiaba, BR 163, km 794, Serra do Cachimbo, Ama- ral et al. 930 (INPA, NY, VDB); Campina do Itajura, Ilha de Colare, 28 Sep. 1954, Black 54-1687a (US, VDB); campina do Palha, Vigia, 10 Aug. 1954, Black 34-16742 (US, VDB); Ilha de Colares, sitio Horizonte, Mun. ban x Sep. 1954, Black 54-16911 (US, V DB); vic. Cachoeira, BR 22, km 98, roadside, 24 Aug. 1964, Pran nce & Silva S. VDB). TERR. RORAIMA: Serra da Lua foothills, 2225-29'N, 69°11-14'W, 12 Jan. 1969, BELIZE: pure sand Volume 75, Number 2 Kral 657 1988 Xyris Lam FIGURE 57A. Xyris paraensis var. paraensis (Black 54-16742).—a. Habit sketch.—b. Leaf tip.—c. Lea sheath—blade junction. — d. Leaf base. —e. Spike. —f. Lateral sepal.—g. Petal blade, stamen.—h. Staminode.— i. Stylar apex.—j. Seed. 658 Annals of the Missouri Botanical Garden FicuRE 57B. Xyris paraensis var. longiceps (Huber 2581) .—a. Habit sketch.—b. Leaf tip.—c. Leaf sheath- blade junction. —d. Leaf base.—e. Spike.—f. Fertile bract.—g. Lateral sepal.—h. Petal blade, stamen.—i. Staminode, enlarged sector of beard hair. —j. Stylar apex.—k. Diagram of capsule and placenta.—l. Seed. Volume 75, Number 2 1988 Kral 659 Xyris Prance et al. 9208 (U). FRENCH GUIANA: Savanes de Kourrou, transect no 12, Institut Francais D'Amerique oe (U). Guyana: “English Guiana, Schomburgk 8" (two specimens, the left-hand one X. savanensis a inter Zanderij I and Hannover, opn. 131, van pondo 4 km E of village pa eni 22 Dec. Donselaar 2841a (U); Zanderi, sand savanna, Flor- schutz 802 rte bis iw anne (fluv. Suriname) tr. . 1956, Heyligers 382 (U); Wilhelmina VDB); via secta ab Moengo tapoe ad Grote Zwiebelzwamp, big swaying swamp near km 18, 21 48, Lanjouw & Lindeman 949 (U); Tibiti I near km 3.2-4.0 second line, Lanjouw & Lindem 789 (U); puc km N of river, 2 km W of d nd 225 July 1963, Maguire et al. T (NY, U, VDB); e Sipaliwini savanna, ca. 2 km '4 Gebroeders Creek, Oldenberger et al. 5954 (U); Fool of Zandery, 31 May Res. “Brinckheuvel” (Saban-Pasi Sav pan sg rasa & IO Per. (U). VENEZUELA. GUA : Morichal Hato Bec 25 km sur de Calabozo, Nov. 1966, L y Dc 6469 (VEN). This variety is the most often misidentified as X. savanensis, which may produce smooth- leaved forms (var. glabrata). Careful ex- amination should solve the problem in that X. savanensis produces no staminodial beard. If flowers are lacking, then seed characters help in that the shorter seed of X. savanensis has the typical truncate and apiculate apex. On the other hand, X. mima would pose a problem except for its uniquely comose seed tips. 57B. Xyris paraensis var. longiceps (Malme) Lyman B. Smith & Downs, Ark. Bot. Estad. Sào Paulo 4(2): 28. 1966. Xyris a Malme, Rec. Trav. Bot. Neerl. 9: 131. 1912. TYPE: “in arenosis inundatis a (Suriname), Splitgerber 978" (lectotype, U). Figure 57B. Xyris leptostachya Malme, Rec. Trav. Bot. Neerl. 9: 32. 1912. Leaf blades often longer and wider than in the typical variety, the leaf tips tending to be less thickened, sharper, the habit sometimes perennial or the plants harder-based, more strongly tufted. Spike outline narrower, most- ly narrowly ellipsoid, or fusiform-cylindric, mostly 1-2 cm long, rarely wider than 0.4 cm. Lateral sepals predominantly lance-ovate, acute, the narrow keels rather uniformly cil- iolate. Seeds mostly 0.5-0.6 mm lon Distribution. Sandy savanna of mostly low-elevation, southeastern Colombia and eastward (occasionally) into Surinam (abun- dantly) and northern Brazil (Amapa, Para). Selected specimens examined. COLOMBIA. AMAZONAS: a in Araracuara savannas, Rio Caquetá, 400 m 6 Sep. 1959, dba et al. 44162 (NY). VAUPÉS: Río Vaupés, cachivera de Yurupari, 400 m, 24-26 Oct 1952, García- Barriga 14936 (NY); Rio Kudujarí: Cem Yapoboda, ca. , 9-6 Oct. 1951, Schultes & Ca- brera 14379-a grin Rio Negro, San Felipe A vic., Schultes & Cabrera 18116 (GH); Circasia: savanna, ca. 800 ft., Nov. 1951, Schultes & Cabrera 19647. SURI- NAM: Sun Zanerij I, Boldingh 3044 (U); zandsa- ct. 1958, van Donselaar 377 (U) Zanderij I, 14 Sep. 1937, Essed s.n. (U); prope Jodensavanne (fluv. Suriname) in savannis arenosis, Hey- ligers & Knoppe 325 (U); Mata, 15 km in dir. occid. a Zanderij, 27 Nov. 1960, Kramer & Hekking 2239 (U); Zanderij I, open wet sand savanna, 4 Nov. 1948, Lan jouw & Lindeman 162 (U, VEN); E of Kopie Venikbà R. distr. Commewijne, 16 July 1953, Lindeman 4407 (U); iter secundum surinamense, July-Sep. 1926, Pulle 58a (U). VENEZUELA. T. F. AMAZONAS: alrededores del Campamento Asisa, margenes del Rio Asisa, trib. del Rio Parú, 8 May 1973, Hoyos & Morillo 126 (US, VEN); alrededores de Puerto Ayacucho, 26 Jan. 1978, Huber & Cerda 1445 (US, VEN); Aeropuerto de Maroa, white sand, 25 Aug. 1978, Huber 2581 (US, VEN); ca. 30 km al N cian Ayacucho, ca. 80 m, 7 Nov. 1979, Huber 4696 (US); Alto Ventuari, Jan. 1959, Infante 44016 (VEN); Río Ventuari, Rio Parú, etc. Sarrania Parú Expedition, 200 m, 15 Feb. 1949, Phelps & Hitchcock 1 (NY, US, VEN); Maroa, 23 July 1982, Stergios & Aymard 4014 (PORT, VDB); Esmeralda Savanna and SE base Cerro dup 200 m, 22 M ; pe ton 57827 (F; M , 80 m, airstrip, 7 1983, du al 129426 (VDB, VEN) ae ca. 325 ft., . 1928, Tate 301 (NY). APURE: betw. Rio Cot near er of Cano San Miguel and southern part of = Galera de Cinaruco, 29-30 Apr. 1977, Davidse & zalez 12348A (US, VEN); E side of Galeras de Cinaruco, ca. 53 airline km o Davidse & c 15571 (MO). Bo orillas del el Salto i Ferry, P 1984, Kral 72190 (MYF, VDB, VEN, and to be dis- doped. along Rio Karuai, NW of Sta. Teresita de Ka- vanayen, 1,220 m, Selena 60823A (US). The petiolelike constriction of old bracts in this and in var. polystachya is created by infolding of the auriclelike bract base. The closest affinity of this variety, other than to 660 Annals of the Missouri Botanical Garden var. polystachya, is to Xyris cuatrecasana, which may turn out to be merely a part of the latter. 58. Xyris mima Lyman B. Smith & Downs, Proc. Biol. Soc. Wash. 73: 250, fig. 4. 1960. TYPE: Brazil. Pará: campo arenoso artificial, Missao Nova, Rio Cururu, re- giào do Alto Rio Tapajos, 12 July 1959, Egler & Raimundo 791 (holotype, US). Figure 58A, B. Xyris d Kral & Lyman B. d Phytologia 53: 433-4 n la-i. 1983. : Venezuela. Boli- e de Cerro Guaiquinima, Salto del Rio m río arriba de 0-25 Jan. 1977, J. mark et al. E TE SA "VEN; iso- types, "Us. VDB). Solitary or cespitose, slender, mostly low annual (0.5-)1-3 dm high, the stems con- tracted. Leaves erect or spreading flabellate- ly, 0.5-2.5 dm long; sheaths entire, strongly keeled, less than ! as long as blades, grad- ually narrowing from the dilated base to the blade, there producing a narrowly triangular erect ligule to 0.5 mm long or eligulate, the surfaces tuberculate-rugose (rarely smooth); blades narrowly linear to linear-gladiate, strongly flattened, sometimes slightly twisted, apically incurved-acute, the margins tuber- culate-scabrid (rarely entire), the surfaces verrucose-scabrid (rarely smooth), strongly nerved, deep reddish green to maroon. Scape sheaths slightly to much shorter than principal leaves, reddish brown below, multicostate, tu- bular, twisted, opening distally, strongly keeled, producing a short blade. Scapes straight or slightly flexuous, twisted, papillate or rugoscabrid, dull green, subterete to slight- ly compressed distally, thus oval or elliptic in cross section, bicostate, the costae strong, usually strongly tuberculate-scabrid or pa- pillate. Spikes ovoid to short-cylindric or nar- rowly ellipsoid, 3-15 mm long, mostly acute, of few-several tightly spirally imbricate bracts with narrow but distinct, usually greenish dor- sal areas, the sterile bracts mostly 4-6, the lowest the smallest, slightly shorter than the fertile bracts, ovate, keeled or carinate, grad- ing into the fertile bracts, these mostly broadly ovate to broadly obovate, 4-4.5 mm long, the apex broadly rounded, sometimes emar- ginate, the margins entire, the backs ecari- nate, convex. Lateral sepals free, slightly in- equilateral, oblong-curvate, ca. 3-3.5 mm long, the broad, thin keel ciliate mostly from just below middle to the blunt apex. Petal blades broadly obovate, ca. 4 mm long, yel- low, the broadly rounded apex lacerodentate. Staminodia bibrachiate, the narrow, flat branches penicillate-ciliate distally with cla- vate hairs. Anthers broadly oblong, ca. 0.5 mm long, shallowly bifid, deeply sagittate, on filaments ca. 0.6-0.7 mm long. Capsules broadly obovoid, planoconvex, ca. 2.5 mm long, the placentation basal, the valves lack- ing septa. Seeds numerous, mostly ellipsoid to fusiform, ca. 1 mm long, pale amber, the ip with a cone of pale bristles, the body finely longitudinally lined. Distribution. Low, mostly white-sand savanna, Amazonian Brazil and contiguous Venezuela (Bolivar), locally abundant. Known n 1960 but only from the type collection. Selected specimens examined. BRAZIL. AMAZONAS: Transamazon Highway, 9 km W of Rio dos Pombos, ca. .5 km E of Igarapé dos Oombos and ca. 64 km E of the Aripuanà, 19 June 1979, Cleofé Calderón et al. 2577 (INPA, US, VDB); base of Serra Araca, 0-3 km south of Central Massif, 3 km E of Rio Jauari, 7 Feb. 1984, Prance et al. 28870 (INPA, NY, US). MATO GROSSO: R. Juruena, Cachoeira Sao Simao, 21 May 1977, Rosa & Santos 1959 (INPA, US, VDB). PARÁ: Mun. Itaituba, arredores da Base Aerea do Cachimbo, proximo ao destacamento km 6 da estrada que vai para o aeroporto km 794, 25 Apr. 1983, Silva et al. 82 (INPA, NY, VDB). VENEZUELA. BOLÍVAR: topotype of X. trisperma Kral & Smith, 20-25 Jan. 1977, Steyermark et al. 113222, 113445 (VDB, VEN). This species very much resembles X. sa- vanensis and X. paraensis var. paraensis. From the former it differs in its strongly bi- costate scape, staminodia with beards, and longer and scaly-comose seeds. From the lat- ter it usually differs in being much more sca- brid of foliage and in its bicostate scape and longer, comose-tipped seeds. In the Venezuelan locality the Steyermark discovery, found in a mixed population with X. savanensis, seemed to depart enough from Volume 75, Number 2 Kral 661 1988 Xyris 4mm FIGURE 58A. Xyris mima (Steyermark & Dunsterville 113445) .—a. yer sketch. = Leaf tip.—c. Leaf sheath-blade junction. —d. Spike, upper scape.—e. Lateral sepal. —f. Petal blade, stamen.—g. Staminode.— h. Stylar apex. —i. Capsule, two valves removed.—j. Seed. —k. Much- tees sector of leaf blade edge. 662 Annals of the Missouri Botanical Garden icm URE 58B. Xyris mima (from type of X. trisperma) .—a. Habit sketch. —b. Leaf tip. —c. Leaf blade, sector at nudi de. —d. Leaf sheath-blade junction. —e. Leaf base. —f. Spike, upper scape.—g. Midscape sector h. Lateral sepal.—t. Seed. the original Smith and Downs description in regard to seed and sepal characters to warrant description. Cryptic species are not rare in Xyris, should be looked for, and at least for now deserve some place in the literature if only to acquaint readers with the “motion” in such a genus. However, it now appears, on consultation with a larger series of X. Volume 75, Number 2 1988 Kral Xyris 663 mima, that X. trisperma fits within the for- mer. 59. Xyris rubrolimbata Heimerl, Ann. k. k. Naturh. Hofmus. Wien 21: 70, pl. 4, figs. 4-6. 1906. TYPE: Colombia. Vaupés: moist sandy woods, Lomo, 100 km northwest of mouth of Casiquiare, Vaupés, Colombia, Spruce 2994 (lec- totype at W destroyed; isolectotype, K). Figure 59. Low, cespitose, small but stiff annual most- ly 0.5-1 dm high, the stems contracted, the roots capillary. Leaves spreading flabellately, -8 cm long, the sheaths from nearly as long as blades to longer, soft, strongly keeled, strongly nerved, entire, tapering gradually from the dilated base to the blades, there producing a narrowly triangular, erect, scar- ious ligule to 2 mm long; blades flattened, straight or rarely slightly twisted, linear-gla- diate, narrowly acute, the margins cartilagi- nous-incrassate, usually forming a distinct, reddish brown band, the surfaces greenish gray or maroon, strongly and palely multi- nerved, smooth. Scape sheaths slightly short- er than principal leaves, tubular and twisted at base, striate, open toward apex, keeled, producing a short, flat, linear-triangular blade. Scapes straight or slightly flexuous, slightly twisted, terete distally, ca. 0.7 mm thick, greenish brown, striate, smooth. Spikes ob- ovoid at maturity ca. 5 mm long, blunt, of several subdecussately arranged brown bracts with thin, usually fimbriolate margins and dis- tinctly papillose dorsal areas; sterile bracts narrowly to broadly ovate, usually 4, the low- est pair ca. 44-% as long as the spike, keeled, navicular; fertile bracts broadly obovoid, 3- 3.5 mm long, the shallowly rounded or sub- truncate apex aging erose, the backs slightly convex, the dorsal areas bisected by a single strong nerve. Lateral sepals strongly curvate, free, subequilateral, 2.5-3 mm long, blunt, yellow-brown, the firm, wide keel ciliolate- scabrid from ca. the middle to the apex. Petal blades obovate, ca. 3 mm long, the broadly rounded apex lacerate. Staminodia slightly bibrachiate, the broad, flat branches distally penicillate-ciliate. Capsule planoconvex, broadly obovoid, ca. 1.5 mm long, the pla- centation basal, the valves without septa. Seeds numerous, ellipsoid, 0.6-0.7 mm long, pale- apiculate, dark, translucent red-brown, lon- gitudinally finely and spirally ribbed, the ribs finely beaded. Distribution. Low Amazon caatinga, wet sand in southeastern Colombia and south- western Venezuela, apparently rare. Additional specimens examined. COLOMBIA. VAUPÉS: Río Guainía, in Colombia (opposite Ven. Town of Maroa) and vic., ca. 800-850 ft., 31 Oct.-2 Nov. 1952, = ei et dk 18195 (GH). VENEZUELA. T. F. AMAZONAS: f Cerro Cucuy, 2 Mar. 1944, Baldwin 3212 (US); ¡Pest Yavita rd., between Río Guiania and Caño Pimichin, at edge of bana (low Amazon caatinga) on wet white sand, 8 Oct. 1978, Clark 6875 (NY); Maroa, 127 m, Guainia Alto Rio Negro en malezas de terreno arenoso, 11 Feb. 1942, Ll. Williams 14280 (F, US, EN). This little but uniformly distinctive annual xyris is so small as to be overlooked easily. It should be looked for in areas of Amazonian Brazil in places along the upper Rio Negro. 60. Xyris cuatrecasana Idrobo & Lyman B. Smith, Caldasia VI(29): 244-245, fig. 30a-d. 1954. TYPE: Colombia. Vaupés: Comisaria del Vaupés. Circasia; mar- genes del Vaupés, sobre granitos, 200 m, 9 Oct. 1939, J. Cuatrecasas 7155 (holotype, COL; isotype, F). Figure 60. Solitary or cespitose, soft-based annual to 3.5 dm high, the stem contracted. Leaves erect or slightly spreading flabellately, 8-12 cm long; sheaths ca. Y2 or less as long as blades, entire, dull at base, pale red-brown or stramineous, tapering gradually to blade, there with a scarious, erect, narrow, acute ligule to 2 mm, the blades flat, ensiform-linear, 1.5- 2.5 mm wide, tapering to narrowly acute apex, the margins subentire, the surfaces smooth and finely multinerved, green with strong ma- roon tints. Scape sheaths loose, shorter than leaves, carinate, short-bladed. Scapes lineal, straight or flexuous, slightly twisted, distally terete, ca. 1 mm thick, finely striate, ecostate, smooth. Spike ovoid-ellipsoid, ca. 1 cm long, acute, the numerous bracts tightly spirally 664 Annals of the Missouri Botanical Garden SU 3 mm K 4 FIGURE 59. Xyris a (J. T. Baldwin 3212).—a. Habit sketch.—b. Leaf apex.—c. Leaf blade- sheath junction.—d. outlined.—e. Spike. —f. Fertile bract.—g. Lateral sepal.—h. Petal blade, stamen.— i. Staminode.—j. St dr apex. —k. Capsule, two valves removed to show placentation. —l. Seed. Volume 75, Number 2 Kral 665 1988 Xyris 1cm FIGURE 60. Xyris cuatrecasana (from the isotype) .—a. Habit sketch.—b. Leaf apex.—c. Sector of blade at midblade.—d. Leaf sheath—blade junction.—e. Leaf base.—f. Spike and upper scape.—g. Fertile bract.—h. Lateral sepal. imbricate, thin, the sterile bracts several, erose edges. Lateral sepals free, subequilat- smaller than the fertile bracts and grading eral, ca. as long as bracts, very thin, lanceo- into them, these ovate, ca. 5 mm long, acute, late, narrowly acute, the firm, narrow keel ecarinate, with strong, dark, lanceolate, ve- subentire. Corolla, staminodes, and stamens nose dorsal areas and scarious, subentire to not seen. Immature capsule ellipsoid, ca. 4 666 Annals of the Missouri Botanical Garden mm long, the valves eseptate, the placentation basal. Immature seeds ellipsoid, ca. 1 mm long. Distribution. Known only from the type. This is perhaps not distinguishable from Xyris paraensis and except for the equilat- eral sepals could be considered a disjunct col- lection of var. polystachya of that species. However, until more information (seeds, flow- ers) becomes available, it seems best to retain the taxon 61. Xyris pectinata Kral, Lyman B. Smith & Wanderley, sp. nov. TYPE: Brazil. Amazonas: Estrada Transamazónica- Capim, Ln aberta, n arenoso, Proj. , 1 June 1976, T. R. Ba- hia 35 jode INPA; isotypes, US, VDB). Figure 61. Herba perennis M RPM din tenella, glabra; radices graciles. Folia line earia, 4-6 cm longa, erecta vel leviter latae, vaginis 3 marginibus in laminas gradatim convergentibus, ad api- cem ligulam acutam o: mm longam fascientes, infime adi dilatatae. Vaginae scaporum laxae, plerumque subteretes, filiformes, plus minusve spiraliter torti, 1.2- 2 dm alti, ca. 0.4-0.5 mm crassi, olivacei, distaliter acute bicostati, costis laevibus. ' Spicae subglobosae vel late ob- ovoideae, 4.5-6 mm longae, pluriflorae, breviter atten- uatae; bracteae steriles 2(-4), paro infimo oblongo, 2- > ongo, areis dorsalibus linearibus bracteam ae duas iue bracteae fertiles arcte spiraliter imbricatae, late diri obova ongae, convexae et lev rus ap q ad apicem utrinque erosae et ci riosae, minute tuberculato- Pond a medi multo crassiores, nitidae, brunneolae, marginibus effuse [o .9 > ratae, luteolae. Staminodia aliquantum redacta, bibrachiata, E ad apicem breviplumosis, pilis moniliformibus. Antherae oblongae, ca. 0.3-0.4 mm longae, loculis pa- rallelis du filamenta ca. 0.5 mm longa. Capsula dorsali-ventraliter compressa, oblongo- sud tenuis- sima, 1 mm longa; placenta basalis. Semen soli- tarium, lenticulariter .2 mm lon gum, translucidum, vallis luteo- joies ne longitudine uber striatum. Delicate, smooth, cespitose perennial; roots slender. Leaves linear, 4-6 cm long, erect or somewhat spreading, longer than the scape sheaths; blades 3-5 times longer than the sheaths, plane or slightly twisted, 0.9-1.2 mm wide, longitudinally distinctly multi- nerved, strongly flattened, ferrugineous to ol- ive green; apices contracted, incurved-acute; margins thickened, minutely ciliate; sheaths carinate, with carinae minutely red-ciliate, in- crassate, the sides strongly longitudinally nerved, pale to deep red-brown, the margins gradually converging into the blade, at apex producing an acute ligule 0.5 mm long, grad- ually dilating below. Sheaths of scales lax, mostly open, twisted, shining toward the base, carinate at the middle, with blades either sim- ilar to those of principal leaves or shorter. Scapes subterete, filiform, + spirally twisted, 1.2-2 dm high, ca. 0.4-0.5 mm thick, oli- vaceous, sharply costate distally, the costae smooth. Spikes subglobose to broadly obovoid or short-cylindric, 4.5-6 mm long, several- owered, short-attenuate; sterile bracts 2 (-4), the lowest pair oblong, 2-2.5 mm long, with dorsal areas linear and equal to them in length; fertile bracts tightly spirally imbricate, broadly ovate, obovate, suborbicular, or re- niform, ca. 3 mm long, convex and slightly carinate, obtuse to subtruncate at apex on either side, erose, scarious, minutely rugu- lose-tuberculate, much thicker from the mid- dle to the base, shining, brownish, with mar- gins effusely and pectinately rigidly fimbriate; dorsal area ovate, ca. 2-2.5 mm long, gray- green. Lateral sepals ca. V? connate, ca. 2 mm long, inequilateral, the lobes acute, scar- ious, the carinal keel narrow, entire. Petal blades narrowly obovate, ca. 1.5 mm long, apically rounded, lacerate, yellow. Staminodia somewhat reduced, bibrachiate, the branches at apex short-plumose with moniliform hairs. Anthers oblong, ca. 0.3-0.4 mm long, the locules parallel, distinct; filaments ca. 0.5 mm long. Capsule dorsiventrally compressed, ob- long-cylindric, very thin, 1.2-1.3 mm long; Volume 75, Number 2 Kral 1988 Xyris 667 RE 61. Xyris p (T. R. Bahia 35) .—a. Habit sketch.—b. Leaf tip.—c. iro of midblade.—d. Blade-sheath junction.—e. Leaf base. "dd Spike.—g. Fertile bract.—h. Lateral sepals. — blade, stamen, staminodial apex.—j. See . Stylar apex, petal 668 Annals of th Missouri En Garden placenta basal. Seed solitary, lenticularly ob- long-ellipsoid, 1-1.2 mm long, filling capsule, translucent, pale yellow-brown, finely longi- tudinally striate. Additional specimens examined. BRAZIL. AMAZONAS: Transamazona Highway, 53 km W of Aripuaná River; "campina" region, common in open campina of white "y soil, 27 June 1979, C. E. Calderón, O. P. Monieiro & J. Guedes oe laa US, VDB); N acima de Terra Preta, campina do Rio Surubím, den do Rio Mi 4°29'S, 58*33'W. Campina aberta, areia branca. Erva de 10 cm de altura; flores amarelas, 4 July 1983, C. A. Cid 4026 (INPA, NY, VDB). This species is easily distinguished by its fringe of strong though slenderly tapering rig- id bristles on the margins of the fertile bracts, nearly separate anther locules, and reduced staminodial condition. It is unusual in its par- ticularly small and thin-walled capsule, this tightly filled by a single large seed. 62. s savanensis Miq., Linn. 18: 605. TYPE: “Surinam, Berlyn, Focke os (holotype, U). Figure 62. Solitary or cespitose, soft-based annual 1- 5 dm high; stem contracted. Leaves spreading flabellately to erect, (2-)5-20(-25) cm long; sheaths 1⁄4 or less as long as blades, entire, red-brown to tan or brown, papillose-rugulose, narrowing gradually from wide base to blade, there with an erect, narrowly triangular ligule to 0.5 mm long, or eligulate; blades flat, slight- ly twisted, gladiate-linear, 1-5 mm wide, nar- rowing gradually above middle to a narrowl acute apex, the margins slightly thickened or not thickened, papillose to scabrid, surfaces mostly rugose-scabrid. Scape sheath shorter than leaves, multicostate and tubular, keeled, open and short-bladed above. Scapes straight or flexuous, twisted, terete or subterete dis- tally, 0.5-1 mm thick, ecostate to tricostate or striate, the costae and surfaces scabridu- lous or papillose (rarely nearly smooth). Spikes ovoid to cylindric, ellipsoid or subglobose, 0.3— 0.7(-1) cm long, with many spirally imbricate bracts, these tan or reddish brown with darker green or maroon dorsal areas; sterile bracts smaller than fertile bracts, grading into them, the fertile bracts broadly obovate to subor- bicular, 3-5 mm long, the apex broadly rounded, entire, backs strongly convex- rounded, ecarinate. Lateral sepals free, very inequilateral, elliptic, ca. 3 mm long, acute to obtuse, the strong, curvate keel irregularly ciliate and/or ciliolate. Petal blades broadly obovate, yellow, 2-2.5 mm long. Staminodia bibrachiate, the lance-linear branches beard- less. Anthers oblong, ca. 0.5 mm long, deeply emarginate and sagittate on filaments about as long. Capsule broadly obovoid, 1.5-2.5 mm long, placentation basal. Seeds numerous, ellipsoid or ovoid, 0.4-0.5 mm long, apically truncate and minutely apiculate, the body pale yellow-brown or red-brown, translucent, faintly 20-24-ribbed and indistinctly cross- lined. Distribution. South America, from the Andean foothills eastward, mostly at low to medium elevations, south into Argentina. Records of this species from Mesoamerica have all turned out to be X. paraensis. It is so commonly collected that even a partial citation of records (as in the case of the equal- ly common X. jupicai, X. laxifolia, X. fal- lax) is better done separately. It is not surprising that a widespread, com- mon, and weedy xyris would show consider- able variation in habit and in character of indumentum, pigmentation, and other fea- tures. Thus two additional varieties, X. sa- vanensis var. glabrata Seubert, (Fl. Bras. 3(1): 217. 1855) and var. procera (Malme) Malme (Ark. Bot. 13(3): 53. 1913), have been de- scribed along with some forms. The former variety, as the name suggests, has totally smooth foliage, while the latter assumes (sometimes?) a perennial habit. The species nearest it taxonomically is X. paraensis, and here would be a great difficulty were it not for a pair of characters that, regardless of variety, appear to hold throughout the range. All specimens of X. savanensis lack bearded staminodes; all have seed uniformly truncate at apex, there with a short but distinguishable apiculus. 4 Kral 669 Volume 75, Number 2 1988 Xyris “Ka NN LILO eee aL tS EEE! » orm RT WOO oe te LU M LA, / A Ea tin t tor EATA , "ass Xyris savanensis (Huber 5744, Wurdack & Adderley 43307) .—a. Habit sketch. —b. Leaf apex. — c. Leaf at junction of sheath and blade.—d. Leaf base.—e. Spike.—f. Fertile bract.—g. Lateral sepal.—h. Petal blade, stamen.—i. Staminode. —j. Stylar apex.—k. Dehiscing capsule.—l. Seed. FIGURE 62. 670 Annals of the Missouri Botanical Garden 63. Xyris surinamensis Sprengel, Tent. Suppl. Syst. Veg. Linn. 2. 1828. TYPE: "Suriname, Weigelt, 1827" fide J. Lan- jouw, Rec. Trav. Bot. Neerl. 34. 1937. Possibly (fide Lanjouw) based on same material as X. eriophylla Reichenb., but still a problem, the marked type not found. Figure 634A, X. eriophylla Reichenb., Pl. Excic. Weigelt, 1827 (?). TYPE: Surinam, Leg. & Exc. Weigelt 1827 (?lec- totype at Delessert Herbarium; isolectotype, MO; phototype, GH). Cespitose, hard-based, often bulbous, stocky perennial 2.5—4 dm high. Leaves often nearly as long as scapes, ascending or slightly spreading, 1.5-3 dm long; sheaths long-cil- iate, usually with brownish hairs, the dilated base ecarinate, red to purple or deep brown, rugose, gradually narrowing to leaf blade, mostly eligulate; blades linear, flattened, 2- 4 mm wide, often twisted, dull green, abruptly blunt-tipped or incurved-acute to incurved- narrowly rounded, the margins mostly pale, incrassate, ciliate or scabrociliate, rarely pa- pillate; surfaces rugulose, strongly nerved, the strongest nerves often pale-incrassate, often hirtellous or scabridulous. Scape sheaths high- ly variable, keeled, from much shorter than leaves and elaminar to nearly as long and similar in blade. Scapes flexuous, twisted, flat- tened, distally 2-3 mm wide with strong cos- tae comprising edges, often with one or both costae pale pilose-ciliate or scabrid, the sides pale yellow-green, strongly rugose, sometimes with additional lower costae. Spikes subglo- bose or broadly turbinate, 0.8-1 cm long, of many spirally imbricate or subdecussate, stiff, dull brown or red-brown, ciliolate or entire bracts with large, ovate dorsal areas. Sterile bracts ovate-triangular, slightly keeled, lower bracts much smaller than the fertile bracts and grading into them, the fertile bracts ob- long to obovate, 6-8 mm long, slightly car- inate, apically obtuse-angled or narrowly rounded, the dorsal area of lower ones venose and sometimes white pilose. Lateral sepals free, subequilateral to inequilateral, lance-lin- ear, curvate, 5-7 mm long, acute, the firm keel lacerate, villosulous or ciliolate from mid- dle to apex. Petal blades broadly obovate, 5- 6 mm long, yellow, the broadly rounded apex erose-denticulate. Staminodia bibrachiate, the slender branches densely long penicillate-cil- iate. Anthers oblong, ca. 2 mm long, deeply bifid and sagittate, on filaments ca. 1.5 mm long. Capsule narrowly to broadly obovoid, 4—5 mm long, the placentation appearing bas- al-central, but the valves dehiscing to reveal strong septa. Seeds few, cylindric-fusiform, ca. 1 mm long, pale amber, faintly longitu- dinally striate and cross-lined, often addition- ally with a few, conspicuous, irregular, dark red-brown ribs. Distribution. Locally abundant in low- to high-elevation savanna from southeastern Colombia eastward across southern Venezuela into French Guiana, southward into Amazo- nas and Pará, Brazil. Selected specimens examined. BRAZIL. AMAZONAS: Serra Araca, 10 Feb. 1975, Pires 15000 (MG, US, VDB); margens de um riacho da serra Araca, 1 Nov. 1978, Rosa & Lira 2343 (US). COLOMBIA. — a scrub savanna, Araracuara, Rio Caqueta p. , Ma- guire et al. 44111(NY, VEN). v AUPÉS: Cerro I dodi. 5 Oct. 1951, Schultes & meas 14234-A (GH); Cerro Kanenda, ca. 800-900 10 Nov. 1952, Schultes & Cabrera (GH). GUYANA: Saesdyke, Nov. 1973, Cooper 51(U);N Rupununi, Apr. 1968, Dave 773(K); Wenamu Trail, Krabu Savanna, 28 Oct. 1966, Forest ps Brit. No. R. B. Guiana Field 139, Record No. 7972 (K); age Distr., Chaakoitou, near Mountain Tu 26 Oct. 1979, Maas & Westra 405 1 (K, U, US); Pakaraima Mts., Mt. Membaru, ca. 400 m, 12 Nov. 1979 as & Westra 4342 (K, U, US); Pokorni Mts., K Nov. 1979, Maas & Westra 4371 Amerindian vill., (U, US); s bs 1, Maas et al. nna, pu. Falls, 2,700 ft., guire et al. 46015 (K), Kaietur ien, L 100 ft., 7 Sep., Sandwith 1420 (U); Kaietur Savanna, 30 Aug. 1933, Tutin 659 (U). SURINAM: Zanderij, 21 Mar. 1959, van Donselaar 479 (U); Zanderij, Kruid, 21 Dec. 1950, Florschutz 737 (U); same locality, 22 July 1933, Lan- jouw 135 (U); Upper Commewijne River, W of Sapende, 14 July 1953, Lindeman 4241 (C, U); Iter secundum surinamense, July-Sep. 1920, 28 July, Pulle 45 (U); Nat. Res. Brinckheuvel (Saban Pasi Savanne), 2 Sep. , Wildschut & Teunissen 11602 (U). VENEZUELA. T. F. AMAZONAS: Serranía del Parú, 1,100 m, 3-4 Oct. 1979, Huber 4287, 4323, 4331 (VEN), 4333 (US); sabana ubicada en la margen derecha (E) del bajo Rio Pacimoni, Huber & Medina 5871 (VEN, VDB); summit Cerro Guanay, 1,800 m, 2 Feb. 1951, Maguire et al. 31701 (NY, Dee vh Hechimoni, 8 km Rx mouth Rio Siapa, 9 Feb. Maguire et al. 37 (GH, Y, V N EN); a r. Esmeralda PACA and Volume 75, Number 2 Kral 671 1988 Xyris ni ENS: y TERM gu IS PE ud uan eu: III) yum PPP EVE IYY T PAROLE A¥} AS 7 A: Lot je vy yr: ES AY SP ^ x AA - E; PASTING s TRAP A Li "TO vitm mme + Sd «WT Den NATA UY WD) [PAR nir EN. FIGURE 63A. Xyris surinamensis (Kral 70568, Steyermark 58405). 70568 at lefi, Steyermark 58405 at right.—c. Leaf at midblade, Kral 70568 at lefi, Steyermark 58405 at right.—d. Leaf at sheath apex.—e. Leaf base.—f. Spike.—g. Lateral sepal.—h. Petal bl —a. Habit sketch.—b. Leaf apex, Kral l sepal.—h ade, stamen.—i. Staminode, enlarged apex of beard hair.—j. Stylar apex.—k. Capsule.—l. Seed. 672 Annals of the Missouri Botanical Garden ma Ñ ia i n A) ` mn ni FIGURE 63B. Xyris surinamensis (Pires 15000). n tion.—d. Leaf base.—e. Spi rode.—j. Stylar apex. Sabana Grande, NE of c 6 Sep. ides i Sia 56405 (F, NY); Grand pem Esm , ca, 5 ft., 1 Nov. 1928, Tate "304 Spes Y). BOLÍVAR: cerca de Sen ae de Camoiran, ca. 1,240 m, 8 Jan. 1982, Cordero y Utrera 5 (PORT); ca. 17 km al NE de Ikabaru, —a. Leaf apex.—b. Sector of leaf midblade.—c. Leaf- (oi ke.—f. Sector of upper scape.—g. Lateral sepal.—h. Petal, stamen.—i. ca. 1,100 m, Huber et al. 6723 (MYF, VDB, PENA cerca de la ribera Norte del km al 1983, 70 (NY); al E del Churi-tepui en el Vallé del Rio Karuay inferior, 18 Nov. 1984, Huber Volume 75, Number 2 1988 Kral 673 Xyris 9794 (MYF, VEN, VDB); ca. 5 km al norte del Poblado San Francisco de Yuruani, 19 Jan. 1985, Huber 9949 (MYF, VDB); el piedemonte septentrional del Cerro Zum- bador, Huber 10271 (MYF, VDB); ca. 35 km al W del caserio de Chiguao, 23 Mar. 1985, Huber 10354 (MYF, B); 1.5 km E of Kavanayen in Gran Sabana, bog, 27 en. 1983, Kral 70536 (BM, F, L, MO, NY, SP, U, US, B, VEN, and others); Salto Yuruani, T July 1983, ioe 70568 (BM, F, K, L, MO, NY, U, US, VDB, VEN, and others); El Pauji, savanna E of town, 2 Nov. 1985, Liesner 19357 (MO, VDB, VEN); savanna vic. Uriman, 300 m, 30 Apr. 1953, Steyermark 75265 (C, F, NY, VEN) As one might suspect from such a wide geographic distribution, the species is highly variable in stature, leaf dimensions, and in indumentum of scapes, leaves, and sepals. Its pale yellow blooms with broad petals expand in morning. 64. Xyris pratensis Maguire & Lyman B. Smith, Mem. New York Bot. Gard. 10: 34, fig. 19A—E. 1963. TYPE: Venezuela. T. F. Amazonas: pubescent, frequent in wet hummocks, Camp Savanna, Campo Grande, 1,500 m, Cerro Sipapo (Pa- raque), 10 Dec. 1948, Maguire & Politi 27581 (holotype, NY; isotypes, GH, US). Figure 64. Cespitose, slender, hard-based perennial 4— 5.5 dm high, the stems short, ascending, cov- ered by chaffy bases of old leaves. Leaves erect, 2-4 dm long; sheaths sparsely pilose- ciliate, as long as blades or longer, abruptly constricted above the dilated, deep red-brown ase, then narrowing upward and keeled to blade, the margins intermittently pilose-cil- iate; blades narrowly linear, flat, 2-3 mm wide, slightly narrowing above middle then abruptly incurved-acute or obliquely acute, the thickened tip often white bristly-ciliate, the margins thin, pale ciliate, the surfaces green with streaks of red, strongly multi- nerved. Scape sheaths somewhat shorter than leaves, above with strong blades like leaves but narrower. Scapes somewhat compressed distally, ca. 1-1.5 mm wide, elliptic in-eross section, with the two costae making edges, strongly pale ciliate or with one costa smooth. Spikes ellipsoid, becoming obovoid, 8-10 mm long, the several bracts subdecussate, rather loosely imbricate, dark red-brown with dis- tinct and usually paler dorsal areas, firm, the sterile bracts narrower and shorter than the fertile bracts, the lowermost keeled, some- times pilosulous-ciliate, grading into fertile bracts, these oblong, bluntly acute or nar- rowly rounded, entire or apically villosulous, 6-7.5 mm long, the backs convex, ecarinate but with dorsal areas bisected by a strong costa. Lateral sepals equaling bracts or slight- ly longer, free, subequilateral, linear-oblan- ceolate, 7-8 mm long, acute, the dark, firm keel entire or above middle low-lacerate and also often minutely ciliate. Petal blades ob- ovate, ca. 5.5 mm long, yellow, apically broadly acute. Staminodia bibrachiate, the flat branches densely penicillate-pilose. Anthers lance-oblong, deeply bifid, sagittate, 1.5 mm long, on filaments 1 mm long. Capsule ellip- soid, 3-3.5 mm long, placentation central, the valves with strong septa toward base, thus actually placentation axile there. Seeds sev- eral, narrowly ellipsoid, ca. 1 mm long, pale amber with a conic white apiculus, the body longitudinally finely but distinctly ribbed. Distribution. High savanna, cerros Si- papo (Paraque) and Marahuaca, southern Territorio Federal Amazonas, Venezuela, in- frequent. Additional specimens examined. VENEZUELA. T. F. AMAZONAS: summit Cerro Marahuaca, 2,685 m, 15 Jan. 1981, Maguire 65635 (NY, VDB); cumbre del Cerro Paraque, 1,600 m, alto Orinoco, Phelps 20 (US); Cerro Sipapo, 17 Feb. 1981, Steyermark 124532 (NY, VDB, VEN). This species is very evidently closely re- lated to X. bicostata Maguire & Lyman B Smith of Cerro Huachamacari in Amazonas, which differs only in that the plants are (most- ly) lower, with smaller and fewer-flowered spikes, and narrower scapes and leaves. 65A. Xyris bicostata Maguire & Lyman B. Smith, Mem. New York Bot. Gard. 10: 30, fig. 14A-F. 1963. TYPE: Ven- ezuela. T. F. Amazonas: frequent on mossy rocky banks, dense woodlands along right fork of Caño de Dios, 1,800 m, Summit Camp, Cerro Huachamacari, 674 Annals of the Missouri Botanical Garden 1 am d \ imm FIGURE 64. Xyris pratensis (from the type) .—a. Habit sketch. —b. Leaf tip. —c. Leaf sheath—blade junction. — d. Leaf base. —e. Spike.—f. Fertile bract.—g. Lateral sepal.—h. Petal blade, stamen.—i. Staminode.—j. Stylar apex.—k. Capsule valve (two views, one on each side of seed sketch) .—l. Seed. Volume 75, Number 2 1988 Kral 675 Xyris Rio Cunucunuma, 13 Dec. 1960, B. Maguire, R. Cowan & J. J. Wurdack 30169 (holotype, NY; isotypes, NY, US, VEN). Figure 65. Cespitose slender perennial 2-3 dm high, the stems short, with base sometimes forming a short, stout rhizome. Leaves erect, 1-2 dm long; sheaths over 4 as long as blades, at very base red-brown ciliate, medially and dis- tally pilose-ciliate with red-brown hairs, ta- pering gradually to blade, eligulate, the sur- faces smooth or papillose-rugulose toward base; blades flat, narrowly linear, 0.8-1 mm wide, the apex abruptly conic-acute or terete- rounded, the margins finely ciliate to scabro- ciliate, the surfaces smooth and punctate, few- nerved, smooth or papillose toward base. Scape sheaths loose, as long as leaves, with long, strong blades. Scapes straight or slightly flex- uous, lineal, 0.8-1 mm wide, oval distally in cross section, strongly bicostate, costae pale- ciliate or scabrociliate. Spikes narrowly ob- ovoid, ellipsoid, aging obconic, 9-10 mm long, red-brown, few flowered; bracts subdecussate, the sterile bracts ca. 6, the lowest ones nar- rowly triangular, strongly keeled, grading into fertile bracts, these oblong or lance-ovate, 7.5-8 mm long, broadly acute or rounded, entire, carinate and keeled apically and with paler, linear-elliptic dorsal areas, often also with faint but evident arcuate lateral nerves. Lateral sepals free, subequilateral, ca. 6-7 mm long, lance-linear, narrowly acute, firm, pale red-brown, the narrow, firm keel entire. Petal blades narrowly obovate, ca. 6 mm long, yellow, the broadly acute apex irregularly dentate. Staminodia bibrachiate, the narrow flat branches penicillate pubescent toward and at apex. Anthers oblong, ca. 1.5 mm long, deeply bifid and auriculate, on filaments ca. 0.5 mm long. Capsule narrowly ellipsoid, ca. 4 mm long, the placentation appearing free- central, but capsule valves with shallow septa at base. Seeds numerous, ellipsoid, ca. 1 mm long, both ends acute, the body amber, finely longitudinally ribbed. Distribution. A high-tepui endemic, thus far found on Cerro Marahuaca and on Cerro de La Neblina, along the Brazil- Venezuela border, as well as at the type locality, Cerro Huachamamacari. The specimens from Ma- rahuaca and Neblina are uniformly different enough from the type to have been recognized as a species by Smith, who related them more to X. tatei Malme and named them X. til- letti. The material seems, however, to rep- resent a varietal extreme of X. bicostata, particularly in its prominent, sharp and prom- inent, scabrociliate pairs of scape costae, and in its general spike and sepal dimensions. Yet the leaves are narrower than in the type, also slightly thicker and blunter, and the spike scales are less prominently keeled (though evidently costate) and are less spreading at maturity. Thus, the following is proposed: 65B. Xyris bicostata var. tillettii (Ly- man B. Smith) Kral, stat. nov. Xyris tillettii Lyman B. Smith, Ernstia 9: 3- 4. 1982. TYPE: Venezuela. T. F. Ama- zonas: Cerro Marahuaca, al NE de, y casi contigua con, Cerro Duida, este im- mediatamente al N de La Esmeralda 3°10'N, 65*31'W, en el Rio Orinoco, ca. 2,750 m, 2 & 9 Feb. 1975, S. S. Tillett, P. Colvee et al. 752-332 (holotype, VEN; isotype, US). Additional specimens examined. VENEZUELA. T. F. AMAZONAS: plateau of Cerro de reat = Salto s Monos on trib. headwaters of Rio Igu 65°23'W, 2,555 m, 25 985, oes sner "17962 ( MO, VEN, VDB); Cerro l enie parte ral de la Meseta Sur-este, 10-12 Oct. 1983, Steye na 129 442 (VDB, VEN); Cerro de La Neblina, Planicie de Zuloaga, Rio Titirico, 2,300 m, 10-15 Oct. 1970, Steyermark 103845 (NY, M pes Valle d Titirico N of Pico Phelps in Cerro Neblina, c , bog, 1 Dec. 1984, Kral et al. 71920, Kral 71927 NY. VDB, VEN, and to be distrib- uted). 66. Xyris globosa Nilsson, Bih. Kongl. Svenska Vetensk.-Akad. Handl. 24(14): 57, pl. 3. 1892. TYPE: Venezuela. T. F. Amazonas: “‘prope Esmeralda ad flumen orinoco, Dec. 1853, R. Spruce 3244" (lectotype, S; possible isolectotype, NY). Figure 66. Solitary or cespitose, slender but stiffish, bulbous-based perennial 3-7 dm high, the 676 Annals of th Missouri Pon Garden FIGURE 65. Xyris bicostata (from the type) .—a. Habit sketch. E Leaf apex.—c. Leaf blade, midsector. — d. my Sd —e. Enlarged small sector of leaf blade edge.—f. Spike.—g. Fertile bract.—h. Lateral sepal.—i. Petal blade, stamen.—j. mimada enlarged sector af beard un ame apex.—k. Dehisced capsule showing placentation and two septa.—l. Seed. Volume 75, Number 2 1988 Kral 677 Xyris stems contracted. Leaves erect, outermost scalelike, castaneous, the principal ones with sheath % or less the length of the blade, the base orbicular-dilated, villous-ciliate, castane- ous, narrowing abruptly above, then tapering gradually into blade, this narrowly lineal, twisted, ca. 1 mm broad, slightly compressed with rounded-incrassate edges, the apex abruptly or narrowly conic, the surface smooth, longitudinally few ribbed, green. Scape sheaths much shorter than leaves, loosely convolute, multicostate, essentially bladeless. Scapes twisted, flexuous, terete, 0.7-1 mm thick, green, smooth, shallowly multiribbed or striate. Spikes broadly ovoid to subglobose or hemispherical, 5-10(-15) mm long, blunt, dull red-brown or tan, of many loosely and spirally imbricate bracts, these with pale-scarious borders, lacerate, white villous-ciliate, with distinct, narrow dor- sal areas. Sterile bracts numerous, the lowest much the smallest, grading gradually to fertile bracts, these oblong to obovate, ca. 5(- 7) mm long, broadly rounded, ecarinate or slightly carinate, the backs slightly convex-rounded. Lateral sepals free, subequilateral, linear-ob- lanceolate or oblong, ca. 5 mm long, obtuse, the sharp, alate keel pale-villous fimbriate or fimbriociliate with pale hairs from middle to apex. Petal blades narrowly obovate or ellip- tic, ca. 4.5-5.5 mm long, yellow, the apex broadly acute, the margins subentire. Stam- inodia bibrachiate, the broadly rectangular branches glabrous. Anthers narrowly oblong, deeply bifid and sagittate, ca. 1 mm long, on filaments ca. 0.7-1 mm long. Capsules deep brown, broadly obovoid, 2.5-3 mm long, the placentation basal-central, the valves lacking septa. Seeds numerous, oblong-ellipsoid, ca. 0.5-0.6 mm long, pale amber, apiculate, lon- gitudinally with fine but distinct and anasto- mosing ribs. Distribution. Mostly in low-elevation white-sand savanna in southwestern Vene- zuela (Territorio Federal Amazonas), partic- ularly along the middle and upper Orinoco and tributaries; possibly in Colombia. Selected specimens examined. VENEZUELA. T. AMAZONAS: La Esmeralda, Alta T Jan.-Feb. 1969, Farinas et al. 571 (VEN); Canaripo, bajo Rio Ventuari, 11 Oct. 1977, Huber 1064 (US); a unos 30 km al SE de la enim Orinoco- Ventuari, 30 Nov.-1 Dec. 1978, Huber & Tillett 2819 (US); SSE de Sta. Barbara del Orinoco, ca. 100 m, 4 Dec. 1978, Huber & Tillett 2852 (US) bajo Río Ventuari, ca. 10 km al E del Caserio de Carmelitas, 20 Feb. 1979, Huber 3241 (US); W del Cario Pimichin un poco al sur del caserio Pimichín, 24 Feb. 1979, Huber 3395 (US); also Cano Yagua, a unos 30 km al W de la Serrania El Tigre, 29 Feb. 1980, Huber 4846 (VEN); 15 km N de Esmeralda, 8 Mar. 1980, Huber 5059 (VEN); 20 km al NW Yavita, ca- beceras Cano Pimichin, 11 Feb. 1981, Huber & Medina 5948 (VDB, VEN); Savana el Venado, left bank Cano Pimichin us Pimichin, 23 Nov. 1953, Maguire et al. 36360 (NY, US); Pu ITO P Rio Ventuari, base . 1951, Maguire et al. 30992 15 Jan. 1951, Maguire et al. 30 cana, Savanna III, 31 Dec. 1950, Maguire et al. 30475 NY, US, VEN); Cerro Moriche, 800 m Maguire et al. 30897 (NY, US); E 125 m, 28 Dec. 1976, Steyermark & Redmond 112801 (MO, US, VDB, VEN); Cerro Yapacana, 8-9 Nov. 1979, Thomas & Rogers 2610 (NY). aaa Cerro de Auyantepui, Cardona 262 (US, VEN); morichal 2 km E of Rio Orinoco between Rio Horeda and Cerro Gavilan (Cerro Carighang), 100 m, 17 Dec. 1955, Wurdack & Monachino 39952 NY, US) ~ — This, fuzzy-spiked plant, common in the Orinoco savannas, appears closest to X. /ac- erata Pohl and X. lanulobractea Steyerm., with the lacerate bract of the former but having villose-ciliate borders, and with the villose-ciliate border of the latter, yet with leaf bases villose-ciliate and more bulbous. It appears to be rare in low-altitude savanna in western Estado Bolivar. The leaf blades, if at all compressed, are thickened so that they have no sharp edge, something hard to reflect in key construction. 67. Xyris arachnoidea Maguire & Ly- man B. Smith, Mem. New York Bot. Gard. 10: 28, fig. 12A-E. 1963. TYPE: Venezuela. T. F. Amazonas: flowers yel- low, occasional, Savanna III, northwest base of Cerro Yapacana, 150 m, alto Rio Orinoco, 17 Mar. 1953, B. Maguire & J. J. Wurdack 34570 (holotype, NY; isotype, US). Figure 67. Cespitose, hard- and fibrillose-based, sub- bulbous perennial 2.5-4 dm high, the stems ontracted. Leaves erect, 1-2 dm long; sheaths dull brown, the very base abruptly 678 Annals of the Missouri Botanical Garden icm xin. FIGURE 66. is globosa das 3395) .—a. d sketch. —b. Leaf tip.—c. 2 sheath— blade junction. — d. Leaf base.—e. Spikes, two types.—f. Fertile bract.—g. Lateral sepal.—h. ~ l, stamen.—i. Staminode.— Cap l. See j. Stylar apex. k. sule, one an removed ea ing placentation. —l. Volume 75, Number 2 Kral 1988 Xyris 679 FIGURE 67. upper blade.— c. Leaf sh h. Petal Made: ud stamen one septum (shaded), PA —Ek. eath— -blade junction. — d. dilated, castaneous, above tapering gradually into blades and white-villous- or lanate-mar- gined, the ligule concealed by cottony hairs; blades evenly linear, at least 5 times as long Xyris arachnoidea (a—e from type, h^ n des Kral 70710).—a. Habit sketch. —b. Leaf tip and base.—e. Spike.—f. Fertile bract.—g. Lateral sepal.— taminode, ae Pl ai enlarged staminodial hair. —j. Capsule, side view of as sheaths, flattened, ca. 1 mm wide, abruptly narrowed to a callused, narrowly rounded apex, the margins entire or scabrid toward base, rather thick, sometimes pale and con- 680 Annals of the Missouri Botanical Garden trasting, sometimes with 1 edge double, the surfaces smooth or papillose-scabrid or ru- gulose toward base, sometimes with a few strong, yellowish nerves, otherwise dull green or maroon. Scape sheaths much shorter than leaves, thin, keeled and open, apically with short, erect, stiff blades. Scapes flexuous and twisted, apically subterete or oval in cross section, ca. 1 mm thick, 2-4-costate, the costae yellowish, low, smooth. Spikes ovoid, drying obovoid, 7-8 mm long, blunt, dull brown, of several spirally imbricate, firm but loose bracts, the sterile bracts narrowly to broadly triangular, grading slightly larger into fertile bracts, these ovate, 4-5 mm long, acute with margins (sometimes also backs) white villose and with strong, red-brown triangular dorsal areas. Lateral sepals free, subequilat- eral, curvate-lanceolate, 4.5-5 mm long, acu- minate, lustrous red-brown, thin, the darker, wide, firm keel with a sparse fringe of villo- sulous or scalelike ascending hairs, villosulous at tip, aging subentire. Petal blades obovate, ca. 5 mm long, yellow, the narrowly rounded tip erose. Staminodia bibrachiate, the branch- es long, slender, long-penicillate from tip to base. Anthers oblong, ca. 2 mm long, ca. 2.5-3 mm long, the placentation free-central, the capsule valves septate. Seeds numerous, narrowly ellipsoid or cylindrical, ca. 1 mm long, apiculate, translucent pale brown with numerous, very fine, wavy, longitudinal ribs and some fine cross-lines. Distribution. Locally abundant in sandy low-elevation savanna along the upper Ori- noco, Territorio Federal Amazonas, Venezue- a. Additional specimens examined. VENEZUELA. T. F. AMAZONAS: SE bank m, Davidse et al. 17403 (MO), 17435 (MO, US); Cerro Yapacana, 22 May 1981, Guán- chez 1164 (TFAV, VDB); el pie occidental del Cerro Yapacana, 14-28 Feb. 1978, Huber 1617 (US); alrede- dores de Canaripo, 30 May 1978, Huber 1914 (US, VEN); Savanna III, Cerro Yapacana, 3 June 1978, Huber 2029 (VEN), 2040 (US); entre el medio Cano Yagua y el bosque al N del Cerro Cucurito, 18 Jan. 1979, Huber 3136 (US); Cano Caname, ca. 18 km arriba (al E) de la boca, 29 May 1979, Huber et al. 3749 (US); alto Cano agua, a unos 30 km al W de la Serrania El Tigre, 29 Feb. 1980, Huber 4846a (VEN); Savanna II, W base Cerro Yapacana, ca. 100 m, 10 Aug. 1983, Kral & Huber 7071 di K, L, MO, NY, ss US, VDB, wes: Cerro Yapacana en la sabana grande el Cano Cotua, 7 May 1970, ama y & Bunting 103243 (US, VEN). This species is fairly common in the sa- vannas around Cerro Yapacana and much resembles the more widespread and associated X. subglabrata Malme (X. garcia-barrigae Idr. & Smith). It differs, however, in having much more copious white villous or arachnoid pubescence on the bracts and in having dis- tinctly flattened (rather than terete) leaf blades. The flowers are open in the morning, closed by mid afternoon. 68. Xyris malmeana Lyman B. Smith, Bol. Inspet. Fed. Obras Contra Secas, Rio 10: 126. 1939. TYPE: Brazil, Para: open sandy soil 2 km S of Vigia, Drouet 2136 (holotype, GH). Figure 68. Xyris E. (Seub.) Griseb. sensu Griseb., Fl. Brit. W. Ind. 525 (Trinidad), not as to ed X. sava- nensis Miq. var. glabrata Seu Slender, solitary or cespitose, glabrous pe- rennial 4-6 dm high, the stems contracted. Leaves erect, 1-3 dm long; sheaths ciliate, pink to deep brown, the abruptly dilated base gradually narrowed, strongly ribbed and ecar- inate into the blade, this flattened, narrowly linear, twisted, 1.5-3 mm wide, gradually narrowed above middle to a narrowly acu- minate or subulate apex, the margins thin, smooth, the surfaces pale green, finely nerved. Scape sheaths much shorter than principal leaves, strongly ribbed, twisted, also keeled, with a cusplike blade apically. Scapes slen- derly linear, straight or flexuous, twisted, sub- terete at apex, ca. 0.5 mm thick, with 3 or more low but sharp and distinct, smooth or papillose costae. Spikes broadly ellipsoid or ovoid, 5-8 mm long, acute, pale red-brown or dull brown, of several spirally imbricate, convex, ecarinate, entire to lacerate bracts with distinct dorsal areas, the sterile bracts slightly smaller than the fertile bracts, grading into them, the fertile bracts 4-6 mm long, broadly elliptic to obovate, broadly or nar- rowly rounded apically, sometimes the inner ones slightly keeled, all with a short, narrow, Volume 75, Number 2 Kral 681 1988 Xyris lam 05 mm FIGURE 68. Xyris malmeana (Steyermark € Dunsterville 113285) .—a. Habit sketch.—b. Leaf tip. —c. Leaf sheath—blade junction.— d. Leaf base. —e. Spike. —f. Fertile bract.—g. Lateral sepal. —h. Petal blade, stamen.— i. Staminode. —j. Stylar apex.—k. Capsule.—l. Seed. 682 Annals of the Missouri Botanical Garden usually greenish dorsal area. Lateral sepals free, slightly inequilateral, thin, elliptic-cur- vate, ca. 5 mm long, acute, the broad, firm keel lacerate-ciliolate from middle to apex, or nearly entire. Petal blades obovate, ca. 4 mm long, yellow, the broadly rounded apex sub- entire. Staminodia bibrachiate, the narrow, flat branches distally penicillate-ciliate. An- thers ca. 1 mm long, lance-oblong, deeply bifid and sagittate, on filaments 0.3-0.4 mm long. Capsule obovoid, firm-valved, 3.5-4 mm long, the placentation basal, the valves with- out septa. Seeds numerous, broadly ellipsoid, ca. 0.5 mm long, 2-apiculate, deep lustrous amber, finely ribbed longitudinally. Distribution. Sandy, low- to high-ele- vation savanna, northern South America from Territorio Federal Amazonas, Venezuela, eastward to French Guiana and in contiguous northern Brazil in Amapá, Amazonas, Gua- pore, and Para; Trinidad. Selected specimens examined. BRAZIL. AMAPÁ: Cam 27.V1.1904, A. Ducke s.n. (MG, US). AMAZONAS: estrada maita-Lábrea, km 17, 10 June 1982, Teixeira et al. 1059 (INPA, NY, VDB). cuAPORE: Porto Velho, Cordeiro 1957, Egler 300 (US); Gu- rupa, campina da Gerenalda, Pires & Silva 4712 (US, VDB). FRENCH GUIANA. Route de Simonmary, environs in 70 km, 7 June 1957 (no collec tor name — U). GUYANA: Rupununi Savanna, ca. 350 ft., — as ug. 1936, Goodland 336 (US). SURINAM: Soesdyke, . 1977, Cooper 354 (U); Gros-savanna (prope km T5 im 309.8, Apr. 1959, Van Donselaar 697 (U); Zanderij I, Aug. 1914, Essed (U); Jodensavanne (fluv iname), Heyligers 235 (U); near Singri Lanti 15 km W of Zanderij, 24 July 1976, Jansma 13 (U); Tapfelberg, distr. Saramacca, Kramer & m 5.8 in third July 1953, Ea 4233 (GH, U); ra savan- na I, 3 Aug. 1944, Maguire 24203 (NY, Uy; Sipaliwini savana area on Braz. Frontier, 255 m, 4 Sep. Ee rs et al. 58 (U); Nat. Res. "Beiukheuvel. i Sep. 1 , Wildschut & Teunissen 11570 (U). za DAD: Arp "Savanna, 21 Apr. 1920, Britton et al. 1996 (GH, NY, US); Aripo-savanne, Manzanilla, 7 Feb. 1962, Hekking 1347 (U). VENEZUELA. T. F. AMAZONAS: 20-25 km W de San Juan de Manapiare, 8 Oct. 1979, Huber 449] (US); Esmeralda, 9 Oct. 1928, Luetzelburg 22498 (US). BoLivar: SE base Auyan-tepui, 24 Nov. 1982, Davidse & Huber 22565 (MO, VDB); 3 km S of El Pauji, 1,050 ns 19 Oct. 1985, Holst & Liesner 2630 ( VDB, VEN); sabanas al SW de Kamarata, 23 Nov. 1982, Huber et al. 6810 (MYF, VDB, VEN); Caserio de Ku- kenan, ESE de la punta SE del Churi-tepui, Huber 9768 (MYF, VDB); flats above Rio Yuruani at Salto Yuruani, 28 July 1983, Kral 70572 (MO, NY, US, VDB, VEN); 1.3 km N of Rio Yuruani Ferry, ca. 750 m, 29 July 1983, Kral dea (NY, US, VDB, VEN); above Ven. cda g 800 m, Dunsterville 113285 (NY, US, VEN); Auyan-tepui, 1,100 m, Dec. 1937-Jan. 1938, Tate 1316 (NY). This species most resembles X. lacerata Pohl ex Seub. However, Xyris malmeana is usually lower, is more slender, and is less bulbous based. Its scapes are more sharply costate, its dorsal areas are narrower, and its lateral sepals are less ciliate. 69. Xyris araracuare Maguire & Lyman mith, . New York Bot. Gard. 10: 34, fig. 20A- F. 1963. TYPE: Colom- bia. Amazonas: frequent in scrub savan- na, Araracuará, Rio Caquetá, 5 Sep. 1959, B. Maguire, C. K. Maguire & A. Fernandez 44132 (holotype, NY; iso- type, US). Figure 69. Slender, somewhat cespitose, bulbous-based perennial 2-3 dm high. Older leaves short, scalelike, covering **bulb." Main foliage leaves few, ascending, 10-20 cm long; sheaths cil- iate, much shorter than blades, abruptly flar- ing at base, castaneous, ligule absent, tapering gradually above into blade, this green, sub- terete, fluted, lineal, twisted, 0.7-0.8 mm wide, smooth or scaberulous along lateral ribs, abruptly bluntly conic. Scape sheaths shorter than leaves, loose, blade short, cusplike. Scape straight or flexuous, twisted, terete and slight- ly grooved or fluted, punctate distally, proxi- mally ribbed and scaberulous. Spike broadly obovoid, 7-8 mm long, blunt, many-flowered, pale dull brown, the bracts loosely spirally imbricate; sterile bracts mostly 6, broadly ovate or triangular, the lowest smallest, nar- rowly scarious-lacerate-bordered, grading gradually to fertile bracts, these ovate, ca. 5.5 mm long, the tips narrowly rounded, lac- erate-bordered, the backs rounded with dark- er, lance to linear dorsal areas; innermost bracts more navicular, more acute, subentire, Volume 75, Number 2 Kral 683 1988 Xyris Zam TM 69. Xyris araracuare from the type) .—a. Habit sketch.—b. Leaf apex.—c. Sector of lead mid- blade.— Leaf at junction of sheath.—e. Two views of leaf base. — f. Spike.—g. Lateral sepal.—h. Petal blade, stamen, oa style branches, enlarged beard hair.—i. Seed. 684 Annals of the Missouri Botanical Garden to 6 mm long. Lateral sepals free, subequi- lateral, elliptic-linear, ca. 4.5(-5.5) mm long, narrowly acute, pale brown except for darker, narrow, firm keel, this ciliate-scabrid from ca. middle to apex. Petal blades yellow, obovate, ca. 5 mm long, narrowly rounded apically, sparsely dentate. Staminodia bibrachiate, the branch apices sparsely long-penicillate. An- thers oblong, ca. 2 mm long, on filaments 0.5 mm long. Capsule narrowly obovoid, 2.5 mm long; placentation basal, funiculi long. Seeds numerous, narrowly lance-ovoid to ellipsoid, 0.6 mm long, translucent, longitudinally with several broad, wavy, distinct ribs. Distribution. Sandy scrub savanna, known only from the type collection. Evidently this species is very close to the widespread X. lacerata Pohl ex Seub., as the descriptions and illustrations show. It is best to include it now with that perspective until the entire genus is revised for South America. 70. Xyris lacerata Pohl ex Seub. in Mar- tius, Fl. Bras. 3(1): 26, pl. 26. 1855. TYPE: Brazil: “Goias, Pohl’ (isolecto- type, BR; phototype, US no. 5752). Fig- 0 ure 7 Tall, bulbous-based, solitary or cespitose perennial 4—6 dm high, the stems contracted, the outer leaves scalelike, castaneous. Prin- cipal leaves 1.5-3 dm long, erect or ascend- ing; sheaths less than !4 length of blades, at base abruptly dilated, ciliate, cochleariform, usually castaneous, papillose-rugulose; above paler, entire, gradually tapering to blade, pa- pillose-rugulose, eligulate; blades narrowly lin- ear, flattened, or rarely thickened, subterete, twisted, 1.5-3 mm wide, acute to narrowly acuminate, often subulate, the edges thin or cartilaginous-incrassate, smooth or papillose, the surfaces dull green, smooth or papillose- rugulose, multinerved. Scape sheaths much shorter than leaves, tight, proximally twisted and multicostate, opening and keeled distally, producing a cusplike blade. Scapes twisted, straight or flexuous, terete distally, 0.5-1 mm thick, striate, ecostate or with 1 low costa, smooth to papillose. Spikes globose, broadly ovoid or hemispherical, 0.5-1 cm long, dull brown, of very many spirally imbricate, con- vex-rounded, lacerate and scarious-edged bracts having distinct, green to red-brown dorsal areas; sterile bracts several, distinctly smaller than and grading into fertile bracts, these oblong to obovate, 6-7 mm long, broad- ly rounded, ecarinate, and with broadly lac- erate scarious borders (innermost fertile bracts narrower, frequently keeled). Lateral sepals free, slightly inequilateral, oblong-curvate, ca. 5.5-6 mm long, blunt, lustrous red-brown, the firm, narrow keel ciliolate from near base to tip. Petal blades obovate, 5.5-6 mm long, yellow, the rounded apex erose. Staminodia bibrachiate, the narrow flat branches long- penicillate from near base to tip. Anthers lance-oblong, ca. 2 mm long, shallowly bifid, deeply auriculate, on filaments ca. 0.5 mm long. Capsule narrowly ellipsoid, ca. 5 mm long, the placentation densely central, the valves lacking septa. Seeds numerous, broad- ly to narrowly ellipsoid or fusiform, 5-6 mm long, deep amber, 15-20-ribbed with much fainter crosslines. Distribution. Sandy or sandy-silty or rocky campos and savannas, eastern Colom- bia eastward across southern Venezuela and probably into Guyana but with no official rec- ord; to the south, from the Andean foothills eastward from northern Brazil south into Ar- : gentina. Selected specimens examined. Only northern rec- ords are listed. COLOMBIA. META: E of Rio Zanza, 2 km above jct. with Rio Cuejar, ca. 500 m, 22 Aug. 1950, Idrobo & S. Smith 1544 (F); Villa Visus 450 m, 26- 31 Aug. 1917, Pennell 1409 (NY); Savan. Boca de Monte, llanos de San Martin, id & Idrobo 1391 (GH, NY); 2 km E Rio Zanza, N end Cordillera Macarena, 22 Aug. 1950, Smith & Idrobo 1543 (GH); Llanos Orien- tales, La Macarena ree: Sur), Rio Guayabero, sabanas, 235-700 m, Jan.- Mar. 1959, Garcia-Barriga & Meji 17078 (NY). SANTANDER: Mesa Killip & A. C. Smith 15139, 15190, 15280 (GH, NY). vAUPÉS: Mesa La Lindosa, ca. 15-20 km sur de San Jose del Guaviare, 13-15 Dec. 1950, Idrobo & Schultes 674 GH, U); lower Rio Pa rucchi 2012 (GH). VENEZUELA. T. F. meralda, 15 July 1951, Croizat 115 (NY E Puerto Ayacucho, 30 Sep. 1960, Foldats 3539 (NY, VEN); Aeropuerto de Puerto Ayacucho, 24 Aug. 1977, ~ Volume 75, Number 2 Kral 685 1988 Xyris k NA AS FIGURE 70. Xyris lacerata — et al. 11152).—a. Habit sketch.—b. Leaf apex.—c. Leaf at gas blade pet —d. Leaf base.—e. Spike.—f. Fertile bract.—g. Lateral sepal.—h. Petal blade, stamen.— Staminode.—j. Stylar apex.—k. Seed. 686 Annals of the Missouri Botanical Garden Huber 1000 (US, VEN) Cerro Morrocoy al Sur Serrania Colmena al Norte, 16 Oct. 1977, Huber 1206 VEN); cuenca del Río Manapiare, 150 m, 15 Aug. 1978, Huber 2239 (US); alto Río Ventuari, sabana del Oso, 17 Aug Ai Huber 2280 (US); iy de Rincones de Chae , 9 km NE de Galipero, ca. 80 m, 17 Aug. 1979, "Huber 4179 (US, VDB, VEN): sabanas de Santa Barbara, 19 July 1980 (VEN); ca. 30 km N of Puerto Ayacucho, Kral et al. 70732 (NY, US, VDB, VEN); Puerto Ayacucho, S side, 22 Nov. 1984, Kral & Boom 7178 zd. "VDB, VEN); Esmeralda Savanna, ca. Aug. . 1929, Tate 265 (F, NY). APURE: Estero la 1979, Garofalo 378 (US); Alto Apure, U, VEN); id ate de Puerto Paez, sabana, 19 Mar. 1973, Ramia & Montes 5159 (US) quist 9d El Rosero, July 1976, Tejos s.n. (US). BOLÍVAR: vic. are, 6 km from Maniapure toward Cai- cara, Boom & "Grillo 6290 (NY, VDB); alrededores de Puerto Nuevo, 14 Nov. 1982, Guánchez & Huber 2469 (TFAV, VDB); 2 km S of Ciudad Piar, 300 m, 18 Oct. 953, Maguire et al. 35834 (GH, NY, US); 14 km SW of Caicara del Orinoco, 2 Sep. 1985, Steyermark et al. 131112 (MO, VDB, VEN); E of Rio Parguaza, 125 km N of Alcabala of Puerto Ayacucho, 8-11 Sep. 1985, Steyermark et al. 131752 (MO, VDB, VEN); morichal just N of Río Chiguirete, 420 m, 11 Oct. 1954, Wurdack & Guppy 20 (NY, US); 2 km E of Rio Orinoco between Rio Horeda and Cerro Gavilan, 17 Dec. 1 & Monachino 39951 (NY, US cional Aguaro-Guáriquito: Morichal Charcote, Dec. 1981, Delascio et al. 11152 (MO, VDB, VEN). This common plant of low- to medium- elevation savanna bears a considerable resem- blance to X. tortula C. Martius of the planalto of Brazil and may indeed intergrade with it. I have seen the phototype of X. fallax, the type supposedly still at M, and am surprised to see that it is comprised of two good spec- imens of X. lacerata. 71. Xyris oblata Kral & Lyman B. Smith, sp. nov. TYPE: Venezuela. T. F. Ama- zonas: en la sombrede arboles pequenos; fls. amarillas; en las sabanetas periodi- camente enegadas cerca de la margem del Cano Temi, Yavita, 128 m, 31 Jan. 1942, Llewelyn Williams 14121 (ho- lotype, F; isotypes, F, US). Figure 71. Planta solitaria vel caespitosa, tenuis, annua aut pe- rennis, 3-4 dm alta, caulibus contractis. Folia principalia vulgo flabellate ni i 1-2 dm onga; vaginae brun neociliatae, kaga nas ca. 2-plo breviora, brunn ia vel atroporphyr sursum in laminas gradatim contractae, Kika. laminae eladiatlincares, complan ae, l- cem gradatim oe ad vel erectae, acutae, leviter incrassatae; margines tenues, scabriduli; sages leviter stratonervosae, marroninae, dense scabrido-rugosae. Va- ginae scaporum foliis breviora, proxime contortae, cari- natae et costatae, carinatis scabridis, ad apicem apertae, laminis dud ciliatis, erectis. Scapi recti vel flexuosi, torti, ad apicem teretes vel in sectione transversali ovales, pillati, LT plani vel leviter striati. Spicae oblata. ca. pd e mea interdum proliferatae, brun- neolae, bracteis numerosis integris, spiraliter imbricatis, area delo dennon bracteae steriles plures, fertilibus m longa, obtusa, pallide p aont ala carinali valde curvata, a medio i api cem s ato lacerata. Lamina petalorum obovat un longs (est.), luteola. Staminodia bibrachiata, brachiis e lon- ira pum Antherae lanceolatae, 1.5 mm longae; filiis longis. Capsula ellipsoidea vel subrotunda, ca. 2.5 mm longa, planoconvexa, placentae axiales. Se- mina numerosa, ellipsoidea, ca. 0.8 mm longa, atrosuc- cinea, longitudine valde et anguste 28- 30-costatae. inaequilatera, ca. 4 mm Solitary or cespitose, slender annual or short-lived perennial 3-4 dm high, the stems contracted. Leaves mostly spreading flabel- lately, 1-2 dm long; sheaths soft, ca. 1⁄2 as long as blades, carinate, brown ciliate, dull brown to deep red-brown, rugoscabrid, nar- rowed gradually to blades and eligulate; blades gladiate-linear, flat, 1-3.5 mm wide, tapering gradually from midblade to an erect or slightly incurved, bluntly acute, slightly thickened tip; margins thin, scabridulous; surfaces striate- ribbed, maroon, densely scabridulous-rugose. Scape sheath shorter than leaves, proximally twisted, carinate and costate, the keel scabrid, open at apex, with short, ciliate, erect blade. Scape straight or flexuous, twisted, distally terete or oval in cross section, papillate, ecos- tate, level or striate. Spikes depressed-glo- bose, ca. 5 mm long, 6-7 mm broad, some- times proliferous, dull brown, of many firm, entire, spirally imbricate bracts with distinct dorsal areas; sterile bracts several, the lowest narrower and slightly shorter than the fertile bracts, scabrid-ciliate, grading into fertile bracts; these obovate, ca. 5 mm long, broadly rounded, subentire, the backs shallowly con- vex, ecarinate, papillate, the dorsal area oval to elliptic, pale punctate. Lateral sepals free, inequilateral, ca. 4 mm long, obtuse, tan, the thin, strongly curvate keel fimbriolacerate from middle to apex. Petal blades obovate, ca. 4.5 mm long (estimate), yellow. Stami- nodia bibrachiate, the branches densely long Volume 75, Number 2 1988 Kral 687 Xyris idm Ñ S> NL- seed on funicle; upper left, detail. hee VTP Ute Jg yj MA y Y P Me n , f Y FIGURE 71. Xyris oblata (from the type) .—a. Habit sketch. —b. Leaf apex.—c. Leaf at midblade.—d. Lea base.—e. Spike.—f. Fertile bract.—g. Lateral sepal.—h. Petal blade, stamen, staminode, beard hair, stylar apex.—i. Dehisced capsule, showing placentae, two valves, oblique view showing septum.—j. Bottom center, 688 Annals of th Missouri Boni Garden penicillate. Anthers lanceolate, ca. 1.5 mm ong, planoconvex, the placentation axile at base. Seeds ellipsoid, ca. 0.8 mm long, dark amber, with 28-30 strong, narrow, irregular longitudinal ribs. Distribution. So far known only from the type collection. Material of this distinctive species in gen- eral character of leaves and aspect of plant much resembles some X. savanensis. It dif- fers markedly in that the leaf sheaths are conspicuously ciliate, the spikes distinctively oblate, and the ornamentation of lateral sepals different. Also, the staminodial branches are densely long-penicillate, a character lacking in X. savanensis. 72. Xyris tenella Kunth, Enum. Pl. 4: 9. 1843. TYPE: Brazil. Sao Paulo: Sul do Estado, Sellow s.n. ue B; pho- totype, US). Figure Synonyms given below are only for Ven- ezuelan material. Xyris steyermarkii Maguire & Lyman B. Smith, Mem. New York Bot. Gard. 10: 30, fig. 15A-E. 1963. TYPE: Caer Bolivar: Cae Mojado between base of upper falls and drop to escarpment, 1,985- 1,910 m, summit, Chimantá Massif, Torono-tepui, p: Feb. 1955, J. A. Steyermark & J. J 096 (holotype, NY; isotype, US). Xyris ibus Maguire & Lyman B. Smith, Mem. New York Bot. Gard. 10: 36, fig. 19A-F. 1963. TYPE: Venezuela. T. F. Amazonas: small savanna and along left fork of Cano Yutaje, 1,250 m, Serrania Yutaje, Rio Manapiare, 12 Feb. 1953, B. Maguire & C. K. Maguire 35206 (holotype, NY; isotype, US). Wurdack Cespitose, delicate, low smooth annual, 0.5-2 dm high, the stems contracted. Leaves mostly erect, 2-8 cm long; sheaths soft, scar- ious, long-ciliate, stramineous or pale brown, less than 1⁄4 blade length, finely costate, evenly narrowed from dilated base to blade, eligulate; blades linear-filiform, slightly twisted, some- what flattened, ca. 0.5 mm wide, narrowed at apex to an incurved-apiculate, callused tip; margins entire, smooth, or ciliolate; surfaces very finely nerved, green to maroon. Scape sheaths shorter than leaves, at base tubular, multicostate, brown, distally open and slightly dilated, keeled, with short, erect blades. Scapes filiform, terete, .4 mm thick, finely striate distally, sometimes unicostate with a low costa, or low-costate, costae papillate to finely scabrid. Spikes ellipsoid or lance-ovoid, drying narrowly obovoid, ca. 5-6 mm long, of a few thin, brown to tan bracts with prom- inent but narrow or streaklike dorsal areas; sterile bracts few, lance-ovate or oblong, na- vicular, smaller than the fertile bracts, grad- ing into them; fertile bracts few, lanceolate, rounded-folded (navicular), ca. 5-5.5 mm ong, acute, subentire or lacerate, the keel darker brown. Petal blades broadly obovate, 3 mm long, yellow, the broadly or narrowly rounded apex with sinuate-erose margin. Staminodia lacking or blade broadly bilobed, beardless (in the Guayanas). Anthers lance- oblong, ca. 1 mm long, deeply bifid and sag- ittate, on broad filaments 0.6-0.7 mm long. Capsule narrowly ovoid, ca. 3 mm long, acu- minate, the placentation basal. Seeds ovoid to ellipsoid, ca. 0.5 mm long, deep amber, finely lined longitudinally. Distribution. Low- to high-elevation sa- vanna from southeastern Venezuela eastward to French Guiana; southward through the planalto of Brazil, thence south into Para- guay. dig ss a Rain examined. oe are — om Ven dies material only. VENE AMAZONAS: ficia d Liesner DB) Cerro de Mahu, ame- 1985, Liesner 17711 (MO, VDB); Cerro uaca, ‘la meseta Sur-Este, 1,560 m, 13-14 Oct 1983, Steyermark 129599 (VDB, VEN). BOLÍVAR: ves yan-tepui, ca. 1,900 m, Apr. 1956, Foldats 2642 (VEN); A El Venado, ca. 20 km E de Canaima, Huber et al. 8257 (NY) Auyan-tepui, sector SSE, 10 Dec. 1983, Huber & Medina 8522; Serrania Guanay, oriental, 20-28 Oct. 1985, Huber 10966 (MYF, VDB, VEN); La Escalera, ca. 1,200 m, 22 July 1983, Kral 70308 (VDB, VEN); top of La Escalera, ca. 1,200 m, 24 July 1983, Kral 70327 (F, K, MO, NY, U, US, VDB, VEN); just S of La Escalera ca. 1,200 m, 24 July 1983, Kral 70372 (VDB, VEN); 6.5 km N of Pioneer Monu- ment by Ven. 10, ca. 1,200 m, 24 July 1983, Kral 70396 (US, VDB, VEN); Salto Yuruani, 1,000 m, 29 July 1983, Kral eel rie VEN); Auyantepui, Panier & Schwabe 5/9 (VEN); Chimanta Massif, middle falls below Summit Camp, 1,925 m, 5 Feb. 195 Venamo, 1,300 m, 25 A 770 (NY, US); Chimantá Massif. Torono-tepui, summit, Volume 75, Number 2 Kral 689 1988 Xyris FIGURE 72. Xyris tenella (Kral 70372) .—a. Habit sketch. —b. Leaf apex.—c. Leaf blade—sheath junction. — d. Leaf base. —e. Spike. —f. Lateral sepal.—g. Petal blade, stamen.—h. Staminode. —i. Stylar apex. —j. Capsule, two valves removed, showing basal placentation.—k. Seed. 690 Annals of the Missouri Botanical Garden Feb. 1955, Steyermark & Wurdack 1096 (Type of X. steyermarkii, F, NY, US, VDB, VEN); Chimantá Massif, Torono-tepui South Cano, summit, 1,955 m, 23 Feb. 1955, Steyermark & Wurdack 1122 (F, NY, US); Pta- ritepui, entre “Drizzly Camp” y “Second Wall," 1,660- o Guaiquinima, sector suroeste- central, 26 May 1978, Steyermark et al. 117480; Cerro Marutani, 1,200 m, 2 Jan. 1981, Steyermark 123895 (NY, US, VDB); parte superior del plato de Auyantepui, 2,300 m, Apr. 1956, Vareschi & Foldats 4958 (NY, VEN). Some insight into the X. tenella situation was given me by examination of a Tate spec- imen (810) from the gorge of Cano Negro, Savanna Hills, Mt. Duida. Malme, working when that part of Venezuela was hardly ex- plored, did see this specimen, annotating it at first X. tenella Kunth var. subtenella, a taxon he later (Ark. Bot. 13(3): 44. 1913) changed to a form of the type variety. This led me to examine a rather large series of materials from Brazil and Paraguay. From these it appears that if transplants of X. te- nella from Paraná or Mato Grosso were put in suitable habitat in Venezuela, such would be indistinguishable from forms named X. steyermarkii and X. yutajensis. The variety from French Guiana (var. leprieurii Malme, “Cajenne legit Le Prieur” isolectotype at L) is taller than the Venezuelan material, has flat, ciliolate-edged leaf blades, and narrowly ovold spikes. The dorsal area character in this species varies troublesomely. Sometimes it appears only on some bracts. In Brazil are specimens named other species that, if given the pencil- thin dorsal area, would again be placeable in X. tenella. The only constant, apart from the low and annual or short-lived perennial habit, seems to be the long brush of ciliae on the leaf-sheath edges. The dorsal area when pres- ent is very narrow, a ready means of distin- guishing this from extremes of X. guianensis, which always has a broad dorsal area. 43. Xyris byssacea Kral, sp. nov. TYPE: Venezuela. T. F. Amazonas: Dpto. Rio Negro, Valle de Titirico N of Pico Phelps in Cerro Neblina, ca. 0?56'N, 65°58' W, ca. 2,200 m; peat bog interspersed with shrub and low rocky but wet ridges, flow- ers opening in the a.M., 1 Dec. 1984, R. Kral et al. 71926 (holotype, VEN; isotypes, MO, NY, US, VDB). Figure 73. Planta perennis, densicaespitosa, glabra, delicatula. Caules breves; ieri graciles, Folia principalia arcte disticha, sube m big vaginis scaporu longiora; ae atae, laminis multi-plo longiores, datim tae, in laminas gr m decrescentes, eligu- latae; 1 compressae, - tae, leviter tortae, ferrugineae, nitidae, antis or in apicem gra atim decrescentes, ad apice latae, plerum m m curto- Fiala area dorsali e on ice bracteae fertiles 2-4, oblongae, 4.5-5 mm longae, sub- d anguste Neue a medio ad apicem a dorsali inseri. Sepala lateralia libera, in- mm longae, integrae. Staminodia bibrachiata, brachiis anguste triangulatis, longipenicillatis. Antherae lanceolatae, va- ose bifidae et sagittatae, ca. 1.5 mm longae, filiis ca. 0.5 mm longis. Capsula matura anguste ellipsoidea, ca. m longa; placenta basalis, funiculis a uses. Semina ellipsoidea vel cylindrica, ca. 1 mm longa, translucentia, pallide ferrugineobrunneola, subtiliter inus spirali- ter lineata. Densely cespitose, smooth, delicate peren- nial. Stems short; roots slender. Principal leaves tightly distichous, suberect, 8-15 cm , long, longer than the scape sheaths; sheaths elongate, many times longer than the blades, cobwebby-ciliate, pale brown, shining, grad- ually dilated toward base, gradually narrowing into blades, eligulate; blades flattened, sub- capillary, 0.3-0.4 mm wide, slightly twisted, entire, red-brown, lustrous, gradually taper- ing, conic-subulate at apex, often setaceous. Scapes slender, 20-30 cm long, brown, terete toward apex, ecostate to slightly striate, ca. 0.5 mm thick. Spikes few-flowered, ellipsoid, later short-cylindric, pale brown, ca. long; bracts spreading slightly, decussate, en- tire, the dorsal areas linear, the lowest 4 bracts sterile, the lowest pair lanceolate, ca. 3-4 mm long, acute, strongly carinate, the inner pair ovate, ca. 4 mm long, with a short carina Volume 75, Number 2 1988 ra Xyris FicunE 73. sheath junction. —d. Leaf base.— at apex, the dorsal area reduced; fertile bracts 2-4, oblong, 4.5-5 mm long, subcondupli- cate, narrowly rounded, carinate from middle to apex, the dorsal area linear. Lateral sepals free, inequilateral, narrowly elliptic, ca. 4.5- Xyris byssacea (from an aed —a. Habit sketch. —b. Leaf apex.—c. Sector of leaf at blade— Spi —f. Two views eum bract.—g. Lateral sepal.—h. Petal blade, stamen.—i. Staminode, stylar apex.—j. ee le.—k. See 5 mm long, acute, slightly curvate, the carinal keel narrow but thick, entire. Petal blades narrowly obovate, yellow, ca. 4 mm long, entire. Staminodia bibrachiate, the branches narrowly triangular, long-penicillate. Anthers 691 Annals of the Missouri Botanical Garden lanceolate, shallowly bifid and sagittate, ca. 1.5 mm long, on filaments ca. 0.5 mm long. Capsule narrowly ellipsoid, ca. 3 mm long; placenta basal, the funicles elongate. Seeds ellipsoid to cylindric, ca. 1 mm long, trans- lucent, pale red-brown, finely spirally lined longitudinally. Distribution. Known only from the type area. This species combines the slender habit of X. delicatula and X. carinata with the smooth, glassy-scaped character of X. setig- era; yet the foliage is utterly smooth and the filiform, evenly tapering leaf blades entire. Notable are the spreading arachnoid-ciliate hairs of the sheath margins. 74. Xyris cryptantha Maguire & Lyman B. Smith, Mem. New York Bot. Gard. 10: 16, fig. 2A-E. 1963. TYPE: Vene- zuela. T. F. Amazonas: Ríos Pacimoni- Yatua—Casiquiare, fls. yellow, locally fre- quent, Sabana Pacimoni on rt. bank of Rio Pacimoni, above mouth 50 km, 100 m, 2 Oct. 1957, B. Maguire, J. J. Wur- dack & C. K. Maguire 41677 (holotype, NY; isotypes, US, NY, VEN). Figure 74. Mostly low, slender, tufted, fibrillose-based and short-lived perennial or annual 0.6- dm high, the stems contracted. Leaves erect, 0.5-2 dm long; sheaths floccose-ciliate or pi- lose-ciliate, less than 12 as long as blades, dilated at very base, broad, castaneous, ta- pering abruptly, then gradually to blades, elig- ulate, the blades filiform, subterete or slightly compressed but not sharp-edged, 0.7-1 mm thick, smooth and pale-puncticulate, conic- subulate or incurved-conic at apex. Scape sheaths much shorter than leaves, loosely tu- bular, with short, erect blades. Scapes linear- filiform, straight or flexuous, slightly twisted, smooth, maroon, pale-puncticulate (stomata), ca. 0.5 mm thick, distally terete, ecostate. Spikes obovoid, 5-9 mm long, reddish brown, few-flowered, the bracts distichous and loose, with large, distinct, pale brown dorsal areas, the sterile bracts ca. 4, the lowest pair dis- tinctly longer than the fertile bracts, narrowly oblong-pandurate, with carinate, cucullate, acute tips arching over spike tip, often there connivent; fertile bracts oblong, 5.5-6 mm long, navicular, erect, the margins entire, scarious, lacerate, the tips rounded. Lateral sepals free, subequilateral, lance-linear, ca. 5.5 mm long, narrowly acute to acuminate, lustrous red-brown, the narrow, firm keel lac- erate or friable-fimbriate above middle, later distantly ciliolate. Petal blades broadly ob- ovate, ca. 5.5 mm long, yellow, the broadly rounded apex erose. Staminodia bibrachiate, the branches densely long-penicillate. Anthers ca. 1.5 mm long, deeply bifid at apex, auricu- late at base, on filaments 1 mm long. Capsule narrowly ellipsoid-cylindric, 2.5-3 mm long, trilocular, the placentation thus axile, ap- pearing free-central as valves detach. Seeds numerous on short funicles, ovoid or ellipsoid, ca. 0.5 mm long, pale red-brown, apiculate, longitudinally prominently but finely ribbed. Distribution. Low, sandy savanna, lo- cally abundant, southeastern Colombia, Ter- ritorio Federal Amazonas in Venezuela, and a disjunction in the Serra Araca, Amazonas, Brazil. Additional specimens examined. BRAZIL. AMAZONAS: plateau of northern massif of Serra Aracá, 0*51-57'N, 63?21-22'W, 1,200 m, S side of North Mt., open plateau savanna, 11 Feb. 1984, G. T. Prance et al. 28981 INPA, NY). COLOMBIA. AMAZONAS: Puerto Huesito, sa- banas del Alto de La Cruz; entre el Cano Chaquita (afluente del Atabapo) y en Cano Gente, 18-20 Aug. 1975, H. García-Barriga 20890 (CH, US, VDB). VENEZUELA. T F. AMAZONAS: la margen izquierda del Rio Sipapo a unos 4-6 km aguas abajo de la boca del Río Guayapo, 8 Oct. 1983, Guánchez & Varadarajan 2524, 2570 (TFAV, ga caño “Cabeza de Manteco” a 3-4 km de la boca, 00 m, al norte del medio Rio Autana, 12 Nov. 1984, a & Melgueiro hi (TFAV, VDB); sabanas en 1 ~ los alrededores de Guarinuma, 95 m, 25 Aug. 8, Huber 2656 (US, VDB): 2 km al W de San Antonio del E Huber & Tillett 5423 y 5 km al Sd aguna Yagua, 22 July 1980, Huber & Tillett 5475 (VDB. VEN); ribera izquier- da (Sur) bajo Río Siapa, poco distante he e desembo- cadura en el Río Casiquiare, 125 m, 81, Huber & Medina 5799 (VDB, Hbi Sabana cae on rt. bank of Rio Pacimoni, 50 bove mouth, 2 Oct. 1957, Maguire & Wurdack ns (NY, US); Sabana El Ven- ado on left bank of Cano Pimichin above Pimichin, 2 July 1959, Wurdack & Adderley 43294 (NY, U, US, VEN). This distinctive xyrid, in its low habit, slen- der foliage, small seeds, and general bract Volume 75, Number 2 Kral 693 88 Xyris WS mm = = E 0 => = A FIGURE 74. Xyris cryptantha (Huber & Tillett 5475).—a. Habit sketch.—b. Leaf tip. —c. Leaf blade-sheath junction. —d. Leaf base.—e. Spike. —f. Lateral sepal.—g. Petal blade, stamen.—h. Stylar apex, staminode. — i. Capsule; at left with one valve removed; a valve at right.—j. Seed. 694 Annals of the Missouri Botanical Garden characters definitely lies near X. oxylepis, a species with which it is associated in the white- sand savannas. 75. Xyris oxylepis Idrobo Lyman B. Smith, basin 6(29): 204, fig. 7a-c. 1954. TYPE: Colombia. S “Rio RA cerro Yapoboda, sabanas FM piedras areniscas," ca. 450 m, 5 1951, R. E. Schultes & I. rid 14234-B (holotype, COL; isotype, US). Figure 75A, B. Xyris tae hat Maguire & Lyman B. Smith, Mem. New . Gard. 10: 24, fig. 8A-E. 1963. TYPE a T. F. Anais Sabana El Venado on left bank of Cano Pimichin, above Pimichin, 140 m, 10 Oct. 1957, Rio Guainia, Maguire et al. 41815 (holotype, NY; isotypes, GH, US). Slender, low, wiry, cespitose perennial, the bases cloaked in fibrillose-chaffy remnants of old leaves, the stems contracted, the plants 1-3 dm high. Leaves mostly erect, quill-like, sometimes exceeding the scapes; sheaths less than Y as long as blades, at base dull brown to castaneous, dilated, above brownish or red- brown, the edges villous-ciliate with brown hairs below, tapering gradually to blades and white-cottony-ciliate above, at junction with blade producing a broad, blunt ligule or es- sentially eligulate; blades subterete, usually with 4 or more rounded ribs and shallow sulci, sometimes with a strong ventral sulcus, nar- rowly linear, 0.5-0.7(-1) mm thick, apically conic-acute or narrowly triangular, smooth, the surfaces smooth or papillate, often white puncticulate. Scape sheaths much shorter than leaves, tubular and multicostate proximally, distally producing a short blade similar to leaves. Scapes twisted and flexuous, about as broad as leaves, terete distally, smooth, prom- inently multicostate, the ribs low, white punc- ticulate. Spikes narrowly ellipsoid or fusiform, aging cylindric, 7-10 mm long, acute, of a few, loosely imbricate, subdecussate, ecari- nate bracts with strong dorsal areas; lower 4 bracts sterile, lance-ovate, 5-8 mm long, 1⁄2 or more as long as spike, broadly acute or narrowly rounded apically, scarious-edged, with large, elliptic, medially 1-nerved dorsal areas; fertile bracts 2-3, usually slightly lon- ger and often with slightly narrower outline than sterile bracts, ca. 8 mm long, with similar dorsal areas. Lateral sepals free, equilateral, lance-linear to elliptic-linear, 6-7.5 mm long, acute, the narrow, firm keel entire or with scattered cilia above middle. Petal blades ob- ovate, 5-6 mm long, yellow, the broadly or narrowly rounded apex erose or dentate-lac- erate. Staminodia bibrachiate, the branches long-penicillate. Anthers oblong, ca. 2-2.5 mm long, nearly 4 bifid, deeply sagittate, on filaments 1.5-2 mm long. Capsule ellipsoid, ca. 3 mm long; valves with strong septa from base to tip, the placentation appearing axile. Seeds short-cylindrical, 0.7-0.8 mm long, pale amber, apiculate, finely anastomosing-ribbed longitudinally. Distribution. | Low-elevation sandy sa- vanna, southeastern Colombia (rare) eastward into southwestern Venezuela, where locally abundant in the Orinoco savannas. Additional specimens examined. VENEZUELA. T. F. AMAZONAS: Santa Cruz, margen del Rio Atabapo, 4 Sep. 1960, Foldats 3675 (US); Cerro Yapacana, Savanna III, 3 June 1978, Huber 2036 (US); sabanas al SE de Car- melitas, 26 Aug. 1978, Huber 2668 (US); Cano Yagua, km rio arriba desde la boca, 6 Dec. 1978, Huber & Tillett 2906 (US); savanna III, 20-25 km al W de San Juan de Manapiare, Huber 4588 (US); Estación de Pis- cicultura de Puerto Ayacucho, 75 m, Maas & Hu id 5164 (U, VDB); 30 km al N de Puerto Ayacucho, al NE de Galipero, 80 m, Huber 5746 (US); same locality, 7 Nov. 1979, Huber 4695 (US, VDB); 1 km N de la Laguna Yagua, 27 July 1980, Huber & Tillett 5574 (US, VDB, VEN); 5 km NE de Galipero, 4 Nov. 1980, “aqa 5746 (US); Savanna I, W base Cerro Yapacana, Aug. 1983, Kral & Huber 70691 (F, K, L, MO, NY, EE TFAV, US, VDB, VEN, and others); savannas Il and III, 10 Aug. 1983, Kral & Huber 70711(US, NY, VDB, VEN); left bank of Cano Pimichin above Pimichin, 140 m, 23 Nov. 1953, Maguire & Wurdack 36356 (NY, Yapacana savannas, 16 Sep. 1957, Maguire et al. 41538 (NY, US); Cerro Yapacana, base, 8-9 Nov. 1979, Thomas & Rogers 2590a (NY). The overlap of this Colombian rarity with the Venezuelan ecological and morphological equivalent, X. fusiformis, necessitates their combination even though ripe seeds are un- described from Colombian material. In the savannas along the upper Orinoco it is a com- mon sight, its pale yellow flowers unfolding in the afternoon. To illustrate reason for the Volume 75, Number 2 Kral 695 1988 Xyris tam FIGURE 75A. Xyris oxylepis ur the isotype) .—a. Habit sketch.—b. Leaf apex.—c. Leaf jc e Bu sq junction. —d. Leaf base.—e. Spike. —f. Lateral sepal. —g. Petal blade, stamen.—h. Staminode.— i. Styla Annals of the Missouri Botanical Garden > Yan rapides ÁN ESAE Aem € tee de ON A P: jue nud del h ARAB e Po, est: mm, VEA . HAS ag VA, i <. `, - i c va. za a ei — Jmm FIGURE 75B. Xyris oxylepis (from type of X. fusiformis and from Kral & Huber 70691) .—a. Habit sketch. — b. Leaf tip. —c. Leaf blade-sheath junction.—d. Leaf base. —e. Spike. —f. Lateral sepal.—g. Petal, stamen.— h. Stylar apex. —i. Staminode.—j. Capsule; at left with valve removed; at right a valve showing the septum.— k. Seed. Volume 75, Number 2 1988 Kral 697 Xyris combination of Colombian and Venezuelan material, Figures 75A (from the type of X. oxylepis) and 75B (from material of X. fu- siformis) are presented. 76. Xyris wurdackii Maguire & Lyman . Smith, Mem. New York Bot. Gard. 10: 24-25, fig. 9A-E. 1963. TYPE: Ven- ezuela. T. F. Amazonas: Yavita-Pimi- chin trail near Pimichin, 140 m, Rio Guainia, 22 Nov. 1953, B. Maguire, J. J. Wurdack & G. S. Bunting 36366A (holotype, NY; isotype, US). Figure 76A. KEY TO THE SUBSPECIES OF XYR/S WURDACKII la. M d by principal leaves; leaf blades and scapes ns Raa white punctic- ulate; fertile bracts pilose-ciliate at tips 76A. X. wurdackii subsp. wurdackii lb. Most scapes overtopping principal leaves; leaf blades and scapes not noticeably white punc- ticulate; fertile bracts completely smooth 76B. X. wurdackii subsp. caquetensis 76A. Xyris wurdackii subsp. wurdack- ii. Cespitose or solitary, perennial, sub-bul- bous and fibrillose at base, the stems con- tracted. Leaves erect or ascending, few, the larger ones 1-3 dm long, twisted and flex- uous, mostly overtopping scapes; sheaths less than Y as long as blades, dilated at very base, dark brown or castaneous, gradually narrow- ing above, long-villous-ciliate, at bristly apex narrowing abruptly into blade, producing a broad, short, rounded-tipped ligule; blades lin- ear, smooth, spirally fluted, terete, 1-1.5 mm thick, ventrally deeply sulcate and with nu- merous spiral lines of punctae, tapering short- ly above middle to narrowly conic apices. Scape sheaths very short, mostly hidden in leaf sheaths, short-bladed. Scapes flexuous and twisted, 6-20 cm high, fluted and punc- tate as in leaf blades, proximally somewhat flattened, distally subterete with several low ridges, ca. as thick as leaf blades. Spikes obovoid, 7-10 mm long, the outer bracts subulate with raised midribs, nearly as long as or as long as the spike, the backs or at least the margins densely villous-tomentose with pale hairs, these often obscuring the bract outlines; bracts slightly widening inward in spike, the lowermost fertile ones the largest, lanceolate, 6-8 mm long, narrowly acute and at upper margins long-villous-pubescent, the convex backs castaneous and with large, lan- ceolate, pale-punctate dorsal areas, the mar- gins strongly convolute apically. Lateral se- pals lance-linear, free, equilateral, ca. 5 mm long, acuminate, the thin keel finely lacerate or flattened-villose toward apex. Petal blades broadly obovate, ca. 5 mm long, yellow, the narrowly rounded apex erose-crenulate. Staminodia bibrachiate, the narrow, flattened branches densely long-penicillate-ciliate. An- thers oblong-linear, ca. 1.5 mm long, on fil- aments ca. 1-1.5 mm long, deeply bifid, deeply sagittate. Capsule short-cylindric, ca. 3 mm long, the placental zone extending near- ly to apex, the narrow valves with strong septa, thus placentation axile. Seeds numer- ous, ellipsoid, ca. 0.5 mm long, yellow-amber, multiribbed with narrow but distinct, often anastomosing, ridges longitudinally. Distribution. Low-elevation savannas, southern Territorio Federal Amazonas, Ven- ezuela. Additional es examined. | VENEZUELA. T. F. AMAZONAS: Ca. al W del bajo Rio Temi, 100 m, 24 Feb. 1979, pa hh (US); 2-3 km al SE del bajo Rio Guasacavi, 10 Mar. 1980, Huber 5113 (US); 20 km al S del medio Cano Caname, ca. 100 m, 10 Mar. 1980, Huber 5144 (US, VDB); 20 km al NW de Yavita, en las cabaceras del Cano Pimichin, 120 m, 11 Feb. 1980, Huber & Medina 5949a (VEN). This subspecies appears to be fairly abun- dant in and around the type locality. It is very distinct in its combination of terete leaves and long, blonde or silvery, bristly and villose indumentum of spikes and edges of leaf sheaths. 76B. Xyris wurdackii subsp. caqueten- sis Maguire & Lyman B. Smith, Mem. New York Bot. Gard. 10: 24-25. 1963. TYPE: Colombia. Amazonas: frequent in scrub savanna, Araracuara, Rio Caque- 698 Annals of the Missouri Botanical Garden SR === SES NE | NS t ) NES FicuRE 76A. Xyris wurdackii (Huber 5144).—a. Habit sketch.—b. Leaf apex.—c. Leaf blade, ca. mid- blade. —d. Leaf base. —e. Spike. —f. Lowermost bract (lefi) ; lowermost fertile bract (right) .—g. Lateral sepal.— h. Petal blade, stamen.—i. Staminode, en beard hair.—j. Stylar apex.—k. Capsule through center longitudinally (left); one valve showing septum and cross section of same valve (right) .—l. Seed. Volume 75, Number 2 Kral 699 1988 Xyris 5mm —— —— am FIGURE 76B. Xyris wurdackii subsp. caquetensis (from the type) .—a. Habit sketch. —b. Leaf apex.—c. Leaf lade-sheath junction. —d. Leaf base. —e. Spike. —f. Lowermost bract (left); lowermost fertile bract (right) .— . Lateral sepal.—h. Petal blade, stamen.—i. Staminode, enlarged beard hair. —j. Stylar apex.—k. Capsule; edian longisection (right), oblique view of valve (lefi), cross section of fruit without placentae (bottom) .— . Seed. 700 Annals rea Eu Garden ta, 5 Sep. 1959, B. Maguire, C. K. Maguire & A. Fernandez 44129 (ho- lotype, COL; isotypes, NY, US). Figure 76B. This subspecies resembles the type sub- species, but most scapes overtop most leaves, and the leaf blades and scapes lack the prom- inent rows of pale punctae; the lowest sterile bracts are much shorter than spike and short- er than the outer fertile bracts; the inner sterile bracts are progressively less ciliate, becoming totally smooth; and the fertile bracts are entirely smooth. The capsule is narrowly cylindric and ca. 4 mm long; the seeds are slightly narrower and longer (0.5-0.6 mm); and the fine longitudinal ribs are more anas- tomosing. Distribution. Known only from the type area. Paratype. COLOMBIA. AMAZONAs: frequent in Ara- racuara Savannas, Rio Caqueta, Maguire et al. 44173 Y). This is another of the “close” vicariads involving the savannas of southern Colombia and Venezuela. 77. Xyris frequens Maguire & Lyman B. Smith, Mem. New York Bot. Gard. 10: 32-33, fig. 16A-E. 1963. TYPE: Ven- ezuela. T. F. Amazonas: sabanita along Cario Pimichin on right bank 1 km above Pimichin, Rio Guainia, 140 m, 24 Nov. 1953, B. Maguire & J. J. Wurdack 36383 (holotype, NY; isotypes, NY, US). Figure 77 Firm-based, cespitose perennial 2.5-3.5 dm high, the stems contracted. Leaves erect, the principal ones 1.5-3.5 dm long, the outer ones broad, short, bladeless brown scales; sheaths at intervals villous-ciliate, less than 1⁄4 the length of the blades, pale red-brown, rugulose or nearly smooth, tapering gradually to green, rigid, flexuous or straight, terete or subterete blades 1 mm thick, apically blunt- conic, punctate, tipped (often) with a pale pilose coma, the surfaces smooth distally, proximally rugulose, strongly ribbed or ecos- tate. Scape sheaths shorter than leaves, open, producing elongate blades similar to leaves. Scapes flexuous and twisted, somewhat com- pressed, oval in cross section, smooth or finely rugulose. Spikes broadly ellipsoid, drying tur- binate, 1-1.4 cm long, bracts several, in sev- eral ranks, loosely subdecussate, reddish brown with strong, paler dorsal areas, aging excur- vate; sterile bracts several, broadly triangular to ovate, ecarinate, often villous-tufted api- cally at base, the lowest bracts much smaller than and grading into the fertile bracts, these broadly oblong, rounded-folded, 8-10 mm long, narrowly rounded or acute apically, sub- entire or erose, when young often villous- ciliate apically and red-bordered, the dorsal areas with a strong midnerve, often with sev- eral subparallel laterals. Lateral sepals free, equilateral, lance-linear, ca. 5 mm long, acu- minate, the dark, wide, firm keel entire or apically ascending-ciliolate or villous-fimbrio- late and coarsely serrate toward base. Petal blades elliptic, ca. 5 mm long, broadly acute and wavy-margined, yellow. Staminodia nar- rowly bibrachiate, the slender branches pen- icillate-pilose. Anthers narrowly oblong, 2.5 mm long, deeply bifid, deeply sagittate, on filaments ca. 1 mm long. Capsule ellipsoid- cylindric, ca. 5 mm long, the placentation appearing free-central from base to apex, but each valve detaching with a wide septum, thus placentation actually axile. Seeds numerous on short funiculi, narrowly ellipsoid, 1.5-1.6 mm long, including a pale, conic apiculus, the body finely striate, pale brown, overlain by a coarse, irregularly anastomosing reticulum of dark brown ridges. Distribution. So far known only from savannas along the Cano Pimichin and those below Cerro Yapacana, Territorio. Federal Amazonas, Venezuela. Other than the type, the only records appear to be the following. Additional specimens examined. VENEZUELA. T. F. AMAZONAS: Savanna III, NW base of Cerro Yapacana, 150 m, 17 Mar. 1953, B. Maguire & J. J. Wurdack 34572 (NY); Savanna III, Yapacana savannas, 10 Au. 1983, Kral & Huber 70706 (MO, NY, US, VDB, VEN). This species most resembles the more wide- spread X. surinamensis but tends to have 701 Kral Xyris Volume 75, Number 2 1988 ; one valve, inner view.—k. Seed. A quts š sz a ZZ; = > > Sf , at right dried spike at anthesis) .—f Lateral sepal.—g. etal, stamen.—h. Staminode.—i. Stylar apex.—j. Capsule, one valve removed ATT bee T TE 2 AE TY did Pea sort tie — RAPT aia ranta s ETT y uE Was Xyris frequens (from the type).—a. Habit sketch.—b. Leaf apex.—c. Leaf at blade-sheath nction.—d. Leaf base.—e. Spike (at left in fruit FIGURE 77. u 702 Annals of the Missouri Botanical Garden more terete scape and leaf blade and a taller habit. Tips of most principal leaves have con- spicuous tufts of pale, bristly trichomes. 78. Xyris subglabrata Malme, Bull. Tor- rey Bot. Club 48: 322. 1931. TYPE: Ven- ezuela. T. F. Amazonas: Grand Savanna, section 1, in swampy ground, petals crimped, orange-yellow. Esmeralda, ca. 325 ft., 1 Nov. 1928, G. H. H. Tate 303 (holotype, NY; phototype, F). Fig- ure 78 Xyris epu -barrigae Idrobo & Lyman B. Smith, Cal- dasia 6: 204, fig. 8. 1954. TYPE: Colombia. Am zonas: aa e la rii u m, Rio aed Araracuara, 2] De Garcia- arriga & R. E. Schultes aie E COL; isotypes, NY, US). Cespitose, hard- and fibrillose-based peren- nial 1-3.5 dm high, the stems short, up from a stocky, horizontal or ascending rhizome. Leaves erect or ascending (0.7)1-2 dm long; sheaths with margins arachnoid or cottony, or pilose-ciliate, abruptly dilated at very base, reddish brown or castaneous, red-brown above, strongly papillose-rugose or nearly smooth, narrowing gradually to blades, ligulate or eli- gulate, the scarious edges often densely cot- tony; leaf blades lineal, terete-fluted, often sulcate and few-ribbed ventrally, 0.5- thick, dull green or maroon tinted, tapering to narrowly conic tips, these sometimes pilose- tufted, the surfaces smooth or papillose-ru- gulose toward base. Scape sheaths roughly equal to leaves and appearing like them. Scapes straight or flexuous, pale red-brown, slightly twisted, subterete distally, 0.7-1 mm thick, ecostate or striate, sometimes sulcate, papillose or smooth. Spikes ovoid to broadly ellipsoid, globose or obovoid, ca. 5-7 mm long, dark brown with several firm, spirally imbricate bracts having strong dorsal areas, also often with tips and base of bracts bearing white-villous patches; the lowest sterile bracts less than Y length of the fertile ones, ovate, ecarinate; fertile bracts broadly obovate, ca. 5-6 mm long, convex-rounded-backed, the apices broadly rounded, entire, the dorsal areas broadly triangular with strong mid and lateral nerves. Lateral sepals free, slightly inequilat- eral, lanceolate, ca. 4.5-5 mm long, acute, deep lustrous yellow-brown, the firm keel vil- losulous or ciliate above middle. Petal blades broadly obovate, 4 mm long, yellow, the broadly rounded apices erose-denticulate. Staminodia bibrachiate, the narrow branches long-penicillate. Anthers oblong, ca. 2 mm long, deeply bifid and auriculate, extrorse, on stout filaments ca. 1 mm long. Capsule nar- rowly obovoid, somewhat compressed dorsi- ventrally, ca. 3 mm long, placentation ap- earing free-central, but valves with complete septa toward capsule base. Seeds oblong-short- cylindric or narrowly ovoid, ca. 0.6-0.7 mm long, apiculate, dark amber, finely lined and very finely cross-lined longitudinally. Distribution. Mostly low-elevation, white-sand savanna, southeastern Colombia, southwestern Venezuela, and contiguous Amazonas, Brazil. Additional specimens examined. BRAZIL. AMAZONAS: base of Serra Araca, 0-3 km S of central massif, 3 km E of Rio Jauari, 0?49'N, 63?19'W, 8 Feb. 1984, G. T. Prance et al. 28885 (INPA, NY, VDB); plateau of north- : . Prance et "b 28979 (INPA, NY, VDB). COLOMBIA. vAUPÉS: Río Kuduyari, Cerro Yapa- boda, ca. 450 m, 5 Oct. 1951, Schultes & Cabrera 14234 (GH); same helo, 17 Nov. 1953, Schultes et E 18495 (GH); Rio Paraná, Pichuna, trib. of Rio Vaupés, 700 ft., June 1953, Schultes & Cabrera 19924, 19947 (GH); Cerro same locality, 9 Sep. VEN); transecta, margen derecha del cano “Cabez de Manteco" afluente del Rio Autana, a 2-3 km de la de- sembocadura, 12 Nov. 1984, Melgueiro & Guánchez 17 (TFAV, VDB); transecto desde comunidad indigena ^ etnia Piaroa, Río Autana en raudal ““seguera,”” 90- m, 6 Nov. 1984, Guánchez & Melgueiro 3306 TRAV. VDB); 2-3 km SE del bajo Rio Guasacavi, 10 Mar. 1980, od s.n. (VEN); savana del bajo Rio Ven- ud 20 a confluencia con el Rio Orinoco, 11 1977, "Huber 1062 (US, Ma alrededores de Ca 978, d 1913 (US o Yapacan 100 m, 3 — a rino & Tillett 2779 (US); sabana ubicada en la ribera NE de Cano Caname, 100 m, 30 June 1979, Huber 4047 (VEN) Volume 75, Number 2 Kral 703 1988 Xyris Vl; FIGURE 78. Xyris subglabrata (Kral 70692, 70709).—a. Habit sketch.—b. Three types of leaf apices.—c. Leaf sheath-blade junction.—d. Two sorts of blade sectors.—e. Leaf base.—f. Spike.—g. Lateral sepal.—h. Petal blade, stamen.—i. Staminode, enlarged beard hair apex.—j. Stylar apex.—k. Capsule; at left with two valves removed to show placentation, at right a single valve showing septum.—l. Seed. 704 Annals of th Missouri Bond Garden 2 km S del medio Rio Puruname, al SE del Caserio de Puruname, 100 m, Huber & Tillet 5454 (US, VDB); 20 m NW Yavita, cabeceras Cano Pimichin, 120 m, 11 Feb. 1981, Huber & Medina 5949 (VDB, VEN); km 11 de la Carretera San Carlos-Solano, 120 m, 16 Sep. 1980, Huber et al. 5673 (VEN); Savanna I, Cerro Yapacana, 10 Aug. 1983, Kral & Huber 70692 (BM, F, K, L, MO, NY, SP, TFAV, U, US, VDB, VEN, and others); Savanna II, 10 Aug. 1983, Kral & Huber 70709 (BM, F, K, L, MO, NY, SP, TFAV, U, US, VDB, VEN, and TER Cerro Moriche, Rio Ventuari, 4,500 ft., , Maguire et al. 30936A (NY, US, VEN); B T ni N San Carlos, 100-120 m, 6 Feb. 1977, Morillo & Villa 5395 (VEN); “Bana de AO 10 km San Carlos i 1982, Ruiz et al. 4025 ase of Cerro Duida, 200 m, 22 Aug. NY); Cerro Yapacana, 7 May DEA Steyerm ting 103253 (US, VDB, VEN); Canaripo, 28 Dec. 1 Steyermark & Redmond 112806 (MO, US, VDB, VEN); S, VEN). BOLÍVAR: ds Piar, llanura del Rio Aqu medio, ca. al NE de Uriman, 600 m 8 Dec. 1983, Huber 8473 3 (NY). This plant, along with X. setigera Oliver, often has the corolla attacked by a fungus, the spore masses lending an orange quality to the flower. This color is sometimes recorded on labels, but I have not seen any Xyris with healthy orange corollas. Consultation of the type material of X. garcia-barrigae and of other material so identified will show no sig- nificant differences from the earlier-named X. subglabrata. 49. Xyris lithophila Kral & Lyman B. Smith, sp. nov. TYPE: Venezuela. Bolivar: Dist. Roscio, arbustales, sabana rocosa y vegetación riberena en los alrededores del Salto ““La Milagrosa,” ca. 15 km al SW de S. Ignacio de Yuruani. Formando pequenas colonias sobre rocas en el lecho del rio, 4%55'N, 61°14’W, 1,000 m, 22 June 1983, O. Huber & Clara Alarcon 7568 (holotype, VEN; isotypes, MYF, VDB). Figure 79 Planta perennis, caespitosa, gracilis. Radices graciles. Caules breves. Folia principalia erecta, subtorta, flexuosa, ae; apices peranguste conicae aut subu latae; vaginae ecarinatae, pallide ferrugineae, ad basin leviter Wer, multicostatae, brunneolae, multicostatae, in laminas gradatim m decrescentes, aciebus dense longivil- losis, LUE pallidis. V i tatae, tubulosae, apicem versus ape r Scapi 2-3.5 dm alti, glabri, leviter torti, nitidi, leviter papillosi, olivacei, ad apicem tereti, leviter striati. Spicae eat ellipticae vel obovoideae, 7-9 mm longae, ra cteae erectae, laxe spiraliter imbricatae, eca- 5 mm longo, rotundato, aaa atae, ca. dorsali anguste t pla atr ralia libera, subaequilatera, oblanceolata, 5.5-6 m ga, leviter curvata, acuta; ala carinali lata a medio ad apicem lacerato- ub fimbriato-ciliata. Laminae pus late obovatae, 4.5-5 mm longae, luteolae, ad late rotun- datum apicem erosae. Staminodia bibrachiata, beach a basi ad apicem longipenicillatis. Antherae anguste o sa sed qu maturas capsulas et seminas vidi: sic, fructificatio cylin- drica, ca. 3 mm longa et placenta. centralis, doin ua massa profunde septatis a basi ad apic ellipsoidea, ca. 0.7 mm longa, ld da leviter multilineata. Slender, tufted perennial; roots slender; stems short. Main foliage leaves erect, slightly twisted and flexuous, 0.8-2 dm long, lightly papillose-rugulose, longer than the scape sheaths; es 2-3 times longer than the sheaths, filiform, slightly compressed or sub- terete, 0.5-0.8 mm wide, longitudinally finely nerved, shining, pale olive green to brownish; tips narrowly conic or subulate; sheaths ecar- inate, pale reddish brown, slightly dilated at base, multicostate, brownish, gradually nar- rowing into blades, the margins densely long- villous, the trichomes pale. Scape sheaths multicostate, tubular, opening toward apex, short-bladed. Scapes 2-3.5 dm high, smooth, slightly twisted, shining, lightly papillose, oli- mm long, obtuse. Bracts erect, loosely spirally imbricate, ecarinate but with conspicuous midrib, entire, pale brown, deep castaneous toward base. Lateral sepals free, subequilat- eral, oblanceolate, 5.5-6 mm long, slightly curvate, acute; keel broad, lacerate- to fim- briate-ciliate from middle to apex. Petal blades broadly obovate, 4.5-5 mm long, yellow, the broadly rounded apex erose. Staminodia bi- brachiate, the branches long-penicillate from Volume 75, Number 2 Kral 705 1988 Xyris FIGURE 79. Xyris lithophila (from the type) .—a. Habit sketch. —b. Leaf apex.—c. Leaf- sheath junction. — d. Leaf base.— e. Spi ike. —f. Fertile bract.—g. Lateral sepal.—h. Petal blade, stamen.—i. Staminode, enlarged beard hair apex.—j. Stylar apex.—k. Capsule, opened, one valve removed, showing placentation and septa (stippled) on two valves.—l. Seed. 706 Annals of the Missouri Botanical Garden base to tip. Anthers narrowly oblong, ca. 2 mm long, deeply bifid and sagittate on fila- ments ca. 1.5 mm long. Nearly mature fruit cylindric, ca. 3 mm long, the capsule valves deeply septate, the placentae central. Distribution. Known only from the type locality. The complex to which this slender plant belongs involves X. setigera, X. delicatula, and X. carinata, all of which are slender- leaved with foliage variously rugulose, the very thin sheath margins delicately pilose or villose. This taxon is distinguished primarily on the basis of smooth scapes (as in X. se- tigera) and conic-subulate tips to its slender leaves. The leaf tips not excentrically pointed, bract edges entire, and bract backs with very distinct dorsal areas are not usual in the com- plex. 80. Xyris carinata Maguire & Lyman B. Smith, Mem. New York Bot. Gard. 10: 36, fig. 21A-E. 1963. TYPE: Venezuela. T. F. Amazonas: occasional in Bonne- tia-grass savanna, summit Cerro Gua- nay, Cario Guaviarito, 1,800 m, Rio Ma- napiare, Rio Ventuari, 4 Feb. 1951, B. Maguire, K. D. Phelps, C. B. Hitchcock & G. Budowski 31773 (holotype, NY; isotypes, US, VEN). Figure 80. Slender, cespitose perennial 2-3 dm high, the stem short. Leaves erect or ascending, 1-1.5 dm long; sheaths mostly less than V4 as long as blades, pilose-ciliate, at very base brown or stramineous and papillose-rugulose, more scabrid and rugose above, reddish, ta- pering to blade, eligulate or with a short, broad, rounded-scarious ligule, ecarinate; blades narrowly lineal, twisted and flexuous, flattened (oblong in cross section), 0.8- 1 mm wide, narrowed above middle, then abruptly bluntly conic-tipped, smooth, the edges thick and rounded, tuberculate-rugose, the sides strongly verrucose-rugose, green with maroon tints. Scape sheaths shorter than leaves, tu- bular, twisted and costate, open distally, tipped with cusplike, erect, blunt blades. Scapes flex- uous and twisted, distally terete, 0.7-0.8 mm thick, papillose-rugulose. Spikes ellipsoid, aging obovoid, 7-8 mm long, dull brown or tan, of several erect but loosely imbricate, firm, spirally arranged bracts with narrowly ipis subapical dorsal areas and papillose, ounded backs; lowest sterile bracts much smaller than the fertile bract, ovate, narrowly rounded, grading into the fertile bracts, the outer fertile bracts narrowly obovate or ob- long, ca. 7-8 mm long, narrowly rounded, subentire, with narrow, short, subapical ca- rinae, the inner bracts progressively more folded and carinate. Lateral sepals free, sub- equilateral, oblanceolate, ca. 6 mm long, acute, tan with the thin, darker keel ascend- ing-ciliate from middle to apex. Petal blades narrowly obovate, ca. 5 mm long, yellow. Staminodia bibrachiate or (anomalously) com- pound, 2- or 4-branched, the branches long- penicillate. Anthers oblong, deeply bifid, deep- ly sagittate, ca. 2 mm long on filaments ca. l mm long. Capsule narrowly obovoid-cylin- dric, ca. 4 mm long, the ovary trilocular with septa breaking from axis at maturity, falling with valves. Seeds numerous, ellipsoid, ca. 0.7 mm long, acute, dark amber, longitudi- nally with ca. 12-14 strong ribs per side. Distribution. High, sandy rocky savan- na at medium to high (over 1,000 meters) elevations, southern Venezuela and contig- uous Brazil. Additional me ime examined. BRA ERR RÓRAIMA: vic. Auaris 4*3'N, 64?22'W, upper is of Serra Parima, Sof Auaris, 1,400-1,520 Py outcrop in seepage, Feb. 1969, Prance et al. 980 8 (F, K, U, US); same locality, 1,200 m, 30 July 1974, Prance et al. 21564 (U, NY, US, GH). VENEZUELA. T. F. AMAZONAS: cumbre del Cerro Autana, 1,230-1,240 m, 20-22 Se ep. 1971, Steyermark 105142 (US); Sierra Parima, cabec- eras del Rio Matacuni, a lo largo de la frontera Vene- zolana- Brasilera, 1,500 m, 19 May 1973, frontera no. 7, 4*5'N, 64?40'24"W, Steyermark 107505 (F, MO, US, VEN). BOLÍVAR: altiplanicie del Auyan-tepui, sector SSW de la meseta, al W de la cumbre sur, ca. 2,070 m, 27 Aug. 1983, Huber et al. 8079 (NY, VDB, VEN); Mount Auyan-tepui, 2,200 m, Dec. 1937, Jan. 1938, Tate 1320 (NY). Yet another of the growing number of species from the Guayana Highlands that, while placed in Nematopus, actually have Volume 75, Number 2 Kral 1988 Xyris 707 iam Li == one a oe 1 I ma, AE 5 — L 4mm FIGURE 80. Xyris carinata (from the type) .—a. Habit sketch.—b. Leaf tip. —c. Leaf she d. Leaf base (more reduced than other leaf sketches) .—e. Spike. —f. Lateral sepal.—g. h. Staminodia, one ath—blade junction. — etal blade, stamen.— of them compound.—i. Capsule outline, placenta superimposed; inner view of one valve showing septum.—j. Seed. 708 Annals of the Missouri Botanical Garden axile placentation. Further study of this species complex, which is around X. setigera, may require reduction of several of these entities to the level of variety or subspecies. 81. Xyris setigera F. Oliver, Trans. Linn. oc. London, Bot. 2: 285, pl. 50A, figs. 1-8. 1887. TYPE: Venezuela. Bolivar: Roraima (Venezuela), 4,000 ft., 4 Dec. 1884, Everard F. im Thurn 62 (lecto- type, K; isolectotypes, BM, US). Figure 81 Xyris setigera var. elongata Steyerm., Fieldiana, Bo t. J. A. Steyermark 59660 (holotype, F; isotypes, GH, NY). Densely cespitose, stiffish perennial, mostly 1.5-6 dm high; stems contracted to slightly elongated. pas erect to spreading flabel- lately or ascending, 0.5-2.5 dm long; sheaths dark to pale, red-brown or purple-red, mostly more than Y as long as blades, the edges long-ciliate, narrowing gradually from the di- lated, scabro-rugose (rarely smooth) base into the blades, there eligulate or in high-elevation forms sometimes producing a ligule to 2 mm long, the blades narrowly linear to filiform (in high-elevation forms), usually flattened, 0.5- 3 mm wide, at apex excentrically short-spi- nulose or narrowly conic-subulate, the mar- gins thick, scabro-ciliate or tuberculate-sca- brid with strumose, antrorsely or retrorsely bent, simple or branched trichomes or tuber- cles (rarely smooth in high-elevation forms), the surfaces deep green to maroon or olive, tuberculate-rugose or smooth, punctate. Scape sheaths shorter than leaves, the bases tubular, multicostate and keeled, often reddish or ro- seate or lucid brown, open distally, keeled, producing a short, strong blade similar to leaves. Scapes straight or flexuous, twisted, terete distally, ca. 0.5-1 mm thick, smooth, often lustrous, punctate, maroon to green. Spikes ovoid to ellipsoid, 0.5-1 cm long, aging obovoid or turbinate, pale to deep brown, of several loosely spirally imbricate, stiff bracts usually without distinct dorsal areas (occa- sional extremes with narrow, indistinct dorsal areas); sterile bracts few, oblong, spreading, ecarinate, smaller than and grading into fer- tile bracts, these oblong to narrowly obovate, ecarinate, 6-7 mm long, narrowly rounded apically, entire, aging lacerate, the backs slightly convex. Lateral sepals free, subequi- lateral, mostly oblanceolate, 5-6 mm long, acute, dark reddish brown, the keel at middle and above lacerate-serrulate to nearly entire. Petal blades broadly obovate, ca. 6 mm long, yellow, the rounded apex subentire. Stami- nodia bibrachiate, the slender branches dense- ly long-penicillate. Anthers 1.7-2 mm long, narrowly oblong, deeply bifid and sagittate, on filaments 0.8-1 mm long. Capsule cylin- dric-ellipsoid, ca. 4.5 mm long, the placen- tation axile, the septa separating from the central (placental) axis as the valves part. Seeds numerous, ellipsoid-ovoid, ca. 0.5-0 mm long, 2-apiculate, dark amber, longitu- dinally finely but strongly ribbed and with several less distinct cross-lines. Distribution. Medium- to high-eleva- tion, sandy or rocky savanna, centering in the massifs of the Gran Sabana, Estado Bo- livar, Venezuela, southward shortly over the border of Brazil, eastward into eastern Guy- ana and northern Territorio Roraima, Brazil. Additional specimens exa BRAZIL. RORAIMA: Mt. Roraima, E. F. rn 62 joi type for the species at K shows that some of the set could have the Brazil side of Roraima); ? Phili p. 5,200 ft., 11 Nov. 1927, Tate 336 (K — ? locality listed for Brazil, British gue and Venezuela, depending on collector at this site). GUYANA: Kaieteur Plateau, 2 May 1944, Maguire & dior inii du (K); southern Paka- raima Mts., 5 a, Kopinang Falls, 2,700 ft., 29 Aug. 1961, More et al. 45988 (K, VEN). VENEZUELA. (Note — this is perhaps the most common and given here. A complete list will be m available later.) BOLÍVAR: Cienagas del Cerro Uaipan, 1,750 m, Rio Caroní, 26 Nov. 1946, Cardona 200g (US, VEN); 2 km S of La Ciudadella, 1,300 m, 3 Dec. 1973, Davidse 4699 (MO, VDB); ca. 17 km NE Ikabarü, 1,100 m, 7 Oct. 1982, Huber et al. 6727 (MYF, VDB, VEN)— this material has dorsal areas; sabanas, sector norte de la Gran Sabana, 1,030 m, 5 Mar. 1982, Huber et al. 7337 (MYF, VDB, VEN); cumbre del Cerro Chirikayne ca. 15 k NW Sta. Elena de Uairen, 1,580 m, 25 June 1983, m Huber et al. 7601 (MYF, VDB); Cerro Venado ca. 20 Volume 75, Number 2 1988 Kral 709 Xyris 6 mm Lam d. Leaf base.—e. Spike.—f. Fertile bract outline.—g. 5777 TG [x e Yer OC SER ES ^ iw FIGURE 81. Xyris setigera (Kral 70630) .—a. Habit sketch.—b. Leafapex.—c. Leaf at sheath-blade junction. — ateral sepal.—h. Petal blade, stamen, staminode, staminodial beard hair.—i. Stylar apex.—j. Capsule, dehisced, showing placentation and side view of one valve.—k. Seed. 710 Annals of the Missouri Botanical Garden km E de Canaima, 1,300 m, 31 Aug. 1983, Huber et al. 8251 (NY, VDB, VEN); Macizo del apr sector px suroriental del Acopan-tepui, 1,920 -16 Feb. 4, Huber et al. 9061 (NY, VDB, TEN Chiman, ud -tepui, 2,250 m, 28 Mar. 1984, Huber 927 1(MYF VDB); Brazo ‘Oacident al del Auyantepui, 1,650 m, 13 Nov. 1984, Huber 9732 (MYF, VDB); ca. 5 km NNW Parupa, 1,400 m Nov. 1984, Huber 9841 (MYF, VDB); Meseta Guaiquinima, sector nororiental, ca. 2,000 m, 21 Nov. 1984, Huber 9877 (MYF, VDB); ca. 30 km Kanca, 900 m, 12 Dec. 1984, Huber et al. 9919 (MYF, VDB); just S of summit La Escalera, 24 July 1983, e & Gonzalez 70364, 70365 (BM, F, K, ; ] , U, US, VDB, VEN, and others); 28 km E d 25 Jul y 1983, Kral & Gonzalez 70432 46 km > of Sta. Elena, 27 July & Gonsales 70557 (F, K, MO, NY, SP, ; 8 km N of San Rafael, 29 July 1983, & Gonzalez 70580 (BM, F, K, L, MO, NY, SP, VEN, and others); 15 km S of San Rafael, 29 1,300 m, 30 n 1983, Kral & Gonzalez 70630 (BM, F U, US, , and others); savanna a n Raphael and Enemasic, 1,200 m 5 Feb. 1952, Maguire 33174 (NY, US, VEN); Eco nayen, ca. 1,200 m, 9 Aug. 1970, Moore et al. 9646 (US, VEN); cumbre de Uaipan, 1,900 m, Jan. 1948, Phelps 375 (NY, US); Plateau of eer tepui, central s section of eastern branch, 1,940 m Aug e & Huber 28270 MYF, NY, VDB); Re. , 25 Sep. 1944, id 611 (F, NY, VEN); Puri. -tepui, 1,600 m, 1 Nov. ion Steyermark in (F, NY, US, VEN); mesa oropan-tepui, 1,615 m, 15- 944, nan 60239 (F, K); Auyan-tepui, 1,660-1,800 m, 11 May 1964, Steyermark 93684 (NY, , , , , , o dana cundis. Sud a E a 109427 (K, , VEN); Cerro Marutari, 1,200 ES 2 Jan. 1981, Steyermark 123857 (NY, VDB); Meseta de Sarisarina- 981, oi 124319 (NY, VDB); Luépa (Ciudadela) savanna of check- point, Thomas 2702 (NY). This species is, not surprisingly, the most variable xyrid in the Guayana Highlands, with populations being found throughout the Gran Sabana at elevations extending from 700 to 2,000 meters or more on tepuis. Those of the lower and middle elevations within this area mostly conform to the described mor- phology pretty well, with the foliage generally coarsely rugose; the leaf blade edges rough- ened with stiff, often branched hairs; the leaf tips excentrically short-spinulose; and the spike bracts many, dark, and erose with age. At the higher elevations in the tepuis, usually at ca. 2,000 meters upward, however, the plants are more slender in scape and leaf; the leaf tips are subulate-conic (rather than excentri- cally spinulose) with the surfaces and edges becoming smoother; and the blades are nar- rower and approaching terete. These are problem plants; many are in loans annotated X. setigera var. elongata Steyerm. Some need further study, as they border on or are the same as X. byssacea. A smut fungus infects flowers to produce a bright orange fruiting mass on the spikes, hence the erro- neous records that some Xyris corollas are "orange." 82. Xyris riparia Maguire & Lyman B. Smith, Mem. New York Bot. Gard. 10: 20, fig. SA-E. 1963. TYPE: Venezuela. Bolivar: frequent in clumps at stream edge, scrub forest near Summit Camp, 1,925 m, Chimantá Massif, central sec- tion, 2 Feb. 1955, J. A. Steyermark & J. J. Wurdack 361 (holotype, NY; iso- types, MO, US, VEN). Figure 82. Cespitose, slender, stiffish perennial ca. 4— 5 dm high, the stems short to elongate, form- ing frondlike plates of leaves. Leaves spread- ing flabellately to erect, 1-2.5 dm long; sheaths coarsely to finely long-ciliate-mar- gined (sometimes sparsely so), as long as blades, deeply dull reddish brown, dull brown or castaneous, scabro-rugose or smooth, ta- pering gradually from the dilated base to the blade, the backs narrowly rounded or keeled and scabrid, usually eligulate or with a short, erect ligule under 0.5 mm long; blades linear, flattened, slightly twisted, 2-3 mm wide, ta- pering gradually from ca. middle to apex, then abruptly narrowed to an excentric, narrowly conic-acute to short-subulate tip; margins thin or pale-incrassate, smooth to papillate, tu- berculate or scabro-ciliate or ciliolate with pale hairs, the surfaces yellow-green to red- dish, few-nerved, above, ciliate-keeled, open, with a short, erect Volume 75, Number 2 Kral 711 1988 Xyris FIGURE 82. e y us A type) .—a. Habit sketch.—b. Leaf apex.—c. Leaf, midblade.—d. ad blade—sheath junctio Leaf base. —f. Spike.—g. Lateral sepal.—h. Petal blade, stamen.—i. Staminod j. Stylar apex. —k. asas qa one frs removed; side view of one valve to show septum and side view Succ — l. Seed. 712 Annals of the Missouri Botanical Garden blade. Scapes straight or slightly flexuous, twisted distally, subterete or broadly oval in cross section, ca. 1 mm wide, ecostate, some- times with few very low ridges, punctate. Spike broadly obovoid, drying turbinate, ca. 1 cm long, reddish brown, of several, firm, papillose bracts without dorsal areas and loosely im- bricate in near vertical rows; sterile bracts with lowermost lance-ovate, much shorter than and narrower than the fertile bracts, carinate, grading into fertile bracts, these numerous, oblong to narrowly ovate, 6-8 mm long, nar- rowly rounded, entire, aging lacerate, the margins often narrowly squarrose. Lateral se- pals linear-oblanceolate, 6.5-7 mm long, acu- minate, dark red-brown, the firm, broad keel slightly lacerate distally. Petal blades broadly obovate, ca. 5 mm long, yellow, the broadly rounded apex crenulate. Staminodia clavate, indistinctly bibrachiate, the irregular margins long-penicillate. Anthers lance-oblong, ca. 2 mm long, shallowly bifid and sagittate, on filaments ca. 1 mm long. Capsule narrowly ellipsoid, ca. 5 mm long, the placentation appearing central but dehiscing valves with strong septa toward base. Seeds numerous on short, stout funiculi, narrowly ovoid, ca. 0.9- l mm long, deep amber, apiculate, longitu- dinally strongly multiribbed and cross-lined. Distribution. Local in high-elevation sa- vanna, the Chimanta Massif and Cerro Guai- quinima, Estado Bolivar, Venezuela. The few known records, in addition to the type are as follows. Additional specimens examine ho ed BOLÍVAR: Cerro Guaiquinima, savanna vic. Cumbre Camp 2,000 m, 25 Dec. b B. Maguire 32751 (NY s e locality, 1 km Cumbre Camp, 1,800 m, c. 1951, B. Maguire 32798 (NY); Macizo del ree sección centro-suroriental del Churi- -tepui, ca. 2,250 m 10-12 Feb. 1984, Huber & Colella ip VEN)— this an extreme bordering on X. setigera - The material from Cerro Guaiquinima should perhaps be considered varietal, being more scabrid in sheath, more ciliate in leaf, and with a shorter range in leaf length, but at this stage of knowledge of the genus it is perhaps best to be conservative. In such a connection, it might be noted that there are many similarities between this taxon and the more common X. setigera, certainly as to the punctate leaves and scapes, the complex harsh hairs on edges of the leaf blade, and the excentrically spinulose-subulate leaf tips. 83. Xyris roraimae Malme in Pilger No- tizbl. Bot. Gart. Berlin 6: 117. 1914. TYPE: Venezuela. Bolívar: “Auf den un- teren Campos,” 1,700 m, Jan. 1910, Ule 8546 (lectotype, B; isolectotype and phototype, US). Figure 83. Tall, solitary or small-tufted, brittle peren- nial to 1 m high or slightly higher, the stems contracted, the foliage totally papillose-ru- gulose, dull yellow-green. Leaves erect or as- cending, twisted, 2-4 dm long; sheaths long- ciliate, 2 as long as blades or longer, abruptly dilated at very base, castaneous, thence abruptly narrowing, dull brown, gradually ta- pering into the blade, there eligulate or with a short, erect, ciliate ligule less than 0.5 mm long; blades strongly flattened, linear, 3-5 mm wide, above middle tapering to an erect, narrowly acute or acuminate, somewhat thickened, erect or incurved apex, the mar- gins pale ciliate and tuberculate, the surfaces dull yellow-green, strongly nerved longitudi- nally. Scape sheaths shorter than leaves, twisted, carinate and strongly costate, open- ing toward apex, keeled, producing a very short, incurved blade. Scapes lineal, twisted and flexuous, flattened and ancipital distally, 2-3 mm wide, 2-costate, the costae reaching the edges and densely pale ciliate. Spikes broadly ovoid, ellipsoid, or cylindric, 1-3 cm long, blunt, pale red-brown or tan, of very many spirally imbricate, thin bracts without dorsal areas and, when young, with narrow but distinctive, reddish, scarious, lacerate bor- ders; lowermost sterile bracts smallest, broad- ly ovate to suborbicular, slightly carinate, grading gradually into the fertile bracts, these broadly obovate to suborbicular, 5-6 mm long, ecarinate, aging lacerate (the red border fri- able and deciduous early). Lateral sepals free, very inequilateral, often exserted apically, ob- lanceolate, 4.5-5.5 mm long, acute, thin, Volume 75, Number 2 Kral 713 1988 Xyris TELS TY ; M FIGURE 83. Xyris jocis (Kral 70560).— a. Habit sketch. —b. Leaf apex.—c. Leaf sheath- ied junc- tion.—d. Leaf base.—e. Spike.—f. Fertile bract.—g. Lateral sepal. —h. Petal blade, stamen.—i. Staminode.— j. Stylar apex.—k. TR D —l. Seed. 714 Annals of the Missouri Botanical Garden with the broad keel lacerofimbriate or ciliate from ca. middle to apex. Petal blades obovate, ca. 5 mm long, yellow, the narrowly rounded apex subentire. Staminodia bibrachiate, the flat, narrow branches long-penicillate distally. Anthers lance-oblong, ca. 2 mm long, shal- lowly bifid apically, the base deeply sagittate, on filaments ca. 1 mm long. Capsule broadly obovoid, 2.5-3 mm long, the placentation basal, the valves without septa. Seeds ellip- soid-cylindric, ca. 1 mm long, amber, apic- ulate, finely but distinctly ribbed longitudi- nally Distribution. Savanna bogs, medium el- evations (ca. 700-1,500 meters), Gran Sa- bana, Estado Bolivar, Venezuela; Guyana; oc- casional in the campos rupestres, the planalto of Brazil in Minas Gerais and Goias. My Bra- zilian specimens tend to be proportionately larger in height (to 1.5 meters) and in spike diameter, but otherwise appear the same as the Venezuelan collections. Additional specimens examined. BRAZIL. MINAS GERAIS: wet slopes over arenaceous rock, above Rio Ita- cambirucu, ca. 8 km WSW of Grào Mogul, 600-750 m, 10 July 1985, Kral et al. 72705 (SP, VDB, and to be distributed) grass-sedge campo, Morro do Onca, 950- 1,000 m N of Jouachim Felicio, 6 July 1985, Kral et al. 72628 (SP, VDB, and to be distributed); below sand- stone bluffs, Morro do Jucao, NNW Joachim Felicio, d to be distributed). f road, 15 9 km E of Kavanayen, Gonzalez 70539 (F, MO, NY, SP, Elena on E side of 70 70560 (BM, F, L, MO, NY, SP, U, US, VDB, VEN, and others); colinas above Ven. 10, 8 km N of El Salto ca. 1,060 m, 15 Dec. 1984, Kral 72089 (MYF, VDB, VEN, and to be distributed); Ilu-tepui, Gran Sabana, Río Aponangao at 1,200 m, 27-28 Mar. 1952, PURE 33646 (NY); Gran Sabana, cabeceras Río Aponguao lo largo Arauta-paru, km 148, 20 Dec. 1970, Mie cenar et al. 104116 (US, VEN); savanna with Stegolepis 16 km from Kavanayen, 1,340 m, 19 Dec. 1979, Steyer- mark with Pruski 121052 (US, VEN). — This species, locally abundant in the ra- pateaceous bogs of the Gran Sabana, is at of Jouaquim Felicio, 7 July 1985, | once one of the tallest of Venezuelan xyrids and the most roughened in foliage, the epi- dermal roughening giving the plants a glazed, dull, yellow-green color which is very distinc- tive in the field. Its spikes are a lovely tint of golden brown and on close inspection show a fine red color of bract border. Éven larger versions of this attractive (for a Xyris) plant are to be found in the planalto of Brazil, where they belong to the most difficult complex around X. ciliata Thunb. 84. Xyris schneeana Lyman B. Smith & Steyerm., Acta. Bot. Venez. 1(7): 149- 150. 1965. TYPE: Venezuela. Bolivar: Auyan-tepui, altiplanicie, 19 Apr. 1956, Ludwig Schnee 1577 (holotype, MY; phototype, US). Figure 84. Cespitose low perennial 1-3 dm high, the stems contracted. Leaves stiff, spreading fla- bellately, 5-10 cm long; sheaths as long as blades or longer, long-ciliate, papillose-rugu- lose, tan, much dilated at very base, thence narrowing gradually to blade, strongly keeled, eligulate, the blades flat, straight, 2- wide, lineal, abruptly narrowed at apex to an obtuse or broadly incurved-acute tip, this fre- quently apiculate, the margins white-ciliolate, the surfaces dull green, low-multiribbed, sca- berulous-rugulose. Scape sheaths overtopping most leaves, the bases tubular, multicostate, carinate, above open, ciliate, the blade a mere apiculus. Scapes dull green, straight or flex- uous, terete or oval distally, ca. 0.5 mm thick, with several low costae, the surface entirely rugulose-papillate or scaberulous. Spikes ovoid-ellipsoid, ca. 5 mm long, drying ob- ovate, dark olive-castaneous, of several loose- ly subdecussate, thin bracts without evident dorsal areas, the sterile ones ca. 4, ecarinate, rounded, shorter than the few fertile bracts, these oblong, ca. 5 mm long, navicular but ecarinate, broadly rounded, entire or erose. Lateral sepals free, very inequilateral, oblong- elliptic to spathulate, ca. 5-5.5 mm long, acute or obtuse, dark and thin, the narrow, darker keel serrulate-ciliolate above middle, the apex often exserted. Petal blades elliptic, 5 mm long, yellow, the apex broadly acute, Kral 715 Volume 75, Number 2 1988 Xyris Xyris schneeana (Steyermark et al. 128918).—a. Habit sketch.—b. Leaf apex.—c. Leaf at blade—sheath junction.—d. Leaf base. —e. Spike. —f. Lateral sepal.—g. Petal blade, stamen.—h. Staminode.— i. Stylar apex.—j. Capsule, dehiscing, showing two valves, placentation. —k. Seed. FIGURE 84. 716 Annals of the Missouri Botanical Garden the margin sinuate. Staminodia bibrachiate, the branches long-penicillate. Anthers lance- oblong, ca. 1.7-2 mm long, deeply bifid and sagittate, on filaments 0.8-0.9 mm long. Cap- sule broadly ellipsoid or nearly obovoid, ca. 3 mm long, placentation basal-central. Seeds narrowly ellipsoid or cylindric, ca. 1 mm long, amber, longitudinally finely but distinctly mul- tiri Distribution. High, sandy, rocky savan- na-plain, Chimantá Massif and related tepuis, Estado Bolivar, Venezuela. Additional specimens examined. VENEZUELA. BOLÍVAR: Kur Ai iS A Pepu: Lage en el sector SW del Macizo, 2,200 m, 31 Jan . 1984, Huber & Dezzeo 8636 (NY) uis Muse s sec- tor sur, ca. 30 de Uriman, ca. 2,500 m, 30 June 1984, Huber ' 9580 (MYF, VDB); NW part of sum- F, NY); sector centro-noreste del Chimanta-tepui, cabeceras orientales del Cano Chimanta, 2,000 m, 26-29 Jan. 1983, Steyermark et al. 128162 Mos VEN); Apacara-tepui, sector Norte del Macizo, 2,200 m, Steyermark et al. 128372 (VDB, VEN); sec- ción onient del Chimantá-tepui, cabeceras del afluente del Río Tirica (Caño del **Grillo"), 2,450 m, CASE et al. 128918 (VDB, VEN). The stubby, low habit of the leaves, dull green and papillose-rugulose foliage, small and dark-bracted spikes, and strongly septate cap- sule valves combine to distinguish this high- tepui endemic. 85. Xyris consolida Kral & Lyman B. Smith, Phytologia 53(6): 434—435, figs. 3, 4. 1983. TYPE: Venezuela. Bolivar: Uaipan-tepui, the summit of west Peak, 1,980 m, small, wet grassy swamp on sandstone, 4 Mar. 1967, G. Agostini & T. Koyama 7462 (holotype, VEN; iso- types, NY, US). Figure 85A, B. Plants perennial, solitary or in small tufts; leaf and scape surfaces scabrid; habit lax; roots slender; stems contracted. Leaves all basal, ensiform-linear, 1.5-2 dm long, longer than scape sheaths, spreading flabellately; blades flat, strongly compressed, 3-5 mm wide, olivaceous, multinerved longitudinally, transversely rugose; apices a ruptly nar- rowed, incurved-acute at tip, slightly thick- ened; margins slightly thickened, minutely tu- berculate-ciliate to strigo-ciliate or entire; sheaths gradually narrowed from base to apex, carinate, the carinae ciliate with spreading to retrorse hairs, the sides transversely rugulose, dull olive, multinerved longitudinally, lustrous at base, deeply red-brown, the edges narrow, pale, hirsute-ciliate with sordid trichomes. Scape sheaths twisted, carinate, prominently multicostate with carinae and submarginal costae strongly tuberculate and ciliate, the sheath apex with a short, erect, flat blade Scapes linear, twisted, 4—6 dm high, subterete or slightly compressed, 1-1.5 mm thick, ol- ive, multicostate, with costae scabrous. Spikes ovoid to broadly obovoid or subglobose, 8- 10 mm long, 6-7 mm wide, several-flowered, the bracts loosely spirally imbricate, ecari- nate, convex-backed, the margins narrowly scarious; sterile bracts several, gradually passing larger into the fertile bracts, 3.5-4.5 mm long, broadly obovate to obovate, round- ed, the margins narrowly scarious; fertile bracts broadly obovate, ca. 5 mm long, round- ed-convex; dorsal area absent. Lateral sepals ca. Y3 connate, inequilateral, ca. 5.5 mm long, the lobes oblong, broadly acute or obtuse; keel wide, thin, entire from base to middle, cilio- late-lacerate from middle to apex. Petal blades obovate, 4-4.5 mm long, yellow, the nar- rowed apex erose. Staminodia bibrachiate, the branches apically long-penicillate. Ripe cap- sule ellipsoid, 3.5 mm long, the valves without septa, the placentation basal-central. Seeds oblong-ellipsoid, ca. mm long, amber, strongly multiribbed longitudinally. Distribution. Known only from the type collection (Uaipan-tepui) and from Ptari-te- pui, in Estado Bolivar, Venezuela. ditio nal specimens examined. VENEZUELA. Santa Teresita de 18, Steyermark et al. 115729 -~ F, MO, US, VEN). The scabrous foliage, small, dark-bracted spikes, and connate lateral sepals distinguish this rather low plant. At the beginning of the study it was thought that the two collections represented tepui-summit endemics; a closer Volume 75, Number 2 Kral 717 1988 Xyris | mim PY pipia dispersal and colonization of aqu ~ m w N m e sg, e tp E D c = m Eno © in the amo suggested that genetic oe nn associated with long- distance with uncertain moisture regimes play major roles in disrupting : road range of morphological specializations associated range from large-flo vp putcrossing ritus es of evi- onmi ful E ade: 8 9 Aisa argely a the maintenance of tristyly. This ud to the breakdown of the polymorphism and the evolution of semi-homostyly. Eichhornia (Pontederiaceae) is a small genus comprised of eight species of freshwater aquatics and palustrial herbs. All species, with the exception of the exclusively African E. natans, are native to the New World tropics. Eichhornia azurea and E. panic mida are ho as A pond ornamentals, and E. with man's Bid: bon lonland rud South America to be- come one of the world's most noxious aquatic weeds. The flowers of Eichhornia are usually mauve- blue and showy, and are pollinated primarily by bees and butterflies, They display a broad range of associated with their pollination mechanisms and breeding systems. The variation ranges from large multicolored tristylous flowers adapted to outcrossing to small uniformly colored self-fertilizing homostylous flowers. This di- versity provides suitable experimental material for studies of breeding system evolution. The present review summarizes work on this topic with special attention to the evolutionary breakdown of tristyly and the responsible ecological and genetic factors. Heterostyly is a genetically controlled floral polymorphism in which plant populations contain two (distyly) or three (tristyly) morphs that differ primarily in style and stamen length, pollen size, and incompatibility behavior. The polymorphism promotes disassortative mating between the floral morphs and is reliably reported from 24 angio- sperm families of which just three (Lythraceae, Oxalidaceae, Pontederiaceae) contain tristylous members (Ganders, 1979). A common feature of heterostylous breeding systems is their propensity to become evolutionarily modified in the direction of increased self-fertilization. The main pathway is by the formation of self-compatible homostyles. Plants in these groups possess anthers and stigmas at the same level within a flower and, as a result, are largely autogamous. In tristylous species homo- styles often have only one of their two anther sets meee to the stigma and as a result are referred as semi-homostylous (Stout, 1925; Ornduff, 1972). The breakdown of heterostyly to homostyly represents a paradigm for studies of the evolution of self-fertilization in plants, because the direction of change is readily interpretable, genetic modifi- cations are often simply inherited, and alterations in the floral polymorphisms that influence mating behavior can usually be detected without difficulty under field conditions (Barrett, 1988a). We have investigated the pathways and mech- anisms of breeding system evolution in Eichhornia nk Jim Eckenwalder, Jennifer Richards, Tim Dickinson, Steve Price, Jan Anderson, Lorne Wolfe, Deborah "oa Tal Shore, Robin Scribailo, Brian Husband, Martin Morgan, Peter Toppings, Petra Donnelly, and Chris Eckert for advice and assistance; Elizabeth Campolin for ei the figures; and the Natural Sciences and Engineering Research Council of Canada for financial su ? Department of Botany, University of Toronto, Toronto, Dias Canada M5S 14A1. ANN. MISSOURI Bor. GARD. 75: 741-760. 1988. 742 Annals of the Missouri Botanical Garden by adopting the comparative approach employed by many plant iu and evolutionists (e.g., bla. 1957; ; Grant & Grant, 1965; Lloyd, 1965; cree & Lewis, 1965; Raven, 1979). Here the working hypothesis is that among closely related taxa of outcrossers and selfers the former condition is usually ancestral and the lat- ter derived. Stebbins (1974) has argued that this pathway has been followed by more lines of evo- lution in the flowering plants than has any other. To test this hypothesis in Eichhornia two types of evidence were sought. Population studies of out- crossing taxa were conducted to establish whether, under present conditions, the shift from outcrossing to selfing is occurring. Of primary importance in these studies is to determine whether genetically based differences in mating patterns occur both within and between populations of the tristylous taxa. A second line of enquiry has involved inter- specific studies of the floral biology of selfing taxa to discover whether residual tristylic traits are ev- ident (see Crowe, 1964). Their occurrence can be taken as evidence of descent from tristylous ances- tors, unless we are prepared to accept that tristyly is in statu nascendi within different selfing lineages of the genus. Given the rarity of the polymorphism in the angiosperms as a whole, this evolutionary scenario seems unlikely (see Eckenwalder & Bar- rett, ) This review summarizes evidence in favor of the derived nature of selfing taxa and evaluates several hypotheses concerned with the selective mecha- nisms responsible for the breakdown of tristyly. Before this is done, however, a brief summary of the systematic and ecological characteristics of the genus is required. SYSTEMATICS AND ECOLOGY Taxonomy. Most floristic works and regional floras follow Solms-Laubach (1883) and Schwartz (1927, 1930) in their treatments of Kichhornia (e.g., Alexander, 1937; Schulz, 1942; Castellanos, 1958; Agostini, 1974; Rosatti, 1987). Schwartz recognized two sections in the genus, the primitive Protoeichhornia composed of E. paniculata and 7. paradoxa, and the more advanced Eichhornia (“Eueichhornia’ `) containing E. azurea, E. cras- sipes, E. natans, and E. diversifolia. More re- cently, two frequently overlooked species, E. het- erosperma and E. meyeri, have come to be recognized. Eichhornia heterosperma, first de- scribed in 1939 from Venezuela by Alexander (Smith, 1939) and very similar in vegetative traits 4 to E. azurea, is widely distributed in Central and South America and is misidentified in most collec- tions (Horn, 1987). Similarly, E. meyeri has most often been treated as a synonym of E. paniculata, which it resembles closely. It is known from a few localities in Paraguay and from the type locality in the Chaco of Argentina (Schulz, 1942). Phylogeny. A recent cladistic analysis of the Pontederiaceae by Eckenwalder & Barrett (1986) is of interest because most cladograms produced were in conflict with Schwartz s sectional treatment of Eichhornia and the genus was paraphyletic un- der most methods of cladogram construction. Of particular relevance to breeding system evolution was the placement of tristylous and homostylous taxa. Two contrasting patterns emerged from the analyses, neither of which is supported by micro- evolutionary studies presented below. Most clado- grams separated homostylous species of Eichhor- nia as a clade not closely associated with tristylous Fichhornia and Pontederia but allied instead to Heteranthera. 'This seems unacceptable because of the clear relationships among species such as £. azurea (tristylous) and E. heterosperma (homosty- lous), and E. paniculata (tristylous) and E. meyeri (homostylous). A second pattern derived homosty- lous taxa from tristylous ancestors, but in this case all homostylous taxa were united as a monophyletic group implying a single origin for homostyly. How- ever, evidence presented below indicates that tris- tyly has broken down repeatedly within Kichhorn- ia, implying multiple origins for homostyly in the genus. These disparities probably result from the choice of characters used in cladogram construc- tion. The parallel evolution of the selfing syndrome in different homostylous taxa may provide enough characters to unite species during tree construc- tion. This difficulty has been recently discussed at length by Wyatt (1988) in his review of phylo- genetic aspects of the evolution of self-pollination. Genetic systems. Breeding systems and chro- mosome numbers of Eichhornia species are given in Table 1. Three of the eight species are primarily tristylous, and the remaining five are semi-homo- stylous. However, because of considerable intra- specific variation in floral traits, it is important to note that these terms refer to the most common condition within each species. In Eichhornia poly- ploidy and aneuploidy are derived from an original base number of x — 8. Unlike other heterostylous groups (e.g., Turnera ulmifolia L., Barrett & Shore, 987), there is no evidence of an association be- tween breeding system and ploidal level. Tristyly and semi-homostyly occur at the diploid and tet- raploid level. Volume 75, Number 3 1988 Barrett Breeding Systems in Eichhornia 743 TABLE 1. are the most common condition within each species. Genetic systems, life forms, and distributions of Eichhornia species. Breeding systems indicated Chromo Major some Clonal Breeding Number Propa- Taxon System (n) Life Form gation Native Distribution E. azurea (Swartz) tristylous 16 floating-leaved aquatic, long- ++ widespread, Neotrop- unth lived perennial i E. crassipes (Mart.) — tristylous 16 free-floating aquatic, long-lived +++ ^ widespread, lowland olms-Laub. ennial th America E. paniculata tristylous 8 emergent aquatic, short-lived = locally abundant, (Spreng.) Solms- perennial or annua northeast Brazil, Laub. uba, Jamaica, Nicaragua, Ecua- or E. heterosperma semi- 15 floating-leaved aquatic, peren- ++ widespread, Neotrop- Alex. homostylous nial ics E. diversifolia semi- 15 floating-leaved aquatic, peren- + widespread, Neotrop- (Vahl) Ur homostylous nial or annual ics E. natans (Beauv.) semi- — floating-leaved aquatic, peren- + widespread, Africa Solms-Laub homostylous? nial or annual E. paradoxa (Mart.) semi- 8 emergent aquatic, annual nd rare, pr olms-Laub. homostylous nezuela, Brazil E. meyeri Schulz semi- 8 emergent aquatic, annual x rare, s Para guay, Ar- homostylous? gentina Distribution. “The three diploids of Eichhorn- ia have narrower distributions than the tetraploids (Table 1) have. In particular, E. paradoxa and F. meyeri are rare with relatively few known localities. 'The distribution patterns of the three diploids sup- port Schwartz's (1927) view that the species are phylogenetically old and that their current distri- butions may represent relict areas of previously wider ranges. The remaining New World Eich- hornia species all have widespread distributions throughout lowland South America, Central Amer- ica, and parts of the Caribbean. Unlike the diploids, which are annual or short-lived perennial emergent aquatics with no clonal propagation, all d exhibit extensive lateral growth either throug branching internode system or by stolons. Frag- mentation facilitates clonal regeneration, and all tetraploid species, particularly Æ. crassipes, ar capable of prolific vegetative spread. The ability to disperse by seeds and floating vegetative frag- ments may contribute to the wider distributions of tetraploid than diploid taxa. The Pontederiaceae are of New World origin with all but one species of Eichhornia native to the Neotropics. The African Eichhornia natans is likely descended from New World ancestors fol- lowing long-distance dispersal. Its homostylous breeding system would favor establishment and subsequent spread in Africa following long-distance dispersal from the New World. The most likely progenitor of E. natans is E. diversifolia, which is very similar in appearance and has been con sidered conspecific by some taxonomists. The small seeds of Eichhornia species and their occurrence in habitats frequented by migratory water birds provide opportunities for long-distance dispersal. This may account for the disjunct and scattered distributions of many of the species. Ecology and life history. There have been no detailed ecological studies of the life TUA demography, or habitat preferences of Eichho species in their native ranges. The following feld observations, while rudimentary, may stimulate more in-depth work as well as providing the nec- essary ecological background from which to discuss the reproductive biology and evolution of breeding systems in the genus. Eichhornia species are exclusively freshwater aquatics. They occupy a diversity of wetland hab- itats ranging from large water bodies, such as major river systems, lakes, and reservoirs, to extensive marshlands, seasonal pools, and low-lying pastures. Several species are also capable of colonizing sites 744 Annals of the Missouri Botanical Garden disturbed by man, such as rice fields, irrigation canals, and drainage ditches. It is not uncommon to find two or three species of Kichhornia at the same site. Significant features of aquatic habitats that determine the presence of individual taxa ap- pear to be permanency of the habitat, water level fluctuations, and overall water depth. These factors not only influence the type of regeneration strategy employed by Eichhornia species, but also the com- position of the aquatic community and hence the degree of interspecific competition. Eichhornia azurea is a large, long-lived, mat- forming perennial which most commonly occurs in permanent water bodies such as rivers, lakes, and extensive marshlands. The long duration of its pre- reproductive phase restricts the species from col- onizing habitats subject to seasonal desiccation. The mat-forming Eichhornia crassipes possesses a sim- ilar ecology, although it is capable of reaching reproductive maturity more rapidly (Barrett, 1980a) and is often found in more seasonal environments. Unlike F. azurea, which is a rooted, floating-leaved aquatic, the free-floating E. crassipes is able to colonize environments that experience large fluc- tuations in water level (Barrett, 1977a, 1979). Eichhornia heterosperma commonly occurs in shallow lakes and ponds. Its abundance in the high- ly seasonal environments of the Llanos of Vene- zuela and in the caatinga of northeast Brazil in- dicates that it is capable of withstanding habitat desiccation, presumably as rhizomes or seed. Eich- hornia paniculata occurs in seasonal pools, rice fields, and low-lying pastures. In comparison with the preceding species, it is capable of colonizing more terrestrial environments that experience only limited periods of inundation. The life history of E. paniculata depends largely on available mois- ture. When available for extended periods, plants perennate; however, at many locations in northeast Brazil and in Jamaica, populations are annual and regenerate from seed (Barrett, 1985a). An annual life history is also implicated for E. meyeri in the seasonal environment of the Chaco of Argentina (Schulz, 1942). wing to the rarity of E. paradoxa, little is known of its ecology. I have observed two popu- lations in northeast Brazil. At one site in Paraiba state the species was growing in a roadside depres- sion with E. paniculata; at the other location it occupied a low-lying floodplain of the Sào Francisco River. These observations suggest that the species prefers seasonally inundated sites. "ichhornia diversifolia is most commonly found in seasonal ponds that experience large water-level fluctuations. During the vegetative growth period it is often in water 1-2 m deep, and as the water level drops, plants flower and mature seed. The species usually behaves as an annual and may require water-level fluctuations for successful seed germination and establishment. Eichhornia diver- sifolia colonizes rice fields and can be considered a fugitive species adapted to temporary aquatic environments. Habitat descriptions on herbarium specimens and flora accounts of E. natans in Africa suggest a similar ecology. The species is reported as a rice field weed in Nigeria (Vaillant, 1967). Reproductive biology. The breeding systems and life histories of Eichhornia species are associated with diverse reproductive attributes (Table 2). In general, the outcrossing species possess large showy flowers, inflorescences contrasting with many flowers, high pollen-ovule ratios, and heavier seeds in comparison with selfing species. It is important to emphasize that each Eichhornia species shows considerable variation in reproduc- tive traits. The values presented in Table 2 are from a single population of each species and were chosen to illustrate overall trends in reproductive traits among species. With the exception of data presented for E. meyeri, all values are based on field-grown plants. Exceptions to the trends dis- cussed above are apparent, however, such as in É. paniculata, where individual flowers are similar in size to several of the selfing taxa. Inflorescences of this species contain many more flowers (up to 300) than other Eichhornia species so that the overall floral display of plants is by no means di- minished because of their smaller flowers. Figure l illustrates the dramatic difference in floral display of tristylous Æ. crassipes and semi-homostylous F. paradoxa. 'TRISTYLY IN THE PONTEDERIACEAE The tristylous syndrome. To understand the evolutionary modifications of tristyly in the Pon- tederiaceae it is necessary to describe the mor- phological, physiological, and functional aspects of trimorphic incompatibility. Populations of tristylous plants are composed of three floral morphs known as the long-, and short-styled morphs (here- after referred to as L, M, S). Each plant possesses flowers with two anther levels that correspond to the stigma levels in the remaining two morphs. Thus, as illustrated in Figure 2, there is a reciprocal positioning of anthers and stigmas in the three floral morphs. Pollinations between anthers and stigmas mid-, of equivalent height result in seed set and are termed “legitimate” following Darwin (1877). The remaining pollinations are referred to as “illegiti- mate” and result in little or no seed set. A unique feature of tristylous plants is the pro- Volume 75, Number 3 1988 Barrett 745 Breeding Systems in Eichhornia Repr di e attribute f New World species of Eichhornia. Values are the mean for a pee TA bonalsiion sample ku species sampled from the native range. Eichhornia meyeri glasshouse-grown pla Flower Number of Breadth Flowers per Ovule Number Pollen-Ovule Seed Weight Taxon (in mm) Inflorescence per Flower Ratio (mg) E. azurea 44.0 46.2 169.7 204.0 1.062 E. crassipes 58.2 17.2 150.0 255.2 0.297 E. paniculata 24.0 82.0 109.7 192.2 0.147 E. heterosperma 16.1 6.8 134.0 60.7 1.104 E. diversifolia 20.6 3.8 225.5 43.2 0.074 E. paradoxa 17.6 2.0 172.8 37.9 0.083 E. meyeri 16.9 26.0 273.1 18.8 0.281 duction, by the two anther levels within a flower, of distinct pollen phenotypes that differ in their size iud incompatible pone vor. Associated with pollen the amount of pollen produced by the three stamen levels. Long-level anthers produce small numbers of large- sized pollen grains; mid-level anthers produce in- termediate amounts of mid-sized pollen grains; and short-sized anthers produce large numbers of small- sized pollen grains. Several other floral polymor- phisms often accompany morphism (e.g., stigmatic papillae length, style coloration, pollen exine sculpturing). These traits the stamen-style tri- often vary among taxa in their occurrence and expression. Genetic studies of the inheritance of tristyly in the three tristylous families indicate that the most common mode of control is by two diallelic loci (5, M) with S epistatic to M (Fig. 2). With this genetic control and legitimate mating among the morphs, an isoplethic equilibrium (1:1: 1) is the only pos- sible condition in large populations, provided that the morphs are of equal fitness (Fisher, 1941, 1944; Heuch, 1979). This expectation provides a "standard" for a fully functional tristylous system, and surveys of style morph frequencies in natural populations can be viewed as the logical starting point for studies directed toward understanding the causes of modification in tristylous systems. In no species of Eichhornia or Pontederia, the two tristylous genera of Pontederiaceae, does the expression of tristyly conform to all of the features described above. The tristylous syndrome of Pon- tederia most closely resembles this hypothetical state with the major departure involving the self- incompatibility system. In the four taxa of Pon- tederia that have been examined experimentally the M morph is moderately self-compatible when pollinated with pollen from long-level anthers, whereas incompatibility expression in the L and S morphs is considerably stronger (Ornduff, 1966; Barrett, 1977b; Glover & Barrett, 1983; Barrett & Anderson, 1985). The significance of this vari- ation to the functioning of tristyly is unclear, since it is not associated with modifications in the sta- men-style polymorphism or features of pollen tri- morphism. While trimorphic incompatibility is a stable fea- ture of populations of Pontederia species (see Price & Barrett, 1982; Barrett et al., 1983), this is not the case in Eichhornia. Only in E. azurea do populations occur that possess self-incompatibility, strong pollen trimorphism, and the three style morphs (Barrett, 1978). Even in this species, how- ever, monomorphic populations are common, and genetic modifications favoring self-fertilization oc- cur in Central America. In the remaining tristylous taxa, self-compatibility occurs and is associated with weak heteromorphisms of pollen size. Differ- ences in the degree of pollen heteromorphism among Eichhornia species are illustrated in Figure 3. The greatest difference in the mean size of pollen orig- inating from the two anther levels within a flower occurs in the self-incompatible E. azurea. Differ- ences are less evident in the two self-compatible tristylous species, Æ. crassipes and E. paniculata. In the remaining homostylous species there is evi- dence of slight differences in pollen size distribu- tions in E. heterosperma and E. diversifolia, whereas in E. paradoxa pollen originating from the two anther levels is uniform in size. Self-compatible tristyly. Im the vast majority of heterostylous plants the floral polymorphisms are associated with a sporophytically controlled self-incompatibility system (Ganders, 1979). Ab- sence of such a system in most Fichhornia species raises the issue as to whether illegitimate matings are frequent under field conditions and whether the stamen-style polymorphism functioning alone is effective at promoting disassortative mating among the floral morphs. Two studies involving 746 Annals of the Missouri Botanical Garden FIGURE l. Floral displays in a tristylous and a semi-homostylous species of Eichhornia.— A. Eichhornia crassipes, Š morph from Boca de Jari, Lower Amazon, Brazil. —B. E. paradoxa, semi-homostylous M morph from Propriá, Sergipe, northeast Brazil. Volume 75, Number 3 1988 Barrett 747 Breeding Systems in Eichhornia S > «4 ssmm ssMm SsMm ssMM SsMM smm Style length genotypes FIGURE 2. pecies. Legitimate pollinations are indica Schematic diagram of the Sey uan of styles and stamens in the floral morphs of a tristylous ows. Genotypes of the floral morphs under the two-locus s model (S, M) for the inheritance Hx e in Eichhornia are shown marker genes indicate that the mating system of floral morphs in self-compatible tristylous popula- tions can involve a high level of outcrossing (Glover & Barrett, 1986a; Barrett et al., 1987). Progeny test data from E. paniculata (Table 3) reveal that the floral morphs are largely outcrossing and that most matings are disassortative in nature. This indicates that self and intramorph matings occur infrequently in this population. Such an effect could result from the operation of a cryptic heteromor- phic incompatibility system of the type demon- strated in Amsinckia (Weller & Ornduff, 1977; Casper et al., 1988). However, this does not appear to be the case. In a controlled pollination experi- ment on E. paniculata that compared the com- petitive ability of self, illegitimately outcrossed, and legitimately outcrossed pollen, using the GOT-3 marker locus, Glover & Barrett (1986a) found no evidence of a residual or cryptic trimorphic incom- patibility system. Outcrossed progeny were always favored over selfed progeny irrespective of whether they were legitimate or illegitimate. The results from E. paniculata are relevant to the question of whether self-compatibility in tristy- lous Eichhornia species is a derived condition in- volving relaxation of self-incompatibility. Following this view we might have anticipated some residual influence of the ancestral incompatibility system favoring legitimate over illegitimate matings. How- ever, this was not detected, perhaps because the action of numerous modifier genes has removed all traces of the functional aspects of trimorphic in- compatibility. The prepotency of cross-pollen over self-pollen might then be explained as a manifes- tation of inbreeding depression operating via ex- tensive pollen-pistil interactions and genetically unrelated to the original incompatibility system (Barrett, 1988b). Comparative evidence from other heterostylous families indicates that self-compatibility is com- monly derived from self-incompatibility (Ganders, 1979). Most workers (e.g., Charlesworth Charlesworth, 1979) have favored the idea that the evolution of self-incompatibility precedes the development of the stamen-style polymorphism in heterostylous groups (although see Richards, 1986). Function of tristyly. Although the progeny test data from Recife demonstrate high levels of disassortative mating in E. paniculata, this does 748 Annals of the Missouri Botanical Garden Eichhornia azurea (2n=32) Eichhornia diversifolia (2n=30) 40 40 30 r 30 20 F 20 Pp 10 H IO F o l l l fe] l l 50 60 70 80 90 100 50 60 70 80 90 100 P Eichhornia crassipes (2n=32) ¡AA heterosperma (2n=30) 4 FREQUENCY m o T o 30 - lor - o l | T 50 "m T 60 TO 80 90 100 Eichhornia paniculata (2n=16) Eichhornia paradoxa (2n=16) 40 30 30 20 20 10 IO (0) l l o 1 l l 50 60 70 80 90 100 50 6 70 80 90 100 POLLEN SIZE (microns) IGURE 3. Patterns of pollen-size heteromorphism in six species of Eichhornia d contrasting breedin d P P P ng ;U systems. The distributions for each species were obtained by measuring the equatorial a is of 200 dry pollen grains ERE from the two stamen levels Ed a bower To facilitate TOMIparisóns am pt species, all mea- surements were conducted on the M morph (E. azurea, E. crassipes, E. paniculata) or semi-homostylous derivatives of the M morph (E. e E. heterosperma, E para "bug Note the overall differences in pollen size between diploid and nd specie not prove that tristyly promotes legitimate polli- ever, stigmatic pollen loads have been examined nation, as Darwin (1877) originally proposed. This in Pontederia species, where size trimorphism is is because in a self-compatible species intermorph well developed, and significant levels of legitimate illegitimate matings contribute towards estimates pollination were recorded in several natural pop- of disassortative mating. Using the progeny test ulations (Glover € Barrett, 1983, 1986b; Price method, it is not possible to distinguish between & Barrett, 1984; Barrett € Glover, 1985). These cross-pollinations involving legitimate and illegiti- — studies therefore provide support for the Darwinian mate pollen that originate from the same plant hypothesis of the adaptive significance of tristyly. (Barrett et al., . However, in some heter- ostylous species, these pollen types differ in size, Daep o eS: and it is therefore possible to measure the m: nitude of legitimate and illegitimate pollination in Disruption of population structure. natural populations by inspecting the pollen loads styly to function effectively, populations should of open- qoum stigmas (Ganders, 1979). The contain the three floral morphs and provide suffi- extensiv lap in the size of pollen that originates cient pollen and nectar rewards to attract special- from diferent nd levels of self-compatible Fich- — ized long-tongued pollinators, usually bees. Several hornia species (Fig. 3) precludes this method. How- — influences can disrupt population structure to yield ` For tri- Volume 75, Number 3 1988 arrett 749 Breeding Systems in Eichhornia populations composed of one or two morphs. In É. crassipes and E. azurea, founder effects and ram- pant clonal propagation are major disruptive influ- ences on tristyly. This is well illustrated by the geographical distribution of floral morphs in F. crassipes (Fig. 4). The S morph is absent from many parts of the New World range as well as from the Old World. In the latter case genotypes of the S morph were not among clones transported to the Old World by man. The M morph predom- inates in most regions, while the L morph appears sporadically (Barrett, 1977). In a survey of 196 sites throughout the New World range of E. cras- sipes, Barrett & Forno (1982) found that 77% of the colonies located were monomorphic for style length, 18.4% were dimorphic, and only 4.6% were trimorphic. The rarity of E. crassipes pop- ulations containing the three floral morphs results from the high dispersal of the free-floating life form coupled with rapid clonal propagation. These fac- tors result in inequalities of representation of found- ing genotypes. In addition, the short-lived nature of many populations and ecological restrictions on seedling establishment (Barrett, 1980a, b) retard further progress toward isoplethic population struc- ture. Founder events have also played a disruptive effect on the maintenance of tristyly in E. pa- niculata. Although clonal propagation is absent from this species, repeated colonizing episodes and stochastic influences on population size play a ma- jor role in determining morph frequencies in pop- ulations. Surveys of population structure in north- east Brazil and Jamaica (Barrett, 1985b and unpubl. data) have revealed a pattern reminiscent of that found in E. crassipes (Table 4). The S morph is absent from Jamaica and underrepresented in many populations from Brazil. The M morph predomi- nates in dimorphic populations and, with one ex- ception (see below), is the only morph that has been observed in monomorphic populations. Absence of the S morph from parts of the range of E. crassipes and E. paniculata and its under- representation in Brazilian populations of E. pa- niculata may be explained by founder events and fluctuations in population size. Since both species are highly self-compatible, polymorphic popula- tions can arise from selfing and segregation of genotypes heterozygous at the S and M loci. How- ever, since the dominant 5 allele governing the expression of short styles is only carried by the S morph, separate introduction(s) of this morph are necessary for it to become established in a region. In contrast, the m allele can be carried by all three morphs and the M allele by the M and S morphs. TABLE 3. Estimates of the mating system of floral morphs in a tristylous population (B5) of Eichhornia paniculata at Recife, northeast Brazil. Outcrossing rates and levels of disassortative mating were estimated from open-pollinated progeny arrays using six isozyme loci or the two loci governing style length, respectively. After Glover & Barrett (1986a) and Barrett et al. (1987) Floral Morphs Measurement L M S Outcrossing rate 0.98 0.977 0.93 N 480 520 480 Standard error 0.021 0.018 0.022 Disassortative matin 0.90 0.93 0.83 652 721 665 Standard error 0.034 0.052 0.052 Computer simulation studies by Heuch (1980) on the effects of random fluctuations of population size in tristylous systems confirm that the S morph is most often lost from populations. The same pro- cesses, on a neighborhood scale, may also account for the low average frequency of the S morph in trimorphic populations of E. paniculata (Table 4). Thus a genetic constraint imposed by the inheri- tance of tristyly interacting with random ecological processes plays a major role in disrupting the main- tenance of population trimorphism. Disruptions of trimorphic population structure in Eichhornia may not necessarily lead to genetic modifications of the breeding system. In E. azurea and E. crassipes, clonal regeneration and the low frequency of sexual reproduction in many mono- morphic and dimorphic populations limit oppor- tunities for evolutionary change. However, under conditions where sexual reproduction is favored, e.g., in strongly seasonal aquatic environments, there may be strong selection pressures to increase fecundity, particularly if pollinating agents are scarce (Barrett, 1979). Semi-homostyle formation. In each of the three tristylous species of Eichhornia there is evi- dence of the breakdown of tristyly and the evolution of semi-homostyly (Barrett, 1978, 1979, 1985a). Semi-homostyles in Eichhornia occur primarily at the geographical margins of the neotropical ranges of species either as local populations (E. azurea, E. paniculata) or as variants in otherwise unmod- ified population systems (E. crassipes). In each species, semi-homostyles are highly autogamous be- 750 Annals of the Missouri Botanical Garden DISTRIBUTION [U] Native Adventive — Larger letter denotes predominant morph The geographical vrais of style morphs in Eichhornia crassipes. Modified from Barrett 82). (1977a) and Barrett & Forno (19 cause of the close juxtaposition of stigmas and anthers. In £. azurea and E. paniculata, semi-homosty- lous populations differ from their outcrossing pro- genitors in several floral traits. They often possess smaller, less showy flowers, reduced numbers of flowers per inflorescence, weak pollen hetero- morphism, and lower pollen-ovule ratios. As de- scribed above, these differences parallel those that distinguish outcrossing and selfing species of Fich- hornia (Table 2). This point is well illustrated by comparing the degree of pollen heteromorphism displayed by populations of E. paniculata with contrasting mating systems (Fig. 5). The difference in size of pollen produced by contrasting stamen levels is large in tristylous populations, whereas dimorphic and monomorphic populations show in- creasing pollen-size overlap. The weakening of pol- len heteromorphism accompanying the evolution of semi-homostyly probably reflects relaxed selec- tion pressures and the random accumulation of small mutations affecting pollen size. These pat- terns suggest that pollen size is a canalized trait in outcrossing populations and that heteromorphism is maintained by strong stabilizing selection. The functional significance of pollen-size heteromorph- ism to pollen-pistil interaction in heterostylous species is still a matter of some conjecture (see Ganders, 1979) emi-homostylous variants are most common in E. paniculata. They predominate in Jamaica and are frequently encountered in dimorphic and mono- morphic populations in northeast Brazil. In tristy- lous populations, however, they occur rarely, suggesting some selective disadvantage. The development of semi-homostyly in E. paniculata occurs most commonly in the M morph (Fig. 6), with populations displaying various floral modifi- cations in different parts of the range of the species (Barrett, 1985a). This suggests that the breakdown of S is | occurring repeatedly in the species. Semi variants in £. paniculata are not exclusively derived from the M morph. Occasion- ally, modified L plants are encountered in popu- lations from northeast Brazil, and material from TABLE 4. Average frequencies of the floral morphs in populations of Eichhornia paniculata from northeast Brazil and Jamaica. The majori o LÀ M s in now level stamens (see text). Number Frequencies Sampled L M S Region Northeast Brazil Trimorphic 58 0.374 0.370 0.256 Dimorphic 21 0.336 0.664 — Monomorphic 5 — 1.00 — Jamaica Dimorphic n 0.211 0.789 — Monomorphic 19 1.00 — Volume 75, Number 3 Barrett 751 1988 Breeding Systems in Eichhornia 30 l. x= 56-4um x = 73-2ym | 2. x :56-.6ym x =71-9um 20 } = IO P = 30 3 x = 59-Ium x =70:8 pm | 4, Xx =60:9um x=70-6um z > 10 = = o z F F u 2 o 30 E l5 K=6l-Sym Xx =-68:4 um 6 xX:60.9ym X=66:6pm u. 30 35 40 45 50 54 30 35 40 45 50 54 MICROMETER UNITS FIGURE 5. breeding systems. The distributions for each population were obtained by measuring the UNIT = 1-538 p Patterns of pollen-size heteromorphism in six populations of Eichhornia paniculata with contrasting equatorial axis of 200 dry pollen grains originating from the two stamen levels ofa flower. To facilitate EDU among populations, all measurements wer e conducted on six flowers from six plants of the M morph in each population. Plants o, the M morph in nontristylous populations possessed varying degrees of genetic E baton of the short stamen level (see text for details) . Populations 1 Populations 1 and 2 are trimorphic, 3 is dimorphic (L, M), an the only reported occurrence of the species in Central America (Haynes 8442, 8603, (ALU) Rio Las Lajas, Department of Rivas, Nicaragua) is com- posed exclusively of semi-homostylous long-styled plants. Ongoing genetic and developmental studies (Richards & Barrett, 1984 and unpubl. data) of the range of semi-homostylous variants in É. niculata are aimed at determining the inheritance patterns and developmental pathways responsible for the breakdown of tristyly in the species In E. crassipes semi-homostyle ai is ap- parently rare. I have observed semi-homostyles in only two populations. Both were dimorphic (L, M), and in each case the modified phenotypes only differed from unmodified forms in the relative lengths of their reproductive parts (Barrett, 1979 and unpubl. data). If these phenotypes became reproductively isolated from their unmodified pro- genitors, it is likely that genetic modifications in other aspects of their reproductive biology would occur. —4 are from northeast Brazil, populations 5 and 6 are from Jamaica. ic (M). d 4—6 are monomorphic The reported semi-homostyles in a population of E. crassipes from Costa Rica are also modifie forms of the M morph (Barrett, 1979). However, in contrast to E. paniculata, where alterations involve elongation of short-level stamens, the breakdown of herkogamy in this population of E. crassipes is primarily the result of shortening of long-level stamens. This indicates that a different developmental pathway is involved. In addition, the recent discovery in E. crassipes of a semi-ho- mostylous L morph in northeast Brazil (S. C. H. Barrett, unpubl. data) indicates that, in common with E. paniculata, semi-homostyle formation can occur in both M and L morphs. The arrangement of reproductive organs in semi-homostylous populations of Æ. azurea suggests that they are M variants with elongated short-level stamens (Barrett, 1978 and unpubl. data). Why the M morph of Eichhornia species appears to be more prone than other morphs to genetic modifi- cations favoring increased levels of self-fertilization 752 Annals of the Missouri Botanical Garden JAMAICA (Dimorphic) M M JAMAICA (Monomorphic) 15mm FIGURE 6. L, M, and S morphs from an outcrossing maica B nid hic rt-level stamens (1 ve (1985a) and Glover & pra (198€ is not clear but may be associated with differences in the development and floral architecture of the morphs. In the Lythraceae and Oxalidaceae, semi- homostyly also appears to occur more commonly in the M morph (Stout, 1925; Ornduff, 1972). The establishment enla in tristylous species of Eichhornia demonstrates that when se- lection pressures that maintain outcrossing are re- laxed or changed in direction, the complex syn- drome of traits that constitute the tristylous syndrome can rapidly break down towards in- The evolutionary breakdown of tristyly to semi- gir ci ld in 5 paniculata. Flowers of the trimorphic inu d Paa : J15, ip ied ie is 3) that are adjacen iini occur in oe perit an from Jamaica. For further details see Barrett ra). . No stamen creased self-fertilization. Furthermore, the wide range of semi-homostylous variants that occur in E. paniculata suggests that the processes respon- sible for the dissolution of floral polymorphism are still active in contemporary populations. This pro- vides an opportunity to investigate directly the selection pressures acting on the mating system and to determine the ecological and genetic con- ditions that favor the evolution of self-fertilization. Fichhornia paniculata is particularly favorable for such studies because it possesses a short life Volume 75, Number 3 1988 Barrett 753 Breeding Systems in Eichhornia TRIMORPHIC Loss of BED S allele L M S Outbreeding FIGURE 7. predominant matings. Note the modifications in shor monomorphic populations. After Barrett (1985b cycle, and population turnover is rapid due to the unpredictable nature of its habitats. These features increase the likelihood of detecting microevolu- tionary changes. This situation contrasts with that in E. azurea and E. crassipes where, because of heavy emphasis on clonal propagation, evolution- ary changes in the breeding system of populations are more difficult to measure. The breakdown process in Eichhornia panicu- lata. Surveys of the patterns of floral-morph fre- quency in populations of E. paniculata in con- junction with studies of their reproductive ecology and genetics have enabled the formulation of a model of the breakdown process (Barrett, 1985b; Glover & Barrett, 1986a, 1987). Figure 7 illus- trates the major stages in the breakdown of tristyly to semi-homostyly in the M morph of E. panicu- lata. The model emphasizes two key stages: loss of the S allele and hence the S morph, and second, loss of the m allele and thus the L morph. Stochastic influences on population size, as discussed above, are thought to be largely responsible for the dis- appearance of the S morph from populations. How- ever, because of its concealed stigma, a loss of specialized long-tongued pollinators in small pop- ulations may also reduce the maternal fitness of the S morph and hence its representation in found- ing populations. Comparisons of the fecundity of floral morphs in populations serviced by either long- tongued bees (Florilegus festivus and Ancyloscelis spp.) or generalist bees (Trigona spp. and Apis mellifera) one evidence in support of this sug- gestion (S. C. H. Barrett, unpubl. data). Loss of the L um and fixation of the M morph DIMORPHIC MONOMORPHIC ] Loss of Y TULS m allele Q M L M m a Increased selfing MI i» Model of the breakdown of tristyly to semi- E in Eichhornia paniculata. Arrows indicate Mm stamen position of the orph in dimorphic and in populations of E. paniculata are associated with the spread of mating system modifier genes and the evolution of semihomostyly. The genes that modify the short-level stamens of the M morph have no significant phenotypic effects when carried by the L and S morphs. As a result, in dimorphic populations, plants of the M morph frequently dis- play altered stamen positions, whereas the L morph remains unmodified (e.g., fig. 3 in Glover & Bar- rett, 1986a, and Fig. 6). This pattern is also evident in E. crassipes (e.g., fig. 2 in Barrett, 1979). This phenotypic difference between the floral morphs of E. paniculata has a profound effect on their mating systems. Unlike tristylous populations where each morph is highly outcrossed, in dimorphic popula- tions the M morph often experiences a high level of self-fertilization, whereas the L morph remains largely outcrossing (Glover & Barrett, 1986a; Bar- rett et al., 1987; Barrett, 1988a). With this mating asymmetry, and no major fitness differences be- tween progeny arising from them, the M morph will likely replace the L morph. This is because the genes that cause an increased rate of self-fertiliza- tion have an “automatic advantage," since the maternal parent will transmit genes via both pollen and ovules to selfed progeny and thus evade the “cost of meiosis” (Maynard Smith, 1978). An additional advantage that modified M plants possess over other floral morphs is their facility for automatic self-pollination in the absence of polli- nating agents. Frequent colonizing episodes result- ing in periods of low density are more likely to favor establishment of semi-homostylous variants, and this likely accounts for the predominance of populations monomorphic for the M morph in Ja- Annals of the 754 Missouri Botanical Garden 'snoungojnp AY TY aid pup su2umis j2a2]-11048 paytpow nqryxo nomumf ui ydiow jy ay) fo squo¡d NF “2861 PUP 'Fg61I “6261 Kapnnuvf ui pajonpuoo 242m s&aaung -"Domurmf woif suonpjndod vie[notued eruoquorgq 4? uo1nqiusip yd 1011-9475 fo ULINDA 'g ANNI Tem cá ` NENS `. `X \ \ - uopuaJe[?) ` 3Uut13Qu]£7) JUIPG i N \ \ \ Ó` I ain "« E N l FN I I AL. ed $81j8UJ0|! F——— Oc Ol 0 se|u oz OL 0 VOIVINVÍ \ \ qieqezug weg ; | 84 j M. ae Puz|əjíounsəAA __ Volume 75, Number 3 1988 Barrett 755 Breeding Systems in Eichhornia maica (Fig. 8). The fecundity advantage of semi- homostylous variants of the M morph over the L morph has been demonstrated in Jamaican popu- lations of E. paniculata (Barrett, 1988a) as well as in a dimorphic (L, M) population of E. crassipes (Barrett, 1979). The reproductive assurance o homostyly under conditions of low density and un- certain pollinator service has been used to explain the geographic patterns of floral variation in several other heterostylous groups (Baker, 1953, 1959; Ganders, 1975; Barrett & Shore, 1987). FLORAL HETEROMORPHISMS IN SELFING TAXA A second source of evidence in support of the derived nature of selfing in Eichhornia comes from investigation of the patterns of variation in floral traits within and among populations of predomi- nantly autogamous taxa. As we have seen, in mono- morphic and dimorphic populations of E. panic- ulata, residual floral heteromorphisms that were originally components of the ancestral tristylous syndrome remain in populations despite their large- ly selfed mating systems. Although the expression of these traits may be considerably modified from their original form, their occurrence in selfing species of Eichhornia is direct evidence that these taxa are descended from tristylous ancestors via evolutionary breakdown of tristyly. Intrapopulation variation. Over most of the range of E. heterosperma and E. diversifolia, pop- ulations are composed of a single floral phenotype with a mid-length style and one set of anthers positioned just above the stigma and another below (Figs. 9 and 10, respectively). Because flower size is reduced in both species, the distance separating the reproductive organ levels is often small. In some populations this makes it difficult to determine the homologous positions of an ancestral tristylous condition and thereby infer the morphs from which the phenotypes descended. In most populations the phenotypes are best interpreted as semi-homosty- lous M plants in which elongation of the short stamen level into the mid position has occurred. Although this origin seems most plausible, based on analogy with semi-homostyle formation in out- crossing Eichhornia species, the possibility that some phenotypes are modified S plants with elon- gated styles cannot be ruled out. Field studies of E. heterosperma in Venezuela and E. diversifolia in northeast Brazil have re- vealed a different pattern of floral variation. In both species a second floral phenotype can be found in populations in addition to the phenotype de- scribed above. The two phenotypes differ in style length, style coloration, pollen size, and the relative positions of their reproductive parts (Table 5). The expression of traits in the second phenotype indi- cates that it is a semi-homostyle derived from the L morph. The two semi-homostyles in E. diversi- folia are bip in P ure 10. A trait of par- ticular g the origin of these floral nad is style toleration. In tristylous Eich- hornia species the three floral morphs differ in the degree of pigmentation of their styles. For example, in E. crassipes the L morph has a purple style, the M morph a lavender style, and the S morph a white style (Barrett, 1977a). In E. diversifolia and E. heterosperma the styles of the two semi-homo- styles are pigmented to different degrees with the semi-homostylous L phenotype being either purple (E. diversifolia) or pink (E. heterosperma) and the semi-homostylous M phenotype light pink or white. In £. heterosperma, the two morphs can also differ in perianth color, with the semi-homostylous L phe- notype possessing dark blue tepals and the M phe- notype pale blue tepals. Differences in perianth color among the floral morphs of Eichhornia cras- sipes have also been reported, although this is not a universal feature of the species throughout its range (Müller, 1883; Haigh, 1936; Barrett, 19777). Populations of E. diversifolia and E. hetero- sperma that contain the two semi-homostylous phe- notypes are largely self-pollinating, and it seems unlikely that the residual polymorphisms have any functional significance. If this is true, we might expect that mutation pressure will eventually break down the discontinuities that currently exist be- tween the forms. Further field studies are required, however, to establish the overall distribution pat- terns of the morphs and to determine whether fitness differences that relate to floral phenotype exist. A recent survey of E. diversifolia populations in Ceará, northeast Brazil, provided no evidence that the two semi-homostylous phenotypes exhibit nonrandom distributions (S. C. H. Barrett, unpubl. ata). The rarity of E. paradoxa has restricted our investigations of its floral biology to surveys of herbarium specimens and to experimental studies of two natural populations from Paraiba and Ser- gipe, northeast Brazil. Even in this small sample several distinct patterns have emerged. The two field populations were each composed of a uniform but different self-pollinating semi-homostylous phe- notype. One of these is illustrated in Figure 1B. The arrangement of reproductive parts and style coloration in the phenotypes is similar to the two semi-homostylous forms described above. It there- 756 Annals of the Missouri Botanical Garden M 3 JJ 5mm 4mm FIGUR E 9. Flowers and reproductive organs of semi-homostylous species of Eichhornia.— A. E. heteros erma p g } p (from Ceará, northeast Brazil) .—B. E. meyeri (from Nueva Asuncion, Paraguay). Note the close proximity o stigmas and anthers in the two species. A sec ie aa phenotype occurs in parts of the range of E. heterosperma (see text Volume 75, Number 3 Barrett 757 1988 Breeding Systems in Eichhornia FIGURE 10. Semi- homostylous L and M morphs of Eichhornia diversifolia from a population in Ceará, northeast Brazil "Note the differences in style length and pigmentation in the floral morphs. 758 Annals of the Missouri Botanical Garden TABLES. Mean style lengths and stamen heights (in mm) in semi-homostylous morphs of Eichhornia diversifolia and E. heterosperma. Measurements were made on field-collected flowers fr om populations at northeast Brazil (N — 24 flowers per morph) and Calobozo, Guárico, Venezuela (N — 10 flowers à per iato ; respectively E. diversifolia E. heterosperma Semi-homo- Semi-homo- Semi-homo- Semi-homo- stylous stylous stylous stylous Trait L M L M Mean style length 253 + 0.8 *** 21] + 0.9 23.7 + 0.7 = 22.8 + 0.6 Mean height of upper stamens 241 + 0.9 *** 25.8 + L2 22.4 + 0.7 *** 24,1 + 0.7 Mean height of lower stamens 18.4 + 0.8 ns 18.8 + 1.2 17.7 + 0.8 7 20.0 + 0.8 *** p < 0.001, ** P < 0.01, ns = not significant following Student's t-tests. fore seems reasonable to assume that they are semi-homostylous L and M morphs. This can be confirmed by controlled crosses between the phe- notypes, followed by genetic analysis of F, and F, variation. Herbarium SA tions of E. paradoxa from a population in Bahia (Harley 21401 (K), Bahia, Brazil) appear to contain three floral phenotypes that correspond to the L, M, and S morphs of a tristylous system. These plants possess inflores- cences that contain many large flowers, and it is possible that the population retains a functionally tristylous breeding system. If this is true, E. par- adoxa would resemble E. azurea and E. panicu- lata in displaying both outcrossing and selfing pop- ulation systems. Because of its rarity and markedly disjunct distribution, it is probable that the species possesses considerable interpopulation differentia- tion, with individual populations displaying different stages in the breakdown of tristyly. Further studies of this interesting taxon are planned. Floral monomorphism. Studies of the re- maining two Eichhornia species (E. natans and E. meyeri) have as yet provided no evidence of poly- morphisms in floral traits associated with the het- erostylous syndrome. Both species exhibit mid- length styles and two stamen levels positioned very close to the stigma, one above and one below (Fig. 9B for E. meyeri). Eichhornia natans has highly reduced uniform blue flowers, and it is possible that the species is ” rather than semi-homostylous (see Ornduff, 1972). In quasi- homostyly the close juxtaposition of anthers and stigmas results from the reduced size of flowers rather than from genetic modifications in the rel- ative positions of reproductive parts. Glasshouse studies eyeri material collected from Paraguay (Billiet & Jadin 3211 (BR), Nueva Asunción, Paraguay) indicate that the species is self-compatible, highly autogamous, and without residual pollen-size heteromorphism. The flowers "quasi-homostylous of this collection are illustrated in Figure 9. Further work on both species is required to establish firmly the relationships of their breeding systems to an ancestral tristylous condition. This may be difficult if both species have lost all of the polymorphic variation associated with the tristylous syndrome. In E. natans this could have occurred through a genetic bottleneck during dispersal and establish- ment on the African continent. In É. meyeri pro- gressive extinctions leading to its current rarity may have had a similar effect. CONCLUSIONS Studies of intraspecific and interspecific patterns of variation in the breeding systems of Eichhornia species indicate that tristyly has broken down re- peatedly in the genus to give rise to predominantly selfing population systems. This conclusion rests on two lines of evidence: 1) the evolution of semi- homostylous forms within each of the three pri- marily tristylous species (E. azurea, E. crassipes, E. paniculata), and 2) the occurrence of residual floral heteromorphisms in several of the largely autogamous semi-homostylous species (E. diver- sifolia, E. heterosperma, E. paradoxa). The path- way of evolution from outcrossing to selfing in Eichhornia appears to be the only major shift in breeding system in the genus. This contrasts with the Lythraceae and Oxalidaceae, where, in addition to semi-homostyle formation, stable distylous breeding systems have evolved from tristyly (Mul- cahy, 1964; Ornduff, 1972; Lewis & Rao, 1971; Weller, 1976; Charlesworth, 1979). The selective pressures responsible for the change from outbreeding to inbreeding are always difficult to identify (Jain, 1976); however, ecological and genetic studies of Fichhornia populations have pro- vided some insights into the conditions that foster the breakdown of tristyly. Genetic bottlenecks resulting from long-distance dispersal, as well as colonization of unpredictable habitats with uncer- Volume 75, Number 3 1988 Barrett 759 Breeding Systems in Eichhornia tain moisture regimes play major roles in disrupting the maintenance of tristyly. Two features of Eich- hornia species make them particularly prone to these effects: 1) their small-seeded habit favoring bird dispersal, and 2) their occurrence in tropical aquatic habitats that are susceptible to frequent droughts. These aspects of Eichhornia ecology result in frequent colonizing episodes and periodic fluctuations in population size. Under these influ- ences, ecological and genetic conditions are likely to favor the establishment and spread of selfing enotypes. In small newly founded populations, serviced by unreliable generalist pollinators, semi- homostylous variants are likely to be at a selective advantage because of the reproductive assurance that autonomous self-pollination provides. In ad- dition, the genetic load of populations may be suf- ficiently low, owing to frequent bottlenecks (see Lande & Schemske, 1985), that epr sion may not be a serious obstacle to the spread of semi-homostylous variants. Our discussion of the breakdown of tristyly in Eichhornia has focused primarily on details of the ecology and genetics of natural populations. How- ever, for a comprehensive model of the breakdown process to be obtained, information from mor- phology and development needs to be integrated with studies from population biology. This is of particular importance in heterostylous groups, be- cause the floral morphs can respond differently to selection and drift as a result of their particular floral morphologies and inheritance patterns. While in Eichhornia mutations affecting floral structure and mating system can arise in each morph, the available evidence indicates that the M morph is more susceptible to evolutionary modification. This may be because developmental constraints restrict the range of floral modifications that can occur in the L and S morphs. In addition, the S morph is frequently lost from populations through stochastic processes and may rarely encounter the selection pressures operating in small populations that favor the evolution of self-fertilization. Thus loss of the S morph may simply reflect a genetic constraint imposed by the two-locus system of inheritance of tristyly. While frequent colonizing episodes and ecological radiations into temporary aquatic habi- tats appear to be the major driving forces respon- sible for the evolution of presume ayes in Fich- hornia, constraints imposed b hology and genetics of the polymorphism have guided the na- ture of the floral modifications that have occurred. LITERATURE CITED š Rud G. 1974. El género Eichhornia (Pontederi- ae) en Venezuela. Acta Bot. Ven. 9: 303-310 ALEXANDER, E. J. 1937. Family Pontederiaceae. North . Flora 19: 51-60. BAKER, H. G Dimorphism and monomorphism in the Plumbaginaceae. III. Correlation of rer ical distribution patterns with dimorphism and mon morphism in Limonium. Ann. Bot. (London) 17: 615- 9. The contribution of autecological and genecological studies to our knowledge of the past migrations of plants. Amer. Naturalist 93: 255-272. 1966. The evolution, functioning and break- down of heteromorphic incompatibility systems. I. The Plumbaginaceae. Evolution 20: 349-368 BARRETT, S. C. H. 1977a. Tristyly in Eichhornia cras- x: ae ) Solms (water hyacinth). Biotropica 9: 21 7b. The breeding system of Pontederia ronde L., a tristylous species. New Phytol. 78: 0 r The floral biology of Eichhornia azu- rea (Swartz) Kunth (Pontederiaceae). Aquatic Bot. 5: 21 28. . 1979. The evolutionary breakdown of tristyly in Eichhornia crassipes (Mart.) Solms (water hy- E Evolution 33: 499-510 98 Sexual reproduction in Eichhornia ERN (water TEL I. T of clones from diverse regions. . 17: 101-112 . 1980 ual reproduction in Eichhornia crassipes (water Thus) II. production in natural populations. J. Appl. Ecol. 17. 113-124, 1985a. Floral trimorphism and monomor- phism i in continental and island populations of Eich hornia paniculata (Spreng.) Solms (Poatedorialole), Biol. E Linn. Soc. 25: 41-60. . 1985b. Ecological genetics of breakdown in — Pp. 261- 27 i A Haeck & J. W. Wol- ), Struc and Functioning of Plant Populations m Phenot ah ee Genotypic Variation in Plant Populations. North-Holland, Amsterdam, The Netherlands. 988a. The evolutionary breakdown of het- ¿Wi In: Y. Linhart & J. Bock (editors), Evolu- tionary Ecology of Plants. Westview Press, Boulder, Colorado (in press). 19 . The evolution, maintenance and loss of sif incompatibility systems. In: J. Lovett-Doust & L. Lovett-Doust (editors), Reproductive Strategies of Plants: Per and Strategies. Oxford Univ. Press, Oxford (in p LM. jet 1985. Variation in expres- sion of patio incompatibility in Eu cor- data L. (Pontederiaceae). Theor. Appl. Genet. 70: 355-362. I. W. Forno. 1982. Style morph distribution in New World populations of Eichhornia crassipes (Mart.) Solms-Laubach (water hyacinth). Aquatic Bot. 13: 299-30 & D. 306. E. GLOVER. 1985. e Darwinian hypothesis of the adaptive za o bai of tristyly. Evolution 39: 7 J. S. SHORE. 1987. Variation and evolution of breeding systems in the Turner assortative mating in tristylous ua ee panic- ulata peer ei ces Heredity 58: , S. D. PRICE & J. S. SHORE. 1983. “Male fertility 760 Annals of the Missouri Botanical Garden pe d e structure in tristylous teder a (Pontederiaceae). Evolution 37: 15- 7 CASPER, B. B., L. S. Savicn & S. S. Lee. 1988. Dem- onstration of cryptic incompatibility i in distylous Am- sinckia douglasiana. Evolution 42: 248-253 Miis ANOS, A. Las Pontederiaceae de Brasil. o de Janeiro Jard. Bot. 16: 149-218, pls. 1-18. neo D. The evolution and break- down of m Evolution 33: 489 ARLESWORTH. 1979 Fem, for the evolution of desi. Amer. Naturalist 114: 467- 498. CRowE, L. K. 1964. The evolution of outbreeding in plants. I. The angiosperms. Heredity 19: 435-457. Darwin, C. 1877. The Different SUR of Flowers on Plants of the Same Species. John Murray, London. ECKENWALDER, J. E . C. H. BARRETT. 1986. Phy- logenetic systematics of Pontederiaceae. Syst. Bot. 391. : 373- FISHER, R. A. . The theoretical consequences of polyploid inheritance for dy id- m form of Ly- thrum salicaria. Ann. Eugen. 11: . Allowances for ‘double reduction i in the calculation of genotype frequencies with polysomic inheritance. Ann. Eugen. 12: 169-171 GANDERS, F. R. 1975. Heterostyly, homostyly and fe cundity in aros spectabilis (Boraginaceae). 6-62. Madrono 23: ; The vi tf of heterostyly. New Zea- land J. Bot. 17: 607-63 GLOVER, D. E ` H. em 1983. Trimorphic e in | Mexican populations of Pontederia Md (Pontederiaceae). New Phytol. 95: 8 Variation in the matin system of Eichhornia p Subs Solms (Pontederiaceae). Evolut : 1122-113 & 986b. Stigmatic um M in populations of P ontederia cordata from the southern U.S. Amer. J. Bot. d 1607-1612. Genetic variation in con tinental and island d e of Eic haben panic- ulata Arare Heredity 59: 7-17 GRANT, V. & K. A. Grant. 1965. Klose Pollination in I Phlox Family. Columbia Univ. Press, Yor HaicH, 3 6 1936. Notes on the water hyacinth aed f hornia crassipes Solms) in Ceylon. Ceylon J. 12: 97-108. HeucH, I. 1979. Equilibrium RTE oi heterosty- lous plants. Theor. Pop. Biol. 15: 43- 1980. Loss of om ca in finite populations of the area plant; Lythrum sa- licaria. Hereditas 92: 53 Horn, C. N. oe In: G. Harling & L. Andersson (editar, Flora of Ecuador No. 29. eborg. . The evolution of inbreeding in plants. . Rev. Ecol. Syst. 10: 173-200. TN d R & D. W. ScuEMsKE. 1985. The evolution of self-fertilization and inbreeding depression in plants. I. Genetic models. Evolution 39: 24-40. Lewis, D. & A. N. Rao. 1971. Evolution irapa A and population Dp in Pemphis Fors. Proc. Roy. Soc. London, Ser. B, Biol. Sci. 178: 79-94 LLovp, D. G. 1965. Evolution of self-compatibility and s differentiation i in Leavenworthia (Cruciferae). ntr. Gray Herb. 195: 3-134. s ia J. 1978. The Evolution of Sex. Cam- ridge Univ. Press, Cambridge Moore, D. M. & H. Lewis. 1965. The evolution of m Rea in Clarkia xantiana. Evolution 19: -11 n AHY, D. 1964. The VER biology of Oxalis . J. Bot. 51: 1045-1050 8 Einige poveri Me der "ichhornia crassipes. Kosmos (Stuttgart) 13: 297- ORNDUFF, R. 1966. The breeding system of gere e "1 Bull. Torrey Bot. Clu -4 The breakdown of i incom- "x in OL section Corniculatae. Evolution Pu . D. & s . C. H. BARRETT. 1982. Tristyly in Pontederia cordata L. (Pontederiaceae). Canad. J Bot. 60: & 1984. The function and adaptive significance of tristylyi in Pontederia cordata L. (Pon tederiaceae). Biol. J. Linn. Soc. 21: 315-329 i N, P.H. 1 . A survey of e biology n Onagraceae. New Zealand J. Bot. 17: 575-593. RICHARDS, A. J. Plant Breeding aay Allen Unw x^ London. & € RICHARDS, J. BARRETT. 1984. The Rot bunk oí tristylyi in Fic citó Ps u- lata (Pontederiaceae). Amer. J. Bot. 71: 1347-1363. Rosatti, T. J. 1987. The genera of Po es Me: ^ the Southeastern United States. J. Arnold Arbor. 6 39-71. Samia, A. G. 1942. . Darwiniana 6: SCHWARTZ, O. E. Sy stematik um gers r Pontederiaceen. Bot. Ja s dn 39: 50. 19 Pontederiace na En gler "ys K. Prantl, x Natürlichen Pflanzenfamilien. a 2, 15a: 181- SMITH, A. C 39. Notes on a EA of plants from British Guiana. Lloydia 2: -218. SOLMS-LAUBACH, Te NM A. . De Candolle, Monographiae Phanerogamarum 4: vU ne de la Argen- 501 -535. STEBBINS, C. L. Self fertilization and population variability in the higher plants. Amer. Naturalist 41: 37-35 Flowering Plant Evolution above the Species Level. Belknap, Cambridge, Massachusetts. Srout, A. B. 1925. Studies in Lythrum salicaria. 2. new form s reai in this species. Bull. Torrey Bot. Club 52 85. VAILLANT, A. To. C Chemical control of annual weeds in rice. World Crops 19: 28-44, WELLER, S. G. 1976 Breeding system polymorphism ina Seige hated Spe cies. W em 30: 442-454. & R. ORN 1977. Cryptic self-incompat- zs in d kia grandiflora. mn 31: 47- no e ° WYATT, R Phylogenetic aspects of the evolution of self- pollination; In: L. D. Gottlieb & S. K. Jain (editors), Plant Evolutionary Biology. Chapman & Hall, London (in press). DISTYLY AND Robert Ornduff? MONOMORPHISM IN VILLARSIA (MENYANTHACEAE): SOME EVOLUTIONARY CONSIDERATIONS! ABSTRACT Distyly occurs in four of the five genera of the Mia Voi as do breeding systems such as dioecy, gynodioecy, and homostyly that are believed to be derived from distyly. Most species of the largely Australian genus Villarsia are distylous, but the Western Australian V. albiflora has monomo rphic (nonheterostylous) flowers and appears to be homostylous. Despite this, individuals of this species are self although most j ENEA of the four populations studied are intercompa atible. The incompatibility system of V. albiflora appears us, this species possesses a self-incompatibility system, phi suggesting that the pedi on of V. parnassiifolia, or a species possessing a floral morphology and breeding system ncestral to the distyly ted. that occurs widely in the Menyanthaceae. A scheme for the origin of distyly in Villarsia. is presente The Menyanthaceae consist of five genera of wetland or aquatic herbs. Menyanthes and Fauria (Nephrophyllidium) are monotypic north-temper- ate genera, both of which bear distylous flowers. Nymphoides is widely distributed in tropical and temperate regions of both hemispheres; distyly is common in this large genus, as are derivative breed- ing systems such as dioecy, gynodioecy, and autog- amy associated with homostyly (Ornduff, 1966, 1970b, 1973; Vasudevan Nair, 1975). Liparo- phyllum is a monotypic genus restricted to New Zealand and Tasmania; its flowers are monomor- phic and the species is self-compatible (Ornduff, 1973). Villarsia occurs in southeastern Asia, South Africa, and Australia, with the largest concentra- tion of species in southwestern Western Australia. Several Australian Villarsia species have distylous flowers and an associated self- and intramorph in- compatibility system (Ornduff, 1974, 1982, 1986, 1988; Fig. 1). My recent work has revealed that some species of Villarsia with distylous flowers have various altered incompatibility systems; one f these species, the Western Australian V. par- nassiifolia (Labill.) R. Brown, is discussed in this paper. Another species, V. albiflora F. Muell., is a Western Australian species which has nonhet- erostylous flowers (Fig. 1), a self-incompatibility system, and essentially full intercompatibility among members of a population. This species is thus not the self-compatible homostyle of the type that com- monly occurs in various other predominantly di- stylous genera (Richards, 1986). This paper doc- uments the nature of the incompatibility systems of V. parnassiifolia and V. albiflora and discusses the possible relationships between the unique breed- ing system of V. albiflora and those that occur in other species of Villarsia. Since work on these two species, as well as others of the genus, is still in progress, this report is preliminary. MATERIALS AND METHODS Seeds were collected in 1983 from four popu- lations of Villarsia albiflora in Western Australia: ' Supported in part by National Science Foundation grant INT 83-03072. I thank Stephanie Mayer for ry af my considerable assistance in findings, and Stephen Weller and Gayle Muenchow commenting on a this project, Deborah Charlesworth for pas dens on an early summa early LA of this paper 2 Department of Botany, ee, of California, Berkeley, California 04720, U.S.A ANN. MISSOURI Bor. GARD. 75: 761-767. 1988. 762 Annals of the Missouri Botanical Garden NM ws Z Long-styled Short-styled Villarsia parnassiifolia (9404) X Wi 2% ¿IN SS YF E E 9296 ° 9369 Villarsia albiflora FIGURE 1. Dissections of distylous flowers of Villarsia parnassiifolia and of monomorphic flowers of V. albiflora. 9296, Gnangara Lake, near Perth; 9365, Medina, near Perth; 9369, the Capel-Boyanup area; and 9397, near Mount Chudalup. This sampling en- compasses most of the geographical range of this species, which is diploid (Ornduff & Chuang, in press). At the same time, seeds were also collected from two populations of V. parnassiifolia: 9404, a diploid population near Walpole, and 9413, a tetraploid population at Parry Beach, near Den- mark (Ornduff & Chuang, in press), both occupying central positions in the range of this species along the southern coast of Western Australia. Plants were grown from these seeds in the greenhouses at the University of California, Berkeley. Thirty- eight individuals of V. albiflora and 10 long-styled plants (Longs) and 13 short-styled individuals (Shorts) of V. parnassiifolia were used in the cross- ing program. Each individual in a progeny was assigned a plant number for reference purposes. Each plant was self-pollinated and crossed with as many other individuals in the population as possible during the spring and summer of 1986 and 1987. At least six pollinations for each type of cross were performed. Nearly mature capsules were collected individually in seed envelopes, and the number of seeds counted. Flowers of self- and cross-pollinated plants of both species were collected 24 hours after pollination, the gynoecium excised, mounted, stained, and observed under ultraviolet light to observe the behavior of pollen and pollen tubes following various types of pollinations (using the method of Martin, 1959). Seeds of intra- and in- termorph crosses of V. parnassiifolia (population 9413) were sown in a greenhouse, the seedlings grown to flowering, and the style lengths of each individual recorded. RESULTS All seed-set figures were assigned arbitrarily to one of three categories (Figs. 2, 3). “High” seed production refers to crosses in which all pollinations produced capsules with large numbers of seeds; "intermediate or variable" seed production in- cludes crosses in which seed production was mark- edly lower than the “high” category of that seed parent, or in which some crosses failed to produce seed; and “low” seed production refers to crosses producing few or no seeds. These categories will not be quantified in this paper because of high variances in the first two categories and the small number of pollinations (a minimum of six) con- Volume 75, Number 3 Ornduff 763 Distyly and Monomorphism in Villarsia 1988 Style length of pollen parent A ^, . No L | Style š n length g of seed S h aren p t B r t Seed production @ High O Intermediate or variable O Low FIGURE 2. Seed production following artificial self. and cross-pollinations of the two morphs of the distylous Villarsia parnassiifolia, using two greenhouse-grown pop ulations. The vertical and horizontal series of numbers refer to individual plants in each population. Seed production figures are described in text. ducted for each cross. Additional work is planned to increase the sample sizes. However, I am con- fident that the “high” and “low” categories rep- resent two distinct types of results and provide a meaningful and relatively consistent basis on which to discuss the nature of the breeding systems in the two Villarsia species discussed in this paper. Whether the “intermediate or variable" category represents something other than a procedural ar- tifact awaits additional data. Villarsia parnassiifolia. Longs and Shorts of this species are strongly self-incompatible; only one of the nine selfed Longs and one of the 13 selfed Shorts produced any seeds following selfing (Fig. 2). Intermorph pollinations produced generally high seed-sets: 50 of the 58 Long x Short pollinations and 51 of the 60 Short X Long pollinations pro- duced high seed-sets. Only three of the 40 Long x Long crosses produced a high seed-set, and an additional six of these crosses produced some seeds, indicating a high level of intramorph incompatibility of Longs. In contrast, 43 of the 71 Short x Short crosses produced high seed-sets, and an additional 13 of the crosses produced some seed, indicating a high level of intramorph compatibility of Shorts. Each Short in the two populations was fully com- patible with at least one other Short in that pop- ulation, and a few Shorts were compatible with most other Shorts in the same population. Most successful Short x Short crosses were successful in both directions, and most that failed did so in both directions. No differences were noted in be- havior between the diploid and the tetraploid pop- ulations of V. parnassiifolia. Crosses between four Longs and four Shorts of tetraploid population 9413 all produced both Longs 764 Annals of the Missouri Botanical Garden A k "1245689 1 /Ojejejeleeje 2 |ejo eee ce 4 0000 5 e.c 6 [e)( Je) 8 ee 9 0008 A A 9296 em e eee (Dejeejee QOUWADO Seed production @ High (D Intermediate or variable O Low Seed production following artificial self- and cross-pollinations of the monomorphic Villarsia albiflora, using four greenhouse-grown populations. The vertical and horizontal series of numbers refer to individual plants in each population. and Shorts in each progeny. Five Shorts of this population used in five Short x Short crosses pro- duced Longs only in one progeny, Shorts only in two progenies, and Longs and Shorts in two prog- enies. Two 9413 Shorts selfed produced Longs and Shorts in each progeny. Two Long x Long crosses produced only Longs in their progenies. Although the progeny sizes are small, these results make it likely that the Shorts carry a dominant allele and Longs are homozygous recessive. Villarsia albiflora. Results of pollinations us- ing 38 individuals in four populations of V. albi- flora provided relatively consistent results. All in- dividuals proved to be self-incompatible, producing few or no seeds upon self-pollination (Fig. 3). Pollen grains on selfed stigmas either failed to germinate or germinated with growth of the pollen tube into stigmatic tissue but not further. Most individuals produced high seed-sets when crossed with other individuals in the population. Most crosses that Volume 75, Number 3 1988 Ornduff Distyly and Monomorphism in Villarsia failed in one direction were successful in the other direction, or provided seed-sets in the diate or variable" category in that direction. Only eet interme- one instance was found of apparent bilateral in- compatibility (between plants 4 and 16 of popu- lation 9369). Th albiflora has a pronounced self-incompatibility sys- tem, but most members of each population are us, the nonheterostylous Villarsia intercompatible. DISCUSSION Villarsia parnassiifolia has morphologically di- stylous flowers and a strong self-incompatibility system. Although Long x Long crosses generally failed to produce seeds, most Short x Short crosses were fully compatible, producing high seed-sets. No differences were observed in the behavior of diploids and tetraploids. Self- and intramorph in- compatibility are commonly associated with distyly, and in most examples where intramorph compat- ibility exists, as in some species of Hedyotis (Ru- biaceae), Melochia (Sterculiaceae), and Amsinckia (Boraginaceae), it is associated with self-compati- bility as well (Ganders, 1979). The occurrence of self-incompatibility but in- tramorph compatibility in distylous species is rare. It has been reported in the borages Anchusa hy- brida Ten. (Dulberger, 1970) and A. officinalis L. (Philipp & Schou, 1981). I am reluctant to consider the examples of Narcissus tazetta L. and Mirabilis froebelii (Behr) Greene cited by these authors to represent heterostyly. In Anchusa of- ficinalis, Schou & Philipp (1984) demonstrated that the morphological features of distyly are con- trolled by a single diallelic locus, with Longs homo- zygous recessive and Shorts with one or two dom- inant alleles, which is the common genetic basis of distyly (Ganders, 1979). However, as Dulberger (1970) suggested is the case for Anchusa hybrida, the incompatibility system of A. officinalis is con- trolled by at least two alleles, and these segregate independently from those controlling the morpho- logical features of distyly (Schou & Philipp, 1984). The condition described in Villarsia parnassü- folia resembles that of the two Anchusa species but differs in that the Long but not the Short morph of V. parnassiifolia possesses intramorph- as well as self-incompatibility. It would appear that in the Long morph of this species the incompatibility al- leles are linked to the “morphological” locus, but in the Short morph they are not. How this is ac- complished (if it is) is unclear. Style lengths of the few small progenies obtained by self- and intramorph pollinations of the tetra- ploid population of Villarsia parnassiifolia are consistent with the notion that the Short morph of this species carries a dominant allele with Longs thus homozygous recessive, but exact interpreta- tion of the scant data may be obscured by tetra- somic inheritance. Short-dominated morph ratios occur in five of the eight field populations sampled of this species (Ornduff, 1986; P = <0.05 with the Wilcoxon’s signed-ranks test), suggesting that under field conditions Short x Short crosses com- monly participate in contributing to the composi- tion of natural populations. The floral morphology of Villarsia albiflora sug- gests that it is a homostyle, but this tentative con- clusion requires examination. In genera or families with both distyly and homostyly, the homostylous condition is usually viewed as the result of genetic recombination in the S *'supergene," leading to production of flowers combining carpel characters of one morph with stamen characters of the other morph. Commonly, such homostyles are Long. homostyles, although Short-homostyles are also known. Homostyly has been recorded in diverse genera such as Armeria (Plumbaginaceae; Baker, 1966), Gelsemium (Loganiaceae; Ornduff, 1970c), Limonium (Plumbaginaceae; Baker, 1953), Nym- phoides (Menyanthaceae; Ornduff, 19702), Ol- denlandia (Rubiaceae; Bir Bahadur, 1970), Pi- riqueta (Turneraceae; Ornduff, 1970a), Primula (Primulaceae; Darwin, 1877; Ernst, 1955), Tur- nera (Turneraceae; Urban, 1883; Barrett & Shore, 1987), Villarsia (Menyanthaceae; Ornduff, 1974), and Waltheria (Sterculiaceae, Bir Bahadur, 1977). Because a single homostyle flower usually bears a combination of pollen of one morph with carpels of the other morph, such homostyles are generally self-compatible and sometimes largely autogamous. Other types of homostyles occur in Amsinckia (Boraginaceae; Ray & Chisaki, 1957), Hedyotis caerulea (Rubiaceae; Ornduff, 1977), Mitchella repens (Rubiaceae; Ganders, 1975), and Primula (Primulaceae; Ernst, 1955). The nature and place- ment of anthers and stigmas of these homostyles vary, but cannot be attributed to genetic recom- bination alone and must involve the additional ac- tion of modifier genes. Such homostyles may be self-compatible (as in Amsinckia and some species of Primula) or self-incompatible (as in Mitchella repens, Hedyotis caerulea, and some species of Primula). In the latter examples, self-incompatible homostyles appear to be very rare or known only from cultivated material (Ganders, 1975). Once homostyly has developed, even as a result of simple genetic recombination, carpel, stamen, and other floral traits may subsequently be altered by modifier genes to accommodate the homostyly in the direc- tion of greater autogamy (as apparently is the case 766 Annals of the Missouri Botanical Garden in Piriqueta cistoides, Ornduff, 1970a) or in the direction of greater xenogamy (as apparently is the case for some races of Turnera ulmifolia; Barrett, 1988). The nature and origin of the floral monomor- phism of Villarsia albiflora are not clear. If we assume for purposes of discussion that this species represents a recombinant homostyle, its self-incom- patibility and essentially full intercompatibility are the idea that style-length and in- compatibility reactions are not uniformly controlled consistent wit by sets of linked alleles in distylous Villarsia species. However, since incompatibility of Short pollen but not Long pollen appears to be unlinked to style length, this may require that the putative prd styly of V. albiflora is Long-homostyly (i.e., style of Longs combined with anther position M incompatibility behavior of Shorts in one flower) rather than Short-homostyly. Whether this sup- position will survive the scrutiny of further study remains to be seen. second, perhaps more attractive (or at least Serials less c hypothesis is that the mono- morphism of V. albiflora represents a situation in which the flowers of this species are fundamentally Shorts, and in which the Longs of a presumed distylous ancestor have been lost. The anther po- sition and stylar morphology of V. albiflora resem- ble those of Shorts of V. parnassiifolia more closely than they do those of Longs of that species (Fig. 1). Short x Short pollinations of V. parnassiifolia are mostly compatible ones that produce vigorous offspring under artificial and, apparently, natural conditions. It is possible that in the evolution of V. albiflora, a postulated distylous ancestor lost the s allele and thus the Long morph, leading ultimately to a condition where populations consist of true- breeding homozygous Shorts carrying only the dominant S allele. If this postulated ancestor pos- sessed the breeding system and Short-dominated morph ratios characteristic of V. parnassiifolia today, periodic severe reductions in population size as a consequence of the cyclic fluctuations in an- nual rainfall that have characterized southwestern Western Australia since mid or late Tertiary times (Hopper, 1979) might have resulted in the loss of the s allele and thus the loss of Longs. This would result in true-breeding homozygous Shorts as the exclusive components of surviving populations. Since Villarsia typically occurs in mesic to aquatic circumstances, recurrent xerothermic periods could have had strong effects on population sizes and distribution of species in this genus. ecause of the occurrence of distyly in four of the five genera of Menyanthaceae, it is natural to assume that the floral condition and incompatibility system of V. albiflora are derivative ones from a distylous antecedent. A third evolutionary scenario contrary to this directionality in evolution is that the floral monomorphism of V. albiflora is primary in the genus and that distyly elsewhere in Villarsia has been derived from this type of monomorphism. Starting with a self-incompatible nonheterostylous species similar to Villarsia albiflora (which today shows some interpopulation erat in position of stigmas relative to anthers, Fig. 1; Ornduff, 1986), selection might operate: to m increased distance between stigmas and anthers as a means of reducing pollen wastage by selfing. One means of achieving this could be via style-length dimor- phism associated with a slight shifting of anther position (the positions of anthers in the two morphs of the distylous V. parnassiifolia are not ver different; Fig. 1). Initially, the alleles controlling floral dimorphism would be unlinked to those mul- tiple alleles controlling the incompatibility reaction (as in Anchusa). Gradually, linkage between these two sets of alleles would develop with a concomitant decrease in the number of incompatibility alleles; in Villarsia parnassiifolia such a linkage appears to occur in Longs but not in Shorts. Ultimately, the alleles controlling floral dimorphism would be tightly linked to those controlling incompatibility, the latter having been reduced to a pair of alleles at a single locus. At this point, distyly of the con- ventional type that occurs in a variety of unrelated genera would have been achieved. In Villarsia, such “conventional” distyly and incompatibility oc- cur in K. capitata Nees (Ornduff, an congestiflora F. Muell. (Ornduff, 1988) Assuming this latter scheme represents an ap- proximation of the sequence of events in the evo- lutionary history of Villarsia, it would explain an apparent anomaly that I commented on many years ago, namely that V. capitata and V. congestiflora "possess an unusual combination of highly ad- vanced characters" with the "primitive" one of distyly (Ornduff, 1982). If distyly is indeed a con- dition that has developed from monomorphism within Villarsia, this anomaly is resolved, since distyly is thus viewed as advanced and not prim- itive. This last suggested series of events is highly speculative. When more available on the breeding systems of other species of Villarsia, this third scenario should be evaluated information. becomes in the context of different suggestions concerning the mode of origin of distyly proposed or discussed by Charlesworth & Charlesworth (1979), Ganders (1979), Muenchow (1982), and Gibbs (1986). Volume 75, Number 3 1988 Ornduff 767 Distyly and Monomorphism in Villarsia Clearly, breeding systems in the Menyanthaceae merit further experimental work and theoretical consideration. LITERATURE CITED Baker, H. G. 1953. Dimorphism and monomorphism in the Plumbaginaceae. II. Pollen and stigmata in the genus Limonium. Ann. Bot. (London) 17: 433- i 966. The evolution, functioning and break- down of heteromorphic incompetiiliy systems. I. The Plumbaginaceae. Evolution 20: 349-368. BARRETT, S. C. H. 1988. e evolution, maintenance and loss of self-incompatibility systems. /n: J. Lovett Doust & tt Doust (editors), Plant Reproduc- tive Ecology: mene and Strategies. Oxford Univ. ress, New ——— & J. S. Saher: 1987. Variation and evolution of breeding systems in ys Turnera ws complex (Turneraceae). Evolution 41: 340-354 Bır BAHADUR. 0. Be ly Ms heterostyly in Idenlandia umbellata L. r s 98 : ieri honesti] a incompat- ibility in indica ac. P. ore (ed Phy Qani wakan. . C odel for the evolution of distyly. Ames. Naturalist : . Floral ora in Anchusa hybrida Ten. Israel J. Bot. 19: 37-41. ERNST, i 1955. Self-fertility in hdd. primulas. Genetica 27: 391-448. ^ meg T. R. 1975. Fecundity in distylous and self- incompatible homostylous pre of Mitchella repens (Rubiaceae). Evolution 29: -188. . The eue 1 E pue: E New Zea- land J. Bot. 17: 607-63 Gibbs, P. E. . Do Um M and heteromorphic self-incompatibility systems have the same sporo- phytic mechanism? Pl. Syst. Evolution 154: 285- 323. Hopper, S. D. 1979. Biogeographical aspects of T ciation in the southwest EUM ralian flora. Ann. Ecol. Syst. 10: 399-42 , F. W. 1959. Steining and observing pollen tubes in the style T means of fluorescence. Stain Technol. 34: 125- Muencuow, G. 1982. A ‘los ss-of-alleles model for the evolution of distyly. Heredity 49: 81-93. ORNDUFF, R. 1966. The origin of dioecism from het- erostyly in ic: p (Menyanthaceae). Evolution 309-314 . 1970a. Relationships in the Piriqueta caro- liniana-P. cistoides complex (Turneraceae). J. Ar- nold Arbor. 51: 492-498. à popa of Nymphoides (Menyanthaceae). Taxon 19: 715-719. 1970c. The systematics had dua system a Gelsemium (Loganiaceae). J. Arnold Arbor. 51: Taxonomy. /n: Menyanthaceae Dum., by s. Nilsson. World Pollen & Spore Fl. 2: 1-2, 11. Cytotaxonomic peg cw in Villar- sia (Menyanthaceae). ef tral. J. a 2: 513-516. 1977. ual hom in Hedyotis caeralen (Robiacohe). PL Syst. Evolution 127: 293- 1982. Heterostyly and incompatibility in Vil- borsa capitata (Menyanthaceae). Taxon 31: 495- 497. 1986. Comparative fecundity and population composition of heterostylous a and non- heterostylous species of Villarsia (Menyanthaceae) in Western Australia. Amer. J. Bot. 73: 282-286. istyly and y Specie pale in Villarsia congestiflora (MB entia dee) y with comparative re- marks o n V. capitata. Pl. Syst. Evolution 159: 81- I. CHUANG. 1988. Chromosome numbers of dns Australian species of Villarsia (Menyan- thaceae). Pl. Syst. p (in press). PHILIPP, M. & O. ScHou. 1981. An unusual hetero- morphic inc habes 03 system. Distyly, self-incom- Sauli pollen load, de fecundity in Anchusa of- Tue M gus? New Phytol. 89: 693-703. . F. CHisaKI. 1957. Studies on Am- E i ud 1 JP of the genus, with a study of het terostyly i in it. Amer. J. Bot. 44: 529-536. RICHARDS, A. J. 1986. Plant Breeding Systems. Allen nwin, Pa E O. € M. PHILIPP. 1984. An unusual hetero- rphic incompatibility system: 3. O E of distyly and self-incompatibility in Anchusa alis L. (Boraginaceae). Theor. Appl. Genet. -144. Ray, 1883. Monographie der | xig der Tur- neraceen. Jahrb. Bot. Gart. Berlin 2: 1-152. VASUDEVAN Nair, R. 1975. “wisa A a breeding mechanism of Nymphoides cristatu oxb.) O. Kuntze. J. Bombay Nat. Hist. Soc. 72: 677- 682. WIND POLLINATION IN AQUATIC ANGIOSPERMS Christopher D. K. Cook! ABSTRACT Aquatic angiosperms _ evolved from de ancestors several times. About 79 angiosperm families and 380 genera contain aquat c species; ra (31.3 119 aquatics, 100 genera dors obvious fenestra nue that are also terrestrial relatives but belong to exclusively anemophilous families; the anemophily, however, only two genera obi m have no obvious d be associated with life in water. In (Hydrocharitaceae), is it likely that the evolut aquatic environment. In Hydrilla (Hydrocharitac a th n ara the male to a aia auno ugh this mec haus aa ce e true anemo where buoy ins are ca oh ea entomophily to anemophily has ol œ) are wind pollinated. Among these wind-pollinated wind pollinated. a genera (4.2%) 0.5%), Brasenia (Cabombaceae) and Li taken plac len grains are heavy and are pesto through ertainly Bui in water, it is not considered by air movements. Wind pollination is not considered ant gra rr an important pud especially zig iated with life in the aquatic environment. Crane (1986) suggested that the seeming sim- plicity of wind pollination has deflected interest in a process that is, in fact, far from straightforward. However, combining the detailed pioneer work on pollen dimensions and pollen production by Pohl (19372, b) with the physical approaches by White- head (1969) and Niklas (1985) and with the re- newed morphological approach of Crane (19806), a better view of wind pollination in gymnosperms and terrestrial angiosperms is emerging. Wind pol- lination and associated characteristics have rarely been studied in aquatic angiosperms. The main purpose of this review is to collect information bearing on the question of which aquat- ic plants are pollinated by wind. By comparing these with their terrestrial ancestors it should be possible to find out which of them have modified their pollination mechanisms. In turn, this should give insight into the especially associated with the aquatic ment? Perhaps the highest development of aquatic angiosperms is the use of water for the transfer of question: is wind pollination environ- pollen (hydrogamy). It is also an aim of this review to see if wind pollination is a prerequisite to hy- drogamy. The term aquatic is used in the sense described by Cook et al. (1974); it includes plants whose photosynthetically active parts are submerged in water or floating on the water surface permanently or, at least, for several months each year. Within the angiosperms the aquatics have evolved from terrestrial ancestors several times, as Sculthorpe (1967) pointed out. It is important to appreciate that most aquatic angiosperms resemble terrestrial flowering plants, not only in gross features of floral morphology, but also in exhibiting similar trends of floral specialization. The aquatic members of predominantly anemophilous families Centrolepi- daceae, Cyperaceae, Hydatellaceae, Juncaceae, and Poaceae from the point of view of their pollination biology resemble the terrestrial members and are thus not described in detail here The first angiosperms were almost certainly in- sect pollinated, so anemophily is a derived state for example, see Crane, 1986). that wind pollination has originated several times It is also clear from diverse stocks. Anemophily has evolved at different times from different morphological back- grounds with differing degrees of efficiency. It is not surprising that it is sometimes difficult to make + a clear-cut disti between wind-pollinated plants and those pollinated by other means. Many pre- dominantly wind-pollinated species may be regu- larly visited by pollen-eating insects (particularly syrphid flies) and may occasionally be pollinated by insects. For suc 7 plants Stelleman (1984) used the term ambophi Quantitative Pa on the efficiency of wind pol- lination or even direct observations are lacking among aquatics. The decision of whether a species is wind pollinated or not is often based on mor- phological criteria. This is justifiable since there is a whole complex of characters associated with ane- mophily. The following list attempts to summarize the most important features of this wind pollination syndrome. ' Institut für Systematische Botanik der Universitat Zürich, Zollikerstrasse 107, CH-8008 Zürich, Switzerland. ANN. MISSOURI Bor. Ganp. 75: 768-777. 1988. Volume 75, Number 3 1988 Cook 769 Wind Pollination in Aquatic Angiosperms 1. The characteristics of wind-borne pollen grains are well summarized by Crane (1986): size 20— 40(-60) um diam., Reynolds number around 0.1 at a terminal velocity of 5 cm sec. ', spread singly, powdery and nonsticky (lipids absent, chemically altered or hidden within the wa surface relatively smooth, resistant to changes in temperature and to desiccation (size and/or number of apertures reduced). The flowers are usually unisexual: the male ba must dispense pollen, and the female must catch pollen. The amount of pollen is increased: this is usually achieved by an increase in anther size and is w often associated with a decrease in number of anthers in each flower and an increase in num- ber of male flowers. The flowers are usually separated from the leaves (temporally or spatially) and held above the water on specialized structures. The perianth and bracts are reduced (to in- crease aerodynamic efficiency). There are sometimes special arresting or ex- plosive mechanisms to ensure that pollen gets into the airstream. . The number of ovules in each flower is usually reduced; the reason for this is not clear. . The stigmas are often specialized to increase “capture” efficiency. e g e ~ ec — . AQUATICS PREVIOUSLY CONSIDERED ANEMOPHILOUS BUT PROBABLY ENTOMOPHILOUS Details of the pollination mechanisms of many aquatic angiosperms are unknown. Many with small and insignificant flowers have been considered to be wind or even water pollinated. The following, which have been thought to be wind pollinated, are probably pollinated by other means; all are illus- trated in Cook et al. (1974). ALISMATACEAE Ten of the eleven genera have large showy pet- als; some are scented and have septal nectaries and are clearly insect pollinated. Wiesneria, a ge- nus with two African and one Asian Lese re- cently redescribed by Sivadasan (1986), has whorled, subsessile, unisexual flowers with pur and reflexed perianths. The male flowers, which are borne above the females, have three stamens, the lowest number in the family. The carpels are reduced to three or four in each female flower. These characters may suggest anemophily, but the inflorescences are shorter than and partly hidden by the leaves; the petals, though small, are cream- colored and relatively conspicuous; the anthers are small (ca. 1 mm long); the filaments are short (ca. 0.5 mm long); the pollen grains are echinate, have numerous apertures (Fig. 6), and are somewhat sticky. These characters strongly suggest insect pollination. It is not known if Wiesneria is scented or if it produces nectar. APONOGETONACEAE Aponogeton, the only genus, sometimes has cat- kinlike inflorescences. A few species have unisexual flowers, and sometimes the perianth is reduced. These features suggest wind pollination. However, the pollen has supratectal spines in all species (see Bruggen, 1985) and most species investigated to date have septal nectaries and are strongly scented. Aponogeton is insect pollinated, but in some species some pollen may be transferred by wind. This is unlikely, however, as the pollen grains stick to- gether in clumps. ARACEAE It has been suggested that Acorus and Orontium are wind pollinated. A considerable amount of work has been published on both genera, cited by Cook et al. (1974), but I have found no observational data on pollination. From their morphology (mostly bisexual flowers, relatively large perianth segments, small anthers, sticky pollen) they are probably en- tomophilous. ERIOCAULACEAE Because the flowers are small, rather insignifi- cant, often unisexual, and apparently dry, they have been considered to be wind pollinated. How- ever, the flowers are structurally complex and elab- orate. In the few species examined, they secrete nectar, have sticky pollen, and are thus most likely insect pollinated or autogamous (see Stützel, 1981, HANGUANACEAE The flowers are small and unisexual with long slender filaments, which suggests anemophily (see Dahlgren et al., 1985). However, the pollen grains are spinulose, the male flowers have conspicuous eshy bodies, and the stigmas are not exposed beyond the perianth, all of which suggests zooga- my. No direct observations on pollination have yet been reported. 770 Annals of the Missouri Botanical Garden FIGURES 1, 2. =6 1. Diagram of flowers of Brasenia schreberi. (a) Female phase. (b) Male phase. Scale bar mm.—2. Diagram of flowers of Callitriche obtusangula. Left, female flower. Right, male flower. Scale bar = 1 mm. LEMNACEAE The flowers are unisexual, very small (stamens ca. 1 mm long, stigma ca. 0.5 mm above the carpel) and not obviously showy. Wind and water have been suggested as pollen vectors, but as Landolt (1986) pointed out, this is unlikely, as the pollen grains are spiny, sticky, often less than 20 um in diameter, and sometimes only 20 grains develop in each locule; these characters all suggest zoog- amy. Although water movements may bring flow- ers on different fronds in contact, the pollen is probably mostly transferred on the legs of different kinds of arthropods (flies, aphids, mites, small spi- ders). At least some races of Lemna minor and Wolffiella oblonga are self-incompatible, while many other species are self-compatible, although some of these are not autogamous and require a pollinator (for details see Landolt, 1986). PODOSTEMACEAE This family has about 45 genera, of which some, such as Mourera and Tulasneantha, are clearly insect pollinated. The majority, however, have very small, almost naked flowers (see Cook et al., 1974). Of these, some are probably autogamous, while others may well be wind pollinated (see Philbrick, 1984b). It is a shame that so little is known about the floral biology of this rather extraordinary fam- ily II. FAMILIES WITH WIND-POLLINATED SPECIES CABOMBACEAE (sometimes included within the NYMPHAEACEAE) Cabomba is insect pollinated, and Brasenia is wind pollinated (Fig. 1; see Osborn & Schneider, this volume: 778-794). Brasenia has most likely evolved to anemophily from entomophily in the aquatic milieu. CALLITRICHACEAE This monotypic family is wind pollinated (Fig. 2) or shows an extreme kind of autogamy above or below the water surface, described by Schotsman (1982, 1985) and Philbrick (1984a). It is doubtful that true hypohydrogamy (transfer of wet pollen through water to wet stigmas) takes place. It is also unlikely that epihydrogamy, with floating pol- len, is an effective means of pollen transfer since the stigmas are usually either submerged or aerial. HALORAGACEAE Haloragis, Laurembergia, Myriophyllum, Proserpinaca, and Vinkia have aquatic species, all of which are well adapted to wind pollination. All have flowers with reduced and often caducous perianths (Fig. 3) and dry, powdery pollen liberated from long-filamented anthers. There is a trend from bisexual to unisexual flowers culminating in dioecy. Patten (1956) suggested without quantitative data that significant quantities of pollen may be trans- ferred by insects in Myriophyllum spicatum. HIPPURIDACEAE This monotypic family has highly reduced bi- sexual or sometimes unisexual flowers. The stamens are reduced to one with a relatively massive anther, and the ovary is reduced to a virtually naked, one- seeded carpel (Fig. 4). The pollen is anemophilous- like. No published data on pollination have been found, but from personal observations it is strongly protogynous (Fig. 4), and seed-set is usually good. can only assume it is wind pollinated. Volume 75, Number 3 1988 Cook 771 Wind Pollination in Aquatic Angiosperms 4a FIGURES 3-5. female below. Scale bar = ] cm. (b) Female flower. of Hippuris vulgaris. (a) Female phase. (b) Male phase. Scale bar — a 75 m —5. D emale inflorescence. Scale 1 cm. (d) Male flower. Scale bar = of Hydrostachys perrieri. (a) (c) Male inflorescence. Scale bar — HYDROCHARITACEAE This exclusively aquatic family with 16 genera is extraordinary in its spectrum of floral structures, showing entomophily, anemophily, epi- and hy- pohydrophily (see Cook, 1982, for a general re- view). In Appertiella, Enhalus, Lagarosiphon (Figs. 7, 11), Maidenia, Nechamandra, and Val. lisneria (Fig. 8) the male flowers become detached from the mother plant and are dispersed by wind or water currents; in these genera the pollen is sticky and transferred directly from anther to stig- ma. In Elodea the pollen is liberated on the surface of the water (Fig. 12) and is dispersed by wind or water currents to the stigma (Cook & Urmi-Konig, 1985). Although wind plays an important role in pollination, this kind of pollination is usually clas- sified as epihydrophilous, since part of the pollen is in contact with water; however, it must be stressed that the pollen and stigma remain dry. In pollination biology Hydrilla is remarkable, as shown by Cook € Liónd (1982). The male flowers are liberated from the mother plant as buds, which then open explosively, shooting pollen grains through the air. The pollen of Hydrilla is inaper- turate (the furrow illustrated by Yeo et al., 1984, is an artifact), spherical, 93 + 5.7 um in diameter, and densely covered with baculae 2 um long, each bearing a small flamelike process (Fig. 9a, b). This pollen is too large and rough to “fit” in the wind 3. Diagrammatic representation of vre g: mu (a) de aper ad hue (c) = eT | above, m of flowers Mos representation bar — 1 cm. (2) “Female ber Scale b ) Male flowe pollination syndrome. The stigmas are rather small and borne at the base of a wineglasslike perianth Fig. 10) exposed to the air but below the surface of the water. For effective pollination the pollen should not get into the airstream, as in other an- emophilous plants, but must drop almost vertically to reach the stigmas, and it must stick and not bounce out again. The pollen is propelled through the air, but wind, as such, plays an unimportant role; I hope it is not necessary to create a new ~ term for this unique pollination mechanism. Described in detail by Cook & Urmi-König (1983), pollination in Limnobium is *normal" when compared with other genera of the Hydrochari- taceae. The sepals of the male flowers act as pollen- arresting organs (Fig. 13). The pollen is then “picked up” off the petals by the wind. The arched petals were suggested by Cook & Urmi-König (1983) to have an aerodynamic function, but after obser- vations in the field it seems clear that their function is to keep the pollen dry when waves come or when it rains; when the flowers are immersed in water the sepals come to rest on the petals, thus enclosing an air bubble. The marine genera Halophila and Thalassia have wettable pollen and stigmas and are pollinated underwater (hypohydrogamous). Within the Hydrocharitaceae, Limnobium is the only genus with dry, powdery, and buoyant pollen that is transported to the stigmas by movements Annals of the Missouri Botanical Garden Me MI w I P a ( > " > A. m S‘ ~ AR M tatu Ry AP... TP A La P> La pt ^n & AP Volume 75, Number 3 1988 773 Wind Pollination in Aquatic Angiosperms of air and thus can be called anemophilous. Hy- drilla has heavy pollen actively propelled through the air and is not dependent on air movements; it is, therefore, not strictly speaking “wind pollinat- ed." Morphological and anatomical evidence in- dicates that the patristic relationships of Limno- bium are to exclusively entomophilous genera Hydrocharis, Ottelia, and Stratiotes. The epi- and hypohydrogamous genera (within the framework of the Hydrocharitaceae) are not patristically close to Limnobium and are probably derived from a Blyxa-like ancestor. The genus Blyxa is essentially entomophilous or highly autogamous (Cook et al., 1981; Cook € Lúónd, 1983) The evolution from entomophily to anemophily in Limnobium has most likely taken place within the aquatic environment. The expanded sepals, which hold pollen until wind carries it off, and the arched petals, which protect the pollen from waves and submergence, may be considered as elabora- tions connected with wind pollination and with the aquatic environment. HYDROSTACHYDACEAE This extraordinary monotypic family, reviewed by Cusset (1973), has about 22 species. The mor- phology of the flowers suggests anemophily (Fig. 5). The perianth is absent and a bract arrests the pollen, analogous to Limnobium. The carpel con- tains numerous seeds, an unusual feature in ane- mophily. No direct observations on pollination have been published. JUNCAGINACEAE and LILAEACEAE These two families are sometimes united. From the floral morphology they are wind pollinated, but I know of no direct observations. Some species have pollen-arresting mechanisms very like those in Potamogeton, and others have highly hetero- morphic flowers (Fig. 14) (Posluszny et al., 1986). PLANTAGINACEAE The aquatic genus Littorella has obvious affin- ities to the terrestrial genera Bougneria and Plan- tago. Some species of Plantago show tendencies toward entomophily, but most are anemophilous. Littorella, however, shows further anemophilous specialization in having stalked male flowers and sessile, uniovulate female flowers. POTAMOGETONACEAE and RUPPIACEAE These two families are sometimes united. Most of the species of Potamogeton are wind pollinated, with erect, many-flowered spikes, spherical pollen grains, bisexual flowers with strong protogyny, and organs generally known as connective appendages to arrest the pollen and liberate it in the airstream after the female receptive phase (Fig. 15). Pota- mogeton filiformis, P. pectinatus, and others are pollinated below the water surface. This kind of pollination is described by Philbrick in this volume (pp. 836-841). The pollination mechanism of Groenlandia has not, I believe, been critically described, but my own observations revealed that in spite of having rather short-stalked, two-flowered inflorescences, it is either autogamous or pollinated by wind and does not seem to be, as sometimes supposed, pollinated by floating pollen (epihydrogamous). Ruppia is better known than the other genera; Verhoeven (1979) described the pollination in R. cirrhosa and R. maritima in detail. Both have curious elongated ‘v’-shaped pollen grains (Fig. 16) that sometimes form chains. This might suggest hypohydrogamy, but the pollen is nonwettable and is liberated in bubbles; the stigmas are protected by the same bubbles. Ruppia cirrhosa is usually pollinated at the surface like Elodea, while R maritima, in Europe, is usually pollinated in bub- bles under water like Potamogeton pectinatus. SPARGANIACEAE A monotypic family having flowers arranged in unisexual heads with male heads above females (Fig. 17). All species are clearly wind pollinated : (Cook & Nicholls, 1986, 1987), in spite of the fact that some syrphid flies specialize on Spar- ganium pollen as a source of food THURNIACEAE It is not certain that this monotypic family de- serves to be called aquatic. The floral structure suggests pollination by wind (Fig. 18) but no ob- servations have been published. FIGURES 6-9. 6. Pollen grain of Wiesneria triandra. Scale bar = 5 um.— 7. Young male flower of uy V muscoides, era and staminodes not fully extended and some pollen grains missing. Scale bar = The male flower of Vallisneria americana. (a) Showing branched stamen. 0 — 9. Pollen of Hydrilla verticillata. (a) A somewhat shrunken jen when showing tepal. Scale bar The same from the other dde fresh they are spherical. Scale ain = 10 um. (b) Detail of surface showing baculae. Scale bar = 774 Annals of the ME Botanical Garden of the flowers of Limnobium laevigatum. Left, female. FicunES 10-13. 10. Diagram of pollination in Hydrilla verticillata. Left, female. tes m Scale bar — aaa bar = 1.2 —11 of pollination in Lagarosiphon m 2. Diagram of pollination in Elodea nuttallii. Left, female. Right, male. Scale bar = pps — 13. Diagram Right, male. Scale bar = 3 mm | PMA4b 4 >) 22 28 Q E ex a = v DX = X CS re = SAS Exe 4 - 4: 16a . Di Lipid s Y Lilaea scilloides. (a) Whole plant with inflorescences. b =5 FicunES 14-16. Scale pui = 8 mm. () B ual flower. Scale bar = 3 mm. (c) Long-styled, female flower. Scale bar mm.— 15. Diagrammatic cede tion of a mraeg (a) Inflorescence during female sag: ini bor = | mm ower at female phase. Scale = Im c) Flouer at male phase. Scale bar = . (d) Longitudinal section of flower at male phase, a ng t de connective appendages as ues rior pet. Scale bar — — 16. rubia of Ruppia maritima. (a) Inflorescence. Scale bar — 1 mm. (b) Chain of five pollen grains. cinis bar — 2 Volume 75, Number 3 1988 Cook 775 Wind Pollination in Aquatic Angiosperms FiGURES 17-19. 17. En mcs below. Scale bar = 2 cm. — 18. Diagram 9. Di iagrammatic representation Typha p bar = = 2 cm. (b) Female flowers; left dons. right sterile. Scale bar — bar TYPHACEAE This monotypic family is relatively well docu- mented and is clearly wind pollinated (Fig. 19; see Krattinger, 1975), even though syrphid flies may collect pollen adhering to the female inflorescence. All species have single ovules in each flower; some species, however, have pollen in tetrads. This contradicts all previous theoretical predictions re- garding the function of pollen tetrads in pollination. Nicholls & Cook (1986) found that pollen tubes from tetrads are capable of traversing from the stigma of one flower through air to the stigma of another (Fig. 20) and fertilizing neighboring flow- ers. This results ] in ingraased rl of gametes ír In effecting as seed-set) com- pared with other seh of Typha having single pollen grains. CONCLUSIONS About 79 angiosperm families and 380 genera contain aquatic species. Most of the data are to be found in Cook et al. (1974). Excluding the Podo- stemaceae, because its floral biology is so poorly known, 31.6% of the families and 42% of the genera are pollinated abiotically. Of the abiotically pollinated genera, 18 (or 19 including Potamogeton) have wettable pollen and are pollinated under water; about seven genera are male a Scale ar = ] mm. ( nt of Thurnia lacera (a) Inflorescence. Scale p»: =Ic angustifolia. (a) Inflorescence male above, female below. . (c) Male flowers. Scale pollinated at the water surface. This leaves 119 genera (35.5%) exclusively wind pollinated. I have tried without success to find geographical corre- lations between pollination and distribution of aquatics; it seems that about one-third wind pol- linated remains reasonably constant when one com- pares the Old and New worlds, Northern and South- ern hemispheres, and Tropics and Temperate zones. Within particular plant communities or associations there are enormous differences in the proportion of wind-pollinated species as also found by Kugler (1971). Reedswamp and sedge-dominated com- munities are mostly made up of wind-pollinated plants. Also the plants of deep and permanent water are mostly abiotically pollinated; all the marine angiosperms, for example, are hydrogamous. Nevertheless, about two-thirds of all aquatic genera (and this probably also applies at the level of species) are biotically pollinated. So little is known about the breeding systems of aquatic species that it is not possible to generalize on in- or outbreeding within the context of the mode of pollination. Aquatic angiosperms have evolved from terres- trial ancestors. All available evidence suggests that the aquatic members of at least six families (Cen- trolepidaceae, Cyperaceae, Hydatellaceae, Junca- ceae, Plantaginaceae, and Poaceae) have evolved from anemophilous ancestors. This translates to an 776 Annals of the Missouri Botanical Garden FIGURE 20. crossing from one stigma to another. b, estimate of at least 100 genera that brought ane- mophily with them into the aquatic milieu. This leaves a mere 19 genera where the anemophily may have evolved subsequent to the invasion of the aquatic habitat. When these 19 genera are examined a little closer, although they do not have any obvious terrestrial relatives, they mostl belong to exclusively anemophilous families (Calli- trichaceae, Haloragaceae, Hippuridaceae, Hydro- stachydaceae, Lilaeaceae, Potamogetonaceae, Ruppiaceae, Sparganiaceae, Thurniaceae, and Ty- phaceae) In only Brasenia (Cabombaceae) and Limno- bium (Hydrocharitaceae) is it very likely that the evolution from entomophily to anemophily has tak- en place in the aquatic environment. The genus Typha latifolia. —a. m dud two grains germinating. Scale bar — 'ale um; c, d, scale bar = 50 um 5 um. —b-—d. Pollen tubes Hydrilla (Hydrocharitaceae) is not strictly “wind” pollinated, because heavy pollen grains are actively propelled from the male flowers to the females. This mechanism relies on water and therefore prob- ably evolved in water. However, it shows an al- ternative pathway to the evolution of anemophily. From a total of 380 aquatic angiosperm genera, a minimum of two and a maximum of 19 may have evolved from entomophily to anemophily sub- sequent to the invasion of the aquatic habitat. From these numbers alone it does not seem likely that pollination by wind is a feature that is especially associated with life in the aquatic environment like, for example, the presence of gas spaces (lacunae) and hydropoten or the absence of lignin and/or stomata. Many aquatics clearly evolved from ter- Volume 75, Number 3 1988 Cook 777 Wind Pollination in Aquatic Angiosperms restrial groups that were already anemophilous; there is no evidence that any of these aquatics have returned to biotic pollen transfer. Taken as a whole, it is not possible to say that anemophily is particularly beneficial or detrimental to plants iving in water. However, it is remarkable, when the different floral types illustrated in this contri- bution are compared, how many different strategies are adopted to guarantee transference of pollen by means of wind. There are no obvious trends among the wind-pollinated aquatics that indicate the path- to hydrogamy. “Aquatic” anemophily is not different from “terrestrial” anemophily. LITERATURE CITED BRUGGEN, H. . Monograph of the genus Aponogeton (Aponogetonaceae). Biblioth. Bot. 137: 1-76. Cook, C. D. K. 1 ee mechanisms in the Hydrocharitaceae. Pp. 1 in ymoens, S. S. Hooper & P. Compére d h Studies on Aquatic odin Plants. Royal Botanical Society of Belgium, Brus CR 82. A revision of the genus Hydrilla ida cas Aquatic Bot. 13: 485- 504. = 83. A revision of the genus T au (Hydrocharitaceae). Aquatic Bot. 15: 1-52 & M. S. NicHOLLSs. 1986. A monographic study of the genus Sparganium (Sparganiaceae), Part 1. Subgenus Xanthosparganium Holmberg Helvet. 96: 213-267 87. A monographic study of Sparganium (Sparganiaceae), Part 2. Subgenus Sparganium. Bot. Helvet. 97: ae & K. Urmi-K6nic. 1983. revision of the genus Limnobium a aqp: Hydronytri (Hydro- charitaceae). Aquatic Bot. . 1985. revision of the genus vig (Hydrocharitaceae). guae Bot. 21: 111- L R. LüóND & B. Nar. 1981. Floral biology : UT, Rix, J. SCHNELLER & M. SEITZ. 1974. Water Plants of the World. W. Junk, The Hague. CRANE, P. R. 1986. Form and function in wind dispersed pollen. /n: S. Blackmore & I. K. Ferguson (editors), Pollen and Spores. Academic Press, London. Sym- osium Series, Number 12: 179-202. Cusser, C. m des Hydrostachyaceae. Adansonia 13: 75-119. DAHLGREN, R. M. T., H. T. CLIFFORD & P. F. Yeo. 1985. The Families of the Monocotyledons. Springer-Ver- lag, Berlin, n New York, Tokyo. ; Genetic mobility in Typha. 1971. KucLER, H. Die Verbreituna der Anemogamie in mitteleuropáischen oss Ber. Deutsch. Bot. Ges. 84: 197-209. LANDOLT, E. 1986. The "ied of Lemnaceae — a mono- graphic study, Volume 1. Veróff. Geobot. Inst. ETH Stiftung Rübel Zürich 71: 1-566. NicHOLLS, M. S. & C. D. K. Cook. 1986. The function of pollen tetrads in Typha (Typhaceae). Veróff. Geo- bot. Inst. ETH Stiftung Rübel Zürich 87: 112-119. NIKLAS, K. J. The aerodynamics of wind polli- nation. Bot. Rev. (Lancaster) 51: 328-386. PATTEN, B. C. 1 Notes E the biology of Myrio- phyllum spicatum L. in a Torrey Bot. Club 83: a PHILBRICK, C. T. 1984a. Pollen tube growth within vegetative tissue of Callitriche (Callitrichaceae). Amer. J. Bot. 71: 882-886. Aspects of floral biology, breeding system, and seed and seedling biology in Podostemum ceratophyllum (Podostemonaceae). Syst. Bot. 9: 4. ew Jersey lake. Bull. Pont, F. 1937a. Untersuchungen zur Morphologie und Biologie des Pollens IV. Die Pollenerzeugung der lag Beih. Bot. Centralbl. 56A: 365-470. 37b. Beitráge zur Morphologie und eee PosLuszNv, U., W. A. CHARLTON & D. K. Jain. 1986. Morphology and development of the reproductive shoots of Lilaea scilloides (Poir. Hauman (Alis- matidae). Bot. J. Linn. Soc. 92: 323-342. SCHOTSMAN, H. 82. Biologie florale = Calli- triche: étude « sur quelques espéces d'Espagne méri- dionale. Bull. Mus. Natl. Hist. Nat., 4e Ser. Sect. , Adansonia 4: 111-1 Biologie orale = Retails II. Etude TEE d’Océanie. Bull. Mus. ipi Hist. Nat.. , Adansonia 7: 357- Sc ULTHORPE, C. D. 1967. "The SEI of Aquatic Vas- ] ndon de Sé. Sec ba cular Plants. E. Arnold, Lo SIVADASAN, 1 Wiesnéria triandra (Dalzell) Mi HAR eine sehr silia und w matacee aus Indien. hnos. P. 1984. Reflections on the transition from wind pollination to ambophily. Acta Bot. Neerl. 33: -508. STÜTZEL, 1. 1981[1982]. Zur Funktion und Evolution kópfchenfórmiger Blütenstánde, insbesondere der Er- iocaulaceae. Beitr. Biol. Pflanzen 56: 439-468 84. Blüten- und infloreszenzmorphologische Untersuchungen zur Systematik der Eriocaulaceen. Diss. Bot. ee 71: 1-108. VERHOEVEN, J. T. 1979. The ecology of Ruppia- dom inated a in western Europe. I. Dis- uppi ia m in relation to their autecology. Aquatic Bot. 6: WHITEHEAD, D. R. 1969. ind polination in | the an- giosperms. ub 23: Yeo, R. R., R. H. Fark € J. R. THUR RSTON. 1984. The morphology of hydrilla Votos p pcr E F.) yle). J. Aquatic Pl. Manag 22: Jeffrey M. Osborn?’ and Edward L. Schneider? MORPHOLOGICAL STUDIES OF THE NYMPHAEACEAE SENSU LATO. XVI. THE FLORAL BIOLOC Y OF BRASENIA SCHREBERT ABSTRACT Observations conducted in East Texas on the pollination biology of DEAR schreberi J. F. Gmel. confirm iurnal, with individual apilla i anded surface area for pollen Bene Seco flowers are uS de fna d and functionally staminate. > Simi al filaments are elongated, elevating de don anthers to a position above the centrally aggregated stigmas. Although self- pollination occurs, roda Sa prevents indivi idual flowers dios self fertilizing. Notiphila of. cressoni (Diptera; Ephydridae) wa or. Base len Evidence from pollination biology, relationship between the Cabombaceae and Nymp ‘pollen transfer. This pollination mechani a floral anatomy, seed anatomy, and embryology indicates a close evolutionary N aeaceae sensu stricto. Genera of the Nymphaeaceae s. str. and Nelumbonaceae exhibit a phyletic elaboration of the flower, whereas the Cabombaceae represent a phyletic reduction. Brasenia schreberi J. F. Gmelin is a monotypic genus with a wide but sporadic distribution in lakes, ponds, and slow streams. It occurs in eastern Asia, Australia, Africa, the West Indies, and South, Cen- tral, and North America. In North America the species ranges from Florida to East Texas, north to Prince Edward Island, southern Quebec, south- ern Ontario, and Minnesota. It also occurs in Idaho, California, north to British Columbia, and Alaska (Wood, 1959). Although Brasenia does not pres- ently occur in Europe, fossil specimens are known (Srodon, 1935; Tralau, 1959; Jessen et al., 1959; Hall, 1978; Collinson, 1980). Brasenia is com- monly known as water-shield, water-target, purple bonnet, and purple wen-dock and is a cultivated food source in Japan (Matsuda & Hara, 1985). Brasenia is one of eight genera within the Nym- phaeaceae sensu lato as circumscribed by Bentham & Hooker (1862). Subsequent workers have grouped the nymphaeaceous genera into various orders, families, and tribes (for review see Gole- niewska-Furmanowa, 1970; Takhtajan, 1980; Cronquist, 1981). The opinions that the genera should be divided among three subfamilies or three families (Nymphaeaceae, Nelumbonaceae, and Ca- ombaceae) within the order Nymphaeales has been widely advocated on the basis of available data. A ninth genus, Ondinea, was described by den Hartog (1970) and placed within the Nymphaeaceae sensu stricto. This taxonomic alignment is supported based by studies of seed anatomy, floral morphology, and floral biology (Schneider, 1978, 1983) In all systematic treatments, Brasenia has been allied with Cabomba on the basis of such shared morphological features as long, slender, sympodial stems; peltate floating leaves; small hypogynous flowers with apocarpous gynoecia, and few floral parts. Moseley et al. (1984), comparing anatomical ! This paper represents a portion of a thesis presented by the first author to the Graduate School of Southwest Texas State University in partial fulfillment of the requirements for the bui of Science degree. The research rch at Southwest Texas State University. A&M University) , Alexander D. Hury n (University of Georgia), Wayne N. Mathis (Smithsonian Institution) , ind Ingolf Askevold (University of Manitob a) r nsect S cd Special rec ognition is offered to Temple EasTex, Inc. for the gratis use of Cypress Pon 3 ned address: Department of Botany, The Ohio State Universtiy, Columbus, Ohio 4. 3210. US. A. ANN. MISSOURI Bor. GARD. 75: 778-794. 1988. Volume 75, Number 3 1988 Osborn & Schneider 779 Brasenia schreberi and morphological features of the vegetative axes and flowers in Cabomba and Brasenia, suggested that the two genera are more distant taxonomically than envisioned by earlier workers (Vinogradov, 1967; Goleniewska-Furmanowa, 1970; Buko- wiecki et al., 1972, 1974), a view also shared by Collinson (1980). Moseley et al. emphasized, how- ever, that the taxonomic distance was not adequate to warrant dismantling the Cabombaceae. Rich- ardson (1969) concluded that the floral vascular system (and by inference, the flower) of Brasenia exhibits a phylesis toward condensation and re- duction from a more primitive, larger-flowered ra- nalian ancestor. Ito (1986a) determined that the ontogeny and anatomical construction of the re- ceptacular vascular plexus, a feature common to all water lily genera, differs in the Cabombaceae from those found in the Nymphaeaceae s. str. Ito also considered the floral vasculature of Cabomba to be derived (via reduction in the number of sta- mens and carpels) from the type of vasculature found in the flower of Brasenia. Numerous studies of Brasenia involving a di- versity of features have been conducted including: seed anatomy and embryology (Weberbauer, 1893, 1894; Chifflot, 1902; Cook, 1906, 1909; Meli- kyan, 1964; Khanna, 1965; Corner, 1976); pollen morphology (Wodehouse, 1932; Ikuse, 1955; Ueno & Kitaguchi, 1961; Ueno, 1962; Meier, 1964; Walker & Doyle, 1975; Walker, 1976a, b; Clark & Jones, 1981; Batygina & Shamrov, 1983); xy- lem anatomy (Kosakai, 1968); karyology (Langlet & Sóderberg, 1927; Okada & Tamura, 1981); chemical analyses (Nakahara, 1940; Riemer & Toth, 1970; Goleniewska-Furmanowa, 1970; Ka- kuta & Misaki, 1979; Sevilla et al., 1984); leaf development and anatomy (Goleniewska-Furma- nowa, 1970; Kaul, 1976; Chen & Zhang, 1986); floral anatomy (Troll, 1933; Moseley, 1958; Khan- na, 1965; Richardson, 1969; Moseley et al., 1984; Ito, 1986a); paleobotany (Hall, 1978; Collinson, 1980; Dorofeev, 1984); and general morphology and taxonomy (Lawson, 1888; Schrenk, 1888; Caspary, 1891; Keller, 1893; Raciborski, 1894; Schilling, 1894; Gwynne-Vaughn, 1897; Hill, 1900; Chrysler, 1938; Wood, TELA Adams, 1969; Kristen, 1974; Ogden, 1974; o & Banerjee, 1979; Matsuda & Hara, 1985). ‘Little attention, however, has been given to the floral (pollination) biology of the genus. Tokura (1937) investigated the blooming of Brasenia in Japan and was the first to record the movements of floral parts during the two-day anthesis period. Tokura suggested that flowers on the first day of anthesis are pistillate, while on the second day staminate, during which time large quantities of pollen are released. Tokura noted the importance of the increased height of second-day, pollen-releasing flowers above those of the first-day, pollen-receptive flowers from the standpoint of pollen transfer but did not suggest the vector(s) for pollination. Schneider & Jeter (1982) reported “Notiphila-like” flies functioning as pollinators in populations of Brasenia growing in East Texas. This investigation is part of a continuing series of studies designed to contribute new evidence for determining relationships among water lily genera. It is the objective of this study to: (1) confirm and amplify observations on floral morphology and flo- ral behavior during anthesis of Brasenia; (2) elu- cidate the mechanism(s) of pollen transfer and re- late floral morphology to pollination syndrome(s); and (3) compare the pollination biology of Brasenia with other genera of Nymphaeaceae s.l. This may contribute to a better understanding of the phy- logeny of this angiosperm family, which occupies a basal, pivotal systematic position in many old and modern classification schemes. MATERIALS AND METHODS SITE DESCRIPTION Observations on the floral biology of Brasenia were conducted during the summers of 1986 and 1987 in Toledo Bend Reservoir, Sabine County, Texas. Extensis m of Brasenia exist in r. This study was conducted in a cove of six ilico hectares adjacent to the Willow Oak Recreation Area. The water level within the reservoir fluctuated widely throughout the study period. The depth of water in which Brasenia grew ranged from 0.5 to 2.5 m. FLORAL CYCLE Ten flowers in various stages of anthesis were tagged with numbered fluorescent orange corks by securing them to peduncles with nylon fishing line. Heights of the floral structure above and below water level were measured over four consecutive days. Measurements were made from the base of the receptacle to the water level. In addition, sepal, petal, staminal, and stigmatal positions were ob- erve CAGING TREATMENTS Exclusion treatments were accomplished by placing floating cages over emergent flowers for the duration of anthesis. Flowers were tagged with fluorescent corks attached to the peduncle for fruit 780 Annals of the Missouri Botanical Garden retrieval. Different experiments were identified by color-coded survey ribbons stapled to corks. Cages consisted of 950-ml styrofoam and plastic drinking cups mounted inversely on 21-cm' styrofoam sheets. The center portion of the styrofoam base was re- moved to allow placement of cages over flowers. Two types of cages were used: one to exclude all abiotic and biotic pollen vectors, and a second to exclude only biotic vectors. Abiotic and biotic ex- clusion cages were made from clear plastic cups (transparent cages). Biotic exclusion cages were produced using styrofoam cups with four rectan- gular windows cut out around their circumferences. Windows were covered with 1-mm? fiberglass mesh screen glued to the styrofoam cups (mesh cages). Transparent cups were mounted on 2.5-cm-thick styrofoam, while mesh cups were mounted on 1.3- cm-thick styrofoam sheeting to maximize flower exposure through the mesh windows above the styrofoam base. Both types of cups were mounted to the styrofoam bases with water-insoluble glue. Three control groups, each consisting of 25 sec- ond-day flowers, throughout the study period to determine natural seed set. An additional control group, consisting of 25 flowers that had morphologically short stigmas, was also tagged. Each exclusion treatment involved were tagged during intervals various floral manipulations of 25 first-day flowers and their subsequent seed production to check for the following: Parthenocarpy (Group A). Flowers were cov- ered prior to anthesis with transparent cages and emasculated. Emasculation involved the removal of undehisced anthers on the first day of anthesis. Cages were observed periodically throughout an- thesis to check flower position and condition. Autogamy (Group B). Flowers were covered with transparent cages and left undisturbed. Stigmatic receptivity (Group C). (1) Flowers were covered with transparent cages and emas- culated. First-day stigmas were dusted copiously with pollen transferred mechanically from uncaged pollen-releasing flowers. (2) Flowers were covered with transparent cages and emasculated. Pollen was transferred mechanically to second-day stig- mas. Allogamy (Group D). (1) Geitonogamy. Flow- ers were covered with transparent cages and me- chanically cross-pollinated on the first day with pollen from uncaged flowers from the same plant. (2) Xenogamy. Flowers were covered with trans- parent cages and cross-pollinated on the first day with pollen from uncaged flowers from different plants. Anemophily (Group E). (1) Flowers were cov- ered with mesh cages and left undisturbed. (2) Flowers were covered with mesh cages and emas- culated. POLLEN DISPERSAL AND TERMINAL SETTLING VELOCITY The presence of wind-borne Brasenia pollen was determined using an Andersen 0101 particle-size air sampler (Andersen, 1958). In addition, dis- persed pollen was quantified by measuring distance and angle of dehiscence from pollen-releasing flow- Measurements were made using calibrated poster boards 71.5 x 56 cm. Black poster boards (pollinometers) were marked in increments of 10 cm and 20° from the midpoint of one edge (Fig. 1). Circles equal in size to an average stigmatic surface area of a first-day flower were drawn into distance-angle segments. Boards were laminated and mounted on 2.5-cm-thick styrofoam. Individual pollinometers were placed downwind from single pollen-releasing flowers with the pe- duncle at the 0? mark between 0800-1200 hours. All additional pollen-releasing flowers from the sur- rounding 3-m radius were removed. Pollinometers were retrieved after total anther dehiscence, an the number of pollen grains within each circle was quantified using a hand lens. Samples were taken on two separate days, six trials per day. Wind- speed measurements were taken using a Taylor anemometer. Results were analyzed using Stu- dent's t-test and regression analyses. Stigmas were viewed with a dissecting micro- scope to detect the presence of pollen. These in- cluded stigmas on uncaged first-day flowers in close proximity and downwind of pollen-releasing flowers and uncaged first-day flowers not adjacent to pol- len-releasing flowers. Comparisons were made of pollen quantity on stigmas of flowers in each con- dition. Additionally, pollen distribution was ob- served (e.g., more grains on leeward or windward surfaces). The terminal settling velocity of freshly collected Brasenia pollen was determined utilizing the stro- boscopic photography techniques of Niklas (1984). POLLEN—OVULE RATIOS AND POLLEN VIABILITY The number of pollen grains/flower was deter- mined by suspending all pollen from 10 anthers in a 0.5-ml solution of aniline-blue in lactophenol and counted using a hemacytometer (Cruden, 1977). Pollen-ovule ratios were calculated assuming a mean number of 28 anthers and 12 carpels (24 Volume 75, Number 3 1988 Osborn & Schneider 781 Brasenia schreberi 70 kama. 70 O sn O O O = O O 60 — 60 O = O O O = ° O 50 = 50 O O = o O O o = O O 40 eus 40 O o = O O re) O — O re) 30 — 30 O o =0 ° O ° o= o 9 ° 20 enn: 20 O O o= o/o O iss O\o =o O O a 10 — 10 (omm o) O O FIGURE 1. Pollinometer. Circles represent the average stigmatic surface area of first-day flowers. Seventy cm cale. ovules) per flower. Pollen viability was determined using aniline-blue in lactophenol staining. FLORAL DENSITY The distributional density of flowers was deter- mined by counting the total number of first- and second-day flowers within Vé-m frames in a random stratified design. FLORAL SECRETIONS The presence of floral secretory tissues was de- termined using neutral red stain as an indicator (Esau, 1965; Vogel, 1966). Fresh first- and sec- ond-day flowers were placed in neutral red for three to six hours. After excess stain was removed by lightly washing with water, the flowers were ex- amined using a dissecting microscope. ULTRAVIOLET REFLECTANCE AND ABSORPTION Ultraviolet (UV) photographs using black and white Kodak Plus X film, 125 ASA, and a Kodak Wratten ultraviolet filter No. 18A were made of first-day and second-day flowers and leaves in sun- light. Second-day flowers were photographed prior and subsequent to anther dehiscence. ° Aa. » >. “2 v i TC. Ae a i rt > Lx - a aum MES - AR we NTN , Missouri Botanical Garden Annals of the Volume 75, Number 3 1988 Osborn & Schneider 783 Brasenia schreberi ANTHOPHILOUS VISITORS Observations were made to determine diversity, frequency, behavior, and extent of pollen loads of various insect visitors. Insects on flowers and leaves were collected using kill jars with ethyl acetate and were preserved in 70% ethanol. Voucher speci- mens are housed at SWTSU SCANNING ELECTRON MICROSCOPY Floral and fruit specimens were investigated with a Cambridge S90 scanning electron microscope M). Tissues were fixed for 24 hours in 2%' E glutaraldehyde in 0.1 M cacodylate buffer, pH 7.2. amples were washed with the buffer and postfixed in 1% osmium tetroxide for three hours, then stored in the buffer. Floral specimens were dehydrated in acidified dimethoxypropane (Postek & Tucker, 1976; Linn et al., 1977) and critical-point dried. Seeds were immersed in acetone and placed in an ultrasonicater for five minutes and air dried. Pollen grains were pipetted onto filter paper and air dried. Floral and seed specimens were sputter-coated with gold or gold-palladium and then mounted on alu- minum stubs with colloidal graphite. Pollen grains were mounted by inverting finely polished stubs covered with Mikrostik® onto filter papers with the dried pollen and then sputter-coated. OBSERVATIONS HABIT Brasenia is a rhizomatous, aquatic perennial. The rhizomes bear axillary buds, adventitious roots, and leaves at each node. Leaves are alternate, long- petioled, and centrally peltate. Young leaves are involute in bud. Mature leaf blades are floating and oval to elliptic with entire margins. Submerged stems, petioles, and abaxial leaf surfaces are heavi- ly coated with a layer of transparent mucilage, as are young plant parts such as axillary buds and juvenile leaves (Fig. 2). Flowers of Brasenia are about 2 cm in diameter, dull purple, and emergent. They are borne singly on long, mucilage-coated peduncles and possess linear-lanceolate perianth parts with three petaloid sepals and three petals. All perianth members bear antrorse, adaxial trichomes (Fig. 3). e androecia of flowers within the study site are composed of 24-33 stamens with filiform fil- aments and four apical microsporangia. Dehiscence is latrorse in the study populations. Pollen grains are elliptic, monosulcate, and have faintly scabrate ornamentation (Fig. 4). The pollen-ovule ratio of Brasenia is 9,238 + 625 (95% C.I., N = 11), and 98% of the pollen is viable. Flowers are hypogynous with apocarpous gy- noecia of 10-14 carpels in Toledo Bend popula- tions. These are characterized by relatively short styles and by linear, extremely papillate stigmas with abaxial stigmatic crests (Figs. 5, 6). Ovaries contain one or two anatropous and crassinucellate ovules. Placentation ranges from laminar to dorsal (Richardson, 1969) to median (Ito, 1986a). Fruits are aggregate and subtended by a per- sistent perianth. Each simple fruit is indehiscent, one- or two-seeded, and surrounded by a leathery pericarp. Peduncles bearing aggregate fruits ab- scise, float to the surface, and drift. Eventually, simple fruits detach from the receptacle, float on the water's surface where they are carried by wind and wave actions, and then sink to the pond bottom. The seeds are ovoid (Fig. 7) and at their apex possess a pyramidal structure with a central mi- cropyle (Fig. 8). The seed coat surface is composed of irregularly digitate cells. Seeds contain small amounts of endosperm, copious perisperm, and haustorial tubes. Embryo broad hemispherical cotyledons (Fig. 9). s are minute, with two FLORAL CYCLE The floral structure of Brasenia is diurnally emergent over a three-day period. On the first day of emergence, the flower is in bud and covered with mucilage (Fig. 10). Individual flowers bloom for two consecutive days. On the first day, flowers are morphologically pistillate; they are character- ized by short, undehisced stamens and elongate, papillate stigmas radiating outward over the re- I 2-8. grain. Scale abundant papillae an crest (SC). Scale bar — 5 Morphological and tsa sbg features of Brasenia. —2. Axill abaxial surface of older leah p Scale bar Ap er surface of vo AME numerous antrorse brc nga) - 4. SEM r= m.—5. SEM of second-day flower showing carpel mo A n Vignati crests. I bar — 1 mm M of seed. Scale ary bud and juvenile leaves; = ] cm.—3. SEM of of monosulcate pollen UP. note linear stigmas with is derg stigma; note abaxial stigmatic = ] mm.—8. SEM of seed apex; note digitate cells of seed surface and polygonal Ke. sr da surface. dnd bar — 500 , um 784 Annals of the Missouri Botanical Garden FicunEs 9-14. Scale bar = 250 um.— 10. undehisced anthers and radiat ting stigmas. —]12. anthers, and centrally aggregated stigmas.— 13. Age variation in lengths of stigmas. (Figures 10—14, scale bars flexed perianth (Fig. 11). Flowers submerge at the end of the first day. On the second day, the flowers are morphologically staminate; the filaments have elongated, elevating the dehiscent anthers to a po- sition. at now centrally aggregated or above the stigmas (Fig. 12). The perianth occupies a position similar to that of first-day flowers. At the end of the second day, flowers again submerge. Third-day flowers remain submerged but occasionally occur at the water surface. These flowers are nonfunc- tional, characterized by closed perianth parts with protruding senescent anthers. Aggregate fruits de- velop below the water surface (Fig. 13). Fruit de- velopment occurs in four to six weeks early in summer and as quickly as two weeks in the late summer. During the course of fieldwork, morphological variation was observed among flowers. Aside from typical features described above, several Brasenia flowers had short stigmas which did not conspic- uously radiate over the perianth (Fig. 14) Individual flowers vary in position above and below the water level. General trends of the floral Embryo and flowers of Brasenia. —9. SEM of embryo with den dia al cotyledons (C). Floral bud; note mucilage-covered peduncle Fir —11 st-day flower showing short, Second-day flower showing al filaments, dehiscent Fir regate i oum st-day flower of Brasenia showing = l cm.) cycle can be identified, however (Fig. 15). Anthesis begins at 0630-0730 hours on both the first and second days. First- and second-day flowers reach maximum height above water at 1300-1400 hours. Second-day flowers are generally elevated to a higher position, with anther dehiscence occurring at 0830-1100 hours. As first-day flowers close, the gynoecia and peri- anths gradually change position. When flowers are at peak height, stigmas begin to arch centrally until they completely aggregate at the time flowers close and submerge. Perianth members begin to close gradually at 1300-1400 hours, with the corolla closing prior to the calyx. CAGING TREATMENTS Results from caging experiments are reported as percentage seed set (Table 1). The percentage values are conservative because they were calcu- lated assuming two ovules (seeds) per carpel (simple fruit) within carpels remaining on aggregate fruits at time of retrieval. They do not take into account 785 Volume 75, Number 3 1988 Osborn & Schneider Brasenia schreberi ‘parmu? 1a3nduoo som aam 11898 "'gg'z = 7) s4omojf Kop-panpi pup “Tp "9 = T7) siəmol/ Xvp-puooas *Z6`¿ = T7) ssamoy &np-isaj *90°9 = T7) spnq ‘Xpp yova wou sanjpa 770 fo sppasaqui aouapifuoo 956 aSD1oap Juasasdas s1Dq 10447 ]2a2] 1930M mo]əq 10 2aoqn uoinsod pup ASojoydsow Josoyf ur saduvyo Buipaisnjn emuoseig fo 2242 ]p10]] “SG INDIA 3/11 SINOH r skeq 1 QL- (ww) LHOI3H 786 Annals of the Missouri Botanical Garden TABLE 1. Percentage seed set from caging treatments conducted on separate trial dates. Experiment (N — 25) 9/23/86 6/19/86 7/03/86 7/31/86 5/18/87 Control 1. Typical flowers 2. Flowers with short stigmas Group A—Parthenocarpy (Transparent cages, emasculated) Group B— Autogamy' (Transparent cages, left undisturbed) Group C— Stigmatic receptivity 1. Transparent cages, emasculated, cross-pollinated first- day 2. Transparent cages, emasculated, cross-pollinated second-day Group D— Allogamy' 1. Geitonogamy (transparent cages, cross-pollinated first- day with pollen from same plant) 2. Xenogamy (transparent cages, cross-pollinated first- day with pollen from different plant) Group E— Anemophily 1. Mesh cages, left undisturbed 2. Mesh cages, emasculated 6.1% == = 28.7% — 1.1% 18.2% 0.5% = = ' Terminology follows Faegri & van der Pijl (1979). any carpels that had abscised prior to retrieval or carpels with only one ovule. The results of exclusion treatments indicate that apomixis and autogamy did not occur (Groups A and B). Self-pollination did occur in treatment B, however, based on the observation of pollen on the stigmas. Flowers artificially pollinated on the first day and second day of anthesis (Group C) produced seeds. Flowers pollinated on the first day of anthesis exhibited greater seed set than those pollinated on the second day. These data indicate first-day flow- ers are receptive to pollen and should be considered functionally pistillate. Second-day flowers are func- tionally staminate. The results of experiment D, determination of allogamy, revealed that Brasenia is compatible with pollen of both geitonogamous and xenogamous origins. Only one of three trials from the mesh cage treatments (Group E) set seed. It is believed that minimal seed set in flowers sub- jected to this treatment occurred because the cages alter the aerodynamics of the floral structure (Nik- las, pers. comm.) and the mesh screen probably inhibits the passage of pollen grains. The three control groups of morphologically "typical" flowers yielded varying seed set (Table 1). This variation may be attributed to differences in wind speed and distributional density of flowers on each sample date. Flowers with short stigmas produced few seeds despite their occurrence in the same population as the “typical” control group. pop yp group POLLEN DISPERSAL AND TERMINAL SETTLING VELOCITY Measurements with the Andersen particle sam- pler revealed that Brasenia pollen was airborne. Pollinometer experiments indicated the mean num- ber of pollen grains dispersed from second-day (pollen-releasing) flowers decreases with distance. he numbers of grains at subsequent 10 cm in- tervals on each trial date were significantly differ- ent, with the exceptions of the 0-10 and 60-70 cm intervals. These differences can be attributed to wind velocity. Average wind speeds on respective sampling dates differed by a factor of five. Dis- persion of pollen can be described exponentially (Fig. 16). A regression analysis of dispersed pollen grains exhibited correlation coefficients of r = 0.99 and r — 0.85 at 8 kph and 1.6 kph, respectively. Pollen on the stigmas of a variety of flowers was also observed. First-day flowers that were down- wind of and within 0.5 m of second-day flowers had abundant pollen on their stigmas. Those grains were primarily restricted to windward stigmas. Pol- Volume 75, Number 3 1988 Osborn & Schneider Brasenia schreberi 787 aso - 300 250 i 200 Ñ 150 + # POLLEN GRAINS BH 80kp.h. o 1.6k.p.h. DISTANCE (cm) Regression analysis of pollen grain deposition from pollinometer experiments. At 8 kph, h, y IGURE 16. 367.9* 10^ (0.02x), r value of 0.99; and at 1.6 k — 112.7*10^ (0.02x) , r value of 0.85. Poit bars represent 95% confidence intervals; Ay curves were computer generated. len-receptive first-day flowers that were 2-3 m from second-day flowers exhibited very low quan- tities of pollen, if any, on their stigmas. The ter- minal settling velocity of Brasenia pollen was cal- culated to be 7.7 + 0.8 (95% C.I., N = 21) cm/ sec. FLORAL DENSITY Of four sample plots, the mean numbers of flow- ers/m? + 95% C.I. were as follows: 110 + 22; 82 + 26; 73 + 17; and 36 + 14. FLORAL SECRETIONS Neutral red staining revealed the absence of nectaries. The numerous perianth trichomes stained indicated a secretory role. ULTRAVIOLET REFLECTANCE AND ABSORPTION The inner surfaces of dehisced anthers are UV reflective. Adaxial (emergent) leaf surfaces are UV absorptive. ANTHOPHILOUS VISITORS A variety of insects visit flowers of Brasenia, including Donacia cincticornis Newman (Coleop- tera; Chrysomelidae), Perigaster cretura Herbst (Coleoptera; Curculionidae), Notiphila cf. cressoni Cresson (Diptera; Ephydridae), Apis mellifera L. (Hymenoptera; Apidae), and various odonates. The most frequent and abundant visitor was Notiphila cf. cressoni. Notiphila primarily visited staminate flowers and there foraged for pollen either directly from the dehisced anther sacs or from the adaxial surface of the reflexed perianth where pollen ac- cumulates (Fig. 17). Grooming and copulatory acts were also frequently observed. Pistillate flowers were rarely visited by flies, but upon their brief visitations, members of Notiphila would typically land on the elongated stigmas or reflexed perianth (Fig. 18). Additionally, flies commonly foraged for 788 Annals of the Missouri Botanical Garden FIGURES 17, 18. foraging for pollen. — 18. Notiphila on stigma of first-day flower. Scale bars — the abundant airborne pollen that had become de- posited on the adaxial surfaces of leaves. Micro- scopic examinations of dipteran bodies indicated that pollen loads were minimal except around the bases of the legs where body hairs are more dense. The stickiness of Brasenia pollen was not deter- ther insects were only occasional floral visitors and cannot be considered pollinators. DISCUSSION FLORAL CYCLE Field observations of Brasenia confirm that an- thesis is diurnal, with individual flowers opening and closing for two consecutive days. First-day flowers are morphologically pistillate; second-day flowers staminate. Caging experiments (Table 1) indicate that stigmas of first-da ceptive to pollen and those of "red flowers y flowers are re- generally are not. Thus floral structure and func- tion are correlated. Caging experiments further reveal that Brasenia flowers set seed only with pollen from an allogamous origin (sensu Faegri & van der Pijl, 1979). Although individual flowers are self-pollinating, seeds are not produced. It is unlikely that a genetic mechanism of self-incom- patibility is responsible for the lack of seed set. Abundant seed production from geitonogamous pollinations is indicative of self-compatibility. In the populations studied, self-fertilization (autog- amy) is prevented by dichogamy. WIND POLLINATION In the populations studied, Brasenia is predom- inantly anemophilous. This conclusion is reached by correlating data and observations of the follow- Insect visitors of Brasenia. — 17. Post-dehiscent second-day p with Notiphila cf. cressoni ing: pollinometer experiments, floral densities, number of viable pollen grains produced, flower position above water level, terminal settling velocity of pollen, and floral morphology. The pollinometer experiments reveal that Brasenia pollen is dis- persed over a relatively short distance. Therefore, for successful pollination by wind, the flowers must be closely grouped and produce a significant amount of viable pollen. As has been shown, floral density can exceed 100 flowers/m* and viable pollen pro- duction > 200,000 grains/flower. The elevated position of second-day flowers above first-day flow- ers, together with a pollen terminal settling velocity of 7.7 cm/sec., further enhances the dispersal distance of Brasenia pollen. Direct observations of pollen on stigmas of first-day flowers downwind of pollen-releasing flowers at distances of < 1 m and > 2 m support the above conclusion. Large pollen—ovule ratios, as discovered in Brasenia, are also characteristic of wind-pollinated taxa (Faegri & van der Pijl, 1979). Whitehead (1983) identified several “idealized conditions” that need to be met by a plant species if wind pollination is to be successful. The floral and vegetative morphologies of Brasenia are adapted for anemophily. Table 2 summarizes those adaptations of Brasenia in comparison with other anemophilous plants. Floral anomalies (e.g., short stigmas), however, are not as well adapted for re- ception of wind-borne pollen. This accounts for the low seed set in flowers with short stigmas (Table 1) compared with typical flowers. Additionally, observations of infrequent insect visitations between first-day and second-day flow- ers, despite insect abundance, and minimal pollen loads negate entomophily as a primary mechanism of pollen transfer in this study site Volume 75, Number 3 1988 Osborn & Schneider 789 Brasenia schreberi TABLE 2. Comparison of Brasenia adaptations with Whitehead’s “idealized conditions” for anemophily. Whitehead Brasenia There is production of large numbers of pollen grains. Pollen grains possess appropriate webs. apud characteristics, are typically 20-40 pu size, and have terminal settling velocities of 2-6 cm/sec. The probabilty of pollen’s entrainment in mov- ing air is maximized due to flower and inflo- rescence structure and their location on the plant. Stigmatic surfaces are structured and positioned to maximize collection efficiency. Pollen release is timed within both the season o maximize the possibility of pollen capture by receptive conspecifics ownwind. There is a relatively close spacing of compatible plants. Plants possess a vegetational structure that is relatively open to minimize filtration of pollen by nonstigmatic surfaces. Greater than 221,700 pollen grains are produced per average flower. Pollen grains are elliptic, smooth, within “ideal” size range (36.3 x 47.6 x 36.9 um),' and have a terminal settling velocity of 7.7 cm/sec. Second-day (staminate) flowers possess elongated filaments that elevate dehiscent anthers; perianth parts are reflexe stigmas are centrally aggregated out of the path of ambient First-day (pistillate) flowers possess elongated, papillate stigmas which radiate outward over the Pom ii an in- ed surface area for pollen adheren Pistillate flowers are fully open and eds elevated lower than staminate flowers at the time of anther dehiscence. Flowers exhibit relatively dense distribution, up to 110/m* during peak anthesis, and are both geitonogamous and xenog- amous. Vegetative organs are either submerged or floating; the only emergent structures are floral. The range of wind velocities ensures pollen transport and minimizes its downwind disper- sion Pollen can be dispersed at least 70 cm at wind speeds as low 1.6 kph. ' Ueno & Kitaguchi (1961). OCCURRENCE OF INSECTS The frequency, abundance, and behavior of No- tiphila flies on second-day flowers and leaves are related to their life history and association with the other nymphaeaceous genera, Cabomba and Nym- phaea (Van der Velde et al., 1978; Van der Velde & Brock, 1980; Willmer, 1982), in the study site. Members of Notiphila, while foraging on nectar and pollen, function as the primary pollinators of Cabomba (Schneider € Jeter, 1982). In Nym- phaea the flies forage on pollen and occasionally sponge stigmatic fluid, but native halictid bees are pollinators of the diurnal Toledo Bend species of this genus. We view the relationship of Notiphila with Bra- senia as one of herbivory in which flies mainly forage for pollen and utilize the flower to groom and mate. The flies forage for pollen in anther sacs only after the majority of pollen grains have been wind dispersed and the inner UV reflective walls, which aid in the attraction of the flies, are exposed. As the flies move about the dehiscent androecium and pollen-covered perianth surface, a small quan- tity of pollen accumulates on their legs. This minute pollen load can be attributed at least in part to the smooth walls of the Brasenia pollen. Occasional visits to pistillate flowers may facilitate transfer of minimal pollen; therefore, Notiphila should only be considered an incidental pollinator. The occur- rence of flies and their role as incidental pollinators, together with floral variations such as shortened stigmas, suggest a possible evolutionary shift from anemophily to partial myophily (ambophily; Stelle- 4) man, POLLINATION BIOLOGY OF THE NYMPHAEACEOUS SENSU LATO FAMILIES This investigation provides the first documen- tation of anemophily in the Nymphaeaceae s.l. Traditionally, wind pollination has been interpreted as derived in the angiosperms (Whitehead, 1969; Faegri & van der Pijl, 1979). Dilcher (1979), Dilcher & Kovach (1986), and Crane & Dilcher (1984), however, have shown that several extinct lower- to mid-Cretaceous flowering plants had the reproductive morphology to accommodate polli- 790 Annals of the Missouri Botanical Garden nation by both wind and insects. The occurrence of anemophily in the Nymphaeaceae s.l., a taxon traditionally viewed as primitive and entomophi- lous, dictates that the phylogeny of this group be re-examined. Cabombaceae. Several studies (e.g., Kosakai, 1968; Goleniewska-Furmanowa, 1970; Buko- wiecki et al., 1972; Okada & Tamura, 1981; Ito, 1986a, 1987) suggest Cabomba and Brasenia share sufficient characteristics to warrant main- tenance of the Cabombaceae. From a classical viewpoint, it could be argued that the presence of perfect flowers, a perianth, the occurrence of for- aging flies, and some UV reflectance in Brasenia are remnants of a former entomophilous condition which has more recently specialized for wind pol- lination. An alternative hypothesis, that anemoph- ily is primitive, could also be argued. Support for this viewpoint comes from the fossil record in which Tertiary pollen of Brasenia exhibits the same mor- phological features as extant Brasenia pollen (Jes- sen et al., 1959). Although pollen alone is not a definitive indicator of pollination biology, modern Brasenia pollen is adapted for wind pollination (Table 2). Because fossil and extant pollen are morphologically similar, comparable functions can be postulated. We do not view these two hypotheses as com- pletely exclusive of each other. A third line of reasoning, that the cabombaceous ancestor had the reproductive morphology to facilitate both wind and insect pollination, could be advocated. An ancestor with dual pollination capabilities would explain the long anemophilous history as suggested by fossil pollen and the occurrence of the ento- mophilous features noted above. An extension of this third concept suggests that the pollination syn- dromes in extant Brasenia and Cabomba are spe- cialized. Phylesis in Cabomba, therefore, would have involved a reduction and stabilization in the number of androecial and gynoecial members, as evidenced by the vasculature studies of Ito (1986a), and the appearance of a distinct, colorful perianth possessing nectaries to enhance a myophilous pol- y (Schneider & Jeter, 1982). Here, pollination is achieved by the appropriate positions of the pollen-receptive stigmas in first-day flowers (Fig. 19L) and pollen-dehiscent anthers in second- day flowers (Fig. 19K) above the petaliferous nec- taries where Notiphila flies sponge secretions. The elongated stigmatic surfaces which radiate above the recurved perianth in first-day flowers of Bra- senia (Fig. 19N) enhance collection of airborne pollen. In second-day flowers (Fig. 19M), the el- lination evation of anthers above the recurved perianth promotes pollen dispersal. A recent cladistic analysis of the Nymphaeales by Ito (1987) indicates Ceratophyllum has asso- ciation with Brasenia and Cabomba. Les (1986) has proposed that the hydrophilous pollination mechanism of Ceratophyllum (Jones, 1931) is de- rived from anemophily. The occurrence of wind pollination within the Cabombaceae warrants con- tinued examination of the suggested relationships among Ceratophyllum, Brasenia, and Cabomba. Nymphaeaceae s. str. Some investigations (Collinson, 1980; Moseley et al., 1984) indicate Brasenia shares more affinities with the Nym- phaeaceae s. str. than with Cabomba. We do not support dismantling the Cabombaceae. It is our opinion that data from pollination biology, floral anatomy, anatomy of fossil and extant seeds, and embryology support a common ancestry hypoth- esis. The Nymphaeaceae s. str., unlike the Cabom- baceae and Nelumbonaceae, possess completely syncarpous gynoecia. When correlated with an evolutionary shift from hypogyny in Nuphar, through perigyny in Nymphaea, to epigyny in Barclaya, Euryale (Kadono & Schneider, in press), and Victoria, flowers of this family exhibit an over- all elaboration. This elaboration consists of an in- crease in size and number of floral parts and stig- matic surface area facilitated by radial extension of the carpels and by the appearance of specialized organs (e.g., staminodia and carpellary append- ages). This family, like the Nelumbonaceae, is en- tomophilous except for the cleistogamous flowers of Barclaya (Fig. 19D), Euryale (Fig. 19C), and some species of Nymphaea. The flowers of Nuphar (Fig. 191, J) have a close relationship with beetles of the genus Donacia, which complete their life cycle in association with the plant, during which time they facilitate polli- nation (Schneider & Moore, 1977). Diurnal flowers of Nymphaea and Ondinea are specialized for washing pollen from the body of foraging native bees. This is accomplished by the production of stigmatic fluid on the first day of anthesis (Fig. 19F, H; Conard, 1905; Schmucker, 1933; Prance & Anderson, 1976; Schneider, 1979, 1982a, b, 1983; Meeuse & Schneider, 1980; Schneider & Chaney, 1981; Capperino & Schneider, 1985). The pollination biology of nocturnally flowering species of Nymphaea is largely unstudied, but can- tharophily is known (Cramer et al., 1975; Wier- sema, 1987). Large, nocturnal flowers of Victoria are adapted for pollination by Cyclocephala beetles. Beetles Volume 75, Number 3 Osborn & Schneider Brasenia schreberi 1988 NYMPHAEACEAE Barclaya Euryale Com «c í AOE oF À Ondinea CABOMBACEAE Nuphar NELUMBONACEAE = s Nelumbo N H, J; secon Brasenia Comparative morphologies of p day (pistillate) x second- fos (staminate) ET of the str. (first-day, B, C, le. FIGURE 19. Cabombaceae (first-day, L, N; second-day ymphaeaceae day, A, E, G, I), and pi ce Nado. Vis d P; second-day, O). LE flowers not drawn. to. scale 792 Annals of the Missouri Botanical Garden are attracted by the white color and fragrance of first-night flowers. The diffusion of fragrance is accelerated by a rise in floral temperature due to the thermogenicity of carpellary appendages (Fig. 19B). Pollination is achieved when pollen- covered within the flower, beetles forage on starch-rich carpellary appendages (Fig. 19A). As beetles emerge from second-evening flowers they become dusted with pollen from the now-dehiscent anthers and, once again, are attracted to first- evening flowers (Knoch, 1899; Valla & Cirino, 1972; Prance & Arias, 1975; Schneider, 1976; Lovejoy, 1978). Nelumbonaceae. | Separation of Nelumbo from the Nymphaeaceae s.l. into the Nelumbonaceae has been justified on numerous grounds, including seed anatomy (Collinson, 1980) and floral anato- my/morphology (Moseley & Uhl, 1985; Ito, 1986b, 1987). Despite this separation, Nelumbo displays a floral behavior similar to that of chas- mogamous flowers of the Cabombaceae and Nym- aceae s. str. in which first-day flowers are pistillate (Fig. 19P) and second-day flowers are staminate (Fig. 190). Flowers of Nelumbo are prin- cipally adapted for cantharophily (Schneider & Buchanan, 1980), although bees and flies (Sohmer & Sefton, 1978) also transfer pollen. Moseley & Uhl (1985) suggested that the flower of Nelumbo represents an evolutionary elaboration in response to cantharophily. Comparative studies of the reproductive biology of Cabombaceae, Nymphaeaceae s. str., and lumbonaceae, when viewed in conjunction with structural aspects of the flower, suggest that evo- lution within these families has involved both floral reduction and elaboration as a result of adaptive evolution in response to abiotic and biotic vectors. This concept is consistent with other angiosperm taxa as indicated by the fossil record (Basinger & Dilcher, 1984). LITERATURE CITED ru F.S. 1969. linus no and function i asenia schreberi. Rhodora 17-433 inu A. A. 1958. in ew Lu for ai collec- tion, sizing and enumeration of viable airborne bac- teria. J. Bacteriol. 76: 471-484. BASINGER, J. F. & D. L. DiLCHER. c Ancient bi- sexual flowers. Science 224: 51 51: grain structure. Bot. Zurn. (Moscow & Leni Us 68: 1177-1183. [In Russian.] BENTHAM, G. & J. D. Hooker. 1862. "ipis Planta- rum, Volume I. Reeve & Co., Londor BuKOWIECKI, H., M. FURMANOWA & H. DD 1972. The numerical taxonomy of Nymphaeaceae Bentham et Hooker. I. Estimation of taxonomic distance. Acta 327 Polon. Pharm. 29: 319 & . The numerical taxonomy of Nymphaeaceae Bentham et Hooker. II. Estimation of similarity coefficients. Acta Polon. Pharm. 31: 385-391. CAPPERINO, M. E. € E. L. ScHNEIDER. 1985. Floral iology of Nymphaea yeas Zucc. (Nymphae- aceae). Aquatic Bot. 23: 83-93. CR R. 1891. eia eae. In: A. Engler & K. Prantl, Die Natürlichen Pflanzenfamilien 3: 1- CHEN, W. & S. ZHANG. 1986. Peltate leaf development in Brasenia schreberi. Acta S: Sin. 28: 341-348 [In Chinese with English summar CHIFFLOT, J. B. J. 1902. tard à l'étude de la classe des PAL M Ann. Univ. Lyon, N.S. 1 Sci. Med. 10 8. CHRYSLER, M. A. ree The winter buds of Brasenia. CLARK, G. C. & M. R. Jones. 1981. The northwest European pollen flora =a i S Rev. Pa- laeobot. Palynol. 33: 51- COLLINSON, M. E. 1980. Reset and Tertiary seeds of Nymphaeaceae sensu lato with a revision of Bra senia ovula "ea ) Reid and Chandler. Ann. Bot. 46: 603-632 es H. S. 1905. The water lilies. A monograph Es genus Vymphaea. Publ. Carnegie Inst. Wash. 19-123 R e T. 1906. The embryogeny of some Cuban Nymphaeaceae. Bot. Gaz. (Crawfordsville) 42: 376- 392. 1909. Notes on the embryology of the Nym- phaeaceae. "d: Gaz. (Crawfordsville) 48: 56-61. Corner, E. J. 1976. The Seeds of Dicotyledons, Volumes x 2. A gd Univ. Press, New York CRAMER, J. M., A. D. J. MEEUSE & P. TEUNISSEN. 1975. note on de pollination of nocturnally flowering species of Nymphaea. Acta Bot. Neerl. 24: 489- CRANE, P. R. & D. L. DILCHER. 1984. Wan an early angiosperm rk. axis us the mid-Creta- ceous. Ann. Missouri Bot. Gar . 384-40 2. 2 OE IST, A. 1981. de al hh of Classi- n of Flowering Plants. Columbia Univ. Press, ; York. CRUDEN, R. W. 1977. Pollen-ovule ratios: a conser- Evolution 31: 32-46. DEN HARTOG, A. 1970. Ondinea ron a new genus for the Nymphaeaceae. Blumea 18: 413- isi DiLcHER, D. L. 1979. Early oa eproduction: an introductory report. Rev. Paleobot. Palynol. p 291-328. & W. L. KovacH. 1986. Early angiosperm reproduction: Caloda delevoryana gen. et sp. nov., a new fructification from the Dakota formation (Ceno- manian) of Kansas. Amer. J. Bot. 73: 1230-1237. DonorkEv, P. I. 1984. The taxonomy and history of los genus Brasenia (Cabombaceae). Bot. Zurn. (Mos- ow & vu 69: 137-148. [In Russian.] m K. 1965. Plant Anatomy, 2nd edition. John i k ork. Farcri, K. & L. VAN DER PUL. 1979. The Principles of Pollination Ecology, 3rd edition. Pergamon Press, New York Volume 75, Number 3 1988 Osborn & Schneider 793 Brasenia schreberi GOLENIEWSKA-FURMANOWA, M. 1970. Comparative leaf anatomy and alkaloid content in the Nymphaeaceae Bentham and Hooker. Monogr. Bot. 31: 5-54 GwYNNE-VaucHN, D. T. 97. On some points in the morphology and anatomy of the Nymphaeaceae. Trans. Linn. Soc in ai Bot. 5: 287-299. Hair, A. 1978. w paleobotanical records for Ren x re layanan cal New Phytol. 81: HiLL, E T "un An observation on the water-shield (Brasenia peltata). Pl. World 3: 153. Ikuse, M. 1955. General p list of pollen grains in Japan (3). J. Jap. Bot. 3 -113. [In Japanese.] o, M. 6a. Studies in in snl morphology and anatomy of Nymphaeales III. Floral anatomy of Bra- enia schreberi Gmel. and Cabomba nen A. hs Bot. D (Tokyo) 99: 169-18 TN die . Studies in the floral S Ss and 1959. The interglacial deposit near Gort, Co., Galway, Ire- roc. Roy. Irish Acad. 60: 1-78. Teas, "E. N. 1931. The morphology and biology of Ceratophyllum demersum. Stud. Nat. Hist. Iowa Univ. 13: 11-55. KADONO, Y. . L. SCHNEIDER. The life history of Eu ryale ferox Salisb. in southwestern Japan with special reference to reproductive ecology. Pl. Sp. Biol. 2 (in press Kakura, M. & y MISAKI. 1979. The A ara mi of “junsai” (Brasenia schreberi J. F. Gmel.) muci- lage; fragmentation analysis by successive Smith deg- siet e partial acid hydrolysis. Agric. Biol. 76. Kaur, R. B. 1978. p observations on floating leaves. Aquatic Bot. 2: -234. issn, I. A. 1893. The echar hairs of Brasenia vw 2e Proc. Acad. Nat. Sci. Philadelphia: ans P. "1968. Morphological and embryological studies in the Nymphaeaceae. II. Brasenia schreberi Gmel. and Nelumbo nucifera Gaertn. Austral. J. Bot. 13: 379-387. KNocu, E. 1899. Untersuchungen über die Morpho- logie, Biologie und Physiologie der Blüte von Victoria regia. Biblioth. Bot. 9: 1-67. Kosakal, H. 1968 of Various Nymphaeaceae. of California, Santa Barbara, Califor KRISTEN, U. 4. Zur Feinstru ktur a submersen Drüsenpapillen von Brasenia schreberi und Cabom- a caroliniana. ue 9: 36-44. E. Söp A The Comparative sum Anatomy M.A. = . University totes O. 8 ERG. 1927. Uber die Chro- mosomenzahlen sins Nymphaeaceen. Acta Horti g. 9: 85-104. Lawson, G. 1888. On the E M Proc. & Trans. Roy. Soc. Canad. 6: 97- Les, D. H 86. The origin and beg of the Cer- atophyllaceae. Amer. J. Bot. 73: 773-774. [Ab- strac d e 5, R. " C FAR & C. R. STOCKING. 1977. light a iin microscopy with pt viu del and 2,2-diethoxypropane. Amer. J. Bot. 64: 601- 605. TE T. E. 1978. Royal water lilies, truly amazo- n. Smithsonian 9: 78- .& Hara. 1985. On the problems related D the pee a of water-shield (Brasenia i and of Haga Tor aris Sci. Rep. Fac. Agric. Ibaraki Univ. 33: In Japanese. PR B. J. D. & E. L. SCHNEIDER. 1980. mphaea Ny revisited: a preliminary communication. larael $ Bot. 964. porre studies of Nym w & Leningrad) 49. Er of the seed coat mphaea "ar radsk. Univ, Ser. Biol 2: 1214125. Morphological studies of the ure of the stamens. Phy- tomorphology 8: 1-29. ——— & N. W. UnL. 1985. Gi ep studies of the Nymphaeaceae sensu lat . The anatomy of the flower of Nelumbo. px Jahrb. Syst. 106: 61-98. — L. J. Menta, P. S. WiLLiAMSON & H. Kosakal. 984. Morphological studies of the Nymphaeaceae sensu lato. X ontributions to the vegetativa and floral structure of Cabomba. Amer. J. Bot. 71: 902- NAKAHARA, H. 1940. Uber den Schleim von Brasenia eri Gmel. Agric. Chem. Soc. Japan J. 16: wind pollination. Amer t OGDEN, E. C. 19 natomical patterns u ic vascular plants of New York. New York State Mus. Sci. Serv. Bull. 424: 1-133. $ . 1981. Karyomorphological study of the Nymphaeales. J. Jap. Bot. 56: 367- 37 PosTEK, M. T. & S. C. Tucker. 1976. A new short chemical dehydration method for light microscopy preparations of plant material. Canad. J. Bot. 54 872-875 PRANCE, G. T. & A. B. ANDERSON. 1976. Studies on the floral biology of Meses Nymphaeaceae. 3 Acta Amazonica 6: 163- & J. R. ARIAS. hu IN study of the floral biology of Victoria amazonica (Poepp.) Sowerby Nymphaeaceae). Acta Amazonica 5: 109-1 RACIBORSKI, M. 1894. Die ee der Cabombeen und d euren Flora 78: 244-279. Rao, T. A. & B. C. BANERJEE. 197 9. Oh foliar sclereids in the Nymphaeaceae sensu lato and their use in familial classification. Proc. Indian Acad. Sci. 88B: 3-422 RICHARDSON, F.C. 1969. Morphological studies of the ymphaeaceae. IV. Structure and development of the flower of Brasenia schreberi Gmel. Univ. Calif. Publ. Bot. T. -101. RIEMER, D. N. f TorH. 1970. Chemical com- position of des species of Nymphaeaceae. Weed Sci. 18: 4-6 SCHILLING, A. J. Anatomisch-biologische Unter- suchungen über die ERI UNUM der Wasser- nzen. Flora 78: pfla SCHMUCKER, T. 1933. Zur Blütenbiologie tropischer 794 Annals of the Missouri Botanical Garden Nymphaea- VU. II. Bor als entscheidender Faktor. Planta 18: 641-650. ScHNEIDER, E. L. io Morphological studies of the Nymphaeaceae. VIII. The flower A Victoria Schomb. J. Linn. Soc., Bot. 72: 115-14 , . Morphological id of the Nym- phaeaceae. X. The seed of Ondinea purpurea den Hartog. Bull. Torrey Bot. Club 105: 192-200. 979. Pollination biology of the Nymphae- aceae. Proc. IVth Int. Symp. on nrw Md. Agric. Exp. Sta. Special Misc. Publ. 1: 419-429. —— ——-. 1982a. The floral biology of s fin ele- gans (Nymphaeaceae) in Texas. Aquatic Bot. 12: 197-200 Observations on the pollination biology Nymphae- phaeaceae. XIV. Gross morphology and floral biology of Ondinea purpurea den Hartog. Austral. J. Bot. 31: 371-382. & J. D. BUCHANAN. 1980. Morphological stud- ies of the Nymphaeaceae. XI. The gis biology of Nelumbo pentapetala. Amer. J. Bot. 6 & T. CHANEY. 1981. The lora biology of haea odorata (Nymphaeaceae). Southw. Nat- uralist 26: ` ens & J. M. JErER. 1982. Morphological studies of the Nymphaeaceae. XII. The floral biology of 69: 141 PFA caroliniana. Amer. J. Bot. 419. L. A. Moore. 1977. Morphological ne of the Nymphaeaceae. VII. The floral biology of Nuphar lutea subsp. macrophylla. Brittonia 29: 88. 99. SCHRENK, J. 1888. On the histology of the vegetative na ot Brasenia peltata Pursh. Bull. Torrey Bot. Club. 15: 28-47. SEVILLA, F., L A. DEL Rio & E. HELLIN. 1984. Su- peroxide dismutases from a Citrus plant. J. Pl. Phys- iol. 116: 381-388. SOHMER, S. H. & D. F. SEFTON. 1978. The reproductive biology of Nelumbo pr inq d on the upper Mississippi Riv nsects as- sociated with the transfer of "bn Brittonia 30: -364. SRODON, A. 1935. Studies of diluvial plant remains of the family Nymphaeaceae. Starunia 7: 1-10. [In Polish. ] STELLEMAN, P. . Reflections on the transition from wind pollination to ambophily. Acta Bot. Neerl. 33: 497-508. TAKHTAJAN, A. 19 Outline for the classification of flowering plants (Magnoliophyta). Bot. Rev. 46: 225- TOKURA, ^ 1937. On the d of Brasenia. Jap. . 13: 51-61. [In Japanese.] 19 TRALAU T. Extinct aquatic plants of Europe. Bot. Not. 112: 385. Trott, W. 1933. Beiträge zur Morphologie des Gy- naeceums. IV. Über das aceen. Planta 21: 447-4 oo der Ny. ae 1962. On the fine structure of the pollen walls Ueno, J. of An mie II. Victoria. J. Biol. Osaka City Univ. 13: 99-104. S. Kiracucut. 1961. On the fine structure of the pollen walls of a Md rx mphae- aceae. J. Biol. Osaka City Univ. 12: 83-90. VALLA, J. J. & D. R. CiRINO. Se jou floral del irupé, Victoria cruziana D'Orb. (Nymphaeaceae). Darwiniana 17: 447-500 Tu. C. M. Brock. 1980. The life history and habits of Notiphila brunnipes " bi tu a "Tijdschr. — Deel 123: 105-127. . HEINE & . PEETERS. 1978. à lowers ul Dutch Nymphaeaceae. as a habitat for cts. Acta Bot. Neerl. 27: 30 nodus r S. 1967. “wami semeistva Nym aeaceae na osnove analiza 5. stroya. ^vi 5 entr.-Kavkazsk. Otd. Vsesojuzn. Bot Obsc. VOGEL, S. p Saal organs of orchid flowers and their relation to insect pollination. Proc. 5th World Orchid Conf., Long Beach, California: 253-259. WALKER, J. W. 1976a. Comparative pollen morphology and phylogeny of the ranalean complex. In: C. B. Beck (editor), Origin and Early Evolution of Angio- die Columbia Univ. Press, New York. 976b. Evolutionary significance of the exine in de pollen of D angiosperms. In: I. K. Ferguson & J. er (editors), The ese say Significance of i. ; Linn. Soc. Symp. Ser. 1 251-308. & J. A. DoyLE. 1975. The bases die er po ne palynology. Ann. Missouri Bot. Gard. 6 a A. 1893. Über die fossilien Nymphae- aceen Gattungen Holopleura und Cratopleura Web. Beitrage zur pamenanstonie der Nym- phaeaceen. Bot. Jahrb. Syst. 18 n D. R. 1969. Wind pollination i in the an- giosperms: evolutionary and environmental consid- erations. Evolution 23: 28-35 1983. Wind pollination: some ecological and evolut utionary ad prn n: editor), Pol- lination s cademic m Orlando, Florida. WIERSEMA, J. 1087, A monograph of Nymphaea subgenus ffir i s Syst. Bot. Monogr. 16 WILLMER, P. G. Tr - Hygrothermal determinants of insect activity ues la n of water-lily leaves. Ecol. Entomol. 1-231. " JSE, R. P. e living representatives of the Bul. Torrey Bot. Club 59: 313- 340 Woop, C. E. 59. The genera of the Nymphaeaceae Zi Ceratophyllaceae in the Em United ates. J. Arnold Arbor. 40: 94-112. 932. 1 en I. Pollen of n River flora. REPRODUCTIVE BIOLOGY John H. Wiersema? OF NYMPHAEA (NYMPHAEACEAE)! ABSTRACT Several reproductive agi have evolved within the genus Nymphaea. Sexual reproduction is mostly o to several days (depending on the species), with pollen release usually considerable variation in the timing of flower opening, floral odor, flower color, and th various flower parts. These differences may contribute to genetic poc between species both through temporal separation of flowering and attraction of different pollinators. While sexual reprod dependent on iiid or uH oon ii ned LAS ave a iis ii to promote autogamy. N. Nymphaea is known to be agamosperm but s l other modes of asexual reproduction are exhibited: detachable tubers, idol formation, and petra a floral and kiar: tissue. Wide-ranging tropical species reproductive alternatives to outcrossing; of these reproductive alternatives is autogamy. W ang on ca for sexual raky probably avoid pollinator limitations by utilizing a broader range of pollinators Nymphaea is the largest and most widely dis- tributed genus in the Nymphaeales. About 40 species of waterlilies, as they are commonly known, are distributed on all continents except Antarctica. The classification of Conard (1905), still generally accepted, recognizes five subgenera: the north-tem- perate subgenus Vymphaea, neotropical subg. Hy- drocallis, and paleotropical subg. Lotos of the syn- carpous group and the Australian subg. Anecphya and pantropical subg. Brachyceras of the apocar- pous group. The success of this genus as deter- mined by number of species and distributional range is due to several factors. Its age, dating probably from mid Tertiary (Collinson, 1980), may be one factor; however, other nymphaeaoid genera of comparable age have not demonstrated the same degree of evolutionary divergence. Another factor is its specialized but widely available ecological niche. Waterlilies almost invariably inhabit still or gently flowing water over rich organic substrates. In stable aquatic habitats they root in water too deep for competing emergent vegetation. Their floating leaves, in addition, outcompete submersed leaves for light. A few other genera are competitors for this niche, including primarily nymphaeaceous genera such as Nuphar, Brasenia, and Nelumbo, and the distantly related genus Vymphoides. None of these, however, has attained the species radiation and global distribution of Nymphaea According to Gupta (1978, 1980) polyploidy, structural chromosome changes, and gene muta- tions have played important roles in the evolution of Nymphaea. The poor development of isolating mechanisms is also mentioned as being responsible for the high incidence of natural and artificial in- terspecific hybridization. However, such mecha- nisms are not altogether absent. Certainly ecolog- ical factors help isolate certain species from others, such as the alkaline-tolerant N. tuberosa Paine rom the more acidophilic N. odorata Aiton, or the slightly halophilic V. rudgeana G. Meyer from less salt-tolerant species, or the riparian N. pota- mophila Wiersema from lacustrine species. Geo- graphical isolation has also been a factor, such as separating N. alba L. of Europe from the North American N. odorata. This paper, in discussing the reproductive bi- ology of Nymphaea, focuses on several potential external e for 3c ara isolation. In- ternal from Grant, 1981) no doubt are important in promoting the hybrid breakdown and sterility often observed in /Vym- phaea (Gupta, 1978) but are outside the scope of this discussion. Differences within and among the = | A portion of the research reported in this study was supported by National Science Foundation grant DEB- 8111 024. * Department of Botany, University of Maryland, College Park, Maryland 20742, U.S.A. ANN. Missouni Bor. GARD. 75: 795-804. 1988. 796 Annals of the Missouri Botanical Garden subgenera with regard to floral odor, flower color, the timing of floral responses, and the form and function of various flower parts may contribute to genetic isolation between species. Isolating mech- anisms are probably more important for tropical and subtropical species, since their densities tend to be greater than those in temperate areas. After focusing on these potential isolating mechanisms, several adaptations relating to asexual reproduction in Nymphaea will be discussed, thereby permitting an evaluation of the various reproductive strategies found among waterlilies. FLORAL BIOLOGY Some general aspects of floral biology in Nym- phaea should be presented. The flowers consist of 4 sepals, 7-40 petals, 20-700 stamens, and 5- 47 carpels, the latter forming a ring embedded in cup-shaped receptacular and appendicular tissue to which the appendicular organs are attached lat- erally. The upper surface of each carpel contributes a ray of stigmatic tissue to the stigmatic disk, which tops the ovary, and this ray usually terminates abaxially in a free appendage termed the carpellary appendage or carpellary style. Sexual reproduction is mostly protogynous. First- day flowers generally open slightly less than on later days, with the stamens and carpellary ap- pendages, if present, spreading to permit insect visitors access to the stigmatic region. The tips of the stamens form a circular wall or palisade around a central pool of stigmatic fluid. Although variable in quantity, I have observed this fluid in all species thus far examined. It reportedly contains a sur- factant primarily responsible for washing pollen from the bodies of insect visitors, but also possibly contributing to their frequently observed drown- ings. In second-day flowers the sepals, petals, and most of the stamens reflex fully. Anther dehis- cence occurs in second-day flowers, with the stig- matic fluid drying up at or by this time. Insects are generally free to forage for pollen due to the dryer nature of the stigma and the incurving of the inner stamens and carpellary appendages over the stigmatic region. Pollen-covered insects that subsequently visit first-day flowers effect pollina- tion. In many species first-day flowers open slightly later than on subsequent days. This ensures that insect visitors will first visit pollen-releasing flowers prior to their entrance into pollen-receptive flowers (Schneider, 1979; Meeuse & Schneider, 1980; and pers. obs.). Floral odor. Although numerous authors have commented on the differences in floral odor among various waterlily species (Conard, 1905; Cramer et al., 1975; Prance & Anderson, 1976; Meeuse & Schneider, 1980), which range from inodorous to lightly fragrant or strongly pungent, nothing is known of the chemistry of the odor-producing com- pounds or their potential in attracting certain taxa of pollinators. The importance of floral odor seems to be greatest for night-blooming species, which can rely less on flower color to attract pollinators. Most members of the night-blooming subg. Hy- drocallis emit considerable odor that is detectable for some distance. These species are probably pol- linated by Coleoptera (Cramer et al., 1975; Prance & Anderson, 1976; Prance, 1980; Wiersema, 1987), in contrast to members of the day-blooming subgenera Anecphya, Brachyceras, and Nym- phaea, which appear to be pollinated by Hyme- noptera or Diptera (Van der Velde et al., 1978; Schneider & Chaney, 1981; Schneider, 1982a, b; Capperino & Schneider, 1985). The latter group has typically far less odorous flowers, with odors appearing to differ qualitatively from those of subg. Hydrocallis. In subg. Anecphya, no odor is known (Conard, 1905; Schneider, 1982b). Strong odors have not been observed in N. lotus L. of the night- blooming subg. Lotos, although nothing is known regarding insect pollinators in this subgenus. Schneider (19822) has hypothesized that in subg. Hydrocallis the highly developed carpellary ap- pendages may help volatilize odors. A darkening of these appendages and sometimes the staminal bases is frequently observed as flowering pro- gresses, suggesting that chemical activity may be taking place. Olfactory tests with N. jamesoniana Planchon support this, but additional tests are need- ed. In the inodorous subg. Anecphya, carpellary appendages are absent. Flower color. Considerable variation in petal coloration is evident among waterlily species. Flow- er color is, however, relatively constant among night-blooming species, with all species of subgen- era Hydrocallis and Lotos having white or nearly white petals. An exception to this trend is found in the red-flowered N. rubra Roxb. ex Andrews, but recent evidence suggests that this is an apo- mictic "species" that may not undergo sexual re- production (Mitra & Subramanyam, 1982). Cer- tain species of the other syncarpous subgenus, the diurnally flowering subg. Nymphaea, have also yielded pink- to red-flowered forms under rare cir- cumstances (Conard, 1905; Elkins, 1970; Erixon, 1980), white being the usual petal color in all species involved. However, N. mexicana Zucc. of this subgenus has yellow petals. The neotropical Volume 75, Number 3 1988 Wiersema 797 Nymphaea Reproductive Biology members of subg. Brachyceras are also primarily white-petaled (pale blue fading to white in N. ele- gans Hook.) those of the paleotropics are pre- dominantly blue-petaled, although both white- and yellow-petaled species are known (Conard, 1905; Mendonça, 1960; Hutchinson & Dalziel, 1966). Little evidence exists to evaluate the importance of flower color in Nymphaea in attracting different pollinators. The presence of blue and yellow color forms may be an adaptation favoring pollination by hymenopterans or dipterans, which pollinate similarly colored flowers in other plant groups (Proctor & Yeo, 1972). Schneider and colleagues have examined pollinators of the white-flowered /V. odorata (Schneider & Chaney, 1981), the blue- flowered N. elegans (Schneider, 1982a), and the yellow-flowered N. mexicana (Capperino & Schneider, 1985). Insect visitors to all of these diurnally flowering species included Hymenoptera, Diptera, and Coleoptera, indicating that the same classes of pollinators appear to be attracted to all three flower colors. As all of these observations have come from Texas, more observations on pol- linators throughout the ranges of these species, especially in situations where they coexist, are needed in order to evaluate better the importance of flower color as a factor in reproductive isolation. Valuable observations toward a resolution of this question could be made in south-central Africa where all three color forms occur. Temporal responses. Timing of floral re- sponses may be important in determining which pollinators are associated with a particular species, insofar as activity levels of potential pollinators may vary throughout the day or night. Species that utilize the same pollinators could still be reproduc- tively isolated through temporal separation of flow- ering, which seems to provide a partial if not total barrier to gene flow between certain species. For day-blooming North American species of subg. Nymphaea (Fig. 1A), flowers of the widespread N. odorata open just after dawn and close around noon or shortly thereafter (Conard, 1905, Penn- sylvania; Schneider & Chaney, 1981, Texas; Wiersema & Haynes, 1983, Alabama). In the northern portion of its range its distribution over- laps that of N. tetragona Georgi, whose flowers are open from just before noon to around 5:00 P.M. (Conard, 1905, cultivation in Pennsylvania). The sequence of floral opening proceeds from the sepals and outer petals gradually inward to the inner stamens and requires 1⁄2—1 hour, as does floral closure. First-day flowers (pollen-receptive) gen- erally open slightly later than those of subsequent days. The important times to consider are the pe- riods when the stigmas of first-day flowers are ac- cessible to potential pollen donors and the stamens of later flowers accessible to potential pollen ac- ceptors. Allowing for the adjustments just men- tioned, it is probable that complete temporal sep- aration exists between /V. odorata and N. tetragona, although field data are needed from areas of sym- patry to support this assumption. Similarly, N. mex- icana, whose flowers open ca. 11:00 A.M. and close ca. 4:00 P.M. (Conard, 1905, cultivation in Penn- sylvania; Wiersema & Haynes, 1983, Alabama; Capperino & Schneider, 1985, Texas), achieves some degree of temporal separation from N. odo- rata, with which it is sympatric over parts of the southeastern United States. overlap appears to be present in the flowering schedules of these two species, which have differ- Less than an hour ent-colored flowers, and occasional natural hybrids occur (Ward, 1977, Florida and Georgia; Wier- sema & Haynes, 1983, Alabama). On the other hand, the flowering schedules of /V. odorata and the closely related N. tuberosa (Conard, 1905, Pennsylvania) overlap considerably (3-4 hours), and not surprisingly, their taxonomic relationship remains confused (Conard, 1917; Monson, 1960; Williams, 1970). Available information on the day-flowering species of subg. Brachyceras in the New World (Conard, 1905; Prance & Anderson, 1976; Schneider, 1982a) indicates considerable overlap in the flowering schedules of the three species (Fig. 1B). Data are lacking for most Old World taxa of this subgenus. Considerable information on flowering schedule in the nocturnally flowering, neotropical subg. Hy- drocallis has been gathered from outdoor culti- vation in Alabama (Wiersema, 1987). Cultivated samples of those species obtained from two or more regions exhibited little variation in flowering sched- ule. Among most species sufficient overlap in flow- ering time was observed to negate the importance of this feature as a barrier to genetic exchange between species (Fig. 1C). This is particularly true of species that complete flowering before midnight. Species, such as N. amazonum C. Martius & Zucc. and N. prolifera Wiersema, whose flowers remain open after midnight, are provided some degree of genetic isolation. In N. prolifera second-day flow- ers, although opening at dusk, remain mostly in- accessible to pollinators due to delayed reflexing of the inner petals and stamens until after midnight, when flowers of most other species have closed. In N. amazonum floral responses reach their greatest specialization. First-day flowers open fully and close during the two hours just preceding dawn. Second- day flowers open at dusk, remain open throughout 798 Annals of the Missouri Botanical Garden A Day of Taxon Flowering 6 7 8 9 10 11 12 1 2 E 4 5 SUBG. NYMPHAEA N. TETRAGONA Unknown 1st day N. ODORATA 2-3 da) 1st day N. MEXICANA 2nd day N. TUBEROSA Unknown .M. P.M. B Day of A Taxon Flowering 6 7 8 9 10 11 12 1 2 3 4 5 SUBG. BRACHYCERAS N. AMPLA Unknown 1st day N. ELEGANS 2-3 da, N. GRACILIS Unknown C Day of P.M. A.M. Taxon Flowering 7? 8 9 10 11 12 1 2 3 4 5 6 7 SUBG. HYDROCALLIS 1st day N. CONARDII 2nd day 1st day N. GARDNERIANA 2nd day 1st day N. GLANDULIFERA 2nd day 1st day N. O A JAMESONIAN 2nd day N. NOVOGRANATENSIS 15t day 2nd day 1st day N. TENERINERVIA 2nd day 1st day N. LINGULATA 2nd day 1st day N. PROLIFERA 2nd day 1st day on e= N. AMAZONUM 2nd day FIGURE 1. Flower-opening times for selected species of Nymphaea. Data sources indicated in text. Dashed lines = partially open flowers, solid lines = fully open flowers.—A. North American species of subg. Nymphaea. — B. Neotropical species of subg. Brachyceras. — C. Subg. Hydrocallis. Volume 75, Number 3 1988 Wiersema 799 Nymphaea Reproductive Biology the night, and close in the early dawn. Their anthers do not dehisce until a few hours before dawn when cross-pollination? of first-day flowers is possible. Thus N. amazonum is effectively isolated repro- ductively from other species of subg. Hydrocallis even though it may make use of the same polli- nators (Cramer et al., 1975). Floral responses in the night-blooming subg. Lo- tos have received little or no attention. Flowers here reportedly open at dusk and remain open until ca. 11:00 A.M. the following morning (Conard, 1905; Hutchinson & Dalziel, 1966; Wiersema, 1982). First-day (pollen-receptive) flowers of /V. lotus 1 observed in cultivation did not open for this lengthy period, but were open for only a few hours around midnight. If these observations are consis- tent with natural populations, for which scrutinous data are lacking, then cross-pollination in some species of this subgenus could occur only at night, with the diurnal portion of flowering on later days, providing no opportunity for successful cross-pol- lination. No evidence suggests that seasonal separation vegetative growth has occurred and continue flow- ering throughout the growing season. In many pop- ulations of N. prolifera and N. lasiophylla C. Martius & Zucc., however, normal flowers are replaced by tuberiferous flowers (see page 801) during part or all of the growing season. This does not seem to be temporal separation of flowering but rather a shift from sexual to asexual repro- duction, the former not being successful in areas where this phenomenon has been observed. In many species the flowering cycle is extended, with anther dehiscence continuing into a third or fourth day. Species with extended flowering cycles increase the percentage of pollen-releasing (or functionally male) flowers relative to first-day (or functionally female) flowers. Thus the percentage of insect visits involving female flowers would be reduced, but larger pollen loads might be found on insects (Capperino & Schneider, 1985). A longer flowering cycle might also attract more pollinators to a population by displaying a greater number o open flowers. A number of ad- Other floral modifications. 3 As some species of Nymphaea, including N. ama- zonum, are clonal, cross-pollination as here used may involve either xenogamy or geitonogamy, in contrast to self-pollination involving autogamy. ditional morphological or behavioral differences in flowers have been noted among waterlilies. The functional significance of many of these has yet to be determined. Perhaps the foremost among them relates to the carpellary appendages, which are absent in subg. Anecphya, triangular to tapering in subgenera Nymphaea and Brachyceras, and most strongly developed in subgenera Lotos and Hydrocallis, the only night-blooming subgenera, where they are linear to highly clavate. They attain a length of nearly 3 cm in some flowers of N. oxypetala Planchon of subg. Hydrocallis. The structural differences in these appendages among species of Nymphaea suggest differences in func- tion. Hypotheses concerning the function of the carpellary appendages have suggested that they may serve as a source of food, a source of heat, or a source of volatile odor-producing compounds (Meeuse & Schneider, 1980; Prance, 1980; Schneider, 1982a, b). Similar structures in the related genus Victoria Lindley, which exhibits a night-blooming beetle-pollination syndrome, are known to have both thermogenic and nutritive functions (Knoch, 1899; Valla & Cirino, 1972; Prance & Arias, 1975). An additional function of undetermined importance for such appendages in Nymphaea may be to control access to the stig- matic disk. In first-day flowers the appendages are erect or spreading, allowing free access to the stig- matic region; however, in second-day and later flowers they are incurved over the stigma. In species such as N. lotus of subg. Lotos or N. amazonum of subg. Hydrocallis, they completely cover the stigma. In many species, especially diurnal species with less-developed appendages, the inner stamens form part of this barrier. It has been suggested that the carpellary ap- pendages may play a role in deepening or broad- ening the pool of stigmatic fluid and thus improving the ability of flowers to wash pollen from the bodies of insect visitors. However, N. gigantea Hook. of subg. Anecphya, which lacks carpellary append- ages, does not appear to exhibit any reduction in = size of the stigmatic pool (Schneider, 1982b). f the quantity of stigmatic fluid among alinak N. odorata, and N. mexicana by Cap- perino & Schneider (1985) contradict this hy- pothesis as well, as flowers of N. elegans, with the least- pios > appendages, produce the greatest amount o In the bene diurnally flowering subg. Brachyceras, the stamens bear prolonged connec- tive appendages in many species. These appear to form the inner surface of an oil-covered and very 800 Annals of the Missouri Botanical Garden slippery circular palisade surrounding the central pool of stigmatic fluid in first-day flowers. Insect visitors that enter this region inevitably fall into the stigmatic pool where pollen is washed from them. flowering subg. Anecphya by means of the large number of weakly supported, nodding stamens hav- e same result is achieved in the diurnally ing narrow filaments, which form an almost insur- mountable wall around the stigmatic region (Schmucker, 1932; Meeuse & Schneider, 1980) lowers of certain species of Nymphaea are characterized by abundant sclereids in some floral parts. In N. caerulea Savigny of subg. Brachy- ceras, elongate bipolar or acicular sclereids are found in the stamens (Chifflot, 1902). In several species of subg. Hydrocallis, such acicular scler- eids are abundant throughout the staminal tissue, and in a few species they are evident in the stig- matic ray tissue as well. In N. oxypetala of this subgenus, tiny spherical sclereids are produced in and apparently released from waqa: tissue ad- jacent to the anther sacs (Ch 02; Wier- sema, . Asterosclereids d dou bcscbere ds are also found in the floral tissues of certain species (Malaviya, 1962). All of the sclereids mentioned above are impregnated with calcium oxalate crys- tals. It has been suggested that the asterosclereids or trichosclereids may provide support (Conard, 1905; Malaviya, 1962) or internal water recovery (Schanderl, 1973). Neither of these explanations is sufficient to explain the spherical sclereids found in stamens of N. oxypetala or perhaps the great abundance of acicular sclereids in stamens in other members of subg. Hydrocallis. That these struc- tures may help deter beetle predation on floral parts has not been investigated. Differences in pollen morphology exist within Nymphaea. The pollen of night-blooming and pre- sumably beetle-pollinated subgenera is smoother than the highly puckered or ornamented pollen of the three diurnally flowering subgenera, which are apparently pollinated by bees and/or flies (Wier- sema, 1987). Waterlily seeds also vary. Those of temperate members of subg. Nymphaea are completely de- void of surface papillae, whereas those of tropical subgenera usually bear numerous surface papillae in various arrangements. A basic difference be- tween the environmental requirements of temper- ate vs. tropical seeds is that the former must be adapted to endure colder (and perhaps freezing) temperatures, whereas the latter must often with- stand periods of drought. Indeed, seeds of most temperate species exhibit little resistance to drought (Conard, 1905), and those of many tropical species do not tolerate freezing. It is hypothesized that the surface papillae may be important in conferring rought resistance to tropical seeds, perhaps by absorbing water. It is interesting to note that N. mexicana, a subtropical member of the mostly temperate subgenus Vymphaea, has papillose seeds. A number of waterlily species produce seeds without cross-pollination. In those thus far exam- ined, autogamy rather than agamospermy has been responsible (Wiersema, 1987). Two different meth- ods are employed in producing seeds autogamously. One involves homogamy, the early dehiscence of stamens on the first day of flowering when the stigma is receptive to pollen. | have observed this in N. jamesoniana and N. lingulata Wiersema of subg. Hydrocallis and N. ampla (Salisb.) DC. of subg. Brachyceras, and it has been reported for N. alba of subg. Nymphaea (Heslop-Harrison, 1955). The second method involves maintaining the receptivity of the stigma in second-day flowers, such that pollen being released at the normal time can effect self-pollination. This method is apparent in several taxa of subg. Hydrocallis, such as N. amazonum subsp. amazonum, N. conardii Wier- sema, and N. rudgeana. Autogamy has also been reported among other members of subg. Brachy- ceras, such as N. caerulea and N. stellata ‘Willd. (Conard, 1905). In the latter species and in /V. ampla (Prance & Anderson, 1976), reports have End pollen release in buds. the exception of the reports concerning Mou in the temperate day-flowering /V. alba, all other reports of autogamy involve tropical diur- nally and nocturnally flowering species. No records of autogamy are known for subgenera Lotos or Anecph ya. ASEXUAL REPRODUCTION A number of asexual modes of reproduction are employed by waterlily species. The most wide- spread is stolon formation, as in N. lotus of subg. Lotos; N. amazonum, N. gardneriana Planchon, N. lasiophylla, N. lingulata, and N. tenerinervia Caspary of subg. Hydrocallis; and N. mexicana of subg. Nymphaea. In N. mexicana, stolons are of greater diameter and develop an unusual ter- minal perennating structure at the end of the grow- ing season. This structure consists of a compact series of leaf buds from which hang several starch- laden roots reminiscent of a bunch of bananas, hence the common name “banana waterlily.” Several other members of subgenus Nymphaea, namely N. odorata, N. tuberosa, N. alba, and N. candida J. S. Presl, possess horizontal rhizomes in Volume 75, Number 3 1988 Wiersema 801 Nymphaea Reproductive Biology contrast to the erect rhizomes of all other species. Several shoots may eventually develop from a sin- gle horizontal rhizome. On the main rhizome of N. tuberosa small tuberous shoots develop which, due a constricted area at their base, are readily detachable and serve as propagules. wo other forms of asexual reproduction have been observed, one involving proliferation of floral tissues, the other proliferation of leaf tissues. In the former, all but the outermost appendages of flowers are aborted, and an enlarged tuber is formed centrally which gives rise to whorls of leaves and additional tuberiferous flowers. A few orders of branching may result, leading to the formation of a large number of small tubers. The tubers readily abscise, float briefly, and eventually become rooted and develop into mature plants. This process occurs regularly in the neotropical N. lasiophylla and N. prolifera of subg. Hydrocallis (Wiersema, 1987). Similar abortive flowers have been reported in other syncarpous Nymphaea (Bose, 1961; Mohan Ram & Nayyar, 1974; Majeed Kak, 1977; Mitra & Subramanyam, 1982) but only as an unusual oc- currence. This is a very effective method of re- production and dispersal, particularly in N. pro- lifera, which commonly inhabits lowland savannas subject to periodic flooding. Proliferation of leaf tissue occurs in N. micran- tha Guillemin & Perrottet of subg. Brachyceras. In this species new plants are formed on the upper surface of a leaf opposite the insertion of the petiole. These develop extensively only after the leaf is detached from the parent plant (Conard, 1905; Hutchinson & Dalziel, 1966). The effectiveness of this reproductive process is unknown. Nymphaea micrantha is found only in west tropical Africa. EVALUATION OF REPRODUCTIVE STRATEGIES Perhaps the best way of evaluating the overall reproductive strategies of waterlilies is to examine distributional ranges of various species as reflec- tions of their colonizing abilities. Reproductive strategy is one of several factors that contribute to distributional success. Table 1 provides a ranking of temperate, subtropical, and tropical species de- rived from their natural ranges. Such a measure cannot account for differences in population fre- quency within ranges, hidden instances of artificial dispersal, or problems of questionable taxonomy; however, it is still useful. Table 1 also provides information on the reproductive alternatives em- ployed by each species, insofar as information is available. All of these species are assumed to be capable of normal sexual outcrossing. Several ques- tionable species from the paleotropics have been exclude Two elements seem to be important in inter- preting the colonizing ability of a given species, its dispersal capability, and its ability to become es- tablished and persist in a new location. In cases involving overland dispersal, seeds are the probable dispersal units and waterbirds the likely agents. Although most aspects of seed release appear to be similar among the various species, the number of seeds per fruit and the size of the seeds are two variables that could affect dispersal capability. Nymphaea tetragona, which is dispersed solely by seeds, has comparatively large seeds with relatively few per fruit, two traits that could potentially retard dispersal. However, N. tetragona has the broadest distribution of any species. Even N. mexicana, having the largest seeds (almost twice as large) with the fewest per fruit of any Nymphaea species, has been successfully dispersed along the Gulf Coast of North America. The smallest and probably most numerous seeds are found in N. jamesoniana, the most widely distributed neotropical species. Thus there appears to be little relationship between these seed variables and dispersal capability, which ap- pears to be adequate for most species. If dispersal factors do not appear to limit overall distribution, then factors relating to establishment must be considered more important. Ecological fac- tors certainly affect germination, seedling devel- opment, and successful maturation of a colonizing species. If these initial barriers can be overcome, establishment and long-term survival in a new en- vironment become heavily dependent on repro- ductive ability. Several key observations can now be made with reference to Table 1. All ten of the most widely distributed tropical species utilize at least one re- productive alternative to total reliance on xenog- amy or geitonogamy. In seven of the ten this al- ternative is autogamy. Of the ten most narrowly distributed species, only one is known to be autog- amous. As already mentioned, reproductive strat- is one of several factors that affect overall species distributions. However, in terms of distri- butional success for tropical Nymphaea species, autogamy seems to be the most important repro- ductive alternative. e importance of autogamy to tropical water- lily species is to provide for seed production in % absence of potential pollinators and to enhan production when xenogamy or geitonogamy be- come inefficient. It also would eliminate the need for a second individual in colonizing situations; how- ever, this obstacle is easily overcome by clonal or 802 Annals of the Missouri Botanical Garden TaBLE l. Distributional success of Nymphaea species in relation to reproductive strategy. All species known or assumed to be capable of xenogamy. Data mostly from Conard (1905), Wiersema (1987), or personal observation. Reproductive alternatives? Species Subgenus Index' Autogamy Asexual TEMPERATE N. tetragona Nymphaea 310 N. candida Nymphaea 130 HR N. alba Nymphaea 110 A HR N. odorata Nymphaea 90 HR N. tuberosa Nymphaea 30 HR SUBTROPICAL N. elegans Brachyceras 40 N. mexicana Nymphaea 35 S TROPICAL N. lotus Lotos 140 S N. jamesoniana Hydrocallis 125 A N. ampla Brachyceras 115 A N. caerulea Brachycera 115 A N. amazonum Hydrocallis 100 A S N. rudgeana Hydrocallis 100 A N. conardii ydrocallis 90 A N. nouchalii otos 90 S? N. prolifera Hydrocallis 90 FLP N. stellata Brachyceras 85 A N. capensis Brachyceras 70 N. gardneriana Hydrocallis 70 S N. oxypetala Hydrocallis 70 N. glandulifera Hydrocallis 65 A N. gigantea Anecphya 60 N. tenerinervia Hydrocallis 40 S N. lingulata Hydrocallis 35 A S N. lasiophylla Hydrocallis 30 FLP, S N. micrantha Brachyceras 30 FOP N. gracilis Brachyceras 20 N. petersiana Brachyceras 20 N. potamophila Hydrocallis 20 N. sulphurea Brachyceras 20 N. novogranatensis Hydrocallis 15 : ele range + longitudinal range (in degrees). * HR = horizontal rhizomes; S = stolons; FLP = floral multiple-flowered individuals, as there is no evi- dence of self-incompatibility in Nymphaea. In the neotropics several night-blooming species of subg. Hydrocallis, such as N. prolifera, N. gardner- iana, N. lasiophylla, and N. tenerinervia, rarely produce seeds in natural populations, although they are capable of seed production if cross-pollinated (Wiersema, 1987). Reports on pollinators in this subgenus have implicated only scarab beetles of the genus Cyclocephala Latreille (Cramer et al., 1975; Prance & Anderson, 1976; Prance, 1980; Wiersema, 1987). Asexual reproduction has seem- proliferation; FOP — foliar proliferation. ingly allowed the four species mentioned to exist in areas where these pollinators are absent. Further overland expansion from such sites is probably more difficult in the absence of seed production. Nonautogamous species without asexual alterna- tives would be completely restricted by pollinator availability. The narrow distribution of most trop- ical species in this latter category suggests that pollinator distribution may be an important element in restricting their spread. The paleotropical and widespread N. lotus and N. nouchalii Burm. f. of the night-blooming subg. Volume 75, Number 3 1988 Wiersema 803 Nymphaea Reproductive Biology Lotos have not been observed to be autogamous, however. To account for the distributional success of these taxa either: a) autogamy has gone unde- tected, b) the pollinators utilized are more widely distributed or a range of pollinators is used, or c) stolons are more effectively employed in dispersal than in other waterlily species. A fourth possibility, that of human dispersal, may account for some of this distribution, as these species are cultivated, and the existence of N. lotus in the neotropics has been attributed to artificial introduction (Wierse- ma, 1982). However, floristic accounts have con- sidered them indigenous throughout most of their paleotropical ranges. n temperate species, all of subg. Nymphaea, autogamy is clearly not as important in conferring distributional success, as it mas been ore uus in one species. With th tubers produced by N. b wr E are the only effective dispersal agent produced in this subgenus. How then are the limits of pollinator availability overcome by wide-ranging temperate species? These diurnally flowering species either employ pollina- tors that are more widely distributed or utilize a range of pollinators. The latter explanation is prob- ably correct, in view of the variety of insect visitors reported for day-blooming species in temperate regions (Robertson, 1889; Conard, 1905; Meeuse & Schneider, 1980; Schneider & Chaney, 1981; Schneider, 1982a; Capperino & Schneider, 1985), but it needs to be confirmed with comprehensive studies of species—pollinator relationships. Pollinator relationships of most tropical day- blooming species in subgenera Anecphya and Bra- chyceras remain to be assessed; however, in subg. Brachyceras autogamy has become an important alternative to total reliance on outcrossing. LITERATURE CITED Bose, R. B. . A note on an abnormal Na Sas pubescens Willd. Bull. Bot. Surv. India 3: CAPPERINO, M. E. & E. L. SCHNEIDER. 1985. Floral EST of Nymphaea mexicana Zucc. (Nymphae ae). Aquatic Bot. 23: 83-93. 902. Contributions à l'étude de la Univ. Lyon, sér. 2, Costos, J. B. J. classe dies Nymphéinées. Ann. 10: 1-294. RN. M. E. 1980. Recent and Tertiary seeds of the Nymphaeaceae sensu lato with a revision of Bra senia ovula (Brong.) Reid and Chandler. Ann. Bot. 46: 603-632 ConarD, H. S. 1905. The waterlilies: a monograph of rs der Nymphaea. Publ. Carnegie Inst. Wash. -219. 1917. The white water lily of Clear Lake, Iowa. Proc. 2D Acad. Sci. 24: FA 454. z £ J, M., A. D. J. MEEUSE & P. . TEUNISSEN. 75. A d on the pollination of “adana w flow- ering species of Nymphaea. Acta Bot. Neerl. 24: 4 l ELKINS, dl 1970. phaea alba f. rosea. Amer. Hort. . 49: 45-46. ERIXON, C. 1980. T a orm av nord- náckros i sódra appland. Svensk. idskr. 74: 1-4. MEE V. 1981. Plant ine 2nd iit Colum- a Univ. Press, New York. Caña P. P. 1978. Cytogenetics of aquatic ornamen- tals. II. Cytology of nymphaeas. Cytologia 43: 477- The Swedish red waterlily — Nym- Ma 1980. EA doge of ii n ornamentals. VI. Evolutiona haea. he d 45: 307-314. HesLoP-HARRISON, Y. Nymphaea L. em. Sm. (nom. conserv.). J. Ecol. 43: 119-734. HUTCHINSON, J. & J. M. DaLzIEL. 1966. Flora of West KNocH, E. Untersuchungen übe - logie, Biologie und pe: der Blüte von Victoria regia. Biblioth. Bot. 9(47): 1-67. MAJEED Kak, A. 1977. Bocni floral morphology in Nymphaeaceae. Geobios E 220 Malaviya, M. 1962. study of sclereids in three species of Nymphaea. RAT Indian Acad. Sci. 56B: 232-231 MEEUSE, B. J. D. & E. L. SCHNEIDER. 1980. Nymphaea revisited: a preliminary communication. Israel J. Bot. : 65-79. MENDONGA, F. A. . Nymphaeaceae. /n A. W. Exell & H. Wild (editors), S Zambesiaca 1(1): 175- for Oversea Governments and (e o M 1982. Is yi rin s an apomict? Bull. Bot Surv. India 24: 83-86 Monan Ram, H. Y. € V. L. NAYYAR. of reversion of flower of Nym 1974. A case mphaea, the white waterlily, in the Itasca State Park r region. Proc. Minnesota Áca 9-6: 26- PRANCE, G. b A note on bi pollination of amazonum Mart. Pa Zucc. (Nymphae- a 32: 505-50 ` ANDEAN 197 6. Studies of the floral biology of neotropical Nymphaeaceae. 3. Acta Ama- zonica 6: 163- & ymphae aceae). feno . R: An 1975. A study of the floral biology of Victoria amazonica phe pel gd Era Acta Amazonica 5: 109-139. Pap TOR, M. & P. Yeo. 1972. The Pollination of Flow- s. Taplinger, New York. Cei a C. 1889. Flowers and insects. I. Bot. Gaz. (Crawfordsville) 14: 120-126. ScHANDERL, H Die physiologische Bedeutung sog. “‘Sternhaare” i Blattund Blattsteilge- d Nuphar. Z. Pflanzenphysiol. 70: 166-172. Miki T. Physiologische und — 8 orn an Blüten tropischer Nymphae n. Planta 16: 376-412. RM E.L. 197% ape tology of the Nym- phaeaceae. Proc. IVt . Sym PIT Maryland Agric. Exp. Sta. Special Misc. Publ. 419-429. Annals of the Missouri Botanical Garden 98 Notes on the floral biology of Nym- phaea elegans (Nymphaeaceae) in Texas. Aquatic Bot. 12: 197 . 1982b. Observation on the pollination biology of Nymphaea A a W. J. T x ymphae- aceae). W. Austral. abad. 15: 71-72. — — & T. e NEY. 1981. The floral biology of mphaea pou (Nymphaeaceae). Southw. Nat- uralist 26: 159- VALLA, J. J. & D. R. Canino, iru ru é, Victoria 1972. Biologia floral = D'Orb. Darwiniana 17: 4 498. VAN DER VELDE, G., T. C. M. Brock, M. HEINE & P. M. P. M. PEETERS. 1978. Flowers of Dutch Nym- py as a habitat for insects. Acta Bot. Neerl. 430. 27: WaRD, D. B. 1977. Keys to the flora of Florida. 4 Nymphaea (Nymphaeaceae). Phytologia 37: 443- WIERSEMA, J. H. 1982. Distributional records for Nym- phaea lotus (Nymphaeaceae) in the western hemi- sphere. Sida 9: 230-234. 1987. A monograph of Nymphaea subgenus Hydrocalis (Nymphaeaceae). Syst. Bot. Monogr. & R. R Haynes. 1983. Aquatic and marsh pes of Alabama. III. Magnoliidae. Castanea 48: 99- WiLiAMS, b R. 1970. Investigations in the white wa- Elie L adda of Michigan. Michigan Bot. 9: REPRODUCTIVE BIOLOGY OF Robert R. Haynes SELECTED AQUATIC PLANTS ABSTRACT Aquatic weeds are species that inhabit bodies of water in such quantities that they either interfere with man's usage or become a health hazard by serving as a breeding area for insects, especially mosquitoes. These plants are most often introduced species. They in increase in vds mostly by vegetative reproduction, including. stem t ap or germina ecies that eproduce sexually, such as Eichhornia azurea, are self-sterile, but others, such as Fi etinm crassipes or Ottelia Dakika, are self- sc eo many a in m almost always is self- ¿iban Before the flowers open, ma from the a ower eventually opens, but fertilization has already wind (P it takes plac the surface n by contact between stigma and a pu bal or by water-borne pollen (Elodea canadensis) ; or pollination may occur underwater (Najas minor Aquatic weeds are species that inhabit bodies of water in such quantities that they either interfere with man's usage or become a health hazard by serving as a breeding area for insects. Such species are often introduced. Cook (1987) discussed 12 species that he considered to be the most notorious aquatic invaders and pointed out that only one, Trapa natans, relies on sexual processes for its reproduction and spread. Of the remaining 11, he noted that only Myriophyllum spicatum, Najas minor, and Pistia stratiotes regularly develop seed in their native and adventive ranges. Salvinia mo- lesta is a sterile hybrid, and the rest have well- developed self-incompatibility mechanisms. An understanding of their reproductive biology, both sexually and vegetatively, is important in de- veloping methods of control for aggressive species. Aquatic flowering plants have pollinating systems ranging from those independent of the aquatic en- vironment, such as wind and animal, to varying degrees of adaptation to the aquatic environment, beginning with pollination at the surface of the water, to underwater pollination with the pollen adhering to the surface of air bubbles, to a totally aquatic system in which pollen sinks in the water. The following selected list of aquatic weeds begins with the least adapted to the aquatic environment and terminates with the most adapted. INSECT POLLINATED Eichhornia crassipes (C. Mart.) Solms-Laub. (Pontederiaceae). Water-hyacinth. Water-hyacinth is a free-floating annual or pe- rennial with rosettes, well-developed stolons, and swollen to bulbous petiole bases. It is native to tropical South America but has been introduced ornamentally into warmer parts of all continents for its showy blue flowers (Barrett, 1982; Cook, 1987). Eichhornia crassipes spreads vegetatively by daughter rosettes that form rapidly on brittle stolons and separate from floating mats quite easily (Barrett, 1979) Water-hyacinth is tristylous and self-compatible (Barrett, 1979). Barrett (1980a) studied fertility of nine clones of Eichhornia crassipes from dif- ferent regions of the world. Eight of the nine pop- ulations flowered regularly during the study period and the ninth flowered frequently following the study period. Artificial crosses, both selfed and out- crossed, were made with each clone that flowered during the study period, totaling 2,546 crosses. Of these, 94.7% produced capsules, with an average of 143.3 seeds per capsule. All populations exhib- ited a high degree of self-compatibility, although degree of seed production varied among clones. Percentage of capsule set was significantly higher ' Department of Biology, The University of Alabama, Tuscaloosa, Alabama 35487-1927, U.S. A. ANN. Missouni Bor. Ganp. 75: 805-810. 1988. 806 Annals of the Missouri Botanical Garden among all of the out-crossed than among all of the selfed plants. A similar amount of seed production has been found in natural populations (Barrett, 1980b). He studied open-pollinated and artificially. pollinated plants from 19 populations, with capsule produc- tion ranging from 72.3% to 100% per month for artificial pollinations, whereas it ranged from 8.1 to 68.775 for open-pollinated plants. Seeds per capsule were considerably fewer in open-pollinated ones, ranging from a mean of 3.1 to 40.8 per capsule, compared with artificially pollinated ones of 74.1 to 188.7 per capsule. According to Barrett (1980b), Kichhornia cras- sipes can begin flowering 10-15 weeks after ger- mination, soon for a perennial. One inflorescence with 20 flowers has the potential of producing 3,000 seeds, and up to four inflorescences can be produced by a single rosette during a 21-day pe- riod. Since flowering in subtropical to subtemperate regions may occur over five to nine months, the species can produce astronomical numbers of seeds. Regardless, most individuals in natural populations are probably produced vegetatively. eedlings were observed in only three of the 19 populations studied by Barrett (1980b). Appar- ently, seeds produced in dense floating mats of water-hyacinth either sink to the bottom or ac- cumulate in the mat. Shading from the mats or low light levels coupled with low temperatures in deep water prevent the seeds from germinating. Eichhornia azurea (Sw.) Kunth (Pontederiace- ae). Rooted water-hyacinth. Rooted water-hyacinth is an attached perennial with distichous, linear submersed leaves and elliptic to obovate floating leaves. The inflorescence is a compact spike of blue flowers, each having erose perianth lobes and a bilobed yellow spot in the center of the upper lobe. Vegetative reproduction is by chance fragmentation of the robustly branched stems (Barrett, 1978). The species is native to the Neotropics (Horn, 1987) and has been introduced to various regions as an ornamental (Barrett, 1978). Unlike Eichhornia crassipes, E. azurea is tri- stylous and partially self-sterile. Barrett (1978) artificially selfed long-styled forms with the low- level anthers. Only 12% fruit set occurred. In contrast, when mid-level anthers were utilized, 94% fruit set occurred. A semihomostylous race has been observed in Costa Rica. This one has one whorl of anthers about the same level as the stigma. High seed set occurs when pollen from this anther whorl is used for selfing. Ottelia alismoides (L.) Persoon (Hydrocharita- ceae). Duck-lettuce. Duck-lettuce is an annual or perennial from attached basal rosettes, with long-petiolate orbic- ular to ovate leaves. The flowers are borne sin- gularly in solitary spathes terminating long scapes. They are perfect or very rarely imperfect (Cook Urmi-Konig, 1984b), fragrant, and have stam- inodia that are important in attracting insect vis- itors. Cook (1982) indicated that, although no field observations have been published, the species is almost certainly insect pollinated. He also indicated that it is highly self-compatible and occasionally cleistogamous. The plants are weeds of rice fields in the southern United States (Dike, 1969), Italy, and Southeast Asia (Cook & Urmi-Konig, 1984b). Increase in population size is apparently by seed, since no specialized means of vegetative propagation occurs ) (Cook € Urmi-Kónig, 1984b Egeria densa Planchon (Hydrocharitaceae). Bra- zilean-elodea. Brazilean-elodea is a rooted perennial with cau- line leaves that are mostly in whorls of four. The flowers are imperfect and solitary on axillary pe- duncles that project the flowers to or above the water surface. Staminate and carpellate flowers contain glistening green functional nectaries. The flowers are frequently visited by small Diptera (Cook & Urmi-Kónig, 1984a), but there is no evidence yet as to whether these insects are important in pollen transfer, since seed-set is so rare in nature and in cultivation. Egeria densa is native to southeastern Brazil, Uruguay, Argentina, and possibly Paraguay. species has been introduced into North America, Europe, Asia, Africa, and Australia. Only in warm temperate and cool subtropical climatic regions has it developed into an aquatic weed. Outside its native range, only in Chile are carpellate flowers known ( Urmi-Konig, 1984a). No specialized over- wintering structures are produced. Stem fragments root readily and develop into new shoots, so rapidly that the species often quickly overgrows a lake. Spread of Egeria densa probably results from its popularity in aquaria. Hydrocharis morsus-ranae L. (Hydrocharita- ceae). European frogbit. European frogbit is a free-floating, stoloniferous perennial with basal, petiolate leaves. The flowers are borne in spathes and terminate pedicels that Volume 75, Number 3 1988 Haynes 807 Selected Aquatic Plants project the flower well above the spathe. Cook (1982) indicated that the carpellate flowers possess staminodia modified into nectaries that secrete a nectar attractive to insects. The staminate flowers appear quite similar to the carpellate flowers but lack these nectaries. They apparently offer no re- ward, but rely on mimicking the nectar-bearing carpellate flowers for insect attraction. Hydrocharis morsus-ranae first appeared in the Western Hemisphere in 1932 at the Ottawa Bo- tanic Garden, where it was cultivated (Dore, 1968). It was first noticed as an escape in 1939 in the Rideau Canal, from which it spread into the Ottawa and St. Lawrence rivers (Dore, 1968) and into the United States in 1974 (Roberts et al., 1981). This spread may have been by seed and by hibernacula. Cook & Lúónd (1982b) indicated that the species can grow from one hibernaculum to cover an area of one meter in diameter in one season. Pistia stratiotes L. (Araceae). Water-lettuce. Pistia is a monotypic genus of rosulate-leaved, free-floating, stoloniferous plants occurring in sub- tropical and tropical Africa, Asia, and America. The leaves are densely short-pubescent and sur- round a single terminal spathe that has the spadix adnate to its median line. The plants are monoe- ceious, with staminate and carpellate flowers on one spadix. Cook (1987) indicated that the species is prob- ably mostly self-pollinated, possibly by insects. Wil- son (1960) stated that in Florida the ovary of water- lettuce tends to enlarge and become inflated, but no seeds are produced. He suggested that this lack of seed-set is probably due to absence of pollinators. His suggestion would tend to support Cook's view of the species being insect pollinated. Pistia is important because of its vegetative reproductive capabilities. New plants are produced at the ends of the stolons and are separated from the parent plant by fragmentation. The species can reproduce rapidly enough to clog waterways. WIND POLLINATED Potamogeton nodosus Poiret (Potamogetona- ceae). Floating pondweed. Potamogeton nodosus is a perennial from elon- gate rhizomes with long-petiolate, lanceolate sub- mersed and floating leaves with cuneate bases. The inflorescence is a compact spike held above the surface of the water (Haynes, 1978). The species can cover huge areas of lakes in the southern United States. There are probably only one or a few clones, as Potamogeton nodosus increases in number mostly by rhizome growth. Pollination is predominantly anemophilous (Phil- brick & Anderson, 1987). As such, the species is adapted for outcrossing. Myriophyllum spicatum L. (Haloragaceae). Eurasian milfoil. Eurasian milfoil is a submersed, rooted, stolon- iferous perennial with whorled pinnately compound leaves and emergent imperfect flowers. The plants are monoecious and wind pollinated. Myriophyllum spicatum is native to northern Eurasia (Cook, 1987) and has spread into the Western Hemisphere from Ontario and Quebec south to Florida and west to Wisconsin, Oklahoma, Texas, and Mexico, as well as to British Columbia, Washington, and California (Aiken, 1981). The species has often choked waterways (Coffey & McNabb, 1974), but in some places dramatic de- cline in the number of plants has occurred (Bayley et al., 1968; Elser, 1969) to the extent that the species no longer poses an environmental problem in these places. Spread of Eurasian milfoil is mostly by vege- tative fragment and seed (Cook, 1987). POLLINATION AT SURFACE OF WATER Hydrilla verticillata (L. f.) Royle (Hydrochar- itaceae). Hydrilla. Hydrilla is a perennial, rooted plant with the lower nodes having opposite leaves and the upper nodes having whorls of three to eight leaves. The flowers are imperfect and are produced singly in the leaf axils. Carpellate flowers are sessile with an elongating hypanthium that projects the perianth and stigma to the water's surface. As the hypan- thium elongates, the perianth is forced open by a gas bubble (Cook & Lüónd, 1982a). Upon reaching the water surface, the perianth lobes open further, forming a funnel that is underwater below and open to the air above. The stigmas are at the bottom of this funnel. The subsessile staminate flower buds are released from the spathe by pedicel abscission and float on the surface at about a 45? angle. An hour or more after the bud is released, the perianth segments retract slightly, with each anther adher- ing to the convex part of the perianth. Soon the perianth spreads horizontally on the water surface; shortly after this the stamens suddenly spring from a horizontal to a vertical position; the anthers burst and scatter pollen in the air. As the pollen falls, most grains land on the water and are lost for 808 Annals of the Missouri Botanical Garden reproductive purposes, but some fall into the fun- nel-shaped perianth of a carpellate flower, con- tacting a stigma. This is apparently a risky method of pollination, but, according to Cook & Lüond (1982a), seed production is adequate to maintain hydrilla populations in areas that dry out and where hibernacula are not produced, or at least have not been observed. Most increase in plant density is by hibernacula, which occur in two forms. The first form is from erect stems and is olive green and ovoid-conical with spreading apices that give the structure the appearance of a sandbur. Hibernacula of the sec- ond type are brown, subterranean structures that appear as minute potatoes at the tips of long, stringy, white, leafless rhizomes. Cook € Lüónd (19822) indicated that the first type has food reserves in leaves, whereas the second has food reserves in swollen stem tissue. Lagarosiphon major (Ridley) Moss (Hydro- charitaceae). African-elodea. African-elodea is a rooted submerged perennial with alternate leaves. The flowers are from axillary, solitary spathes. The staminate spathes enclose many pedicellate flower buds, whereas the carpel- late spathes contain only one sessile flower. The staminate flower buds are released and rise to the water surface, remaining closed for a short while before eventually opening by the reflexing of perianth parts. The open flower, with three fertile horizontal and three sterile erect stamens, floats on the reflexed perianth. The staminodia function as a sail. Carpellate flowers are projected to the water surface by an elongate hypanthium (Healy & Ed- gar, 1980). A meniscus is formed on the water surface by the carpellate flower, which then is slightly below the surface but with the styles pro- truding above the surface film. As staminate flowers move along the water, propelled by staminodia sails, one flower eventually tips into the meniscus, thereby causing a aiu fertile stamen to con- tact an erect stigm Lagarosiphon "m become an important weed in New Zealand, where it is replacing Elodea can- adensis that has attained its maximum density and is declining (Healy & Edgar, 1980). Only carpel- late plants are known in New Zealand, however, where it spreads by vegetative fragments. Elodea canadensis Michaux (Hydrocharita- ceae). Canadian-pondweed. Canadian-pondweed is an attached species with cauline leaves in whorls of three. The flowers are axillary, imperfect, and elevated to the water sur- face by an elongate pedicel on the staminate flower and by an elongate hypanthium on the carpellate flower. The carpellate flower opens by recurving of the sepals and petals, which float on the water surface. The styles spread between the sepals and then recurve, with the tips usually becoming sub- mersed. Upon reaching the surface, the staminate flower opens by the sepals and petals spreading there. The anthers dehisce in an upright position, scattering the pollen onto the surface of the water Pollen floats on the water surface until it contacts a stigma, initiating germination. "lodea canadensis is widespread in North America, where it is not known to become weedy. It was introduced into Europe in the early nine- teenth century (Coo Urmi-Kónig, 1985), whereas the first record in Australia is 1931 (Aston, 3). The European material has almost exclu- sively been carpellate plants. Vegetative reproduc- tion is mostly from stem fragmentation. Not all nodes, however, are capable of rooting and devel- oping a new plant. SUBMERSED POLLINATION WITH POLLEN ADHERING TO AIR BUBBLES Ruppia maritima L. (Potamogetonaceae). Ditch- grass. Ditch-grass is an annual or perennial of brackish or saline waters with alternate leaves, these having the blades adnate to the stipules for the entire length of the stipules. The flowers are perfect and are produced in a capitate spike that is first en- closed by the sheathing leaf bases. The pollen is four times as long as broad, arcuate, swollen at the ends and at the center on the convex side, and three celled. Its exine is reticulate (Haynes, 1978). The gynoecium is of four or five distinct, stipitate carpels that have the gynophore elongating after anthesis. Pollination is mostly underwater. Follow- ing maximum peduncle elongation, the anthers de- hisce underwater, releasing pollen that is trapped in air bubbles (Verhoeven, 1979). As the bubbles remain with an inflorescence for several hours, the pollen grains can only contact a stigma of that flower (Verhoeven, 1979), making d almost certain. Such a system insures ample se set, which is important for an often annual species. Out-crossing does occur occasionally in Ruppia maritima. Air bubbles, with their trapped pollen grains, break free from the inflorescence occasion- ally and rise to the water surface. Once on the surface, the bubble breaks, liberating the pollen grains, which float on account of trapped air in the reticulations of the exine. Should they contact a Volume 75, Number 3 1988 Haynes 809 Selected Aquatic Plants stigma that is at the water surface, cross-pollination occurs. More likely, however, these grains are blown from the Ruppia zone. UNDERWATER POLLINATION Zannichellia palustris L. (Zannichelliaceae). Horned-pondweed. Horned-pondweed is an annual rooted plant with alternate, opposite, and occasionally whorled, lin- ear leaves on the same plant. The flowers are imperfect, both staminate and carpellate in the same leaf axil. The carpellate flower is surrounded by a spathelike envelope and consists of four or five separate carpels, each with a funnel-shaped stigma. The staminate flower is outside the enve- lope, with a filament that projects the anther over the stigmas. Pollen is released in a gelatinous mass (Haynes & Holm-Nielsen, 1987) and falls directly into the funnel-shaped stigma, thus insuring self pollination. annichellia is an annual without any vegeta- tive perennating structure. It therefore depends entirely on seed production for surviving the un- favorable season. Najas minor All. (Najadaceae). Water nymph. Najas minor is a rooted, submersed annual with subopposite, serrulate leaves. The plants are mon- oecious, with flowers solitary in the leaf axils, the staminate flowers mostly above the carpellate ones. Pollination is entirely underwater. The pollen is heavier than water, and, after being released, it slowly descends through the water column, possibly contacting a stigma. Although this is not a system that insures pollination, seed set is very good The species occurs in northern Africa, Europe, Asia, and North America (Triest, 1987). In the past 50 years, it has become widespread in eastern North America (Haynes, 1977, 1979; Merilàinen, 1968), where it has become a troublesome weed in some areas. Increase in numbers is mostly from seed. Meriläinen (1968) suggested that the species has been spread by migrating waterfowl. DISCUSSION Some species of aquatic vascular plants have become weedy, especially in areas outside their natural ranges. These weedy species have posed environmental problems, either by interfering with recreation, such as by clogging waterways or by lowering the quality of fishing, or by forming breed- ing areas of noxious insects. Reproduction sufficient to overtake a body of water most often is vege- tative, either by stem fragmentation or by pro- duction of hibernacula. Almost all species also, at least occasionally, undergo sexual reproduction. Most have very little adaptation for adequate trans- fer of pollen. Instead, they project the inflorescence above the water for either animal or wind polli- nation. À few, however, have developed methods of pollen transfer either at the water surface or underwater. Among these few, some, such as Najas minor, are annuals and depend entirely on seed production for maintaining the population. LITERATURE CITED AIKEN, S. G. 19 A conspectus of Myriophyllum (Haloragaceae) in n North America. Brittonia 33: 57- ASTON, H. I. 1973. Aquatic Plants of Australia. Mel- ourne Univ. Press, Melbourne. BARRETT, S. C. H. 1978. Floral biology of Eichhornia azurea Cu Kunth (Pontederiaceae). Aquatic Bot. 5: 217 Er "The evolutionary breakdown of tristyly in Eichhornia crassipes (Mart.) Solms (water hy- iia Evolution 33: 499-510. 98 Sexual reproduction in Eichhornia ue (water hyacinth). I. EN z Feri from diverse regions. J. Appl. Ecol. 17: 19 Sexual edil in > BichNprhia crassipes (water hyacinth). II. Send production in natural populations. J. Appl. Ecol. 17: 113-124 Style morph distribution in New World populations of Eichhornia crassipes (Mart.) Solms- Laubach (water hyacinth). Aquatic Bot. 13: 299- 306 BAYLEY, S., H. RABIN & C. H. SourHwick. 1968. Re- cent decline in Us distribution and abundance of Eurasian milfoil in Chesapeake Bay. Chesapeake Sci. -181. Correy, B. T. & C. D. McNaBB. 1974. Eurasian water- milfoil in bi Michigan Bot. 13: 159-16 Cook, C. D. K. 1982. Pollination ex in the Hydrocharitaceae. Pp. 1 -15 in J. J. Symoens, S. S. oope ompére (editors) Stúdisė on yy i Vascular Plants. Royal Botanical Society of Belgium, Brussels. 1987. Vegetative growth and genetic mobility in some aquati weeds. Pp. 217-225 in Differentia- c tion Patterns in Higher Plants. Academic Press, Lon- —— . LÜÓND. 1982a. A revision of the genus oo (Hydrocharitaceae). Aquatic Bot. 13: 485- 504 & —————. 1982b. A revision of the genus ud (Hydrocharitaceae). Aquatic Bot. 14: A revision of the c Bot. 19: . Unmi-KÓNIG. I genus Egeria (Hydrocharitaceae). Aquatic 73-96. € ——, 4b. A revision of the genus Ottelia (Hydrocharitaceae). 2. The species of Eur- asia, Australia and America. Aquatic Bot. 20: 131- ns A revision of the n = 1985. Fades (Hydrocharitaceae). Aquatic Bot. 21: 111- DIKE, D. H. 1969. Contributions to the Biology of 810 Annals of the Missouri Botanical Garden Ottelia alismoides (Hydrocharitaceae). M.S. Thesis. pu d of Southwestern Louisiana, Lafayette, i W. Yo 1968. Progress of the PHEA F ped n Canada. Canad. Field-Naturalist 82: , ErsER, H. J. 1969. Observations on the decline » the water milfoil and other aquatic ae enn 1962-1967. Hyacinth Control J. 8 Haynes, R. R. ; . in ds South- eastern United States. J. Arnold Arbor. 58: 161- 170 1978. The Potamogetonaceae in the South- mes United States. J. Arnold Arbor. 59: 170- 979. Revision of North pr Central Amer- ican | Najas risa Sida 8: 34-56 —— & L M-NIELSEN. The Zanni- chelliaceae in ihe S E United States. J. Ar- nold Arbor. 6 -268. HEALY, .& Flora of m Zealand, Volume III. P. D. Hasselberg, Wellingto Horn, C. N. 1987. 205. Pontederiaceae. In: G. Harling & L. Andersson (editors), Flora of Ecuador, Number MERILRINEN, J. 1968. d minor All. in North Amer- hodora S 161- R C. T. & G.J. pe m 1987. ei erie of pollen/ a ratios and pollen size for the - ductive biolog otamogeton n autogamy in aquatic Mi: Ri Syst. Bot. 1 RoBERTS, M. L., R. L. idees & R. S. MCN. 1981. Hydrocharis morsus-ra ia Lope iid new to the United Sta Rhodora 7-148. TRIEST, L. nes revision of e genus /Vajas L (Najadaceae) ir in Africa and surrounding islands. Mém cad. Roy. Sci. Belgique, Cl. Sci. (8°) 21(4): 1-88 + pl. VERHOEVEN, J. T. A. 1979. The ecology id Nos dominated communities in Western Europe. I. Dis- tribution of Ruppia representatives in Eidos to their p Aquatic Bot. 6: 197-268. WirsoN, K. A. 1960. The genera of the Arales in js jr ms United States. J. Arnold Arbor. POLLINATION POSTULATES AND TWO-DIMENSIONAL POLLINATION IN HYDROPHILOUS MONOCOTYLEDONS! Paul Alan Cox? and R. Bruce Knox? ABSTRACT large search vehicles w found to be more likely to hit sus than small o Although the determination of the vectors re- sponsible for transporting pollen from flower to flower is a central task of pollination biology, no standard criteria for vector determination exist. Historically, methodologies for vector determina- tion range from careful ecological studies to con- jecture based upon superficial examination of dried herbarium specimens or worse. As a result, the literature of pollination ecology is uneven in rigor. Perhaps since anthecology originated as a spe- cialization of plant taxonomy, textbooks and re- views tend to favor claims with historical priority. Thus, even incorrect ideas are notoriously difficult to remove from the literature. An example may be found in the literature con- cerning the pollination ecology of Freycinetia ar- borea Gaud. (Pandanaceae) in Hawaii. Pollination by rats was suggested by Degener (1930), who made no experiments or observations of pollinator visitation. No consideration was given to how F. arborea was pollinated before the introduction of rats to Hawaii by Polynesian colonists in 400 A.D (Jennings, 1979). However, despite several pre- vious detailed accounts of Freycinetia pollination in the Dutch, German, and English literature (see reviews in Cox, 1981, 1984), Degener's rather anecdotal sketch emerged as authoritative. Fre- quently cited in textbooks, but sometimes attributed to a secondary source (Proctor & Yeo, 1973), the story of rat pollination of Freycinetia thus began a life of its own. Each recounting of the story added to both its detail and apparent authenticity. Al- though the original report did not indicate the time of day that pollination supposedly occurs, this im- portant detail later appeared in a major compen- dium (Faegri & van der Pijl, 1979). Degener (pers. comm.), however, never witnessed rat visitation but only inferred it from scratches on the floral bracts. Subsequent studies demonstrated such scratches to be caused by the introduced white- eyes that now pollinate F. arborea. Electron mi- croscope analysis of bird specimens collected in the nineteenth century indicates that formerly polli- nation was mediated by now extinct honeycreepers Cox, 1983a Such confusion in vector identification is remi- — niscent of disputes concerning pathogen determi- nation in medicine in the 1850s. The prevailing theory was that diseases like tuberculosis are caused by a variety of climatic, environmental, and organic factors. However, Robert Koch (1880) proved that tuberculosis is caused by a single bacterium. His ! This study was supported by a National Science Foundation Presidential Young Investigator Award ( BSR 8452090) , a University of Melbourne Research Fellowship, and a grant from the Data General Cor pordi, . K. k, S. C. Ducker We thank S. Banack, D. Banack, ook, , T. Hough, C. McConchie, L. O'Rourke Tomlinson, M. Tuohy, J. West, and an anonymous reviewer for assistance, and M. Rothman for his illustration of Amphibolis. 2 pr of Botany and Range Science, Brigham ung University, Provo, Utah 84602, U.S.A. You 3 Plant Cell Biology Research Centre, School of Botany, University of Melbourne, Parkville, Australia 3052. ANN. Missouni Bor. GARD. 75: 811-818. 1988. 812 Annals of the Missouri Botanical Garden TABLE l. Postulates. grain on a random trajectory will not hit any given Koch's Postulates l. e must be demonstrated in all cases of the 2. Pathogen must be cultivated in pure cultur 3. Cultured pathogen must cause disease in bas in- dividua 4. Pathogen must be re-isolated from these infected in- dividu Pollinaton Postulates In the field: 1. Pollen transfer from anther or pollen presentation ap- ratus to vector must be observed. 2. Pollen transport by vector must be observed. 3. Pollen transfer from vector to stigma must be ob- served. 4. Pollen deposited by vector on stigma must be dem- onstrated to effect fertilization. rigorous methodology has since been encapsulated in a series of postulates named after him (Table 1). Brock (1966) and Harper (1978) have sug- gested that ecology could benefit from similar rigor. Using Koch’s postulates as a model, we propose a standard set of postulates for vector determi- nation in pollination biology (Table 1). Our pos- tulates require confirmation of each stage of the interaction between anther, vector, and stigma. We submit that unless these postulates have been met, the efficacy of a presumed vector in pollination has yet to be proven. We additionally suggest that all such determinations be made in the natural pop- ulations rather than in the laboratory, where un- natural conditions can deceive even the most care- ful investigators. We used these pollination postulates to guide our studies of pollination in hydrophilous mono- cotyledons and to test previous theoretical predic- tions concerning two-dimensional pollination sys- tems (Cox, 3b) THEORETICAL PREDICTIONS OF Two-DIMENSIONAL POLLINATION SYSTEMS The probability of a pollen grain reaching the stigma can be modeled for aquatic regimes where wind, currents, and orbital wave motion drive it along an essentially random path. In the simple case of a pollen grain tracing a Brownian path, it can be shown that the pollen grain will eventually hit any coplanar stigma, given enough time (Cox, 1983b; Hersch & Griego, 1969). However, in three dimensions there is a probability that a pollen stigma even given an infinite amount of time. Since such fractal motion is recurrent in two dimensions but not three, it was predicted that selection will favor the evolution of two-dimensional pollination systems where such systems are possible, i.e., in aquatic regimes such as ponds, and intertidal re- gions (Cox, 1983b) The theoretical effectiveness of two-dimensional pollination systems can be examined through the use of search theory (Cox, 1983b; Koopman, 1956). Of particular interest are the effects of pollen mor- phology on stigma encounter rates. The probability P of a pollen grain encountering a fixed stigma, if the pollen grain traces a random path in the same plane, can be shown to be P= ] — va, (1) where w is the width of the path swept by the pollen grain, L is the length of the search path, and 4 is the search area (Cox, 1983b). Hence at low encounter probabilities, even a small increase in the pollen grain diameter will result in a very large increase in probability of encountering a stig- ma. Of particular interest is the exponential nature of this function: at low encounter probabilities a doubling of pollen grain diameter will result in a more than two-fold increase in the probability of hitting a stigma. From these theoretical considerations two gen- made concerning hydroph- ily (water-pollination): 1) a 2-D pollination system will be more efficient than a 3-D system, and thus favored by natural selection if pollen and stigmas can be dispersed in the same plane; and 2) in a 2- D pollination system, pollen grains or aggregations of large dimension will be much more likely to hit stigmas than small ones. Following the nomencla- ture of search theory (Koopman, 1956; Cox, 1983b), such aggregations will hereafter be termed "search vehicles." eral predictions can be EMPIRICAL PROPERTIES O Two-DIMENSIONAL POLLINATION SYSTEMS These theoretical predictions concerning two- dimensional pollination were tested in five different genera of marine and freshwater aquatic plants. In accordance with our postulates, we observed in each species the release of the pollen from the anther to the surface, transport of the pollen by the water surface, and deposition of the pollen from the water surface onto the stigma. We are cur- rently studying the efficacy of fertilization under field conditions. Such confirmation is important Volume 75, Number 3 Cox & Knox 813 1988 Hydrophilous Monocotyledon Pollination O <+ o Qo o o = E dg Kk à p = JK SA. 2774 777 £ d iz Li ULL, Amphibolis 2 Halophila Af 1^ 9,7777. Mus ird ZZZ Ruppia Lepilaena 5 10 15 20 25 Diameter (mm) Size distribution of floating search vehicles as aggregationt from five genera of hydrophilous 14), Ha monocotyledons: Halodule pinifolia (median = am is um lophila ovalis (median — 8,000 um, N = 17), Ruppia spiralis, Altona, Victoria, popu n (median = 1,361 um, N = 300), yar cylindrocarpa (median = 338 um, N = 172), and Amphibolis pet pes — 8,500 um, N — 40). because of the possibility of apomixis in some dioe- rine and essentially occurs along the plane of the cious seagrasses, such as Halophila stipulacea substratum (Forsk.) Aschers. (Hydrocharitaceae), where seed In the species we studied, e.g., Halodule pi- has been found to develop in pistillate cultures — nifolia (Miki) den Hartog (Cymodoceaceae), Hal- (McMillan, 1980). ophila ovalis (R. Br.) Hook. f. (Hydrocharitaceae), Our study focused on genera in which pollen is Ruppia maritima L. ex Dumort (Ruppiaceae), Le- transported in direct contact with the water. How- pilaena cylindrocarpa (Koernicke ex alp. ever, plants in which pollination occurs by the Benth. (Zannichelliaceae), and Amphibolis ant- collision of floating staminate flowers with buoyant arctica (Labill.) Sonder € Aschers. ex Aschers. pistillate flowers, such as some genera of the Hy- (Cymodoceaceae), we found pollination to occur on drocharitaceae, e.g., Vallisneria, Nechamandra, the water surface. In these species the pollen is Enhalus, Lagarosiphon, or Appertiella, can also hydrophobic and forms floating search vehicles that be said to have in effect two-dimensional pollination collide with the buoyant stigmas. However, the systems (Cook, 1982; Ernst-Schwarzenbach, 1945; search vehicles are formed differently and are of Troll, 1931; Wylie, 1917; Cox, 1988a). Taxa also different size in each species (Fig. 1). The pollen exist that have submarine two-dimensional polli- grains of Halodule pinifolia, for example, are fi- nation systems, such as Thalassia testudinum liform and link together in search vehicles that Banks ex Konig (Hydrocharitaceae), where pollen resemble snowflakes, while the pollen grains of is released at the substrate surface and dispersed — Lepilaena cylindrocarpa are small, spherical, and in negatively buoyant search vehicles (Cox € Tom- dispersed in a floating mat of mucilaginous slime. linson, 1988; Cox, in press). Pollination is subma- The pollen grains of Halophila ovalis are oval and 814 Annals of the Missouri Botanical Garden of medium size, but are dispersed in long floating mucilaginous tubes of thecal origin (Cox & Knox, 1986; Pettitt, 1980, 1981). These tubes link to- gether to form featherlike search vehicles. Ruppia spiralis pollen grains have a boomerang shape, forming search vehicles by lining up side-to-side (Cox & Knox, 1986; Gamerro, 1968; Verhoeven, 1979; Van Vierssen et al., 1982). Similar evolu- tionary convergences can also be found in stigma morphologies: the marine genera have filamentous, smooth stigmas while the freshwater genera have small indusiate stigmas that create small depres- sions on the water surface. It therefore appears that there has been convergent evolution towards a surface pollination syndrome. 'Two-DIMENSIONAL POLLINATION IN AMPHIBOLIS ANTARCTICA The pollination ecologies of Halodule pinifolia, Halophila ovalis, Ruppia spiralis, and Lepilaena cylindrocarpa will be reported in detail elsewhere Cox & Knox, in press). However, many of the essential features of surface-pollinated taxa are il- lustrated by the pollination ecology of Amphibolis antarctica, a dioecious seagrass found in the waters of Western Australia, Southern Australia, Victoria, and Tasmania (Ducker et al., 1977; Aston, 1973). Amphibolis antarctica plants produce solitary floral units (for discussions of floral terminology see McConchie et al., 1982; Tomlinson, 1982) at the ends of short lateral leafy branches with distichous phyllotaxis. The staminate floral units are bipartite, with two fused stamens borne on a short pedicel. The pistillate floral units consist of two free carpels, each with a sessile ovary bearing three slender styles (McConchie et al., 1982; Tomlinson, 1982). A few days prior to dehiscence, the pedicel bearing the staminate floral units elongates, pushing the fused anthers up from between the bracts. How- ever, the floral units (Fig. 2A) are still hidden by the leafy shoots until they abscise and float to the surface. Within the anthers, the mature tricellular pollen grains are filamentous and 3,000-5,000 um long, with forked tails (Ducker et al., 1978). The mature pollen “noodles” lack a developed exine (Pettitt et al., 1983). Details of pollen—stigma in- teractions and pollen tube growth have been in- vestigated by Pettitt et al. (1980, 1983). In pis- tillate plants the mature styles protrude from the shoots in a plane orthogonal to the plane of phyl- lotaxis. Pollination in Amphibolis antarctica has been described as submarine, i.e., occurring beneath the surface of the water (Ducker et al., 1978; Pettitt et al., 1980, 1981, 1983). However, the possibility of variation is indicated by previous observations that the male flowers may be shed, releasing their pollen on the surface of the sea, with the floating pollen forming large aggregates (Ducker et al., 978). bas We studied the pollination ecology of Amphi- bolis antarctica in an intertidal population at Point Lonsdale, Victoria, Australia, attempting to verify each of the previously discussed pollination pos- tulates. At 3:04 P.M. on December 13, 1986, dur- ing a low (0.3 m) tide, numerous staminate and pistillate plants in the population were examined and found to have mature flowers. At low tide, the tops of the plants were observed floating on the surface, with the plane of the distichous phyllotaxis coplanar with water surface. As a result, one set of stigmas from each pistillate floral unit penetrated the water surface. A few abscised staminate flowers that had already released their pollen were found along the beach, but no pollination events were observed. Between 4:40 P.M. and 5:20 P.M. on December 15, 1986, during an even lower tide (0.2 m), several thousand staminate floral units were ob- served to abscise (Fig. 3A) and float to the surface Fig. 3B). Once on the surface, the anthers dehisced longitudinally and extruded flocculent masses of floating pollen (Figs. 2A, 3C). The pollen mass from each male floral unit rapidly expanded on the water surface into large floating search vehicles (median diameter — 8,500 um, N — 40) of fractal geometry (Fig. 3D). The tops of nearly all of the mature pistillate plants in the population were exposed during this extremely low tide so that one set of stigmas in each floral unit penetrated the water surface (Fig. 3E). Pollination was observed to occur on the water surface through the collision of the floating search vehicles with the stigmas (Figs. 2B, 3E). As the tide came back in, thousands of empty staminate floral units were washed up along the beach. Field experiments with mature and slightly immature pollen showed mature pollen to float, — while slightly immature pollen proved to be neu- trally buoyant. By observing actual pollination events at the water surface, we can with confidence ascribe a two-dimensional surface pollination syndrome to Amphibolis antarctica. 'Two-dimensional surface pollination is probably characteristic of A. amphi- bolis in the intertidal zones where the stigmas have the potential to be exposed during low tides. The species can grow and flower at greater depths, however, where pollination, if it occurs at all, must of necessity be submarine. Although we believe Amphibolis antarctica to be primarily surface- pollinated, submarine pollination cannot be ex- Volume 75, Number 3 1988 Cox & Knox 815 Hydrophilous Monocotyledon Pollination FIGURE 2. with the filiform pollen e. (scale = 1 mm).— exposed natural pollination event in the Point Lonsdale, Victoria, a (scale = 100 um cluded on the basis of our current observations and may also occur in the species. The floral morphology of A. antarctica permits only one of the two sets of stigmas in each pistillate floral unit to encounter the water surface and hence be pollinated. This may relate to the fact that only one of the two ovaries develops, while the other one aborts (Tomlinson, 1982). This condition is analogous to the situation in Ruppia, where the inflorescence frequently floats on the surface in such a way that only one of the sets of stigmas is exposed to floating pollen while the other is sub- merged. Germina tion in A. antarctica is viviparous (Fig. 3F) (Black, 1913), with the mature seedling being released together with an unvascularized “grappling apparatus" (Fig. 3G) ‘tal develops as an outgrowth of the female floral unit (McConchie et al., 1982). This grappling apparatus presumably assists in attaching the seedling to a substrate. SEARCH VEHICLE SIZE AND SUCCESS in RUPPIA SPIRALIS Although two-dimensional search theory pre- dicts that large search vehicles are much more likely to encounter stigmas than small search ve- hicles (Cox, 1983b), there has yet to be an em- pirical test of this prediction. We therefore studied pollen/stigma encounters in a Ruppia maritima Reproductive structures of Amphibolis antarctica.— A. Staminate floral unit prior to dehiscence, Fili rm pollen grain on an Amphibolis stigma from a ). population growing in a brackish pond near Queenscliff, Victoria. Using an immersible camera stand, we filmed natural pollination events with a video camera equipped with a macro lens. The video tape was then analyzed frame by frame with a high-resolution freeze-frame video deck attached to a Zeiss Videoplan digital imaging system. Di- mensions of all search vehicles were calculated by digitizing their images on the video screen. The size distributions (largest diameters) of 1,000 such search vehicles chosen at random from the video ootage were calculated and recorded. Then the video tape was analyzed again to determine the sizes of search vehicles that hts collided with stigmas. This was accomplished by finding a frame in which a search vehicle could be seen to hit the stigma, and then reversing the tape several frames so that the dimensions of the search vehicle prior to collision could be determined. This procedure was repeated for 100 such search vehicles. The respective size distributions of successful search vehicles together with the size distributions of all search vehicles are shown in Figure 4. The median size of all search vehicles was found to be 812 um (N = 1,000), while the median size of successful search vehicles was much larger (1,388 um; N = 100). A U-test for the difference between the medians was significant at the 0.05 level. This ‘duypaas pasiosqn fo ]psiədsiq e) — quad jousajpu uo Fuimos? Sun]pəəs snoanday4 -y—‘spw3ys ynn 2]onjoa yosvas fo 1201u02 samaisngpn 142su1 Ya] addy ‘asvfins 12910m voy Suipnaj04d . s2]K1$ 270N ‘aovfins 1210 ynm 10uv]doo AxnjoyAyd snoyousip fo aunjd yna əəofans 12100 uo Sunvoy 1uvjd amjnisid fo doy “g—-vas fo aavfins uo Jumo uajjod eonoseyue “y fo sajoryaa Y91DIS 9911019044 q —'uajjod fo ssmu Auonos Fuispajas *aouaosn[ap 10 nun puo ayourumis pasi9sqy 7)—22vfins 07 Suyvoy nun [0104 231punumjs pasiosqp 'g — uoissiosqp 01 4014d nun 70104 aypurunys ynn junjd ayurunig "p — "DUDAS" 01101214 “9DPSUO] jujoq. 10 uoiv]ndod pprap uD ui eonoreyue sioqiqdury fo uoiuiod “€ INDI J Missouri Botanical Garden Annals of the 816 Volume 75, Number 3 OX 817 1988 Hydrophilous Monocotyledon Pollination e N e 10 o e gG au e€ o 0 = 0 = O successfu o search vehicles all search un vehicles 10 20 30 40 Diameter (mm) FIGURE 4. Size distributions so x Pike “all search ue icles' > represents the size distribution of 1, 00 med near Queenscliff, Victoria. The maqa on 0 search vehicles chosen at random, while the distribution labeled "successful search vehicles" represents the size distribution of 100 search Delicias that collided with stigmas suggests that larger search vehicles do, in accor- dance with theoretical predictions, have a much greater chance of dini id dis $E: a stig- ma than small search vehic Many questions, however, remain concerning two-dimensional pollination. What is the role of stigma morphology on search vehicle encounter rates? Computer simulations (P. A. Cox & J. Seth- ian, unpubl. data) indicate filamentous stigmas to be far more efficient in pollen capture, but the relative importance of stigmatic texture is un- known. The stigmas of the five surface-pollinated genera we studied are smooth, while the stigmas of the two-dimensional submarine-pollinated species Thalassia testudinum are densely papillate (Cox & Tomlinson, 1988). Major questions also remain concerning mixed modes of pollination. Is it possible that some seagrasses have both surface and sub- marine modes of pollination as has been suggested by Hartog (1970) for Phyllospadix and Zostera (Cock, 1980)? If so, what are the relative impor- tances of these different modes? Another important question concerns the effect of two-dimensional pollination systems on breeding systems. If the ollen shadows are far smaller than the average clone size of the species, will there be strong se- lection for obligate outbreeding systems (Cox, 1988) such as dioecism? It is of interest in this regard that 75% of the seagrass genera are dioe- cious. Gene flow between populations mediated by pollen exchange must be exceedingly rare in sea- grasses (Cox, 1983b) and even rarer in freshwater hydrophilous plants, since the pollen cannot move from one pond to another (Cook, 1987). The evo- lutionary ecology of such genetically isolated pop- ulations merits further investigation. Finally, con- vincing evolutionary explanations have yet to be made of lack of a developed exine in the pollen of many hydrophilous plants (Ducker et al., 1978). We believe that more light will be shed on these questions as further ecological studies are made of two-dimensional pollination systems. 818 Annals of the Missouri Botanical Garden LITERATURE CITED ASTON, fia I. 1973. Aquatic Plants of Australia. Mel- e Univ. Press, Carlton, Victoria. BLACK, J "M. The flowering and fruiting of Pec- tinella antarctica (Cymodocea antarctica). Trans. oy. Soc. Australia 37: Brock, T. D. 1966. Princ ciples of ida oe — Pw por Cliffs, New Jer Cock, A. . M. 980. Flowering, pollination, and E in uu marina L. Aquatic Bot. 9: 220, Cook, C. D. K. 2. der mechanisms in the n Pp. 1 in J. J. Symoens, S. S. Hoope P. Compére ea Studies on Aquatic Vascular Plants. Royal Botanical Society of Belgium, Brussels. 1987. Vegetative growth and p dre in some aquatic weeds. Pp. 217-225 M banska (editor), Differentiation laus in Hehe: Plants. Academic Press, London. Cox, P. A. . Vertebrate Pollination and the Main- tenance of Unisexuality in Freycinetia. Ph.D. Dis- sertation. Harvard University, Cambr ridge, Massa- chusetts. 1983a. Extinction of the Hawaiian avifauna resulted in a change of jon for the ieie, Frey- cinetia arborea. Oikos 41 199. 198 Search theory, random motion, and the convergent evolution of pollen and spore mor- phologies in aquatic plants. Amer. Naturalist 121: 9- 1984. Chiropterophily and ornithophily in Freycinetia in Samoa. Pl. Syst. Evol. 144: 277- 1988. Monomorphic and "gru sexual strategies: a modular approach. P 0-97 in J Lovett Doust & L. Lovett Doust hi eus Plant Re- productive Ecology: Strategies and Patterns. Oxford Univ. Press, Oxford. Hydrophilous pollination. Ann. Rev. Ecol. Syst. (in press). R. B. Knox. 1986. d werte: postulates nd two- eae are ue Pp. 48-57 in E. G. Williams, R. B. D. Irvine Verdad ASE nation ‘86. Univ. Melbourne Press, Melbou aol pollination ir in | hy- drophilous plants: c nt evolution in the genera Halodule Cymodoceaceae) Halophila (Hydrochar- itaceae), R Va eine and Lepilaena (Zan- nichelliaceac) he . J. Bot. (in press). & P. B. Tow. 1988. The pollination Thalassia testu- 75: 9 J. Bot. ecology of a seagrass, dinum at. Amer. DEGE NER, O. Plants of Hawaii am Park. Braun-Brumfield, Ann Arbor, Michig Dr Kruir, P. 1927. Microbe Hunters. RA Whilst- able. DUCKER, S. . J. Foorp & E B. KNox. 1977. Biology of Jua n seagrasses: the genus Amphi- bolis C. Agardh “ataq payani q Austral. J. Bot. 6 E TT & R. B. Knox. 1978. Biology of m seagrasses pollen development and sub- marine pollination in Amphibolis antarctica and Thalassadendron ciliatum (Cymodoceaceae). Aus- tral. J. Bot. 26: 265-285 ERNST-SCHWARZENBACH, M. 1945. Zur re einiger Hydrocharitaceen. Ber. Schweiz. Bot. Ges 3- Sheen K. & L. VAN DER Pur. 1979. The Principles Pollination. Ecology, 3rd edition. Pergamon, Ox- ro. J.C. 1968. aa alot sobre la biologia oral y morfologia de la mogetonacea Ruppia currhosa (Potag. bases (= R spiralis L. ex Dum.). 78. Populsuon Biology of Plants. Ac- emic ie London. ji. C. DEN. 1970. The Sea-Grasses of the World. on. N . Wetensch., Afd. Natuurk., Tweede Sect. 59. North Holland, Amsterdam. HrnscH, R R. J. GRIEGO Brownian motion and potential theory. Sci. Amer. 220(3):66- 74. JENNINGS, J 1 e Prehistory of Polynesia Kocu, R. 1880. UM into the Traumatic Infectious Diseases. Transla "ii by W. Cheyne. The Sydenham Socie ty, Lon Koiran, B.O. 1956. The theory of is II: Target detection. Operations Res. 4: 503-531. McCowcHiE, C. A., S. C. DUCKER & R. B. Knox. de Biology of Australian seagrasses: floral developme and morphology in uic aM i nde ustral. J. Bot. 30: 251- MEMIAN. C. 1980. Flo pte under controlled con- an oe hemprichii from Kenya. Aquatic Bot. 8: ; E Di M 1980. Reproduction in seagrasses: na- ture of the pollen and receptive surface of the stigma in the Hydrocharitaceae. Ann. Bot. (London) 45: 257-271. 1981. Reproduction in seagrasses: pollen de- velopment i in Thalassia hemprichii, Halophila sti- ulacea and “w ssp ciliatum. Ann. Bot m 48: Me CKER 3 R. es iot 2 marine pont on. Sci. Amer. 244: 43. , C. A. McCoNcHIE S C buka ë K B ER 1980. Unique adaptations y ae pollinstion in seagrasses. Nature 286: 489. Sub- . 1983. Re duction in seagrasses: pollination in 4mphibolis ant- oy. Soc. London, Ser. B, Biol. Sci. PROCTOR, . M. & P. Yeo. 1973. The Pollination of Flow- ers. Collins, London. iie P.B. 19 Anatomy of the Mor ledons. VII. Holobis (Alismatidae) Gate jd Puce. Clarendon Press, Oxfor TROLL, W. 1931. Botanische Mitteilungen aus den Tro- pe n II. Zur Morphologie und Biologie von Enhalus coroides (Linn. f.) Rich. Flora 125: 427- VAN Minos EN, W. R. J. van WiK & J. i VAN DER ZEE. 1 On the pollination mechanism of some eu- rysaline Potamogetonaceae. iiec Bor. 14: 339- VERHOEYEN, J. T. 1979. The ecology of Ruppia- minated ORENSE in Western n Europe. I. Dis- gies of Ruppia representatives in relation to Bot. 6: 197-268. their ouis Aquatic Wyer, R. B. 1917. The pollination of Vallisneria spiralis. Bot. Gaz. (Crawfordsville) 63: 135-145. BREEDING SYSTEMS, POPULATION STRUCTURE, AND EVOLUTION IN HYDROPHILOUS ANGIOSPERMS' Donald H. Les? ABSTRACT True hydrophily occurs in 18 submersed angiosperm genera. X 17 are monocots, 12 are marine, and 5 contain an nuals. include dioecious spec of genetic Hie in these plants. Many w ree genera necessar y for outcrossing. In hydrophiles, decreased see e predominant ies of asexu ine ond sexually t ermaphroditic species, 8 have s. The prevalence of dicliny in hydrophiles has sd to assumptions of outcrossing and high levels water- cimi species, howev is “i oe of species. E. monoecious species, and 13 have ver, may often fail to meet all conditions utput associated with dicliny increases the probability ve. Annual Gisk qm genera have higher species diversity and possibly greater Ede variability uie sh [inde populatio Aquatic plants are notorious for their anatomi- cal, morphological, and physiological peculiarities. A question relating to their unusual biological fea- tures is whether the course of evolution in aquatic plants differs fundamentally from that of terrestrial plants. This question is difficult to address because water plants are not a monophyletic group; there- fore, the assessment of **peculiarities" would nec essarily involve interpreting patterns of convergent evolution. Certainly, there is no reason to suspect that hydrophytes do not follow evolutionary paths dictated by the same basic factors that have influ- enced terrestrial plant evolution. On the other hand, there are evolutionary patterns associated with the "biological group" of submersed hydrophilous an- giosperms that warrant consideration. A particu- larly striking feature of these species is their slow rate of evolutionary diversification, a conclusion revealed by several lines of evidence. Data tabulated for 31 principally aquatic fam- ilies (from Cook et al., 1974) furnish an average of about eight species per genus. The number of extant species in most hydrophilous angiosperm genera, however, is fewer than eight (Table 1). In comparison, some genera of nonhydrophilous sub- mersed aquatic plants (e.g., Potamogeton, Myr- iophyllum, Ranunculus) may contain 35-100 or more species. Not all hydrophilous genera, how- ever, have few species; Najas contains 35-50 species, the largest number among hydrophilous genera (Table 1). Furthermore, the fossil recor indicates that at least several hydrophilous angio- sperms have undergone a prolonged period of mor- phological “‘stasis.”’ In an evolutionary study of the genus Ceratophyllum, Les (1986a: 32) observed: "One interesting aspect of the fossil record is that most of the extinct taxa can be associated with the extant genus Ceratophyllum, and often with extant species ...." Similar statements were made by Hartog (1970) with respect to the hydrophilous seagrasses, e.g. (p. 15): “It is noteworthy that these Tertiary [seagrass | fossils all belong to still existing genera and that at least two of them can be iden- tified with still existing species." And (p. 30): “° evidence for the great age of recent species . . . is supported by the fact that fossil remains of Cy- odocea from the European Eocene e can be iden- tified with still existing species.” Hartog (1970) ! Į express sincere thanks to Tom Philbrick, Kent Holsinger, and Spencer Barrett for their helpful advice and discussions ofthe to < presented i in this paper. l also thank Photographic Services at the University of Wisconsin- Milwaukee for help in preparing the figures. 2 Department of f Biological Sciences, The University of Wisconsin- Milwaukee, Milwaukee, Wisconsin 53201, S.A. ANN. MISSOURI Bor. GARD. 75: 819-835. 1988. 820 Annals of the Missouri Botanical Garden TABLE l. Synopsis of p de genera, with submersed hydrophilous species. H — ee tbc E = si i ip Mo = monoecious; D = dioecious; Hm = hermaphroditic; P = perennial; A = annual; M = ne; B = brackish; F = fesh Tanod from Aston, 1973; Cook, 1982; Cook et al., 1974; Cox, 1983; Edwards, 1976; Hartog, 1970; Haynes, 1977; Haynes & Holm-Nielsen, 1987; Hutchinson, 1975; Les, 1986a; Sculthorpe, 1967). Num- r of Sexual Sexual Life Species Reproduction Condition Form Habitat Pollen DICOTYLEDONS Ceratophyllaceae Ceratophyllum (H) 6 rare to common Mo P B,F precocious MONOCOTYLEDONS Hydrocharitaceae Elodea (E) 5 rare D, Mo P F globose Halophila (H) 8 rare to common D, Mo P M chains & precocious Thalassia (H) 2 rare to common D P M chains & precocious Najadaceae Najas (H) 35-50 prolific Mo, D AP BF precocious Posidoniaceae Posidonia (H) 3 rare to common Hm P M filiform Ruppiaceae Ruppia (E) 1-7 common Hm AP M,B,F chains Zosteraceae Heterozostera (H) 1 common Mo P M filiform Phyllospadix (H) 5 common D P M filiform Zostera (H) 12 rare to common Mo AP M filiform Zannichelliaceae Althenia (H) 2 common D P B,F lo Amphibolis (H) 2 common D P M filiform Cymodocea (H) 4 rare D P M filiform Halodule (H) 6 rare D P M filiform Lepilaena (E, H) 4 common Mo, D AP BF lo Syringodium (H 2 mo D P M filiform Thalassodendron (H) 2 infrequent D P M filiform Zannichellia (H) 1-5 common Mo, D, Hm A,P BF lobose & precocious attributed the slow rate of evolution in seagrasses to the relative reis ity of the marine environ- ment, and Les (1986a) related stasis in Cerato- phyllum to interactions of hydrophily and aspects of the breeding system. Are the slow evolutionary rates that apparently characterize various unrelated hydrophile species a consequence of their unique pollination system? This question has prompted the present study to review not only the pollination system but the over- all reproductive biology of hydrophilous angio- sperms. Specifically, the intent of this paper is to hypothesize possible evolutionary implications as- sociated with peculiarities of hydrophile reproduc- tive biology. Establishment of a theoretical basis will provide a means for testing hypotheses em- pirically. Although this symposium focuses on freshwater angiosperms, consideration must also be given here to marine angiosperms, which dom- inate this biological group. SALIENT FEATURES OF HYDROPHILY True hydrophily includes hyphydrophily where pollen is transported exclusively under water, and ephydrophily where pollination occurs at the sur- face (Faegri & van der Pijl, 1979). Various mech- anisms by which pollen is transported above the water surface (e.g., Enhalus, Vallisneria) mimic hydrophily but are not considered here Volume 75, Number 3 1988 Les Hydrophilous Angiosperms Hydrophily is viewed as a derived condition in angiosperms and probably developed from both anemophily and entomophily (Faegri & van der Pijl, 1979); however, the immediate precursor to hydrophily in most instances appears to have been anemophily (Les, 1988b). Hydrophily is unique to submersed aquatic angiosperms and occurs only within 18 genera which represent seven families d which constitute a heterogeneous group phy- logenetically. The taxonomic distribution and se- ected features of these genera are summarized in Table 1, from which several associations are ap- parent. Except for Najas, there are 1-12 species in hydrophilous genera. The frequency of sexual reproduction ranges from common to rare. Nearly all genera are principally perennial, and all (even annuals) possess mechanisms for vegetative repro- duction. An overwhelming consistency is the di- clinous sexual condition (monoecy or dioecy), with hermaphroditic flowers occurring in only three gen- era. Most hydrophilous genera are marine; sevén genera occur in freshwater. Like anemophily, hydrophily is an abiotic pollen- transfer mechanism and therefore inherently “wasteful” (i.e., much of the pollen produced does not contact a receptive stigma) because of nondi- rectionality (Faegri & van der Pijl, 1979; Cox, 1983). The three-dimensionality of hyphydrophily leads to high pollen wastage, although higher ef- ficiency may be attained in shallow water where pollen loads may concentrate. Ephydrophily re- duces pollen wastage by confining the dispersal of grains to the two-dimensional water surface (Faegri & van der Pijl, 1979). Because of the stochastic nature of hydrophily, the highest level of efficiency is probably achieved with autogamous pollinations, where the shortest transport distance is involved, and presumably decreases with the greater dis- tances involved in geitonogamous (involving dif- ferent flowers on one individual) or xenogamous (involving flowers on different individuals) polli- nations. Underwater pollination in some species is enhanced by the reduction of water currents over plant beds (which assists pollen deposition) and the relatively large area of pollen influence around female flowers (Ackerman, 1983, 198 Hydrophile pollen exhibits structural modifica- tions that tly maximize transport efficiency. The surface area of most hydrophile pollen is in- creased variously (Table 1). In some species, glob- ular pollen grains form filiform chains, or the grain itself is highly elongate (Pettitt & Jermy, 1975; Cox, . In other species, precocious germi- nation of the pollen tube (Fig. 1A) may increase surface area to maximize capture by the stigma (Sculthorpe, 1967; Faegri & van der Pijl, 1979; Cox, 1983; Les, 1986a). In Ceratophyllum, the surface area is increased further by the occasional branching of pollen tubes (Fig. 1 B), and grains with precocious pollen tubes have been observed to mass together (Sehgal & Ram, 1970), thereby mimick- ing pollinia. Typically, the exine of hydrophilous pollen is highly reduced (Pettitt & Jermy, 1975). Hydrophily and anemophily are the major abiot- ic pollination systems in plants. Understandably, the two systems share certain similarities in their floral syndromes, such as frequent dicliny, reduced perianths, high pollen/ovule ratios, reduced pollen exines, and enlarged receptive surfaces (Faegri & van der Pijl, 1979; Les, 1986a; Sculthorpe, 1967; Whitehead, 1969). Both systems lack the speci- ficity and constancy associated with biotic polli- nation systems. There are, however, various dif- ferences between the two abiotic systems. Anemophily relies on the availability of wind, whereas availability of water is seldom a liability to submersed plants. Pollen dispersal distances of anemophiles are limited only by the longevity of grains and their ability to be carried aloft. Dispersal distances in hydrophiles are restricted entirely to the dimensions of the body of water they inhabit. For freshwater species, this distance may be quite small, e.g., a pond, pool, or small lake. A further consideration is that hydrophiles are generally con- fined to shallow depths, and pollen transported to deeper water has little chance of encountering plants. With respect to ephydrophily, Cook & Urmi- Konig (1985: 118) stated: “Pollen transfer on the efficient over = ® surface of the water may well distances best measured in centimeters, but for distances measured in meters or kilometers it will become inefficient or even ineffective.” Hydrophile pollen is not subject to desiccation, but studies are needed to determine the duration of its viability. In Zostera, pollen retains ET for over 48 hours (De Cock, 1980). Although pollen of hyphydrophias frequently contains starch to pr (Sculthorpe, 1967; Les, 1986a), it is conceivable that water currents may facilitate resuspension of grains (particularly in very shallow water), which may increase opportunities for con- tact with submerged receptive stigmas. A HYPOTHESIS OF OUTCROSSING IN HYDROPHILES Most angiosperms are hermaphroditic (Bawa & Beach, 1981), with dioecy (depending on the re- gion) occurring in only 2-28% of species (Bawa, 1980) and monoecy equally rare (Grant, 1975). 822 Annals of the Missouri Botanical Garden FicURE 1. Pollen tube modifications in Ceratophyllum demersum. Fu, — — A. Precocious germination of grains 5 um. showing elongated pollen tubes. —B. Branching of pollen tubes (at arrow) . Scale bars — In contrast, 13 of the 18 genera (72%) of hydroph- ilous angiosperms have dioecious species; 8 genera (44%) have monoecious representatives; and only about 8.7% of hydrophile species are hermaph- roditic (Table 1). The association between dioecy and hydrophily was noted by Sculthorpe (1967), who did not provide an evolutionary explanation for the co-occurrence of these traits. Others, however, have viewed the predominance of dicliny in hydrophiles as evidence of an inevitable association with outcrossing and production of ge- netically variant progeny (Hartog, 1970; Pettitt et al., 1981). Although dioecy is associated with re- duced prolificity because of the presence of males iis 1980), it has been postulated (in seagrasses) that **. .. gains in survival attributable to contin- uous out- broad outweigh the disadvantages of diminished seed production” (Pettitt et al., 1981: 137). These conclusions can be interpreted as a hypothesis that embraces three components: 1) hydrophiles are characterized by outcrossing, which leads to the production of genetically variable off- spring; 2) the selective advantage of outcrossing in hydrophiles offsets the evolutionary costs asso- ciated with transitions to dicliny; and 3) dicliny has evolved in hydrophiles as a mechanism for pro- moting outcrossing. The appealing rationality of such a hypothesis belies the fact that it rests entirely on circumstantial evidence associated with sexual conditions and is not substantiated by more em- pirical evidence. Furthermore, it is difficult to ra- tionalize the slow evolutionary diversification of this group with a supposed history of outcrossing and prolific genetic variability. It is possible that few species have made the transition to hydrophily in recent time, and therefore the hydrophile genera have simply not had sufficient time to diversify. This interpretation, however, is incongruent with Volume 75, Number 3 1988 Les 823 Hydrophilous Angiosperms fossil evidence that indicates a great age of many hydrophile genera. Obviously, it is necessary to consider other sources of data that may be more pertinent in assessing the extent of outcrossing in hydrophiles. TESTING THE “OUTCROSSING HYPOTHESIS” In hydrophiles, the possibility of outcrossing (natural crossing as defined by Grant, 1975) exists only when several conditions are satisfied. One requirement is for sexual reproduction. A second Bo is for xenogamy, which assures that m of one individual reach the eggs of different individuale. Thirdly, the parents contributing ga- metes must differ genetically and their offspring must survive. By evaluating the ability of hydro- philes to satisfy these conditions, it may be possible to assess better the role of outcrossing in these species. To interpret fully the significance of out- crossing in hydrophiles, it is also necessary to un- derstand the relative level of inbreeding that may occur in outcrossing species. THE EXTENT OF SEXUAL REPRODUCTION IN HYDROPHILES There has been much discussion regarding the costs and benefits of sexuality in organisms. Using a group selection model, Lloyd (1980) argued that dioecious populations experience a cost of sex, whereas hermaphroditic populations do not. Fur- thermore, he pointed out that asexual reproduction results in greater prolificity than dioecy, raises the potential rate of increase, and may aid in the long- term persistence of asexual populations and species. He concluded, however, that asexuality does not appear to have been significant in group selection against dioecy because of the relatively lower evo- lutionary success of asexual lines. Nevertheless, hydrophyte reproduction occurs both sexually and asexually, most species have well-developed means of vegetative reproduction, and most reproduction in perennial hydrophytes is estimated to be asexual (Hutchinson, 1975). Approximately half of hydrophile genera are characterized by rare sexual reproduction (Table 1). A quote from Hartog (1970: 34) illustrates this point: **. . . inflorescences, flowers, fruits and seeds . . are not often found in most seagrasses, and in some species they are not known at all or only incompletely.” The rarity of fruiting in the fresh- water genera Elodea and Ceratophyllum is also widely recognized (Cook & Urmi-Konig, 1985; Les, 19862). One facet of sterility in dioecious hydrophiles is the low percentage of flowering in several species. The seagrasses Halophila stipulacea, Halodule beaudettei, H. bermudensis, H. ciliata, Cymo- docea rotundata, C. serrulata, C. angustata, and some Posidonia species are rare-flowering accord- ing to Hartog (1970). McMillan (1976, 1979, 1980) noted that environmental conditions have wide effects on the reproductive biology of many seagrasses. Even when flowering, dioecious hydro- philes may exhibit another facet of sterility. In Thalassia testudinum, fewer than 1% of plants may flower simultaneously in **beds," and beds are often unisexual (Hartog, 1970; Durako & Moffler, 1987). Such conditions may result in frequent ste- rility of the species, such as that reported by Ed- wards (1976). Similarly, plants of Thalassoden- dron ciliatum and Elodea species typically exist in unisexual colonies with plants of both sexes rare- ly coexisting (Hartog, 1970; Cook & Urmi-Kónig, 1985). Such circumstances surely contribute to low fruit production in these species. Cook & Urmi- Konig (1985) attributed the unisexuality of Elodea populations to differential competition for habitat by the sexes. Alternate explanations, however, in- clude the possibility that unisexual populations are derived clonally, or that sex expression is affected by environmental rather than by strictly genetic factors. A further restraint of sexuality is the rarity of seedling production in some plants with high seed output, e.g., Zostera noltii (Hartog, 1970). Agamospermy is possibly mistaken at times for sexual reproduction in some hydrophiles. In Hal- ophila stipulacea, cultured plants induced to flow- er produced no males, yet the females produced seed (McMillan, 1980), a good reason to suspect agamospermy. In H. hawaiiana, male and female plants are not known to coexist (Herbert, 1986). Although agamospermous reproduction is geneti- cally equivalent to vegetative propagation, one ob- vious difference is production of fruits allowing for "normal" dispersal. It would be an important con- tribution to test experimentally for agamospermy among other hydrophilous species. Although gene recombination via sexual repro- duction is viewed as important for response to changing or heterogeneous environmental condi- tions, genetic uniformity enforced by vegetative reproduction may be more advantageous for a species already adapted to uniform habitat condi- tions (Grant, 1981; Lloyd, 1980). Therefore, the relative uniformity of freshwater and marine en- vironments (Hartog, 1970; Tiffney, 1981) pro- vides one explanation for the ubiquity of efficient vegetative reproduction in hydrophytes. Annals of the Missouri Botanical Garden Demands on parental energy for reproductive effort have been implicated in compromises be- tween sexual and asexual avenues in some plants. In many instances, the relationship is inverse, i.e., high fruit production with low vegetative propa- gation and vice versa (Salisbury, 1942; Harper, 1977). In hydrophiles, free-fruiting species often exhibit less vegetative development than rare-fruit- ing species. Of the freshwater hydrophiles, poor vegetative development occurs in the genera Al- thenia, Lepilaena, Ruppia, Najas, and Zanni- chellia, which are all typically free-fruiting (Table 1). Of these genera, Najas, Zannichellia, and Lepilaena are mainly annual. In Ruppia, little biomass is allocated to reproductive structures in perennial species, whereas allocation to reproduc- tive structures in annuals is always much higher (Brock, 1982). The freshwater genus Elodea has coarse vegetative growth and low seed output. All species of Ceratophyllum are perennial, yet much higher seed output occurs in species with fine fo- liage than in the coarse-leaved species C. demer- sum (Les, 1986a). Sometimes in Zannichellia rel- atively robust plants have been associated with lower flower and fruit production and behave like perennials, whereas slender plants have higher flower and fruit production (Uotila et al., 1983). Similar associations are not as evident in marine hydrophiles, possibly because a well-developed veg- etative system for anchoring plants against forces of tidal currents and wave action is essential for marine existence (Hartog, 1970). In the marine genus Zostera, however, a relationship exists be- tween fruiting and shoot development. Arber (1920: 127) observed: “In Zostera marina... the fertile and sterile plants are readily distinguishable from one another, since in the fertile plant the stem is slender, erect, and much branched, while that of the sterile individual is thick, creeping, more lux- uriantly leafy, and anchored to the soil by adven- titious roots... .” Fu rthermore, annual individuals vegetative shoots and rhizomes (Keddy & Patri- quin, 1978). Although most species within the prin- cipally annual genus Najas have relatively fine foliage, the perennial species /V. marina is char- acterized by very coarse leaves and has the ability to form vegetative turions (Agami et al., 1986). Differential resource allocation to reproduction has been studied in terrestrial plants in some detail (Silvertown, 1982) and provides a convenient ex- planation for the association of high vegetative development with low sexual reproduction in some hydrophytes. It is possible that the development of effective perennating mechanisms in hydrophytes was accompanied by a sacrifice of sexual repro- duction. It could be argued that vegetative growth and reduced reproductive effort in perennials vs. high sexuality and reproductive effort in annuals are merely adaptive life history traits. On the other hand, the shoot dimorphism described above be- tween both fertile/sterile and annual/perennial in- dividuals of Zostera marina indicates that tradeoffs in reproductive allocation do occur. In Zostera marina, the different reproductive strategies (an- nual, perennial, sexual, asexual) are employed de- pending upon environmental circumstances (Phil- lips et al., 1983). Although other factors may be involved, it is possible that the relationship between low sexual reproduction and high vegetative de- velopment noted in hydrophiles and other hydro- phytes is due in part to constraints related to re- source allocation. From the above discussion, it is evident that sexual reproduction in water-pollinated plants is not commonplace. This conclusion is important because, despite any other factor, outcrossing can- not occur in hydrophile populations that reproduce only asexually. THE EXTENT OF XENOGAMY IN HYDROPHILES Hermaphroditic, monoecious, and dioecious sex- hydrophiles allow for three possible means of gametic exchange: autogamy, geitonog- amy, and xenogamy. Although outcrossing can oc- cur only when gametic exchange is xenogamous, xenogamy is possible with all three sexual condi- tions. Autogamy in many hermaphroditic terrestrial plants is prevented or restricted by mechanisms such as dichogamy, herkogamy, heteromorphy, and self-incompatibility (Faegri & van der Pijl, 1979; Lewis, 1979). Heteromorphy is associated with biotic pollinators (Lewis, 1979) and is unknown in al cand) hydrophiles. Incompatibility mechanisms have not been reported in hydrophiles, presumably because of constraints imposed by the water-liability of rec- ognition substances and by reduction of the exine in hydrophilous pollen (Pettitt & Jermy, 1975). Herkogamy, the spatial separation of sexes, is not characteristic of hermaphroditic hydrophiles. This leaves only the possibility of dichogamy, the tem- poral separation of sexes, as a means of preventing autogamy in hermaphroditic hydrophiles. Autogamy is possible only in three genera (Po- sidonia, Ruppia, Zannichellia) which have mem- bers with hermaphroditic flowers. Sexual repro- duction in these genera is common (Table 1). Aston Volume 75, Number 3 1988 Les 825 Hydrophilous Angiosperms (1973) described fruit production as “prolific”” for Posidonia, and the numerous fruits of Ruppia and Zannichellia taken from waterfowl stomachs (up to 4,000 and 10,000 per stomach, respectively) indicate high fruit production in these genera (McAtee, 1939). A single plant of Zannichellia is capable of producing more than two million seeds in six months (Yeo, 1966). Discounting apomixis, the prolific fruit production in these genera reflects their successful adaptation to hydrophily. Much of this prolificity, however, probably results from autogamy. In hermaphroditic Zannichellia (Aston, 1973), autogamy is possible due to the enclosure of stamen and carpels within the cuplike perianth, although the flowers may be dichogamous. Most species of Zannichellia, however, are monoecious and even some dioecy has been reported (Muhlberg, 1982). In monoecious Zannichellia, the arrangement of flowers results in functional bisexuality. Male and female flowers are typically adjacent and appear *, . . to rise as a group in a leaf axil” (Sculthorpe, 1967: 298). Although the anther of the male flower is raised above the carpels, pollen grains have a higher specific gravity than water and “... sink on to the peltate or tongue-shaped stigmas .. .” (Sculthorpe, 1967: 299). Arber (1920: 71) re- ported a similar scenario for Zannichellia. She noted that, “The anther dehisces and the pollen grains fall into the open mouths of the cornucopia- shaped stigma . . . ." These accounts suggest that dichogamy does not occur in Zannichellia and that selfing would be commonplace. Hutchinson (1975: 232), however, presumed that local tur- bulence would displace the descending pollen of Zannichellia, **. . . so that occasional cross-polli- nation can occur even when ... the male and female flowers [are] very close together." The de- scription of pollination in Zannichellia by Haynes & Holm-Nielsen (1987: 264) renders Hutchinson's presumption untenable: **... the anther of the staminate flower arches over the funnel-shaped stigmas of the carpellate flower. Pollen transfer is entirely underwater: it is released from the anther in a gelatinous mass and falls directly into the stigma." Although monoecious Zannichellia can- not be categorized as autogamous, these descrip- tions of pollination in the genus indicate predom- inant geitonogamy (virtually the genetic equivalent of autogamy). Even the effects of turbulence that Hutchinson proposed would probably result in gei- tonogamy rather than xenogamy, as the gelatinous pollen masses would be more likely to settle than to be transported laterally. Furthermore, the floral morphology of Zannichellia does not appear to be adapted for xenogamy. Pollen/ovule ratios are not known specifically in Zannichellia; however, they are probably relatively low, with only a single sta- men for every cluster of four one-ovuled female flowers. The sculptured exine of Zannichellia pol- len is atypical of hydrophilous angiosperms (Pettitt & Jermy, 1975). The pollen shape is globular and the grains do not form chains (Table 1), an indi- cation that it is not as well-adapted for transport over distances as that of other hydrophiles. For Zannichellia, the high percentage of fruiting is likely a result of autogamy in hermaphroditic plants or of geitonogamy in monoecious plants, rather than of xenogamy. Exceptions may occur with Zannichellia con- torta and Z. peltata, in which the anther filaments are much longer than those of other species, and their pistillate and staminate flowers arise at dif- ferent nodes (Talavera et al., 1986; Van Vierssen & Van Wijk, 1982). With the greater spatial sep- aration of anthers and pistillate flowers in these species, the opportunity for xenogamy is en- hanced. Haynes & Holm-Nielsen (1987) concluded that generally the pollination system in Zanni- chellia limits outcrossing but is valuable for the annual habit because pollination is essentially as- sured. The pollination system of Ruppia has been de- scribed in some detail. In Ruppia cirrhosa, the hermaphroditic-flowered inflorescence reaches the surface of the water but remains submersed; the anthers dehisce, are carried to the surface by air bubbles, and release pollen explosively when con- tacting the atmosphere. Cohering pollen grains cov- er the surface in chainlike strings and eventually reach the carpels, which are raised to the surface by bending of the inflorescence. In R. maritima, the discoid, peltate stigmas form a canopy above the anthers and trap the ascending pollen chains; the grains drift around the stigma to its surface, where some adhere and germinate. (For details see Arber, 1920; Gamerro, 1968; McCann, 1945; Sculthorpe, 1967.) In Ruppia it is difficult to ascertain the extent of autogamy. Frequent fruit production can be explained either by autogamy (such as described or R. maritima above) or by xenogamy. The observance of spreading surface pollen masses of Ruppia (Faegri & van der Pijl, 1979) demon- strates the potential for xenogamy. The two-di- mensional nature of the ephydrophilous mechanism “concentrates” the pollen, thereby increasing the chances for pollen capture. In aquarium studies, however, pollen of R. megacarpa remained mostly near the stigmas of the flower from which it was j= 826 Annals of the Missouri Botanical Garden released (Van Vierssen et al., 1982), an indication that selfing may occur frequently in the species. In any case, sexual reproduction in Ruppia plays a large role in its propagation (Edwards, 1976). Information on the reproductive biology of the marine Posidonia is too general to infer much of its breeding system. The flowers consist of three or four sessile anthers surrounding the simple car- pel, which terminates in a feathery, lacerate stigma (Sculthorpe, 1967). When the anthers dehisce, clouds of filamentous pollen are released into the water (Pettitt et al., 1981). Because the carpel is surrounded by anthers, the potential for autogamy might be quite high. Without direct observations and knowledge of possible dichogamy, however, this conclusion can only be implied. In monoecious plants, autogamy is prevented but geitonogamy is not. Hartog (1970), however, attributed cross-fertilization to all monoecious sea- grasses (Zostera, Heterozostera, Halophila de- cipiens) due to their protogyny. In the monoecious Zostera marina, the potential for xenogamy is indeed increased by protogyny (Arber, 1920). Pollen is shed in cloudy masses, with pollen tubes already beginning to protrude (Arber, 1920). Pollen is released either slowly un- der water or quickly in floating masses, which readi- ly adhere to any object coming in contact with the grains (De Cock, 1980). Despite widespread pro- togyny in Zostera (Sculthorpe, 1967), the prox- imity of male and female flowers within a spathe (Aston, 1973) suggests the possibility of occasional geitonogamy among flowers within the spathe. In fact, De Cock (1980) observed that self-pollination in Zostera will occur in the absence of cross-pol- lination and takes place widely in plants cultured in aquariums. The extent of geitonogamy in natural populations of Zostera is not clearly known. A similar situation exists in the related genus Heterozostera. Although male and female flowers occur within the same spathe, the flowers are pro- togynous, with the stigmas falling off prior to the opening of anthers within the same spadix (Aston, 1973; Hartog, 1970). Halophila decipiens, the sole monoecious species of the principally dioecious genus, is likewise protogynous, with male and fe- male flowers occurring within the same spathe (Hartog, 1970) n the monoecious Ceratophyllum demersum, xenogamy is probably infrequent. The rarity of flowering and the aggressive vegetative growth of this species greatly limit sexual reproduction (Les, 1985). When sexual reproduction occurs, the self- compatibility of the species allows for geitonogamy (Les, 1980, 1985). Although xenogamy is possible, transport of pollen tends to be within large clones (Les, 1986b Flowers of the monoecious Lepilaena australis are similar to those of Zannichellia but appear to be more conducive to xenogamy. Female flowers are borne on the upper parts of the plant and occur at the ends of peduncles up to 14 cm long. The male flowers are very short-stalked and occur on the lower portions of the plant (Aston, 1973). In this arrangement, geitonogamy would require the upward transport of pollen, and lateral movement of pollen would be more likely to facilitate xenog- amy. In monoecious Lepilaena preissii, however, the male and female flowers are clustered together (Aston, 1973) and here geitonogamy is more likely. Pollination in the monoecious Lepilaena cylindro- carpa is ephydrophilous. Stigmas create depres- sions in the water surface into which floating pollen grains released from submersed anthers were drawn (Van Vierssen et al., 1982). This arrangement would facilitate geitonogamy (as pollen from the same plant would be in closest proximity to the stigmas), although xenogamy may occur as well. The freshwater genus /Vajas is predominantly monoecious (with only one dioecious species) and annual (Haynes, 1977). The pollination biology of Najas is not known in detail, but several aspects have been described. Sculthorpe (1967) observed that pollen tube germination is precocious, often as the microspores are released from the floral envelope. He reasoned that the dense growth of plants would place male and female flowers to- gether, and that liberated pollen would be (p. 301) *... caught haphazardly on the elongated stig- mas." Aston (1973) observed that pollen is dis- charged apically through an opening in the floral envelope and is transported through the water to the stigmas. There is no indication that /Vajas possesses any mechanisms to prevent geitonogamy. Dioecy is the only sexual condition that ensures xenogamy, and all dioecious hydrophiles must be regarded as possessing a high outcrossing potential. Dioecy does not guarantee outcrossing, however, which will occur only when all conditions have been satisfied. THE EXTENT OF GENETIC VARIATION IN HYDROPHILE POPULATIONS The genetic structure of a hydrophile population ultimately determines the extent of outcrossing. In sexual and xenogamous populations, outcrossing will occur whenever sexual reproduction is con- summated between genetically different individu- als. How different genetically are individuals in Volume 75, Number 3 1988 Les Hydrophilous Angiosperms TABLE 2. Flowering frequency and short-shoot sex ratios for five populations of Thalassia testudinum. M — male; F = female (adapted from Durako & Mofjler, 1985a). TABLE 3. Flowering frequency and plant sex ratios in three species of dioecious Elodea; F — female, M — male (computed from data in Catling & Wojtas, 1986 (A) ; Cook & Urmi-Konig, 1985 (B)). Flowering Sex Ratio % in Sex Ratio Sites Frequency M:F Species Flower F:M Cockroach Bay 38% 1:1 A. Elodia bifoliata 53 1.8:1 Egmont Key 26% 1.7:1 B. E. bifoliata 87 1.2: Big Coppit Key 25% 1:3.1 A. Elodea nutallii 32 1.62 Lassing Park 22% 1.4:1 B. E. nutallii Ma 1.2:1 Dp Name Bey 376 28 A. Elodea canadensis 32 1.4:1 B. E. canadensis — 1,23 hydrophile populations? Few published studies have addressed this essential question. Dioecy and outcrossing are typically equated despite unresolved questions whether heterosexual individuals of dioecious plant species always differ genetically. Although sex in the majority of di- morphic plant species is believed to be determined solely by genetic factors (Lloyd & Bawa, 1984), diphasic responses have been reported in many dioecious plants, including species reported to have sex chromosomes (Freeman et al., 1980). If di- phasic sex changes can occur in hydrophiles that reproduce extensively by vegetative growth, then there would be at least the potential for clonally derived individuals to express different sexes. Un- fortunately, there is little available evidence to pro- vide a satisfactory resolution to the question of whether such changes occur in dioecious hydro- philes or not. In addition to direct observation of population sexuality over several seasons, Lloy Bawa (1984) regarded consistent sex production in widely spaced ramets or branches and sex ratios of unity as circumstantial evidence of phase stability in dioe- cious plants. Few studies of dioecious hydrophiles have provided direct observation of phase stability or change. Grey & Moffler (1978) reported an overall female-biased sex ratio of 3: 1 for Thalassia testudinum, although they observed a range of ratios from 21:0 to 1: 1. Durako & Moffler (1985a) determined that three out of four populations of T. testudinum were characterized by male-biased ratios (Table 2). Further studies of Thalassia dem- onstrated that male-biased ratios occurred only in one of three years of observation and correlated with highest seed output (Durako & Moffler, 1985b). There was no apparent relation of the ratios to flowering frequency; male bias occurred at flowering frequencies from 3-38% (Table 2). In Thalassia testudinum, the variation of sex ratios among populations and their deviation from unity may indicate that sex expression in this species is affected by environmental conditions. Durako & Moffler (1985a, b), however, attributed the yearly differences in the sex ratios not to sex change but to annual variation in the density of males; the female densities remained fairly constant tempo- rally. Biased sex ratios also occur in other dioecious hydrophile species. Estimates of flowering frequen- cy and sex ratios computed from data published by Cook & Urmi-Kónig (1985) and Catling & Wojtas (1986) for three dioecious and ephydroph- ilous species of Elodea indicate female-biased ratios of 1.2-1.8:1 (Table 3). In Elodea, the highest female bias was found in E. bifoliata, the most frequently flowering species (Table 3). Greater fe- male-biased sex ratios occur in dioecious marine hydrophiles, epitomized by Phyllospadix, with an approximately 12:1 ratio of female to male plants (Dawson, 1966). Using specimen lists cited in Har- tog (1970), flowering frequencies and sex ratios were computed for seven dioecious species of hy- drophilous seagrasses (Table 4). There are some indications of stability in the sex ratios of these species. The 11:1 ratio calculated for Phyllo- spadix approximates the 12:1 ratio reported in- dependently by Dawson (1966); both the frequency of flowering and the sex ratio are nearly equal for two species of Syringodium. Although sex ratios of three species of Elodea are similar, Cook & Urmi-Kónig (1985) have discussed various aspects of sex instability in the genus. Sex ratios of unity were noted only in Thalassia testudinum, Halo- dule uninervis, and Thalassodendron ciliatum; most other ratios were female-biased (Tables 3, 4). An interesting trend in the data is a possible re- lationship between sex ratios and flowering fre- quency. À decrease in the frequency of flowering is associated with a lower female bias in the sex ratios; lower floral frequencies appear to be asso- 828 Annals of the Missouri Botanical Garden TABLE 4. Flowering frequency and plant sex ratios in seven dioecious marine hydrophiles; F = female, M = male (computed from data in Hartog, 1970). TABLE 5. Sex ratios in a population of Cerato- phyllum demersum from Okauchee Lake, Wisconsin, U. Flowe Sex Ratio Species (seasonal) F:M Phyllospadix scouleri 83 List Phyllospadix torreyi 78 4.2:1 Syringodium filiforme 39 1.4:1 Syri ngodium um 38 1.3:1 Halodule uninervis 18 1:1 Thalassodendron ciliatum 12 1:1 Cymodocea nodosa 12 1:3 ciated with equal or male-biased ratios (Table 4). This trend is also apparent in Elodea, but to a much lesser eee Gaia 3 ie ratios of the dioecious t bias at low flowering nnda of 15-19% (Kay, 1971). Syringodium flowers commonly, and pro- duces mostly female flowers (Kay, 1971). An ex- ception to this trend occurs in the infrequently male flowering genus Cymodocea, which has been ob- served to produce only female flowers (Kay, 1971). The absence of male flowers, however, was possibly due to the sampling of a single clonal population (Kay, 1971). In other populations of Cymodocea, however, sexes co-occur but are partitioned spa- tially into unisexual zones (Caye & Meinesz, 1985). It is possible that the relationship of reduced fe- male-bias and low flowering frequency may result in part from sampling error. Sex-ratio data must be interpreted cautiously. The different methods of assessment (e.g., plant ratios, short-shoot ratios, single population ratios, multiple population ratios) and wide variability in sample size are only two shortcomings. Further- more, there have been no real efforts to distinguish between genets and ramets in populations where studies have been conducted. Because of the ef- fective system of vegetative reproduction in hy- drophiles, sex-ratio data may be misleading. Apo- mixis and other factors that can influence sex ratios have been summarized elsewhere (e.g., Opler & Bawa, 1978). Ultimately, a precise knowledge of sex expression in dioecious hydrophiles will be re- quired before any reliable conclusions can be drawn from sex-ratio data. Less is understood of sex ratios in monoecious hydrophiles. In Ceratophyllum demersum, plants fruit rarely, but fertile specimens are typically male- biased (Les, unpubl.). Sex ratios computed for a A. M = male; F = female; * = female-biased. Number Average Num of Com- ber of Female Plant bined Male Flowers Sex Sex a Flow = Ratio Ratio Number ers fruits (M:F) (M:F) 1 22 3 7.3:1 7.3:1 2 0 0:2* 4.4:1 3 0 1 (1) 0:2* 3.1:1 4 15 2 7.5:1 4.1:1 5 12 5 2.4:1 3.5:1 6 51 6 8.5:1 5.0:1 T 7 9 0.8:1* 3.7:1 8 39 10 3.9:1 3.7:1 9 14 9 1.6:1 3.3:1 10 l 2 0.5:1* 3.2:1 ll 1 0 1:0 3.2:1 12 15 7 2.1:1 3.1:1 13 7 3 2.3:1 3.1:1 14 3 8 (3) 0.3:1* 2.6:1 15 1 0 1:0 2.7:1 16 1 2 0.5:1* 2.6:1 17 10 1 10.0:1 2.7:1 18 5 3 1.7:1 2.411 19 0 6 0:6* 2.031 20 0 5 0:5* 2.3:1 21 2 5 0.4:1* 2.2:1 22 6 1 6.0:1 2.3:1 23 10 5 2.0:1 2.2:1 24 28 38 0.7:1* 1.8:1 25 24 12 2.0:1 1.8:1 26 6 7 0.9:1* 1.8:1 27 16 5 3.2:1 1.8:1 28 26 10 2.6:1 1.9:1 29 12 6 2.0:1 1.9:1 30 12 7 1.791 1.9:1 Total 346 184 — 1.9:1 fertile population of C. demersum from Wisconsin, U . provide some insights into its sex ratios. The ratios were calculated by counting all flowers on 30 dissected plants. The range of sex-ratio vari- ation in this population is extensive. Most plants are male-biased (up to 10:1); however, nearly a third of the sample is female-biased (Table 5). Al- though the species is monoecious, unisexual plants were observed for both sexes. The overall sex ratio was male-biased at 1.9:1. Fruiting was low (2%), and fruits were found only on plants with female- biased ratios. In species of Ceratophyllum where fruiting is more common, sex ratios appear to be less male-biased (Les, unpubl.). Reasons for the Volume 75, Number 3 1988 Les Hydrophilous Angiosperms wide amplitude of sex distribution on Ceratophyl- lum plants are not clear, but it is apparent that sex ratios are not rigidly fixed genetically. The potential for environmental influence of sex expres- sion in Ceratophyllum was demonstrated by the sole production of male flowers on plants of C. demersum and C. echinatum grown under contin- uous illumination (Les, 1980). Plants from the same population observed in the field were normally mon- oecious. In Zostera marina, a consistent sex ratio of one pistil to one anther was observed from a population in France, despite wide variability in the flower number per spathe (Jacobs & Pierson, 1981). In monoecious Zannichellia, the close association of male and female flowers indicates a probable 1:1 ratio. Such consistent values may indicate tighter genetic regulation of sex expression in these species. Sex ratios of other monoecious hydrophiles have not been studied in any detail. Although hydrophily is often equated with out- crossing, a critical point is made by Faegri & van der Pijl (1979: 41): “*. . . hydrophilous mechanisms give no guarantee against autogamy. However, the gregarious habit of the plants in question will gen- erally cause allogamy and counteract auto- and geitonogamy unless the whole meadow represents a single clone" [emphasis mine]. The extent of clonal growth in hydrophiles is surely an important determinant of their population structure. Hutchinson (1975) recognized that most repro- duction in hydrophytes is asexual, and as a result (p. 238) “... large clonal populations are likely to be very common in lacustrine angiosperms." He also observed that for perennial species (p. 233), *...any given specimen is much more likely to have arisen by asexual than by sexual processes." An important difference between asexual repro- duction in terrestrial plants and aquatic plants is that vegetative buds, fragments, turions, etc. in the latter can facilitate dispersal without the plant- ing requirement necessary for fragments of most terrestrial species (Hutchinson, 1975). This feature is evidently one of the most important mechanisms for dispersal in aquatic plants, particularly those that deii small janisa. of seeds. The effi- tot n ciency o is E emplifica by the spread of Elodea canadensis across Europe, which occurred within 50 years following its introduction from an uncertain North American source. This feat was accomplished en- tirely by vegetative means, as the species is dioe- cious and (with one exception) only female plants were introduced (Cook & Urmi-Kónig, 1985; Grant, 1981). Cook & Urmi-Kónig (1985: 118) remarked that the **. . . aggressive vegetative growth [of Elo- dea] leads to . . . the consequence that one habitat becomes fully occupied by one genotype." Similar conclusions have been drawn for other hydrophiles. In Cymodocea serrulata, the “only” means of medium- or long-range dispersal is by vegetative fragments, and clonal growth patterns are typical due to the vigorous vegetative growth (Kay, 1971). In C. nodosa, seeds are not dissem- inated, and sexually derived offspring must be spa- tially and temporally restricted (Caye & Meinesz, 1985). Likewise, vegetative reproduction is more prevalent than sexual reproduction in the main- tenance and spread of Thalassia (Kay, 1971; Grey & Moffler, 1978). In Halophila, Thalassia, Cym- docea, and Halodule, sexual reproduction is rare and reproduction is mostly vegetative (Edwards, 1976; Jacobs & Dicks, 1985). For Halophila ha- waiiana, Herbert (1986: 101) stated that **. . . it is possible that the monospecific meadows of Hal- ophila in the Hawaiian islands are made up of plant material from a single genetic individual." According to Obermeyer (1966), propagation of Halophila ovalis is mainly vegetative and may lead to the formation of **homogeneous colonies.” Zostera marina reportedly flowers infrequently and maintains stable population sizes by vegetative re- production (Harrison, 1979). Both Zostera and Heterozostera reproduce asexually by unusual vegetative propagules (Cambridge et al., 1983). Ceratophyllum demersum is rare-flowering and highly clonal (Les, 1986b). Haynes & Holm-Niel- sen (1987) stated that all Zannichelliaceae “grow clonally.” Widespread vegetative growth in hydrophilous species may lead to the formation of large, genet- ically uniform populations in which outcrossing would be thwarted even among sexual, xenogamous individuals. Hutchinson (1975), however, noted that the few aquatic annuals such as Najas are excep- tions to the generalization that most hydrophytes reproduce asexually. Development of an annual habit in hydrophiles, where consistently high levels of seed production are essential for survival, may have provided an escape from constraints on out- crossing imposed by dense clonal growth. Although vicinism may also occur in diclinous annuals such as Najas, there is at least a much higher potential for dispersal of sexually derived propagules to other sites, which betters the chances that future gen- erations will breed with genetically different indi- viduals. The foregoing discussion cites only circumstan- 830 Annals of the Missouri Botanical Garden TaBLE 6. Summary of enzyme variability in species from 13 hydrophilous angiosperm genera. * = en zym displaying intra- or interpopulational variability at one or more loci; all other enzy mes monomorphic (compiled from A: McMillan et al., 1981; B: Les, 1986b; C: McMillan, 1981; D: McMillan & Williams, 1980; E: McMillan, 1982; F: Triest et al., 1986; G: McMillan & Phillips, 1981; H: Gagnon et al., 1980). Species Enzymes Surveyed Amphibolis antarctica (A) A. griffithii (A) Ceratophyllum demersum (B) Cymodocea rotundata (C) C. serrulata (C) — d (A) H. un H. wrightii i Halophila decipiens (D) ovalis (A . stipulacea (D) . sp. (A) Heterozostera tasmanica (E) Najas marina (F) Posidonia australis (A) P. sinuosa (A) Phyllospadix scouleri (G) ADH*, Syringodium filiforme (E) S. isoetifolium (E) Thalassia hemprichii (E) T. testudinum (E) Thalassodendron ciliatum (E) Zostera capensis (E) Z. capricorni (E) Z. marina (H) APH, EST, G-6-PD, GOT, MDH, PER, PGI, PGM APH, EST, G-6-PD, GOT, MDH, PER, PGI, PGM APH, ADH, EST, GOT, G-6-PGD, GDH, IDH, LAP, MDH, PGI, PGM APH, EST, G-6-PD, GOT, MDH, PER, PGI, PGM APH, EST, G-6-PD, GOT, MDH, PER, PGI, PGM APH, EST, G-6-PD, GOT, MDH, PER, PGI, PGM APH, EST*, G-6-PD, GOT, MDH, PER, PGI, PGM APH, EST, G-6-PD, GOT, MDH, PER, PGI, PGM APH, GDH, GOT, MDH, PER, PGI, APH, GDH, GOT, MDH, PER, PGI, PGM APH, GDH, GOT, MDH, PER, PGI, PGM APH, GDH, GOT, MDH, PER, PGI, PGM APH, EST, GDH, GOT, MDH, PER, PGI, PGM APH, GDH, GOT, MDH, PER, PGI, PGM APH, EST, GDH, GOT, MDH, PER, PGI, PGM APH, G-6-PD, GOT, ME*, SkDH, XDH APH, EST, GDH, GOT, MDH, PER, PGI, PGM APH, EST, GDH, GOT, MDH, PER, PGI, PGM ADH*, APH, EST, G-6-PD, GDH, GOT, MDH, PER, PGI, PGM ADH, APH, EST, G-6-PD, GDH, GOT, MDH, PER, PGI, PGM ADH, APH, EST, G-6-PD, GDH, GOT, MDH, PER, PGI, PGM APH, G-6-PD, GOT, APH, G-6-PD, GOT, APH, G-6-PD, GOT, APH, G-6-PD, GOT, APH, EST, G-6-PD, GOT, MDH, PER, PGI, PGM ADH, APH, GOT, MDH, PER, PGI, PGM APH, ADH, GOT, MDH, PER, PGI, PGM ADH, CAT, DIA, G-6-PD, G-6-PGD, GDH, GTR, GOT*, HK, IDH, MDH, PGM MDH, PER, PGI, PGM MDH, PER, PGI, PGM MDH, PER, PGI, PGM MDH, PER, PGI, PGM MDH, PER, PGI, PGM PGI*, PGM, PMI, SDH, SOD Z. marina (E) APH, GOT, MDH*, Z. muelleri (E) Z. novazelandica (E) PER, PGI, PGM* APH, ADH, GOT, MDH, PER, PGI, PGM ADH, APH, GOT, MDH, PER, PGI, PGM tial evidence for assessing the degree of genetic variation in hydrophile populations; however, there is some empirical evidence that can be brought to bear on this issue. The use of enzyme gel electro- phoresis has allowed for the direct estimation of genetic variation in plant populations (Gottlieb, 1981; Brown, 1979; Hamrick et al., 1979). F tuitously, most electrophoretic studies on aquatic plants have been carried out with seagrasses (Wain et al., 1985), all of which are hydrophilous. Approximately 30 species representing 12 gen- era of seagrasses have been studied electrophoret- ically (Wain et al., 1985). These analyses consis- tently report genetic uniformity and provide little evidence of electrophoretically detectable genetic Or- variation in hydrophile populations (Table 6). Ac- cording to Wain et al. (1985: 43), “In most species [of seagrasses] there exists no intraspecific varia- ton in banding patterns, even across large geo- graphic distances." Furthermore, the level of het- erozygosity in seagrasses is apparently extremely low (Wain et al., 1985; McMillan, 1982). Of the 32 hydrophile species studied, enzyme variability has been reported only in Zostera marina, Hal- odule uninervis, Phyllospadix scouleri, and Na- Jas marina (Table 6). As Crawford (1983) dis- cussed, enzyme uniformity of this sort is more characteristic of self-pollinating plants than of out- crossers. Preliminary studies of Ceratophyllum have giv- Volume 75, Number 3 1988 Les 831 Hydrophilous Angiosperms en similar results. Populations of the rare-flowering C. demersum are relatively uniform both morpho- logically and genetically, and are probably clonal (Les, 1988b). Populations of the more sexual species C. echinatum are more variable morphologically and display patterns of intrapopulational variability that are quite representative of the species as a whole (Les, 1988a). An electrophoretic study of populations of both species is under way in hopes of providing further insight into the genetic struc- ture of populations of the predominantly sexual vs. asexual species of Ceratophyllum. Enzyme polymorphisms have been reported in electrophoretic studies of the annual species /Vajas marina (Triest et al., 1986), a possible indication of greater genetic diversity in this genus. Further electrophoretic surveys are necessary to determine the extent of intrapopulational genetic variation in other Najas species and annual hydrophiles in gen- eral. An informative study would be to compare populations of Najas and Zannichellia electro- phoretically. Although both genera are annual, annichellia has few species compared with /Va- jas, a possible outcome of the predominantly geito- nogamous/autogamous breeding system of Zan- nichellia. Despite the large number of electrophoretic studies carried out on hydrophiles, an adequate understanding of population structure in this group is far from being reached. A major difficulty has been that most studies have reported genetic data qualitatively rather than quantitatively (Table 6). Because data have not been presented in allelic form, it is not possible to compute appropriate quantitative measurements of population structure such as fixation indices, average heterozygosity, proportion of loci polymorphic, number of alleles/ locus, and gene diversities (Nei, 1987). Further- more, no study has measured outcrossing rates quantitatively in any hydrophilous species. Because models for estimating outcrossing rates should be evaluated in terms of the pollination system of a species (Shea, 1987), it may first be necessary to develop a specific model for hydrophilous plants. Electrophoretic approaches provide a powerful means of estimating population structure and out- crossing rates. The scarce amount of genetic vari- ation detected in hydrophiles thus far is an impetus for continuing investigations in this area. A satisfactory resolution to the question of pop- ulation structure in hydrophiles will not come until rigorous genetic analyses have been carried out. On the other hand, it is important to emphasize that the widely held assumption of extensive genetic variability in hydrophile populations is not sup- ported by available data. The high degree of clonal growth attributed to hydrophiles may result in ge- netically uniform populations. Even in instances where sex expression of dioecious hydrophiles may be under strict genetic control, clonal growth allows for the possibility of the establishment of extensive unisexual populations. Because of these factors, clonal growth may thwart outcrossing in many hy- drophiles. In support of this conclusion are prelim- inary electrophoretic data which indicate little de- tectable genetic variation in hydrophile populations. ssumptions that hydrophiles are outcrossing and produce genetically variable offspring should not be taken for granted. The inefficiency of hy- drophily, reduced sexuality, of autogamy and gei- tonogamy, and widespread clonal growth have the potential to restrict outcrossing greatly in this group. Because no studies have characterized genetically the actual level of inbreeding vs. outcrossing in any particular hydrophyte species, however, it is premature to argue too strongly either for or against the first component of the outcrossing hypothesis. COSTS VS. BENEFITS OF DICLINY IN HYDROPHILES The second component of the outcrossing hy- pothesis, that outcrossing in hydrophiles offsets evolutionary costs associated with transitions to dicliny, is even more difficult to assess. One diffi culty with pursuing this possibility is that it remains to be proven conclusively that hydrophiles are high- ly outcrossing. Plant breeding systems involve three general mechanisms: inbreeding, outcrossing, and apomixis (Briggs & Walters, 1984). Historically, outcrossing has been associated with enhanced genetic vari- ability and heterozygosity, whereas inbreeding and obligate apomixis are related to low genetic vari- ability (Faegri & van der Pijl, 1979; Briggs & Walters, 1984). With adapted gene complexes pre- served by well-developed asexual reproduction sys- tems, it is reasonable to assume that it would be advantageous for sexual reproduction in hydro- philes to provide a means of outcrossing that could respond facultatively to changing environmental conditions. Because dicliny may promote or en- force xenogamy, the prevalence of the unisexual condition in hydrophiles has been linked to out- crossing and widespread genetic variability. Pre- sumably, a rich gene pool would facilitate adap- tation and response to environmental changes. Therefore, enhanced genetic variation is inter- preted as a major possible benefit of dicliny. Grant (1975) pointed out the liability of ineffi- 832 Annals of th Missouri ES Garden cient sexual reproduction in diclinous plants, a cost related to reduced seed output in populations. If sex ratios : are és half of the flowers would not bear see itl ious or dioecious species. This outcome could have serious consequences in hydrophiles in which seed output is suppressed by other factors such as infrequent sexual reproduc- tion and extensive vegetative growth. Female-biased or equal sex ratios, however, may maximize seed production in dioecious species (Grant, 1975; Opler & Bawa, 1978). As discussed above, female-biased sex ratios have been noted in Elodea, Phyllo- spadix, Syringodium, and Thalassia, and ratios of unity have been found in Halodule and Tha- lassodendron. It is important to emphasize that female-biased sex ratios do not always alleviate problems of low seed production. A surplus of fe- males in a population (in favor of reduced males) may also result in low seed output because of in- adequate pollination (Grant, 1975). This limitation may have been responsible for a 28% loss in seed set observed in a population of the monoecious Zostera marina (Churchill & Riner, 1978). A com- promise between seed output and adequate polli- nation may be reflected in male-biased sex ratios reported in Ceratophyllum, Cymodocea, and Tha- lassia. Low seed output in hydrophiles may be specially implicated in their dispersal mode. Dispersal has been described as the physical basis of gene flow and in most plants occurs via transport of pollen or fruits (Grant, 1981). In aquatic plants, dispersal of vegetative propagules is also of great impor- tance. In hydrophiles, reduced seed output may result in a greater dependence on the transport of pollen and vegetative propagules as avenues of dispersal and gene flow. This relationship may be quite significant for freshwater hydrophiles where pollen gene flow is restricted to the single body o water in which the population occurs. In such in- stances, interpopulational gene flow may be entirely by transport of vegetative propagules. In contrast to perennial species, seed output of diclinous annual hydrophiles is extremely high, reaching levels of 100 seeds/m? in Lepilaena (Vollebergh & Cong- don, 1986). Evidently, adaptation to the annual habit has overcome seed limit costs of dicliny. In some diclinous annuals, such as monoecious Zan- nichellia, however, seed set is ensured by geiton- ogamy (Haynes & Holm-Nielsen, 1987), thereby negating any possible advantages of outcrossing associated with monoecy. These circumstantial data provide no satisfying resolution to the question of whether or not the costs of dicliny are outweighed by its selective advantages. Better insight into this matter may be gained by studying the relative fitness of closely related species in genera such as Zannichellia that possess monoecious, dioecious, and hermaphroditic sexual conditions. EXPLANATIONS FOR THE EVOLUTION OF DICLINY IN HYDROPHILES The third component of the outcrossing hy- pothesis is that dicliny evolved in hydrophiles prin- cipally as a mechanism of promoting outcrossing. This element of the hypothesis can be challenged outright because dicliny does not always guarantee outcrossing. Lewis (1979: 4) emphasized that “° separation of sexes . . . offers no protection against sib-mating, because pollen from a male plant is equally effective on a female whether it be a sib or non-sib," and that **. . . separation of sexes has a limited value as an outbreeding device in static plants." It is also important to realize that inbreed- ing is not restricted to hermaphroditic species but can also result in diclinous species as a result of geitonogamy, and possibly even from xenogamy due to vicinism (Grant, 1981). Furthermore, anom- alies such as geitonogamy in monoecious Zanni- chellia make it difficult to accept outcrossing as the compelling force in the evolution of dicliny. ome consideration has been given to other fac- tors influencing sexual conditions in plants such as relative resource allocation to maternal and pater- nal function, and constraints imposed by the dy- namics of the pollination system (Bawa & Beach, 1981). Unfortunately, these discussions have not considered water-pollinated plants. If outcrossing has been important in the evo- lution of hydrophiles, then why are so many species evolutionarily important in this group, then why has dicliny been conserved? This dilemma may be resolved by taking into account the early evolution of hydrophilous plants. Kimura & Ohta (1971) observed that the extent of genetic variability de- rived from sexual recombination could be attained by mutation alone given sufficient time, and they emphasized that gene recombinations are broken up by sexual recombination as quickly as they are made. They believed that the greatest advantage of sexual reproduction is to enhance the rate of evolution. Although the process of adaptation accompa- nying the transition to hydrophily may have oc- curred slowly by mutation alone, the combination of sexual recombination and an outcrossing breed- Volume 75, Number 3 1988 Les 833 Hydrophilous Angiosperms ing system would have facilitated the process by generating pools of genetic variability more rapidly. In this way, the ancestors of modern hydrophiles may have benefited greatly from sexuality and the outcrossing potential conferred by dicliny, espe- cially if their progenitors lacked other mechanisms to prevent autogamy. In species where hydrophily was derived mainly from lines of self-compatible anemophilous plants, dicliny may have preceded hydrophily. Once species had become well adapted to hy- drophily and a submersed existence, greater fitness may have resulted from maintaining genetically uniform rather than highly varias püspring. Be- ly constant, Cause alketativo pets + would maintain adaptive gene complexes and prevent their breakup by sexual reproduction. The phasing out of sexual reproduc- tion may have been accompanied by greater re- source allocation to vegetative growth. A shift to asexual reproduction would likely result in slow evolutionary rates and low species diversity in the group, precisely the pattern that we see in the fossil record of several modern species. By this scenario, dicliny in extant hydrophiles is essentially a relic- tual condition of uncertain consequence to their present reproductive biology. There are additional appealing aspects of this interpretation. Because dispersal in rare-flowering perennial hydrophiles is likely to be predominantly by vegetative propa- gules, many populations would be expected to be clonal and genetically uniform. The available data show that these patterns commonly occur in water- pollinated plants. The maintenance of sexuality in extant annual hydrophiles may account for their higher species diversity, the result of accelerated evolutionary rates associated with sexual recom- bination and outcrossing. SUMMARY Because of the prevalence of dicliny in hydro- philes, it is widely believed that water pollination in aquatic plants is linked to an outcrossing breed- ing system. Actually, there is little evidence that three necessary conditions for outcrossing (sex- uality, xenogamy, genetically variable populations) are met by many extant species. Interrelationships of inefficient pollen transfer, reduced sexuality, widespread clonal growth, and diminished seed pro- duction in hydrophiles may reflect a major adaptive shift towards asexuality as a means of preserving adaptive gene complexes in stable aquatic envi- ronments. This hypothesis explains the apparent slow evolutionary rates and low species diversity of some hydrophilous species noted in the fossil record, as well as the greater species diversity observed in sexual, annual, potentially outcrossing species. If this interpretation is correct, then dicliny persists in modern perennial hydrophiles as a re- lictual condition associated with their early evo- E = Although the course of evolution in water-pol- linated plants does not appear to differ fundamen- tally from that of other plants, the complex inter- actions of hydrophily and other aspects of their reproductive biology are likely to have profoundly influenced the patterns of their present diversity. As we improve our methods of evaluating the breeding systems and population structure of hy- drophilous species, so will our understanding of evolution in this unusual group of plants be ex- panded. LITERATURE CITED ACKERMAN, J. D. 1983. Current flow around Zostera pla nts ubmar ir ic Hole alae Biol. Bull. Mar. Biol. Lab. Woods 5: 504. . 1986. Mechanistic oe for pollination in the marine angiosperm Zostera marina. Aquatic Bot. 24: 343-353 Acai, M., S. BEER & Y. WAISEL. 1986. The mor- phology and physiology of bis in Najas marina L. in Israel. tic Bot. 1- ARBER, À Vater Pants ^ Study of Aquatic Angiosperms. Reprint, J. Cramer, Lehre. 197 od AQUAM Plants. of Australia. Mel- ASTON, H. I. bourne Univ Bawa, K. S 980. km of dioecy in flowering plants. Ann. Rev. Ecol. Syst. 11: 15-39. — — € J. H. Beach. 1981. Evolution of sexual systems in i plants. Ann. Missouri Bot. Gard. 68: 254-274 Bniccs, D. & S. M. WALTERS. 1984. Plant Variation and as 2nd edition. Cambridge Univ. Press, Nes or Brock, M. A. 1982, Biology of the salinity tolerant genus Ruppia L. in saline lakes in south Austra uppi II. et a ia ecology and reproductive biology. Aquatic Bot. 13: 249-268. Brown, A. H. 1979. Enzyme polymorphism in plant populations. Theor. Populat. Biol. 15: 1-42. CAMBRIDGE, M. L., S. A. Carstairs & J. Kuo. 1983. An unusual method of vegetative propagation in Aus- tralian Zosteraceae. Aquatic Bot. 15: 201-203. CaTLinc, P. M. & W. WOJTAS. The waterweeds (Elodea and M np Hydrocharitaceae) in Canada. Canad. J. Bot. 64: 1525-1541 CAYE, C. & A. MEINESZ. 1985. Observations on the velopment, ing and seeding of Cy- modoctt bo s= (Ucria) a on the Mediter- ranean coasts of France. Aquatic Bot. 22: 277-289. & 86. Experimental peni of seed germination in the seagrass Cymodocea nodosa Aquatic Bot. 26: 79-87. CHURCHILL, A. C. & M. I. RINER. 1978. Anthesis and 834 Annals of the Missouri Botanical Garden seed production in Zostera marina L. from Great n th Bay, New York, U.S.A. Aquatic Bot. 4: 83- 1982. Pollination mechanisms in the oens, S. S. UE r: D. K. Hydrocharitaceae. Pp. 1-15 ¿n J. J. Sym Hooper & P. Compére (editors), Studies on Aquatic Vascular Plants. Roy. Bot. Soc. Be lgium, Brussels. K. Urmi-KóniG A revision of the genus Elodea (Eisdroebaritacesd]. Aquatic Bot. 21: -156. , B. J. Gut, E. M. Rix, J. SCHNELLER & M. SEITZ. e Water Plants of the World. W. Junk, The Cox, j oe 1983. Search theory, random motion, and the convergent evolution of pollen and spore mor- phology i in aquatic plants. Amer. Naniwala 121: 9- Gave D. J. 1983. Phylogenetic and systematic inferences from ar studies. Pp. 257- 2 ) . J. Orton ao rt A 66. Marine Botany. An Introduction. e Rinehart & Winston, New Yor De Coc W Flowering, pollination Rey fruiting i in Zostera marina L. Aquatic Bot. 9: Durako, M. J. & M. D. MorrLER. 1985a. Observations on the reproductive ecology of Thalassia testudinum (Hydrocharitaceae) II. Leaf width as a secondary sex character. Aquatic Bot. 21: 265-275. ho . 1985b. Observations on the re- productive ecology of Thalassia testudinum (Hy- oda nao Hil. Spatial and soon variations 1. Aquat ic Bot. 22: 265- 276. 987. Factors affecting the re- productive ecology of Thalassia testudinum (Hy- drocharitaceae). Aquatic Bot. 27: 79-95. Epwarps, P. 1976. Illustrated Guide to the Seaweeds the Seagrasses in ii Vicinity of dd Aransas, Texas. Univ. Texas , FaEGRI, K. & L. VAN DE The Principles of Pollination Ecology, 3rd edition. Pergamon Press, New York. FREEMAN, D. C., K. T. HARPER & E. L. CHARNOV. 1980. ain change i in plants: old and new observations and w hypotheses. Oecologia 47: 222-232. ic NON, P. S. . VADAS, D. B. Burpick & B. May. 1980. Genetic identity of annual and Maletas forms Zostera marina L. Aquatic Bot. 8: 62. GAMERRO, J. C. Observaciones e Ja biología floral y morfología de la Pota SER Eia cirrhosa (Petag.) Grande. Darwiniana 14: 575-608. GOTTLEI 981. Electrophoretic CUR and Phytochem. 7: 1-45. plant populations. Progr. Grant, V. ] Genetics of Flowering Pianta Colum- ia Univ. Press, New York. 1981. Plant qoem 2nd edition. Columbia Univ. Press, New Grey, W. F. & M. D. Marin the seagrass Thalassia testudinum (Hydrocharita- ceae) in the Tampa Bay, Florida area. Aquatic Bot. 3 9: 2 HAMRICK, J. E Y. B. Linnart & J. B. Mitton. 1979. Relationships between life-history characteristics and 1978. Flowering of electrophoretically detectable genetic variation in plants. Ann. Rev. Ecol. Syst. 10: 173-200. iri J.L. 1977. Population Biology of Plants. Ac- mic Press, New HARRISON, P. G. Reproductive strategies in in- dam . R. . The Nejdassss in the South- eastern United States. J. Arnold Arbor. 58: 161- 17 & L. B. Horw-NiELsEN. 1987. The Zanni- chelliaceae in the Southeastern United States. J. Ar- d -2 o BERI d D. A. waiiana: description and notes cola y. Aquatic Bot. bas - Hime neo G. A Treatise on Limnology, olume 3. A Botany. John Wiley & Sons, New York. Jacobs, R. P. W. M. & B. Dicks. 1985. Seagrasses in the Zeit Bay area and at Ras i: (Egyptian Red Sea ep Ti Bot. 23: & E. S. Pierson. 19 ^ pui of repro- ductive Seed of eelgrass, ee marina L., at Roscoff (France). "ides Bot. 10: 45- y Floral seid in the m giosperms lp ai serrulata and Thalassoder- . So 86. Staminate flowers of Halophila i on its flowering v PATRIQU HN. 1978. An annual form of eelgrass in Nova Scotia. Aquatic Bot. 5: 163-17 Kimura, M. ^ T. Outa. 1971. Population Genetics. Princeton Univ. Theoretical Aspects of ress, New Contributions to the Biology and m in the Eastern United States. M bad Eastern Michigan Univ., Ypsi- lanti, Michiga = 1985. ia Pp. 97- in G. Leach & P. Osborne (editors), The ues Plants of Papua New Guinea. Univ. Press, Papua New Guinea Systematics and Evolution of Cera- tophyllum enn dear A as a ii Ph.D. Thesis. Ohio State Univ., Columbus, Ohio. The evolution of hee morphology in Ceratophyllum (Ceratophyllaceae), I. Fruit-spine variation and relationships of C. demersum, C. sub- ersum, and C. apiculatum. Syst. Bot. 11: 549- 1988a. The evolution of achene morphology in Ceratophyllum (Ceratophyllaceae), II. Fruit vari- ation and M of the “spiny-margined” group. Syst. Bot. 13: 73-86. 1988 . The origin and affinities of the Cer- atophyllaceae. Taxon 37: 326-345. Lewis, D. 1979. Sexual Incompatibility in Plants. Ed- Arnold Press, London 1980. Benefits and erii of sexual reproduction. Evol. Biol. 13: 69-11 & K. S. Bawa. 19 Modification of the gender of seed oo in varying conditions. Evol. Biol. 17: 255-338. Volume 75, Number 3 Les 835 1988 Hydrophilous Angiosperms McATEE, W. L. 1939. Wildfowl Food Plants. Their apnd. pp of eelgrass (Zostera marina Value, Propagation, and Management. Collegiate L.). Aqua -20. Press, Inc., , lowa Zi ae M Capacity of Plants. McCann, C. 1945. "Notes on the genus Ruppia. J. Bombay Nat. Hist. Soc. 45: 396-402. McMILLAN, c. 1976. Experimental studies on flowering and reproduction in seagrasses. Aquatic Bot. 2: 87- . 1979. Flowering under controlled conditions by Cymodocea iiis from ps aai Islands, A Aquatic Bot. 6: 397-4 1980. ring under c Bam conditions by Cymodocea pec Ay ec y mn Cb ue e isoetifolium, Zostera capensis and Thalassia hemprichii from Kenya. Aquatic P 8: 323-33 š 1981. Morphological variation and isozymes under d conditions in Cymodocea serrulata. Aquatic Bot. 10: 365-370. Isozymes in seagrasses. Aquatic Bot. 14: 231-24 ——— & R. C. Pumps. 1981. Morphological vari- ation and isozymes of North American Phyllospadix Se ee Canad. J. Bot. 59: 1494-1500. & S. C. WILLIAMS. 1980. stematic impli- cations of 1 ~~ in "aai diu section Halophila. Aquatic Bot. 9: 21- , L. Ex R & O. Zapata. 1981. Isozymes, G compounds and experimental cultures = —- ian seagrasses in Halophila, Hal- dule, Zostera, Amphibolis and Paloma. Austral. 260. .B E MUHLBERG, H. 1982. The Complete Guide to Water Plants. EP Publishing, Ltd. NEI, M. 1987. Molecular Evolutionary Genetics. Co- lumbia Univ. Press, New York. MN A. A. 66. Hydrocharitaceae. Pp. 101- 102 in Flora of Southern Africa, Volume I. Dept. of ide Technical Services, Pretoria, South Africa. Oper, P. A. 8 K. S. Bawa. 1978. Sex fe in tropical forest Tp aru 32: 812-82 Perrrrr, J. . C. JERMY. is „Polen in hy- a ab bio Micron 5: 377-4 , S. DuckeR & B. Knox. 1981. pollination. Sci. Amer. 244(3): 135-143. PuiLLIPs, R. C., W. S. GRANT & C. P. McRoy. 3 M 1983. SALISBURY, E. J 194 G. Bell & Sons, London. ScuLTHORPE, C. D. 1967. The Biology of Aquatic Vas- cular Plants. Edward Arnold Publishers, London. SEHGAL, A. & H. Y. MOHAN T Co omparative developmental morphology lati f Cer- atophyllum L. des tephyllacces) and their taxon- . J. Linn. Soc. Bot. 82: 343-356 1987. Effects of population structure and SILVERTOWN, J. W. Introduction to Plant Pop- ulation Ecology. Longman , New Yor TALAVERA, S., P. Gancía Mu RILLO & H. Smit. 1986. Sobre el genero Zannichellia L. (Zannichelliaceae). Lagascalia 14: 7 eds its and seeds of the Brandon lignite. VI. P ihe bim (Lythraceae). J. Arnold Ar- bor. 62: 487-516. TRIEST, L., J. VAN GEYT & V. Ranson. 1986. Isozyme ponen in several populations of Najas ma- c Bot. 24: 373-384. VAN Wik. 1983. chellia in Turkey. A VAN VIERSSEN, W. & R. J. van Wik. 1982. On the identity and autecology of Zannichellia peltata Ber- tol. in Western Europe. Aquatic Bot. 13: 367-383 & J. 1 "t etonaceae. Aquatic Bot, 14: 339-34 pilaena cylindrocarpa in ephem Westernport Bay, Victoria. Aquatic Bot. 26: Wain, R. P., W. T. HALLER & D. F. Martin. 1985. Med in studies " nd plants. J. Aquatic Pl. Man . 23: 42+ ama Dp. R. 1969. Wind pollination in the a giosperms: evolutionary and environmental consid- erations. Evolution 23: 28-35. YEO, R. X 1966. Yields of propagules of certain aquatic plants. I. Weeds 14: 110-113. EVOLUTION OF UNDERWATER OUTCROSSING FROM AERIAL POLLINATION SYSTEMS: A HYPOTHESIS! C. Thomas Philbrick? ABSTRACT It is evident that underwater E (hypohydrophily) arose from aerial pollination systems. Howeve no mechanism to explain bn tran n has dnd er, n proposed. Herein l suggest a system involving bubble der d pollinatio similar to hydroautogamy in Potamogeton, systems. Such an a o ould ar go opportunity for the gradual evolution of « characters that are needed for underwater outcrossing without sacrificing seed production during the transition ere is general agreement that water pollina- tion (hydrophily) is derived from aerial pollination, for the aquatic habit itself is derived in angiosperms (Arber, 1920; Daumann, 1963; Sculthorpe, 1967). Although we have some understanding of the mech- anisms of pollen transfer in hydrophiles, there has been little published to explain the origin and sub- sequent radiation of hydrophilous systems. Hydro- phily has been most widely investigated in marine angiosperms (e.g., Ducker & Knox, 1976; Ducker et al., 1978; Cock, 1980; Pettitt, 1984; Pettitt et al., 1980, 1981). Our understanding of the floral biology of freshwater hydrophiles is narrower than in marine groups, even though the unusual mech- anisms of pollen transfer in some, e. he Hy- drocharitaceae, have received considerable atten- tion (e.g., Cook, 1982; McConchie, 1982; Wylie, 1917). Arber (1920) and Sculthorpe (1967) proposed that marine angiosperms arose from freshwater stock (Hartog, 1970, presents a contrasting view), hence hydrophily likely arose in freshwater sys- tems. Thus, study of hydrophily in freshwater groups could be central to understanding the origin of and selective pressures behind hydrophily, matters that remain to be adequately addressed. Herein I pro- pose a mechanism by which underwater outcrossing evolved from aerial pollination systems. This hy- pothesis could serve to reorient investigations of the evolution of hydrophily and its role in diver- sification of aquatic angiosperms. BACKGROUND Over 90% of aquatic angiosperms bear aerial flowers and have the same manner of pollination as their terrestrial ancestors (Arber, ; Scul- thorpe, 1967). Far fewer exhibit a system whereby water is the vector for pollen transfer, i.e., hydroph- ily. Currently two classes of hydrophily are rec- ognized: 1) epihydrophily; pollination at the water surface, i.e., in two dimensions, and 2) hypohy- drophily; pollination below the water surface, i.e., in three dimensions. However, it is evident that epihydrophily embraces at least two rather dissim- ilar subtypes. In one, the flowers undergo anthesis above the water surface and the reproductive struc- tures (e.g., stigmata, pollen) remain dry: dry-epihy- drophily. Vallisneria represents the best-known example of dry-epihydrophily (Cook, 1982, and references therein). In contrast, many seagrasses exhibit wet-epihydrophily, where the reproductive structures are wet at anthesis but in close associ- ation with the water surface, e.g., the pollen floats just below the water surface but not upon it. Hy- pohydrophily is more similar to wet- than dry- ' I wish to thank Gregory J. Anderson, Paula K. Busse, Kent E. Holsinger, and Donald H. Les for their helpful omments on the manuscript. I thank haet Barlow formulated during discussions with Gre J. A necessarily share the views presented. See in us ana e - Many of the ideas on this subject were n and Richardson, although they do not ya National Science Foundation grant (BSR 82- 07125) to Gregory J. Anderson and a National Science Foundation Dissertation grant (BSR 87-01285) to T.P. ° Department of Ecology and Evolutionary Biology, U-43, The University of Connecticut, Storrs, Connecticut 06268, U.S.A ANN. Missouni Bor. Garb. 75: 836-841. 1988. Volume 75, Number 3 1988 Philbrick 837 Underwater Outcrossing epihydrophily. In fact, wet-epihydrophily may be but an evolutionary “refinement” of the more gen- eral hypohydrophily. It is clear that wet-epihydrophily and hypohy- drophily require greater modification and adapta- reflect specialization to facilitate pollen release and capture in water, a medium that in addition to being wet is significantly more viscous than air. Features that are associated with wet-epihydrophily and hypohydrophily are summarized in Table 1. Hypohydrophily occurs in relatively few angio- sperm families (Arber, 1920; Cox, 1983; Dau- mann, 1963; McConchie, 1982; Sculthorpe, 1967); most are monocotyledons. The Ceratophyllaceae are the only dicotyledonous exception. taxonomic distribution it is evident that hypohy- drophily is polyphyletic, with convergence toward a similar overall morphology The Najadales (sensu Cronquist, 1981) exhibit a complete range of pollination systems (aerial, epihydrophily, hypohydrophily) and thus are an ideal group in which to search for clues to the evolution of hypohydrophily. This order comprises 10 families and nearly 200 species (Cronquist, 1981). Phylogenetic relationships in this and re- lated orders are largely unclear, but the families From its Dahlgren & Clifford, 1982; Tomlinson, Eight of the 10 families are T up of freshwater or marine aquatic taxa. The species in the re- maining two families grow as emergents in marshy habitats. Aerial pollination characterizes most species of Potamogetonaceae, Aponogetonaceae, Scheuchzeriaceae, and Juncaginaceae. Dry-epi- hydrophily is found in Ruppia (Ruppiaceae) (Ver- hoeven, 1979) and Lepilaena (Zannichelliaceae) (Vierssen et al., 1982). Wet-epihydrophily occurs in a number of seagrasses. Hypohydrophily is found in the largely freshwater aes (Najsdacine) (Seg thorpe, 1967) and Z (Vierssen et al., 1982) in ddditiog to some sea- grasses. Given our limited understanding of the distinctions between wet-epihydrophily and hypo- hydrophily, no attempt will be made to distinguish them. For brevity, the term hypohydrophily will be used in the following discussions to include both wet-epihydrophily and hypohydrophily. It is generally believed that floral biology has played an important part in the evolution of an- giosperms (Baker, 1963; Crepet, 1983, 1984; Grant, 1949, 1963; Grant & Grant, 1965; Steb- bins, 1970; and others). Therefore, it is not un- reasonable to propose that diversification of hy- TABLE 1. Features X associated with hypohy- drophily and wet-epihydrophily. Feature References! General Similarities with anemophily : 6, 14 Specialized aad stigma rec- 111,12 ognition Reduction in joue size 9 Reduced perianth 9, 11, 15 Single ovule/ JT (usually) 3, 14, 15 3,5,9, 11, 14, 15 Unisexual flow Lack of scent Es nectar; col- less Reduction in stamen number / 15 ower Reduced anther wall 6, 14, 15 Pollen High pollen/ovule ratio 3, 6, 13, 14 Reduced exine 3, 4, 7, 10, 11, 14, 15, 16 Elongate (by various means) 1, 2, 5, 8, 9, 11, , 15 Precocious pollen tube produc- 7, 8, 14, 15 tion Wettable 3, 4, 10 Stigma/Style Large, rigid, and simple (linear) 3, 5, 14 Wettable 3, 10, 11 l — Arber, 1920. 2— Cox, 1983. 3— Daumann, 1963. Jaeger, 19 : 1982. 9— Percival 1965. 10— Pettitt, 1984. 11 — Pet- titt et al., 1981. 12— Pettitt et al., 1980. 13— oy Tam & Anderson, 1987. 14— Proctor & Yeo, 1972 cer " 967. 16— Wodehouse, 1935. pohydrophilous systems has played an integral role in speciation, e.g., /Vajas (ca. 40 species) (Haynes, 1977). Mechanisms that ensure pollination during evo- lutionary transitions in floral structure have been proposed as being primary in the evolution of pol- lination systems (Baker, 1963; Stebbins, 1970, 1974). Given the almost universal occurrence of aerial flowers among angiosperms, adaptations for an aerial floral biology are undoubtedly well fixed in the angiosperm genome. An impediment to the formulation of hypotheses regarding the evolution of hypohydrophily has been a lack of obvious in- termediate pollination systems. Hypohydrophily re- quires the abandonment or modification of a suite of characters that are intimately tied to the dry, aerial flowering condition. The mechanical and bio- chemical ramifications entailed in adapting to un- 838 Annals of the Missouri Botanical Garden FIGURE Scanning electron micrograph of a flower of Potamogeton pusillus that illustrates the overall morphology cla i pue and the orientation of the stigmata (S) , anthers (A), and perianth segments (PS). Scale ba derwater release, transport, and capture of pollen raise significant adaptive obstacles. The fact that flowers of hypohydrophiles are wet when anthesis occurs makes them unique. The rapid decrease in viability of pollen when wetted (Daumann, 1963; Jones, 1967) illustrates the sen- sitivity of aerial flowers to the influence of water. Means by which selection for wettability could act upon an aerial flower and yield a ““half-wet” in- termediate while retaining seed production and sex- uality are not evident. Strong selective pressure to retain aerial flowers, or at least perhaps to avoid flowering under water, is evidenced by the retention of aerial flowers in a clear majority of aquatic angiosperms. This phe- nomenon is particularly well illustrated in species that exhibit pronounced modification of submerged vegetative structure yet exhibit aerial flower pro- duction, e.g., Utricularia and Myriophyllum (Ar- ber, 1920; Sculthorpe, 1967). Nonetheless, some aquatics have essentially abandoned aerial flowers. These can be organized into two general groups: 1) those that produce submerged, aerial-type flow- ers that self-pollinate, and 2) those that have ap- parently dispensed with flower production or with sexual reproduction altogether, and reproduce by a variety of asexual means. Of course many species combine asexual and sexual reproduction. The abundance of species in these two classes under- scores the evolutionary importance of alternatives to aerial flower production. The occurrence of hy- pohydrophilous groups in the extant flora reveals an additional evolutionary pathway, where modi- fication has taken place to allow the employment of the aquatic medium as a pollen vector. Given that hypohydrophily evolved from aerial- flowered systems, there can be little argument that its evolution would require the submergence of aerial flowers. When typical aerial flowers become submerged, closing of the perianth usually serves to trap and maintain a small bubble of air around the reproductive structures. Therefore, pollination within such flowers is equivalent to selfing in a functionally dry flower (Arber, 1920; Hutchinson, 1975; Sculthorpe, 1967). Thus, little selection for wettability would occur, for during anthesis there is no contact between the reproductive structures and the water. It appears that the evolutionary submergence of flowers per se would not neces- sarily provide the circumstances under which se- lection leading to hypohydrophily would proceed. The flowers of a progenitor of hypohydrophily would have to have been somehow predisposed such that selection could act on flowers that are open un- derwater, while the ability to produce seed is re- tained. That is, an intermediate pollination system "bridging" of Baker, 1961, 1963) was likely im- portant during the transition. HYDROAUTOGAMY IN POTAMOGETON I propose that a reproductive system similar to hydroautogamy in Potamogeton occupied a key intermediate stage between aerial systems and hy- pohydrophily. Numerous species of Potamogeton exhibit underwater flowering and seed production, Volume 75, Number 3 1988 Philbrick Underwater Outcrossing 839 FIGURE 2. su a derived condition in the genus (Philbrick & An- derson, 1987, and references therein). Flowers of Potamogeton open while submerged; thus pollen and stigmata are exposed to water during anthesis. Self-pollination via hydroautogamy (the movement of bubble-borne pollen from anther to stigma within an open flower) rather than hypohydrophily seems to be the principal mechanism of pollination in submerged flowers (Philbrick & Anderson, 1987). During anthesis, bubbles are produced as gas is released from the dehiscing anthers (Figs. 1, 2). Preliminary study with T. Taigen suggests the gas is carbon dioxide. Pollen travels from the anther onto the outer surface of the bubble (Fig. 2). The bubble increases in size until it extends from the anther to the stigma. Pollen is then deposited onto the stigma from the bubble surface (Fig. 2) and self-pollination and subsequent fertilization results. The bubble continues to enlarge until it breaks free from the flower and rises to the surface. Additional bubbles are formed as each anther opens. An im- portant consequence of this system of self-polli- nation is that the pollen and stigmata are wet during anthesis. In addition, close inspection reveals that individual pollen grains often drift off the bubble surfaces. In systems where transition is occurring within a pollination-vector category, e.g., one insect species to another, relatively minor changes in floral struc- ture that accommodate a new vector would be more probable than when the shift is from one broad category of pollen vector to another. The hypoth- esis of fortunate accidents” formulated by Baker (1963), which suggests a chance preadaptation to an available vector, is more likely to hold when the gross structure of the flower remains un- changed. In contrast, floral modification that allows a change from one type of pollination vector to another, e.g., wind (dry) to water (wet), is less likely to be sudden because of the numerous modifications Idealized diagrams of a portion of a Potamogeton pusillus flower that illustrate the progressive enlargement of a bubble (B) during anthesis and the deposition of pollen (P) onto the stigma from the bubble rface. involved (Table 1). A hydroautogamous system would provide an opportunity for the operation of selective pressures leading to the gradual accu- mulation of hypohydrophilous features while main- taining seed production. Lacunae systems such as those in the plant body of Potamogeton may have played an important role in the acquisition of hypohydrophily. Lacunae are common in submerged hydrophytes and allow movement of gases within the plant body (Arber, 1920; Hutchinson, 1975; Sculthorpe, 1967). La- cunae are also evident in the inflorescence and floral structures of Potamogeton (U. Posluszny, pers. comm.). Gases that are transported via the lacunae build up within the anther and seem to play a role in its dehiscence. It is reasonable to propose that early in the acquisition of a hydroautogamous-like system the contents of the anther were dry at dehiscence. In such a case, the pollen would be shed dry and presumably remain on the inside surface of the bubble. This is little different from the behavior of airborne pollen when it becomes trapped upon the surface film of water. The water/atmosphere in- terface is similar in both instances. Similarly, the stigmata remain within the bubble during pollina- tion. However, the surface of the bubble is impor- tant because it provides an opportunity for selection to operate. Although the pollen is virtually dry when inside the bubble, humidity levels there would be high, favoring pollen that could withstand the effects of increased humidity, i.e., increased wet- ting. Similarly, stigmata that could withstand wet- ting would be favored. Thus, the gradual selection of two major characters that distinguish hypohy- drophily from aerial systems would result: wettable pollen and wettable stigmata. The bubble system provides a setting wherein major obstacles in the evolution of hypohydrophily could be surmounted. Stebbins (1970, 1974) discussed the importance 840 Annals of the Missouri Botanical Garden of character syndromes, i.e., correlations between characters, in the evolution of pollination systems. Selective modification of one feature is often man- ifested in a change in related features. Thus, se- lection for wetting might have affected pollen and stigmata simultaneously. Selection for wettability would result in the pollen no longer being limited to the inside of the bubble. The pollen could then move to the outside surface of the bubble, as in the extant species. Pollen on the outside of the bubble would be subject to continued selection, as well as “loss” by drifting away; the stage would then be set for the initiation of outcrossing. Unisexual flowers characterize all of the docu- mented cases of hypohydrophily. Arguments for the evolution of unisexual flowers in terrestrial plants (e.g., Anderson & Stebbins, 1984; Bawa, 1984; Charnov, 1982; Lloyd, 1982) would be equally applicable to hypohydrophilous aquatics. I propose that unisexuality was acquired not before, but after the initial submergence of a bisexual flower and initial selection toward hypohydrophily was on a bisexual flower. Evolution of hypohydrophilous characters from aerial, unisexual flowers seems less likely due to the requirement for simultaneous and somewhat independent acquisition of hypohydroph- ilous features in both staminate and pistillate flow- ers (Table 1). Further, seed production via sexual means would be more difficult to maintain. High pollen production (pollen/ovule ratios) characterizes abiotic pollination systems (e.g., Cru- den, 1977; Philbrick & Anderson, 1987; White- head, 1969). Lower pollen/ovule ratios are a gen- eral feature of predominantly self-pollination sys- tems (Cruden, 1977). Aerial-flowered species of Potamogeton exhibit high pollen production and are believed to be anemophilous (Philbrick & An- derson, 1987). Hydroautogamous potamogetons exhibit lower pollen production than aerial taxa but not as low as would be expected for selfers (Phil- brick & Anderson, 1987). The maintenance of relatively high pollen production in largely autog- amous lineages may be due either to genetic con- straints, the relatively recent acquisition of selfing, or the fact that hydroautogamy itself is a somewhat stochastic system. The propensity for continued high pollen production in hydroautogamous taxa may have been instrumental as a preadaptation in subsequent selection for hypohydrophily, a sto- chastic system that relies, as does anemophily, on high pollen/ovule ratios. CONCLUSION There is general agreement that hydrophily evolved from aerial pollination systems. In contrast with some previous ideas, I consider hypohydroph- ily to be significantly different from dry-epihydro- phily, where pollination is effectively aerial. Hypohydrophilous taxa exhibit many more spe- cializations to the aquatic medium in reproductive features. A major obstacle to the evolution of hypohy- drophily is the accumulation of characters that allow for seed set with wetted pollen and stigmata. I suggest that such characters were acquired grad- ually, and offer a hypothesis where a system that utilizes bubble pollination, similar to hydroautog- amy in Potamogeton, could serve as an interme- diate between aerial pollination systems and hy- pohydrophily. Hydroautogamy could provide an opportunity for the gradual selection of hypohy- drophilous characters without sacrificing seed pro- duction. This hypothesis is the first to address explicitly the issue of the evolution of hypohydrophily from aerial systems. Corollaries of this hypothesis are testable via studies of hydroautogamy in Pota- mogeton (e.g., viability of pollen that is released from the bubble, ability of submerged stigmata to capture water-borne pollen). Further investigation of the mechanisms of hypohydrophily and phylo- genetic relationships among hydrophilous and non- hydrophilous taxa will provide information that can be used to evaluate this hypothesis. LITERATURE CITED ANDERSON, G. J. & G. L. STEBBINS. 1984. Dioecy versus ciao d FIDCORQpABRIEN a test. Amer. Nat- uralist 124: 428. ARBER, A. 1920. "Water Plants. Cambridge Univ. Press, Cambridge BAKER, H. G. 1961. The adaptation of flowering i to nocturnal and crepuscular pollinators. Quart. Rev Biol. 36: 64-73. 1963. Evolutionary mechanisms in pollination biology. Science 139: 877-883. Bawa, K. S. 84. The evolution of monty — Concha ing remarks. Ann. Missouri Bot. Gard. 71: 294- CHARNOV, E. L. 1982. The Theory of Sex Allocation. Princeton Univ. Press, Princeton, New Jerse Cock, A. W. A. M Flowering, pollination and fruiting in Zostera marina L. Aquatic Bot. 9: 201- 1982. aer mechanisms in the . Symoens, S. S. 220. Cook, C. D. K. dede genit du -15 i ooper ëre (editor >) Sh; dies on Aquatic . Plants. Royal Botanical Society of Belgium, ussels. Cox, T A. 1983. Search theory, random motion, and the conv 'ergent evolution of pollen and spore mor- pho ology in aquatic plants. Amer. Naturalist 121: 9- és W. evolution of the angiosperms. Pp. 31- 1983. The role of insect pollination in the 0 in L. Real Volume 75, Number 3 1988 Philbrick 841 Underwater Outcrossing (editor) Pollination Biology. Academic Press, New ork. 84. deba (constant) insect pollination mechanisms: patter evolution and implications vis-à-vis angio: E nerd ity. Ann. Missouri Bot. Gard. 71: 607-630. Tu A. 81. An Integrated System of Classi- n Flowering Plants. Columbia Univ. Press, CRUDEN, R. w. 1977. Pollen-ovule ratios: a conser- 1: P $ a E. 1 + Evolution 31: 32-46. DAHLGREN, R. M. T. & H. T. CLirFORD. 1982. The Monocotyledons: A Comparative Study. Academic Press, New York. DaAuMANN, E. 1963. Zur Frage nach dem Ursprung der Hydrogamie. Zugleich ein Beitrag zur Blüten- okólogie von Potamogeton. Preslia 35: 23-30. 976. Ducker, S. D. . KNox. Submarine pol- lination. Nature 263: 705- , J. M. Perritr & R. B. Knox. 1978. Biology of a ae pollen ss agua S. and sub- marine pollination in ibolis antarctica and Thalassodendron ciliatum (Grodan ey Aus- AN DER | Pun. 979. The Principles of Pollination pore Pergamon Press, New York. GRANT, V. 194 ollination systems as isolating mech- anisms in angiosperms. Evolution 3: 82-97. 1963. The Origin of Adaptations. Columbia Univ. Press, New York. K. RANT. 1965. Flower Pollination in the Phlox Family. Columbia Univ. Press, New York. Hanroc, C. DEN. . The seagrasses of the world. erhandl. Kon. Ned. Akad. Wetensch. Nat. 59: 5- 275. North-Holland, Amsterdam. HavNES, R. R. The Najadaceae in the south- eastern United States. J. Arnold Arbor. 58: 161- 170. HUTCHINSON, G. E. . A Treatise on Limnology, Volume III. Limnological Botany. John Wiley & k. . The Wonderful Life of Flowers. E. . Dutton & Co., New York. Jones, C. E. Some evolutionary aspects of a water stress on flowering in the .tropics. Turrialba 17: 188-190. LLoyp, D. G. 1982. Selection of combined versus se arate sexes in seed plants. Amer. Naturalist 120: 571-585 McCocniE, C. A. 1982. The Diversity of edo pd Po lina tion in Monocotyledons. Pollination '82. Pro the Symposium Held at the School of ne. Sp ores and pollen grains of water d and their dispersal. Rev. Palaeobot. Pa- lyn PERCIVAL, M. R 1965. Floral Biology. Pergamon Press, Ha J. M. 1984. Aspects of flowering and polli- nation in marine angiosperms. Oceanogr. Mar. Biol. 22: 315-342. , S. DUCKER & R. B. Knox. Sai pollination. Sci. Amer. 244: 134-14 , C. A. McConcuie, S. C. Du nA R. B. Knox. 1980. Unique adaptations for submarine pollination seagrasses. Nature 286: 487-489. Pana C. T. & G. J. ANDERSON. 1987. Implications of pollen/ovule ratios and pollen size for the repro ductive biology of Potamogeton and autogamy in aquatic angiosperms. Syst. Bot. 12: 98-105. 1972. The Pollination of Flow- Submarine Proctor, M. & P. YEO. k. 967. The Biology of Aquatic Vas- cular Plants. St. Martin's Press, New York. Adaptive radiation of reproduc- s. I: pollination 4. Flowering Plants: Evolution Above the Species Level. rie aca Press of Harvard Univ., Cam- bridge, Massachusetts ToMLINSON, P. B. 1982. Volume VII. Helobiae ml matidae). /n: C. R. Metcalfe (editor), Anatomy of the al ig igs Clarendon Press, Oxford. hgp er EN, J. T. A. 1979. The ecology i br (od ated communities in western Europe. I. Dis- rysaline Potamogetonaceae. Aquatic Bot. 14: 339- 3 WHITEHEAD, D. R. 1969. Wind pollination in the a gios sperms evolutionary considerations. Evolution 23. WODEHOUSE, R. P. 1935. Pollen Grains. McGraw-Hill, R. 917. The pollination of Vallisneria bo Bot. Gaz. (Crawfordsville) 63: 135-145. INSECT FORAGERS ON Gregory J. Anderson? and David Symon? SOLANUM FLOWERS IN AUSTRALIA! ABSTRACT Eighteen native insect species were found on flowers of 18 Solanum pe ina an study in Australia. All of the Solanum species studied are endemic to Australia, and about one-half of them are andromonoecious or dioecious. Fifteen of the insect species and 93% of total floral visitors were bees. New records of activity on Solanum flowers in Australia are reported for Braunsapis and Xylocopa (Anthophoridae) , Leioproctus (Colletidae) , and Trigona (Apidae). Two species each of the dq en-collecting bees Amegilla ( Anthophoridae) , Nomia (Ha- lictidae) , and Trigona are considered the n gnificant floral visitors. This conclusion is based on the distribution, bos e, and behavior of the bees, bs on the high perc entage of Solanum pollen in pollen loads. Amegilla and Nomia extract pollen by "buzzing" it out i are hypothes Solanum flowers are of the “‘dish-bowl” type of Faegri & van der Pijl (1979) and as a consequence do not physically exclude floral visitors. However, these wide-open flowers do not represent the cor- nucopia to bees and other floral visitors that some other dish-bowl species do (e.g., Tilia, Anderson, 1976) because floral rewards are limited and there are specialized requirements for pollen extraction. Ithough extrafloral nectaries have been described in Solanum, floral nectar is absent (Anderson & Symon, 1985); and pollen, the only reward offered, is not easily accessible to all floral visitors. Solanum is the exemplar of the more than 540 genera whose anthers open by terminal pores rather than by longitudinal slits (Vogel, 1978; Buchmann, 1983). Solanum pollen is typical for species with poricidal flowers (Buchmann, 1983) in that it is relatively dry (not sticky) and has a smooth, granulate tectate exine (Anderson & Gensel, 1976). To remove pol- len, floral visitors can either “milk” the anthers by stroking them from base to apex with their mandibles (e.g., Thorp & Estes, 1975), dig it out of the terminal pores, steal pollen by biting holes in the sides of the anthers (Buchmann, 1983), or buzz the pollen out of the terminal pores. Thorp & Estes (1975) described buzz or vibratile polli- nation succinctly as "shivering the indirect flight muscles of the thorax while the wings [are] in repose." Buchmann (1983) estimated that about 60% of angiosperm species with poricidal anthers, including Solanum, are buzz pollinate The general syndrome of Solanum allinsi. as described above, is well known. However, there is little known about specific pollinators and polli- nation. This is particularly true for Australia, where even some of the floral visitors are unknown (see below). Michener's (1965) major study of the bees of Australia reported collections of only three species from three genera on a single species of Solanum in southern Queensland. Symon's (1979) review of Solanum pollinators includes reports of seven taxa of bees. Armstrong's (1979) thorough over- view of biotic pollination in Australia includes only one citation beyond Michener's and Symon's stud- ies. Thus, we present information that expands the data base on Solanum pollinators. Also included is an analysis of insect pollen loads and relative abun- dance of insects on flowers to address the question of their importance as effective pollinators of Sola- num. Finally, we speculate on the role of pollinators ' We are especially grateful to Terry Houston of the Western Australian Museum (Perth) for pag Shek the vast majority of the insects mentioned herein. We thank Mary Ja erson, P. Bernhardt, K. Holsinger, T.H pn ; pu z ANA gy and Evolutionary Biolo ane Spring for Apnd the figure, Rose Karosi ston, fs T. Philbrick, R. W. ributed observations ational Science Foundatio y U-43, U Iniversity of Ti ticut, Storrs, Cones 062068, U.S.A. ' State Herbarium, North Terrace. Adelaide, South Australia 5000, Australia ANN. Missouni Bor. GARD. 75: 842-852. 1988. Volume 75, Number 3 1988 Anderson & Symon 843 Insect Foragers on Australian Solanum Tropic of 9 Alice Capricorn Springs 40 37 42 Perth 36 35 Adelaide FIGURE 1. in the evolution of nonhermaphroditic breeding sys- tems. METHODS Insects were collected only from open Solanum flowers (i.e., not from the extrafloral nectaries) primarily from natural populations in 1979-1980. The study was centered on andromonoecious or dioecious solanums, the distribution and biology of which are given by Symon (1981). The species studied are as follows. The five-letter abbreviations are those used in Appendix I; the single letters indicate whether the species bears only hermaph- roditic flowers (H), or whether it is andromonoe- cious (A) or dioecious (D): Solanum asymmetri- phyllum Specht (asymm) (D) S. beaugleholei Symon (beaug) (A), S. cinereum R. Br. (ciner) (A), S. cunninghamii Bentham (cunni) (D), S. dioicum W. Fitzg. (dioic) (D), S. diversiflorum F. Muell. (diver) (A), S. eburneum Symon (eburn) (A), S. ellipticum R. Br. (ellip) (H), 5. esuriale Lindley (esuri) (H), S. hoplopetalum Bitter & Summerh. (hoplo) (H), S. leopoldense Symon (leopo) (D), 5. lucani F. Muell. (lucan) (H), S. parvifolium R. Br. Field collections. Numbers correspond to the study areas cited in Appendix I. (parvi) (H), S. petrophilum F. Muell. (petro) (H), S. quadriloculatum F. Muell. (quadr) (H), 5. stur- tianum F. Muell. (sturt) (H), S. tudununggae (tu- dun) (D). Because these Solanum species occur primarily in northern Western Australia and the Northern Territory, fieldwork was concentrated there. The study in this region ranged over some 5,000 km. Collections were also made in South Australia, in- cluding a small sample taken from garden-grown native plants in Adelaide, and from Western Aus- tralia (by R. Thorp). Flowers were open, and col- lections and observations were made throughout the daylight hours. Thousands of individuals were observed, but not all floral visitors were taken. After vouchers were taken within a given area, only observations of behavior were recorded. Specimens netted from flowers were stored with cotton in separate vials to prevent contamination of pollen loads. The plant species, time of day, and notes on insect behavior were recorded at each site. Pollen loads removed from insects were mounted on slides in aniline blue in lactophenol and analyzed for the percentage of Solanum pollen and the num- 844 Annals of the Missouri Botanical Garden TABLE 1. Summary by genus of insect visitors to Solanum flowers. See Appendix I for detailed information. * = see Discussion; others have reported vibratile activity. Num % Sola- ber o of Vibra Sec num Num- Sola Num Total tile onds % Pollen r of Num- num ber of | Sample Pollen Each Activity in ther of ecies Locali (indi- Re Flower after Pollen Polleno Species Visited ties viduals) moval (range) | Noon Load morphs Infrequent floral visitors Diptera 2 3 no Braunsapis 1 2 2 < no ylaeus 2 1 1 < no Lasioglossum 1 1 1 « no* Leioproctus l 1 1 < no l l l < yes Abundant floral visitors Amegilla 3 7 11 yes 1-3 5 79 1.1 Nomia 4 11 20 34 yes 1-10 9 79 1.1 Trigona 2 7 17 44 no 1-90 16 99 ber of other pollenomorphs. The fidelity to Sola- num flowers was estimated from these calculations. Plant vouchers are in AD and insect vouchers in the collections of the Western Australian Museum in Perth. RESULTS INSECT DIVERSITY AND DISTRIBUTION Specific data on time and place of collection and on pollen loads carried are presented in Appendix I, and the locations of the collection sites given there are plotted in Figure 1. The data from the Appendix are summarized in Table 1. More that 150 insects, 93% of which were bees, were collected, representing at least 18 species in two orders (Appendix I). All of the bees were fe- males except one Amegilla pulchra taken from S. ellipticum in South Australia. In addition to collections from Western Australia and the North- ern Territory, a few individuals of Nomia and Ame- gilla were taken in South Australia. The following four bee genera are reported here for the first time from Solanum flowers in Australia: Braunsapis, Leioproctus, Trigona, and Xylocopa. The em- phasis in the following analyses centers on the three most frequently collected genera: Amegilla, Nom- ta, and Trigona. The collection sites (Fig. 1) of the three primary bee genera correspond to the general range of the solanums in the study area (Symon, 1981). As able 1 shows, individuals of Nomia and Trigona were found at nearly twice as many sites as Ame- gilla, and individuals of Amegilla were less abun- dant than either Nomia or Trigona. The two species of Amegilla were equally abundant and wide- spread; one species was more widely distributed and was locally abundant in Nomia (N. flavovir- idis) and in Trigona (undescribed species “B”). While these data accurately reflect the rank-order of occurrence of the genera as we observed them in the field, the frequency figures are biased against Nomia and Trigona, which were much more abun- dant than reflected in the collections. Trigona was particularly abundant, with sometimes tens of in- dividuals foraging over the flowers of a single plant. FORAGING BEHAVIOR As indicated in Table 1, Xylocopa were observed to vibrate or “buzz” the Amegilla, Nomia, and pollen out of the flowers; the remaining five bee genera were not. Lasioglossum, however, has been observed to buzz flowers in other studies (e.g., Bernhardt, 1986; Buchmann, 1983). Of the five genera not observed to utilize vibratile extraction, only Trigona was found on more than three oc- casions (Table 1); thus, the other four are not considered primary pollinators. Individuals of Tri- gona collect pollen from the anthers by digging it out of the terminal pores and by scavenging for it on floral parts such as the corolla and stigma (Fig. 2). No obvious differences in behavior were noted for insect visitors to hermaphroditic flowers, to staminate flowers of the andromonoecious species, or to the staminate or pistillate flowers of the dioe- cious species. Volume 75, Number 3 1988 Anderson & Symon 845 Insect Foragers on Australian Solanum FicURE2. A Trigona (within circle) foraging on the s.v of a hermaphroditic flower of the andromonoecious Solanum beaugleholei. The flower is about 30 mm in diameter. The length of time individual bees stay on flowers is correlated with their size and capability to buzz flowers. The bees that vibrate pollen out of the flowers are generally larger (Amegilla—about 13 mm; /Vomia—8- 10 mm) and stay for only a short time (one to a few seconds) on each flower (Table 1). On the other hand, individuals of Trigona (smallest of the three species, about 5 mm) spent up to one minute on each flower. Individuals of Trigona frequently visited several flowers in the same inflorescence. This is in contrast with Ame- gilla, where an individual usually visits only a single flower per plant and then flies some distance, often away from the population under study. Individuals of Nomia most often foraged within and/or among inflorescences of the same populati Most Amegilla visits take place "m 10 A.M. (Table 1; Appendix I, column 5). The majority of visits by Nomia and Trigona also occur during this period, but significant proportions of the individuals of these two genera are also active after noon. INSECT FIDELITY The average fidelity for all six major insect vis- itors (two species each from Amegilla, Nomia, and Trigona) is high (Table 1; Appendix I, column 3). For the genera overall, however, the fidelity esti- mate for Trigona is 20% higher than for either Nomia or Amegilla. These figures are, as one would expect, paralleled by the estimates of the number of other species visited (Table 1; Appendix I, column 4): Nomia and Amegilla pollen loads include about seven times as many other species as found in pollen loads of Trigona. The pollen loads from the scopae or corbiculae respectively of the Nomia and Trigona were some- what more sticky than those carried by Amegilla. 846 Annals of the Missouri Botanical Garden Amegilla scopae are covered with long hairs, among which the drier pollen loads were packed. DISCUSSION Solanum flowers present a relatively rich pollen resource for the bees that can exploit them. AI- though they lack nectar and restrict access to pollen (having only terminal anther pores), they are heavily visited by at least a few species. This is likely a tribute to their local abundance and the relatively large quantity of pollen available per flower (more than one million grains in some species, Anderson & Symon, in press). Although a range of insects was observed to visit the flowers, not all were anaes oo dogs (Tahle 1). Ba the range of the pies five Ph the bee genera and other insect groups are considered relatively insignificant pollinators, but these cannot be ruled out as oc- casional pollinators (see below). Two species each in Amegilla, Nomia, and Trigona are the primary floral visitors, and, as a consequence of this and their behavior, we propose that they are the major pollinators of the Solanum species studied. Michener (1965) suggested that a large per- centage of the Australian bees are oligolectic on the Myrtaceae but proposed that this is due largely to the overwhelming abundance of species in this family. In addition, he listed several genera, in- cluding Nomia, Trigona, and Amegilla, as ex- amples of bees visiting a wide variety of species. In fact, some of the same genera of pollen-col- lecting bees (Nomia, Trigona, Braunsapis, Xy- locopa) are reported as pollinators of another species from nearby Indonesia that is dioecious and offers only pollen as a reward, Decaspermum parviflo- rum (Myrtaceae) (Kevan & Lack, 1985). Although Michener (1965) did not treat the bee species we studied, this generalist behavior also likely applies to them. On the o and, the proportion of Solanum pollen in pollen loads (79-99%) implies fidelity high enough to consider these polylectic genera to be important pollinators and to be tem- porally and spatially specialized on Solanum. This supports the contention by Thorp (1979) and Arm- strong (1979) that most pollen-collecting bees show a high degree of diurnal or temporal constancy. Prance (1985) made a similar suggestion for poly- lectic bees from the Amazon rain forest. Although the sticky pollen masses carried by individuals of Nomia and Trigona might indicate collection of nectar from other species (Thorp, 1979; Buchmann, 1983), the nearly monotypic nature of the loads (especially of the Trigona) suggests that some other substance could be in- volved (perhaps stigmatic exudate, see below). Al- ternatively, the Trigona species could be moisten- ing pollen loads with honey carried in their crops. Obviously nectar or honey from some source other than the Solanum flowers provides the carbohy- drate resource to subsidize the pollen-collecting activities of the bees. Nomia and Amegilla are ground-nesting, soli- tary, larger bees and are well represented in Aus- tralia (85 and 69 species, respectively; T. Houston, pers. comm.). These genera also share the ability to vibrate pollen out of the anthers. Michener (1965) indicated that the highly social, tree-nesting 7ri- in although not as diverse in Australia (only ut 14 species, T. Houston, pers. comm.), visits a side range of monocots and dicots. However, this mostly tropical genus (Bernhardt, 1987) does not vibrate pollen out of the anthers and is too small to simultaneously contact stigmata while working anthers of a few of the large-flowered diclinous solanums. Are species of Trigona then simply pollen thieves removing the reward without effecting pollination? In this instance, that conclusion does not seem warranted. We never observed them biting holes in anther bases to steal pollen, and anthers of dried voucher specimens do not have holes. Some indi- viduals of Trigona opportunistically collect pollen spread over the flower, taking advantage of the activities of the vibratile pollinators. However, oth- ers are active on flowers not visited previously; we regularly observed them digging pollen out of the terminal pores of anthers. Members of Trigona were also observed foraging on stigmatic surfaces (Fig. 2). They may have been gathering pollen, but it is also possible that they were collecting stigmatic fluid to cement pollen grains together, as Baker et al. (1973) reported for other angiosperms. In either case, with such behavior they could effectively transmit pollen from anthers to stigmata in even the large-flowered species. Given this, and that Trigona species were omnipresent diurnally, were more abundant on Solanum flowers than all the other species combined, and showed 99% fidelity, we propose that these little bees are significant pollinators. In view of the fact that representatives of Tri- gona tended to visit flowers in the same inflores- cence, those on the same plant, or those within a population, we suggest that most often they effect self-pollination. This is reinforced by the colony behavior of social bees like Trigona, where indi- viduals in various parts of a colony tend not to sample widely but continue to visit one area or Volume 75, Number 3 1988 Anderson & Symon 847 Insect Foragers on Australian Solanum population repeatedly (T. Seeley, pers. comm.). Two features of the plants are relevant to this hypothesis as well. First, it is possible to self because most species of Solanum that have been studied (outside of the tuberous solanums and their rela- tives), including those from Australia, are self-com- patible (Anderson & Symon, in press; Whalen & Anderson, 1981). Secondly, we observed that pop- ulations of most of the Solanum species studied are small in size and are widely separated from each other. These populations are even smaller (genetically) when one takes into account that most of the species reproduce vegetatively and form large clones, thus most of what appear to be genets in an area are actually ramets (Symon, 1981; Anderson & Stebbins, 1984; Anderson & Symon, in press). As a result, even many foraging visits between “plants” simply constitute visits to differ- ent ramets. The largest but least abundant bees (Amegilla) visit flowers for only a very short time, supporting Buchmann's (1983) suggestion that the length of buzzing time is inversely correlated with bee size. These bees behave like “‘trap liners" (e.g., Janzen, 1971); that is, most visited only a single flower in an area and then flew off and out of sight, pre- sumably to another Solanum flower (based on the nearly 80% purity of pollen loads). Such behavior supports the contention that Amegilla species are outcrossing agents transmitting pollen among gen- ets. The bees in the first section of Table 1 were not abundant within any population at any site studied. Thus, if they are significant pollinators, they are also likely to effect outcrossing. The Nomia species are intermediate between the small Trigona and large Amegilla in temporal pattern, abundance, size, visitation times, and be- havior (Table 1). Individuals of Nomia visited more than one flower per plant and often visited other plants within the population before flying out of sight. As a consequence, we predict that these species generally effect inbreeding. t seems that most flowers are visited, and ob- servations of fruit set from the previous season showed a high seed set. Thus, we conclude that seed set is likely not pollen limited. Snow (1986) implied this is often the case for insect-pollinated species. The nectarless flowers of Solanum may promote interplant or interpopulation foraging as suggested by Bernhardt (1987) for Australian Acacia (which also has no floral nectar). Bees collecting pollen have to find other species as nectar sources; this interruption raises the possibility of returning to a different plant or population of Solanum, thereby increasing the chance of effecting outcrossing. On the other hand, given the population structure of the solanums and the pattern of visitation (espe- cially that of Trigona), it is likely that more than three-fourths of interfloral visits result in self cross- es. This strong likelihood lends strength to the arguments (Anderson & Stebbins, 1984; Anderson & Symon, in press) that dioecy, which promotes genetically wider crossing, may have been selected in response to several features of the hermaphro- ditic-flowered progenitors of the andromonoecious and dioecious species. These features include self- compatibility, vegetative reproduction, scattered population distribution, and the behavior of the pollinator assemblage as described herein. This exclusive pollen-collecting behavior of in- sects on Solanum flowers has been accommodated in the Australian diclinous species. In both andro- monoecious and dioecious species, all flowers are morphologically hermaphroditic (Anderson & Sy- mon, in press). In staminate flowers of the andro- monoecious and dioecious species the gynoecia are present but reduced. This reduction is presumably of little consequence to the pollen-collecting for- agers. Of more importance is the fact that in the dioecious species the pistillate flowers bear anthers with pollen. This pollen reward differs from the usual tricolporate pollen of Solanum in that it is inaperturate (Anderson & Gensel, 1976). The in- aperturate pollen is fully viable but incapable of hus, the Australian dioecious species have a par germination (Levine & Anderson, system that maintains pollinator visitation, but one that also disallows the self-pollination promoted by the plant biology and pollinator behavior. LITERATURE CITED Ap G.J. 1976. The aa biology of Tilia. Amer. J. Bot. 63: 1203-12 — & P. G. GENSEL. 1976. Pollen ¡mesphalog " " P fS Pollen & Spores 17: 533-552 — n & G. L. SrEBBINs. 1984. Dioecy versus ga- metophytic sali incompatibility: a test. Amer. Natu- ralist 124: 423-428. . E. Symon. 1985. Extrafloral nectaries in Solanum. Biotropica 17: 40-45. Functional nad and andromon- in So lanni Evolution (in press). Awstsm0sG, J.A. 1979. Biotic pollination mechanisms n the peser flora—a review. New Zealand J. Bot. 4: -508 BAKER, H. G., ki Baker & P. A. OPLER. 1973. Stigmatic exudates and pollination. Pp. 47-60 in N. B. M. Sten (editor), Pollination and Dispersal. Depart- any, Univ. Nijmegen, Nijmegen, Neth- ph s. E P. 1986. Bee-pollination in Hibertia fas- 848 Annals of the Missouri Botanical Garden ciculata (Dilleniaceae). Pl. Syst. Evol. 152: 231- joy (editors Key Environments: Amazonia. Perga- 24]. mon Press, Oxford. A comparison of the diversity, density, Snow, A. 1986. Pollen dynamics in Epilobium canum and foraging behavior of bees and wasps on Austra- (Onagraceae): [e rig. for gametophytic selec- lian Acacia. Ann. Missouri Bot. Gard. 74: € m tion. Amer ot. 73: 139-151. BUCHMANN, S. L. 1983. Buzz pu in angios Symon, D. E. 1979. Sex forms in Solanum (Solanaceae) p. 73-113 in C. E. Jones & R. J. Little (editors) and the role of pollen ora eae insects. Pp. 385- Handbook of nd eun Biology. Var Nostrand Reinhold, New FAEGRI, K. L. & L. VAN DER Po. "1979, The Principles of Pollination Ecology. Pergamon Press, New York. JANZEN, D. H . Euglossine bees as di distan pollinators of tropical plants. Science 171: -205. KEVA Pk P. G. & A. J. Lack. n Rm in cryptically dioecious d (Dec orum (Lam.) A. J. Scott Macaco) by pullin collecting bees in Sulawesi, Indonesia. Biol. J. Linn. Soc. 25: 319-330. Levine, D. A. & G. J. ANDERSON. 1986. Evolution of dioecy in an American Solanum. Pp. 264-276 in W. G. D'Arcy (editor), Solanaceae: Biology and Sys- tematics. Columbia Univ. Press, New York. MICHENER, C. D. 1965. A classification of the bees of the Australian and South ja regions. Bull. Amer. Mus. Nat. Hist. 130: 1- Prance, G. T. 1985. c M of Amazonian plants. Pp. 166-171 in G. T. Prance & T. E. Love- z 397 i n J. G. Hawkes, R. N. Les A. D. Skelding (editors), The Biology and DU of the Sola- naceae. Linnean iiid Symposium Series, 7. Ac- = d à 81. n A revision of the genus Solanum in -367. . Structural, behavioral, and phys- iological adaptations of bees M ara for collecting pollen. Ann. Missou is Gard. 66: 788-812 . ESTES 175; n behavior of bees on fHowens of Casi fasciculata. J. Kansas Entomol. Soc. 48: 175-184. VocEL, S. : NC shifts from reward to deception in pollen flowers. Pp. 89-96 in A. J. Rich- ards (editor), The Pollination of Flowers by Insects. Linnean Society Symposium Series, 6. Academic Press, London. Wuaten, M. D. & G. J. ANDERSON. 1981. Distribution of self-incompatibility and infrageneric classification in Solanum. Taxon 30: 761-767. Volume 75, Number 3 1988 Anderson & Symon Insect Foragers on Australian Solanum 849 * APPENDIX I. Insects on Solanum flowers. See Methods for full names and authorities for the Solanum species. d sp. = species not identified, s.n. = without number, = plant cultivated. Number of Time Insect Other of Day Study Site Solanum Voucher % Species ap- see Insects Species Number Fidelity Visited tured Fig. 1) Bees Anthophoridae Amegilla (Amegilla) aeruginosa cunni 341 7:00 8 (Smith) 348 7:00 8 dioic 369 100 0 7:30 18 eburn 300 12 1 9:30 1,2 301 69 1 9:30 1,2 304 8:30 30 305 8:30 30 435 100 0 8:30 30 436 8:30 30 437 8:30 30 438 87 8:30 30 X = 85 X = 0.8 A. (Amegilla) pulchra (Smith) diver 306 37 3 15:00 2a 313 37 2 9:30 3 398 3 7:30 2a ellip 460 100 0 15:00 37 416 100 0 15:00 37 472 100 0 40 esuri 457 4] hoplo s.n 42 sturt s.n. 37 sp 474 41 X = 73 X = 1.33 A. sp.? dioic 363 900 16 364 900 16 Genus (Amegilla) X = 79 X = 1.09 11 sites Braunsapis sp. ? dioic 351 100 0 8:00 14 eburn 450 10:30 31 Xylocopa (Koptortosoma) aruana eburn 444 99 1 10:30 31 itsema Apidae Trigona (Plebeia) sp. A cunni 344 7:00 8 345 7:00 8 346 7:00 8 347 7:00 8 dioic 383 100 0 11:00 21 384 100 0 11:00 21 diver 393 100 0 7:30 2a 396 100 0 7:30 2a lucan 408 9:00 24 409 9:00 24 410 9:00 24 411 9:00 24 412 9:00 24 413 9:00 24 414 9:00 29 415 100 0 9:00 24 416 100 0 9:00 24 417 100 0 9:00 24 850 Annals of the Missouri Botanical Garden APPENDIX I. Continued. Number of Time Insect Other of Day Study Site Solanum Voucher % Species Cap- (see Insects Species umber Fidelity Visited tured Fig. 1) quadr 418 11:00 25 419 11:00 25 XA=100 AXA=0 Trigona (Plebeia) sp. B beaug 370 9:00 20 371 9:00 20 372 9:00 20 373 9:00 20 374 9:00 20 375 9:00 20 376 99 1 9:00 20 dioic 326 100 0 15:00 6 327 100 0 15:00 6 328 15:00 6 329 15:00 6 330 15:00 6 352 100 0 9:00 16 353 100 0 9:00 16 354 9:00 16 355 9:00 16 368 12:00 17 382 11:00 21 385 100 0 11:00 21 386 100 0 11:00 21 diver 307 308 100 0 9:30 3 309 96 l 9:30 3 310 9:30 3 311 9:30 3 312 9:30 3 324 100 0 10:00 5 325 100 0 10:00 5 394 100 0 7:30 2a 395 100 0 7:30 2a 397 90 1 7:30 2a 400 7:30 2a 401 7:30 2a eburn 303 99 1 10:00 2 429 15:30 29 430 100 0 15:30 29 439 100 0 8:30 30 lucan 337 15:00 9 451 100 0 11:30 32 452 12:30 33 453 12:30 33 454 100 0 12:30 33 X-99 X=0.22 Genus (Trigona) X=99 X=0.16 17 sites Colletidae Hylaeus (Prosopisteron) sp. 1 cunni 334 12:00 8 335 12:00 8 H. (Rhodohylaeus) sp. 2 cunni 336 95 l 12:00 8 Leioproctus (Leioproctus) sp. hoplo s.n. 42 Volume 75, Number 3 1988 Anderson & Symon y Insect Foragers on Australian Solanum 851 APPENDIX I. Continued. Number of ime Insect Other of Day Study Site Solanum Voucher % Species ap- see Insects Species Number Fidelity Visited tured Fig. 1) Halictidae Lasioglossum sp. leopo* 459 79 2 10:00 36 parvi* s.n. 36 Nomia (Austronomia) dimissa (?) beaug 380 93 1 9:00 20 Cockerell N. (Austronomia) flavoviridis beaug 379 83 9:00 20 Cockerell 381 100 0 9:00 20 ciner* s.n. 36 dioic 315 75 2 11:00 4 331 100 0 10:00 6 332 99 1 10:00 6 358 100 0 9:00 16 359 100 0 9.00 16 360 100 0 9.00 16 361 65 1 900 16 362 50 2 9:00 16 diver 317 95 2 10:00 5 318 98 Z 10:00 5 319 54 2 10:00 5 320 10:00 5 321 10:00 5 322 10:00 5 eburn 431 100 0 15:30 29 432 83 1 15:30 29 433 39 2 15:30 29 434 84 1 15:30 29 446 65 1 10:30 31 447 9 1 10:30 31 448 10 1 10:30 31 449 13 3 10:30 31 ellip s.n. 37 esuri 458 14:00 35 lucan 407 100 0 9:00 24 455 34 3 12:30 33 petro s.n. 37 quadr 420 11:00 25 421 99 1 11:00 25 422 100 0 11:00 25 sturt 473 100 0 40 tudun 428 100 0 11:00 26 X=78 X=1.0 N. (Hoplonomia) rubroviridis cunni 342 33 1 7:00 8 Cockerell 343 46 3 7:00 8 349 98 1 12:30 11 eburn 302 T3 2 10:00 2 lucan 402 97 2 9:00 24 403 99 1 9:00 24 404 100 1 9:00 24 X=80 X=1.6 N. sp. ? dioic 387 99 10:00 22 388 10:00 22 389 10:00 22 852 Annals of the Missouri Botanical Garden APPENDIX I. Continued. Number of Time nsect e of Day Study Site Solanum Voucher % Species Cap- see Insects Species Number Fidelity Visited tured Fig. 1) 390 10:00 22 diver 314 37 2 9:30 3 Genus (Nomia) X=79 X= 1.14 20 sites Floral Visitors Other than Bees Eumenidae (Hymenoptera) asymm 0 34 Ants dioic 333 15:00 6 Diptera quadr 423-427 10:00 25 cunni 338 7:00 8 339 7:00 8 TWO NEW SPECIES OF CALYPTROCARYA (CYPERACEAE: SCLERIEAE) FROM VENEZUELA AND OBSERVATIONS ON THE INFLORESCENCE MORPHOLOGY OF THE GENUS: Gerrit Davidse? and Robert Kral? ABSTRACT Two new species, C. montesi multiple, sessile, headlike os Calyptroc rhizomes. Calyptrocarya montesii is the second Ga inflorescence morpholog separate inflorescence units in all species of the gen i and C. delas Sena are described from Guárico, Venezuela. Both species have ya — is the first species in the genus with elongated the genus with three stigmas. A detailed discussion of y is present ted, demonstrating kai compound male and female spikelets are borne in A recent review of specimens of Calyptrocarya for a treatment in the Flora of the Venezuelan Guayana has brought to light two distinctive gath- erings of Calyptrocarya from a region just north of the flora area. This tropical American genus has been revised by Koyama (1967, 1969b), who recognized six species. The presence of two undescribed endemic taxa from a small region in the state of Guárico may seem suspicious, yet seems to be justified by their distinctive morphologies. The low mesas of Parque Nacional Aguaro- -Guariquito that rise out of the llanos harbor a number of d elements. One of these is Rhynchospora eso W. Thomas (1984), a species recently described from Montana de Guardahumo, in Parque Nacional Aguaro-Guariquito. Ironically, the type of C. mon- tesii was collected as a mixture in the type gath- ering (Delascio, Montes & Davidse 11342) of R. papillosa. ti 1 ULSUJIU UL DESCRIPTION OF NEW SPECIES Calyptrocarya montesii Davidse & Kral, sp. ov. TYPE: Venezuela. Guárico: Distrito Mi- randa, Parque Nacional Aguaro-Guariquito, Montana de Guardahumo, ca. 8?88'-8?92'N, 67*44'W , 40-60 m, Dec. 1981, F. Delascio, R. Montes & G. Davidse 113424 (holotype, MO). Figure 1. Species affinis C. delascioi a qua habitu caespitoso, culmis rae 2, bracteis capitulis longioribus et stig- matibus 3 differ Cespitose annual with lax culms and leaves. Roots fibrous, slender. Lowermost leaves mostly sheaths, short, keeled, multinerved, dull brown, friable and soon lost; principal leaves well overtopping the in- florescence, the largest lowest and more approxi- mate with short, dull brown sheaths; blades 10- 30 cm long, 3-5 mm wide, linear, thin, flat, gla- brous, 3-costate, the midrib projected on the lower surface, the lateral costae projected on the upper surface, the apex abruptly acute, the margins sca- bro-ciliate only distally, otherwise smooth; upper leaves gradually reduced and more distant, with all but the uppermost blades elongate and over- topping the inflorescence. Culms 6-16 cm long, ca. 0.5 mm wide, all fertile, centrally borne, trig- onous; nodes usually 3, the upper 2 usually bearing inflorescences; internodes glabrous with scaberu- lous culm angles. Inflorescence a congested head ' Fieldwork was supported by NSF grant INT 80-0982 (United States) and CONICIT (Venezuela). Davidse thanks Francisco Delascio po and Rubén A. Montes for inviting him to join in their exploration of Parque n, P.O. Box * Herbarium, P.O. Box 299, St. Louis, Missouri 63166, U.S. A. 1705, Station B, Vanderbilt University, Nashville, Tennessee 37235, U.S.A. ANN. MISSOURI Bor. GARD. 75: 853-861. 1988. Annals of the Missouri Botanical Garden IGURE 1. Calyptrocarya montesii Davidse & Kral (from holotype) .—a. Habit sketch.—b. Upper surface of leaf blade tip.—c. Upper surface of mid blade.—d. Lower surface of mid blade.—e. Portion of inflorescence showing two heads composed of globose clusters of compound spikelets. —f. Staminate spikelet. —g. Young female compound spikelet. —h. Dissection of female compound spikelet, from left: lowest female spikelet bract; inter- mediate bract (prophyll) ; adaxial view of upper bract and sterile spikelet; young fruit with attached upper bract and its sterile spikelet. —i. Left, upper bract complex; right, mature fruit. Volume 75, Number 3 1988 Davidse & Kral New Venezuelan Calyptrocarya 855 5-8 mm long and wide, composed of smaller glo- bose clusters of compound spikelets; globose clus- ters sessile or nearly so; involucral bracts 1-3, leaflike, 0.5-8 cm long; lower 1-3 compound spikelets in a globose cluster male, the upper 4-8 compound spikelets female. Male compound spike- lets 1.3-2.2 mm long, usually unbranched; lowest bract 1.8-2.2(-8) mm long, membranous or in the lowest spikelet of the head sometimes herbaceous and leaflike, several-nerved; prophyll 1.3-1.6 mm long, 2-keeled, membranous, truncate, inconspic- uously many-nerved; bracts subtending the solitary stamens 5-7, the uppermost usually reduced and rudimentary and not subtending a stamen; male flowers solitary stamens, the filaments as long as or slightly longer than the subtending bracts at anthesis, the anthers 0.7-1.2 mm long, linear. Female compound spikelets 3.1-4.2 mm long, composed of 5 primary bracts; lowest bract 1.8— 4.2 mm long, ovate, 9-15-nerved, acute or awned to 1 mm; prophyll 1-1.1 mm long, often split into 2 parts, 2-keeled, many-nerved; upper bracts 3, 1.6-2 mm long, 1.4-1.7 mm wide, herbaceous, broadly ovate, 15-19-nerved, prominently apic- ulate, each subtending 1 sterile, axillary spikelet and surrounding a central, terminal female spikelet; sterile spikelets 2.9-4 mm long including the ped- icel, much exceeding the achene, long attenuate, the lowest bract 1.4-2.1 mm long, 2-keeled and clasping at the base, membranous, inconspicuously many-nerved, the upper bract 2.4-3 mm long, —0.5 mm wide at the base, lanceolate-linear, strongly 2-keeled, the tip of the keels free as awn points 0.1-0.3 mm long, the lateral nerves 2-4; female spikelet a solitary female flower; achenes 1.5-1.6 mm long, ellipsoid; utricle with the base triangular in outline, somewhat thickened and forming a low 3-lobed annulus, the upper part a hyaline, pale brown, smooth to minutely puberulent sac closely appressed to the achene; beak ca. mm long; stigmas 3, ca. 1 mm long. Calyptrocarya montesii is known only from the type collection. It is named in honor of Dr. Rubén A. Montes, a co-collector of this new species. This species is most clearly characterized by its lax foliage, sessile, multiple inflorescences, and three stigmas. The species appears to be annual, which would be unique in the genus. However, the avail- able specimens are too few to be certain of this interpretation e only other species in the genus with three stigmas is C. luzuliformis T. Koyama, from which C. montesii differs by having relatively broader, laxer leaves, sessile heads, and minutely puberulent rather than pubescent utricles. Two other species have sessile headlike inflo- rescences, C. monocephala Hochst. ex Steud. and C. delascioi. Calyptrocarya montesii differs from C. monocephala in its multiple rather than solitary inflorescences, broader and laxer leaves, and three versus two stigmas. Calyptrocarya montesii differs rom C. delascioi in the lack of rhizomes, less lec] numerous inflorescences per culm, much longer involucral bracts, smaller achenes, and three rather than two stigmas. E delascioi Davidse & Kral, sp. TYPE: Venezuela. Guárico: Distrito Mi- Sa N Parque Nacional Aguaro-Guariquito, Montana de Guardahumo, ca. 8°88'—8°92'N, 67°44'W, 40-60 m, Dec. 1981, F. Delascio, R. Montes & G. Davidse 11336 (holotype, MO; isotype, VEN). Figures 2, 3 Species rhizomatibus productis gracilibus a congene ribus diversa; folia laxa, culmi capitulis 3-8 aa 8 stigmata 2. Rhizomatous perennial with lax culms and leaves; aerial shoots up to 7 clumped together, all fertile; roots fibrous, slender. Rhizomes 1-10 cm long, ca. 1.5 mm thick, scaly, slender; scales 6-10 mm long, purple, multinerved, slightly dilated and open distally at the acute, sometimes bifid apex, usually slightly overlapping, passing into scalelike basal eaves. Principal leaves longest and most crowded toward the culm base, there with short, red to purple loose sheaths to 3 cm long, these ventrally brownish, nearly open, passing at the orifice into strongly folded dull-green blades; blades 30-60 cm long, 3.5-7 mm wide, well overtopping the inflorescence, linear, very lax, thin, flat, gla- brous, 3-costate, the midrib projecting on the lower scarious, surface, the lateral costae projecting on both sur- faces, the tip obtuse, the margin scabro-ciliate dis- tally, otherwise a smooth nerve; upper leaves grad- ually shortening, but even the upper ones overtopping the inflorescences. Culms to 35 cm long, ca. 1 mm wide, lax, centrally borne, glabrous, trigonous, each angle with a strong costa, each concave face with several low ribs, nodes 3-9 (-10), all usually bearing inflorescences. Inflores- cence a congested head -7 mm long, 3.5-9 mm wide, composed of globose clusters of com- pound spikelets; globose clusters sessile or with rays to 5 mm long; involucral bracts inconspicuous, l- 3 mm long, linear-lanceolate; lowest 2-3 com- pound spikelets in a globose cluster male, the upper 5-8 compound spikelets female, larger and firmer than the male compound spikelets. Male compound spikelets 1.5-2.1 mm long, usually unbranched; lowest bract 1.9-2.1 mm long, 7-11-nerved, lan- Missouri Botanical Garden Annals of the 856 ómm 1 ) = AM SS < Seat y — AS > = EE > Bc or fe c > Z = AMS KA, 1 A FD NN AN A Volume 75, Number 3 1988 Davidse & Kral New Venezuelan Calyptrocarya ceolate, acute; prophyll 1.1-1.3 mm long, mem- branous, 2-keeled, 5-7-nerved; bracts subtending the solitary stamens 6-8, 1-1.6 mm long, lanceo- late to linear-lanceolate, membranous, especially the inner, the uppermost 2-3 reduced or rudi- mentary and not subtending a stamen; male flowers solitary stamens, the filaments flattened, slightly longer than the subtending bracts at anthesis, the anthers 0.5-0.8 mm long, oblong-linear. Female compound spikelets 2.5-4 mm long, composed of 5 or rarely 6 primary bracts, the tips and/or upper margins conspicuously purple, otherwise pale green; lowest bract 1.9-2.3 mm long, ovate, herbaceous, 15-19-nerved, acute; prophyll 1-1.2 mm long, usually split into 2 parts, 2-keeled, many-nerved; upper bracts 3 or rarely 4, 2-2.3 mm long, 1.2- mm wide, herbaceous, ovate, acute, 9-11- nerved, acute, each subtending an axillary spikelet and surrounding a central, terminal female spikelet; axillary spikelets sterile or female, 2-3.2 mm long including the pedicel, when sterile composed of a 2-keeled prophyll and 1-3 bracts, when fertile, bearing 3 bracts above the prophyll, each bract subtending a solitary, axillary spikelet, and sur- rounding a central, terminal female spikelet; female spikelet a solitary female flower; achenes 1-1.2 mm long, 0.8-1 mm wide, lenticular; utricle with the base elliptic in outline, slightly thickened and forming a low, trilobed collar around the achene, the upper part a hyaline, pale brown, puberulent sac closely appressed to the smooth stramineous achene; beak 0.1 mm long; stigmas Calyptrocarya delascioi is known only from the type collection. It grew in shallow water of a small stream in the shade of a low gallery forest through the Trachypogon savanna that covers most of the mesa top. At the time of the collection, the leaves and culms were mostly floating on the sur- face of the water. This species is named for Mr. Francisco Delascio Chitty, a co-collector of the type collection. Calyptrocarya delascioi differs from all other species in the genus in the development of long, slender rhizomes; all other species are cespitose. In C. irwiniana Koyama, however, lateral culms are produced that eventually arch and bend toward the ground. Spikelets often become proli produce new shoots that take root (Koyama, erated 1969b). This species is thus functionally stolonif- erous. The central culms of C. glomerulata (Brongn.) Urban also occasionally reproduce plant- lets in the inflorescence through proliferation of spikelets. Calyptrocarya delascioi may be most closely related to C. montesii. See the discussion of that species for a comparison. INFLORESCENCE MORPHOLOGY The inflorescence morphology of Calyptrocar- ya has been consistently misinterpreted in recent years (Eiten, 1976; Koyama, 1967, 9a, b, 1971). For example, Koyama (1967) described the spikelets as “compound, sessile or short-pe- duncled; glumes 6 (including a prophyll); prophyll and the lower 2 empty; the upper 3 bearing an axillary staminate floret; fructification solitary, ter- minal.” A similar interpretation was accepted by Eiten (1976). The typical compound spikelet, such as in Ca- lyptrocarya glomerulata, that is borne toward the tip of a globose head and that matures an achene bears, in fact, 5 bracts (Fig. 4). The lowest bract subtends a reduced axillary branch system that bears a 2-keeled prophyll as its first foliar bract. Both the lowest bract and prophyll do not subtend flowers. Three multinerved bracts are borne suc- cessively above the prophyll, each subtending an axillary sterile spikelet. Koyama (1967) considered these axillary spikelets to be staminate spikelets but, after examining spikelets of all species except C. irwiniana, of which we have not seen authentic material, we have never observed any stamens or even the remains of filaments in these spikelets. Even C. irwiniana, judging from the published illustration (Koyama, 1969b), appears to conform to the general pattern in the other species. The sterile axillary spikelets consist of a pedicel bearing a basal two-keeled prophyll, usually followed by two or three bracts. These two or three bracts are clearly separate in some species (e.g., some spec- imens of C. poeppigiana Kunth, C. delascioi, and C. bicolor (Pfeiffer) Koyama, and many specimens of C. luzuliformis and C. monocephala); however, in certain species (e.g., C. montesii, C. glomeru- lata) and some specimens of most other species, the upper two bracts appear to have their margins — FIGURE 2. Lower side of mid blade of leaf. —d. No Calyptrocarya delascioi Davidse & Kral (from holotype).—a. Habit sketch.—b. Leaf tip.— de from mid culm with attached leaf sheath, adaxial view.—e. Infloresieabe head composed of globose clusters of compound spikelets. 858 Annals of the Missouri Botanical Garden FiGURE 3. Calyptrocarya delascioi Davidse & M» (from holotype) .—a. Inflorescence head with male branch at lower left.—b. Female compound spikelet.—c. Female compound spikelet: left, showing upper pseudowhorl of bracts, sterile spikelet whorl, fruit; right, yon with base of one upper bract pulled downward to show the abaxial side of the sterile spikelet whic A it subtends.—d. Utricle and enclosed achene.—e. Stigma apparatus, lefi; stamen, right —f Staminate bract and its stamens, vie Vai ed male compound spikelet, right. —g. Side- oblique view w of an inner pistillate bract and a sterile sj —h. Abaxial view of sterile spikelet (from g), lefi; dl view of sterile spikelet, right.—i. Mature pw Volume 75, Number 3 1988 Davidse & Kral 859 New Venezuelan Calyptrocarya fused, thus appearing to be a single two-keeled structure. In any case, the compound spikelets near the tip of the globose heads are functionally female with never any evidence of stamens The functionally male compound spikelets are borne separately at the base of each inflorescence head and often also near the tip of each head, depending on the species. The number of parts is much more variable in the male compound spikelets than in their female counterparts. As in the female compound spikelets, the typical male compound spikelet is borne in the axil of a bract and the first bract on the axis is a two-keeled prophyll. The axis beyond the prophyll may be branched or not. When it is unbranched (Fig. 5), the bracts are spirally arranged, and each subtends a solitary stamen which represents the male flower. Typically, the upper- most one to six bracts in a male compound spikelet are reduced or rudimentary. There is no morpho- logical evidence of a pistil in any of the male compound spikelets that we have observed. In branched male compound spikelets (Fig. 6), the branch within the spikelet is borne in the axil of a bract, and the first bract on the branch is a two-keeled prophyll. All subsequent bracts except the uppermost subtend solitary stamens. In some male spikelets of C. poeppigiana that are branched once again, the distinction between the lowest basal bract and prophyll from the bracts subtending sta- mens is difficult to observe, because the bracts become smaller and narrower towards the tip of each compound spikelet. An additional difficulty is that either the lowest basal bract or prophyll may sometimes be absent. It is unlikely, as maintained by Eiten (1976) and Koyama (1967, 1969a, b, 1971), that the axillary spikelets borne below the terminal pistil in the female compound spikelet of Calyptrocarya (Fig. 4) are staminate or rudimentary staminate spikelets. Rather, it seems certain that these sterile axillary spikelets represent rudimentary female spikelets of a branched female compound spikelet. This interpretation is based on three observations: (1) We have not been able to confirm the presence of stamens in such spikelets, contrary to the as- sertions of Koyama (1967, 1969a, b, 1971) and Eiten (1976). (2) The branched male compound spikelets are morphologically equivalent to the branched female compound spikelets. The main difference is that almost all female compound spike- lets uniformly contain three branches, whereas the male compound spikelets are commonly two- branched, less uncommonly unbranched, and least commonly more than two-branched. (3) Compar- i 8 2 FIGURES 4-8. Schematic diagrams of Mo: la und spikelets of Calyptrocarya species. —4. Sim l circles = veo te ns open triangles = fertile male flow ved lines = bracts; hooked curved lines = se oy ison of a gradation series of female compound spikelets in different Calyptrocarya species shows that the axillary spikelets represent female spike- lets. That the relatively simple structure of the female compound spikelet in C. glomerulata (Fig. 4) and C. montesii (Fig. 1g) represents a reduction may be seen by comparing them with the female compound spikelets of C. bicolor, C. delascioi, C. monocephala, and C. luzuliformis. In C. bicolor and C. delascioi, the axillary spikelets commonly bear larger and additional bracts compared with those of C. glomerulata. Such axillary spikelets in C. bicolor are still nonfunctional and usually do not bear a pistil. In C. delascioi, C. luzuliformis, and C. monocephala, on the other hand, the ax- 860 Annals of the Missouri Botanical Garden illary spikelets that surround the solitary terminal pistil are frequently fully developed, each in turn bearing a fully developed functional terminal pistil surrounded by three lateral bracts, and each o these bearing a solitary axillary sterile spikelet (Fig. 7). Instances in which these third-order axillary spikelets are functional may also be observed in C. luzuliformis (Fig. 8). However, in such cases of third-order branching, the number of branches is often less than three, as in Figure 8 where there = are two third-order branches with one of them fertile. From these three lines of evidence, but especially the last where we have demonstrated intergrading series of axillary spikelets in female compound spikelets from rudimentary to fully developed, the conclusion is inescapable that the axillary spikelets borne below the terminal pistil always represent female spikelets. This means that the female and male compound spikelets of Calyptrocarya are always borne on completely separate units of the inflorescence, exactly as in many species of Scleria Bergius, a closely related genus in the tribe Scle- rieae The major feature of this interpretation, namely, that the male and female spikelets of Calyptro- carya are borne on separate inflorescence units, is not new, since Nees (1842) had already correctly described and illustrated the genus as having sep- arate male and female partial inflorescences. His KEY TO THE SPECIES OF CALYPTROCARYA la. Inflorescence composed of headlike, sessile, or nearly sessile clusters of spikelets. 2 Heads solitary per culm characterization seems to have been misinterpreted many subsequent students of the genus, prob- ably through false analogy with other sclerioid and mapanioid genera. Eiten's diagnosis of Calyptrocarya was slightly different from Koyama's interpretation in that she noted and, in fact, illustrated (Eiten, 1976, fig. 14) axillary spikelets lacking male flowers. However, she also indicated that these axillary spikelets usu- ally bear male flowers composed of one stamen. At anthesis, prophylls in male and female spike- lets are often torn and sometimes appear to be two distinct structures. However, this seems to result from the expansion and growth of spikelets at an- thesis, and in the female spikelets especially from expansion of the achene. This commonly torn pro- phyll seems to be responsible for the assertion by Nees (1842), Koyama (1967, 1969a, 1971), and Eiten (1976) that the ultimate unit of the Calyp- trocarya inflorescence is composed of six bracts. In fact, the intact prophyll can be observed readily in very young spikelets of all species and is even easily visible intact in the narrow, postanthesis male compound spikelets of C. luzuliformis and C. mon- ocephala. It is true, as in an unusual spikelet of C. delascioi that we observed, that four rather than three bracts may rarely subtend the central, terminal pistil. But such cases are clearly anom- alous and do not affect our interpretation of the structure of the prophyll. C. monocephala 2b. Heads several per culm Stigmas 3; plants cespitose; involucral bracts conspicuous, longer than the head; heads usually 2 Im per culm montesit 3b. Stigmas 2; plants rhizomatous with elongated rhizomes; involucral bracts inconspicuous, shorter l C. than the head; heads cum delascioi 3-8 per lb. Inflorescence composed of ti gmas 3; leaf blades less than 5 mm w some peduncled globose clusters of spikelets arranged in a cymose ign wide; utricle pubescent C. luzuliformis 4b. Stigmas 2; leaf blades sometimes wider than 5 mm; utricle puberulent. C. irwininana 6b. wasa 1.6-2.2 m nes 1.2-1.5 mm wide, 1-1.5 mm long m wide, 1.7-2.2 mm C. glomerulata lon 7a. IM basal ne 5-9(-12) mm wide, attenuate at the apex, mostly longer than 25 poeppigiana 7b. Upper basal leaves (8-)10-28 mm wide, abruptly acute at the apex, mostly shorter than C. bicolor LITERATURE CITED Erren, L. T. 1976. Inflorescence units of the Cyper- aceae. Ann. Missouri Bot. Gar 1-112. ` T. 1967. Cyperaceae —Mapanioideae. Mem. w York Bot. Gard. 17: 23-79. Neid Mu and CES of the Cyper — nioideae. Pp. 201-228 in J. E. da (e PES ‘Current Pie in Plant Science. Academic Press, 1969b. The (eee of tropical America: Volume 75, Number 3 Davidse & Kral 861 1988 New Venezuelan Calyptrocarya some new or critical species. Jap. J. Bot. 20: 123- Nees AB ESENBECK, C. G. 1842. Cyperaceae. In: C. F. 134. Martius (editor), Flora Brasiliensis 2(1): 1-226. 1971. Systematic interrelationships among THoMas, W. W. 1984. The systematics of Rhynchos- Sclerieae, Lagenocarpeae and Mapanieae (Cypera- pora section Dichromena. Mem. New York Bot. ceae). Mitt. Bot. Staatssamml. München 10: 604- Gard. 37: 1-116. 617. NIGERIAN SOLANUM SPECIES OF ECONOMIC IMPORTANCE Z. O. Gbile' and S. K. Adesina? Abstract any Solanum species that occur in Nigeria are species especially serve sources of food and medicinal products. as sources of edible fruits and v vegetables. Many diosgenin and solasodine, chemicals of great importance in the steroid ae The domesticated saa species remain good sources of The genus Solanum is represented by some 25 species in Nigeria, including five introductions: S. mammosum, S. tuberosum, S. melongena, S. wrightii, and S. seaforthianum var. disjunctum (Gbile, 1987). Solanum macrocarpon, S. aethio- picum, S. scabrum, S. melongena, S. gilo, S. indicum, S. anomalum, S. americanum, S. ni- grum, and others are domesticated, and their leaves or fruits or both are eaten as vegetables and used in traditional medicine. Many other Solanum species grow wild and are less known or used. Chemical information on the Nigerian Solanum species is scanty and it is difficult to assess the values of these species in this regard. The present article reports on the protein con- tent of some domesticated Solanum species and also reviews the economic importance of the Ni- gerian Solanum. SOLANUM SPECIES AS Foop PLANTS The common Solanum species that are used for food include S. tuberosum, “Irish potato," which grows well in the highlands; S. melongena, “au- bergine” or “eggplant,” which flourishes in the lowlands; S. americanum; S. nigrum; and the ad- ditional species listed in Table 1. Solanum anom- alum, S. gilo, and S. melongena provide edible fruits. Fruits and leaves of S. aethiopicum and S. macrocarpon are edible, and only the leaves of S. S. nigrum, S. americanum are eaten as vegetables. While the fruits of 5. aethiopicum and 5. gilo are usually eaten raw or are steamed before eating, the leaves of 5. aethiopicum, S. ameri- scabrum, canum, S. macrocarpon, S. nigrum, and S. sca- brum are usually boiled. The relative bitterness of the leaves and fruits dictates to a great extent which is edible vs. poisonous. Bitterness has been attrib- uted to steroidal alkaloids in these plant parts. DETERMINATION OF PROTEIN AND ASH (Mc, P) iN SOLANUM EDIBLE SPECIES MATERIALS AND METHODS The Solanum species were interplanted in the same experimental plot behind the Forestry Re- search Herbarium, in Ibadan. After fruiting, each species was sampled and separated into edible leaves and fruits. Plant materials were dried in an oven to constant weight at 60°-70° C and ground to pass through a 1-mm sieve in a Thomas-Wiley meal in preparation for chemical analyses. All de- terminations were prepared in duplicates. Nitrogen was determined by the semi-micro Kjeldahl procedure using selenium as catalyst. Per- centage of crude protein content was obtained by multiplying N, content by 6.25. In the determi- nation of phosphorus and magnesium, samples of 0.5 g were digested using a mixture of nitric acid and perchloric acid. Phosphorus was determined by colorimetry using vanadomolybdate yellow color development, mined colorimetrically by the titan-yellow method. The results are detailed in Table 1 while magnesium was also deter- SOLANUM SPECIES AS MEDICINAL AGENTS Many Solanum species are used in indigenous medicine to counter ailments as listed in Table 2. Many of these species are employed as tonics, antirheumatics, remedies for colds, fevers, and diz- ziness, and are eaten as vegetables for their high nutritive values or as potherb as mild anticonvul- sants. Modern research has shown that some Sola- ' Forestry Research Institute of Nigeria, Private Bag 5054, Ibadan, Nigeria * Drug Research and Production Unit, Faculty of Pharmacy, Obafemi ps University, Ile-Ife, Nigeria. ANN. Missouni Bor. Garb. 75: 862-865. 1988. Volume 75, Number 3 1988 Gbile & Adesina 863 Nigerian Solanum TABLE l. Protein and ash content of some edible Solanum species. % Crude Species Organ Protein % P % Mg Solanum macrocarpon L. fruit 1.4 0.25 0.12 S. macrocarpo leaf 2.4 0.44 0.40 S. aethiopicum L fruit 1.6 0.38 0.26 S. aethiopicum leaf 3.2 0.37 0.38 S. scabrum Miller fruit 1.8 0.36 0.16 S. scabrum leaf 2.9 0.40 0.44 S. melongena L. fruit 1.6 0.25 0.08 S. gilo Raddi conical fruit 1.2 0.47 0.17 S. gilo spherical fruit 1.3 0.45 0.22 S. indicum L. subsp. distichum Thonn. fruit 1.4 0.46 0.27 num species have antiviral, anticancer, anticon- vulsant, and anti-infective agents. Antiviral activity has been demonstrated in ex- tracts of Solanum melongena, S. nigrum, and S. tuberosum (Roychoudhury, 1980). Weak anticon- vulsant activity has also been demonstrated in ex- tracts of S. dasyphyllum fruit (Adesina, 1985), S. aethiopicum leaf, S. americanum leaf and un- ripe fruit, S. melongena root, and S. scabrum leaf and fruit (Adesina et al., 1985). Besides, all the extracts examined for anticonvulsant activity ex- hibited. an interference on the functions of the CNS to gr nticonvulsant activity has been rated E the presence and concentration of sco- poletin and related coumarins found present in most Solanum species examined (Adesina et al., 1985). The anticonvulsant, sedative, hypotensive, and an- tipyretic properties of scopoletin and scoparone have been reported before by many workers (Jam- wal et al., 1972; Adesina et al., 1981; Ojewole $ Adesina, 1983a, b; Adesina et al., 1985). Chemical and biological work on immature ber- ries of Solanum nigrum showed that the berries possess anticancer activity. 3-O--lyco-tetraoside, desgalactotigonin, and solamargine isolated from the berries showed inhibitory activity against JTC- 26 (100, 97.9, tion of 15 ug/ml) (Saijo et al., been shown that the crude alkaloid fraction isolated from the leaves of Solanum melongena exhibited significant analgesic effect and some CNS depres- sion in mice but no anticonvulsant action (Vohora et al., 1984). This effect was also noted for S. scabrum alkaloidal fraction (Adesina & Gbile, 1984). Molluscicidal activity was examined in some Solanum species. All the parts of S. americanum were found toxic to Biomphalaria glabrata and B. globosus used as test snails and could possibly be used to check schistosomiasis. SOLANUM SPECIES AS SOURCES OF PHARMACEUTICALLY IMPORTANT CHEMICALS Some Solanum species have recently assumed great importance as rich sources of precursors of steroid drugs. Steroidal raw materials have been found useful in cardiovascular therapy, as human abortifacients, as anti-inflammatory agents, and as menopause regulants and are now known to influ- ence the CNS. Many researchers have investigated Solanum species for their steroidal sapogenin and alkaloid content with a view to determining the quantities of these compounds. Indrayanto et al. (1985) recently examined the fruit of Solanum wrightii chemically for its sola- sodine content. Pharmaceutically important compounds dios- genin and solasodine were isolated from the tissue samples of Solanum verbascifolium in appreciable amounts (Jain & Sahoo, 1981a, b). The leaf was found to contain solasodine (0.26%), tomatidine (0.05%), solaverbascine (0.01%), progesterone (0.001%), 16-pregnenolone, and other compounds (Adam et al., 1979, 1980). Telek (1979) found a very good yield of crude solasodine, suitable for the commercial synthesis of 36-acetoxy-5, 16- pregnadiene-20-one, in S. mammosum. Studies on Solanum nigrum berries by Bose & Ghosh (1980) revealed that solasodine content of the berries varies from 5-6% in ripe berries to 4— 5% in unripe berries and that this could be exploited or commercial synthesis of new drugs. Tigogenin and diosgenin have also been reported from the plant. The unripe fruit of S. incanum, on exami- nation by Segal et al. (1977), led to the identifi- 864 Annals of th Missouri Botanical Garden TABLE 2. Some medicinal and food uses of Solanum species. Species Medicinal and Food Values Solanum aculeatissimum Jacq. S. aethiopicum S. americanum (L.) Jacq. Fruit used in enema, constipation Ripe fruits edible raw or when cooked, fruits remedy for colic and flatulence; therb Fruit and leaf used as digestive tonic, diuretic, depurative, and antiparasitic; plant has high nutritive values and eaten as vegetable or in soup after cooking; whole plant used to remove dizziness due to epilepsy and other S. anomalum Thonn. S. erianthum G. Don S. gilo isorders Serves as vegetable, laxative, and treatment of ear sores and infections oots and fruits deliriant, purge, diuretic, and cholagogue Restorative, fruit eaten raw as vegetable, has high nutritive values; remedy for fevers and dizziness, weak anticonvulsant S. incanum L. S. macrocarpon S. melongena Used to treat syphillis, fruit for patients with hi Bitter fruit edible when cooked; plant cultivated as potherb, fruit and leaf soups and s h blood pressure Root and boiled fruit se as antirheumatic, digestive tonic, and for veteri- nary purposes; all three varieties of fruit with high nutritive values as veg- etable when cooked; plant used for various skin diseases and infections and to relieve excitement in nervous diseas S. nigrum L. Anticonvulsant, African remedy for alarin, fever, dysentery; antispasmodic, oki diaphoretic, and m ripe fruit and leaf eaten after coo d as tive tonic; who S. scabrum L. nt used as medicine for eye, heart, and live Anticonvulsant, digestive tonic, leaf boiled and eaten as vegetable; jT high utritive values. Whole plant sedative, depressant, anticonvulsant and anti- S. torvum Sw. parasitic Ripe Suis edible (eaten in India) and used DE for liver and spleen complaints; fruits expectorant and sedativ cation of diosgenin and yamogenin in fairly large concentrations suitable for chemical development. Solanum macrocarpon furnished solasodine, to- matidine, diosgenin, and sitosterol on chemical hy- drolysis. Recent chemical examination of some oth- er Nigerian Solanum by Adesina & Gbile (1984) and Adesina (1985) revealed large amounts of so- lasodine, diosgenin, and tomatidenol in S. scabrum and S. dasyphyllum fruits. CONCLUSION From Table 1, it can be seen that the edible tissues examined contain a high percentage of pro- tein and remain a good source of ash for use as vegetables. Wild Solanum species are less studied than do- mesticated species. Despite this, the Nigerian Sola- num remains a good source of pharmaceutically important chemicals and of vegetable for the teem- ing population. LITERATURE CITED ADAM, G., H. T. Huong € N. H. Kuori. 1979. o stituents of the Maru nai drug S. verbasci- folium L. Pl. Med. 36: ,————— € ——. 1980. Solaverbascine — a new 22,26-epiminocholestane pe from S. ver- abico. y edes Seid 19: 1002-1003 ADESINA, S. K. nstituents of Solanum dasy- phyllum a Lords 48: 147. & Z. O. GBILE 1984. Steroidal constituents of S. a abrum subsp. nigericum. Fitoterapia 55: 362-3 FASANMADE & I. M. ONONIWU. 1985. FS à Chemical investigation of some Nigerian io Pp. 65-70 in Ess e Plants. Drug Research and Production Unit, Uni- versity of Ife. —T J. A. O. Owo & V. O. Marquis. 1981. On the molluscicidal and anticonvulsant properties of He fruit of T. tetraptera Taub. J. African Med. Pl. 4: 27-31. Bose, B. & C. GHosH. 1980. Studies on the variation eld 5 INDRAYANTO, G. N. Cu S & WAHYUDI. e of fruit size of Solanun wrightii on its so- la sodine content. Pl. : 470. Jain, SS C. € S. L. SAHOO. 1981a . Solanum verbas- cifolium suspension cultures. Pharmazie 36: 714- 715. & 198 1b. Isolation and character- Volume 75, Number 3 1988 Gbile & Adesina Nigerian Solanum 865 ization of steroidal sapogenins and glycoalkaloids from dap verbascifolium L. Chem. Pharm. Bull. 29: 1765-1767. JAMWAL, K. S., M. L. SHARMA, N. CHANDHOKE & B. J. 1972. es action of sco- ria Waldst. and Kit. : 763. a ADESINA. 1983a. Cardio- vascular and caca: actions of scopoletin iso- lated from the fruit of T. tetraptera Taub. Pl. Med 3b. Mechanism of the h tensive ga of scopoletin isolated from T. tetraptera uit. ay aca R. 1980. Effect of extracts of certain solanaceous plants on plant virus infection. Acta Bot. Indica 8: 91-94. Saio, R., K. MURAKAMI, T. NOHARa, T. TOMIMATSU, A. Sato & K. MaTsuoKA. 1982. Studies on the con- stituents of Solanum plants. II. On the constituents of the immature berries of S. nigrum L. Yakugaku Zasshi 102: 300-305. SEGAL, R., I. MILO-GOLDZWEIG & D. V. ZarrscHEK. 1977. Diosgenin and yamogenin from S. incanum. Lloydia : 604 TELEK, L. 1979. Preparation of solasodine from the fruits of Solanum species. Pl. Med. 37: 92-94 VonoRA, S. B., K. Isuwan & M. S. Y. KHAN. 1984. Effect of alkaloids of Solanum melongena on the CNS. J. Ethnopharmacol. 11: 331-336. CHROMOSOME NUMBERS OF GRASSES (POACEAE) FROM SOUTHERN AFRICA. I.' Takuji Hoshino? and Gerrit Davidse? ABSTRACT Chromosome numbers and meiotic behavior are p à oe and 55 South African grass collections representing 30 genera and 45 species. First chromo Andropogon amethystinus, n = ca. 30; Anthephora argentea, n = - 9. nts are a for the following 13 species: aria chusqueoides, n = 9; B. glomerata, ra n = 9; Centropodia glauca, n = 24; Danthoniopsis parva, n = 12; o arena n= 18; Ehrharta longigluma, LE n = /2; Miscanthidium capense, n = 15; T. ramosissima, n = 10. Chromosome counts Panicum monticolum, n — differing from . schinzii, n = 9; Triraphis fleckii, n = any nou reported numbers were obtained id. for six species. Of the 45 species, 57% are polyploid and 43% are diploi This paper is part of a series contributing to a broader knowledge of chromosome numbers of Af- rican grasses. In the first we reported new chro- mosome counts for Zimbabwean grasses (Davidse et al., 1986). In this report we present chromosome counts for 63 collections representing 45 species and 30 genera of South African and Namibian grasses. The major studies dealing with chromosome numbers of South African grasses are those of Moffett & Hurcombe (1949), Pienaar (1955), De Wet (1954a, b, 1958, 1960), De Wet € Anderson (1956), Spies & Du Plessis (1986a, b, 1987a, b, 1988), and Spies & Jonker (1987), although other smaller scattered reports, mostly dealing with in- dividual genera, have also been made MATERIALS AND METHODS All cytological samples studied were collected and fixed in the field January to March 1974. The methodology is the same as explained in Davidse et al. (1986). Voucher specimens (Table 1) are deposited at MO and PRE. The suprageneric clas- sification used in this paper follows the one of Clayton & Renvoize (1986) except that we rec- ognize the tribe Brachypodieae. RESULTS AND DISCUSSION A complete list of the species studied, their chro- mosome numbers, the generic base number derived from the determined number, and the voucher specimens is given in Table 1, where totally new counts and counts differing from any previous count for the same taxon are also identified. We illustrate only new counts (Figs. 1-4, 7, 9, 10, 15-20) and counts different from any other for a given taxon (Figs. 5, 6, 8, 11, 13, 14). Unless otherwise in- dicated, meiosis was regular for all taxa listed in Table 1. Comments on chromosome or base num- bers without reference to original sources are based on the indices of Fedorov (1969), Moore (1973, 1974, 1975), and Goldblatt (1981, 1984, 1985). TRIBE ANDROPOGONEAE Diheteropogon amplectens was previously re- ported to be tetraploid 2n — 40 from a Zimbabwean population (Moffett & Hurcombe, 1949). We found two collections from the Transvaal to be diploid with n — 10 (Fig Heteropogon melanocarpus was previously known only as an aneuploid (2n = 22) from Zim- babwe (Moffett & Hurcombe, 1949). Our count establishes the existence of a eudiploid population (n = 10; Fig. 14) in the Transvaal. We determined both Miscanthidium capense (Fig. 15) and M. junceum to have n = 15. This confirms earlier counts for M. junceum (De Wet & Anderson, 1956; De Wet, 1958; De Wet, 1960, as M. teretifolium). In addition, Brett (1954) re- ported M. violaceum to have 2n — 28. This strong- ' Fieldwork by Davidse pie supported by NSF grant GB 40630. Seed s work in de Hu: was supported by the Foundation of Private S fieldwork, particularly Bernard De Winter, ey of the determinations of the voucher specim ? Biological Laboratory, Okayam school Personnel (Japan) and by Dr. Peter H. Raven. We of the Botanical Research Institute, Pretoria, a bela and logistic al support that Pat the ^. Ellis, and Al Loxton. I thank Lynn Fish for Miele some bi AE kas to the staff ss of the ma Hades of Science, Ridai-cho, Okayama, 700, Japan. A. ory, 3 Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166, U.S. ANN. MISSOURI Bor. GARD. 75: 866-873. 1988. Volume 75, Number 3 1988 Hoshino & Davidse 867 Chromosome Numbers of Southern African Grasses TABLE 1. African) grasses. Chromosome numbers of South African (without country designation) and Namibian (South West Taxon Locality and Voucher ANDROPOGONEAE Andropo, Ws ode Steud. Cymbopogon pog excavatus (Hochst.) Stapf ex Burtt Davy validus mn Stapf ex Burtt Dav Diheteropogon amplectens (Nees) Clayton Eulalia villosa (Thunb.) Nees Heteropogon tate (Ell.) Benth. Hyparrhenia hirta (L.) Stapf Ischaem afrum ( ni F. Gmel.) Dandy Miscanthidium capense (Nees) Stapf junceum (Stapf) Stapf onocymbium ceresiiforme (Nees) Stapf ARUNDINEAE Centropod glauca (Nees) T. A. Cope ARUNDINELLEAE Danthoniop parva i B. Phipps) Clayton AVENEAE Agrostis lachnantha Nees Helictotrichon turgidulum (Stapf) Schweick. x=] n = ca. 30* x= 10 n = 10 n=10 x=10 n= 10 x=10 n=10 x=10 n= 10! x=10 n= 20 x=] n = 1 x= 15 n= 15: n= 15 x= 10 n = 10 x= 12 n= 24 x= 12 n = 12: x= n = 21° x= 7 n= 14 Orange Free State: 33 km SW of Witsieshoek, Davidse 6993 Transvaal: 5 km NE of Haenertsburg, Davidse & Ellis 5839 Transvaal: 5 km NE of Haenertsburg, Davidse & Ellis 5840 Transvaal: 0.5 km NE of Haenertsburg, Davidse & El- lis 5832; 15 km NE of Cullinan, Davidse 6005 Natal: 33 km S of Nqutu, Davidse 6852. Transvaal: 5 km NE of Haenertsburg, Davidse & Ellis 5836 Transvaal: Kruger National Park, Dzundwini Hills, 20 m N of Babalala, Davidse 5853 Natal: Belelasberg, 6 km S of Wakkerstroom, Davidse 758 Transvaal: Kruger National Park, Babalala, Davidse Natal: 7 km N of Kranskop, Davidse 6923 Natal: 11 km NW of Utrecht, Davidse 6803 Transvaal: Magoebaskloof, 3 km NE of Haenertsburg, Davidse & Ellis 5812 Cape Province: 59 km W of Olifantshoek, Davidse & Loxton 6436. Namibia: Gibeon District, 41 km E of Gochas, Davidse & Loxton 6367 Transvaal: Zoutpansberg, Davidse & Ellis 5930 pay 2.5 km NW of Wakkerstroom, Davidse 42. Orange Free State: 33 km SW of Witsie- jou) Davidse 6988 Orange Free State: 33 km SW of Witsieshoek, Davidse 6967 868 Annals of the Missouri Botanical Garden TABLE l. Continued. Taxon Chromo some (n) and Generic Base Number (x) Locality and Voucher Koeleria capensis (Steud.) Nees BRACHY PODIEAE Brachypodium flexum Nees EHRHARTEAE Ehrharta erecta Lam. longigluma C. E. Hubb. ERAGROSTIDEAE Triraphis fleckii Hack. ramosissima Hack. ORYZEAE Leersia hexandra Swartz Prosphytochlo prehensilis (Nees) Schweick. PANICEAE Alloteropsis semialata (R. Br.) Hitchc. Anthephora argentea Goossens Brachiaria chusqueoides (Hack.) Clayton deflexa (Schumach.) C. E. Hubb. ex Robyns eruciformis (J. E. Sm.) Griseb. glomerata (Hack.) A. Camus nigropedata (Munro ex Fical. & Hiern) Stapf x= n= x=9 n=9 x=] n=l n = 12: x= 10 n = 10 n = 10: x= 12 n= 24 x= 12 n=12 x=9 n=9 x=9 n= 9 x=9 n= 9 n=9 n=9 n= 9 n= 36 Orange Free State: 33 km SW of Witsieshoek, Davidse ies 6 km NE of Haenertsburg, Davidse & Ellis atal: Belelasberg, 6 km S of Wakkerstroom, ous 6787 Transvaal: Magoebaskloof, 3 km NE of Haenertsburg, Davidse & Ellis 5810 Orange Free State: 33 km SW of Witsieshoek, Davidse 6974 Namibia: Gibeon District, 74 km E of Gochas, Davidse & Loxton 6 Namibia: Keetmanshoop District, 2-3 km E of Groot Karasberge, Davidse & Loxton 6252; Warmbad District, 36 km W of Ariamsvlei, Davidse & Loxton 6416 Transvaal: 11 km WSW of Koster, Davidse & Loxton 6012; 2 km S of Vanderyst, Davidse 6691 Transvaal: Magoebaskloof, 3 km NE of Haenertsburg, Davidse & Ellis 5811. Natal: 42 km S of Silut- shana, Davidse 6898 Transvaal: paced 3 km NE of Haenertsburg, Davidse & Ellis 5 Cape Province: 34.4 km NE of Kuruman, Davidse & Loxton 6063 Natal: Tinley Manor Beach, 55 km NE of Durban, Da- vidse 6938 Transvaal: Kruger National Park, 8 km N of Babalala, Davidse & Ellis 5847 Transvaal: 27 km SE of Bethal, Davidse 6708 Cape Province: 21 km WSW of Keimoes, Davidse & Loxton 6124. Namibia: Gibeon District, 25 km E of Gochas, Davidse & Loxton 6358 Transvaal: Kruger National Park, Dzundwini Hills, 20 km N of Babalala, Davidse & Ellis 5852 Volume 75, Number 3 1988 Hoshino & Davidse 869 Chromosome Numbers of Southern African Grasses TABLE 1. Continued. Chro some (n) and Generic Base Taxon Number (x) Locality and Voucher Digita x= 9 een (Nees) Stapf n= 18: Natal: Tinley Manor Beach, 55 km NE of Durban, Da- vidse eriantha Steud. n=9 Cape Province: 25 km SW of Olifantshoek, Davidse & Loxton 6102. Namibia: Keetmanshoop District, 2-3 km E of Groot Karasberge, Davidse & Loxton 6279 n= 18 Cape Province: 75 km SW of Vryburg, Davidse & Loxton 6040 longiflora (Retz.) Pers. n=9 Natal: 3 km S of Kingsley, Davidse 6839. Transvaal: Kruger National Park, 12 km NW of Punda Milia, Davidse & Ellis 5924 ternata (A. Rich.) Stapf n — 18 Transvaal: 27 km SE of Bethal, Davidse 6712 Echinochloa x=9 haploclada (Stapf) Stapf n=9 Transvaal: Kruger National Park, 14 km SE of Punda Milia, Davidse & Ellis 5856 n= 27° Transvaal: Kruger National Park, Machayi Pan, Da- vidse & Ellis 5869; Kruger National Park, 14 km SE of Punda Milia, Davidse & Ellis 5857 Panicum x=9 coloratum L. var. coloratum n=9 Namibia: Keetmanshoop District, 2-3 km E of Groot Karasberge, Davidse & Loxton 6226 n=18 Namibia: Gibeon District, pe River, 100 km E of Gochas, Davidse & Loxto maximum Jacq. n=16 p Kruger National ofa Dzundwini Hills, 20 of Babalala, Davidse & Ellis 5854 monticolum Hook f. n= 27" Eu Woodbush Forest Reserve, Davidse & Ellis 5826 schinzii Hack. n= 9 Transvaal: 2 km S of Vanderyst, Davidse 6692; 2 km NE of Haenertsburg, Davidse & Ellis 5820; 5 km of Morgenzon, Davidse 6719 stapfianum Fourc. n = 9° Cape Province: 75 km SW of Vryburg, Davidse & Loxton 6043 Paspalidiu x=9 ea (Del.) Simpson n=18 Transvaal: Kruger National Park, Machayi Pan, Da- vidse & Ellis 5868 Pennisetum x=9 villosum R. Br. ex Fresen n= 27 Transvaal: just SE of Amersfoort, Davidse 6731 Rhynchelytru =9 nerviglume (Franch. ) Chiov. n=18 Natal: 11 km NW of Utrecht, Davidse 6806; 33 km S of Nqutu, Davidse 6861 Setaria x29 megaphylla (Steud.) Dur. & n=27 Transvaal: 2 km NE of Haenertsburg, Davidse & Ellis Schinz 5817 Urochloa x=9 panicoides Beauv. n= 18 Natal: 39 km S of Utrecht, Davidse 6825 * First chromosome count for the species. ` Chromosome count differing from any previous count for the species. 870 Annals of the Missouri Botanical Garden FIGURES 1-11. Camera lucida drawings of meiotic chromosomes of South African grasses. —1. Andropogon amethystinus, n = ca. 30, diakinesis.—: entropodia glauca, n = 24, diakinesis.—: Triraphis fleckii, n = 10, diakinesis.—4. Triraphis ramosissima, n = 10, diakinesis. —5. Agrostis lachnantha, n = 21, metaphase L — 6. Brachiaria nigropedata, n = 36, diakinesis.— 7. Digitaria diversinervis, n = 18, diakinesis.— 8. Echinochloa haploclada, n = 27, diakinesis. —9. Panicum monticolum, n = 27, diakinesis. — 10. Panicum schinzii, n = 9, diakinesis.— 1 1. Panicum stapfianum, n = 9, diakinesis. Scale line = 10 um. > Ficures 12-20. Photomicrographs of meiotic chromosomes of South African grasses.— 12. Cymbopogon excavatus, n= 10, diakinesis.— 13. Diheteropogon amplectens, n = 10, diakinesis.— 14. Heteropogon melanocarpus, n = 10, diakinesis.— 15. Miscanthidium capense, n = 15, diakinesis. — 16. Danthoniopsis parva, n — 12, diaki- Volume 75, Number 3 Hoshino & Davidse 871 1988 Chromosome Numbers of Southern African Grasses 18 — 49 Ex 20 nesis.— 17. Ehrharta longigluma, n = 12, metaphase 1.—18. Anthephora argentea, n = 9, diakinesis. — 19. Brachiaria chusqueoides, n = 9, diakinesis. — 20. Brachiaria glomerata, n — 9, diakinesis. Scale lines = 10 um. 872 Annals of the Missouri Botanical Garden ly indicates that the genus has a base number of — 15, which itself was probably derived by poly- ploidization from x — 5, the base number for the tribe (Clayton & Renvoize, 1986). The number in M. violaceum, if it can be confirmed, was probably derived by secondary aneuploidy from n — 15. also gives support for the continued recognition of this genus from the related Miscanthus, which has x = 19 (Clayton & Renvoize, 1986) TRIBE ARUNDINEAE The n = 24 (Fig. 2) count for one population of Centropodia glauca is consistent with the 2n = 24 reported by De Wet (1954a) and Sokolovskaya & Probatova (1978) for C. forskalii (Vahl) Trin., as well as with the prevalent base number x = 6 for the tribe (Davidse, 1988). TRIBE AVENEAE De Wet (1958) reported a Transvaal population of Agrostis lachnantha to be tetraploid with 2n = 28, but our sample had n = 21 (Fig. 5) and is thus hexaploid. Meiosis in Helictotrichon turgi- dulum (n = 14) was slightly irregular with the common occurrence of a single quadrivalent. All other chromosomes paired as bivalents. TRIBE PANICEAE Brachiaria nigropedata has been reported as diploid from South Africa (De Wet, 1954b; De Wet & Anderson, 1956) and tetraploid from Zim- babwe (Moffett & Hurcombe, 1949). We now add an octoploid count (n = 36) based on our analysis of a Transvaal population (Fig. 6). Echinochloa haploclada has up to now been known as a diploid with 2n = 18 from Tanzania (Tateoka, 1965) and as diploid (Malik & Tripathi, 1969) and tetraploid (2n = 36) from Kenya (Ya- buno, 1966). We confirmed the diploid number for a Transvaal population and also found two nearby populations to be hexaploid (n = 27; Fig. 8). As presently circumscribed, E. haploclada is morphologically variable, and broadly based cy- totaxonomic studies may help interpret this vari- ation. Since diploids have never been found outside Africa in Echinochloa, Yabuno (1973) considered Africa to be one of the centers of origin for the genus. Our results strengthen this interpretation. Spies & Du Plessis (1988) reported Panicum stapfianum as tetraploid (n — lation in the southern Cape Province, whereas we determined a northern population to be diploid (n — 9; Fig. 11). 18) from a popu- Pennisetum villosum, a native of northern Af- rica, now widely naturalized in the tropics and subtropics, has been reported as a eudiploid to euhexaploid. Bridges and fragments were observed at anaphase I in the hexaploid (n = 27) plant that we examined. The occurrence of triploids, penta- ploids, and hexaploids with irregular meiosis sug- gests the likelihood of apomixis in this species. CONCLUSIONS The basic chromosome numbers calculated for all the genera sampled in this study agree with those previously reported. Aneuploid numbers turned up in seven species. Based on this report and chromosome numbers previously published for South African grasses, ploidy levels were determined for all the species included in this study from any part of their dis- tributional range. On this basis 24 species (53%) of the 45 are polyploid in some part of their range. This is somewhat on the low side for grasses in general since most estimates for polyploidy among grasses are higher than 60% (Davidse et al., 1986). Analyzing this further, 47% of the species we stud- ied are known only as diploids, 24% only as poly- ploids, and 29% as both diploids and polyploids. Although this sample is small (7% of the 895 species listed by Gibbs Russell et al., 1985) and may therefore not be very representative of the southern African grass flora, the percentages of species only known as diploids and only as poly- ploids are the reverse of that found in the Zim- babwean grass flora (Davidse et al., 1986). Wheth- er this represents a real geographical trend or is random variation awaits further intensive sampling of the rich African grass flora. LITERATURE CITED Brett, P. G. C. 1954. Saccharum- Miscanthidium hybrids. J. Genet. 52: 542-546. DH. W. D. & S. A. RENvoizE. 1986. Genera Graminum. Kew Bull., Add. Ser. 12: 1-38 DavipsE, G. . A revision of the genus Pr. rionan- thium (Poaceae: Arundineae). Bothalia 18 (in press). I & B. K. Simon. 1986. mhahu (Po an analysis of polyploidy E the grass fore. of Tan babwe. S. African J. Bot. 52: 521-528. De Wer, J. M. J. "19542. The genus oo in grass phylogeny. Amer. J. Bot. 41: 204-21 Chromosome numbers of a dm South African grasses. Cytologia 19: 97-103. 1958. Additional ap pee numbers in animal grasses. Cytologia 23: 11 18. 96 hromosome numbers pie some mor- phological attributes of various South African grasses. Amer. : ; J. Bot. 47: 44-50 Volume 75, Number 3 1988 Hoshino & Davidse Chromosome Numbers of Southern African Grasses & L. J. ANDERSON. 1956. bd aps num- bers in DujARDIN, M. mber "a some tropical African grasses from western Zaire. Canad. Bot. 56: 2138-2152. Feporov, A. A. 1969. Chromosome Numbers of Flow- a Plants. V. L. Komarov Patani] Institute, Len- rad. Giles | in G. E., C. Rep, J. Van Rooy & L. SMook. 19 ist of o of southern African plants, i edition, part 1. Mem. Bot. Surv. S. Africa 51 -152. D pa P. 1981. Index to plant chromosome num- bers 1975-1978. Monogr. Syst. Bot. Missouri Bot. 5: 1-553. ard. - 1984. Index to plant chromosome numbers 1979- 1981. Monogr. Syst. Bot. Missouri Bot. Gard. 8: 1-427. 1985. Index to plant chromosome numbers up 1983. Monogr. Syst. Bot. Missouri Bot. Gard. vp "T C. E. & R. C. TRIPATHI. 1969. Cytology of some Echinochloa species. Chromosome Inform. Serv 10: 9-1 MOFFETT, A. A. & R. HurcomBE. 1949. Chromosome rs of South African grasses. Heredity 3: 369- 3. E, R.J. 1973. Index to plant "e er num- Fie ers 1967-1971. Regnum Veg. 90: 1-539. 1974. Index to m onibus numbers 1972. Regnum Veg. 91: 08. — 975. Index to plant chromosome numbers 1973- 1974. Regnum Veg. 96: 1-257. PIENAAR, R. DE V. 1955. The chromosome numbers of some indigenous South African and introduced Gra mineae. Pp. 551-570 in C. Meredith (editor), The Grasses and Pastures of South Africa. Central News Agency. SokoLovskAYA, A. P. & N.S. PROBATOVA. 1978. Chro- mosome numbers of some grasses (Poaceae) of the XE flora. II. Bot. Zhurn. SSSR 63: 1247- SPIES, J. E H. Du PLessis. 1986a. Chromosome studies on African plants. 1. Bothalia 16: 87-88. 1986b. Chromosome studies on Bothalia 16: 269-270 1987a. ea studies on Bothalia 17: 131-135. 198 ae studies on Bothalia 17 (in press). 1988. omosome studies on African plants. 6. Bothalia 18 (in press). A. JONKER. 1987. Chromosome studies on African plants. 4. Bothalia 17: 135-136. TATEOKA, T. 1965. Chromosome numbers of some East African grasses. Amer. J. Bot. 52: 864-869. YABUNO, T. 1966. Biosystematic study of the genus Echinochloa. Jap. J. Bot. 19: 277-323. 13. des gun p between some ochloa spec Ken e haploclada men Stapf. ni ha "38: 131- 35. African plants. 2. African plants. 3. African plants. 5. A PRELIMINARY LIST OF THE MOIST FOREST ANGIOSPERM FLORA OF MWANIHANA FOREST RESERVE, TANZANIA! Jon C. Lovett,? Diane M. Bridson,’ and Duncan W. Thomas: ABSTRACT A preliminary d of the Mwanihana Forest Reserve, Tanzania, has yielded 440 angiosperm species. These tically, and by habitat and ha are listed system ibit. The forests are rich in species of restricted distribution, and have affinities to de Guineo-Congolian forests of western Africa. The Mwanihana Forest Reserve has been the focus of biological study since the discovery of a new taxon of the primate Cercocebus galeritus there in 1979 (Homewood & Rodgers, 1981), which prompted a proposal for the area to be given na- tional park status (Rodgers & Homewood, 1982a). In the course of further study, a new species of sunbird (Nectarinia rufipennis) has been recorded from the forest (Jensen, 1983), and more than 40 ew species of plants have been found. These in- clude Seychellaria africana, which represents the first record of the family Triuridaceae from East Africa, and new species in the genera Uvariopsis and Omphalocarpum, which were not previously known from eastern Africa. The following account is the result of botanical exploration of the forest in 1984 as part of a National Geographic Society expedition to the area in collaboration with the World Wildlife Fund Tan- zania Forest Habitat Evaluation Project. Other as- pects of the botany of the forest have been treated elsewhere (Lovett & Thomas, 1986). The main area of study was the forest on the steep east-facing escarpment slopes above Sanje village at latitude 7?50'S and longitude 36°55’E. These slopes are the eastern edge of the Gologolo Mountains, which are the northern end of the Uzungwa range. In the study area the continuous altitudinal range of the forest is from 450 m to 1,760 m, but it is greater elsewhere on the es- carpment. The rainfall is approximately 2,000- 2,500 mm a year, with one wet season from No- vember to May, when on average more than 100 mm of rain a month is received. There is a four- month dry season from July to October, when each month receives 50 mm of rain on average, an months of no rain are frequent. At higher altitudes there is a rich epiphytic bryophyte cover, indicating that a mist effect may be important in the water balance. The escarpment is dissected by steep-sided valleys separated by sharp ridge tops. The ridge tops are exposed, and so species normally occurring outside the forest habitat can be found within it. Some of the valley bottoms are flat, though nowhere particularly broad, and it is here that the forest is best developed. Elephant damage is common in the higher altitudes, and buffaloes are also present. Timber is being extracted by pit sawing, and in some areas there is considerable disturbance. The angiosperm species recorded in the forest are arranged below according to habitat, habit, and in a systematic list. The habitat is divided into four types. The first three—Zanzibar- Inhambane Low- Forest, Zanzibar-Inhambane Transitional Forest, and Afromontane Rain Forest — correspond to White's (1983) phytogeographic vegetation classification. The fourth division, Ridge Top For- est, is a specific vegetation type of restricted extent, ' We gratefully acknowledge the assistance of the staff at the Royal Botanic Gardens collections. The Tanzanian National Scier very kin gave us permission to wor Mwanihana For t Reserve. was supported by the National ple, a ea 2 and World Wildlife Fund. Langson Kusoma and Henry assisted in the field. den, P.O. U.S.A. Pre 2 Missouri Botani sarden, P.O. Box 299, St. L Louis, Missouri 63166-0299, any, elo of Dar es Salaam, P.O. ie sent address: De- 35060, Aa es Salaam, Tanzania m on an ' Royal Botanic Gardens, Kew, Richmond, Surrey TW9 ! Miri Botanical Garden, P.O. Box 299, St. Louis, us Lu 0299, U.S.A. ANN. MISSOURI Bor. GARD. 75: 874-885. 1988. Volume 75, Number 3 1988 Lovett et al. 875 Mwanihana Forest Angiosperm Flora being the highest ridge investigated in this study. This habitat is treated separately, as there are a number of species found only there, and it indicates which species of the Afromontane Rain Forest tend towards areas of greater exposure. The systematic list is divided into dicotyledons and monocotyle- dons, with families arranged alphabetically. Species and genera are arranged alphabetically within the families. The list of species is far from complete, partic- ularly for herbaceous and shrubby species. Many of the large trees were identified from sterile ma- terial gathered during the course of a quantitative ecological survey, and the determinations are in- cluded here because the canopy species are not often collected. SPECIES ARRANGED ACCORDING TO HABITAT AND HABIT The habitats follow the phytogeographical def- initions given by White (1983) with the exception of Ridge Top Forest. Each habitat is given an altitudinal range and a short description of forest structure. The habit of each species is self-explan- atory, except that it should be noted that a tree is a single-stemmed woody plant 10 m tall or at least 20 cm in diameter at breast height. The species are arranged alphabetically. ZANZIBAR—INHAMBANE LOWLAND FOREST Altitudinal range 450-750 m, canopy height 25-30 m with emergents to 40 m tall. Middle and shrub layer well developed. Herb layer sparse ex- cept in disturbed areas, where it is dominated by Olyra latifolia. TREES: ene cerasifera, Albizia adi- errima labrescens, A. hylus melliodorus, Angy- localyx braunii, Aningeria pseudoracemosa, An- thocleista grandifolia, Antiaris toxicaria, Baph- ia semseiana, Bequaertiodendron natalense, Brachystegia microphylla, Calycosiphonia spathicalyx, Campylospermum sacleuxii, Cas- earia battiscombei, C. ?runssorica, Celtis durandii, Chaetacme aristata, Chionanthus mildbraedii, Craibia brevicaudata subsp. brevi- audata, Croton macrostachyus, C. cf. mega- locarpoides, Dialium holtzii, Dichapetalum stuhlmannii, Diospyros brucei, D. mespiliformis, D. zombensis, Dracaena usambarensis, Drypetes natalensis, Elaeis guineensis, Enantia kummer- iae, Erythrophleum suaveolens, Ficus cyathisti- pula subsp. cyathistipula, F. kirkii, F. lutea, F. var. quibeba, F. sur, F. thonningii, Filicium de- cipiens, Funtumia africana, Garcinia buchan- ani, Haplocoeliopsis africana, Haplocoelum iodiscus fraxinifolius, Lettowianthus stellatus, udia mauritiana, Macaran Monodora grandidieri, Odyendea zimmermannii, Pachystela brevipes, Paramacrolobium coeruleum, Parinari excelsa, Parkia filicoidea, Phyllanthus inflatus, Placod- iscus aff. amaniensis, Pteleopsis myrtifolia, Pter- ocarpus milbraedii subsp. usambarensis, Rhodognaphalon schumannianum, Ricinoden- dron heudelotii, Rothmannia fischeri subsp. fis- cheri, R. manganjae, Sapium ellipticum, Schef- flerodendron usambarense, Sorindeia madagascariensis, Stereospermum kunthianum, Suregada zanzibariensi abernaemontana pachysiphon, Terminate sambesiaca, Treculia africana, Trichoscypha ulugurensis, Trilepisium madagascariense, Uvariodendron gorgonis, Vangueriopsis longiflora, Vitellariopsis cuneata, Ziziphus pubescens. SMALL TREES AND SHRUBS: Achyrospermum car- valhi, Allophylus pervillei, Campylospermum sa- cleuxii, Carvalhoa campanulata, Chytranthus prieurianus subsp. longiflorus, Clerodendrum capitatum var. capitatum, Cremaspora triflora bsp. confluens, Crotonogynopsis usambarica, Didymosalpinx norae, Dorstenia kameruniana, Ixora tanzaniensis, Justicia interrupta, Leptac- tina platyphylla, Maytenus undata, Memecylon aff. myrtilliodes, Olax gambecola, Pavetta ?mze- leziensis, P. stenosepala subsp. kisarawensis, Pleioceras orientale, Psychotria schliebenii, Psy- drax livida, Rhus natalensis, Rinorea arborea R. ferruginea, R. subintegrifolia, MT SPA obtusisepalus, Sloetiopsis usambarensis, Taren- na pavettiodes subsp. affinis, Tricalysia pallens. LIANES AND CLIMBERS: Cissus producta, Cnestis confertiflora, Culcasia orientalis, Entada pur- saetha, Oncinotis tenuiloba, Rhoicissus revoilii, Salacia lovettii, Strychnos angolensis, Tetracera cf. litoralis. HERBS: Aeolanthus holstii, Begonia wakefieldii, Brillantaisia madagascariensis, Bulbophyllum concatenatum, Cheirostylis lepida, Chlorophy- tum sparsiflorum, Coelachne africana, Crasso- cephalum crepidiodes, Diodia sarmentosa, Dis- peris uzungwae, Geophila obvallata subsp. iodes, 876 Annals of the Missouri Botanical Garden Habenaria trilobulata, Impatiens joachimii, I. walleriana, Justicia glabra, J. interrupta, Lo- belia cymbalariodes, L. inconspicua, Maranto chloa leucantha, Mellera lobulata, Oldenlandia affinis, O. corymbosa var. caespitosa, Olyra la- tifolia, Pentas longituba, Phaulopsis imbricata, Pseuderanthemum tunicatum, Puelia olyrifor- mis, Saintpaulia ionantha, Scadoxus multiflorus, Scleria isostephana, S. racemosa, Sclerochiton obtusisepalus, Sida urens, Solanecio angulatus, Streptocarpus glandulosissimus, Utricularia striatula. ZANZIBAR-INHAMBANE TRANSITIONAL RAIN FOREST Altitudinal range 750—1,200 m. Canopy height 25-30 m with emergents to 40 m tall. Taller trees are found in the valley bottoms than on the ridge tops. Middle and shrub layer well developed. Herb layer sparse except in disturbed areas, where it is dominated by a dense growth of Zingiberaceae and Piperaceae. TREES: Afrosersalisia cerasifera, Aidia micran- tha var. msonju, Albizia adianthifolia, A. gum- mifera, Allanblackia stuhlmannii, Aningeria pseudoracemosa, Anisophylla obtusifolia, An- nonaceae gen. nov., Anthocleista grandifolia, Antiaris toxicaria, Antidesma vogelianum, Beilschmiedia kweo, Cassia angolensis, Cassi- pourea gummiflua, Celtis africana, Cephalo- sphaera usambarensis, Chrysophyllum gorun- gosanum, C. cf. lanceolatum, Cleistanthus polystachyus, Cola uloloma, Croton sylvaticus, Dalbergia boehmi subsp. Dialium holtzii, Diospyros amaniensis, D. cf. troupinii, Dracaena usambarensis, Drypetes cf. arguta, D. reticulata, D. cf. roxburghii, D. usambarica var. trichogyna, Eugenia capensis, Funtumia afri- cana, Garcinia volkensii, Glenniea africana, Grewia cf. barombiensis, Heinsenia diervil. leoides, Homalium longistylum, Isolona hexalo- ba, Ixora scheffleri, Lagynias rufescens subsp. angustilobus, Leptactina platyphylla, Macar- anga capensis var. capensis, Maesa lanceolata, Maranthes goetzeniana, Maytenus pos Memecylon ?brenanii, Memecylon sp. oe mil, sogyne insignis, Milicia excelsa, Milletia * slon: gistyla, Mimusops aedificatoria, Myrianthus holstii, Newtonia buchananii, Ochna holstii, Oc- toknema orientalis, Odyendea zimmermannii, Omphalocarpum sp. nov., Pancovia golungensis, Parinari excelsa, Parkia Alida Phyllanthus inflatus, Polyceratocarpus scheffleri, Porteran- dia penduliflora, Rawsonia lucida, Rothmannia urcelliformis, Sapium ellipticum, Sibangea ple- ioneura, Sorindeia madagascariensis, Strombos- ia scheffleri, Syzygium guineense subsp. afro- montanum, Tabernaemontana pach siphon, Tarenna pavettoides subsp. affinis, Trichilia dre- geana, Trichoscypha uc pus Trilepisium iopsis longiflora, Xylopia aethiopica, X. parvi- folia, Xymalos monospora, Zanha golungensis. SMALL TREES AND SHRUBS: Acalypha psilo- stachya var. psilostachya, Alchornea laxiflora, Allophylus pervillei, Baissea sp., Brachysteph- anus africanus, Campylospermum reticulatum, Carpolobia cf. goetzei, Chassalia discolor, Clau- sena anisata, Dicranolepis usambarica, Dra- caena laxissima, Erythrococca polyandra, La- gynias rufescens subsp. angustiloba, Lasianthus sp. nov., Leptonychia usambarensis, Lobelia lon- gisepala, Mostuea brunonis var. brunonis, Or- mocarpum sp. nov., Oxyanthus pyriformis subsp. tanganyikensis, O. speciosus subsp. stenocarpus, Pancovia holtzii, Pauridiantha sp. nov., Pavetta sp., Piper capense, P. umbellatum, Psychotria lauracea, P. schliebenii, P. tangan yicensis var. ferruginea, Pyrostria sp. nov., Rauvolfia mannii, Rhus longipes, Rinorea ferruginea, R. ilicifolia var. amplexicaulis, R. subintegrifolia, Schizo- zygia coffaeoides, Tricalysia pallens, T. sp. nov. LIANES AND CLIMBERS: Agelaea heterophylla, Baissea myrtifolia, Culcasia falcifolia, Keetia venosa, Monanthotaxis buchananii, Salacia lov- ettii, S. madagascariensis, Tetracera cf. litoralis, Uncaria africana var. africana, Uvaria tanza- niae. HERBS: Alectra kirkii, Begonia oxyloba var. kummeriae, Cheirostylis lepida, Costus subbiflo- rus, Dorstenia tayloriana var. laikipiensis, D. warneckei, Geniosporum africanum, Impatiens confusa subsp. longicornu, Isoglossa lactea, Jus- ticia aff. nyassana, J. pseudorungia, Leptaspis cochleata, Lobelia baumannii, L. inconspicua, Peperomia molleri, P. rotundifolia, Plectranthus sp. = Schlieben 4215, Pollia condensata, Rungia sp. ? = Mwasumbi 2720, rhueróstenfianos da subsp. africanus, Stenandriopsis warneckei, Streptocarpus caules- nov., Solenostemon sp cens var. pallescens, Utricularia livida. HEMIPARASITES: — Englerina inaequilatera. AFROMONTANE RAIN FOREST For ease of data presentation, Afromontane Rain Forest is taken here to include the forest type Volume 75, Number 3 1988 Lovett et al. 877 Mwanihana Forest Angiosperm Flora defined as Afromontane Undifferentiated Forest by White (1983). Altitudinal range 1,200-1,700 m. In valleys and sheltered areas canopy height 25- 30 m with emergents to 35 m tall. On ridges and exposed sites canopy height 15-25 m. Middle and shrub layers well developed in tall forest, middle layer not pronounced in exposed forest. Herb layer sparse in tall forest, dominated by Acanthaceae and Gramineae in more open areas. TREES: Agauria salicifolia var. pyrifolia, Albi- zia gummifera, Allanblackia stuhlmannii, A. ulugurensis, Alsodeiopsis schumannii, Aphloia theiformis, Beilschmiedia kweo, Vh s ee dron magalismontanum, abyssin Brachylaena huillensis, Bridelia brideliiflia. Caloncoba welwitschii, Canthium captum, earia battiscombei, Cassipourea gummiflua, C. malosana, Chrysophyllum gorungosanum, Cleistanthus polystachyus, Coffea mongensis, ola greenwayi, Craibia brevicaudata subsp. brevicaudata, Craterispermum longipeduncula- tum, Cryptocarya liebertiana, Diospyros amani- ensis, Drypetes cf. arguta, Faurea saligna, Fi- calhoa laurifolia, Garcinia kingaensis, G. volkensii, Heinsenia diervilleoides, Hirtella megacarpa, Isoberlinia scheffleri, Isolona hex- aloba, Ixora scheffleri subsp. scheffleri, Lasio- discus usambarensis var. usambarensis, Macar- anga capensis var. kilimandscharica, Maesa lanceolata, Maranthes goetziana, Maytenus ac- uminata, Memecylon ?brenanii, M. aff. semsei, Myrianthus holstii, Myrica salicifolia, Mystrox- ylon aethiopicum, Newtonia buchananii, Ochna holstii, O. kenyensis, Ocotea usambarensis, Olea Cas- scheffleri, Polyscias fulva, Psychotria aff. tri- choclada, Psydrax parviflora subsp. rubrocos- tata, Rapanea melanophloeos, Rawsonia retic- ulata, Schrebera alata, Strombosia scheffleri, Strychnos mellodora, S. mitis, Syzygium cor- tum, S. guineense subsp. afromontanum, S BORSS, Trichocladus goetzei, Trichoscy- phya ulugurensis, Uvariodendron sp. nov., Uva- riopsis bisexualis, Vepris stolzii, Xylopia ae- thiopica, ymalos monospora, Zanha golungensis, Zanthoxylum sp. SMALL TREES AND SHRUBS: Acanthopale pubes- Casearia gladiiformis, Chassalia parvifolia, Clu- tia abyssinica var. abyssinica, Coffea mufindien- sis, Craterispermum longipedunculatum, Cro- talaria unicella, Crotonogynopsis usambarica, Dissotis princeps var. princeps, Erythrococca sanjensis, E. ulugurensis, Garcinia volkensii, Iso- glossa bracteosa, Lasianthus kilimandscharicus subsp. kilimandscharicus, L. peduncularis, La- siodiscus usambarensis var. usambarensis, Lep- tonychia usambarensis, Maytenus mossambicen- sis subsp. mossambicensis, M. undata, Memecylon aff. amaniense, Memecylon sp. nov., Micrococca cf. holstii, Mimulopsis arborescens, M. kili- mandscharica, Pauridiantha paucinervis subsp. holstii, P. sp. nov., Pavetta lynesii, P. mshi- geniana, P. nitidissima, P. stenosepala subsp. kisarawensis, Peddiea polyantha, Pseuderan- themum campylosiphon, Psychotria crypto- grammata, P. faucicola, P. goetzei, P. megalo- pus, P. zombamontana, P. sp. nov., Rauvolfia mannii, Rinorea arborea, Rytigynia caudatis- sima, R. hirsutiflora, R. lichenoxenos subsp. gla- brituba, Sclerochiton obtusisepalus, Sericanthe odoratissima var. ulugurensis, Solanum aff. schu- mannianum, Strychnos myrtoides, Tarenna uzungwaensis, Teclea nobilis, Tricalysia pallens, Trichilia sp. nov., Turrea holstii, Uvariopsis sp. ?nov., Xymalos monospora. LIANES AND CLIMBERS: Asparagus asparagoides, Embelia schimperi, Schefflera myriantha, Tilia- cora funifera. HERBS: Aframomum laxiflorum, Alectra kirkii, Anisosepalum humbertii, Ardisiandra sibthorp- ioides, Begonia sp. ?nov., Bulbophyllum cf. im- bricatum, Cheirostylis lepida, Cincinnobotrys oreophila, Cyperus T5 indus, Disperis elaphoceras, Dissotis polyantha, Dorstenia den- ticulata, D. holstii, D. schliebenii, Gladiolus ru- picola, Gymnosiphon usambaricus, Impatiens E kentrodonta, I. rubromaculata subs , L sp. nov. aff. rubromaculata, pa FAR Justicia pseudorungia, Liparis sp. ?nov., Lobelia baumannii, Medini engleri, Olden- landia rupicola, Plectranthus sp., Polystachya ulugurensis, Saintpauliopsis lebrunii, Sclerochi- ton obtusisepalus, Seychellaria africana, Sole- nostemon sp., Streptocarpus glandulosissimus. HEMIPARASITES: Englerina inaequilatera. RIDGE TOP FOREST Altitude 1,760 m. Canopy height 3-6 m. Middle layer absent, the shrub layer composed of many small-diameter poles when under the canopy. Herb layer sparse under canopy, in open areas domi- 878 Annals of the Missouri Botanical Garden TABLE 1. Forest Reserve. Taxa recently described from Mwanihana Disperis ltd Verdc. uzungwae pee estais (Pax) Hutch. var. trichogyna A. Sm. [n m sanjensis A. R.-Sm. Lagynias rufescens (E. A. Bruce) subsp. angustiloba erdc. Pavetta nitidissima Bridso arenna u elie Lua Bridson Uvaria tanzaniae Uvariopsis "ines sid Verde. nated by Gramineae and Pteridophyta. Rich epi- phytic bryophyte cover TREES: Afrosersalisia sp. ?nov., Allanblackia ulugurensis, Aphloia theiformis, Apodytes di- midiata subsp. acutifolia, Bequaertiodendron magalismontanum, Cryptocarya liebertiana, Diospyros whyteana, Faurea saligna, Garcinia kingaensis, Isoberlinia scheffleri, Maytenus acu- minata, Newtonia buchananii, Ochna holstii, Ocotea kenyensis, O. usambarensis, Olea capen- sis, Olinia rochetiana, Rapanea melanophloeos, Strychnos mitis, Syzygium cordatum, Ternstroe- mia polypetala var. ?nov., Trichocladus goetzei. SMALL TREES AND SHRUBS: Agauria salicifolia var. pyrifolia, Clutia abyssinica var. abyssinica, In- ula stuhlmannii, Pauridiantha paucinervis subsp. holstii, Phyllanthus hutchinsonianus. LIANES AND CLIMBERS: Schefflera myriantha. HERBS: Cryptotaenia calycina, Cynorkis ana- camptoides, C. buchwaldiana subsp. braunii, C. Polystachya cultriformis, P. transvaalensis, pleistadenia, ?goetzeana, P. melantha, P. Saintpauliopsis lebrunii, Solanecio epidendri- cus, Stolzia leedalii, S. nyassana, Streptocarpus caulescens var. pallescens, Tridactyle sp. nov., T. sp. nov. = Leedal 6138. HEMIPARASITES: Thesium triflorum. DISCUSSION The moist forests of eastern Tanzania and south- eastern Kenya are well known for the high number of species that are found only in that area (Lovett, 1988; Rodgers & Homewood, 1982b). The Mwa- nihana Forest Reserve is no exception, and a num- ber of new species have been described recently from the locality (Table 1). Many other species are ABLE 2. Taxa in the Mwanihana Forest Reserve currently inadequately known to be described, or taxo- nomic research not yet completed. The following ab- breviations are used to refer to collectors: DT = Duncan Thomas, DB = Diane Bridson, RP = Roger Polhill, JL = Jon Lovett. sn = sine numero sterile collections from the ecological survey. Afrosersalisia sp. ?nov., JL sn Annonaceae ? nov. = es 65, JL sn egonia sp. ?nov., 79 Bulbophyllum cf. uu E DT 3876 3758 47 Impatiens joachimii var. ?nov., DB 602, DT 3798, JL 271 l. aff. rubromaculata, DT 3846 Justicia aff. nyassana, DB 644 Lasian thus sp. nov., DB 650, DT 3759, JL 235 . ?nov., 3874 aff. amaniense, DT 3850 === Onphalocarpun E nov., de 3652 Ormocarpum sp. nov., J 384 Pauridiantha sp. nov., DB 648, DT 3680B, DT 377 3841 778, DT 3826, DT 3 Pavetta sp. nov., JL 291 Placodiscus aff. amaniensis, DT 3724 23 m aff. K. Hoffm., DT 3636 ntidesma vogelianum Mue , DT E Bridelia brideliifolia (Pax) ng DT 3 Cleistanthus polystachyus Hook. f. ex Planchon, JL sn Clutia abyssinica Jaub. & Spach var. à seme DT L 300 chy us pr Re x Del., JL sn i s & M. Gilbert, DT 3727 DT 3707, Croton macrosta . cf. C. sylvaticus Hochst. ex Sl ii A usambarica Pax, DT 3761, DT 3924, P 51 bee natalensis HM Hutch., JL sn D. reticulata Pax D. Es. (Paz). Bod var. trichogyna A. R.-Sm., L 219 D. ?arguta (Muell. Arg.) Hutch., DT 3913 D. cf. roxburghii (Wall.) Hurusawa, JL sn Erythrococca polyandra (Pax & K. Hoffm.) Prain, DT 3751 E. sanjensis A. R.-Sm., DT 3677, DT 3774, DT 3775, T 3782, DT 3783 E. ulugurensis A. R.-Sm Bel 3762, DT 3770 Macaranga capensis Baill. v r. capensis, DT 3739 M. capensis Baill. var. Le y ip paya (Pax) Friis & M. Gilbert, JL sn Micrococca cf. holstii (Pax) Prain, DT 3833 Phyllanthus hutchinsonianus S. Moore, DT 3803 P. inflatus Hutch., DT 3609, JL 181 Ricinodendron heudelotii (Baill.) n. ex Pax, JL sn Sapium ellipticum bspw ) D JL s Sibangea pleioneu ,DT 3651, JL 221 Suregada pesti Ball. 'DT 3613 Uapaca paludosa Aubrév. & Léandri, JL sn FLACOURTIACEAE Aphloia theiformis (Vahl.) Benn., DT 3705 Caloncoba en ee Gilg, JL sn Casearia battiscombei R. E. Fr., C. gladiifo is Mast., DT C. ?runssorica Mildb., DT 3773, DT 3755 Homalium longistylum Mast., JL sn Ludia mauritiana ES Dt 3936 3650 R. aff. lucida ` imaq ` Sonder, DT 3638, DT 3966 R. reticulata Gilg, DT 3835, JL 290 GESNERIACEAE Saintpaulia ionantha H. Wendl., RP 5130 Streptocarpus caulescens Vatke var. pallescens Engl., DT 3620, DT 3847 S. glandulosissimus Engl., DB 604, DT 3848, RP 5155 GUTTIFERAE Allanblackia stuhlmannii (Engl.) Engl., JL sn A. u . kingaens : G. volkensii Engl., DT 3649, DT 3912 Harungana madagascariensis Poiret, JL sn HAMAMELIDACEAE Trichocladus goetzei Engl., JL sn ICACINACEAE Alsodeiopsis schumannii (Engl.) Engl., DT 3915 Apodytes dimidiata E. Meyer ex Arn. subsp. acutifolia (Hochst. ex A. Rich.) Cuf., DT 3715, DT 3766 LABIATAE chyrospermum carvalhi Guerke, DB 598 daolan hii holstii Guerke s.l., DB 617 Geniosporum africanum P. Beauv., DB 635 piena sp., DT 3823 P. sp. = Schlieben 4215, DB 640 = Mwasumbi 2720, DB 641, DT i ais s 3771 LAURACEAE Hou kweo (Mildbr.) A. Robyns & Wilczek, DT 388 C a liebertiana Engl., JL sn Ocotea kenyensis (Chiov.) A. Robyns & Wilczek, DT 3907 JL 301 O. usambarensis Engl., LEGUMINOSAE Caesalpinioideae Brachystegia micros M pc DT 3612 Cassia angolensis Hiern, JL s Dialium holtzii Hara DT 3732, DT 3740 Erythrophleum suaveolens (Guill. & Perr.) Brenan, JL sn Isoberlinia scheffleri (Harms) Greenway, JL sn Paramacrolobium coeruleum (Taubert) Leonard, JL sn Mimosoideae Albizia adianthifolia (Schumach.) W. F. Wight, DT 3646 A. glaberrima (Schumach. & IPS Benth. var. gla- brescens (Oliver) Brenan, D 734 A. T (J. Gmelin) C. A. Sm., DT 3735A, DT I zimmermannii Harms, DT 3640, DT 3731 Entada pursaetha DC., JL sn Newtonia buchananii (Baker) Gilb. & zm N. paucijuga (Harms) Brenan, Parkia filicoidea Welw. ex Oliver, Dr 3735 JL sn 882 Annals of the Missouri Botanical Garden Papilionoideae MONIMIACEAE [o ai de diea Harms, DT id Baphia semseiana Brummitt, DT 3 Craibia ed (Vatke) Sdn suben brevicau- data, 730 Crotalaria unicella Lam., RP 5139 Dalbergia boehmii Taub. subsp. “araya DT 3623 Millettia ee ata Gill, DT 3746 Ormocarpum sp. nov., JL 3 Pterocarpus mildbracdii Harms subsp. usambarensis (Verdc.) Polh., JL sn Schefflerodendron usambarense Harms, JL sn LENTIBULARIACEAE Utricularia livida E. Meyer, DT 3648 U. striatula Sm., DB 605, RP 5121 LOBELIACEAE Lobelia baumannii Engl., DB 642, DT 3804 L. cyambalarioides Engl., DB 606 L. inconspicua A. Rich., DB vu DT 3621 L. longisepala Engl., DT 36 LOGANIACEAE rota leista Linie say Gilg, JL s Mostuea brunonis Didr. var. MN DT 3637 L 276 ce hans ies Gilg. S. mellodora S. a e, JL sn S. mitis S. Moo «m S. myrtoides Gilg & Bus DT 3829 LORANTHACEAE Englerina inaequilatera (Engl.) Balle ex Polh. & Wiens, T 3830, JL 288, JL 299 MALVACEAE Sida urens L., DB 633 MELASTOMATACEAE Cincinnobotrys bii ae Gilg, DT 3877 Dissotis polyant ilg, 794 D. princeps (Kun th. Triana see prine eps, RP 5141 Medinilla ?engleri Gilg, DT 3 Memecylon ?brenanii A. & R. oe DT 3760, DT 3916 M. aff. amaniense (Gilg) A. & R. e DT 3850 M. aff. myrtilloides Markgraf, RP 5 M. aff. semsei A. & R. Fernandes, s DR M. sp. nov. DT 3773 M. sp. nov. JL 215 MELIACEAE Khaya nyasica Stapf ex Baker f., JL sn Trichilia dre, geana Sonder, JL sn | sp. nov., JL 23 Turraea holstii Guerke, DT 3714 MELIANTHACEAE Bersama abyssinica Fresen., JL sn MENISPERMACEAE Tiliacora funifera (Miers) Oliver, DT 3914 Xymalos monospora (Harvey) Baill., DT 3786, DT 3787, DT 3843 MORACEAE Antiaris toxicaria Lesch., JL sr Dorstenia denticulata E ‘Petr, RP 5151 D. holstii Engl., RP 5 . kameruniana Eng. RP 5119 D. schliebenii Mildbr., DT 3796 D. tayloriana Rendle var. laikipiensis (Rendle) Hijman, JL 213 D. warneckei Engl., DT 3660 Ficus cyathistipula pesa subsp. cyathistipula, DT 3666 m E E sur Forsskal, JL sn T Trilepisium Sarena ur DC., JL sn MYRICACEAE Myrica salicifolia Hochst. ex A. Rich., JL sn MYRISTICACEAE Cephalosphaera usambarensis (Warb.) Warb., DT 3619 MYRSINACEAE Embelia schimperi Vatke, JL sn Maesa lanceolata Forsskal, DT 3897 Rapanea melanophloeos (L.) Mez, JL sn MYRTACEAE Eugenia capensis (Ecklon & ea Sonder, JL sn Syzygium cordatum Hochst Krauss, JL sn S. guineense (Willd.) DC. arie afromontanum F. White, JL sn S. masukuense (Baker) R. E. Fries, JL sn OCHNACEAE Campylospermum reticulatum (P. Beauv.) Farron, JL 214 C. sacleuxii (Tieghem) Farron, DT 3939, JL 217, RP 5122 Ochna holstii Engl., Ouratea schusteri Gilg ex Engl., DT 3738, DT 3922 DT 3834 OLACACEAE Octoknema orientalis Mildbr., JL 287 Olax gambecola Baill., RP 5117 Strombosia scheffleri Engl., JL sn OLEACEAE Chionanthus mildbraedii (Gilg & Schellenb.) Stearn, RP 9123 Olea capensis L., JL s Schrebera alata (Hochst. ) Welw., DT 3904 Volume 75, Number 3 1988 Lovett et al. 883 Mwanihana Forest Angiosperm Flora OLINIACEAE Olinia rochetiana A. Juss., JL sn PIPERACEAE Peperomia molleri C. DC., DT 3692, RP 5142 P. rotundifolia (L.) Kunth., DT 3622 iper capense sn P. umbellatum L., JL 234 POLYGALACEAE Carpolobia cf. goetzei Guerke, DT 3758 PRIMULACEAE Ardisiandra sibthorpioides Hook. f., DT 3799 PROTEACEAE Faurea saligna Harvey, DT 3703 RHAMNACEAE Lasiodiscus usambarensis Engl. var. usambarensis, DT Ziziphus pubescens Oliver, JL sn RHIZOPHORACEAE Anisophyllea iier i Engl. & ips JL sn Ls Cassipourea gummiflua Tul., C. malosana (Baker) Alston, JL sn RUBIACEAE Aidia micrantha (K. Schum.) F. White var. msonju 3679 , DT 3842, DT ey Pas spathicalyx (K. Schum. ) Robbrecht, DB 8, JL 269 mnm captum Bullock, JL Coffea mongensis Bridson, JL sn C. mufindiensis Hutch. ex Bridson, DT 3819, DT 3914A Craterispermum longipedunculatum Verdc., DT 3784, 21 pene triflora (Thonn. ES Schum. subsp. confluens m.) Verdc., DB 624 DoD mosalnits norae (Swynnerton) Keay, DB 623 Diodia sarmentosa Sm., 632 Geophila d (Schumach) F. Didr. subsp. iodes . Schum Ed onis diervilleotdes K. Schum., DT 3 Ixora scheffleri K. Schum. & K. Krausse er. schef- fleri, DT 3852 I. tanzaniensis Bridson, DB 626, RP 5116 Keetia venosa (Oliver) Bridson, DT 3608 Lagynias pallidiflora Bullock, DB 612 Bruce) Verdc. subsp. angustiloba 9 Lasianthus kilimandscharicus K. Schum. subsp. kili- mandscharicus, JL sn L. pus E. A. Bruce, DT oe L. sp. B 650, DT 3759, JL 2 Leptactina platyphylla (Hiern) Web DB 653, DT Oldenlandia affinis (Roemer & Schultes) DC., DB 597 osa L. var. caespitosa (Benth.) Verdc., DB O. rupicola (Sonder) Kuntze, DT 3800 Oxyanthus pyriformis (Hochst.) Skeels subsp. tangan- yikensis Bridson, DB 636, DT 3745 O. eee subsp. stenocarpus (K. Schum.) Brid- , DB Fus fuc LM Verdc., DB s P. paucinervis (Hiern) Bremek. subsp. holstii (K. Schum.) y 31 P. sp. nov., DB 648, DT 3680B, DT 3778, DT 3826, DT 3841 Pavetta lynesii Bridson, DT 3717, DT 3837 P. mshigeniana Bridson, DT 8678 P. nitidissima Bridson, DT 3851 P. stenosepala K. Schum. subsp. kisarawensis (Bremek.) Bridson, DB 599, JL 210, JL 308 P. ?mzeleziensis Bridson, DB 625 P. sp., 91 Pentas longituba K. Schum., DB 5 Porterandia penduliflora (K. 3 ; Keay, DT 3611 Psychotria cryptogrammata Petit, JL 227 ] 384 1 P. meridiano-montana Petit var. angustifolia Petit, DT 3710 P. schliebenii Petit, DB 627 P. tanganyicensis Verdc. var. Tue Verdc., JL 278 P. aff. tric a ee Petit, DT 3836 P. sp., Psydrax mis (Hiern) Bridson, DB 586, DB 587, DB P. Ee oe ) Bridson subsp. rubrocostata (Robyns) on, Pyro ostria sp. Ro thmannia fischeri ide pd ) Bullock subsp. fischeri, J R. manganjae (Hiern) Keay, JL R. SE E ai (Schweinf. ex Hiern) Bullock ex Robyns, 374 . caudatissima li .. DT 3781 R. hirsutiflora Verdc., 381 R. lichenoxenos S. SCR M Robyns subsp. glabri- tuba Verdc., R. o ou EN BI 3780 Schum.) Robbrecht var. ulu- gurensis Robbrecht, DT 36804, DT 3711 Tarenna pavettiodes (Harvey) Sim subsp. affinis ( Schum.) Bridson, DB 629, DT 3614, DT 3750 T. uzungwaensis Bridson, DT 3 5 B 595A, DB 600, 78 p nM pallens Hiern, DB 595, D D , DB 630, DT 3607, DT 3615, DT 3712, DT 37 T. sp. nov., Uncaria africa n var. rh DT 3758A Vangueriopsis huh Verdc., RUTACEAE Clausena anisata (Willd.) Benth., DT 3941 Teclea nobilis Del., DT 3906 Vepris stolzii Verdoorn, DT 3713 Zanthoxylum sp., J , 884 Annals of the Missouri Botanical Garden SANTALACEAE ULMACEAE Thesium triflorum Thunb., DT 3809, JL 309 SAPINDACEAE Allophylus macrobotrys Gilg, DT 3792 A. melliodorus Gilg ex Radlk., JL sn A. pervillei Blume, DB 656, DT 3635 Chytranthus prieurianus Baill. subsp. longiflorus (Verdc.) Halle, DT 3940, JL 189 Deinbollia sp., JL sn Filicium decipiens (Wight & Arn.) Thw., E sn F. Davis in MS, "n sn Haplocoelum foliolosum (Hiern) Bullock, JL sn Lecaniodiscus fraxinifolius Baker, JL sn Pancovia golungensis (Hiern) Excell & Mendini; JL sn P. holtzii Radlk., DT 3633 Placodiscus aff. amaniensis aga DT 3724 Zanha golungensis Hiern, SAPOTACEAE P Sensing mu dii ) Aubrév., DT 3628 Afrosersalisia sp. iunea. Sek bc JZ Hemsley, DT 3725 Bequaertiodendron magalismontanum (Sonder) Heine . Hemsley, JL sn B. náialinte (Sonder) Heine & J. D. DT 3606 Chrysophyllum gorungosanum pe l., JL s C. cf. lanceolatum (Bl.) DC., Inhambanella henriquesii (Engl. & Wan, ) Dubard, JL sn Malacantha alnifolia (Baker) Pierre, JL sn Mimusops — ie uc Mildbr., DT vm Omphalocarpum sp. nov., Pachystela ene (Baker) Engl., JL s Vitellariopsis cuneata (Engl.) Aubrév., DT 3662 SCROPHULARIACEAE Alectra kirkii Hemsley, DT 3631, RP 5140 SIMAROUBACEAE Odyendea zimmermannii Engl., JL sn SOLANACEAE Solanum aff. schumannianum Dammer, DT 3896 STERCULIACEAE Cola greenwayi Brenan, JL sn C. uloloma Brenan, JL sn i ci a ini K. Schum., DT 3657, DT 3795 THEACEAE Ficalhoa laurifolia Hiern, JL sn Ternstroemia polypetala Melchior, var. ?nov., DT 3767 THYMELAEACEAE Dicranolepis usambarica Gilg, JL sn Peddiea polyantha Gilg, DT 3701 TILIACEAE Grewia cf. barombiensis K. Schum., DT 3747 Celtis africana Burm. a JL sn C. durandii Engl., Chaetacme aristata Planchon, JL sn UMBELLIFERAE Cryptotaenia calycina C. C. Townsend, DT 3806 VERBENACEAE Clerodendrum capitatum Schum. & Thonn. var. capi- tatum, DB 631 VIOLACEAE Thouars) Baill., DB 654, DT 3665, — Rinorea arborea JL 296 R. aff. arborea (Thouars) Baill., DT 3894 R. ferruginea Engl., DB 609 9, DB 622, DB 652, JL 185 R. ilicifolia (Welw. ex Oliver) Kuntze var. amplexicaulis Grey-Wilson, DT 3691 R. e ri ag (P. Beauv.) Kuntze, DT 3634, DT 3938, JL 223 VITACEAE Cissus producta Afz., DB 6 Rhoicissus revoilii Planchon, "br 3720 ANGIOSPERMAE, MONOCOTYLEDONS AGAVACEAE Dracaena laxissima Engl., JL sn usambarensis Engl., JL 181 AMARYLLIDACEAE Scadoxus multiflorus (Martyn) Raf., JL sn ARACEAE Amorphophallus sp., JL Callopsis volkensii Engl., Carmichael 127 Culcasia falcifolia Engl., C. orientalis Mayo, JL sn BURMANNIACEAE Gymnosiphon usambaricus Engl., JL 298 CYPERACEAE Cyperus pseudoleptocladus BM DT 3897A Scleria iostephana Nelmes S. racemosa Poiret, DB 61 9 GRAMINEAE Coelachne africana Pilger, DT 3935 Leptaspis E oi JL sn Olyra latifolia L Puelia olyriformis ee Clayton, DT 3722 IRIDACEAE Gladiolus rupicola Vaupel, DT 3706 LILIACEAE Asparagus asparagoides (L.) Wight, DT 3704 Volume 75, Number 3 1988 Lovett 885 et al. Mwanihana Forest Angiosperm Flora Chlorophytum sparsiflorum Baker, DB 608 C. sp., DT 3892, DT 3893 MARANTACEAE Marantochloa leucantha (K. Schum.) Milne-Redh., DB 615, RP 5127 ORCHIDACEAE Bulbophyllum concatenatum Cribb & P. Taylor, DB 590 B. cf. imbricatum Lindl., DT 3876 Cheirostylis lepida (Reichb. f.) Rolfe, DB 589, DT 3647, DT 3700 Cynorchis anacamptoides Kraenzl., JL 304 C. buchwaldiana Kraenzl. subsp. braunii (Kraenzl.) Sum- merh., 3 C. pleistadenia (Reichb. f.) Schltr., JL 303 Disperis elaphoceras Verdc., D. uzungwae Verdc., JL 27 Habenaria trilobulata Schltr., JL 277 Liparis sp. 874 Proba piden Mos : en DT 3875 P. ?goetzeana Kraenzl., P. melantha Schltr., DT ae P. JACO Schlechter, DT 3768, DT 3812, JL 302 P. ulugurensis Cribb & ipe DT 3793 Stolzia leedalii Cribb, DT 3871 S. nyassana Schltr., DT s Tridactyle sp. nov., DT 3763A sp. nov. = Leedal 6138, DT 3764 PALMAE Elaeis guineensis Jacq., JL sn Phoenix reclinata Jacq., JL sn SMILACACEAE Smilax kraussiana Meisen., JL sn TRIURIDACEAE Seychellaria africana Vollesen, DB 634, JL 226 ZINGIBERACEAE pbi angustifolium (Sonn.) K. Schum., JL sn 4. laxiflorum Loes ex Lock, JL 228 Costus subbiflorus K. Schum., JL sn EXPERIMENTAL STUDIES ON David A. Neill: SPECIES RELATIONSHIPS IN ERYTHRINA (LEGUMINOSAE: PAPILIONOIDEAE)! ABSTRACT Erythrina L. — em 112 species aie e throughout the tropics and subtropics. Most species are trees or shrubs, and m e diploids with n 1. All are adapted to bird ceo some by passerine birds and others by sous Erythrina is iudi "e into five subgenera and 27 sections. Research concentrated on sect. Erythrina, with 36 species centered in Mexico and Central America; aa 1 species in other sections were also studied. Experimental interspecific hybridizations and self-compatibility trials were conducted using cultivated trees at several botanic a sakes in Hawaii Comparative morphological onion were made of the hybrids and their paren population structure and natural hybridization were carried out in natural populations of toni urb ira pollinnied sect. Erythrina. Erythrina species are self.com ut some inbreeding depression is associated with selfing. Within sect. Erythrina, interspecific are obtained just as readily as are progeny from ded sisi dn rosses. The hybrids are vigorous, fertile, and by several measures exhibit interspecific heterosis. At g mic distances between the parental species (between sections and subgenera) , crossability, viability, and erii of the hybrid progeny are generally lower than in intrasectional aan Some hybrids we ained between species of different subgenera indigenous to different continents. There are probably no a d internal barriers to hybridization among all the diploid species of Erythrina. The genus may be sino zed e ally as a homo ogamic complex. Interspecific idi a are intermediate between their parental species in morphological traits, including macroscopic features of the es experimental hybridization, together with studies o comparative morphology and distribution patterns, suggest that some species of Erythrina are stabilized hybrid derivatives. Experimental hybridization studies have been a efforts to define taxa and taxonomic categories on cornerstone of research in plant biosystematics since the basis of reproductive barriers as revealed by the emergence of this synthetic field. Much of the experiment, an approach exemplified by the studies work of early biosystematists was directed toward of Clausen et al. (1939, 1940) and their proposal tudy was undertaken as part of a doctoral dissertation at Washington University, St. Louis, Missouri. ete ven first suggested the research on Erythrina biosystematics as a dissertation topic and helped in sia ways during a researc h His iia on plant cuptution ug n role ot Ibis asa in the rt Barneby Gerald Carr at the University of Hawaii provided use of his cytological laboratory and excellent advice on the handling of chromosomes. Michael Veith provided assistance and advice with the sc anning electron microscope at Washington University. iin Tobe E the gel sections "ari he in Figures 19, 21, and 23. Conversations with and comments by Héctor Hernández eter Hoch helped to improve the manuscript. Alina Chacón and Mary Merállo an 2 in the preparation perd manusc ae and figures. As a graduate student was supported by fellowships from the Foundation, Washington University Division of Biology and Biomedical Sciences, and Missouri Botanical Garden. js research was funded by grants from the National Scienc e Foundation (DEB 81-20386) and Elizabeth Neill ? Missouri Botanical Garden, St. Louis, Missouri 63166, U.S.A. ANN. Missouni Bor. GARD. 75: 886-969. 1988. Volume 75, Number 3 Neill 887 Erythrina to establish ecotypes, ecospecies, coenospecies, and comparia as universal units of classification to re- place the traditional ones. The criteria used to define taxa on the basis of fertility relationships and the enormous labor required to obtain the experimental results on a broad scale proved im- practical for a general-purpose classification sys- tem, and most present-day biosystematists have rightly abandoned the earlier efforts to “meddle” with the traditional taxonomic hierarchy. Experimental hybridization studies continue to play a central role in research on the nature of species relationships, however, and their usefulness extends far beyond the requirements of formal taxonomy. They provide the material for a broad spectrum of integrated studies in the genetics of evolutionary divergence. Research on long-lived perennials, although it requires patience and a long- term commitment of labor and resources, ticularly amenable to hybridization programs be- cause the parentals and several generations of offspring can be grown side-by-side, and many com- is par- parative studies can be carried out with the living plants. This paper describes the results of experimental investigations into the biosystematics and repro- ductive biology of Erythrina (Phaseoleae), a wide- spread genus comprising more than 110 species, most of them tropical trees. The research was concentrated on species of sect. Erythrina, a com- plex of 36 species centered in southern Mexico and Central America (Mesoamerica), but other taxa of Erythrina were included in some phases of the investigation. The results of the research are used to establish a series of hypotheses regarding species relationships and evolutionary history of Erythri- na. The hypotheses are presented here in se- uence: each is dependent on the validity of the previous hypotheses. Hypothesis 1. The species of sect. Erythrina can all hybridize freely with each other, and the resulting hybrids are as fertile as the parents. The section is a homogamic complex, and internal, post- mating isolating barriers between the species are absent. Hypothesis 2. complex encompassing sect. Erythrina extends, to a greater or lesser degree, to species in other sec- tions and other subgenera of the genus. Any diploid Erythrina species can mate with any other to form a viable F, hybrid, but hybrids between widely divergent species may exhibit varying degrees of sterility. The genus as a whole may be character- ized CUBA EN as a series of interfertile homo- The interfertile homogamic gamic complexes with weak to moderate repro- ductive barriers between the complexes. Hypothesis 3. The widely foraging humming- birds that pollinate trees in sect. Erythrina ensure effective outcrossing even in the isolated popula- tions of small neighborhood size and low density characteristic of these species. Self-compatibility and occasional autogamy allow establishment of a population from a single founder individual. When two species of sect. Erythrina are sympatric, the pollinating birds do not discriminate between them, and interspecific pollen transfer is likely. Hypothesis 4. The species in sect. Erythrina are mostly restricted in geographic range and are usually allopatric, separated by habitat differences. For the most part, these factors are effective bar- riers to interspecific gene flow. However, some- times different species do come into contact in nature, and then fertile hybrids are formed. Hypothesis 5. Patterns of distribution and phenetic variation in sect. Erythrina indicate that some distinct forms recognized as species are sta- bilized derivatives resulting from hybridization of two parental species. In the changing climates and dynamic geomorphologic landscape that have char- acterized Mesoamerica since the Miocene, and with the consequent migration of vegetation types and mixing of floristic elements, formerly allopatric species may have come into contact a number o times. With the temporary breakdown of external isolating barriers, the interfertile species hybrid- ized, and the subsequent segregation and stabili- zation of hybrid derivatives has contributed to the proliferation of Erythrina species in Mesoamerica. The first two hypotheses can be tested directly by experimental hybridization programs. The third and fourth can be substantially confirmed by ob- servations of mating behavior and patterns of vari- ation in natural populations. The fifth hypothesis is historical and can only be inferred by drawing on information obtained by testing the first four. The “level of confidence" (Gottlieb, 1972) in the final hypothesis of hybrid speciation in Erythrina is dependent upon the strengt the evidence presented in this paper in support of the four an- tecedent hypotheses. his research was made possible by the existence of the extensive living collections of Erythrina at three botanical gardens in Hawaii: Pacific Tropical Botanical Garden in Lawai, Kauai; Waimea Ar- boretum in Haleiwa, Oahu; and Ho'omaluhia Bo- tanic Garden in Kanehohe, Oahu. The cultivated Erythrina collections were assembled, beginning 888 Annals of the Missouri Botanical Garden in the early 1970s, through the efforts of the Erythrina monographer B. A ff. The gar- dens collectively now have in cultivation more than 90 of the 112 recognized species in the genu the remaining species are gradually being obtained s, and through requests to botanists around the world for seed. SECTION 1. EVOLUTION IN HOMOGAMIC COMPLEXES: A REVIEW The following section provides an overview of the conceptual framework of the experimental work on Erythrina and a literature review of the role of hybridization in the evolution of homogamic complexes. Grant (1953) coined the term “hybrid complex” for groups of related species linked by occasional or frequent hybridization, and he classified different types of hybrid complex based on their reproduc- tive mode and the means of stabilization of the hybrids. In two of these complexes the hybrid de- rivatives are mostly or entirely apomictic: in a clonal complex the hybrids are sterile and repro- duce vegetatively, and in an agamic complex they reproduce by agamospermy. In the remaining three types of hybrid complexes, the hybrid derivatives reproduce sexually: (1) in a heterogamic complex they are permanent structural heterozygotes or permanent odd polyploids; (2) in a polyploid com- plex they are amphiploid with respect to the pa- rental species; and (3) in a homogamic complex the hybrid derivatives exhibit normal meiosis and are sexual diploids, homoploid with respect to the parental species. In some groups forming homogamic complexes, internal reproductive barriers may be present and the hybrid derivatives may be partially intersterile with the parents and with each other, as revealed by Grant's studies of annual Gilia (summarized in Grant, 1981). More frequently, though, particu- larly in complexes of perennials and woody plants, the derivatives are highly interfertile with the par- ents, with each other, and with all the other species in the complex: the only barriers to gametic ex- change between any ppm or any pair of taxa in the group are external. Grant (1953) and rim (1972) pointed out a paradox inherent in the recombinational system of the homogamic complex that sets it apart from the other types of hybrid complex. In clonal, agam- ic, heterogamic, and polyploid complexes, the cy- togenetic features or reproductive systems of the hybrid derivatives are important criteria of hy- bridity and distinguish them from the parents. In homogamic complexes the derivatives are fertile and cytogenetically homogeneous with the parents, so these criteria cannot be used as a test of hy- bridity. Consequently, homogamic complexes are more difficult to analyze and may pass undetected. Grant (1953) contended that in the long term the "evolutionary potential" of homogamic com- plexes is much greater than in other types of hybrid complex. In the clonal, agamic, and heterogamic complexes, and to a lesser extent in polyploid com- plexes, favorable gene combinations are stabilized in the hybrid derivatives at the cost of a severe restriction in recombination. When environmental conditions change, the derivatives are less flexible in their capacity for genetic adaptation than are sexual diploids. Should the progenitors of the com- plex, the original sexual diploids, become extinct, an important source of new variation in the com- plex is lost. These restrictions do not apply to homogamic complexes, however. Since the derivatives are sex- ual diploids, recombination is unrestricted. They are able to backcross freely with the parentals, and the original species may become extinct without jeopardizing the evolutionary potential and flexi- bility of the complex. Relative to the other types of hybrid complex, the homogamic complex is, in the words of Grant, an “open-ended genetic sys- em.” The maintenance of the ability to hybridize gains importance because it extends the pool of natural — variation available for recombination and selection far beyond that present in any single species. In environments undergoing climatic and/or geologic change, that for adaptive adjustment of the organisms, and hy- brid recombinants from two or more species ma have greater fitness in the newly created habitats than either of the parental species. Evolution in an open-ended homogamic complex may follow a re- ticulate pattern, with cycles of divergence and dif- "extra" genetic pool may be crucial ferentiation alternating with hybridization and re- combination as environmental conditions change (Raven & Raven, 1976; Raven, The paradox is that while hybadicanon’ is most difficult to detect and analyze in homogamic com- plexes, on a broad scale homogamic complexes may be much more important in plant evolution than other types of hybrid complex. Grant (1953) even speculated that hybridization in homogamic com- plexes may account for much of the diversity of the angiosperms, and for the reticulate nature of variation and lack of clear discontinuities between the major phyletic lines of flowering plants. Grant concluded that “the ancestral stocks may have Volume 75, Number 3 1988 Neill Erythrina 889 been hybridizing on the diploid (or diploidized) level since the earliest stages of angiosperm evolution." Whether or not homogamic complexes have played such an important role in the evolutionary history of flowering plants, it is now well accepted that they are characteristic of the genetic structure of many large and ecologically dominant genera of trees and shrubs, at least in temperate regions. The only really thorough biosystematic study of a homo- gamic complex in a genus of woody plants, com- bining fossil evidence, experimental hybridizations, and careful field studies, is Nobs's (1963) exem- plary work on Ceanothus in California. Many species of Ceanothus are dominant shrubs in the chaparral vegetation of that region, and all are diploid with n — 12. Nobs showed that since the Miocene, certain wide-ranging species in Ceano- thus sect. Cerastes have formed hybrid swarms in areas where they have intermixed. In novel habitats created by an increasingly arid climate and by the exposure of new substrates such as serpentine out- crops, some of the hybrid derivatives have become stabilized as new self-perpetuating species. Numerous studies have also been carried out on natural hybridization in Quercus, and this enor- mous homoploid genus (n = 12), which dominates the forests of much of the north-temperate zone, is generally agreed to comprise a homogamic com- plex (Muller, 1952; Hardin, 1975; Van Valen, 1976) or, perhaps more accurately, several homo- gamic complexes corresponding to its subgenera, with strong but incomplete barriers between them. The results of Cottam's long-term Quercus hy- bridization program (Cottam et al., 1982) consid- erably strengthen the experimental evidence (most of the previous hybridization studies in the genus merely analyzed morphological variation in natural populations). mong other genera of trees and shrubs that probably comprise extensive homogamic com plexes are Eucalyptus (Pryor, 1959), Prosopis (Simpson, 1977), and Ribes (Keep, 1962). Most of the world’s flora is made up of tropical woody plants, and the role of hybridization and the presence of homogamic complexes in these groups is largely unknown and remains a matter of dispute. Many systematists who work on tropical woody genera evidently believe that hybridization is absent unimportant in the organisms they study, e.g., Ashton’s (1969) comments on Dipterocarpaceae in Southeast Asia. Ehrendorfer (1970) thought that narrower “niche width" restricted gene flow, and a higher incidence of polyploidy and apomixis in tropical tree species made them much less likely to hybridize than their temperate-zone counter- parts. It does appear logical that in species-rich tropical forests population density is low and neigh- borhood size is small for any one species, as well as for groups of sympatric congeners, so the op- portunities for hybridization may be fewer and the hybrids harder to detect than, for example, in a temperate forest with large populations of sym- patric Quercus species. At any rate, the critical experimental hybrid- ization trials have not been carried out for tropical woody plants, except for a few economically im- portant genera. For example, strong sterility bar- riers have been found between Amazonian species of Theobroma (Addison & Tavares, 1952). In con- trast, Hevea in the same region is probably a homo- gamic complex. In this homoploid genus (n — 18), fertile d were easily obtained in experimental gardens, and numerous natural hybrids were re- ported (Seb 1947). In many genera of tropical woody plants, all or most species share the same relatively high chro- mosome number and may be considered diploidized paleopolyploids. Thus, most genera of Bignoni- aceae have the same chromosome number of n — 20; they are probably paleohexaploids based on x — 7 (Goldblatt & Gentry, 1979). Such plant groups may be prime candidates for the formation of homogamic complexes. Until the present study, however, a thorough biosystematic investigation on the scale of Nobs's work on Ceanothus had not been carried out on any large tropical woody genus or, in fact, on any other woody genus. SECTION 2. HISTORY AND RELATIONSHIPS OF THE GENUS Erythrina L. comprises about 112 species dis- tributed throughout the tropical regions of the world and extending into warm-temperate areas such as South Africa, the Himalayas and southern China, the Rio de La Plata region of Argentina, and the southern United States (Krukoff & Barneby, 1974) (Fig. 1). Most species are trees or shrubs, but about 10 species occurring in climates with pronounced dry and/or cool seasons are perennial herbs with large woody rootstocks. Erythrina species occur in a very wide variety of habitats, from lowland tropical rainforest to very arid subtropical deserts to highland coniferous forests above 3, m The distinctiveness of Erythrina has long been recognized by legume systematists. Following Ben- tham (1865), the genus has been placed tradition- ally in the subtribe Erythrininae of the tribe Phas- eoleae, a relationship based principally on the characteristic trifoliolate leaves that Erythrina 890 Annals of the Missouri Botanical Garden ^v k= v > < q ^ p ASIA & OCEANIA % 12 Species °. $c NEOTROPICS =... n s 70 Species z - ú AFRICA 31 Species B 2 FiGURE 1. Distribution of Erythrina. shares with the rest of the Phaseoleae. In a recent — generic treatment of Phaseoleae, Lackey (1981 maintained the traditional classification but re- marked that “the relationship of [Erythrina] to the remainder of the Papilionoideae is an absolute mystery ... the genus would have long ago been accommodated outside the Phaseoleae had not the foliage suggested this tribe. In many significant characters, Erythrina stands alone among the Phaseoleae.” Besides Erythrina, the subtribe Erythrininae contains the genera Mucuna, Strongylodon, Bu- tea, Apios, Spatholobus, Cochlianthus, Rhodop- sis, and Neorudolphia. The relationships among these genera are not close and the erection of the subtribe is largely a matter of convenience to ac- commodate a loose assortment of genera not easily placed in other subtribes of the Phaseoleae. Butea alone, the only other arborescent genus in the Phaseoleae, appears to have some true af- finities with Erythrina. Baretta-Kuipers (1982) found that the unusual wood anatomy of Erythrina is very similar to that of Butea. In many other important traits, however, Erythrina is distinct from Butea and from all other legumes. The base chromosome number of x = 21, shared by all 86 Erythrina species that have been counted, is unique in the Leguminosae and indicates no direct rela- tionship with Butea with n = 9. The unusual, high activity—low affinity nitrate reductase system pres- ent in all Erythrina species that have been ex- amined differs in some respects from known nitrate reduction patterns in other angiosperms (Orebamjo et al., 1982). Neither Mucuna nor Butea (G. R. Stewart, pers. comm. to P. Raven, 1984) shares this trait with Erythrina. The Erythrina alkaloids, structurally complex isoquinolines nearly universal in the seeds of the genus, are found in no other legumes (Mears & Mabry, 1971) Ithough Erythrina is quite distinct from the rest of the Leguminosae, and despite its great eco- logical and morphological diversity, the cytological and phytochemical evidence cited above and the interfertility relationships presented in this paper indicate that the genus is unusually close-knit for its size (Raven, 1974), and there is no doubt that it is monophyletic. As such, the genus is an ideal subject for the biosystematic study of diversification of an entire evolutionary clade. The origin of Erythrina, like its relationship to the rest of the Leguminosae, is obscure. No fossil record of the genus has been reported. In the light of its distribution patterns, pollination, and dispersal mechanisms, and the known history of nosae as a whole, Raven (1974) postulated an ,egumi- Upper Eocene to Upper Oligocene origin for the genus (40-30 m.y. BP), followed by ocean-drift and/or other long-distance dispersal among the Volume 75, Number 3 1988 Neill 891 Erythrina three principal tropical regions of America, Africa, and Asia-Oceania. Much diversification of Ery- thrina has occurred independently in Africa and America and to a lesser extent in Asia. The place of origin of Erythrina is unknown, but South America appears most likely since the majority of the putative ancestral groups (as con- sidered by Krukoff & Barneby, 1974) within the genus are found there. Africa is also a possible candidate, since it likewise contains a number of endemic groups. Although Erythrina almost cer- tainly originated well after the breakup of West Gondwanaland, it has a basically South American— African distribution that is shared by many an- giospermous groups, including the Leguminosae itself (Raven & Axelrod, 1974). The Erythrina taxa in “tropical Laurasia,” i.e., Asia and Meso- america, are clearly derived groups. In Mesoamer- ica the genus has undergone extensive recent spe- ciation within a single lineage. PATTERNS OF DIVERSIFICATION: POLLINATION Erythrina species exhibit a great diversity in floral structure, inflorescence orientation, fruit morphology, seed coat coloration, and vestiture and epidermal ornamentation of foliage and calyces. The infrageneric classification of Erythrina is based principally on these characters. The diversity of floral structure reflects adaptive radiation in Erythrina with respect to pollination mechanisms. All Erythrina species have red or orange flowers and copious nectar, and are adapted to pollination by nectarivorous birds. There are two distinct syndromes of ornithophily in the genus. All 42 Old World species and 15 of the 70 New World species are pollinated by “perching birds" of sev- eral families in the order Passeriformes. Passerine birds cannot hover efficiently or for any length of time, and the inflorescences of passerine-pollinated Erythrina are oriented in such a way that the birds can perch while feeding on floral nectar. The co- rolla standard is usually broad, and the flowers are open, with exposed reproductive parts. Pollen is deposited on the feeding bird's breast. The flowers of passerine-pollinated species of Erythrina are diverse in size, form, and orientation, which ap- pears to reflect the variation in size, morphology, and behavior of the pollinators, which range from sunbirds and white-eyes weighing 8-10 g to orioles weighing over 35 g. The remaining 55 New World species of Ery- thrina (nearly half the genus) are pollinated by hummingbirds (Trochilidae). Hummingbirds are the most specialized of nectarivorous birds and the only ones that hover while feeding. The corolla standard of hummingbird-pollinated Erythrina is narrow and conduplicately folded to form a ““pseudotube,”” con- cealing the wing and keel petals as well as the reproductive parts. The flower resembles the tu- bular corollas of many gamopetalous hummingbird- pollinated plants, but in Erythrina the pseudotube is not sealed on the ventral side where the margins of the corolla standard meet. The inflorescence axis of the hummingbird-pollinated species is erect, and the flowers are oriented outward, providing no perch for the hovering hummingbirds. Only a small number of the 315 neotropical hummingbird species are Erythrina pollinators, and these are all similar in size, bill length, and behavior. The Erythrina pollinators are principally special- ized “high-reward trapliners," nonterritorial species that follow regular daily foraging routes between widely separated individual plants (Neill, 1987). The flowers of the hummingbird-pollinated Ery- thrina species are much more uniform in size and shape than those of the passerine-pollinated species, and this probably reflects the relative uniformity of pollination mechanisms among the former group. FRUITS, SEEDS, AND DISPERSAL The diverse fruit and seed characteristics of Erythrina species are indicative of adaptation for different dispersal mechanisms. The putative an- cestral species (Krukoff & Barneby, 1974) inhabit coastal, estuarine, or riverine environments and have dull brown floating seeds transported by oceanic or fluvial currents. These species are ef- fective colonizers: Erythrina fusca and E. varie- gata both became established on the island of Krakatoa a few years after the cataclysmic erup- tion of 1883 (Guppy, 1906). Since the review of Raven (1974), new anecdotal evidence has come to light concerning the dispersal of these seeds and their viability following long exposure to salt water. A drift seed of Erythrina variegata was recorded after a storm on the beach of Canton Island, a low coral atoll at 3°S latitude in the western Pacific, where Erythrina does not occur. The nearest pos- sible source for the drift seed is Samoa, 700 km to the south. The seed, planted in Hawaii, grew into a 20-m tree (from herbarium label of Degener 35066, BISH). Alone in Erythrina, E. subumbrans of Asia- Oceania has winged, wind-dispersed fruits. The Tanzanian endemic E. greenwayi has unusual fruits with narrow winglike ridges, but the fruits are heavy and do not appear to be effectively wind dispersed. Most of the putative derived species of Ery- 892 Annals of the Missouri Botanical Garden thrina have bright red seeds, which persist in con- spicuous display on the pods after dehiscence. Red seeds have evidently evolved independently in sev- eral lineages of Erythrina; one or two species in each lineage have bicolored red and black seeds. The red or red-and-black seeds are presumed to be **imitation arils" (Ridley, 1930) or mimetic ber- ries. According to this theory, they are eaten by frugivorous birds attracted by the bright colors and are dispersed when they pass through the digestive tract unharmed, but there are few actual reports of such “mistake” dispersals. Skutch (1971) re- corded an overwintering yellow-throated vireo (Vir- eo flavifrons) eating red Erythrina seeds in Costa Rica. I have seen this phenomenon on only one occasion, when in March 1983 in Chiapas, Mexico, I observed a migrant wood thrush (Hylocichla mustellina) ingest several red seeds of Erythrina folkersii displayed on the pods. A major autumn food item of this bird in eastern North America, before it migrates south, is the bright red, fleshy fruit of Cornus florida, which Erythrina seeds resemble quite closely (E. Morton, pers. comm.). Thus the dispersal of Erythrina seeds as mimetic berries by “naive” migrant birds does seem to be a real, though perhaps infrequent, phenomenon. The **mimetic berry" theory is fraught with all of the conceptual difficulties common to consid- erations of the evolution of mimicry. The alkaloids in the seeds of Erythrina are toxic, and the de- ceived bird must survive the passage of the seed through its gut if it is to produce subsequent gen- erations of birds that will disperse subsequent gen- erations of Erythrina. (The alkaloids are not re- leased unless the seed coat is broken, and frugivorous birds do not have strong gizzards to grind seeds.) Additionally, the mimic should be rare relative to the model, and the deception must occur frequently enough so that natural selection can act upon it. The question of mimetic seed coloration in Erythrina and other legumes is discussed in ) McKey (1975 FEATURES OF THE EPIDERMIS A great variety of special epidermal structures occurs in Erythrina, particularly on the abaxial leaf surfaces. These include hairs of many types, epidermal papillae and various ““lamellae,” and ep- icuticular wax deposits. The adaptive significance of these features is not known, but they are often diagnostic for particular species or species groups and often aid in identification of sterile material. Patterns of leaf epidermal features and their in- heritance in interspecific hybrids are discussed lat- er. SUBDIVISIONS OF ERYTHRINA The first formal subdivision of Erythrina was established by Harvey (1861), with subsequent treatments by Harms (1915), Louis (1935), and Krukoff (1939a, for the American species; 1939b, for the Asiatic-Polynesian species). In the 19th century a number of generic segregates were pro- posed based on the distinctive floral morphs of certain groups of species: e.g., Chirocalyx Meisn., Micropteryx Walp., Duchassaingia Walp., and ypaphorus Hassk. These segregates were treated as sections or subgenera of Erythrina by later monographers. The modern classification of the genus was established by Krukoff & Barneby (1974), who recognized 5 subgenera and 26 sec- tions. I accept their treatment as the systematic basis for the present work; a few taxonomic changes to be published later are anticipated in this paper prior to their formal designation. The infrageneric classification of Erythrina is summarized in Table 1. A list of the currently recognized species, with authorities and with changes in synonymy made since Krukoff & Barneby's (1974) conspectus of the genus, is included in Appendix I. The sections of Erythrina are well delimited morphologically and biogeographically, and each appears to be monophyletic. The subgenera also are delimited by several good characters and ap- pear monophyletic, except for the large and het- erogeneous subg. Erythrina, which includes 70% of the species in the genus. The relationships of the sections comprising subg. Erythrina to one another still present a number of unresolved taxo- nomic and phylogenetic questions. he following is a short narrative synopsis of the infrageneric classification of Erythrina and an outline of evolutionary and biogeographical trends; in the discussion, the taxa used in the experimental studies are emphasized. Subgenus Micropteryx is restricted to South America, except for Erythrina fusca in the mono- typic sect. Duchassaingia. With floating seeds dispersed by ocean currents, É. fusca is the only species in the genus to occur in both the Old World and the New World. It is widely distributed along coasts and rivers in the Neotropics and Asia- Oceania, as well as in Madagascar and the Mas- carene Islands, but its present native range does not include continental Africa. It often occurs in extensive pure stands in seasonal swamps. With its Volume 75, Number 3 Neill 893 1988 Erythrina TABLE l. Infrageneric classification of Erythrina. Number of Distribution Sections Species America Africa Asia-Oceania I. Subg. Micropteryx 1. Duchassaingia 1 X X (Madagascar) X 2. Cristae-galli X 3. Micropteryx 4 II. Subg. Erythrina Suberosae 4 X 9. Arborescentes 1 X 6. Hypaphorus 1 X 7. Breviflorae 4 X 8. Edules 2 X 9. Stenotropis 1 X 10. Pseudo-edules 2 X 11. Leptorhizae 4 X 12. Erythrina 36 X 13. Gibbosae 1 X 14. Corallodendra 9 X 14a. Fidelens 1 X 15. Cubenses 1 X 16. Olivianae 1 X 17. Caffrae 2 X 18. Humeanae 2 X 19. Acanthocarpae 1 X III. Subg. Tripterolobus 20. Tripterolobus 1 X IV. Subg. Chirocalyx 21. Bruceanae 1 X 22. Macrocymbium 2 X 23. Dilobochilus 1 X 24. Dichilocraspedon 1 x 25. Chirocalyx 14 X V. Subg. Erythraster 26. Erythraster 13 x x X ide distribution and presumably primitive features (Krukoff & Barneby, 1974), E. fusca or a fusca- like ancestor may represent the original progenitor of the entire genus. Section Cristae-galli includes two species, F. crista-galli, which forms extensive populations along the estuary of the Rio de La Plata in extra- tropical South America, and E. falcata, which in- habits the “Yungas” forest of the eastern Andean foothills and similar subtropical forest vegetation in southeast Brazil. The four species of sect. Mi- cropteryx inhabit riverine or upland forests of the Amazon and Orinoco basins and the Planalto of Brazil. Subgenus Erythrina, with 79 species in 17 sec- tions, is distributed throughout the three major tropical regions of America, Africa, and Asia, but no single section occurs in more than one of these areas. The subgenus includes all 55 of the Amer- ican hummingbird-pollinated species in six different sections which I believe to have been derived from passerine-pollinated groups by convergent evolu- tion in several independent lineages. Erythrina speciosa of coastal Brazil, in the monotypic sect. Stenotropis, is geographically and phylogenetically isolated from the rest of the hum- mingbird-pollinated species. The herbaceous, hum- mingbird-pollinated species of sect. Leptorhizae, endemic to central Mexico, are probably derived directly from the passerine-pollinated shrubby /ar- 894 Annals of the Missouri Botanical Garden 1 o 200 400 600 800 1000km ¿wama L L L J r T T Y ROME O 100 200 300 400 500 600 miles FIGURE 2. Distribution of Erythrina sect. Erythrina. The vist aie indicate the numbers of species known to R occur in each geopolitical region bounded by the heavy black line borescent sect. Breviflorae endemic to the same region. In a parallel manner, the Andean hum- mingbird-pollinated sect. Pseudo-edules may be derived from the Andean passerine-pollinated sect. Edules. large Mesoamerican-centered, humming- bird-pollinated sect. Erythrina (36 species) is closely allied with the remaining hummingbird-pollinated sections; Corallodendra, with 9 species in South America and the Antilles, and the monotypic sec- tions Gibbosae in southern Central America, and Cubenses endemic to Cuba. The relationship of these advanced arborescent hummingbird-pollinat- ed groups to the rest of the genus is not clear, however. shows the distribution of sect. thrina and the number of species know Figure 2 Ery- o occur in geomorphologically and politically delimited subregions of its range. The greatest concentration of species is in nuclear Central America, particu- larly in the Mexican state of Chiapas and in Gua- temala. Geologically, nuclear Central America is much older than southern Central America. It has been connected to the North American continent since the Cretaceous, whereas southern Central America was only a chain of volcanic islands until the close of the Panamanian isthmus in the Pliocene (Raven & Axelrod, 1974; Coney, 1982). It is most probable that sect. Erythrina originated in nuclear Central America following migration of its progen- itor from South America, an event that could have occurred either before or shortly after the final formation of the southern Central American land ridge. Species of this section, which comprises nearly one-third of the entire genus, inhabit nearly every forested habitat in the geologically active and climatically complex Mesoamerican region. In contrast to Erythrina fusca and other species that Volume 75, Number 3 1988 Neill 895 Erythrina form extensive monospecific stands, the species of sect. Erythrina generally occur at low population densities. Many have a restricted geographic range and occur in a single vegetation type, in a rather narrow altitudinal belt, or only on particular sub- strates, such as outcrops of calcareous rock. Sym- patry among species in the section is rare, but when it does occur, natural hybrids are generally found. All available evidence indicates that sect. Erythrina is an outstanding example of rapid adap- tive radiation and speciation in the recent geological past. The remaining sections of subg. Erythrina oc- cur in the Old World, and their affinities to the American sections are not apparent. The South African endemic sects. Caffrae, Humeanae, and Acanthocarpae are the only representatives of the subgenus on that continent. Certain floral, fruit, and seed features of sect. Caffrae do suggest an affinity with the monotypic Mexican sect. Olivia- nae, but a plausible explanation of such a con- nection is difficult to imagine. The Asian sects. Suberosae, Arborescentes, and Hypaphorus are an autochthonous group with stly primitive" features and do not appear to be closely allied with the American and African sections of subg. Erythrina. The species of sect. Suberosae possess one singularly ture: complex reticulate ““lamellae” formed by epi- dermal cells of the abaxial leaf surfaces (this paper, Section 5). The monotypic subg. Tripterolobus, consisting of E. greenwayi and endemic to a small area in the Rift Valley of Tanzania, is an evolutionary anomaly. The three-winged follicular pod is unique in the genus, while the flower, as Kru off & Bar- neby (1974) indicated, seems constructed from disparate elements of different subgenera. Subgenus Chirocalyx, with 5 sections and 19 species, is restricted to sub-Saharan Africa. Section Chirocalyx comprises 14 species which inhabit environments as diverse as the Kalahari Desert, the lowland rainforests of Cameroon, the vast sa- vannas of the Sahel, and the montane forests of eastern Zaire—a radiation reminiscent of sect. Er- ythrina in Mesoamerica, although with fewer species. The remaining sections in subg. Chiro- calyx are mono- or ditypic, each quite distinct morphologically. he final subgenus is Erythraster, with 13 species in the sole sect. Erythraster. It is basically an Old World group with two disjunct, derived species in the Neotropics. Erythrina variegata, the coastal- strand, ocean-dispersed species, occurs from Tan- zania and Madagascar around the shores of the Indian Ocean and westward through Indonesia, New Guinea, Polynesia, and Micronesia to the Mar- quesas. The remaining species inhabit upland areas, including four in East Africa and one in Australia. There is one endemic species on each of the islands or island groups of Madagascar, Java-Bali, New Guinea, Tahiti, and the Hawaiian archipelago, and each of these may be derived independently from E. variegata. The disjunct E. velutina, widely distributed in dry forests of northern South Amer- ica, the Galapagos, and the Antilles, and its Cuban endemic derivative, E. grisebachii, form a distinct species complex together with the Tahitian £. tahi- tensis and the Hawaiian E. sandwicensis. All the species of this Polynesian- Neotropical complex have “mimetic berry" red seeds, unlike E. variegata and most of the other species in sect. Erythraster. Erythrina variegata is present on Tahiti but not in Hawaii or the Neotropics. With this pattern of distribution, it appears most likely that these Poly- nesian- Neotropical disjuncts were established fol- lowing long-distance dispersal by birds across the Pacific, and not by ocean-drift of E. variegata or a variegata-like ancestor. SECTION 3. CHROMOSOME NUMBERS AND MeloTic BEHAVIOR IN DIPLOID AND POLYPLOID SPECIES Erythrina is well known to be relatively uniform cytologically; polyploidy is rare, and aneuploidy is unknown (Lewis, 1974; Goldblatt, 1981a, 1984). The basic chromosome number of the genus is x = 21, unique in Leguminosae. Of the 65 species counted prior to the present study, 61 are diploid 2n = , two are tetraploid (2n = 84), one has reports of both diploid and tetraploid races, and one has reports of hexaploid (2n = 126) and oc- toploid (2n = 168) races. e base number for Phaseoleae and probably for subtribe da c is x = 11, and reduction o n — 10 is common e tribe. naikin is x either an alloted based on n = 11 + = 10 or a hypotetraploid n = (11 x 2) — 1 (Goldblatt, "1981b), and thus a paleopolyploid ge- nus. UN i MATERIALS AND METHODS Floral buds were collected from trees in culti- vation at three botanical gardens in Hawaii: Pacific Tropical Botanical Gardens in Lawai (PT); Waimea Arboretum in Haleiwa (WA); and Ho‘omaluhia Bo- tanic Garden in Kaneohe (HO). Floral buds and/ or seeds were collected from wild populations of certain species in Mexico and Costa Rica. 896 Annals of the Missouri Botanical Garden For gametic counts and meiotic analyses, floral buds in developmental series were fixed either in 3:1 ethanol:acetic acid or in 6:3:1 chloroform: ethanol: acetic acid, which generally provided bet- ter fixation. After 1-2 weeks in the fixative at room temperature, buds were transferred to 70% ethanol and stored Anthers were squashed in acetocarmine with Hoyer's solution added (Beeks, 1955) to make permanent slides. For somatic counts, seeds obtained from wild elow populations were germinated on filter paper. The primary root tips were pretreated in 0.003 M 8-hydroxyquinoline for 4 hours at room temper- ature, fixed in 3:1 ethanol:acetic acid for 2-12 hours, and hydrolyzed in 10% HCl for 10 minutes at 60°C. Root tips were squashed in FLP orcein (Jackson, 1973). Slides were examined under phase contrast with a Zeiss Universal microscope; chromosomal con- figurations were photographed with Zeiss MC63 equipment using Kodak Technical Pan film devel. oped for high contrast. RESULTS Chromosome Numbers. Chromosome counts and voucher data are listed in Table 2. For cultivated material the original wild-collected voucher is cited if it exists; if not, a voucher made from the garden progeny is cited. All vouchers are deposited at Missouri Botanical Garden (MO) unless otherwise noted. The gametic count of n = 42 for E. amazonica reconfirms earlier somatic counts of 2n — 84 (Atchison, 1947; Goldblatt & Davidse, 1977) for this tetraploid. This species is distributed through- out the northern Amazon basin and in the Guianas, but all the chromosome counts to date have been obtained from populations in the Brazilian state of Maranhao. A more complete sampling of the species range may reveal infraspecific variation in ploidy level, as has been determined for other species with polyploid strains. My count of n — 21 for the tropical Asian Erythrina suberosa is diploid and agrees with 13 previous reports for the species. Mehra (1976), however, reported n — 42 in three populations in the western Himalayas, at its geographic margin and altitudinal upper limit. As with £. amazonica, a cytogeographic survey of ploidy level in E. sube- rosa is desirable With the exception of Erythrina macrophylla, the remaining 22 chromosome counts listed in Ta- ble 2 are all first reports for species. All are diploid = 21 or 2n = 42) except the octoploid burana (n — 84). This Ethiopian endemic is closely related to E. burttii, which ranges from Ethiopia south to Tanzania, and for which both hexaploid 2n — ca. 126; Atchison, 1947) and octoploid 2n = ca. 168; Goldblatt, 198 la) counts have been obtained. Chromosome numbers are now known for 86 of the 112 species of Erythrina recognized here. Eighty-one species (94%) are diploid, with the re- maining 5 species (6%) polyploid or variable in ploidy level. The polyploid species are all in dif- ferent sections and are not closely related to one another, with the exception of E. burttii and E. burana. Polyploidization has thus occurred at least four times independently in Erythrina. Given the rarity of polyploidy in the genus, it seems likely that the closely related E. burttii and E. burana were derived from a common polyploid ancestor. Chromosome counts have yet to be obtained from 25 species of Erythrina. Eleven of these are from Africa where three of the five known poly- ploids occur. Seven uncounted species are South American, where E. amazonica is the only poly- ploid known. Polyploidy is unknown for Erythrina in North and Central America, where 42 of the 45 native species have been counted. Meiosis in Diploid Species. Chromosome size, morphology, and meiotic behavior were similar in all species examined. Observations of individuals of two typical diploid species, Erythrina berenices (WA 815505) and FE. macrophylla (PT 750420001), are described and illustrated here. Observations of chromosome pairing at zygotene and pachytene are desirable in meiotic analyses on. 1984), but these were not feasible in Erythrina because of its high chromosome num- bers. At diakinesis, 21 bivalents were regularly formed. Each bivalent had either one or two ter- minal chiasmata. In E. macrophylla, the average number of chiasmata per cell was 31.5 + 1.8 out of 10 cells sampled. This accords with the figures of 31.35 + 0.54, 31.25 + 0.61, and 32.08 + 0.7 chiasmata per cell reported by Jalil et al. (1982) for, respectively, E. variegata, E. resupinata, and their F, hybrid E. x resuparcellii. At early to mid diakinesis, the bivalents were generally well separated, thus the gametic chro- mosome counts listed in Table 2 were usually made at this stage. Toward the later stages of diakinesis and as the nucleolus began to disintegrate, groups of two or more bivalents appeared clumped to- gether in the cell (Fig. 3). Thin strands of chromatin were frequently observed to stretch between bi- valents Volume 75, Number 3 1988 Neill 897 Erythrina TABLE 2. Chromosome counts of Erythrina species reported in this paper. Vouchers are housed at MO unless otherwise indicated. Species n= 2n= Voucher Data E. amazonica Krukoff 42 WA 76s449, cultivated. Brazil. Maranhão: Lapela, N. T. Silva 4238 (NY). E. batolobium Krukoff & 21 Missouri Botanical Garden, cultivated (from wild-collected rootstock). Barneby Mexico. Guerrero: Filo de Caballo, 6,300 ft., oak forest, /Veill 5647. E. berenices Krukoff & 21 WA 815505, cultivated. Mexico. Veracruz: Tlalnelhuayocan, H. Perales s.n. in 1981 ( : E. ire bud A. DC. 21 Mexico. Guerrero: H. Iltis 28655. E. burana R. Chiovenda 84 PT 740435001, cultivated. Ethiopia. ve unknown: F. Meyer s.n. in 1974. Voucher from cult.: š E. cochleata Standley 21 Costa Rica. Heredia: La Virgen, 10 bs ‘SW of Puerto Viejo, 200 m, Neil E. elenae Howard & 21 WA 80s614, cultivated: Cuba. Cienfuegos: Centro de Investigación Briggs Forestal s.n. Voucher from cult.: Neill 5078. E. florenciae Krukoff & 42 Mexico. Chiapas: Motozintla, Cerro Mozotal, 6,600 ft., Neill 5600. E. gibbosa Cuf. 21 Costa Rica. Alajuela: Cordillera de Tilaran, upper Penas Blancas valley, below Monteverde reserve, 1,300 m, Neill 5028. E. globocalyx Porsch & 21 Costa Rica. San José: Las Nubes, 1,700 m, Neill 5033. uf. E. hondurensis Standley 21 HO 80.037, cultivated. MI Tela: Hazlett s.n. in 1980. Voucher from cult.: Neil E. horrida A. DC. 42 Mexico. Oaxaca: 2 km E of ave road to Yavesia, 2,030 m, M. Sousa 12634 (MEX E. leptorhiza A. DC. 21 Mexico. México: ‘penne Ixtapaluca, old Hwy. 190, km 25, 8,300 ft., Neill 5646. E. macrophylla A. DC. 21 PT 750420001, cultivated. Guatemala. Sololá: Godinez, 6,145 ft., B. A. Krukoff 1975-4 (NY). E. mexicana Krukoff 42 Mexico. Oaxaca: 14 mi. SW of San Jeronimo Miahuatlan, 4,800 ft., Neill 5423. E. oaxacana (Krukoff) 42 — Oaxaca: 9 km N of Diaz Ordaz, road to Cuajimolaya, Krukoff 700 ft., Neill 5409. E. pudica Krukoff & 21 Pon Chiapas: 15 mi. E of Cintalapa, Hwy. 190, 2,000 ft., Neill Barneby 5440. E. sacleuxii Hua 21 WA 74p1296, cultivated. Kenya: Arabuko forest, near coast, La- vranos 11225 (NY). E. sigmoidea Hua 21 PT 740192001, cultivated. India: locality unknown, cultivated, D. A. Millington s.n. in 1974. Voucher from cult.: Neill 5715. E. smithiana Krukoff 21 PT 740329001, cultivated. Ecuador. Guayas: Manglaralto. Mac- Bryde & Herrera-MacBryde 690 (NY). E. sousae Krukoff 42 Mexico. Oaxaca: 14 km S of San Miguel Suchixtepec, 2,100 m, Neill 5425. WA 755960, cultivated. India: Matrimandir Gardens, cultivated. E. suberosa Roxb. 21 Voucher from cult.: Neill 5273. E. tahitensis Nad. 21 PT 770442001, cultivated. Tahiti: Manupa Ridge, 2,000 ft., Perl- man s.n. in 1977. Voucher from cult.: Neill 5177. E. tuxtlana Krukoff & 21 Mexico. Chiapas: 26 km N of Ocozocuautla on road to Malpaso, Barneby 2,100 ft., Neill 5486. The characteristic clumping of two or three bi- chromosomes, even after the main bodies of the valents became even more common in metaphase chromosomes were separated by a considerable I. At anaphase I, long strands of chromatin were distance on either side of the equatorial plate (Fig. frequently observed stretching between disjoining — 4). Annals of the Missouri Botanical Garden q ( » a Á me, | > — a = ` 4 aoe ld 5 V et; ` we 10um € 4 10um 10um ° . . ° °. ° ` ° * 2" i `` ° e @ ` * " mots. 7 o... By ^ > a ” o . I © . .. ” ` , e. a f ~ t ow = " b a oa 0,0. 4 e. ° P 10um 10um 7 — 8 s 3-8. 6 d ha pollen mother cells, diploid species of Erythrina.—3. Late rrr quis E phase, E. aaa is. WA 81s505. Sticky Clumping of bivalents. pu Anap chromatin bridges stretch bue ^en 2 disjoine ed chromosomes.— 5, a ase, E. ma Late disjunction of some dioe —7. Mitosis in root tip cell of diploid E. honda, Sousa 12634 ; clumped. i Diakinesis in gis mother cell of tetraploid E. amazonica, ophylla, PT o (n bereni nices, A 815505 (n = (2n WA 765449 (n = 42). Two la are circled. Volume 75, Number 3 1988 Neill 899 Erythrina The chromatin connections between disjoining chromosomes appeared to be in all cases simply "matrix bridges" caused by chromosome “‘sticki- ness" (Beadle, 1932). Based on observations at subsequent stages, it is unlikely that any of the observed *'bridges" were true dicentric bridges or any other pie lea resulting from chromo- somal inversions or translocations. Another frequently observed phenomenon was late disjunction of one or several bivalents at ana- phase I (Figs. 5, 6). One or two lagging bivalents were often present even at late anaphase I when most chromosome pairs were completely separated. However, observations of cells at later stages re- vealed no evidence of nondisjunction or unequal assortment of chromosomes. In contrast to the chromosome *'stickiness" and secondary association of bivalents at meiosis, so- matic pairing of homologues was not observed in mitotic root-tip cells. A typical mitotic configuration of the diploid species Erythrina horrida (2n = 42) is shown in Figure 7, where there is no evidence of pairing or of sticky chromatin connections be- tween chromosomes. Meiosis in pollen mother cells of the Asian KÉ. variegata (as E. indica Lam.) was depicted b Sundar Rao (1945) and in several additional Asian species by Mehra (1976). Both reported postsyn- aptic secondary association of chromosomes at metaphase I and subsequent stages to be common in some species. Mehra (1976) reported aberrant meiosis with 13-21 bivalents, 0-16 univalents, and 0-2 B chromosomes in a diploid strain of £F. suberosa, while a tetraploid strain of the same species exhibited normal meiosis with 42,. Jalil et al. (1982) reported normal meiosis with 21, in the artificial hybrid E. Xresuparcellii (E. resupina- ta X E. variegata). Pollen fertility, as estimated by Alexander's dou- was uni- used as parentals in the experimental hybridization trials, had a mean pollen fertility of 95% (at least 500 grains counted per sample). Such high fertility suggests that the chromatin "bridges" and late disjunction of bivalents observed in most cells at anaphase I are not indicative of meiotic aberrations, do not result in a high frequency of aborted cells, and therefore most cells receive the correct com- plement of 21 chromosomes following meiosis I and II Meiosis in Polyploid Species. Among the tet- raploid species of Erythrina, only a single strain of E. amazonica was available for analysis (WA 765449; PT 760356001). In several individuals of this strain, 42 bivalents were observed in some cells at diakinesis and metaphase I, while in others one or two quadrivalents were clearly visible (Fig. 8). In contrast to the merely *'sticky" postsynaptic associations seen in the diploids, the quadrivalents in E. amazonica appeared to be true multivalents resulting from synaptic pairing at prophase. The configuration of this species at meiosis I, then, is 38-42, and 2-0,. The formation of occasional quadrivalents did not disrupt normal disjunction, however, as no cells at telophase I or subsequent stages were observed with other than 42 chro- mosomes. At diakinesis and metaphase I of the octoploid E. burana, considerable clumping of chromosomes was evident in all cells examined. With a large number of chromosomes crowded together, the configurations were not completely resolvable and it was not possible to determine whether synaptic multivalents were actually formed. DISCUSSION The postsynaptic secondary association of bi- valents at meiotic metaphase observed in diploid Erythrina pus b^ vw re, from many other plant groups. According to a theory intro- duced by Dida (1930) and amplified by Law- rence (1931), secondary pairing is due to attraction of homologous or homeologous chromosomes when the degree of homology is not close enough to result in synaptic pairing, and is presumed to be indicative of allopolyploidy. In a recently formed autotetra- ploid, homology between the two pairs of chro- mosomes will be nearly complete and a multivalent will be formed at pachytene. In an allotetraploid or in an “old” tetraploid in which the genes of homeologous chromosomes have diverged to some extent, two bivalents result; they may later form a secondary association at metaphase I or late diakinesis due to attraction between the homeolo- gous chromosomes making up the two bivalents. Secondary pairing does not always occur, however; it is a relatively loose association and does not affect disjunction at anaphase I. is interpretation of secondary pairing and its relation to allopolyploidy has been borne out by quantitative studies of the spatial distribution at metaphase I of marked homeologous chromosomes in the allohexaploid Triticum aestivum (Kempanna & Riley, 1964). In other plant groups such a rigorous quantification of secondary pairing has not been obtained, but a number of workers have in- 900 Annals of the Missouri Botanical Garden ferred a history of polyploidy in groups with meiotic secondary pairing, particularly for those with high chromosome numbers suspected to be paleopoly- ploids. For example, Venkatasubban (1944), in a cytological study of Bignoniaceae, found a base number of n — 20 for the family and a presumed ancestral base number of x — 10, since up to 10 "secondarily associated" pairs of bivalents were present at metaphase Í in many species. The evi- dence from secondary pairing is not unequivocal, however; in the case of Bignoniaceae, Goldblatt & Gentry (1979) believed n — 20 to be a paleohexa- ploid of a base number x — from n — 21 Both Sundar Rao (1945) and Mehra (1976) and the latter author cited it as evidence for an ancestral lower base number for the genus. On the basis of present knowledge, however, it is not possible to state unequivocally that the observed meiotic pat- terns in diploid Erythrina species are due to sec- ondary pairing of specific homologous or homeol- ogous chromosomes, and not simply to random nonhomologous 7, with. aneuploidy noted secondary pairing in Erythrina, of chromosomal ma- trix material. Multivalent formation in É. amazo- nica, a neopolyploid, does appear to be a result of true synaptic pairing of homologous chromosomes. The hybridization trials carried out in Erythrina (described below) reveal a high degree of structural and genic homology in the chromosomes of all species, and it is probable that virtually any Ery- thrina genome can combine with that on the same ploidy level of any other species in the genus to form a viable F, hybrid. Whether tetraploid FK. amazonica is of autoploid or alloploid origin, then, “stickiness” it must have two highly homologous sets of chro- mosomes. It is somewhat surprising that more than two quadrivalents are not usually formed in meiosis I of E. amazonica. It is possible that the species contains a specific gene that suppresses multivalent formation and promotes strict homologous pairing of bivalents, similar to the Ph gene which performs this function in hexaploid Triticum aestivum (Riley & Chapman, 1958). SECTION 4. EXPERIMENTAL HYBRIDIZATION AND SELF-COMPATIBILITY MATERIALS AND METHODS Experimental hybridizations and self-compati- bility trials were conducted at Pacific Tropical Bo- tanical Garden and Waimea Arboretum February- July 1982 and February- April 1984. Although the living Erythrina collections at the two gardens share many accessions from the same sources, the species complement of mature, flowering individ- uals was different at each garden. The use of both gardens allowed a broader inclusion of taxa in the ie studies than would have been possible otherw In as self-compatibility trials and inter- specific hybridizations were conducted with natural populations of Erythrina chiapasana and E. gold- manii at El Sumidero Canyon National Park in Chiapas, Mexico in February 1983. The two species are parapatric at El Sumidero and hybridize nat- urally (this paper, Section 6). In all, 32 species were used in the interspecific hybridization trials, in 155 hybrid combinations including reciprocals. Species from throughout the worldwide distribution of Erythrina were used in the trials; four of the five subgenera and 12 of the 27 sections were represented. The monotypic Af- rican subg. Tripterolobus was the only subgenus not included. Eighteen species were tested for self- compatibility. All species used in the trials are diploids (n = 21) except E. amazonica, a tetraploid (n = 42). Attempts were made to hybridize É. amazonica as the pollen parent with several diploid species. The hybrid combinations were selected to rep- resent different “taxonomic distances" between the female and male parental species: "narrow hy- bridizations" between species of the same section, “medium hybridizations” between species of dif- ferent sections in the same subgenus, and **wide hybridizations" between species of different sub- genera. The narrow hybridizations involved mostly species within sect. Erythrina. The medium and wide hybridizations included crosses of sect. Ery- thrina to other sections and subgenera, as well as ect. Er- e the maximum taxo- nomic diversity and geographic range of the genus. representative hybridizations not involving se ythrina, selected to inclu Constraints on the Experimental Protocol. Short- ly after the initiation of the pollination trials, certain constraints imposed by Erythrina breeding systems became apparent. Other constraints were imposed by the fact that the experimental subjects were trees exposed to the vicissitudes of the weather and to local, uncontrolled variation in other factors that may affect reproductive success, such as soil fer- tility and moisture, and insolation. These consid- erations required a somewhat different experimen- tal protocol and a different statistical treatment of the results than has been customary with biosys- tematic studies of greenhouse-grown herbaceous plants. Volume 75, Number 3 1988 Neill Erythrina The proportion of fruit set in intraspecific and interspecific pollinations was quite low (see Results, below) and the incidence of postfertilization abor- tion of young fruits was very high. Pollination suc- cess and fecundity varied greatly among individuals the same species. Some trees were effectively “female sterile": hey produced no fruits either spontaneously (i.e., from “open-pollinated” flow- ers) or from controlled pollinations. At the same time, conspecifics and even individuals from the same accession, which were presumably at least half-siblings of the “female sterile" individuals, pro- duced fruits spontaneously in abundance and pro- duced fruits quite readily from both intraspecific and interspecific controlled pollinations. For many species only a single tree was available, so the use of intraspecific outcrossing success rate as a control was not possible for those species. Another constraint was the often limited number of flowers per tree that were accessible each day. On many trees only one or a few inflorescences producing three or four new flowers each day were accessible for hand-pollination. The time required to emasculate, isolate, and pollinate each flower individually also limited the number of flowers that could be treated each day. Another practical con- sideration was the amount of land and labor re- nations, so with limited resources large F, families of any particular combination could not be accom- modated. One positive aspect of Erythrina reproductive systems that influenced the experimental protocol was the relatively high viability of the seed. Among the hybrids, 45% of the seeds germinated and produced healthy F, plants. This high viability meant that large seed lots of any particular combination were not necessary to ensure that at least some progeny would survive to maturit e above considerations and constraints led to phasis on any particular combination. Once several well-formed maturing fruits were produced for any combination, pollinations of that combination were ceased and new combinations were attempted. When possible, species combinations were repeated using several different individuals as female and/ or male parents. For self-compatibility trials as well, pollinations within an individual were terminated once several semimature fruits had developed. n the course of the pollination trials it soon became apparent that certain individual trees of several species were more fecund, successful fe- male parents than others. To the extent possible, pollination trials were concentrated on the more successful females, within the limitations imposed y the number of flowers available. Hybrid com- binations or self-pollinations that failed to set fruit were repeated up to times or more, but for many combinations fewer than 10 flowers were pollinated due to limitations of time and available flowers. Pollination Techniques: Hybridization. The development of suitable techniques to emasculate, isolate, and pollinate the flowers involved consid- erable trial and error. Nylon mesh bags of several types were used initially to isolate the flowers, but these proved to be too unwieldy, requiring elabo- rate, heavy wire frames or other means of support so the mesh did not touch the flowers. Also, in rainy weather the high humidity within the mesh bags tended to cause all the flowers to abort. A simple alternative technique that proved suc- cessful was to isolate each flower individually. Flo- ral buds were emasculated at the latest possible stage of development, i.e., on the day before an- thesis and pollen release. The tightly closed corolla standard was carefully peeled open, and the anthers were e with dissecting scissors sterilized in 95% ethanol between each emasculation. anther Eos pollen before removal, the flower the pistil and sealed with plastic Scotch tape. This effectively protected the stigma from any chance pollen deposition and also prevented it from drying out. The following day the corolla was reopened and the standard excised. After pollination a small cone of aluminum foil, formed over the point of a pencil, was placed over the stigma and pinched lightly onto the style. This helped to hold the pollen on the stigma in the face of rain and wind, and isolated the stigma from any other pollen deposi- tion. The cap remained on the stigma throughout the development of the fruit. For the open-corolla, homogamous species of Erythrina adapted to pollination by passerine birds, this technique ensured that the pollen was applied while the stigma was receptive. In most species receptivity was signalled by presence of a wet, sticky exudate on the stigmatic surface on the day of anthesis. For the closed-corolla, protandrous species (primarily sect. Erythrina) adapted to hum- mingbird pollination, the style had not elongated fully and the stigma was not yet receptive on the 902 Annals of the Missouri Botanical Garden day following emasculation. The stigma was thus pollinated prematurely by this method, but the pollen held in place by the aluminum cap evidently remained viable at least until the next day when the stigma became receptive, and these species did set fruit with premature pollination. An alternative method, to wait two days following emasculation to pollinate the protandrous species, yielded no better results and was logistically more complicat- ed. Stephenson (1981) presented evidence from an extensive literature review that in many plant species, particularly massively blooming trees, only that can be matured is usually limited by resource availability, not by pollination. Furthermore, flower and fruit abortion is selective: some species selec- tively shed self-pollinated flowers. They mature fruits from self-pollinated flowers only when fruit set is low and/or when “higher quality" fruits from cross-pollinated flowers are removed. Therefore there may be “mate competition" within a plant among fruits of different paternity. In this study attempts were made to reduce, to the extent practical, the effects of resource limi- tation and competition on mating success. All flow- ers except the hand-pollinated ones were removed from the inflorescence. Once a few fruits were set n an inflorescence, all flower buds were removed. Until fruits were set, buds were left to develop into flowers available for further pollination trials. Most inflorescences bloomed continuously for several weeks, producing a few new flowers each day, so failed matings could be attempted repeatedly. An individual inflorescence was treated with pol- len from a single source. This eliminated mate competition among the flowers within the inflores- cence. An individual tree often had several inflo- rescences, each pollinated with a different species of male parent, so there could have been interin- florescence competition among mates o further reduce resource competition and channel available nutrients into the hand-pollinated owers, most untreated inflorescences and spon- taneous, open-pollinated fruits (those accessible with clipper poles) were removed from the crowns of the trees. In all species, seeds matured approximately 60 days after pollination. At maturity the hybrid fruit was removed and the number of mature seeds, aborted seeds, and undeveloped ovules was re- corded. Length and width of each mature seed were measured for comparison with seeds produced from intraspecific matings. Tests for Self-Compatibility, Autogamy, and Apo- mixis. or self-pollinations and intraspecific out- crosses, anthers were not emasculated, but the corolla standard of the flower bud was sealed with tape prior to dehiscence to prevent chance depo- sition of nonself pollen on the stigma. When the stigma became receptive, pollen from the same tree (for selfs) or from different conspecific trees (for outcrosses) was applied, and a cap of aluminum foil was placed over the stigmas in the same manner as in the intraspecific hybridizations. The pollen for the outcrosses was a mixture from all available conspecific trees in the botanical gar- den, including individuals from the same accession as the female parent as well as from different acces- sions. For the self-compatibility trials carried out in the natural populations of Erythrina chiapa- sana and E. goldmanii at El Sumidero, the pollen for the outcrosses was a mixture of at least five different individuals in the population. The treated flowers of an individual inflorescence were either all selfed or all outcrossed to eliminate within-inflorescence mate competition. he abortion of young fruits during the first two to three weeks following fertilization was very high for selfs and intraspecific outcrosses, as well as for interspecific hybridizations. Fruit set data were tak- en at least four weeks following pollination, after which abortion of the developing fruits was negli- gible. Complete data on intraspecific reproductive success, including mature seed production and seed size, germination success, and viability of the prog- eny, were obtained only for cultivated Erythrina guatemalensis and E. crista-galli. Several indi- viduals of were available for the trials, and they were the most successful female parents in the interspecific hybridizations. For analyses, then, the intraspecific data were pl desirable for these species from different accessions these two species. Because of space and labor lim- itations, intraspecific progeny could not be raised for all species. In the flowers of sect. Erythrina and the other hummingbird-pollinated sections of the genus, the anthers and stigma are positioned close to one another. though the flowers are protandrous, autogamy may Initial observations indicated that, al- sometimes take place. Autogamy was tested by isolating entire inflorescences in wire-framed nylon mesh bags. After all the flowers had either aborted or set fruit, the mesh was removed. Six species were tested this way at Pacific Tropical Botanical Volume 75, Number 3 1988 Neill 903 Erythrina Garden during a period of relatively dry weather (May 1982) to minimize abortion of flowers caused by high humidity inside the mesh bags. Autogamous fruits were obtained only on the most distal flowers (the last to open) on inflores- cences of two individuals of Erythrina guatema- lensis (Results, Table 4). These individuals were tested for agamospermy. On three inflorescences of each plant, all the flowers on the distal one-third of the inflorescence were emasculated before de- hiscence, and the stigmas were covered with alu- minum foil caps to prevent any pollen deposition on the stigma. Fruit set was monitored in the same manner as in the pollination trials. Statistical Analysis of Results. scribed above was necessitated by the flowering patterns and reproductive traits of Erythrina, by practical limitations of breeding trees in the erogeneous environments of open-air botanical gar- dens, and by the goal of obtaining viable progeny of as many “° ” “medium,” wide” hybrid combinations as lis: The resulting small and very unequal sample sizes for different hybrid combinations as well as for selfings and intraspecific outcrosses meant that outcomes of particular com- binations could not be compared statistically. In- stead, for statistical analyses hybrid combinations and intraspecific matings were pooled into broad categories based on taxonomic distance (assumed for the purposes of the study to be a true repre- sentation of relative genetic and phylogenetic dis- tance) between the female and male parents. The five experimental treatments are: self-matings, in- traspecific outcrosses, and the three categories of > webp combinations— narrow," “medium,” and wide” hybridizations. u For statistical analyses, mating success for each treatment was expressed as the proportion of hand- pollinated flowers producing mature fruit (i.e., a fruit with at least one fully developed, normal-sized seed). The commonly used analysis-of-variance (ANOVA) tests (e.g., Sokal & Rohlf, 1969; Sta- tistical Analysis Institute, 1982) are designed to test the significance of differences between means of continuously variable data. ANOVA tests are inappropriate for categorical (either /or) data, such as mating success, where the outcome of a polli- nation attempt falls into one of only two categories. A multiple comparison test for differences between proportions, appropriate for categorical data, was devised by Alan R. Templeton for these analyses. Templeton's test allows for pairwise comparisons of all combinations of the five treatment categories; also, categories can be pooled to test various hy- The protocol de- potheses regarding mating success (e.g., all intra- specific vs. all interspecific matings). The null hypothesis for the test was that there is no difference in proportion of mature fruits pro- duced among any of the pollination treatments. This is a corollary of the central hypothesis of this research: that there are no interspecific or self- incompatibility barriers to mating within Erythri- na, that any pair of gametes from any species in the genus are equally likely to pair successfully, form a viable zygote, and grow into a healthy adult sporophyte regardless of the infrageneric taxonom- ic position or putative phylogenetic distance be- tween the parents. Templeton’s test is an inequality that compares the differences between proportions with their vari- ances. Proportions are subjected to an arcsine— square root transformation to set the variance in- dependent of the mean; the variance is inversely proportional to the sample size. The 95% confi- dence limits of the proportion are: F 1 = : +] X = arcsin VE: .96 VA: where F — number of mature fruits (successful matings); N — number of flowers pollinated (at- tempted matings) or small sample sizes (N < 50), the arcsine— square root transformation is corrected: F 1 X=- (ari " i F +1 arcsin NE 1): For the general case, the null hypothesis is re- jected at P = 0.05 if the inequality is true: | 2 a, X, where X, = the arcsine-square root transformed proportion of successful matings in the ¿th cate- gory; N, = sample size (total number of flowers pollinatedi in the ith eni and a, — a weighting factor set so that | a| = a? > 0.98 —, VE The a, for each category is ecc to the sample size N,. For pairwise comparisons between categories ¿ and j, the inequality is simplified; the H, is rejected at P — 0.05 if it is true that ELE IX, - X| > 098 V/x tN For a test of “highly significant” difference at 904 Annals of th Missouri Boos Garden P = 0.01, the term “0.98” on the right side of the inequality is replaced by the value “1.28.” Several individuals of Erythrina guatemalensis and E. crista-galli were the most fecund, suc- cessful females in the interspecific hybridizations as well as the intraspecific matings. For all cate- gories employing these two species as female par- ents, separate multiple comparison tests were used to compare mating success. These comparisons included both selfings and intraspecific outcrosses for E. guatemalensis but only selfings for E. cris- er which had only one individual in flower at each garden (Pacific Tropical and Waimea) when the pollinations were conducted. Separate statistical tests were carried out for the self-compatibility trials of individual species. For those species with analyzable data on fruit set of self-pollinations vs. intraspecific outcrosses, the data were ordered into 2 X 2 contingency tables. With small sample sizes and values of less than 5 in many cells of the contingency tables, the standard chi-square test was not appropriate; so Fisher's or the outcomes or the pooled self-com- patibility data including all species tested, the sam- ple size was large enough for a chi-square test. F, Hybrid Viability. planted within a few weeks after harvest. The F, hybrid seeds were To the extent possible, seed lots of each hybrid combi- nation were divided for propagation at two sites. The F, plants were raised by the horticulturists at Pacific Tropical and Ho‘omaluhia Botanical Gar- dens, who monitored germination success, growth, and vigor of the hybrids. Evaluations were made approximately once each six months using a stan- rdized form. Survivorship, growth rates, and in- dications of chlorosis or other abnormalities were recorde The multiple comparison test described above for analysis of fruit set was employed for a statis- tical evaluation of hybrid viability, defined as the proportion of seeds in each category that germi- nated and survived as healthy plants for six months (after which mortality in the garden was negligible). Viability of the **narrow, and "wide" F, hybrids was compared, together with that of the "narrow" F, hybrids in sect. Erythrina (see below). Seeds from controlled intraspecific matings of Er- ythrina guatemalensis and E. crista-galli were planted along with the hybrids. For each of these species a separate multiple comparison test was conducted for viability of “intraspecific” seed vs. hybrid seed having these two species as female parents. F, Hybrid Fertility. Many of the narrow hybrids between species in sect. Erythrina, produced in 1982, grew to be 4-m trees and produced flowers by February 1984. By several different measures of fertility, these F,s were compared with the pa- rental species and their meiosis examined. As an estimate of pollen fertility, percentage of stainable (nonaborted) pollen was determined for the F, hy- brids and their parents using Alexander's double stain technique (Alexander, 1969) (at least 500 grains counted per sample) and compared with a one-tail t-test Fecundity of F, Hybrids. As discussed earlier, one goal of the hybridization study was to assess the relative fitness of the The viability, vigor, meiotic ybrids in comparison with their parents. regularity, and pollen fertility of the hybrids are indicators of fitness, but a more direct measure is their relative reproductive success vis-à-vis that of the parental species. During the period of this study the F, hybrids, although some of them produced flowers and fruits, did not grow into full-sized adult trees, so a thorough assessment of hybrid fecundity and fitness was not possible. However, a prelimi- nary indication of reproductive success was ob- tained from the two-year-old narrow hybrids in sect. Erythrina that flowered in the spring of 1984. Controlled self-pollination of some of these F, hybrids was conducted in order to obtain seed for a limited number of F, families. The multiple com- parison test was employed for pairwise comparisons of mating success (proportion of hand-pollinated flowers producing mature fruits) between the selfe Fs and their parental species. Different categories of parental matings varied among themselves in mating success, and several types of parental mat- ings were compared with the selfed F,s. The pair- 1) selfed Vs vs. their own parental matings, i.e., the original hybridizations that produced the F,s used in the trials; 2) selfed F,s vs. all paired combinations of the parental species, including the reciprocals that failed to produce F, hybrids; 3) selfed F,s vs. selfed parental species. The logic for using these partic- wise comparisons of fruit set included: ular groupings of parental matings in the compar- ative assessment of F, reproductive success is dis- cussed in the Results. Viability of F, Hybrids. Many of the Fs also produced fruit and mature seed spontaneously on open-pollinated inflorescences, almost certainly the result of autogamy. F, seed lots from selfed flowers, and some from open-pollinated flowers, were plant- ed along with the 1984 F, hybrids. The viability Volume 75, Number 3 1988 Neill 905 Erythrina of the F, progeny was compared with that of the Fs. Studies of Previously Synthesized Hybrids. A few Erythrina hybrids have been produced in the past by horticulturists and are commonly grown in tropical and subtropical regions. Two were available at the Hawaiian gardens: Erythrina x bidwillii and E. X sykesii. Studies of meiosis, pollen fertility, and fruit set from controlled self-pollinations were conducted on these plants with the methods de- scribed above. Hybrid Names. In this paper horticultural con- vention (Brickell et al., 1980) is followed for the hybrid names. For artificial hybrids when the fe- male parent is known, the female parent is first in the hybrid formula name. When the female parent is not known, as in natural hybrids, the order of the constituent species names is alphabetical. RESULTS AND DISCUSSION The results of the experimental hybridizations, self-compatibility trials, and studies of viability and fertility of the progeny are presented in summary form for the statistical analyses of the data. In addition, more complete data sets, listing the results obtained from individual plants, are presented in certain of the tables below. There are several reasons for this more thorough reporting of the data. The first is to provide the most complete information available on the ances- try of each individual in the F, and subsequent hybrid generations. The full documentation is nec- essary for the studies on the inheritance of various traits in the hybrids. Studies of morphological in- heritance were initiated in this paper (Section 5), and research on the inheritance of micromolecular and macromolecular traits in the interspecific hy- brids is anticipated for the future. In addition, some of the hybrid plants with their colorful flowers are likely to be propagated widely as ornamentals; the tables presented here serve as public documenta- tion of the parentage of these cultivars. Finally, it is hoped that some of these experiments will be repeated with the same parental and hybrid trees in the Hawaiian botanical gardens. The documen- tation of the results for individual plants of traits probably indicative of reproductive success, such as pollen fertility and fruit set, will allow investi- gation of the possibility that such traits may change through time with the maturation of the plant. Self-Compatibility. The results of the self-com- patibility trials are shown in Table 3. The low percentage of mating success in the pooled data for all species— 6^6 fruit set for selfings and 10% for outcrosses—is due to a high incidence of postzy- gotic abortion of young fruits and to failure of pollen tubes to reach the ovules, but the relative impor- tance of these two factors is not known. For the pooled totals, the difference in fruit set between selfs and outcrosses is nonsignificant. Statistical comparison of fruit set in selfs vs. outcrosses was possible in four species. In only one of these, the natural population of Erythrina gold- manii at El Sumidero, was fruit set significantly higher in outcrosses than in selfs, and then but marginally so at P — 0.05. In six additional species, self-pollinated flowers set fruit, but only one individual of the species was available, so the outcrossing control was not pos- sible. In nine species, no fruits were set from self- pollinated flowers, but in four of these the out- crossing controls yielded no fruits either. Failure of fruit set, then, is evidently a consequence of overall low fecundity in Erythrina and not of self- incompatibility per se. Self-incompatibility has previously been report- ed for seven species of Erythrina: E. senegalensis and E. speciosa (East, 1940); E. crista-galli (Fryxell, 1957); E. mitis and E. poeppigiana (Ar- royo, 1981); E. leptorhiza (Hernández & Toledo, 1979); and E. montana (Hernández, 1982). Only for E. montana was the assertion of self-incom- patibility supported by evidence from experimental self-pollinations and outcrossing controls. Calcu- lation of Fisher's exact probability for the data presented in Hernández (1982), however, reveals that the difference in fruit set between selfs and outcrosses in E. montana is nonsignificant (P = 0.25). My evidence for self-compatibility in E. sen- egalensis and E. crista-galli contradicts the ear- lier reports of self-incompatibility in these species, which were based merely on the failure of isolated cultivated trees to produce seed spontaneously. Feinsinger et al. (1979) provided evidence from experimentally controlled pollinations that E. fusca and E. pallida are self-compatible. There is thus no reliable evidence for genetic self-incompatibility in any species of Erythrina. It appears safe to assume that genetic self-incom- patibility—at least the classical single-locus, mul- tiple S-allele, stigma- or style-mediated model of self-incompatibility (Nettancourt, 1977)—is com- pletely absent from the 112 species in the genus. If this is true, it would invalidate some of the evidence that Arroyo (1981) advanced to support her assertion that tropical woody Papilionoideae are predominantly self-incompatible. Five of the 906 Annals of the Missouri Botanical Garden TABLE 3. Self-compatibility trials in Erythrina. Self Outcross Probability, Species! Flowers Fruits? Flowers Fruits? Self vs. Cross E. berteroana (4) 27 0 9 0 — E. chiapasana (2) 14 0 — — — E. chiapasana* (6) 32 4 (0.13) 15 1 (0.07) 0.48* E. crista-galli (1) 27 7 (0.26) = — E. elenae (1) 27 — — — E. falcata (1) 5 1 (0.20) — — — E. folkersii (1) 2 2 (1.0) — — — E. fusca (2) 82 1 (0.01) 85 1 (0.01) > 0.05** E. goldmanii® (6) 27 3 (0.11) 23 8 (0.35) 0.05* . guatemalensis (5) 33 3 (0.09) 28 7 (0.25) 0.09* E. latissima (1) 10 0 = — — E. lysistemon (1) 25 4 (0.16) — — — E. perrieri (2) 27 0 l 0 == E. sandwicensis (1) 49 2 (0.04) — — — E. senegalensis (1) 28 2 (0.07) — — — E. speciosa (2) 21 0 8 0 — E. standleyana (1) 9 0 — — — E. tahitensis (2) 64 0 _ _ — E. variegata (1) 6 0 6 Total 515 29 (0.06) 175 17 (0.10) > 0.05** ' In parentheses: number of individuals used in trials. * [n parentheses: proportion of pollinated flowers producing mature fruits. * Pollinations conducted in natural population at El Sumidero, Chiapas, Mexico. * Fisher's exact probability. ** Chi-square probability. 27 species Arroyo listed in that habitat/life form category as self-incompatible were Erythrina species. There are very few comparable studies on other genera of tropical woody Papilionoideae. With the information presently available, it is not known if Erythrina is an anomaly, or if self-compatibility is common in this group of plants. Because low fecundity and high rates of flower and fruit abortion are probably characteristic of these plants, greater caution is required in carrying out and interpreting self-incompatibility tests than has customarily been taken. It is true that fruit set is frequently lower in self-matings than in outcrosses. This may be due not to genetic self-incompatibility, but rather to multiallelic inbreeding depression, expressed either in the progamic phase as failure of pollen tubes to reach the ovules (Mulcahy & Mulcahy, 1983) or as postzygotic abortion of young fruits. Although Erythrina species are genetically self- compatible, the production of seed from selfed flow- ers in natural populations may be quite limited. A flowering tree visited by pollen-bearing birds will receive many geitonogamous pollinations (pollen from a different flower on the same individual) as well as xenogamous pollinations (pollen from a dif- ferent individual). The reduced fruit production from self-pollinations, as well as the relatively poor viability of selfed seed (see section on F, viability below) suggests that progeny derived from selfed relative to progen derived from outcrossed flowers. The selective abortion of low-quality selfed fruits, cited by Ste- phenson (1981), may be operative in Erythrina. Interfruit competition may be very intense under natural conditions, since such a small proportion flowers are low in of pollinated flowers develops into mature fruits. Therefore it is possible that most successful prog- eny are derived from outcrossing, and that the level of inbreeding in most Erythrina populations is quite low in spite of self-compatibility and a large proportion of geitonogamous pollinations. This is still speculative; the significance to mating success of mate competition among male parents has not been explored in Erythrina. In regard to flower and fruit abortion, the at- tempts to increase mating success by eliminating the effects of competition and resource limitation in the experimental pollination trials were only par- tially successful. Certainly the fruit maturation rates of 25% or more obtained in some of the outcrossing trials represent an increase in fruit production over Volume 75, Number 3 1988 Neill Erythrina TABLE 4. Tests for autogamy and agamospermy in Erythrina at Pacific Tropical Botanical Garden. I. Test for autogamy Inflorescences Species Accession Number Bagged Flowers Fruits Seeds E. abyssinica 770034001 3 90 0 0 E. berteroana 700044001 3 124 0 0 E. crista-galli 740283001 3 152 0 0 E. guatemalensis 720999001 2 165 1 2 E. guatemalensis 720999002 3 96 0 0 E. poo eue 150419001 3 148 5 8 E. hum 740283001 4 156 0 0 E. poo lola 750420001 4 86 0 0 E. salviiflora 721000002 2 63 0 0 II. Test for agamospermy (in individuals exhibiting autogamy) Inflorescences Flowers Species Accession Number Treated Emasculated Fruits Set E. guatemalensis 750419001 3 30 0 E. guatemalensis 720999001 3 66 0 the percentages found in natural populations. Usu- ally the percentages of fruit maturation were much lower, however, and in all cases the majority of pollinated flowers were aborted early in develop- ment. The factors promoting flower and fruit abor- tion are several, including nutrition and resource limitation, competitive effects, possible damage to the flowers caused by emasculation, and factors such as adverse weather conditions, in addition to the factor under consideration here: the genetic compatibility of the female and male parents. Nei- r the self-compatibility trials nor for the experimental hybridizations was it possible to sort out all of these variables. Autogamy and Agamospermy. The results of the tests for autogamy and agamospermy are shown in Table 4. Autogamous fruits were produced only two individuals of Erythrina guatemalensis. Significantly, these were rather “fecund” trees with relatively high mating success from controlled hand- pollinations. The autogamous fruits were produced only from the uppermost three fascicles of flowers on an inflorescence—the last flowers to bloom. They evidently were produced, in part, because of the occasional breakdown of protandry, which pre- vents autogamy on most flowers of E. guatema- lensis and the other species that are adapted to hummingbird pollination (Neill, 1987). Although autogamous fruits were produced in my limited trials only on Erythrina guatemalensis, I believe that occasional autogamy is widespread in the genus. Cultivated trees of many species produce some fruits spontaneously in the absence of evident pollen vectors. The fact that autogamous fruits are produced only on the latest flowers of an inflorescence sug- gests that the breakdown of protandry may be an adaptive mechanism that allows some seed set in the absence of the appropriate avian pollen vectors. Each inflorescence, although it may produce 75 or more flowers, will mature only a few fruits, so fruit set on the lower, earlier-blooming flowers of the inflorescence must inhibit the formation of fruits on the upper, later-blooming portion. It is likely that autogamous fruits from the ultimate flowers of the inflorescence will be produced only if some allogamous fruits (from either xenogamous or gei- tonogamous pollinations) have not already been produced on the lower portion of the inflorescence. The two Erythrina guatemalensis trees that produced autogamous fruits were tested for agamo- spermy (Table 4). No fruits were produced when stigmas were isolated from pollen deposition, so agamospermy is not indicated. Agamospermy is almost unknown in the Leguminosae Tr 1981) and it is unlikely to occur in Erythrin Hybridization Trials: Mating Success (Fruit Maturation). The complete results of the h bridization trials are listed in Appendix 1,671 hybridization attempts in 155 hybrid com- binations, 98 mature fruits were produced in 47 hybrid combinations, for an overall hybrid mating success of 6%, in 30% of the attempted combi- nations. 908 Annals of the Missouri Botanical Garden TABLE 5. Proportion of hand-pollinated flowers producing mature fruit: all diploid Erythrina species. Flowers Fruits Proportion Pollination Treatment Pollinated Matured Fruit Set Selfed 515 29 5.6% Intraspecific outcr 175 17 9.7% Narrow yban (within section) 540 50 9.3% Medium hybridization (between sections, within subgenus) 350 22 6.3% Wide hybridization (between subgenera) 705 25 3.6% Total 2,285 143 6.3% Multiple comparison test for differences between treatments in proportion of fruit set Self vs. outcross N.S.! Self vs. narrow hybrid P « 0.05 Self vs. medium hybrid N.S. Self vs. wide hybrid N.S. Outcross vs. narrow hybrid N.S. Outcross vs. medium hybrid N.S. Outcross vs. wide hybrid P « 0.01 Narrow hybrid vs. ineditis hybrid N.S. Narrow hybrid vs. wide hybrid P « 0.01 Medium hybrid vs. wide hybrid N.S. Intraspecific vs. hybrids N.S. Outcross + narrow hybrids vs. self + medium + wide P « 0.01 ' N.S. = not significant (P > 0.05). For the statistical analysis of mating success in selfs, intraspecific outcrosses, and hybridizations (Table 5), data from the diploid species only were included. The tetrapoloid Erythrina amazonica as male parent, after numerous pollination attempts with the diploids E. guatemalensis and E. crista- galli as female parents, produced one hybrid seed with each of the females. Neither of the seeds germinated, however, so there are no successful hybrids between Erythrina species of different ploi- dy levels. Since the results from E. amazonica are not germane to the hypotheses of the interfertility of diploid species and the formation of homogamic Mes aridi they were excluded from the statistical analys The pe for the diploid species in Table 5 indicate that the highest mating success was ob- tained with intraspecific outcrosses and "narrow" hybridizations (within sections). "Medium" hybrid- izations (intersectional, intrasubgeneric) and self- and " (intersubgeneric) hybrids were the least successful of the five treatment classes. A general trend, then, is interspecific matings be- tween closely related species (within sections) are evident: nomic distance” between the parents (intersection- al and intersubgeneric hybridizations). Mating suc- cess is also somewhat lower in selfings than either intraspecific outcrosses or hybridizations between closely related species. his overall trend, shown by the percentages of fruit maturation in Table 5, is not a strong one; the differences between treatments are for the most part nonsignificant. The multiple comparison test revealed only three significant differences among the ten possible pairwise combinations. Fruit mat- uration in narrow hybridizations was significantly higher than in self-mating (P < 0.05). Intraspecific outcrosses and narrow hybridizations also had higher fruit maturation than wide hybridizations; in these comparisons the difference was highly significant (P < 0.01 Because the pooled fruit maturation data for all species may obscure the heterogeneity in results among different species, it is instructive to examine the patterns of mating success in a few selected species. Erythrina guate s as female parent accounted for 30% of all mature fruits in the pollination trials and 33% of all the hybrid fruits. Thirty-three hybrid fruits were produced from E. guatemalensis as female; of these, 30 (9095) were from a single genetic individual, a clone represented by one tree at each garden (PT 700018001, WA 74c1453). The pattern of mating success for Er- ythrina guatemalensis (Table 6) is very similar to the overall results for the combined species trials. Volume 75, Number 3 1988 Neill 909 Erythrina TABLE 6. malensis; male parents — diploid species. Proportion of hand-pollinated flowers producing mature fruit: female parent — Erythrina guate- Flowers Fruits Proportion Pollination Treatment Pollinated Matured Fruit Set Selfed 33 3 9% Intraspecific outcross 28 7 25% Narrow hybridization (within section) 86 21 24% Medium hybridization (between sections, within subgenus) 54 5 9% Wide hybridization (between subgenera) 185 7 4% Total 386 43 11% Multiple comparison test for differences between treatments in proportion of fruit set Self vs. outcross N.S Self vs. narrow hybrid N.S. Self vs. medium hybrid N.S. Self vs. wide hybrid N.S. Outcross vs. narrow hybrid N.S. Outcross vs. medium hybrid N.S. Outcross vs. wide hybrid P < 0.01 Narrow hybrid vs. medium hybrid N.S. Narrow hybrid vs. wide hybrid P < 0.05 Medium hybrid vs. wide hybrid N.S. Outcross + narrow hybrid vs. self + medium + wide P < 0.01 The E. guatemalensis trees were unusually fecund; fruit maturation was much higher than the overall average for intraspecific outcrosses (25%) and nar- row hybridizations (24%). As in the combined species results, fruit maturation was significiantly igher in intraspecific outcrosses and narrow hy- bridizations than in wide hybridizations (P « 0.01 for both pairwise comparisons), and for narrow hybridizations vs. medium aoe gui the dif- ference was marginally significant (P < 0.05). The data set for Erythrina j as fe- male parent (Table 7), although not so extensive, shows that the neat congruence of mating success and taxonomic distance evidenced by FK. guate- malensis does not always apply. Erthyrina crista- galli as female, represented by one genetic indi- vidual at each of the two gardens, produced 16% of all the hybrid fruits in the trials, but it produced 49% of the “medium” and "wide" hybrid fruits. Fruit maturation was higher in the medium an wide hybridizations than in the few narrow hybrid- izations. (Only one species, Erythrina falcata, is in the same section with E. crista-galli, so the opportunities for narrow hybridization were limit- ed.) There are no significant differences, however, in any of the pairwise comparisons between polli- nation treatments for E. crista-galli; the variances are large because the sample sizes are rather small. For intersectional hybridizations (the “medium” and “wide” categories combined) mating success in Erythrina crista-galli as female parent was significantly higher than in E. guatemalensis (mul- tiple comparison test for proportions, P « 0.05). This is illustrated by the results of attempted re- ciprocal hybridizations between these two species, which are in different subgenera. Sixty-four polli- nation attempts to produce the hybrid E. guate- malensis $ X E. crista-galli à yielded a single fruit with two seeds, neither of which germinated. Only ten attempts at the reciprocal cross of E. crista-galli € x E. guatemalensis à yielded four fruits, 15 seeds, and eight vigorous F, plants. In all, E. crista-galli as female parent produced seeds from seven hybrid combinations with species in six sections and three subgenera. Five of these com- binations in all three subgenera survived as healthy F, plants. Erythrina crista-galli, in short, was a singularly successful **wide hybridizer.” Summing up the contrasting results in fruit mat- uration for Erythrina guatemalensis and E. cris- ta-galli, E. guatemalensis hybridized very readily with species in the same section, much less so with more distantly related species. Erythrina crista- galli, in contrast, hybridized with a number of species in different subgenera with apparently equal facility, regardless of the formal taxonomic rela- tionships and presumed phylogenetic affinities be- tween E. crista-galli and the male parents. These 910 Annals of the Missouri Botanical Garden TABLE 7. Proportion of hand-pollinated flowers TABLE 8. Viability of hybrid Erythrina: proportions producing mature fruit: female parent = Erythrina cris- ta-galli; male parents = diploid species. of seeds germinating and growing into healthy plants (at 6 months). Flow ers Fruits portion portion Polli- Ma- Frui Seeds Live Via- Pollination Treatment nated tured Set Type of Hybrid Seed Sown Plants bility Selfed 27 7 26% Narrow F, hybrid 167 86 51% Narrow a Narrow F, hybrid (all (within section) 22 2 9% within sect. Erythrina) 66 7 11% Medium hybridization (be- Medium F, hybrid 62 16 26% tween sections, within Wide F, hybrid 44 18 41% subgenus) 69 11 16% Total 339 127 37% Wide hybridization (be- tween subgenera) 64 10 16% Multiple comparison test for differences in Total 182 30 16% viability between types of hybrid Multiple comparison test for differences between treatments in proportion of fruit set Self vs. narrow hybrid N.S Self vs. medium hybrid N.S Self vs. wide hybrid N.S Narrow hybrid vs. medium hybrid N.S Narrow hybrid vs. wide hybrid N.S. Medium hybrid vs. wide hybrid N.S. differences in mating success probably reflect in- dividual variation rather than real and consistent interspecific differences in crossability. F, Hybrid Viability. Viability of the F, hybrids was high and equal to or higher than normal via- bility of progeny within species. Overall, 143 (52%) of the 273 F, hybrid seeds germinated; 120 of these (44% of the total) survived as healthy F, plants. In other words, most of the F, seeds either germinated and grew vigorously or they did not germinate at all: the survival rate of those that germinated was 84%. There we hybrids. Two individuals completely lacked chlo- re few instances of weakness in the roplasts and died soon after germination: a narrow hybrid, Erythrina macrophylla x E. berteroana, and a wide hybrid, E. crista-galli x E. speciosa. In both cases, though, siblings from the same cross grew into healthy green plants. About 10 other hybrid plants in different combinations were chlo- rotic, with yellow-green foliage, and died within 1— 2 months. Several others at first appeared chlo- rotic, but after several months they recovered the normal green color and grew vigorously. The comparative viablity of narrow, medium, and wide hybrids is summarized in Table 8. Among the F, hybrids viability (defined as successful ger- F, narrow vs. F, narrow P « 0.01 F, narrow vs. F, medium P « 0.01 F, narrow vs. F, wide N.S. F, narrow vs. F, medium N.S. F, narrow vs. F, wide P « 0.01 F, medium vs. F, wid N.S. All F, hybrids vs. F, narrow hybrids P < 0.01 mination and survival of the plant for at least six months) was highest for narrow hybrids, interme- diate for wide hybrids, and lowest for medium hy- brids. The difference between viability of narrow and medium hybrids was statistically significant P < 0.01). but between narrow and wide hybrids it was not. Included in Table 8 is the viability data for the "narrow F,” hybrid seed produced in 1984 from the two-year-old narrow Fs within sect. Erythrina. — Germination success of the F,s was very low, an unexpected and anomalous result; the difference in viability between both the narrow F, hybrids (51%) and the wide F,s (41%) vs. the narrow F,s (1176) was highly significant (both comparisons, P « 0.01). Viability data for the intraspecific and hybrid progeny of maternal Erythrina guatemalensis and E. crista-galli, respectively, are presented in Ta- bles 9 and 10. The intraspecific progeny of these two species were grown for two purposes: to com pare their viability with that of the hybrids from the same female parents, and to carry out a study of intraspecific variation of morphological traits among siblings. The second goal was thwarted, however, because of the poor germination of the intraspecific seeds. Although the seed lots were not large to begin with, viability of the seed derived from selfings was particularly low: all seven selfed Volume 75, Number 3 Neill 911 1988 Erythrina TABLE 9. Viability of seed produced from Erythrina TABLE 10. Viability of seed produced from Ery- guatemalensis as female parent (intraspecific and hy- brids). Proportion of seed germinating and growing into healthy plants (at 6 months) . thrina crista-galli as female parent (selfs and hybrids) . Proportion of seeds germinating and growing into healthy plants (at 6 months) . portion Seeds Live Via- Paternity of Seed Sown Plants bility Selfed 12 1 8% Intraspecific outcross 14 1 29% Narrow hybrid (within sect. Erythrina) 93 39 42% Medium ie: wide hybrids 19 2 11% Total 138 46 33% Multiple comparison test for differences in viability between seed of different paternity; female parent = Erythrina guatemalensis Self vs. outcross N.S. Self vs. narrow hybrid P < 0.05 Self vs. medium & wide hybrid N.S. Outcross vs. narrow hybrid N.S. Outcross vs. medium & wide hybrid N.S. Narrow hybrid vs. medium & wide hy- brid P « 0.01 Self vs. all hybrids P « 0.01 Outcross + narrow hybrid vs. self + E diua + wide hybrids P « 0.01 seeds of E. crista-galli failed to germinate, as did all but one of 12 selfed seeds of E. guatemalensis. This could be an expression of inbreeding depres- sion in the self-progeny, but this possibility must be corroborated with larger samples. Among the hybrids derived from Erythrina gua- temalensis and E. crista-galli females, the pattern of F, viability (Tables 3: 10) and its oa: similar to the pattern of mating success ol earlier (Tables 6, 7) for the same two species. mong the progeny of E. guatemalensis, viability of narrow hybrids was significantly higher than that of medium and wide hybrids. Among the progeny of E. crista-galli, by contrast, there was no cor- relation between F, hybrid viability and the degree of relatedness of the parentals. A complete listing of the F, hybrid plants pro- duced during 1982-1984 is contained in Tables 11-13. In all, there are 120 individuals in 33 hybrid combinations (2176 of the 155 attempted combinations): 22 narrow hybrid combinations (34% of 65 attempted combinations); four medium hy- brid combinations (15% of 27 attempts); and seven ropor- Paternity of Seeds Live tion Via- Seed Sown Plants bility Selfed y 0 0% Narrow hybrid 4 1 25% Medium hybrid 33 7 21% Wide hybrid 24 10 42% Total 70 18 26% Multiple comparison test for differences in viability between seed of different paternity; f emale parent = Erythrina crista-galli Self vs. narrow hybrid N.S. Self vs. medium hybrid P « 0.01 Self vs. wide hybrid P « 0.01 Narrow hybrid vs. medium hybrid N.S. Narrow hybrid vs. wide hybrid N.S. Medium hybrid vs. wide hybrid N.S. Self vs. all hybrids P « 0.01 wide hybrid combinations (11% of 64 attempts). The number of individual F, hybrids for each com- bination ranges from one to nine Nineteen of the 22 narrow hybrid combinations are between species in sect. Erythrina. There is one narrow hybrid combination in each of sections Cristae-galli, Chirocalyx, and Erythraster. The medium and wide hybrids include species combi- nations in nine of the 26 sections of Erythrina and four of the five subgenera. In seven of the hybrid combinations, one parental species is native to the New World and the other is native to the Old World. In summary, the viable F, hybrids obtained be- tween the diploid species of Erythrina include rep- resentative crosses that bridge the entire range of taxonomic diversity and geographic distribution of the genus. Interspecific crossability appears to be largely a function of individual variation in fecun- dity of the female parent and only partially a func- tion of taxonomic/phylogenetic distance between male and female parents. Given th Its obtained in these experiments, it may be expected that with sufficient time, perseverance, and selection of com- patible and fecund individual genotypes, any diploid Erythrina species could be crossed with any other to produce a viable F, hybrid. Sexual Maturation and Fertility of F, Hy- brids. Some F, hybrids not only were very rapid 912 Annals of the Missouri Botanical Garden TABLE 11. Artificial Erythrina hybrids: narrow (intrasectional) . Hybrid Hybrid Live Accession Parental Accession Hybrid Number Plants Numbers* Numbers* Sect. Cristae-galli E. crista-galli x E. falcata HO 84.284 PT 740283001 (F) PT 750086001 (M) Sect. Erythrina E. americana X E. berteroana 25x PT 820420 WA 75c1171 (F) WA 74s864 (M) E. berteroana x E. guatemalensis 53x43-1 6 HO 82.647 WA 78s564 (F) PT 820549 WA 74c1453 (M) E. chiapasana X E. berteroana 36x53-1 8 HO 82.278 PT 721005001 (F) PT 820283 PT 700044002 (M) E. goldmanii x E. chiapasana 29 x 36-1 1 WA 84c560 Neill 5617 (F) Neill 5497 (M) E. guatemalensis x E. berteroana 43x53-1 2 HO 82.289 PT 700018001 (F) PT 700044001 (M) E. guatemalensis x E. berteroana 43x53-3 6 HO 82.641 PT 750419001 (F) PT 820493 PT 730711001 (M) E. guatemalensis X E. berteroana 43x53-4 l HO 82.642 PT 720999001 (F) PT 700044001 (M) E. guatemalensis X E. chiapasana 43 x 36-1 4 HO 82.283 PT 700018001 (F) PT 820254 PT 721005001 (M) E. guatemalensis X E. chiapasana 43 x 36-2 5 HO 82.284 PT 700018001 (F) PT 820278 PT 730710001 (M) E. guatemalensis x E. folkersii 43x31-1 2 HO 82.282 PT 700018001 (F) PT 700010001 (M) E. guatemalensis x E. macrophylla 43x42-1 4 HO 82.285 PT 700018001 (F) PT 750420002 (M) E. guatemalensis x E. macrophylla 43x 42.3 2 PT 820276 PT 750420002 (M) E. guatemalensis x E. macrophylla 43x42-4 2 HO 82.288 PT 750420002 (M) E. guatemalensis X E. salviiflora 43x50-1 1 PT 820492 PT 750419001 (F) PT 721000002 (M) E. guatemalensis X E. standleyana 43x 23-3 2 HO 82.765 WA 74c1453 (F) WA 76s1056 (M) E. guatemalensis X E. tajumulcensis 43x40-1 1 HO 82.335 WA 74c1453 (F) WA 74c1448 (M) E. guatemalensis x E. tajumulcensis 43 x 40-4 2 HO 82.640 WA 74c1448 (M) E. guatemalensis x E. tajumulcensis 43 x 40-5 5 PT 820546 WA 74c1448 (M) E. guatemalensis X E. tajumulcensis 43x40-6 1 PT 820547 WA 74c1448 (M) E. herbacea x E. americana 22x 25-2 2 HO 82.759 WA 75c1103 (F) WA 75c1171 (M) E. herbacea x E. berteroana 22x53-1 l PT 820541 WA 75c1103 (F) WA 74s864 (M) E. herbacea x E. guatemalensis 22x 43-1 2 PT 820421 WA 75c1103 (F) WA 74c1453 (M) E. macrophylla x E. americana 42x25-1 1 PT 820543 WA 75s1136 (F) WA 75c1171 (M) E. macrophylla x E. berteroana 42x53-1 2 HO 82.280 PT 750420002 (F) PT 820253 PT 700044001 (M) E. macrophylla x E. berteroana 42x53-2 2 HO 82.281 PT 700044001 (M) E. macrophylla x E. folkersü 42x31-1 1 PT 820337 PT 750420002 (F) PT 700010001 (M) E. macrophylla x E. guatemalensis 42x 43-1 l PT 820281 PT 750420002 (F) PT 700018001 (M) E. macrophylla x E. guatemalensis 42 x 43-2 3 HO 82.763 WA 75s1136 (F) PT 820544 WA 74c1453 (M) Volume 75, Number 3 Neill 913 1988 Erythrina TABLE ll. Continued. ybrid Hybrid Live Accession Parental Accession Hybrid Number Plants Numbers* Numbers* E. tajumulcensis x E. guatemalensis 40 x 43-1 4 HO 82.761 WA 74c1448 (F) PT 820599 WA 74c1453 (M) Sect. Chirocalyx E. abyssinica X E. latissima 95 x 94-1 1 HO 82.867 PT 770034001 (F) PT 750281004 (M) Sect. Erythraster E. perrieri x E. variegata 106 x 96-1 5 HO 82.768 WA 755857 (F) PT 820550 WA 745892 (M) E. perrieri x E. variegata 106 x 96-2 3 HO 82.769 WA 745892 (M) PT 820551 WA 745892 (M) E. perrieri X E. variegata 106 x 96-3 1 HO 82.770 WA 745892 (M) * HO = Ho‘omaluhia Botanic Garden; PT = Pacific Tropical Botanical Garden; WA = Waimea Arboretum; F = female parent; M = male in growth rates, but they also produced flowers at individuals had pollen fertilities higher than either an exceptionally early age. Many of the narrow of their parents. For this trait at least, the narrow hybrids between species in sect. Erythrina grew hybrids in sect. Erythrina clearly exhibited inter- to be 4-m trees within two years after the seeds specific heterosis. were sown, and most flowered within that time. Meiosis in pollen mother cells was examined in Such sexual precocity is unknown in the parental several of the F, hybrids in sect. Erythrina. An species. Seeds from intraspecific matings were sown example is Erythrina guatemalensis X E. macro- concurrently and in the same nursery with the — pAylla, HO 82.288-A (Figs. 9, 10). Meiotic be- hybrids; none grew as rapidly or flowered as early havior in this hybrid can be compared with meiosis as most of the hybrids. None of the parental species, in its male parent E. macrophylla, PT 750420002 in cultivation, have been known to produce flowers (Figs. 5, 6). in less than three years from seed. As in the parental species, meiosis in the hybrids An even more phenomenal case of precocious was characterized by clumping of bivalents at late flowering was the intersectional (medium) hybrid diakinesis and metaphase I and by “sticky” chro- Erythrina crista-galli x E. fusca. Two sibling matin bridges and late disjunction of some bivalents individuals of this combination (PT 840231001 at anaphase I. The normal meiotic process was, and -002) produced flowers when they were still however, not disrupted. Nondisjunction or unequal small plants in nursery pots less than five months — assortment of chromosomes during meiosis I was after the seeds were sown. The parentals are both not observed, and all cells examined at telophase large- to medium-sized trees and are not known to I or subsequent stages had the expected number flower in the wild or in cultivation before at least of 21 chromosomes. Meiotic behavior in the F, several years of growth. hybrids within sect. Erythrina, in short, was iden- Twenty-five of the two-year-old F, hybrid plants tical to the behavior described above for the pa- flowered during February-March 1984. All were rental species. narrow hybrids within sect. Erythrina, and rep- The only intersectional F, hybrid to flower by resented nine hybrid combinations. These are listed | November 1984 was the five-month-old Erythrina in Table 14 with the pollen fertility of each indi- crista-galli x E. fusca. Pollen fertility in this vidual. Also included in Table 14 is the pollen hybrid (PT 840231001) was 81%. This was lower fertility of each of the parental individuals from than the pollen Wr i of either parent (E. crista- which these hybrids were derived. galli, WA 74p840, 96.1%; E. fusca, WA 74599, With two exceptions, the pollen fertility of the 96. 3%) but t not low enough to affect sub- F, hybrids was above 95%. The pollen fertility of stantially fertility and mating success of the hybrid. the hybrids (X = 97.6%) was slightly but signifi- Only limited material was available for analysis of cantly higher (P < 0.05) than the fertility of the meiosis in pollen mother cells of E. crista-galli x parentals (X = 95.0%). Eighteen of the 25 hybrid E. fusca. In some cells, several quadrivalents ap- 914 Annals of the Missouri Botanical Garden TABLE 12. Artificial Erythrina hybrids: medium (between sections, within subgenera) . ybrid Parental Hybrid Live Accession Accession Hybrid’ Number Plants Numbers umbers Subg. Duchassaingia E. crista-galli (2) x E. fusca (1) 2x]-l 2 HO 84.234 PT 740283001 (F) PT 840232 PT 740230005 (M) E. crista-galli (2) x E. fusca (1) 2x1-2 5 HO 84.235 WA 74p840 (F) PT 840231 WA 74599 (M) Subg. Erythrina E. herbacea (12) x E. humeana (18) 22x73.1 2 HO 82.863 WA 765187 (F) PT 820697 WA 74p1382 (M) E. lysistemon (17) x E. speciosa (9) 72x16-1 1 HO 84.238 PT 750280003 (F) PT 730708001 (M) E. lysistemon (17) x E. speciosa (9) 72x16-2 2 HO 84.243 PT 750280002 (F) PT 730708001 (M) E. speciosa (9) x E. lysistemon (17) 16x 72-1 l HO 84.236 PT 730708003 (F) PT 750280003 (M) E. speciosa (9) x E. lysistemon (17) 16x 72-2 3 HO 84.237 PT 730742002 (F) PT 750280003 (M) peared to be formed at metaphase I (Fig. 11). In other cells, meiosis was normal with 21 bivalents at metaphase I. Without more thorough cytological analyses, it is not possible to state whether or not meiosis is significantly disrupted in this hybrid. Nondisjunction and unequal segregation of some chromosomes may contribute to the partial reduc- ' Number in parentheses after each species denotes section (see Table 1). tion in fertility of the pollen. Fecundity of F, Hybrids. the controlled self-pollinations of the two-year-old F, hybrids in sect. Erythrina (Table 15) was very Fruit maturation from low, less than 3%, and nine of the 12 F,s produced no fruits from controlled selfing. In common with the usual pattern of results in experimental polli- nations of Erythrina parentals, much of the failure in fruit maturation was due to postzygotic abortion of young fruits, within one or two weeks after fertilization. Most of the F,s did produce a few fruits spontaneously, on open-pollinated inflores- cences. Animal pollen vectors were not present in the garden plots, and it is most likely that these open-pollinated fruits were produced by autogamy. TABLE 13. Artificial Erythrina hybrids: wide (intersubgeneric) . Hybrid Hybrid Live Accession Parental Accession Hybrid' Number Plants Numbers Numbers E. caffra (17) x E. fusca (1) 71x1-1 2 PT 820422 WA 74c1456 (F) WA 74s99 (M) E. crista-galli (2) x E. guatemalensis (12) 2x43-3 8 HO 82.758 WA 74p840 (F) PT 820598 WA 74c1453 (M) E. crista-galli (2) x E. speciosa (9) 2x16-1 l HO 82.860 WA 740283001 (F) PT 730708001 (M) E. crista-galli (2) x E. variegata (26) 2x 96-2 1 HO 82.495 WA 74p840 (F) WA 768996 (M) E. guatemalensis (12) x E. abyssinica (25) 43x95-1 1 HO 84.287 PT 700018001 (F) PT 731006002 (M) E. guatemalensis (12) x E. senegalensis (22) 43x79-2 1 HO 82.766 WA 74c1453 (F) WA 745100 (M) E. herbacea (12) x E. fusca (1) 22x1-1 4 HO 82.634 WA 75c1103 (F) PT 820542 WA 74599 (M) ' Number in parentheses after each species denotes section (see Table 1). Volume 75, Number 3 Neill 915 1988 Erythrina TABLE 14. Pollen fertility of artificial F, hybrids TABLE 14. Continued. within sect. Erythrina and of their parents. At least 500 grains counted for all samples. II. Parentals ; Percent L. Hybrids Accession Normal Per- Species Number Grains : de E. americana WA 75c1171 97.3 Accession Normal Hybrid Combination Number Grains E. berteroana WA 748864 96.6 E. berteroana WA 78s564 97.2 E. americana x E. berteroana PT700044001 96.4 E. berteroana PT 820420001 994* E. berteroana PT 700044002 98.1 E. berteroana x E. berteroana PT 730711001 97.4 E. guatemalensis PT 820549001 96.7 E. chiapasana PT 730710001 96.3 E. berteroana x E. chiapasana PT 721005001 94.4 E. guatemalensis HO 82.647-A 86.3 E. folkersii PT 700010001 85.3 E. chiapas E. guatemalensis PT 70018001 97.7 E. berteroana HO 82.278-A 98.1 E. guatemalensis WA 74c1453 97.7 E. chiapasana x E. guatemalensis PT 750419001 91.4 E. berteroana HO 82.278-B 99.6% E macrophylla PT 750420002 94.4 E. chiapasana E. macrophylla WA 7551136 96.9 E. berteroana HO 82.278-C 99.6* E. standleyana WA 7551056 84.9 E. chiapasana E. tajumulcensis WA 74c1448 98.0 E. berteroana HO 82.278-D 99.5* . = Hybrids: mean pollen fertility = 97.6% + E ply Ú: PT820493001 97.3* Parentals: mean pollen fertility = 95.0% + p 1%. Serter : Differences in pollen fertility, hybrids vs. parentals: t — E guatemalensis x 2.28; DF = < 0.05. E. bertero PT 820493002 96.3 * [ndicates beh with higher pollen fertility than either E. guatemalensis x parent E. chiapasana HO 82.283-A 99.8* E. guatemalensis x E. chiapasana PT820254002 96.6 Also in Table 15 are comparisons of fruit mat- E. guatemalensis x uration in the selfed F, hybrids vs. their parents. E. chiapasana HO 82.284-A 98.6% Several combinations of parental matings are in- E. guatemalensis T , cluded in the analyses. Fruit maturation was much - up 1 HO 82.284.B — 98.8" lower in the selfed F,s (3%) than in the original guatemalensis ju : | ni cmm HO 82.284-F 99.1* hybridizations which produced these F,s (22%) (P. < E nal L 0.01) guatemalensis . h , E. folkersii HO 82.282.A 99.4* The second pairwise comparison of fruit matu- E. guatemalensis x ration in Table 15, selfed F, hybrids vs. all of their E. folkersii HO 82.282.B 98.4* parental hybrid combinations (including recipro- E. guatemalensis x cals), may be more biologically meaningful than E. macrophylla HO 82.285-B 99.6* the first comparison for the following reason: the E. guatemalensis x female parents of the F, hybrids were very fecund, E. macrophylla HO 82.288-A 97.8% with higher than average fruit maturation. The E. guatemalensis x male parents (pollen donors) of the F,s generally E. macrophylla PT 820276001 98.2* h : š . ad lower fruit maturation when employed as fe- E. guatemalensis x les in the hybriia f th E. standleyana HO 82.765-A 89.8 ma es Ip ine A us ira dad > P p E. guatemalensis x ciprocal crosses produce | no hyl rid fruit at all. ' tajumulcensis PT 820547001 99.* it is assumed that fecundity (fruit maturation) is a E. macrophylla x quantitatively heritable trait, then an F, hybrid E. berteroana HO 82.281-A 98.8* might be expected to be intermediate in fecundity E. macrophylla x between its two parents, providing there is no re- berteroana HO 82.281-B 98.2* duction in fruit maturation in the hybrid caused E. macrophylla by incompatibilities between its constituent ge- 3 E. g ee Aarón I nomes. The proportion of fruit maturation expected macrophy | geboten HO 82.763.B 9gggs in the Fs, then, should approximate the proportion in all the parental hybrid combinations, including Annals of the Missouri Botanical Garden 916 s.s... ` ° ^e e i p L| r a * P A 10um "s ° 9 $ ° "aa .. ` m T a! do» : l - «1 ^e, ^ o * 10um 10 wah v, DE ) ° 5 ¿z ` J" * 4 4 10um 11 FIGURES 9-11 wis in Erythrina hybrids (pol- len mother cells) .— Late anaphase, E. guate- malensis x E. m E A "HO 82.288 ( chromatin bridges and late disjunction of some biva- lents (compare with meiosis in male parent i macro- phylla, Figs. 5, 6).— 11. Metaphase, E. crista-galli x E. fusca, PT 840231001 (n — 21). At least Du d nues rivalents are visible. the failed reciprocal hybridizations. By this mea- sure, the second pairwise comparison in Table 15, fruit maturation in the F, hybrids, is still much lower than the 15% fruit maturation in the parental generation; the difference is highly significant (P < (n = 21). Sticky There are several possible reasons for the re duced fruit maturation in the ybrids in sect. Erythrina. The first is that the low fecundity is in fact a consequence of hybridity caused by genic incompatibility between the parental genomes. It is evidently not, however, a matter of “hybrid sterility” in the usual sense of the term, in which the microgametophytes (pollen) and/or megaga- metophytes (embryo sacs) borne on the F, sporo- phyte are abortive and nonfunctional (Grant, 1953; Stebbins, 1958). The pollen fertility of the F, hy- brids, as discussed above, was exceptionally high; the pistil and ovules also appeared to develop nor- mally in the hybrids. Much of the failure of fruit set in the selfed F,s was at the postzygotic stage (abortion of young fruits). If the reduced fecundity was truly a consequence of hybridity and inter- genomic aer it is probably best consid- ered as a case of “hybrid breakdown" (Grant, 1953; Stebbins, 1958) expressed as low viability of the F, embryos. There are other possible explanations for the low fecundity of the selfed F,s that do not invoke hybrid breakdown or other effects of hybridity. The first is that it may be a consequence of self-mating, the opposite effect from the apparent heterosis evidenced by the exceptional vigor of the F, plants. Fruit set in the selfed F,s was significantly lower than the one in the selfed parentals, which in turn were significantly lower in fruit set than the hy- bridizations. For both parental and F, selfings, the high incidence of fruit abortion may be an expres- sion of inbreeding depression, a result of the homo- zygous pairing of deleterious recessive alleles in the genomes of the embryos. This possibility could be tested by controlled cross-pollinations between Fs, a step that was not taken initially because the goal of the F, selfings was to produce F, plants with no more than two constituent genomes. Another possible reason for the low fecundity in the F,s may simply be the juvenility of the F, plants themselves. Although the F,s were very vigorous and flowered precociously at two years of age, they were not yet full-sized trees. At their size, they might not be able to draw on sufficient resources for the full fruit crop of a larger adult. In short, the variables accounting for reduced fecundity in the narrow F, hybrids still need to be sorted out. This should be possible once the F, trees attain. their. full adult size and several categories of matings within and between individuals are car- ried out. Viability of F, Hybrids. The viability of the F, hybrid seed, obtained from selfed and open-polli- nated F, plants in sect. Erythrina (Table 16), was Volume 75, Number 3 Neill 917 1988 Erythrina TaBLE 15. Fruit and seed maturation from controlled self-pollinations of narrow F, hybrids in sect. Erythrina. Total Flowers umber Hybrid Combination Accession Number Pollinated Mature Fruits of Seeds E. berteroana x E. guatemalensis HO 82.674-A 18 0 0 E. guatemalensis x E. berteroana PT 820493002 13 0 0 E. guatemalensis X E. chiapasana HO 82.284-A 32 3 T7 E. guatemalensis x E. chiapasana HO 82.284-B 6 0 0 E. guatemalensis x E. chiapasana HO 82.283-A 38 0 0 E. guatemalensis x E. chiapasana PT 820254002 7 2 3 E. guatemalensis X E. chiapasana PT 820278002 10 0 0 E. guatemalensis X E. macrophylla HO 82.288-A 21 0 0 E. guatemalensis x E. macrophylla HO 82.285-B 16 0 0 E. guatemalensis X E. tajumulcensis HO 820547001 20 1 1 E. macrophylla x E. berteroana HO 82.281.B 23 0 0 E. macrophylla x E. guatemalensis HO 82.763-A 15 0 0 Total selfed F, hybrids 225 6 (3%) 11 Parental hybridizations 51 21 (22%) All parental hybrid combinations (including reciprocals) 171 25 (15%) Selfed parentals (sect. Erythrina) 144 12 (8%) Multiple comparison test for differences in fruit maturation Selfed F,s vs. parental hybridizations Selfed F,s vs. all parental hybrid combinations Selfed F,s vs. selfed parentals Parental hybridizations vs. selfed parentals All parental hybrid combinations vs. selfed parentals P « 0.01 P « 0.01 P « 0.05 P « 0.01 N.S. significantly lower than viability of the F, hybrids. This was shown in Table 8, where the F,s were compared with all the narrow F, hybrids; the dif- ference was highly significant (P « 0.01). In Table 16 the viability of the F, seed is compared specif- ically with that of their own parents, i.e., with the F, seed lots producing the parents of the F,s. The viability of the F,s (13%) was significantly lower « 0.01) than that of their F, parent generation (61% viability). Table 16 the viability of the F, seed and of their F, parents is also compared with seed from intraspecific matings in Erythrina guatemalensis (including seed from selfings and intraspecific out- crosses, the only intraspecific viability data avail- able for sect. Erythrina). The F, hybrid seed was significantly higher in viability than the intraspe- cific seed (P < 0.05). The viability of the intra- specific seed (19%) was somewhat higher than that of the F, seed, but the difference was nonsignificant. In summary, the viability of F, hybrid seed was significantly higher than F, seed derived from selfed , matings and higher than seed derived from in- traspecific matings. If the very high F, viability is truly an expression of interspecific heterosis, this hybrid advantage is not retained in the F, gener- ation, when the F,s are derived from selfed F, ybrids. It is possible to interpret the reduction in F, viability with respect to F, viability as "hybrid breakdown." However, with the data presently available, the reduced viability of the F,s derived from selfed F,s could also be interpreted as an expression of inbreeding depression. It could also be interpreted simply as an absence of the heterotic advantage possessed by the Fs, since the viability of the F,s was not significantly lower than that of the intraspecific progeny. These three alternatives cannot be differentiated with the presently available information. Additional progeny trials of F,, F,, and intraspecific seed lots are needed to test the em that hybrid breakdown may be expressed n the F, generation of Erythrina hybrids. In any case, the lowered average viability of the F.s was a function only of poor germination of the seed. The seeds that did germinate produced healthy plants with normal growth and vigor at six months of age. There were no indications of chlorosis or other debilities in the F, plants. Studies of Previously Synthesized Hybrids. Nine artificially produced Erythrina hybrids, all between 918 Annals of the Missouri Botanical Garden TABLE 16. Viability of F, hybrids within sect. Erythrina. Accession Female Parent Pater- Seeds Live F, F, Hybrid Combination Number (F, Hybrid) nity Sown Plants E. guatemalensis X E. berteroana HO 84.288 HO 82.641-A pen 9 0 E. guatemalensis X E. chiapasana PT 840234 PT 820254002 Self 3 0 PT 840235 E. guatemalensis X E. chiapasana HO 84.289 HO 82.284-A Self 7 1 E. guatemalensis x E. macrophylla HO 84.290 HO 82.288-A Open 5 l E. guatemalensis x E. tajumulcensis PT 840243 PT 820547001 Self l 0 E. herbacea x E. guatemalensis PT 840241 PT 820421001 Open 5 0 E. macrophylla x E. berteroana HO 84.245 HO 82.281-A Open 4 1 E. macrophylla x E. berteroana HO 84.291 HO 82.281-B Open 4 4 E. macrophylla x E. guatemalensis HO 84.244 HO 82.763.B Open 18 1 HO 84.292 PT 840242 Total F, hybrids in sect. Erythrina 56 7 (1390) F, hybrid parents o 28 17 (61%) Intraspecific progeny, Eas guatemalensis 26 5 (19%) Multiple comparison test for differences in viability of seed F, hybrids vs. their F, hybrid parents P « 0.01 F, hybrids vs. intraspecific E. guatemalensis N.S. F, hybrids vs. intraspecific E. guatemalensis species of different sections, have been reported were obtained but they invariably aborted before prior to this study (Table 17). Krukoff & Barneby (1974) described most of these; in the same paper they described some putative natural hybrids be- tween sympatric Mesoamerican and African species. The parentage of only two of the artificial hybrids is known for certain; both of these F,s are "wide" intersubgeneric hybrids and are reported to be fer- tile. The oldest and best-known Erythrina hybrid is E. x bidwillii Lindley, synthesized from E. her- bacea (sect. Erythrina) 9 and E. crista-galli (sect. Cristae-galli) à in Australia in the 1840s and since spread around the world as a cultivar by propa- gation of cuttings. Krukoff & Barneby (1974) re- ported E. xbidwillii to produce viable seed and also that “no Mendelian segregation of phenetic characters is observed in the F, or subsequent generations." They further claimed that this hybrid had naturalized in Fiji and was therefore a stabilized “neospecies.” I examined E. x bidwillii in cultivation at Foster Garden, Honolulu (FG 64.2035). Meiosis in pollen mother cells was normal with 21 bivalents at meta- phase I. Pollen fertility was 63%, comparable to Graham & Tomb's (1974) report of 76% normal pollen for this hybrid. I attempted to produce an F, generation by controlled self-pollination of 60 flowers over a period of several weeks. Young fruits two weeks of development. I have not seen mature spontaneously produced fruits on any cultivated plants or herbarium specimens of E. x bidwillii, so the reports of its viable seed production are questionable made limited attempts (12 trial pollinations) to backcross E. x bidwillii to one of its parents, E. crista-galli. The pollinations all failed, but given the reasonably high pollen fertility of E. x bidwillii, it is likely that with perseverance some backcross progeny could be obtained. e other previously reported hybrid of known parentage is Erythrina X resuparcellii Srivastava (a nomen nudum, not validly published), a hybrid between the perennial herb E. resupinata (sect. Suberosae) 9 and E. variegata (sect. Erythraster) ô (Jalil et al., 1982). The F, is a branched shrub, and in other morphological traits is also interme- diate between the two parents. The flowers, how- ever, resemble those of the female parent much more closely than those of the male. This hybrid was not available to me, but Jalil et al. (1982) reported that it had normal meiosis in pollen mother cells with 21, at metaphase I, pollen fertility of 62%, and viable seed. Erythrina Xsykesii Barneby & Krukoff was the only other hybrid among those listed in Table 17 available to me for experimental studies. This Volume 75, Number 3 Neill 919 1988 Erythrina TABLE 17. Previous reports of artificial Erythrina hybrids.' l. Erythrina x bidwillii Lindley, Bot. Reg. 33: pl. 9. 1849. E. herbacea 9 (12) x E. crista-galli 6 (2) 2. Erythrina x bellangeri Focke, Die Pflanzen-mischlinge. 110. 1881. ? E. crista-galli 2 (2) x E. herbacea à (12 3. Fares X crassifolia Koorders ex Backer, Schoolflora voor Java 1: 360. 1911. ? E. su ans (6) x E. variegata ) ? E. fusca (1) x E. variegata (26 4. Erythrina x fluminensis Barneby & Krukoff, Lloydia 37: 446. 1974. ? E. speciosa (9) x E. sp. (subg. Micropteryx 5. Enea X hennesyae Barneby & Krukoff, Clovis 37: 448. 1974. ? E. humeana (18) x E. lysistemon (17 6. Erythrina X orba Barneby & Krukoff, Lloydia 37: 449. 1974. E. lysistemon (17) x E. speciosa (9) T. Eryttrina X sykesii Barneby & Krukoff, Lloydia 37: 447. 1974. americana (12) x E. lysistemon (17) ? E. speciosa (9) x E. lysistemon (17) 8. Erythrina x vlissingensis Waby ja Barneby & Krukoff, Lloydia 37: 446. 1974. ? E. fusca (1) x E. variegat ? E. fusca (1) x E. suberos 9. E. resupinata 9 (4) x E. variegata à To a Erythrina X resuparcellii Srivastava, Allertonia 3: 19. 1982. nomen nudum. ) ' Known or presumed parental species combinations are listed below each hybrid binomial; question mark preceding hybrid combination indicates uncertain parentage. Numbers in parentheses following species refer to sections to which species belong (Table 1 hybrid was reputedly produced under cultivation in Australia in the 19th century, but its parentage is unknown. Krukoff & Barneby (1974) believed the parents to be E. lysistemon (sect. Caffrae) and E. americana (syn. E. coralloides) (sect. Ery- thrina). Based on study of floral and leaf mor- phology, I believe instead that the parents are E. lysistemon and E. speciosa (sect. Stenotropis). Since I have obtained both reciprocal hybrids of E. speciosa X E. lysistemon (Table 12), these F,s can be compared with E. x sykesii when they come into flower. I examined cytologically several individual ra- mets of E. X sykesii (WA 76p864, WA 75s1706, Foster Garden FL.669) and attempted to produce F, plants by controlled self-pollination. Meiosis in MCs was apparently normal, with 21 bivalents at metaphase I. Pollen fertility was 81-84%, whic agreed closely with results reported earlier for the same taxon by Graham & Tomb (1974). However, no mature fruits were obtained from 65 attempts at selfing. Young fruits with partially developed seeds wer dance as with E. x bid- willii, but these always aborted within two to three weeks following pollination. Erythrina X bidwillii, E. X resuparcellii, and E. Xsykesii are the only intersectional or inter- subgeneric hybrids that have been tested for fe- cundity at this time. An F, generation was report- edly obtained from E. x resuparcellii (Jalil et al., 982), but the other two may be incapable of producing F, progeny, at least from selfing of the F,. Zygotes, embryos, and young fruits are formed, but the fruits abort before maturity. These “wide” hybrids, then, may be subject to hybrid breakdown expressed as inviability of the F, hybrid embryos borne on the F, hybrid sporophyte. As discussed in the previous section, the cause of mating failure in the F, hybrids is subject to different interpre- tations. Whether or not F, hybrid breakdown is a general phenomenon in Erythrina remains to be investigated. CONCLUSIONS: EXPERIMENTAL HYBRIDIZATIONS AND SELF-COMPATIBILITY TRIALS From the information presented in this section, a series of generalizations regarding breeding sys- tems and species relationships in Erythrina can be outlined: . Even under the most carefully controlled conditions, mating success (proportion of pollinated flowers producing mature fruits) is low in all Er- ythrina species. This is true even when the effects of “resource competition" are eliminated by re- moving most flowers from an inflorescence as well as all of the spontaneously produced fruits on the tree, and pollinating only a few selected flowers per inflorescence. Mating failure results partly from prefertilization abortion of pollinated flowers, but also to a large extent from postfertilization abortion of young fruits. Annals of the Missouri Botanical Garden 5um 100um Aa e S RET sa cs. 100um SIIR AAE Volume 75, Number 3 1988 Neill Erythrina 2. Gametophytic or sporophytic self-incompat- ibility systems of the “classical”” model, mediated by inhibition of pollen tubes in the style or stigma and governed by a single-locus, multiallelic S-gene, are evidently not present in any of the Erythrina species examined. Self-incompatibility in this strict sense is probably absent from the entire genus. There is considerable individual variation in fecun- dity, and some individuals may be cryptically fe- male-sterile, but if an individual produces seed from outcrossing it will also produce some seed from self-mating. For some individuals and some pop- ulations, mating success is lower in selfing than in outcrosses, but much of the mating failure is ex- pressed postzygotically by abortion of young fruits. This is probably an expression of inbreeding depres- sion, the consequence of increased homozygosity for any number of deleterious recessive alleles, rather than the action of a specific S-allele. In- breeding depression may also be expressed in the progamic stage as inhibition of pollen tubes. 3. Spatial separation of anthers and stigma in some Erythrina species, and protandry in other species, limits autogamous pollinations. For the pro- tandrous species at least, this mechanism is not absolutely effective; autogamous fruits are occa- sionally produced. Autogamy occurs only with the ultimate flowers on an inflorescence and may be an adaptive feature of the breeding meten to pro- duce some seed as a a dest resort if no “high- quality” (i.e., outl on earlier flowers of the inflotescinds . The hybridization trials indicate that matings between closely related species (within sections) are just as likely to produce viable progeny as are matings within species. Mating success is usually higher, in fact, in interspecific hybridizations within sections than in self-mating. At increasing taxo- nomic distance between parental species, there is a general trend to lower mating success in the hybridization trials. This trend is not universally applicable, however. Viable F, hybrids have been produced between the most distantly related groups of species in the genus—between species of dif- ferent subgenera indigenous to different continents. It is probable that viable F, hybrids can be obtained between any two diploid species in the genus Er- ythrina. 5. F, hybrids between the closely related species in sect. Erythrina exhibit interspecific heterosis by several measures: viability of the F, seed is higher, and the F, plants are more vigorous, sexually pre- cocious, and have higher pollen fertility than the parental species. Pollen fertility is somewhat lower in hybrids between more distantly related species, but these hybrids are generally comparable in vi- ability and vigor with the parental species. reduction in fecundity is exhibited by the F, hybrids, at least when the F,s are selfed. This lowered mating success is not due to “hybrid ste- rility" per se, since the gametes produced by the F, hybrid function normally. Mating failure is ex- pressed postzygotically by abortion of young fruits and evidently is a consequence of inviability of the F, hybrid embryo. An alternative explanation may be that mating failure in the selfed F,s is a con- sequence of inbreeding depression. These experiments support the first two hy- potheses presented in the introduction: 1) the species in sect. Erythrina can hybridize freely with each other, and there are no internal qualitative or quan- itative postmating isolating barriers between the species; and 2) hybridization between more widely divergent species is also possible; there is generally a quantitative reduction in mating success in the wider hybridizations, but this probably does not constitute an absolute barrier to the formation of F, hybrids. The only major unanswered question regarding interspecific compatibility among diploid Erythrina is the possibility of hybrid breakdown in the F, and subsequent generations. F, break- down, if it exists, does not form an absolute isolating barrier within sect. Erythrina, but it may form an absolute barrier in hybridizations between more widely divergent species. SECTION 5. INHERITANCE OF PHENETIC TRAITS IN INTERSPECIFIC HYBRIDS The fact that plant species with large morpho- logical discontinuities can be hybridized, and that large hybrid progenies can be grown together in a common garden, has allowed for analyses of the genetic basis of these phenetic differences (review in Gottlieb, 1984). A thorough genetic analysis, of course, requires at least the study of segregating — FIGURES 12-17. WA 45s960.— 13. Epicuticular wax platelets, E. O a 7000 pasana, WA 74s876.— 15. Dendritic hairs, E pe rrieri, WA 7 78s225.— 17. Ribbonlike hairs, E. leptorhiza, Neill 5646. SEM images ofabaxial - aes of Erythrina.— 12. ag iyu wax De us E. suberosa, Two 044001.— med hairs, E. chia- 876.— 16. BUD Ru. La. E. arborescens, WA Missouri Botanical Garden Annals of the 922 Volume 75, Number 3 Neill 923 1988 Erythrina "100um Ro d REN be FIGURES 24-27. SEM Tues abaxial leaf surfaces of Erythrina. All images at 60° tilt. —24, 25. Lamellae, E. stricta, WA 74s897.— 26, 27. Lamellae and two-armed hairs, E. suberosa, WA 755960. — FIGURES 18-23. p rat of zq ga leaf surfaces of Erythrina. 18, 20, 22.—SEM images. 19, 21, 23.—Anatomical section , E. guatemalensis, PT 750419001.— 19. DA E. folkersii, PT 700010001.—20. Lamellae E E beg PT 721346001.—21. a llae, E. suberosa, WA 755960.—-22. Glandular hair, E. salvia, PT 721346001.—23. “Glandular” hair, papillae, and lamellae, E. berteroana, WA 745864. 924 Annals of the Missouri Botanical Garden TABLE 18. Comparison of leaf epidermal characters in Erythrina hybrids and parents. Female Parent F, Hybrid Male Parent Figures Hairs Epidermal sculpturing Epicuticular Figures Hairs Epidermal sculpturing Epicuticular wax Figures Hairs Epidermal sculpturing Epicuticular wax Figures Hairs Epidermal sculpturing Epicuticular wax Figures Hairs Epidermal sculpturing E. chiapasana PT 721005001 HO 82.278 E. berteroana PT 700044002 28-30 dense covering of two-armed hairs low ridges around stomata; no lamellae wax absent E. guatemalensis PT 700018001 31-33 sparse two-armed hairs open, irregular network of discontinuous lamellae; un- even in height, less than 15 um tall wax present HO 82.289 34-36 hairs absent dense network of discontin- uous lamellae, up to 40 um wax present E. berteroana PT 700044001 37-39 hairs absent dense papillae, to 40 um tall wax present E. guatemalensis PT 700018001 40-42 hairs absent sparse covering of papillae nd 2-6-celled lamellae; short, less than 25 um tall wax present HO 82.283 43-45 hairs absent dense network of discontin- uous lamellae, to 40 um all wax present oo E. chiapasana PT 721005001 46, 47 hairs absent dense papillae to 40 um tall wax present 48, 49 sparse scattering of two- armed hairs incipient papillae: low cres- cent-shaped ridges outlin- ing anticlinal walls of epi- dermal cells wax present 50, 51 dense covering of two-armed airs low ridges around stomata: no papillae wax absent E. guatemalensis E. abyssinica PT 700018001 HO 82.647 PT 731006002 52-54 95-57 58-60 hairs absent sparse covering of two-armed dense covering of two-armed airs hairs dense papillae to 40 um tall wax present E. guatemalensis WA 74c1453 low epidermal ridges, less than 10 um tall wax present HO 82.766 low epidermal ridges, less than wax present E. senegalensis WA 745100 61-63 hairs absent dense papillae, to 40 um tall 64-66 sparse scattering of balloon- like hairs, up to 50 um x 100 in size low, crescent-shaped epider- mal ridges, less than 5 um tall 67-69 sparse scattering of balloonlike hairs, up to 50 um x 100 in size low stellate papillae, less than 10 um ta Volume 75, Number 3 Neill 925 1988 Erythrina TABLE 18. Continued. Female Parent F, Hybrid Male Parent Epicuticular wax present wax present wax absent wax E. lysistemon . speciosa PT 75028003 HO 84.238 PT 730708001 Figures 70, 71 72. 73 14, 75 Hairs hairs absent sparse scattering of two- sparse scattering of two-armed armed hairs hairs Epidermal low papillae, less than 10 um low epidermal ridges, less low epidermal ridges, less than sculpturing tall a tall Epicuticular wax present wax present wax present wax E. crista-galli E. OPEP WA 74p840 (4 individuals) WA 74c1453 Figures 76, 82 78-81, 84-87 77,83 Hairs hairs absent hairs absent hairs absent Epidermal low, discontinuous lamellae, variable: papillae or discon- dense papillae, to 40 um tall sculpturing less than 10 um tall, tinuous lamellae, to 15 forming reticulate pattern wax absent um ta wax present in all wax present Epicuticular PT 840231 E. crista-galli A E. fusca WA 74p840 (3 individuals) PT 74599 Figures 88, 94 90-92, 96-98 89, 95 Hairs hairs absent hairs absent hairs absent Epidermal low discontinuous lamellae, variable: scattered low pa- irregular convoluted surface sculpturing less than 10 um tall, , 3-4-celled lamellae with deep cavities, knobs, forming reticulate pattern or dus flat and protrusi Epicuticular wax absent wax present in all wax absent wax progeny in the F, generation. In the absence of large F, families, however, preliminary character- ization of the inheritance of morphological char- acters can be obtained from F, hybrids. A study of the inheritance of phenetic traits in artificially produced hybrids serves several pur- poses beyond that of genetic analysis. Firstly, it allows for confirmation of hybridity in the hybrid progeny. In any experimental hybridization, there exists the possibility that the cross may be spurious; the progeny could result from contamination of self-pollen on the stigma, or from agamospermy. However, if the progeny possess a character pres- ent in the male parent but absent in the female, their hybrid nature is reasonably confirmed A second purpose for studying the inheritance of morphological traits in artificial hybrids is to generate information on the patterns of variation to be expected when hybridization occurs in nature. If, as Raven (1980) and Grant (1981) have sug- gested, there is a great deal of hybridization in flowering plants that passes undetected as such, then study of the products of artificial hybridization may help in the discovery and confirmation o hybrids in natural populations. This method was used effectively, for example, by Nobs (1963) in his biosystematic study of Ceanothus. Some of the artificial F, hybrid segregates of Ceanothus closely resembled stabilized populations with restricted ranges recognized as species. Nobs used this evi- dence to support his hypothesis that these species were of hybrid origin and were nite from pairs of extant, more wide-ranging speci A similar study, combining artifical hybridiza- tion and analysis of natural hybridization, con- ducted by Gillett & Lim (1970) on Bidens in Volume 75, Number 3 Neill 1988 Erythrina Ficures 37-45. SEM images, abaxial leaf surfaces of d orn) guatemalensis X E. berteroana and parents Each horizontal row x equal magnification. 38, 41, 45 at tilt. —37—39. E. guatemalensis, PT 700018001, female parent. —40—42. E. guatemalensis x FE, berteroana, HO 82.289.—43-45. E. berteroana, PT 700044001, male parent. 928 Annals of the Missouri Botanical Garden Hawaii, has been questioned by Ganders & Nagata (1984). They showed that some of the putative natural hybrids were merely intraspecific variants, and Ganders & Nagata concluded that adaptive divergence was more important than hybridization in the evolution of Bidens in the Hawaiian Islands. Although all of the 17 are in fact interfertile, natural hybridization is rare because they are mostly allopatric. Evidently Gillett & Lim used too few characters and ignored intra- Hawaiian Bidens species population variation. An important caveat is to avoid a too facile interpretation of hybridization results when applying them to the study of pro- cesses in natu In this ach. two sets of phenetic traits were examined in the Erythrina hybrids and their parent species: 1) features of the epidermis of abaxial leaf FIGURES 46- SEM images, abaxial leaf a of Erythrina guatemalensis x E. « Each horiz pes row at equal magnification : 19. E. guatemalensis X E. chiapasana, HO 82.2 283. 50, eu E. chiapasana, PT 721005001, male parent. iapasana and parents. guatemalensis, PT UL female parent. — 48. surfaces, and 2) morphology and color of the flow- ers. EPIDERMAL FEATURES OF ERYTHRINA LEAVES Characters of the leaf are second only to those of flowers in their use and value in taxonomic studies (Stace, 1984). Studies of the inheritance of leaf surface characters in interspecific hybrids have recently been carried out in Aloe and Gas- teria (Liliaceae) (Cutler, 1972), in /lex (Baas, 1978), and in Quercus (Cottam et al., 1982 Erythrina species possess a wide variety of leaf surface characters. These have figured promi- nently in the taxonomic delimitation of the species (Krukoff, 1939a, b: Krukoff & Barneby, 1974) In these previous works the surface characters 50um = 3 o hus Q 00600 Annals of the Missouri Botanical Garden 861-69. SEM images, abaxial leaf surfaces E Erythrina guatemalensis X E. senegalensis and parents. Eac n [seb row at equal magnification. —6 1-63. E. guatemalensis, WA 74c1453, female pre — 64-66. E. guatemalensis x E. senegalensis, HO 82.766. hid E. senegalensis, WA 74s100, male paren Volume 75, Number 3 1988 Neill 931 Erythrina 100um E pus 70-75. zontal row at equa magnification. — hor al m iyak Dens x E. speciosa, HO 84.283. — 74, p^ E. olini PT nee. male parent were not studied with high magnification or ana- tomical sectioning, however, and the structure of some of the surface characters was misinterpreted. This is discussed below in the description of epi- dermal characters. Leaf epidermal features of a few species of Erythrina have also been surveyed using scanning electron microscopy (Ayensu, 1977). Materials and Methods In this study, only the abaxial surfaces of leaves were examined. All samples were obtained from mature, fully expanded leaves, which were pressed and dried as in preparation of herbarium speci- mens. The specimens were gold-coated with a Po- laron E5000 sputter-coater and observed with an Hitachi 450-S scanning electron microscope. For a few selected species, anatomical sections of par- SEM i images, abaxial b usce Lieben" ar x E. speciosa in parents. ud 71. E. lys PIT I 0280003, odes parent.— 72, affin-embedded leaves were prepared by Dr. Hi- roshi Tobe of Chiba University, Japan. Results Survey of Leaf Epidermal Features in Erythrina Epicuticular Wax. Platelets of epicuticular wax cover the abaxial leaf surfaces of many Erythrina species. The wax gives a whitish, glaucous ap- pearance to the leaf observed without magnifica- tion. The platelets are 1—3 um in size, are oriented randomly on the leaf surface, and vary in density (Figs. 12, 13). Epicuticular wax is consistently present in some species, but in others its presence or absence is variable even among individuals of a single population. Multicellular Branched Hairs. Several types of Annals of the Missouri Botanical Garden 50 um ^ pde Volume 75, Number 3 1988 Neill 933 Erythrina hair occur on the abaxial leaf surfaces of Erythrina species. The most common, and the only type found in sect. Erythrina, is a multicellular, two- armed hair (Fig. 14). This consists of several short basal cells, one or two longer cells forming the stalk, and one cell forming each of the arms, which may be 1,000 um or more in length. Two-armed hairs are present on the young leaves of most species in sect. Erythrina and in species of other sections as well, but in many species they are deciduous and are absent from fully expanded leaves. In the other species, the hairs are retained on mature leaves and form a dense tomentum. Multiple-branched, dendritic hairs (Fig. 15) are restricted to sect. Erythraster. They occur in all species of that section and are found on calyces, inflorescence branches, and on leaves. Each branch of the dendritic hair is 50-100 um long and is formed by a single cell. The dendritic hair has about 8-12 branches and extends up to 300 um above the surface of the epidermis. “Glandular”? Hairs. Multicellular, uniseriate hairs occur sporadically on leaf surfaces of many Erythrina species (Figs. 22, 23). They appear to be glandular, but what substance these hairs se- crete, if any, is not known. They are squat, rounded hairs about 50 um long and comprised of five or six cells. Observed with a microscope, they glisten with a translucent amber color Unicellular Hairs. The most common type of multicellular hair in Erythrina is formed by a rounded or elliptic, thin-walled cell which loses its cytoplasm and collapses at leaf maturity or upon drying (Fig. 16). These I refer to as “balloonlike”’ hairs. They occur in several sections of Erythrina. Long, flat, ribbonlike hairs 500 um or more in length are found particularly along the principal veins of the leaf in some species (Fig. 17). This type of hair is predominant in the Mexican sects. Breviflorae and Leptorhizae. Epidermal Sculpturing: Papillae and Lamel- lae. The remaining types of Erythrina trichomes are considered separately from hairs because they are more integrally a part of the foliar epidermis. These are papillae and lamellae, which I refer to collectively as ““epidermal sculpturing.” Papillae are single-celled, fingerlike trichomes. Each papilla is formed by the protrusion of an epidermal cell above the leaf surface. Papillae are up to 40 um tall and 15 um in diameter (Figs. 18, 19). Papillae and the epidermal surface between them are usually covered with epicuticular wax platelets. Under low magnification, a leaf surface with papillae appears covered with whitish gran- ules. Krukoff (Krukoff & Barneby, 1974) termed such leaf surfaces or ““gran- ular-ceriferous," but the ‘“‘granules” he described are wax-covered papillate cells, not individual par- ticles of wax. Papillae similar to the ones found in Erythrina occur on leaf surfaces in many groups of plants, but the structures I have termed not to my knowledge been reported from leaf epi- dermis of any angiosperm besides Erythrina. La- mellae, like papillae, are formed by protrusions of epidermal cells, but in lamellae the cells are joined edge-to-edge to form continuous “walls” one cell thick that stand above the surface of the leaf (Figs. 20, 21). Leaf surfaces with lamellae are also usually covered with epicuticular wax. “farinose-ceriferous”” “lamellae” have Lamellae occur in several species of sect. Er- ythrina. In these species the lamellae are discon- tinuous; each lamella is composed of several to twenty cells standing edge-to-edge. The lamellae form a dense, discontinuous network with a char- acteristic pattern when observed at low magnifi- cation (Fig. 34). Krukoff (Krukoff & Barneby, 1974) referred to leaves with wax-covered lamellae as "reticulately ceriferous.” A unique pattern of lamellae occurs only in the Asian sect. Suberosae (Figs. 24-27). The lamellae are tall (50 um) and continuous. Parallel rows of several lamellae, each leaning at a different angle with respect to the leaf surface (Fig. 25), are joined to form an open network of interconnected poly- gons (Fig. 24). Shorter lamellae extend into the center of the polygons. The distribution of the polygons is associated with the vascular tissue of the leaf. Erythrina suberosa (Figs. 26, 27) has both polygon-forming lamellae and two-branched airs. Trichome characters are generally quite con- stant within a species and are useful taxonomic markers, often allowing species identification from — Ficures 76-81. photos at equal magnification. — 76. SEM images, abaxial leaf surfaces of E. crista-galli x = o and parents. All E. crista-galli, WA 7 uatemalensis, 74p840, female par . gu 74c1453, male parent.—78-81. E. crista-galli x E. guatemalensis, four F, siblings, PT 820548 and WA 82.278. 934 Annals of the Missouri Botanical Garden sterile material. [In contrast, presence or absence of epicuticular wax is variable within populations and is not a useful marker. As will be seen in the discussion of the hybrids below, epicuticular wax is evidently a simply inherited trait. Krukoff (Kru- koff & Barneby, 1973) separated the Mexican species Erythrina americana Miller and E. cor- alloides A. DC. (sect. Erythrina) solely on the basis of presence or absence of epicuticular wax. This trait is variable and is not well correlated with geographic distribution. For this and other reasons, Erythrina coralloides is here considered a syn- onym of E. americana. Inheritance of Leaf-Surface Characters in Interspecific Hybrids Each of the six plates comprising Figures 28- 75 illustrates the leaf surface features of a single F, hybrid and its two parents. On each plate, the female parent is illustrated on the left, the male parent on the right, and the hybrid in the center. Each horizontal row of photographs is a comparison of the three individuals at equal magnification (in- dicated by the bar in the left-hand photograph). Table 18 summarizes the features present in the parents and hybri our of the six hybrids illustrated were derived from the same genetic individual as female parent, Erythrina guatemalensis PT 700018001 and WA 74c1453. It is particularly instructive to note the pattern of inheritance in the hybrids produced from the combination of this genome with those of four different species: E. berteroana, E. chiapasana, E. abyssinica, and E. senegalensis. Erythrina guatemalensis has well-developed papillae on the abaxial leaf surface, each composed of a single epidermal cell; the male parent KÉ. ber- teroana has well-developed lamellae, each com- posed of about 4-5 cells forming a ‘“‘wall-like” structure. The F, hybrid has lamellae intermediate in length between the two parents, composed of 2-3 cells, but these are lower in stature and less developed than the epidermal sculpturing of either parent (Figs. 37-45). Erythrina guatemalensis lacks hairs on mature leaf surfaces. The male parents E. chiapasana and E. abyssinica have dense covering of two- armed hairs. The hybrids derived from these males with E. guatemalensis as female also possess two- Wm hairs, but at a much lower density than in the male parents (Figs. 46-51, 52-60). Erythrina senegalensis has scattered balloonlike hairs, and these are also inherited in the F, hybrid E. gua- temalensis X E. senegalensis (Figs. 61-69). The male parents E. chiapasana and E. senegalensis lack epicuticular wax; this trait is present in the female E. guatemalensis and in the hybrids. (Other individuals of E. chiapasana than the one used in this cross do have epicuticular wax.) Similar patterns of inheritance are exhibited by the other F, hybrids, for example, Erythrina chia- pasana X E. berteroana (Figs. 28-36). The fe- male parent E. chiapasana has a dense covering of hairs and lacks lamellae and epicuticular wax. The male parent E. berteroana lacks hair but possesses lamellae and wax. The F, hybrid has scattered hairs, lamellae reduced in stature, and epicuticular w In the du. hybridization that produced Erythrina lysistemon X E. speciosa, the male parent possesses two-armed hairs, which are lack- ing in the female parent; hairs are present in the F, hybrid, but again, at a rather lower density than in the male parent (Figs. 70-75). These results demonstrate that it is possible to confirm hybridity in the progeny by examination of leaf epidermal characters. Many of the hybrids possess characters present in the male parent but absent in the female parent. There is no evidence FicunEs 82-87. photos at equal magnification. — SEM images, abaxial leaf surfaces of E. crista-galli x E. o and parents. All 82. E. crista-galli, WA 74p840, E => emale paren uatemalensis, E 74c1453, male parent. —84—87. E. crista-galli x E. guatemalensis, four F, Prid PT Po and WA 82.2 Ficures 88-93. e. All photos at equal magnification. — 74599, male parent. — 93. E. dominguezii, PT 740234001. FIGURES 94-99. dominguezii. All photos at equal m 74599, male parent. — 99. E. dominguezii, PT 740234001. agnification.— 94. SEM images, abaxial leaf surfaces of Erythrina crista- yap X E. fusca and a and E. E. crista-galli, WA 7 90-92. E. crista-galli x E. fusca, three F, in PT 840231 a HO 84.235. p840, female parent. —89. E. fusca, SEM images, abaxial leaf surfaces of Erythrina crista- o X E. fusca and parents, and E. E. crista-galli, WA 74p 96—98. E. crista-galli x E. fusca, three F, ag PT 840231 br HO 84.235. 0, female parent. —95. E. fusca, Volume 75, Number 3 Neill 1988 Erythrina Annals of the Missouri Botanical Garden Volume 75, Number 3 Neill 1988 Erythrina Annals of the 938 Missouri Botanical Garden x sisuo[eurojeng ^ JOI — 1u24nd ajpuaf € 1u210d ?]»u *[00010002 Ld 51930) “A "ZOI — "28728 OH SIMOJ 73 100810002 Ld 'sssuep[euroyena ^q "99r —'srua4nd pun usiəylo] ^4 x stsuoep[eurojen2 euy, Ag fo s22u22s240)u[ “ZOL-001 SINNANI Volume 75, Number 3 Neill 939 1988 Erythrina BLE Comparison of flowers of F, hybrids and their parents in sect. Erythrina. Color code in "Standard Color" refers to Berlin & Kay color chart; see Croat (1983). Female Parent F, Hybrid Male Parent E. berteroana E. no Figure 103 WA 18s564 HO 82.647 WA 74c1453 CALYX Shape oblong oblong to elliptic elliptic Apex shape irregular; bilabiate or Length (cm) oblique; longer on cari- nal side oblique; longer on cari- nal side oblique Vexillar side 2.2-2.3 2.2-2.4 2.3-2.1 Carinal side 2.6-2.7 2.7-2.8 2.6-2.9 Width (cm) Greatest 0.8 1.0-1.1 1.2-1.3 Middle 0.8 1.0-1.1 1271.3 Indumentum glabrous sparsely puberulent glabrous Texture minutely papillate minutely papillate olor pale red red to reddish brown reddish brown COROLLA Standard olor red 8/7.5 red 8/7.5 red 6/7.5 Length (cm) 8.7 7.5 6.5 Greatest width (cm) 1.6 1.6 1.8 Wings Shape apex acute apex acute to rounded apex rounded Length (cm) 1.2 1.1 1.3 Width (cm) 0.3 0.4 0.3 Keel Shape emarginate; each half apex short-apiculate apex rounded acute at apex Length (cm) 1.0 1.1 1.3 Width (cm) 0.8 1.0 1.0 E. guatemalensis E. tajumulcensis Figure 104 WA 74c1453 PT 820547001 WA 74c1448 CALYX Shape elliptic oblong oblong Apex shape Length (cm) Vexillar side Carinal side Width (cm) Greatest Middle Indumentum Texture Color COROLLA Standard Length (cm) Greatest width (cm) variable; bilabiate or oblique 2.3-2.7 2.6-2.9 1.2-1.3 1.2-1.3 glabrous minutely papillate reddish brown red 6/7.5 6.5 1.8 oblique; longer on cari- nal side 0.9 glabrous smooth red 7/7.5 1.5 oblique; longer on carinal side glabrous smoot red 6/7.5 9.3 1.3 940 Annals of the Missouri Botanical Garden TABLE 19. Continued. Female Parent F, Hybrid Male Parent Length (cm) Width (cm) Keel Shape Length (cm) Width (cm) oblong; apex rounded 1.3 0.4 apex rounded 1.3 1.0 oblong; apex rounded 1.0 0.4 emarginate; each half short-apiculate at apex 1 1.8 oblong, apex rounded 1.0 0.3 apex long-acuminate 1.3 1.0 Figure 105 E. guatemalensis PT 700018001 HO 82.282 E. folkersii PT 700010001 CALYX Shape Apex shape Length (cm) elliptic; broadest in mid- dle irregular; oblique or bila- i cuneate; broadest at apex oblique to truncate cuneate; broadest at apex oblique; longer on side Vexillar side 2.3-2.7 1.8-1.9 1.8-1.9 Carinal side 2.6-2.9 24-29 2.1-2.2 Width (cm) Greatest 1.2-1.3 1.2-1.3 1.2 Middle 1.2-1.3 1.0 0.8 Indumentum glabrous sparsely puberulent densely puberulent Texture minutely papillate minutely papillate minutely papillate Color reddish brown reddish brown to purple- light brown rown COROLLA Standard red 6/7.5 red 6/10 red 7/7.5 Length (cm) 6.5 7.0-7.8 Greatest width (cm) 1.8 2.2 2.4 Wings Shape apex rounded apex rounded apex rounded Length (cm) 1.3 1.0 Width (cm) 0.4 0.5 0.5 Keel Shape apex rounded apex variable; rounded emarginate, each half ob- or emarginate Length (cm) 1.3 -1.3 1.0 Width (cm) 1.0 0.9 1.3 HO 82.284 E. guatemalensis PT 820278002 E. chiapasana Figure 106 PT 700018001 (2 individuals) PT 73071001 CALYX Shape liptic oblong to elliptic ong Apex shape variable; bilabiate or truncate to oblique; 5 truncate; 5 apical lobes in Length (cm) Vexillar side Carinal side oblique 2.3-2.7 2.6-2.9 apical lobes in bud, absent in anthesis 1.7-1.9 1.7-2.1 ud, absent at anthesis 1.5-1.6 1.5-1.6 Volume 75, Number 3 Neill 941 1988 Erythrina TABLE 19. Continued. Female Parent F, Hybrid Male Parent Width (cm) Greatest 1.2-1.3 0.9 0.8 iddle 1,2-1.3 0.8-0.9 0.7 Indumentum glabrous sparsely puberulent densely puberulent Texture minutely papillate smooth to indistinctly 5-angled Color reddish brown reddish brown to green green COROLLA Standard Color red 6/7.5 red 5/5 to 5/7.5 red 5/7.5 Length (cm) 6.5 6.5 7.4 Greatest width (cm) 1.8 1.5-1.8 1.4 Length (cm) Width (cm) Keel Shape oblong; apex rounded 1.3 0.4 apex rounded oblong; apex rounded 1.5 0.4 variable; apex rounded oblong; apex rounded 1.3-1.4 0.4 emarginate; each half emarginate short-apiculate Length (cm) 1.3 1.0-1.3 E Width (cm) 1.0 0.9-1.0 0.9 HO 82.285 E. guatemalensis HO 82.288 E. macrophylla Figure 107 PT 700018001 PT 820276001 PT 750420002 CALYX Shape elliptic; broadest in mid- oblong to elliptic cuneate; broadest at apex le Apex shape truncate to slightly truncate; 5 prominent, Length (cm) Vexillar side Carinal side Width (cm) Greatest Middle Indumentum Texture Color COROLLA Standard Color Length (cm) Greatest width (cm) P Length (cm) Width (cm) Keel Shape Length (cm) Width (cm) variable; bilabiate or ique 2.3-2.1 2.6-2.9 1.2-1.3 1.2-1.3 glabrous minutely papillate reddish brown oblong; apex rounded 1.3 0.4 apex rounded 1.3 1.0 oblique; 5 irregular apical lobes 1.8-2.5 2.3-2.6 1.0-1.2 1.0-1.2 sparsely to densely pu- berulent obscurely 5-angled reddish brown to green red 6/7.5 to 6/10 6.4-7.0 1.7-2.1 oblong; apex rounded 1.3-1.6 0.4-0.5 apex short-apiculate 1.1-1.3 1.1-1.2 blunt lobes 2.0 2.0 1.2 (at apex) 0.9 densely puberulent longitudinally 5-angled brown to green red 5/7.5 6.4 1.7 oblong; apex rounded 1.3 0.4 apex short-apiculate 1.3 1.1 942 Annals of the Missouri Botanical Garden TABLE 19. Continued. Female Parent F, Hybrid Male Parent Figure 108 E. macrophylla PT 750420002 (2 individuals) E. guatemalensis PT 700018001 CALYX Shape Apex shape Length (cm) cuneate; broadest at apex truncate; 5 prominent apical lobes elliptic variable; bilabiate or uncate; 5 indistinct apical lobes present or 8 lackin elliptic; bilabiate in middle variable; bilabiate or oblique Vexillar side 2.0 1.9-2.2 2.3-2.7 Carinal side 2.0 2.0-2.4 2.6-2.9 Width (cm) Greatest 1.2 (at apex) 1.1-1.2 1.2-1.3 Middle 0.9 1.1-1.2 1,2-1.3 Indumentum densely puberulent densely puberulent glabrous Texture longitudinally 5-angled smooth to minutely papil- minutely papillate late Color brown to green reddish brown to green reddish brown COROLLA Standard Color red 5/7.5 red 5/7.5 red 6/7.5 Length (cm) 6.4 6.5-6.7 Greatest width (cm) 1.7 1.8 1.8 Wings Shape oblong; apex rounded oblong; apex rounded oblong; apex rounded Length (cm) 1.3 3 1.3 Width (cm) 0.4 0.4-0.5 0.4 Keel Shape apex short-apiculate apex rouuded apex rounded Length (cm) 1.3 1.0-1.1 1.3 Width (cm) 1.1 1.1-1.2 1.0 E. chiapasana ; E. beteroana Figure 109 PT 721005001 (2 individuals) PT 700044002 CALYX Shape oblon oblong Apex shape truncate; 5 indistinct truncate oblique; longer on carinal apical knobs side Length (cm) Vexillar side 1.5-1.6 2.2-2.3 2.8 Carinal side 1.5-1.6 2.2-2.4 3.0 Width (cm) Greatest 0.8 0.9 0.8 Middle 0.7 0.8-0.9 0.7 Indumentum Texture indistinct longitudinal indistinct longitudinal smooth ridges ridges Color green pale green to red pale red COROLLA Standard olor red 5/7.5 red 6/5 to 8/7.5 red 8/7.5 to 9/7.5 Length (cm) 6.8 8.5 8.3 Greatest width (cm) 1.4 1.6-1.7 1.6 Volume 75, Number 3 Neill 943 1988 Erythrina TABLE 19. Continued. Female Parent F, Hybrid Male Parent Wings Shape oblong; apex rounded oblong; apex acute to oblong; apex acute rounded Length (cm) 1.5 - 1.0 Width (cm) 0.4 0.4 0.4 Keel Shape emarginate, each half apex rounded to emar- deeply emarginate; each short apiculate ginate half acute at apex Length (cm) .8 Width (cm) 0.9 0.9 0.9 E. macrophylla š E. beteroana Figure 110 PT 750420002 (2 individuals) PT 700044001 CALYX Shape cuneate; broadest at oblong oblong, narrow ex Apex shape truncate; 5 prominent truncate; 5 indistinct oblique; longer on carinal ical lobes apical lobes i Length (cm) Vexillar side 2.0 2.2-2.4 2.8 Carinal side 2.0 2.2-2,4 3.0 Width (cm) Greatest 1.2 1.0-1.1 0.7 Middle 0.9 0.7-0.8 0.7 Indumentum densely puberulent sparsely puberulent glabrous Texture longitudinally 5-angled indistinct longitudinal ridges Color brown to green green to red pale red to green COROLLA Standard Color red 5/7.5 red 6/7.5 red 8/7.5 Length (cm) 6.4 7.3-7.4 8.3 Greatest width (cm) 1.7 1.5-1.6 1.3 Wings Shape oblong; apex rounded oblong; apex acute to oblong; apex acute rounded Length (cm) 1.3 1.0-1.2 1.0 Width (cm) 0.4 0.4 0.4 Keel Shape apex short-apiculate apex apiculate or emar- deeply emarginate; each ginate half acute at apex Length (cm) 1.3 LU-LI 8 Width (cm) 1.1 0.8-1.0 0.9 of matrocliny or female-dominant inheritance in the Erythrina hybrids. The results, although preliminary, also suggest a difference in the genetics of inheritance of hairs on the one hand, and papillae and lamellae on the other. Hairs are inherited in the hybrids as discrete characters—that is, they are fully formed and of normal size, although they occur at low densities, in crosses between hairy and hairless parents. The formation of hairs may thus be controlled by a single gene or supergene, with modifiers controlling density of the hairs. On the other hand, in crosses between papillate (or lamellate) and nonpapillate d papillae (lamellae) may be present but they much reduced in stature. This suggests that iB stature of papillae and lamellae are continu- ously variable, typical of morphometric traits, and controlled by many genes, each of small effect. B 944 Annals of the Missouri Botanical Garden TABLE 20. Comparison of flowers of Erythrina atitlanensis and F, hybrid, E. macrophylla x E. berteroana. (Photo, Fig. 111.) E. atitlanensis E. macrophylla x E. berteroana HO 82.281 WA 74s98 WA 75s1141 CALYX Shape oblong Apex shape truncate; 5 indistinct apical lobes oblong truncate; 5 indistinct apical lobes Length (cm) Vexillar side 1.7-1.8 Carinal side 1.8-1.9 Width (cm) Greatest 1.0 Middle 0.7 Indumentum sparsely puberulen Texture indistinct longitudinal ridges Color pale green to re COROLLA Standard Color red 6/7.5 Length (cm) 6.5-6.8 Greatest width (cm) 1.7 Wings Shape oblong; apex rounded Length (cm) 1.0 Width (cm) 0.4 Keel Shape apex short-apiculate Length (cm) 0.8 Width (cm) 1.0 2.2-2.4 2.2-2.4 1.0-1.1 0.7-0.8 sparsely puberulent indistinct longitudinal ridges green to re red 6/7.5 to 7/10 7.3-7.4 1.5-1.6 oblong; apex acute or rounded 1.0-1.1 0.4 apex apiculate or emarginate 1.0-1.1 0.8-1.0 This suggestion requires confirmation by analysis of segregation in the F, generation The variation in the four F, hybrid siblings de- rived from a single cross, Erythrina crista-galli x E. guatemalensis, is illustrated in Figures 76-87. The female parent has an irregular network of low lamellae less than 10 um tall and lacks epicuticular wax. The male parent has a dense covering of unicellular papillae up to 40 um tall and has epi- cuticular wax. The F, hybrids exhibit a narrowly segregating array of these characters: they have papillae and/or lamellae intermediate in form and stature between the two parents. Some Fs resemble the female parent more closely and some resemble the male parent. All the F,s have epicuticular wax, evidently derived from the male parent. (Other individuals of E. crista- a besides the one used in this cross, do posse The leaf epidermis of d F, siblings derived from the cross Erythrina crista-galli x E. fusca, together with their parents, are illustrated in the two plates comprising Figures 88-99. Also includ- ed on these plates are photos of E. dominguezii, a species that, based on floral characters, may be a stabilized derivative of hybridization between K. crista-galli and E. fusca (see discussion of hybrid flowers below). The male parent Erythrina fusca has a very unusual epidermal surface with deep and irregularly sized and shaped cavities, knobs and protrusions, appearing under SEM much like the surface of a limestone cavern. The stomata are at the bottom of the cavities. The F, hybrids vary somewhat in surface con- figuration, but none of them possess either the regular network of lamellae present in K. crista- galli or the complex, irregular cavity structure of E. fusca. Two of the Fs have scattered low papillae and one is nearly flat on the abaxial surface. All three F,s have epicuticular platelets, which are not present in either of the parents. Erythrina dominguezii (Figs. 93, 99) has scat- tered balloonlike hairs, not present in EF. crista- galli or E. fusca. Otherwise, the epidermal surface on E. dominguezii has no distinctive features. Low Volume 75, Number 3 1988 Neill Erythrina LIUC UU KL K aa FIGURE 103. Flowers and buds of Erythrina berteroana x E. guatemalensis and parents. Left: E. berteroana, WA 785564, female parent. Center: E. berteroana X E. EK HO 82.647. Right: E. guatemalensis, WA 74c1453, male parent. epidermal convolutions similar to the lamellae of E. crista-galli are visible at high magnification (Fig. 99) but these are not organized into a regular reticulate pattern. FLORAL FEATURES: INHERITANCE IN INTERSPECIFIC HYBRIDS The inheritance of floral morphology and color was examined in the hybrids that flowered by No- MAT "E FIGURE 104. malensis, WA 74c1453, female parent. Center: E. gu tajumulcensis, W/A 74c1448, male parent. lek le hs Flowers and buds of ul m guatemalensis x E. K py was, n and parents. Lefi: E. guate- uatemalensis x E. vember 1984. These included the hybrids within sect. Erythrina and the intersectional hybrid Er- ythrina crista-galli x E. fusca Materials and Methods Fresh flowers of the hybrids and parents were fixed in FAA, which preserved their three-dimen- sional shape, and later measured, described, and photographed, each hybrid together with its par- 104 ajumulcensis, PT 820547. Right: E. Annals of the Missouri Botanical Garden SAC LT LL E DF LC |] M 1 2 37455 '6 37 '8 10 FIGURE 105. Flowers and buds of Erythrina guatemalensis x E. folkersii and parents. Left: E. dd aris PT 700018001, female parent. oe E . guatemalensis x folkersii, PT 700010001, male paren ents. Color was determined from fresh flowers at the time of collection. A Berlin & Kay color chart (Berlin & Kay, 1969) was used for color descrip tions of corolla standards. For use of the Berlin & Kay color chart in botanical descriptions see Croat | Y” — “> "t m a | | | | | | m | A a E 4 s A |. |, B T FIGURE 106. . Center: E. X E. folkersii, two F, siblings, HO 82.282. Right: (1983). Colors are reported in the form “Red 5/7.5.” The number preceding the slash refers to brightness (1-9; 1 is brightest) and the number after the slash refers to the hue. Inflorescences and floral details were photo- 106 Flowers ne n of Erythrina guatemalensis x E. chiapasana and parents. Lefi: E. guatemalensis, PT 700018001, female p hi guatemalensis x E . chiapasana, two F, siblings, HO 82.284 and PT 820278002. Right: E. d. PT 730710001, male parent. Volume 75, Number 3 Neill 947 1988 Erythrina | | | P FicURE 107. Flowers and buds of Erythrina guatemalensis x E. macrophylla and parents. Left: E. guatemalensis, PT 700018001, female parent. Center: E. guatemalensis x E. mac rophylla, £ hs F, siblings, HO 82.285, HO 82.288, PT 820276001. Right: E. macrophylla, PT 750420002, "in paren o — H 9 10 EU 107 E s. als A P TAM ap MAR 108 1 2 13 4 FIGURE 108. Flowers and buds of a gie xeu i ase x E. guatemalensis and parents. Left: E. macrophylla, PT 750420002, female parent. Center: E. A: lla x E. guatemalensis, two F, siblings, HO 82.763. Right: E. guatemalensis, PT 700018001, male paren a 948 Annals of the Missouri Botanical Garden IU LL a LE ho 109 FIGURE 109. Flowers and S Erythrina ela X E. berteroana and parents. Left: E. chiapasana, PT 721005001, female parent. Center: E. chiapas X E. berteroana, two F, siblings, HO 82.278. Right: E. berteroana, PT 700044002, boc nion — ml | |, | Á k l. ñ |. |. |. ka 110 FIGURE 110. Flowers and buds of Erythrina macrophylla x E. berteroana and parents. Left: E. macrophylla, PT 750420002, female parent. Center: = macrophylla x E. berteroana, two F, siblings, HO 82.281. Right: E. berteroana, PT 700044001, male paren Volume 75, Number 3 Neill 949 1988 Erythrina O I!) CM 12 3 4 i5 '6 7 8 9 10 111 FIGURE lll. Comparison of flowers of Erythrina macrophylla x E. berteroana and Erythrina atitlanensis. Left: E. macrophylla x E. berteroana, HO 82.281, two F, siblings. Top right: E. atitlanensis, WA 74598. Bottom right: E. atitlanensis, WA 75s1141. graphed at the time of collection. Herbarium vouchers are deposited at Missouri Botanical Gar- den (MO). Results Hybrids within Sect. Erythrina Inflorescence and Flower Orientation. Species of sect. Erythrina all have erect inflorescences, but they differ in the arrangement of the flowers on the inflorescence axis (congested or open), length of the axis, and orientation of the flowers (ascend- ing, horizontal, or descending). These traits are generally intermediate in the hybrids (e.g., Figs. 100-102). In Erythrina guatemalensis, the fe- male parent, the flowers are horizontal on an open inflorescence. In E. folkersii, the male parent, the flowers descend to nearly vertical on a congested inflorescence. The F, hybrid is intermediate in both these traits. Floral Characters. A comparison of floral char- acters of the hybrids and their parents within sect. Erythrina is summarized in Table 19. The flowers are illustrated in Figures 103-110. In all characters—color, indumentum, shape, and morphometric dimensions—the F, hybrids are intermediate between the two parents. The F, sib- lings from a single cross vary to some extent. There is no evidence of matrocliny or maternal domi- nance. Subjectively, some of the hybrids resemble the male parent more closely than the female parent. This is evident in the progeny of the reciprocal titure and shapes of the calyces of the hybrids are 950 Annals of the Missouri Botanical Garden Comparison of flowers of Erythrina crista-galli x E. fusca, its parents, and E. dominguezii. E. fusca PT 840231001 E. dominguezü WA 74s865 TaBLE 21. E. crista-galli x E. crista-galli E. fusca WA 74p840 PT 84021001 Orientation resupinate; stan- standard semi-cleis- rd beneath, togamous, folded open over reproduc- tive parts CALYX Color red reddish brown Shape bowl shaped bowl shaped; slight- asymmetric ngt Carinal side 1.8 cm 1.6 cm Width at apex 1.4 cm Apex ornamen- 1.0 cm large subulate narrow tooth, ca- id tation tooth, carinal rinal side side COROLLA Standard Color red orange Length 5.0 cm 6.1 cm Width 3.3 cm 4.9 cm Wings Shape asymmetric; broad- obovate, rounded est at base cucullate Length 1.3 cm 1.8 cm Width 0.9 cm 1.0 cm Keel Color red pale red Shape falcate, acute at ovate-falcate, rounded at apex Length 4.5 cm 3.4 cm Width 1.0 cm 1.5 em standard reflexed brown asymmetrically bowl shaped large blunt tooth, carinal side orange-pink 9 cm 9.9 cm obovate, cucullate, broadly rounded ivory at base, red ovate-falcate, broadly rounded standard semi-cleis- togamous, folded over reproductive parts pale orange-pink asymmetrically bowl shaped Hlan tooth, carinal side pale orange-pink 4.8 cm 3.5 cm small, obovate, cucul- late 0.6 cm 0.4 cm pale red falcate, acute at apex 1.5 cm 1.0 cm intermediate and variable, but in both reciprocals the hybrids resemble the male parent somewhat more than the female. osses involving species with tomentose and glabrous calyces, the hybrids are tomentose, but sometimes sparsely so. In color values and mor- phometric dimensions, the hybrids are generally intermediate between the two parents. e flower of the hybrid Erythrina macro- phylla x E. berteroana is intermediate between the two parents, and it closely resembles a third recognized species, E. atitlanensis (Fig. 111, Ta- ble 20). The principal difference is that the calyx of the F, hybrid is somewhat longer than that of E. atitlanensis. The natural distribution of Ery- thrina atitlanensis is confined to a small area near Lake Atitlan in western Guatemala, and it is geo- graphically and ecologically intermediate between E. macrophylla and E. berteroana. The possibility that the form known as E. atitlanensis represents either a hybridizing population or a stabilized species of hybrid origin will be discussed below. Flowers of Erythrina crista-galli x E. fusca The inflorescence and flowers of this intersec- tional hybrid and its parents are illustrated in Fig- ures 112-117 and described in Table 21. A third species, E. dominguezii, is included in the illus- trations and descriptions for reasons discussed be- o = The morphometric dimensions and proportions of the floral parts of the two parental species are relatively similar, considering the total range of Volume 75, Number 3 1988 Neill Erythrina FIGURES 112-115. dominguezii.— 7 12. E. crista-galli, 840231001.—115. E. dominguezii, PT 74023 4001. variation of these traits in the genus Erythrina, but the way in which these parts are arranged and the overall appearance of the flowers are very different. The flower of E. crista-galli is resupinate (inverted from the usual position, with the standard below the keel) and the red standard is flattened out, an unusual trait in Erythrina. The orange standard of E. fusca is reflexed from the clawed base, exposing the reproductive parts, and is broad- ly folded down the middle. The flower of the F, hybrid E. crista-galli x E. fusca is different from either parent. The hybrid flower is semicleistogamous, with the standard tightly folded over the wings, keel, and reproductive parts. In this semicleistogamous form, in the orientation Inflorescences da natural position of Erythrina crista-galli x E. fusca, its parents, and E. d 3. E. fusca, WA 74s9 9. — 114. E. crista-galli x E. fusca, PT of the flowers and the inflorescence, and in the pale pink-orange color of the corolla standard, E. crista-galli x E. fusca bears a striking resem blance to E. dominguezii (Figs. 114-117). Qer- tainly the hybrid resembles E. dominguezii more closely in overall appearance than either of its parents. The dimensions of the floral parts are not identical in the F, hybrid and in E. dominguezii (Fig. 117, Table 21). The overall similarity be- tween the two could be a coincidence, but it is so striking and so unexpected that it raises the pos- sibility that Erythrina dominguezii is in fact a hybrid derivative of E. crista-galli and E. fusca. The distribution of E. dominguezii is geographi- cally intermediate but ecologically distinct from E. 952 Annals of the Missouri Botanical Garden FIGURE 116. Flowers of Erythrina crista-galli x E. fusca, its parents, and E. dominguezii, showing approximate natural orientation. Top row, lefi to right: E. crista-galli, WA 75p840, female parent; E. crista- E . fusca, PT 840231001; E. fusca, WA 74s99, male parent. Bottom center: E. dominguezii, PT 74023400 crista-galli and E. fusca. This will be discussed in greater detail in Section 6. Conclusions: Inheritance of Phenetic Traits in Interspecific Hybrids The results of the morphological studies of the F, hybrids clearly demonstrate that the progeny are indeed of hybrid origin. Almost universally, the F, progeny meet the criterion of intermediacy, and frequently they possess traits present in the male parent but absent in the female parent. Matroc p is not indicated in Erythrina hybrids. Some of t F, hybrids closely resemble forms occurring in nat- ural populations and recognized as species. SECTION 6. NATURAL HYBRIDIZATION AND HYBRID SPECIATION The Mexican state of Chiapas has great geo- graphical diversity and complexity and a very large flora for an area its size. Climate ranges from and elevation from sea level to over 4,000 m. The flora of Chiapas contains more than 8,000 plant species and 13 major vege- semidesert to rainforest, tational formations recognized by Breedlove (1981). Chiapas, together with adjacent western Gua- temala, is also the center of diversity of Erythrina sect. Erythrina. Eleven species are known to occur in the state, and six of these are endemic or shared only with western Guatemala. Although they are not found in abundance or in large populations, species of Erythrina occur in virtually every vege- tation type in Chiapas except the upper belts of cloud forest and elfin forest on the highest peaks n common with the usual pattern of distribution in Erythrina, the Chiapas species of the genus are mostly allopatric. However, at some localities, par- ticularly at the margins of distribution of the species, different species do come into contact, and there natural hybrids are formed. One phenomenon that has apparently occurred with Erythrina in Chiapas and perhaps elsewhere is spontaneous hybridization in man-made popu- lations. Throughout Mesoamerica many species of Erythrina trees are used by the local populace as “living fenceposts." Erythrinas take root readily from woody cuttings and the trunks are ideal posts for stringing barbed wire. Extensive fencerows of the plants line roads and fields in many areas. Volume 75, Number 3 1988 Neill 953 Erythrina FIGURE 117. left to right: standard, wings, keel). Sometimes two species are cultivated together, and hybrids, apparently produced spontaneously in situ, are occasionally found in these fencerows. An analysis of hybridizing populations involving three species of Erythrina in central Chiapas, Er- ythrina chiapasana, E. goldmanii, and E. pu- dica, is presented below. Distributions of these species and their hybrid populations are shown in 8 Figure 11 Erythrina chiapasana X E. goldmanii Erythrina chiapasana is a tree of the pine- oak forests of the Central Plateau of Chiapas, oc- curring primarily above 1,500 m. Erythrina gold- manii inhabits the dry tropical deciduous forests of the Central Depression of Chiapas, formed by the highland-rimmed valley of the Rio Grijalva. At El Sumidero National Park a few km north of the city of Tuxtla Gutierrez, where the Rio Grijalva cuts through the limestone of the Central Plateau on its way to the Atlantic Ocean and forms a spectacular 800-m-deep canyon, the two species occur parapatrically and a hybrid zone is found Dissected petals lira crista-galli x E. fusca, its parents, and E. dominguezii (each flower, crista-galli, WA 74p840, female parent. Center left: E. crista-galli x E. fusca, PT 840231001. Center right: H dominguezi, PT 740234001. Bottom: E. fusca, WA 74599, male parent. about 2 km wide and extending about 300 m along an elevational gradient (Fig. 119) Throughout their respective distributions, Ery- thrina chiapasana and E. goldmanii exhibit some intraspecific variation, but the two species are readily distinguishable morphologically. The leaves of F. chiapasana are densely tomentose ay two-armed hairs on the abaxial surface (Fig. 120). The leaves of E. goldmanii are glabrous or nearly so at ma- turity and are aculeate along the midvein and pri- mary veins of the abaxial surface (Fig. 122). The calyx of E. chiapasana is green to reddish, densely puberulent, and truncate at the margin without a prominent tooth on the carinal side; the corolla standard is dark red. The calyx of E. goldmanii is broader, dark purple brown to nearly black, glabrous, and provided with a prominent apical tooth on the carinal side; the corolla standard is usually pale red ] Sumidero both species are at the altitudinal and geographical limits of their ranges. Only in- dividuals with the “pure” E. chiapasana pheno- type are found in the oak-dominated forest at the plateau summit above 1,100 m; only individuals 954 Annals of the Missouri Botanical Garden ^ VERACRUZ ^ ^ OAXACA | ° ^ e d ° MED ^ e on ' Mm. Me e de . Tuxtla ! " Gutiérrez 9 16 FIGURE 118. Mexico. Squares—E. chiapasana; circles—E. goldmanii; solid stars—E. goldmanii x E. pudi with the “pure” E. goldmanii phenotype are found in the dry scrub forest below 800 m. In the tran- sition zone near the top of the escarpment between 800 m and 1,100 m there are plants with inter- mediate phenotypes, or displaying in one individual various combinations of traits of both species. Some individuals, for example, have leaves that are sparsely tomentose on the abaxial surface and are also aculeate on the midvein (Fig. 121). Others have flowers with characters intermediate between parental traits such as a puberulent, dark purple- brown calyx with an apical tooth (Fig. 123). Both parental types are also present in the transition zone. This pattern of variation in the intermediate zone at El Sumidero establishes with reasonable cer- » ." a e s C a a Ban md. * Las Casas a wa e he a a e a " Comitán O a a g- “= - > / ° / ° ° / / / / ° / " ° Í GUATEMALA ° / / Motozintiag” e ° / ` Distribution of Erythrina chiapasana, E. goldmanii, E. pudica, and hybrid populations in Chiapas, g0 ra diamonds—E. pudica; open s ars—E. chiapasana X E. tainty that the population is a hybrid swarm of Erythrina chiapasana X E. goldmanii. The in- termediacy of the traits in this population resembles the patterns of inheritance expressed in the ex- perimentally produced hybrids as discussed in Sec- tion 5 As indicated in Section 4, I attempted to syn- thesize hybrids between Erythrina chiapasana and E. goldmanii in the field at El Sumidero using the same techniques of controlled hand-pollination em- ployed in the experimental gardens in Hawaii. Hy- brid fruits in both reciprocal crosses were obtained, ut the fruit of Erythrina chiapasana 9 x E. goldmanii à was destroyed in a brush fire. The reciprocal E. goldmanii 9 X E. chiapasana ó produced one mature hybrid seed. The F, was viable and is now growing in cultivation in Hawaii Volume 75, Number 3 Neill 955 1988 Erythrina CANON DEL SUMIDERO, CHIAPAS ELEV ERYTHRINA CHIAPASANA 1200 — z = 1100 HYBRID ZONE 1000 LOWER MONTANE MOIST FOREST 900 800 ERYTHRINA GOLDMANII 700 600 «mm NORTH 500 m i TROPICAL DRY FOREST | FiGURE 119. Cross section of slope at El Sumidero, Chiapas, Mexico, showing distribution of Erythrina chiapasana, E. goldmanii, and hybrid zone. alongside accessions of both parental strains (Table 11). When this artificial hybrid flowers it will be possible to compare it with o di of the pu- tative natural hybrids from El Sumide As discussed in a separate paper (Neill, 1987), the hummingbird Heliomaster constantii polli- nates Erythrina chiapasana and E. goldmanii at El Sumidero and is therefore implicated as the agent directly responsible for interspecific gene flow in the hybridizing Erythrina population. Erythrina goldmanii X E. pudica Erythrina pudica is a locally endemic species that is restricted to the dry valley of the Río de La Venta, a tributary of the Rio Grijalva, at the western end of the Central Depression of Chiapas. This is an unusual species, with the flowers drooping to nearly parallel with the erect axis of the inflo- rescence (Fig. 126). The calyx is truncate without an apical tooth and covered with a dense grayish tomentum; the corolla is very pale pink or orange- pink. In the vicinity of Ocozocuautla, Chiapas, at the eastern margin of its small range, Erythrina pu- dica occurs sympatrically with Erythrina gold- manii. In disturbed scrub forest small hybrid pop- ulations are found, with individuals of both parental species as well as intermediates (Figs. 124-126). Along the highway 5 km east of Ocozocuautla are living fencerows of Erythrina containing both E. goldmanii and E. pudica, and occasional in- termediate and evidently hybrid individuals occur hybrid populations. The fencepost hybrids are very likely the progeny of other fencepost trees that received interspecific pollen from foraging hum- mingbirds moving down the line of mixed species fenceposts. The hybrid seed thus probably germi- nated directly below its female parent and grew up to become part of the fencerow itself. Erythrina berteroana x E. folkersii On the Atlantic coastal plain of northern Chiapas and adjacent states the natural vegetation has been almost entirely destroyed and replaced with pas- tures. There, as elsewhere in Mesoamerica, the pastures and roadsides are commonly lined with living fencerows of Erythrina trees. On the Atlan- tic plain the most frequently used species are E. berteroana and E. folkersii, which are both native to the region. Trees morphologically intermediate between Er- ythrina berteroana and E. folkersii in shape and vestiture of the calyx and orientation of the flower occur in northern Chiapas. None of the interme- diates set seed. Pollen stainability from four col- lections of the intermediates (Alexander's stain; 500 grains per sample: Neill 5533, 5540, 5543, 5544) was 73.2% (range 60.9-86.1%), an un- 956 Annals of the Missouri Botanical Garden usually low figure for Erythrina. These individuals are almost certainly hybrid Erythrina berter- oana X E. folkersii. The reason for the low level of stainable pollen and lack of fruit set is not known; the experimentally produced hybrids within sect. Erythrina (Section 4) all had very high pollen fertility. These intermediates closely match the type specimen of Erythrina caribaea Krukoff & Bar- neby as well as other collections determined by Krukoff as this species. Despite a protracted search, I never found this form occurring in a natural population and never found any seed set on the fencepost trees. It seems reasonable that Erythrina caribaea is in fact a hybrid E. berteroana x E. folkersii and probably occurs only as a cultivated fencepost tree. HYBRID SPECIATION In this paper it has been demonstrated that diploid Erythrina species are interfertile, that the hybrids are viable and fertile, and that hybridization sometimes occurs in natural populations. What has not yet been shown is the validity of the final hypothesis set forth in the introductory chapter: that hybrid speciation has taken place in Erythri- na, that some distinct forms recognized as species are stabilized derivatives resulting from hybridiza- tion of two parental species, and that this process has been an important element in the evolutionary history of the genus. Direct and unequivocal evidence relating to this hypothesis is difficult to obtain. Phylogenies based on molecular data of the taxa involved, including studies on isoenzymes and on nucleic acid restric- tion sites, might in the future provide such evi- dence. The best evidence available at present is the morphological congruence between certain ar- tificially produced hybrids and certain naturally occurring forms that evidently are stabilized and self-perpetuating populations. In considering the hypothesis of hybrid specia- tion, there is no reason to assume that the stabilized derivatives, especially if they became stabilized sev- eral generations or more after the original hybrid- ization event, should be precisely pte EOI pa: Fıcures 120-122. SEM images, abaxial leaf sur- faces of Erythrir a chiapasana, E. pM and hybrid tween the parental species or should from a diio at El Sumidero, Chiapa . Mexico.— the F, hybrids. A limited number of F, sce is, ). E. iere a ill 5617. —121. geo asana X however, generally the only material available for E. werde, Neill 5 — 122. E. goldmanii, Neill 5616. comparison. ERIS Darse Lo ume n Erythrina some of the artificial F, hybrids do resemble naturally occurring forms recognized as species, according to the results of the mor- phological studies presented in Section 5. Ery- Volume 75, Number 3 1988 Neill 957 Erythrina c |, ñ k: A L FIGURE 123. Chiapas, Mexico. Left: E. chiapasana, Neill 5 "ME v of Erythrina É um E. goldmanii, ps hybrid from a population at El Sumidero, 455. Center top: E. 123 asana X E. goldmanii, Neill 5493. Center iap bottom: E. a x E. goldmanii, Neill 5466. Right: E. zoldmani, Neill 5495. thrina macrophylla x E. berteroana bears a close resemblance to E. atitlanensis, and E. galli x E. fusca resembles in certain features E. dominguezii. Field studies, which would be valu- able for determining whether hybrid speciation could have occurred, were not conducted in either of these situations. The known geographical and eco- logical distribution of the taxa involved is outlined below. Erythrina macrophylla is distributed through- out the highlands of Guatemala and western El Salvador, growing in the pine-oak forests above 1,500 m elevation. Erythrina berteroana, the most widespread species in sect. Erythrina, is common in the Pacific coastal plain of Guatemala and on the lower slopes of the volcanic range that lead up from the plain to the highlands. The intermediate known as Erythrina atitlanensis is known onl from the vicinity of Lake Atitlan on the southern edge of the highlands. In terms of geography and elevational distribution, E. atitlanensis is precisely intermediate between the putative parental species. If hybridization is really implicated in this case, E. atitlanensis could be merely an early generation segregate rather than a stabilized, self-perpetuating derivative. Based on comparison of herbarium crista- specimens, the progeny cultivated in Hawaii grown from seed obtained from the population in Gua- temala closely resemble the parents. Therefore sta- bilization of the hybrid form may have taken place. The case of Erythrina dominguezii and its pu- tative parental species E. crista-galli and E. fusca is more problematic because the three taxa are so morphologically distinct. They are also ecologically distinct. Erythrina crista-galli and E. fusca are both riparian or estuarine species. Erythrina cris- ta-galli is common along the estuary of the Rio de La Plata and its tributaries and along the coast of southern Brazil. The more tropical E. fusca is distributed widely throughout the Amazon basin and south along the coast of Brazil. The ranges of the two species evidently do overlap in southern Brazil. The putative derivative Erythrina domin- guezii also occurs in southern Brazil and westward through Paraguay and northern Argentina to east- ern Bolivia, but it is an upland species of the dry Chaco forest and cerrado. Erythrina dominguezii would never have been suspected as a hybrid de- rivative of E. crista-galli X E. fusca were not its resemblance to the artificially produced F, so com- pelling. This situation appears to merit further in- vestigation. SUMMARY AND CONCLUSION In the introduction, a set of five hypotheses was stated regarding the species relationships and evo- lutionary history of Erythrina: 1) The numerous species of sect. Erythrina can all cross freely with one another, producing fully fertile hybrids. The section forms a homogamic complex in which in- 958 Annals of the Missouri Botanical Garden FIGURES 124-126. Ocozocuautla, os qun i 126. E. pudica, Neill 5 ternal barriers to hybridization are absent. 2) The interfertile homogamic complex of sect. Erythrina extends, to a greater or lesser degree, to species in other sections and subgenera of Erythrina. Any diploid Erythrina species can hybridize with any other, but crosses between widely divergent taxa are generally difficult to obtain and the resulting F,s may exhibit varying degrees of sterility. The genus as a whole may be characterized as a series of interfertile homogamic complexes with weak to moderate reproductive barriers between the com- eli ences á Erythrina goldmanii, 24. E. goldmanii, Neill 5510.— 725. E. goldmanii x E. pudica, and hybrid Em a d = E. pudica, Neill 5 — plexes. 3) The widely foraging hummingbirds that pollinate species of sect. Erythrina are capable of effecting interspecific pollen flow between sympat- ric species of sect. Erythrina. 4) Sympatry at the local community level is rare among species of sect. Erythrina. Most species are restricted in geo- graphic range and ecological amplitude and are allopatric, separated by habitat differences. How- ever, sometimes different species do come into con- tact in nature, and then hybridizing populations are formed. 5) Patterns of distribution and phenetic Volume 75, Number 3 1988 Neill 959 Erythrina variation in sect. Erythrina indicate that some distinct forms recognized as species are stabilized derivatives resulting from hybridization of two pa- rental species. As a consequence of changing cli- mates and dynamic geomorphological processes, and the consequent migration of vegetation types and mixing of floristic elements, formerly allopatric species may have come into contact a number of times. With the temporary breakdown of external isolating barriers, the interfertile species hybridized and the subsequent segregation and stabilization of hybrid derivatives have contributed to the prolif- eration of species of Erythrina. The data presented in this paper have been marshalled in support of this set of hypotheses. The cytological studies (Section 3) and the exper- imental hybridization and self-compatibility trials (Section 4) present evidence in support of the first two hypotheses. In spite of the considerable mor- phological, ecological, and geographic differentia- tion of Erythrina, the species have retained a high degree of chromosomal (structural and genic) ho- mology. Within sect. Erythrina, this homology, as evidenced by interspecific compatibility, is virtually complete: there is no detectable difference in the success of interspecific matings as compared with intraspecific matings. At greater taxonomic dis- tances between the two parents (intersectional and intersubgeneric matings), mating success declines to some extent, but the number of successful “wide hybridizations” obtained in the experimental trials indicates that even the most morphologically and eplogicnly ii i of í diploid REA FIR species th genic homology and qum not evolved substantial barriers to hybridization in concert with morphological dif- ferentiation. Erythrina forms a homogamic com- plex of interfertile species, or perhaps a series of homogamic complexes with weak to moderate bar- riers between the complexes. Erythrina shares this pattern of species relationships with many tem- perate-zone genera of trees and shrubs. The evi- dence from Erythrina suggests that the patterns of species relationships in predominantly or exclu- sively tropical groups of woody plants may not differ significantly from the patterns found in their better-known temperate-zone counterparts. For- mation of homogamic complexes may be a common phenomenon in tropical woody plants and may be an important factor in the evolution of these taxa. The patterns of tanos of phenetic traits in the artificially produce brids (Section 5) confirm the true hybrid nature of these plants and dem- onstrate that matricliny, a potential complicating factor in the inheritance of these traits and in the interpretation of hybridization patterns, is not in- dicated in Erythrina hybrids. The patterns of in- heritance in the artificial hybrids reveal the pat- terns to be expected in the detection and analysis of natural hybridization: for morphometric char- acters, a rather narrowly segregating array of in- termediate types among the hybrids; and for dis- crete characters such as trichomes, possessed exclusively by either the female or male parent, the inheritance of the character in some of the hybrid offspring, the character being often reduced in size or density Evidence for the third hypothesis, concerning the pollination of sect. Erythrina by relatively spe- cialized, widely foraging hummingbirds and the re- lation of this pollination system to Erythrina breed- ing systems, is presented in a separate paper (Neill, 1987). The pollination studies indicate that inter- specific pollen flow and potential natural hybrid- ization are likely to occur among sympatric species of sect. Erythrina. Evidence for the fourth and fifth hypotheses, concerning natural hybridization and hybrid spe- ciation in Erythrina, is presented in Section 6. Natural hybridization was detected among several co-occurring species of sect. Erythrina in Chiapas, Mexico, at the geographical center of diversity of the section. The natural hybrids display the same patterns of inheritance of phenetic traits as the artificial hybrids described earlier. The evidence for hybrid speciation itself is somewhat more equiv- ocal. As stated in the introduction to this paper, the final hypothesis is historical and cannot be tested directly, but can be inferred only by drawing on information obtained by testing the first four. The information presented throughout this pa- per does make oot the hypothesis of hybrid speciation in Erythrina. Moreover, the research reported here RUP a unique base of information for further studies of species relationships and the evolutionary history of Erythrina, as a model of evolutionary processes in flowering plants that may be common to many tropical woody genera. Per- haps the most incisive research that could be car- ried out at this point in the continuing biosystematic investigation of Erythrina would entail studies of isoenzymes and particularly of nucleic acid restric- tion sites among the taxa, as well as the inheritance of these molecular character states in the hybrids of known origin, followed by the construction of hylogenies combining molecular data with the presently available evidence on morphological and biogeographic patterns and the data from crossing experiments. The collection of Erythrina species and hybrids now available in cultivation at the 960 Annals of the Missouri Botanical Garden Hawaiian botanical gardens provides an ideal re- source for such studies, and it is my hope that my colleagues specializing in chemosystematics and molecular phylogenetics do take advantage of this resource to investigate further the patterns of evo- lution in this interesting genus. LiTERATURE CITED ADDISON, G. A. & R. Tavares. 1952. en €; grafting in species of eiat in which oc n Ama . Evolution 6: 380-386. raices, M. P. ; Differential staining of aborted and non-aborted pollen. Stain Technol. 14: ARROYO, M. T. K. 1981. Breeding systems and polli- nation biology in the Leguminosae. Pp. 723-770 in R. M. | & P. H. Raven (editors), Advances in egume Systematics. Royal Botanic Gardens, Kew. ASHTON, P. S. 1969. Speciation among tropical trees: some sole seg a in oa light of recent evidence. Biol. n. Soc. 1: -196. mete E. 1947. Studies in the Leguminosae. I. romosome numbers in Erythrina L. Amer. J. Bot. 4. 34: 407-4 AYENSU, E. S. 1977. Scanning electron microscopy of jin features in Erythrina (Fabaceae). Lloydia -4 43 Bias, P. 1978. luherttnes of foliar and nodal anatom- sia characters in some /lex hybrids. Bot. J. Linn. 1-52. Bane kunana T. 2. Wood structure of the us Erythrina. Allertonia 3: 53-69. BE Nas . G. W. 19 gene for sticky chromosomes in Zea agde Z. Indukt. Abstammungs-Vererbungsl. 195- BEEKS, R. M. “1985. Improvements in the squash tech- nique for plant o e 3: 131-134. BENTHAM, G. 1865. In: G. Bentham & J. D. Hooker, G me , B. Ba asic Color Terms, their Universality and Evolution. Univ. California Press, Berkeley. BREEDLOVE, D. E. Introduction to the Flora d vin M s. California Academy of Sciences, San Fra Y, R. H. RicHENS, F. SCHNE E. E Pes (editors) 1980. niinus Code of oe for Cultivated Plants— 1980. Regnum Veg CLAUSEN, J., D. P. yo & W. HieseY. 1939. The concept J Bap based on experiment. Amer. J. 106. Bot. 26: Buc KELL, n D., RÁ & 1940. Experimental studies on the nature of species. I. Effect of varied environments on western a American plants. Publ. Carnegie Inst. Wash. Coney, P. J. . Plate tectonic constraints on the biogeography of Middle America and the Car re Region. Ann. Missouri Bot. Gard. a hase . M. Tucker & F. S. SANTAMOUR, on Oak hybridization at the University of Uta State Arboretum of Utah Publ. No. Croat, T. B. 1983. A revision of the genus Anthurium (Araceae) of Mexico and Central America. Part I: Mexico and Middle America. Ann. Missouri Bot. Gard. 70: 211-420. CUTLER, D. F. 1972. Leaf anatomy of certain Aloe yi Gasteria p Fpa and their hybrids. Pp. 103-122 i A. K G e (e ditor), Research Trends in Plant w Delhi. 1977. Experimental evi- dence for the genetic control of leaf surface char- acters in hybrid Aloineae (Liliaceae). Kew Bull. 32: 23- DARLING TON, TJ D. LM Recent Advances in Cytology. "The distribution of self- d in oe plants. Proc. Amer. Phil. Soc 449- 518. EHRENDORFER, F. 1970. Evolutionary patterns and strategies in seed plants. Taxon 19: S c FEINSINGER, P., Y. B. LINHART, L. A. Sw WOLFE 1979. Aspects of the sa biology j^ three Erythrina ic Hj Trinidad and Tobago. n. Missouri Bot. Gard. 66: 451-471. non P. A. Mode of reproduction i in higher plants. Bot. Rev. (Lancaster) 25: 135-233. GANDERS, F. R. € K. M. Nacata. 1984. The role s hybridization i in the evolution of Bidens on the Haw ian Islands. Pp. 179-194 in W. Grant (editor), Plant o Academic Press, Toronto. GILLETT, G. W. & E. K. S. Lim. 1970. An experimental study of the genus Bidens i n the Hawaiian Islands. Univ. Calif. Publ. Bot. 56 3. GOLDBLATT, P. i Chrom mosome numbers in le- gume x = Ann. Missouri Bot. Gard. 68: 551-557. 81b. Index to plant chromosome numbers 1975- ioe Monogr. Syst. Bot. Missouri Bot. Gard. 5 1984. Index to plant chromosome numbers 1979- 1981. Monogr. Syst. Bot. Missouri Bot. Gard. 8. & G. Davipse. 1977. Chromosome numbers in legumes. Ann. Missouri Bot. Gard. 64: 121-128. & A. Gentry. 1979. Cytology of Bignoni- aceae. Bot. Not. 132: 475-482. GOTTLIEB, L. D. 1972. Levels of confidence in the analysis of hybridization in plants. Ann. Missouri Bot. ard. 59: 435-446. ——- 1984. Genetics and morphological evolution plants. Amer. Naturalist 123: 681-709. TERRA A. & A. S. Tomb. 1974. Palynology of Er- ythrina (Leguminosae: Papilionoideae): preliminary survey of the subgenera. Lloydia 37: 465-481. 1977. Palynology of Erythrina (Fabaceae: Kabnideass: the subgenera, sections, and generic relationships. Lloydia 40: 413-435 Grant, V. 1953. The role of hybridization in the evo- lution of the leafy-stemmed gilias. Evolution 7: 51- 64. 81. S Peine. 2nd edition. Columbia w Yor í servations of a Naturalist i in "à in Quercus alba. J. Arnold Arbor. 56: 336-363. Erythrina. In: H. G. A. Engler & . Drude, Die b cd der Erde 9(3): 656- ser Harvey, W. H. 1861. a aa Volume 2. Hodges, Smith & Co., Dublin Volume 75, Number 3 1988 Neill 961 Erythrina HERNÁNDEZ, H. M. 1982. Vrae sj in Erythrina montana. Allertonia 3: 72- & V. M. ToLepo. 1979. The role of nectar robbers and pollinators in the reproduction o E Ery- thrina leptorhiza. Ann. Missouri Bot. Gard. 66: 512-5 JACKSON, R. C. 1973. Chromosomal evolution in Hap- lopappus gracilis: a centric transposition race. Evo- lution 27: 243-256. Chromosome pairing in species . 1984. and hybrids. Pp. 67-86 in W. Grant (editor), Plant Bio- OSHOO. Cytogenetics of Erythrina E ecd Srivastava. Allertonia 3: 19-24. Keep, E. 1962. Interspecific hybridization in Ribes. Genetica 33: 1-2 jud KEMPANNA, C. & R. 1964. Secondary associ- Se ation between Mer equivalent bivalents. He- redity 19: 289-299. KRnukorr, B. A The American species of Er- thrina. Brittonia 3: 205-337. 1939b. Preliminary notes on Asiatic- Polyne- sian species of Erythrina. J. Arnold Arbor. 20: 225- 233. R. C. BanNEBY. 1973. Notes on the species of Erythrina. by Phytologia 27: 108-141. & 74. A conspectus of the genus Erythrina. Lies 37: 332-459. LACKEY, : A. 1981. Phaseoleae. Pp. 301-328 in R. M. Polhill & P. H. Raven DET pads in Legume Systematics. Royal Botanic Gardens, Kew. LAWRENCE, W. 193 he perunt. association of chromosomes. mae 2: 352-384 Lewis, W. H. 1974 mosomes and phylogeny of Erythrina decal pues a 37: 460-464. Louis, J. 193 evision des espéces congolaises du genre Erythrina L. Bull. Jard. Bot. État 13: 295- 31 McKey, D. pa Sa ecology of co-evolved seed dis- persal system . 159-191 in L. E. Gilbert & P. E Raven e Cosvelition of Animals and Plants. niv. nas Press, Austin. Mas J. A. . MaBRY. 1971. Alkaloids in the ped Pp. 73-178 in J. B. Harborne, D Butler & B. L. Turner nito Chemotaxonomy of the Leguminosae. Academic Press, London. MEHRA, P. 1976. Cytology of Himalayan Hardwoods. Sree Saraswaty Press, Calcutta. Mutcany, D. L. € G. B. MuLcany. Gameto- phytic self-incompatibility reexamined. Science 220: 1247-1251. MULLER, C. H. 1952. Ecological control of pie ie di in Quercus: a factor in the mechanism of evolutio Evolution 6: 147-161. NEILL, D. A Trapliners in the trees: humming bird pollination of Erythrina sect. Erythrina (Le guminosae: Papilionoideae). Ann. Missouri Bot. Gard. 74: 27-41. NETTANCOURT, D. DE. 1977. yo aga in Angio- sperms. Springer-Verlag, New Noss, 963. Experimental jum on species relationships in Ceanothus. Publ. Carnegie Inst. Wash. 623: 1-94. OREBAMJO, T. O., G. Porteous & G. R. STEWART. 1982. Nitrate reduction i in the genus Erythrina. Allertonia 11-18. 3: Pryor, L. D. 1959. Species distribution and association in Eucalyptus. Monogr. Biol. 8: -471. Raven, P. H. 1974. Erythrina (Fabaceae): achieve- ments and opportunities. Lloydia 37: 321-331. 77 Erythrina (Fabaceae: Faboideae): in- troduction to symposium II. Lloydia 40: 401-406. 19 ybridization and the nature of species in higher aaa Canadian Bot. Assoc. Bull. Suppl. Vol. 13: 3-10. — € D. I. AxELROD. 1974. psi lay bio- geo ography a and past continental movements. Missouri Bot. Gard. 61: 539-673. . E. RAVEN. 1976. The genus Epilobium in Australasia: a systematic and ge qe 16: nod Zealand Dept. Sci. Industr. Res. Bull. 2 Ru, H N. 1930. The peu id “Pam Do e & Co., ord, Ken - CHAPMAN. "1958. be ep da of cytologically PE behavior of hexaploid wheat. Nature 182: 713-7 SEIBERT, R. J. 1947. À Es of Hevea (with its eco nomic aspects) in the republic of Peru. Ann. Mialouri Bot. Gard. 34: 261-353. SIMPSON, : B. (editor). 1977. Mesquite: its biology in two desert ecosystems. ne Hutchison & Ross, Siria Pennsylva SkuTCH, A. 19 A Naturalis in Costa Rica. Univ. Florida Press, Gekker Sokar, R. R. & F. J. > ipte 1969. Biometry. W. H. Freeman & Co., San Francisco. SrACE, C. A. 1984. The taxonomic importance of the leaf surface. Pp. 67-94 in V. H. Heywo od ias (editors), di Concepts in Plant Tax- onomy. demic Press, London ST ATISTICAL is ysis INSTITUTE. 1982. SAS e: Statistics. Statistical Analysis Institute, Cary, North Carolina. SrEBBINS, G. L., JR. 1958. The inviability, weakness and sterility of interspecific hybrids. Advances Genet. 9: 14 6. STEPHENSON, À. G. 1981. Flower and fruit abortion: proximate causes x ultimate functions. Ann. Rev. Ecol. Syst. 12: 253-280. SUNDAR = Y. 1945. Chromosomes of s^ ig indica Lamk. J. Indian Bot. Soc. 24 VAN VALEN, L. 1976. mre species, aid PR and oaks. Taxon 25: 233-239. VEREATASU IBBAN, K. R. 1044. Cytological studies in AS Annamalai Univ., Annamalaingar, In- dia Erythrina montana Standley . Erythrina Gómez 962 Annals of the Missouri Botanical Garden APPENDIX I. Species, sections, and subgenera of APPENDIX I. Continued. Erythrina. All recognized taxa in Erythrina are included in this list. | do not recognize Mie din d taxa in 20. Erythrina leptorhiza A. DC. Erythrina. Proposed taxonomic changes p 21. Erythrina horrida A. DC. here, prior to their formal designation. The numberin 21a. Ethene sousae Krukoff sequence of Krukoff & Barneby (1974) is followed for 12. Sect. Erythrina reference to that work, and because the numbers were 2 Erythrina herbacea L. used to designate the hybrids. There are gaps in the 23. Erythrina standleyana Krukoff number sequence because of reduction of species to 24. Erythrina flabelliformis Kearney synonymy. Species reduced to synonymy since Krukoff 25. Erythrina americana Miller & Barneby (1974) are indicated at the end of this 27. Erythrina pudica Krukoff & Bar- list. neby 27a. Erythrina krukoviana Neill, sp. Erythrina L. nov. ined. L Subgenus Micropteryx (Walp.) F. G. Baker 28. Erythrina lanata Rose 1. Sect. Duchassaingia (Walp.) Krukoff 29. Erythrina goldmanii Standley l. Erythrina fusca Lour. 31. Erythrina folkersii Krukoff € 2. Sect. Cristae-galli Krukoff Mold. Erythrina crista-galli L. 32. Erythrina tuxtlana Krukoff & Bar- 3. Erythrina falcata Benth. : 33. Erythrina smithiana Krukoff 3. Sect. Micropteryx š . : u 34. Erythrina cochleata Standley 4. Erythrina dominguezii Hassler : . i . 35. Erythrina hondurensis Standley 5. Erythrina ulei Harm i 36. Erythrina chiapasana Krukoff 6. Erythrina verna Vel ! : 37. Erythrina atitlanensis Krukoff & 7. Erythrina poeppigiana na (Walp. ) O. F. Cook Barneb 38. Erythrina cobanensis Krukoff & II. Subgenus Erythrina Barneby 4. Sect. Suberosae Krukoff 39. on williamsii Krukoff & 8. Erythrina suberosa Roxb E n s Krukof & 9. Erythrina microcarpa Koord. & 49. laa faJuemuleensis Kruk Valeton ic D. 10. Erythrina stricta Roxb. 41. Erythrina chiriquensis Krukoff ; 42. Erythrina macrophylla A. DC. ll. Erythrina resupinata Roxb. ' 43. Erythrina guatemalensis Krukoff 5. Sect. Arborescentes Krukoff : 44. Erythrina globocalyx Porsch € 12. Erythrina arborescens (Roxb.) Cuf. Walp. 45. Erythrina steyermarkii Krukoff & 6. are M as aay (Hassk.) Krukoff arneby Erythrina subumbrans (Hassk.) 46. Erythrina florenciae Krukoff & Merr Barneby 7. Sect. Breviflorae Krukoff 47. Erythrina berenices Krukoff & Bar- l4. Erythrina breviflora A. DC. by l4a. Erythrina petraea Brandegee 48. Erythrina huehuetenangensis Kru- 14b. Erythrina oaxacana (Krukoff) Kru- koff & by koff 49. Erythrina lanceolata Standley l4c. Erythrina batolobium Barneby & 50. Erythrina costaricensis M. Micheli Krukoff 51. Erythrina barqueroana Krukoff & 8. Sect. Edules Krukoff Barneb 15. Erythrina edulis Triana ex M. 53. Erythrina berteroana Urban Micheli 54. Erythrina rubrinervia H.B.K. l5a. Erythrina megistophylla Diels 55. Erythrina mexicana Krukoff 9. Sect. Stenotropis (Hassk.) Krukoff 56. Erythrina salviiflora Krukoff & 16. Erythrina speciosa Andrews T — S ritar 10: Seah Pieudocdüles Krukeff & Brady a. Erythrina santamartensis Kruko 17. Eryth lychaeta H oo Ine porycnastn Harma 57. Erythrina castillejiflora Krukoff & 18. Erythrina schimpfii Diels para ll. Sect. Leptorhizae Krukoff 57a thyrsiflora Gómez & Volume 75, Number 3 Neill 963 1988 Erythrina APPENDIX I. Continued. APPENDIX I. Continued. 13. Sect. Gibbosae Krukoff & Barneby 84. Erythrina baumii Harms 58. Erythrina gibbosa Cuf. 85. Erythrina decora Harms 14. Sect. Corallodendra off 86. Erythrina livingstoniana Baker . Erythrina amazonica Krukoff 87. Erythrina tholloniana Hua 60. Erythrina similis Krukoff 88. Erythrina addisoniae Hutchinson & Dalziel 61. Erythrina peruviana Krukoff 62. Erythrina mitis Jacq. 89. Erythrina droogmansiana DeWild. 63. Erythrina pallida Britton & Rose & T. Durand 64. Erythrina corallodendrum L 65. Erythrina eggersii Krukoff 91. Erythrina sacleuxii Hua 66. Erythrina buchii Urban 92. Erythrina haerdii Verdc. 67. Erythrina leptopoda Urban & Ek- 93. Erythrina sigmoidea Hua 94. Erythrina latissima E. Meyer 95. Erythrina abyssinica Lam. 90. Erythrina orophila Ghesq. [e man 14a. Sect. Fidelenses Neill, sect. nov. ined. 68. Erythrina elenae Howard & Briggs V. Subgenus Erythraster Barneby & Krukoff 15. Sect. Cubenses Krukoff 26. Sect. Erythraster 9. Erythrina cubensis C. Wright 96. Erythrina variegata L. 16. Sect. Olivianae Krukoff & Barneby 97. Erythrina tahitensis Nad. 97a. Erythrina sandwicensis Degener 98. Erythrina euodiphylla Hassk. 99. Erythrina vespertilio Benth. 100. Erythrina merrilliana Krukoff 101. Erythrina velutina Willd. 70. Erythrina oliviae Krukoff 17. Sect. Caffrae Barneby & Krukoff 71. Erythrina caffra Thunb. 72. Erythrina lysistemon Hutchinson 18. Seat. Humeange Barneby E Erukoff 103. Erythrina grisebachii Urban T3. Beythreng icono Sprengel 104. Erythrina burtii Baker f. TA. Reyne yayhari Harvey 105. Erythrina burana R. Chiovenda 19. Sect. Acanthocarpae Barneby & Krukoff 106. Erythrina perrieri R. Viguier 75. Erythrina acanthocarpa E. Meyer 107. Erythrina schliebenii Harms ex HI. Subgenus Tripterolobus Barneby & Krukoff Mildbr. 20. Sect. Tripterolobus Barneby & Krukoff 108. Erythrina melanacantha Taubert 76. Erythrina greenwayi Verdcourt ex Harms IV. Subgenus Chirocalyx (Meisner) Harvey s reduced to synonymy since Krukoff & Barneby 2 Sect. Bruceanae Barneby & Krukoff Erythrina caribaea Krukoff & Barne 77. Erythrina brucei Schwein. = E. berteroana Urban x E. Fees Krukoff & 22. Sect. Macrocymbium (Walp.) Barneby & Mold. Krukoff Erythrina coralloides A. DC. — E. americana Miller 78. Erythrina vogelii Hooker f. Erythrina insularis F. M. Bailey = E. vespertilio 79. Erythrina me A. DC. Benth. The first two reductions to synonymy are proposed for 23 $55 Dilobochilus Har the first time in this paper. The third reduction follows Erythrina Peri Baker Krukoff's treatment in his post-1974 publications on Er- 24. ES Dichilocraspedon Harms ythrina. 81. Erythrina mildbraedii Harms 25 Sect. Chirocalyx 82. Erythrina pygmaea Torre 83. Erythrina mendesii Torre 964 Annals of the Missouri Botanical Garden APPENDIX II. bridization trials for each species combination. For many of the species combinations, more than one ac Erythrina hybridization trials. This appendix summarizes the results of the interspecific hy- cession was employed as the male and/or female parents. The identity of the individual parents is presented in Tables 11—13 only for the successful trials resulting in viable hybrid plants. The hybridization trials are grouped into five categories: I. Narrow hybridizations within sect. Erythrina. I. Narrow hybridizations, excluding sect. Erythrina. III. /ntersectional hybridizations: female parent in sect. Erythrina. IV. Intersectional hybridizations: male parent in sect. Erythrina. V. Intersectional hybridizations: excluding sect. Erythrina. In categories III-V, the number in parentheses after the species name refers to the section to which the species belongs (see Table 1). An asterisk indicates a wide (intersubgeneric) hybridization. Hybrid # — a number assigned to each hybrid parental species as listed in Krukoff & Barneby (1974 Pol — number of flowers hand-pollinated in the hybrid combination Frt — number of pollinations producing mature fruits Sds = total number of normal-sized seeds produced in the hybrid combination er — number of seeds that germinated Liv — surviving progeny; number of live F, plants in the hybrid combination a combination, made up from the numbers assigned to the Female Parent Male Parent Hybrid # Pol Frt Sds Ger Liv I. Narrow hybridizations within sect. Erythrina americana berteroana 25x53 3 l l 1 1 americana herbacea 25x22 7 0 0 0 0 atitlanensis berteroana 37x53 4 0 0 0 0 atitlanensis guatemalensis 37x43 1 0 0 0 0 berteroana chiapasa 36x53 4 0 0 0 0 berteroana folkersii 53x31 14 0 0 0 0 berteroana uatemalensis 53x43 21 1 8 6 6 berteroana rubrinervia 53x54 3 0 0 0 0 berteroana salvüflora 53x56 3 0 0 0 0 berteroana standleyana 53x23 2 0 0 0 0 berteroana tajumulcensis 53x40 4 0 0 0 0 chiapasan berteroana 36x53 7 1 12 8 8 chiapasana folkers 36x31 1 0 0 0 0 chiapasana s us 36x43 16 0 0 0 0 chiapasana macrophylla 36x42 2 0 0 0 0 chiapasana tajumulcensis 36x40 3 0 0 0 0 costaricensis berteroana 50x53 12 0 0 0 0 folkersii berteroana 31x53 22 0 0 0 0 folkersü guatemalensis 31x43 9 0 0 0 0 goldmanii chiapasa 20x53 12 1 1 1 1 guatemalensis berteroana 43x53 13 4 25 14 8 guatemalensis chiapasana 43x36 10 2 15 9 9 atemalensis folkersii 43x31 4 1 5 5 2 guatemalensis erbacea 43x22 12 0 0 0 0 guatemalensis rophylla 43x42 7 4 17 12 8 guatemalensis salviiflora 43x56 2 l T l 1 guatemalensis standleyana 43x23 15 3 7 3 2 guatemalensis tajumulcensis 43x 40 23 6 17 9 9 herbacea eric 22x25 2 2 4 3 3 herbacea berteroana 22x53 14 2 3 1 1 herbacea chiapasa 22x36 1 0 0 0 0 herbacea guatemalensis 22x43 4 1 2 2 2 herbacea standleyana 22x23 8 0 0 0 0 herbacea tajumulcensis 22x40 2 0 0 0 0 macrophyl americana 22x25 4 1 3 2 1 macrophylla atitlanensis 42x37 1 0 0 0 0 macrophylla berteroana 2x53 4 2 8 4 4 Volume 75, Number 3 Neill 965 1988 Erythrina APPENDIX II. Continued. Female Parent Male Parent Hybrid # Pol Frt Sds Ger Liv macrophylla chiapasana 42x36 6 1 3 0 0 macrophylla folkersii 42x31 7 1 2 1 1 macrophylla guatemalensis 42x43 17 2 6 4 4 macrophylla herbacea 42x22 8 0 0 0 0 macrophylla salviiflora 42x56 8 0 0 0 0 macrophylla standleyana 42x23 2 0 0 0 0 macrophylla tajumulcensis 42x40 9 1 1 0 0 rubrinervia berteroana 54x53 2 0 0 0 0 salviiflora berteroana 56x53 19 0 0 0 0 salviiflora guatemalensis 56x43 3 0 0 0 0 standleyana berteroana 23x53 19 0 0 0 0 tandleyan guatemalensis 23x43 8 0 0 0 0 standleyana herbacea 23x22 10 0 0 0 0 tajumulcensis berteroana 40x53 12 0 0 0 0 tajumulcensis guatemalensis 40 x 43 8 1 4 4 4 tajumulcensis bacea 40 x 22 2 0 0 0 0 tajumulcensis macrophylla 40x42 5 0 0 0 0 I. Total 417 39 142 90 75 II. Narrow (intrasectional) hybridizations: excluding sect. Erythrina 1. Sect. Cristae-galli crista-galli falcata 2x3 22 2 4 1 1 falcata crista-galli 3x2 14 4 7 0 0 2. Sect. Chirocalyx abyssinica latissima 95x94 3 1 2 1 1 abyssinica sacleuxii 95x91 2 0 0 0 0 3. Sect. Erythraster perrieri variegata 106 x 96 10 + 12 10 9 sandwicensis variegata 97a x 96 8 0 0 0 0 tahitensis sandwicensis 97x97a 7 0 0 0 0 tahitensis variegata 97x96 26 0 0 0 0 tahitensis velutina 97x 102 19 0 0 0 0 variegata rri 96x106 6 0 0 0 0 variegata vespertilio 96 x 99 6 0 0 0 0 II. Total 123 11 25 12 11 III. Intersectional hybridizations: female parent in sect. Erythrina berteroana* fusca (1) 53x1 21 0 0 0 0 chiapasana* sca 36x1 5 0 0 0 0 folkersii* fusca 31x1 8 0 0 0 0 guatemalensis* fusca 43x1 35 0 0 0 0 herbacea* fusca 22x1 10 1 4 4 4 macrophylla* fusca 42x1 10 1 3 0 0 berteroana* crista-galli (2) 53x2 2 0 0 0 0 guatemalensis* crista-galli 43x2 64 1 2 0 0 herbacea* crista-galli 22x2 10 0 0 0 0 macrophylla* crista-galli 42x2 20 0 0 0 0 rbacea* dominguezii (3) 22x4 3 0 0 0 0 guate nsis stricta (4) 43x10 9 0 0 0 0 guatemalensis arborescens (5) 43x12 3 0 0 0 0 guatemalensis speciosa (9) 43x16 10 0 0 0 0 herbacea | 22x16 2 0 0 0 0 macrophylla 42x16 11 0 0 0 0 berteroana pallida (14) 53x63 4 0 0 0 0 966 Annals of the Missouri Botanical Garden APPENDIX II. Continued. Female Parent Male Parent Hybrid # Pol Frt Sds Ger Liv guatemalensis amazonica (14) 43x59 58 1 1 0 0 guatemalensis corallodendrum (14) 43x64a 2 2 6 0 0 guatemalensis caffra (17) 43x71 1 0 0 0 0 ME lysistemon (17) 43x 72 23 2 3 1 0 herbac caffra 22x71 2 0 0 0 0 guatemalensis humeana (18) 43x73 5 l l 0 0 herbac humeana 22x73 5 1 2 2 2 oui senegalensis (22) 43x 79 8 3 3 1 l guatemalensis* abyssinica (25) 43x95 9 2 2 1 1 guatemalensis* latissima (25) 43x94 6 0 0 0 0 macrophylla* abyssinica 42x95 2 0 0 0 0 macrophylla* latissima 42x94 4 0 0 0 0 guatemalensis* perrieri (26) 43x106 12 0 0 0 0 guatemalensis* sandwicensis (26) 43x097a 25 0 0 0 0 guatemalensis* variegata (26) 43 x 96 26 0 0 0 0 guatemalensis* vespertilio (26) 43 x 99 4 1 2 0 0 herbacea* perrieri 22x106 8 0 0 0 0 herbacea* variegata 22x96 8 0 0 0 0 macrophylla* sandwicensis 42x97a 4 0 0 0 0 macrophylla* variegata 42 x 96 8 0 0 0 0 macrophylla* vespertilio 42x99 1 0 0 0 0 HI. Total 444 16 29 9 8 IV. Intersectional hybridizations: male parent in sect. Erythrina fusca (1)* berteroana 1x53 70 0 0 0 0 fusca* folkersii 1x31 8 0 0 0 0 usca* guatemalensis 1x43 11 0 0 0 0 crista-galli (2)* guatemalensis 2x43 10 4 15 10 8 stricta (4) guatemalensis 10x43 6 0 0 0 0 arborescens (5) guatemalensis 12x43 24 1 1 0 0 speciosa (9) berteroana 16x53 3 0 0 0 0 corallodendrum (14) berteroana 64x53 3 0 0 0 0 pallida (14) berteroana 63x53 4 0 0 0 0 pallida (14) fusca 63x1 2 0 0 0 0 Mimeana i) berteroana 73x53 3 0 0 0 0 )* guatemalensis 95 x 43 3 0 0 0 0 tal (26)* berteroana 106 x53 2 0 0 0 0 perrieri (26)* guatemalensis 106 x 43 8 0 0 0 0 variegata (26)* guatemalensis 96 x 43 7 0 0 0 0 variegata (26)* herbacea 96 x 22 8 0 0 0 0 IV. Total 176 5 16 10 8 V. Intersectional hybridizations: excluding sect. Erythrina fusca (1) crista-galli (2) 1x2 38 0 0 0 0 fusca* lysistemon (17) 1x72 1 0 0 0 0 fusca* variegata (26) 1x96 17 0 0 0 0 crista-galli (2) fusca (1) 2x1 35 11 33 8 7 a-galli dominguezii (3) 2x4 34 0 0 0 0 crista-galli* arborescens (5) 2x12 1 0 0 0 0 crista-galli* speciosa (9) 2x16 14 2 2 2 1 crista-galli* amazonica (14) 2x59 18 1 1 0 0 crista-galli* abyssinica (25) 2x95 3 0 0 0 0 crista-galli* perrieri (26) 2x106 8 0 0 0 0 crista-galli* sandwicensis (26) 2x97a 5 2 3 0 0 crista-galli* variegata (26) 2x96 23 2 3 2 1 dominguezii (3) crista-galli (2) 4x2 8 0 0 0 0 Volume 75, Number 3 Neill 967 1988 Erythrina APPENDIX II. Continued. Female Parent Male Parent Hybrid # Pol Frt Sds Ger Liv arborescens (5)* crista-galli (2) 12x2 15 1 1 0 0 arborescens umeana (18) 12x 73 7 0 0 0 0 arborescens* sandwicensis (26) 12x97a 15 0 0 0 0 speciosa (9)* fusca (1) 16x1 3 0 0 0 0 speciosa* crista-galli (2) 16x2 33 0 0 0 0 speciosa lysistemon (17) 16x72 64 3 v 4 4 caffra (17)* usca (1) 71x1 5 1 2 2 2 lysistemon (17)* fusca (1) 72x1 10 0 0 0 0 lysistemon speciosa (9) 72x16 46 3 8 4 3 lysistemon abyssinica (25) 72x95 11 0 0 0 0 lysistemon* latissima (25) 12x94 4 0 0 0 0 senegalensis (22)* fusca (1) 79x1 8 0 0 0 0 abyssinica (25)* fusca (1) 95 x1 5 0 0 0 0 abyssinica crista-galli (2) 95x2 2 0 0 0 0 abyssinica* humeana (18) 95x 73 5 1 1 0 0 abyssinica* sandwicensis (26) 95 x97a 1 0 0 0 0 abyssinica* variegata (26) 95 x 96 1 0 0 0 0 latissima (25)* lysistemon (17) 94x 72 16 0 0 0 0 sima* humeana (18) 95x 73 5 0 0 0 0 perrieri (26)* fusca (1) 106x1 3 0 0 0 0 variegata (26)* usca (1) 96x1 32 0 0 0 0 variegata crista-galli (2) 96 x 2 2 0 0 0 0 variegata speciosa 96x16 6 0 0 0 0 variegata* senegalensis (22) 96x 79 7 0 0 0 0 V. Total 511 21 61 22 18 968 Annals of the Missouri Botanical Garden APPENDIX III. Sources of cultivated Erythrina used as parentals in successful interspecific hybridizations. (W): Accession obtained from known wild populations; (NW): Accession obtained from cultivated source, or otherwise not from a known wild population. Vouchers (from plants cultivated in Hawaiian id are deposited at MO. Location of voucher from original wild collection of seed is indicated here if know Erythrina abyssinica Lam. PT 770034001 Kenya: Nairobi. E. Taylor 17. (NW) jn abyssinica Lam. PT 731006002 Kenya: Nairobi, cultivated tree in yard of Cunningham van Someren. (NW) CMM americana Miler WA 75c1171 exico: Mexico City, o tree. L. S. Ayres s.n. (Waimea received as cutting from Los Angeles State & County peii # 565874) d p nin berteroana Urban PT 7 1001 Guatemala: Suchitepequez. o D El al B. A. Krukoff 1973-13 (NY). (W) pees berteroana Urban PT 700044001, -0 Panama: Canal Zone. Tree cultivated at Summit nim W. S. Stewart s.n. (NW) Erythrina berteroana Urban WA 74s864 Guatemala: Suchitepequez. jou m B. A. Krukoff 1968-508 (NY). (W) Erythrina berteroana Urban WA 78s Panama: Canal Zone. Between Madden Dam and Chilibre. J. Folson 3661 (MO). (W) Erythrina caffra Thunb. WA 74c1456 South Africa: Cape Province, Grahamstown, elev. 2,400 ft. Roy Bayliss s.n. (Waimea received as cutting from Foster Garden # 69.265) ( i Ain chiapasana Krukoff PT 721005001 Guatemala: Huehuetenango, near La aaa B. A. Krukoff 1969-68 (NY). (W) “amanta chiapasana Krukoff PT 73071000 Guatemala: Huehuetenango, near La im z A. Krukoff 1973-16 (NY). (W) Erythrina crista- dn n L. PT 74028300 Paraguay: near Asunción. Conrad & ie 2191. (W) Erythrina crista- aes L. WA 74p840 South Africa. Cultivated tree; seed received from Wm. J. Tijmens, Univ. of Stellenbosch. (Waimea received as live ven from PT 725352) (NW) Erythrina falcata Benth. PT 750086001 Argentina. Thays Botanical Garden, cultivated tree. E. Pingitore s.n. (NW) Erythrina folkersii Krukoff & Mold. PT 700010001 Guatemala: Izabal, at junction of road to Puerto Barrios and Mathias Calves. B. 4. Krukoff 1969-109 (NY). (W) = Erythrina fusca Lour. PT 740230005, WA 74599 Guatemala: Escuintla. B. A. Krukoff 1972-12 (NY). fe i Erythrina guatemalensis PT 700018001, WA 74c14 Guatemala: Alta Verapaz, along Cobán-Salama ia near Santa Cruz, elev. 1,280 m. B. A. Krukoff 1969-195 (NY). (W) Note: The tree at Waimea Arboretum WA 74c1453 was grown from a cutting taken from PT 700018001, so the two accessions are genetically identical. Erythrina guatemalensis Krukoff PT 720999002 Guatemala: Huehuetenango, near Barillas. B. 4. dd 1969-200 (NY). (W) Erythrina Vp db qnd Krukoff PT 75041900 Guatemala: Huehue kp near Barillas. B. A. Krakoff s.n. (W) Erythrina herbacea PT 7 5 California: Los Angeles E s Coun Arboretum # 5451201, cultivated. (NW) Erythrina herbacea L. WA 7 7 Florida: Miami, Fairchild s cultivated. (NW) Erythrina hum Sprengel WA 74p1382 South Africa: oca b cg D. Millington s.n. (NW) Erythrina in ima E. Meyer PT 750281004 South Africa: Natal. Cultivated tree at e Town Gardens. lan Whitton 750401. (NW) Erythrina isistemon Hutchinson PT 750280002, -003 South Africa: Natal, Durban. Jan Whitten s.n. (1975) Na Erythrina macrophylla A. DC. PT 750420002, WA 75s Guatemala: Sololá, near Godinez. Elev. 6,145 ft. B. A. Krukoff 1975-4 (NY). (W) Volume 75, Number 3 Neill 969 1988 Erythrina APPENDIX III. Continued. gie die R. Viguier WA 74s857 Madagascar: Maintirano, near Bekopaka. Fred Meyer s.n. (W) Elan uos Krukoff & Barneby PT 721000002 Guatemala: iuf dolia Municipio Chicacao, Finca El Naranjo. Elev. 1,070 m. B. A. Krukoff 1969-58 (NY). (W) Erythrina senegalensis A. DC. WA 74s100 igeria: Coastal area. Seeds received from B. A. ideae rd unknown. (W) Erythrina speciosa Andrews PT 730708001, PT 73074 Brazil: Sao Paulo. Cultivated tree at Sao imd Botanical Um B. A. Krukoff 1973-20 (NY). (NW) 3 one standleyana Krukoff WA 76s10 Califor ` ed Cultivated tree. (NW) Originally collected as seed by Fred Meyer from wild tree, Yucatán, exico. Erythrina Miche ine Krukoff & Barneby WA 74c1448 Guatemala: San cos, near Aldea Feria, along road from San Marcos to San Rafael de La Costa. B. A. Krukoff 1969. 249 (NY). (Waimea received as cutting from PT 700015001) (W) prpthring variegata L. WA 74s892 Hawaii: Honolulu, Mid-Pacific Country Club, cultivated (white-flowered form). Beatrice Krauss s.n. (NW) Erythrina variegata L. WA 765996 Mariana Islands: Saipan Unai, Laulau Beach. Derral Herbst s.n. (W) THE BIOSYSTEMATICS OF LUDWIGIA SECT. MICROCARPIUM (ONAGRACEAE)! Ching-I Peng? ABSTRACT dwigia sect. Microcarpium is a polyploid complex of 14 species distributed primarily in the southeastern Lu United States. Several of the spec ies are variable logic diploid are vigorous but produce only 6% stainable pollen and at meiotic metaphase I, with up to three sometimes heteromorphic bi not been studied biosystematically in any combination, prod one of which is identical with that of L. mi L. linearis- L. linifolia lineage was involved in and taxonomically difficu ybridization between a tetraploid mbers are present in the L. curtissii complex . The hexaploid jo died of L. simpsonii appears to include three different diploid genomes, ocarpa. Present data are not sufficient to determine if the diploid arentage of the hexaploid L. simpsonii, but morphological evidence L. simpsonii, n — 24; suggests it was not. It is highly probable, ess 5 un the octoploid L. curtissii was derived after hybridization between the diploid L. linearis-L. linifolia lineage and the hexaploid L. simpson Ludwigia contains some 82 species, which are classified into 23 sections (Raven, 1963; Rama- moorthy, 1979; Ramamoorthy & Zardini, 1987). It is the only member of the monotypic tribe Jus- siaeeae and is the fourth largest genus in Onagra- ceae (after Epilobium, Oenothera, and Fuchsia). It has existed since at least the Eocene (Eyde & Morgan, 1973), and increasing evidence suggests it represents the earliest surviving evolutionary off- shoot of the family (review in Ramamoorthy & Zardini, 1987; Eyde, 1981), or, in cladistic terms, the “sister group” of all other genera of the family. Study of this genus is therefore critical to under- standing the overall evolutionary pattern in Ona- graceae, a subject to which Peter H. Raven and his associates have devoted much time and energy over the past 30 years. Thirteen species of Ludwigia are restricted to the Old World, 59 to the New World, and 10 are common to both hemispheres (Ramamoorthy, ' This study was part of my doctoral studies at Washington University, St. Louis, Missouri. I am indebted to Peter H. Raven for suggesting and supporting this project an 85 18906 to Peter H. Raven. I tha nk Ge for valuable suggestions, comments, and criticism Microcarpium; Steven rad Warren Vi pr aud Hir Popenoe, who supplied living material x Ludw d to the Bi wigia; Bruce Carr, who helped with the cytological analysis; James Henrich and Emily C ee for aid in the greenhouse; Ji-Yui Guo for aid in preparing ab John Myers for preparing the maps; Glo a Hoch who typed the manuscript; and, finally, my wife, anling, for assisting in the field, preparing ine Gain s, and for supporting my work, without which my és ons could not have been completed. ? [ns itute of Botany, Academia Sinica, Nankang, Taipei, Taiwan 11529, Republic of China. ANN. Missouni Bor. Garp. 75: 970-1009. 1988. Volume 75, Number 3 1988 Peng Ludwigia sect. Microcarpium 1980). Most sections are tropical and subtropical, but Ludwigia also has several well-developed tem- perate offshoots, particularly in North America (Raven & Tai, 1979 Most native North American Ludwigia species have four sepals, four stamens, pluriseriate and free seeds with narrow raphes, and a herbaceous habit (Raven, 1963). With a few exceptions they are confined to the Atlantic and Gulf coastal plains of the United States. The 23 haplostemonous species represented here are placed in three sections: sect. Ludwigia, with 4 species (Munz, 1944, 1965; Raven, 1963); sect. Dantia, with 5 species (Schmidt, 1967); and sect. Microcarpium, a di- verse group of 14 species (Table 1), which con- stitutes the subject of this study. ection Microcarpium was selected for the pres- ent study because it represents a diverse polyploid complex (Raven & Tai, 1979) in which abundant natural hybridization occurs and in which the re- lationships among species are not clear. In a com- parative study of the reproductive structures in Ludwigia, Eyde (1978) considered sect. Micro- carpium as “remarkably diverse," at least with respect to surface-cell orientation of the seeds. The fact that several polyploid species were reported to have two different chromosome numbers (Raven & Tai, 1979) may reflect either difficulties in iden- tification or their diverse genetic backgrounds. Duke (1955) noted the varying degree of intergradation exhibited between North Carolina Ludwigia species, and he suspected one population to be of hybrid origin between two species of sect. Microcarpium. A critical study of herbarium material has revealed many specimens exhibiting various intermediate morphological characteristics. Two such natural hybrids, L. lanceolata X L. pilosa and L. suffru- ticosa X L. pilosa, have even been given specific and varietal names (L. simulata Small and L. suf: fruticosa Walter var. pubens Torrey & A. Gray, respectively). Conclusive evidence of intergrada- tion among species, however, can only be obtained through experimental hybridization. Schmidt's (1967) work in establishing the phy- logenetic relationships among species belonging to Ludwigia sect. Dantia has proved helpful in the study of sect. Microcarpium. Both of these sections were grouped, along with the monotypic East Asian section Miquelia, in the ** Microcarpium complex" on the basis of reproductive structures and overall evolutionary patterns in Ludwigia (Eyde, 1981). Schmidt (1967) demonstrated through artificial hy- bridization and cytological observations of hybrid microsporocytes in sect. Dantia that hybrids be- tween plants of different ploidy levels consistently exhibited typical **Drosera-type" chromosome pairing; i.e., homologous genomes in the F, gen- erations always formed bivalents and nonhomolo- ous genomes remained unpaired. Schmidt (1967) also studied two naturally oc- curring intersectional hybrids. Meiotic observations of the hybrid between L. palustris (n — 8; sect. Dantia) and L. glandulosa (n = 16; sect. Micro- carpium) revealed very little association between chromosomes, with 0-3 weakly joined bivalents. The other natural hybrid examined, involving L. simpsonii (n — 24, cited as L. curtissii by Schmidt; sect. Microcarpium) and L. repens (n — 24; sect. Dantia), yielded 48 unpaired chromosomes at mei- osis. These observations suggest that in the related sect. Microcarpium, **Drosera-type" chromosome pairing also occurs. This characteristic is helpful in studying genetic relationships among species in such a polyploid complex. The purpose of the present study was to deter- mine the evolutionary relationships of the taxa comprising sect. Microcarpium. Toward this end I have considered extensive cytological evidence from my experimental hybridizations. This has been supplemented with data on morphology, pollination biology, crossing relationships, geographical distri- bution, and field observations of morphological variation and habitat preference. In addition, more than 7,000 herbarium specimens have been ex- amined. MATERIALS AND METHODS Specimens were collected from wild populations and were propagated from seeds or clonal trans- plants in the experimental greenhouse at the Mis- souri Botanical Garden (with the exception of some strains that were obtained as seeds from the Kew Seed Bank). Members of sect. Microcarpium can be cloned easily from their vegetative parts and cultivated in the greenhouse by standard proce- dures. Parental seeds usually germinated readily one or two weeks after they were sown. However, it took one or two months or even longer for some of the hybrid seeds to germinate. Plants of sect. Microcarpium normally flower the first year. In an insect-free greenhouse, several flowers of each parental strain were (1) artificially self-polli- nated to test for self-compatibility, (2) left alone to test for self-pollination, and (3) emasculated and left alone to test for apomixis. Plants of all species studied were found to be self-compatible and non- apomictic. Except for a few species, in which selfing is prevented physically, members of sect. Micro- carpium are capable of mechanical self-pollination. 972 Annals of the Missouri Botanical Garden All experimental crosses were therefore made by first emasculating the ovulate parent before self- pollination could occur, and then applying pollen from the pollen parent to its stigmas. Generally, several different parental strains were used in each cross; in a given trial all pollen parents were from the same population. The seeds resulting from suc- cessful hybridizations were sown early the following spring. Usually six seedlings of each artificial hybrid were randomly selected and transplanted into five- inch pots. A few of the F, families were also grown to maturity. About five to six months were required to produce a flowering hybrid after germination. Flower buds to be examined for meiotic behavior were fixed in a 3:1 mixture of 95% ethanol and glacial acetic acid and stored in the refrigerator. Prior to staining, the buds were hydrolyzed for 5- 8 minutes at 60°C using a 1:1 mixture of con- centrated HCl and 95% ethanol. They were then squashed in FLP orcein (Jackson, 1973). Somatic chromosome counts for some of the parental strains were obtained from actively growing root tips pre- treated for three to four hours in 8-hydroxyquin- oline, then fixed as above for at least ten minutes. The root tips were then hydrolyzed in 1 N HCI for 8-10 minutes at 60°C and squashed in the FLP orcein. Semipermanent slides were prepared and preserved in the freezer. Cytological observations were made using a Zeiss Universal Large Research Microscope. All analyzable chromosome configu- rations (mostly diakinesis or metaphase I) were documented with camera lucida drawings or pho- tomicrographs using Kodak Panatomic-X film. Negatives and drawings are deposited at the In- stitute of Botany, Academia Sinica, Taipei. Fertility of greenhouse parental individuals, ex- perimental hybrids, and suspected naturally oc- curring hybrids was estimated by determining the percentage of stainable pollen using the malachite green-acid fuchsin-orange G stain of Alexander (1969), which stains pollen walls green and cyto- plasm red. Pollen grains with uniformly red cyto- plasm were scored as fertile; partially stained or unstained grains were considered sterile. In many hybrids, however, and especially those of hetero- ploid crosses, the stainable pollen grains differ sub- stantially in size, and some probably are not func- tional, as was suggested by Uhl (1976). At least 200 tetrads (when pollen shed as tetrads) or 400 single grains (when pollen shed singly) per plant were scored. Seeds of members of sect. Microcarpium are so small (ca. 0.4-0.7 mm long) that studying the shapes and orientations of their surface cells under a dissecting microscope is very difficult. However, they are transparent enough to be examined under a light microscope. Photographs of seeds were tak- en with strong back illumination using Kodak Pan- atomic-X film. Polarized light was sometimes used. Herbarium specimens prepared from all exper- imental plants, both parental and hybrid (with the exception of the very small individuals such as those which did not develop much further than the cotyle- donous stage), are deposited at MO. Experimental hybrids are designated by a formula consisting of the acronym (Tables 3-15; Figs. 5, 21) for the two parents, connected by the multiplication sign (X), with the ovulate parent listed first. DIAGNOSTIC FEATURES Discussion of the morphological features is lim- ited to those useful in delimiting members of sect. Microcarpium and in recognizing hybrids. A more complete discussion of morphological variability in this group can be found in Peng (in press). Char- acters that are functionally related to the pollina- tion biology will be examined in more detail later in this paper. Some relevant diagnostic characters are summarized in Table 1. HABIT Plants of this section are all erect perennial herbs about 15-100 cm tall. They produce sprawling, leafy stolons along the surface of the ground in winter, although some species may also send out stolons in the summer while they are still flowering. In Ludwigia suffruticosa, however, underground rhizomes with scalelike leaves are also produced. FLOWERS The flowers have four sepals, four stamens, and a four-loculed ovary. Only L. linearis, L. linifolia, and L. stricta, all diploid, consistently have four yellow petals. The only other diploid species, L. microcarpa, and tetraploid, hexaploid, and octa- ploid species are apetalous. Vestigial petals are, however, occasionally present in normally apetal- ous species, especially L. curtissii. Loss of petals apparently represents the derived status, which is reflected in the fact that all polyploid species lack petals. Absence of petals does not in itself indicate autogamy, however, since nectary discs on top of the ovaries of all species produce various amounts of nectar that is fed on by insects. Furthermore, in the apetalous species— L. alata, L. suffruticosa, L. pilosa, and L. sphaerocarpa—the sepals are either cream-colored or yellow and are quite showy. One aspect of the floral morphology of particular Volume 75, Number 3 1988 Peng 973 Ludwigia sect. Microcarpium TABLE 1. Some characters of Ludwigia sect. Microcarpium. Petals Pollen Shed Present (+) Singly (S) or Seed Surface Chromosome or Absent as Tetrads ell Shape and Taxon Number 73 Orientation* L. alata Elliott n= 24 = S T L. curtissii Chapman n = 32 = S T L. glandulosa Walter subsp. glandulosa n — 16 Ee T P subsp. brachycarpa (Torrey & A. Gray) Peng n — 16 = 4 T L. lanceolata Elliott n= 16 s T I L. linearis Walter n=8 te T P or T L. linifolia Poiret ns 4 T I L. microcarpa Michaux n*8 e S T L. pilosa Walter a= 16 = T I L. polycarpa Short & Peter n — 16 = T P L. ravenii Peng n = 16 = T T L. simpsonii Chapman n = 24 = S T L. sphaerocarpa Elliott n= 16 = T P and T* L. stricta (Wright ex Griseb.) Wright n= + T I L. suffruticosa Walter n= 16 = S I * Letter codes I, P, T indicate seeds with surface cells more or less isodiametric, in columns parallel elongate to erve * Seed surface cells elongate parallel to the seed length are predominantly exhibited by the subglabrous populations, whereas cells elongate transversely to the seed length are prevalent in more strigillose populations. surface comprised of a mixture of cells both parallel and transversely elongate to the seed length. Sometimes the orientation of these columnar cells is irregular. interest in L. linearis is the presence in the anthers of transverse septa composed of tapetum and parenchyma which divide the sporogenous tissue into packets (Eyde, 1977; Tobe & Raven, 1986). This character is shared only by an unrelated South American species, L. latifolia (Benth.) Hara, and five other genera of Onagraceae. POLLEN Pollen morphology of Onagraceae has been stud- ied intensively by Ting (1966), Brown (1967), Skvarla et al. (1975, 1976, 1978), and Praglowski et al. (1983). Unique palynological features of the family include: protruding papillose apertures; mechanisms of tetrad and polyad cohesion; the fine structure of the exine; and viscin threads, which are extensions of the exine that tend to bind the grains together in masses. Pollen of sect. Micro- carpium is quite uniform, being characterized by isopolar grains frequently with prominent colpal extensions and with a psilate exine (Praglowski et al., 1983). Most species shed pollen in tetrads, although in L. alata, L. curtissii, L. microcarpa, L. simpsonii, and L. suffruticosa grains are shed singly (in monads), a characterisic found sporad- ically in other species of Ludwigia (Praglowski et al., 1983). Pollen shed as monads is thought to be the ancestral condition in Ludwigia (Praglowski et al, 1 , but in view of the relationships sug- gested by Eyde (1977, 1978, 1981), the monad pollen in sect. Microcarpium was probably derived secondarily from tetrad pollen. In any event, this character, along with the seed-surface cell pattern, is useful in distinguishing closely related species like L. alata (n = 24; monads) and L. lanceolata (n = 16; tetrads), which have different chromosome numbers but are otherwise difficult to distinguish. It is also highly helpful in detecting natural hybrids when their suspected parents differ in this char- acter. CAPSULES Fruit anatomy of Ludwigia has been studied by Eyde (1978), who reported that the fruit wall in L. alata is thickest in the placental radii, contrast- ing markedly with the fruit wall in sect. Dantia and in most other species of sect. Microcarpium. The shape, size, and vestiture of capsules are very 974 Annals of the Missouri Botanical Garden FicuRES 1-4. Photographs of seeds of some members of Ludwigia sect. Microcarpium. —1. L. lanceolata, Florida: Highlands Co., Peng 4183 (MO). — 2. L. glandulosa subsp. glandulosa, Florida: Santa Rosa Co., Dille 412 L. alata, Florida: Wakulla Co., Morar 11 (FSU).— 4. L. sphaerocarpa, South Carolina: Jasper Co., Dille 348 (MO). Scale bar = 0.4 mm diverse within sect. Microcarpium. Capsule shape ranges from obpyramidal to subcylindrical, oblong- obovate, turbinate, or subglobose, and length ranges from | to 12 mm. In L. alata and L. lanceolata, the capsules are narrowly to markedly four-winged. The surface vestiture ranges from glabrous to strig- illose or hirtellous. These characters are of para- mount importance in distinguishing the species and detecting hybrids. Furthermore, the lengths of per- sistent bracteoles at or near the capsule base is often a diagnostic character in species delimitation and hybrid recognition. SEEDS The seeds are small, 0.4—0.7 mm long, and are cylindric, ellipsoid, reniform, or ovoid in shape. Their surface cells are diverse in shape and ori- entation (Eyde, 1978). They are either more or less isodiametric (Fig. 1) or are in parallel columns that are predominantly either elongate parallel (Fig. 2) or transversely elongate (Fig. 3) to the seed length, with minor variation on the two ends and areas near the raphe (Table 1). Like capsule mor- phology, seed-surface pattern provides excellent diagnostic characters for identifying the species and detecting natural hybrids. In L. sphaerocarpa the seed-surface cells (Fig. 4) are less regularly oriented than in the other species. They are arranged in columns both trans- versely elongate and parallel to the seed length, with the former alignment often dominant in the central part of the seed. In this species, it is also not uncommon to have some seeds with variously oriented cells, a pattern that supports the sugges- tion of a hybrid origin for L. sphaerocarpa (see elow). REPRODUCTIVE BIOLOGY Raven (1979) thoroughly reviewed reproductive biology of Onagraceae. Two-thirds (56 of 82) of all species in Ludwigia modally self-pollinate, and of the 26 that modally outcross, most accomplish that by separation of the stigma and anthers (Ra- ven, 1979; Ramamoorthy & Zardini, 1987). There are no known instances in Ludwigia of protandry, protogyny, or male sterility, such as are found in other genera of Onagraceae (Raven, 1979). In nine species of Ludwigia, however, outcrossing is rein- forced by genetic self-incompatibility, which occurs in about a quarter of all outcrossing species in the family (Raven, 1979). ll members of sect. Microcarpium are genet- ically self-compatible perennials that are mostly facultatively autogamous. Shortly after the flowers open in the morning, the anthers dehisce and the stigma becomes receptive. The anthers spread and are held away from the stigma shortly after an- thesis, but in most species, petalous or apetalous, the anthers arch over a few hours later and attach firmly to the sticky stigma, thus effecting self- pollination, Self-pollination, as well as cross-pollination, how- ever, can also be achieved by insect vectors without having the anthers attached to the stigma me- chanically. Bumble bees, honeybees, wasps, moths, and ants were observed visiting populations of L. Volume 75, Number 3 1988 Peng 975 Ludwigia sect. Microcarpium pilosa in the field (Peng, 1984). Raven (pers. plentiful confirmed natural hybrids (see below) sug- gests substantial cross-pollination by insects in the e CYTOLOGY The first cytological study of a member of sect. Microcarpium was made by Gregory & Klein (1960) who, in the course of investigating the meiotic chromosomes of several onagraceous gen- era, recorded five counts for four species (cited as five species). These authors were the first to doc- ument polyploidy in the genus. One diploid pop- ulation of L. linifolia that they studied was sub- sequently examined mitotically by Kurabayashi et al. (1962), who called attention to the fact that the chromosomes of Ludwigia and those of tribe Epilobieae are the smallest in Onagraceae, and also that they may differ conspicuously in size within a single genome. The proximal ends of the chromo- some arms are heavily pycnotic and appear even in interphase nuclei as very distinct and definite chromocenters. Based on a review of the literature and the study of 302 individuals from 282 naturally occurring populations from throughout the range of the ge- nus, Raven & Tai (1979) presented a comprehen- sive chromosome number report for 38 of the 45 species of Ludwigia exclusive of sect. Myrtocar- pus sensu lato. The basic chromosome number for the genus was established as x = 8 with no aneu- ploidy but extensive polyploidy. In my description of a new species of sect. Microcarpium, L. ravenii (Peng, 1984), I reported 2n — 32 for this species, which has been misidentified as L. pilosa in the past. Ludwigia stricta, a Cuban endemic, is here reported as a diploid, with n — 8. Through these efforts, chromosome counts are now available for all taxa recognized in sect. Microcarpium. Section Microcarpium is shown to be a diverse polyploid complex, with four diploids, eight tetraploids, two hexaploids, and an octoploid. Ludwigia alata, L. curtisii, and L. suffruticosa, reported as having more than one chromosome number (Raven & Tai, 1979), will be discussed below. In the present paper, I am reporting 78 more counts representing 75 populations of 14 species and one additional subspecies in sect. Microcar- pium (Table 2). The chromosome number of L. stricta is here reported for the first time. These, along with 69 previously reported counts by Raven & Tai (1979), constitute our present knowledge of cytology of sect. Microcarpium. Because sev- eral taxonomic changes have been made, including a new combination (Peng, 1986), and because a new species, L. ravenii, has recently been recog- nized (Peng, 1984), I have checked the identifi- cation of voucher specimens cited by Raven & Tai (1979) and included these previously published counts in Table 2, using the currently accepted d inal counts reported by Raven & Tai (1979) are indicated by asterisks; those reported by others are accompanied by ref- erences Raven S Tai (1979), while reporting two chro- mosome numbers for L. alata, stated that **Chro- mosome counts are now available for all taxa except . L. stricta ... and, if it is distinct from L. ate L. lanceolata Ell.” Indeed, the last two species are similar in many aspects, especially in sharing obpyramidal capsules with winged corners, which is the key character in recognizing this species pair. However, upon examination of microscopic characters, such as seed surface architecture and the way pollen grains are shed, characters which previously have not been used by monographers (Munz, 1944, 1965), it has become clear that the two are distinct species. Ludwigia alata consis- tently sheds pollen singly (Praglowski et al., 1983) and has seed-surface cells in parallel columns trans- versely elongate to the seed length (Fig. 3), whereas L. lanceolata sheds pollen as tetrads (Praglowski et al., 1983) and invariably has nearly isodiametric seed-surface cells (Fig. 1). Their seeds differ in size and shape as well. Furthermore, these differences are correlated with chromosome number. Based on the 15 counts available (Table 2), L. alata is hexa- ploid with n = 24, whereas L. lanceolata is tet- raploid with n = 16. The only exception is a spec- imen collected from Collier Co., Florida (Raven 18672), which Raven & Tai (1979) correctly iden- tified as L. alata and reported to have n = 16. In order to verify this report, the same population was sampled again in 1980 (Peng 4242). Plants from this population yielded a count of n — 24. It seems likely, therefore, that the reported count of n — 16 resulted from confusion or interchange of samples. The questionable count has therefore been omitted from Table 2. or L. suffruticosa there are 11 chromosome counts available at present, ten with n — 16, an one with n — 24 for a collection (Raven 18651) from Hillsborough Co., Florida (Raven & Tai, 1979). Four other populational counts obtained from the same county consistently show n = 16. It is possible that plants with n = 24 have arisen 976 Annals of the Missouri Botanical Garden TABLE 2. Chromosome numbers in Ludwigia sect. TABLE 2. Continued. Microcarpium, with voucher information. Voucher spec- imens for original counts are at the Missouri Botanical Garden (MO); those of earlier reports, ao by an asterisk, are indicated in Raven & Tai (1979). Ludwigia pte Elliott (n = 24) RIDA: Collier Co., Peng 4242 (2n = 48), Peng 4267 n = 48); Franklin Co., Godfrey 70575, Peng ries di Co., Dille 392; Martin Co., Peng 4203 (n = 24; = 48); Wakulla = Raven 18608*. GEO- RGIA: din Co., nr. Cravens Hammock, Raven in 1974* (2n = 48). SOUTH CAROLINA: Horrey Co., Raven 8719”. Ludwigia Erie Sie n (n = 32) BAHAMA ISLANDS. Grand Bahama Island, Correll & Popenoe 51315. U.S.A. FLORIDA: Collier Co., Peng 4231, 4276, 4283; Dade Co., Godfrey 63396* (2n — 64) Franklin Co., Godfrey 71148; Hendry Co., Peng 4285, 4287; Hills- borough Co., Dille 435; Martin Co., Peng 4199; Monroe Co., Godfrey 63519* (2n = 64); Palm Beach Co., Po- penoe 1962; Sarasota Co., Raven 18662* (2n — 64). Ludwigia glandulosa Walter subsp. glandulosa (n — 16 U.S.A. ALABAMA: Macon Co., Raven 18562*. AR- KANSAS: . ip c (Raven Oe: 42, 2n = 32). FLORIDA: Jefferso , Raven 1 18617*; Leon Co., God- 1960); ul Co., Raven Pa ~ M 18576*, 1857 7". Paci : Jones Co., Raven 18569*. NORTH Columbus Co., Broome 865 pe TEXAS: Liberty Co., Raven EU Ludis glandulosa subsp. brachycarpa (Torrey & A. ps y) Peng (n = 16) S.A. LOUISIANA: nn Parish, Peng 4367. TEXAS: E Ben d Co., Raven 19398* (as L. glandulosa), Raven 19405* (as L. posee Ludwigia m Elliott (n = 16) S. PU A: Highlands 2 Dille 370 (2n = 32), Peng 418. 4193, Raven 18681* (as L. alata), Raven 18684* x alata), Raven 1 PEE. A Raven 17927, typographic error, 2n — 32, as L. alata). xc linearis Walter (n = 8) BAMA: Baldwin Co., Raven 18590*. AR- 46 65-43, 2n = 16). nl * GEORGIA: Emanuel 16). LOUISIANA: St. Tammany en 16579*. MISSISSIPPI: irr nd Co., Rav 8585*. NORTH CAROLINA: Cumberland Co., Lloyd 10369. ran Co., Lloyd 1121*. soUTH CARO- LINA: Horry Co., Raven 1 18721*, Jasper Co., Dille 350, Peng 3935 (2n = Ludwigia linifolia Poiret (n — EXICO. EA EREET I S Cowan 2632 (also 2n = 16), C U.S.A. FLORIDA: Franklin ud Peng 4343; borough Co., Dille 427; Okaloosa Co., Raven 18593*; Wakulla Co., Godfrey 77091 (2n = 16). MISSISSIPPI: Hood Co., Ruwan 18581*; Jackson Co., Demaree 37879 (n = 8. Pessoa & Klein, 1960; 2n = 16, Kurabayashi et al., 962). icy microcarpa Michaux (n = 8) "LORIDA: Charlotte Co., Peng 4 U.S.A. 4294; mag ve - Dille 359, Raven 18692*; Franklin Co., Peng Hillsborough Co., s n 18641* ; Jackson Co. M dne. 77093; Wakulla Co., Rav en 18601*, 18610*. NORTH CAROLINA: Jones Co., Pen i di in Ludwigia T eats Walter (n — 16) U.S.A. FLORIDA: Franklin Co., Peng 4345; Leon Co., Kral in 1963* (Raven 65-44, 2n = 32); Madison Co., Raven 18625*; Walton Co., [om 18594*. GEORGIA: Camden Co., Raven 18701*; Emanuel Co., Peng 4025 n = 32). MississiPPr: Hancock Co., Dille 419 (2n = 32), Raven 18580*; Jackson Co., Raven 18583*; Jones Co., Raven 18568*. SOUTH CAROLINA: Colleton Co., Ra- ven 18717*; Horry Co., Dille 342; Jasper Co., Raven 18712*. goce polycarpa Short & Peter (n — U.S.A. MASSACHUSETTS: Middlesex Co., eias 16514*. MICHIGAN: Washtenaw Co., Raven 16523*. MISSOURI: Franklin Co., Dille 328 (2n = 32), se 436 (2n = 32); Lincoln Co., Dille 443 (also 2n = 32). oy er ravenii Peng (n = 16) U.S.A. RIDA: Clay Co., Raven 18690* (as Raven 19690, la error; as L. pilosa). SOUTH CAROLINA: Berkeley Co., Peng 4402 (2n — 32, Peng, 1984). Ludwigia simpsonii Chapman (n = 24) U.S.A. FLORIDA: Charlotte Co., Peng 4293; Collier Co., Dille 378, Munz & Gregory 23476 (Gregory & Klein, ven 18649* (also 2n — : T Peng 4289; Martin Co., Munz & Gregory 23481 (Greg- ory & Klein, 1960); Sarasota Co., Dille 383, Peng 4313, Raven 18664* (as Raven 18640, typographic error). Ludwigia sphaerocarpa Elliott (n — 16) .S.A. Without definite locality, Monari beg - Klein, 1960). FLORIDA: Franklin Co., Dille 402; Mad- n Co., Raven 18626*, 18630*; Taylor Co. Raven 18620*. Wakulla Co., Dille 401, Peng 4339. INDIANA: Starke Co., Raven 16525*. MASSACHUSETTS: Plymouth Co., Raven 16516*. SOUTH CAROLINA: Jasper Co., Dille 348 (2n = 32). Ludwiga stricta (Wright ex Griseb.) t (n = 8y IBA. A. Leiva s.n. in 1982 (2n = 16). Ludwigia suffruticosa Walter (n -S.A. FLORIDA: Glades Co., Rave Hillsborough Co., Dille 423, 424 "434, Peng 4327; L Co., Raven 18637; Leon Co., Dille 421, Raven oe (as Raven 18585, typographic error); Polk Co., Lakela 24806* (Raven 19704, 2n = 32); Taylor Co., Raven 18619*. ) n 18678* (2 n= 32) * Chromosome number here determined for the first e directly from tetraploids by fusion of an d gamete from one parent with a normal gam from the other, as Raven & Tai (1979) postulated 1979), and pollen from the voucher specimen was fully viable Meiosis was normal (Raven & Tai, Volume 75, Number 3 1988 Peng 977 Ludwigia sect. Microcarpium as judged by staining results. The count of n — 24 should be reconfirmed and investigated further if additional individuals are located; it is omitted from Table 2 Ludwigia curtisii is another species Raven & Tai (1979) reported as having two chromosome numbers. They considered this species, along with L. simpsonii and L. spathulifolia, to comprise a single species complex, within which diagnostic characters such as capsule size and leaf shape (Munz, 1944, 1965) were not correlated with chro- mosome numbers. The present study, however, indicates that L. simpsonii is specifically distinct rom L. curtissii and that L. spathulifolia should be treated as a variant of L. curtissii with slightly larger capsules. In addition to being high polyploids (hexaploid and octoploid), plants of this complex are unique among species of sect. Microcarpium in having the capsules split along four longitudinal lines op- posite the loculi at maturity (Peng & Tobe, 1987). Capsules from other species in sect. Microcarpium dehisce either by separation of the walls from the indurate nectary disc (Munz, 1944, 1965; Raven, 1963) or by irregular disintegratoin of the fruit wall (Peng & Tobe, 1987). Technical characters, such as the way pollen grains are shed and seed- surface cell shape and orientation, are not useful in distinguishing these taxa; they have both pollen grains shed singly and seed-surface cells in parallel columns transversely elongate to the seed length. They have turbinate or slightly broadly turbinate capsules 1.5-4.5 mm long and are extremely vari- able in cauline leaf shape, which ranges from ob- ovate to spatulate-oblanceolate, narrowly oblan- ceolate, or sublinear. No consistent correlation was found between leaf shape and capsule size. When plants of distinct leaf shape and capsule size were collected and cultivated in an experimental green- house, the leaf shapes tended to converge and become spatulate-oblanceolate, but the capsule sizes remained constant. Based on counts from 30 populations, including seven reported by Raven & Tai (1979), differences in chromosome number are indeed correlated with mature capsule size. All plants with n = 24 have mature capsules 1.5-2(-2.5) mm long, while all plants with n = 32 have mature capsules (2-)2.5— 4(-4.5) mm long. The morphology of plants with = 24 fits very closely the description of L. simpsonii, whereas the morphology of plants with n = 32 clearly corresponds to that of L. curtissii (including L. spathulifolia). Compared with plants with n = 32, plants with 4 have generally diminutive floral parts: shorter bracteoles, sepals, stamens, and ovaries. Further, plants with n — 24 rarely exhibit vestigial petals, whereas those with n — 32 frequently have 1—3 caducous petals on at least some flowers. Veg- etatively, plants with n — 24 are erect to ascending, or sometimes are prostrate and rooting at the nodes. They are pale green with slender stems often much branched from below or above. The leaves are sometimes quite small, tending to be opposite or Mp at the lower nodes. By contrast, plants n — 32 are usually dark green or sometimes purplish, have stouter erect stems branched above, and leaves usually alternate throughout. Although L. simpsoni and L. curtissii fre- quently are sympatric, they are ecologically distinct and seldom actually intermix. Plants with n = tend to grow along roadsides with other weeds in moist sandy soil. Plants with n — 32 grow far from roadsides in black muck and often in deep standing water, often mixed with tall grasses or sedges. In view of this evidence, L. simpsonii is con- sidered a hexaploid with n = 24, and L. curtissii (including L. spathulifolia) an octoploid with n = 2 A collection from Clay Co., Florida (Raven 18690) was originally identified as L. pilosa and determined to be tetraploid with n = 16 (Raven & Tai, 1979). A more detailed morphological study reveals that this is best treated as L. ravenii (Peng, 1984). Nearly all previous collections of L. ravenii were initially identified as L. pilosa by their re- spective collectors: the two species are similar in being densely hirtellous, a distinct character not shared by any other member of sect. Microcar- pium. Ludwigia ravenii may be distinguished readily from L. pilosa by having oblong-obovoid capsules and seed-surface cells predominantly in parallel columns transversely elongate to the seed length, and by having shorter sepals, filaments, and styles. Furthermore, unlike L. pilosa, which is nearly always hirtellous on the lower half of the style and between the lobes of the nectary disk, all 28 collections of L. ravenii are completely glabrous in these areas. Among the voucher specimens for L. glabrous (n = 16, Raven & Tai, 1979), a collection from Fort Bend Co., Texas (Raven 19398), with short capsules and seed-surface cells in parallel columns transversely elongate to the seed length, has been placed under subsp. brachycarpa in Table 2. Another collection (Raven 19405) from the same county is preserved in two duplicate specimen one of them (FLAS) belongs to subsp. sheets: 978 Annals of the Missouri Botanical Garden ABLE 3. Strains ont used in artificial hy- bridization experiment L. alata (ALAY (a) Franklin Co., Florida, M 70575. (b) Levy Co., Florida, Dille L. curtissii (CUR) (a) Hillsborough Co., Florida, piat 435. (b) Martin Co., ach Co., Florida, Popence 1962. (f) Franklin Co. Florida, Godfrey 48. L. glandulosa subsp. glandulosa (GLA) (a) Santa Rosa Co., Florida, cig ers (b) Emanuel Co., Georgia, Peng (c) Columbus Co., North i a 897. (d) Columbus Co., North Carolina, Broome 865. L. glandulosa subsp. brachycarpa (BRA) (a) Cameron Parish, Louisiana, Peng 4367. L. lanceolata (LAN) (a) Highlands Co., Florida, Dille 370. (b) Highlands Co., Florida, Peng 4193. L. linearis (LIE) (a) Jasper Co., South Carolina, Dille 350. (b) St. Tammany Parish, Louisiana, Dille 420. (c) Jasper Co., South Carolina, Peng 3935. (d) Emanuel Co., Georgia, Peng 4023. L. linifolia (LIF) (a) rs Huimanguillo, Tabasco, Mexico, Cow- 263 (b) Hillsborough Co., Florida, Dille 427. (c) Wakulla Co., Florida, Godfrey 77091. (d) Franklin Co., Florida, Peng 4343. L. microcarpa (MIC) (a) Clay Co., Florida, Dille (b) Jackson Co., Florida, Godficy 77093. (c) Jones Co., North Carolina, Peng 3800. (d) Franklin Co., Florida, Peng 4348. L. pilosa (PIL) (a) Horry Co., South Carolina, Dille 342. (b) Hancock Co., Mississippi, Dille 419. (c) Emanuel Co., Georgia, Peng 4025. L. polycarpa (POL) n Co., Missouri, Dille 328. n Co., Missouri, Dille 436. (c) Lincoln í Co., ‘Missouri. Dille 443. L. simpsonii (SIM) (a) Collier Co., Florida, Dille 378. (b) Sarasota Co., i O. 2 (e) Sarasota Co., Florida, e 4313. L. sphaerocarpa (SPH) (a) Jasper Co., South Carolina, Dille 348. (b) Wakulla Co., Florida, Dille 401 (c) Wakulla Co., Florida, Dille 402. TABLE 3. Continued. L suffruticosa (SUF) (a) Hillsborough Co., Florida, Dille 423. (b) Hillsborough Co., Florida, Dille 424. (c) Hillsborough Co., Florida, Dille 434. * Three-letter abbreviations are used for each taxon in Tables 3-14 and Figures 5, 6, 21, brachycarpa; the other (DS), however, is a mixture of two plants of subsp. L. glandulosa with one of subsp. brachycarpa. This is arbitrarily placed un- der subsp. brachycarpa in Table 2; both taxa have the same chromosome number. The following collections counted by Raven & Tai (1979) as n — 16 are considered to represent hybrid populations of L. pilosa x L. sphaerocarpa and have been excluded from Table 2: Florida: Highlands Co., Raven 18683 (as L. pilosa); Clay Co., Raven 18680 (as L. sphaerocarpa); Colum- bia Co., Raven 18634 (as L. sphaerocarpa). South Carolina: Beaufort Co., Raven 18716 (as L. sphaerocarpa); Colleton Co., Raven 18718 (as L. sphaerocarpa). EXPERIMENTAL HYBRIDIZATION An extensive artificial hybridization program was carried out among members of sect. Microcarpium with the following objectives: (1) to determine chro- mosome homologies and relationships of taxa with the same and with different ploidy levels; (2) to study whether chromosome repatterning has played a role in the diversification of the species; (3) to study the genetic isolating mechanisms that may have permitted preservation of the genetic integrity of the taxa; and (4) to examine the variation pattern generated by hybridization, which is useful in study- ing natural hybrid populations. Over 1,000 reciprocal crosses have been made among the 12 species and one additional subspecies of sect. Microcarpium. Ludwigia ravenii and L. stricta were not included because living plants were not available at the time these studies were con- ducted. Most of the attempted crosses resulted in seed set. Some of the failures probably resulted from damage by the sharp forceps used to emas- culate the flowers and to transfer the pollen to these usually small-flowered plants. In these cases, additional attempts were made using either the same or different parental strains, and seed set was usually obtained. The only consistent failure oc- curred in some crosses involving L. microcarpa as pollen parent. This species is the smallest in Volume 75, Number 3 1988 Peng Ludwigia sect. Microcarpium 979 LE4. Percentages of stainable pollen in hybrids between species of Ludwigia sect. Microcarpium. Ludwigia pilosa group, diploid group, and L. curtissii complex are separated by lines. $ ó ALA BRA GLA POL SPH SUF LIF MIC SIM CUR 71 63 1 0 81 stature and in flower size and has the shortest style and smallest pollen grains among members of sect. Microcarpium. 'The few ovulate parents that failed to set seed in these crosses were of taxa with styles 3-6 times longer than those of L. microcarpa. Rather than genetic disharmony, it is likely that these failures were due to the improbability of small grains containing sufficient food reserves to support pollen tube growth through long styles (review in Lee, 1978), since the reciprocal crosses using shorter-styled L. microcarpa as ovule parent in- variably set seeds. As all plants in sect. Microcarpium are self- compatible and modally autogamous, self-pollina- tion sometimes interfered with the hybridizations, especially in crosses involving a small-flowered ovu- late parent. Part of the seed set may thus reflect a small to moderate amount of self-pollination de- spite regular emasculation. The seedlings resulting from self-pollination, however, can almost always be distinguished from those resulting from hybrid- ization by the leaf color or shape, or even by growth rate, when both types of seedlings were grown together. The strains utilized in the crossing experiments are presented in Table 3. The results of the artificial hybridizations are summarized in Figure 5 and indicate three more or less interfertile groups in sect. Microcarpium: (1) all the tetraploid taxa plus L. alata, a hexaploid; (2) L. linearis and L. lin- ifolia, both diploid species; and (3) the L. curtissii complex (hexa- and octoploids). Data on the per- centages of stainable pollen in hybrids (Table 4) support this interpretation. The crossing results will be discussed in the following order: (I) the diploid group (including L. linearis, L. linifolia, and L. microcarpa); (II) the L. pilosa group (including all tetraploids and a single hexaploid, L. alata); (III) the L. curtissii complex (including L. curtissii and L. simpsonii); (IV) crosses between the diploid group and the L. pilosa group; (V) crosses between the diploid group 980 Annals of the Missouri Botanical Garden N ALA BRA GLA LAN PIL POL SPH SUF|LIE LIF MIC | SIM CUR ài e e 9 9 @ @ O O — x x i o © O) X X X le e € ` @-+ @ O x O O e - ° 00001: x o]/0 Oo "le e e e . e 9: o- o P mo | O ` @-- "IO x x © a e ° ° e o ° X X — LIE n= 8 X X X O O X @ kasi O LIF i" ms} X X OO x x x x|Q X X MIC 810 O x x O O O O O © SIM n=24 | O) X X X O © © e CUR n=32 Xx © x x © © OJO x @ e F, hybrids producing abundant seeds (usually >80% of potental number) @ Fi hybrids producing fewer seeds (e.g., often about 1/2 of potential number) © F4 hybrids producing very few seeds (« 596 of potential number) Q F, hybrids flowering, but setting no seeds X P hybrid seed formed failed to germinate or died Soon after germination Hybridization resulted in abundant seed set , but seeds not sown — Hybridization failed to result in seed set . Summary of artificial hybridization between species of Ludwigia sect. Microcarpium. Ludwigia pilosa group, diploid group, and L. curtissii complex are separated by lines. Volume 75, Number 3 Peng 981 1988 Ludwigia sect. Microcarpium TABLE 5. Meiotic configurations and pollen stainability in interspecific Ludwigia diploid hybrids. Number of Cells with Modal Configuration/ Pollen Hybrid Modal Meiotic Configuration Total Number of Stainabilty Combination Observed Cells Examined (%) LIE d x LIF c 1IV + (4-6)II + (4-0)I 6/20 48 LIF c x LIE d 1IV + (5-6)II + (2-0)I 5/9 47 MIC b x LIEc (0-1) + (16-14)I 5/10 0 MIC b x LIF c (2-3) + (12-10)I 4/4 6 and the L. curtissii complex; and (VI) crosses between the L. pilosa group and the L. curtissii complex. HYBRIDS WITHIN THE DIPLOID GROUP (LUDWIGIA LINEARIS, L. LINIFOLIA, AND L. MICROCARPA) These interspecific crosses resulted in significant seed set except for a few cases in which L. mi- crocarpa was pollen donor. As suggested salian, these failures were probably due to the inability of the small pollen grains of L. microcarpa to grow through the long styles of L. linearis and L. lin- ifolia. Four hybrid combinations were obtained (Table 5) and grown to flowering. Meiotic analysis and pollen stainability Meiosis was normal in the examined natural populations of the diploid parental strains. The chromosomes regularly formed eight bivalents, with no univalents or multivalents observed. In the re- ciprocal crosses between L. linearis and L. lini- folia, meiotic configurations ranged from eight bi- valents to four bivalents and eight univalents; a ring or chain quadrivalent was observed in 11 of the 29 cells examined. In a single cell, a hexavalent was observed; trivalents were observed occasion- ally. In the translocation heterozygote, chiasma frequency and quadrivalent frequency are inter- dependent to a certain extent, as McCollum (1958) suggested. If chiasmata are formed in each of the four paired arms of the cross-shaped configuration at pachytene, then the four chromosomes form a ring at metaphase I; progressively lower numbers of chiasmata will give a chain of three plus a univalent, two bivalents, or four univalents. The observation of the quadrivalent in meiosis of the diploid hybrid indicates that the strains of L. lin- earis and L. linifolia hybridized differ by at least one reciprocal translocation, which may help to account for the inviability in about half of the pollen in the hybrid. This is the first demonstration of a reciprocal translocation in Ludwigia. The inter- pretation of the single cell recorded as having a hexavalent is uncertain, and the observation should be confirmed. The reciprocal hybrids between L. linearis and L. linifolia produced many aborted seeds. How- ever, L. linearis has the highest number of seeds per capsule (about 700) among species of sect. Microcarpium. F, plants resulting from reciprocal crosses involving L. linearis and L. linifolia were also very fecund, and their output of viable seeds per capsule was therefore quite substantial, al- though reduced. An F, family of nearly 100 plants was grown for each of the reciprocal crosses. Some 25% of these plants died in the seedling stage, and another 25% were very weak. About 10-15% of the F,s flowered. Pollen stainability of the seven F, plants examined ranged from 1 to 52% (1, 7, 10, 15, 22, 32, 52%). None of the F,s that sur- vived were as vigorous and floriferous as the F, plants. In diakinesis and first metaphase of L. micro- carpa X L. linearis and L. microcarpa X L. linifolia univalents predominated. The bivalents ranged from O to 3, the rest being univalents. Heteromorphic bivalents were observed in at least a few cells. The paired chromosomes were some- times held together by a matrix connection instead of by a chiasma. However, a maximum of three true bivalents, held together by chiasmata, were observed in a few cells. The few paired chromo- somes did not always line up in the equatorial plane, more often being randomly placed like univalents. Micronuclei were present at the tetrad stage. Pollen grains were shed both singly and as (loose) tetrads. Their sizes varied. The stainability was 0% in microcarpa X L. linearis and 6% in L. micro- carpa X L. linifolia, which seemed to accord with the fact that no or one bivalent(s) were observed 982 Annals of the Missouri Botanical Garden in the former and two or three were observed in the latter. Morphology of the hybrids Reciprocal hybrids between L. linearis and L. linifolia resembled each other morphologically. They were vigorous and floriferous. In overall habit and capsule shape they were more similar to L. linearis. In floral details, however, the hybrids were intermediate between the parents. For example, the divided sporogenous tissue in the anthers, which is characteristic of L. linearis, also appeared in the hybrids, although the packets were fewer and shallower. Seed set was variable from capsule to capsule, but was generally fairly high. Ludwigia microcarpa, a small, highly autoga- mous herb with minute, apetalous flowers and ob- ovate-spatulate to spatulate leaves, differs from both L. linearis and L. linifolia in nearly every morphological aspect. Hybrid combinations be- tween L. microcarpa and L. linearis and those involving L. microcarpa and L. linifolia were vig- orous and morphologically intermediate between the parents. Both produced small flowers with four yellow petals and reached the maximum height of the taller parents. Hybrids between L. microcarpa and L. linearis were about 70 cm tall and had very narrowly elliptic leaves, whereas those be- tween L. microcarpa and L. linifolia were about 35 cm tall and had oblanceolate to narrowly ob- ovate leaves. Neither hybrid set fruit—the ovaries simply withered. HYBRIDS WITHIN THE LUDWIGIA PILOSA GROUP Parental strains of this group consisted of one hexaploid species, L. alata (n — 24), and seven tetraploid entities (n = 16). All interspecific recip- rocal crosses attempted invariably yielded abun- dant seed. As many hybrid combinations as green- house space and time allowed were grown (Fig. 5; Tables 6, 7). Hybrid seedlings were nearly always vigorous. Occasionally a few plants assumed stunt- ed growth or exhibited some morphological abnor- mality. These are considered to represent chance combinations of disharmonious genotypes from the two parents. For each hybrid combination, only the healthy F, individuals were grown to maturity for study. Meiotic analysis and pollen stainability Meiosis was normal in all parental strains of the tetraploid species as well as in L. alata, the only hexaploid in this group. From diakinesis to meta- phase I, 16 and 24 bivalents, respectively, were invariably observed in these taxa; no univalents or multivalents were detected. Disjunction and mi- crospore formation also appeared to be normal. The pollen stainability of the parental strains was 95-100%. As all the hybrids between tetraploid taxa ex- hibited a similar pattern of meiotic behavior, the data for these are pooled and discussed together. Hybrids between the tetraploid taxa and Ludwigia alata (hexaploid) were likewise similar with respect to meiotic behavior; these data will also be treated as a group. (1) Hybrids between tetraploid taxa (Table 6) In very many cases, meiosis in these hybrids appeared normal, exhibiting 16 bivalents from dia- kinesis to metaphase I. The majority of cells ex- hibited complete chromosome pairing, but 2-6 (very rarely 8-10) univalents were often encountered in prepared slides. In many cases these univalents appeared to have resulted from precocious sepa- ration of the bivalents (probably because no chias- mata developed). This was suggested by their shapes and nonrandom locations on the metaphase plates. Nevertheless, meiotic anaphase I was generally regular, and no micronuclei were observed in the tetrad stage. Occasionally bivalents had a slightly stretched or attenuated appearance. This stretch- ing may result from a failure of terminal chiasmata to disjoin properly (Grant, 1952). A few cells with 2-3 bivalents associated side by side were seen. This association is apparently not due to chiasmata but rather to the "sticky" matrix bands typical of the “pseudobivalents” in hybrids of other taxa (e.g., Bromus, Walters, 1954). Despite the few meiotic irregularities observed, the pollen stainability of the tetraploid hybrids was generally high (Tables 4, 6). Of the 17 hybrid combinations examined, 6 had estimated pollen stainabilities of 95-98%. With a few exceptions, most other hybrids exhibited 75-87% stainable pollen. Two hybrids failed to flower in the closed green- house: L. glandulosa subsp. brachycarpa x L. lanceolata and L. sphaerocarpa x L. glandulosa subsp. glandulosa. These were then moved to an open, netted greenhouse. Here, the hybrids be- tween L. sphaerocarpa X L. glandulosa subsp. glandulosa produced a single flower; its pollen stainability was not studied. The hybrid L. glan- dulosa subsp. brachycarpa X L. lanceolata was very vigorous and produced many flowers two Volume 75, Number 3 1988 Peng 983 Ludwigia sect. Microcarpium TABLE 6. Meiotic configurations and pollen stainability in interspecific Ludwigia tetraploid hybrids. Number of Cells with Modal Configuration/ olle Hybrid Modal Meiotic Configuration Total r o Stainability Combination Observed Cells Examined (96) BRA x GLA b 1611 10/10 97 BRA x LAN a 1511 + 2I 4/12 56 BRA x POL c 67 GLA b x BRA 1611 15/25 96 GLA d x POL b Fruiting specimen showed plump capsules with abundant seeds. GLA b x SUF b LAN a x PIL a 1611 12/13 96 LAN a x POL c 1611 14/19 83 LAN a x SPH b 1611 9/13 95 LAN a x SUF a 1411 + 4I 4/6 95 PIL e x BRA 161 5/5 87 PIL b x GLA d (14-16)II + (4-0)I 5/6 84 PIL b x LAN a 16II 14/15 75 PIL b x SPH b 1511 + 2I /7 98 PIL b x SUF a 78 POL c x LAN a 83 POL c x PIL b 83 POL c x SPH b Fruiting specimen showed plump capsules with abundant seeds. SPH b x GLA c Fruiting specimen showed plump capsules with abundant seeds. SPH c x LAN a 87 months after transfer. Its pollen stainability was estimated to be only 56%, significantly lower than one would expect, since this plant has as much chromosome pairing (13—16 pairs) as other hybrids that exhibited 75-98% stainable pollen. Jones (1976) noted that environmental factors may exert considerable influence on production of presumably normal stainable pollen. She found that plants of some Aster species moved out from the greenhouse to the field consistently yielded signif- icantly reduced fractions of normal pollen. Wheth- er this is the case for L. glandulosa subsp. bra- chycarpa X L. lanceolata remains to be confirmed by growing them in a closed experimental green- house. It is also possible that the low observed pollen stainability is the result either of genic in- teraction between the parents or of the pronounced meiotic irregularities discussed earlier. However, the present sample size is not large enough to establish that hybrids from the cross L. glandulosa subsp. brachycarpa x L. lanceolata indeed ex- hibit a higher degree of meiotic irregularity than other hybrids. Reciprocal hybrids between tetraploid taxa gen- erally exhibited similar values for pollen stainability (Tables 4, 6). Different values were obtained for the reciprocal crosses between L. lanceolata and L. pilosa (96% and 75%), but different strains of L. pilosa were used in these crosses. These data suggest that the tetraploid species in sect. Microcarpium in general have high chro- mosome homology, and complete pairing is fre- quently observed in almost all interspecific hybrids. The few meiotic irregularities occasionally observ (a few univalents, slightly attenuated bivalents, and sticky bivalents) have little effect on the hybrid fertility (as estimated by pollen stainability). This was further substantiated by abundant seed sets of all the hybrids. A small family of vigorous F, individuals was reared for two randomly selected hybrids: L. pi- losa x L. glandulosa subsp. glandulosa (17 plants) and L. lanceolata x L. suffruticosa (19 plants). These populations were small, owing to the limited space and time available to handle them. They generally received less care than the F,s. There- fore, the several plants that were weak or died may not necessarily reflect F, weakness or breakdown. For example, the organic potting soil that was used was easily spoiled since all pots were continuously kept in standing water and were sprayed with in- secticide periodically. Had the potting soil not been changed as needed, the plants might have remained vegetative, become weak, or even died. 984 Annals of the Missouri Botanical Garden TABLE 7. Ludwigia species and the hexaploid species (L. a Meiotic configurations and pollen oe in hybrids resulting from crosses between tetraploid ata). Number of Cells with Modal Configuration / Pollen Hybrid Modal Meiotic Configuration Total Number of Stainability Combination Observed Cells Examined (90) ALA a x GLA c 1611 + 8I 10/16 47 GLAd x ALAa (15-16)11 + (10-8)I 7/9 46 ALA a x LAN a 1611 + 8I 15/29 62 LAN a x ALA a (15-16) + (10-8)I 4/4 64 ALA a x PIL a (12-16)11 + (16-8)I 10/10 52 PIL a x ALA a (15-16)II + (10-8)I 6/9 43 ALA a x POL a 50 ALA a x POL b 56 POL a x ALA a (11- pt * (18-10)I 8/11 43 ALA a x SPH b loll + 13/22 71 SPH a x ALA a (13- Us * (14-8)I 2/1 4T ALA a x SUF a 1611 + 8I 5/6 63 SUF c x ALA a (14-16)II + (12-8)I 1/7 61 All F,s that did grow were vigorous. In the cross L. pilosa X L. glandulosa subsp. glandulosa, the only individual that flowered was quite unlike either the parents or the F,s in morphology but had 91% stainable pollen. In L. lanceolata x L. suffruti- cosa, about 10 F, individuals flowered, some of which resembled the intermediate F,s, whereas oth- ers were similar to L. lanceolata. Pollen stainability was quite different from plant to plant, ranging from 1% to 97% (1, 12, 51, 63, 74, 97%) in the six plants studied. Meiosis of an F, individual (pollen stainability not assessed) was nearly normal. Com- lete chromosome pairing was nearly always ex- hibited, although 2-4 univalents were occasionally seen. Some of the univalents were formed as a result of precocious disjunction of the bivalents. (2) Hybrids between tetraploid taxa and hexaploid species (Table 7) The hexaploid L. alata (n — 24) was crossed reciprocally to nearly all of the tetraploid (n = 16) taxa. All crosses resulted in vigorous, floriferous, pentaploid hybrids. Meiosis in the F, individuals typically exhibited a maximum of 16 bivalents and 8 scattered univalents, although exceptional con- figurations of 17 bivalents and 6 univalents were observed in L. alata x L. lanceolata (one cell), L. alata x L. sphaerocarpa (four cells), and L. pilosa X L. alata (one cell). One or two trivalents were seen occasionally in diakinesis and metaphase I of the following hybrids: L. alata x L. glan- dulosa subsp. glandulosa (one trivalent in two cell), L. alata x L. lanceolata (one trivalent in four cells; two triva- cells; two trivalents in one lents in one cell), L. alata x L. sphaerocarpa (two trivalents in one cell), and L. polycarpa x L. alata (one trivalent in two cells; two trivalents in one cell). Groups of 2-4 bivalents (often of similar shape and size) "sticking" to one another sidewise were observed in meiotic metaphase but not in diakinesis. These associations were obviously not a result of chiasmata formation. Eight univa- lents would normally be expected in these hybrids. Observations of additional univalents at metaphase I either represent precociously separated bivalents or chromosomes which have no homologue. Pres- ent observations suggest that in most cells which had more than eight univalents, these “extras” are often precociously separate bivalents, judged by their shapes and locations. Pseudobivalents (Walters, 1954) composed of two univalents held together by matrix bands were seen at least in a few metaphase I cells. One to six (rarely to ten) bivalents of stretched appearance were also seen occasionally. Some metaphase I cells alata X L. suffruticosa were peculiar in having attenuated ends on the bivalents. Lagging univalents were uniformly found in meiotic ana- in L. phase I and II of all hybrids, and micronuclei occurred in nearly all the sporads. Pollen stainability was in the range 43-71%, rather high for pentaploid hybrids exhibiting the above meiotic irregularities. All F, plants produced Volume 75, Number 3 1988 Peng 985 Ludwigia sect. Microcarpium some aborted and some viable seeds in the plump capsules. An F, family (L. glandulosa subsp. glan- dulosa X L. alata) of 32 vigorous plants was grown and many individuals probably flowered but were not studied in detail. One morphologically inter- mediate plant was examined, however, and had 31% stainable pollen. Morphology of the hybrids Because of an apparently chance distribution of dominance between parents, some hybrids are like- ly to possess some characters of one parent, some of the other, and some (due to incomplete domi- nance or polygenic inheritance) that are interme- diate (Stace, 1975). The net result is that hybrids are nearly always intermediate in overall mor- phology between the two parents. This is the sit- uation within sect. Microcarpium. Genomic inter- action resulting in a new character state has not been observed in these plants. With the exceptions of L. glandulosa subsp. glandulosa x L. poly- carpa and L. sphaerocarpa X L. glandulosa subsp. glandulosa, all hybrids were very vigorous and flowered over a period of at least a month. Even in these two exceptional hybrids, if additional seeds were sown, or if different parental strains were crossed, vigorously growing hybrids would, I believe, be expected. This is because of the oc- currence of a natural hybrid population of L. sphaerocarpa X L. glandulosa subsp. glandu- losa. As there are very many naturally occurring hybrid populations between members of the L. pi- losa group, it is valuable to discuss the morphology of artificial hybrids between these taxa. Since all reciprocal hybrids resembled each other, the dis- cussion of characters that follows is limited to cross- es in only one direction. (1) Tetraploid hybrids Ludwigia glandulosa subsp. glandulosa X L. polycarpa. 'These were dwarfs 5-20 cm in height at maturity, with an ascending habit. Despite their abnormal appearance, all produced a few flowers, set plump capsules with abundant seeds, and sur- vived in the experimental greenhouse for two con- secutive years. In general they were intermediate in vegetative and reproductive features, especially in shapes and sizes of floral parts and capsules. Like their parents, the hybrids shed pollen grains as tetrads and had seed-surface cells in parallel columns elongate to the seed length. Ludwigia glandulosa subsp. glandulosa x L. glandulosa subsp. brachycarpa. The intermedi- acy of these hybrids was apparent in capsule size (4.5-5.5 mm long) and seed surface, which ex- hibited a mixture of columnar cells elongate and transversely elongate to the seed length (Fig. 6). Ludwigia lanceolata x L. pilosa. The parents did not differ in seed-surface cell pattern or in the way their pollen grains were shed. The hybrids showed intermediate morphology in all aspects. Diagnostic characters include hirtellous pubes- cence and oblong-obpyramidal, 4-angled, unwinged capsules. Ludwigia lanceolata x L. polycarpa. The hy- brids were morphologically intermediate between the parents. Diagnostic characters include the 4-angled capsules and seed-surface cells (Fig. 7) basically similar in orientation to those of L. lycarpa (Table 1) but arranged in shorter columns. Some of them appeared more or less isodiametric in shape. Ludwigia lanceolata x L. sphaerocarpa. The hybrids were again intermediate. The obpyramidal capsules were puberulent and very slightly winged, and the seed surface was composed of some more or less isodiametric cells and numerous, variously oriented short columnar cells (Fig. 8). Ludwigia lanceolata x L. suffruticosa. The hybrids were generally intermediate in morphology. The stems were much branched as in L. lanceo- lata, but with a slightly congested and branched terminal inflorescence. The lower margins of se- pals, bracteoles, and peduncles were slightly hir- tellous. The capsules were weakly 4-winged and the seed-surface cells were subisodiametric as in both parents. Pollen grains were shed singly as in L. suffruticosa. Ludwigia pilosa x L. glandulosa subsp. glan- dulosa. These hybrids were similar to L. glan- dulosa subsp. glandulosa in having subcylindric capsules, which were, however, somewhat shorter (5-5.5 mm long). Otherwise the plants were in- termediate in general appearance, pubescence, and shape and size of floral parts. Seed-surface cells Fig. 9) were predominantly columnar and elongate to the seed length like those of L. glandulosa subsp. glandulosa (Fig. 2). Yet few columnar cells were transversely elongate or oblique to the seed length. Some subisodiametric cells characteristic of L. pilosa were also present. Ludwigia pilosa X L. sphaerocarpa. The hy- MUS 4: R. 2 1 PL A A 1 | LA | ~ strigillose) and shape and size of floral parts. Seed surface pattern (Figs. 10, 11, 12) was irregular and variable within the same capsule. A small yel- low petal was observed in a single flower. Vestigial Annals of the Missouri Botanical Garden NS lu wp tue 1i $ M M E Ç. NY N ` Ss =N v SN SYS -— N I dirt 4-17 f LL hy 15 "VIE Uu o A Aas * A AL A MM Volume 75, Number 3 1988 Peng 987 Ludwigia sect. Microcarpium petals, however, occur occasionally in both parental species. Ludwigia pilosa X L. suffruticosa. The hybrids were intermediate in leaf shape and habit, were similar to L. pilosa in being hirtellous, and resem- bled L. suffruticosa in having a congested inflo- rescence. Fruiting specimens were not studied. Ludwigia pilosa x L. brachycarpa. The hybrids were similar to L. pi- losa X L. glandulosa subsp. glandulosa in gen- eral, although they were less robust and had shorter subcylindric capsules 3-3.5 mm in length. The plants were strigillose. Their seed surface was com- posed predominantly of columnar cells, which were transversely elongate to the seed length (Fig. 13) as in L. glandulosa subsp. brachycarpa. Some glandulosa subsp. more or less isodiametric cells typical of L. pilosa were also present, however. Also, columnar cells that were either parallel to the seed length or ran- domly oriented were observed. Ludwigia polycarpa X L. pilosa. Hybrids were morphologically intermediate. Diagnostic features included the overall strigillose pubescence and the cubic-turbinate capsule with long bracteoles. These plants exhibited seed-surface cell patterns similar to those of L. pilosa X L. glandulosa subsp glandulosa (Fig. 9). Ludwigia sphaerocarpa X L. glandulosa subsp. glandulosa. The hybrids were not as vig- orous as other hybrid combinations involving mem- bers of this group. The plants were reddish and less than 20 cm high. They were glabrous except for the fruits, which were sparsely and minutely strigillose. Only a single individual flowered, which later set plump capsules oblong in outline. Seed- surface cell pattern was basically similar to that of L. glandulosa subsp. glandulosa, but with some irregularities. Ludwigia glandulosa subsp. brachycarpa X L. lanceolata. These hybrids were morphologically intermediate between the parents. The plants were nearly glabrous. The capsules were about 3 mm long, elongate obpyramidal, minutely strigillose on the sepal margins, and weakly winged on the an- gles. The bracteoles were short (about 1.2-1.5 mm long) and, as in L. glandulosa subsp. brachycar- pa, the seed surface (Fig. 14) consisted mostly of columnar cells elongate transversely to the seed length. Some cells were arranged parallel to the seed length. Isodiametric cells, as characteristic of L. lanceolata, were uncommon. (2) Pentaploid hybrids Ludwigia alata x L. glandulosa subsp. glan- dulosa. These hybrids were similar to L. alata in habit, leaf shape, and in having winged capsules. Intermediate characters included capsule shape (elongate obpyramidal) and pollen grains shed sin- gly and as tetrads. The seed surface in the hybrid consisted mainly of columnar cells elongate trans- versely to the seed length, as in L. alata, although the cells were much smaller and were sometimes more nearly isodiametric than on seeds of that species (Fig. 15). Columnar cells elongate parallel to the seed length and similar to those of L. glan- dulosa subsp. glandulosa were also observed. Ludwigia alata X L. lanceolata. The parental species were themselves very similar in floral mor- phology and habit. The hybrids can be identified only by seed surface (Fig. 16), which consists of a mixture of columnar cells elongate transversely to the seed length, as in L. alata, and more or less isodiametric cells characteristic of L. lanceolata. Pollen grains were shed singly as in L. alata. Ludwigia alata X L. pilosa. These hybrids were generally intermediate. They were neither glabrous as in L. alata nor hirsute as in L. pilosa, but were minutely villous all over. The capsules, like those of L. alata, were winged. Seed-surface cell pattern (Fig. 17) was similar to that of L. alata X L. lanceolata. Pollen grains were shed mostly as tetrads, although single grains were seen occasionally. Ludwigia alata X L. polycarpa. With their winged capsules, the hybrids were more similar to L. alata in general appearance. They had the minutely strigillose leaf margins characteristic of L. polycarpa, however. Conspicuously interme- diate characters include: sepal shape, seed surface FicunEs 6-19. El sphaerocarpa. — 13. L. pilosa x L. glandulosa subsp. brachycarpa. — 15. L. alata x L. glandulosa subsp. pda p piis L. lanceolata x L. alata 0.5 m L. lanceolata. — Photographs of seeds ec from experimental hybrids in Ludwigia sect. Microcarpium. — a andulosa subsp. glandulosa. — 7. L lanceolata x L. polycarpa. — peces x L. glandulosa subsp. glandulosa. — 10-12. L. pilosa x landulosa subsp. erar y x —17. L. alata x L. pilosa. —18, 19. L. alata x L. polycarpa. Scale bar = 988 Annals of the Missouri Botanical Garden pattern in which columnar cells were randomly oriented (Figs. 18, 19), and pollen grains shed as loose tetrads. Ludwigia alata X L. sphaerocarpa. These hy- brids were similar to L. alata in habit and in having winged capsules. However, they shed pollen grains as loose tetrads, and their capsules were strigillose, both intermediate character states. Seed-surface cell pattern is not of diagnostic value here, as both parents were similar in this respect. Ludwigia alata x L. suffruticosa. These hy- brids generally resembled L. alata. The flowers were loosely arranged along the apices of the stems, and the capsules were winged and glabrous. The seed-surface cell pattern was intermediate between that of the two parents, being similar to that shown in Figure 16. The hybrids shed pollen grains singly, as did both parents. In summary, artificial hybrids between members of the L. pilosa group were easily produced and were nearly always vigorous and floriferous. Mor- phologically, F, plants were more or less inter- mediate between the parents. The only character that exhibited some consistent degree of dominance is the winged capsules of L. alata. When tetraploid species with rounded capsules were crossed wit L. alata, the resultant hybrids invariably had dis- tinctly winged capsules. In contrast, when L. lan- ceolata, a tetraploid with winged capsules, was crossed with tetraploid species having rounded cap- sules, the F, individuals usually exhibited inter- mediate capsule shape—their capsules were 4-an- gled or at most slightly winged. Therefore, it seems that in hybrids involving L. alata, the apparent dominance of the winged-capsule character may be due to a multiple dose of genetic information received from this hexaploid. The single most important diagnostic character for hybrids within the Ludwigia pilosa group is the shape and size of the capsules. Other diagnostic characters include overall pubescence, seed-sur- face pattern, and to some extent whether pollen grains are shed singly or as tetrads. This pollen character is of limited value, as most species in the L. pilosa group shed pollen as tetrads. When one of the taxa that do shed pollen singly was used as a parent, loose tetrads or a mixture of single and tetrad pollen grains were commonly found in the hybrids. This pattern was also clearly shown by the diploid hybrids L. microcarpa X L. linearis and L. microcarpa X L. linifolia. Similarly, hy- brids between species that differ in their seed-sur- face pattern always showed a mixture of cell types or cells with intermediate shapes and/or of random orientation. Likewise, a hybrid resulting from cross- ing a hirtellous species with a glabrous species always showed minutely villous or strigillose pu- bescence. Examination of the above-mentioned di- agnostic characters permits very accurate deter- mination of the parentage involved in hybrids within the L. pilosa group. HYBRIDS WITHIN THE LUDWIGIA CURTISSII COMPLEX Two species, L. curtissii (n = 32) and L. simp- sonii (n — 24), are included in this complex. Re- ciprocal hybridizations (Table 8) between them re- sulted in vigorous, floriferous F, individuals. Meiotic analysis and pollen stainability Meiosis was normal in both parental species. Diakinesis and metaphase I cells consistently re- vealed only bivalents. Multivalents were never ob- served despite the polyploidy of both parents. The two parents have pollen stainability of 97-100%. Although reciprocal hybrids were made, meiosis was examined only in F,s in which L. curtissii was the ovulate parent and L. parent. Of the seven clear, analyzable cells, three were at diakinensis and four in metaphase I. They consistently showed a configuration of 24 bivalents and 8 univalents. Nearly all of the bivalents in the simpsonii the pollen metaphase I cells were regularly rod-shaped, and The eight univalents were scattered at random along a con- were oriented on the equatorial plate. tinuous bipolar spindle; some were probably lost at anaphase I and II, since micronuclei were observed in many sporads. In spite of the presence of scat- tered univalents, the hybrids produced 89% stain- able pollen grains and a moderate number of viable seeds. Eleven F, plants grew up to flower. All them branched profusely and set abundant fruits and seeds. The reciprocal F, hybrid, L. simpson- ii X L. curtissii, had a similar value for stainable pollen (81%). Morphology of the hybrids Ludwigia curtissii and L. simpsonii are often difficult to distinguish; the only consistent diag- nostic character available to separate them is the size of mature capsules. Thcse of L. simpsonii are 1.5-2(-2.5) mm long, those of L. curtissii (2-) 2.5-4.5 mm long. Artificial hybrids between them had robust, erect stems up to 85 cm high and capsules about 2.5 mm long. Such plants are likely to be identified as L. curtissii in the field or the herbarium, although their capsule size is at the Volume 75, Number 3 1988 Peng 989 Ludwigia sect. Microcarpium TABLE 8. Meiotic configurations and pollen stainability of hybrids between members of the Ludwiga curtissii complex. Number of Cells with Modal Configuration/ Hybrid Modal Meiotic Configu- Total Number of Pollen Combination ration Observed Cells Examined Stainability (%) CUR d x SIM d 2411 + 8I TAT 89 SIM e x CUR d 81 lower limit for L. curtissii. The hybrid plants set plump capsules with moderate amounts of viable HYBRIDS BETWEEN THE DIPLOID GROUP AND THE LUDWIGIA PILOSA GROUP Of the 42 hybrid combinations in which recip- rocal crossing attempts have been made involving diploid taxa and members of the L. pilosa group (Fig. 5, Table 9), only three failed to result in seed set. These crosses involved the small-flowered L. microcarpa as pollen parent and one of the larger- flowered species as ovulate parent. Eight of the 39 successful crosses yielded offspring that did not germinate. Of these, 15 resulted in weak F, indi- viduals that died soon after germination or re- mained sterile for the entire season. Only the re- maining 15 hybrid flowered successfully, but three of them were weak and died soon after anthesis. The 12 remaining combinations were vigorous and floriferous. Despite this, none of them set any seed. Their ovaries simply turned yellowish and dropped off or were somewhat per- sistent but shriveled after mechanical self-pollina- tu combinations on. The results of artificially hybridizing the different species of the above two groups ranged from seeds that were unable to germinate to vigorous and floriferous F, plants. A few general comments are appropriate here. First, the inability of hybrid seeds to germinate as observed in the present study is not necessarily a reliable indication of reproductive isolation. In some cases, seeds obtained from a particular cross did not germinate the first year, but additional seeds planted the following year did. Seed dormancy, however, is not characteristic of Ludwigia, at least of sect. Microcarpium. In a few rare cases, hybrid seeds were observed to ger- minate three or four months after they were sown. These would have been scored as germination fail- ures if the experiments had been terminated after two months, as was generally done. Second, on several occasions the majority of seeds failed to germinate or produced weak or stunted seedlings, although one or few vigorous and floriferous hybrid individuals grew to maturity. For example, only a single, healthy plant of L. microcarpa X L. alata was obtained from 18 seeds sown; none of the others germinated. Such failures to germinate are not mentioned in the above dis- cussion, and records were not kept, because the seeds of Ludwigia species are very small, and usually 50-200 seeds were sown for each hybrid combination. Third, weak or sterile hybrids could sometimes be brought to flower if they were grown very care- fully or if alternative parental strains were utilized to vary the genetic composition of the hybrids (for examples see Table 9). Meiotic analysis and pollen stainability Cytological data are available for ten hybrid combinations. These include seven hybrids result- ing from crosses between the tetraploid species and each of the three diploid species, including one reciprocal cross (Table 10), and three hybrids re- sulting from crosses between L. alata (hexaploid) and each of the three diploid species. (1) Hybrids resulting from crosses between the diploid and tetraploid species Crosses between diploids and tetraploids would normally be expected to form triploid hybrids. It is most interesting, therefore, that upon sowing 15 seeds resulting from crossing L. linifolia (n — 8) with L. lanceolata (n = 16), only 5 plants were obtained, all of which were tetraploid (2n — 32). That the diploid L. linifolia was used as the ovulate parent suggests that these 2n = 32 plants were not simply the result of self-pollination in the tet- raploid. This chromosome number was apparently produced by the union of an unreduced egg from the diploid L. linifolia with a normal sperm nucleus 990 Annals of the Missouri Botanical Garden TABLE 9. Summary of crossing results between taxa of the diploid group and the Ludwigia pilosa group. L. linearis as female parent LIE b x ALA a Seeds failed to germinate LIE a x LAN a Plants weak, with reddish leaves, died soon after germination. LIE a x PIL b Seeds failed to germinate. LIE b x POL c Plants weak, barely flowered; no viable pollen. LIE a x SPHc F, hybrids vigorous and flowered (see Table 10) LIE a x SUF b Low germination percentage; weak plants died when ca. 4 cm high. L. linearis as male parent ALA a x LIE a F, hybrids M arr: and flowered (see Table 11). BRA x LIE b Formed a mat in 3-inch pots; remained sterile. GLA a x LIE a Plants dus about 10-30 c qs GLA a x LIE b Plants sterile, about 5-30 c GLA a x LIE d F, hybrids vigorous and flowered (on Table 10). POL c x LIE b Seeds failed to germina SPH c x LIE a F, hybrids vigorous and poer (see Table 10). SUF c x LIE b Some died in cotyledon stage; some produced a few pinkish leaves and soon withered. L. linifolia as female parent LIF b x ALA a Seeds failed to germinate. LIF c x ALA b Seeds failed to germinate. LIF c x BRA C en expanded a month bd radicles had protruded; plants weak and died when ca. 1 cm high with 4-6 lea LIF c x GLA c Seeds tailed to germinate LIF c x GLA b Plants weak, mob died 2- 3 cm high; one plant barely flowered and showed 0% stain- able pollen LIF b x LAN a F, hybrids vigorous and flowered, but with an unexpected chromosome number of 2n — 32 (see T 0). LIF c x PIL b Very low Aria ee 3 weak plants obtained, which died when ca. 1 cm high with about 10 leave LIF b x POL c Seeds failed to germinate. LIF c x SPH c Plants weak, sterile, about 5-7 cm high. LIF b x SUFc Five out of 20 seeds germinated; all very weak, died soon. LIF c x SUFa Numerous seeds germinated, but all died in 4-6-leaved stage. L. linifolia as male parent ALA a x LIF c F, hybrids vigorous and flowered (see Table 11). BRA x LIF c Plants remained small, died when ca. 2 cm high. GLA b x LIF a Plants weak, with reddish brown leaves, sterile. LAN a x LIF b Six out of 30 seeds germinated; all died in aa stage. PIL b x LIF b F, hybrids vigorous and asawa Pas; Table 10). PIL b x LIF c Cotyledons reddish, soon wither POL c x LIF b Died in cotyledon stage. POL c x LIF c Two weak plants barely flowered and soon died; 0% stainable pollen. SPH c x LIF c Plants with pinkish yellow leaves, 10-15 cm high, sterile. SUF b x LIF b Seeds failed to germinate. L. microcarpa as female parent MIC a x ALA a F, hybrids vigorous and flowered (see Table 11). MIC a x GLA b Plants healthy, flowered when 10-15 cm high. MIC b x LAN a Seeds failed to germinate. MIC a x PIL b Seeds failed to germinate. MIC a x POL c F, hybrids vigorous and flowered (see Table 10). MIC c x SPH c Plante flowered, but meiosis was not studied; 4% stainable pollen. L. microcarpa as male parent ALA b x MIC a Crosses failed to set s GLA b x MIC a F, plants healthy 24 ce (see Table 10). LAN b x MIC d Plants recently grown; healthy, but have not yet flowered. PIL c x MIC b Crosses failed to set seed. SPH a x MIC a Very low germination rate; soon died. SUF b x MIC a elis failed to set seed. Volume 75, Number 3 1988 Peng Ludwigia sect. Microcarpium TABLE 10. Meiotic configurations and pollen mane dn in seven hybrids resulting from crosses between the diploid with tetraploid species of Ludwigia sect. Microcarpium Number of Cells wit a Configuration/ Pollen Hybrid Modal Meiotic Configuration Total Number o Stainability Combination Observed Cells Examined (920) L. linearis as one parent GLA a x LIE b (4-7)11 + (16-10)I 5/5 2 LIE a x SPH c (4-1) + (16-10)I 6/9 1 SPH c x LIE a (6-7) + (12-10)I 19/24 1 L. linifolia as one parent LIFb x LANa 811 + 161 4/8 1 PIL b x LIF b (6-7) + (12-10)I 6/6 0 L. microcarpa as one parent GLA b x MICa (2-6)I + (20-12) 9/11 0 MIC a x POL c 81I + 81; 411 + 161 2/2 13 Hybrids with 2n = 32 , a number apparently produced by the union of an unreduced egg from the diploid (L. ) linifolia) with a normal sperm nucleus from the tetrapoloid (L. lanceolata). from the tetraploid L. lanceolata. These plants are, following the terminology of Harlan & deWet (1975), Class I Polyploids. Diakinesis observed in one plant resulting from this cross clearly showed a configuration of 8II + 161. Configurations at metaphase I, however, ranged from 6II + 20I to 8II + 161, with two cells each exhibiting a single trivalent. The bivalents were apparently formed between chromosomes of the duplicated L. linifolia genome. This unexpected but significant finding suggests strongly that (a) chromosomes in the two genomes in the tetraploid L. lanceolata are not homologous and thus remained unpaired; and (b) the genome of L. linifolia does not pair with either of the genomes of the tetraploid species when true chromosome homologues are present. he cytological behavior of all other hybrids generally followed a consistent pattern (Table 10). Although the number of bivalents formed was vari- able, it never exceeded eight, the haploid chro- mosome number of the diploid species. The number of bivalents ranged from three to seven in most hybrids, although values ranging from one to six were observed in L. glandulosa subsp. glandu- losa X L. microcarpa. The configuration 811 + 8lI was observed in only a single cell in L. micro- carpa X L. polycarpa. However, in this meta- phase I cell, two of the bivalents are in a somewhat perpendicular orientation to the other six bivalents and appear to be attached univalents. A single trivalent was seen in three of the 33 cells examined of the reciprocal hybrids between L. linearis and L. sphaerocarpa. One to three heteromorphic bi- valents, in which the two chromosomes differed in shape or size, were observed on occasion. Atten- uated bivalents and precociously separated biva- lents were also seen in some cells. Pseudoassocia- tion resulting from stickiness between two univalents, two bivalents, or one of each, was also observed in a few cases. It is observed that the two to seven chromosome pairs observed in these diploid x tetraploid hybrids indicate that the diploid genome is homologous with only one of the tetraploid genomes. The cytological abnormalities and the high pro- portion of univalents observed in these hybrids are reflected in the low pollen stainability. Six of the seven hybrids had less than 2% stainable pollen; only L. microcarpa X L. polycarpa had greater than 1046 stainable pollen (Table 10), presumably due to chance events. In L. microcarpa x polycarpa, the pollen was shed as single grains and as tetrads, a combination of the characters seen in the two parental taxa. Most stainable pollen, however, was shed as single grains, an observation which cannot be explained at present. (2) Hybrids resulting from crosses between the diploid and hexaploid species All hybrids involving diploid and hexaploid taxa were made (Table hen L. alata was crossed to L. linearis and L. linifolia, the resulting hybrids typically exhibited a configuration of (1-2)III 992 Annals of the Missouri Botanical Garden TABLE 11. — Meiotic configurations and pollen stainability in three hybrids resulting from crosses between the diploid and hexaploid species of Ludwigia sect. Microcarpium. Number of Cells with Configuration/ olle Hybrid otal Number of — Stainability Combination Modal Meiotic Configuration Observed Cells Examined (%) ALA b x LIE a (1-2)II + (4- iden + (21-16)I 7/12 1 ALA b x LIF c 111 + 611 + 1/6 0 MIC a x ALA b lIV + (7- d + (14-8)I 5/14 15 (4-6)II + (21-16)I in both diakinesis and meta- phase I, although a configuration of 8II + 16I was occasionally seen. There is no unequivocal evi- dence to indicate whether the trivalents represent intergenomic homology in the hexaploid L. alata or associations between chromosomes of L. alata and those of L. linearis or L. linifolia. In meta- phase I of many cells, one to three bivalents ap- peared to be attached univalents and were often observed away from the equatorial plate. Other cytological aberrations, including heteromorphic bivalents and elongate, attenuated bivalents, were very common. Normal rod or ring bivalents were infrequent. In L. microcarpa X L. alata the modal meiotic configuration was lIV + (7—10)II + (14-8)L, with maximum pairing of either 11V + 10H + 8I or 1111 + 101. In the latter case, the quadrivalent was not seen, as it probably separated into two 2 L. curtissii complex bivalents because of a lack of sufficient chiasmata o hold them together. Rings or chains of four chromosomes were more commonly seen in dia- kinesis than in metphase I. Trivalents were not noted. It is uncertain whether the quadrivalent indicated that, in addition to the difference in ploidy level, the two species differ by a reciprocal trans- location, as the multivalent can also result from associations of one chromosome from L. micro- carpa and three chromosomes each from one of the three genomes of L. alata. The bivalents observed in the meiotic first meta- phase of L. microcarpa x L. alata were typically normal, being ring- or rod-shaped. Attenuated bi- valents were very seldom found. Heteromorphic bivalents or attached univalents were not observed. The reduced occurrence of cytological aberrations and the higher number of chromosome pairings in these hybrids probably account for their higher Summary of crossing results between the diploid species of Ludwigia sect. Microcarpium and the L. curtissii as female parent CUR a x LIE a Seeds failed to germin CUR a x LIE b CUR a x LIE c Seeds failed to germina CUR e x LIE b CUR a x LIF b Hybrids vigorous and floriferous. CUR f x MIC a L. curtissii as male parent LIE b x CUR f ered ultimately. Seeds failed to germinate. LIF b x CUR a L. simpsonii as female parent SIM d x LIE b Hybrids stunted but flowered. SIM d x LIF c Seeds failed to germinate SIM e x LIF c Seeds failed to germinate SIM f x LIF c Seeds failed to germinate. L. simpsonii as male parent LIF c x SIM d MIC d x SIM d Hybrids vigorous and floriferous. ion. Hybrids vigorous and floriferous. ate Most seeds failed to germinate; those that did germinate died at the cotyledon stage. Hybrids stunted at first, ere and floriferous eventually. Numerous plants remained in the cotyledon stage and were reddish, four of which flow- Plants remained small and rosettelike, ca. 10-leaved, 1 cm high 7 months after germina- Volume 75, Number 3 1988 Peng 993 Ludwigia sect. Microcarpium TaBLE 13. Meiotic configurations and pollen stainability in hybrids resulting from crosses between diploid species of Ludwigia sect. Microcarpium and the L. curtissii complex. Number of Cells with Modal Configuration/ ollen brid Modal Meiotic Configuration Total Number of Stainability Combination Observed Cells Examined (90) CUR e x LIE b 0 CUR a x LIF b 8H + 24I 13/31 10 CUR a x MIC a (14-15) + (12-10)I 5/5 2 LIE b x CUR e 2 MICb x CURa 2 MICd x SIM d 9 SIM d x LIE b 4 pollen stainability (15%) than was observed in L. alata X L. linearis and the reciprocal hybrids of L. alata X L. linifolia (0-1%). Morphology of the hybrids In diploid species of sect. Microcarpium, the presence of petals appears to be dominant over the apetalous condition, judged from the consistent presence of petals in the petalous X apetalous hybrids. It is therefore of interest to examine this character in hybrids resulting from crosses between the petalous diploids and the apetalous polyploids. It was observed that all hybrids between diploids and tetraploids exhibited a variable number (0—4) of vestigial petals on different flowers of the same plant, whereas all hybrids between the diploids and hexaploids lacked petals completely. When two of the diploid species L. linearis or . linifolia were crossed to the tetraploid taxa, the resulting hybrids were somewhat intermediate in overall pubescence, leaf shape, capsule shape and size, sepal shape and size, and bracteole length. However, when L. microcarpa, the third diploid, was crossed with tetraploids, the F, hybrids were generally similar to their tetraploid parent in aspect but were diminutive in height and in leaf and flower ze. When all three of these diploids were each crossed to the hexaploid, L. alata, the resultant F, hybrids were more similar to L. alata, particularly in ex- hibiting its characteristic winged capsules. The hy- brids L. alata x L. linearis and L. alata x L. linifolia resembled each other in their slightly nar- rower and longer capsules as compared with those of L. alata. Neverthelesss, L. alata x L. linearis had sparsely strigillose capsules, short sepals, and bracteoles shorter than the ovary, whereas L. ala- a X L. linifolia was completely glabrous, had elongate acuminate sepals, and had bracteoles long- er than the ovary. Hybrids between L. microcarpa and L. alata had small leaves and flowers, but were at least as robust and tall as L. alata. This is in sharp contrast to the situation in hybrids resulting from crosses between L. microcarpa and the tetraploid taxa (see above). When diploid species (with pollen shed as tet- rads) were crossed to hexaploid species (with pollen grains single), the resulting hybrids had single pol- en grains only. Hybrids resulting from crosses between diploid species having single grains and tetraploid species having tetrad pollen showed a mixture of tetrad and single pollen in mature an- thers. Crosses between diploids and tetraploids that both shed pollen as tetrads yielded hybrids that also produce tetrads. In the hybrid L. linifolia x L. lanceolata, however, where unreduced gametes of the diploid L. linifolia united with normal pollen of the tetraploid L. lanceolata to produce tetraploid hybrids, the pollen grains were shed mixed as tetrads and single grains. This result was not ex- pected. In this hybrid, the morphology was inter- mediate between L. linifolia and L. lanceolata. HYBRIDS BETWEEN THE DIPLOID GROUP AND THE LUDWIGIA CURTISSII COMPLEX A summary of the crossing results is shown in Table 12. Some of these data are supplemented by study of natural hybrid populations, as is dis- cussed below. Meiotic analysis and pollen stainability Six hybrid combinations were examined to de- termine percentage of stainable pollen. Two of these were also studied cytologically (Table 13). 994 Annals of the Missouri Botanical Garden TABLE 14. Summary of crossing results between members of the L. curtissii complex and the L. pilosa group. L. curtissii as female parent CUR a x BRA Only 4 4 seeds sown, none of which germinated. Plants vigorous and floriferous. ed. an 10 cm; set a few flowers. CUR a x PIL b Seeds failed to germinate. CUR a x POL c Plants flowered but were not dum CUR a x SPH b Plants very vigorous and floriferou CUR a x SUF b Plants dwarf, but were healthy Ried flowered: L. curtissii as male parent ALA a x CUR a Plants with reddish leaves, rosettelike, i when 6-leav LAN a x CUR a Plants started losing leaves when dd PIL a x CUR f Plants very vigorous and floriferou POL a x CUR a Plants flowered but were not denm SPH a x CUR a L. simpsonii as female parent SIM a x ALA a Plants vigorous and xa SIM c x LAN a Seeds failed to ger SIM a x PIL a Died in cotyledo res e. SIM a x SUF b L. simpsonii as male parent Seeds failed to germinate. ALA a x SIMa Plan GLA b x SIMa Plants vigorous and florife LAN a x SIMa TOR x SIM b Seeds failed to germinate Plants very vigorous and floriferous. ts weak, leaves Hii did died when 6- or 8-leaved. Plants with reddish leaves jen much branched below; flowered recently. Of the 31 analyzable cells in L. curtissii (n = 32) x L. linifolia (n — 8), 22 formed 1j bivalents and univalents, ranging from (8-12) (24-16)I, with a modal configuration of 811 + bur nine cells formed multivalents, in which four cells had a configuration of 1111 + (6-9)II + (25-19)I, and five cells showed 11V + (6-9)II + (22-18)I. An exceptional cell with only 5II + 301 was also observed. In some cells chromosomes were so sticky that cytological analysis was impossible. In L. curtissii X L. microcarpa (n — 8), only five analyzable cells were studied, the results being (14-15)II + (12-10)I. One or two sticky or pre- cociously disjunct bivalents were noted in two cells. Pollen stainability in all hybrids examined was very low, ranging 0-10% (Table 13). Despite this problem, a single F, individual of L. curtissii x L. linifolia was raised and another was raised from L. curtissii X L. microcarpa. The mained sterile and died, whereas the latter died shortly after anthesis; it had 17% stainable pollen. former re- Morphology of the hybrids Reciprocal hybrids of L. curtissii X L. linearis — 8) resemble each other. They are erect and branched on the upper stems. The tallest individual reached a height of 65 cm. The plants were ex- tremely similar to L. curtissii in aspect, and before anthesis were thought to be selfed progeny of L. curtissii. In fact, even their flowers resembled those of L. curtissii except that their ovaries were slightly narrower and four-angled and their bracteoles were slightly shorter. Most flowers were apetalous; in a few cases one or two vestigial petals were present. After anthesis, the ovaries turned yellow and fell off. Pollen was shed as loose tetrads and single grains. Hybrids between Ludwigia simpsonii (n = 24) and L. linearis could not be distinguished from L. curtissii X L. linearis. Plants of L. curtissii X L. linifolia exhibited more morphological intermediacy than those of L. curtissii X L. linearis, although both resulted from hybridization between an octoploid and a diploid. Plants of L. curtissii X L. linifolia were smaller than either parent (about 25-30 cm high), muc branched, and very floriferous (resembling L. lin- ifolia). The flowers commonly had one to four petals (intermediate) and were congested at ends of branches (uncommon in both parents). The leaves were generally similar to those of L. curtissii in shape but were slightly smaller. The ovaries were about as long as those of L. curtissii, but were not accrescent and were narrower and four-angled. Sepals were intermediate in shape and size. Pollen was shed mostly as tetrads, and with some single grains as well. These hybrids did not usually set Volume 75, Number 3 1988 Peng 995 Ludwigia sect. Microcarpium TABLE 15. of the L. curtissii complex and the L. pilosa group. Meiotic configurations and pollen stainability in hybrids resulting from crosses between members Number of Cells with Modal Con figuration/ otal Num ollen Hybrid ber of Cells Stainability Combination Modal Meiotic Configuration Observed Examined %) L. curtissii as one parent CUR a x GLA d 5 CUR a x POL c III + 411 + 371 1/1 32 CUR a x SPH b 31 CUR a x SUF b (11-13)II + (26-22)1 5/5 9 LAN a x CUR f 4 PILa x CUR a 1I + (10-13)II + (25-19)I 4/16 6 POLa x CURa omosomes very sticky, not analyzable 25 SPH a x CUR f (LIV) + (1-3)HI + (6-13)11 + (14-30)I 6/6 17 L. simpsonii as one parent GLA b x SIMa 3 LAN a x SIMa 11 + 511 + 271 1/3 11 SIM a x ALA a 20 seed, although one or two viable seeds were oc- casionally obtained. Plants of L. curtissii X L. microcarpa were very vigorous and floriferous, up to 80 cm high, and much branched. The leaves were similar to those of L. curtissii, but smaller. Similarly, all floral parts were reduced and somewhat resembled those of L. microcarpa. In the hybrid, however, the nectary discs were always distinctly raised, unlike those of L. microcarpa, in which they were nearly flat. The ovaries usually shriveled after anthesis, although in exceptional cases one to three seeds were produced. Plants of L. microcarpa X L. simpsonii were also very vigorous. Their floral parts were similar to those of L. curtissii X L. microcarpa, and their flowers were sterile. These hybrids were smaller (up to 50 em high) and had slightly broader leaves. As plant height and leaf shape are somewhat vari- able characters, it is difficult to distinguish between these two hybrids when they occur together in nature. HYBRIDS BETWEEN THE LUDWIGIA CURTISSII COMPLEX AND THE L. PILOSA GROUP As with crosses between the diploid group and the L. pilosa group, these crossing results were quite variable, ranging from total failure to ger- minate to vigorous and floriferous F, hybrids (Table 14). None of these F, plants were observed to set any seed, however, while most of the hybrids re- sulted from crosses between the L. curtisssii com- plex and the L. pilosa group set at least a few viable seeds. An F, family of 12 vigorous plants of L. simpsonii X L. alata was established. Meiotic analysis and pollen stainability In hybrids with L. simpsonii as one of the par- ents, meiosis was studied only in L. lanceolata x L. simpsonii (Table 15). Only three metaphase I cells were obtained, which had three to five biva- lents, most of which were chromosomes connected by a chromatin thread and were aligned randomly. A trivalent was seen in one of the cells. In hybrids with L. curtissii as the male parent, L. polycarpa X L. curtissii had sticky chromo- somes, which rendered study of meiosis difficult. Four other hybrid combinations were studied cytologically. In L. curtissii X L. polycarpa, only one first metaphase cell was analyzable; it showed a configuration of 1III + 4II + 371 with two attenuated bivalents in the equatorial plate. he other three hybrids resulting from crossing L. curtissii and the tetraploid species showed sig- nificantly higher chromosome associations. A max- imum of 13 bivalents were seen in at least some of the cells, and one tetravalent and a maximum of three trivalents were observed in others (Table 15). The meiotic metaphase figures of these hybrids generally consisted of bivalents or multivalents 996 Annals of the Missouri Botanical Garden aligned in the equatorial plate with many univa- lents scattered throughout the cell. The following cytological aberrations were occasionally observed: attenuated bivalents, precociously separated bi- valents, attached univalents, and sticky chromo- somes. The pollen stainability was, surprisingly, higher in Ludwigia curtisii X L. polycarpa (32%), which showed the least chromosome associations, and in L. polycarpa X L. curtissii (2595), which had stickier chromosomes than in other hybrids known to have higher levels of chromosome pairing. Since in L. curtissii X L. polycarpa only one meiotic cell was studied, the observed configuration pos- sibly could represent the lower limit of chromosome association in this hybrid. Variability of chromo- some pairing is a common phenomenon in species hybrids. The stickiness of meiotic chromosomes of L. polycarpa X L. curtissii, however, may have a genetic basis, as it was rarely shown by either of the parental species, or the stickiness might be attributable to environmental factors. My unpub- lished study of meiosis in a sterile natural hybrid between L. spathulata and L. palustris (both sect. Dantia) was not successful in 1979 due to pro- nounced chromosome stickiness. Nevertheless, very clear chromosomal configurations for the same clone were obtained the next year. In spite of the prevalence of univalents in mei- osis, these hybrids usually set at least a few seeds. They were quite unlike the hybrids between the L. curtissii complex and the diploid group, which were completely sterile. This difference is probably at- tributable to the fact that, when both parents are polyploids, the development of functional pollen and ovules is better able to withstand the random segregation or loss of some chromosomes (as lag- ging univalents) in meiosis because of genetic re- dundancy. Morphology of the hybrids The hybrids were generally intermediate mor- phologically. The leaves were characteristically ob- lanceolate or narrowly oblanceolate. The inter- mediate nature of the size, shape, and pubesence of the capsules could be diagnostic, although the winged capsules characteristic of L. alata and L. lanceolata were not taxonomically useful, as the capsules of all the hybrids shriveled to some extent and thus appeared winged. The differences between hybrids with L. curtissii and with one parent were strictly quantitative, with the latter being slightly smaller in their floral features and occasionally in height also. These differences were obvious only when the two hybrids were brought L. simpsonii as together and compared. When L. curtissii and L. simpsonii, both of which shed pollen singly, were crossed with species in the L. pilosa group, which shed pollen as tetrads, the resultant hybrids con- sistently produced a mixture of pollen in loose tetrads and single grains. ECOLOGY AND GEOGRAPHICAL. DISTRIBUTION With the exceptions of L. stricta, which is en- demic to Cuba, and L. polycarpa, which is dis- tributed mainly in the north-central United States, sect. Microcarpium is confined primarily to the Coastal Plain of the United States (Fig. 20). The detailed geographical distribution of each taxon in Ludwigia sect. Microcarpium is presented in a companion taxonomic paper (Peng, in press). The Coastal Plain is defined geologically as the flat area between the Atlantic and Gulf coasts and the Piedmont, and extending from the Gulf of Mex- ico to southern New England (Peattie, 1922). The soils of this area are chiefly gray sands and sandy loams, except in the swamps where the prevailing sands are covered by muck or peat (Cooke, 1 On the Coastal Plain, especially in areas close to the coast, the water table is seldom very far below the surface, and many areas are periodically or permanently flooded (Gleason & Cronquist, 1964). At one end of the Coastal Plain, in southern Texas, the climate is semiarid and the soils are alkaline, containing a high proportion of clay (Hunt, 1974). n addition to this general distribution of species of sect. Microcarpium in North America, several species extend further south. Ludwigia alata oc- curs in Jamaica, L. simpsonii in Cuba and Jamaica, and L. curtissii in the Bahamas. L. linifolia is disjunct to Tabasco, in the Yucatán Peninsula of Mexico, and /. microcarpa ranges to the Baha- mas, Cuba, and Jamaica. Ludwigia stricta, en- demic to Cuba, is the only species of the section that does not occur in the United States. Like Ludwigia species occurring in other parts of the world, plants of sect. Microcarpium grow in at least seasonally wet habitats. They are com- monly found along alluvial ground or in the shallow water of many areas, including ponds, lakes, rivers, streams, lagoons, sloughs, backwaters, swales, wet meadows or prairies, open swamp forests, drain- ages, and irrigation ditches. All species grow in sandy or occasionally peaty soils. SYMPATRIC OCCURRENCE AND NATURAL HYBRIDIZATION The results of experimental hybridizations reveal that vigorous and floriferous hybrid individuals can Volume 75, Number 3 1988 9 Ludwigia sect. Microcarpium UNA: b^ A 5 FicurE 20. Distribution of Ludwigia sect. Microcarpium in North America, with shading to indicate the combined distribution of all species of this section except L. polycarpa and L. stricta. Distribution of L. polycarpa is indicated by do i polycarpa in Kootenal Co., Idaho, are not mapped. readily be obtained between most members of sect. Microcarpium. Exceptions involved some crosses between members of the L. pilosa group (the tet- raploid taxa plus the hexaploid L. alata) and either the diploid species (L. linearis, L. linifolia, and L. microcarpa) or members of the Le complex (L. curtissii and L. simpsontt) (Fig. 5). These crosses resulted in seeds that failed to ger- minate or in inviable hybrids. Even in these in- curtissii stances, if alternative parental strains were utilized hybrids, vigorous F, individuals could sometimes be ob- tained. The general lack of postzygotic barriers, in conjunction with the facts that most Ludwigia species have overlapping geographic ranges, sim- to vary the genetic composition of the dashed line. ilar habitat requirements, and similar flowering pe- riods (during the summer), and that they are at least facultatively outcrossing, favor natural hy- bridization. Field observations and examination of herbarium specimens suggest that natural interspecific hy- bridization involving species of sect. Microcarpium occurs frequently. Intersectional hybridization is also quite common; at least seven hybrid combi- nations bridging sect. Microcarpium and sect. Dantia have been observed. Observation of individuals exhibiting a combi- nation of characters intermediate between distinct taxa initially suggests the possibility of natural hy- bridization. However, the members of two species 998 Annals of the Missouri Botanical Garden Q Species known to occur together ° Species that occur together and hybridize O) Species that occur together and probably hybridize, but such hybrids would be almost impossible to detect ° olojo ð ° O | O ° O| O| O 174 ° @ Q ALA | BRA [GLA | LAN | PIL | POL RAV SPH | SUF [LIE [LIF [MIC | SIM | CUR FIGURE 21. Sympatric occurrence of species in Ludwigia sect. Microcarpium. Acronyms are those given in ble 3. pairs, L. alata (n — 24)/L. lanceolata (n — 16) and L. curtissii (n = 32)/ L. simpsonii (n = 24), are usually so similar in appearance (especially as herbarium specimens) that their hybrids cannot be recognized readily. As indicated earlier, capsule shape and size are the most important characters for detecting hybrids in sect. Microcarpium. Other useful features in- clude overall pubescence, seed-surface cell shape and orientation, presence or absence of petals, whether pollen grains are shed singly or in tetrads, and pollen stainability. Seed-surface pattern is use- ful only for hybrids within the tetraploid group (including L. alata), as these nine taxa are inter- fertile, yield abundant seeds, and are diverse with respect to this character. Absence of developing fruits and low levels of pollen stainability are char- acteristic of hybrids resulting from all other inter- ploid crosses, with the exception of L. curtissii (n = 32) x L. simpsonii (n = 24). Chromosome number and meiotic chromosome behavior are use- ful indicators of natural hybrids when the parents involved have different chromosome numbers or differentiated genomes. ield experience indicates that the occurrence of natural hybrid populations is fairly common wherever two or more species grow together. Hy- brids resulting from crosses between members of the Ludwigia pilosa group are very common. Also, many hybrid combinations have been observed in- Volume 75, Number 3 1988 Peng 999 Ludwigia sect. Microcarpium volving the following species: L. microcarpa, L. curtissii, L. simpsonii, L. palustris (n — 8; sect. Dantia), and L. repens (n — 24; sect. Dantia). Anderson (1948) suggested that disturbed hab- itats often afford conditions suitable for establish- ment of natural hybrids. This does not seem to be the case for Ludwigia hybrids, since the parental species themselves nearly always grow with the hybrids. Figure 21 illustrates the sympatric oc- currence and known cases of hybridization in na- ture for all taxa of sect. Microcarpium except the Cuban endemic L. stricta. Sympatry was deter- mined primarily from personal observation (in 1979, 1980, 1982) and supplemented by field notes of Peter H. Raven and from mixed herbarium col- lections. Voucher information for the suspected natural hybrids is presented below with comments when appropriate. Chromosome numbers of the parents are indicated in parentheses after their names. The sequence of epithets in each formula is alphabetical. HYBRIDS INVOLVING MEMBERS OF LUDWIGIA SECT. MICROCARPIUM Ludwigia alata (n = 24) x L. pilosa (n = 16). U.S.A. FLORIDA: Franklin Co., 38.8 mi. W of jet. US 98 and US 319, Peng 4346 (MO). Wakulla Co., 5 mi. S of Sopchoppy on US 319, Lazor 4984 (FSU, VDB); 1 mi. 5 of Sopchoppy on US 319, Morar 29 (FLAS, CH, MSU, USF). Walton Co., Freeport, Godfrey 57653 (FSU). GEORGIA: Charlton Co., of Folk- ston, Okefenokee Swamp, Camp Cornelia, Jones 22996 (GA). MISSISSIPPI: Hancock Co., along Jordan River S of Kiln, Jones 9539 (MISS). Notes. Peng 4346 from Franklin Co., Florida, was found in a very wide roadside drainage ditch (ca. 10-15 m across) at the edge of a pine forest. Hybrids were very abundant and mixed with a large population of L. alata and a few individuals of L. pilosa. Large populations of L. linifolia and L. microcarpa were also present. This hybrid exhib- ited meiotic configurations of 15-16 bivalents and 10-8 univalents. It had 84% stainable pollen. Ludwigia alata (n — 24) x L. suffruticosa (n = 16). U.S.A. FLORIDA: Hillsborough Co., on E side of FL 581, 2.8 mi. N of FL 582, Peng 4329 (MO). Lake Co., Tavares, Biltmore Her- barium 41704 (DS); vicinity of Eustis, Hitch- cock in 1894 (F), Nash 1154 (NY); 5 mi. SE of Lebanon Station, Kral 7807 (GA, GH, NCU, US, USF). Taylor Co., N edge of US 27 at Fenholloway River Bridge, Nelson 667 (USCH); ca. 20 mi. NW of Cross City, God- frey & Houk 60296 (FSU, MSU, NCU, SMU). Notes. More or less congested infloresence and winged capsules are characteristic of this hybrid combination. The hybrids were often identified as L. alata when they had lax inflorescences. In these cases, however, reduced levels of stainable pollen and intermediate seed surface cell pattern were useful in revealing the hybrids. Peng 4329 showed a meiotic configuration of 15 bivalents and 10 univalents. Ludwigia curtissii (n = 32) x L. linifolia (n = 8). U.S.A. FLORIDA: Monroe Co., Pine Crest, Moldenke 856* (MO, NY). Pasco Co., 1 mi. E of Gowers Corner off US 41, Ray et al. 9932 (USF). Notes. This hybrid is very similar to L. cur- tissii in aspect, but its pollen is not stainable, and its ovaries abort. One putative parent, L. linifolia (Ray et al. 9934, USF), was collected at the same locality as the hybrid (Ray et al. 9932). Ludwigia curtissii (n — 32) x L. microcarpa ). U.S.A. FLORIDA: Martin Co., 4.3 mi. E of Okeechobee and Martin Co. line, on FL 710, at Brady Ranch, Peng 4202 (MO). Notes. The hybrid population was found in a wide swampy depression between the highway and a railroad, growing with both putative parents. The hybrid had a single analyzable metaphase I cell which showed 8 bivalents and 24 univalents. Pollen stainability was 16%. Ludwigia glandulosa subsp. glandulosa (n = 16) x L. pilosa (n = 16). U.S.A. ALABAMA: Covington Co., along Co. Rd. 42, 15 mi. E of Brooklyn, Kral 40992 (FLAS, GH, MO, NCU, NY, US, USF, VDB). GEORGIA: Grady Co., 13 air mi. SW of Cairo, 5 air mi. NE of Concord, Florida, with L. pilosa, Anderson 4044 (MO, FSU). MISSISSIPPI: Jackson Co., Ocean Springs, Demaree 32174 (RSA, SMU); on MS 90, 2 mi. W of US 10 and MS 90, with L. pilosa, Peng 4354-4 (MO). NORTH CAROLINA: Hyde Co., 1.1 mi. N of Scranton Creek on US 264, with both putative parents, Duke 54-232B (NCU); 1.2 mi. N of Scranton Creek on US 264, with both putative parents, Duke 54-276, 54-277, and 54-278 (NCU). 1000 Annals of the Missouri Botanical Garden Ludwigia glandulosa subsp. glandulosa (n — 16) x L. linearis (n = 8). U.S.A. GEORGIA: Long Co., 4.2 mi. SW of jct. of US 301 and 25, and GA 99, on US 301 and 25, Peng 4118 (MO). NORTH CAROLINA: Craven Co., 0.8 mi. N of US 17 on Co. Rd. 1224 (road to Tuscarora), Boufford et al. 21443 (MO), Peng 3740 (MO) Notes. Both hybrid populations were found along with the putative parents. Peng 3740 from Craven Co., North Carolina, was found in a wet drainage ditch about 60-80 cm wide. The popu- lation consisted of ca. 15-20 floriferous individu- als, some of which were even more robust than the putative parents, which grew next to and on either side of the hybrid population. The modal meiotic configuration of this hybrid was 3-5 bivalents and 18-14 univalents. Heteromorphic pairs were sometimes observed. A trivalent was seen in one of the 26 cells studied. The pollen stainability was 2%. Peng 4118, from Long Co., Georgia, consisted of a few scattered individuals. The putative parents were growing nearby on a grassy roadside shoulder. Ludwigia glandulosa subsp. glandulosa (n — 16) x L. sphaerocarpa (n — 16). U.S.A MISSOURI: Butler Co., swamps, Eggert in 1893 (MO). SOUTH CAROLINA: Clarendon Co., shore of Lake Marion, ca. 4.5 mi. SW of St Paul off US 15, Bradley & Sears 3561 (BOON, East Carolina Univ., NCU, WCUH). Notes. Bradley & Sears collection; all four specimens contain a mixture of hybrids and individuals of L. glandulosa subsp. glandulosa. Although evident hybrids between L. sphaero- 3561 is a mixed glandulosa subsp. glandulosa and L. carpa were found in Butler Co., Missouri, speci- mens of L. sphaerocarpa have not yet been col- lected from that state. Ludwigia lanceolata (n = 16) x L. pilosa (n = 16) [Ludwigia X simulata Small]. U.S.A. FLORIDA: Franklin Co., Apalachicola, Chap- man s.n. (F, US). Highlands Co., Bear Point, Lake Childs, Brass 15532 (GH, US). West Florida, Biltmore Herbarium (NY, holotype of Ludwigia X simulata Small). Notes. The holotype of L. x simulata is char- acterized by being densely strigillose throughout and having four-angled or slightly winged capsules, isodiametric seed-surface cells, and pollen shed in tight tetrads. Such a combination of characters clearly suggests L. lanceolata (which is glabrous, has winged capsules, isodiametric seed surface cells, and pollen shed in tetrads) and L. pilosa (which is densely hirsute, has rounded capsules, isodiametric seed-surface cells, and pollen shed in tetrads) as putative parents. Furthermore, L. Xsimulata is comparable to the experimental hybrids obtained from reciprocal crosses between those species. Ludwigia lanceolata (n — 16) x L. suffru- ticosa (n = 16). U.S.A. FLORIDA: Charlton Co., Okefenokee Swamp, Harper 1483 (GH, MO, NCU, NY, US). Hillsborough Co., 1.5- 1.7 mi. S of FL 674, on E side of Taylor Gill Dr., Peng 4324, 4328 (MO) Notes. Both putative parents as well as L. linearis were present in the same drainage ditch where Peng 4324 and Peng 4328 were collected. Peng 4324 showed 16 bivalents in diakinesis and metaphase I cells. One precociously separating bi- valent was occasionally observed. Peng 4324 had 85% stainable pollen. Ludwigia linearis (n = 8) x L. linifolia (n = 8). U.S.A. FLORIDA: Palm Beach Co., along the S side of Co. Rd. 74, 1.5 mi. W of the Turnpike, Palm Beach Popenoe 1957 ( Gardens, Notes. This plant is only 25-30 cm tall and has crowded leaves. It showed features somewhat intermediate between those of L. linearis and those of L. linifolia in its flowers and capsules. The pollen was shed as tetrads, many of which are unstainable. This plant probably is a hybrid between L. linearis and L. linifolia, although the experimental hybrids obtained between these taxa have been more ro- Ludwigia linearis (n — 8) x L. sphaerocarpa (n — 16). U.S.A. ALABAMA: Covington Co., Conecuh National Forest, SW Andalusia, Kral 44732 (ENCB, SMU, VDB). Notes. The plants are very vigorous and florif- erous. One to four vestigial petals are present in most flowers. Seed set is very low or possibly non- existent. Ludwigia microcarpa (n = 8) x L. simpsonii (n = 24). U.S.A. FLORIDA: Charlotte Co., Pun- ta Gorda City, on US 41, ca. 1 mi. S of jet. of US 17 and 41, Peng 4297 (MO). Clay Co., 5 mi. W of Penny Farms on FL 16, Peng 4160 (MO). Collier Co., 4.8 mi. W of Monroe Station, on N side of US 41, Peng 4263 (MO). Volume 75, Number 3 1988 Peng 1001 Ludwigia sect. Microcarpium Sumter Co., Cedar Hammock, 1894, Lewton n. (NY). Notes. Peng 4297 from Charlotte Co., Flor- ida, was found intermixed with both parents. Also occurring here were L. microcarpa, L. repens (sect. Dantia), and the intersectional hybrid be- tween them. Of the nine meiotic metaphase I cells examined from Peng 4297, seven showed 8 bi- valents and 16 univalents, and two showed 9 bi- valents and 14 univalents. In the latter, one of the bivalents separated precociously. Pollen stainability was 4%. Peng 4160, from Clay Co., Florida, was collected from a large population located in a wa- terlogged roadside ditch along the margin of a pine woodland. The hybrids were intermixed with one of the putative parents, L. microcarpa. Across the highway in the similar habitat was another very large, pure population of L. microcarpa. Peng 4160 had 9% stainable pollen. Peng 4263, from Collier Co., Florida, was found in an open pal- metto-cypress forest, with both putative parents growing nearby. One somewhat analyzable meta- phase I cell from this plant showed 8 bivalents and 16 univalents. Pollen stainability was 24%. Ludwigia pilosa (n = 16) x L. sphaerocarpa (n = 16) Before discussing this hybrid combination, a few comments on L. sphaerocarpa are appropriate. This species is quite variable and has a widely scattered distribution (Fig. 22). Populations of L. sphaerocarpa consist of individuals that are ex- tremely varied in overall pubescence, leaf shape and size, fruit size, and density of fruits on branch- es. Three varieties (var. jungens, var. macrocar- pa. and var. deamii) have been recognized pre- viously within this species (Fernald & Griscom, 1935) based on various combinations of the above characters. Study of numerous herbarium speci- mens not available to Fernald & Griscom, however, revealed that correlations between these charac- teristics are not consistent. It is of interest to note that, although seed-surface cell pattern is generally very regular within populations of members of sect. Microcarpium (Figs. 1—3), this is not the case for L. sphaerocarpa (Fig. 4). The seed surfaces are arranged in columnar cells both transversely elon- gate and parallel to the seed length, with the former alignment often predominant in the central part of the seeds (Fig. 4). Seeds with variously oriented surface cells are also seen in some populations. A comparison of the irregular seed surface pattern in L. sphaerocarpa with that of various artificial hybrid combinations (Figs. 6-19) strongly suggests that earlier hybridizations within the interfertile tetraploid group of sect. Microcarpium may have resulted in production of this widespread series of populations that have more or less stabilized in some of their characteristics. Morphological variation in L. sphaerocarpa is further complicated by its frequent natural hy- bridization with L. pilosa (and perhaps with L. ravenii as well, although it would be difficult to distinguish these hybrids from those involving L. pilosa), which has apparently resulted in many hybrid swarms or introgressed populations. These plants are generally neither typical of L. pilosa nor of L. sphaerocarpa and exhibit varying de- grees of intermediacy between the two species. The diagnostic characters for the hybrids include over- all pubescence, leaf shape, bracteole size and lo- cation, sepal shape and size, and color of abaxial leaf venation. Examples of populations of such intermediates are too numerous to cite here. In- stead they have been mapped (Fig. 22). It is of interest to note that some of the hybrid populations occur in central and southern Florida where typical L. pilosa and L. sphaerocarpa are absent; this suggests that physiological characteristics, and thereby ecological tolerances, may recombine into novel adaptive combinations in the hybrids also. Artificial hybrids between L. pilosa and L. sphaerocarpa were synthesized in an experimental greenhouse. Plants of this hybrid combination showed 15-16 bivalents and exhibited the highest level of stainable pollen (98%) among all the tet- raploid hybrids. Ludwigia pilosa (n = 16) x L. suffruticosa (n = 16) [L. capitata B pubens Torrey & A. Gray ]. U.S.A. FLORIDA: Citrus Co., 5 mi. S of Homosassa, Kral 777 1 (FLAS, GH, both mixed with L. suffruticosa). Gadsden Co., along Old Bainbridge Rd. (Rte. 173); 0.5 mi. NW of Ochlockonee River bridge, NW of Tallahas- see, Anderson 7486 (FSU). Seminole Co., W shore of Prairie Lake, Schallert 16009, 28652 (S). GEORGIA: McIntosh Co., on Sapelo Isl., ca. 1.4 mi. WNW of S tip of Blackbeard Isl., Duncan 20445 (DUKE, F, GH, NCU, SMU, TEX, US, USF). Wayne Co., near Jessup, Biltmore Herbarium 4 174* (US). GEORGIA (7): Baldwin Herbarium (NY, holotype of Lud- wigia capitata Michaux 8 pubens Torrey € A. Gray, mixed with L. suffruticosa Walter). SOUTH CAROLINA: Darlington Co., Hartsville, Smith 44 (NCU). Georgetown Co., North San- tee, Radford 28678 (NCU, NY, VDB). 1002 Annals of the Missouri Botanical Garden 7 i I FIGURE 22. dots) Ludwigia polycarpa (n — 16) x L. sphae- rocarpa (n = 16). U.S.A. INDIANA: Starke Co., border of Bass Lake, 5 mi. S of Knox, Kriebel 5715 (SMU); SW corner of Bass Lake, Friesner 16306 (CAS, NY). Notes. Ludwigia polycarpa, which occurs in the central Midwest (Fig. 20), is effectively isolated geographically from all other species of sect. Mi- crocarpium, with the exception of L. sphaerocar- pa. Where the two species come into contact, hybridization occurs. The following is a list of collections of hybrids for which the identity of the putative parents is not certain: Distribution of Ludwigia pilosa (shading), L. sphaerocarpa (stippling), and their natural hybrids ?Ludwigia linearis (n = 8) x L. ravenii (n = 16). U.S.A. NORTH CAROLINA: Duplin Co., 3 mi. E of Sarecta, Beal 3674 (NCSC). Notes. This is a much-branched, densely vil- lous plant with sublinear leaves and small, elongate- pyramidal ovaries that abort after anthesis. The pubescence suggests that either L. pilosa or L. ravenii is involved in the parentage. The aborted ovaries indicate that hybridization involved one of the above species and a taxon outside of the in- terfertile L. pilosa group. The elongate ovaries and narrow leaves indicate clearly that either L. linearis or L. linifolia is the other parent. Floral characters as well as geographical distribution of these species, however, indicate that Beal 3674 Volume 75, Number 3 1988 Peng 1003 Ludwigia sect. Microcarpium from Duplin Co., North Carolina, is probably a hybrid between L. linearis and L. ravenii. Ludwigia microcarpa (n — 8) x L. curtissii (n = 32)/L. simpsonii (n = 24). U.S.A. FLORIDA: Flagler Co., 6 mi. E of Co. line, Hwy. 28 near Andalusia, West in 1940 (FLAS). Lake Co., 7 mi. SW Okahumpka, Kral 7611 (FLAS, FSU, GH). Notes. Kral 7611 represents a mixture of sev- eral species and hybrids. Specimens deposited in GA, GH, US, VDB are L. microcarpa, whereas the specimen at FSU contains both the hybrid and either L. curtissii or L. simpsonii, the separation of which is difficult, since mature capsules are not present. ?Ludwigia pilosa (n = 16) x L. suffruticosa (n = 16). U.S.A. FLORIDA: Jackson Co., Ochee- see Lodge Landing S of US 90, near Sneads, Jones 23569 (GA). Co. unknown, S Florida, Chapman Herbarium (NY). GEORGIA: Lee Co., near US 19, ca. 4 mi. S of Smithville, Moran 2551 (GA) Notes. These specimens are less pubescent than typical L. pilosa X L. suffruticosa hybrids. They are for the most part densely strigillose only in the branched and somewhat lax inflorescence and may represent backcrossed populations or seg- regates of advanced generation of the hybrid L. pilosa X L. suffruticosa. HYBRIDS BETWEEN MEMBERS OF LUDIVIGIA SECTS. MICROCARPIUM AND DANTIA Hybrids between members of sects. Microcar- pium and Dantia are easy to recognize; members of sect. Microcarpium are erect plants with alter- nate leaves, whereas plants belonging to sect. Dan- tia are [oen and have li gua MUS mater mediacy in these tw g between the two sections. Ludwigia arcuata (n = 16; sect. Dantia) x L. pilosa (n = 16). U.S.A. ALABAMA: Mobile Co., Audubon Bird Sanctuary, Dauphin Isl., Der- amus D643 (DS, UNA). Notes. Hybrids of this combination were syn- thesized in the experimental greenhouse. They showed a modal meiotic configuration of 12-15 bivalents and 8-2 univalents; 1-2 trivalents were sometimes observed. The high degree of chromo- some pairing observed between species of the two sections was quite unexpected. This artificial hy- brid, however, had only 38% stainable pollen. Since selfing in the hybrid is physically impossible, as in L. pilosa, artificial pollination was attempted in order to investigate seed set. None of the pollination attempts yielded any seed set, even though some seeds were observed in mature capsules of the natural hybrids. Ludwigia curtissii (n — 32) x L. repens (n — 24; sect. Dantia). U.S.A. FLORIDA: Glades Co., 4.4 mi. SE of jct. of FL 29 with US 27, with both parents, Raven 18680 (MO). Lee Co., on US 41, 5 mi. N of Ft. Meyers, with both parents, Dille & Dille 379 (MO). Ludwigia curtissii (n = 32) and L. simpsonii (n — 24) hybridize with. L. repens in nature. Unless chromosome numbers are counted, it is unlikely that one would be able to distinguish between these two hybrid combinations morpho- logically. Notes. Ludwigia curtissii (n — 32)/L. simpsonii (n — 24) x L. repens (n — 24; sect. Dantia). U.S.A. FLORIDA: Charlotte Co., 12 mi. S of Punta Gorda, Kral 18058 (VDB). Lee Co., 5 mi. S of Bonita Springs, Crevasse in 1940 (FLAS). Manatee Co., Bradentown, Cuthbert in 1926 (FLAS). Wakulla Co., between Hwys. 365-367, N of Spring Creek, Lazor 4561 (NCU). Ludwigia repens (n = 24; sect. Dantia) x L. simpsonii (n = 24). U.S.A. FLORIDA: Char- lotte Co., Punta Gorda City, on US 41, ca. 1 mi. S of jet. of US 17 and 41, Peng 4296 (MO). Notes. This hybrid grew intermixed with both putative parents along a roadside field. It formed 0-1 bivalent(s) and 48-46 univalents in metaphase I cells. This result corroborates an earlier report by Schmidt (1967) [FLORIDA: Glades Co., 8.4 mi. SE of jct. of FL 29 on US 27, Raven 18849 (DS)]. Ludwigia glandulosa subsp. glandulosa (n — x L. palustris (n = 8; sect. Dantia). U.S.A. ARKANSAS: Clark Co., maree 16120 (DS, GA, GH, MO, NY, SMU, TENN). NORTH CAROLINA: Co. unknown, stag- nant water just S of Upper Littel River on US 401, Lloyd in 1962 (MO). Johnston Co., 4.9 mi. W of NC 210 and US 70, on US 70, E of Clinton, Ahles 59736 (NCU), 61803 (DS, 1004 Annals of the Missouri Botanical Garden NCU, SMU). oKLAHOMA: Johnston Co., Devil's Den, 4.6 mi. NW of Tishomingo, Crutchfield 2882 (LL). VIRGINIA: Fluvanna Co., just Š of Rt. 696, 1 mi. S of Rt. 250, Diggs & Diggs 353 (NCU). Notes. Plants of Lloyd in 1962 from North Carolina showed 24 univalents in meiosis (Raven, pers. comm.). Schmidt (1967) reported “at most three weakly joined bivalents" in the hybrid plant from North Carolina [Harnet Co., 5.1 mi. S of Lillington, Lloyd 1022 (DS) Artificial hybrids re- cently synthesized, however, showed higher chro- mosome associations in meiosis, in the range five to eight bivalents modally. Ludwigia microcarpa (n = 8) x L. palustris (n = 8; sect Dantia). U.S.A. FLORIDA: Franklin Co., 41.7 mi. W of jet. of US 98 and 319, with both putative parents, Peng 4349 (MO). Hamilton Co., off 1-75, ca. 1 mi. N of Colum- bia Co. line, Bowers & Wofford 71-550-F (TENN, a mixture of /. L. micro- carpa, and L. microcarpa X L. palustris). Lake Co., 3.7 mi. S of Mascotte city limit, on Co. Rd. 33, with L. microcarpa and L. pa- lustris, Peng 4167 (MO). GEORGIA: Camden Co., 6.6 mi. S of Woodbine on US 17, with L. microcarpa and L. repens, Raven 18704 (MO). NORTH CAROLINA: Jones Co., near NC 41, 0.7 mi. E of Taylor's Corner, Radford 37152 (NCU). repens, Notes. Peng 4349 from Franklin Co., Florida, was collected from a large population growing in- termixed with both putative parents along a swampy ditch. Peng 4167 from Lake Co., Florida, was found growing with both putative parents in a nar- row ditch. Pollen stainability was 5% in Peng 4349. Artificial hybrids of this combination were synthe- sized and showed 2-5 bivalents and 12-6 univa- lents in meiotic metaphase I. Observations from experimental hybridization indicate that hybrid combinations of L. microcarpa (n = 8) x L. palustris (n = 8; sect. Dantia) and L. microcarpa (n = 8) x L. repens (n = 24; sect. Dantia) are prostrate herbs with similar leaf shape and minute flowers that abort after anthesis. In L. microcarpa X L. palustris, the flowers are apetal- ous as in both parents, and the phyllotaxy is in- termediate; the plants have alternate, opposite, and subopposite leaves. By contrast, in L. microcar- a X L. repens the leaves are always opposite as in L. repens, and some of the flowers have one to four vestigial petals. The apparent dominance of the traits of the hexaploid parent over those of the diploid parent is probably due to a multiple dosage of genetic information from the hexaploid. In two instances, Bowers & Wofford (V DB) and Raven 18704 (MO), the hybrids were collected along with L. microcarpa and L. repens; L. lustris was not observed locally. Although these plants, which have aborted ovaries, might have been considered as hybrids between L. microcarpa and L. repens on this basis, detailed examination of their flowers and leaves reveals that they are L. microcarpa X L. palustris. Ludwigia polycarpa (n = 16) x L. palustris (n = 8; sect. Dantia). U.S.A. KENTUCKY: Bal- lard Co., at intersection of Kelley Branch Creek Rd. and KY 473, Athey 1158 (NCU, NY, VDB). onto: Erie Co., in bottom of South Quarry at N edge of town on Kelley's Isl., Stuckey 7400 (PH). Notes. Although one of the putative parents, L. polycarpa, has not been recorded from Ballard Co. in extreme southwest Kentucky, it was col- lected from adjacent Massac Co., Illinois. Ludwig- ia palustris is common in Ballard Co., Kentucky. Plants of Athey 1158 are comparable to artificial hybrids between these species synthesized in an experimental greenhouse. In summary, geographical isolation and self-pol- lination are the primary factors limiting natural hybridization in species of sect. Microcarpium. For example, L. microcarpa, an extreme selfer, has been hybridized successfully with most species of sect. Microcarpium in the greenhouse, and the resulting hybrids were vigorous. It grows sym- patrically with many other species (Fig. 21), but natural hybrids have been found only with L. simp- sonii, L. curtissii, and members of sect. Dantia. In general, however, natural hybrids in sect. Microcarpium are frequently found wherever two species occur together. This is particularly so for plants in the tetraploid group (including the hexa- ploid L. alata; Fig. 21), which were often found intermixed with their putative parents. Especially evident in L. pilosa and hybrid pop- ulations involving L. sphaerocarpa were the effects of backcrossing and introgression, which may pro- vide the hybrid populations with increased evolu- tionary flexibility, thus enabling them to grow in areas where neither of the parents are found. As outlined above, natural hybridization is not limited to species within sect. Microcarpium. At Volume 75, Number 3 1988 Peng 1005 Ludwigia sect. Microcarpium least seven hybrid combinations, some of which occur commonly in nature, have been found be- tween members of sect. Microcarpium and sect. Dantia. Most of these have also been synthesized in the experimental greenhouse. Most intersection- al hybrids, however, do not set seed, although they are usually very vigorous and appear to compete well with their parents. Once established, these sterile hybrids may be able to persist in a given location due to their perennial habit. New colonies may be established vegetatively if entire parents or fragments are transported to suitable habitats, most likely by water. SPECIES RELATIONSHIPS AND EVOLUTION Among the four diploid species, L. microcarpa is quite distinct from L. linearis, L. linifolia, and L. stricta. Plants of L. microcarpa are small herbs with spatulate or obovate-spatulate leaves, minute apetalous flowers from which the pollen is shed singly, and short, tiny capsules; whereas the other three species have linear leaves, somewhat showy flowers with four petals from which the pollen is shed in tetrads, and elongate capsules. They appear to be relatively closely related to one another. eight taxa in the tetraploid (n — 16) group, including L. alata (n — 24), are apetalous. Their leaves range from linear-lanceolate to lanceolate, elliptic, or oblanceolate. They differ from each other in a number of characters, including capsule morphology, seed-surface pattern, pubescence, and the way in which pollen is shed. Although L. alata and L. simpsonii are both hexaploids (n — 24), morphological characters sug- gest that they are not closely related. Ludwigia alata, which has winged capsules, is most similar . lanceolata of the tetraploid group, although it can be distinguished by having pollen shed singly, seed surfaces consisting of columnar cells elongate transversely to the seed length (Fig. 3), and by being modally outcrossing. Ludwigia lanceolata sheds its pollen in tetrads, has isodiametric seed- surface cells (Fig. 1), and is autogamous. By contrast, hexaploid Ludwigia simpsonii re- sembles octoploid L. curtissii (n — 32). These two species can be distinguished relatively consistently only by the size of their mature capsules, although even this character is not always reliable. Both exhibit a specialized type of capsular dehiscence unique in sect. Microcarpium. Moreover, L. simp- sonii and L. curtissii, along with the diploid L. microcarpa, are the only species in the section with spatulate or obovate-spatulate cauline leaves. There appears to be a close relationship between L. microcarpa and the L. curtissii complex. Results from the crossing program and chro- mosome analysis of the artificial and natural hy- brids confirm these general observations and pro- vide additional evidence for some of th y relationships discussed below. RELATIONSHIPS AMONG THE DIPLOID SPECIES (L. LINEARIS, L. LINIFOLIA, L. STRICTA, L. MICROCARPA) The presence of a quadrivalent during meiosis of reciprocal hybrids of L. linearis x L. linifolia indicates that the genomes of the strains hybridized differ by a reciprocal translocation. Although re- ciprocal translocations are very frequent in the tribe Onagreae, they are rare in the remainder of the family; their presence here indicates that chro- mosomal repatterning has occurred between the strains hybridized. The artificially produced F, hy- brids were very vigorous and set abundant seeds, although many of the F, plants were either weak or inviable. Presumably L. stricta will show a com- parable degree of differentiation. The hybrids between L. microcarpa and either L. linearis or L. linifolia showed very few (0-3) bivalents in meiosis. The bivalents sometimes ap- peared to be held together by matrix connections rather than by chiasmata; some of them were het- eromorphic, and the chromosomes did not always line up in the equatorial plane. Taken together, these phenomena indicate that L. microcarpa has a diploid genome essentially different from that of either L. linearis or L. linifolia, and presumably L. stricta as well. RELATIONSHIPS AMONG THE TETRAPLOID TAXA (L. GLANDULOSA SUBSP. GLANDULOSA, L. GLANDULOSA SUBSP. BRACHYCARPA, L. LANCEOLATA, L. PILOSA, L. POLYCARPA, L. RAVENII, L. SPHAEROCARPA, AND L. SUFFRUTICOSA) This is a group of eight diverse and morpholog- ically well-delimited taxa. Artificial hybridization between any two species nearly always resulted in vigorously growing individuals with nearly com- plete chromosome pairing, high levels of pnl pollen, and abundant seeds. Many vigorous F, plants were raised that exhibited various degrees of in- termediacy between the parents. This group is in- terpreted as representing an assemblage of inter- fertile tetraploid species that have two genomes in 1006 Annals of the Missouri Botanical Garden common and thus represents a homogamic com- plex. ORIGIN OF THE TWO GENOMES IN THE TETRAPLOID SPECIES Most of the tetraploid species have been crossed with each of the three diploid species included in this study in order to assess whether one or more of the extant diploid species has been involved in their formation. Ludwigia linearis and L. linifolia were shown to share a similar genome that has undergone some chromosomal repatterning. Hy- brids between either of these diploid species and any tetraploid species produced 3-7 bivalents in meiosis. Hybrids between L. microcarpa and the tetraploid species showed 1-6 bivalents in meiosis. Two lines of evidence suggest that the chromosomal pairing in these hybrids is the result of pairing of chromosomes between the diploid and the tetraploid rather than pairing between the two genomes pres- ent in the tetraploid. First, hybrids between L. linifolia (9; n = 8) and L. lanceolata (8; n = 16) were themselves tetraploid, modally forming 8 bi- valents and 16 univalents in meiosis. This tetraploid chromosome number apparently resulted from the union of an unreduced egg from the diploid parent with a normal sperm nucleus from the tetraploid. This unexpected result strongly suggests that the 8 bivalents observed represent paired chromosomes from the duplicated L. linifolia genomes, while the 16 univalents are the chromosomes from the two genomes of the tetraploid parent, which are suffi- ciently different to remain unpaired. Pairing be- tween chromosomes of L. linifolia and those of the tetraploids is evidently precluded by prefer- ential pairing between the two sets of chromosomes derived from L. linifolia. econd, when L. glandulosa (n = 16) was crossed with L. alternifolia (n = 8), a less closely related species belonging to sect. Ludwigia, the resulting hybrids exhibited 0-1 bivalent(s) in mei- osis; the one bivalent that was occasionally ob- served was only loosely associated. This lack of pairing also suggests that the tetraploid species are alloploids with two unlike genomes. As noted earlier, two distinct genomes appear to be represented among the diploid species; one shared by L. linearis, L. linifolia, and probably L. stricta, and another found in L. microcarpa. Hybridization between any of these species and the tetraploid taxa consistently results in F, off- spring that show on -7 bivalents, some of them heteromorphic, in meiosis. This strongly suggests that neither of the two genomes present in the tetraploids was derived from an existing diploid. ORIGIN OF GENOMES IN THE HEXAPLOID LUDWIGIA ALATA Nearly all of the reciprocal hybrids between the hexaploid L. alata and the tetraploid species ex- hibited a modal meiotic configuration of 16 biva- lents and 8 univalents. This suggests that they share two genomes: the 16 chromosomes of the tetraploid species pair with their homologues pres- ent in the genome of the hexaploid species, while the additional eight chromosomes in L. alata main unpaired and appear to represent a third genome present in L. alata. Ludwigia alata was also crossed with the diploid species in an attempt to assess whether this “third genome”” is in fact homologous to one of those present in the existent diploid taxa. Hybrids be- tween L. alata and either L. linearis or L. linifolia typically exhibit a configuration of 1-2 trivalents and 4—6 bivalents with the rest of the chromosomes univalents. Again, heteromorphic pairs are com- mon here. These results are similar to those ob- served in hybrids between L. linearis and L. lin- ifolia and the tetraploid taxa. Since the two genomes of the tetraploid species appear to be present in L. alata, it is presumed that the bivalents are formed between chromosomes derived from the parental taxa. By contrast, meiosis in hybrids between L. microcarpa and L. alata shows a high degree of chromosome pairing, with a modal configuration of one quadrivalent, 7-10 bivalents, and a cor- responding number of univalents. The maximum amount of pairing seen was 1111 + 101 or 1III + 10I + 8I. These chromosome associations involve bivalents of normal appearance, either rings or rods, a situation that contrasts with the sorts of loosely associated bivalents characteristic of hy- brids between L. microcarpa and the tetraploid taxa. The higher level of chromosome pairing (more than eight bivalents) and the normal appearance of the pairs suggest that the hexaploid L. alata may have been derived following hybridization be- tween L. microcarpa and one of the tetraploids. Indeed L. alata is similar to L. microcarpa in that (1) its seed surface consists of columnar cells trans- versely elongate to the seed length, and (2) its pollen grains are shed singly. These characters are rarely found among members of the tetraploid species group, with which L. alata shares two genomes. Volume 75, Number 3 1988 Peng 1007 Ludwigia sect. Microcarpium ORIGIN OF GENOMES IN THE LUDWIGIA CURTISSII COMPLEX The L. curtissii complex consists of L. simp- sonii (n = 24) and L. curtissii (n = 32). Hybrids between the two species consistently reveal 24 bivalents and 8 univalents in meiosis. This indicates that chromosomes of the three genomes in L. simp- sonii pair with homologues in L. curtissii, leaving the eight additional chromosomes in this species unpaired. Naturally occurring intersectional hy- brids between L. repens (n — 24; sect. Dantia) and L. simpsonii produce a meiotic configuration of 48 univalents or 46 univalents and two loosely associated chromosomes. This strongly suggests that the three genomes in L. simpsonii are distinct from each other and not homologous with any of the genomes present in L. repens. For ease of discussion, letters will be used to designate distinct, nonhomologous genomes. Since the five genomes present in Ludwigia sect. Dantia have been designated as A, B, C, D, and E (Schmidt, 1967), the genomic formula GGHHII will be used for L. simpsonii and FFGGHHII for L. curtissil. The natural hybrid L. microcarpa x L. simp- sonii has a meiotic configuration of 8 bivalents and 16 univalents. The eight chromosomes of L. mi- crocarpa thus have 8 homologues in L. simpsonii, and its genome is designated as GC. A configuration of 8 bivalents and 24 univalents is expected in meiotic cells of the hybrid L. curtissii (FFGGHHII) x L. microcarpa (GG). In a natural hybrid of this combination, the single analyzable cell did indeed show this configuration. In an ar- tificial hybrid, however, more than 8 bivalents have been observed: five cells exhibited 14-15 bivalents and 12-10 univalents. Some intergenomic inter- action must have led to the observed configurations. In the meiosis of L. curtissii (FFGGHHII) x L. linifolia (n — 8), a modal configuration of 8 bivalents and 24 univalents was observed. The enome common to L. linearis and L. linifolia is designated FF, since it differs from that present in L. microcarpa (GC). As in the case of L. curtis- sii X L. microcarpa, however, a few meiotic cells in L. curtissii X L. linifolia exhibited more than 8 bivalents; up to 12 have been observed. Figure 23 summarizes the chromosomal ho- mologies in the L. curtissii complex. The hexaploid L. simpsonii appears to have been derived from three different diploid lines. Morphological as well as chromosome pairing data indicate that the dip- loid L. microcarpa has been involved in the for- mation of L. simpsonii. Based on morphology, it seems unlikely that a member of the L. linifolia/ . linearis lineage could have been a parent of L. simpsonii. Nevertheless, it will be necessary to examine meiosis in the artificial hybrid between L. simpsonii and L. linearis or L. linifolia. Although plants of L. simpsonii X L. linearis were available, my attempts to study meiosis in them were unsuc- cessful. The octoploid L. curtissii was probably derived following hybridization between a diploid similar to L. linearis or L. linifolia with the hexa- ploid L. simpsonii, based on morphological data and crossing relationships. To study the genetic relationships among plants from the tetraploid group (including the hexaploid L. alata) and the L. curtissii complex, 11 artificial hybrids were produced, 5 of which have been stud- ied cytologically. The number of bivalents observed ranged from 4 to 13; a few trivalents and quad- rivalents were also frequently seen (Table 15). Chromosome associations between the polyploids of the tetraploid group and the three diploid species, on the one hand, and those observed between the L. curtissii group and the diploid species, on the other hand, are consistent with these results. Following the differentiation of diploid species of sect. Microcarpium, some have evidently be- come extinct. Natural hybridization between the diploid lineages followed by polyploidy has played a major role in the evolution of this group. Postzy- gotic genetic barriers do not exist between most of the extant species in the section. Rather, the major limiting factor to natural hybridization appears to be the modally autogamous breeding system of most species. Geographical isolation is important only with respect to L. polycarpa, which is dis- tributed well to the north of nearly all the other ct £g a. Natural hybridization is prevalent within sect. Microcarpium, and hybrids often occur in more or less undisturbed habitats where the parental taxa also grow. All interploid hybrids are sterile except for crosses between L. alata (n = 24) and the tetraploid species (n = 16), and crosses between L. curtissii (n = 32) and L. simpsonii (n = 24). Even sterile hybrids can persist and form large colonies, at least locally, and compete effectively with their parents, because of strong vegetative reproduction by means of stolons. Natural hybrids are especially common among members of the tet- raploid group (including L. alata) and are nearly always vigorous and fertile. Particularly complex is the pattern of variation in the tetraploid L. sphaerocarpa, which is ap- parently comprised largely of widespread stabilized 1008 Annals of the Missouri Botanical Garden Y L. microcarpa (n-8; GG 8 Tr + 24 I 8 Tr 16 I L. simpsonii 24 T * (n-24; GGHHII) 8I ` ` ` b N ( 0- L. curtissii (n=32; FFGGHHII) L. linearis/L. linifolia (n=8; FF) FIGURE 23. Chromosomal homologies in the Ludwigia curtissii complex. Dotted line RA artificial hybrids were obtained, but meiotic analysis was not successful. Illustrations of the plants are drawn to sca hybrid populations that exhibit a combination of characters distinguishing them from other taxa. As in the evolution of Epilobium in New Zealand (Raven & Raven, 1976), recombination of genetic information from somewhat differentiated popula- tions followed by maintenance of well-adapted ge- netic strains by a combination of autogamy and vegetative reproduction appears to have played a central role in the evolution of the polyploid mem- bers of Ludwigia sect. Microcarpium. LITERATURE CITED ALEX XANDER, M. P. Differential staining of aborted and nonaborted Pellen. Stain Technol. 44: 117-122. ANDERSON, "s : 1l- 1948. Hybridization of the habitat. Evo- lutio 9. BROWN, D. 1967. Pollen morphology of the Onagra- ceae. Rev. Paleobot. Palynol. 3: 163-180. Cooke, W. 1 The Coastal Plain. Pp. 19-54 Physical Geography of Georgia. Geol. Surv. etn Bull. 42: 1-189. Duke, J. A. 1955. Distribution and speciation of the genus m in North Carolina. J. Elisha Mitchell Sci. Soc. 71: 255-269. Eype, R. H. 19 Reproductive structures and ev lution in Ludwigia (Onagraceae). I. Andyoochum, placentation, merism. Ann. Missouri Bot. Gard. 64 ) 55. . 1978. Reproductive structures and evolution in Ludwigia (Onagracea 2, Me Fruit and seed. Ann. Miscouri Bot. Gard. 65: Volume 75, Number 3 1988 Peng 1009 Ludwigia sect. Microcarpium . Reproductive structures and evolution in Ludwigia (Onagraceae). III. Vasculature, nectar- ies, conclusions. Ann. Missouri Bot. Gard. 68: 379- 412. J. T. Morcan. 1973. Floral structure and evolution in Lopezieae (Onagraceae). Amer. J. Bot 60: 771-787. FERNALD, M. L. & L. Griscom. 1935. Three days of tanizing in southeastern Virginia. Rhodora 37: 167- GLEASON, H. A. & A. CRONQUIST. 1964. The Natural Geography of Plants. Columbia Univ. Press, New ork. GRANT, V. 1952. e of ba hybrid Gilia mil- lefoliata x _achillae efolia. riations in meiosis a by nutritional and genetic conditions. Chromosoma 5: 372-390. Grecory, D. P. & W. M. Kie. 1960. Investigations Ona of meiotic chromosomes of six genera in the graceae. iu 4: 505- HanLaN, J. R. & J. M. J. DEWET. 1975. On Ó Winge and a prayer: the origins of polyploidy. Bot. Rev. (Lancaster) 41: 361-390. Hunt, C. B. 974. Natural Regions of the United States and Canada. W. H. Freeman & Co., San Francisco. JACKSON, R. C. 1973. Chromosomal evolution in Hap- lopappus gracilis: a centric transposition race. Evo- lution 27: 243-256. Jones, A. G. Environmental effects on the per- yis of stainable and presumed normal ord in Aster (Compositae). ere J. Bot. 63: 657-66 KURABAYASHI, 1, M., H. Lew x P. H. Raven. 1962. A A mer. J. Bot. 49: 1003-1026. LEE, S. 1978. A E ls study of the functional significance of angiosperm pollen. Syst. Bot. 3: 1- MCCOLLUM, G. D. 1958. Comparative studies of chro mosome pairing in natural and induced istud Dactylis. Chromosoma (Berl.) -60 Munz, P. A. 44. Studies in Oe ae— XIII. The American specigs of Ludwigia. Bull. Torrey Bot. Club 71: 5. 1965. Onagraceae. N. Amer. Fl. II. 5: 1- 278. IT D.C. 1922. The Atlantic Coastal Plain ele- t in the flora of the Great Lakes. Rhodora 280: 7- 88. ` I. 1984. Ludwigia ravenii (Onagraceae), a new species from the Coastal Plain of the sout eastern United y odio Syst. Bot. 9: 129-132. 198 ew combination of Ludwigia sect. Microcarpium (Ona O ceae). Ann. Missouri Bot. d 490. Gard. 3 systematics and evolution of Ludwigia pes sakra qa (Onagraceae). Ann. Missouri Bot. Gard. (in press). OBE. 1987. Capsule wall anatomy in relation to capsular dehiscence in Ludwigia sect. epee Por. (Onagraceae). Amer. J. Bot. 74 1102- Im p J. ]. SKVARLA, P. H. RAVEN & J. Pac IC 1983 DAL Jus Rwiwoormim. T sect. Myrtocarpus s. lat. (Onagraceae). Ann. Missouri Bot. Gard. 66: 893-896. jd Systematics and evolution of Ludwigia carpus sens. lat. (Onagraceae). Ph. Disenatio. Washi ington University, St. Louis, Mis- ouri, M. ZARDINI. 1987. The systematics and evolution of Ludwigia sect. Myrtocarpus sensu lato (Onagraceae). Monogr. Syst. Bot. Missouri Bot. Gard. 19: 1-120. Raven, P. H. 1963. The Old World species of Lud- wigia (including Jussiaea), with a synopsis of the genus perdi Reinwardtia 6: 327-427 19 A survey of oo biology in Onagraceae. EN Zealand J. Bot. 17: 575-593. T vEN. 1976. The i: Epilobium in eterne a Mie aeu and evolutionary study. "N Zealand Dept. Sci. Industr. Res. Bull. 216: 1- "e W. Tar. 1979. Observations of chro somes in Ludwigia pee. Ann. Missouri bo. d. 66: 862-879 SCHMIDT, CoL. pu E biosystematic B of Lud- ja sect. a (Onagraceae) Ph.D. Thesis. Stanford E Sanford eaten SKVARLA, J. J., P & J. PRAcLOWsKI. 1975. The sidus E e s Eis in Onagraceae. Amer. J. Bot. 62: 6-35. ——. 1976. Ultrastructural survey of Onagraceae pollen. Linn. Soc. Symp. Ser. 1: 447-419. , W. Cuissoe & M. SHARP. 1978. An learn study of viscin threads in Onagraceae . Pollen & Spores 20: 1-14 Hybridization and the Flora of the British Isles. Academic Press, L Tinc, W. 1 . Pollen morphology ot Onagraceae. Pollen & Spores 8: 9-3 Tope, H. & P. H. Raven. 1986. Evolution of en, a . J. Bot. 73: eia anthers i in Onagraceae. Amer UHL, x H 1976. C hromosomes, hybrids, and ploidy Sedum cremnophila and Echeveria linguifolia (Crassulaceae). Amer. A Bot. 63: 806- WaLTERS, M. S. 1954. study of pseudobivalents in in meiosis T two € hybrids of Bromus. Amer J. Bot. 41: 160- RECONSTRUCTIONS OF Greg J. Retallack? and David L. Dilcher? SELECTED SEED FERNS' This Paper is Dedicated to the Memory of Sergei V. Meyen ABSTRACT Seed ferns (pteridosperms) make up a Lidl, Ue qa group of dec 8 gymnosperms. Our attempts to reconstruct these extinct plants here summarize research o ny years on the best-known seed ferns. We have named each reconstructed plant after its best-preserved ils ii because these are the most reliable Earliest Late Carboniferous fused 320 ln FA old) Lagenostoma lomaxii is reconstructed as a bushy understory shrub in swamps of arborescent lycopods. Latest Late Carbon Pr sss (about 296 million years old) Pachytesta illinoensis was a tree probably growing on elevated and nutrient-rich areas in and around permanently waterlogged swamps of marattiaceous tree ferns. Pachytesta illinoensis had larg d le see probably dispersed by insects. Its fleshy ovules may have been dispersed by large a amphibians, af a sh. Another seed fern of these latest Carboniferous swamps, Callospermarion pusillum, is reconstructed as an pa successional scrambling vine. Its pollen probably was dispersed by wind, and its numerous small seeds scattered widely by wind and ofi intermontane valleys in the southern Vehicle Large air chambers in its roots enabled it to grow in waterlogged soils. i a tree of lowland mixed conifer—broadleaf forest in a subtropical, seasonally wet paleoclimate. Its ovules were enclosed in berrylike cupules, which may have been pollinated and dispersed by small animals. From these examples, it is apparent that seed ferns were exceptionally diverse broadleaf plants which occupied a variety of niches now occupied by angiosperms. Reconstructions of extinct plants from dispersed drawings presented here are visual expressions of fossil organs have not been attempted to the same some of these hypotheses for especially well-known extent as restoration of vertebrate fossils. Presum- seed ferns. We intend these drawings to be working ably this is because of the modular construction of hypotheses of reconstruction in the same way as plants, because of their various deciduous organs, the written accounts on which they are based. and because of the variety of ways and places in Hypotheses expressed in this way are understood which fossil plants are preserved. Nevertheless, more readily than pages of scientific text. By the many hypotheses have been published concerning same token, however, such drawings also make which parts of fossil plants belong together. The ^ mistakes of interpretation more obvious. A further e reconstructions have been assembled over so many years and with such varied assistance in the field, E collection and with ideas, that it is not possible to goe empi all those who have helped. We espec ” y hav e of. Drs. W. D. D le (Smithsonian Institution, Washington) , oyle (bna of Hara po KR. E. Gould ine jn Brisbane) , T. M. Harris (University of Reading England), C. R. Hill (British Museum Natural History, Londo n), S. V. Meyen (Geological Institute, Moscow), G. KE Ro thwell Fun on Athens), and T. N. Taylor (Ohio State University, Columbus) . noun and | S.F. Grants DEB7910720 and BSRES 16657 to D. L.D. and EAR8206183 to G.J.R. * Department of Geology, University of Oregon, Eugene, Oregon 97403, U.S.A. * Department of Biology, Indiana University, Bloomington, Indiana 47405, U.S.A. ANN. Missouni Bor. GARD. 75: 1010-1057. 1988. Volume 75, Number 3 1988 Retallack & Dilcher 1011 Seed Ferns difference between written and illustrated recon- structions of fossil plants is the difficulty of ex- pressing uncertainty and presenting detail in a drawing. With special regard to this problem, the poses of various organs have been selected care- fully, and enlargements and insets have been used liberally both to reveal detail and obscure uncer- tainties. The plants reconstructed here are those re- garded traditionally as “seed ferns,” a group of plants being increasingly recognized as an evolu- tionary grade rather than a clade. This is not to say that the terms “seed fern" or “pteridosperm”” should be abandoned, any more than the term "dinosaur," another often loosely defined assem- blage, should be discontinued. The discovery that these fernlike fossil leaves belonged to seed-bearing plants reshaped understanding of the ire ria relationships between seed plants and [ (Potonie, 1899; Oliver & Scott, 1904), Ne rim importance to understanding vascular plant evo- lution has remained undiminished. The first seed plant and the ancestors of m o gym sperms were probably seed Tun (Rothwell, 1982) Seed ferns or allied plants remain most likely ances- tors of angiosperms (Dilcher, 1979; Retallack & Dilcher, 1981; Doyle & Donoghue, 1986). The various organs and preservational styles of each of these fossils have separate botanical names, following accepted paleobotanical nomenclature. Our gathering together of various organs and names for different parts of the same plant should not be construed to mean that each part is equally defin- itive of the whole plant. A fossil species of root, for example, may have belonged to several different species distinguishable among reproductive struc- tures. It is likely that different organs of plants have evolved at different rates. In gathering to- gether these names we merely imply that there once existed a plant for which each of these names is appropriate for its various fossilized parts. Our choice of a single name for each recon- structed plant does not follow the /nternational Code of Botanical Nomenclature (Stafleu, 1978) in its rules of priority, because there are indications in the code that these do not apply between form genera for different kinds of plant parts. In the case of the present compilation, strict adherence to priority would result in naming eight of the plants reconstructed here from leaves, one from wood, and one from a root. Instead, we use the name of the best-preserved ovular fructifications consis- tently as the name for these reconstructed seed plants (following Retallack, 1980a, and Anderson & Anderson, 1985). Modern seed plants are clas- sified mostly according to their ovulate fructifica- tions, and a new suprageneric classification of seed plants based on ovular fructifications has recently been proposed by Meyen (1984). We do not agree with all of the criteria for his classification, but it is a welcome replacement for the preexisting mis- cellany of suprageneric taxa based on wood for Carboniferous groups, leaves for Permian groups, and ovular fructifications for Mesozoic groups. Meyen (1984) outlined a new botanical nomen- clature for seed plants, but here we use well.es- tablished terms such as sporophyll (phyllosperm of Meyen), sporoclad (branching polysperm), head (ultimate segment of cladosperm), and cupule (also used by Meyen). The word ovule is used here in the strict sense for unfertilized integumented mega- sporangia. In most gymnosperms, seeds represent dispersed propagules, whereas ovules are found in place in reproductive organs. Thus the word ovule appears in this account of reproductive organs more frequently than the word seed. In much paleobo- tanical writing, the latter term being understood more easily in general English usage, is used loosely for ovule or seed. We have also used the term prepollen for the microspores of very early seed ferns. Prepollen look more like spores of pterido- phytes than pollen of seed plants, and, like spores, prepollen germinate from the side of the grain originally oriented within the parent tetrad (Chal- oner, 1976) We have examined type material of all the seed ferns reconstructed here and a good deal of ad- ditional material. One of us (G.J.R.) also has visited the type or comparable localities of these fossil plants in order to assess and obtain new evidence for their geological occurrence. In pd to the drawings, we have written an inter y as an introduction to each ua The ions of each account details both evidence and argu- ments for and against various aspects of our re- construction. The plants are discussed in order of geological age. STAMNOSTOMA HUTTONENSE Hypothetical reconstruction. We envisage this plant as a tall forest tree (Fig. 1) with coni- ferlike wood (Pitus primaeva Witham, 1833). It was a prominent tree of well-drained soils of ele- vated river terraces surrounding lagoons of a broad coastal plain to the south of a hilly and volcanic region now forming the Southern Uplands of Scot- land. During the Early Carboniferous (late Tour- naisian or about 352 million years ago) the climate of this region probably was subtropical, with a 1012 Annals of the Missouri Botanical Garden aw. ve. m. G] NIE Sue ke A Ex Jod e Z Volume 75, Number 3 1988 Retallack & Dilcher Seed Ferns 1013 pronounced dry season. During this season, the pinnules of its large fronds (Aneimites acadica Dawson, 1860) may have been deciduous. In some ways these fronds resembled the plagiotropic short shoot systems of progymnosperms, but the petiole of the plant (Lyginorachis papilio Kidston, 1923) was dorsiventrally differentiated and its vascular structure like that of true leaves. Stalks arising from within the dichotomy of the rachis of the frond terminated in cupules (Calathiops sp.) bearing ovules (Stamnostoma huttonense Long, 1960a). The prepollen organ (Telangium sp. when petrified and Telangiopsis sp. in compression) probably was borne in a similar manner. It consisted of a copi- ously branched aggregate of clusters of elongate pollen sacs. The fernlike prepollen grains (Cola- tisporites decorus (Bharadwaj & Venkatachala) Williams in Neves et al., 1973) were released through an elongate dehiscence slit. These plants may have been wind pollinated at a stage when the ovules were immature. The prepollen was sealed in an apical chamber of the ovule (lagenostome) at the base of a trumpetlike opening (salpinx) by the upward and outward growth of a central plug of tissue. There it waited as the multicellular game- tophyte and archegonia developed. Once fertiliza- tion was achieved, perhaps after shedding and wind dispersal of the ovule, there appears to have been little interruption in the rotting of the seed integ- uments and germination of the embryo. Evidence for reconstruction. The main lo- cality for our reconstruction is the quarries in the steep eastern banks of the Crooked Burn west of Newton Farm, near Foulden, Berwickshire, south- ern Scotland (Wood & Rolfe, 1985; Scott & Mey- er-Berthaud, 1985). This is in the lower Cement- stone Group of the Calciferous Sandstone Series, of Early Carboniferous age (late Tournaisian, or late Courceyan in the regional stratigraphic scheme; Scott et al., 1984; Clayton, 1985) or 352 million years old (following Palmer, 1983). The attribution of these various remains to a single plant is based on anatomical similarity of petrified petioles and small branches; on attachment of petrified stalks within the dichotomy of petioles as in other fertile seed ferns; on the similarity of pollen found in pollen sacs, in ovules, and dispersed; and on the associ- ation of both petrified and compressed remains at the same localities, often with few other associated plants (Long, 1960a, 1962, 1963, 1979a; Neves etal., 1973; Wood & Rolfe, 1985; Scott & Meyer- Berthaud, 1985). Most of this evidence has been presented by Long (1979a), but we doubt that the leaf type, Sphenopteris affinis, or petrified stem, Tristichia ovensii, belonged to this plant, for the following reasons. Aneimites acadica is a very common leaf at the main locality for this plant (Newton Farm), where it is represented by abundant isolated pin- nules, which were probably deciduous. The overall frond morphology thus remains poorly known, and Long’s (1979a) objections on this basis against this being the foliage of the petrified petioles have little substance. There are several problems with Long’s view that Sphenopteris affinis was the leaf of the plant in question. Sphenopteris affinis was prob- ably a shrubby plant growing in quite different waterlogged soils (of the Oil Shales: Andrews, 1948). Only one very poorly preserved fragment (British Museum of Natural History specimen 16865) of S. affinis has been found at the main locality for our reconstruction (Newton Farm) of Stamnosto- ma huttonense, and this equally could be a badly lacerated specimen of Aneimites acadica or a rag- ged specimen of Sphenopteridium pachyrrachis, also reconstructed here as part of Lyrasperma scotica. In addition, the prepollen of Telangium affine found in association and attached to Sphe- nopteris affinis in the Oil Shales has a clear or- nament and trilete mark (Kidston, 1924), quite different from the almost featureless prepollen found in the prepollen chambers of ovules and in prepollen organs attributed to Stamnostoma huttonense (Long, 1962, 1979a). Further, the prepollen of Sphenopteris affinis is somewhat smaller (52 um according to Kidston, 1923) than that of the re- constructed plant (54-69 um according to Long, 1979a). This would have been a more substantial difference if the prepollen grains of Stamnostoma huttonense are lacking their outer wall, as Long suspected. — FIGURE 1. ewton Farm, Xylem model of branches, petiole, and cu A reconstruction of Stamnostoma huttonense E Early Carboniferous (Tournaisian) age, FT near Foulden, po da Scotland.—A. Habit ule.—D-F. Sche of well-drained soils.—B, C. anches and petioles, showing as a large tree matic cross sections of bra primary xylem (black dots) , secondary en (surrounding areas) , sclerotic nests (asterisks) , and mecha ^u Cellular structure in tangential and radial (respectively) sections of wood.— vulate T tai . Ovule .—O-S. Sections of ovule, her 1014 Annals of the Missouri Botanical Garden Scott & Meyer-Berthaud (1985) reinterpreted Tristichia ovensii from the main locality for our reconstruction (Newton Farm) as a separate small seed fern with Rhodea-like leaves, “small uncu- pulate seeds," and “small, lateral, pedunculate male organs." They also questioned whether the speci- mens of Tristichia ovensii between the fork of Lyginorachis papilio and aligned with Stamno- stoma huttonense (Long, 1963) are really at- tached. We accept their reconstruction of this oth- er plant but see little reason to question the attachments described by Long, especially in view of the very generalized anatomy of T. ovensii. Very small axes of several species of seed ferns may have been comparably anatomically simple. Habit. | Stamnostoma huttonense was a large tree, with trunks up to 25 m long and 1.4 m in diameter (Long, 19792). On some of the petrified trunks, branches and petioles have been found attached in a closely spaced helix, whereas other axes lack branches for considerable lengths. This is a common branching pattern in modern forest trees (Rauh's model of Hallé et al., 1978). This plant also had the long and short shoot organization found in many modern conifers, such as Scots pine (Pinus sylvestris). Unlike the leaves of most conifers, its leaves were fernlike. These fossil trees were superficially similar to extinct progymnosperms such as 4 chaeopteris ovata (see Beck, 1981). These had leaves similar to the pinnules of Aneimites acadica, although these were arranged in large dorsoven- trally planated shoot systems quite different from the true leaves of 4. acadica (Kidston, 1924). Ovules were attached on stalklike structures which formed the central axis of a trichotomy of the rachis of the frond. A petrified terete stalk with a triarch stele has been found in the dichotomy of a petrified petiole (Long, 1962). This was a com- mon mode of attachment of ovular and pollen- bearing structures, seen in other Early Carbonif- erous compression fossils, such as Sphenopteris affinis (Kidston, 1924), Sphenopteris bifida (Long, 1979a), and Diplopteridium teilianum (Walton, 1926, 1931). Compared with modern gymno- sperms in which axillary branching and clear dif- ferentiation of stem and leaf are the rule, this epiphyllous sporoclad is peculiar. Presumably this arrangement was inherited from progymnosperm ancestors that lacked consistently axillary branch- ing (as shown by Scheckler, 1976, 1978). The sporoclads of Stamnostoma huttonense may have been erect. Their stalks are terete and gently curved, rather than the dorsiventrally planated and = flexuous as in pendent fructifications. Niklas (1981) has shown that an erect orientation would have been more effective for pollination, because the rachides of the fronds would have remained in the way of pendent ovules even if all the pinnules were abscised at the time of pollination. Long (1965) and Walton (1964) argued that cupules of other species were pendent because this would protect the pollination drop from rain. However, the open form of this particular cupule would have been a rain guide, rather than protection. Reproduction. The life cycle of this plant was probably similar in general outline to that of a number of associated Early Carboniferous seed ferns (Rothwell, 1986). Irregularly branching cupulate structures of other species sometimes are found with poorly developed ovules or prepollen sacs (Long, 1969, 1975, 1977a, 1979b). We interpret these as fructifications at the stage of pollination or earlier, analogous to modern fructifications of cycads (Dioon edule) and maidenhair trees (Gink- go biloba) at this stage in their development (Chamberlain, 1935). The smooth prepollen were produced in great quantity. These are features of modern wind-pollinated plants (Faegri & van der Pijl, 1966; Whitehead, 1969). The size of the prepollen (54—69 um in diameter) approaches that of modern insect-dispersed pollen, but their aid in pollination is unlikely in view of other features of the prepollen. At pollination stage the funnellike nucellar apex (salpinx) may have retained a pol- lination drop for entrapment and withdrawal of prepollen (Walton, 1964) as in some modern co- nifers. The best-preserved fossils of mature ovules have a central column filling much of the salpinx and a multicellular megagametophyte, occasionally with one to three apical archegonia (Long, 1960a). These may have been ovules during the long period after pollination and before or shortly after fertilization. Judging from the abundance of remains of this stage, this was a long period in the life cycle, perhaps taking several months, as in modern cy- cads and Ginkgo (Chamberlain, 1935). Even ma- ture ovules have three small cutinized megaspore remnants at the apical end, representing aborted spores of the parent tetrad (Long, 1975). This feature, presumably a legacy of heterosporous pro- gymnosperm ancestors, is no longer seen in modern seed plants. Mature seeds were small (up to 3.75 mm long by 1.5 mm wide) and had a thin, dense seed coat. Compared with modern seeds (Van der Pijl, 1972), they were unspecialized and possibly scattered by Volume 75, Number 3 esa : Dilcher 1015 1988 Seed F m m 7 i y bu hey of R coastal swamp well drained soil of forested soil seasonally t, and may ree flood deposits distal levee of e of stream successional peaty soil of vegetation coastal swamp soil of seasonally ah i [o n. XN owland es water swam ores B. STALYBRIDGE i i 3 weakly developed,” F- -H Se Eg EH * eutrophic ` : — E=] oxbow lake | 2 e1-—- = D. CAYTON BAY Q coastal lagoon -H channel Late/Early oo osais L—] lag Triassic oil of Y disconformity «æ fossil logs sind wer, A. NEWTON FARM fossil stumps C. UPPER UMKOMAAS VALLEY $^ fossil root traces grainsize scale : EN coal = © marine fossils Sar 2 8 [==] shale, claystone SFF $ Š ae fossil fish siltstone — planar bedding mi red color FE) sandstone - ripple marks orange color oo oc o 909 00 oo oo conglomerate ~~~ erosional surface FIGURE 2. near Foulden, Scotland Stalybridge, Cheshire, England (for Lagenostoma lomaxi maas Valley, Bay, Yorkshire, England (for Caytonia nathorstii) . Section <= wavy bedding e^ ferruginized surface Natal, South Africa (for Peltaspermum thomasii and e sh ta) .— Zw sideritic band C» calcareous nodules Stratigraphic sections and interpreted paleoenvironments q st M at:—A. New (for Stamnostoma huttonense is Lyraspe erma scotica n Far. Hough jew Colliery near —C. Near V r Umko- ere measured i eld b G.J.R., except for B, 1). which was compiled from sections described by Stopes & Watson (1909) M Tonks et al. (193 shaking from the cupules, followed by wind and water dispersal. No seeds with embryos have been found among many examined in Early Carboniferous rocks. Those few possible seed fern embryos found have two seed leaves and are free of their seed integuments (Long, 1975). It is likely that germination followed rapidly after fertilization, again as in modern cy- cads and Ginkgo (Chamberlain, 1935). If most mbryos grew without pause into small seedlings, this would explain their rarity. Habitat. The gray shales at Newton Farm (Fig. 2A) have yielded a variety of fossil sharks, palaeoniscid and acanthodian fish, crustaceans, bi- valves, and plants (White, 1927; Long, 1960a, 1016 Annals of the Missouri Botanical Garden 1962, 1964; Wood & Rolfe, 1985; Scott & Mey- er-Berthaud, 1985). Apart from arborescent ly- copods and rare sphenopsids, the best-known plants from here are the seed ferns reconstructed as Stamnostoma huttonense and Lyrasperma sco- tica, and a small plant with Rhodea-like foliage. Remains of all these plants have been found mixed together in shales of what was once a large brack- ish-to-freshwater coastal lagoonal system (Clark- son, 1985). Several fossil soils (paleosols) in this sequence allow more precise understanding of where Stam- nostoma huttonense grew. Overlying the lagoonal shales are sequential, thin, waterlogged paleosols (gleyed Inceptisols of Soil Survey Staff, 1975). The size and nonanastomosing cortical striations of fossil root traces in these are most like those of Lyra- sperma. scotica, also reconstructed here. A better- drained paleosol containing larger and more deeply penetrating root traces underlies the lagoonal shales. Stamnostoma huttonense probably grew in such well-drained soils, considering the deeply pene- trating roots under its fossil stumps, its well-cuti- cularized seeds, and the growth rings in its wood, not seen in coeval arborescent lycopods of swamps (Gordon, 1935). The well-drained paleosol thought to have sup- ported Stamnostoma huttonense has a carbona- ceous surface horizon: an indication that it was not always entirely dry. Its subsurface zone of iron- staining and deep (60 cm) horizon of calcareous nodules (caliche) mark the minimum usual depth of the water table. It was a young soil (probably formed in only hundreds of years before covered), because development did not proceed to the extent that the original bedding was entirely destroyed. It probably formed on an alluvial terrace, a meter or so above the level of the nearby lagoonal system. Additional evidence of Early Carboniferous for- ests of Stamnostoma huttonense can be seen at other British localities. Along the King Water be- tween Spadeadam and Gilsland in Cumbria, north- ern England, there are ten petrified stumps of Pitus primaeva preserved in growth position within 200 m of outcrop (Long, 1979a). Considering the likely canopy of these trees and the size of the stumps, they would have dominated a forest with closed canopy. In addition to forests of Stamnostoma hutto- nense and swamp woodlands of Lepidodendron and shrubby Lyrasperma scotica, documented at Newton Farm, seed ferns lived in a variety of other habitats in southern Scotland during Early Car- boniferous time. The oil shale north of Cove Har- bour (Craig, 1975) contains abundant Sphenop- teris affinis, as well as narrow, shallowly penetrating root traces. This was a weakly developed, water- logged paleosol (Aquent of Soil Survey Staff, 1975), supporting scrubby vegetation similar to modern fen carr. Other seed ferns are now preserved in calcareous nodules (caliche) of paleosols (Incepti- sols of Soil Survey Staff, 1975) flanking deposits of a small (about 10 m wide) creek in the Ce- mentstone Group deposits exposed in the southern sea cliffs of Oxroad Bay, south of Tantallon Castle, East Lothian (Long, 1976, 1979b; Barnard & Long, 1973, 1975; Matten et al., 1980; Scott et al., 1984). This fossil assemblage, including Cal- athospermum fimbriatum reconstructed here, ap- pears to have been in large part scrubby, early successional vegetation on the well-drained, ashy soils of a nearby volcano. Even at this early time in the geological history of seed ferns, they appear to have been varied in habit and habitat, and in- cluded stately forest trees such as Stamnostoma huttonense. Paleogeographic setting. This plant and its various parts are widely distributed in the lower Cementstone Group of Early Carboniferous age southern Scotland. At this time, the Cementstone Group accumulated in a coastal plain south of a hilly region including basaltic and rhyolitic volca- noes, and north of a shallow marine shelf, a few large islands, and open equatorial ocean (Anderton et al., 1979). Southern Scotland enjoyed a warm tropical climate, in which corals and large fusuline foraminifera flourished. Statistical analysis of paly- nological data provides evidence of dry climate at this time (van der Zwan et al., 1985). Calcareous nodules in paleosols of the Cementstone Group (Fig. 2A) are evidence of a climate at least as dry as subhumid, and probably seasonally dry. A dry sea- son is also apparent from growth rings in fossil wood of this sequence (Long, 1979a; Creber & Chaloner, 1984) and the development of marine evaporites in tropical seas to the south (Ramsbot- tom, 1973). Although close to the equator, this area was in the rain shadow of large mountain ranges to the west (Bambach et al., 1980). LYRASPERMA SCOTICA Hypothesized | reconstruction. Lyrasperma scotica is reconstructed as a small bush, with stiff, coriaceous, fernlike leaves (Fig. 3). It probably formed a shrubby understory to swamp woodland of arborescent lycopods around coastal lagoons of the same age and areas already described for Stam- nostoma huttonense. Its leaves (Sphenopteridium pachyrrachis (Goeppert) Potonié, 1899) were Volume 75, Number 3 1988 Retallack & Dilcher Seed Ferns 1017 strengthened by a thick cuticle and, in the rachis and petiole, by a cortical mechanical tissue (dic- tyoxylon cortex), formed by radially arranged blades of sclerenchyma. Their petioles (Kalymma tue- diana Calder, 1938) had a pulvinus and numerous vascular strands. They were borne stiffly and hor- izontally, in a well-spaced helix on slender erect stems (Stenomyelon tuedianum Kidston in Scott, 1909). The stems were smooth and barkless, with an outer dictyoxylon cortex and an inner paren- chymatous cortex, and a soft, mixed pith within the central cylinder of secondary xylem. The pre- pollen organs of this plant are not known, but some ovules were found containing prepollen (Colati- sporites denticulatus Neville in Neves et al., 19773). The ovules (Lyrasperma scotica (Calder) Long, 1960b, when petrified, and Samaropsis bicaudata (Kidston) Kidston, 1902, in compression) were len- ticular and had two prominent horns on either side of the apex. Ovules only have been found isolated. They may have been enclosed within epiphyllous cupules (Alcicornopteris convoluta Kidston, 1887). Like other Early Carboniferous seed ferns, polli- nation was presumably by wind, and dispersal by wind and water. The distinctive large horns of the ovule may have aided dispersal over water. Evidence for reconstruction. Our reconstruc- tion of this plant is based on the same Early Car- boniferous (Tournaisian) locality near Newton Farm, southeastern Scotland, already discussed for Stam- nostoma huttonense (Wood & Rolfe, 1985; Scott & Meyer-Berthaud, 1985). The attribution of these various remains to one plant is based on the sim- ilarity of dispersed prepollen to that found in the ovule; attachment of petrified petioles to petrified stems; the similar size, shape, stelar arrangement, long internodes, evidence of sclerotic nests, and nonanastomosing woody cortex in petrified and compressed petiole-bearing stems; the similar size and shape of petrified and compressed ovules; and consistent association at the same localities (Long, 1960b, 1964). A most important specimen for our reconstruction is the large compressed trunk with attached leaves excavated by Long (1964). Al- though pinnae associated with this compressed trunk agree with Sphenopteridium pachyrrachis, the petioles of the specimen lack the pinnae below the fork and the rough transverse bars usually found in that species (Long, 1964). Comparably anom- alous compression fossil leaves were referred to the same species by Walton (1931), and it is uncertain whether this is part of the natural variation of this species or represents a distinct new form. Habit. This plant had an unbranched erect trunk (Corner’s architectural model of Halle et al., 1978). Its stems were quite succulent, as they had a wide parenchymatous cortex, a mixed pith, and inflated petiole bases (Seward, 1917: 184). Compression of soft tissues in some specimens has given the misleading appearance of “multicellular projections" (Taylor, 1981: 363). These slender stems were strengthened by an outer zone of woody mechanical tissue—bands of sclerenchyma form- ing a dictyoxylon cortex, which in this plant was not anastomosing. There is no indication of bark, and the stem probably was smooth. The leaves were coriaceous and strengthened by sclerenchyma within their petioles. The base of the petiole was expanded into a broad, fleshy, pul- vinuslike structure. Attached leaves were arranged stiffly and horizontally on the stem. The arrangement of ovules in this plant is un- certain because they only are known isolated. In similar Late Carboniferous plants described by De- levoryas & Taylor (1969) and Corsin (1928), synangia and ovules were borne bipinnately on what appears to be a fertile frond, in a similar fashion to our reconstruction of Lagenostoma lo- maxii. In contrast, Long (1977a) has argued that the ovules of Lyrasperma scotica were borne in complexly branched cupules, similar to our recon- struction of Stamnostoma huttonense. Long's ar- gument is based mainly on the widespread asso- ciation of ovules (Samaropsis bicaudata) and cupules (Alcicornopteris convoluta), the very sim- ilar anatomy (especially of the lagenostome) of oth- er ovules (Eurystoma angulare, E. burnense, and Hydrasperma longii) known to be borne in cupules (Long, 1965, 1969, 1975, 1979b; Matten et al., 1980), and the existence of compressed fronds (of Sphenopteridium pachyrrachis) showing a tri- chotomy of the rachis (Kidston, 1923, pl. 39, fig. 5; Long, 1960b). It is most likely that the ovules, and perhaps also sporangia, of this plant were borne within infolded cupules as in Alcicornopteris. We regard many of Long's Alcicornopteris-like spec- imens of Hydrasperma longii (Matten et al., 1980) as immature and have modeled our reconstruction of the cupule after that of Eurystoma angulare (Long, 1969). As in the generalized reconstruction of similar plants by Camp & Hubbard (1963), the cupule is shown erect on the frond for reasons similar to those given for our reconstruction of Stamnostoma huttonense. Reproduction. The early development and pollination of Lyrasperma scotica was probably similar to that of Stamnostoma huttonense. Ly- rasperma scotica had a wide, shallow salpinx, un- Annals of the 1018 Missouri Botanical Garden AIR E C Emm TERE TTN esM Pe Rete ne wewe. uie Kas see eM M Tte: DU ANETNNUT DATED QE yg. sam NOTTE E ry is va i? TUUM v ` ede tot a lOcmEE——A A MANDA q TAAL Se . Mes SP [| Qf HANS | | | | B Emm ————— — CUT] — — — — .°.h o — llin SA € SoMa teg rn ys ER E Volume 75, Number 3 Retallack & Dilcher 1019 Seed Ferns like many other Early Carboniferous seed ferns. The mechanism of sealing the pollen chamber after pollination could not have been by upward growth of a plug of tissue, as envisaged for Stamnostoma huttonense or Lagenostoma lomaxii. It may have been by lateral growth of the central column or buckling of the central column by growth of the tent pole. The prominent horns of the ovule are somewhat reminiscent of those in the living water chestnut (Trapa natans), in which barbed spines may serve as floats during dispersal over water and deter consumption by fish or other aquatic vertebrates. The fossil ovule does not have an especially woody or fleshy integument of the kind found in modern fish-dispersed seeds (Gottsberger, 1978). Habitat. A lowland habitat is indicated by the occurrence of Lyrasperma scotica at the same locality on Newton Farm (Fig. 2A) sd dis- cussed for Stamnostoma huttonense (Wood & Rolfe, 1985; Scott & Meyer-Berthaud, 1985). aa seed fern roots with nonanastomosing striations, like those of this plant, as well as poorly preserved lycopod cones, have been found in paleosols over- lying the lagoonal shale. These paleosols lack coal, and fossil root traces penetrate them more deeply than usual for permanently waterlogged soils. Nor do they have the reddish oxidized minerals of well- drained soils or the prominent relict bedding of very young soils. They were weakly developed, periodically waterlogged, clayey lowland soils (gleyed Inceptisols of Soil Survey Staff, 1975), probably marginal to the lagoon. Fossil fish and shrimp in the underlying lagoonal deposits (Clarkson, 1985) are evidence that this was a brackish-to-freshwater inland part of a large coastal lagoonal system, con- nected to the ocean. Lyrasperma scotica may have formed a shrubby understory to this lagoon-margin woodland. he suggested succulence and thick leaf cuticles of this fossil plant could be considered indications of a locally or regionally arid climate, but consid- ering geological evidence for its habitat, these fea- tures more likely allowed its growth in salty or nutrient-poor, stagnant groundwater. This inter- pretation is also compatible with the elaborate, fleshy, open-mouthed seeds of this plant, which would have required moist but not necessarily nu- trient-rich conditions for germination and early seedling growth. Paleogeographic setting. This plant is found at several localities in the Cementstone Group of Early Carboniferous age in southern Scotland, in the same region and time as Stamnostoma hut- tonense (Wood & Rolfe, 1985; Scott & Meyer- Berthaud, 1985). CALATHOSPERMUM FIMBRIATUM Hypothesized reconstruction. Also of Early Carboniferous age (Tournaisian or 352 million years old), this small herbaceous-to-shrubby plant was possibly an early successional colonizer of the banks of gullies and ephemeral creeks in ash at the foot volcanoes within a large rift Ww now the Midland Valley of Scotland (Fig. 4 (Sphenopteridium capillare Walton, 1931) were dichotomously forked, and the pinnae copiously divided into terete and filiform segments. The stem (Calathopteris heterophylla Long, 1976) had a wide cortex with narrow medullated rings of sec- . Its leaves ondary xylem. Ovules were borne within large cu- pulate structures (Calathospermum fimbriatum Barnard, 1960), which had a series of pinnae on the petiole like those of ordinary foliage leaves. ithin the cupule were about 16 erect-growing, elongate ovules (Salpingostoma dasu Gordon, 1941). The prepollen organ (not yet known) may have been attached to the central stalk of a tri- chotomously divided leaf rachis. It may have been a copiously branched structure with numerous elongate sporangia, as in allied seed ferns. The prepollen of this plant (Perotriletes tessellatus (Staplin) Neville in Neves et al., 1973) were so large (104 um in diameter) that this plant may have been pollinated by small animals. The long integumented ovules may have been shaken from the cupules by wind. — FIGURE 3. A reconstruction of Lyrasperma scotica of Early Carboniferous (Tournaisian) age, from Newto .H land. R Farm, near Foulden, Berwickshire, Scotland. — stem and petioles. —C, D. Sche points) , secondary x cortex (radial shading). as a shrub of coastal swamplan matic cross section of stem and petiole (rese Ni ¿As xylem i LE. Detail of leaf. —F. Ovule.— —B. Xylem model of sterisks) , mechanical itudinal se T s of ovule, showing vascularization (heavy lines), woody iesus layer (stipple) , modus os (cellis pat- tern) , and archegonium (circle with stippled center) .—L. Prepollen. 1020 Annals of the Missouri Botanical Garden Volume 75, Number 3 1988 Retallack & Dilcher 1021 eed Ferns Evidence for reconstruction. The various parts of Calathospermum fimbriatum have been put together from similarities in anatomy and hairs between isolated ovules and cupules; from the sim- ilar anatomy of petioles of petrified cupules and petioles of petrified stems; and from close associ- ation at one especially well-studied locality (Bar- nard, 1960; Long, 1976; Scott et al., 1984). These are the calcareous, nodular layers of a buried creeklike feature in sea cliffs south of Tantallon Castle, in Oxroad Bay, East Lothian, Scotland (Long, 1976). This is part of the Cementstone Group, Calciferous Sandstone Series, of Early Carbonif- erous age (late Tournaisian or Courceyan of Scott et al., 1984), or 352 million years ago (following Palmer, 1983). The prepollen of this plant have been found in petrified ovules (Long, 1976) and are similar to a common type of dispersed grain (Neves et al., 1973). Similar prepollen are found in a fructifi- cation of another closely allied species (Staphy- lotheca kilpatrickensis Smith, 1962) from another Scottish locality of comparable age (Loch Hum- phrey Burn). The prepollen organ of Calatho- spermum fimbriatum may have been similar in some respects but remains unknown. Only the pet- ioles likely to have borne the prepollen organ of Calathospermum fimbriatum are known. o compression fossils have been found in as- sociation with the petrified remains, but the pet- rified pinnae below the fork of the cupule agree in all respects with Sphenopteridium capillare (G.J.R., pers. obs.), a leaf known from other lo- calities in Fife, Scotland (Walton's, 1931, material from Burntisland; British Museum (Natural His- tory) specimen v31831 from Ardross). Unlike Oxroad Bay, these Fife localities are within the Oil Shale Group or its equivalents (MacGregor, 1968) and are slightly younger (Visean) within Early Car- boniferous time (equivalent or older than Pettycur fossils from this area; Scott et al., 1984). Other kinds of fronds are unlikely to have be- longed to this plant, as they are thought to have had quite different fructifications (Stamnostoma huttonense and Lyrasperma scotica, as recon- structed here, and fertile Sphenopteris bifida of Long, 1979b), or they differ in having few (Sphe- nopteris affinis: see Kidston, 1924, pl. 100, fig. 1) or no pinnae below the fork (Adiantites ma- chanekii, Diplotmema bermudensiforme and Spathulopteris ettingshausenii of Walton, 1931). The other common kind of leaves of this age (Rha- copteris spp.) were unforked, although their micro- sporangiate axes were forked (Walton, 1926). Compression fossils showing these leaves attached to stems (Walton, 1926) are very similar in size, phyllotaxy, angle of attachment of petioles, and spacing of pinnules to petrified plant remains (re- ferred to Buteoxylaceae by Barnard & Long, 1973, 1975), different from the plant reconstructed here. Habit. From a distance these plants probably looked like umbelliferous weeds, such as hemlock (Conium maculatum) and anise (Pimpinella ani- sum). The stem of Calathospermum fimbriatum was weak, with medullated secondary wood (Long, 1976). No growth rings were seen. The single stem found was densely clothed in leaves and showed no branches within its 17 cm length (Long, 1976). It was not a copiously branched plant and had much weaker wood than seed ferns such as Stam- nostoma huttonense. We have reconstructed it as a small shrub, but the availabl may have been a young plant of a species usually more com- plex in architecture (G. W. Rothwell, pers. comm., 1985 Reproduction. The enlargement and differ- entiation of the megagametophyte between the time of pollination and fertilization was not synchronous. There are small, poorly developed ovules on short stalks near the base of the cupule, as well as large, mature ovules on long stalks near the mouth of the cupule (Barnard, 1960). Such variation in de- velopment is not nearly so marked in other Early Carboniferous seed ferns, although these do include occasional *aborted" ovules (Long, 1960a). **Con- tinuous flowering" of this kind is commonly found in modern plants of disturbed habitats, where op- portunities for seedling establishment are unpre- dictable (Heinrich, 1976). Several features of this plant suggest a syndrome now associated with pollination by animals (Faegri & van der Pijl, 1966). These features are especially striking in comparison with co-existing wind-polli- — FIGURE 4. Oxroad Box East Lothian, Scotland.—A. Habit a stem and petioles.—C. Sc rounding cellular pattern) , Pa ded nests (asterisks) , and mechanical cortex (radial sha —L. Transverse and longitudinal sections of ovule leaf. —E. Ovulate cupules.—F. Ovul A reconstruction of uos fimbriatum of Early Mp diy mm age, in a weedy shrub of inland streamsides.— B. hematic cross section aur primary xylem (black Too) , rm pica (sur- model of —D. Detail of M Prepollen. 1022 Annals of the Missouri Botanical Garden nated seed ferns such as Stamnostoma huttonense and Lyrasperma scotica. The cupules were large and almost radially symmetrical, and so might have been easily recognized by animals. They were borne erect, as can be seen from the kink where the cupule was bent farther upwards (adaxially) on the attached petiole (Barnard, 1960; Long, 1975). The prepollen of this plant were large (104 um), much larger and heavier than is effective for wind-pol- linated plants today (Whitehead, 1969). Finally, the ovules and the interior of the cupule were covered with numerous glandular hairs (Gordon, 1941; Barnard, 1960) of unknown function. They may have offered a nutritional reward to animals or protected the ovules from them. It is difficult to be certain what kinds of creatures could have been pollinators—few likely terrestrial animal fossils are known from this geological time. There is a poor Late Devonian and Early Carbon- iferous fossil record of canopy-dwelling spiders and mites. Winged insects are not found as fossils in rocks older than earliest Late Carboniferous (Na- murian: Rolfe, 1980) The propagules of this plant appear to have been individual ovules, because only a few mature ovules are found in any particular cupule, and in some cases cupules contain none. Mature ovules were elongate and of moderate size (6 mm diameter and 5 em long including nucellar extensions). There were zones of resinous hairs on the inside of the nucellar extensions and between the ribs of the body of the ovule, which was enveloped in a hard sclerotesta. The whole structure is larger, heavier, and more elaborate than propagules usually dis- persed by wind or water alone, such as the pappus of modern Compositae which Walton (1964) thought comparable. They were probably shaken loose from the cupule by wind (a “wind ballist” of van der Pijl, 1972). Also possible is dispersal by ground-dwelling spiny millipedes and similar large arthropods (Rolfe, 1980). Habitat. This fossil plant at Oxroad Bay is found encased in calcareous nodules (caliche) of weakly developed paleosols (Inceptisols of Soil Sur- vey Staff, 1975) on the aggrading banks of a small (10 m wide) creek deposit (Gordon, 1941, pl. 1 fig. 1; Scott et al., 1984). The stabilized banks (rooted calcareous nodules) can be seen to be lat- erally impersistent, eroded on top, and covered alternately on each side of the creek. The area was well drained and quite near the volcano, as can be seen from the nature of these paleosols and associated bouldery volcanic mudflows. In such well-drained, frequently disturbed soil one would expect small, xeromorphic, and early successional plants. Such a habitat agrees with several adaptive fea- tures of this plant: small size (Long, 1976), dis- sected pinnae (Walton, 1931; Barnard, 1960), cupular protection of ovules (Barnard, 1960), long extensions of the ovular integument, and hairy coat and woody integument of the ovule (Gordon, 1941). This also is compatible with what is known about the other fossil plants found there (listed by Scott et al., 1984). Lycopods and remains of arborescent plants such as Stamnostoma huttonense are rare and fragmentary. Most common are small seed ferns and enigmatic plants (family Buteoxylaceae). Paleogeographic setting. This reconstruc- tion is based mostly on petrified fossils of Early Carboniferous (late Tournaisian) age, from Oxroad Bay, East Lothian, Scotland (Long, 1976; Scott et al, 1984). Only the basal parts of pinnae are known from Oxroad Bay, and the remainder of the frond is modeled after compressions from Ardross and Burntisland, Fife, Scotland (Walton, 1931; British Museum specimen v31831). All of these localities were in a large rift valley, now occupied by the Midland Valley of Scotland. This was flanked by the Grampian Highlands to the north and the Southern Uplands to the south, and opened out to the sea toward the southwest (Anderton et al., 1979). It wasa volcanic landscape with extensive flows from fissure eruptions and some imposing alkali-basaltic volcanoes (Francis, l Calathospermum fimbriatum may have lived more inland and at higher elevation than the coastal lagoons colonized by Stamnostoma huttonense and Lyrasperma scotica, and probably in a similar subtropical and seasonally dry climate. As Gordon (1941) argued, it may have lived in a drier climate than these other plants. It was within the rain shadow of rift valley walls, volcanoes, and the mountain ranges of the Grampian Highlands and contiguous ranges now in North America (Bambach 1980) et al., LAGENOSTOMA LOMAXII Hypothesized reconstruction. We envisage Lagenostoma lomaxii as a shrub with an irregular crown of large leaves (Fig. 5). It grew in extensive, permanently waterlogged swamps, dominated by arborescent lycopods, on deep peaty soils. During the early part of the Late Carboniferous (early Westphalian A or about 320 million years ago) Volume 75, Number 3 1988 Retallack & Dilcher Seed Ferns 1023 this area was in the subhumid tropics. The trunk (Lyginopteris oldhamia (Binney) Potonie, 1899) was slender (3-4.5 cm diameter) and strengthened by a thick outer zone of sclerenchyma forming an anastomosing system of radially arranged plates (dictyoxylon cortex). The lower part of the stem was anchored by numerous slender prop roots (Ka- loxylon hookeri Williamson, 1875), which ran di- rectly down from the stem and branches but did not sheath the stem as they do in many modern tree ferns and palms. Its petioles (Rachiopteris aspera Williamson, 1874) were flattened and had two vascular strands near the base. The leaves (Sphenopteris hoeninghausii Brongniart, 1828) were large and spreading, with a dichotomously forked rachis and numerous small, orbicular, third- order pinnules. The frond rachis and the stems were clothed in prominent stalked glands. Cupulate ovules (Lagenostoma lomaxii Williamson in Oliver & Scott, 1903, when petrified, and Calymmato- theca hoeninghausii (Brongniart) Stur, 1877, in compression) were borne pinnately, in modified api- cal parts of fronds. Prepollen organs are not cer- tainly known, and prepollen are represented only by badly corroded specimens within the pollen chamber of the ovules. The cupulate cover to the ovules was liberally studded with stalked glands like those on the stem and leaves. Evidence for reconstruction. Our reconstruc- tion of Lagenostoma lomaxii is based principally on structurally preserved fossils in coal balls from the Hough Hill Colliery, presently abandoned on the south face of the hill 1 km south of Stalybridge, near Manchester, in Cheshire, England (Stopes & Watson, 1909). They come from the “Six Inch Mine Coal" of the uppermost Millstone Grit, of early Late Carboniferous age (early Westphalian A; Tonks et al., 1931; Phillips, 1980) or about 320 million years ago (in time scale of Palmer, 1983). The distinctive stalked glands on these fossil leaves, petioles, stems, and cupules were the main evidence used by Oliver & Scott (1903, 1904) in their reconstruction of this plant, which was the first indication that some gymnosperms included plants with fernlike leaves. There are additional anatomical similarities between the various parts, which are closely associated in coal balls and shales (Benson, 1904; Oliver & Scott, 1903, 1904; Sew- ard, 1917; Jongmans, 1952; van Amerom, 1968). There has been considerable debate about the likely prepollen organ of this plant. Benson (1904 argued that it was Telangium scottii, but T. scottit is nonglandular, and so more likely to have be- longed with Lagenostoma ovoides. Kidston (1905, ~ 1906) thought the prepollen organ was the fossil now known as Crossotheca kidstonii (Hemingway) Jongmans (1952) found attached to leaves in sid- erite nodules of slightly younger geological age (Westphalian B) in the Lancashire Coalfield. How- ever, Jongmans (1952) pointed out that the at- tached foliage belongs to another species of Sphe- nopteris, lacking the characteristic stalked glands. A third possibility figured by Seward (1917) is a fragment of a petrified pinnule with an attached sporangium on the abaxial side, and a nearby emer- gence where it appears that another sporangium may have fallen off. A similar specimen with glands and stalks was figured by Kidston (1906, fig. 2). Until this material is studied further, the prepollen also remain poorly known, because those found in the ovules are too badly corroded to be identified with dispersed species. Habit. Because of the small size of these stems up to 4.5 cm diameter), this plant has been con- sidered a vine (Phillips, 1981). This was a common habit for some Carboniferous seed ferns such as Callospermarion pusillum (reconstructed here). y comparison Lagenostoma lomaxii has a much — more prominent zone of mechanical tissue in the cortex and a lesser development of secondary xy- lem, and lacks asymmetric wood. This plant is also very abundant locally (it may comprise up to 3876 by volume of some coal balls: Phillips, 1981). It may have grown in waterlogged soils, because there are lacunae within the cortex of the stems (Blanc- Louvel, 1966). It was branched copiously, with some very slender branchlets (Blanc-Louvel, 1966). Both axillary and nonaxillary lateral branching have been observed. For all these reasons we do not think that Lagenostoma lomaxii was a vine. Nor do we follow the reconstruction of Scott (1900), showing the plant leaning for support on adjacent trees. Instead, it was probably a shrubby plant with a tangle of prop roots (Attim's model of Hallé et al., 1978). Reproduction. Judging from compression fos- sils (Jongmans, 1952) and the anatomy of cupule stalks (Oliver & Scott, 1904), the cupulate ovules were borne pinnately on fertile fronds or parts of fronds. After pollination, the prepollen chamber was occluded by growth of the central column against the overarching sclerotesta of the ovule. A considerable delay between pollination and fertilization is likely. This would account for the abundance of either ovules or of pollen organs of allied species, but not both, in association with compressed foliage. Jongmans (1952) has argued 1024 Annals of the Missouri Botanical Garden 272 WW eM. PES Za La < a. K t A SST p 4 fe r URN AX AY 202602; e vo > = FIGURE 5. A reconstruction of Lagenostoma lomaxii of earliest Late Carboniferous (Westphalian A) age, from Hough Hill Colliery, near Stalybridge, Cheshire, England. — A. Habit as a swampland shrub.—B. Xylem model of stem and petioles.—C, D. Schematic cross sections of stem and of root (respectively), showing mixed pith Volume 75, Number 3 1988 Retallack & Dilcher Seed Ferns 1025 that pollen organs and mature ovules were pro- duced, abscised, and then decayed during different parts of the year. Compressions of conical cupules lacking prominent ovules (Jongmans, 1952; van Amerom, 1968) may have been fertile fronds with immature ovules at or close to the time of polli- nation. Petrified ovules containing megagameto- phytes and archegonia are a good deal larger than these remains and set within a bell-shaped cupule with flaring lobes (Oliver & Scott, 1904) Lagenostoma lomaxii may have been pollinated by insects, considering the large size (averaging 55 by 70 um) and coarse ornament of the prepollen (Oliver & Scott, 1904), within the range found in modern insect-pollinated plants (Whitehead, 1969). A case also could be made that the capitate glands were insect attractants, but we find this unlikely. The distribution of glands over almost all known parts of the plant is most compatible with inter- pretation as organs to deter insect herbivory. Winged insects were around in some diversity and abundance by this geological period (Rolfe, 1980). By the time the ovule was ready for dispersal, it was about 5.5 by 4.5 mm in size (Oliver & Scott, 1904). The cupule appears to have spread open and the capitate glands withered, judging from compression specimens (Benson, 1904; Seward, 1917). Since many of these lacked attached ovules, it is doubtful that the cupule played a role in dis- persal. Habitat. This fossil plant is best known in coal balls from Hough Hill Colliery near Stalybridge, England (Stopes & Watson, 1909). Coal balls are calcareous or dolomitic nodules found within coal seams (Scott & Rex, 1985). Although modified somewhat during burial (Rao, 1985), coal balls developed as the peat accumulated, as a kind of caliche nodule in peaty soils (Retallack, 1986). In most swamps the development of such nodules is limited by acidity. Coals containing coal balls form under fen or carr vegetation of neutral to alkaline wetlands, where acidity is buffered by a subhumid, seasonally dry climate and nearby limestone bed- rock. The coal containing coal balls in Hough Hill Colliery is interpreted here as the less decayed (mor) humus layer of the organic horizon of a peaty paleosol (Histosol) of permanently waterlogged ground (Fig. 2B). It is underlain by a thick zone 1931) representing a more decayed (mull) humus, which may have formed at a time during the development of this soil when it was periodically better drained. Below this dark clay is gray, leached clay with root traces (“fireclay””). A thin zone of siderite (““iron- of carbonaceous claystone (Tonks et al., stone") represents a deep gley horizon of this older paleosol (comparable to cases discussed by Retal- lack, 6). This paleosol changes character along strike, and in places the organic horizon (Six Inch Mine Coal) directly overlies levee (Rough Rock Flags) and channel deposits (“massive current bedded grit” of Rough Rocks; Tonks et al., 1931) of a former stream. Clayey soils (Entisols of Soil Survey Staff, 1975) of the stream levee supported vegetation dominated by seed ferns, such as Alethopteris lon- chitica and Mariopteris muricata (Stopes & Wat- son, 1909, discussed this “flora in shales”), quite different from vegetation of the swamp (the flora in coal balls). The former stream may have flowed into a la- goon or bay, which ultimately inundated the swamp. The shale overlying the coal contains marine fossils such as goniatites (Gastrioceras sp. aff. G. listeri) and scallops (Pterinopecten papyraceus, ?Posi- donom ya insignis and Posidoniella sp.: Tonks et al., 1931) as well as plant fragments (Lepido- dendron lycopodioides: Stopes & Watson, 1909). In general, Lagenostoma lomaxii is found in coal balls dominated by the remains of arborescent lycopods (more than 90% by volume of the coal ball assemblage: Phillips, 1981) and a variety of understory ferns (Phillips, 1980). We envisage it as a small bush of these lycopod-dominated wet- lands. Vegetation of these permanently water- logged woodlands was distinct from that of stream- sides, which were dominated by other seed ferns. There was not any detectable marine influence in these lycopod woodlands, although coastal lagoons may not have been far away. Paleogeographic setting. | Lagenostoma lo- maxii is best known from coal balls in the upper- most Millstone Grit (Stopes & Watson, 1909; Tonks 1931; Phillips, 1980). The same species is et al., — (clumped eg primary ay (black), secondary xylem qa pattern), and mechanical cortex (radial shading) .—E. Detail of leaf. —F, G. Cupules with ovules gitudinal and transverse sections of cupulate ovules, x nde vasc cularisation. (heavy lines), madlticellidar Boni our dupli (cellular pattern) and arche- gonium (circle with central stipple) .—1L. Longitudinal section of capitate gland. 1026 Annals of the Missouri Botanical Garden TA Za Petre NARA YA K i $ 29 AX : IA 4 Ve y H 2. ERY SY Z: a AE LE. — AAA " Ñ NY y Z. E- à e I +> Volume 75, Number 3 1988 Retallack & Dilcher 1027 eed Ferns found also in shales of about the same age elsewhere in the British Isles and western Europe (Jongmans, 1952). In Cheshire, the coal accumulated as peat in a coastal swamp on the northern margin of the Wales— Brabant landmass (Anderton et al., 1979), which was a promontory extending west into the moun- tains of eastern North America (Ziegler et al., 1979). During most of the Late Carboniferous this was a subhumid part of the tropics, in the rain shadow of mountains to the west (Bambach et al., 1980). There were dry periods (perhaps seasonal) and forest fires, which left charcoal in the swamps (Scott, 1979). PACHYTESTA ILLINOENSIS Hypothesized reconstruction. | Pachytesta il- linoensis was similar to modern tree ferns in overall appearance (Fig. 6). It probably grew on river and deltaic levees and other open and slightly elevated or disturbed areas of midcontinental North Amer- ican coastal swamps of late Pennsylvanian (Ste- phanian) age, about 296 million years ago. By this time the dominant swampland plants were no longer arborescent lycopods but marattiaceous tree ferns. By comparison with these and modern tree ferns, this plant had less copiously divided and more leath- ery leaves (Alethopteris lesquereuxii Wagner, 1964). Its trunk (Medullosa noei Steidtmann, 1944) was formed largely of closely adpressed leaf bases. It had a wide fleshy cortex and little woody tissue. Large ovules of this plant (Pachytesta il- linoensis (Arnold & Steidtmann) Stewart, 1954) hung pendulously from beneath terminal pinnules. Synangiate prepollen organs (Bernaultia formosa (Schopf) Rothwell & Eggert, 1986) containing large prepollen (Schopfipollenites ovatus (Schopf) Po tonié & Kremp, 1954) dangled in pinnate struc- tures replacing parts of fronds. These fructifica- tions were among the largest and most conspicuous of Carboniferous swamps, and the prepollen grains so unusually large and heavy that insect pollination is likely. Among the great variety of Late Carbon- iferous insects known, Paleodictyoptera such as Homaloneura | dabasinkasi Carpenter (1964), shown in our reconstruction, were the most likely pollinators of these plants. After pollination, the megagametophyte differentiated and the pollen chamber was sealed by continued growth of the outer integument. The large (2.5 by 4.5 cm) ovules had well-developed fleshy layers of the kind at- tractive to animal dispersers. Their sclerotesta could have protected them from crushing and acids of the digestive tract. Animals of that time sufficiently large to swallow such ovules included reptiles, am- phibians, fish, and sharks. With large food reserves, the young seedlings could grow in the shade and tolerate other adverse influences of competing vegetation. Although this plant is associated with disturbed, nutrient-rich parts of swamps, it was probably not the earliest successional colonizer, but a later species in plant succession. Evidence for reconstruction. Our reconstruc- tion is based on remains preserved in coal balls from the bed of Sugar Creek 3.7 km northwest of Berryville, Lawrence County, Illinois. Another well- known locality is a tributary of Bonpas Creek 4.5 km east of Calhoun, in nearby Richland County, Illinois. These are both from the Calhoun Coal of the Mattoon Formation in the McLeansboro Group, of Late Pennsylvanian (Stephanian) age (Phillips, 1980), or about 296 million years ago (in time scale of Palmer, 1983). Our reconstruction ma also be valid for fossils from coal balls of the Herrin No. 6 coal of slightly older (Westphalian D) Penn- sylvanian age, where many of the species gathered together here have been found in association (Phil- lips & DiMichele, 1981). Because of the probability of mosaic evolution, we do not mean to imply that the whole plant existed in every locality where some small part of it has been found. The attribution of these various remains to one plant is based on the — age, from Berryville, Illinois, CS AS Schematic cross section of xylem and petiole traces of stem.—D. Det , H. Lower and upper (respectively) cuticles. —I. Atta sections of ovule, showing sclerotesta (stipple and rectilinear cellular pattern) and va apical en gagametophyte supposed sperm (black and subtriangular, above) .— Q. Prepollen on hai g of insect. — R. 2 M and internal secretory glands (black FIGUREÓ. A reconstruction of dus d illinoensis of Late Pennsylvanian (Late Carboniferous or Do" . Habit as a palmlike tree o of adventitious ho showing secondary xylem (cellular pattern) .—C. Cut well-drained parts of swampland.—B taway model eaf.—F. Cross section of stomate. chán longitudinal and transverse scular strands (black) .— —N. Detail of prepollen chamber, verse sectio section epollen organs and Cross section d prepollen organ, showing glandular hairs (outer ). 1028 Annals of the Missouri Botanical Garden anatomical similarity of petioles attached to leaves, stems, and prepollen organs (Ramanujam et al., 1974); the similar pollen found in prepollen organs and ovules (Taylor, 1965); and close association of different organs in coal balls (Schopf, 1948; Taylor, 1965; Phillips, 1981). Although preserved as petrifactions, splitting and degaging of coal balls has revealed the nature of the leaf, which conforms in shape and venation to Alethopteris lesquereuxii var. cerverae Wagner, 1968 (Stidd, 1981), and has a cuticle (Ramanujam et al., 1974; Oestry-Stidd & Stidd, 1976; Reih- mann & Schabilion, 1976, 1978) generally similar to that of the compression species 44. davreuxii (Barthel, 1962). There are other alethopterid leaves in coal balls at the main locality for this plant (Berryville, Illinois), and these have blunter, short- er, and straighter pinnules, more like the impres- sion species A. bohemica and A. grandinoides var. subzeilleri (Mickle & Rothwell, 1982). Similar impression fossils have been found in nearby In- diana with numerous small ovules attached (Taylor, 1981, figs. 13-18A, B), and it is likely that this second type of frond is the foliage of other small fructifications, such as Pachytesta berryvillensis and Dolerotheca villosa, also found in Berryville coal balls (Schopf, 1948; Taylor, 1965; Phillips, 1980). There is only one other allied species in these coal balls (Pachytesta hexangulata), and this is much rarer than the other two (known from only one specimen: Taylor, 1965). Habit. linoensis follows that of Stewart & Delevoryas (1956), which is drawn as if it were a young plant. Fossil stems of this plant range from 10 to 50 cm Our reconstruction of Pachytesta il- in diameter. They probably were about 5 m high and occasionally attained heights of 10 m (Wnuk & Pfefferkorn, 1984). I branches have not been found on long stem 1984), so it prob- t is a common plant, yet compressions (Pfefferkorn et al., ably conformed to Corner’s architectural model (of Halle et al., 1978), as do modern tree ferns such as mamaku (Cyathea medullaris) of New Zealand. As in modern tree ferns, Pachytesta illinoensis probably had a crown of about ten large leaves, oriented so that a good deal of the trunk was visible from the side (Wnuk & Pfefferkorn, 1984). Like these modern plants also, it was perennial, growing slowly by the development of new leaves above the leaf bases of the old crown. The trunks were strengthened somewhat by several bundles of wood. These were not separate steles, but rather a single eustele with secondary xylem and phloem devel- oped in several separate bundles (Basinger et al., 1974; Stewart, 1983; Smoot, 1984b). The base of the plant was invested in adventitious roots, although these did not cover the stem to the same extent as in associated fossil tree ferns, and it was more like New Zealand mamaku in this respect. Like this modern plant also, the fossil trees lacked lacunar spaces in their roots (Phillips, 1981). e large, hemispherical, compound prepollen organ (microsynangium) included numerous elon- gated sporangia oriented perpendicularly to the flat side. The pattern of their open ends in this flat — lower) side is almost radial and has been inter- preted to be homologous to an infolded stack of bladelike synangia (Rothwell & Eggert, 1982, 1986) and, alternatively, to a branch of pinnately dee — a sporangia (Dufek & Stidd in Stidd, 1981). ering "d number of vascular strands in its petiole Ramanujam et al., 1974), each large synangium is the morphological homolog of a primary pinna of a frond. We do not think that prepollen organs were scattered randomly among otherwise sterile y either arrangement and consid- — fronds, an impression that could be gained from uncritical inspection of the reconstruction of Ra- manujam et al. (1974). They were probably borne pinnately in special fertile fronds or parts of fronds. Such an arrangement has been demonstrated for other synangia of closely comparable structure (Po- toneia), but containing different prepollén (Taylor, 1982, pl. 6, fig. Ovules of EE illinoensis have not been found attached, but their mode of attachment can be inferred from similar compressions of ovules found attached under the terminal pinnule of the leaves (Crookall, 1959). We do not think that they formed at the end of the pinna (as reconstructed by Darrah, 1939: 95). Close inspection of these specimens often reveals the terminal pinnule twist- ed into a less-conspicuous upright position beside the heavier and larger ovule (Crookall, 1959: 27, figs. 1-5). Other related species of ovule have been found attached farther back on the rachis of the pinna (Halle, 1929; Wagner, 1968; Zodrow & McCandlish, 1980). Reproduction. "The prepollen of Pachytesta illinoensis had an alveolate exine ultrastructure and was so large (300-350 um long by 200-250 um wide) and heavy that it would not have been dispersed far by wind. It was more likely dispersed by insects (Taylor, 1978). The glandular hairs of the microsynangium and the resinous internal glands of both the microsynangium and ovules may have produced a nutritional reward for pollinating insects or have deterred their herbivory. Both organs also Volume 75, Number 3 1988 Retallack ds Dilcher Seed Fer 1029 were enclosed in fleshy tissue, which may have been nutritious. It is uncertain which among the great variety of Carboniferous insects known could have polli- nated this plant, but several circumstantial lines of evidence implicate Paleodictyoptera. These were superficially like modern dragonflies, but, unlike these modern carnivorous insects, had only narrow sucking mouthparts (Kukalova, 1970). The ex- ample we have shown in our reconstruction is Ho- maloneura dabasinkasi Carpenter, 1964, known largely from wings with camouflage color-banding in siderite nodules of slightly older Late Carbon- iferous age (Westphalian D) near Braidwood, IMi- nois. Paleodictyopterans were active, flying insects, with large compound eyes and hairy legs: all fea- tures of modern insect pollinators. Their sucking mouthparts match in size and shape the fine borings seen in petrified microsynangia and ovules (Schopf, 1948). The claim of paleodictyoptera as pollinators of these plants could be regarded as diminished by the discovery of prepollen like that of Pachytesta illinoensis lodged in the leg joints of a large fossil millipedelike creature (Scott & Taylor, 1983), but it is unlikely that these extinct arthropods were any more adventurous in the canopy than modern litter-feeding millipedes (Rolfe, 1980). Coprolites containing these prepollen also have been found (Scott, 1977), but these cannot be attributed to any particular animal. A variety of animals prob- ably ate these grains, both in the canopy and on the ground. As in all the seed ferns considered here, there was a period of time between pollination and fer- tilization. For ovules with such large megagame- tophytes this was probably at least a few weeks. Fossil prepollen grains within ovules of Pachytesta illinoensis have been found with two large black bodies (Stewart, 1954), which have been compared to the ciliated, motile sperm of modern cycads (Chamberlain, 1935). Such opaque "spots" in pet- rified cells also could be remains of decayed cy- toplasm, as paleobotanists have learned from pain- ful experience in the study of other fossils (Knoll & Barghoorn, 1975). However, there is other pos- sible cytoplasmic material in the medullosan pollen grain in question. The ovule of Pachytesta illinoensis was con- spicuous in its large size (about 2.5 cm in diameter and up to 4.5 cm long). Like many modern stone fruits, the fossil ovules had well-differentiated sar- cotesta and sclerotesta. The sarcotesta was espe- cially thick near the om which was early to decay or be eaten, judging fro imens ( 1954). This succulent "flesh mies have attracted tewart animal dispersers. The sclerotesta can be seen to seal the micropyle in some specimens (Taylor, 1965). Similar stones in modern fruits serve to withstand crushing and acidity during passage through the guts of animals (van der Pijl, 1972). Of the array of known Late Carboniferous reptiles and amphibians (Moodie, 1916; Reisz et al., 1982), edaphosaurs, temnospondyls, and microsaurs were large enough to have been able to swallow a fruit this size. Considering the likely habitat of this plant, the ovules could have been dispersed by fish, as are some modern angiosperm fruits in the varzea swamp forests of Amazonian Brazil today (Gotts- berger, 1978). Large palaeoniscid fish and pleu- racanth sharks were common within parts of the Late Carboniferous swamps of Illinois (Zangerl & Richardson, 1963; Zangerl & Case, 1973). Individuals of Pachytesta illinoensis and relat- ed plants appear to have borne fruit in massive bursts, judging from local accumulations of petri- factions and molds-and-casts of similar large ovules Seward, 1917, fig. 423; Tonks et al., 1931, pl. II; Taylor, 1965; Jennings, 1974). Such a repro- ductive effort probably was not begun until after a few years in the life of the plant. In modern angiosperms, copious production of fruit ensures — that some remains unmolested by animals and is especially common in trees with large fruits (Jan- zen, 1978). Habitat. in the manner already described for Lagenostoma lomaxii. Although Pachytesta illinoensis grew in and near swamps, it lacked root lacunae usually found in plants of waterlogged ground and thrived in slightly elevated and nutrient-rich parts of the swamp (Phillips, 1981). A similar habitat is indi- cated by the bundle sheath cells (Krantz anatomy) found around veins within petrified leaves, thought by Baxter & Willhite (1969) to indicate the C, photosynthetic pathway for this extinct plant. Like modern plants using this pathway (Chazdon, 1978), Pachytesta illinoensis may have been able to en- This plant was preserved in coal balls dure hotter, sunnier, and drier sites than many associated plants. Leaves of the plant had thick cuticles, especially on the adaxial surface. Stomates and hairs were confined to the abaxial surface, and each stomate was overhung by papillae borne on the subsidiary cells (Ramanujam et al., 1974; Oestry-Stidd € Stidd, 1976; Reihmann & Schabilion, 1976, 1978). These are all features of sun leaves of plants subject to deficiency of water or nutrients (Mickle & Roth- well, 1982) Such large ovules would have been able to main- 1030 Annals of the Missouri Botanical Garden a! E SA Ss) D * e OLI ANO €// / neis Ó + WI, sim, (50 RALIS "Pr, c > ... A 2s ` “ L “ay Lh TFAL TITELN 4, THB re ^, m JI TU) 0% "i WY a SAN ^h ' 109.6 HH Tre Volume 75, Number 3 1988 Retallack & Dilcher 1031 Seed Ferns tain the growth of seedlings, even under the shade and other adverse influences of other plants, as in modern plants of the “competitive strategy" (of Grime, 1979). Although Pachytesta illinoensis is geologically associated with disturbed and more- elevated parts of the swamps, it may have been a late species in succession, growing through and then shading out the earliest successional plants. In better-drained areas outside the swamps, similar plants may have formed stable, long-term com- munities, but within the swamp itself these plants were less persistent than associated tree ferns and lycopods (Phillips & DiMichele, 1981). udging from paleobotanical and palynological studies of Calhoun coal balls (Phillips et al., 1974; Phillips, 1980), Pachytesta illinoensis formed a minor part of vegetation dominated by marattia- ceous tree ferns (with Psaronius spp. trunks and Scolecopteris spp. and Asterotheca spp. fructifi- cations). Lycopods (Sigillaria ichthyolepis) and horsetails (Calamites retangularis) were moder- ately common, and the rest of this diverse flora was made up of rare cordaites, other ferns, and seed ferns Ecologically, Pachytesta illinoensis probably slightly dryer, nutrient-rich parts voforred elevate r of the swamp where there was either sandy or clayey soil or exposed dry, dusty peat. It was not the earliest successional colonizer of such habitats. In shale floras of stream and lake margins (Scott, 1979), calamites and small ferns appear to have occupied that niche. In coal balls of peat swamps, successional patterns are less clear (Phillips & DiMichele, 1981), but weedy plants of erratic oc- currence include other saed fernis Gea as Callo- re), and small spermarion pusillum herbaceous lycopods and Erde Pachytesta illinoensis or allied plants part of ma- rine-influenced swamps (mangroves) at this time. This habitat was dominated by cordaites (Raymond & Phillips, 1983). Nor were Paleogeographic setting. The principal lo- cality for our reconstruction of Pachytesta illi- noensis is in Sugar Creek near Berryville, Illinois, of late Pennsylvanian (Stephanian) age (Phillips, 1980). Illinois was then part of a large central lowland within North America, bordered on the west by mountains in what is now Nevada and Montana, on the north by low hills and limestone plateaus on the Canadian Shield, and on the east by the Appalachian Mountains. These mountains were much more imposing than they are today, and more like the present European Alps (Heckel, 1977). Much of the mid continent was inundated by shallow seas opening out on deep ocean in the area of modern Texas. At the time the Calhoun Coal formed, the sea extended only as far as west- ern Illinois and Missouri. Most of central Illinois was covered by a swamp, dissected by a large stream system draining the granitic shield and car- bonate plateau country to the north. This ancient river is usually called the “Michigan River" but is in some senses a forerunner of the Mississippi (Stan- ley, 1985). To the west in Indiana and Ohio were more swamps and riverine lowlands. The present area of Illinois was about 5? north of the equator at this time, on the margin of the tropical humid and subtropical subhumid belts (Heckel, 1977). Evaporites formed in the dry climate of the present area of the Black Hills of South Dakota. Climate became increasingly humid southeastward toward the equatorial Appalachian Mountains (Schopf, 1975) CALLOSPER MARION PUSILLUM Hypothesized reconstruction. We envisage Callospermarion pusillum as a scrambling fernlike vine (Fig. 7). It probably formed dense tangles in moist, disturbed sites within and around Late Penn- sylvanian swamp forests of tree ferns. Leaves of this plant (Dicksonites pluckenetii (Schlotheim) — FiGURE 7. A reconstruction of Callos Tr pusillum of Late Pennsylvanian (latest Late Carboniferous or Bich hanian) age, from near Cien Illinois, U.S.A.—A. Habit as a swampland early successional scrambling vine.—B, C. Schematic cross se and root, showing primary xylem (black), secondary xylem (cellular pattern) , and m root traces.—E. Fertile prora bearing leaf.—F. Pollen organ.—G. pigesemerophos ne in til of ov pollinatio Overview and cross sectio du d prats (hachured). multicellular megagametophyte (equant cellular p Monosaccate pollen from pollen (circle with central stippl ons ui cortex radial shading). — D. Xylem model of a node with petiole, bud, and Monosaccate pollen with 2- and 4-cell a“ pee leaf. — I. Longitudinal section of young ovule.—J. Longitudinal section on of mature ovules, showing vascular stran ds chamber of ovule, showing a 1 pollen ) .—O, P. ross section fle shoving vascular bundles (hachured) , ino cells (open ps. and palisade tes ube. 1 layer (ena lines) .—R, S. Upper r and lower cuticles, both with stoma 1032 Annals of the Missouri Botanical Garden Sterzel, 1881) had stomates on both sides and little differentiation between the upper and lower sur- face. The clambering stem (Callistophyton porox- yloides Delevoryas & Morgan, 1954) was woody and perennial. In addition to horizontal runners, there were slender erect leafy stems and, at inter- vals, stout adventitious roots, which arose in the axils of foliage leaves or from within bifurcations of the main axis. Pollen-bearing organs (/dano- thekion callistophytoides (Stidd & Hall) Rothwell, 1980) and ovules (Callospermarion pusillum Eg- gert & Delevoryas, 1960) were borne in zones on the abaxial surface of foliage leaves otherwise nor- mal in appearance. The numerous pollen organs released abundant small saccate pollen (Vesicas- pora schaubergeri (Potonié & Klaus) Jizba, 1962). Pollination may have been by wind and possibly occurred before the megagametophyte or pollen chamber was fully differentiated. After fertilization, which may have occurred on the ground, the seeds were already worn back to their sclerotesta. The numerous small seeds were probably dispersed largely by wind and water. Evidence for reconstruction. This plant is best known at the locality near Berryville, Illinois, al- ready discussed for Pachytesta illinoensis, and also is known from nearby coal-ball localities such as the one near Calhoun, Illinois. Reconstruction of Callospermarion pusillum was based on the similarity of secretory cavities in stems, sterile fo- liage, ovules, and microsporophylls; on the occur- rence of similar pollen in the pollen organs, in the ovule pollen chamber, and dispersed; and on close association of its parts (Stidd & Hall, 1970a, b; Millay & Taylor, 1974; Rothwell, 1975, 1980, 1981). Habit. drical mass of secondary xylem, a thin zone of The young stems have a nearly cylin- secondary phloem, and a subsurface zone of me- chanical tissue formed from anastomosing bands of sclerotic cells (dictyoxylon cortex: Rothwell, 1975; Smoot, 1984a). These were probably erect, leaf-bearing stems. Older stems do not all have such well-developed mechanical tissue, but rather a thick bark and a large central asymmetric mass of secondary xylem. These were probably horizon- tal runners. Stout adventitious roots arose from these runners, usually in the axils of leaves, but also within the branches of the main axis (Kidston, 1924; Rothwell, 1975). Buds, petioles, and ad- ventitious roots often arise from within the di- chotomies of stems and leaves. These are false dichotomies (Kidston, 1924) arising by develop- ment of opposite branches with suppression of the terminal meristem. This branching system is similar to Leeuwenberg’s architectural model of plants pro- posed by Hallé et al. (1978), although the two are not strictly compara ble b se the leaf-borne fruc- tifications of seed ferns are very different from those of most modern plants. A similar architecture is seen in the modern Australian and Southeast Asian fern Gleichenia dicarpa. The synangiate pollen organs were attached on the underside of the leaves in a manner similar to sporangia of marattiaceous ferns (Kidston, 1924; Stidd & Hall, 1970a). They contain, however, monosaccate pollen rather than fern spores. The saccus is laterally expanded, so that it is superfi- cially similar to bisaccate grains (Hall & Stidd, 1971; Millay & Taylor, 1974). As in modern bi- saccate grains, the saccus of the fossil pollen formed by detachment and expansion of the sexine from the lamellated nexine. These fossil pollen also were similar to those of modern gymnosperms in ger- minating through the side of the grain opposite the point of attachment in the original tetrad, unlike the prepollen of the other Carboniferous seed ferns already considered (Eggert & Millay, 1976). Ovules and their attachment scars are found over large areas of compressed fronds (Grand'Eury, 1905; Kidston, 1924). Ovules were attached at the ends of veins near the abaxial margin of the pinnules, and their micropyles faced inwards to- ward the pinnule midribs (Loubiére, 1929; Roth- well, 1980). Reproduction. | Callospermarion pusillum was probably wind pollinated, considering the moderate size (37-54 um long by 30-49 um wide: Millay & Taylor, 1974) of its pollen and the large amount of it produced. Saccae sometimes have been con- sidered the “wings” of wind-dispersed pollen, but they are now thought to have served more for orientation and flotation in an inverted pollination drop (Doyle, 1945). Microgametophytes at the two- and four-celled stage have been found in some synangia (Millay & Eggert, 1974), indicating that these were fully formed at the time of pollination. In some coal balls, ovules have been found which vary considerably in degree of maturity (Rothwell, 1971). Large fructifications of modern weeds of disturbed habitats show comparable “continuous flowering," thus improving chances that some propagules will mature at an appropriate time (Heinrich, 1976). The youngest ovules found are small and weakly differentiated. At what is thought to have been pollination stage, the ovule was small, with an imperfectly differentiated seed coat and no indication of a megagametophyte. A fossilized pol- Volume 75, Number 3 1988 Retallack & Dilcher 1033 Seed Ferns lination drop has been reported in a closely related species (Callospermarion undulatum: Rothwell, 1977), but it appears rather more resinous than modern pollination drops (described by Doyle & O'Leary, 1935b) and contains spores known to belong to other plants (Rothwell, 1980). It may be an exudate from the micropyle of a partly decaying ovule rather than a pollination drop. Nevertheless, larger ovules than this do contain pollen grains and have a sealed micropyle, thus suggesting that ovules were at this stage of development at the time of pollination. With further growth after pollination, the micropyle was occluded by growth and differ- entiation of sclerotesta and sarcotesta as the mega- metophyte developed. In the most mature ovules, presumably already dispersed, there were several archegonia at the apical end of the megagameto- phyte, and the sarcotesta was abraded away. A germinated pollen grain with a branched pol- len tube has been found in the pollen chamber of an immature ovule of a related species (C. undu- latum; Rothwell, 1972). It could be that this pollen tube merely served for nutrition and stabilization of the pollen grain during its long wait until fer- tilization was achieved by motile gametes. Alter- natively, this pollen tube, or at least one branch of it, may have delivered nonmotile sperm nuclei to the archegonium (siphonogamy), as in living conifers such as black pine (Pinus nigra — P. laricio in Chamberlain, 1935). Rothwell (1980) was impressed with the coniferlike development and morphology of this plant's ovules, pollen, and pollen tubes, and thought that siphonogamy was more likely. The small ovules were dispersed individually, as can be inferred from compressed leaves showing only the attachment scars where ovules had ab- scised (Kidston, 1923). The ovules were numerous, small, and somewhat flattened, with narrow wings. They were probably dispersed largely by wind and water. The ovules were made in quantity, rather than quality, thus maximizing the chance that at least a few would find places suitable for germi- nation, as in modern weedy plants (*ruderals" of Grime, 1979) Habitat. This reconstruction is based on spec- imens from coal balls found in Sugar Creek north- east of Berryville, Illinois (Rothwell, 1975, 1980), and it is known from coal balls of comparable age at several other localities in the mid-continental U.S.A. (Phillips, 1980). This species of compressed leaves (Dicksonites pluckenetii) has long been known from compression floras of Britain and France (Grand'Eury, 1905; Kidston, 1923), where allied species of petrified plants have also been found (Loubiére, 1929; Rothwell, 1981). Like Pachytesta illinoensis reconstructed here, it was locally abundant in swampy lowlands. Several features of Callospermarion pusillum can be construed as evidence that it was a plant of moist, somewhat shady understory habitats. The pinnules are attached along the middle of the rachis of the frond (Rothwell, 1975), and there is little differentiation between the upper and lower cuticle, both of which are stomatiferous (Barthel, 1962). Considering this and its weedy reproductive fea- tures, we hypothesize that Callospermarion pu- sillum was an early successional plant of disturbed ground. This may have included stream margins, tracts of forest devastated by hurricanes, locally well-drained peat, forest defoliated by fire, or light gaps around fallen trees. It lived among swamp forests dominated by tree ferns during the late Pennsylvanian (Stephanian: Phillips, 1980). Paleogeographic setting. This plant is known from the locality of Pachytesta illinoensis already described. Callospermarion pusillum lived in very similar climate and general environment. DICTYOPTERIDIUM SPORIFERUM Hypothesized reconstruction. In our view, Dictyopteridium sporiferum was a swampland tree, widely distributed in cool-temperate regions of the Gondwana supercontinent during Late Permian time, some 245 to 253 million years ago (Fig. 8). n Queensland and New South Wales, it grew in an extensive system of intermontane valleys west of a volcanic mountain range like that of the mod- ern South American Andes, and east of the plains and hill ranges of inland Australia. Its tongue- shaped leaves (Glossopteris communis Feistman- tel, 1876) were seasonally deciduous. They had a thick adaxial cuticle and stomates sunken into the abaxial surface. Their venation was reticulate, but not organized into veins of different thickness, as in modern angiosperms. Its wood (Araucarioxylon bengalense (Holden) Maheshwari, 1972) was mas- sive and coniferlike, with clear growth rings. Large roots spread out horizontally from the trunk, and the rootlets (Vertebraria australis McCoy, 1847) had internal chambers, which may have allowed growth in oxygen-poor, waterlogged, peaty soils. Ovule-bearing (Dictyopteridium sporiferum Feist- mantel, 1881) and pollen-bearing structures (£r- etmonia sp. cf. E. hinjridaensis Surange & Ma- heshwari, 1970) were borne on the midrib of leaflike structures. These were arranged helically on fertile 1034 Annals of the Missouri Botanical Garden x = PL IU - G x X x x= Volume 75, Number 3 1988 Retallack & Dilcher Seed Ferns 1035 short shoots, but it is not certain whether these shoots were bisexual or unisexual. Pollen sacs (Ar- beriella africana Pant & Nautiyal, 1960) were borne at the ends of slender, copiously branched stalks. They dehisced longitudinally to release nu- merous bisaccate, striate pollen grains (Protohap- loxypinus limpidus (Balme & Henelly) Balme & Playford, 1967). These plants were probably wind pollinated. Numerous ovules (Stephanostoma crystallinum (Pant) Pant & Nautiyal, 1960) were borne on the underside of a leaflike organ and loosely enclosed by its inrolled margins. The ovules were interconnected by a meshwork of multicellular hairs and probably also a good deal of mucilage. y the time of fertilization the leaflike structure bearing ovules probably was unfurled, shrivelled, and decayed lightweight. They probably were scattered by wind. . Seeds were numerous, small, and Evidence for reconstruction. This reconstruc- tion is based on petrified material from near the Homevale-Elphinstone road at the southwest prop- erty boundary of Homevale Station (or "cattle ranch"), 87 km west-southwest of Mackay, Queensland, Australia (Isbell, 1955; Gould & De- levoryas, 1977). These petrified peats are part of the Fort Cooper Coal Measures, Blackwater Group, of Late Permian age (Jensen, 1975) or about 245 to 253 million years old (in time scale of Palmer, 1983). In these petrified fossils, sterile leaves are anatomically identical to leaflike structures bearing ovules; pollen is present in pollen sacs and in ovules; the same kind of tracheids are found in leaves and trunks; and identical secondary xylem is found in trunks and around large septate roots (Gould & Delevoryas, 1977 These petrified fossils have not been named but appear identical to several compression fossils from the same region. The petrified fructifications are the same size and elongate shape and enclose nu- merous wingless ovules of the same size as com sion fossils of Dictyopteridium sporiferum en mantel (1881) from Late Permian coal measures of India (Surange & Chandra, 1975), Queensland (as “Cistella bowenensis" of White, 1964, and “Plumsteadia microsacca” of Rigby, 1971, 1978), and New South Wales (Holmes, 1974; White, 1978). The petrified material also agrees closely in size and anatomy with compression fossils of ovules and pollen sacs comparable in age from India (Pant & Nautiyal, 1960). The petrified pollen or- gans have up to 17 pollen sacs in a cluster (Gould & Delevoryas, 1977), more like the compression genus Eretmonia than Glossotheca (as defined by Surange & Chandra, 1975). Impressions of Fr- etmonia from New South Wales (White, 1978) and a variety of “scale fronds" from Queensland (White, 1964) have a thickened tip, most like Eretmonia hinjridaensis Surange & Maheshwari 1970). Most of the pollen from petrified remains (Gould & Delevoryas, 1977) is identical to the broadly defined dispersed species Protohaploxy- pinus limpidus (Balme & Henelly) Balme & Play- ford, 1967 (Foster, 1975; Rigby & Hekel, 1977). Impressions of Dictyopteridium sporiferum from Queensland have been found attached to leaves of Glossopteris communis Feistmantel, 1876 (White, 1964; Rigby, 1971). Glossopteris communis is a common fine-meshed leaf, like G. indica, G. lin- —. earis, and G. angustifolia. This last-mentioned species also has been considered the leaf of Dic- tyopteridium sporiferum, although on less secure evidence of association and venation density (White, 1964; Maheshwari, 1965). The petrified ovular heads and associated leaves (Gould & Delevoryas, 1977; G.J.R., pers. obs.) were also fine-meshed (veins about 0.2 mm apart), but they have a clearly defined midrib of a size most like that in Glossop- teris communis. Associated coarse-meshed Glos- sopteris leaves in the Late Permian coal measures of Australia belonged to quite different plants, as shown by their attachment to nearly circular ovu- late fructifications bearing winged seeds (Holmes, 1974; White, 1978). The petrified leaves of the — A reconstruction y Doa paea MN of Late Permian age, from near Homevale Station, . Hab FIGURE 8. Queensland, Australia. as a sw tangential and a 4n sections eens — secondary wood.— E Xylem model of small chambered root.—H. oodland tree. 'ulaway reconstruction of secondary wood of basal trunk —B-D. Cellular details of wood structure in m at points between arms of of leaf, dnd vascular bundles of midrib (circles with curved hachure) , their sheathing sclerenchyma (black ellipses), u stomate, showing cuticle (heavy outline) and cell walls (dotted and broken lines). ovulate fertile structure nin ovules, showing vascular traces (black) , woody integumentary e (irregular pattern), and a ongituc in ons layer ppl mull megagamet ipple) .— alisade cells (vertical lines), and spongy mesophyll (irregular pattern) .—J. Cross section of Transverse and —K-M. archegonium (circle with Pod r and upper ae cuticles of leaf, with stomates on under side only.—P. Po A M sida —Q. Dehisc ent sporangia.—R. Str iate bisaccate pollen 1036 Annals of the Missouri Botanical Garden plant reconstructed here are hypostomatic, with low papillae on the abaxial side, similar to the cuticles of the compression species Glossopteris waltonii Pant & Gupta o after which our cuticular restoration was moc etrified wood of this lant. both within trunks and its distinctive chambered roots, includes late wood like Araucarioxylon arberi (Seward, 1919) Maheshwari (1972) and early wood like 4. ben- galense (Holden, 1917) Maheshwari, 1972 (Gould, 1975). We have chosen the latter name for our reconstruction on the grounds of priority. There have been some nomenclatural problems also with the name for the root of this plant, which Schopf (1982) decided should be Vertebraria aus- tralis McCoy (1847). Only one species of Verte- braria is recognized. This fossil species probably includes remains of what were roots of numerous species of glossopterid plants recognized from their reproductive structures. Habit. reconstructed from specimens of trunks showing alternate and whorled branching (Gould & Dele- voryas, 1977). The whorled specimen conforms to Rauh's architectural model of modern trees (Hallé et al., 1978). The other trunks may represent other growth forms but also could be old or damaged The overall habit of the plant has been trunks. Regularly whorled young trees and exten- sively repaired and irregular old trees are char- acteristic of many living conifers, of Ginkgo bi- loba, and some angiosperms (Hallé et al., 1978; Retallack & Dilcher, 1981). The wood of this plant was like that of modern softwoods, with very narrow unicellular rays and abundant pitting in the cross eld. It was a kind of wood apparently conservative in gymnosperms and still found in many conifers and in Ginkgo biloba (Beck, 1971) Leaves were arranged in close helices on short shoots, which in turn were arranged on long shoots (Pant & Singh, 1974). The venation of the leaves was a fine mesh, with a zone of much narrower meshes and a sclerenchyma sheath forming a mid- rib. There is a good deal of evidence that these plants were seasonally deciduous: the noncoria- ceous nature of the leaves compared with those of associated and presumably evergreen conifers, well- developed abscission scars at the bases of the pet- ioles, and well-marked growth rings in its fossil wood (Gould & Delevoryas, 1977; G.J.R., pers. obs.). A variety of triangular scalelike leaves have been found in association with these fossils (Wal- kom, 1922; White, 1964, 1978). Some of these were fertile scales, but others may have been young leaves or protective scales of dormant winter buds. Fertile structures were attached to leaflike or- gans arranged in a closely spaced helix (Pant & Singh, 1974). Fertile scales arranged on the same short shoot as sterile leaves would have formed distinct clusters (White, 1978). shoots there were up to three kinds of scales or leaves (White, 1978), but there is no evidence that any of these shoots were bisexual. Pollen sacs were borne at the end of copiously On some short dichotomizing stalks arising from the midrib of a scale leaf. Although much reduced, this epiphyllous structure is similar to that found in seed ferns such as Telangium affine (Kidston, 1923) and Diplo- pteridium teilianum (Walton, 1926, 1931). In many of these Early Carboniferous plants, the spo- rangia were fused into bell-like synangia, but in Dictyopteridium sporiferum the sporangia were free, as in other enigmatic Early Carboniferous plants (Skog & Gensel, 1980) and Late Devonian progymnosperms (Beck, 1981). The walls of each pollen sac were only one cell thick (Gould & De- levoryas, 1977). They opened by way of a long, sinuous slit (Pant & Nautiyal, 1960), which pre- sumably developed because of diagonal stresses arising during drying. Ovules were borne on the underside of an in- folded leaflike structure which was attached by a long stalk to the midrib of the adaxial side of what looks like an ordinary foliage leaf. This epiphyllous structure also can be compared with fructifications of Early Carboniferous seed ferns, such as Sphe- nopteris bifida (Long, 1979b). The structure of these fossils was anticipated in some studies of impression fossils (Schopf, 1976) but did not be- come clear until well-preserved n fossils were studied (Gould & Delevoryas, 77). In earlier studies of impressions, they were sies to have been bivalved cupules (Plumstead, 19582) or brac- teate cones (Surange & Chandra, 1975). The first view is now thought incorrect, but the second in- terpretation has gained some support from exper- imental compaction of model structures (Rex, 1986) and from cuticular studies of compressed speci- mens (Chandra & Surange, 1976). In our view, however, the three separate cuticles can be ex- plained as (1) a central abaxial hair-bearing cuticle with holes marking positions of ovule attachment; as (2) a peripheral abaxial stomatiferous cuticle with hair bases; and as (3) an adaxial thick non- stomatiferous cuticle. A comparison of experimen- tally deformed structures with compression fossils indicates that some real biological diversity may be reflected in the large number of generic names for glossopterid fructifications (reviewed by Rigby, 1978), but some of them may be merely different developmental and preservational states of the same kinds of fructifications (Gould & Delevoryas, 1977). Volume 75, Number 3 Retallack & Dilcher 1037 1988 Seed Ferns Reproduction. Each pollen sac contained nu- and its roots riddle the drab underclays to coal merous pollen grains. The combined release of pol- len from a stand of trees could have produced clouds of yellow dustlike grains. Considering the amount of pollen produced and its moderate size (32-46 um in breadth and 14-26 um in corpus diameter: Gould & Delevoryas, 1977), this plant was probably wind pollinated. As in Callosper- marion pusillum, the saccae may have served to orient and float the grains in a pollination drop. Stout striae on the body of the pollen grain also may have functional significance, because similar striate bisaccate grains were also produced by ap- parently unrelated seed ferns in Permian coal swamps of Siberia (Meyen, 1984). Mormon tea (Ephedra spp.) has pollen with bands that strength- en the grain against stresses arising from desic- cation in a dry climate (Hughes, 1976). Striae on the pollen of Dictyopteridium sporiferum may have served to withstand stresses associated with moisture losses during wind pollination. At the time of pollination the ovule-bearing head was infolded to enclose the ovules and an interovu- lar mesh of unicellular hairs. Rigby (1978) has suggested that these filaments were fungal hyphae, from decay of the fossil. This seems unlikely con- sidering their radiation from the micropyles of the ovules without plugging them, the apparent lack of damage tissue in petrified specimens (Gould & Delevoryas, 1977), and the cuticularized hair bases seen in macerated preparations (Pant & Nautiyal, 1960; Chandra & Surange, 1976). The filamen- tous meshwork may have produced or trapped fluid which bulged from the gaping lips of the underside of the ovule-bearing structure, like pollination drop- lets. Pollen of Dictyopteridium sporiferum en- trapped in this material and floated and oriented by their inflated saccae would have been pulled into the micropyles of the ovules inside as the fluid dried back along the guiding filaments. Ovule-bearing heads have been found withered and open, partly or wholly naked of ovules (Gould & Delevoryas, 1977). The seeds of this plant lacked wings or fleshy layers found in associated fossil plants (Walkom, 1922; Holmes, 1974; Plumstead, 1963). They were numerous, small (0.8-1.5 mm long: Gould & Delevoryas, 1977), and possibly were shaken from the fructifications by swaying in the wind (a ““wind ballist” of van der Pijl, 1972), then dispersed widely by wind and water. The hair bases appear to have been quite persistent on dis- persed seeds (Pant & Nautiyal, 1960) and may have aided dispersal. Habitat. Leaves of this plant are abundant in roof shales and clastic partings within coal seams, seams (Jensen, 1975; Retallack, 1980b hese plants lived in waterlogged muds, silts, and sands (gleyed Entisols of Soil Survey Staff, 1975), as well as in peaty organic substrates (Histosols of Soil Survey Staff, 1975). Trunks of this plant (Jensen, 1975, fig. 45; Gould, 1975) and of glossopterids in general (Da- vid, 1907; Plumstead, 1958b) had shallow root systems. This kind of tabular root system is com- monly found in modern plants of waterlogged hab- itats (Jenik, 1978). Such a habitat also may explain the peculiar construction of the roots of these plants, in which xylem is restricted to narrow radial arms and transverse platforms that enclose empty spaces within the root. There has been some concern whether these spaces were filled with parenchyma, but calluslike groups of parenchyma cells on the walls of mature examples is evidence that they were originally empty (Gould, 1975). They may have served for aeration of the root in oxygen- poor ground, as in living crack willow (Salix fra- gilis: Kawase & Whitmoyer, 1980). The texture of the leaves, with an adaxial thick cuticle and an abaxial stomatiferous and papillate cuticle (Gould & Delevoryas, 1977) is similar to sun leaves of modern angiospermous trees (Salis- bury, 1927). This, together with the abundance of these leaves, is an indication that these probably were dominant canopy trees. Modern riverside weed trees such as sycamore Platanus occidentalis) and sheoak (Casuarina cunninghamiana) produce seeds in comparable — numbers and sizes to those of Dictyopteridium sporiferum. Associated fossil seeds of other glos- sopterid plants (Walkom, 1922) are generally larg- er and more elaborate, thus supporting the view that D. sporiferum was weedy in comparison. Judg- ing from their sedimentary context, the earliest successional plants were horsetails (Phyllotheca spp.) and lycopods (Selaginella harrisii). Other species of glossopterids were probably the dominant plants of stable swamp woodlands. Associated os- mundalean tree ferns (Gould, 1970) may have formed understory shrubs and small trees. In con- trast to those of Glossopteris, the leaves of these tree ferns were delicate, like those of shade plants. Higher land to the west, and perhaps also the volcanic highlands to the east, were forested by conifers (Walkomiella australis: Retallack, 1 980b). Paleogeographic setting. Dictyopteridium sporiferum is exquisitely preserved in Late Perm- ian silicified peat near Homevale Station and is found as compression fossils in shales elsewhere in the northeastern Bowen Basin (Jensen, 1975). This 1038 Annals of the Missouri Botanical Garden Lt FL XS. ill. de : QE m W REX ^ le SEA £ P : 4 E AW ND aN ' j fe FIGURE 9. A reconstruction of “napi thomasi of Late Triassic (Carnian) age, from the upper Umkomaas Valley, Natal, South Africa. . Habit as a shrub of seasonally wet bottomlands.— B. Abscission scar and leaf traces of petiole base. —C. ps shoot. —D, E. Upper and lower (respectively) leaf cuticles, with Volume 75, Number 3 1988 Retallack & Dilcher 1039 Seed Ferns basin is continuous to the south with the Sydney Basin of New South Wales. Coal measures in the Sydney-Bowen Basin accumulated in a broad in- termontane depression between the plains and hill ranges of inland Australia to the west and a vol- canic, Andean-style mountain range to the east (Jensen, 1975). The eastern Australian part of the Gondwana supercontinent was at very high paleolatitudes dur- ing the Late Permian (Herbert, 1980). There is no clear evidence of glaciation in the Bowen Basin at this time, but there were probably alpine glaciers on high mountains to the east. There are several lines of evidence for a seasonal, cool temperate climate throughout these lowlands. Growth rings are seen not only on fossil wood (Rigby, 1971) but also in the shells of marine shellfish, which were much less diverse in this region of the world during Late Permian time than in other formerly tropical regions (Runnegar & McClung, 1975). lacustrine shales have been found with rafts of leaves at the very top of the silty layer, overlain by the thin shaley layer of the varve. This may be explained as coarse material washed out by brisk spring runoff and summer storms, followed by au- tumn leaf fall and slow winter accumulation of clay (Retallack, 1980b). Winters were not harsh by this time, because osmundalean tree ferns were wide- spread. They are represented by silicified trunks and foliage in the Bowen Basin (Gould, 1970). arved PELTASPERMUM THOMASII Hypothesized reconstruction. This plant is thought to have been a low-growing perennial shrub, vegetating stream, pond, and lake margins within temperate mesophytic woodlands and forests of South Africa and other southern continents during Late Triassic (Carnian) time, about 225 to 230 million years ago (Fig. 9). Its leaves (Lepidopteris stormbergensis (Seward) Townrow, 1956) were coriaceous, stiff, and fernlike. They varied from pinnatifid to bipinnatifid and in the proportion of stomates on either side of the leaf. This and the blisterlike hydathodes covering the frond rachis and the stem are features found in some modern water- side plants. Leaves had clear abscission scars, so it probably was a long-lived, perennial plant and may have been seasonally deciduous. The pollen- bearing (Antevsia extans (Frenguelli) Townrow, 1960) and ovulate organs (Peltaspermum tho- masii Harris, 1937) were pinnately arranged, and each frondlike structure formed a large paniclelike fructification. Pollen (Monosulcites minimus Cook- son, 1947) was moderately sized (23-40 um) and produced in great quantity: both features of modern wind-pollinated plants. The reconstructed plants may have had a pseudostigmatic kind of pollination, in which pollen adhered to the ovule or head an only pollen tubes entered the micropyle, as in mod- ern coniteus va relatively inaccessible ovules and eeds of the fossil plant probably were dispersed Sauk by water. HOMSaACCALe p Evidence for reconstruction. This reconstruc- tion is based largely on fossil compressions from the black shales of “Burnera Waterfall," where a tributary creek of the Umkomaas River drops over a scarp of the basal Molteno Formation, 4 km southeast of Vergelegen Nature Reserve, Natal, South Africa (our Fig. 2C; locality Umk 111 of Anderson & Anderson, 1983). These shales form the base of the Molteno Formation and are Late Triassic in age (Yabeiella oppel-zone of Retallack, 1977; or Carnian in the marine time scale), which is about 225 to 230 million years old (in time scale of Palmer, 1983). The various fossil organs at- tributed to this species have also been found in association at other localities in South Africa, South America, and New South Wales (Retallack, 1977; Anderson & Anderson, 1983). The remains dis- cussed were attributed to one plant by Townrow (1960) for several reasons: the close similarity of the cuticles of leaves, microsporophylls, and mega- sporophylls; pollen found dispersed and in micros- porophylls; and close association at several local- ities. Habit. This plant is mainly known from co- riaceous, bipinnatifid to pinnatifid leaves. Unipin- nate leaves also have been included in the genus (Townrow, 1960). These have identical cuticle and may be closely allied plants but are now placed in the genus Pachydermophyllum (Retallack, 1981). Also variable is the distribution of stomates on both sides of the leaf in Lepidopteris stormbergensis. They are usually more numerous on the abaxial side but occasionally are more common on the adaxial side, which can be distinguished because _ stomates overhung by n —F. Cross section of stoma onstructed e mu section of ovule, H, I. Ovulate fructification e.—G. Blisterlike structure on cuticle of petiole. — showing vascularization (heavy Rec lines) and resinous glands (black ellipies]- —K, L. Pollen-bearing organ.— 9M. Monocolpate pollen 1040 Annals of the Missouri Botanical Garden of its thicker cuticle and less-bulging venation. Such a stomatal distribution is found in leaves of under- story shrubs (Salisbury, 1927) and of waterside herbs (Townrow, 1960). The peculiar blisterlike swellings under hair bases on the stem and rachis of this plant are similar to hydathodes in waterside herbs such as water dock (Rumex hydrolapathum: Townrow, 1960). A waterside understory habitat seems likely, but we reconstruct this plant as a small, much-branched, perennial woody shrub for the following reasons. The leaves have thick cu- ticles and were probably stiff and coriaceous. Leaf bases with clear abscission scars and two leaf traces have been found (Townrow, 1960). These were abscised, perhaps seasonally. Slender stems of this plant have also been found which have stomates and blisters like those of the leaves (Townrow, 1960). Such stems are unlikely to have been the main stem of a plant with a single palmlike stem. It is more likely that these were short shoots of a branched plant. Pollen sacs and ovules were arranged radially on more or less laminar organs, which were ar- ranged pinnately. These pinnate structures formed large, lax, paniclelike structures (Anderson & An- derson, 1983, pl. 23, and possibly also their “fe- male cone gen. C. sp. A." of pl. 26). Reproduction. Wind pollination of this plant may be indicated by its copious production of smooth, moderately sized (23-40 um) pollen. How- ever, its large (5 mm long by 2 sacs, with glandular bumps and borne in groups of mm wide) pollen four, are distinct from the more numerous, non- glandular, smaller pollen sacs of other Mesozoic seed ferns usually regarded as wind pollinated, such as Umkomasia granulata (also reconstructed here). vulate fructifications have been found in sev- eral different stages of maturation. The least-de- veloped examples have only bumps rather than obvious ovules underneath the ovular hea e- tallack et al., close to the time of pollination, when the ovules 1977). These may be remains at or and their elongate micropyles were not as obvious as in other remains with ovules attached (Townrow, 1960). Pollination by means of a pollination drop is unlikely, considering the nature of the pollen grains. Unlike the bisaccate pollen of related peltasperm seed ferns such as Pteroma (Harris, 1964) and Townrovia (Retallack, 1981), Peltaspermum tho- masii had nonsaccate pollen grains. This is also true of some modern conifers such as western hem- lock (Tsuga heterophylla) and Prince Albert's yew (Saxegothaea conspicua), compared with closely allied conifers with bisaccate pollen (Doyle, 1945). These modern conifers with nonsaccate pollen have pseudostigmatic pollination, in which pollen grains adhere to parts of the plant near ovules, and long pollen tubes enter the ovules. The time between pollination and fertilization in these conifers is no less brief, compared with other conifers such as pines (Doyle & O'Leary, 1935b), because the tip of the pollen tube overwinters in the nucellus of the ovule (Doyle & O'Leary, 1935a, c). The largest ovules seen are attached singly or in pairs to fructifications with abscission scars in- dicating a former complement of about six ovules (Townrow, 1960). Th ovules, and the most common remains are of iso- e most complete remains lack lated ovular heads only rarely with attached ovules (Anderson & Anderson, 1983). Even at these stages, associated ovules lack differentiation of woody and fleshy integuments to the degree seen in associated seeds, such as those of Umkomasia granulata. Neither the ovular heads, which may have been dispersal units, nor the ovules were adapted to specialized modes of dispersal, and they probably were scattered by wind and water. Habitat. The black shale in the Umkomaas Valley (Umk 111 of Anderson & Anderson, 1983) is the deposit of a poorly oxygenated lake. The lacustrine shale overlies a thin conglomerate, which disconformably overlies floodplain deposits of the atberg Formation, of Early Triassic age (Lystro- saurus zone). Such conglomerates at the base of a shale are found often in abandoned channels (oxbow lakes) of meandering streams (McDonnell, — At the Umkomaas Valley locality the shales are overlain by a variety of paleosols in which the plants may have lived (Fig. 2C). The so-called “oil shale” overlying the black shales is a coal (du Toit, 1954) and was probably a peaty soil (Histosol of Soil Survey Staff, 1975) fringing the Oxbow Lake. Overlying this are some very weakly developed fossil soils (Entisols of Soil Survey Staff, 1975), consisting of little-modified alluvium riddled with red, clayey, noncalcareous paleosols with slightly sandier surface horizons (probably Alfisols or Ul- tisols). Their root traces, profile differentiation, and degree of oxidation are typical for well-drained forested soils. these various possible habitats, we think that Peltaspermum thomasii preferred the peaty lake- side soils. Unlike other associated fossil plants, this species lived very close to the lake, because it is Volume 75, Number 3 1988 Retallack & Dilcher Seed Ferns 1041 equally abundant through the entire thickness of the lacustrine shales (Townrow, 1960). are reasons for regarding it as a waterside plant (as already discussed). Peltaspermum thomasii has been found in place in the organic layer and im- mediately overlying carbonaceous shales of water- logged, peaty paleosols at other localities, such as Konings Kroon, South Africa (Kon 111 of Ander- son & Anderson, 1983; G.J.R., pers. obs.) and in Nymboida Open Cut Mine, New South Wales, Aus- tralia (Retallack, 1977). In all of these localities Peltaspermum thomasii is associated with Umko- masia. granulata. Although a waterside and lowland plant, there are indications that it could tolerate deficiencies of water or nutrients. The stomates, for example, are ere also overarched by papillae on the subsidiary cells. The blisters on the stems and leaves are comparable to water control structures (Townrow, 1960), such as salt glands or hydathodes (discussed by Esau, 1977). Judging from its geological occurrence, it is more likely that these peculiar features of the plant re- flect growth in nutrient-deficient habitats rather than lack of water. It may have been suited to siliceous, nutrient-poor, streamside sands of point bars, or acidic swampy lowlands The other local habitats revealed by paleosols at the Umkomaas Valley locality seem less likely for this plant, because there is evidence from other localities that they were vegetated by other plants. Entisols of streamsides may have been colonized by broad-leaved conifers (Heidiphyllum) in drier, sandy, and elevated areas, and by ferns (Clado- phlebis sterile, or Asterotheca when fertile) and horsetails (Neocalamites) in wetter places. Alfisols and Ultisols of well-drained floodplain forests and woodlands in contrast were dominated by other seed ferns (Dicroidium elongatum and D. cori- aceum) and a variety of plants with ginkgolike leaves (Retallack, 1977). Within this mosaic of vegetation, Peltaspermum thomasii is envisaged as an understory plant of woodlands dominated by Umkomasia granulata in periodically waterlogged lowlands. Paleogeographic setting. During Late Trias- sic time the Molteno Formation formed an exten- sive lowland piedmont north of the mountainous Cape Fold Belt (Turner, 1978). The lower part of the formation in the Umkomaas Valley lies dis- conformably on older alluvial rocks, which may have formed low hills flanking low ranges of much older (Precambrian), resistant rocks to the north, west, and east. This far north of the mountains, streams laid down sequences of sediment of a type formed in loosely sinuous modern streams trans- porting a mixed load of clay and sand (Turner, 1978; compare general models of Schumm, 1981). Climate in this region was humid and cool tem- perate, perhaps seasonally snowy. This can be in- ferred from the high paleolatitude of this part of Gondwana (Anderson & Schwyzer, 1977). Late Triassic fossil wood from South Africa (Walton, 1923) has well-marked growth rings, an indication of strong seasonality. A humid climate is indicated by the quartz-rich composition of Molteno sand- stones (Dingle et al., 1983) from which most of the easily weathered minerals have been lost. High rainfall probably also accounts for noncalcareous fossil soils and the dearth of vertebrate fossils in the Molteno Formation (G.J.R., pers. obs., follow- ing general models of Retallack, 1984). UMKOMASIA GRANULATA Hypothesized reconstruction. This was prob- ably a woodland tree which dominated seasonally waterlogged floodplains of extensive lowlands north of mountains in the Cape region of South Africa, and of other humid regions of Gondwana during Late Triassic time, some 225-230 million years ago (Fig. 10). The leaves of this plant (Dicroidium odontopteroides (Morris) Gothan, 1912) had cal- lused abscission scars, so probably were deciduous. Its wood (Rhexoxylon tetrapteridoides Walton, 1923) was coniferlike, but with exceptionally wide rays. Ovulate (Umkomasia granulata Thomas, 1933) and pollen-bearing structures (Pteruchus johnstonii (Feistmantel) Townrow, 1962b) of this plant were pinnately organized, and the pinnate structures were arranged helically in large panicu- late fructifications. The copious production of mod- erately sized (corpus averaging 46 by 32 um), bisaccate pollen (Alisporites australis de Jersey, 1962) is compatible with pollination by wind and possibly with a pollination drop. At ovulation, the ovules were weakly developed, and their bifid, elon- gate micropyles extended out below the margin of the cupulate head that enclosed their inverted bas- es. During subsequent development the ovule elon- gated well below the cupule margin and gained a sclerotesta with three broad ribs as the formerly elongate micropyle withered away. Seeds were re- leased as their stalk abscised and the bivalved cu- pule spread open. The moderately-sized (3.7-7 m X 2.2-5 mm) seeds were produced in great numbers and protected by a sclerotesta from des- iccation or damage during possible ingestion. They probably were dispersed in a variety of ways, main- ly by wind and water. 1042 Annals of the Missouri Botanical Garden Í y S TT ac ao la fae: S: < bai Tm r ifi FiGURE10. A reconstruction of Umkomasia granulata of Late Triassic (Carnian) age from the upper Umkomaas Valley, Natal, South Africa. — A. Habit as a tree of seasonally wet bottomlands.—B. Diagrammatic cross section Volume 75, Number 3 1988 Retallack & Dilcher 1043 Seed Ferns Evidence for reconstruction. This plant has been put together on the basis of compressed re- mains of leaves, ovulate and pollen-bearing organs with similar cuticle, and similar pollen grains in pollen sacs and ovules (Thomas, 1933; Townrow, 1962a-c). These remains were first demonstrated to be closely associated in the same locality (in the Umkomaas Valley, South Africa) that also serves as a basis for our reconstruction of Peltaspermum thomasii. The association of these fructifications and leaves is now known at most localities where it is found in South Africa, South America, Aus- tralia, and New Zealand (Retallack, 1977, 1980c). Evidence that this plant was a tree of deciduous woodlands and forests includes associated fossil trunks, root traces, and the callused abscission scar at the base of its leaves. Other lines of evidence have been used in support of the idea that its wood, at least in South Africa, was Rhexoxylon tetrapter- idoides Walton (1923). There is a close association of Dicroidium with petrified wood of Rhexoxylon pianitskyi at two localities in Argentina (Archan- gelsky, 1968). Of the two species of Rhexoxylon found in South Africa, R. tetrapteridoides is the only one found in the Molteno Formation near, but not at, the same localities (such as the Umkomaas Valley; Walton, 1923; Anderson & Anderson, 1983) from which our reconstructions of Umko- masia granulata and Peltaspermum thomasii came. A petrified leafy shoot of Rhexoxylon pian- itskyi from Argentina (Archangelsky & Brett, 1961) formed the basis for our reconstruction of R. tetrapteridoides here, which we have assumed also to have six vascular traces leading into the petioles of the leaves. Because of this, Rhexoxylon could not be the wood of associated conifer (such as Rissikia or Heidiphyllum) or ginkgolike foliage (such as Sphenobaiera), which had only one or two vascular strands in their petioles. Nor does Rhexoxylon show the girdling leaf traces or inti- mately: admixed parenchyma and xylem seen in the wood of cycads and cycadeoids, which presum- ably produced associated foliage of Pseudoctenis and Taeniopteris. Among associated seed ferns in South Africa, Peltaspermum thomasii (as recon- structed here) had only two leaf traces in the petiole (Townrow, 1960) and is here reconstructed as an understory shrub. Glossopteris verticillata is a rare element with multistranded petioles of this Triassic fossil flora, but it has been found attached to stems in a much closer helix than apparent for petrified Rhexoxylon pianitskyi or natural groups of impressions of Dicroidium odontopteroides leaves (du Toit, 1927; Thomas, 1952; Anderson & Anderson, 1983) Thomas’s (1933) numerous species of pollen organs have been revised and reduced in number by Townrow (1962a, b). This task has not yet been completed for ovulate organs, although most of these remains have been transferred to the genus Umkomasia (Holmes, 1987). From our examina- tion of Thomas’s specimens and preparations (in the British Museum of Natural History), we think that there are probably only three species. The type species of Thomas’s three genera can be taken as representative of these species, which can be recognized by deeply incised cupular lobes (Umko- masia macleanii), weakly lobed cupule margins (Umkomasia sewardii), and smooth, bivalved cu- pule margins (Umkomasia granulata). These three taxa also show cuticular differences comparable to those found in the three associated species of Pteru- chus by Townrow (1962a). Thus, the species name Umkomasia granulata is used broadly here to include several other named species (“ Pilopho- "P." geminatum, “P.” paucipartitum, burnerense, "P." nata lense, * P." sp. type A, “P.” sp. type B, and ep u sp. type C), which we regard as different stages of development and different parts of a large fructi- rosperma” fication. In the case of one of these species (“P geminatum), Thomas (1933, 1934, fig. 12F) thought that it had paired cupules, back to back. From comparison of the branching pattern of this specimen with the others, as well as our exami- nation of its cuticular preparations, we interpret this as a dehisced, bivalved cupule. The branching pattern and small bractlike organs were regarded by Thomas (1933) as evidence that these fructi- fications were stalked, like the flowers of angio- sperms. We prefer Townrow's (1962a) interpre- tation, developed for the pollen organ, that eac — of wood showing funde parenchymatous ios (black) and secondary wood (concentric dotted lines) .—C. X ioles. — D. oot and peti ss section of stomate.— showing vascular traces (white circles) and pollen sacs ; (stipple ). Leafy shoot.—E, F. Upper and lower em cuticles ofl ‘eat J. Cross section of head of pollen organ, M. b ovule at time of pollination. —N. Cupulate ovule at time of fertilization. —O. Dehisced c tudinal section of cupulate ovule, showing hairy inner side of cupules (irregular extensions) and 3 wc (stipple) . 1044 Annals of the Missouri Botanical Garden cupule is a homolog of a pinna, each "bracteole" a sterile pinna, and the whole pinnate structure homologous with a frond. This structure is in some ways similar to the ovule-bearing cupule arising from the dichotomy of a frond in early Carbonif- erous seed ferns such as Sphenopteris bifida (Long, 1979b) Habit. Judging from the trunks of Rhexox- ylon tetrapteridoides found (up to about 16 cm in diameter), this plant was a woodland tree (Ar- changelsky & Brett, 1961; Petriella, 1985). A trunk showing branch scars (Walton, 1923) pro- vides evidence that the large branches were ar- ranged in tiers that were not quite whorled. Like Dictyopteridium sporiferum reconstructed here, Umkomasia granulata probably had tiered branching which became more irregular with old age. This, and a long shoot and short shoot or- ganization apparent from small petrified leafy stems and leaf groups, is very similar to our envisaged habit of Dictyopteridium sporiferum. The leaves of this plant were variable in mor- phology, as documented in detail by Anderson & Anderson (1983). They may have belonged to a complex of hybridizing species. The Andersons have identified aberrant fossil leaves with narrow seg- ments (Dicroidium coriaceum and D. elongatum) in some parts of the frond, and wide ones (D. odontopteroides) elsewhere on the same frond. ese may represent hybrid leaves. Disregarding these as sports and considering variation in natural populations of these leaves (Anderson & Anderson, 1983, figs. 3, 4), Umkomasia granulata probably included leaves referred to Dicroidium odoniopte- roides, as well as those which have been referre to “D. lancifolium" and “D. obtusifolium." Stomates were distributed on both sides of the leaf but were much more abundant on the abaxial side, which had a thinner cuticle covered in papillae (Anderson & Anderson, 1983; Petriella, 1985). Fossil leaves have been found in radial groups and with asymmetric basal petioles (Anderson & An- derson, 1983). The ovulate and pollen-bearing structures were pinnately organized and helically arranged on a larger stem (shown well in “Umkomasia sp. A” and “Pteruchus sp. B" illustrated by Anderson & Anderson, 1983, 1985). Only the ovule-bearing structure has been found in an apical position on slender shoots (Thomas, 1933; Holmes, 1987). Reproduction. — Individual laminar heads of the pollen organ bore hundreds of pendent, elongate pollen sacs (Townrow, 19622). Pollen grains were bisaccate, nonstriate, and of moderate size (corpus averaging 46 by 32 um in size: Townrow, 19622). It is likely that the saccae were areas of porous exine (as indicated by Taylor et al., 1984, for a closely allied fossil species, really from the Ipswich oal Measures near Dinmore, Queensland, ac- cording to these authors' erratum) rather than com- pletely detached from the corpus. These are all indications that Umkomasia granulata probably was wind pollinated. Only in specimens with small ovules does the elongate micropyle protruding from under the cu- pule appear to have been turgid and open. This was probably the stage of pollination, because in specimens with larger, more elongate ovules, the micropyle is withered and torn. The pollen chamber was narrow and tubular. It has been compared (by Thomas, 1933) to the salpinx of Paleozoic seed ferns such as Lagenostoma lomaxii reconstructed here. However, there is no sign of a central column, and it was not exposed like the salpinx. In these respects it was more like the pollen chamber of Callospermarion pusillum. Both were passive re- ceptacles, not sealed from within by a plug of tissue, but sealed by the overarching sclerotesta of the integument. As the ovule enlarged and elongated, three broad ribs developed in the sclerotesta. This period of growth may have been when the mega- gametophyte and archegonia developed during the time between pollination and fertilization, as al- ready described for Callospermarion pusillum. ature ovules and seeds were shed from the cupules by abscission from their stalk and by the cupule splitting into two lobes. Cupules at this open stage, revealing fully the degree of lobation of their margins, are the easiest of these remains to identify (Retallack, 1977). The branching fructifications found as fossils were presumably aborted or torn from the trees during storms. Much more common are isolated seeds, which were presumably the prop- agules. The seeds were neither exceptionally large nor small (3.5 to 7 mm long by 2.2 to 5 mm wide: Thomas, 1933), nor especially woody or fleshy, winged or spiny, even compared with the variety of associated fossilized propagules (Anderson & Anderson, 1983). They were probably dispersed by generalized methods, mainly by wind and water. Habitat. Late Triassic locality in the Umkomaas Valley of South Africa (Fig. 2C), already described for Pel- taspermum thomasii. Remains of these species drifted into a lake and were mixed with a variety This plant is best known from the of other plants. Fossil leaf litters preserved in pa- leosols at Konings Kroon, South Africa (locality Kon 111 of Anderson & Anderson, 1983) and at Volume 75, Number 3 1988 Retallack & Dilcher 1045 Seed Ferns Nymboida Open Cut, New South Wales, Australia (Retallack, 1977) are better evidence of its habitat and associated plants. As already discussed for Peltaspermum thomasii, the soils were periodically (probably seasonally) waterlogged in lowland flood- plains. High soil acidity and deeply weathered par- ent material of this humid region depleted nutrients in these fossil soils. Stumps and root traces pene- trate deeply into these carbonaceous paleosols Hatch & Corstorphine, 1909). This is unlike roots of modern plants in permanently waterlogged soils (Jenik, 1978). Although waterlogged for some part of the year, thus retarding decay of carbonaceous material in the soil (Retallack, 1984), these soils were moderately well drained at other times. Also relevant to the former habitat of this plant are the distinctive wide parenchymatous rays of the wood here included in our reconstruction. This wood structure has been compared to that of mod- ern vines, in which wide parenchymatous rays be- stow flexibility (Walton, 1923). These plants are unlikely to have been vines, considering their size, i some flexibility may have been advantageous during seasonal fluctuations in water availability (Carlquist, 1975). Judging from its abundance and likely stature, Umkomasia granulata was probably a dominant canopy tree of lowland mesophytic woodlands in humid, cool temperate regions of Gondwana (Re- tallack, 1977). Other trees included rare conifers (Rissikia media) and plants with ginkgolike leaves (Sphenobaiera stormbergensis). Among the understory plants were other seed ferns (Pelta- spermum thomasii), cycadophytes (Pseudoctenis and Taeniopteris), ferns (Cladophlebis), and horsetails (Neocalamites). These last-mentioned pteridophytes also may have colonized disturbed parts of the forest, because they are found in other nearby localities (such as Konings Kroon localities Kon 211 and 221 of Anderson & Anderson, 1983) where they form distinctive early successional as- semblages in very weakly developed paleosols (En- tisols of Soil Survey Staff, 1975: G.J.R., pers. obs.). Paleogeographic setting. || Umkomasia gran- ulata is best known from the same locality and region as Peltaspermum thomasit. CAYTONIA NATHORSTII Hypothesized reconstruction. We recon- struct this plant as a forest tree of an extensive Middle Jurassic (Bajocian or 175 to 183 million years old) coastal plain, in what is now northeastern England (Fig. 11). Climate in England at this time was subtropical, monsoonal, and seasonally dry, with an annual rainfall within the humid range. The compound leaves (Sagenopteris phillipsii (Brongniart) Presl in Sternberg, 1838) of this plant had four leaflets paripinnately arranged but so close together that they appear palmately compound. Leaves developed by elaboration of the apical re- gion of an initial scale into the four leaflets and by modification of the body of the scale into a petiole. The adaxial cuticle of the leaflets was thick and lacked stomates. The abaxial side had a thin cuticle, stomates, and papillae. Both ovulate (Ca ytonia na- thorstii (Thomas) Harris, 1940) and pollen-bearing structures (Caytonanthus arberi (Thomas) Harris, 1941) were organized imparipinnately and ar- ranged in a helix on slender axes. The pollen (Vi- treisporites pallidus (Reissinger) Nilsson, 1958) were moderately sized (22-28 um wide), bisaccate, and produced in large quantities. These are typical features of wind-pollinated plants. Pollen grains in coprolites provide evidence that these were also eaten by small animals that may have been effective as pollinators. The ovules (Amphorispermum pul- lum Harris, 1943) were borne on the inside of fleshy, berrylike cupules. At the time of pollination, ovules were small and poorly developed, and the ovulate structure consisted of cuplike inrolled flaps. Pollen were probably entrapped by a pollination drop at the opening of the ovulate structures and drawn back into the micropyles of the ovules along a series of guiding filaments as the droplet dried out. In mature cupules the ovules were completely enclosed by the overarching pressure of the turgid fleshy "fruit" and by sealing the entrance with bands of cutin and perhaps also of other substances. The fleshy, berrylike cupules have been found burst and split. They may have spilled ovules as they fell on the ground. More effective in their dispersal may have been reptiles and small mammals at- tracted by these fleshy fructifications. The seeds were fortified against crushing and acidity during ingestion by their sclerotesta. Evidence for reconstruction. Caytonia na- thorstii is best known from a thin (1 m) shale bed Fig. 2D) in the rock platform south of Cayton Bay, Yorkshire, England (Harris, 1964). This is the Gristhorpe Bed of the Cloughton Formation, Ravenscar Group (Kent, 1980), and is Middle Ju- rassic (Bajocian) in age, or some 175 to 183 million years old (in the time scale of Palmer, 1983). This plant has been reconstructed from compression remains of leaves, buds, pollen organs, and ovule- bearing organs with similar cuticles; from the sim- ilarity of pollen in pollen sacs and ovules; and from close association at several localities (Harris, 1964). ~ 1046 Annals of the Missouri Botanical Garden === QUE 2 N iW VB cm TAJ S y. lod / NIH RASO CALS UTA Volume 75, Number 3 1988 Retallack & Dilcher 1047 Seed Ferns At the best-known locality, Cayton Bay, the fruc- tifications of most of the other associated fossil plants are known to be different (Harris, 1961, 1964, 1969, 1979; Harris et al., 1974). Habit. poorly preserved fossil stumps (Thomas, 1925; Black, 1929; Harris, 1971) in a forested paleosol (Spodosol or Inceptisol, Fig. 2D) that this plant was part of a mixed fern-cycadophyte- conifer for- est. Its exact habit within this vegetation is less certain. Harris (1971) has argued that Caytonia nathorstii was a tree, as shown by finds such as leaves attached to stout woody shoots and of bud- like short shoots of young leaves. As he argued, few shrubby plants have such stout leafy limbs, and those that do are often more succulent or more completely invested with leaf bases. Buds are typ- ical of a perennial deciduous plant. The four leaflets of the compound leaves appear palmately compound but are paripinnately ar- ranged with very close points of contact (Harris, 1951). The diffuse, anastomosing venation of leaf- lets of Caytonia nathorstii is superficially like that in leaves of Dictyopteridium sporiferum, also re- constructed here. Sagenopteris differs from Glos- sopteris in having compound leaves, a true midrib There is evidence from association with to the leaflets, and secondary venation in which radiating and dichotomizing veins are undeviated by confluent strut veins. These differences from Glossopteris have been indicated also by Delevor- yas & Person (1975) for Mexiglossa leaves, which we regard as closely related to, if not congeneric with, Sagenopteris. Fossil buds (Harris, 1964) indicate that short shoots of leaves developed synchronogeyy presum- ably after a period of leaf fall, then d y during a harsh season. Leaves developed from the scalelike organs by the differentiation of two orbicular apical leaflets, followed by a second pair on either side behind them (Harris, 1940). Most of the scales became differentiated into the petiole of the com- pound leaf. The developing leaves were pilose, as are many young organs (Esau, 1977), but mature leaves were glabrous. The ovules and pollen sacs were arranged pari- pinnately on leaflike structures on either side of an axis, which had a different adaxial and abaxial cuticular structure, more like a petiole than a stem. The ovulate structure was unipinnate (Harris, 1964), and the pollen-bearing structure was bi- pinnately arranged (Harris, 1951). These fructi- fications probably formed paniculate structures, like those of other Mesozoic seed ferns reconstructed here. Pollen sacs of this plant are thought to have been attached to the abaxial side of the leaflike microsporophyll (Thomas, 1925; Harris, 1964), ut the orientation of the ovules is more contro- versial. For the following reasons we believe that they were also morphologically abaxial, as were the openings to the cupules. The most important evidence is a single specimen of Caytonia na- thorstii attached to another axis (Thomas, 1925, pl. 12, fig. 16). This specimen was thought lost (Harris, 1971) but was discovered during 1980 by us in the paleobotany teaching collection of Cam- bridge University. We confirm that the megaspo- rophyll at right angles to the axis is attached, as are two berrylike cupules displayed with their oc- cluded opening crushed on top of their stalks. From our interpretation of the creasing at the attachment points, clearly visible in Thomas's excellent pho- tograph, the attachment does not appear twisted into an unusual orientation. The occluded opening of the fruit appears to be abaxial, and the ovules enclosed by the fruit wall are also morphologically abaxial. This is because, as Harris (1964) and Reymanowna (1974) argued, the cupule is ho- mologous with an infolded leaf. By our inter- pretation, papillae and hair bases consistently characterize the abaxial surface of leaves, micro- sporophylls, and megasporophylls (Thomas, 1925; Harris, 1940, 1964). In addition, the outer cuticles of the cupules lack stomates and have a structure in thinner areas (probably lenticels) very similar to t as a tree of bot E ll. Vorkshir re, England. SS ppesss, cuticles of leaf, with stomates on gu side. —H. Pollen or, —I. Cutaway view of quadrilocular synangiu —J. FE. at time of pollination. —L. Mature ovulate organ.— M. - Longitudinal sections of ovule, s "aleurone layer" views of mature fruits and ovules wall cells) , "spotted layer" (haaay dimpled line) , micropyle. A reconstruction of Caytonia T paqar of Middle dri (Bajocian) age from Cayton Bay, —A. Habi mland forest.—B. Leafy shoot.—C, D. Uppe r and lower Sueco stages in the growth " young isaccate po^ n.— uae H time of disp showing Erde yim es id. double (black rectangles), and hair bases around == E »» "Fruits 1048 Annals of the Missouri Botanical Garden the adaxial cuticles of the leaves (Harris, 1940). Interpretation of the ovules as abaxial brings these plants into line with closely allied plants and with seed ferns as a whole. We have taken pains to justify our reconstruc- tion on this point, because it runs contrary to the opinion of the leading authority on Caytonia na- thorstii (Harris, 1940, 1964). Since the thickest cuticle on the rachis of this fructification is on the narrow side, Harris argued that the ovules of Cay- tonia were morphologically adaxial. He interpreted this as similar to the thicker adaxial cuticle of leaves and pollen organ of this plant. We, on the other hand, have pointed out other ways of viewing cu- ticular homologs between different organs of this plant and regard the thick narrow cuticle of the abaxial side of the rachis as a structural thickening that prevented drooping of the heavy cupules. Reproduction. It is difficult to determine whether fertile shoots of this plant were bisexual or unisexual. Thomas (1925) argued that individual plants were unisexual, because the best-preserved fructifications of either sex were found in different parts of the same bed. We confirmed this during our own collecting. This is compatible with the view that this plant, like modern plants that have uni- sexual fructifications, was wind pollinated. The fos- sil plant produced copious amounts of weakly sculp- tured, moderately sized (22-28 um wide: van Konijnenburg-van Cittert, 1971), bisaccate pollen. Shriveling of the saccae on release, as observed in modern pine pollen by Wodehouse (1935: 254), would have greatly reduced the size of the pollen grains. The central body of the grains is only 12- 18 um long by 6-12 um wide (van Konijnenburg- van Cittert, 1971). They could not have shrunk this much because the sacci contained an internal mesh of sporopollenin, in some ways like that found in pollen from the corystosperm microsporophyll Pteruchus (Pedersen & Friis, 1986). Some possible evidence for animal pollination of this fossil comes from small (4 mm diameter) coprolites full of these pollen grains. These were produced by small ani- mals that lacked the ability to digest sporopollenin (Harris, 1964), perhaps a small arboreal mammal or lizard. Australian honey possums (Tarsipes spencerae), and a variety of other small animals are known to be effective pollinators, even though they are unable to fly, as is usual among the most important modern pollinators (Rourke & Wiens, 19777). Apart from dinosaur footprints (Kent, 1980), no fossil vertebrates or insects have been found in association with these plants, but from what is known of Middle Jurassic life on land elsewhere in Britain (Clemens et al., 1979; Savage, 1984), there were varied small mammals and arboreal gliding and flying reptiles. Birds had not yet appeared, and despite the diversity of insects at this time, the fossil coprolites are too large for them. As is common in modern conifers with bisaccate pollen (Doyle, 1945), pollination may have been aided by pollination droplets. In this case it probably emerged from the mouth of the cupules at an early stage of development, when they were little more than infolded flaps (Harris, 1943, 1964). As in Dictyopteridium sporiferum reconstructed here, there may have been a reticulum of filaments, which guided pollen toward the micropyles of ovules inside. Filaments were observed in preparations of ovules by Thomas (1925) and Krassilov (1977) but were not noticed by Harris (1933). Pollination must have occurred when ovules were young and the cupule was open, because later in development the opening was sealed by inflation of the fleshy fruit wall and the development of cutinous bands, and perhaps also waxy or other substances (Harris, 1933, 1943). This subsequent development may have disrupted internal filaments so that they were difficult to see in mature fruits. This closure also prevented penetration of clay in the fallen mature fruits. Early ideas (Thomas, 1925) about stigmatic, angiospermlike pollination of these plants were long ago disproved by the discovery of pollen grains within the ovules inside the cupules (Harris, 1933, 1940). At maturity the berrylike cupules were swollen and completely enclosed the ovules. Each ovule A ] + + A 4 is] well g y within the ovular coat, the “spotted layer," is quite unlike a true cuticle, but nevertheless somewhat maceration-resistant (Harris, 1958a). It could be the basal layer of a zone of osteosclereids or lage- nosclereids, as found in the seed coat of modern soybean (Glycine max: Esau, 1977). Inside the nucellus of the ovule was a layer of poorly pre- served brown cell contents. Harris (19582) called it an “aleurone layer," which is a distinctive outer enzyme-producing layer of the endosperm well known in modern angiospermous cereals, such as barley (Hordeum vulgare: Esau, 1977). Dispersal of seeds may have been by bursting on impact or by rotting of the cupule. Many com- parably fleshy modern fruits with stony seeds are animal dispersed. The same kinds of small reptiles and mammals already discussed as possible polli- nators of this plant may also have aided in its dispersal. Habitat. "The fossiliferous shale (Gristhorpe Bed) at Cayton Bay (Fig. 2D) is a sequence of flood deposits overlying a thick (1.3 m), moderately de- Volume 75, Number 3 1988 Retallack & Dilcher 1049 Seed Ferns veloped sandy paleosol with a white surface (E) and orange-red subsurface (Bs) horizon (Thomas, 1925; G.J.R., pers. obs.). It was probably a base- poor, forested soil (Cley Podzol of Stace et al., 1968, or Spodosol or spodosolic Inceptisol of Soil Survey Staff, 1975) of periodically waterlogged lowlands. Stumps and large roots have been found within this paleosol (Black, 1929). These are sur- rounded by a layer of carbonaceous shale, which when macerated yields many fruits and seeds of Caytonia nathorstii and other plant debris. We interpret this as an imperfectly decayed (mor) hu- mus layer of leaf litter in a periodically (perhaps seasonally) wet soil. Plants preserved in the over- lying shales represent other leaf litter entrapped as clay and floodwaters were slowed around the trees, as in cumulic horizons of modern floodplain soils (modern examples are discussed by Birkeland, 1984). The shale layer becomes more silty and sandy higher towards the next overlying paleosol. This reflects an increase in the power and fre- quency of flood disturbance, which would have introduced early successional species of plants not usually found in undisturbed lowland forest. A great variety of plants have been found in the Gristhorpe Bed (Harris, 1961, 1964, 1969, 1979; Harris et al., 1974). The commonest tree species from near the base of the bed are the seed fern Caytonia nathorstii and the taxodiaceous conifer Elatides williamsonii. Other common remains likely to have been understory plants include the osmundalean fern Todites williamsonii, the cycad Nilssonia compta, and the cycadeoid Nissoniop- teris vittata. Caytonia nathorstii was not common in other habitats in the coastal plains of Yorkshire. Fluvial levees were colonized largely by ginkgoes (Ginkgo huttonii: van Konijnenburg-van Cittert, 1971; Nami, 1976) and other nearstream, early succes- sional environments by horsetails (Equisetum co- lumnare: Harris, 1961). Near-coastal peat swamps were vegetated by cycads (Nilssonia kendallae), cycadeoids (Ptilophyllum pectinoides), and czek- anowskialeans (Sphenobaiera gyron); and the most marine-influenced vegetation of tidal creeks was dominated by another seed fern (Pachydermo- phyllum papillosum: Thomas & Bose, 1955; Spi- cer & Hill, 1979). Compared with these other plants, Caytonia nathorstii, although a lowland plant, lived in more inland and stable sites. Paleogeographic setting. During Middle Ju- rassic time Yorkshire was a flat deltaic coastal plain between moderately elevated plateaus of older rocks to the north, south, and east. To the west was a shallow sea, which connected to the south past several low islands and a broad shelf of tropical reefs to the open Tethys Ocean (Sellwood, 1978). Yorkshire at this time was at a paleolatitude of about 35° north (Briden et al., 1974). Climatic zonation ` was less marked than at present, gr this 1975). Overall rupit was probably i in the humid range, considering the lush vegetation of pteridophytes (Harris, 1961), noncalcareous and podzolic nature of paleosols (Fig. 2D), and corrosion of the umbos of fossil unionid bivalves (Wilson, 1948). Growth zones in fossil shoots (Harris, 1971, 1979) are evidence of seasonality. The harsh season was prob- ably dry and a time of forest fires (Harris, 1958b). As in modern continents at about this latitude and orientation, climate was probably monsoonal and winter dry (Hallam, 1975). daa ibtropi cal eh DISCUSSION The reconstructions given here range from vines to stately forest trees, from plants of intertidal swamps to those of well-drained soils in locations ranging from the tropics to the high latitudes. These extinct broadleaf plants appear to have occupied many niches. In the face of such diversity it is difficult to characterize seed ferns. Certainly, they were not merely small swampland shrubs. They were a prominent component of much late Paleo- zoic and Mesozoic vegetation. The seed ferns reconstructed here, representing well-understood examples of these plants, have a few features in common. They are all characterized by fernlike leaves, ovule-bearing structures which are leaflike or adnate to leaves, and gymnosper- mous anatomy, more or less as originally envisaged by Potonié (1899) and Oliver & Scott (1904) in proposing the terms “eycadofilicales” and “pteri- dosperms." However, we think it unlikely that they form a natural group or clade, because the same features distinguish other groups of plants, such as cycads. The earliest gymnosperms were probably seed ferns (Rothwell, 1982, 1986). Cycads, cy- cadeoids, pentoxylans, ginkgoes, czekanowskians, cordaites, conifers, and other gymnosperms all probably were derived from seed ferns (Meyen, 1984). Seed ferns were thus central to the evo- lutionary radiation of seed plants, and in this sense represent a grade of evolution embracing the evo- ime roots of these various other gymno- erms. The phylogenetic relationships of gymnosperms can be expected to become clearer as lineages of seed ferns are better understood. Phylogenetic in- ference is beyond the scope of this article, but we 1050 Annals of the Missouri Botanical Garden point out two quite different groups of seed ferns to underscore the artificiality of seed ferns as a taxon. The plants with fernlike prepollen (Stam- nostoma huttonense, Lyrasperma scotica, Cala- thospermum fimbriatum, Lagenostoma lomaxii, and Pachytesta illinoensis) were different in many ways from those with true pollen (Callospermarion Pelta- spermum thomasii, Umkomasia granulata, and Caytonia nathorstii). The group with fernlike pre- pollen was similar in many ways to modern cycads, whereas the others were more like conifers. The second group includes both Paleozoic and Mesozoic plants, some of which show a number of oy len biological similarities (Reymanowna, 1 Th supposed morphological gap between Ta and Paleozoic seed ferns, apparent in many general pusillum, Dictyopteridium | sporiferum, discussions of these plants, may be seen as a study ias. It arose from the varied experience of dif- ferent investigators of these plants and from quite distinctive styles of preservation and geological set- tings for Paleozoic and Mesozoic plants, rather than from the plants themselves. A variety of seed ferns have been considered as possible ancestors of angiosperms: especially forms allied to the plants here reconstructed as Stam- nostoma huttonense (Long, 1966, 1977b), Dic- tyopteridium sporiferum (Retallack & Dilcher, 1981), 1978). Another gymnospermous group thought to be an- cestral to angiosperms, the cycadeoids (Arber & Parkin, 1907) were probably also derived from seed ferns. The origin of angiosperms is still a and Caytonia nathorstii (Doyle, puzzle, for which more detailed understanding of seed ferns is needed. In our role as arbitrators of conflicting views on the nature of the plants reconstructed here, we have become aware of the ease with which mistakes can be made. We do not imagine that our recon- structions will be the last word (or image) for these plants, and hope that they will serve as hypotheses that stimulate as well as instruct. LITERATURE CITED ANDERSON, J. M. & H. M. e iq 1983. Peleeoflata of Southern Africa. Molteno Formation (Triassic), Volume 1, Part 1. nos Part 2. Dic ae A. A. Balkema, Rotterdam 19 = Southern s, Devonian 5. Pa la eellora - Africa. Pokom of South African Flor to Lower Cretaceous. A . U. SCHWYZER. of the Permian and Triassic. Part 4 4. Palaeomagnetic evidence for large-scale intra- Condwanan plate move- ments during the O to Jurassic. Trans Geol. Soc. S. Africa 80: -234 ANDERTON, R., P. H. BRIDGES, M. P. LEEDER & B. W. SELLWOOD. 1979. A d Stratigraphy of the , London 948. as evolutionary trends in the pteridosperms. Bot. Gaz. (Crawfordsville) 110: ARBER, E. A. N. & J. Parkin. 1907. On the origin of angiosperms. J. Linn. Soc., Bot. 38: 29-80. ARCHANGELSKY, S. 1968. Studies on Triassic fossil plants from Argentina. IV. The leaf genus Dicroidium and its possible relation to Rhexoxylon stems. Palaeon- tology 11: 500-512. & ETT. 1961. Studies on Triassic fossil plants from Argentina. I. Rhexoxylon from the Ischigualasto Formation. Philos. Trans. B244: BarME, B. E G. PLAYFORD. 1967. Late Permian plant microfossils from the Prince ces Mountains, Antarctica. Rev. Micropaléont. 10: 179-192. BAMBACH, R., C. R. SCOTES SE & A. M. Pe LER. 1980. Before Pangaea: the geographies of the Paleozoic 26-38. . Calathospermum fimbria- ov., a Lower Carboniferous pteridosperm cupule from Scotland. Palaeontology 3: 265-275. & A. G. Lonc. . On the structure of a petrified stem and some associated seeds from the Lower Carboniferous rocks of East ea Scotland. Trans. Roy. Soc oe 69: ———— & leds —a new ge- nus of Lower C Sud petrified stems and pet- ioles together with a review of the classification of early Pterophytina. Trans. Roy. Soc. Edinburgh 69: 249. BanrHEL, M. 1962 nigen inkohlten karbons a Perms. Beih. Geolog BASINGER, J. Romweik & S. s STEWART. 1974. Cauline vasculature and leaf production in salas pteridosperms. Amer. J. Bot. 61: 10 Epidermus-Untersuchungen an ei- A rn des Ober- 4 1015. Baxter, R. W. € M. R. WILLHITE. 1969. The mor- phology and anatomy of Alethopteris lesquereuxii agner. Univ. Kansas Sci. Bull. 48: 767-783. Beck, C. B. 1971. Problems - Lese: delimitation in TRA Proc. N. Amer. Paleontol. Conv. 1: : . Archaeopteris and its role in vascular plant ele Pp. 193-230 in K. J. Niklas (editor), Paloob oon Paleoecology and Evolution. Praeger, New BENSON, M. 904. Telangium scottii, a new species of eg e (Calymmatotheca) showing structure. . Bot. oo 18: 161-177. jie P. W. š pe and Geomorphology. Oxford Univ. As , New Brack, M. 1929. Drifted sae ju n the e Es- tuarine Series of Yorkshire. Qua . Geol. Soc. London 85: a 43 BLanc-LouveL, C. 966. Étude anatomique compa- rée des tiges et gus pétioles d'une Pte day t du Carbonifére du genre Lyginopteris Potonié. Mém rud n Hist. Nat. Ser., Sect. C, Sciences Terre . E. ,DREWRY & A. G. SMITH. 1974. 18: BRIDEN, j. d PI ual-area world maps. J. Geol. 82: 555- 574. BRONGNIART, A. 1828. Histoire des Vegetaux Fossiles. Dufour € D'Ocagne, Paris Volume 75, Number 3 1988 Retallack & Dilcher Seed Ferns 1051 CALDER, M. G. 1938. On some undescribed species m the Lower Carboniferous flora of ado Ae iie: with a note on the genus rip ston. Trans. Roy. Soc. Edinburgh 59: 309- 331. Camp, W. H. € M. M. HUBBARD. 1963. On the origins of rh d Am . Bot. 50: 235- 24 3. dan. S. 1975. Ecological Strategies of Xylem a Puts Univ. California Press, Berkeley, Califor- 1964. Studies on North American . 9. A spilapterid from the vi- raw F. M. d insects cinity of Maz h Creek Illinois (Palaeodictyoptera). Psyche 71: n 24. CHALONER, W. G. 1976. The evolution of adaptive features in fossil exines. Pp. 1-14 in I. K. Ferguson & J. Muller (editors), hs Evolutionary Significance Exine. Academic Press, London CHAMBERLAIN, C. J. Gymnosperms. Univ. Chi- ago Press, Chicago, ioi. CHANDRA, S. & K. R. SURANGE. 1976. Cuticular studies of the reproductive organs of Glossopteris, Part II. Cistella-type fructification sie anu el- ix en. et sp. nov., Port ioides Feistimantel. Éelseckotsiil 23 161-17 Ecological aspects of the dis at C, grasses in selected habitats of Costa ica. Biotropica 10: 265-269. CLARKSON, E. N. K. 1985. Palaeoecology of the Dinan- of Foulden, Berwickshire, Scotland. Trans. Roy. Soc. Edinburgh Earth Sci. 76: 97-100. CLAYTON, G. 1985. Plant miospores from the Dinantian of Foulden, Berwickshire, Scotland. Trans. Roy. Soc Edinburgh Earth Sci. 76: 21-24. CLEMENS, W. A., J. A. LILLEGRAVEN, E. H. Linpsay & G. G. Simpson. 1979. When, where and what— a survey of known Mesozoic mammal distribution. Pp. 7-58 in J. A. Lillegraven, Z. Kielan-Jaworowska & W. A. Clemens (editors), The First Two Thirds of Mammalian History. Univ. California Press, Berke- ley, California. p ge I.C. 1947. Plant microfossils from the lignite the Kerguelen Archipelago. Rep. British, Austra- Ñan & New Zealand ae Research Expedition (1929-1931) A2: 129-14 Consi, P. 1928. Sur les P uat et la dieat systématique de Sphenopteris zamioides ertr. Ann. Soc. Géol. du N. 53: 222-229. CRAIG, G. Y. 1975. Siccar Point, v Catcra 107-177 in G. Y. Craig & P. . D. Duff dior The Geology of the Lothians d Southeast Scotland Scottish Academic Press, Edin CREBER, G. T. & W. G. CHALONER. 5984. of environmental factors on the wood structure of living and fossil trees. Bot. Rev. (Lancaster) 50: 357- 448 Influence L. R. 1959. Fossil = of the Carboniferous ocks of ë Britain. ppa eol. Surv. Great Brit- ain, Palaeontol. 4: 85- DARRAH, 1939. s i iiid Chronica Baunies: Leiden, Netherlar Davip, T. W. E. 1907. Geology of db e River Coal A N.S.W. Mem. Geol. Surv. N. S. W., Geol. 4 Dawson, A ir 0. On an undescribed fossil fern from the pon. Coal gue of Novia Scotia. Canad. Naturalist 5: 460-461. De Jersey, N. J. 19 62. “Triassic spores and pollen grains from the Ipswich Coalfield. Publs. Geological Survey Dic rh ES DELEvoRYas, T. & J. Morcan. 1954. A new pterido- sperm e the upper Pennsylvanian pe: of North America. joa iore a B96: 23. : E P. exiglossa varia gen. et sp. nov., a new genus of Eileen leaves from e usay: of vas Deed baa Abt. B, br his 11 — & T. N. TAYLOR. 1969. A probable pterido- sperm with dedi ud foliage from the Allegheny Group of northern Pennsylvania. Postilla 133: 1- Dic HER, D. L: 979. Early angiosperm reproduction: an xo report. Ann. Rev. Palaeobot. Paly- nol. 27: 291- DINGLE, R v. W. G. Sesser & A. R. NEWTON. 1983. Mesozoic and Tertiary Geology of Southern Africa. A. A. Balkema, Rotterdam DoviE, J. 1945 evelopmenta al lines in pollination mechanistiti in the Coniferales. Sci. Proc. Roy. Dub- lin Soc O LeABy. 1935a. Pollination in Saxe- gothaea. Sci. Proc. Roy. Dublin Soc. 21: 175-179. b Pollination in Pinus. Sci. Proc. Roy. Dublin Soc. 21: 181-1 & ————. 1935c Pollination i in Tsuga, Ce- eut di Ee and Larix. Sci. Proc. Roy. Dublin Y 21: = fent 3 A. js cin of angiosperms. Ann. Rev. col. EX 9: 3 Si M. J. DovocHUE. 1986. Seed plant phy- logeny and origin of angiosperms: an experimental cladistic approach. Bot. Rev. (Lancaster) 52: 321- Du Torr, A. L. 1927. The fossil wp a the upper Karroo aed Ann. S. African Mus. 22: 289-420. . H. Ha Been). re “The Geology mario sylvanian of Illinois. Phytomorphology 10: 131-138. & M. A. MiLLaY. 1976. $a «ies (plant). McGraw-Hill Yearb. Sci. Technol. 15: 53 Esau, K. Anatomy of Seed Plants. "Wiley. New bar FAEGR . VAN DER Pur. 1966. Principles of Pollination Biology. Pergamon, Oxford. FEISTMANTEL, O. 76. On some fossil plants from the amuda Series in the Raniganj Coalfield collected by Mr. J. Wood-Mason. J. & Proc. Asiat. Soc. Bengal 45: 329-382. . 1881. The fossil flora of the Gondwana Sys tem, Part 2. The flora of the Damuda and Panchet Divisions. Mem. Geol. Surv. India, Palaeontol. Indica 12: 78-149 Foster, C. B. 197 Permian plant microfossils from the Blair Athol Coal Measures, central Queensland, Australia. Palaeontographica, Abt. B, Palàophytol. B154: 121-171 Francis, E. H. 1983. ect eL Permian igneous maf Pp. 297-324 in G. Y. Craig (editor), Ge- of deren Wiley. New York. 935. The genus Pitus Witham emend. 11. - a new Car- banifero ous seed from East Lothian. Trans. Roy. Soc. Edinburgh 60: 427-464. 1052 Annals of the Missouri Botanical Garden GorHAN, W. 1912. Über die Gattung Thinnfeldia Et- PUPA Abh. Naturhist. Ges. Nürnberg 19: 67- NM ER, G. 1978. Seed dispersal by fish in the inundated regions of the Humaita, Amazonia. Bio- tropica 10: 170-183. Gourp, R. E. . Palaeosmunda, a new genus of siphonostelic osmundaceous trunks from the Upper Permian of Queensland. desinis 13: 10-28. 975. A preliminary report o etrified axes of Y 'ertebraria from the Permian of B. Australia. Pp. 109-123 in K. S. W. Campbell (editor), Gon- dwana Geology. pleas National Univ., Canberra. . DELEVORYAS. 77. The biology of Glos- sopteris: evidence from petrified seed-bearing and ollen-bearing organs. oe 1: 387-399. GRAND’ Eury, C. 19 és s graines trouvées atta- chées au Pecopteris Pius ke n Schl. Compt. Rend. Hebd. Séances kad Sci : 920-923 GRIME, J. P. 1 Plant ea and Vegetation Processes. Wiley, New En rk. Hau, J. W. & B. M. Sripp.. 1971. Ontogeny of Ves- icaspora, a late Pe mies pollen grain. Palaeon- tology 14: 431-436. 1975. Jurassic Environments. Cambridge . Press, Cambridge Haré, F., R. A. A. OLDE MAN & P. B. TOMLINSON. Tropical Trees and Forests. Springer, Berlin. HarLE, T. G. 1929. seed-bearing JE ali from the Permian of Chim. Konigl. Svenska Vete skapsakad. Handl. 6: 1-24. Harris, T. M. 33. A bod member of the Caytoniales. New Phytol. 32: 97-11 937. The fossil ^ut of Scoresby Sound, East Greenland, Part 5. Stratigraphical sepe of the ps beds. Meddel. Grenland 112(2): 1 940. Caytonia. Ann. Bot. nr 713- 1978. š d the pen M of Caytonia. Ann. Bot (London) 5: 47-58. 1943. Notes on s dom e of Yorkshire, 9. Ann. Mag. Nat. Hist. 10: 850-852. 951. The eren of the Caytoniales. Phytomorphology 1 —— H 19580. The seed + c tania. Palaeobotanist : 93- Te . 1958b. Forest fire in the Mesozoic. J. Ecol. 46: 447-453 . 1961. The Yorkshire Jurassic Flora. I. Thal- lophyta- ig a British Museum Natural His- tory, Lor 964. "The Yorkshire Jurassic Flora. II. Cay- toniales, Cycadales and ie E British Mu- seum Natural History, Lon 9 The Yorkshire Tusc Flora. III. Ben- nettitales. British Museum Natural History, London 1971. The stem of Caytonia. Geophytology l: 23- 9. The ao Jurassic Flora. V. Conif- re British Museum Natura . MILLINGTON & J. MILLER Yorkshi ire Jurassic Flora. anowskiales. British Museum Natural History, pes don. Harch, F. H. & G. S. CORSTORPHINE. ology of South Africa. Ma HeckeL, P. H. 1909. The Ge- cmillan, London n 1977. Origin of phosphatic black shale facies in duin oe of mid-continen- tal North America. Bull. Amer. Assoc. Petrol. Geol. 61: 1045- Heinrich, B. 1976. qiii. poo bog, wood- and and disturbed habitats. Ecology 57: 890-899. Hemos RT, C. 1 very br a ae in the Syd- ney Basin and Tamworth Synclinorial zone. Bull. Geol. Surv. N. S. W. 26: 274-293. HorpEN, R. 1917. On the anatomy of two Paleozoic stems from India. Ann. Bot. (London) 31: 315-326. Homes, W. B. K On some fructifications of the "pug rn from the upper Permian o S. W. Proc. Linn. Soc. N. S. W. 98: 131-141. New corystosperm ovulate fructifica- tions from the Middle ME of eastern Australia. Alcheringa 11: 165-17 Hucues, N. F. 1976. Palseobiology of Angiosperm Origins. Cambridge Univ. Press, Cambri du L, R. F. 55. The geology of the BD section f the Bowen Basin. Pap. Univ. Queensland, Dept. en 4(11): 1-43. JANZEN, D. H. 1978. Seeding patterns in tropical trees. Pp. 83-128 in P. B. Tomlinson & M. H. Zimmerman (editors), Tropical Trees as Living Systems. Cam- Univ. Press, Cambridge. Roots and root systems in tropical trees: morphologic and ecologic aspects. Pp. 323-349 in P Tomlinson & M. H. Zimmerman (editors), Tropical Trees as Living Systems. Cambridge Univ. Press, Cambridge JENNINGS, J. R. 1974. Lower Pennsylvanian plants of Illinois. I. A flora from the Pounds Sandstone Member p the Caseyville Formation. J. Paleontol. 48: 459- 413. JENSEN, A. R. 1975. Permo-Triassic stratigraphy and sedimentation in the Bowen Basin, Queensland. Bull. Bur. Min iis Resources, Geol. & Geophys., Canberra JizBa, K. M. M. 1962. Late Paleozoic bisaccate pollen from the United States mid-continent area. J. Pa- leontol. 36: 871-887 Joncmans, W. J. 1952. The female fructification "à Sphenopteris hoeningshausii and the supposed r lation of this species with Crossotheca. Paleobotórie: 1: 267-274. Kawase, M. & R. E. WHITMOYER. 1980. hae niuis TEY uM in waterlogged plants. Amer. J. Bot. KENT, P Eastern England from the Tees to the ash. Institute Geological Sciences, London. Mision. R. 1 On the fructifications of some ferns rom the Carboniferous formations. Trans. Roy. Soc. Edinburgh 33: 137-156 he flora of the parois on period, Part E Proe. Yorkshire Geol. Soc. -Preliminary note on ie: occurrence of eeu ia in the foliage of Lyginodendron. Proc. Roy. Soc. rad 76: 358- 360. 1 the microsporangia of the Pter dospermeae, S remarks on their relations hip to existing groups. Philos. Trans., Ser. B 198: 446. 1923. Fossil plants of the Carboniferous rocks of Great Britain. E Geol. Surv. Great Britain, Palaeontol. 2(1-3 1924. Fossil tie a the Carboniferous rocks Volume 75, Number 3 1988 Retallack & Dilcher Seed Ferns 1053 of Great Britain. Mem. Geol. Surv. Great Britain, Palaeontol. 2(5): 379-522. Kno tt, A. H. & E. S. BARGHOORN. 75. ys senqa eukaryotic ERE a reassessment of the evi- dence. Science 190: 52-54. KrassiLov, V. A. Tt. C iuda to the knowledge | f the Caytoniales. Rev. Palaeobot. Palynol. 24: 155- 78. KUKALOVA, J. 1970. Revisional study of the order Pa- laeodictyoptera in the Upper Carboniferous shales of ommentry, France, Part III. Psyche 77: 1-44 A.G. 19 Stamnostoma huttonense gen. et sp. ridosperm seed and cupule from the Calciferous Sandstone yug rd dos. ig Trans. ru Edinburgh 64: 5 On the oe of Samaropsis sco- tica (Calder) emended and Eurystoma angulare gen. et sp. nov., petrified seeds from the Calciferous Sand- stone Sonia: of Berwickshire. Trans. Roy. Soc. Edin- burgh ve 261-280 “Tristichia ovensi” gen. et sp. nov., a due Lower Carboniferous pteridosperm from Berwickshire and East Lothian, with an account of some associated seed and cupules. Trans. Roy. Soc. ae 64: 478 š 196 n some specimens of s Luis papilio Kidston associated with stems of Pitys. Tra Roy. pue Edinburgh 65: 211-224. Some specimens of Stenomyelon and Ka La from the Calciferous Sandstone Series of Berwickshire, Trans. Roy. Soc. Edinburgh 65: 435- LONG, o mn. ` 1965. On the cupule structure of Eurystoma angulara Trans. Roy. Soc. Edinburgh 66: 111- 128. 66. Some Lower Carboniferous fructifica- tions from Berwickshire, together with a pe aia account of the evolution of ovules, cupules and c ps Trans. Roy. Soc. Edinburgh 66: 345-375. 19 urystoma trigona sp. nov., pterido- sperm o ovules borne on a frond of Alc vp eA ——À us Roy. Soc. Edinburgh 68: . Further — ons on some vnm cupules. Trans. Roy. Soc. e dol seeds a Edinburgh 69: 267- 293. — 1976. Calathopteris heterophylla gen. et sp. nov., a Lower Carboniferous pteridosperm bearing two kinds ^e Minn Trans. Roy. Soc. Edinburgh 69: 327- 19770. Some Lower Carboniferous pterido- sperm cupules bearing ovules m microsporangia. Trans. T Soc. Edinburgh 70: 1-11. 77b. Lower usn pteridosperm cu- pules d the € of angiosperms. Trans. Roy. Soc Edinburgh 70: 35. 79a. Gig: N on the Lower Carbon- iferous EE duis Witham. Trans. Roy. Soc. Edin- burgh 70: -127. 97 “The resemblance between the "sd Carboniferous cupules of Hydrasperma cf. tenu ess and Sphenopteris bifida Lindley ve Hohe s. Roy. Soc. Edinburgh 70: 129-1 Tonite, A. 1929. Pride ia et JAHRE du Leptotesta grand’eu . gen. ., graine sili- cifée du Pecopteris 25 Schlot. Revue Gen- erale Bot. 41: 593-605. McCoy, F. 1847. On the fossil botany and zoology of the rocks associated with the coal of Australia. Ann. Mag. Nat. Hist. 20: 145-157. McDonneELL, K. L. 1974. Depositional environments of the Triassic, Gosford dcum Sydney Basin. J. Geol. Soc. Australia 21: 10 2. MacGnEGOR, A. R. Blackwood, Edinburgh. MAHESHWARI, H. K. 1965. Studies on the Mi dan flora of India. 23. On two fructifications from t Raniganj Stage of pA Panigen) Coalfield, Bengal. Pici 13: 144-147. 19 "bu wood from Antarctica and revision a some Lower Gondwana wood taxa. Pa laeontographica, Abt. B, Palaophytol. B138: 1-43. MaTTEN, L. C., W. S. Lacey & R. C. Lucas. 1980. Studies on the cupulate seed genus Hydrasperma Long, from Berwickshire and East Lothian in Scot- land. and County Kerry in Ireland. Biol. J. Linn. Soc. London 81: 249-273. 1984. Basic features of gymnosperm asiente ps phylogeny as evidenced by the fossil ev. (Lancaster) 50: 1-111. . W. BÉ 1982. Perminer- alized Alethopteris from the Upper Pennsylvanian of Ohio ju Illinois. J. ee 56: 392-402. MiLLAY, M. A. & D. A. EGGERT. 1974. Microgame- tophyte deco. in the Paleozoic seed fern family Callistophytaceae. Amer. J. Bot. 61: 1064-1075. T. N. TaYLOR. 1974. Morphological studies of Paleozoic saccate pollen. Palaeontographica, Abt. B, Paláophytol. B147: 75-99. Moonpir, R. L. 16. Coal measures amphibia E EE. America. Publ. Carnegie Inst. Wash. 238: Nami, M. 1976. An exhumed Jurassic nr ik Fife and Angus Geology. from TEE England. Geol. Mag. 113: 47-5 Neves, R., K. J. G e. rg CLAYTON, N. S. DIS R.S. W. NEVILLE & K. KRUSZEWSKA. 1973. Paly- nological Mekar l within the Lower Carboniferous of Scotland and no Em England. Trans. Roy. Soc Edinburgh 69: 23-7 NikraAs, K. J. 1981. a wind pollination and airflow around ovules of some early seed plants. Sci ence 211: 275-277 Nitsson, T. 1958. Uber das Vorkommen eines meso- zoischen Psp oa ap in Schonen. Acta Univ. Lunds 54: OESTRY ts L L E B. M. Sripp. 1976. Paracytic Ed vice e in Carboniferous seed ferns. Bu 193: OLIVER, F. W. € D. H Nori . On masm P 39 lomaxi, a be: A [ Lygina enam. Proc. Roy. Soc London 481. "21904. n the structure of the anis seed po e ee te lomax ent of Lyginodeaaien Philos. Trans. B197: 193-247. The Decade of North American P acd 1983 geologic time scale. Geology 11: 503- ER D. D. & K. L. GUPTA. 1968. Cuticular structure of some Indian Lower Gondwana species of Glossop- teris Brongniart, Part I. Palaeontographica, Abt. B, Paláophytol. B124: 45-81. D. NaurIYAL. 1960. Some seeds and spo- rangia of the Glossopteris flora from Raniganj Coal- field, India. Palaeontographica, Abt. B, Paláophytol. B107: 42-64. 1054 Annals of the Missouri Botanical Garden R. S. SINGH. 1974. On the stem and at- tachment of Glossopteris and Gangamopteris leaves, Part II. Structural ews aii a Abt. B, Paláophytol. B14 "Tu. PEDERSEN, K. R. & E. M. inne 1986. Caytonanthus pollen from the Lower and Middle Jurassic ben Essays Dept. Geol. Aarhus Univ. Denmark 1: 255- — PETRIELLA, B. 85. Caracteres viu ales y autoeco- ogia de las Corystospermeae. Pp. 57 in R. We- ber (editor), Simposio sobre floras del Triasico tardio, su fitogeographica y paleoecologia. Mem. Inst. Geol. Univ. Nac. Anton c PrEFFERKORN, H. W., W. H. Giese D. A. RESNICK & M. H. Sc HEIHING. 1984. onstruction and architecture of medullosan a AE cc (Pennsyl- m Mosasaur 2: 1-8. an). Pur E L. rope. & T. N. Taylor (editera), ru iy A of Fossil Plants. om Hutchinson oss, Stroudsburg, Pennsylva 9 Stratigraphic occurrences and vege- tational patterns of Pennsylvanian apa assasi in Euramerican coal swamps. Rev. Palaeobot. Palynol. 32: 5-26. & W. A. DIMIcHELE. 1981. Paleoecology of Middle a age coal swamps in southern Illinois. Herrin Coal Member at Sahara Mine no. 5. Pp. 231-284 in K. J. Niklas (editor), Paleobotany, Paleoecology and Evolution, Volume 1. Praeger, New York. , R. A. PEPPERS, M. J. AvciN & D. F. LAUGHAN. 1974. Fossil plants and coal: patterns of change in coal swamps of the Illinois Basin. Science 184: 1367- 1369. PLumsTEAD, E. P. 1958a. Further fructifications of the e and a provisional classification based n them. Trans. Geol. Soc. S. Africa 61: 51-76. 958b. The habit of growth “of the oe teridae. Trans. Geol. Soc. S. Africa 61: 81- 1963. The influence of plants and enviro ments on the developing animal life of Karroo um S. African J. Sci. 59: 147-152. Potonié, H. 1899. Lehrbuch der Pflanzenpaliontolo- gie. Berlin. Potonié, R. & G. Kremp. 1954. Die Gattungen der Palüozooischen Sporae ae p ^i Stratigra- phie. Geol. Jahrb. Beih. 69: 111- RAMANUJAM, C. G. K., G. s HWELL & W. N STEWART. 1974. Probable attachment of the Do- lerotheca campanulum to a M Transgressions and regressions in the Dinantian: a new synthesis of Brit- ish Dinantian stratigraphy. Proc. Yorkshire Geol. -607. 85. Origin of coal balls in the Illinois asin. Compt. Rend. 9th Int. Congr. Stratigr. Geol. Carboniferous, Washington and Urbana 4: 393-406. & T. L. P RAYMOND, A. HILLIPS. bin 83. Evidence for an Upper Carboniferous ve community. Pp. 19-30 in H. J. Teas aps Tasks for Vegetation cience, iud 8. W. he Hague. REIHMANN, M. A. & J. T. MU 1976. Cuticles of two species of Alethopteris. Amer. J. Bot. 63: 1039-1046. A reconsideration of the stomatal structure of Alethopteris sullivantii. Bot. Soc. Amer. c. Ser. Publ. 156: 33. REISZ, R. R., R. HEATON & B. R. 1982. Vertebrate fauna of Late Pennsylvanian Rock Lake hale, near Garnett, Kansas: Pelycosauria. J. Pa- 750. . 1976. Triassic palaeosols in the upper Narrabeen Group of New South Wales, Part 1. Fea- tures of the palaeosols. J. Geol. Soc. Australia 23: 399, 1 Reconstructing Triassic vegetation of eastern Australasia: a new approach for the biostra- bug of Gondwanaland. Alcheringa 1: 247-278. ——M 1980a. Towards a natural classification of seed ferns. P. 46 in Abstracts Ist Int. Palaeobot. Conf., Reading, England 1980b. Ee Carboniférous to Middle Triassic megafossil floras from the Sydney Basin. Bull. Geol Surv. N. S. W. 26: 384-430. 1980c. Middle Triassic megafossil plants and trace a from Tank Gully, eres New Zea- y. Soc. New Zealand 10: 31-63. 1981. Middle Triassic reaala plants from Long Gully, near Otematata, north Otago, New Zea- land. J. Roy. Soc. New Zealand 11: 167-200. —— ——-. 1984. Completeness of the rock and fossil record: some estimates using fossil sr Paleobiology 10: 59-78. 1986. The Holocene, Hitchcox limey peat soil as an analog for Pennsylvanian coals bearing coal balls. Pr Le 99th Annual Meeting Geol. Soc. Amer : L Dit CHER. 1981. Arguments for a glosopterid ancestry of angiosperms. Paleobiology 7: 54-67. , R. E. Goutp & B. RUNNEGAR. 1977. Isotopic dating of a Middle Triassic megafossil flora from New South Wales. Proc. Linn. Soc. N. S. W. 101: 77- Rex, G. M. 1986. Experimental modelling as an aid to E the gom three- P deus structures compressions. 17-38 in R. A. Spicer & B omas dica mido and Tax proaches erc ta Fionn: M. of Caytonia. P. 50-57 in M. N. Bose (e Symposium on Morphological and Stratigraphical Pa- inp Birbal Sahni Inst. Palaeobotany, Luck- now, Ricsy, J. F. . A revision of some plants from the Permian of the Bowen Basin, el qe oa our Geol. Surv. oun 349, Palaeontol. Pay : 1-8. urv. Queenslan . Al: HEKEL. 1977. Palynology of the Permian sequence in the Springsure Anticline, central Queens- land. Publ. Geol. Surv. Queensland 363, Palaeontol. Pap. 37: 1-76. Rorre, W. D. I. 1980. Early invertebrate terrestrial faunas. Pp. 117-157 in A. L. Panchen (editor), The Terrestrial Environment and the Origin of Land Ver- tebrates. Academic Press, London Volume 75, Number 3 Retallack & Dilcher 1055 1988 Seed Ferns ROTHWELL, G. W. fluvial system, c. Publ. Ontogeny of the sgh ovule Callospermarion pusillum. Amer. J. Bot. 58 706-715 2. Evidence of pollen E in Paleozoic pteridosperms. Science 175: 772- 1975. The c L n spermopsida): I, vegetative structures. Palaeonto- pe Abt. B, Paláophytol. B151: 171- 196. —— ———. 1977. Evidence for a pollination-drop mech- anism in Palaeozoic pteridosperms. Science 198: 1251-1252. The Callistophytaceae (Pterido- SHORE Abt. B, P 1981. The Callistophytales D E sida): reproductively sophisticated Paleozoic gym nosperms. Rev. Palaeobot. Palynol. 32: 103- 121. 1 New interpretations of the earliest co- nifers. Rev. ' Palaeobot. Palynol. 37: 7-28. 1986. Classifying the earliest gymnosperms. Pp. 137-161 in R. A. Spicer & B. A. Thomas (editors), Systematic and Taxonomic Approaches in Palaeobotany. Clarendon, Oxford. & . EccERT. 1982. What is the vascular architecture = complex medullosan pollen organs. A 9: 641-643. — ; 86. A monograph of Dolero- theca Halle, and related complex permineralized médullosan pollen organs. Trans. Roy. Soc. Edi s. 1977. Convergent floral evo- lution in South African and Australian Proteaceae and its possible bearing on RP in non-flying nn. Missouri Bot. Gard. 6 G. McCLUNc. 1975. ^ Foma time scale for Gondwanaland. Pp. 425-441 in K. S. W. Campbell (editor), Gondwana Geology. Australian Na- ti niv. Press, Canberra. SALISBURY, E. J. On the causes and ecological signification of stomatal frequency, with special rer erence to the woodland flora. Philos. Trans. B216 1-65 SavaGE, R. J. G. 1984. Mid-Jurassic mammals from Scotland. Pp. 210-213 in W.-E. Reif & F. Westphal (editors), Short Papers from the Third Symposium on Mesozoic Terrestrial Ecosystems. Attempto, Tu- bingen. SCHECKLER, S. E. 1976. Ontogeny of progymnosperms. I. Shoots of Upper Devonian Aneurophytales. Canad. J. Bot. 54: 202-219. 1978. Ontogeny of progymnosperms. II. Shoot of Upper Devonian Archaeopteridales. Canad. J. Bot. 56: 3136-317 ScnorF, J. "M. 1948. Pteridosperm male ei niaaa American species of Dolerotheca, with n re- garding certain allied forms. J. Paleontol. 22: '681- E — — 1975. Pennsylvanian climate in the United States. p Pap. U.S. Geol. Surv. 853: 23-31. —————. 1976. Morphologic iple apa of fertile structures in glossopterid gymnosp . Rev. Pa- laeobot. Palynol. 21: 25-64. 1982. Forms and facies of Vertebraria in boa to Gondwana coal. Amer. Geophys. Union, arc. Res. Ser. 36: 37-62 ScHUMM, S. A. 1981. Evolution and response of the Soc. Econ. Paleontol. uo rd 31: 19- 29 Scott, A. C. 1977. Coprolites containing plant material from the Carboniferous of Britain. Palaeontology 20: 59-68. 9. The ecology of coal measure floras from northern Britain. Proc. Geologists Assoc. 90: 97-1 & B. MEYER-BERTHAUD. 1985. Plants from the War of Foulden, Berwickshire, Scotland. Trans. Roy. Soc. Edinburgh Earth Sci. 76: 13-20. & G. ee 1985. The formation and signifi- cance of Carboniferous coal balls. Philos. Trans. B31 1: 123-137 . N. TaYLOR. 1983. Plant/animal inter- actions during pare Carboniferous. Bot. Rev. (Lancaster) f , J. GAL áfa: 4 TM 1984. Distribution of ener m preserved floras in the Lower Car boniferous in western Europe. Trans. Roy. Soc. Edin- burgh WU ^ 75: 311-3 Scorr, D. H. 19 Studies in Fossil Botany, lst edition. Adam * Cirio Black, London ; . Studies in Fossil Botany, Volume I. Pelion 2nd edition. Adam & Charles Black, London 1909. Studies in Fossil Botany, Volume II. Spermophyta, 2nd edition. Adam & Charles Black, London. SELLWoop, B. W. 1978. Jurassic. Pp. 204-279 in W. S. McKerrow (editor), The Ecology of Fossils. Mas- sachusetts Institute of Technology Press, Cambridge. SEWARD, A. C. Fossil Plants, Volume III. Pte- ridosperms, Cycadofilicales, Cordaitales, Coniferales, C oniferales, Gnetales. Cambridge Univ. Press, Cam- ridge. J. E. & P. G. GENSEL. 1980. A fertile specimen of Triphyllopteris from the Early Carboniferous (Mississippian) of southwestern Virginia. Amer. 0-4 SKOG, SMITH, D. L. 1962. Three fructifications from the Scot- tish Lower Carboniferous. Palaeontology 5: 225- SMOOT, E. L. 1984a. Secondary phloem anatomy in Callistophyton boyssetit Renault & Rothwell and histological changes in the outer phloem. Bot. Gaz sale ge 145: 395-406. — Phloem anatomy of the Carboniferous me ade Medullosa and evolutionary trends in gym m phloem. Bot. Gaz. (Crawfordsville) 1 45: 0-564. SoiL E IRVEY STAFF. 1975. E Taxonomy. Handb. U.S. * 1979. Principal compo- nents and correspondence analyses of quantitative data from a Jurassic plant bed. Rev. Palaeobot. Paly- 9 G. D. HungBLE, R. Brewer, K. H G. HaLLswORTH. 1968. A Handbook of Australian Soils. Rellim, Adelaide. STAFLEU, F. A. 1978. International Code of Botanical Nomenclature. Boh, Scheltema & Holkema, Utrecht. 1056 Annals of the Missouri Botanical Garden STANLEY, S. M. 1985. Earth and Life Through Time. 944. The anatomy and affinities of Medullosa noei Steidtmann and associated foliage, roots and seeds. Contr. Mus. Paleontol. Univ. Mich igan 6: 131-166. STERNBERG, G. K. 1838. Pp.81- 220 in Versuch einer Flor botanischen D a der Vorwelt. Tentamen, Leipzig. STERZEL, J. T. 1881. Paláontologischer Charakter der oberen Steinkohlenformation und des Rotliegenden in ergebirgischen Becken. Naturwiss. Ges. Chemnitz Ber. 7: 155-170. STEWART, W. N. 1954. The structure and affinities of Pachytesta illinoense comb. nov. Amer. J. Bot. 41: -50 f Plants. 983. eobotany and the Evolut Cambridge Univ. Press, Cambridge. —— ——— & T. De gg e 1956. The MD eridos ap Bot. Rev. (Lancaster) 22: 45-80 bs B. M. 1981. B status of nm seed ferns. Rev. Palaeobot. Palynol. 32: 63-101 & J. W. Hari sig be ¡nerd callis- tophytoides, gen. et sp. he probable pollen- bearing organ of e seed fern THO UH Amer. 7: 394-403. J. Bot. 1970b. The natural affinity of the Sepre seed Callospermarion. Amer. J. Bot 7-836. E M. C. & D. M. S. Watson. 1909. On the present distribution and origin of the calcareous con- cretions in coal seams, known as coal balls. Philos. rans : 167-218 Srun, D. R.J. 1877. Die Culm-Flora der Ostrauer und Wa soupa Schichten. Kgl. K. Geol. Reichsanst. . 8: 1-366. A SURANGE, K. R. & S. CHANDRA. 1975. Morphology of gymnospermous fructifications of the Glossopteris flora and their eee Pi a ae Abt. B, Paláophytol. B149: & MED ARI. EA Some male and female fr ucllications of Glossopteridales from India. Palaeontographica, Abt. B, Paláophytol. B129: 178- TAYLOR, T. As 1965. Paleozoic seed studies: a mono- grap e American species of Pachytesta. Pa- leontographica. Abt. B, Palaophytol. B117: 1-46. 1978. The ultrastructure and saura a Monoletes A pollen. Canad. J. B 56: 310 y. McGraw-Hill, New York. . 1982. Uliracteuce ral studies of Paleozoic seed fern pollen: — m development. Rev. Palaeobot. Palynol. 37: 29-5 M. A. Musis 1977. dq i ini dip dn fossil cell contents. Trans. r. Microscop. Soc 96: 390-393. . A. CICHAN & A. M. BaLponi. 1984. The ultrastructure of Mesozoic pollen: Pteruchus dubius (Thomas) Townrow. Rev. Palaeobot. Palynol. 41: 319-327. MENS H. H. 19 The Caytoniales, a new group angiospermous plants from A urassic rocks of York- shire. € Trans. B213: 2 —— 1933. On some se s, ous plants from the Pod of South Africa. Philos. Trans. B222: 193-265 934. The nature and origin of the stigma. New Phyl 33: 173-198. —— 1952. A Glossopteris with whorled leaves. Picea 1: 435-438. & M. N. Bose. 1955. Pac añ rmophyllum papillosum gen. et sp. nov., from the Yorkshire Jurassic. Ann. Mag. Nat. Hist. 12: 535-543 Tonks, L. H., R. C. B. m W. LLovp & R. L. SHERLOCK. 1931. The geology of Manchester and the southeast Lancashire Coalfield. Mem. Geol. Surv. England & Wales, p of Sheet 85: 1-240. Townrow, J. A . The genus pe aiti and its southern hemisphere species. us rske Vi- densk.-Akad., Mat.-Naturvidensk. KI. 1960. The Peltaspermaceae, a e er ois sper family of Permian and Triassic age. [n nie tiis 33-361. 1962a. of the Seng oce da 316. Hist.), Ane 6: 285 On Pteruchus, the arri cda . Mus. ayes nomenclature of Pteruchus mo (FeistmanteD) ii nov. & Proc oy. Soc. Tasmania 98: -118 1962c. On some 2 ccate pollen grains of Permian to Middle Jurassic age. Grana Palynol. 3: 13-44. Turner, B. R. 1978. Palaeohydraulics of clast trans- port during deposition of the Upper Triassic Molteno Formation in the main Karroo Basin of South Africa. sichtigung ihrer igr c e Sudlinberger Kohlenreviers. Meded. Rijks Geol. Dienst 3-1(7): 1-208. VAN DER PIJL, L. 1972. Principles of Dispersal in Higher Plants. Springer, Berlin. VAN DER ZWAN, L. J., s ,BOUDTER. & R. N. L. B. HUBBARD. 1985. Climatic rboniferous in a based on multivariate statistical analysis of palynological data. Palaeogeogr. Palaeoclimatol. Palaeoecol. 52: 1-20. VAN KONIJNENBURG-VAN CirrERT, H. H. A. gymno ny pollen from the Middle Jurassic of situ Yorkshire. a Bot. N eerl. 20: 1-96. Wacner, R. H. 19 Stephanian flora in NW Spain with special reference to the nd D.Ste- phanian A bou pio De ? Congr. Car- bonifére je 2: 835 1 Upper worin and Stephanian species of. Alethopteris from Europe, Asia Minor and rth America. Meded. Rijks Geol. "Dienst C3-1(6): 18. ^ WALKOM, A. B. 1922. Palaeozoic floras of i ad “sasan Part I. The flora of the Lower and U ontributions to the kno owledge of Low- er Carbonifasatis plants. Philos. Trans. B215: 201- 224. Contributions to the knowledge of the Lower Carboniferous plants (cont'd). III. On the fossil flora of the Black Limestones in Teilia Quarry, Gwaenysgor, near Prestatyn, Flintshire, with special Volume 75, Number 3 1988 Retallack & Dilcher Seed Ferns 1057 reference to Diplopteridium teilianum Kidston sp. (gen. nov.) and some other fern-like fronds. Philos. Trans. B219: 347-379 . 1964. The ae Trans. Bot. Soc. Edinburgh 39: 449-459. WHITE, E. I. 1927. The fish fauna of the Cementstones of Foulden, Berwickshire. Trans. Roy. Soc. Edi burgh 55 WHITE, M. E. 1964. Reproductive structures in Aus- tralian Upper Permian Glossopteridae. Proc. Linn. Soc 1. 978. "Reproductive structures of the Glos- sopteridales i in the plant fossil collection of the Aus- tralian Museum. Rec. Austral. Mus. 31: 473- a WHITEHEAD, D. R. 1969. Wind pollination in the a poss evolutionary and environmental saber] rations. Evolution 23: 28-35. Wi LIAMSON, W. C. 1874. On the organization of the fossil plants of the Coal Measures— Part 6, Ferns. Philos. Trans. 164: 675-703. 1875. On the organization of fossil plants of the Coal Measures— Part 7. Myelopteris, — and Kaloxylon. Proc. Roy. Soc. London 23: 69- 3. WiLsoN, V. 1948. East Yorkshire and Lincolnshire. British Geological Survey & Museum, London WITHAM T 1 . The Internal Striature of Fossil Vegetables Found in the Carboniferous and Oolitic Deposits of Great Britain. Adam & Charles Black, Edinburgh WNuK, C. & H. W. PFEFFERKORN. 1984. The life habits C Rev. Palaeobot. Palynol. 41: 329- WODEHOUSE, R. P. 1935. Pollen Grains. McGraw-Hill, ew York. Woop, S. P. & W. D. I. RoLFE. to the palaeontology of 1985. Introduction the Dinantian of Foulden, Berwickshire, Scotland. Trans. Roy. Soc. Edinburgh arth Sci. 76: ZANGERL, R. & G. R. T 1973. Iniopterygia, a new order of Chondrichthyan fishes from the Pennsyl- vanian of North America. Fieldiana Geol. Mem. 6 1-6 E. S. RICHARDSON. 1963. The paleoecolog- ical history of two inr black shales. Field- iana rar y ZIEGLE R. SCOTE W. . McKerrow, M. E. pee & R. K. i rdc H. 1979. Paleozoic ron Ann. Rev. Earth Science 7: 473- us. E. L. & K. McCapusu. 1980. On a Tri- gonocarpus species attached to Neuropteris (Mix- oneura) flexuosa from the Sydney Coalfield, Cape Breton Island, Nova Scotia, Canada. Rev. Palaeobot. Palynol. 30: 57-66 FLORA OF THE VENEZUELAN GUAYANA—V' Julian A. Steyermark? ABSTRACT In the preparation of various families for the Flora of the Venezuelan Guayana, many new taxa have been detected. These in eii aee imens collected from ipsc completed expeditions and earlier herbarium material. The following new ta insculpta, M. longistipitata M. neblinae, M. pustulat longipes subsp. yutajensis, M. eii var. ellen dulifera, T. heterodoxa, T. antha, T. san are described: Portulaca — ia Rhamnus longipes, R. dug (Rhamnaceae) ; Pochota ewelii, P. fuscolepidota, P. gracilis subsp. bolivarensis, densi : : P. liesneri, P. mawarinumae, P. redm naceae); Remijia sessilis, Simira ignicola (R aceae). In addition, comments are supplied, where MAE. on the affinities of various taxa in the several families treated. PORTULACACEAE PORTULACA Portulaca insignis Steyermark, sp. nov. TYPE: nezuela. Bolivar: laja granitica con bosque y matorral seco, cerca del afluente del Rio Aro, carretera Ciudad Bolivar-Maripa, 100 m W of Ciudad Bolivar, 100-200 m, 26 Feb. 1980, G. Morillo & G. Carnevali 8260 (holotype, VEN). Figure 1. nta herbacea annua diminutiva 4.5-10 alta, a pilis. lanosis albis "od mnino ii eran ernis aminis bee d herbe xir era basi in poe me an- gustatis 4.5-6 mm lon atis; petiolis 0.8 mm longis 0.6 mm sd "tli involucralibus 1 10-12 fo- liorum caulinorum similaribus; sepalis 4 mm longis; petalis 4-5 mm longis; staminibus 13-15; se stigmatosis quat- tuor; capsulis 2.5 mm longis 1.5 mm diam. prope mediu circumscissilibus; seminibus numerosis lenticularibus 0.5- .6 mm diam., a latere compressis conspicue tuberculatis, tuberculis apice oe anguste elevatis basi 4-5 ra- iatis angustis ornati Herbaceous diminutive annual, 4.5-10 cm tall with an elongated slender tap root. Stems several, procumbent to erect-ascending or curving, simple or sparingly branched, densely foliose up to the apex, densely covered and nearly concealed by a dense, white, woolly mass of elongate hairs 8-15 mm long; internodes conspicuously abbreviated. Leaves alternate, densely imbricate to the apex of the stem, fleshy, flattened, lance-linear, obtuse at ex, 4.5-6 mm long, 0.5-1 mm wide, base at- tenuate. Petioles 0.8 mm long, 0.6 mm wide. In- volucral leaves 10-12, similar to the cauline, 4— 7 mm long, 1 mm wide. Sepals 4 mm long, sub- acute. Petals 4-5 mm long, slightly connate at base. Stamens 13-15; filaments ca. 2 mm long, attached 0.5-0.8 mm above base of petals. Stig- matic lobes 4; style filiform, 0.8 mm long. Capsules 2.5 mm long, 1.5 mm diam., circumscissile near the middle. Seeds numerous, lenticular, 0.5-0.6 mm diam., laterally compressed, conspicuously tu- berculate, the tubercules with a subacute apical projection, at the base with 4-5 spreading narrow rays Paratype. VENEZUELA. BOLÍVAR: Dist. Paez, Laguna Larga, 8 km of Upata, sabanas con chaparrales, 350 m, July 1978, Delascio & Liesner 6882 (MO, VEN). In this species the entire plant is more or less concealed by a dense cover of white, woolly or cottony hairs up to 15 mm long. It differs from Portulaca elatior Mart. in its much smaller size, shorter stems, shorter cauline and involucral bracts, fewer stamens, 4-fid style, and much denser woolly covering over the entire plant. Portulaca pygmaea Steyermark, sp. nov. TYPE: Venezuela. T. F. Amazonas: 1 km E of Hotel Amazonas, Puerto Ayacucho, gravel pockets ' Flora of the Venezuelan Guayana was carried out under National Science Foundation Grant (BSR) 85152085. lam greatly indebted to t 2 Missouri Botanical Garden, P.O. Box the curators of F, es US, MER, MY, PORT, and VEN for th 99, St. Louis, Missouri 63166-0299, U.S. 4 e loan of critical material. ANN. Missouri Bor. Garb. 75: 1058-1086. 1988. Volume 75, Number 3 1988 Steyermark Venezuelan Guayana Flora 1059 1.5mm K BAKE ING CR i-o») ¿ URS DK UC. &@ SS FIGURE 1. view.—C. Single leaf, detached, the venation as seen in t tubercules.—E. Habit of younger plant.—F. Detail of surface of seed showin stellulate rays. on laja, along contact between laja and Ori- noco gallery forest, 100 m, 12 Nov. 1953, Bassett Maguire, John J. Wurdack & George S. Bunting 36188 (holotype, NY). Figure 2. Planta herbacea diminuta 2-3 cm alta glabra vel fere; axillaribus inconspicuis 0.2-0.6 mm longis; petiolis 0.7-1 mm longis; sepalis suborbicularibus obtusis, 3 x 3 mm; petalis luteis 4.5-5 mm longis; capsulis US "t Portulaca insignis.— 4A. Habit of fully mature plant with woolly aspect.—B. Capsule, exterior in trans parent light. —D. Seed with small projecting g tubercule raised at apex above depresso-hemisphaericis prope basem circumscissilibus longis 1.7-2.3 mm latis; seminibus griseis . am., superficie tuberculato, tuberculis tessel- latis haud projectantibus marginibus arcte contiguis. Diminutive herbaceous plant 2-3 cm tall, gla- brous or nearly so; roots napiform, the short tu- berous root 1-1.5 cm long, 4-8 mm thick. Leaves subtending the capsules opposite, otherwise alter- nate; leaf blades suborbicular, oval, or obovate, rounded at the apex, acute at base, 1.5-6 mm long, 1.5-3 mm wide, glabrous or sometimes with minute hairs present in the axils; petioles 0.7-1 1060 Annals of the Missouri Botanical Garden A ES P LX PATH] ESSA TIC 0.5mm ¿A A 2 x s SS T an zm Cc IRE 2. Portulaca pygmaea. — 4 d Habit, showing tuberoid root.—B. Capsule subtended by uppermost bracts. —D. Ovate-triangular bracteole. —E. Petal subtended by sepal 3mm , opposite leaves. —C. Pistils subtended by epal, dorsal view.—G, H. Leaves, showing different shapes.—l. Seed, showing platelike, lateral view.—F. subflattened tubercules. mm long. Flowers bibracteate, slightly substipitate, the bracts 1 mm long. Sepals suborbicular, broadly obtuse, 3 X 3 mm. Petals yellow, ligulate-oblong, rounded at the apex, 4.5-5 mm long, 2 mm wide in the upper half, 0.7-1 mm wide at the base. Stamens 5; filaments 2.7 mm long; anthers subor- bicular, 0.5 x 0.5 mm. Style 3 mm long; stigmas 3, spathulate, 1.2 mm long. Capsules hemispher- epressed at the apex, 1.5-2.3 mm long, ical, subd 1.7- mm wide, circumscissile near the base. Seeds gray, 0.5 mm diam., tessellate-tuberculate, the tubercules scarcely projecting, their margins strongly contiguous. This is a rarely collected species, occurring on the lajas in the vicinity of Puerto Ayacucho with Portulaca pusilla H.B.K. and P. sedifolia N. E. Brown. It differs from the very similar dwarf P. pusilla in the yellow petals; napiform, short, thick- ened tuberous root; 5 stamens; longer and broader Steyermark 1061 Volume 75, Number 3 1988 Venezuelan Guayana Flora 1.8mm FIGURE 3. corolla with disk and longitudinal section through ova turned back to show minute puberulence on corolla tube.— views leaf blades; depressed-hemispherical capsule open- ing very close to the base; and seeds with gray, flattened, contiguous tubercules that do not pro- trude above the general surface contour. It differs from P. sedifolia and P. teretifolia H.B.K. in its general glabrity and leaf shape. OLACACEAE SCHOEPFIA Schoepfia clarkii Steyermark, sp. nov. TYPE: Venezuela F. Amazonas: Mari's bana (low Amazon caatinga), 10.8 km NE of San Carlos on Solano road, 1°56'N, 6793' W, 119 m, 16 uug', Schoepfia clarkii.— A. Habit of flowering branch.— B. Flower. —C. Inflorescence. —D. Interior of h ry in position.—E. Anther, ventral view.—F. SN G, H.— United bracteoles and bract (epicalyx), t Aug. 1987, H. L. Clark 8111 (holotype, MO). Figure Frutex 2 m; ' folio orum laminis anguste lanceolato-ellip- m acutis, 4.5- lateralibus obsoletis utroque latere 3-4; p longis; inflorescentia axillari spiciformi simplici usque a 7-flora 6-7 mm longa: pedunculis ut videtur solitariis; floribus — bracteolis e e supra medium connatis 1.5-1.8 m apicem atis extus don papillato- ‘puberulentibus: corals lait cylindrico, tubo ongo l. o glabro, lobis 1.2 x 1.2 mm; ovario infero 0.8 mm Eon Shrub 2 m tall. Leaf blades narrowly lance- elliptic, acute or subacute at apex, cuneately acute 1062 Annals of the Missouri Botanical Garden at base, 4.5-5.5 cm long, 1.2-1.5 cm wide, gla- brous, entire; lateral nerves obsolete, 3-4 each side; petioles 4-5 mm long. Inflorescence axillary, spiciform, simple, up to 7-flowered, 6-7 mm long; peduncle solitary, 2 mm long. Flowers sessile. Bracts and bracteoles (epicalyx) connate more than half- way, forming a 3-lobed involucre 1.5-1.8 mm long, 1.5 mm wide, narrowing to 0.9 mm wide at base, densely papillate-puberulent without; 1 larger lobe broadly triangular-ovate, abruptly acute at apex, 0.8 mm long, 0.9 mm wide at base; 2 smaller lobes narrower, broadly triangular, slenderly acu- minate-attenuate, 0.5 mm long, 0.4 mm wide; tube of epicalyx shallowly campanulate, 1 mm long, 1 mm broad at summit, 0.6 mm broad at base. Co- rolla greenish yellow, thick-cylindric, the tube 1.8- 2 mm long, 1.6 mm wide, glabrous without, gla- brous within except for a minute tuft of papillate hairs behind the anthers; lobes 4, spreading-squar- rose, broadly triangular-ovate, obtusely acute or subacute at apex, 1.2 x 1.2 mm. Stamens 4, subsessile, suborbicular, 0.2 mm long, 0.3 mm wide; filaments 0.2 mm long. Disk depressed-sub- globose, fleshy, annular, 0.7 mm long, 1 mm wide. Ovary inferior, 0.8 mm long, 1.3 mm broad at the truncate summit, narrowed at base to 0.3 mm wide. Style 1.8 mm long; stigma capitate, 0.4 mm long, 0.7 mm wide. This taxon is characterized by small corollas with a short tube 1.8-2 mm long, bracts and bracteoles of the epicalyx upwardly connate more than half- way, narrowly lance-elliptic leaf blades up to 1.5 cm wide, and spiciform solitary inflorescence with sessile flowers on a short rachis. CELASTRACEAE MAYTENUS In preparing an account of the genus Maytenus for the Flora of the Venezuelan Guayana, the near uniformity of the floral structures and general inflorescence types has necessitated greater reli- ance on vegetative characters, especially those of leaf venation. [n the majority of the taxa studied, the calyx lobes are uniformly suborbicular and rounded with erose-fimbrillate or ciliolate margins. Rarely, as in M. apiculata Steyerm. and M. ka- nukensis A. C. Smith, they are acute or apiculate. Likewise, the fruit does not vary sufficiently to separate taxa. [n a few cases, such as in M. apic- ulata, the apex is noticeably prolonged to form a more prominent beak, while in M. longistipitata Steyerm. the base is greatly extended into a longer stipitate portion. In both M. longistipitata and M. oblongata Reiss., the fruit is larger than in the other species studied. In M. neblinae Steyerm. the leaves are smaller, while the leaf margins are more prominently dentate with more numerous teeth than in the other taxa examined. An insculpted type of venation characterizes the leaves of M. insculpta Steyerm., whereas a pustulate lower leaf surface is characteristic of M. laevis Reiss., M. huberi Steyerm., and M. pustulata Steyerm. In M. sie- beriana Krug & Urban, the upper leaf surface has a finely lineolate-striolate character. A partial treatment of the genus was published by Reissek (1861) for the Brazilian species. An account of the West Indian taxa by Urban (1904), together with descriptions of additional South American species y Briquet (1919) and A. C. Smith (1939a, b) e the remainder of the known aes from South America and the West Indies. The numerous collections from the Vene- zuelan Guayana have yielded the new taxa de- scribed here. LITERATURE CITED 1919. Celastraceae, XVII. P ei ioa . Bot aire Conserv. & Jar BRIQUET, J. tarum novarum. Annua Genéve 20: 342-367. REISSEK, S. 1861. AR In: C. Martius, Flora Brasiliensis 11(1): 1 me A.C. 1939a. Arbor. 20: 294-2 1 b. Notes on a collection of plants from Br itish Guiana. Lloydia 2: 189-1 Urban, I. 1904. Symbolae Antillanae 5: 53-72. a Krukovianae VI. J. Arnold 295. Maytenus huberi Steyermark, sp. nov. TYPE: Venezuela. T. F. Amazonas: Dept. Atures, cuenca del Rio Manapiare, sabanas situadas en los cerros ubicados entre el Cerro Morrocoy al sur y Serrania Colmena al Norte, 5?21'N 66°10'W, 200-300 m, 29 Jan. 1977, Otto Huber 441 (holotype, VEN). 2-6 metralis; foliorum laminis crasso-coriaceis ap acutis basi rotundatis b e ut pP E vel rar- iter acutis (4.5-)6-15.5 cm longi .9-)3-9 cm latis o: "pedicellis sub irit m longis fructu 3.5-5 mm longis; ibd lobis cues pie rotundatis ciliolatis; capsulis majoribus obovoideis apice rotundatis 10-11 mm longis 7-8 mm latis Shrub 2-6 m tall. Leaf blades thick-coriaceous, often pruinose or subglaucous on one or both sur- faces, oblong, elliptic-lanceolate, or ovate-oblong, broadly obtuse or obtusely acute at the apex, Volume 75, Number 3 1988 Steyermark 1063 Venezuelan Guayana Flora rounded, subcordate, subobtuse, or rarely acute at the base, (4.5-)6-15.5 cm long (2.5-)3-9 cm wide, entire; lateral nerves obsolete or inconspic- uous, 8-10 each side; tertiary venation obsolete; both surfaces ceca Petioles 4-11 mm long, 1.5-2 mm e. Inflorescence 1-8-flowered, fas- ciculate; dicas 1.5 mm long in anthesis, 3.5-5 mm long in fruit. Calyx lobes suborbicular, round- ed, ciliolate. Petals broadly ligulate-oblong, round- ed apically, 1.1-1.5 mm long, 0.8-1 mm wide. Capsules obovoid, rounded at summit, mature ones 10-11 mm long, 7-8 mm wide. Paratypes. VENEZUELA. ATURES: cuenca del Rio Manapiare, sabanas al los cerros al N de Cerro pie de Morrocoy, alrededores del sitio Pozo de la Carolina, 12 m E de Sanariapo, 5?15'N, 67 16764 (MO, VEN); vegetación de laja y Luo granitico en rauda eza en el Rio , 4°48’ 67°18'W, 100-115 m, Guánchez $ j^ aa 3403 (MO, TFAV). This species is distinguished by the thick-cori- aceous, pustulose or verruculose leaf blades; these are usually pruinose on one or both surfaces, often drying discolored, and have scarcely evident or obsolete venation adaxially. It differs from the new- ly described M. pustulata Steyerm. in the pruinose leaf blades, inconspicuous or obsolete nervation of the upper leaf surface, and the mainly rounded or subcordate leaf base. Maytenus insculpta Steyermark, sp. nov. TYPE: Venezuela, T. F. Amazonas: Dept. Atures, cuenca del Rio Manapiare, sabanas al pie de los cerros al N del Cerro Morrocoy, alrede- dores del sitio “Pozo de la Carolina," 12 km W of San Juan de Manapiare, 05?19'N, 66?6'W, 225 m, 16 Oct. 1977, Otto Huber 1229 (holotype, VEN; isotype, MO). Frutex 3-metralis, ramulis juvenilibus teretibus tenuiter coru foliorum laminis elliptico-oblongis apice obtuse acuminatis basi su a vel subacutis 11-18 cm longis s haud revolutis integerrimis, ner- decurrentibus; petiolis 6-9 mm m lycis lobis suborbicularibus rotundatis 0.8 x 0. ciliato-erosis; petalis ligulato-oblongis 1.5 mm longis 0.8 mm latis. Shrub 3 m tall, the young stems terete, finely corrugated. Leaf blades subcoriaceous, drying con- colorous, elliptic-oblong, obtusely acuminate at apex, the acumen 3-7 mm long, 3-4 mm wide, subobtuse or subacute at base, 11-18 cm long, 5- 7 cm wide, the margins not revolute, entire; both lateral and tertiary nervation completely finely in- sculpted on both sides; lateral nerves very faint, 7-10 each side; lower and upper surfaces pustu- late. Petioles 6-9 mm long, the base of leaf blade inconspicuously decurrent. Inflorescence few-flow- ered, shortly pedunculate; peduncles 1.5-2.5 mm long; pedicels 1 mm long. Calyx lobes suborbicular, rounded, 0.8 x 0.8 mm, strongly erose-ciliolate. Petals ligulate-oblong, 1.5 mm long, 0.8 mm wide. This species is distinct in having the venation of both sides of the leaf blade delicately insculpted with a prominent pustulate surface between the veins. It is related to Maytenus nitida of Brazil, from which it differs by having rounded, suborbic- ular, strongly ciliolate calyx lobes, and by having larger leaf blades that are more conspicuously pus- tulose on the lower surface. Maytenus laevis Reissek, in Mart., Fl. Bras. 11(1): 27. 1861 This taxon, as Reissek noted, has the upper portion of the young stems distinctively carinate- angled and leaves “sub lente marginatis repan- disque” and “‘subtus opacis scrupulosis." The leaves, petioles, and the young branches often take on a grayish color upon drying. While many of the specimens can be satisfactorily delimited by the application of such vegetative differences, one en- counters atypical specimens on the higher slopes and summits of the table mountains in the Terri- torio Federal Amazonas and Estado Bolivar that indicate possible introgression from another source, such as Maytenus guianensis Kl. Specimens from the summit of Cerro Guaiquinima, Estado Bolivar (Steyermark et al. 113333, 113424, 113428, 117216), from Cerro Sarisarinama, Estado Bolivar (Steyermark et al. 109252 —type of M. jauaensis Steyerm.), and from Cerro Yutaje, Territorio Fed- eral Amazonas (Maguire & Maguire 35441, 35455; Holst & Liesner 3272, 3357; Liesner & Holst 21807) generally preserve the grayish dried leaf blades, petioles, and young stems as well as the thickened repand leaf margins, but the cari- nate-angled young stem is either lacking or scarcely apparent. These specimens may represent a vari- ation of this taxon at higher altitudes, or may indicate introgression from M. guianensis, which possesses terete young stems and occurs in the Paragua and the Caura river basins, where Cerro Guaiquinima and Cerro Jaua are located, respec- tively. 1064 Annals of the Missouri Botanical Garden Maytenus longistipitata Steyermark, sp. nov. TYPE: Venezuela. Bolivar: Cerro Uroi, summit, north portion, Rio Uroi, Rio Chicanán, 700 m, 12 Sep. 1962, B. Maguire, J. Steyermark & C. K. Maguire 53731 (holotype, VEN; isotype, MO) Arbor 10 metralis, foliorum laminis lanceolato-ellipticis apice basique acutis 15-18 cm longis 5.5-6 cm lati integerrimis; nervis lateralibus principalibus utroque | latere 8-10 tenui d ane rotundatis 21-26 p stipite 5-7 mm longo. Tree 10 m tall. Leaf blades lance-elliptic, acute at apex and base, 15-18 cm long, 5.5-6 cm wide, entire; main lateral nerves 8-10 each side, with finer intermediate nerves, impressed above, slightly elevated below, arcuate-ascending at an angle of 35—45*, scarcely anastomosing; tertiary venation fine, laxly irregularly reticulate and subelevated below, inconspicuous above. Petiole 8-9 mm long. Fruiting pedicels fasciculate, 9-10 mm long. Cap- sules broadly obovoid, rounded at summit, 21-26 mm long, 12-14 mm wide at and above the middle, conspicuously long-stipitate, the stipitate portion 5-7 mm lon The conspicuously stipitate fruit readily distin- guishes this species Reiss from Maytenus oblongata Maytenus neblinae Steyermark, sp. nov. TYPE: Venezuela. T. F. Amazonas: Cerro de la Ne- blina, ridge at divide between Brazil and Ven- ezuela, 26 km ENE of Base Camp, wet cloud forest on steep SW-facing slopes, 0?53'N, 65°56'W, 2,000 m, 15 Apr. 1984, Timothy Plowman & Wayt Thomas 13610 (holotype, MO; isotypes, F, MO, VEN). Arbor 8-metralis; foliorum laminis ovatis vel elliptico- ovatis vel oblongo-elli ipticis apice obtusis vel obtusiusculis basi acutis vel subacutis 4 5 cm qiix 2.5-3.5 cm latis, utroque margine a 30- 40 crenulato- den- tatis praedito, dentibus 7-8 en supra manifesta subtus paullo manifesta; pagina inferiore subpustulata; petiolis 6-9 mm longis. Tree 8 m tall. Leaf blades coriaceous, ovate or elliptic-ovate or oblong-elliptic, obtuse at apex, acute or subacute at base, 4.5-9.5 cm long, 2.5-3.5 cm wide; margins uniformly 30-40-crenulate-den- tate, the dentations 7-8 per cm, conspicuous, 1— mm wide. Lateral nerves 8-13 each side, im- pressed or subelevated and more conspicuous on the upper side, scarcely evident beneath, ascending at an angle of 45—50°, anastomosing 3-6 mm from the margin. Petioles 6-9 mm long. This taxon differs from some of the variations of Maytenus guianensis Kl. and M. ficiformis Reiss. with crenulate margins, in the smaller, dif- ferently shaped leaf blades with more numerous teeth per centimeter. Maytenus pustulata Steyermark, sp. nov. TYPE: Venezuela. T. F. Amazonas: Cerro de la Ne- blina, north branch of river in canyon, Camp IV, 15 km NNE of Pico Phelps, 0?51'N, 65°57'W, 780 m, 5 Mar. 1984, Ronald Lies- ner 16720 (holotype, MO; isotype, VEN) Arbor 5- — foliis PX a oblongis apice obtusis, acumine 3-4 mm lato, tusis vel subacutis 14-17 cm longis 6- 75. cm latis noes nervis lateralibus utroque latere gus supra paullo elevatis subtus obsoletis; venatione tertiaria ubique obsoletis, in superficie inferiori fus jure petioli 11-18 mm longis; pedicellis fruc- 10 mm longis; fructu elipscideo- eee 15- 16 1 mm Dien! 8-9 mm lato, apice rotundat ct =. a Tree 5 m tall. Leaf blades coriaceous, 14- 17 x 6-7.5 cm, elliptic-oblong, narrowed to an obtuse apex with an acumen 3-4 mm wide, obtuse to subacute at base, decurrent on the petiole; lateral nerves 9-13 on each side, slightly elevated on upper surface, obsolete on lower surface; tertiary veins obsolete both sides, the lower surface pus- tulate. Petiole 11-18 mm long, strongly canali- culate, 2-2.5 mm wide. Fruiting calyx lobes 0.5 mm long, rounded, unequally long-ciliolate around apex. Fruiting pedicels 6-10 mm long, 1-1.5 mm wide. Fruit ellipsoid-obovoid, 15-16 mm long, 8- 9 mm wide toward summit, narrowed to 2.5 mm wide basally, rounded at summit. The pustulate lower leaf surface, obsolete ter- tiary venation, large leaf blades, and stout pedicels distinguish this species from Maytenus huberi Steyerm., in which the tertiary and lateral nerve venation is insculpted. ZINOWTEWTA Zi ardii Steyermark, sp. nov. TYPE: Venema. Bolivar: Dist. Sifontes: bosques hú- medos intervenidos por actividades mineras del sector “La Hoya” (Peray-tepuy), 7 km NW of Caserio El Pilón, 58 km W of Sta. Elena de Uairén, 4?40'N, 61°33'W, 850 m, 21 Oct. 1986, Gerardo Aymard 4631 (ho- lotype, MO; isotype, PORT Volume 75, Number 3 1988 Steyermark 1065 Venezuelan Guayana Flora Arbor 10-45 metralis, foliorum laminis ovatis vel lan- ceolato- o wed obtuse acuminatis basi attenuato- acutis vel a s decurrentibus 5.5-10 cm longis 2.5-3.5(-4. 5 c cm n la ud nervis lateralibus utroque latere (5-)8-12 supra subelevatis, venatione tertiaria ubique manifestis grosse reticulatis; petiolis 4-10 m ngis; xillari terminalique 5-6-plo ramosa cm longa 4-6 cm lata, axibus Ens w 3-9 mm longis; floribus 5-meris; pedicellis 1-1.5 mm pr pp bie lobis suborbicularibus dpi d 0.5 mm longis 0 ique obovato 14-20 mm longo ihid 9 mm lato arcte venoso; seminibus anguste tis. ver ellipsoideis 6.5-7 mm longis 2-3 mm m lat inflorescentia cymosa a Tree 10-45 m tall. Leaf blades ovate or lance- elliptic, obtusely acuminate at apex, narrowly acute to acuminate at the shortly decurrent base, 5.5- 10 cm long, 2.5-3.5(-4.5) cm wide; lateral nerves (5-)8-12 each side, ascending at an angle of 45— 60°, anastomosing 3-5 mm from margin, conspic- uous and impressed or subelevated above, less con- spicuous below; tertiary venation manifest both sides, grossly reticulate, more conspicuous below; midrib impressed above with a slender narrow rib- bonlike border on each side. Inflorescence cymose, axillary and terminal, 5-6 times branched, 2.5 cm long, 4-6 cm wide, the primary axes 3-9 mm long. Flowers 5-merous, the pedicels 1-1.5 mm long. Bracts subtending opposite pairs of the di- chotomous axes. Calyx papillose-pustulose without, the lobes suborbicular, rounded, long 0.7 mm wide, margins slightly uneven but eciliate, the apex ending slightly higher than the lateral margins and crowned by 3 darker minute appendages. Pet- als broadly ovate-oblong, rounded at apex, 1.1 mm long, 0.5-0.8 mm wide. Anthers orbicular, 0.2- 0.3 mm x 0.2-0.3 mm; filaments 0.3-0.4 mm long, slightly wider basally. Disk annular-cupuli- orm, lance-deltoid, acute, 0.5-1 mm long, the margins with dark squamellate appendages. Fruit obliquely obovate, broadly rounded at apex, 14- 20 mm long, mm wide toward the distal end, about 3 times longer than broad, strongly venose, the main nerves at first ao Seeds narrowly ellipsoid, 6.5-7 mm long, 2 .5 mm —3 mm wide. Paratype. VENEZUELA. BOLÍVAR: lower portion of dd iras ie ae wage to Rio epic below San- ta Teresita de Kavanayén, 915-1,065 m, 25 Nov. 1944, Steyermark 60580 (F. VEN). T. F. DELTA AMACURO: bos- que pluvial, E of Rio Grande, ENE of El Palmar, near limits of Bolivar, 13 Jan. 1965, Marcano Berti 562 (MER, MO, NY, VEN Common names. guatacare montanero. Ata-caramá-yek (Arekuna); This taxon differs from Zinowiewia australis Lundell of the northern Coastal Range of Venezuela in the more numerously flowered, larger inflores- cence with longer primary and secondary axes, and especially in the shorter seed body and shorter fruit with the wing narrower in proportion to the width. Also the leaf blades are more shortly and less abruptly decurrent. RHAMNACEAE GOUANIA Gouania wurdackii Steyermark, sp. nov. TYPE: Venezuela. Bolivar: Dist. Cedeno: Cerro San Borja, middle Orinoco River, 100-300 m, 12 Dec. 1955, J. J. Wurdack & J. V. Monachino 39810 (holotype, MO; isotypes, NY, VEN). Frutex scandens; foliorum laminis discoloribus anguste ovatis vel oblongo-ovatis apice acutis vel subacutis basi cordatis vel subcordatis 4-7 cm longis 2.5-4 cm latis supra tenuiter TADA sulcinerviis nion i dan pu- sime arcte cine ramineo- LUE te arcte serratis, pilin e 15-20 late deltoideis subacutis apice in glan subtus manifeste elevatis, nervis lateralibus utroque 6-8; petiolis 5— 10 mm longis; floribus extus dense cinereo- tomentosis; calycis lobis ovatis subacutis 1.2 mm longis extus albido- iris intus glabris; disco dense minuteque pubescenti, circa stylum elevato pilis non munito, I triangulari- ae ad apicem atten su natis subobtusis 0.5 mm longis; capsulis bestie bon alatis 9 9-10 : x 9-10 mm, omnino dense pilosulis alis sublunatis longioribus quam latioribus 9- 10 mm longis .5-4 Woody vine or climbing shrub with striate, densely fulvous-tomentellose branches. Leaf blades discolored (dry), olive green above, cinereous or stramineous below, narrowly ovate or oblong-ovate, acute to subacute at apex, cordate or subcordate at base, 4-7 cm long, 2.5-4 cm wide, finely ru- gulose and sulcate-nerved above, minutely ap- pressed-pubescent in the sulcations, below densely tomentellose with cinereous or stramineous hairs completely covering the surface and nerves, the hairs on the lower surface densely intertwining and matted, the margins closely and rather uniformly serrate with 15-20 broadly deltoid, subacute teeth 2—4 mm wide and averaging 4 per cm, terminating in a small, brown or maroon-brick-colored, thick- ened callosity, glabrous on the lower side and with axillary hairs on the upper leaf surface; lateral nerves 6-8 each side, sulcate above; midrib and lateral and tertiary nerves conspicuously elevated, the tertiary ones conspicuously transverse between the finer elevated veinlets. Petioles 5- 10 mm long, densely pubescent with stramineous or pale yellow- brown hairs. Inflorescence interruptedly spiciform, the lower inflorescences simple and axillary, the upper becoming paniculately branched; rachis 1066 Annals of the Missouri Botanical Garden deeply stramineous or pale brown pubescent. Flow- ers sessile or nearly so. Calyx lobes + densely pilose without, with pale hairs, ovate, subacute, 1.2 mm long, 1.2-1.3 mm wide at the base, glabrous within; hypanthium obconic, 1.2 mm long, densely pubescent. Petals cucullate, about length of calyx lobes. Disk completely and densely short pubes- cent, elevated into an annulus near the style, where densely setose; disk lobes triangular-lanceolate, at- tenuate to a subobtuse subemarginate apex ca.0.5 mm long, about 1⁄4 the width of the calyx lobe. Capsules fulvous brown or gray-brown, sessile or up to 2 mm pedicellate, suborbicular, about as broad as long, 9-10 x 9-10 mm, + pilosulous throughout; axis of fruit 7-9 mm long, the wings sublunate, longer than broad, 9-10 mm long, 2.5- i body densely mm wide, about as wide as the central atypes. VENEZUELA. BOLÍVAR: Dist. Cedeño: east slopes of Cerro Pijiguao, N end of Serranía Suapure, above Pijiguao, ca. 70 km from mouth of Rio Suapure, 110- 520 m, 19 Jan. 1956, Wurdack & Monachino 41310 (MO, NY, VEN). T. F. AMAZONAs: Dpto. Atures, 18 km al SE de Puerto Ayacucho, laja cerca de la Piedra con Petroglifos, + 2 km al N del pueblo de Pintado, 5%32'N, 67°32'W, 1,000 m, 8 Dec. 1977, Huber 1370 (NY, VEN) This taxon is distinguished vegetatively from the common and widely distributed Gouania mollis H.B.K. by the narrowly lance-ovate, acute leaves with conspicuous discolored nerves on the stra- mineous lower surface and by the finely rugulose sulcate nerves of the upper surface. Also, the ter- tiary venation abaxially is conspicuously elevated and prominent transversely and the marginal teeth are closer together, shorter, narrower, and ter- minate more abruptly in smaller callosities. The taxon is restricted to the granitic sector of the Distrito Cedeno of northwestern Estado Bolivar and adjacent northwestern Territorio Federal Amazo- nas, a region of high endemism. RHAMNUS Rhamnus longipes Steyermark, sp. nov. TYPE: Venezuela. T. F. Amazonas: Serrania Part, cumbre, SSE to edge of descent, to tributary of Cano Asisa, mostly rocky sabanita open areas, 2,000 m, 10 Feb. 1951, R. S. Cowan & John J. Wurdack 31388 (holotype, NY). Frutex 2-metralis, a oense sparsim pilosulis; foliorum laminis ovatis vel oblongo-ova dy acutis vel i m longis pando- crenulatis, costa media subtus sparsim pilosula, aliter ubique glabris; nervis la- ves utroque latere 5-8; petiolis 4-8 mm longis paullo dien. calyce fructifero sparsim pilosulo pilis adpressis nito; infructescentia 1-2-fructifera; pedunculo fructi- " ro maturo 10-15 mm longo puc pilosulo; pedicellis fructiferis maturis 10-15 mm longis sparsim pM fructibus subglobosis 5-7 mm longis 6-7 mm latis Shrub 2 m tall. Young stems sparsely pilosulous with pale hairs. Leaf blades ovate or oblong-ovate, acute to subacuminate at apex, obtuse at base, the larger ones 5-6.5 cm long, 2.5-3.2 cm wide, the margins subrevolute, obscurely repand-crenulate with ca. 3 depressed crenulations per cm, the leaf surface glabrous both sides; midrib above sulcate, glabrous, below sparsely pilosulous with pale hairs. Principal lateral nerves 5-8 each side, inconspic- uous above, slightly elevated below; tertiary ve- nation slightly sulcate above, more conspicuously elow. Petiole 4-8 mm long, slightly pilosulous with pale hairs. Fruiting calyx sparsely pilosulous, the lobes triangular-lanceolate, subacute, 1.3 mm long. Infructescence with 1-2 fruits; mature fruit- ing peduncle and pedicels 10-15 mm long, pilo- sulous with pale hairs. Fruit subglobose, 5-7 mm long, 6-7 mm broad, glabrous. This species is characterized by the relatively elongated, sparsely pilosulous fruiting peduncle with 1-2 elongate, sparsely pilosulous pedicels 10-15 mm long; subrevolute, shallowly repand-crenulate leaf margins with an average 3 crenulations per cm; and the slightly elevated tertiary venation of the lower leaf surface. Rhamnus sipapoensis Steyermark, sp. nov. TYPE: Venezuela Cerro Si- papo (Paraque), rim head of South Basin, oc- casional in woodland, rugged terrain, 1,970 m, 26-28 Jan. 1949, Bassett Maguire & Louis Politi 28656 (holotype, NY; isotype, MO) mazonas: A 3-metralis, ramulis junioribus dense brunneo- tomentosis; stipulis haud persistentibus; foliorum laminis oblongis elliptico-oblongis vel ovato-oblongis apice obtuse acutis vel abrupte breviter acutis mucronatis basi rotun- datis vel obtusis 2.5-7.5 cm lon cm latis, supra glabris subtus pallido-s stramineis dense velutinis, marginibus revolutis integerrimis vel fer ivi odia, costa media nervisque supra sulcatis subtus dense to tellis; nervis lateralibus utroque latere 7-11; etiolis Pi 10 mm longis dense tomentellis; inforescentia epedun- culata vel pedunculo usque ad 6 mm longo; pedicellis sub i ue ad 11 mm e 3.5-4 mm longo extus dense brunneo-tomentelloso; ovario dense pu- bescenti. gis 1.5-3.8 Tree 3 m tall, the young branches densely dull brown tomentellous. Stipules not persistent. Leaf blades oblong, elliptic-oblong, or ovate-oblong, mu- Volume 75, Number 3 1988 Steyermark 1067 Venezuelan Guayana Flora cronate at the obtusely acute to abruptly short acute apex, obtuse to rounded at base, 2.5-7.5 cm long, 1.5-3.8 cm wide, glabrous above, densely pale buff velutinous below with the tomentum con- cealing the leaf surface; margins revolute, entire or nearly so; midrib and lateral nerves sulcate above, below densely brown tomentellous. Lateral nerves 7-11 each side, spreading-ascending at an angle of 25-45°, ending at margins. Petioles 4- 10 mm long, densely tomentellous. Inflorescence epedunculate or with a peduncle up to 6 mm long. Pedicels 6-9 mm in anthesis, up to 11 mm long after anthesis, densely brown pubescent. Calyx campanulate, 3.5-4 mm long, 3.2-3.5 mm broad, densely brown tomentellous; lobes 5, lanceolate- ovate, acute, 1.5-2.1 mm long, 1-1.5 mm wide. Petals unguiculate, bilobate, 1.5 mm long, the lam- ina suborbicular-obovate, 0.8 mm long, 1.3 mm wide. Anthers oblong-subquadrangular, 0.5 mm long; filaments 1 mm long, glabrous. Ovary subglo- bose, 1.5 mm long, densely pubescent; style 0.4 mm long, glabrous. This species has a close affinity with Rhamnus marahuacensis Steyerm. & Mag., from which it differs in the mainly epedunculate inflorescence, or the peduncle may be developed to 6 mm in length. It differs further in the densely tomentose calyx, pedicels, and stems; the mainly entire leaf margins; the obtusely acute to abruptly short-acute, mucronate leaf blades; and the more densely to- mentose lower leaf surface. SAPINDACEAE MATAYBA Matayba affinis Steyermark, sp. nov. TYPE: Ven- ezuela. T. F. Amazonas: Dept. Atabapo. Cu- curital de Caname, southern bank of the mid- dle part of Cano Caname, 3?40'N, 67?22'W, 100 m, 30 Apr.-1 May 1979, Gerrit Da- vidse, Otto Huber & Stephen Tillett 17011 (holotype, MO; isotype, VEN). Figure 4. Arbor 4- metralis; foliis 3 dm longis, foliolis 3-5 oblongo- rimis; calycis lo vel subacutis 1-1.2 m latis; antheris pilosis; disco sparsim piloso; stylo xen 1.2 mm longo; ovario 3-loculari. Tree 4 m, the branches strongly lenticellate. Leaves, including the petiole, 3 dm long. Petioles 3-4 cm long, microscopically puberulent. Leaflets 3-5, alternate, oblong-obovate to elliptic-oblong, rounded to obtusely acute at apex, acutely nar- rowed at the base, the upper larger ones 14-15 cm long, 6-7 cm wide, the lower smaller ones 7- 11 em long, 3-5 cm wide, glabrous except for the microscopic rufous glands especially abundant along the midrib and lateral nerves, entire; lateral nerves 6-8 on each side, inconspicuous, ascending at an angle of 45—-50*; tertiary venation finely and ob- scurely reticulate both sides; petiolule 5-6 mm long, glabrous. Panicle terminal, 15-18 cm long; rachis moderately puberulent with 2-4 unbranched axes 2.5-7 cm long, bearing numerous sessile clus- ters of pedicellate flowers. Pedicels 1.5-3 mm long, densely puberulous. Calyx moderately appressed- puberulous without, the margins cut 25 the length; calyx lobes broadly suborbicular-deltoid, rounded or subacute at apex, 1-1.2 mm long, 1.2-1.5 mm wide at the base. Petals narrowly oblong-ovate, obtuse, 1.2 mm long, 0.4 mm wide; petaliferous scales rhomboid-oblong, rounded, slightly exceed- ing the petal, 1.3-1.5 mm long, 0.5 mm wide, densely villous as is the petal. Anthers suborbicular, 5 mm, densely pilose; filaments 2 mm long, pilose except in uppermost portion. Disk sparsely pilose. Style 1.2 mm long, elongated, sparsely strigillose. Ovary ovoid-subglobose, tri- onous, 2 mm long, 1.5 mm wide, 3-celled, mod- erately strigillose. This species is allied to Matayba macrostylis Radlk., but the anthers are pilose, the disk is sparse- ly pilose, the calyx lobes are rounded or subacute, the ovary is 3- instead of 2-celled, the petals are shorter, and the leaflets are rounded or obtusely acute, and only 3-5. Matayba longipes Radlk., Sitzungsber. Bayer. Akad. IX. 536. n. 479. p. 626. n. 7. 1879; in Engler, Das Plfanzenreich, Heft 98e (IV. 165). Sapindaceae. 1085. 1933. Matayba tovarensis Radlk., Sitzungsber. Bayer. Akad. IX. 536. n. 494. p. 626. n. 8. 1879. Radlkofer (1933) distinguished Matayba to- varensis Radlk. from M. longipes Radlk., both described from the area of Colonia Tovar, Vene- zuela, on vegetative characters only, such as sup- posed differences in texture (membranous in M. longipes contrasted with coriaceous in M. tovar- ensis), leaf shape (oblong- or subacute-lanceolate in M. longipes vs. lanceolate in M. tovarensis), and degree of narrowing of the base of the leaflet blade into the petiole (abruptly attenuate in M. longipes, but gradually narrowed in M. tovaren- sis). Study of type material and of additional col- 1068 Annals of the Missouri Botanical Garden FIGURE 4. C. Petal, dorsal view.— D. Petal, ventral view. ——E. Matayba n — A. Habit of flowering branch.— B. Detail of small portion eina rachis.— iagrammatic transverse section dd y.—F. Pistil and calyx.—G. Basal un of leaf blade, abaxial view, with portion of petiole. —H. Sta lections from near the type locality and elsewhere in the coastal mountains of northern Venezuela indicates that these differences break down and that the two taxa must be united. Even on the type specimen of M. longipes (Fendler 1748), a few leaflets show their bases gradually tapering into the petiole as in M. tovarensis. The leaflet shape varies within a given population, and there appears to be no difference in the texture of the two taxa. | am uniting them under M. longipes. The fruit of Matayba longipes has a stipe de- scribed as being 1.8-2 cm long (Radlkofer, 1933). However, an isotype specimen at MO has stipes mostly only 1.5 cm long. Fruiting specimens collected in the Venezuelan Guayana from Cerro Yutaje, Serrani Parü, and Volume 75, Number 3 1988 Steyermark 1069 Venezuelan Guayana Flora the Brazilian side of the Cerro de La Neblina in Territorio Federal Amazonas agree in all essential respects with Matayba longipes of the Venezuelan coastal mountains, except for the constantly short- er fruiting stipe, only 5-9 mm long, whereas in typical M. longipes the length varies from 9-18 (7-20) mm. Other minor differences are in the some- times smaller number of leaflets, relatively larger length and width of the leaflets, and in their slightly longer, more slender leaf apex. This difference in foliage is somewhat more marked in the specimens from Serrania Yutaje and Neblina than those from the vicinity of Cerro Yutaje. In view of the differences in stipe length and the geographically isolated distribution on the sand- stone mountains of the Venezuelan Guayana, to- gether with some tendency in foliar divergence, I consider that the Guayana population shows a suf- ficient degree of separation as to warrant a sub- specific status. LITERATURE CITED RADLKOFER, L. . Sapindaceae. /n: A. Engler, Das Pflanzenreich IV. 165: 1085. (Heft 98e). Matayba longipes Radlk. subsp. tepuiensis Steyerm., subsp. nov. TYPE: Venezuela. T. F Amazonas: Serrania Parü, Cano Asisi, Top Camp to Cano Camp, talus slopes, 1,400 m, 13 Feb 1951, R. S. Cowan & J. J. Wurdack 31450 (holotype, MO, as 2997923; isotype, NY) . longipes praesertim fructus stipite 5-9 mm longo recedit; foliolis quattuor quinque vel sex 9-21 cm longis -8 cm latis; infructescentiis 2-12 cm longis. Tree 8-20 m tall. Leaves 4—6-foliolate; leaflets elliptic-ovate to oblong-lanceolate, obtusely to slen- derly acutely acuminate, 9-17(-21) cm long, 3- 6.5(-8) cei wide, the acumen 1.5-2.5 cm long, abruptly to gradually acutely narrowed to the base; main lateral nerves 8-10(-12) each side, elevated below; tertiary venation prominently reticulate. Petiole 4-8 cm long; petiolule 2-10 mm long. Disk tomentose. Infructescence 2-12 cm long, the pe- duncle and rachis mostly densely strigose-pubes- cent. Fruiting pedicel 4—7 mm long, sparsely strig- illose. Stipe of fruit 5-9 mm long, sparsely strigillose. Fruiting capsule lobed, horizontally divaricate, 7— 12 mm high, 1.5-2.2 cm broad, glabrous or very sparsely puberulent without, the valves densely tomentose within. Style persistent in fruit, 2-3 mm long. Paratypes. VENEZUELA. T. AMAZONAS: Dept Atures, valley of Rio Coro-Coro, pes of Serrania de Yutaje, east base of forested mountain 5 km W of river, 5?4]'N, 66%9'30"W, 1,100 Liesner 3405. BRAZIL: da Neblina, between Palmito and Tatú Camp, 400-600 m, 19 Dec. 1965, Silva € Brazão 6069 (MO, NY). Matayba oligandra Sandwith var. oligandra Matayba oligandra Sandw., Kew Bull. 1935: 123. 1935. Trichilia ptariana Steyerm., Fieldiana, Bot. 28(2): 278. Matayba oligandra var. e ai ), Bol. Soc. en i. Nat. 26 M 19 Matayba pho Steyer ' Bal ‘Soe Venez. Ci. Nat. 33 (132/133): 347, fig. 15. 1976. Matayba oligandra Sandw. var. occidentalis Steyermark, var. nov. TYPE: Venezuela. T. Amazonas: West Mountain Cano Grande, Rio Cuao, Rio Orinoco, 125 m, 17 Jan. 1949, Bassett Maguire & Louis Politi 28399 (ho- lotype, MO; isotype, NY). A M. oligandra foliolis pou elliptico-ovatisve 2.5- 4.5 cm latis (2.8-)3-3. ongioribus quam latioribus, acumine longitudine Y,-Y, dA iras partes aequanti recedit. Leaves 10-12-foliolate. Leaflets ovate or ellip- tic-ovate, abruptly slenderly and obtusely acumin- ate at apex, acute at base, 5.5-10.5 cm long, 2.5- 4.5 cm wide, (2.8-)3-3.6 times longer than broad, the acumen 10-20 mm long, YY, the length of the leaf blade; petiolules 5-10 mm long. Petioles 3-)5.5-8 cm long. Infructescence 9-17 cm long. Fruit 1.5-1.8 cm long, 1.5-2 cm diam., the valves glabrous without and within. — Paratypes. VENEZUELA. T. F. AMAZONAS: Cerro Si- 25-150 m, 12 Apr. 1970, Stey- ermark & Bunting 102560 (VEN). Matayba oligandra Sandw. var. oligandra, de- scribed from Guyana (Sandwith, 1935: 123), has the leaflets principally oblong or oblong-lanceolate, -3.8 cm wide, chiefly 2.2-3 times longer than broad, and with an acumen /—'/ the length of the leaf blade. It is confined to Estado Bolivar in the eastern half of the Venezuelan Guayana. Var. oc- cidentalis, on the other hand, is restricted to the western part of the Venezuelan Guayana in Ter- ritorio Federal Amazonas. Matayba | jauaensis Steyerm. and M. oligandra var. ptariana Stey- erm. appear to represent merely minor variations of leaflet size and cannot be maintained apart from var. oligandra The type da of Matayba oligandra and M. chimantensis Steyerm., the latter possibly syn- onymous with M. oligandra, have only 4-6 sta- 1070 Annals of the Missouri Botanical Garden mens, whereas M. jauaensis Steyerm. and M. oli- gandra var. ptariana have 8 stamens. However, no additional characters have been found to sep- arate those collections having 8 stamens from typ- ical M. oligandra. LITERATURE CITED SaNDWITH, N. Y. 1935. Contributions to the Flora of Tropical America: XXIII. Bull. Misc. Inform. 1935: 1-132 Matayba yutajensis Steyermark, sp. nov. TYPE: Venezuela. T. F. Amazonas: Serrania Yutaje, Northwest Ridge, 1,400 m, 11 Feb. 1953, Bassett & C. K. Maguire 35143 (holotype, MO; isotypes, NY, VEN). Figure 5. Arbor 3-5-metralis; foliis 2-4-foliolatis; foliolis alternis, oblongis vel oblanceolato-oblongis apice rotundatis retu- sisque 10-18 cm longis 2.5-8 cm latis; nervis lateralibus principalibus utrinque latere 8-15; venulis tertiariis utrin- que conspicue elevatis reticulatisque; petiolis 1-1.7 cm longis; calyce 3.5-4 mm lato extus glabro; filamentis ubique dense pilosis; fructu 2 cm longo 2.3 cm lato extus glabro; endocarpio intus glabro. Tree 3-5 m. Leaves 2-4-foliolate. Leaflets al- ternate, oblong to oblanceolate-oblong, rounded and retuse at apex, subacute to acute at base, 10-18 cm long, 2.5-8 cm wide, glabrous throughout; principal lateral nerves rather inconspicuous, 8- 15 each side, spreading at an angle of 10-20*; tertiary venation prominently elevated and retic- ulate on both sides. Petioles 1-1.7 cm long, gla- brous; petiolules 3-8 mm long, glabrous. Inflores- cence terminal or subterminal, including the peduncle, 15-18 cm high, 5 cm wide, ael branched, the individual axes ascending, 2-5 c long, sparsely strigillose. Rachis sparsely les to glabrescent. Flowers pedicellate. Pedicels 1-1.5 mm long, sparsely strigillose. Bract subtending ped- icels and axes lanceolate, acute, 0.7-1 mm long, sparsely strigillose. Calyx 3.5 mm long, 3.5-4 mm wide, glabrous without or sparsely strigillose near base; et suborbicular, rounded with narrowed to subacute apex, 1 mm high, 2 mm wide. Petal seed flabelliform, 1.5 mm long, 1.6-1.7 mm wide, glabrous without, pilose within; petalif- erous scales obovate-oblong, rounded-truncate at summit, | mm long, 0.5 mm wide, with long brown- ish hairs, the margins (especially around the apex) laciniate-pectinate, pubescent both sides. Anthers suborbicular, basally bilobed, 0.7 x 0.7 mm; fil- aments 3 mm long, 0.3 mm wide at base, densely pilose throughout. Fruit 3-lobed, shortly stipitate, the stipe 3 mm long, 4 mm wide; fruit body 2 cm long, 2.3 cm wide, glabrous without; endocarp gla- brous within. Paratypes. | VENEZUELA. T. F. AMAZONAS: Serranía Yutaje, Cerro Yutaje, left hand fork of Cano Yutaje, 1,300-1,400 m, 15 Feb. 1953, B. & C. K. Maguire 35242 (MO, NY); 1-2 km E of Rio Coro-Coro, W of Holst 21484 (MO, NY, VEN); valley of Rio Coro-Coro, W of Serrania Yutaje, W of valley, 5?42'30"N, 66°10'W, 1,300 m, Holst & Liesner 3368 (MO, NY, VEN); Dept. Atabapo, Cerro Marahuaca, above branch of Cano Negro, aka, 3*38'N, 65°28'W, 1,225 m, 17-18 Feb. 1985, Liesner 17614 (MO, VEN). This species is related to Matayba atropurpu- rea Radlk., from which it differs in the retuse, rounded apex of the leaflets; fewer leaflets; shorter petioles; filaments densely pilose throughout their length; and the much larger fruit. Further, Ma- tayba yutajensis occupies higher altitudes of mon- tane forest of the sandstone table mountains, whereas M. atropurpurea is a lowland tree TALISIA Talisia amaruayana Steyermark, sp. nov. TYPE: Venezuela. Bolivar: Amaruay-tepui, 5%54'N, 62°15'W, 550-810 m, 26 Apr. 1986, Ron- ald Liesner & Bruce Holst 20394 (holotype, MO; isotype, VEN) Arbor 3-metralis; foliis 12-foliolatis; puting lanceolatis apice acuminatus basi acutis vel subac s 26-38 cm i 5 an sparsim puberulentam glabris; nervis lateralibus een latere 6-25; panicula usque 8.5 dm longa; Beds lobis intus glabris extus puber ulis longiciliatis; petalis 6 mm longis 2.5- mm latis longiciliatis; squamis petaliferis parte dorsali praeter margir nes inferiores dense sericeos glabr ra; staminibus 5; fi : disco hirsutulo. 1 I 1 1 Tree 3 m, with simple unbranched stem. Leaves 1 2-foliolate. Leaflets chartaceous, lanceolate, acu- minate at apex, acute to subacute at base, 26-38 cm long, 4-8 cm wide, glabrous except sparsely puberulent on the midrib below; lateral nerves 16- 25 each side, conspicuously elevated beneath, im- pressed above, anastomosing 4-7 mm from mar- gin; tertiary venation minutely reticulate above, grossly reticulate below with elevated veins. Pet- iolules 7-9 mm long, hirtellous. Panicle terminal, up to 8.5 dm long, the rachis and axes hirtellous. Axes simple or branched, 12-20 cm long. Flowers crowded in several-flowered fascicles, sessile. Calyx Volume 75, Number 3 1988 Steyermark 1071 Venezuelan Guayana Flora URE 5. Fic out, e view.—D. Upper portion of stamen.—E. Sta 3.5 mm long, cut 25 way down; lobes ovate-oblong, rounded at summit, 2 x 2 mm, puberulous without, glabrous within, long-ciliate. Petals lance-oblong, obtuse at summit, 6 mm long, 2.5-2.8 mm wide below the middle, glabrous both sides, long-ciliate. Petaliferous scales lanceolate, obtuse-subtruncate, slightly shorter than the petal, 3-3.5 mm long, 2 mm wide at base, the ventral portion densely se- riceous above the glabrous du sector, the dorsal portion densely sericeous along the lower margins, Matayba yutajensis. — 4. Habit caca bd branch.—B. Caly. U x. "a^ Petal with 2 scales spread en.—F. Petal, dorsal view. glabrous elsewhere. Stamens 5; filaments 1.3 mm long, pilose except in the uppermost part; anthers linear-oblong, 1.5-1.8 mm long, 0.2 mm wide, the appendage broadly lanceolate, acute, 0.3 mm long, the base bilobed. Disk shallowly undulate-lobulate, 2.8 mm diam., hirsutulous except on the outer concavities. Style 3 mm long, strigose. Fruiting edicels 2 mm long, densely pubescent. Fruit ovoid- subglobose, 2 cm long, 1.7-1.8 cm wide, minutely appressed-puberulent. 1072 Annals of the Missouri Botanical Garden Paratype. VENEZUELA. BOLIVAR: Amaruay-tepui, 9*55'N, 62°15'W, 550-800 m, 20 May 1986, Liesner & Holst 20935 (MO, VEN). From the closely related Talisia tiricensis Stey- erm., T. amaruayana differs by having ciliate pet- als and puberulous lower midrib. Talisia caudata Steyermark, sp. nov. TYPE: Ven- ezuela mazonas: Cerro Sipapo, trail from Base Camp, 125 m, 25 Jan. 1949, Bas- sett Maguire & Louis Politi 286 15 (holotype, MO; isotype, NY). Figure 6. rbor 2-metralis; foliis 6 dm longis, 25- foliolatis; Pe alternis lanceolatis apice longicaudatis basi acutis 9.5 cm longis 2. die cm latis, acumine 17-25 mm Tp subtus praeter costam mediam dense hirtellam "DS eglandulosis; nervis lateralibus utroque dulosa; petiolulis 7- 10 mm longis minute dens seque pu- berulentibus pilis diva niculata 14 cm longa datis 2.2-2.4 cm longis 1- adpresso puberulis; pus sub fructu deciduo vel persistenti 2 mm longo dense strigoso. Tree 2 m. Leaves 6 dm long, petiolate. Petiole 17 cm long, minutely and densely puberulent with divaricate hairs. Leaflets mainly alternate, 25, lan- ceolate, long-caudate at apex, asymmetrically acute at base, 9.5-14 cm long, 2.5-3 cm wide, the caudate portion 17-25 mm long, 3 mm wide at base, the upper surface glabrous with impressed or slightly raised, minutely puberulent midrib, the lower surface glabrous except for the densely mi- nutely hirtellous, eglandular midrib. Lateral nerves -17 each side, slightly sulcate above, elevated below, slightly ascending at an angle of 25-30?, strongly anastomosing below 3-5 mm from margin; tertiary venation reticulate below, slightly elevated, inconspicuous and scarcely manifest on upper side. Leaf rachis densely puberulous with divaricate non- glandular hairs. Petiolules 7-10 mm long, minutely and densely puberulent with divaricate hairs. In- fructescence paniculate, 14 cm long, 7 cm wide, the rachis densely appressed-pubescent; axes 8-9, unbranched, the lower and middle ones 2-3 cm long, the upper ones 0.5-1 cm long, densely ap- pressed-pubescent. Fruit obovoid-oblong, rounded- subtruncate at summit, rounded at base, 2.2-2.4 cm long, 1-1.2 cm wide, obtusely trigonous, ap- pressed-puberulous. Style in fruit deciduous or per- sistent, 2 mm long, densely strigose. Calyx lobes in fruit persistent, strigose without, ciliate. This taxon differs from the related Talisia erecta Radlk. in the much longer caudate, lanceolate leaf- lets with longer petiolules, more numerous lateral nerves, and eglandular pubescence without stipitate glands. The leaflets are longer than broad, but less so than in 7. erecta, which has more closely spaced leaflets. Talisia glandulifera Steyermark, sp. nov. TYPE: French Guiana. Saul: onte La Fumée, 3°37'N, 5321 2W, 200-400 m, 1 Oct. 1982, Scott Mori et al. 15027 (holotype, MO; iso- type, NY). Figure 7. Arbor usque 15 m altis; foliis 25-40 cm longis; foliolis 4- 1-jugis plerumque oppositis elliptico- lanceolatis apice A ace ie. basi acutis " us 10-18 cm longis .5-5 cm latis, costa media supra impress Mech elevata minute aridus pilis donis praedita; petiolulis 2- mm longis; calyce 3. m longo extus dense glanduloso- piloso, lobis ovato- -oblongis apice rotundatis; petalis sub- ovato- POR AR apice rotundatis extus glabris intus sparsim prope basim secus mediumque sparsim adpresso-g landu- loso- itso pal oh filamentis glabris; p Tree 15 m tall. Leaves 25-40 cm long, the petiole 6.5—10 cm long. Leaflets 4-7 pairs, mainly opposite, elliptic-lanceolate, long-acuminate at apex, asymmetrically acute at base, the larger ones 10- 18 cm long, 3.5-5 cm wide, the midrib impressed above, elevated below, minutely puberulent wit short, divaricate hairs, otherwise glabrous; lateral nerves 13-18 each side, conspicuous and elevated below, slightly ascending at an angle of 15-250; tertiary venation finely reticulate above, more grossly reticulate below; rachis terete, minutely puberulent with a few glandular hairs; petiolule 2- 5 mm long. Inflorescence subterminal, paniculate, densely flowered, 13-23 cm long, 8-20 cm wide, with 6-12 ascending axes, the lower ones 4-15 cm long, densely hirtellous intermixed with glan- dular hairs. Peduncle 1 cm long or none. Flowers pedicellate or subsessile; pedicels 1.5 mm long, densely hispidulous. Bracts ovate-lanceolate, acute, 0.75-1.5 mm long, densely hirtellous with some glandular hairs. Calyx tance, densely glandular-pilose; calyx lobes ovate- oblong, rounded at summit, 2.5 mm long, 1.5-1.8 mm in upper half, ciliate. Petals subovate-oblong, 5 mm long, cut 25 dis- rounded at summit, 5.5 mm long, 2.2 mm wide in upper half, glabrous without, ciliate in lower half, within sparsely appressed-glandular-puberulent near base and along median line; petaliferous scale about equaling or slightly shorter than petal, 3.5 mm long, 1.5 mm wide, densely barbate-villous for most of length on inner face, the outer face glabrous basally, pubescent in upper half, densely appressed- ciliate. Stamens 8; anthers oblong, obtusely apic- ulate, 1.3-1.4 mm long; filaments 3.8 mm long, Volume 75, Number 3 1988 Steyermark 1073 Venezuelan Guayana Flora FIGURE 6. glabrous. Disk lobulate, 2 mm diam., densely his- pid. Fruit not seen. Paratypes. VENEZUELA. BOLIVAR: El Paraiso LE m Ros , San Félix, 1-10 Jun 77 (MER). FRENCH GUIANA. SAUL: Monte La Fumée P3TN. 53?12'W, 200-400 m, 24 Sep. 1982, Mori a al. 14988 (MO, NY). This species is apparently most closely related to Talisia cupularis Radlk. within the group of species 32-36 of Radlkofer's sect. III “Eutalisia” (sect. Talisia). It is characterized by calyx lobes — Talisia caudata. — A. Habit of fruiting branch. — B. Fruit, lateral view.—C. Fruit, apical view. densely glandular externally, minutely puberulent rachis with a few gland-tipped hairs, glabrous outer surface of the petals, glabrous filaments, densely hirsutulous disk, and petals glandular on the inner surface. From T. cupularis Radlk. it may be dif- ferentiated by the glandular-puberulent inner sur- face of the petals and the strongly glandular pu- bescent outer surface of the calyx lobes. Talisia heterodoxa Steyermark, sp. nov. TYPE: Venezuela. Bolivar: Represa Guri, islands 6- 1074 Annals of the Missouri Botanical Garden 5.5mm E 7. Talisia glandulifera. — 4. Habit of flowering branch.— B. Pubescent scale with interior surface of petal reflexed. —C. Exterior view of petal. —D. Interior view of petal. o apiculate connective. —H. Two of the calyx lobes sh branc 18 km S of dam, 7°38'N, 62°58'W, 220- 240 m, Ronald Liesner & Angel González 11151 (holotype, MO; isotype, VEN). kK ru LN E bor 10 8 oppositis vel suboppositi elliptico-oblongis vel oblanceolato-oblongis apice rotun- datis saepe retusis 9-16 cm longis 3.5-6.7 cm latis gla- —E. Disk. —F. Stamen.— Anther with wing glandular hairs, exterior view.—1. Portion of flowering bris; petiolulis 5-6 mm longis glabris; calycis lobulis su- borbiculari-ovatis dense ciliatis extus puberulis intus adpresso-p ; squamis petaliferis extus in quarta p te superiore pubescentibus prope basim glandulosis; stam- inibus 8, filamentis glabris; disco glabro. yheruh Tree 10 m tall. Leaves 6-8-foliolate. Leaflets opposite, subopposite, or slightly alternate, elliptic- Volume 75, Number 3 1988 Steyermark 1075 Venezuelan Guayana Flora oblong or oblanceolate-oblong, rounded and often retuse at apex, cuneate at base, 9-16 cm long, 3.6-6.7 cm wide, glabrous; lateral nerves slender, slightly ascending at an angle of 25-45°, 12-13 each side, finely prominulous both sides; interme- diate nerves finer; tertiary venation finely reticulate both sides, the reticulations more elevated and larg- er below; rachis of leaf 7-15 cm long, minutely puberulous to glabrous. Petiolule 5-6 mm long, glabrous. Petiole 4-9.5 cm long. Inflorescence ter- minal, 38 cm long with 7-8 widely spreading, sparingly branched axes, the lower ones 10-16 cm long, 2 mm wide, minutely puberulous with short spreading hairs; peduncle 3-4 cm long. Calyx cut % distance, the lobes suborbicular-ovate, rounded at summit, 2.2 mm long; 1.5-2 mm wide, densely ciliate, puberulous without, within ap- pressed-puberulous 34 distance upwards. Petals ovate-oblong, broadly rounded at summit, 4.5 mm long, 2.5 mm wide, glabrous both sides, the margins slightly papillate. Petaliferous scale ligulate, obtuse, 3.5 mm long, 1 mm wide, densely hirsutulous ven- trally, pubescent within dorsally in the upper 1⁄4 and pubescent basally, elsewhere glabrous. Sta- mens 8; anthers 1.5 mm long with a short, tri- angular, obtuse apical appendage, rounded at base; filaments 3 mm long, glabrous. Disk lobulate, gla- brous. Fruit 3-3.5 cm long, 3 cm wide, subgla- rous Paratype. VENEZUELA. BOLÍVAR: Represa Guri, is- lands 6-18 km S of dam, 7?38'N, 62°58'W, 220-240 m, Liesner & Gonzalez 11116 (MO, VEN) This species is closely related to Talisia retusa Cowan but differs in having the interior of the calyx lobe pubescent only in the middle portion rather than densely sericeous throughout and in having the petals glabrous instead of partly or wholly re- trorsely ciliate. Moreover, in T. retusa the petio- ules are mainly shorter, and the leaflets smaller and narrower. Talisia pentantha Steyermark, sp. nov. TYPE: Venezuela. Bolivar: Canaima, W of Avensa Camp, gallery forest, 500 m, 4 Oct. 1974, F. Ehrendorfer 74104-23 (holotype, VEN). Arbor; foliis 10- foliolate; foliolis oppositis, lanceolato- ellipticis apice acuti utis majoribus 13.5-16 cm longis 3.7-4.5 c m longis 3.5 cm latis Mone glabris; nervis lira Bia principal utrinque 8-11; inflorescentia cm longa, peduncu ulo 8 cm longo; floribus pedicellatis; calyce extus spa berula, lobis intus glabris; petalis minute ciliatis aliter glabris; squamis dorsaliter glabris; staminibus 5; antheris ji uns appendiculatis, apice subacuto 1.8 mm longis; disco profunde lobato glabro. Tree. Leaves 10-foliolate. Leaflets opposite, lan- ceolate-elliptic, acute at apex, acute at the slightly asymmetric base, mostly 13.5-16 cm long, 3.7- 4.5 cm wide, the lowest smaller, 8.5-10.5 cm long, 3.5 cm wide, glabrous both sides, ascending at an angle of 20-25”; tertiary venation inconspicuous. Inflorescence 37 cm long, bearing interrupted short groups of flowers borne on short axes 5-15 mm long; peduncle 8 cm long; rachis ridged, with ver- tical lines of spreading pilosity on the angles, 2.5 mm diam. Flowers pedicellate, pedicels 1 mm long, densely pilosulous with spreading hairs. Bracts of inflorescence lanceolate, subacute, 0.75 mm long, puberulous. Calyx 3 mm long, cut % distance; lobes oblong-lanceolate, subacute or subobtuse, 2 mm long, mm glabrous within, imbricate, densely ciliate. Petals ligulate, obtuse, 4.5 mm long, 1.3 mm wide in late bud, minutely ciliate on the margins, otherwise glabrous; petaliferous scale 3 mm long, dorsally glabrous, densely sericeous ventrally. Stamens 5; anthers linear, appendiculate with a subacute lan- ceolate apex 0.2 mm long, 1.8 mm long. Disk deeply 5-lobed, glabrous. wide, sparsely puberulous without, The glabrous disk relates this species to Talisia guianensis Aubl., which, however, has eight sta- mens. Talisia sancarlosiana Steyermark, sp. nov. TYPE: Venezuela. T. F. Amazonas: between San Car- los and Solano, 11-17 Mar. 1970, Luis Mar- cano-Berti & P. Alcedo 119-979 (holotype, MER) , Frutex; foliis 17-foliolatis ee suboppositis vel alter- s, oblongo-lanceolatis apice acuminatis basi subacutis beara s 30-35 cm longs 8.5-9 cm latis utrinque glabris; nervis lateralibus utrinque 7-9; petio iol 3-8 mm longis; inflorescentia 13 cm bu onga 15 cm lata; sisi lobis ovato- oblon ngis late obtusis extus s Ds osis pr. raeditis pests cum ciliatis; squamis pe- taliferis . pun dorsaliter glabris; staminibus 5; filamentis glabris; antheris linearibus 2 mm longis apice obtuse ap- crt ida ame : Shrub. Leaves 17-foliolate. Leaflets subopposite or alternate, oblong-lanceolate, acuminate at apex, subacute at the base, the upper leaflets larger, 30— 35 cm long, 8.5-9 cm wide, the smaller alternate lower ones 19 cm long, 6 cm wide, glabrous both sides; lateral nerves 7-9 each side, elevated below; rachis glabrous; petiolules 3-8 mm long, glabrous; petiole 13-15 cm long, glabrous. Inflorescence paniculate, widely branching, 13 cm long, 15 cm broad, the rachis minutely pubescent in lines. Calyx 1076 Annals of th Missouri im Garden lobes ovate-oblong, broadly obtuse, 2.3 mm long, .9 mm wide, sparsely pubescent without, glabrous within, conspicuously ciliate. Petals lanceolate, broadly obtuse at apex, 5-6 mm long, 1.5-1.8 mm wide below middle, glabrous without, glandular within, the outer ones ciliate. Petaliferous scale ligulate, 2.5 mm long above the glabrous basal 2-mm portion, densely sericeous ventrally, gla- brous dorsally. Stamens 5; filaments 2.5 mm long, glabrous; anthers linear, 2 mm long, the obtusely oblong appendage 0.1 mm long, bilobed at base. isk glabrous. Ovary obovoid, 3 mm long, gla- brous; style 3.5 mm long, mainly glabrous, stig- matic in the apical 1.5 mm From Talisia guianensis Aubl. this species dif- fers in having five stamens, short inflorescence, fewer lateral nerves, and pubescent inflorescence rachis. TOULICIA Toulicia anomala Steyermark, sp. nov. TYPE: Venezuela. Bolivar: Rio Suapure, Middle Ori- noco, along river between Raudal Budare and Raudal Punta Brava (70-80 km from mouth), 110-120 m, 17 Jan. 1956, J. J. Wurdack & J. V. Monachino 41253 (holotype, MO; isotype, NY). Frutex vel arbor 2-3.metra e pinnatis, 6-10-foliolatis, foliolis oppositis is, ramulis glabris; foliis vel al- apice acutis vel acuminatis basi asymmetricis acutis vel obtusis 4-8.5 cm longis 1.5-2.5 cm latis obscure re undulatis glabris; petiolis 1.5- 4 cm longis 1 m tundatis m longis 1.2- L 5 mm peus extus vip is intus ae Eve pubescentibus EAA ciliatis; i irs 4 esquamatis obovatis 0.8-1 mm longis 0.7-1 mm latis 0.5 mm unguiculatis integri vel 2-3- jr marginibus ongis parte basali 1 mm pilosa; disco regulari leviter MS glabro; pistillodio 1 mm longo; pistillo 3.5 mm long 1.5 mm lato, stylis duobus, ovario 2-loculari, ovulo uno in quoque lo- culo. Shrub or tree 2-3 m tall, the branches glabrous. Leaves abruptly pinnate, 6-10-foliolate, the leaf- lets opposite or alternate, subsessile or 1-2 mm petiolulate, chartaceous, lanceolate, acute to acu- minate at apex, acute to obtuse at the asymmetrical base, 4-8.5 cm long, 1.5-2.5 em wide, obscurely repand-undulate, glabrous; lateral nerves 10-12 each side, ascending at an angle of 15-25", ter minating near the margin, there anastomosing; ter- tiary venation finely reticulate and prominulous on both sides; midrib elevated below, impressed or sulcate above, bordered by a slender, ribbonlike lateral extension 0.25 mm wide on each side; lower leaf surface often with foveolate depressions at the junction of the midrib and lateral nerves accom- panied by thickened portions of the lower epider- mis. Rachis glabrous, 1 mm wide. Petioles 1.5-4 mm long, 1 mm wide. Inflorescence paniculate, 5— 7 cm long, 2-3.5 cm wide, with 8-12 short axes 2-17 mm long, the rachis and axes appressed- pubescent; axes 3-8-flowered, sparsely branched. Peduncle 0.7-3 cm long. Sepals 4, unequal, the outer ones smaller, suborbicular, rounded, 1-1.2 mm long, 1.2-1.5 mm wide, glabrous without, within pilose near the base, long-ciliate on margins. Petals 4, esquamate, unguiculate, obovate, 0.8-1 mm long, 0.7-1 mm wide, the unguiculate portion 0.5 mm long, entire, 2-cleft or 2-3-lobate, densely long-fimbrillate with elongate hairs 0.5 mm long. Stamens usually 6-7; filaments hypogynous, 2.5 mm long, pilose in the basal 1 mm; anthers dor- sifixed, suborbicular, 0.5 x 0.5 mm. Disk contig- uous but not adnate to the ovary, regular, shallowly lobulate, glabrous, 2 mm across, 0.1-0.2 mm high. Pistillode ovate-elliptic, 1 mm long, 0.8 mm wide. Pistil 3.5 mm long, 1.5 mm wide; styles 2; ovary 2-locular, with 1 ovule in each cell. This species is anomalous in having 4 sepals, and usually 6-7 stamens. The esquamate petals align it in the genus to section Aphanolepis Radlk. I thank Dr. Aaron Goldberg of the Smithsonian Institution for valuable help in critical observations and for comments. BOMBACACEAE POCHOTA In Steyermark & Stevens (1988), the generic name Pochota was shown to have priority over Bombacopsis, and Rhodognaphalopsis A. Rob- yns was synonymized with Bombacopsis. Thus the following taxa are assigned to Pochota. Pochota amazonica (A. Robyns) Steyerm. & W. D. Stevens, Ann. Missouri Bot. Gard. 75: 396-398. 1988. Bombacopsis amazonica A. Robyns, Bull. Jard. Bot. Etat 1963. Ai i A. Robyns, Mem. N.Y. Bot. Gard. 17(1): 19 967. Bombacopsis amazonica and B. wurdackii in- tergrade and cannot be maintained as two distinct taxa. In his original description of B. amazonica, Volume 75, Number 3 1988 Steyerma 1077 eh Guayana Flora Robyns (1963: 186) stated that the pedicels vary from 3.5 to 9.5 cm long. This description was based on the US holotype. The isotype (Foldats 3794) at VEN, however, has pedicels only 2.5-3 cm long. Moreover, the calyx was described from the US holotype as 1-1.2 cm wide, whereas the VEN isotype has the calyx only 1.1-1.2 cm wide. This disparity in the mea- surements of the holotype and isotype is manifested again in the measurements of the calyx of B. wur- dackii, which are described from the MO holotype as 0.6 cm long, whereas the VEN isotype measures 0.8-1 cm. Additionally, dimensions of leaflet blades (and their apex shapes), petioles, and petiolules overlap and real differences are not discernable. In both taxa the leaflet blades are thick-coriaceous with a prominent midnerve beneath. Both possess a fine reticulate tertiary venation on the lower surface of the leaflets. One of the principal characters em- ployed by Robyns in separating Bombacopsis wur- dackii from B. amazonica was the presence of glands on the receptacle in the former. Removal of the mass of stamens that had previously hidden the receptacle area on the VEN isotype of B. amazonica revealed glandular depressions, thus eliminating one of the key separating characters between the two taxa. Robyns differentiated B. wurdackii further from B. amazonica by the rel- atively smaller staminal tube of the former, which is described as “circa 5-7 mm long." Examination of the VEN isotype reveals a length closer to 8 mm. cm long and 1.6-1.8 LITERATURE CITED ROBYNS, " 1963. Essai de monographie du genre Bom- bax sl. (Bombacaceae). Bull. Jard. Bot. Etat 33: STEYERMARK, J. A. & W. D. Stevens. 1988. Notes on o aie and Bombacopsis (Bombaca- ceae) in the Guayanas. Ann. Missouri Bot. Gard. 75: 98. Pochota ewelii Steyermark, sp. nov. TYPE: Bra- zil. Dept. Amazonas: near Venezuela frontier, camino al Cerro Neblina desde Rio Tucano (afluente del Rio Cauaburi), 1,250 m, 23 Apr. 1964, J. Ewel 135 (holotype, MY). Arbor foliis 5-foliolatis, den coriaceis late de tis apice acutis vel obtuse acutis basi obtusis vel ro d wm majoribus 9.5-12.5 cm A Kork 4. 5-7. cm latis supra sparsim lepidotis basi conspicue petiolatis; petiolulis 9-12 cm longis; pedicellis 3 cm longis lepidotis; rece m inibus ca. 150, 8 cm longis; staminali tubo 13 mm longo basi glabro deinde superne stellato-piloso; stylo 10 cm longo glabro; ovario lepidoto Tree with glabrous branches. Leaves 5-foliolate, the leaflets thick coriaceous, broadly elliptic-ovate, acute to obtusely acute at apex, obtuse to rounded at base, the larger 9.5-12.5 cm long, 4.5-7 cm wide, the upper surface sparsely lepidote except the more densely lepidote midrib, the lower surface and midrib strongly lepidote; primary lateral nerves on each side 12-14, impressed below, inconspic- uously above, anastomosing 5-8 mm from margin; tertiary venation reticulate and manifestly im- pressed or subelevated below, inconspicuous above; midrib prominently elevated below, slightly elevat- ed above. Petiole 3.6 cm long, 3 mm thick; pet- iolules 9-12 mm long. Flowers solitary, axillary. Pedicels 3 cm long, 3 mm wide, moderately lepi- dote. Receptacle conspicuously glandular. Calyx 10 mm long, 20 mm wide at subtruncate summit, moderately lepidote without, densely sericeous within. Petals linear, acute, 1 2—12.5 cm long, 1.8- 1.9 cm wide, minutely stellate-squamose without, more densely stellate-puberulent within with slight- ly longer trichomes. Stamens numerous, ca. 150, 8 cm long, with 10 phalanges 1 cm long; staminal tube 13 mm long, glabrous below, moderately stel- late-pilose above. Ovary lepidote; style 10 cm long, glabrous. This taxon differs from Pochota gracilis (Rob- yns) Steyerm. & W. D. Stevens in the pubescent upper part of the staminal tube, broader and short- er calyx, and the thick-coriaceous, longer and broader, conspicuously lepidote leaves. The leaflets resemble in their size and thickness those of P. amazonica (Robyns) Steyerm. & W. D. Stevens, which, however, has narrower leaflets less rounded at the base, and shorter petals. The new species is dedicated to Dr. John Ewel, one of the early ex- plorers of Cerro de La Neblina, who reached the summit from the Brazilian side. Pochota fuscolepidota Steyermark, sp. nov. TYPE: Venezuela. T. F. Amazonas: Dept. Ata- bapo, Cerro Marahuaca, forested slopes 1-2 km N of Sima Camp, 3?40'N, 65°31'W, 1,100 m, 8-9 Mar. 1985, Ronald Liesner 18452 (holotype, MO; isotype, VEN) r 5-8-metralis; foliolis 3-5 oblongo-ovatis vel el- buc pen apice acétis basi. obtusis vel subacutis tis 16- m longis 7-10 fusco-ferrugineo-lepidotis, pus subrevolutis; nervis lateralibus principalibus "pa latere 12-15 subtus ma cm longis; petiolulis 1- ifeste elevatis; petiolis 5.5-9.5 1078 Annals of the Missouri Botanical Garden 2.5 cm longis; fructu pedicello fructigero 5 cm longo 5 mm lato densissime fusco-lepidoto; calyce fusco- Lepido Pa obovoideo apice rotundato densissime fusco-fer rugineo lepidoto 6 cm longo apice 3 cm lato basi 0.7 cm lato. Tree 5-8 m tall, the young stems densely fer- rugineous-fuscous lepidote. Leaflets 3-5, oblong- ovate or elliptic-oblong, acute at apex, obtuse or subacute at base, 16-22 cm long, 7-10 cm wide, glabrous above, densely dark brown-ferruginous low, glabrous above with impressed nerves, densely and thickly covered below by dark brown, ferruginous lepidote scales, the principal lateral nerves 12-15 each side, prominently elevated, anastomosing 7-15 mm from the margin; margins subrevolute; midrib subsulcate above, prominently elevated below. Petioles 5.5-9.5 cm long; petio- lules 1-2.5 cm long, densely fuscous lepidote. Flowers not seen. Fruiting pedicels 5 cm long, 5 mm thick, densely fuscous lepidote. Calyx tube in fruit 10 mm long, 17 mm wide, subtruncate at apex, dark lepidote without, densely buff sericeous within. Fruit obovoid, rounded above, 6 cm long, 3 cm wide at summit, 0.7 cm wide at base. Paratypes. VENEZUELA. T. F. AMAZONAS: Dept. Ata- bapo, Cerro Marahuaca, Sima Camp, S-central portion of forested slopes along E branch of Cano Negro, 3?43'N, 65?31'W, 1,140 m, 21-24 Feb. 1985, Steyermark & Holst 130515 (MO, VEN); 9.2 km NE of San Carlos on Solano road, 1°56’N, 67?3'W, 119 m, Clark 6920 (MO, NY, VEN) This species is well characterized by the large leaflets that are densely dark brown-ferruginous lepidote beneath, the prominently elevated lateral nerves on the lower leaf surface, and the densely dark brown lepidote fruits. Pochota gracilis (Robyns) Steyerm. & W. D. Stevens This species was originally described (Robyns, 1967) from specimens collected on the Rio Paci- moni of the Territorio Federal Amazonas of Ven- ezuela at an altitude of 100-140 meters. Subse- quent collections manifest variations in the apex and shape of the leaflets, and in the abundance and distribution of scales on the lower leaflet sur- face. Specimens collected from the eastern portion of the range in Estado Bolivar, Venezuela, have ditionally, the scales on the lower surface of the leaflets are dark and close together. Specimens having these characteristics all come from localities associated with the sandstone table mountains at altitudes of 400-1,000 meters. In contrast, the collections originating from the Territorio Federal mazonas all are found at low elevations of 95- 140 meters in usually white sand savannas bor- dering streams of the lowland areas. The leaflets of these lowland populations are narrowly elliptic- oblong or narrowly oblong with broadly obtuse to rounded, rarely emarginate apex, and the scales on the lower surface are more scattered and less conspicuous than those of the Estado Bolivar pop- ulations. Since these differences can be segregated in eastern and western geographically separated populations, they may be considered to represent two subspecies as follows. Pochota gracilis subsp. gracilis Rhodognaphalopsis maguirei A. Robyns, Mem. N.Y. Bot. Gard. 17: 200. 1967. Leaflets narrowly elliptic-oblong to narrowly ob- long, broadly obtuse to rounded at apex, this rarely emarginate; scales on the lower surface of the leaflets small, scattered, rather noticeably sepa- rated Distribution. T. F. Amazonas, Venezuela, in usually white sand savannas bordering streams at altitudes of 95-140 meters. Derimens ev d. VENEZUELA. T. F. AMAZONAS: Rio Pacimoni, 12 km below mouth of Río Yatua, 100-140 m, Maguire, Wurdack & C. K. Maguire 41653 (holo- pe, MO; isotype, NY); Caño Caname del medio Río hona sabanita, 3?40'N, 67°27'W, 100 m, Huber, Tillett = Davidse 3747 (VEN) nu Rio Ventuari, 10 km al N why 66°16! i 110 m, Huber 6116 (MO); sabana, 1 km al del caserío de Ont n ° o Marueta E ig opposite Cucurital de Canam 9 , Davidse, Huber & Tillett 17035 (MO, VEN): sae nien Rio Sipapo, Maguire & Politi 27849 (NY). Pochota gracilis subsp. bolivarensis Steyer- mark, subsp. nov. TYPE: Venezuela. Bolivar: wooded knoll in savanna, Rio Kanarakuni, southern base of Cerro Sarisarinama, Meseta de Jaua, 400 m, 17-29 Mar. 1967, Julian Steyermark 98206 (holotype, VEN; isotype, NY). Local name. Wanabana (Maquiritare). A ssp. gracilis foliolis anguste obovatis vel elliptico- obovatis apice subtruncato-rotundatis emarginatis, subtus squamis lepidotis conspicuis fuscatis confertis recedit. Volume 75, Number 3 1988 Steyermark 1079 Venezuelan Guayana Flora Leaflets narrowly obovate to elliptic-obovate, subtruncate-rounded at the emarginate apex; scales on the lower surface conspicuous, darker, and clos- er together. arope. VENEZUELA. BOLÍVAR: Cerro Guaiquini- ma, mit, riverine forest, Salto de Rio Szczerbanari (Rio PUES 5?44'N, 63?41'8"W, central prit 750 m, Steyermark, G. & E. Dunsterville 113236 (VEN); Auyan-tepui, 1,100 m, Tate 1155 (NY, VEN); upper Caura River, Cerro Marajanu, 550 m, Cardona 2951 (VEN). Rhodognaphalopsis maguirei A. Robyns was JPR described as having a calyx 15 mm long mm wide, while R. gracilis was described H a calyx 12 mm long and 5 mm wide. Yet on the NY isotype of R. gracilis the older calyces attain a width of 6-6.5 mm; those in bud are 4- 5 mm wide. In the holotypes of both R. maguirei and R. gracilis the staminal tubes are glabrous. Although the style in R. maguirei is described as glabrous and that of R. gracilis as sparsely stellate- puberulous at the base, it is doubtful if this last character can be used to separate the two taxa, since in all other respects they are alike. The type collection of R. maguirei from the lower Rio Sipapo occurs within the general range of Pochota gracilis subsp. gracilis. LITERATURE CITED RoBvNs, A. 1967. Pochota gracilis. In: B. Maguire € collaborators, The Botany of the Guayana High- land — Part VII. Mem. N.Y. Bot. Gard. 17: 198 Pochota liesneri Steyermark, sp. nov. TYPE: Venezuela. T. F. Amazonas: Dept. Atabapo, Salto Yureba, Cerro Yureba, lower Ventuari, 4°3'N, 66?1'W, 350 m, 14 Mar. 1985, Ron- ald Liesner 18637 (holotype, MO; isotype, VEN). Figure 8. Arbor 15-metralis; foliis 5-6, foliolis rugulosis Era obovatis apice a shrupte cutis basi acutis majoribus 26 cm longis 14-20 subtus prominente elevatis supra sulcatis, venulis tertiariis eo e grosseque reticulatis; petiolis 13-20 cm longis glabris, petiolulis 7-13 mm longis E E dicello H cm lon - mm crasso; pia e 3.5-3.7 longo .5 cm lato extus densissimo: fulvo stellato- densissim 25 cm longis; tubo stamineo 6-7 cm longo parte a glabra dimidia parte inferiori 4.5 cm dense stellulat tomentoso in 10 phalangibus anto: stylo 25-30 cm E inferne dense stellato- villosa superne seio: tomentoso; ovario conico 17 mm costis dense ferrugineo-squamosis ‘sie minute arene, i Tree 15 m tall. Leaves 5-6-foliolate, leaflets rugulose, elliptic-obovate, abruptly acute at apex, acute and long decurrent at base, the larger ones 12-26 cm long, 8-1 1 cm wide, upper surface with dark glandular dots d the lateral nerves and midrib, softly pilose below wi brown stellate hairs 0.5-1 mm long on both ce and nerves; principal lateral nerves 14— 20 each side; tertiary veinlets prominently grossly reticulate below, subsulcate above. Petioles 13-2 cm long, glabrous; petiolules 7-13 mm long, gla- brous. Flowers large, axillary, solitary, opposite on the stem. Pedicels 3 cm long, 5-7 mm thick, densely dark brown stellate-tomentose. Calyx tu- bular-campanulate, 3.5-3.7 cm long, 2-2.5 cm wide at the subtruncate, scarcely repand apex, densely fulvous stellate-tomentose without, the stel- late tomentum with numerous small hairs and a few larger ones, densely buff-sericeous within. Pet- als tan (fide Liesner), linear, 25-31 cm long, ca. 1 cm wide, on both sides densely stellate-tomentose. Stamens about 50, 18-25 cm long; staminal tube 6-7 cm long, the upper portion glabrous, the lower portion 3.5-4.5 cm long with pale, minute, stellate tomentum, separating above into 10 phalanges 3- 4 cm long. Style 25-30 cm long, densely stellate- pilosulous, especially in the lower half. Ovary conic, 17 mm long, 10 mm wide at base, conspicuously 5-costate, the ridges densely ferruginous squamose with minute pale appressed hairs in the sulcations. This unusual species is at once distinguished by the softly brown long-pilose lower surface of the rugulose leaflets, relatively large flowers with mark- edly elongated petals and staminal tube, and mi- nutely dense brown-stellate tomentose, elongated calyx tube. The collections of Ronald Liesner have contributed greatly to our knowledge of the rich flora of the Venezuelan Guayana. Pochota mawarinumae Steyermark, sp. nov. TYPE: Venezuela. T. F. Amazonas: Dept. Rio Negro, near Cerro de La Neblina Base Camp, Rio Mawarinuma, 0°50'N, 66°10’W, 140 m, 2 Mar. 1984, Ronald Liesner 16355 (ho- lotype, MO; isotype, VEN) Arbor 15-metralis; foliis 5- 7-foliolatis; foliolis coriaceis subtus glaucis argenteis late obovatis apice rotundatis minute mucronatis basi longiattenuatis 20-32 cm longis 8.5-10.5 cm latis subtus minute lepidotis, nervis late- ralibus principalibus utroque latere (6-)8-10; petiolis 2 4 cm longis, petiolulis 1-4 cm longis; pedicellis 1.5- 3.5 cm longis dense minuteque fulvo-stellato-puberulo; receptaculo manifeste glanduloso fulvo-stellato-puberulo; calyce breviter campanulato apice truncato 10-13 mm longo 15-20 mm lato extus dense minuteque fulvo-stel- 1080 Annals of the Missouri Botanical Garden IRE 8. Pochota liesneri.— A. Por B rtion of flowering stem.—B. Single leaf attached to stem.—C. Showin larger and smaller stellulate hairs on staminal tube.—D. e Showing one large and numerous smaller stellulate tal. hairs on calyx tube and receptacle.—E. Stellulate pubescence on pet lat ed he et intus dense sericeo-tomentoso; petalis li- nearibus 13-16.5 cm longis 1.3-2 cm latis; staminibus 8-13 cm longis, tubo stamineo 20 mm longo dense mi- natoque i cinereo- nem “puberula: stylo parte basali 2.5 cm aliter glabro; ovario superne dense sericeo tomentoso aliter minute puberujenti: capsula nondum visa. Tree 15 m tall. Branches robust. Leaves 5-7- foliolate. Leaflets coriaceous, pale or glaucous be- low, broadly obovate, rounded at the mucronate apex, attenuate to the acute base, the larger ones 20-32 cm long, 8.5-10.5 cm wide, minutely lep- idote below. Petioles 2.3-3.4 cm long; petiolules unequal, 1-4 cm long. Principal lateral nerves (6—) 8-10 each side, arcuately ascending, conspicu- ously elevated on lower surface, subelevated on upper surface, branched, and anastomosing 1-2 Volume 75, Number 3 1988 Steyermark 1081 Venezuelan Guayana Flora cm from margin. Tertiary venation finely reticulate between larger areoles, impressed below. Midribs strongly elevated below. Flowers solitary, axillary. Pedicels in anthesis 1.5-3.5 cm long, densely brown stellate-tomentose. Receptacle conspicuously glan- dular, brown tomentose, 1 cm long, 1. Calyx shortly campanulate, 10-13 mm long, 15- mm wide, densely brown tomentose without, sericeous within. Petals linear, 13-16.5 cm long, 1.3-2 cm wide, minutely densely tan stellulate- cm wide. tomentose without, densely and paler stellulate- tomentose within. Stamens 8-13 cm long, numer- ous, ca. 80 in 10 phalanges 10 mm long; staminal tube 20 mm long, stellate-puberulent most of the length; anthers hippocrepiform-linear, 2.5-3.5 mm long. Style 10.5(immature)- 14 cm long, glabrous most of the length, densely pale stellate-tomentose in the basal 2.5 cm. Ovary conic, 5-carinate, dense- ly sericeous-tomentose in the upper 3, minutely appressed-puberulent in the lower 25 on the ridges and in the sulcations. Paratype. | VENEZUELA. T. F. AMAZONAS: Cerro de La Neblina, same data as type, 6 Mar. 1984, Liesner 16460 (MO, VEN). 'This species is most closely related to Pochota obovata (Robyns) Steyerm. & W. D. Stevens but differs in the longer petals, completely stellate- pubescent staminal tube, manifestly glandular re- ceptacle, predominantly larger glabrous portion of the style, and more rounded, merely mucronulate apex of the somewhat narrower leaflets. The Base Camp along the Rio Mawarinuma of Cerro de La Neblina served as a working base for the numerous scientists who participated in the recent expedition to that sandstone table mountain, and is the locale of many noteworthy species new to Venezuela or new to science. Pochota redmondii Steyermark, sp. nov. TYPE: Venezuela. T. F. Amazonas: Dept. Atures: stream 0.5-2 km E of Rio Coro-Coro, W of Serrania de Yutaje, 3 km N of Yutaje settle- ment, 5?38'N, 66?30'W, 200 m, 19 Feb. 1987, Ronald Liesner & Bruce Holst 21248 (holotype, MO; isotype, NY, VEN). Arbor 6 m; foliolorum laminis late obovatis Res ro- tundatis 10.5-48 cm longis 5.5-15 cm latis haud ma feste lepidotis subtus haud glaucis; floribus solitariis, d dicellis 1.5-3 cm longis glabris; calyce late campanulato 8-10 mm longo 8-13 mm lato su hoi iode petalis li- gulato- -spathulatis 8 cm longis superne 9-10 mm latis basi 5-5 mm latis; tubo stamineo 13 mm long mm d. lineatis, lineis 5 minute stellato- puberulentibus cum zonis latioribus glabris 0.9-1 mm latis alternantibus mu- nito; ovario dense tomentoso Tree 6 m tall, the branches glabrous, the ter- minal portion 1-1.2 cm diam. Leaves 5-foliolate; leaflets articulate, subcoriaceous, broadly obovate, rounded at apex, sometimes emarginate to abruptly cuspidate, cuneately acute and slightly decurrent at the base, 10.5-48 cm long, 5.5-15 cm wide, glabrous, not manifestly lepidote; lateral nerves 10-16 each side, impressed to subelevated above, subelevated below, anastomosing 5-8 mm from margin; tertiary venation conspicuously reticulate below, the larger areoles enclosing a minute net- work of elevated veins, less conspicuously reticulate above; petiole 18-22.5 cm 4 mm diam., glabrous, terete, not lepidote; petiolules unequal, 1-5 cm long, glabrous, articulate. Flowers solitary; pedicels 1.5-3 cm long, glabrous. 5-glandular, glands oval, 1-1.5 mm long. Calyx broadly campanulate, 8-10 mm long, 8-13 mm wide at summit, truncate, densely buff stellate- long. Receptacle tomentose without, densely pale sericeous within. Petals cream-tan, subcoriaceous, ligulate-spathu- late, subobtuse at apex, 8 cm long, 9-10 mm wide in upper half, 4.5-5 mm wide at base, densely stellate-tomentose within, more densely stellate-to- mentellose without. Staminal column 13 mm long, 2.5 mm wide, glabrous above and at base, with 5 vertical minutely stellate-puberulent lines in be- tween alternating with broader glabrous zones 0.9— ] mm wide, branched into 10 phalanges, the main basal branches 10-13 mm long, each main basal branch again branched into several divisions with ultimate 18-24 filaments on each phalange, a total of ca. 240 filaments 4—4.5 cm long. Anthers linear- oblong, 2.5-3 mm long. Stigma 5-lobed, the lobes ovate-lanceolate, 1 mm long, 0.7 mm wide. Ovary suborbicular, densely tomentose. Paratype. | VENEZUELA: T. F. AMAZONAS: Dept. Atures, Rio Coro-Coro, river and pres forested slopes, W of Serrania de Yutaje, 6-8 of Yutaje settlement, 5°41'N, 66?7' 30"W, 23 Feb. 1987, 3 Holst 21337 (MO, VEN). 20 m, Liesner & This species is related to Pochota obovata Rob- yns but differs in the shorter pedicels, narrower calyx and petals, alternately glabrous and stellate- pubescent vertical zones of the staminal tube, and nonlepidote, nonglaucous lower leaf surface. It is a pleasure to name this species for Parker Red- mond, who kindly took care of the logistics for the expedition to the Serrania de Yutaje. Pochota tepuiensis Steyermark, sp. nov. TYPE: Venezuela. Bolivar: Meseta de Jaua, Cerro Sari- sarinama, summit, 4?41'40"N, 64?13'20"W, parte NE, afloramiento arenisco con zanjones 1082 Annals of the Missouri Botanical Garden en formación de bosque achaparrado y árboles enanos, 1,380 m, 11-12 Feb. 1974, Julian Steyermark, V. Carreño Espinoza & C. Brewer-Carias 108938 (holotype, VEN). rbor 2.5-metralis; foliolis 5 obovatis apice rotundatis emarginatisque basi cuneatim acutis 3-6 cm longis 1.5- petiolis 1.3-2 cm longis glabris, petiolulis 2-3 mm longis glabris; pedicellis immaturis 4 mm longis; calycis tubo immaturo 5 mm m go apies S ma lato extus dense fusc co- tomentoso intus : gl petalis immaturis nei id -sericeis. Tree 2.5 m tall. Leaves 5-foliolate; leaflets ob- ovate, rounded and emarginate at apex, cuneately acute at base, 3-6 cm long, 1.5-2 cm wide, grayish below, not lepidote below, 1.6-2.7 times longer than broad; lateral nerves subhorizontal, anasto- mosing near the margin, 6-10 each side, promi- nently impressed below, slightly sulcate above; ter- tiary venation prominent and impressed below. Petiole 1.3-2 cm long, 1.5-2.5 mm wide, gla- brous, not lepidote; petiolules 2-3 mm long, gla- brous, not lepidote. Pedicels (immature) 4 mm long, brown tomentose. Calyx (immature) 5 mm long, 6 mm wide at summit, densely brown stellate-to- mentose without, tawny sericeous within. Recep- tacle (immature) prominently glandular. Petals in early bud densely brown sericeous. This species is distinguished by the small size of the strongly revolute, apically rounded leaflets; ci- nereous nonlepidote lower leaf surface; and short petioles, petiolules, pedicels, and calyx. It is known only from the summit of the isolated Cerro Sari- sarinama. ARALIACEAE SCHEFFLER A Schefflera simplex Steyermark & Holst, sp. nov. TYPE: Venezuela. F. Amazonas: Cerro de La Neblina, ridge line on Venezuela- Brazil border, 1,900-2,100 m, 17 Apr. 1984, Al Gentry & Bruce Stein 46694 (holotype, MO; isotype, VEN). rbor 6-metralis; foliis 4—5-foliolatis, foliolis oblongo- 5.5 cm latis supra praeter costam mediam minute pu- bescentem glabris subtus densissime velutinis; nervis la- -P n- tibus; EES 2- da brevipedunculatis, pedunculis sub anthesi 2.2-5 cm is densissime tomentosis; inflo- rescentiis reed solitaria capitatis, capitulis densi- floris ca. 20-floris 9-12 mm longis 10-12 mm latis dense tomentosis; floribus sub alabastro sessilibus 2.5 mm longis dense pubescentibus; stylo 3-lobulato. Tree 6 m tall. Leaves 4—5-foliolate; leaflets co- riaceous, oblong-lanceolate to narrowly elliptic-ob- long, obtuse to acute at apex, obtuse or subobtuse at base, 6.5-12 cm long, 2-5.5 cm wide, glabrous above except for the minutely pubescent midnerve becoming glabrous except at base, densely veluti- nous beneath; midrib elevated on both sides; prin- cipal lateral nerves 15-23 each side, 4-6 mm apart, extending to margin, subhorizontally spread- ing to slightly ascending at an angle of 10-20?, subsulcate above; tertiary venation subsulcate, sub- reticulate above. Petioles 9-14 cm long, 2.5-3 mm wide, minutely appressed-tomentellose. Pet- iolules unequal, 1-3 cm long, 2-2.5 mm wide, bicarinate above, minutely tomentellose. Peduncles 2—4, in anthesis 2.2-5 cm long, 3-4 densely buff-tomentose; inflorescence simple, soli- tary, capitate, each one terminating a peduncle; heads densely ca. 20-flowered, 9-12 mm long, 10- 12 mm wide, densely tomentose. Bracts subtending inflorescence deltoid-ovate, acute, 3-4 mm long, 2.5 mm wide at base, densely gray-buff tomentose without, glabrous within. Flowers in bud 2.5 mm long; calyx teeth acute, unequal, tomentose with- out, glabrous within; petals 2.5 mm long, 1.7-2 style 1, shortly mm wide, wide, As lobulat cid carinate; This species is unusual among the Venezuelan Guayana taxa of Schefflera in having the inflores- cence consisting of solitary heads terminating un- branched, short peduncles. Schefflera globulifera, a related taxon with capitate flowers, has a long- pedunculate inflorescence with the heads arranged in one or more verticils. Schefflera yutajensis Steyermark & Holst, sp. nov. TYPE: Venezuela. T. F. Amazonas: Dept. Atures, valley of Rio Coro-Coro, west of Ser- rania de Yutaje, E base of forested mountain 9 km W of river, 5°41'N, 66?9'30"W, 1,100 m, 11 Mar. 1987, Bruce K. Holst & Ronald L. Liesner 3426 (holotype, MO; isotype, MO). Arbor 28-metralis; foliis 6-7 -foliolatis, foliolis oblongo- lanceolatis vel late elliptico-oblongis apice abrupte breviter caudato-acuminatis, basi obtusis vel subobtusis 15-25 cm longis 7-12 cm latis supra glabris subtus minute densis- florescentiis late racemoso-paniculatis, pedunculis inclusis sub anthesi 15 cm longis 10-12 cm latis sub fructu 22- 25 cm longis 14-25 cm latis, axibus lateralibus primariis Volume 75, Number 3 1988 Steyermark 1083 Venezuelan Guayana Flora late divaricatis sub anthesi 5-11 cm longis sub fructu 6- 15 cm longis racemose dispositis 15-30 umbellas sim- plices 10-20-flores gerentibus, omnino ferrugineo-sericeo- tomentellis; axibus umbelliferis sub alabastro 6-9 mm longis 1 mm latis sub fructu 3-6 cm longis; dcs sub alabastro subsessili bus 0.5 mm longis sub. fru axibus pedicellisque minute densequ o- tomentellis; bracteis sub axibus primariis late lanceolato- ovatis acuminatis 3-5 mm longis extus dense ferrugineo- sericeo-tomentellis; stylis y fructu 5; fructibus globosis in sicco conspicue 5-carinatis maturis 1.5 x 1.5 cm praeter apicem minute pubescentem glabris. Tree 28 m tall, the bark slightly rough with longitudinal rows of elevated brown lenticels and widely separated petiolar scars. Leaves 6- 7-folio- late; leaflets coriaceous, dark green and shining above, bright copper-colored below when young, turning gray in age, oblong-lanceolate to broadly elliptic-oblong, abruptly short caudate-acuminate at apex, obtuse to subobtuse at base, 15-25 cm long, 7-12 cm wide, glabrous above, minutely and densely ferruginous-sericeous beneath. Lateral nerves 8-11 each side, forking 5-16 mm before reaching the margin. Petiolules unequal, 3-9 cm long, minutely and densely ferruginous-sericeous. Petioles of mature leaves 49-50 cm long, 6 mm wide, densely and minutely ferruginous-sericeous. Inflorescence paniculately branched with 10-13 widely spreading primary axes bearing 15-30 ra- cemosely arranged, simple umbels, each of these 10-20-flowered; inflorescence pedunculate, in- cluding the peduncle, 15 cm long, 10-12 cm wide in anthesis, 22-25 cm long, 14-25 cm wide in fruit, the peduncular portion 7.5 cm long in an- thesis, 14-15 cm long in fruit, all parts of the inflorescence and infructescence densely ferrugi- nous-sericeous. Primary lateral axes alternate to subverticillate, widely spreading, 5-11 cm in bud, 6-15 cm long in fruit; bracts subtending primary axes broadly lance-ovate, acuminate, 3-5 mm long, ferruginous-sericeous, umbelliferous secondary axes 6-9 mm long, 1 mm wide in bud stage, 3-6 cm long, 1-2 mm wide in fruit; umbels simple, 10- 20-flowered. Flowers (staminate) subsessile in bud, 4-6 mm-long pedicellate in fruit; calyx in bud shallowly 5-denticulate, the teeth deltoid, acute, densely ferruginous-sericeous. Styles 5, in fruit spreading over the summit and appressed, 1 mm long. Fruit pale green when mature, globose, sharp- ly 5-carinate when dried, 1.5 x 1.5 cm, the apex minutely pubescent, elsewhere glabrous, 5-celled. Schefflera yutajensis of Territorio Federal Amazonas is allopatric with S. quinquecarinata Steyerm. of eastern Venezuelan Guayana, having similar 5-carinately angled fruits with 5 styles and a paniculately branched inflorescence. The former differs by having the primary axes bearing nu- merous racemosely arranged many-flowered um- bels, fruiting pedicels 4-6 (vs. 18-25) mm long, mostly glabrous (vs. tomentose) fruits, obtuse or subobtuse leaflet bases, and oblong-lanceolate or elliptic-oblong leaflets. The species inhabits sea- sonally dry, evergreen forested slopes of the Ser- rania Yutaje. GENTIANACEAE SYMBOLANTHUS Symbolanthus huachamacariensis Steyer- mark, sp. nov. TYPE: Venezuela. T. F. Ama- zonas: Cerro Huachamacari, diagonal ledge, cumbre, 1,700 m, 4 Dec. 1950, B. Maguire, R. S. Cowan & John J. Wurdack 29859 (holotype, VEN; isotype, NY). Planta herbacea 1-metralis, caulibus bono o foliorum laminis lanceolato-ellipticis apice ten minatis basi E id dp cuneatis, maj 5-14 4.5 cm longis 3- atis; nervis lateralibus principalibus utroque latere i bo petio -17 mm ongis; bracteis sub pedicellis subulatis 9— 13 mm longis 1-1.5 mm latis; calyce 30-32 mm longo, lobis lanceolatis acuminatis vel breviter caudatis ph anthesi 25-27 mm longis sub fructu usque ad 35 mm longis basi 10 mm ihe M hypo- Ha pm 5.2-5.5 cm longa, tubo 3-4 cm longo 8- m lato; lobis late ovato- ride M acutis 15- 17 m mm longis prope basin 15 mm l Perennial herb 1 m tall. Stems quadrangular. Leaf blades lance-elliptic, slenderly acuminate at apex, long-attenuate cuneate at base, long-petio- late, the larger 8.5-14.5 cm long, 3-5 cm wide; main lateral nerves 2 on each side, the uppermost pair arising about % of the length of the leaf blade above its base; midrib and lateral nerves incon- spicuous, at most impressed. Petiole 10-17 mm long. Flowers 2, terminal; pedicels slender, 15-35 mm long. Bracts at base of pedicels subulate, cau- date at the apex, 9-13 mm long, 1-1.5 mm wide. Calyx 30-32 mm long, the lobes lanceolate, acu- minate, or shortly caudate, 25-27 mm long in anthesis, elongating in fruit to 35 mm with caudate tips, 10 mm wide near the base. Corolla magenta- rose red with purple lines in the white throat, sal- verform, 5.2-5.5 cm long, the tube 3-4 cm long, 8-10 mm wide; lobes broadly ovate-suborbicular, acute, 15-17 mm long, 15 mm wide near the base. Anthers, including the subulate appendage, 7-8 mm long. Ovary lance-ovate, 11 mm long, 5 mm wide, with 5 squamellate truncate scales 1.5 mm long, 1.2 mm wide at base of ovary; style 3 cm long; stigmas ovate, obtuse, 3 mm long. 1084 Annals of the Missouri Botanical Garden Paratype. | VENEZUELA. T. F. AMAZONAS: Cerro Hua- minutely cuspidate, 17 mm long, 21 mm wide at chamacari, upper escarpment, 3 Dec. 1950, Maguire, Cowan & Wurdack 29801 (NY, VEN); above diagonal ledge, Cerro Huachamacari, Camp II to escarpment and ,500 m, 5 Dec. 1950, Maguire, Cowan & Wurdack 29879 (NY, VEN). This species is distinguished in having attenuate- acuminate calyx lobes that attain 25-34 the length of the corolla tube, and in having a relatively small, tubular corolla. Symbolanthus yaviensis Steyermark, sp. nov. Venezuela. mazonas: Dept. Atures, summit of Cerro Yavi, headwaters of Rio Parucito, eastern affluent of Rio Mana- plare, 5?43'N, 65?52'W, 100 m, 24 Oct. 1986, Otto Huber 11649 (holotype, MO; iso- type, VEN). Suffrutex 1-1.5-metralis, caule quadrangulari 4-4.5 mm diam. glabro; foliorum laminis elliptico-ovatis apice subacutis basi cuneatim acutis 3-3.8 cm longis 1.2-1.6 cm latis ubique glabris; nervis lateralibus utroque latere duobus subtus elevatis in parte inferiori 1⁄4 laminae exo- rientibus; petiolis 6-7 mm longis; floribus terminalibus wipe pedicello 6 mm longo; calyce 18 mm longo, m longo 10 mm lato, sepalis late ovatis apiculatis parte basali angusta 12 mm longa 5 mm lata; ] mm endo inserto, libero 5-6 mm; fila- mentis paullo inaequalibus 3-3.5 cm longis glabris, squamis late ligulatis apice truncatis late 2-denticulatis 4 x 4 mm; antheris 7 mm longis, appendicibus 0.5 mm longis; stylo 3.8 cm bun ovario lanceolato-ovoideo 9 mm longo 4.5 mm lato. Suffruticose plant with subherbaceous stems, 1— 1.5 m tall; stems quadrangular, 4-4.5 mm diam., glabrous. Leaves short-petiolate; leaf blades firmly membranous, elliptic-ovate, subacute at apex, cu- neately acute at base, 3-3.8 cm long, 1.2-1.6 cm wide, glabrous both sides, decurrent; midrib sul- cate above, elevated below; lateral nerves 2 on each side, arising within the lower 1⁄4, elevated below, obsolescent above; tertiary venation obscure or obsolescent. Petioles 6-7 mm long. Flowers terminal, solitary; pedicel 6 mm long, 3 mm wide. Calyx 18 mm long; tube 4 mm long, 10 mm wide; sepals broadly ovate, apiculate with a minute ap- pendage 0.2 mm long, 15 mm long, 14 mm wide, scarious-margined, entire. Corolla pale green in lower 25, roseate within, creamy white on lobes, broadly infundibuliform, 6.4 cm long (tube and limb cm long, the constricted basal portion 12 mm long, 5 mm wide; lobes depressed-suborbicular, middle). Staminal tube 17 mm long, attached 10— 11 mm above the base of the corolla tube, free 5- 6 mm above the attachment; filaments somewhat unequal, the longer ones 3.5 cm long, the shorter 3 cm long, glabrous. Squamellate scales at base of filaments broadly ligulate, 2-denticulate at the trun- cate summit, 4 mm long, 4 mm wide. Anthers curved, 7 mm long, 1.7-1.8 mm wide, dorsifixed in the lower 1⁄4, terminating in an attenuate ap- pendage 0.5 mm long. Ovary lance-ovoid, 9 mm long, 4.5 mm wide; squamellate scales at base of ovary truncate, 1.5 mm long; style 3.8 cm long; stigmas ovate, 3 mm long. From Symbolanthus sessilis Steyerm. & Mag. of the Meseta de Jaua (Cerro Sarisarinama) of Estado Bolivar, Venezuela, this species differs in the smaller corolla, calyx, and leaves; shorter ped- icels; entire calyx margins; corolla color; and leaf shape. It is distinguished from S. calygonus (R. & P.) Griseb. by the smaller corolla, calyx, and leaves; shorter pedicels and petioles; and minutely mucronate corolla lobes. RUBIACEAE REMIJIA Remijia sessilis Steyermark, sp. nov. TYPE: Ven- 6°10'W, 400-500 m, 20 Feb. 1985, M. Nee 30967 (holotype, MO; isotypes, NY, VEN). Figure 9 Frutex 2-metralis, ramulis petiolis foliorum laminisque longihirsutis pilis brunneis praeditis; petiolis alatis inflatis bulliformibus 2 js m longis; laminis cm longis 5-11 m latis extus hirsutis intus glabris. Hollow-stemmed shrub 2 m tall, the branches brown-hirsute. Petioles winged, inflated-bladdery, 2.5-3 cm long with long, brown hairs. Leaf blades rugose, especially above, firmly membranous, broadly obovate, shortly acuminate at apex, con- spicuously attenuate to the base, 36 cm long, 16 cm wide, brown hirsute with spreading hairs 2.5- 4.5 mm long; lateral nerves 13-14 each side. Inflorescence sessile, densely many-flowered, 2 cm Volume 75, Number 3 1988 Steyermark Venezuelan Guayana Flora 1085 NW 7 ^ N x NON ` AS hy EZ = a Wak » Ups hg Z y N A P CONSER SA AV ~ D ^/. 3 =$ 47, 7 XS Y M Y ⁄Z SN SZ Z INIA B NU FiGURE 9. Remijia sessilis.— 4. Portion of stem with infructescence. —B. Calyx and hypanthium in post- lob I anthesis.—C. Calyx lobe. long, 1.5-2 cm broad; flowers sessile, 12-15 in each cluster between the stem and base of petiole. Calyx tube campanulate, 3.5-5 mm long, 3.5-5 mm wide, brown hirsute with spreading hairs; lobes 5, broadly triangular, attenuate, 1-1.2 mm long, hirsute-ciliate. Hypanthium obconic, 5 mm long, 2-2.5 mm wide, densely brown hirsute. Capsules loculicidally dehiscent, the valves linear-oblong, 3.2-3.8 cm long, 5-11 mm wide, short-hirsute without, glabrous within. This species is unique in the genus in having a congested, sessile inflorescence. It strongly simu- lates Remija physophora Benth. ex Schum. of the Rio Vaupes, Colombia, in the conspicuous, long, brown-hirsute pubescence of the stem, petiolar leaf 1086 Annals of the Missouri Botanical Garden blades, and calyx, as well as shape of leaves and bladderiform, inflated petioles, but that taxon has conspicuously long-pedunculate, shortly cymosely ranched inflorescences with scattered flowers. The calyx lobes of R. sessilis are also relatively short- er and more broadly deltoid. Except for the dif- ferences in the inflorescence, the two taxa are strikingly similar. SIMIR A Simira ignicola Steyermark, sp. nov. TYPE: Ven- ezuela. Bolivar: Dist. Cederio: shaded canyon on igneous cerro, 1 km S of Quebrada La Flore, affluent of Rio Ore, affluent of Rio Par- guaza, 6*17'N, 67%5'W, 85 m, 9 Sep. 1985, Steyermark, Holst & Manara 131659 (ho- lotype, MO; isotype, VEN). Arbor 15-metralis, foliis ovato- -ellipticis apice acumi- natis base cordatis 17-19 cm longis 11 cm latis subtus dense tom centia trichotome _eymosa, pedunculo 2-2. 3m E Jongo m longo; lobis calycinis 4 ae. corolla in N wass 3-3.2 mm longa 4-lobata extus minute hirtella. Tree 15 m tall, the wood dull rose when cut. Leaves subcoriaceous, rugose above, dull green below with elevated nerves. ovate-elliptic, acumi- nate at apex, cordate at base, 17—19 cm long, 11 cm wide, glabrous above, soft tomentose below, densely so on midrib and lateral nerves; lateral nerves 15-16 each side, slightly ascending at an angle of 25-45”, terminating near the margins without forming any common anastomosing nerve; tertiary veins subsulcate above. Inflorescence ter- minal, cymosely trichotomous with 3 main axes, 5-5.9 cm long, 5-6.5 cm broad; peduncle 2-2.3 cm long, 3-3.5 mm wide, sparsely pubescent; low- er main axes 2.5-3 cm long, densely brown to- mentose, the upper main axes 1.5 cm long, densely brown tomentose; ultimate axes 3-6-flowered. Ca- lyx and hypanthium 2.5-3 mm long in bud; calyx 4-lobed, the lobes suborbicular, rounded, ciliate. Hypanthium 2 mm long, 1-1.2 mm wide, subcla- vate, minutely hirtellous without. Corolla urceolate, 3-3.2 mm long in bud, 4-lobed, minutely hirtellous without. Stamens 4; anthers oblong. Style 2 mm long in bud, glabrous The small flowers and the leaf shape resemble Simira cordifolia, but the leaf blades and petioles are densely pubescent. From S. rubescens the new taxon differs in the pubescence of leaves and floral parts. REVISED SYNOPSIS OF PANAMANIAN EUPHORBIACEAE! Grady L. Webster? and Michael J. Huft’ ABSTRACT Collections made in the last two decades have added 9 genera and 42 species to the 35 genera and 100 species included in the Flora of Panama treatment ulis combinations proposed in this paper are Richeria Huft, Tetrorchidium microphyllum Hufi, Croton pach cunensis Webster, Croton draco Cham Schldl. essleri Webster, Phyllanthus gentryi Webster, Hufi, Dalechampia canescens Kunth subsp. friedrichsthali (Muell. Arg.) Webster & ypodus subsp. panamensis (Klotzsch) Webster, Gol bilake n 1968. The new taxa and Tragia correae & Hufi, Er aes 8 a roton speciosus Mue subsp. uphorbiaceae published in ebster, Muell. Arg. subsp. pyramidalis (J. D. Smith) Webster, Croton santaritensis Huft, Sebastiania panamensis Webster, Gymnanthes dressleri Webster, and Gymnanthes farinosa (Griseb.) Webster. In addition, new or updated ks are provided where appropriate, as well as descriptions and specimen citations When the treatment of the Euphorbiaceae for the Flora Panama was originally published (Webster & Burch, 1968), it was anticipated that it would prove to be incomplete, but collections made during the past 20 years show that it was even more provisional than we had thought. Not only has Pausandra been found as predicted, but no fewer than nine other genera new to Panama: Adenophaedra, Astrocasia, Croizatia, Drypetes, Gymnanthes, Maprounea, Richeria, and Sene- feldera, and an unpublished genus from Cerro Ta- carcuna in Darién; this brings the number of native genera to 45. In addition, species new to Panama have been found in a number of genera, including Acalypha, Alchornea, Cleidion, Croton, Dale- champia, Euphorbia, Hyeronima, Mabea, Man- ihot, Sapium, Sebastiania, Tetrorchidium, and Tragia. The arrangement of genera within the family has become obsolete since the publication of a new classification (Webster, 1975); there are now five subfamilies recognized, of which four oc- cur in Panama. This new treatment, with the order of the genera now following the revised classifi- cation, includes the taxa new to Panama as well as references to recent publications on these taxa. In order to incorporate all of these additions and changes, the generic key has been revised, and new keys to species have been made for several genera. Citations of specimens are not given for species already included in the original treatment unless they represent new records for provinces or considerable range extensions. Descriptions are provided for most species new to Panama, but in a few cases, if the Panamanian material is not adequate for description or if the species has been recently described elsewhere, a literature reference is given in lieu of a description. The preparation of this paper has involved us in the study of much extra-Panamanian material and has led to the resolution of a number of an- cillary taxonomic and distributional problems. We have not hesitated to discuss these additional items where appropriate. The central position of Panama in the Neotropics certainly renders this account of Panamanian Euphorbiaceae of value to an under- standing of the family throughout Central America and northern South America. It seems appropriate, therefore, to include peripheral items that, w not concerning Panamanian species directly, are definitely of relevance. ! We thank the curators of the several herbaria for the loan of specimens. Part of the work of the senior author was supported by grants from the Nat California, Davis; considerable assistance was provided by the staff of the Missouri B. McPherson, who pido mae of some critical taxa. Drawings were provided by Wan-Ling Dr. Gordon Peng, Clara Richardson, and Ste ional Science Foundation and from the Graduate Division, University of articularly vri! CCIL, 2 Department of Biology, University of ‘California, Davis, Be 95616, U.S.A.; sara dey ipii N for 3 p key to genera and new treatments of genera um 27592, 34, 36, treatment of genus 28, Cnidoscolus, is furnished by Dr. Gary pu lid of Puerto Rico, Mis 3 Missouri Botanical Garden. Mailin dress: Departm Chicago, Illinois 60605, U.S.A.; primarily responsible for new treatments of genera gue ment of Botany, Field p of Miel History, 12, 13, 14, 17, 19, 20, 26, 30, 33, 39, and 42; the treatments of genera 16, 22, and 35 are the joint 5 m of both authors. ANN. Missouni Bor. GARD. 75: 1087-1144. 1988. Missouri Botanical Garden Annals of the `šuiqun|ə jou ‘sqniys Jo səən ‘aeueo Ápuoururo1d jou ‘g 10 Z sjadieo t[enxəstun saoueosa1ogut :ojeuuoo Á][&seq Jo 3313 sajÁ1g "egT 'juesqe suey Suldung “e, | *ssoloe WD Y UB] sso| IMI} ‘ysydind 10 ysippal Apounstp jou xaje] (pa3repua you saAnoouuoo ‘z J) o1our 10 g Á[isour suaurejs :ojeA[PA spedas ojeururejg ‘eg “pews `qçI AJaa Ə|9un:e9 ƏY} JO ələe[nəunieoə spəəs təstw1əu1o soouaoso1ogut tpəqo| Ajayeuyed 10u ajad sayitau saava] tpəqoueiq jou suaureys ` eje[nounieO spaas ‘[eIsIP siƏwoJ snuiny- RU ay) *eje[norued euru saouoosa1ogut :peqo| ÁJereured ‘oead seara] :sno1eumu s1oqgjue 'pagrurei-9je[nOtose] SUSUIBIS 'ec[ 'sjo*1q pautoa Á[ereured ç Aq peurroj a1onoaut ajetqe[Iq e ur paso[pua jou s1oMo[q ep] "Áutgs 10 *xoe[q ‘ayejnounseo jou spaas *a1onjoaut pax[eis e ur pəso|ouə 10u 9[ISSIS 1oqjrou s19MO[] “PET '"juasqe səd `qII piumDyiAFip ‘OI ^ epojustd inoqa s1940g ayeutureis (41ed ur 1seo[ 18) snoaoeryStdyeur seq :[o[ered Apounsip jou sura [e1eje[ Əy} *əmuə soAvaT] `qZI vruosado) `6 Əpollüstd yım siəwog ojeuturejs ‘rempug? 10 apduns sureq ‘aered Ápounsip sutea [8.1918] Əy} *oje11os Ájəuy saAe9] ‘eZ 'e1e[0940] SSA| 10 IIOU spoos :()| sueurejs :seuraoe1 ut SISMOY *juesaid spejaq ‘eT | [ eeeproqdápeoy Aqrurejqng] *(seporued ojut po1e32188e soumouros asay}) saxids 10 *sourooei ‘sdajsnjo Krejuxe ut siamoy :(n4244 ut ajopido[) snoave ideu 10 apduns umjuoumpur :(D19q pue snuto1y ut Ájuo oje[nounie2) oje[noun1e2a Ápisow spaas :eje1od[oo-g Ápisour sure12 uojod *(juasaid uejjo spue[8 1euture| peppoquro) juasqe spue[2 1eonad payjeis tpoutea A[o1euutd saAee[ uay1 juosa1d jt 10 *juasqe səd “Ol "punoduroo 10 peqo[ A[oieurped oi əjduns soAeo[ ssnowea umiuəurputi : *juesqe 10 juesa1d sjejed :pnq ur s1oqjue əy} Suuo4o2 X[Ə1ə[duroə K[|ensn *oieA[pA 01 aje»uqurt sjedas :(wn1p14210473] ut 1dooxo) 'snoureA umjueurnput :spue[8 Iejonad 10 reuture| uit tÁ1eAO Əy} Jo [[29 yoea ur Arenos sama q1 uə1Jo *peqo[ Jo pautaa Ajayeurjed 10/pue ojejuap uaijo səApə[ tÁx[rur Jo pa1o[oo uojjo des :aje[nounieoo 10 9je[noungeo spaes smua q cg SNOLIBA JPY əy} 'snorooorp JO snovou ou i(juasqe 3stp Ua) * Jt 10) £ 10 g suaurejs :Xuoq 10u AYsay 1euiteu spəəs :(2m1572 `J ut 1daoxo) seA[eA oiu jueostqep '(z A[orei) £ sde) :qg DUDNIDFID p: pue SIP re[nuue YM 190g ojeururejs “1809 IUU! Áuoq pue 1209 19jno usm[q Áysay yım spaas 5jueosmgap Áe (9 10 € Á[o1e1) ç 10 y sadie?) “eg "e[n2o[ 10 G-E sjedieo :xsrp [ENUI e MOUM srəwog əleurure1ç "92 lod Z spaas *juaosiyap Á[1sour sima] :peie[p 0} 1aopue[s seursus ey) ‘pyig Ajjounsip sajÁ1s :a1oui s33adXa(T ` A yore ut Áreos spaas ‘snosoednip sima] :peje[p seurgns ayi *eje[osqo sajá1s ‘g 10 T s|ədreə *xsip ([pururejse1jur) [enu95 e YIM s19^op ojeurujg “e, ‘apoyusid jueuruo1d e Suryov| ((snonproap sie[qoue1q uəu1 *asouraoei Jt 10) s1ajsn[o ÁKre][IXe ut SIƏMOY ojeuturejs tsnot9əotp 10 SNODIOUO 'qc p snoasednip sima] :Z s[adueo :sno[npued soes sayjue :sno[edosoure8 Apounsip x&[eo :ojopido[ Apureur urnjueurnpu] -qg £ 10 Z slədreə ‘snojnpuad jou soes 19y1ue tsnopedastioyo [eau ‘paqoy Ájdəəp xÁ[eo ‘ajduns umjueumpuy “eg 'epo[[ustd 1uəururord qii ‘saurases 10 saxids ut siəwop a1eururejs 'snomeoq] “PC 'juesqe sed 'qz DUNUOLIA Y `ç DUIYNMY p wa Áusəg spəəs :re[nsdeo simu : DIJVZIOL) `£ == SALA sduim K1eded £ out papuedxa Á[eisip vj[ournjoo :pyq aot«1 sajis quəosəqnd sjejad turajs pue ajoued usemjaq Sunoalord jou se[ndns ‘snowecrq `qp pesuim Kreotde jou ‘aseq uro1j Sutsodey e[pourn[oo ‘pytq *perepip sajÁ1s tsno:rqe[8 sperod :1eponade.nur so[ndus *snor ouo] “ey "uasqe 10 Ájueos uriadsopuo YIM spaes :ajeuuoo 10 3313 sjuaure[g *s[edas uey, 1911098 yonu sed “qe unadsopua snordoo qi spəəs :ajeuuoo sjueure[y *spedos ayy uey, 198uo[ 10 Suqpenbo seraq “eg 'jusseid sjejag “ez ['seaptoqyuey¡4yg Ápurejqng] ‘ajoprday 10 o[duirs urnjuoumpur DISDIOLISY/ ^I "el 1088 'eje[noun1e29 spəəs :KreAo ay} jo [[22 qoeo ut pated so[na() *spue[8 reurure| pəppəqurə inoy1m Áppensn *payioo] Ápemosqo 10 əmuə *paqojun seA?9| “juasqe xoje| Ax < VWVNVgd NI dAVaOVIGUOHdNY JO VYANA AHL OL AIN 1089 Webster & Huft Volume 75, Number 3 1988 Panamanian Euphorbiaceae umprn2104ja[ “0% sit snoooergsid[eur pue spue[2 iejoned payers YUM S9ABA| {saad} JO SQNIYS snotəəorp ‘Xysey spass :oje1od[oo-g surei18 uo[[od :eie[[ooop-p Suuveodde pue ayeyed siayjue oui ‘g sueurejs juesqe sjed “ez [ &&oprouojor) Apiureyqng] T T EI dM Un CE A EU RM SEDE MM UU DOMAINE ADAM UNDC ND asso soAv9| pai2asstp 10 *poqo| “pausa X|əleuu[ed ym səuta Surequre[o 10 Zuu; ‘siey 2unvjur YUM powe jmjj pue Á1eA0 :seurdus aui 01 ejeuuoo sajÁ1s :sjoeq (peqo| 10) pautaa A[oieured Z jo o1on[oAut eietqe[Iq e Aq pepueiqns s1o^og *[enxestq qenguepnesd seoueose1ogu[ ‘qI vajoydw “pz ssoJ98 ul) ¿1-8 ‘wopwod 11m] “ysippa. 10 ysudind xər] *Aysoy pue pa3uepua Apeai3 soAnoouuoo 1eqjue ayi ‘g sueureis “(s[IOGM Z Ul) e1eouqur “p sjedas aeunue1s “qQT DISDAL “ZZ v Suiduns yım jueoseqnd ÁA[snordoo ‘aseq ie oiep1oo səAeə[ ‘saura BUIUTM] snoro9ouour “pajina Ajyeotde jou siəuqtuy `qoz uoyp»op “1G : - juaosa1qe|8 *oseq je ojeouno soA?o[ :s991] 10 sqnius snorooorp tsarey Sutguns jo 1jni pearde qi s1oquy ‘29g *eje[nounueooa *qjoours spaes 5juesqe ASIP MO[9Q o1euuoo *peqo[un səs squaseid sey 2utgung “q) | DNaUayNd “OZ seuer toseq 1e 1e[npue[giq saaea] to1eutreo Apuoururoud “y s|ədre5 t re]pixe ‘J Iq pur: [o9 Suo| e o1ut 91 I&S “81 nydXqvop ^61 aSOIMIISA Á[osuep ÁIBAO əy} Əs[9 JO 'sjoeJq ejeururejs aq Uey} 19318] yonu sjoe1q ajeppusid :ejeotds soouooso1ogut ‘ayerutoR] sə[K1s *surrojturi9A sIOyUY `q61 pauloyyp S UU (qed ut seag 18) e1ej[ois Ápojnurur urnjuournput ‘Aip 'eje[nounieoo 'oje[no1eqni spees :g Ayyensn suaureys “aus ‘33I ‘ayeBuoja sajÁ1s ‘z sjediey :q0Z uoipio]) “LT 91e1npur səmdys ‘yuasqe xsrp Ə1e|[üstd *pesre[uo aatjoouuod JoyUe ‘QG uet ILOW SUSUIBIS "q pz pyəpp Sp er ne tee M 1e[npue[ | (urur QZ 1seo[ 18) s[oorped Buoj yım s19^0g ayeyusid :ajeruroe[ sapá1s `qçZ DIPADULI i| ^ 7 aseq 1eou spurj? Ieo} snonotidsuoo yum soAeo[ ‘(Zuo wur ç uey} ssə[) sjaoipad 1104s YIM JO a[issas SIIMO eie[[ustd “pyiq sewn Z Jo I SIÁS “ecz "ut so[ndnus 51uese1d xsrp eiejpisid :pos1e[ua jou eAn2euuoo 1eqjue ‘QF uet SSA] suaure]g "ep *juajsts1ed səuməuros so[ndus :01 Uey, o1our sueureig “q[Z Q nipanydouapp cg UU Zuo] ur» [ uey} SSA] spəəs :ur1ojneur -Sus ‘papp X[peoiq so[Kis tg sueurejs 5juesqe ASIp əleuturels tie[npue[8ə saAe9T 'qgc uo1ipuapoKav,) "cl Suo| ur» [ 1940 spoes to1eZuo[a sajÁ1s *) -p sueurejs tur1oputa[nd ‘aatsseur ASIP ojeururejs tspue[g reuture[ YIM SIABIT “PEZ ‘poutea A[oieuutd saava, ¿Aysoy jou 3809 paas :(ired ul ise9| 18) [eurur19) saoueose1ogut :juasqe opo[mstd 'qcc sisdoousoygpy "yg UU paaaur din soaBa] “Áysoy 1809 poos “ÁIe[[Ixe soouaose1oput tsrəwog ajeumuejs ur juəsə:d epo[[nstq “ezz 'snot2aorp SNOIMPEI Jo juesqe səmdys :Q[ UY] 1949] suauTRIg ‘LIZ -(Aysay spəəs uay1 *peqo[un J 10) pyiq sajáas ‘g sjadiey ‘207 (vyapp ut 1deoxo oje1ooe| pue iepue[s jou Inq) paqojun Jo pytq uejjo ‘SNOVA so[Áis “s19e.1q ojeunue]s uoy padeys Apueiogrp 10 193181 Á[SunLos jou sjoesq oje[[nstd mopa jou SIayuy “e6] Missouri Botanical Garden Annals of the 1090 DINH “lh 777 ç uey} el0ur sjedies :əuoo Xusəg e ojut poje2a183e siəwog ojeuturejs pim wo ç-Z ysip Aysay 'peqo[ e ut Sureururio; ‘Buoj wo c'z isee| ye uumjoo JejA1s 'q9e aunuoddig ‘OF peired-g xÁpeo əle[[qstd :oje[nounueoo spəəs :snoaoednup simuj *Ie[n20[-6-9 AIBAQ “LE umdos “6¢ ` Z sueurejs ‘snoyAydoures Ápounsrp saodjeo :spue[8 ayeydys 10 Sunoalord jo ned e yum ÁAqpensn sajoned :Kusəg 1809 ayi 'eie[nounueo2o spaas “age DIFUNINS ^gg aJ9UNIBI jews yum (smo ur) spees :spue[8 uniojrgeÁo [eseq poured ym səAeə| ‘snordos Ájfensn xaje] (yur :aseqouÁ3 poguoid-g juajsisiad yum e[[oum[o? :qgg gpunies ojendá[eo naunoudpW ^;€ 93.8] YUM *eje[ooA0J spaas tojeuuoo Ápounusip saJÁ1s pue suourejs təsoqo|8qns *po1oeuiuoo a2ue2sei1ogut Jo yed ajeurujg 'qop sayjuDUWA “OF a Zuo] uu 01 uey} a1our spoorped 2unmuj *ijoours KreAo 5juesqe 10 1uəur8əs [ Jo xÁ[eo ojeururejs tÁreguxe sayidg :qI+ DIUDIISDAIS “CE Suo| ww ç uey} sse s[oorped Sumy :ejnu100-9 Áeao :peqop-g xÁpeo ajeumue]s sara] əy} aisoddo 10 purua} saxidg ‘erp `əl9un:e2 pews YIM spəəs :pejeguo[e oouoosoiogut jo ed ojeuruig “ey “spurs pax|e1s JO paste moyim seAee[ :(1uəsqe sourjauros sdeqied 10) Ájueos xoje| Ayu toseqouA3 poSuoud-g juajsisiad 1nothtw euno “e6€ 'as1y s[edes oje[usid :re|mpue[8ə sajoned ‘Xap 1809 pəəs ayy *eje[nounieo spasg “egg "1e[nsdeo simi] tiepnoop-g(-z) Areag “PLE "sioqiue ye sjoe1q 2uumidni jou *so[n1ouro[8 ut s1amoy ojeururejs tsdy 1opuo[s ozur Sut&1oAip ‘Buoj ur ç-Z uey; sse| uumq[oo IeJÁ1S "egg "pedo[e4ep [Jom Ápsour sjueurepg yım sueureis təsourəəe: JO ojeoids S9JUIISIJOYU] “GPE Diapjafauag "ve ` € (smo ut) suaureijs topou qoeo 1e Árejos *e[rssasqns siomoy ajeumue]s təə1J Á[1eau so|Kis tjuosqe umjuaumpu] “GE D2qUjy `E ^ SOUL 10 QT (smo ur) suoureijs :opou sad o1our 10 Z *eiej[oorpad-3uo| siamoy a1eururejs tpajeSuo[o uurnjo2 1e|4is *onnpuep umjueumpu] “PCE 'e[»e1deoe1 pareaaja ue uo o[rssasqns sueurejs foje[norued 10 prosiKui səouəosəiogu] “BPE "Peqo[un se[Ájs taxonjoaur snop[&udoure8 e ym ¡eryiuepnasd jou “ozemomed 10 *osourooei ‘ayeotds saguaosalonur -e poqo I [oaur sno[[4y WA [erq1uep [nal - BUI EEE dL CREE e[durs umjuauinput ‘saan Jo sqniys snotoəouour :ojep[nounjeoo spass :e[-9 sperd ajeumue]s talou 10 001-0€ sueureis *sjuourgos aeaea Olur Zuunijdni xÁ[e) ‘qrg DIPUDBSNDA '0g ` ` snoave 31d -Jew umjuaumpur :ur1opoids *Krejrxe soouooseiogut əy} “SNOI93OIP “39.17 sjuaure[y eui ‘L-g sueurejs tpauraa Ájejeuuid seaeo7 “AZ€ n»ydoamf '6z 9vjes re[npue|8 payoue1q ssa] 10 aou jo 10 əjduns umnjueurnput :*[erseuorp ‘euru; səouəosəiogut əy; 'snor»eouour :ojeuuoo sjuaure[y aui *ZI-9 Á[isour sueureis “paqo] 10 peuteA ÁjJoieured SIALIT “BZE "'eje[nounae? spəəs ¿q Ayfeutiou spezad ojeururejs ‘gp ueu1 o1our jou suəure1s ‘sjedas ojeouqun jo xe?) qg `ə1opidə[ Jou əle[[ə1s 19yuou umauəumput :pnq ut 199.13 ss9[ JO 91our SISYIUY LOE ‘ayeanqiadeut sure18 uə[|od “Ax Apounsip jou xoje[ *(ə1optidə[ urmuəumput asja 10) 1uosa1d spe1oq snjoosopiu)) “BZ [Puturejse1r)xe ASIP :ojeuuoo sueurejs fysiyym xÁ[eo i(osreds K1oA səurnəuros) 1uosa1d suey Suigung ‘qez qoylUuDy “LZ [pururejse1jut pue pegua ysip “39.17 sueurejs tysidimd o} ustuəə:8 10 ystmoyad xÁ[e» uasqe sureq 2uigung epz 'eje[norued Jo [erseqorp *[Eurur19] so2uo2sa1ogut au; ‘SNo ouou ‘aye10d sures uə||od :Áxprur xa]e] ta¡duns urnjueurput :pəqo| Á[o1eurped sarta] *prope1ad xAyeo sjuasge s[elag “PBZ “Aysay you ‘(D12409 ut 1deoxo) ajepnouneo spaas :oje1nj1odeur 10 ayeiod surei3 uə[[od :eiej[ooo| p Suueadde jou s1egjue əy} ‘aiour Jo g sueurejs ‘(ajoptday umnjueumput 10 plojeyed xÁpeo asta 10) yuasaid sped 'q/c a2 © G Volume 75, Number 3 1988 Webster & Huft 1091 Panamanian Euphorbiaceae 42. Euphorbia ) obsolete or glanduliform; veins main axis aborting just above the cotyledons .... 43. Chamaesyce an taxa) ecarunculate or minutely carunculate. cyathium; pistillate flower solitary, terminal, the staminate flowers each with a (lobes) around rim; styles free or nearly so. ; seeds (in Panamani (but if opposite then not inequilateral at base); stipules (in Panamanian taxa , with glands alternating with bract tips , ; styles usually bifid; milky latex copious 43b. Leaves opposite, stipulate, inequilateral at base, 42b. Cyathia reddish, bilaterally symmetrical, the involucral glands hidden within t , Le., the 5 involucral bracts fused into a cupular Leaves alternate, opposite, or whorled not chlorenchyma-sheathed; 42a. Cyathia more or less radially symmetrical 43a. single stamen, in 5 lateral cymes 33b. Inflorescences pseudanthial main axis not precociously aborting 44. Pedilanthus he nectar spur; styles connate into a long column .................... the veins with chlorenchyma sheaths; ENUMERATION OF TAXA Subfamily I. PHYLLANTHOIDEAE Asch. l. Astrocasia Astrocasia Robinson & Millsp., Bot. Jahrb. Syst. 6, Beibl. 80: 19. 1905. TYPE: Astrocasia phyllanthoides Robinson & Millsp. = Astro- casia tremula (Griseb.) Webster. Dioecious, glabrous trees or shrubs. Leaves al- ternate, petiolate; stipules ribbed, deciduous; blades entire, pinnately veined. Inflorescences axillary; flowers in clusters. Staminate flowers pedicellate; sepals 5, sometimes unequal; petals 5, longer than the sepals; disk annular; stamens 3 or 5, the fil- aments connate into a column, anthers extrorse in bud, dehiscing horizontally; pollen grains tricol- porate, reticulate; pistillode dilated at tip into a peltate disk capping the staminal column. Pistillate flowers long-pedicellate; sepals 5, articulated, de- ciduous; petals 5, longer than sepals; disk cupuli- form, surrounding the ovary; ovary of 3 or 5 car- pels; ovules 2 per locule, anatropous; styles free, bifid. Fruits capsular; columella slender, persistent; seeds 1 or 2 per locule, ecarunculate; seed coat dry, thin, smooth; raphe ESOO endosperm thin, flat, much copious; embryo straight; longer and broader than the radicle. This neotropical genus of four species was not reported from Panama in the original treatment. Astrocasia is one of the more primitive genera of Euphorbiaceae and has its closest relatives in Africa and Madagascar (Heywoodia Sim, Wielandia Bail- ) lon 1.1. Astrocasia tremula (Griseb.) Webster, J. rnold Arbor. 8. 1958. Phyllanthus tremulus Griseb., Fl. Brit. W.I. 34 TYPE: Jamaica: Purdie, Wullschlaegel (syn- types, Astrocasia plot Robinson & Millsp., Bot. Jahrb. Syst. Beibl. 19. 1095. TYPE: "Mexico. Yu- catán: AM cok 3943 (holotype, F). Shrub or tree 2-10 m high; branches terete or obscurely angled, pale; foliage deciduous. Leaves with slender petioles 2—6 cm long; stipules lanceo- late, chartaceous, 4-6 mm long; blades charta- ceous, ovate, acute or obtuse at tip, broadly cu- neate at base, 5-12 cm long, 3-7.5 cm broad; major veins 5-8 on a side, ascending, brochidod- romous; veinlets prominulous beneath; margins narrowly revolute. Flower clusters axillary, sta- minate and pistillate on separate plants (or on sep- arate branches of the same plant). Staminate flow- 1092 Annals of the Missouri Botanical Garden ers with pedicels 8-15 mm long; sepals broadly elliptic to obovate, entire, 1.2-1.5 mm long, 1.2- 1.8 mm broad; petals elliptic-lanceolate, 2.4-2.7 mm long, 0.8-1.1 mm broad; disk cupuliform, fluted, 0.4-0.5 mm high, 0.9-1 mm broad; an- droecium 0.7-0.9 mm across; stamens 5; anthers 0.4 mm across; pistillode head circular, 0.5-0. mm across. Pistillate flowers with slender pedicels becoming 2.5-5.5 cm long; sepals suborbicular to 2-2.2 mm long, 1.8-2 mm broad; disk cupuliform, its margin undulate, ca. 1 mm high and 2 mm broad; styles thickened, 0.6 mm long, bifid, the tips clavate. Fruits oblate, 3-angled, re- ticulate-venose, cocci ribbed on back; columella 3.2-3.5 .9 mm long; seeds plano-convex, elliptic, cylindric, smooth, yellowish, 4.4-5 mm long, 3.8-4 mm road. The recent discovery of Astrocasia in Panama is one of the most surprising additions to the flora, particularly since it was found near Madden Dam in what is surely one of the most heavily botanized locations in the country. Astrocasia tremula has a broad but greatly disjunct distribution from Mex- ico and Jamaica to Colombia, Venezuela, and Bra- zil. The Madden Dam locality, however, is the only known station in Central America south of Belize and Guatemala. Specimens examined. PANAMA. COLÓN: forests along shores of Eu 1 Lake, near Madden Dam, 50 m, vA 1299 (DAV, F, MO), Witherspoon 8805 (DAV, MO); 6 km N of na along Madden Lake, Knapp 2715 (MO). 2. Amanoa Amanoa Aublet, Hist. Pl. Guiane 256. 1775. TYPE: Amanoa guianensis Aublet. 2.1. Amanoa guianensis Aublet, Hist. Pl. Guiane 256. 1775. TYPE: French Guiana: Au- blet (possibly at BM, not seen). Additional specimens examined. PANAMA. COLON: orenzo turnoff, 9940'N, (F, ). SAN BLAS: along Rio o Obaldia. ca. 8°40'N, 77?25'W, sea level, Mc e doña (F, MO). 3. Croizatia Croizatia Steyerm., Fieldiana, Bot. 28: 308, fig. 57. 1952. erm. TYPE: Croizatia neotropica Stey- Dioecious trees or shrubs; indumentum simple. Leaves alternate, petiolate; stipules persistent or deciduous; blades entire, pinnately veined, without embedded glands. Flowers in axillary clusters. Sta- minate flowers pedicellate; sepals 5, imbricate; petals 5, much shorter than sepals, pubescent; disk annular; stamens 5, free or connate, the anthers + introrse; pollen grains 3-colporate, the sexine echinate; pistillode 3-fid. Pistillate flowers pedi- cellate; sepals 5, imbricate; petals 5, much shorter than sepals, pubescent; disk annular; ovary pu- bescent; styles free, twice bifid; ovules paired in each locule, hemitropous. Fruits capsular; colu- mella distally expanded into 3 broad papery wings; seeds paired or solitary in each locule, smooth, not fleshy, ecarunculate; endosperm absent; cotyledons greenish, contortuplicate, much broader than and about as long as the radicle. LITERATURE WEBSTER, G. L., L. GILLESPIE & J. STEYERMARK. 1987. Systematic of Croizatia (Euphorbiaceae). Syst. Bot. r 1-B, The affinities of this small neotropical genus of three species have remained questionable because of fragmentary material. The recent discovery of staminate flowers of Croizatia naiguatensis Stey- erm. (Webster et al., 1987) has not made it te to determine the affinities of the genus m In the protologue to the elena! eS i Croizatia, Steyermark proposed a relationship to the Old World genus Actephila Blume on the basis of a suggestion by Dr. Leon Croizat. That sugges- tion seems very perceptive, as there is a clear resemblance to that genus in details of habit, flower, and fruit. On the basis of gross morphology, Croizatia can be maintained as a genus distinct from Actephila, especially by virtue of its pubes- cent petals and ovary and its twice-bifid styles. In this latter. character it is similar to the African Pentabrachium Muell. Arg.; however, in the Af- rican genus the seeds have abundant endosperm and the embryo is not contorted as in Croizatia. The echinate pollen grains of Croizatia are very different from those of Actephila or Pentabrach- ium and indicate a possible closer affinity to genera in subfamily Oldfieldioideae. 3.1. Croizatia panamensis Webster, Syst. Bot. 12: 7. 1987. TYPE: Panama. Panamá: primary forest along road from El Llano to Carti-Tu- pile, 300-500 m, 30 Mar. 1973, Liesner 1279 (holotype, MO; isotype, DAV). Shrub or small tree 1-6 m high, usually with a single main stem. Leaves with petioles 0.5-1 cm long, 3-4 mm thick; stipules + persistent, oblong- lanceolate, acuminate, ribbed, sericeous, 10-20 Volume 75, Number 3 1988 Webster & Huft Panamanian Euphorbiaceae 1093 mm long, 6-7 mm broad; blades chartaceous, gla- brous or sparsely strigose-hispidulous beneath, ob- ovate, abruptly short-acuminate, basally attenuate, 22-47 cm long, 5-15 cm broad, with ca. 15 arcuate-ascending lateral nerves connected by in- tramarginal loops, the veins and (to some extent) veinlets prominulous beneath. Staminate flowers with sparsely pubescent pedicels 3-4 mm long; sepals (4—)5, elliptic, entire, 1.7-2.5 mm long, 1- 1.5 mm broad; stamens 5, the filaments 2-4 mm long, connate at the base for 0.5-1.5 mm, the column long-pubescent; anthers 0.6-0.8 mm long; pistillode 1.5-2.5 mm long. Pistillate flowers with pubescent pedicels ca. 1.5 cm long, becoming 2.5- 3.5 cm long in fruit; sepals 5, elliptic-lanceolate, + acute, 8-12 mm long, 3-4 mm broad, hispid- ulous without, persistent and becoming reflexed in fruit; ovary 3.5-5 mm diam., densely hirsutulous; styles 3, 3-4 mm long, connate basally into a column ca. 1 mm high, three times bifid. Fruit capsular, 10—15 mm broad; columella ca. 8-9 mm high, 10-11 mm broad; seeds trigonous, smooth, brownish, 7.2-10 mm long, 5.3-6.5 mm broad. Rainforests, Panama and Colombia. This more complete species description of Croiz- atia panamensis has been made possible by re- cently collected flowering specimens and data pro- vided by Dr. Gordon McPherson. It is now apparent that C. panamensis is clearly different from C. naiguatensis in floral characters: staminate flowers with stamens connate in C. panamensis (free in C. naiguatensis), staminate petals more long-cil- iate and styles more divided in C. panamensis. Since flowering material of C. neotropica is still unknown, it remains difficult to assess its relation- ships with C. panamensis. Additional specimens examined. PANAMA. SAN BLAS: near road from El Llano to Carti, beyond Disp n along divide m to east, is N, 79°00'W, ca. 300 m (fr), McPherson 11037 V, MO), (fl, piso) 11040 (DAV, MO), (fl, ei 11041 (DAV, M 4. Richeria Richeria Vahl, Eclog. Amer. 1: 30, tab. 4. 1797. TYPE: Richeria grandis Vahl. Trees or shrubs; dioecious; indumentum simple or absent. Leaves alternate, petiolate; stipules de- ciduous; blades entire or distantly crenulate, pin- nately veined, sometimes with basal laminar glands. Inflorescences axillary, racemose or spicate; sta- minate flowers several per bract in sessile or pe- dunculate glomerules; pistillate bracts subtending solitary flowers; flowers apetalous. Staminate flow- ers sessile; calyx 3—5-lobed, the lobes imbricate; disk segments 3-5; stamens 3-6, free; filaments exserted from calyx; anthers introrse, + versatile, dehiscing longitudinally; connective not enlarged; pollen grains prolate, 3-colporate, semitectate, re- ticulate; pistillode present. Pistillate flowers ped- , the lobes imbricate; disk cupulate; ovary 3-locular, glabrous or pubescent; styles short, bifid; ovules 2 per locule, anatropous. Fruits capsular (somewhat fleshy and tardily de- icellate; calyx 3-5-lobe hiscent); columella slender, upwardly dilated, with papery wings; seeds solitary in each locule, ecarunculate, the outer testa fleshy; endosperm present; cotyledons broad, plane, basally cordate. A neotropical genus of five closely related species, previously unreported from mainland North Amer- ica. Richeria appears to be most closely related to the African genus Maesobotrya Benth. and to Aporusa Blume of southeastern Asia and Malaysia. The circumscription of the genus adopted here differs from that of Mueller (1866) and Jablonski (1967), since section Podocalyx (Klotzsch) Muell. Arg. (based on Richeria loranthoides (Klotzsch) Muell. Arg.) should be segregated as the monotypic genus Podocalyx Klotzsch, which, in fact (as in- dicated by the spinose pollen), belongs in the subfamily 'Oldfeldioideae rather than the Phyllan- thoideae. There are two species of Richeria in Panama, neither previously reported. KEY TO THE SPECIES OF RICHERIA IN PANAMA la. Carpels 3; capsules glabrous; styles suppressed (stigmas sessile); leaves obtuse or rounded a tip, glabrous or corr so; stipules m than 1 cm on g, lanceola R. obovata lb. duro 2; os distinctly pierden styles ca. 1.5 mm long; leaves acuminate at tip, pu- bescent abaxially; stipules over 1 cm long, fo- liaceous .2. R. dressleri 4.1. Richeria obovata (Muell. Arg.) Pax & K. offm., Pflanzenreich IV. 147. XV(Heft 81): 29. 1922; Jablonski, Mem. New York Bot. Gard. 17(1): 126 . 1967. Richeria grandis £ obovata Muell. Ms: n DC., Prodr. 15(2): 468. 1866; Fl. Bras. 11(2) 16. 1873. TYPE: "Brazil," Rio Casiquiari, Spruce 3526 (not seen). A species description is not offered here, since the Panamanian specimens are incomplete, and it is not possible to expand the description of Pax & Hoffmann. In the absence of flowers, it is not entirely certain that the Panamanian specimens belong with those cited by Jablonski from montane 1094 Annals of the Missouri Botanical Garden rain forests in the states of Bolivar and Amazonas, Venezuela. PANAMA. PANAMÁ: ca. 5-6 mi. N of El Llano, 1,300 ft., Gentry 5796 (GH, MO, SCZ). VERAGUAS: cloud forest, Cerro Tute, NW of Santa Fe, Mori & Kallunki 5264 (DAV, MO). Specimens examined. 4.2. Richeria dressleri Webster, sp. nov. TYPE: anama. Panama: Santa Rita Ridge, road to Estación Calibrar el Agua Clara, 9°22'N, 79?42-45'W, 1,000-1,500 ft, 26 June 1971, Webster & Dressler 16744 (holotype, DAV; isotype, MO). Species haec ab congeneribus differt stylis elongatis, capsulis 2-locularis; foliis acuminatis E integris, subtus puberulis; ovario sericeo-hispido Tree to 15 m high, 3.5 dm thick; twigs terete, mostly densely appressed-hirtellous when young, eventually glabrate; foliage evergreen. Leaves with hirtellous petioles 1.5-5 cm long; stipules lanceo- ate, 1-1.5 cm long, densely sericeous, caducous; blades chartaceous, obovate, mostly abruptly acu- minate, at base narrowly cuneate and decurrent on the petiole, 10-30 cm long, 4-14 cm broad; major veins mostly 10-12 on a side, straight, brochidodromous, the midrib saliently raised be- neath; secondaries archingly and irregularly sca- lariform; ultimate veinlets fine, scarcely prominu- lous; surface of blade above glabrous and flecked or pitted with minute colored spots, beneath bronze- colored and densely to sparsely hirtellous (becom- ing glabrate in age except along midrib and larger veins); margins entire, plane or recurved. Inflo- rescences spiciform; staminate spikes 1.5-5.5 cm long, pistillate spikes 1.5-7 cm long; axes densely irsutulous without. Staminate flowers sessile; ca- lyx deeply 4—5-lobed, densely hirsutulous without; calyx lobes oblong to obovate or suborbicular, un- equal, the larger ones 1.4-1.8 mm long, 1.2-1.5 mm broad, the smaller ones 1.2-1.5 mm long, ca. l mm broad; disk segments 5, erect, kasd prismatic, apically hirtellous, 0.2— m high; stamens 5(-6); filaments free, 2-3 mm des an- thers ellipsoid, ca. 0.4 mm long; pistillode cylin- drical, densely hirtellous, 1-1. high, 0.6- 0.9 mm broad. Pistillate flowers subsessile; sepals mostly 4(-5), elliptic, tomentulose outside, seri- ceous within, 1.5-2 mm long, 0.8-1.3 mm broad; disk entire, adnate, ciliate-margined, ca. 1.5 mm across; ovary of 2 carpels, sericeous; styles stout, 2- or 3-fid, 1.5-1.7 mm long. Capsules ellipsoid, reddish, smoothish (not venose), ca. 10-1 long, 6-9 mm broad; columella flattened, papery- winged, 11-12 mm long; seeds somewhat asym- metrically ovoid-ellipsoid, tapering to an obtuse beak, with reddish, fleshy ribbed-striate exotesta, 8.3-9 mm long, 4.8-5.3 mm broad. This species is sharply characterized within Richeria by its 2-carpellate densely sericeous gy- noecium with distinct styles; the staminate flowers are similar to those of Richeria grandis but differ in the more slender cylindrical pistillode. The acu- minate leaves often copiously hirtellous beneath and the large foliaceous stipules also appear dis- unctive. The collections from Costa Rica are mor- phologically divergent but may tentatively be grouped with the Panamanian plants. It seems ap- propriate to name this species in honor of the co- collector of the type specimen, Dr. Robert Dressler, formerly of the Smithsonian Tropical Research In- stitute, since he has made a significant contribution to our knowledge of Mesoamerican Euphorbiaceae through his many collections and his monographs of Pedilanthus (Dressler, 1957) and Euphorbia subgenus Poinsettia (Dressler, 1961). Additional imens examined. Costa RICA. HERE- DIA: Finca La Selva, the OTS Field Station on the Rio Puerto ao just E ofi its junction with the Rio Sarapiqui, 100-450 m, Grayum & Perry 1447 (F), Hammel 8689, 8833 x Jacobs 21 79 (F). PUNTARENAS: on road to radio and telecommunications tower 6 km N of Golfito, 300- 400 m, Utley & Utley 4902 un PANAMA. eee near El Valle de Antón, ca. 8°37'N, 80%07'W, ca. 550 m, lise uu 7616 (F). COLON: Santa Rita lumber road, 5 km E of Colon, Dressler & Williams 3968 (MO). PANAMA: rainforest along El Llano-Cartí road, 4.6-8.2 mi. N of Panamerican Highway, 350-450 m, D'Arcy 11496 (MO, dupl. at SCZ seen by M. Huft), Gentry 5076 (MO), Hammel 7350 (MO), Knapp 5929 (DAV, F, MO), McPherson 9959 (F, MO), Mori & Kallunki 5607 (DAV, MO). SAN BLAS: Cerro Brewster, 9°18'N, 79°16' W, 850 m, de Nevers et al. 5414 (F) 5. Hyeronima Hyeronima Allemào, Pl. Novas Brasil 1. 1848. TYPE: Hyeronima alchorneoides Allemao. A number of collections made in cloud forests in Panama indicate that there is at least one ad- taxonomically; the following revised treatment is highly tentative pending revisionary studies of this poorly understood genus. KEY TO THE SPECIES OF HYERONIMA IN PANAMA la. Stamens 4; pistillode slender, bifid, 0.7-0.8 mm high; leaves sparsely lepidote; blades mostly 10- 30 cm long with petioles of 3-9 cm; stipules Volume 75, Number 3 1988 Webster & Huft 1095 Panamanian Euphorbiaceae 5-15 mm long; ovary densely lepidote; endo- carp of fruit not over 3.5 mm lon to den sely lepidote be- glabrous to lepidote; endocarp of fruit at least 2: oblonga 5.1 Hyeronima laxiflora (Tul.) Muell. Arg., Linnaea 34: 67. 1865. Stilaginella laxiflora Tul., Ann. Sci. Nat. Bot. III, 15: 244. 1851. TYPE: Guyana: “British Guiana," Schom- burgk 879, Hostmann 391 (syntypes, P). The specimens originally cited under this name were correctly referred to H. laxiflora, which is apparently widespread in lowland rain forests in northern South America. All of the lowland pop- ulations of Hyeronima in Panama belong to this species. 5.2 Hyeronima oblonga (Tul.) Muell. Arg., Linnaea 34: 1865; in DC., Prodr. 15(2): 271. 1866. [OMM oblonga Tul. Ann. Sci. Nat. Bot. III, 15: 248. 1851. TYPE: Guyana: “British Guyana," Schomburgk 805 (P, not seen). di vetu benthamii Tul., Ann. Sci. Nat. Bot. III, 15: 247. 1851. H. oblonga (Tul.) Muell. Arg. var. ben- thamii ien Muell. Arg., Linnaea 34: 66. 1865. xico. Oaxaca: Hartweg 513 (P), Gal- eotti 7240 rs Hieronyma guatemalensis J. D. Smith, Bot. Gaz. (Craw- elis 54: 241. 1912. TYPE: Guatemala. Alta Verapaz: Tuerckheim 423, II 2228 (not seen). Tree to 10 m high; young twigs angled, densely lepidote (scales ca. 0.15-0.25 mm across). Leaves with petioles mostly 10-15 mm long; stipules ap- parently absent; blades mostly obovate, abruptly cuspidate or short-acuminate, cuneate at base, gen- erally 4-8 cm long, 2.5-5 cm broad; major veins ca. 5-7 on a side, divergent, straight, brochido- dromous; midrib and veins raised beneath and + hirsutulous, the veins and veinlets distinctly prom- inulous above (upper surface scabrous to the touch); lepidote scales on upper surface scattered to absent, a. 0.1-0.2 mm across, with reddish center, be- neath sparse to dense and overlapping, ca. 0.2- 5 mm across, with pale center (lower leaf sur- face much paler than upper). Panicles densely lepidote with whitish scales; lateral axes mostly 2— 4, the staminate ones 5-10 cm long, the pistillate ones ca. 1.5-2.5 cm long; bracts densely lepidote, acute, ca. 0.7-1 mm long. Staminate flowers with rigid stout pedicels ca. 0.4-1.2 mm long; calyx cupulate, shallowly 5-lobed, 1-1.4 mm high, densely lepidote; disk massive, 0.6-0.8 mm high, densely lepidote on top. Pistillate flowers subsessile (ped- icels equaling or shorter than the bracts); calyx cupulate, shallowly 5-lobed, densely lepidote, 1- 1.3 mm high; disk cupulate, subentire, glabrous, ca. 0.4-0.5 mm high; ovary ovoid, ca. 1.5 m high, glabrous or nearly so; stigmas punctiform. Fruits ellipsoid, acute at both ends, coarsely bul- late-rugose, 5-6 mm long (endocarp 4-5.5 mm long). Montane rainforests, Guatemala to Panama and South America. With some reluctance we are referring all of the high-elevation (cloud forest) populations of Hy- eronima in Panama to a single species. There is a striking amount of variation in pubescence, and the literature might lead one to recognize two, three, or even more species. Plants with densely lepidote leaves, pale inflorescence axes, and the ovary glabrous or nearly so could be referred to H. scabrida (Tul.) Muell. Arg., and plants with T lepidote leaves and densely lepidote ovary oblonga s. str. However, specimens from Darión in particular have the pale 1 axes of H. scabrida combined with the sparsely lepidote leaves of H. oblonga. Both “species” oc- cur in the vicinity of El Valle. Plants from the vicinity of Cerro Campana, divergent in having densely lepidote leaves (with prominulous venation above) and larger flowers, appear to match the descriptions of H. oblonga var. benthamii (Tul.) uell. Arg. However, it is not clear whether that variety can be satisfactorily delimited from other populations. Only critical field studies can establish whether the broad delimitation of H. oblonga kui here is correct. qE Pe fa specimens examined. PANAMA. BOCAS DEL TORO: between Criollo and Quebrada a Higueró ón Chiriquí Trail, Kirkbride & Duke 783 (MO). sai ea erro Hornitos, ca W of Gualaca, 2,238 m Mori & lten 7505, 7514 (DAV, MO); Cerro Pate Macho, 4 km of me Sytsma et al. 4868 (MO) COCLÉ: La km alle, 850 m, Mori e esa, 2. al. 6610 (DAV, MO); hill 3 E: E of El Valle, 2,500 ft., Hammel 4776 (MO); swampy area 5 mi. from El Valle, 3622 (DAV, MO) foothills of Cerro ). DARIÉN: Cerro Mali, Tacarcuna, 1,800- 14025 (DAV, MO). TA Cerro Campana, Webster & Breckon 16490 (DAV). VERAGUAS: 3-4 km W of Santa Fe, 2,500 ft., Nee 11315 (DAV, MO); summit of Cerro Arizona, N of Santa Fe, 4,700 ft., Hammel 4741 (MO); Cerro Tute, just W of Santa Fe, Knapp & Dressler 5390 (MO). 1096 Annals of the Missouri Botanical Garden 6. Drypetes Drypetes Vahl, Eclog. Amer. 3: 49. 1807. TYPE: Drypetes glauca Vahl. Trees or shrubs, dioecious; indumentum absent or of simple hairs. Leaves alternate, short-petiolate, stipulate; the blades often coriaceous, entire to serrate. /nflorescences axillary; flowers in axillary clusters, sometimes cauliflorous. Flowers apetal- ous; sepals usually 4 or 5, imbricate, deciduous. Staminate flowers sessile to pedicellate, with in- trastaminal disk; stamens mostly 4-5(-50), fila- ments free; anthers basifixed, extrorse to introrse; pollen grains tricolporate, reticulate; pistillode pres- ent or absent. Pistillate flowers pedicellate; disk cupuliform; ovary of 1 or 2 (rarely 3 or 4) carpels; styles obsolete or nearly so, dilated stigmas capping the ovary; ovules 2 in each locule, anatropous. Fruits indehiscent, + drupaceous, the exocarp fleshy or leathery, the endocarp crustaceous or bony; seeds usually solitary in each locule, eca- runculate, the testa smooth; endosperm copious; embryo straight, the cotyledons broad and flat A large circumtropical genus of about 150 species, best represented in the Old World; about 20 neotropical species have been described. The single Panamanian species was discovered on Barro Colorado Island shortly after the publication of our original treatment. LITERATURE WEBSTER, G. L. 1977. A new species of Drypetes (Euphorbiaceae) from Panama. Madroño 24: 65- 6.1. Drypetes standleyi Webster, Madroño 24: 65, fig. 1. 1977; Croat, Fl. Barro Colorado I. 529, fig. 321. 1978. TYPE: Panama. Canal Zone: Barro Colorado I., Armour Trail, Fos- ter & Croat 2307 (holotype, DAV; isotypes, DUKE, F, F neg. 62358, MO). Recent collections indicate that Drypetes stand- leyi may occur over a broad area in Panama. barren collection from the Burica Peninsula, Chi- riqui Province (Busey 602, MO) may possibly rep- resent D. standleyi, although it differs from the other collections in its stiffer leaves with a more prominent | reticulum. The species may also occur in Cos ica; a specimen from La Selva (Hartshorn 1009. DAV) resembles D. standleyi, although it is divergent in having more slender pistillate pedicels. Additional ra ecl examined. PANAMA. PANAMA: Barro Colorado I., Armour Trail, Foster & Croat 2308 (DAV); vicinity of Armour Trail, Croat 14843, 14849, 16516 (DAV, MO); S of Zetek 11, Foster 1122 (DAV DUKE, MO). COLÓN: Santa Rita lumber road, 9.4 kai from Transisthmian Highway, Dressler 3810 (MO). vE- RAGUAS: Alto Piedra Santa Fe, Lao € Maasola 480 (MO); Cerro Tute, Mori et al. 7541 (MO). 7. Margaritaria Margaritaria L. f., Suppl. Pl. 66. 1782. TYPE: Margaritaria nobilis L. f RECENT LITERATURE WEBSTER, G. L. . A revision of eC rH Uem ced | Arnold Arbor. 60: 403- 8. Phyllanthus Phyllanthus L., Sp. Pl. 981. 1753. LECTOTYPE: Phyllanthus niruri L. (chosen by Small in Britton & Brown, Ill. Fl. N. U.S. edition 2, 2: 453. 1913) RECENT LITERATURE BANCILHON, L. 1971. Contribution à l'étude taxono- e a genre | Phyllanthus (Euphorbiacées). Bois- siera 18: 1-81 8.11. Phyllanthus anisolobus Muell. Arg. in DC., 15(2): 382. 1866. TYPE: Peru: Pavón (holotype, G). Prodr. The collection from La Palma, Darién (Pittier 6600, US) mentioned with doubt in 1968 now appears to represent Phyllanthus anisolobus on the basis of its resemblance to the Darién specimens cited below. e specimens examined. PANAMA. BOCAS DEL Puerto Palenque, Kirkbride & ARIEN: Manené, Kirkbride & Bristan Santa Fe on road to Rio C alavébora: 500- 700 m, Hernández et al. 744 (F). 8.12. Phyllanthus gentryi Webster, sp. nov. TYPE: Panama. Darién: lower slopes of Cerro Pirre, 200-500 m, Gentry & Clewell 7017 (holotype, F, a neg. 62354; isotypes, DAV, MO). Figure cies haec aff. P. he ek paq sed. ab subsp. ju- 2 differt foliis seminibus cornifolio differt staminibus 3, ambibus differt disco non rugoso, Se en stylari breviore. Volume 75, Number 3 Webster & Huft 1097 1988 Panamanian Euphorbiaceae FIGURE 1. Phyllanthus gentryi.— a. Habit. —b. Columella of fruit, showing seed scars and persistent calyx.— c. Seed, dorsal view.—d. Seed, ventral view. Based on Gentry 4589. Illustration by Wan-Ling Peng. 1098 Annals of the Missouri Botanical Garden Shrub or small tree to 5 m high; monoecious; deciduous branchlets pinnatiform, at least 5 dm long, obtusely angled, brownish, minutely scabrid- ulous. Leaves with petioles 5-8 mm long; stipules cordate, acuminate, brownish, scarious, ca. mm long; blades chartaceous, elliptic-lanceolate, 12-28 cm long, 6-7.5 cm broad, acuminate, rounded to cuneate at base, glabrous on both faces. Cymules each of 1 pistillate and several staminate flowers. Staminate flowers with pedicels 15-20 mm long; sepals 5, broadly elliptic or oblong; 2.5- 3 mm long, 2.2-2.7 mm broad; disk entire, angled, not pitted, 1.8-2.2 mm across; stamens 3, the filaments connate into a column 0.7-0.8 mm high and ca. 0.5 mm broad; anthers suborbicular, flat- tened, dehiscing horizontally, 0.8-0.9 mm long and broad. Pistillate flowers with pedicels 8-13 mm long; sepals 5, elliptic-oblong, blunt, mostly 3-4 mm long, 2.5-3 mm broad; disk massive, angled, not pitted, 2.2-2.5 mm across; ovary smooth, of 3 carpels; styles nearly free, dilated, 1-1.2 mm long, 0.7-1 mm across. Capsules reddish, valves 12.5-13 mm long; columella massive, 4.5- long, 4.2-4.5 mm broad; seeds trigonous-umbo- nate, 6.5-7.1 mm long, 5.2-5.3 mm broad, smooth, with irregular wavy horizontal dark brown bands on a light brown background, the apex some- times with a small whitish caruncular outgrowth (ca. 0.5 mm across); hilum triangular, broad, ca. 3 mm long and broad. Additional specimens examined. PANAMA. DARIEN: trail up Cerro Pirre, Gentry 4589 (MO); razorback ridge on Cerro Pirre, Duke 6556 (MO); Serrania de Pirre, trail from Q. Perecingo to Cerro Pirre, ca. 10 km airline SSE of El Real, in subtropical moist-to-wet forest, 300-750 m, Reveal & Duke 4919 (MARY, MO); around Rancho rio, halfway up slope of Cerro Pirre from Piji Vasal, Folsom 6245 (F, MO); S of El Real on trail up Cerro Pirre, ca. 8°00'N, 77?45'W, 550-1,030 m, McPherson 7051 (F. MO). Phyllanthus gentryi is the first representative of subgenus Xylophylla (L.) Pers. discovered in Panama. It clearly belongs in section Asterandra (Klotzsch) Muell. Arg. by virtue of its confluent staminate disk and dilated styles, and it resembles P. juglandifolius Willd. in general aspect. Al- though it could be interpreted as a subspecies of P. juglandifolius, it is distinctive in its large seeds, smooth (nonpitted) disk, and nearly free styles. In order to accommodate this species in the Flora of Panama treatment, the key on p. 221 must be revised as follows: e. eae aa pinnatiform. e'. Plants (in Panama) herbaceous; leaves less than 3 cm long; seeds less than 3 mm long; pollen grains prolate, 3-4-colporate (subg. Phyllanthus) : vicus won leaves more than 3 cm long; eeds o m long; pollen ont globose, neers e Xyllophylla) ... 12. P. gentryi c c Subfamily I]. ACALYPHOIDEAE Asch. 9. Caperonia Caperonia A. St. Hil., Hist. Pl. Remarq. Bresil 244. 1826. LECTOTYPE: Caperonia castanei- folia (L.) A. St. Hil. (Croton castaneifolius L.) (chosen by Britton & Wilson, Bot. Porto Rico 6: 486. 1925). No additional Panamanian species have been discovered. However, examination of additional specimens indicates that we may not have gone far enough in reducing the taxa proposed under C. paludosa Klotzsch. It is extremely difficult to separate that species from C. castaneifolia (L.) A. St. Hil., and we now believe that our Panamanian specimens of C. paludosa probably represent only forms of that more wide-ranging species. However, the narrower leaves of plants referred to C. pal- udosa are distinctive, and further study in the field is required to establish whether that species concept can be upheld. 10. Argythamnia Argythamnia P. Browne, Civ. Nat. Hist. Jamaica 338. 1756. TYPE: Argythamnia candicans Sw. Alchorneopsis iiis Muell. Arg., Linnaea 34: 156. 65: : Alchorneopsis floribunda (Benth.) Muell. Arg. (Alchornea glandulosa var. ?floribunda Benth.). 11.1 Alchorneopsis floribunda (Benth.) Muell. Arg., Linnaea 34: 156. 1865; in DC., Prodr. 15(2): 765. 1866. Alchornea glan- dulosa Poeppig var. floribunda Benth., Hook- er's J. Bot. Kew Gard. Misc. 6: 331. 1854. TYPE: Brazil. Amazonas: Spruce 2681 (holo- type, K, not seen). The single collection of this species cited in the original treatment, from Darién, was the only Cen- tral American record known at the time and was likewise a range extension of over 800 miles. The specimens cited below extend the range to western Panama as well as to Costa Rica. Volume 75, Number 3 1988 Webster & Huft Panamanian Euphorbiaceae 1099 Additional specimens examined. COSTA RICA. CAR- TAGO: 24 km of Turrialba on hwy. to Limón, then E at Tres Equis on jeep road 1.5 km, 9%58'N, 83°34' W, 450-525 m, Liesner et al. 15354 (MO). HEREDIA: Finca La Selva, the OTS Field Station on the Rio ji Viejo just E of its junction with the Rio Sarapiqui, ca. 100 m, Hammel 9425, 11083 (F, MO); Istarú Farm, 1. Franca, 10?19'N, 83'34'W, Grayum et al. 3520 (F, MO). SAN JOSÉ: 2 km N of Dominical along CR 223, 40- 100 m, Utley & Utley 4938 (F). PANAMA. BOCAS DEL TORO: Cerro Pila de Arroz, along road to Chiriqui Grande, 10 road-mi from Continental Divide and 2 mi. along pipeline access road E of highway, ca. 8*55'N, 82?08'W, 350-500 m, McPherson 8750 (F). 12. Caryodendron Caryodendron Karsten, Fl. Columb. 1: 91, tab. O. TYPE: Caryodendron orinocense Karsten. 12.1. Caryodendron angustifolium Stan- dley, Publ. Field Columbian Mus., Bot. Ser. 4: 217. 1929. TYPE: Panama. Chiriqui: Pro- greso, Cooper & Slater 192 (holotype, F, F neg. 59913). The collection cited from Darién under this name in the original collection is now known to be Se- nefeldera testiculata Pittier (q.v.). No additional collections of Caryodendron angustifolium are known from Panama or elsewhere 13. Adenophaedra Adenophaedra (Muell. Arg.) Muell. Arg. in Mart., Fl. Bras. 11(2): 385. 1874. Bernardia sect. Adenophaedra Muell. Arg., Linnaea 34: 172. 1865; in DC., Prodr. 15(2): 918. 1866. TYPE: ?megalophylla Muell. Arg. — Adenophaedra megalophylla (Muell. Arg.) Muell. Arg Bernardia Dioecious trees and shrubs; indumentum of sim- ple trichomes. Leaves alternate, petiolate, stipu- late; blades pinnately veined, without embedded laminar glands, dende eria i axillary or often compound; terminal, spiciform, bracts eglandular, ies l pistillate or several staminate flowers. Staminate flowers pedicellate; calyx splitting into 3 valvate lobes at anthesis; petals and disk absent; stamens 2(—3); filaments short; anthers with enlarged connectives, dehiscing introrsely and longitudinally; pistillode absent. Pis- tillate flowers pedicellate; calyx lobes 6, biseriate, imbricate; petals absent; disk 3-lobed; ovary of 3 carpels; ovules 1 per locule; styles contracted into sessile stigmas. Fruits capsular, 3-lobed; seeds per locule, smooth, ecarunculate. LITERATURE CroizaT, L. Nomenclatural transfers and cor- 1 rections in the Euphorbiaceae. Trop. Woods 88: 30- 32. This poorly known genus, hitherto considered to be South American, includes only three species. Croizat (1946) reported Adenophaedra from Pan- ama on the basis of Adenophaedra woodsoniana; but in the original treatment (Webster & Burch, 1968: 278) it was pointed out that his original generic disposition (J. Arnold Arbor. 24: 167. 194 of this plant as Cleidion woodsonianum was cor- rect, although that species is now known to be synonymous with C. membranaceum Pax & K. Hoffm. (q.v.). Several recent collections of A. gran- difolia from Panama and Costa Rica, however, firmly establish the presence of Adenophaedra in southern Central America. In addition, it now ap- pears that the plant called Bernardia denticulata in the original treatment is actually 4. grandifolia. 13.1. Adenophaedra grandifolia (Klotzsch) Muell. Arg. in Mart., Fl. Bras. 11(2): 386. 1874. Pip - pin Klotzsch, London : 46. 1843 : not seen). Bernardia ?grandifolia (Klotzsch) Muell. Arg., Linnaea 34: 173. 1865; in DC., Prodr. 15(2): 918. 1866. Cleidion Em Standley, Publ. Field Columbian 5 : 29. Bernardia den inva (Standley) Webster, Ann. Missouri Bot. Gard. 00. 1967. TYPE: Panama. Bocas del Toro: Chiriqui Trail, Buena Vista Camp, 1,250 ft., Cooper 606 (holotype, F, F neg. 52608; isotypes, NY, Y). Shrub or small tree to 8 m; twigs smooth, red- dish, thinly puberulent, tardily glabrate. Leaves with petiole 3-6 mm long; stipules lanceolate, ca. ] mm long, ca. 1 mm broad, caducous; blades chartaceous, oblanceolate, acuminate at the tip, attenuate at the base, 12-35 cm long, 3-12 cm wide, 2.5-4.5 times as long as broad, glabrous, or sometimes thinly villous below on the principal veins, the secondary veins 7-9 per side, arcuate, prom- inent below, the tertiaries reticulate, prominulous; margins remotely denticulate. /nflorescences ax- illary, densely villous, the bracts deltate, 1-2 mm long, densely villous; staminate spikes slender, flex- uous, to 15 cm long, the glomerules widely spaced, 1100 Annals of the Missouri Botanical Garden to 25; pistillate spikes thicker, not flexuous, 5-12 cm long, with 4—7 solitary flowers. Staminate flow- ers to 12 per glomerule, early dehiscent, leaving persistent pedicels ca. | mm long; calyx lobes mem- branous, deltate, spreading at anthesis, ca. 0.5 mm long. Pistillate flowers not seen; pedicels at ma- turity ca. 4 mm long, reflexed, sericeous. Capsules depressed-globose, deeply 3-lobed, 6-8 mm high, 12-18 mm diam., sericeous, glabrate at maturity, the persistent calyx lobes ca. 2 mm long, deltate, sericeous; seeds subglobose long, ca. 0.7 mm diam., yellowish, aia . to ovol ca. mm Additional specimens examined. COSTA RICA. LIMÓN: 7 km SW of Bribri, 100-150 m, Gómez et al. 20422 (F); camino entre la finca de don Calixto Kiamble y el EN camino a Katsi, subiendo hasta el Cerro Kikir- en bata, Gómez et al. 23800 (F); camino de Fila Dimat sa de TS » — hasta Soki pasando por la ena Sha, Gó al. 23859 (F). PANAMA. BOCAS DEL TORO: along ud rom pine a to Chiriqui e ca. 8°45'N, 82°15'W, McPherson ) (F); Cerro Pila pe rroz pia roa d to eru ceni 10 road-miles from C ontinental Divide along pipeline ee roac ighway, ca. “go on 2°08'W, 350-500 m, Mc Pherson 8752 (F), 8775 (F, distributed as Cleidion woodsonianum). VERAGUAS: 5 mi. NW of Santa Fe, 700-1,200 m, Liesner 976 (GH, MO). ~ 14. Bernardia Bernardia Miller, Gard. Dict. abr. ed. 4, 28. 1754. LECTOTYPE: Bernardia carpinifolia Griseb. (see Buchheim, 1962) Following the reduction of Bernardia denticu- lata to the synonymy of Adenophaedra grandi- folia (q.v.), this genus is now represented in Pan- ama by a single species. LITERATURE apii G. 1960. Nomenklatorische und systema- che Bemerkungen über die Gattung Bernardia (Euphorbiaceae) Willdenowia 2: 291-318. —— ———. 1962. Uber die Typusart der Im Ber- me (Euphorbiaceae), Willdenowia 3: 217-220. KEY TO THE SPECIES OF ALCHORNEA IN PANAMA la. Pistillate spikes less than 25 beneath), with 2 o 2a. Leaves (ex iar goin bene L cm long; leaves less than 25 cm lon or more basal laminar glands; seeds (where eius. not over cept on sprout shoots) not over ca. 20 cm long, glabrate, the veinlet reticulum only moderately ath. 14.1. Bernardia macrophylla a J. Wash. Acad. Sci. 15: 103. 1925. TYPE: Pa- nama: Rio Tocumen, near sea level, P Jan. 1924, Standley 29389 (holotype, US). No Panamanian specimens of this species have been found in addition to those cited in the original treatment. The recent collection cited below ex- tends the range to Costa Rica. That specimen has somewhat more acuminate leaf apices than the Panamanian collections and was taken at a con- d higher altitude but seems otherwise iden- cal. As indicated in the original treatment, the ck relative of Bernardia macrophylla seems to be B. jacquiniana Muell. Arg. of Venezuela, but the latter differs in having retrorse rather than ascending pubescence on the stems, more promi- nent venation on the undersurface of the leaves, nd 9-12 vs. 14 stamens. Additional specimen examined. Costa RICA. PUN- TARENAS: foothills of the Cordillera de Talamanca, around Tres Colinas, 9%07'N, 83%04'W, 1,800-1,850 m, Da- vidse et al. 25611 (F). 15. Adelia Adelia L., Syst. Nat. ed. 10, 1298. 1759, nom. cons. TYPE: Adelia ricinella L. (typ. cons.). 16. Alchornea Alchornea Sw., Prodr. 98. 1788; Fl. Ind. Occ. 2: 1153. 1800. TYPE: Alchornea latifolia Sw. Several recent collections of Alchornea indicate that there are some additional taxa in Panama, but the material is still inadequate for a satisfactory treatment. The taxa that appear to be present may be treated as follows. LITERATURE Pax, F. & K. HoFFMANN. 1914. Euphorbiaceae — Aca- ly »heae — Mercurialinae. /n: A. Engler, Das - zenreich ne 2 VII(Heft 63): 1-473 (Alchornea, pp. 220-2 g (or else copiously stellate-pubescent es aaa ae abruptly cuspidate-acuminate; vein axils not barbate beneath; basal foliar l. A. lungs usually 2; staminate spikes unbr anched, mostly axillar costaricensis 3b. Leaves not abruptly cuspidate (or else coriaceous); vein axils often barbate beneath. 4a. Styles 5 Leaves coriaceous, usually 5-20 mm long, relatively slender. with 2-4 basal glands; spikes mostly cauliflorous. 6a. Leaves mostly 8-20 cm long, acuminate, with mostl or crenate-dentate; foliar glands mostly 2(-4); Sistillate sepals 2-2.8 m 5-8 main lateral veins, entire $i latifolia Volume 75, Number 3 1988 Webster & Huft Panamanian Euphorbiaceae 1101 6b. Leaves mostly 2-7 c 5b. Leaves chartaceous, usually 4b. Styles 3-7 mm glands 2-4; spikes axillary ong, thick; leaves coriaceous, pedes with 5-7 main lateral veins; foliar . A. m long, acute, with mostly 3 or 4 main lateral veins; foliar lands 2-4; pistillate sepals shorter than 2 mm lon 3. A. with 5-10 = glands, with 5-8 main lateral hip spikes mostly E pistillate xA 1-1.5 m . A. triplinervia glandulosa granero 2b. pri over 20 cm long, copiously pits -pubescent beneath, with and distinctly raised beneath; spikes cauliflorou lb. Pistillate aie mostly 25 l 16.1. Alchornea costaricensis Pax & K. offm., Pflanzenreich IV. 147. VII(Heft 63): 235. 1914. TYPE: Costa Rica: Palmar, Tonduz 6757 (not seen). Recent collections show that Alchornea costari- censis is not confined to western Panama; it extends into South America, where there is at least one record from Colombia (Choco: Ordóñez et al. 58, ). The plant described from Colombia by Croizat (Caldasia 2: 357. 1944) as A. umboensis may prove to be a form of A. costaricensis. Additional specimens examined. PANAMA. una vicinity of San Bartolo Limite, 11 mi. W of Puerto Ar- muelles, Croat 21973a (MO). cocLÉ: along Rio San od below its junction with Rio Tife, Hammel 3405 (MO). Quebrada Venado, Bristan Duke 11859, Lewis et al. 2208 (MO). PANAMÁ: Chiltepe, Holdridge 6471 (MO). SAN BLAS: Canagangi, forest up- stream of village, 9°24’N, 79?24'W, 100 m, de Nevers et al. 5720 (F 16.2. Alchornea latifolia Sw., Prodr. 98. 1788. TYPE: Jamaica: Swartz (not seen). Southern Mexico to Venezuela and Peru. Further botanical exploration has shown that Alchornea latifolia is widely distributed in Pana- ma, including Barro Colorado Island, whence it was correctly recorded by Croat (1978). On several peaks and ridges in central Panama occur montane forms that appear very different from typical A. latifolia of lowland Central America and the West Indies. For example, plants with entire leaves and unusually short petioles are found on Cerro Jefe and Santa Rita Ridge (e.g., Gentry & Dwyer 5536, Croat 15309). These specimens somewhat suggest the South American 4. pearcei Britton, but their relatively long petioles and short spikes bring them closer to A. latifolia. Specimens with very unusual narrowly obovate leaves have been collected on Santa Rita Ridge (e.g., Croat 13844, Duke 15264); 6. cm long or longer, cauliflorous; samana coriaceous, oblanceolate, ca. 25-40 ong, glabrescent beneath; basal foliar glands obscure or absent; seeds 9-11 mm long .... LI essed p A. cpu . Á. seculo piis however, since plants with this leaf form occur in the same area as plants with more typical leaves, it seems probable that they are merely local vari- ants. At present, it seems best to include all o these variants within 4. latifolia, but it must be admitted that the species so conceived has an ex- traordinary amplitude of foliar variation; until crit- ical studies in the field are made, the situation will remain unsatisfactory. Additional specimens examined. PANAMA. CHIRIQUÍ: San Félix, Croat 33416 (MO). PANAMÁ: Cerro Campana, Croat 14673, Duke 10742, Sullivan 434 (MO), Méndez 19, 49 (F); Cerro Jefe, Dwyer et al. 5048, 5049, Gentry 4938 (MO), Gentry & Dwyer 5536 (GH, MO), Webster & Dressler 16454 (DAV); between Cerro Jefe and Cerro Azul, Tyson et al. 4325, Mori et al. 6543 (MO); Cerro Azul, Dwyer 5042 uo Fn E Holdridge 33 (DAV, MO), Stimson et al. 515 H, MO), Tyson & Blum 4081 (MO); N of El Llano, [ne 5105 (MO); El Llano- Carti road, 7.8-8.6 mi. from Pan-American Hwy., Fol- som 3572, Mori & Kallunki 6405 (MO) 16.3. Alchornea triplinervia (Sprengel) Muell. Arg. in DC., Prodr. 15(2): 909. 1866; Pax & K. Hoffm., Pflanzenreich IV. 147. VII (Heft 63): 227. 1914. Antidesma triplinervium Sprengel, Neue Entdeck. 2: 116. 1821. TYPE: Brazil. Rio de Janeiro: Serra do Mar, Gardner 617 (neotype, G; chosen here). This species is tentatively added to the Pana- manian flora on the basis of two recent collections from a single locality. The collection of Knapp has only staminate flowers, but the small coriaceous leaves with only three or four main veins and the cauliflorous spikes match those of Alchornea trip- linervia better than those of any of the species of Alchornea previously known from Panama. Another collection that may represent 4. triplinervia is Hammel 7252 (MO) from Cerro Sapo, Darién; this has much larger leaves and somewhat resembles some of the aberrant forms here treated as latifolia; for the present, its assignment must be regarded as dubious. The collections from Coclé do not fit any of the varieties recognized by Pax & Hoffmann (1914: 228-230), but the variation within A. triplinervia 1102 Annals of the Missouri Botanical Garden has not yet been critically studied, and it would certainly be premature to assign the Panamanian material to a new variety. The typification of Al- chornea triplinervia requires some comment, since Sprengel apparently left no type specimen. Mueller (1866: 909) designated what may be regarded as the typical element of the species as Alchornea triplinervia var. genuina forma psilorhachis. Since there is a good microfiche image (G, Prodromus Herb.) of Gardner 617, probably from the general area of the collection that was available to Sprengel, it seems appropriate to designate that as neotype. Specimens examin iin PANAMA. COCLÉ: hills N of El Valle, E slope o o Gaital, 900-1,000 m, Knapp 5351 (MO), FS din 11242, 11260 (MO) 16.4. Alchornea glandulosa Poeppig var. pittieri (Pax) Pax, Pflanzenreich IV. 147. VII(Heft 63): 235. 1914. Alchornea pittieri Pax, Bot. Jahrb. Syst. 33: 291. 1903. TYPE: Costa Rica: Canas Gordas, Pittier 11101 (is- otype, US) Since the Darién collection of Terry & Terry was reported in our original treatment, a number of additional specimens have accumulated; these confirm the widespread occurrence of Alchornea glandulosa in montane forests of Panama. Ex- amination of this expanded suite of specimens now shows that the Panamanian plants represent var. pittieri, originally described from Costa Rica. This variety is very similar to var. glandulosa of the upper Amazon but differs in the smaller glandular spots at the base of the leaf (mostly 0.5 mm long or less in var. pittieri, reaching 1-1.5 mm long in var. glandulosa). At present, var. pittieri is known only from Costa Rica, Panama, and adja- cent Colombia (Choco). Additional specimens examined. PANAMA. BOCAS DEL TORO: headwaters of Rio Mali, between Q. Gace and La Zorra, Kirkbride & Duke 726 (MO); along pipeline near end of road, ca. , 82°15'W, 900-950 m, McPherson 8691, 8699 (F). DARIEN: Cana-Cuasí Trail, Terry & Terry 1575 (MO); Cerro Pirre, Bristan 620 (MO); Alto de Nique, Cerro Pirre massif, 1,300-1,520 m, Gentry et al. 28647 (DAV, ); Cerro Tacarcuna, lower montane wet forest, 1,500 m, Gentry & Mori 13793 (DAV, F, MO, PMA). vE- RAGUAS: lower montane wet forest, 7 km W of Santa Fe, a. 900 m, Nee 11183 (MO, US); NW of Santa Fe, 2.8 m from Escuela Agricola, Alto de Piedra, Mori & Kal- lunki 6219 (MO). 16.5. Alchornea grandiflora Muell. Arg., Linnaea 34: 170. 1865; in DC., Prodr. 15(2): 907. 1866. SYNTYPES: Venezuela: Fendler 1272 (G), Moritz 1497a (C). Costa Rica: Hoffman 530 (G, not seen). Although it has been confused with 4. glan- dulosa, the relatively short thick styles and stiff glandular leaves distinguish A. grandiflora from A. glandulosa and from A. latifolia. Mueller (loc. cit.) reported A. grandiflora from Costa Rica and Venezuela, so its occurrence in Panama is not dns peci m m ined. PANAMA. CHIRIQUÍ: Cerro Pd 1,690 m, Croat 37195 (MO). DARIEN: Cerro Tacarcuna, ds. forest, 1,800-1,850 m, Gentry & Mori 13995 (DAV, F, MO) 16.6. Alchornea grandis Benth., Bot. Voy. Sulphur 164. 1844. TYPE: Colombia. Naririo: Tumaco, Barclay & Hinds (K, not seen). The specimens cited below, and several collec- tions from Chocó Province, Colombia (Fernández 206, Killip & Cuatrecasas 39076, both UC) fur- nish previously unknown characters for the sta- minate plant: staminate spikes mostly compound, 4-12 cm long, with 1-6 lateral axes; staminate calyx glabrous, sepals ca. 1.2 mm long; stamens 8, anthers 0.7-0.8 mm long, blunt. Specimens iy cie PANAMA. VERAGUAS: Isla de Coiba, road from Cam ento Juncal to Colonia Penal, Antonio 2417 (MO); Playa Hermosa, Antonio 2342 (MO). 16.7. Alchornea megalophylla Muell. Arg., Flora 47: 343 64; in DC., Prodr. 15(2): 911. 1866. TYPE: Colombia. Antioquia: Pur- die (K, not seen). Tree to ca. 10 m high; trunk 1.5 dm diam.; twigs subterete, smooth, glabrous. Leaves with stout petioles 0.5-1.5 cm long, glabrous, plicate; stipules inconspicuous, ca. 1.5 mm long or shorter, dark, triangular, pubescent; blades becoming subcoria- ceous or coriaceous, elliptic to obovate, abruptly short acuminate at apex (the acumen ca. 1-2 cm long), ca. 25-40 cm long, 7-15 cm broad, glabrous or glabrescent (minute scattered stellate hairs on underside of lamina or confined to midrib); basal foliar glands obscure or absent; major lateral veins ca. 10-15 on a side, straight, ascending, raised beneath, connected by ladderlike prominulous vein- lets, contracted to an obtuse base; margins distantly crenulate-dentate (ca. 10—15 glandular teeth on a side), apex abruptly short-acuminate (acumen ca. -2 cm long). Spikes cauliflorous, pendulous, stel- late-pubescent; staminate spikes not seen; pistillate spikes ca. 50-75 cm long, with ca. 20-30 flowers. Staminate flowers not seen. Pistillate flowers sub- sessile; calyx ca. 3.5 cm broad, 4-lobed, pubescent; ovary copiously pubescent with minute stellate hairs; styles slender, unlobed, ca. 20-25 mm long, Volume 75, Number 3 1988 Webster & Huft Panamanian Euphorbiaceae 1103 basally connate for 2-4 mm, basally stellate, api- cally smooth and long-attenuate. Capsules reddis brown, stellate-pubescent, not seen entire; seeds elliptic, plump, pale brown, coarsely tuberculate, 9-11 mm long. Rainforests, Panama and Colombia. This striking species stands out from all other Panamanian taxa by virtue of its long, pendulous, cauliflorous inflorescences and its large, coriaceous, more or less oblanceolate leaves. It resembles A. grandis in a number of respects but differs in leaf shape and sparseness of the laminar pubescence. Specimens examined. PANAMA. DARIEN: La Laguna, ridge between Pucuro and Tapalisa rivers, 820-840 m, Gentry & Mori 13560 (DAV, MO); top of Cerro Mali, 1,400 m, Gentry & Mori 13693 (DAV, MO); Cerro Tacarcuna, Gentry & Mori 13938 (MO); Alturas de Nique, S of El Real, 900-1,250 m, McPherson 11614 MO). 17. Cleidion Cleidion Blume, Bijdr. Fl. Ned. Ind. 612. 1826. TYPE: Cleidion javanicum Blume. Recent collections in Panama and further study of the South American species have greatly altered the picture of Cleidion in Panama. Largely as a result of problems encountered in the preparation of this account, the junior author has undertaken a revisionary study of the neotropical species of Cleidion, and until its completion, some of the conclusions expressed here must remain tentative. n addition to the two species treated here, a recent collection in Darien by Dr. Gordon Mc- Pherson may belong to Cleidion prealtum Croizat (J. Arnold Arbor. 24: 167. 1943), a species oth- erwise known only from material collected in the basin of the upper Rio Madeira in Amazonian Bra- zil. The poorly known Polyandra bracteosa Leal (Arch. Jard. Bot. Rio de Janeiro 11: 64. 1951), described from staminate material, also from the Rio Madeira, now appears to be synonymous with C. prealtum. LITERATURE VAN DER WERFF, H. & A. R. SmitH. 1980. Pterido- phytes of the State of Falcon, Venezuela. Opera Bot. 56: 1-34. KEY TO THE SPECIES OF CLEIDION IN PANAMA la. Fruiting RE PE LM )8-18 cm long; cap- sules 3-6, r m in diameter; leaves up to 13 cm i Des vein axils barbate be- ES veins 7 or 8 on a side; staminate thyrses a. l cm | 1. C. membranaceum lb. Fruiting spikes thick, 2-3 cm long; capsules 1- 3, 13- m in diameter; leaves over 15 cm long; leaf vein axils not barbate beneath; veins 8-10 on a side; staminate thyrses 5-9 cm lon 2. C. castaneifolium 17.1. Cleidion membranaceum Pax & K. gler, Pflanzenreich IV. 147. XIV(Heft. 68) 23. 1919. TYPE: Venezuela. Lara: around Palmosola, in forest along Río Aroa, near sea level, 26-28 June 1913, Pit- tier 6375 (US, photo F neg. 44609). Cleidion woodsonianum Croizat, J. Arnold Arbor. 24: 167. 1943. TYPE: Panama. Panamá: vicinity of Sal- amanca Hydrographic Station, Rio Pequeni, ca. 80 , Woodson et al. 1587 (holotype, p. isotypes, F, F neg. 62417, MO, F neg. 62356, NY). There appear to be no differences between the Panamanian plants and the Venezuelan collections of Cleidion membranaceum. The three known Venezuelan collections are all from a restricted area near the junction of the provinces of Falcon, Ya- racuy, and Lara. No substrate is indicated on the labels of these collections, but it is known that much of this area is underlain by limestone (van der Werff & Smith, 1980), which is also true of the Pana- manian collections. The recent disjunct collection from Peru strengthens the probability that this species is restricted to limestone, and that this accounts for the peculiar disjunctions in its range. Further study may show that Cleidion mem- branaceum is synonymous with C. tricoccum (Ca- sar.) Baillon, of eastern Brazil from Bahia to Sao Paulo, which has leaves similar in shape and size to those of the Panamanian plant, similar long, nearly filiform pistillate inflorescences, and similar capsules. Additional E examined. PANAMA. PANAMA: wth on limestone hillsides, along Choco Reserva Forestal “Rio ampto. *Canelon," via Tucacas, Tocuyo,” a 4 km del Blanco 895 (MO). 17.2. Cleidion castaneifolium Muell. Arg., Linnaea 34: 184. 1865. TYPE: Peru: Pavon (holotype, G, F neg. 7159). Lig ve dpt Standley, Carnegie Inst. Wash. 461 (Botany of the Maya Area 4): 66. 1935. ) Croizat, J. Arnold 24: 166. 1943. TYPE: Guatemala. Petén: eidion Arbor. 1104 Annals of the Missouri Botanical Garden Camp 35, boundary with Belize, 750 m, Schipp 5-279 (holotype, F, F neg. 52594). Trees to 10 m high; dioecious; twigs glabrous. Leaves with petioles 1.5-4 cm long, glabrous or sparsely pubescent with short, white, appressed hairs, swollen toward apex; stipules not evident; blades membranous or thinly chartaceous, elliptic, abruptly caudate-acuminate at tip, acute or cu- neate at the somewhat inequilateral base, 15-26 cm long, 6.5-9.5 c road, minutely pustulate, glabrous, the veins 8-10 on a side; margins shal- lowly dentate, the teeth callose, 15-28 on a side. Inflorescences unisexual, axillary; pistillate ra- cemes to 16 cm long, widely divergent from the stem, the rachis glabrous, or puberulent toward the apex, with ca. 3 flowers occurring singly (3 fruits on only complete pistillate raceme seen); staminate thyrses 5-9 cm long, with 7-12 flowers crowded at each of the 20-50 nodes, the rachis densely puberulent. Staminate flowers on pedicels to 1 mm long; calyx lobes cucullate, reflexed, 1.5-2 mm long, glabrous, ovate, the apex acute or acu- -80. Pistillate flowers not seen; bracts narrowly lanceolate, rigid, divergent, minate; stamens ca. ca. 2 mm long, puberulent; fruiting pedicels 10- 12 mm long, slightly clavate, puberulous, jointed; calyx lobes (in fruit) 3-5, somewhat reflexed, del- tate, 2-3 mm long, acute, canescent below, ciliate on the margins; styles (persistent on mature fruits) 7-12 mm long, deeply bifid, densely strigose. Cap- sules 3-locular (of which often only 2 fully devel- oped), deeply lobed, dorsally carinate, ca. 1 cm high, 14-18 mm diam., densely puberulent, drying black; columella 6-8 mm long, trigonous, narrowly winged, the seed scars elongate, conspicuous; seeds globose, smooth, not beaked, ca. 9 mm long, mot- tled light and dark brown. Rainforests, southern Mexico to Panama, Ec- uador. The Panamanian collections of Cleidion cas- taneifolium match perfectly the type photograph as well as the original description except for the curious statement in the latter that the capsule is six-lobed. Since all other species of Cleidion have three-lobed capsules (as do the Panamanian col- lections), and since the one capsule on the type photo appears crushed and misshapen (and thus made to appear six-lobed), it seems that Mueller was merely careless in his description. Since these collections are in perfect agreement with all other distinguishing characteristics of this species (large elliptic-ovate leaves, petioles 3-5 cm long and gla- brous except for the subpuberulous tips, long un- divided pistillate racemes, and large capsules with dorsally carinate lobes), their identification seems certain. The only other Panamanian species of the genus, C. membranaceum, and the Peruvian C. amazonicum Pax both differ from C. castaneifo- lium in their smaller leaves, much shorter petioles 2-5 mm long), and smaller capsules. Cleidion prealtum Croizat of Amazonian Brazil differs from all of these in its obovate and coriaceous leaves. Cleidion castaneifolium was described from _ "Peru," but a possible isotype sheet at F (Ruiz & Pavón s.n.) is labeled as having been collected at Guayaquil, Ecuador (another Ruiz & Pavón spec- imen at F has no locality data). The species is not definitely known from Peru, but there are two modern collections from between Santo Ee and Quinindé in Esmeraldas Province, Ecua (Acosta Solis 13649, Little 6196, both at F, "n distributed as Alchornea). There appear to be no salient differences sep- arating Cleidion castaneifolium from Panama and South America from the Mexican and Central American populations that have long been referred to C. oblongifolium. In addition to the common characters given in the key, these populations all frequently exhibit a characteristic purplish cast to the leaves. Specimens examined. PANAMA. DARIEN: silos de Pirre, on the NW slope of the mountain range dominated by Cerro Pirre, along Q. Perecingo (Parasénico), a spas utary to Rio Pirre, ca. 10 air iig S of El Real, 8°03'N, 77°43'’W, Reveal & Duke 4875 (MARY, MO): 2-3 mi. SE of Pijibasal on Rio Perasénico, ca. 9-10 mi. S of El Real, Hartman 12038 (F, MO). 18. Ricinus Ricinus L., Sp. Pl. 1007. 1753. TYPE: Ricinus communis 19. 4calypha Acalypha L., Sp. Pl. 1003. 1753. LECTOTYPE: Acalypha virginica L. (chosen by Small in Britton € Brown, Ill. Fl. N. U.S. ed. 2, 2: 457. 1913 19.10. Acalypha cuneata Poeppig in Poeppig & Endl., Nov. Gen. Sp. Pl. 3: 22. 1841. TYPE: Peru. Maius Yurimaguas, Poeppig (not seen). Acalypha obovata Benth. in Seemann, Bot. Voy. Sulphur 163. 4. TYPE: Ecuador. Esmeraldas: Atacames, Hinds (BM, not seen). 4. cuneata Poeppig var. obovata (Benth.) Muell. Arg. in DC., Prodr. 15(2): . 1866. Volume 75, Number 3 1988 Webster & Huft Panamanian Euphorbiaceae 1105 Shrub or small tree 2-5(-8) m high; monoe- cious; stems nearly glabrous. Leaves with petioles 1-7 cm long, glabrous; stipules lanceolate, 4—7 blades obovate or obovate-oblong, cuspidate-acuminate at the tip (the acumen 1.5-3 cm long), acute at the base, 15-30 cm long, 5-13 cm broad, 2-3.2 times as long as broad, glabrous, pinnately veined, the sec- mm long, strongly keeled, caducous; ondary veins 11-15 per side, arcuate, prominent above and below, connected by a prominulous re- ticulum; margins shallowly crenate-denticulate. /n- florescences axillary, spicate, unisexual; staminate spikes to 15 cm long, densely flowered, densely 2-4 and sessile on a slender rachis; pistillate spikes 7-15 cm long, 4-7 mm thick, loosely flowered with 15- 50 bracts, the rachis glabrous to densely puberu- lent. Pistillate flowers solitary; bracts + reniform, 3-4 mm long, 6-7 mm wide, inconspicuously 8 10-lobed, each lobe with a short tuft of bristles, the bracts otherwise glabrous or lightly short-stri- gose; calyx lobes obscure; ovary densely hispid, the styles free, lightly strigose, pinnatifid along entire length into puberulent, solitary and pedunculate or 8-12 narrow segments. Capsules -5 mm diam., hispid, verrucose; seeds obovoid- ellipsoid, 3-3.5 mm long, ca. smooth, brown, the caruncle nearly obsolete. m diam., This is a widespread species of lowland rainfor- ests in northern South America, and its discovery in eastern Panama is not surprising. It is easily recognized by the long-petiolate obovate leaves with pinnate venation and axillary pistillate inflores- cences. In order to accommodate this species in the Flora of Panama treatment, the key on p. 300 must be revised as follows: ee. Leaves pinnately TUE ante bracts sub- entire or shallowly den e'. Spikes mostly omy 3-8 cm long, with 1 or 2 pistillate bracts at base, these sub- tending 2 or 3 flowers 0. z A. diversifolia : iy unisexual, to 15 cm long, the pis tillate ones toward the apex of the branch, the staminate below; pes bracts sub- tending a single flow A. cuneata o a Specimen examined. PANAMA. DARIÉN: S of El Real on trail to Cerro Pirre, disturbed forest along Cerro Per- recénega, ca. 8%00'N, 77°45'W, ca. 50 m, McPherson 6977 (F). 20. Plukenetia Plukenetia L., Sp. Pl. 1192. 1753. TYPE: Plu- kenetia volubilis L. The discovery of a second species of Plukenetia in Panama makes it necessary to provide the fol- lowing key. KEY TO THE SPECIES OF PLUKENETIA IN PANAMA la. Leaves palmately veined; stylar column cylin- HP ue slender, 10-25 mm long; c 3.5 mm broad 1. P. volubilis lb. Per pinnately veined; stylar column obovoid, 2 mm long; capsules to 1.5 mm broad ......... 2. P. penninervia 20.1. Plukenetia volubilis L., Sp. Pl. 1192. 1753. TYPE: West Indies, Plumier (perhaps at BM). This species is more widespread in eastern Pan- ama than was indicated by the single collection cited in the original treatment. These new collec- tions also confirm the identity of the Panamanian species with Plukenetia volubilis of the Antilles and South America. Recent collections have also extended the range to Costa Rica (Gómez et al. 191841, F), Nicaragua (White 5323, F), and Ve- racruz, Mexico (Calzada 1034, F Additional specimens examined. PANAMA. COLÓN: upstream from bridge over Rio Guanche, 0-100 m, An- tonio 3351 (F, MO); near Portobelo, Croat 12969 (MO); Portobelo, along stream running into Rio Buena Ventura, S of Portobelo, 0-10 m, Foster 2060 (F, MO); along Rio Guanche, 6 km S of Portobelo, 0-10 m, Nee & Gentry 8686 (MO). DARIEN: Cerro Pirre, valley between Pirre ae next most piae | pea ak, Folsom 4426 (F, MO). PANAMÁ: 4-5 hours walk upriver from Torti Arriba, 200- 300 m m, Folsom et al. 6845 (F, MO). 20.2. Plukenetia penninervia Muell. Arg., Linnaea 34: 158. 1864; in DC., Prodr. 15(2): 770. 1866. TYPE: Venezuela, near Biscaina, Fendler 2412 b ds G, not seen; photo F neg. no. 7110). Plukenetia — pri era a pr Columbian Mus., Bot. Ser. 4: 314 TYPE: Honduras. Atlántida: Lancet. Sol 4 Mar. 1928, Standley 56708 (holotype, F, F neg. 52742). Liana; twigs spreading-puberulent, glabrescent. Leaves with petioles ca. 1 cm lo stipules brownish, glabrous, rigid, deltate-lanceo- late, 1-1.7 mm long; blades chartaceous-oblong, oblong-elliptic, or oblong-lanceolate, acute to acu- minate at tip, abruptly acuminate-truncate at base, 5-10 cm long, 2-4.5 cm broad, glabrous and shining above, glabrate or with a few hairs along the nerves and paler below, with 2 prominent glands above at the base, often with 1—3 pairs of smaller ng, puberulent; ones in a row above them, pinnately veined, the midrib and secondary veins (6-11 on a side) prom- inent below; margins shallowly crenate-denticulate. Inflorescences axillary, bisexual or staminate, 0.5— 3 cm long; pistillate flowers solitary at lower nodes of bisexual inflorescences, the staminate flowers 1106 Annals of the Missouri Botanical Garden few at the distal nodes. Staminate flowers with pedicels short-pilose, 4—7 mm long; calyx segments generally 3, obovate, acute, 1.2-1.6 mm long; receptacle cylindrical, 1.6-1.8 mm high; disk ob- solete; stamens ca. 18-25, inserted spirally on the receptacle, the filaments ca. 0.1 mm long, the anthers 0.1-0.2 mm long. Pistillate flowers with pedicels becoming 12-20 mm long, these narrowly clavate, strigose when young, glabrate to sparsely short-pubescent at maturity; calyx lobes lanceolate, -1.2 mm long, ca. 0.5 mm broad, strigose in a band along the center; ovary of 4 carinate carpels, strigose on the keels, otherwise glabrous, the stylar column to 2 mm long, obovoid, the stigmas thick, unlobed. Capsules deeply 4-lobed, oblate, to ca. 1 cm high, 1.5 cm broad, the cocci thick and rigid; seeds subglobose, only slightly compressed later- ally, reticulate-venose, brownish mottled, ca. 5 mm long, 3-4 mm thick. Lowland evergreen rainforests, Mexico (Oaxaca, Yucatan Peninsula) to Colombia and Northern Bra- zil (Para); here reported from Panama for the first time. The discovery of Plukenetia penninervia in Panama is not surprising, although it is still un- known in Costa Rica, and the reason for its ap- parent scarcity in southern Central America is not clear. e populations from northern Central America, where the species is much better known, were described by Standley as P. angustifolia, but no salient differences between these plants and those of South America are apparent. Specimens examined. PANAMA. COLÓN: Santa Rita Ridge, x ravine Mise near Agua Clara rainfall station, 400-500 m er HA e 2222 ii ); Santa Rita Ridge c "9°20'N W, ca. 500 asa 8461 (P. PANAMÁ: de fa mi. S of Pan piti Highwa 3.0 mi. E of Canazas checkpoint, foothills of Serranía d Cañazas, 8%52'N, 78°15'W, 0-50 m, Knapp 3887 (F, MO). 21. Acidoton Acidoton Sw., Prodr. 84. 1788. TYPE: Acidoton Sw urens 21.1 Acidoton nicaraguensis (Hemsley) Webster, Ann. Missouri Bot. Gard. 54: 191. 1967. Cleidion ?nicaraguensis Hemsley, Biol. Cent.-Amer., Bot. 3: 130. 1883. TYPE: Nic- aragua. Chontales: Tate 352, 455 (syntypes, presumably K, not seen). Additional specimens examined. PANAMA. COLON: Santa Rita, Correa & Dressler 607 (MO); Santa Rita Ridge Rd., 20 km from Transisthmian Hwy., 9%24'N 79°39'W, Sytsma 1117 (F, MO). PANAMA: along the El Llano-Carti rd., ca. 10 mi. N of Pan Am Hwy., 500 m, Gentry et al. 8878 (MO). SAN BLAS: El Llano-Carti road, km 26.5, 9?19'N, 78*55'W, 300-400 m, de Nevers et al. 5292 (F); Nusagandi, along trail to Quebrada de Nu- sagandi, van der Werff 7029 (F). 22. Tragia Tragia L., Sp. Pl. 980. 1753. LECTOTYPE: 7ragia volubilis L. (chosen by Small in Britton & Brown, Ill. Fl. N. U.S. ed. 2, 2: 458. 1913). The discovery of three additional species in Pan- ama makes it necessary to provide a key to the Panamanian species. LITERATURE Pax, F. & K. HOFFMANN. 1919. Mu d m OM cd lypheae — Plukenetiinae. /n: A. Engler, Das n- zenreich IV. 147. IX (Heft 68): 1-108 (Tragia, pp. KEY TO THE SPECIES OF TRAGIA IN PANAMA la. Leaf blades 12-25 cm long, 11- T cm broad, often 3-lobed; stamens ca. 40 ... T. OMM m long, less le 10 e broad, unlobed; stamens 2-12. a. Er aia bifurcate, the pistillate flow- s 5-10 on lower branch; stamens 2 . T. fendleri 2b. Inflorescence racemose, not bifurcate; pis- tillate flowers solitary at basal node; sta- mens 2 or 3. 3a. ded flower long-pedicellate; sty T. nm 3b. Pistillate flower subsessile; styles free | correae 22.1. Tragia bailloniana Muell. Arg., Linnaea 34: 178. 1865; in DC., Prodr. 15(2): 927. 1866. TYPE: Mexico. Tabasco: Teapa, Linden (P, not seen). Zuckertia cordata Baill., Etud. Euphorb. 496, pl. 4, 1858, not Tragia cor- data Michx., 1803 Twining vine; stamens and foliage + densely covered with stinging hairs. Leaves with petioles 8-14 cm long; stipules ovate-lanceolate, acumi- nate, greenish, 7-10 mm long; blades membra- nous, broadly ovate, unlobed, with a single lateral lobe, or shallowly 3-lobed, acuminate or caudate at tip, deeply cordate at base, 12-25 cm long, 11-18 cm wide, sparsely beset above and below with stinging hairs, usually 7-veined at the base, the margins remotely denticulate. Inflorescences opposite the leaves, bifurcate; peduncle 3-10 cm long; staminate branch to 20 cm long, many-flow- ered; pistillate branch 15-25 cm long, 7-15-flow- ered. Staminate flowers 1-3 per bract; bracts Volume 75, Number 3 Webster & Huft 1107 Panamanian Euphorbiaceae foliaceous, ovate-lanceolate, acuminate, 3-5 mm long, reflexed; pedicels 8-10 mm long, divergent, glabrate; sepals 5, linear-lanceolate, acuminate, ca. 6 mm long; stamens ca. 40; buds pyriform, acute. Pistillate flowers solitary in the axil of each bract; bracts similar to those of the staminate flowers; pedicels 1—4 mm long, hirsute; sepals deltate to lanceolate, acute to acuminate, 4—5 mm long; mar- gins ciliate with long, stiff hairs; ovary densely hirsute with stiff hairs ca. 1 mm long; styles black, 6-8 mm long, fused ca. 7, their length, the style branches slightly spreading. Capsules deeply 3-lobed (one lobe sometimes abortive), ca. 15 mm diam., ca. 8 mm high, densely hirsute with stiff hairs; columella 6-7 mm long, with 3 prominent, narrow wings at tip; seeds nearly globose, ca. 6 mm diam., smooth, with 8 or 9 light longitudinal striations. Forests, southern Mexico to western Panama. This distinctive species is the only member of Tragia section Zuckertia (Baill.) Muell. Arg., dis- tinguished from all other American species of the genus by its numerous stamens (ca. 40 vs. 2 or 3 or rarely up to 20) and its large, usually lobed leaves. The Panamanian collection represents a considerable range extension, for the species had previously been known only from southern Mexico (Veracruz, Chiapas, Yucatán Peninsula) to Hon- duras. Earlier reports of Tragia bailloniana from Costa Rica (Standley, 1937: 622) are erroneously based on collections of Dalechampia shankii (see discussion in Huft, 1984); however, several recent collections establish the presence of the former Specimens examined. Costa RICA. ALAJUELA: along upper Río Sarapiqui, near Cariblanco and along the road to Colonia Virgen del Socorro, 10°18'N, 84?10'W, ca. m, Burger et al. 11850 (F); lower NE slope of Arenal Volcano, 10?29'N, 84?42'W, 500 m, Lent 2947 (F). LIMÓN: hills 2 airline km SSE of Islas Buena Vista in the Rio Colorado, 14 E km SW of Barro del yv os 83°40'W, 1040' -120 m, Davi idse 31023 (F, N i ro NW-facing slope, Le S of the Rio Ek 10°40’ SUN w ro (` 83°39'30” m na avidse & Herrera 31388 (F, MO); Cerro Coronet E of Laguna Danto; 10?41'N, 83°38'W, 20-170 m, Stevens 24383 (F, MO). PANAMA. CHIRIQUÍ: MM Dam site, 1,400-1,600 m, Folsom et al. 5612 ( 22.2. Tragia fendleri Muell. Arg., Linnaea 34: 178. 1865; in DC., Prodr. 15(2): 928. 1866. TYPE: Venezuela: Biscaina, Fendler 1208 (C). Twining vine; stems and petioles hirsutulous or strigose with mostly nonstinging hairs. Leaves with petioles 4-11 cm long; stipules lanceolate, 5 mm long or more; blades thinly chartaceous, oblong- or elliptic-obovate, rather abruptly short-acuminate at tip, distinctly cordate at base with open to closed sinus, mostly 8-16 cm long, 4-7 cm broad, sparse- ly hispidulous on both faces with stinging and non- stinging hairs, mostly 5-nerved at base; the margins bluntly and coarsely crenate (teeth 15-25 on a side). Inflorescences opposite the leaves, becoming ca. 10-15 cm long, distinctly bifurcate and pro- togynous, the lower pistillate branch with 5-10 flowers; pistillate bracts entire, 3-4 mm long, the staminate ones entire, 1.5-2 mm long. Staminate flowers with minutely hispidulous pedicels ca. 1.5— 2 mm long, articulate near the base (stumps re- maining after dehiscence of flower much shorter than subtending bract); sepals 3 or 4, obovate, acute, strigose without, 1.5-3 mm long, 1.8-2.2 mm broad; disk glands 5, erect, cylindric, thicker than filaments, 0.5-0.8 mm high; stamens 8-12; filaments free; anthers 1-1.2 mm long. Pistillate flowers with hispidulous pedicels up to 2.5 mm long in fruit; sepals 6, lanceolate, asymmetric, green, reflexed in fruit, becoming 3.5-6 mm long, 1-1.5 mm broad; ovary densely hispidulous with stinging hairs; styles basally connate or nearly free, 2-2.5 mm long, distinctly papillate. Capsules copiously hispid with stinging hairs, cocci ca. 8 mm long; columella m long; seeds globose, mottled brownish and gray, 3.6-3.7 mm across The single specimen of this species, previously unrecorded from Panama, is in poor condition and without flowers, so there is some doubt regarding its assignment. The Bristan collection matches a photograph of the type specimen from Venezuela, although the basal leaf sinus is not as open in the Panamanian plant. There is also some resemblance to T. japurensis Muell. Arg., described from Am- azonian Brazil. However, it seems probable that the Brazilian species is synonymous with the one from Venezuela; at least, no convincing differences are given by Pax & Hoffmann (1919: 36) ecimen examined. PANAMA. DARIÉN: Rio Uru- ceca, Bristan 1444 (MO) 22.4. Tragia correae Huft, sp. nov. TYPE: Pan- ama. Panamá: Picada da Estrada Panamá- San Blas entre 320-420 m, 9.1.1973, Sucre, Braga, Dressler & Correa 9832 (holotype, RB-165572, F neg. 62359). Caulis volubilis lignosus; Sate rubelli dense phe Folia alterna elliptica-oblonga 6-12 cm longa septemner- via, infra dense pilosa supra Bras, basi cordata, mene remote denticulata. Inflorescentia racemosa unico flore 1108 Annals of the Missouri Botanical Garden femineo RR ceteris floribus (20-25) masculis; flores masculi calyc o staminibus 3, filamentis crassis, antheris extrorsi ores feminei SUME 5-lobo, ovario dense bius. stylis papillatis Twining woody vine; twigs reddish, densely pi- lose, tardily glabrate, the older twigs with loose, exfoliating bark. Leaves with petioles 0.5-4 cm long, densely pilose; acute, 6-10 mm long, pilose below, glabrous above; blades membranous, elliptic-oblong, acuminate at tip, cordate at base, 6-12 cm long, 3-4.5 c broad, 2.2-2.7 times as long as broad, sparsely pilose above, more densely so below, usually stipules deltate-lanceolate, 7-nerved at base; margins remotely denticulate (teeth 18-22 on a side). Inflorescences opposite the leaves, racemose, ca. 3 cm long (immature), with a single basal pistillate flower, the remaining nodes (ca. 20-25) with staminate flowers; bracts trifid; bracts entire. Staminate flowers on short, ee pedicels; calyx lobes 3, obovate, acute, bo . 1.3 mm long, ca. mm broad, hispidulous ted cucullate; stamens 3, the filaments thick and fleshy, free, ca. 0.8 mm long; anthers elliptic, .2-0.3 mm long, extrorse. Pistillate flowers with pilose pedicels ca. 2 mm long; calyx lobes 5, lan- ceolate, acute, ca. 3 mm long; ovary densely his- pidulous with stinging hairs, the styles free to the base, spreading, papillate, ca. 2 mm long. Mature capsules not seen In aspect Tragia correae resembles the species that Pax & Hoffmann (1919) placed in section Bia, particularly such species as T. fendleri, T. Japurensis Muell. Arg., and T. fallax Muell. Arg. The new species is excluded from that affinity, however, by its racemose rather than bifurcate inflorescences and staminate flowers with three (vs. 8-20) stamens. The entire sepals and extrorse anthers dictate its placement in section Tragia KEY TO THE SPECIES OF DALECHAMPIA IN PANAMA la. 3 aa or unlobed, never compound. S (section Eutragia Muell. Arg. of Pax & Hoffmann, 1919). This species bears some similarity to 7. volubilis but differs by a much woodier habit, densely pilose leaves, and persistently sessile pis- tillate flowers. It is a pleasure to name this distinctive new species for Profesora Mireya Correa of the Uni- versity of Panama. Tragia correae is known only from the type collection. Dra. Correa has kindly searched for a duplicate at the herbarium of the University of Panama (PMA), but so far, unfor- tunately, none has been found. 23. Dalechampia Dalechampia L., Sp. Pl. 1054. Dalechampia scandens L. 1753. TYPE: LITERATURE ARMBRUSTER, W. S. 1984. Two new species of Dale- champia D dade from Mesoamerica. Syst. Bot. 9: 272-278. ¿RZIG. 1984. Partitioning and sharing of pollinators by four sympatric species of Dale- Tei sss M pers MEE in Panama. Ann. Missouri ot. . 71: 1-16. Hurt, M. L 1984. A new iier ti in Dalechampia (Euphorbiaceae). Ann. Missouri Bot. Gard. 71: 341. Pax, F. & K. HOFFMANN. 1919. Euphorbiaceae “a In: A. Engler (editor), Das lancen IV. 147. XII (Heft 68): 1-59. WEBSTER, " L & B. WEBSTER. 1972. The morphology and relationships of Dalec pur scandens (Eu- phorbiaceae). Amer. J. Bot. 59: 573-586 As a result of additional collecting in Panama, the species of Dalechampia are now considerably better understood, and the number of species has increased from five to seven, necessitating a new key. The order of species has been modified to reflect better their systematic relationships (Web- ster & Armbruster, unpubl. synopsis). s and a densely golden-hirsute; leaves unlobed to 3-lobed, mostly 15-30 cm long, with he ad, open at base; stamens 60-70 (in trifid, iie hey wis patent of staminate involuc l free) sed to spreading hairs, these never golden; leaves mostly N c . Stem ti inflorescences with appre smaller, iban shaped; stam E. 3a. Bracts of staminate involucel fre apically lacerate; leaves u creamy or pinkish, marginally hc see 4a. Stigma asymmetric, slightly dilated, not over ; hairs of stem spreading at least in part; leaves persistently anata beneath, attenuate- acuminate, basal sinus narrow or lobes neo volucral bracts narrowly spathulate and apically l. D. shankii and disarticulating . pA bractlets within staminate involucel nlobed eee onally mall lateral teeth); involucral bracts s more ui less Mi ose. 1.2 mm across; involucral bracts creamy o ng canescens subsp. friedrich sthalii EN c . Stigma peltate, 1-3.5 mm across; involucral bracts pink or pu rple to white with pink veins (rarely white with greenish veins); hairs of stem appressed; leaves glabrate beneath, cuspidate d open sinus dioscoreifolia it 3b. Bracts of staminate involucel more or less connate into a cup, not disarticulating separately; leaves Volume 75, Number 3 1988 Webster & Huft 1109 Panamanian Euphorbiaceae lobed, at least in part; involucral bracts creamy or white, with entire denticulate margins; seeds h smooth. 5a. Unlobed leaves often present with lobed leaves; involucral bracts merely 3-dentate at apex, 7-9-nerved, 3-6.5 cm long; bractlets within staminate involucel apically lacerate D. tilüfolia en T Unlobed leaves rare or absent, the leaves all 3- or 5-lobed; involucral bracts 3- load at least to the middle, 5-nerved, 1.5-3 cm long; bractlets within staminate involucel entire, laminar 7. Dos candens lb. Leaves compound, 3- e (rarely 5-foliolate). oung stems and low r leaf surfaces spreading-hirsute; leaflet tips cuspidate-acuminate; involucral bracts we ea n ela of 4 separate bracts, staminate bractlets lacerate; central fruiting 4. D . websteri pedicel c 6b. n eds roirorsely pubescent; leaf blades glabrate; leaflet tips acute to evenly acuminate; involucral | 2-li bracts green; staminate involuce central ‘fruiting pedicel ca. 1 cm lon ipped (bracts confluent); staminate bractlets truncate, lamin 6. D ar; . cissifolia subsp. panamensis 23.1. Dalechampia shankii (A. Molina) Huft, nn. Missouri Bot. Gard. 71: 341. 1984. Tragia shankii A. Molina, Ceiba 11: 68. 1965. TYPE: Costa Rica. Limon: Rio Reventazon, 15 m, 23 Oct. 1951, Shank & Molina 4427 (holotype, F). This species, originally described from Costa Rica, has now been recorded from Nicaragua to Colombia. It may easily be distinguished from our other Panamanian species by its large and distinc- tive golden hairs, as well as yellowish, narrowly spathulate, trifid involucral bracts. The free bracts of the staminate involucel and the lacerate sta- minate bractlets indicate that the species belongs to section Dioscoreifoliae in the emended sense (Webster & Armbruster, ined.), along with the two following Panamanian taxa. The collections of Bar- ry Hammel show that D. shankii is polymorphic in leaf shape, since leaf blades from the single locality vary from unlobed to having one lateral lobe to three-lobed. Additional collections from Cos- ta Rica, Panama, and Colombia are cited by Huft (1984) Specimens examined. PANAMA. COCLÉ; near sawmill, 16.7 km N of turnoff to Coclesito from L 700 ft., Hammel 1811, 1812, 1813 (MO); Llano nde 200 m, Churchill et al. 4148 (F, MO). 23.2. Dalechampia canescens Kunth subsp. friedrichsthalii (Muell. Arg.) Webster & Huft, stat. nov. Dalechampia friedrichsthalii Muell. Arg., Flora 55: 45. 1872. TYPE: Nic- aragua. Río San Juan: Friedrichsthal 683 (not seen; locality erroneously cited by Mueller as Guatemala). Several additional collections of this plant have now been made in Panama. Specimens examined. PANAMA. COLÓN: along Rio Mendosa, 8 km NW of Gamboa /Vee & Smith 11370 (MO); Rio Fato, Pittier 3866 (GH, NY, US); Rio Bo- querón, 6-8 km upstream from Peluca Hydro Station, Siri 1005 (DAV). SAN BLAS: Puerto Obaldia, sea level, Knapp & Mallet 4627 (DAV, MO). It now appears that Dalechampia friedrichs- thalii is excessively close morphologically to D. canescens Kunth (Nov. Gen. 2: 98. 181 The collection from San Blas in particular seems somewhat intermediate, and it makes more sense biologically to treat the two taxa as allopatrically replacing subspecies of a single species. The two Sp may be keyed out as follows: . Leaf blades mostly attenuate-acuminate, sp ly to moderately p Pee ame pensat (hairs most- 3 0.1-0.2 m p? ong, not ias ing in sinuses between veinlet p. friedrichsthalii lb. Leaf blades at s dcr pre i softly pubescent beneath (many hairs over 0.2 mm long, overlapping in sinuses between veinlets) subsp. canescens The Colombian taxon, subsp. canescens, was described from Tolima Province (Mariquita) and cited from Narino by Pax & Hoffmann (1919: 52). The additional collections cited below indicate that subsp. canescens is widely distributed in the low- lands and foothills of the western Andean region in Colombia. Specimens examined. COLOMBIA. CALDAS: Quebrada m N of Honda, 300 m, Gentry et al. 18167 : Rio Patia, 590 m, Plowman & Vaughan 5354 (DAV). SANTANDER: 29 km Vicente de Chucurí, 200 m, Gentry & Aguirre 15429 MO). ~ 23.3. Dalechampia dioscoreifolia Poeppig in Poeppig & Endl., Nov. Gen. Sp. Pl. 3: 20 1841. TYPE: Peru. Maynas: Poeppig 2163 (W). One additional locality merits noting: PANAMA. DARIÉN: near Rio Canglón, Duke & Bristan 378 (MO) An unusual specimen from Playón Chico, San 1110 Annals of the Missouri Botanical Garden Blas (Gentry 6365, MO) appears to be intermediate between D. dioscoreifolia and D. canescens subsp. friedrichsthalii, having the broad stigma of the ormer and the pubescent attenuate-acuminate leaves of the latter. 23.4. Dalechampia websteri Armbruster, Syst. Bot. 9: 272. 1984. TYPE: Costa Rica. Heredia: La Selva, 3 km SE of Puerto Viejo, Armbruster & Herzig 79-207 (DAV). This species, recently described from Costa Rica, has been identified from Panama on the basis of the single record that was attributed (with doubt) to D. cissifolia in our treatment of 1968 Specimen examined. PANAMA. BOCAS DEL TORO: Chi- riquicito, Lewis et al. 2123 (MO). 23.5. Dalechampia cissifolia Poeppig subsp. panamensis (Pax & K. Hoffm.) Webster, Ann. Missouri Bot. pd 54: 193. 1967. D. panamensis Pax Hoffm., Pflanzenreich Iv. 147. XII (Heft 68): 19. 1919. SYNTYPES: Costa Rica: Tonduz 8089, Guatemala: Cub- ilgüitz, Tuerckheim Il. 244, 7978, Mexico. Chiapas: Escuintla, Donnell Smith 2079. Panama: Oersted; Pittier 2311, 3775 This species still requires additional study. A variant with simple, unlobed leaves mixed with the compound ones, to which the name Dalechampia heteromorpha Pax & K. Hoffm. has been applied, occurs in Panama and throughout Central America but does not appear to be specifically distinct from the South American subsp. cissifolia. There is one new provincial record for D. cis- sifolia subsp. panamensis. Additional specimen examined. PANAMA. BOCAS DEL TORO: 10 mi. NW of Almirante, D'Arcy 11204A (MO). 24. Omphalea Omphalea L., Syst. Nat. ed. 10. 1264. 1759. Nomen conserv. TYPE: Omphalea triandra L. (typ. conserv.). 24.1. Omphalea diandra L., Sp. Pl. ed. 2. 1377. 1763. TYPE: Jamaica, Browne (pre- sumably BM, not seen). Additional specimen examined. — PANAMA. CHIRIQUÍ: Burica Peninsula, 2-4 mi. SW of Puerto Armuelles, Croat 22045, Liesner 409 (MO). 25. Pera Pera Mutis, Kongl. Vetensk. Akad. Nya. Handl. 5: 299, tab. 8. 1784. TYPE: Pera arborea Mutis. Subfamily III. CROTONOIDEAE Pax 26. Tetrorchidium Tetrorchidium Poeppig in Poeppig & Endl., Gen. Nov. Sp. Pl. 3: 23, tab. 227. 1841. TYPE: Tetrorchidium rubrivenium Poeppig. The discovery of two distinctive new species in western Panama makes it necessary to provide a revised key. New province records are also re- corded. KEY TO THE SPECIES OF TETRORCHIDIUM IN PANAMA la. Stems knobby, the apex densely hirsute or stri- gose. 2a. Stems with persistent swollen stipules; leaves less than 7 cm long, the paired glands toward the apex of the petiole L. icrophyllum . Stems with raised leaf scars, ie. stipules persistent or not, rarely swollen; leave usually over 10 cm long, the pues podio near the middle of the petiole N c MM EM ELE Te AA: “T. costaricense lb. Stems smooth, the apex appressed- pubescent or or glabrescent, leaves mostly more than 7 cm long usually more Nee 4 long. 3a. Basal foliar glands ca. 0.5-0.6 mm thick, lara attached on laminar tissue; leaves h mostly 1 long or more, wit or 7 sitas lateral veins; iu flowers essile ea RONDE REA euryphyllum w c . Basal foliar glands ca. 0.15- 0. 3 mm thick, pl at junction of lamina and petiole r well down on "s leaves 7-12 cm n ostly with only 4 or 5 prominent lateral an. lalla Joves with distinct pec wid 1.5-2 mm long 0... . 4. T. gorgonae 26.1. Tetrorchidium microphyllum Huft, sp. nov. TYPE: Panama. Chiriqui: 3.5 mi. NE of Boquete, end of road along Río Alto, 6,200 ft., 18 Nov. 1978, Hammel 5721 (holotype, MO; isotype, F, F neg. 62357). Figure 2. rbor mediocris, gracilis, dioecia; ramulis junioribus fragilibus, praeter raso dense hirsutum, ER tiw totis oblanceolatis, 2.5-7 c ngis, sparsim pubescentibus utrinque pilis malpighiaceis vel glabratis, margine integris, apic s, dense a bescentibus pilis malpighiaceis; stipulis glanduliformibus, tumidis, persistentibus; floribus masculis in thyrsis axil- laribus ad 3 cm longis; floribus femineis et capsulis ignotis. Dioecious tree to 10 m; branches brittle, gla- brous except at tips where densely hirsute, ap- pearing warty by the presence of persistent, swollen Volume 75, Number 3 1988 Webster & Huft Panamanian Euphorbiaceae 1111 E2. Tetrorchidium ee —a. Habit.—b. Detail of staminate inflorescence. Based on Hammel Fic 6039. "Rai by Wan-Ling Pen, stipules. Leaves short-petiolate, crowded near ends of branches; petioles 0.5-1.2(-1.5) cm long, densely appressed-pubescent with malpighiaceous hairs to glabrate, with massive, paired, thick-stalked, opposite or subopposite glands near the tip, these 0.6-0.8 mm long, 0.8-1 mm thick, stipules glan- duliform, tumid, broadly triangular, 1.5-2 mm long, densely pubescent, persistent, glabrate soon after leaf-fall; blades chartaceous, oblanceolate, acumi- nate at tip, cuneate at base, (2.5-)4-5.5(-7) cm long, (0.9-1)1.2-2 cm broad, sparsely pubescent with malpighiaceous hairs to glabrate on both sides, the midrib and primary veins (3-4 on a side) prom- inently raised below, the veinlets forming a prom- inent reticulum; margins entire. /nflorescences ax- illary; staminate thyrses unbranched, 1.5-3 cm long, the rachis densely strigose; pistillate inflores- cences unknown. Staminate flowers subsessile; ca- lyx lobes 3, triangular, glabrous, ca. 1-1. long; petals lacking; anthers subsessile, 0.8-0.9 mm long. Pistillate flowers unknown. Fruits un- known. Cloud forests, western Panama. This distinctive new species of Tetrorchidium appears to be most closely related to T. brevifolium Standley & Steyerm., described from the province of Alta Verapaz, Guatemala (Publ. Field Mus. Nat. Hist., Bot. Ser. 23: 126. 1944), from which it differs by the smaller leaves (7-12 cm long in T. brevifolium), densely pubescent shoot apices and inflorescences (both glabrous in 7. brevifolium), and shorter inflorescences (4-7 cm long in T. brevi- folium). The type specimen of the Guatemalan species (Rubelpec, Finca Seamay, Wilson 188, F) lacks the persistent tumid stipules that are so char- acteristic of T. microphyllum, but these are pres- ent on several collections of the former species made in 1974 and 1975 from Baja Verapaz, Gua- temala (Lundell & Contreras 19173, 19436 (both 1112 Annals of the Missouri Botanical Garden F, LL); Williams et al. 43277, F). The type spec- imen consists only of branchlet tips with a few leaves and staminate inflorescences that appear to have been taken from rapidly growing long shoots. Tetrorchidium molinae L. Williams, described from cloud forests in the mountains above San Juancito, Honduras (Fieldiana, Bot. 29: 348. 1961, based on Williams & Molina 17068, F), is similar in all respects to 7. brevifolium and should be relegated to the synonymy of that species. A paratype spec- imen of 7. molinae from the same area, Williams & Molina 13980 (F), has long shoots without stipules attached to a normal shoot with persistent stipules. This matches the pattern of the type col- lection of T. brevifolium, thus confirming the sus- picion voiced above concerning the nature of that collection. Additional specimen examined. PANAMA. CHIRIQUÍ: end of road past Palo Alto NE of Boquete in forest along ridge, 6,200-6,800 ft., Hammel 6039 (F, MO). 26.2. Tetrorchidium costaricense Huft, sp. nov. TYPE: Costa Rica. Puntarenas: Cordillera de Tilarán, Monteverde Reserve, near Con- tinental Divide on Pacific side, 1,520-1,580 m, Dryer 1403 (holotype, CR, F neg. 62351; isotypes, F, F neg. 62350, MO, F neg. 62349). rbo 16 m alta, dioecia; ramulis junioribus dense strigosis; foliis anguste oblongis, 8-16 cm longis, apice abrupte cuspidatis; petiolis longis, prope medium glandibus binatis; floribus masculis sessilibus, in thyrsis axillaribus -9 cm longis; floribus femineis subsessilibus, ad fructus maturitatem brevipedicellatis, segmentis disci discretis lig- ulatis; ovariis 2-locularibus; seminum ovoideis, grosse re ticulatis. Dioecious tree to 16 m; branches densely stri- gose toward tip, appearing knobby from the raised leaf scars and occasionally from persistent, indur- ate stipules. Leaves long-petiolate, not crowded toward ends of branches; petioles 2.5-6 cm long, glabrous or minutely strigose, with paired, subop- posite, sessile, patelliform glands near the middle, these ca. 1 mm (rarely to 2.2 mm) diam.; stipules oblong, 1.5-2.5 mm long, 1-1.5 mm broad, dense- ly strigose, persistent, sometimes indurate after leaf-fall; blades membranous, narrowly oblong, 8— 16 cm long, 3-7 cm broad, 2.2-3.3 times as long as broad, abruptly cuspidate at tip with an acumen 5-10 mm long, acute to attenuate at base, minutely puberulent below with scattered short malpighia- ceous hairs, glabrous or nearly so above, the midrib ate, prominent below, obscure above; margin entire, eglandular. Inflorescences axillary, the axes densely strigose with short malpighiaceous hairs; staminate thyrses 4-9 cm long, freely branched, the lateral branches to 3.5 cm long; pistillate racemes to 5 cm long. Staminate flowers in glomerules of 2—5, sessile; sepals 3, obovate, cucullate, glabrous without, pilose within; anthers subsessile, 1.3-1.5 mm broad. Pistillate flowers subsessile, the pedicels becoming 0.5-3 mm long in fruit; sepals broadly ovate, obtuse, 3-3.5 mm long, glabrous or sparingly short-strigose without, densely hispid toward base within; disk segments free, narrowly ligulate, ca. 2 mm long; ovary smooth, 2-locular, glabrous above, densely long- strigose below; style cap at maturity 0.5-0.8 mm high, 1.5-1.8 mm diam. Capsule 3-5 mm high, 5-6 mm diam., glabrous, oblate to globose; seeds ovoid-lenticular, 5-6 mm long, prominently and coarsely reticulate, the caruncle an irregular yellow papery keel running halfway from the hilum to the apex. Known only from Costa Rica and extreme west- ern Panama, this distinctive species is easily dis- tinguished by the large, dark green, lanceolate leaves that are conspicuously venose, the paired glands near the middle of the rather long petioles, and the densely and minutely strigose branchlet tips, petioles, and leaves. Tetrorchidium costaricense belongs to the group of species with free ligulate disk segments in the pistillate flower that includes 7T. rotundatum Stand- ley and T. brevifolium Standley & Steyerm. in northern Central America. Like 7. T. rotundifolium has paired glands near the middle of long petioles, but differs in its completely gla- brous stems and leaves, mostly unbranched sta- minate thyrses that have larger glomerules, densely pubescent pistillate calyces, and distinctly pedicel- late fruits. Tetrorchidium brevifolium differs in its glabrous stems, leaves, and calyces, short petioles with paired glands near the tip, and unbranched staminate thyrses. The widespread South American costaricense, species T. rubrivenium Poeppig also belongs to this group but has glabrous stems and leaves, crenate or denticulate leaf margins, sometimes long petioles with the paired glands near the tip, densely pu- bescent pistillate calyces, uniformly puberulent ovary and capsule, and pedicellate fruits. Additional specimens examined. Bag Rica. ALAJUELA: Cordillera bí Tilarán, Monte Atlantic side, 1,500-1,580 m CARTAGO: Reserva de Tapanti 1,300- 18752 (F). PUNTARENAs: Cordillera de T lapan, Monte- verde Reserve, Pacific side, en orilla de Pantano Chomogo, 1,600-1,620 m, Dryer 659 (CR), 887 (CR, F); Mon- teverde Reserve, at field station, 1,500 m, Haber 491 (F); Monteverde Reserve, 1,570 m, Haber & Bello 1640 (F), 1,500 m, Haber & Bello 2457 (F). ALAJUELA/PUNTA- RENAS: on and near the Continental Divide, ca. 2-5 Volume 75, Number 3 1988 Webster & Huft Panamanian Euphorbiaceae 1113 E and SE of Monteverde, 10?18'N, 84?46'W, 1,580- 1,700 m, Burger & Gentry 8608 (F); Monteverde, Dryer 1731 (F). SAN JOSÉ: bajo de La Hondura, Poveda 862 (CR, USJ). PANAMA. BOCAS DEL TORO: along Continental Divide, trail to headwaters of Rio Mali, to W of Oleoducto Road, 8°47'N, 82?13'W, 1,200 m, Churchill 5276 (F); Fortuna Dam region, along Continental Divide W of high- way pass, ca. 8°45'N, 82°15'W, ca. 1,200 m, McPherson 9695 26.3. Tetrorchidium euryphyllum Stand- ley, Publ. Field Columbian Mus., Bot. Ser. 4: 219. 1929. TYPE: Panama. Bocas del Toro: vicinity of Almirante, 1928, Cooper 621 (ho- lotype, F). When the original treatment was written, this species was known only from Costa Rica and ex- treme western Panama, but recent collections in Panama have now extended its range eastward to Darién. Additional specimens examined. PANAMA. BOCAS DEL TORO: between Quebrada and Buena Vista, Kirkbride & Duke 662 (MO). CHIRIQUÍ: Fortuna Dam dar T along Quebrada Arena, ca. 8°45'N, 82°15'W, ca. 1,100 m McPherson 8394 (F). COCLE: slopes of Cart Pilón near x p 100-900 m, Duke 12196 (MO, 2 e Cerro lón, Dwyer 8330 (MO); La Mesa, 8.5 mi. from Club Camana Valle), Dwyer 10515 (MO); Margarita near chicken farm, Dwyer uke 8280 (MO); La Mesa, above El Valle, Dwyer & Nee 11938 (MO); La ac km N of El Valle, 850-875 m, Nee & Dwyer 9212 (MO). DARIEN: Cerro Sapo, ca. 2,500 ft., Hr 1240 (MO). SAN BLAS: Cerro Brewster, 9°18'N, 79?16'W, 850 m, de Nevers et al. 5408 (F). VERAGUAS: Caribbean bid above Rio Primero Brazo, 5 mi. NW of Santa Fe, 7 1,200 m, Croat 23233 (MO); NW of Santa Fe, 4.2 km from Escuela pi Alto de Piedra, Mori & Kallunki 4831 (MO); ca. 2.7 km from Escuela Agricola Alto de Piedra, Mori & Kallunki 5359, 6208 (MO); 7. km W of Santa Fe on road past agricultural school, 2,900 ft., Nee 11216 (MO). 26.4. Tetrorchidium gorgonae Croizat subsp. robledoanum (Cuatrec.) Webster, Ann. Missouri Bot. Gard. 54: 199. 1967. T. roble- doanum Cuatrec., Brittonia 9: 81. 1957. TYPE: Colombia. Antioquia: 23 Jan. 1947, Gutiérrez 35556 (holotype, CAL, not seen). This species is still unknown in Central America outside of Central Panama. Several recent collec- tions allow a description of the fruit to be made for the first time. They are on densely strigose pedicels 7-10 mm long and jointed below the mid- dle. The capsule is green, drying to brown, globose, shallowly 3-lobed, rugulose, 4-6 mm high, 4.5-7 mm in diameter, more or less densely strigose with short (0.2-0.6 mm) malpighiaceous hairs. The 3 styles are deeply bifid, 0.6-0.7 mm long, and tu- mid. The seeds are ovoid, 4-5 mm long, 3-4 mm in diameter, black, coarsely and shallowly pitted, and completely surrounded by bright red, juicy arils. Additional specimens examined. ms COLÓN: Santa Rita Ridge Road 1.5 mi. from Transisthmian Hwy Dwyer & Gentry 9338 (MO, 2 sheets); Santa Rita Ridge Road 4 Agua Clara weather iau ca. 5 sheets); in forest along Rio Guanche 3- b 300-700 ft., Hammel 4894 (MO). PANAMA: El Llano- Carti highway, 17-20 km N of El Llano, Dressler 4629 F, MO); 10 km N of Margarita on road to Madrono, then 3 km W o n ridgetop road, 1,800 ft., Hammel 6014 (MO); Cerro Jefe region 2.5 mi. N of turnoff to radio , Hammel 6300 (MO); El — 350 F Mori & Kalluaki 3531 (MO); 5-10 km NE of Altos de Pacora on trail at end of road, 700-800 m, Mori i Kallunki 6058 (MO); El Llano-Carti road, 8 km N of Pan Am. Hwy. at El Llano, ca. 450 m, Nee & Warmbrodt 10391 (MO). 27. Manihot Manihot Miller, Gard. Dict. abr. ed. 4. 1754. TYPE: Manihot esculenta Crantz (Jatropha manihot L.). RECENT LITERATURE Rocers, D. J. & S. G. APPAN. 1973. Manihot; Man- ihotoides. Fl. Neotropica 13: 1-272. & H. S. FLEMING. 1973. A monograph of Manihot esculenta ith an s of the taxi- metric methods used. Econ. Bot. 27: 1-113. The discovery of an additional Panamanian species necessitates a revised key and enumeration. KEY TO THE SPECIES OF MANIHOT IN PANAMA la. Leaves glabrous, mostly with 7-9 lobes; calyx glabrous within (12-14 mm long; disk bia. anthers glabrous; ovary not mer ribbed o winged) M. aesc ulifolia lb. Leaves pubescent or with cane lobes; calyx pubescent within. 2a. Leaves pubescent, mostly with 5 or more lobes; ovary distinctly ribbed or winged; a) staminate calyx 34.5 m VO ONT . M. esculenta Leaves glabrous, strictly 3- lobed: ovary not ribbed or winged; staminate calyx 10-20 mm long; anthers glabrous; gra as vine . M. brachyloba N d 27.1. Manihot aesculifolia (Kunth) Pohl, Pl. Bras. Icon. Descr. 1: 55. 1827. Janipha aes- ips Men Kunth, Nov. Gen. Sp. 2: 85, tab. 1817. TYPE: Mexico. Campeche: Hum- re & Bonpland (P, not seen). This plant was called M. gualanensis Blake in our original treatment, but that name has been 1114 Annals of the Missouri Botanical Garden reduced to a synonym of M. aesculifolia in the recent monograph of the genus by Rogers & Appan cited above. 27.3. Manihot brachyloba Muell. Arg., Fl. Brasil 11(2): 451. 1874; Rogers & Appan Fl. Neotrop. 13: 190-192. 1973. TYPE: Dra: zil. Para: Martius (syntype, G; microfiche seen). The Bristan specimens have very young inflo- rescences but vegetatively match the description and illustration given by Rogers & Appan. How- ever, contrary to their description and that of Muel- ler, the staminate buds in the Panamanian speci- mens are pubescent externally. Specimens examined. PANAMA. DARIEN: headwaters of Rio Tuqueza, between Quebrado Venado and Peje Swamp, Bristan 1001 (DAV, MO); between Manené and Rio Coasi, Hartman 12127 (MO). 28. Cnidoscolus* Cnidoscolus Pohl, Pl. Brasil. Icon. Descr. 1: 56. 1827. LECTOTYPE: Cnidoscolus hamosus Pohl (chosen by Small in Britton & Brown, Illust. Fl. N. U.S. ed. 2, 2: 462. 1913). 28.1. Cnidoscolus urens (L.) Arthur, Tor- reya 21: 11. 1921. Jatropha urens L., Sp. Pl. 1007. 1753. TYPE: “America calidiori, in Brasilia & c." (not seen, possibly in Hortus Cliffortianus Herbarium, BM). Since the treatment of 1968, further study of the Panamanian specimens of C. urens leads to the conclusion that the two variants discussed there merit taxonomic recognition. Pending a more de- tailed revision of the C. urens complex (Breckon, ined.), the Panamanian plants may be disposed of as follows. 28.1a. Cnidoscolus urens (L.) Arthur subsp. urens. Specimens examined. PANAMA. PANAMA: Farfan Beach area, Correa et al. 1586, Dwyer 3065, Tyson 1803 (MO). cocLÉ: o Aguadulce and San Antón, el al, 7 al. qe (MO); Sal Salinas de Chitré, Croat 9692 (MO). PANAMÁ: San Carlos, de McPherson 11 (MO). * Contributed hl Dr. Gary Breckon, University of Puer- to Rico, Mayag 28.1b. Cnidoscolus urens subsp. adenophi- lus (Pax & K. Hoffm.) Breckon, stat. nov. Jatropha adenophila Pax & K. Hoffm., Pflanzenreich IV. 147. VII(Heft 63): 409. 1914. Cnidoscolus adenophilus (Pax & K. Hoffm.) Pax & K. Hoffm., Nat. Pflanzenfam. ed. 2, 19c: 166. 1931. TYPE: Panama. Pa- namá: Chepo, Pittier 4740 (isotype, US). Specimens examined. PANAMA. CANAL ZONE: Mira- flores Locks, Stern et al. 81 (MO); Pipeline Road, Croat 12732 (MO); Curundú, McDaniel 5180 (MO), Tyson 1045 (MO) Ft. Amador Islands, Tyson 5413 (MO). DARIÉN: El Real, Lazor & Correa 3364 (MO), Stern et al. 454 (MO). Los saNTOs: 5 mi. NW of Guararé, Wilbur et al. 12054 (MO); Monagre Beach, Lewis et al. 1673 (MO). PANAMA: Jenine, Río Canita, Duke 3821 (MO); between Las Margaritas and Río Mamoni, Duke 5867 uni Puente de Pacora, De Hoyos 18 (MO). vERAGUAS: 4 mi. E of Santiago, Duke 12354 (MO); 12 km E of Redes Dwyer & Kirkbride 7450 (MO). 28.2. l itifolius (Miller) I. M. Ea Contr. Gray Herb. 68: 86. 1923. subsp. aconitifolius. Jatropha aconitifolius Miller, Gard. Dict. ed. 8. 1768. TYPE: Herb. Miller (presumably at BM, not seen). In the present interpretation, the Panamanian specimens of Cnidoscolus aconitifolius all belong to the nominate subspecies, which is not native to Panama. 29. Jatropha Jatropha L., Sp. Pl. 1006. 1753. LECTOTYPE: Jatropha gossypiifolia L. (see McVaugh, 1944: 459). RECENT LITERATURE DEncaN, B. 82. Comparative anatomy of the petiole and infrageneric relationships in Jatropha (Euphor- biaceae). Amer. J. Bot. 5. Son eee e. of interspe- A 1 1 t Bot. 9: 467-478. WEBSTER. 1979. Morphology and infrageneric €— of the genus Un (Eu phorbi iuh alif. Publ. Bot. 74: 1- MERE. R. e genus Cnidoso olus: generic limits and gei groups. Bull. Torrey Bot. Club 71: 457-474. I 30. Pausandra Pausandra Radlk., Flora 53: 92, tab. 2. 1870. TYPE: Pausandra morisiana (Casar.) Radlk. (Thouinia morisiana Casar.). Dioecious trees or shrubs; stems with reddish latex; indumentum malpighiaceous. Leaves alter- Volume 75, Number 3 88 Webster & Huft Panamanian Euphorbiaceae 1115 nate, simple; petioles swollen distally; blades pin- nately veined, biglandular at base, the margins serrate. Inflorescences axillary, spiciform; stami- nate flowers in glomerules; pistillate flowers solitary at each node; bracts inconspicuous, eglandular. Staminate flowers subsessile; calyx lobes 5, im- bricate; petals 5 or rarely 6, connate (at least below), adaxially villous; disk extrastaminal, ur- celoate, lobate, glabrous; stamens (3-)5- 7; fila- ments free; anthers dehiscing introrsely and lon- gitudinally, the connective not enlarged; pollen grains globose, inaperturate, clavate; pistillode ab- sent. Pistillate flowers subsessile; sepals 5, imbri- cate; petals 5, free, adaxially villous; disk urceolate, sometimes lobate, glabrous; ovary of 3 carpels; ovules 1 per locule; styles free, bifid. Fruits cap- sular; seeds smooth, carunculate; endosperm co- pious; embryo straight, cotyledons palmatinerved, much longer than radicle. This primarily South American genus is here reported from Panama for the first time. LITERATURE BaiLLoN, H. 1873. Nouvelles observations sur les Eu- phorbiacées. Adansonia 11: 72-138. Lanjouw, J. 36. The genus Pausandra Radlk. Re- cueil Trav. Bot. Néerl. 33: 758-769. = c: 30.1. Pausandra trianae Baillon, Adansonia 11: 92. 1873, proposed without reference to Pogonophora trianae Muell. Arg. TYPE: Co- lombia: Bogotá, plains of San Martin, Rio Meta, Triana 2597 (holotype, P, not seen; isotypes, G, not seen, photo F neg. 24574, K, not seen, holotype of Pogonophora trianae Muell. Arg.). Nerd trianae Muell. Arg., Flora 47: 434. 1864. :: Colombia. Bogota: plains of San Martin, Rio Meta, Triana 2597 (holotype, K, not seen, isotypes, G, not seen, photo neg. 24574, P, not seen, e of Pausandra trianae Baillon). Pausandra quadriglandulosa Pax & K. Hoffm., Pflan zenreich IV. 147. XIV(Heft 68): 43. 1919. TYPE: Brazil: Ri "om Serin ngal S. Francisco, Ule 9538 (holotype, B, not seen, photo F neg. 5406). Pausandra extorris Sta ndley, Trop. Woods 17: 24. Mar. 1929; Publ. Field Ehon Mus., Bot. Ser. 4: 219. ct. 1929. TYPE: Nicaragua: Bragman’s Bluff, En- glesing 216 (holotype, F E 52719; wood sam- ple, Y no. 13301). Clavija septentrionalis L. O. Williams, Fieldiana, Bot. 32: 205. 1970. TYPE: Nicaragua: Gracias a Dios, Laimos Creek, ca. 15 km SW of Waspam, 7 ar. 1961, Bunting & Licht 390 (holotype, F; isotypes, NY, US). Tree to 30 m; branching pagodiform; sap dark red-brown; twigs and buds + densely brown-pu- berulent with malpighiaceous hairs; leaf scars prominent. Leaves horizontally aligned; petiole 1.5- 3.5 cm long, 2-4 mm thick, terete, strigose with short (to 0.5 mm long) malpighiaceous hairs; glands at apex of petiole (3-)4, cylindrical, 1-1.5 mm long, ca. 1 mm thick; blade chartaceous, obovate to oblanceolate, rounded, obtuse, or abruptly short- cuspidate at tip, long-attenuate at base, 20-50 cm long, 7-18 cm broad, 2.6-4 times as long as broad, glabrous above, glabrate or thinly puberulent with short malpighiaceous hairs below, the secondary eins prominent, 15-23 per side; margin remotely Pieces Inflorescences spicate, arising singly 10-25 cm long, the rachis densely puberulent. Staminate flowers 6-12 per glomerule, these sessile, widely spaced; sepals 5, imbricate, densely puberulent, obovate, ca. 1.5 mm ong, c mm broad, rounded at apex; petals 5, white, narrowly obovate, ca. 6 mm long, gla- brous without, densely hirsute toward base within; stamens 6, exserted, the filaments 5-6 mm long; disk cupulate. Pistillate flowers not seen. Capsule smooth, strigose with short malpighiaceous hairs, apparently subglobose, ca. 1 mm diam. (fragments only seen); columella 7-8 mm long; seeds sub- globose, brown with irregular white striations, 8— 9 mm long, ca. 6 mm diam.; irregular mass near hilum. in the upper axils, caruncle a flattened Rainforests, Honduras to western Brazil. The description is based on the Panamanian specimens cited below and supplemented by col- lections at F from Costa Rica and Nicaragua. The genus Pausandra remains poorly under- stood. The most recent revision (Lanjouw, 1936) was based upon only 33 collections and resulted in the recognition of nine species, most known from only a single sex. Lanjouw admitted that several of the species might need to be united as more material became available. Pausandra trianae does appear, however, to be one of the better-delimited species in the genus and is certainly the most widespread. The Central American plants readily key to that species in Lanjouw's revision, where P. extorris and P. quadriglandulosa are reduced to synonymy. The identity of Clavija septentrio- nalis as P. trianae was first pointed out by R. L. Liesner of the Missouri Botanical Garden (pers. o ° The author citation for Pausandra trianae has been almost universally given for the last century as Pausandra trianae (Muell. Arg.) Baillon, car- rying the implication that Baillon had transferred Pogonophora trianae Muell. Arg. to its correct place in Pausandra. It is clear from Baillon`s text, 1116 Annals of the Missouri Botanical Garden however, that he intended to describe a new species; there is no indication that he was aware of Mueller's name. Baillon's paper (1873) is a series of mis- cellaneous notes on the Euphorbiaceae. Under Pausandra (pp. 92-93), he stated that the genus was described in 1870 and was known by a single species, P. morisiana. He then indicated that a collection from Colombia (Triana 2597) repre- sented an additional species in the genus, which may be given the name P. trianae. (“Ce genre est aussi représenté à la Nouvelle-Grenade, par une plante qui est bien voisin de celle de Brésil, qui n'en est peut-étre méme qu'une forme; je lui don- nerai provisoirement le nom de P. Trianae.") No reference whatever was made to Mueller's name, and none would be expected. One would not nor- mally look in Pogonophora for species pertaining to Pausandra; the genera are too different for that. The fact that the names of Mueller and of Baillon are both based on the same collection, although on different specimens, is merely a coincidence. That both chose the same epithet is also coincidental, but not very surprising, given the common practice of naming species after the collector of the type. Thus, although Baillon ideally should have made a transfer of Mueller's name, he was understand- ably unaware that Mueller had previously described the species at hand, and did not make the transfer. Had the two authors chosen different epithets, then the rules would mandate a transfer of Mueller's epithet to create a new combination in Pausandra supplanting Baillon's name. But that solution is closed because it would create a later homonym. Article 63 of the Code (Voss, 1983) might mislead one to reject Baillon's name as superfluous, because it seems to be based on the type of a name whose epithet ought to have been adopted under the rules. A correct reading of the Code, however, makes it clear that a type is a specimen, and not a gathering, which usually consists of several duplicate speci- mens. An isotype has no official standing as long as the holotype exists, although its value is un- questioned. Thus, for a name to be rejected as KEY TO THE SPECIES OF CROTON IN PANAMA superfluous, it must be based upon the same spec- imen as an earlier name whose epithet ought to have been adopted. This is not the case with Pau- sandra trianae Baillon. There is therefore no bar to the acceptance of that name as the correct one for this species. Additional specimens examined. dida BOCAS DEL TORO: premontane rainforest between and Cer- ro Bonyic, near Terebe, 300-900 ft., Kirkbride & Duke 647 (MO). SAN BLAS: Río Taindi (Taimdi of maps) 6 km above mpm with Rio Mandinga, 9°25'N, 79%11'W, 30-100 m, de Nevers & Herrera 7674 (F); along newly cut road from El Llano to Carti-Tupile, near Continental Divide, 300-500 m, Liesner 1289 (DAV, F, MO), s Kallunki 5535 ; seasonal low- land rainforest on the Aila Tilar (Rio Acla), 8°48'N, 77°40'30"W, 25-100 m, Sugden 424, 588 (MO). 31. Garcia Garcia Vahl in Rohr, Skr. Naturhist.-Selsk. 2 217. 1792. TYPE: Garcia nutans Vahl. 31.1. Garcia nutans Vahl in Rohr, Skr. Na- turhist.-Selsk. 2: 217, tab. 9. 1792. TYPE: Colombia. Magdalena: near Santa Marta, von Rohr (C, not seen). Additional specimens examined. PANAMA. CANAL ZONE: Barro Colorado I., Knight 1090 (MO). LOS SANTOS: between Tonosi and Guánico, Tyson et al. 3126 (MO). 32. Croton Croton L., Sp. Pl. 1004. 1753. LECTOTYPE: Cro- ton aromaticus L. (chosen by Webster, J. Arnold Arbor. 48: 354. 1967). Seven species new to Panama (including two new species and one new subspecies) are reported here; a few new province records are indicated as well. These additions make it necessary to provide a revised key. LITERATURE Lanjouw, J. 1931. The Euphorbiaceae of Surinam. Amsterdam. , la. Staminate flowers apetalous; stipules and petiolar glands absent; leaves with stellate-lepidote indumentum above, pud La (stamens m lon 10-12; filaments hirsutulous; styles usually 3 times bifid; seeds deciduously stellate, 5-6 — c ntum lepidote 2a. Indum C. fid tatus . Staminate dover un a leaves stipulate (stipules sometimes early deciduous); indumentum various 3a. puis without basal a or petiolar glands; petals in flowers not reduced; pistillate calyx lobes valvate but not reduplica 4a. Stamens 9-12; a glabrous; anthers 0.6-0.8 mm long; petals = staminate flowers not lepidote or with only 1 or 2 scales; seeds mostly 7 mm long or shorte Volume 75, Number 3 1988 Webster & Huft 1117 Panamanian Euphorbiaceae 5a. Leaves more or less oblong-elliptic, pinnately veined; inflorescences mostly 2 cm lon longer; iig of pistillate flowers mostly 10 r longer ............... . C. schiedeanus 5b. Leaves ovate, more or less cordate, (3-)5-veined at base; inflorescences 1 cm long or shorter; onal of pistillate flowers 1-2 mm lon Le seudoniveus 4b. Stamens 14-17, or if fewer, then filaments hirsutulous; anthers 1-1.2 mm long; petals of staminate flowers lepidote; seeds r longer 6a. Stamens (10 or)11-13; leaves pinnately nal pistillate flowers pan 3 or 4 pe raceme; seeds 20-22 mm lon . tenuicaudatus Ób. Stamens = Meg 3-5-veined at base; pistillate flowers mostly 1 or 2 per raceme; seeds 15-18 m . pyriticus 3b. Leaves with basal ens gl ] ; pet lsi I istillate fl ] d pi tillat ly licat 7a. Seeds 3-5 mm long; fruiting pedicels 8-15 mm lon ng 5. C. lanjouwensis 7b. Seeds 16-17 mm long; fruiting pedicels not over long OSO E 6. C. pachypodus 2b. Indumentum not lepidote, the trichomes mostly or entirely ail or dendritic; petals in intilla flowers distinctly reduced or absent a fairly well idis in C. draco); petioles with prominent glands at junction with blade (except sometimes in C. hircinus); inflorescence (at least in part) terminal. 8a. Leaves -lobed; staminate calyx lobes deinen. imbricate in bud; staminate receptacle glabrous; sedi ta tetragonal; annual herb . C. lobatus 8b. AR unlobed or shallowly lobed (less than halfway); staminate calyx lobes mostly alesia i in bud; staminate receptacle sparsely to densely villose (glabrous in S. santaritensis); seeds not tetragonal. 9a. 5m a ove styles mor m long, twice bifid; stipules tomentose, dentate, 3-6 mm long; seeds ventrally ribbed 7. C. speciosus 9b. Stamens fewer than 50; styles less than 5 mm long. ower cymules of inflorescence with both staminate and pistillate flowers at the same nodes; seeds coarsel lla. Styles bifid; pistillate calyx lobes not reduplicate-valvate; stamens 13 or filaments glabrous or nearly so; leaves unlobed, without scattered laminar glands. 12a. Pistillate flowers distinctly Yoke sq the pedicels mostly 3-6 mm long; eie mostly 2- ody stamens 13-20; styles sr or 8. ore; early so; seeds less than 5 mm . draco 12b. Pistillate flowers subsessile, the ed, cels in fruit no t in alae mostly 1-2 dm long; stamens 25-45; iie distinctly stellate- cent; seeds at least 5 m . Styles mui, stellate-pubescent; pistillate calyx lobes reduplicate- -valvate; sta m ; filaments hirsutulous; leaves more or less 3-lobed, with ud tate o glands above smithianus 10b. Lower oe of bisexual inflorescences with acad eee flowers (not with staminate flowers at the same nodes), or inflorescences unis 13a. 2 twice bifid to multifid. 14a. - e ca gens rubs or trees; Lap s calyx lobes not strongly unequal, the abaxial ones xx deeply lacer . kana irc not deeply lacerate; staminate res villose. s 14-16; seeds costate or verruco: 7a. *Pistillate calyx lobes elliptic to EE valvate, neither reduplicate nor accrescent, cm iim within, less than 5 mm broad; staminate petals not over 4 mm lo - leaves OT T 7 veined at base, entire; petioles mostly 5- ong . C. billbergianus 17b. Pale calyx lobes ovate, e caga os accres- c m stellate-tomentose within, becom m long broad; staminate petals 4.5-5 flee ng; leaves most- y 5-veined at base, entire to denticulate; in 1-4 ong C. fragrans 16b. Sc 10-12; seeds smooth; staminate E ‘sparsely villose. 18a. Leaves mostly alternate (occasionally opposite at 1 or distal nodes n E at bas dentate (teeth l small or rahe iones, bracts, and pistillate calyx lobes coarsely glandular-dentate; inflorescence with 3- tillate flowers; filaments densely hirsutulous below; ovar usually densely stellate-tomentose .................... . C. hircinus posite or ternate distally, pin- nately veined or obscurely triplinerved, the basal laminar glands long-stipitate; stipules and bracts entire; pistillate 1118 Annals of the Missouri Botanical Garden calyx lobes entire or obscurely dentate; pinna m. l or 2 (rarely 3) pistillate flowers; filament y sparsely stellate-pubescent apically 15b. Pistillate Sus deb lacerate; staminate receptacle glabrous; sta 14b. Annual herbs; pile calyx lobes very unequal, t red , the abaxial ones lacerate, accrescent (6-8 mm lon 16. . santaritensis the adaxial ones grea C. argenteus eaves coarsely and ai serrate (major teeth usually not over 10 per adin side); stamens 8-10; styles less than 1.5 mm long, sprea C. trinitatis 19b. Leaves more finely and/or bluntly toothed; stamens mostly 11 or 12; styles ng or more, ascending or erect; seeds minutely beaked. 20a. Stems coarsely hispid; leaves mostly ovate, pointed at the tip, petiolar C. SUM 20b. Stems not coarsely hispid; leaves mostly elliptic or oblong, blunt 32.3. Croton tenuicaudatus Lundell, Phyto- logia 1: 451. 1940. TYPE: Costa Rica. San José: vicinity of El General, Skutch 2575 (ho- lotype, MICH). Additional sau examined. PANAMA. BOCAS DEL TORO: along road to Chiriqui Grande, 10 road mi. from Continental ai an mi. along pipeline access road E of highway, ca. 8%55'N, 82°08’W, 350-500 m, McPherson 8767 (F). 32.5. Croton lanjouwensis Jabl., Mem. New York Bot. Gard. 12: 158. 1965. C. matou- rensis Aublet var. benthamianus Muell. Arg., Linnaea 34: 95. 1865. C. benthamianus (Muell. Arg.) Lanjouw, Euphorb. Surinam 17. . 11(2): 106. 1874. TYPE: Brazil: Rio Negro, Spruce Croton 2 (isotype, NY). Tree to ca. 12 m high; monoecious; twigs densely lepidote. Leaves with petioles lepidote, 1.5-3.5 cm long, distally with 2 large yellowish, subsessile, patelliform glands ca. 1.5-2 mm across; stipules linear-lanceolate, entire, densely lepidote, 8-10 mm long, early deciduous; blades chartaceous, el- liptic to elliptic-oblong, acute to short-acuminate at tip, cuneate at base, mostly 8-15 cm long, 3.5- 6.5 cm broad, smooth and glabrous above, densely lepidote beneath (scales denticulate-margined, ca. 0. mm across and nearly or quite contig- uous); venation distinctly pinnate, the major lateral veins (ca. 11-15 on a side) straight; margins entire and lepidote-marginate. Inflorescences terminal, racemose, 13-15 cm long, bisexual; pistillate flow- ers solitary at 3-6 proximal axils; staminate flowers in cymules of 2 or 3 at distal axils; bracts entire, lepidote, up to 3 mm long. Staminate flowers with lepidote pedicels 2-5 mm long; calyx lobes 5, val- vate, ovate-triangular, lepidote, 2-2.5 mm long; [C. iia] receptacle villose; petals obovate or narrowly ellip- tic, ca. 2 mm long, glandular-punctate, densely villose on margins, glabrous on back; stamens 12 or 13; filaments glabrous or sparsely hirsutulous; anthers elliptic, 0.6-0.7 mm long. Pistillate flow- ers with stout lepidote pedicels becoming 8-15 mm long; calyx lobes 5, equal, triangular, + redupli- cate-valvate, lepidote without, + stellate-lepidote within on the recurved margins, 4.5-5 mm long; disk 5-lobed, adnate to calyx; petals rudimentary (shorter than 1 mm long); ovary densely lepidote (scales ca. 0.6-0.8 mm across); styles twice-bifid, ca. 4 mm long, stellate proximally (branches gla- brous). Capsules not seen entire; cocci lepidote, ca. 5 mm long; columella slender, ca. 4 mm long; seeds plump, scarcely compressed, brownish, smooth, ca. 3.5 mm long The Panamanian plants appear to agree in most respects with C. lanjouwensis as defined by Lan- jouw (1931: 12-17; as C. benthamianus); the broad pistillate calyx with adaxially cine lobes is apparently diagnostic in separating the species m C. matourensis Aublet. However, the Pan- amanian plants occurring in cloud forest at 800- 1,000 m would appear to differ ecologically from the Amazonian plants, which have been collected in lowland rainforests. Jablonski (1965: 157-158) cited C. matourensis from comparable altitudes in Venezuela; further comparisons of plants from Pan- ama with the South American plants evidently are needed. The specimens from Panamá Province dif- fer rather strikingly in their duplex petiolar glands, sparsely lepidote upper leaf surfaces, and shorter pistillate pedicels. Specimens examined. PANAMA. BOCAS DEL TORO: along road to Chiriquí Grande, ca. 1.5 mi. along road E highway, ca. 8?55'N, 82*10^W, 250-350 m, Me- Pherson & Allen 9640 (F). CocLé: Cerro Pilón, near El Volume 75, Number 3 1988 Webster & Huft Panamanian Euphorbiaceae 1119 Valle, ca. 900 m, Duke & Correa 14718 (DAV, MO); cloud forest, hills N of El Valle de Anton, Dressler 4083 1 (MO). PANAMÁ: primary forest, road from El Llano to Carti-Tupile, 200-500 m, Croat 22905 (DAV, MO). 32.6. Croton pachypodus Webster, sp. nov. TYPE: Panama. San Blas: km 18 of El Llano- Carti road, 9?19'N, 78*55'W, 350 m, 1 Oct. 1984, de Nevers & Herrera 3980 (holotype, MO; isotypes, DAV, F). Species haec ab C. pyritico differt foliis supra glabris petiolo glandulato, ab C. lanjouwense differt seminibus majores, ab aliis speciebus dau BM pa differt pe icellis valde incrassatis. Tree 7-25 m high; twigs obtusely angular, gla- brate. Leaves with petioles lepidote, 0.8-1.5 c long, distally (near base of blade) with 2 subsessile or short-stipitate (to 0.7 mm) blackish patelliform glands 0.5-1 mm across; stipules linear-lanceolate, entire, densely lepidote, 4-10 mm long, 0.9-1.2 mm broad, + persistent; blades chartaceous, ellip- tic-oblong to somewhat obovate, subacute or acute to abruptly short-acuminate at tip, cuneate to rounded at base, 8-22 cm long, 3.5-7 cm broad, smooth and glabrate above (with sparse scales on major veins when young), evenly and sparsely lep- idote beneath (scales 0.25-0.4 mm across, den- ticulate, with ca. 50 radii), the scales widely sep- arated; venation distinctly pinnate, the major lateral veins (9-13 on a side) straight or slightly curving, the veinlet reticulum prominulous beneath; margins entire, smooth, without lepidote rim. n ci terminal and axillary, racemose, (5- 5 long, bisexual or staminate; pistillate io soli- tary at lowermost (1-)2-4(-5) nodes of bisexual inflorescences, staminate flowers solitary or paired at distal axils; bracts triangular, blackish, sparsely lepidote, 0.5-1 mm long. Staminate flowers (buds only observed) with lepidote pedicels 1.5-2.5 mm long; calyx lobes 5, valvate, triangular, lepidote, ca. 3 mm long; receptacle villose; petals narrowly elliptic, ca. 2.5 mm long, densely villose on mar- gins, sparsely lepidote (often a single scale) on the back, densely hirsutulous adaxially; stamens 14- 16, the filaments glabrous, the anthers 0.8-0.9 mm long. Pistillate flowers with stout lepidote ped- icels ca. 2.5-3.5 mm long, becoming in fruit 4.2— 6.5 mm long, 3.2-4 mm broad; calyx lobes 5, equal, triangular-ovate, reduplicate-valvate, dense- ly lepidote without, densely tomentose-villose with- in, 3-3.5 mm long, 2.5-3 mm broad; disk shal- lowly 5-lobed, nearly 4 mm across, smooth and glabrous; petals obsolete, represented by whitish tufts of hairs; ovary densely lepidote (scales 0.5— 0.8 mm across, denticulate, 50—70-radiate, with 20-30 darkened radii); styles blackish, twice-bifid, 3.5-4 mm long, nearly glabrous. Capsules not seen entire; valves of cocci ca. 23-25 mm long; colu- mella slender, ca. 20 mm long; seeds elliptic, some- what compressed, flattened and obscurely carinate on the back, distinctly keeled on inner face, grayish brown, smooth and shining, 16.2-17.2 mm long, 10.3-11.3 mm broad; caruncle hippocrepiform, obscure, tenuous, ca. 2.5 mm long, 1.5 mm broad. Collections of this species have been determined as C. lanjouwensis, to which indeed it is related and superficially very similar. However, it is dis- tinguished by leaves very sparsely lepidote beneath and with margins free of scales, shorter petioles with smaller darker glands, and especially by the much larger fruits borne on greatly thickened ped- icels. Among species earlier reported from Panama, the new species resembles C. tenuicaudatus; how- ever, that species has eglandular leaves lepidote on both faces, strongly lepidote staminate petals, and more slender pistillate pedicels. Croton pyri- ticus appears to be even less similar because of its ovate palmately veined eglandular leaves, longer and more slender (1.5 mm or thinner) pistillate edicels, and verruculose capsules; however, the seeds, although larger, are similar in shape to those of C. pachypodus. Additional Cada examined. PANAMA. SAN BLAS: mi. above Pan-American Hwy., 200-500 m, Croat 22905 (MO); 20. 7 km from Pan-American Hwy., 350 m, Mori & Kallunki 5116 (MO). 32.7. Croton speciosus Muell. Arg. [Linnaea 34: 83. 1865] subsp. tacarcunensis Web- ster, subsp. nov. TYPE: Panama. Darién: Cerro Tacarcuna, S slope, premontane wet forest on ridge below summit, 1,250-1,450 m, Gentry & Mori 13925 (holotype, MO; isotype, DAV). a subsp. specioso differt stipulis minoribus, glan- dulis paki: brevioribus, carunculo seminis ca. 2 mm la Monoecious tree 5 m high; twigs subterete, densely tawny-villose with dendritic hairs. Leaves with petioles (2-)3-11 cm long, tomentose, api- cally with 4—6 stalked glands on ventral side, the glands ca. 0.8-1.5 mm long, 0.3-0.4 mm across; stipules lanceolate, densely tomentose, toothed, 3- 6 mm long; blades membranous, mostly ovate, long-acuminate at tip, rounded to subcordate at base, the larger ones shallowly 3-lobed, 12-21 long, 7-12 cm broad; lamina above copiously pu- 1120 Annals of the Missouri Botanical Garden bescent with stellate-tufted hairs, beneath copiously tomentose with whitish dendritic hairs ca. 0.5- mm across, 3—5-nerved at base, with 7-10 major lateral veins on each side, connected by a scalari- form reticulum of straightish veinlets; margins sub- entire (obscurely denticulate). /nflorescences ter- minal (or pseudolateral), bisexual, racemose, 2.5- 6 cm long, with 1 or 3 proximal flowers and 3-7 distal flowers; flowers solitary at each node; bracts 5-7 mm long, attenuate-acuminate, tomentose, with subulate stipules ca. 3-5 mm long. Staminate flow- ers with stellate-tomentose pedicels 5-8 mm long; receptacle densely tomentose; calyx lobes 5, ful- vous-tomentose, obtuse, entire, 3.5-5 mm long; petals obovate, 4.5-5 mm long, densely appressed- pubescent without, glabrous within, woolly-villose along margins; stamens ca. ; flaments slen- der, glabrous, ca. 4-5 mm long; anthers oblong, apiculate, 1.2-1.6 mm long, 0.5-0.7 mm broad. Pistillate flowers with stout tomentose pedicels ca. 2-4 mm long at anthesis (becoming up to 1 cm long in fruit); calyx segments 5, valvate, oblong, densely whitish- to fulvous-tomentose without, sparsely tomentose in distal third within, ca. 10 mm long, 3-5 mm broad; disk inconspicuous, ad- nate to base of calyx, crenulate, stellate-pubescent, ca. 5 mm across; petals rudimentary, densely hir- sute, ca. 1-1.5 mm long; ovary densely fulvous- tomentose; styles twice-bifid near the base, densely stellate-pubescent below (and with scattered stellate hairs distally nearly to tips), ca. 7-9 mm long, the branches dark reddish, dilated and crenulate at tips. Capsules subglobose, fulvous-hispidulous, ca. l cm long and broad; seeds plump, plumbeous brown, distinctly costate ventrally, obscurely cos- tate on back, ca. 7 mm roundish, low, ca. long, 5 mm broad; caruncle 2 mm across. This striking Croton from the cloud forests at the crest of Cerro Tacarcuna appears to be con- specific with C. speciosus, which was described from specimens collected near Caracas, Venezuela. I have examined Moritz 1329 (A, GH) from Gali- pán, near Caracas, one of the syntypes cited by Mueller (incorrectly located by him as in Colombia rather than Venezuela), as well as three other col- lections from near Galipan (Allart s.n., Pittier 221, 9577, A, GH). These plants in general rather closely resemble the Cerro Tacarcuna specimens in leaf shape and pubescence, flower configuration, and seeds; there can be little doubt that we are dealing with a single species. The stamen number in the Venezuelan plants varies from 40 to 80 and hence includes the number for the Cerro Tacarcuna plants. Mueller gave the stamen number of C. speciosus as ca. 150, which certainly does not agree with Moritz 1329, in which two buds yielded ca. 70 and ca. 80 stamens. Possibly the number may become higher in some Venezuelan plants, as Muel- ler (1866: 529) also cited two other collections from near Caracas, Fendler 34 and 231 (cited as Linden 34 and 201), which we have not seen. The publication of a separate subspecies for the Tacarcuna population is made diffidently, as in- tervening collections may close the gap. However, the Panamanian plants differ strikingly in their much smaller and less lacerate stipules, as well as in having distinctly shorter stalked glands at the apex of the petiole. Furthermore, the caruncle in seeds from Panama is roundish and ca. 2 mm broad, whereas it is distinctly laterally expanded and ca. 3 mm broad in seeds from Venezuela. Provisionally, therefore, it seems best to designate the Cerro Tacarcuna plants as a distinct subspecies. Additional specimens examined. PANAMA. DARIÉN: 5 slope of westernmost peak of Cerro Tacarcuna, 1,100- 1,300 m, Gentry et al. 16877 (MO); ridgetop below Alto de Nique base camp, Gentry et al. 28727 (MO) 32.8. Croton draco Cham. & Schldl. [ Linnaea 6: 360. 1831 | subsp. panamensis (Klotzsch) Webster, stat. nov. Cyclostigma panamense Klotzsch in Seem., Bot. Voy. Herald 105. 1853. Croton panamensis (Klotzsch) Muell. Arg. in DC., Prodr. 15(2): 546. 1866. TYPE: Panama. Chiriqui: Volcán Chiriqui, Seemann (K, not seen). (see Webster & Burch, 1968: 254 for additional synonymy.) Further examination of Mexican and Central American specimens of Croton draco indicates that the Panamanian plants cannot reasonably be main- tained as a separate species. Except for the larger, broader stipules (mostly 2 mm or more across), the Mexican populations here assigned to subsp. draco show no essential differences from plants with nar- row stipules that occur from Guatemala (and spo- radically in southern Mexico) to Panama and Co- ombia. 32.10. Croton smithianus Croizat, J. Arnold Arbor. 21: 93. 1940. T tander: Mesa de los Santos, Killip & Smith 15283 (holotype, A; isotype, US Tree to 18 m high; twigs angled or sulcate, yellowish-scurfy with pedicellate stellate hairs. Leaves with petioles mostly 5-15 cm long; patel- Volume 75, Number 3 1988 Webster & Huft Panamanian Euphorbiaceae 1121 liform glands at apex of petiole sessile, 0.9-1.2 mm across; stipules linear-lanceolate to spathulate, 5-9 mm long, 1 -2 mm broad; blades mostly ovate, sometimes 3-lobed, blunt to acuminate at the tip, cordate to subcordate at base, the larger ones 15- 35 cm long, 10-30 cm broad; venation palmate, with 5(- each side above the base, the laterals sometimes dichotomizing towards the margin; veins and vein- lets prominulous on both sides, the veinlets sca- 7) major veins at base, 5-8 laterals on lariform; trichomes on upper surface pedicellate- stellate, 0.2—1 mm across, with stalks 0.1—0.5 mm high, beneath denser and + floccose; small patel- liform glands (0.4-1 mm across) occasional on upper surface; margins denticulate, with occasional small stalked glands. /nflorescences terminal, most- ly 20-50 cm long; 5-15 proximal cymules bisex- ual, distal ones staminate; staminate bracts subu- late, 1-2.5 mm long, subtending several flowers. Staminate flowers with stellate-tomentose pedicels 1.5-5 mm long; calyx entire, deciduous, ca. distinctly gamophyllous, 3.5-4 mm long; calyx lobes ovate, acute, valvate, 2.2-3.7 mm long, 2-2.8 mm broad; petals narrowly spathulate, 3.2-4.5 mm long, 0.5-1 mm broad, densely villose ven- trally, strigose-hirsutulous dorsally; receptacle densely villose; stamens 11 or 12; filaments flat- tened, densely hirsutulous in lower Z, 3-4.5 mm long; anthers elliptic, the connective glandular-pus- tulate, 1.1-1.5 mm long. Pistillate flowers with stellate-lepidote pedicels becoming 9-14 mm long; calyx lobes 5, valvate (not distinctly reduplicate), elliptic to ovate, entire, stellate-lepidote (trichomes 0 mm across) outside, stellate inside near tip and along margins, 5-7.5 mm long, 3.5-6.5 mm broad; disk entire, thickish, ca. 1.7 mm across; ovary yellowish-stellate or stellate-hispid, trichomes 0.5-1.5 mm across in fruit; styles free, multifid, sparsely to copiously stellate-hispid, ca. 5 mm long. Capsules subglobose, yellowish with appressed scales; columella ca. 4 mm long; seeds plump, lenticular, brownish, finely costate-rugulose, blunt- 3 pointed at both ends, 3.9-4.1 mm long, 3.3- mm broad; caruncle flat, bilobed, 1.7-2.1 mm E Croton smithianus is found in lowland and lower montane forests, up to ca. 1,500 m elevation, Nicaragua to Colombia, flowering July to Septem- r. The Panamanian representative of the wide- spread and variable South American species com- plex centering on Croton palanostigma Klotzsch is here referred to C. smithianus Croizat because of its characteristic indumentum of pedicellate stel- late trichomes. The populations in Colombia are rather variable and poorly understood; the isotype at US, which is a fruiting specimen, has the char- acteristic indumentum, and a specimen from Chocó (Archer 2062, US) has the characteristic leaf form and margin, but the trichomes are not distinctly pedicellate as in the Panamanian plants. The Pan- amanian and Colombian plants somewhat resemble Croton killipianus Croizat, described from Bo- yacá; however, the type collection of that species (Lawrance 588; isotype, US) has subentire leaf margins and an appressed, rather sparse indumen- tum more characteristic of C. benthamianus Muell. Arg. Croton nuntians Croizat from Guyana is somewhat similar but differs in its smaller fruiting calyx and shorter fruiting pedicel. Until this species complex is revised, it seems best to refer our plants to C. smithianus. Several collections from Nica- ragua and Costa Rica are also referred to that species. Specimens examined. COSTA RICA. HEREDIA: Finca La Selva, Hammel & Trainer 12849, 13044 (DAV, 400 m, Webster 21883 (DAV, MO); rainforest 17 mi. SE of San Isidro General, 700 m, Webster & Miller 12394 (DAV). NICARAGUA. RÍO SAN JUAN: Sábalo, Ara- quistain 3223 (DAV, MO). zELaYa: Bluefields, Neill 2598 MO). PANAMA. COLON: Rio Salud, Howell 128, Lao & Holdridge 224 (MO). DARIEN: between Manené and Tusijuanda, Duke 13576 (DAV). PANAMA: SE slopes of Cerro Trinidad, Kirkbride & Duke 1665 (MO). 32.11. Croton billbergianus Muell. Arg., Lin- naea 34: 98. 1865; subsp. billbergianus. TYPE: Panama. Colon: Portobelo, Billberg 316 (not seen). This species has been collected in three addi- tional provinces. Additional specimens examin ned. PANAMA. BOCAS DEL TORO: Sürsuba, Río Changuinola, Dwyer s.n. (MO). SAN BLAS: Puerto Obaldia to La Bonga, Knapp & "Mallet 4667 (MO). vERAGUAS: Coquyito beta Río Barrera, Hammel 5221 (MO); Santa Fe, Folsom & Edwards 3392 (MO). The description in the original treatment (Web- ster & Burch, 1968: 257-258) applies only to subsp. billbergianus. As suggested at that time, C. pyramidalis J. D. Smith, extending from Ve- racruz, Mexico, to Honduras, does not appear to be a distinct species. It may be retained at the subspecific level because of its apparently larger seeds (5.7-6.2 mm in the Veracruz population vs. 4.3-5.5 mm in the Panamanian plants) and longer 1122 Annals of the Missouri Botanical Garden Volume 75, Number 3 988 Webster & Huft 1123 Panamanian Euphorbiaceae stipules (7-15 mm long vs. 5-7 mm in the Pan- amanian plants). A new combination for the Mex- ican plant is therefore necessary.^ 32.13. Croton hircinus Vent., Jard. Malmai- son 1: 50, pl. 50. 1804. TYPE: cultivated specimen, Ventenat (presumably at G, not seen). An additional provincial record is cited. Additional specimen examined. PANAMA. CHIRIQUÍ: Río Cobre bridge, 25 mi. W of Tole, Dwyer & Hayden 7542 (MO 32.14. Croton santaritensis Huft, sp. nov. TYPE: Panama. Colón: Santa Rita Ridge Road, 21-26 km from Transisthmian Highway, tropical wet forest, 500-550 m, 9?25'N, 9°37'W, 4 July 1982, Knapp 5882 (holo- type, MO; isotypes, DAV, F, F neg. 62353, PMA). Figure 3 rutex monoecus dense villosus, pilis castaneus stellatis. Folia ovata-oblonga vel lanceolata, basi subcordata ve rotundata, apice Serene glandulae petiolares aliquot e-deltatae margine fimbriatae. u d ni Erin racemosae, masculinae ter- s, femineae axillares; qe femineae Par onde laciniatae flores includente — Shrub ca. 1.5 m high; monoecious; twigs sub- terete, densely villous with brownish stellate hairs. Leaves with petioles densely villous as the twigs, 5-20 mm long; petiolar glands several at apex of petiole, stipitate, trumpet-shaped, 0.5-1.5 mm long, 0.3 m across; stipules ovate-deltate, mem- branous, eglandular, appressed, 8-11 mm long, 4-6 mm broad, the margins fimbriate; blades char- taceous, ovate, ovate-oblong, or lanceolate, long- caudate at apex, rounded to subcordate at base, 10-15 cm long, 4-8 cm broad, 1.7-3 long as broad, sparsely to moderately stellate pu- bescent above, sparsely tomentose below, 3-7- times as ° Croton SPEI AME per pyramidalis (J. D. Smith) Webster, stat. nov. Cro da * xs Bot 3. Verapaz: Rio De near Cubilgüitz, Tüsrchheit 7974 (holotype, US; not seen). nerved at base, the secondary veins 5-8 per side; margin entire to remotely denticulate. /nflores- cences unisexual, racemose, the staminate ones terminal, the pistillate ones axillary, occurring only at the two subopposite nodes immediately below the terminal staminate raceme; staminate inflores- cences 17-22 cm long, densely brown stellate- villous; nodes 15-30; flowers 1-3 at each node on stellately pubescent pedicels 6-9 mm long, the bracts subulate, 2-3 mm long, stellate below, gla- brous above; pistillate inflorescences 5-11 cm long, densely brown-villous; bracts flabellate, deeply la- ciniate, 10-12(-14) mm long, at least the lower ones loosely enclosing the flower buds. Staminate flowers: sepals 5, deltate, joined at base, valvate, stellately pubescent, the lobes 2.5-3 mm long, ca. 2 mm broad; petals 5, only slightly exceeding the alyx lobes, ca. 6 mm long, ca. 2 mm broad below the tip, the tip abruptly expanded, ca. 2.5 mm broad, coarsely erose; stamens 12-15; disk con- sisting of 5 nearly separate glands; receptacle gla- brous. Pistillate flowers ie aaa E or 5, del- tate, fleshy; ovary densely stellate-hispidul tyles twice-divided. Mature fruits and seeds not seen. This species may easily be distinguished from all other Central American species of Croton by the peculiar arrangement of the inflorescences, the conspicuous fimbriate stipules, and the oblong, la- ciniate bracts that loosely enclose the young pis- tillate flowers. No close relative of C. santaritensis is known. Because of its combination of pentamerous calyces in both pistillate and staminate flowers, five petals and glabrous receptacles in the staminate flowers, and the large fimbriate stipules and bracts, it is not easily accommodated in any of the sections rec- ognized by Mueller (1866: 511-700) in the most recent worldwide account of Croton. No species closely resembling Croton santaritensis has been found among the large holdings of South American Croton at the Field Museum or the Missouri Bo- tanical Garden. 32. A re brevipes Pax, Bot. Jahrb. Syst. 290. 1903. TYPE: Costa Rica. Rio del vH Pittier 12117 (isotype, US, photo, F) ;URE 3. bles gem ism from st tem.—F. Pa Richardso Croton santaritensis. — A. Habit, with staminate and pistillate 47 ap ird — B. Detail of leaf wer leaf surface. —E. Trichome ht. —H. die nal . Trichome from lo 1124 Annals of the Missouri Botanical Garden Shrub 1-3 m high; monoecious; twigs pale, ap- pressed-stellate. Leaves alternate below, mostly op- posite or ternate above; petioles densely appressed- serene 3-20(-30) mm long (less than Y, length blade); petiolar glands (at base of blade) con- saa cylindrical, apically truncate and dilated, 1-2.5 mm long, 0.3-0.5 mm across; stipules subu- late to narrowly lanceolate, dark, entire, eglan- dular, stellate-pubescent, 1.4-2.8 mm long; blades thinly chartaceous, elliptic to ovate-elliptic or ob- ovate, acute to acuminate at tp, cuneate to obtuse at base, 4-13 cm long, (1-)2-5 stellate or appressed-hispid above with few-rayed > em broad, sparsely trichomes, sparsely appressed-stellate and incon- spicuously glandular-punctate beneath, pinnately veined (or inconspicuously triplinerved) with mostly 5-7 veins on each side; margins subentire to rather coarsely and irregularly dentate (teeth ca. 8-15 on a side), with stalked glands between some of the teeth. Inflorescences mostly terminal and bisexual (some also lateral and staminate), racemose, 1.5- 3(-4.5) cm long, with 1 or 2 (rarely 3) basal solitary pistillate flowers, the staminate flowers 1 or 2 per bract at distal axils; bracts narrow, entire, eglan- dular, stellate-pubescent, mostly 1.5 shorter. Staminate flowers with sparsely stellate mm long or or nearly glabrous pedicels 1-2 mm long; calyx lobes 5, elliptic-lanceolate, acute, stellate-pubes- mm long, 0.9— cent, glandular-punctate, 1.2- 1.7 .1 mm broad; receptacle moderately villose; petals obovate-spathulate, 1.4-1.8 mm long, glandular- punctate, barbate-hirsute on lower margins; sta- mens 10-12; filaments glabrous, 1.8-2.5 mm long; anthers ovate, 0.5-0.7 mm long. Pistillate flowers with stout appressed-stellate pacion becoming 1.3- 3.5 mm long; calyx lobes 5 oblong-lanceolate or spathulate, subentire (with 1 r 3 obscure teeth), sparsely stellate outside, gla- 5-6.5 mm long, 1-1.5 mm broad; 5, subequal, narrowly brous inside, 3. petals rudimentary, subulate; disk entire, glabrous; ovary sparsely stellate- ps ent apically, glabrous m long, twice bifid, pue subglobose, below; styles free, ca. glabrous to hispidulous. sae stellate-pubescent or glabrescent, ca. 5 iam.; columella slender, 3.2—4 mm long; seeds apically beaked, brownish, nearly smooth (minutely striolate), 3.8- broadly ellipsoid, compressed, 4.1 mm long, 2.8-3.4 mm broad, the caruncle small, ca. 0.5-0.8 mm across. Rainforest below 1,000 m, Costa Rica and Pan- This plant bears a considerable resemblance in habit to C. hircinus but differs in having (distally) opposite or ternate pinnately veined leaves with larger laminar glands; the stipules, bracts, and ca- lyx lobes lack the glandular serrations of C. hir- cinus. Examination of material of Croton macro- dontus Muell. Arg. from Mexico shows that it is extremely close to C. brevipes. Although specimens from Costa Rica and Panama may be easily rec- ognized by their distally opposite, less coarsely toothed leaves with more rounded bases and shorter petioles, they are very similar to the Mexican plants in most details, including pubescence, floral details, and fruits. The seeds of the Mexican plants are somewhat larger, but this difference may disappear upon further sampling. Provisionally, the two species may be kept distinct on the basis of the foliar characters, and because no intermediate popula- tions have yet been discovered in Central America between Costa Rica and Mexico. Specimens examined. PANAMA. COLÓN: Santa Rita Ridge, ca. 300 m, Antonio 3739 (DAV, F, MO), Correa & Dressler 912 (F, MO), Croat 13898 (MO), Duke 15291 (MO), Dwyer 8543 (MO), Dwyer & Gentry 9395 F, MO), Foster 1751 (DAV, DUKE, F), Gentry 1874 (DAV, F, MO), Kennedy 2756 (MO), Knapp 5845 (DAV, F, MO), Sytsma 2047 (MO), 2054 (F, MO), Webster & Dressler 16727 (DAV, MO, US); East I. Duke 15291 (DAV). PANAMÁ: Cerro Jefe, 700-750 m, Dressler 3844 e MO, US), Webster & miden 16477 (DAV, DUKE, MO, US); Torti Arriba, Folsom et al. 6644 (DAV, MO). pu — 32.16. Croton argenteus L., Sp. Pl. 1004. 1753. dr roton argenteus (L.) Didr., Vi- Meddel. Dansk N Foren. Revista Argent. ag 10: 125. 1943; C Cor rell & Johnston, Man. Vasc. Plants Texas 939. 1970. TYPE: America (not seen; presum- ably in Hortus Cliffortianus Herbarium, BM; 1140.8 in LINN). Annual herb 2-10 dm high; stems pseudodi- chotomizing, with long internodes and pseudover- ticels of leaves, appressed stellate-puberulent. Leaves with petioles 1-5 cm julio apical glands; stipules subulate, (2.5- ): m long; blades chartaceous, ovate or the upper 1 long, these without -1 ones oblong-ovate, obtuse or rounded to subacute at tip, cuneate to rounded at base, mostly 3- 7(-15) cm long, 2.5-5 cm broad, 5-veined at base; lateral veins above base 3 or 4 on each side, not prominent; above green and finely appressed-stel- late, beneath grayish and more densely stellate; margins finely serrulate. /nflorescences terminal, bisexual, ca. 1-4 cm long; bracts subtending sol- itary flowers, the pistillate flowers 4— spike; staminate bracts subulate, entire, ca. 3 mm Volume 75, Number 3 1988 Webster & Huft Panamanian Euphorbiaceae 1125 long. Staminate flowers with pedicels ca. 1.5-2.5 mm long; calyx lobes lanceolate, acute, valvate, ca. 1.5-2 mm long; petals linear, 2.1-2.3 mm long, 0.3-0.4 mm broad, glabrous except for the ciliate margins; receptacle copiously villose; sta- piously appressed-hirsutulous, ca. 2-2. anthers elliptic-oblong, 0.6-0.8 mm long. Pistil- late flowers with short pedicels ca. 1 - 1.5 mm long, becoming 3-5 mm long in fruit; calyx lobes 5, imbricate, very unequal, the 3 abaxial lobes much larger, in fruit 6-8 mm long, 2.5-6 mm broad, oblong, laciniate, provided on each side with 5-10 teeth ca. 0.5-3 mm long, the 2 abaxial lobes much smaller, nearly or quite obsolete; petals absent; disk strongly asymmetrical, with larger adaxial lobes 0.9-1.2 mm long, 0.5-0.7 mm broad, the 2 adax- ial lobes very small; ovary stellate-tomentellous; styles erect, distally quadrifid, hispid-stellate, ca. 2-4 mm long. Capsules ca. 5 mm long; columella 3-4.5 mm long; seeds ellipsoid, smooth, mottled gray and brown, apically beaked, 3.1-4 mm long, 2.4-2.9 mm broad; caruncle ca. 1.5 mm broad. Scattered in weedy habitats from ext th ern Texas to Panama, reappearing in Venezuela, Paraguay, and Argentina, but not reported from most of tropical South America. The two Pana- manian collections may represent recent introduc- tions. Specimens — ipm PANAMÁ: marsh area 2 mi. S of Tocumen Air yson & Clewell 5899 (MO); Rio Tapia, Bartlett & E 16629 (MO). Subfamily IV. EUPHORBIOIDEAE 33. Mabea Mabea Aublet, Hist. Pl. Guiane 867. 1775. TYPE: Mabea piriri Aublet. RECENT LITERATURE Hurt, M. J. Notes on Mabea (Euphorbiaceae) 1987 in Central America, together with comments on sect Apodae in Brazil. Ph eee 62: 339-34 STEINER, K. E. 1983. Pollination of Mabea occidentalis (Euphorbiaceae) in Panama. Syst. Bot. 8: 105- The discovery of a distinctive new species of Mabea in Panama makes it necessary to provide a new key to the three species now known from the country. KEY TO THE SPECIES OF MABEA IN PANAMA la. Leaves acute or subacuminate at apex, usually imes as long as mra iced glands of bs staminate cymules not elevated above the rachis; stamens 15-20 per staminate flower ... 1. M. montana lb. Leaves cuspidate to long-acuminate at apex, usually less than 3 times as long as broad. 2a. ppl glands of the staminate cymules portion of the style at maturity 12-20 m long 2. M. occides . Bracteal glands of the staminate cymules 2b elevated above the rachis; stamens 10-15 per staminate flower; undivided portion of the style at maturity 4-9 mm long ............ 3. M. jefensis 33.1. Mabea montana Muell. Arg. in DC., Prodr. 15(2): 1151. 1866. TYPE: Colombia: Schlim 1132. Venezuela: Fendler 24. Pan- ama: Sutton Hayes 715 (syntypes, not seen). PANAM Additional collections examined. 10314 (MO); 5 km S of Santa Fe, below 500 m & Collins 1644 (MO); a 1 km del puente sobre el desvio del Rio San Juan, Luna 47 (MO). 33.3. Mabea jefensis Huft, Phytologia 62: 341. 1987. TYPE: Panama. Panama: newly bull- dozed trail off Cerro Jefe Road, 0.4 km beyond turnoff to Alto de Pacora, 29 Sep. 1975, J. T. & F. Witherspoon 8570 (holotype, MO, F neg. 62352). Figure 4. Montane and premontane rainforests of Central Panama, 350-1,000 m. Mabea jefensis is known from abundant collec- tions from both the Cerro Jefe area and from the Continental Divide north of El Llano in eastern Panamá Province, as well as from a single collec- tion from the Cariazas mountain range in the west- ern part of the province. It seems likely that this species will prove to be common in the mountainous region along the border of Panamá and San Blas provinces as this area becomes more thoroughly Additional specimens examined. PANAMA. PANAMA: Cerro Jefe area, Antonio et al. 3399 (F), Correa et al. 1601, 1610 (MO), Croat 13031, 14438 (MO), D'Arcy & D'Arcy 6253 (MO, 2 sheets), D'Arcy 12185 (MO), 12201 (F), D'Arcy & Sytsma 14733 (F 9474 (MO), Dwyer et al. 7296 ( (MO), Folsom et al. 6709, 7105 (MO), (MO), Knapp 867 (F), Liesner 531 (MO) 5072 (MO), Sytsma 1475, 4112 (F), Sytsma et al. i Hufi & Knapp 1594, 1613 (MO), Huft et al. 1868 1126 Annals of the Missouri Botanical Garden 30M FIGURE 4. Mabea jefensis. —a. Branch with inflorescence at anthesis and with young fruits.—b. Branch with mature fruits. Based on Correa et al. 1601 (a), D'Arcy & Sytsma 14733 (a & b). Illustration by Steve Wilson. Volume 75, Number 3 1988 Webster & Huft 1127 Panamanian Euphorbiaceae (MO), Knapp 1396 (F), Knapp et al. 4728 (F), Liesner 303 (MO), Maas et al. 1758 (MO), Mori & Kallunki 1864 (MO), Nee et al. 8752 (MO), Sytsma 960 (F, MO); añazas mountain chain, near Rancho Chorro, above Torti Arriba, 400-700 m, Folsom et al. 6709 (MO). 34. Senefeldera Senefeldera C. Martius, Flora 24 (Beibl.): 29. 1841. TYPE: Senefeldera multiflora C. Mar- tius. Monoecious shrubs or trees without evident milky latex; glabrous throughout. Leaves alternate (or pseudoverticillate at ends of branches), simple, pet- iolate; stipules deciduous; blades + entire, pin- nately veined, usually glandular on midrib at base; margins entire. Inflorescences terminal, paniculate (of compound spikes or racemes), bisexual. Sta- minate flowers solitary or in glomerules at distal axils of inflorescence axes, subtended by biglan- dular bracts; calyx 3-5-lobed, sometimes asym- metrical, not covering anthers in bud; petals and disk absent; stamens 5-12; an elevated receptacle, extrorse, dehiscing longi- tudinally; pollen grains subglobose, tectate, 3-col- porate; pistillode absent. Pistillate flowers solitary at proximal nodes of inflorescence, sessile; calyx 3-parted, segments distinctly imbricate; petals and disk absent; carpels 3, unappendaged, each with a single ovule; styles unbranched, free, or basally connate. Fruits capsular, thin-walled; columella slender, usually not persistent; seeds solitary in each locule, plump, carunculate; endosperm co- anthers subsessile on pious. As treated by Jablonski (1965: 171-174), Se- nefeldera is a genus of nine rather poorly under- stood South American species. It is here recorded from North America for the first time. 34.1. Senefeldera testiculata Pittier, Contr. Fl. Venez. 2: 31. 1923. TYPE: Venezuela. Zu- lia: Perijà, Pittier 10910 (US). Shrub or small tree to ca. 4 m high; twigs subterete, channeled, smooth. Leaves with petioles variable in length, 0.5-4 cm long, adaxially chan- neled; stipules triangular, ca. 2 mm long, deciduous (leaving conspicuous scars); blades chartaceous, elliptic-lanceolate, acuminate at apex, cuneate at base, 12-24 cm long, 4-10 cm broad, usually with a ventral median swollen gland ca. 0.5-1 mm long; major lateral veins ca. 10-15 on a side, slightly curving to margins, the veinlets prominulous on both sides; margins plane to slightly reflexed, en- tire. Inflorescences usually bisexual, mostly 7-20 cm long; lateral axes 5-8, the peduncles 5-11 mm long. Staminate flowers 2 or 3 per node; bracts —1.5 mm long, acute, the glands infolded on adaxial side; pedicels 0.5 mm long or shorter, sub- tended by 1 or more bractlets within the bract; calyx segments 3, unequal, acute, as broad as or broader than long, ca. 0.5-0.7 mm long; stamens 5; anthers apiculate, ca. 0.4-0.5 mm long (much longer than the very short filaments). Pistillate flowers solitary at the 2 or 3 lowermost nodes of each lateral axis, sessile; bracts apiculate, 1-1.3 mm long, with glands 0.8-1 mm across; calyx segments 3, slightly imbricate, broadly ovate, apic- ulate, ca. 1.2-1.3 mm long; ovary smooth and unappendaged, 3-locular, carinate; styles nearly free, ca. 1.5 mm long, falcate, thickened below and tapering to an acute tip, ventrally papillate. Capsules oblate, ca. 11.5-12 mm diam., 3-lobed, inconspicuously reticulate and ribbed, cocci ca. 7 m long; columella ca. 5 mm high, unthickened, not persistent; seed plump, ovoid, ca. 5-6 mm long, brownish, streaked or mottled, smooth, + notched at base; caruncle small; hilum subapical, raphe conspicuous. Lowland evergreen rainforest, Panama to Ven- ezuela; here reported from Panama for the first time. The Panamanian specimens are a rather good match for collections of S. testiculata from Zulia, Venezuela (Steyermark 99576, 99917, VEN). Al- though they differ from the South American plants in having more acute leaf bases and shorter inflo- rescences, there seems little doubt that they are Lacu dd PANAMA. DARIÉN: Rio Can O); folium in the original treatment); Estero Grande off Rio vig Duke 10962 (F, MO); Rio Ucurganti, Bristan daa Manené to Rio Coasi, Hartman 12213 35. Sebastiania Sebastiania Sprengel, Neue Entd. Pflanzenk. 2: 18 . 1821. TYPE: Sebastiania brasi- liensis Spreng. The discovery of an additional species of Se- bastiania makes it necessary to provide a key to the two Panamanian species. A third Panamanian species, too fragmentary for identification, is dis- cussed but not included in the key. 1128 Annals of the Missouri Botanical Garden KEY TO THE SPECIES OF SEBASTIANIA IN PANAMA la. Annual herb; leaves linear-lanceolate; staminate igi gs iue a on the rachis; seeds not o l. S. corniculata ; Shrub °> 3m apres ee elliptic-lanceolate; staminate flowers borne spirally on the rachis; seeds ca. 4 mm long 2. S. panamensis — = 35.2. Sebastiania panamensis Webster, sp. nov. TYPE: Panama. Chiriqut: N of San Félix at Chiriqui-Bocas del Toro border, on Cerro Colorado copper mine road, 5,000-5,500 ft., 3 May 1975, Mori & Kallunki 5786 (holo- type, DAV: isotype, MO). À speciebus sect. Microstachydi recedit floribus mas- culinibus spiralibus, seminibus longioribus; a speciebus sect. Elachocrotoni differt habitu; a speciebus sect. Se- bastiania differt ovario armato Shrub ca. 2-3 m high, the trunk + unbranched; twigs slender, subterete, antrorsely + appressed pubescent. Leaves with petioles 3-7 mm long, ap- pressed-pubescent; stipules triangular-lanceolate, dark, 0.8-1.2 mm long; blades thinly chartaceous or membranous, caudate- acuminate at tip, cuneate at base, (2.5-)5-12 cm long, (1-)2-4 cm broad, concolorous, without lam- inar glands, glabrous to distinctly hirsutulous on both faces; midrib plane above, distinctly raised elliptic-lanceolate, + beneath; major lateral veins ca. 8-20 on a side, straightish, slightly prominulous beneath, distally anastamosing into intramarginal loops; forming a delicate inconspicuous reticulum; mar- gins finely crenulate with ca. 8-25 appressed teeth on a side. Inflorescences opposite leaves (some- veinlets times pseudoterminal), spiciform, usually bisexual, 2-3 cm long, rachis + hirtellous; pistillate flowers solitary at base, staminate flowers 1 or 2 per distal ract; bracts spirally arranged, lanceolate, dark, ca. 1 mm long, on each side with a short-stipitate (ca. 0.5 mm) cyathiform gland 0.5-0.8 mm across. Staminate flowers: pedicel less than 0.5 mm long; calyx 3-lobed, the lobes obovate, ca. 0.6-0.8 mm long, the anthers ca. 0.3 mm long. Pistillate flow- ers: subsessile or the pedicel up to 1.5 mm long at anthesis, becoming up to 4 mm long in fruit; calyx lobes 3, imbricate (covering ovary in bud), 0.7-1 mm long, eglandular within; ovary glabrous, 3-carpellate, each carpel with a pair of subapical horns; styles nearly free, spreading, tapering, ca. 1.5-2.5 mm long. Capsules ca. 6 mm high, ca. 7 mm broad, with 6 subapical lower triangular processes; columella ca. 4.5 mm high; seeds ovoid- ellipsoid, reddish brown and mottled, smooth, ca. mm long; caruncle nearly 1 mm broad. Montane rainforests or cloud forests, western Panama. This shrubby species, now represented by sev- eral collections from montane rainforests in Chi- riqui and Veraguas, does not appear to have been previously described. In overall appearance and floral characteristics it resembles the weedy S. cor- niculata of section Microstachys (Adr. Juss.) Muell. Arg. but differs in its woody habit. Its spirally arranged staminate flowers separate it from the woody species of that section. In the treatment of Pax (Pflanzenreich 85: 89ff. 1912) it would key to section Elachocroton (F. Muell.) Pax, but it does not resemble any of the species in that section except possibly S. stipulacea (Muell. Arg.) Muell. Arg.; with broader leaves and multicornute ovaries. The that species, however, is entirely glabrous, Panamanian plants do not fit into section Adeno- gyne (Klotzsch) Benth. because the staminate ca- lyx is not asymmetric, the fruit is echinate, and the branches are not spiny. Nor do the Panamanian plants agree with species in section Sebastiania, because of their leaf-opposed spikes, broader sta- minate calyx lobes, and echinate fruits. The sectional divisions used by Pax do not seem to be very well founded, so that a satisfactory systematic placement of S. panamensis does not seem to be feasible until the genus is monographed. At present, it appears that 5. panamensis may represent a possible connecting link between the species of section Microstach ys, with leaf-opposed spikes and echinate fruits, and the sections with terminal or axillary spikes and usually unarmed fruits. Additional specimens examined. PANAMA. CHIRIQUÍ: Cerro Colorado, 1,200-1,500 m Mori & Dressler 7827 (MO, dupl. at SCZ seen by M. Huft), d 391 (MO), Antonio 1409 (MO, dupl. a at PMA n by M. Huft); above San Félix along mining ro Ca l; aaa 1,500 m, Croat 33044 (DAV, m near San Félix, 1,200 m, Croat 33437 (MO); Chiriqui Trail, pennae rainforest between Pinola and Quebrada Honda, bride & Duke 898 (DAV, MO), Churchill & Churchill 6083 (F, MO); La Fortune hydroelectric project, 1,100- 1,200 m, Hammel 2165 (DAV, MO), Knapp 4980 (MO), Mendoza et al. 110 (MO); E of Fortuna campsite, Folsom & Dressler 5299 (DAV, MO). vERAGUAs: Cerro Tute, [up 1,400 m, Antonio 1817 (MO), Knapp & Kress 4359 (F, MO), Knapp & Sytsma 2562 (F, MO), Mori & Kallunki 5232 (DAV, MO), Mori et al. 7609 (DAV, MO). 35.3. Sebastiania sp. A. Several fruiting collections from the dry Pacific coast of Panama and Costa Rica and from thorn Volume 75, Number 3 1988 Webster & Huft Panamanian Euphorbiaceae 1129 scrub on the Caribbean coast of Colombia represent a species otherwise unknown in Central America. In its membranous, venose, broadly ovate to rhom- biform leaves with crenate to crenulate margins, it closely resembles such species of sect. Sebasti- ania from southern Brazil and Argentina as 5. brasiliensis Sprengel, S. anisandra (Griseb.) Lillo, and S. warmingii (Muell. Arg.) Pax. Even closer is $. macrocarpa Muell. Arg. of Ceara in north- eastern Brazil, with which it shares a large capsule (to 15 mm in diameter). Until flowering material becomes available, it will remain uncertain whether the Panamanian plant represents an undescribed species or belongs to one of the Brazilian species. The Costa Rican specimen cited below is the basis for the record of Ophellantha spinosa Stand- ley cited by Standley from Costa Rica (Standley, 1938: 1557). That species is definitely known only from northwestern Mexico to Honduras. One of the Colombian collections (Gentry & Cuadros 47466A, MO) has an immature inflorescence with a single pistillate flower and the multiparted brac- teal glands that are characteristic of several species of section Sebastiania. _ Spe "cimens examined. Costa RICA. ALAJUELA: vic. of Loros, Brenes 22679 (CR, NY). Pan of Candelaria, Duke 12443 (MO, US); Las Tablas, Dwyer 1100 M Los ini E 30 m, Lao 320 (MO). WA A Peet »: Puerto lombia, 50-100 m, Dugand 626 (F, decode as ç granatensis Muell. Arg.). cl Galerazamba, N ti of Bolivar, thorn scrub forest, 10°48'N, 75?15'W, Gentry & Cuadros 47456, 47466A, 47474 (F, MO). 36. Gymnanthes Gymnanthes Sw., Prodr. 95. 1788. LECTOTYPE: Gymnanthes lucida Sw. (chosen by Grise- bach, Fl. Br. W. Ind. 50. 1859). Aginostemon Klotzsch, Arch. Naturgesch. 7: 184. 1841. . concolor (Sprengel) Muell. Arg. (Gussonia co nA sae Sprengel). me Phatma Klotzsch, Arch. Naturgesch. 7: 181. 1841. TYPE: not designated. Monoecious (rarely dioecious) shrubs or trees; latex scanty and scarcely milky; indumentum ab- sent or of simple hairs. Leaves alternate, simple, petiolate; stipules small, mostly persistent; blades pinnately veined, entire or crenulate, glandular or eglandular. Inflorescences terminal or axillary, bi- sexual (or less commonly unisexual), spiciform, of | -several basal solitary pistillate flowers and many distal staminate cymules; bracts mostly biglandular, subtending solitary pistillate flowers and 1-several staminate flowers. Staminate flowers mostly ped- icellate; calyx of 1 lobe or rudimentary or absent; petals and disk absent; stamens mostly 2-5; fila- ments free or basally connate; anthers extrorse, dehiscing longitudinally; pollen grains subglobose, tectate, 3-colporate; pistillode absent. Pistillate flowers sessile or pedicellate; calyx mostly 3-lobed, the lobes sometimes reduced or obsolete, eglandular within; petals and disk absent; carpels 3, each with a single ovule; styles free or basally connate, un- branched. Fruits capsular; columella + persistent; seeds solitary in each locule, carunculate; testa smooth; endosperm copious. As here circumscribed, Gymnanthes includes Actinostemon Klotzsch and Dactylostemon Eloise: Although most authors have upheld at least Actinostemon as a distinct genus, this seems to have been due more to inertia rather than to critical assessment of diagnostic characters. Pax & Hoffmann (1912: 13) did not provide any con- vincing distinctions in their key, nor did Jablonski (1967: 164, 178), who candidly stated that “the distinction between Actinostemon and Gymnan- thes is very vague." Even when Actinostemon is merged with Gymnanthes, the enlarged genus is difficult to distinguish from Sebastiania. Mueller (1866: 1164-1165), in fact, combined the two, but called the composite genus Sebastiania be- cause the name Gymnanthes (which has priority) seemed inappropriate to him. The distinctly re- duced staminate calyx of Gymnanthes furnishes only a tenuous difference from Sebastiania, in which the staminate calyx is presumably usually well developed. As Pax & Hoffmann (1912: 89- 90) pointed out, Sebastiania may be an unnatural genus, part of which should be combined with Gym- nanthes. Clarification of these difficulties will have to await the efforts of a very intrepid monographer. Rothmaler (1944), following a suggestion by Hallier (1918), adopted the name Ateramnus P. Browne (1756: 338) in place of Gymnanthes. However, this is not justifiable, as has been shown recently (Webster, 1983); Ateramnus is best dis- posed of by lectotypifying it so that it becomes a synonym of Sapium. As here construed, Gymnanthes is then a neo- tropical genus of about 40 species; the two Pan- amanian species are reported here for the first time. LITERATURE Browne, P. 1756. The Civil and UM History of Jamaica. Osborne & Shipton, Hauer, H. 1918. Über Fairidk Browne’ s Gattungen 1130 Annals of th Missouri E ua Garden zweifelhafter Stellung. Meded. Rijks-Herb. Leiden 36: 1-6. 1969. Notes on neotropical Euphorbi- . Monograph of the genus Actinostemon. cri 18: 213-240. Pax, F. € K. e 1912. Euphorbiaceae — Hip- oma m e. In: A. Engler, Das Pflanzenreich IV. 147. V(H "ron Y 5319 (Gymnanthes, pp. 81-88). onam W. ina generica neglecta 1753- 3. Repert. Spec. Me Regni Veg. 53: WEBSTER, G. L. 1983. A botanical gordian knot the case of Ateramnus and eee (Euphorbi- aceae). Taxon 32: 304-305 JABLONSKI, P KEY TO THE SPECIES OF GYMNANTHES IN PANAMA la. Staminate bracts each with 1 gland on each side, subtending 3 flowers; leaves glandular on margins; staminate calyx of 1 segment; fruiting pedicels 13-21 mm long; Ea 6.3-6.6 mm lon G. actinostemoides . Staminate bracts (at least in gb with paired glands on each side, each bract subtending 1 flower; leaves eglandular on margins; staminate calyx absent; fruiting pedicels 40-50 mm long; seeds ca. 5.5 mm long . 2. G. dressleri — = 36.1. Gymnanthes actinostemoides Muell. Arg., Linnaea 32: 103. 1863; Pax & Hoff- mann, Pflanzenreich IV. 147. V(Heft 52): 85. 1912 (as G. actinostemonoides). Sebastiania actinostemoides Muell. Arg. in DC., Prodr. 15(2): 1184. 1866. TYPE: Mexico. Veracruz: Zacuapan, Linden 1357 (holotype, G, mi- crofiche seen Tree to 10-12 m high; twigs of current year obtusely angled, short-puberulent (glabrate in age). Leaves with petioles 3-10 mm long, + puberulent; stipules lanceolate, ca. 1-2 mm long; blades char- taceous, elliptic-oblong, mostly caudate-acuminate at tip (the acumen acute, ca. 0.5-1.5 cm long), acutely cuneate at base, ca. 6-16 cm long, 2-6 cm broad, without laminar glands but with 1—3 depressed cyathiform glands at proximal marginal crenulations; midrib raised on both sides and pu- berulent proximally; major lateral veins ca. 10- 15 on a side, arcuate; veinlets distinctly promi- nulous beneath (slightly so above), forming irreg- ular areoles; margins plane, remotely crenulate (teeth mostly 12-17 on a side), the distal teeth with a minute deciduous glandular tip, the proximal teeth with depressed cyathiform glands. /nflores- cences axillary, racemiform, unisexual or bisexual, 2-7 cm long, the rachis puberulent. Staminate flowers in cymules of 3, subtended by umbonate bracts ca. 1 mm long, each bract with a pair of cyathiform glands 0.6-0.8 mm across attached above the base (subapical); pedicels 0.5-1.5 mm long (longer in central flower); calyx usually of 1 triangular segment ca. 0.5-0.7 mm long; stamens or 3; filaments free, ca. 0.8-1.2 mm long; an- thers ca. 0.4 mm long. Pistillate flowers solitary, ] or 2 per raceme; bract eglandular; pedicel pu- berulent, ca. 2-4 mm long at anthesis, increasing 13-21 mm long in fruit; calyx segments 3, pointed, not imbricate, 0.6-0.7 mm long; ovary unappendaged, smooth, puberulent; styles 3, ba- sally connate for ca. /—Y, their length, 2.5-5 mm long. Capsules not seen entire; cocci 11-13 mm long, smooth; seeds plump, ovoid, ca. 6.3-6.6 mm long, 5.8-6 mm broad, brownish, smooth, obscure- ly beaked; caruncle 0.8-1.3 mm broad. Montane and perhaps also lowland rainforests, eastern Mexico (Veracruz) to Panama. The single Panamanian collection of this species bears only fruits, so the identification of Bristan's plant with a Mexican species must be provisional. However, the characteristic leaf venation and es- pecially the distinctive marginal foliar glands sug- gest that our plant is conspecific with specimens from Veracruz, Mexico (such as Purpus 37995, 8060 from Zacuapan, 4410 from Fortin, all at UC, from which the floral characters have been taken). ecimen examined. PANAMA. DARIÉN.: Rio Pirre, p Bristan 1466 (DAV, MO) 36.2. Gymnanthes dressleri Webster, sp. nov. TYPE: Panama. Panamá: La Eneida, region of Cerro Jefe, 3 Jan. 1968, Dressler 3323 (ho- lotype, MO; isotype, DAV pecies haec aff. G. granatensi, differt foliis integris late ellipticis, cuspidato-acuminatis, glandulis bractearum masculinis duplicatis, calyce nullo. Tree 2 m or more, glabrous; twigs slender, sub- terete, smooth, brownish. Leaves with petioles 3- 5 mm long; stipules ovate, rounded, scarious, ca. | mm long; blades chartaceous, broadly elliptic, rather abruptly cuspidate-acuminate at tip (the acumen obtuse, 0.5-1 cm long), cuneate at base, 3-8 cm long, 1.5-4 em broad, appearing eglan- dular but sometimes with a few scattered minute (diam. ca. 0.2 mm) embedded laminar glands, dis- tinctly paler and glaucous beneath; major lateral veins ca. 6-8 on a side, straightish or distally arcuate; veinlets distinctly prominulous beneath, often as prominent as the laterals, forming areoles partly parallel to the laterals; margins entire, with a plane or slightly reflexed subcartilaginous rim. Inflorescences axillary, mostly at the base or lower axils of annual increments, racemiform, unisexual or bisexual, where bisexual with | or 2 basal pis- Volume 75, Number 3 1988 Webster & Huft Panamanian Euphorbiaceae 1131 tillate flowers and ca. 8-13 distal staminate flowers; bracts all subtending solitary flowers, ovate, scar- ious, ca. 0.5-0.7 mm long, mostly with paired subsessile cyathiform glands on each side at base, the larger gland of each pair ca. 0.3 mm across. Staminate flowers with pedicels ca. 0.4-1.2 mm long, articulated at the top; calyx, petals, and disk absent; stamens 2-4; filaments free or basally united, 0.4-0.7 mm long; anthers 0.4-0.5 mm long. Pistillate flowers with pedicels ca. 3-5 mm long at anthesis, increasing to 40-50 mm long in fruit; calyx lobes 3, ovate, not overlapping, ca. 1 mm long, eglandular within, the margins minutely crenulate; ovary smooth and unappen- daged; styles 3, ca. 1.5-2 mm long, thickish, re- curved, slightly connate at base. Capsules not seen entire; columella 5.5-6 mm long, subpersistent; seeds ovoid-oblong, ca. 5.5 mm long, ca. mm broad, essentially smooth, dark brown, shiny, apically beaked; caruncle ca. 1 mm across. This new species from Cerro Jefe appears to be closely related to G. granatensis Muell. Arg., which was described (Linnaea 32: 107. 1863) from the vicinity of Ocana in northern Colombia. Although the type collection of G. granatensis (Schlim 586) has not been examined, study of a photograph of the type and Mueller's description (in DC., Prodr. 15(2): 1189-1190. 1866) suggests that the Pan- amanian plant differs in some important particu- lars: the leaves are broader, more abruptly cus- pidate, distinctly glaucous beneath, and entire at the margins. The staminate flowers completely lack a calyx, whereas there are two subulate calyx lobes in G. granatensis, and the staminate bracts have duplex glands on each side, while in the Colombian species (judging from Mueller's description), the bracteal glands are single on each side. In the system of Pax & Hoffmann, G. grana- tensis would probably fit best into the Actinostemon, although they listed it under Se- bastiania (Pax & Hoffmann, 1912: 150). How- ever, because of its glabrous inflorescence and well- developed pistillate calyx, it would not readily fit into either of the sections of Actinostemon. Within Gymnanthes (sensu Pax), perhaps the species most similar to G. dressleri and G. granatensis is G. farinosa (Griseb.) Webster”. That West Indian species has somewhat similar leaves but differs in its three-flowered staminate bracts with a single ^ Gymnanthes farinosa (Griseb.) Webster, comb nov. Excoecaria farinosa Griseb., Abh. Ges. Wiss. Goettingen 7: 169. 1857. TYPE: Guadeloupe: Duchassaing (presum- ably GOET, not seen). gland on each side, and in its well-developed sta- minate calyx. Among the species with bracts sub- tending solitary flowers, the closest to G. dressleri appears to be the Cuban species G. albicans (Gri- seb.) Urban; however, in that species the leaves are more elongated and lack cuspidate tips, the bracts are eglandular, and the staminate flowers e 5-12 stamens. 37. Maprounea Maprounea Aublet, Hist. Pl. Guiane 2: 895. 1775. TYPE: Maprounea guianensis Aublet. Shrubs or trees, glabrous throughout; latex nei- ther copious nor milky; monoecious. Leaves alter- nate, simple, petiolate; stipules small, persistent; blades pinnately veined, entire, glandular or eglan- dular. /nflorescences terminal, usually bisexual, of 1—4 solitary, pedicellate, pistillate flowers at basal nodes, the staminate flowers densely aggregated in a strobiliform mass at the end of the fleshy enlarged rachis, separated from the pistillate portion by an elongated internode (pseudopeduncle); bracts bi- glandular. Staminate flowers mostly 3 per bract; pedicel very short; calyx + 3-lobed, distinctly gam- ophyllous, the lobes imbricate, + covering the sta- mens in bud; petals and disk absent; stamens usu- ally 2; filaments completely connate into a slender tube that is exserted from the calyx at anthesis; anthers bluntly apiculate, dehiscing extrorsely and longitudinally; pollen grains subglobose, tectate- perforate, 3-colporate, colpi marginate; pistillode absent. Pistillate flowers solitary to each bract, distinctly pedicellate; calyx 3-parted, segments im- bricate, eglandular within; petals and disk absent; carpels 3, each with a single ovule; ovary unap- pendaged; styles connate into a column, the tips unbranched, spreading. Fruits capsular; columella not persistent; seeds solitary in each locule, the testa distinctly foveolate; caruncle large and partly occluding top of seed; endosperm copious. A well-marked genus of three or four species, one or two in tropical America and two in tropical Africa, easily distinguished from the genera in the Gymnanthes—Sebastiania complex by its char- acteristic headlike staminate inflorescence, elon- gated staminal column, and hypertrophied carun- cles on the seeds. reported from North America for the first time. Maprounea is here RECENT LITERATURE ALLEM, A. C. 1976. Uma especie unica de Maprounea : Bo» na America do Sul. Acta Amazonica 1132 Annals of the Missouri Botanical Garden 37.1. Maprounea guianensis Aublet, Hist. Pl. Guiane 2: 895, tab. 342. 1775. TYPE: French Guiana, Aublet (not seen). Trees toca. 12 m high, glabrous; twigs subterete, smooth, brownish. Leaves with petioles ca. 0.5- 1.5 mm long, slender; stipules triangular to lan- ceolate, scarious, ca. 0.5-1 mm long, persistent; blades thinly chartaceous, ovate to elliptic, rather abruptly short-acuminate at tip, cuneate at base (and minutely auriculate at junction with petiole), ca. (2-)3 l or 2 elliptic laminar glands on the underside near midrib (occasionally with a few small circular glands towards the tip, or sometimes entirely eglandular); -7 cm long, 1.5-4 cm broad, usually with midrib plane above, distinctly raised beneath; major lateral veins ca. 10-15 on a side, straightish; vein- lets prominulous on both sides, forming areoles + parallel to lateral veins; margins plane or recurved, entire. /nflorescences terminal mostly on short lat- eral branches, 1-2 cm long; pistillate flowers sol- itary at 1-4 basal nodes; staminate flowers in heads ca. 3-9 mm long; staminate bracts rather fleshy, the tip acute, less than 0.5 mm long, the cyathiform glands subsessile, ca. 0.3-0. Sta- minate flowers articulate above very short pedi- cels; calyx gamophyllous, basally contracted into a stipe ca. 0.3-0.8 mm long; staminal column slender, mostly 0.7-1 mm long; anthers 0.3-0.4 mm long. Pistillate flowers on ascending or re- mm across. curved pedicels (1—)2—5 mm long at anthesis, these becoming 6-13 mm long in fruit; calyx lobes 3, ovate, pointed, ca. 0.7-1 mm long; ovary smooth; styles 3, 2-3 mm long, united nearly or quite halfway into a stout column. Capsules + oblate, not lobed, 6-6.5 mm diam.; compressed, grayish brown, shiny, distinctly beaked, seeds ovoid, somewhat deeply and coarsely foveolate on both sides, 2.9- 3.6 mm long, -3.5 mm broad (including caruncle); caruncle large, covering nearly half the face of the seed ventrally, appearing 2-armed dor- sally. Rainforests at low elevations, Panama and Trin- idad south to Peru and Brazil (localities in South America summarized by Jablonski, 1967: 180). It is curious that this distinctive plant has been collected only a single time in a well-known area in the center of the Canal Zone. Presumably it will eventually turn up in various lowland areas toward the Colombian border. ecimen examined. PANAMA. CANAL ZONE: 1 mi of summit on road to FAA radar tower, Tyson et al. 2761 (MO, US) 38. Stillingia 12. 2: p Seid ex L., Syst. Nat. ed. 637. Stil- ; Mant. Pl. 19. 1767. TYPE: PR sy nn a L. 39. Sapium Sapium P. Browne, Civ. Nat. Hist. Jamaica 338. 1756. TYPE: Sapium jamaicense Sw. In the original treatment, the account of Sapium was avowedly tentative, pending the appearance of Jablonski's study of the Caribbean and Central American species. Jablonski's work, as well as a study of more recently collected specimens, ne- cessitates an entirely new treatment of the Pana- manian species. Of the four species recognized in the original treatment, only 5. eglandulosum re- mains unchanged. We have followed Jablonski in referring the species called S. aucuparium by Burc to 5. jamaicense and in uniting 5. caudatum and S. biglandulosum under the name S. aucuparium. Jablonski's action in restoring the name 5. jamai- cense seems straightforward and is adopted here. The nomenclatural subtleties connected with 5. aucuparium are still unresolved, however, and Ja- blonski's choice is followed as a tentative conclusion ° = Three additional species of Sapium are now known from Panama, bringing the total to six. Recent sterile collections of a seventh, possibly undescribed, species have been made on Barro Colorado Island. RECENT LITERATURE a L. 1943. Peu in American Euphorbi- ae. J. Arnold Arbor. 24: 165-189. Hurt, M J. Four new species of Sapium (Eu- phorbiaceae) from Central and South America. Phy- tologia 63: 441-448. oe a 1968. Notes on neotropical Euphorbi- ace oe of Caribbean Sapium. Phytologia 16: "303 434 KEY TO THE SPECIES OF S4PIUM IN PANAMA la. Petioles eglandular 2. S. eglandulosum S. rigidifolium mature capsule; capsule €— secondary foliar veins arcuate-asc pm hy stachys 2b. Leaf tips inflexed or cucu ullat 4a. Spikes clustered at tips of branchlets; Volume 75, Number 3 1988 Webster & Huft Panamanian Euphorbiaceae 1133 pisi foliar veins straight, pcd uate near margin ..... S. jamaicense Spikes solitary at tips of hatte: secondary pied veins strongly ar- cuate-ascen da. Leaves eius or oblong-elliptic, less than 2.5 times as lon broad; the cts) veins usu- ally 10-15 per s as v S. oligoneurum . Leaves usually elliptic- -lanceolate, more than 3 times as long as broad, usually longer than 10 cm, the secondary veins usually mor than 20 per side .. 1. S. aucuparium en e 39.1. Sapium aucuparium Jacq., Select. Stirp. Amer. Hist. 249. 1763. LECTOTYPE: Jacquin, Select. Stirp. Amer. Hist., pl. 158 (chosen here). ?Sapium i riii (L.) Muell. Arg. (Linnaea 32: 116. 1863) sensu auctt. Hippomane biglandulosa L., = Pl. ed. 2, TA 1762, correction of H. glandulosa L., Sp. Pl. ed. 1, 1191. 1753. Excoe- caria biglandulosa E ) Muell. Arg. in DC., Prodr. 15(2): 1204. 1866. Sapium pde amu Kunth, Nov. Gen. Sp. Pl. 1817 E: Colombia: Humboldt & Bonpland (P. not see Sapium moritzianum Klotzsch in Seem., Bot. Voy. Her- ald 100. 1853. Sapium biglandulosum (L.) Muell. Arg. var. moritzianum (Klotzsch) Muell. Arg., Lin- naea 32: 119. 1863. Excoecaria biglandulosa (L.) Muell. Arg. var. moritziana (Klotzsch) Muell. Arg in DC., Prodr. 15(2): 1206. 1866. Sapium aucu parium Jacq. subsp. moritzianum (Klotzsch) Pittier, Contr. U.S. Natl. Herb. 20: 127. 1918. TYPE: pre- sumably Colombia: Moritz 236. Panama: Seemann 1243 (syntypes, not seen); no specimens cited in protologue. d a caudatum Pittier, Contr. Yw " Natl. Herb. 20: 127. 1918. LecrorYPE: Panama. Canal Zone: hill near Gamboa, 25 June 1911. Pittier 3713 (US) (chosen by Jablonski, 1968). E t giganteum Pittier, Contr. U.S. Natl. Herb. 20: 128 8. TYPE: Panama. Colón: Fato, sea level, 10 Aug . 1911, Pittier 4141 (holotype, US puesta haematantha Standley, Ann. Missouri Bot. : 314. 1940. TYPE: j e ma. Coclé: N rim of EI Valle, 9 July 1939, Allen 1915 (holotype, F; isotype, MO, F neg. Monoecious tree to 10 m. Leaves membranous to chartaceous, rarely coriaceous; petiole 1-5 cm long, the 2 apical glands prominent, cylindrical, ca. 1 mm long; stipules ovate-deltate, 1-1.5 mm long, 1.5-2 mm broad, persistent; blades elliptic- lanceolate to oblong-obovate, 5-40 cm long, 1.5- 8 cm broad, 3-5 times as long as broad, the base rounded to acute, the margins entire, obscurely toothed or coarsely serrate, the apex short-acu- minate or cuspidate, strongly cucullate. Spikes sol- itary, terminal, to 22 cm long, bisexual or stami- nate. Staminate flowers in groups of 7-10, the subtending bract flabellate, ca. 0.5 mm long, bi- glandular, the glands circular to oblong, 0.5-3 mm long, 0.5-1 mm broad; calyx ca. 1 mm long, cupular, 2-lipped; stamens 2. Pistillate flowers to 10, solitary at basal nodes, the bracts and glands as those of the staminate flowers; calyx cupular, 2-lobed; ovary orbicular, the style simple, the style- branches strongly reflexed, the tips expanded. Cap- sules ovoid, to 1 cm long, smooth; seeds ovoid, flattened laterally, the surface warty. There has been considerable controversy con- cerning the proper name of this species. The name S. aucuparium Jacq. had long been applied to it, but Croizat (1943: 175), whom Burch followed in the original treatment, referred that name to the species usually known as S. jamaicense Sw. and resurrected S. biglandulosum for this species. Ja- blonski correctly restored Jacquin's name to the present species, but rejected the name 5. biglan- dulosum as a nomen confusum, leaving S. aucu- parium as the earliest available name. It would seem, however, that the application of 5. biglan- dulosum can be fixed by a proper lectotypification. More study is needed before that can reasonably e done, and, in any case, it is not clear to the present author whether any of the elements orig- inally cited under S. biglandulosum actually be- long to this species. The best course, then, seems to be the tentative acceptance of 5. aucuparium Jacq. for this species until the matter can be sat- isfactorily resolve 'e are following Jablonski (1968) in uniting Sapium caudatum and S. biglandulosum (sensu Croizat and Burch) under S. aucuparium. Ex- amination of a wide range of collections from Pan- ama and from northern South America shows that the leaf characters used by Burch to distinguish these species are continuously variable and do not correlate with other characters. This is a commonly collected, highly variable species of lowland tropical 39.2. Sapium eglandulosum Ule, Bot. Jahrb. Syst. 35: 673. 1905. TYPE: Brazil. Amazonas: Bom Fin on the Rio Juruá, Nov. 1900, Ule 5356 (holotype, B, F neg. 5522) The following collection records a considerable westward range extension for this species, which was previously known in North America only from eastern Darién. 1134 Annals of the Missouri Botanical Garden Additional specimen examined. PANAMA. PANAMA: Cerro Jefe region, 2 km N of turnoff to radio tower on roadside of Alto de Pacora, 2,600 ft., Hammel 4868 (F, MO). 39.3. Sapium jamaicense Sw., Adnot. Bot. 62. 1829. TYPE: Jamaica: Swartz s.n. (BM, not seen). Sapium aucuparium sensu Croizat, J. Arnold Arbor. 24: 174. 1943, sensu Burch, in Missouri Bot. Gard. 54: 325. 1967, non Jacq. 1763. Sapium pleiostachys Schumann & Pittier, Contr. U.S. Natl. Herb. 12: 164. 1908. TYPE: Costa Rica. Pun- tarenas: Golfito de Osa, near seashore, Mar. 1896, Pittier s.n., Inst. Fis. Geog. Costa Rica no. 9906 OH cd US-578902; isotypes, F, F neg. 62366, GH, F ne 0). Sapium iiec Pittier, Contr. U.S. Natl. Herb. 12: 164. 8. TYPE: Costa Rica. Cartago: Hacienda Valverde at Orosi, 1, m Mar. I s.n., Inst. Fis. Geog. Cos a Joi: pa. 16366 (ho- lotype, US-578045; Sen F, F neg. 62364). Rainforests, West Indies, Mexico to Panama, and recently discovered in Colombia. The correct name of this species has also been a matter of some confusion. Sapium jamaicense had been used universally until Croizat (1943: 174) asserted that 5. aucuparium sensu Jacq., Enum. Pl. Carib. 31, 1760, properly refers to this plant, in contradistinction to Jacquin’s intention stated in his Selectarum Stirpium Americanum Historia (1763), which had been followed by most other authors, who had applied the name S. aucuparium as in this paper. Jablonski (1968), however, cor- rectly pointed out that Jacquin’s publication of S. aucuparium in 1760 is invalid, since a description is lacking, and that the traditional application of S. aucuparium as published by Jacquin in 1763 is correct. Jablonski (1968) recognized Sapium pleiosta- chys as distinct from S. jamaicense, referring all collections from Panama and Costa Rica, as well from Guatemala and Chiapas, to the former species, and restricting the latter to the West Indies and northern Central America. The only difference he adduced was the presence of petiolar glands on S. pleiostachys and their absence on S. jamai- cense. Even granting this difference, recognition of S. pleiostachys would be tenuous indeed, given the inadvisability of maintaining a species on the basis of a single morphological feature and the assertion that both variants occur in northern Cen- tral America (indeed, specimens collected by Ma- tuda at Escuintla, Chiapas, are cited under each species by Jablonski). A thorough examination of collections from throughout the range, however, shows that even these meager grounds are unten- able. Plants with petiolar glands occur commonly in the West Indies (cf. Ekman 5512 from His- paniola, Harris 9156 from Jamaica, and Pringle 104 from Cuba, all F), and thus there is no mor- phological discontinuity. Sapium jamaicense is simply a variable species in regard to the presence or absence of petiolar glands, and there is no jus- tification for recognition of S. pleiostachys. A few new provincial records are recorded be- low. Additional specimens examined. PANAMA. DARIEN: Rio Pirre, Bristan 1475 (M ca. 2 mi. E of El Llano, 2 oste (F, NY). VERAGUAS: Isla de Coiba i dale Colony), mith 1566 (MO). COLOMBIA. ANTIOQUIA Canon del Rio Claro, 330-425 m, Cogollo 965, 1232 MO). — 39.4. Sapium oligoneurum Schumann & Pit- tier, Contr. U.S. Natl. Herb. 12: 168. 1908. Sapium biglandulosum (L.) Muell. Arg. var. oligoneurum (Schumann & Pittier) Monach., Bull. Torrey Bot. Club 67: 772. 1940. TYPE: Costa Rica: near San Rafael on road from Cartago to Cot, 1,500 m, July 1899, Pittier s.n., Inst. Fis. Geog. Costa Rica no. 13403 (holotype, US-578903). Sapium Mm Pittier, Contr. U.S. Natl. Herb. 12: 69. 1908. Sapium biglandulosum (L.) Muell. Arg. var. Pod ae Monach., Bull. Torrey Club 67: . 19 osta Rica. La ay 1,500 m, E Nie 1898, Fondas s.n., Ins Fis Geog. Costa Rica no. 12428 (holotype, LS 77588: isotype Sapium Pbi Cr oizat, Amer. Midl. Nat. 29: 477. 1943 : Belize. Toledo District: Forest Home, Punta Corda, Sc hipp pee "EUM A, F neg. 62361; isotypes, F, F neg. 62365, MO, F neg. 62363). Monoecious tree to 20 m. Leaves membranous or chartaceous; petiole 1-3(-6) cm long, the two glands near the apex opposite or subopposite, cy- lindrical, 1-2 mm long; stipules ovate-deltate, oblique, 2-3 mm long, 1.5-2 mm broad, ap- pressed, persistent; blade oblong or elliptic-oblong, 4—10(-18) cm long, 2.5-4.5(-8) cm broad, 1.2- 2.4 times as long as broad; midvein prominent, the secondary veins 10-15(-20) per side, somewhat inconspicuous; base rounded to obtuse; margin ap- pearing entire, remotely denticulate with minute glandular teeth; apex acute or more often abruptly short-cuspidate, conspicuously and tightly cucul- late. Spikes solitary, terminal, to 22 cm long, bi- sexual or staminate. Staminate flowers in groups of 5-7, the subtending bract short, broad, 1- 1.2 Volume 75, Number 3 1988 Webster & Huft Panamanian Euphorbiaceae 1135 mm long, 1.8-2.1 mm broad, rounded, hyaline, slightly erose, biglandular, the glands suborbicular to oblong, 1.8-3 mm long, 1.8-2.5 mm broad, flattened; calyx cupular, 1.7-2 mm long, 2-lipped; stamens 2, the filaments free. Pistillate flowers 10-22, borne singly at basal nodes of bisexual spikes; bracts and calyces as in the staminate flow- ers; ovary globose; styles simple. Capsules sub- globose to slightly obovoid, subsessile, 5-9 mm long, 5-12 mm diam., smooth; seeds subglobose, slightly compressed, yellowish, the surface warty. Montane and cloud forests, forest edges, and clearings, Chiapas, Mexico, to Panama. Specimens examined. PANAMA. BOCAS DEL TORO: re- ion of Cerro Colorado, 3.3 mi. above Camp Chami, 8*35'N, 81?45'W, ca. 1,350 m, McPherson 9587 (F). CHIRIQUÍ: E of Boquete on Cerro Azul near Quebrada Jaramillo, 1,500-1,620 m, n tru (MO, NY); along road between Gualaca and Fortuna Dam site, 10.1 m NW of Los Planes de Nornito, 8045; N, 82°17'W, 1, 250 m, Croat 50032 (MO); Boquete, 4,000 ft., ke ot 852 (MO, US; cited in the original treatment as S. cuparium sensu Burch); near Cerro Colorado, ca. 3. 5u mi. along road from Chami Camp, ca. 8°35'N, 81?45'W, ca. 1,350 m, McPherson 8997 (F); pastures Fei El Bo- quete, 1,000-1,300 m, Pittier 2880 (F, US); valley of the upper Rio Chiriquí Viejo, G. & P. Whit ite 95 (MO). These specimens have been distributed under the names S. caudatum, S. aucuparium, S. oli- goneurum, or S. sulciferum. 39.5. Sapium pachystachys Schumann & Pit- tier, Contr. U.S. Natl. Herb. 12: 168, tab. 16. 1908. TYPE: Costa Rica. San José: Dota Mountains, El Copey, 1,800 m, Feb. 1898, Tonduz s.n., Inst. Fis. Geog. Costa Rica no. 11875 (holotype, US-333961; isotype, F, F neg. 62367). Monoecious tree to 25 m; older twigs covered with crowded persistent stipules. Leaves membra- nous or chartaceous; petiole 2-5 cm long, the 2 glands near the apex subopposite, cylindrical, 1— 3 mm long; stipules deltate, 4-6 mm long, 2-3 mm broad, appressed, persistent; blade elliptic, el- liptic-obovate, or elliptic-lanceolate, 5-20 cm long, 2.5-7.5 cm broad, 1.6-2.4(-4.5) times as long as broad; base rounded or obtuse, rarely acute; mar- solitary at the apex of smooth lateral shoots, to 20 cm long, bisexual. Staminate flowers in groups of 7-10(-12), the subtending bract short, broad, to 2 mm long, hyaline, erose, biglandular, the glands oblong, 2.5-3 mm long, 1-1.5 mm broad, flat- tened, calyx 1-1.5 mm long, cupular, 2-lipped; stamens 2, the filaments free. Pistillate flowers 14-22, solitary at basal nodes; bracts as in the staminate flowers; calyx 1-1.5 mm long, cupular, 2-lipped; ovary globose; styles simple, fused for /,— Y, their length, the free portion strongly coiled. Capsules globose, subsessile, 7-10 mm long, smooth; seeds subglobose, flattened laterally, ca. 4 mm diam., the edges short-winged, the surface somewhat warty. Montane rainforest and cloud forest, 700-2,000 m, Costa Rica and Panama. The twigs densely covered with persistent stip- ules and the large, broad leaves with noncucullate apices are characteristic features of this species. A related species, S. allenii Huft, has recently been described from eastern Costa Rica (Huft, 1987) and may eventually be discovered at lower elevations in western Panama. It differs from S. pachystachys in having axillary spikes, smaller, stipitate capsules, and a small membranous calyx that does not persist on the mature capsules. Specimens examined. PANAMA. BOCAS DEL TORO: border with Chiriqui, Cerro Colorado mine area, from Chami Station to ca. 9 mi. along road, 8?35'N, 81°54’W, 1,100-1,700 m, Hammel & Trainer 15004 (F). CHIRIQUÍ: Boquete, Finca Collins, Blum & Dwyer 2558 (MO y SO de campamento Fortuna (8?45'N, 82°15'W), sitio de pre- sa, desde la finca Pitti hasta e filo del Cerro Fortuna, 0 00 m Cerro Colorado, along ro io deis 1 km beyond turnoff to DRE 1,390 m, 6 dos 37305 (MO); Cerro Punta, 0 m, a 328 (MO); above Los Llanos, ST N, 82°38! W, ca. 2,100 m dy abe eke Colorado, 50 pum N of S Félix ha Continental Divide, ,200-1,500 m, Mori & pl 7819 (MO, NY); be- tween Rio Ladrillo and Las Siquas Camp, southern slope of Cerro de la Horqueta, 1,200-1,700 m, Pittier 3165 (US); slopes of Volcán Barü near town of Cerro Punta, 6,300 ft., Stern & Chambers 97 (MO, US). cocLé: N of El Copé on road past sawmill, 2,400 ft., Antonio 3264 (F, MO). PANAMÁ: 5-10 km NE of Altos de Pacora, on trail at end of road, 700-800 m, Mori & Kallunki 6065 (MO, NY, 2 sheets). 39.6. Sapium rigidifolium Huft, Phytologia 63: 444. 1987. TYPE: Costa Rica. Heredia: pastures above San Rafael, 30 km W of Vaca Blanca, 1,750 m, 8 Aug. 1971, Lent 2041 (holotype, F; isotypes, MO, NY, US), distrib- uted as Sapium thelocarpum Schumann & Pittier. Figure 5. Sapium rigidifolium, which is known only from high altitudes in Costa Rica and Chiriqui Province in Panama, belongs to the otherwise wholly South American subsection Emmenostylum (Hemsley) 1136 Annals of the Missouri Botanical Garden I 9. Sapium rigidifolium.— a. Bra doni staminate flowers at anthesis and pistillate flowers in young fruit. Based on Lent 2041. Illustration by Steve Wilsor Pax (Pflanzenr. IV. 147. V(Heft 52): 211. 1912), characterized by styles that are connate for most of their lengths and whose columns persist on the mature capsule, and by leaves with planar apices and nearly horizontal, prominent, closely spaced secondary veins. The South American represen- tatives are restricted to high altitudes in the north- ern Andes and include such species as 5. verum anch with mature capsules. —b. Det ail of portion of inflorescence, Hemsley, S. stylare Muell. Arg., mayense Croizat. and S. putu- Spe . PANAMA. CHIRIQUÍ: Guadalupe Avia, above Cerro Punta, 852 N, 82°33' W, 2,100 m, )); PMA); slopes sd Volcán Bart, near town of Cerro aoe 6,000 ft., Sterr & Chambers 85 (A, MO, US) Volume 75, Number 3 1988 Webster & Huft Panamanian Euphorbiaceae 1137 40. Hippomane Hippomane L., Sp. Pl. 1191. 1753. TYPE: Hip- pomane mancinella L. 41. Hura Hura L., Sp. Pl. 1008. 1753. TYPE: Hura cre- pitans 42. Euphorbia Euphorbia L., Sp. Pl. 450. 1753. LECTOTYPE: Euphorbia antiquorum L. (chosen by Mills- paugh, Publ. Field Columbian Mus., Bot. Ser. 2: 306. 1909). Poinsettia Graham, Edinburgh New Philos. J. 20: 412 uphorbia sect. Poinsettia (Graham) Baillon, Etud. 284. 1858. Euphorbia subg. Poinsettia (Gra- ham) House, New York State Mus. Bull. 254: 472. KEY TO THE SPECIES OF EUPHORBIA IN PANAMA la. Leaves more than 15 cm long, leathery; cyathia 4-6 mm long lb. Leaves less than 8 cm long, membranous; cyathia less than 3 mm long. 2a. Gland of cyathium 1; floral bracts usually with pale or colored spo 3a. x olucral gland cup-shaped; floral leaves green, white, or purple-spotted at base, never red; se k gulate, coarsely tuberculate . TYPE: Poinsettia pulcherrima (Willd. ex Klotzsch) Graham. Four species new to Panama are reported here. In addition, the Panamanian endemic, Euphorbia apocynoides, which was merely mentioned in pass- ing in the original treatment, is here treated in full. These changes make it necessary to provide a new key. Both of the present authors are agreed that Poinsettia, treated as a separate genus in the orig- inal treatment, is best regarded as a subgenus of Euphorbia; thus Poinsettia is suppressed in the generic key, and the two Panamanian species are included in the key to Euphorbia. LITERATURE DRESSLER, R. L. 1961. A synopsis of d (Eu- phorbiaceae). Ann. Missouri Bot. Gard. 48: 32 41. 3. E. elata ts; seeds coarsely tuberculate. eds . heterophylla 3b. icr gland bilabiate; floral leaves red, at least at base; seeds ovoid-cylindrical, finely and A. Be a ° y tuberculate cyathophora N c ° D d a, un o 5 a < E g. £ z EN ° mi rarely 2). ; stems articulate; trees or large shrubs. 4. E. leucocephala 5b. Floral leaves green (unknown in £. apocynoutes): 6a. Leaf blades ovate, 3-5 cm broad; petioles 2-6 cm 9. 6b. Leaf blades oblong or narrowly obovate, 1.5-2 cm ers petioles to 1 cm lon 6. E. cotinifolia 4b. ca aha herbs. cres and capsules pubesce 8b. Infl 9b. Cy aibi sla 2 Sabio llanta aves jaa Sa leaf blades mostly 1 cm long or shorter ar; leaf blades mostly longer than 2 cm r 5, plane E. apocynoides ENS 7. E. ocymoidea 8. E. xalapensis 9. E. oerstediana 7b. Involucres and cap es gla 10a. Cyathial Be petra ae te; glands with 2 conspicuous lateral horns .......... 10. E. peplus ages pre 10b. rs appe enda s terete, the s 1 = Sedi ee} an m dei esent; glands elliptic urface smooth; cya athia endages pa glands, ciliate- -pubescent above a ca. 2 mm in diameter; glands 4, dark; 11. dwyeri rows; cyathia y punctate with pits in regular longitudinal . | mm in cab glands 4 or 2, green; appendages obsolete (Panamanian paa] to prominent and white, several times size of gland, 2 glabrous E. graminea 42.3. Euphorbia elata Brandegee, Univ. Calif. Publ. Bot. 6: 55. 1914. TYPE: Mexico. Chia- pas: Finca Irlanda, Purpus 7026 (holotype, UC; isotypes, A, BM, F, GH, MO, F neg. 62362, UC, US). > a ony Standley, J. Wash. Acad. Sci YPE: Costa Rica. Guanacaste: Las id near Tilarán, 600-700 m, Standley & Valerio 45338 (holotype, US; EUR US). Shrub or small tree to 3.5 m high, glabrous; branches few or none. Leaves alternate, clustered at apex of stem; petioles 2-4(-6) cm long, ca. 2 mm thick; stipules fleshy, light-colored, deltate to rounded, 3-4 mm long, 3-4 mm broad; blades 1138 Annals of the Missouri Botanical Garden coriaceous, glabrous, oblanceolate or narrowly el- liptic, the apex rounded to bluntly short-acuminate at tip, acute to cuneate at base, 15-35 cm long, 4-10(-12) cm broad, 3-5(-7) times as long as broad; midrib prominent below, to 2 mm thick; lateral veins 17-30 pairs on a side, obscure, nearly at right angles to midrib; margin entire. /nflores- cences single, terminal or rarely axillary, long- pedunculate, cymose; peduncle 10-30 cm long, glabrous, minutely brown-puberulent, or covered with a waxy brown reticulum; cyme up to 5 times divided, the cyathia ultimately in 2-4 compact groups; bracts opposite, scalelike, attached to the stem along a broad base, deltate, ca. 4 mm long, 2.5-3 mm broad; margin entire, sometimes ciliate, hyaline; apex blunt, somewhat cucullate. /nvolu- cres campanulate, glabrous, minutely brown-pu- berulent or covered with a waxy brown reticulum, green, drying to brown, 4-6 mm high, (3-)4-5.5 mm diam.; pedicels 1.5-3 mm long, 1-1.5 mm thick; lobes 5, 1.5-2 mm long, ca. 2 mm broad, erose to fimbriate, rarely entire, erect, concolorous or dark red; glands 5, round, placed vertically on rim of involucre, looking out- wards, but positions of lobes making glands appear below rim; gland sometimes erect and thick-stalked so that surface is flat with respect to orifice, the margin then crisped; appendages none. Capsules exserted from the cyathium no more than 1 mm 1.7-2 mm diam., (the gynophore erect), green, glabrous, 8-9 mm high, 10-11 mm diam.; united at base for ca. /, of their length, strongly recurved, bifurcate; seeds subglobose, truncate at apex, 4.7-5 mm diam., ca. 4.5 mm long, ecarun- culate, dark brown, with lighter, low, broken, i- regular longitudinal ridges. styles ca. 1.5 mm long, Euphorbia elata, which occurs in moist ever- green forests of low to middle elevations, is defi- nitely known to range from southern Veracruz, Mexico, to Colombia, and may occur as far south as Bolivia. It is the most widespread species of sect. Adenorima (Raf.) Webster, a group of several highly divergent species of trees and shrubs from the West Indies, western and southern Mexico, and the northern Andes. The species closest to É. elata are all highly restricted in range and are poorly understood. Euphorbia sinclairiana Benth. (in Seemann, Bot. Voy. Sulphur 163. 1844), known only from the island of Gorgona off the Pacific coast of Colombia, is characterized by a deeply bifurcate and much-branched inflorescence with conspicuous foliaceous bracts, but is otherwise sim- ilar to E. elata. The Peruvian E. tessmannii Mansf. (Ber. Deutsch Bot. Ges. 46: 674. 1929; Notizbl. Bot. Gart. Berlin-Dahlem 11: 137. 1931) is known only from the type collection, and no original ma- terial is definitely known to be extant. Until either new or original material is available, the status of that species cannot be determined. Euphorbia ca- pansa Ducke (Arq. Inst. Pesq. Agron. 1: 21. 1938), described from western Amazonian Brazil, appears to be synonymous with E. elata. Another collection from Bolivia (Bang 619, MO, NY, US) was given an unpublished name but will probably also prove to be E. elata. Specimens examined. PANAMA. BOCAS DEL TORO: road idge Rd. to Ri Pedros, 600 m, Antonio 3750 (F, MO); Distr. Portobelo, stream off N slope of Rio Gatun, 2,200 ft., Antonio 3804 (MO) Santa Rita, E of mountainous zone, Correa & ecd 973 (PMA); Santa Rita Ridge, E of Colón, Dress- ler 3338 (PMA), 3348 (F, PMA); Santa Rita lumber road, ca. 15 km E of (MO, PMA); Santa Rita Ridge, 4 hour walk from end of road, Hammel 6318 i. up i Guanche, 10-20 m, Kennedy € Foster 2170 (MO); Santa Rita Ridge Rd., ca. 6 Hwy., Mori & Kallunki 2151 (MO, NY); Ridge, on fork in road on main Santa Rita Ridge Road. Mori & Kallunki 3053 (MO, NY). DARIÉN: Cerro Pirre, valley between Pirre and next most southerly peak, sloping hillside, Folsom 4385 (MO). PANAMA: on road near slopes of Cerro Jefe, 2,400 ft., Antonio et al. 3426 (F, MO); just before La Eneida along new trail beside Lopez House, Correa & Dressler 724 (DUKE, 2 sheets), Correa et al. 816 (MO); El Llano-Carti road, 14-18 km from road to Chepo, 400 m, Correa et al. 1867 (PMA); El Llano- Cartí road, 12 mi. above Pan-Am Hwy., Liesner 1244 (MO); El Llano-Cartí road, 11 km from Pan-Am Hwy., Agricultura, Alto Piedra near Santa Fe, 0.3 mi. b fork in road near school, toward Atlantic slope along trail to top of Cerro Tute, 3» 400-3, ~ ft., Antonio 3498 (MO); 6.4 t oad past e dna school, 2 the cordillera, Folsom 2970 (F, MO). 42.4. Euphorbia leucocephala Lotsy, Bot. Gaz. (Crawfordsville) 20: 350, pl. 24. 1895. LECTOTYPE: Guatemala. Huehuetenango: Cuil- co, Dec. 1891, Shannon 305 (presumably US, not seen; chosen by Standley & Steyer- mark, Fieldiana, Bot. 24(6): 107. 1949). Shrub to 3 m high; branches terete, glabrous, swollen at the nodes. Leaves verticillate; petioles (1-)2-6 cm long, slender, glabrous; stipules glan- duliform, 0.3-0.5 mm long; blades glabrous, ellip- tic to linear-elliptic, mucronate and rounded or bluntly acute at tip, acute at base, (2-)3-7 cm long, (0.7-)1.5-2.7 cm broad, 2-3(-4) times as long as broad; margin entire. /nflorescences ter- minal, cymose; bracts white, narrowly spatulate; petioles 3-7 mm long, thinly pilose; blades 5-10 Volume 75, Number 3 1988 Webster & Huft Panamanian Euphorbiaceae 1139 mm long, 1.2-2 mm broad; base narrowly acute; margin entire; apex rounded, mucronate. Cyathia on peduncles 1-2.5 mm long; involucre campan- ulate, 1-1.8 mm high, 1.1-1.7 mm diam., densely tomentose just below the glands, otherwise sparsely pilose; glands 5, green, 0.8-1 mm long parallel to the rim of the cyathium, 0.4-0.5 mm broad, the center of the inner margin strongly inflexed forming a deep convex trough; appendages white, linear- lanceolate, exceeding glands by 2-3(-3.5) mm, 0 road, the apex narrowly rounded. Gynophore erect, exserted from cyathium 1-1. high, shallowly 3-lobed, the cocci mm. Capsules glabrous, smooth, 5-6 mm 4.5-5.5 mm diam., distinctly 3-angled; styles 0.6-0.7 mm long, united at base, deeply trifid, the style branches strongly recurved; seeds (immature) ca. 3.5 long, trigonous, ca. 1.8 mm wide, carunculate. The reports of this species from central Panama represent a considerable range extension; the pre- viously known range is from western and southern Mexico to Honduras. Euphorbia leucocephala is a commonly cultivated ornamental in Central America, and the Panamanian collections may be from cultivated trees, but the label data are not clear on this point. Capsules are unknown in Panamanian collec- tions and are scarce in the numerous collections from northern Central America. Our description of the capsules is taken from a specimen collected in Depto. Huehuetenango, Guatemala (Molina 21389, F Specimens examined. PANAMA. COCLÉ: El Valle Antón, 1,000-2,000 ft., Lewis et Am 2570 (MO); E Valle, Ramos 19 (M A). PANAMÁ: Panama Viejo, Girón 1 (MO); near Corra Azul, Laden 49 (MO, PMA). 42.6. Euphorbia apocynoides Klotzsch in Seemann, Bot. Voy. Herald 99. 1853. TYPE: Panama. Darién: Punta Garachiné, Seemann 1096 (holotype, BM; isotype, K; photo of isotype, K, MO). Shrub to 2 m high; stems glabrous, terete, swol- len at nodes; internodes 2-4 cm long. Leaves ter- nate; petioles slender, 7-10 mm long, very sparsely pilose; blades membranous or chartaceous, dark green above, lighter or even somewhat glaucous below, oblong to narrowly obovate, rounded at tip, acute at base, 4.5-6 cm long, 1.5-2 cm broad, 3-3.5 times as long as broad, glabrous or with very few hairs below; margin entire. /nflorescences unknown. Cyathia turbinate, ca. 2.5 mm high, ca. 2.5 mm diam. below the appendages, sparingly to evenly appressed-pubescent, the hairs short, straight, nonoverlapping; peduncles 3-4 mm long, subglabrous; appendages 5, narrowly flabellate, white, pubescent as the involucre below toward the base, ciliate on the margins near the juncture with the involucre, otherwise glabrous, entire in the lower half, deeply and coarsely crenellate along the distal margin. Flowers and fruits not seen. This poorly known species is apparently endemic to Panama and is still known only from the frag- mentary type collection and an equally fragmen- tary recent collection from the type locality. The type is sufficiently complete, however, to enable the species to be placed with reasonable certainty in section Alectoroctonum (Schldl.) Boissier, a group characterized by verticillate branching and swollen nodes, and to say that it is unlike any other species in the section. Contrary to the statement in the original treatment (Webster & Burch, 1968: 335), the cyathium is quite unlike that of sect. Dichilium Boissier, which is characterized by a reduced num- ber (usually 2) of bilabiate glands and small, erect, or often obsolete, Appendages. The species of sect. Dichilium are also characterized by alternate leaves and stems that are pinched just above the nodes. A plant vegetatively very similar to Euphorbia apocynoides, collected near Puerto Colombia on the Caribbean coast of Colombia (Elias 1197, F), has been identified as E. nudiflora Jacq., a West Indian species. The cyathium, however, differs from that of E. apocynoides in that it is campanulate, the pubescence is crisped with overlapping hairs, and the appendages are completely glabrous, ob- ovate, and with smaller and more numerous cren- ellations on the distal margin. In all of these characters it matches E. nudiflora, which differs vegetatively in its more highly branched habit and shorter, broader leaves. An elucidation of the re- lationship of the Colombian plant to the Panama- nian one, and of both to E. nudiflora, must await fuller collections from Panama and Colombia. We are indebted to Mr. A. Radcliffe-Smith of Kew for providing a description and a sketch of the cyathium from the isotype of E. apocynoides at K, from which our description has been taken. Additional collection examined. PANAMA. DARIEN: thorn forest near Punta Garachiné, Duke 10485 (MO, 2 sheets). 42.7. Euphorbia ocymoidea L., Sp. Pl. 453. 1753. TYPE: Mexico. Campeche: Houston s.n. (BM, not seen). E. astroites Fisch. & Mey., Index Sem. Hort. Petrop. 1140 Annals of the Missouri Botanical Garden Mexico: Tampacoala, Kar- 10: 44. 1845. TYPE: winsky (LE, not seen). A thorough examination of collections from a the range of this species (western Mex- ico to Panama) makes it clear that Euphorbia astroites cannot be separated from E. ocymoidea. e only consistent character separating the two is the glandular-pilose stems of the former, as op- posed to the glabrous or short-pilose, eglandular stems of the latter. The two forms occupy roughly the same geographical range (the glandular form is not yet known from western Mexico, and the eglandular form has not been collected south of Nicaragua) and the same habitats, and they exhibit similar variation patterns, particularly in the shape of the leaves, which range from broadly ovate or deltate to somewhat reniform. As suggested by McVaugh (1961 collections from western Mexico, particularly FK. subreniformis S. Watson, undoubtedly belong here. The only known Panamanian collections are glandular-pilose, a fact that was omitted from the original treatment, and thus would have been placed 7), several names based on under F. astroites. The species has apparently not been collected in Panama since the appearance of the original treatment. 42.8. Euphorbia xalapensis Kunth, Nov. Gen. Sp. Pl. 2: 61. 1817. Poinsettia xalapensis (Kunth) Klotzsch & Garcke, Monatsber. Kó- nigl. Preuss. Akad. Wiss. Berlin 1859: 253. 1859. TYPE: Mexico. Veracruz: near Xalapa, Humboldt & Bonpland s.n. (P, not seen). SUP enalla Brandegee, Univ. Calif. Publ. Bot. 6: 14. TYPE: Mexico. Chiapas: Cerro del Bo- ae, Purpus 7035 (holotype, UC; isotypes, F, F neg. 60269, GH, MO, NY; distributed under an unpublished name) dle FEH TR Publ. Field Colum- bia dos a 313. 1929. TYPE: Honduras. Cor sasa aqa hag os Feb. 1928, Standley 56. 341 (holotype, F; isotypes, F, F neg. 60244, US). Perennial rhizomatous kerb to 50 cm high, the stems shaggy brown-pilose. Leaves opposite or ver- ticillate above, alternate below; petioles 1.5-2.5 cm long, brown curly-pubescent; stipules glandu- liform, brown or black, minute, 0.1-0.2 mm long; blades ovate, acute at tip, rounded to obtuse at base, dark green, 1.5-3.5 cm long, 1-2 cm broad, 1.5-1.8 times as long as broad, densely shaggy- pilose below, more sparsely so above, base rounded to obtuse; margins entire, ciliate, acute. /nflores- cences terminal, cymose, often appearing one-sided by the abortion of one branch at a node, shaggy- pubescent; bracts similar to the leaves but greatly reduced, often aborting. Cyathia on pedicels 1-3 mm long; involucre campanulate, 0.7-1.3 mm high, 0.5-1.3 mm diam., green, elliptic to reniform; appendages whitish or crisp-pubescent; glands 5, greenish, broadly ovate, exceeding the gland by 1-2 mm, 1.5-2 mm broad, the margin entire or crenate; gynophore glabrous, erect or somewhat recurved, exserted from the cyathium 1.5-2 mm. Capsules sparsely to densely pilose, 1.5-1.9 mm high, 1-1.5 styles 0.7-0.8 mm long, free to the base, deeply bifid, thinly pilose; seeds ca. 1.5 mm long, 0.9-1 mm diam., ovoid, grayish, coarsely pitted, tuberculate, ecarunculate. mm diam.; Euphorbia xalapensis is a common species of forest borders and thickets that ranges from west- ern Mexico to Honduras, and thus the new reports cited here from Panama and Costa Rica represent a considerable range extension. This species was erroneously treated in the Flora of Guatemala (Fieldiana, Bot. 24(part 6): 108. 1949) as E. oer- stediana (Klotzsch & Garcke) Boissier, a very dif- ferent species that has mostly glabrous stems, ar- ticulated nodes, two cyathial glands that are more or less bilabiate and that have inconspicuous or obsolete appendages, and densely white-pubescent capsules. The latter species, which belongs to sec- tion Dichilium Boissier, is very rare in Central America and is somewhat better known from the West Indies and northern South America. It is still unknown from Guatemala, and nearly all of the putative collections from there are referable to £. xalapensis. It has not been re-collected in Panama since the collection cited in the original treatment. Euphorbia xalapensis is also frequently con- fused with E. graminea, a highly variable and widespread species of the same section, Cyttaro- spermum Boissier, and the two are indeed very similar in aspect. Euphorbia xalapensis, however, can be distinguished by the distinctly perennial and often strongly rhizomatous habit; the shaggy brown pubescence of the stems, leaves, and inflorescence; the one-sided appearance of the inflorescence, which is due to the frequent abortion of one branch at a node; and the five glands that have ample, usually greenish appendages. Additional up e examined. Costa RICA. SAN JOSÉ: along Quebrada Tablazo and on forested d slope above creek, NE part of Altos Tablazo, 9°50'N, 84°03'W, 1,700 1,800 m, Grayum & Sc nate eee ). PANAMA. CHIRIQUÍ: Volcán Chiriquí above Boquete, roadside, D'4rcy 9805 ee 2 distributed as “Euphorbia graminea Jacq. (s. lat 42.10. Euphorbia peplus 1753. TYPE: L., Sp. Pl. 456. Europe (presumably in Hortus Volume 75, Number 3 1988 Webster & Huft Panamanian Euphorbiaceae 1141 Cliffortianus Herbarium, BM, not seen; 630.24 in LIN Annual glabrous herb 18-35 cm high. Leaves alternate below, opposite or ternate above, sessile or short-petiolate, numerous, early deciduous be- low; stipules obsolete; blades bright green, mem- branous, spatulate, rounded at tip, acute or cuneate at base, 10-14 mm long, 8-12 mm broad; margins entire; floral leaves slightly smaller, somewhat re- flexed, congested. Cyathia solitary in forks of up- per branches, on peduncles 0.6-1 mm long; in- volucre campanulate, light green, ca. 1 mm high, 0.5-0.7 mm diam.; glands 4, green, crescent- shaped, exappendiculate, ca. 0.5 mm long parallel to the rim of the involucre, with prolonged narrow horns 0.5-0.7 mm long; gynophore exserted, ca. 1.3 mm long, recurved. Capsules green, 1.8— mm high, 2-2.2 mm diam., broadest below the middle, shallowly 3-lobed, the cocci each with 2 narrow longitudinal ridges; styles ca. 0.1 mm long, bifurcate, the style branches bifurcate; seeds ovoid- oblong, weakly 6-angled, gray, ca. 1.5 mm long, ca. 0.8 mm diam., carunculate, deeply pitted, mi- nutely white-tuberculate. This is apparently the first report from southern Central America of this cosmopolitan weed of tem- perate Eurasian origin. Specimens examined. PANAMA. CHIRIQUÍ: trail from Paso Respingo to Bajo, Chorro Cerro Punta to Boquete, along stream near Guadalupe, Hammel et al. 7077 (MO); forest remnant Ding o Syo s, 1 mi. from road near Cerro Punta dairy, , D'Arcy et al. 13193 (F, MO); Cerro Punta, Tyson 215 (MO, PMA). 42.11. Euphorbia dwyeri Burch, Ann. Mis- souri Bot. Gard. 54: 182. 1967. TYPE: Pan- ama. iriqui: Cerro Horqueta, NW of Boquete, Dwyer i al. 434 (llalatipé; MO; isotypes, GH, K, A second collection of this species has been found, misidentified as E. caracasana Boissier. Additional specimen examined. PANAMA. CHIRIQUÍ: humid forest of Cuesta de Las Palmas, southern slope o Cerro de la Horqueta, 1,700-2,100 m, Pittier 3216 US). 43. Chamaesyce Chamaesyce Gray, Nat. Arr. Brit. Pl. 2: 260. 1821. TYPE: Chamaesyce maritima Gray = C. peplis (L.) Prokh. (Euphorbia peplis L.) (See Wheeler, 1943: 461, for a discussion of the type.) LITERATURE WHEELER, L. C. 1943. The genera of the living Eu- phorbieae. Amer. Midl. Naturalist 30: 456-503. Pedilanthus Necker ex Poit., Ann. Mus. Natl. ist. Nat. 19: 388. 1812. TYPE: Pedilanthus tithymaloides (L.) Poit. (Euphorbia tithy- maloides L.). GENERAL REFERENCES Croat, T. 1978. Flora of Barro n Island. Stan- ford Univ. ut Stanford, Californ DRESSLER, R. L. The genus Pedilanthus (Eu- purs cae Gray Herb. -188 . A synopsis of pie ‘(Bachar ae). Ann Missouri Bot. Gard. 48: 329-341. Hatt, T. 1971. Architecture and growth of tropical trees exemplified by the Euphorbiaceae. Biotropica Hans, A. S. 1973. Chromosomal conspectus of the uphorbiaceae. Taxon 22: 591-636. JABLONSKI, E. 1965-1967. Euphorbiaceae. In: B. Ma guire (editor), Botany of the Guayana Highland VI. Mem. N.Y. Bot. Gard. a 150-178. 1965. VII. op. cit. 17(1): 80-190. McVaucH, R. 61. ` Tanapa novae Novo-Ga- licianae. Brittonia 13: 145-205. MUELLER, J. Š uphorbiaceae. In: DeCandolle, Prodromus 15(2): 1- 1286 (Euphorbieae, pp. 3-188, by E. Boissier). Seymour, F.C. 1979. Acalypha, Croton, and Sapium in Nicar id Phytologia 43: 133-195. A 37- b Flora Lu ipee usi Publ. , Bot. Ser. 18: 1-15 Voss, E. G. (Chairman, " Editorial i SORS dba, LA kg code of botanical nomenclature. Reg- L Wekerdk. G. 1 1975. Conspectus of a new classifi- cation of the Euphorbiaceae. Taxon 24: 593-601. — € D. Burcu. 1968. Euphorbiaceae. In: Flora of Panama. Ann. Missouri Bot. Gard. 54: 211-350. & E. RUPERT. 1973. Phylogenetic significance of pollen nuclear number in the Euphorbiaceae. Evo- lution 27: 524-531 1142 Annals of the Missouri Botanical Garden INDEX TO LATIN NAMES Numbers in boldface type refer to descriptions or prin- cipal entries; numbers with ! refer to illustrations; names in boldface type refer to new taxa or combinations; names in italic refer to synonyms. Acalypha L. itl rs cuneata Poe r. "emsa ‘Benth Muell. Arg. 1104 as Bent 6 trae Hen DEAE Webster 1106 urens Sw. Actephila e] M Actinostemon Klotzsch 1129 concolor ae Muell. Arg. 1129 Adelia L. mea Adenophaedra (Muell. Arg.) Muell. Arg. mel oe grandifolia (Klotzsch) Muell. Arg. 1 megalophylla (Muell. Arg.) e ooh tesa” woodsoniana (Croizat) (Croizat 087, 1 latifolia Sw. 1100, 1101, 1102 megalophylla Muell. Arg. 1102 oblongifolia Standley 1103 i egi oye Muell. Arg. 1101 a f. psilorhachis Muell. Arg. 1102 umboensis ou 1101 Alchorneopsis Muell. Arg. 1098 floribunda (Benth. E Muell. Arg. 1098 Amanoa Aublet guianensis Aublet 1092 Antidesma L. a ee 1101 Aporusa Blume 1093 Argythamnia P. "uud 1098 candicans Sw. 10 Astrocasia Robinson & Millsp. 1087, 1091 phyllanthoides Robinson & Millsp. 1091 tremula (Griseb.) Webster 1091 Ateramnus Browne 1129 Bernardia P. Miller 1100 section Adenophaedra Pod Arg. 1099 i 1 ndley) Webster dii 1100 ?grand ifolia (Koltzsch) Kies Arg. 1099 yl a Le Muell. Arg. 1099 VAN NA St castaneifolia r ) Y Hil. 1098 098 angustifolium Standleyy 1099 9 orinocense Karsten 109 Chamaesyce Gray 1141 aritima Gray 1141 septentrionalis L. O. Williams 1115 Cleidion Blume 1087, 1103 amazonicum Pax 1104 castaneifolium Muell. Arg. denticulatum iari Webster 1099 javanicum Blume 11 membranaceum Pax x Hoffm. 1099, 1103, 1104 ?nicaraguensis Hemsle oblongifolium (Standley) va 1103, 1104 prealtum Croizat 11 tricoccum (Casar.) Belo WT anum Croizat 1099, 1103 ERAN 14 aconitifolius | (Miller) I. M. em 1114 subsp. aconitifolius 1 adenophilus (Pax & K. Hoffm.) Pax & K. Hoffm. 1114 hamosus Pohl 1114 urens (L.) Arthur 1114 subsp. adenophilus (Pax & K. Hoffm.) Breckon 11 i subsp. urens 1114 Croizatia Barn 1087, 1092 naiguatensis Steyerm. 1092, 1903 neotropica Steyerm. 1092, 1093 anamensis Webster 1092 Croton L. 11 argenteus L. 1124 aromaticus L. 1116 be eiu Muell. Arg. 1118, 1121 benthamianus eye Soi ) Lanjouw 1118 ipsi Muell. qa 121 ni a pyramidalis (J. D. Smith) Webster 1123 ene Pax castaneifolius L. 1098 draco Cham. & Schldl. 1120 subsp. draco 1120 subsp. panamensis en Webster 1120 Bid Vent. 1123, 1124 killipianus Croizat Th lanjouwensis Jabl. 1118, 1119 1118 benthamianus Muell. Arg. 1118 nuntians Croizat 1121 pachypodus Webster 1 r Klotzsch 1 ande (Klotzsch) Ml is 1121 is J. D. Smith 1 subsp. tacarcunensis Webster 1119 enum Lundell 1118, 1119 section Dioscoreifoliae Pax & K. Hoffm. 1109 canescens Kun su bsp. friedrichsthalii (Muell. Arg.) Webster & Volume 75, Number 3 1988 Webster & Huft Panamanian Euphorbiaceae 1143 Huft 1109, 1110 ubsp. canescens "di cissifolia Poeppig 11 subsp. panamensis Pax & K. Hoffm.) Webster 1110 dioscoreifolia Poeppig 1109 friedrichsthalii Muell. Arg. 1109 heteromorpha Pax & K. Hoffm. 1110 p de D K. Hoffm. 1110 scandens shankii (a. ud Huft 1107, 1109 websteri Armbruster 111 Drypetes "Vahl 1087, 1096 glauc 96 anda Webster 1096 Euphorbia L. ed 1137 section Adenorima (Raf.) Webster 1138 section i ate ( chldl. ) Acl 1139 section Poinsettia (Graham) Baillon 1137 subgenus Poinsettia (Graham) House 1094, 1137 amphimalaca Standley 1140 antiquorum L. 1137 apocynoides Klotzsch 1137, 1139 1139 astroites Fisch. & Mey. ocymoidea L. oerstediana (Klotzsch & Garcke) Boissier 1140 peplus L. 1140 peplis L. 1141 sinclairiana Benth. 1138 subreniformis S. Watson 1140 xalapensis Kunth 1140 Excoecaria biglandulosa (L.) Muell. Arg. 1 var. moritziana RIAM Pu Arg. 1133 gel concolor i 112 s L. 1137 Hyeronima Allemào 1087, 1094 alchorneoides Allemao 1094 guatemalensis J. D. Smith 1095 laxiflora (Tul.) Muell. Arg. 1094, 1095 — (Tul.) Muell. Arg. 1095 ar. benthamii (Tul.) Muell. Arg. 1095 mos ir ) Muell. Arg. 1095 Janipha Kun seculi Kunth 1113 Jatropha L. 1114 econ Miller 11 adenophila Pax & K. Hoff. 1114 eel L. 1114 manihot urens Julocroton C. Martius Maesobotrya Benth. 1093 Manihot Miller 1087, 1113 aesculifolia (Kunth) Pohl 1113 brachyloba Muell. Arg. 1114. triandra L morisiana (Casar.) Radlk. 11 quadriglandulosa Pax & K. Hof. 1115 trianae Baillon 1115 Pedilanthus Necker ex Poit. 1094, 1141 tithymaloides (L.) Poit. 1141 Pentabrachium Muell. Arg. 1092 Pera Mutis 1110 arborea Mutis 1110 is L. 1096 subgenus X ylophylla de 4 Pers. tzsch) van Arg. 1098 sa > _ 2 © r ~Se = e o ==] ntryi Web juglandifolius Willd. 1098 niruri 6 malus Griseb. 1091 Pluken netia L. 1105 aiden stifolia Standley 1105 a Muell. Arg. 1105 acu A Podocalyx Klotzsch 1093 Poinsettia Graham pulcherrima (Willa ex Klotzsch) Graham 1137 (or dad Dr Klotzsch & Garcke 1140 Polyandra racteosa Leal 1103 Richeria Vahl 1087, 1093 dressleri Webster 1094. grandis Vahl 1093, 1094 1144 Annals of th Missouri BE Garden grandis Vahl £ obovata Muell. Arg. 1093 loranthoides (Klotzsch) Muell. Arg. 1093 obovata (M 11 uell. Arg. ) Pax & K. Hoffm. 1093 04. SEPA 1129, 1132 subsection Emmenostylum (Hemsley) Pax 1135 allenii Huft 1135 anadenum cad 1134 np pis Jacq. 1132, 1133, 1134, 1135 moritzianum (Klotzsch) Pittier 1133 3 mo asi sensu Croizat biglandulosum (L.) Muell. Arg. ar. moritzianum (Klotzsch) Muell. Arg. 1133 var. oligoneurum (Schumann & Pittier) Monach. 1134 var. sulciferum ( ren PORE E caudatum Pittier cere 1133. sulciferum Pittier 1134, 1135 is dis ae & Pittier 1135 erum Hemsley 6 Sabastiania a ee 1127, 1129, 1131 section Adenogyne 112 section Elachocroton E Muell.) Pax 1128 section Microstachys (Adr. m ) Muell. Arg. 1128 section Sebastiania” 1128, actinostemoides (Muell. he ) Muell. Arg. 1130 p corniculata (Vahl) Muell. Arg. 1128 macrocarpa Muell. Arg. 1129 namensis Webster 1128 Tragia L. stipulacea (Muell. Arg.) “wa Arg. 1128 29 095 laxiflora (Tul. | Muell. Arg. 1095 oblonga Tul. 1095 Stillingia ex L. 1132 haematantha Standley 1133 13 sylvatica Te trorchidium Poeppig 1087, 1110 dley & Steyerm. 1111, 1112 is nis Huft 1112 euryphyllum Standley 1113 gorgonae Croizat subsp. robledoanum (Cuatr.) aa 1113 microphyllum Huft 1110, molinae L. O. Williams n bii Arn Cuatr. : ; 13 undatum Standley 1112 a Poeppig 1 i 10, 1112 Meu oit. siana Casar. 1114 87, 1106 section Bia (Klotzsch) Le Arg. 1108 section Fut tragia Muell. 1108 section Tragia 1108 section Zuckertia pese Muell. Arg. 1107 . 1106 bailloniana Muell. A co fallax Muell. Arg. 1108 fendleri Muell. Arg. 1107, 1108 grandifolia Klotzsch 1099 japurensis Muell. Arg. 1107, 1108 shankii A. Molina 1109 volubilis L. 1106, 1108 Wielandia Send 1091 Zuckertia cordata Baillon 1106 NOTES A NEW SPECIES OF HYMENOLOBIUM (LEGUMINOSAE— PAPILIONOIDEAE) FROM CENTRAL AMERICA Among the numerous trees of the Amazonian Hylea, the genus Hymenolobium is considered to be one of the main suppliers of excellent timber. It is closely related to the genus Andira (tribe Dalbergieae) but is distinguished by having a flat, reticulately veined fruit with one or two prominent submarginal nerves. Although delimitation of the species still presents various difficulties due to in- sufficient collections in herbaria, the 15-17 taxa previously recognized (Mattos, 1979; Lima, 1982) are restricted to South America from Venezuela and Surinam to southeastern Brazil. During examination of material in the herbarium of the Missouri Botanical Garden (MO), some new collections from Central America were encoun- tered. These represent a new species: Hymenolobium mesoamericanum Lima, sp. nov. TYPE: Costa Rica. Heredia: alt. 220 m, Rio Tirimbina, Istarú Farm, Tiribina, Sara- paqui, 16 July 1971 (fl), Roy W. Lent 2003 (holotype, MO: isotypes, NY, US). Figure 1. A omnibus specibus fructibus alatis adhuc cognitis flo- ribus majoribus (18-20 mm) et calycibus tenue-coriaceis pubescentis differt. Large tree to 40 m, dbh to 90 cm. Trunk with smooth, grayish bark; wood yellow. Terminal branches grayish, with numerous scars of caducous leaves, pubescent on the newest growth. Stipules caducous, ovate-lanceolate, 1 —1.4 x —0.9 cm, stipels filiform, 0.1-0.2 cm long. Leaves impari- pinnate, aggregated at the branch apices, 8-17- jugate (up to 25-jugate on regrowth branches or young plants); petiole pubescent, 2.5-4 cm long (to 6 cm long on regrowth branches or young plants); rachis pubescent, 12-28 cm long (to 56 cm long on regrowth branches or young plants); leaflets oblong, oblong-elliptic or ovate-oblong, the terminal one elliptic, 2.5-7 x 1.5-3 cm, the base rounded, subcordate or obtuse, the apex retuse or obtuse, mucronate, the lower face pubescent, the upper face sparsely pubescent to subglabrous; peti- olules 1.5-2.5 mm long. Panicles 10-15 x 15- 22 cm, the branches pubescent; bracts and brac- teoles caducous, the bracteoles ovate-lanceolate, pubescent, 1-1.5 x 0.5-0.7 mm. Flowers 18-20 mm long; pedicels 5-8 mm long; calyx campan- ulate, slightly coriaceous, pubescent, 5-toothed, 8— mm long; corolla rose, the petals papery; vex- illum 16-18 x 14-15 mm, wings 15-17 x 5-6 mm; keel 14-16 x 5-6 mm; stamens 10, mon- adelphous, 15-16 mm long; anthers 0.8-1 x 0.3- 0.4 mm; ovary long-stipitate, pilose along the mar- gins; ovules 2-3; style sparsely pilose, the stigma punctiform. Immature fruit flat, with wide lateral wings. Habitat and distribution. Emergent tree of gallery forests of low altitude (50-300 m), or left uncut in pasture land, in Costa Rica, Nicaragua, and Panama. Additional mater examined. COSTA RICA. HERE- : d ad, after entrance to Starke's ha- 976 (fl, Ivs), G. S. Hartshorn 1843 reno 23688 (MO, RB); 23 Mar. m vs) E. L. Little, Jr. 25272 (MO, US). ANAMA. CANAL ZONE: hills north of Frijoles, 19 Dec. 1923 (lvs), P. C. Standley 27584 (US). Hymenolobium mesoamericanum is distin- guished from other species in the genus principally by its large flowers (18-20 mm) with slightly co- riaceous, pubescent calyces. It shows affinities with H. heterocarpum, which possesses slightly larger flowers (22-23 mm) with coriaceous, tomentose calyces and suborbicular to oblong-reniform fruits with rudimentary lateral wings. The young fruit of H. mesoamericanum is provided with wide lateral wings, which suggest a samaroid fruit type as found in the majority of the species in the genus. How- ever, collections of completely mature fruit are necessary to confirm this character. Size and consistency of leaflets frequently vary in species of Hymenolobium (Lima, 1982). The ANN. Missouri Bot. GARD. 75: 1145-1147. 1988. 1146 Annals of the Missouri Botanical Garden Ñ T LN iro ES A HN 10 mm e ay n => Sie * Vra av) Bak 2 A KO? SM S u* «n [vd tbh Ó IM Sho LoT EI. Ww PENI ONES uM Coa SIA ay y CUR X] a I E a! Uy». I os ' Lt ei S do, , 14 EN esa q A MN RA SN " he my pa 2 ATUM, 7 YY D Code ? Y At y š kJ 21 MS ue Volume 75, Number 3 1988 Notes 1147 variation is related to leaf fall and/or to environ- mental factors, mainly intensity of solar radiation. The latter element determines the development of smaller, more rigid leaflets on the upper branches of the crown of adult individuals than on lower (regrowth) branches or on young individuals (Fig. ,8 The wood of this species supplies planks for construction and is commonly called “Golepac” (Costa Rica, Prov. Heredia) and “‘carolillo” or **no- gal" (Nicaragua, Dept. Zelaya). I am grateful to Dr. Gwilym P. Lewis, who reviewed and translated the manuscript from Por- tuguese. I also express appreciation to Dr. Warren D. Stevens for cooperation in lending the MO ma- terial and to Maria Helena Pinheiro for the illus- tration. LITERATURE CITED DuckE, A. 1936. Notes on the species of Hymenolo- 2 giant trees of Brazilian Amazonia. Trop. Woods == T. LIMA, p c DE. 1982. Considerações taxonómicas sobre ero Hymeno Y va Bentham (LEG. FAB.). Wetu on 12: Marros, 'N. F. o, género o Hymenolobium Ben- ham (Leguminosae) no Brasil Roessleria 3: 13-53. —Haroldo C. de Lima, Jardim Botánico do Rio de Janeiro, Rua Pacheco Leao 915, CEP 22460, Rio de Janeiro, Brazil. — FiGURE Al —a. Flow Hymenolobium mesoamericanum. (Lent 2003) .—f. individual yee 27584) —b. Bud.—c. Androecium.—d. P —e. Corolla stil. Leaflet DON the upper branches badd Taki icons 23688) .—g. i from young A NEW SPECIES OF VANTANEA (HUMIRIACEAE) FROM PANAMA Recently collected flowering material of the cen- tral Panamanian species of Vantanea hitherto con- sidered conspecific with the Colombian V. occi- dentalis Cuatr. (Gentry, 1975; Croat, 1978) has forced a reconsideration of the taxonomic status of the Panamanian entity. It is here recognized as a new species, V. depleta, differing most impor- tantly in stamen number and ovary pubescence from V. occidentalis, which it otherwise strongly resembles. Of significance to the current generic concept of Vantanea is the low number of stamens diag- nostic of the new species. Cuatrecasas, in his re- vision of the family (1961), characterized Van- tanea (sole member of the Vantaneoideae) as having 50-180 stamens, among other attributes, but V. depleta rejoices in no more than 15-18. Never- theless, since the new species has the bilocular anther thecae, biovulate ovary locules, and the drupe morphology typical of the genus, there can be little doubt that it is correctly placed in Van- tanea. Vantanea depleta McPherson, sp. nov. TYPE: Panama. Panama: Cerro Jefe, 650 m, 2 May 1987, McPherson & Stockwell 10892 (ho- lotype, PMA; isotypes, F, MO). Figure 1. Species floribus parvis (6 mm longis), petalis glabris, staminibus paucis (15-18), et ovario puberulo dignoscen- da. Tree 9-40 m; twigs often strongly angled, gla- brous, marked by elongate lenticels. Leaf blades elliptic or elliptic-ovate, (5.5-)8.5-16 cm long, (3-)4-8 cm wide, entire, coriaceous; base acute, often somewhat reflexed abaxially, extending down the petiole as a pair of tapering wings; apex obtuse, not or only obscurely acuminate; midrib prominent on upper surface, somewhat raised on the lower surface; secondary veins (6—)7—9(—1 1), only slight- ly raised; both surfaces dull, glabrous, 1 or 2 small sunken laminar glands in most cases associated with each of the secondaries. Petiole somewhat poorly delimited from blade, 3-9 mm long, swollen ANN. MISSOURI Bor. Garb. 75: 1148-1149. at the base, glabrous. Inflorescences terminal as well as sometimes also from the uppermost axils, broadly paniculate, the branches puberulent. Ped. icels 1.5-3 mm, less puberulent than the rachises. Sepals semicircular, 1 mm long, 1.5-2 mm wide, obtuse, puberulent, mostly bearing one centrally placed raised crateriform gland. Petals narrowly triangular-ovate, slightly imbricate in bud, 5 mm long, 2 mm wide, glabrous on both surfaces, white. Stamens 15-18, the filaments 3-5 mm long, fused basally for ca. 1 mm, glabrous, white; anthers ca. 0.7 mm long, the thecae bilocular, about as long as the distal prolongation of the connective. Disk ca. ] mm high, sharply dentate, glabrous. Ovary 2 mm long, densely puberulent, the hairs much shorter than the width of a filament; style 3 mm long, geniculate. Fruit 2.5-3.5 cm long, 1.5-1.7 cm diam., puberulent, rounded basally, acute dis- tally; endocarp smooth, with 5 broad ribs alter- nating with 5 oblong valves, 2.4-3.3 cm long, 1.5- cm in diam. Additional specimens examined. PANAMA. VER- AGUAS: Cerro Tute, 1,200 m, Lao & Gentry 530 (MO). CANAL AREA: Barro Colorado Inland, Garwood 440 (MO); un de Gentry 1931, 7406 (MO). PANAMÁ: Cerro Jefe, , McPherson 11008 (MO), 11296 (MO, PMA). SAN BLAS: between Rio Irgandi and Rio Carti Senni, de Nevers & Herrera 6597 (MO). Vantanea depleta is known from central Pan- amanian forests from near sea level to 1,200 m. I thank John K. Myers for the illustration. LITERATURE CITED Croat, T. 1978. Flora of Barro Colorado Island. Stan- ord Univ. Press, Stanford, California. CUATRECASAS, J. 19 A taxonomic iw of the .S. : : 25-214. . 1975. Family 87A. Bor In: Flora Panama. Ann. Missouri Bot. Gard. 62: 35-44. —Gordon McPherson, Missouri Botanical Gar- den, P.O. Box 299, St. Louis, Missouri 63166, U.S.A 1988. Volume 75, Number 3 Notes 1149 1988 FIGURE l. 4-1 Vantanea d McPherson (A-G drawn from the type, McPherson & Stockwell 10892; H, Twi, Fl D. I from McPherson id . Buds 2n opened flower.—C. Flower with corolla removed.— Gynoecium.—E-G. Stam n5.—H. Dra upe mar Endocarp. J, K. V. occidentalis Cuatracasas (drawn from Gentry 24054) .—J. Gynoecium. Tk Flower with four t removed. NOMENCLATURAL CHANGES IN THE GENUS FUCHSIA (ONAGRACEAE) In Berry (1982), ten new species of Fuchsia were described. A typographical error was made in one of these descriptions (the first **i" was omit- ted) and is corrected as follows: Fuchsia coriacifolia P. Berry, Ann. Missouri Bot. Card. 69: 150. 1982. Examination of specimens from Cambridge Uni- versity revealed that the type of F. parviflora Lindley, treated by Breedlove (1969) as a species of the Mexican and Central American sect. En- cliandra, does not belong to that group. Instead, it belongs to the monotypic sect. Kierschlegeria and is conspecific with the earlier described F. lycioides. As a result, the following nomenclatural changes are nee Fuchsia lycioides Andrews, Bot. Rep. 2: pl. 120. 1800. TYPE: plate 120 of the Botanists Re- pository, vol. 2 (lectotype, here designated). nme parviflora Lindley, Bot. Reg. 13: 1048. 1827. e Canning in 1824, without collector, July l2 lesbos CGE). The type of F. parviflora has alternate leaves and both series of stamens erect, clearly excluding it from sect. Encliandra, in which all members have opposite leaves and the antipetalous stamens reflexed back into the floral tube (Breedlove, 1969). Lindley, in fact, noted the close similarity of F. parviflora to F. lycioides, distinguishing his species mainly by the smaller flower size and the longer petioles. Fuchsia lycioides is subdioecious, how- ever, with the pistillate flowers nearly half the size of the hermaphrodite ones (Atsatt & Rundel, 1982). Lindley's type was from a pistillate individual, whereas Andrews's type of F. lycioides was from ANN. Missouni Bor. Garp. 75: 1150. 1988. a hermaphrodite plant with larger flowers. Leaf size in F. lycioides, on the other hand, is too variable to distinguish it from F. parviflora. Quite likely Lindley's report of F. parviflora as a native of Mexico was in error, since F. lycioides is re- stricted to a narrow coastal area of central Chile, and his type was from a specimen cultivated in England. Fuchsia cylindracea Lindley, Bot. Reg. 24: 66. 1838. TYPE: cultivated at the Horticultural Society, London, England, raised from Mex- ican seeds presented by George Barker, with- out collector (lectotype, CGE). Fuchsia parviflora Lindley, sensu Breedlove, Univ. Calif. Publ. Bot. 53: 56. 1969. The type sheet of F. cylindracea has two sep- arate branches, one male and the other female. The male portion is here designated as the lecto- type, since the species is dioecious (Breedlove, 1969), and an illustration of a male branch ac- companies the type description. LITERATURE CITED ArsaTT, P. R. & P. RuNDEL. 1982. Pollinator main- tenance vs. fruit production: partitioned reproductive effort in subdioecious Fuchsia lycioides. Ann. Mis- souri Bot. Gard. 69: 199-208. BERRY, P. E. 1982. The systematics and evolution of Fuchsia sect. Fuchsia (Onagraceae). Ann. Missouri t. Gard. 69: 1-198. BREEDLOVE, D. E. 1969. The systematics of Fuchsia section Encliandra (Onagraceae). Univ. Calif. Publ. Bot. 53: 1-69. —Paul E. Berry, Universidad Simón Bolívar, Departamento de Biología de Organismos, Apar- tado 69000, Caracas 1081, Venezuela. BACTRIS DIVISICUPULA AND BACTRIS FUSCOSPINA REEXAMINED Work on a forthcoming paper on the palms of western San Blas, Panama required the determi- nation of similar-looking specimens labeled Bactris divisicupula Bailey (Palmae, Cocoeae) or B. fus- cospina Bailey. Bailey (1943a) described both species, from one specimen each, from semi-iso- lated peaks in west-central Panama. The type col- lections actually intergrade, and collections since then also bridge the ines da indicating that the two taxa are synonym Bailey (1943b) uer he following couplet to distinguish the types of B. divisicupula (Allen 1817) and B. fuscospina (Allen 2086): e. Pinnae short-caudate or only acuminate, marked between the nearly entire margin and Rn ee. Pinnae long DRM cupule divided into deep lobes in both series B. divisicupula Collections amassed since 1943 demonstrate that leaf shape and venation in this species are variable and do not provide consistent separation. Allen 2086 bears an inflorescence just past flowering, while Allen 1817 has mature fruits. The difference in developmental stages may have been a reason Bailey described two species. He stated (1943a: 230) that the setae of the calyx of B. fuscospina (from the less-developed material of Allen 2086) e "likely to perish with handling," which does occur, thus rendering the vestiture indistinguish- able from that of the more-developed B. divisi- cupula. Allen 1817 actually retains some setae on the calyx, identical to those of Allen 2086. The lobing of the “cupule” (corolla) is also an artifact of development, the corolla being parted into lobes by the expanding fruit. I here include B. fuscospina in synonymy with B. divisicupula. Bactris divisicupula Bailey, Gent. Herb. 6: 230. 1943. TYPE: Panama. Cocle: Él Valle de An- ton, 21 May 1939, Allen 1817 (hololecto- type, MO; isolectotypes, BH, GH, here des- ignated). B. NEM Bailey, Gent. Herb. 6: 228. 1943. TYPE: a. Panamá: Cerro Campana, 31 Dec. 1939, Allen 2086 (hololectotype, MO; isolectotype, BH, here designated). Bailey did not designate a holotype. The BH and GH specimens are fragments. The description appears to be based on the MO sheet, which is the only one containing the information included in the description. For these reasons the MO duplicate is here chosen as lectotype. When Bailey described the two palms, the flow- ers were unknown. The collections de Nevers 6127, Moore 6531, and Johnston 1552 allow the fol- lowing description: Flowers crowded on the rachillae; pistillate flow- ers in triads with 2 staminate flowers; the triads mixed with numerous solitary staminate flowers in proximal third of rachillae; distal 7; of rachillae with exclusively staminate flowers; staminate sepals 3, unequal, linear, connate at base, 1 mm long; sta- minate petals connate in lower third, free and val- vate above, 5 mm long, thick, apically acute, ine- quilateral; stamens 6, the filaments adnate to the peur basally, free distally, twice folded or bent, a. 2 mm long; anthers ovate, dorsifixed just below middle, dehiscing longitudinally; pistillode mi- nute; pistillate sepals connate into a vertically striate tube 3-4 mm long, minutely 3-lobed at apex; pis- tillate petals connate into a tube 2 mm long, trun- cate to obscurely 3-lobed apically, minutely spi- nescent without, striate within; staminodes absent; stigma sessile, truncate, 0.5 mm long; ovary ob- long. Ovule basal. Bailey (1943a) described the rachillae of the two species as ““pubescent” (B. divisicupula) and “indifferently pubescent” (B. fuscospina). The pu- bescence is strongly eroded in the material he saw: the indument of the rachillae is a dense mat of multicellular hairs, each reminiscent of a string of beads. Moore 6531 has these hairs well developed and is unique among the flowering specimens seen in having the spinescent indument of the corolla tube continue onto the rim of the tube as a fringe. Bactris divisicupula ranges from the provinces of Limón and Puntarenas in Costa Rica to the Department of Valle in Colombia. It inhabits trop- ical moist forest, tropical wet forest, premontane wet forest, and premontane rain forest (sensu Hold- ANN. Missouni Bor. Garb. 75: 1151-1152. 1988. 1152 Annals of the Missouri Botanical Garden ridge et al., 1971) between sea level and 1,000 1986, de Nevers, Herrera & Charnley 7697 (MO); Aila meters. The Colombian material has larger fruits than the Panamanian material and may tie B. divisicupula to additional specimens at MO from Amazonian Peru and Venezuela. Comparison with the Amazonian material is deferred to a later date pending collections from Amazonian Colombia and Ecuador. Additional specimens examined. Costa RICA. LIMÓN: woodlands S of La Lola on the railroad, 120 m, 15 Apr. 1953, Moore 6711 (BH). PUNTARENAs: Palmar, 6 Mar. 1956, Schubert 1184 (A). PANAMA. CANAL AREA: Skunk Hollow, Caribbean side, 22 Oct. 1965, Blum 1496 (MO); Pavón Road W of Gatün Locks, 4 Aug. 1955, Johnston 1538 (BH); near Marú Towers W of Gatún e John- ston 1552 (BH); Pipeline road near Gam 0 m, 7 Nov. 1973, Nee 7846 (MO, NY); Frijoles, 20 pow 1923, Stevens 1185 (US); Agua Salúd, 13 July 1923, Cook & Martin 63 (US); Barro Colorado "mm a July 1931, Bailey 505 (BH). COLON: Santa Rita Ridge, 8 Apr. 1971, Croat 14182 (MO). DARIEN: 2€ River, Duke & Bristan 220 (MO, US). PANAMA: Cerro Campana, 31 Mar. 1969, Dwyer, Croat & Castillón Ha 9 (BH, MO); Cerro yin 1 (MO); 3 mi. NE of A m, 10 Me 1973, Croat 22767 a Cerro Jefe, 23 pa 1946, Allen 3440 (BH, MO); El lano-Carti road km 8, 5 Mar. 1974, Nee & Warmbrodt 10400 (MO). COMARCA DE San Bras: El Llano-Cartí road km 19, 350 m, 9'19"N, 78'55"W, 2 Nov. 1985, de Nevers, j; & Charnley 6127 (MO); ari 30 m, 9'24"N, 79'24"W, 10 Feb. 1986, de Nev > rera 7121 (MO); Cangandi, swampy flats, 10. m, 5 Apr. Tiwar (Rio Acla), 25-100 m, 8'48"N, 77'40"W, 12 Feb. 1 an, 76'37"W, 6 Nov. 1983, Juncosa 1319 (MO): Hoya del Rio San n Quebrada La Sierpe, x 4'10"N, 7710” W, 25 Mar. 1979, Forero et al. 3966 (MO). VALLE: Rio Naya did from Puerto Merizalde, 10 m, 3'15"N 77'25"W, 23 Feb. 1983, Gentry & Juncosa 40679 (MO). Fieldwork in Panama and Costa Rica was sup- ported by a fellowship from the Smithsonian In- stitution. Logistical support and permission to work in the Comarca of San Blas were kindly granted by the Kuna Indians through PEMASKY. Her- caclio Herrera assisted with fieldwork. LITERATURE CITED HE be H. 1943a. New palms in Panama, and s. Gent. Herb. 6: 198-264. "1 Palmaceae. /n: R. E. Woodson, Jr. Schery (editors), Flora of Panama. Ann. Missouri Bot. Gard. 30: 327-396. HorpnipcE, L. R., W. C. GRENKE, W. H. Har T. Lianc & J. A. Tost, JR. 1971. Forest Envi- ronments in Tropical Life Zones. Pergamon Press, New York THEWAY, —Gregory C. de Nevers, California Academy of Sciences, Golden Gate Park, San Francisco, Cal. ifornia 94118, U.S.A. CHROMOSOMAL OBSERVATIONS ON FUCHSIA SPECIES AND ARTIFICIAL HYBRIDS The two largest sections of Fuchsia, sects. Fuch- sia and Hemsleyella, are concentrated in the trop- ical Andes and together comprise 75 of the nearly 105 species of the genus. Both sections have been recently revised by Berry (1982, 1985), who re- ported chromosome numbers for 52 species. Species from both sections are primarily diploid (n = 11; 43 species), but seven are tetraploid (n = 22), and one species has both diploid and tetraploid popu- lations. In an effort to obtain counts for the re- maining species and for interesting new collections or interspecific hybrids in these sections, we ex- amined eight specimens cultivated by members of the Dutch Circle of Fuchsia Friends in 1986. Young floral buds were fixed in Carnoy’s solution and stained in 1% aceto-orcein, as described in Berry (1982). Photomicrographs were taken with a Nikon Biophoto camera using Kodak Technical Pan film. Results of the chromosomal observations are presented in Table 1 and Figure 1. The diploid TABLE l. Additional chromosome counts in Fuchsia.’ counts for F. decussata, F. furfuracea, F. sca- briuscula, and F. simplicicaulis are the first re- ports of chromosome numbers for these species, all belonging to sect. Fuchsia. They lend further support to Berry's (1982) finding of predominant diploidy in that section. The collection of F. cinerea studied, however, proved to be tetrapoloid, unlike an earlier diploid count for another population of this species (Berry, 1982). Fuchsia cinerea occurs in the same high-elevation areas of northern Ec- uador as F. corollata and F. vulcanica, two mem- bers of the F. petiolaris species group that also have tetraploid populations, as well as problemat- ical species limits. A more extensive cytological sampling of the Fuchsia populations in this area would be helpful in resolving the complex variation patterns observed in this group and to determine if tetraploidy has arisen repeatedly in these taxa. Fuchsia magdalenae (sect. Fuchsia) was intro- duced into cultivation just over ten years ago Meiotic hromo- some Taxon Number Collection Data? F. cinerea P. Berry n= 22 Berry 004-86, from seed of Koenen 153-06-81, Prov. Carchi, Ecuador, 6 km NE of El Angel F. decussata R. & P. n-—]l Berry 014-86, from seed of Fg 3049, Dept. Ayacucho, Peru F. furfuracea Johnst. n=11 Berry 010-86, from seed of Solomon 12573, Dept. La Paz, Bolivia F. scabriuscula Benth. n= ll Berry 012-86, from seed of Berry 3574, Prov. Pichincha, Ecuador F. simplicicaulis R. & P. n=11 Berry 016-86, from plants long established in cultivation in Europe, originally from Peru F. magdalenae Munz n-— 22 Berry 017-86, from progeny of the type collection of F. lam- padaria J. O. Wright, originally from Santa Marta, Colom- bia F. magdalenae x F. denticulata 2n = 33? Berry 009-86, artificial cross made by D. Reiman F. magdalenae x F. pilaloensis 2n = 33* Berry 018-86, artificial cross made by D. Reiman ' All species belong to sect. Fuchsia except for F. pilaloensis, from sect. Hemsleyella. ? All == from plants cultivated by Mrs. Drude Reiman-Dietiker in Hollandsche Rading, the Netherlands, with vouchers * Many piii and bridge chromosomes at Anaphase I. ANN. Missouni Bor. GARD. 75: 1153-1154. 1988. 1154 Annals of the Missouri Botanical Garden C FIGURE 1. of Fuchsia. — A. F. magdalenae (2n = 221 (Wright, 1978) and has since been used to produce a novel series of attractive interspecific crosses in England and the Netherlands. The original, paren- tal stock of F. magdalenae was cytologically re- examined, showing it to be tetraploid with normal bivalent formation (Fig. 1A). This agrees with pre- vious counts by Wright (1978) and Berry (1982). The first F. magdalenae hybrid, with the diploid F. denticulata (Berry, 1982), yielded triploid prog- eny with meiotic irregularities such as bridges and laggard chromosomes (Fig. 1B-D). The second F. magdalenae hybrid was with F. pilaloensis, a member of the apetalous sect. Hemsleyella. The chromosome number of this species was not re- ported in Berry's (1985) revision of the section, but the triploid chromosome number of the F, hybrid with F. magdalenae indicates that it must be ae; Unless spontaneous or induced chro- doubling occurs in the F, of these triploids, they are likely to prove infertile and will need to be propagated vegetatively. y ç: ; wi te SN ec D Photomicrographs of meiotic metaphase (A, B) , anaphase (C) , and telophase (D) chromosomes 1). . F. ma iW i enae X F. denticulata (2n = 33) ; note the numerous laggard and bridge chromosomes in C and D. Scale = 10 We wish to thank Mrs. Drude Reiman-Dietiker for providing buds from her extensive collection of living Fuchsia species and hybrids, and Lois Brako for assistance in collecting the bud material. LITERATURE CITED Berry, P. E. 1982. The systematics and evolution of Fuchsia D. ders (Onagraceae). Ann. Missouri Bot. Gard. 69: 1-198. 85. "The systematics of the apetalous fuch- sias of South America, Fuchsia sect. Hemsle bor (Onagraceae). Ann. Missouri Bot. Gard. 72: WRIGHT, J. O. 1978. A a species of Fuchsia L. Bot. J. Linn. Soc. 77: 113-115. —Takuji Hoshino, Biological Laboratory, Oka- yama University of Science, 1-1 Ridai-Cho, Oka- yama 700, Japan; and Paul E. Berry, Univer- sidad Simón Bolívar, Biología de Organismos, Apartado 89000, Caracas 1081, Venezuela. UNA NUEVA COMBINACION EN BAUHINIA (FABACEAE— CAESALPINIOIDEAE) La especie Amaria sessilifolia fue descrita por debería ser considerada como sinónimo de Bau- De Candolle (1825) con base en material colec- hinia petiolata (Mutis ex DC.) Triana ex Hook. cionado por Mutis (Mutis 2724) en Colombia. f. Sin embargo, el examen de las iconografias de Wunderlin (1976, 1983) sugirió que esta especie la Real Expedición Botanica del Nuevo Reino de mem 32 33 34 35 36 37 38 39 40 41 42 43 44 45 5 L š : E 8 a Ë B 2 x - i = t a EE t <=: E : Li 3 R > a butón.—c. Estambre con glándula.—d. Esta- lla. —b. Flor minodio.—e. Pistilo.—f. Frutos.—g. Corte transversal del fruto mostrando la semilla glándulas bien desarrolladas y sésiles a ambos lados cerca de la base del filamento; anteras con 4 ca- vidadees; estambres arreglados en 4 verticilos, los 2 verticilos externos con anteras introrsas, rectan- gulares, con las celdas en 2 planos, las bases de las 2 celdas superiores lateralmente tangenciales a los ápices de las 2 celdas inferiores; las anteras en el 3 verticilo interno oblongas y extorsas; el 4 verticilo formado por 3 estaminodios relativamente grandes (2-2.8 mm de largo), sagitados y con corto filamento, densamente pilosos. Ovario 1.9 mm de largo, 1.5 mm de ancho, obovoido, con escasos pelos seríceos; estilo aprox. 3 mm de largo, con pelos seríceos esparcidos en la base. Frutos glo- bosos, 18-20 mm de largo, hasta 17 mm de ancho, glabros, brillantes, verde amarillentos y algo mo- teado rojizos al madurar; pedicelo 8 mm de largo; mesocarpo delgado y suave; endocarpo delgado y dura. Semilla 1, esférica 12 mm de largo, hasta 12 mm de ancho, cubierto por una delgada testa 1162 Annals of the Missouri Botanical Garden CUADRO 1. de Perú. Algunas características morfológicas de diagnóstico entre Caryodaphnopsis Burgeri y C. fosteri Caracteristica C. Burgeri C. fosteri Pubescencia del envés de la hojas Apice de las hojas Tamano de la panicula Pubescencia de la panicula Longitud y pubescencia de los pe- dicelos Longitud de los tépalos externos Longitud y oo de los té- palos interno Longitud y pubescencia del ovario Tamano del fruto diminuta y esparcidamente sericeas agudo, acuminado, o cuspidado 1-3 cm diminuta y esparcidamente sericea 2.5-9 mm, diminuto sericeos 1.9 mm 7-10 mm, diminuto-sericeos exter- mas largos, argen- teo-sericeos internamente 1.9 mm, con escasos pelos sericeos 1.7 cm diam. densamente pilosas agudo a redondeado 3-6 cm densamente pilosa 5 mm, pilosos l mm 5-6 mm, ferrugineo-pilosos en am- bas superficies 1 mm, glabro + 1 cm diam. que al madurar queda adherida al endocarpo, que- dando un espacio aéreo entre estos y la semilla. Ejemplares examinados adicionales. SAN sag boc de Puriscal, 400 m z V. y otros 3908 TARENAS: Bebo de Fauna Silvestre Golfito, Golfito, 150- 200 N. Zamora V., P. Sánchez m, N. Zamora V. 1070 Dedicamos esta especie en honor al Dr. William C. Burger, editor de Flora Costaricensis, por su gran aporte en las investigaciones botánicas de El actualmente está elaborando la monografia de la familia Lauraceae de Costa Rica. La identificación a nivel genérico fue posible nuestro pais. CUADRO 2. ifolia y C. Burgeri. gracias al análisis anatómico de maderas y hojas realizado por el Dr. Henk van der Werff en 1984 El descubrimiento de esta nueva especie es de COSTA RICA. , L. J. Poveda A., (CR). PUN- Distribución y habitat. especie es endémica para Costa Rica, conocido en Santa Rosa y Mastatal de Puriscal (Prov. San José) y la Peninsula de Osa (Prov. Puntarenas), de 50 a 400 m de elevación, ocupando la Zona de Vida Bosque Muy Hümedo Tropical (Tosi, 1969). Se mucha importancia pues amplia la distribución co- nocida del género en América (ver van der Werff & Richter, 1985; van der Werff, 1986), y abre nuevas posibilidades para entender la distribución geográfica de las especies de esta familia. Algunas características anatómicas comparativas entre Caryodaphnopsis inaequalis, C. theobrom- Caracteristicas del xilema C. inaequalis C. theobromifolia C. Burgeri Vasos Frecuencia Diámetro tangencial de vasos á es Perforaciones Fibras septadas Aliniamiento (sección transversal) Distribución de parenquima Radios Altura de los más largos Ancho de los radios grandes Punteado radio vascular Contenido inorgánicos 9-12 (nr./mm?) 200-250 um exclusivamente simples septos frecuentes irregular (sección transversal de las fibras poligonal) paratraqueal escaso 2-3 mm 5-6(8) células grandes y redondeados sentes 6-7 (nr./mm?) 200-300 um exclusivamente simples septos frecuentes radial a irregular (sección transversal e las fibras poligonal) paratraqueal escaso 2-3 mm 3-6 células grandes y redondeados a ovalados sentes Hasta el momento la Notes 1163 Volume 75, Number 3 1988 sopn118s1;mui u012238 ‘q— ‘Sorpo so] ua (Z) spWOs . SO1PD! OPUDLISOU ]o12u28u0] UQIDIIS 7)—'sDjja ap SVUNÍ]D ap OMUIP (I) souo sp, X sviqyf sv] ap pound vsansZ vj 12 (I) sepa? uoo sajdigjmu sorod opupsjsow jps12asunaj] UOIDIIS `F E ee LU Qe... — A e € (T) spsafiajo soyma) "q— (I) a apand as anb n) ua jvs12asuna] 1»12u28u» X ppssaasups} s2u01228s ‘Zg sisdouydepodsey ap vapo 'c vun514 1164 Annals of the Missouri Botanical Garden — P q —4 e a id t Ti i 2 — = = o. [ Wn . ot I^ p ` Madera de Caryodaphnopsis Burgeri, sección radial.—A. Punteado radio-vascular (1).—B. Punteado intervascular alterno (1).—C. Gomas (1) en las cavidades de las fibras. —D. Radios heterogeneos (1). Ficura 3. Volume 75, Number 3 1988 Notes 1165 PUT = ees > LM FiGURA 4. vascular ( encuentra en laderas de pendientes medias y sitios planos. Se ha observado buena regeneracion na- tural, principalmente en claros de luz. Produce flo- res desde finales de enero hasta principios de fe- brero, y hemos Snes frutos maduros a mediados de ma En las localidades de Santa Rosa y Mastatal de bre de "carne" o dera tienen excelentes propiedades para la cons- truccion de viviendas. Se compara C. Burgeri con C. fosteri en Cua- dro 1 DESCRIPCIÓN ANATÓMICA DE LA MADERA DE CARYODAPHNOPSIS BURGERI Propiedades generales Madera dura y pesada. Duramen color pardo- claro a pardo-rojizo, con vetas longitudinales os- curas, casi negras. Presenta olor caracteristico, pero ningun sabor en especial; textura media y lustre de mediano a elevado Descripción macro- y microscópica Presenta anillos de crecimiento visibles debido a bandas (zonas) tangenciales oscuras, de espesor irregular con poca concentración de poros. -r ~ -€ 4 Madera de Caryodaphnopsis Burgeri, platinas de perforasion simple (1) y punteaduras del segmento Poros pequenos, visibles a simple vista, con dia- metro tangencial promedio de 225 um, abundantes (60 o más por 5 mm), distribuidos en una porosidad difusa con ligera tendencia a formar hileras oblicuas con apud a los radios de más de 4 poros (rango e ). su mayoria son solitarios, algunos multiples radiales de 2 a 4, y ocasionalmente ra- cemiformes. Hay abundancia de tilides. En las caras lebt Dinos los vasos a simple vista como estrias claras sobre un fondo ligera- mente más oscuro. Segmentos vasculares cortos con platinas de perforación simples, éstas de poco a medianamente inclinadas. Su punteado intervas- cular alterno, con punteaduras aeroladas de ta- mano mediano (10 um) y de aperturas angostas. El parenquima axial invisible a simple vista o con lupa. Paratraqueal muy escaso, con algunas pocas células alrededor de los poros. El parenquima radial (radios), apareciendo a simple vista en sec- ción transversal como lineas finas de color claro sobre un fondo más oscuro, hasta 1 mm de altura. Los radios multiseriados en su mayoria, de 3 a 6 células de ancho. Los uniseriados escasos. Hete- rogéneos tipo II (segün clasificación de Kribs, 1968) con 2 o 3 hileras de células marginales erectas y el cuerpo formado por células procumbentes. Hay presencia de radios agregados. Células oleiferas frecuentes, y con gran cantidad de gomas rojizas 1166 Annals of the Missouri Botanical Garden en las células del cuerpo. Punteado radio-vascular de alterno a opuesto, con punteaduras semi-aero- ladas grandes de redondeadas a oblicuas. Fibras del tipo fibro-traqueidas con punteaduras areoladas pequenas, de pared muy gruesa y fre- cuentemente septadas. Contenidos gomozos rojizos en el lumen de las mismas. No presenta marcas de estratificación. Nuestro sincero agradecimiento a Henk van der Werff por la revisión del manuscrito y sus valiosas sugerencias al mismo, a Frank Almeda por facili- tarnos literatura reciente, y a Francisco Hodgson F. por su excelente ilustración. BIBLIOGRAFÍA KosrERMANS, A. J. G. H. 1974. A monograph of Car- yodaphnopsis A. Shaw. Reinwartia 9: 123-137. Kris, D. 68. Commercial Foreign Woods on the Amanoa Market. Dover, New Yor Tost, J. A. 1969. Mapa ecológico de Costa Rica. Centro Cientifico Tropical, San Jose, Costa Rica VAN DER WERFF, A new species ‘of Caryo- oe (Lauraceae) from Peru. Syst. Bot. 11: 415-418 & H. G. RICHTER. 1985. Caryodaphnopsis ed ry Shaw pix a genus new to the Neotrop- s. Syst. Bot. 10: -173. — Nelson Zamora V., Herbario Nacional, Museo Nacional de Costa Rica, Apdo. Postal 749-1000, San José, Costa Rica; Luis J. Poveda A., Escuela de Ciencias Ambientales, Universidad Nacional, Apdo. Postal 86-3000, Heredia, Costa Rica, Herbario Nacional, Museo Nacional de Costa Rica, Apdo. Postal 749-1000, San José, Costa Rica; y Edwin Canessa A., Departamento de Ingeniería en Maderas, Instituto Tecnológico de Costa Rica, Apdo. Postal 159, Cartago, Costa Rica, Cartago. Una Nueva Combinacion en Bauhinia (Fabaceae — Caesalpinioideae) Luz Mila Quinones 1155 Una Nueva Especie de Guettarda L. (Rubiaceae, Guettardeae) para Costa Rica Nelson Zamora V. & Luis J. Poveda A. 1157 Una Nueva Especie de Caryodaphnopsis Airy Shaw (Lauraceae) para la Región Neotropical Nelson Zamora V., Luis J. Poveda V. & Edwin Canessa A. .....— 1160 Book Reviews 1167 CONTENTS Reproductive Biology of Freshwater Aquatic Angiosperms: An Introduction C. Thomas Philbrick Evolution of Breeding TUNE in Eichhornia (Pontederiaceae): A Review Spencer C. H. Barrett ; Distyly and Monomorphism in Villarsia (Menyanthaceae) Some Evolutionary Considerations Robert Ornduff Wind Pollination in Aquatic Angiosperms Christopher D. K. Cook ........... | Morphological Studies of the Nymphaeaceae Sensu Lato. XVI. The Floral Biology f Brasenia schreberi Jeffrey M. Osborn & Edward L. Schneider „n T Reproductive Biology of Vymphaea (Nymphaeaceae) John H. Wiersema uo» Reproductive Biology of Selected Aquatic Plants Robert R. Haynes „n. 4 Pollination Postulates and Two-Dimensional Pollination in Hydrophilous = otyledons Paul Alan Cox & R. Bruce Knox GAR m ib Breeding Systems, Population Structure, and Evolution in Hydrophilous ; Angiosperms Donald H. Les PERSIUS Evolution of Underwater Outcrossing from Aerial Pollination Systems: A Hypothesis C. Thomas Philbrick — Insect Foragers on Solanum Flowers in Australia Gregory J. Anderson & Dee ; Symon 836 LE Two New Species of Calyptrocarya (Cyperaceae: Sclerieae) from Venezuela and Observations a . on the Inflorescence Morphology of the Genus Gerrit Davidse & Robert Kral. 855 : Nigerian Solanum Species of Economic Importance Z. O. Gbile & S. K. Adesina ...... 862. Chromosome Numbers of Grasses (Poaceae) from Southern Africa. I. . Takuji P ei & Gerrit Davidse Ho - Ue a Preliminary List of the Moist Forest Angiosperm Flora of Mwanihana Forest Reserve, 3 Ç Tanzania Jon C. Lovett, Diane M. Bridson & Duncan W. Thomas . Le Experimental Studies on Species Relationships in Erythrina (Legunt P pilo D noideae) ` David A. Neill MÀ Vb The Biosystematics of Ludwigia Sect. Mic/oedr plas (Onagraceae) ` mu Pe mec Reconstructions of Selected Seed Ferns Greg J. Retallack de David L. Dilcher —— 1010 Flora of the Venezuelan Guayana — V Julian A. Steyermark HEURES 22° — 5 Revised Synopsis of Panamanian Euphorbiaceae _ ted L. Webster & Miski J Hut ARS EN : ipte M A New Species of H ( America. Haroldo C. de Lima T At S $ New Species o of Vantanea (Humiriaceae) stan ca : = McP herson Annals of the Missouri Botanical Garden oe y Volume 75, Number 4 Annals of the Winter 1988 Missouri Botanical Garden The Annals, published quarterly, contains papers, primarily in systematic botany, con- tributed from the Missouri Botanical Garden, St. Louis. Papers originating outside the 1 Garden will also be accepted. Authors should write the Editor for information concerning | arrangements for publishing in the ANNALS. Instructions to Authors are printed in the back | of the last issue of each volume. Editorial Committee George K. Rogers | Marshall R. Crosby Editor, Missouri Botanical Garden Missouri Botanical Garden jess Wilson | Gerrit Davidse Editorial Assistant, Missouri Botanical Missouri Botanical Garden Garden , John D. Dwyer Missouri Botanical Garden & Saint Louis University - Peter Goldblatt — Missouri Botanical Garden Henk van der Werff NE Pitaties Garden x For moet information. contact Department ro D ANNALS OF THE MissouRI BOTANICAL € Eleven, P.O. Box 299, St. Louis, MO 63166. Sub- — (ISSN. 0026-6493) is published quarter scription price is $75 per volume U.S., $80 Canada - .. Missouri Botanical Garden, 2345 Tower Grove ' ce: i and Mexico, pou other countries. Airmail deliv. — eme, St. Louis, pes me Second esu: P. SOS pasta SER don changes to us THE MISSOURI BOTANICAL GARDEN, né Eleven, P.O: Box 299, S Lai NO es j Volume 75 Annals Number 4 of the NZ 1988 Missouri Botanical Garden TOWARDS MOLECULAR L. D. Gottlieb? GENETICS IN CLARKIA : GENE DUPLICATIONS AND MOLECULAR CHARACTERIZATION OF PGI GENES"? ABSTRACT I review the phylogenetic implications of eight duplications of nuclear genes encoding isozymes in Clarkia (Onagraceae) . These include ADH, cytosolic PGI, and both we stid and cytosolic isozymes of PGM, 6PGD, and TPI. The PGI Ep anb has been studied intensively from biochemical and genetic standpoints. Recent results have identified two levels of regulation that operate in species with this duplication, one that reduces cytosolic impact of the duplication on abolic function. I also describe our recent cloning and sequencing of tw genes encoding PGI obtained from a genomic DNA library of C. unguiculata, a species with the duplication. The code proteins of 548 and 543 am ds, respectively, an ir predicted amino acid sequences a i muscle PGI. Both p lack introns. The two genes are the first nuclear genes sequenced in wild plants. They are being studied as part of a research program. on gene evolution and the application of nuclear gene sequences for phylogenetic reconstruction in higher plants Questions about phylogeny have the form, “Is with evidence from morphology, cytology, chem- A more closely related to B than to C?" For flow- istry, reproductive compatibility, and other fields ering plants, the best phylogenies are thought to somehow combined. However, accurate phyloge- take into account the “maximum number of at- netic reconstruction is more often a goal than an tributes possible" (Davis & Heywood, 1967: 485), achievement because of problems brought about ' This and the following three papers comprise the proceedings of the Missouri Botanical Garden's 34th Annual Systematics Symposium— Macromolecular Approaches to Phylogeny. The symposium took place in St. Louis, Missouri on October 9 and 10, 1987. ? The molecular genetics results (library construction, gene cloning and qa described in this report were hier in my lab by we R. C. Tait, Debbie Laudencia, and Byron Froman. The molecular genetics research was supported by National Science Foundation grant BSR 86-07054 and ‘USDA 86-CRCR-1-2139. * Department of Genetics, Universit of California, Davis, California 95616, U.S.A ANN. Missouni Bor. GARD. 75: 1169-1179. 1988. 1170 Annals of the Missouri Botanical Garden by character convergence, functional and devel- opmental correlations, and unequal rates of evo- lution in different lineages. The essential difficulty is that little or nothing is known about how genetic changes that affect developmental processes result in differences in character expression. The consequence is that no present procedure can translate the extent of morphological diver- gence into a measure of the closeness of phylo- genetic relationship. I believe the way out of this impasse is to utilize a new source of evidence to assess phylogenetic relationships. The data of mor- phology, the traditional source of information about phylogeny, should be viewed as relevant to studies of plant development. There is good reason to believe that information derived directly or indirectly from the structure and sequence of protein and DNA can be used to settle many phylogenetic questions. Interestingly, in this context, the molecular data are self-sufficient in that their usefulness does not depend on con- cordance with other lines of phenotypic evidence. For example, certain types of changes, particularly duplications of nuclear genes encoding enzymes (Gottlieb & Weeden, 1979; Gottlieb, 1983; Odrzy- koski & Gottlieb, 1984) and large inversions of the chloroplast genome (Jansen & Palmer, 1987) appear to occur only once within a lineage. Thus taxa that now possess them probably descended from a single common ancestor and can be con- sidered monophyletic without regard to their pres- ent morphological and cytological divergence. In addition to phylogenetic inferences made on the basis of unique genetic and molecular traits, cladograms based on shared derived mutations or the extent of overall similarity can be constructed by comparing nucleotide sequences of genes or the size pattern of fragments cut from homologous DNAs by restriction endonucleases. The increasing availability of molecular data suggests that biosys- tematics no longer has to be considered an “un- ending synthesis" (Constance, 1964). hylogenetic relationships can now be deter- mined accurately and reliably at many taxonomic levels. When this is done, the phylogeny can be used as a framework to ask important questions in other areas of biology. For example, how the at- tributes of species reflect both genetic legacy and selected and other changes since their origins, how genetic changes lead to specific modifications of ontogeny that result in new characters, and how and whether these new traits facilitate adaptation to different habitats. From this perspective, phy- logeny can begin to inform both developmental and ecological analyses by providing evidence that A indeed derived from B and not from C. In this paper, I review genetic and biochemical studies from my laboratory on gene duplications in Clarkia with emphasis on their phylogenetic implications. In addition, I describe very recent studies in which we have cloned and sequenced several genes encoding the glycolytic enzyme phos- phoglucose isomerase from genomic libraries of Clarkia DNA infer correct phylogenetic relationships in this well- studied plant genus. When the beginning and end points of species’ genealogies are identified, we can ask about the steps in between. . One purpose of these studies is to BACKGROUN Previous to our studies and, indeed, making them appropriate, were the intensive investigations by Professor Harlan Lewis and his students and colleagues in the 1950s and 1960s (Lewis, 1953, 1962, 1973; Lewis & Lewis, 1955; Lewis & Ra- ven, 1958). They correlated evidence from field studies, morphology. and a major program of hy- bridization and cytogenetical analysis. Clarkia was found to comprise at least 43 species, 33 being diploid. The diploid species were distinguished by substantial amounts of chromosomal repatterning in addition to aneuploidy. The extent of morpho- logical divergence varied from a difference in a single character between some pairs of species to differences in entire suites of traits that might serve as evidence of generic distinction in other plant groups. The degree of morphological resemblance was frequently not concordant with the amount of chromosomal rearrangement. Nevertheless it was possible to discern meaningful phylogenetic pat- terns among the diploid species, and they were assigned to seven taxonomic sections (Lewis Lewis, 1955). Allopolyploid species linked several sections so that as a whole the genus was considered a natural unit. Lewis formulated an elegant model of speciation to account for these relationships. The critical fea- tures of this model included the following: (1) species were regarded as progenitor and derivative and not as siblings; (2) a new species differed from its parent by gross chromosomal rearrangements and some- times by a change in basic number; (3) the spe- ciation process was rapid and abrupt; (4) speciation was independent of the evolution of new adapta- tions and therefore was largely fortuitous; (5) spe- ciation, in general, occurred at the xeric margin of the distribution of the parent species. Volume 75, Number 4 1988 Gottlieb Molecular Genetics in Clarkia 1171 Lewis’s model and his proposed examples of progenitor-derivative species made Clarkia appro- priate for the first studies carried out in plants that applied electrophoretic analysis of enzymes to as- sess the amount of genetic divergence correlated with speciation (Gottlieb, 1973, 1974a). The ratio- nale behind these studies has been reviewed by Crawford (1983, 1985) and by me (Gottlieb, 1977, 1981, 1986). In addition to information about variation (pres- ence, number, frequency) and divergence of alleles at loci coding enzymes, electrophoretic patterns provide evidence about the number of isozymes of particular enzymes and, thereby, the number of pos loci. a" more and more species of Clarkia xamin d, it became apparent that they some- times differed among themselves or from other sulted from duplications of the coding genes (Gott- lieb, 1977; Gottlieb & Weeden, 1979; Pichersky & Gottlieb, 1983). Examination of the number of isozymes in a broad array of higher plants, including conifers and angiosperms, showed that isozyme number was highly conserved and depended on the number of subcellular compartments in which a particular cat- alytic reaction occurred (Cottlieb, 1982). For ex- ample, in diploid plants, enzymes of glycolysis and the oxidative pentose phosphate pathway are en- coded in the nucleus and are generally found as two isozymes, one located in the plastids and the other in the cytosol. When the number of isozymes within a particular compartment is more than one, it probably results from duplication of the structural gene or, in polyploid plants, from additive expres- sion of the genes in the several constituent ge- nomes. Since the conserved number of isozymes reflects the metabolic requirements of plant cells, a reduced number is not possible because it would be lethal. (Failure to observe bands of enzyme activity following electrophoresis of plant extracts should not be taken as evidence that the enzyme is not present in the extract, a common error in many surveys of electrophoretic variation in plants that report the absence of an expected enzyme band as a null allele.) The rules for recognizing duplicate isozymes, following electrophoresis of plant extracts, have been thoroughly described (Cottlieb, 1982). It is worth noting again that subcellular location furnishes the best criterion for recognizing the homology of isozymes from different species, and that the rules apply only to enzymes assayed with natural in vivo substrates. No regularities have been identified in the number of isozymes of en- zymes such as esterases, phosphatases, and per- oxidases that are generally assayed with artificial substrates. GENE DUPLICATION IN CLARKIA The first duplicate isozyme discovered in Clark- ia was that of alcohol dehydrogenase (ADH) in C. franciscana (Gottlieb, 1974b). Its absence from the closely related C. amoena and C. rubicunda, along with the very low genetic identity between C. franciscana and these species (Gottlieb, 1973), helped reject the hypothesis (Lewis & Raven, 1958) that C. franciscana was a recent derivative of C. rubicunda. The genetic evidence for duplication of ADH in C. franciscana was based on its exhib- iting a true-breeding, three-banded electrophoretic pattern, whereas similar three-banded patterns in the related species resulted from heterozygosity at a single locus as evidenced by segregation patterns in progeny. Since C. franciscana did not display polymorphism for ADH, the duplication model was tested by making interspecific hybrids between it and C. amoena. homozygous at a single locus for an allele that encoded a slow ADH variant. The F, hybrids dis- played a five-banded pattern that could only have resulted from the dimeric associations of three dif- ferent polypeptides and, consequently, they must have possessed three genes (Gottlieb, 1974b). The ADH duplication was the second duplication of a gene encoding an enzyme discovered in plants. The first, in maize, was also an ADH (Schwartz & Endo, 1966). Seven additional duplications of genes in Clarkia have since been described and, for each, the taxo- nomic distribution within the genus has been de- termined (Table 1). These duplications are cytosolic phosphoglucose isomerase (PGI) (Gottlieb, 1977; Gottlieb & Weeden, 1979), plastid and cytosolic triose phosphate isomerase (TPI) (Pichersky & Gottlieb, 1983), plastid and cytosolic 6-phospho- gluconate dehydrogenase (6PGD) (Odrzykoski & Gottlieb, 1984), and plastid and cytosolic phos- phoglucomutase (PGM) (Soltis et al., 1987). De- tailed information about them is available in the individual reports. Five of the seven duplications (plastid and cytosolic 6PGD, plastid and cytosolic TPI, and plastid PGM) are present in species of all diploid sections of Clarkia (Table 1), suggesting they are at least as old as the genus. But only the duplicated plastid TPI was found in every species. e C. amoena plants used were 1172 Annals of the Missouri Botanical Garden PLASTID AND CYTOSOLIC 6PGD Four species of Clarkia appear to lack one or both 6PGD duplications (Odrzykoski & Gottlieb, 1984). Clarkia rostrata and C. epilobioides have a single plastid isozyme and a single cytosolic one and, consequently, lost both duplications. Clarkia lewisii and C. cylindrica have duplicated plastid 6PGDs but only a single cytosolic 6PGD (Table 1). The four species have been assigned to sect. Peripetasma, with the morphologically similar and crossable (Davis, 1970) C. rostrata, C. lewisii, and C. cylindrica to one subsection and the dis- tinctive and highly self-pollinating C. epilobioides to a monotypic subsection (Lewis & Lewis, 1955). The close relationship of the former three species suggested that the loss of the duplicated cytosolic 6PGD occurred in their common ancestor and was subsequently followed in C. rostrata by an addi- tional mutation or chromosomal deletion that si- lenced a duplicated gene encoding a plastid 6PCD. Since C. epilobioides also lacked both duplications, it seemed reasonable to suggest that it was closely related, although it was not possible to decide if the loss of its plastid 6PGD duplication was inde- pendent of the loss in C. rostrata. The matter was settled by a restriction endonu- clease analysis of chloroplast DNA carried out on all the species in this section, which revealed that C. rostrata and C. epilobioides were sister species and that C. lewisii and C. cylindrica comprised a second pair of sister species (Sytsma & Gottlieb, 1986a). The chloroplast DNA study also showed that the two pairs of species share a common ances- tor well removed from the other species of the section. Thus, even though C. rostrata is not mor- phologically similar to C. epilobioides and was placed two species have a close genealogical relationship. Since this phylogenetic inference was based on evidence from both nuclear genes and chloroplast DNA, it is par- ticularly strong. in a different subsection, the PLASTID AND CYTOSOLIC TPI Both TPI duplications appear to be present throughout Clarkia (Table 1), although some un- certainty remains in regard to the cytosolic TPI in sect. Eucharidium for which the genetic analysis is incomplete (Pichersky & Gottlieb, 1983). Elec- trophoretic studies of TPI have also been carried out on a number of species of other genera of Onagraceae to ascertain the taxonomic distribution of the duplications outside of Clarkia. Since suf- ficient (or appropriate) material was not available to conduct genetic analysis, three criteria had to be met to warrant the hypothesis that a given species possessed a TPI duplication. The minimum number of electromorphs per individual for each isozyme had to be at least three (TPI is dimeric), the multiple isozymes had to be located in the same subcellular compartment, and a side-by-side com- parison of leaf and pollen extracts had to show the same number of cytosolic isozymes (the criteria are discussed in detail in Gottlieb, 1983). On the basis of satisfying all of these criteria (although sample sizes were very limited), the cytosolic TPI dupli- cation was identified in five of the seven tribes of the family, including Jussiaeeae (Ludwigia), Fuch- sieae (Fuchsia), Hauyeae (Hauya), Onagreae (Clarkia, Heterogaura, Camissonia, Calylophus, Gongylocarpus, and Oenothera), and Epilobieae (Boisduvalia) (Pichersky & Gottlieb, unpubl.). The presence of the duplication in both Fuchsia and Ludwigia, the two most ancient lineages in the family (Raven, 1979), suggests its great antiquity. In contrast, the plastid TPI duplication was not identified outside of Clarkia and must have arisen much more recently. Although these results should be regarded as exploratory, they point out the possibility that certain taxonomically widespread duplications may ne uselul to erp genera (and eventually f. semblages However, since the time spans int are great, it would be appropriate and ONE to validate the conclusions by examination of the nucleotide sequences of the duplicated genes. PLASTID AND CYTOSOLIC PGM In contrast to the situation in 6PGD in which the absence of duplicated genes could be assigned to some type of mutation in common ancestors of extant species, the loss of the plastid PGM dupli- cation (Table 1) in C. concinna and in C. lasse- nensis (Soltis et al., 1987) must be regarded as independent events in lineages directly ancestral to these species but to no others, since the two species belong to distantly related sections of Clark- ia (Lewis & Lewis, 1955). The presence of the cytosolic PGM duplication in C. arcuata (sect. Rhodanthos) and in all species of sections Godetia and Myxocarpa (Table 1) is consistent with a taxonomic assignment previously made by Lewis & Lewis (1955). They proposed p. 261) that sect. Rhodanthos (then designated Lek Primigenia) was “probably directly ances- tral" to sect. Godetia and “perhaps” to sect. Myxocarpa. Within sect. Rhodanthos, the rele- vant lineage is now represented by C. arcuata 5 Volume 75, Number 4 1988 ottlieb Molecular Genetics in Clarkia 1173 TABLE 1. The phylogenetic distribution of duplicate isozymes in diploid species of Clarkia. The PGI data are from Gottlieb & Weeden (1979), the PGM data from Soltis et al. (1987), the 6PGD data from Odrzykoski & Gottlieb (1984), and the TPI data from Pichersky & Gottlieb (1983). The numeral 1 indicates the species has a single isozyme and the numeral 2 indicates duplicated isozymes. For each enzyme, plastid (Pl) and cytosolic (Cy) isozymes are indicated. Isozyme Number : PGI PGM 6PGD TPI Section e Species Cy Pl Cy Pl Cy Pl Cy Eucharidium C. breweri 2 2 1 2 2 2 T C. concinna 2 1 1 2 2 2 ? Fibula C. bottae 2 2 1 2 2 2 2 Peripetasma C. cylindrica 2 2 1 2 1 2 2 C. lewisü 2 2 1 2 1 2 2 C. epilobioides 2 2 1 1 1 2 2 C. rostrata 1 2 1 1 1 2 2 C. biloba subsp. australis 2 2 1 2 2 2 2 C. dudleyana 2 2 1 2 2 2 2 C. lingulata 2 2 1 2 2 2 2 C. est 2 2 l 2 2 2 2 Heterogaura heterandra 2 2 l 2 2 ? 2 Phaeostoma C. xantiana 2 2 l 2 2 2 C. unguiculata 2 l 2 2 2 2 Godetia C. imbricata l 2 2 2 2 2 2 C. nitens l 2 2 2 2 2 2 C. speciosa subsp. polyantha 1 2 2 2 2 2 2 C. williamsonii 1 2 2 2 2 2 2 Myxocarpa C. mildrediae 1 2 2 2 2 2 2 C. virgata 1 2 2 2 2 2 2 Rhodanthos C. arcuata 1 2 2 2 2 2 2 C. lassenensis 1 1 1 2 2 2 2 C. amoena subsp. huntiana 1 2 1 2 2 2 2 C. franciscana 1 2 1 2 2 2 2 C. rubicunda 1 2 1 2 2 2 2 which, together with C. lassenensis, was placed in a distinct subsection. The other subsection con- taining diploid species includes C. amoena, C. rubi- cunda, and C. franciscana, and it then would represent the lineage from which the other four sections of Clarkia (Table 1) eventually evolved. Alternatively, the cytosolic PCM duplication may ave had independent origins in C. arcuata and sections Godetia and Myxocarpa. Sequence com- parisons of the PGM genes will make it possible to distinguish these models. Regardless of the outcome of such comparisons, the taxonomic distribution of the cytosolic PGM duplication is independent of the sectional phylog- eny suggested (Lewis, 1980) following the discov- ery of the cytosolic PGI duplication (Gottlieb, 1977; Gottlieb & Weeden, 1979), since the two dupli- cations are not present together in any species (Table 1). The PGM evidence suggests that the four sections that have the PGI duplication (Table 1) arose from the lineage within sect. Rhodanthos that also gave rise to C. amoena, C. rubicunda, 1174 Annals of the Missouri Botanical Garden and C. franciscana. It is also an interesting pos- sibility that since the two enzymes catalyze adjacent reactions in glycolysis and gluconeogenesis (PCI interconverts fructose-6-phosphate and glucose-6- phosphate, and PGM interconverts the latter and glucose-1 phosphate), there may be metabolic rea- sons that select against the occurrence of both duplicated enzymes in the same cytosol. Overall, the genetic and biochemical evidence from the several gene duplications provides a re- markably consistent and coherent picture of the phylogenetic relationships within Clarkia. The evi- dence is also consistent with the recent discovery based on restriction endonuclease patterns in chlo- roplast DNA that the monotypic Heterogaura het- erandra (Table 1) is actually a Clarkia and closely related to C. dudleyana (Sytsma & Gottlieb, 1986b PGI The PGI duplication in Clarkia has been studied intensively because it was one of the first dupli- cations identified that is present in some but not all species of a single genus. Thus, it is possible to compare duplicate PGI genes and their products with their nonduplicate homologues, and the com- parisons can be done in species having a relatively similar genomic background. The example provides rtunity to examine the critical early an unusual opp stages of gene evolution and to test the general model that major changes in gene regulation, struc- ture, and function cannot evolve without the avail- ability of duplicate sequences. The duplication characterizes all of the species (except C. rostrata) in the morphologically advanced and diverse sections Kucharidium, Fib- ula, Phaeostoma, and Peripetasma, and is absent from sections Godetia, Myxocarpa, and Rhodan- thos (Table 1). Consequently it identifies a specific branching point in the phylogeny of Clarkia and serves to group the former four sections into a monophyletic lineage (Gottlieb & Weeden, 1979; Lewis, 1980). The realignment was effected with- out having to move any species into or out of any section (Lewis, 1980). Genetic studies revealed that the duplicate PGI genes assort independently (Gottlieb, 1977; Gott- lieb & Weeden, 1979; Weeden & Gottlieb, 1979), which is thought to mean that they arose by a process involving overlapping reciprocal translo- cations or insertional translocations rather than unequal crossing-over. The relevant arguments were presented in Gottlieb (1983). Many other duplicate genes in plants also assort independently (Tanksley, 1987). The mode of origin is important for phy- logenetic reconstructions because chromosomal rearrangement is much more likely than unequal crossing-over to occur only once for a particular chromosome segment in a particular linkage. Link- age relationships for the other duplications in Clarkia have not been studied in similar detail, although we do know that the duplicate genes en- ts plastid TPIs and one of them and a cytosolic ene also assort independently (Pichersky & Gottlieb, 3). A number of biochemical studies were carried out to determine how much and what type of di- vergence marked the duplicate PGI isozymes. Three results are noteworthy, one having to do with the molecular weight of PGI subunits and the other two with the evolution of regulatory factors that appear to modulate the expression of the duplicate PGI genes. PGI subunits encoded by the duplicate genes have different apparent molecular weights (appar- ent because the values were obtained from their electrophoretic mobility on SDS gels), with PGI-2 being 60,400 and PGI-3 59,000, or values closely similar (Gottlieb & Higgins, 1984a). Species in sect. Myxocarpa that lack the duplication have PGI subunits with molecular weight of 60,400, and PGIs from sections Godetia and Rhodanthos weighed in at 59,000. The presence of two mo- lecular weight forms in species with the duplication and each molecular weight form by itself in species without the duplication was unlikely to have come about by chance. The result suggested the novel possibility that the PGI locus in an ancestal Clarkia was translocated to different nonhomologous chro- mosomes, that the genes then accumulated mu- tational changes that encoded different molecular weight subunits, and that lines carrying the differ- ent chromosomes eventually hybridized with both PGI genes becoming segregated into a single ge- nome by a process originally documented in maize (Burnham, 1962) involving overlapping reciprocal translocations. The scenario seems feasible for Clarkia, in which species are distinguished by gross amounts of chromosomal rearrangement, and which all have a self-compatible breeding system permit- ting chromosomal heterozygotes to be made homo- zygous and true-breeding by self-pollination. The merits of this speculation can be directly tested by comparing nucleotide sequences of PGI genes from species with and without the duplication (see below). After it became apparent that the catalytic prop- erties of the duplicate and nonduplicate PGIs were alike (Higgins & Gottlieb, 1984), studies turned to questions about increased gene dosage and whether Volume 75, Number 4 1988 Gottlieb Molecular Genetics in Clarkia 1175 it caused increased levels of cytosolic PGI activity and protein. The PGI levels in clarkias with and without the duplication were assessed by immu- nological means using an antiserum specific to cy- tosolic PGI (i.e., one that does not cross-react with plastid PGI). The result was clear-cut. The two types of species had the same levels of cytosolic PGI activity and protein, suggesting that some form of regulation had evolved that “compensated” for the duplicated genes (Gottlieb & Higgins, 1984b). The activity level proved to be the same as that in a number of diploid vegetables, indicating that green plants generally maintain a similar PGI level. This finding provided an important rationale for the evolution of dosage compensation because it re- stored an activity level characteristic of typical diploid plants having a single cytosolic PGI. Thus a regulatory gus had evolved that reduced the impact o plica To determine whether the regulation operated via metabolic or genetic factors, a series of null activity mutants of each duplicate gene was in- duced by ethyl methanesulfonate (EMS) treatment of seedlings of C. xantiana (Jones et al., 1986). Metabolic factors would be implicated if lesions induced in either gene did not change PGI levels. In homozygous state, each mutant completely lacked the homodimer activity normally specified by the affected gene. The mutants were back- crossed to wildtype for five generations, making it possible to assign changes in PGI activity directly on metabolic function. to the mutation and not to unknown factors in the background. Immunological analysis revealed that they reduced PGI activity in direct proportion to the normal contribution of each gene. The ho zygous mutants at Pgi2 reduced cytosolic PGI activity to 36% of wildtype, and the mutant at Pgi3 to 64%. The effects of the mutations at the two loci were additive. Thus, Pgi2""2^, Pg 3s 3nut plants synthesized in an F, progeny from experi- mo- mental hybrids between the two mutants exhibited only 1446 of wildtype activity. The double homo- zygous null was lethal. The results demonstrated that PGI activity in plants having the duplication is not directly regulated by metabolic factors, war- ranting the suggestion that the dosage compensa- tion depends on factors that regulate the levels of transcription or translation (Jones et al., 1986). Since Pgi3 contributes less than Pgi2 to the total cytosolic PGI activity, the regulatory factors ap- pear to operate to a greater extent on the former locus. Thus, two levels of regulation were identified, one that reduces cytosolic PGI activity in species with the duplication to the level characteristic of species without the duplication, and the second that results in differential accumulation of the products of the duplicate genes. The genetic and biochemical analyses of PGI in Clarkia identified a number of interesting questions that can be answered only with evidence from the sequences of the coding genes. For example, in terms of phylogenetics, it is necessary to test the major hypothesis that the duplication had a unique origin, with the consequence that the four sections that possess it are monophyletic. A corollary hy- pothesis is that the origin of the duj involved hybridization between lineages now represented by Myxocarpa (which has the higher molecular weight subunit) and Godetia/ Rhodanthos (with the low molecular weight subunit). The hypotheses can be tested by comparing the sequences of duplicate and nonduplicate PGI genes. On the hypothesis, Pgi2 from a species with the duplication should be similar to Pgi from Myxocarpa, and Pgi3 from the species with the duplication should be similar to Pgi from Godetia/ Rhodanthos. In other words, the duplicate genes should be more similar to genes from different species than they are to each other. Other questions of interest in a context of evo- lutionary biology have to do with the extent of PGI sequence divergence in species with the duplication versus those without the duplication, the extent of polymorphism for PGI genes in natural populations of Clarkia, and the value of the sequences to demonstrate phylogenetic relationships outside of Clarkia, particularly among the diverse genera included in tribe Onagreae. A different set of questions must be answered to explain how the cytosolic PGI level is reduced in species with the duplication to that characteristic of those without it, to determine the basis for the near 2:1 difference in PGI activities attributable to the duplicate genes, to learn why Pgi2 encodes a higher molecular weight unit than Pgi3, and the nature of the mutations that eliminated PGI activity in the EMS-induced null mutants. CLARKIA PGI GENE SEQUENCES Headway on these questions can now be made because we have cloned and sequenced PGI genes from several Clarkia genomic libraries. Here I describe how these genes were obtained, evidence that they encode PGI, and their general structure. Detailed characterizations and sequences will be -Eai separately. To my knowledge, the Clark- a PGI genes are the first nuclear genes from wild EM that have been sequenced. Our first genomic library was constructed with DNA isolated from seedlings of a horticultural strain 1176 Annals of the Missouri Botanical Garden B H SpP S Bg SEPN PK P Py KPS PP P B LÀ 000000 | n pe " sharps map d sequencing strategy for the Clarkia rai acpi U2 gene which encodes PGI. mH] fra ent. Restriction sites shown e BamH1 (B), Hpal (H), Sph sent on a 4.45-kb Ba gm Gor “Pst! (P). Sal/ (S), Bg 1111 (Bg) EcoR 1 (E), Pvull (Pv), Ncol (N), per Kpn/ (K). The. arrows above and below the restriction map show e direction and extent of sequencing for the individual M13 subclones. 3.8 kb including the entire coding ipi was sequenced on one strand, and 1.6 kb on the complementary strand. of C. unguiculata (Northrup King “Clarkia Double Mixed Colors"), a species with the PGI duplication. Horticultural material was used because very large amounts of seed could be purchased, permitting us to fine-tune our techniques prior to studying natural populations. The DNA was extracted by a proce- dure modified from Fischer & Goldberg (1982) that yielded a nuclear pellet that, after lysis, pro- vided high molecular weight DNA fragments (greater than 100 kb). The DNA was partially restricted with Sau3a and fragments between 15 and 23 kb obtained by fractionation on a sucrose gradient. After determining optimal ratios of chro- mosomal DNA and vector arms, the DNA frag- ments were cloned into the BamHl site of the lambda replacement vector Charon 35. The re- sulting library is estimated to contain 1.8 x 10° phage with 88% recombinants and represents about seven Clarkia genomes. The library was screened at low stringency (51°C, 5 X SSPE) with an 800-bp DNA fragment of a yeast gene encoding PGI, kindly provided by a biotechnology firm. Since we expected low to very low homology between the yeast and Clarkia PGI TABLE 2. Homology between predicted amino acid sequences from nucleotide sequences of U2 and U8, cloned from a genomic library of Clarkia unguiculata and amino acid sequences of cyanogen-bromide pep- tides purified from pig muscle PGI (Achari et al., 1981). Number Identical Amino Acids Total N umber of Am Homol- Sequence Acids ogy U2 vs. U8 319/548 58% U2 vs. Pig 110/166 66% U8 vs. Pig 108/166 65% U2 vs. U8 (in sequences covered by Pig peptides) 89/165 54% sequences, the screening conditions were deter- mined in a prior experiment in which the probe was hybridized on a Southern blot to genomic C. unguiculata DNA digested with several restriction enzymes. Two positive clones were obtained from the first 30,000 plaques examined. They were pu- rified, and DNA prepared from each was restricted with several enzymes, subjected to agarose gel elec- trophoresis, and analyzed by Southern blots using the yeast PGI DNA fragment as probe. The two clones had inserts of 13.7 and 15 kb, which proved different. Hybridizing fragments of the former clone, designated U2, were cloned into M13mpl10 and partially sequenced. The sequences showed ho- mology to that of the yeast gene. A 4.45-kb BamH1 fragment (Fig. 1) was then subcloned into pUC19 and deletion fragments constructed using the ex- onuclease III-S1 protocol of Henikoff (1984). One strand of 3.8 kb including the entire coding region was completely sequenced, and 1.6 kb was se- quenced on the complementary strand by the di- deoxy sequencing protocol (Messing, 1983). The U2 sequence revealed an uninterrupted open read- ing frame of 1,644 nucleotides encoding a protein of 548 amino acids. The identity of U2 was established by comparing its predicted amino acid sequence with the amino acid sequences of five cyanogen-bromide peptides obtained from pig muscle PGI (Achari et al., 1981). These are the only PGI sequences, protein or DNA, that are published for any organism. The five pig peptides identify a total of 166 amino acids, about 30% of the complete protein. The U2 gene encodes amino acids that are identical to those in pig PGI at 110 of these 166 residues, or 66% of the total (Table 2). A second PGI gene, called U8, also obtained from the C. unguiculata genomic library, using U2 as the probe, was found that contains the same sequence present in the 15-kb insert noted above. A similar isolation and sequencing strategy was used to characterize the U8 clone as was used for U2. U8 proved to have a 65% ho- Volume 75, Number 4 1988 Gottlieb Molecular Genetics in Clarkia 1177 NES TENE RENE FG Pic " U 2 318-407 uim RRR K P 408-467 468-527 FIGURE 2. genes with the amino acid sequences from five cyanogen-bromide peptides from pig musc 1981). An open box drawn across the black bars indicates the same amino acid 528-518 Comparison of the predicted amino acid sequences encoded by Clarkia popa U2 and U8 e PGI (Achari et al., Eus in the corresponding position on two or three of the sequences. The amino acids are numbered on the right beginning with the first U2 sequence. U2 encodes 548 amino acids, U8 encodes 543 amino acids, and the total number i in the pig peptides is 166. The diagram represents the best fit by eye, taking into account several short insertions and deletions in the sequences. mology to pig PGI (Table 2), and encodes a protein of 543 amino acids. The predicted amino acid sequences show that U2 and U8 are 58% homologous over their entire coding regions. Comparing U2 and U8 only in the regions covered by the pig peptides, the two se- quences are 54% identical (Table 2). Thus, the two Clarkia PGI genes differ more from each other than either does from pig PGI. The homology of the Clarkia and pig sequences is diagrammed in Figure 2. The two Clarkia proteins exhibit large blocks of very high amino acid identity as well as 1178 Annals of the Missouri Botanical Garden many shorter regions of nonidentity. Several lengthy portions of the three sequences show complete iden- tity. Overall, the high homology between the pig GI amino acid sequences and the predicted Clark- ia amino acid sequences establishes with certainty that both Clarkia genes encode PGI. On the basis of lack of interruption in their open reading frames and the lengths of their sequences, which encode proteins that have closely similar molecular weights to that previously determined for Clarkia PGI, neither Clarkia gene appears to include introns (and see below). Otherise, both genes have many features expected of eukaryotic genes, including potential TATA boxes and other up- stream regions similar to known regulatory se- quences. A complete transcriptional characteriza- tion of the genes will be reported separately. Clarkia unguiculata possesses the PGI dupli- cation, and its genome must include two loci en- coding cytosolic PGIs and one locus encoding plas- tid PGI. Since a heterologous probe was used to obtain the U2 and U8 PGI genes, it was necessary to determine which isozyme is encoded by each gene. A priori, the expectation was that sequences encoding the cytosolic PGIs would be more similar to each other than either would be to the plastid PGI. Genes encoding plastid and cytosolic glycolyt- ic isozymes have been cloned and sequenced in plants only for tobacco glyceraldehyde-3-phos- phate dehydrogenase (G3PD) (Shih et al., 1986), and the results of that study are closely relevant to our research with PGI. Comparison of predicted amino acid sequences from cDNAs showed that the tobacco cytosolic G3PD was more similar to other eukaryotic G3PD enzymes, with about 65% homology, than it was to the tobacco plastid iso- zyme, with 45% homology. The homology of U2, U8, and pig PGI are roughly similar to these values, but we were able to compare only a few sequences. Our initial attempt to identify the isozymes en- coded by the Clarkia genes centered on the search for correlation between restriction length fragments and allelic segregation. This could be followed by PGI activity staining on starch gels following elec- trophoresis of leaf extracts and correlated with the RFLP segregation. To date, we have examined a number of DNAs from single C. unguiculata plants by restriction analysis followed by electrophoresis and Southern blotting. The DNAs proved highly polymorphic, but we have been able to match re- striction fragments to U2. U8 and several other genes cloned from the C. unguiculata library have not yet been similarly matched, but this is not unexpected, since the library was made from DNA isolated from a bulk of hundreds of individual seed- lings and this commercial strain is highly poly- morphic. However, a different procedure suggests that U8 encodes the slowly migrating allozyme PGI- 3B, a cytosolic isozyme. The U8 sequence was inserted in pUC18, downstream from the beta- galactosidase promoter. When the operon was in- duced by IPTG, the E. coli host synthesized very large quantities of PGI protein. The PGI was cat- alytically active and had a very slightly faster elec- trophoretic mobility on starch gels than the slow allozyme PGI-3B of C. unguiculata, a difference robably caused by different post-translational pro- tein modification between Clarkia and F. coli. By the same procedure, a large quantity of protein with the molecular weight of PGI was also synthe- sized from U2, and its electrophoretic mobility was similar to that of Clarkia plastid PGI. The expres- sion of these genomic clones in E. coli, apparently by virtue of fortuitous promoters in their 5' non- coding region, provides convincing evidence that introns are not present in these genes. Whether other PGI genes also lack introns remains to be determined. Their absence is surprising, since other genes encoding glycolytic enzymes in plants such as maize Tpi has eight introns (Marchionni & Gil- bert, 1986) and maize 4dh has nine introns (Dennis 1984). To summarize the molecular studies, we have cloned and sequenced two PGI genes from a geno- mic library of C. unguiculata, a species with the PGI duplication. The genes have a homology of 58%; one of them (U8) appears to encode a cy- tosolic PGI-3 isozyme; the other is thought to en- code a plastid PCI. We have also constructed geno- mic libraries from Clarkia species without the PGI duplication and have obtained clones of a number of sequences homologous to the PGI probes from et al., C. unguiculata. The molecular genetics studies of PGI in Clarkia constitute one of the first analyses of the evolution of a plant nuclear gene. Many additional molecular studies are called for to un- derstand gene evolution and to improve phyloge- netic reconstructions. LiTERATURE CITED ACHARI, A., S. E. MARSHALL, H. MUIRHEAD, R. H. PAL- MIERI & E. A. NOLTMANN. 1981. Glucose-6- Vis Pen isomerase. Philos. Trans., Ser. B 293: Burana, C. H. 1962. Discussions in Cytogenetics. Burgess Publ. Co., Minneapolis, Minnesota. pit: E, L. . Systematic botany: an unending nthesis. Taxon 13: 257-273. (Ci FORD, D. 1983. Phylogenetic and systematic in- ferences from electrophoretic studies. Pp. 257-287 Volume 75, Number 4 1988 ottlieb 1179 Molecular Genetics in Clarkia in S. O. Tanksley & T. J. Orton (editors), Isozymes in Plant Genetics and Breeding, Part A. Elsevier, Amsterdam. . 1985. — BP data and plant specia- tion. Syst. Bot. 10: -416. Davis, P. H. & V. H. nter 1967. Principles of Angiosperm Taxonomy. Van Nostrand, Princeton, w Jersey. Davis, W.S. 1970. The systematics of Clarkia bottae, cylindrica, and a related new species, C. rostrata Brittonia = 210-2 DENNIS, E. S., W. L. GERLACH, de ja PRYOR, J. L. T ZEN WELLYN, M. M. Sacus, R. J. 1984. Molecular. dris of the alcohol t (Adh1) gene of maize. Nucl. Acids Res. 1 . Fiscugn, R. L. & R. B. GOLDBERG. 1982. Structure and flanking regions of soybean seed protein genes. Cell 29: 651-660. GOTTLIEB, L. D. 73. Enzyme differentiation and ph apar in Clarkia franciscana, C. rubicunda, and na. Evolution 27: 205-214 — "1974. Genetic confirmation of the origin of Clarkia a e Evolution 28: 244-250. ———. 197 e duplication and fixed hetero- zygosity for esr dehydrogenase in the diploid plant Clarkia franciscana. Proc. Natl. Acad. U.S.A. 71: 1816-1818. 977. Evidence for duplication and diver- gence of the structural gene for phosphoglucose isom- erase in diploid species of Clarkia. Genetics 86: 289- 307. 81. Electrophoretic e y and plant pop- ulations. Progress in Phytochem. 7: 1-46. 1982. Conservation and Ma of iso- zymes in Eu Science 216: 373-380. E sozyme number and phylogeny. Pp. 209-22 in U. Jensen D. E. Fairbrothers (edi- tors), o and Nucleic Acids in Plant System- atics. Springer-Verlag, Berlin, Heidelberg. 986. Genetic differentiation, speciation and phylogeny in Clarkia Win pem ). Pp. 145-160 n K. Iwatsuki, P. H. Raven & W. J. Bock (editors), Modern Sog of deut Univ. of Tokyo Press, Tokyo . C. Hiccins. 1984a. Evidence from sub- unit molecular weight suggests hybridization was the source of the phosphoglucose isomerase gene dupli- cation in Clarkia. Theor. Appl. Genet. 68: 369- 373. & ao . 1984b. Phosphoglucose i isomerase expression in species of Clarkia with and without a duplication of the coding gene. Genetics 107: 131- 140. & N. F. WEEDEN. phylogeny in Clarkia. HENIKOFF, S. 1984. Unidirectional dan on Men exo- nuclease III creates targeted breakpoints for DNA sequencing. Gene 28: 351-359. HicciNs, R. C. & L. D. GOTTLIEB. 1984. Subunit hy- bridization and immunological studies of duplicated phosphoglucose isomerase isozymes. Biochem. Ge- net. 22: 957-979, JANSEN, R. K. & J. D. PALMER. 1979. Gene eeu and Evolution 33: 1987. A chloroplast DNA inversion marks an ancient evolutionary split in the rs family (Asteraceae). Proc. Natl. Acad. : 5818-5822. ., E. PICHERSKY & L. D. GOTTLIEB. 1986. e mechanism of evolution in the genus Clarkia. Evolution 7: 1-20. —— 1962. Catastrophic selection as a factor in speciation. Evolution 16: 257-271. 1973. The origin of diploid neospecies in Clarkia. Amer. Toe 107: 161-170. 0 mode of evolution in Clarkia (On- PAS Paves “NEA at ICSEEB II, Vancou- ver, British Colum & M. E. peg 1955. The oe Clarkia. Univ. Calif. Publ. Bot. 20: 241-39 & RAVEN. 1958. Ranid evolution in arkia. Exolutids 12: 319-336 Marci HIONNI, W. GILBERT. 1986 The triose- sphate isomerase alze: introns an- o the plant-animal aM Cell 46: 133- 141. Messine, J. 1983. New M13 vectors for cloning. Meth. Enzymol. 101: 20-78. ODRZYKOSKI, I. & L. D. GOTTLIEB. 1984. Duplications of genes coding 6- phosphogluconate deh ydrogenase in Clarkia (Onagraceae) and their phylogenetic im- plications. Syst. Bot. 9: 479-489. PiCHERSKY, E. & L. D. Gorriies, 1983. Evidence for duplication of the structural genes coding plastid and SCHWARTZ, D. & T. ENDO. 1966. Alcohol dehyllrge nase polymorphism in maize: simple and compound loci. Genetics 53: 709-715. Sun, M.-C., G. Lazar & H. M. GoopMan. 1986. Evi- dence in favor of the Hus ig 2 bic anan primary structure and e of t Eme 3-phosphate T. cal Call 47 ne P. , D. E. Sorris & L. D. GOTTLIEB. 1987. Ph ue edad: ^: gene duplications in Clarkia (Onagraceae) and their “ananas implications. Evolution 41: 667-671. SyTsMa, K. J. & L. D. GOTTLIEB. 1986a. Chloroplast A evolution and phylogenetic relationships in Clarkia sect. Peripetasma (Onagraceae). Evolution 40: 1248-1261 & 1986b. Chloroplast DNA evidence for the origin of the genus Heterogaura from a ey of Clarkia (Onagraceae). Proc. Natl. Acad. . 83: 5554-5557. s. se 1987. Organization of the nuclear mato and related iploid species. Amer. Naturalist 130 suppl.: b WEEDEN, N. F. € L. D. Gorra. "ur 1 979. Distinguishing by electrophoretic comparison ‘of pollen and somatic tissues. Biochem. Genet. 17: 287-296. CHLOROPLAST DNA VARIATION AND PLANT PHYLOGENY' Jeffrey D. Palmer,? Robert K. Jansen,’ Helen J. Michaels? Mark W. Chase?* and James R. Manhart?* ABSTRACT Several features, foremost its conservative mode of evolution, make chloroplast DNA an extremely valuable into the origin of hybrid and polyploid complexes, as illustr chidaceae, restriclion site mapping can also be used to ships. The greater expense of DN. at those higher taxonomic levels—above the family le e. restriction site mappi families of angiosperms, such as the Asteraceae and Or determine intergeneric relations s conservatism is also its Co omparative pein: site mapping is currently the bae only serious drawback, as this can limit the amount is level, one typically encounters less than 5 es unique nus sperms, the rbcL gene appears to be the chloroplast gene of choice for phylogenetic studies. Twenty-five rbcL sequences have already been accumulated, and several laboratories are ma among angiosperms and gymnosperms. The mise for resolving the deepest branchings of plant evolution and, indeed, have already this gene widel d the greatest pro aking a coordinated effort to sequence e more conservatively Bos ribosomal RNA hol settled the ultimate question of chloroplast evolution, namely, its endosymbiotic origin. A third approach to extracting Wheat a ation from chloro earrangements, suc cha o , point . their powerful characters. Examples Asteraceae, rearran green algal ancestors of land plants. s to be discussed inclu great rarity and e an inversion defining the most ancient branching in the ements that mark several major divisions within the Fabaceae, and events that identify the last DNA is by analyzing the distribution of major structural versions and the loss or gain of genes and introns. Although such rearrangements are reedom from homoplasy also make them extremely The confounding diversity and plasticity of plant morphologies have prompted plant systematists to search for more reliable characters to use in phy- logenetic reconstruction. In recent years, these characters often have been chemical ones. A large ody a plant chemicals (reviewed in Harborne & Turner, 1984; nasi & Crawford, 1986, in press), yet the Minds systematic value of these compounds remains unclear, particularly at higher taxonomic levels (Ciannasi & Crawford, 1986, in press). The examination of plant protein variation of comparative data has been gathered on by enzyme electrophoresis continues to be a pow- erful tool for inferring relationships within species and for studying mechanisms of speciation (Gott- lieb, 1981; Crawford, 1983; Harborne & Turner, 1984; Giannasi & Crawford, 1986, in press). How- ever, early attempts to infer broader patterns of plant phylogeny by amino acid sequencing were largely unsuccessful (Boulter, 1980; Harborne & 1984; Giannasi & Crawford, 1986, in press). While a recent review concluded that plant Turner, "amino acid sequencing is an approach whose time has come and gone" (Giannasi & Crawford, 1986), recent studies by Martin and collaborators (e.g., Martin & Dowd, 1986; Martin et al., 1986) dem- onstrate that thia approach does indeed have phy- logenetic merit. ank D. Crawford, C. DePamphilis, J. Doyle, critic a ending, of the manuscript and J. sequence dat (DCB 8451660 and BSR-8717600) and the USDA (85-CRCR-1-1778) to J.D E. Knox, R. Olmstead, K. Sytsma, and E. Zimmer for Doebley, W. had and T. Dyer for providing unpublished rbcL . The chloroplast DNA research of this laboratory has been diii ia by grants from the NSF ., and from the NSF to J.D.P. and R.K.J. (BSR- 8415934). H.J. M. (BSR-8700195) , M.W.C. (BSR-8600179) , per R.M. (BSR-8600167). * Department of Biol ogy, University of Michigan, Ann Arbor, Michigan 48109, a of Ecology and Evolutionary Biology, University of TAS Storrs, Connecticut 06268, * Present address: Department of Biology, University of North Carolina, Chapel Hill, North Carolina 27599, U.S 5 Present address: Department of Biology, Texas A&M University, College Station, Texas 77843, U.S.A. ANN. Missouni Bor. Garp. 75: 1180-1206. 1988. Volume 75, Number 4 1988 Palmer et al. 1181 Chloroplast DNA Variation The tremendous power of modern molecular biology has revolutionized our ability to analyze and compare large stretches of chromosomes an DNA sequences. Literally millions of discrete DNA characters, i.e., base pairs, are present in the ge- nomes of all organisms. Moreover, the underlying basis of changes in the states of these characters can be understood in the most satisfying way pos- sible—as mutations in the genetic material itself. The application of DNA comparisons to systematic questions is now being actively pursued for most life forms. The power and promise of DNA sys- tematics has already been demonstrated in studies whose emphases range from the primary lines of descent of life on Earth (Pace et al., 1986) to population dynamics and the speciation process in higher animals (Avise et al., 1987; Moritz et al., 1987). Although the field of chloroplast DNA (cpDNA) systematics is only 12 years old (Atchison et al., 1976; Vedel et al., 1976), it has already provided significant insights into a number of phylogenetic problems, and, together with parallel studies of nuclear DNA variation, offers the best hope of producing an accurate phylogeny of the major lines of plant descent. Several reviews of the field have appeared recently (Cattolico, 1985; Palmer, 1985a, 1986a, 1987; Giannasi & Crawford, 1986, in press). This article will review approaches for ex- tracting useful phylogenetic information from chlo- roplast DNA comparisons as practiced in this lab- oratory and also results obtained from such studies. THE CHLOROPLAST CHROMOSOME An understanding of certain basic properties of the cpDNA molecule is critical to a proper appre- ciation of its utility in systematic studies. The fol- lowing overview is based primarily on several re- view articles (Sears, 1980; Gillham et al., 1985; Palmer, 1985b, c, 1987; Rochaix, 1985; Ritland & Clegg, 1987; Zurawski & Clegg, 1987; Wolfe et al., 1987), which should be consulted for details beyond this brief synopsis and for relevant refer- ences. In addition, the recently completed se- quences of the entire chloroplast genomes of an angiosperm (/Vicotiana tabacum; Shinozaki et al., 1986) and a liverwort (Marchantia polymorpha; Ohyama et al., 1986) provide a wealth of specific information on chloroplast gene content and or- ganization. STRUCTURE AND GENE CONTENT Figure 1 shows the structure and gene arrange- ment of a representative and well-understood chlo- roplast genome, that of tobacco. This circular chro- mosome is 156 kilobase (kb) pairs in size (155,844 base pairs, to be exact) and is arranged in four parts; there are two identical 25-kb segments that form an inverted repeat separating the rest of the molecule into single-copy regions of 87 kb and 18 kb. The typical chloroplast genome of land plants is densely packed with approximately 120 genes, some of which are shown in Figure 1. These genes encode four ribosomal RNAs, 30-31 transfer RNAs, approximately 55 proteins of known func- tion, and about 30 unidentified proteins. Chloro- plast gene products function primarily in photo- synthesis and in transcription-translation. The former category includes many of the critical poly- peptides of the major thylakoid protein complexes (see legend to Fig. 1) and also the large subunit of the primary CO, fixing enzyme, ribulose-1,5-bis- phosphate carboxylase. Gene products involved in transcription-translation include the ribosomal and transfer RNAs, approximately a third of the chlo- roplast ribosomal proteins, four subunits of RNA polymerase, and initiation factor 1. Surprisingly, tobacco and Marchantia cpDNAs contain six genes similar in sequence to those encoding subunits of mitochondrial NADH dehydrogenase, although their function in the chloroplast remains a mystery. Both strands of the chloroplast genome are ac- tively expressed, indeed transcription. switches strands over 30 times. Many chloroplast genes are grouped functionally i , Suc as those containing the four: ribosomal RN A genes, atpI-H-F-A, atpB-E, and a cluster of eight ribo- somal proteins extending from rp/23 through rps8 (Fig. 1). The order and mode of expression of genes in these operons are highly similar to those found in prokaryotes. A major structural difference be- tween certain chloroplast genes and those of pro- karyotes is the presence of introns, which in to- bacco occur in six tRNA genes and ten protein genes, but which are absent from eubacterial genes. EVOLUTION AND INHERITANCE Land plant cpDNAs evolve quite slowly in all respects. They vary in size less than two-fold (120— 217 kb), with most of this variation resulting from a few major expansions or contractions of the large inverted repeat, so that the range of sequence complexities is only 110-150 kb. Most length mu- tations are quite short (1-10 bp), although larger ones on the order of 50—1,200 bp occur frequently enough to be a major component of the variation encountered in restriction fragment and map com- parisons (see below). Gene content is highly con- Annals of the Missouri Botanical Garden 1182 FIGURE 1. monocot (Oncidium excavatum; Orchidaceae). Heavy black lines centered on the circles indicate large inverted repeats. Ribosomal RNA genes are indicated by 16S and 23S. Genes for the 50S and 30S ribosomal proteins Physical and gene maps of typical cpDNAs from a dicot (Nicotiana tabacum; Solanaceae) and are given as rpl and rps, respectively, followed by the number of the corresponding E. coli protein. Genes for RNA polymerase subunits are indicated by rpo, followed by a subunit-specific letter. InfA encodes initiation factor 1. RbcL encodes the large subunit of ribulose-1,5-bisphosphate carboxylase. Subunits of the thylakoid membrane complexes ATP synthase, photosystem I, photosystem Il, and cytochrome b,/f complex are encode by atp, psa, psb, an the completely sequenced cpDNA of tobacco; see Shinozaki et al. are from M. Chase & J. Palmer (unpublished data). served; only two differences are known among the 120 genes present in the cpDNAs of tobacco and Marchantia, which diverged some 400 million years ago. The order of chloroplast genes is also highly conserved. The tobacco gene order is found in most other angiosperms (e.g., an orchid, Fig. 1) and in at least one fern and one gymnosperm, and differs by only one inversion from the Marchantia order. With the exceptions so far of only two groups of plants (some legumes and Pelargonium), most of the changes in gene order found among angiosperm cpDNAs can be accounted for by one or two simple inversions. Sequence comparisons of several kinds reveal a low rate of nucleotide substitution in land plant cpDNA as a whole, although direct sequence stud- ies reveal rate differences among specific. chloro- plast genes. On average, the rate of silent substi- tution in chloroplast genes is two to three times lower than in nuclear genes and 20 times lower than in animal mitochondrial genes, but three to four times higher than in plant mitochondrial genes. Transitions outnumber transversions in chloroplast genes by a factor of somewhat less than two relative to random expectations, a much smaller bias than that found in animal mtDNA. A critical feature of cpDNA from the standpoint of phylogenetic studies at lower taxonomic levels is its mode of inheritance. In all land plants ex- d pet, respectively, followed by a subunit-specific letter. Only selected genes are shown for (1986) for a more complete map. Orchid data amined thus far, it is inherited clonally, through the maternal parent in most angiosperms and the paternal parent in gymnosperms. In those plants where both parents contribute chloroplasts to their offspring, the chloroplasts and their genomes have never been seen to recombine, but simply sort out somatically. APPROACHES TO DETECTING AND ANALYZING CHLOROPLAST DNA VARIATION Mutations in cpDNA are fundamentally of two types— point mutations (single nucleotide pair sub- stitutions) and rearrangements, with several kinds of rearrangements recognized. By far the most frequent mutations are point mutations and dele- tions/insertions in noncoding regions. Whereas point mutations can profitably be used for phylo- genetic studies at all taxonomic levels, the system- atic use of noncoding deletions/insertions is often inappropriate, as will be discussed in the section on restriction site mapping. Other classes of rear- rangements (inversions and deletions/insertions of genes, introns, and one copy of the large inverted repeat) occur rarely during cpDNA evolution. Be- cause of their rarity, these changes are often ex- tremely useful in phylogenetic reconstruction. In the following sections, we discuss the kinds of approaches available for comparing cpDNAs and Volume 75, Number 4 1988 Palmer et a : 1183 Chloroplast DNA Variation revealing phylogenetically informative mutations. e three major approaches will be discussed in terms of the methodologies involved, the kinds of mutations detected, and the taxonomic levels at which they are most appropriate. Before discussing these approaches, we first review methods for pre- paring cpDNA and total DNA for use in phylo- genetic studies. ISOLATION OF CHLOROPLAST AND TOTAL DNA The most generally applicable procedure for pre- paring purified cpDNA is the sucrose gradient pro- cedure described in Palmer (1986b). For some plants, such as members of the Geraniaceae, sub- stantially higher yields are obtained by substituting for the aqueous blending used in Palmer (1986b) a two-step homogenization procedure, consisting of powdering of liquid nitrogen-frozen tissue in a cof- fee mill (Calie & Hughes, 1987a), followed by Polytron-grinding of the powder in aqueous buffer. A large number of alternative procedures for pre- paring cpDNA are reviewed in Palmer (1986b). We particularly call the reader's attention to a NaCl-isolation technique (Bookjans et al., 1984) that works extremely well with many recalcitrant legumes and Malpighiaceae, to recently developed techniques for extracting cpDNA from conifers (White, 1986) 19872), and to a promising modification (Dally & Second, submitted) of the nonaqueous procedure of Bowman & Dyer (1982), which eliminates the use of highly dangerous organic chemicals. n many cases it may be convenient, if not preferable, to analyze cpDNA variation in prepa- rations of total cellular DNA, rather than purified cpDNA. The major advantages of using total DNA are yield (total DNA extraction efficiencies are and mosses (Calie & Hughes, nearly quantitative and provide 5-100 times higher yields of cpDNA than cpDNA extraction proce- dures; 1-2 grams of fresh leaf tissue will usually provide sufficient total DNA for 10-30 restriction enzyme digests, enough for most comparisons of cpDNA variation), flexibility (total DNA prepa- rations can obviously be used to study variation in all three plant genomes), and adaptability (total DNA has been extracted from several groups of plants in which common methods of cpDNA ex- traction did not work). The small amount of starting material needed for total DNA extractions has proved critical in our studies on orchids, many of which are rare or minute plants that at maturity may weigh only 0.3-0.6 grams. Since cpDNA is present in several thousand copies in the typical leaf cell (Bendich, 1987), it is easy to visualize and map cpDNA fragments by hybridizing cloned cpDNA fragments to filter blots containing restriction enzyme digests of total DNA. Indeed, most current studies of cpDNA restriction site variation have employed total DNA (e.g., Syts- ma & Gottlieb, 1986a, b; Coates & Cullis, 1987; Jansen & Palmer, 1988). The use of purified cpDNA is of somewhat greater advantage in DNA sequencing and rearrangement studies, as de- scribed in the relevant sections below. An extremely effective procedure for preparing total cell DNA is Doyle & Doyle's (1987) modi- fication of the CTAB isolation method of Saghai- Maroof et al. (1984). This method typically yields 20-100 micrograms of high molecular weight DNA from 0.5-2 grams of fresh (or fresh frozen) leaves. We find that higher yields of total cellular DNA are often obtained from species, particularly or- chids, with succulent tissues by increasing the CTAB buffer from 2x (Doyle & Doyle, 1987) to 3x concentration. Also, fibrous leaves are often ho- mogenized more efficiently by grinding in liquid nitrogen (either with mortar and pestle or in a coffee mill), followed by addition of CTAB buffer, rather than directly in the buffer. The CTAB method sometimes gives an enrichment of cpDNA (relative to nuclear) if the alcohol-precipitated DNA is spun down rather than spooled out of the aqueous so- lution. The CTAB method has also proven suc- cessful in the extraction of usable DNA from re- cently dried herbarium specimens (Doyle & Dickson, 1987). A variation on this CTAB pro- cedure was reported by Rogers & Bendich (1985) to give good results with fresh tissues from a wide variety of plants, but to yield only degraded DNA from older herbarium specimens and mummified tissues. RESTRICTION SITE MAPPING The cpDNAs of closely related species (from the same or related genera) are most easily compared by examining the pattern of fragments produced upon digestion of the DNAs with restriction en- donucleases and subsequent electrophoresis in aga- rose gels. The restriction fragment phenotypes that are diagnostic for three types of mutations— base substitutions (point mutations), deletions/insertions (length mutations), and inversions—have been de- scribed in earlier papers (Upholt, 1977; Palmer et al., 1985; Palmer, 1986a). Point mutations (de- tected as the gain or loss of restriction enzyme cleavage sites) and length mutations (detected as fragment size differences) are common, whereas inversions are rare. The length mutations detected Annals of the Missouri Botanical Garden 1184 SE EHHHHFEFF 4.0- C. 9899.9 3.0- * y $ . 2.5- FicURE2. | Hypervariation in cpDNA resulting from "m. mutations. DNAs from 1-4 populations of each of four species in the genus Pisum (P. sativum, lane 5; P. elatius, lanes E; P. humile, lanes H; P. fulvum, lanes F) were digested with the restriction enzymes Xhol, gel to a nitrocellulose filter and hybridized with a *P- labeled Xhol fragment of 3.2 kb from pea cpDNA. Sizes indicated at left are in kb. Modified from Palmer et al. 985). in restriction enzyme studies usually occur in spacers between genes and pose problems for phylogenetic analysis because they tend to cluster in "hotspot! regions having high levels of variability and because the assignment of exact homology is often difficult (e.g., Fig. 2). In combination, these two factors can result in high degrees of homoplasy (i.e., par- allelism and convergence). Therefore, spacer length mutations are usually not included in a formal phylogenetic analysis (Perl-Treves & Galun, 1985; 1985; Sytsma & Gottlieb, 1986a), although it is critical that they be properly rec- ognized and distinguished from restriction site mu- tations. In early studies cpDNA restriction site mutations were inferred indirectly by the inspection of re- striction fragment profiles of total, purified cpDNA (e.g., Palmer & Zamir, 1982; Erickson et al., 1983; Clegg et al., 1984). This approach is limited to situations in which 1) cpDNA can be readily prepared (requiring both large amounts of tissue, 10-100 g fresh weight, and amenability of the tissue to cpDNA purification techniques) and 2) Palmer et al., levels of base sequence divergence are low enough (less than 0.5- 1.046) to permit the critical inter- pretation of fragment pattern differences in terms of specific mutations. Currently, most researchers in the field analyze cpDNA variation by a filter hybridization approach, in which cloned fragments of one chloroplast genome are hybridized to filter blots containing digests of all the DNAs under study (e.g., Sytsma & Gottlieb, 1986a, b; Coates & Cul- lis, 1987; Jansen & Palmer, 1988). The hybrid- ization approach enables the direct ordering, or mapping, of restriction enzyme cleavage sites and permits a more critical analysis of mutations and discrimination between site mutations and length mutations. In addition, it allows one to use total DNA, which offers several advantages over the use of purified cpDNA (see previous section). The choice of cpDNA clones to use as hybrid- ization probes in mapping studies depends on sev- eral factors (reviewed in Palmer, 1986b). “Ho- mologous"' clones, i.e., those from one of the taxa under study, give the best results but can be a lot of work to make if not already available. The use of “heterologous” clones requires sufficient con- servation of 1) q level of cross-hybridization and 2) linear arrange- ment of the chromosome to permit alignment of the cross-hybridizing fragments and interpretation of the hybridization signals. Fortunately, the great majority of cpDNAs are highly conserved in se- quence and arrangement (Palmer, 1985b, c; Zu- rawski & Clegg, 1987), and heterologous probes have been shown to work effectively across sub- classes of angiosperms (e.g., Perl-Treves & Galun, 1985; Sytsma & Gottlieb, 1986a). The relevant properties of 15 angiosperm cpDNA clone banks were tabulated in Palmer (1986b). Additional clone banks have been developed for a number of other angiosperms (tobacco, Sugiura et al., 1986; to- mato, Phillips, 1985; potato, Heinhorst et al., 1988; soybean, Singh et al., 1984; lettuce, Jansen & Palmer, 19872; a poplar, Populus nigra, R. Smith & K. Sytsma, unpublished data; sorghum, Dang & Pring, 1986; an orchid, Oncidium excavatum, M. Chase & J. Palmer, unpublished data), two ferns (Stein et al., 1986; D. Stein, unpublished data), a conifer (C.-H. Tsai & S. Strauss, unpublished data), and the liverwort Marchantia polymorpha (Ohya- 1986). Most of these clone banks are freely available for use. An outline of the approach we currently use for comparative mapping of cpDNA restriction site mutations is as follows: 1) Total DNAs are prepared from 1-2-g leaf samples by the procedure of Doyle & Doyle (1987) or by one of the modifications of this procedure described in the preceding section. 2) The DNAs are digested with each of 10-20 different restriction enzymes that cut the genome 20-100 times. 3) The digests are separated elec- trophoretically in 1.0% agarose gels in which the bromophenol blue dye marker is run 10 cm. (Most to ensure a significant ma et al., of the details of the electrophoresis and hybridiza- tion conditions are given in Palmer, 1986b; plans for the gel rigs we use are available upon request of the senior author.) The use of narrow tooth gel combs permits 35 samples (e.g., 33 plant DNAs Volume 75, Number 4 1988 Palmer et al. 1185 Chloroplast DNA Variation and two lanes of size markers) to be run across a 20 cm width corresponding to the width of standard 8" x 10" X-ray film. The DNAs are arranged on the gel (or gels) according to presumed relatedness. By running the gels no farther than 10 cm, we accommodate two gel-sized filters (each 20 cm x 12.5 cm), or 66 experimental lanes, on each X-ray film exposure. 4) Two replica filters of each gel are made using the bidirectional blotting procedure of Smith & Summers (1980) as modified for greater sensitivity by the alkaline transfer procedure (Reed nn, 1985). We use durable nylon filters (Zetabind; manufactured by AMF Cuno), which in our experience can be readily reused 15 or more times. 5) The two identical sets of 8-12 filters (one filter per enzyme in the case of a 33-DNA study) are hybridized in plastic buckets using the BLOTTO (nonfat dry milk) hybridization buffer of Johnson et al. (1984). Hybridization probes consist of cloned fragments of an appropriate chloroplast genome (see above). Using two probes at a time, and taking a week for each cycle of hybridization and auto- radiographic exposure, we require about two to three months to perform the entire set of hybrid- izations once all the gels have been run and blotted (each hybridization cycle requires only about two person-days of work per week). The most time-consuming and difficult part of such studies, particularly those in which levels of divergence are on average greater than 1%, can be the interpretation in terms of individual muta- tions of the fragment pattern differences revealed by the autoradiographs. We recommend the con- struction of complete restriction site maps for each enzyme for one of the taxa under study. This comprehensive mapping is aided by including on each enzyme gel a double digest of the reference DNA with the enzyme specific to that gel and an enzyme used in common in all the double digests. The double digest hybridizations allow one to place the single enzyme maps in register with one another and thereby to construct a single unified restriction map. Mutations relative to this basic map can then be recognized by comparing the patterns across the autoradiographs and then identifying and lo- calizing on the map fragment changes consistent with the loss/gain of specific cleavage sites (i.e., the presence in one taxon of two adjacent frag- ments which are replaced in a second taxon by a single fragment equal in size to the two missing ones). Restriction site mutations are studied by grouping the X-ray films by enzyme and “walking” along the chloroplast chromosome from cloned hy- bridization probe to adjacent probe fragment. Re- sorting the films according to probe fragment fa- cilitates detection and confirmation of length mutations, as well as the resolution of most am- biguities noted in the first level of analysis. The phylogenetic analysis of restriction site mu- tations is easily accomplished within the framework of parsimony-based cladistic analysis. Each site can be treated as a single two-state character, whose polarity can be assessed by outgroup comparisons (Watrous & Wheeler, 1981). Algorithms based on both Wagner parsimony (PAUP, developed by D. Swofford) and Dollo parsimony (PHYLIP, devel- oped by J. Felsenstein) can be employed to analyze cpDNA data. Wagner parsimony gives equal weight to parallel gains and losses, whereas Dollo parsi- mony prohibits the former class of events. The justification for the use of Dollo parsimony is that parallel site losses are in fact much more probable than parallel site gains (DeBry & Slade, 1985). The bootstrap algorithm of Felsenstein (1985) is useful for evaluating the statistical significance of the monophyletic groups defined by parsimony analysis. The application of parsimony and boot- strap programs to large sets of cpDNA site mu- tations is best exemplified by the studies of Sytsma & Gottlieb (1986a) and Jansen & Palmer (1988). DNA SEQUENCING DNA sequence analysis allows one to compare bases individually and results in much lower levels of homoplasy than site mapping, where changes at any of six positions can cause a site loss. By se- quencing a gene, one effectively avoids the problem of length mutations and also gains resolution at greater phylogenetic distance, since many genes are more conserved than the genome as a whole. The trade-off relative to restriction site mapping is that sequencing is currently a significantly slower and more expensive way of gathering phylogenetic information. Therefore, sequencing is presently most appropriate at the family level and above, where restriction site mapping is plagued by excessive homoplasy and length mutation. No major studies on chloroplast gene sequences and plant phylogeny have been published. How- ever, one chloroplast gene (rbcL, encoding the large, catalytic subunit of ribulose-1,5-bisphos- phate carboxylase) has, because of its fundamental importance in photosynthesis, been sequenced widely enough to demonstrate the phylogenetic util- ity of this approach. Members of each of the four angiosperm families for which multiple rbcL genes have been sequenced cluster as natural groups, while representatives of three subclasses of dicots group together at an appreciable molecular dis- 1186 Annals of the Missouri Botanical Garden 30% 25% 20% 15% 10% 5% 0% Lactuca Senecio Asteraceae Helianthus Valeriana —— Valerianaceae N. tabacum N. acuminata N. otophora Solanaceae Lycopersicon Spinacia — Chenopodiaceae Pisum | Fabaceae Lm to ° 3 a = = 3 T. aestivum Poaceae Aegilops T. urartu Hordeum Marchantia C. reinhardtii — RR C. moewusii Chlorella Anacystis Anabaena FIGURE 3. Phenogram of rbcL sequence relationships. The o is based solely on the percent nucleotide a ce divergence (top scale) eem rbcL genes of the indicated taxa. No corrections have been applied m multiple substitutions at a posi As yet, no phenetic or cladistic nisoria have been applied to this data set. References for eight of m sequent es are given in Ritland & Clegg (1987); the bus are Nicotiana acuminata and N. otophora (Lin et al., 1986), Lycopersicon (W. Gruissem, pers. comm.) ; Petunia and Medicago ( Aldrich et al., 1986a, b) ; Lactuca, end Helianthus, and Valeriana ES eco R. Jansen & J. Palmer, unpublished data) ; Oryza (Nishizawa & dr 1987) ; Sorghum (J. Doebley pers. comm.) ; Aegilops (Terachi et al., 1987) ; Triticum aestivum and T. urartu (T. Dyer, unpublished data) ; Marchant (Diana et al., 1986) ; Chlamydomonas moewusii (Yang et al., 1986) ; and Chlorella (Yoshinaga et al., 1988). tance from the grasses (Fig. 3). A liverwort, three of plant molecular systematists have started to or green algae, and two cyanobacteria appear as suc- are planning on using rbcL (we are sequencing cessive outliers to the angiosperms. With certain rbcL in the Asteridae and other subclasses of dicots; minor exceptions, discussed below in "Familial Re- M. Clegg, D. Giannasi, D. Soltis, and P. Soltis are lationships,” all of the relationships expressed in conducting a broad survey of rbcL in several sub- Figure 3 are consistent with widely accepted con- classes of angiosperms; M. Chase in Rosidae and cepts of plant phylogeny. Such correspondence is Liliidae; K. Sytsma in Myrtales; R. Jansen in As- encouraging relative to the problematic early his- — teraceae; G. Furnier in gymnosperms). In a few tory of protein sequencing and plant phylogeny years, accumulated rbcL data should permit the (see introductory paragraphs). construction of a broad framework of phylogenetic or a variety of reasons, we feel that rbcL is relationships among angiosperms and gymno- the chloroplast gene of choice for examining phy- sperms. 3) The rbcL gene is large enough (1,431 logenetic relationships within vascular plants: 1) bp; only eight of the 120 chloroplast genes are The rbcL gene is by far the most widely sequenced significantly larger) to provide a sufficient number chloroplast gene. This provides not only a reservoir of characters (ie., base pairs) for phylogenetic of data that can be compared with data from future studies. 4) Its rate of evolution appears appropriate studies but also confidence that the gene is a reliable for questions of angiosperm phylogeny, whereas one for phylogenetic purposes (Fig. 3). 2) A number other obvious candidates such as the 165 and 235 Volume 75, Number 4 1988 Palmer et al 1187 Chloroplast DNA Variation ribosomal RNA genes change too slowly (the 165 gene is only 4% divergent between monocots and dicots, compared with 3-6% rbcL divergence with- in each of three angiosperm families in Fig. 3). Our current rbcL sequence studies rely on three technical factors to enable collection of large amounts of sequence data in a rapid and efficient manner. First, rather than screen large shotgun clone banks of total cellular DNA or total cpDNA for rbcL-containing clones, we isolate directly a rbcL-containing fragment (first identifying it by blot hybridization) from a cpDNA digest separated in a low-melting-temperature agarose gel and then ligate the fragment directly in the melted agarose to a suitable cloning vector. In the case of most taxa examined in the Asteridae and Rosidae, diges- tion with EcoRI, BamHI, SacI, Xhol, or various double-digest combinations of these enzymes pro- — duces a conveniently sized fragment (1.5-5 kb containing the entire rbcL gene. The use of purified cpDNA considerably simplifies this cloning step as compared with total DNA. The second savings re- sults from our use of the plasmid-phage cloning vector BLUESCRIPT M-13 (Strategene Inc.). This allows us to clone the gel-isolated rbcL-containing fragment in a double-stranded form and then obtain single-stranded DNA for dideoxy chain termination sequencing without the need for recloning into a single-stranded phage vector. Third, we use syn- thetic oligonucleotides (20-25 bases in length) complementary to conserved regions of rbcL as primers in dideoxy sequencing reactions. These primers were synthesized by C. Zurawski, who generously has made them widely available to mo- lecular systematists. By using synthetic primers of this nature, we avoid perhaps the most time-con- suming aspect of sequencing, i.e., preparing mul- tiple subclones from each primary clone. The technology for DNA sequencing is rapidly advancing at present. Significant future savings in time should derive from the use of 1) double-strand- ed dideoxy sequencing protocols (Korneluk et al., 1985; Zhang et al., 1988); 2) optimized sequencing reactions permitting sequence reading up to 1 kb from a priming site (Johnston-Dow et al., 1987); 3) the polymerase chain reaction to amplify a DNA fragment many thousands of times in vitro, which either facilitates the cloning step or obviates it altogether (Erlich et al., 1988; Oste, 1988); and 4) automated DNA sequencing machines (Connell et al., 1987; Prober et al., 1987). Once obtained, DNA sequence data are aligned and compared using standard DNA sequence anal- ysis software programs, a large number of which are now commercially available. Cladistic analysis of the sequence data (each nucleotide position is treated as a single four-state character) can be performed using the same parsimony programs and statistical tests as described for restriction site data. In addition, maximum likelihood programs have been developed specifically for nucleotide sequence data (Felsenstein, 1981, 1983), and Nei (1987) reviewed a number of tree-building algorithms that use sequence data. REARRANGEMENT ANALYSIS A second class of DNA mutations useful for phylogenetic inference is major rearrangements (the first class being the base substitutions detected by sequencing and site mapping). Major rearrange- ments are defined here as inversions and deletions/ insertions of introns, of the coding region of genes, and of one segment of the large inverted repeat characteristic of most chloroplast genomes. The analysis of major cpDNA rearrangements is in sev- eral ways a complementary approach to compar- ative sequencing for studying higher levels of plant phylogeny. An advantage of sequencing is that large numbers of characters (1,431 in the case of rbcL) can be surveyed with reasonably predictable expectations as to the number of differences that will be found (Fig. 3). Many fewer DNA rearrange- ments can be expected and a preliminary survey is needed to find them. However, once found, they are often easy to survey widely (Fig. 4). The ex- treme rarity and lack of homoplasy of major rear- rangements (Fig. 5) makes each one a signal char- acter in a way that nucleotide substitutions, which inevitably will be afflicted with certain levels of homoplasy, can never be. Each rearrangement is therefore a much more powerful character, one that in our opinion should be weighted much more heavily than a single nucleotide substitution or re- striction site mutation. Large amounts of sequence data may not suffice to resolve relatively ancient and compressed evolutionary radiations, whereas each rearrangement generally resolves with con- fidence a particular branching point in a phylogeny. The existence of inversions and deletions of the cpDNA inverted repeat has been known for sev- eral years (Palmer, 1985c; Palmer et al., 1987). However, it is only with the recent acquisition of the complete sequences of the chloroplast genomes of the angiosperm Nicotiana tabacum (Shinozaki et al., 1986) and the liverwort Marchantia poly- morpha (Ohyama et al., 86) that it became possible to uncover and exploit significant numbers of the two other major classes of rearrangements, 1188 Annals of th Missouri Nu" Garden a Chloroplast DNAs Nunber Species Family Subclass 123 456 7 8 9 1011 12 13 14 15 16 17 — —KÑ—Y-J.— —s——y 1 Zea mays Poaceae Commelinidae 2 Narcissus tazetta Amaryllidaceae Liliidae 3 Aristolochia durior Aristolochiaceae Magnoliidae 4 Delphini Ranunculaceae Magnoliidae 5 Echachicl esia californica Papaveraceae Magnoliidae 6 Pilea microphylla Urticaceae 7 Spinacia oleracea Chenopodiaceae Caryophyllidae 8 obtusifolius Polygonaceae Caryophyllidae 9 Glycine max Fabaceae Rosi 10 icago sativa Fabaceae ll Trifolium Fabaceae Rosidae 12 Pisum sati Fabaceae Rosidae 13 us californica Hippocastanaceae Rosidae 14 Pelargonium Xho: Geraniaceae 15 Brass. campestris Brassicaceae Dilleniidae 16 Nicotiana taba Solanaceae i7 Lactuca sativa Asteraceae Asteridae C rpi2 intron b rpl2 exon 12345 6 7 8 9 1011 12 13 14 15 16 17 “° . NL. o e° = Y m ," Tr ' E EET. s “< e 123 45 67 8 9 10 11 12 13 14 15 16 17 d rpoA e rpl22 123 45 67 8 9 10 t1 12 1314 15 16 17 1234 56 7 8 9 10 11 1213 14 15 16 17 - @ FIGURE 4. Detection of intron and gene losses during angiosperm cpDNA evolution.—a. Electrophoresis in a 0.9% agarose gel of cpDNA fragments d eu M digestion with EcoRI (lanes 1 and 11), Sacl-Pvull (lanes 2-6, 8, 9, 13, and 14), Sacl-Psstl (lanes 7, 12, and 15), and Hindlll (lanes 10, 16, and 17). Two Zetabind filter replicas of the gel were made by debe blotting and then hybridized jene un with the gene probes indicated in panels b-e.—b. Hybridization with a 772-bp fragment internal to and containing 90% of the coding region of the rpl2 gene from spinach.—c. Hybridization with a 545-bp fragment internal to and Volume 75, Number 4 1988 Palmer et al. 1189 Chloroplast DNA Variation namely gene and intron losses/gains. All four of these rearrangement classes are usually detected by Southern hybridization experiments using de- fined segments of chloroplast genes as probes. In some cases, follow-up DNA sequencing analysis may be called for if the results of the hybridization experiments are ambiguous. As illustrated in Figure 4, gene and intron losses/ gains are easily detected by a simple presence/ absence test. For example, a cloned fragment in- ternal to the intron of the gene rp/2 hybridizes to all angiosperm cpDNAs tested except to that of spinach (Fig. 4d). This result and similar hybrid- izations to other cpDNAs have led to the conclusion that this intron was lost in the common ancestor of the order Caryophyllales (J. Palmer & G. Zu- rawski, unpublished data). Similarly, hybridization studies reveal that the gene rpoA is absent from the chloroplast genome of Pelargonium X hor- torum (Fig. 4d) and all other species of Pelargoni- um (P. Calie & J. Palmer, unpublished data), while the rpl22 gene is absent from the chloroplast ge- nomes of all legumes (Fig. 4e; J. Palmer, B. Mil- ligan, J. Doyle, unpublished data). It should be noted that all three of these intron/gene absences have been confirmed by sequencing the region of the suspected absence in at least one of the relevant taxa (Zurawski et al., 1984; B. Milligan, P. Calie, J. Palmer, unpublished data). The “gene losses" should, in the larger biological context, be viewed s "gene transfers," since genes that are missing from the chloroplast genome appear to have been transferred to the nucleus (S. Baldauf, S. Gantt, J. Palmer, unpublished data). The detection of inversions and inverted repeat loss/gains is somewhat more complicated than for gene or intron loss/gains. Inversions can be di- agnosed in two ways. Two nearby fragments in an uninverted genome that have become separated by virtue of inversion will consistently hybridize to different fragments in the inverted genome. Con- versely, two fragments that are widely separated in a genome lacking an inversion will each hybridize to the same two fragments in a genome containing a derived inversion, indicating a new linkage re- lationship (e.g., see figs. 2 and 3 of Jansen & Palmer, 1987a). The presence or absence of the large inverted repeat shown in Figure 1 is diagnosed as follows (Palmer et al., 1987; Lavin et al., 1988). In digests produced by enzymes that cut rarely (10-20 times per genome), short, single copy frag- ments flanking a deleted repeat will normally hy- bridize to different fragments in genomes retaining the repeat, but to the same fragment in genomes lacking the repeat. Also, with most rarely cutting enzymes, short fragments near the end of the re- peat will hybridize to two fragments in genomes retaining the repeat, but to only one fragment in those lacking it. In a typical survey for rearrangements, we di- gest each cpDNA with four enzymes only (com- pared with the 10—20 used for restriction site stud- ies), each cutting on the order of 40-100 times. This provides resolution sufficient to detect small inversions without unduly increasing the size of the study. In contrast to restriction site studies, where it is most useful to arrange all the digests for a given enzyme on a single gel or set of gels, all four digests for a given DNA are placed together on the same gel. This facilitates rapid diagnosis of rearrangements. Eight DNAs can be analyzed con- veniently on a single gel with 35 lanes, including three for size markers. Double-sided blotting (Smith & Summers, 1980) allows one to make two filter replicas for each gel, which then can be probed sequentially over a period of weeks, months, and even years. For hybridization probes, we use small cloned fragments 0.2-3.0 kb in size that we have prepared from several angiosperm chloroplast genomes (Jan- sen & Palmer, 19872). These clones are available upon request of J. Palmer for use in molecular systematic studies. In addition, we currently are making probes specific for many of the genes from the completely sequenced tobacco chloroplast ge- nome (Shinozaki et al., 1986), using a starting set of large cloned fragments provided by M. Sugiura (Sugiura et al., 1986). Smaller probes (0.2-1.0 b) are most useful as internal, gene- and intron- specific markers to detect gene and intron losses. These smaller probes, plus somewhat larger ones (1-3 kb) containing spacer and tRNA genes, are useful to detect inversions because many cpDNA inversions have their endpoints in tRNA-spacer- rich regions of the genome (Palmer, 1985c; Jansen & Palmer, 19872). — containing 82% of the intron of the rpl2 gene from tobacco.—d. Hybridization with a 1,040-bp pem containing 9676 of the coding region of the rpoA gene from spinach and 78 and containing 45% Ld the coding region of the rpl22 gene Unpublished jh of J. Palmer, B. Milligan and P. Calie ybridization with a 209-bp fragment internal to from tobacco. of 5' noncoding sequence.— Annals of the Missouri Botanical Garden 1190 Barnadesiinae (36)* All other Asteraceae (576) Lobeliaceae (15) Campanulaceae (2G) A inversion (in kb) | [ Goodeniaceae (2G) . i 30S) Geranium et al. (4G, @ intron loss/gain ———4 a POA A many ASTERIDAE Pelargonium (40 (gene name) ) Caesalpinioideae (5G, 5S) @ inverted repeat loss (IR) ROSIDAE CARYOPHYLLIDAE 50 rpl2 | Mimosoideae (7G, 7S Most Papilionoideae (35G, 50S) Most Phaseoleae (34G, 42S) Most Phaseolinae (16G, 235) Milletieae (9G, 9S) Wisteria (1S) Four tribes (9G, 13S) Pinacea Caryophyllales (4F, 86)" e (26) Ferns (1G, 3S) Liverworts (1S) s (1S) Charophyceae (3F, FIGURE 5. rearrangements are shown (se Asterisks denote genomes with the consensus P ular pon gene order. | of taxa examined (F = family; G = 4G) Chlorophyceae, Ulvophyceae (6F) Evolutionary tree based on the distribution of selected cpDNA rearrangements. Not all known ^e text and Palmer, 1985c, umbers in parenthe Question marks denote rearrangements of uncertain genus nature and/or phylogenetic distribution a direction p charophytes have been examined for the tuf A gene loss, but only one for the tRNA-ile intron gair APPLICATIONS OF CHLOROPLAST DNA DaTA TO SYSTEMATIC QUESTIONS Most cpDNA comparisons have been made with- in and between congeneric species using the ap- proach of restriction site analysis. The utility of this approach at these levels has already been well documented in a number of papers and reviews (for reviews, see Palmer, 1985a, 1986a, 1987; Giannasi & Crawford, 1986, in press). Conse- quently, the following discussion will emphasize re- cent developments and results obtained in studies at higher taxonomic levels. INTRASPECIFIC RELATIONSHIPS The ultimate utility of cpDNA as a marker within and among populations of a species remains un- clear. Relatively little cpDNA differentiation was obtained in the two most extensive intraspecific studies (of 371 individuals and 147 populations of 1987) and 100 individuals and 21 populations of Lupinus texensis (Banks & Birky, 1985)). two pine species (Wagner et al., In a number of more limited studies (reviewed in Palmer, 1987; Giannasi & Crawford, in press), multiple populations from the same species often had indistinguishable cpDNAs. However, when found, intraspecific dif- ferences have been quite informative, particularly regarding the origins of several crop plants (Palmer et al., 1983; Clegg et al., 1984; Palmer et al., 1985; Doebley et al., 1987). A recent study dem- onstrated extensive intraspecific variation within Heuchera micranthra sufficient to document mul- tiple origins of autopolyploidy (Soltis et al., in press). DNA variation actually provided greater resolution at the populational level than In this case, nuclear isozyme markers. It is not surprising that intraspecific variation in cpDNA is often limited, given the overall conserva- tism of the chloroplast molecule (Palmer, 1985b, c; Zurawski & Clegg, 1987). By pushing the mo- lecular approaches to their extreme in terms of the number of nucleotides sampled, it may be possible to address many microevolutionary questions using cpDNA. By using more enzymes, especially ones that cut frequently, the sample size could readily Volume 75, Number 4 1988 Palmer et al. Chloroplast DNA Variation 1191 have been increased by a factor of ten or more in many of the studies in which little or no variation was found. It is unlikely, however, that the cpDNA molecule will ever be as useful for populational studies as the rapidly evolving mitochondrial ge- nome of animals (Avise et al., 1987; Moritz et al., 1987), although the Heuchera case (Soltis et al., in press) begins to approach the animal mitochon- drial situation. INTERSPECIFIC RELATIONSHIPS Over 40 studies have now been published in which cpDNA restriction site variation has been used to assess qp relationships at the interspecific level (many of these are reviewed in Palmer, 1986a; Giannasi ^ Crawford, 1986, in press), and at least 30 laboratories are now engaged in such pursuits. The molecular phylogenies con- structed using cpDNA data have been remarkably untroubled by homoplasy, which ranged from 0% in Pisum (Palmer et al., 1985) and Lisianthius (Sytsma & Schaal, 1985), to 2.5% in Lycoper- sicon-Solanum (Palmer & Zamir, 1982), 3.3% in Zea (Doebley et al., 1987), 3.8% in Brassica (Palmer et al., 1983), 4.8% in Clarkia sect. Peri- petasma (Sytsma & Gottlieb, 19862), and 4.9% in Cucumis (Perl-Treves & Galun, 1985). The overall level of homoplasy in these seven studies is 3.9%, i.e., only 12 convergences and parallel- isms were necessary to account for the observed distribution of 299 variant cpDNA restriction sites. To illustrate the kinds of insights cpDNA anal- ysis can provide, we briefly review results obtained in the genus Brassica. This is an excellent group to examine in this context for a number of reasons. First, relationships within the genus have already been studied using a wide variety of morphological, genetical, and biochemical approaches (reviewed in Vaughan, 1977; Prakash & Hinata, 1980), and therefore an excellent opportunity exists to eval- uate the merits of a cpDNA analysis. Second, Bras- sica features one of the classic cases of a polyploid, hybrid species complex, one that was already well understood at the nuclear level, but where addi- tional, sometimes surprising, ings emerged from the cpDNA analysis. Finally, the Brassica work exemplifies the reproducibility of most molecular analyses. Two different groups independently stud- ied cpDNA restriction site variation in Brassica (and coincidentally published their results in the same issue of the same journal) and reached largely similar conclusions (Erickson et al., 1983; Palmer 1983). Figure 6 is a cladogram based on the analysis et al., B.carinata [2] B.nigra [1] B.nigra [1] B.nigra (1) B.hirta [2] R.sativus [1] B.oleracea [41 B.junc B.campestris [3l B.campestris [11 B.napus [21 FIGURE 6. Evolutionary tree of eight cultivated Numbers at termination of branches indicate nuclear chromosome number, followed by genome designation. Numbers in parentheses indicate number Ma qum examined. Redrawn from Palmer et al. (1983). of cpDNAs from eight species and 21 populations of Brassica and Raphanus sativus (Palmer et al., 1983). Six of the eight species (those whose genome designations are given in Fig. 6) comprise the three diploids and three tetraploids whose nuclear rela- tionships were first expressed in the famous triangle diagram of U (1935). One unexpected finding is the placement of Raphanus sativus (radish) within one of the two major sections of Brassica. This placement was first made (Palmer et al., 1983) without the use of an outgroup by invoking the molecular clock assumption (Wilson et al., 1977), which now appears to not hold for cpDNA (see Fig. 8 and “Directions and Prospects") but has since been validated by outgroup comparison (J. Palmer, unpublished data). A similar contrast be- tween molecular and morphological data, suggest- ing the need for reassessment of generic bound- aries, has recently been made in the case of Clarkia and the monotypic genus Heterogaura. Hetero- gaura heterandra is clearly a derivative of Clarkia at the molecular level but features a very divergent morphology from all Clarkia species (Sytsma & Gottlieb, 1986b; also see papers in this volume by K. Sytsma & J. Smith and L. Gottlieb). Both Erickson et al. (1983) and Palmer et al. 1983) obtained considerable insight into the origins of the amphidiploid species B. carinata, B. juncea, and B. napus. The cpDNAs of B. carinata and B. juncea were essentially identical to those of the diploids B. nigra and B. campestris, respectively (Fig. 6). Since cpDNA is maternally inherited in ~ 1192 Annals of the Missouri Botanical Garden Brassica (Erickson et al., 1983), one can conclude that these latter two species served as the maternal parents in the interspecific hybridizations that gave rise to the two amphidiploids. By subtraction, B. oleracea and B. nigra must have served as the paternal parents in these crosses. The cpDNA data permitted two quantitative conclusions regarding the timing of hybridization, although it should be kept in mind that these conclusions rest on an implicit assumption of rate constancy. Brassica carinata and B. juncea were judged to result from recent hybridization events since their cpDNAs were identical at all 3,000 base pairs compared with those of specific parental lines (Fig. 6; Palmer et al., 1983). However, these recent hybridizations must have taken place after a substantial period of separation and diversification of the parents, since the two pairs of parental cpDNAs were sub- stantially diverged (29-30 mutations distinguish the B. nigra genome from those of B. oleracea and B. campestris; Fig. 6). Surprisingly, the chloroplast genome of the third amphidiploid, B. napus, is less closely related to those of its uncontested nuclear parents, B. cam- pestris and B. oleracea, than they are to each other (Fig. 6). To resolve this conflict between nuclear and cytoplasmic phylogenies, Palmer et al. (1983) hypothesized that B. napus gained its cy- toplasm by introgression from some unidentified species. In contrast, in Helianthus concordant re- sults obtained using both cpDNA and nuclear DNA do not support a morphologically based hypothesis of introgression (Rieseberg et al., 1988). FAMILIAL RELATIONSHIPS A great challenge to plant sytematists is to re- construct phylogeny among genera and at higher levels, where relationships are in general much more poorly understood than among congeneric species. Although little has yet been published on higher level cpDNA systematics, results are en- couraging from several studies in progress, and therefore we feel it important to discuss these in some detail. In addition, the reader should consult the article in this volume by Sytsma & Smith, which discusses their ongoing work on the Ona- graceae (also see Sytsma & Gottlieb, 1986a, b). Asteraceae. the largest and most successful flowering plant fam- ilies, consisting of 12-17 tribes, approximately 1,100 genera, and 20,000 species (Cronquist, 1981). A combination of specialized floral char- acters (capitula, reduced and modified floral parts, inferior ovaries, basal and erect ovules, and syn- The Asteraceae make up one of genesious anthers) support the monophyly of the family. Recent classifications (Thorne, 1983; Dahl- gren, 1980; Takhtajan, 1980; Cronquist, 1981) emphasized the distinctness of the family by placing it in a monotypic order at the most advanced po- sition in the Dicotyledonae. Although there is some controversy concerning the age of the family (Turner, 1977), fossil evidence (Cronquist, 1977; Muller, 1981) and biogeographical considerations (Raven & Axelrod, 1974) suggest that the Aster- aceae originated in the middle to upper Oligocene (30 million years ago) and subsequently underwent rapid and extensive diversification. During the past 30 years, six pan schemes of phylogenetic relationships amon milies and tribes have been proposed (oin 1955, 1977; Carlquist, 1976; Wagenitz, 1976; Jeffrey, 1978; Thorne, 1983; Bremer, 1987). Most of these recent classifications agree that two distinct subfamilies (Asteroideae and Cichorioideae) should be recognized; however, there is no consensus con- cerning the circumscription of the subfamilies, the number of monophyletic tribes, and the relation- ships among tribes. Two reasons account for the lack of agreement on intrafamilial relationships in the Asteraceae. First, previous studies have relied almost completely on morphological characters, which have undergone repeated parallel and con- vergent evolution. Second, only the most recent reassessment of relationships (Bremer, 1987) has applied cladistic approaches to phylogenetic re- construction, and even that study is limited by high levels of homoplasy and a lack of statistical testing of alternative trees. To provide new characters to aid in clarifying relationships in this complex family, we have ana- lyzed cpDNA variation within the Asteraceae and putatively related families. We have completed studies using two approaches to the study of cpDNA evolution, the assessment of genome arrangement and comparative restriction site mapping, and re- cently initiated a comparative study of rbcL se- quences. Our studies have revealed two genome arrange- ments in the Asteraceae that differ by a single inversion (Jansen & Palmer, 1987a, b). Chloroplast DNAs from the subtribe Barnadesiinae (tribe Mu- tisieae) are colinear with those of most other land plants, including ten families putatively related to the Asteraceae. All other Asteraceae examined (57 genera from 16 tribes) share a derived 22-kb in- version. This rearrangement defines a basal evo- lutionary dichotomy within the family and has two important phylogenetic implications. First, the Mu- tisieae are paraphyletic, as previously hypothesized Volume 75, Number 4 1988 Palmer et al. 1193 Chloroplast DNA Variation on the basis of morphological evidence (Small, 1918; Wodehouse, 1928; Cabrera, 1977; Bremer, 1987). Second, the Barnadesiinae represent the sister group to the rest of the family, which resolves one of the most controversial systematic issues within the As- teraceae. Five different tribes—Cardueae, Helian- theae, Mutisieae, Senecioneae, and Vernonieae— have previously been suggested as the most prim- itive lineage (Cronquist, 1955, 1977; Carlquist, 1976; Wagenitz, 1976; Jeffrey, 1977). The iden- tification of the earliest lineage provides indirect support for previous hypotheses concerning the origin of the Asteraceae in the Andes of northern South America (Raven & Axelrod, 1974; Turner, 1977) and the primitive woody habit and bilabiate flowers of the ancestors of the family (Carlquist, 1976; Jeffrey, 1977). Our restriction mapping study has been carried out in two stages. We first examined taxa primarily from the tribe Mutisieae (Jansen & Palmer, 1988) and then studied representatives of the entire fam- ily (R. Jansen, H. Michaels, J. Palmer, unpublished data). Initially, cpDNAs from 13 genera of the Mutisieae, one genus from each of three other tribes, and two genera from two outgroup families were analyzed with ten restriction enzymes. A total of 211 restriction site mutations were detected, 55 of which were phylogenetically informative. Wag- ner and Dollo parsimony trees constructed with these data were very similar; only the Wagner tree is discussed here. The Wagner parsimony analysis resulted in a single most parsimonious tree (Fig. 7, top) with 247 steps and 15% homoplasy. Four major phylogenetic relationships are depicted in this tree. The most significant is the initial dichot- omy separating the subtribe Barnadesiinae (Mutisi- eae) from the rest of the Asteraceae, including the three other subtribes of the Mutisieae. This is the same dichotomy defined by the inversion described above. The robustness of this initial dichotomy has strong statistical support (98% confidence interval) by the bootstrap analysis of Felsenstein (1985). Furthermore, a recent cladistic analysis (Bremer, 1987) based primarily on morphological data, but also including the 22-kb cpDNA inversion (Jansen & Palmer, 1987b), placed the Barnadesiinae at a basal position within the Asteraceae. Also in accord with the inversion result, the restriction site data indicate the Mutisieae as paraphyletic, since three of the four subtribes are more closely related to the three other examined tribes of the Asteraceae than to subtribe Barnadesiinae. The molecular phy- logeny (Fig. 7, top) also provides support for the monophyly of three of the four currently recog- nized subtribes of the Mutisieae (sensu Cabrera, 1977). Only the morphologically diverse and geo- graphically widespread subtribe Gochnatiinae is shown to be paraphyletic, a conclusion also reache by Bremer (1987). To clarify further relationships within the Mu- tisieae, a more extensive analysis was performed in which 12 of the 16 genera of Asteraceae ex- amined above were analyzed using 19 restriction enzymes (Jansen & Palmer, 1988). A total of 390 restriction site mutations were detected, 117 of which were phylogenetically informative. Wagner and Dollo parsimony analyses again gave similar results. The Wagner analysis (Fig. 7, bottom) re- sulted in a single most parsimonious tree of 454 steps and 14% homoplasy. The tree provides fur- ther support for the relationships indicated by the ten-enzyme tree (Fig. 7, top). This is reflected in the higher confidence intervals for a number of monophyletic groups, including all taxa that share the cpDNA inversion, the three subtribes of the Mutisieae with the inversion, and the subtribes Mutisiinae and Nassauviinae. The 19-enzyme tree also indicates that subtribe Gochnatiinae is para- phyletic. More detailed phylogenetic comparisons were performed using 11 enzymes and 57 genera of the Asteraceae, representing all currently recognized tribes. A total of 926 restriction site mutations were mapped, 328 of which were phylogenetically informative. Wagner parsimony analyses using the global swapping option of PAUP generated 12 equally parsimonious trees of 1,315 steps, all of which support the monophyly of the subfamily As- teroideae (sensu Thorne, 1983). Phylogenetic anal- yses of the data using the same options in PHYLIP gave six equally parsimonious trees of 1,316 steps, all of which support two monophyletic subfamilies (Asteroideae and Cichorioideae sensu Thorne, 1983). A bootstrap analysis (Felsenstein, 1985) produced a majority rule consensus tree (Fig. 8) with 1,318 steps and 30% homoplasy. Although this tree is three steps longer than the most par- simonious tree, it is presented here because it shows the groups that are best supported statistically. The most significant implications of the cpDNA phylogeny for the Asteraceae (Fig. 8) concern the circumscription of tribes and subfamilies and phy- logenetic relationships among tribes. There is strong support for the monophyly of the subfamily As- teroideae, which includes the eight tribes Tageteae, Heliantheae, Eupatorieae, Calenduleae, Seneci- oneae, Inuleae, Anthemideae, and Astereae. This group occurs in all most parsimonious trees gen- erated by both PAUP and PHYLIP and has a confidence interval of 86%. The remaining tribes 1194 Annals of the Missouri Botanical Garden Dipsacaceae Rubiaceae Barnadesia Chuquiragua Barnadesiinae Dasyphyllum Ainsliaea | Gochnatia Stifftia Onoseris Gerbera Ge DIEN Mutisiinae Mutisia J Acourtia Perezia Nassauviinae Gochnatiinae Trixis Heliantheae 54% 1 Cichorieae 53% Cardueae Rubiaceae Barnadesia Dasyphyllum Barnadesiinae Ainsliaea Gochnatia Stifftia Onoseris | Gerbera Leibnitzia Mutisia Acourtia Gochnatiinae Mutisiinae Nassauviinae Perezia Cichorieae FIGURE 7. Wagner parsimony trees of genera in the Mutisieae and ud — eae. The —€— » each node and along each lineage indicate the number of site mutations. The ntages indicate the er of times that a monophyletic grap occurred in 100 bootstrap samples i 1985). The arrows sic ate the occurrence of a sie pDNA inversion “(dans en & Palmer, oe bce? species was examined from each genus, which are as ee n for the sieae, and for the other taxa are: Dipsacaceae, Cephalaria and Dipsacus; Rubiaceae, Pass and Psychotria; [arte da Helianthus; Cichorieae, Lactu uca; Cardueae, Carthamnus. (Top) Single most parsimonious tree for 16 species of Asteraceae using 211 cpDNA restriction site mutations identified with 10 enzymes. ae he pe fos sid steps and 15% homoplasy, including 27 parallel losses, two parallel gains, five s t gas and tw s Botto m) Single isis parsimonious tree for 12 species of Asterac pa using 390 cpDNA restriction site m utations identified with 19 enzymes. The tree has 454 steps and 14 muc iun inc EON 41 p losses, six parallel gains, nine gains/losses, and eight losses / gains. ded e with permission from Jansen & Palmer (1988). Volume 75, Number 4 1988 Palmer et al. Chloroplast DNA Variation 1195 form a paraphyletic group, indicating that the Ci- chorioideae as circumscribed by most recent work- ers (Carlquist, 1976; Wagenitz, 1976; Jeffrey, 1978; Thorne, 1983) may not be a strictly natural group. This conclusion is consistent with the recent morphologically based cladistic study of Bremer (1987). The molecular phylogeny indicates that 11 of 14 currently recognized tribes are monophyletic, with the Heliantheae, Mutisieae, and Tageteae being paraphyletic. The cpDNA data (Fig. 8) reveal that the Liabeae and Vernonieae are monophyletic, in agreement with the finding of Bremer (1987). Re- lationships among the eight tribes in subfamily As- teroideae are clearly resolved in some instances. The previously recognized tribes Cotuleae and Ur- sinieae (sensu Jeffrey, 1978; Robinson & Brettell, 1973) are closely allied to the Anthemideae, which agrees with their placement in this tribe by Bremer & Humphries (in press). The Tageteae are para- phyletic with respect to the Heliantheae and both tribes are paraphyletic with respect to the Eupa- torieae. The very close relationship between the Eupatorieae and Heliantheae does not agree with Bremer’s (1987) placement of the former tribe close to the Astereae. Except for the close rela- tionship of the Calenduleae and Senecioneae, there is little resolution of tribal affinities for the re- maining members of the Asteroideae. Further phy- logenetic analyses using Dollo parsimony and a careful reassessment of character homology are under way to resolve more fully the relatedness of these tribes. For example, an analysis in which statistical approaches were used to eliminate the six most homoplasious characters produced a single most parsimonious tree which again strongly sup- ported the Asteroideae as monophyletic but also weakly supported the monophyly of the Cichorioi- deae The substitution rate in cpDNA (as measured indirectly by the number of restriction site muta- tions) is markedly faster in the derived subfamily Asteroideae than in the basal, paraphyletic Cicho- rioideae (Fig. 8). At the extremes, tracing back to the node defined by the 22-kb inversion, only 14 site mutations are derived in the lineage leading to Stifftia (Gochnatiinae), compared with 116 in San- tolina (Anthemideae). This greater than eight-fold discrepancy is not in accord with the molecular clock hypothesis (Wilson et al., 1977) but is con- sistent with recent findings of rate inequities in the evolution of nuclear DNA (Britten, 1986) and mi- tochondrial DNA (Wolfe et al., 1987; Palmer & Herbon, in press). The existence of significant rate differences in cpDNA evolution within the Asteraceae means that a phenetic (distance) analysis may produce a tree that conflicts substantially with the true phylogeny, particularly if the proper algorithm is not used (Nei, 1987). Many lineages within the Asteroideae are phenetically more closely related to members of the Cichorioideae than to other members of the Asteroideae. Such a relationship is also apparent upon a phenetic analysis of the limited rbcL se- quences (only three) for the Asteraceae (Fig. 3), which shows Helianthus (Asteroideae) as an outlier with respect to Lactuca (Cichorioideae) and Se- necio (Asteroideae). This phenetic analysis of se- quence data (Fig. 3) is at odds with cladistic anal- yses of both the same data (H. Michaels, R. Jansen, J. Palmer, unpublished) and also restriction site data (Fig. 8; note the high confidence interval, 86%, of the Asteroideae lineage). Orchidaceae. The focus of our molecular studies in the Orchidaceae has been the neotropical orchid subtribe Oncidiinae. This large subtribe 1,500 species; 75 genera) is one of the most — diverse orchid subtribes in terms of floral and veg- etative morphology and chromosome number and has long been viewed as "natural" and isolated relative to other orchid subtribes in the New World. he taxonomic circumscriptions of the two largest genera, Oncidium (450 species) and Odontoglos- sum (250 species), have been considered problem- atic since the 1840s. We hoped that a cpDNA- based study of phylogenetic relationships among the groupings identified as natural units (genera and various subgeneric categories) would aid in taxonomic realignment. We ( ase & J. Palm- er, unpublished) used ten restriction enzymes to analyze cpDNA variation in 99 species that rep- resent most of the currently recognized generic and subgeneric groupings in the Oncidiinae and an additional 33 species from other subtribes in the Epidendroideae sensu lato. One problem we hoped to address with a phy- logeny reconstructed from molecular data was the evolution of chromosome number in the Oncidiinae, which ranges from n = 5-30 among species that are not obvious polyploids. Prior hypotheses pos- tulated that this range results from hybridization among species of lower numbers followed by epi- sodes of chromosome doubling (Garay, 1963; Charanasri & Kamemoto, 1975). Chase (1986) studied morphological trends and concluded that hybridization and polyploidy were not involved and that taxa with lower numbers were the results of several independent lineages experiencing reduc- tion from primitively high numbers. The number of loci coding for soluble enzymes was found to be 1196 Annals of the Missouri Botanical Garden 1: arnadesia Barnadesiinae* um. 1 - poetics = Ainsliaea Gochnatiinae* Stifftia ] 1 16 Mutisia 7 Mutisiinae* 16 Gerbera 1—2 Acourtia isi 14 Perez ] Nassauviinae* 12 Centaurea 19 Carthamnus 3 3 C Cardueae £ Silybum A 18 Echino 6 == Lychnophora al g Piptocarpha Vernonieae 5 kes 4 2 Vernonia 7 > Cacosmia ] Lisbsas lab 4 12 Lactuca 1 : 24 Sonc Cichorieae Hieracium 3 22 Tragopogon = $ tot Z Arctotideae and 1 Astereae ed chrysanthemum Anthemideae hi lle 5 Inuleae i Senecioneae io ari a pt pe Calenduleae ] Eupatorieae me | Heliantheae Tageteae Dyssodia Ficure 8. Molecular phylogeny of the Asteraceae. Shown is a nu rule consensus tree generated by the bootstrap option o P using 926 restriction site mutations. The numbers at each node and along each lineage indicate the number of restriction site mutations. The tree Rus a tot add 1.318 ane and 30% homoplasy and is rooted dgio to the Barnadesiinae. Bracket s show Mir current ins coat of 13 of the tribes, while the four suliribe s of the Mutisieae are indicated with asterisks. The a rrow indicates the occurrence of a 22-kb cpDNA inversion n. (Jansen & Palmer, 19875). Unpublished Vin of R. Jansen, H. Michaels, and J. sien the same among species exhibiting the complete range of chromosome number (Chase & Olmstead, 1988), making gei ba Lied an "HOURS expla- nation to account for th ul e cp phylogeny (M. Chase & J. Palmer, unpublished) strongly supports the idea that members of several lineages have ci parallel reduction in chro- mosome nu The ol idea that the Oncidiinae comprise an isolated subtribe was shown by a study of seed Volume 75, Number 4 1988 Palmer et al. 1197 Chloroplast DNA Variation morphology (Chase & Pippen, in press) to be fal- lacious (i.e., a number of other subtribes have the same seed morphology). The molecular phylogeny has identified as close relatives many of the same subtribes as did the study of seed morphology. Among these other subtribes are the Bifrenariinae, Lycastinae, Maxillariinae, Xylobiinae, and, in p ticular, the Lockhartiinae and Ornithocephalinae. Most systems of orchid classification have used pollinium number to place all but the Lockhartiinae in another tribe from that of the Oncidiinae. Pol- linium number appears not to be a reliable indicator of relationships in these orchids (Chase & Pippen, in press; M. Chase & J. Palmer, unpublished). The cpDNA phylogeny is concordant with the evidence from morphology that Oncidium and Odontoglossum are not monophyletic genera. On- cidium in particular is paraphyletic rather than a genus held together by synapomorphies. The mor- phologically distinct derivatives of this assemblage have been recognized as different genera. The ap- parent phylogenetic radiation that produced this large species complex has a strong geographic com- ponent. Members of several sections of Oncidium endemic to the mountains of southeastern Brazil losely related to genera, such as Gomesa, endemic to this same region, than to morphologi- cally similar sections of Oncidium from other geo- graphical areas. We failed to find cpDNA mutations that could establish relationships among several well-defined (both by molecular and morphological data) lin- eages of the Oncidiinae. One explanation for this result is a slowed rate of molecular change during the period when these lineages arose. This expla- nation also requires a corresponding decrease in the rate of morphological change followed by in- creases in molecular and morphological rates dur- ing subsequent evolution in several now-separate lineages. A more likely explanation we feel is that a large number of lineages diverged in a short period of evolutionary time from a polymorphic ancestral stock. This process must have happene so rapidly that either no mutations exist at this level or the few mutations that do exist convey conflicting ideas of relationships because of ho- moplasy. The implications of such a phylogenetic radiation are significant in evolutionary and taxo- nomic terms. Radiations of this type have long been suspected to be responsible for the large num- bers of closely related, taxonomically difficult species and genera found in most large families of vascular plants, but this is the first time molecular evidence documenting such a phenomenon has been uncov- ered (for a similar case in Clarkia, see symposium paper by K. Sytsma in this volume). Fabaceae. Chloroplast DNA variation has been used to explore interspecific relationships in several legume genera (reviewed in Doyle, 1987), and studies using restriction site mapping are currently in progress to address relationships in several tribes of legumes (J. Doyle, pers. comm.). However, the high level of cpDNA sequence divergence in certain legumes (Palmer & Thompson, 1982) suggests that such efforts would be largely fruitless in these groups. Several cpDNA rearrangements have been well characterized in legumes and appear to be useful phylogenetically. Two rearrangements, a 50-kb inversion and the loss of the rp/22 gene from the chloroplast genome (and its concomitant transfer to the nucleus), are shared by all members examined of the three subfamilies of legumes (Figs. 4, 5; Palmer et al., 1987; J. Palmer & J. Doyle, unpublished data). These two events may constitute legume-specific markers, as they are absent from the cpDNAs of such putatively related families as Rosaceae and Sapindaceae. However, other close relatives, such as the Connaraceae and Chrysobalanaceae, remain to be examined. A 78-kb inversion that includes most of the large single copy region of the chloroplast genome (Palm- er et al., in press) has been surveyed in 50 genera of papilionoid legumes, principally in the tribe Phas- eoleae (Bruneau et al., shared by all members examined of the subtribe Phaseolinae sensu Lackey (1981), and its distri- bution also helps elucidate the placement of several problematic genera. For example, Clitoria and Centrosema lack the inversion and therefore ought to be excluded from a narrowly defined Phaseolinae for which this chapan 3 is a apa pomopii In lacement his inversion is contrast, Macrotyloma, has been disputed, has the inversion and on this basis is best placed in the Phaseolinae. gume phylogeny, is the loss of the large cpDNA inverted repeat. This inverted duplication, which typically is on the order of 25 kb in size (Fig. 1), is a hallmark of land plant cpDNAs (reviewed in Palmer, 1985b, c). The repeat structure has been lost (one of the two repeat elements has been de- leted, without the loss of genetic information) only twice among all of the nearly 1,000 species of land plants whose cpDNAs have been investigated. One of these absences characterizes at least two genera 1198 Annals of th Missouri BOIS Garden of Pinaceae (Fig. 5; Strauss et al., 1988) and the other several tribes of legumes (Palmer et al., 1987). A total of 136 species of legumes have now been surveyed for the presence of the inverted repeat, including 124 species representing 76 genera and 24 of the 30 tribes of subfamily Papilionoideae (Lavin et al., 1988). The inverted repeat is present in most legumes, including members of all three subfamilies and a taxonomically diverse group of papilionoid tribes (Fig. 5). It is absent in all mem- bers examined from the Galegeae, Vicieae, Cicer- eae, Trifolieae, Carmichaelieae, and Hedysareae, a group of temperate herbaceous papilionoid tribes characterized by a combination of features, such as reduction or loss of the pulvinus and a base chromosome number of n — 7 or 8. In this respect our results are consistent with phylogenetic hy- potheses based on traditional characters. However, two areas of apparent disagreement exist. First, Wisteria is the only member of the tropical tribe Milletieae that lacks the inverted repeat. Wisteria is also unusual within the Milletieae in its entirely temperate distribution and its base chromosome number of n — 8, both characters that link it with the Galegeae, which also lack the inverted repeat. Secondly, Loteae and Coronilleae, which have the inverted repeat, are commonly grouped on the basis of morphology with temperate herbaceous tribes that lack it. groups of tribes is, however, The distinction berweon. these two ] y some nonmolecular characters, such as tool nodule mor- phology. Circumstantial evidence and theoretical specu- lation suggest that the absence of the inverted repeat leads to an unstable chloroplast genome prone to more frequent rearrangement (Palmer & Thompson, 1982; Palmer, 1985b; Strauss et al., 1988). In some legumes, other factors, such as the presence of large dispersed repeats, also contribute to an unstable, actively rearranging chloroplast genome (Palmer et al., is acceleration in rearrangement rate can be dramatic; for ex- ample, genomes in Pisum and Trifolium differ from those of other genera in the same tribes by eight or more inversions (Fig. 5). Contrast this with the colinearity of cpDNAs of major groups of vascular plants (Fig. 1; Palmer & Stein, 1986) and with the single inversion that distinguishes cpDNAs of vascular plants and some bryophytes (Fig. 5). While interesting from the standpoint of mechanisms of molecular evolution, extensive rearrangement makes it impossible to align much or all the chlo- roplast chromosomes of different taxa, thus largely precluding comparative cleavage site analysis. In addition, the use of the rearrangements as char- acters is usually not justified owing to the enormous amount of work required to define each rearrange- ment as an individual, polarizable mutation. Solanaceae. The chloroplast genomes of species in the Solanaceae are among the best known at the molecular level thanks to the complete se- quencing of cpDNA from tobacco (Nicotiana ta- bacum; Shinozaki et al., 1986). A number of stud- ies have utilized cpDNA restriction fragment variation to examine interspecific relationships in Lycopersicon, Solanum, and Nicotiana (Palmer & Zamir, 1982; Hosaka, 1986, and references therein), but none have addressed higher level is- sues in the family. The limited divergence observed in rbcL sequences of three disparate genera of Solanaceae bodes well for such an approach (Fig. 3; this level corresponds to that seen in the As- teraceae, where site mapping has proven success- ful), and we have recently initiated a comparative mapping study of 100 representative species and genera of Solanaceae. The rbcL sequence data, limited as they may be, are tantalizing in placing Nicotiana and Ly- copersicon closer to each other than either is to Petunia (Fig. 3). Although in conflict with standard schemes (D’Arcy, 1979), which place Nicotiana and Petunia in the same tribe (Nicotianeae) and Lycopersicon in a different subfamily (Solanoi- deae), this relationship is supported by sequence data for several nuclear genes (Pichersky et al., 1986) and by chromosome numbers (n — 12 for tobacco and tomato and n — 7 for petunia). It will be important to see whether the controversial re- lationships suggested by the phenetic treatment of rbcL sequence data (Fig. 3) are supported by the large-scale mapping study in progress, as well as by a cladistic analysis and a more sophisticated phenetic analysis (Nei, 1987) of the same sequence data. Poaceae. Chloroplast DNA has been used ex- tensively to investigate interspecific relationships in grasses (particularly in cereals), more so than in any other plant family (reviewed in Hilu, 1987). However, with few exceptions (e.g., Doebley et al., 1987), these studies have yielded limited phylo- genetic insights owing to insufficient sampling of taxa and cpDNA sequences and also to nonrigorous methods of phylogenetic analysis. Similar problems exist for two studies of intergeneric and intertribal relationships of grasses (Lehvaslaiho et al., 1987; Enomoto et al., 1985). Despite these problems, a surprisingly reasonable phylogeny was derived for ten grass genera from an analysis of cpDNA re- striction fragment patterns (Enomoto et al., 1985). Volume 75, Number 4 1988 Palmer et al. 1199 Chloroplast DNA Variation The rbcL gene has been sequenced more ex- tensively in grasses than in any other plant family. The resulting phenogram (Fig. 3) shows good agreement with previously held concepts (Watson et al., 1985) and with a cladogram based on nuclear ribosomal RNA sequences (Hamby & Zimmer, 1988). This latter congruence is comforting, given the possibility for reticulate evolution to produce different phylogenies for maternally inherited chlo- roplast genes and biparentally inherited nuclear ones. The closer molecular relationship of Triticum aestivum to Aegilops crassa than to T. urartu (Fig. 3) is not surprising, since 7riticum is an unnatural group with strong affinities to Aegilops and to several other genera in subtribe Triticinae (Kerby & Kuspira, 1987) Several cpDNA inversions have been charac- terized in one or a few grasses, and at least one of these is shared by members of different subfamilies (Palmer & Thompson, 1982; Quigley & Weil, 1985; Hirai et al., 1985). These inversions are lacking in the four other families of monocots whose cpDNAs have been mapped (Fig. 1; de Heij et al., 1983; J. Palmer, unpublished). Careful exploration of the phylogenetic distribution of these inversions should, therefore, define several major branchings of grass evolution. HIGHER-ORDER RELATIONSHIPS The major avenue for exploring relationships among families and at higher levels is quencing. Supplementary information will also come from the rare and powerful rearrangements that lead to alterations in chloroplast genome structure and gene content. Land plants. As discussed in the section on DNA sequencing, rbcL is the chloroplast gene of nogram in Figure 3 offers little revelation con- cerning higher-level questions of angiosperm phylogeny, this situation will soon change owing to the coordinated rbcL sequencing efforts under way in several laboratories. The correspondence of the rbcL phenogram and also those based on more limited sampling of the genes atpB (Ritland & Clegg, 1987) and psbA (Wu et al., 1987) with generally held concepts of angiosperm phylogeny is encouraging. [n contrast, a phylogenetic analysis based on sequences of the chloroplast rp/2 gene conflicts with generally accepted classification schemes and with other molecular data (Zurawski & Clegg, 1987). The explanation for this discrep- ancy appears to lie in the unusual evolutionary properties of rpl2, which is subject to a substantial selective constraint on third-position substitutions, perhaps owing to the presence of a gene on the strand opposite to that encoding rpl2 (Zurawski & Clegg, 1987). It is important to emphasize that this anomalous evolutionary behavior of rp(2, which makes it such a poor choice for phylogenetic study, was observed directly from consideration of the dynamics of nucleotide substitutions (Zurawski & Clegg, 1987), an observation independent of the realization that the gene is a poor phylogenetic marker. Despite the lack of any systematic effort to identify and survey major rearrangements among land plant cpDNAs, several already have turned up that promise to mark dichotomies among major plant groups. The best documented is the afore- mentioned loss of the rp/2 intron (Figs. 4, 5), an event that unites the Caryophyllales (Centrosper- mae) in a manner consistent with the analysis of Rodman et al. (1984). Other rearrangements for which further study promises to illuminate specific branchings include the loss of the large inverted repeat in two genera of Pinaceae (Strauss et al., 1988) and a 30-kb inversion of unknown evolu- tionary polarity that distinguishes vascular plants from the single liverwort and moss examined thus far (Fig. 5; Ohyama et al., 1986; Calie & Hughes, 1987b). Algae. Our knowledge of the structure and sequence of algal chloroplast genomes is quite lim- ited compared with those of land plants. No sig- nificant comparative sequence study or data base yet exists for algal chloroplast genes, although sev- eral laboratories are now engaged in sequencing chloroplast ribosomal RNA genes in major algal groups (pers. comm. from S. Giovannoni & K. Field, R. Chapman & E. Zimmer, R. Cattolico, and J. Manhart & J. Palmer). Rates of structural rearrangement and size change in cpDNA appear to be substantially higher in algae, particularly green algae, than in land plants (e.g., Lemieux & Lemieux, 1985; reviewed in Palmer, 1987). These high rates probably preclude the use of inversions as phylogenetic characters at higher levels among algae, although they may be useful within genera. Other organizational features, in particular the con- tent of genes and introns, hold promise for defining major events of algal evolution. The use of two such markers for defining the green algal origins of land plants is described in the next two para- aphs. The gene tufA, encoding the chloroplast protein 1200 Annals of the Missouri Botanical Garden synthesis factor EF-Tu, has previously been shown to be encoded by cpDNA in the phytoflagellate Euglena gracilis (Montandon & Stutz, 1983) and suggested to be cpDNA-encoded in the green alga Chlamydomonas reinhardtii (Watson & Sur- zycki, 1982). However, this gene is absent from the cpDNAs of an angiosperm and a bryophyte, as demonstrated by complete sequencing of the two genomes (Ohyama et al., 1986; Shinozaki et al., 1986). Recent studies (Fig. 5; J. Manhart, S. Baldauf & J. Palmer, unpublished data) reveal that tufA is indeed encoded by cpDNA in Chlamydo- monas and all other green algae examined, with the possible exception of Coleochaete, the pre- sumed closest relative to land plants (Graham, 1985). In contrast, we find tufA to be a nuclear gene in land plants. The transfer of tufA from the chloroplast to the nucleus thus defines land plants as a natural, monophyletic group, with the dispo- sition of Coleochaete with respect to this character in need of further study. The chloroplast gene for tRNA-isoleucine con- tains a large intron in all land plants examined (Ohyama et al., 1986; reviewed in Palmer, 1985c) but lacks introns in all eukaryotic algae, including the green algae Chlamydomonas (Schneider & Rochaix, 1986) and Chlorella (Yamada & Shimaji, 1986). We find this intron to be present in Nitella cpDNA and are examining other charophytes and other classes of green algae (Fig. 5; J. Manhart & J. Palmer, unpublished data). This intron gain thus has the potential to define a specific lineage of charophytes as the sister group of land plants, consistent with ultrastructural and other evidence (Mattox & Stewart, 1984). A signal accomplishment of molecular system- atics is the unambiguous resolution of the long- standing controversy regarding the very origin of the chloroplasts. Findings from numerous molec- ular approaches, including the analysis of rRNA gene sequences (Fig. 9) and of gene organization (e.g., Cozens & Walker, 1987), have established unequivocally that chloroplasts are derived, en- dosymbiotic, photosynthetic bacteria (most likely cyanobacteria; reviewed in Gray, 1983). The chal- lenges that remain for molecular systematists are to elaborate the full history of chloroplast endo- symbiosis. Have different chloroplast lineages been derived from just one or from multiple independent and/or serial endosymbioses, and what are the sources of the plastid and nucleus in each of these associations (Whatley, 1983; Palmer, 1985b)? Analysis of ribosomal RNA sequences (Giovannoni et al., 1988) and consideration of chloroplast ge- nome size and gene content (Palmer, 1985c) favor the idea of a single initiating endosymbiosis (i.e., one between a photosynthetic prokaryote and a nonphotosynthetic eukaryote), followed by several strictly eukaryotic symbioses. DIRECTIONS AND PROSPECTS Chloroplast DNA is now routinely used to in- vestigate interspecific relationships among angio- sperms and other plants. Its popularity at this level reflects both the general preoccupation of plant systematists with interspecific questions and also a set of factors that make cpDNA analysis especially appropriate at this level. These include the special utility of cpDNA for providing insights into the directionality of the hybridization events that are so common in plants, the relative ease of comparing cpDNA molecules at this level, the sufficiency of readily detectable mutations, and the low levels of homoplasy associated with the cladistic analysis of these mutations. With current practices, the value of cpDNA analyses to reveal patterns of intraspecific variation and processes of speciation is marginal owing to the slow rate of cpDNA evolution. A concentrated effort is necessary to develop technical strategies that will allow the routine accumulation of sufficient numbers of cpDNA mutations to make the molecule predictably useful below the species level. In the short term such an enlargement of the cpDNA data base will involve increasing the number of restric- tion sites scored in a typical survey. This will mean using more retriction enzymes, particularly ones that cut frequently (100—400 times per genome), utilizing electrophoretic techniques (e.g., acryl- amide gels of the type normally used for DNA sequencing) adequate to the task of resolving mul- tiple small fragments, and hybridizing blots made from these gels repeatedly with cloned fragments spanning the genome. In the long term, it is con- ceivable that sufficient advances in the technology of DNA sequencing (e.g., of the type presently being sought to allow the sequencing of the three billion base pair genome of humans) will occur to permit efficient intraspecific study by this ap- proach. The analysis of phylogenetic relationships among genera and tribes is currently a nexus for the various approaches used to compare cpDNA mol- ecules. Restriction site mapping, the easiest ap- proach, is of demonstrated utility within families showing moderate levels of cpDNA divergence, such as the enormous family Asteraceae. In fam- ilies of greater molecular diversity, such as the Fabaceae and Onagraceae (see symposium paper Volume 75, Number 4 Palmer et al. 1201 1988 ick ie DNA Variation Halobacterium volcanii 0.1 fixed A: Sulfolobus solfataricus Methanospirillum hungatei sequence position alienos Thermoproteus tenax Methanobacterium formicicum Methanococcus vannielii Homo sapiens P laevis A m mays S , loarn, cere KA “Ny "Slog, y “By, “Cy Oy gcc jn 4 OU aise a, “S Y m e o 2: Gz y Té, oy?" yi TN id pic w Ra ^6 KÇ p 2. " 10 % EUKARYOTES EUBACTERIA FIGURE 9. An unrooted dendrogram based on a phenetic analysis of 16S and 18S ribosomal RNA sequences. See Pace et al. (1986) for details p tree construction and references to the primary sequence literature. Reprinted L (1986). with permission from Pace et a by Sytsma & Smith in this volume), direct se- quencing (e.g., of rbcL) is a more fruitful approach. It is presently unclear, and hence unpredictable, as to why the cpDNAs of certain families are too diverse to permit comparative restriction site map- ping. Factors such as the size and age (see final paragraph) of the family may in some cases be unrelated to its molecular diversity. In certain fam- ilies, such as the Fabaceae, Asteraceae, and per- haps Poaceae, major rearrangements may also make significant phylogenetic contributions. Perhaps the major accomplishment of cpDNA systematics, and molecular systematics in general, will be the resolution of the hotly debated major branchings of plant evolution, such as relationships among the families of flowering plants. DNA se- quencing will be the major tool in this effort, with rearrangements serving to mark with exquisite clar- ity a subset of branchings. It remains to be seen whether rbcL will be a reliable phylogenetic indi- cator in all lineages. In some cases, particularly very old lineages, more conservative genes such as those encoding rRNAs may be more useful. In other cases, such as rapid radiations, sequence data from multiple genes may be necessary to provide statistically adequate phylogenetic resolution. An increasingly important component of molecular analyses as one moves back in evolutionary time will be the scope and sophistication of the computer programs that construct evolutionary trees and evaluate their statistical significance. A frustrating but fascinating conclusion from recent cpDNA comparisons is that the rate of cpDNA sequence change may vary markedly in different, even closely related, plant lineages. These rate differences were discussed earlier in reference to the Asteraceae phylogeny shown in Figure 8 and have also been inferred from studies in the Fabaceae and Juglandaceae (J. Doyle, pers. comm.), Populus (K. Sytsma, pers. comm.), and Lactuca (E. Jandourek & J. Palmer, unpublished data). The frustrating aspect of these rate differences is the attendant inability to use molecular variation as a “clock” gence times in plant evolution. The fascination stems from the fact that we pres- ently have no satisfactory understanding of the biochemical and genetic bases of such rate heter- ogeneities. In conclusion, cpDNA is in general both conservative and predictable in its evolution, fea- tures which make it extremely useful for inferring phylogeny but of concomitantly lesser interest from the standpoint of molecular evolution. However, aspects of its evolution (not only differences in substitution rates, but also such remarkable mu- 1202 Annals of th Missouri E ESI Garden tations as the transfer of its genes to the nucleus) provide grist for the mill of the molecular evolu- tionist. LITERATURE CITED ALDRICH, J., B. RNEY, E. MERLIN & J. D. PALMER. 1986a. hup of the rbcL gene for le large subunit of ribulose bisphosphate wae paga -oxygen- ase from petunia. Nucleic Acids Res. 14: 4. 1986b. quence of the rbcL gene for the large subunit of ylase-oxygenase from al- falfa. Nucleic Acids Res. 1: 9535. Arcuison, B. A., P. R. Waitretp & W. BOTTOMLEY. l Comparison of chloroplast DNAs by specific I endonuclease. Molec. Gen. fragmentation with Genet. 148: 263-269 AVISE, J. C., J. ARNOLD, R. M. BALL, E. BERMINGHAM, Intraspecific phylogeography: the mitochon- drial bridge between population genetics and system- atics. E al Rev. Ecol. Syst. 18: 489-522. Banks, J. A. . W. Birky, Jr. 1985. Chloroplast DNA — is low in a wild plant, Lupinus tex- ensis. Proc. Natl. Acad. U.S.A. 82: 6950-6954. Benpicu, A. J. 1987. Why do chloroplasts and mito- chondria contain so many copies of their genome? Bioessays 6: 279-282. BookJAns, G., B. M. STUMMANN & K. W. HENNINGSEN. 1984. Preparation of chloroplast DNA from pea plastids isolated in a medium of Y a ionic strength. Analytical Biochem. 141: 244-247 BourrER, D. The evaluation of present results and future possibilities of the use of amino acid se quence data in phylogenetic agn with specific ref- erence to plant proteins. Pp. 235-240 in F. A. Bisby, aughan & C. A. eh (editors), pen tematics: Principles and Practices. Systematics As- analysis of DNA from wheat m isolated in nonaqueous media. Analytical Biochem. : 108 18. l BREMER, K. 1987. Tribal interrelationships of the As- teraceae. Cladistics 3: 210-253. . HUMPHRIES. Generic oen of the Anthentidas (Asteraceae) (in press BRITTEN, R.J. 1986. Rates of DNA sequence evolution differ between taxonomic groups. Science 231: 1393- 1398. BRUNEAU, A., J. J. DOYLE & J. D. PALMER. 1988. A chloroplast DNA inversion as a subtribal character in the tribe Phaseoleae (Leguminosae: Papilionoi- deae). Amer. J. Bot. 75: 162. [Abstract.] CABRERA, A. L. 1 Mutisieae — systematic review Pp. 1039-1066 in V. H. Heywood, J. B. Ha rborne B. L. Turner (editors), The Biology and ee a. ^W “eficient protocol for the isolation and purification of chloro- plast DNA from s aa Dd tissues. Pl. Molec. Biol. Rep. 4: 206- . The consensus land plant chloroplast gene order i is present, with two alterations, in the moss Physcomitrella patens. Molec. Gen. Genet. 208: 335-341. CARLQUIST, S. 1976. Tribal interrelationships and phy- ogeny of the Asteraceae. Aliso 8: 465-492. CATTOLICO, R. A. 1985. Chloroplast biosystematics: as uai DNA as a molecular probe. BioSystems 306. Dm. rising m anda CHARANASRI, U. & H. BAMEMOTO, n Oncidiu Amer. Orchid Soc. Bull. 44: bg Chase, M. W. 1986. A reappraisal of the oncidioid orchids. Syst. Bot. 11: 477-491. & R. G. OLMSTEAD. 1988. Isozyme number in subtribe Oncidiinae (Orchidaceae): an evaluation of po Amer. J. Bot. = 1080-1085. & J. S. Pippen. Seed morphology in the On- cidiinae abd related subtribes (Orchidaceae). Syst. CLEcc, M. T., A. D. H. Brown & P. R. WHITFELD. 1984. Chloroplast DNA diversity in wild and cul- tivated barley: implications for genetic conservation. Genet. Res. 43: 339-343. Coates, D. & C. A. Cuntis. 1987. Chloroplast vs variability among Linum species. Amer. J. Bot. 260-2 CONNELL, C., S. Func, C. HEINER, J. BRipGHAM, V. CHA- KERIAN, E. Nee B. Jones, S. MENCHEN, W. Mor- i, M. Rarr, M. m KNOR, L. ue J. SPRING ER, sequence analysis Biotechniques 5: 342-358. COZENS, à L. & J. E. W; 87. The organization and sequence of the genes for ATP synthase subunits in the cyanobacterium Synechococcus 63 port for an endosymbiotic origin of chloroplasts. J. Molec. Biol. 194: 359-383. CRAWFORD, D. J. 1983. Phylogenetic and oM inferences from iid qs studies. 257- S anksle J. Orton deem Isozymes in Plant eie ast Breeding. Elsevier, Amsterdam Cronguist, A. 1955. Phylogeny and taxonomy of the ompositae. Amer. Midl. Naturalist 53: 478-511. l The Compositae revisited. Brittonia 29: 87 in S. 137-153. 1 An Integrated System of Classification of Flowering Plants. Columbia Univ. Press, New York. DAHLGREN, R. A revised system of classification of the angiosperms. Bot. J. Linn. Soc. 80: 91-124. Improvements to the non- aqueous method of chloroplast DNA preparation as applied to rice. Analytical Biochem. (submitted). L. H. & D. R. PRING A physical ed a the ue nd chloroplast genome. Pl. Molec. Biol. 119-12 DaNc, "a W. x . 3-47 in J. G. Hawkes, R. N. Lester & A D. Skelding (editors The Biology and Taxonomy of the Solana cademic Press, Lon DEBry, A. SLADE. 1985. Cladistic analysis of vasificdon um clease cleavage maps within a maximum-likelihood framework. Syst. Zool. 34: 21- 34 ral The classification of the gar De Hen, H. T., H. Lusto, D. J. M. Moeskops, W. A BovENBERG, C. Bisanz & G. S. P. GROOT. 1983. Chloroplast DNAs of g net Petunia and Spi- Volume 75, Number 4 1988 Palmer et al. Chloroplast DNA Variation 1203 rodela have a similar gene organization. Curr. Genet. T: DOEBLEY, ox W. RENFROE & A. BLANTON. 1987. Re- striction site variation in the Zea chloroplast genome. Genetics 117: 139-147. 1987. Variation at the DNA level: uses and potential in legume systematics. Pp. 1-30 in C H. Stirton (editor), Advances in Legume Systematics, Part 3. Royal Botanic Gardens, Kew. . E. Dickson. 1987. Preservation of plant samples fus DNA restriction endonuclease analysis. Taxon 36: 715-722. & J. L. Doyte. 1987. A rapid DNA isolation procedure for small quen of fresh leaf tissue. Phytochem. g 19: 11-15. Enomoto, S., HARA & K. TSUNEWAKI. i p species by restriction endonuclease analysis of organellar DNA. I. Phylo- genetic relationships among ten cereals ia by the ig oae fragment patterns of chloroplast DNA. Jap. J. Genet. 60: 411-424. ERICKSON, L. R. A. SrRAUS & W. D. BEVERSDORF. 1983. i patterns reveal origins of chlo- roplast genomes in Brassica baia cM Theor. 1985. . H. GELFAN R. K. Saiki. 1988. ecific DNA amplification. Nature 331: 461-462. ran: J. Evolutionary trees from DNA sequences: oo likelihood approach. J. Molec. Evol. 17: —_—, | 3 Inferring evolutionary = ora DNA sequences Pp. in B. Ae t Statistical Analysis of DNA Sequence bs Mar Dekker, New York. 9 Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39: 783- 791. Garay, L. A. . Oliveriana and its position in the Oncidieae. Amer. Orchid Soc. Bull. 32: 19-24. GiANNAsI, D. E. & D. J. CRAWFORD. 1986. iuga s II. A reprise. Pp. 25-248 i K , B. "Wal lace & G. T. Prance EUN Evo- erani Biology, Volume 20. Plenum Press, New York. . Biochemical Systematics (in press). GILLHAM, N. W., J. E. Boynton & E. H. DATAE 1985. Evolution of plastid DNA. Pp. 299-351 in T. Cav- alier-Smith (e nat DNA and Evolution: Natural Se- lection a e Size. Wiley, Pe Yor GIOVANNONI, S. EN, S. BAR Lane & N. R. bici 1988. "Mei es relationships among cyanobacteria wee green chlo- roplasts. J. Bacteriol. 170: 3584-3592. GOTTLIEB, L. D. a ae di evidence and plant gp Prog. Phytochem. 7: 1-46. 985. de L The os p the life cycle of land plants. pea Sci. 73: 178-186. Gray, M. W. The bacterial ancestry of plastids and ioaad, BioScience 33: 693-699. Hampy, R. K. € E. A. ZIMMER. 1988. Ribosomal RNA sequences for inferring phylogeny pers the grass family (Poaceae). Plant Syst. Evol. 9-37 HARBORNE, J. B. & B. L. TURNER. jue Plant Che- mosystematics. Academic Press, Orlando, Florida. HEINHORST, S., G. C. GANNON, E. GALUN, L. KENSCHAFT & A. WEISSBACH. 1988. Clone bank and Sole and genetic map of potato chloroplast DNA. Theo Appl. apes T 244-251. HiLu, K. W. 7. Chloroplast DNA in the systematics and Mix of the Poaceae. Pp. 65-72 in T. R. Soderstrom, K. W. Hilu, C. S. Campbell & M. E. Barkworth (editors), rin Systematics and Evolu- tion. Smithsonian Press, Washington, D. Hiral, A., T. ISHIBASHI, A. MORIKAMI, N. Iwat UKI, K. SHINOZAKI & M. SuciURA. 1985. Rice ice jam DNA: a physical map and the location of the genes for the large subunit of ribulose-1,5-bisphosphate car- boxylase and the 32 K 6. Who is the mother of the potato? — Restriction endonuclease analysis of chloroplast DNA of cultivated potatoes. Theor. Appl. Genet. 72: 606- 618 JANSEN, R. K. & J. D. PALMER. 1987a. Chloroplast DNA from lettuce and Barnadesia (Asteraceae): structure, gene localization and characterization of a large inversion. Curr. Genet. 11: 553-564. & 1987b. A chloroplast DNA in- version marks an ancient evolutionary split in the sunflower family (Asteraceae). Proc. Natl. Acad. U.S.A. 84: 5818-5822. 88. Phylogenetic WERE dU of chloroplast DNA restriction site variation in the Mutisieae (Asteraceae). Amer. J. Bot. 75: 151-764. JEFFREY, C. 1977. Corolla forms in Compositae —some evolutionary and taxonomic speculations. "Pp. 11l- 118 in V. H. Heywood, J. Harborne & B. L. Turner (editors), The M and Chemistry of the Com- positae, . Academic Press, New Y ork. š Com mpositae. Pp. 263-268 in V. H. Heywood (editor), Flowering Plants of ie World. Mayflower Books, New York. . W. GaurscH, J. R. SPORTSMAN & J. Í 1984. Improved technique utilizing non- fat dry milk for analysis of proteins and nucleic acids transferred to nitrocellulose. Gene Anal. Tech. 3-8 JOHNSTON-DOW, L., E. Manpis, C. HEINER & B. A. Rok. 87. Optimized methods for fluorescent and radio polyploid wheats Triticum aestiv and Triticum turgidum (macaroni wheat). Genome 9: 122—131. KoRNELUK, R. G., F. QuaN & R. A. GRAVEL. 1985. apid and reliable dideoxy ae ria of double- stranded DNA. Gene 40: 317 Lackey, J. A. 1981. Phaseoleae. Pp. 301-3 27 in R. M. Po lhill & P. H. p Sy (editors), Advances in 1. Royal Botanic Gardens, = Kew. Lavin, M., A. BRUNEAU, J. J. DoyLE & J. D. PALMER. 1988. Loss of the chloroplasts DNA inverted repeat as a tribal character in the Leguminosae. Amer. J. Bot. 75: 189. [Abstract.] pr wq H., A. Saura & J. Lokki. 1987. Chlo- t DNA variation in the grass tribe Festuceae. Theor. Appl. Genet. 74: 298-302. LEMIEUX, B. & C. LEMIEUX. 1985. Extensive sequence rearrangements in the chloroplast genomes of the green algae Chlamydomonas eugametos and 1204 Annals of the Missouri Botanical Garden Chlamydomonas reinhardtii. Curr. Genet. 10: 213- Lin, C. M., Z. Q. Liu € S. D. Kunc. 1986. Nicotiana chloroplast genome: X. Correlation between the DNA sequences and the isoelectric focusing patterns of the LS of Rubisco. Plant Molec. Biol. 6: 81-87. Martin, P. G. & J. M. Dowb. 1986. Phylogenetic studies using protein sequences within the order Myr- tales. Ann. Missouri Bot. Gard. 73: 442-448. . Morris & D. E. Symon. 1986. The study of ‘plant phylogeny using amino acid se- quences of ribulose-1,5-bisphosphate carboxylase. VI. ome Solanum and allied species from different con- inents. Austral. J. Bot. 34: 187-195. & K.D 1984. Classification of the green algae: a concept based on AA cytology. Pp. 29-72 in D. E. G. Irv . M. ci ll uw of the Green LE Ac- c Press, London Atan DAN. P. E. & E. Sturz. 1983. Nucleotide sequence of a Euglena gracilis chloroplast genome region coding for the elongation factor Tu; evidence for a spliced mRNA. Nucleic Acids Res. 11: 5877- 5892. Moritz, C., T. E. DowLinG € W. M. BROWN. 7. Evolution of animal eae DNA: relevance for ep biology and systematics. Ann. Rev. 292. Ecol. Syst. 18: 269- MULLER, J. 1981. Fossil je records of extant an- giosperms. Bot. Rev. (Lancaster) 47: 1-142. Nel, M. Baie Molecular Evolutionary Genetics. Co- lumbia Univ. Press, New Ce Y. & A. HIRal. 1987. ‘Nucleotide sequence and expression of the gene for the large subunit of rice ae ws: 1,5-bisphosphate carboxylase. Jap. J. Genet. 62: -395. OHYAMA, K., ^ Fuca, T. Konci, H. Surat, T. SANO, S. SANO, K. UMESONO, Y. SHIKI, M. TAKEUCHI, Z. » CHANG, S. AOTA, H. INOKUCHI dx H. OZEKI. 1986. sequence of liverwort Mare hantia 5 in roplast DNA. Nature 322: 572-574 Osre, C. 1988. Polymerase chain reaction. Biotech- ae 162-168. Pace, N. R., G. J. OLSEN & C. R. Woese. 1986. Ri- dpa RNA phylogeny and Len primary lines of evolutionary descent. Cell 45: 325-326. PALMER, J. D. 1985a. Chioroplast DNA and molecular A BioEssays 2: 267. 85b. cete = chloroplast and mito- me DNA i in plants and algae. Pp. 131-240 in R. J. MacIntyre (editor), Monographs in Evolutionary Biology: Molecular Evolutionary Genetics. Plenum Press, p York. 985c. Comparative organization of chloro- plast pia Ann. Rev. Genet. 19: 325-354 1986a. e xd and phylogenetic relationships. 5 i ystomatics, ‘Volume I. Plants. Boca Raton, Flori b. AE a apris exui: of chloroplast DNA. Meth. Enzymol. 118: 167-186. 987. Chloroplast DNA e volution and bio- uo. uses of chloroplast t DNA. ER Amer. Naturalist 130: S6-S29. & L. A. HERBON. CRC Pre Plant mitochondrial DNA evolves rapidly in structure, but slowly in sequence. Molec. 8. & D. B. STEIN. 1986. Conservation of chlo- roplast genome structure among vascular plants. Curr. Genet. 10: 823-833. & W. F. Thompson. 1982. Chloroplast DNA rearrangements are more frequent when a large in- verted repeat sequence is lost. Cell 29: 537-550. ———— & D. Zamir. 1982. Chloroplast DNA evolution and E eco relationships in Lycopersicon. Proc. Natl. Aca 5010. ORGENSEN & W. F. beaa 1985. Chloroplast DNA variation and evolution in Pisum: patterns of change and phylogenetic erai Ge- netics 109: du 3. , B. Osorio € W. F. THompson. Evolutionary significance oti inversions in legume chloroplast DNAs. Curr. Genet. (in E RICH & W. F. THOMPSON. 1987. Chloroplast DNA evolution among legumes: loss of a large inverted repeat occurred prior to other sequence rearrangements. Curr. Genet. 11: 275- 286. ————, C. R. SuIELps, D. B. Conen & T. J. ORTON. 1983. Chloroplast DNA evolution and the origin of amphidiploid Brassica. Theor. Appl. Genet. 65: 181- 189 PERL-TREVES, R. € E. Gatun. 1985. The Cucumis plastome: al map, intrageneric variation and oo relationships. Theor. Appl. Genet. 71: d hi L. 1985. Restriction map and clone bank of tomato plastid DNA. Curr. Genet. 10: 147-152 PICHERSKY, BERNATZKY, S. D. TANKsLEY & A. R. CASHMORE. 198 Eviden for selection as a n of Lycoper- sicon esculentum (tomato) genes encoding the small ne of ribulose-1,5- pror carboxylase /ox- nase. Proc. Natl. Acad. U.S.A. 83: 3880-3884. PRAKASH, S. axonomy, cyto- genetics pu origin of crop Brassica, a review. Opera Bot. 55: 1-57. PROBER, J. M.G . L. TRAINOR, R. J. Dam, F. W. HOBBS, SE ROBERTSON, R. J. ZaGURSKY, A. J. Cocuzza, . A. JENSEN & K. unstable 1987. A system A rapid DNA sequencing w orescent chain- terminating f iris S Science 238: 336- 341. H. WEIL. QuicLEY, F. & J. 1985. Organization and dence for gene rearrangements during the evolution of chloroplast genomes. Curr. Genet. 9: 495-503. Raven, P. H. & D. I. AXELROD. Angiosperm uo and past wen movements. Ann. ri Bot. Gard. 61: 539-673. REED, K. °C. & D. A. Mann. 1985. Rapid transfer of NA from agarose gels to nylon membranes. Nucleic Acids Res. 13: 7207-7221. isa L., D. Sottis & J. D. aa 1988. A molecular reexamination of introgression between Helianthus annuus and H. bolanderi (Composite Evolution 42: 227-238. RiTLAND, K. & M. T. CLecc. 1987. Evolutionary anal- ysis of plant DNA sequences. Amer. Naturalist 130: 100. Volume 75, Number 4 Palmer et al. 1205 1988 Chloroplast DNA Variation p H. & R. D. BRETTELL. 1973. Tribal revisions L. D. GorrLiEB. 1986a. Chloroplast n the Asteraceae. III. A new tribe, Liabeae. Phy- - 107. ER. 25: 104- RocHarx, J. D. 1985. Genetic organization of the chlo- roplast. Int. Rev. Cytol. 93: 57-91. RopMaN, J. E., M. vER, R. R. NAKAMURA, J. U. McCLAMMER, Jr. & A. H. BLEDSOE. 1984. A taxo- spermae. T Bot. 9: 297-323. A. J. BENDICH. 1985. Extraction of DNA from eei amounts of fresh, herbarium and mummified plant tissues. Plant Molec. Biol. 5: -76. SacHar-Manoor, M. A., K. M. SoLIMAN, R. A. JORGENSEN & R. W. ALLARD. 1984. Ribosomal DNA spacer- length polymorphisms in barley: Mendelian inheri- = tRNAala genes and extensive secondary struc- ure. Plant Molec. Biol. 6: 265-270. Eo B. B. 1980. The elimination of plastids "ning spermatogenesis and fertilization in the plant king A, T. W AKASUGI, N. Z " cleotide sequence of tobacco chloroplast genome: its gene organization and expression. EMBO J. 5: 2043- 2049. SINGH, G., G. DE LANVERSIN, N. A. Straus & D. T. N. PILLAY. 984. Characterization of soybean chlo- roplast ribosomal RNA genes. Curr. Genet. 9: 99- 105 SMALL, J. 1918. The origin and era of iu Compositae. IV. The corolla. New Phytol. 1 40. SMITH, G. E. & M. D. SUMMERS. 1980. The bidirec- tional transfer of DNA and RNA to nitrocellulose or oe L. Anal. Biochem. 109: SOLTIS, D. E. P. S. SoLris & B. D. Ness. Chloroplast DNA variation and multiple origins of autopolyploidy in Heuchera micrantha (Saxifragaceae). Evolution (in press). STEIN, D. B., J. D. PALMER & W. F. THoMPsON. 1986. Structural evolution and flip-flop recombination of chloroplast DNA in the fern genus Osmunda. Curr. enet. 10: 835-841. Strauss, S. H., J. D. Parmer, G. T. Howe & A. H. OERKSEN. 1988. Chloroplast genomes of two co- nifers ac al rearranged. Proc. Natl. Acad. . 85: 3898- SUGIURA, M., K. SHINOZAKI, N. Zarra, M. Kusupa & M. Kumano. 1986. Clone bank of the a (Nicotiana tabacum) chloroplast genome as a set of rlapping restriction endonuclease fragments: mapping of eleven ribosomal protein genes. Pl. Sci. 4: 211-216. 98- WiLsoN, A. C., S. S. CARLSON & T. J. WHITE. Sytsma, K. J. & DNA evolution and phylogenetic relationships in Clarkia sect. Peripetasma (Onagraceae). Evolution 40: A . 1986b. Chloroplast DNA evidence for 2 origin of the genus Heterogaura from a species of Clarkia (Onagraceae). Proc. Natl. Acad. U.S.A. 83: 5554-5557. B. A. SCHAAL. 1985. Phylogenetics of: the Lisianthi Lus skinneri ( in Panama utilizing DNA restriction Sii: y is sis. Evolution 39: 594-608. TAKHTAJAN, A. L. 1980. Outline of the classification of flow Mies plants tp sma al Bot. Rev. (Lan caster) 4 25- TERACHI, T., Y ous K. TsuNEWAKI. 1987. The molecular basis of genetic o among cyto- plasms of Triticum. and Aegilo I. Complete nu- e rbcL dones encoding H-and L- sis Rubisco d. subunits in common wheat and Ae. crassa 4x. Jap. J. Genet. 62: 375-387 THORNE, R 1983. pre new ae in the giosperms. Nordic J. Bot. 3: 85- TURNER, B. L. 1977. Fossil e and pee Ñ Pp. 21-39 in V. H. Heywood, J e & B. L. Turner (editors), The Biology and Chemistry of the Compositae, Volume 1. Academic Press, New York. U, N. 1935. Genomic analysis in Brassica with special reference to the experimental formation of B. napus and peculiar mode of fertilization. Jap. J. Bot. 7: 5 UprnoLr, W. B. 1977. Estimation of DNA sequence divergence from comparison of restriction endonu- clease digests. Nucleic Acids Res. 4: 1257-1265. VAUGHAN, J. G. 1977. A multidisciplinary study of the taxonomy and origin of Brassica crops. BioScience : 35-40. VEDEL, F., F. QUETIER € M. BavEN. 1976. Specific cleavage of chloroplast DNA from higher Porn by EcoRI restriction nuclease. Nature 263: 440-442. Wac p G. 1976. Systematics and Pipes of Sron positae (Asteraceae). Pl. Syst. Evol. 125 WAGNER, D. B., G. R. FURNIER, M. A. SAGHAI-MAROOF, . M. WILLIAN 1s, B. P. Dancik & R. W. ALLARD 1987. Chloroplast DNA polymorphisms in p and jack pines and their hybrids. Proc. Natl. U.S.A. 84: 2097- W ATROUS Q. W ER. 1981. The out-group comparison method of uer analysis. Syst. Zool. 30: 1-11. WATSON, r: C. & S. J. Sunzvcki.. 1982. Extensive sequence e: in the DNA coding for elongation factor Tu from Escherichia coli and the Chlam- ydomonas reinharti chloroplast. Proc. Natl. Acad. U.S.A. 79: 67. Warson, L., H. 1. ji 0d & M. J. Dattwirz. 1985. The classification of Poaceae: subfamilies and su- pertribes. os ral. J. Bot. 33: 433-484. WHATLEY, J. 1983. Pistüdi—pan, present, and future. dum Rev. Cytol. Suppl. 14: 329-374. WhriTE, E. E. 1986. A method for extraction of chlo- roplast DNA from conifers. Pl. Molec. Biol. Rep. 4: 101. 1977 1206 Annals of th Missouri EDI Garden Biochemical evolution. Annual Rev. 73-639. bip e R.P. 1 cation and classification of plants. II. Barnadesia. 449-462. P. M. SHARP. 1987. Rates of nucleotide o vary greatly among Asia mitochondrial, chloroplast, and nuclear DNAs. Pro Natl. Acad. U.S.A ; Wu, N. H., J. C. Core € R. Wu. 1987. Structure of the chloroplast psbA gene encoding the QB protein Genet. 8: 339-350. 1986. Peculiar features of rRNA genes of the en ME DNA. Nu- cleic Acids Res. 14: 3827 . Dove, V. L. s. C. LEMIEUX, M. . A. NARANG. 1986. Complete nu- cleotide sequence and mRNA-mapping of the large subunit of ribulose-1,5-bisphosphate carboxylase /ox- ygenase a dud Chlamydomonas moewusii. Gene 50: Biochem. 46: 928. Pollen grains in the identi- YosHiNAGA, K., T. Oura, Y. Suzuki & M. SUGIURA. 1988. Chlorella chloroplast DNA sequence con- taining a gene for the large subunit of ribulose-1,5- bisphosphate carboxylase-oxygenase and a part of a possible gene for the beta' subunit of RNA polymer- ase. Pl. Molec. dye 10: 245-250. ZHANG, H., R. SchoLL, J. Browse & C. SOMERVILLE. 1988. Double Bane E DNA sequencing as a choice for DNA sequencing. s € Res. 16: 1220. ZuRAWskKI, G. € M. T. CLEGG. . Evolution of higher plant chloroplast DNA- encoded genes: impli- cations for structure- function and Peete stud- ies. Ann. Rev. i W. BOTTOMLEY . R. NOU. 1984. Junctions of the large single copy region and the inverted repeats in Spinacia oleracea and Nicotiana debneyi chloroplast DNA: sequence of the genes for tRNA" and the ribosomal proteins S19 and Nucleic Acids Res. 12: 6547-6558. RIBOSOMAL DNA VARIATION WITHIN AND AMONG PLANT POPULATIONS! Barbara A. Schaal? and Gerald H. Learn, Jr.?? ABSTRACT DNA sequences that code for the 17S, 5.8S, and 25S subunits of the ribosome have e bei be in th The study of plant evolutionary biology . The coding sequences are egos: = conservative e pr r iwidad ve also elucidated kishi level tion occurs within studies indicate that rDNA provides a good genetic marker for the study of microevolutionary processes. Many of the current questions in population biology center on the levels of genetic variability within populations and on the factors that influence genetic variation. Levels of genetic variability are central to the study of population biology and evo- lution because the amount of variability directly influences the evolutionary potential of populations and species. Much attention has been given to the problems of measuring genetic variation for dif- ferent features, such as morphology, life-history traits, chromosomes, and various types of mole- cules (e.g., Lewontin, 1 In recent years, a ise technique. for examining genetic variation has bee trophoresis. This technique has ever E our understanding of the genetic processes that occur in plant species, and without it we would have information on the genetic structure of only a handful of noncultivated species. In spite of its usefulness and widespread application, there are some well-known limitations. Most frequently, only genes of a single class, those encoding soluble en- zymes, are analyzed, and they often are selected on the basis of the ease of their products extraction and ability to migrate on a starch gel. Only nu- cleotide differences in genes that lead to changes in product amino acid composition can be detected, and then detection is usually limited to those changes in amino acid composition that result in a net change in charge of the product molecule. These genes may not be representative of the genome in general. For instance, there is evidence that many commonly studied allozymes are more variable than other categories of gene products, and that this variability may stem in part from such processes as post-translational modification (Johnson, 1979). Moreover, changes in allele migration were thought to result from single codon changes; in fact, dif- ferences may be the result of several changes in DNA structure (Sachs et al., 1986). Many of these concerns can be avoided if DNA that encodes dif- ferent types of genes is analyzed ant DNA is relatively simple to extract and purify, and DNA representing different portions of the genome can be studied by hybridization to cloned probes. In addition, current techniques of DNA analysis are many times more sensitive to changes in gene structure than are other macro- molecular assays. Variation in fine structure can be detected at several levels: in nucleotide se- quence, in sequence length, and in gene copy num- ber. Finally, straightforward and require only small amounts of tissue. It is thus feasible to analyze the large num- bers of individuals required for populational studies. current DNA technologies are The use of restriction site and sequence data offers ! This work was miaii by NSF grants DEB 82-07020 and BSR 8501215. We thank an anonymous reviewer for commen ? Department e Dialog Washington University, St. Louis, Missouri 63130, U.S ? Present address: Department of Botany and Plant Sciences, University of Eus Riverside, California 92521, U.S.A. ANN. Missouni Bor. Garb. 75: 1207-1216. 1988. 1208 Annals of the Missouri Botanical Garden the potential to reexamine once-problematic as- pects of population biology. Accurate measures of genetic variation in specific portions of the genome, determination of the amount of genetic change associated with speciation, and assessment of the amount of somatic variation and its influence on population variation are examples of areas where DNA data can provide new insights. Populational analysis offers new insights into molecular biology as well, for example, in the study of concerted evolution. Concerted evolution takes place when the members of a multigene family are more similar to each other than expected had they evolved independently from the time of the initial gene duplication that gave rise to the multigene family (Zimmer et al., 1980; Arnheim, 1983). If concerted evolution were not occurring, each in- dividual would have a large amount of variation among the copies of a multigene family; virtually none of the copies within an individual would be the same. Early hybridization studies and subsequent populational analysis indicated that this is clearly not the case for ribosomal DNA (rDNA). Most of the rDNA copies within an individual are much more similar than would be expected had they evolved independently of one another (a lim- ited amount of variation is seen within individuals, see below, but most rDNA repeats in an individual contain very similar sequences). It is thought that gene conversion, unequal crossingover, or a com- bination of these are likely responsible for con- certed evolution of ribosomal genes. The balance of the processes governing both concerted evolu- tion and the turnover of multiple copy DNA families (Dover, 1982, 1987; Dover & Flavell, 1984) and their interaction with population level phenomena (e.g., gene flow, genetic drift, and organismal se- lection) are complex and require much additional study. The pattern of variation in specific DNA sequences within and among populations will pro- vide information on the pattern of molecular changes among rDNA repeats. In the following we will examine variation in DNA sequences within and among populations of plant species. Our purpose is two-fold: to quantify the type and levels of variation at specific DNA sequences and to relate levels of variation to pop- ulational features. We will concentrate on one spe- cific DNA sequence, ribosomal DNA, that has re- ceived a great deal of attention from molecular biologists and, most recently, population biologists. Before turning our attention to ribosomal DNA variability, we will discuss the structural and func- tional aspects of rDNA that affect levels of vari- ation STRUCTURE AND FUNCTIONAL ASPECTS OF rDNA Ribosomal DNA is a mid-repetitive sequence with from 500 to more than 40,000 copies per genome arranged in tandem repeats (Long & Da- wid, 1980; Rogers & Bendich, 1987a). Ribosomal gene repeating units are composed of a number of regions that vary in functional constraint and, con- sequently, in evolutionary rate (for a review see Gerbi, 1986). Figure 1 shows the segments of the rDNA that will be discussed here. Each rDNA repeat contains a transcription unit (a through f), from which the rRNA precursor (pre-rRNA) is transcribed, and a so-called nontranscribed (or in- tergenic) spacer (g) between the transcription units of adjacent repeats. The pre-rRNA is cleaved after transcription into the mature rRNAs: the 17S (b), 5.8S (d), and 25S (f). The sequences of the rDNA that correspond to the mature rRNAs are the cod- ing regions. The 5' leader sequence (a) is the ex- ternal transcribed spacer (ETS). The internal tran- scribed spacers separate the 17S, 5.85, and 25S RNA coding sequences (ITS-1 [c] and ITS-2 [e], respectively). The pre-rRNA is transcribed and processed into the various rRNAs. Ribosome sub- units are assembled from these gene products along with the 5S rRNA and the ribosomal proteins. The different evolutionary rates observed for the various regions of rDNA is a likely reflection of the differences in the functional constraints that govern these regions. Portions of the coding regions have a high degree of evolutionary conservation, being invariant in all organisms examined to date. The nontranscribed spacer (NTS), on the other hand, diverges among closely related taxa. Other regions of the repeat show a range of intermediate rates. The rates of divergence determine a specific sequence's utility for studying the variation among populations or higher taxonomic levels. We will review the functions of the various coding and noncoding regions and then discuss the sorts of variation seen for portions of the rDNA repeats in plants CODING REGIONS The coding regions, segments of the rDNA re- peat ultimately incorporated into the cytoplasmic ribosomes, are expected to vary the least. Although this is true as a rule, limited variability is possible because of a range of functional constraints. Se- lection appears to act to conserve functionally im- portant secondary structure (Wheeler & Honey- cutt, 1988). Higher amounts of variation at the sequence level are seen among closely related taxa Volume 75, Number 4 1988 Schaal & Learn Ribosomal DNA Variation 1209 NNSNNNNNNNNNNNNNNNSN. a b NN NNN cde FIGURE 1. rDNA repeat unit organization. The line corresponds (ITS-1, s: and ITS-2, and 17S regio TS, a) and interna by the adjacent ETS (a') for the portions of the rRNAs constituting the he- lical ‘stem’ portions of the molecule. Changes may occur by compensating substitutions, in which base- paired nucleotides in opposite strands of the helix change ‘in response’ to one another. A degree of mismatch is apparently tolerated. Presumably oc- casional mismatches may slow down the rate of formation of stem and loop structures without pre- venting the formation of the helix. In addition to guanine-cytosine and adenine-uracil pairing, pair- ing between guanine and uracil is possible and does not inhibit helix formation. Furthermore, even short stretches of one or a few nucleotides of mismatch o not prevent formation of helical structure as long as they are flanked by regions of base com- plementarity. The most conserved sequences are in single-stranded regions (Wheeler & Honeycutt, 1988). These sequences either act enzymatically or bind to proteins (either ribosomal proteins or protein translational cofactors) or other RNAs (tRNAs and mRNAs). Some stretches of rRNA sequence do not vary in any organism for which the sequence is known. At the other extreme, portions of the large ribo- (Gonzalez e et ab 1985) or present in some species of a genus but are lacking in others (Chan et al., 1983; Hadjiolov et al., 1984 Ithough the evo- lutionary dynamics of these changes is not under- stood, obviously there is polymorphism within pop- ulations for rRNA coding region variants. Most studies of rDNA variation analyze restriction en- donuclease sites, and little variation has been re- ported for sites within coding regions. Sequence analysis would be much more sensitive for detecting variation, but with the exception of Gonzalez et al. (1985), most rDNA sequences are known for only a single gene copy of each species. The expected level of variation is too low to justify such a cur- rently expensive and labor-intensive survey. TRANSCRIBED SPACER REGIONS The transcribed spacer regions are the portions of the rRNA transcription unit that are not seen as mature rRNAs. of variability in interspecific studies (Appels & Dvo- ey show intermediate levels RRR f g à to the nontranscribed spacer region ( g), . The repeating structure of these genes is represented rak, 1982b; Sytsma & Schaal, 1985; Hillis & Davis, 1986; Davis, 1986), an observation con- sistent with the intermediate E d of functional hile some portions of the transcribed spacers may act merely as spacer DNA with the length of the sequence being more important than its information content, analysis of ITS sequences shows substantial conservation among moderately closely related species. Presum- ably this reflects the presence of processing signals, for which a degree of conservation is expected. Sequence conservation is also seen for portions of the ETS, again presumably due to processing sig- nals in this region. [n addition, it has been pos- tulated that the intermediate level of conservation in transcribed spacer regions may reflect RNA- mediated gene conversion; Appels & Dvorak (1982b) suggested that the conservation seen for rDNA may be due in part to ‘correction’ from the rRNA transcript of differences among ribosomal gene copies within the same nucleolus. constraint on t NONTRANSCRIBED SPACERS This region between adjacent transcription units is in fact transcribed to a degree, so it is becoming apparent that this term is a misnomer. Transcrip- tion proceeds from the 5' transcription unit, through the spacer, to the initiation site of the adjacent repeat; these transcripts are rapidly processed to the rRNA precursor and the ephemeral nature of NTS transcripts led to this region being so-named. S is the most rapidly evolving portion of the rDNA. Since it shows the greatest amount of variation within and among plant populations, it is the region most useful as a genetic marker for analyzing microevolutionary processes. The dy- namics of molecular evolution in the nontran- scribed spacers in plants have begun to be eluci- dated recently. From sequencing studies (Appels & Dvorak, 1982a; Yakura et al., 1984; Lassner ., 1986; Toloczy- ; Rogers et al., 1986), it appears that the NTS consists of at least three regions that may differ in function and may evolve at different rates. 1210 Annals of the Missouri Botanical Garden A series of subrepeating elements in the non- transcribed spacers is seen in all higher eukaryotes for which the sequence is known. In addition, the presence of subrepeats in a number of plant species is inferred from length variability in the NTS (e.g., Cluster et al., 1984, reviewed by Rogers & Ben- dich, 19872). A degree of sequence similarity has been demonstrated between NTS subrepeats of wheat (Appels & Dvorak, 1982a) and maize (McMullen et al., 1986; Toloczyki & Feix, 1986). This presumed conservation has been interpreted as evidence that the subrepeats have a function. It has been demonstrated that one class of sub- repeating elements acts as enhancers of transcrip- tion in Xenopus (Reeder et al., 1983; Reeder, 1984), and evidence suggests that some types of subrepeats function similarly in plants (Flavell & O’ Dell, 1979; Martini et al., 1982; Flavell, 1986). No function has been demonstrated for the re- gion of the NTS 5' to the presumed enhancer subrepeats. This sequence is 144 base pairs in maize (Toloczyki & Feix, 1986) and at least 241 bp in wheat (Lassner & Dvorak, 1986) but may The region 3' to the enhancer subrepeats is 135 to 240 bp in maize (McMullen et al., 1986; Toloczyki & Feix, 1986); in wheat it is considerably less than 960 bp (Lassner & Dvorak, 1986). This region is assumed to contain the promoter for transcription of the pre-rRNA. Although the NTS is presumed to code for no gene products, there is good evidence for functional constraints, and the NTS is therefore potentially subject to selectional forces. It is not be considerably longer in other plant taxa. clear how strongly these constraints govern the evolution of the NTS region of the rDNA multigene family, but they clearly differ from those governing evolution of the coding regions (see Jorgensen, this volume, for further discussion). VARIATION AMONG INDIVIDUALS OF A POPULATION Because of these differences in the levels and kinds of functional constraints, variation of rDN is very different for the transcription unit versus the so-called nontranscribed spacer. In general, within a species there appears to be only little variation in the coding regions. Such variation ap- pears to be predominantly developmental variation due to methylation (see Jorgensen & Cluster, this volume). We will concern ourselves here with vari- ation in the nontranscribed spacer region. When the individual plants within a population show some type of rDNA variation, it is within this region. Length variation is most common and is due to a series of repetitive elements in the nontranscribed spacer region. In Triticum spp. and Vicia faba, length heterogeneity is due to copy number dif- ferences of a series of 135-bp or 325-bp elements, respectively (Appels & Dvorak, 1982a; Yakura et 19 A variable number of copies of the same or highly similar DNA sequence gives rise to the different length classes. Because rDNA is a repetitive DNA sequence within the genome, individuals can contain several different length variants. A single V. can have up to 20 different length variants of rDNA 1986). Native populations of Lu- pinus texensis contain plants with up to 11 length faba plant (Rogers et al., variants, although most commonly there are three or four variants per plant (D. Baum, pers. comm.). In Phlox divaricata, the mean number of repeats per plant is 1.98 (Schaal et al., 1987). Clematis fremontii has an average of 2.65 variants per individual (Learn & Schaal, 1987), whereas Hor- deum spontaneum contains on average 2.28 vari- ants per plant (Saghai-Maroof et al., 1984). Such length variation is not ubiquitous. Solidago altissima is highly variable in the nontranscribed spacer region but this variability is limited almost exclusively to restriction site variation (Schaal et al., in prep.). Length variation is restricted to a 300-bp insertion present in low frequency within some populations. Table 1 shows the variation in restriction sites of S. altissima. Variation of rDNA occurs often within individual plants. Plants are most commonly polymorphic for rDNA variants that have sites. Another feature of S. altissima rDNA is the genetic differentiation among portions of a clone. Several plants showed variation in rDNA types within a clone for variants based on different EcoRI or EcoRV restriction sites. Such within-clone differences may occur via somatic mutation or rapid increase of a rare vari- ant. The occurrence of variation within plants adds a level of analysis not previously possible in pop- ulational studies. Other species show no rDNA variability. Rud- beckia missouriensis, an endemic of isolated rocky habitats in Missouri and Arkansas, shows no length heterogeneity nor does it show restriction site vari- ation in a survey of six populations of glade habitats L. King, pers. comm.). Gaura demareei, a hybrid- derivative species with a narrow range, contains two length variants of 10.5 and 11.3 kb. Each plant examined in a survey throughout its range was identical for the two length variants (Schaal — & Raven, in prep.). Moreover, there was no re- striction site variation. Similarly, species of the Volume 75, Number 4 1988 Schaal & Learn 1211 Ribosomal DNA Variation Lisianthius skinneri complex show little restriction site variation or length heterogeneity (Sytsma & Schaal, 1985). t this time no clear correlations emerge be- tween levels of rDNA length variability and char- acteristics of the population biology of various species. In the three cases where no variation is observed, the species are narrowly endemic. The R. missouriensis populations are isolated, although population size can be in the thousands. Gaura demareei has a highly restricted species range, occurring predominantly within a single Arkansas county in populations often fewer than 50 individ- uals. Likewise, Lisianthius species have a very narrow range and often consist of few populations with low plant numbers. On theoretical grounds one expects that narrowly endemic species would have little variation, due to genetic bottlenecks. Variation would be expected to be lost due to sam- pling, in these cases either by small population size, by founder events due to repeated colonizations, or by a recent species origin after hybridization. The generation of length heterogeneity may not occur very rapidly in these species since none of them have accumulated variation; even A. mis- souriensis, where number of individuals per pop- ulation is high, remains depauperate for rDNA length variation. In marked contrast to the results obtained for R. missouriensis are the levels of variation de- tected in Clematis fremontii. Populations of Cle- matis fremontii, like the Rudbeckia, occur on islandlike glades in Missouri. These populations are predominantly limited to two counties and are often much smaller in population number than R. mis- souriensis. On theoretical grounds one would ex- ect C. fremontii to show less variation than R. missouriensis, but this is not the case. The Cle- matis plants contain up to four length variants, and the length variation shows apana diferentia: tion (see below). Thus, there is n attern here between ribosomal DNA variation E? pop- ulation size or species range (but see Flavell et al., 1986). Since the two species with the greatest observed length heterogeneity are both legumes (Vicia faba and Lupinus texensis, see above), one might suspect that something in the ancestry or in the biology of legume species leads to such high numbers of rDNA length variants, but length het- erogeneity is not great in some other legume species (soybean and its relatives, Doyle & Beachy, 1985; other Vicia species, Rogers & Bendich, 1987b). Clearly, associations between levels of rDNA length vari- uch more research is necessary before TABLE 1. Variation of rDNA in Solidago altissima. Be- E ud ithin tween Individ- Within Individ- red Site uals! Clones? uals tions Sstl + _ — — Dell + — = = Bell + _ = + Xmnl + = + = Hincll + — — — hol + - am BamHI + = — — EcoRI T t = - EcoRV + + + Pa Xmnl + = + + HindIII + — — + nsert + = + st ! Genome contains more than one rDNA variant. ? Differences in rDNA type occur among the parts of a clone. ation and populational characteristics, such as gene flow, population size, breeding system, or founder events can be established. A further aspect of rDNA variation within pop- ulations is apportionment and distribution in space. One feature that distinguishes many plant popu- lations from most animal populations is the frequent occurrence of genetic population substructure in the former. Plant populations often show significant local genetic differentiation, many times on a mi- crogeographic scale. Such local differentiation can result from selection on a very local scale. Local differentiation may also occur via genetic drift. Such drift can be the consequence merely of n random mating due to restricted pollinator behavior the subdivisions, us leading to significant local genetic heterogeneity (Turner et al., 2) ocal differentiation within plant populations has been documented for genes that cause heavy metal tolerance (Jain & Bradshaw, 1966), that result in different flowering times (McNeilly & Antonovics, 1968), that cause morphological differences (Lin- hart, 1974), and that encode different allozymes (Schaal, 1975). Our study of Clematis fremontii has docu- mented nonrandom geographical distribution of ri- bosomal DNA variants within populations. A single population of Clematis fremontii has been analyzed for spatial variation in the frequency of rDNA- length variants. Many of the length variants that 1212 Annals of the issouri Volume 75, Number 4 1988 Schaal & Learn Ribosomal DNA Variation 1213 occurred in high frequency showed no significant spatial differentiation (Fig. 2A). However, two of the variants showed statistically significant micro- geographic differentiation; the variants do not tend to be distributed randomly in space within the population, but rather are confined to specific areas within it (Figs. 2B, C). Such local differentiation of rDNA variants is consistent with population sub- division due to restricted gene flow, or perhaps it may be a consequence of a recent origin or dis- persal of an rDNA variant into a population. VARIATION AMONG POPULATIONS Few studies have examined the pattern and ap- portionment of rDNA variants among the popu- lations of a species. Most of the work to date has centered on cultivated species and is reviewed in Appels & Honeycutt (1986). Here we look at levels of variation in natural, noncultivated plant species. Levels of differentiation for rDNA variants vary among populations of a species. Some plant species show no significant heterogeneity within or among populations. Those species having low levels of rDNA variation within populations show little or no differentiation among populations. No significant genetic differences in rDNA types were detected among populations of Gaura demareei, Rudbeckia missouriensis, or members of the Lisianthius skin- neri complex. Judging from their ranges and/or other determinations of genetic variability, it is likely that these species have undergone genetic bottlenecks and variation has been lost within and between populations. In the few highly variable examined species, significant genetic heterogeneity is detected among populations. The best-studied example to date is the wide- spread woodland perennial Phlox divaricata (Schaal et al., 1987), in which there is clear dif- ferentiation of rDNA variants. Populations often contain unique rDNA variants and may be distin- guished by the number of variants (2-6) they con- tain (Schaal et al., 1987). There is clear differ- entiation between population systems. Phlox divaricata subsp. laphamii shows less rDNA di- versity than subsp. divaricata. The subspecies dif- fer in the numbers and types of variants they contain, and in the overall genetic diversity (Table The variation in Phlox divaricata provides cor- roborative information on the origin of the sub- species. Based on morphological criteria, subsp. laphamii is considered derived from subsp. di- varicata. This hypothesis appears to be supported by the apportionment of rDNA variation; rDN variability in subsp. laphamii is a subset of the variability seen in the other, more widespread sub- species. Variation among populations also has been ana- lyzed in the old-field perennial Solidago altissima (Schaal et al., in prep.). Differentiation in this species occurs for a 200-bp sequence which is fixed in one population and is present in low frequencies in other populations. Populations are also differentiated for restriction sites. As with intrapopulation variation, too few species have been analyzed to draw con- clusions about levels of rDNA variation and such populational characteristics as size, gene flow, or bottlenecks. Clearly, Ppa are differentiated for levels and kinds of rDNA variation. Whether the differentiation is Ad to selection, genetic drift, gene flow, or any other population-level ge- netic process remains to be determined. Although the mechanisms responsible for generating rDNA length variation obviously require further study, such variants can and have been used to recon- struct aspects of the evolutionary history of plants see also Sytsma & Schaal, 1985; Doyle et al., 1984, 1985). — CONCLUSIONS AND PROSPECTS The use of DNA sequences in studies of popu- lation biology is in its infancy and holds a great deal of potential for understanding processes and ] š + answering persistently sequences is a major technical advance. Virtually any segment of DNA can be studied, whether it is a coding or noncoding sequence, or is single copy, — 2. i a from Haran y (Fig. 2B) and E A. Variant C, 113 kilobase pairs (Kb). Num alon, The left o rdinate and the histogram bars right ordinate and < a e are proportion a pi 5 is the most common ant in the population A (10.2 kb). Axes as in Hue mean freq LUE within a quadrat bearing a var oes not show significant local differentiation —B. Var 2A.—C. Varad E (11.9 kb). Axes as in Figure Frequencies of some rDNA repeat length variants in Clematis T Individuals were sampled O-m transect at Victoria Glade, Jefferson County, Miss (Fig. 2C) Vade REN ‘significant local differentiation along the tra ri (see Learn & Schaal (1987) for details) . nsect.— 1214 Annals of the Missouri Botanical Garden TABLE 2. Population variation in rDNA repeat-type frequency of Phlox divaricata. Repeat-Type Frequency Subspecies and Population V-1 V.2 V-3 V-4 V-5 V-6 V-7 V.8 Subspecies laphamii CC 0.19 0.50 0.31 T 0.09 0.27 0.55 WH 0.18 0.55 0.42 0.25 0.08 Results for subspecies overall 0.08 0.38 0.04 0.24 0.10 0.02 Subspecies divaricata TC 0.29 0.14 0.21 PH 0.50 0.20 0.30 B 0.19 0.19 0.10 0.19 0.21 0.08 0.13 0.12 0.17 Results for subspecies overall 0.11 0.13 0.16 0.19 0.06 0.10 Results for all populations 0.070 0.078 0.031 0.239 0.141 0.132 0.10 0.003 mid-repetitive, or highly repetitive DNA. In fact, one strategy of population analysis is to clone ran- dom portions of the genome, and study variation of restriction sites in these random sequences (e.g., Hofker et al., 1986). From the studies discussed above, some DNA sequences, specifically segments of rDNA, vary at the appropriate levels for studies of population pro- cesses. This is in contrast to other sequences, such as chloroplast DNA, where variation is usually seen at the interspecific or intergeneric level, and is most informative for phylogenetic studies. There are clear differences among individuals and populations in added dimension to studying variation occurs with the use of mid-repetitive sequences, since single individuals contain many copies of a sequence and thus can themselves be polymorphic. Studies of ribosomal DNA provide an additional level of analysis, that of the individual; the appor- tionment of variation can be examined at the with- in-plant as well as the between-plant levels. Another potentially important aspect of rDNA studies is the ability to detect somatic mutations. There currently is much speculation in the liter- ature on the role of somatic mutation in plant population biology (Whitam & Slobodchikoff, 1981; ill & Halverson, 1984; Walbot, 1985; Walbot & Cullis, 1985). Several workers suggest that so- matic mutation and subsequent variation leads to differences in ecological parameters, such as sus- ceptibility to insect predation. It is argued that somatic variation in DNA sequences may have an adaptive function. These ideas are contested, in part because the frequency of somatic variation is not known. With current methods of analyzing DNA sequences, it is possible to document un- equivocally the occurrence and frequency of so- matic variation nal new area where DNA analysis is poten- tially important for population biology is rapid ge- nomic change. Many organisms alter their DNA in response to stress, as in the case of gene am- plification in response to toxic agents (Schimke, 1983). McClintock (1984) suggested that genome change is a way in which plants routinely deal with stress. Walbot & Cullis (1985) suggested that flex- ibility is an important feature of the plant genome. Genome flexibility has been demonstrated in flax, where heritable variation in rDNA cistron number is induced by environmental changes (Cullis, 1986). Such heritable changes in genome size have pro- found implications for population biology. Altera- tion of genomes in response to environmental vari- ation may contribute to the genetic adaptation of a plant species. Such a process can alter the genetic characteristics of populations and, on a practical level, can confound such experiments as reciprocal transplants. There is much future work on the interface between plant molecular biology, popu- lation genetics, and ecology; blending among these disciplines promises to add greatly to our under- standing of plant evolution. LITERATURE CITED APPELS, R. & J. Ta ee bs eps inea NA spacer region: i and v on in populations s am s pines Theor, Appl rer 63: 337-348. . 1982b. Relative rates of diver- gence of spacer and g within the rDNA Volume 75, Number 4 1988 Schaal & Learn 1215 Ribosomal DNA Variation TABLE 2. Continued. Mean Number of Repeat-Type Frequency Repaqta/ V-9 V-10 V-11 V-12 Plant 2.0 0.09 1.2 0.18 0.09 1.2 0.25 1.5 0.12 0.02 1.5 0.15 0.21 2:3 2.0 0.04 0.30 2.6 .12 0.17 3.0 0.03 0.05 0.17 2.5 0.051 0.016 0.031 0.108 1.98 region of species in the Triticeae: implications for the maintenance of homogeneity of a repeated gene family. Theor. Appl. Genet. 63: 361-365. & R. L. Honeycutr. 1986. is evolution over a billion years. Pp. 81-136 i utta (editor), DNA Systematics, Volume IL Plant DNA. Press, Boca Raton, Florida. 19 Concerted evolution of multigene families. Pp. 38-61 in M. Nei K. Koehn (editors), Evolution of > and Proteins. Sinauer, Sunderland, “sea use CHAN, Y.-L., J. AGI. Woo 1983. The struc- ture of rat 28s Da buie acid inferred from the sequence of nucleotides in a gene. Nucl. Acids Res. 11: 7819-7831. , R. A. JoRGENSEN, R. BERNATSKY, A. ALLARD. 1984. The ge- netics and geographical distribution of ribosomal DNA ild oat Avena bar- bata. Genetics 107: s21. [Abstract.] Plant DNA variation and stress. adler Genetics Symposium 17: 55 Davis, S. K. 19 opulation Structure and Patterns of Speciation in Geomys Analysis Using Mitochondrial a Ph.D. Dissertation. Washington University, St. Louis, Missouri. Dover, G. A. 1982. Molecular drive: a er model of species evolution. Nature 299: 11 987. DNAt ia and the coe clock. J. Molec. Evol. 26: R. B. iat eon Molecular pak n: DNA 2 n = the maintenance of fun tion. Cell 38: 622 Doy_e, J. J. & R. N. Ben AC HY. 1985. Ribosomal = variation in Theo: Appl. Genet. 70: 369-376. S W. H. Lewis. 1984. Evolution mplex Po. 21- of rDNA i in Claytonia polyploid 341 in W. F. Grant (edor) Plant Biosystematics. Academic Press, Toront —., D.E. So LTIS & P. "s. SOLTIS. 1985. An in- Zw tergeneri e the ribosomal RNA genes. Amer. J. E 72: 1388- Neid R. B. 1986. TY RNA genes and control of their expression. Pp. 251-276 in B. J. Miflin (editor), Oxford Surveys of Plant Molecular arid Cell Biology, Volume 3. Oxford Univ. Press, Oxford, En- gland. O'DELL. 1979. The genetic control of nucleolus formation in wheat. Chromosoma 71: 135- 152 — ———. P. SHARP, E. Nevo € A. BELES. 1986. Variation in the eae spacer of ribo- somal DNA of wild wheat, as dicoccoides, in Israel. Molec. Biol. Evol. 3: 547-558. GERBI, S. A. The uin of eukaryotic ribo- somal DNA. BioSystems 19: 247-258. GILL, D. E. & T. G. HALVERSON. 1984. Fitness variants among branches within trees. Pp. 105-11 od Siri Evolutionary Ecology. Blackwell Scientific, Oxfor GONZALEZ, I i Gorski, T. J. CAMPEN, D. J. Dorney, J. M ERICKSON, J. E. SYLVESTER & R. D SCHMICKEL 5. Variation among human 28S ribosomal RNA genes. Proc. Nat. Acad. U.S.A. 82: -7670. HADJIOLov, A. A., O. I. GeonciEv, V. V. Nosikov & L. . YAVACHEV. 1984. Primary and secondary struc- ture of rat 28S ribosomal RNA. Nucl. Acids Res. E Davis. 1986. Evolution of ribosomal DNA: fifty million year ded histor in the frog genus Rana. Evolution 40: 1275-1288. HorkER, M. H., M. I. SKRAASTAD, A. A. B. BERGEN, M. C. WAPENAAR, E. BAKKER, A. 1. WARD, G. J. B. VAN AMMEN & P. L. PEARSON. 1986. The chromosome shows less genetic variation at re- striction sites than the autosomes. Amer. J. H Genet. 39: 438-451. Jain, S. & A. BRADSHAW. 1966. Evolution in closely adjacent plant ils iem x piar and its the- oretical analysis. Heredity 41. inei p den orphism: genetic variation in the i sr gn EAE M Pp. 62-83 in O. Solbrig, S. Jain, G. Johnson & P. n. n (ed- itors), Topics in Plant Population Biology. Columbia Univ. Press, New York. Lassner, M. & J. Dvorak. 1986. Preferential homog- enization between adjacent and alternate subrepeats in wheat rDNA. Nucl. Acids Res. 14: 5499-5512. LEARN, G. H. & B. A. ScHAAL. 1987. Population sub- Turon for ribosomal DNA repeat variants in Cle- atis m ntii. Evolution 41: 433-438. BE ax. R. C. 1974. The Genetic Basis of Evolu- tionary Change. Columbia Univ. Press, New York pub Y. 1974. Intra- resin differentiation in nnual plants I. Veronica peregrina L. raised under non-competitive conditions. Evolution 28: 232-243. E. O. € I. B. Dawip. perg pee in euka aryotes. Ann. Rev. Biochem. 49: 72 Mc p B. 1984. The significance di jun the genome to challenge. Science 226: 792-801. McMuut LEN, M. D., B. HUNTER, PHILLIPS & I. RUBENSTEIN. 19 The structure of the maize ribosomal DNA spacer region. Nucl. Acids Res. 14: 4953-4968. MCNEILLY, T. & J. ANTONOVICS. LoNc, 1968. Evolution in 1216 Annals of the Missouri Botanical Garden closely adjacent m pla as IV. Barriers to ne flow. Her ci MS R B. Fav ELL. 1982. ice nucleolus organiser chromosomes from Aegilops um- bellulatus. Chromosoma 84: 687-700 REEDER, R. H. Enhancers and ribosomal gene spacers. Cell 38: 349-351. . G. Roan & M. Dunaway. 1983. Spacer regulation of Xenopus ribosomal e transcription: competition in oocytes. Cell 35: 6. Rocers, S. O. & . BENDICH. 1987. Ribosomal RNA genes in plants: variability in copy number and in the intergenic spacer. Pl. Molec. Biol. 9: 509- 520 — r & ———. 1987b. Heritability and variability in ribosomal RNA genes of Vicia faba. Genetics 117: -295. , S. Honpa & A. J. BenbicH. 1986. Variation in the ribosomal RNA genes among individuals of icia faba. Pl. Molec. Biol. 6: 339- se Sacus, M. M., E. S. Dennis, W. L. GERLACH & W. J. PEACOCK. 1986. Two alleles of maize p de- hydrogenase 1 have 3' structural and poly(A) addition ymorphisms. Genetics 113: 449-467. SAGHAI- Mandar M. A., K. M. SoLIMAN, R. A. JORGENSEN & ics. Proc. Natl S.A. 014- ScHAAL, B. A. 1975. Local hates nn popu- lation structure in Liatris cylindrac r. Nat- uralist 109: 511-528. , W. J. Lane H & J. Niero-SOTELO. 1987. Ribosomal DNA variation in the native Fa Phlox divaricata. Molec. Biol. Evol. 4: 611- ScHIMKE, R. T. 1983. Gene amplification i in mammals somatic cells. Pp. 235-251 in K. F. Chater, C. A. Cullis, D. A. Hopwood, A. W. B. Johnston & H. W. Wollhouse (editors), Genetic Rearrangement. The Fifth John Innes Symposium. Croom Helm, U.K Sytsma, K. J. & B. A. ScHaaL. 1985. bet dcs of the pets skinneri (Gentianaceae) species com e in Panama utilizing DNA restriction os t analysis. Evolution 39: 594-6 Tales zYKi, C. & G. Feix. 1986. Occurrence of 9 homologous repeat units in the external spacer region of a nuclear maize rRNA gene unit. Nucl. Acids Res. 14: 4969-4986. Turner, M. E., J. C. STEPHENS & W. W. ANDERSON. 1982. Homozygosity and patch structure in plant populations as a result of nearest-neighbor pollination. Proc. Natl. Acad. U.S.A. 79: 203-207. WALBOT, V. 1985. On the life strategies of plants and animals. Trends in Genet. 1: 165-169. & C Rapid genomic change ONEYCUTT. 1988. Paired sequence differe n ribosomal RNAs: evolutionary and S L asno E Molec. Biol. Evol. 5: 90-96. WurrHAM, T. G. & C. N. SLOBODCHIKOFF. 1981. Evo- lution by individuals, plant-herbivore interactions, and mosaics of genetic variability: the adaptive sig- nificance of somatic mutations in plants. Oecologia 49: 287-292. YaKURA, K., A. Karo & S. TANIFUJL. 1984. Length heterogeneity of the large spacer of Vicia faba rDNA is due to the differing number of a 325 bp repetitive i element. Molec. Gen. Genet. 193: 400- a E. A., S. L. Martin, S. M. BEVERLEY, Y. W. Kan 8 A. C. Witson. 1980. Rapid duplication and loss of genes coding for the a chains of hemo- globin. Proc. Natl. Acad. U.S.A. 77: 2158-2162. DNA AND MORPHOLOGY: COMPARISONS IN THE ONAGRACEAE! Kenneth J. Sytsma and James F. Smith? ABSTRACT Comparisons of systematic information generated from both sled systematic approaches and from DNA analysis at a number . Phylogenetic results from chloroplast a (= Peripe etasma) are not entirely congruent with results from morphology, but are congruent with on a duplications of isozyme-coding loci. Chloroplast is discussed with respect to morphological divergence between the two genera. Detailed chloroplast DNA restriction site mapping within the seven diploid sections of Clarkia and subsequent preliminary intersectional phylogenetic analysis are presented. These DNA-based relationships are compared with a morphological and cytological model of relationships, and to various gene duplication-based models. Section Godetia is implicated as the sister group to the rest of Clarkia, a result concordant with preliminary cladistic analysis of morphological and i isozymic e other sections of Fuchsia A restriction site Eb Gm iii a cytological, and isozymic variability are reviewed for the Onagraceae and other angiosperm Phylogenetic analysis of plants using molecular techniques is increasingly providing detailed and often unexpected evidence of phylogenetic rela- tionships among populations, species, sections, gen- era, and tribes (Gottlieb, 1977a, b; Gottlieb & Weeden, 1979; Odrzykoski & Gottlieb, 1984; Sytsma & Schaal, 1985a; Sytsma & Gottlieb, 1986a, b; Jansen & Palmer, 1987, 1988; Rie- , 1988; Soltis et al., in press). A major hese new molecular tech- niques—e.g., chloroplast DNA (cpDNA) restriction fragment analysis—is that they provide numerous independent characters that can be used as his- torical markers to define more rigorously the phy- logenetic relationships of the plants (see Sytsma & Gottlieb, 1986b, and Jansen & Palmer, 1988, for examples). Those molecular techniques that pro- vide but a single piece of information, such as analysis of cpDNA inversions or duplications of iso- zyme encoding genes, are still powerful, since the underlying genetic or structural bases can be clear- ly demonstrated and the essentially neutral char- acter changes can be argued to be strictly homol- ogous and rare (see Jansen & Palmer, 1987, and Gottlieb & Weeden, 1979, for examples, respec- tively). The Onagraceae provide inique opportunities for the application of these and genetic techniques, especially those involving > molecular proteins and nucleic acids. The Onagraceae are a well-defined family of seven tribes, 16 genera, and approximately 650 species (Raven, 1988). An abundant and detailed information base for the family has been generated already using morphol- ogy, anatomy, chromosomal features, and flavo- noid chemistry. Ongoing systematic studies using proteins, nucleic acids, and formal cladistic anal- yses that complement the information already available are making the Onagraceae the best-stud- ied plant family of their size (Raven, 1979, 1988). Given the large information base generated from the more “classical”” systematic approaches on On- agraceae, phylogenetic analyses using proteins and/ or nucleic acids are especially applicable to the ! Supported in part by a grant Wes the National Science Foundation (BSR-85 16573) . . H. . Stein, J. M. Affolter, K. Holsinger, and L. D. Gottlieb for their assistance; Raven, P. C. Hoch, P. = Berry, B. A om L. Taylor for artw We thank P. H. otany aent, Vuinersity of Wisconsin, Madison, Wisconsin 53706, U.S.A. ANN. MISSOURI Bor. GARD. 75: 1217-1237. 1988. 1218 Annals of the Missouri Botanical Garden study of relationships among Onagraceae. First, detailed genotype-based phylogenies can be con- structed for taxa within Onagraceae. Second, phy- logenies resulting from morphology, anatomy, and other phenotypic characters can be compared with molecular phylogenies to determine which studies are providing similar or congruent phylogenies. This will permit identification of consistently mono- phyletic lineages in these independent studies. Ad- ditionally, certain kinds of characters might be viewed with suspicion if they suggest relationships at odds with those provided by other types of char- acters. Third, incongruencies found among these independently derived phylogenies can point to fur- ther research along either of two lines: (1) a reex- amination of specific data sets or the techniques themselves to identify possible reasons for the in- congruencies (e.g., nonhomologous characters, high levels of homoplasy, rapid or uneven rates of char- acter divergence among lineages, and hybridiza- tion/introgression); and (2) reassessment of rela- tionships not previously supported or even suspected with other available information. The classic series of studies by Gottlieb and his associates using the distribution of isozyme-encod- ing gene duplications within Onagraceae (Gottlieb, 1977b; Gottlieb & Weeden, 1979; Odrzykoski & Gottlieb, 1984; Soltis et al., 1987) illustrate well how molecular techniques can be used in this fash- ion. These studies have generated phylogenetic hy- potheses, made comparisons with morphologically and cytologically based phylogenies with which they often differ substantially, questioned several models of phylogenetic relationships previously supported by classical studies, and lastly initiated several new and rewarding lines of research (e.g., genetic stud- ies of duplications and subsequent silencings, ef- fects of different isozyme number and activity in plants, and verification of progenitor /derivative species relationships). n this paper, we use evidence from cpDNA restriction fragment analysis and site mapping to produce phylogenies within Onagraceae, compare these with other molecular and morphological phy- logenetic hypotheses, reexamine a number of lin- eages that are either not supported by molecular evidence or not supported by morphological evi- dence, and finally raise questions that these cpDNA studies now permit us to ask. Previous cpDNA phylogenetic studies in Clarkia sect. Sympherica (= Peripetasma) and the genus Heterogaura (Sytsma & Gottlieb, 1986a, b) will be reviewed and additional nuclear rDNA evidence introduced. Preliminary cpDNA phylogenetic analysis of sec- tional relationships within Clarkia will be described and compared with relationships based largely on morphology, chromosome number, and crossing relationships (Lewis & Lewis, 1955), and to rela- tionships based on gene duplications (Gottlieb & Weeden, 1979; Soltis et al., 1987). Preliminary cpDNA phylogenetic analysis of sectional relation- ships within Fuchsia will be then presented and compared with those described by Berry (1982) and Raven (1979, 1988). Lastly, systematic results of DNA versus morphology will be reviewed in the Onagraceae and other angiosperms. PHYLOGENETIC ANALYSIS OF HETEROGAURA AND CLARKIA SECT. SYMPHERICA INTRODUCTION amc r most of which are restricted to California, but with C. pulchella Pursh, the type species, confined to the northwest U.S. outside California. The C. te nella polyploid complex exhibits a disjunct distri- bution in California and Argentina and Chile. The largest section of Clarkia recognized by Lewis & Lewis (1955) is sect. Sympherica, the valid name for the former sect. Peripetasma (Holsinger € Lewis, 1986). Section Sympherica is comprised of three morphologically well-defined diploid sub- sections and the tetraploid subsect. Prognatae, the CI of approximately 44 species, latter comprising only C. similis Lewis & Ernst, which is believed to be an allopolyploid derived from hybridization between subsects. Lautiflorae and Micranthae (Lewis & Lewis, 1955). Relationships within sect. Sympherica based on morphology and crossing experiments (Lewis & Lewis, 1955; Davis, 1970) are illustrated in Figure 1. Subsection Micranthae consists of one strictly self-pollinating species with small, inconspicuous, and white flowers, whereas subsects. Sympherica (three species) and Lautiflorae (four species) con- sist of primarily outcrossing species with large, showy, and colorful flowers. Petals of subsect. Lau- tiflorae are more or less uniform in color with some flecking, whereas petals of subsect. Sympherica have distinct areas of color. In addition, subsect. Lautiflorae has terete or grooved immature cap- Sympherica has deeply eight-ribbed immature capsules. Isozyme analysis challenged certain relation- ships within sect. Sympherica. Odrzykoski & Gott- lieb (1984) found that the distribution of gene duplications and subsequent silencings for isozymes of 6-phosphogluconate dehydrogenase (6PGD) in- dicated that the plastid isozymes are coded by two loci in all diploid species examined except for two sules, whereas subsect. Volume 75, Number 4 1988 Sytsma & Smith Onagraceae—DNA & Morphology 1219 ç o 48 x E * < S $ e ` | ° x ` | 4 9 $ e FIGURE 1. Relationships within Clarkia sect. Sympherica based on morphological and crossing studies ç d «v SS S | A | S 9^ S rM of Lewis a Lewis (1955) and modified afier Davis (1970). Subsections are indicated above the species abbre- viatio species in sect. Sympherica, C. epilobioides (Nutt.) Nels. & Macbr. Micranthae), and C. rostrata Davis (subsect. Sympherica), which have a single locus coding for the plastid isozymes. In addition, all species of Clarkia have a single locus coding for the cytosolic isozyme of 6PGD except for four species of sect. Sympherica— C. epilo- bioides, C. rostrata, and the other two species of (subsect. subsect. Sympherica, C. cylindrica (Jeps.) Lewis & Lewis, and C. lewisii Raven & Parnell (formerly C. bottae (Spach) Lewis & Lewis). The most parsimonious explanation for the dis- tribution of these character states as suggested by Odrzykoski & Gottlieb (1984) is illustrated in Fig- ure 2. The duplications of the two genes coding the plastid and cytosolic 6PGD isozymes are an- cestral in Clarkia and retained in the four species of sect. Sympherica subsect. Lautiflorae: C. bi- loba (Dur.) Nels. & Macbr., C. lingulata Lewis & Lewis, C. modesta Jeps., and C. dudleyana (Abrams) Macbr. The loss of one of the duplicated cytosolic 6PGDs occurred in the common ancestor of the four species of subsects. Micranthae and Sympherica. This cytosolic 6PGD loss was then followed by the loss of one of the duplicated plastid 6PGDs in the common ancestor of C. epilobioides and C. rostrata. Thus, subsect. Sympherica is paraphyletic with one species sharing a more recent common ancestor with a species from another sub- section than it does with species in its own sub- section. This isozyme-based phylogeny clearly con- tradicts the morphological model in Figure 1 in that C. rostrata is placed as the sister species to the distinctive selfer C. epilobioides rather than with C. cylindrica and C. lewisii, species it closely resembles and with which it can experimentally produce fertile hybrids. À more dramatic difference in results between the classical and molecular techniques was seen when Heterogaura heterandra (Torr.) Cov. was used as the outgroup in preliminary cpDNA anal- ysis of sectional relationships in Clarkia (see Phy- logenetic Analysis of Intersectional Relationships within Clarkia). Heterogaura is a monotypic genus closely related to Clarkia, based on floral mor- phology, stigma surface, seed coat structure, an- ther anatomy, and flavonoids (Raven, 1979, 1988; Tobe & Raven, 1985, 1986; Averett et al., 1982). Heterogaura heterandra is a strictly self-pollinat- ing annual limited to the slopes of the Sierra Nevada in California and Oregon. It differs markedly from Clarkia in having only four fertile anthers (four are sterile), an unlobed stigma, and a round nutlike indehiscent fruit with one or two seeds. In contrast, members of Clarkia generally have eight fertile anthers, four-lobed stigmas (although self-pollinat- 1220 Annals of the Missouri Botanical Garden ç <ç o Gi e > S ~ v N A e S 3 $ Y y . An š | | ` Di e $ e =P -C +P +C FIGURE 2. Relationships within Clarkia sect. Sympherica based on gene number for 6- p indicat —, resp i a i sasa bns bid a Gottlieb, 1984). Plastid (P) and cytosolic (C) duplications and losses are ed b ctively. Subsections are indicated above the ene abbreviations. hae that relationships ndis a tee. are not resolved with the isozyme dat ing species have short lobes), and elongated, many- seeded capsules that dehisce along four septa. The floral and fruit differences between the two genera are so distinctive that they have been maintained as separate genera since 1866, when H. heter- andra was first named. The long-standing idea that Heterogaura is the sister group to Clarkia was questioned when it became apparent that restriction fragment analysis of cpDNA indicated that Heterogaura was not aligning itself as a basal clade to all of Clarkia. Instead, Heterogaura exhibited synapomorphies with certain lineages within section Sympherica (Sytsma & Gottlieb, 1986a). Section Sympherica is relatively advanced in Clarkia based on mor- phology (Lewis Lewis, 1955) and on ine pres. ence of a PGI ( l i (Gottlieb & Weeden, 1979). Thus, cpDNA ul suggested that Heterogaura was not an appropri- ate outgroup for Clarkia and, more importantly, indicated that the genus might be derived more recently from within Clarkia. n extensive restriction fragment and site anal- ysis of cpDNA in Clarkia sect. Sympherica and Heterogaura heterandra was initiated to address these discrepancies between the classical results and the results from both isozyme gene duplication and cpDNA restriction fragment analysis (Sytsma & Gottlieb, 1986a, b MATERIALS AND METHODS Seeds of Heterogaura heterandra and the eight species of Clarkia sect. Sympherica were germi- nated, grown for four to seven weeks, and total DNA extracted using the protocol of Zimmer et al. (1981). Two populations each of C. biloba, C. epilobioides, C. lewisii, C. modesta, and H. het- erandra were assayed; one population was ex- amined for all other species. Only one site differ- ence was seen within a species (C. biloba), and this character state was autapomorphic to this one population. Clarkia xantiana Gray (sect. Phaeo- stoma) and C. amoena (Lehm.) Nels. & Macbr. (sect. Rhodanthos) were used as outgroups. DNAs were digested with 29 restriction en- zymes, electrophoresed in agarose gels, and South- ern blotted to reusable nylon membrane. The entire clone bank of the Petunia (Solanaceae) cpDNA genome was used successively to probe the nylon membranes for homologous cpDNA fragments. De- tailed protocols of prehybridization, nick-transla- tion, hybridization, and washes are provided in Sytsma & Schaal (19852). Volume 75, Number 4 1988 Sytsma & Smith Onagraceae —DNA & Morphology 1221 FIGURE 3. by the “branch-and-bound” option of PAUP. The tre Most parsimonious (Wagner) tree of Clarkia sect. Sympherica and Heterogaura heterandra generated e was rooted with Clarkia xantiana and C. amoena . The tree is 125 steps long and includes two parallel gains, three parallel losses, and one gain/loss. Numbers indicate numbers of restriction site mutations along each line age. Percentages along branches reflect the number of times at the monophyletic group defined by that branch occurred in 100 bootstrap samples. Based on Sytsma & t Gottlieb (1986b) Methods of phylogenetic analysis, explained in detail elsewhere (Sytsma & Gottlieb, 1986a, b), included Wagner parsimony (Farris, 1970) (PAUP version 2.4.0, Swofford, 1985), Dollo parsimony (Farris, 1977) (PHYLIP version 3.0, Felsenstein, 1985), and the Fitch & Margoliash (1967) phe- netic approach using p values of Nei & Li (1979). Felsenstein's (1985) bootstrap method (in PHY- LIP) was utilized to place confidence intervals on resulting phylogenies. A majority-rule consensus Wagner parsimony tree was used to construct a phylogeny indicating all inferred monophyletic lin- eages determined by bootstrap analysis. PHYLOGENETIC ANALYSIS AND DISCUSSION The 29 restriction enzymes used to digest the DNAs recognize approximately 605 restriction sites in each of the cpDNAs. Because all 29 enzymes recognize six base pair sequences, about 3,630 nucleotide base pairs were sampled in each of the species. A total of 119 site changes were docu- mented within Clarkia sect. Sympherica and Het- erogaura (these restriction site mutations are listed as table 3 in Sytsma & Gottlieb, 1986b), and 55 of these mutations are shared by at least two but not all members of the ingroup (including Hetero- gaura) and were used as the data matrix in the phylogenetic analyses (Table 1). The PAUP (BANDB option) and PHYLIP (MIX option) pro- grams found a single most parsimonious (Wagner) tree of 125 steps (Fig. 3). This tree requires an additional six steps beyond the 119 site changes to account for the observed variation. The most parsimonious Dollo tree is exactly congruent to the Wagner tree but is four steps longer. The unrooted Fitch & Margoliash network based on nucleotide sequence divergences is topologically congruent to the shortest Wagner tree (see fig. 4 in Sytsma & Gottlieb, 1986b). The shortest cpDNA phylogenetic tree provides unambiguous evidence for relationships within Clarkia sect. Sympherica. The cpDNA analysis substantiates Odrzykoski & Gottlieb’s (1984) sug- gestion that C. rostrata is indeed phylogenetically closer to C. epilobioides than to its morphologically related species, C. cylindrica and C. lewisii. These results are the first alternative genetic confirmation of phylogenetic relationships based on gene dupli- cation data, and they greatly Pen the utility of the latter approach in systematics. Lewis & Lewis (1955) first irm the pop- ulations now recognized as C. rostrata to be un- usual northern members of C. cylindrica, but Da- vis (1970) later separated out C. rostrata and indicated that it was more similar to C. lewisii than 1222 Annals of the Missouri Botanical Garden TABLE l. Data matrix of 55 restriction site characters for Heterogaura heterandra and the eight species of Clarkia sect. Sympherica. Outgroup states were determined from examination of C. amoena and C. xant 1 indicates absence of a restri presence of a site. Details concerning each jo are presented in Sytsma & Gottlieb (1986b). 0" autapomorphies are listed. The character state iana. No iction site and “1” indicates ) Outgroup 1100010110110100010110001001101000100011000100101100100 Clarkia epilobioides (subsect. Micranthae) 010001011 111000110001001110101000110 C. rostrata a Sympherica) 0100010 111000110001001110101000110 C. lewisü etd Sympheri ca) 0110010100111011001110011000101011100011 10 C. d (subsect. lite 110011000101011100011 10 C. ues (subsect. b os 1100000110100100010111001001101100000010010100001111101 Heterogaura hetera ndra 1100000110110100010111001001101 111101 C. modesta (subsect. Lautiflorae) 1000010010110100010111001001101100100011000100101111101 C. lingulata (subsect. Lautiflorae) c d aaa aa C. biloba (subsect. Lautiflora a ETE ORO RnR CTC IG to C. cylindrica. Moreover, C. rostrata can be crossed successfully with the former but not the latter (Davis, 1970). Davis concluded that the close morphological similarity among these three species of subsect. Sympherica suggested a common origin or even the derivation of one species from another. Clarkia rostrata is found the farthest north in foothills of the Sierra Nevada in Stanislaus, Merced, and Mariposa counties; C. lewisii is found only in the Coast Ranges in Monterey and San Benito counties; and C. cylindrica is found farther south and in more xeric habitats along the foothills of the southern Sierra Nevada and Tehachapi Moun- tains (subsp. clavicarpa) and in the southern Coast Ranges (subsp. cylindrica) (see Fig. 4). All of the progenitor-derivative species pairs examined in Clarkia have indicated that the direction of evo- lution is from north to south or from mesic to more xeric habitats (Lewis, 1962; Lewis & Roberts, 1956; Lewis & Raven, 1958; Vasek, 1958; Gottlieb, 1974). Because of the northern distribution and wide separation of C. rostrata and C. lewisii and because of the continuous and more southern dis- tribution of the two subspecies of C. cylindrica, Davis (1970) suggested that the former two species may have become restricted in their distribution and perhaps preceded C. cylindrica or have been involved in its origin. The gene duplication and the cpDNA data in- dicate that the evolutionary events within section Sympherica are more complex than data based on morphological similarity and crossing relation- ships indicate. The similarity between Clarkia ros- trata and the lineage encompassing C. cylindrica and C. lewisii strongly suggests that the common ancestor of the lineage comprising these three species plus C. epilobioides almost certainly re- sembled the former three species. Alternatively, strong phenetic convergence in C. rostrata towards C. lewisii and C. cylindrica would have to be invoked. Clarkia epilobioides is clearly closely related to C. rostrata and exemplifies a lineage that has undergone a tremendous amount of mor- phological divergence relative to other species. Clarkia epilobioides ranges from San Francisco to Baja California and has a disjunct range in Arizona (Fig. 4). Because of its exclusively inbreed- ing mode of reproduction, but despite its northern (and southern) distribution, it is almost certain that C. epilobioides has been derived from an out- crossing taxon and does not represent an ancient lineage as might C. rostrata and C. lewisii. NA restriction fragment and site analysis have demonstrated clear phylogenetic relationships among the four extant species of subsects. Mi- cranthae and Sympherica (Fig. 3). Further studies Volume 75, Number 4 Sytsma & Smith 1223 1988 Onagraceae —DNA & Morphology E Clarkia cylindrica 4 Clarkia epilobioides PAAA Clarkia lewisii Clarkia rostrata FIGURE 4. Distribution range of the four species in Clarkia sect. Sympherica subsects. Sympherica and Micranthae. Ranges in California, Arizona, and Baja California are provided for C. cylindrica, C. rostrata, C. lewisii, and C. epilobioides. 4dapted from Lewis & Lewis (1955) and Davis (1970). involving additional populations from throughout tions remaining include: Are C. rostrata and C. the ranges of these four species and involving bi- lewisii ancestral in this lineage? Did C. epilo- parentally inherited DNA (nuclear genome) are — bioides and C. cylindrica diverge independently needed to clarify how these species evolved. Ques- (or together) from that lineage? What evolutionary 1224 Annals of the Missouri Botanical Garden forces permitted the rapid morphological diver- gence in C. epilobioides? Can C. epilobioides be crossed with three species currently placed in sub- sect. Sympherica, especially C. rostrata? The most parsimonious tree (Fig. 3) also clarifies the relationships within subsect. Lautiflorae and the relationship of Heterogaura heterandra to Clarkia. The cpDNA analysis verifies the close relationship of the proposed progenitor-derivative species pair of C. biloba and C. lingulata (Lewis & Roberts, 1956; Gottlieb, 1974). Only six mu- tations separate the two species. The placement of C. modesta is the only portion of the phylogenetic tree that is not statistically documented using the bootstrap analysis. The cpDNA analysis conclu- sively places it within subsect. Lautiflorae, but its exact position in the subsection is not certain. The position of C. modesta as shown in Figure 3, how- ever, is supported by chromosome numbers. This most parsimonious tree indicates that only one aneuploid decrease from the more widespread n — 9 of the section to n — 8 (only C. biloba and C. modesta) has to be invoked. The subsequent re- versal to n — 9 in C. lingulata has been amply demonstrated. he most striking conclusion of this phylogenetic analysis is the documentation that the genus Het- erogaura is actually derived within Clarkia (Syts- ma & Gottlieb, 1986a). Indeed, H. heterandra is placed firmly within subsect. Lautiflorae with C. dudleyana as its sister species. The two species share nine cpDNA synapomorphies despite the ex- tensive morphological divergence between the two. The derivation of H. heterandra from a common ancestor with C. dudleyana is supported by the next three most parsimonious trees as well. The extreme floral and fruit reduction in Heterogaura relative to Clarkia have masked the close phylo- genetic relationship of Het M rM to an ad- vanced subsection within Clar Flavonoid analysis of risu heterandra and five species of Clarkia indicates that the four compounds present in the Heterogaura flavonoid profile are also found in the few species of Clarkia examined (Averett et al., 1982). Besides cpDNA and nrDNA analyses, therefore, the only evidence that supports a relationship of Heterogaura to spe- cific lineages within Clarkia is chromosome num- ber. Raven (1979) speculated that Heterogaura, with n = 9 (found elsewhere only in some Clarkia and Boisduvalia), “might have been derived from a species of Clarkia with the same chromosome Morphological or cytological evidence, however, could not place Heterogaura near any number." particular Clarkia species because their morpho- logical divergence had completely obscured the relationships. This study raises additional questions that can or should be addressed in the future: What are the evolutionary forces that permit such rapid mor- phological divergence as seen in Heterogaura? Is the morphological divergence seen in Heterogaura (and also C. epilobioides) common and thus in- dicative of what might occur frequently in plants? Can H. heterandra be crossed with C. dudleyana? How many genes were necessary to get expression of the extreme fruit and floral reduction in Het- erogaura? Should additional mono- and ditypic genera be suspected as similar derivations from within related and more speciose genera instead of as sister genera to these genera (the monotypic Stenosiphon and closely related Oenothera, for example)? Is the relationship of Heterogaura to Clarkia actually more complex, involving hybrid- ization and/or introgression, and thus the cpDNA results described here incomplete by not also using biparentally inherited nuclear DNA he last question is particularly important be- cause other cpDNA studies have yielded unusual results that suggest apr a by in- trogression (Palmer et al., 1983, 1985). In these instances, results from cpDNA iu can be quite different from results from nuclear DNA analysis. For these reasons, restriction site mapping of nu- clear ribosomal DNA (rDNA) has been initiated within Clarkia and Heterogaura to determine if nuclear DNA analysis provides phylogenetic results similar to cpDNA analysis. Methodology for rDNA analysis follows that described in Sytsma & Schaal (1985a). One preliminary piece of information di- rectly concerns the issue of the relationship of Heterogaura to Clarkia. A partial Sst / restriction site map for rDNA in Clarkia is shown in Figure 5. Sst Í fragments b and c are conserved across the four genera of Onagraceae examined. Frag- ment a, however, is found only in species of Clarkia sects. Sympherica (C. biloba, C. lewisii), Phaeostoma (C. xantiana), Fibula (C. bottae), and Heterogaura heterandra. The Sst I site in the 18S gene that delineates fragment a is absent in other sections of Clarkia and in Oeno- thera and Lopezia. These Clarkia and other gen- era thus lack fragment a and instead have a large lingulata, C. Sst I fragment that encompasses fragment a, the 18S gene, and an undetermined portion of the nontranscribed spacer region (NTS region in Fig. 5). Outgroup analysis (using Oenothera and Lo- pezia) would indicate that the absence of the Sst I site in the 18S gene (and thus absence of fragment a) is the plesiomorphic condition. This preliminary 1225 "uo1821 1422Dds paquosupajuou ay) fo suoi40d Suiuupds sjuaugpaf u uoimina q18u23]-1220ds 0j anp sí squauigpaf yyFiam iv noajow 194214 fo u1231d 231142ppv'] ‘auoj2 qwadas YNGA eutoA[9 Dyna (uonpndod sad sjypnpiaipui snoa2urnu 07 q042a98 fo Fuyood v suasardas YN 202) SYNO 19101 2uiqo4d wos paynsas umagoippaomp eizado] pup 'eiogjoue() *e1ne2019]9] “ente fo sə19əds ur pAjq4 1023]onu sof dou ans uo112141$24 pu susaod ju2uigvaf uoi2141524 | 158 fo WDIFOPDIOMY `ç 34n914 Sytsma & Smi Volume 75, Number 4 Onagraceae —DNA & Morphology 1988 sel o» 1226 Annals of the Missouri Botanical Garden nuclear rDNA evidence confirms that Heterogaura is indeed derived from within the genus Clarkia because the former shares a synapomorphy with sections Sympherica, Phaeostoma, and Fibula, sections of Clarkia now believed to be related based on gene duplication data (Gottlieb & Weeden, 1979) and cpDNA restriction site mapping data (see Phy- logenetic Analysis of Intersectional Relationships within Clarkia). A word of caution should be noted here con- cerning the phylogenetic use of a single molecular character as done here with rDNA. One could argue that nuclear rDNA evidence for the place- ment of Heterogaura within Clarkia is based solely on a single restriction site character that exhibits homoplasy. Clarkia rostrata has lost the Sst Í site that is found in other members of sect. Sympher- ica, even though the weight of morphology (Davis, 1970), isozyme gene duplications (Odrzykoski & Gottlieb, 1984), and cpDNA analysis (Sytsma & Gottlieb, 1986b) fully supports its inclusion in sect. Sympherica. Thus, C. rostrata has lost the Sst / site secondarily. Does this homoplasy involving C. rostrata cast doubt on the relationship of Hetero- gaura to Clarkia using nuclear rDNA? No, be- cause the loss of the Ss: / site in C. rostrata (thus producing the plesiomorphic character state) is a statistically likely gain/loss type of convergence (Templeton, 1983). On the other hand, there is little such support that the gain of the Sst I site in Heterogaura is also due to convergence rather than to shared common ancestry with certain lin- eages within Clarkia. Such a convergent restriction site gain is an order of magnitude less likely to occur than a convergent loss or a gain/loss (Tem- pleton, 1983). PHYLOGENETIC ANALYSIS OF INTERSECTIONAL RELATIONSHIPS WITHIN CLARKIA INTRODUCTION The monograph of Clarkia by Lewis & Lewis (1955) was a landmark study in classical biosys- tematics. Prior to their work, Clarkia was divided into a number of distinct genera (Clarkia, Godetia, Phaeostoma, and Eucharidium). Using extensive population collections in which they examined floral and vegetative morphology, chromosome number, and crossing relationships, Lewis able to define convincingly 11 natural sections in & Lewis were n b subsequent researchers (Raven, 1979, 1988). and monstrated on molecular grounds (Pichersky & de Gottlieb, 1983). Relationships among the seven predominantly diploid sections, as viewed by Lewis & Lewis (1955), are depicted in Figure 6. Ancestral clarkias were viewed as having relatively large, lavender-pink, bowl-shaped (godetia-type) flowers with petal mark- ings, being self-compatible but outcrossed, pos- sessing the chromosome number n — 7, and having a northern distribution (Lewis, 1980). Evolution in Clarkia has occurred primarily from north to south and involved development of floral tubes (sects. Eucharidium and Clarkia), extensive repatterning of the chromosomes with subsequent aneuploid in- creases and decreases from the primitive haploid number 7 or polyploidy (all sections), and formation of autogamous breeding systems (many sections). The large classical information base for Clarkia and the spectacular evolutionary changes that are seen in the genus have made it a model system for subsequent evolutionary studies (Lewis, 1980; Raven, 1988) Recent isozyme analysis, however, has chal- lenged certain aspects of the phylogeny proposed on morphology, chromosomes, and crossing stud- ies. Gottlieb & Weeden (1979) provided evidence that a duplication of the cytosolic gene for PGI (phosphogluco isomerase) defines four diploid sec- tions (Sympherica, Phaeostoma, Fibula, and Eu- charidium), previously not placed together by Lewis & Lewis (1955), as a monophyletic lineage. Indeed, sect. Eucharidium, with its distinctive stamen re- duction, elongated floral tube, pollen features (Small et al., 1971), and lepidopteran pollination syn- drome, represents the greatest morphological di- vergence from putative ancestral clarkias (Lewis, 1980). Section Eucharidium originally was re- tained within Clarkia only because it was strongly suspected of being involved in the intersectional derivation of the polyploid C. pulchella (Lewis & Lewis, 1955). The weight of this molecular evi- dence led Lewis (1980) to accept the argument of Gottlieb & Weeden (1979) and to propose the phylogeny illustrated in Figure 7. An analysis was initiated using restriction site mapping of cpDNA from representatives of the diploid sections to test these alternative models of sectional relationships in Clarkia. Presented here are preliminary phy- logenetic results from this analysis; a more detailed phylogenetic analysis on an expanded data set is in progress (Sytsma et al., in prep.). MATERIALS AND METHODS Total DNAs of two species of each of the seven diploid sections (one species in the monotypic Fib- ula) were obtained as described above. These rep- Volume 75, Number 4 Sytsma & Smith 1227 1988 Onagraceae —DNA & Morphology FIBULA RHODANTHOS SYMPHERICA GODETIA FIBULA SYMPHERICA PHAEOSTOMA RHODANTHOS PHAEOSTOMA MYXOCARPA Krems p. EUCHARIDIUM GODETIA NI EUCHARIDIUM FiGURE 7. Relationships among sections of Clarkia FIGURE 6. Relationships among the seven diploid based on the inclusion of the distribution of the dupli- logy, chromosome o logeny modified after Lewis & Lewis (1955 danthos is the former Primigenia, and Sympherica is the former Peripetasma.) resentative species were Clarkia amoena and C. lassenensis (Eastw.) Lewis & Lewis (sect. Rho- danthos), C. imbricata Lewis & Lewis and C. williamsonii (Dur. & Hilg.) Lewis & Lewis (sect. Godetia), C. mildrediae (Heller) Lewis & Lewis, and C. virgata Greene (sect. Myxocarpa), C. breweri (Gray) Greene and C. concinna (Fisch. & Mey.) Greene (sect. Eucharidium), C. xantiana and C. unguiculata Lindl. (sect. Phaeostoma), C. bottae (sect. Fibula), and C. biloba and C. rostrata (sect. Sympherica). Preliminary surveys of restriction fragment vari- ation using methods described above indicated that substantial amounts of site mutations and length mutations are evident at the intersectional level in Clarkia. 'The amount of both types of variation was so high that even the small portions of the chloroplast genome examined with individual probes often exhibited fragment patterns too complex to interpret accurately. Numerous insertions and dele- tions made it difficult to decide what were homol- ogous fragments. The increased frequency of mul- tiple mutations in a given fragment often generated complex fragment patterns. For these reasons, re- striction site mapping using only seven enzymes, rather than restriction fragment comparisons of numerous enzymes, was selected as the method to examine the phylogenetic relationships among sec- tions in Clarkia. This method involves reciprocal double digests to map exactly an enzyme's restric- tion sites relative to sites of other enzymes. Al- though this method is more laborious and less pro- ductive in terms of numbers of mutations screened, cated PGI Eos with morphology, chromosome numbers d features, and c M red relationships. Phylogeny modified n Lewis (1980). the method produces precise cpDNA maps for groups in which variation is often too great and interpretation of homology too difficult for simple fragment pattern analysis. The restriction enzymes Pst I, Sal I, Sma I, Kpn I, Pou Il, Xho I, and Nru I were used to digest total DNAs alone and in double digests with at least two other enzymes. Replicate filters were successively probed with the entire Petunia clone bank (courtesy of J. Palmer & E. Clark), portions of the Lactuca clone bank (courtesy of R. Jansen), and rbcL and B subunit atpase gene clones (courtesy of G. Zurawski). Map positions for these probes are illustrated elsewhere (Sytsma & Gottlieb, 1986b, fig. 1). Alignment of the maps was facilitated by the conservative nature of the chloroplast genome (Palmer, 1985a, b, 1986a, b; Palmer & Stein, 1986) and of specific restriction sites, especially sites of Pst / within the rbcL gene and Pou Il and Sma I within the rRNA genes of the inverted repeat (Sytsma & Gottlieb, 1986b, and J. Palmer, pers. comm.). da panes analysis utilized Wagner parsi- mony (PAUP) to derive the most parsimonious trees. Because Heterogaura is no longer an ap- propriate outgroup to Clarkia, Oenothera (tribe Onagreae) and Epilobium (tribe Epilobieae) were examined. Oenothera biennis, currently mapped (Sytsma & Smith, unpubl. data), is not an appro- priate outgroup even though it is placed within the same tribe as Clarkia, because it contains a large inversion in the large single copy region of the chloroplast genome. Epilobium brachycarpum esl (=E. paniculatum Nutt. ex Torr. & A. Gray), although placed in a separate tribe, was used as 1228 Annals Missouri Prin Garden TABLE 2. Data matrix of 23 restriction site characters for representatives of the seven diploid secioni ki Clarkia (13 species) and one ae Epilobium). No autapomorphies are listed. The character sta indicates absence of a restriction site a d “1” indicates presence of a site. Details concerning each cud will be presented elsewhere (Sytsma & Gottlieb, in prep.). Epilobium brachycarpum Clarkia biloba (Sympherica) rostrata (sect. Sympherica) xantiana (sect. Phaeostoma) unguiculata (sect. Phaeostoma) bottae (sect. Fibula) concinna (sect. Eucharidium) brewerii (sect. Eucharidium) imbricata (sect. Godetia) williamsonii (sect. Godetia) amoena (sect. Rhodanthos) lassenensis (sect. Rhodanthos) virgata (sect. Myxocarpa) mildrediae (sect. Myxocarpa) AAAANAAAAAANA 11111001000010000001110 00111001000010000000000 00111001000010000000000 10110001000011100100000 00111001000010000000000 00010001001011000100000 01100000011010010010000 01100001011011110010000 00100001000010011001111 00100001000010011001111 00110001000000000000000 00110001000000000000000 00110111100110000000000 00110110100110000000000 the outgroup for Clarkia. The PAUP option BANDB (Hendy & Penny, 1982) was run to find most parsimonious trees. These trees were used to construct a strict consensus tree using the PAUP option CONTREE PHYLOGENETIC ANALYSIS AND DISCUSSION Restriction site maps of cpDNAs from the seven sections of Clarkia and from Epilobium brachy- carpum are presented elsewhere (Sytsma et al., in prep.). The seven restriction enzymes mapped to date recognize + 100 sites on average in each cpDNA. This represents 0.4% of the total nucleo- tide sequence of each cpDNA. A total of 51 re- striction site mutations were found among the 13 Clarkia species examined. An additional 14 site mutations were found in Epilobium relative to all Clarkia, but these are not further analyzed here because they do not provide additional information concerning relationships among Clarkia sections; they are being used in a family-wide ein analysis (Sytsma & Smith, in prep.). Of the restriction site mutations documented, 23 are phy- logenetically informative; that is, they are shared by at least two but not all of the OTUs. The data matrix for these 23 restriction site characters in the seven sections examined makes up Table 2. No autapomorphies were included in the PAUP analysis. To simplify the A m analysis fur- ther, C. rostrata and C. imbricata were removed from the phylogenetic analysis because for these 23 characters they are identical in character states to C. biloba and C. williamsonii, respectively. These species pairs are identical for the phyloge- netically informative characters examined here, al- though a number of autapomorphies were seen for Individual species. The Wagner analysis using option BANDB found 201 most parsimonious trees of length 32. With the autapomorphies included, these trees have a total length of 60 steps and a consistency index Kluge & Farris, 1969) of 0.850. The strict con- sensus tree derived from 100 of these most par- simonious trees is illustrated in Figure 8A. The high level (15%) of homoplasy in the Clarkia data set, which is similar to that found within the entire Asteraceae (Jansen & Palmer, 1988, pers. comm.) could be due in part to (1) multiple changes between the outgroup Epilobium and the ingroup Clarkia, (2) the more rapid divergence of cpDNA in the strictly annual Clarkia, and/or (3) the greater age of Clarkia relative to the Asteraceae. The first situation would give rise to trees where errors are made in the determination of the plesiomorphic character state. Two additional PAUP analyses were thus performed: (1) scoring the plesiomorphic state of characters involved in multiple changes from outgroup to ingroup as unknown (characters l and 2 scored as missing in the outgroup in this instance) and (2) removal of the outgroup taxon and use of midpoint rooting. Midpoint rooting can clearly only be justified if rates of character change throughout lineages are nearly equal. Clocklike evolution of chloroplast DNA could not be statis- tically rejected for Clarkia sect. Sympherica (Syts- ma & Gottlieb, 1986b), thus lending support to the use of midpoint rooting within Clarkia. Scoring the outgroup state of characters 1 and 2 as missing and allowing PAUP to determine the plesiomorphic states resulted in the same 201 most parsimonious trees, but homoplasy was reduced by two conver- Volume 75, Number 4 Sytsma & Smith 1229 1988 Onagraceae—DNA & Morphology gences in each tree (58 total steps including autapo- a ER SYMPHERICA morphies, C.I. = 0.879, 12% rate of homoplasy). A 7” Removal of Epilobium completely and resorting UNG PMAEOSTOMA to midpoint rooting resulted in 15 most parsimo- nious trees, each five steps shorter than trees gen- erated with Epilobium present (55 total steps in- cluding autapomorphies, C.I. = 0.927, 7% rate of homoplasy). The strict consensus tree of these 15 trees is depicted in Figure Phylogenetic relationships among sections of Clarkia as depicted in the consensus tree of Figure 8A suggest that sect. Godetia is monophyletic and the sister group to the rest of Clarkia. The early divergence of sect. Godetia is also seen when Epi- lobium is removed as an outgroup and midpoint rooting is performed (Fig. 8B). Section Godetia consists of diploid and polyploid species similar in many respects to sect. Rhodanthos and, according to Lewis & Lewis (1955), almost certainly derived from the “primitive” related to any other section. Several morphological features constant in sect. Godetia, notably the erect buds and rachis and the conspicuously eight-ribbed ovary, are found in portions of sect. Rhodanthos (Lewis & Lewis, 1955). Derivation of sect. Godetia from elements within sect. Rhodanthos is consis- tent with respect to chromosome number, since all diploid species of sect. Godetia apparently have evolved with increase in chromosome number from n = 7 (found in sect. Rhodanthos) to both n = 8 and n = 9 (found in sect. Godetia). Further support for this early split of sect. Godetia comes from a preliminary cladistic analysis of 38 characters en- compassing morphology and isozyme gene dupli- cations, which places sect. Godetia as the sister group to the rest of Clarkia (K. Holsinger, pers. comm.). It is possible that sect. Godetia was indeed the first lineage splitting off from ancestral clarkias, but it also appears that this ancestral group, per- haps now encompassing sect. Rhodanthos, con- tinued to evolve and subsequently split off the other Rhodanthos and not clearly sections. A larger survey of species within sect. Rhodanthos would be needed to detect the pos- sibility that this section is paraphyletic, with dif- ferent elements giving rise independently to Gode- tia and to the other sections. This scenario is implicitly suggested by the phylogenetic model of Lewis & Lewis (1955; also see Fig. 6) and by the distribution of phosphoglucomutase (PGM) gene duplications (Soltis et al., 1987). Relationships within the second lineage com- prising the other six diploid sections is not clear, as all sections split at a polychotomous node (Fig. 8A). Of these six sections, all but sect. Phaeostoma are monophyletic lineages. Clarkia xantiana and C. unguiculata (sect. Phaeostoma) consistently do BOT FIBULA mm] CON EUCHARIOIUM BRE AMO RHODANTHOS LAS VIR MYXOCARPA MIL jl om gx TIA L. Ros B — L. SYMPHERICA BIL UNG : === a A he RHODANTHOS L us VIR | MYXOCARPA MIL Í e EUCHARIDIUM IIQZQIISt!SIƏIIGGGII | - GODETIA oO Wi FIGURE 8. Strict consensus trees of relationships mong sections in Clarkia based on chloroplast DNA fiction site mapping.— E tree is rooted with Pr cia and is ned from yses using 100 most parsimonious trees. — is phy- logenetic tree E iuo on the full data set minus Epi- lobium, rooted using the MIDPOINT option in PAUP, and using 15 most parsimonious trees. See text for discussion. not form a monophyletic clade in most of the 100 most parsimonious trees examined in detail (data not shown). Indeed, C. xantiana often is aligned with C. bottae of sect. Fibula, whereas C. ungui- culata. often is aligned with C. biloba and C. ros- trata of sect. Sympherica. These relationships are seen in the consensus tree when the outgroup is omitted (Fig. 8B). Lewis & Lewis (1955) postulated that C. bottae (formerly C. deflexa), the only species of sect. Fibula, represented a diploid hybrid be- tween sects. Sympherica and Phaeostoma. If this is true, the maternal genome of C. bottae most likely came from a species similar to C. xantiana of sect. Phaeostoma and not C. unguiculata as postulated by Lewis & Lewis (1955). The proposed monophyletic nature of sects. Eu- charidium, Sympherica, Phaeostoma, and Fibula based on the presence of the PGI duplication (Gott- lieb & Weeden, 1979) cannot be rejected or sup- ported with the results of cpDNA restriction site 1230 Annals of the Missouri Botanical Garden mapping when using Epilobium as an outgroup, since all four sections, along with sects. Rhodan- thos and re split from a polychotomous node (Fig. 8A). However, when Epilobium is re- moved as an outgroup and midpoint rooting is used, sect. Eucharidium is placed in a lineage with sect. Godetia and is separate from the other three sec- tions with the PGI gene duplication (Fig. Preliminary nuclear rD evidence supports the separation of sect. Kucharidium from sects. Sympherica, Phaeostoma, and Fibula. As detailed earlier, sects. Sympherica, Phaeostoma, and Fib- ula share a synapomorphic gain of an Sst / site in the 18S gene of nrDNA (Fig. 5). Section Fucha- ridium retains the plesiomorphic condition also found in sects. Myxocarpa, Godetia, an 0- danthos and in Oenothera and Lopezia. This rDNA site mutation could also be argued as arising within a monophyletic lineage of the four sections but after the split of sect. Fucharidium. An obvious problem that is apparent with the Clarkia cpDNA data set is the low number of synapomorphic characters shared by two or more sections. A rapid and early divergence of most of the sections within Clarkia could account for the relatively low numbers of phylogenetically infor- mative site changes encountered among sections. The slowly evolving chloroplast genome would not be expected to exhibit numerous changes in the short time intervals when the sections shared com- mon ancestors. Conversely, most cpDNA sequence changes would be expected in the long time periods after most sections had already diverged. Addi- tional restriction enzymes that recognize low num- bers of sites in cpDNA (Bgl I, Sst l, for example; see Palmer, 1986a) are now being examined for variation within Clarkia. Additional analyses are also under way to examine individually the sets of most parsimonious cpDNA-derived trees for the number in each tree of unlikely convergent gains and loss/gains relative to the more likely conver- gent losses and gain/losses (Templeton, 1983). A subset of more likely trees (i.e., with fewer con- vergent gains or loss/gains) might then be found to demonstrate more rigorously phylogenetic re- lationships within Clarkia. PHYLOGENETIC RELATIONSHIPS AND DIVERGENCE WITHIN FUCHSIA INTRODUCTION Fuchsia is a genus of 102 species belonging to ten sections (Berry, 1982; Raven, 1988; al., 1988). Most species of this genus of outcrossing shrubs and some trees occur in South America, erry et including 60 species of sect. Fuchsia (two in His- paniola), 14 species of sect. Hemsleyella, 8 species of sect. Quelusia, the monotypic sect. Kierschle- geria, and an undescribed monotypic section from northern Peru. Twelve species of sects. Ellobium, Encliandra, Jimenezia, and Schufia occur in Mexico and Central America. The four species of sect. Skinnera are found in New Zealand (3) and Tahiti (1). Most lines of evidence point to an origin of Fuchsia in austral temperate forests of South America in Paleogene times (Berry, 1982; Raven, 1988). The species of the large South American Quelusia, southeastern Brazil, with one species in Chile, pos- sess the largest suite of generalized characters in the genus and may represent the extant section most similar to ancestral fuchsias. These characters include shrubby habit, bisexual flowers, well-de- veloped petals, bird-pollination, numerous seeds, and segmented-beaded viscin pollen threads (Skvarla et al., 1978; Berry, 1982; Nowicke et al., 1984; Averett et al., 1986; Raven, 1988). Raven (1988) postulated that F. /ycioides Andr. (sect. Kier- schlegeria) is related to sect. Quelusia, based on their polyploidy and temperate South American distribution. But F. /ycioides has a number of derived characters, including its dry coastal scrub sect. restricted to the mountains of habitat, summer deciduousness, functional dioecy, small flowers, few seeds, and smooth viscin pollen threads. The first offshoot in Fuchsia probably involved the lineage that dispersed to New Zealand an subsequently Tahiti (Fig. 9). This lineage, now com- prising four species in sect. Skinnera, separated from the rest of Fuchsia at least 25 million years ago since fossil pollen has been recorded in late Oligocene and early Miocene deposits from New Zealand (Mildenhall, 1980; Daghlian et al., 1985) and eastern Australia (P. Berry, pers. comm.). Skinnera is the most distinctive section in the ge- nus, with the advanced conditions of male sterility (Godley, 1955), reduced petals, and varying life forms that include a tree, a scandent shrub, and an almost herbaceous creeper. Other early dispersal events probably included the ancestor to the morphologically related sections Encliandra, Jimenezia, and Schufia of Mexico and Central America (Fig. 9). These sections share unusual characters of small flowers, smooth viscin pollen threads, male sterility, and lobed adnate nectaries (Breedlove et al., 1982; Berry, 1982). These three sections are so distinctive that it is difficult to relate them to their South American relatives (Raven, 1988). The two most speciose Volume 75, Number 4 Sytsma & Smith 1231 1988 Onagraceae—DNA & Morphology (SomurIa) North America Andes C» North America New Zealand ahiti C» Temperate _A__ South America Ü ^ Ne Z = = KIERSCHLEGERIA SOUTH AMERICAN “PROTO” FUCHSIA FIGURE 9. Schematic diagram illustrating probable eer events in the genus Fuchsia (based almost entirely on ideas presented in Berry, 1982; depicted as ancestral to the genus. Section ‘Quelus pu et al., , however, is god cte 2; and Raven, 1988). o extant section : ted as retaining the inm number T states that such an ancestral fuchsia odd probably have had. See text for discussio sections, Fuchsia and Hemsleyella, occurring mostly on moist slopes of the tropical Andes, most certainly evolved rapidly as the Andes uplifted to their present height over the past few million years (Berry, 1982; Raven, 1988). The fourth Mexican and Central American section, Ellobium, is related to these two Andean sections and represents an additional and probably Neogene dispersal event of Fuchsia northward (Fig. Flavonoid analyses of many Fuchsia species have provided useful information concerning evolution within the genus (Williams et al., 1983; Averett & Raven, 1984; Averett et al., 1986). Flavones, otherwise rare in the Myrtales (Gornall et al., 1979), are found in Onagraceae only in Circaea and Fuch- sia. The presence of flavones is presumed to be ancestral in Fuchsia, since they are found in all species of sect. Skinnera, in the primitive F. pee dens of sect. Ellobium, in the primitive F. ma uelusia, in sect. Kier- (Averett et al., 1986). The presence of sulfated flavones only in sect. Skinnera again emphasizes the distinctiveness of the section within Fuchsia (Williams et al., 1983) A chloroplast DNA restriction fragment and site analysis was begun in Fuchsia to look at a number of systematic and evolutionary questions in the genus. Can sect. Skinnera be shown to have di- verged first in the genus? Do cpDNA divergence data indicate large genetic differences among the four species of the Old World which exhibit ex- tremes in plant form? Are sects. Quelusia and Kierschlegeria, which are polyploid and inhabit putative ancestral biogeographic habitats, closely related? Does cpDNA analysis support the mono- phyletic origin of sects. Encliandra, Schufia, and Jimenezia from an early dispersal event? Does the Central American section Ellobium relate to the Andean sects. Fuchsia and Hemsleyella in a phy- logenetic sense? What are the closest relatives of the recently discovered monotypic and tuber-bear- ing section from western Peru? Initial surveys of cpDNA restriction fra tatives of all sections, tentative phylogenetic inter- pretation, and answers to some of these questions ment variation in represen- are presented here. MATERIALS AND METHODS Total DNA from 16 taxa of Fuchsia was ex- tracted from leaf tissue as described above. Rep- resentative species from the sections in Fuchsia included F. excorticata L. f. (sect. Skinnera), F. jimenezii Breedlove, Berry & Raven (sect. Ji- menezia), F. arborescens Sims and F. paniculata Lindley (sect. Schufia), F. thymifolia H.B.K. (sect. Encliandra), F. splendens Zucc. (sect. Ellobium), F. lycioides Andr. (sect. Kierschlegeria), F. ma- gellanica and F. regia (Vand. ex Vell.) Munz (sect. 1232 Annals of the Missouri Botanical Garden TABLE 3. Data matrix of 46 restriction site characters for representatives of the nine described sections of Fuchsia and F. pachyrrhiza of the new monotypic section from western Peru. Outgroup character states were derived from both Clarkia and Epilobium. bod LA e are listed. The character state “0” indicates absence of a restriction site and “1” indicates presence of a si Outgroup 10000000110011010 Fuchsia splendens (sect. Ellobium) 10000000110010010 F. se ie ie SE d 00000 F. bo i n se a) 0010001 100000001 1000000010101 100000001 10010010 F. nigricans (sect. Fuchsia) 110000100110010110 F. verrucosa (sect. Fuchsia) 0010001000000001 1000000010001 100000001 10010010 F. tillettiana (sect. Hemsleyella) 10000010111010010 F. jimenezii (sect. Jimene ia) 001110100010011111011 1010001 L1100011100101001 F. lycioides (sect. Kierschlegeria) 0010001000011001100000001100010010010110010010 F. magellanica (sect. Quelusia) 1010000011 110010000 F. regia (sect. Quelusia) 100000001 10010010 F. arborescens (sect. Schufia) 10001000110010010 F. paniculata (sect. Schufia) 10001000110010010 F. excorticata (sect. Skin ra) 11 100010100001010000001 10000110000000110011010 F. pachyrrhiza (sect. Pachyrrhiza) 110000000110010010 Quelusia), F. tillettiana Munz (sect. Hemsleyel- la), F. boliviana Carriére, F. nigricans, and F. verrucosa Hartweg (sect. Fuchsia), and F. pachy- rrhiza Berry & Stein (sect. Pachyrrhiza). Eleven restriction enzymes were utilized (Sst I, Hind Ill, Eco RV, Pvu Il, Bgl I, Kpn 1, Pst I, Sma I, Sal I, Bst Ell, and Sph D. Filters were sequentially probed with Petunia probes that cover y the large single copy region. Phylogenetic analysis followed procedures described above for the analysis of cpDNA variation within Clarkia sect. Sympherica. Polarity of restriction site changes was determined from cpDNA maps of Clarkia and Epilobium in which most of the sites for these eleven restriction enzymes have been determined (see above). Restriction fragment pat- terns were examined in these outgroup genera for restriction enzymes not completely mapped. Wag- ner analysis (PAUP) was used to determine most parsimonious trees and the CONTREE program utilized to generate a strict consensus tree. RESULTS AND DISCUSSION Approximately 100 restriction sites were ex- amined in each taxon with the combination of probes and restriction enzymes. Forty-six restriction site mutations were seen among the 16 ingroup taxa surveyed. Only nine of these mutations are shared by at least two but not all species of Fuchsia examined and thus are phylogenetically informa- tive (Table 3). Nucleotide divergence within Fuch- sia is high, with p values ranging up to 4.8%. Fuchsia jimenezii (sect. Jimenezia) exhibited the most site divergence, followed by F. excorticata (sect. Skinnera). PAUP analysis generated over Volume 75, Number 4 Sytsma & Smith Onagraceae —DNA & Morphology 1233 SPL Ellobium BOL NIG Fuchsia THY Encliandra PAC Pachyrrhiza ARB Schufia PAN TIL Hemsleyella eO Kierschlegeria MAG | Quelusia REG JIM Jimenezia EXC Skinnera COMITÉ DO ID VUI'nur 10. Strict consensus tree depicting relationships among the sections of Fuchsia based on cpDNA restriction site mutations. ree was generate from a subset of most parsimonious trees that lacked an unlikely convergent gain. Species are M hi by abbreuiation (see Table 3). 100 most parsimonious trees of 49 steps (C.I. = 0.94). Inspection of the 100 trees retained in mem- ory by PAUP showed that three lineages were implicated as the sister group to the rest of Fuchsia: (1) sect. Skinnera, (2) sect. Skinnera with sect. Jimenezia, and (3) sect. Jimenezia. Inspection of all 100 trees by the CHGLIST option in PAUP further indicated that all trees in categories (2) and (3) above required an additional unlikely conver- gent gain or loss/gain (see Templeton, 1983). T unlikely convergence is not found in trees in cat- egory (1). A consensus tree was obtained for all the trees not requiring this unlikely convergence and is depicted in Figure 10. cpDNA restriction fragment analysis indicates that most mutations encountered (39 out of 49) are autapomorphies. Relationships among sections based on the available phylogenetically informative cpDNA data are thus tentative and subject to change as additional restriction site mutations are found in the ongoing phylogenetic analysis of Fuchsia. e consensus tree indicates that the Old World sect. Skinnera and the monotypic Central Amer- is preliminary ican sect. Jimenezia are the first lineages to split off from the presumed ancestral Fuchsia stock in temperate South America. The consensus tree places sect. Skinnera as the sister group to all other Fuchsia sections, including Jimenezia. 'This placement, however, is based on a single and homo- plasious restriction site synapomorphy for all sec- tions excluding Skinnera. As indicated above, the ancient split and long evolutionary divergence of sect. Skinnera is supported by the presence of 30- million-year-old fossil Fuchsia pollen, the spectac- ular morphological divergence within the section, and the presence of sulfated flavones. The large cpDNA divergence evident in the monotypic sect. Jimenezia relative to all sections is surprising and merits further research. Based on floral structure and reproductive characters, Fuch- sia jimenezii has been suggested to occupy an intermediate position between the Central Ameri- can sections Encliandra and Schufia (Berry, 1982). The large cpDNA divergence between sect. Ji- menezia and sects. Encliandra and Schufia is thus contrary to results from floral characters. Berry (pers. comm.) indicates, however, that many such floral characters appear to have evolved several times independently in the genus Fuchsia based on preliminary cladistic analysis of floral and veg- etative morphology. The distant relationship of sect. Jimenezia to these two other sections, as suggested by the cpDNA cladistic analysis, indicates that substantial morphological convergence is indeed present in Fuchsia. Again, additional cpDNA data are needed to determine more firmly the position of F. jimenezii within Fuchsia. The remaining eight sections of Fuchsia share three synapomorphies, but relationships among these sections are not clear based on the prelimi- nary cpDNA consensus tree in Figure 10. Three distinct lines diverge early: (1) a lineage comprising 1234 Annals of the Missouri Botanical Garden sects. Kierschlegeria and Quelusia, which is de- fined by three synapomorphies; (2) sect. Hemsley- ella, defined by two autapomorphies; and (3) an unresolved lineage comprising five sections (Schu- fia, Encliandra, Fuchsia, Ellobium, and Pachy- rrhiza), defined by one homoplasious site mutation. The lack of substantial numbers of cpDNA site synapomorphies linking any of these sections (ex- cept for Kierschlegeria and Quelusia) is very suggestive that the genus Fuchsia diverged rapidly and profusely following the early separation of sect. Skinnera into the Old World and sect. Jimenezia into Central America. Indeed the three species examined from sect. Fuchsia do not even form a monophyletic lineage within the strict consensus tree depicted in Figure 10. The close phylogenetic relationship of the tem- perate South American sects. Kierschlegeria and Quelusia is supported by cpDNA restriction frag- ment analysis. These two sections have already been suggested to be closely related (Raven, 1988), despite the unusual derived vegetative and floral characters of F. lycioides of sect. Kierschlegeria. The lack of resolution among other sections of Fuchsia in this cpDNA restriction site analysis does not provide much evidence for or against the pre- vailing phylogeny. The phylogenetic relationships of the new sect. Pachyrrhiza from Peru to other Fuchsia sections also are not resolved. These and additional representatives of Fuchsia are currently being surveyed with larger numbers of restriction enzymes and with an entire cpDNA clone bank to maximize the numbers of site mutations for phy- logenetic analysis. However, if most sections of Fuchsia did indeed diverge quickly and at about the same time, as suggested by this study, there may not be substantial and thus statistically useful numbers of cpDNA synapomorphies. A preliminary cladistic analysis of morphological and cytological characters in Fuchsia likewise demonstrated the early divergence of sect. Skinnera and also failed to resolve relationships among the remaining New World sections (P. Berry and J. Crisci, pers. comm.). DNA versus MonPHOLOGY: A REVIEW Chloroplast DNA restriction site comparisons in the Onagraceae have substantiated many relation- ships based on morphology, cytology, and exper- imental crosses. Most relationships in Clarkia sect. Sympherica are congruent with the earlier results of Lewis & Lewis (1955) and Davis (1970). For example, subsect. Lautiflorae is shown to be a natural lineage, with C. lingulata and C. biloba as a close sister species pair, supporting morpho- logical and cytological evidence (Lewis & Roberts, 1956) and isozymic evidence (Gottlieb, 1974). Large chloroplast DNA differences among sections of Clarkia are consistent with the results of early work, suggesting that the genus, although natural, is composed of at least several evolutionarily dis- tinctive sections (Lewis & Lewis, 1955). Likewise, the preliminary cpDNA analysis in Fuchsia pro- vides evidence in support of the early divergence of the Old World sect. Skinnera, an event also suggested by morphological and phytochemical studies. However, each of these separate cpDNA studies in the Onagraceae also provides strong evidence that DNA and morphology can result in different phylogenetic conclusions. The DNA results place Clarkia rostrata with the morphologically dissim- ilar C. epilobioides rather than with C. lewisii and C. cylindrica, species with which C. rostrata is barely distinguished morphologically, an unex- pected result supported by isozyme evidence (Od- rzykoski & Gottlieb, 1984). Even more unexpected is the discovery that cpDNA characters, as well as nuclear rDNA characters, provide compelling evi- dence that the monotypic Heterogaura is actually derived within Clarkia and has C. dudleyana as its closest relative. This relationship is clearly at odds with the great morphological differences be- tween the two genera involving taxonomically im- portant floral and fruit characters. The cpDNA analysis of Fuchsia indicates that F. jimenezii is one of the most ancient lineages within the genus but provides no evidence for a relationship of F. jimenezii to sects. Schufia and Encliandra, sec- tions with which it shares several derived charac- ters. A survey of published phylogenetic studies using chloroplast DNA in angiosperms is presented in Table 4. Although a number of these studies en- countered no incongruity between the relationships generated with cpDNA and morphology, crossing studies, or isozymes, many of the studies have found at least some instances of incongruity. Var- ious explanations for these discrepancies are pro- vided in these studies: (1) morphology or repro- ductive isolation may not be good measures of phylogenetic relatedness; (2) reliance on phenetic rather than cladistic studies of morphological vari- ation; (3) unequal rates of morphological or geno- mic divergence; (4) unknown levels of molecular variation within ancestral species; and (5) cyto- plasmic exchange through introgressive or second- ary hybridization. The last explanation has been used in addressing incongruities in cpDNA studies of Brassica and Pisum (see Table 4). Cytoplasmic Volume 75, Number 4 1988 Sytsma & Smith Onagraceae—DNA & Morphology 1235 TABLE 4. Comparison of chloroplast DNA restric- tion data versus morphological, cytological, and/or isozymic data in phylogenetic studies within angio- sperms. I. Studies indicating congruence A. Citrus (Green et al., 1986) B. Coffea (Berthou et al., 1983 C. Cucumis (Perl-Treves & Galun, 1985; Perl- Treves et al., 198 D. Linum (Coates & Cullis, 1987) E. Nicotiana (Kung et al., 1982) F. Solanum (Hosaka et al., 1984; Hosaka, 1986) G. Triticum (Bowman et al., 1983; Tsunewaki & m. ihara , 1983) Zea (Doebley, 1987; Doebley et al., 1987) II. Studies indicating incongruencies or unexpected re- lationships A. Asteraceae subtribe Barnadesiinae (Jansen & Palmer, 1987, 1988) B. Clarkia rostrata and C. epilobioides (Sytsma & Gottlieb, 1986b) C. Daucus capillifolius and D. carota subsp. sa- tivus (DeBonte et al., 1984 D. Helianthus annuus and H. bolanderi (Riese- berg et al., 1988) E. Heterogaura heterandra and Clarkia dudley- ana (Sytsma & Gottlieb, 1986a) F. Heuchera micrantha (Soltis et al., in press) G. Lisianthius (Sytsma & Schaal, 1985a, b) H. Nicotiana debneyi and N. repanda (Salts et al., 1984) I. Populus nigra and P. alba (Smith & Sytsma, in prep.) III. ena ene introgression or secondary hy- bridiza A. pde napus (Palmer et al., 1983) B. Lycopersicon chmielewskii (Palmer & Zamir, 1982 C. Pisum sativum (Palmer et al., 1985) exchange via hybridization and introgression, giv- ing rise to different and sometimes unexpected organeller- and nuclear-based phylogenies, is prob- ably common in angiosperms. Clearly, a molecular phylogenetic study would be more thorough (and also more willingly accepted by the systematic com- munity) if both the biparentally inherited nuclear genome and a predominantly uniparentally inher- ited organeller genome are examined and com- red. m (1987), in a perceptive review of the promises and pitfalls of plant systematics at the DNA level, stated that the available DNA studies (both nuclear and organeller) suggest that **sweep- ing statements that a particular molecular phylog- eny is ‘right’ and that more traditional approaches, such as morphology, are *wrong' when the two do not happen to agred arb unwarranted without fur- ther investigations” [italics added]. If, however, further investigation using alternative genetic or molecular methods consis tently provide results con- trary to the traditional approaches, it is then time to reexamine these traditional approaches. The examples presented here of Clarkia epi- lobioides/ C. rostrata and C. dudleyana/ Hetero- gaura heterandra might well be some of the most definitive examples of where molecular phylogeny could be considered ‘right’ and the traditional ap- proach ‘wrong.’ In these two cases, data from chloroplast DNA, nuclear rDNA, and isozymes pro- vide independent and congruent phylogenies, con- trary to phylogenies using morphology. Doyle (1987) further stated that “it is just such instances of incongruence that are likely to lead to major revelations about the evolution of the taxa being studied—or of the molecules being used in the analysis." Thus, disparity between DNA and mor- phology (1) is expected to occur in some or many systematic studies, (2) should be the basis for in- cluding additional systematic procedures to ex- amine the disparity, (3) will provide insight into relative rates of molecular and morphological di- vergence, (4) should provide insight into what char- acters (morphological or molecular) in a given group are particularly prone to convergence or parallel- ism and thus less phylogenetically useful, and (5) will undoubtedly permit new or previously nontra- ditional questions to be asked and answered. LITERATURE CITED AVERETT, J. E. & P. H. Raven. 1984. poer ^g nagraceae. Ann. Missouri Bot. Gard. 0 W. J. Ha . RAVEN. 1982. The flavonoids of Heterogaura (Onagraceae). Phyto- chemistry 21: 1834. . BERRY & P. H. Raven. 1986. Füsebolds id üisrorhid evolution in B (On- agraceae). Amer. J. Bot. 73: 1525-15 The systematics and AE E of Fuchsia sect. debo (Onagraceae). Ann. Missouri Bot. Gard. di -198. , B. EIN, S. CARLQUIST & J. W. NOWICKE. Fuchsia x (Onagraceae), a tuberous new Fuchsia from western Peru. BERTHOU, F., C. MATHIEU & F. VeDEL. 1983. Chlo- roplast and mitochondrial DNA variation as indicator of phylogenetic relationships in the genus Coffea L. Theor. Appl. Genet. 65: 77-84. , G. BoNNARD & T. A. Dyer. 1983. Chloroplast ‘DNA variation between species of Trit- icum and Aegilops. Location of the variation on the chloroplast genome and its r o the inheri- tance and classification of the eae Theor. Appl. Genet. 65: 247-262 1236 Annals of the Missouri Botanical Garden — OVE, D. E. The systematics of Fuchsia vui (Onagracese). Univ. Calif. Publ. Bot. 53: X RRY & P. H. Raven. 1982. The Mexican and rude nius species of Fuchsia t. Encliandra. Ann. Mis- souri Bot. Gard 60: 20 9- 234. Coates, D. & C. A. Cuntis. 1987. Chloroplast DNA variability among Linum species. Amer. J. Bot. 74: -268. DAGHLIAN, C. P., J. J. SKvaRLA, D. POCKNALL & P. H. RAVEN. 1985. Fuchsia pollen from the early Mio- . 72: 1039-1047. 970. The systematics of Clarkia bottae, indrica, and a related species, C. rostrata Brittonia 22: 270-284. DeEBoNTE, L. R., F. Marrugws & K. G. WILSON. 1984, araon of plastid and jeg DNAs enus Daucus. Amer. J. Bot. 71: 932-940 iis T 1987. Evolution of the Zen chloroplast genome. dient Trends in Plants 1: 3-6. xem, W. RENFROE & A. BLANTON. 1987. Restric- n site variation in the Zea chloroplast genome. ABRE 117: 139-147. DovLE, J. J. 1987. Plant systematics at the DNA level: puse and pitfalls. Int. Org. Plant Biosyst. News- lett. 8: 3-7 Fans J S. 1970. rae for computing Wagner s. Syst. Zool. 19: 83-92. í Phylogenet analysis under Dollo’s w. Sy st. Zool. 26: 77-88. Viii NSTEIN, J. 1985. pari limits on phylogenies: an approach using the bootstrap. Evolution 39: 783- 1 1967. eli 155: 279 Fitch, W. M. & E. MancoLiAsH. of phylogenetic trees. Science ioci ^ J. 1955. Breeding systems in Nan pe s. I. Fuchsia. Ann. Bot. (London) n.s. 19: 549- GornaLL, R. J., B. A. Boum & R. DAHLGREN. 1979. he poe of flavonoids in angiosperms. Bot. Not. : 1-29. GOTTLIEB, T D. 1974. Genetic confirmation of the origin p Plage lingulata. Evolution 28: 244-250. m—— Ann. Misso 1977b. “Evidence for duplication and diver- ence of the st erase in diploid : species of Clarkia. Genetics 86: 289- 307. . F. WEEDEN. 1979. Gen O and phylogeny i in Clarkia. Evolution 33: 1024-1039. GREEN, R. M., A. VARDI & E. Gatun. 1986. The plastome of Citrus. Physical map, variation among n with 1982. Branch and bound algorithms to determine minimal evolutionary trees. Math. Biosci. 59: 277-290. HorsiNGcER, K. E. € H. Lewis. a new section and subsection in Clarkia Description of Onagra- is the mother of the potato? e j ase analysis of chloroplast DNA of cultivated aede Theor. Appl. Genet. 72: 606-618 „Y. Oc IHARA, M. MarsUBAYASHI & K. Tsun WAKI. 1984. Phylogenetic relationships between the tuberous Solanum species as revealed by restric- tion endonuclease analysis of indies DNA. Jap. . Genet. 59: 349- JANSEN, R. K. & J. D. PALMER. Sigk A chloroplast A inversion marks a ient evolutionary split in the sunflower family (steraceae) Proc. Natl. fe ad. U.S.A. 84: 5818-5822 Phylogenetic oc rn of chloroplast DNA restriction site variation in the Mutisieae (Asteraceae). Amer. t. 79; 751- 7164. KLUGE, A. G. & J. S. Farris. 1969. La s. phy- letics and the evolution of anurans. Syst. Zool. 18: -32. Kune, S. D., Y. S. Zuu & G. F. SHEN. 1982. Nicotiana chloroplast ace S ra e DNA evolution. Theor. Appl. Gen 9. Lewis, H. 1962 lo selection as a factor in speciation. Evolution 16: 257-271. 1980. The mode of evolution in Clarkia. Sarasa paper presented at Int. Congr. Syst. and Evol. Biol. II, Vancouver, British Columbia & M. EWIs. 1955. The genus Clarkia. Univ. Calif. Publ. Bot. 20: 241-392. & AVEN. 1958. Rapid evolution in Clarkia. Evolution 12: 319-336 & M. R. RopERTs. 1956. The origin of Clarkia lingulata. Evolution 10: 126- MILDENHALL, D. C. 80. New Ae late Cretaceous a: O = ° N ° ° EN m = e g. o 0% c — 0 a 1 £g ye] = < f e 3 Lad =. = = 2. e = e c. Natl. Acad. U.S.A. 75: 5269-5273. T J. SkvaRLa, P. H. Raven & P. E. 1984. A palynological study of the genus uu (Onagraceae). Ann. Missouri Bot. Gard. 71 5-9]. NUM ICKE, J. W BERR oe I. J. & L. D. Gorriizs. 1984. Duplica- tions of genes coding 6- -phosphogluconate dehydro- genase in Clarkia (Onagraceae) and their phyloge- netic werte Syst. Bot. 9: 479-489 PALMER, J. 1985a. Evolution of chloroplast and SR DNA in plants and algae. Pp. 240 in R. J. MacIntyre (editor), Moni da in Evolutionary Biology Evolutionary Genetics, Plenum Press, D 985b. po organization of chloro- plast buxo Annual Rev. Genet. 19: 325-354. . 198 Isolation and structural vue of 67 a. i d DNA. Methods in Enzymol. 118 . 1986b. ra DNA and phylogenetic om Pp. 63-80 in S. K. Dutta (editor), DNA matics, ied E Plants. CRC Press, Boca Raton. Florida. & D. B. STEIN. 1986. Conservation of chlo- ] vascular plants. Curr. Genet. ‘10: 823-833. —— & D. Zamir. j962, Chloroplast DNA evolution and phylogenetic relationships in Lycopersicon. Proc. wl 10. Natl. Aca o . Jo ORGENSEN & pecas 1985. Chloroplast DNA variation and ev n in Pisum: patterns of TA and spese. analysis. Ge- netics 109: "2 , C. R. cl D. B. Conen & T. J. ORTON. Volume 75, Number 4 1988 Sytsma & Smith 1237 Onagraceae—DNA & Morphology 1983. Chloroplast DNA evolution and the Z " PATET Brassica species. Theor. Appl. Gene : -189. PERL- TREVES, R. & E. Gatun. 1985. The Cucumis plastome: al map, intrageneric variation and ria ia relationships. Theor. Appl. Genet. 71: 429. > D. Zamir, N. Navor & E. Gatun. 1985. Phylogeny of p sad: based on isozyme variability and its peg ee with plastome phylogeny. Theor. enet. 71: 1983. a Ev idence for diploid — of Clarkia. Genetics 105: 42 Raven, P. H. 1979. A survey of P gru s in Onagraceae. New Zealand J. Bot. 17: 575-593 . 19 nagraceae as a model of plant evo- lution. Pp. 85-107 ¿n L. D. Gottlieb & S. K. Jain (editors), Plant Ex Biology. Chapman & Hall, London. RiEsEBERG, L. H., D. E. Sorris & J. D. PALMER. 1988. A molecular reexamination of introgression between Helianthus annuus and H. bolanderi (Compositae). Evolution 42: 227-238 SALTS, Y., R. G. HERRMANN, N. PELEG, U. Lavi, S. IZHAR, R. FRANKEL & J. S. BECKMANN. 1984. Physical mapping of plastid DNA variation among eleven Ni- cotiana species. Theor. Appl. Genet. 69: 1-14. SKVARLA, J. J., P. H. Raven, W. F. Cuissok & M. SHARP. 1978. An ultrastructural study of viscin threads in Onagraceae d ains Pollen & Spores 20: 5-143 SMALL, E., I. J. Bassett, C. W. Crompton & H. Lewis. 1971. Pollen ceri in Clarkia. Taxon 20: 739- 46. Sorris, D. E., P. S. Sorris & B. D. Ness. Chloroplast DNA variation and multiple origins of autopolyploidy in = micrantha (Saxifragaceae). Evolution (in p SOLTIS, P. s. . SoLTIS & L. D. GOTTLIEB. 1987. Salt rent gene duplications in Clarkia their phylogenetic implications. PAUP, Phylogenetic Analysis Using Parsimony. Version 2.4. Illinois Natural His- tory Survey, Champaign, Illinois. Sytsma, K. J. & L. D. GOTTLIEB. 1986a. Chloroplast DNA evidence for the origin of the genus Hetero- gaura from a species of Clarkia (Onagraceae). Proc. Natl. Acad. U.S. E 83: 5554-5557. & Chloroplast DNA evolution and phylogenetic relationships in Clarkia sect. Peri- M a (Onag ae). Evolution 40: 1248-1261. rae i 1985a. Phylogenetics of the ) i (Genti ) plex in Pan- L iuis ama utilizing DNA triction fi g t ly Evo- lution 39: 594-608. . 1985b. Genetic variation, differ- entiation, and evolution in a species complex of trop- ba ical shrubs based on isozymic data. Evolution 39: 582-593. TEMPLETON, A. R Convergent evolution and DNA sequences. Pp. 151-179 Statistical Analysis of DNA Sequence Data. Marce Dekker, Inc., New York. Tose, H. € P. H. Raven. 1985. The histogenesis and evolution of integuments in Pia psi . Ann. Mis- 468. souri Bot. Gard. 72: 451- & Evolution of polysporan- giate anthers in Onagraceae. Amer. J. Bot. 73: 475- 488. TSUNEWAKI, K. & Y. OGIHARA. 1983. The molecular basis of genetic diversity among cytoplasms of Trit- icum and Aegilops species. II. On the origin of polyploid wheat cytoplasms as suggested by chloro- plast DNA restriction fragment patterns. Genetics 104: 155-171. VasEK, F. C. 1958. The relationships of Clarkia exilis to Clarkia unguiculata. Amer. J. Bot. 45: 162. WILLIAMS, C. A., J. J. FronczYk & J. B. HARBORNE. 83. Leaf ‘flavonoid and other oe shua, as indicators of parentage in six ornamental Fuchsia species and their hybrids. Consuet 22:1953- 7 ZIMMER, E. A, C. RIVIN & V. Warmor. 1981. A DNA isolation procedure suitable for most higher plant species. Pl. Molec. Biol. Newslett. 2: 93-96. MODES AND TEMPOS IN THE EVOLUTION OF NUCLEAR RIBOSOMAL DNA: NEW CHARACTERS FOR EVOLUTIONARY STUDIES AND NEW MARKERS FOR GENETIC AND POPULATION STUDIES! Richard A. Jorgensen** and Paul D. Cluster? ABSTRACT tempo of evolutionary change determines in what manner any class of characters is informative levels ranging from the intraspecific to the interfamilial, we show that pla eleven classes of characters that can be distinguished by comparisons at the DNA ich taxonomic levels. Here we describe and summariz e some fundamen ant ribosomal level. These classes are temporal NA and physical subsets of three basic modes of variation: length variation, single base pair substitution, and D modification. We also discuss the impact of length variants on population genetic studies and the implications of these studies for understanding the molecular mechanisms of rDNA evolution. Because DNA is the richest and most unambig- uous source of genetic variability, information on its evolution is fundamentally important to evolu- tionary biology. Research into the evolution of DNA is still in its infancy, and workers studying DNA variation are still faced with (1) cataloging both the classes of DNA sequences (characters) that are found in the genomes of various organisms and the ways (modes) in which th ha vary among organisms, and (2) ne the approximate rate (tempo) of change in the different character classes. From this fundamental information it is possible to begin to ascertain at which phylogenetic level a particular character is useful in reconstructing phy- logeny. Here we examine modes and tempos of evolution in nuclear-encoded ribosomal DNA. The available technology allows us to estimate the tem- po of evolution in three basic modes: length vari- ation, base pair substitution, and nucleotide mod- ification. PHYSICAL AND GENETIC DESCRIPTION OF rDNA Ribosomal DNA, or rDNA, is the set of DNA sequences that directs the synthesis of ribosomal RNA. Each haploid nuclear genome of a higher plant cell typically contains 1,000 to 10,000 copies of ribosomal DNA (Ingle et al., 1975), a range roughly twenty-fold higher than in animal genomes. Copies of rDNA exist in long tandem arrays at one or a few chromosomal locations. Within a species, the number of copies of rDNA varies by as much as four-fold (Cullis & Davies, 1975; Long & Dawid, 1980). Unequal crossing-over is one likely mech- ' Editor's note: This and the "pri article by H. J. Price are based upon the authors’ presentations at the of a sy r the manusc ks S. meetings as part 982 A.I.B. Botanical Garden. Afte were pre osium lcd planned for publication in the Annals of the Missouri red, plans were changed, a mposium was never taste Publication here of the 1987 Missouri Botanical Garden Annual Systematics Symposium, dealing with the s dinis matter, provides an opportunity to present these updated papers as complements to the more Sube sym ? De epartment of Canis, University of California, Davis, California 95616, 3 Present address: Advanced Genetic Sciences, Inc., U.S. A ANN. U.S.A. 6701 San Pablo Avenue, Oakland, California 94608, Missouni Bor. GARD. 75: 1238-1247. 1988. Volume 75, Number 4 Jorgensen & Cluster 1239 1988 Evolution of Nuclear rDNA ——— — P i do ¡04 ) "d M P d dis 7 ka Í "s P ñ ie (í 18S 5.85 258 > — RIA os ITS IGS FiGURE 1. Schematic representation of rDNA repeat structure. 18S, 5.8S, and 25S refer to ribosomal RNAs. ITS and IGS refer to internal transcribed spacer and intergenic spacer, respectively. anism responsible for variations in copy number (Szostak & Wu, 1980). Alth heterogeneity among copies of rDNA within indi- viduals (see below), the rDNA repeat units of an individual plant are highly homogeneous. That is, while several types of rDNA repeat unit may be found in a single plant, many hundreds of repeat units are identical as assayed by Southern blot analysis. This homogeneity was first observed in comparisons within species contrasted with com- parisons between species and is presumably the result of concerted evolution of rDNA repeat units, as explained by Arnheim et al. (1980). Similarly, in species where rDNA is found to reside at two or more genetic loci, repeat units are found to be quite homogeneous within each locus. Thus, loci can usually be distinguished by their repeat types, and homogeneity is greater within loci than between loci (Dvorak & Appels, 1982; Saghai-Maroof et al., 1984). The physical structure of higher plant ribosomal DNA (Fig. 1) is similar to that in other higher eucaryotes (Long & Dawid, 1980, for review). The three ribosomal RNA coding regions lie in the or- der 5’, 18S, 5.85, 25S, 3', and are transcribed as a single large precursor, which is processed sub- sequently to the mature rRNA forms. Several hundred base pairs of DNA separate the 18S cis- tron from the 5.85 cistron and the 5.85 cistron from the 25S cistron. These two intercistronic re- gions are referred to as internal transcribed spacers (ITS). The region separating the transcription units of adjacent rDNA repeats is called the intergenic spacer (IGS, formerly NTS or nontranscribed spacer; Dover et al., 1982) and in most plants ranges in length from one to eight kilobase pairs (kb). A tandemly repeating sequence comprises part of the IGS region. This sequence varies in- terspecifically in length, ranging generally from ough there is some 100 to 200 bp, while within species its length varies only slightly. The length of this subrepeat has been shown to be 130 bp in wheat, 180 bp in peas, 325 bp (comprised of two copies of a 155-bp sequence and one 14-bp sequence) in broad bean, 115 bp in both barley and oats, and 200 bp in maize Appels & Dvorak, 1982; Jorgensen et al., 1982; Yakura et al., 1984; Saghai-Maroof et al., 1984; Cluster et al., in prep.; and McMullen et al., 1986, respectively). The number of these elements within a given rDNA repeat unit is variable, and thus the overall length of the IGS is variable, within and between populations. This variability in length of the IGS is discussed in detail below An individual plant's rDNA array is often het- erogeneous with respect to the three basic modes of variation: length, nucleotide sequence, and base modification (e.g., Siegel & Kolacz, 1983; Appels & Dvorak, 1982; Waldron et al., 1983; Jorgensen et al., 1982, 1987). It should be noted that a fourth mode of rDNA variation occurs, namel ~ variation in the copy number of rDNA per haploid genome; because it is a quantitative character it is rarely measured. rDNA copy number is unlikely to be informative taxonomically because it is ex- tremely variable within species, although in genetic analyses it may have some utility. In Figure 2 the three principal modes of variation are illustrated for a single individual of the garden pea (Jorgensen et al., 1987). This individual carries a minimum of three types of rDNA repeat units, and each of the three is distinguished by the three modes. First, each type of repeat has a different overall length due to variation in the number of 185-bp subre- peats in the IGS. Second, nucleotide substitutions are evident in several regions of the rDNA repeat unit. Type “L” repeats (see Fig. 2) carry EcoRI sites in two of their nine subrepeats that are not present in types *'S" or “C,” and in the nonsubre- 1240 Annals of the Missouri Botanical Garden (50%) (50%) Nco Bgl I Bg! II Sph kb i L . 96 (10%) (10%) Nco Bgl II BglII EcoRI M | “aná L (80%) (78096) BglII Bgl II Sph 88 mimi _ ¢ (50%) (50%) Nco Bgl II Bgl II Sph 8.6 | | | S FIGURE 2. Maps of four pea rDNA repeat units. L, C, and S refer to distinct repeat types found in a single Al. pea plant. E refers to another repeat type found in Pisum elatius. Nco and Sph refer to Ncol and Sp peat region of the IGS types “C” and “S” carry an Sphl site not found in type “L.” Within the rRNA coding sequence, an Ncol site in the 185 gene is present in both type "L'and "S," but absent in the cloned type “C” repeat. Third, the BglII cleavage sites of types “L”and "S" differ in their degree of apparent base modification such that only 10% of the BglII sites of “L” repeats are cleaved, while 50% of the BglII sites of “S” repeats are cleaved. TEMPORAL ANALYSIS OF PLANT rDNA VARIATION For a given mode of sequence evolution, the taxonomic level at which any segment of DNA is useful for making phylogenetic determinations is determined by the tempo at which that segment of DNA varies. Tempo can be estimated by ana- lyzing DNA variation in species from several levels within an accepted taxonomic hierarchy. We have chosen nine legume genera for study: three (Vicia, Pisum, and Lathyrus) are from the tribe Vicieae and the rest (Medicago, Trifolium, Lupinus, Wis- teria, Cytisus, and Phaseolus) are each from a different tribe. Seven genera are represented by a single species. Vicia is represented by five species, Pisum by four. For V. sativa and P. sativum, four and twenty, respectively, distinct isolates were examined. Outside the legume family we have com- pared the rDNA of wheat (Triticum aestivum) and pumpkin (Cucurbita pepo). Postulated phyloge- netic relationships between species in this hierarchy are depicted in Figure Using the cloned rDNA repeat unit from Pisum sativum and detailed physical maps of both the cloned repeat and the pea nuclear genomic repeats described above, we performed nitrocellulose blot analysis (Southern, 1975) of rDNA sequences in the genomic DNA of each species. “Southern anal- ysis”” requires the use of a specific probe homol- ogous to DNA sequences being analyzed. Different regions of the rDNA repeat unit were analyzed independently by use of seven different purified restriction fragments as probes (Fig. 4). The ITS region was analyzed by DNA sequencing because it was too small to analyze effectively by Southern analysis. MODES AND TEMPOS OF VARIATION IN DIFFERENT rDNA REGIONS A. Base Modifications We have characterized in some detail two types of base modification in plant rDNA. They are dis- tinguished both by sequence specificity and by de- gree of variability among taxa, as will be explained here. Most common is an evolutionarily conser- vative type of base modification typified by the BamHI site in the 25S gene. This site is modified in about one-half of the rDNA repeats of all the legume species in our survey, as well as in several other species (Gerlach & Bedbrook, 1979; Golds- borough et al., 1981; Jorgensen et al., 1982; Siegel & Kolacz, 1983). Siegel & Kolacz (1983) have postulated that this methylation is due to a CCG sequence of which the BamHI site (GGATCC) is a part. (That only one-half the sites are cleaved by Volume 75, Number 4 1988 Jorgensen & Cluster 1241 Evolution of Nuclear rDNA length point voriation methylation mutation IGS es ITSrRNA NC C IGS rRNA ae a | | i| | ET *......1..... v * ç FIGURE 3. Schematic drawing of evolutionary doors ` SUM among species analyzed and e taxonomic levels at which the eleven characters classes are pr igs sed E E aha in [nah yd phylo e taxonomic levels indicated are intraspecific dics sativum (s), us (e), e (h), P. fulvu m (f) , and Vicia sativa (s)], intrageneric (Vicia and Pisum), aon bee pr LI (Fabaceae) , within an giosperms. The cladogram is not meant to repr servative; C to conservative. BamHI appears to be due to the fact that meth- ylation occurs at either but not both C residues.) Methylation of CG and CXG sequences has been observed in all plants investigated to date. It is presumed that many of the other enzyme cleavage sites that contain CG or CXC sequences could be subject to evolutionarily conservative modification. However, the extreme conservatism of the 25S BamHI site modification is a specific example of the general phenomenon of plant cytosine modi- fication, and it is unwise to generalize from this. In fact, within or near various structural genes a substantial number (and perhaps a large fraction) of CG and CXG sequences are unmethylated in a variety of plants, and the possibility of variation in plant XG modifications certainly exists. In contrast to the BamHI modification, modi- fication of the BglII sites in pea rDNA is apparently variable by degree among individual pea plants, just as it is variable among rDNA repeats of the same plant (Jorgensen et al., 1982, 1987). Bglll sites do not contain CG or CXG sequences but could be part of CXG sequences. It is not clear whether the variable modification of BglII sites in rDNA is due to variation in (a) modification by the resent relationships but is presente levels being compared. IGS refers "s intergenic spacer; ITS refers to internal transcribed space ed only to bans D taxonomic r; NC to noncon- CG, CXG system, (b) modification by another sys- tem, or (c) sequences adjacent to the site. Adenine modification prevents cleavage by certain restric- tion enzymes, but these have not been analyzed for variability among plants. B. Single Base Pair Substitutions 1. Coding Regions. The coding regions for mature rRNAs were compared in two ways: b comparing restriction maps of cloned repeats from pea, wheat, and pumpkin (Jorgensen et al., 1987) and by comparing Southern blots of legume species rDNA using probes A, B, C, and D (Fig. 4). The 18S genes of pea and wheat were found to differ at three of ten six-bp cleavage sites (at least three of 60 bp), while the genes of wheat and pumpkin differ at five of nine, and the genes of pea and pumpkin at two of eight. The 5’ end of the 255 gene shows no site conservation in comparisons of these three species, which is consistent with the fact that this is one of the last-conserved regions in comparisons among frog, yeast, and slime mold rDNA (Gerbi et al., 1982). The rest of the 25S gene shows substantial similarity among species: 1242 Annals of the Missouri Botanical Garden Hin d Xmn Xba Tth Tth Tth Tth — va ic, oa i 18S 5.8S 255 A—!460 E 1440 x E 2300 E60 Map indicating restriction PRU (A-G) used to study rDNA variation. Numbers indicate rs. Xmn, X FIGURE 4. bud of fragments in base pai pea and pumpkin differ at three of ten sites, pea and wheat at eight of twelve, and wheat and pump- kin at five of ten. Although it is possible to align and compare the restriction maps of the pea, wheat, and pumpkin coding regions, a statistically signif- icant number of mutations have not been analyzed to permit phylogenetic conclusions. outhern analysis of different legume species reveals very little seq divergence in the rRN coding sequences, even after use of all available enzymes insensitive to base modification (Jorgensen et al., 1982). Figure 5 summarizes this analysis with a comparison of pea, vetch, and bean, illus- trating the only two cleavage site mutations that could be detected in this survey of 19 cleavage sites (for a survey of 114 bp). The degree of se- quence divergence in this rDNA region is at least several-fold less in the legume family than among peas, pumpkins, and wheat. It is important to recognize the limitations in- herent in the use of restriction enzyme analysis of plant rDNA for phylogenetic investigations. First, the choice of restriction enzymes for nuclear DNA analysis in plant genomes is more limited than for animal genomes or for plant organellar genomes due to the fact that plant nuclear DNA is meth- ylated at most CG dinucleotides and CXG trinu- cleotides (Gruenbaum et al., 1981), and because a, Rsa, and Tth refer to Xmnl, Xbal, Rsal, and Tth1111. many restriction enzymes that cleave sequences containing CC, CXG sequences do not cleave if these sequences are methylated. Thus, analysis of genomic rDNA by Southern blot is quite limited relative to analysis of rDNA clones using restriction enzymes with respect to the number of variants that can be detected. Cloned sequences are not necessarily a good alternative because the use of single clones from an array of thousands entails the risk of not being representative. Second, be- cause rRNA coding sequences are only 5.5 kb long, in contrast to chloroplast DNA which amounts to about 150 kb (see Palmer et al., relatively few cleavage sites are available and rel- this volume), atively few variants can be detected. The obvious solution to this problem is to utilize rRNA sequenc- E ing so, Zimmer (this symposium, not pubhehed here) has shown that the conservative nature of the coding region is extremely useful in phylogenetic comparisons between distantly related genera and closely related families. 2. ITS Region. The ITS region, because it is small, also cannot be analyzed well with restriction enzymes. The DNA sequence of the 5.85 gene and its surrounding ITS sequences has been determined in pea and lupine (Jorgensen & Hess, unpubl.; Rafalski et al., 1983). A schematic comparison of - jr d d 1l Tiwi van My lll)’ DL LL LLLI. se dI HL Y ji 11 L! FIGURE 5. Teh 1111; K, Kpnl. Maps comparing coding regions of pea (Pisum sativum) , vetch (Vicia sativa) , and b vulgaris) rDNA. Symbols are coded as follows: N, Xmnl; ean (Phaseolus A, Xbal; M, BamHI; Q, Sstl; D, BstEll; E, EcoRI; T, Volume 75, Number 4 1988 Jorgensen & Cluster 1243 Evolution of Nuclear rDNA Eco Cla RV I t y % divergence: 7 5.85 RNA 5'ITS FIGUR between Dum sativum and Lupinus luteus rDNA clones these sequences is shown in Figure 6. DNA in the ITS was observed to change at two distinct rates, both faster than that within the 5.85 gene. The 5.85 rRNA of Vicia faba has been sequenced (Tanaka et al., 1980) and differs from the pea 5.8S rRNA by 2%. 3. IGS Region. of the nonsubrepeat segment of the IGS region reveals many cleavage site variants within legume genera (Jorgensen et al., appear to have some utility in assessing phyloge- netic relationships and in genetic studies of inter- fertile populations and species (Zimmer et al., in prep.). In the IGS subrepeats, DNA sequence and restriction pattern analyses indicate that subrepeat sequences are variable within and among individual genomes in a species (Appels & Dvorak, 1982; Jorgensen et al., 1987). This variation in single base pair substitutions in the ICS is of interest primarily within species. Because this variability is very difficult to detect with restriction analyses of total genomic DNA, it is of limited utility, except in studies of the evolution of the subrepeat itself. Restriction enzyme analysis These variants C. Length Variation 1. Coding Regions. The ribosomal RNA transcription unit was surveyed for length variants by monitoring four restriction fragments (A-D) in- dicated in Figure 4. Fragment A, a 1,460-bp Xbal- TthI fragment, lies entirely within the 18S gene; fragments B (2,300-bp Tth fragment) and D (420- bp Tth fragment) lie entirely within the 25S gene; fragment C, the 1,075-bp TthI fragment, carries the entire ITS region and the 5.85 gene within it, as well as short portions of the 18S and 258 genes. Within the legumes, no length variations were ob- served (detection limit, 50 bp) in any of the three fragments that lie within the rRNA genes. The 5.85 rRNA coding regions of pea, broad Map of 5.8S rRNA gene and ITS regions. | Numbers indicate % divergence in nucleotide sequence bean, and lupine have now been sequenced (Hess & Jorgensen, unpubl.; Tanaka et al., 1980; Ra falski et al., 1983). Comparison of these sequences shows that the 164-bp 5.85 gene differs in length between pea and broad bean by only one base pair and between pea and lupine by two adjacent base pairs. Length variation of this sort is very likely to be found also within the 18S and 25S genes by DNA sequence comparisons, but not by restriction fragment comparisons, as the differences are too small to detect by agarose gel electrophoresis. 2. ITS Region. Overall length variation in the ITS, as monitored by changes in the 1,075-bp Tthl fragment, is much more prevalent than are length changes in the three coding sequence restriction fragments. The pea ITS 1,075-bp Tthl fragment detects fragments varying in size from 1,000 to 1,200 base pairs among the nine legume genera surveyed. At least six of these size classes are distinct from all the others. Within Pisum and Vicia no length variation was observed. Thus, length variants of 50 bp or greater appear to be restricted mostly to the intergeneric level, at least in the tribe Vicieae. Whether small variants ever occur within these genera and whether observed variation re- sults from the accumulation of many small variants or few large variants remains to be determined. For particular ITS length variants to be of practical use in studying relationships among genera, they will have to be large and rare, rather than small and common. Particular length variants in chlo- roplast DNA have been quite useful (in conjunction with point mutations) in developing chloroplast DNA phylogenies (Palmer et al., this volume). 3. IGS Region. By far the most variable re- gion of the rDNA repeat unit is the subrepeat- containing region. Length variants of restriction fragments carrying this region almost always differ by a multiple of the length of the subrepeat. For 1244 Annals of th Missouri EOS Garden example, 14 distinct fragment lengths were ob- served in a sample of 12 pea lines, each length differing from the others by a multiple of 180 bp. Similarly, detailed intraspecific surveys of IGS vari- ability among hundreds of individuals have dem- onstrated variation produced by 15 increments of 115 bp in barley (Saghai-Maroof et al., 1984) and by 17 increments of 115 bp in wild oats (Avena barbata) (Cluster et al., in prep.), indicating that the subrepeat length is 115 bp in each of these species. This large number of variants results in a large number of IGS phenotypes observed within and among populations (see below). For example, 11 distinct phenotypes were observed among the 12 pea lines. The next most length-variable region is the part of the IGS region without subrepeats. We have monitored this region in Pisum with a 960-bp re- striction fragment (G) produced by combined cleav- age with HindIII and Xbal and with two Xmnl fragments (E and F) of 310 bp and 670 bp (Fig. 4). Among ten pea lines that showed extensive variation in length of IGS-a, we found only two IGS-b types. These probably resulted from a single substitution event, this resulting in several restric- tion site differences as well as a 100-bp length difference. This finding illustrates the need for cau- tion in interpreting restriction patterns and consid- eration of the possibility that restriction site vari- ants in the IGS-b may be the result of deletions, insertions, or substitutions rather than point mu- tations. Interpretation of the molecular basis of mutations in this region in the absence of direct DNA sequence information is prone to error due to the small length changes potentially involved. D. Summary of Tempos and Modes Based on the results described above, Figure 3 illustrates the taxonomic levels at which the 11 identified character classes may have some utility in evolutionary genetic studies. It should be noted that characters in the single base pair substitution (point mutation) mode will be best detected by sequencing of rRNA or cloned rDNA, not by re- striction analysis, as is demonstrated by Zimmer (this symposium, not published here). Clearly the size of the rRNA coding region indicates that this region will provide the greatest number of char- acters and so will be the most informative. Further, we would expect this region to be useful at levels ranging from the intergeneric to the interfamilial. It would be interesting to test the utility of the nonrepeated IGS region for intrageneric compar- isons, although length variations (i.e., additions and deletions) might preclude this possibility if they are found to occur so frequently as to obscure sequence similarities. Characters in the base modification and length modes will be best detected by restriction analyses of total genomic DNA. It appears that character classes in these two modes will be useful primarily in intraspecific genetic studies, as de- scribed in the next section. VARIATION WITHIN AND AMONG POPULATIONS OF A SINGLE SPECIES The rDNA spacer length (sl) phenotype of in- dividual wild oat plants usually is comprised of 4— 10 variants, out of 17 variants known in the species, often in widely varying copy numbers. From among over 500 individuals sampled, at least 40 distinct phenotypes were distinguished by scoring the most abundant sl variants (Cluster et al., in prep.). Vari- ation of rDNA sl phenotypes among populations of wild oats in California closely tracks previously established patterns of differentiation identified by allozymes, morphological characters, and quanti- tative characters. Furthermore, the degree of vari- ability suggests that it may be possible to identify and differentiate populations on the basis of rDNA variability alone nearly as accurately as with the available set of variable allozyme loci. Similarly, in 75 samples of barley, 15 distinct sl phenotypes could be distinguished (Saghai-Maroof et al., 1984). Most of these were comprised of two or three sl variants. The level of intraspecific polymorphism in the IGS is, therefore, extremely high. In the case of wild oats, the ability of sl variants to dis- tinguish in detail among and between populations is probably the result of two genetic properties of rDNA variants in this species: (1) that these vari- ants lie at a minimum of four independently seg- regating loci and (2) that each locus contains hundreds or thousands of repeat units which can be of more than one sl variant type Another result of genetic analysis of rDNA is the observation of nonrandom distribution of sl variants in several species, which suggests that genetic exchange occurs less frequently between than within nucleolus organizer regions. This sit- uation occurs in barley (Saghai-Maroof et al., 1984), wheat (Dvorak & Appels, 1982), pea (Ellis et al., 1984; Polans et al., 1986), and mouse (Arnheim et al., 1982); however, random distribution has 1981). In wild oats the most abundant sl variants were been reported in humans (Krystal et al., present in nearly all isolates, including both parents of the single F2 analyzed, and so it is not possible to assess accurately whether these variants are Volume 75, Number 4 1988 Jorgensen & Cluster 1245 Evolution of Nuclear rDNA distributed randomly or nonramdomly among loci, even though five less-abundant variants are non- randomly distributed. In species where a degree of nonrandomness is observed, rDNA appears to be a new and useful genetic marker (Saghai-Maroof 1984). However, the multilocus nature of rDNA spacer length variation may place a severe limitation on its use in population genetics because of the difficulty in determining each plant's ge- notype, except for those genomes possessing only a single major nucleolus organizing region (e.g., tomato and corn). Still, the great amount of phe- notypic diversity will clearly be useful. Also noteworthy is the observation that the com- posite frequency distribution of rDNA sl variants in California wild oats shows a nearly Poisson dis- tribution of sl variants centered at sl variant 8. There at at least three ways to explain this distri- bution. First, it could be the result of classical forces in population divergence such as genetic drift and/ or selection on loci at or correlated with rDNA. Second, it could be the result of stochastic molec- ular processes, perhaps involving DNA rH or repair. Third, it could be determined by t function of the subrepeat. The first explanation is based on the fact that in populations that reproduce substantially by self- ing, a correlational structure is imposed on all com- ponents of the genome, allowing selection at each locus to affect allelic frequencies at all other loci in the genome. Thus, sl variant Foroen at rDNA loci must reflect the sel t y other loci and could be determined by these forces (Saghai-Maroof et al., 1984). The second explanation considers whether the observed distribution could be simply a conse- quence of the molecular mechanisms that create new spacer length variants. New variants can ap- pear in evolutionary time by mechanisms such as unequal crossing-over, resulting in repeated cycles of amplification and contraction of arrays of both repeats and subrepeats of rDNA. Accordingly, one can hypothesize that the number of subrepeats in the intergenic spacer region would be determined by a feature(s) of these mutational mechanisms whereby very long subrepeat arrays are more likely to be shortened than lengthened and short arrays are more likely to be lengthened than shortened, resulting in a balance in wild oats, for instance, at slv-8. Since various cellular processes might affect this mechanistic optimum, it could be possible for et al., ol man species to differ in their optimum number of sub- repeats, were such a mechanism the only one op- erating on the distribution. The third explanation considers what function the IGS might have, based on recent observations on the transcriptional and structural nature of the subrepeat elements (Reeder, 1984; Flavell et al., 1987). Briefly, it is hypothesized that the IGS subrepeats function analogously to enhancer se- quences, increasing the transcription of the repeat unit(s) to which they are adjacent. It has been observed that rDNA repeat units with more sub- repeats are transcribed with strong preference over units with fewer subrepeats, probably due to an interaction between subrepeats and some positive transcription factor. Furthermore, loci with repeat units having more subrepeats show nucleolar dom- inance over loci having fewer repeats. These ob- servations provide a simple explanation of how natural selection for longer sl variants might occur. Of course, subrepeats apparently do not increase to many tens or hundreds of copies. Therefore, we must also hypothesize that too large a number of subrepeats can be deleterious. Perhaps multiple subrepeats would sequester a transcription factor not only from rRNA promoters in unlinked loci but also from promoters in adjacent repeat units (see Reeder, 1984, for detailed discussion of this model). This situation would likely lead to a reduction in efficiency of rRNA transcription and thereby be deleterious to the individual. Thus, it is possible that natural selection can directly mold the rDNA sl variant pattern and influence the frequency dis- tribution seen in wild oats. SUMMARY Eleven classes of useful characters have been identified for plant nuclear ribosomal RNA genes. These classes and their approximate rates of evo- lution are as follows: (1) The length of the plant ribosomal DNA re- peat unit is highly variable within most species and this variability has great utility in studies of pop- ulations. A 100-200-bp sequence that is repeated many times in tandem in the IGS region of rDNA forms the molecular basis for this variation in that the number of tandem copies of this sequence dif- fers among individuals as well as among rDNA repeats within an individual. Studies of these vari- ants appear to be helpful in elucidating the molec- ular mechanisms of rDNA evolution. (2) The nonrepeating portion of the IGS region is less variable than is the subrepeat region, but is variable in length within the genera Pisum and Vicia; this variability might have utility in assessing specific relationships within such genera. (3 and 4) Ribosomal RNA coding sequences are invariant in length in restriction pattern analyses, 1246 Annals of the Missouri Botanical Garden but length variation is observed in the internal transcribed spacer (ITS) region. The frequency at which length variations in the ITS are detected is lower than that for length variation in either seg- ment of the IGS. It is not likely that length variation in this region will be useful phylogenetically. (5 and 6) Modification of Bg 5-merous Ovary, style Tectum of pollen wall Subepidermal floral laticifers Nucellar tissue consumed early Endothelium present endospermous relatively thick narrow with particular protein bodies reported in several genera pauciflorous, cymules, solitary flower superior to inferior, a single style present, different types wider with starch only present none plas Gi pes or panicle-derived inferior, ES. stylodia relatively thin one persisting longer none exendospermous nous, and the stamens are generally numerous. Furthermore, the ovary generally has parietal pla- centae. No single genus of Flacourtiaceae to my knowledge has the combination of character con- dition of the Anisophylleaceae, and it is solely due to the variability (or heterogeneity) of the Fla- courtiaceae that it agrees, in most respects, with the Anisophylleaceae. ANACARDIACEAE Anacardiaceae are members of Sapindales. They are rich in tannins and have well-developed schi- zogenous or lysigenous ducts or channels with res- ins, which is not the case with the Anisophylleaceae. The leaves are variable in Anacardiaceae but are more frequently compound than simple, as in An- isophylleaceae. The small flowers are reminiscent of those in Anisophylleaceae but are more often pentamerous; they vary from hypo- or perigynous to epigynous and are often diplostemonous, with free stamens and a well-developed disc. The carpels are often solitary (sometimes several and free from each other) but when fused are generally three. The fruit is usually drupaceous. Anacardiaceae, apart from the resin canals, compound leaves, and numerical conditions of the flower (especially gy- noecium), agree fairly well with Anisophylleaceae. CONCLUSION Anisophylleaceae are difficult to place, as most of their characteristics are of very common oc- currence. Rosales(-Saxifragales-Cunoniales) seem to be the group in which they would be fairly well placed, but without obvious links. These orders have a clearly temperate(-boreal) concentration, phological (Tobe & Raven, 1987c) and embryo- logical conditions in Anisophylleaceae agree completely with those common in Rosaceae sensu stricto, for example: exendospermous seeds and a bitegmic, crassinucellate ovule with nuclear en- dosperm formation. That in both families the ovules are bitegmic in some and unitegmic in other genera is a coincidence. The basic chromosome numbers, x = 7 or x = 8, are also present in Rosaceae sensu stricto, especially the former number (Raven, It is probable that Anisophylleaceae comprise a rather isolated family evolved from ancestors shared between those in Rosales, Cunoniales, and Saxi- fragales. It is also probable that these were not very remote from the ancestors of Myrtales, al- though the last order is distinctive in several re- spects, including the anatomical features men- tioned above PROPOSED CLASSIFICATION Ordinal composition around Rhizophoraceae: GERANIALES: Zygophyllaceae, Nitrariaceae, Pegan- aceae (position uncertain), Balanitaceae, Vivi- aniaceae, Geraniaceae, Ledocarpaceae, Bieber- steiniaceae, Dirachmaceae, Ixonanthaceae, umiriaceae, Erythroxylaceae, Linaceae, Lepidobotryaceae, Hugoniaceae, Ctenolophonaceae, xalidacea CELASTRALES: Celastraceae, Elaeocarpaceae, Rhi- zophoraceae 1276 Annals of the Missouri Botanical Garden Ordinal composition around Anisophylleaceae: CUNONIALES: Cunoniaceae, Baueraceae, Brunelli- aceae, Davidsoniaceae, Eucryphiaceae ROSALES: Crossosomataceae, Rosaceae, Neurada- ceae, Anisophylleaceae, Malaceae, Amygda- laceae (plus perhaps some smaller families such as Rhabdodendraceae) SAXIFRAGALES: Saxifragaceae, Penthoraceae, Vah- liaceae, Francoaceae, Greyiaceae, Brexiaceae, Grossulariaceae, Iteaceae, Cephalotaceae, Cras- sulaceae, Podostemaceae. LITERATURE CITED Airy SHaw, H. K. 1966. J. C. Willis: A Dictionary of the Flowering Plants and Ferns, 7th edition. Cam- bridge Univ. Press, Cambridge. 3. J. C. Willis: A Dictionary of the Flow- ering Plants and Ferns, 8th edition. Cambridge Univ. s, Cambridge. Am A. 1931 der . Studien über die Embryologie Familien Celastraceae, Oleaceae e Apo cyna- ceae. Acta Univ. Lund N. F., Avd. 2, 27: 1-112. ARENA, M. R. & F. Orsini. 1973. das Florae ‘yea a 24. Rhizophoraceae. Webbia 28: 135- BALGOOY, ‘M. M. J. v 1982. A revision of Sericolea Schlechter (Elaeocarpaceae) Blumea 28: 103-141. BEHNKE, H.- . Sieve element characters. Nordic J. Bot. 1: 381-40 1982. owls element plastids of Cyrillaceae, cane of dria PV plastids. Pl. Syst. Evol. -39. 984. Ultrastructure of the sieve-element "e of Myrtales and allied groups. Ann. Missouri Gard. 71: 824-831. : Sieve- element plastids and systematic relation- ships of RI and allied groups. Ann. Missouri Bot. Gard. (this a Bick, I. R., J. B. BREMNER, N. W. Preston & J. C . 1971. Peduncularine, an indole- ar dina ‘alkaloid. J. Chem. Soc. Chem. Comm. 1971: BOESEWINKEL, F. D. a of Linum usitatissimum L. Acta 29: 17-32 Development of ovule and Bot. Neerl. 1985. Development of ovule and seed-coat in peers (Oxalidaceae) with some v on related genera. Acta Bot. Neerl. 34: 413 ————— & J. GEENEN. 1980. De Rd of ovule and seed-coat of riae coca Lamk. Acta Bot. Neerl. 29: 231- BOSSER, J. & R. PERNET. 1957. Etudes chimiques sur e ge hehe ie de Madagascar. Naturaliste Malgac BREWBAKER, 1 L. 1967. The distribution and phylo- genetic significance of binucleate and trinucleate d ces grains in the angiosperms. Amer. J. Bot. 069-1083. BULL, L. B., C. . CULVENOR & A. T. Dick. The Pyrrolizidine Alkaloids. Amsterdam CHENERY, E. M. 1948. Aluminium in the plant world, I. Kew Bull. 1948: 173-183. 1968. Cook, M. T. 1907. The embryology of Rhizophora mangle. ui Torrey Bot. Club 34: 271-277. Corner, E. J. The Durian Theory or the origin of di modern tree. Ann. Bot. II, 13: 367- 4 —— 76. The Seeds of Dicotyledons, 1-2. Cam- bridge Univ. Press, Cambridge CnoNQUIST, A. 1968. The Evolution and up ein of Flowering Plants. Houghton Mifflin Co., Bos —— 1981. An Integrated System of Gaston of Flo owering Plants. Columbia Univ. Press. New York. CULVENOR, C. C. J. 1978. Pyrrolizdine Sika oc- currence and systematic importance in angiosperms. Bot. Not. 131: 473-486 DAHLGREN, R. 1975. A system of classification of the angiosperms to be used to demon es the distri- bution of characters. Bot. Not. 128: -147 1980a. A revised system of ais of the angiosperms. Bot. J. Linn. Soc. 80: 91-124 . 1980b. The taxonomic significance of chlo- rophyllous embryos in angiosperm seeds. Bot. Not. 133: 337-341. ————— & R. F. Tuorne. 1984. The order Myrtales: circumscription, ues and relationships. Ann. Missouri Bot. Gard. 71: 633-699. Davis, G. L. giosperms. Wiley & Sons, New DUMORTIER, B.C. 1829. Analyse des Familles des dern avec l'Indication des Principaux Genres qui s'y tachent. J. Casterman, Tournay. 66. "icon icc d of the An- ErpTMAN, G. 1952. Pollen Morphology and Plant Tax- onomy. Angiosperms. Almqvist & Wiksell, Stock- holm. ErrLINGER, M. G. & A. KjaER. 1968. Sulphur com- pounds in plants. /n: T. J. Mabry, R. E. Alston & V. C. Runeckles evel. Recent Advances in Phy- ; ge the major al- kaloid of Carallia brachiata rta daa paq Aus- ral. J. Chem. 18: 589-591. ibis J. J. 1974. Comiphyton, E nouveau ga- onais, i a Pa MR eae. Adansonia, Sér : 499-506. GEESINK, R., A. J. M. LEEUWENBERG, C. E. RIDSDALE & J. F. VELDKAMP. 1981. Thonner's Analytical Key to i Families a Flowering Plants. Leiden Univ. Press, The GIBBs, R. D. 1974. cT of dais Plants, 1-4. McGill & Queens, Montreal, London GORNALL, R. J., . Boum & R. DAHLGREN. 1979. The distribution of flavonoids in the angiosperms. 1-30 1966. Morphology of the androecium in Malvales, "fomes 13: 177-19 HEGNAUER, R. 1964. oa der Pflanzen, Birkháuser, Basel and Stuttgart. 1973. Chemotaxonomie der Pise 6. Birk- blas, Basel and Stuttgart. Hickey, L. J. & J. A. Worrk. 1975. The bases of angiosperm M Vegeta morphology. Ann. Missouri Bot. Gard. 62: 538-589. Hou, D. 1958. ie Flora Malesiana Series 29-493. 60. A new species of Caratia Roxb. (Rhi- zophoraceae). Nova Guinea, N. S 19 Crossostylis in Solomon jeans and the New Hebrides (Rhizophoraceae). Blumea 16: 129- 132. Volume 75, Number 4 88 Dahlgren 1277 Relationships HUTCHINSON, J. 59. The Families of Flowering Plants, I. Dicotyledons, 2nd edition. Clarendon Press, Ox- or HYLAND, B.P. M. & M. J. E. CoopE. 1982. A second species for Australian eae eens (Elaeo- carpaceae). Kew Bull. 36: 745. Jouns, S. R., J. A. LAMBERTON š re SiouMIs. 1967. The occurrence of (+)-hygroline in Gynotroches ax- illaris Bl. (Rhizophoraceae). Austral. J. Chem. 20 04. & ———. New tropane Pes 6R)-30a- -acetoxy- -68-hydroxynor- d (4 )2a-benzoyloxy-38-hydroxynortro- ane from Peripentadenia mearsii (Euphorbiaceae). Austral. J. Chem. 24: 2399-2405. Juncosa, A. M. 1982. Developmental morphology of the embryo and seedling in Rhi mangle s rera Amer. J 1984a. Embryogenesis and see s devel- m D (Rhi- alkaloids (4 t ane opment in Cassipourea Wero zophoraceae). Amer. 1: 84b. Embr ryogenesis zd. (i PMPP morphology of the seedling in Bruguiera exaristata Ding Hou (Rhizophoraceae). Amer. J. Bot. 71: 180- 191. & P. B. TomLINSON. 1987. Floral P capot in mangrove Rhizophoraceae. Amer. J. Bot 1263-1279. istorical and taxonomic syn- opsis of Penge and Anisophylleaceae. Ann. Missouri Bot. Gard. (this volume). & stematic comparison and some biological characteristics of Rhizophoraceae and An- isophylleaceae. Ann. Missouri Bot. Gard. (this vol- ume). KaPiL, R. N. & S. C. Tiwana. 1978. The integumen- tary tapetum. Bot. Rev. (Lancaster) 44: 457-490. KarsTEN, G. 1891. Ueber die re ia d^ agers in Malayischen Archipel. Bibl. Bot. KeaTING, R.C. 1984. Leaf histology kam its nai d to Vae in ra Pin Ann. Missouri Bot. Gard. 71 & Ra NDRIANASOLO. Leaf architecture and deere in the Rhizophoraceae and Anisophyl- lea . Ann. Missouri Bot. Gard. (this volume). T 1 50. Lepidobotrys Engl., type d'une famille nouvelle de spermatophytes: les Lepidobotry- aceae. Bull. Jard. Bot. Etat 20: 31-40. Le epidoboiryaceae. Flore du Congo osi et du Ruanda- Urundi 7: 58-61. LN.E.A.C., en N. R. & J.D. Curtis. 1974. Colleter anatomy in red mangrove, pa ets mangle (Rhizopho- 2221 (+) -Tropine-1, a new alkaloid from Bru rabedron Lett. 6327- 6329. «Duis i B. 1968. ution and potential taxo- nomic value of e ellagic acids. Phytochem- y 7: 1803- S Nen e J. iu. "Zur pom P der Elaeocar- paceae. Ark. Bot. 26A(10): 1 1938. Contributions to the embryology of the orders Td and Myrtales. Acta Univ. Lund 2, 35: 1-12 METCALFE, C. R. & L. CHALK. 1950. Anatomy of the icotyledons, 1-2. Clarendon sis Oxford. MM L. 19 y of two Simarou- a | Sci. 26: 970. Etico diens In: Proc. Symposium ae Embryology of oo Bull. In- dian Nat. cad. 41: 135. OLTMANN, O. 1971. Pollenmorphologisch- he on tische AET innerhalb der Geraniales. Diss Bot. 11: 1-163. Raven, P. H. 1975. The bases of angiosperm phylog- eny: cytology. Ann. Missouri Bot. Gard. 62: 724- 764. 984. The order Myrtales: a symposium. Ann. Missouri Bot. Gard. 71: 631-632. RipLEv, H. N. 1922-1925. The Flora of the Malay Peninsula, 1-5. L. Reeve & Co., London. RoMEIKE, A. 1978. Tropane alkaloids: occurrence and systematic importance in angiosperms. Bot. Not. 131: SCHIMPER, A. F. W. 1892. Rhizophoraceae. /n: A. Engler & K. dew: uade Die Natürlichen Pflan- zenfamilien SEIGLER, D. S. t Ë systematics and alkaloids. The Alkaloids 16: 1 ^in THORNE, R. F. nopsis of a putatively phylo- genetic classification of the flowering plants. Aliso 6: TIEGHEM, Pu. VAN & J. CONSTANTIN. 1918. Éléments de Botanique, 2, ae ar Paris? Tose, H. H. 198 „An embryological analysis les: it it isti of Ann. Missouri Bot. Car 70: oe ———. 7a. The embryology and re- lationships of Cas Ms and Sterigmapetalum (Rhizophoraceae — Macarisieae). Opera Bot. 92: 253- 64. ——— & —— ——. 1987b. Systematic embryology of y Anisophylleaceae. Ann. Missouri Bot. Gard. 74: -26. ——. Floral rod and evolution in Anisophylleaceae. Bot. J. L oc. Seed 1 Rhizophoraceae, and i tionships. Ann Missouri Bo. Gard. TOMLINSON, P. B., . PRIMACK Preliminary ebaerentians on floral biology in man- grove Rhizophoraceae. Biotropica 11: 256-277. VENKATA Rao, C. 3. Floral anatomy and embryol- ogy of two species of Elaeocarpus. J. Indian Bot. Soc. 32: 21-33. VERDCOURT, B. A new species of VPN pou z ythroxylaceae) from Tanzania. Kew Bull. 36: 4 nl, ¡3 V. P. Suan, J. J. SkvarLa & P. H. Raven. Morphology and s e Pe of Rhizophoraceae pollen. Ann. Missouri Bot. Gard. (this volume). VLIET, G. J. C. M. van. 1976. Wood #natomy of the Rhizophoraceae. Leiden Bot. Ser. No. 3: 20-75. ide iy R. 58. Linaceae. In: Flore du Congo Belge t du Ruanda-Urundi 7: 29-53. LN.E.A.C., Bru- fos (PR apoq M. S. & G. Y. ZHukova. 1980. hg ing in embryos of angiosperm seeds, a review. Bot. 133: 323-336. A HISTORICAL AND TAXONOMIC SYNOPSIS OF RHIZOPHORACEAE AND ANISOPH YLLEACEAE'! Adrian M. Juncosa and P. Barry Tomlinson? ABSTRACT Even from its time of formal rec dasha by Robert Brown in 1814, the Rhizophoraceae has been known as a family with both terrestrial and m family” is inappropriate, even though one of its major of mangroves. This associati enlargement of our knowledge of the fami angal representatives so that “m the tendency to regard it as the “mangrove r subdivisions, the tribe Rhizophoreae, is made up exclusively on of terrestrial and mangrove species adds a of the group because it allows the diagnostic usefulness of esc to the fs idis systematic s s to be as e progressive mily is reviewed, including the status “of “Anisophyllea and related genera, now regarded as constituting a separate family Anisophylleaceae. Brief triba anc d ge enerte provided, with id illustration of geographical distribution. Some portraiture of the more common genera is attempted. This symposium had as an objective the seg- regation of the small family Anisophylleaceae from the Rhizophoraceae (sensu lato), with a discussion of the evidence upon which this distinction was based (cf. Dahlgren, this volume). A further ob- jective has been a discussion of where the two families might be placed in a natural system. Al- though the mangrove Rhizophoraceae are familiar to botanists and laymen alike, the systematic char- acteristics and even the existence of the inland genera are not well known. The Rhizophoraceae and de together include about 18 genera and some 140 species of tropical shrubs and trees (Table 1). The families exhibit a wide variety of character states and have had a check- ered taxonomic history. It seems now generally agreed that the Anisophylleaceae are a distinct, probably unrelated family; for a detailed systematic comparison see our other contribution in this vol- ume. Unless otherwise noted, “Rhizophoraceae” is therefore used sensu stricto, i.e., without the four genera now removed as Anisophylleaceae. In ad- dition to describing this history, we present a syn- 18 appears re elsewhere, so it should prove a a useful irana for opsis of genera. No such plet this symposium volume. A detailed comparison of the systematic and biological characteristics of the two families appears separately (Juncosa & Tom- linson, this volume). TAXONOMIC HISTORY OF RHIZOPHORACEAE (SENSU LATO) The family Rhizophoraceae, in a broad sense, originates with the genus Rhizophora of Linnaeus (1753), preceded by Mangium of Rumphius (1741-1755). Both of these authors used their generic names to designate a group of species grow- ing in tropical tidal swamps, which modern ecol. ogists would call “mangal” (Macnae, 1968). Of the seven species named by Linnaeus in Rhizoph- is still valid and forms the type species of the genus and hence of the family. Mangium has disappeared entirely, ora, only one, R. mangle L., except in Acacia mangium, a terrestrial plant. A list of the Rumphian and Linnean names with their ! We thank Paul P. K. Chai and the Forest Department, Kuch and the n Institute of Marine Science; the Nacionales, Cost Fisher, Hiroshi Tobe, and Peter H. Raven for s support has come directly from N.S.F. 82-1627 1 Additional suppo Pewee PA on a gran come from the Forest Department, Brune Rica, for permission to n and for assistance in plying material of several gener Ed Seling for technical assistance. Distribution base ma lissertation improvement award DEB 80-16 from the National Geographie jr to P. B. Cabot Foundation and Atkin ing, Sarawak; John S. Bunt, Norman C. Duke, ei; and the Servicio de Parques | field studies. We also thank pos B. a, and Mo Unversity of Chicago Pr 635 to A. Juncosa and BSR Tomlinson. ds of Harvard University. ps are by the s Garden Eun ? Harvard iMd Harvard Forest, Petersham, Massachusetts 01366, ANN. Missouri Bor. GARD. 75: 1278-1295. 1988. Volume 75, Number 4 1988 Juncosa & Tomlinson Historical & Taxonomic Synopsis of Rhizopt /Anisophyll 1279 M Summary of genera of the Rhizophoraceae (s. str.) 1958; Melchior, 1964; Floret, TABLE 1. Alston, 1925; Ding Hou, and Anisophylleaceae Dcos Ridley, 1922; 1974; van Vliet, 1976; Willis, 1973; Steyermark & Liesner, 1983). Tribal classification of Rhizophoraceae follows that accepted in the ORA recent studies indicate that changes are necessary (Tobe & Raven, this volume). Number o Family or Tribe Genus Species Distribution ANISOPHYLLEACEAE (= Anisophylleae) Anisophyllea 25 Africa to South America and Malesia Combretocarpus 1 Borneo Poga 1 Africa Polygonanthus 2 Brazil RHIZOPHORACEAE Gynotrocheae' Carallia 9 Madagascar to Australia Crossost ylis 12 South Pacific Gynotroches k Indochina, iom Pacific Islands Pellacalyx 8 Malesia to S Chin Macarisieae' Ano 2 Africa Blepharistemma l India Cassipourea (including Lasi- 55 Sri Lanka to Central America osepalum and Weihea) Comiphyton l Equatorial Africa on 15 Africa, Madagascar Macari 7 Madagascar pala 3 South America Rhizophoreae Bruguiera 6 Africa to Australia Ceriops 2 Africa to Australia Kandelia 1 Malaysia to Japan Rhizophora 8* pantropical ! Tobe & Raven (this sale segregate Crossostylis as its own tribe and create two subtribes each in Macariseae and their modified Gynotroc * Ding Hou (1958) i lel much variation in Gynotroches and only reluctantly accommodated it in a single ecies. Juncosa & Tobe (this volume) describe some characteristics of two of the several distinct taxa * Floret (1976) presented evidence for the elevation of Dactylopetalum to generic rank. ). 1 Includes putative hybrids or varieties (Tomlinson, 1986 modern equivalents forms Table 2. This demon- strates an intrinsically ecological concept for their initial circumscription. amily was formally designated (as the “or- der" Rhizophoreae) by Robert Brown in his account of the botany of Terra Australis (Brown, 1814) and included the genera Rhizophora Linnaeus, Bruguiera Lamarck, and Carallia Roxburgh “all of which are found in the equinoctial part of New Holland." Consequently, even at its erection the family already included plants of terrestrial as well as mangal habitats. Robert Brown drew attention to the “affinity of Rhizophoreae to Cunoniaceae" and rejected Jussieu's attempt to combine some of its elements with loranthaceous genera (cf. Du- mortier, 18 e subsequent development of our ue of the family is summarized in Ta- a (1840) subdivided the family into Leg- notidae (Cassipourea and Dryptopetalum = Gy- notroches) and Rhizophoreae sensu stricto, which still included Carallia. Blume (1849) raised these two divisions to family status but transferred Car- allia to Legnotidae so that the first recognition i.e., a named taxon capable of ecological designation, is from this date, even though Blume's “families” have scarcely been recognized by subsequent authors. Corner (1976) did recognize Legnotidaceae in his description of seeds of dicotyledons. Bentham & Hooker (1865), treating the Rhizophoraceae as a family, main- tained Rhizophoreae and Legnotidae as tribes but added Anisophylleae as a third tribe to include the terrestrial genera Anisophyllea and Combreto- ca T 3. stem was essentially followed by Baillon (1876), s split the Legnotidae into two separate 1280 Annals of the Missouri Botanical Garden TaBLE 2. Present probable status of some Linnean and Rumphian names in Rhizophora and Mangium (see Salvoza, 1936). Present A. Rhizophora Linnaeus (1753) R. mangle R. mangle L. R. norhiza Bruguiera gymnorrhiza (or gymnorhiza) (L.) Lamk. R. cylindrica Bruguiera cylindrica (L.) Bl. R. ca Kandelia candel (L.) Druce R. caseolaris Sonneratia caseolaris (L.) En R. majus Aegiceras corniculatum (L.) Blanco R. corniculata Aegiceras corniculatum (L.) Blanco B. Mangium Rumphius (1741-1755) M. candelarium Rhizophora apiculata Bl. M. celsum Bruguiera gymnorrhiza (L.) Lamk. M. minus (p.p.) Bruguiera gymnorrhiza (L.) Lamk. M. digitatum Bruguiera cylindrica (L.) Bl. M. minus (p.p.) Bruguiera cylindrica (L.) Bl. M. caryophylloides Bruguiera cylindrica (L.) Bl. M. caryophylloides parvifolium Ceriops tagal (Perr.) C. B. Rob. latifolium M. caseolare rubrum Sonneratia caseolare (L.) Engl. M. floridum Aegiceras floridum R.&S. tribes, Macariseae and Barraldeieae (= Gynotro cheae) from Barraldeia (= Carallia). Ridley (1922) raised the three tribes to families (his “orders,” viz. Rhizophoraceae, Legnotidae, and Anisophyl- leae) but considered only the Malayan genera. Schimper (1898) retained the traditional one- family concept but departed dramatically from the general consensus by splitting the family on mor- phological characters that he considered to be more fundamental than the structure of the fruit and seedling, since he considered their specializations to be features adapting the plants to their habitat (Anpassungsmerkmal) and therefore comparable to those found in other isolated mangrove taxa like Aegiceras, Avicennia, Lumnitzera, and Sonnera- tia. This may be seen as a complete reversal of the Linnean view. Schimper developed an elaborate subdivision, cutting across ecological characteris- tics as follows: subfamily I. durs aic tribe 1. Gynotrochea subtibe la. Cyaorrosliiase [Crossostylis (as Crossostyles), ee Rhizophora, Ceriops, Kande subtribe 1b. icum [ Pellacalyx, guiera, Carallia tribe 2. Macarisieae [Blepharistemma, Dac- tylopetalum, Macarisia, Weihea, Cassi- Bru- pourea Anisophylloideae [ Anisophyllea, Combretocarpus | subfamily II. This arrangement has been summarily rejected by all subsequent authors, in particular it leads to Ding Hou’s statement (1958), which might well be placarded in all institutes of systematic botany: "Schimper's clearly wrong classification provides again a good example of what danger is involved if a taxonomist introduces pre-occupied theoretical ideas into classification." In fairness to Schimper, it should be pointed out that he had a better idea of fruit and seed construction than most of his predecessors, since he recognized that the family Rhizophoraceae, which Brown had characterized as exalbuminous, did indeed possess well-developed endosperm (Juncosa, 1982), and that Bruguiera was appreciably different from the other mangrove taxa in some basic features. However, the sum of all vegetative features and especially of wood anat- omy (Marco, 1935; van Vliet, 1976) shows that the habitat distinctiveness of the Rhizophoreae, even though it can be said to be primarily based upon “adaptive” or is well founded at the systematic level. The family therefore provides an interesting example for phy- letic study since the Rhizophoreae are clearly de- rivative and must have had an ancestry in taxa "ecological characteristics, that are now represented by the terrestrial genera. The most recent accepted intrafamilial classifica- tion (Table 1) was initially presented by Melchior Volume 75, Number 4 1988 Juncosa & Tomlinson 1281 Historical & Taxonomic Synopsis of Rhizophoraceae/Anisophylleaceae (1964). It recognizes three tribes: Macarisieae, with a center of distribution in Africa; Gynotrocheae, centered in Malesia; and Rhizophoreae, also cen- tered in Malesia. Monographs and partial revisions of varying util- ity exist for some genera of Rhizophoraceae. Alston (1925) discussed the largest genus, Cassipourea, including within it Dactylopetalum, Weihea, and Lasiosepalum. The related genus Comiphyton was described by Floret (1974), who subsequently dis- cussed relationships of African Macarisieae further (Floret, 1976). Sterigmapetalum was recently re- vised by Steyermark & Liesner (1983). Several systematic treatments of Rhizophora or portions thereof exist (Salvoza, 1936; Ding Hou, 1960; Breteler, 1969, 1977). Yet only in the western Pacific, where the plants have been studied largely in the field rather than as dried material, is our understanding of the taxa adequate (Tomlinson, 1978; Duke & Bunt, 1979), although many basic questions about distribution, hybrids, and specific limits still remain unanswered. Excellent revisions of other Malesian genera, both mangrove and in- land, are found in Ding Hou (1958). Our present understanding of the mangrove Rhizophoraceae is summarized in Tomlinson (1986). In looking at the history of this family, one is impressed by the relative recentness of our present taxonomic and nomenclatural understanding. For example, the clarification of the full range of mor- phology in Bruguiera only comes with Ding Hou (1958); Ceriops decandra was correctly desig- nated for the first time in this monograph. The difference between the two species of Ceriops is quite striking when floral function is considered. ome African taxa have been discussed by J. J. Floret (1976), while the New World taxa have been treated incompletely by a few authors (e.g., Prance et al., 197 One might conclude that any discussion of the phylogeny of this family is premature in the ab- sence of detailed information about many species, but the situation reflects our ignorance of tropical plant families generally. If we can devote at least as much time to observation as we are doing to speculation about phylogenies, then our under- standing is likely to improve. AFFINITIES WITH OTHER FAMILIES Rhizophoraceae (sensu lato) have traditionally been placed in the Myrtales (Bentham & Hooker, 1865; Melchior, 1964; Takhtajan, 1980), al. though more recently they have sometimes been aligned with Cornales (Cronquist, 1968; Thorne, 1976). These assignments and the long-obsolete suggestions of affinities with Loranthaceae or San- talaceae were based on the incorrect assumption that Rhizophoraceae (s. str.) characteristically have an inferior ovary; on the contrary, this is a rare and derived condition in the family. Furthermore, there are many fundamental vegetative differences between Rhizophoraceae (s. str.) and all of the aforementioned groups. Suggestions of relationship with Rubiaceae, Cunoniaceae, or Dialypetalantha- ceae are based solely on the common possession of interpetiolar stipules and are as indefensible as are most systematic judgments based upon a single character. Cronquist's (1981) separation of Aniso- phylleaceae and Rhizophoraceae and assignment of the latter to its own order, Rhizophorales, is a wiser phylogenetic policy but still begs the question of affinities. We strongly favor Dahlgren's ap- proach that leads to the surprising but very well. supported suggestion, proposed and discussed in detail elsewhere in this volume, of affinities between Rhizophoraceae, Celastraceae, Elaeocarpaceae, and possibly several other families. FIELD RECOGNITION OF RHIZOPHORACEAE Because of the great infrafamilial variation in most morphological characters, the family diag- nosis, given below for completeness, is unwieldy and nearly useless to both field and herbarium botanists. Much of the year the mangrove genera are easily recognized by the hypocotyls protruding from the fruits. Vegetatively, the family is recog- nized by having opposite (or verticillate) lea interpetiolar stipules; the leaves are generally bit- ter-tanniniferous and usually minimally toothed in inland genera. In mangal, only one nonrhizopho- raceous genus has interpetiolar stipules (to only 3 mm long—the weak-stemmed Old World shrub Scyphiphora (Rubiaceae)). Inflorescences are ax- illary and basically cymose, condensed in many common genera. The sole diagnostic floral feature, present in all species of which we are aware and otherwise known only in Rhamnaceae, is that each petal (fringed in all but two genera) individually encloses either a single antipetalous stamen or a group of 2-6 stamens, depending mainly upon whether the androecium is diplostemonous or poly- androus. Rhamnaceae are easily distinguished by their isomerous stamens and entire petals and of course many other characters. FAMILY RHIZOPHORACEAE R. BROWN Shrubs or trees (to 50 m) of dry to wet forests or mangrove swamps; leaves opposite or verticil- late, toothed, crenate, or entire. Stipules interpetio- lar, valvate and pubescent or imbricate and gla- 1282 Annals of the Missouri Botanical Garden TABLE 3. History of tribal classifications of Rhizophoraceae (s.l.). This Article, 1987 — Family and Tribe Brown, 1814— Family Endlicher, 1840 — Tribe RHIZOPHORACEAE (i) Rhizophoreae Rhizophoreae (i) Rhizophoreae Rhizophora (1753) Rhizophora Rhizophora Bruguiera (1797) Bruguiera Bruguier Kandelia (1834) Carallia Kandelia Ceriops (1838) Ceriops Carallia (ii) Gynotrocheae Crossostylis (1776) Carallia (1811) Pellacalyx (1836) Gynotroches (1844) (ii) Legnotidae* Cassipourea Gynotroches (ii) Macarisieae Cassipourea! (1775) Macarisia (1836) Blepharistemma (1858) Dactylopetalum (1859) Sterigmapetalum (1925) Anopyxis (1960) Comiphyton (1974) ANISOPHYLLEACEAE Anisophyllea (1824) Combretocarpus (1865) Poga (1896) Polygonanthus (1932) Crossostylis not classified Bentham Blume,’ 1849 — Family Hooker, 1865 — Tribe Baillon, 1876 — Tribe (i) Rhizophoreae (i) Rhizophoreae (i) Rhizophoreae Rhizophora Rhizophora Rhizophora Bruguiera Bruguiera Bruguiera Kandelia Kandeli Kandelia Ceriops Ceriops Ceriops (ii) Legnotideae (i) Legnotideae (ii) Gynotrocheae (as Barraldeieae) Carallia Carallia Cassipourea Cassipourea Gynotroches Gynotroches Gynotroches Crossostylis Crossostylis Pellacalyx Pellacalyx acari Blepharistemma (iii) Macariseae Cassipourea Macarisia (iii) Anisophylleae (iv) Anisophylleae Anisophyllea Anisophyllea Combretocarpus Volume 75, Number 4 Juncosa & Tomlinson 1283 1988 Historical & Taxonomic Synopsis of Rhizopt /Anisophyll TABLE 3. Continued. Schimper, 1898—See Text Ridley,’ 1922 — Family Melchior, 1964. — Family and Tribe I. Rhizophoroideae (i) Rhizophoreae RHIZOPHORACEAE 1. Gynotrocheae Rhizophora (i) Rhizophoreae (a) Gynotrochineae Bruguiera Rhizophora J Kandelia Bruguiera Kandelia Ceriops Kandelia Ceriops Ceriops Gynotroches Crossostylis (b) Carallinae (ii) Legnotidae (ii) Gynotrocheae Bruguiera Carallia arallia Carallia Gynotroches Gynotroches Pellacalyx Pellacalyx Crossostylis Pellacalyx 2. Macarisieae (ii) Macarisieae Cassipourea Cassipourea Dactylopetalum Macarisia "eihea Macarisia Blepharistemma II. Anisophylloideae (ii) Anisophylleae ANISOPHYLLEACEAE Anisophyllea Anisophyllea Anisophyllea Combretocarpus Combretocarpus oga | Cassipourea includes wir. el a Dactylopetalum (p.p.), and Petalodactylis. rom Legnotis a synon * Miquel (1855) added ba Md paraa (as Anisophyllum) to Legnotidae. ed. * Non-Malaysian taxa not consider Inflorescences or fasciculate. brous, always bearing colleters. axillary, cymose, dichotomous, Flowers actinomorphic, bisexual or rarely unisex- ual, clearly articulated at juncture with pedicel. alyx valvate, 4-many-lobed; petals equaling the number of sepals, usually with both a terminal arista and filiform appendages on the two lobes (rarely entire), each petal individually enclosing 1- 5 stamens. Androecium diplostemonous to polyan- drous, the filaments sometimes connate at base; nectariferous ring (“disc”) intrastaminal, entire or . Ovary superior to inferior, 2-many-carpel- late (locules often incompletely or not at all sep- arated by septae at anthesis); stigma capitate or with pronounced lobes, generally papillate. Ovules 2 or many per carpel, anatropous, usually apically inserted. Fruit capsular or baccate; seeds 1 —many, naked, arillate, or winged, nium. hii green, usually straight, with laminar epigeal germination, or with thick MER a Hn cylindrical cotyledonary body) and viviparous ger- mination. Genera. 15 (names listed by tribes); about 145 species Distribution. . Pantropical. In the synopsis that follows, genera are orga- nized into the traditionally recognized tribes (e.g., Melchior, 1964), despite evidence already in hand that suggests some rearrangements (Tobe & Ra- ven, this volume). Complete diagnoses are not giv- en; instead, only some of the more distinctive char- acteristics of each tribe or genus are mentioned. References are likewise selectively cited. Listing of tribes and genera is alphabetical. TRIBE GYNOTROCHEAE Shrubs to large trees, some species weedy and characteristic of highly disturbed vegetation. Prom- inent aerial roots known in all genera but Pella- calyx. Leaves bijugate (not decussate), generally crenate. Stipules glabrous and imbricate (except in Pellacalyx). Flowers bisexual (except in Gyno- troches); stamens twice the number of petals; ovary Missouri Botanical Garden Annals of the 1284 es — — SK La EN a Volume 75, Number 4 1988 Juncosa & Tomlinson 1285 Historical & Taxonomic Synopsis of Rhizopt /Anisophyll F inferior (except superior in Gynotroches), 5- or many-carpellate; ovules 2 or many per carpel (loc- ules incompletely separated). Fruit + baccate, 1— many-seeded; seeds naked (except arillate in Cros- sostylis). This highly variable tribe has been the source of much confusion and is undoubtedly para- phyletic. Its proposed subdivision by Tobe & Raven (this volume) aids in clarification. Genera. Carallia, Crossostylis, Gynotro- ches, Pellacalyx istribution. Centered in Malesia, except Crossostylis (South Pacific) (Fig. 1). Carallia Roxb., 9 species (Brandis, 1911; Ridley, 1922; Ding Hou, 1958). Figure 2. Distribution. Indochina, Malesia, Philippines to New Guinea; C. brachiata also ranges to Mad- agascar, India, Nepal(!), S China, Solomon Islands, and N Australia. Stilt roots reported only in C. brachiata. Tine stalked glands (up to 5 mm) present outside stipu in several (all?) species. Inflorescences usually ir cymose. Flowers 5-8-merous, diplostemonous, the ovary fully inferior. Seeds 1-several. Embryolog- ically similar to Rhizophoreae. Crossostylis J. R. & G. Forst., about 12 species (Smith, 1981). Figures 3, 4. S Pacific Islands. Distribution. Inflorescences dichotomous. Largest flowers in the family (to 6 cm wide) in C. grandiflora (Fig. 4). Petals with very reduced appendages, even appearing entire at maturity. Stamens variable in number, sometimes basally connate and bearing odd appendages that retain the copious nectar in pendulous flowers. Ovary multicarpellate (up to about 20), nearly superior to inferior; stigma with long lobes. Fruit dehiscent or a partially dehiscent "salt-shaker" This genus shares some characteristics (e.g., appendaged seeds) with Macarisieae, but many oth- ers with Gynotrocheae. It is placed in its own tribe, Crossostylieae, by Tobe & Raven (this volume). capsule. Gynotroches Blume, 2-4 species (Ding Hou, 1958; Backer & Bakhuizen, 1963). Figures 5, 6. Distribution. Burma through Malesia to Mi- cronesia and Melanesia. Weedy tree; branches often drooping. Plants dioecious. Inflorescences fasciculate. Flowers 4—5- merous, diplostemonous (Fig. 6). Ovary superior, 4—5-carpellate, with 3-8 ovules per locule; stig- matic lobes sometimes elongate. Fruit a berry; seeds several to many. The most recent revision (Ding Hou, 1958) rec- ognizes only one species (G. axillaris Blume) but points out that the variation in floral characters is such that with further study several distinct species will be recognized. Juncosa & Tobe (this volume) provide some details separating two of these taxa. Pellacalyx Korth., 8 species (Ding Hou, 1958). igure Distribution. Burma and South China to Ma- lesia. Some species weedy; branches often drooping. Stipules valvate, the edges folded sharply inward. Indumentum of stellate hairs, unique among Rhi- zophoraceae. Pairs of bracteoles fused into a toothed cup. Number of stamens and of carpels twice the number of petals, this usually 4 or 5. Each carpel or locule with 8-25 ovules. Despite its distinctive indumentum and super- ficially very different flowers, this genus shares many vegetative and embryological synapomor- phies with Gynotroches; the two are clearly sister genera. TRIBE MACARISIEAE Shrubs to large (50 m) trees, generally of moist primary forest; but some species found in dry or deciduous forest. Stilt roots absent or weakly de- veloped. Leaves toothed, at least in juvenile growth phases, verticillate or opposite, decussate in bud but often reoriented on branches. Stipules valvate, pubescent, also bearing colleters. Inflorescences fasciculate or lax-cymose. Flowers bisexual, except in Rat ae hanna and PIepRanitemma (?). An- only in Cas- E Ovary superior’ in four or five a the six genera, not in only two as sometimes stated; locules 2-6. Fruit a capsule, sometimes indehiscent; seeds arillate or winged. Detailed information is lacking for most genera, and characterizations reported here are based largely upon the literature, much of which is very — FIGURE 1. Geographical distribution of genera of tribe Gynotrocheae. Annals of the Missouri Botanical Garden Volume 75, Number 4 1988 Juncosa & Tomlinson 1287 Historical & Taxonomic Synopsis of Rhizopl /Anisophyll |» LI J incomplete and not illustrated. This basal tribe is the most poorly known group of Rhizophoraceae, largely due to the unavailability of fixed material and the rarity of certain key taxa. The three sub- genera of Cassipourea are distinguished by several characters of calyx, androecium, and petal ap- pendages, although three species exhibit combi- nations of character states of more than one sub- genus (Floret, pers. comm.). Dactylopetalum, here re-elevated from a subgenus of Cassipourea to generic rank, and Blepharistemma are dubiously distinct at the generic level. Anatomical and mono- graphic work in progress promises to clarify the relationships of all these arillate-seeded Macari- sieae. Distinctions between the winged-seeded gen- era (Anopyxis, Macarisia, and Sterigmapetalum) are clear. Genera. Anopyxis, Blepharistemma, Cassi- pourea, Comiphyton, Dactylopetalum, Macari- sia, Sterigmapetalum. Distribution. Centered in Africa, extending to Madagascar and India (Sri Lanka) and to South and Central America (Fig. 8). Anopyxis eda Engl., 1-3 species (Sprague & Bo ; Hutchinson & Dalziel, 1954; Irvine, E Distribution. Tropical Africa. Tallest inland genus (to 50 m), often dominant. nate over their entire length. Petals sometimes entire. Fruit woody, indehiscent (?); seeds winged. Blepharistemma Wall. ex Benth., (Schimper, 1898; Gamble, 1919). SW India. ] species Distribution. Bracteoles absent (?); flowers polygamodioe- cious, 4-merous. Ovary 3-locular; fruit fleshy, in- dehiscent (?); seeds arillate. Cassipourea Aublet (including subgenera Weihea and Lasiosepalum), about 55 species (Alston, 1925). Figures 9, 10. Distribution: subgenus C TA tropical Americas, West Indies, West Afric subgenus as West Africa. subgenus Weihea: Africa, Madagascar, India, Sri Lanka. Many species occurring in dry habitats; com- monly shrubby, also some tall (30 m) trees (Fig. 9). Inflorescences usually condensed. Flowers (4-)5(-6)-merous. Stamens indefinite in number (15-40), in one (or more?) whorls. Ovary superior, 3(-4)-locular. Seeds 1-4; aril white, yellow, or orange. Comiphyton Floret, 1 species (Floret, 1974, Distribution. Gabon to E border of Zaire; not yet known from the belt defined by 2°N or 5, thus one of the most narrowly equatorial ranges of any plant. Distinguished from Cassipourea by its inflores- cence, diplostemonous androecium, anthers, and placentation (Floret, 1974, 1976); not sharply dif- ferentiated from Dactylopetalum. Dactylopetalum Benth., about 15 species (Al- ston, 1925; Floret, 1976). Distribution. Equatorial Africa, Madagascar. After consideration of remarks by Floret (1976) and other literature (Bentham & Hooker, 1865; Oliver, 1871; Dale & Greenway, 1961), we prefer to recognize this genus as distinct from Cassipou- rea; distinguishing characteristics include the diplo- stemonous androecium, 2-carpellate ovary, and in- dehiscent (?) fruit. Macarisia Thouars, 7 species (Schimper, 1898; Arènes, 1954) Distribution. | Madagascar. Diplostemonous, 4—5-carpellate, seeds winged. Sterigmapetalum Kuhlmann, 7 species (Stey- ermark & Liesner, 1983). < FicunEs 2-7. u e) rest typical car inland Rhiz cophoracene = 4 uds are white to red, petals w i P opposite bijugate leaves reorie Gynotroches sp.; they were being vis ite.—5. Ceno. sp. gro ted into Ps (pollinated?) by numerous small ants Habit and flowers of Gynotrocheae. — 2. Fully open flowers of Carallia borneensis growing in w in their pistillate phase (stigmas receptive) ; those growing in New Caledonia in iowland rain of Pellacalyx cristatus growing in Sarawak; overall habit in this genus is similar to that of Gynotroches. Annals of the 1288 Missouri Botanical Garden 'apaisi4DnoDpy fo viaua? fo uoimqiuisip ¡vary dodo) g 340214 / ZITTA LEPE E AA J / PƏTSTIPOPKR f | Y \ / | | \ / un[e3edo[A3»eq > ox ee | srxKdouy Iv CM O DES ⁄ pcm P ios) Y e | d EN U GLASS UA | CN | un[ejedewSlieig ' ka a 37 Volume 75, Number 4 Juncosa & Tomlinson 1289 Historical & Taxonomic Synopsis L r FIGURES 9, 10. Monteverde, Costa Rica. Smooth unbuttressed bole Cassipourea sp. cf. killipii (Chocó, Colom appendages du increase apparent size of flow NW South America. Distribution. Plants dioecious. Leaves usually verticillate. Flowers 4—6-merous, diplostemonous. Seeds winge TRIBE RHIZOPHOREAE Shrubs or trees of mangrove swamps; flowering plants 1.5-50 m tall. Aerial stilt roots always pro- duced, prominent only in Rhizophora. Leaves en- tire, bijugate. Inflorescences variable, generically diagnostic. Flowers 4—multimerous, mostly diplo- stemonous, the petals variously specialized for di- verse pollination mechanisms. Ovary half to fully inferior, 2—3-carpellate; ovules 2 per carpel. Fruit baccate, fibrous, 1 -seeded. Germination viviparous, the huge seedling axis (to 1 m) emerging from both seed coat and fruit up to 9 months before abscis- sion. This familiar tribe is unfortunately morpholog- ically atypical of the family; this has led to con- siderable confusion in phylogenetic decisions. Habit a Pega of Cassipourea ee ne —9. Cassipourea sp. growing by pasture on crown are typical of the genus. —10. Flower of bia) at anthesis showing both side and top views; plumose petal Genera. Bruguiera, Ceriops, Kandelia, Rhi- zophora. Distribution. | Pantropical (Fig. 11). Bruguiera Lam., 6 species (Ding Hou, 1957, 1958). Figure 12. Distribution. SE Africa through Malesia to Pacific Islands and Northern Australia. Inflorescences cymose or reduced to 1-3 flow- ers, sometimes ebracteolate. Flowers polymerous, diplostemonous; ovary deeply inferior, 3-carpel- late. Each petal encloses two stamens, releasing them explosively when stimulated. Petal append- ages often reduced, the petals variably pubescent abaxially. Ceriops Arn., 2 species, 1 variety (Ding Hou, 1958). Figures 13, 14 Distribution. SE Africa through Malesia to Australia. Missouri Botanical Garden Annals of the 1290 jou) nomo ui pazijpanjpu Suruo22q 240 pup pagnposjur uəəq aany e1omSnag pup eioqdozupy -avasoydoz1y y 2q141 fo viaua? fo uoiunquisip 1»onjdpa8oo*) LS, Zi L F a / 4 | e» / + , - d / / | / / ] | a — "(umowys “11 330914 [f 7, pu x ?eioudozr Z= u o LI ° E ` sdoT1ə2 ` Zu uh ME. aes AIN Wr. os EE 74 p EN Al i ` X LN T ——— —3À a E y eiornSnig Vv di vi ES - - A `. ` `. - N Ea Ë Sl Habit, flowers, and bns ci seedlings of Rhizophoreae. — 12. Flower and blunt-tipped 'uiera exaristata (NW nsland, Australia) .— 13. Small trees (the usual habit) of Ceriops ling of Bru tagal (Sarawak), showing = cone of stilt roots that ae: to form buttresses; leaves of this very drought- and salt-tolerant species are curled and upright.— 14. Branch tip with fasciculate inflorescences of Ceriops tagal | sei ak) .— 15. Branch Ba of Kandelia candel (Sarawak) with mature viviparous seedlings. The long-acuminate pow of this genus have a mode of establishment unlike that of other Rhizophoreae. — 16. Habit of Rhizophora apicula ata (New Caledonia). In this species, highly adapted to anemophily, flowers are borne well below leaves, and petals and dehisced stamens may abscise before the flower opens. Missouri Botanical Garden Annals of the 1292 ‘younxa aq mou Apu sndieooləiquio2 *nisKn]ppy 4D nsuruag u] ‘avaavayAydosiup Apnunf fo piaua? fo uoinqiaisip ponydvis0ay — 7; | 380914 2 X WC PP / JJ | \ \ AERE \ a "a C LF— — -----þ--- snu3ueuo8AToq [ b ` | ; | Bey [Aydostuy ‘Le eo snd180331qu0) N AG A N Š Volume 75, Number 4 1988 Juncosa & Tomlinson of RI 1293 Historical & Taxonomic Synopsis u T J FicurEs 18, 19. understory shrub in Malaysia; large trees in this genus have similar architecture and phyllotaxis, albeit less obvious.— 19. Foliage and paniculate inflorescence of Combretocarpus rotundatus (Also from Brunei) . Most salt-tolerant genus of the tribe; frequently shrubby; bark pale (unlike all other genera of the tribe). Inflorescences fasciculate (Fig. 14), tricha- sial changing to monochasial. Flowers pentamer- ous, diplostemonous; ovary half-inferior, 3-carpel- late. Hypocotyl generally ridged. Kandelia (DC.) Wight & Arn., 1 species (Ding Hou, 1958). Figure 15. Distribution. Bangladesh to S Japan, through Indochina to Malaysia, Sumatra, Borneo. Char- k acteristically found on riverbanks. Inflorescences dichotomous; bracteoles connate and corky. Flowers 4—5-merous; petals to 2 cm, 3- 7-lobed nearly to base; stamens about 30; ovary half-inferior, 3-carpellate. Hypocotyls long-acu- minate. Rhizophora L. Either 8 species or 4 species, 2 distinct varieties, 3 hybrids (Salvoza, 1936; Ding Hou, 1958; Tomlinson, 1978). Figure 16. Habit and inflorescence of Anisophylleaceae.— 18. Anisophyllea disticha (Brunei) , a common Distribution. | Pantropical, barely extending to subtropics. Large stilt roots. Inflorescences dichotomous, bracteoles of most species tiny. Flowers 4-merous; ovary half-inferior, 2-carpellate. Petals entire (this unique in the family) but usually densely pubescent, the edges barely enclosing each antipetalous sta- men. Stamens 8, in two whorls, except 3 times the number of petals in R. apiculata; multilocellate in all species. Wind pollinated. FAMILY ANISOPHYLLEACEAE RIDLEY Trees and shrubs of wet primary forest; leaves alternate, dimorphic (except in Combretocarpus), exstipulate or with highly reduced stipular homo- logues. Inflorescences axillary, racemose to panicu- late. Flowers mostly unisexual (plants monoecious), except bisexual in Combretocarpus. Calyx and pet- als valvate, 3—-5-merous. Petals lobed or laciniate (except entire in Polygonanthus). Androecium di- plostemonous. Nectary crenate. Ovary inferior, 3— 1294 Annals of the Missouri Botanical Garden 4-locular, the styles separate. Ovules 1-2 per car- pel. Fruit a drupe or dry, winged (Combretocar- pus), usually 1-seeded. Endosperm lacking; em- bryo with reduced or no cotyledons. Germination hypogeal. Genera. Anisophyllea, Combretocarpus, Poga, Polygonanthus. Distribution. South America, Africa, India to Malesia (Fig. 17). Anisophyllea R. Brown ex Sabine, 25 species utchinson & Dalziel, 1954; Ding Hou, 1958). Figure 18 Distribution. South America, Africa, India to Malesia. Shrubs be m tall through MES trees. Ani strongly most species anisophyllous, with unique tetrastichous phyllotaxy (Fig. 18). Serial axillary buds numerous. Staminate and pistillate flowers either in separate inflorescences or mixed, often not strongly hetero- morphic, usually 4-merous. Drupes 2-15 cm, usu- ally 1-seeded. Combretocarpus Hook. f., 1 species (Ding Hou, 1958). Figure 19. Distribution. Borneo. Large tree of peat swamps. Leaves isomorphic, distichous. 19). owers bisexual, usually trimerous. Petals linear or irregularly 3-4-lobed. Fruit winged. Inflorescence paniculate (Fig. Poga Pierre, 1 species (Hutchinson & Dalziel, 1954) Distribution. | Equatorial Africa. Large tree. Leaves anisophyllous. Inflorescences on specialized leafless branches. Flowers unisexual, strongly heteromorphic. Fruits with 3-4 edible oily seeds. Polygonanthus Ducke, 2 species (Prance et al., 1975) Distribution. |. Brazil (Amazonia). Small trees. Leaves anisophyllous. Flowers uni- sexual, strongly heteromorphic; petals unlobed, nearly linear. LITERATURE CITED ALSTON, A. H. G. 1925. Revision of Cassipourea. Kew Bull. 1925: 241-276. ARÈNES, J. 1954. Rhizophoracées. In: H. Humbert (ed- itor), Flore de Madagascar et des Comores. Firmin- Didot, Paris. (Rhizophorac Backer, C. A. & 1963. Flora of Java, Volume Groningen, The Netherlands. 76. Natural History of Plants (English edition). L. Reeve, London. . D. HOOKER. R. C. BAKHUIZEN VAN DEN BRINK, JR. . N. V. P. Noordhoff, 1865. El Planta- BERE s L. 1849. Museum Bonon. Pp. 126- 13 E D. 1911. Indian Trees. Constable, London. BRETELER, F. J. 1969. The e eie ve f Rhi- zophora. Acta Bot. Neerl. 18: l. — P 1977. America’s Pie cou of Rhizo- . Acta Bot. Neerl. 26: -230. 1814 General nuu geographical and The Seeds of Dicotyledons. 2 olumes. Cambridge Univ. Press, Cambri CRONQUIST, e 1968. The iue and Classification iai juge Houghton Mi An Integrated i of € lassification of Flowering Pus Columbia Univ. Press, New DAHLGREN, R. M. T 1961. Kenya Trees and Shrubs. Buchanan’ s Kenya Estates, Nairobi. Dinc Hou. 1957. A ponies of the genus Bruguiera (Rhizophoraceae). Nova Guinea, n.s. 8: 163-171 19 H Sane Flora Malesiana, se- ving 1 5: 429-493. 1960. A review of the genus Rhizophora ud special merou to the Pacific species. Blumea 10: ppp Dure, N. C. & J. S. Bunt. 1979. The genus Rhizoph- ora (Mie boris in north-eastern Australia. Aus- tral. J. Bot. 27: 657-678. mM B.C. 1829. Analyse des Familles des Plantes c l'Indication des dod Genres qui s'y Rat hent. J. Casterman, Tournay. bici i E. 1840. DE Plantarum. [Frederick] Beck, Vienna. Vindobonae. FLORET, J.-J. 1 Comiphyton, genre nouveau ga- bonais oe eee isieae. Adansonia, ser. A propos de Comiphyton gabonense (Rhizophoraceae — Macarisieae). Adansonia, ser. 2, 16: 39-49. E J.S. 1919. Flora of the Presidency is Madras, part 3. Botanical Survey of India, Calcut HUTCHINSON, J. . DALZIEL. 1954. lora ‘of West xi ropical Africa, 2nd edition. Crown Agents for Over- eas Governments and hice inii London. JUNE F. R. 1961. Wood Plants of Ghana. Oxford Jniv. Press, London JUNCOSA, A. 2 Developmental morphology of the embryo n: ve in Rhizophora ne L, ae). Amer. J. Bot. 69: 1599 neu SON. Systematic compariso e some biological characteristics of Rhizophora- eae and Anisophylleaceae. Ann. Missouri Bot. Gard. (this volume). Bui i uA C. 1753. Species Plantarum. L. Salvii, Stock- Miu W. 1968. A general account of the fauna and flora of mangrove swamps and forests in the Volume 75, Number 4 1988 Juncosa & Tomlinson Historical & Taxonomic Synopsis of Rhizophoraceae/Anisophylleaceae 1295 Indo-West Pacific region. Advances Mar. Biol. 6: 73-270 Marco, H. F. 35. Systematic anatomy "à m woods of the Rhizophoraceae. Trop. Woods 44: 1-20. MELCHIOR, H. (editor). 1964. A. Engler's Syllabus der Pflanzenfamilien, 12th edition, Volume 2. Gebrüder idi us ar Berlin. MIQUEL, F. . 1855. bs van Nederlandische Indië. < der Post, Amsterdam OLIVER, D. 1871. Flora of Tropical Africa, Volume 2. . Reeve, London. kir G. T., M. FREITAS DA SILVA, B. W. ALBUQUERQUE, I. DE J. DA SILVA ARAUJO, L. M. MEDEIROS CARREIRA, . NOGUEIRA Braca, M. Macepo, P. N. DA Co ONCEICAO, P. L. BRAGA LisBÓA, P. I. BRACA ,R.C. LoBATO LisBÓA & R. C. QUEIROZ VILHENA. 1975. Revisáo taxónomica das espécies amazónicas de Rhi- : 5-22. 1922. The Flora of the Malay Peninsula, Volume 1. L. Reeve, London. RUMPHIUS, G. E. 1741-1755. Herbarium Amboinense. msterdam SALVOZA, F. M. 1936. Prec Nat. Appl. Sci. Bull. ui Philipp. 5: 179-2 SCHIMPER, A. F. W. 1898. Rhio barajas : En gler & K. Prantl (editors), Die Natürlichen Pflan- zenfamilien. Engelmann, Leipzig SMITH, A. C. 1981. n A Tropical B Flora Vitiensis Nova, Volume 2. Lawai, Kauai, Ha- Ual ucil, nes T. A. & L. A. BoopLe. 1909. Kokoti (An- opyxis ealaensis, Sprague). Kew Bull. 1909: 309- 312. STEYERMARK, J. A. & R. LIESNER. 1983. Revision of the e genus Sterigmapetalum (Rhizophoraceae). Ann. Missouri ^ui ard. 70: 179-193 198 TAKHTAJAN, . Outline of the classification of ono pod 5 i a Bot. Rev. (Lan- caster) 46: THORNE, R. F. p pe erences classification of the Angiospermae. Evol. Biology 9: 3 Tose, H. € P. H. RAVEN. Seed morphology and anatomy of Rhizophoraceae, inter- and infrafamilial re- ationships. Ann. Missouri Bot. Gard. (this meh TOMLINSON, P. B. 1978. Rhizophora in Australas some clarification of taxonomy and distribution. J. ii Arbor. 59: 156-169. — 86. The Botany of Mangroves. Cambridge Un x Press. Cambridge VLIET, G. J. C. M. VAN 1976. Wood anatomy of the Rhizo Sra Leiden Bot. Ser. No. 3: 20-75. WiLLIS, J. 1973. A Dictionary of the Flowering Plants i Ferns, 8th edition (revised by H. K. Airy Shaw). Cambridge Univ. Press, Cambridge. SYSTEMATIC COMPARISON AND SOME BIOLOGICAL CHARACTERISTICS OF RHIZOPHORACEAE AND ANISOPHYLLEACEAE' Adrian M. Juncosa? and P. Barry Tomlinson* ABSTRACT Systematic and biological characteristics of Rhizophoraceae and Anisophylleaceae are detailed. Comparison of a wide variet two families. bcn dn such a e basically di choripetalous dicot yledons as to be 2 nía ant. One a l unrelated families Ee and is ob fie ed to be a homoplasy. or appendiculate petal margins, occurs in s ist de mod deri a Te analysis of Wak alero shows that, p surprisingly, information y of vegetative a qe bu uctive characters reveals virtually no points of agreement between p iplostemonous pane ium are of such wide occurrence morphy found i in both families, divided ow in han a e genera and clades of ihe tribe. The remainder of the The Rhizophoraceae (used sensu stricto through- out this article) and Anisophylleaceae have often been treated as a single family, although some recent phylogenetic treatments of the angiosperms have placed them separately (Cronquist, 1981; Dahlgren, 1980), as originally suggested by Ridley (1922). We believe that this | stems largely from the absence of detailed infor- ack of consensus mation on a wide variety of systematic characters of the two families, rather than from differences in interpretation. Detailed discussions of several specific suites of characters, such as pollen and leaf architecture, appear elsewhere in this volume, so we have emphasized other, mostly morpholog- ical, characters that are either not widely under- stood for these two families or not generally con- sidered in systematic comparisons. A summary of the systematic differences between Rhizophoraceae and Anisophylleaceae in these characters forms Table 1. Despite the fact that most botanists are some- what familiar with the mangrove Rhizophoraceae, particularly the genus Rhizophora, the biology of these plants is widely misunderstood. Accordingly, in a second section of this article, we discuss some of the biological adaptations to the mangrove hab- itat that are found in Rhizophoraceae, with as much comparative reference to the inland genera as the current state of our knowledge permits. SYSTEMATIC COMPARISON DISTRIBUTION AND HABITAT Rhizophoraceae and Anisophylleaceae are trop- ical families of shrubs and trees; only a few species of mangrove Rhizophoraceae stray beyond 22° lat- ' We thank Paul P. K. Chai and the Forest Department, Kuching, Sarawak; John S. Bunt, Norman C. Duke, and the Australian Institute of Marine Science; the Forest Aaa Vind i; an nd the Servicio E p wass Nacionales, Costa Rica, for permission to collect and assistance in Ang es. We E Mas Jack B. Fisher, iroshi Tobe, and Peter aven for supplying material of sev and Mon la and Ed Seling for technical assistance; Nancy Dengler and Malcolm Gill kindly ed cease and photographs, es guod W nk James Doyle for advice and computer time. Re ppo s come directly from ` dissertation M 1 DEB 80-16635 to A. Juncosa and research grant BSR 82-16271 (P. B. Toe P.L) « 1 grant vie Geographic Societ . B. Tomlinson. Additional support has c tome te fron n the ae Foundation and Atkins 956 16. arvard Univers it) nt Aie Department of Agronomy snd Range Science, University of California, Davis, California ). A. 3 owed Forest, Petersham, Massachusetts 01366, U.S.A. ANN. Missouni Bor. Garb. 75: 1296-1318. 1988. Volume 75, Number 4 1988 Juncosa & Tomlinson 1297 Systematic Comparison of Ficures 1, 2. Aer Rhizophora harrisonii (Costa Rica) .— 2. Stilt of knee roots is illustrated in Figure 29 itude, and then only on coastlines with warm cur- rents. Taxa of both families occur in most of the major moist-tropical floristic regions. The prepon- derance of genera with only one or two species and the narrow geographic ranges of those genera (Juncosa & Tomlinson, this volume) suggest that both may be relictual families. For its size, the Rhizophoraceae have an ex- ceptionally wide ecological and geographic range. e mangrove taxa are found on virtually all trop- ical coasts, and inland species grow in many moist forest types, both primary and successional. A few species, mostly in the genus Cassipourea, grow in drier habitats. Several genera of Rhizophoraceae (both inland and mangrove) may form very large trees (to 50 m), but most species are small trees and may begin flowering at heights of only 1-2 m. Anisophylleaceae are characteristically large trees of wet lowland primary forest, although the genus Anisophyllea also includes some elegant small shrubs (e.g., A. disticha). Combretocarpus is a dominant tree of Bornean (fresh water) peat swamp forest, v pend now extinct in peninsular 1) Malaysia (F. S. g, pers. comm., AERIAL ROOTS Members of at least seven genera of Rhizo- phoraceae characteristically form aerial stilt roots, which have been described in greatest detail for the mangrove taxa (Troll, 1943). The genus Rhi- zophora (““root-bearer”) is justly famous for its ial roots of mangrove Rhizophoraceae. Localities given in parentheses.— 1. Stilt roots of and knee roots of Bruguiera gymnorrhiza (Queensland) . Development remarkable stilt roots (Fig. 1), the development, anatomy, and function of which were not under- stood until recently (Cill & Tomlinson, 1971, 1977; Scholander et al., 1955). Ceriops tagal and species of Bruguiera also form stilt roots on the hypocotyl and base of the trunk, which coalesce to form the fluted, conical trunk base that is seen in older plants (Fig. 2). Among the inland genera, Gynotroches and Crossostylis (Gynotrocheae) normally form thick stilt roots on the lower trunk; Crossostylis gran- diflora is known as New Caledonia for this reason (Fig. 3). At least one species of Carallia, C. brachiata, forms abun- dant stilt roots in peat swamps (Ding Hou, 1958). ““palétuvier de montagne" in Thus, prominent aerial roots occur in three of the four genera of Gynotrocheae, the inland tribe that is probably most closely related to the mangroves. umidity of the air rather than the inundation of the soil that is a factor in the development of aerial roots in inland taxa nother root character that is not often consid- ered in systematic analyses is the presence of root hairs. These are formed on roots of Cassipourea seedlings but are not found in any of the members of the Gynotrocheae or Rhizophoreae that we have studied. Absence of root hairs from Rhizophoreae Annals of the Missouri Botanical Garden FIGURES 3-6. Stilt roots Ms Ro odds e bud morpholog) ' of inland Rhizophoraceae.—3. Stilt roots of Crossostylis grandiflora, one of at least three inland genera in which they occur (New Caledonia) .—4. Transverse section of branch tip of Pu EE erties showing decussate phyllotaxy and valvate stipules. —5. Transverse Volume 75, Number 4 Juncosa & Tomlinson 1299 Systematic Comparison of ' J may have a simple functional explanation if, as suggested by Tomlinson (1986), the endodermis is the site of the salt excluding mechanism, rendering an elaborated root surface unnecessary. Conse- quently, the root surface instead is elaborated by production of capillary rootlets (Attims & Cremer, 1967). This does not account for the absence of root hairs from Gynotrocheae, however. Only one genus of Anisophylleaceae is known to possess any kind of aerial roots. Combretocarpus sometimes produces unique, dimunitive (1—4 cm), negatively geotropic aerial roots on the trunk, usu- ally 1-2 m above the soil (or water) level. However singular these may be, they bear no resemblance to the stilt roots of Rhizophoraceae, neither in development nor in mature anatomy and mor- phology, and cannot be considered a synapomor- phy between the two families. WOOD ANATOMY A more detailed consideration of the wood anat- omy of Rhizophoraceae and Anisophylleaceae ap- pears elsewhere (Keating & Randrianasolo, this volume), but several specific points merit brief men- tion here. doni wood anatomical character states are very variable within Rhizophoraceae (Marco, 1935; van ` Vliet, 1976); however, we deem it poor systematic practice to use this variability as license to draw a relationship between the woods of these two families without any consideration of adaptive significance of wood structure. A number of significant differences could be discussed, but we wish to cite only a few. All Anisophylleaceae have alternate intervessel pitting with coalescent apertures, which are not found in any Rhizopho- raceae. A limited amount of alternate pitting occurs in some species of Carallia, clearly as a special- ization that has arisen within that genus and is thus not relevant to interfamilial relationships. The nar- row vessels and scalariform perforation plates of Rhizophoreae are distinctive and related to the low negative pressures induced by the saline environ- ment, as discussed in Tomlinson (1986). At the request of P. Baas, E. Wheeler (pers. comm.) compared wood anatomical characteristics of Anisophylleaceae with those of the Gynotrocheae and her computerized data base of 5,000 species. Although the character set was not specified and clearly did not include coalescent apertures, absent from Rhizophoraceae (van Vliet, 1976), the coded characters of woods of Anisophylleaceae match those of at least some species of Carallia and, to a lesser extent, Gynotroches and Crossostylis. Sig- nificantly, woods of Anisophylleaceae are similar neither to those of the mangrove genera nor, more importantly, to those of any Macarisieae. There- if the wood anatomical similarities between Anisophylleaceae and Carallia were taken as syn- fore, apomorphies, it would then be necessary to include the Anisophylleaceae as a subtribe of Gynotrocheae (see Fig. 27), which in turn requires us to hypoth- esize parallel reversals in at least 20 (probably closer to 50) vegetative, chemical, embryological, floral, fruit, seed, embryo, and seedling characters (see Table 1 and Dahlgren, this volume). As this is both extremely unparsimonious and biologically completely implausible, we conclude that the sim- ilarities in wood anatomy represent homoplasies and are not systematically significant in this case. PHYLLOTAXY AND NODAL ANATOMY Phyllotaxis is the one systematic difference be- tween Rhizophoraceae and Anisophylleaceae that seems to be widely known: Rhizophoraceae have opposite leaves with interpetiolar stipules, whereas most Anisophylleaceae have alternate, exstipulate leaves. Certain additional details may ultimately prove helpful in understanding infrafamilial sys- tematics. In all Rhizophoraceae, the interpetiolar stipules bear colleters that secrete gummy substances onto the buds. In Rhizophora stylosa, this secretion contains galactose (Primack & Tomlinson, 1978), but whether its primary function is to deter her- bivory (either through direct toxicity to insects or by attracting insectivorous birds) or merely to lu- bricate the expanding leaves while protecting them from desiccation is uncertain. There are tribal distinctions in bud morphology within Rhizophoraceae. In Cassipourea, the stip- ules are valvate and pubescent (in addition to bear- ing colleters), and the leaves are truly decussate, that is, successive pairs of leaf primordia are ini- tiated at exact right angles to one another (Fig. 4). Some reorientation of the leaves may occur during and after expansion. However, the usual descrip- — section of branch tip of Carallia borneensis, showing imbricate stipules and bijugate phyllotaxy (successive pairs of primordia not perpendicular. 3 primordium; Sp, stipule; T, sicellilas trichomes UR d e glands of Carallia borneensis. C, colleter; G, gland; L, leaf Annals of the Missouri Botanical Garden Volume 75, Number 4 1988 Juncosa & Tomlinson 1301 Systematic Comparison of Rhizophoraceae/Anisophylleaceae tion of phyllotaxis in all Rhizophoraceae as decus- sate holds only for the tribe Macarisieae and for Pellacalyx (Gynotrocheae). Tomlinson & Wheat (1979) showed that phyllotaxis in all genera of Rhizophoreae is actually bijugate, with successive leaf pairs offset by angles of xcept for Pellacalyx, which has many autapomorphic fea- tures, all genera of Gynotrocheae also have bijugate phyllotaxy (Fig. 5). At least several species of Car- allia bear extrastipular glands in addition to the colleters (Fig. 6). Interestingly, all genera with bi- jugate phyllotaxy have imbricate stipules, whereas those with decussate leaves have valvate stipules. As discussed in greater detail later, overall tree architecture in Rhizophoraceae is variable, to some extent in relation to habitat differences. However, the basic architectural character of systematic in- terest is that the trunk and branches are onl minimally or not at all differentiated; plagiotropic axes may be lacking altogether. Anisophylleaceae are universally characterized as having alternate, exstipulate leaves, but at least one species of Anisophyllea, A. disticha, has structures interpretable either as minute stipules or large glands (Figs. 8, 11). The phyllotaxy of plagiotropic branches of this genus is unique among angiosperms: there are four orthostichies of leaves, two of reduced leaves on the upper side, and two of full-sized leaves along the lower side (Figs. 7- 9). Orthotropic axes, in contrast, bear helically arranged reduced leaves in a conventional 2/5 helix (Figs. 10-12; Vincent & Tomlinson, 1983); thus, differentiation of axes is pronounced, at least in this genus. Based primarily on study of herbarium material, Ding Hou (1958) stated that anisophylly of this kind is characteristic of only two species of Aniso- phyllea. However, our fieldwork shows that ani- sophylly is also the rule in plagiotropic axes of at east A. cinnamomoides, A. ferruginea, A. grif- fithii, and A., sp. nov. Ding Hou (1958), but that the reduced leaves in these species are sometimes very small and caducous and hence are rarely seen in herbarium specimens. (Their scars, ordinarily lacking axillary buds, can often be detected.) It thus seems likely that both anisophylly and the unique tetrastichous phyllotaxy are uniform for Asian species of the genus. Anisophyllea cinna- momoides, which forms a moderate-sized tree, has the same architecture as a sapling that has been described in detail for adult 4. disticha (Vincent & Tomlinson, 1983). Consequently, one may in- terpret the latter species as a permanently juvenile form. Herbarium study of virtually all other de- scribed species of the genus revealed that aniso- phylly occurs in all but a group of two or three closely related species (Juncosa, pers. obs.), a con- clusion corroborated by the field studies of Floret (pers. comm.). He additionally communicated that habitat subject to both severe seasonal drought and periodic fires, axis differentiation is seasonal rather than architectural. Our knowledge of phyllotaxy and architecture in other genera of Anisophylleaceae is even poorer than in Anisophyllea; no reference mentions an- isophylly in the other three genera. It does occur in Poga and Polygonanthus, although the strict alternation of leaf types seen in Anisophyllea is not preserved in all parts of the axes of Polygonan- thus: several small leaves occur in succession at the bases of some branches, but these reduced leaves may be lacking distally (Juncosa, pers. obs.). A more detailed discussion of leaf anatomy ap- pears elsewhere in this volume, but a few characters bear mention here. It is not generally understood that in most inland Rhizophoraceae, juvenile and usually also the adult leaves are variously toothe or crenate, not entire as is usually stated. Only in Rhizophoreae are the leaves consistently entire, one of many characters in which derived states (within the family) are exhibited by the specialized mangrove genera. Other leaf characteristics of this tribe, such as the succulent hypodermal layer, ter- minal tracheids, and frequently abundant sclereids, are probably all adaptations to the mangrove hab- — FIGURES 7-12. leaved Sarawak variety of A. dis e pseudowhorls of plagiotropic branche s and o apex of orthotropic axis. din leaf primordia e of sc ( Micrograph courtesy of Nancy Dengler.) — leaves m arranged in a al phyllotactic d Habit and development of axes of cm dcm e WE ticha. Both ranks of sc s (ar 2 2 ophyll. — 10. Top of plant tof À. Pons px Rn raphed in rthotropic axis bear scale leaves. Transverse section of orthotropic axis at level of shoot apex; y? de ca branch of large- cale lea re on upper side of branch.—8. wakas although they do a Transverse section of plagiotropic branch cells; folia age leaves (L) have wasl. SENE ing only scale leaves m: ows) .— M of "Glands" are very evident at RUE g sites. Annals of the Missouri Botanical Garden a FIGURES 13-17. ed ence ae ae of Rhizophoraceae.— 13. Open-branched cymose inflorescence of er of maturation of flowers is indio ated by numbers. First branching event is dichasial, others generally abis di — 14. Bifurcate in other ma up to 32 flowers occur in ev enly bifurcating Vue of R. Coon aborted bracteole and terminal floral apex a. ket) between the a where maturing F Bruguiera parviflora (Malay: angle Phasa were removed deae ‘asciculate inflorescence of Gynotroches sp. (Sarawak) .— 17. SEM of ultimate inflorescence branch of Pellacalyx cristatus. Opposite bractlets are fused into a toothed cup; Pica flower has been removed (site d E). Axillary branch (A) is enclosed by fused prophyllar bractlets Volume 75, Number 4 1988 Juncosa & Tomlinson 1303 Systematic Comparison of T T Z itat. An achlorophyllous hypodermal layer is also found in at least three genera of Gynotrocheae, but not in Cassipourea (Macarisieae), so that char- acter may be of phylogenetic significance as well. Nodal anatomy also seems to distinguish the two families, as the Rhizophoraceae are characterized by multilacunar nodes with split-lateral traces (Howard, 1970, 1979), whereas Anisophylleaceae have unilacunar nodes (Anisophyllea: Geh & Keng, 1974; Vincent & Tomlinson, 1983; Combreto- carpus: Juncosa, unpubl. obs.). INFLORESCENCE The inflorescence in Rhizophoraceae is funda- mentally cymose, both in open-branched and fas- ciculate forms. Flowers of large-flowered species of Bruguiera are solitary, presumably by reduc- tion. The first branching event is dichasial (some- times trichasial in Ceriops and in Carallia bra- chiata); subsequently, branching is usually monochasial (Fig. 13). In Rhizophora, Kandelia, and Crossostylis, inflorescences bifurcate through- out (Fig. 14), but the division appears to be pseu- dodichotomous. Although the apices terminating the sympodial units do not develop into flowers in these genera, their vestiges can often be found (Fig. 15). Interestingly, both open-branching and fasciculate forms (in which the branch internodes do not elongate; Fig. 16) are found in all three tribes. It therefore appears that condensation of the inflorescence evolved in parallel three times. A single bract subtends each branch, thus there is a pair of bracts at each node, even when further development of one of the two branches is sup- pressed in the monochasial portions of the inflo- rescence. Pairs of bracteoles also subtend each flower. The bracteoles bear colleters similar in dis- tribution, development, and mature anatomy to those of the stipules and secrete a sticky, rubbery coating over the floral primordia. In Pellacalyx, the pairs of bracts are fused into a toothed cup- shaped structure superficially resembling an epi- calyx, but subtending minute axillary buds in ad- dition to the terminal flower (Fig. 17). Inflorescences in the Anisophylleaceae are pa- niculate or racemose and usually somewhat open- branched. A single bract subtends each branch and Bowen In beu NA disticha, male and female , but in most allas species on genera in the family, the two floral types are mixed in all inflorescences. Thus, although the differences are to some extent an extension of the phyllotactic differences of the veg- etative shoots, there is no inflorescence character state common to both Rhizophoraceae and Ani- sophylleaceae. FLORAL MORPHOLOGY AND DEVELOPMENT In most Rhizophoraceae the flowers are bisexual, but exceptions are found in both inland tribes. Gynotroches (Gynotrocheae) is dioecious; in male flowers, the ovary and ovules develop and at least the early stages of megagametogenesis occur, but the style atrophies distally instead of elongating. In female flowers, anthers and sporogenous tissue develop, but normal microsporogenesis seems to be interrupted shortly after meiosis. At least some species of Crossostylis (Gynotrocheae) are poly- gamodioecious, with many individuals bearing only functionally female flowers (Smith, 1981). Among the Macarisieae, Blepharistemma and Sterig- mapetalum are reported to be polygamodioecious and dioecious, respectively, but developmental de- tails are lacking. though mature floral structures in Rhizopho- raceae are remarkably diverse, especially in rela- tion to contrasted pollination mechanisms (Juncosa & Tomlinson, 1987), early developmental stages are generally very similar. Floral characters that unify the family include petal development and mature morphology, the generally diplostemonous androecium, and the presence of laticifers. These generalizations are based upon study of nine of the fourteen genera of the family, including all genera of tribes Gynotrocheae and Rhizophoreae and one of Macarisieae (Cassipourea). Inasmuch as the mature morphology of flowers of Macarisieae varies very little, especially in comparison with that of genera of the other two tribes, this survey may be taken as encompassing nearly all aspects of floral evolution in the family. Petals in rhizophoraceous taxa are fundamen- tally bifid, with a prominent terminal arista, and they enclose groups of one or more stamens in- dividually, rather than forming a whorl that col- lectively surrounds the androecium as a whole (Fig. 18). This distinctive petal vernation is to our knowl- edge found in only one other family of dicotyledons, Rhamnaceae. Usually, several to many filamentous appendages develop on the distal margins of the two main lobes of the petal (Fig. 19). Mature petals of Crossostylis appear to lack appendages, but this results from the suppression of development of appendages that are initiated in exactly the same mode and position as in other genera. This may also be the case in Anopyxis, which is described as having entire petals. Only in Rhizophora, the most specialized genus in the family, are the mar- ginal petal appendages truly absent, even though 1304 Annals of the Missouri Botanical Garden Volume 75, Number 4 1988 Juncosa & Tomlinson 1305 Systematic Comparison of r u J petals still bear a terminal arista and enclose in- dividual stamens. Although fringed petals are found in a number of other families, including Aniso- phylleaceae, their mature morphology and prob- ably also early development do not resemble those of petals of Rhizophoraceae. Another distinctive characteristic of rhizopho- raceous flowers is the presence of a layer of lati- ciferous cells in the ovary and calyx. These cells may form a more or less continuous layer (Cas- sipourea and Rhizophoreae; Fig. 20) or may grow radially, usually dividing periclinally (Crossost ylis and Carallia; Fig. 21). In Gynotroches and Pel- lacalyx, the laticifers are further modified into disconnected canals or idioblastic cells. In most genera, the epidermis surrounding the laticiferous layer divides periclinally, becoming 5- 7-seriate. e androecium in most Rhizophoraceae is dip- lostemonous, and the antisepalous whorl of stamens is ordinarily initiated earlier than the antipetalous whorl. Nearly all Macarisieae and at least one genus in each of the other two tribes exhibit this basic floral pattern, which we believe to be ancestral for the family. However, significant modifications have arisen in parallel in all three tribes. For example, in Carallia and Pellacalyx (Gynotrocheae), the antipetalous stamens are initiated earlier than the antisepalous stamens (Juncosa, in press). Also, one genus in each of the three tribes has polyandrous flowers, probably a homoplasy (see Fig. 27). This indirect conclusion, based upon the likely cladistic relationship of the genera, is also supported by the diversity of developmental pathways that give rise to the numerous stamens in these three genera. In Kandelia, the polyandrous mangrove genus, the supernumerary stamens result from the initiation of about five stamens on a large common primor- dium that also produces a petal (Juncosa & Tom- linson, 1987). In the distantly related inland genus Cassipourea (Macarisieae), additional stamens are not initiated together with the petals, but the vas- culature of each petal is closely associated with that of several nearby stamens. In several species of Crossostylis, the groups of stamens are found opposite the sepals instead of the petals (Smith, 1981) All Rhizophoraceae have a single style and stig- ma, although in Gynotroches the stigmatic lobes may be rather long. The separate stigmatic lobes may be discernable at gynoecial initiation (e.g., Gynotroches, Pellacalyx), or the gynoecium may arise as a single toroidal primordium (e.g., Car- allia, Bruguiera). Most Macarisieae have fully superior ovaries, but two genera of this tribe and most other genera in the family have half-inferior or completely inferior ovaries. Details of the de- velopment and vasculature of the ovary appear in Juncosa (in press). The occurrence of a superior ovary in female flowers of Gynotroches is probably a reversion from the half-inferior condition, which occurs in the functionally male flowers of the same genus. (As detailed above, unisexuality occurs in this genus by very late-developmental changes; the basic morphology of the two kinds of flowers is quite similar.) Additional support for this hypothesis comes from the diversity of placentation types in Gynotroches (Ding Hou, 1958); placentation in all other genera of Rhizophoraceae is invariably api- cal-axile. Rhizophoraceae typically have 3-5 carpels (whether the locules are completely separate or not) and only two ovules per carpel, but an increase in the total number of ovules and seeds has evolved in several ways. In Crossostylis, this is achieved by multiplication in the number of carpels, up to Pellacalyx both the number of ovules in each is increased. The presence of a floral disc is often used as a number of carpels and the systematic character, but an imprecise understand- ing of the initiation and development of this organ — FicunEs 18-23. Floral morphology and anatomy of Rhizophoraceae and Anisophyllea.— 18. Transverse section of flower bud of Cassipourea elliptica; a petal, largely represented by its filamentous appendages, and the small group of stamens that it encloses are indicated enclosed by i - but stand between their lateral abaxial sides (arrows) .— SEM g filamentous appendages (d ‘and terminal arista (Ar) .— cristatus at a y developmental stage, showin 20. Continuous ir laticifer of e rie exaristata at an early -seriate, but laticifer cells seem only to expand somewhat are beginning in the epidermis, which will become 5-7 by Note d pro aen stamens are not of petal of Pellacalyx a bracket. y stage of development. Periclinal divisions periclinally, not Ph divide. — 21. Longitudinal section of flower of Crossostylis biflora, showing anticlinally expanded laticiferous cells (L); these seem also t o divide periclinally, but this is difficult to establish. —22 e ial longitudinal section of flower of Carallia dust showing that the putative nectary (N) is strictly intrast j P St, stamen; P, petal.—23. Epi-illumination light micro, graph of c nal. crenate nectary (N) of Anisophyllea obtusifolia. ) Divisions extend to the base of the nectary. (Material courtesy of Hiroshi Tobe. 1306 Annals of the Missouri Botanical Garden may have led to considerable confusion and mis- interpretation in the study of angiosperm phylog- eny. In Rhizophoraceae, a nectarial ring arises inside the androecium late in floral development (Juncosa & Tomlinson, 1987). Thus, the stamens are not inserted on this ring, nor is it part of the androecium (Fig. 22). This important distinction is clearly illustrated by the genus Bruguiera, in which the androecium is initiated as a toroidal primor- dium, upon which separate stamen primordia later develop. initiated internally (centripetally) to the androe- cium and ultimately develops into the nectary. Some significant modifications to this basic pattern occur in certain inland genera in tribe Gynotrocheae, but the oft-cited character state “stamens inserted on is certainly incorrect for nearly all Rhi- Later, a separate toroidal primordium is a disc" zophoraceae. Unfortunately, detailed information on devel- opment of flowers of Anisophylleaceae is not avail- able, but descriptions of their mature morphology (Ding Hou, 1958; Tobe & Raven, 1987b) reveal several major differences from Rhizophoraceae. Petals of several genera of Anisophylleaceae are fringed (Poga, Anisophyllea) or weakly divided (Combretocarpus), but the distinctive morphology described above for Rhizophoraceae does not occur here. In particular, the prominent terminal arista is absent, and even in the few cases where the petals are bifid (some species of Anisophyllea), they do not individually enclose one or more sta- mens. Other floral differences are seen in gynoecial morphology. The several styles are separate all the way to their bases in all Anisophylleaceae. The inferior ovary is evidently ancestral for the family (not superior, as in Rhizophoraceae). The mor- phology of the nectary in Anisophylleaceae is un- usual. The mature nectary is both inter- and in- trastaminal and is cleft to its base in many places (Fig. 23). Information on its development is lack- ing, but it is safe to speculate that it is likely to be very different from that of nectaries of Rhizo- phoraceae. FRUIT AND SEED Within the Rhizophoraceae, a clear evolutionar trend can be seen in fruit structure. All Macarisieae have capsular fruits, although these vary in their degree of sclerification and in how readily they dehisce. The tough-fleshy capsules of Cassipourea elliptica (Sw.) Poir., for example, often (always?) drop from the trees unopened, dehiscing on the ground as they dry slightly; the arillate seeds are then carried off in great numbers by leaf-cutter ants. Field observation on the Osa Peninsula of Costa Rica revealed that the ants carry the seeds into the nest, remove the arils, and discard the stripped seeds (Juncosa, pers. obs.). This may seem an unadaptive destination, but Prance (pers. comm.) correctly pointed out that the dispersal function is satisfied by those seeds dropped along the way by accident or as a result of such common distur- bances as rain showers. In Cassipourea, presence of the aril does not inhibit germination. The com- paratively large (3.5 cm), woody capsules of An- opyxis have been described as indehiscent, but we imagine that the description was based on immature fruits; it seems overwhelmingly likely that a canopy tree with winged seeds would also have dehiscent fruits. Consideration of the probable cladistic re- lationship of genera yields the interesting conclu- sion that some or all of the arillate-seeded genera were derived from winged-seeded ancestors. The homology of wing and aril is established by Tobe & Raven (this volume) on the basis of comparative morphology of mature seeds; these authors con- sider the aril to represent the ancestral condition, primarily because of its strong correlation with a superior ovary. Fruits of Crossostylis (Gynotro- cheae) are also capsular, often only partly dehis- cent, the small seeds falling out through the slots. Fruits of all Gynotrocheae other than Crossosty- lis and those of all Rhizophoreae are baccate, and the seeds are naked. In Gynotroches and Pella- calyx, the many small seeds are embedded in a nearly homogeneous juicy berry. In Carallia and in Rhizophoreae, a distinctive loose spongy region develops below the ovules, allowing for the rapid early expansion of the seed(s). In Ceriops, Kan- delia, and Rhizophora the superior portion of the ovary develops greatly in fruit. 'The seed coats of all Macarisieae and most Gy- notrocheae are characterized by a sclerified exo- tegmen (Corner, 1976; Juncosa, 1984a and un- publ.; Tobe & Raven, 1987b), but this structure is absent from seeds of Carallia and the Rhizo- phoreae. That this absence probably represents an evolutionary loss is supported by developmental evidence: the outer epidermis of the inner integ- ument of Carallia ovules is histologically distinc- tive, as in other inland genera, but the cells fail to expand and sclerify. Fruits of most Anisophylleaceae are drupaceous, unlike those of any Rhizophoraceae. The fruit of Combretocarpus is light and three-winged, so it is not surprising that the development of a heavily sclerified endocarp is suppressed. The seed coats of Anisophylleaceae lack a sclerified exotegmen and differ in other respects from those of Rhizo- Volume 75, Number 4 1988 Juncosa & Tomlinson 1307 Systematic Comparison of FicunEs 24-26. Seedling germination in Rhizophoraceae. —24. Epi tica.—25. Viviparous germination by growth of hypocotyl in Bruguiera exaristata. — igeal germination in Cassipourea ellip- 26. Viviparous germination by expansion f eee in Rhizophora mangle. C, cotyledon; E, endosperm; Emb, embryo; H, hypocotyl; Te, esta or seed c phoraceae (Tobe & Raven, 1987a). Seeds of Ani- sophylleaceae also lack endosperm (Floret, 1979), which is present in all Rhizophoraceae. EMBRYO AND GERMINATION As far as is known, all inland Rhizophoraceae have epigeal germination and foliaceous cotyledons (Fig. 24) (Cassipourea: Juncosa, 1982a, b, 1984a; Carallia, Pellacalyx, and Gynotroches: Geh Keng, 1974; Ng, 1980; Ng & Sanah, 1979; Ng, pers. comm.). Development of the embryo to a large seedling while still attached to the parent plant (vivipary) characterizes the Rhizophoreae and offers interesting material for comparative study. Germination morphology in these (mangrove) gen- era has been classified as “Durian-germination,” in which the hypocotyl elongates, but the cotyle- dons remain hidden (Ng, 1978). Both outgroup comparison (with all inland Rhizophoraceae) and comparative developmental morphology indicate that this is a modification of ancestral epigeal ger- mination. Among the Rhizophoreae, cotyledonary mor- phology varies. Bruguiera species have two or three cotyledons, which are thick and fleshy and remain permanently within the seed coat through- out the viviparous seedling development. The development of these cotyledons and their median- plus-split-laterals vasculature reveals their folia- ceous ancestry; they also subtend axillary buds (Juncosa, 1984b). In the more advanced Rhizo- phoreae, however, the cotyledons arise as a single toroidal primordium and form a solid cylindrical body, sometimes becoming 2—3-lobed distally; it is vascularized by many separate traces, evenly spaced. Vasculature of the cotyledonary body of Kandelia is intermediate between that of Bru- guiera and that of Ceriops and Rhizophora. n addition to being the only genus in the tribe with separate cotyledons, Bruguiera also exhibits the intermediate condition in the evolution of vi- vipary: the embryo grows to fill the seed complete- ly, then germination occurs by elongation of the hypocotyl, much as in the nonviviparous inland genera (Fig. 25). Endosperm is not involved in viviparous germination. However, in Ceriops, Kan- delia and Rhizophora, the micropyle is forced open by growth of the endosperm, which in Rhi- zophora may even carry the undifferentiated proembryo partially or entirely out of the seed (Fig. 26). In all genera, the cotyledonary body subse- quently grows to fill the seed coat, with only a thin layer (usually one cell thick) of intervening endo- sperm. This persistent endosperm forms transfer cells (Juncosa, 1982a, b). A considerable amount 1308 Annals of th Missouri Bold Garden TaBLE l. Summary comparison of Rhizophoraceae and Anisophylleaceae. Refer to text for important details and exceptions. Rhizophoraceae (14 genera) Anisophylleaceae (4 genera) Growth continuous rhythmic Branch differ- minimal extreme entiation ood very heterogeneous large vessels, alternate pitting with coalescent apertures Aerial stilt roots formed in Gynotrocheae and Rhizophoreae not formed ai leaves opposite (bijugate or decussate) leaves alternate (distichous or tetrastichous) Stipu present, interpetiolar vestigial or absent Leaf bie toothed margins in inland genera entire margins Nodal anatomy tri- or multilacunar, split laterals unilacunar Inflorescences cymose (dichasial changing to monochasial) racemose (to paniculate) Floral histology subepidermal laticifers no laticifers Petals bifid, fringed, with terminal aristata, convo- sometimes fringed or divided, not folded ute Ovary superior to inferior inferior Fruit capsular to baccate drupaceous or winged Seed coat sclerified ui Md in most inland genera sclerified oo absent d album exalbuminou Cotyledons Senen (modified in mangroves) minute or absent Germination epigeal or viviparous hypogeal Sieve-tube plas- Pv-type S-type tids Chromosome = 14, 18, 21, 32 n=7,8 number of endosperm emerges from the seed coat in all three genera. In Rhizophora, merely forms a collar around the (intercalary) mer- istematic upper portion of the hypocotyl (Juncosa, 1982a), but in Ceriops and Kandelia it grows invasively into the ovary wall, forming an irregu- this endosperm larly branched haustorium. mbryo anatomy and germination in Aniso- phylleaceae differ sharply from that of any Rhi- zophoraceae. The embryo in Anisophylleaceae has a massive hypocotyl, with the cotyledons repre- sented by minute scales (Anisophyllea) or even entirely absent, not even detectable under micro- scopic examination (Poga; Floret, 1979). Germi- nation is hypogeal, with all of the morphological characteristics associated with that growth habit (e.g., earliest epicotyledonary leaves cataphylls, not foliaceous as in epigeal germination). In Aniso- phyllea disticha, shoots may arise from both ends of the seed (Geh & Keng, 1974). Regrettably, anatomical information is lacking. CYTOLOGY Chromosome numbers are often regarded as sys- tematically important, but only a few genera of these two families have been counted. Rhizopho- reae uniformly have n — 18 (Sidhu, 1968; Yoshio- ka et al., 1984). Crossostylis has n = 14 (Tobe & Raven, pers. comm.); Anopyxis, n = 32 (Man- genot & Mangenot, 1958). Counts of n — 18 and n — 21 have been reported in Cassipourea (Weiss, 1973; Juncosa, unpubl.); further data from this large genus are urgently needed. Chromosomes throughout Rhizophoraceae are extremely small, most of them being about 1 um long Chromosome numbers in Anisophylleaceae have been counted only in sectioned material (Tobe & Raven, 19872) but show quite different numbers (n — 7, 8), which have not yet been observed in Rhizophoraceae. Other systematic characters that have recently proved revealing tube plastids. Behnke (1982) found that Anisophylleaceae have S-type (starch-containing) plastids, whereas Rhizophora- ceae have the rare PVc-type, in which crystalline protein inclusions are found. Further discussion of this and several important chemical characters ap- pear in Dahlgren (this volume). SYSTEMATIC CONCLUSIONS A summary of the systematic characters of An- isophylleaceae and Rhizophoraceae (Table 1) re- Juncosa & Tomlinson 1309 Systematic Comparison of Volume 75, Number 4 u T J Characters used in cladistic analysis (Fig. 27). See text for full descriptions of many of the one- ABLE 2 word character states. Condition Ancestral Derived More Derived Character (01/10) (11) l. Stilt roots absent present 2. Root hairs present absent 3. Phyllotaxy decussate bijugate 4. Stipule vernation valvate imbricate 5. Leaf margin serrate entire 6. Hypodermis absent present 7. Salt-tolerant? no yes 8. Inflorescence open-branched fasciculate 9. Terminal flower present absent 10. Breeding system hermaphroditic or monoecious dioecious 11. Floral laticifers la radially Mug idioblastic' 12. Petal arista as laterals nie ted 13. Lateral appendages present abor never initiated 14. Petal orientation reflexe erect 15. Androecium diplostemonous polyandrous 16. Hypant absent present 17. Ovary position superior half-inferior inferior 18. Locule formation enclosure schizogeny 19. Carpels 5 3/many 2 20. Ovules per locule 2 5-8 21. Receptacle solid spongy 22. Exoteg present vestigial absent 23. Nucellu crassinucellate tenuinucellate 24. Integument vascularized ulariz 25. Seed appendage wing? none 26. Seeds per fruit several to many 1 27. Fruit dehiscent baccate 28 tyledons se connate 29. Cotyledonary node unilacunar trilacunar multilacunar 30. Extra-ovular endosperm absent present 31. Endosperm transfer cells absent present 32. Viviparous? no yes 33. Germination process by hypocotyl by endosperm expansion 34. Seedling establishment by radicle by lateral roots Autapomorphy Genus 35. Multimery Bruguiera 36. Explosive pollination system‘ Bruguiera 37. Cotyledonary axillary buds Bruguiera 38. Coleorhiza Bruguiera 39. Extrastipular glands Carallia 40. Androecial appendages Crossostylis 41. Stellate pubescence cal 42. Trichosclereids Rhizophora 43. Anemophily Rhizophora 44. Multilocellate anthers Rhizophora 45. Abortion of first epicotyledonary leaves Rhizophora ' The idioblastic laticifers or mucilage cells of Gynotroches and Pellacalyx are here interpreted as derived from the radially expanded ones of other Gynotrocheae. ios may actually be instead a cell type sui generis within the family, but this has no effect whatever on the cladog ? Used here as a perianth/androecium character, eae TE of ovary positi ? Separate analyses were carried out with the aril as the ancestral condition, e the resulting trees were essentially topologically equivalent and equally (not more) parsimonious. * Evolved independently in Ceriops tagal; not present in C. decandra. 1310 Annals of the Missouri Botanical Garden veals virtually no points of agreement. The very few points of vegetative similarity are clearly homo- plasies—for example, entire leaves in tribe Rhi- zophoreae and in Anisophylleaceae. The most fre- quently cited floral character state common to both families is fimbriate petals (e.g., Tobe & Raven, 19872), but this also occurs in many dicotyledo- nous families and is even a diagnostic character of several of these (e.g., Cunoniaceae, Elaeocarpa- ceae). Moreover, as emphasized above, the petals of Rhizophoraceae have a unique pattern of de- velopment and mature morphology and vernation. The ancestral androecial condition in both Rhizo- phoraceae and Anisophylleaceae is diplostemonous, but this can hardly be regarded as a synapomorphy, as it is found in many other apopetalous dicoty- ledons. We feel that the overwhelming number of differences in all aspects of growth, anatomy, and reproductive cycle of Rhizophoraceae and Aniso- phylleaceae make it clear that these two families are not at all closely related. It is to be hoped that the data assembled in addressing this question will also aid in assigning them to their correct phylo- genetic positions. Although more information about the poorly known African genera of Macarisieae will be re- quired before the family can be rigorously revised, a well-supported cladogram (Fig. 27) has been de- rived from analysis of the 45 characters listed in Table 2. Many other characters that we were un- able to study in crucial taxa, or the coding of which could not be satisfactorily resolved, or whose states were unknown in as many as half of the genera, were not included in this preliminary analysis. For the sake of consistency, all anatomical character states noted were derived from our own observa- tions. Computer analysis was carried out using the Phylogeny prue Package (PHYLIP; Felsen- stein, 1985); the PENNY branch-and-bound al- gorithm was utilized to ensure that all most par- simonious trees were found. Characters were weighted equally and all reversals were permitted. Primarily due to the paucity of characters distin- guishing genera of Macarisieae, numerous alter- native most-parsimonious trees were found, but all were essentially topologically equivalent and re- flected only different placements of the clades and genera of this tribe. Specifying the ancestral con- dition of the seed appendage as arillate, as would be indicated by comparison with the hypothetical out-groups (Dahlgren, this volume), neither re- solves this point nor results in a more parsimonious tree. We emphasize again that more characters from a variety of aspects of the plants are required in order to elucidate the relationships of the Ma- carisieae. A complete character analysis and discussion of generic relationships will be presented elsewhere, but several comments are appropriate here. Tribe Gynotrocheae is clearly paraphyletic and is main- tained here and in our synopsis (Juncosa & Tom- linson, this volume) merely for convenience. De- spite the occurrence of arils in Crossostylis, character traditionally given much weight, it is clear that this genus is much more closely related to other Gynotrocheae than it is to Macarisieae. It is also evident that Carallia is more closely allied to the mangroves than is any other inland genus. In fact, C. characters with the mangroves (vascularized integ- ument, frequently single-seeded fruits, complete disappearance of exotegmic cell layer, etc.) that borneensis shares several additional we have conservatively regarded as homoplasies at present. If the extra-stipular glands, the only apo- morphy distinguishing the genus, do not occur in all species, then the genus may have to be regarded as paraphyletic, with C. borneensis and possibly other species placed further up on the clade leading to Rhizophoreae. It is instructive to note which characters are homoplastic in Rhizophoraceae. Polyandry occurs in Kandelia, Crossostylis, and in two subgenera of Cassipourea, all independent origins of this character. As described above, this is also reflected in the different modes of development of the ad- ditional stamens or, in the case of species of Cros- sostylis not yet studied developmentally, in the positions of those stamens. Condensation of inflo- rescences also arose at least three times, as did the suppression of later development of the lateral ap- pendages of the petals; this is hardly surprising, considering the minor developmental changes that are involved. Traditionally, the number of carpels or locules and the position of the ovary have been regarded as systematically important, but it is clear that substantial changes in both of these characters have occurred within the Rhizophoraceae. How- ever, the trend from a superior ovary to half- inferior to inferior is reversed only once, in the female flowers of Gynotroches. In these, a sec- ondary expansion of the superior portion of the ovary occurs, accommodating the increased num- ber of ovules. Male flowers of the genus also form an ovary and (slightly fewer) ovules, but the ovary is half-inferior. In our judgment, between the unbiased cladistic analysis and the developmental evidence strongly supports our sys- the agreement tematic conclusions. Volume 75, Number 4 Juncosa & Tomlinson 1311 1988 Systematic Comparison of u u Z = 2 < = - = > =< = uu nm Er Ww < XQ GC vv u o > > I < c (ou < = = a rm E L o x < e = N = e a e > 2] < o m LJ = Nn I >< — [de] a =< =l I o o a -l — -l a a > oí = — e > a çO < = = paa] uJ o o a =< a N a = = n 4 O < c5 a m N e c uJ o uy o = o 4 = [a4 = = a = z < E < - x O a Lu > < a < uJ I T = v S m A o o a oO ° co x o = 4l* +13 10 2. 4AM* 418 TEN 19 19 Lip +17* 1-35 Fl +36 +37 +38 19 m +11 +20 +23 +39 19 +13 +15 T1 + t T14 +17 +21 T 22 25 27 +1 +2 +3 +4 16 +11 +12 +16 +17 pa RE27. Probable phylogenetic relationshi era of R most-parsimonious cladogram repalting from analysis of the characters listed i in fols 2 by PHYLIP de € “iterative ——— rsimonious trees reverse the positions of the two basal clades and the genera A term them; all thes don basically represent trichotomies. On the diagram, each mark indicates a one-step Mak in character state ius two marks for the exappendiculate petals of Rhizophora) ; characters Bos as changing in more than one place may be multistate or homoplastic or both. Asterisks indicate reversals. 1312 Annals of the Missouri Botanical Garden ECOLOGY AND BIOLOGY The ecological rips of mangroves is presented by Tomlinson (1986: 23) as that of species com- bining the vee li of pioneer species (e.g., early flowering, wide distribution, extended or even con- tinuous flowering and fruiting, short period of dor- mancy) with attributes of mature-phase species (dense wood, slow growth, large seeds). It is sug- gested that this is because there is no real succes- sion in the community (although zonation is pro- nounced) so that individuals have to be both pioneers and climax constituents at different stages in their life span within communities that are inherently unstable. The vegetation itself (mangal) largely shows the characteristics of a pioneer community. Little is known about the ecology of terrestrial taxa in both families; this makes difficult the search for possible ancestral traits of mangrove taxa within extant rain forest plants. The following paragraphs outline some of these morphological and biological features, with inevi- table emphasis on mangrove taxa, which have been most extensively studied. These plants are unusual in the possibilities they offer, not only for systematic out-group comparison, but for ecological out-group comparison, since it is useful to establish those features that occur in unrelated mangrove taxa, i.e., in mangrove members of other families, and which may suggest themselves as features of direct functional significance, and therefore of limited systematic and phyletic value. Aerial roots are an obvious example. In contrast, some features occur in mangrove taxa but not in their terrestrial or ecological relatives. AERIAL ROOTS The aerial roots of Rhizophora are all initiated from above-ground parts and develop as a series of sympodial loops, which branch aerially only when they are damaged or when the root apex becomes anchored at the end of a loop (Fig. 28). Sympodial branching is adventitious and is of two contrasted kinds. After damage it is distal, i.e., immediately behind the damaged portion, but after anchoring it is proximal, i.e., some distance behind the point of anchorage. A remarkable anatomical transfor- mation occurs in aerial roots as they become sub- merged distally. Aerial roots have extensive de- velopment of trichosclereids, develop secondary xylem, and branch infrequently lack aerenchyma, and adventitiously. Submerged roots lack tricho- sclereids and other mechanical tissues, develop a lacunose aerenchymatous cortex, have little sec- ondary xylem, and branch abundantly and non- adventitiously. This kind of root system seems unique to Rhizophora. On old trees the trunk- borne roots develop as massive flying buttresses, the trunk base itself being obconical (Fig. 1 As described in the previous section, Ceriops tagal and Bruguiera sexangula develop a fluted base formed by the coalescence of clusters of stilt roots that are initiated on the hypocotyl after the seedling has taken root (Fig. 2). This kind of de- velopment is also found in the quite unrelated New World mangrove Pelliciera rhizophorae (Pelli- cieraceae) (Tomlinson, 1986). Flutings of this kind seem also to be the maximum extent of aerial root development in Kandelia. However, in Ceriops and some species of Bruguiera an emergent por- tion of the subterranean root system further de- velops by the periodic looping of the major hori- zontal roots, each loop becoming the site of a woody pneumatophore from which branch roots subse- quently arise (Fig. 29). erial roots in tropical woody plants have a specific function in supplying oxygen to the sub- merged roots by the shortest possible pathway in anaerobic, waterlogged substrates. Consequently, such structural variations can be seen as conver- gences in which developmental modifications result in an identical suite of clearly nonhomologous func- tional components (Gill & Tomlinson, 1975; Tom- linson, 1986). The functional components always include anchoring and absorption, connecting the separate units horizontally (cable component), and providing for aeration. Consequently, we are pro- vided with a clear picture of the limited value of aerial roots in phyletic analysis. At the same time, this aspect of functional morphology could be a necessary preadaptation in a hypothetical ancestor for the Rhizophoreae. Kandelia, which lacks pneu- matophores, makes implausible any direct state- ment about the root system of ancestral mangrove Rhizophoraceae, since it could represent either an ancestral or a derived state. Troll & Dragendorff (1931) denied that aerial roots have a respiratory function and preferred to see them as allowing the trees to root in sediments whose level may fluctuate. The aerating and layering functions are, of course, not mutually exclusive. WOOD ANATOMY The relevance of wood anatomy to ecological as distinct from phyletic considerations is again well borne out by a comparison between the mangrove and nonmangrove Rhizophoraceae. The former are uniformly characterized by relatively narrow ves- sels with scalariform perforation plates, the latter Volume 75, Number 4 1988 Juncosa & Tomlinson 1313 Systematic Comparison of u y J by wider vessels with partly or exclusively simple perforation plates. One can ascribe the differences in vessel diameters to direct ecological causes be- cause they render the wood of mangroves safer in environments where vessel embolisms are likely to be more frequent as tensions are increased because of the low water potential of sea water (Scholander et al., 1965). Safety is maximized by the production of large numbers of narrow elements (Tomlinson, 986; Zimmerman, 1983). On the other hand, scalariform perforation plates, whose function is quite unknown, do not seem to have any necessary significance in relation to water stress as suggested by “ecological out-group comparison.” Vessel ele- ments in all other true mangroves have simple 1950; Panshin, perforation plates (Janssonius, 1932). Significantly, Sperry (1985) provided ex- perimental evidence that scalariform perforation plates in the palm Rhapis excelsa restrict the size of bubbles in recently embolized vessels as they refill with water under positive pressures. Restrict- ing bubble size may facilitate recovery of vessels after water columns are broken during exceptional water stress in any plant. Mangrove Rhizophora- ceae could then have a functional advantage not found in their ecological associates from other fam- ilies. ARCHITECTURE Phyllotactic differences between Rhizophora- ceae and Anisophylleaceae, so far as they are understood, are but that lead to architectural differences of quite a fundamental nature. In the Rhizophoraceae there one of a suite of characters is a strong tendency toward continuous growth and the expression of Attims’s model, whereas Aniso- phyllea suggests rhythmic growth and Massart’s model. Continuous growth has been suggested as an adaptive feature in mangrove taxa simply be- cause the tree must remain permanently active metabolically in order to maintain its salt balance (Scholander, 1968; Hallé et al., 1978; Tomlinson, 1986). Trunk axes are therefore monopodial, al- though they may branch continuously or diffusely. Branching seems always to be by syllepsis (Fig. 31; cf. Wheat, 1981). The branches themselves may then repeat the structure of the parent axis (Attims's model, as in Rhizophora) but progres- sively become plagiotropic by apposition (Fig. 30). Their incipient orthotropy is, however, demonstrat- ed when TE are jelàssed from apical control; reiteration in Rhizophora occurs chiefly by de- differentiation of existing sylleptic branches. Species of Bruguiera, notably B. gymnorrhiza, more nearly conform to Aubréville’s model, since from inception the branches are plagiotropic by apposition. Dedifferentiation of axes is less com- mon. Although distal branches in trees belonging to these contrasted models all become superficially similar, there is a strong underlying difference: Bruguiera tends to remain narrow-crowned, while Rhizophora is very plastic in its crown form, es- pecially as lower branches are readily supported by aerial roots. Very likely part of the ecological success of Rhizophora as compared with Bru- guiera lies in this greater plasticity of form. Little is known about the architecture of inland Rhizo- phoraceae, but superficial study suggests a limited range of crown form comparable to that in Bru- guiera. Carallia and Gynotroches seem to con- form to Attims's model. Anisophyllea, in contrast, provides an extreme example of Massart's model (Vincent & Tomlinson, 1983), since the trunk grows episodically, produc- ing pronounced tiers of branches, which are them selves strongly plagiotropic. The trunk axis sup- ports only scale leaves. The differences in phyllotaxis that underlie this contrast have already been men- tioned. Multiple serial buds occur in Anisophyllea spp. and some other Anisophylleaceae. The extent to which these are committed at inception to become either reproductive or vegetative branches is un- known. Serial buds also occur in Rhizophoraceae but are more strictly committed at the inception of the first primordia on the axillary apex; the pair of bracteole primordia that form on a future inflo- rescence apex can be distinguished from leaf pri- mordia that form on a future vegetative apex. It should be emphasized that in most Rhizophoraceae that we have studied, vegetative branching is by syllepsis, additional vegetative “reserve” buds may be formed as part of a primary branch complex, but their further development is very limited (Wheat, 1981). In both inland and mangrove gen- era, when the plant is in reproductive condition, an inflorescence develops in the axil of each leaf of a pair, and, at least on vigorously growing shoots, a single vegetative bud develops above this, al- though it is often not evident even with a hand lens. In adult Rhizophora, reiteration can occur from these residual buds, but more usually reiter- ation is the result of dedifferentiation of existing branches. Adventitious buds, however, do occur on the hypocotyl of damaged seedlings (Larue & Mu- zik, 1954; Gill & Tomlinson, 1969). In summary, we recognize that Rhizophoraceae and Anisophylleaceae differ in architecture and reiterative ability in ways that support their seg- Botanical Garden issourl H ° < P — o 2 G c c « M Volume 75, Number 4 1988 Juncosa & Tomlinson Systematic Comparison of 1315 Li u J FIGURES 32, 33. enclosed under tension b: regation as two families. If this discussion seems imperfectly formulated, it is simply because the problem of morphological plasticity in woody plants of the tropics itself is a subject that remains little understood, despite its considerable ecological im- portance. FLORAL MECHANISMS Understanding of floral structure and develop- ment stands in an interesting relationship to our knowledge of floral mechanism, which is fairly com- plete for the dps d taxa, from the work of Tomlinson et al. (1979) and Roue et al. (1987), but is scarcely investigated for either terrestrial Rhizophoraceae or Anisophylleaceae. In the Rhi- zophoreae there is an initial contrast between Rhi- zophora, which is wind-pollinated, and the other genera, which are animal-pollinated. Evidence for wind pollination comes from floral mechanism, pol- len-ovule ratios, dispersibility of pollen, and infre- quency of insect visitors, even though the flowers do not display a conspicuous wind-pollination syn- Floral oo in Bruguiera gymnorrhiza.—32. Flower before being tripped; stamens are etals.—33. Similar flower afier being ie i petals spring open, releasing the stamens and pollen nM. Stamens are now visible at center of flowe drome as it is familiarly understood for temperate trees. Animal pollinators of other genera include birds, bees, moths, and butterflies, at least, with each species or group of species visited by a par- ticular type of flower visitor. Floral specialization involves a very distinctive explosive mechanism that physically projects the light pollen onto the biotic interaction relate to differences in flower size, orientation, and attractant (nectar and odor). Ceri- ops decandra and Kandelia lack any specialized mechanism and seem to be pollinated by rather generalized visitors. Of interest is that Ceriops ta- gal and C. decandra are strongly contrasted in their floral mechanism. The existence of light pow- dery pollen in the taxa with animal pollination is an example of the way in which pollen character- istics may be misleading about the method of pol- lination; most animal-pollinated plants have heavy, sticky pollen. In the Rhizophoreae one may spec- ulate that this pollen type is a preadaptation for wind pollination in Rhizophora, which represents == FIGURES 28-31. Rhizophora; branch roots arise from t > ra M. Gill). by 4 sill). Successive stages 30. Older branch of Rhizophora de which Root and branch iind ques in Rhizophoreae. = Development of aerial root system of arches after the main — 29. ‘Development of knee roots in ni tagal (Philippines, from a transparency e indicated A-D. Similar development occurs in h has become plagiotropic by apposition. —31. Sylleptic branching apex enters the substrate (Queensland, from a Bruguiera (see Fig. 2).— (arrows) in Ceriops tagal; branches are elongating even before stipule has fallen from the node. 1316 Annals of the Missouri Botanical Garden a derived condition, and is itself an adaptation to animal pollination by the distinctive explosive pro- jection of pollen. Flower orientation itself provides contrast with reference to classes of visitor. Small- flowered species of Bruguiera have erect flowers, suitable for the approach from above of small, delicate-winged visitors, such as butterflies; large- flowered species of Bruguiera have pendulous flow- ers on recurved pedicels and are approached from below by large visitors like birds. There is evidence that pollination can occur in conditions of autog- amy, allogamy, and gectonogamy, but none for apomixis (Kondo et al., 1987) he general ecological conclusion is that this group of related mangrove plants partitions the available pollinator resource by adopting either dif- ferent mechanisms, or, where there is a common mechanism, by varying it to suit different types of pollinator. CONCLUSIONS The phyletic and systematic conclusions of the various contributors to this symposium are drawn from an examination of a remarkable diversity of characters, ranging from features of gross mor- phology to ultrastructural details of sieve-tube plas- tids. Consciously or unconsciously there is a good deal of weighting of these characters; consciously because there can be bias towards acceptance of a character state if it agrees with a pre-existing position, unconsciously because a single biological character (e.g., wind-dispersed seeds) may be re- flected in several morphological characters (dehis- cent fruits, winged seeds, thin seed coat). Rarely is a character sufficiently understood at a functional level for the bias to be rational. The taxonomist will emphasize that it is impractical to consider the possible biological significance of all attributes of a character in making systematic hypotheses, but the consequences of this restriction should at least be understood. The more information one has about a feature used in systematic analysis, the more likely that its systematic or phyletic significance can be correctly assessed. Subjectivity is an in- evitable consequence of the empiricism of system- atic methodology—the papers in this symposium are replete with examples—and any claim of ob- jective neutrality is particularly inappropriate where the evolutionary polarity of character states is con- tinually invoked. There should be some apprecia- tion of what may be termed “biological” attributes of characters if they are to be manipulated suc- cessfully. Within this broad spectrum of attributes, specific attention needs to be given to three classes of information: correlation among characters; func- tional aspects of characters; developmental history of characters, particularly at the primordial level. An understanding of the way in which seemingly different characters are correlated is necessary since an apparent complex that may or may not be structurally connected may all be subsumed as a single character if the interdependence is appre- ciated. For example, phyllotaxis, stipular mor- phology and nodal anatomy are interlinked; thus, aspects of nodal vasculature may be a direct expression of phyllotaxis, as in the presence or absence of split laterals, distinguishing Rhizopho- raceae from Anisophylleaceae. More subtle phys- iological connections need to be sought, as in the relation between leaf succulence and salt-excluding mechanisms in Rhizophoraceae. The contrast between mangrove and terrestrial taxa in such features as root morphology, leaf anatomy, wood structure, and embryo development could well be cited as characters with little system- atic weight because their functional attributes are at least perceived, if not totally understood —they are the ““Anpassungsmerkmal”’ of Schimper. It is well established that the tribe Rhizophoreae is an advanced group, even though a diagnostic feature is the presence of scalariform perforation plates in its wood. An evolutionary scenario sees this simply as the retention of a putatively primitive character. If this is true, it is surely helpful to know why it has been retained. If the preferred explanation of the function of scalariform plates in restricted air- bubble size in embolized vessels is accepted, we can appreciate its occurrence in a more informed = £5 Developmental information may simply add to the range of characters made available, an attribute that strongly justifies embryological study (cf. Tobe & Raven, 19872, b), but it can also clarify struc- tural, functional, and correlative attributes; for ex- ample, floral development suggests that pleiomery is derived in the Rhizophoraceae. The best example is provided by Kandelia. Here the feature is related to a rather unspecialized floral mechanism that may be derived, not an ancestral feature. Currently we have no evidence for a specialized pollinator group in this genus. Comparative study of embryo de- velopment shows a trend of modification leading from epigeal germination to vivipary, with Bru- guiera the least specialized within the viviparous group. “Vivipary”” becomes more useful as a sys- tematic character when viewed developmentally, even though we do not understand its functional significance. Developmental study may also reveal Volume 75, Number 4 Juncosa & Tomlinson 1317 1988 Systematic Comparison of the extent of convergence, in which superficially red mangrove (Rhizophora mangle L.). 2. Growth similar structures show contrasted developmental pathways; the diversity of root systems is a very gross example These examples relate to “primordial develop- ment" (Tomlinson, 1982); “ontogenetic develop- ment" provides an independent set of attributes. The distribution of axes with contrasted kinds of phyllotaxis in Anisophylleaceae clearly has an on- togenetic component that is still incompletely ex- plored. Once it has been done and its correlation with stem vasculature worked out, we are likely to be in a position to make evolutionary statements, because some of the phyllotactic patterns in this species are clearly derived, as indicated by their uniqueness. Inevitably these considerations of analysis" to which we "character ave drawn attention may lie in the realm of “‘consummations devoutly to be wished” as far as practicing systematists are con- cerned. Nevertheless we hope that investigations of functional, developmental, and correlative at- tributes can be seen to play a central role in sys- tematics. LITERATURE CITED ATTIMS, Y. & G. CREMER. 1967. Les — capil- RM ek palétuviers dans une mangro e Cóte d'Ivoire. Adansonia, ser. 2, 7: 547- 551. BEHNKE, H. -D. 1982. Sieve-element plastids ü V laceae, Faythroxylaoeae and Rhizophorace de- cance of subtype PV ene PI. -39. 1976. The Seeds of Dicotyledons, 2 s. Cambridge Univ. Press, oT ge. docs A 1 . An Integrated System of Classi- cation of Flowering Plants. Columbia Univ. Press, ew Yo rk. DAHLGREN, R. M. T. 1980. A revised system of clas- sification of the angiosperms. Bot. J. Linn. Soc. 80: Rhizophoraceae and Anisophylleaceae: sum- mary "oh relationships. Ann. Missouri Bot. Gard. (this volume). Dinc Hou. CH „Rhizophoraceae. Flora Malesiana, ser. 1, 5: 429-49: FELSENSTEIN, J. Pu netics Confidence limits on. phyloge- s: an approach using the bootstrap. Evolution 39: 783-791. FLORET, J.-J. 1979. A propos du contenu séminal dans les genres Anisophyllea et Poga (Rhizophoracées Anisophylloidées). Adansonia, sér. 2, 19: 109-115. GEH, S. Y. & H. Kenc. 1974. Morphological studies on some inland oe Garden Bull. Straits Settlem. 27: 183-220. GILL, A. M. & P. B. di 1969. Studies on the growth of red mangrove (Rhizophora sia L.). I. Habit and ld morphology. Biotropica 1: 1-9. € —— ———. 1971. Studies on the iwi of and differentiation of aerial roots. Biotropica 3: 63- TT. & Aerial roots: an array of form and functions. Chapter 12 in J. G. Torrey & D. T. Clarkson us. The Development and Func- tion of Roots. Academic Press, London. & Studies on the growth o . The adult red mangrove Man Ard Ts root system. Biotropica 9: 5. HALLÉ, F., R. A. A. OLDEMAN & P. B TOMLINSON. 1978. Tropical Trees and Forests— An Architectural Anal- ysis. Springer Verlag, Berlin Howarp, R. A Some observations on the nodes of woody plants with special reference to the problem of 'split-lateral' versus the *common-gap'. In: N. K. B. Robson, D. F. Cu 53 & M. Gregory (editors), New Research in Plant Da Bot. J. Linn. Soc. 63, Suppl. 1: 194-214 1979. The stem-node-leaf continuum of the Disoryledansdis Chapter 8 in C. R. Metcalfe & L. Chalk (editors), Anatomy of the Dicotyledons, 2nd edition. Clarendon Press, Oxford. ii aina H. H. 1950. The vessels in the wood of an mangrove trees. Blumea 6: 465-469. Juncosa, A. M. 1982a. Developmental morphology of the embryo and seedling in Rhizophora mangle L. (Rhizophoraceae). Amer. J. Bot. 69: 1599- 1611. 1982b. Embryo and m development in the Rhizophoraceae. Ph.D. Thesis. Duke University, Durham, North Carolina 1984a. Embryogenesis and developmental morphology of the seedling in Bruguiera p rm Ding Hou (Rhizophoraceae). Amer. a Bot. 71: 180- 19]. . 1984b. Embryogenesis and seedling devel. opment in Cassipourea elliptica a (Sw) Poir. (Rhi- zophoraceae). Amer. J. Bot. 71 . Floral development and danan —Ó in Rhizophoraceae. /n: P. Leins, S. C. Tucker Endress (editors), Aspects of ipie Development. Gebrüder Borntraeger, dw lin. - Tou 1987. "Hi | devel t in Mancia e d Sai eae. Amer. J. Bot. 74: 1279. KEATING, R. C. & V. RANDRIANASOLO. Leaf aic ud norrhiza and Rhizoph ceae) in pou per the Ryukyu Islands, Japan. iiie E ARUE, C. T A ies 1954. Growth regener- ation and uen rooting in Rhizophora mangle. Pap. Michigan Acad. Sci. (Part 1, Bot. and Forest.) S -29. MANGENOT, S. € G. MANGENOT. 1958. Deuxiéme liste de nombres chromosomiques nouveaux chez diverses dicotylédones et monocotylédones E DUO Occi- dentale. Bull. Jar. Bot. État 28: 31 Marco, H. F. 35. CAE i HEN at the woods of the Rhizophoraceae. Trop. Woods 44 0. LFS Strategies ‘of aa in Ma- layan forest trees. Chapter 5 in P. B. Tomlinson & M. ), Tropical Trees as pag a Cambridge Univ. Press, Cambridge 1318 Annals of the Missouri Botanical Garden 1980. Germination ecology of Malaysian woody plants. Malayan Forester 43: 406-437. & Mar ASRI BIN NGAH S 1979. Ger- mination of fresh seeds of Vt trees IV. Ma- layan Forester 42: 221-224 PANSHIN, A. J. 1932. An anatomic 'al study of the woods of the dip cda mangrove swamps. Philipp. J. Sci. 48: 14 PRIMACK, R. B. P P. B. ToMLINSON. 1978. Sugar secretions from the buds of Rhizophora. Biotropica : 74-15. EE A "x 1922. bu pe ix the Malay Peninsula, . L. Ree SCHOLANDER, P. F. M e oa desalinate r. Physiol. Plant. 21: 25 , L. Van DAM ek a 1955. Gas exchange in the roots of mangroves. Amer. J. Bot. 42: 92-98. . T. HAMMEL, E. D. BRADSTREED & E. A. HEMMINGSEN. 1965. Sap pressure in vascular plants. Science 148: 339-340. SIDHU, S. S. 1968. deg studies on the e | a. Caryologia 21 i im Vitiensi Nova, e K 1, Lawai, Kauai, Ha- SMITH, A. C. 1981. ds "d Tropical B Shean, ‘Ls 1985. Xylem ere in the palm Rha- pis excelsa. pas Bull. 6: 283-292. Tope, H. & P. H. Rav | Systematic embryol- ogy of the jee, Ann. Missouri Bot. 26. 74: 1- . The embryology and re- lationships of Cassipourea an eri ie ion Tris La deitas Goa Bot. 92: 253 264. & —————. Seed morphology and anatomy of Rhizophoraceae, inter- and infrafamilial rela- tionships. Ann. Misouri Bot. Gard. (this volume). ToMLINSON, P. B. 1982. Chance and design in the construction of plants. Pp. 162-183 in R. Sattler (editor), Axioms and eure of Plant Construction. Martinus Nijhoff/W. Junk, The Hague —. 1986. The Botany of Mangroves. Cambridge Univ. Press, Cambridge. & D. W. WHEAT. 1979. Bijugate oe in r. (Rhizophoraceae). Bot. J. Linn. Soc 78: 317 .B. pon MACK & J. S. Bunt. 1979. Prelim- inary observations on floral biology in mangrove Rhi- zophoraceae. Biotropica 11: 256-277. TROLL, W. 3. ar ue Morphologie den ho- heren Pflanzen. Bd. . Vegetationsorgane. Ge- brüder Borntraeger, Be eS & O. DRAGENDORFF. 1931. Ueber die Luft- wurzeln von Sonneratia a E ihre biologische Be- deutung. Planta 13: 311-473. VINCENT, J. R. & P. B. TOMLINSON. . Architecture and phyllotaxis of Anisophyllea de eri ho- raceae). Gard. Bull. Straits Settlem. 36: 3-18. VLIET, G. J. Wood anatomy of Rhizophoniceas. Leiden Bot. Ser. 3: 20-75. Weiss, H. 1973. Contribution à la cytotaxonomie et à s caryologie des palétuviers de Madagascar et du onde. Rev. Gén. Bot. 80: 209-240. Mana. D. W. 1981. Sylleptic branching in the Rhi- zophoreae of d Rhizophoraceae. Bot. Gaz. (Craw- fordsville ) 142 123. YOSHIOKA, H., K. Kon DO, M. LEGRAND, K. NEHIRA & S. MaxEDA. 1984. K G bel studies in five Rhizophoraceae. species of mang a Kromosomo II-3 ZIMMERMANN, M. H. ve ia Structure and the Ascent of Sap. Springer, Berlin SEED MORPHOLOGY AND ANATOMY OF RHIZOPHORACEAE, INTER- AND INFRAFAMILIAL RELATIONSHIPS! Hiroshi Tobe? and Peter H. Raven? ABSTRACT We present an overall study of the seed morphology and anatomy of all Rhizophoraceae (10 inland and 4 mangrove gene orphologically seeds are arillate, winged or nonappendaged; both arillate nsi winged seeds are borne in capsular fruits, and nonappendaged seeds in baccate or indehiscent hard-walled c amis es. Seed coat anatomy is diversified in correlation with the seed iie morphology, but a well-develop ud exotesta and a fibrous exotegmen are common to all inland genera. Despite certain minor divergences, pls o different genera of Rhizophoraceae is defined as exotestal, exotestal-exotegmic, or undifferen jated. y) combinations of seed morphological and anatomical features characterize different "boni or aro s of genera An overall comparison ^, seeds and other reproductive characters confirms that in Rhizo otegmen are plesiomorphic features that can families. Seed morphology and anatomy also support manng hizophoraceae wit and the presence of a fibrous ex Celastraceae, and the exclusion of Elaeocarpaceae from that nA been assigne (particularly a superior ovar whereas the first t three have arillate seeds. Among th (i.e., an um ovary) ; it retains many plesiomorphies. a nonappendaged seed and a persistent meso- and endotegmen, both clearly "s and Pellacalyx “further share some distinct synapomorphies, a their close affinit genera—Bruguiera, Ceriops, Kan ey rallia, Gynotroches, and Pellacalyx ios to Macarisi infrafamilial classification is revised, and a new tribe proposed. o oubt. Cladistically Ne E genera ceae arillate seeds e used f searching for related aeocarpaceae an ales. Our comparison further suggests that the seed y of Rhi zophoraceae have €) as the result of adaptation to different methods of er. We carried - Blepharistemma, herd Comiphyton, Anopyxis, Macarisia, and Sterigmapetalum, o Macarisieae, are characterized by having many plesiomorphies The last e Lgs have winged seeds and a thinner seed coat (apomorphies) , e four remaining inland genera, no ave been assigne to Gynotrocheae, Crossostylis (with arillate seeds) differs greatly from the others in having only one apomorphy In contrast, gon allia, Gynotro sis bg ud Pellacalyx T d nec ae and p s. The four ma i oa whic e been segregated as Rhiz ophoreae, share and e Y niya lack of the tegmen). ue efe are closely related to sieae. on our cladistic bini the traditional Eus dee, which consists of Crossostylis only, is Traditionally, Rhizophoraceae have been broad- ly defined to contain one mangrove tribe Rhizopho- reae (4 genera) and three inland tribes: Macari- sieae (6 genera), Gynotrocheae (4 genera), and Anisophylleae (4 genera) (e.g., Melchior, 1964) The comparative study of wood anatomy (van Vliet, 1976) and leaf architecture and anatomy (Keating & Randrianasolo, this volume; Baas, pers. comm.) has supported this broad definition of the family. In contrast, embryological evidence (Tobe & Ra- ven, 1983), as well as an overall comparison based on various systematic characters (Dahlgren & Thorne, 1984), enar) suggested that the overall group was heterogeneous and indicated the need for further embryological studies of the constituent genera and tribes as an important key to their The study was supported by pa from the U.S. National Science Foundation to P. H. Bird, Barbara Browning, Lorence , K. S. Manilal, Gordon McPhe BSR-8518902. We are grateful to Ann Nair, Peter C. Hoch, Betsy R. Jackes. L. Liben Raven, most a Dorr, Jack Fisher, K. Gopinathan erson, Bruce W. Nelson, Juan V. Pancho, Ching-l Peng, N. Sasidharan, Benjamin C. Stone, Duncan W. Thomas, and Robert Wingfield for collecting the materials that were used in this s * Biological Laboratory, Yoshida College, Pis University, Kyoto 606, yigg * Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166, U.S. A. ANN. Missouni Bor. GARD. 75: 1319-1342. 1988. 1320 Annals of the Missouri Botanical Garden relationship. Earlier embryological studies were limited to 9 of 18 genera: Bruguiera, Ceriops, and Rhizophora of Rhizophoreae (Karsten, 1891; Cook, 1907; Carey, 1934; Mauritzon, 1939; Jun- cosa, 1982, 1984b); Cassipourea of Macarisieae (Juncosa, 1984a); Carallia and Gynotroches of Gynotrocheae (Karsten, 1891; Haberlandt, 1895; Mauritzon, 1939; Corner, 1976); phyllea, Combretocarpus, and Poga of Aniso- phylleae (Karsten, 1891; Hou, 1958; Vaughan, 1970; Geh & Keng, 1974). Most of these studies were concerned with relatively few embryological features. Thus we recently presented an overall embry- and Aniso- ological study of *Anisophylleae" (Tobe & Raven, 987b) and, with support of evidence from other sources, justified the separation of Anisophylle- aceae as a distinct family from the rest of Rhizopho- raceae, as proposed earlier by several authors (e.g., Ridley, 1922; Cronquist, 1981, 1983; Dahlgren, 1983). Subsequently, we presented the first com- prehensive embryological study on two inland gen- era, Cassipourea and Sterigmapetalum of Ma- carisieae (Tobe & Raven, 1987a), which were recently separated from Macarisieae as constitut- ing a new tribe Hypogyneae (Steyermark & Leis- ner, 1983), and we provided some discussion on infrafamilial relationships based on data available. Juncosa (pers. comm.) has surveyed the respective embryological features of the remaining tribes, Gynotrocheae and Rhizophoreae. Consequently, Rhizophoraceae are becoming one of the most well- known families with respect to their embryological characters. Previous embryological studies of Rhizophora- ceae have lacked comprehensive, comparative in- formation on the seed morphology and anatomy of the whole family. Concerning the seed morphology, Schimper (1893), Hou (1958, 1968), Floret (1974, 1976), and Tobe & Raven (19872) described some of the constituent genera and suggested that Rhi- zophoraceae are diverse in this feature. However, no overall studies, except for the preliminary dis- cussions in our previous paper (Tobe & Raven, 1987a), have been made. Carey (1934) provided fragmentary descrip- tions on the seed coat anatomy of Rhizophora and Ceriops of Rhizophoreae. Later, Corner (1976) described some details of the seed coat anatomy of Carallia, Gynotroches, and Pellacalyx (all Gynotrocheae); Tobe & Raven (19872) described those of Cassipourea and Sterigmapetalum of Ma- carisieae; and Juncosa studied those of Carallia and Gynotroches. These works contained impor- tant suggestions inducing further studies. For ex- ample, Corner (1976, 1: 161) defined the seed coat of both Gynotroches and Pellacalyx as “ex- with the exotegmen fibrous and, on the basis of seed coat anatomy, transferred the two genera into Legnotidaceae, a family first described by Endlicher (1840) as “Legnotidae” to comprise Cassipourea and Gynotroches. Tobe & Raven (1987a) showed that Cassipourea and Sterig- mapetalum (Macarisieae) have a thick and a thin testa, respectively, and are clearly distinct from each other based on seed coat anatomy. Juncosa » ,*5 otegmic, (pers. comm.) reports that the seed coat of Carallia borneensis differs from those of other inland genera in having no persistent tegmen and a vascularized testa. Thus, the works of Corner (1976), Tobe & Raven (19872), and Juncosa strongly suggested the utility of seed coat anatomy in considering relationships of genera in Rhizophoraceae and in- dicated that an overall study of seed coat anatomy was needed for further understanding of infrafamil- ial relationships. The utility of the seed morphology and anatomy in such considerations has already been demonstrated in studies on several unrelated families (e.g., Cruciferae— Vaughan & White- 1971; Polygalaceae— Verkerke, 1985), leading to a revision of conventional tribal classi- house, fications in each case. This paper presents the features of seed mor- phology and anatomy for the whole family Rhi- zophoraceae, which consists of 14 genera (exclud- ing the genera of Anisophylleaceae). These results, which have revealed a considerable degree of di- versity in these features, are then used together with other evidence to clarify infrafamilial phylo- genetic relationships. Dahlgren (this volume) has used these features extensively in searching for relatives, and this analysis is of fundamental im- portance for such comparisons. MATERIALS AND METHODS Twenty-one species representing all 14 genera of Rhizophoraceae were investigated. Collection data are provided in Table 1. For microscopic obser- vations mature and immature seeds were micro- tome sectioned following standard paraffin methods described elsewhere (Tobe & Raven, 1987b). Some hard specimens, like those of Ceriops, were embed- ded in glycol methacrylate, sectioned with glass knives, and stained with 0.1% Toluidine Blue (e.g., Figs. 28, 29). Scanning electron micrographs were also used in observing seeds of Gynotroches and Pellacalyx, and they were prepared following the standard method using a JEOL 255 instrument Volume 75, Number 4 1988 Tobe & Raven 1321 Rhizophoraceae Seed Morphology TABLE l. Studied taxa, collections, and materials. Asterisk (*) indicates that dry herbarium materials were investigated. Tribal positions of genera follow Melchior (1964) and Floret (1976). Taxa Collections and Materials Macarisieae Anopyxis klaineana (Pierre) Engl. Blepharistemma membranifolia (Miq.) Ding Hou Cassipourea gummiflua Tul. var. verticillata (N. E. Br.) J. Lewis C. guianensis Aubl. C. malosana (Bak.) Alston *Comiphyton gabonense (J.-J. Flo- ret Macarisia pyramidata Thou. Sterigmapetalum heterodoxum Steyermark & Liesner Gynotrocheae Carallia brachiata (Lour.) Merr. C. eugenioidea King. "Crossostylis biflora Forst. grandiflora Brongn. & Gris *C. multiflora Brongn. & Gris Gynotroches axillaris Bl. Pellacalyx lobbii (Hook. f.) Schimp. P. cf. saccardianus Scort. Rhizophoreae Bruguiera gymnorrhiza (L.) am. Ceriops tagal (Perr.) C. B. Rob. Kandelia candel (L.) Druce Rhizophora mangle L. R. stylosa Griff. Cameroon. D. Thomas 3464 (MO)— buds & fruits India. Quilon, Kerala, K. Manilal s.n. in 1984 (MO)— female buds India. Grichur District, Kerala, N. Sasidharan s.n. in 1986 (MO)—fruits Zimbabwe. Cultivated, National Botanic Garden, Harare, Th. Müller 3558 (SRGH); pica collection: Mt. Inyangani, Th. Müller 698 (SRGH)— buds and fru Brazil. Mui p Nelson 1324 (MO, NY)— buds & fruits Zimbabwe. Cultivated, National Botanic Garden, Harare, Th. Müller 3557 (SRGH); original collection: Chirinda Forest, Mt. Selinda, B. Gold- smith—buds & fruits Zaire. Mt. Homas, Irumu, Germain 5213 (BR)— fruits Madagascar. L. Dorr 4392 (MO) — fruits Venezuela. Sierra de San Luis, Falcón, R. Wingfield 13692 (MO) — female buds, 13696 (MO)— fruits apr Jourama Falls National Park, North Queensland, B. Jackes s.n. in 1983 (JCT) — buds & fruits Malasia. pius B. Stone 15114 (KLU)— buds & fruits — Islands. Tahaa, Mt. Purauti, H. St. John 17346 ae w Caledonia. G. McPherson 6331 (MO)— buds & fruit New ae pus River valley, ca. 12 km NE "alisa G. McPherson 7 (MO)—fr a Maxwell "Hill Perak, B. Stone 15397 (KLU)— uei & fruits Malaysia. Sarawak, P. Chai s.n. in 1986, no voucher — frui Malaysia. Maxwell Hill, Perak, B. Stone 15396 (KLU)— buds & fruits eror Maputo, Costa u Sol, J. de Koning & M. C. Groenaedyk 9243 (LMU)— buds & f — Tong hid Mahmud Sider s.n. in 1983, no vouch- s & fruits pao Dagbilao, Quezon, Luzon, Hernaez CA 29249 (CAHUP)— buds epublic of PE icem Tanshuei, Taipei Co., C. Peng 4504 (HAST)—bu ruits U.S.A. eo Fairchild boum Garden, Florida. H. Tobe s.n. in 1981, no voucher — buds & frui U.S.A. Cultivated, Fairchild Tropical Garden, Florida. FG 69-111. H. Tobe s.n. in 1981, no voucher —buds; 4. Bird s.n. in 1983, no voucher — fruits (Tobe & Raven, 1987b). Comparisons among gen- OBSERVATIONS era were made on the basis of mature seeds, and when materials were available, seed coat ontogeny The seeds of rhizophoraceous genera have either was investigated to understand the mature struc- — an aril or a wing, or they lack appendages. Arillate ture more completely. The terminology on seeds seeds occur in Blepharistemma, Cassipourea, and and seed coat anatomy follows Corner (1976) and Comiphyton (all in Macarisieae), which have loc- Schmid (1986); Schmid elaborated Corner's ter- — ulicidally (?) or septicidally dehiscent or indeshis- minology. cent capsular fruits (Floret, 1976), and in Cros- 1322 Annals of the Missouri Botanical Garden sostylis (Gynotrocheae), which has capsular fruits of unknown dehiscence mode. Winged seeds occur in Anopyxis, Macarisia, and Sterigmapetalum (all in Macarisieae), which have septicidally dehis- cent capsular fruits (Floret, 1976; contrary to Flo- ret, Arénes (1954) described fruit dehiscence in Macarisia as loculicidal, but we confirmed septi- cidal dehiscence in M. pyramidata). Nonappend- aged seeds occur in Carallia, Gynotroches, and Pellacalyx (all in Gynotrocheae), which have bac- cate fruits, and in Bruguiera, Ceriops, Kandelia, and Rhizophora (all in Rhizophoreae), which have indehiscent hard-walled fruits. Both the aril and the wing develop as an outgrowth of the exostome: compare the young arillate seed of Cassipourea malosana (Figs. 1, 2) with the young winged seed of Anopyxis klaineana (Figs. 3, 4). In the case of the aril, a raphal tissue, which continues from the outer integument, may also join the aril for- mation; however, a funicular or hilar tissue never joins there. Therefore, as discussed in a previous paper (Tobe € Raven, 19872), the aril and the wing are homologous to each other, and the his- togenetic origin of both structures are regarded substantially as exostomal. We have also confirmed that the seed coat anat- omy correlates with the external seed morphology type (ie., arillate, winged, and nonappendaged and, in addition, that within the nonappendaged — seed category the seeds of Gynotroches and Pel- lacalyx are distinct from those of Carallia, as already described by Corner (1976) to some de- gree. The details are documented below. ARILLATE SEEDS Blepharistemma. It has been previously un- certain whether Blepharistemma has an aril, a wing, or neither, because fruits and seeds of this genus are undescribed. We found for the first time that B. membranifolia, the only species of the genus, has a fleshy aril (Fig. 5), which covers nearly the upper half of the seed. The mature seed is ellipsoid, with a somewhat conspicuous raphe, and ie slightly depressed toward the lateral side; it .3-4.5 mm long and 2.4-2.5 mm thick, as MR from the raphe to antiraphe (R-A), and 2.2-2.3 mm thick from side to side (L-L) (see Tobe & Raven, 1987a, fig. 14, for directions of width measurement). The oldest seed coat available is 0.20-0.22 mm thick in total, comprising the testa 135-142 um thick and the tegmen about 50 um thick. When the embryo sac is mature, the outer and inner integuments are 3-4 cells and 4-5 cells thick, respectively. The outer integument increases its thickness in postfertilization stages and eventually differentiates into a l-cell-layered exotesta, a 3- 6-cell-layered mesotesta, and a 1-cell-layered en- dotesta (Fig. 6). There is no clear histological dif- ference between the mesotesta and the endotesta. Exotestal cells are radially enlarged, thick-walled, and tanniferous, forming a palisade (Fig. 6). Meso- and endotestal cells are much smaller than the exotestal cells, but also somewhat thick-walled and tanniferous. The endotestal cells often contain crystals. On the other hand, the inner integument develops into a l-cell-layered fibrous exotegmen and 2 or 3 underlying unspecialized cell layers, the latter of which apparently disintegrates further. (The seeds investigated in this study seem still somewhat immature, and we believe that the un- derlying cell layers below the exotegmen eventually disappear completely.) Of the constituent cell lay- ers, the exotesta is the most conspicuously devel- oped as a mechanical structure; therefore, the seed coat of Blepharistemma is exotestal. Cassipourea. Cassipourea is a large and vari- able genus (4 subgenera and 80 species (Alston, 1925; Airy Shaw, 1973)), which shows a range of variation in seed size and anatomical structure. We described some details of seed size and seed anat- omy earlier (Tobe & Raven, 19872), and therefore these are only briefly summarized here. Only data on the total thickness of the mature seed coat were added. The mature seed is ellipsoid (Cassipourea gui- anensis and C. gummiflua var. verticillata) to broad ellipsoid (C. malosana) and slightly de- aere to the raphe. In C. guianensis, the seed s 8.9-9.2 mm long and 2.6-2.8 mm thick (R- A) to 3.3-3.6 mm thick (L-L); in C. gummiflua var. verticillata it is 4.2-4.4 mm long and 1.6- 1.8 mm thick (R-A) to 2.2-2.4 mm thick (L-L); in C. malosana it is 5.0-5.2 mm long and 2.8- 3.1 mm thick (R-A) to 4.5-4.8 mm thick (L-L). The aril is fleshy and wholly covers the seed except on the chalazal and antiraphe side. In the three examined species, the mature seed coat is 0.16-0.22 mm thick in total and is com- posed of a 1-cell-layered exotesta, a 2—6-cell-lay- ered mesotesta, a l-cell-layered endotesta, and a 1-cell-layered exotegmen; all other cell layers of the tegmen are crushed and disappear, although both the outer and the inner integuments are mul- tiple cell-layered at the mature embryo sac stage. exotesta comprises enlarged cuboid, thick- walled cells; the meso- and endotesta are composed of much smaller cells, and endotestal cells may Volume 75, Number 4 Tobe & Raven 1323 1988 Rhizophoraceae Seed Morphology FIGURES 1- , 2. Scanning electron micrograph (SEM) and longitudinal Vies (LS) of a mature ovule of € /assipourea epu showing the early « evelopment of an aril.— . SEM and LS of a mature ovule of Anopyxis klaineana showing early development of a wing. All scales — 100 um. ar, pen w, wing. 1324 Annals of the Missouri Botanical Garden Volume 75, Number 4 1988 Tobe & Raven 1325 Rhizophoraceae Seed Morphology contain crystals; the exotegmen is composed of longitudinally elongate, thick-walled fibrous cells. cells are mostly tanniferous. In all three species, the exotesta is most conspicuously devel- oped as a mechanical structure, and therefore the seed coat of Cassipourea is exotestal. Comiph yton. The mature seed of the only species of the genus, C. gabonense, is narrowly ellipsoid (Fig. 7) and nearly circular in cross section; it is 7.0-7.5 mm long and 2.0-2.8 mm in diameter. The aril is restricted to a micropylar top, and its tissue is apparently irregularly folded (Fig. 7). Com- pared with those of the species examined of Cas- sipourea, the seed coat surface of Comiphyton is more undulated, although it is uncertain whether this difference distinguishes Comiphyton from all species of Cassipourea. he mature seed coat is 0.14-0.18 mm thick in total and is composed of a 1-cell-layered exo- testa, an 8- 10-cell-layered mesotesta, a 1-cell-lay- ered endotesta, and a 1-cell-layered exotegmen (Fig. 8). There is no clear difference between the mesotesta and the endotesta. The exotesta is com- posed of somewhat enlarged, thick-walled tan- niferous cells; the meso- and endotesta comprise much smaller cells, and they do not contain crys- tals; the exotegmen is composed of longitudinally very narrow, fibrous cells. No other cell layers of the tegmen persist. Considering the structure of the entire seed coat, only the exotesta is relatively well developed as a mechanical structure. There- fore, the seed coat of Comiphyton is exotestal. Crossostylis. The mature seeds of the three examined species, C. biflora, C. grandiflora (Fig. 9), and C. multiflora, are ellipsoid, bearing a raphe as a very narrow longitudinal ridge. The mature seed is 2.1- long and 1.4-1.6 mm in diameter in C. biflora; 3.6-3.8 mm long and 1.8- mm in diameter in C. grandiflora; and 1.8- 1.9 mm long and 1.0-1.2 mm in diameter in C. multiflora. The aril is membranous, and its tissue is irregularly folded. The aril spreads over the micropylar top and does not tightly cover the seed proper. The mature seed coat structure differs from species to species. In Crossostylis grandiflora, the mature seed coat is relatively thick— 0.19-0.20 mm in total—and is composed of a 1-cell-layered exotesta, a 2-3-cell-layered mesotesta, a 1-cell- laine endotesta, and a | -cell-layered exotegmen (Fig. . There is no histological difference be- tween us mesotesta an At the mature embryo sac stage, the outer and the inner integuments are 4-5 cells thick and 9-11 cells thick, respectively. As the seed matures, therefore, the inner integument completely disappears except for the outer epidermis—i.e., exotegmen—- while the outer integument remains nearly persistent. The exotesta comprises enlarged, extremely thick- walled, tanniferous cuboidal endotesta are formed by unspecialized smaller cells; and the exotegmen is composed of longitudinally elongate, thick-walled, fibrous cells. Since the ex- otesta is most conspicuous as a mechanical struc- ture, the seed coat of C. grandiflora is exotestal. he mature seed coat structures of Crossostylis biflora (Fig. 13) and C. multiflora (Figs. 11, 12) are very similar. The total thickness is about 0.09 mm (C. biflora) or 0.06-0.07 mm (C. multiflora). The mature seed coat is basically composed of a l -cell-layered exotesta, a 1-cell-layered mesotesta, a l-cell-layered endotesta, and a 1-cell-layered exotegmen; however, the mesotesta and even the endotesta may be lacking in C. multiflora (Figs. 11, 12). Although we did not confirm the thickness of the integuments, they seem to be multiplicative, as we saw more clearly in our sample of C. gran- diflora. 'The exotesta is covered with a thick cu- ticle; its cells are longitudinally elongate and con- tain tanninlike pigments. The meso- and endotestal cells are much smaller than those of the exotesta and contain crystals. The exotegmen is coraposed of extremely sclerotic, longitudinally elongate, fi- brous cells. Compared with that of C. grandiflora, the exotegmen of these species is much more con- spicuously developed as a mechanical structure. Therefore, the seed coat of C. biflora and C. mul- tiflora is exotestal-exotegmic. the endotesta. cells; the meso- and WINGED SEEDS Anopyxis. The species examined, A. klaineana, has a large mature seed; it comprises the seed proper and a membranous wing (Fig. 14). The seed proper is oblanceoloid but extremely depressed lat- erally, and it is 13.4-13.7 mm long and 5.2-5.7 mm thick (R-A) to 1.6-2.1 mm thick (L-L). The — FIGURES 5-8.— (L5) fi do seed coa As E arillate seed available of Blepharistemma membranifolia and longitudinal section . Mature arillate seed of Comiphyton gabonense and LS of its seed coat. Scales = Figs. 5, 7) ae E um (Figs. 6, 8). exts, exotesta; rts, meso- and endotesta; extg, exotegmen 1326 Annals of the Missouri Botanical Garden wing is always larger than the seed proper and is 23.9-25.0 mm long and 10.6-11.4 mm wide. The mature seed coat is thin and 0.09-0.11 mm thick; for the most part it is apparently com- posed largely of a l-cell-layered exotesta and a l -cell-layered exotegmen (Fig. 15). In the mature embryo sac stage, however, the outer and the inner integuments are 4—5 cells thick and 6-7 cells thick, respectively. As the seed matures, therefore, all integumentary cell layers except for the outer epi- dermis of both integuments seem to degenerate or collapse. The exotesta is composed of enlarged, thick-walled, tanniferous cells, and the exotegmen of longitudinally elongate, thick-walled, fibrous cells. oth the exotesta and exotegmen are conspicuous as mechanical layers, and therefore the seed coat of Anopyxis is exotestal-exotegmic. Macarisia. The mature seed of this genus, like that of Anopyxis, comprises a seed proper and a membranous wing (Fig. 16). Thes size of seed varies within the genus; the sp d, M. pyrami- data, is known to have the largest Seeds in the genus (Arenes, 1954), which, however, are much smaller than those of Anopyxis. In M. pyrami- data, the seed proper is ellipsoid but extremely depressed laterally as in other species of the genus; it is 3.5-3.9 mm long and 2.1-2.5 mm thick (R- A) to 0.4 mm thick (L-L). The wing is always larger than the seed proper and is 8.5-9.1 mm long and 3.6-4.2 mm wide. The mature seed coat of Macarisia pyramidata is 0.05-0.07 mm thick in total, and it comprises mainly a 1-cell-layered exotesta and a 1-cell.lay- ered exotegmen (Fig. 17). Crystalliferous (endo- examine the thickness of integuments because we lacked material. The exotesta is composed of en- larged, thick-walled, and tanniferous cells; the exotegmen comprises longitudinally elongate, thick- walled, fibrous cells. Based on M. pyramidata, the seed coat of Macarisia is exotestal-exotegmic. Sterigmapetalum. Although the seed structure of this genus (consisting of seven species according to Steyermark & Liesner, 1983) has not been emphasized as a systematic character, it agrees with those of Anopyxis and Macarisia in having a seed proper and a membranous wing on the micropylar top. The morphology and anatomy of the mature seed of S. heterodoxum were discussed by Tobe & Raven (1987). In this paper, only characteristic features of the species are briefly summarized, and data on the total thickness are dded. The seed proper is oblanceoloid and extremely depressed laterally; it is 5.0-6.0 mm long and 1.6- 2.0 mm thick (R-A) to 0.8-1.0 mm thick (L-L). The wing is 7.4-10.2 mm long and 3.9-5.0 mm wide. (For other species of the genus, Steyermark & Liesner (1983) described the seeds of S. obova- tum as oblong, plano-convex, 8 mm long and 3.5- mm wide. The mature seed coat is 0.08-0.10 mm thick in total and is composed only of a 1-cell-layered exotesta and a |-cell-layered exotegmen, although the outer and the inner integuments were originally 2-4 cells thick and 8-10 cells thick, respectively. The exotesta comprises enlarged, somewhat radi- ally elongate, thick-walled, tanniferous cells; the exotegmen is composed of longitudinally elongate, thick-walled, fibrous cells. The seed coat of Sterig- mapetalum is exotestal-exotegmic. NONAPPENDAGED SEEDS Gynotroches. The mature seeds of the only species of the genus, G. axillaris, are ellipsoid and small with an areolate surface (Fig. 18). They are — FIGURES 9-13.— 9, 10. Mature arillate seed of Crossostylis grandiflora and longitudinal section (LS) of its . LS of m seed coat.— 11, 12. o biflora. Scales = ra mm Bu cr, crysta FIGURES 14-17.— coat.— 16, 17. Mature pae seed of 16) and 50 um (Figs. 15, 17) LS and transverse section of mature seed coat of C. multiflora. — 13 (Fig. 9) and 50 um (Figs. 1 mature seed coat 0-13). exts, exotesta; rts, meso- and endotesta; extg, 14, 15. Mature winged seed of Anopyxis klaineana m longitudinal section (LS) of its Vi acarisia pyramidata and LS o Scales . exts, exotesta; extg, exotegmen; cr, ie s seed coat. = ] mm (Figs. al. FicunES 18-21.— 18, 19. Scanning electron micrograph Pa of mature iios a Se seed of Gynotroches axillaris and longitudinal section (LS) of its seed coat cf. saccardianus and LS of its seed coat. Scales = —20, 2 of mature nonappendaged seed of Pellacalyx 200. um (Figs. A 20) and 50, um (Figs. 19, 21). exts, exotesta; ents, endotesta; extg, exolegmen; mtg, mesotegmen; entg, endotegmen. Volume 75, Number 4 Tobe & Raven 1327 1988 Rhizophoraceae Seed Morphology © "Yee 739 à *7' dà — DN "CAL IY 1 £ 1328 Annals of the Missouri Botanical Garden 1329 Tobe & Raven Volume 75, Number 4 1988 Rhizophoraceae Seed Morphology 1330 Annals of the Missouri Botanical Garden Volume 75, Number 4 988 Tobe & Raven 1331 Rhizophoraceae Seed Morphology about 1.5 mm long and 0.5 mm thick (R-A) to 0.3 mm thick (L-L). A raphe is relatively con- spicuous. The mature seed coat is 0.12-0.14 mm thick and comprises a thick testa and a thick tegmen (Fig. 19). At the mature embryo sac stage, the outer and the inner integuments are 2 cells thick and 6-8 cells thick, respectively; a nonmultipli- cative outer integument is characteristic of the enus. Even in the mature seed coat, all integu- mentary cell layers persist and form a 1-cell-lay- ered exotesta, a 1-cell-layered endotesta, a | -cell- layered exotegmen, a 4-6-cell-layered mesoteg- men, and a l-cell-layered endotegmen (Fig. 19). The exotesta is composed of remarkably enlarged, thick-walled, and tanniferous cells; the endotesta of much smaller and unspecialized cells; the exo- tegmen of radially elongate, thick-walled, fibrous cells; the underlying meso- and endotegmen of nonspecialized but somewhat thick-walled cells. Be- cause of the very conspicuous development of the exotesta, the seed coat of Gynotroches is exotestal. Corner (1976, 1: 161; 2: 260, fig. 315) re- ported that the mesotegmen (= “‘mesophyll’’) is eventually crushed, and interpreted the cell layers below the exotegmen as the “nucellus.” However, since the nucellus disintegrates earlier, even in ovular stages (Juncosa, unpubl.), Corner seems to have misunderstood the persistent meso- and en- dotegmen as the “‘nucellus Pellacalyx. The mature seeds of the genus are elliptic-oblong and small, often with an areolate surface. The shape and | size of the mature seeds differ somewhat f to species (Hou, 1958). The mature seed of P. lobbii and P. cf. saccar- dianus (Fig. 20) that tigated in this study are both ellipsoid. Their size is 1.7-1.8 mm long and 0.7-1.1 mm in diameter in P. lobbii and 1.2- 1.4 mm long and 0.7-0.8 mm in diameter in P. cf. saccardianus. The mature seed coat of Pellacalyx lobbii and P. cf. saccardianus is nearly the same, 0.26-0.2 mm thick in total. The seed coat, like that of Gynotroches, is composed of a thick testa and a thick tegmen (Fig. 21). The outer and the inner integument are 2 cells thick and 6-8 cells thick, respectively; all cell layers persist up to the mature seed coat stage, although the tegmen may further increase its thickness. Thus the mature seed coat comprises a l-cell-layered exotesta, a l-cell-lay- ered endotesta, a |-cell-layered exotegmen, a 6- 9-cell-layered mesotegmen, and a l-cell-layered endotegmen (Fig. 21). Exotestal cells are enlarged, thick-walled, and tanniferous; endotestal cells are much smaller and not specialized; cells of the exo- tegmen are longitudinally elongate, thick-walled, and fibrous; and those of the underlying meso- and endotegmen somewhat thick-walled. An undulation r "ribbon-like" structure (Corner, 1976) of the endotesta and the exotegmen is characteristic of the genus. As in the case of Gynotroches, Corner (1976, 1: 161, 2: 261, fig. 316) erroneously de- scribed the ‘‘nucellus” as persistent. But the nu- cellus that he considered is evidently the persistent meso- and endotegmen, because the nucellus com- pletely disappears at a much earlier stage (Juncosa, unpubl.). Carallia. The shape of the mature seed differs from species to species: oblong, oblong-ellipsoid, oblong-ovoid, ovoid, or reniform; the seed surface is areolate or corrugate; the size varies between 3 1.5 mm and 11 x 4 mm (Hou, 1958). The mature seeds examined of C. brachiata and C. calophylloidea are reniform, with the micropylar end close to the chalazal end and curved linear embryos (Fig. 22). The size is 5.5-6.0 mm long and 5.0-5.2 mm in diameter (when measured along the longest direction and the thickest middle part of the seed body). The mature seed of C. euge- nioidea, another species investigated, is oblong- ovoid, and, probably like those of most other species, has a straight linear embryo; it is 3.0 mm long and 1.5 mm in diameter. The mature seed coat of Carallia brachiata is 0.33-0.40 mm thick and is apparently composed only of a thick testa with exotestal cells extremely enlarged, thick-walled, and tanniferous (Fig. 24). Unlike the seed coat of C. no tegmen, as Corner (1976) indicated. At the mature embryo sac stage, however, the outer and the inner integuments are 3-4 cells thick and 6- 7 cells thick, respectively. Later, the young seed coat is evidently composed of a thicker testa and a thicker tegmen, where the differentiation of a eugenioidea, there is — FIGURES 22-25.— 22. Longitudinal be section of mature nonappendaged seed of Carallia brachiata. — 23. Longitudinal sec e of young seed co C. brachiata.—25. TS of mature seed coa 23-25 mesotegmen; entg, endotegmen e df C. brachiata. t of C. eugenioidea, with persistent neon Scales = 1 mm ). em, embryo; exts, exotesta; rts, meso- and endotesta; extg, exotegmen; mtg, mature seed coat of (Fig. 22) — 24. Transverse section (TS) of 1332 Annals of the Missouri Botanical Garden de R udi Bini . xç Pee wore A kÇ er Ya "4 (eos jt hi x . 4 4 E 7 e a t k: " PX XL P. d an y . r S € nas e T Me ae a. mer. hr ` d LI Sl zr 1L p " - É WAT NT 1 T. FIGURES 20-29. Ceriops tagal.—26. Mature seed covered with overflowing endosperm.—27. Longitudinal hand section of mature seed showing thick seed coat and endosperm.—28. Longitudinal section (LS) ofa young ovule showing that both inner and outer integuments are multiplicative. —29. LS of a mature ovule with an Volume 75, Number 4 Tobe & Raven 1333 1988 Rhizophoraceae Seed Morphology TABLE 2. Comparison in ovule, seed and fruit morphology, and putative methods of seed dispersal. Putative Ovules Methods of per Seeds per eec Tribe /Genus' Carpe!’ Fruit? Seed Size Seed Form Fruit Dispersal Tribe Macarisieae Blepharistemma 2 6 medium arillate capsular ants (birds?) Cassipourea 2 (4-)6(-8) | medium arillate capsular ants (birds?) Comiphyton 2 l medium arillate capsular ants (birds?) Anopyxis 2 10 mediur inged capsular wind Macarisia 2 10 medium winged capsular wind Sterigmapetalum 2 10-12 medium winged capsular wind Tribe Gynotrocheae Crossost ylis 2 several- small- arillate capsular ants and/or many medium birds Gynotroches many many nonappendaged baccate birds or mammals Pellacalyx many many small nonappendaged baccate birds or mammals Carallia 2 1(-5) small- nonappendaged baccate birds or medi mammals Tribe Rhizophoreae Bruguiera 2 l large nonappendaged indehiscent, sea water ard-walled Ceriops 2 1 large nonappendaged indehiscent, sea water walled Kandelia 2 l large nonappendaged indehiscent, sea water hard-walled Rhizophora 2 1 large nonappendaged indehiscent, sea water walled ! Tribal positions of genera follow Melchior (1964) and Floret (1976). 2 Data from Alston (1925), Arénes (1954), Floret (1974, 1976), Geh & Keng (1974), Hou EM 1968), Sprague & Boodle (1909), Steyermark & Liesner (1983), Tomlinson et al. (1979), and our observations 1-cell-layered exotegmen apparently occurs (Fig. 23). The structure of the exotegmen in the young seed coat in C. brachiata looks the same as those of other genera that have the persistent exotegmen. In C. brachiata the whole tegmen seems to de- generate as the seed develops. The meso- and en- dotesta, 10-20 cells thick in total, comprise much smaller cells than those of the exotesta. The mature seed coat of Carallia eugenioidea, like that of C. brachiata, has a thick exotesta that comprises extremely enlarged, thick-walled, tan- niferous cells (Fig. 25); the total thickness of the seed coat is 0.21-0.39 mm. The meso- and en- dotesta are histologically very similar and about 4-8 cell layers thick in total. Their cells may contain crystals. Carallia eugenioidea, unlike C. brachiata but like Gynotroches and Pellacalyx, has a persistent tegmen, which comprises a 1-cell- layered fibrous exotegmen and a 4—5-cell-layered underlying mesotesta. The endotegmen appears to have nearly collapsed at maturity. At the mature embryo sac stage, the outer and the inner integ- ument are only 3 cells thick and 5-6 cells thick, respectively. Therefore, even in postfertilization stages, nearly all cell layers of the inner integument appear to remain uncrushed. Both Carallia brachiata and C. eugenioidea have no vascular bundles in the integuments or seed coats (except where a raphal vascular bundle is continued from a funicle). Because of the very e ' Sais ed embryo sac. The inner integument is being broken by an enlarging embryo sac. Se end, endosperm; il, inner integument; oi, outer i tel din , 50 um (Fig. 28), = 100 um (Fig. 29). es, gei in sac; nuc, nucellu n (Figs. ales = 1m 1334 Annals of the Missouri Botanical Garden TABLE 3. Comparison in e anatomy. Abbreviations: oi, outer integument; ii, inner it; exts, exotesta; mts, mesotesta; ents, endotesta; extg, exotegmen; mtg, mesotegmen; entg, PURUS p testa. Seed coat categories follow Schmid (1986) š Thickness of Mature Seed Coat Integuments? Total (Cell Layers) Thickness Tribe /Genus' (Seed Form) oi ii (mm) Composition Tribe Macarisieae Blepharistemma (arillate) 3-4 4-5 0.20-0.22 exts + mts + ents + extg (+ mtg?) Cassipourea (arillate) 3-6 5-8 0.16-0.22 exts + mts + ents + extg pieni. iy (arillate) ? ? 0.14-0.18 exts + mts + ents + extg Anop (winged) 4-5 6-7 0.09-0.11 exts (+ mts + ents) + extg Mac (winged) ? T 0.05-0.07 exts (+ ents) + extg ida (winged) 2-4 8-10 0.08-0.10 exts + extg Tribe Gynotrocheae Crossost ylis (arillate) 4-5 9-11 0.19-0.20? exts + mts + ents + extg' 0.06-0.09* exts (+ mts + ents) + extg' Gynotroches (nonappendaged) 2 8 0.12-0.14 exts + ents + extg + mtg + entg Pellacalyx (nonappendaged) 2 6-8 0.26-0.28 exts + ents + extg + mtg + entg Carallia (nonappendaged) 3-4 5-7 0.21-0.40 exts + mts + ents (+ extg + mtg + entg) Tribe Rhizophoreae Bruguiera (nonappendaged) 14-19 5-9 0.79-0.85 ts Ceriops (nonappendaged) 8-16 4-5 0.65-1.30 ts Kandelia (nonappendaged) 11-16 3-5 0.70-0.90 ts Rhizophora (nonappendaged) 13-18* 4-6° 0.60-1.00 ts ' Tribal positions of genera follow Melchior (1964) and diode (1976). ? [n Cassipourea and Sterigmapetalum, inner an nteguments are both initially two cells thick and later increase their thickness (Tobe & Raven, 19872); this Bec uia nature of the integuments was further confirmed mma, Ceriops, and Kandelia in the prese in Blepharistem nt stu y and therefore is probably true of all other genera of Rhizophoraceae when they have thick integuments. Measurements of the thickness of the integuments were made on mature es with organized embryo sacs ultiflora * [n Carallia there appears to be a variation apre respect to the seed coat structure and vasculature. See text for details. * Data from Carey (1934: 393, fig. 3). conspicuous development of the exotesta as a me- chanical structure, the seed coat of Carallia is exotesta NONAPPENDAGED VIVIPAROUS SEEDS Bruguiera, Ceriops, Kandelia, and Rhizopho- ra. The mature seed taken from the indehiscent hard-walled fruit is cylindrical, 6.0-8.0 mm long, and 3.0-3.5 mm in diameter (C. tagal; Figs. 26, 27). The mature seed contains ample endosperm, as do those of the inland genera, but the endosperm of all the mangrove genera except Bruguiera char- acteristically extrudes from the micropyle and overflows the seed to fill up the space between the fruit wall and the seed (Haberlandt, 1895; Cook, 1907; ane 1934; Juncosa, 1982, 1984b). Such an endosperm structure is most conspicuous in Ceriops (Figs. 26, 27), least so in Kandelia. any diversity and is nearly t mangrove genera (Cook, 1907; e 1934; To unpubl.). The mature seed coat is 0.6-1.3 mm thick in total and comprises only a thick testa; it completely lacks the tegmen. Although the massive inner integument is present at ovular stages (Fig. 28), it completely disintegrates in pre- and post- fertilization stages (Fig. 29; see also Cook, 1907; Carey, 1934). The testa is 20-60 cells thick but it is apparently undifferentiated and merely com- Volume 75, Number 4 1988 Tobe & Raven 1335 Rhizophoraceae Seed Morphology TABLE 3. Continued. Endosperm Overflow from Embryo Sac (+) or Not (—) Mature Seed Coat Seed Coat Vasculature Present (+) or Absent (—) Category exotestal — — exotestal-exotegmic = = exotestal-exotegmic > = exotestal-exotegmic = — exotestal* = — exotestal-exotegmic* exotestal = m exotestal — — exotestal =5 = undifferentiated e€ = = = % O ®© c = + + + + + + + + undifferentiated prises many tanniferous or nontanniferous cells and vascular tissues. According to Carey (1934), in Rhizophora mucronata “the outer integument [= testa] is differentiated into two zones, an outer which contains the extensive vascular supply, and an inner in which ue cells are regularly arranged and meristematic,’ ' but in Ceriops candolleana the vascular strands, which are profusely branched in the testa, are distributed in the middle part of the testa in all the species examined, but they certainly do not always demarcate the inner zone. Since the seed coat anatomy of mangrove genera is not directly comparable with those of inlan genera, it can be designated for our discussion as an “undifferentiated” seed coat. DISCUSSION As described above, the seeds of Rhizophoraceae show certain fundamental differences in their mor- phology and anatomy. These features are su marized in Tables 2 and 3. In Table 2, the i ed of seed dispersal are also indicated as a basis for subsequent discussions. Despite the diversity of these characters, there are several major character co- incidences in certain groups of genera. All Ma- carisieae with a superior ovary (i.e., unspecialized ovary position), and Crossostylis (Gynotrocheae), with an inferior ovary (i.e., derived ovary position), always have either an aril or a wing on the seed; in contrast, the remaining genera, which have an inferior or semi-inferior ovary, lack any sort of seed appendage. Anatomically, all Macarisieae and Gynotrocheae (all inland groups) basically have a similar mature seed coat structure and consistently have a well-developed exotesta and (fibrous) exo- tegmen; their seed coat is either exotestal or exo- testal-exotegmic. In contrast, all Rhizophoreae (the mangrove group) do not have a histologically dif- ferentiated seed coat, and they entirely lack a tegmen. AFFINITIES OF RHIZOPHORAEAE The accumulating data strongly confirm that Rhizophoraceae are monophyletic. In particular, the combination of subdermally initiated laticifers* in the gynoecial wall (and sometimes even in other floral parts) with colletors? is unique to Rhizopho- raceae (see also Juncosa & Tomlinson, this volume b). Dahlgren (this volume), on the basis of a cladistic approach incorporating various vegetative, repro- ductive (including embryological), and chemical characters at Rhizophoraceae are closely related to r les of Malvales and Celas- traceae of Celastrales and possibly with Erythrox- tl Dahlgren selected as plesiomorphies of Rhizopho- * Juncosa € Tomlinson (1987) reported the i w wa of the subdermally EEN Dpi in ani oe era; Floret (1974: 502, pl. 2, figs. a-c) red this pa bolas cell layer in the ovary w hr “of Chiniphytan gabonense, although he did not specify it in = text. We confirmed the presence of laticifers in the ovary wall or calyx wall of Gynotroches ie) yen Blepharis- temma, Ca assipourea, and Sterigmapetalum (Macari- sieae) but not in Pellacalyx. Material of Macarisia was not available to check this character. 5 The occurrence of colleters has been recorded in a (Alston, 1925), Rhizophora (Gill & Tom- 1969; Lersten & Curtis, 1974), Bruguiera and Kanal (Tomlinson et al., 1979; Metcalfe & Chalk, 1950, 1: 602; Hou, 1958). We confirmed the presence of s ass. in all other genera available: Carallia (Gyn- otrocheae), Blepharistemma, Macarisia, and Sterig- mapetalum (Macarisieae). 1336 Annals of the Missouri Botanical Garden raceae the presence of an endothelium, the presence of an aril, the exotegmic seed coat, a chlorophyllous embryo, a Pvc-type sieve-element plastid, the pres- ence of certain types of alkaloids, and the occur- rence of a combination of some embryological fea- tures. We add the possession of deeply incised etals, because such petals are also undoubtedly plesiomorphic in Rhizophoraceae and Elaeocar- paceae. In addition, the occurrence of multipli- cative inner and outer integuments may be more strongly emphasized as another symplesiomorphy shared by all of the families mentioned above. From the viewpoint of seed morphology and Cassipourea, and Comiph yton of Macarisieae best agree with Elaeo- anatomy, Blepharistemma, carpaceae and Celastraceae in having arillate seeds and a fibrous exotegmen (and albuminous seeds and linear embryos; Weibel, 1968; Corner, 1976). he only conspicuous difference between these three genera of Macarisieae and Elaeocarpaceae- Celas- traceae is the absence of a persistent meso- and/ or endotegmen in the latter. In Rhizophoraceae, a persistent meso- and endotegmen occurs only in Gynotroches, Pellacalyx, and certain species of Carallia, all of which are undoubtedly specialized genera in the family as discussed later. Seed mor- phology and anatomy have not been investigated fully in Elaeocarpaceae and Celastraceae, and therefore it cannot be determined with certainty that this difference will hold up when more infor- mation is available. Critical, however, is the fact that the arillate seeds occur in the genera of Ma- carisieae that have a superior ovary (a plesiomor- hy), and that a fibrous exotegmen is common to all inland genera, which are less advanced in gen- eral than the mangrove genera. These features support Dahlgren's suggestion that the presence of an aril and a fibrous exotegmen is plesiomorphic in Rhizophoraceae and justify the use of those seed features in searching for related families. Comparisons with Anisophylleaceae may also be needed. Even though embryological evidence as well as various other lines of evidence suggest that this family is distinct (Tobe & Raven, 1987b), wood anatomy (Keating & Randrianasolo, this volume; Baas, pers. comm.), and floral morphology (Tobe Baas, pers. comm.), and flora morphology (Tobe & Raven, in press) link it with Rhizophoraceae. bs ign leaf anatomy partienlarly suggest that y be closely related: Our ends however, indicate that Anisophylleaceae lack subdermally initiated latic- ifers and colleters, both of which are characteristic of Rhizophoraceae and support their interpretation as a closely linked monophyletic evolutionary unit not directly related to Anisophylleaceae. The pres- ent study further shows that, even apart from the presence or absence of an aril, the seed morphology and anatomy of Anisophylleaceae differ greatly from those of Rhizophoraceae in completely lacking a tegmen (see Tobe & Raven, 1987b, for data on Anisophylleaceae). Therefore, Anisophylleaceae seem clearly to be much more distantly related to Rhizophoraceae than to Elaeocarpaceae and Celas- traceae; Rhizophoraceae and Anisophylleaceae are evidently more distantly related than we have sug- gested elsewhere (Tobe & Raven, in press). Dahlgren (this volume) has suggested that Elaeo- carpaceae have probably been misplaced in Mal- vales and ought to be transferred near Celastraceae of Celastrales sensu Dahlgren, along with Rhizopho- raceae. In terms of seed coat anatomy, most Mal- vales other than Elaeocarpaceae (e.g., Bombaca- ceae, Malvaceae, Sterculiaceae, and Tiliaceae) are characterized by a palisadal structure of exoteg- men, which is unknown in Elaeocarpaceae (see Corner, 1976). Thus the seed coat anatomy also supports the exclusion of Elaeocarpaceae from Malvales and, as already discussed above, their close relationships with Rhizophoraceae and Cel- astraceae. EVOLUTION OF SEEDS IN RHIZOPHORACEAE As discussed above, the arillate seeds of Bleph- aristemma, Cassipourea, and Comiphyton (Ma- carisieae) and Crossostylis (Gynotrocheae) appar- ently represent an archaic, ancestral state in Rhizophoraceae. Judged from the distribution of seed characters in Rhizophoraceae (Tables 2, 3), some of the specialized seed types seem clearly to have evolved more than once. In Macarisieae, arillate seeds (in Blepharistem- ma, Cassipourea, and Comiphyton) and winged seeds (in Anopyxis, Macarisia, and Sterigma- petalum), both types borne in capsular fruits, are probably dispersed by ants and wind, respectively; arillate seeds might also be dispersed by birds (see Ridley, 1930; van der Pijl, 1969, for general dis- cussions of seed dispersal). What may have induced an evolutionary change from ant dispersal to wind dispersal is uncertain. Anatomically, in contrast with a relatively thick seed coat of arillate seeds (ca. 0.2 mm thick), the seed coat of winged seeds is thinner (less than 0.1 mm thick). The thinner seed coat might be adaptive in lightening the seeds for wind dispersal. Despite this innovation in seed dispersal, however, the area of distribution of the genera with winged seeds is at present restricted. Anopyxis is restricted to West Africa, Macarisia Volume 75, Number 4 1988 Tobe & Raven 1337 Rhizophoraceae Seed Morphology to Madagascar, and Sterigmapetalum to the Am- azon (see Juncosa & Tomlinson, this volume b, for a distribution map). In contrast, Cassipourea, which has arillate seeds, is widely distributed in Africa, India, and South America. Blepharistemma and Comiph yton, also with arillate seeds, are confined to the Kerela district of southwestern India and West Africa, respectively The seeds of Crossostylis, which traditionally has been assigned to Gynotrocheae, resemble those of Blepharistemma, Cassipourea, and Comiphy- ton in having an aril and in being produced in capsular fruits. The distribution of Crossostylis (comprising 13 species) at present is restricted to and scattered in Polynesian islands, considerably separated from that of Macarisieae; it partly over- laps with the distribution area of the rest of Gy- notrocheae and of Rhizophoreae. These facts may suggest that the arillate, small-medium-sized seeds of Crossostylis are now or formerly were dispersed by birds from one island to another. Ridley (1930: 423-424) discussed bird-dispersed arillate seeds. Van der Pijl (1969: 30) gives an example: in In- donesia, the fruit-pigeons (Carpophaga) eat nut- meg with its aril (““arillode””) and disseminate it outside the region. Variation in seed size and seed coat anatomy within the genus may reflect complex methods of seed dispersal. The nonappendaged seeds of Carallia, Gyno- troches, and Pellacalyx (Gynotrocheae), which are borne in baccate fruits, are much smaller than the arillate seeds of Macarisieae in general. The num- ber of seeds per fruit is either 1(-5) (Carallia) or ca. 20-40 (Gynotroches and Pellacalyx). Those seeds are very hard and were difficult to section with a microtome. Fruit and seed structure of these genera seem to indicate that the seeds are endo- zoochorous and dispersed by birds or mammals. Bird or bat dispersal seems to provide a likely explanation of the wide distribution of these genera throughout the islands of southeastern Asia, west to India and Madagascar (Carallia). Seeds of the four mangrove genera— Bruguiera, Ceriops, Kandelia, and Rhizophora of Rhizopho- reae— produce viviparous seedlings in indehiscent hard-walled fruits. They do not exhibit histological differentiation in their seed coats, probably because the seeds are protected by the fruit wall and there- fore are not specialized against external environ- ments. When they fall from the parent plant, the dropped fruits or seedlings are undoubtedly dis- persed by sea water, as discussed by many workers (e.g., Ridley, 1930; Hou, 1958). The mangrove genera are at present best represented from the western Pacific to the Indian Ocean; Rhizophora is pantropical (Graham, 1964). Rhizophora, the most widely distributed genus of the family, is also the only one that is wind pollinated (Tomlinson et al., 1979), a system that might be well suited to newly colonized habitats. To sum up, the seeds of Rhizophoraceae seem to have evolved as the result of adaptation to changes in seed dispersal methods, that is, from ant dispersal to dispersal by wind, bird, mammal or water. The diversity in the seed morphology and anatomy, in conjunction with the diversity of fruit structure, is well explained by such changes in the methods of seed dispersal, and vice versa. However, the actual methods of dispersal of the seeds of the nonman- grove genera are very poorly known and should be studied in the field. PHYLOGENETIC RELATIONSHIPS WITHIN RHIZOPHORACEAE The accepted infrafamilial classification of Rhi- zophoraceae is based primarily on androecial po- sition (i.e., perigynous or epigynous), the number of carpels or ovarian locules, and the number of ovules per carpel (e.g., Melchior, 1964). Data from seed morphology and anatomy and fruit structure generally support the traditional classification in separating Macarisieae, Gynotrocheae, and Rhi- zophoraceae as distinct units. For the cladistic analysis, we chose 16 char- acters whose character-state evaluations were pos- sible; these include characters of embryology, seed morphology and anatomy, and floral morphology (Table 4). The character-state evaluation in Rhi- zophoraceae was made on the basis of outgroup comparison with Elaeocarpaceae and/or Celastra- ceae. Data on Rhizophoraceae, Elaeocarpaceae, and Celastraceae were obtained from the following references: Rhizophoraceae—Schimper (1893), Haberlandt (1895), Cook (1907), Carey (1934), Melchior (1964), Floret (1974, 1976), Corner (1976), Tomlinson et al. (1979), Juncosa (1982, 1984a, b), Tobe & Raven (1987a), and present study; Elaeocarpaceae— Mauritzon (1934), Ven- kata Rao (1953), Corner (1976), Cronquist (1981), and Hyland & Coode (1982); Celastraceae— Mauritzon (1936), Berkeley (1953), Adatia & Gavde (1962), Copeland (1966), Corner (1976), and Cronquist (1981). Results of character-state evaluation within the Rhizophoraceae are provided in Table 4, their distribution in the family in Table 5, and a cladogram based on these features in Fig. The cladogram (Fig. 30) indicates that, except for six genera of Macarisieae, the eight remaining 1338 Annals of th Missouri Botanical Garden TABLE 4. Character-state evaluation of some selected reproductive characters in Rhizophoraceae. Character Plesiomorphy' Apomorphy 1. Ovary position superior inferior or semi-inferior 2. Numbers of ovules per carpel 2 many 3. Structure of nuce crassinucellate tenuinucellate 4. Number of uod cells in ovule many l 5. Formation of endothelium occurs absent 6. Nature of outer integument multiplicative not multiplicative 7. Integumentary (or seed coat) vasculature absent resent 8. Development of endosperm not overflow from overflow embryo sac 9. Seedling not viviparous viviparous 10. Fruit morpholog capsular baccate or indehiscent 11. Type of seed i m aril win 12. Presence or absence of seed appendage present absent 13. Total thickness of mature seed coat > 0.1 mm < 0.1 mm 14. Histological differentiation of seed coat occurs does not occur 15. Development of exotegmen persistent, fibrous early disintegration 16. Development of meso- and endotegmen early disintegration persistent ' All plesiomorphies occur not only in some or all inland genera of Rhizophoraceae but also are common to the outgroups, e.g., Elaeocarpaceae and Celastraceae. References for data on each family are presented in the text. genera of the family share an inferior or semi- inferior ovary (a synapomorphy). The genera of Macarisieae retain many plesiomorphic features, including a superior ovary. Within Macarisieae, Floret (1976) recognized three subgroups on the basis of floral and seed characters: 1) Anopyxis, Macarisia, and Sterigmapetalum; 2) Blephari- stemma, Comiphyton, Plesiomorphy (—); ABLE 5. and Cassipourea subg. Dactylopetalum; and 3) the three remaining sub- genera of Cassipourea (see Tobe & Raven, 1987a, for revision). Later, emphasizing probable coinci- dences in the seed morphology and anatomy, but on the basis of meager data, we suggested closer affinities among Anopyxis, Macarisia, and Sterigmapetalum on the one hand, and amon Blepharistemma, Cassipourea, and Comiphyton Distribution of character states of some selected reproductive characters in Apomorphy (+). Character numbers correspond to those given in Table 4 Rhizophoraceae. Character Genus 1 2 3 x 10 11 12 13 14 15 Blepharistemma (BLE) = = = Cassipourea (CAS) — = — Carallia (CAR) Bruguiera (BRU) iops (CER) Kandelia (KAN Cer — Rhizophora (RHZ) + — = - | | | | + + + + + + + + ! Three-letter dS pet given in parentheses are used in a cladogram (Fig. 30). onal observations on C die os grandiflora. l 2 Data ips per ¿The n mber EN archesporial cells in an ovule e mangrove genera is not well documented; Karsten (1891) described t two megaspore mother cells in Ceriops NES This character must be confirmed in this and other mangrove gene Volume 75, Number 4 1988 Tobe & Raven 1339 Rhizophoraceae Seed Morphology on the other (Tobe & Raven, 1987 study confirms the relationships that we suggested then. Character-state comparisons indicate that Anopyxis, Macarisia, and Sterigmapetalum share the synapormorphies winged seeds and a thinner seed coat, and thus constitute a single clade. On the basis of seed morphology and anatomy, there- fore, we propose the division of Macarisieae into two subtribes: 1) Cassipourinae, Blepharistemma, Cassipourea, and Comiphyton; and 2) Macarisinae, comprising Anopyxis, Ma- carisia, and Sterigmapetalum. Wood anatomy (van Vliet, 1976) and leaf architecture (Keating Randrianasolo, this volume) do not provide a clear distinction between the two subtribes; wood anatomical comparison rather suggests that Comi- a). The present comprising phyton is intermediate between Macarisia (Ma- carisinae) and Cassipourea (Cassipourinae) (van Vliet, 1976). An overall comparison of their veg- etative morphology (Sprague & Boodle, 1909), however, suggests close affinities between Ano- pyxis and Macarisia. Leaf anatomy strongly sup- ports our suggested subtribal classification (Baas, Anopyxis- Macarisia(-Sterigma- petalum) have a nonpluriseriate epidermis and a differential hypodermis, ma-Cassipourea(-Comiphyton) have a plurise- riate epidermis but lack a discernible hypodermis. whereas Blepharistem- For further elucidation of the relationships in this group, a comprehensive study of the largest and most widely distributed genus, Cassipourea, seems essential. Crossostylis, which has been placed in Gyno- trocheae, shares only one apomorphy (an inferior ovary) with any group other than the six genera of Macarisieae. Apart from the ovary position, in contrast, Crossostylis agrees nearly completely wil Macarisieae in many plesiomorphic features. Jun- cosa & Tomlinson (this volume b) summarize that Crossostylis shares stilt roots, roots having no hairs, bijugate phyllotaxy, imbricate stipules, flat floral apices, and several-layered laticifers with other Gynotrocheae (sometimes excepting Pellacalyx), and shares most of these features also with Rhi- zophoreae. It is uncertain whether those shared structures represent apomorphic character states or not, but at least some of them are probably synapomorphies, suggesting phylogenetic affinities of Crossostylis with the rest of Gynotrocheae and all Rhizophoreae. Crossostylis, however, is clearly distinguished from other Gynotrocheae and all Rhizophoreae in not sharing the apomorphies noncapsular fruit and the nonappendaged seed. The cladistic analysis thus indicates mutual closer affinities between the rest BLE ANO CAS MAC COM STR CRS pet F E 30. A cladogram illustrating evolutionary E e +w of the gene h nera are shown in Tables 4 and 5, respectively. of Gynotrocheae and all Rhizophoreae, and leaves Crossostylis in a distinct evolutionary line. Juncosa (unpubl. suggests an intermediate position for Crossostylis between Macarisieae and Gynotro- cheae, and we agree with this interpretation. Cros- sostylis seems to be more appropriately assigned to its own tribe, Crossostylideae, as also suggested by Juncosa (pers. comm.). Carallia, Gynotroches, and Pellacalyx have been grouped in Gynotrocheae; this treatment seems reasonable because as a group they have no syn- apomorphies with Rhizophoreae. The three genera share two synapomorphies: a one-celled ovuled ar- chesporium and a persistent meso- and endotegmen (although the latter feature may be inconsistent in Carallia). Gynotroches and Pellacalyx closely re- semble each other in sharing the following addi- tional apomorphies: tenuinucellate ovule, and a nonmultiplicative outer integument; in contrast, Carallia is characterized by retaining the plesiomorphic states of those char- acters. Thus, we suggest that Gynotroches and Pellacalyx should be segregated as a subtribe Gy- notrochinae, and Carallia should be treated as the monogeneric subtribe Carallinae. In considering the phylogenetic relationships of Carallia, the diversity within the genus must be taken into account. For example, with respect to the mature seed coat structure, Corner (1976) described "tegmen without trace" in Carallia a multiovulate carpel, a 1340 Annals of the Missouri Botanical Garden brachiata; likewise, Juncosa (unpubl.) did not see a tegmen in C. borneensis. We did not observe a tegmen in the mature seed coat of C. brachiata, either (Fig. 24). In contrast, in the mature seed coat of C. eugenioidea (Fig. 25) and in the younger seed coat of C. brachiata (Fig. 23), we observed thick, distinct tegmic cell layers, the outermost of which even assumed the fibrous, exotegmenlike structure that is characteristic of all other inland genera. In its lack of a tegmen in the mature seed coat, Carallia might be compared to Rhizophoreae; however, when they are compared throughout their ontogeny, they appear to be quite distinct. Thus, in Rhizophoreae, the inner integument or young tegmen soon disintegrates because of the enlarge- ment of the embryo sac in postfertilization stages, but in Carallia it does not disintegrate until much later, and it may even persist, as in C. eugenioidea. Furthermore, in Carallia the outermost cell layer of the inner integument differentiates into the fi- brous exotegmen, but in Rhizophoreae such a his- tological differentiation has not been observed. e thickness and vasculature of the outer in- tegument in Carallia seems to be diverse. Ac- cording to Juncosa (unpubl.), C. borneensis has a 7-15-cell-layered outer integument at anthesis, which is vascularized as it is in Rhizophoreae. How- ever, in the samples of C. brachiata and C. eu- genioidea we observed, there were only 3-4-cell- layered outer integuments at the mature embryo sac stage; the outer integument of those two species has not space for vascularization at that stage, and no vascular bundles were observed in testae even in later stages, although the raphe always contained some vascular tissues. To sum up, we suggest that in Carallia, as well as in Gynotroches and Pellacalyx, the tegmen was originally persistent, but that a tegmenless seed coat (as in C. brachiata and C. borneensis) ap- parently evolved, almost certainly independently from Rhizophoreae. Likewise, the vasculature of the testa in Carallia seems to have been acquired only in certain species of the genus with an ex- tremely thickened outer integument or testa, but probably independently of the evolutionary line leading to Rhizophoreae. Although Carallia is diverse, it apparently should continue to be placed in Gynotrocheae, along with Gynotroches and Pallacalyx. he four mangrove genera— Bruguiera, Ceri- ops, Kandelia, and Rhizophora—share the fol- lowing synapomorphies: endothelium not formed, outer integument vascularized, endosperm over- flowing, viviparous seedlings, testa not differen- tiated histologically, and tegmen lacking. Thus, the mangrove genera undoubtedly form a coherent group, Rhizophoreae. Wood anatomy (van Vliet, 1976), leaf architecture (Keating & Randriana- solo, this volume), leaf anatomy (particularly sto- matal type; Baas, pers. comm.), and consistent distinct chromosome number (2n = 36; Yoshioka et al., 1984) also support the coherence of Rhi- zophoreae. The cladistic analysis indicates that Rhi- zophoreae have direct relationships with Gynotro- cheae (not including Crossostylis), rather than with Macarisieae. On the basis of our cladistic analysis, we suggest the following revised classification: Family Rhizophoraceae (not including Anisophy]l- eaceae Tribe Macarisieae Subtribe Cassipourinae Blepharistemma Cassipourea Comiphyton Subtribe Maca risinae Sterigmapetalum Tribe Crossostylideae ;rossostylis Tribe Gynotrocheae Subtribe Carallinae Carallia Subtribe Gynotrochinae Gynotroches Pellacalyx Tribe Rhizophoreae ruguiera Ceriops Kandelia Rhizophora LITERATURE CITED ADATIA, R. D. & S. G. GavpE. 1962. Embryology of the Celastraceae. In: Plant Embryology — A S the klowering Plants ‘and Ferns, 8th edition, Cam- bridge Univ. Press, Cambridge. ALSTON, A. H. G. 1925. Revision of the genus Cassi- pourea. Bull. Misc. Inform. (Kew Bull.) 1925: 241- 216. ARENES, J. . Fam. 150. Rhizophoracées. /n: H. — (editor), Flore de Madagasc car et des Co- mores. Typog. Firmin-Didot, Paris. a E. 1953. Morphological studies in the Cel- astraceae. J. Elisha Mitchell Sci. Soc. 69: 185 CAREY, G. 3 i bryology of i apri see South Wales 59: 3 : ,M. T. : The pe M of Rhizophora mange. Bull. Torrey Bot. Club 34: 271-277. c. Linn. Soc. New Cook Volume 75, Number 4 1988 Tobe & Raven Rhizophoraceae Seed Morphology 1341 y dece H. F. 1966. Morphology and embryology uonymus japonica. Phytomorphology 16: 326- du, E. J.H. 1976. The Seeds of Dicotyledons. 2 volumes. Cambridge Univ. Press, Cambridge. b dies IST, A. 1981. An Integrated System of Classi- n of Flowering Plants. Columbia Univ. Press, ew 1983. ew realignments in the dicotyledons. c J. Bot. 3: 75-83. s REN, R. 1983 General aspects of angiosperm evolution and macrosystematics. Nordic J. Bot. 3: -149. Gard. (this volume). R. F. THoRNE. 1984. The order Myrtales: circumscription, variation, and relationships. Ann. ouri Bot. Gard. 71: 633-699. 5. 1840. Genera Plantarum 974. Comiphyton, nais. Adansonia 14: 499- 500. 976. A propos de Comiphyton gabonense (Rhizophoraceae — Macarisieae). Adansonia 16: 3 49. . Vindobonae. genre nouveau Ga GEH, S. Y. & H. Kenc. 1974. Morphological studies on some inland Rhizophoraceae. Gard. Bull. Straits Settlem. 28: 183-220. GILL, A. M. & P. B. ToMLINSON. 1969. Studies on the growth of red mangrove (Rhizophora mangle L.) I. and Combretaceae i in the Southeastern United States. old Arbor. 45: 285-301. e pene G. 1895. Über die Ernagrung der bei: linge und die Bedeutung des Endospermes bei vivi- paren Mangrovepflanzen. Ann. Jard. Bot. Poona 12: 102-105 Hou, D. 1958. Rhizophoraceae. /n: C. G. G. J. van Steenis MN Flora V repone 9: i 493. ossostylis 1 lomon ny Cr e So and p New Hebrides rodar ades Blumea dus B. P. M. & M. J. E. Coone. 1982. n species for Australian genus Pe Pur ^ntadenia (Elaeo- carpaceae). Kew Bull. 3: 741-745 š Developmental morphology of the died and seedling of Rhi UN ee mangle L. ligi cra Amer. J. Bot. -1611. Embryogenesis M n devel- opment in Cassipourea elliptica (Sw.) P: (Rhi- zophoraceae). Amer. J. Bot. 71: 170-1 1984b. Embryogenesis and deena diosphalony of the seedling in Bruguiera pur Ding Hou (Rhizophoraceae). Amer. J. Bot. 71: 180- 191. & P. B. TOMLINSON. 1987. Floral development in mangrove Rhizophoraceae. Amer. J. Bot. 74: 1263-1279. A historical and taxonomic synopsis of Rhizophoraceae and Anisophylleaceae. Ann. Mis souri Bot. Gard. (this bec a). & atic comparison and some biological ds. of Rhizophoraceae and A isophylleaceae. Ann. Missouri Bot. Gard. (this s KARSTEN, G. 1891. Über die Mangrove-Vegetation im Rhizophoraceae and Anisophylleaceae: sum- mary statement, relationships. Ann. Missouri Bot. malayische Archipel. Eine morphologische-biologische Studie. Biblioth. Bot. (Stuttgart) 22: 11-18, 31-4 KEATING, R. C. & V. RANDRIANASOLO. Leaf architecture and relationships in the ur ag See and Ani phylleaceae. Ann. Missouri Bot. . (this volts) en N. R. € J. D. Curtis. 1974. " Colleter anatomy a mangrove, Rhizophora mangle (Rhizopho- . Canad. J. Bot. 52: 2277-2278. Maurarzon J. 1934. Zur Embryologie der Elaeocar- paceae. Arkiv Bot. 26: 1-8. 1936. Zur Embryologie und Yagua ie oe Abgrenzung der Dun arco aan und Celas trales. Bot. Not. 1936: 939. LI to p lb of the orders Rosales and Myrtales. Acta Univ. Lund. 35: T e 1-12 MELC HIOR, m Rhizophoraceae. /n: H. Melchior (editor), A. Engler's Syllabus der Pflanzenfamilien. II. z 357-359. METCALFE, C. R. & L. CHALK. 1950. Systematic Anat- omy of the Dicotyledons. 2 volumes. Clarendon Press, xford. RipLEv, H. N. 1922. The D of Malay Peninsula. Leeve & Co., Londor —— ———. 1930. The Teel e Plants Throughout the World. L. Leeve & Co., London SCHIMPER, A. F. W. o In: Engler & K. Prantl, Die Natürlichen aie is 6. On Cornerian and other pir id of anglospermous and gymnospermous seed c historical perspective E. terminological recommen- 6-4 . BoopLE. 1909. Kokoti (An- 2 ealaensis Sprague). Bull. Misc. Inform. (Kew Bull.) 1 sr t A. & R. LIESNER. 1983 [1984]. Re- of the genus Ste De id aac S rri Saal a. Ann. Missouri Bot. Gard. 70: 193. Tone, H. € P. H. RAVEN. An creo analysis of the Myrtales: its definition and char teristics. Ann. Missouri Bot. Gard. 70: 71-94 987a. The embryology a nd re- lationships of Cassipourea and Steri moren ca a ii Opera Bot. 92: 2 ` A —————. 1987b. Systematic embryology of ie w Anisophylleaceae. Ann. Missouri Bot. Gard. 74: ip ——. Floral a and evolution in Anisophylleaceae. Bot. n n. Soc. 97 (in press). ToMLINSON, P. B., R. B. PRIM & J. S. BuNr. 1979. Preliminary observations on id biology in man- grove Rhizophoraceae. Biotropica 11: 256-277. VAN DER dur 969. Principles of Dispersal in Higher Plan Longer Verlag, Berlin ae T G. 1970. The S and Utilization of Oil Seeds. eae le & Hall Ltd., London . M. WHITEHOUSE. 1971. Seed structure and the a! of the Cnieiferas, Bot. J. Linn. Soc. 64: VENKATA ides rx pee Floral anatomy and embryol- al ss ei of Elaeocarpus. J. Indian Bot. -33. ogy 9 Soc. 32 VERKERKE, . Ovules and seeds of the Poly- galaceae. J. Arnold Arbor. 66: 353-394. 1342 Annals of the Missouri Botanical Garden Vurr, G. J. C. M. van. 1976. Wood anatomy of the — Yosuroka, H., K. Konno, M. SEcAWwA, K. NEHIRA & S. M Biss locus. Leiden Bot. Ser. 3: 20-75 AEDA. 1984. Karyomorphological studies in five WEBEL, R. 1968. Morphologie de l'embryon " de la species of mangrove genera in m Rhizophoraceae. graine des Elaeocarpus. Candollea 23: 101-108. La Kromosomo 11-35-36: 1111-1116. THE CONTRIBUTION OF | LEAF ARCHITECTURE AND WOOD ANATOMY TO CLASSIFICATION OF THE RHIZOPHORACEAE AND ANISOPHYLLEACEAE! Richard C. Keating? and Voara Randrianasolo? ABSTRACT Leaf-architectural and wood-anatomical data on the 18 genera traditionally considered to comprise the Rhizophoraceae were analyzed for insight on the intergeneric affinities. The four genera of the tribe Anisophylleae, considered on other lines of evidence to comprise the family Anisophylleace eae, are not new data from leaf structure, or from a new look at the wo Wood d features, analyzed from the literature, do not readily distinguish the two families. Possible affinities of the Anisophylleaceae are not easily established using vegetative data. The Rhizophoraceae share a number of features with Celastraceae, Eleaocarpaceae, Hugoniaceae, Lepidobotryaceae, and Oxalidaceae. The Rhizophoraceae are a family of tropical- forest and mangrove trees variously considered as comprising either 14 or 18 genera, a disparity coming from inclusion of the genera Anisophyllea, Combretocarpus, Poga, and Polygonanthus as a tribe Anisophylleae vs. their exclusion as the family Anisophylleaceae. Tobe & Raven (1987a) dis- cussed the taxonomic history of the Anisophylle- aceae, while van Vliet (1976) and Juncosa & Tom- linson (this volume) provided a detailed taxonomic background on the Rhizophoraceae (sensu lato). However, to place this contribution in perspective, a brief review follows. mong modern systems, the genera of Aniso- phylleae have been included as a tribe in the Rhi- zophoraceae by Melchior (1964) and Takhtajan (1980), and by Thorne (1983) as a subfamily. Dahlgren (1983) and Cronquist (1983) recognized the separate family Ansiophylleaceae. In a review of the taxonomy and ecology of the Asian Rhi- zophoraceae, Hou (1958) included the genera of Anisophylleae. Prance et al. (1975) also followed this approach in a revision of Amazonian Rhizopho- raceae. Van Vliet (1976), in a detailed analysis of the wood anatomy, concluded that the Rhizophora- ceae comprise 18 genera arranged in four tribes (see Table 1), including the four genera assigned to the tribe Anisophylleae. Van Vliet (1976) con- cluded that wood-anatomical data provide no con- vincing case for excluding the Anisophylleae. This is no doubt partly due to the unusual amount of variation in the three tribes of the Rhizophoraceae s. str. In Table 1, we recognized a fifth tribe, Hypogyneae, which was separated from the tribe Macarisieae by Steyermark & Liesner (1983) on the basis of several morphological features (but cf. Juncosa and Tomlinson, this volume). Tobe & Ra- ven (1987a) found that floral, seed, and embryo- We are glad to acknowledge the loan of specimens from the National Cleared Leaf Hes, rubies pl pl Museum, Yale University. Peter Raven, Leo Hickey, and Hiroshi Tobe contributed v aluable discus pue was partially supported by the Office of Research & Projects, Southern lllinois University, Fdwardiellie, Illin š jm of Biological Sciences, Southern Illinois University, Edwardsville, Illinois 2 U.S.A. * Department d'Histoire Naturelle, Parc Botanique et Zoologique Tsimbazaza, B. P. 5 Madagascar. , Tananarive 101, ANN. MISSOURI Bor. GARD. 75: 1343-1368. 1988. 1344 Annals of the Missouri Botanical Garden TABLE l. Selected leaf characters of the genera of Rhizophoraceae and Anisophylleaceae.' l. 2. 3. 4. 5. 6. 1. 8. 9. 10. Mar- Vena- Highest yl- Stip- Mar- gin tion vein Are- Free Trache- lotaxy ules gin glands type Rank order oles endings oids Macarisieae Anopyxis O + E + Eb 3r 5-6 r 1-7 Blepharistemma O + E, T + Eb 3r0-3r3 6-7 P 0-5 — Comiphyton O + E + E 3r 6 P 0-2 == Macarisia O + E, T + E, Eb 2r0 6-7 P 0-3 F Cassipourea O + E, T (+) B,R 1r3-3r0 5-7 I 0-8 +, — Sterigmapetalum O + E + B 3r0-4r0 6 I 0-3 um Gynotrocheae Carallia O + E, T + B, E 1r2 5-7 I 0-6 F Crossostylis O + E, T + B, Eb 1r3-3r0 7 P 2-9 _ Gynotroches O + E ? E 3r0 7 I 0-9 -— Pellacalyx O + E 4 Eb 2r3 6 P 3-10 - Rhizophoreae Bruguiera O + E + B 2r0-3r0 5-7 I 0-8 C leri O + E ? B 3r2 5-6 I 0-4 ? Kandelia O + E B 3r0 6 I 0-6 C Rhizophora O + E > B 3r0 6-8 I 0-4 ? Anisophylleaceae Anisophyllea A — E + A,B 2r0-4rl 6-10 I 0-18 —,F Combretocarpus O - E = Eb 4r2 6 I 0-4 F Poga A -— E = Eb 2r0 6-10 I 3-20 C Polygonanthus A = E = Eb 4rl 5 I 0-6 = ' Symbols used: i Phyllotaxy: O = opposite, A = alternate; 2. Stipules: + = pre — = absent; 3. Margin: E = entire, T = toothed; 4. Marginal glands: + = present, (+) = uncommon or get l, — = not detected; 5. Vanatibn: A= alas B = brochidodromous, E = eel Eb = basally eucamptodromous and distally brochidodromous, R = reversed yasa at qud 6. Rank: e.g., 1r2 = first rank, second of three subranks (see Hickey, 1977); 7. Highest vein order present; eoles PP = polyg onal, I = irregular; 9. Number of Hie vein endings per areole; 10. Tracheoids (terminal futi db e = , — = absent, F = few common; 11. Marginal venation: F = fimbriate, I = incompletely looped, L = looped, Im = = intramarginal v n; 12. Sclereids (diffuse): — = absent, A = astrosclereids; 13. Adaxial epidermis surface cell shape: P = po ay S = sinuous, I = interlocking, * = some larger cells present with radiating neighbors; 14. Hypoderm: + = present, — = absent; 15. Abaxial epidermis surface cell shape: P = polygonal, S = sinuous, I = interlocking, * = surface papillate; 16. Stomata: B = aa ipi ks P — paracytic, C — td Ai = anisocytic , An = anomocytic, D = diacytic; 17. Fibers at veins: + = Bd esent; E Crystal type: D — druse, P — prismatic, S = crystal sand, — = pe Crystal distribution: R = ra , V = clustered ay veins, Mv = = along midvein . a equi , Ve = very common, qs = common, F = ; 19. Trichomes (simple): C = = few ay logical characters do not support segregation of in the Rhizophorales. Dahlgren (1983) placed the this tribe. Anisophylleaceae in the Corniflorae (Cornales) and The ordinal position of the 18 genera is even the Rhizophoraceae in the Myrtiflorae (Rhizopho- less agreed upon than the family composition. Mel- rales). Thorne (1983) placed his Rhizophoraceae chior (1964) and Takhtajan (1980) assigned the s. l. in the Corniflorae (Cornales). While a a Rhizophoraceae, including Anisophylleae, to the literature has developed on various aspects of m Myrtales. However, the consensus of the sympo- grove biology and ee (see Rollet, 1981), sium on Myrtales at Sydney in 1981 (see Dahlgren y relationships of the mangrove gen- & Thorne, 1984) was that the family should be era and their relatives remain much less studied. excluded from that order. Cronquist (1983) re- Leaf architecture of samples of all genera of tained the two families apart in his Rosidae with Rhizophoraceae and Anisophylleaceae is examined Anisophylleaceae in Rosales and Rhizophoraceae in this study to assess the contribution of foliar Volume 75, Number 4 Keating & Randrianasolo 1345 1988 Leaf Architecture & Wood Anatomy TaBLE l. Continued. n 12. 13. 14. 15. 16. 17. 18. 19. axial Abaxial Fibers Crystal Marginal Scle- epider- Hypo- epider- istri- re- venation reids mis derm i Stomata veins Type bution quency chomes LL = P = P 7 + P R, V C C L = P = S Ai, € + P R, V C F L = P = P An, B + P R A F L = P * qe ? + P R, V Ve —, Ç L.I _ P = P B? * P R, V F, A =, € LL = P Eod P B, Ai * P R, V CF = LL = pt = P B? + D R F zx L = P = P B, An + D R, V F, A =: LL = P = P B, An + D V A — L = P = P * D V F = LL A, — P - P C 4 D R, V A = L A? P = P C + D R A = L, F = P = P C + S R C = L A P + P C + D R, V C = I, L, Im = P, S, I — P,S,I An, D, P + D R, V C, A - LL ni P = P ? + — ? ? = I = P = P An, D, B + D R, V CF = L = P* == P P, An + = ? ? _ characters to several taxonomic problems. Histo- logical features detectable from cleared whole leaves are included in the observations. The leaf-archi- tecture scheme developed by Hickey (1973, 1979), and later applied by Doyle & Hickey (1976), Hick- ey & Wolfe (1975), and Hickey & Doyle (1977) has supplied a practical basis for gathering and interpreting these data in an evolutionary context. In addition, we have reviewed the wood-anatomical data from the comprehensive studies of Marco (1935) and van Vliet (1976). Using these data we will assess the intergeneric variability, the degree to which tribal groupings can be supported, and whether the exclusion of Anisophylleaceae from Rhizophoraceae can be sustained. These are the easier tasks. More complex is an assessment of the position of the Ansiophylleaceae among the angio- sperms. MATERIALS AND METHODS The leaf sample, including 53 specimens rep- resenting all 18 genera, was obtained from the herbaria of the Missouri Botanical Garden (MO), Paris (P), and some individual collectors. In ad- dition, descriptions were amplified after inspecting all Rhizophoraceae slides deposited in the National Cleared Leaf Collection currently deposited at the eabody Museum of Yale University. All speci- mens are cited following generic descriptions. All of the leaves were studied as safranin-stained clearings. The permanent specimens prepared for this study were cleared in 5% NaOH followed when necessary by 5.25% NaHCO, (commercial laundry bleach). Chloral hydrate was avoided, as it tends to destroy the differential stainability of venation. The clearings were dehydrated to 95% ethanol, stained in 0.5% safranin-O in 95% ethanol, de- hydrated to toluene, and mounted between glass plates in cover glass resin. Hickey’s (1979) de- scriptive protocol was followed. Leaf rank on a scale of 1—4 with three subdivisions within each rank, as presented by Hickey (1977 and pers. comm.), was scored for each specimen. The term "domain" is used to denote the area bounded by veins of a given rank, except that intercostal area is used instead of “secondary domain." The shape of intercostal areas and higher-order domains is a useful measure of the regularity of architectural organization (Hickey, pers. comm.). Except where descriptions from other literature 1346 Annals of the Missouri Botanical Garden are specifically noted, all of the descriptions given here and used for discussions and conclusions are restricted to the specimens cited. No claim is made that the sample observed encompasses all of the variation, particularly in the larger genera. OBSERVATIONS Anisophyllea R. Br. ex Sabine (Figs. 32-39) Leaves are chartaceous to membranaceous, en- tire, markedly asymmetrical in most species, ovate on one side, elliptic to obovate on the other. Those of A. pomifera are symmetrical and suborbiculate. Vincent & Tomlinson (1983) described the marked dimorphy in A. disticha. Venation is acrodromous, basal or suprabasal; the primary vein is slightly curved and is the same size as or larger than the acrodromous secondaries. Secondaries are perfect, extending more than 25 the distance toward the apex. Simple intersecondaries are present or ab- sent. The uniformly curved secondaries originate from the primary at a narrow or wide acute angle, the upper more obtuse or acute than the lower. Tertiaries are percurrent and forked or unforked and may be acute or perpendicular to the primary vein; angle of origin is acute or right, exmedially or admedially to the secondaries. Quaternary veins may be percurrent or not and are random orthog- onal to the tertiaries. Stomata appear to have 2 large paracytic guard cells surrounded by 8-10 smaller epidermal cells with radiating anticlinal walls. The margin in Á. meniandii has shallow convex glands that are not approached by any larger veins. Minor veins in the area often have tracheoidal endings but do not closely approach the gland. Other species have no detectable glands. Specimens examined. Anisophyllea apetala Kin SARAWAK. Clem ans & Clemans 21596 (M ). A. boehmii 2: A. cinnamo- Lanka, Meijer 709 (M disticha Tn aillon: Sumatra, Toroes 2610 (US; LJH 1856); Borneo, vii (M 20409 (MO). A. griffithii is , Anderson 4290 (MO). A. m i lac rin: Liberia, Jacques- Por 25941 (MO). A. poggei Engl. ex De Wild & T. Durand: ongo, Makany 1054 (MO). A. pomifera Engl., Malawi, Pawek 7414 (MO) Anopyxis (Pierre) Engl. (Figs. 1, 2) Leaves are symmetrical, oblanceolate, with an attenuate apex and cuneate base. The texture is membranaceous and the margin entire. The eu- camptodromous pattern tends toward brochidod- romy distally. The primary vein is moderately thick and straight. Secondary veins have a widely acute divergence, are uniform base to apex, and curve uniformly toward the margin. Long sinuous inter- secondaries are common. Tertiary veins are acute exmedially and right or obtuse admedially. They are often percurrent with intersecondaries and, dis- tally, with secondaries. Tertiary veins are generally oblique to the primary vein. Quaternary and quin- ternary veins produce orthogonal domains of ir- regular shape. The polygonal islets are often formed in interquinternary domains. Occasional marginal glands are vascularized by short straight veins that originate on a secondary arch. The gland is a shallow mound on the edge of the entire margin. cimen examined. Anopyxis calaensis Sprague: O Spec Nigeria, Kennedy 1561 (MO). Blepharistemma Wall. ex Benth. (Figs. 3-6) The leaf is symmetrical and elliptic, obovate, or lanceolate with an acute apex and an acute or obtuse base. Texture is chartaceous or membra- naceous. The margin is shallowly toothed in B. corymbosum or entire in B. membranifolia. Eu- camptodromous secondaries originate from the straight primary vein at a uniform wide acute or acute angle. In the distal half of B. membranifolia the secondaries are brochidodromous. Occasional simple intersecondaries are present and may be exmedially forked. Tertiary veins are sinuously per- current, often forked, forming polygonal domains with intersecondaries. Tertiary angles of origin are variable, but the tertiaries tend to be oblique to the primary vein. They originate mostly orthogonal to the secondaries but may be exmedially acute. Qua- ternaries and quinternaries form irregular orthog- onal domains. Teeth are shallow, strongly asym- metric, and curved inward to a small concave sinus. An asymmetrical turbinate gland having a super- ficial columnar epidermis protrudes from the apex parallel to the margin or incurved toward the sinus. A single vein the diameter of a secondary originates from a secondary arch and curves apically to the sinus, ending in the tooth apex, just below the gland. Just inside the margin, tertiary and quaternary veins branch off from the tooth vein and become part of the looped marginal venation. Specimens examined. Blepharistemma corymbos um Wall. & Benth.: India, Metz 713 (P). B. membran- ifolia (Miq.) Ding Hou: India (CAZ). Bruguiera Lam. (Fig. 27) Leaves are coriaceous or chartaceous, entire, symmetrical, elliptic or oblong, with an acute apex Volume 75, Number 4 1988 Keating & Randrianasolo 1347 Leaf Architecture & Wood Anatomy and base. Venation, not easily seen, is brochidod- romous with tertiary and quaternary arches or with a tendency toward a strong (sub)marginal second- ary collector vein. Secondaries diverge from the stout, straight primary vein at a wide-acute to acute-decurrent angle and meet the superadjacent secondaries at an obtuse or right angle. The upper secondaries may be more obtuse than the lower. Intersecondaries are simple or composite and are often admedially ramified or “perfect admedially.' Tertiary arches may enclose the secondaries and form an intramarginal vein. Tertiaries originate from either side of the secondaries at acute, right, or obtuse angles. Tertiaries may be randomly forked and may form large polygonal domains elongated parallel to the secondaries. Quaternary veins have a random polygonal relationship to the tertiaries. The margin has occasional shallow glands, ob- scurely vascularized in the existing preparations. Specimens examined. Bruguiera gymnorrhiza (L.) .: cultivated (FTG) [RCK 1443]; Ryukyu apes Kokuhara & Sunagawa 83 (US; LJH 1851). B. p viflora (Roxb.) Wright & Arn.: Australia, Blake 16995 (US; LJH 1852). Carallia Roxb. (Fig. 21) Leaves are chartaceous to membranaceous, symmetrical, entire or toothed, elliptic, obovate or ovate, mucronate or acute at the apex and acute to decurrent at the base. Venation is brochidod- romous or eucamptodromous basally and brochi- dodromous distally. Secondary veins diverge at a narrow or wide acute angle from the straight pri- mary, at a uniform angle, or more widely acute angle approaching the apex. Secondaries in broch- idodromous segments join the superadjacent sec- ondaries acutely and may be enclosed by tertiary and higher-order arches. Tertiaries may diverge at right angles or be acute exmedially and obtuse admedially. Tertiaries ramify randomly and may be retroflexed; the branches are sometimes aligned with the secondaries and intersecondaries. Inter- secondaries are basally simple, distally composite, or forked. In C. brachiata, intersecondaries evenly divide the intercostal areas and join the superad- jacent secondary high in the intercostal area. Teeth in C. fascicularis are acuminate, closely spaced, and are basally and distally concave. Veins origi- nate from eucamptodromous arches (tertiary and higher sized) and enter the teeth symmetrically. higher-order loops and veinlets ob- Carallia brachiata had no expressed teeth but regular, papillate, nonstaining glands. Vasculari- zation is often by an approaching marginal loop or several small anastomosing veins that often flare just below the gland. Hou (1958) reported the teeth to be dense in juvenile leaves and quite variable for the genus in general. Specimens examined. Carallia brachiata (Lour.) Merr.: ns China, Lau 488 (MO); Indochina, Pierre 683 (A, MO). C. fascicularis Guillaumin: Cochinchina, Poilane 116 (US; LJH 1848). C. integerrima DC.: India, Saldana 15267 (US; LJH 1850). C. lucida Roxb.: Bur- ma, Gallatly 783 (US; LJH 1849 Cassipourea Aubl. (Figs. 13-16, 19, 20) The leaf is symmetrical and elliptic to wide ovate (to orbiculate in C. rotundifolia). Texture is char- taceous to coriaceous, and the margins are entire or shallowly toothed. The straight primary vein is attached to brochidodromous secondaries, which in turn pass off tertiary and quaternary loops. Sec- ondary arches attach to superadjacent secondaries at acute, right, or obtuse angles. In C. rotundifolia, the venation pattern is reverse eucamptodromous with retroflexed secondaries. Secondary veins have a wide-acute, uniform angle of origin from the primary vein. They form regular intercostal areas. Intersecondaries are simple, short, and forked, or long, then extending to the secondary arches. There may be several per intercostal area. Tertiary veins are obtuse or random admedial to the secondaries, and acute or random exmedially. Tertiaries are forked and not often percurrent. Tertiary domains are irregularly polygonal, with several extending across an intercostal area. Quaternary veins orig- inate randomly, forming irregular, well- or poorly efined domains. Teeth are very shallow with a slightly convex glandular surface that may have a tuft of simple trichomes. No sinus is distinguished distal to the apex. Teeth are vascularized by a single arching vein originating from a secondary arch. As the vein curves apically toward the tooth, it passes off tertiary and quaternary loops. No marginal glands were detected in C. ceylanica. pecimens examined. Cassipourea brittoniana Fawcett & Rendle: Jamaica, Award & Proctor 14422 (GH, US: LJH 1198). C. barteri (Hook. f.) N. E. Br. Cote d'Ivoire, Chevalier 19981 (P); C. cey dual I | Alston: Sri Lanka, Mueller et a 69042737 (MO), S Lanka, dedi d 70040204R (US: LJH 1854). C. «linteo Poir.: British Honduras, Schipp 254 (GH; 2) C. deeds Aublet: o Rico, eee 10575 (GH; LJH 1199); Brazil, Dahlgren & Sella 86 (US: LJH 1853). C. gummiflora Tul. var. gummiflora: Madagascar, Boi- ne p^ (1851) (P). C. nmm Alston: Peru, Croat 8 (MOX. Peru, Klug 2235 (GH: LJH 1301). C. ) Alston, ea Schlieben 3552 (GH; Lin T 1348 Annals of the Missouri Botanical Garden Ceriops Arn. (Figs. 28, 29) Leaves are coriaceous, entire, symmetrical, ob- long, with a rounded apex and an acute, decurrent base. Venation is obscure, brochidodromous. Sec- ondaries diverge from the straight primary mod- erately acutely, the upper ones somewhat more obtuse than the lower. Secondary arches join the superadjacent secondaries obtusely. Intersecond- aries are simple, equally dividing the intercostal areas. Tertiary veins originate at obtuse or right angles admedially and exmedially and may be ad- medially ramified. Quaternary veins branch from transverse tertiaries and run parallel to the sec- ondaries. Occasional convexities in the margins do not look glandular and are not obviously vascular- ized. p examined. Ceriops boiviniana Tul.: Madagascar, Gentry 11901 (MO: LJH 4512); Comoro Isl., D 2872 (MO). Combretocarpus Hook. f. (Fig. 40) Leaves are membranaceous, entire, symmetri- cal, elliptic, with an obtuse apex and acute base. Secondaries arising from the straight primary vein are brochidodromous distally and eucamptodro- mous basally. Secondaries arise at a uniform, nar- rowly acute angle and are uniformly concave. The brochidodromous secondaries join superadjacent secondaries at right angles. Simple intersecondaries are present. Tertiary veins are percurrent and arise from both sides of secondaries at right angles. Ter- tiaries are oblique to the midvein at a uniform angle. Quaternary veins are orthogonal to tertiaries. ooped venation is typical at the margin but a straight collector vein may be present. No teeth or AH glands were detected. cimen examined. Combretocarpus rotundatus ub ) Danser: North Borneo, van Niel 427 1 (MO). Comiphyton Floret (Figs. 7, 8) Leaves are symmetrical, narrowly oblong, with an acuminate apex and obtuse base. Texture is membranaceous, and margins are entire. Eucamp- todromous secondaries arise from a straight pri- mary vein at a moderately acute divergence angle with the upper veins more obtuse than the lower ones. The secondaries are uniformly curved with lower secondaries being longer and more strongly ascending. The few short intersecondaries are ex- medially forked. Tertiary veins are sinuously per- current and oriented at approximately right angles to the primary vein, especially the outer ones. They are acute exmedially and obtuse admedially. Qua- ternary and quinternary veins arise orthogonally and form irregular polygonal domains. No teeth are expressed, but shallow, marginal, domelike glands are present. Short, single veins arise from secondary or tertiary loops and end directly below a gland. The leaf venation drawings of Floret (1974) agree with my observations. Specimen examined. Comiphyton gabonensis J. J. Floret: Gabon, Le Testu 5918 (P). Crossostylis Forster & G. Forster (Fig. 23) Leaves are chartaceous, entire or toothed, sym- metrical, wide oblong to narrow obovate, with an acute or obtuse apex and an acute to subdecurrent base. Venation is brochidodromous or sometimes basally eucamptodromous. Secondaries arise with a narrow- or wide-acute divergence from the straight primary vein. Secondaries are uniformly curved, join the superadjacent Secondanes s at right angles, and are enclosed by tertiary and y arches. Tertiary veins are usually forked, oncinating at random angles. Quaternary and quinternary veins form irregularly polygonal domains and are often retroflexed. Teeth are shallowly crenate with a fine glandular apiculum. A single, conspicuous tooth vein originates at a secondary or higher-order arch, which becomes smaller in diameter distally. Smaller veins of the marginal reticulum attach decurrently to the tooth vein. Specimens examined. — Crossostylis biflora Forster: Samoa, Veupel 493 (US: LJH 2256); Tabi, Balgooy 1715 (MO); C. multiflora Brongn. & Gris.: New Cale- donia, McPherson 2344 (MO). Gynotroches Blume (Fig. 24) Leaves are chartaceous, entire, elliptic with an acuminate apex and an acute base. Venation is eucamptodromous with secondaries uniformly curved after diverging at a wide acute angle from the straight primary. Intersecondaries are adme- dially simple and exmedially composite. Tertiary veins are acute exmedially, obtuse admedially, and admedially and transversely ramified. Quaternaries and quinternaries are polygonal to irregular. Un- common marginal, glandlike, shallow protuber- ances are not consistently vascularized. An en- larged vein may end in the vicinity, but it may not originate as deeply as a secondary or tertiary arch. cimens examined. Gynotroches axillaris Blume: Spe Philippines, Wenzel 1323 (MO); MO 2227076 (MO). Kandelia Wight & Arn. (Figs. 30, 31) Leaves are chartaceous, entire, symmetrical, ob- ovate, with an obtuse apex and an acute base. Volume 75, Number 4 1988 Keating & Randrianasolo 1349 | Leaf Architecture & Wood Anatomy Venation is obscure, brochidodromous, with sec- ondaries arising at a moderately acute angle from the straight primary vein. Upper secondaries are more obtuse than lower secondaries. Secondary arches join the superadjacent secondaries at right or obtuse angles. Simple or composite intersecond- aries are present. Tertiary veins diverge from sec- ondaries at right angles from both sides and may be transversely ramified. No marginal glands were detected. Specimens examined. | Kandelia candel (L.) Druce: Taiwan, Murata & Nishimura 31214 (MO); Taiwan, Peng 6042 (MO). Macarisia Thouars Leaves are symmetrical or asymmetrical, elliptic to suborbiculate. The apex is emarginate, or round- ed, or acute or acuminate, and the base is acute to decurrent. The texture is chartaceous or cori- aceous and the margin entire or shallowly toothed. Eucamptodromous secondary veins diverge mod- erately acutely and uniformly from the straight primary vein. They are gradually apically curved and unbranched. Secondaries in the distal half of the leaf may be brochidodromous. The few inter- secondaries are short and ramify into tertiaries. Tertiary veins may be sinuously percurrent, and connections to secondaries are at right angles or variable. In M. lanceolata the tertiaries are ad- medially obtuse and exmedially acute. Tertiaries are oriented relatively uniformly at right angles to the primary vein. Quaternary veins are orthogonal to tertiaries, forming irregular polygonal domains. Higher-order veins are indistinguishable in a retic- ulum. Areole veinlets often anastomose to form smaller suspended islets within an areole. Teeth are shallow, and strongly asymmetric; they have rounded convex bases leading to shallow, rounded, concave sinuses above the glandular apex. The turbinate gland points apically and is surrounded by a tuft of simple trichomes. A single vein orig- inates from a secondary arch and curves toward the tooth apex where it flares slightly. It does not supply the sinus. Tertiary and smaller veins that form marginal loops or the submarginal reticulum merge decurrently or randomly with the tooth vein. Specimens examined. Macarisia lanceolata p lon: Madagascar, Capuron 11.339SF (M ramidata Thouars: Madagascar, Serv. Forestier 11R460 (MO); s sapaq Dorr 4495 (MO). Pellacalyx Korth. (Fig. 22) Leaves are chartaceous, entire, symmetrical, ob- long, with an acuminate apex and an obtuse base. Venation is eucamptodromous basally and looped brochidodromous distally. Uniformly curved sec- ondary veins diverge from the straight primary at a uniform, widely acute angle. Secondaries are enclosed by tertiary and higher-order arches. In- acutely, both admedially and exmedially. Quater- nary and quinternary veins are random. Depres- sions in the leaf margin contain round, flat-topped glands. A single strong vein arises from a tertiary eucamptodromous loop and ends just at the margin below each gland. cimen examined. — Pellacalyx frustulata Merr.: O). Speci Philippines, Wenzel s.n. (1915) (M Poga Pierre (Figs. 42, 43) Leaves are chartaceous or membranaceous, en- tire, symmetrical, elliptic, oblong or lorate, with an acuminate or emarginate apex and an acute base. Venation is brochidodromous with tertiary and higher-order arches, or eucamptodromous at the base becoming brochidodromous distally. The straight or curving primary vein, often distally forked, produces secondaries at a wide acute angle, with upper secondaries often more obtuse than the lower. Secondary arches join the superadjacent secondaries at acute angles. Intersecondaries are simple at the primary vein, becoming composite exmedially. Tertiary veins are sinuous and oblique- ly percurrent, joining secondaries or intersecond- aries. Tertiaries arise at acute angles admedially from secondaries and at right angles exmedially, or their course may be irregular. Quaternary veins are random, often retroflexed, forming incomplete domains. Quinternary veins may be retroflexed in any orientation. No marginal glands were detected. Specimens examined. Pogo oleosa Pierre: Camer oon, Zenker s.n., (1909) (US: LJH 1858); Nigeria, 2 waodo s.n. (1983) (FHI). Polygonanthus Ducke (Fig. 41) Leaves are membranaceous, entire, elliptic, with an acuminate apex and an asymmetric, acute base. serandaries arice at a wide acute E. from the straight primary vein. A few of the ascending secondaries connect brochidodromously. Some dominant secondaries show an incipient su- prabasal acrodromy. Simple intersecondaries are straight, percurrent, and oblique to the midvein at a constant angle. Tertiaries diverge at right angles from both sides of secondaries. Quaternary veins 1350 Annals of the Missouri Botanical Garden form regular polygonal domains, often elongated parallel to the secondaries. Quinternary veins arise randomly from lower-order veins. No marginal glands were detected. At the leaf tip, the midvein ends at the margin similarly to tooth vasculature in the Macarisieae. cimen examined. Polygonanthus amazonicus Spe Ducke: Brazil, daSilva 4486 (MO). Rhizophora L. (Figs. 25, 26) Leaves are coriaceous, entire, symmetrical, ob- long or elliptic with an obtuse apex and an acute base. Venation is brochidodromous, and the sec- ondary arches may form a strong composite col- lector vein. Secondaries diverge from the straight primary at a right or acute (decurrent) angle. The divergence angle is uniform, or the upper and lower secondaries may be more acute than the middle sets. Secondary arches join superadjacent second- aries at acute, right, or obtuse angles. Intersecond- aries are absent or simple and arising from primary veins or from secondary loops; they are obmedially ramified parallel to the secondaries. Tertiaries arise at a right angle from both sides of secondaries, are not percurrent, are often orthogonal with intersec- ondaries, and often are admedially ramified forming a random reticulum with no directional orientation to the primary vein. No marginal glands were de- tected. Specimens examined. Rhizophora mangle L.: Flor- ida, uni 5438 (US: LJH 1846); Hawaii, Degener & ‘ielbese 3363 (MO). R. mucronata Lam.: Caroline Isl., Anderson 1049 (US: LJH 1847). R. stylosa Griffith: Australia, eu 220-4 (MO). Sterigmapetalum Kuhlm. (Figs. 17, 18) The leaves are symmetrical, oblanceolate or el- liptic, with an acute or emarginate apex and a rounded to acute decurrent base. The leaves are entire-margined and chartaceous. Venation is looped brochidodromous, and secondaries have an acute to wide acute divergence from the straight primary vein. Secondaries are uniformly spaced with the arches fusing with the superadjacent secondaries at obtuse or right angles. Composite intersecond- aries may be present. Tertiary veins originate most- ly at right angles exmedially and admedially, and are oblique to the primary vein at a constant angle. Tertiaries are forked percurrent, producing irreg- ular but evenly sized domains. Quaternary veins are orthogonal to tertiaries. Quaternary and higher- order loops festoon the secondary arches. Veinlets in the areoles often anastomose to form small sus- pended islets. Rarely a flattened marginal gland is found that may be directly vascularized by a single vein arising from a tertiary loop. The illustration of leaf architecture by Steyermark & Liesner (1983) agrees with these observations. Specimens examined. | Sterigmapetalum guianense Steyerm. subsp. ichunense Steyerm Liesner: Vene- yerm. zuela, Liesner 7314 (MO); Venezuela, Clark 7275 (MO). DISCUSSION VENATION PATTERNS Brochidodromy and eucamptodromy appear to be quite closely related since intermediate forms are often found. In intermediate leaves, the distal secondary veins are always brochidodromous and the proximal secondary veins always eucamptod- romous. The ontogeny of lamina tissue in many dicot leaves begins with the blocking out of the distal portion of the lamina, while the proximal portion is produced later following intercalary growth of the midrib. In early developmental stages of Fuchsia and Hauya (Onagraceae; Keating, un- publ.), brochidodromous leaves appear to have more or less simultaneous lamina inception with unified marginal growth. The cross attachment of second- ary arches to the superadjacent secondaries occurs as the marginal expansion is slowing down. In eu- camptodromous leaves, the later basipetal incep- tion of secondaries on the midrib is followed by an acropetal marginal expansion of the lower lamina. This produces apically arching secondaries that eventually dissipate in submarginal loops parallel to the margin. Transitions between distal brochidodromy and basal eucamptodromy are commonly found in An- opyxis, Carallia, Crossostylis, and Pellacalyx of the Rhizophoraceae, as well as Combretocarpus, Poga, and Polygonanthus of the Anisophylle- aceae. The developmental transition from brochidod- romy to eucamptodromy suggests a phylogenetic recapitulation, but further, it is certainly clear that the two venation patterns are closely related with eucamptodromy always appearing later in devel- opment. The data of Hickey & Wolfe (1975) show that brochidodromy is stratigraphically earliest, and it tends to have the lowest rank order (organiza- tional complexity and regularity) of any type of angiosperm venation pattern. Rhizophoreae, having the most coriaceous leaves, are exclusively brochidodromous with well-devel- oped secondary arches meeting the superadjacent secondaries at an obtuse angle. These arches have tended to become straightened, forming a sub- marginal collector vein, a tendency shown by no Volume 75, Number 4 1988 Keating & Randrianasolo 1351 Leaf Architecture & Wood Anatomy other tribe. This implies a leaf ontogeny with little if any basal acropetal expansion. Instead, it can be hypothesized that the leaves have a uniform marginal expansion, followed by a more abrupt cessation of expansion and a more synchronized final marginal differentiation. Tertiary veins in the Rhizophoraceae generally do not show predictable patterns in most genera. They are generally forked, not percurrent, and have irregular domains. Branching from second- aries is acute to obtuse and seldom sufficiently well developed to be called orthogonal, or regularly polygonal. Areolation is most commonly imperfect, al. though it may be well developed in Blepharistem- ma, Comiphyton, and Macarisia of the Macari- Combretocarpus of the Anisophylleaceae. Free vein endings are variable within each tribe but generally fewer than seven. The highest number, to greater than 20 in An- isophyllea, is large for both families. The most distinctive pattern of secondary ar- chitecture is found in Anisophyllea, most species of which show basal and suprabasal acrodromy (Fig. 32). Some species (A. griffithii, Fig. 33, and a specimen of 4. pomifera) are not acrodromous but rather eucamptodromous with irregular spacing of the pinnate secondaries on the midvein. All spec- imens of the other three genera of Anisophylleaceae are eucamptodromous, often with distal brochido- dromy. Polygonanthus has some secondaries that show the tendency toward strong ascending cur- vature parallel to the margin as seen in Aniso- phyllea sieae, an in MARGINAL TEETH AND GLANDS As leaf teeth in the Rhizophoraceae are best developed in the Macarisieae, we will refer to them as the Macarisioid type. The teeth are best devel- oped and largest in Blepharistemma (Figs. 3-5), although conspicuous expression can also be found in the genera Macarisia (Figs. 10-12), Cassipou- rea (Figs. 13, 14), and in Carallia of the Gyno- trocheae. In Macarisieae, the teeth are markedly asymmetric, having a gently convex margin prox- imal (basal) to the tooth apex. The apex is incurved facing directly onto an abruptly or gradually con- cave sinus. Tooth vascularization is always by a single vein usually originating from an ascending secondary vein or tertiary loop. It may begin below the tooth and curve gradually toward the apex, entering the tooth symmetrically (Figs. 11, 12). In other cases, the vein originates directly opposite the tooth apex and follows a straight course toward the distal mar- gin of the tooth, supplying the sinus as well as the apex (Blepharistemma, Fig. 5). The tooth vein is usually more strongly developed than the neigh- boring veins, which often join it oriented randomly or decurrently. The tooth vein ends slightly flared at the margin, which is surmounted by a turbinate, rounded or flattened gland. On expressed teeth, glands are often incurved toward the sinus. They often show no stainability and often have a surface of columnar cells. No visible epithem appears with- in the lamina distal to the vein. On leaf margins with glands but no expressed teeth, the relationship between veins and glands is not basically different from those leaves with expressed teeth. Carallia fascicularis shows an unusual tooth for the family. It is concave on both sides, very frequent and conspicuous, and is vascularized by one vein with tertiary garland connections to the neighboring tooth veins. Lersten & Curtis (1974) reported on the struc- ture of colleters as found at the base of stipules in Rhizophora mangle. They are not regularly as- sociated with vascularization, and they appear from the evidence presented to have no obvious rela- tionship to glandular teeth in other genera of the family. Within the available data, there are no strong correlations between the degree of expression of glands and teeth with the habit or habitat. It can be noted in general that the four specialized man- grove genera have more coriaceous leaves than the inland genera. Givnish (1979) reviewed the statistical data showing a correlation between toothed (nonentire) leaves and a thin, noncoria- ceous lamina. He also noted that deciduous leaves are toothed more frequently than evergreen leaves. If the ancestors of Rhizophoraceae were thin-leaved and possibly from seasonal habitats, the nontoothed leaves are clearly derived. On the other hand, the best-developed teeth are found only on plants with eucamptodromous venation, while brochidodro- mous leaves have very reduced teeth/glands or no marginal features at all. Hickey & Wolfe (1975) presented stratigraphic evidence that the brochi- dodromous pattern is primitive in angiosperms. While it is tempting to regard teeth in Rhizopho- raceae as a reduction series, this cannot be taken as proven. EPIDERMIS AND STOMATA The paradermal outline of the anticlinal walls of epidermal cells is not particularly useful system- atically in these genera. In nearly all specimens, 1352 Annals of the Missouri Botanical Garden cells are polygonal (isodiametric or elongated up to 2:1 length/width ratios) and are the same on both surfaces. Blepharistemma has sinuous abax- ial epidermal cells, Macarisia and Poga have pap- illate abaxial epidermal surfaces. Anisophyllea species vary in cell outline from polygonal to un- dulate to interlocking shapes with the two surfaces matching or not. In some genera, occasional larger rounded cells occur that have more radiating neigh- bors than most of the smaller cells. Such large cells were found in Carallia and Pellacalyx of the Gyn- otrocheae, and in Anisophyllea and Polygonan- thus of the Anisophylleaceae. Observations on stomatal types are difficult in some cases, as they may be quite obscure in clear- ings. Nevertheless, a few conclusions seem valid. The Rhizophoreae are quite distinct with their ap- parently cyclocytic patterns (Figs. 27, 29, 31). The subsidiary cells appear narrow and differen- tiated from other epidermal cells. They vary in number from 5 to 8-10 cells surrounding the guard cells. Stomata are exclusively abaxial in all genera observed. While the brachyparacytic pattern is most common in all other genera of both families, anomocytic and anisocytic types were noted as well. IDIOBLASTIC CELLS The Macarisieae and Hypogyneae are distin- guished by having prismatic crystals with small druses rarely present (Blepharistemma). All other genera studied have only druses, no prismatics. Occasional specimens have no crystals. Druses in the Rhizophoreae are normally birefringent but have centers that appear dark or extinct under polarized light. This phenomenon was also noted in Pellacalyx, a genus otherwise clearly belonging to Gynotrocheae. In the Gynotrocheae, druses tend strongly to cluster along veins with only a few scattered ran- domly in the mesophyll. In other tribes, the dis- tribution of crystals varies among the genera. Crys- tals may cluster nonexclusively at the veins, and many are scattered randomly. In this respect, the Anisophylleaceae are unexceptional. Foliar sclereids of several types were noted in this study in both families. A series of studies by Rao & Bhattacharya (1978), Rao et al. (1978), and Rao & Das (1979) have also called attention to these cell types, particularly in the tribe Rhi- zophoreae. “Diffuse polymorphic sclereids” (Rao et al., 1978), here called astrosclereids, are found in Bruguiera, Ceriops, and Rhizophora (Figs. 25, 26) of the Rhizophoreae and in Pellacalyx of the Gynotrocheae. A second type, the terminal tra- cheoid, is characteristic of Macarisia, Cassipou- rea, Crossostylis, Bruguiera, and Kandelia (Fig. 30) of the Rhizophoraceae, as well as in Aniso- phyllea, Combretocarpus, and Poga (Fig. 43) of the Anisophylleaceae. Only in Rhizophoreae are astrosclereids and terminal tracheoids commonly found together. FAMILY AND TRIBAL DELIMITATION: RHIZOPHORACEAE A review of the available leaf-architectural and histological data (Table 2) demonstrates much over- lap in most features and an apparent relatedness among all of the genera. Yet a few characters readily distinguish the tribes. The Macarisieae and Hypogyneae have prismatic crystals in the leaves, and the Hypogyneae are further differentiated by having mostly looped brochidodromous leaf vena- tion instead of mostly eucamptodromous as found in the Macarisieae. In several genera of the Macarisieae (Anopyxis, Blepharistemma, Comiphyton, Macarisia), the leaf-rank organization is unusual: the tertiary veins and areolation show a high level of organization while the secondaries are less well organized. Nor- mally, rank order increases proceed in ascending order from secondaries through tertiaries to the organization of areoles (Hickey, 1977). Following current practice (Levin, 1986; Hickey, pers. comm.) we have ranked the leaves according to the highest level obtained so that the maximum evolutionary advancement is reflected in the clas- sification. In his study of the leaf architecture of Euphorbiaceae: Phyllanthoideae, Levin (1986) hy- pothesized that venation where secondaries are less well organized than tertiaries represents a regres- sion of rank order. He found this type among leaves from arid, arctic, and alpine habitats. What its significance is in the Macarisieae remains obscure, as these genera do not face such environmental stresses. The Gynotrocheae and Rhizophoreae are distin- guished from the Macarisieae by having druses in the leaves. The Rhizophoreae have coriaceous leaves with brochidodromous venation and a cyclocytic stomatal pattern unique in the family. Also, three of the four genera in this tribe have astrosclereids in the leaves. The Gynotrocheae have chartaceous leaves and brachyparacytic or other noncyclocytic stomata. In many respects the Rhizophoreae are the most unified and specialized group of genera due to their mangrove habit, vivipary, chromosome morphology (Yoshioka et al., 1984), bijugate phyl- lotaxy (Tomlinson & Wheat, 1979), and the leaf 1353 Keating & Randrianasolo Volume 75, Number 4 Leaf Architecture & Wood Anatomy 1988 azeue Ápsow *juslmoJad pex1oj JO SNONUIS ‘aynoe JO 14211 ejsoduroo Aye} -sip 01 apduns ‘Zuo] “110ys *uoululoo JO ma] 'euou Ut ‘042 repn39.1.1 Jo Je¡n3a1 IML IPIM 01 Zutpusose Á[Suo.]s pue yeseqeidns oj [eseq snouloJpoloe 'snoulo1poi -dureone *snourojpopru204q [2.19 -A8S 0] auo '[euriou “1y3te.11S snoo3ej1eqo əmuə pəpunoi 10 ƏMƏ 0] jual1noop ojeuid.1euro JO pepunoi 0j ojeurumoe [e9rnəururÁse JO [eərnəu -uiÁs *eje[notq1oqns “onda 9jno? Jo osnjqo Ájje -uotseooo tÁA][erpeurxo 29 K[[etpeurpe 14311 Ajrensn peytures ÁA[perpeurqo uejjo eisoduroo 10 ə|durts 61£-016 1e[n2o1 Ápsour ƏMƏL MOIIEU 0} ƏPIM snouloJ1poptq20.Iq posse] -uə 0} [BUIOU *1ySte.]s sno399B1I09 IO SNOIIB]IBUYO 219 U9 ƏMƏL 0] JUVIINIEp asn1q0 01 ƏMƏL 9]PAO -qo Áqpeuorseooo “onda pax.103 10 peururei1 tsanepuodas 0] peed 1o os1oAsueq] Á] -[Prpeurxo ƏM 01 14311 tKj[erpeurpe asniqo 0i 14311 po1oj uajjo eisoduroo 10 o[duris 01£-04T Ie[nga.u Jo re[n3ə: 3INIB MOJIIEU 01 IPIM snouro1popmq2o1q Á[[ei -sIp 01 snouro1poidureo -nə 'snouroJpoprq20q pa3.8] -uə 0} ewou *jugreujs sn093B]I18yo 311U9 JO pəu1oo1 asn3qo Á[[peuors -2390 0] 9jnà? *juelIn2op asnjgo 0j 9jeutumoe (Mo1reu uəlJo) aao “ajeaoqo ndia sourepuooas 0} [o([e1ed Aqpeuoiseooo “jualimoJad 10 poexioj ‘ajue asn1qo 0] ə1noe pax.10] 10 Zuo] eysoduros 10 a¡duns 01F-£€1I Jen -3911 Áj[euorseooo 're|n3ə: 9jno? umrpəur 0j 9pt« snoulol -pojdureona as19491 ouo *snouloJpoptu20Jq 0} snouro1poptuoo1q padoo] |puijou BULL Snoo2PIIO2 JO SNOIIBIIBUYO əmvuə JO pəu1oo1 pe -punoa 0} 3M “JUSIMISP papuno. 1o ojeutZ1eure 0j 9jeurumoe 91e[notq1o (MOJIBU IO DPIM) 34840 'eje[oeoue[qo “onda pax 10] 10 jus1mo1ad snonuls ‘ajsue uropuei snon -uts ‘Buoj 0} pex10j ‘oys [1£-046 Ie¡n39.1.1 JO ire[ngoi ƏMƏL MOJIBU 0] 9DIA snouro1popruooJq Á[pei -stp 0} snouropojdureona pes1e| -uə 0] ewou *1y3te.ns sno99 -BLIO) JO SNOUB.IQUIQUI əmuə JO payiool asniqgo 0] ƏM “JUILINIIP pepuno. 1o ojeuig.1euro 0j ojeurumoe ponu -UlÁse JO *|eornəuruÁs 'eje[notqroqns *oje| -oooue[qo “ajeaoqo “onda SUIA Á1enio] sauepuoo -os19]u Japio yuey spəIp [81509.191u] ut -LIO ULSA Áepuoyas Áreuiugd 3.M3X3 J, uB adeys aeoo2ea[[Aqdosiuy aevaloydoziyy IBIYIOMOUÁA) aesuádodAy IBIISLIBIBIA ‘apaops0ydoz1y y fo saqiy pup avaavayAydosiup ayi fo s121204m2 fos] fo Xuownumç `Z 3I8vL Annals of the 1354 issouri Botanical Garden M (suu -1əpidə [erxepe se aures) Sutxoo[191ut 0} j[euodA[od Uc-ID TI s[¡89 pa1o1ua2 2318] Á[¡puorse290 :gurqoo[1o1 -ur 0} aje[npun o} ¡euo3A¡od Id -osəur uropuei 'suioA guo[e |euo3 od |euoS od náyd -Osgu1 uropuel *SUuIƏA guo[e soneustid euoZAjod D6-I s]]29 pepunos 193.18] Ajjeuotsea90 :[euogA[od sturiəpidə 10 [Aud -OSƏUuI WOpueI ¿SUISA guo[e [euo od Uc-I [euo Kqod srur1opido ut [|pEuorse22o “¡Aydosaur ut ulopuei JO sutƏA uo[e ¡euo3Ájod DS I-I ¡euo3 [od [¡4ydosaur ul UIOpUEI IO SUI9A Suo¡e sturiəp -tdə [eixeq v on "EI pM /p3uə] SUULISp -ida [erxepy uonnquisiq sasnip jews pue pues ‘sasnip umipeur oj [[eurs ‘sasnip son eustid sasnip aez ‘soneuistid əd} [eisÁr”) uta [eut -1eure1jut uajjo :padoo| padoo] padoo] padoo] padoo] &[ajo[duroo uonpuəA Álə1ə[duroout o1 padoo] A[eie[duroout Jo pədoo| Á[e1e|duroou: 10 pedoo| Á[e19[duroout 10 padoo| -ur A[[euorse220 ípadoo| [guts IEA D21uDjX22 Dat s3urpua %(S < 01 Maj 0} auou UOUIUIOD JO əuou auou -nodissD”) ut ÁJUO 2/0, 01 auou [|pproaqo?a [, 0c « sut -ST '81-0I '6-€ ‘Z-0 8-0 'F-0 OI-€ ‘6-7 '9-0 ‘E-I v-0 > Apsour *g-0 'z-0 S-I *2-1 ‘2-0 quet 9914 [euo3Á] 1e[nZu?1221 adeys jeuoZdjod o1 re[ngo.1ut ¡euo3Ajod o1 rejndoxzs -od A[yeuorseooo “ena. 91 ¡euo3Ájod o1 1e[nZo. [euo [od *repngo. it 9[oa1y juaur pedojaaap [jam pedojaaap [jam pedo padojaaap -dojaa 0} 129j1edurt *ejeduroout 01 129j1edur *ajo[duroout 1ajladun 01 ə1ə[duroout -|[949p JPM 01 129jJ12dumn [pom May *'129j1edur Apsour -əp e[oaiy Jap OI-S 8-5 L5 L-S ¿=S -10 {says} saurepuooas 0} [o[[ered Appeuorseo [puo sureui ¡euo34Ajod repn3311 -d0 ‘yeuoSAjod Kjrepn2o.1t Jengein — -SAjod re[n8əi Jo repə [puosA[od Áprepndous -op Krenio[ eeaoeoa[^udosruy aea1oudoziy 3BIYIONOUÁS) oeauASodÁH oeoarsLIE2P]A enuuguo) `g 318v[ Volume 75, Number 4 1988 Keating & Randrianasolo 1355 Leaf Architecture & Wood Anatomy Continued. TABLE 2. Hypogyneae Gynotrocheae Rhizophoreae Anisophylleaceae Macarisieae same or smaller than adaxi- same as adaxial Relative size same or smaller than same or usually smaller same or smaller than adaxial than adaxial adaxial brachyparacytic, few an- brachyparacytic, anisocy- Stomate brachyparacytic, anisocytic, obscure but probably all brachyparacytic, anomo- diacytic, paracytic, an- cyclocytic with 5-10 omocytic subsidiary cells omocytic, anisocytic cytic, paracytic (few diacytic, anisocytic) type tic, few diacytic, anom- ocytic abaxial abaxial abaxial abaxial abaxial Position absent absent astrosclereids in 3 of 4 absent absent Sclereids absent to occasional, simple, absent absent Trichomes absent to occasional, sim- absent to very common short to long long; near veins L] ple simple; near veins or random absent occasional, 2 layered absent usually present, 2 layered occasional, 2 layered Hypoderm features mentioned above. Juncosa (1984b) found that Rhizophora has the most specialized shoot apical morphology and hypocotylar anatomy, with Bruguiera more transitional to inland genera of Rhizophoraceae. Leaf samples of Bruguiera showed the lowest rank order and the highest number of free vein endings for the tribe, but they are not otherwise distinctive architecturally. The principal morphological diagnostic features of the Anisophylleaceae are the alternate leaves and the absence of stipules. No microscopic fea- tures absolutely distinguish its four genera from the Rhizophoraceae. Features not shared by Rhi- zophoraceae are asymmetrical leaves, teeth and marginal glands almost entirely absent, acrodrom- ous venation, highest vein order to ten, and free vein endings per areole to greater than 20. But within the Anisophylleaceae these features char- acterize fewer than all of the species of one genus. All remaining recorded features are also found within the range of the variable tribes of Rhizopho- raceae and supply no basis for separating the An- isophylleaceae from the Rhizophoraceae. A good case can be made that the Rhizophoreae are the most distinctive group based on their various spe- cializations, including cyclocytic stomata, unusual druse crystals, high-ranking brochidodromous ve- nation, vivipary, and mangrove habit. Yet much other available evidence argues for the obvious relatedness of the 14 genera of Rhizophoraceae s. str. and for their general similarity to the Aniso- phylleaceae. The only “teeth” in the Anisophyl- leaceae, the shallow marginal glands in Aniso- phyllea meniandii, are too simple structurally to allow an opinion on their origin or homologies. COMPARISONS WITH LEAVES OF OTHER FAMILIES The leaf architecture of the Rhizophoraceae i is of the families mentioned below are distinctly unlike the Rhizophoraceae. Each of the families was cho- sen by Dahlgren (this volume) for closer scrutiny after extensive comparisons using a broad data base. We compared all of the available cleared specimens in the National Cleared Leaf Collection, singling out those genera and species in particular that had the most elaborate, character-rich archi- tecture. he most useful characters appear to be of com- plex genetic ner melucing sien arces: ture, leaf shap 1 1356 Annals of the Missouri Botanical Garden gland type, and orientation of the related venation). The marginal configuration of the Cunonioid/ro- soid leaf, with the internal epithem and the strongly converging marginal veins producing a hydathodal tooth, appears complex and monophyletic. It ap- pears nowhere else than in the rosoid alliance as understood in all of the major current phylogenetic systems. However, there is insufficient information on the Macarisioid tooth type. The apiculate, de- ciduous gland serviced by a usually small single vein seems widespread and is probably not mono- phyletic, at least as the structure is detected in clearings. Celastraceae The ovate or elliptic leaves of this family are brochidodromous or eucamptodromous and of gen- erally low rank order; the margins are faintly toothed and structurally Macarisioid. Elaeodendron ae- thiopicum and Celastrus pringlei have a single dominant vein that arises from a secondary or tertiary loop. It ends just below an apicular gland that is oriented adjacent to or points toward a sinus. All associated minor veins are decurrent on the dominant veinlet and do not participate in vascu- larizing the gla tooth shape is Macarisioid but no dominant vein approaches the substantial gland. Two or three smaller marginal veins approach the gland but not closely. A number of leaf characters of this family, par- ticularly the secondary and higher order vascula- ture and the tooth type, are quite compatible with the Rhizophoraceae. nd. In Celastrus racemosus, the Cunoniaceae The leaves have prominent teeth with three con- verging veins leading to an internal epithem. The converging veins have perpendicular cross braces, giving the Cunonioid tooth a unique appearance but not hiding an obvious similarity to the slightly simpler rosoid tooth. This type of tooth was held by Hickey & Wolfe (1975) to be ancestral to the s no similarity to the Rhizopho- raceae at all. Cyrillaceae The two specimens available here, representing Purdiaea nipensis and Cyrilla nipensis, have el- liptic to obovate leaves with entire margins. The brochidodromous secondaries are straightened into a collector vein or are looped, respectively. While Purdiaea is unexceptional in its higher-order ve- nation and other features, the Cyrilla specimen does have incomplete venation with 15-20 vein endings per undefined aerole. It has stellate hairs f a type not seen in the Rhizophoraceae. The absence of teeth makes it impossible to compare marginal features productively, but the trend to- ward development of a strong submarginal collector vein in Purdiaea is not noted in the Rhizophora- ceae. Elaeocarpaceae Leaves of this family show broad variation in the level of specialization (rank order), ranging from the lrl brochidodromous arrangement of Elaeocarpus arnhemicus to the 4r leaves of the genus Sloanea. Margins are toothed in all speci- mens seen, and the secondary architecture ranges from brochidodromous to forked craspedodromous. None of the observed patterns of higher-order ve- nation or areolation can negate relationships with the Rhizophoraceae, nor does the tooth type. teeth are typically Macarisioid with the strong me- dial vein servicing a prominent apiculate gland that often points to the sinus. The decurrent branching veins attached to the median vein diverge, often in a falcate curve, and they show no tendency to be involved in tooth venation. This family shows the closest resemblance to Rhizophoraceae on the basis of the leaf architecture. Erythroxylaceae The leaves examined have a secondary archi- tecture of looped brochidodromous veins with ir- regular higher-order domains with a few unique or specialized features. Their level of organization is low first rank. Since the margins are entire, no tooth data are available. In most Erythroxylon species examined there is a rounded projecting apical gland vascularized by the midrib. They look the most macarisioid-like in a species from Java. Most of the glandular tips do not resemble the marginal features in the Rhizophoraceae. The fam- ily has trends of secondary and tertiary architec- ture unlike the Rhizophoraceae, including helicoid and admedially dendritic fourth- and fifth-order veins (E. Coli: bara from Sri Lanka) and sec- ly spaced with higher-order veins par- allel within tlie intercostal areas (E. urbanii from Puerto Rico). Flacourtiaceae Leaves of this large family vary greatly in size, shape, and secondary-venation patterns. Among Volume 75, Number 4 1988 Keating & Randrianasolo 1357 Leaf Architecture & Wood Anatomy the toothed leaves, two patterns are found, neither resembling the Rhizophoraceae. One of these has a Cunonioid vascular pattern with a strong median bundle that merges with two strongly converging lateral veins just below an apiculate tooth. The three converging veins are perpendicularly cross- braced with three or four minor veins. The other tooth type has one strong medial vein ending at the tooth apex, which may or may not have once had an apicular gland. Only one branch on the distal side passes through the sinus margin, pro- viding it with conspicuous vascularization. No fla- courtiaceous leaves seemed to approach the Rhi- zophoraceae architecturally. Geraniaceae Leaves of this family are often toothed and broadly elliptic. A principal characteristic of the family is parallelodromous secondary venation. Higher-order venation is not distinctive, and the low rank order (2r1) reflects the paucity of dis- tinctive comparative architectural characters. The teeth are typically rosoid throughout all samples examined and therefore quite different from the Rhizophoraceae. Viviana crenata (Vivianaceae or Geraniaceae) is typically geranioid, including the presence of the rosoid tooth. Hugoniaceae The four Hugonia leaf specimens examined are generally compatible with Rhizophoraceae. The secondary architecture is eucamptodromous or looped brochidodromous, and the rank order ranges from 2r to 4r. The crenate margins have teeth very similar to the Macarisioid type. The tooth is generally incurved with the deciduous glandular apiculum pointing toward or located in the sinus. A single large vein, apically curved, arises from the secondary arch or loop and ends below the apiculum. It is associated with branching loops and minor veins that recurve or otherwise show no tendency to converge on the tooth. No features negate relationship with the Rhizophoraceae. Ixonanthaceae The three genera (five species) sampled here generally resemble the Rhizophoraceae, although the secondary architecture includes semicraspe- dodromous as well as brochidodromous types. Teeth are not universal but are found in Octhocosmus, and in a reduced form in /xonanthes. Best-devel- oped toothed margins are crenate with the glan- dular apiculae so “*distal”” on the tooth as to be on the proximal side of the sinus. A prominent vein arises from a secondary loop, sometimes in an organized semicraspedodromous pattern, and ends just below the gland. Some of the minor veins branching from this trace tend to converge toward the pad below the gland. This represents a signif- icant difference from the Macarisioid tooth. Con- vergence of several veins toward the gland is a major characteristic of the Cunonioid/Rosoid/ Fuchsioid type, although it appears to be an in- dependent trend in the Ixonanthaceae. Other dis- tinctive trends in this family include development of an entire leaf with a fibrous marginal vein in Klainedoxa and in Irvingia, and semicraspedod- romous venation in Octhocosmus. Lepidobotryaceae The materials examined here are ovate leaves with entire margins and eucamptodromous sec- ondary architecture. Intercostal areas are irregu- lar, as are most of the tertiary domains and areoles. Some percurrent tertiaries are present. These un- specialized examples, Papania scandens and Sar- cotheca ferruginea, are generally compatible with the Rhizophoraceae. Oxalidaceae Oxalis acuminata and the other species ex- amined are trifoliate with entire ovate leaflets, which may be ciliate, looped brochidodromous, and with percurrent or irregular tertiaries. The reduced number of vein orders, five or six, and generally reduced appearance are compatible with the Rhi- zophoraceae. Linaceae The samples representing seven species are all more specialized than any Rhizophoraceae de- scribed. The secondary architecture is brochidod- romous or eucamptodromous, but this ends the similarities. Tertiary and quaternary venation is virtually parallel, being perpendicular to the axis of the intercostal areas (Roucheria griffithiana) or parallel to them (Roucheria calophylla). Linum shows tendencies toward acrodromy and has fine, papillate/serrate, unicellular teeth. No multicel- lular teeth are present. Tooth glands in Roucheria are somewhat Macarisioid in appearance, as a me- dian vein arises from a submarginal loop, ending just below the gland. However, there is also a definite tendency for lateral veins to converge to- ward the gland, a condition not seen in Rhizopho- raceae. In Linum and Reinwartia, the small mar- ginal projections are multicellular but not glandular or vascularized. It seems doubtful that the Rhi- 1358 Annals of the Missouri Botanical Garden zophoraceae leaf type has any close relationship with leaves of this family. Rosaceae This family is quite variable, and several types of secondary venation and teeth are found that bear much closer examination. Fragaria species generally have rosoid teeth, as does Holodiscus discolor, although the latter's medial vein is dom- inant, with the lateral ones being nearly absent. The marginal teeth of Cotoneaster pyricantha and Ka- geneckia lanceolata have deciduous apiculae and prominent medial tooth veins. Convergent lateral veins are present at most teeth or may not be obvious. The incurved appearance of the Coto- neaster tooth form is remarkably similar to the Macarisioid tooth. Saxifragaceae All leaves examined show a typical Rosoid hy- dathodal tooth with converging marginal veins that end in a flaring submarginal vascular plexus. They show no relationship to structures of the Rhizopho- raceae. Zygophyllaceae The ovate leaves of this family are generally not toothed, and they have brochidodromous venation with a different appearance of areolation than Rhi- zophoraceae. One specimen, representing Guaia- cum sanctum L., is craspedodromous with two or three small pointed teeth. A tooth is vascularized by a craspedodromous secondary vein that ends below its tip. There is apparently no apiculum or epithem at the tooth. Some minor veins form a converging buttress on the tooth vein. Overall, the tooth does not resemble Macarisioid architecture nor do the leaves resemble those of Rhizophoraceae in other respects. WOOD ANATOMY In the comparative wood anatomical literature, only two studies were found that were based on a comprehensive generic sample, those by Marco (1935) and by van Vliet (1976). It should be noted that neither author developed a major operating hypothesis that the genera of Anisophylleae should be segregated as a separate family nor did their data and analyses support that view. Among the wood features they listed, only those mentioned below may have value in distin- guishing the families. Table 3 provides a list of features we believe to have promise in differen- tiating the tribes and families. Vessel tangential diameters of Anisophylleaceae overlap the high end of the range of Rhizophora- ceae values, but they have the highest average and absolute values for this feature. Average vessel element length of the two families overlaps entirely, but Anisophylleaceae are at the short end of the range. Pores per square millimeter are the lowest for Anisophylleaceae with almost no overlap in the ranges for Rhizophoraceae. This correlates well with the high vessel tangential diameters. A com- putation of relative parabolic flow rates per mm? (Zimmermann, 1983) sh equivalent ranges with the Anisaphyllcacese producing their flow rates arger diameters on the av- with fewer pores and erage. The calculations were made using pooled data from the literature and are only indicative. Macarisieae and Gynotrocheae have simple and scalariform perforations, while these in the Rhi- zophoreae are exclusively scalariform. Rhizopho- reae also have the shortest vessel element lengths for the family, a character normally correlated with simple perforations (Dickison, 1975). It may be that efficiency of conduction (high flow rates) is not selected for in the mangroves. The generally small statures, coriaceous leaves, and restriction to saltwater habitats suggest the need to conserve water rather than to maximize its flow. No other features— including fiber-tracheids, parenchyma or rays—seem to provide distinctive or mutually exclusive characters for the two fam- ilies. Many of the characters are quite diagnostic for the genera, but differentiation of the tribes and of the two families is less readily accomplished. This is mostly due to very wide variation among the genera of Rhizophoraceae. The obvious dis- tinctiveness of the Rhizophoreae is paralleled by data and can no doubt be explained by habitat specialization. n other characters studied, the Anisophylle- ae show no similarity to the Rhizophoraceae. Chenery (1948) and Kukachka & Miller (1980) noted a positive aluminum test for the family, which is not shared by the Rhizophoraceae. Behnke (1981, 1984) noted S-type sieve element plastids in An- isophylleaceae and Myrtales but P-type in the Rhi- zophoraceae. In their embryological paper, Tobe & Raven (1987a) pointed out strong similarities of the Anisophylleaceae to the Myrtales and distinct differences with the Rhizophoraceae. However, they (1987b) later concluded that Anisophylleaceae and Rhizophoraceae do share enough deve elopmenta al homologies to hypothesize common ancestry. 1359 Keating & Randrianasolo Volume 75, Number 4 1988 Leaf Architecture & Wood Anatomy juesqe 0} pe1equireqo Áj[ensn ‘Areyyos eurAyouased ¡erxe /Áei s[[29 yieays [Puorse22o 'ojeuesu[nur peoq 9jeuos Qg-[ 'sezis younsip Z ‘Jp 1e[nj[2201919q 0cc'£-0T8 o1eduo[a 0j punol 91e[nonal 0] IJLUIAJL ‘IILI so1njiode juoo Árejgos eurdyouaied Kei s[¡p9 u1eəus 1uanbougut Ə1pLIƏS L-T i1e[njooouroq Ayyeuots -B220 THI ‘Il Ie[n[[2201919q OOF'Z-OTL uo ur1ojuqm ajduns 01 pa1op1oq [ey ə1e[nəonə: o} aytsoddo juenboaj '*pe1ejsn[o 10 Árejjos eurqouozed ¡erxe / ei soo yieays Aj[euorseooo ‘sprey ou ƏJLIIƏS Z€-] :səzIs jounsip Z ‘Jp 1e[nj[o20191eq 08T'£-OIc'I ayeBuoja 0} emn ayejnonei ‘aytsoddo ‘ajyeusayye ULIOJLIe[eos ‘qe juenbeajut 10 juonbojg Axe] -yos eur&qouaaed [eixe / Ae s[[2» 12219 10 aienbs oç-1 aye “Was C- [ ‘JJ “y 19[nj[2201919q 06€£'£-096 o1eZuo[o 0} emn wo} -ue[eos “¡euonisue.] *ajrsoddo [e s]e1sA10) sujeaus ‘spe Ary uoneuos ‘addy Áew (ur) y13U3]| prauoea-19qt q 3unud Kei-[ossa A -se[peoo juanba1] uit« ‘eurae ]euoursueg] *uriope[eos -uonisue1j *ajtsoddo *ojeuaoj[e -uoytsuey ‘aytsoddo *ojyeuieq[e Suuid AI ə|duris ULIOJLIR[eos ur1ogtre[eos / o[duris ULIOJLIe eos /ə|duns suon?1oj1aq 03€'1-033 099*1-033 062*1-03? OvE'Z-03S (um) iuo juəurə|ə [ossa A (war?) Otv-v8 O0tl-T€ 00€-I1Z Ov£- TG 19jourerp [enuague] [asso A cl-I 89-9 66-€ (8€ D IF-F (zuu) &ouanbaaj [assa A oeoo2vo[Audosiuy Əeəioudoztu1] əeəuoonouÁ9 aeauA3odAp :ovorsueov]q l9]O9P1E7) ‘avaonsoydozry y fo saqi pup avasvayAydosiup 211 fo səməf poom po122]2s fo Kinunung `ç TWV L 1360 Annals of the Missouri Botanical Garden perspective of vegetative anatomy alone is insuf- ficient at present to resolve substantially this prob- lem. On the basis of their lack of vestured pits and intraxylary phloem and because of their possession of the Macarisioid tooth, the Rhizophoraceae s. str. appear to have no relationship to the Myrtales. Leaf teeth as known in all of the Rosales and Myrtales thus far examined are metrically vascularized by two to several converging veins surmounted by a secretory epithem, a sub- marginal foramen, and a connection to the surface through stomatelike pores. The tooth sinus is never reported to be included in the tooth vascularization. As described earlier here, the Macarisioid tooth is different in all respects. Thus far, the closest sim- ilarities to the Macarisioid tooth structure, and the general leaf architecture, can be found in the Dil. leniid line. The Celastraceae (Celastrales), Elaeo- carpaceae (Malvales), and Cyrillaceae (Ericales or Theales) are commonly classified in that lineage. Similarities seem more distant with the Hugoni- aceae (Linales of Cronquist), Lepidobotryaceae (Geraniales of Dahlgren), and Oxalidaceae (Gera- niales). Considering that the field of comparative wood anatomy has a long established tradition, the results of comparisons among the families mentioned in this study are uncomfortably ambiguous. The new- er field of leaf architecture has an equally well- drawn glossary and some organized preliminary synthesis regarding the character syndromes and their evolution. Here, too, we are left with equally large ambiguities. While the treatments of leaf architecture of families are far fewer than those based on comparative wood anatomy, existing com- prehensive collections of cleared leaves should at least partially compensate. Some of these ambi- guities may vanish in the future when the devel- opmental bases of architectural forms becomes bet- ter understood and when the homologies of features of some of the larger phylads have been studied. LITERATURE CITED BEHNKE, H.-D. 1981. Sieve element characters. Nordic Bot. 1: 381-400. 1984. Ultrastructure of sieve-element plastids of Myrtales and allied groups. Ann. Missouri Bot. Gard. 71: 824-831. CHENERY, E. M. 1948. Aluminum in the plant world. I. Kew Bull. 1948: 173-183. CRoNQUIST, A. 1983. Some sz E in the dicot- yledons. Nordic. J. Bot. -83. DAHLGREN, R. Ge do s of angiosperm evolution and macrosystematics. Nordic J. Bot. 3: 9. Rhizophoraceae and Anisophylleaceae: sum- mary statement, cia Ann. Missouri Bot. Gard. ue volum R. F. THORNE 1984. The order Myrtales: circumscription, variation, and relationships. Ann. Missouri Bot. Gar ^ -699. Dickison, W. C. 1975. The bases of angiosperm phy- logeny: vegetative anatomy. Ann. Missouri Bot. Gard. DOYLE, 1 A. & L. J. Hickey. 1976. Pollen and leaves from the Mid-Cretaceous Potomac Group and their boaring on Sai angiosperm evolution. Pp. 139-206 in C. B. k (editor), Origin and Early Evolution Yo : bonais Rhizophoraceae-Macarisieae. Adansonia, Sér 2, 14: 499-506. GivvisH, T. 1979. On the adaptive significance of leaf for Population Biology. Colum Hickey, L. J. . Classification of the architecture of dicotyledonous leaves. Amer. J. Bot. 50: 17-33. Stratigraphy and paleobotany of the Golden Valley d (Early Tertiary) of western Nor Rn a. Geol. Soc. Amer. Mem. 150: 1-181. A revised classification of the archi- tecture e dicotyledonous leaves. Pp. 25-39 in C. R. Metcalfe & L. Chalk. Anatomy of the Dicotyle. dons, 2nd edition, Volume 1. Clarendon Press, Ox- ford. & J. A. DoYLE. 1977. Early Cretaceous fossil .43:3 oe for angiosperm evolution. Bot. Rev 10 . A. WOLFE. 1975. The bases of angio- sperm phylogeny: vegetative morphology. Ann. Mis- souri Bot. Gard. 62: 538-589. Hou, D. 1958. Rhizophoraceae. Pp. 429-493 in C. G. G. J. van Steenis (editor), Flora Malesiana, Ser. I. Spermatophyta, Volume 5. Noordhoff-Kolff. Djar- arta. Juncosa, A. M. 1984a. development in Cassipourea elliptica r Se Amer. J. Bot : 170-179. Embr ryogenesis “and developmental endis of the seedling in Bruguiera exaristata ena Hou (Rhizophoraceae). Amer. d Bot. 71: 180- 19 Embryogenesis and seedling (Sw.) Poir. in. A. M. & P. B. ToMLINSON. A historical and taxonomic synopsis of Rhizophoraceae and Aniso- phylleaceae. Ann. Missouri Bot. Gard. (this volume). Kukacuka, B. F. & R. B. MILLER. 1980. A chemical spot-test for aluminum and its value in dus iden- tification. I. A. W. A. Bull. 1: 104-109. Is. 1974. Colleter anatomy Systematic foliar morphology of Phyllanthoideae (Euphorbiaceae). I. Conspectus. Ann. Missouri Bot. Gard. 73: 29-85. Manco, H. F. 35. Systematic anatomy of M woods of the Rhizophoraceae. Trop. Woods. 44: 1-20. Melchior, H. 1964. Myrtiflorae. Pp. are in À. Engler’s Syllabus der Pflanzenfamilien. Aufl. 12, Band 2. Borntraeger, Berlin. PRANCE, G. T., M. F. pa SiLva, B. W. ALBUQUERQUE, I. DA SILVA ARAÚJO, L. M. M. Carreira, M. M. N. Volume 75, Number 4 1988 Keating & Randrianasolo Leaf Architecture & Wood Anatomy 1361 BRACA, M. Macepo, P. N. pa Conceição, P. L. B. . Lispóa & R. C. Revisão taxonómica das espécies amazonicas de Rhizophoraceae. Acta Amazonica 5: -22. Rao, T. A. & J. BHATTACHARYA. 1978. eview of foliar sclereids in angiosperms. Bull. Bot. ie India 91- & S. Da s. 1979. Typology of foliar — in angiosperms. Proc. Indian Acad. Sci. 88B(2): 3 345. ———, J. BHATTACHARYA & J. C. Das. lord in Rhizop plications. Proc. In lan - “ens ai on Mangrove Research -1975. Unesco, . A. € R. 1978. Foliar em 1983. Revision u ouri Bot. Gard. 70: 179-193. : 80. Outline of the classification of flowering plants (Magnoliophyta). Bot. Rev. 46: 225- THORNE, RF. . Proposed new realignments in angiosperms. Nordic J. Bot. 3: 85-117. Tose, H. & P. H .RAVEN. 1987a. Systematic embryol- ogy of the Anisophylleaceae. Ann. Missouri Bot. Gard. 74: 1-26. .—————. 1987b. The embryology and re- lationships of Cassipourea and Sterigmapetalum hizophoraceae- Macarisieae) Opera Bot. 92: 253- 264. TOMLINSON, P. B. & D. W. WHEAT. 1979. Bijugate i ). J. Linn. Soc. Bot. 78: 317-321 VINCENT, J. R. & P. B. TOMLINSON. 1983. Architecture and phyllotaxis of eae disticha (Rhizopho- al Garden's Bull. 36: 3-18. VLIET, G. J. C. M., van. 1976. aid rcd of the Rhizophoraceae. Leiden Bot. Ser. 3: 20-75. YosHIoKA, H., K. Konpo, M. NA K NEHIRO & S.-I. MaEDA. 1984. Karyomorphological studies in pis species of mangrove genera in the Rhizophora- ae. La Kromosomo 11-35-36: 1111-1116. UN unas M. H. Xylem Structure and the Ascent of Sap. Springer- Verlag, New York. 1362 Annals of the Missouri Botanical Garden dromous venation ns.— Ficures 1-6. Cleared leaves of Rhizophoraceae. — 1. Anopyxis calaensis, showing eucampto J calaensis, showing marginal venation and irregular higher-order d ) j or- and entire, irregular margin. Figure 4 showing the vein borderin Median portion of lamina of B. corymbosum demonstra onal tertiary and quarternary domains. Scale lines: Figures 5 mm. e ing irregular polyg — | mm; Figure 4 — 1, 3 = 1 cm; Figures 2, 5, 6 Volume 75, Number 4 Keating & Randrianasolo 1363 1988 Leaf Architecture & Wood Anatomy "m Y 3) q * y ees 7-12. Cleared leaves of Rhizophoraceae.— 7. Comiphyton gabonensis, with strongly ascending todromous secondaries and irregular intercostal areas.— 8. C. gabonensis, margin with reduced "tooth" Sith ee gland (arrow) .—9. Macarisia lanceolata, with a a: teeth.—10. M. pyramidata, eucampto- dromous leaf with well- developed teeth. —11. M. pyramidata, margin showing single tooth veins arising from the secondary loops. — 12. M. pyramidata, showing vein entering a ciliate tooth from a Noten rical angle. Scale lines: Figures 7, 10 = 1 cm; Figures 8, 11 = 1 mm; Figure 9 = 5 mm; Figure 12 = 500 1364 Annals of the Missouri Botanical Garden Els < xL 2 g y ALIIA Ps AS 2 Oe 72 v» % nu 4 LAS = PA FIGURES 13-18. Cleared leaves of Rhizophoraceae. — 13. Cassipourea guianensis, showing looped brochidod- romous venation with wide, regular intercostal areas and well-formed intersecondaries.— 14. C. guianensis, showing single vein leading to small marginal tooth. Note that no other veins converge or participate in vascularizing the tooth. — 15. C. ceylanica, showing the least organized intercostal areas among any of the cassipoureas examined. The margin is entire.— 16. C. ceylanica, entire margin, irregular higher-order domains, and imperfect areolation with a highly variable number of free vein endings.— 17. Sterigmapetalum guianense, showing looped brochidodromous venation and narrow intercostal areas.— 18. S. guianense, showing the entire margin with looped venation. Scale lines: Figures 13, 15, 17 = 1 cm; Figures 14, 16, 18 = 1 mm Volume 75, Number 4 Keating & Randrianasolo 1365 1988 Leaf Architecture & Wood Anatomy e 1 y, Om ; sb | xe Cleared leaves of Rhizophoraceae.— 19. Cass -— Crossostylis multiflora, showing irregular quaternary domains and imperfect areolation. — 24. Gynotroches axillaris, with eucamptodromous venation and regular percurrent quaternaries. Scale lines: Figure 9 — ] cm; Figures 20-24 — 1 mm. 1366 Annals of the Missouri Botanical Garden FIGURES 25-31. Cleared leaves of Rhizophoraceae. —25. Rhizophora mangle, showing obscure areolation. — 26. R. mangle, showing astrosclereids in the mesophyll. —27. Bruguiera gymnorrhiza, showing cyclocytic sto- mata.— 28. Ceriops boiviniana, leaf margin showing parallel secondaries and tendency toward development of C. boi a collector vein. —29. C. boiviniana, showing cyclocytic stomata.—30. Kandelia candel, areoles showing veinlets with tracheoidal endings.—31. K. candel, cyclocytic stomata. Scale lines: Figure 28 = 1 mm; Figures 27, 29, 31 = 100 um; Figure 26 = 250 um; Figures 25, 30 = 500 um. Volume 75, Number 4 Keating & Randrianasolo 1367 1988 Leaf Architecture & Wood Anatomy >. “ Al AS BSS EN ARA es eA: SEE Ç EE ES SS 7 E VEN eS l LSN x ie di S T pro A e f AY "C FIGURES 32-38. Cleared leaves of Anisophylleaceae. —32. Anisophyllea cinnamomea, leaf with asymmetrical base and basal acrodromous venation. —33. A. griffithii, leaf with brochidodromous looped margin and irregular higher-order venation.—34. A. boehmii, showing margin of eucamptodromous leaf. —35. A. meniandii, leaf with eucamptodromous entire margin.— 36. A. poggei, mesophyll showing irregular higher-order venation.—37. A. disticha, mesophyll showing large, incomplete areolation. — 38. A. disticha, veinlets showing differentiated sheath cells. Scale lines: Figure 32 = 1 cm; Figures 33-37 = 1 mm; Figure 38 = 100 um. 1368 Annals of the Missouri Botanical Garden ETT Casa AVISA T7 a FicunES 39-43. Cleared leaves of Anisophylleaceae.—39. Anisophyllea pomifera, showing looped margin.— 40. Combretocarpus rotundatus, margin of eucamptodromous leaf. —41. Polygonanthus amazonicus, margin with brochidodromous secondaries and imperfect areolation.—42. Poga oleosa, margin with brochidodromous sec- ondaries and imperfect areolation.—43. P. oleosa, tracheoidal vein endings and druses. Scale lines: Figures 39— — | mm; Figure 43 — 100 um. MORPHOLOGY AND PHENETICS OF RHIZOPHORACEAE POLLEN! Edward L. Vezey,? Varsha P. Shah? John J. Skvarla,?* and Peter H. Raven‘ ABSTRACT Pollen morphologic data from light, scanning, and transmission electron microscopy were used in a phenetic analysis to assess variation within and am leeae, Gynotrocheae, Macarisieae, and ong the four tribes traditionally included in Rhizophoraceae: Anisophyl- Rhizophoreae. Principal components analysis revealed that pollen of Anisophylleeae is phenetically divergent from that of ouai aud Macarisieae, and Rhizophoreae, and therefore ollen of Maca doapertures Vossii ie an pollen, Rhizophoraceae possess endoapertures with some amily level, Anisophylleaceae. In contrast Rhizophoraceae sensu stricto forms y phy P risieae intermediate between diese and Rhizophoreae. En- when present, are circular and oorly defined, whereas all species of degree of fusion. Pollen ofi both families has a generalized s Mida up thereby providing no piece basis for assessing relationships to Myrtales or other gro Comparative palynology in Rhizophoraceae has focused on the mangrove genus Rhizophora, pri- marily in connection with the recognition and study of paleo-shorelines (Kuprianova, 1959; Langen- 1967; Assemien, 1969; Rakosi, 1978; Sowunmi, 1981). Consequently, several species of Rhizophora have been well documented. Langen- heim et al. (1967) used light microscope data (pol- len shape and a unique endoaperture system) to characterize R. mangle, R. samoensis, R. race- mosa, and R. harrisonii. Muller & Caratini (1977) expanded the study of modern Rhizophoraceae by employing transmission electron microscopy (T in addition to light microscopy ( and scanning electron microscopy (SEM). Their analysis included three species studied by Langenheim et al. (1967), R. mangle, R. racemosa, and heim et al., R. harrisonii, as well as R. mucronata, R. stylosa, R. apiculata, R. lamarckii, and R. brevistyla. Muller & Caratini (1977) essentially confirmed the findings of Lan- genheim et al. (1967), but underscored that most LM characters exhibit too much overlap to separate species. Although their study lacked the benefit of comparison with other Rhizophoreae (Bruguiera, Ceriops, and Kandelia), as well as other members of the family, they subdivided the Rhizophora pollen type into four groups by combining LM data with an SEM analysis of sculpture patterns. Typically, other pollen studies of the family were accomplished as part of floristic or general mor- phologic surveys (Erdtman, 1952; Kubitzki, 1965; Huang, 1968; Guers, 1974; Geh & Keng, 1974; Sowunmi, 1974; Straka & Friedrich, 1984; Than- ikaimoni, 1986a, 1987). SEM studies have been centered on the tribe Rhizophoreae (Tissot, 1979; atin 1983; Ludlow-Wiechers & Alvarado, 983). For complete references to pollen studies in the family, see Thanikaimoni (1972, 1973, 1976, 1980, 6b). Little palynologic attention has been directed specifically to the taxonomic integrity within and among the four tribes traditionally included in Rhi- zophoraceae sensu lato: Anisophylleeae, Gynotro- cheae, Macarisieae, and Rhizophoreae. Using LM, SEM, and TEM, we investigated the pollen mor- phology of all genera in these taxa. Of particular interest is Anisophylleeae, which on the basis of a broad array of characters has been considered to constitute a distinct family, Anisophylleaceae (see other symposium papers). With this in mind, we ! This study was partially supported by grants from the National Science Foundation to J.J.S. (BSR-83155 186) and P.H.R. We gratefully acknowledge the help of Daniel J. Houg h and Henry Loconte, University of Oklahoma, for many suggestions in developing the n analysis, and jr; ial D. Schnell and James R. Estes, University of | Oklahoma, for ‘arable the completed ma ? Department of Botan * Oklahoma Biological: Survey, Pu of * Missouri Botanical Garden, P.O. Box ANN. ript. robiology, University of Oklahoma, dé dun Oklahoma 73019, U.S.A. oma, Norman, Okla U.S. A. 299, St. Touts Missouri 63166- 0299, US. A. Missouni Bor. Ganp. 75: 1369-1386. 1988. ma 73019, 1370 Annals of the Missouri Botanical Garden TABLE l. Taxa examined, collection data, and plate references. Figures OTU Taxa Location Collector / Herbarium SEM TEM Anisophyllea buttneri Engl. Gabon Thollon 4130 (MO) 2 Al 4A. disticha (Jack.) Baill. Singapore Kiah & Leong s.n. in 1984 4 36 (no voucher) A2 A. fallax Scott Elliot Madagascar Reserves Nat. 2619 (TAN) 40 A. laurina R. Br. West Africa Fairchild s.n. in 1927 (US) 1 41 A3 A. obtusifolia Engl. & Brehmer Tanzania (Usam- Amani River Institute s.n. 3 38, 39 bara Mountains) (no voucher), Bukit Timah Natural Reserve Combretocarpus rotundatus N Borneo Tandom 2816 (K) 37 (Miq.) Dans. A4 C. rotundatus Brunei dia s.n. in 1983 (no 7,8 ucher) A5 Poga oleosa Pierre Cameroon Thomas 2273 (MO) 5 P. oleosa Nigeria Coombe 186 (K) 43 A6 Polygonanthus amazonicus Brazil Zarucchi 3138 (US) 9 Ducke P. amazonicus Brazil Pires 1281 (NY) 42 P. amazonicus Brazil S. R. Hill 12922 (MO) 6 G1 Carallia brachiata (Lour.) Australia Jackes s.n. in 1983 (JCT) 35 44 Merr. 62 C. eugenioides King Malaysia B. C. Stone 15114 (KLU) 33,34 45 Crossostylis biflora. Forst. (C. Society Islands St. John 17346 (MO) raiateensis J. W. Moore) G3 C. grandiflora (Pancher ex New Caledonia McPherson 6331 (MO) 31 Brogn. & Gris. C. grandiflora New Caledonia McPherson 1898 (MO) 46 G4 . Gynotroches axillaris Blume Malaysia B. C. Stone 15397 (KLU) 30 47 G5 . Pellacalyx cf. saccardianus Malaysia B. C. Stone 15396 (KLU) 32 49 Scortech. P. pustulata Merr. Philippines Wenzel 1497 (MO) 48 Anopyxis ealeaensis Sprague Belgian Congo Germain. o (MO) 23 50 Ml A. kleineana (Pierre) Engl. Cameroon Thomas 4 (MO) 22 51 Blepharistemma membranifolia India Wallich p (K) 59 (Miq.) Ding Hou M2 B. membranifolia India Manilal s.n. in 1984 (no 17 voucher) Cassipourea afzelii (Oliv.) Al- Liberia Baldwin 10609 (MO) 18 ston M3 C. elliptica (Sw.) Poir. Panama Kirkbride & Duke 1322 58 (MO) C. guianensis Aubl. Brazil Nelson 1324 (MO, NY) 19 C. gummiflora Tul. var. verti- Zimbabwe (cul- Muller 3558 (SRGH) 21 cillata (N. E. Br.) J. Lewis tured Harare Bot. Gard.) Compiphyton gabonense Floret Gabon Le Tetsu 5918 (P) 10 M4 C. gabonense Zaire Germain 5213 (BR) 62 M5 Dactylopetalum sessiliflorum Madagascar Reserves Nat. 4327 (TAN) 20 60 Benth. D. zenkeri Engl. Cameroon Zenker 4701 (MO) 61 M6 Macarisia ellipticifolia Arènes Madagascar Service For. 1972 (TAN) 14 55 M. humberti Arénes Madagascar Humbert 23505 (P) 15 56 M. lanceolata Baill. Madagascar Service For. 9366 (P) 13 54 M. lanceolata Madagascar Serv. Eaux & Forét 2955 (TAN) Volume 75, Number 4 1988 Veze et al. Rhizophoraceae Pollen 1371 TABLE l. Continued. Figures OTU Taxa Location Collector /Herbarium SEM TEM M? M. pyramidata Thou. Madagascar Service For. 9741 (P) 16 57 Petalodactylus obovata Arènes Madagascar Alaotra Agric. Sta. 3868 63 (TAN) M8 Sterigmapetalum heterodoxum Venezuela Wingfield 13245 (MO) 11 53 Steyerm. M9 S. obovatum Kuhlm. Brazil Maguire et al. 56502 (MO) 12 52 Bruguiera gymnorrhiza (L.) Madagascar Alaotra Agric. Sta. 27552 68 Lamk. (TAN) RI . gymnorrhiza Madagascar Reserves Nat. 9255 (TAN) 28 Ceriops tagal (Perr.) C. B. Rob. Madagascar Hervien s.n. (TAN) R2 C.tagal Madagascar Dorr & Koenders 3063 21 69 (MO) R3 Kandelia candel (L.) Druce Japan Murata & Nakamura 1142 29 67 (MO) R4 Rhizophora mangle L. Florida (cultivated) Tobe s.n. in 1981 (no 24 66 voucher; Fairchild Botani- cal Garden) R5 R. mucronata Lamk. Mozambique Torre & Paiva 11483 (MO) R6 R. mucronata Madagascar Bosser 9947 (TAN) 64 R. mucronata Madagascar ervien s.n. in 1964 (TAN) R. mucronata Madagascar Marot 2602 (TAN) 25 R7 R. stylosa Griff. Florida (cultivated) Tobe s.n. in 1981 (no 26 65 voucher; Fairchild Botani- cal Garden FG69-111) conducted a phenetic analysis to assess overall morphologic variation within and among these groups and to test the hypothesis of separate fa- milial status for Anisophylleaceae. MATERIALS AND METHODS POLLEN MORPHOLOGY In our investigation, pollen from 51 collections (representing 39 species of Anisophylleaceae and Rhizophoraceae) was examined by light and elec- tron microscopy (Table 1). Hereinafter, Rhizopho- ture of Erdtman were mounted in glycerine jelly and observed with a Leitz Ortholux microscope using transmitted light. Measurements (um) were based on 10-50 undis- torted grains when possible. Pollen for SEM was either air dried from 95% ethanol or critical-point dried, sputter coated with gold, and examined with either an ISI Super II SEM or ETEC Autoscan SEM. Pollen for TEM was processed as reported earlier (Skvarla, 1966) and examined with either a Philips model 200 TEM or Zeiss 10 TEM PHENETIC ANALYSIS Operational Taxonomic Units (OTUs). Data from 27 collections (Table 1) were subjected to a numerical phenetic analysis as Operational Taxo- nomic Units (OTUs; Sneath & Sokal, 1973). These 27 OTUs represent 26 species (Rhizophora mu- cronata has two OTUs) and were selected if data were obtained for all 33 characters (Table 2). Characters. The 33 characters (Table 2) used in the phenetic analysis are based on standard palynological data (Erdtman, 1952; Faegri & Iver- sen, 1975), and include 10 LM (1-10), 13 SEM (11-23), and 10 TEM (24-33) characters. Al- though most are self-explanatory, several merit elaboration. The variability in polar axis (P) and breadth (E) is often expressed in terms of range or minimum and maximum measurements. We used standard deviation (characters 2 and 4) in- 1372 Annals of the Missouri Botanical Garden TaBLE 2. Characters used in numerical analysis. Polar axis (P) 1. Mean 2. Standard deviation (SDP) Greatest breadth (E) 3. Mean 4. Standard deviation (SDE) 5. Polar axis/greatest breadth (P/E) 6. Endoaperture fusion (EF) Completely unfused Mostly unfused (< 40%) Mixed (approx. 50%) Mostly fused (> 60%) w NO 7. Mean polar length of endoaperture (PL) 8. Endoapertural index (El) 9. Mean distance between colpal ends (DCE) 10. Polar area index (PAI) Sculpture of mesolcolpia Sculpture of poles 11. Psilate (PS) 15. Psilate (PS) 12. Punctate (PU) 16. Punctate (PU) 13. Rugulate (RU) 17. Rugulate (RU) 14. Striate (ST) 18. Striate (ST) Sculpture of mesocolpial margins 19. Psilate (PS) 20. Punctate (PU) 21. Rugulate (RU) 22. Spinulate (SP) 23. Striate (ST) Exine structure 24. Tectum thickness (TT) 25. Height of columellae (CH) 26. Maximum width of columellae (CW) 27. Foot layer thickness (FL 28. Endexine thickness (EN) 29. Tectum thickness ratio (TT/TET*) 30. Columellar height ratio (CH/TET) 31. Foot layer thickness ratio (FL/TET) 32. Endexine thickness ratio (EN/TET) 33. Intercolumellar granulation (IG) No granulation 0 Incipient granulation (< 50%) 1 Abundant granulation (> 50%) 2 * TET = total exine thickness. stead, because it is based on all grains measured and is less affected by sample size or aberrant grains. Character 6, endoaperture fusion, expresses information usually included for Rhizophoraceae pollen (e.g., er & Caratini, 1977). I ordered multistate character based on increasing percentage of endoaperture fusion and was pro- cessed in the same way as quantitative characters (Sneath & Sokal, 1973). Character 8, endoaper- t is an tural index (EI), is the ratio of polar length of the endoaperture (PL) to polar axis (P). It is a ratio we have constructed to express the relative width of the e The 13 SEM characters describe the sculpture of Tu parts of the pollen surface: mesocolpia (11-14), poles (15-18), and mesocolpial margins (19-23). Although this information can con- densed into three unordered multistate characters, only ordered multistate characters are acceptable with principal components analysis (Gower, 1966). Therefore, as Gower (1966) recommended, we used binary (0, 1) characters to indicate absence (0) or presence (1) of a particular sculpture pattern. M characters 24-28 were recorded as means of measurements (um) taken from an average of six negatives. All measurements were mid-meso- colpial on equatorial sections perpendicular to the polar axis. Character 33 is an ordered multistate character based on increasing percentage of inter- columellar granulation. The remaining characters (29-32) are expressions of exine shape, that is, the relative contribution of tectum, columellae, foot layer, and endexine to the total exine thickness. n five cases it was necessary to take TEM data from a different collection of the same species (Table 3). This procedure was followed to insure adequate representation of all taxa. Numerical Analysis. Phenetic variation was analyzed using NT-SYS (Rohlf et al., 1982), a package of multivariate computer programs de- signed for use in systematics. The data (Table 3) were standardized (i.e., each character being trans- formed to have a mean of zero and standard de- viation of one), followed by calculation of a Pearson product-moment correlation matrix and principal components analysis (Sneath & Sokal, 1973). A minimum spanning tree (Dunn & Everitt, 1982) was calculated using an average taxonomic dis- tance matrix (Sneath & Sokal, 1973) produced from the standardized data set. Principal components analysis also produces a matrix of eigenvectors showing character loadings on each component (Table 5). If a character has a loading of 0.9 on component I, then (0.9) or 0.81 (81%) of that character is expressed or sta- tistically “explained” on component I and the re- maining 19% on other components. Table 5 lists all characters with loadings greater than 0.5. If a character has a positive (+) loading, then OTUs with higher values for that character tend to be found toward the positive end of that component. Higher values include larger quantitative mea- Volume 75, Number 4 1988 Vezey et al. 1373 Rhizophoraceae Pollen surements, higher-numbered multistate character states, and binary (0, 1) characters with character state RESULTS POLLEN MORPHOLOGY Detailed LM, SEM, and TEM measurements and observations for 27 collections (Table 1) are given in Table 3, and summarized for Anisophyl- leaceae and each tribe of Rhizophoraceae in Table 4. The supplemental descriptions below include characters not considered in the numerical anal- ysis, as well as unique morphological features in need of emphasis. Morphological delineation is based on all 51 collections (Table 1). Anisophylleaceae LM. ollen is mainly tricolporoidate (occa- sionally tricolporate), rarely with two apertures, radially symmetrical, and isopolar. Syncolpate grains are common. Endoapertures, when present, are circular but poorly defined. SEM (Figs. 1-9). phyllea disticha, which differs markedly from all other species in this study by having a striate Of special note is Aniso- surface. - TEM (Figs. 36-43). In A. disticha and A. obtusifolia, a narrow and highly undulating colu- mellae layer is present in sectional planes near the equator (Figs. 36, 38). In a different sectional plane of A. obtusifolia, and in A. laurina, the columellae layer is straight, and the foot layer is thinner than the tectum (Figs. 39, 41). Gynotrocheae LM. Pollen is tricolporate, radially symmet- rical, and isopolar. Endoapertures in all grains are fused laterally, the only tribe so distinguished. SEM (Figs. 30-35). A psilate-punctate sur- face is dominant in this tribe. TEM (Figs. 44-49). In Crossostylis grandi- flora a trace of granular matrix similar to that described for Macarisieae (see below) is present at the lower tectum margin (Fig. 46). Pellacalyx differs from other Gynotrocheae by having a thin tectum and tall, branched columellae that become shorter at the poles (Figs. 48, 49); granules are suggested beneath the lower tectum margin and the distal parts of the columellae. Macarisieae Pollen is tricolporate, rarely dicolporate and tetracolporate, radially symmetrical, and iso- polar. Shape has the greatest range of the tribes, from suboblate to prolate (character 5, Table 3). SEM (Figs. 10-23). This tribe processes vari- able pollen sculpture (Table 4). TEM (Figs. 50-63). either partially or completely filled with a matrix of granules (Figs. 50, 52, 54, 55, 57, 59). Some columellae also are granular distally. In taxa with prominent fused (zonorate) endoapertures (char- acter 6, Table 2), the endexine is granular in the mesocolpia in the vicinity of the endoaperture. Intercolumellar spaces are Rhizophoreae LM. Pollen is tricolporate, radially symmet- rical, and isopolar (except some grains in Rhizoph- ora mucronata). SEM (Figs. 24-29). Pollen of R. mucronata has a basically punctate-rugulate surface. How- ever, this pattern varies among the five collections examined (Table 1), as well as within collections. In the collection A. Torre & J. Pavia 11483 rugulate elements are distinct; some grains in this collection showed a punctate-rugulate surface on one hemisphere and a psilate—punctate surface on the other. Dicolporate grains were common. In the collection J. Bossier 9947 the surface has a faint rugulate-punctate sculpture, while in the collection Greve 290 it is psilate—punctate. In the collection Hervien s.n., made in 1964, the pollen grains are psilate-punctate, faintly rugulate-punctate, or dis- tinctly rugulate-punctate. Some dicolporate grains are also present in this collection. In the collection P. Marot 2602 the grains are rugulate-punctate; some also have spinules. Note that spinules were present on the mesocolpial margins of Carallia brachiata (Fig. 35), C. eugenioides (Figs. 33, 34), and Gynotroches axillaris (Fig. 30). TEM (Figs. 64-69). Tectum thickness is the most variable of the four groups (character 24, Table 3). PHENETIC ANALYSIS Principal components analysis reveals two phe- netically distinct groups of OTUs (Fig. 70). One group, consisting of all Anisophylleaceae OTUs, is located toward the positive end of component I and the negative end of component II. The first two components account for 25.5 and 20.7% of the total variation. The other group is an elongated continuum formed by the three tribes of Rhizopho- raceae, with Macarisieae OTUs distributed across phenetic space between Rhizophoreae and Gyno- trocheae. The minimum spanning tree indicates 1374 Annals of the Missouri Botanical Garden TABLE 3. Data set used in numerical analysis.’ Characters OTU l 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Al 22 1.2 16 0.8 135 0 2 .08 3 .17 0 0 0 l 0 0 A2 28 3.5 17 3.0 164 0 0 .00 2 .09 0 1 0 0 0 1 A3 26 1.7 20 1.7 131 0 2 .09 3 .14 0 1 0 0 0 1 A4 20 1.2 17 1.7 118 0 0 .00 3 .16 0 1 0 0 0 1 A5 29 2.6 22 1.7 133 0 3 .10 4 .17 0 1 0 0 0 1 A6 28 1.6 25 1.6 114 0 0 .00 5 .17 0 1 0 0 0 1 Gl 14 0.6 12 0.6 117 3 2 al 3 :22 1 1 0 0 1 1 G2 17 1.5 14 1.0 125 3 5 aT 3 .18 l l 0 0 1 l G3 15 0.9 14 0.8 105 3 5 .30 2 .14 1 1 0 0 1 1 G4 15 3.0 12 2.3 122 3 4 .29 3 .23 1 1 0 0 l 1 G5 11 1.2 8 0.7 124 3 2 .23 ] ad 1 0 0 0 l 0 M1 24 1.3 22 1.8 106 3 4 .16 4 .15 1 0 0 0 1 0 M2 19 1.6 18 1.1 107 3 4 .19 5 .27 0 ] 1 0 0 1 M3 20 1.7 15 1.3 133 2 3 .13 4 26 0 1 0 0 0 1 M4 17 0.6 13 1.4 129 l 2 .10 2 .11 0 1 0 0 1 0 M5 17 1.2 15 1.4 109 l 2 .10 3 .20 0 1 0 0 0 l M6 12 0.9 13 0.9 88 3 l .11 2 :17 1 0 0 0 1 0 M7 13 1.5 14 0.9 97 1 1 .07 3 .26 0 1 1 0 1 l M8 19 2.5 20 1.7 85 3 5 .25 4 .19 1 0 0 0 1 0 M9 14 1.1 16 1.4 86 3 2 .16 3 :21 0 1 1 0 0 1 RI 18 1.0 19 1.1 93 2 2 13 5 A. 0 1 1 0 0 1 R2 13 1.0 14 1.2 92 3 3 19 4 2d 0 1 1 0 0 1 R3 20 1.2 22 1.5 92 1 3 .16 4 .21 0 1 1 0 0 1 R4 19 1.0 19 0.9 103 3 3 .17 5 .28 0 1 1 0 0 1 R5 23 1.3 21 1.4 107 3 3 ll 6 ¿32 0 1 l (0) 0 l R6 24 1.8 21 1.5 113 3 4 .18 5 .25 0 1 l 0 0 ] R7 23 1.2 23 1.4 100 1 4 .17 5 we 0 l l 0 0 1 ! ‘OTU wami us Table 1 (column 1), characters as in Table he TEM 4, A5, G3, and M2 were taken from the biba? collection of the same species (Table 1). The TEM data for AG were taken from the Pires 1281 collection. considerable distortion within Gynotrocheae (Fig. on overall size, tectum thickness, collumellae height, 70). Relationships within this tribe are clarified by and other characters (Table 5). Thus the larger including component III, which accounts for an grains of Anisophylleaceae and Rhizophoreae are additional 13.0% of the total variation (Fig. 71). to the right (Fig. 70), and the smaller Gynotrocheae Component III also reveals marked divergence be- pollen is to the left. Anisophylleaceae and Rhizoph- tween Anisophyllea disticha (A1) and the other orea OTUs separate along component II because Anisophylleaceae OTUs. of differences in sculpture, shape, and other char- Us are distributed across component I based acteristics (Table 5). TABLE 4. Summary of data in Table 3 for Anisophylleaceae and each tribe of Rhizophoraceae.' Characters OTU l 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 AA 26 2.0 20 1.7 133 0 l 05 3 15 0 1 0 0 0 1 GG 14 1.4 12 1.1 119 3 4 .27 2 .19 l 1] 0 O l l MM 17 1.4 17 1.3 104 3 3 4 3 20 0 1 0 0 1 1 RR 20 1.2 20 1.3 100 3 3 16 5 .26 0 1 1 0 0 1 ' AA = Anisophylleaceae, GG = G , MM = Macarisieae, RR = Rhizophoreae. All measurement characters have been averaged. For binary (SEM) characters, the predominant state is given and underlined to pae one or more exceptions within the family or tribe. For multistate characters 6 and 33, the median is give Volume 75, Number 4 Vezey et al. 1375 1988 Rhizophoraceae Pollen TaBLE 3. Continued. Characters 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 0 l 0 0 0 0 1 58 18 15 38 .13 46 14 30 10 0 0 0 0 1 0 0 0 47 28 23 34 .14 38 23 28 11 0 0 0 0 l 0 0 0 68 28 23 43 .12 45 19 28 08 0 0 0 1 0 0 0 0 37 36 19 ot .07 35 34 25 07 0 0 0 0 1 0 0 0 41 46 15 09 .14 37 42 08 13 0 0 0 0 1 0 0 0 71 28 26 98 .18 33 13 46 08 0 0 0 0 0 0 1 0 16 08 11 20 .14 28 14 34 24 0 0 0 0 0 0 1 0 25 05 11 26 .23 32 06 33 29 0 0 0 1 l 0 0 0 27 05 07 14 .13 46 08 24 22 1 0 0 0 0 0 1 0 28 06 14 26 22 34 07 32 27 0 0 0 1 0 0 0 0 16 07 08 12 .13 33 15 25 27 ] 0 0 1 0 0 0 0 36 09 19 51 .08 35 09 49 08 2 0 0 0 1 1 0 0 " 09 10 30 .07 37 12 41 10 2 0 0 0 1 0 0 0 20 11 12 15 .08 .37 20 28 15 0 0 0 1 0 0 0 0 40 07 11 43 .09 .40 07 43 09 2 0 0 1 0 0 0 0 30 08 14 31 .10 38 10 39 13 1 0 0 l 0 0 0 0 30 10 11 30 .07 39 13 39 09 2 0 0 1 1 0 0 0 17 08 08 20 .05 34 16 40 10 2 0 0 l 0 0 0 0 54 19 13 77 .05 35 12 50 03 2 l 0 0 l l 0 0 3l 12 11 37 .01 38 15 46 01 2 l 0 0 l l 0 0 31 11 14 35 .03 39 14 44 04 0 l 0 0 l l 0 0 17 08 11 26 .09 28 13 43 15 0 1 0 0 1 l 0 0 40 12 12 59 .14 32 10 4T 11 0 l 0 1 1 1 0 0 36 13 19 56 .05 33 12 51 05 0 l 0 0 1 1 0 0 35 21 27 57 .03 30 18 49 03 0 l 0 0 1 l 0 0 51 19 22 4] .06 44 16 35 05 0 l 0 1 0 0 0 0 38 14 19 46 .11 35 13 42 10 0 The minimum spanning tree interconnects Gy- but only one OTU (Dactylopetalum sessiliflorum, notrocheae OTUs with two short and two long links M5) links with more than two other OTUs within (Fig. 71). The short connections are Carallia eu- the tribe. The congeneric OTUs Sterigmapetalum genioides (G2) to C. brachiata (G1) and Gyno- heterodoxum (M8) and S. obovatum (M9) are sep- troches axillaris (G4) at distances of 0.630 and arated by a distance of 1.377. Sterigmapetalum 0.636, respectively. Pellacalyx cf. saccardianus obovatum (M9) is actually more similar (0.603) to (G5) joins C. brachiata (G1) and Crossostylis Bruguiera gymnorrhiza (R1) than to any OTU grandiflora (G3) at distances of 1.035 and 1.023. of its own tribe. onnections within Macarisieae average 0.793, Most Rhizophoreae are directly linked to Bru- TABLE 4. Continued. Characters rn -J js ec -— No] N c N — N N N ° N P. h2 = N °. N ~] bo ° N No) [9v] © Ww — Ww bo ° w =l © ° > O Olo III Imiolo!l— I-lo 00 © olm ° ooolo bo N © a — c N © -J w on — © w © N an on ° O 1376 Annals of the Missouri Botanical Garden TABLE 5. Character loadings from principal com- similarity) than those within Rhizophoreae. Even ponents analysis.' discounting A. disticha (A1), links within Aniso- phylleaceae average 0.923, higher than any tribe Direc- Load. of Rhizophoraceae. Al joins A3 from a distance ps i i " of 1.796, the longest link on the minimum spanning + .80-.89 13 tree. Anisophyllea disticha is actually more sim- -70-.79 1,3,26 10,17, 21 16 ilar to three Macarisieae OTUs than to Combre- .60-.69 9, 20, 24, 6,31 12 tocarpus rotundatus (A4), the next closest OTU 50-.59 aa 9 $5. 5d. 35 within Anisophylleaceae. The closer Macarisieae bad diia OTUs are Cassipourea elliptica (M3), Comiphy- — ,80—.89 11,15 ton gabonense (M4), and Dactylopetalum sessi- .70-.79 5 liflorum (M5). .60-.69 8,32 Intertribal links are shorter than many intratrib- 50-.59 6 25, 28 14,18, 23, al connections. Blepharistemma membranifolia 33 (M2), for example, is more similar to R1 (0.702) ' Numbers under each component refer to characters than to either M3 (0.833) or M9 (0.718). Likewise, outlined in Table 2. Only characters with loadings greater — Macarisia ellipticifolia (M6) is more similar to than 0.5 are shown. G5 (0.819) than to either M1 (0.906) or M4 0.850) The phenetic gap between Anisophylleaceae and guiera gymnorrhiza (R1), with two exceptions (R5 Rhizophoraceae is spanned by a link between Com- and R7) which are two links away. The links within bretocarpus rotundatus (A4) and Dactylopetalum Rhizophoreae are the shortest of any tribe, aver- sessiliflorum (M5) at a distance of 0.913. Although aging 0.639. In a similar manner, all Anisophyl- this is greater than intertribal connections within leaceae OTUs are joined to Anisophyllea obtusi- Rhizophoraceae, it is shorter than several links folia (A3) but at much longer distances (lower within Anisophylleaceae. FIGURES 1-12. Scanning electron micrographs of Anisophylleaceae (1-9) and Macarisieae (10-12) p en.— I. Anisophyllea laurina, lateral view, surface punctate.— 2. A. buttneri, polar view.—3. A. obtusifolia. Portion of mesocolpium showing punctate surface.—4. A. disticha, lateral view. This species differs from the others in having a striate surface. —5. Poga oleosa, polar view. —6. salen amazonicus, polar view.— 7, tus. b view. mesocolpia but psilate at the poles and mesocolpial margins. — 1 1. Sterigmapetalum heterodoxum, lateral view, psilate surface. — 12. S. obovatum, portion of a mesocolpium showing a punctate—rugulate surface. Scale bars = ] um. FIGURES 13-23. Scanning electron micrographs of Macarisieae pollen.— 13. Macarisia lanceolata, polar and humbertiana, lateral view. Surface rugulate, becoming smoother near the colpi.— 16. M. pyramidata, polar view.— - Blepha ristemma kir inh lateral view. HH e afzelii, lateral view, surface p 9 guianensis, lateral view. The surface is punctate but n Figure 18.— 20. actylopotahim sessiliflorum, lateral view.—21. Cas ssipourea gummiflora var. verticillata, polar view.—22. Anopyxis kleineana, lateral view. The surface is psilate; e mu of the large endoaperture is visible. — 23. A. ealeaensis, lateral view. Scale ars = 1 um unless otherwise indicated. FIGURES 24-35. Scanning electron micrographs of Rhizophoreae bie 29) and Gynotrocheae (30-35) pollen.—24. Rhizophora mangle, lateral view, surface punctate—rugulate — 25. R. mucronata, lateral view.— 206. R. stylosa, lateral view.— 27. Ceriops tagal, polar view.— 28. Bruguiera pee za, lateral view.—29. Kandelia candel, lateral view. Note the outline of the endoaperture.—30. Gynotroches axillaris, lateral view. The colpus membrane and the vi e of the mesocolpia have a granular-spinulate surface; the rest of the grain has psilate-punctate surface Crossostylis grandiflora, lateral view.—32. Pellacalyx cf. saccardianus, lateral view.—33. Carallia Laas polar view. — 34. C. eugenioides, lateral view. The surface is psilate— punctate with ped us spinulate mesocolpial margins and colpial munda — 35. C. brachiata, lateral view. Similar to Figure 34 but fewer granules—spinules. Scale bars = Volume 75, Number 4 1988 ezey et al. Rhizophoraceae Pollen - www d a. 18 0 ek AN. REESE A à k q Meat, i M Annals of the Missouri Botanical Garden Volume 75, Number 4 1988 ezey et al. Rhizophoraceae Pollen 1380 Annals of the Missouri Botanical Garden FIGURES 36-43. Transmission electron micrographs of Anisophylleaceae pollen.—36. Anisophyllea disticha. In the center of ded mesoc olpium the columellae layer is undulating and the corresponding foot layer shows "hills" and "valleys." Toward the colpi the foot layer becomes thin and the columellae are straight. The outer margin of the tectum » appears lobed due to perpendicularly sectioned striae. —37. Combretocarpus rotundatus. Section of a mesocolpium near an endoaperture (arrow). The tectum is incomplete, the columellae simple, the foot layer Nibh and the endexine disrupted. — 38, 39. Anisophyllea obtusifolia. The undulating columellae (as in Fig. 38) part of the mesocolpium also has straight columellae (as in Fig. 39, to the lefi) .—40. A. fallax. The tectum is thick and incomplete, the columellae n and short, and the foot layer and indi uniform.—41. A. laurina. The columellae are well bi e — 42. Polygonanthus amazonicus. The foot layer is thick and tapering toward the endoaperture (left) .— 3. Poga dc The tectum is incomplete and thick; the columellae are tall, becomin gra pa distally; the foot pel is thin; and the endexine is thicker than the foot layer and is uniform. Scale = 1 um. Volume 75, Number 4 Vezey et al. 1381 1988 Rhizophoraceae Pollen ` &%w < Ey M, is Es a; a FIGURES 44-53. Transmission electron eb. Au of Tip deed (44—49) and Macarisieae (50—53) pollen. —44. Carallia brachiata and 45. C. e ioides. In ae the tectum is thin, the columellae are short and unbranched, the foot layer is uneven in Peep and the endexine is via but has long, narrow gaps in Figure 44.— 40. esee grandiflora. A thin granular bo s present just below the tectum. The endexine is thicker and has irregular open spaces near a colpus (lefi) .—47. Gynotroches axillaris. The exine is similar to that in Carallia. — 48. Pellacalyx pustulata. The tectum is thin, the short columellae are extensively branched, and the foot layer is thin and irregular. A line separates the foot pii from the endexine. The al becomes very thick near the colpus, where it has a large, irregular gap.—49. P. cf. saccardianus. The exine is similar 1382 Annals of the Missouri Botanical Garden DISCUSSION milial status for Anisophylleaceae. On the other Lack of endoaperture fusion is the only feature of Anisophylleaceae pollen that does not overlap variation within Rhizophoraceae. Therefore, the phenetic gap between the two families results from the combined effects of many characters. With LM, for example, Anisophylleaceae OTUs have the highest or lowest average (or median) in 8 of 10 characters (1, 2, 4-8, and 10) and the same high average as Rhizophoreae for character 3. Aniso- phylleaceae OTUs also have the highest or lowest average for six TEM characters (24-26, 29-31) and the second highest average for characters 27 and 28. Also with TEM, Anisophylleaceae pollen can be separated from Gynotrocheae by a muc lower endexine thickness ratio, and from Macari- sieae by lack of intercolumellar granulation. SEM analysis reveals the punctate-only sculp- ture of four Anisophylleaceae OTUs. Combreto- carpus rotundatus (A4) differs slightly by having psilate mesocolpial margins. The striate sculpture of Anisophyllea disticha (A1) is strikingly differ- ent from any other in this study and is primarily responsible for isolating A. disticha along com- ponent III. In contrast, punctate pollen within Rhi- zophoraceae is usually psilate-punctate or punc- tate-rugulate. Exceptions include Cassipourea elliptica (M3), which is punctate-only, and Dac- tylopetalum sessiliflorum (M5), which has the same SEM character states as C. rotundatus (A4). The similarity in exine sculpture between D. sessiliflo- rum and C. rotundatus partially accounts for their connection on the minimum spanning tree. Pollen of Anisophylleaceae can therefore be dis- tinguished from Rhizophoraceae by the following combination of character states: larger polar axis and breadth, greater variation in polar axis and breadth, higher P/E ratio (in the subprolate-pro- late range), narrow or nonexistent endoapertures, lack of endoaperture fusion, relatively small polar area (including some syncolpate grains), punctate sculpture, thicker tectum, taller and wider colu- mellae, the highest tectum thickness and columel- lae height ratios, the lowest foot layer ratio, a low endexine ratio, and absence of intercolumellar granulation. These results agree with separate fa- and, Anisophylleaceae, even without Anisophyl- lea disticha (Al), is a relatively variable taxon with longer average links on the minimum spanning tree than the tribes of Rhizophoraceae. Pollen of Gynotrocheae, Macarisieae, and Rhi- zophoreae can be generally characterized, but these tribes cannot be separated based on palynological evidence. Characteristics generally distinguishing pollen grains of Rhizophoreae from those of carisieae and Gynotrocheae are larger size, lower P/E ratios (averaging 100), greater distances be- tween colpal ends, higher polar area indexes, punc- tate-rugulate sculpture, thicker tecta, taller and wider columellae, and thicker foot layers. At the opposite end of the continuum are Gynotrocheae with the smallest grains, highest P/E ratios (av- erage 119), psilate-punctate sculpture, highest en- doapertural indexes, lowest foot layer ratios, and highest endexine ratios. Results of the phenetic analysis are not consistent with the proposed sep- aration of Crossostylis grandiflora (G3) from Gy- notrocheae (see other symposium papers, this vol- ume). In both the ordination and minimum span- ning tree, C. grandiflora is located within the group of Gynotrocheae O' Macarisieae OTUs are widely distributed in phe- netic space, partly because of the many exceptions to the predominant SEM character states. There is essentially no phenetic gap between Macarisieae and either Rhizophoreae or Gynotrocheae. Macar- isieae is also most similar to Anisophylleaceae. In many measurements and ratios, the Macarisieae average is between Rhizophoreae and Gynotro- cheae, including polar axis and breadth, P/E ratio, distance between colpal ends, polar area index, tectum and foot layer thickness, height and width of columellae, columellar height ratio, and foot layer and endexine thickness ratios. Macarisieae ollen is also characterized by the presence of intercolumellar granulation. Only two OTUs out- side Macarisieae, Crossostylis grandiflora (G3) and Pellacalyx cf. saccardianus (G5), have this feature, and only one Macarisieae OTU, Cassi- pourea elliptica (M3), lacks granulation. to that in Figure 48 but no line is evident. —50. Anopyxis ealeaensis. The narrow columellae layer has a granular ick. l. A. e ex matrix. The foo S. heterodoxu > kleineana. the endexine is highly ahs iri d Turic iia Mr j | m. The distinct granular layer is present below the tectum. ine is similar to that of the previous species, but The granular matrix is well a and The foot layer is very thick; de . is thin and disrupted in the mesocolpium but is well iiis below the colpi. Scale bars — 1 um Volume 75, Number 4 Vezey et al. 1383 1988 Rhizophoraceae Pollen TW C. py $ a.C. TE FIGURES 54-63. Transmission electron micrographs A Sae pollen. —54. Macarisia it ie ta. Middl part of a mesocolpium. Note that the intercolumellar spaces are filled with a granular matrix iie elipticifolia Mesocolpium near a colpus (on the right).—56. M. humbertiana. xs layer quite thick, ume e granu due to proximity of circular endoaperture.—57. M. pyramidata. /n 55, 56, and 57 the granular | im iberia) is not as extensive as in 54.—58. Cassipourea elliptica. An oblique section showing part of a mesocolpium near a colpus (to the right). Here the endexine is greatly thickened and is g igi r on the lefi due to the proximity of an endoaperture.— 59. Blepharistemma membranifolia. Granular ains ps es endexine thin or absent except near the colpus, where it is very thick (to the right) .—60. Dactylopetalum A oram. Middle part of a mesocolpium; a thin t layer below the tectum.—61. enkeri. /ncomplete tectum suggests a reticulate or meshlike surface. — omiphyton gabonense. Columellae 2r is narrow and has some granules below the tectum.—63. ela obovata. Exine similar to those in 55, 56, and 57. Scale bars = 1 um. 1384 Annals of the Missouri Botanical Garden FIGURES 64-69. (9947), section of an entire mesocolpium endoapertures (arrows) the thick.—66. R. mangle. Tec isa incomple circular endoaperture.—67. Kandelia candel. Note distally.—68. Transmission electron micrographs of Rhizophoreae pollen.—64. Rhizophora mucronata ndexi n ete in some areas; the thick granular Rose is probably due to the e massive horse Colum ruguiera gymnorrhiza. Á few la similar the intercolumellar spaces. The thin endexine is difficult to distinguish — 69. Ceriops tagal. Tectum inc omplate, foot layer very thick, and endexine thin and disrupted. Scale bar um. CONCLUSION Principal components analysis using palynologi- cal data clearly separates Anisophylleaceae from Rhizophoraceae and supports the hypothesis of sep- arate familial status for Anisophyllaceae. If Ma- carisieae pollen data were not considered, Rhizoph- oreae and Gynotrocheae OTUs would also form discrete phenetic groups. Including Macarisieae pollen data changes this picture to one of contin- uous phenetic variation from Rhizophoreae through Macarisieae to Gynotrocheae. The majority of pollen morphological charac- teristics in Anisophylleaceae and Rhizophoraceae occur in a broad range of families throughout the angiosperms. Therefore, it was not possible within the limits of this study to suggest relationships with other taxa. It seems particularly significant that neither family we investigated can be connected to Myrtales on palynological grounds. Light and ultrastructural data on Myrtales pollen (Patel et al., 1985) are generally comparable to data in this study. The colporoidate or fused endoapertures possessed by Anisophylleaceae or Rhizophoraceae have no counterpart in Myrtales, however, and the pseudocolpi of Myrtales pollen do not correspond to features of either Anisophylleaceae or Rhizopho- raceae pollen. Volume 75, Number 4 1988 Vezey et al. Rhizophoraceae Pollen 1385 0.8 R2 R4 0.4 7] — 007 c ° c o 4 oa E A6 Oo O -047 7| o Anisophylleaceae Q Gynotrocheae -0.8 7] © Macarisieae A2 =| 9 Rhizophoreae -1.2 T T T T T T T T T -1.2 -0.8 -0.4 0.0 0.4 0.8 Component | E 70. total variation explained tree was calculated from an average taxonomic distance matrix or componen Projection of 27 OTUs onto the first two principal components (data from Table 3). Percent of is 25.5 for component I and 20.7 t II. The superimposed minimum spanning 0.8 2 A5 Ó,. 04 - - e O R2 E e A3 A4 R3 e. = M2 R5 A6 = 00 y m7 M5 O e R1 R6 e t OO w 79 9 ° @ = 4 O R4 o M4 E O O -0.4 7 M1 P o 7 e Anisophylleaceae Gynotrocheae bon "T © Macarisieae J $ 3 Rhizophoreae -1.2 T T T T T T T T T T -1.2 -0.8 -0.4 0.0 0.4 0.8 Component | FicURE 71. — Projection of same 27 OTUs (Fig. 70) onto principal components | and III. Percent of variation expressed by component III is 13.0. A partial minimum spanning tree has been included to clarify relationships within Gynotrocheae. 1386 Annals of the Missouri Botanical Garden LITERATURE CITED iii c P. 1969. Fossil pollen of Rhizophora at que (Basso Valley of the Senegal). Pollen & Spores id. R. 1983. Pollen from four common New World mangroves in Jamaica. Grana 22: 147-151 DuNN, G. & B. TT. 1982. An Introduction to Mathematical Taxonomy. Cambridge Univ. Press, or Piya, G. 1952. Pollen Morphology and Plant Tax- omy. Angiosperms. Almqvist and Wiksell, Stock- 960. The acetolysis method. A revised de- scription. Svensk. Bot. Tidskr. 39: 561-564. FAEGRI, K. & J. IvVERSEN. 1975. Textbook of Pollen heap: Blackwell Scientific Publications, London GEH, S. Y. & H. KENG orphological ziudiss of some inland Rhizophoraceae. Gard. Bull. Straits Settlem. 27: 183-220. GOWER, J. C. 1966. Some distance properties of latent Biometrika 53: 325-3 Guers, J. Combreiacéne, Lecythidaceae, Ly- thraceae, Melastomataceae, Myrtaceae, Rhizophor- aceae. In: Pollen et Spores d'Afrique Tropicale. As- sociation des Palynologues de Langue Frangaise. Pollen grains of Formosan plants (4). Taiwania 14: 133-270. KuBiTZKI, K. 5. Palynologia Madagassicas et Mas- ios Im 147-154: hraceae-Oenothera- 7. Koro. L A. Pollen and spores of West African mangroves. Dokl. Akad. Nauk SSSR 129: 430. 428- LANGENHEIM, J. H., J. HACKNER & A. BARTLETT. 1967. Mangrove pollen at the depositional site of Oligo- Miocene amber from Chiapas, Mexico. Bot. Mus Leafl. 21: vip LupLow- WIECHER J. L. ALVARADO. 1983. Pal. 11. Rhizophoraceae family. Biotica 8: 7-1 MULLER, J. € C. CARATINI. 1977. Pollen of Ricoh: ra (Rhizophoraceae) as a guide fossil. Pollen & Spores 19: 361-389. PaTEL, V. C., J. J. SkvarLa & P. H. Raven. 1985. Pollen characters in relation to the delimitation of Myrtales. Ann. Missouri Bot. Gard. 71: 858-969 Rakosi, L. 1978. Palynologic data from the Eocene mangrove swamps of Hungary. Magyar Allami Foldt. Intéz. Évk. 1976: 357-374. RonuLr, F. J., J. KisiPAUGH & D. Kir 982. merical Taxonomy System of Mulivavata Statistical Programs (NT-SYS). The State University of New York at Stony Brook, New Yor . Techniques ‘of pollen electron microscopy. Part I. Staining, dehydration, and embedding. eee: Geol. Notes 26: 179-186. SNEATH, P. H. A. & R. R. SokaL. 73. Numerical Taxonomy. Y. H. Freeman and Company, San Fran cisco. SowUNMI, M. A. 1974. Pollen grains of Nigerian plants Woody a Grana 13: 145-186. 1981. Late Quaternary environmental changes in Nigeria. Pollen & Spores 23: 125-148. STRAKA, H. & B. FRIEDRICH. agp: Palynologia Mad- agassica et Mascarenia. Fam. 147-154. Microscopie electronique à balayage et rsen Trop. Subtrop. Pflanzenwelt 51: 58-60. THANIKAIMONI, G. 1972. Index Bibliographique sur la Morphologie des Pollens d'Angiospermes. € t. Franç. — Trav. Sect. Sci. Tech. 12: 1973. Index Bibliographique sur s Mor- gliologie des Pollens d'Angiospermes. Supplement I. n. ins Pondichéry, Trav. Sect. Sci. Tech. 12: ——., Index Bibliographique sur la M nos p Pollen ns d'Angiospermes. Supplement 1. Inst. Franç. Pondichéry, Trav. Sect. Sci. Tech. 13 dia Quatriéme Index Bibliographique sur la Mae des Pollens d'Angiospermes. Inst. Franc. Pondichéry, Trav. Sect. Sci. Tech. 17: 1- 338 . 1986a. Pollen n form and function. Pp. 119-135 in S. Blackmore & I. K. Ferguson (editors), Pollen and e [^ and Function. Academic Press, London. Cinque Index Bibliographique sur la Morphologie des Pollens d' Angiospermes. m t. Franc. Pondichéry, Trav. Sect. Sci. Tech. 22: "203. 1987. Mangrove Palynology. nt xu ondichéry, Trav. Sec i. Tech. 24: 1979. a. j evolution de i man- groves du Tamil Nadu (Inde). ag oe de la 3éme Sect. E.P.H.E. Montpellier: 1 SIEVE-ELEMENT PLASTIDS H.-Dietmar Behnke? AND SYSTEMATIC RELATIONSHIPS OF RHIZOPHORACEAE, ANISOPHYLLEACEAE, AND ALLIED GROUPS! ABSTRACT One hundred fifty-five species | iss families. inira icd to the proposed ordinal composition around and including the families Rhizophoraceae and Anisophylleaceae have been studied with respect to their sieve- element plastids. The great male “of taxa, including the Anisophylleaceae, contain S-type plastids. P-type sieve-element plastids were y aene in Humiriaceae, Rhiz ophoracese, and Ery th roxylaceae (all with yeu raceae (P5c) is proposed to be para alleled by another P-type containing sequence Lepidobotryaceae (S-type) — Hypseocharitaceae (S) -Oxalidaceae (S, Pc) -Averrhoaceae (Pcfs) , both being linked to the S-type Linaceae s.l. Sieve-element data do not support the inclusion of hizophoraceae in the Celastrales; however, such data corroborate the exclusion of the new celastralean family Elaeocarpaceae from the Malvales. Among the taxa PI by Dahlgren, Anisophylleaceae would be best placed in iia to the S-typ e families of the Rosales, ot in close association to P-type Neuradaceae. The presence i type sieve-element plastids in Zygophyllaceae, Neuradaceae, and Humiriaceae is reported here for the first ti ! This investigation was ipm possible only by the unselfish ict help of my colleagues who collected and sent fresh or fixed samples from various areas of the wor rld—V. Armer (Kuwait), J. Aronson (Beer-Sheva, Israel), H. Balslev (Aarhus, Denmark), K. m p (Chicago, Illinois, U.S.A.) , P. Berry ro Venezuela), D. E. Boufford (Pittsburgh, Pennsylvan U.S.A.), P. Chai (Kuching, Malaysia), A. C. Gibso iie i Arizona, U.S.A.), G. & I. S. pedi iu (Giessen F.R.G.), M. H. Grayum (San José, Costa Rica), I. A. U. N. Gunatilleke (Peradeniya, Sri Lanka) , C. Huckins Dm Arizona, U.S.A.) , S. B. Jones ( Athens, e Ph .S.A.), A. Juncosa (Petersham, ddr dien U.S , W. J. Kress ates North Carolina, U.S.A.) , K. .G.) , C. Marticorena SpA Chile), G. J. Muller (Cape Ferguson, Australia) , B. Nelson (Manaus, Brazil), G. Persch (Bonn, F.R.G.), L. J. Poveda (San p Costa Rica) , G. T. Prance (Bronx, New York, U.S.A.), J. S. Pringle (Hamilton, m Canada), A. E. Radford (Chari Hill, North Carolina, U.S.A.), A. Schaltz (Porto HB. Brazil), W. Thomas (Kumba, Cameroon), F. dst Ëq (Claremont, California, U.S.A.) , R. Tra a A Australia) , E. A. Ulrich (Mexico Lu: oolliams (Hal a = al I = [2] = — ~ ~ a 3 o E m x E co (abbreviations given under e . Ferre d quara S.A.) , P. Hind (Sydney, Australia), J. Jacobs (Sydney, yum alia) , J. L. S. Keesing (Kew Gardens, K. Kramer (Bonn, F.R.G.), Lee Ying Fah pu Malaysia) , W. Stubbe (Düsseldorf, F.R.G.) , 3 wm (Bogor, Indonesia). P. H. Raven (St. Louis, Missouri, U.S.A.) gave invaluable support in finding sources and collectors for fresh material. Ute Kiritsis her W. Siller helped with collections and fixations durin the author's stay at Bogor (Indonesia) ; Marianne von der Decken, Doris Laupp, and Liliana Pop (all Heidelberg) processed the material on its long way from the fixation to the electron microgra aph. The Deutsche Forschungs- q raqa gave continued travel and research support for these investigations. My sincere gratitude is extended all the aforementioned persons and institutions without whose help this research would not have been possible. ? Zellenlehre, Universitat Heidelberg, Im Neuenheimer Feld 230, D-6900 Heidelberg, Federal Republic of Germany. ANN. MISSOURI Bor. Garb. 75: 1387-1409. 1988. 1388 Annals of the Missouri Botanical Garden The family Rhizophoraceae is distinct from other dicotyledon taxa by the formation of rather ex- traordinary P-type sieve-element plastids (Behnke, 1982a). Their specific form-P5c plastids contain some twenty more or less rectangular protein crys- tals—i.e., an accumulation of proteins to a degree found nowhere else in P-type plastids—and were originally reported for seven species of the family and an additional four species of the Erythroxy- laceae. A closely related form-P5cf (containing pro- tein filaments in addition to the crystals) was found in the family Cyrillaceae (Cliftonia and Cyrilla). These unique subtype-P5 plastids raised questions about the systematic position of the three families (Behnke, 1982a) and initiated further research on sieve-element plastids and other characters. A first study of the distribution of types of sieve- element plastids of Myrtales and allied groups (an association of taxa into which the family Rhizopho- raceae had been placed most commonly) revealed that (1) all core families of the Myrtales and all of those closely related contained S-type plastids, and (2) within the Rhizophoraceae (an additional seven species were investigated) the two genera Aniso- phyllea and Combretocarpus also contained S-type plastids (Behnke, 1984). This gave support to var- ious efforts to separate the tribe Anisophylleeae from the Rhizophoraceae and to erect the family Anisophylleaceae (Cronquist, 1981; Dahlgren, 1983; Tobe & Raven, 1987). The present additional report on sieve-element plastids in Rhizophoraceae, Anisophylleaceae, and allies is an extension of the previous investigations taking also into account all the higher taxa, i.e., ordinal compositions and their associates, to which the two families have been affiliated (see Dahlgren, this volume). MATERIALS AND METHODS One hundred fifty-five species of 41 families, all proposd by Dahlgren (this volume) for placement around Rhizophoraceae and Anisophylleaceae were investigated (see Table 1). Living material recently removed from the plant or shipped within a few days under special care is a prerequisite for a fixation of sieve elements and the eventual investigation of their plastids with the transmission electron microscope. Thin hand sec- tions were made with a razor blade from preferably young herbaceous shoots or end parts of tree branches less than 1 cm in diameter. The sections were immersed into a fixing solution containing formaldehyde and glutaraldehyde and processed according to standard methods (see Behnke, 1982b). Material made available by collections at original locations was sent to Heidelberg either fresh (caus- ing a delay of up to a week between sampling and start of fixation) or as formaldehyde/glutaralde- hyde prefixed hand sections (causing an equally long delay between primary and postfixation). RESULTS A SHORT OUTLINE OF CHARACTERS OF SIEVE-ELEMENT PLASTIDS USED TO CHARACTERIZE THE TAXA INVESTIGATED Sieve-element plastids are separated into two types by presence (P-type) or absence (S-type) of protein crystals and/or filaments, while starch grains may or may not be present. Subtypes of P-type sieve-element plastids are identified by any unmis- takable feature of their protein inclusions, e.g., the subtype-P5 by a high number of generally rect- angular protein crystals. Forms of sieve-element plastids are defined by any combination of the three inclusions: c — protein crystals, f — protein fila- ments, s — starch grains, e.g., P5cf. In addition, all sieve-element plastids within a family will be characterized by their average diameter and av- erage amount of protein vs. starch content (Table 2), both calculated from the respective data of the different species listed in Table 1. Recent studies of the sieve-element plastids of the Acanthaceae (Behnke, 1986a) and within all families of the Magnoliidae (Behnke, 1988)—the latter for the first time taking into account diam- eters and quantitative data of the plastids—re- sulted in a general model for the interrelationships between the different forms of plastids. It was con- cluded that at least in these groups, P-type plastids may have derived from S-type plastids (for details see Behnke, 1988, but compare with Behnke, 1981). In his summary statement of Rhizophoraceae and Anisophylleaceae and their systematic rela- tionships, Dahlgren (this volume) proposes a revised ] iG ti d hete th nu ] ots, d each of the two families. The following description of the sieve-element plastids in these taxa follows his sequence of families. DISTRIBUTION OF THE DIFFERENT SIEVE-ELEMENT PLASTIDS AMONG THE FAMILIES GROUPED AROUND RHIZOPHORACEAE Zygophyllaceae (ZYG; Fig. 1: Guaiacum, Lar- rea). Five species in four genera investigated, one with P-type, the others with S-type plastids. Plastid diameter is 1.2 um. Larrea divaricata contains form-Pcs sieve-element plastids with two protein crystals of different diameters (0.4 and 0.3 um) and different crystal spacing. There are about Volume 75, Number 4 1988 Behnke 1389 Sieve-Element Plastids five typically disc-shaped starch grains in addition. The S-type plastids of other species studied contain up to ten starch grains of different diameters and shapes, including typically disc-shaped ones. In this family the plastids of the mature sieve element are often disrupted, making it impossible to record some protein crystals. Therefore, after the detection of P-type plastids in both collections of Larrea (Table 1), all species have been studied once more. Nitrariaceae (NIT; Fig. 1: Nitraria). This monogeneric family, represented here by Nitraria retusa, contains S-type plastids with characteristics slightly different from those of ZYG, i.e., with a diameter of 1.0 um and about five more or less globular starch grains. Peganaceae (PEG; Fig. 1: Peganum). Pegan- um harmala likewise contains S-type plastids with a diameter of 1.2 um and about five starch grains. Balanitaceae (BLT; Fig. 1: Balanites). Two species investigated in the monogeneric family both contain S-type plastids with about ten globular starch grains. Plastid diameters are 1.7 um in one and 1.1 um in the other (Table 1). Vivianiaceae (VIV; Fig. 1: Caesarea). Caesa- rea albiflora contains S-type plastids with up to five typically disc-shaped starch grains. Plastid di- ameter is 1.1 um. (See also Behnke & Mabry, 1977.) Geraniaceae (GER; Fig. 1: Pelargonium). Five species in three genera investigated, all with S-type plastids. There are about five disc-shaped starch grains within a plastid, the average diameter of which is 1.2 um (range 1.0-1.6 um). (See also Behnke & Mabry, 1977.) Ledocarpaceae (LDC; Fig. 1: Wendtia). Wend- tia gracilis contains small S-type plastids (diameter 0.9 um) with a few starch grains of variable sizes. The plastids appear distinctly different from those of the Geraniaceae. Ixonanthaceae (IXO; Fig. 2: Ixonanthes). The two species of [xonanthes investigated contain S-type plastids of an average diameter of 1.1 um and with an average of ten small and large, more or less globular starch grains. Humiriaceae (HOU; Fig. 2: Humiria, Endo- pleura, Sacoglottis). Four species in three gen- era investigated: all with P5cs sieve-element plas- tids. This form is characterized by numerous (average more than ten) irregular to rectangular protein crystals and about ten small starch grains. The average measurements are: plastid diameter 2 um and protein crystals 0.3 um. Hugoniaceae (HUG; Fig. 2: Indorou- chera). The two species investigated represent two genera, and both contain S-type plastids with about ten starch grains, among them one or two large ones. The average plastid diameter is 1.1 um. Erythroxylaceae (ERX; Fig. 2: Erythroxy- lum). e four species of Erythroxylum inves- tigated are characterized by P5c sieve-element plastids. This highly specific form contains about ten up to 0.7 um large rectangular protein crystals (and no starch), which are densely packed within the comparatively small plastids (diameter 1.1 um). (See also Behnke, 1982a.) Linaceae (LIN; Fig. 2: Linum). Three species in two genera investigated, all containing S-type plastids with an average of about five starch grains (often including a large globular one), which may disintegrate into small particles. The average plas- tid diameter is 1.2 um (range 1.0-1.5 um). Lepidobotryaceae (LPB; Fig. 2: Lepidobot- rys). The monotypic Lepidobotrys staudtii con- tains S-type plastids with up to ten starch grains disintegrated into small particles. The plastid di- ameter is 1.2 um. Oxalidaceae (OXL; Fig. 3: Averrhoa, Oxalis, Sarcotheca). Twelve species in five genera in- vestigated. Sieve-element plastids not uniform: two types including three different forms occur in the family. Averrhoa and Sarcotheca contain form-Pcfs plastids with an average diameter of 1.1 um. This P-form, not specified to belong to a distinct P-sub- type, contains protein filaments (f), two rectangular or cubic protein crystals (c) up to 0.4 um in di- ameter, and about five globular starch grains (s), of which one may be very large. Oxalis (8 species tested) is characterized by very small (average diameter 0.8 um) form-Pc sieve- element plastids with two protein crystals, a very prominent (diameter about 0.5 um) cubical one and a second, smaller one. Biophytum and Hypseocharis contain S-type plastids of different sizes and different starch con- tents (see Table 1 and Behnke, 1982c). Form-Pcfs and -Pc sieve-element plastids (and/ or -Pcs not found in the Oxalidaceae) are reported in other dicots, e.g., in Vitaceae, Rhabdodendra- ceae, Connaraceae, Eucryphiaceae, Acanthaceae, Gunneraceae (Behnke 1974, 1976a, 1982c, 1985, 1986a, b), and Neuradaceae (see below). Missouri Botanical Garden Annals of the 1390 (F861) 0 LO OI TI 96d 12 MQNY C]HWOOH TH 1S CV unyod aun] Axo1yi4a 62 XA UOJAY (Sti10]A) (FB6D 0 S0 OI €I 96d ariH ariH-9g asuajpunad-oaou UNJAXOLY IAI 6 XW (F86D 0 90 Ol TI 96d dIHH ariH-98 "ure 7202 wNjAXOIY IAI 6t; XW OI 0 0 TI S GIH 09:94 PH `H %uviy/f43 dsayonosopul NIICTZ OAH OI 0 0 OT S GIH aajpipun») *exueT us "pxeru sruod8ng NIIT Stc SNH Z992 g cø OI €I socdg GrH ‘OW wNA DIO) HW ry iso) "nen? nuXgoyo1 $110]8020g OFZ NOH 8 ZO 9 60 scd GIH ‘OW GOST uos]aw ^g ‘zeig ds sijj0]2o2ng 9f; NOH G £0 07 €I sgg ariH uos]aw ^g ‘zeig 19/MUDS]DG viuumg] 9f; OH Sl £0 OI ZI Sd ON 99S] uosjow `g ‘zeig ds »1na]dopu 96% NOH OI 0 0 OT S GIH 7220£8 227 Y `g'q'u “PPS “Yoel n70]n21124 səyyupuox] 8bz OXI OI 0 0 €I S CIAH 09-94 “IYO p2ojfipup48 soyjupuox] 8b OXI z 0 0 TI S qH uD1]211002) “YSS(] *eueurry ‘Od »pyimu pu21512022g crc sgg F9IZ np) £ 0 0 60 S ONOO [ Y 22NFUPOY “Y “AUD USÁS]A X1)19D18 DIJPUO y Iz Odi "WY xo “119H 7] (2261) 8 0 0 91 S dIaH ariH-o8 CFT) unuogvaj01 ummios1p]ag 62 Ma S 0 0 ZI S GIH ariH-o8 Up wnauingups wuss 6€o MW G 0 0 TI S dIdH ariH-9g “AR UNILIUI) UNIUDAIL) 6£7 MAD S 0 0 TI S ariH GIriH-98 '8907) NADIsauDW umpo47 6£2 Wad G 0 0 Ol S ariH Z0F0£98 `g'q H ‘APY `PHLAN $9p10Yy 90D umipo23 672 Wat (2261) S 0 0 TI S ariH zjmyjog "p “Ze Ig `ssəqureO D10]/1Q/D DILDSID:) Obs AIA OI 0 0 TI S dD2-58 sisuajosun SINUD]DE tho lH OI 0 0 ZI S CIAH T58 aeq n2n1d452p sapupjng PPS 119 G 0 0 ZI S 3-58 "1 D9]DULDY wnundaq L£2 dd 9 0 0 O'I S uosuo4p T *[9 91s] 's19Q5sy Dsni3l DUDIN BEC LIN G 0 0 €I S dH ariH-og ‘ds un Aydo3x7 98% AZ 8 0 0 31 S d0-98 Aed Y zm Duo 01121404] 9€6 DAZ S 0O 2 TI səd qIHH GI3H-98 “PHN ‘xIW AB) D)D211DA1D D241U'] 9z — 9AZ 8 0 0 €l S (IHH 'q-H9 GIH: “8-99 T Pujo umnovin:) 9€ DAZ 8 0 0 ZI S (IHH anH-o8 ÁRIN "V Mayno 1UNIDIDNI) 9€% AZ un NS VWd N via *díL umtieq1oH IPHI jo uiu) satoads 6861.0 mue, -e3nqng u123104d fo 1212umip umunxvu = FWd *si»is&ao u12104d fo 1oquinu = wni10942H : Apurpf,, woaf quasaffip fr wKuo420 Apruvf pup. (ssaad un) usihyog ur Apruvf fo 12qumu = 12q2, 199f0 swXuossn (mu * (aumjoa snj1) uso Aq paziuSoo24 sauf = Apu ‘saizads porogijsoaui ay) fo vip puspjd juauaja-aaaig Zuwmopof ayy *s1212noa fo nsodap = 4011np ay? fo sy10m snoia24d 07 siafas | uoioi]qnq,, suma q21vis fo 12quinu aS12ap = NS “ur ui sqpis&12 N “ur u 4212umip = Y] : (uaof pup) addy :spnsojd 1u2u2]2-2a21$ ay) 221421201042 DIDP 6861.0 * (£861) "I TY L Behnke 1391 Volume 75, Number 4 1988 Sieve-Element Plastids S 0 O rl S ggs- ‘[puy 241u202D4Kd v1321142111d 9cz T32 S 0 0 OI S GIAH xg ‘dye A. snippidsnoi sníjs0]220421d ocz TID quis '9 OI 0 0 +'I S an ‘ALog 12219Q5110%) `Ç `] ‘zeg “Y (`ssəqure9) s11959duivo Dsso M3 7 osz TAD 0 0 0 Ol S ariH d-98 APM `M tpapuona vjj2112ussao'] 98% TAD G 0 0 9I S arHog "MG DSOUO) n3jDn120ddig] 9SZ T49 c1 0 0 YI S IAH uosjoy ^g ‘zeg ‘ds nidnoy 98% TD "ZZeWN OI 0 0 91 S dIAH dIAH:9q4 “Pue (zoun]) 12un310f snuXuon;] 9cz TD S 0 0 Ul S MOG Ig unGdoiəpus uoipuəpoən];] 9cz; TAD snjnjound S 0 0 I S ariH AIAH-94 "IBA "qunq[ $NID]NI1QUO $N1I8D]9) 9c; TD OI 0 0 €I S ariH-og "1 smyofixng snajspja?) 9e; TD S 0 0 €I S ariH-28 "48104 SIMP DYW) 9e; TAD S 0 0 TI S GIH Saq-58 "110 | DYJUDID]OY DIJOUD/) 9cz TD S eo c Cl spd IAH ZZ20£8 227 X "H'G'H ‘PPS IEH `H 9170/181901p 022110240$ 1Sz TXO 0 SO Z 60 9d ariH-og phoaqieg sisuoTa:p]oa si]px() IS; TXO (973861) 0 Yo c 80 od ariH-od "T 97201418. SIX) Ist XO (93861) 0 SO c 80 9d ord ‘boef D1Jof1iquoya sijpx() IZ TXO (?2286D 0 S0 z 01 od 9445-94 ‘uneg n9jun218 s1]DxO IS; TXO (973861) 0 SO c 80 9d ariH aniH-od 199Mg Xo "ppoT 1addop si]px() 167 TXO (93861) 0 SO c 80 od ariH ariH-9g boef DIDUILI SDX() 1st TXO (92860) 0 c0 cz 80 od ariH ariH-od JOW »sou4no s1]DxO 1S7 TXO (2860 0 ro c 80 ad qH £€IF062 “UH '33əqdləptəH "T 2112801290 s1]DxO 1st TXO (973861) SI 0 0 €I S d298 “uw nofñəudund suooəsdXH ISZ TXO (03860) € 0 0 80 S CIAH NNOY" 'IT3H-28 “Od CT wnansuas wnjidydorg IS; TXO (93861) € vo z Ol spd dIaH dIAH-94 "[ D]OQUIDIDI DOYLLIAY Isc TXO EFA A UNIDA] OI 0 0 zI S GIH ‘OW "p SDUWOY ‘q ‘Noorweg ‘Bug mipnvis sXsjogopidaT 0Sz è dT eleH (uog ‘q x? G 0 0 S'I S GariH WO ‘uep yong) DqOun212 DIJpADNU1d y cp NIT 8 0 0 CI S GariH ariH-od ys unapy nur] S+ NIT € 0 0 01 S ariH-og 'qiog um21;1urojop umur] SZ NIT (+861) 0 vo OI TI Gd gan 'aLog 42842081105) `ç "] zeg TH IS ^V wnsosaqns um]Axo1443] 6 Xua uou NS VWd N via sd umireq1oH PPH Jo utu) satoadg 6861.1 4puey -eonqng enuuguo) “| XI8V[ Annals of the 1392 Missouri Botanical Garden OI 0 0 ZI S Y) DS[EE Dpaaog ‘LIY PISO) "nen? DUDÁS]PUDIS pi]ounig Sol TN8 OI 0 0 LT S 60539 ae]sppg *1openo ‘ds nifjounig S9] “ING OI 0 0 €l S YO “4 I£6c 122U1110g ‘eny viso?) Aa[purig SISUIIILDISOS pi]jaunig SOI TNA G 0 0 80 S X98 718018 'H'G'H ‘MSN "puy sapioigns DIINDG +91 nva Š 0 0 TI S ssnd-og "ABD Du42ds01214) puupuu gt €91 NNO 0 0 0 ZI S 618018 “HH ‘MSN uoq `d 21040 puəuornoç €91 NM) OI 0 0 €l S GIH uq-58 "1 s1suədoo viuoun) €91 NNO (S861) 8 0 0 01 S ssna-94 "uoq `d DIDNIMDA pianjop];) €91 NM) 9 0 0 71 S dIdH 3-98 "pug wnsouisa4 wnjpjadounydp £901 NOD FOER lousearT] by "urna (F86D 0 £0 (Oc TI 26d IaH —pjey2uigp P Lusag 'e[onzeuoA -KAG umxopoaoj2y umjvjadvwuguo1g LSZ ZHN (z861) 0 €0 (Oz bl 26d GIH ariH-98 7] 21guvui vioydoziy y LSZ ZHN (z861) 0 €0 073 SI Id NNOS8-28 7] vivgnluos “yo vioydoziy y LSZ ZHN (DSOJÁIS "Y =) (87861) 0 £0 (OZ FI 26d 822187 42201] "M ‘GTO "UY Y 1Q20A, Npaays pijapuny LSZ? ZHY FI9600c "sur (t86D) 0 €0 Oz €I %d IAH ‘ANNA Dsosunf *etuopo[e?) MƏN x "udoig pjojipup42 si)K150ss047) LSZ ZHU (87861) O €0 07 €I Gd 4-98 "15104 D40jfiQ SNAISOSSOL) LSZ ZHY (F861) O 90 Oz €I 26d dIdH 42ə]]m ^f 7») ‘ATO uosutqoy “g `D 752» sdo17) LSZ ZHMW GIH ‘OW (F86DÜ O vO OZ ZI 96d “109 'Wavft OPSZ vsoounf “equino”, sese2o1jen) ndi: pa1nodissp;) LSZ ZHN (F861) 0 So Oc CI %d IAH ‘ANNA 628096 vsoounf 'eon] viso?) tog (MS) »21d1jj2 na1nodissp?) LSZ ZHN ug 0 VO 02 SI 26d dIdH Ya-9g “AN CJ 99H) 1422209. vasnodissv77 LSZ ZHU (ez86D O VO Oc €I 3Gd 82318Z Á22D4] "y ‘GTO [419] D701,2D4Q D1010) LSZ ZHY (87861) 0 Yo 07 — 26d d20:98 MOT pjngupxos piəmənig LSZ ZHY (87861) 0 S0 Oc ZI 26d g-H5 gd wey] pznj4J40uuA8 p421ng nig LSZ ZHM OI 0 0 9'I S umjaloqiry-eaulte AA “IPAPH “qxoy snımun? snd.1n2037]3 857 9TA OI 0 0 ¿1 S ariH 09-94 eum[g p?pouo42n sndupooən];] 8 J14 S 0 0 TT S GariH NNOS-28 ‘OW »n£»;»d uospuapouns) 8 DT S 0 0 FI S ariH NNO8-98 ZJUMIS (JOW) $15U3]1Y9 12101514 8S2 JT G 0 0 ZI S GariH 04:99 = WIS ‘Dy unpEgd&spp um y 8Sz DT OI 0 0 TI S MSN:94 "gueg s1]D13snD uopouoydig 98% TAD ç 0 0 OT S NNO8-98 ÍNƏZORAN `M s2suəp2gəim DIID]DG 9z TAD uon NS VWWd N vid addy umureqioH PHL jo utZu() satsadg 6861.0 Ápuegq -eoyqnd penuguo) “| a318v[ Behnke 1393 Volume 75, Number 4 1988 Sieve-Element Plastids OI 0 0 PI S OIN uosjaw ^g '[zeig IJMA SNI1UOZDUD smj1unuos 10d PBI SNV OI 0 0 €'I S OW 6119 svuoy] “q 'uoo1oure?) NOYIS DSOA]O v3od +81 SNV (F861) OI 0 0 UI S unpuy 1g ‘yeMeleg — 7j “00H 142101 “yo snda4020321Qui07) +81 SNV (v861) OI 0 0 zI S (IH Iz20€8 QCH ‘IS "eg sa3opr0zado17 vayAydosiup vg8l SNV ‘zed x *`sutqo1nH OI 0 0 El S OW 0c19 spuoy ] `d *uoo1əureo suao9svoindind “yo payAydosiup PSI SNV Ol TO I SI Sod dIdH sojnog “7 ‘Wwemny q suaquinzosd pppinagy II (HUN S 0 21 S CIH dIAH-D4 anpnq muospim vaviid¢ 0801 SOM 0 0 0 60 oS GIH qIH-58 doog 10uru Dq10smaáuns 0801 SOM 0 0 0 TI oS dIdH NNOS-98 "urxe snisnjooruaoyd. snqny ost soy 0 0 0 01 oS GIH dIadH-94 ‘ds psoy ost sou Ol 0 0 ZI S ariH dIAH:99 ‘IEW Cqunq[) suapunos sndA10poy y 081. SOY 0 0 0 TI oç ariH- og "WR sisuoipispaq vfojpim() 081 SOY 0 0 0 80 oS GIH AIAH-94 7] suu3sadna vjniuajod 081. SOY S 0 0 €I S dIdH NNOg:94 Avi "Y SISUZUDAD]D DISTIAIN] ost SOY OI 0 0 OI S ggs-g “#:5 `v snpunquoy snuumijouoc] 081 SOY G 0 0 TI S qH 128082 (TCH B19PPWH '9d CT) PoruodoÍ nuy 081 SOY 0 0 0 90 oS qH ariH-9g8 "] 210a14 uma?) Ost SOY 1 0 0 60 S dIdH dariH-og T p3səa pippa] 0801 SOY 0 0 0 80 oS d3- "uixejy (7T) 0140un. onpuadyyg Ost SOM 0 0 0 60 oS dIdH dIdH-94 23204 (PUY) noipu? DIUSIYIN(] ost SOY 0 0 0 80 oS dIdH GariH-og "urapng muupuiapons SVÁL] Ost sou S 0 0 60 S dIAH-99 "unpoy DLOPYIOYIUDX njpnuau]p 0801 . SOM 9433.16) 0 0 0 OI S uosq19 7) `V *euozuy (Ke15) sisuəppaəu nnsə]]əs10.] s8l OND Ol 0 0 iI S VSU Sggs- 'vSu-58 "HnN uwmnorofi]oo DUOSOSSO4/) S81 OND Ol 0 0 Ll S VSU VSY-99 “uosq19 7) `V "SIEM `S 12401391Q DULOSOSSOL/) Sst OND 0 €0 cz 90 əd CIH x28 ‘yosneg sisuasupwdu x onjdXaon7 L9 YA (s861) 0 — Zz 90 9d X8 ona] vpron) viyddiong 1.90 W4 "(jeg ('Ipuq (S861) 0 ZO Z 90 od GI3H ssnd:og 19 *ddaog) nsouim]2 viyddsong 291 WO (S861) 0 €0 Z 90 9d GrH ssna-94 AB) Dyofipioo onjd aon 190 DA (S861) 0 v0 Z L0 od w-9d "yoedg 1492p1D]2q pn d/Xaən;] 1901 Y4 OI 0 0 Ft S ASN:28 ‘Pen ^4 $u214n4d pruospiav(] 991 SAd uo NS VNd N via d umtreqiaH PHAP Jo uiu) satoadg 6861.0 Apurey -eoyqng ‘penuljuo’) "I Tv L Annals of the 1394 Missouri Botanical Garden S 0 0 ZI S GI3H GIJH- 'NNOS-O8 `AH ie `3ooH Npudjuayims Də OLI AYO £ 0 0 TI S ariH X: 'NNOS-28 "AR DIJOfIYIVOS bosuna] 691 024 OI 0 0 cl S ariH qud-9d 'qunup sisuadoo vyvody XVS 891 THA AVH-9g pue 8 0 0 ZI S NON piofpoy ^4 `V ‘uyog 'N "T sapi0pas wnioyjuag XVS 891 HLd Keg “y 01 0 0 €I S ariH QIJH-98 12 uo, ('qosung) nsarzuar pənu]o |, 891 XVS 8 0 0 OT S GariH ariH-98 "IMA 2727n21u0d VIDAD 891 XVS 8 0 0 FI S GIH NNO8-58 eg D170/qn9s90 vis193poy 891 XVS Ol 0 O0 91 S dO-98 [uq (1107) uunzoyad wnyíydiyəd 891 XVS ‘uq ('surou, S 0 0 71 S dIdH GIdH-O8 BY J 09H) suaosvindind viuasiag 891 XVS 8 0 0 cl S qH dlaH-94 APA 920/1097 duns 2q]usp 891 XVS 06cl uos]əA; `g pue "qnH Cyg xa aonidc) (89261) S £0 I Tl Sod dIaH ‘OW 28103 22uvaq `] “9 ‘Teg wunjAydosspu Uuospuapopqny y L81 gHH "NH ("yueg xa aonidc) (89/61) S €'0 I OT Sod AN ‘u's 32uD4d `L «4 ‘zeig WNNNUOZDUD uo1puapopquuaM L8I qHu OI 0 0 60 S arniH-98 iepreuqog smpdoos snunig £8I KWY S 0 0 OT S dIdH qIdH-58 "1 snpod snunig £8I AWV 0 0 0 ZI S GIH ariH-9g "yy n42/1$0422 snunig €8I AWV € 0 0 60 S dIaH ariH-98 assay np]p41d Dp10Y90X 3 €8I AWV Ol 0 0 ZI S 288s-98 "Sieg vonuofi]v» vuii]onbnp y Z81 TVW OI 0 0 zI S CIAH ariH-og "1 p1updməno snqiog c8l TVW Ol 0 0 OI S qI3H-958 pur] »o:pu: sidajorydoy c8l 1VIN OI 0 O0 Ol S 128082 ‘FH '31əqləptəH "1 $Tunuruo2 snig c8l TVW Ol 0 0 Ft S dIdH 128082 H'Q'H ?19q[opreH "ureoy `f “JA DIUIDIO) DYJUDIDIA q 781 TVIN G 0 0 OT S VNAIS-99 "1 Pruvu? snpdsopy 781 VW OI 0 0 91 S arniH-98 "IMA s222səa]Xs sapo 781 VW Ol 0 0 ZI S QIIH-O8 ‘PUT (quny y) pətuodoí vA110q0143 781 VW G 0 0 €I S dIdaH ariH-98 "IMA 23uojqo »iuop&?) c8l 1VIN S 0 0 VI S icis dqriH-98 edueT vupyaddip x snd2D1D1,) 781 VW [s 0 0 ZI S QIJH-OH PUT xe TEM snpi2uf 491spəuolo9 Z8I TVW S 0 0 ZI S qH qI3H-:958 1əpuəl] (“ys1ie]a) vyofiunsd viuoíp c8l VW q 0 0 OI S II3H-98 APAN CT ssuapouno saryouvjawp c8l TVW uou NS VWd N VIG adÁ 1] UNI 13H [eee] jo utdu() saradg 6861.0 Apwe y -eoyqnd "anuquo) “| 318V[ Volume 75, Number 4 1988 Behnke 1395 Sieve-Element Plastids ontinued. > C TABLE 1. Publica- tion DIA N PMA SN Type Herbarium Origin of Material Species D'1989 Family BG-K, -BERN Brexia madagascariensis (Lam.) 171 BRX 1.2 BG-HEID BG-K Ribes bracteosum Doug]. ex Hook. Choristylis rhamnoides Harv. GRS 7 HEID BG-K Cephalotus follicularis Labill. BG-HEID HEID Cotyledon orbiculatum L. BG-HEID Kalanchoe laciniata (L.) DC. 0.8 HEID BG-HEID Kalanchoe uniflora (Stapf) R. Ha- 75 1 CRS met. Podostemum ceratophyllum Michx. CM Georgia, S. B. Jones; and D. 176 PDS E. Boufford 22047 ' Erroneously listed in Table 1 of Behnke (1984) as PVc. Celastraceae (CEL; Fig. 4: Catha, Pterocelas- trus). Fourteen species in 13 genera investigat- ed, all with S-type plastids. There are 5-10 mostly globular starch grains recorded within these plas- tids, but their diameter is not uniform (average: 1.3 um; range 1.0-1.8 um). The sieve elements of Goupia contain crystalline, persistent p-protein bodies, a feature that characterizes a number of different dicotyledonous taxa (see Behnke, 1981) but is not found elsewhere in Celastraceae. Hip- pocratea and Salacia, as well as Siphonodon, sometimes separated as Hippocrateaceae and Si- phonodontaceae, respectively, do not differ signif- icantly in sieve-element characters. Elaeocarpaceae (ELC; Fig. 4: Aristotelia, Elaeocarpus). The five species in four genera investigated contain S-type plastids. Their sizes (1.1-1.7 um; average 1.4 um) and number of globular starch grains (5-10) resemble those in Rhizophoraceae (RHZ; Fig. 5: Cassipourea, Ceriops, Crossostylis, Rhizophora, Sterigmape- talum). Thirteen species of eight genera inves- tigated, all with the specific form-P5c sieve-element plastids. Twenty or more rectangular to irregular protein crystals (0.2-0.5 um) fill the plastid inte- rior. The average plastid diameter is 1.4 um. In Rhizophora the protein crystals are irregular, only rarely showing exact rectangular outlines. It is demonstrated, at least for R. mangle, that during the development of a sieve-element plastid, protein accumulates first as a large granular body (see Fig. 5, lower left micrograph) and only thereafter *crys- tallizes’ into several distinct parts. In this study of their sieve-element plastids, all of the tribes recognized within the family were covered. Except for the crystal outlines mentioned for Rhizophora, there is almost no distinction pos- sible between the plastids of the different species. The comparatively large protein crystals depicted in Ceriops (Fig. 5), which come very close in size to those shown in Erythroxylum (Fig. 2), are not restricted to this species. Similar views could have een chosen from other Rhizophoraceae. (See also Behnke, 1982a, 1984.) DISTRIBUTION OF THE DIFFERENT SIEVE-ELEMENT PLASTIDS AMONG THE FAMILIES GROUPED AROUND ANISOPHYLLEACEAE Cunoniaceae (CUN; Fig. 6: Cunonia, Wein- mannia). Five species of five genera investigat- ed, all with S-type plastids. The diameter of the plastids is about 1.2 um; their contents are up to Missouri Botanical Garden Annals of the 1396 8v6 OXI OI CI ç S I I Sve 9MH OI UI ç S əəeəəoeruo8snH Svc NIT L eI € S əLəveur] crc sgg [4 UI I S 9?ooetutojs1oqotg lve OI € 6'0 I S aeared.1e90pa] OG AIA S UI I S 9P92PIUPIAI A 6tc UI 9 eI S S 9?92PIUPI27) lez Odd S GI I S oeooPuegoq 8t6 LIN 9 01 I S SPORE IN 9tc DAZ L £I Y S LAU r4 UI I sod ‘S aeaoe[Audo347 bre LIH OT VI 6 S a?oaoejne[eg 921 Süd 8 £c I S Əeəozurə1sopoq 881 NND 8 S0 r4 Cl e od 9?aoPIouun-) 181 GUN OI vO I CI I od IBIJPPLININ L81 4HY S £0 I cI Z sod aeaoe Ipuspopqeyy £8l AWV 9 01 t S əeəoe[ep3Xury c81 IVW 8 GI £I S ILIJE 081 SOY 60 6 9 cil L oS ‘S IBIJESOY v81 SNV OI GI S S aeaoeo[udosiuy S81 OY) OI S'I t S 9?92P]PUIOSOSSOI) SLI SUD 6'0 € oS oPao?[nsse17) PLI HdO S UT I S oeaovjo[eudo^) €¿LI ALI S VI ç S | GLI SUD OI cI I S IBIDBLIB|NSSOC) ULT XY4 S UI I S 9?ooIXa1g 021 AUD S GI I S LIMI) 691 O9A $ VI I S 9P22POOUPI] 891 THA OI eI I S ILIIYL A 891 HLd 8 ol I S 9?ea2e1oqjuaqd 891 XVS 8 €I 9 S QBIIEFEIPIXES 291 HOU g'o e 9'0 G 3d eeaoerqd&i1on T9I NVA S 8'0 I S araorioneg £91 NND L CI S S IPIIPIUOUN?) 991 SAC OI VI I S aeaoetuosprA edq S9T 'INH 01 VI € S 3PIMRISUNIJ uaidpqeq IdOS Os NS VIAS dss Sd VWd N VIGd idSd sadÁ J, Ápue y `D1Dp pisv]d Juawmaja-aaais 07 Zuipí0220 pasuvssv 2099010ydoz1y y pun avaovayAydosiup punoap Asaydiiad jpuipao ay) fo saipuvpf əy; `g 318v] Volume 75, Number 4 Behnke 1397 1988 Sieve-Element Plastids 10 globular starch grains, often surrounded by d additional tiny granules. (See also Behnke, 1985.) & = = E Baueraceae (BAU; Fig. 6: Bauera). Bauera S| S C R 2 Z 3 5 ses = bioid : f th : Ë Ga qaqas asas Sales rubioides as representative of the monogeneric = dJdd2wN ejs family contains very small (diameter 0.8 S-type A|EEREouunid»e|e3s y nen y (di um) S-typ aoccozmumouo|-«. plastids with up to five small irregular starch grains. om Ek an x . . BÍ Brunelliaceae (BNL; Fig. 6: Brunellia). The EA E j = ae t three species of Brunellia investigated contain = ° P = S-type plastids. Their globular starch grains (about 5 E Š 10 in the average) often seem to disintegrate into c A PE: tiny pieces. The plastid diameter is 1.4 um. | gA A - : : ip ER- Davidsoniaceae (DVS; Fig. 6: David- wo 8 ` sonia). The monotypic Davidsonia pruriens zl onm ES EN" E contains S-type plastids with about 10 strictly glob- un — w >a . è b * DES ular starch grains. The plastid diameter is 1.4 um. gag š : ; z E s 2 Eucryphiaceae (ECR; Fig. 6: Eucryph- - ° š I . . £| ame ny 58 ia). Four of the five species of this monogeneric p|^e9 == [9578 family and one hybrid were found to contain form- FR : Pc sieve-element plastids with two protein crystals, " 5% >> one with a diameter of about 0.3 um. The sieve- EI rm m 2 Š element plastids of Eucryphia are among the tiniest S. recorded within the dicotyledons (average diameter o = = > Cg of 0.6 um). (See also Behnke, 1985.) Sos un tt Oo oS e M zl Crossosomataceae (CRO; Fig. 8: Crossoso- vd m ma). Three species in two genera investigated, o EL all with S-type plastids. The two Crossosoma species < zv 5 = et Soe ° | ° are identical, both in respect to their plastid di- A oooóococ o B us a.s š ameter (1.8 um) and the starch content (about 10 2 o Š globular grains). Forsellesia has smaller plastids - SANDS als E e (diameter 1.0 um) and fewer starch grains. DW . š 2” @ Rosaceae (ROS; Fig. 7: Duchesnea). Sixteen g | = PE species in 16 genera investigated; all species with = o aa o bs Gs S-type plastids, but nine of them without starch A O m men n PE (with form-So plastids). The average diameter of > E E the plastids is 1.0 um, with an average of 0.9 um à. i rE for the So form and of 1.2 um for those with starch FE AT d < 2 E grains. The average number of grains in the starch- D $ containing species is six. "am sol Neuradaceae (NRD; Fig. 8: Neurada). D n : š ° Bo PPP Ez E- = Neurada procumbens contains form-Pcs sieve-ele- T| wow vins Ba Ba Qa un un a 23% ment plastids with one rectangular protein crystal d È $ (diameter 0.4 um) and up to five large starch 3 = 2, 3 grains. The diameter of the plastids is 1.5 um. A e. . A j ES n 5 With these characteristics, there is resemblance to N Š S uu 3 a ° Cy a E = š the sieve-element plastids of Gunnera. (See Behnke, $8 sS s 6 s |288 1986b . [El %- ¿ç = e S Š ç Š Ë = s .) (El ESS298T559£&8|!.9 = d| g£9588852s9585]la en Anisophylleaceae (ANS; Fig. 8: Anisophyllea, a |> | sc 3 s 5 e E S s| > P g ZAZE 28 2 Š = |> a E Combretocarpus, Poga, Polygonanthus). ive = oda < s e SES Zm neo] sz species in four genera investigated; all have S-type plastids. With diameters of about 1.2 um and some ten globular starch grains, the sieve-element plas- 1398 Annals of the Missouri Botanical Garden Ei > dl n e ^ 2 a b »- A B 7 P -Caesafga +. ^ Pelargonium FIGURE 1. S-type sieve-element plastids of Nitraria retusa, Balanites aegyptiaca, Guaiacum coulteri, Peganum harmala, Caesarea albiflora, Pelar ni rne d and Wendtia = and P-type plastids of Larrea divaricata. All x 30,000. c = protein crystals, s = starch grains. Scale b m. Volume 75, Number 4 1399 Behnke Sieve-Element Plastids HT $a | 70 Se Lepidobaiwys -— Humiria Lut "d" ` ` In@orouchera S P T ^ P: CK, Sacoglottis sp. sErythroxylum ` Ixonanthes FIGURE 2. Upper left: longitudinal section through sieve elements of Humiria balsamifer with compound sieve plate (arrows) and several P-type sieve-element plastids (p) , x 10,000. Other photographs: sieve-element plastids of Lepidobotrys staudtii, Linum flavum, Indorouchera griffithiana, Ixonanthes grandiflora; and P-type plastids of Humiria balsamifer, Endopleura sp., Sacoglottis sp., and Erythroxylum coca. All x 30,000. c = protein crystals, s = starch grains, M = mitochondrion. Scale bars = 1 um. 1400 Annals of the Missouri Botanical Garden ` E ` ` N » x 5 , As ` d ied ng l O | l. a E OMEN FR] | Oxa is "5 m. 1 E JM FIGURE 3. P-type sieve-element plastids of Averrhoa carambola, Sarcotheca diversifolia, and Oxalis deppei. All x 30,000. c = protein crystals, f = protein filaments, s = starch grains. Scale bar = 1 um tids of the Anisophylleaceae are rather homoge- neous and very distinct from those of the Rhizopho- raceae. The difference in the plastid types supports elevation of the former tribe Anisophylleeae to family status. (See also Behnke, 1984.) Malaceae (MAL; Fig. 7: Amelanchier). Thirteen species in 13 genera investigated; all have S-type plastids. The average diameter of the plas- tids is 1.2 um, and the average number of starch grains eight. These data do not differ from those of the S-type Rosaceae. No form-So plastids are found in the Malaceae. Amygdalaceae (AMY; Fig. 7: Prunus). Four species in two genera investigated, all with S-type plastids. The plastids in this family are smaller (average diameter 1.0 um) than those in ROS and MAL, although their starch content is similar. Rhabdodendraceae (RHB; Fig. 8: Rhabdoden- dron). The two species investigated of this mono- generic family contain form-Psc sieve-element plastids. Their single protein crystal is rectangular and about 0.2 um in diameter. There are about five irregular starch grains. The average plastid diameter is 1.2 um. (See also Behnke, 1976a.) Saxifragaceae (SAX; Fig. 7: Bergenia). Six species in six genera investigated, all with S-type plastids. The average plastid diameter is 1.3 um; there are about eight irregularly shaped starch grains in the plastids. Penthoraceae (PTH; Fig. 7: Penthorum). Penthorum sedoides as representative of the mono- generic family contains S-type plastids with char- acters almost identical to those in SAX, i.e., with a diameter of 1.2 um and containing about eight starch grains. Vahliaceae (VHL; Fig. 7: Vahlia). The S-type plastids recorded for Vahlia capensis show the same pattern as found in SAX—diameter 1.2 um and containing about ten starch grains. Francoaceae (FCO; Fig. 7: Francoa). The in- vestigated Francoa sonchifolia contains S-type plastids only slightly different from those in SAX; their diameter is 1.4 um; the number of starch grains is about three. Greyiaceae (GRY; Fig. 7: Greyia). Greyia sutherlandii contains S-type plastids with about five globular starch granis. The plastid diameter is 1.2 um. Brexiaceae (BRX; Fig. 7: Brexia). Brexia madagascariensis was shown to contain S-type plastids about 1.1 um in diameter with about five starch grains. Grossulariaceae (GRS; Fig. 7: Ribes). Ribes Volume 75, Number 4 1401 1988 Behnke Sieve-Element Plastids bracteosum contains S-type plastids with about ten globular starch grains and a diameter of 1.2 um. Iteaceae (ITE; Fig. 7: Itea). Two species in the two genera investigated; both contain S-type plastids with about five starch grains that disinte- grate into tiny particles. The average plastid di- ameter is 1.5 um. Cephalotaceae (CPH; Fig. 7: Cephalotus). The monotypic Cephalotus follicularis contains S-type plastids with about five starch grains and a diameter of 1.1 um. The starch grains are slightly disc-shaped and surrounded by tiny particles. Crassulaceae (CRS; Fig. 7: Cotyledon). Three species in two genera shown to contain form- So sieve-element plastids. Their average diameter is 0.9 um. Many more species were investigated, but although the fixation of the material was sat- [ : isfactory, repeated screening failed to detect sieve- a * e element plastids. Most likely, the So-plastids easily Bd. + break down during the differentiation of the sieve rota elements. Similar conclusions were made from stud- = — ies with Cucurbita species that also contain form- So plastids (Buvat, 1963; Esau & Cronshaw, 1968). In Crassulaceae the form-So plastids sometimes contain small inclusions, which rarely show a crys- talline composition. Since they are probably related to protein crystals, sieve-element plastids of CRS were also defined as Po/So. (See Behnke, 1981.) Podostemaceae (PDS; Fig. 7: Podoste- mum). | Podostemum ceratophyllum contains S-type plastids with about eight starch grains, often surrounded by tiny particles. The diameter of the plastids is 2.3 um, by far the highest found among the taxa studied for this report. DISCUSSION The results from investigations of the sieve-ele- ment plastids reported here unambiguously support separation of the tribe Aniosphylleeae from the Rhizophoraceae and its elevation to the rank of a family. The very distinct and remarkable form-P5c j : sieve-element plastids of Rhizophoraceae s. str.— 1 4 | w without any incorporation of even traces of starch— has no direct relationhip to the pure S-type (lacking AA un any amount of protein) as found in the Aniso- B EX phylleaceae. According to a model demonstrating š ev m E the interrelationships between the different forms Elaeocarpus of sieve-element plastids (proposed by Behnke, FIGURE 4. S-type sieve-element plastids of Catha f edulis, Pterocelastrus tricuspidatus, Aristotelia chilensis, necessary to change the S-type into a form-Pc and Elaeocarpus ganitrus. All x30,000. s = starch sieve-element plastid. Without the presence of in- — grains. Scale bar = 1 um. 1988), at least two evolutionary steps would be 1402 Annals of the Missouri Botanical Garden Sterigmapetalum Sy Cassipourea killipii Cassipourea barteri ; 5. sieve- iai plastids of the Rhizophoraceae. Upper left: MM section fs primary es P. f Rhizoph ra mangle showing two members of a sieve tube (SE) « ted by plate rows) and apum ceat P-type plastids (p); Xx5,000. Lower left: P. oes plastids of a sei jn element of R. mangle. Plastid matrix filled with granular protein material (p), not yet differentiated vin crystals; — dictyosomes; X 30,000. Other photographs, from top to bottom: P-type pla B of R. mangle, Sterigmapetalum anti ne Crossostylis grandiflora, Ceriops tagal, Gamer killipii, and Cassipourea barteri. All x 30,000. c = protein crystals. Scale bar = 1 um. Volume 75, Number 4 1988 Behnke 1403 Sieve-Element Plastids ‘Bauera FIGURE 6. S-type sieve-element plastids of Cunonia capensis, Weinmannia trichosperma, Brunellia sp., Bauera rubioides, Dsvulsonia pruriens, and form-Pc sieve-element plastids of Eucryphia billardieri. All x 30,000. c = protein crystals, s = starch grains. Scale bar = 1 um termediates the coexistence of both types within one family is not very likely. ile sieve-element plastids help discriminate nisophylleaceae, possible only to a limited extent. Therefore, the position of the two families relative to the taxa proposed by Dahlgren (this volume) to constitute the ordinal periphery will now be discussed. GERANIALES SENSU LATO In his last version of the system of classification of dicotyledons, Dahlgren (in press) divided the Geraniales s.l. into two orders, the Geraniales s. str. and the Linales. The following discussion makes use of this separation. 1. (= Geraniales s. str.). Dahlgren (in press) listed Zygophyllaceae, Peganaceae, Nitrariaceae, Geraniaceae, Vivianiaceae, Ledocarpaceae, Bie- bersteiniaceae, Dirachmaceae, and Balanitaceae in this alliance. Among the families available for our studies (see Table 2) S-type sieve-element plastids are most common and P-type plastids are foun only in Larrea (Zygophyllaceae). The diameter of the plastids is rather uniform, varying around 1.1 um. Some families contain disc-shaped starch grains as a specific marker: Geraniaceae, Vivianiaceae, and Zygophyllaceae (in part). The S-type plastids of Balanitaceae diverge more from the above pattern (see also Fig. 1). Both Cronquist (1981) and Thorne (1983) placed Bal- anites within Zygophyllaceae; Takhtajan (1987) transferred to Rutales the Zygophyllaceae and those families that, like Balanitaceae, Nitrariaceae, and Peganaceae, were split off earlier. Neither of these assignments is strongly supported by the plastid data 1404 Annals of the Missouri Botanical Garden Ni Bergeniá Ye Volume 75, Number 4 1988 Behnke 1405 Sieve-Element Plastids 2. (= Linales). ed Linaceae, Hugoniaceae, Humiriaceae, Cteno- lophonaceae, Ixonanthaceae, Erythroxylaceae, Lepidobotryaceae, and Oxalidaceae in the Lin- ales. With four S-type families, two P-type fam- ilies, and one family containing both S-type and two different forms of P-type sieve-element plas- tids, this suborder is very heterogeneous (fresh material from Ctenolophon was not available). The S-type plastids (in Linaceae, Hugoniaceae, Ixonanthaceae, Lepidobotryaceae, Biebersteini- aceae, and Oxalidaceae) are rather small: their average diameter is about 1.1 um. From five to ten starch grains, often disintegrating into small particles and sometimes including a very large one, are found within these plastids (see Figs. 2, 3; Behnke, 1982c); a few species differ slightly from this pattern, e.g., in Reinwardtia (Linaceae) (Table Dahlgren (in press), includ- The two different forms of P-type plastids re- corded within the Oxalidaceae are restricted to different genera: Averrhoa and Sarcotheca contain form-Pcfs, while Oxalis has highly specialized and very small form-Pc sieve-element plastids. Diam- eters and compositions of these two forms are so different (see Table 1) that it seems justified from the plastid data to support the separation of the families Averrhoaceae (see Hutchinson, 1959) and Hypseocharitaceae (see Takhtajan, 1987). The two remaining P-type families, Humiriaceae and Erythroxylaceae, contain P-forms not directly related to those of the Oxalidaceae. The form-P5cs plastids found in Humiriaceae are similar to the P4cs plastids of Fabales (cf. Fig. 2 with Behnke & Pop, 1981, figs. 5-15) and can be regarded as transitional between S-type and form-P5c plastids of the Erythroxylaceae. The latter are extraordi- narily distinct from all other sieve-element plastids, and the only other family reported to contain this form is the Rhizophoraceae. Cronquist (1981), Thorne (1983), and Takh- tajan (1987) incorporated the Oxalidaceae and from each other only slightly. The patterns of the S-type plastids in the Linales and Geraniales are not sufficiently different to favor one or other treat- ment—and the plastids of Lepidobotrys, Bio- phytum, and Hypseocharis are intermediate. Therefore, in the familial sequence given in Ta- ble 2, which is arranged according to the data obtained with the sieve-element plastids, the order Geraniales s.]. is maintained. CELASTRALES Celastraceae and Elaeocarpaceae contain S-type sieve-element plastids to some extent alike in pat- tern but not very specialized. The families are not very uniform in their sieve-element plastids (e.g., see Fig. 4: Aristotelia and Elaeocarpus). Rhizophoraceae contain form-P5c plastids, a highly specialized pattern that is found throughout all genera investigated. In addition, variation in the diameter of the plastids is very small. There seem to be no common sieve-element plastid characters between the Rhizophoraceae and the other two families of this order; the closest similarities are with plastids of Erythroxylaceae. Thorne (1983) placed Rhizophoraceae in his Cornales; Cronquist (1981) and Takhtajan (1987) regarded their order Rhizophorales as a close ally of the Myrtales. Dahlgren in his last version (in press) put Rhizophoraceae together with Elaeocar- paceae into his order Rhizophorales and gave it a position after Geraniales/ Linales and his newly de- fined Celastrales (including S-type families only). Sieve-element plastid data suggest a close as- sociation of Rhizophoraceae with Humiriaceae and Erythroxylaceae (see Table 2). CUNONIALES The S-type sieve-element plastids present in four of the five families of this order are heterogeneous. Plastid diameter and starch content range from large with ten grains to very small with little starch (see Table 1). The sequence given in Table 2 sug- gests an evolution from the large unspecialized to the small specialized plastid and enables a connec- tion to the only P-type family (Eucryphiaceae). Exactly the same five families constitute Takh- tajan's (1987) Cunoniales. Thorne (1983) added Staphyleaceae and Corynocarpaceae to his sub- order Cunoniinae, while Cronquist distributed them among his Rosales. SAXIFRAGALES All families within this order contain S-type sieve- element plastids, of which Crassulaceae is special- — FIGURE 7. S-type So), Pein orbiculatum (form sonchifolia, Greyia sutherlandii, Ribes bracteosum, Brexia sieve-element plastids of Amelanchier canadensis, -— padus, Duchesnea indica (form- -So), Bergenia purpurascens, Penthoru sedoides, Vahlia capensis, Francoa madagascariensis, wawawa . Itea ilicifolia, and um Podostemum ceratophyllum. All x 30,000. s = starch grains, o = form-So plastid. Scale b 1406 Annals of the Missouri Botanical Garden AnisdDhytteg trap. Anisophyllea purp. Rhabdod: 'Polygonanthus Combretocarpus SURE 8. Upper left: longitudinal section through sieve elements (SE) of rs purpurascens connected by ed sieve areas (arrows) and containing many S-type plastids (s) ; X6 phloem protein. Other Volume 75, Number 4 1988 Behnke 1407 Sieve-Element Plastids ized by form-So plastids. Sizes and starch content of the S-type plastids are rather homogeneous. The nly exception is Podostemum, which has very large plastids (Fig. 7, Table 1) and does not fit into this order, nor in the entire alliance. Cronquist (1981) and Takhtajan (1987) separated the Podo- stemaceae in its own order. Thorne (1983) placed this family within his Saxifragineae. Until further evidence from other characters emerges, we prefer the treatment as a se ripheral to the Saxifragales/Rosales. parate order, somewhat pe- ROSALES S-type sieve-element plastids are recorded ex- cept for the two families Neuradaceae and Rhab- dodendraceae. The pattern of the S-type plastids (diameter and starch content) is similar to that of the Saxifragales. One family includes form-So plas- tids: the Rosaceae. The presence of these So plas- tids in at least some genera (see Table 1 for iyu e.g., smaller diameter) makes the Rosaceae s. distinct from the Malaceae and Amygdalaceae. The plastid pattern of the latter family is not different from that of the S-type genera in the Rosaceae. Sieve-element plastids of the four genera tested from the Anisophylleaceae display a rather uniform pattern: while their sizes conform with that of both Saxifragales and Rosales, the amount of starch within a plastid is much higher than in the other taxa (cf. Figs. 7, 8). Therefore, on account of the plastid data, an association of Anisophylleaceae with either Saxifragales or Rosales is not excluded, but a positive decision cannot be made. The remaining two P-type families of Rosales both contain Psc sieve-element plastids, but of dif- ferent pattern. Rhabdodendraceae contain in their sieve-ele- ment plastids a tiny rectangular protein crystal and up to five irregular starch grains, a pattern re- peatedly found within the Magnoliiflorae (see Behnke, 1988). Neurada is more distinct because of its larger crystal (diameter 0.4 um) and higher starch con- tent. [ts sieve-element plastids come very close to those of the Gunneraceae (see Behnke, 1986b). CONCLUSIONS DRAWN FROM THE PLASTID DATA Given the periphery of families and orders around the Rhizophoraceae—including a few additional ones discussed during the preparation of this Rhi- zophoraceae symposium—and the distribution of types and forms of sieve-element plastids, the fol- lowing annotations to the relationships between the different taxa can be ma The ordinal placement of the Aniosphylle- aceae. This is still uncertain as far as sieve-element plastids are concerned. Pattern similarities exist to S-type plastids in the Saxifragales-Rosales groups (not to the Cunoniales), but affinities to other taxa are not ruled out. If a closer relation to the Rhizophoraceae is still considered, the diameter of the plastids (average of 1.3 um in both families) would be the only supporting plastid data; otherwise their contents, as discussed, differ by at least two evolutionary steps. The ordinal placement of the Rhizophora- ceae. The identical sieve-element plastids in Rhi- zophoraceae and Erythroxylaceae, together with the fact that within the dicotyledons the form-P5c is exclusive to these two families, strongly favors their close alliance (see also the ordinal restriction of the subtype-P3 sieve-element plastids, Behnke, 1976b). Related plastid forms are found in the Cyrillaceae (P5cf) and the Humiriaceae (P5cs). Sieve-element plastids of the Cyrillaceae are dis- tinguished from pei of the Rhizophoraceae by the presence of protein filaments in addition to protein crystals Gebnke, 1982a) and a larger di- ameter (average of 1.6 um). Nevertheless, their similarity is reason enough to propose at least dis- tant relationships. Traditionally, Cyrillaceae have been placed into Celastrales, Theales, and (recently more often) Ericales (cf. Behnke, 1982a). In view of Dahlgren’s (this volume) proposal to associate closely the Rhizophoraceae with the Celastraceae, the inclusion of the Cyrillaceae within the Celas- trales (see e.g., Melchior, 1964) may be worth reconsidering. — np show S-type sieve-element plastids in the five m we species s] dui quida ais (A. trape- zoidales, purpurascens, Combretocarpus motleyi, Polygonanthus ama onicus, Poga oleosa) an rossosoma bigelovii, as well as form-Pc plastids of Neurada poe and Ses alee macrophyllum. All x 30,000. c — protein crystals, s — starch grains. Scale bar 1408 Annals of the Missouri Botanical Garden FIGURE 9. S-type sieve-element plastid of Bieber- steinia multifida. x 30,000. s = starch grain. Scale bar = ] um. The first and only record for form-P5cs plastids in the Humiriaceae links the form-P5c plastids of the Rhizophoraceae and Erythroxylaceae to the S-type families in the Geraniales s.l. It has further potential in bridging the entire subtype-P5 to the subtype-P4 of the Fabales, thus making the sub- type-P4/P5 a characteristic pattern of sieve-ele- ment plastids restricted to the Rutinae (sensu Dahl- gren, this volum The form- P5cs plastids found in the Humiri- aceae connect the form-P5c to the S-type plastids in the Geraniales: their number of protein crystals (more than ten on average) is the second highest recorded in the dicotyledons (after those in RHZ and ERY), their number of starch grains and av- erage plastid diameter are compatible with the S-type plastids in the Geraniales. The shape of the protein crystals is not as distinctly rectangular as in the form-P5c plastids (cf. Figs. 2, 5), but even within Rhizophora the crystals have no sharp edges (Fig. 5). The P5cs pattern is very close to that of the form-P4cs plastids present in the Fabales: both contain five or more irregular protein crystals in addition to a variable number of starch grains. It is suggested that from a common ancestor with the plastid inheritance several parallel lines lead to Fabales, Geraniales, Rhizophorales, and probably Celastrales. However, data from sieve-element plastids do not contribute to the placement of Celastrales un- less the inclusion of the Cyrillaceae (cf. Hutchinson, 1959; Melchior, 1964) is followed. Dahlgren transferred Elaeocarpaceae from Mal- vales to either the newly defined Celastrales (Dahl- gren, this volume: together with RHZ and CEL) or to his Rhizophorales (Dahlgren, in press: as the only other family of this order in addition to the RHZ). The data from sieve-element plastids support neither of the two arrangements. However, another phloem character corroborates the exclu- sion of Elaeocarpaceae from the Malvales: Bom- bacaceae, Malvaceae, Sterculiaceae, and Tiliaceae have within their sieve elements so-called persis- tent, crystalline p-protein bodies, which are absent from the Elaeocarpaceae. The persistent p-protein bodies are a typical character of the Malvanae/ Violanae and a few other taxa (see Behnke, 1981). In summary, data from sieve-element plastids suggest the following parallel sequences of families (those not yet investigated are in parentheses; cf. Table 2): 1. Balanitaceae, Zygophyllaceae, Nitrariaceae, Peganaceae, Geraniaceae, Vivianiaceae, Le- docarpaceae, (Biebersteiniaceae), (Dirach- maceae bo . Linaceae, Hugoniaceae, (Ctenolophona- ceae), Ixonanthaceae 2.1 Lepidobotryaceae, Hypseocharitaceae, Oxalidaceae, Averrhoaceae 2.2 Humiriaceae, Erythroxylaceae zophoraceae 3. Celastraceae, Elaeocarpaceae 4. Cyrillaceae Families excluded. On the basis of the sieve- element characters two families discussed during the preparation for this symposium as putative allies are to be definitely excluded: the Flacourti- aceae and Podostemaceae. The Flacourtiaceae contain S-type sieve-ele- ment plastids, but their persistent p-protein bodies (cf. Behnke, 1981) place them in the Violales. Podostemaceae differ from the discussed orders by their large S-type plastids and the pattern of starch grains. NOTE ADDED IN PROOF Fresh rhizomes of Biebersteinia multifida DC. kindly have been made available by E. Gabrielian (Erevan, USSR). The following paragraph should be read after Ledocarpaceae (on page 1389): Biebersteiniaceae (BBS; Fig. 9) Biebersteinia multifida contains S-type plastids with one or few globular starch grains, often Ba MN into tiny particles. The plastid diameter is 1. LITERATURE CITED BEHNKE, H.-D. 1974 [1975]. P- und S-Typ Siebele- ment-Plastiden bei Rhamnales. Beitr. Biol. Pflanzen 50: 457-464. 1976a. Sieve-element plastids of Fouquieria, Volume 75, Number 4 1988 Behnke Sieve-Element Plastids 1409 Frankenia deseada. and Rhabdodenron (Ru- taceae), tax metimes allied with Centrospermae (Caryophyllales) Taxon 25: 265-268. — — —. 1976b. Ultrastructure of sieve-element plas- tids in Caryophyllales (Centrospermae), evidence for the delimitation and ir ssification of the order. Pl. Syst. Evol. 126: 31- 1981. Sieve- an characters. Bot. 5 381-400. 82a. Sieve-element plastids of Cyrillaceae, etii ads and Rhizophoraceae: description and significance of subtype PV plastids. Pl. Syst. Evol. 141: 31-39. 1982b. Sieve-element plastids, exine sculp- turing and the acne affinities of the Buxaceae. Pl. Syst. Evol. 139: 257-266. . 1982c. Sieve- p plastids of seria and Oxalidaceae. A contribution to the knowledge P-type as in dicotyledons and their Bic Bot. Jahrb. S 3: 1-8. 84. ae of sieve-element plastids of Myrtales and allied groups. Ann. Missouri Bot. Gard. 71: 824-831. . Contributions to the knowledge of P-type sieve-element plastids in dicotyledons. II. Eucryphi- aceae. Taxon 34: 607-610. 19 Contributions to the knowledge of P-type sieve-element plastids in dicotyledons. IV. Acanthaceae. Bot. Jahrb. Syst. 106: 499-510. Contributions to the knowledge of sieve-element plastids in Gunneraceae and allied fam- ilies. Pl. Syst. Evol. 151: 215-222. ———. 1988. Sieve element plastids, phloem-protein and evolution of the flowering plants: III. Magnoli- idae. Taxon 37: 699-732 T. J. Masry. 1977. S-type sieve-element plastids and anthocyanins in Vivianiaceae: evidence against its eae into Centrospermae. Pl. Syst. 375. Evol. 126: Nordic J. 981. Sieve-element plastids and crystalline P(hloem)- -protein in Leguminosae: micro morphological characters as an aid to the circum- scription of the family and subfamilies. Pp. 707-715 in R. M. Polhill & P. H. Raven (editors), Advances n Legume Systematics. Academic Press, London. es R. 1963. Les infrastructures et la différenciation des cellules criblées de Cucurbita pepo L. Portugaliae Acta Biol., Ser. A, 7: 249-299. CRONQUIST, A. 1981. An Integrated System of Classi- cation of Flowering Plants. Columbia Univ. Press, New York. DAHLGREN, G. The last Dahlgrenogram— ea em of clas- sification of the Dicotyledons. /n: K. Tan (editor), The Davis and Hedge Festschrift. ied: Univ. Press. (In press. DAHLGREN, R. 1983. General aspects of angiosperm evolution and macrosystematics. Nordic J. Bot. 3: 9. Rhizophoraceae and Anisophylleaceae: sum- mary statement, relationships. Ann. Missouri Bot. , K. & J. CRoNsHAW. 1968. Plastids and mito- chondria in the phloem of Cucurbita. Canad. J. Bot. HUTCHINSON, J. 1959. The Families of Flowering Plants. 2 volumes. Academic Press, London & New York. MELCHIOR, H. (editor). 1964. A. Engler's Syllabus der Pflanzenfamilien, Band II, 12th edition. Borntager, Berlin. TAKHTAJAN, A. 1987. Systema Magnoliophytorum. Nauka, Leningrad. [In Russian.] THORNE, R. F. Proposed s eT in the angiosperms. Nordic J. Bot. Tose, H. & P. H. Raven. 1987. ACER aa ae of mi lide Der Ann. Missouri Bot. Gard. 74 WEBER, W. E 1982. Mnemonic three-letter acronyms for the families of vascular plants: a device for more effective herbarium curation. Taxon 31: 74-88. EMBRYOLOGY OF TRIBE GYNOTROCHEAE (RHIZOPHORACEAE) AND ITS DEVELOPMENTAL AND SYSTEMATIC IMPLICATIONS! Adrian M. Juncosa? and Hiroshi Tobe? ABSTRACT 4 complete embryological description of the four genera traditionally circumscribed as xen be oberen ( Carallia, Crossostylis, Gynotroches, and Pellacalyx) is presente ed. Most of the character states are consistent w those known for other genera of the family that have been studied, but several are not. In Crossoatylis, Gynotroches, and Pellacalyx, microsporogenesis occurs by bo single flower. This is apparently previously unreported in any angiosperm anther m ) h simultaneous and successive meiotic cytokinesis, even in a and suggests that meiosis in the phenomena: in female flowers sporopollenin pollen walls fail to form, and in male flowers late ovule development Rhizophoraceae in ha mbryological and other morphological data indicate that Gynotrocheae and seed development. Crossostylis is intermediate between these two genera and the ancestral tribe Macarisieae, and Carallia is probably intermediate between all other inland genera and the mangrove tribe (Rhizophoreae) . The family Rhizophoraceae is generally accept- ed as comprising three tribes: Macarisieae (six or seven inland genera), Gynotrocheae (four inland genera), and Rhizophoreae (four mangrove gen- era). The four genera previously included as tribe Anisophylleeae or as a a are now epe gated as an unrelated family, a disp rigin suggested by Ridley (1922) and now puni by many lines of evidence (Behnke, 1982; Tobe & Raven, 1987a; Juncosa & Tomlinson, 1987, this volume). Embryological evidence (Tobe & Raven, 19872) and some aspects of vegetative anatomy (Juncosa & Tomlinson, this volume) suggest that the Aniso- hylleaceae may be related to Myrtales, but the infrafamilial systematics and extrafamilial phylo- genetic relationships of the Rhizophoraceae remain uncertain. Much of this uncertainty results from the great variability in many systematic characters in this family and the paucity of information about T certain key genera. While many later-develop- mental and sporophytic characters exhibit consid- erable adaptive radiation, embryological characters are usually more conservative. Furthermore, be- cause embryological characters are inherently de- velopmental, homologies are more reliable and the polarity of the characters can often be determined, lending additional weight to a phylogenetic hy- pothesis. The Gynotrocheae are of pivotal systematic im- portance within the context of the family for several reasons. There is greater variability in conventional taxonomic characters within this tribe than in the other two: phyllotaxy, wood, flowers, fruits, and seeds all afford good examples (Schimper, 1893; van Vliet, 1976; Juncosa & Tomlinson, this vol- ume). It has even been suggested that Pellacalyx could be excluded from the family (Marco, 1935; Dahlgren, pers. comm.). In another direction, the floral morphology of the Rhizophoreae is very sim- ' We gr iind. acknowledge the assistance of individuals and agencies that permitted and facilitated fieldwork aul P. and collections K. Chai and the O.R.S.T.O.M., iced ledon Mattmuller for technical assista ance. This DEB-8016635 to A.M.J. a 95616 Forest Department, Sarawak, M ial was also supplied nd B. C. Stone. nk study was made possible by NSF Dissertation Improvement Award d NSF R Minds Grant BSR-8216271 to P. B. * Present address: Donee of Agronomy and Range Science, University of California, Davis: California » A Malaysia; Philippe Morat and the by G. McPherson a We thank Monica Tomlinson and A.M J. j Biological Laboratory, Yoshida College, Kyoto University, Kyoto 606, Japan. ANN. Missour! Bor. GARD. 75: 1410-1424. 1988. Volume 75, Number 4 Juncosa & Tobe 1411 1988 Gynotrocheae Embryology TABLE l. Collection localities and voucher information for species studied. Location of Species Locality Collection Number Voucher Carallia borneensis Oliver Andulau Forest Reserve, Brunei Juncosa 464 (lost) Crossostylis grandiflora Mt. Panie, New Caledonia Juncosa 388 NOUM, DUKE ngn. & Gris Yate, New Caledonia Juncosa 413 NOUM, DUKE New Caledonia G. McPherson 1617 MO Gynotroches axillaris Padawan, Sarawak, Malaysia Juncosa 481, 482 SAR, DUKE Blume! Kuching vicinity, Sarawak, Malay- Juncosa 440, 442 SAR, DUKE sia Maxwell Hill, Perak, Malaysia B. C. Stone 15397 KLU Pellacalyx cristatus Hemsl. Kampong Tepoi vicinity, Sarawak Juncosa 488 SAR, DUKE P. lobbii (Hook. f.) Schimp. P. cf. saccardianus Scort. P. symphiodiscus Stapf sia Andulau Forest Reserve, Brunei Maxwell Hill, Perak, Malaysia Kuching vicinity, Sarawak, Malay- Juncosa 465 (lost) B. C. Stone 15396 86 Juncosa 4 KLU SAR, DUKE ' Material from the two Sarawak localities represented two distinct species (see text for details). ilar to that of Carallia (Juncosa & Tomlinson, 1987; Juncosa, pers. obs.); so clues to the evolution of vivipary might be found indirectly among the embryological characteristics of this and other gen- era of the Gynotrocheae. Although data are not yet available for all genera of Macarisieae, they comprise a relatively homogeneous tribe in other respects (Schimper, 1893; Hutchinson & Dalziel, 1954; Floret, 1976; van Vliet, 1976; Juncosa $ Tomlinson, this volume), and comprehensive em- bryological data for Cassipourea and Sterigma- petalum are available (Juncosa, 1984a; Tobe & Raven, 1987b). Fragmentary data are available for two genera of Gynotrocheae (Mauritzon, 1939), and seed coat anatomy of all four genera is now well known (Corner, 1976; Tobe & Raven, this volume). MATERIALS AND METHODS Species studied and collection data are listed in Table 1. Identification and nomenclature follow Ding Hou (1958). In this treatment, Gynotroches is considered to consist of one rather variable species, but material used for this study appeared to represent two very distinct ecotypes or species; differences are noted below. Adequate embryolog- ical material was available for only one species in each of the genera Carallia (eight species) and Crossostylis (six species). Only four of the nine species of Pellacalyx were examined, and although complete series of all developmental stages were not available for all four, the data obtained suggest that the genus is probably embryologically uniform. Material was generally fixed in the field in for- malin-acetic acid-60% ethanol (10:5:85 or 5:5: 90), but some Brunei collections were fixed in whis- ey and formalin (about 10: 1). Most material was dehydrated in a tertiary butanol/ethanol series and embedded in paraffin, then sectioned at 8-10 um and stained with either Heidenhain's iron hema- toxylin or safranin followed by fast green (Johan- sen, 1940). Other material was dehydrated in eth- anol, embedded in Polysciences JB-4 resin, sectioned with glass knives at 4-6 um, and stained with iron hematoxylin. OBSERVATIONS Embryological character states are summarized in Tables 2 and 3. Detailed descriptions follow, including notes of occasional variations in character states and some features not always included in embryological summaries. Carallia borneensis Oliver Anthers of this species are medifixed, and the connective and the tips of the two halves of the anther are slightly prolonged. The anther consists of four microsporangia. The sporangial wall de- type, that is, having two middle layers, one sharing a common origin with the endothecium; the other, with the tapetum (Fig. 1). Neither the epidermis nor the middle layers persist to anthesis. The endothecial cell walls have very few thick bars of secondary wall (so-called “fibrous” thickenings; Fig. 2). The tapetum is glan- dular; its cells have two nuclei. Cytokinesis in mi- crospore mother cells is simultaneous, producing tetrahedral tetrads. Anther dehiscence is introrse and occurs by longitudinal slits. 1412 Annals of the Missouri Botanical Garden TABLE 2. Anther and pollen characteristics of tribe Gynotrocheae. Carallia Crossostylis Gynotroches Pellacalyx umber of microsporangia 4 4 4 Anther wall development basic basic monocotyledonous c Epidermis at anthesis degenerate persistent degenerate persistent, collapsed iddle layers degenerate degenerate degenerate generate Endothecial wall thickenings present present present present petum glandular glandular glandular glandular Tapetal cell nuclei 2 2 2 Meiotic cytokinesis simultaneous simultaneous or simultaneous or simultaneous or successiv ccessi successive Pollen tetrads tetrahedral tetrahedral or tetrahedral or tetrahedral or decussate decussate decussate Pollen nuclei 2 Ovules of C. borneensis are bitegmic, and even at early stages of development each integument consists of more than two cell layers (Fig. 3). Early cell divisions in the integuments are irregularly oriented, as would be expected in a multiseriate, parenchymatous structure; by contrast, the anti- clinal divisions in the biseriate integuments of other inland Rhizophoraceae are usually uniformly per- pendicular to the surface, as would be expected in a structure consisting of two protodermal layers. The outer integument of C. borneensis thickens to 7-15 cell layers by the time of anthesis; the inner integument, to about 10 cell layers (Fig. 4). Both integuments contribute to forming the micropyle, and the endostome and exostome are not aligned. The outer integument is vascularized, but the inner integument is not (Fig. 5). The outer epidermis of the inner integument, which forms the prominent exotegmen in other genera, is distinctly differen- tiated, and a pronounced endothelium is formed as megagametogenesis begins; its cells are densely staining and palisadelike by the two-celled stage of gametogenesis (Fig. 4). At anthesis, the endo- thelium becomes tanninized and thick-walled. The archesporial cell divides once periclinally, d the parietal derivative usually divides once anticlinally (Fig. 6). The ovule is thus crassinu- cellate. A cell wall is formed after meiosis I, and a linear, T-shaped, or irregular tetrad of mega- spores is formed. The gametophyte develops from the chalazal megaspore. The observation of two-, four-, and eight-nucleate stages confirmed that ga- metogenesis occurs according to the Polygonum- type pattern. Three antipodal cells are formed, but their nuclei appear condensed or degenerate at anthesis. The mature synergids are pyriform and stain very densely. The polar nuclei are closely appressed but do not fuse to form a secondary nucleus before fertilization. Most of the nucellus, particularly its micropylar half, degenerates during megagametogenesis (Fig. 4). Fertilization is porogamous, and endosperm for- mation is initially free-nuclear. Endosperm devel- opment and expansion of the fertilized ovule both transverse in all specimens studied (over 100 seeds in early developmental stages). However, the ori- entation of the second division was variable. In most cases it was parallel to the axis of the proem- bryos (““vertical” or longitudinal, although the axis of the proembryo was not usually parallel to the long axis of the ovule), but in many other specimens it was oblique (Figs. 7, 8). Although the basal cell subsequently divides transversely and these deriv- atives divide both transversely and longitudinally, none of the resultant cells contribute to the embryo proper. Thus, allowing for a slight relaxation of the plane of division of the terminal cell, embryogenesis in C. borneensis can be described as being fun- damentally of this type. During the early globular stage (proembryo consisting of up to about 20 cells in median longitudinal section), both the basal cell of the suspensor and the cells closest to the embryo proper divide longitudinally, producing a mostly multiseriate suspensor, sometimes with a uniseriate section in the middle (Fig. 8). The mature embryo consists of two laminar cot- yledons and a straight hypocotyl, this making up about two-thirds of the length of the embryo. The axial vascular cylinder is medullated throughout its ength. The embryo is green and enveloped by abundant endosperm. The seed coat is mostly testal and 20-50 cell layers thick; its surface is bullate Volume 75, Number 4 1988 Juncosa & Tobe Gynotrocheae Embryology 1413 TABLE 3. Ovule and seed characteristics of tribe Gynotrocheae. See text for additional details. Carallia Crossostylis Gynotroches Pellacalyx Inner integument cell layers 2-3/many! 2/20 2/5-6 2/8-10 at inception/anthesis Outer integument cell layers 3 or more/many! = 2-3/4-5? 2/2 2/2 at inception/anthesis Integument vascularization outer only? none none one Micropyle diplostomic diplostomic diplostomic diplostomic Endothelium present present present present Archesporial cell(s) several severa 1 Megasporangium crassinucellate crassinucellate tenuinucellate tenuinucellate Megaspore tetrad T-shaped or linear linear T-shaped linear Megagametogenesis chalazal-monospor- chalazal-monospor- _chalazal-mono- chalazal-mono- ic ic sporic sporic Antipodals at anthesis condensed condensed persistent persistent Synergid shape i pyriform rifor ifo Secondary nucleus not formed not formed not formed not formed Endosperm development ndosperm transfer cells Embryogenesis xotegmen Cotyledons free-nuclear variable unsclerified 2, foliaceous — en ent — variable variable — sclerified sclerified 2, foliaceous 2, foliaceous ' Integuments of species other than C. borneensis are thinner at maturity and possibly also at inception. ? Strictly biseriate outer integument was never observed, possibly because sis x: divisions begin very early. oidea * Presently known only in C. borneensis; not found in C. brachiata or C. eugen due to localized growth of hypodermal parenchyma and expansion of the overlying epidermis (Fig. 9). Acne the exotegmen is a distinctive cell layer in the unfertilized ovule, it does not persist to form a prominent sclerified layer in the mature seed, as in other inland genera of Rhizophoraceae; instead, the tegmen gradually degenerates and/or is crushed during later seed development. Crossostylis grandiflora Brongn. & Gris. The anther consists of four microsporangia. Periclinal divisions in the outer parietal cell layer produce the endothecium and a primary middle layer, which in turn divides to form two cell layers. Later, the inner parietal cells divide, producing the tapetum and a third middle layer (Fig. 10). Thus, anther wall development follows the basic type. The middle layers degenerate during microsporo- genesis. The epidermis persists to anthesis, al- though tannins are deposited in its cells. A few thick bars of secondary wall are formed by the endothecial cells (Fig. 11). As in other Rhizophora- ceae, the tapetum is glandular, and its cells are binucleate. However, unlike other genera, the ta- petal cytoplasm in C. grandiflora stains more densely with hematoxylin and may include several to many unstained globules of uncertain compo- sition. Microsporogenesis occurs by both simultaneous and successive cytokinesis, but most cytokineses in a single microsporangium occur by only one pattern (Figs. 12, 13). Although any one cell in which simultaneous nuclear division (but not wall formation) has occurred may resemble a cell at the end of the nuclear division of meiosis I, careful examination at various focal levels of hundreds of meiotic figures in man different flowers revealed virtually no cells with tetrahedrally arranged nuclei in "successive" the- cae, such as the one shown in Figure 13, and revealed virtually no binucleate cells or portions of cells in “simultaneous” thecae, such as that shown in Figure 12. The very large number of observa- tions makes us confident that this conclusion is not due to orientation of the cells relative to the plane of sectioning, but rather reflects actual variability in the meiotic division pattern. In successively di- viding cells, wall formation occurs after meiosis I but is not documented here for this genus (see also Fig. 25, Pellacalyx). Pollen tetrads of C. gran- diflora are either tetrahedral or decussate. The mature anthers are strongly reniform, even semi- circular, and dehisce introrsely by longitudinal slits. The ovule is bitegmic. At inception and during early development, the inner integument is bise- riate (Fig. 15). Later, numerous peripheral divi- 1414 Annals of the Missouri Botanical Garden FIGURES 1-6. Carallia xov Basic-type anther wall development. ML, middle layers; Sp, sporog- enous cells; T, tapetum. x 1,140. ere just before anthesis; of the wall layers, only the endothecium (ETC) persists. Longitudinal radial wadii vous at Lii one wall thickening (arrows). X455.— 3. Ovule, early devel- Volume 75, Number 4 1988 Juncosa & Tobe 1415 Gynotrocheae Embryology FIGURES 7-9. (1) and oblique second Ew om (2). cells (arrows) . E Dee Eie se and hypodermis Wn testa sa (T9. —— of wees (Tg) persists at this stage but degenerates later; its outer epidermis does not sc lerify. x sions occur, and it becomes about 20 cells thick by the time of anthesis (Fig. 16). The exotegmen and endothelium are sharply differentiated throughout megagametogenesis. The outer integ- ument is probably biseriate at its inception, al- though this stage was not observed. In the pre- meiotic stages that we studied in which the outer integument (or at least its abaxial portion) was present, its basal half was triseriate and the distal portion was biseriate. During the remainder of ovule development very few periclinal divisions occur in the outer integument, and at anthesis it is only about four cell layers thick. Its outer epidermal cells expand considerably and are filled with tannin. Gynotroches axillaris Blume Collections used for this study came from two sites in southern Sarawak. Study of plants from these and a number of other sites suggests that at least two ecotypes, probably constituting distinct species, are found in this region. This lends strong support to the suggestion by Ding Hou (1958) that, with further study, this highly variable and widely (II above left) is piri biseriate. Outer but its abaxial portion (Ol a nuclei, some in Pii Megag of endothelium (ETL) and differentiated outer epidermis (E V). seed, showing kr a of outer integum tegmen persists (Tg) , but this degenerates as js ier develops. x 20. In this very immature 6. Nucellus with megaspore mother cell (MMC) separated from epidermis by two parietal cells (PC). x 1, 140. 1416 Annals of the Missouri Botanical Garden Volume 75, Number 4 88 Juncosa & Tobe 1417 Gynotrocheae Embryology distributed genus will be amenable to subdivision into several distinct species. Accordingly, brief de- scriptions of the taxa collected are given so that future readers will be able to ascertain to which species the embryological data pertain. Trees found in secondary vegetation on low altitude kerangas sites or peat swamps had smooth leaves about 6— 9 cm long and reddish fruits about 4 mm in di- ameter at maturity, usually largely covered by a cracked periderm. Trees found along the banks of small streams in hill forest at 50-200 m elevation had bullate leaves 20-25 cm long and yellow, noncorky fruits at least 5-6 mm in diameter at maturity. In addition to these nonoverlapping color and size differences, the trees had very different habits and leaf colors. Both species produce aerial stilt roots, which are more numerous on trees grow- ing in very swampy sites. Both taxa observed ap- peared to be dioecious. Absence of developing or mature fruits on trees bearing male flowers suggests that the apparent dioecy did not result merely from a temporal separation of the two types of flowers. Almost all of the embryological data reported herein were derived from study of specimens collected in the hill forest. The anther of Gynotroches consists of four microsporangia. Only one middle layer is present between the endothecium and tapetum; because this layer shares its origin with the tapetum, anther wall development is of the “monocotyledonous” type (Fig. 17). The tapetum is glandular; its cells have two nuclei. In some microsporangia, meiotic cytokinesis was successive, producing isobilateral or decussate tetrads (Fig. 18), but in others, even within the same stamen, simultaneous cytokinesis occurred, producing tetrahedral tetrads (Fig. 19). As with Crossostylis (see above), numerous meiotic figures were examined to insure that we were not misled by an artifact of the plane of sectioning, although only a few cells are shown in the figures. Mature pollen grains in male flowers are binucleate. In female flowers, the anther as a whole develops more or less normally, having at least a partially functional tapetum and an endothecium with sec- ondary wall thickenings (Fig. 20). Meiosis also oc- curs, forming numerous microspores, but the nor- mal sporopollenin wall does not form around them (Fig. 20) Functional ovules of Gynotroches are bitegmic and anatropous. Early in development, both integ- uments are two cell layers thick (Fig. 21). The inner integument thickens to 5-6 cell layers by the time of anthesis, but the outer integument remains only two cell layers thick in most places, except for its expanded micropylar portion (Fig. 22). A funicular (raphe) vascular bundle extends along the adaxial side of the outer integument, but neither integument proper is vascularized. An en- dothelium is formed early in megagametogenesis (Fig. 22). The micropylar half of the nucellus de- generates before gametogenesis is complete. The micropyle is quite short and is formed by both integuments. Occasionally, however, extension of the inner integument makes the micropyle en stomic. The single archesporial cell functions di. rectly as the megaspore mother cell; thus, the ovule is tenuinucellate (Fig. 21). A cell wall forms after meiosis I, and the megaspore tetrad is usually T-shaped. The chalazal megaspore forms a normal 8-nucleate megagametophyte via Polygonum -type development. The synergids are large and pyri- form, and the polar nuclei do not fuse to form a secondary nucleus. Antipodals are formed and per- sist to anthesis (Fig. 22) In male flowers, early ovule development is more or less similar to that in female flowers, although the nonfunctional ovules are somewhat smaller and irregularly shaped compared with the fertile ones of female flowers, primarily due to irregularities in integument development (Fig. 23). A normal mi- cropyle can rarely be seen. No obvious pr ities in megasporogensis were noted in the non- functional ovules, and ER ticos may either proceed normally through at least the four- nucleate stage (Fig. 23) or may be somewhat dis- ordered. Gynoecial development is also modified in e FIGURES 10-16. form (ML); periclinal divisions of inner parietal cells are seen here (arrows) . Ep, epiderm anther wall. Epidermis (Ep) persists, and radial walls of endothecial cells have one or 2. Micro aie haba in a th io x 1,140.—11. Mature two secondary wall thickenings (arrows) . x 265.— were simultaneous. X 1,140—13 ovule with at least two archesporial cells (Ar divided ally to form a Meiosis l in a ms "a mos megaspore mother cell (MMC) Crossostylis grandiflora.— 10. Developing anther wall. Typically, at least three middle layers mis; ETC, endothecium. most meiotic hr ad 4. Early eca where ns were successive. X 1,140.— ovule, in which the Ne oe 2 has zw a parietal cell (PC). x 800.— 16. Ovule at anthesis. Inner integument has thickened considerably; outer integument, only slightly, to about four cell layers. Endothelium is prominent (arrow), and aril primordium (A) has begun to develop. Ap, condensed antipodals. 100. 1418 Annals of the Missouri Botanical Garden FIGURES 17-21. Gynotroches axillaris, anther and early ovule. — 17. Monocotyledonous-type development of anther wall; arrows indicate divisions forming the tapetum and single middle layer. x 1,060. Ep, epidermis; SIC, endothecium.— 18. Meiosis I in a microsporangium where most divisions were successive. X 1,500. — 19. Microsporogenesis in a theca where most meiotic cytokineses were simultaneous. Arrows indicate cells with tetrahedrally arranged nuclei, just prior to wall formation. x 1,500.— 20. Anther of female flower at anthesis. Normal endothecium is formed, but pollen development is arrested before deposition of sporopollenin wall, which would stain darkly in safranin (used here). x 310.— 21. Developing tenuinucellate ovule; archesporial cell functions as megaspore mother cell (MMC) without first dividing periclinally. Both inner (Il) and outer integuments (Ol) consist of two cell layers. x 900. Volume 75, Number 4 1988 Juncosa & Tobe O 1419 Gynotrocheae Embryology FIGURES 22-24. at anthesis me of nucellus has degenerated; endothelium (ETL) is 2 dud diferente Antipodals persist (Ap). Gynotroches axillaris, mature ovule and proembryo.— 425.—23. Sterile ovule from male flower at anthesis. Megagam vule from female flower metogenesis oceeded to the four-celled HB but ovule morphology is abnormal; micropyle was not evident in this or adjacent sections. 425 5. —24. Proembryo forme first division; 2, second division. X 1,0 male flowers. In particular, the style may become necrotic or may elongate abnormally, with the stig- matic lobes Pure. inserted at its base (Juncosa, pers. obs.). Further details of floral development will appear separately. In fertilized ovules of female flowers, both cell layers of the outer integument persist and form a distinctive testa, in which the outer epidermal cells are very large and tanniniferous, and the inner epidermal cells are small and compact, with prom- inent nuclei and dense cytoplasm. This latter layer of cells may play an important role in controlling the process of imbibition. The exotegmen is very prominent and sclerified; radial elongation of some of its cells makes its outline undulate. Several in- ternal cell layers of the inner integument also per- sist but remain unspecialized. Mature endosperm cells contain many large inclusions of undetermined composition. These inclusions stain weakly with safranin and are not birefringent. As in other Rhi- zophoraceae, the embryogenetic cell division pat- tern in Gynotroches is variable. However, in this genus, the overwhelming majority of proembryos ed by vor onagrad-type embryogenesis, which predominates in this genus. I, observed had developed according to the typical onagrad-type pattern (Fig. 24), and only a few exhibited an oblique division of the terminal cell. The mature embryo is straight, and both the hy- pocotyl and cotyledons are well developed; they are of roughly equal length. The provascular cyl- inder in the radicle is not medullated. Pellacalyx species £. ub + h ] ldat were deena Doa studies of p. cristatus Hemsl. Using the scanty available material of P. sym- phiodiscus Stapf, we confirmed that in anther wall, pollen tetrad, and ovule integument characters, it agreed exactly with P. cristatus. Ovules of P. c saccardianus Scort. were examined, and seeds of P. cristatus and P. lobbii (Hook. f.) Schimp. were studied. The anther in species of Pellacalyx has four microsporangia, the exterior walls of which develop according to the basic-type pattern. The two middle layers degenerate before pollen develops, but the 1420 Annals of the Missouri Botanical Garden "wo w" wq A FIGURES 25-28. — Pellacalyx species.—25. P. cristatus, microsporogenesis by predominantly successive meiotic division (M2); some simultaneous divisions are also seen (arrow). x485.—26. P. cristatus, microsporogenesis by almost exclusively simultaneous cytokinesis. x 485.— 27. P. cf. saccardianus, early ovule. Both integuments initially consist of only two cell layers. Archesporial cell functions directly as megaspore mother cell (MMC); thus, ovule is tenuinucellate. x 900.— 28. Ovule at anthesis. Except for proliferation of micropylar region, outer integument (Ol) remains biseriate, but inner integument (II) has thickened. Nucellus has degenerated; endo- 2. thelium (ETL) is indicated. x31 Volume 75, Number 4 1988 Juncosa & Tobe 1421 Gynotrocheae Embryology epidermis persists to anthesis, although its cells are somewhat collapsed by that time. Secondary wall thickenings are found in the endothecial cells. The tapetum is glandular and composed of binucleate cells. Both successive and simultaneous meiotic cytokineses occur (Figs. 25, 26), producing de- cussate and tetrahedral pollen tetrads, respectively. Generally, only one type of cytokinesis occurred in any one microsporangium, but incongruent di- visions were observed (Fig. 25). Pollen grains are binucleate when dispersed. Anther dehiscence is introrse, occurring by longitudinal slits. Both integuments of the ovule of P. cristatus and P. saccardianus are initially two cell layers thick (Fig. 27), but divisions in the inner layer of the inner integument soon form three, then eight to ten cell layers (Fig. 28). The outer integument remains biseriate. Neither integument is vascular- ized, and both contribute to forming the micropyle. A prominent endothelium is formed during mega- gametogenesis. The single archesporial cell func- tions directly as the megaspore mother cell (Fig. 27); thus, the ovule is tenuinucellate. A cell wall is formed after meiosis I, and the megaspore tetrad is usually linear. The chalazal megaspore forms a normal megagametophyte via Polygonum-type megagametogenesis. Antipodals are formed but be- come condensed by the time of anthesis. The polar nuclei remain separate until fertilization. The syn- ergids are pyriform, with prominent chalazal vac- oles. DISCUSSION Although this study was limited to one of the two inland tribes of the familiar but poorly under- stood family Rhizophoraceae, some of the char- acteristics observed are strongly discordant with embryological dogma and may have significant im- plications in the study of the control of critical stages of plant development. These features are discussed separately at the end of this section. SYSTEMATIC SIGNIFICANCE The data reported here combined with infor- mation from other (Karsten, 1891; Maurit- zon, 1939; Tobe & Hass. 1987a, b, this volume; Juncosa, 1982b, 1984a, b) permit a relatively complete embryological characterization of the family Rhizophoraceae. Reports for African genera of tribe Macarisieae are not yet available, but data from Cassipourea and Sterigmapetalum (Tobe & Raven, 1987b) are taken as representative of that tribe. Despite the occurrence of certain derived char- acter states in some genera, the family is remark- ably homogeneous embryologically. This contrasts somewhat with the wide range of variation in wood anatomy, mature floral morphology, and chromo- some number (Juncosa & Tomlinson, this volume). Allowing for some exceptions, a generalized em- bryological summary is as follows: Anthers tetrasporangiate, wall development ba- sic-type (monocotyledonous-type in Gynotroches), endothecium with few secondary wall thickenings (more numerous in Macarisieae), middle layers de- generate, tapetum glandular, its cells binucleate; microspore cytokinesis simultaneous (or also suc- cessive, in most Gynotrocheae); pollen tetrads tet- rahedral or decussate, pollen binucleate. Ovules anatropous, bitegmic, crassinucellate (tenuinucel- late in Gynotroches and Pellacalyx), integuments two cell layers thick at inception (multiseriate in Carallia and Rhizophoreae), usually thickening s sequently, endothelium present, micropyle diplo- nucellus ephemeral; megagametogenesis chalazal-monosporic (Polygonum-type), synergids pyriform with large chalazal vacuoles, antipodals usually degenerate (persistent in Gynotroches and Pellacalyx). Fertilization porogamous, endosperm initially free-nuclear, embryogenesis variable with- in species. Mature seeds usually with sclerified exo- stomic, tegmen and abundant nonstarchy endosperm; em- bryo chlorophyllous, straight, with prominent cot- yledons and earth radicular vascular cylinder usually medullated; g type. For most of these characters, the states given epigeal or Durian- above are merely the most common ones among dicotyledons and are regarded as plesiomorphic and therefore not phylogenetically diagnostic. How- ever, the presence of an endothelium is relatively rare in bitegmic ovules (Kapil & Tiwari, 1978). Also, variable orientation of the first and second embryogenetic cell divisions has almost never been reported in families other than Rhizophoraceae, although it may be common in woody plants with large proembryos (Nast, 1941; Souéges et al., 1967) ed temperate pni it is apparently common only in Dipsaca u , 1957). Nu- cellar ae which occurs in all Rhizopho- raceae yet studied, is of some diagnostic value. Sclerified exotegmen is not common (Corner, 1976). e new data reported here support the con- tention that Rhizophoraceae and Anisophylleaceae are not related (Juncosa, 1982a; Juncosa & linson, 1987, this volume; Tobe & Raven, 1987a). This latter family differs sharply from the former in all of the apomorphic character states just enu- merated (endothelium, degenerate nucellus, exo- 1422 Annals of the Missouri Botanical Garden tegmen) as well as in others that are of more general occurrence in dicotyledons. Also, as previously pointed out (Tobe & Raven, 1983), Rhizophora- ceae are embryologically out of place in either Myrtales or Cornales, two orders in which the fam- ily has often been placed by phylogeneticists (Cron- quist, 1968; Thorne, 1976; Takhtajan, 1980). Assignment of the Rhizophoraceae to nus own order (Cronquist, 1981; Dahlgren, 1983) merely raises the question of the proper placement of that order. As suggested and discussed by Dahlgren (this volume), the alignment of the Rhizophoraceae with the Celastraceae, Elaeocarpaceae, Erythroxyla- ceae, and possibly other families is well supported by the embryological data. These data also have clear implications for in- trafamilial systematics. Among the ten genera of Rhizophoraceae so far investigated, only Gyno- troches and Pellacalyx have tenuinucellate ovules and an outer integument that remains two-layered throughout development; the histology of the seed coats in these two genera is similar and differs from that of all other genera in the family (see also Tobe & Raven, this volume). These embryological syn- apomorphies also correlate with distinctive char- acteristics of the inflorescence and floral anatomy that are unique to these two genera, such as fas- ciculate monochasial inflorescences, multiovulate carpels, and idioblastic laticifers (Juncosa & Tom- linson, this volume). Therefore, we conclude that Gynotroches and Pellacalyx are sister genera, de- spite the pronounced differences in mature floral morphology that have led some to question even the inclusion of the latter in the family. As judged by embryological criteria, the system- atic position of Crossostylis is likely to be inter- mediate between the Macarisieae and the Gyno- troches/Pellacalyx clade. The outer integument of the mature ovules of Crossostylis is about four cell layers thick, as in Cassipourea and Sterig- mapetalum (Juncosa, 1984a; Tobe & Raven, 1987b). Further, seeds of Crossostylis are arillate (Fig. 16; Corner, 1976; Smith, 1981; Tobe & Raven, this volume); seeds of Macarisieae are in- variably winged or arillate. This suggests the pos- sibility of merely reassigning Crossostylis to the Macarisieae. However, the number of secondary wall thickenings of endothecial cells of Crossostylis and other Gynotrocheae is quite low (one to three), whereas five or more such bars are found in Cas- sipourea (Juncosa, pers. obs.). Also, microsporo- genesis in Crossostylis grandiflora occurs by both simultaneous and successive cytokinesis, a circum- stance reported here for Gynotroches and Pella- calyx but otherwise unknown in the family. Fur- thermore, there are many important vegetative and floral morphological characters common to Cros- sostylis and other Gynotrocheae that do not occur in Macarisieae (Juncosa & Tomlinson, this vol- ume). Therefore, a reassignment of Crossostylis based solely upon the outer integument and ari seems incorrect, despite the importance tradition- ally ascribed to this character. Carallia seems also to be phylogenetically in- termediate, standing between the ide ip ca Pellacalyx clade and the mangrove tribe (Rh zophoreae). Within the Gynotrocheae, only Car. allia has a prolonged anther connective and acute sterile tips on the two halves of the anther, char- acters also found in some Rhizophoreae (Juncosa & Tomlinson, 1987). Nonappendaged seeds and baccate fruits are found throughout the “higher” Gynotrocheae (that is, excepting Crossostylis) and Rhizophoreae and thus do not help clarify rela- tionships among these seven genera. Seed coat anatomy of C. eugenioidea resembles that of other Gynotrocheae (Tobe & Raven, this volume), but that of C. borneensis is vascularized and quite thick, like those of the Rhizophoreae. The seed coat of C. brachiata is intermediate between these two in at least some respects (Tobe & Raven, this volume). The seed coat apomorphies of C. bra- chiata (degenerate tegmen) and especially C. bor- neensis may be homoplastic with those of the Rhi- zophoreae, but it is also possible that the genus Carallia is paraphyletic. Embryological data do not resolve this question, but Carallia and the Rhizophoreae do share several floral synapomor- phies (Juncosa & Tomlinson, this volume and un- publ. obs.), and only one putative autapomorph distinguishes the genus Carallia (stalked extrastip- ular glands). This character is known to occur only in C. longipes (Ding Hou, 1960) and C. borneensis (Juncosa & Tomlinson, this volume), but other species have yet to be examined critically; if these glands prove not to be ubiquitous in the genus, its monophyly will be questionable. Wit the possible exception of the vascularized integu- ment of C. borneensis, we observed no embryo- logical peculiarities in Carallia that could be con- strued as any kind of preadaptation to vivipary, which therefore appears to have arisen entirely within the mangrove tribe. DEVELOPMENTAL IMPLICATIONS It is generally believed that nearly all funda- mental embryological characters are invariant for a given species (Davis, 1966). For example, the extensive literature on angiosperm embryogenesis Volume 75, Number 4 1988 Juncosa & Tobe Gynotrocheae Embryology 1423 is based upon this supposition (Schnarf, 1929; Jo- hansen, 1950). However, variability in cell division pattern has been reported in two genera with rel- atively large embryos (Juglans: Nast, 1941; Lau- rus: Soueges et al., 1967) and is characteristic of d ag acini yet studied (Juncosa, 1982b, 4a, b). Interestingly, among Rhizophoraceae, = genus that most consistently exhibits the fa- miliar onagrad-type pattern (as traditionally de- scribed) is Gynotroches, which has quite small em- bryos in comparison with those of other genera. This suggests a correlation between small embryo size at cotyledonary initiation and consistency in orientation of cell divisions that has not previously been recognized. Unfortunately, most of the em- bryogenetic literature pertains to temperate, her- baceous species with small seeds and embryos, so it is not clear from this limited example whether this correlation is generally true or not. nother embryological characteristic that may have developmental implications is variability in meiotic cytokinesis in the anthers of all Gynotro- cheae except Carallia. This has not previously been reported in any angiosperm, although the variation in pollen tetrad configuration that has occasionally been noted (Davis, 1966) suggests that variable cytokinesis may occur widely. However, it is not merely the occurrence of this variation in Gynotrocheae, but especially the pattern of vari- ation that is developmentally significant: nearly all meioses in a single microsporangium occur by only one cytokinetic pattern, yet adjacent sporangia in a single stamen may exhibit different patterns (Figs. 12, 13, 18, 19, 25, 26). This suggests strongly that the pattern of cytokinesis is controlled by tapetal secretions, not by the individual microspore urther investigation of this system cant new insights into the mother cells. is certain to yield signific process of meiosis in angiosperms. LITERATURE CITED BEHNKE, H.-D. 1982. Sieve-element plastids of Cyril- laceae, Erythroxylaceae, and Rhizophoraceae. Pre- sentation and significance of subtype-PV plastids. Pl. Syst. Evol. 141: 31-39. 1976. Seeds of Dicotyledons. Cam- bridge Univ. Press. CRONQUIST, A. 1968. The Evolution and Classification of Flowering Plants. Houghton Mifflin Co., Boston, Massachusetts. An Integrated System of Classification of Flo owering Plants. Columbia Univ. Press, New York. jig pe R. 19 General aspects of angiosperm volution and macrosystematics. Nordic J. Bot. 3: 1 19- 149. Rhizophoraceae and Anisophylleaceae: sum- mary statement, relationships. Ann. Missouri Bot. Gard. (this volume). Davis, C. L. 1966. Systematic Embryology of the An- giosperms. Wiley, New York. Dinc Hou. 1958. Rhizophoraceae. Flora Malesiana 5: 429-493. 60. A new species of Papaya Roxb. (Rhi- zophoraceae). Nova Guinea, Bot. -23. FLORET, J.-J. 1976. A propos de Tum ton gabo- nense (Rhizophoraceae— Macarisieae). Adansonia, 1954. Flora of West . 2, 16: HUTCHINSON, J. 8 M. DALZIEL. Lon Jonassen, D. A. 1940. A Man Me Plant Meraai: SE Hill, New 50. Plant Embryology. Chronica Botanica, A Uat Massachuse 1 Embryo and Seedling Devel- h.D. Sid ud Duke University, Durham Developmental le ‘of the em- bryo and seedling of “ag redi mangle L. (Rhi- zophoraceae). Amer. J. Bot. 69: 1599-1611. 1984a. Embryogenesis and seedling devel- opment in Cassipourea elliptica p Poir. (Rhi- zophoraceae). Amer. J. Bot. 71 l Embryogenesis i Pg morphology of the seedling in Bruguiera exaristata Ding Hou. Amer. J. Bot. 71: 180- 191. P. B. TOMLINSON. 1987. Floral devel t in mangrove Rhizophoraceae. Amer. J. Bot. 74: 1263-1279. . To OMLINSON. Systematic and biological he integumentary tapetum. Bot. Rev. 44: 457- KARSTEN, C. 1891. Uber die Mangrove- bacs d im Malayische Archipel. Biblioth. Bot. 22: Manco, H. F. 1935. Systematic anatomy r ic vods of t M ro vss rop. Woods 44: 46-69. Menos, J. 19 Contributions to the embryology of the orders Rosales and Myrtales. Acta Univ. Lund. 2. 35: 1-121. Nast, C.G. 1941. The embryogeny and seedling mor- phology of Juglans regia L. Lilloa 6: 163-206. Riptey, H. N. 1922. The Flora E = = Peninsula, Volume 1. L. Reeve & Co., Sc | A. F. W. 1893. Rh ee PON In: A. ngler & K. Prantl, Die Natürlichen coqueta: Pt le 3: 42-56 SCHNARF, K. Embryologie der Angiospermen. Pflanzenanatomie, Volume 10. Ge- rlin. r SMITH, A. C. 1981. Flora Vitiensis Nova, Volume 2. Pacific Tropical Botanical Garden, Lawai, Kauai, Ha- wail. SouEGES, R. 1957. Embryogénie n Dipsacacées. Dé- bean ws Mw z le Scabiosa col- umbaria L. C t. Rend. Hebd. grees Acad. Sci. ; J. L. GuIGNARD & J. C. MESTRE. 1967. Dé- ve loppement t de l'embryon chez le Laurus nobilis L. (Lauracées) Phytomorphology 17: 225-261 TAKHTAJAN, A. . Outline of the classification of flowering plants (Magnoliophyta). Bot. Rev. 46: 225- 359. 1424 Annals of the Missouri Botanical Garden AE R. F. 1976. A Aue classification. of e Angiospermae. Evol. -106. Tone, H. & P. H. RAVEN. 1983. M embryological analysis of the Myrtales: its definition and charac teristics. Ann. Missouri Bot. Gar = & Systematic embryology of "ui Anisophylleaceae. An: Missouri Bot. Gard. 74: -26. & 1987b. The embryology and re- lationships of Cassipourea and Sterigmapetalum (Rhizophoraceae — Macarisieae). Opera Bot. 92: 253- 264. ————. Seed morphology and anatomy of Rhizophoraceae, and inter- and infrafamilial rela- tionships. Ann. Missouri Bot. Gard. (this volume). VLIET, G. J. C. M. van. Wood anatomy of the Rhizophoraceae. Leiden Bot. Ser. 3: 20-75. ADDITIONAL NOTES ON THE Hiroshi Tobe? and Peter H. Raven? EMBRYOLOGY OF POLYGONANTHUS (ANISOPHYLLEACEAE) AND RELATIONSHIPS OF THE FAMILY! ABSTRACT Polygonanthus is reported here to have a Polygonum-type embryo sac, like Anisophyllea but unlike Combre- tocarpus (with an Allium-type embryo as ee together with the results of comparisons of other embyrological character states reported earlier, indica isolated position for Combretocarpus within the family qe phylleaceae. On the basis of the av nilable Borde data, we suggest that Anisophylleaceae appear distinct from both Rosales sensu stricto and Saxifragales. The family shares many embryological features ris eae i 's and may be regarded, at least for the time being, as constituting a distinct order in that phylogenetic line In the course of our earlier study of the em- bryology of Anisophylleaceae (comprising Aniso- phyllea, Combretocarpus, Poga, and Polygonan- thus), we were unable to determine several important characters for Polygonanthus (Tobe & Raven, 1987a). The collection of an additional sample of Polygonanthus amazonicus has made it possible for us to report supplemental embryo- logical results here. A een we have already discussed the em- bryology and the floral morphology and anatomy of peri seeds (Tobe & Raven, 1987a, 1988a), our new results, together with the analysis of wood anatomical characters made by Dr. Elis- abeth A. Wheeler on the basis of computerized databases and the suggestion by Dahlgren (this volume) that the family belongs in his narrowly defined order Rosales, necessitate further analysis. We have, therefore, returned to the question of the relationships of Anisophylleaceae in the present aper. The fixed female flower buds of Polygonanthus amazonicus Ducke used in this study were col- lected by Bruce W. Nelson at Maue, Amazonas, Brazil (voucher J. L. Zarucchi 3138, MO) and fixed in FAA. Microtome sections for observations were made following the standard methods dis- cussed in the previous paper (Tobe & Raven, 1987a) EMBRYO Sac FORMATION IN POLYGONANTHUS As previously reported, the ovule is anatropous and crassinucellate. At least one parietal cell is observed above a megaspore mother cell, and the occurrence of periclinal cell division in the nucellar apical epidermis is also confirmed (Fig. 1). The megaspore mother cell divides into two cells, with the upper micropylar cell much smaller than the lower chalazal cell (Fig. 2). Subsequent division occurs only in the chalazal cell, giving rise to a triad of megaspores (Fig. 3). Only the chalazal megaspore functions, developing into a monosporic eight-nucleate embryo sac; therefore, embryo sac formation of Polygonanthus conforms to the Po- lygonum-type, in agreement with that of Aniso- phyllea but not with that of Combretocarpus, which has a bisporic Allium-type embryo sac (Tobe & Raven, 1987a). With respect to other embryological characters, ! This study was supported by grant BSR-85 18902 from the U.S. National Science Foundation to P. H. Raven. We are grateful to Bruc e W. Nelson for collection of the ^y Met iun material used in this study, to Pieter my, and to Elisabeth A. lion | Sd n wood a ANN. Missouni Bor. Garb. 75: 1425-1428. 1988. 1426 Annals of the Missouri Botanical Garden URES 1-3. Longitudinal sections of young ovules of Polygonanthus showing m — 1. Me spore esp e ll db d — 2. Dyad of megaspores.— 3. Triad of megaspores. Abbreviations: mc, megaspore ke er cell; c, megaspore; fc, functioning sh cae aah Arrows in Figure 3 indicate dege a megaspores. Scale bars equal 50 um, 10 um, and 10 um, respective Volume 75, Number 4 1988 Tobe & Raven Embryology of Polygonanthus 1427 Polygonanthus and Poga share many plesio- morphies, most of which are common even to An- isophyllea (see cladogram in Tobe & Raven, 1987a, fig. 71). In view of these features, the occurrence of Polygonum-type embryo sac for- mation in Polygonanthus (also in Anisophyllea) indicates an isolated position for Combretocarpus within the family. Combretocarpus is character- ized by many apomorphies, including Allium-type embryo sac formation. Polygonanthus, like Poga, seems to be a relictual genus whose embryological features were mostly inherited from ancestral An- isophylleaceae (Tobe & Raven, 19872). RELATIONSHIPS OF ANISOPHYLLEACEAE We discussed the relationships of Anisophylle- aceae (and Rhizophoraceae) earlier (Tobe & Ra- ven, 1987a, b, 1988a), as have other authors in this symposium (e.g., Juncosa & Tomlinson, this volume). Historical views on these relationships are reviewed in these papers, and it seems unnecessary to repeat them here. Instead, we shall use recent suggestions (Tobe & Raven, 1987a, 1988a; Baas, pers. comm.; Dahlgren, this volume) as our point of departure. Anisophylleaceae have traditionally been as- signed to Rhizophoraceae as a subfamily or a tribe. The close resemblance with Rhizophoraceae (par- ticularly with Carallia) has been strongly sup- ported by evidence from wood anatomy (van Vliet, 1976; Baas, pers. comm.). Additionally, the result of the computer search by Dr. Wheeler, which incorporated wood anatomical data of about 5,000 dicotyledonous species representing all major and many minor woody genera, confirms that Aniso- phylleaceae agree completely with Carallia and largely with Crossostylis and Gynotroches; all three of these genera are Rhizophoraceae sensu stricto. In terms of wood anatomy, therefore, Rhizophora- ceae undoubtedly agree most closely with Aniso- phylleaceae, and Baas (pers. comm.) suggested that this evidence rules out many other families as close relatives. Nonetheless, overall evidence from many other lines of investigation, including embryology, makes it absolutely clear that Rhizophoraceae and Anisophylleaceae belong to different evolutionary lines (Tobe & Raven, 1987a, 1988a; Dahlgren, this volume). What then are their relatives? Dahlgren (this volume) proposed that Rhizophoraceae be place in Celastrales along with Celastraceae and Elaeo- carpaceae, and we agreed with this suggestion on the basis of embryological evidence (Tobe & Ra- ven, this volume). It seems, therefore, to be the best available hypothesis. Concerning the affinities of Anisophylleaceae, we proposed Myrtales on the basis of embryological evidence (Tobe & Raven, 7a); in contrast, Dahlgren (this volume) sug- gested that they belonged in Rosales sensu stricto. On the basis of our analysis (Tobe & Raven, 19883), Dahlgren concluded that the floral structure of Anisophylleaceae agreed completely with that of Rosales sensu stricto Rosales sensu Dahlgren (= Rosales sensu stricto in the following discussion), for comparison with Anisophylleaceae, comprise Crossosomataceae, Rosaceae, Malaceae, Neuradaceae, and Amygda- laceae (Dahlgren, 1983, this volume). Crossoso- mataceae have often been placed in Dilleniales (e.g., Melchior, 1964; Takhtajan, 1980). On the basis of embryological evidence, Kapil (1970) sup- ported this traditional treatment and rejected the inclusion of Crossosomataceae in Rosales. Despite this, we are not aware of any essential difference in embryological features between Crossosomata- ceae and Rosaceae, and therefore disagree with Kapil's view. Except for Crossosomataceae, these families are closely related; they are often grouped into a broadly defined family Rosaceae (e.g., Thorne, 1983). Among them, Crossosomataceae, Rosa- ceae, and Malaceae are relatively well known em- bryologically, but Neuradaceae and Amygdalaceae are poorly known. The embryological features of Rosales sensu stricto, on the basis of available data, are surveyed in our paper on the embryology of Rhabdodendraceae (Tobe & Raven, 1988b), which is also assigned by Dahlgren (1983) to Rosales sensu stricto. If we compare the embryological features of Anisophylleaceae (see Tobe & Raven, 19872, for data) with those of Rosales sensu stricto (see Tobe & Raven, 1988b, for data), we find that although Anisophylleaceae share many embryolog- ical features with Rosales sensu stricto, the family is distinguished from Rosales sensu stricto in having vascularized integuments, no hypostase, no persis- tent nucellar tissue in the mature seed, a two-cell- layered thin inner integument (mostly thicker in Rosales sensu stricto), no obturator, and no en- dosperm in the mature seed. These embryological features suggest strongly that Anisophylleaceae, even though there are some points of similarity to Rosales sensu stricto, do not belong in that order. e have searched for combinations of embry- ological features similar to that found in Aniso- phylleaceae among groups related to Rosales sensu stricto such as Saxifragales sensu Dahlgren. This order comprises 11 families, including several fam- ilies of **Glossulariineae," a group to which Cron- quist (1981, 1983) considered Anisophylleaceae 1428 Annals of the Missouri Botanical Garden to belong. We also considered Cunoniales sensu Dahlgren (five families) as possible relatives. We found that Anisophylleaceae are clearly distinct from Saxifragales in lacking the Cellular- or He- lobial-type endosperm formation and a persistent endosperm in the mature seed, characteristic fea- tures of Saxifragales. As regards Cunoniales, the fi nstituent families (Cunoniaceae, Baueraceae, Brunelliaceae, Davidsoniaceae, and Eucryp iaceae) are so poorly known embryologically that an adequate comparison with Anisophylleaceae on this basis is not possible at present. As discussed above, Anisophylleaceae differ strongly from Rosales sensu stricto and Saxifra- gales in their very different combinations of em- bryological features. Once more, we emphasize that Anisophylleaceae, on the basis of their embryolog- ical features, most closely resemble Myrtales (see Tobe & Raven, 19872), although Anisophylleaceae do not share the characteristic wood anatomical features of Myrtales (i.e., intraxylary phloem and vestured pitting; see van Vliet & Baas, 1984). Myrtales are generally considered to be allied to Rosales(-Saxifragales-Cunoniales) (e.g., Takhta- jan, 1980; Cronquist, 1981; Dahlgren & Thorne, 1984). Compared with Rosales sensu stricto and with Saxifragales, Anisophylleaceae and Myrtales apparently share at least one apomorphy, namely, the lack of endosperm in mature seeds. In contrast, Rosales sensu stricto and Saxifragales are appar- ently more specialized than Anisophylleaceae and Myrtales in other embryological characteristics. For example, in Rosales sensu stricto the inner integ- ument is thicker than the outer one, a feature found only in derived families (Boesewinkel, 1981); and Saxifragales have a Cellular- or Helobial-type en- dosperm formation, which predominantly occurs in sympetalous groups with a tenuinucellate ovule (Dahlgren, 1975). Therefore, even though the flo- ral morphology of Anisophylleaceae closely resem- es that of Rosales sensu stricto(-Saxifragales) (Dahlgren, this volume; see also Cronquist, 1981, 1983), Anisophylleaceae seem, on the basis of their embryological features, evolutionary line. to represent a different n conclusion, we need further information on the embryology of Rosales sensu stricto and par- ticularly of Cunoniales for more critical compari- son. Meanwhile, however, considering evidence of various kinds, it seems appropriate to conclude that Anisophylleaceae, like Myrtales, are one of the derivatives from the common ancestor of Rosales- Saxifragales~Cunoniales. Anisophylleaceae may or may not be directly related to Myrtales, but they do appear to be at approximately the same evo- lutionary level as that order, at least with respect to embryological features. It might be most ap- propriate, at least for the time being, to regard Anisophylleaceae as constituting a distinct order in this general phylogenetic lineage. LITERATURE CITED Ms eue F. D. 1981. Development of Ovule and Seed Coat in the Rutales-Geraniales Assembly. The- sis. Amsterdam University. Hugo de Vries Labora- torium Contribution No. 9. CRONQUIST, A. An Integrated System of Classi- fication of Flowering Plant. Columbia Univ. Press, . 1983. Some realignments in the dicotyledons. Nordic J. Bot. 3: 75-83. The distribution of characters within an angiosperm system. Bot. Not. 128: 181- 197. General aspects of angiosperm evolu- tion P macrosystematics. Nordic J. Bot. 3 49, Rhizophoraceae and Anisophylleaceae: sum- mary statement, RUM Ann. Missouri Bot. . (this volur R. F. SHORE 1984. The order Myrtales: circumscription, variation, m p USE Ann. Missouri Bot. Gard. 71: 633 Juncosa, A. M. & P. B. Dum `À historical ne taxonomic synopsis of Rhizophoraceae and Ani phylleaceae. Ann. Missouri Bot. Gard. (this a aang KaPIL, N. 1970. Crossosomataceae. /n: Proceedings of the Symposium on Comparative renin sa of Angiosperm Bull. Indian Natl. Sci. Acad. 41: MELC HIOR, H. 1964. Cuttiferajes. In: H. Melchior (ed- TAKHTAJAN, A. L. . Outline of the so of flowering plants (magnoliophyta). Bot. Rev. 46: 225- THORNE, R. F. 198: 12 new realignments in angiosperms. Nordic J. Bot. 3 5-917 Tope, H. & P. H. Raven. 1987a Systematic embryol- ogy t ps nee Ann. Missouri Bot. Gard. 74 —. b. The embryology and re- lationships of Cassipourea and ddl ar Ar (Rhizophoraceae — Macarisieae). Opera Bot. 92: 25: 264. Floral morphology and evolution in Anisophylleaceae. Bot. J. Linn. Soc. (in s). The embryology and sys- tematic position at Ria (Rhabdodendra- ae). Aliso (in pa e 19 Se [^ Seed morphology and anat- omy of Rhizophoraceae, and inter- and intrafamilial relationships. Ann. Missouri Bot. Gard. (this volume). RP 976. Wood anatomy of the Rhizophoraceae. Leiden Bot. Ser. 3: 20-75. & P. BAAS. 1984. Wood anatomy and clas- ves of "m iii dune Ann. Missouri Bot. Gard. "t VLIET NEW SPECIES AND A NEW COMBINATION FOR PLANTS FROM TRANS-ANDEAN SOUTH AMERICA! Alwyn H. Gentry? ABSTRACT Eight new species from the ates d B of Colombia, Ecuador, and Peru are described, each i different family. The are: Bonamia new spec a (Meliaceae) , Desm megistoc nii (C apantin d Prockia pentamera Maconha nti, Marila deberi (Guttiferae) , lonis glabrata (L acistemataceae) , p cus cirrhifera (Palmae) , and Allophylus dodsonii eases eae) . In addition, Rouchera monsalveae (Linaceae), Carapa a new ton ayama for Albizia auspice = Bitbecellohbium paucipinnatum—is propos In the course of preparing local florulas in west- ern Colombia (A Checklist of Plants of Choco Department, Colombia (Forero & Gentry, in press), Flora de Bajo Calima (Gentry & Monsalve, in prep.), and Ecuador (Flora of Capeira (Dodson & Gentry, in press), Flora of the Rio Palenque Sci- ence Center, Revised edition (Dodson & Gentry, in prep.)) we have encountered the following eight novelties, each in a different family. In addition, a new combination is needed for one of the species included in the Capeira florula. CONVOLVULACEAE Bonamia leonii A. Gentry & Austin, sp. nov. TYPE. Colombia. Choco: Municipio de Riosu- cio, Parque Natural Nacional “Los Katyos," Camino Tilupo Alto via Sautatá, desviando por el camino a Tilupo Salto parte baja, 250-100 m, bejuco, flor lila, 25 Feb. 1976, H. León 525 (holotype, COL; isotype, MO). Fru tex scandens, ramulis tomentosis. Folia ovata a densa, axillaris, lo culo 1-3 cm longo. Sepala 6-7 mm longa, extima elliptica, obtusa, puberula; corolla infanibulformis ca. 1.5 cm longa, extra pilosa. Fructus ignotu Liana, the stems densely tannish-tomentose, be- coming partially glabrescent. Leaves ovate, sharply acuminate, rounded at base, 8-15 cm long, 3-6.5 cm wide; densely tomentose with golden tannish trichomes, these forming a slightly thickened base, erect with curving tips, appearing macroscopically + sericeous; petiole 0.3-2 cm long, tomentose. Inflorescence a compound axillary cyme, rather dense, ca. 3 cm across, the peduncle 1-3 cm long; bracts and bracteoles narrowly oblanceolate, tan- nish tomentose, to ca. 8 mm long. Flowers with the sepals 6-7 mm long, subequal (outermost some- what longer), the outermost elliptic with obtuse tips, the inner sepals ovate with + acute tips, appressed puberulous; corolla (only 1 seen) infundibuliform, lilac, ca. 1.5 cm long, pilose outside, slightly lo the lobes ca. 2 mm long; stamens 5, the anthers 2-2.5 mm long, on slender glabrous 3-4 mm long filaments; styles 2, free to near bur glabrous, the stigmas biglobose. Fruit n Known only from Paus Nena Nacional “Los Katyos" in northern Chocó Department of Colom- ja. This plant was originally identified tentatively as Tetralocularia pennellii O'Donell; however, it does — Science Fin ork in western Ecuador was in collaboration with C. Dodson and supported by the U.S. National nce ' Foundation (INT-7906840; BSR-8342764). Fieldwork in Chocó was in collaboration with E. Forero y funde d by Colciencias and the U.S. National Science Foundation (OIP-7518202; INT-7920783) . collaboration with M. Monsalve, has been aided by Cartón de Colombia, with de Cauca. Fieldwork in Tumbes, Peru was funded by the Mellon Foundation. I thank C. Paba S. McDaniel, B. Styles, J. rin and R. Bernal for commenting on the sec ‘tions of this paper related to their respective taxonomic spe ectaliti ° Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166, U.S.A. ANN. Missouni Bor. GARD. 75: 1429-1439. 1988. 1430 Annals of the Missouri Botanical Garden not closely resemble that monotypic genus and clearly belongs to Bonamia. In Myint & Ward's key to Bonamia (Phytologia 17: 121-239. 1968) this new species will key out with three species that they recognize from Southeastern Brazil — B. agrostopolis (Vell.) Hall. f., B. burchellii (Choisy) Hall. f., and B. tomentosa Hassler. All of these have the upper leaf surface densely pubescent and belong to sect. Trichantha. Recently, Austin & Staples (unpubl.) examined the types of these Bra- zilian names and concluded that they represent a single variable species that should be known as B. agrostopolis. This Brazilian species differs from B. leonii in having slightly smaller leaves; scalelike, linear, caducous bracts only 2-3 mm long rather than oblanceolate, persistent bracts ca. 8 mm long; corolla 2-3 cm long filaments; and a reniform to sub-bilobed rather than biglobose stigma. Bonamia leonii appears to be most closely re- lated to B. trichantha H. Hallier of northwestern South America and Mesoamerica and to B. apu- rensis Austin of Amazonian Venezuela. Although B. trichantha and B. leonii overlap geographi- cally, they appear to be ecologically separated, with B. trichantha usually found in drier, more seasonal habitats. Bonamia trichantha differs further in having white flowers, more glabrescent leaves with glabrous or very sparsely puberulous upper sur- faces, and densely glandular pubescent rather than glabrous filament bases Allopatric B. apurensis is the only Bonamia described (Flora de Venezuela 8(3): 40. 1982) since Myint & Ward's monograph. That species differs from B. leonii in the more broadly ovate leaf shape, cordate base, obtuse apex, more gla- brescent upper leaf surface, simple capitate stigma, and especially the larger, more openly corymbose inflorescence with elongate peduncles (12-14 cm long in the MO isotype vs. 1-3 cm long in B. leoni). It is a pleasure to dedicate this distinctive species vs. 1.5 cm long; villous lower to its collector, Henry Léon, who made extensive collections in the Katios Park area of northern Chocó Department, where it is apparently endemic. FLACOURTIACEAE Prockia pentamera A. Gentry, sp. nov. TYPE. eru. Tumbes: Huasimo, Quebrada Ucumares, 550 m, 12 Feb. 1976, T. Plowman 5443 (holotype, USM; isotype, GH, photocopy MO). Prov. Contralmirante Villar, Arbor parva. Folia late ovata, acuta vel breviter acu- minata, ad basim truncata vel cordata. Flores 2-2.8 cm diametro, sepalis 5, ovatis, ca. 10 mm longis, petalis 5, dense tomentosis, stigmate 5-lobato. Small tree 5-6 m tall, the bark rough, dark brown, the branchlets appressed puberulous when very young, soon glabrescent except at nodes, len- ticellate, the stipules tiny and apparently (only one seen) very early caducous, linear, less than 1 mm long, with a pair of thick yellow glands in lower half. Leaves ovate to broadly ovate, acute to short- acuminate, the base truncate to broadly and shal- lowly cordate, with 2-4 basal glands at petiole insertion above, membranaceous, glabrous above except for small appressed trichomes on midvein, below rather sparsely hirtellous to glabrous over surface, persistently pubescent at least in and above axils of lateral nerves, serrate, 2.5-15 cm long, 1.3-12 cm wide, 5-nerved from base; petiole 0.5— 3.5 cm long, pubescent with appressed or erect trichomes. Inflorescence of 2-3 flowers at end of lateral branches, the slender peduncle ca. 5 cm long, the pedicels 0.5-1 cm long, puberulous with mostly subappressed trichomes. Flowers green when fresh, the sepals five, densely grayish tomentose, ca. 10 mm long, 3-8 mm wide; the petals 5, narrowly oblong, acute, rather densely tomentose, about as long as sepals; stamens inserted on re- ceptacle, the filaments glabrous; ovary subglobose, glabrous, the style distinctly 5-lobed. Fruit not seen. ca. 4 mm long, the stigma Endemic to the now mostly destroyed dry forest of southwestern Ecuador and extreme northwest- ern Peru. Additional specimens examined. ECUADOR: GUAYAS: 25 km SE of Cherrelique, 600 forest along Quebrada Los een 4"9' S, 80? June 1987 (st), Gentry & Díaz 58225 MO USM); Cerros de Amotape, Quebrada Los Conejos ca. 25 km SE of Cherrelique, 820 m, premontane moist forest, 4°9’S, 80°37'W, 9 June 1987 (st), Gentry & Díaz 58245 (MO, USM). Prockia pentamera, only the third species of Prockia, is very distinct from its closest relative, P. crucis L. Prockia crucis (fide Sleumer, 1980 is extremely polymorphic but always has smaller flowers (8-14 mm in diameter, with sepals and petals 4-7 mm long) with 3 (rarely 4) sepals, 3 petals (sometimes none by abortion), and a 3-lobed stigma, whereas all these parts are in 5s in P. pentamera. Prockia pentamera is the only pen- Volume 75, Number 4 1988 Gentry 1431 New South American Species tamerous species of Prockia, necessitating changes in Sleumer's ther, the leaves of P. pentamera are generally larger than in P. crucis, although only the largest exceed the largest extremes (to 10(-15) x 5(-10) cm) of the latter species. In P. crucis, the stipule is rather foliaceous, 5-8(-17) mm long, and per- sistent, unlike the minute (less than 1 mm long), early caducous stipule of P. pentamera This species was first collected as a sterile tran- sect voucher at Capeira, near Guayaquil. During fieldwork for the Florula of Capeira (Dodson & Gentry, in press), we were unable to discover it in fertile condition despite repeated visits to the single tree. I tried comparing this sterile material in sev- eral herbaria with Prockia and related genera of Flacourtiaceae (as well as with Morus to which it has a superficial resemblance), but was unable to match it and decided that it must be undescribed. When first found, there were numerous seed- lings under the single capeira tree, but in 1985 both tree and seedlings were burned in one of the fires that devastate the remnants of dry forest in coastal Ecuador during the dry season. It seemed (1980) generic circumscription. Fur- possible that an undescribed species had gone ex- tinct. However, in the Universidad de San Marcos herbarium (USM) in Lima, I came across the fertile Plowman collection, which is here designated the type of P. pentamera. Subsequent fieldwork in Tumbes, Peru, shows that P. pentamera can be locally very common in dry forest remnants. Indeed, it turns out to be the tenth-commonest species in a study site in the Cerros de Amotape, where there were eight indi- vidual plants of this species at least 2.5 cm dbh in a 0.1-ha sample; the largest tree measured 17 cm dbh (Diaz & Gentry, in prep.). GUTTIFERAE Marila parviflora A. Gentry, sp. nov. TYPE. Co- lombia. Valle: Bajo Calima, ca. 10 km N of Buenaventura, Carton de Colombia conces- sion, transition between tropical wet and plu- vial forest, ca. 50 m, 3?56'N, 77?08'W, H. Mazuero 47 (holotype, CUVC; isotype, MO, fragment and photocopy, IBE). Figure 1. ul Folia oblongo-ovata, acuta ue breviter acumi- Inflorescentia spicato-racemosa 3 cm longa, pedicellis 1-2(-3) mm longis. Flores 2 ongi, petalis mm mm longis, caducis, staminibus numerosis, ovario glabro. Tree. Leaves simple, opposite, rigid-coriaceous, oblong-ovate, acute to short-acuminate at apex, obtuse at base, 14-19 cm long, 6-10 cm wide, drying dark gray-brown above, tan below, densely appressed-puberulous below on main veins, sparse- ly and + glabrescently so over surface, the sec- ondary nerves almost at right angle to midvein, 23-25 on a side, 4-8 mm apart, anastomosing with a strong submarginal collection vein; petiole 1.5-2.5 cm long. Inflorescences spicate-racemose, usually 3 per axil, the slender rachis 9-13 cm long, ca. 1 mm diam., puberulous with suberect trichomes, the pedicels 1-2(-3) mm long. Flowers tiny for the genus, 2 mm long, the sepals subap- pressed puberulous, 2 mm long; petals caducous, thinly membranaceous, strap-shaped, 2 mm long; stamens many, free, about as long as sepals, the minute anthers subglobose with the connective thick and patelliform-glandular; pistil ca. 2.5 cm long, the ovary ovoid, glabrous, the style linear, the stigma truncate, subcapitate. Known only from the type locality, which ap- pears to be in an area of high species richness for the genus. Cuatrecasas (1949) described five new species and a new variety of Marila from the central Pacific coastal region of Colombia. Two of these, M. micrantha and M. geminata, are closely related to M. parviflora by their very small flowers; indeed, M. parviflora and M. micrantha have the smallest flowers in the genus. Unique among de- scribed species of Marila is the almost spicate inflorescence of M. parviflora. Marila parviflora differs from M. micrantha, presumably its closest relative, by having shorter pedicels (1-2 mm vs. 3-6 mm long) and consequently spiciform inflo- rescence, inflorescences several per node (rather than solitary), larger, obtuse-based leaves with over twice as many straight (rather than arcuate-as- cending) lateral veins, and subcapitate stigma. The other close relative of M. parviflora is M. gemi- nata, which has similar multiple inflorescences at each node but distinctly larger flowers (sepals 3.5- 4 mm long) and longer pedicels (4-7 mm long). The leaves of M. geminata differ in being larger and having acute bases and longer petioles and especially in the lateral nerves averaging almost twice as far (8-10 mm) apart: n undescribed species of Marila from Ama- zonian Peru has a similarly spicate inflorescence but is amply distinct from M. parviflora (Mc- aniel, pers. comm.). LACISTEMATACEAE Lozania glabrata A. Gentry, sp. nov. TYPE. Co- lombia. Chocó: north ridge of Alto de Buey above Dos Bocas del Rio Mutata, tributary of 1432 Annals of the Missouri Botanical Garden LEM p Sinis ihe —A. asas — B. Post-anthesis flower with petals fallen. —C. Open nther. Rio El Valle, ESE of El Valle, 200-500 m, cemosa, 6-10 cm lo ga. Flores marronini, sepalis 4, tropical and premontane wet forest, 8 Aug. ovatis, filo singulas ipie Fructus ellipsoideo-trigonus, 1976, Gentry € Fallen 17425 (holotype, dehiscens, semine singular COL; isotype, F, MO, NY, U, UTD). Tree 6 m tall. Branchlets glabrous. Leaves ob- Arbor, ramuli glabri. Folia oblongo-elliptica, acuminata, long-elliptic, acuminate, cuneate at base, mem- glabra, subintegra. Inflorescentia axillaris, peranguste ra- branaceous, completely glabrous above and below, Volume 75, Number 4 1988 Gentry 1433 New South American Species subentire to very inconspicuously serrulate, 9-16 cm long, 3.5-5 cm wide, the tertiary veins per- pendicular to the midvein and + parallel (but not as strikingly so as in other Lozania species); petiole 0.7-1.2 cm long, grooved above, glabrous or very inconspicuously puberulous with a few minute scat- tered trichomes. Inflorescence a long, slender, sub- spicate, axillary raceme, mostly in fascicles of sev- eral per node, 6-10 cm long, sparsely puberulous, the adjacent flowers separated by ca. 5 mm, the pedicels glabrous, ca. 1 mm long, subtended by a bilobed cupule formed by two 0.3-mm-long, sessile, basal bracteoles. Flowers with the sepals 4, ovate, spreading, ca. 1 mm long, maroon when fresh, drying dark brown with a brown-flecked cartilagi- nous apex and margin; petals absent; stamen 1, the short thick filament ca. 0.3 mm long, deeply forked apically, each side with a subglobose anther .3 mm long, the ovary broadly ovoid, ca. 0.5 mm long, glabrous, the blunt apex with 3 slender, reflexed style branches ca. 0.2 mm long; ovary and stamen surrounded by a thick extra-staminal + 4-lobed disk 1 mm across. Fruit irregularly ellipsoid-trigonous, 6-7 mm long and 4-6 mm wide, splitting incompletely at apex into 3 valves, with a single round orangish 5-mm-long and 4-5- mm-wide seed. This is easily the most distinctive species of Lozania, a small genus traditionally assigned to Lacistemataceae but placed by Sleumer (1980) in Flacourtiaceae. It is closest to L. mutisiana J. A. Schultes on account of its very short filament and glabrous pedicels and sepals, although the relatively large fruit is closer to that of Amazonian L. klugii (Mansfeld) Mansfeld. Lozania glabrata is the only species of Lozania with glabrous leaves; the sub- entire leaf margin is also unusual, being approached only by occasional variants of L. klugii. The long- + canes | are the longest in bue genus, lled by those of the 1 type specimen of L. hin anata L. B. Smith (a species merged into L. mutisiana by Sleumer but perhaps better maintained as distinct). Lozania glabrata is unique in the genus in the deeply split filament apex, which gives the illusion of two sta- mens. LINACEAE Rouchera monsalveae A. Gentry, sp. nov. TYPE: Colombia. Valle: Bajo Calima, Pulpapel concession, 100 m, bosque pluvial tropical, 3°55'N, 77°W, 14 Dec. 1984, M. Monsalve 631 (holotype, CUVC; isotypes, MO, and to be distributed). Arbor. Folia obovata, obtuse cuspidata vel acuminata, basim cuneata, margine subtiliter crenulata. Inflores- centia axillaris, fasciculata vel perbreviter racemosa. Flo- res lutei, petalis glabris, filis 10, ad basim connatis. Fructus ignotus Tree 6(-30?) m tall, mostly glabrous, incon- spicuously puberulous on young branches, elenti- cellate, the stipules tiny, 1 mm long, caducous. Leaves alternate, obovate, oblanceolate when young, 2-8 cm long (-15 cm in juvenile state), 1-3.5 cm wide, obtusely cuspidate to acuminate at apex, the base cuneate, the margin finely and conspicuously crenulate, chartaceous to subcoriaceous, drying dark gray or olive gray above, tannish olive below, the secondary and intersecondary veins indistin- guishable, close together, finely parallel, with an inconspicuous collecting vein ca. 0.5 mm from the margin, minutely glandular-punctate below, also with somewhat larger scattered disk-shaped glands, subsessile, the poorly defined petiole 1-3 mm long. Inflorescence a sessile or subsessile axillary fascicle of few flowers, sometimes extended as a contracted raceme or spike to 10 mm long with 2-3(-5)-mm- long peduncle, the flowers subtended by bracteoles 1-2 mm long, these glabrous except the + ciliate margin. Flowers yellow; sepals 5, oblong, 2-3 mm for the + ciliate margin; petals narrowly obovate, contracted at base, to 5 mm long, completely glabrous; stamens 10, the filaments ca. 3 mm long, fused into a ca. 0.5-mm- long basal tube, the anthers flattened-globose, ca. 0.6 mm long; pistil glabrous, the ovary ovoid, ca. ] mm long, the 3 styles separate, each with a discoid stigma. Fruits unknown. long, glabrous except Known only from Bajo Calima. Additional specimen examined. COLOMBIA. VALLE: Bajo Calima, ca. 10 km N of Buenaventura, ca. 3°56'N, 77°08'W, 13 Dec. 1981, Gentry 35624 (CUVE. Common name. Juana se va. The genus Rouchera has been used in a broad sense to include Asian and African lianas with hooked processes for climbing and in a narrow sense to exclude Hebepetalum, which differs in having hairy petals with a clawed base. Rouchera sensu stricto and Hebepetalum (now mostly in- cluded in Rouchera) are South American. This new species is most closely related to R. calophylla Planch., the type of the genus, and to R. parviflora (Ducke) Ducke. These are the only other species with reduced inflorescences and sessile or subsessile axillary flowers. Rouchera calophylla, which oc- curs in central and eastern Amazonia, differs in having more membranaceous narrower, narrowly 1434 Annals of the Missouri Botanical Garden oblong-elliptic leaves with less conspicuously cre- nate margins, better-defined petioles, and much longer (to 1 cm) petals. Rouchera parviflora, which occurs near Manaus in central Amazonian Brazil, has leaves similar to R. calophylla but smaller. According to the original description, it differs from R. monsalveae by its suborbicular, distinctly gland- completely glabrous young branches, and apparently larger flowers ("floribus dimidio brevioribus" compared with R. calophyl- lum). Although I have seen no material of R. parvifolia and the description is incomplete, its habitat (wet forest on sand) and distribution strong- ly argue against conspecificity with the coastal Co- lombian plant. nly two other species of Rouchera are known from coastal Colombia. One is the very different R. humirüfolia (sometimes segregated as Hebe- petalum), which has terminal paniculate inflores- cences and white flowers with hairy petals. In view margined sepals, of the controversy over generic limits, it is inter- esting that at Bajo Calima R. humiriifolia bears the same common name as A. monsalveae. The second coastal Colombian species, R. colombiana Hall., differs according to the original description by its chartaceous leaves, petioles 6-8 mm collecting vein 2 mm from the leaf margin, flowers in a lax short-pedunculate axillary cyme, and es- pecially by the conspicuous persistent stipule 5 mm ong, MELIACEAE Carapa megistocarpa A. Gentry & Dodson, sp. nov. TYPE. Ecuador. Pichincha: Centinela, km 12 carretera Patricia Pilar-24 de Mayo, cima de las Montanas de Ila, 650 m, 26 July 1984 (fl, fr), Dodson, Gentry, Palacios & Zaruma 14492 (holotype, MO; isotypes, MO, QCNE) Figure 2 Arbor 15-20 m. Folia foliolis 7—multi-jugatis, oblongis, atis vel acuminatis, ad basim rotundatis. Inflores- onga, anguste paniculata. staminorum tubo cylindrico. Fruc- tus permagnus, a as apiculatus, 17-29 cm longus. Slender unbranched or few-branched tree 15- O m tall, to cm diam., the branch apices conspicuously bracteate. Leaves with 7 or more leaflet pairs, the petiole and rachis woody, glabrous, finely longitudinally ridged, the leaflets oblong, apiculate to abruptly acuminate at apex, rounded at base, 17-55 cm long, 8-16 cm wide, coria- ceous, completely glabrous. Inflorescence cauliflo- rous, borne usually several together from short shoots on main trunk, 27-50 cm long, irregularly scaly, otherwise glabrous, very narrowly panicu- late, the longest (basal) side branches occasionally to 8 cm long. Flowers slender-pedicellate, white with greenish petals and yellowish nectary, func- tionally unisexual, the calyx 4-lobed to base, the lobes less than 1 mm long, the petals 4, broadly ovate, 4-5 mm long, glabrous, not ciliate, the staminal tube broadly cylindric, not urceolate, 4— 5 mm long, apically split into 8 narrowly triangular acute or acutish lobes, the sessile anthers (or an- therodes) alternating with lobes of staminal column, the pistil (or pistillode in male flowers) with a con- spicuously discoid style-head, the nectary annular- pulvinate, yellowish when fresh. Fruit very large, ellipsoid, not at all tetragonal nor angled and lack- ing verrucose ridges, 17-29(-30?) cm long with- out the 1-3 cm-long apiculus, 13-16 cm diam., the surface uniformly brownish, even when young, covered with dense scalelike papillae. Additional specimens examined. | ECUADOR: LOS RIOS: km 12 road from Patricia Pilar to 24 de Mayo, 540 m 7 Oct. 1976 (fl), Dodson & Gentry 6593 (MO, SEL). PICHINCHA: type locality, 30 Jul 1984 ed oe Gen- ry, Palacios, Zaruma 14676 (MO, QCN Only two species were recognized in Carapa in the recent Flora Neotropica Monograph of Me- liaceae (Pennington & Styles, 1981). One of these, C. procera DC., occurs in the Guianas and Central Amazonia, and in Africa; the other, very hetero- geneous C. guianensis Aubl., is widespread in the Neotropics. One collection of the new species (Dod- son & Gentry 6593) was cited under C. guianen- sis, although the description of C. guianensis dis- agrees with the plant described above in numerous characters. Typical C. guianensis and C. megistocarpa grow sympatrically in our study area in western “cuador, and we are convinced that they cannot possibly be conspecific. At least in our study area, Carapa guianensis is large, freely branching, ram- iflorous (never dev dri) and lacks conspicuous bracts at the branch apices. Its flowers (Fig. 2) have a subtle but al different, dis- tinctively more urceolate shape, with the apices of the otherwise fused filaments bent inward and more truncate and more closely appressed than in C. megistocarpa. The staminal tube of our population of C. guianensis has orange suture lines and the nectary is orange, whereas in C. megistocarpa the staminal tube is uniformly white and the nectary is yellowish. Moreover, the fruits of the two species are consistently and dramatically different, as in- dicated in Table 1. This whole suite of consistent differentiating characters (Table 1, Fig. 2) in two 1435 Volume 75, Number 4 Gentry 1988 New South American Species : (u10210q) el ¡Aydo1sidau :9 pun (do3) stsuauem3 :7 fo suamoy fo uosiuwduos dnaso]) (q — sisusuemi :9 fo 32u22s240jfu1 Sursamozy )— &dreoojstdour 77) fo sasuaosaso fur Futsamozy *g —' (8uo] wo Qg asd [8 19 uospoq]) edueoojsídour edeg ‘g ANNA SIDIYS 21/1 gu uo sisuguem3 77) *;/9] uo edaeoojsidour uw -a us ^ xin AE e ts j b a” fua or 9) simi Y (C0LOP1 ‘Je 19 uospoq) sisuauem3 7) pun (c6tvTI 1436 Annals of the Missouri Botanical Garden TABLE l. Features differentiating Carapa megistocarpa from C. guianensis at the Centinela study site. C. guianensis C. megistocarpa Height mostly 20-40 m 15-20 m freely branchin abit g Bracts at branch lacking or inconspicuous apices Inflorescence ramiflorous on small branchlets orolla strongly urceolate Filament apices truncate, bei strongly inward and closely appressed to eem apices Staminal tube orange suture lin Nectary Fruit shape ly tetragonal Fruit surface when young, each valve usually with a verrucose medial ridge 11-14 cm long and wide! truncate or depressed Fruit size Fruit apex subglobose to depressed-globose and slight- greenish with numerous brown lenticels unbranched or few-branched conspicuous cauliflorous urceolate-tubular narrow, bent slightly inward and barely or not at all touching uniformly white yellowish ellipsoid, not at all tetragonal uniformly brownish, even when young, the valves unridged 17-30 cm long, 13-16 cm diameter apiculate (apicule 1-3 cm long) ! 5-10(-12) x 6-8(-10) cm fide Styles (in Pennington & Styles, 1981). sympatric populations clearly mandates specific recognition. Although B. Styles (pers. comm.) maintains that the characters noted above and in Table 1 are variable elsewhere in the range of Carapa, they are quite constant in western Ec- uador, where C. megistocarpa and C. guianensis clearly pass the test of sympatry. In some characteristics C. megistocarpa is clos- er to C. procera than to C. guianensis, e.g., con- sistently slender-pedicellate flowers and ellipsoid fruit with valves lacking medial excrescences, but the 4- rather than 5-parted symmetry of its flowers and fruits seemed to relate C. megistocarpa de- finitively to C. guianensis. The fruit is the largest in the genus, the minimum length and width both always exceeding the maximum dimensions given for either species in the Flora Neotropica mono- graph, hence the epithet * Carapa megistocarpa was the fourth-common- est species in the Centinela forest, with 18 plants at least 2.5 cm dbh in a 1,000-m? sample area. Its amazing football-like cauliflorous fruits were one of the most characteristic features of this forest, famous for its high local endemism (Gentry, 1986; Dodson & Gentry, in prep.). Sadly, it may now be extinct, since the last remnants of the Centinela ‘megistocarpa.’ ridgetop have been converted to banana planta- tions. PALMAE Desmoncus cirrhifera Gentry & Zardini, sp. nov. TYPE. Colombia. Valle: Bahia Malaga, steep banks at edge of Mora swamp over- looking tidal stream, O m, 4?2'N, 76?15'W, 16 Dec. 1985, A. Gentry. Favino a nd J. Cae 53392 (holotype, CUVC; isotypes, MO(2), COL, K). Figure 3. onsalve, C. utex scandens, sparse spinosus, spinis brevibus cur- vatis. . Folia 25-60 cm longa, foliolis 5-9-jugatis, lanceo- lato- Pune oo. Inflorescentia semel ramosa 15-19 rachillis cm longis. Fructus ellipsoideus vel iban D cm longus, 1-1.5 cm latus Spiny climbing palm (+ erect when young) with stems (including the enclosing leaf sheath) 1-2 cm thick, the leaf sheath with occasional short thick- based spines < 4 mm long, these mostly + re- curved, glabrous except for small, appressed, ir- regularly branched reddish scales, extended above the node into an ocrea ca. 5 cm long, this unarmed or with an occasional small spine. Leaves 25-60 or more cm long, with 5-9 pairs of lanceolate- elliptic to narrowly elliptic, subopposite to definitely alternate, caudate-acuminate pinnae, these 6-21 cm long (not counting the acumen), 2-7 cm wide, the tendril-like linear acumen 4-11 cm long, pen- dent and often somewhat twisted when fresh, the surface glabrous except for minute scattered + peltate scales, these mostly sessile and reddish, sometimes in part stalked, also with minute whitish scales or scalelike enations, the larger leaflets with somewhat larger, irregularly branching, appressed reddish trichomes near margin below, the midvein rominent above and below, the other longitudinal veins relatively inconspicuous, the transverse vein- lets slightly prominulous above in older leaflets; rachis with irregularly scattered thick-based straight Volume 75, Number 4 1988 Gentry 1437 New South American Species P. ( l V UNO ca or recurved spines 2-5 mm long; leaf apex linear and unarmed in juvenile state, in mature plants armed with small, strongly recurved spines 2.5-4 mm long, sometimes a few of these thicker and terminated with a reduced, very narrow, vestigial leaflet 4-6 cm long, this always thin and mem- branaceous; thick, elongate grappling hooks com- pletely lacking at leaf apex. Inflorescence axillary, the peduncular inflorescence bract (spathe) ca- ducous in fruit, the persistent basal part of the prophyll ca. 6-8 cm long, thin, unarmed, fibrous, fragmenting into the individual fibers; peduncle 1-14 cm long, the rachis once-branched with 15-19 rachillae, each 1-3 cm long, the lowermost progressively larger, the flowers not seen, loosely clustered along upper half to two-thirds of rachilla, this flattened and somewhat zigzag or twisted be- tween adjacent flowers. Fruit 1-seeded, red, ellip- soid to subglobose, 1.5-2 cm long, 1-1.5 cm wide, subtended by a sessile 3-lobed cupule formed from the 3 persistent basally fused tepals, this ca. 5 mm across. Endemic to the wet part of lowland coastal Co- lombia in Valle, Chocó, and (fide Galeano & Bernal, 1985) extreme northwestern Antioquia depart- ments. Additional specimens examined. COLOMBIA. VALLE: Bahía Malaga, Quebrada Alegria, trail along proposed route of new road to military base, 50 m, 4%2'N, 77°22' W, 15 Dec. 1985, Gentry, Monsalve, Restrepo & Ganibes 53319 (CUVC, MO). chocó: Taparalito, Quebrada Ta- paral, N of res: Ape wet forest, 30 m, 4?15'N, 77°12'W, 30 Mar 5, Gentry, Zardini, Moneádib & Lape 53795 el MO); Quibdo-Tutunendó road m W of Tutunendó, iuis forest, 80 m, 5?46'N, ds Pu 8 Jan. 1981, Gentry, Mulampy, Hikes, Li- benson, Olson & Cogollo 30363 (COL, MO). Common name. Matamba. The outstanding feature of the plant is its ex- ceedingly long tendril-like leaflet apices. It belongs to Burret's (1934) section Campylacanthium, as indicated by its short curved spines, and it is the first record of that section in the trans-Andean Neotropics. John Dransfield, who examined a du- plicate of one of the sterile collections cited above, identified it as a new species. Subsequently, this species was also recognized as undescribed by Gal- = FIGURE 3. Desmoncus cirrhifera.— A. Leaf and in- yqa wea (Gentry et al. 53392) . — B. Cut stems wait- g to be made into baskets, Taparalito, Chocó, Colom- bia (Gentry et al. 53795).— C. Basket made from D. cirrhifera, Docordó, Colombia. 1438 Annals of the Missouri Botanical Garden eano & Bernal (1985) in their treatment of the palms of Antioquia Department. Again, sterile ma- terial prevented its description. Discovery of fruit- ing material finally makes possible its description. Desmoncus cirrhifera is one of the most utilized palm species of the Pacific coast region of Colom- bia. It is used to make nets and shrimp traps (catangas) in the Bahia Malaga and Rio San Juan delta areas (sub Gentry et al. 53392) and is prized by the Chocó Indians at Taparalito and Docordó who make their strongest baskets from it (Fig. 3C). SAPINDACEAE Allophylus dodsonii A. Gentry, sp. nov. TYPE. Ecuador. Los Rios: Rio Palenque Field Station, halfway between Quevedo and Santo Domingo de los Colorados, wet forest, 200 m, 21 Feb. 1974. Gentry 10098 (holotype, MO; isotype, QCA; additional duplicates distributed as 4. us). cf. amazonic mer F olla unifoliolata, elliptica vel obovato-elliptica, vel subacuta, a cuneata, margine el Inflorescentia axillaris, paniculata, ramis flo- riferis anguste racemosis, ascendentibus. Flores minuti, petalis intra pubescentibus. Fructus ellipsoideus, ca. le longus. Tree 15-20 m tall, the branchlets + glabres- cently puberulous with mostly appressed trichomes. Leaves unifoliolate, elliptic or obovate-elliptic, the apex obtuse to acutish, the base cuneate, 6-28 cm long, 3-12 cm wide, membranaceous, the mar- gin entire or inconspicuously crenulate-serrate near apex, almost glabrous, very sparsely and incon- spicuously puberulous along midvein above an elow, sometimes with few trichomes in axils of lateral nerves, with 10-14 secondary veins on a side, these curved and ascending; petiole 1-2 cm long to petiolule insertion, + appressed puberulous, at least adaxially, apically jointed with flexed petiol- ular leaflet base. Inflorescence axillary, paniculate, the usually 3 elongate branches ascending and narrowly racemose, puberulous. Flowers (seen only in female condition) small, ca. 1 mm long and 2 mm across at full anthesis, white, on pedicels 1— 2 mm long, the very broadly ovate sepals ap- ressed-puberulous, the petals strongly pubescent inside, slightly pubescent outside, usually also cil- ong, the tiny anthers presumably nonfunctional; pistil almost 2 mm long, the style apically forked to form 2 exsert- ed stigmas almost 0.5 mm long. Fruits broadly ellipsoid, ca. 1 cm long, essentially glabrous. iate-margined; stamens ca. Known only from the remnant patch of coastal Ecuadorian lowland wet forest at the Rio Palenque Field Station. Additional specimens examined. ECUADOR. LOS RÍOS Río Palenque Field Station, 16 Feb. 1974 (st), Gentry 9957 (MO), 7 Oct. 1976 (st), Dodson & Gentry 6587 MO, SEL), without date (fl), Dodson 7343 (MO, SEL), 4 Apr. 1980 (fr), Dodson & Gentry 10176 (MO, SEL). ~ This species was treated as Allophylus cf. ama- zonicus (Mart.) Radlk. in the Flora of Río Pa- lenque (Dodson & Gentry, 1978). However, it differs conspicuously from that Amazonian species in the larger fruits, sparsely appressed puberulous branchlets, entire or subentire leaves, and espe- cially the 3-branched rather than simply racemose inflorescence. This was one of very few species in the Rio Palenque flora interpreted as having a trans-An- dean range disjunction, i.e., occurring on both sides of the Andes but not reaching Central America. Thus it is perhaps not surprising that the additional collections now available from both sides of the Andes prove the coastal plant specifically distinct. LEGUMINOSAE Pithecellobium paucipinnatum (Schery) Gen- odson, comb. nov. Albizia paucipin- nata Schery, Ann. Missouri Bot. Gard. 37: 400. 1950. TYPE. Ecuador. El Oro: Portovelo, Steyermark 54035 (MO)... . as “Albizzia.” This is a common tree species of the dry forests of the Guayaquil area of southwest Ecuador, es- pecially in juvenile condition. At Capeira, 20 km N of Guayaquil (Dodson & Gentry, in press), there are about 10 trees at least 2.5 cm dbh per ha in the remnant patch of now highly disturbed dry forest. Although previously unreported from that country, it also occurs in adjacent northwestern Peru. At Capeira its common name is “‘compono”’; in Peru it is called "angolo." This species was described as an Albizia by Schery, in the absence of fruiting material, and related by him to Pithecellobium multiflorum (HBK.) Benth. and P. coripatense Rusby of what is now generally known as Pithecellobium section Arthrosamanea (sometimes segregated as the ge- nus Cathormion). These species grow mostly in swampy or riverine habitats and are characterized by a flattened segmented fruit that breaks apart transversely into numerous lomentlike segments, presumably adapted for water dispersal. Generic delimitation in Mimosoideae is noto- riously difficult, and Schery considered section Ar- throsamanea as related to mostly wind-dispersed Albizia rather than mostly animal-dispersed Pithe- cellobium. Sometimes these taxa are accorded ge- neric rank as Arthrosamanea (Britton & Rose, 1936) or Cathormion (Burkart, 1964). Others Volume 75, Number 4 1988 Gentry New South American Species 1439 (Brenan & Brummitt, 1965) suggested merging these taxa with Enterolobium instead. in nearly all neotropical floristic treatments, the indehiscent-fruited species of this alliance are now- adays included in Pithecellobium sensu lato (Mac- bride, 1943; Woodson & Schery, 1950), following Bentham (1875). Although a more recent summary (Nielsen, 1981) returns these plants to Albizia (along with some other segregates of Pithecello- bium), we prefer to retain the indehiscent-fruited non-wind dispersed relatives in Pithecellobium fol- lowing Barbosa (1984 and pers. comm.) and the long-standing tradition. Vegetatively the taxa of Pithecellobium section Arthrosamanea are characterized by oblong, ses- sile, asymmetric leaflets with strongly ascending, almost palmate venation. Pithecellobium section Arthrosamanea is represented in the Guayaquil area by P. daulense Spruce ex Benth., which is vegetatively strikingly similar to P. paucipinnatum but has more-elliptic leaflets that are glabrous rath- er than puberulous as in P. paucipinnatum. That species, like other bona fide members of section Arthrosamanea, occurs in swampy, poorly drained areas along the Rio Daule at Capeira, while paucipinnatum occurs in the upland dry forest. We were at first inclined to treat the upland plant at Capeira as a variant of P. daulense; how- ever, a collection of its fruits (Dodson & Gentry 12652; MO, GUA, SEL) proves that it is only diltantiy related to P. multiflorum and its allies. Indeed, the indehiscent fruits are quite unlike those of any Pithecellobium known to us, although prob- ably most closely related to the very different but similarly indehiscent fruits of Pithecellobium sa- man. Superficially, there is a surprising resem- blance to the fruits of Hymenaea, especially in the texture, color, and composition of their resinous- owever, secreting surface. The fruits of P. paucipinnatum are 12-14 cm long, 1.7-2 cm pressed, subwoody, reddish brown, glabrous, and pitted on the surface with minute resinous glands. Although obviously indehiscent, these fruits are interesting in that they break transversely rather easily, potentially indicating a shared ancestor with wide, straight, somewhat com- section Arthrosamanea. Since the fruits of this species are completely unlike the thin wind-dis- persed fruits of Albizia, transfer to Pithecellobium seems appropriate. d. ECUADOR. GUAYAS: 79°58'W, 20-150 Additional specimens examine N of Guayaquil 2°S, 7 & Gentry 12652 (MO). MANABI: base of Montecristi, 180 m, Dodson & Thien 1716 (MO). EL oro: Rio Amarillo upstream from Portovelo, Steyermark 54035 (MO). PERU. TUMBES: Zarumilla, Dtto. Matapalo, 550 m, Camino Cau- cho-Campo Verde km 79, I. Canales P. 15 (MO, MOL). LITERATURE CITED BanBosa, C. 84. Revisión Taxonomica de la Sección C BRENAN, J. New and little known species from the Flora Zambesiaca Area. 19: Leguminosae — Mimosoideae. Bol. Soc. Brot., Ser. 2A, 39: 189-205. Britton, N. L. & J. N. Rose. 1936. Mimosaceae and Caesalpinaceae of Colombia. Ann. N.Y. Acad. Sci. 35: 101-208. BURKART, A. 1964. Leguminosas nuevas o criticas VI. Darwiniana 13: 428-448. Burret, M. 1934. Die Palmengattung Desmoncus Mart. Fedde Rep. 36: 197-221. Cu ATRECASAS, J. 1949. Gutiferas nuevas o poco con- as en Mei Anales Inst. Biol. Univ. Nac. México 20: 91-112 Dopsow, C. pa UM 1978. Flora of the Rio Palenque eap Center, Los Rios Province, Ec- uador. Selbyana 4: 1-628. & k Florala de Capeira. Banco Nacional e Ecuador, Quito (in press GALEANO, G. & R. BERNAL. 1985 Palmas del dee GENTRY, À Hime in tropi perate communities. Pp. 153-181 in M. Soulé i The Science of S 507. 2 Po P. Raven st. E Le NIELSEN, I. 1981. Ingea den lhill & P. e Systematics. Royal (e id Advances in oe otanic Gardens, PENNINGTON, T. & B. donis: 1981. Neotropica Monograph 28: 1-4 SCHERY, R. 1950. In: cpi Taxonomica. I. Ann. Missouri Bot. Gard. 37: 400. aor H. ; ds Flora Neotropica nograph 22: 1-499. Migne R. E., ScHERY. 1950. mosoideae. le Pica of Pinawa, Ann. Missouri Bot. 37: 185-314 Meliaceae. Flora 70. STUDIES IN NEOTROPICAL PALEOBOTANY. V. THE LOWER MIOCENE COMMUNITIES OF PANAMA— THE CULEBRA FORMATION' Alan Graham? ABSTRACT lower Miocene Culebra Formation of central í consists of lignites and oe shales in a predom- B estuarian sandstone sequence. Forty-one palynom s have been identifiec m the lignites, and the following ten are most abundant: Monolete fern spore type ? (20%) , Manicaria-type um heiss (10%) , Cyathea (8%) , Cryosophia-type palm = (8%), a (7%) , Synechanthus-type palm pollen (6%), Rhizophora 6%), monolete fern spore type munities were the tr an only moderately developed mangrove swamp) genera), and possibly some form of the o arid habitats (including savannahs similar to those of the p The rainfall probably dec E ed s between about 21C a the general paleote mperature curve derived from '* Nees al data from other Centra distinctly Central m North American, ih al ii, Pollen of t formations from ien Central Ameri 1 (5%), Hampea/Hibiscus (4%), 25% of the fossil px palms 2496, and bold vegetation types totaled 7 tropical moist forest represented by 30 genera t p). premontane wet forest (25 dad ; M wet a (22 premontane moist forest ( tly with elevation, similar to the pattern shown in to about 1,172 m from present- > se Atlantic Costa Rica (3 53.6 nd 32 and Vi ae ha (2%). Ferns constituted most prominent paleocom- hat can occur in this vegetation type (including studies Balada invertebrates, and with emerging paleo- eng floras. 1 e Gramineae and s of en drier habitats continue to be rare or absent in Tertiary The geologic formations of central Panama pro- vide an opportunity to trace vegetational and pa- leoenvironmental history through five segments of Tertiary time. The Gatuncillo Formation exposed near Alcalde Diaz is of middle(?) to late Eocene age, and study of this paleoflora has been completed (Graham, 1985). The Culebra, Cucaracha, and La Boca formations are of lower Miocene age, and the Gatun Formation is upper Miocene/ Pliocene in age (a recent estimate by Vokes, pers. comm., 1988, is that the Gatun Formation may be as young as middle Pliocene). All are known to contain fossil pollen and spores. These studies, together with those of Bartlett & Barghoorn (1973) on Quater- nary vegetation, and information on the modern vegetation (Croat, 1978; D’Arcy, 1987; the re- cently completed Flora of Panama; and the Flora Mesoamericana project) will eventually provide a more complete data base for the vegetation of Panama and its Cenozoic evolution than for any other area in the Neotropics. Concepts on the history of neotropical vegetation are being developed at a time when important new information is also becoming available on other aspects of the biotas. This allows comparison of the paleobotanical data with global paleotempera- ture and sea-level curves (Savin, 1977; Douglas, 1985; Savin et al., 1975; Vail & Har- denbol, 1979; Vail et al., 1977; Haq et al., 1987) and studies on marine invertebrates (Jones & Has- son, 1985) and terrestrial vertebrates (e.g., see papers in Stehli & Webb, 1985). This allows in- ' The author gratefully ee R. H. Stewart, J. L. mmission, for many useful bi A pos for facilitating fieldwork Vasquez, Panama Canal wart, Pastoria Franceschi S., and Numan in Panama in 1963-1964, 1968, 1980, 1983, and 1986. 1 supported by NSF grants GB-11862, DEB-8007312, DEB- 8205926, BSR 8500850, and BSR 8619203. ? Department of Biological Sciences, Kent State University, Kent, Ohio 44242, U.S.A. ANN. Missouni Bor. GARD. 75: 1440-1466. 1988. Volume 75, Number 4 Graham 1441 1988 Culebra Formation—Paleotropical Communities terpretation within an increasingly precise paleo- | SYSTEMATICS physiographic context (Buskirk, 1982; Gose, 1985; Smith, 1985). Other thai a few megafossil studies done earlier in the century (see literature in Graham, 1973, 1979, 1982, 1986), the plant microfossils of the Culebra Formation provide only our second insight 1985; Coney, into lower Miocene vegetation of northern Latin America. The other is an assemblage from the lower Miocene Uscari sequence of Costa Rica (Graham, 19872). THE COLLECTING LOCALITY The Culebra Formation belongs to a complex of lower Miocene strata in the Canal region of Panama (Graham et al., 1985: 489). Material was obtained from a well (Hole No. GH-9) drilled in January 1958 by the Panama Canal Commission. The site was near Gold Hill on the west side of the Canal at latitude 9°02'N, longitude 79°38' W. Fifty-seven samples were taken from lignites or lignitic shales along the 154-m core between levels 491.6 and 377, and 21 yielded pollen and spores. Eleven were selected for study from the following depths (in feet, following the original log data): 377, 407, 415.5, 425, 456, 469.8, 470.6, 488, 490.6, 491, and 491.6. The 57 samples totaled approximately 9.5 feet of coalified lignite and lignitic shale (avg. ca. 2 inches/sample), with numerous narrow bands add- ing another ca. 1.5 feet (9.6%). The rest of the core was mostly sandstone (Graham et al., 1985: 489-49], tables 1, 2), and constituted ca. 103 feet (90.4%). Swift (1977) has studied Holocene sedimentation rates in the Panama Basin, but there have been no studies on the Culebra or other Ter- tiary formations. In general, near-shore sandstones and lignites deposited in an estuarian environment in tectonically active regions accumulate rapidly. In the absence of more precise data, the time span represented by the section is estimated at a few to several tens of thousands of years. Other details on the geology of the Culebra and related forma- tions are given in Graham et al. (1985). MATERIALS AND METHODS Extraction and processing techniques are de- scribed in Graham (1985). Slides are labeled Pan Core, Culebra, with depth and slide number cited. Location of specimens on the slides is by England Slide Finder coordinates (e.g., ESF L-39, 1). All materials are deposited in the palynology collec- tions at Kent State University. Forty-one palynomorphs were identified from the Culebra Formation, and a number of others were recovered whose biological affinities could not be established. Some of the more distinctive and/ or common ones are listed as Unknown types l- 14 in Table 1. Further details of the identification procedures are given in Graham (1985: 507-508). LYCOPODIACEAE Lycopodium (Figs. 1-3). Amb oval-triangu- lar, apices rounded; trilete, laesurae straight, nar- ow, ca. 20 um long, extending to and occasionally branching at spore margin, inner margin entire; distal surface with numerous circular punctae ca. l um diam., proximal surface laevigate; wall ca. 1.5-2 um thick; 35-37 um These spores are similar to the modern L. re- flexum Lam./L. linifolium L. type (cf. Tryon € Tryon, 1982: 811, fig. 22), presently growing in moist shaded habitats in Panama. Similar spores have been reported from the lower Miocene Uscari shale sequence in Costa Rica (Graham, 1987). SELAGINELLACEAE Selaginella (Fig. 4). lateral, amb circular to reniform; trilete, laesurae Spherical to nearly bi- frequently obscured by wall thickness and sculpture elements, straight, narrow, ca. 20-22 um long, extending nearly to spore margin, inner margin entire; echinate, echinae short (ca. 2 um), occa- sionally curved, dense, bases broad; wall ca. 2 um thick (excluding echinae); 26-30 um Microspores of Selaginella vary in size, and specimens from the Culebra Formation are rela- tively small (26-30 um) and thick-walled. They are common in the Gulf/Caribbean Tertiary and have been reported from all formations studied in the present project (Eocene Gatuncillo Formation of Panama, Graham, 1985; Oligocene San Sebas- tian Formation of Puerto Rico, Graham & Jarzen, 1969; lower Miocene Uscari sequence of Costa Rica, Graham, 1987a; lower Miocene Culebra For- mation of Panama, present report; and upper Mio- cene Paraje Solo Formation of Veracruz, Mexico, Graham, 1976). Selaginella is widely distributed in tropical regions in shaded, humid habitats. CYATHEACEAE Cyathea (Figs. 7-9). apices rounded; trilete, laesurae straight, narrow, 14-16 um long, extending to spore margin, inner Amb oval-triangular, 1442 Annals of the Missouri Botanical Garden TaBLE l. Identification and numerical representation of fossil palynomorphs from the lower Miocene Culebra Formation, Panama. Figures are percentages based on counts of 200, except levels 377—425, which are based on 100 due to lower concentration and only fair preservation of the specimens; these levels also contain high percentages of fungal spores. 491.6 491 490.6 488 470.6 469.8 456 425 415.5 407 377 Lycopodiaceae Lycopodium 12 1 = — 3.5 — — 3 4 — — Selaginellaceae Selaginella 13 5 65 65 7 9 2 12 10 5 3 Cyatheaceae Cyathea 9 95 12 7.5 5.5 7 7.5 5 8 == == Pteridaceae Pteris — — 2 _ — 15 — — — — Schizaeaceae Lygodium 2 3.5 — l — — — — Vittariaceae cf. Antrophyum 2 l5 2 2 2 — 1 = = = a Trilete fern spores Type 1 — = 1 = 1.5 — — — 1 — — Type 2 3 - 2 — 1.5 0.5 l — — — — Type 3 — — = — 2 2 = — = — Type 4 — - — - l — — — — — Marattiaceae Danaea š — = - 05 — l — — — Monolete fern spores Type 1 5 8 5 8.5 2.5 4 1 2 = Type 2 15.5. 37 28 15.5 125 19 125 24 12 32 20 Type 3 = = — 0.5 1 — — — — Gramineae = 0.5 — — — 0.5 — — 1 — Palmae Cryosophila-type 7.5 4.5 7.5 5 8.5 9 10 13 10 8 5 Desmoncus-type = — — l 05 — — — — — — Manicaria-type — 8.5 7.5 10 5 22 7.5 10 15 15 15 Synechanthus-type 5.5 10.5 2.5 4.5 2 1.5 12 11 3 30 Aquifoliaceae Ilex — — 0.5 2.5 0.5 2.9. = l — = Chenopodiaceae / Amaranthaceae == = = = l = = = — = = Combretaceae Combretum/ Terminalia — — 0.5 3 = l — — = Compositae = — 0.5 — — — — — — = Connaraceae cf. Rourea — — = 15 05 1 — = = Dilleniaceae cf. Doliocarpus — — — 0.5 05 — — — — Dioscoreaceae Dioscorea/ Rajania — — 0.5 — — — — — — Volume 75, Number 4 Graham 1443 1988 Culebra Formation—Paleotropical Communities TABLE l. Continued. 491.6 491 490.6 488 470.6 469.8 456 425 415.5 407 377 Euphorbiaceae Alchornea — — — = == 1.5 0.5 — — => 1. Sapium = = = — — — 239 c — — — Tetrorchidium — — — l 1.5 — 1.5 E — 1 — Flacourtiaceae Casearia — — — -— 1.5 — — — — — — Leguminosae Mimosoideae Acacia — — — 0.5 — — 0.5 — — — — Malpighiaceae — — — 2 1 — -— e 1 — — Malvaceae Hampea/ Hibiscus 7.5 4 4.5 4 3.5 5 1 2 1 3 3 Myrtaceae Eugenia/ Myrcia — — — 0.5 1.5 — — -— — — — Rhizophoraceae Rhizophora 10.5 4 3.5 8.5 5.5 1.5 2.5 5 9 8 6 Rubiaceae Sabicea — — — — — — 0.5 — — — -— Sapindaceae Allophylus = = l = = 1 0p = — a = Cupania 05 = = = = x CE — = = — Matayba 0.5 — — — 0.5 — — — = -— — Sapotaceae cf. Pouteria — — — — 0.5 — — -— — — =- Sterculiaceae cf. Guazuma — — — — — — 0.5 — — — — Unknown Type 1 - -— = == 095 — = 1 =a == on Type 2 — — 0.5 — — -— 1 — — — Type 3 — 0.5 4.5 4.5 6 11.5 7.5 2 — 4 4 Type 4 _ ae — 0.5 — — — -— — Type 5 = — = = 155 — 2 = == = — Type 6 — — — -— — — 0.5 — — — -— Type 7 — — — — 0.5 -— — — — — Type 8 = E = 05 = 15:5 = = = = Type 9 — — — — — — 0.5 — — — — Type 10 = es £= 0.5 pes 1 1.5 = — = pe Type 11 — — 0.5 2 1.5 5 3.5 1 2 — — Type 12 — = = -— — — 0.5 — — — — Type 13 = — = = 1 = 19 . = — = = Type 14 — — — 1 1.5 1 1.5 — — — — Other unknowns 4 6 9 7 7 8 8 6 14 18 14 margin entire, bordered by lip 2-3 um wide with punctae | um diam.; distal surface finely punctate, proximal surface more laevigate near laesurae; wall 1.5-2 um thick; 32-36 um. The classification of tree ferns differs among various authors, and reference slides with spores similar to the fossil specimens come from herbarium material labeled as Alsophila and Cyathea. As 1444 Annals of the Missouri Botanical Garden Volume 75, Number 4 1988 Graham 1445 Culebra Formation—Paleotropical Communities noted by Tryon & Tryon (1982: 204), “The name Cyathea has been variously applied to genera of widely different scope and definition, sometimes including nearly all members of the family." In illustrations of spores of neotropical tree ferns (Gas- tony & Tryon, 1976; Tryon & Tryon, 1982: 207), the micropunctate forms are referred to Cyathea. The genus consists of 40 species of the American tropics growing primarily in montane forests and cloud forests, and also in low rain forests in Central America, usually at 1,500-2,000 m, but as low s 500 m (Tryon & Tryon, 1982: 204-205). Fossil spores are common in Gulf/Caribbean Ter- tiary deposits. PTERIDACEAE Pteris (Fig. 10). Amb triangular, apices rounded; trilete, laesurae straight, narrow; 18-22 um long, extending to spore margin, inner margin entire; distal surface with coarse, irregular verru- cae, proximal surface more laevigate, flange ca. 2-6 um wide, hyaline; wall 2 um thick; 42-48 Pteris includes about 200 species, with approx- imately 55 occurring in the American tropics (Tryon & Tryon, 1982: 334). The plants typically grow in wet forests or in openings along stream banks at low elevations (sea level to 2,000 m). Fossil spores are frequent, but in low percentages in Gulf/ Caribbean Tertiary deposits. SCHIZAEACEAE Lygodium (Figs. 5, 6, 11, 12, 17). Amb tri- angular to concavo-triangular, apices rounded; tri- lete, laesurae + straight, narrow, 34-38 um long, extending nearly to spore margin, apices frequently branched, inner margin entire; laevigate to faintly verrucate; wall 2-4 um thick; 80-100 um Lygodium is a genus of about 30 species, with six to eight in the American tropics. According to Tryon & Tryon (1982: 71): “Lygodium charac- teristically occurs in open forest especially along the borders where the climbing leaves can reach well-lighted situations. In tropical America, it some- times grows in rain forests, but more commonly in gallery forests, shrubby savannahs or along the borders of streams or river banks. It most com- monly occurs from sea level to about 350 m, as a characteristic element of the low, humid tropics, and less often grows up to 1,000 m." The specimens are large (up to 100 um in di- ameter) and somewhat diverse in morphology; more than one biological species may be represented. The most notable variation is in the dark coloration that can occur at the apices. This is not found in our modern reference material, but in the fossil specimens it ranges from absent (Figs. 5, 6) to quite prominent (Fig. 17). Spores of Lygodium, but of a different morphological type, have been reported from the upper part of the lower Miocene in northern South America (Germeraad et al., 1968) under the artificial generic name Crassoretitri- letes. Spores similar to the Culebra specimens are also known under the name Matonisporites de- scribed by Couper (1958; see especially pl. 20, figs. 15, 16) from the Jurassic and Lower Creta- ceous of England. Dettmann (1963, pl. 11, figs. 1, 2) emended the description to include the thick- ened apices and reported the genus from Lower Cretaceous deposits of southeastern Australia. She noted (p. 59) possible affinities with Dicksonia (D. D. antarctica Labill.), but our modern reference material of these species, and others in the genus, is different (see illustrations in Tryon & Tryon, 1982: 148). Recently Jameos- sanaie (1987, fig. 10, #5) described similar spores (Matonisporites) from the Upper Cretaceous of New Mexico. The Culebra specimens are the first of the type illustrated in Figures 5, 6, 11, 12, and 17 for the Gulf/Caribbean region. sellowiana Hook., VITTARIACEAE Cf. Antrophyum (Figs. 18, 20). Amb trian- ular, apices rounded; trilete, laesurae relatively small in relation to spore diameter, straight, nar- row, 10-14 um long, extending ca. % distance to spore margin, inner margin entire; laevigate; wall ca. 1.5 um thick; 52-65 u Approximately ten of the ca. 40 species of An- trophyum occur in the Neotropics from Hidalgo, Mexico, through Central America and the Antilles, — Ficures 1-10. slide 3a, ESF L-3 470.6, slide 3a, ESF K-28, 1.—5, a Login Pan core 488, slide 1, ESF Q- = 4. 7-9. Cyathea. . Pan core 456, slide 3, ESF B-13, 304.— 10. Pteris. Pan core 488, ESF M-16, 1-2.— D-18, 3 a dm the Culebra Formation, idu 1-3. Lycopodium.— . Pan ESF W-31, 2.— 4. Selaginella. [ipe core 470.6, slide 3a, 2. Pan core 470.6, — 7. Pan core 456, slide 3, slide 1, ESF 1446 Annals of the Missouri Botanical Garden , 12. uiro Pan core 488, slide FIGURES A 16. Fossil spores dien the Culebra eripi P ,ESF L — 13. Trilete ki spore type 1. Pan core 490.6, slide 1, ESF H. 36, 1.— 14. Trilete fern spore ‘pe E Pan core y 56, slide 4, ESF F- 22, 3.— 15, 16. Trilete fern spore type 3. Pan core ye o 6, slide 2a, ESF E-3 1-2 Volume 75, Number 4 1988 1447 Culebra Formation—Paleotropical Communities to northern Argentina and southeastern Brazil. The genus grows in rain forests and cloud forests, usu- ally at elevations of 100-1,500 m. The specimens are laevigate, while modern spores often have a delicate sculpture (*echinate rods and surface strands," Tryon & Tryon, 1982: 360-361). Tryon & Tryon believe, however, that a perispore is pres- ent, and in fossil specimens this ornamental cov- ering would not be preserved. Similar large, trilete, laevigate spores are known from the Eocene of Panama (Graham, 1985, figs. 13, 16) and the Miocene of Veracruz, Mexico (Graham, 1976, fig. 227). OTHER TRILETE FERN SPORES Several trilete fern spores were recovered for which biological affinities could not be determined. Four of the more distinctive ones are described below. Type 1 (Fig. 13). laesurae straight, narrow, ca. 20 um long, extend- ing to spore margin, inner margin + entire; finely Amb oval-triangular; trilete, reticulate to micropunctate (width of muri equals diameter of lumen, ca. 1 um), sculpture less dense near laesurae; wall ca. 1.5 um thick; 42 um. Type 2 (Fig. 14). Amb oval-triangular to + circular; trilete, laesurae straight, narrow, 20-22 um long, extending ca. % distance to spore margin, inner margin entire; finely reticulate; wall 1.5 um thick; 45-55 um. Type 3 (Figs. 15, 16). Amb oval-triangular to wedge-shaped, apices rounded; trilete, laesurae straight, narrow, 22-26 um long, extending to spore margin, inner margin entire; punctate, punc- tae circular (ca. 1 um) to elongated (2-4 um) and slitlike and sinuous; wall 2 um thick; 42 x 36 um (slightly compressed). Type 4 (Fig. 19). angular, apices rounded; trilete, laesurae straight, narrow, 34-36 um long, extending to spore mar- gin, inner margin entire, conspicuous lip 12 um wide (maximum, conspicuousness augmented by folds?); low, moundlike verrucae ca. 2-3 um diam.; wall ca. 3 um thick; 72 x 64 um (compressed). Amb irregular, + oval-tri- MARATTIACEAE Danaea (Fig. 21). Amb circular to + reni- form; monolete, laesurae straight, narrow, 18-20 um long, extending nearly to spore margin, inner margin entire; echinate, echinae ca. 2 um long, relatively dense; wall 1 um thick; 28-32 um (ex- cluding spines). Danaea is a neotropical genus of about 20 species, which presently grows from southern Mex- ico, through Central America and the Antilles, into Venezuela, the Guianas, Bolivia, and the Amazon Basin (Tryon & Tryon, 1982: 47). It grows in moist habitats, wet forests, and rain forests from sea level to about 2,300 m. Fossil spores have not been reported previously from Tertiary deposits in the Gulf/Caribbean region. OTHER MONOLETE FERN SPORES Type 1 (Fig. 22). straight, narrow, 22-26 um long, extending ca. ?4 Reniform; monolete, laesura spore length, inner margin entire; laevigate; 41 x 22 um. These spores likely represent the Blechnaceae (e.g., Blechnum) and Polypodiaceae (e.g., Aspleni- um, Thelypteris) without the delicate ornamented perispore. They are common in fossil deposits, rec- ognized under the artificial generic name Laevi- gatisporites, and range from Paleozoic to Recent. Type 2 (Fig. 23). straight, narrow, 22-26 um long, extending ?4 Reniform; monolete, laesura spore length, inner margin entire; verrucate, ver- rucae moderately low and conspicuous, shape ir- regular, ca. 2 X 4 um; wall 2-3 um thick; 56- 58 x 39-41 um. These spores are produced by species of several genera of the Polypodiaceae and Blechnaceae (e.g., Blechnum, Microgramma), bear the artificial ge- neric name Verrucatisporites, and range from the Paleozoic to Recent. Type 3 (Figs. 24-26). Reniform; monolete, laesura straight to curved, narrow, 32-36 um long, extending ca. 144-% spore length, inner margin entire; verrucate, verrucae distinct peglike struc- tures, apices often flat, moderately to sparsely dense, 5 ca. 2-3 um diam.; 52-65 x 27-51 um GRAMINEAE (Fig. 27). ate, pore circular, 3 um diam., inner margin entire, Spherical, amb circular; monopor- annulus ca. 2 um wide, outer margin entire; tectate, wall ca. 2 um thick; scabrate; 36-40 um. j i t l and the spec- e Gramineae Lu Z imens cannot be referred to any single genus. Grass pollen has been reported from the Miocene of Ve- racruz (Graham, 1976) but not from the Eocene Gatuncillo Formation of Panama (Graham, 1985) or the lower Miocene Uscari sequence of Costa Rica (Graham, 1987a). Germeraad et al. (1968) listed its range in the Gulf/Caribbean region as base of the middle Eocene to Recent, but Muller 1448 Annals of the Missouri Botanical Garden 22 FicurEs 17-23. Fossil spores from the Culebra Formation, Panama.— 17. Lyg Pus Pan core 456, slide 4, ESF K-11, 2. 18, 20. Cf. Antrophyum.— 18. Pan core 456, slide 1, ESF U. 14, 3—4.— 20. Pan core 456, slide Volume 75, Number 4 1988 Graham Culebra Formation—Paleotropical Communities 1449 (1981) cited reports extending it back into the Paleocene of Brazi PALMAE Palm pollen is one of the most common plant microfossils in the Culebra Formation. Four types are described, and it is likely that more than one genus may be represented by each. Cryosophila-type (Figs. 29-32). Prolate, amb oval; monocolpate, colpus straight, 24-26 um long, extending entire length of grain, margin entire; tectate-perforate, wall 2 um thick; reticulate, muri relatively broad (ca. 1-1.5 um), flat, lumen ca. 2 um diam. on distal side, smaller approaching col- pus; 34-38 x 24-32 um. This is one of the most common types of palm pollen in the Culebra Formation (Table 1). It is very similar to that of Cryosophila warscewiczii (H. Wendl.) Bartl., a tree up to 10 m tall distributed from Belize to Panama. In Panama it is known from the “‘tropical moist forest on both slopes of the Canal Zone and in Panamá and Darien; known also from premontane wet forest in Coclé (El Valle) and Panama (Cerro Campana) and from tropical wet forest in Darien" (Croat, 1978: 171). Palm pollen is usually abundant in Gulf/Caribbean Ter- tiary deposits, but this is the first report of the Cryosophila type. Bartlett & Barghoorn (1973: 230, figs. 5, 6) reported a more finely reticulate form from the Quaternary of Panama. Desmoncus-type (Figs. 33, 34). Prolate; monocolpate, colpus straight, broad, boatlike, api- ces rounded, 26-28 um long, extending nearly entire length of grain, inner margin entire; tectate- perforate, wall ca. 1.5 um thick; finely reticulate (diameter of lumen ca. 1 um or slightly less); 30 X 18 um. This kind of pollen is produced by several palms, and Desmoncus is used primarily as an example of a morphologically similar type. A climbing palm known from Mexico to South America, it grows in Panama in tropical moist forests in the Canal re- gion, in the provinces of Panamá and Darién, and in the tropical wet forest in Colon (Croat, 1978: 171). Manicaria-type (Figs. 37, 38). Prolate; monocolpate, colpus straight to occasionally bent, 36-40 um long, extending nearly entire length of grain, inner margin entire; finely reticulate (di- ameter of lumen ca. 1 um or slightly less); tectate- perforate, wall ca. 1 um thick; 40-45 x 2 um, widest part just off equator of grain. Manicaria is used as an example of a palm producing pollen morphologically similar to the specimens. It is a tree up to 10 m tall with three species in the Antilles, Central America, and South America (fide Bailey, 1943) and often occurs in dense groves in wet places. Similar pollen of this general palm type has been reported from the Eocene of Panama (Graham, 1985, fig. 26). Synechanthus-type (Figs. 35, 36). Prolate; monocolpate, colpus straight, 28-30 um long, ex- tending entire length of grain, inner margin entire; scabrate (to possibly microreticulate); tectate, wall 1.5-2 um thick; 28-34 x 22-26 um. Synechanthus is representative of modern palms producing morphologically similar pollen (cf. also Aiphanes, the West Indian Thrinax). It is a tree to 6 m tall distributed from Costa Rica to Colombia and Ecuador, usually at low elevations. In Panama it is known from “tropical moist forest on both slopes in the Canal Zone and in San Blas and Darién, from premontane wet forest in Veraguas (Cerro Tute) and Panamá (Lago Cerro Azul), and from premontane rain forest in Panama (Cerro Jefe) and Darien (Cerro Pirre)" (Croat, 1978: 178). AQUIFOLIACEAE Ilex (Fig. 28). Oblate-spheroidal, amb circu- lar; tricolporoidate, colpi straight, 9-11 um long (pole to equator), tapering to acute apex, equato- rially arranged, meridionally elongated, equidis- tant, extending within 7 um of pole (P.I. 0.33), inner margin diffuse, pores obscure, diam. ca. 2- 3 um, circular, situated at midpoint of colpus; intectate, clavate, wall 3 um thick; 25-30 um Ilex is one of the most common microfossils in the Gulf/Caribbean Tertiary, and its stratigraphic range and ecology have been discussed previously (e.g., Graham, 1985: 514). Briefly, its geologic range is from Late Cretaceous to Recent (Muller, 1981), and in northern Latin America pollen has been recovered from all formations studied in the ESF J.27.— 19. Trilete fern spore type 4. Pan core 470.6, slide 2a, ESF K-3 . Monolete fern spore type a a core 470.6, slide 2 ESF T-31, 2.— 23. Monolete E slide 3, ESF K-7, 2.—22 fern spore type 2. Pan core 470.6, slide 2a, ESF S-28, 1.—21. Danaea. Pan core 1450 Annals of the Missouri Botanical Garden LL Lg P. E, [A1 T ^. , `" Ve Tu e Fada "v FIGURES 24-38. Fossil spores and pollen from the Culebra Formation, Panama.—24-26. Monolete fern spore type 3. Pan core 456, slide 1, ESF R-25, 1-3; Pan core 470.6, slide 2a, ESF L-18; Pan core 470.6, slide 2a, Volume 75, Number 4 1988 1451 Culebra Formation—Paleotropical Communities present project (Eocene to Recent). In the modern flora it grows at mid altitudes in moist to slightly drier habitats. CHENOPODIACEAE/ AMARANTHACEAE (Fig. 39). Spherical, amb circular; periporate, pores circular, evenly distributed, distance between pores ca. 3 um, inner margin entire; tectate, wall relatively thick (3 um); psilate; 21-23 Pollen of genera in these two families cannot be distinguished consistently, and the specimens serve only to record the family (or families) in the lower Miocene vegetation of Panama. —8 um, estimated number ca. 40, diam. COMBRETACEAE Combretum / Terminalia (Figs. 40, 41). to prolate-spheroidal; tricolporate with 3 pseudo- 15 um long, apices blunt, equatorially arranged, meridionally elongated, equidistant, inner margin entire, pore frequently obscure, circular, diam. 2-3 um, situated at mid- point of colpus; tectate, wall 1.5-2 um thick; psi- late to faintly scabrate; 14-16 x 18-20 um The occurrence of Combretum/Terminalia in Gulf/Caribbean deposits has been discussed re- cently by Graham (1985). Pollen of the two genera cannot be distinguished consistently (Graham, Prolate colpi, colpi straight, found in moist to wet habitats, although individual species may range into drier habitats. COMPOSITAE (Figs. 42, 43). amb circular; tricolporate, colpi straight, short (ca. 4-6 um pole to equator), equatorially arranged, meridionally elongated, equidistant, pore obscure, ca. 2 um diam., situated at midpoint of colpus; vesiculate (air cavities evident in equatorial me- socolpial region); tectate, outer wall 3—4 um thick; echinate, spines men ice 3 um), base broad, mod- er ately d a. 3-4 um); ` r Spherical to oblate-spheroidal, These Jaala Compositae pollen grains are typical of later Cenozoic deposits and have not been reported previously in our studies, although other Compositae pollen types are known from the upper Miocene Paraje Solo Formation of Veracruz, Mex- ico (Graham, 1976). The Culebra specimens are of the Heliantheae type (e.g., Ambrosia, Franse- ria, loa) but cannot be referred to any one genus. CONNARACEAE Cf. Rourea (Figs. 44, 45). Prolate; tricolpate, colpi straight, 16-18 um long, equatorially ar- ranged, meridionally elongated, equidistant, inner margin entire to appearing minutely dentate due to overlying sculpture elements; tectate, wall ca. 1 um thick; finely reticulate; 20-24 x 14-18 um. These prolate, tricolpate reticulate grains are a generalized type but compare most closely to mem- bers of the Connaraceae, especially Rourea. The genus is represented in Panama by three species (fide Woodson & Schery, 1950a), the most com- mon being R. glabra Aubl., distributed from south- ern Mexico, through Central America and the An- tilles, to the Guianas and Brazil. In Panama it is ‘known principally from tropical moist forest in the Canal Zone, San Blas, Veraguas, Los Santos, Panamá, and Darién; known also from tropical dry forest in Coclé and Panamá (Taboga Island), from premontane moist forest in the Canal Zone and Panamá, and from premontane wet forest in Colón and Panamá" (Croat, 1978: 423). DILLENIACEAE Cf. Doliocarpus (Figs. 46-48). Prolate to prolate-spheroidal; tricolporate, colpi straight, 23- 25 um long, equatorially arranged, meridionally elongated, equidistant, inner margin entire, costae colpi ca. 3 um wide, pore obscure, circular, ca. 3 um diam., situated at midpoint of colpus; tectate perforate, wall 2.5-3 um thick; reticulate, muri smooth, about as wide as lumen; 30-34 x 20- 24 um. Four species of Doliocarpus presently occur in the vicinity of the collecting site: D. dentatus (Aubl.) Standley, D. major J. F. Gmel., D. multiflorus Standley, and D. olivaceus Sprague & L. O. Wil- liams ex Standley. The specimens are most similar to D. dentatus, but there is some variation in the — ESF R-22, 1.—27. Gramineae. Pan core 456, slide 4, ESF N-9, 3-4.—28. Ilex. Pan core 488, slide 1, ESF S-9, 1-2.—29-32. Cryosophila-type. Pan core 488, slide 1, ESF S-10, 2-4; Pan core 456, slide 3, ESF V-10, 1.—33, 34. 1 Ly, slide 1, ESF U-35, 2—4; Pun core 469.8, slide 2, ESF ESF S-12, 1; ia core 470.6, slide 3a, ESF N-27, 3-4. pe. Pan core 470.6, slide 3a, ESF J-24.—35, 36. Synechanthus-type. Pan core 456, 1.16, 3.—37, 38. Manicaria-type, Pan core 456, slide 3, Annals of the 1452 Missouri Botanical Garden A tunt SED B . ? "dir. W^ - Ee HII p? DA * t Volume 75, Number 4 1988 1453 Culebra Formation—Paleotropical Communities modern pollen among individual herbarium collec- tions. In the collection Lewis et al. 722A (Chiriqui, Panama, MO) the reticulum is slightly coarser (muri broader) like the specimens, while in the collection Proctor et al. 27037 (Nicaragua, CR) the retic- ulum is finer. Consequently the fossil specimens through Central America to Paraguay and is scat- tered in the Antilles. In Panama it grows in the tropical moist forest, tropical dry forest, premon- tane moist forest, and premontane wet forest (Croat, 1978: 599). DIOSCOREACEAE Dioscorea/ Rajania-type (Fig. 51). Prolate; 1. colpi frequently sinuous, irregular, 21— m long, equatorially arranged, meridionally elon be ates equidistant, inner margin entire; tectate- perforate, wall Speed thin (ca. 1.5 um); finely reticulate; leven native species of Dioscorea are listed for Panama by Morton (1945), and Croat (1978) listed five native Central American species for Barro Colorado Island. Dioscorea is found primarily in the tropical moist, premontane moist, premontane wet, and tropical wet forests. Rajania is a West Indian segregate of Dioscorea. Similar pollen has been reported from the Miocene of Veracruz, Mex- ico (Graham, 1976) EUPHORBIACEAE Alchornea (Fig. 52). colpate, colpi straight, 6-8 um long (pole to equa- tor), equatorially arranged, meridionally elongated, equidistant, extending within 5-6 um of pole, inner margin entire, distinct operculum; tectate, wall 1.5 Oblate, amb circular; tri- N 7i um thick; psilate to faintly scabrate; Pollen of Alchornea is frequent in Gulf/Carib- bean Tertiary deposits and ranges from the lower and middle Eocene (Colombia; González Guzmán, 1967) to Recent (Muller, 1981). Its distribution and ecology for the region have been summarize by Graham (1987a), based on Croat (1978) and Webster & Burch (1967). Briefly, it grows in the tropical moist, premontane wet, and premontane rain forest, with a wide altitudinal range of 300 to 2,000 m. Pollen of the family has been described by Punt (1962). Sapium (Figs. 49, 50). Prolate; tricolporate, colpi straight, 62-66 um long, equatorially ar- ranged, meridionally elongated, equidistant, inner margin entire, costae colpi 4-6 um wide, margo present (formed by reticulum becoming finer bor- dering the colpi), pore large (12-14 um diam.), oval, situated at midpoint of colpus; tectate-per- forate, wall 3 um thick, individual columellae dis- tinct; finely reticulate; 76-80 x Sapium is a genus of about 120 species dido Willis, 1966) of trees and shrubs widely distributed in tropical and subtropical regions of the New World Webster & Burch (1967) listed four species for Panama, and Croat (1978) recorded two for Barro Colorado Island. It grows in the tropical moist, premontane wet, and lower montane wet forests. Hartshorn (1983: 143-144) described its occur- rence within the Monteverde Reserve in Costa Rica (cove, leeward cloud, windward cloud, and swamp forests). There is considerable range in size of Sapium pollen, and the smaller forms (48 um) have been described from the Quaternary of Panama (Bartlett & Barghoorn, 1973). Slightly larger specimens (58 um) occur in the upper Miocene Paraje Solo For- mation of Veracruz, Mexico (as cf. Sapium; Gra- ham, 1976). The Culebra specimens are excep- tionally large (80 um) and are matched by a collection labeled 5. haematospermum Muell.-Arg. (Rocha 3666, Argentina, TEX). It is not possible to refer the specimen to any one modern species, but pollen identical to the specimens in size and fineness of the reticulum apparently does occur in the genus. — FicunEs 39-58. T core 470.6, slide Ps a Ahad core 456, sli ESF W-15. pii Fossil m the b iw Formation, Panama.— 39. Chenopodiaceae/ Amaranthaceae. , 41. Pap allie eia ena Pan core 470.6, slide 2a, ESF O- "5 3. Compositae. Pan core 488, cnni 1, ESF J-14, 4.— 44, G vu ie bie core 456, oo 4, reed LIZ. 2- ha Pan core 456, slide 3, ESF W —46-48. Cf. meia Pan core slide 2, ESF K-23, 1—2; Pan core 456, slide 1, ESF L-26, 2-4. m^ 50. Sapium. Pan core 456, e ~ E F ; Pan core yc slide 4, ESF G-12, 3.—51. Dioscorea/Rajania type. Pan core 470.6, slide 2a, ESF F-38, 3. E NER nea. Pan core 469.8, slide 2, ESF T- 15, 2-4. 3. Casearia. Pan core 470.6, Pg 2a, ESF L-21.— cia. Pan core 4868, slide 1, Em 34.—55-58. Tétrorchillum. Pan core 488, slide T-33, 2-4; Pan c core 470.6, slide 2a, ESF Q-35, 1454 Annals of the Missouri Botanical Garden "67 '€«*68. FIGURES 59-70. gas pollen from the Culebra Formation, Panama.—59, 60. Hampea/ Hibiscus. Pan co 488, slide 1, ESF C —61, 62. Malpighiaceae. Pan core 488, slide 1, ESF B-26, 3.—63, 64. Sabicea. Pon Volume 75, Number 4 1988 Graham Culebra Formation—Paleotropical Communities 1455 Tetrorchidium (Figs. 55-58). Oblate-sphe- roidal, amb circular; tricolpate, colpi straight, 12- 14 um long (pole to equator), equatorially ar- ranged, meridionally elongated, equidistant, inner margin finely dentate; intectate, finely baculate to nearly echinate, wall ca. 1.5 um thick; 26-30 um. According to Webster & Burch (1967), there are about ten species of Tetrorchidium in the New World (five in Africa), and two are listed for Pan- ama (T. euryphyllum Standley, rainforests of Costa Rica and Panama; 7. gorgonae Croizat). Hartshorn (1983) mentioned Tetrorchidium sp. in the swamp forests of the Monteverde Reserve in Costa Rica (associated with Alchornea and Sapium). Pollen wall thickness varies among the species of Tetror- chidium, and the specimens are similar to the thinner-walled forms in our collection (e.g., 7. ro- tundatum Standley, Nicaragua). Similar pollen has been reported from the Oligocene of Puerto Rico (Graham & Jarzen, 1969) and the Miocene of Veracruz, Mexico (Graham, 1976). FLACOURTIACEAE Casearia (Fig. 53). Prolate; tricolporate, colpi straight, 18-20 um long, apices acute, equatorially arranged, meridionally elongated, equidistant, in- ner margin entire, narrow costae colpi, pore equa- torially elongated (colpi transversalis), 1.5-2 x 4- 5 um, situated at midpoint of colpus; tectate, wall 1.5 um thick; sculpture subdued, appearing psilate; 23-25 x 16-18 um. Casearia pollen records in the Gulf/Caribbean Tertiary have recently been reviewed by Graham (1985). Briefly, it is known from the middle(?) to late Eocene (Gatuncillo Formation, Panama) to Re- cent. The modern plants are trees and shrubs widely distributed in tropical and subtropical regions, with eight species listed for Panama (Robyns, 1968). They are most typical of moist forest types but can range into somewhat drier habitats (Croat, 1978). Pollen of the family has been studied by Keating (1973). LEGUMINOSAE-MIMOSOIDEAE Acacia (Fig. 54). | Solid sphere, 16-celled poly- ad, amb circular; nonaperturate; individual cells + 1.5 um thick; cubical, ca. um; tectate, wall faintly scabrate; 52-54 Woodson & Schery (1950b) listed 12 species of Acacia for Panama, and Croat (1978) recorded five for Barro Colorado Island. They grow mainly in the tropical moist forest (in contrast to the fa- miliar savannah habitats of African and Australian species) but can range into drier vegetation types. Acacia pollen has been reported from the Oligo- cene San Sebastian Formation of Puerto Rico (Gra- ham & Jarzen, 1969), and a similar grain was recovered from the Quaternary of Panama (as Mimosoideae; Bartlett & Barghoorn, 1973). MALPIGHIACEAE (Figs. 61, 62). Spherical, amb circular; peri- porate, pores circular, 4-5 um diam., inner margin entire; tectate, wall thick (3-4 um); scabrate; 38- 2 um. Only two poorly preserved specimens of the Malpighiaceae were recovered from the Culebra Formation and serve only to record the family in the assemblage. MALVACEAE Hampea/ Hibiscus (Figs. 59, 60). Spherical, amb circular; periporate, pores circular, 3-4 um diam., inner margin entire, narrow annulus; tec- tate, wall 2 um thick; echinate, spines 4-5 um long, broadened at base, moderately dense (dis- tance between spines ca. 6-8 um); 48-70 um. Robyns (1964a) and Croat (1978) listed one species of Hampea, H. appendiculata (Donn. Sm.) Standley, for Panama and Barro Colorado Island. It is a shrub to mid-size tree growing in Honduras, Costa Rica, and remontane moist, tropical wet, and premontane rain forests. Hibiscus is represented by 14 species in Panama (fide Robyns, 1965), and on Barro Colorado Island by three species (Croat, 1978). Hibiscus sororius L.f. is a herb to small suffrutex that is common in marshy areas as a component Panama in the tropical moist, of floating masses of marshy and swamp vegetation (Croat, 1978: 583). Hartshorn (1983) listed it as a member of the coastal strand vegetation in Costa Rica. Although insect pollinated, with large (to 70 um) echinate pollen, the plant grows in habitats that would be immediately marginal to the Culebra depositional basin, and fossil specimens are fre- quent in the deposits. T 0.6, slide 3a, ESF D-26, 2-4.—67- , ESF Q-11, 3—4; Pan core 488, slide 1, ESF V-1 — core 470.6, slide 3a, ESF U-31, 3.—65, 66. Eugenia / Myrcia. Pan core 470.6, slide ? ESF Q-22, 4; Pan core 70. ate Pan core 456, slide 4, ESF J.1 , 1; Pan core 456, slide 1456 Annals of the Missouri Botanical Garden The Hampea / Hibiscus pollen type is recorded under the artificial generic name Echiperiporites and ranges globally from the upper Eocene to Re- cent. It is known from the upper Eocene of Ven- ezuela (Muller, 1981), upper Miocene of Veracruz, Mexico (Graham, 1976), and the Quaternary of Panama (Bartlett & Barghoorn, 1973). MYRTACEAE Eugenia / Myrcia (Figs. 65, 66). Oblate; amb triangular; tricol(poroid)ate; colpi straight, 6-8 um long (pole to equator), equatorially arranged, me- ridionally elongated, equidistant, inner margin en- tire, syncolpate, pores vaguely defined, located at apices of grain at midpoint of colpus; tectate, wall ca. 1.5 um thick; scabrate; 12-16 um. Occurrences of Eugenia / Myrcia in the Gulf/ Caribbean Tertiary have been discussed recently by Graham (1985, 19872). It is frequent in the sediments, although not in high percentages, and ranges from the middle Eocene to Recent. Study of the modern pollen (Graham, 1980) indicates that isolated microfossils cannot be referred con- sistently to any one genus of the family, and the specimens serve only to document the Myrtaceae as a prominent component of the lower Miocene vegetation of Panama. RHIZOPHORACEAE Rhizophora (Figs. 67-70). late-spheroidal; tricolporate, colpi straight, 14-1 um, equatorially arranged, meridionally elongated, Prolate to pro- equidistant, inner margin entire, costae colpi ca. 2-3 um, pores elongated equatorially (colpi trans- versalis), 1 x 4 um, constricted at midpoint of colpus; tectate-perforate, wall 2—3 um thick; finely reticulate; 16-20 x 14-19 um. Records of Rhizophora pollen in the Gulf/Ca- ribbean Tertiary have been summarized by Graham (1985: 519, 1987a). The Culebra specimens fur- ther document its widespread occurrence and often dominant percentages in sediments such as lignites deposited under warm-temperate to subtropical, coastal, brackish-water conditions. It ranges from late Eocene to Recent in tropical American sedi- ments (often reported under the artificial generic name Zonocostites), and in older deposits is re- placed by its presumed ecological counterpart, Brevitricolpites of unknown biological affinity. Other aspects of modern and fossil mangrove pollen have been discussed by Langenheim et al. (1967), Leopold (1969), and Muller & Caratini (1977). RUBIACEAE Sabicea (Figs. 63, 64). Oblate, amb oval-tri- angular to nearly circular; tricolpate/porate (ap- ertures short, slitlike, ca. 2:1 length/width ratio), 4-6 x 2-3 um, equatorially arranged, meridio- nally elongated, equidistant, inner margin faintly dentate (due to overlying sculpture elements), faint costae colpi; tectate-perforate, wall 2-3 um thick; finely reticulate, muri smooth, width about same as diameter of lumen (ca. 0.5-1 um); 32-36 um. The presence of Sabicea pollen in the Culebra assemblage has been reported recently by Graham (19882). Briefly, the genus presently grows from Mexico to northern South America and is repre- sented in Panama by three species and two varieties common in tropical moist and premontane wet forests. Sabicea has not been reported previously in the fossil record. SAPINDACEAE Allophylus (Figs. 71-75). Oblate to per- oblate, amb distinctly triangular; triporate, pore slightly elongated meridionally, ca. X 3 um, equatorially arranged, equidistant, inner margin entire, faint costae colpi; tectate, wall 2 um thick; scabrate to microreticulate; 14 x 24 um. Allophylus is a mostly South American genus of about 190 species of shrubs or small trees with six species in Mexico and Central America, and three in Panama (fide Croat, 1976). It typically grows at low to moderate elevations (ca. 1, and ranges through several forest types— tropical dry, tropical moist, tropical wet, premontane moist, and premontane wet forests. It has been reported previously from the upper Miocene Paraje Solo Formation of Veracruz, Mexico (Graham, 1976). Cupania (Fig. 76). Oblate to peroblate, amb triangular; tricolpor(oid)ate, colpi straight, 7-8 um long, equatorially arranged, meridionally elongat- ed, equidistant, inner margin entire, syncolpate; tectate, wall 2 um thick; psilate to faintly scabrate; Cupania is a genus of about 45 tropical Amer- ican species of trees and shrubs. The nine species listed for Panama (Croat, 1976) grow primarily in the tropical moist, tropical wet, and premontane wet forests. The genus has been reported from the upper Miocene Paraje Solo Formation of Mexico (Graham, 1976) and from the Quaternary of Pan- ama (Bartlett & Barghoorn, 1973). Matayba (Fig. 71). Oblate, amb triangular; Volume 75, Number 4 1988 Graham 1457 Culebra Formation—Paleotropical ti Communities tricolpor(oid)ate; colpi straight, 12-14 um long, apices branched with triangular sexine area in- cluded at poles, equatorially arranged, meridionally elongated, equidistant, inner margin entire; tectate, wall 2 um thick; scabrate; 23-27 um. Croat (1976: 433) noted that Matayba is not always easily separable from Cupania. The pollen is also similar. Matayba pollen, however, com- monly has branched colpi that include a triangular segment of the sexine at the poles, while in Cupania the colpi most frequently are unbranched or, when branched, do not usually include the triangular segment of sexine. Matayba consists of about 45 species of trees and shrubs in tropical America, four of which are recorded for Panama (Croat, 1976), where it is found in the tropical dry, tropical moist, tropical wet, premontane moist, and premontane wet for- ests. Matayba has been reported from the upper Miocene Paraje Solo Formation of Veracruz, Mex- ico (Graham, 1976). SAPOTACEAE Cf. Pouteria (Figs. 78, 79). ate, colpi straight, equatorially arranged, me- ridionally elongated, equidistant, inner margin en- tire, narrow costae colpi, pores elongated equatorially, 2-3 x 4-5 um, situated at midpoint of colpus, narrow annulus; tectate, wall relatively thick (2-3 um); faintly scabrate; 26-28 x 20 2 um. Pouteria is a genus of about 100 species of trees and shrubs found chiefly in tropical America. In- cluded are a number of segregates that “are prob- ably sound and deserve recognition at least at the subgeneric level. Unfortunately many of the seg- regate genera themselves are poorly defined, being Prolate; tricolpor- based on relatively few specimens. As more ma- terial is collected, the generic limits may have to be shifted. This is particularly true among Amer- ican species” (Pilz, 1981: 186) The pollen is also similar to other members of the family, including Sideroxylon, which is a genus of trees and shrubs of the New and Old World pe growing in Mexico, northern Central Amer- and Colombia but is not listed for Panama (Blackwell 1968; Croat, 1978; D'Arcy, 1987). Species previously assigned to Sideroxylon from Panama are now mostly referred to Pouteria (e.g., S. sapota Jacq. = P. sapota (Jacq.) Moore & Stearn; S. uniloculare Donn. Smith = P. uniloc- ularis (Donn. Smith) Baehni; Pilz, 1981). Collec- tions at MO are rare from Mexico (one collection) and South America (two collections) but common from Asia, Africa, and Oceanica (Ricketson, pers. comm., 1987), suggesting a possible Old World origin. It is the only genus in the Culebra assem- blage with this pattern of distribution, hence the provisional reference of the fossil specimens to Pouteria. STERCULIACEAE Cf. Guazuma (Fig. 80). Prolate; tricolporate, colpi straight, 16-18 um long, equatorially ar- ranged, meridionally elongated, equidistant, inner margin entire, narrow costae colpi, pores obscure, slitlike, ca. um, situated at midpoint of colpus; tectate-perforate, wall 2 um thick; finely reticulate; 22-24 x 20-22 um. Guazuma is a genus of trees and shrubs con- sisting of three species with only G. ulmifolia Lam. listed for Panama (Robyns, 1964b). The latter species grows from Mexico to Panama and in the West Indies. It is ecologically variable and in Pan- ama grows in the tropical moist, premontane wet, premontane moist, premontane dry, and tropical dry forests (Croat, 1978: 594) UNKNOWNS A number of specimens were recovered that could not be identified. Some of the more distinctive and/or abundant ones are illustrated and briefly described below Type 1 (Fig. 81). nonaperturate; intectate, wall thin (ca. 1.5 um); echinate, spines short (ca. 2-3 um), dense; 36 um. Type 2 (Figs. 82-86). Prolate; tri(?)colpate, colpi sinuous, 28-46 yum long, equatorially ar- ranged, meridionally elongated, equidistant, inner margin entire; tectate, wall 2-3 um thick, homo- geneous and hyaline; psilate to faintly textured (scabrate?); 32-50 x 19-34 um These specimens are characterized by a glass- like, hyaline wall and possibly represent an un- known fungal spore. Type 3 (Figs. 87-90). colpi straight, 40-46 um long, equatorially ar- ranged, meridionally elongated, equidistant, inner margin entire, costae colpi 5-9 um wide, pore slightly oval, 3 x 5 um, situated at midpoint of colpus; tectate-perforate, wall 2 um thick; finely reticulate; 42-55 x 32-43 um. These specimens are similar to pollen of several genera of the Anacardiaceae but also resemble Spherical, amb circular; Prolate; tricolporate, 1458 Annals fice) a Garden FIGURES 71 = Fossil pollen from the Culebra Formation, Pd ei — 71-75. Allophylus. Pan core 469.8, SF re 1, ESF J-14, 1-3; Pan core 2 slide 2, ESF W-24, 2; Pan core 456, slide 3, ESF G-27, 1—3.— 706. la Pan core n 56, slide 3, ESF F-30. — 77. Matayba. s core 470.6, slide 3a, ESF T-25, 2.—78, 79. Cf. Pot . Pan core 470.6, slide 3a, ESF He Fi 3.—80. Cf. Guazuma. Pan core 456, slide 4, ESF 0-29, 4.— 81. jos n type 1. Pan core 470.6, slide 3a, ESF Q-18, 4.—82-86. Unknown type 2. Pan core 490.6, slide 1, ESF Y-32, A Pan core 456, slide 1, ESF H-12, 4; Pan core 456, slide 3, ESF V-18, 4; Pan core 456, slide 1, ESF R-19, 3.—87, 88. Unknown type 3. Pan core 469.8, slide 1, ESF R-23, 1-2; Pan core 469.8, slide 1 ESF Q-30, 3— 4 Volume 75, Number 4 Graham 1459 1988 Culebra Formation—Paleotropical Communities FIGURES 89-104. re pollen from the Culebra Formation, Panama.—89, 90. Unknown type 3. Pan core 456, slide 1, ESF U-22, 1; Pan core 456, slide 3, ESF L-14, 2.—91. Unknown DL. 9. Pan core 456, slide 3, ESF R-26, 4.— 92. S type 14. Pan core 470.6, slide 3a, ESF F-36, 3-4.— 93. dar type 13. Pan core 456, slide 1, ESF Q-37, 4.—94, 95. Unknown type 5. Pan core 456, slide 4, ESF F-17, 3-4; Pan core ESF P-25, 2-4.—99, 100. Unknown type 7. Pan core 469.8, slide 2, ESF U-25, 3-4.— 101. Unknown type 12. Pan core 456, slide 4, s4 N-10.— 102-104. Unknown type 11. Pan core 456, slide 3, ESF G-33; Pan core 488, slide 1, ESF V-27, 1460 Annals of the Missouri Botanical Garden some Bignoniaceae and Euphorbiaceae. They are common in the Culebra assemblage and in several other Gulf/Caribbean Tertiary formations (e.g., Eocene Gatuncillo Formation, Panama; Graham, 1985, figs. 138-144 Type 4 (Fig. 107). Oblate-spheroidal, amb cir- cular; tetracolporate, colpi straight, equatorially arranged, meridionally elongated, equidistant, in- ner margin entire, pores equatorially elongated, ca. l X 4 um, situated at midpoint of colpus; tectate, wall 2 um thick; psilate to faintly scabrate; 23 um. Type 5 (Figs. 94, 95). Oblate to oblate-sphe- roidal; amb circular; stephanocolpate, colpi 6, straight, 6 um long, equatorially arranged, merid- ionally elongated, equidistant, inner margin entire, pores equatorially elongated, ca. 1 X 4 um, situ- ated at midpoint of colpus; tectate, wall 2 um thick; psilate; 14-16 um. Type 6 (Figs. 105, 106). Oblate-spheroidal, amb circular; tricolpor(oid)ate, colpi straight 15 um long, equatorially arranged, meridionally elongat- ed, equidistant, margin diffuse; tectate, wall 3 um thick; scabrate, some sculpture elements more pointed, approaching echinae; 36 um. These specimens are generally similar to Vale- riana, but no exact match could be found among the species examined. Type 7 (Figs. 99, 100). Prolate-spheroidal; tricolpate, colpi straight, 22-24 um long, equa- torially arranged, meridionally elongated, equidis- tant, inner margin entire, costae colpi ca. 3 um wide, pores circular, 3 um diam., situated at mid- point of colpus; tectate-perforate, wall 3 um thick; reticulate, muri smooth, slightly narrower than the lumina; 36 um These rare specimens may represent obliquely preserved specimens of Coccoloba Type 8 (Figs. 96-98). Oblate, amb spherical; tricolporate, colpi straight, short (4-6 um), equa- torially arranged, meridionally elongated, equidis- tant, inner margin entire, pores elongated equa- torially (colpi transversalis), 1 x 5-6 um, situated at midpoint of colpus; tectate, wall 1.5 um thick; psilate to faintly scabrate; 14-20 um. Type 9 (Fig. 91). Oblate, amb triangular; tri- colpate/porate (length / width ratio ca. 2:1), equa- torially arranged, meridionally elongated, equidis- tant, inner margin entire, situated at apices of triangular grain; tectate, wall 2-3 um thick; sca- brate; ] ` Type 10 (Fig. 108). Oblate to peroblate, amb oval-triangular to circular; tricolpate, colpi short (3-4 um), equatorially arranged, meridionally elon- gated, equidistant, inner margin entire, costae colpi ca. 3 um wide; tectate-perforate, wall 2 um thick; finely reticulate; 24-28 um These specimens are similar to several Bom- bacaceae and related families (Tiliaceae, Sterculi- aceae), but a precise identification could not be made. Type 11 (Figs. 102-104). Oblate, amb cir- cular; triporate, pores circular, 2-3 um diam., equatorially to slightly subequatorially arranged (grains heteropolar), equidistant, inner margin en- tire, annulus 3-4 um wide; tectate-perforate, wall 2 um thick; finely reticulate; 43- These distinctive specimens are common in the Culebra Formation and probably represent a plant that was and likely still is common in the tropical American flora. The specimens are similar to Boc- conia arborea S. Wats., but the modern pollen is periporate with only an occasional triporate grain, while the numerous specimens are all triporate. Type 12 (Fig. 101). Oblate-spheroidal to spherical, amb circular; tri- to periporate, pores circular, 2 um diam., conspicuous annulus 2-3 um wide; tectate-perforate, wall thick (ca. 4 um); finely reticulate; 35 um. Type 13 (Fig. 93). Spherical, amb circular; tricolpate (colpi short, obscure); tectate-perforate, wall thick (4-5 um); reticulate, muri smooth, fine and slightly sinuous; 25-28 um Type 14 (Fig. 92). ilar to Type 13, but the muri and walls are slightly These specimens are sim- thicker. Both are similar to pollen in some Rubi- aceae (Chomelia, Terebraria) but do not match exactly any species in our reference collection. In addition to these unknowns, there is also present the usual assortment of generalized small, prolate, tricolpate/colporate, finely reticulate forms common in Gulf/Caribbean Tertiary deposits. These are illustrated in Graham (1985, figs. 117-128). It is difficult to distinguish among these similar, intergrading types and they cannot presently be identified NUMERICAL REPRESENTATIONS A total of 1,800 palynomorphs were tabulated from the Culebra material (200 each from levels 491.6-456, and 100 each from levels 425-377; Table 1). The ten most abundant types are as follows: Monolete fern spore type 2: 367 specimens (20%) Volume 75, Number 4 1988 Graham 1461 Culebra Formation—Paleotropical Communities FicunEs 105-108. core 456, slide 3, ESF M-1 type 10. Pan Core 456, 5, 3-4.—1 Manicaria-type palm pollen: 176 (10%) Cyathea: 140 (8%) Cryosophila-type palm pollen: 140 (8%) Selaginella: 128 (7% Synechanthus-type palm pollen: 109 (6%) Rhizophora: 100 (6%) Monolete fern spore type 1: 84 (5%) Ham P usya 68 (4%) Lycopodium: 40 (2%) Total ferns: a Total cryptogams (ferns, Lycopodium, Selagi- nella): 40% Total palms: 24% Total lowland vegetation types (ferns, cryptogams, palms, Rhizophora, Hampea/ Hibiscus): 71%. other The composition of the Culebra flora clearly reflects a lowland tropical vegetation. Since the flora is derived from lignites deposited under low- land, coastal, brackish-water, warm-temperate to tropical conditions, there is a tendency to empha- size, possibly to over-emphasize, the potential bias toward this vegetation type. The actual bias is more in quantitative representation, with lowland com- munities over-represented and upland communities usually under-represented. Direct comparison of the fossil assemblage with modern vegetation types can be complicated further by differential preser- vation, insect versus wind pollination, and other factors. Thus attempts to quantify the paleocom- munities in terms of relative abundance or aerial extent would be speculative, particularly in the absence of modern studies on pollen rain. Any impression, however, that major components of the vegetation, such as upland communities or savan- vada pollen from the Culebra Formation, Panama.— 105, 106. Unknown type 6. Pan 07. TER type 4. Pan core 488, slide 1, ESF J-13, 4.— 108. Unknown sfide 3, ESF F-2 nahs, are a priori missing from paleofloras derived from lignites is inconsistent with existing data. The middle to upper Oligocene San Sebastian flora of Puerto Rico (Graham & Jarzen, 1969) and the upper Miocene Paraje Solo of southeastern Mexico (Graham, 1976), both preserved in lignites, contain many representatives of inland and highland com- munities in the form of pollen and spores blown or washed into the lowland basins of deposition. The data show rather convincingly that virtually any tropical vegetation type can be represented in and recognized from palynofloras such as the Culebra. An exception may be paramo, since the pollen of some important components, such as Gramineae, Compositae, and Cyperaceae, cannot be differen- tiated from non-paramo species. In general, how- ever, when the characteristic and defining members of a community are missing from a diverse and well-preserved palynoflora, this is likely due to the absence or poor representation of the community in the region rather than to an absolute bias in the depositional process Features traditionally examined in core material are the first and last appearances of distinct types, changes in percentages of individual types through the section, and similar data on groups or assem- blages of types. The purpose is to assess the pos- sibility of subdividing or zoning the section for more precise comparison with other sections or forma- tions (stratigraphic correlation). Such analysis is difficult for the Culebra section because of the uneven quality of preservation along the core. For example, the lowermost (oldest) samples from 491.6 and 491 feet contain palynomorphs of only fair preservation and of low diversity (14 and 12 types of palynomorphs, respectively; Fig. 109). There 1462 Annals of the Missouri Botanical Garden 28 F 24} NUMBER OF TAXA L L J 1 l 1 1 L l 1 491.6 491 490.6 488 470.6 469.8 456 425 415.5 407 DEPTH IN FEET FIGURE 109. ungal spores. Levels Number of palynomorphs in each of the 11 samples from the Culebra core. Diversity and preservation were pius at levels 491.6, 491, 490.6, 425, 415.5, 407, a 0.6 and especially 456 contained the largest number of well-preserved palynomorphs nd 377, which also contained abundant and are probably most representative of the Culebra vegetation. are abundant thick-walled fern spores, suggesting possible differential preservation. Beginning with level 490.6, both the quality of preservation and diversity increase, which continue through level 456 (40 types), with the exception of a drop in diversity at level 469.8 (19 types). The remaining four samples (425, 415.5, 407, 377) are poor, with large numbers of thick-walled fern spores and fungal spores. It has been my experience in Gulf/ Caribbean Tertiary deposits that abundance of fun- gal spores usually coincides with a decrease in the quality, abundance, and diversity of pollen. In light of the differences in preservation along the core, the facts that Lycopodium first appears at level 490.6, or Pteris and Malpighiaceae at level 488, probably do not carry much stratigraph- ic or paleoecological significance. In instances where a particular microfossil appears confined to a single level or narrow zone (e.g., Desmoncus-type, Che- nopodiaceae/Armaranthaceae, Compositae, cf. Doliocarpus, Alchornea, Casearia, Acacia, Eu- genia / Myrcia, Sabicea, cf. Pouteria, cf. Gua- zuma), the percentages are very low, and the zone is usually between levels 488 and 456, where di- versity and preservation are generally high. One quantitative feature of the section that does appear real is the very high percentage of Synechanthus- type palm pollen at level 377. Not only are these grains numerically abundant, but many occur in dense, anther-size clusters, indicating that this layer likely was deposited under or immediately adjacent to a dense stand of these palms. Similar clusters of Rhizophora, Pelliceria, and monolete, verru- cate fern spores (Polypodiaceae, Blechnaceae) oc- cur in other Gulf/Caribbean Tertiary deposits. Other than the unusual, and probably fortuitous, abundance of Synechanthus-type palm pollen at level 377, the quantitative features of the assem- Volume 75, Number 4 Graham 1463 Culebra Formation—Paleotropical Communities TABLE 2. modern community types most range through more than one community. Distribution of taxa identified from the lower Miocene Culebra Formation among comparable in Panama. Placement is according to principal or most common occurrence (s) , and TROPICAL MOIST FOREST (30 genera) Lycop type, Manicaria-type, Syn Dioscorea, cf. Doliocarpus, Eu, Sapium, Terminalia, “awan dim PREMONTANE WET FOREST (25 g odium, Selaginella, cf. Antrophyum, Cyathea, Danaea, Lygodium, Pteris, pex, o type, Desmoncus- nechanthus- -type, Acacia, Alchornea, Allophylus, Case b up , Combretum, a, Hampea, Matayba, Myrcia, cf. Pouietin. Prata cf. Rourea, Sabicea, ra) Lycopodium, Selaginella, d. pon Cyathea, Danaea, v era de | Cryosophila-type, Manicaria- type, Synechanthus-type, Alchornea, Allophylus, Casearia, Combretum, Cu Eugenia, cf. Guazuma, Matay TROPICAL WET FOREST (22 genera) Lycopodium, Selaginella, cf. Ant Manicaria-type, Allophylus, a Combre Myrcia, cf. Pouteria, Terminalia, da PREMONTANE MOIST FOREST (12 gene cf. Antrophyum cf. Rourea, Terminalia TROPICAL DRY FOREST (7 genera) upan a, cf. Rourea, Sabicea, Sapium, disp e (à reñir a, Dioscorea, cf. Doliocarpus, ophyum, iii Lygodium, Pteris, Cryosophila-type, Desmoncus-type, um, Cupania, Dioscorea, Eugenia, Hampea, Hibiscus, Matayba, ra) , Lygodium, Allophylus, Combretum, Dioscorea, Eugenia, cf. Guazuma, Hampea, Ilex, Matayba, Lygodium, Allophylus, Casearia, Combretum, cf. Guazuma, Matayba, cf. Rourea LOWER MONTANE WET FOREST (6 genera) Lycopodium, Selaginella, Cyathea, Pteris, Manicaria-type, Sapium LOWER MONTANE MOIST FOREST (5 genera) Lycopodium, Selaginella, Cyathea, Pteris, Manicaria-type PREMONTANE RAIN FOREST (5 genera) Synechanthus-type, Alchornea, Casearia, Hampea, cf. Pouteria PREMONTANE DRY FOREST (4 genera) Casearia, Combretum, Eugenia, cf. Guazuma LOWER MONTANE RAIN FOREST (none) MONTANE WET FOREST (none MONTANE RAIN FOREST (none) blage do not suggest any distinct zones or significant change in vegetation other than the expected spa- tial reshuffling of local communities with minor changes in the landscape. The assemblage is best considered as a unit, with levels 488 to 456 most representative of the Culebra vegetation. The tab- ulations in Tables 1 and 2 provide the most com- plete listing presently available of plants and pa- leocommunities occurring on the volcanic islands constituting Central America ca. 20 Ma PALEOCOMMUNITIES Genera identified from the Culebra Formation are arranged according to paleocommunities in Ta- ble 2. These are only approximations because the microfossils are mostly identifiable to genus, and many genera range through more than one com- munity. On the other hand, recognition of the paleocommunities is based on the presence of char- acteristic genera and on the total number of genera indicative of a given assemblage rather than on single “key” members. Thus, the impact of any one assignment is reduced. Also, as data accu- mulate from other Tertiary fossil floras in northern Latin America, there is increasing consistency, both between floras and between the paleobotanical data and evidence derived from other independent lines of inquiry (see discussions in Graham, 1987a, b, 1988c, d) regarding the kinds of vegetation likely present in the Gulf/Caribbean region. As noted earlier, the prominent representation of wet- to moist-lowland communities, compared with higher- altitude and drier savannah-type vegetation, can- not be ascribed solely to bias in the depositional environment, because the latter communities are recorded in Gulf Caribbean Cenozoic palynofloras. 1464 Annals of the Missouri Botanical Garden Four paleocommunities are prominent in the Culebra flora (Table 2). The tropical moist forest is represented by 30 genera that presently occur in this community. Included is Rhizophora, es- tablishing the presence of the mangrove swamp, although pollen was not abundant in this section. he maximum percentage was 10.576 at level 491.6, with an average of 5.876 for the 11 levels. This is in contrast to percentages as high as 96% in one sample from the Paraje Solo Formation of southeastern Mexico (Graham, 1976: 803, table 1). The absence of other mangrove genera such as Avicennia, Conocarpus, Laguncularia, and Pelliceria suggests: 1) the community was present but not dominant in the vicinity of the depositional basin at this time, and 2) that these sediments accumulated in an estuarian fern/palm marsh en- vironment behind the mangrove zone. The record of Rhizophora can be quite variable through a section, however, reflecting changes in land/sea relationships in this tectonically active region. In January 1984, core material was received from another well drilled on the east bank of the Panama Canal just north of Gold Hill. The base of the core penetrated into the uppermost part of the Culebra Formation. The microfossils were similar to those from the core used in this study, but at level 171.65-171.9, near the transition between the Culebra and the overlying Cucaracha formations (slightly higher in the Culebra section than the material reported here), the assemblage was vir- tually 100% Rhizophora. These rapid changes in the abundance of mangrove pollen are useful in defining the position of former shorelines, and were used by Bartlett & Barghoorn (1973) to trace the history of Quaternary fluctuations in sea level in Panama. Mangrove vegetation was likely common throughout the Gulf/Caribbean region from the late Eocene onwards, even though its representa- tion at a given locality varied with physiographic conditions. The premontane wet forest potentially includes 25 genera, and the tropical wet forest 22 genera. Some form of the premontane moist forest (12 genera) was probably present, although the evi- dence is not as strong as for other lowland forests. Representation of vegetation types characteristic of higher altitudes and drier habitats (including savannahs) drops off significantly, and there is no palynological evidence for their presence. This is also true for the middle(?) to late Eocene Gatuncillo flora (Graham, 1985), and the Cucaracha (Gra- ham, 1988b) and La Boca floras of Panama (Gra- ham, in prep.). Thus, the vegetation of the low-lying volcanic islands constituting present-day southern Central America ca. 20 Ma consisted of a fringing zone of mangrove vegetation, with fern and palm swamps occupying the estuaries where fresh water diluted the brackish-water habitat of the mangroves. On adjacent, better-drained slopes were versions of the tropical wet, tropical moist, premontane wet, and premontane moist forests. Very local, edaphically controlled, temporal habitats supporting drier vege- tation may have been present, but there is no paleobotanical evidence for these communities in the Culebra flora. PALEOENVIRONMENTS Of the 41 taxa recognized for the Culebra flora, all grow in southern Central America at present. The vegetation clearly grew under climatic con- ditions comparable to those presently prevailing in the coastal, lowland, and moderate-altitude habi- tats. It is not possible to quantify precisely the lower Miocene climates of the region, but the data from Barro Colorado Island cited by Croat (1978: 3-5) and from Costa Rica cited by Coen (1983) and Fleming (1986) provide approximations of the rainfall and temperature. For Barro Colorado Is- land, **Under the Koppen system of climatic clas- sifications, BCI’s climate is Am, or tropical Mon- soon Climate. Annual rainfall on BCI ranges from 190 to 360 cm (76 to 143 inches); between 1924 and 1962 it averaged 275 cm (107.3 inches). This compares with an average 328 cm (128 inches) at Colón, on the Atlantic coast of the Canal Zone, and 177 cm (68 inches) at Balboa on the Pacific coast, during the same period" (Croat, 1978: 3; data from Rubinoff, 1974). The vegetation of the Culebra Formation is most similar to that of the present Atlantic coast of Panama. It is obvious that the existing climatic differentiation between a moist Atlantic side and a drier Pacific side did not prevail, at least to the same degree, in the lower Miocene when the present-day isthmus consisted of a series of islands. Regarding temperature, “The atmo- spheric temperature may vary from as low as 16.5°C (61.7°F) to as high as 35.5?C (95.9°F), with the lowest temperature being recorded within the forest during the rainy season and the highest at the Laboratory Clearing in the dry season. With rare exceptions the temperature ranges between 21 and 32°C (70? and 90°F) throughout the year, and the average ambient temperature in the Laboratory Clearing is 27°C (77°F). The seasonal variation in monthly averages is just 2.2?C" (Croat, 1978: 3). Fleming (1986) presented similar rainfall data for ten sites in Costa Rica. Three of these form a Volume 75, Number 4 1988 Graham Culebra Formation—Paleotropical Communities 1465 transect from the Atlantic coast (site 10, Limon, elevation 5 m) up the Cordillera Central (site 8, Turrialba, elevation 602 m) to the Central Valley (site 6, San José, elevation 1,172 m). The mean annual rainfall is 353.6, 264.1, and 188.9 cm for the respective sites. These elevations and climatic conditions are considered similar to those prevailing on the islands of southern Central America during the lower Miocene. A generalized illustration of my concept of the spatial distribution of lower Miocene communities in southern Central America is shown in Wake (1987: 255, fig. 14), truncated at about 1,000-1,500 m elevation. Estimates of general climate can be evaluated to some extent by comparison with ocean surface- ature curves derived from '*O analysis of foraminifera and other marine invertebrates (Savin, 1977; Savin & Douglas, 1985; Savin et al., 1975). The curves have been discussed re- cently by Graham (1987b) in relation to emerging paleobotanical data for northern Latin America. The curve shows that during the lower Miocene water temper temperatures were generally high, consistent with the tropical conditions suggested by the Culebra (Panama) and Uscari (Costa Rica) floras. A sharp drop occurred at the end of the Miocene, and this is clearly reflected in the composition and paleo- ecology of the late Miocene ie Solo flora of southeastern Mexico (Graham, 1 The affinities of the Culebra flora are distinctly Central and North American. There are no exclu- sively South American elements present, and those growing in South America are plants widespread throughout the neotropics (e.g., Rhizophora). This is the expected pattern of geographic affinity be- cause the latest connection between the North and South American continents formed only in late Pliocene times, about 3 Ma (Stehli & Webb, 1985). ese data illustrate the consistency beginning to emerge between results from paleobotanical studies in the Neotropics and data derived from other independent sources. The paleoenvironmen- tal reconstructions and biogeographic patterns, presently based on the Uscari and Culebra paly- nofloras, can be further evaluated and refined when studies on two additional lower Miocene floras are completed. These are the Cucaracha and La Boca floras from the Canal region of central Panama. LITERATURE CITED BAILEY, L. H. 1943. Palmaceae. In: R. E. Woodson & R. W. Schery era Flora of Panama. Ann. Missouri Bot. Gard. 30: 327-396. BARTLETT, A. S. & E. s BARGHOORN. geographic history of the Isthmus of Pa the past 12,000 years (a history of vegetation, cli- mate, and sea-level change). Pp. 203-299 in A. Graham (editor), Vegetation and Vegetational ovs of Northern Latin America. Elsevier Publ. Co. sterdam aired W.H. 1968. Sapotaceae. /n: R. E. Wood- n & R. W. Schery rni ee of Panama. . Missouri Bot. Gar -169. 198 EE R. E. ; 2 ds patterns and tec- tonic history of M a and the du rriv wa region. Bull. . Soc. Amer. 54: E Coen, E. 1983. Climate. Pp. 35-46 in D. H. 1” (editor), Costa Rican Natural iei Univ. Chicago Press, Chicago. Coney, P. J. 1982 [1983]. Plate tectonic constraints on the biogeography of Middle America and t P E region. Ann. Missouri Bot. Gard. 69: 43 32- 443. COUPER, R. A. 1958. British Mesozoic microspores and pollen grains. Palaeontographica, Abt. B, Palaophy- tol. 103: 75-179. Croat, T. B. 1976. Sapindaceae. /n: R. E. Woodson & R. W. Schery (editors), Flora of Panama. Ann. Missouri Bot. Gard. 19-540. . 19 Flora of Barro oe Island. Stan- ford Univ. ee Stanford, Califor D'Arcy, W. G. Flora of Bd diia and Index. 2 ae Missouri Botanical Garden, St. Upper Mesozoic microfloras from southoaptuñ Australia. Proc. Roy. Soc. Victoria : 1-148. FLEMING, T. H. Secular changes in Costa Rican jen correlation with elevation. J. Trop. Ecol. 2: 87 ET E) e J. & R. Tryon. 1976. Spore morphology in the Cyatheaceae 2. The genera Lophosoria, Me- taxya, “phar Alsophila, and Nephelea. Amer. J. Bot. 63: 738-758. GERMERAAD, J. H c A. Hoppinc & J. MULLER. 1968. Palynology of Tertiary sediments sr tropical areas. Rev. Palaeobot. i ar 6: 189-348. GONZÁLEZ GUZMÁN, A. E. 1967. A Palynological Study on the Upper Los Cuervos and Mirador Formations (Lower and Middle Eocene; Tibú Area, Colombia). E. J. Brill, Leiden. GosE, W. A. 1985. Caribbean tectonics from a paleo- magnetic perspective. Pp. 285-301 in F. G. Stehli S. D. Webb (editors), The Great American Biotic pcd Plenum, New York. GRAHAM, ALAN. 1973. Lite dg on vegetational his- tory in Latin America. Pp. 237-282 in A. Graham (editor), Vegetation pa Ve getational History of Northern Latin America. Elsevier Publ. Co., Am- sterdam. Ç 2916. Studies i in sav ch ains paleobotany. II. The ^ eracruz, Mexico. Ann. = E Gard. 63: 787-842. Literature on vegetational history in Latin MUN Supplement I. Rev. Palaeobot. Pal- ynol. 27: 29-52. 1980. Morfología del polen de Eugenia / Myr- cia (Myrtaceae) y Combretum/ Terminalia (Com- bretaceae) en relación a su alcance estratigráfico en el Tertiario del Caribe. Biotica 5: 5-14 1982 pen id on vegetational history in t II. Rev. Palaeobot. Pal- Latin America. T d ynol. 37: 185- 1466 Annals of th Missouri imm Garden 1985. Studies in m paleobotany. IV. The Eocene communities of Panama. Ann. Missouri Bot. Eo" 12: 504-534. Literature on vegetational history in Latin America Supplement III. Rev. Palaeobot. Pal- ynol. 48: 19 d ie Miocene communities and n vironments oi en Costa Rica. Amer. J. Bot. 74: 1501-1518. —— — —. 1987b. Tropical American floras and paleoen- vironments: Mexico, Costa Rica and Panama. Amer. J. Bot. 74: 1519-1531. "e a. Fossil pollen of Sabicea (Rubiaceae) from the lower Miocene Culebra Formation of Pan- ama. Ann. Missouri Bot. Gar 8 . 1988b. Studies in neotropi ical paleobotany VI. The lower Miocene communitie s of Pan ama—the Cucaracha Formation. Ann. Missouri Bot. Gard. 75: 1467-1479. o oO + Cc: 1988c. Some aspects of Tertiary vegetational history in the Gulf/Caribbean region. Proc. 11th Caribbean Geol. Conf. (Barbados, 1986) (in press). 1 Lower Miocene floras and biogeog- raphy of Central America. J. Geol. Soc. Jamaica (in dis m M. JARZEN. 1969. Studies in bi a see A I. The Oligocene communities of Puerto Rico. Ann. Missouri Bot. Gard. 56: 308 357. . STEWART & J. L. STEWART. 1985. Stud- ies in n neotropical pn III. The Tertiary com- munities of Panama — geology of the pollen-bearing sediments. Ann. Missouri Bot. Gard. 72: 485-503. Hag, B. U., J. HAnpENBOL & P. R. VAIL. 1987. Chro- nology of fluctuating sea levels since the Triassic. Science 235: 1156-1166. HARTSHORN, G. S. 1983. Plants. Pp. 118- : H. Janzen (editor), Costa Rican Natural History. Univ. Chicago Press, Chicago. "MEE A. 1987. Palynology and age of South ah coal-bearing deposits, McKinley County, New au Mines Mineral Re- Mexico Bull. 112: 1-65. & Jones, D. S. F. Hasson. 1985. History and development of the m marine invertebrate To sep arated by the Central American Isthmus 325- 355 in F. G. Stehli & S. D. Webb (editoro). The Great American Biotic Interchange. Plenum, New ork. Kasma, R. C. Pollen morphology and relation- ad d the Flacourtiaceae. Ann. Missouri Bot. Gard. 60: s Todi J. . L. HACKNER & A. S. BARTLETT. 1 ade pollen at the depositional site of from Chiapas, Mexico. Bot. 4. Mus LEOPOLD, E. B. 19 Miocene pollen and spore flora of Eniwetok Atoll, Marshall Islands. Profess. Pap. eol. Surv. 260-II: 1133-1185 Morton, C. V. 1945. Dioscoreaceae. In: R. E. Wood- R. vod (editors), Flora of Panama. An nn. Missouri Bot. 6-33. MULLER, J. 1981. Fossil peius records s an- giosperms. Bot. Rev pase 47: & C. CARATINI. 1977. en 5 "v zophora "rs rio as a guide e Pollen & Spores 19 389. Pilz, G. E Ts Sapotaceae of Panama. Ann. Missouri Bot. Gard. 68: 172-203. Punt, W. 1962. Pollen morphology of d i ien aceae with special reference to taxo . Wen : l- ROBYNS, A. 1964 a. Bombacaceae. /n: R. E. Woodson & R. W. Schery (editors), Flora of Panama. Ann. Missouri Bot. Gard. 7-68. 196 Sterculiaceae. In: R. E. Woodson & R. W. Schery (editors), Flora of Panama. Ann. Mis- souri Bot. : 69-107. 19 Malvaceae. /n: R. E. Woodson & R W. Schery (editors), Flora of Panama. Ann. Missouri Bot. Gard. 52: 497-578. . Flacourtiaceae. /n: R. E. Woodson & A W. Schery epu Flora of Panama. Ann. Mis- uri Bot. Gar -144. Riwi s R. W A d in NAR Monitoring and Baseline Data. Smithsonian Inst. Press, Washington, SavIN, S. M. 1977. The history of the Earth's surface temperature during the past 100 million years. Ann. Rev. Earth Planet. Sci. 5: 319-355. . G. DoucLas. 1985. Sea level, climate, and the Central American land bridge. Pp. 303-324 in F. G. Stehli & S. D. Webb (editors), The ae American Biotic Interchange. Plenum & EHLI. 1975. Te ety marine paleotemperatures. Bull. Geol. Soc. Am 86: 1499-1510 SMITH, D. E 1985. Caribbean plate relative move- 48 in F. G. Stehli & S. D. Webb coro, Ne Great y odes Biotic Interchange. Plenum, New Yor STEHLI, F. G & S. D. WEBB a American v Interchange. P SwIFT, S. A. 1977 anpra in the Pan 1985. The Great rk. ma Basin, J. Geol p Tryon, R. eed 1982. Ferns and Allied Plants, i Special Reference to Tropical America. Springer-Verlag, New York. VarL, P. R. & J. HARDENBOL. 1979. Sea.level changes during the iba Oceanus 22: 71-79. UM, R. G. Topp, J. M. WipMiER "Seite Pier cian and global ges in sea level. Pp. 49-212 . E. Payton (editor) Stratigraphic Interpretation of fedt Data. . Assoc Mem. 26. WAKE, "D. B. 7. Adaptive radiation of salamanders In E A Es cloud forests. Ann. Missouri Bot Gard. 2 WEBSTER, c. L & D. BURCH. 1967. Euphorbiaceae. In: R. E. Woodson & R. W. Schery (editors), Flora of Panama. Ann. Missouri Bot. Gard. 54: = WiLLis, J. C. 1966. A Dictionary of the Flowering Plants and Ferns, 7th edition, revised by H. K. Airy ses ene Univ. Press, Cambridge, Massa- ET AL. 1977. S NA NE R. E. & R. W. ScHERY. Connara eae. In: R. E. Woodson & R chery (editors, di of Panama. Ann. Missouri La Gard. 37: Tipus —-. Leguminosae. In: R. E. Woodson & R. W. Schery Mp ies of Panama. Ann. Missouri Bot. Gard. 37: STUDIES IN NEOTROPICAL PALEOBOTANY. VI. THE LOWER MIOCENE COMMUNITIES OF PANAMA — THE CUCARACHA FORMATION! Alan Graham? ABSTRACT The Cuc tral America (Uscari—Costa Rica; Cu and North American to those presently prevailing in nea extensive volcanic activity doc mosaic of short-term, more open communities wlan mented jor the region likely po the vege may account for a paleobotanical record reflecting primarily aracha microfossil flora is the third in a series of four lower Miocene assemblages studied from southern Cen (U: lebra, Cucaracha, La Boca—Panama been identified (Selaginella, monolete fern spores types 1-3, Cyathea, Ceratopteris, Pteris, cf. Antro rysophila and Manicaria-type palm pollen, lle ex, Composit ineteen palynomorphs have hyum, trilete ae, Alchornia, Alfaroa / Engel- marsh, with associated tation, iid in a shifting op . This dense forests on the slopes, while the few fossil mammalian faunas contain remains of browsers and grazers, suggesting more open forests and savannahs. The Cucaracha Formation is lower Miocene in age and outcrops along both sides of the Panama Canal between Hodges Hill and the Pedro Miguel Locks (Stewart & Stewart, 1980). It belongs to a complex of three Tertiary formations similar in age and lithology known to contain plant microfossils. e lowermost is the Culebra, and study of these palynomorphs has been completed (Graham, 1988). e Cucaracha Formation lies directly on the Cu- lebra in local areas, and thus stratigraphic rela- tionships are relatively clear, even though the re- gion is considerably faulted. The position of the La Boca Formation is more difficult to determine be- cause nowhere does it lie directly on the Cucaracha. However, the Pedro Miguel does overlie the Cu- caracha, and in other areas it interfingers with the La Boca. Thus the three formations are presently considered sequential in age, with the Cucaracha being intermediate between the slightly older Cu- lebra and slightly younger La Boca formations. For more detailed discussion of the geology of the pol- len-bearing strata see Graham et al. (1985). The Cucaracha Formation consists mainly of bentonitic clay shales, tuffaceous siltstones, and sandstones with lenses of conglomerates, carbona- ceous shales, and lignite. A well was drilled through the formation in 1958 (Hole No. PA-33, latitude 9°01'N, longitude 79°38’W, Cucaracha Reach Widening Studies, Panama Canal Commission), and the log shows the above sequence repeated many times through the 40.8-m section. Near the base is a conglomerate containing abundant oyster shells. The sequence is typical of a tectonically active, coastal, estuarine environment, and the lignites ! The author ined e eei n R. H. Stewart, mission, for many useful PESE and for facilitating fieldwork in Panama in 1963-1964, 1968, pet 1983, and 1986. David Lellinge Vasquez, Panama material, and David Dilcher, unir. University, J. L. S made helpful editorial suggest 8205926, BSR-8500850, dae BSR. 8619203. J. L. Stewart, Pastoria Franceschi S., and Numan r, Smithsonian Institution, provided spore reference o and Ralph Taggart, Michigan State University, alp s. The research was supported by NSF grants GB-11862, DEB-8007312, DEB- ? Department of Biological Sciences, Kent State University, Kent, Ohio 44242, U.S.A. ANN. Missouni Bor. GARD. 75: 1467-1479. 1988. 1468 Annals of the Missouri Botanical Garden TABLE l. Identification and numerical represen- indicate warm-temperate to tropical conditions. Ex- tation of fossil palynomorphs from the lower Miocene Cucaracha Formation, Panama. Figures are percent- ages based on counts of 100. Samples 57—59 are from the lower lignite in the section, and samples 62-66 are from the upper lignite. Samples 57 and 66 contained abundant fungal spores, and preservation of pollen and other spores was fair to poor Sample Number 57 58 59 62 63 65 66 Selaginellaceae Selaginella See i Monolete fern spore Type 1 16 6 62 7 14 1 3 Type 2 3. — se CCU; T7 Type 3 6 65 16 72 77 77 38 Cyatheaceae Cyathea A M a ee CF Pteridaceae pee — — l — — — — Pteris E ev Vittariaceae cf. Antrophyum — 3 3 — — — — Trilete fern spore el — — — — — — 4 Type 2 — — — — — — 8 Palmae rysophila-type — — — — — — 2 œ a [e TN — w Pus aria-type ll 13 Aquifoliaceae Ilex ue E e | Compositae —— s cs Euphorbiaceae Alchornea — — 1 ] — ] — Juglandaceae Alfaroa/ Engel- rardia == ness = Lo == = Leguminosae Caesalpinioideae Crudia == Myrtaceae Eugenia/Myrcia — — — — — — 2 Rhizophoraceae Rhizophora 60 13. 7 9 — ll 17 Unknowns tensive vulcanism is indicated by the tuffs (water- lain volcanic ash) and the basalt that caps the section. THE COLLECTING LOCALITY Samples were obtained from a along road K-2, about 0.8 km ioithwest o ié intersection with K-15 in the Gaillard Cut section of the Canal. At this site there is a small utility building, and about 4 m above and to the left is a conspicuous conglomerate layer about 1 m thick (see figs. 7, 8 in Graham et al., 1985). This layer terminates abruptly and continues on the other side of the building about four meters lower in the section. This is one of the many minor faults in the region and is a convenient marker for the locality. Ten samples were collected (our locality C, samples 57a-66) and seven contained plant microfossils. Three of these (57-59) came from a .6-m layer of lignitic shale about six meters below the conglomerate and to the right (facing the slope) of the utility building. Four (62, 63, 65, 66) came from a similar lignitic shale about four meters above the conglomerate; all samples were spaced hori- malian fauna from the Cucaracha Formation, all with distinct North American affinities. MATERIALS AND METHODS Extraction and processing techniques are de- scribed in Graham (1985). Slides are labeled Pan C, Cucaracha, sample, and slide number (e.g., Pan C, Cucaracha, , 1). Location of specimens on the slides is by England Slide Finder coordinates (e.g., ESF R-34, 3-4). All materials are deposited in the palynology collections at Kent State Uni- versity. SYSTEMATICS Ni hs have been identified from the rra uie: (Table 1), and two oth- ers were recovered whose biological affinities could not be established (unknowns types 1 and 2). lustrations, descriptions, and other data are pro- vided for each pollen/spore type, but since all have been recovered from other Gulf/Caribbean Ter- tiary formations, the information is synoptic, and references are provided to more detailed discus- sions. These : ormations, with references, are as follows: Gatuncillo (middle(?) to late Eocene, Pan- Volume 75, Number 4 1988 Graham 1469 Cucaracha Formation — Lower Miocene Communities ama; Graham, 1985), San Sebastian (middle to late Oligocene, Puerto Rico; Graham & Jarzen, 1969), Uscari (early Miocene, Costa Rica; Graham, 19872), Culebra (early Miocene, Panama; Gra- ham, 1988), and Paraje Solo (late Miocene, south- eastern Veracruz, Mexico; Graham, 1976). A map of the geographic distribution and a chart sum- marizing the age of the formations are given in Graham (1987b; for other aspects of the identifi- cation procedures see Graham 1985: 507-508). Present ranges of the modern analogs within the Neotropics and ecological data are summarized af- ter each description, with more detailed summaries provided in the paleobotanical publications previ- ously cited. These data are based on field obser- vations, personal communication with specialists in the various plant groups, and the literature, es- pecially Croat (1978), D’Arcy (1987), Hartshorn (1983), Tryon & Tryon (1982), and Woodson & Schery (1943-1980). Terminology for vegetation types follows Holdridge (1947; Holdridge et al., 1971), used by Croat (1978) and Hartshorn (1983) for describing the plant communities of Panama and Costa Rica. SELAGINELLACEAE Selaginella (Figs. 1, 2). Spherical, amb cir- cular to oval-triangular; trilete, laesurae frequently obscured by dense sculpture and appearing mono- lete, straight, narrow, ca. 20-24 um long, ex- tending nearly to spore margin; echinate, echinae short (ca. 2-3 um), occasionally curro dense, bases broad; wall ca. 2 um thick (excl chinae); 26-30 um. Other occurrences. Gatuncillo, San Sebastian, Uscari, Culebra, Paraje Solo formations. Distribution. Widespread; usually moist, shaded habitats; typically low to mid altitudes but widespread altitudinally. MONOLETE FERN SPORES These spore types are produced by many mem- bers of the Blechnaceae and Polypodiaceae and cannot be referred to any one modern genus, es- pecially in the absence of the ornamented exospore or perine. They are often assigned to the artificial genera Laevigatosporites (smooth forms) or Ver- rucatosporites (verrucate forms), and these range from Paleozoic to Recent. They occur in all of our Gulf/Caribbean Tertiary deposits studied to date, and multiple biological species are likely repre- sented by each of the spore types described below. Type 1 (Figs. 3, 4). Reniform; monolete, lae- surae straight, narrow, 24-28 um long, extending ca. % spore length, inner margin entire; laevigate; 40-50 x 30-40 um This is one of the most abundant spores in the Cucaracha Formation, although percentages vary widely among the samples (Table 1). Figure 4 is a low-magnification view of part of a single field (neg- ative-size portion) typical of samples where the spore is dominant. Ten specimens are evident in this field, and a thousand or more may occur on a slide Type 2 (Figs. 5, 6). This spore differs from Type 1 in having a slightly thicker wall and, there- fore, is more consistently reniform in shape. The specimens range in size from 35 X 20 um (Fig. 5) to 42 x 32 um (Fig. 6). Type 3 (Figs. 7-11). sura straight, narrow, 30-45 um long, extending 34 spore length, inner margin entire; verrucate, verrucae moderately low, conspicuous and dense, grading into less dense, widely spaced verrucae, X 6 um; wall 2-3 um thick; Reniform; monolete, lae- shape irregular, ca. 35-55 x 25-35 um. This is another spore that is dominant in several samples, almost to the exclusion of other micro- fossils (Table 1). In addition to individual speci- mens, many clusters of 5-20 loosely aggregated spores were evident on the slides (Fig. 7). This indicates that the Cucaracha sediments at this lo- cality were accumulating directly under or im- mediately adjacent to a fern marsh, with little trans- port of the specimens. The spores vary in size and density of the ver- rucae. Figure 8 illustrates a small specimen (ca. 37 um), and Figure 9 a larger one (ca. 54 um), while comparison of Figure 8 with Figure 11 shows the variation in the number of verrucae. Similar spore variation occurs within species as well as among species and genera of modern Blechnaceae and Polypodiaceae. As noted previously, more than one biological species, or genus, is likely repre- sented by this spore type. TRILETE FERN SPORES Cyatheaceae Cyathea. Amb oval-triangular, apices round- ed; trilete, laesurae straight, narrow, 14-16 um long, extending to spore margin, inner margin en- tire, bordered by lip 2-3 um wide with punctae 1 um diam.; distal surface finely punctate, proximal surface more laevigate near laesurae; wall 1. jm thick; 30-35 1470 Annals of the Missouri Botanical Garden Volume 75, Number 4 1988 Graham 1471 Cucaracha Formation— Lower Miocene Communities A few poorly oriented specimens of Cyathea were recovered from sample 66, but the genus is frequent in other Caribbean Tertiary deposits. Ac- cording to Gastony & Tryon (1976) and Tryon & Tryon (1982: 207), the micropunctate forms rep- resent Cyathea, while similar but smooth (laevi- gate) types are referred to the closely related Al- sophila. Other occurrences. Paraje Solo formations. Distribution. forests and cloud forests; in Central America low rain forests usually at 1,500-2,000 m, but as low as San Sebastian, Culebra, Widespread, primarily montane Pteridaceae Ceratopteris (Figs. 14, 15). Amb oval-trian- gular, spore margin undulating due to projecting sculpture elements; trilete, laesurae straight, nar- row, 28-32 u inner margin entire; wall coarsely and conspicu- m long, extending to spore margin, ously striate, striae psilate, 3-4 um wide, promi- nently developed on distal surface, less distinct approaching laesurae; wall 2-3 um thick; 75-90 um Other occurrences. Gatuncillo, Paraje Solo formations. Distribution. Widespread, southeastern U.S. (Texas to Florida), southern Mexico, Central Amer- ica, Antilles, South America; frequently aquatic, ditches, brackish waters; sea level to ca. 300 m lagoons, river/lake margins, Pteris (Fig. Amb triangular, apices rounded; trilete, laesurae straight, narrow, 15-18 um long, extending to spore margin, inner margin entire; distal surface with coarse, irregular verru- cae, proximal surface more laevigate, Lr. ca. 6 um wide, hyaline; wall 2 um thick; 42-4 Other occurrences. Gatuncillo, San s Uscari, Culebra, Paraje Solo formations. Distribution. Mexico, Central America, the Antilles, and South America; openings or along margins of wet or cloud forests; sea level to ca. 2,000 m. Vittariaceae Cf. Antrophyum (Fig. 12). apices rounded; trilete, laesurae relatively small in relation to spore diameter, straight, narrow, 12- 15 um long, extending ca. % distance to spore margin, inner margin entire; laevigate; wall ca. 1.5 um thick; 47-54 um Other occurrences. San Sebastian (not figured in Graham & Jarzen, 1969), Culebra, and Paraje Solo formations. Distribution. Hidalgo, Mexico, Central Amer- ica, and the Antilles, to northern Argentina and southeastern Brazil; rain and cloud forests; eleva- tions usually 100-1,500 m Amb triangular, OTHER TRILETE FERN SPORES Type 1 (Fig. 16). laesurae straight, narrow, 14-16 um tending to or nearly to spore margin, inner margin entire; finely reticulate, reticulum becoming more Amb oval-triangular; trilete, long, ex- irregular near laesurae; wall ca. 1.5 um thick; 27- 30 um. Other occurrences. tion (larger specimen). Type 2 (Figs. 17, 18). trilete, laesurae straight, narrow, 15-17 um long, Possibly Culebra Forma- Amb oval-triangular; extending to or nearly to spore margin, inner mar- gin entire; punctate, punctae circular (ca. 1 um) to elongated (2-3 um), slitlike and sinuous; wall 2 um thick; 33-36 um Other occurrences. Possibly Culebra Forma- tion (larger specimen with more conspicuous slitlike punctae). PALMAE Crysophila-type (Fig. 19). Prolate, amb oval; monocolpate, colpus straight, 22-24 um long, ex- tending entire length of grain, margin entire; tec- tate-perforate, wall 2 um thick; reticulate, muri relatively broad (ca. 1-1.5 um), flat, lumen ca. 2 um in diameter on distal side, smaller approaching colpus; 31-33 x 18-20 um Other occurrences. Culebra Formation. — FIGURES 1-8. Pan C-66, 1, ESF R-34, 3-4.— 3, 4. Monolete fern is type 1.—3. Pan C-57, 1, ESF C Pan C-59, 2 showing 10 monolete ted spore ne Í in Ds frame nariak feld of view) .—5. spore ‘ype 2 (small) . Pan C-63, 1, ESF K-21, G-21, 2—4.— 7. Loose cluster of monii fern s pes 3. jo C-62, 1, type 3. im C-57, 1, ESF G-2 Fossil 1 from the Cucaracha Formation, Panama.— 1. ii oen (trilete scar evident) . 2. Selaginella (trilete scar obscure, E emer ied an C-66, 1, ESF T-31. 24, 4.—4. Overview (10x) in upper left corner of Monolete fern onolete fern spore type 2 (large). Pan C-57, 1, ESF G-33, 2.—8. Monolete fern spore 1472 Annals of the Missouri Botanical Garden FIGURES : 15. dog pie from the ie ha ae irae Panama. . Monolete fern spore type 3 E. Pan C-65, 1, ESF P-22, 3-4; Pan C-63, 1, ESF K-11, 2; Pan C-63, 1, ESF H i 1-2.— 12. Cf. Antrophyum. Volume 75, Number 4 Graham 1473 1988 Cucaracha Formation — Lower Miocene Communities Distribution. Belize to Panama; tropical wet, margin entire, distinct operculum; tectate, wall 2 tropical moist, premontane wet forests; low alti- tudes Manicaria-type (Figs. 20-22). Prolate; monocolpate, colpus straight, 30-36 um long, ex- tending nearly entire length of grain, inner margin entire; e d possibly microreticulate); tri- lete, wall m thick; 36-45 x 20-24 um. Other occurrences. Gatuncillo, Culebra for- mations. Distribution. Antilles, Central and South America; wet places; low altitudes. Another palm pollen of the Manicaria-type was recovered but differs in size (25 vs. 36-45 um; Figs. 23, AQUIFOLIACEAE Ilex (Fig. 25). Oblate-spheroidal, amb circu- lar; tricolporoidate, colpi straight, 18 um long, equatorially arranged, meridionally elongated, equidistant, inner margin diffuse, pores obscure, situated at midpoint of colpus; intectate; clavate, wall ca. 3 um thick (length of clavae); 27 x 22 um. Other occurrences. | Gatuncillo, San Sebastian, Uscari, Culebra, Paraje Solo formations Distribution. Widespread; mesic to slightly drier habitats; low to mid altitudes; Costa Rica— tropical and premontane wet lowlands, montane rain forest. COMPOSITAE (Figs. 26-28). Spherical, amb circular; tri- colporate, colpi straight, short (ca. 10 um), equa- torially arranged, meridionally elongated, equidis- tant, pore oval, ca um, situated at midpoint of colpus; tectate, wall 2-3 um thick; echinate, echinae short (ca. 2-3 um), base broad, moderately dense (distance between spines ca. 3-4 um); 23- um. Other occurrences. Uscari (rare), Culebra (rare), and Paraje Solo (relatively common) for- mations. EUPHORBIACEAE Alchornea (Fig. 29). colpate, colpi straight, 8-10 um long (pole to equa- tor), equatorially arranged, meridionally elongated, equidistant, extending within 6-7 um of pole, inner Oblate, amb circular; tri- um thick; psilate to faintly scabrate; 15-21 um ther occurrences. San Sebastian, Uscari, Culebra, Paraje Solo formations. Distribution. Widespread; Panama—tropi- cal moist, premontane wet, premontane rain for- ests; Costa Rica—alluvial soil in tropical wet low- lands, mid-altitude wet and rain forests, altitude range 300-2,000 m JUGLANDACEAE Alfaroa/ Engelhardia (Fig. 30). Oblate, amb oval-triangular; triporate, pores circular, ca. 2 um, inner margin entire, equatorially arranged, equi- um thick; psilate; 21 um. Gatuncillo, San Sebastian, distant; tectate, wall 1 Other occurrences. Paraje Solo formations. Distribution. ically associated with lower to mid-altitude tem- perate forests. Mexico, Central America; typ- LEGUMINOSAE Caesalpinioideae Crudia (Fig. 31). Prolate; tricolporoidate, col- i narrow, straight, 25 um long, extending nearly entire length of grain, equatorially arranged, merid- ionally elongated, equidistant, pore area faint, cir- cular, situated at midpoint of colpus; tectate but with occasional separation between sculpture ele- ments, wall 1.5 um thick; distinctly and coarsely striate, striae generally oriented parallel to long axis of grain, surface psilate, margins entire, oc- casionally appearing beaded from underlying pores in foot layer/endexine; 32 x 20 um Gatuncillo Formation. ainly Amazonian, often riv- erine; low altitude. MYRTACEAE Eugenia / Myrcia (Fig. 32). Oblate to per- oblate, amb triangular; tricolporate, colpi narrow, straight, 9 um long, equatorially arranged, meridio- nally elongated, equidistant, syncolpate, pore ca. l um diam., situated on equator at midpoint of colpus; tectate, wall thin (ca. 1.5 wm); faintly sca- brate; 18 um Other occurrences. Gatuncillo, San Sebastian, Uscari, Culebra, Paraje Solo formations. E Pan C-59, 2, ESF X-20, 1-2.— 13. Pteris. Pan C-59, 2, ESF R-34, 2-4.— 14, 15. Ceratopteris. Pan C-59, 2, ESF X.45, 1-3; Pan C-59, 2, ESF H-33 1474 Annals of the Missouri Botanical Garden ` ‘> ean h + " * um dcm m e quaa 4 ES TOG š - e `Y> á - or ‘8 ye FIGURES 16-36. Fossil ope and p I the Cucaracha Formation, Panama.— 16. Trilete i spore type 1. Em C-66, 1, ESF K-35, 1.— 17, 18. Trilete fern spore type 2. Pan C-66, 1, ESF G-53, 3; Pan C-66, 1, ESF X-31, 1-3.— 19. oe -type ‘pale pollen. Pan C-66, 1, ESF P-33, 3.— 20-22. Manicaria- pe palm Volume 75, Number 4 1988 Graham 1475 Cucaracha Formation—Lower Miocene Communities FIGURES 37-39. 41, 1; Pan C-59, 2, ESF V RHIZOPHORACEAE Rhizophora (Figs. 33, 34). spheroidal; tricolporate, colpi narrow, straight, 14— 16 um, apices acute, equatorially arranged, merid- ionally elongated, equidistant, costae colpi ca. 3 um, pores elongated equatorially (colpi transver- salis), 1 X inner margin entire; tectate-perforate, wall 2-3 um thick; finely reticulate; 18- 6-20 um. Other occurrences. |. Gatuncillo, San Sebastian, Uscari, Culebra, and Paraje Solo formations. Distribution. Widespread, southeastern U.S. to South America; warm-temperate to tropical hab- itats, coastal brackish waters; sea level. Prolate to prolate- 4 um, constricted at midpoint of colpus, UNKNOWNS Type 1 (Figs. 35, 36). tricolporate, colpi narrow, straight, 6-8 um long, Oblate, amb circular; equatorially arranged, meridionally elongated, equidistant, syncolpate, pores circular to slightly ted equatorially, costae pori, situated on equator at midpoint of colpus; tectate-perforate, wall ca. 2 um thick; finely reticulate; 16-20 um Type 2 (Figs. 37-39). Prolate; tricolporate, Fossil pr m the Cucaracha an ia Unknown type 2. Pan C-59, 2, ESF 3-4; Pan C-57, 1, ESF M-1 colpi straight, 40-48 wm long, equatorially ar- ranged, meridionally elongated, equidistant, inner margin entire, costae colpi 5-7 um wide, pore slightly oval, 3 x 5 um, situated at midpoint of colpus; tectate-perforate, wall 2 um thick; finely reticulate; 43— x 30-42 um. Mher occurrences. Gatuncillo, Culebra, and Paraje Solo formations. NUMERICAL REPRESENTATIONS A total of 700 palynomorphs were tabulated from the Cucaracha material (100 from each of the seven samples; Table 1). Normally 200 pollen grains and spores are counted, but in the Cucaracha assemblage each sample was dominated by a single type, with all other forms collectively constituting only a small percentage of the e most abundant microfossils were monolete fern spore type 3, monolete fern spore type 1, Rhizophora, and Manicaria-type palm pollen. Vascular cryp- togams (ferns and Selaginella) totaled 71% of the flora, Rhizophora 17%, palms 9%, and all other angiosperm pollen 3%. Clearly, the sediments ac- cumulated in a fern marsh with palms surrounding itc (large). Pan C-59, 2, ESF N-43, 1-3; Pan C-59, 2, ESF K-28, 1-3; Pan C-59, 2, H-33, 4.—23, 24. Manicaria-type palm pollen (small). Pan 1, ESF E-34, 3-4.— i Alchornea. Pan C-59, 2, ESF H-21.— C-66, 1, ESF H-40, 3-4.—32. Eugenia/ Myrcia. Pan 1, ESF C-27, 1; Pan C-57, 1, ESF D-1 2, ESF S-41, 1-3 C-59, 2, ESF N-25; Pan C-57, 1, ESF U-11, 3.—25. Nex. Pan C-66, kd , 3.—35, 36. Db» bins 1. Pan € 173 1, ESF F. 19, 4; Pan C-59, 1476 Annals of the Missouri Botanical Garden and possibly intermingling with the ferns and with mangrove (Rhizophora) growing along the sea- ward margin of the depositional basin. It was a simple ecosystem in terms of community types, with little pollen of other associations being blown or washed into the basin from the surrounding landscape. Another quantitative feature of the Cucaracha flora is the rapid change in this coastal vegetation over very short periods of time. Samples 57, 58, and 59 were from a lower lignite bed about 0.6 m thick. All samples were taken from about the middle of the seam and horizontally were about two meters apart. Yet sample 57 contained about 60% Rhi- zophora pollen, sample 58 had 65% type 3 mono- lete fern spores (with only 13% Rhizophora), and sample 59 had 62% type 1 monolete fern spore (with only 7% Rhizophora and 16% type 3 fern spore). The vertical difference between the samples was, at maximum, only a few centimeters, and in rapidly accumulating, estuarine sedimentary ba- sin this likely represents only a few to several hundred years. Within this brief time span, three different taxa dominated the site. The upper lignite layer was more uniform in composition, with mono- lete fern spore type 3 dominant in all samples (Table 1). Among the seven samples from the two lignite layers, Rhizophora ranged from absent (sample 63) to 60% (sample 57); monolete fern spore type | from 62% (sample 59) to 1% (sample 65); and monolete fern spore type 3 from 77% (samples 63, 65) to 6% (sample 57). There were apparently some habitat differences within and be- tween the lignite layers, as well as differences in ecological preference between the ferns, because the percentages of monolete fern spore types 1 and 3 were reciprocal. These fluctuations evident in the pollen and spore assemblage are typical of estuarine habitats in tectonically unstable regions. light elevation of the land surface (or lowering of sea level) drains the lowlands marginal to the coast of salt water, and ferns dominate the freshwater habitats provided by inflowing rivers. With land subsidence marshes are inundated, the waters ren- dered brackish, and mangroves dominate the site. This sequence is repeated many times through the Culebra Formation, as evidenced by the alternating layers of lignites and lignitic shales; mud. silt-, and sandstones; water-lain volcanic ash (tuffs); and the basal oyster-rich conglomerate. PALEOCOMMUNITIES AND PALEOENVIRONMENTS A limited number of paleocommunities are re- vealed by the Cucaracha flora. A fern marsh, with associated palms and the floating fern Ceratopteris, are well represented. This freshwater assemblage was fringed seaward by mangroves (Rhizophora). Some indication of the inland vegetation is provide by a few spores of the tree ferns Cyathea and Pteris, and small amounts of pollen of Alchornea, Alfaroa/Engelhardia, Compositae, Crudia, Eu- genia/ Myrcia, and Ilex. These suggest tropical wet, tropical moist, and premontane forests on the adjacent slopes. It is likely that higher-altitude vegetation was meager, and its absence to poor representation in the microfossil record not just the result of exclusion from the sedimentary process. In other palynofloras (e.g., the is Sebastian flora of Puerto Rico and the P à Solo flora of Mexico), representatives from mid- to Le com- munities are common. Their poor representation in the Cucaracha flora is probably an accurate reflection of the kinds of communities occupying the low, insular landscape in the region at the time. Pollen grains of the Gramineae, other species associated with savannah habitats, and dry to arid vegetation were not recovered, consistent with re- sults from other Tertiary floras in Central America. The paleoenvironment can only be characterized generally as tropical and probably similar to that presently prevailing at lower altitudes in the re (see Graham, 1985: 531-532). All te recovered from the Cucaracha Formation occur in the modern vegetation of Panama. TERTIARY FLORAS, FAUNAS, AND VOLCANISM IN SOUTHERN CENTRAL ÁMERICA s data accumulate on the Tertiary floras of northern Latin America, it is becoming evident that little paleobotanical evidence is emerging for ex- tensive savannah or open-forest habitats. The lim- ited data do not preclude, however, more local, shifting, temporal stands of these communities in the Central American landscape. Such accommo- dation is necessary because of a seemingly anom- alous situation developing between the kind of vege- tation reflected in the Tertiary floras of Central America and the kind of habitats required by the few mammalian faunas known from the area. The palynofloras suggest that tall, dense tropical forests were prevalent on adjacent upland slopes, while the vertebrate faunas contain significant numbers of browsers (low open forests) and grazers (savan- nahs). Tertiary palynofloras from Central America (see chart in Graham, 1987b) contain little or no grass pollen, or pollen of trees or shrubs characteristically associated with open forests or savannahs. An ex- ception may be the few grains of Acacia pollen Volume 75, Number 4 1988 found in the Paraje Solo and Culebra floras. Bartlett & Barghoorn (1973) found little evidence for sa- vannah or drier open forests in Quaternary sedi- ments from Gatun Lake, Panama, but Leyden (1984) reported the presence of more arid vege- tation from Pleistocene deposits in Guatemala. The latter report, and the very small amounts of grass and Acacia pollen in Tertiary sediments, indicate that the list of Cenozoic paleocommunities for the region is still being developed. Nonetheless, the principal communities in the Cucaracha flora were likely the tropical moist, tropical wet, and pre- montane forests, with mangroves and fern and palm marshes occupying coastal and swamp habitats. In contrast, browsers and grazers are the prom- inent components in the two principal Tertiary vertebrate faunas reported from Central America. Olson & McGrew (1941) described a fauna from the Pliocene Gracias Formation, in the Mejocote Valley of Honduras. In addition to Amphicyon, Procamelus (camel), and Blickotherium (mast- odon), the most common remains were of the horses Pliohippus and Neohipparion. Whitmore & Stewart (1965) described a fauna from the Cu- caracha Formation, ca. 0.7 km from the plant microfossil locality. In addition to the marsh- and swamp-inhabiting turtle and alligator remains, they reported five browsing ungulates: Equidae (Anchi- therium, Archaeohippus), Rhinocerotidae (Dicer- atherium), Merycoidodontidae (Merycochoerus), and Protoceratidae (a selenodont artiodacytyl). The situation wherein fossil floras suggest forest while faunas, including the near-contemporaneous and adjacent Cucaracha fauna, reflect open forest and savannah on the uplands, is similar to that in the Tertiary biota of the western United States. A tentative explanation suggested for that region may apply to southern Central America. In a study of the Miocene Trout Creek flora of southeastern Oregon (Graham, 1963, 1965), megafossil species were not uniformly distributed through the 17-m section. In the middle part of the section there were numerous layers of sand and volcanic ash. Toward the top and bottom, the number and thickness of the ash layers decreased (Fig. 40). Megafossils were abundant in the upper and lower portions and were much less common in the middle part, where volcanic activity was greatest. Clearly the plant communities were dis- rupted by ejection of quantities of volcanic ash, the deforestation allowing sands to wash into the depositional basin from the surrounding landscape. Taggart et al. (1982; Taggart, pers. comm., 1987) suggested that the browsing and grazing faunas that characterize the middle and late Tertiary of western North America were exploitive, flourishing representative Graham 1477 Cucaracha Formation—Lower Miocene Communities TROUT CREEK CUCARACHA Depth Thickness Depth Thickness 0 === 71.5 1'8" 73.2 š 3 6'10” 80.0 7 pex 9 — rp 5” 24' Eure 1.5" 13 Ex 6" 104.0 15 [XU 8" 1' 2'8" 48'3" 1' P 152.3 Li ” 15" 1547 24 1' 10" 32'1" 95 m: 1856 & 40 FIGURE 40. Creek (late Miocene, Oregon) a Stratigraphic sections from the Trout c ash; bla borras sam siltstone; RN volcanic ash) or clay-silt- n both sections the plant megafossils (Trout Creek) and microfossils (Cu- lebra) are associated with sediments indicative of fre- quent and extensive volcanic activity. during recovery periods when open habitats were more extensive. The reduced vegetation, unfavor- able preservation conditions in the accumulating sandy ash layers, and the relatively brief time spans involved minimized the representation of this re- covery vegetation in the plant fossil record. During intervening times when volcanic activity was less, 1478 Annals of th Missouri Eco Garden FIGURES 41, 42. Panama.—41. Portable drill from truck be d. provide the descriptive lithologies (e.g., Fig. 40) u Drilling operations and resulting core from the s doris drm feu —42. Core from the Cucaracha Form . Lo sed to characterize the general PA dae iet of 6. region, m these cores the microfossil assemblages. Photographs taken nus, 198 forests became more extensive and contributed the abundant plant remains preserved in the diatomite. Another factor is the resistant nature of the faunal remains favoring preservation, transport, and ac- cumulation in the depositional basin. Figures 41 and 42 illustrate the coring device and the cores derived from the Cucaracha and other Tertiary formations in the Canal region of Panama. A description of the lithologies along these cores is kept by the Panama Canal Commission in the form of drill logs. A portion of one log from the Cucaracha Formation is presented by Graham et al. (1985: 495, table 3) and is summarized in Figure 40. It is clear that volcanic tuffs (water- lain volcanic ash) are common throughout the sec- tion (e.g., at levels 154.7', 104.0', 80', and 73.2"), and the same is true for other Tertiary formations in Panama. Basalt also caps many of the sections. If the pattern of Tertiary faunas with significant numbers of browsing and grazing elements, and Tertiary floras reflecting forested vegetation, per- sists in future studies, a shifting mosaic of brief, open, recovery vegetation in this volcanically ac- tive region may afford a reasonable explanation for the apparent differences in habitats suggested by the floral and faunal evidence. LITERATURE CITED BARTLETT, A. S. S. BARGHOORN. 1973. Phyto- peoga history of the Isthmus of Panama during the pas y mate, ds usine change). Pp. 203-299 1 Graham (editor), Vegetation and Vegetational History of Northern Latin America. Elsevier Publ. Co., Am- sterdam Volume 75, Number 4 1988 Graham 1479 Cucaracha Formation—Lower Miocene Communities Croat, T. B. 1978. Flora of Barro Colorado Island. Stanford Univ. Press, Stanford, California. D’Arcy, W. G. 1987. Flora of Panama Checklist and Index. 2 volumes. Missouri Botanical Garden, St. Louis, Missouri. Gastony, G. J. & R. TRYoN. 1976. Spore morphology in the Cyatheaceae 2. The genera Lophosoria, Me- taxya, Sphaeropteris, Alsophila, and Nephelea. Amer. J. Bot. 63: 738-758. caes iade revision of the Sucker Creek and Trout Creek oras of southeast- ern Oregon. Amer. J. Bot. 50: 921-936. The Sucker Creek and Trout Creek Miocene floras of bi aim Oregon. Kent State Univ. Res. Bull. 9: 1-147. 1976. Perl in Vins is qi paleobotany. II. The Mi f Veracruz, Mexico. Ann. Missouri Bot. Gard. 63: 787-842. 1985. Studies in ppt paleobotany. IV. The Eocene e es of Panama. Ann. Missouri Bot. Gard. 7 04-534. Bo poss communities and paleoen- vironments of southern Costa Rica. Amer. J. Bot. 74: 1501-1518. . 1987b. Tropical American floras and paleo- environments: Mexico, Costa Rica and Panama. Amer. J. Bot. 74: 1519-1531. 1988. Studies in neotropical pe V. The lower Miocene communities o — The Culebra Formation. Ann. Missouri Bot. di 15: 1440-1466 & 1969. Studies i in n nee core e Oli f Puerto y. LT Rico. > Missouri Bot. | Gard. 56: — ^R H. STEWART & J. L. STEWART. 1985. Stud- ies in neotropical paleobotany. III. The Tertiary com- munities of Panama — geology of the pollen-bearing eposits. Ann. Missouri Bot. Gard. 72: 485-503 308-357. HARTSHORN, G. S. 1983. Plants. Pp. 118-157 in D. H. Janzen (editor), Costa Rican Natural History. Univ. Chicago Press, Chicago. HorpnipcE, L. R. 19 Determination of world plant formations from simple climatic data. Science 105: 367-368. as me ENKE, W. H. HATHEWaY, T. LIANG & J.A 1971. Forest Environments in Trop- ical Life pad A Pilot Study. Pergamon Press, New Yor LEYDEN, B. W. 1984. Guatemalan forest m Fa Pleistocene aridity. Proc. Natl. Acad. U.S.A 4856-4859. Orson, E. C. P. O. McGrew. 41. Mammalian fauna from the Pliocene of Honduras. Bull. Geol. Soc. Amer. 52: 1219-1244 STEWART, R. H. & J. L. STEWART (with the xr of W. P. Woodring). 1980. Geologic Map Panama Canal and Vicinity, Republic of ue Scale: 1:100,000. U.S. Geol. Surv. Misc. Invest. Map I-1232. [Map also included in Woodring, 1982, a Surv. 306F. s & L. S. SATCHELL. 1982. Effects of petiodie Enn on Miocene vegetation distribution in eastern Oregon and western Idaho. Proc. III N. Amer. Paleontol. Conv. (Montreal, 1982) 2: 535-540 Tryon, R. M. & A. F. Tryon. 1982. Ferns and Allied Plants, with Special Reference to Tropical America. Springer-Verlag, New Yor WHITMORE, F. C. & R. H. STEW RT. 1965. Miocene mammals and Central saka seaways. Science 148: 180-185. Woopson, R. E. & R. W. ScHERY. 1943-1980. Flora of Panama. Ann. Missouri Bot. Gard. 30 (1943) et CUPULAR STRUCTURE IN PALEOTROPICAL CASTANOPSIS (FAGACEAE)! Robert B. Kaul? ABSTRACT ructure and some developmental aspects of the cupules of 22 species of paleotropical Castanopsis are The disc lr and illustrated. Some species have cupul are less spiny, and some are wb pe ooth. p the valvular scales and basally gus to them. Columns of scales straddling the sutures between the cupular valves do not become spines, but t assively invested with stro spines, but others n the valves become somewhat spiny with age. f some species have species are often indehiscent or only irregularly dehiscent and usually show nearly complete adnation of the cupule to the nut. Castanopsis is the third largest genus in the Fagaceae, after Quercus and Lithocarpus in their broad sense. Most of the approximately 120 species recognized by Camus (1929) and Barnett (1944) are paleotropical, but a few occur as far north as northeastern China, Korea, and Japan, and some ascend the Himalayas. For Malesia (the Malay Peninsula and all the islands from Sumatra to New Guinea and the Philippines) Soepadmo (1968, 1972) recognized 34 species and mapped species density. In Malesia, Borneo has the most species (21) and the most endemics (10); Sumatra and Java have 11 and 4 species, respectively, none endemic; and the Philippines have 4, including one endemic. Only two— C. buruana and C. acuminatissima — occur east of Wallace's Line between Bali and ombok. Barnett (1942) found only one endemic species among the 16 she recognized for Thailand. About 25 are known from China, mostly from the southern provinces. The eight or nine mostly en- demic Taiwanese species are taxonomically well studied (Li, 1963; Lin & Liu, 1965; Liao, 1971; Liu & Liao, 1976). One species (C. cuspidata), represented by two varieties, occurs in Japan (Ohwi, 1965). The one or two species outside Asia and nearby islands occur in western North America and are sometimes given their own genus, Chry- solepis, based largely upon cupular structure; pal- ynologically they are indistinguishable from Cas- tanopsis (Erdtman, 1943; Crepet & Daghlian, 1980), and foliar distinctions are weak (Jones, 1986). Lithocarpus, closely related and with per- haps 300 species, has a remarkably similar geo- graphic distribution, including a single species in western North America. Castanopsis in Malesia is characteristic of for- ests without strongly seasonal climates, but in Java and Thailand some species occur in seasonal cli- mates (Hjelmqvist, 1968; Soepadmo, 1972). Twelve of the 34 Malesian species are restricted to lowland forests, and the others are in lowland and montane forests. Barnett (1942) and Hjelmqvist (1968) not- ed that the species of Thailand are found mostly in the uplands. Most paleotropical species of Cas- tanopsis grow in mixed forests that often include Lithocarpus and Quercus, but C. acuminatissima forms nearly pure stands in New Guinea (Soepad- mo, 1972; Whitmore, 1975; Paijmans, 1976). The genus is prominent in subtropical forests of China (Wang, 1961). The northeastern Asiatic and western North American species are sometimes abundant in the forests and are, of course, sub- jected to more seasonal climates than are most paleotropical species. Pollination by small insects is suggested by their abundance around the rather strong-smelling, con- spicuous staminate flowers, but some pollen might be wind-transported. It is not clear how the incon- ' Research te: als by National Science Foundation grants DEB 8206937 and BSR 8508046. Grateful LG ne B. Abbe, A. L. and acknowledgment is made to cited earlier (Kaul & Abbe, 1984 ) Bogle, and all those persons and institutions r their Fé ose due in the field and laborator * School of Biological Sciences, rai, of Nebraska, Lincoln, Nebraska 68588- 0118, U.S.A. ANN. Missouni Bor. GARD. 75: 1480-1498. 1988. Volume 75, Number 4 1988 Kaul 1481 Castanopsis Cupular Structure spicuous, presumably odorless pistillate flowers at- tract insects; perhaps they are pollinated only in- cidentally to insect movements among the far more numerous staminate flowers around them. The published infrageneric classifications of Cas- tanopsis are based mostly upon fruit characters, as is the case for other fagaceous genera. Camus (1929) recognized three sections in Castanopsis: Eucastanopsis (cupule spiny, dehiscent or inde- hiscent, the nuts not fused to it); Callaeocarpus (cupule with or without spines, the nuts fused to it for most of their length); and Pseudopasania (cu- pule thin, dehiscent, containing a single free nut). The last group includes C. acuminatissima. Barnett (1944) saw the genus as comprising 11 groups in Asia and nearby islands, but she did not give them formal taxonomic rank. She defined them by degree of adnation of the cupule to the nuts, ornamentation of the cupule, size of the scar of the nuts, size of the fruits, and foliar characters. She included the small “fissa-group” in Casta- nopsis, noting that it forms a connecting link to Lithocarpus; Forman (1966a, b) also aligned the fissa-group with Castanopsis, noting its strong alliance with C. acuminatissima, but Camus (1929) placed those species in Lithocarpus subg. Pseu- docastanopsis. Thus these authors recognized the close affinity of Castanopsis to Lithocarpus, and Barnett (1940) suggested that recognition of two genera is perhaps more artificial than natural. Jones (1986) supported the placement of the fissa-group in Castanopsis, based upon his detailed study of foliar characteristics in the family. Camus (1929) suggested that Castanopsis is closer to Lithocarpus than Lithocarpus is to Quer- cus, and recent foliar and palynological evidence supports her interpretation (Jones, 1986; Zavada & Dilcher, 1986). (Lithocarpus and Quercus are usually interpreted as distinct from each other, but exhibit many parallel traits.) The affinity of Cas- tanopsis to Castanea is undoubtedly close. Hjelmqvist (1948, 1968) and Forman (1966a, b) saw Castanopsis and Lithocarpus as distinct, noting the differences between the dichasium-cu- pule, which encloses all the flowers of a cymule in Castanopsis, and the flower-cupule enclosing in- dividual flowers of a cymule in Lithocarpus. They interpreted the dichasium-cupule as a product of phylogenetic fusion of flower-cupules, and the di- chasium-cupule of the one-flowered cymules of the fissa-group and other species thus as indistinguish- able from a flower-cupule. The valveless cupules of Lithocarpus and Quercus are apparently mor- phologically identical also, but Forman (1966a) suggested that they are convergently derived, the cupule of Lithocarpus being a true flower-cupule and that of Quercus being a one-flowered dicha- sium-cupule whose lateral flowers have disap- peared. However, there is no direct evidence for that in Quercus; rather, it was based upon com- parative study of other genera, especially Trigo- nobalanus daichangensis. Some species of Litho- carpus show partial to nearly complete fusion of flower-cupules in the cymules, resulting in partial dichasium-cupules (examples illustrated in Kaul, 1987), which Forman (1966a) interpreted as phy- logenetically parallel to the dichasium-cupules of „astanopsis. In Castanopsis the cupular primordia are rather well developed by anthesis (but less so than in Castanea). The mature cupule encloses the one to seven nuts and often has distinct vertical sutures that define the cupular *'valves," at the edges of which dehiscence often occurs (such features are not found in Lithocarpus and Quercus). However, variations in these and other characteristics occur within and between species. The origin and structure of the fagaceous cupule have had various interpretations; Abbe (1974), Okamoto (1982), and Fey & Endress (1983) re- viewed the literature, in which it is implied that the cupule has arisen but once. Most workers in- terpret it as derived from pre-existing structures, not as a structure de novo. Whatever the origin of the cupule, it now has some qualities and func- tions unlike those of its ancestral parts. Fey & Endress (1983) found that ontogenetic observa- tions in Castanea, Fagus, and Quercus revealed more orderly arrangement of cupular appendages than is evident at maturity. This is true for Cas- tanopsis too, in which the mature cupule is often massively invested with rigid spines and other struc- tures that obscure positional relationships. Casta- Li nopsis is largely unstudied developmentally, but floral development in C. cuspidata var. sieboldii of northeastern Asia and Japan has been shown in detail (Okamoto, 1983). Castanea has been the subject of several developmental studies, most re- cently by Fey & Endress (1983). any species of Castanopsis have spiny cu- pules, but in others spines are barely evident or absent, and the cupule is then smooth and figlike. Forman (1966a) interpreted the spines as emer- gences, not foliar homologues. Barnett (1940) not- ed that they do not appear to be the first cupular appendages to form, but develop later, often in the axils of the "scales." Fey & Endress (1983) inter- preted spines of Castanea as axillary branch sys- tems. The patterns of distribution and elaboration of Annals of the 1482 Missouri Botanical Garden ES = NVS — O£88I UDyDpuDg qeqes pjjuydopisd PS ‘ZS SN ‘NIW '1'V SN ‘NI “1 ‘Y ZZIZ `P 19 moy seutddimqq deos »jj&udojisd 06 — 9£66 NL ja 2097 WIEN JIA snusofind 69-09 = SA ‘NIW “1 'V bES6 1P 12 204p WEN PIA snure) `V 2 ay snuzofuuid S9 = CSEG 1" 12 299p pue[teu 124121d cL cD: ts | SA ‘SNIS ‘NIW “T 48 ‘V PREG ]" 12 əqqF pue[eu aoueH 124421d 8c '9c-0c = NVS 289b€ UDYDPUDS qeqes smusofiao LE — NVS £EII£ UDYDPUDS yqeqes daog snuzofiao 91-141 '21-6 sn “T (TI V 02031 T? 1? 2097 Jemes DunÁ2]joui 61-21 ‘£I sa “T ‘a ‘Vv SN “1 q ‘AM V 9£0cI 1P 1? 2997 APMPIES Bury punA2]jour cg T ‘AMG ‘V ONIS 1'4N8'V — 9I86 P 12 299K azodeBu1s ‘dang (sean) Dp19m] 16 ‘69-99 V SA '9NIS NIN “A ‘AM ‘V 6596 7? 12 2399F eayondure y snui SD) "Y Y “YO 07010172d12u0] 68 SA NIN “1 ‘AM ‘V SN “ONIS ‘NIW “T V €I 26 “1? 12 23097 ekee sru21u1 IS-LP SA “ONIS ‘A ‘AM “Hd ‘V 6696 7? 12 2097 pue[reu, "y "1004 m "queg (PUT) snu42u? U V SN “9NIS ‘NIN ‘4 “AMA ‘V CEE6 1" 12 299p pue[eug ‘Od `V npu 6L TV TV SZZ “1D 12 2920g eipug "DA “Y xuisy L£-St = = 2166 `P 12 2997 runig "daos CWG `A) varrusoydod Xy v8 “vo ‘E9 = =a ZES6 1P 12 2097 UIEN| 191A snure) “y 14əum2 vL = — b803l `D 12 29qqy YeMBIEG 1Ay110mxof S7 '8-S == Sn “T ‘a ‘Vv I60cI 1P la 2097 APMPIES 1 y110mxof v-1 — TV F90cI 1D 1? ?qqp APMBJES Kynoqos 14y110mxof 18 SA ‘NIW “T “a V 999 `p 12 ə]5og UBMIB L ejeÁeg (uexs) vunsow.of 9v-Yvt SA “ONIS '1 ‘AM ‘V SN NIS ‘NIN “T “A “ANA 'V 6££6 “1D 12 20qp pue[reu L yoedg xosaf 96 “SS ONIS “1 “ANE 1ZL6 “1D 19 ?qqp kere Bury 11s124no 88 1 ONIS ‘AME ‘V 8S£6 “1D 12 2Qqp pue[eur qoeds pvu 8 SA T “q V SN ‘NIN “T “q ‘V 6bZ P 19 2]gog euung Bury 09744 doB 9L SN T q ‘AM ‘V ISb6 “19 12 2099F pue[eur Dunssimurumop rE ‘EE ANA £486 `P 12 əqqF puepeur DUNSSIJDUVUNID c£ = SNA l'V OI£0I T? 1? əqqF eauns) may endeg DUNSSIJDUNUNID LL ‘1€-62 = 1 ‘a ‘Hd ‘V ZOEOI “1D 12 20907 gəumo May] ended 'Puey (19) Vuaissivununoo 1ədeq su sueuredg mig suəunoədç umureqion $10]29][[o7) ƏoueuəAoid sei edc ut samiy jo seuoijisode( 19qonoA jo seuojsodeq pamaisnjpn sisdouejse?) fo sa12ads ayi fo suaun2ads aynandnp fo sa1101sodop pup əoupnuəao g ‘1 318v] Volume 75, Number 4 1988 Kaul 1483 Castanopsis Cupular Structure Continued. TABLE 1. Figures in This Paper Depositories of Fruit Specimens Depositories of Voucher Herbarium Specimens Collectors Provenance Species US A, BKF, SING, US BKF, L, SING Abbe et al. 9689 Thailand purpurea Barn. Malaya schefferiana 38, 39, 70 BKF, SING Thailand tribuloides (Smith) A. DC. tribuloides tribuloides Bogle et al. 241 Burma Bogle et al. 260 Burma cupular scales, spines, and sutures vary among the species, and the functional and phylogenetic sig- nificance of the various patterns is not known Forman (1966a) interpreted as primitive in Cas- tanopsis the three- or more-flowered cymule; the four-valved cupule dehiscing between the valves; the spines branched, well developed, and in definite rows; and the cupule not adnate to the nut(s). Those characteristics are most common in paleotropical species. He saw derived conditions in the one- flowered cymule, the two-valved cupules, irregular dehiscence not along valve sutures, the spines ab- sent or reduced and not in obvious rows, and the cupule fused to the nut(s). Those attributes occur in paleotropical and palearctic species. He also interpreted the American species as primitive be- cause the valves are free, and some occur between the flowers of a cymule—characteristics used to distinguish Chrysolepis from Castanopsis. Hjelm- qvist (1948) reported remnants of walls be- tween the nuts in the cupules of Castanopsis hul- lettii and C. fabr Hjelmqvist (1948) interpreted one-flowered, three-parted cupules as originating by the union of three flower cupules, one middle and two lateral cupules. He suggested that in four-parted cupules the middle cupule is reduced and the remaining common (dichasium) cupule is four-parted, the parts corresponding to two bipartite, lateral cupules. De- hiscence then occurs along the longitudinal fissures through the middle of the part-cupules and at the border between the two anterior cupules. MATERIALS AND METHODS My colleagues and I collected Castanopsis and other Fagaceae from India to New Guinea and Japan, gathering developmental reproductive ma- terial wherever possible. We preserved specimens originally in various liquid fixatives and later trans- ferred them to glycerine-alcohol for permanent storage. For three species from the lowlands of Borneo (C. foxworthyi, C. motleyana, and C. ovi- formis) we took samples at intervals from marked trees through most of a year. The species reported upon in this paper are listed in Table 1. The study of reproductive structures of Cas- tanopsis presents unusual technical difficulties. Sectioning and clearing, even of young flowers, do not produce photogenic results because of the hard spines and cupules, the numerous hairs, and the heavy sclerification of tissues. Specimens were ex- amined with light and scanning electron microscopy (often following removal of obscuring structures), drawings were produced with the camera lucida, and photographs were taken with incident light. 1484 Annals of the Missouri Botanical Garden The dense vestiture obscures many details, and so the line drawings are shown without hairs. Various sets of herbarium voucher specimens were distributed to A, BH, BKF, E, L, MIN, SAN, SAR, SING, and US (Table 1). The fruits illustrated here are in my research collection, and not all the distributed vouchers have fruits at all stages illus- trated in this paper. Nomenclature for paleotropical species follows Soepadmo (1972), who identified most of my pa- leotropical specimens. I considered the fissa-group with Lithocarpus (Kaul, 1987), following Camus's system, but a good case can be made for including it in Castanopsis if the two genera are recognized as distinct (Forman, 1966a, b). OBSERVATIONS PISTILLATE CYMULES The pistillate cymules (often called dichasia) are borne on androgynous, pistillate, or androgyne- candrous spikes (Kaul & Abbe, 1984; Kaul, 1986). They contain one, two, three, or sometimes more flowers (Figs. 1, 4-6, 10, 11, 13, 14, 17, 20, 25, 29-31, 38-40, 47, 52, 55, 56, 66, 70-77). The number of flowers per cymule is more stable in some species than others. Often there are fewer flowers in more distal cymules, and in some species that ordinarily bear one-flowered cymules, occa- sional two-flowered cymules appear. When three or more flowers occur in a cymule, the central flower often develops slightly earlier than the others (Fig. 10). Each cymule is subtended by a primary bract (shown in black in Figs. 1, 5, 14, 20, 25, 29, 30, 38, 47, 52, 55, 56, 60, and evident in Figs. 70, 72, 76) and secondary, tertiary, and sometimes quaternary or even higher-order bracteoles (Figs. 72-77). Distinctions between these bracteoles and those of the cupule (hereinafter called scales) are not always obvious. The pattern of bracteolation in Castanopsis is closely similar to that shown for Lithocarpus (Kaul, 1987 and in press). The apparent pedicel of a flower in anthesis (Figs. 1-3, 12, 21, 22) is actually the very un- developed inferior ovary, and the flower is thus sessile in the cupule. All species examined at anthesis have prominent staminodia (e.g., Figs. 30, 31, 77), which are some- times basally adnate to the perianth (Fig. 12). FRUITS The fagaceous nut develops from an inferior ovary, and the cupule is an accessory part. If all nuts of a cymule are abortive (probably due to lack of pollination or fertilization, or to competition for resources from other fruits on the rachis), cupular growth stops at about the stages shown in Figures 9, 7, 14, 18, 25, 31, 33, 35, 36, 39, 40, 48, and 53. But if one or more nuts mature, the cupule matures. The mature cupule is not fused to the nuts in some species; it is partially or completely fused in others. Some species have notably scaly and/or spiny cupules, others have rudimentary spines, and some are spineless or scaleless. A cupule surrounds all flowers of a cymule, and there are no well-developed valves (segments of cupules) between the flowers except in the Amer- ican species. Rudimentary interfloral valves were reported by Hjelmqvist (1948) in C. hullettii and C. fabri, but I have not studied those species and have not seen such valves in others. Spiny cupules with one nut typically have two valves and dehisce along lines between them (*'su- in the anatomical sense). In indehiscent fruits with one nut, the valves and sutures are evident at maturity in some species but not others. Fruits with three nuts typically have four valves, dehiscence occurring along the sutures between them (e.g., C. hystrix, Fig. 79; C. schefferiana, Fig. 82), but variations occur in both valve number and location of dehiscence lines. The mature cupule of C. motleyana shown in Figure 9, for example, has three fruits (one apparently abortive), but only two valves are evident. In C. tribuloides some cupules are one-flowered and have two valves (Figs. 1, 80), but other cupules, even on the same rachis, are three-flowered and have four valves (Figs. 42, 43). In the latter, some sutures between valves may not rupture. Species with one and three nuts per cupule occur at all elevations in southeastern Asia and south- western Pacific islands, but in Taiwan they are only at middle and higher elevations. The size ranges (1-6 cm) and averages (ca. 4 cm) of maximum cupular diameters in all the species are about the same in the two areas, except that in southeastern Asia and southwestern Pacific islands there are two lower-elevation species with very large cupules (to 10 cm in diameter). Average maximum cupular diameter in the species with one-nut cupules is about that of those with three-nut cupules, again excepting the extremely large cupules of two south- eastern Asiatic species, both of which have one nut per cupule. SPECIES WITH SPINY CUPULES In early anthesis the cupule is little developed and is hidden by the bracteoles (Figs. 10, 11, 29, 38, 47, 55, 72-76), but the cupular scales quickly Volume 75, Number 4 aul 1485 Kau Castanopsis Cupular Structure CASTANOPSIS FOXWORTHYI C. MOTLEYANA " > y j "X AN v bee y 5 sc D AR, K N ^ "e ITE > ff fiiv 277777777». de = (a , E showing three cymules oles are removed in the middle cymule, cupule with the spines now in bands and somewhat confluent. The sutural scales (SSC) have not enlarged as much as the valvular scales (VSC) , but the axillary spine clusters (SP) are now mostly much larger than the SC. J valves shown is indicated by SU, 1.9x.—9. C. motleyana. Mature, dehisced cupule in polar view, revealing one abortive and two functional nuts. The cupule has split on only two of its four sutures, 1.9x. 1486 Annals of the Missouri Botanical Garden CASTANOPSIS MOTLEYANA FicunES 10-28. 10-19. Castanopsis motleyana.— 10, 11. Three-flowered cymule in anthesis, the primary bract and the bracteoles removed in Figure 11 to reveal early cupular scales, the lower two (arrow) being the first sutural scale pairs, 7.5 X.— 12. A single flower with perianth removed, revealing staminodia basally adnate to the tepals and showing the very undeveloped inferior ovary, 7.5 x .— 13. Three-flowered cupule in late anthesis, has formed between the lower the sutural scales (dashed lines) of the right lateral suture removed to reveal the valvular scales and their 15, 16. i immature, a e the axillary spines (black), 15x.— 17. Cupule about one-third mature, the sutural scales (SSC) and valvular scales (VSC) clearly differentiated, and the axillary spines (black) just emerging from among the scales, x.— 18. Longitudinally halved immature cupule revealing two of the three nuts, 1.9x.— 19. Two mature Volume 75, Number 4 1988 Kaul Castanopsis Cupular Structure 1487 become evident and soon exceed the bracteoles (Figs. 5, 14, 25, 30, 39, 40, 52, 56, 77-83). In the species examined for preanthesis develop- mental detail (Castanopsis foxworthyi, C. motley- ana, C. oviformis), the cupule primordium arises as an undulate ring at the base of the flowers, well within the bracteoles (Figs. 13, 23). In the three- flowered cymules of C. motleyana, four growth centers (valve primordia) of cupular scales are ev- ident in the early stages (Fig. 13, scale dcn shown in black), but growth centers in C. fox- ape and C. s iple: are di obvious, although g various parts of the early cupule E 4, 23, 24). Distinctions develop among the cupular scales during and after anthesis, especially in cupules that will be spiny at maturity. Columns of shorter, broader, sometimes connate or paired scales alter- nate with masses of apparently random, pointed scales (Figs. 5, 6, 14, 17, 28, 31-33, 48, 49, 51, 58). The columns mark the sutures between valves and are potential sites of dehiscence, at which time sutural scales become separated onto adjacent valves (Figs. 34, 41-43, 54). In the three- flowered cupules of Castanopsis foxworthyi, C. motleyana, and C. inermis there are four such columns: one adaxial, one abaxial, and two lateral (Figs. 6, 13, 51). The columns define the four valves. The first pair of sutural scales is clearly evident in Figure 11, in which the primary bract has been removed to reveal them and the earliest valvular scales. Forman (19662) found that dehis- cence of the cupule of C. inermis (Figs. 48-51) does not occur along the sutures between the valves. The valvular scales between the columns of su- tural scales elongate, becoming indurate and spine- tipped with age. They do not comprise the majority of spines of the cupule; instead, most of the spines develop later, in groups axillary to the valvular scales; such spines are shown in black in Figures 6, 14-17, 26-28, 31, 32, 35, and 36, and are evident in Figures 72, 77-83. Figures 15 and 16 show that the spines in each group arise nonsyn- chronously and free, but they become basally con- nate and ultimately adnate with the subtending scale. By maturity, each subtending scale becomes spine-tipped, and the axillary spines are as large as or larger than it (Figs. 8, 19). Thus, in these spiny-cupuled species, the armament of the cupule is provided mostly by the axillary spines, with some contribution by the original valvular scales (Figs. 8, In some species with less cupular armament (e.g., Castanopsis tribuloides, Figs. 38-43, 80; C. ferox, Figs. 44-46; C. psilophylla, Figs. 52- 54) the sutural scales are less obviously paired but instead occur in irregular groups. Such groups are evident in Figures 41-44 and 54, which show dehiscence of the cupule through them. In these species, the cupular scales become indurate and spine-tipped, forming much of the cupular arma- ment; the mature axillary valvular spines are of similar size to the scales. In Castanopsis curtisii (Figs. 55-59, 87) the asymmetric cupule is totally fused to the nut, the surface of which is therefore entirely scarred (Fig. 87) except for a tiny portion near the perianth. The flower and young cupule are rather symmet- rical at anthesis (Fig. 55), but asymmetry is es- tablished soon thereafter (Fig. 56); massive abaxial growth of the cupule turns the fruit upward about 90°, and it becomes anatropous. Thus at maturity all the sutures appear to be on the adaxial side (Fig. 59). Sutural scales are evident only in the early stages of cupular development, when they can be seen to be entirely adnate to the body of the young cupule (Fig. 56, left cupule). The cupular scales in this species are also largely adnate, and are irregularly disposed at maturity (Fig. 59). In dry fruits, rupturing begins in the adaxial suture, but other ruptures radiate from the cupular pore (Fig. 59); those lines apparently do not represent spine clusters, each consisting Vies spine-tipped T š scale (VSC) with its axillary cluster odas (SP), — 3.8x. 20-28. C. oviformis. — pike in anthes anthesis, the lower with t staminate, 7.5% undeveloped, but staminodia prominent, 7.5 Xx.—23, 2 removed to reveal the young cupule, 15x. cupule now emergent from the bracteoles; two pairs of large, sutural scales (arrow) a most cymule; cymule on lower right removed to reveal persistent bracteole (black), side of the upper 26. Still older cupule with 12 valvular scales removed to rev 8. Immature fruit in lateral (Fig. 27) and i en (Fig. 2 7 25X.—27, 2 Four one-flowered pistillate cymules are shown in i e flower ud to reveal the D developed cupule. Unopened ve owns are .—21, 22. Pistillate flower in porey the tepals removed in Figure 22; inferior ovary very 4. Cymule in later anthesis, Kus primary bract and bracteoles — 225. Segment of pistillate portion of spike well afier anthesis, the re evident on the Re i eal tiny upyan (black) of the pir id spines, 8) view, the axillary spines (black) now as among the valvular scales; Figure 28 also shows the abaxial column of large sutural scales, 2.8X 1488 Annals of the Missouri Botanical Garden C. ACUMINATISSIMA C. TRIBULOIDES FiGURES 29-46. 29-34. Castanopsis acuminatissima.— 29. One-flowered cymule in anthesis, the primary bract in black and a few cupular scales evident, 2x.— 30. Late anthesis, the cupular scales evident and overtopping the basal bracteoles; perianth and staminodia obvious, 4x .—31. Sti j ] j , spines (black) emerging from si -— scales, the abaxial suture indicated by the central, vertical depression abutted by sutural scales, Ax. 32 . Still later stage of a specimen showing three sutures defined by appressed scales; between the sutures are pi of axillary spines (black) and valvular scales, 4x.—33, 34. sci 35-37. “sear — Abaxial view ides same cupule showing the prominent recurved cupular scales and the emerging n spines ig i 2x.—37. Mature cupule, the scales and spines now widely separated and equally DE tribuloides, — 38. One-flowered cymule in early anthesis, the primary bract in black, 2x after anthesis, the relatively ge cop adaxia ain eviden ej in Figure 39, the Figure 40, 2x.—41. Mature, dehiscing cupule in abaxial view, the rounded sutural scales evident adjacent to the suture, but the rows of valvular inus are no ra orderly, 2x.—42, 43. Adaxial (Fig. 42) and abaxial ig. 43) views of dehiscing cupule containing three nuts. eem scales of the partially opened adaxial a eu ^ue ap rows i sc sales obvious in Volume 75, Number 4 1988 Kaul 1489 Castanopsis Cupular Structure intervalvular sutures, or at least such sutures are not evident in earlier stages (Figs. 57, 58). These fruits are possibly indehiscent in the wild, where they perhaps never dry, but my dry specimens show regular patterns of splitting. Four related species reveal comparative details of reproductive structure: Castanopsis armata, C. inermis, C. lucida, and C. pierrei. The cupule is almost entirely fused to the nut in C. pierrei (Figs. 65, 86). Figure 65 shows the massive scar (hatched) and two small abortive nuts near the distal end of the single mature nut. The small free portion of each nut is indicated in black. The relationship of the mature and immature nuts to the cupule is evident in Figure 86, a polar view wherein the partially open cupule reveals two abortive nuts on the right. In this species the cupular scale-spine complexes are massively thickened (Figs. 65, 86). The spine-tips are not much elongated, but they are very sharp. The scale-spine complexes are arranged in slanting rows on the cupule, but that is not evident in Figures 65 and 86. Castanopsis armata is similar to C. pierrei in that the cupule is entirely fused to the nut, but my specimens have no abortive nuts, and the single mature nut has about 20% of its distal end free (nonscarred) (Fig. 88). The mature cupule is mas- sively thickened, and the cupule scale-spine com- plexes are not aligned in evident rows at maturity but are even more massive than those of C. pierrei. Each spine has a short, indurate, subulate tip, and the cupule is therefore formidably armed. Imma- ture fruits display cupular sutures (cf. Fig. 51), along which dehiscence probably occurs eventu- ally. In Castanopsis inermis the cupule contains one to three nuts and is largely free from them (Fig. 89). Forman (1966a) noted a tendency toward reduction of the central flower of each cupule in this species, but I cannot confirm or deny that from observations of specimens from the Malay Penin- sula, which show no regular pattern of reduction. The mature cupule displays horizontal rows of groups of low spines (Fig. 89, center). Such regular arrangement is more obvious on immature cupules, where the sutural scales are also readily distin- guished (Figs. 48-51). The latter are permanently adnate to the cupule and do not become spiny. The cupular armament is largely made of spine- tipped cupular scales; axillary spines, if present, contribute little. Usually four sutures form on three- flowered cupules, and two form on one-flowered cupules. Variation occurs; some one-flowered cu- pules have four sutures (Fig. 51), for example. Dehiscence does not follow the sutures entirely and is therefore rather irregular. he large fruits of Castanopsis lucida (Fig. 85) resemble those of C. inermis, but the cupular scale groups are proportionally larger and, at least in my specimens, the cupules seldom contain fewer than three nuts. The four sutures of a three-nut cupule are clearly evident in Figure 85 (upper photograph). Dehiscence usually begins with the opening of the adaxial and abaxial sutures, which are the shorter ones; later the longer lateral sutures open and reveal the free nuts within (Fig. 85). Mature spikes sometimes bear as many as ten fruits that collectively weigh 70 g when dry. SPECIES WITH SMOOTH CUPULES Some species of Castanopsis have nearly smooth, figlike cupules. Examples are shown of C. pirifor- mis, C. guineri, and C. longipetiolata (Figs. 60- 64, 66-69, 84, 89-91), all placed in section Cal- laeocarpus by Camus (1929). In these the inde- hiscent cupule is fused to the single nut and is unadorned except for low ridges and small, mostly adnate scales. No axillary spine complexes form. The cupular scale-bearing ridges are widely sep- arated and not concentric. In Castanopsis piri- formis and C. guineri they converge distally on the cupule and there form a continuous spiral around the aperture (Figs. 62, 63, 84, 90). The ridges of C. longipetiolata are vertical in places (Figs. 67- 69, 91) and converge near the aperture, but do not form a spiral (Fig. 67). In adaxial view (Figs. 61, 64, 68), the adaxial suture is barely evidenced by a broad longitudinal depression. A few sutural scales show in Figures 61 and 64, but there are none in Figure 68. The cupular scales enlarge little as the cupule swells, and thus they become somewhat remote. In Cas- — suture are seen to intergrade with the pointed valvular scales in Figure 42, but sutural scales on the abaxial side are confined to the base of the cupule (Fig. 43), in abaxial, lateral, and adaxial views, respectively. — 44-406. C. ferox showing mature, dehiscing cupule The adaxial side is essentially naked, and the suture there is without adjacent scales; the abaxial suture (Fig. 44) has a few scales, and the valvular scales are irregularly 2X. disposed at maturity, 1490 Annals of the Missouri Botanical Garden C. INERMIS C. PSILOPHYLLA C. CURTISII FIGURES 47-59. 47-51. Castanopsis inermis. — 47. Three-flowered cymule in pd anthesis, the primary bract shown in black, and only two cupular scales visible between the basal bracteoles, 2x.— 48. Immature cupule in lateral view, the abaxial suture on the left shaded, the valvular scales still in rows, 2x.— 49. Nearly mature cupule in adaxial view, the rounded sutural scales of the adaxial suture in tiers, the valvular scales less orderly than in Figure 48, 2x .— 50. Lateral view of nearly mature cupule, showing rows of valvular scales, 2x .— 51. Oblique polar view of nearly mature cupule, the four sutural areas evident and defined by appressed, more or less rounded veda scales, and the valvular scales fully enlarged and pointed; there are no evident axillary spines, 2X, 52-54 a — 52. Three-flowered cymule in anthesis, the primary bract in black and the large basal oral in white; numerous cupular scales are evident in rows above the bracteoles, 2x.—53 Postanthesis, the cupule her its scales enlarging, the latter yet in evident rows; perianths and staminodia of the three flowers persisting, 2X.—54. Mature, dehiscing cupule in abaxial view, showi few elongate, appressed sutural scales to the left of the suture; valvular scales yet in rows but somewhat separated by diametric growth of the cupule, 2x —59. C. curtisii. — wered cupule in early anthesis, the primary bract ne- r shown in black, the cupular scales present under the basal bracteoles but not evident in this figure, 2x .— Two cymules soon afier anthesis, the cupule already prominent above the primary bract and basal alla. Volume 75, Number 4 Kaul 1491 1988 Castanopsis Cupular Structure C. PIRIFORMIS pou 60-69. 60-62. EIA piriformis.—60, 61. Half mature rper. cupule in abaxial (Fig. and semiadaxial (Fig. 61) v - The primary bract is shown in black in Figure 60. Cupular scales are Ln torn and widely spaced "ie this stage. The potential adaxial suture is seen on the lef in Figure 61, although the fruit is probably Aya oaa 2x.—62. Mature cupule in oblique view, the adaxial side to the right. Upper rows of scales becoming spiral, 2x .—63, 64. C. guineri, nearly mature fruits in abaxial (Fig.63) and adaxial (Fig. 64) views. The strong spiral formation of cupular ridges is evident in Figure 63, and Figure 64 shows the adaxial vertical depression marking a potential suture, although the fruit is indehiscent, 2x 65. C. pierrei. Mature fruit with foreground part of cupule removed to expose the massive scars (white) M the three nuts, two of which Logros only partially; free portion of the nuts shown in black. nuts are not fuse to each other, 2x. 66-69. C. longipetiolata. —66. Two one-flowered cupules in anthesis, iis primary bract of each shown in black; us row of cupular scales is apparent above the basal bracteoles, 2x .—67-69. Nearly mature cupule in polar, adaxial, and lateral views, respectively. The few surviving cupular scales are mostl qos to the curving cupular ridges, and there is no obvious indication of potential sutures on this indehiscent cupule, 2x. == the scales m or less in orderly rows; the rather smooth, unadorned adaxial face of m: fa a is easily Fi seen, 2x .—57, 58. Fruit about half grown, in lateral (Fig. 57) and adaxial (Fig. 58) views. Massive adaxial growth has S strong asymmetry, and the pore through which the tip of the nut can n be seen is now more than 90° from the vertical. Scales wi no longer in rows, and some slight differentiation is beginning Ara to the right and lefi of the pore, 2x.—59. Mature dehiscing cupule in adaxial view. Five lines of dehiscence are shown on this specimen, the vertical one through the upper, unadorned area visible in Figure 58 aoe the two upper, lateral ones between the barely differentiated scales, 2x Annals of the Missouri Botanical Garden FIGURES 70-73. 70. Castanopsis tribuloides, lateral view of cymule in anthesis. The po, pray bract is lowermost, but no cupular scales show between it and the prominent perianth (arrou), 65x.—71. C. indica, Volume 75, Number 4 1988 Kaul 1493 Castanopsis Cupular Structure tanopsis longipetiolata there are few such scales from the beginning, so by maturity the cupule is nearly scaleless (Figs. 67-69, 91). No axillary spine complexes form. DISCUSSION Castanopsis shows primitive and advanced character states for the Fagaceae. If multiflowered cymules are primitive in the family, as suggested by Soepadmo (1970, 1972), Forman (1966a), and Kaul (1987), and by widely accepted interpreta- tions of inflorescence evolution in the angiosperms, then it is likely that some associated character states are primitive too. Among the latter possi- bilities are four-valved cupules, dehiscence between the valves, the cupule free from the nut(s), and spines present on the cupule (Forman, 19664). The general association of one-flowered cupules with spinelessness, adnation of the cupule to the nut, and irregular or no dehiscence thus suggests those states to be advanced. Forman (196062) pos- tulated such interpretations based upon other species, and they are supported by my observa- tions. If the interpretations of Hjelmqvist (1948) and Forman (19662) are correct, the dichasium cupule evolved by phylogenetic fusion of adjacent flower-cupules in the cymule and is thus highly derived. Strong cupular asymmetry approaching anatropy, as in C. curtisii, is certainly a derived state, although most species show at least some asymmetry. There is some developmental evidence that ad- nation of cupule to nut is only apparent. Rather, it is possible that the basal scar enlarges enormously as the nut grows in species with apparently exten- sive adnation and that the basal part of the cupule expands simultaneously. The result is wide spacing of the lower cupular lamellae, but distally they remain rather crowded. With this interpretation the figlike, one-nut cupules can be viewed as the products of massive basal but little distal enlarge- ment of the cupule. The distal part of the nut, which in other species enlarges more than the basal part, remains free, as it is in the species without apparent cupular adnation. Detailed developmental investigations, though technologically formidable, could provide further insights into this idea. In some species, the scales and spines are ar- ranged in concentric rows, but in others there is no apparent order in mature cupules. However, in the latter instances observed for this study, early developmental stages show rather regular rows of cupular scales, and I regard such rows as the primitive condition in the genus. In some species with figlike cupules (e.g., Castanopsis guineri, C. piriformis) the parallel rows of scales on the lower part of the mature cupule contrast with the distal, spiral rows, which are the last to form. The cupules of some species of Quercus subg. Cyclobalanopsis also show spiral lamellations, but most species of that subgenus, all of subg. Quercus, and nearly all of Lithocarpus have concentric lamellae (Kaul, 1985, 1987, and in press). Thus spiral lamellations are exceptional in the Fagaceae and are likely to have been derived by of the concentric pattern through alteration of later ontogenetic LG stages. Forman (1966a) interpreted the sutural scales as derived from branched spines because he be- lieved there are intermediates between scales and He thus considered the cupular scales of Lithocarpus and Quercus also to be modified spines. He saw the spines as single branched entities, but Barnett (1940) interpreted them as axillary to the first-formed cupular ap- pendages. I also interpret the sutural and valvular scales to be homologous and synchronous, and the spines in some species. axillary spines to arise above the valvular but not the sutural scales. Thus intermediates between the two scale types are expected and, in fact, have been observed by Forman (1966a) as well. lant spines are sometimes emergences, not stem or leaf homologues, and that was Forman's (1966a) interpretation of the cupular appendages of Cas- tanopsis. Evidence in favor of the spines of Cas- tanopsis being modified stems are their branching capability (which sometimes occurs in other fam- ilies in spines originating as emergences) and their position axillary to the scales. However, there are not even ephemeral subtending bracts at the nodes of branching on the spines, as could be expected in a reduced branching system. Furthermore, it is possible that the scales and spines, as defined here, are homologues and therefore not fundamentally different as suggested by the scale/axillary spine p^ view of cymule x.—72. m the lanas basal bracteoles and below the perianths, 29 . pierrei, abaxial view of three-flowered cymule in anthesis. . soon after anthesis, five spines (arrow) npo pis h the bracteoles and hairs, vis ad ales (arrow) are evident x.—73. é allein abaxial view of cymule soon after anthesis, the styles still showing. The cupular se are now prominent above the basal bracteoles, dx. Annals of the Missouri Botanical Garden FicURES 74-77. 74, 74. Abaxial view of three-flowered cymule in late anthesis The subtending primary id is seen in the lower right corner, but the bracteoles, emerging cupular scales, an« much of the perianths are obscured bs ae 32x.— 75. Semiabaxial view kh a c nm not long after anthesis, the abundant cupular scales e emerged from RE the basal 2 ract of the subulate structures are probably emerging spines, 24x. 76, 77. C. acuminatissima.— 76. One ede, c uus at anthesis, in nearly abaxial view. The prominent. primary bract is helo. and ie equally Din subte nding bracteoles are just above it; only the perianth and styles emerge at this sta ;ymule not long after anthesis, in nearly abaxial view. The Nu primary bract below is [Ben in focus, but the numerous c 'upular scales are pl lainly merging subulate spines ee which are yet appressed to the young cupule. Some x. 75. Casta — foxworthyi.— aml d seen, as are a few e shriveled staminodia are seen below the styles, 1495 Volume 75, Number 4 Kaul 1988 Castanopsis Cupular Structure R Mature rie i) uium — 78. C. argophylla. Cupule dehiscing on two sutures, the 8x.—80. C. FicunEs 78-84. nut free from it, 1.6x.—79. C. . Three-nut cupules der ap n four sutures, an nuts free, O. tribuloides. One-nut cupules Poe id on two sutures, the nut free, 1.2x.— 81. C. for osana. Abaxial and qe views (upper and lower cupules, respectively) . The n on adaxial surface is seen in the lower cupule, 1 x .— C. schefferiana. Lateral and polar views (upper and dl ca of a three-nut a dehiscing on bur sutures; the basal scar (arrow) or a nut shows in the lowest feu , 1.2x.—83. armed with heavy, curved spines; the unadorned suture is evident in the center, 1.3x .— 84. C. slightly asymmetric cupule in lateral views, showing the lentas with the few persisting scales, 1x (cf. Figs 1496 Annals of the Missouri Botanical Garden 88 SURES 85-91. Mature fruits of Castanopsis. — 85. C. lucida. Mature, dehiscing cupule in polar (upper) and abaxial (lower) views. This specimen has three nuts free from the cupule, the central one apparently abortive (upper figure), 1x.—86. C. pierrei. Mature fruit in lateral (upper figure) and polar (central figure) views. In js Fk figure, two abortive nuts are revealed to the right of the large, fertile nut (cf. Fig. 65). The fertile nut is shown removed from the cupule in the lowest figure; only the dark, upper part was free from the cupule, 0.8x .— 87. C. curtisii. Upper figure: mature cupule in nearly adaxial view, the entire fruit anatropous, i apical pore visible lefi-central. Lower fe ure: mature nut r moved . rom cupule and viewed laterally. The scar covers nearly the entire surface, 1X. . C. armata. Maure nut upper figure) entirely scarred from fusion to the cupule, and lateral view laur] ure) of mature cupule with its massive but sparse armament, 1.3x .— 89. inermis. More or less polar views (upper two figures) o due cupule ipi along two sutures and Mone only weak definition of sutural (unadorned) and valvular areas (cf. Figs. 47-51). Lowest figure is the mature nut, its basal scar (arrow) clearly evident, 1 Xx .— 90. C. piriformis. Lateral view of mature cupule enda: fused Š ~ Volume 75, Number 4 1988 Kaul 1497 Castanopsis Cupular Structure interpretation. For Castanea, Fey & Endress (1983) interpreted the valvular scales to be residual pherophylls of a condensed, cymose system and the spines to be axillary, cymose branches, based upon their comparative and developmental studies. I favor that interpretation for Castanopsis as well. If the branched spines are relictual axillary branches— vestiges of a condensed, fused branch- ing system that formed the cupule—then their presence is the primitive condition in the family. Their loss, both within Castanopsis and elsewhere in the family, would be the derived state. Spininess due to induration of cupular scales in a few species of Quercus and Lithocarpus would then be a sec- ondary development. Whatever interpretation is correct, there are unanswered questions about evolution of the var- ious patterns of spine elaboration and distribution on the cupules. While protection of the immature nuts is a likely function of the spines, the fact that some nonspiniferous species of Castanopsis grow in the same forests with spiniferous species suggests that other protective measures are effective too. Detailed studies of life histories of Castanopsis and of its fruit predators and disseminators could reveal important information about cupular function. It is likely that the solitary pistillate flowers of some species of Lithocarpus and Castanopsis and all species of Quercus are the results of phyloge- netic elimination of all flowers but one (presumably the terminal one) of a cymule, as suggested by Forman (1966a). However, it is by no means ev- ident that abortive flowers are M phylogenet- ically abortive flowers that failed to lack of pollination or fertilization, or due to com- petition from other fruits on the same rachis. Such ——sometimes they a merely normal produce fruit, probably for failed flowers are frequent even among species with solitary flowers. In this paper and elsewhere (Kaul, 1987 and in press) examples are illustrated showing failed flowers of multiflowered cymules in every position in the cymules, varying greatly even on a single spike. If there are tendencies for lateral flowers in pistillate cymules of these genera to be phylogenetically abortive, I have seen little con- vincing evidence. However, such evidence appears in a few staminate spikes of Lithocarpus and pos- sibly also of Castanopsis. There are more differences in cupular structure within Castanopsis than between it and some species of Lithocarpus, and the distinctions between flow- er-cupules of the latter and dichasium-cupules of the former are not always convincing, especially in Lithocarpus species with cupules partially of both types. Thus, once again the generic and in- frageneric taxonomy of Fagaceae is questioned: the long history of taxonomic doubt expressed in the literature is not eased by such observations. However, it is premature to propose new align- ments because of the confusing levels of apparent parallelism and convergence in these large genera. LITERATURE CITED E. C. Flowers and inflorescences of the “*Amentiferae.”” Bot. Rev. 40: 159-261 BARNETT, E. C. 19 A Survey of the Genus Quercus and Related Genera of the Fagaceae in Asia, with a More Detailed Account of the Siamese Species of These Genera. D.Sc. Thesis. University of Aberdeen — ). ABBE, e Fagaceae of Thailand and their — 1 distlbuco: Trans. Bot. Soc. Edinburgh 7-343 OO. «€ Keys to the species groups of Quercus, Lithocarpus, and Castanopsis of eastern Asia, with notes on their distribution. Trans. Bot. Soc. Edin- burgh 34: 159-204 Camus, A. 1929. Les chátaigniers. Monographie des genres Castanea et Castanopsis. Encycl. Econ. Syl- vic. 3: 1-604. CREPET, W. L. & C. P. DacHLIAN. 1980. Castaneoid inflorescences from the Middle Eocene of Tennessee and the diagnostic value of pollen (at the subfamily level) in the Fagaceae. Amer. J. Bot. 67: 739-757 ERDTMAN, G. 1943. An Introduction to Pollen Analysis. Ronald Press, New Yor Frey, B. S. & P. K. ENDRESS. 3. Development and morphological interpretation of the cupule in Faga- ceae. Flora 173: 451-469 . 1966a. On the evolution of cupules in the Fagaceae. Kew Bull. 18: 385-4 1966b. mi EN Bil aaa in the Castaneoi- deae. Kew Bull. 18: 426. HiELMQvisT, H. 1948. Sm on the floral morphology Pa phylogeny of the Amentiferae. Bot. Not. Suppl. Tod Studies in the flora of Thailand 44: Fagaceae, Betulaceae and Corylaceae. Dansk Bot. Ark. 16. NES, J. Aa A Evolution of the Fagaceae: the implications of foliar features. Ann. Missouri Bot. rd. 73: 228-275 < to the nut), more clearly in Figure 62, 1x.— 2. longipetio the weakly defined, remote lamellae essentially without scales; the distal lamellar spiral is seen ata. Matur e cupule in lateral view (entirely fused to the nut) showing he fs hoked lamellar ridges. The three dark spots are areas apparently damaged by chewing animals, 192 X, 1498 Annals of the Missouri Botanical Garden Kaul, R. B. 1985. p pedi e a of Quer- cus. Amer. J. Bot. 62- 19 Evolution and sd M biology of inflorescences in Lithocarpus, Castanopsis, and Quercus (Fagaceae). Ann. Missouri Bot. Gard. 73: 296. P 1987. Reproductive structure of Lithocarpus sensu lato (F nara eymules sl faite: J. Arnold Arbor. 68: 73-104. Fruit structure and ecology in Linen dn Lithocarpus. In: S. Blackmore & P. Crane (editors), Evolution, Rose and Fossil History of the Ha- mamelidae. Oxford Univ. Press, Oxford, England. (In press ay — & E. C. Aspe. 1984. Inflorescence architec- ture "s evolution in the Fagaceae. J. Arnold Arbor. 65: 375-401. Li, H.-L. un Woody Flora of Taiwan. Livingston Publ. Co., Narberth, Pennsylvania. Liao, J.-C. 1971. Morphological studies on the jan and fruits of the genera Fagus, Castanea, Cast nopsis and Limlia in Taiwan. Mem. Coll. Agric. Natl Taiwan Univ. 12: 83-113. Lin, W.-F. & T.-S. Liv. 1965. Studies on the classi- fication of Fagaceae in Taiwan. Bull. Taiwan Forest 1-59, Liu, T.-S. & J.-C. Liao. 1976. Fa agaceae, Yo) H.-L. Li et al. (editors), Flora of Taiwan 2: 49 Onuwi, J. 1965. Flora of Japan. Smithsonian Institution, Washington. [English translation.] OKAMOTO, M. consideration on the cupules of ; 93: 321- 327. pid var. sieboldii. Acta Phytotax. Geobot. 34: PAIJMANS, K. 1976. New Guinea Vegetation. Elsevier North-Holland, New York. SOEPADMO, E. 1968. ue Malesianae i-em XLVII. Census of Malesia ). Reinwardtia 7: 383-410. 1970. Florae Malesianae praecursores XLIX. Malesian ue of Lithocarpus Bl. (Fagaceae). Reinwardtia 8: 197-308. fd Fagaceae. Flora Malesiana I. 7(2): The Forests of China. Maria Moors 5. Harvard Univ., Cam- 265-403. Wane, C.-W. 1961. Cabot Found. Publ. No bridge, Massachusetts. WHITMORE, T. C. 1975. Tropical Rain Forests of the Far East. Clarendon Press, Oxfor ZAVADA, M. S. & D. L. DILCHER. Comparative pollen morphology and its relationship to Mi of pollen in the Hamamelidae. Ann. Missouri Bot Gard. 73: 348-381. MALVACEAE OF JAMMU AND KASHMIR STATE, INDIA! A. R. Naqshi, G. H. Dar, G. N. Javeid, and P. Kachroo? ABSTRACT e Malvaceae of the Jammu and Kashmir State are reviewed with a complete synopsis of the taxa recorded e m ne species xa ° micranthus is a new reco Jammu and Kashmir State. All the taxa are keyed, and the species are provided ity. with descriptions and usually followed by brief notes on distribution and economic utili In Hooker's Flora of British India (1874), Mas- ters recognized 108 malvaceous species in 27 gen- era and sorted these into four tribes: Malveae, Ureneae, Hibisceae, and Bombaceae. As the tribe Bombaceae is now referred to family Bombacaceae, 97 species in 19 genera remain from Masters's listing. Of these, ten species in seven genera were cited from Jammu and Kashmir State. Many ad- ditional species have been described from this area since Hooker's publication, and the nomenclature of most of the species included there has changed. Lambert (1933), in his list of trees and shrubs for the Jammu and Kashmir forest circles, listed the then-known arborescent taxa of the family. Stewart (1972) catalogued 70 species in 16 genera in West Pakistan and Kashmir. Of these, 30 species in 12 genera, including those based on literature, were listed from the Jammu and Kashmir State. Following Stewart, Abedin (1979) monographed the Malvaceae of West Pakistan, listing 94 specific and infraspecific taxa in 19 genera. However, Jam- mu and Kashmir materials are poorly represented in this work. Many of the collections referred by Stewart to our state are not mentioned by Abedin. Lately, many local workers have included the family in floristic works (Javeid, 1970; Javeid & Naqshi, 1973; Singh & Kachroo, 1976; Sharma & Kachroo, 1981; Naqshi & Kachroo, 1982; Dar et al., 1983; Dhar & Kachroo, 1983), but none has given a descriptive account of all the taxa. Therefore, a complete synopsis of the family as it occurs in the Jammu and Kashmir State is given here. The state is situated on the northern fringe of India between 32?10' and 37?10'N latitudes and 72°30' and 80%30'E longitudes. The eastern, northern, and western boundaries of the state com- prise a segment of the border of India. To the east of the state lies Tibet, to the north lies China (with a very small portion of the border touching Af- ghanistan), to the west is Pakistan, and to its south is Himachal Pradesh and a very small part of the Punjab. It covers an area of about 222,000 km”, which, except for a short belt in Jammu and the valley of Kashmir, is wholly mountainous, from ca. 270 m in Jammu and extending to the heights of the Himalaya in Kashmir and Ladakh (up to ca. 8,128 m at Nanga Parbat). The rock formations in the entire state belong to three broad groups: the Panjal, the Zanskar, and the Tertiary. The Panjal includes the outer hills, outer plains, and the middle mountains; the Zanskar includes the whole of the eastern region from Spiti to Lahul and to the lofty Karakoram in the north; and the Ter- tiary includes the valley of Kashmir and other river valleys (Wadia, 1953). ' Dr. Paul A. Fryxell, Research Botanist, Agronomy Field Laboratory, Texas A&M University, U.S.A., has e vey Aute Thanks included in this ntre of Plant Dune University of Kashmir, etd f India ANN. MISSOURI Bor. GARD. 75: 1499-1524. 1988. 1500 Annals of the Missouri Botanical Garden Tm x y ë m 7° E : ! SCALE 1:2,000,000 20100 Qo 120 I mm = 2 km 240 km A — » ico o e Pa Che er 5 Khardungla "" T» K Ho. P _he ` CHINA * Ç [TIBETJ + i d AP UPSHU 4 —]3 P u H I MA CHA h. A `. end AD E SH ge wa tor SPITI RO | _ A4 | | | " ` 74 75 76 78° 79° FIGURE 1. Map of Jammu and Kashmir State. Geographically the state can be divided into three distinct regions, Jammu, Kashmir, and La- dakh (Fig. 1), embracing considerable variation in topography, physiography, and climate. Jammu has mostly a subtropical climate (moist temperate in higher reaches of Chenab Valley), with the south- west monsoons resulting in an average annual rain- fall of over 1,100 mm. The mean maximum tem- perature during summer is as high as 40°C, and the mean minimum during winter as low as 6?C. Floristically the region is largely dominated by broad-leaved, deciduous and evergreen woody ele- ments. Kashmir (separated from the Jammu region by the lofty Pir Panjal range, which also acts as a barrier to the southwest monsoons) is predomi- nantly dry temperate, with an average annual rain- fall of ca. 660 mm. The maximum temperature in summer reaches 35?C, and the minimum in winter (usually with heavy snowfalls) decreases to as low as —10?C. The fewer woody genera in Kashmir are evergreen— broad-leaved arborescent species are usually lacking. Ladakh, an extremely barren land with high elevations (above 3,000 m), has a cold, arid climate, approaching arctic cold in win- ter. The average annual rainfall varies between 80 mm (Leh) and 650 mm (Drass). d gig ied in summer goes as high as 30°C a a — 40°C in winter. The region latur a desert flora largely dominated by xerophytic elements and, ex- cept for more humid valleys, almost lacks natural tree cover. MATERIALS AND METHODS Almost all the collections cited in this work were critically examined in the herbarium of the Uni- versity of Kashmir (KASH) by the authors. The herbarium studies were supplemented by extensive observations in the field. Representative specimens of all our determinations were kindly seen by Dr. Paul A. Fryxell, U.S.A. However, a few collections, mostly made in the Pakistan-occupied part of Jam- mu and Kashmir State and deposited in various herbaria of Pakistan, were not accessible, and they Volume 75, Number 4 Naqshi et al. 1501 1988 Malvaceae of Jammu & Kashmir State TABLE l. Distribution of the malvaceous species recorded from Jammu and Kashmir State. . Number of Species No. of Species stal Recorded. “esa Common Between Number of Jammu Kashmir Jammu Jammu, pecies Kash Kashmir & Genus Recorded Jammu mir Ladakh Kashmir Ladakh Ladakh Ladakh Fioria ] (—)' 1 — — — — — — Hibiscus 6 (2) 9 2 — 1 — — — Abelmoschus 3 (1) 3 1 — 1 — —- — dia 1 (—) 1 — — — — -— — Gossypium 2 (2) — 2 — = — — — Thespesia 1 (—) 1 — — — — — — Malva 9 (3) 4 8 3 6 3 3 Althaea 2(—) = 2 — — — — — Alcea 3 (3) l 3 — 1 — — — Lavatera 1 (—) 1 1 — 1 — — — Abutilon 4 (—) 3 1 — — — — — Sida 5(—) 5 — — — — — Malvastrum 1 (—) 1 — — — — — — Urena 1 (—) 1 — — — — — Sidalcea 1 (—) =- 1 — — — — — Grand total 41 (12) 27 21 7 6 3 3 ' Numbers within parentheses indicate the number of cultivated species and/or escapes from cultivation. have been included with almost full citation of their collectors and places of collection and deposition. RESULTS The Malvaceae have a moderate representation in the Jammu and Kashmir State, but only a few species are indigenous. Altogether, 41 species in 15 genera are recorded in this study. Out of these, 12 species are cultivated, with at least half of them escaped from cultivation. Among the escaped cul- tivated species, Sidalcea neomexicana, Alcea la- vateriflora, A. pallida, and Malva verticillata var. rafiqii are reported for the first time from India. Turning to wild species, Malva bucharica and Althaea broussonetiifolia are first recorded for the Indian subcontinent; Malva microcarpa, M. ambigua, and M. mohileviensis are new rec- ords for India; and Hibiscus micranthus is a first record for the Jammu and Kashmir State. Twenty-seven of the 41 species occur in Jammu Province, 21 in Kashmir, and six in Ladakh. Twelve of the 15 genera occur in Jammu, nine in Kashmir, and only one (Malva) in Ladakh. It is evident that there is a marked decline in the number of genera and species as one proceeds from the sub-Hima- layan Jammu through the Himalayan Kashmir to the trans-Himalayan Ladakh (Table 1). The num- ber of species in common is highest between Jammu and Kashmir, while it is the lowest between Jammu and Ladakh. Almost all our arborescent species are restricted to Jammu, although Hibiscus syriacus is commonly grown ornamentally in Kashmir. None of our taxa except Lavatera kashmiriana and a few species of Malva reach subalpine and alpine ranges. None of the taxa recorded here is endemic to the Jammu and Kashmir State. All our genera except Sidalcea (a North American plant, collected as an escape from cultivation only once in Kashmir) are shared with Pakistan. As for our other neigh- oring countries, Fioria does not occur in Af- ghanistan, U.S.S.R., China, and Iran; Kydia and Thespesia are absent from Afghanistan, U.S.S.R., and Iran; Malva and Lavatera are not found in a istan, U.S.S.R (38) are also in Pakistan, followed by 22 in Iran, 19 each in the U.S.S.R. and China, 11 in Burma, and 10 in Afghanistan. Both our new records for the Indian subcontinent (Malva bucharica and Althaea broussonetiifolia) grow in the U.S.S.R. MALVACEAE A. L. DE JUSSIEU, GEN. PL. 271. 1789. Plants annual, biennial, or perennial herbs, to small trees, mucilaginous, usually pubescent with stellate, furcate, and simple hairs, rarely with fer- 1502 Annals of the Missouri Botanical Garden ruginous peltate scales. Leaves alternate, simple, stipulate, petiolate, unlobed to deeply lobed. Flow- ers axillary, solitary, or fasciculate, or subracemose to paniculate, usually bracteate with an epicalyx of 3-13 free or basally connate segments, some- times ebracteate, actinomorphic, usually perfect, sometimes polygamodioecious. Calyx usually cam- panulate or tubular, 5-lobed or -toothed, rarely spathaceous and 2- or 3-lobed, valvate, usually persistent. Corolla polypetalous, 5-merous, adnate basally to the staminal tube, contorted. Stamens numerous, monadelphous with filaments coherent to form a staminal tube, this wholly or partially antheriferous; anthers dorsifixed, monothecous, lin- ear to horseshoe-shaped, solitary, rarely in clusters of 3-5. Carpels (3—)5—many, syncarpous in a sin- gle whorl (in ours) around the columella; ovary superior, with as many locules as carpels; placen- tation axile; style usually divided at the apex into as many (or twice as many) branches as carpels, or sometimes unbranched; stigmas sessile, linear or capitate or discoid. Fruit a dry, loculicidal (or indehiscent) capsule or a schizocarp separating into usually 1-seeded mericarps, rarely follicular and 2—3-seeded. Seeds reniform, ovoid or obovoid, gla- brous or pubescent with short and long hairs. Considerable embryological work has been done on this family (Schnarf, 1931; Venkata Rao, 1954, 955; Winter, 1960; Ramchandani et al., 1966), and optimistic views are being held about the taxo- nomic significance of such studies. The structure and development of the seed and seed coat anatomy have also been shown to be of great taxonomic and phylogenetic value (Reeves, 1936; Wunderlich, 1967; Bouman, 1971; Mohana Rao, 1978; Ku- mar, 1981) A family of 88 genera and ca. 2,300 species, most abundant in the tropics, common in subtrop ical and temperate regions, and usually absent from arctic regions. The Ureneae have style branches and stigmas twice as many as the carpels. The Malopeae have carpels irregularly arranged in two or more whorls around the carpophore and the style branches and stigmas are as many as carpels; the fruit is schizocarpic. In the Hibisceae the car- pels are regularly arranged in a single whorl around the carpophore, the style branches and stigmas are as many as carpels or the style is unbranched, and the fruit is capsular. The Malopeae do not occur in our area. The Malveae are represented here by Malva, Lavatera, Alcea, Althaea, Abutilon, Sida, Malvastrum, and Sidalcea; the Hibisceae by Fio- ria, Hibiscus, Abelmoschus, Kydia, Gossypium, and Thespesia; and the Ureneae by Urena KEY TO THE GENERA OF MALVACEAE IN JAMMU AND KASHMIR STATE la. Flowers polygamous or polygamodioecious. Plants trees; flowers white, polygamous; epicalyx present; anthers in globose head of 3-5; carpels not 4. K dia beaked 2b. Plants herbs; flowers rose-purple, polygamodioecious; epicalyx absent; anthers solitary; iu m . Sidalcea v Flowers bisexual 3a. Style branches and stigmas twice the number of carpels, always 10; mericarps glochidiate-spiny ......... 14. Urena 3b. Style branches and stigmas as many as carpels or style unbranched; mericarps never glochidiate-spiny. 4 ruit a capsule, the xL at maturity not separating from one another. Style unbranched, + 6a. Plants herbaceous to suffrutesce d; epicalyx segm soa loculicidally dehisc clavate, or superficially alse into very short branches nt, not co ments um oa Sct eee calyx with oil glands; capsule with ferr ruginous peltate scales; leaves Gossypium 6b. Plants small trees, "hei id ous po + woody, indehiscen ortions covered with ferruginous peltate scales; leaves lobed epicalyx segments lanceolate, not foliaceous, caducous; calyx without oil glands; t 6. espesia a v ule Sue divided into 5 divergent bran Calyx ches spathaceous, irregularly 2- or 3-lobed, falling together with corolla and staminal elmoschus tube 7b. Calyx brea campanulate, regularly 5-lobed, persistent, not falling as a unit with corolla nal tube. and stam 8a. Capsule with 5 conspicuous, scarious, and strongly veined wings ooo... 1 . Fioria Hibiscus Capsule not winge 4b. Fruit a M ocarp, the carpels at maturity (mericarps) separating from one another bun 4 ella distinct central colum 9a. Epicalyx prese 10a. Eos E 3; staminal tube antheriferous only in the apical par 1 eaves ovate to lanceolate-oblong, unlobed; flowers yellow; stigmas capitate; mericarps tricuspidate (in ours) olaa Volume 75, Number 4 1988 Naqshi et al. Malvaceae of Jammu & Kashmir State llb. Leaves orbicular-reniform or cordate, mostly lobed or angled; flowers pink-lilac (or white); stigmas linear, decurrent; mericarps awnless 2a. Stipules foliaceous; epicalyx segments ovate- orbicular, foliaceous, connate ce 10b. Epicalyx segments 6-12; ma tube antheriferous almost to the ES Epicalyx segments 7-12; c den cylindrical, the BR paetos purple; mericarps 12-25, pesi es wing- 8. rolla 2-3 cm in diameter, 0.8-2.2 m long; staminal Althaea 13b. dirum segments 6-7; corolla 5-8 cm in diameter, 3.5-7 cm lo tube 5-angled, the anthers yellowish; mericarps 20-40, sub-bilocular, often winged 9. ng; staminal Alcea 9b. Epicalyx absent. 1 Calyx cupular; mericarps 10-20, follicular, 2-3-seeded 14b. Calyx campanulate; mericarps 5-10, not follicular, 1-seeded 1. FIORIA Mattei, Boll. Reale Orto Bot. Giar- dino Color. Palermo 2: 71. 1916 Four species, distributed in tropics and subtrop- ics of the Old World; represented in our area by a single species. Fioria vitifolia (L.) Mattei, Boll. Reale Orto Bot. iardino Color. Palermo 2: 1916. Hibis- cus vitifolius L., Sp. Pl. 696. 1753; Masters in Hook. f., Fl. Brit. India 1: 333. 1874; Rakshit & Kundu, Bull. Bot. Surv. India 12: 166. 1970; Ngwe, Union Burma J. Life Sci. 4: 204. 1971; Stewart in Nasir & Ali, Ann. Cat. Vasc. Pl. W. Pak. & Kashm. 480. 1972; Sharma & Kachroo, Fl. Jammu 1: 112. 1981. H. vitifolius L. var. genuinus Hochr., in An- nuaire Conserv. Jard. Bot. Genéve 4: 169. 1900. TYPE: Herb. Hermann, Vol. IV, Fol. 39, Linn. no. 265 (lectotype, BM); see Brenan & Exell, Bol. Soc. Brot. Ser. 2. 32: 72. 1958. e pla obtusifolius Willd., Sp. Pl. 3: 829. 1801. ncatus Roxb., Hort. vues 51. 1814; Fl. Ind. ig 1832. 3: 200. 1832 H. cuspidatus Edgew., J. Asiat. Soc. Bengal 21: 168. Fioria vitifolia ( (L.) Mattei subsp. vulgaris (Brenan & Exell) e Pakistan J. Bot. 9: 59-66. 1977; Fl. W. Pak. 1 1979. Hibiscus vitifolius L. subsp. a AR & Exell, Bol. Soc. Brot. ser. 2. 32: 58. H. hererorrichis DC., Prodr. 1: 450. 1824. H. d x. Aetersarichüs (DC.) Hochr., Annuaire serv. Jard. Bot. Genève 4: 170. 1900 Annual, suffrutescent herbs. Stems and inflo- rescence axes usually densely tomentose with stel- late and glandular hairs. Leaves ovate-cordate, en- tire or shallowly 3-5-lobed, coarsely serrate, stellately tomentose beneath or on both surfaces. Flowers axillary, solitary or clustered at the ends of branches, usually drooping; pedicels shorter than 11. Abutilon 12. Sida petioles, articulate at or below the middle. Epicalyx of 7-12 linear segments. Calyx campanulate, 5-lobed; lobes ovate, acute, 3—5-nerved, simple and 2-rayed pubescent within, stellate-tomentose out- side. Corolla twisted, 3-5 cm diam., yellow with a purple center; petals obovate, glabrescent. Stam- inal tube truncate, shorter than corolla, antherifer- ous throughout. Capsule suborbicular, apiculate, 5-winged, hirsute. Seeds 2-4 in each cell, reniform, minutely tubercled. Distribution. India, Pakistan, Burma, Sri Lanka, Australia, and tropical Africa. In our area it occurs in Jammu Province only. Sharma & Kachroo (1981: 113) wrongly reported it as a new record for Jammu and Kashmir. Additional specimens examined. INDIA. JAMMU: Mandal, common along banks of irrigation channels, B. M. Sharma 765 (KASH); Poonch Dist. Nawal Nadi, A. Rashid s.n. pra Rajouri, Jacquemont 1428 (fide Stewart, 1972: 480). Abedin (1979) followed pem & Exell (Bol. Soc. Brot. Ser. 2. 32: 73. 1958) in placing the specimens of Fioria ost from Pakistan and Kashmir under subsp. vulgaris, which is said to differ from subsp. vitifolia in the density of in- dumentum and leaf incision. However, the depth of leaf incision, the shape of the leaf lobes, and the density and rigidity of hairs have been found to be highly variable in the species, even within a single plant (Rakshit & Kindu, 1970). Subdivision of the species on the basis of the above-mentioned char- acters does not, therefore, seem to be satisfactory. Most previous authors included Fioria in Hi- biscus. This seems almost justified, especially when we consider their only superficial differentiating character of winged (Fioria) and nonwinged (Hi- biscus) capsule. However, for a better understand- ing of these plants, it is now believed to recognize the two as distinct genera. 1504 Annals of the Missouri Botanical Garden 2. HIBISCUS L., Sp. Pl. 693. 1753; Gen. Pl. 5th Edition. 310. 1754 Ketmia Miller, Gard. Dict. Abr. 4th Edition. 28. 1754. Pariti Adans., Fam. Pl. 2: y Paritium re ung in St. Hilaire, Fl. Bras. Mered. 1(2): 255. A genus of some 3,000 species distributed chief- ly in the tropical and subtropical regions of both hemispheres. Six species occur in our area, of which two are cultivated. KEY TO THE SPECIES OF HIBISCUS IN JAMMU AND KASHMIR STATE la. Flowers drooping; petals deeply laciniate; staminal tube much longer than corolla, exserted „u de H. s lb. Flowers not drooping; petals usually entire; staminal tube shorter than or equaling the corolla, ar d. schizopetalus 2a. Epicalyx absent or rarely represented by minute teet . H. lobatus 2b. Epicalyx present, represented by conspicuous segm Ja. Epicalyx segments radiate; staminal tube Vica e in the upper = only o 6. H. caesius 3b. EE segments not radiate; staminal tube ERR age i an 4a. Annual herbs; calyx inflated, more so in frui 4b. Shrubs; calyx not inflated; seeds villous o . Plants usually glabrous; diis eliptic- -rhombic, often 3-lobed; pedicels equal to or ni 6 cm in diameter; seeds with a line of long white hair e w than petioles; flowers 4- seeds glabrou 1. H. trionum r with a line of a: white hairs. H syriacus [21 S Plants scabrous-bristly; leaves + 1-1 ill m in diameter; seeds villous 1. Hibiscus trionum L., Sp. Pl. 697. 1753; Masters in Hook. f., Fl. Brit. India 1: 334. 1874; Ijin in Shishkin & Bobrov, Fl. U.R.S.S. 15: 159. 1949; Hu, Fl. China, Malvaceae. 57. 1955; Rakshit & Kundu, Bull. Bot. Surv. India. 168. 1970; Stewart in Nasir & Ali, Ann. Cat. Vasc. Pl. W. Pak. & Kashm. 480. 1972; Riedl in K. H. Rechinger, Fl. Iran. 120: 30. 1976; Singh & Kachroo, Forest FI. Srinagar. 151. 1976; Abedin in Nasir & Ali, Fl. W. Pak. 1300. 11. 1979. TYPE: Linn. Herb. no. 875/39, photo (LINN). HE e i d eee Miller, Gard. Dict. 8th n. HIB. Hun sicar Lu Te ed 64, f. 2. 1787. H. dissectus s Wallich, Cat. no. 3696. 1831, nom. nud. Annual hispid herbs. Stems with simple and tu- berculate-stellate hairs. Leaves orbicular-ovate, the ower leaves usually undivided, the upper leaves palmately 3—5-lobed, the lobes obovate or oblong, pinnatisect, punctate, nearly glabrous or sparsely stellate-pubescent, especially abaxially. Flowers solitary, axillary; pedicels longer than petioles, ar- ticulate above the middle. Epicalyx of 8-13 linear, long-hispid segments. Calyx campanulate, 5-lobed, inflated in fruit; lobes deltoid, acute, membranous; with many hispid, green-purplish, raised veins. Co- rola 1.5-3 cm diam., pale yellow with a dark purple center; lobes glabrous. Staminal tube shorter than corolla, purplish, antheriferous throughout. Capsule oblong, obtuse, strigose-hispid, black, en- closed in the inflated persistent calyx. Seeds + ovate, unlobed; pedicels longer than petioles; flowers 4. H. micranthus reniform, tuberculate-rugose when mature, gla- brous Distribution. India, Pakistan, Afghanistan, southern U.S.S.R., Burma, China, Iraq, Iran, Tur- comania, Transcaucasia, Mediterranean Region, southern Europe to southern Africa, Australia; nat- uralized in America. In our area it is common in Kashmir Province, often growing as a weed es- caped from cultivation. Stewart (1972) gives Poonch, Jammu. Specimens examined. , A. R. Naqshi 195 (KASH); Gulmarg, 4. R INDIA. KASHMIR: university mpus . Naqshi 514 (KASH); Kokernag, I. M. Nahvi s.n. (KASH); KASH); Ganderbal, G. H. Dar 3004 (KASH); Shuhama (Ganderbal), G. H. Dar 2943-44 (KASH), 2942 (PF); Sarich aio G. H. Dar 1767-68 (KASH); Narbal, A. R. qshi & G. N. Dar 8173-75 (KASH). This is a species of the eastern Mediterranean Region, now widespread in almost all the conti- nents. Despite the extensive distribution, the species retains its essential morphological characters throughout. However, a number of species related to it have been described. Hochreutiner (1900) retained H. trionum s. str. and reduced to varieties some species described by earlier workers. We follow this conservative approach. A diaphoretic syrup is prepared from its leaves, which contain 0.3% rubber substances. The seeds contain 23.8% oil. In South Africa the plant is said to be used for treatment of round worm, while in Volume 75, Number 4 1988 Naqshi et 1505 spe lei Bi Jammu & Kashmir State China and Malaya the dried leaves are considered stomachic. Àn infusion of the flowers is used for itch, for painful skin diseases, and as a diuretic. It is reported to be poisonous to stock, particularly orses. 2. Hibiscus schizopetalus (Masters) Hook. f., Bot. Mag. 106, tab. 6524. 1880; Hu, FI. China, Malvaceae. 46. 1955; Rakshit & Kun- du, Bull. Bot. Surv. India 12: 166. 1970; Ngwe, Union Burma J. Life Sci. 4: 205. 1971; Stewart in Nasir & Ali, Ann. Cat. Vasc. Pl. W. Pak. & Kashm. 479. 1972; Abedin in Nasir & Ali, Fl. W. Pak. 130: 12. 1979. Hibiscus rosa-sinensis L. var. schizopetalus Masters, Gard. Chron. n.s. 12: 272, f. 45. 1879. TYPE: Gard. Chron. n.s. 12: 272, f. 45. 1879 Glabrous shrubs with spreading-drooping branches. Leaves elliptic, glabrous, shining, pal- mately nerved, entire in basal half, serrate in apical half. Flowers solitary, axillary, pendulous; pedicels slender, longer than petioles, articulate at the mid- dle. Epicalyx of 5-8 subulate, ciliate segments 1— 2 mm long. Calyx spathaceous, tubular, irregularly 2—-5-lobed. Corolla 4-9 cm diam., pinkish; petals deeply laciniate and recurved. Staminal tube much longer than corolla, filiform, pendulous, red, an- theriferous in the upper half only. Capsule oblong- cylindrical. Seeds smooth, glabrous. Distribution. Native of Kenya and Tangan- yika (Exell, Fl. Zambes. 1: 470. 1960), cultivated elsewhere. It is commonly cultivated in gardens throughout India, Pakistan, Burma, and a few coastal cities in southern China. In our area it is rarely grown, collected only at Banihal in Jammu Province. cimen examined. INDIA. JAMMU: Banihal, Sira- Spe jud-din s.n. (KASH It is reported that the flowers in H. schizopet- alus drop after anthesis and that fruits seldom form. According to Wilcox & Holf (Hawaii Agr. Exp. Sta. Bull. 29. 1913) it has been used as "male" parent in crosses with H. rosa-sinensis L. and its varieties. In 1984 Fryxell labeled the spec- imen cited above as "apparently a hybrid or a hybrid-derivative of Hibiscus rosa-sinensis L. and H. schizopetalus (Masters) Hooker.” 3. Hibiscus syriacus L., Sp. Pl. 695. 1753; Masters in Hook. f., Fl. Brit. India 1: 344. 1874; Iljin in Shishkin & Bobrov, Fl. U.R.S.S. 15: 152.1949; Hu, Fl. China, Malvaceae. 50. 1955; Kitamura, Fl. Afghan. 270. 1960; Rakshit & Kundu, Bull. Bot. Surv. India 12: 170. 1970; Stewart in Nasir & Ali, Ann. Cat. Vasc. Pl. W. Pak. & Kashm. 480. 1972; Ried] in K. H. Rechinger, Fl. Iran. 120: 29. 1976; Abedin in Nasir & Ali, Fl. W. Pak. 130: 13. 1979. TYPE: Syria: Linn. Herb. no. 875/24 (LINN). Ketmia v: ie Scop., Fl. Carniol. 2nd Edition. 2: 45. 1772 K. syrorum Medikus, Malvaceae: 45. 1787. K. arborea Moench, Suppl. Meth. 617. 1794. Hibiscus floridus Salisb., Prodr. 383. re H. acerifolius Salisb., Parad. Londin. 1: tab. 33. 1805. H. syriacus L. var. sinensis Lemaire, Jard. Fleur. 4: tab. 370. 1854. H. chinensis sensu Forbes & Hemsley, J. Linn. Soc., Bot. 88. 1886. Glabrous, branched shrubs. Leaves elliptic- rhombic, acute at the apex, cuneate at the base, irregularly dentate, often 3-lobed. ciis solitary, axillary, single or double; pedicels equaling or shorter than petioles. Epicalyx of 6-8 linear, single-nerved segments. Calyx campanulate, densely stellate-to- mentose, shallowly 5-lobed; lobes triangular-lan- ceolate, acute. Corolla campanulate, 4-6 cm diam., usually lilac with a purple center; petals obovate, ciliate and stellately villose outside. Staminal tube shorter than corolla, antheriferous to the base. Capsule oblong-ellipsoid, yellowish, stellate-tomen- tose, shortly beaked at the apex. Seeds reniform, glabrous except for a line of long white hairs. Distribution. China, Grown in gardens throughout India, Pakistan, Af- ghanistan, China, Iran, and other countries. In our cultivated elsewhere. area this species is extensively grown as an orna- mental shrub or as a hedge plant in the Kashmir Valley. Additional specimens examined. INDIA. KASHMIR Beehama (Ganderbal), G. H. Dar 2866 (KASH), 2866a de Srinagar University campus, 4. R. Naqshi 8176 KASH); Nehru Bot. Garden (Cheshma Shahi), 4. R. Naqshi s.n. (KASH). = A number of single and double horticultural va- rieties of this species have been described mainly on the basis of flower color, which ranges from purple-pink through pink to pure white. Hu (1955) recognized nine varieties mainly on the basis of floral size and color Although the plant is mainly used ornamentally, its stem is said to yield a strong fiber. The seeds contain 24.6% oil. In China, the flowers are re- 1506 Annals of the Missouri Botanical Garden portedly eaten, and the tender leaves are used as a substitute for tea and as a shampoo. 4. Hibiscus micranthus L. f., Suppl. Pl. 308. 1781; Masters in Hook. f., Fl. Brit. India 1: 335. 1874; Rakshit & Kundu, Bull. Bot. Surv. India 12: 171. 1970; Stewart in Nasir & Ali, Ann. Cat. Vasc. Pl. W. Pak. & Kashm. 479. 1972. Hibiscus micranthus L. f. var. gen- uinus Hochr., Annuaire Conserv. Jard. Bot. Genéve 4: 83. 1900. TYPE: Linn. Herb. no. 875/2 (holotype, LINN). H. gossypinus DC., Prodr. 1: 453. 1824, non Thunberg, 1800. Erect shrubs with slender, terete branches, sca- brid with scattered stellate bristles on almost all parts. Leaves + ovate, 1.9-4. —4 cm, acute or obtuse, serrate, eglandular. Flowers axillary, solitary; pedicels slender, longer than petioles, ar- ticulate above or below the middle. Epicalyx of 6- 8 filiform, stiff, pubescent segments. Calyx 5-lobed; lobes triangular-lanceolate, pubescent. Corolla 1- .9 cm diam., white or pink; petals often reflexed, stellately pubescent outside. Staminal tube to 5 mm long, shorter than corolla, antheriferous throughout. Capsule globose. Seeds reniform, black, villous. Distribution. India, Sri Lanka, Pakistan, tropical Africa, South Africa, Madagascar, Arabia. In our area confined to Jammu Province. Additional specimens examined. INDIA. JAMMU: Ram Nagar forest, common among hedges and on open road- side slopes; erect pubescent shrubs up to six feet tall with rosy-red flowers, B. M. Sharma 249 (KASH Three varieties of H. micranthus are recognized from India and Pakistan (Abedin, 1979), but only var. micranthus occurs in our area. The plant is reportedly valued as a febrifuge in Sri Lanka. This is the first record for Jammu and Kashmir State. 5. Hibiscus lobatus (Murray) Kuntze, Revis. Gen. Pl. 3rd Edition. 2: 19. 1898; Rakshit & Kundu, Bull. Bot. Surv. India 12: 169. 1970; Stewart in Nasir & Ali, Ann. Cat. Vasc. Pl. W. Pak. & Kashm. 479. 1972; Abedin in Nasir & Ali, Fl. W. Pak. 130: 18. 1979; Sharma & Kachroo, Fl. Jammu 1: 112. 1981. Solandra lobata Murray, Comment. Soc. Re- giae. Sci. Gott. 6: 20, tab. 1. 1785. TYPE: Comment. Soc. Regiae. Sci. Gott. 6: 20, tab. 1. n solandra L'Hér., Stirp. Nov. 1: 103, tab. 49. 1789, nom. illeg.; Masters in Hook. f., Fl. Brit. India 1: 336. 1874. H. solandra var. genuinus Hochr., Annuaire Conserv. Jard. Bot. Genéve 4: 128. 1900. H. pumilis Roxb., Fl. Ind. 3: 203. 1832. Lagunea lobata Willd., Sp. Pl. 4th Edition. 3: 733. 1800. L. sinuata Hornem., Hort. Bot. Hafn. 2: 645. 1851. Annual, erect, pubescent or somewhat hispid herbs. Leaves polymorphic, the lower leaves or- bicular-ovate, the upper leaves deeply 3-lobed, the uppermost 3-fid, all cordate, crenate or coarsely serrate, pubescent with simple and stellate hairs on both surfaces. Flowers solitary and axillary, or in terminal racemes; pedicels equal to or longer than petioles, articulate near the apex. Epicalyx segments absent or rarely represented by minute teeth. Calyx 5-lobed, pubescent without; lobes lan- ceolate, prominently 3-nerved. Corolla 1-2 cm diam.; white to pale yellow; petals obliquely ob- cordate. Staminal tube shorter than or equaling corolla, pink, antheriferous throughout. Capsule ovoid, beaked, + wrinkled and pubescent. Seeds + reniform, black, usually granulated, minutely pu- bescent. Distribution. India, Pakistan, Sri Lanka, Lac- cadive Islands, tropical Africa, Madagascar, and Java. In our area confined to Jammu Province. Specimen examined. INDIA. JAMMU: Ram Nagar common under shade on roadside slopes, often associated with Triumfetta rhomboidea Jacquem., Bidens bipinnata L., and other annuals; erect herbs up to 2 feet tall, flowers white with sticky calyx, B. M. Sharma 204 (KASH). 6. Hibiscus caesius Garcke, Bot. Zeit. 7: 850. 1849; Rakshit & Kundu, Bull. Bot. Surv. India 12: 173. 1970; Stewart in Nasir & Ali, Ann. Cat. Vasc. Pl. W. Pak. & Kashm. 478. 1972; Abedin in Nasir & Ali, Fl. W. Pak. 130: 21. 1979. H. caesius Garcke var. gen- uinus Hochr., Annuaire Conserv. Jard. Bot. Geneve 4: 160. 1900 Hibiscus Ps an F. Muell., Fragm. 2: 13. 1859, xb., H. gibsonii cm ex Harvey, Fl. Cap. 2: 587. 1859- 1860; Masters in Hook. f., Fl. Brit. India 1: 339. H. heptaphyllus Dalz. & A. Gibson, Bombay Fl. 20. 1861. Erect, usually suffrutescent herbs; branches bristly or with minute bristle-pointed prickles. Leaves palmately 3-5-lobed, the lobes oblong-elliptic, sharply serrate, glabrous or stellately pubescent. Flowers solitary, axillary; pedicels longer than leaves, articulate near the apex. Epicalyx of usually 10 radiate, needlelike, subspiny segments. Calyx 5-lobed, the lobes lanceolate, acuminate, strongly Volume 75, Number 4 1988 Naqshi et al. 1507 Malvaceae of Jammu & Kashmir State 3-nerved, distantly ciliate. Corolla 3-5 cm diam., yellow with purple center or completely purple, rarely white with purple center; petals obovate, sparsely stellate-pubescent outside. Staminal tube shorter than corolla, purple, antheriferous in upper half only. Capsule ovoid, beaked; valves setose. Seeds dark brown, pilose Distribution. India, Pakistan, Afghanistan, south tropical Africa and north Australia. Rare in our area, reported from the Pakistan-occupied part of Poonch District in Jammu Province. Specimen examined. INDIA. JAMMU: Poonch Mirpur, in bushes, Stewart 27244 (RAW). One more species, Hibiscus hirtus L. is reported from Kashmir by Stokoe (fide Stewart, 1972) and probably is based on misidentification and remains yet to be seen. The specimen B. M. Sharma 249 (KASH) under this name turned out to be H. mi- cranthus. 3. ABELMOSCHUS Medikus, Malvaceae 45. 1787; Schumann, Nat. Pflanzenfam. III. 6: 49. 1890; Hochr., Candollea 2: 83. 1924; Taxon 4: 188. 1955 rune Cav., Diss. 3: 173. 1787. Bamia R. Br. ex dE I. Asiat. Rar. 1: 39. 1830. Six species (Borssum-Waalkes, 1966), distrib- uted in temperate and warm regions. Three species have been recorded form our state, one cultivated. KEY TO THE SPECIES OF ABELMOSCHUS IN JAMMU AND KASHMIR STATE la. pda segments caducous before anthesis; rolla white — pinkish at maturity), with a ‘dul purple center c iculneus Epicalyx segments pan sas until dehiscence of fruit; — yellow to yellowish white, with a purple c 2a. de E 7-12, linear to narrow- ly lanceolate, 1-2.5 mm wide; corolla 5- 7 cm diam.; capsule pend 7-25 cm long; seeds labra ous s. A. esculentus Epicalyx segments 4-5, ovate- Viris 5-10 mm wide; corolla 7-10 cm diam.; capsule ovate- Ser M 3.5-6 cm long; seeds glabrescen 3. A. pungens = t > - . Abelmoschus esculentus (L.) Moench, Methodus 617. 1794; Hu, Fl. China, Mal- vaceae. 39. 1955; Stewart in Nasir & Ali, India 1: 343. 1874; Iljin in Shishkin & Bob- rov, Fl. U.R.S.S. 15: 165. 1949; Ngwe, Union Burma J. Life Sci. 4: 203. 1971; Riedl in K. H. Rechinger, Fl. Iran. 120: 32. 1976. TYPE: "Habitat in Indiis," Linn. Herb. no. 875/31 (LINN). Hibiscus ficifolius Miller, Gard. Dict. 8th Edition. 15. H. longifolius Willd., Sp. Pl. 4th Edition. 3: 827. 1800. Annual, erect herbs, strigose-hirsute through- out. Leaves aceriform, wider than long, cordate, angular or palmately 3-7-lobed; lobes ovate to lanceolate, dentate. Flowers solitary and axillary. Epicalyx segments 7-12, linear to narrowly lan- ceolate, persisting until dehiscence of fruit. Calyx 2-toothed, spathaceous, bilabiate, caducous. Co- rolla 5—7 cm diam., yellow to yellowish white with a purple center; petals obovate. Staminal tube 2— 2.5 cm long, included, antheriferous throughout. Capsule SA E 25 cm long, 5-angled, acu- e. Seeds + globose, glabrous. wre) minate stri gose- Distribution. Cultivated as vegetable in most tropical and many temperate countries; also in the Jammu and Kashmir provinces, known there as “Bhindi.” Additional specimen examined. INDIA. KASHMIR: Chatterhama (Srinagar), 1,700 m, G. H. Dar 8963-66 SH). According to Abedin (1979), the species is Asian in origin, as the whole genus is mainly of Asiatic distribution. On the basis of its close resemblance to A. tuberculatus Pal & Singh, a northern Indian species, Borssum-Waalkes (1966) considered the latter as one of the possible ancestors. This is taken to imply that 4. esculentus originated in India (Hu, 1955 and Abedin, 1979). Babu (1977), however, thought it originated in Africa. The species is primarily cultivated for the young fruits, which are eaten fresh or cooked as vege- tables. Chopped capsules are used in making mu- cilaginous soups and sauces. Roasted seeds are also edible and are used as a substitute for coffee. The seeds contain 18% oil and ca. 15% water-soluble proteins. The plant is used in medicine for its di- uretic and anticatarrhal effects. The stems furnish fiber and serve as raw material for paper produc- tion. 2. Abelmoschus ficulneus (L.) Wight & Arn. ex Wight, Cat. 14. 1833; Stewart in Nasir & Ali, Ann. Cat. Vasc. Pl. W. Pak. & Kashm. 475. 1972; Abedin in Nasir & Ali, Fl. W. Pak. 130: 26. 1979. Hibiscus ficulneus L., Sp. Pl. 695. 1753; Masters in Hook. f., Fl. 1508 Annals of th Missouri Botanical Garden Brit. India 1: 340. 1874; Ngwe, Union Burma J. Life Sci. 4: 204. 1971. TYPE: Dillenius, Hort. Elth., tab. 157, f. 190. Laguna aculeata Cav., Diss. 3: 173, tab. 71, f. 1. 1783. Hibiscus sinuatus Cav., Diss. 3: 147, tab. 52, f. 2. 1787. Abelmoschus alboruber F. Muell., Fragm. 1: 67. 1859. Annual, sometimes prickly herbs. Leaves orbicu- lar with cordate base and serrate margin, palmately 3—5-parted, scabrous on both surfaces; lobes ob- ovate to spathulate. Flowers solitary, axillary or in terminal racemes. Epicalyx segments 5-6, lanceo- late, falling before expansion of corolla. Calyx 5- toothed, spathaceous, rarely bilabiate, tomentose, caducous. Corolla 3-6 cm pinkish at maturity, with a dark purple center; petals obovate. Staminal tube ca. 1.5 cm long, included, wholly antheriferous. Capsule pyramidal- ovoid, 3-4 cm long, 5-angled, hispid. Seeds ovoid to clavate-globose, black, striated with pilose stel- late hairs. iam., white, turning Distribution. Northern Australia, southern Asia, Malaysia, eastern Africa, and Madagascar. Very rare in our area, this species has been col- lected from Jammu only (Stewart, 1972: 475). > Abelmoschus pungens (Roxb.) Voigt, Hort. Calc. 119. 1845; Abedin in Nasir & Ali, Fl. W. Pak. 130: 27. 1979. Hibiscus pungens Roxb., Hort. Bengal. 52. 1814, nom. nud., Fl. Ind. ed. 1832. 3: 213. 1832; Masters in Hook. f., Fl. Brit. India 1: 341. 1874. Hi- biscus manihot L. var. pungens (Roxb.) Hochr., Annuaire Conserv. Jard. Bot. Genéve 4: 155. 1900. Abelmoschus manihot (L.) Medicus, var. pungens (Roxb.) Hochr., Can- dollea 2: 87. 1924. A. manihot var. pungens (Roxb.) Hochr. sensu Hu, Fl. China, Malva- ceae. 36. 1955. TYPE: Roxburgh's Icone no. 1585 (K). Annual or perennial herbs, densely covered with long, yellow, bristly hairs. Leaves orbicular to broadly ovate, cordate at base, palmately 3-7- lobed or -parted; lobes variable, ovate to lanceolate, oblong-lanceolate, obovate or elliptic, entire to coarsely serrate. Flowers solitary, axillary, some- times subracemose towards the stem apex. Epicalyx segments 4—5, ovate-lanceolate, persisting until de- hiscence of fruit. Calyx 5-toothed, spathaceous, caducous. Corolla 7-10 cm diam., yellow with a purple center; petals usually obovate. Staminal tube 1.5-3 cm long, included, antheriferous to base. Capsule ovate-ellipsoid, 3.5-6 cm long, 5-angular, short-beaked. Seeds globular or reniform, black, scabrid on the back, glabrescent. Distribution. Northern India, Pakistan (rare), China, Malaysia, Philippines, and northern Aus- tralia. In our area this species has been collected only in the Pakistan-occupied part of Poonch Dis- trict in Jammu Province and is no doubt very rare. pecimen examined. INDIA. JAMMU: Poonch District, Nawal Nadi, 11-9-1953, 4. Rashid, E. Nasir & R. R. Stewart s.n. (RAW). According to Hu (1955), this species is found wild in China and northern India on grassy banks or along roadsides at altitudes of 1,500-1,600 m. She further reported that this plant (like her Abel- moschus manihot “typicus””) is extensively culti- vated in China for its flowers and roots. The flowers are said to be used in soup for the conservation of health during hot summer months. The root may be used fresh or dried. It is boiled with pork, and the preparation taken internally to cure abscesses. It is also soaked in rape seed oil and used for dressing boils. The genus Abelmoschus was established by Medicus (1787) to accommodate species of Hi- biscus with caducous calyces. The new genus was subsequently adopted by Gaertner (1791) and Moench (1794). Notwithstanding this, most work- ers on the Indian flora (Roxburgh, 1832; Masters, 1874; Prain, 1903; Duthie, 1903; Gamble, 1957; Cooke, 1958; and others) did not recognize the new genus but considered Abelmoschus as a section of Hibiscus. Hochreutiner (1924), however, stressed the need of placing Abelmoschus apart rom Hibiscus, because in the former the calyx, corolla, and stamens are adnate basally and fall together after anthesis, although earlier (1900) he considered Abelmoschus only to be a section of Hibiscus. Most recent workers on Malvaceae (Hu, 1955; Borssum-Waalkes, 1966; Abedin, 1979) have treated Abelmoschus as a distinct genus. We also take this approach and agree with Paul A. Fryxell (1984, pers. comm.) in believing that the species of Fioria, Abelmoschus, and Hibiscus are better understood when separated into distinct gen- era. 4. KYDIA Roxb., Pl. Coromandel 3: 11, tab. 215-216. 1819. Four or five species, distributed in India, Pa- kistan, Burma, and China; one species recorded from our area. Kydia calycina Roxb., Pl. Coromandel 3: 11, tab. 215. 1819; Fl. Ind. ed. 1832. 3: 188. 1832; Masters in Hook. f., Fl. Brit. India 1: 348. 1874; Hu, Fl. China, Malvaceae. 71. Volume 75, Number 4 1988 Naqshi et al. 1509 Malvaceae of Jammu & Kashmir State 1955; Ngwe, Union Burma Life Sci. 4: 200. 1971; Stewart in Nasir & Ali, Ann. Cat. Vasc. Pl. W. Pak. & Kashm. 480. 1972; Abedin in Nasir & Ali, Fl. W. Pak. 130: 28. 1979; Sharma & Kachroo, Fl. Jammu 1:110. 1981. TYPE: Roxb., Pl. Coromandel 3: 11, tab. 215. 1819 Kydia fraterna Roxb., Pl. Coromandel 3: 12, tab. 216. 1819. Medium-sized trees, the herbaceous portions with stellate pubescence. Leaves broadly cordate to su- borbicular, entire or usually 3-5-angled, stellate- pubescent, the midrib (sometimes adjacent nerves also) with a basal gland abaxially. Flowers panicu- late, polygamous. Epicalyx segments 4-6, oblong, in fruit obovate-spathulate, spreading stellately, densely stellate-villous, 5-parted; lobes triangular, incurved, persistent, + enclosing fruit. Corolla ro- tate, connate at base, 1-2 cm diam., white; petals obovate-obcordate, barbate at the base. Staminal tube 3-5 mm long, included, divided in the apical half into 5 branches, each with a cluster of 3-5 sessile anthers, rudimentary in the carpellate flow- ers. Ovary globose, villous, 3-carpellate; style with 3 terminal branches, each with a peltate stigma, rudimentary in the staminate flowers. Capsule subglobose, loculicidally 3-valved, stellate-pubes- cent. Seeds reniform, glabrous. India, Pakistan, Burma, and China. In our area this species occurs, uncom- Distribution. monly, in Jammu Province. Specimens examined. INDIA. JAMMU: Nandni, un- common, collected from a slope opposite Brig. Atma Singh's Memorial, small tree, bark gray, young shoots and Pon icles grayish, leaves hoary, flowers white, aa 728 (KASH); also reported from Mirpur ad Billawar 2s Lambert (1933). The wood of this species is straight-grained and good for house building. The liber yields fiber, and the leaves are said to be used as an embrocation. 5. GOSSYPIUM L., Sp. Pl. 693. 1753; Gen. Pl. 5th Edition. 309. 1754 About 40 species distributed in tropical and sub- tropical regions; two cultivated species are recorded from our area KEY TO THE SPECIES OF DOM M IN JAMMU AND KASHMIR STA la. Perennial suffrutescent herbs; stipules linear; epicalyx segments connate at the base, + en tire, with 3 small teeth at the apex; Danan equal in l. arboreum lb. Annual harta; stipules ovate-falcate; epicalyx segments free, laciniate, with 7-9 long and acuminate teeth at the apex; filaments unequal in length, the upper ones longer ...2. G. hirsutum pi . Gossypium arboreum L., Sp. Pl. 693. 1753; Masters in Hook. f., Fl. Brit. India 1: 347. 1874; Hu, Fl. China, Malvaceae. 62. 1955; Stewart in Nasir & Ali, Ann. Cat. Vasc. Pl. W. Pak. & Kashm. 477. 1972; Abedin in Nasir & Ali, Fl. W. Pak. 130: 30. 1979. TYPE: Linn. Herb. no. 874/3 (holotype, LINN). Gossypium rubrum Fórskal, Fl. Aegypt.-Arab. 125. 1775. Perennial, pubescent, suffrutescent herbs with purple branches. Leaves ovate to orbicular or sub- reniform, 5- 7-parted; lobes oblong-lanceolate, acute, stellate-pilose adaxially, sparsely villous abaxially; stipules linear, caducous. Flowers soli- tary, axillary. Epicalyx segments 3, foliaceous, cor- date, connate at the base, + entire, with 3 small teeth at the apex, stellately hirsute and villose on the nerves. Calyx cupular, + 5-dentate. Corolla pale yellow, usually with a maroon center, some- times all purplish. Staminal tube included, anther- iferous throughout; filaments equal in length. Cap- sule fibrous, ovoid, beaked, with persistent, accrescent epicalyx and calyx. Seeds densely cov- ered with long and short hairs. Distribution. can; widely cultivated in tropical and subtropical regions of the Old World. According to Hu (1955), it has escaped from cultivation in Hainan and southwestern Sichuan in China. In our area it was occasionally grown in the Kashmir Valley until recently. Vernacular name: **Kapas." Origin uncertain, possibly Afri- Additional specimen examined. INDIA. KASHMIR: Pampore, 6 June 1970, G. N. Javeid s.n. (KASH) 2. Gossypium hirsutum L., Sp. Pl. 2nd Edi- tion. 975. 1763; Prokhanov in Shishkin & Bobrov, Fl. U.R.S.S. 15: 178. 1949; Hu, Fl. China, Malvaceae. 66. 1955; Abedin in Nasir & Ali, Fl. W. Pak. 130: 31. 1979. Gossy- pium herbaceum L. var. hirsutum Schumann, Nat. Pflanzenf. III. 6: 51. 1890. TYPE: Miller's description (see Fryxell, 1968: 882). G. religiosum L., Syst. Nat. 12th Edition. 2: 462. 1767. Annual, erect, hirsute herbs with green or red- tinged branches. Leaves broadly cordate, + orbicu- lar, 3(-5)-lobed, upper ones sometimes entire and ovate; lobes triangularly ovate, abruptly acumin- ate, glabrescent with simple and stellate hairs on both surfaces; stipules ovate-falcate, caducous. 1510 Annals of the Missouri Botanical Garden Flowers solitary, axillary. Epicalyx segments as in the preceding species but free and with 7-9 long and acuminate teeth at the apex. Calyx cupular, 5-toothed. Corolla pale yellow. Staminal tube as in the preceding species but filaments unequal in length, the upper ones longer. Capsule fibrous, ovoid, beaked. Seeds thickly covered with white pubescence. Distribution. A native of Central America, acclimatized from Guatemala northwards to the cotton belt of the southern United States and cul- tivated in all cotton-growing countries. [n our area it was grown in. Kashmir iore until recently. “Kapa Vernacular name: ecimen examined. INDIA. KASHMIR: exact locality S not given, M. Y. Khan s.n. (KASH) A third species, Gossypium herbaceum L., has been reported from the Kashmir Valley by Stewart (1972: 478), but we have not seen any specimens. It is said that cotton thrives in Kashmir, but the practice of growing it here has now nearly ceased. The various species of cotton are abundantly cultivated in many countries of the world for their seeds, which are densely covered with short and long hairs, forming the cotton of commerce. 6. THESPESIA Solander ex Correa, Ann. Mus. Natl. Hist. Nat. 9: 290, tab. 8, f. 2. 1807, nom. cons. Azanza Alef., Bot. Zeit. 19: 298. 1861 About 15 species, distributed in tropics of both hemispheres; a single species has been reported from our area m populnea s Solander ex Correa, s. Natl. Hist. 9 1: 345. 1874; Hu, Fl. China, Malvaceae. 69. 1955; Ngwe, Union Burma J. Life Sci. 4 200. 1971; Stewart in Nasir & Ali, Ann. Cat. Vasc. Pl. W. Pak. & Kashm. 484. 1972; Abedin in Nasir & Ali, Fl. W. Pak. 130: 32. 1979. Hibiscus populneus L., Sp. Pl. 694. 1753. Malvaviscus populneus (L.) Gaertner, Fruct. Sem. Pl. 2: 253, tab. 135, f. 3. 1791. TYPE: Herb. Herm. Volume V, fol. 208, tab. 258 (lectotype, BM). Hibiscus bacciferus Forster f., Fl. Ins. Austr. 48, 1786. Thespesia macrophylla Blume, Bijdr. 2: 73. 1825. Medium-sized bushy trees; herbaceous portions covered with ferruginous, peltate scales. Leaves ovate, cordate at base, shortly acuminate, entire; stipules linear-lanceolate, caducous. Flowers soli- tary, axillary; pedicels 1-5 cm long, articulate at base. Epicalyx segments 3, lanceolate, caducous. Calyx cupular, truncate, minutely 5-toothed, co- riaceous, persistent. Corolla convolute, campanu- late, pale yellow with a crimson center. Staminal tube cylindric, included, 5-dentate at the apex, antheriferous for most part; filaments paired, the anthers horseshoe-shaped. Ovary 5-loculate; style + clavate, unbranched; stigma elongate, scabrous. Capsule + globose, almost woody but easily com- pressed, indehiscent. Seeds obovoid, angular, pi- lose. Distribution. A coastal plant, common in tropical countries. In Jammu and Kashmir State it has been reported only from Udhampur in the 1933: 3; Sharma & Jammu Province (Lambert, Kachroo, 1981: 113) According to Ngwe (1971), the bark of this plant is used in treating piles, dysentery, and skin dis- eases. 7. MALVA L., Sp. B 687. 1753; Gen. Pl. 5th Edition. 308. 175 Bismalva Medicus, Malvaceae. 39. 1787. Over 100 species native to Europe, Asia, and Africa. Several species are naturalized in America, Australia, and New Zealand; represented in our area by nine species, three being cultivated or escapes from cultivation. KEY TO THE SPECIES OF MALVA IN JAMMU AND KASHMIR STATE la. Epicalyx segments oblong, oblong-lanceolate, ovate, or ovate-lanceolate. 2a. v 7-lobed; staminal tube with simple or 2-rayed, retrorse hairs N of š des 3-5-lobed Diode ovate-lanceolate; flowers 5-15 ex; mericarps 10-14, glabrous ; staminal tube stellately Lira nt. (rarely fewer than 5) in fascicles; petals 1.5-2 cm wide 3. M. 4. M. bucharica mauritiana w c ent. 4a. Petals oblong-obovate, 0.7-2 x 0.5 cm; mericarps pubescent cm; mericarps glabrous 4b. Petals obovate, 2-3 x 1 a ; Spa lanceolate; flowers solitary or 2-4 in fascicles; petals 1 cm or less wide at apex; mericarps l. M. ambigua 2. M. sylvestris Volume 75, Number 4 1988 Naqshi et al. 1511 Malvaceae of Jammu & Kashmir State lb. Epicalyx segments linear to linear-lanceolate a. Plants biennial or perennial; mericarps 12- 15, pubescent, smooth throughout; seeds pubescent ........... 5. M. neglecta 5b. Plants mostly lot a biennial; mericarps 9-12, glabrous, striate-rugose at least on sides and margins; seeds gla alyx 6a. lobes + diet and neto in fruit; corolla shorter than, equaling, or slightly — res staminal tu m long, glabrous; mericarps with raised reticulation on back, + winged and acute along margins 7a. Flowers solitary or paired, rarely more than 2 but never compact; pedicels distinctly visible, -2 cm long; calyx ed, + entire 5-6 mm long, slightly enlarged in fruit; margins of mericarps slightly 6. M. iata ps 7b. Flores usually many, compactly a pedicels generally not visible, 3-5 m .1 calyx 3-5 m undulate-toothed e c 3-5 mm mar m long; mm long, enlarged in fruit to c cm; margins of mericarps distinctly winged, . M. parviflora . Calyx lobes incurved and enclosing the fruit; corolla 174-2 times as long as calyx; stami inal tube m long, glabrous or pubescent; mericarps smooth on back, unwinged and rounded along 8a. Plants annual, pubescent; leaves suborbicular, 5- 7-lobed; petioles 2-9 cm ent fruiting calyx less than 10 mm long; petals r retuse, ca. 2 tim staminal tube gee pubescent; fruit ca. 5 m mes y length of sepals, the cla iam.; mericarps 12 . ubescent; . e ee Me 8b. Plants annual to biennial, glabrescent; leaves orbicular. usually 5-lobed; petioles (1.5-)4-2 (-24) cm long; fruiting calyx 10-15 mm long; petals scarcely notched, 1% times or less ie length of sepals, the claw glabrous; staminal tube glabrous or pubescent with simple hai iam.; mericarps 10-12 9. M towards the apex; fruit 5-7 mm 1. Malva ambigua Guss., Fl. Sicul. Prodr. 2: 331. 1828; Iljin in Shishkin € Bobrov, Fl. U.R.S.S. 15: 48. 1949. TYPE: described from Sicily (NAP). Malva sylvestris L. var. eriocarpa Boiss., Fl. Orient. 1: 819. 1867. Biennial to perennial herbs. Stems erect or as- cending, usually weak, sparsely pubescent. Leaves + semiorbicular, usually truncate (to cor- date) at base, 3—5-lobed, serrate, glabrescent; stip- ules lanceolate, 2-4 mm long; petioles 2- long. Flowers axillary, solitary or in fascicles of 2- 4; pedicels 1-3 cm long, with simple or stellate hairs. Epicalyx segments narrowly ovate or oblong. Calyx 3-6 mm long, stellately pubescent, slightly accrescent in fruit; lobes broadly triangular. Petals lilac, oblong-obovate, 0.7-2 x 0.5 mm, claw pu- bescent. Staminal tube 3-5 mm long, pilose with stellate hairs. Fruit 6 mm diam.; mericarps 9-12, pubescent, with raised reticulation on back. Seeds reticulate, glabrous. Distribution. U.S.S.R., western and eastern Mediterranean, Iran, Afghanistan, and Pakistan. In our area infrequent in the Kashmir Valley, col- lected on house tops, waste places, moist sites, and sides of water courses. Additional specimens examined. INDIA. KASHMIR: Ganderbal, G. H. Dar 2499 (PF); Srinagar, G. N. Javeid 590 A (KASH); Srinagar, around graveyards in associ- ation with Urtica dioica L., 4. R. Naqshi 8165 (KASH). First record for India. : nilo 2. Malva sylvestris L., Sp. Pl. 689. 1753; Mas- ters in Hook. f., Fl. Brit. India 1: 320. 1874; Ijin in Shishkin & Bobrov, Fl. U.R.S.S. 15: 41. 1949; Kitamura, Fl. Afghan. 171. 1960; Stewart in Nasir & Ali, Ann. Cat. Vasc. Pl. W. Pak. € Kashm. 481. 1972; Abedin in Nasir & Ali, Fl. W. Pak. 130: 37. 1979. TYPE: described from western Europe, Linn. Herb. no. 870/22 (holotype, LINN). Biennial to perennial herbs. Stems erect, pu- bescent to glabrescent with simple (or bifid) hairs. Leaves + suborbicular, truncate to broadly cordate at base, usually 3-lobed, crenate-dentate, sparsely pilose; stipules lanceolate, scarious, ca. 5 mm long; petioles 2-7 cm long, pilose. Flowers axillary, sol- itary or in fascicles of 2-4; pedicels ca. 2 cm long. Epicalyx segments ovate-oblong. Calyx 3-6 mm long, glabrescent with stellate hairs; lobes broadly triangular. Petals pink-purple, obovate, emargin- ate, 2-3 x 1 cm, the claw ciliate. Staminal tube ca. 3 mm long, pilose with stellate hairs. Fruit 5— 6 mm diam.; mericarps 10-12, glabrous, reticu- late. Seeds sparsely punctate. Distribution. Western Europe, northern Af- rica, and Asia. Occasionally cultivated for greens in the Kashmir Valley and at certain places evi- dently escaped from cultivation, also collected from Ladakh (Stewart, 1972: 481) Additional specimen examined. INDIA. KASHMIR: Srinagar, A. R. Naqshi 8153 (KASH). The species resembles M. ambigua Guss. and 1512 Annals of the Missouri Botanical Garden M. mauritiana L. but differs from the former in having glabrous fruits and from the latter in having narrower and emarginate petals, and fewer flowers in the fascicles. Riedl (1976) followed Boissier (1867) in treating all of the above three species as M. sylvestris varieties sylvestris, eriocarpa Boiss., and mauritiana (L.) Boiss. Malva sylvestris is believed to have been cul- tivated by ancient Greeks and Romans as a me- dicinal and edible plant. An infusion of flowers and leaves is used internally and as a gargle. The in- fusion, mixed with honey, is taken in case of ca- tarrhal ailments, inflammatory conditions of the digestive tract, and constipation. A paste of leaves and flowers is applied in case of external inflam- matory conditions. Flowers are also used in coloring medicine, liquors, and wool. The coloring properties are said to be due to glucoside malvin and diglu- coside malvidin in the petals. 3. Malva mauritiana L., Sp. Pl. 689. 1753; Ijin in Shishkin & Bobrov, Fl. U.R.S.S. 15: 49. 1949; Abedin in Nasir & Ali, Fl. W. Pak. 130: 38. 1979. Malva sylvestris var. mauri- tiana (L.) Boiss., Fl. Orient. 1: 819. 1867; Masters in Hook. f., Fl. Brit. India 1: 320. 1874; Stewart in Nasir & Ali, Ann. Cat. Vasc. Pl. W. Pak. & Kashm. 481. 1972; Riedl in K. H. Rechinger, Fl. Iran. 120: 19. 1976. TYPE: Linn. Herb. no. 870/24 (holotype, LINN) Biennial to perennial herbs; stems mostly rigid, erect, rarely ascending, glabrescent with simple and bifid hairs. Leaves orbicular to suborbicular, truncate to shallowly cordate at base, 3—5-lobed, coarsely crenate, sparsely pilose with simple and bi- (tri-)fid hairs; stipules ovate-lanceolate, 3-6 mm long; petioles 4-12 cm long, with a line of dense hairs apically. Flowers axillary, in fascicles of 5— 15, rarely fewer than 5; pedicels 1-4 cm long, unequal in length. Epicalyx segments ovate-lan- ceolate to ovate or oblong. Calyx 5-8 mm long, pilose with stellate hairs; lobes triangular, plicate at angles, accrescent in fruit. Petals dark pink to purple, 2-3 x 1.5-2 cm, obovate, retuse, claw pubescent at base. Staminal tube ca. 5 mm long, pilose with stellate hairs. Fruit 5-7 mm diam.; mericarps 10-14, glabrous, reticulate-wrinkled. Seeds finely punctate. Distribution. Mediterranean region, western Europe, and U.S.S.R., elsewhere cultivated. In our area collected from Ladakh, where it grows, rare and handsome, along borders of cultivated fields, also reported from Kashmir by Stewart (1972: 481). Additional specimen examined. INDIA. LADAKH: Nu- bra, along borders of cultivated fields, A. R. Naqshi & G. N. Dar 7218 (KASH). Most authors follow Boissier (1867) in treating this species as a variety of Malva sylvestris L. Whitmore (1979) considered M. mauritiana as synonymous with M. sylvestris. e species is used for the same purpose as M. sylvestris. In the Iranian pharmacopoeia it is em- ployed in a mixture with violets, Nymphaea can- dida, Ziziphus jujuba, Alhagi camelorum, and other species for preparation of the purgative in- fusion (Hooper, Useful Plants and Drugs of Iran and Iraq, 1937). s Malva bucharica lljin, Bot. Mater. Gerb. Glavn. Bot. Sada RSFSR 5: 4. 1924; Shishkin & Bobrov, Fl. U.R.S.S. 15: 55. 1949; Riedl in K. H. Rechinger, Fl. Iran. 120: 19. 1976. TYPE: Kurgan-Tyube, Roshewitz 175 (lecto- type, LE). Perennial herbs; stems erect or ascending, te- rete, glabrate or with simple to 2-branched hairs, sometimes completely glabrous. Leaves semiorbic- ular, truncate or + cordate at base, 5- 7-lobed, serrate-dentate, subglabrous, the nerves pilose with usually simple hairs; stipules ovate to broadly lan- ceolate; petioles 3-15 cm long. Flowers usually 3 in axils, crowded towards apices of branches; ped- icels 1.5-5 cm long, much shorter than the sub- tending leaf; epicalyx segments oblong-lanceolate; calyx 3-7 mm simple and trifid hairs, the lobes accrescent and enclosing the fruit, triangular-ovate. Petals pur- plish, obovate or oblong-obovate, 0.8-2.2 x 0.4- 1.2 cm, deeply notched, claw + densely pilose. Staminal tube 4-6 mm long, with simple and 2-rayed, retrorse hairs. Fruit 6-7 mm diam.; mer- icarps (8-)10(-12), glabrous or pubescent, retic- ulate-rugose on back. Seeds puncticulate. m o long, glabrous or pubescent with U.S.S.R. and Kashmir. In our area it occurs frequently in the Kashmir Province but has not been collected in Jammu and Ladakh. Distribution. Additional specimens examined. INDIA. , KASHMIR: Naqshi 8152 (KASH); ai S roadsides near university gate, G. H. Dar KASH); Srinagar, ae wari, in graveyards among de uc bushes, 4. R. Naq 8178 (PF), 8179-81 (KASH). Volume 75, Number 4 1988 Naqshi et 1513 Malvaceae pi Jammu & Kashmir State In habit this species resembles M. sylvestris but differs from it mainly in having simple and two- rayed, retrorse hairs on the staminal tube. First record for Indian subcontinent. n . Malva neglecta Wallr., Syll. Ratisb. 1: 140. 1824; Iljin in Shishkin & Bobrov, Fl. U.R.S.S. 15: 56. 1949; Hu, Fl. China, Malvaceae. 6. 1955; Kitamura, Fl. Afghan. 271. 1960; Stewart in Nasir & Ali, Ann. Cat. Vasc. Pl. W. Pak. & Kashm. 481. 1972; Singh & Kachroo, Forest Fl. Srinagar. 151. 1976; Riedl in K. H. Rechinger, Fl. Iran 120: 24. 1976; Abedin in Nasir & Ali, Fl. W. Pak. 130: 38. 1979. TYPE: without exact locality, Germany, Wallroth s.n. (E). Malva EEA sensu Maxim., Acta Hort. Petrop. 11: 78. 1890, non L.; Masters i in Hook. f., Fl. Brit. India 1: 320. 1874. M. vulgaris Ten., Fl. Napol. Suppl. 1: 62. 1811-1815. M. lignescens Iljin, Bot. Mite Gerb. Glavn. Bot. Sada RSFSR 2: 173. 1921 Biennial to perennial herbs with woody bases. Stems prostrate or decumbent, pubescent with stel- late hairs, especially on younger parts. Leaves or- bicular-reniform, cordate at base, crenate-dentic- ulate, occasionally shallowly 5(-7)-lobed on elongated branches, sparsely pilose with simple and stellate hairs adaxially and densely so with stellate hairs abaxially; stipules ovate-lanceolate, ca. 5 mm long; petioles 3-15 cm long, stellately villous. Flow- ers axillary, 3-4 in a fascicle, those on the lower branches occasionally solitary; pedicels, 0.5-5 cm long, unequal in length, much longer than the flower but shorter than the subtending leaf; epi- calyx segments linear to linear-lanceolate; calyx 5-8 mm long, stellately pilose; lobes triangular. Petals purplish to pinkish, sometimes white, 10— 13 x 3-4 mm, oblong-obovate, retuse, claw pu- bescent on the margin. Staminal tube 4-5 mm long, pubescent with simple, + retrorse hairs. Fruit —7 mm diam.; mericarps 12-15, pubescent, smooth. Seeds decent Distribution. Native to Old World, natural- ized in America. In our area occurring widely in Kashmir Valley along wastelands, meadows, and cultivated fields from 1,600 m altitude to the alpine zone; also reported from the Jammu and Ladakh regions. It is commonly used as a wild vegetable under the local name *'Sotsal." Additional specimens examined. INDIA. KASHMIR: Srinagar, Habak, G. N. Javeid 121 (KASH); Womens’ College campus, D. p 101(KASH); Harwan, G. Singh 1986 depu Tangmarg, A. R. Naqshi 599 (KASH); Rainawari, A. R. Naqshi 8159-60 (KASH); Gandia, exposed hill slopes, G. H. Dar 2 akura, 1, m, Dar "es 57 (KASH), 1058 (PF); Srinagar (Lal RN A. R. Naqshi 8166 (KASH); So- namarg, G. H. Dar 7679— 7682 (K ASH); Gund-Haknar, G. H. Dar 8644 (KASH); Hang (Sonamarg), G. H. Dar 8643 (KASH); Chatterhama (Srinagar), G. H. Dar 8841 (KASH); Manigam (Lar), G. H. Dar 5328-31 (KASH). A number of varieties and forms of this species can be distinguished in our area. However, their recognition is deferred until a satisfactory mono- graphic work on the species within the area is available. These plants are said to be used in medicine due to the high content of mucilage in the foliage and roots. The leaves contain vitamin C and provitamin A, while the seeds contain about 18% of a light green oil. A decoction of leaves and roots is used as a gargle for treatment of inflammatory condi- tions of the respiratory tract and as fomentations for external treatment of skin inflammations, ul- cerations, and swellings. An infusion of leaves with milk is used for its diuretic effect. A decoction of the leaves is used as an enema to relieve consti- pation (Dar et al., 1984). The petioles are used for treating babies, instead of glycerine clysters. 6. Malva microcarpa Pers., Syn. Pl. 2: 251. 1806; Riedl in K. H. Rechinger, Fl. Iran. 120: 27. 1976; Abedin in Nasir & Ali, Fl. W. Pak. 130: 41. 1979. Malva parviflora var. microcarpa (Pers.) Loscos, Trat. Pl. Ar- agon 2: 203. 1877; Stewart in Nasir & Ali, Ann. Cat. Vasc. Pl. W. Pak. & Kashm. 481. 1972. TYPE: Herb. Persoon s.n. (holotype, L). nual herbs. Stems prostrate or ascending, stellately villous. Leaves orbicular-reniform, cor- date at base, 3- 7-angular, crenate to serrate, gla- brescent; stipules ovate or lanceolate, 2-4 mm long; petioles 1—10(—25) cm long. Flowers axillary, solitary or paired, rarely more but then never com- pact; pedicels 0.3-2 cm long, distinctly visible; epicalyx segments linear; calyx 5-6 mm long, slightly enlarged and rotate in fruit; lobes trian- gular, acute to acuminate. Petals white (sometimes with pinkish tips), equaling or slightly exceeding the calyx. Staminal tube 1-2 mm long, glabrous. Fruit 3-6 mm diam.; mericarps 9-10, glabrous, reticulate, with slightly winged margins. Seeds gla- rous. Distribution. Native to the Mediterranean Region, Malaysia, Iran, Afghanistan, and Pakistan. 1514 Annals of the Missouri Botanical Garden In our area it has been collected, infrequently, from Jammu. Additional specimen examined. INDIA. JAMMU: Tilo Talab, common along drains; annual herbs, mucilaginous when bruised, flowers white, B. M. Sharma 389 (KASH). First record for India. 7. Malva parviflora L., Demonstr. Pl. 18. 1758; Sp. Pl. 2nd Edition. 269. 1763; Masters in Hook. f., Fl. Brit. India 1: 321. 1874; Iljin in Shishkin & Bobrov, Fl. U.R.S.S. 15: 63. 1949; Kitamura, Fl. Afghan. 271. 1960; Stewart in Nasir & Ali, Ann. Cat. Vasc. Pl. W. Pak. & Kashm. 481. 1972; Riedl in K. H. Rechinger, Fl. Iran. 120: 23. 1976; Abe- din in Nasir & Ali, Fl. W. Pak. 130: 42. 1979, TYPE: described from Barbary (north Africa), Linn. Herb. no. 870/17 (holotype, LINN). Annual herbs; stems erect or prostrate-ascend- ing, sparse y pubescent with stellate hairs to gla- brescent. Leaves orbicular-reniform, cordate at base, oun ere often shallowly 3-7-lobed, sparsely pubescent with simple or 2-fid hairs on the adaxial surface, usually stellately pilose abax- ially; stipules lanceolate to ovate, 2-5 mm long; petioles 3-17 cm long, longer than blade, simple and stellately pubescent, especially apically. Flow- ers usually in compact axillary fascicles; pedicels 3-5 mm long, generally not visible, subglabrous. Epicalyx segment linear. Calyx 3-5 mm long, pi- lose with stellate hairs, accrescent in fruit to ca. | m; lobes triangular, mucronate. Petals white, sometimes pinkish at the tips, usually shorter than or equaling the calyx, oblong, slightly narrowed at base, scarcely notched at apex, glabrous. Staminal tube ca. 2 mm long, glabrous. Fruit 5- 6 mm diam.; mericarps 9-10, glabrous, with raised reticulation and toothed, winglike margins. Seeds glabrous. claw Distribution. Mediterranean Region, Anato- lia, Iran, Iraq, Afghanistan, Pakistan, India, and Arabia. Many workers (Javeid, 1970; Stewart, 1972; Sharma & Kachroo, 1981) have reported this species from our area, but there is no authentic specimen, because all the specimens under this name turned out to be either Malva neglecta or M. microcarpa. 8. Malva mohileviensis Downar, Bull. Soc. Mosc. 1: 177. 1861; Iljin in Shishkin & Bob- rov, Fl. U.R.S.S. 15: 64, pl. 3, f. 1. 1949; Abedin in Nasir & Ali, Fl. W. Pak. 130: 43. 1979, TYPE: vicinity of Mogilev (LE). Annual herbs; stems erect, + purplish, pubes- cent with simple and 2-fid hairs, especially towards the apex. Leaves suborbicular, cordate at base, 5- 7-lobed, serrulate to crenulate-dentate, sparsely pubescent with simple or 2-fid hairs on the upper surface, more pubescent and with mixed stellate hairs beneath; stipules lanceolate-ovate, 3-5 mm long; petioles 2-9 cm long. Flowers axillary, in fascicles of 4—7(—many); pedicels 5-10 mm long. Epicalyx segments linear, green to purple. Calyx ca. 5 mm long, green to purple, glabrescent, ac- crescent and scarious in fruit, the lobes triangular. Petals pinkish, about twice the length of calyx, obovate, retuse, claw slightly pubescent. Staminal tube ca. 4 mm long, retrorsely pubescent. Fruit ca. 5 mm diam.; mericarps 12, glabrous, smooth on the back, transversely rugose along the rounded margins, radially wrinkled on the sides. Seeds mi- nutely puncticulate. Distribution. Japan, China, U.S.S.R., Paki- stan. In our area it occurs infrequently in Kashmir and Ladakh Additional specimens examined. INDIA. KASHMIR: Sonamarg, along roadsides just near the Shutkar Bridge, G. H. Dar 8483 (PF), 8484 (KASH). LaDAKH: Leh, A R. Naqshi & G. N. Dar 7217 (KASH). In Tibetan medicine the flowers of this species are used as a diuretic. The leaves and young shoots can be used as a salad or as a vegetable. It is a valuable forage plant due to a high protein content and tender consistency. It has been estimated that this species contains twice as much protein as any other forage plant, so it increases the yield and quality of milk when fed to cows. First record for India. 9. Malva verticillata L., Sp. Pl. 689. 1753; Masters in Hook. f., Fl. Brit. India 1: 320. 1874; Iljin in Shishkin & Bobrov, Fl. U.R.S.S. 15: 68. 1949; Hu, Fl. China, Malvaceae. 5. 1955; Stewart in Nasir & Ali, Ann. Cat. Vasc. Pl. W. Pak. & Kashm. 481. 1972; Riedl in K. H. Rechinger, Fl. Iran. 120: 22. 1976; Abedin in Nasir & Ali, Fl. W. Pak. 130: 43. 1979. LECTOTYPE: Linn. Herb. no. 870/26 (LINN). 9a. Malva verticillata var. verticillata [see Abedin in Nasir & Ali, Fl. W. Pak. 130: 45. 1979]. M. chinensis Miller, Gard. Dict. 8th Edition. 670. 1768. Annual or biennial herbs; stems erect, green to purplish, sparsely stellately pubescent. Leaves usu- ally orbicular, cordate to subtruncate at base, usu- Volume 75, Number 4 1988 Naqshi et al. 1515 Malvaceae of Jammu & Kashmir State ally 5-lobed, coarsely crenate-dentate, glabrescent on the adaxial surface with simple, or 2-fid hairs, more pubescent abaxially with stellate, simple, or 2-fid hairs; stipules lanceolate, 5 mm long; petioles 4-20(-24) cm long, glabrescent, grooves. Flowers axillary, subsessile, in dense and compact fascicles of 5-many; pedicels 5-8 mm long, + of equal length, all hidden by flowers or fruits. Epicalyx segments linear or linear-lanceo- late, acute. Calyx 5-6 mm long, sparsely hirsute with stellate hairs, prominently reticulate-veined, with villous accrescent in fruit to 10-15 mm, the lobes tri- angular, with long-ciliate margins. Petals purplish, 7-9 mm long, scarcely notched, the claws gla- brous. Staminal tube 3-5 mm long, glabrous or pubescent with simple hairs towards the apex. Fruit 5-7 mm diam., enclosed in accrescent calyx, the mericarps 10-12, glabrous, smooth on the back, rugose along the rounded margins, radially striate on the sides. Seeds glabrous. Distribution. China, Europe, Asia, Ethiopia, Egypt. In our area commonly cultivated in the Kashmir Valley for its leaves which are used as a vegetable, also collected from Jammu and Ladakh. Vernacular names: **Parim sotsal”, “Bagh sotsal.” Additional specimens examined. INDIA. KASHMIR Naqshi 107 (KASH); Chat ar), ASH, PF); Beehama, 7563- 65 (KASH); Hi pur (Pir Panjal), G. A. Clone š s.n. (K, fide Abedin, 1979). jAMMU: Poonch Dist., 4. Rashid, E. Nasir & RRS 25586 (RAW). 9b. Malva verticillata var. rafiqii S. Abedin in Nasir & Ali, Fl. W. Pak. 130: 45. 1979. TYPE: Hazara District, Pakistan, S. Abedin & M. Qaiser 9109 (holotype, KUH). Malva d var. chinensis sensu Hu, Fl. Chin Malvaceae. 6, tab. 15, f. 5. 1955. Not M. M MS Mille. - Differs from the preceding variety in having smaller habit, comparatively thinner stems, smaller and shallowly 3—5-lobed leaves with shorter (1.5— 10 cm) petioles, flowers in looser fascicles of 2-6, pedicels of unequal length, 10-20(-25) mm long, the longer ones not hidden by the clusters of flowers or fruits. Distribution. China and Pakistan and in the Kashmir Valley. Additional specimens examined. INDIA. KASHMIR Ganderbal, G. H. Dar 1762-66 (KASH); Haknar (Gund), 2,050 m, G. H. Dar 8640 (PF), 8641-8642 (KASH). 8. ALTHAEA L., Sp. Pl. 686. 1753; Gen. Pl. 5th Edition. 307. 1754 About 12 species, distributed in Africa, Asia, and Europe; two species are recorded from our area. la. Leaves entire to E oe 3-lobed; corolla al- most twice the length o cere staminal tube cent; anle , pubescent jin dim. seeds smooth, not esca EDEN officinalis . Leaves deeply 3-5-fid or parted; en usually less than twice the lengt yx; stami inal tube almost glabrous; mericarps 12-18, pu- bescent towards apex, glabrous at base; seeds minutely whitish-verrucose on bac 2. A. broussonetiifolia he = 1. Althaea officinalis L., Sp. Pl. 686. 1753; Masters in Hook. f., Fl. Brit. India 1: 319. 1874; Iljin in Shishkin & Bobrov, Fl. U.R.S.S. 15: 131. 1949; Kitamura, Fl. Afghan. 270. 1960; Stewart in Nasir & Ali, Ann. Cat. Vasc. Pl. W. Pak. & Kashm. 417. 1972; Riedl in K. H. Rechinger, Fl. Iran. 120: 39. 1976; Abedin in Nasir & Ali, Fl. W. Pak. 130: 46. 1979. LECTOTYPE: Linn. Herb. no. 863/1 (LINN) Perennial herbs, stems erect, with densely to- mentose branches. Leaves triangular to broadly ovate, acute, base rounded or truncate, the margin irregularly serrate-dentate, sometimes superficially 3-lobed, densely pubescent on both surfaces, es- pecially beneath; stipules linear-lanceolate, ca- ducous. Flowers axillary, borne on many-flowered peduncle; pedicels 2-10 mm long. Epicalyx seg- ments 8-12, linear. Calyx with the lobes connate below the middle, 5-lobed, 6-12 mm long, persis- tent. Corolla pinkish to white, 2-3 cm diam.; petals broadly obovate to oblong-obovate, 1.2-2.2 cm long, slightly notched at apex, the claw with ciliate margin. Staminal tube cylindric, with short-papil- lose hairs. Mericarps 15-25, minutely stellate-pu- bescent throughout. Seeds smooth, glabrous. Distribution. Europe, Palestine, Syria, Tur- key, Iran, Afghanistan, Pakistan, India. Though the species has been collected in the Kashmir Val- ley, it appears to be rare in our area now. A report from Baramulla, Kashmir is in Stewart (1972). Additional specimens examined. INDIA. KASHMIR: Bandipur, Jacquemont 1082; Pampore, Drum 15029 Abedin (19779) described the fruit of this species as glabrous in his key, perhaps an error. 2. Althaea broussonetiifolia Iljin in Shishkin & Bobrov, Fl. U.R.S.S. 15: 678. 1949. TYPE: Stalingrad (Olim Zarizyn), Wunderlich 1839 (LE). 1516 Annals of the Missouri Botanical Garden Perennial herbs; stems + erect, densely stellate- pubescent on almost all parts. Leaves deeply 3-5- fid or -parted with irregularly dentate, oblong-lan- ceolate lobes, densely pubescent on both surfaces, especially beneath; stipules linear, caducous. Flow- ers on axillary and terminal racemose-paniculate peduncles; pedicels much shorter than calyx. Epi- calyx segments 7-9, lanceolate. Calyx lobes 5, connate below the middle, 6-10 mm long, persis- tent. Corolla pink, 2-2.5 cm diam.; petals obovate to oblong-obovate, 8-15 mm long, slightly notched at apex, the claw fringed-pubescent. Staminal tube cylindric, almost glabrous. Mericarps 12-18, stel- late-pubescent except at base. Seeds sparsely and minutely whitish-verrucose, especially in the lower part. Distribution. U.S.S.R. The taxon, hitherto considered endemic to the U.S.S.R., has been col- lected in the Kashmir Valley. Apparently it is rare in our area. Additional specimens m e m. KASHMIR: Narbal-Suzeath, on the bor vegetable gardens, 4. x Naqshi 8167 (PF), HI Shakar (Srinagar), G. N. Javeid 596 (KASH). First record for Indian subcontinent. 9. ALCEA L., ui i 687. 1753; Gen. Pl. 5th Edition. 307. About 60 species distributed chiefly in eastern Mediterranean Region; represented in our area by three species, all cultivated or escapes from cul- tivation. KEY TO THE SPECIES OF ALCEA IN KASHMIR AND JAMMU STATE la. "ir leaves undivided or shallowly lobed. 2a. Stem and branches sparsely setose with stellate hairs when young, + glabrous at maturity . rosea ; Ste mand aya tebs uus bristly with per- sistent stellat A. pa allida Upper ie kept lobed almost to the bas 2. A. isum m N c 2d — . Alcea rosea L., Sp. Pl. 687. 1753; Iljin in Shishkin & Bobrov, Fl. U.R.S.S. 15: 126. 1949; Zohary, Israel J. Bot. 12: 12. 1963; Abedin in Nasir & Ali, Fl. W. Pak. 130: 49. 1979. TYPE: Linn Herb. no. 869/ 1 (LINN). Biennial (or perennial) herbs; stem erect, sparse- ly setose with stellate hairs when young, + glabrous at maturity. Leaves orbicular-ovate, cordate at base, obtuse at apex, the lower ones shallowly 5—7-lobed, the upper leaves undivided or shallowly 3-lobed, crenate-dentate, scabrous with stellate pubescence on both surfaces; stipules ovate, tricuspidate. Flow- ers axillary, solitary or subfasciculate, the inflo- rescences spikelike towards the apices due to short- er pedicels and gradual diminution of the subtending leaves into leaflike bracts. Epicalyx 6—7-lobed, the lobes ovate-lanceolate. Calyx campanulate, 5-lobed, densely stellate-pubescent like the epicalyx. Corolla 5-8 cm diam., of various colors but usually red; petals obovate-cuneate, notched at apex, the claw barbate. Staminal tube 5-angled, glabrous. Fruit depressed, covered by persistent calyx; mericarps 20-40, stellate-pubescent, channeled and winged dorsally. Seeds reniform, pubescent. Distribution. According to Zohary (1963), “Wild A. rosea L. seems to be indigenous almost exclusively on the Aegean islands and the adjacent alkan Peninsula. The areas of its origin are no doubt the northeastern Mediterranean countries, but not China which is beyond the natural range of the genus." Alcea rosea is cultivated as an ornamental almost everywhere. In the Kashmir Valley it often grows as an escape from cultivation, together with the two following species. Sharma & Kachroo (1981) reported this species from Batote, ammu. Additional specimens examined. INDIA. KASHMIR: Srinagar, G. N. Javeid 33 (KASH); Sind Valley, Prang, G. H. Dar 6217 (KASH); Ganderbal, G. H. Dar 6619 pus, A. R. Naqshi & G. N. The flowers and seeds of this species are said to have diuretic properties, and the roots and seeds are used as demulcents. According to Dar et al. (1984), a decoction er the roots boiled with water and milk is applied lly for treating dermatitis and goiter; it is also given to pregnant women to ease delivery. A decoction of flowers with milk and "gud" is reportedly applied for boils. 2. Alcea lavateriflora (DC.) Boiss., Fl. Orient. 1: 828. 1867; Riedl in K. H. Rechinger, FI. Iran. 120: 65. 1976; Abedin in Nasir & Ali, Fl. W. Pak. 130: 50. 1979. Althaea lava- teriflora DC., Prodr. 1: 437. 1824; Kita- mura, Fl. Afghan. 270. 1960. TYPE: Prope Seydeol rodius Lipain, Meryon. Alcea persarum Bornm. sensu Zohary, Israel. J. Bot. 12: 15. 1963. This species resembles 4. rosea L. in habit and flowers but differs chiefly in having deeply palmate- Volume 75, Number 4 1988 Naqshi et al. 1517 Malvaceae of Jammu & Kashmir State lobed leaves, especially towards the apices of branches, and in having narrower calyx lobes. Distribution. Turkey, Greece, and Bulgaria; occasionally cultivated in Iraq, Iran, Pakistan, and elsewhere. Additional specimens examined. INDIA. KASHMIR: university campus, 4. R. Naqshi & G. N. Dar 8170 (KASH); Ganderbal, Power House, G. H. Dar 6620— 6621 (KASH); Prang, G. H. Dar 6216 (KASH, PF). 2 Alcea pallida (Waldst. & Kit.) Besser, Enum. Pl. 2: 872. 1822; Iljin in Shishkin € Bobrov, Fl. U.R.S.S. 15: 118. 1949; Zohary, Israel J. Bot. 12: 11. 1963; Abedin in Nasir & Ali, Fl. W. Pak. 130: 50. 1979. Althaea pallida Waldst. & Kit. in Willd., Sp. Pl. 3: 773. 1800; Stewart in Nasir & Ali, Ann. Cat. Vasc. Pl. W. Pak. & Kashm. 477. 1972. TYPE: described from Hungary (Prague). This species also resembles A. rosea L. in gen- eral habit, leaves, and flowers but differs in having dense, persistent, stellate hairs on the stem and branches and in having transversely rugose wings on the mericarps. Distribution. Central Europe, Balkan Penin- sula, and Asia Minor; cultivated elsewhere. Additional specimens examined. INDIA. KASHMIR: Sind Valley, Ganderbal, G. H. Dar 6618 (KASH, PF). All species of Alcea L. are raised as ornamentals in the Kashmir Valley under the local name of “Saz Posh.” They are often grown in close asso- ciations. The three species were, until now, referre to Althaea rosea (L.) Cav. However, the genera Alcea and Althaea differ so markedly, particularly in the structures of the staminal column and car- pels, that they are no longer considered a single genus. In fact Alcea approaches more closely Lavatera than Althaea.The species of Alcea con- tain 12-14% fiber in the stem and are suitable for paper production. The flowers contain mucilage and are used for gargling. They are also taken internally as an emollient for treatment of catarrhal gastric complaints. A dye extracted from the petals is used for coloring wines, vinegar, liquors, food products, silk, and wool. 10. LAVATERA L., Sp. Pl. 690. 1753; Gen. Pl. 5th Edition. 308. 1754 Some 45 species, chiefly Mediterranean, but extending to the Canaries, northwest Himalaya, central Asia, eastern Siberia, Australia, and the U.S.A. (California); represented in Jammu and Kashmir by a single species. Lavatera kashmiriana Cambess. in Jacquem., Voy. Inde 4: 29, tab. 32. 1844; Masters in Hook. f., Fl. Brit. India 1: 319. 1874; Iljin in Shishkin & Bobrov, Fl. U.R.S.S. 15: 78. 1949; Riedl in K. H. Rechinger, Fl. Iran. 120: 14. 1976; Singh & Kachroo, Forest Fl. Srinagar. 151. 1976; Abedin in Nasir & Ali, Fl. W. Pak. 130: 51. 1979. TYPE: India. Kashmir: Jacquemont s.n. (K). Malva cachemiriana (Cambess.) Alef., Osterr. Bot. Z. 12: 258. 1862. Lavatera cachemiriana var. haroonii v Abedin in Nasir & Ali, Fl. W. Pak. 130: 52. 9 Lavatera thuringiaca L. var. macromera ed Russk. Bot. Zurn. 7: 117. 1922 huringiaca vx macromera (Litw.) Iljin, Bot. Syst. Leningrad 5: 1924. Perennial, densely stellate-pubescent herbs with erect terete stems. Leaves orbicular, the base trun- cate to slightly cordate, margin crenate-serrate, 3— 7-angled or palmatifid to parted; stipules foliaceous, persistent. Flowers solitary, axillary; pedicels 3-6 cm long, articulate near the apex. Epicalyx seg- ments 3, foliaceous, connate in the lower half, ovate-orbicular, mucronate, accrescent in fruit. Calyx 5-lobed, longer than epicalyx, the lobes tri- angular to deltoid, accrescent in fruit. Corolla 4— 7.5 cm diam., pink-lilac; petals obovate, deeply notched. Staminal tube densely pubescent at base, antheriferous in the upper half. Fruit discoid; mer- icarps 20-25, glabrous, + rugose. Seeds glabrous. Distribution. India (Himalayan Mountains), Pakistan, Afghanistan, U.S.S.R., Iran. Stewart (1972) reported it for Poonch, Jammu. Additional specimens examined. INDIA. KASHMIR: Sonamarg, A. R. Naqshi 3975 (KASH); Aharbal, A. R. d 7523 (KASH); Baltal, 4. R. Naqshi 4010-12 ASH) Harwan, G. Singh 1984 (KASH); Hadurah oo G. H. Dar 2176 (KASH); Naranag, G. Dar 4142-43, 45 (KASH), 4144 (PF); Soraphraw [Sind Valley), G. H. Dar 8594 (PF), 8595-96 (KASH); Najwan (Kangan), A Dar 5787 (KASH), 7788 (PF), Harwan, G. N. Javeid 361 (KASH); Ferozpur Nullah (Gulmarg), U. Dhar 1258 (KASH). In Kashmir, this species serves as a vegetable to hill tribes under the name of “Wan Sotsal.” It can be raised as an ornamental. Its seeds contain about 12% oil, and a small amount of vitamin C is present in the stem and leaves. The plant has the capacity of yielding a good quality of fiber for binder twine, string, and ropes. 1518 Annals Ade i Garden Based on leaf pubescence Abedin (1979) rec- ognized two varieties as follows: adaxial surface of leaf with dense stellate hairs (var. cachemiriana) and adaxial surface of leaf except veins with simple and fascicled hairs (var. haroonii). Close scrutiny of a number of collections from different parts of our state, however, revealed that the varietal dis- tinction does not hold up. The difference in the pubescence of upper leaf surface does not seem to be constant, and intermediate conditions are not uncommon. 11. ABUTILON Miller, Gard. Dict. Abr. 4th Edition. 1: AB. 1754; 8th Edition. 1768. Over 150 species distributed in the tropics and subtropics of both hemispheres; represented in our area by four species. KEY TO THE SPECIES OF ABUTILON IN JAMMU AND KASHMIR STATE la. Corolla 2.5-3.5 cm diam.; staminal tube 5-8 mm long; — a (14-)15-20 . Corolla 1 m diam.; staminal tube 1-4 mm long; mericarps 10- 2a. Annual jf stimdbal tube ades "T A. theophrastii 2b. Perennial suffruticose herbs; vidi be stellate-pubescent to puberulent. 3a. Calyx lobes deltoid-ovate, erect in fruit; 10-12 mm ~ c ramosum . Calyxlobes lanceolate, nücrali in Lii etals 7-9 mm long; fruit ovoid; mer- icarps 13-16; seeds stellately pilose . Á. w c bidentatum m . Abutilon ramosum (Cav.) Guill. € Perr. in Guill. et al., Fl. Seneg. Tent. 1: 68. 1831; Masters in Hook. f., Fl. Brit. India 1: 328. 1874; Stewart in Nasir & Ali, Ann. Cat. Vasc. Pl. W. Pak. & Kashm. 477. 1972; Abedin in Nasir & Ali, Fl. W. Pak. 130: 55. 1979; Sharma & Kachroo, Fl. Jammu 1: 111.1981. Sida ramosa Cav., Diss. 1. 28, tab. 6, f. 1. 1785. TYPE: Senegal, 4danson (holotype, MA, photo; isotype, P). Abutilon sparmanoides Guill. E Perr. in Guill. et al., Fl. Seneg. Ten 83 A. elaeocarpoides Webb, Fragm. Fl. Pan 53. 1854. A. sidoides A. Gibson, Bombay Fl. Perennial, suffruticose herbs, tomentose with stellate and long, spreading hairs. Leaves broadly ovate, the base cordate, the apex acuminate, the margin coarsely crenate-serrate, often 3-lobate; stipules filiform to linear. Flowers axillary or ter- minal, solitary or denis or divided above dichot- omously as in a cyme; pedicels shorter than peti- oles. Calyx sided 5. lobed; lobes deltoid- Pus acuminate-cuspidate. Corolla yellow, ca. 1.7 diam.; petals 10-12 mm long. Staminal tube very short, puberulous. Fruit cylindrical; mericarps usu- ally 10, biaristate with awns ca. 2 mm long. Seeds 2-3 per mericarp, furfuraceous-dotted. Distribution. Tropical Africa, Arabia, Paki- stan, and India. Confined in our area to the Jammu Province. For a Billawar report see Lambert (1933). Additional specimen examined. e JAMMU: Ram Nagar, among Carissa spinarum L., ntose green shrub with yellow flowers, B. M. Sharma € 674 (KASH). 2. Abutilon theophrastii Medicus, Malven- fam. 28. 1787; Hu, Fl. China, Malvaceae. 31. 1955; Kitamura, Fl. Afghan. 269. 1960; Stewart in Nasir & Ali, Ann. Cat. Vasc. Pl. W. Pak. & Kashm. 477. 1972; Riedl in K. H. Rechinger, Fl. Iran. 120: 7. 1976; Abedin in Nasir & Ali, Fl. W. Pak. 130: 61. 1979. LECTOTYPE: India: Herb. Cliff. (BM). Sida abutilon L., Sp. Pl. 685. “ws j 1 . Sem. Pl, A. behrianum F. ide in Trans. & Proc. Philos. Inst. Victoria 1: 13. 1855 Annual, velutinous herbs. Leaves orbicular or broadly ovate, the base deeply cordate, the apex acuminate, the margin crenulate or undulate; stip- ules caducous. Flowers axillary, solitary or in few- flowered terminal racemes; pedicels shorter than petioles. Calyx 5-lobed; lobes ovate or lanceolate, acute. Corolla yellow, 1-2 cm diam.; petals 1-1.2 cm long, obovate. Staminal tube 2-4 mm long, glabrous. Fruit hemispherical; mericarps 12-16, strongly birostrate with awns 3-5 mm long. Seeds 3 per mericarp, stellate-pilose. Distribution. Native to India, introduced and naturalized in northern America, northern Asia, and westward to southern Europe. Confined in our area to the Kashmir Valley, where this is the only Abutilon growing now, although some other species have been ian by earlier workers. Vernacular name: “Yac Additional dein examined. INDIA. KASHMIR university campus, 4. R. Nagshi 189 (KASH); Srinagar, . R. Naqshi 3981 (KASH); Narbal, 4. R. shi & G. N. Dar 8171 (PF), 8172 E "redis G. N. us s.n. (Sind Valley) (KASH); H. Dar 8820 KASH); Jhelum River, cultivated ata R. R. & I. D. e 4980 (RAW). EN Volume 75, Number 4 1988 Naqshi et al. 1519 Malvaceae of Jammu & Kashmir State Abutilon, especially A. theophrastii, has been long cultivated for its coarse fiber suitable for mak- ing ropes, sackcloth, binder twine, string, and fish- ing nets. Its fiber is fairly tough, water resistant, and brittle. In America this is said to be preferred over jute and Manila hemp. The stem is used for paper manufacture and as fuel. The seeds yield up to 30% of a yellow, tasteless, and odorless oil, which approaches cotton oil, sesame oil, and peanut oil in its chemical composition. The oil is suitable for use in food and for hydrogenation. Inferior grades may be used as varnish oil and in soap manufacture. The flowers are used for coloring wines and, in China, for making ink. A decoction of roots and infusion of flowers is used internally and externally against inflammatory conditions. o . Abutilon bidentatum Hochst. ex A. Rich., Tent. Fl. Abyss. 1: 68. 1847; Masters in Hook. f., Fl. Brit. India 1: 326. 1874; Ki- tamura, Fl. Afghan. 269. 1960; Stewart in Nasir & Ali, Ann. Cat. Vasc. Pl. W. Pak. & Kashm. 475. 1972; Riedl in K. H. Rechinger, Fl. Iran. 120: 5. 1976; Abedin in Nasir & Ali, Fl. W. Pak. 130: 63. 1979. TYPE: Abys- sinia, prope Aguar, prov. Modat, Schimper 1003 (K). Abutilon cornutum Dalz. ex T. Cooke, Fl. Bombay 1: 98. 1901. A. pakistanicum Jafri et Ali in Jafri, Fl. Karachi. 220. 1966. Perennial, suffruticose herbs, canescent-tomen- tose with stellate, weak, spreading hairs. Leaves broadly ovate, the base deeply cordate, the apex acute-acuminate, the margin crenate-dentate, rarely 3-angular; stipules filiform. Flowers solitary, axil- lary; pedicels longer or shorter than petioles. Calyx 5-lobed, slightly accrescent; lobes lanceolate, ul- timately reflexed. Corolla pale yellow to yellow, 1— .5 cm diam.; petals 7-9 mm long, obovate. Stam- inal tube 2-3 mm long, stellate-pubescent. Fruit ovoid; mericarps 13-16, acute-acuminate, awn l- 2 mm long. Seeds 3 per mericarp, sparsely pilose with stellate hairs. Distribution. India, Pakistan, Iran, tropical Africa, and Arabia. Apparently confined to the Jammu region in our area. Stewart (1972) reported it from the Kashmir Valley, but we have not seen any specimens. Also note: Jammu: Billawar (Lam- bert, 1933: 3; Sharma & Kachroo, 1981: 111); Poonch, Mirpur. Kashmir: Jhelum Valley road; Barsala (Stewart, 1972: 476). 4. Abutilon indicum (L.) Sweet, Hort. Brit. 1st Edition. 54. 1826; Masters in Hook. f., Fl. Brit. India 1: 326. 1874; Hu, Fl. China, Mal- vaceae. 32. 1955; Ngwe, Union Burma J. Life Sci. 4: 207. 1971; Stewart in Nasir & Ali, Ann. Cat. Vasc. Pl. W. Pak. & Kashm. 476. 1972; Riedl in K. H. Rechinger, Fl. Iran. 120: 6. 1976; Abedin in Nasir & Ali, Fl. W. Pak. 130: 69. 1979; Sharma € Kachroo, Fl. Jammu 1: 111. 1981. Sida indica L. in Tor- ner, Cent. Pl. 2: 26. 1756. LECTOTYPE: Linn. Herb. no. 866/29 (LINN). Sida Lid a sem "e cu ag 1: 7. 1783. Abutilon Wight ie Arn. ex Masters in Hook. f., Fl. Brit. India 1: 326. 1874. A. indicum var. microphyllum Hochr., Annuaire Con- serv. . Bot. Genéve 6: 20. 1902. A. badium Husain & Baquar, Phyton 15: 229. 1974. Perennial, suffruticose herbs to subshrubs, ca- nescent with stellate hairs; branches and petioles generally purple on one side. Leaves broadly ovate, the base cordate, the apex acute or acuminate, the margin coarsely dentate, stipules linear. Flowers solitary, axillary; pedicels longer than petioles. Calyx 5-lobed, not or slightly accrescent in fruit; lobes lanceolate to ovate, mu- cronate, ultimately reflexed. Corolla orange-yellow to yellow, 2.5-3.5 cm diam.; petals 1-1.5 cm long, obovate. Staminal tube 5-8 mm long, stellate-hir- sute. Fruit ovoid-truncate; mericarps (14-)15-20, very short-awned, erect at maturity. Seeds 3 per mericarp, minutely stellate-pilose. sometimes 3-angular; Distribution. Tropics and subtropics of the New and Old worlds. Confined to the Jammu region in our area. Additional specimens examined. INDIA. JAMMU: Ram Nagar forest, common along hedges; erect shrubs, pro- fusely branched, flowers orange-yellow, B. M. Sharma 110 (KASH). KASHMIR: Falconer 273 (fide Stewart, 1972: 6). According to Ngwe (1971), the bark of this plant is used as an anthelmintic and the roots as a di- uretic. The seeds are utilized in the treatment of piles and coughs, and as emollients and demulcents. The leaves are also said to be medicinal. 12. SIDA L., Sp. Pl. 683. 1753; Gen. Pl. Sth Edition. 306. 1754 About 150 species distributed in the tropics and subtropics of both hemispheres; represented in our area by five species. 1520 Annals of the Missouri Botanical Garden KEY TO THE SPECIES OF SIDA IN JAMMU AND KASHMIR STATE la. Branches, petioles, ima and calyx pubes- cent with stellate hairs mixed with simple, eaves cordate, palmately nerved; calyx 4-5 mm long; fruit 3-4 mm diam.; mer- icarps s. muticous Won pilose mucro ...... LS 2b. Leaves usually ovate, suborbicular-subcor- e nninerved; calyx jam.; meri- carps 9-10, anos with apical awns 3- 5 mm long covered with stiff reflexed hairs S. s lb. Branches, petioles, p and calyx pubes with stellate hairs 3a. Leaves ovate- Rose. calyx tube subangu- lar; corolla white, 1-1.5 cm diam.; meri- carps 7-8, cor pini awns connivent, | $ inflexed; seeds farinose ................ . S. ovata 3b. Leaves elliptic, obovate, oblong or rhom- boid; c ay pie angular; corolla yel- low, up to ala mericarps 5, mem branous; awns not connivent, erect; seeds 6-5 cm; flowers sublasciculate pedicel in fruit up to m lon igh genns cw stellate- S all o . S. 4b. Leaves l- 2 x : solitary (- dept pedicels in fruit i- m long; mericarps pubescent at apex only 3. S. alba 1. Sida cordata (Burm. f.) Borssum Waalkes, lumea 14: 182. 1966; Stewart in Nasir & Ali, Ann. Cat. Vasc. Pl. W. Pak. & Kashm. 483. 1972; Abedin in Nasir & Ali, Fl. W. Pak. 130: 77. 1979; Sharma & Kachroo, Fl. Jammu 1: 109. 1981. Melochia cordata Burm. f., Fl. Indica 143. 1768. TYPE: (C). Sida veronicifolia v Encyl. 1: 5. 1783; Hu, Fl. China, s . 23. 1955. S. dus var. ve ronicifo olia m.) Masters i in Hook. f., Fl. Brit. bids iS EEE E S. radicum Cur. veronicifolia Lam. IM humilis (Cav. ) K. Sehu. : Fl. Bras. 12(3): 320. 1891. S. humilocularis L' Hér., Sr». Nov. 1: 117. bis 56. 1789 Perennial, Branches, petioles, pedicels, and calyx stellate-pu- bescent and villose with long, spreading hairs. Leaves cordate, palmately nerved, stellate-hispid on both surfaces, the hairs adaxially often simple, the apex prostrate to procumbent herbs. acuminate, the margin evenly crenate to serrate. Flowers axillary, solitary, fasciculate or subpanicu- late; pedicels 1.5-2.5 cm long, in fruit up to 3.5 cm. Calyx 4-5 mm long; lobes deltoid, acuminate. Corolla yellow, 7-10 mm diam., slightly exceeding the calyx. Staminal tube 2-3 mm long, hirsute with simple hairs. Fruit depressed-globose, 3-4 mm iam.; mericarps 5, muticous, membranous, trun- cate and pilose at apex. Seeds glabrous. Distribution of the world. Confined in our area to the Jammu region. Tropical and subtropical regions Additional specimens examined. INDIA. JAMMU: Ga- jansu, common; herbs with long trailing branches among hedges, B. M. Sharma 569 (KASH); Jammu & Kashmir road, R. R. Stewart s.n. (RAW). A distinctive species distinguishable from all oth- er local species by its cordate leaves with palmate venation and by its mucronate mericarps. The species shows great variation in its indumentum and inflorescence. At least two taxa, Sida veron- icifolia and S. humilis, are occasionally recognized on the basis of variation in the inflorescence. In the former the flowers are described as fasciculate or subpaniculate, whereas in the latter they are solitary. However, a close study of herbarium spec- imens has shown variation in the inflorescence to be continuous, so any division on this basis is not justified. Similarly Cavanilles (Class. Diss. Dec. 5: 277. 1788) distinguished Sida multicaulis Cav. and S. humilis on the basis of stem pubescence. The former has tomentose-canescent stems, while the latter has scabrous ones. However, Masters (1874) did not see this difference in the Indian material and, accordingly, interpreted S. multi- caulis as synonymous with S. humilis. 2. Sida yunnanensis Hu, Fl. China, Malvaceae. 16, tab. 16, f. 7. 1955; Abedin in Nasir & Ali, Fl. W. Pak. 130: 79. 1979. TYPE: China. Yunnan: Forrest 11088 (K, BM). Sida obovata Wallich, Cat. no. 1864. 1828, nom. nud. S. rhombifolia var. obovata Wallich ex Masters in Hook. f., Fl. Brit. India 1: 324. 1874. Perennial, suffruticose herbs or undershrubs; al- most all parts stellate-pubescent. Leaves variable; elliptic, oblong, oblong-elliptic, rhomboid or ob- ovate, 1-6.5 x 0.6-5 cm, penninerved, the base obtuse or cuneate, the apex obtuse or acutish margin serrate-crenate, the abaxial surface velu- tinous, adaxially green and glabrescent. Flowers subfasciculate, axillary and terminating the upper branches; pedicels 3-6 mm long, in fruit up to 1 cm. Calyx 4-6 mm long; lobes triangular, acu- minate, carinate at base. Corolla yellow, + 1 cm diam., slightly longer than calyx. Staminal tube ca. Volume 75, Number 4 1988 Naqshi et 1521 Mc 5 Jammu & Kashmir State 3 mm long, sparsely hirsute or subglabrous. Fruit depressed-globose, 4-5 mm diam.; mericarps 5, membranous, densely stellate-pilose, birostrate with convergent apical awns less than 1 mm long. Seeds glabrous. Burma, China, India, and Pa- kistan. [n our area confined to the Jammu region, where it is common along roadsides in the Nagrota- Salora area, in orchards at Tilo Talab, Udhowalla, and elsewhere. Distribution. Additional specimen examined. INDIA. JAMMU: andni, common on roadsides; erect shrubby perennial with yellow flowers, leaves whitish-gray beneath, B. M. M akin 688 (KASH). In having subfasciculate flowers and pubescent birostrate mericarps, this species is close to Sida spinosa L. However, the broad elliptic or obovate leaves with very short petioles of the former can easily be differentiated from the ovate-lanceolate, long-petiolate leaves of the latter species. The plants are said to yield good bast fiber. The roots are used to relieve rheumatism. 3. Sida alba L., Sp. Pl. 2nd Edition. 960. 1753; Stewart in Nasir & Ali, Ann. Cat. Vasc. Pl. W. Pak. & Kashm. 483. 1972; Abedin in Nasir & Ali, Fl. W. Pak. 130: 81. 1979. TYPE: Linn. Herb. no. 866/2 (LINN). Perennial, suffruticose, stellate-pubescent herbs. Leaves small, elliptic to obovate, 1-2 x 0.5-1.2 cm, penninerved, evenly crenulate-serrate. Flow- ers mostly solitary, sometimes paired; pedicels 5— 6 mm long, in fruit 1-2 cm. Calyx 4-6 mm long; lobes triangular, acute to acuminate. Corolla yel- low, 6-10 mm diam., slightly longer than calyx. Staminal tube 2-3 mm long, hirsute. Fruit de- pressed-globose, 4-6 mm diam.; mericarps 5, membranous, pubescent at apex, birostrate with convergent apical awns less than 1 mm long. Seeds glabrous. Distribution. India, Pakistan, and China. In our area apparently confined to Pakistan-occupied area in Jammu Province. Stewart (1972: 483) reported Sida alba from Mirpur, Poonch, in Jam- mu. The S. alba of Sharma & Kachroo (1981: 110) has turned out to be 5. yunnanensis. The differences between Sida alba, S. spinosa, and S. alnifolia L. seem to be few. In fact, Riedl (1976) cited S. alba and S. alnifolia as synonyms % S. spinosa. However, a thorough study of the of these three species, together with field E O and analysis of enough herbarium material from different geographical regions is needed to gain an insight into their ranges of vari- ation and to ascertain their relationships. 4. Sida cordifolia L., Sp. Pl. 684. 1753; Mas- ters in Hook. f., Fl. Brit. India 1: 324. 1874; Hu, Fl. China, Malvaceae. 25. 1955; Ngwe, Union Burma J. Life Sci. 4: 207. 1971; Stew- art in Nasir & Ali, Ann. Cat. Vasc. Pl. W. Pak. & Kashm. 483. 1972; Riedl in K. H. Rechinger, Fl. Iran. 120: 3. 1976; Abedin in Nasir & Ali, Fl. W. Pak. 130: 83. 1979. TYPE: Linn. Herb. no. 866/12 (LINN). Sida herbacea Cav., Diss. 1: 19, tab. 13, f. 1. 1785. S. holosericea Willd. ex Sprengel, Syst. Veg. 3: 112. 1826. Perennial, suffruticose herbs; branches, petioles, pedicels, and calyx stellate-hispid and long-villose with simple, spreading hairs ca. 3 mm long. Leaves usually ovate, suborbicular-subcordate or lanceo- late, penninerved, the base subcordate or rounded, the apex obtuse or acute, the margin irregularly crenate, uniformly stellate-hirsute on both sur- faces, the hairs longer beneath. Flowers axillary, solitary or paired or fasciculate; pedicels 4-7 mm long, in fruit up to 2 cm. Calyx 5-10 mm long; lobes triangular or deltoid, acute to acuminate. Corolla yellow, 1-1.3 cm diam., slightly longer than calyx. Staminal tube 3-5 mm long, hirsute. Fruit subdiscoid, 5-8 mm diam.; mericarps 9-10, stellate-strigose at apex, birostrate with divergent apical awns 3-5 mm long covered with stiff reflexed hairs. Seeds glabrous except hilum. Distribution. tropical countries. Confined to the Jammu region in our area. Jammu: Rajouri (Stewart, 1972: 483); Jammu (Sharma & Kachroo, 1981: 110; Lambert, 1933: 3) Common in tropical and sub- The plant is said to be used as a tonic, emollient, and astringent. The bark is considered useful in urinary troubles. 5. Sida ovata Forskal, Fl. Aegypt.-Arab. 124. 1775; Stewart in Nasir & Ali, Ann. Cat. Vasc. Pl. W. Pak. & Kashm. 483. 1972; Riedl in K. H. Rechinger, Fl. Iran. 120: 4. 1976; Abedin in Nasir & Ali, Fl. W. Pak. 130: 86. 1979. TYPE: Arabia, Surdud, Forskal (C). Sida rie Guill. & Perr. in Guill. et al., Fl. Seneg. Ten 1830; Masters in Hook. f. Fl. Brit. bus I E 1874 Perennial, suffruticose herbs with all parts stel- late-pubescent. Leaves ovate-oblong, penninerved, 1522 Annals of the Missouri Botanical Garden the base cuneate or rounded, the apex obtuse, the margin + entire towards base, obtusely crenate or crenate-serrate elsewhere, stellate-pubescent on both surfaces. Flowers axillary, solitary or paired; pedicels 5-8 mm long, in fruit 1(-1.5) cm. Calyx 4-6 mm long, the tube slightly angular; lobes tri- angular-deltoid, acuminate. Corolla white, 1-1.5 cm diam., slightly longer than calyx. Staminal tube to 3 mm long, pubescent. Fruit depressed-globose, 3-5 mm diam.; mericarps 7-8, coriaceous, retic- ulate toward the margins, glabrous except awns; awns 2, connivent, ca. 0.5 mm long, + inflexed. Seeds farinose. Distribution. Im drier parts of Africa, Arabia, India, and Pakistan. In our area reported from Mirpur (Pakistan-occupied area) in Jammu Prov- ince. Note the following reports: Jammu: Poonch, Mirpur (Lambert, 1933: 3; al 1972: 483; Sharma & Kachroo, 1981: 110). Sida spinosa L., Sp. Pl. 683. 1753, has been reported from Jammu by Lambert (1933). This is probably based on misidentification, because Shar- ma 688 (KASH) under this name turned out to be Sida yunnanensis. 13. MALVASTRUM A. Gray, Mem. Am. Acad. Arts n.s. 3: 21. 1849, nom. conserv. Fourteen species (Hill, 1982), distributed in tropical and subtropical America and in Australia; one species is known from our area. Malvastrum coromandelianum (L.) Garcke, Bonplandia 5: 297. 1857; Hu, Fl. China, Mal- vaceae. 11. 1955; Stewart in Nasir & Ali, Ann. Cat. Vasc. Pl. W. Pak. & Kashm. 481. 1972; Riedl in K. H. Rechinger, Fl. Iran. 120: 36. 1976; Abedin in Nasir & Ali, Fl. W. Pak. 130: 89. 1979; Sharma & Kachroo, Fl. Jammu 1: 109. 1981. Malva coroman- deliana L., Sp. Pl. 687. 1753. LECTOTYPE: Linn. Herb. no. 870/3 (LINN). Malva carpinifolia Desr. in Lam., Encycl. 3: 754. 1789. Malva tricuspidata Aiton, Hort. Kew. 2nd Edition. 4: 11. Malvastrum tricuspidatum (Aiton) A. Gray, Plantae Wright. 1: 16 52; Masters in Hook. f., Fl. Brit. India 1: 321. 1874. Erect to suberect, herbaceous or suffruticose, sparsely pubescent plants with simple and 4-fid, appressed hairs. Leaves ovate to lanceolate-oblong, coarsely serrate, acute or obtuse; stipules linear- lanceolate, acuminate. Flowers axillary, solitary or fascicled, sessile or with pedicels to 5 mm long. Epicalyx segments 3, linear, persistent. Calyx 5-lobed; lobes ovate, acuminate. Corolla yellow, 1.5-2 cm diam.; petals obliquely obovate, pubes- cent at base. Staminal tube 2-4 mm long, glabrous, antheriferous only at the apical end. Fruit discoid, ca. 6 mm diam.; mericarps 8-14, reniform, sparse- ly stellate-pilose, tricuspidate with an apical and 2 dorsal awns at the middle, l-seeded. Seeds reni- form, glabrous. Distribution. Native to North America, dis- tributed in tropical regions of both New and Old worlds. In our area it occurs in the Jammu region only. Lambert (1933: 3) reported the species from Billawar, Jammu, Mirpur. Additional specimens examined. INDIA. JAMMU: Ri- hari, common in waste places; semierect gregarious herbs with yellow flowers, B. M. Sharma 453 (KASH); Poonch, Bufliaz, 2,000 m, G. Singh & H. Kiran 775. 14. URENA L., Sp. Pl. 692. 1753; Gen. PI. 5th Edition. 309. 1754 About six species distributed in the warmer re- gions of both hemispheres; represented in our area by a single species. Urena lobata L., Sp. Pl. 692. 1753; Masters in Hook. f., Fl. Brit. India 1: 329. 1874; Hu, Fl. China, Malvaceae. 73. 1955; Ngwe, Union Burma J. Life Sci. 4: 206. 1971; Stewart in Nasir & Ali, Ann. Cat. Vasc. Pl. W. Pak. & Kashm. 484. 1972; Abedin in Nasir & Ali, Fl. W. Pak. 130: 92. 1979; Sharma & Kach- roo, Fl. Jammu 1: 111. 1981. LECTOTYPE: Linn. Herb. no. 873/ 1 (LINN). Urena monopetala Lour., Fl. Cochinch. 418. 1790 U. — TNT Diosa Vidensk.-Selsk. Nat. -Math. Afd. 4 829. + stellate- tomentose plants. Leaves stellate-pubescent on both surfaces, densely so and + tomentose abaxially, evenly serrate, with a basal nectary on 1-3 middle nerves beneath, variable in shape and size: the Erect, herbaceous to suffruticose, lower leaves usually subcordate or suborbicular, angled or shallowly 3-lobed at the apex; the middle leaves ovate; the upper leaves ovate to ovate-el- liptic; stipules linear, caducous. Flowers axillary, usually solitary or in fascicles of 2-3; pedicels 2- 3 mm long, in fruit up to 5 mm. Epicalyx segments 5, connate basally, linear-lanceolate. Calyx 5-part- ed, almost as long as the epicalyx; lobes ovate or ovate-lanceolate, keeled. Corolla pink with a darker center, ca. 2 cm diam.; petals obovate, 1-1.5 long. Staminal tube ca. 1.5 cm long, antheriferous in the apical part; anthers subsessile. Carpels 5, cm Volume 75, Number 4 1988 Naqshi et al. 1523 Malvaceae of Jammu & Kashmir State style branches and stigmas 10. Fruit subglobose, 1 cm or less diam.; mericarps 5, triangularly ob- ovoid, coriaceous, stellate-pubescent and glochidi- ate-spiny, shortly awned. Seeds pubescent to gla- brescent. Distribution. Tropical regions of both hemi- spheres. Confined to the Jammu region in our area. Udhampur (Stewart, 1972: 484) Additional specimen examined. INDIA. JAMMU: Udhowala Ashram, common in orchards; erect shrubs, 1.2 m high, with pink flowers, leaves variable, B. M. Sharma 111 (KASH). A variable species sometimes divided into a num- ber of varieties. Hu (1955), mainly on the basis of leaf shape and the nature and density of hairs, recognized five varieties from China. As we have very little material at hand, no attempt is made to segregate the various varieties at present. An important fiber-yielding plant. The bast-fiber from the stems is said to be more lasting than jute. In Cuba, Madagascar, Nigeria, and Brazil it is cultivated for making coffee sacks. The roots are used in rheumatism, and the twigs are chewed for relieving toothache. 15. SIDALCEA A. Gray, in Benth. Pl. Hartw. About 35 species distributed in western North America, chiefly in California and Oregon; one species has been recorded from our area. Sidalcea neomexicana A. Gray subsp. thur- beri (Robinson ex A. Gray) Hitchcock, in Univ. Wash. Publ. Bot. 18. 1957. S. par- viflora Greene var. thurberi Robinson ex A. Gray, Syn. Fl. N. Amer. 11: 305. 1897. TYPE: described from Santa Monica, Los Angeles County, California, U.S.A. Perennial, erect, glabrescent-glabrous herbs, glaucous throughout. Leaves orbicular, palmately lobed or divided; stipules small, deciduous. Inflo- rescence terminal, racemose; bracts mostly bifid. Flowers usually polygamodioecious, the pistillate flowers smaller than the perfect or staminate ones. Epicalyx absent; pedicels and calyces stellate-pu- bescent. Ca long; lobes triangular- ovate, acuminate. Corolla rose-purple, 10-16 mm long. Staminal tube biseriate, glabrous except for a few retrorse hairs on the apical portion. Mericarps beaked, reticulate on angles, smooth on back, in- dehiscent, 1-seeded. Distribution. From southern Monterey Coun- ty (Jolon) to eastern San Bernardino County, Cal- ifornia, U.S.A Additional specimen examined. INDIA. KASHMIR: Srinagar, Emporium Garden, in E lands, an escape y from cultivation, 4/5/1967. G. N. Javeid 6 (KASH). The species is restricted to America, and its occurrence in Kashmir is interesting. It is probable that seeds were brought and raised ornamentally in our area in the 1960s. We have a single spec imen identified as “S. malvaeflora A. Gray." The specimen comprises a single upper leaf Re inflo- rescence and has been determined by Paul A. Fryxell, as S. neomexicana subsp. thurberi. LITERATURE CITED ABEDIN, S. 1979. Malvaceae. Pp. 1-107 in E. Nasir & S. I. Ali (editors), Flora of West Pakistan. Karachi. Flora of Dehra Dun. CSIR, New Basu, C. R. elhi. Boissier, E. 1867. Flora Orientalis. Basileae. 1: 816- bisce WAALKES, J. VAN. 1966. Malesian Malvaceae evised. Blumea 14: 1-213. Bin IMAN, F. 1971. The application of tegumentary stud- ies A taxonomic and p po problems. Ber. Deutsch. Bot. Ges. 84: 169- COOKE, T 8. The Flora Fd ns Presidency of Bom- bay (Repr. E 1: 110-12 CORNER, E. J. 1967. The Seeds of Dicotyledons, Volumes I p^ II. sd : Dar, G. H. € P. KacHroo. 1982. Plants of Karnah (Kashmir, rs + Econ. Taxon. Bot. 3: 695- . DHAR. 1983. Weed flora of cultivated e Da Srinagar, Kashmir valley. Trop. Res. 1: 167-17 — — — , Vir JEE, P. een & G. M. BurH. 1984. Ethnobotany `of rwr ae Sind valley. J. Econ. 1988, Alpine Flora of Kash- imalaya. Saale Publishers, Jodhpur, India. Dui F. 1903. Flora of the Upper Gangetic Plains, . R. s 1969. Embryology of Thespesia lampas Dalz. & Gibs. syn. T. macro 968. The typification and application of the Linnaean binomials in Gossypium. Brittonia 20: 378-386. d pis pe 1791. De Fructibus et Seminibus Plan- tar : 249-253 e T. s 1957. The I of the Presidency of Madras (Repr. ed.) 1: Hii, S. R. 1 A sais of the genus Malvas- trum. Rhodora 84: 1-83, 159-264, 317-409 Hitcucock, C. L. 1957. A study of the perennial species of Sidalcea, Part I. Taxonomy. Univ. Wash. Pub 8: 1-79, HOCHREUTINER, B. P. G. 1900. Revision du genre Hi- iscus. Annuaire Conserv. Jard. Bot. Genéve 4: 23- — — n". 1924. Genres nouveaux et discutées de la famille des Malvacées. Candollea 2: 79. 1524 Annals of the Missouri Botanical Garden Hu, Seg 1955. Malvaceae. /n Flora of China, fam. 153. Arnold Arboretum of Harvard University. ILLJIN, M. M. 1949. Malvaceae. Pp. 23-170 in B. K. Shishkin & E. G. Bobrov (editors), Flora of the U.R.S.S., Volume 15 (treatment of Gossypium by Ya. I. Prokhanay; pp. 170-184). [Translated from Russian into English by N. Landau, Jerusalem, Sis .] Javerp, G. N. 1970. Flora of Srinagar: A Phyt graphic and Systematic Study of the Urin P of Srinagar. Ph.D. Thesis, = Universit A. R. NAQsHI. Flora of e UD camps (Kashmir), I. J. Sci. Univ. Kash. 1(1-2): KITAMURA, s, 1960. Flora of Afghanistan. Kyoto Uni- versity, Japan. Kumar, P. 1981. Studies on the Structure and De- velopment of Seed Coat E Malvaceae. Ph.D. Thesis. opal University, Bho LAMBERT, W. J. 1933. List t trees and shrubs for the 2 mir and Jammu dos circles, Jammu & Kash- r State. Forest Bull. n 0. ES M. T. i Mina In: J. D. Hooker, Flora of British India 1: 317-353. Mepicus, F. C Uber einige Kunstliche Geschlechter aus der Malvenfamilie . MoENCH, C. 1794. Methodus Plantas Horti Botanic et Agri Marburgensis, a Staminum Situ Describendi. is m R. R. 1978. Seed and fruit anatomical s in Malvaceae T. Thespesia populnea. Phy- 4. P. KacuRoo. 1982. Floristics. Pp. 29-50 in P. Kachroo (editor), eran on Ladakh Survey. University of Kashm Newer, T. M. 1 The Burmese Malvales. Union Burma J. Life Sci. 4: 196-208. Prain, D. 1903. Bengal Plants 1: 262- 269. Raksurr, S. C. & B. C. Kunpu. 1970. Revision of the e species of Hibiscus. Bull. Bot. Surv. India 12: -175. lu S., P. C. Josur & N. S. Punpir. 1966. Seed development i in Gossypium L. Indian Cotton J. 20: 97-106. REEVES, R. G. . Comparative anatomy of the seed of cotton and other malvaceous plants. I. Malveae and Ureneae, II. Hibisceae. Amer. J. Bot. 23: 291- 296, 394-405 RIEDL, I. 1976. Epo In: K. H. Rechinger, Flora lanes: No. 120. Akademische Druck- u. Verla sanstalt, Graz, 1 RoxBurcH, W. 1832. Flora Indica. 3: 522-530. SCHNARF, K. . Vergleichende Embryologie der An- in. 1982. Flora of Jammu and Plants of Neighborhood, Volumes I and II. Bishen Singh Mahendra Pal Singh, Dehra Dun. SiNcH, B. 1967. The structure and od e Abelmoschus moschatus Medic. seed. Phyton phalogy 1 — & D H. 1986. Development and structure of seed ea in Malvaceae — II Thespesia populnea : 11-15. H. 1985. kre aan and structure of seed coat in Malvastrum A. Gray, Malachra L. and Kydia Roxb. Geobios New Rep. 4: 7-13. SiNcH, G. & P. KacHRoo. 1976. Forest Flora of Sri- nagar and Plants of Neighbourhood. Bishan Singh Mahendra Pal Singh, Dehra Dun. STEWART, R. R. 72. In: E. Nasir & S. I. Ali ia n Annotated Catalogue of the Vascular Plants West Pakistan and Kashmir. Fakhri Press, set VENKATA Rao, C. 1954. Embryological studies in Mal- vaceae I. Development of cnn Proc. Natl. Inst. Sci. India 20: 127-150. . Embryological studies in Malvaceae II. Fertilization € seed development. Proc. Natl. Inst. Sci. India 2 : 93-67. 1959. Geology of India ed. s. Macmillan, M iris u Wu T.C. 1979. ‘Malvaceae. In: H. Hara & L. H. J. Williams (editors), An Enumera tion of the Flow- ering Plants of Nepal, Volume 2: 73. London. Winter, D. M. 1960. The Seveopmen: of seed of Abutilon theophrastii. 1. O ) . Seed coat. Amer. J. Bot. 47: 8-14, scene WUNDERLICH, R. 1967. So the nomic significance of seed coat. olaa -311 17: 30 ZOHARY, M. 1963 Taxonomical studies in Alcea of southwestern Asia. Part I. Bull. Res. Council Israel 11D: 210-229. Part II. Israel J. Bot. 12: 1-2 THREE NEW SPECIES OF NECTANDRA FROM THE VENEZUELAN GUAYANA’ Jens G. Rohwer? ABSTRACT e new species f Nectandra from the Venezuelan Guayana are described in advance of a complete E of the genus During work towards a monograph of the gend Nectandra Three of isis occur within hai area of the Pon of the Venezuelan Guayana, currently in prepa- ration at the Missouri Botanical Garden. Since the treatment of the Lauraceae for this flora is sched- uled to appear before the monograph of Nectandra will be completed, the new species are published here in anticipation of the more complete work. All three new species are somewhat similar to Nectandra globosa (Aubl.) Mez, and two of them seem to be closely related to it. They differ from N. globosa and from each other mainly by striking characters of the indumentum in addition to less conspicuous differences in venation, leaf shape, flowers, and fruits. While any one of these char- acters alone would hardly warrant recognition on the specific level, their combination makes the en- tities described below quite distinct. Nectandra aurea Rohwer, sp. nov. TYPE: Ven- ezuela. Apure: Dtto. San Fernando, mouth of the Rio Arauca at its intersection with the Rio Orinoco, 66°36'W, 7?24'N, 35 m, 14-15 May 1977 (fl, immature fr), Davidse & Gon- zález 13215 (holotype, MO). Figure 1 ad 15 m alta. ides novelli + dd vel longitudinaliter exarati, dense rufo- vel aurantio-tomen- res teretes, dense pete demum r ceolato-elliptica, (7-)9-17(-19) cm longa, 2.8-5(-6) cm lata, wataq lateralibus utroque costae latere 5-7(-9) sub angulo 35-50? prodeuntibus, in sicco rigide chartacea vel subcoriacea, laevia, apice in acumen angustata, basi acuta b)plano, folia novella supra + dense et molliter (sed. o inconspicue) flavido-tomentella, subtus dense et + a nervorum rufo-tomentosa, vetustiora tam leviter pubescentem glabrescentia, subtus indumento fere persistente vel tarde deminuente sed demum sparso. Inflorescentiae axillares, breviter thyrsoideae, 2-5(-9) cm longae, pedunculo 1-2(-3) cm longo, dense rufo- vel aurantio-tomentosae, cymis lateralibus 2-4, 1 -3(-4)-fur- catis. Pedicelli 0.5-2.5 mm longi. Flores 5-8(-9.5) mm diametro. Tepala elliptica, intus papillosa, exteriora in- terioribus paulo majora, ad 4 mm longa. Stamina papillosa, subsessilia, 0.7-0.9 mm longa, exteriores 6 late pentagona adpresse pubescens. Fructus (non nisi i ovato-ellipticus. Cupula profunda, campanulata, iam ca. 11 mm diametro et 9 mm profunda, margine integro Tree to 15 m. Twigs at first + angular or lon- gitudinally furrowed, densely covered with curled reddish longer hairs (0.5-1.5 mm long) over an extremely dense layer of very short yellowish hairs, older twigs roundish with the short hairs persistent for a long time and becoming grayish. Leaves al- ternate, lanceolate or rarely lanceolate-elliptic, (7-)9-17(-19) em long, 2.8-5(-6) cm wide, with 5-7(-9) pairs of lateral veins diverging at 35—50^, dried laminae stiffly chartaceous to subcoriaceous, smooth (secondaries slightly prominent below), the apex tapering towards a narrow acumen, the base acute, flat or nearly so, young leaves above + Most a the e for this paper was conducted at the Missouri S cal Garden during a long-term visit in 1986—1 uthor is most grateful to the director and the st ‘of this institution for friendly reception and continuous ps The visit was made possible through a das toral scholarship grant from the Deutsche Forschungsgemeinschaft. ? Institut für Allgemeine Botanik, Ohnhorststrasse 18, D-2000 Hamburg 52, Federal Republic of Germany. ANN. Missouni Bor. GARD. 75: 1525-1532. 1988. 1526 Annals of the Missouri Botanical Garden FIGURE l. Nectandra aurea.— A. bit, flowering branch.—B. Habit, branch.—C. Flower, seen obliquely from above. —D. Stamen of = first whorl, seen from inside.—E. Stamen of the meas whorl, seen from inside. —F. Stamen of the third whorl, seen from outside.—G. Receptacle with ovary, staminode, and two inner stamens, cut open, tepals and outer stamens removed.—H. Fruit. Vouchers: A, Davidse & González 14492; B-H, Davidse & González 13215. Volume 75, Number 4 1988 Rohwer 1527 Nectandra densely yellowish pubescent (often inconspicuous but soft to the touch), below densely and often shining golden-pubescent, hairs predominantly ap- pressed but with a few + erect reddish hairs, these rarely forming indistinct tufts in the vein axils, older leaves above glabrescent except for some hairs on midrib, below with an almost persistent indumentum, or very slowly subglabrescent. Inflo- rescences short thyrsoids, axillary to distal leaves or occasionally some of them axillary to cataphylls below the terminal bud, covered with the same kind of indumentum as the young twigs, 2-5(-9) cm long with a peduncle of 1-2(-3) cm, with 2-4 lateral cymules below a terminal cluster of cymules, the cymules branched 1-3(-4) times. Pedicels 0.5- 2.5 mm long. Flowers 5-8(-9.5) mm diam. Tepals elliptic, up to 4 mm long, papillose on the inside, the outer ones slightly larger than the inner ones. Stamens papillose, subsessile, 0.7-0.9 mm long, the outer 6 broadly pentagonal to broadly rhombic, with a slightly prolonged, broadly triangular, obtuse to almost rounded apex, those of the first whorl (opposite the outer tepals) usually slightly wider than those of the second whorl, the 3 inner stamens broadly rounded to almost truncate at the apex. Staminodes small but thickish, papillose at the tip, hidden between the stamens of the third whorl and united with them at the very base. Pistil + slender, glabrous, 1.5-2 mm long, the style relatively thick and about as long as the ovary. Receptacle deeply urceolate, on the inside covered with tightly ap- pressed hairs (sometimes only in patches). Fruit (known only immature) ovoid-elliptic, its cupule deep, campanulate, already about iam. and 9 mm deep, with an entire margin. ] mm Additional Med examined. VENEZUELA. APURE: Dtto. San Fernando, banks of the Rio Arauca, 5 km n tly (in a E line) SW of El Faro, 7?19'N, 54' W, 35 m -19 May 1977 (fl), Davidse & Gon- D. 13412 Dio. Dtto. Pedro Camejo, banks of the Rio bii eat B a Pantallo, 48 airline km NE of Puerto Páez, 3'N, 67?09' W, 40 m, 24-25 Feb. 1978 (fl). Davidse pu Gonz vis 14492 (MO). BOLÍVAR: Dtto. Cedeño, + 20 km al E de Türiba, 6-11 Dec. 1970 (fl), Marcano-Berti 2595 (MER); margen del Caura, Tem- blador, 100 m, 28 Mar. 1939 (fl), Ll. Williams Pp F, US); a n 80 m, 11 June 1984 (fl, 1 mature fr), S. López P. et al. 4598 (NY); same datu. (fl), S. López P. i p^ 4664 (NY). — Nectandra aurea is a small to medium-sized tree of gallery forest formations on the the Orinoco River and its tributaries. It is easily recognized by the characteristic pubescence on young leaves, young twigs, and inflorescences. The anks of tertiary venation is rather inconspicuous above and below, and the young leaves tend to dry with a much lighter and more yellowish color than the mature leaves, which dry medium brown above. The deep campanulate cupule is also quite unusual, but it seems to become wider towards maturity and therefore may not be a reliable specific character. Allen (1964) cited the collection Williams 11612 under N. pisi Miq. (= N. globosa (Aubl.) Mez). With only this specimen at hand, one might indeed think of it as an unusually hairy variant of N. globosa. In that species, however, the leaves are generally more elliptic and much wider when ma- ture, the veins are more distinct, and the immature cupules are constricted, not expanded towards the margin. Nectandra fulva Rohwer, sp. nov. TYPE: Vene- zuela. Amazonas: Dep. Rio Negro, 0-1 km E of Cerro de Neblina Base Camp on Rio Ma- warinuma, 140 m, 0%50'N, 66?10'W, 27 Nov. 1984 (fl, fr), Liesner & Kral 17338 (holo- type, MO). Figure 2. Arbor, ad 20 m alta (raro frutex). Ramuli novelli + angulati, apice ipso + den nse cupreo- -subtomentosi, ce- leriter autem t nigrica ntes. Folia alterna, coal gp nine vel (o b)lanceolato- elliptica, raro elliptica, 12-26 cm longa, 3.3-8.5 cm lata, nervis as pare 33 costae latere (6-)7-9(-10) a angulo 50-55(-65)? Ee in sicco rigide char- lo i impressis venulis thyrsoideae, 7-12(-15) cm longae, pedunculo 3.5-6(-7.5) cm dice cupreo- et floribus den tog Ha (2-)4(-6), 2-4-furcatis. Pedicelli 3.5-7 longi. Flores (7 -)9-10 mm diametro. Tepala erbe st an uenis -elliptica, intus papillosa, aequalia vel exteriora interioribus aliquantum majora, ad 4.5 mm longa. Stamina papillosa, 0.9 trullata apice acuto vel parabolico, serie tertia + rectan- gularia, apice truncata vel late rotundata. Staminodia parv ula sed crassiuscula, apice papillosa, inter stamina seriei tertiae celata et basi eis conn iata . Pistillum elon- u m dia metro. Cupula + hemispherica, ca. 10 mm diametro et 5 mm profunda, margine paulo 6-dentato. 1528 Annals of the Missouri Botanical Garden FIGURE 2. Nectandra fulva.— 4. Habit. —B. Leaf base. —C. Flower, seen from side.—D. Stamen of the first whorl, seen from inside. —E. Stamen of the second whorl, seen from inside. —F. Stamen of the third whorl, seen from outside. —G. Ovary.—H. Stamens of the third whorl, with stigma and staminode between them, glands and scars of outer stamens at the periphery. —[. Receptacle with two inner stamens, cut open; tepals, outer stamens, and ovary removed; semischematic.—J. Fruit. Vouchers: A, C-J, Liesner & Kral 17338; B, Gentry & Stein 46954 Volume 75, Number 4 1988 1529 Nectandra Tree (rarely shrub) to 20 m. Twigs at first + angular, immediately below the terminal bud with a relatively dense copper-colored indumentum of very short to moderately long (ca. 1 mm) curled to almost straight hairs, these quickly becoming sparser below the tip and revealing the blackish epidermis. Leaves alternate, (ob)lanceolate-oblong to (ob)lanceolate-elliptic, B elliptic, 12-26 cm long, wide, with (6-)7-9(-10) pairs of lateral veins diverging at 50—55(—65)°, dried laminae stiffly chartaceous, above usually with the secondary veins slightly impressed and the tertiary veins very slightly raised, below with midrib and secondaries prominent and the tertiary veins no- ticeably raised, + at right angles to the secondary veins, the apex acuminate, the base acute or almost cuneate, at the very base often extended into a short projection along the petiole, margin flat; young leaves (only those developed around flowering time, see below) very densely covered above with pre- dominantly short strongly curled yellowish hairs, soft to the touch, below with a (moderately) sparse indumentum of short appressed hairs between the secondary veins and slightly longer more reddish curled hairs on the veins, the indumentum of older leaves first worn off between the secondary veins above, later slowly subglabrescent. Inflorescences + lax thyrsoids, axillary to distal leaves, 7-12 (7-15) cm long with a peduncle of 3.5-6(- 7.5) cm, this with a moderately dense to sparse indumentum similar to that of the twigs, becoming shorter, dens- er and more yellowish towards the flowers, with (2-)4(-6) lateral cymules below the terminal cy- mule or cluster of cymules, the s branched 2-4 times. Pedicels 3.5-7 (7-)9-10 mm diam. Tepals elliptic e elongate- elliptic, up to 4.5 mm long, papillose on the inside, + equal or the outer ones somewhat larger than the inner ones. Stamens papillose, 0.9-1.4 mm long, with very short filaments, these in the outer whorls sometimes adnate to the tepals, the stamens of the first whorl suborbicular to broadly rhomboid, mm long. Flowers with a rounded or obtuse tip, those of the second whorl slightly narrower, + ovate to slightly an- gular, with an acute to parabolic tip, the third whorl almost rectangular, truncate to broadly rounded at the apex. Staminodes very small but thickish, pa- pillose at the tip, hidden between the stamens of the third whorl and basally united with them. Pistil elongate, 1.7-2.7 mm long, the style papillose and about as long as the glabrous ovary. Receptacle deeply urceolate, glabrous on the inside. Fruit (known only immature) elongate to ovoid-elongate, its cupule nearly hemispherical, ca. diam. and 5 mm deep, with a tendency to develop 6 mm thick teeth on the margin, each corresponding to the midvein of a tepal. Additional specimens examined. VENEZUELA. AMAZONAS: Dep. Rio Negro, Neblina Base Camp on the io Mawarinuma, 0%50'N, 66?10'W, 140 m, 17 July 1984 (fl), Davidse & Miller 27417 (MO); same data, 120 m, 17 July 1984 (fl), Davidse 27520 (MO); Cerro Neblina, between base camp and Puerto Chimo along Río Mawarinuma, 150-180 m, 0%50'N, ca. 66?08'W, 26 Apr. 1984 (fl), Gentry & Stein 46954 (MO); upstream end of large island in Río Mawarinuma just upstream from Neblina Base Camp, 0%50'N, 66?10'W, 160 m, 27 Nov. 1984 (fl, immature fr), Kral 7 1844 (MO); 1- 3 km E of Cerro de Neblina Base Camp on Rio Mawarinuma, 140 m, 0?50'N, 66°10'W, 8 Feb. 1984 Minis fr), Liesner 15739 (MO); same data (post fl), Liesner 15755 (MO); along Rio Mawarinuma, 0-5 km E of Cerro de La Neblina Base Camp, 140 m, 0%50'N, 66?10'W, 10 Mar. 1984 (immature fr), Liesner & Funk 16521 (MO); along Rio Mawarinuma, 2- - km E of Base p between Base Camp and "Puerto Chimo" ca 160 m, 09?50'N, 66?08'W, 26 hes “1985 (fl), Thomas 3189 (MO, NY, US). Nectandra fulva is known from only one pop- ulation near the Neblina Base Camp on the Rio Mawarinuma. The tendency to develop dentate or even thickly lobed cupules is a very rare character in Nectandra, and it suggests a close relationship with the group around N. acutifolia (Ruiz & Pa- vón) Mez. The structure of the indumentum and the presence of often distinct gland dots in the leaves further support this conclusion. /Vectandra fulva differs, however, from the other species in this group (among other characters) by less elon- gate anthers and the lack of an inrolled leaf base. Nectandra fulva shows an interesting dimor- phism in the indumentum of leaves developed dur- ing the flowering and fruiting periods. Only those from the flowering period show the dense indu- mentum on the upper surface described above. In those from the fruiting period there are only some- what curled reddish hairs, moderately dense on the major veins and sparse from the beginning between them. While these sparsely hairy new leaves are developed, some of the older ones may still show remnants of the dense indumentum developed dur- ing flowering time. Nectandra ruforamula Rohwer, sp. nov. TYPE: Venezuela. Amazonas: 0.5 to 2 km N of San Carlos de Rio Negro, ca. 20 km S of conflu- ence of Rio Negro & Brazo Casiquiare, 1?56'N, 67?03'W, 120 m, 10 May 1979 (fl), Liesner 7318 (holotype, MO). Figure 3. rbor, ad 10 m alta. Ramuli novelli + angulati, dense TN tomentelli, vetustiores plerumque + teretes et pilis coactis inconspicuis fumosi, demum glab in typo 1530 Annals of the Missouri Botanical Garden autem ramuli permanifeste angulati et pubescentia persis- tente). Folia alterna, o, Ao s lanceolato- elliptica, 8 8.5- m lata, nervis lateralibus p a Hiro subcoriacea, supra €— pra interveniis glabra vel paulo pallido-tomentella, subtus indumento densiore, interveniis plerumque pilis brevibus adpressis pallidis modice sparsis et pilis d + erectis rufescentibus (ut in venis) sparsis, ventustiora supra (in- m A ele basin e glabrescentia, ra subgla- brescentia, pilis brevibus adpressis pallidis valde mini " cuis = perstemibue In Inflorescentiae ien breviter a lateralibus (0-)2 4. | De 2(-: 3) furcatis, pose lli 13 longi. Flores 5.5-7.5 mm diametro. Tepala elliptica, intus papillosa, aequalia vel exteriora deae ic paulo majora, ad 3.5 mm lor , 0.7-0.9 mm dioc filamento evidente eviore, exteriores 6 de- presse obtrullata a fere transverse elliptica, ers paulo producto obt secunda plerumque paulo intrinsecus paulo > glandulosa, extrinsecus papillosa, inter stamina seriei tertiae + celata. Pistillum crassiusculum glabrum vel stylo aliquantum papillosum, 1.4-1.8 mm longum, stylo crasso ovario breviore. Receptaculum late urceolatum (hemisphericum sed supra staminibus stami- nodiisque fere praeclusum), intus glabrum vel pilis sparsis adpressis. Fructus (immaturus, ex Maguire & Politi 28001) ellipticus. Cupula profunda, subhemispherica, iam ca. 11 mm diametro et 6 mm profunda, margine integro. Tree to 10 m. Twigs at first + angular, densely covered with a rusty red indumentum of erect and somewhat curled longer hairs (0.3-0.8 mm) over an extremely dense layer of very short hairs, older twigs usually roundish and with much of the in- dumentum persistent for a long time as a matted gray layer over the epidermis, finally glabrescent (in the type collection all twigs strongly angular and covered with a dense reddish indumentum). Leaves alternate, elliptic to lanceolate-elliptic, 8.5- 20(-26) em long, 2.7-8(-10) cm wide, with 5- 7(-9) pairs of lateral veins diverging at 35-50 (—55)°, dried laminae subcoriaceous, above grayish green (older leaves rarely olive), with impressed midrib and secondaries, the tertiary veins usually slightly impressed, below yellowish brown to reddish brown, midrib and secondaries prominent, the ter- tiary veins noticeably to distinctly raised, + at right angles to the secondary veins, the apex narrowly acuminate, the base acute or rarely obtuse but at the very base attenuate or at least extended into a short projection along the petiole, almost flat or slightly curved downwards; young leaves on both sides with erect curled reddish hairs on midrib and secondaries, the intercostal fields glabrous or with some paler hairs above, below indumentum denser both on veins and in the intercostal fields, the latter usually with pale, short, appressed hairs and with longer reddish hairs similar to those on the veins; older leaves glabrescent above (some hairs on base of midrib often persistent), subglabrescent below, with the inconspicuous, pale, short, appressed hairs persistent for a long time. Inflorescences short and + dense thyrsoids, axillary to distal leaves, covered with an indumentum similar to that of the young twigs (but becoming shorter and more grayish brown towards the flowers), 2-7 cm long with a peduncle of 1-2.5(-3.5) em, with (0-)2-4 lateral cymules below a terminal cluster of cymules, the cymules branched 1-2(-3) times. Pedicels 1-3 mm long. Flowers 5.5-7.5 mm diam. Tepals ellip- tic, up to 3.5 mm long, papillose on the inside, equal or the outer ones slightly larger than the inner ones. Stamens papillose, 0.7-0.9 mm long, the outer 6 with short but distinct filaments; anthers broadly obtrullate to almost transversely elliptic, with a slightly prolonged obtuse apex, those of the first whorl (opposite the outer tepals) usually slight- ly wider than those of the second whorl, the 3 inner stamens broadly rounded to almost truncate at the apex. Staminodes small but thickish with a small heart-shaped head, slightly glandular on the adaxial side, papillose on the abaxial side, hidden between the stamens of the third whorl. Pistil relatively stout, glabrous or somewhat papillose at the style, -1.8 mm long, the style relatively thick, reach- ing about 2—% the length of the ovary. Receptacle broadly urceolate (hemispherical but above closed by stamens and staminodia), on the inside glabrous or with a few tightly iens hairs. , after Maguire & Politi 28001, s elliptic, its cupule deep, Pd beue but slightly contracted at the margin, already about mature 11 mm diam. 6 mm deep, with an entire margin. Additional specimens examined. VENEZI AMAZONAS: Dep. Atabapo, Cucurital de Caname, oen bank of the middle part of Cano Caname, 3?40'N, -] May 1979 (fl), Davidse et al. 17 7008 (MO); fo ae 1-10 km below San Fer- nando de m 150 m, 11 May 1954 (fl), Level 65 JS); 4-7 km NE of San Carlos de Rio Negro along pond, ca. 20 kr Brazo Casiquiare, 1979 (fl), Liesner 7550 (MO); Rio Atabapo, along Yavita Pimichin trail near Yavita, 125-140 m, 10 June 1959 (fl), Wurdack & Adderley 42902 (MO, NY); Cerro Si- papo (Paráque), camp, 125 m, 28 Dec. 1948 (fr), Maguire & Politi 28001 (US). — "T oO A9 ° c Volume 75, Number 4 1988 Rohwer 1531 Nectandra FIGURE 3. Nectandra ruforamula.— 4. Habit. —B. Flower, seen from side. —C. Stamen of the first whorl, seen e.— SUR from insid outsid D. S "em Stamen of the second whorl, seen from insi —F. Staminode with glandular patches, seen from inside. —G. Receptacle with ovary, staminode, and de.—E. Stamen of the third whorl, seen from two inner stamens, cut open; tepals and outer stamens removed.—H. Fruit. Vouchers: A-G, Liesner 7318; H, Maguire & Politi 28001. Nectandra ruforamula occurs in the lowland region that connects the basins of the Orinoco and the Amazon. At present it is only known from the Venezuelan side, but it is likely to occur in Colombia as well. Nectandra ruforamula is recognized main- ly by its dense reddish indumentum on the young twigs, but the leaves have a characteristic ap- pearance, too. They usually show a marked dif- ference between a grayish green upper surface with impressed reticulation and a reddish brown lower surface with raised reticulation. Despite these rather obvious characters, there 1532 Annals of the Missouri Botanical Garden is a problem with the delimitation of N. ruforamula against /V. globosa (Aubl.) Mez. The type collection of N. ruforamula is not fully representative of the species but has been selected because it is so strik- ingly unlike any other species of Nectandra, in- cluding N. globosa. In the other collections, how- ever, the reddish indumentum (which in the type collection covers the entire twigs) is restricted to the apical part of the branchlets. Further from the tip it quickly turns into a gray matted mass in which individual hairs can hardly be resolved, and which imperceptibly intergrades with the grayish bark still further down. Where this gray indumen- tum prevails over the reddish hairs, the specimens may become similar to N. globosa, and three of the collections included here (Level 65, Maguire & Politi 28001, Wurdack & Adderley 42902) have been cited under its synonym N. pisi Miq. by Allen (1964). Nectandra ruforamula is treated as a separate entity here because in /V. globosa the tertiary reticulation is less distinct, even the youngest leaves never show erect curled reddish hairs, and the dry leaves are usually medium to dark brown on both sides. The fruiting collection, Maguire & Politi 28001, is placed here with some doubt. It shows the char- acteristic color of the leaves, but it altogether lacks the reddish hairs, and the tertiary venation is some- what less conspicuous than should be expected in leaves of this size. The indumentum, however, is lost or significantly altered with age in most species of Nectandra. LiTERATURE CITED ALLEN, C. K. Lauraceae. /n: B. Maguire et al., The Botany of the Guayana Highland, Part V. Mem. New York Bot. Gard. 10(5): 44-123 A SYNOPSIS OF MATELEA SUBG. E ASCLEPTADOIDEAE): Warren Douglas Stevens? ABSTRACT Matelea subg. Dictyanthus comprises thirteen currently known species as from northwestern Mexico to southern kasq One of the Da is provided as new, a, M. hamata, M.1 and a ke with a new combination, M. aenea, and five are a, M. macvaughiana, and M. aah The synopsis provides descriptions e described The genus Dictyanthus was described by De- caisne in de Candolle's Prodromus in 1844. The description was based on a Sessé & Mocino col. lection which had been distributed by Pavón. A few years later, sometime in the late 1840s, Dic- tyanthus pavonii was introduced into European botanical gardens and became relatively well known. During this period the species was illustrated in horticultural journals and was provided with several new names. The next treatment of the genus was that of Bentham & Hooker in their Genera Plan- tarum (1876). They considered the genus to com- prise three or four Mexican species. Six years later, Hemsley (1882) treated the genus in Biologia Centrali-Americana, Botany and recognized four species, one of which was described as new. In Engler & Prantl's Die natürlichen Pflanzenfami- lien, Schumann (1895) again considered Dictyan- thus to be a Mexican genus of three or four species. Next, Standley in his Trees and Shrubs of Mexico (1924) included six species, one of which was de- scribed as new. Woodson (1941), in providing a generic revision of the North and Central American Asclepiadaceae, reduced Dictyanthus to a sub- genus of Matelea and made new combinations for the ten species he recognized. These were listed with partial synonymy to document the subgenus. In the present treatment there are recognized seven of the species Woodson listed, two are reduced to synonymy, and one is excluded from the subgenus. To these are added five new species and one species resurrected from Woodson’s synonymy. Finally, Standley & Williams (1969) included the two Gua- temalan species of the subgenus in their treatment of Asclepiadaceae in Flora of Guatemala. This summarizes the taxonomic history of Dictyanthus, except for the description of individual species. Since Woodson’s generic revision, no species have been added to or removed from the subgenus, and no species belonging to the group have been de- scribed since atelea belong to the New World tribe Gon- olobeae, which can be distinguished within the subfamily by having the pollinia partially sterile and excavated on one or both faces and oriented more or less horizontally along the margin of the style apex. Woodson (1941) recognized three gen- era in the tribe, Matelea, Gonolobus, and Fisch- eria, reducing many previously recognized genera to synonymy. While the larger generic concepts offer significant advantages, it seems clear that reinstating some segregates would improve all the circumscriptions. Matelea is the least satisfactory of the concepts but, although certain small seg- regates, notably Macroscepis and Pherotrichis, can easily be recognized, the bulk of the species require more careful consideration than has yet been possible. Most of Woodson’s subgenera of Matelea are ill-conceived, but Dictyanthus is such a distinctive group that Woodson expressed some misgivings about submerging it. He correctly noted, however, that Matelea altatensis (Brandegee) ; Preparation of the illustrations was supported in part by NSF Grant GJ-573 to Michigan State University rde and in part by BSR- 8417791 to Missouri Botanical Ga en. ? Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166, U.S.A. ANN. Missouni Bor. GARD. 75: 1533-1564. 1988. 1534 Annals of the Missouri Botanical Garden Woodson, which he included in subgenus Dic- tyanthus, is intermediate with the larger group of Matelea. To M. altatensis, which is here excluded from the subgenus, can be added the additional intermediate species M. aspera (Miller) W. D. Stevens and M. sepicola W. D. Stevens. Subgenus Dictyanthus is separated from the rest of Matelea by having the corona deeply five-lobed with the axis of each lobe entirely adnate to the corolla; also partially diagnostic are 1) simple inflores- cences, 2) a mixed indumentum with at least some of the trichomes glandular and at least some of the long trichomes uncinate, and 3) narrowly fusiform follicles with thickish projections. Thus, while ad- equate generic and subgeneric circumscriptions within the Gonolobeae are yet to be established, Dictyanthus is an easily defined but not entirely discrete group within the current concept of Ma- telea and is here treated as such. To interpret properly the following key and de- scriptions, the following should be considered. 1) The description of the indumentum has been sim- plified and, to a certain extent, generalized by the convention of referring to all trichomes as either short, glandular, or long and by modifying these terms as appropriate. These trichomes are, unless otherwise indicated, uniseriate and multicellular and have straight or uncinate tips. Short trichomes are less than 0.1 mm long, typically about 0.05 mm. Short trichomes on the inner surface of the corolla, when present, are somewhat different in form and have a glassy appearance when dried. Glandular trichomes are the same length to slightly shorter than the short trichomes, with which they are al- most always mixed, and have a short stalk, an inflated middle, and a short apiculum. The glan- dular trichomes are probably not secretory, but the inflated part frequently collapses on drying, giving these the appearance of normal capitate glandular trichomes. Long trichomes are typically much lon- ger than 0.1 mm. The maximum length of long trichomes is given only for the stem. Trichomes on other structures tend to be shorter. When only long trichomes are present on a structure, as is often the case on the leaf blade, they often occur in two discrete lengths, giving much the same ap- pearance as mixed long and short trichomes. 2) The terminology used for describing the surfaces of the leaves and seeds is according to Stearn (1966). 3) The leaves are described essentially according to Hickey (1973). The same terminology is employed to describe the shape of the bracts, calyx lobes, and corolla lobe apices. The leaf length has been considered to be the length of the mid- rib. In all cases the leaves are described on the basis of the largest leaf of each specimen examined. The largest leaves, especially on specimens of the erect species, tend to be near the middle of the stems; the lower leaves tend to be broader and the upper leaves tend to be narrower. 4) The inflores- cence and floral characters are described only on the basis of examples in anthesis. The bracts are described on the basis of the largest bract of each inflorescence. The first bract (opposite the first flower) tends to be the largest, and the subsequent bracts are gradually reduced in size. 5) The corolla lobes are considered to be distinct from the limb; thus the corolla is composed of the tube, limb, and lobes. The descriptions of flower colors have been much simplified. In general, only the basic color pattern of the corolla has been described. This color pattern applies only to the inner surface of the corolla, and considerable care should be exercised in attempting to discern the pattern by examining the outside of pressed flowers. 6) Measurements of bipollinia are taken in lateral view and in the normal orientation they assume when removed; the depth or thickness is ignored. The length of the pollinium is measured from the point of attachment of the corpusculum to the tip, including the caudicles of most other asclepiad bipollinia; in most species of the Gonolobeae there is no sharp demarcation of the caudicles from the pollinia. In the preparation of this synopsis, 694 speci- mens of 302 collections from 41 herbaria were examined. A list of the specimens examined is available on request. Ten of the thirteen species were studied in the field and nine were cultivated. TAXONOMIC TREATMENT Matelea subgenus Dictyanthus (Decaisne) Woodson, Ann. Missouri Bot. Gard. 28: 236- 237. 1941. Dictyanthus Decaisne in de Can- dolle, Prodr. 8: 605. 1844. TYPE: Dictyan- thus pavonii Decaisne. ds eque Hasskarl, Flora 47: 258-259. 1847. TYPE: nthe suberosa Hasskarl. und ona fienda Bull. Soc. Imp. Naturalistes Moscou 25(2): 319- 320. 1852. TYPE: Rytidoloma reticulatum Turczanin Plants erect, trailing, or twining, herbaceous or woody, with or without a woody or fleshy caudex. Woody parts typically with thick, fissured, corky bark. Indumentum variable and often mixed; tri- chomes multicellular, uniseriate, simple, straight or uncinate, of 3 general types: short nonglandular, short glandular, and long nonglandular. Leaves ovate in general outline, apex mostly acuminate to attenuate, base lobate, with acropetiolar colleters; exstipulate but with an interpetiolar fringe of long trichomes and colleters. Inflorescence extra-axil- Volume 75, Number 4 1988 Stevens 1535 Matelea subg. Dictyanthus lary, a condensed, simple, helicoid cyme or reduced to a single flower with or without an apparent peduncle. Calyx 5-lobed nearly to base, with 1 or 2 colleters below each sinus within. Corolla deeply to shallowly campanulate; tube mostly convoluted, with raised parts opposite corona lobes and sacs formed between them. Corona digitately 5-lobed, lobes connate below or free, adnate to gynostegium and adnate for their entire length to corolla. Gyno- stegium stipitate, apex + pentagonal and con- KEY TO SPECIES or MATELEA SUBG. DICTYANTHUS cave to apiculate, terminal anther appendages cov- ering margin of apex. Corpusculum sagittate; caudicles winged, hardly dent from pollinia; pol- linia flattened, excavated and hyaline along upper margin, obliquely obovate. Follicles fusiform, with few to numerous, thick to thin, straight to arcuate projections. Seeds obovate, flattened, with a raised, smooth or radially grooved, entire or toothed mar- gin, the surface otherwise verrucate to rugose, pale to dark brown; with a white apical coma. la. Corolla tube with parallel vertical lines, these only occasionally with a few cross connecti 2 edicels less than 6 mm ong; corolla lobes less than 7 mm long; corona lobes libi, in — less ine 2. M. tuberosa than half the length of the corolla 2b. Pedicels more than spathulate in outline, more t 3a. in x and outer surface of corolla glabrous; corolla tube with ca. 5 lines between each corona 6. M. a lobes more than 7 mm long; corona lobes linear to linear- be. lauta 3b. Calyx and outer surface of corolla distinctly hairy; corolla tube with 15 or more lines between each corona lobe. 4a. Corolla base-sinus length more than 12.5 mm, with a narrow band of short trichomes around corona lobes within; long trichomes of peduncles and pedicels mostly uncinate; twining woody 4. vines without thickened caudices M. pavonii EN = . Corolla base-sinus length less than 12.5 mm, glabrous around corona lobes within; long trichomes of peduncles and pedicels straight; erect or weakly twining herbaceous vines with 5. M. thickened caudices macvaughiana lb. Corolla tube with circular lines or with distinct reticulations, o or without a distinct pattern. 1 5a. Corolla entirely glabrous within; corona lobes basally bracts more than 1.5 mm wide, elliptic in general shape ast on limb; corona lobes basally connate or ie but 1.5 mm wide, line 5b. bia with dense short trichomes within, at le ot forming a disk; inflorescence bracts less than . M. hamata ear or ovate in general shape. a. Corona lobes spathulate, with prominent, purple-black, deeply rugose tips; long trichomes of internodes uncinate. 7a. Peduncle less than 1 mm long; calyx lobes less than 5 mm c corolla sharply dr from tip of corona lobe to sinus; plants of the Isthmus of Tehuant suffruticosa than 5 mm long; corolla not lc dr h (sinus-sinus) ratio greater than 0.80, margins revolute, limb and lobes patent or sore reflexed; corolla densely gray-purple- M. reticulate . yu ucatanensis 8b. Corolla lobes 5-9 mm long, length to width (sinus-sinus) ratio less than 0. 80, margins not pinsy limb and lobes ascending; corolla yellow-green when fresh, sometimes drying 2. r and somewhat retic M. aenea ulate 6b. Corona lobes of various shapes but never modified as above; long trichomes of internodes rarely uncin ona lobes more than 102. Corolla tube with circular lin ate. 9a. idan deeply campanulate, sca -sinus length g, linear or ta spathulate. m or more, margins strongly revolute; 7. M. standleyana 10b. Corolla tube with a oue hen or without a distinct patte apic ri twining vines bx Airani caudices; plants of the Isthm of Tehua 12a. Corolla limb with codd bane or AGE Ti indumentum within stricted to the reticulations; septa connecting corona lobes to oL aen eac h with a prominent tooth; calyx lobes 6 mm or less wi “fr M. ceratopetala 12 = selves form Corolla limb with circular lines only occasionally merging, is veins them a regular angular reticulum, indumentum within uniformly ing a inis ibuted, not —— to veins; septa of corona entire; calyx lobes 6 M. eximia m or more llb. daa apex pee ira concave; erect or weakly twining vines, mostly from thickened caudices; plants from northwest of the Isthmus of Tehuantepec ............ 1 . M. dictyantha 9b. Corolla shallowly campanulate to nearly rotate, base-sinus length 1 ] mm or lon margins ightly or not at all revolute; corona lobes less than 4.5 mm long, short- MARC with acute . M. hemsleyana apices ...... 1536 Annals of the Missouri Botanical Garden 1. Matelea hemsleyana Woodson, Ann. Mis- souri Bot. Gard. 28: 237. 1941, based on Dictyanthus parviflorus Hemsley. Dictyan- thus parviflorus Hemsley, Biol. Cent. Amer., Bot. 2: 329. 1882, not Matelea parviflora (Torrey) Woodson. TYPE: México. Chiapas: "les montagnes pres in village indien de Can- cunc" [?Cancuc, Mpio. Chilón, Chiapas], June, year not given (fl), Ghiesbreght 663 (holo- type, K, not seen; isotypes, GH, MO, NY). Figure 1. Dictyanthus rre Brandegee, Univ. Calif. Publ. Bot. 7: 20, not Matelea prostrata (Will- de Boi Wood son. eee baleen Woodson, Ann. Missouri Bot. Gard. 1941, based on Dic- tyanthus ei cta ted TYPE: México. Ve- racruz: or “Acasonica” [Acaxónica], Aug. 191 19 (8). j de 8411 pro parte (holotype, UC; isotypes, GH, MO, NY, US(2), VT). Plants erect to trailing or rarely weakly twining. Stems 20-60(-90) cm long, with a woody caudex to 4 cm long and 2 cm wide, this with thin to thick corky bark, also often with short woody stems above caudex, these with or without corky bark, the herbaceous stems with dense short and glan- dular trichomes and sparse to dense, mostly straight long trichomes to 3 mm long. Leaf blade ovate to very wide-ovate, 13-34 mm long, 13-36 mm wide, with mostly uncinate long trichomes and often with scattered glandular trichomes below, surface smooth, smaller veins sharply raised below, apex acuminate to attenuate or rarely obtuse, base lo- bate, lobes overlapping to divergent, with 2—6(—8) acropetiolar colleters, margin often somewhat thickened and revolute; petiole 7-18(-26) mm long, with dense short and glandular trichomes and sparse to dense, mostly uncinate long trichomes. Peduncle 1-4 mm long, with dense short and gladular tri- chomes and sparse to dense, straight or uncinate long trichomes; bracts linear or lorate to lanceolate, 2-4 mm long, with indumentum of leaf or nearly glabrous; pedicel 3-5 mm long, with indumentum of peduncle. Calyx lobes narrow-ovate or occa- sionally lanceolate, 4-6 mm long, 1.5-2.5 mm wide, apex acute to attenuate, with one colleter below each sinus, abaxial surface with scattered glandular trichomes and scattered to dense, straight or uncinate long trichomes, adaxial surface gla- brous. Corolla shallowly campanulate, base to sinus length 3-6 mm, limb not distinct, margin slightly revolute; lobes (3-)4-6(-7) mm long, apex acute or occasionally rounded, patent or slightly reflexed at tip, margin slightly revolute; glabrous within except with dense short trichomes on limb and lobes and these sometimes extending down raised ridges within tube, indumentum on outside of straight long trichomes or sometimes the limb and lobes nearly glabrous; tube convoluted, with the raised parts opposite the corona lobes, forming shallow pockets between them, with corona lobes in distinct pockets in bases of raised parts; moderately to densely brown-purple-reticulate, becoming clear pale pur- ple on and around corona lobes. Corona lobes 1- 1.5 mm long, basically short-spathulate with an acute apex, the upper surface with a narrow ridge extending as a short spur to edge of gynostegium. Gynostegium 1-1.5 mm high and 1.5-2 mm wide at apex, short-stipitate, apex convex and slightly bilobed. Corpusculum 0.18-0.22 mm long, 0.08- 0.10 mm wide, pollinia 0.58-0.91 mm long, 0.26- 0.34 mm wide. Follicles 48-70 mm long, 10-18 mm wide, green with white markings, glabrous or with sparse short and glandular trichomes, with 28-54 projections to 2 mm long, arcuate and somewhat reflexed proximally, straight and leaning forward distally. Seeds ca. 4 mm long, ca. 3 mm wide, with a raised margin, this irregularly toothed distally, inside this margin slightly convex on one side and slightly concave on the opposite side, both sides verrucate, concave side with a narrow ridge from apex to near center, apparently pale brown; 25 mm lon Found in Mexico (Michoacán, Estado de Mé- xico, Morelos, Veracruz, Chiapas), Guatemala, and El Salvador (Fig. 2). Collected at elevations of somewhat below 800 m to nearly 2,600 m, but mostly 1,000- 1,500 m, on slopes and hills, mostly in grasslands, but sometimes in open pine-oak for- ests, on volcanic cinder and rocky clay soils. Flow- ering mostly from June through September, but flowering specimens also collected once each in April and November; specimens with mature-sized fruits collected August-December. coma ca. Only those specimens of Purpus 8411 with the locality given as ““Acanoxica” or “Acasonica” [ Acaxónica ] and a date of August 1919 are to be considered types of Dictyanthus prostratus. Collections from the part of the range centered around the state of Morelos differ somewhat from plants in the rest of the range. Woodson, according to his annotations, considered the Morelos plants to be Matelea hemsleyana and the others M. dif fusa. Standley & Williams (1969) also considered the plants from southern Mexico to El Salvador to be M. diffusa, but perhaps without seriously con- sidering the plants from Morelos. Standley (1924) considered the two species to be synonymous. Plants from around Morelos tend to be shorter and more erect and to have thicker caudices, larger, more distinctly veined leaves, larger flowers (to nearly Volume 75, Number 4 1988 Stevens 1537 Matelea subg. Dictyanthus c3 LI HH Lh SS — j = EE m) XA >F ^» A SY” que pr KIAN Wa Zz, e ` FIGURE 1. Flowers.—D. Bipolliniu twice as large), and proportionately longer corona obes. In describing Dictyanthus prostratus (= Matelea diffusa), Brandegee considered it to be different from D. parviflorus (= M. hemsley- ana) in having “five minute scales attached to the middle of the gynostegium representing an inner corona." There seems to be no such character, and the *'scales" were most likely the remains of the attachments of the corona lobes to the gyno- stegium, which are typically torn free when the flower is pressed. Despite the differences described above, no known characters faithfully differentiate the two elements. It may well be that further col- lection will demonstrate that some level of taxo- nomic recognition is preferable. In this regard, it Matelea hemsleyana (Stevens C-162, a cultivated specimen of Stevens 1399) .— A. Habit. —B, C. m. is curious that this is the only species in the sub- genus inhabiting both sides of the continent, a distinctly uncommon phenomenon among the viney milk weeds. 2. Matelea tuberosa (Robinson) Woodson, Ann. Missouri Bot. Gard. 28: 237. 1941. Dic- tyanthus tuberosus Robinson, Daedalus 27: 180-181. 1891/1892 [1893]. LECTOTYPE (here chosen): México. Jalisco: slopes of bar- ranca near Guadalajara, 10 Sep. 1891 (fl), Pringle 3568 (lectotype, GH; isolectotypes, F, VT). Lectoparatype: México. Jalisco: Guadalajara, in ravines, (15 July-3 Aug. fide McVaugh, 1956, p. 215) 1886 (fl), Palmer 1538 Annals of the Missouri Botanical Garden 2 p m 0 == = E i 0 e° M.hemsleyana Sie i o N . ° ) i * M.tuberosa go i / a je . T. "e, / ° I ( E ol 400 KM ‘he ` i” a FIGURE 2. Distributions of Matelea hemsleyana and M. tuberosa. 251 (lectoparatype, GH; isolectoparatypes, ENCB, G, K, MO, ND, NY(2), P, PH, US, WU). Figure 3. Plants erect to trailing or sometimes weakly twining. Stems 10-70(-100) cm long, with a woody caudex to 5 cm long and 3 cm wide, this with thick corky bark, otherwise typically herbaceous and lacking bark (rarely subshrubs with erect, branched woody stems), with dense short trichomes, very sparse glandular trichomes, and sparse to dense, mostly straight long trichomes to 2 mm long. Leaf blade ovate to very wide-ovate, 17-45 mm long, 17-40 mm wide, with mostly uncinate long tri- chomes, surface smooth, smaller veins sharply raised below, apex acuminate to attenuate, base lobate, lobes mostly convergent to descending, with 3-6 (-9) acropetiolar colleters, margin often somewhat thickened and revolute; petiole 7-31 mm long, with dense short trichomes, very sparse glandular trichomes, and sparse to dense, mostly uncinate long trichomes. Peduncle 0.5-9 mm long, with dense short trichomes, very sparse glandular tri- chomes, and sparse to dense, straight or uncinate long trichomes; bracts linear or lorate to lanceolate, 2-8 mm long, with mostly uncinate long trichomes; pedicel 4-5 mm long, with indumentum of pedun- cle. Calyx lobes lanceolate to narrow-ovate or el- liptic, 5-9 mm long, 1.5-3.5 mm wide, apex acute to attenuate, with one colleter below each sinus, abaxial surface with sparse to dense, straight or uncinate long trichomes, adaxial surface glabrous. Corolla deeply campanulate, base to sinus length mm, limb revolute; lobes mm long, apex acute, slightly to strongly reflexed, margins strongly revolute; glabrous within except with dense short trichomes on limb and lobes, indumentum outside of short trichomes on tube and of straight or uncinate long trichomes on limb and lobes, oc- casionally with a few long trichomes scattered along tube and occasionally distal third of lobes glabrous; with a pair of ridges within tube opposite each corona lobe, ridges of adjacent pairs almost coming together at base and forming pockets at base of corolla, with the corona lobes in distinct pockets in the bases of the furrows between the paired ridges; within the tube with fine gray-brown vertical lines, limb densely gray-brown-reticulate. Corona lobes ca. 2 mm long (but borne distinctly above base of corolla), shape elaborate but basically sag- ittate in outline, adnate to gynostegium by a thin septum. Gynostegium ca. 2 mm high and ca. 2 mm wide at apex, stipitate, apex broadly and shal- Volume 75, Number 4 1988 Stevens 1539 Matelea subg. Dictyanthus — = ne Sees == ` Le Matelea tuberosa (Stevens C-163 and C-164, cultivated specimens of Stevens 1458 and 1473, l réspectiuely, and Stevens 1473) .—4. Habit. —B, C. Flo lowly concave with the corpuscula as high points and slightly convex and bilobed in center. Corpus- culum 0.14-0.22 mm long, 0.08-0.13 mm wide, pollinia 0.63-0.86 mm long, 0.29-0.37 mm wide. Follicles 55-65 mm long, 11-19 mm wide, mottled pale and dark green, with scattered short and long trichomes, with 50-110 arcuate projections to 2 mm long. Seeds nearly circular, 5.5-6 mm long, 4.5-5 mm wide, with a raised, radially grooved wers.— 0D. Bipollinium. margin, this entire to shallowly toothed distally, inside this margin slightly convex and verrucate on both sides, one side with a narrow ridge from apex to near center, pale brown; coma ca. 25 mm long. Collected from southern Sonora to southern Ja- lisco (Fig. 2) at elevations of 500-1,600 m, in open oak and pine-oak forests and adjacent grass- 1540 Annals of the Missouri Botanical Garden lands, usually in shallow, red, clay soil. Flowering specimens have been collected from late July to early October, and the one specimen with mature seeds was collected in March. The nearly tubular corolla of this species readily distinguishes it from the other species of this sub- genus and is probably unique in the genus Matelea. 3. Matelea hamata W. D. Stevens, sp. nov. TYPE: Mexico. Guerrero: La Unión, 50 m, 29 July 1898 (fl), Langlassé 257 (holotype, US; isotypes, CH, P). Figure 4 Matelea hamata W. D. Stevens; a speciebus ceteris subgeneris Dictyanthi pagina interiore corollae glabra et corona disciformi carnosa lutea lobis brevibus (circa 1 mm) corollae adnatis clare distinguenda. Plants pining vines. pue woody below, with corky bark, herb tel a and glandular trichomes and sparse to very sparse straight or uncinate long trichomes to 1.5 mm long. Leaf blade narrow-ovate to wide-ovate, 52-113 mm long, 26-52 mm wide, with sparse to dense uncinate long trichomes, surface smooth to minutely pusticulate, apex acute to attenuate, base lobate, lobes slightly convergent to divergent, with 4-9 acropetiolar colleters, margin somewhat thickened and revolute; petiole with moderately dense short and glandular tri- chomes and sparse, mostly uncinate long tri- chomes. Inflorescence relatively elongate; pedun- cle 14-40(-53) mm long, with indumentum of stem; bracts narrow-elliptic to elliptic or lanceolate to narrow-ovate, 4—10 mm long, with indumentum of leaf; pedicel 7-19 mm long, with indumentum of stem. Calyx lobes lanceolate to narrow-ovate or elliptic, 10-17 mm long, 4-6.5 mm wide, apex acute or acuminate, with one colleter below each sinus, abaxial surface with moderately dense, straight or uncinate long trichomes, adaxial surface glabrous. Corolla urceolate-campanulate, base to sinus lengt -16 mm, limb broad, patent; lobes 8-13 mm long, apex obtuse or rounded, slightly recurved, margin patent; glabrous within, indu- mentum on outside of dense short trichomes and of moderately dense long trichomes on limb and lobes; tube shallowly convoluted, with a pair of ridges opposite and a shallow pit alternate with each corona lobe; with fine, faint, reticulate lines within the tube, limb with fine, distinct circular lines, these becoming reticulate on lobes, these lines gray-brown on a pale yellow-green background. Corona lobes connate and forming a fleshy yellow disk ca. 6.5 mm wide, lobe tips subulate and ex- tending ca. 1 mm above rim of disk, disk adnate J dense 3-61 mm long, to corolla base and lobe tips adnate to corolla be- tween paired ridges, with a fleshy septum from each lobe to the gynostegium, each septum with a fleshy ornate hump. Gynostegium ca. 3 mm high and ca. 3 mm wide at apex, stipitate, apex broadly and shallowly concave with the corpuscula as high points and a slight hump in center. Corpusculum 0.22-0.24 mm long, 0.18-0.19 mm wide, pollinia 1.24-1.49 mm long, 0.41-0.48 mm wide. Fruit and seeds unknown. Paratypes. MÉXICO. GUERRERO: near el Tuzal, ca. 9 mi. SE of Petatlán, ca. 80 m, 25 July 1976 (fl), Stevens et al. 2538 (ARIZ, DS, DUKE, ENCB, F, G, GH, L, MICH, MO, MSC, NY, P, SD, SMU, TEX, UMO, US, WIS). oaxaca: Distrito de Pochutla, camino a la Bahía de Santa Cruz, 2 km al S de la desviación, 50 m 25 July 1982 (fl), Cedillo et al. 1697 (MEXU, MO). Known only from near the coast (less than 100 m elevation) in Oaxaca and Guerrero (Figure 6), growing in clayey soil. Known flowering in July. n many ways this is intermediate between Ma- telea hemsleyana and M. tuberosa on one hand and the other species of the subgenus on the other. The overall aspect of the plant, the general size and shape of the corolla, the connate corona lobes, and the size and shape of the bipollinia are essen- tially comparable to the larger group of species. The corona lobes, in this case the tips of the lobes, are more comparable in size, shape, and method of adnation to the corolla of M. hemsleyana and M. tuberosa. The fleshy corona disk is, however, unique. Also unique are the large, nearly foliaceous bracts, the large elliptic calyx lobes, the broa patent corolla limb, and the relatively short and blunt corolla lobes. The holotype of this species previously had been tentatively determined as Dictyanthus stapeliiflo- rus; see the discussion of this name under Matelea pavonit. 4. Matelea pavonii (Decaisne) Woodson, Ann. Missouri Bot. Gard. 28: 237. 1941. Dictyan- thus pavonii Decaisne in de Candolle, Prodr. 8: 605. 1844. TYPE: “Pavón” (Sessé, Mocino et al. s.n.) (holotype, FI, not seen; fragment of holotype, P). Figure Stapelia campanulata Pavón ex i E in de Candolle, Prodr. . 1844, A vea sii ag c Mec "phos 47: 258-259. 1847. unknown. Dictyanthus. o Reichenbach, Selectis e Sem- inar E eu Ene = 4. 1850 AN 24: 1851), n TYPE: unknowr Dictyanihus stap eli iflorus Reichenbach, ll.cc. Matelea eliiflora (Reichenbach) Woodson, Ann. Mis- souri Bot. Gard. 28: 237. 1941. TYPE: unknown. Volume 75, Number 4 1988 Stevens Matelea subg. Dictyanthus URE 4. Bipollinium. Stapelia campanulata Sessé & Mocino, Pl. Nov. e 41. 1888. TYPE: unknown (Sessé, Mociño et al. s.n from mountains of Mazatlán, Guerrero). Plants twining vines. Stems woody below, with thick or occasionally thin corky bark; herbaceous stems with sparse to dense short and glandular Matelea hamata (Stevens et al. 2538).— 4. Section of flowering stem.—B, C. Flowers.—D. trichomes and sparse to dense, straight or uncinate long trichomes to 1.5 mm long, these brittle and often missing from specimens. Leaf blade ovate to wide-ovate or rarely very wide-ovate, (49—)60— 128 mm long, 29-100 mm wide, indumentum of sparse to dense uncinate long trichomes, surface 1541 1542 Annals of the Missouri Botanical Garden pusticulate to minutely pusticulate or occasionally nearly smooth, smaller veins occasionally slightly raised below, apex acuminate to attenuate, base lobate or very rarely cordate, lobes mostly de- scending to widely divergent, with (0-)1-6(-11) acropetiolar colleters, margin often slightly thick- ened and revolute; petiole (16-)22-65(-81) mm long, with indumentum of stem. Peduncle 9-60 (-90) mm long, with indumentum of stem or oc- casionally long trichomes absent; bracts lanceolate or occasionally lorate, narrow-oblong, very narrow- elliptic, narrow-ovate, or ovate, 4-13 mm long, with sparse to dense short and uncinate long tri- chomes or sometimes nearly glabrous; pedicel (7—) 10-25(-32) mm long, with indumentum of stem or occasionally long trichomes absent. Calyx lobes lanceolate to narrow-ovate or rarely ovate, 9-18 mm long, 3-6(-9) mm wide, apex attenuate, with one colleter below each sinus, abaxial surface with sparse to moderately dense uncinate long tri- chomes, adaxial surface glabrous. Corolla campan- ulate, base to sinus length 13-25 mm, limb rev- olute; lobes 11-25 mm long, apex rounded or occasionally acute or obtuse, patent to slightly re- flexed, margin revolute; glabrous within except with moderately dense short trichomes around corona lobes and on limb and lobes, indumentum on outside of very sparse to dense uncinate long trichomes except lobes distally to entirely glabrous; tube con- voluted with the raised parts opposite the corona lobes and deep sacs formed between them; with brown, purple, or red vertical lines within tube, these lines becoming finer and circular on base of limb and finely to densely reticulate on distal part of limb and on lobes. Corona lobes 7-13 mm long, linear to linear-spathulate in outline, connate at base, adnate by a thin septum to gynostegium, this septum continuing as a narrow ridge nearly the length of the lobe and often with 1 or rarely 2 distinct teeth on upper margin. Gynostegium 3- 6(- 7) mm high and 3-4.5 mm wide at apex, stip- itate, apex apiculate, the apiculum 0.5-1.5 mm long, slightly shorter than to slightly exceeding corpuscula, appearing to be papillate when dried. Corpusculum 0.31-0.38 mm long, 0.12-0.18 mm wide; pollinia 1.45-1.62 mm long, 0.42-0.48 mm wide. Follicles 70-106 mm long, 22-23 mm wide, green with pale green streaks, with dense short trichomes, bearing 18-54 projections, these thick, straight or slightly arcuate, to 4 mm long. Seeds ca. 6.5 mm long, ca. 3.5 mm wide, with a raised margin, this irregularly crenate distally, inside this margin slightly convex on one side and slightly concave on the opposite side, convex side tuber- culate, concave side longitudinally verrucate, dark black-brown; coma ca. 25 mm long. Collected from southernmost Sinaloa to Oaxaca (Fig. 6). Found mostly at elevations of 900-2,000 m, but also once at 2,500 m and three times at about 600-750 m. Mostly occurring in mountain- ous areas where pine-oak forests occupy the more exposed sites and tropical deciduous forests occupy the more protected slopes and barrancas. Found in either vegetation type but especially common in disturbed places. Apparently tolerant of a variety of substrates, including limestone, lava, weathered metamorphics, and alluvium. Flowering primarily July-September but also collected in June, Octo- ber, and November. Mature fruits known only in cultivation (March). Although there can be no question as to the proper name of this, the type species of Dictyan- thus, the treatment of the Sessé & Mocino names and specimens has considerably complicated the synonymy. A summary of the Sessé & Mociño specimens relating to Matelea pavonii is provided in Table 1. A curious aspect of this summary is that two of the herbarium numbers, 3580 and 3581, are mixtures of Matelea pavonii and M. standleyana. Since these species are not known to be sympatric, the mixing probably occurred at some stage of herbarium handling. After both Sessé and Mocino had died, Pavon apparently distributed specimens from their herbarium. It was upon one of these specimens that Decaisne based his genus Dictyanthus. The specimen was in the Webb her- barium, which at the time was in Paris (now at FI), but Decaisne kept a fragment which is now at P. I have not seen the specimen at FI, but Dr. Rogers McVaugh has examined it. Decaisne attributed the specimen and the label name, Stapelia campan- ulata, to Pavon. The name was probably actually the same Sessé & Mocirio name that was published posthumously in their Plantae Novae Hispaniae (1888). In that publication a locality, mountains of Mazatlán, Guerrero, and a plate, “Fl. Mex. Ic. 255," are both cited, but these cannot be associ- ated with any particular one of the Sessé & Mocino herbarium numbers nor with the specimen distrib- uted by Pavón. The plate is apparently the same as the de Candolle plate 804, labeled “255,” which is at G (F neg. 30763). A small line drawing taken from the flowers of the de Candolle plate 804 is also at G (F neg. 30406) and is labeled “Eurybia stapeliaeflora." 'This may or may not have been a Sessé & Mocino name, but was never published. Reichenbach added to the proliferation of names Volume 75, Number 4 Stevens 1543 1988 Matelea subg. Dictyanthus Matelea pavonii (A-D from Stevens C-160, a cultivated specimen of Stevens 1375; E from Stevens FIGURE 5. 1421) .—A. Section of flowering stem.—B, C. Flowers.—D. Bipollinium.—E. Base of stem. 1544 Annals of the Missouri Botanical Garden TABLE l. Sessé & Mociño collections pertinent to Matelea pavonii. Sessé & Mocirio F Herbarium Herbar- Negative Sessé & Mocino Woodson’s Determinations Accord- ber i Number Label Names Determinations ing to This Treatment 829 MA 41470 Stapelia campanulata Matelea pavonii not determinable from photo, probably not subgenus Dictyan- thus, perhaps not Matelea 835 MA 41471 Cynanchum violaceum M. pavonii probably at least partly M. pavonii 838 F s= == M. pavonii M. pavonii 838 MA 41472 C. campanulatum M. pavonii M. pavonii 1253 MA 41473 C. campanulatum & M. pavonii M. standleyana C. punctatum 3580 F — M. pavonii M. standleyana 3580 MA 41474 s. campanulata M. pavonii M. pavonii 3581 F = M. pavonii M. pavonii plus M. standleyana 3581 MA 41475 C. campanulatum M. pavonii M. pavonii plus M. standleyana s.n. FI — S. campanulata — M. n (not seen) s.n P — — = M. pavo by describing Dictyanthus campanulatus and D. stapeliiflorus. The former, apparently as an early attempt to apply the idea of priority, is a super- fluous name, since Reichenbach cited “Stapelia campanulata Pavón. D. pavonii DC. prodr. Tym- pananthe suberosa Haskarl.” Dictyanthus sta- peliiflorus is a most problematic name. Reichen- bach described both of his species from plants growing at a botanical garden in Dresden from seeds collected in Mexico at the foot of the Sierra Madre, near Durango. Apparently no specimens were prepared and neither description is alone ad- equate for identifying the species, but both could apply to Matelea pavonii and the source area would be more appropriate for that species than for any of the other larger-flowered species of sub- genus Dictyanthus. Partly on the basis of this weak evidence and partly because of a later reference (Anonymous, 1857), I have tentatively considered D. stapeliiflorus to be synonymous with Matelea pavonii. This Anonymous (1857) reference pur- ports to provide the first illustration of D. stape- liiflorus. It is implied but not stated that the illus- tration, taken from a living plant, is from the original material. Considering that the plant was apparently also growing in a German botanical garden and that only seven years had passed since Reichen- bach's description, it could well have been from original material. The illustration does not precisely fit any species of Matelea, but most resembles M. pavonii. It is conceivable that this represents a distinct species that has never been re-collected, but, in the absence of specimens, I consider it an atypical representative of the variable M. pavonii. Although Woodson (1941) made a new combina- tion, Matelea stapeliiflora, one cannot be certain as to what he intended the name to apply. In some cases he annotated specimens of M. tuberosa with this name, probably following Gray's misapplication of the name (in Watson, 1887). In one other case, Woodson applied the name to a specimen of M. yucatanensis, a duplicate of which he properly determined. He also almost certainly had examined the specimen of Langlassé 257 at US, which had been tentatively determined as D. stapeliiflorus, but which is here described as a new species, Ma- telea hamata. Standley (1924) apparently (but tentatively) described the equivalent of my M. ha- mata under D. stapeliiflorus, and this could also have accounted for Woodson’s concept of the species. In the late 1840s this species was introduced into European botanical gardens and was appar- ently a popular plant for about ten years. During this period, at least seven illustrations, mostly col- ored plates, were published in horticultural journals (Anonymous, 1852; Anonymous & Beaton, 1852; Morren, 1852; Planchon € Van Houtte, 1852- Volume 75, Number 4 Stevens 1545 1988 Matelea subg. Dictyanthus k. us y O y N P S SW: k ' / M 4 b. pa ya ` , H U \ e. K su 5 ` [] Tas 1 ' e a M. lauta `, š e M. pavonii A M LJ s M. hamata Ae s š ° e e, A 400 KM S .? E e ` i ° .° gode Fe FIGURE 6. Distributions of Matelea lauta, M. pavonii, and M. hamata. 1853; Anonymous, 1853; Anonymous, 1857; Anonymous, 1862). The plants probably originated from one or two introductions, but little reliable information was provided. This is the most common species of subgenus Dictyanthus and exhibits considerable floral vari- ation throughout its range. The most conspicuous variation is in the background color of the corolla and in the color and density of corolla reticulations, but the basic color pattern is essentially constant. The presence or absence of teeth on the septum connecting the corona lobe to the gynostegium appears to have some geographical basis, all of the toothed specimens occurring from Jalisco and northward, but untoothed specimens occur throughout the range. In some populations, ex- amples can be found with prominent teeth, with very small teeth, and with no teeth. 5. Matelea macvaughiana W. D. Stevens, sp. nov. TYPE: México. Jalisco: moist slopes near Guadalajara (between El Castillo and Juana- catlan, fide Davis, 1936, p. 199), 5 Aug. 1902 (fl), Pringle 8629 (holotype, MSC; isotypes, ENCB, F, G(4), GH, L(2), MEXU, MO, NY, P, PH(2), POM, UC, US(2, VT, W). Figure 7. Matelea macvaughiana W. D. Ste s; M. pavo co tibus lateralibus loborum coronae duobus parvis et pro- minentiis folliculorum numerosioribus Plants erect to occasionally twining. Stems 20— 85 cm long, with an herbaceous or woody caudex to 4 cm long and 2 cm wide, this with thin to moderately thick corky bark, occasionally with short woody stems above caudex, these with or without thin corky bark, herbaceous stems with dense short trichomes, sparse to dense glandular trichomes, and sparse to dense straight long trichomes to 3 mm long, these thin and often broken off on lower and older stems. Leaf blade ovate to wide-ovate or rarely narrow-ovate or very wide-ovate, 30-95 mm long, 21-72 mm wide, with sparse to dense uncinate long trichomes, surface smooth, smaller veins sharply raised below, apex acuminate to at- tenuate, base lobate, lobes mostly descending to widely divergent, with 2-6(-8) acropetiolar col- leters, margin often somewhat thickened and rev- olute; petiole 9-37(-48) mm long, with dense short trichomes, sparse to dense glandular trichomes, 1546 Annals of the Missouri Botanical Garden Volume 75, Number 4 1988 Stevens 1547 Matelea subg. Dictyanthus and sparse to dense uncinate long trichomes. In- florescence often reduced to a single flower (then with or rarely apparently without a rudimentary peduncle); peduncle absent-16 mm long, with in- dumentum of stem or occasionally with long tri- chomes nearly absent; bracts linear to lorate or lanceolate, (2-)4-7 mm long, abaxial surface with dense short trichomes, sparse to dense glandular trichomes, and sparse to dense, straight or uncinate long trichomes, adaxial surface glabrous or with scattered short trichomes distally; pedicel (5.5—)8— 20 mm long, with indumentum of peduncle. Calyx lobes lanceolate to narrow-ovate, 8-12 mm long, 3-4.5 mm wide, apex attenuate, with 1(2) colle- ter(s) below each sinus, abaxial surface with dense short trichomes, sparse to dense glandular tri- chomes, and sparse to dense, straight or uncinate long trichomes, adaxial surface glabrous. Corolla campanulate, base to sinus length 9-12 mm, limb revolute; lobes 9-17 mm long, apex rounded or occasionally obtuse, patent, margin revolute; gla- brous within except limb and lobes with moderately dense to dense short trichomes, indumentum out- side of dense short trichomes on tube and limb and sparse to moderately dense uncinate long trichomes on limb and lobes; tube convoluted with the raised parts opposite the corona lobes and sacs formed between them; with gray or black vertical lines within the tube, these becoming circular on base of limb and reticulate on distal part of limb and lobes. Corona lobes (6—)7—9(—10) mm long, linear to linear-spathulate in outline, connate at base, adnate by a thin septum to gynostegium, this sep- tum continuing as a narrow ridge about half the length of the lobe and with a pair of small thick teeth lateral to the upper margin near center. Gynostegium (2.5-)3-4 mm high and (3—)4— wide at apex, stipitate, apex broadly convex or nearly apiculate, the center apparently slightly shorter than to equaling the corpuscula. Corpus- culum 0.23-0.26 mm long, 0.12-0.15 mm wide, pollinia 1.46-1.68 mm long, 0.43-0.49 mm wide. Follicles ca. 83 mm long, ca. 20 mm wide, striped pale and dark green, with dense short and glandular trichomes, with 118-144 arcuate projections to 3 mm long. Seeds 5.5-6 mm long, ca. 4.5 mm wide, with a raised, faintly radially grooved margin, this 5 mm entire or weakly toothed distally, inside this margin both sides flat or slightly convex, both sides ver- rucate to rugose, one side with a slight ridge from apex to near center, pale brown; coma 25-30 mm long. Paratypes. MÉXICO. JALISCO: Huejotitán, pu 1912 (fl), Diquet s.n. vere (2, one mixed with telea pa- vonii), P, US); ranch near Coyula (near Tonalá) ca. 12 mi. E of Cunas, July 1963 (fl), Faberge s.n. grs wet seepage area 23 mi. S of Guadalajara on Hwy. 5,300 ft., 13 July 1963 Bu Molseed & Rice 220 RIZ MEXU, MICH, MO, NY, UC); wet meadows near Gua- dalajara "m Castillo, according to Davis, 1936, p. 118), 22 Aug. 1893 (fl), ips i 5431 (GH, VT). Michoacan: culated fields 6-7 km N of Jaripó, roadside thickets, 1,60 , 1 Dec. 1970 PA McVaugh 24934 (MICH, The known collection localities are essentially centered in the region of Lago de Chapala at an elevation of about 1,600 m (Fig. 10). Apparently growing in seasonally wet meadows and grasslands; the erect or weakly twining habit and ground-level perennating parts are consistent with this open type of vegetation. Flowering July-August. Specimens with mature-sized fruit collected in December. This species is named in honor of Dr. Rogers McVaugh for his extraordinary contributions to the flora of the part of Mexico in which this species is found, not the least of which are the many fine specimens of Matelea. It is something of a quirk that this species re- quires description. Woodson recognized the species, but according to his annotation of the MO specimen of Pringle 8629, he considered it to be Matelea dictyantha Woodson, a new name based on Ry- tidoloma reticulatum Turcz. This apparently re- sulted from the fact that the two Pringle collections of this species were misdetermined as Dictyanthus reticulatus (Turcz.) Bentham & Hooker f. ex Hem- sley (actually “Dictyanthus reticulatus Turcz. (ex char.)" in the case of Pringle 5431 and “Dic- tyanthus reticulatus B. & H." in the case of Pringle 8629). Woodson, in providing the new name, cited both Turczaninow's name and type (Jürgensen 692), leaving no question as to the application of the name. Jürgensen 692, which Woodson apparently never examined, represents another species of Matelea, which according to his E FIGURE p flower, ca. same scale as A, negati (McVaugh 24934, sor d indicate the presen Caudex (McVaugh 24934, MICH) . Representative niei A Matelea macvaughiana. e taken from color rs uice r. Robert W. Crud D. ea reduced to a single flow a peduncle—contrast with more dy esee of A (Pringle 8629, VT).— —A. Flower (Pringle 8629, o . "od —C. Fru r (past anthesis), withou E a bract to 1548 Annals of the Missouri Botanical Garden annotations he did not recognize, but which must nevertheless bear the name M. dictyantha. This left the species he did recognize without a descrip- tion, type, or name, which are herewith provided. Matelea macvaughiana is likely to be confused with Matelea pavonii because of the similarity of the shape and color pattern of the corolla but is amply distinct, most prominently by having a cau- ex, an erect or weakly twining habit, straight rather than uncinate long trichomes on several structures, smaller flowers on more reduced inflo- rescences, paired lateral teeth on the corona lobes, indumentum absent around the corona lobes, and more numerous and arcuate projections on the follicles. o . Matelea lauta W. D. Stevens, sp. nov. TYPE: Mexico. Colima: steep ravines in gorge of Rio Cihuatlan, near bridge 13 mi. N of Santiago, 200-300 m, 27 July 1957 (fl), McVaugh 15826 (holotype, MICH). Figure 8. ied = led p a differt ordinatio natio rolla, une pedunculis pedicellque [in m dice floribusque parvioribus; a speciebus ibus subgeneris Dictyanthi calyce et pa- gina exteriore corollae glabra facile dignoscenda. a M. pavonii Plants twining vines. Stems woody below, with thick corky bark, herbaceous stems with dense, short, sparse glandular trichomes and sparse straight or uncinate long trichomes to 1 mm long. Leaf blade ovate to wide-ovate, 107-130 mm long, 71- 88 mm wide, indumentum of sparse uncinate lon trichomes above and dense uncinate long trichomes below, surface pusticulate along the veins, apex acuminate, base lobate, lobes descending to widely divergent, with 4—6 acropetiolar colleters; petiole 47-81 mm long, with indumentum of stem. Pe- duncle 1-4 mm long, with moderately dense short and glandular trichomes; bracts lanceolate, 2.5-5 mm long, with sparse short trichomes abaxially, adaxially glabrous; pedicel 3-4.5 mm long, with moderately dense short and glandular trichomes. Calyx lobes lanceolate, 8-9.4 mm long, 2-3 mm wide, apex acute, with one colleter below each sinus, glabrous. Corolla campanulate, base to sinus length 15-18 mm, limb patent; lobes 14-18 mm long, apex acute, patent, margin revolute; glabrous within except with sparse short trichomes on lobes and in a line around corona lobes, glabrous without; tube convoluted with the raised parts opposite the corona lobes and deep sacs formed between them; with gray-brown vertical lines within tube, these becoming + angularly and uniformly reticulate on limb and lobes. Corona lobes 7-8 mm long, linear- spathulate in outline, connate at base, adnate by a thin septum to gynostegium. Gynostegium ca. 4.5 mm high and 3 mm wide at apex, stipitate, apex apiculate, the apiculum ca. 0.5 mm long, equaling corpuscula, shriveled and apparently pap- illate when dried. Corpusculum ca. 0.33 mm long, 0.20 mm wide, pollinia ca. 1.27 mm long, 0.40 mm wide. Fruit and seeds unknown. Known only from the type collection in lowland Colima (Fig. 6), flowering in July This new species appears to be closest to M. pavonii but can be immediately distinguished by the pattern of reticulations of the corolla. Most inflorescence and floral measurements of this species are notably smaller than the corresponding mea- surements of M. pavonii, and the glabrous calyx and outer surface of the corolla are unique in the subgenus. Additionally, the single collection was made at a significantly lower elevation than any known collection of M. pavonii. 7. Matelea standleyana Woodson, Ann. Mis- souri Bot. Gard. 28: 237. 1941, based on Dictyanthus tigrinus Conzatti & Standley. Dictyanthus tigrinus Conzatti & Standley in Standley, Contr. U.S. Natl. Herb. 23: 1183- 1184. 1924, not Matelea tigrina (Grisebach) Woodson. TYPE: México. Oaxaca: Dist. Tux- tepec, Laguna de Ojitlán, 350 m, 31 Oct. 1919 (fl), Conzatti 3760 (holotype, US; iso- type, GH). Figure 9. Plants twining vines. Stems herbaceous and lacking bark, or sometimes rhizomes slightly woody and with thin corky bark; rhizomes thin, horizontal; stem indumentum of sparse to dense short and glandular trichomes and sparse to very sparse straight long trichomes to 1.5 mm long, these very brittle and mostly missing from specimens. Leaf blade wide-ovate to very wide-ovate or occasionally ovate, 48-104 mm long, 36-102 mm wide, in- dumentum of sparse uncinate long trichomes above and dense uncinate long trichomes below, surface smooth, apex acuminate to attenuate, base lobate, lobes mostly convergent to descending, with 1-7 acropetiolar colleters; petiole 35-112 mm long, with sparse to dense short and glandular trichomes and sparse to very sparse, straight or uncinate long trichomes. Peduncle 5-18(-25) mm long, with in- dumentum of stem or often with long trichomes absent; bracts linear to lanceolate, 2-6 mm long, with dense short and sparse straight long trichomes; pedicel 7-16 mm long, with indumentum of stem. Calyx lobes narrow-ovate or occasionally lanceolate Volume 75, Number 4 1549 1988 Stevens Matelea subg. Dictyanthus FIGURE 8. Matelea lauta (McVaugh 15826) .— A. Section of flowering stem.—B, C. Flowers.—D. Bipollin- ium.—E. Base of stem. 1550 Annals of the Missouri Botanical Garden 3 yy Sy D TR LI ES QU FAA S 2/7 ES WO X M FicURE 9. Matelea standleyana (Stevens C-161, a cultivated specimen of Stevens 1392).—A. Section of flowering stem.—B, C. Flowers.—D. Bipollinium.—E. Base of stem.—F. Base of adaxial surface of leaf blade, or ovate, (8.5-)12-18 mm long, 4-6.5 mm wide, revolute; lobes 17-28 mm long, apex acute, patent apex attenuate, with one colleter below each sinus, to reflexed, margin revolute; glabrous within except abaxial surface with dense short trichomes, margin with moderately dense to dense short trichomes on with sparse straight or uncinate long trichomes, lobes, limb, and around corona lobes, indumentum adaxial surface glabrous. Corolla deeply campan- on outside of sparse to dense short trichomes; tube ulate, base to sinus length (14-)17-31 mm, limb convoluted with the raised parts opposite the co- Volume 75, Number 4 1988 Stevens 1551 Matelea subg. Dictyanthus * M.standleyana M. macvaughiana o M.yucatanensis 400 KM FIGURE 10. rona lobes and sacs formed between them; with thick, brown-red, circular lines within tube, these becoming thinner and reticulate on distal part of limb and lobes. Corona lobes 9-13 mm long, linear in outline, connate at base, adnate by a thin septum to gynostegium. Gynostegium (3-)4-5 mm high and wide at apex, stipitate, apex with a blunt projection (formed from apices of anther wings) below each corpusculum and exceeding them laterally, apex convex with tip flattened and slightly bilobed and slightly exceeding corpuscula, terminal anther appendages covering ca. 1/3 of apex. Cor- pusculum 0.48-0.55 mm long, 0.23-0.28 mm wide, pollinia 1.54-1.88 mm long, 0.45-0.63 mm wide. Mature follicles unknown, immature follicles fusiform, to 85 mm long, to 28 mm wide, appar- ently green, with dense short trichomes, with ca. 50 very thick, straight projections to 7 mm long. Seeds unknown. mm Apparently a plant of moist thickets restricted to northern Oaxaca and adjacent Veracruz, with one collection in northern Chiapas (Fig. 10), at elevations up to O m. Flowering specimens collected mainly in July and August, but also once each in June and late October. The one immature fruit was collected in August. Flowers of a green- Distributions of Matelea standleyana, M. macvaughiana, M. yucatanensis, and M. aenea. house-grown specimen were noticed to produce a faint foetid odor in late afternoon. This species is readily identifiable because of the large, deeply campanulate corolla with circular markings within the tube. It is likely the largest- flowered New World asclepiad. Well-formed leaves of this species are, along with those of Matelea pavonii, the largest of the subgenus and have a uniquely angular sinus. For a discussion of the Sessé & Mocino collec- tions of this species, see Matelea pavonii. 8. Matelea ceratopetala (J. D. Smith) Wood- son, Ann. Missouri Bot. Gard. 28: 236. 1941. Dictyanthus ceratopetalus J. D. Smith, Bot. Gaz. (Crawfordsville) 18: 208. 1893. TYPE: Guatemala. Santa Rosa: Santa Rosa, 3,000 ft., Aug. 1892 (fl), Heyde & Lux ex J. D. Smith 3999 (holotype, US; isotypes, G, GH, K, NY, US). Figure 11 Plants twining or occasionally trailing or erect, frequently rooting at lower nodes. Stems 25-55 cm long when erect, woody and with thin to thick corky bark below or occasionally entirely herba- ceous, herbaceous stems with dense short tri- 1552 Annals of the Missouri Botanical Garden chomes, lacking to dense glandular trichomes, and very sparse to moderately dense, mostly straight long trichomes to 2 mm long. Leaf blade ovate to very wide-ovate, 25-63 mm long, 15-55 mm wide, indumentum above of sparse to occasionally dense straight or uncinate long trichomes and occasion- ally also sparse short trichomes, or rarely glabrous, indumentum below of sparse to dense mostly un- cinate long trichomes, glandular trichomes lacking to sparse, surface smooth, smaller veins often mod- erately to sharply raised below, apex acute to at- tenuate, base lobate, lobes mostly convergent, with 1-4(-7) acropetiolar colleters, margin often some- what thickened and revolute; petiole 13-60(-70) mm long, with dense short trichomes, sparse to dense glandular trichomes, and very sparse to mod- erately dense, mostly uncinate long trichomes. In- florescence sometimes reduced to a single flower but with a distinct peduncle; peduncle 1-15(-27) mm long, with dense short trichomes, sparse to dense glandular trichomes, and very sparse to mod- erately dense, straight or uncinate long trichomes; bracts linear to lanceolate, 3-5(-7) mm long, abax- ial surface with short, glandular, and long tri- chomes, adaxial surface with short and occasionally also scattered long trichomes; pedicel 3-12(-15) mm long, sometimes accrescent in fruit, with in- dumentum of peduncle. Calyx lobes narrow-ovate or occasionally lanceolate, 9-13 mm long, 3.5-6 mm wide, apex acute or attenuate, with one colleter below each sinus, abaxial surface with scattered straight or uncinate long trichomes, adaxial surface glabrous. Corolla campanulate, base to sinus length 1 m, limb revolute; lobes 11-18 mm long, apex acute, patent to reflexed, margin revolute; glabrous within except with scattered short tri- chomes on lobes, on limb (on veins only), and around corona lobes, indumentum on outside of sparse to dense short and scattered long trichomes; tube convoluted with the raised parts opposite the corona lobes and deep sacs formed between them; brown- or red-purple-reticulate, reticulations wider within tube. Corona lobes 8-11 mm long, linear- spathulate in outline, connate at base, adnate by a thin septum to gynostegium, this septum with a distinct tooth near center of upper margin. Gyno- stegium 3.5-5.5 mm high and 3-4.5 mm wide at apex, stipitate, apex apiculate, the apiculum 1— 2 mm long and exceeding corpuscula. Corpusculum 0.41-0.46 mm long, 0.18-0.22 mm wide, pollinia 1.52-1.63 mm long, 0.40-0.49 mm wide. Folli- cles 75-90(-103) mm long, 20-27 mm wide, striped dark green and pale green or white, with short and glandular trichomes, with (36-)48-60 straight or occasionally arcuate projections to 7 mm long. Seeds ca. 4.5 mm long and 2.5-3.5 mm wide, with a raised, radially grooved margin, this weakly toothed distally, inside this margin convex and verrucate on one side, concave and verrucate to rugose on the opposite side, concave side with a slight ridge extending ca. 1 mm from apex; coma 20-30 mm long. Collected from southern Guatemala to southern Nicaragua (Fig. 13), mostly at elevations of 800— 1,000 m but occasionally up to 1,400 m and down to near sea level, both extremes occurring in Nic- aragua. Apparently not found on forests but rather in moist or dry fields, thickets, fencerows, stream- sides, and roadsides. Mostly associated with rocky volcanic soils but once noted as occurring on a salt at. Flowering mostly mid July to early October but as early as June and as late as November. Collected with mature-sized fruit from mid Septem- ber to December. Until now, the plants here considered to be Ma- telea dictyantha have been included in this taxon. The two species are actually quite distinct, and M. ceratopetala has its closest affinities with M. pa- vonii. This problem is further discussed under M. dictyantha. 9. Matelea eximia W. D. Stevens, sp. TYPE: México. Chiapas: Fca. Fuarez, Esc. (Fin- ca Union Juarez, fide M. Sousa S., pers. comm.), 12 Aug. 1937 (fl), Matuda 1778 (holotype, MICH; isotype, MEXU). Figure 12. nov. Matelea eximia W. D. Stevens; differt a M. cerato- petala ordinatione ve enationis super limbo inr ae o nate circulari (in illa culum), septis coronae integris, apiculo styli bre- viore, floribus ubique grandioribus. Plants twining or trailing, rooting at lower nodes. Stems slightly woody below and very sparse straight long trichomes to 0.5 mm ae Leaf blade ovate, 45-58 mm long, 36-41 mm wide, indumentum above of sparse straight or uncinate long trichomes, indumentum below of dense uncinate long trichomes, surface smooth, apex acuminate to attenuate, base lobate, lobes convergent, with 1—4 acropetiolar colleters, margin somewhat thickened and revolute; petiole 44-65 mm long, with dense short trichomes, sparse glan- dular trichomes, and sparse to dense (only distally) uncinate long trichomes. Peduncle 3-8 mm long, with dense short trichomes; bracts lanceolate to narrow-ovate, 2-3 mm long (probably larger, pri- mary bracts unknown), with dense short and sparse long trichomes; pedicels 7-9 mm long, with dense Volume 75, Number 4 1988 Stevens Matelea subg. Dictyanthus 1553 š NS š N E WSS Y FIGURE 11. pollinium.—E. Section of old stem. short trichomes and sparse straight long trichomes (only distally). Calyx lobes narrow-ovate to ovate, -15 mm long, 6-7.2 mm wide, apex acute or attenuate, with 1 colleter below each sinus, abaxial surface with scattered short trichomes and sparse straight or uncinate long trichomes, adaxial surface glabrous. Corolla campanulate, base to sinus length Matelea ceratopetala (Stevens 1245) .—4. Section of flowering stem.—B, C. Flowers. —D. Bi- 16-20 mm, margin apparently revolute; lobes 14— 23 mm long, apex acute, apparently reflexed with margin revolute; with uniformly distributed, mod- erately dense short trichomes within except gla- brous in tube between corona lobes, indumentum outside of dense short trichomes; tube convoluted with the raised parts opposite the corona lobes and 1554 Annals of the Missouri Botanical Garden FIGURE 12. Matelea eximia (Breedlove 28682 and Matuda 1778).—A. Section of flowering stem.—B, C. Flowers. deep sacs formed between them; apparently purple- reticulate. Corona lobes 6-7 mm long, linear-spath- ulate in outline, connate at base, adnate by a thin, entire septum to gynostegium. Gynostegium 4.5- 5 mm high and 2.7-3 mm wide at apex, stipitate, apex apiculate, the apiculum ca. 0.8 mm long and exceeded by corpuscula. Corpusculum ca. 0.37 mm long, 0.23 mm wide, pollinia ca. 1.17 mm long, 0.30 mm wide. Fruits and seeds unknown. Paratype. | MÉxico. CHIAPAS: Mpio. de Motozintla de Mendoza, 25-27 km NE of Huixtla along road to Mo- Volume 75, Number 4 1988 Stevens 1555 Matelea subg. Dictyanthus v , j ER hl = E d m in „=! i 0 i: ! s F ' i qom HS ia Ç e fe M i e° br - *« Fi * M.ceratopetala C5 49. t * M. dictyantha 73$ o M. suffruticosa < « M. eximia = 400 KM 2 FIGURE 13. Distributions of Matelea ceratopetala, M. dictyantha, M. suffruticosa, and M. eximia. tozintla SW of Toliman, 700 m, 7 Oct. 1972 (fl), Breed- love 28682 (DS). Known only from the type collection and one other, both from the southern corner of Chiapas (Fig. 13). This new species is clearly related to Matelea ceratopetala, with which it has a contiguous but apparently allopatric range. These two species are quite similar in general aspect and are the only ones in the subgenus where the trailing stems typ- ically root at the nodes. The flowers of the two known collections of M. eximia are rather different in appearance, perhaps because the Matuda col- lection was bleached of corolla coloration during preservation. The Matuda collection thus has a corolla limb with the regular circular veins inter- secting the vertical veins and producing a regular angular reticulation; the Breedlove collection has the same pattern of veins but the circular veins are pigmented while the vertical are not, resulting in a circular pattern. Both are different, however, from the irregular roundish pigmented reticulations of M. ceratopetala. The additional differences in limb indumentum, corona septa, and style apex apiculum leave little doubt that neither collection pertains to M. ceratopetala, nor to any other species of subgenus Dictyanthus; but until M. eximia is better known, there must remain some doubt that the two collections are properly associated. 10. Matelea dictyantha Woodson, Ann. Mis- souri Bot. Gard. 28: 236. 1941, based on Rytidoloma reticulatum Turczaninow. Ry- tidoloma reticultum Turczaninow, Bull. Soc. Imp. Naturalistes Moscou 25(2): 319-320. 1852, not Matelea reticulata (Engelmann ex A. Gray) Woodson. Dictyanthus reticulatus (Turczaninow) Bentham & Hooker f. ex Hem- sley, Biol. Cent.-Amer., Bot. 2: 329. 1882. TYPE: México. Oaxaca: “Sierra San Pedro No- lasco, Talea, etc.,” 1843-1844 (fl), Jurgen- sen 692 (holotype, KW, not seen; isotype, K). Figure 14. Plants erect to trailing or twining. Stems 15- 70(-150+) cm long, with a woody caudex to 5 cm long and 3 cm wide, this with thin to thick corky bark, or occasionally with an elongate woody rhizome, also often with woody stems above caudex or rhizome, these usually with thin corky bark and up to ca. 5 cm long but occasionally up to 20 cm 1556 Annals of the Missouri Botanical Garden long, herbaceous stems with dense short and glan- dular trichomes and sparse to dense, mostly straight long trichomes to 3 mm long. Leaf blade wide- ovate to very wide-ovate, 26-62(-103) mm long, 24-52(-76) mm wide, with long uncinate tri- chomes and also often glandular trichomes on veins below, surface smooth, smaller veins sharply raised elow, apex acute to attenuate or rarely obtuse, base lobate, lobes mostly convergent to descending, with 1-6(-8) acropetiolar colleters, margin often somewhat thickened and revolute; petiole (11—)14— 34(-57) mm long, with dense short and glandular gle flower with or apparently without a rudimentary peduncle; peduncle absent-10 mm long, with in- dumentum of petiole; bracts linear to lanceolate, 2.5-7 mm long, abaxial surface with indumentum of stem, adaxial surface glabrous; pedicel 5-12 (-16) mm long, sometimes markedly accrescent in fruit, with indumentum of stem. Calyx lobes nar- row-ovate or occasionally lanceolate or ovate, 6— 11 mm long, 2.5-6 mm wide, apex acute to at- tenuate, with one colleter below each sinus or oc- casionally these somewhat above sinus near margin of lobe, abaxial surface with indumentum of stem, adaxial surface glabrous. Corolla campanulate, base to sinus length (7-)9-16 mm, limb revolute; lobes 8-14 mm long, apex acute or sometimes rounded, patent to strongly reflexed, margin revolute; gla- brous within except with sparse to dense short trichomes around corona lobes and on limb and bases of lobes or sometimes over whole surface of lobes, indumentum on outside of short trichomes and occasionally also with long trichomes on limb and bases of lobes, or occasionally tube and tips sacs formed between them; faintly to densely gray- purple-reticulate, reticulations wider within tube. Corona lobes m long, linear in outline with a raised margin, connate at base, adnate by a thin septum to gynostegium, this septum con- tinuing as a narrow ridge nearly the length of lobe. Gynostegium 3-4 mm high and 3-3.5 mm wide at apex, stipitate, apex broadly and shallowly con- cave with corpuscula as high points. Corpusculum 0.22-0.35 mm long, 0.09-0.17 mm wide, pollinia 1.17-1.45 mm long, 0.29-0.38 mm wide. Folli- cles (45-)55-70 mm long, 10-22 mm wide, pale green with a few dark green stripes, with short and glandular trichomes, with (30-)50-1 10 straight to arcuate projections to 3 mm long. Seeds ca. 5.5 mm long and 4—4.5 mm wide, with a raised, radially grooved margin, this weakly toothed distally, inside this margin flat or slightly concave and verrucate on one side, convex and verrucate on the opposite side, the flat side with a slight ridge from apex to near center, pale brown to brown; coma 25-30 mm long. Collected in the mountains of four more or less discrete areas: around Cuernavaca (Morelos and adjacent state of Mexico), around Chilpancingo (Guerrero), around Oaxaca (Oaxaca), and in south- western Puebla (Fig. 13). Found at elevations of ca. 1, O m. About a third of the collec- tions are noted as being on or associated with limestone, and many of the other localities are in limestone areas, but whether or not Matelea dic- tyantha is restricted to limestone is uncertain. Mostly in low, open oak, pine, or pine-oak forests, especially where disturbed. Flowering mid June to mid September. Mature-sized fruits collected Au- gust- December, mature seeds only in December. As mentioned under Matelea macvaughiana, Woodson provided the epithet for M. dictyantha quite by accident. He did not distinguish it from M. ceratopetala, to which it bears some resem- blance in the shape and color pattern of the corolla and the size and shape of the leaves. The major characteristics distinguishing M. dictyantha from M. ceratopetala include the woody caudex and predominantly shorter habit, the smaller flowers, the lack of a tooth on the septum connecting the corona lobe to the gynostegium, the concave rather than apiculate gynostegium apex, and the smaller follicles with smaller and more numerous projec- tions. The geographic ranges of the two species are also distinct. The distinctiveness of M. tyantha from M. ceratopetala has been noted on dic- two specimens, Conzatti 2168 at F (unsigned and undated) and Pringle 1768 at GH (J. M. Green- man, 18 Sep. 1890). Standley (1924) considered Rytidoloma reticulatum to be synonymous with Matelea pavonii. The form of Turczaninow’s protologue (1852) has led to some confusion. The apparent generic description is actually a description of the genus and its single species and cites Jurgensen 692. Immediately following the genus-species descrip- tion is the entry “18. R. reticulatum. Altera species hujus generis, quantum e flore unico, Ptino cor- rupto, dijudicare possum, adest in collectione Gale- ottiana ex Oaxaca sub n. 1563.” This led Langman 1964: 748) to state that Rytidoloma reticulatum was based on a Galeotti collection. Turczaninow — was actually indicating that he recognized another Volume 75, Number 4 Stevens 1557 1988 Matelea subg. Dictyanthus ES 39. FIGURE 14. Matelea dictyantha (4—D from Stevens C-105, a cultivated specimen of Graham 1231; E from Stevens 1311).— 4. Section of flowering stem.—B, C. Flowers. —D. Bipollinium.—E. Caudex. 1558 Annals of the Missouri Botanical Garden species in his new genus, but he neither named nor described it. Galeotti 1563 is indeed a different species, Matelea standleyana. As with M. hemsleyana, two somewhat distinc- tive elements of M. dictyantha can be recognized, but it does not seem that they warrant taxonomic recognition on the basis of currently available ma- terial. The flowers of the Morelos and Guerrero element tend to be larger, more shallowly cam- panulate, and much lighter in color than the Pue- bla-Oaxaca element. In this case the ranges are apparently distinct, but I have found no objective way of describing the difference in dried specimens. Perhaps with additional field study their separation will become more feasible. In this connection, it is most likely that the specimens of Ghiesbreght s.n. from “province d'Oaxaca" were actually collected at Cuernavaca, Morelos; such mixing of labels ap- parently often occurred with Ghiesbreght speci- mens. 11. Matelea suffruticosa W. D. Stevens, sp. nov. TYPE: México. Oaxaca: 4 mi. W of junc- tion of Mex. 185 with Mex. 190, near La Ventosa, 17 Aug. 1971 (fl), Wunderlin, Dwyer, Spellman & Vaughn 800 (holotype, MO; iso- type, MEXU). Figure 15. Matelea tuffrutionsa. Ns b. Stevens; M. yucatanensis atque M. species snag Dietyanthi forma corollae (tubo uste campanulato, limbo in sinibus abrupte reflexo, lobis erectiusculis) notabilis. Plants apparently suffrutescent with twining tips. Woody stems with thin corky bark, herbaceous stems with dense short and glandular trichomes and moderately dense uncinate long trichomes to 1.5 mm long. Leaf blade ovate to wide-ovate, 40— 51 mm long, 27-39 mm wide, indumentum of short and glandular trichomes on and near major veins and moderately dense (above) to dense (be- low) uncinate long trichomes, surface smooth, apex acuminate, base lobate, lobes descending, with 5- 7 acropetiolar colleters, margin somewhat thick- ened, revolute, and crispate; petiole 36-53 mm long, with indumentum of stem. Inflorescence sometimes with a smaller adjacent cyme apparently produced lateral to the axillary bud, that is, orig- inating between the axillary bud and the normal extra-axillary inflorescence; peduncle to 1 mm long, with dense short and glandular trichomes; bracts linear to lorate, margins mostly revolute, 1-2 mm long, with short, glandular, and uncinate long tri- chomes abaxially, glabrous adaxially; pedicel 1.5- 2.5 mm long, with indumentum of stem. Calyx lobes decurrent on pedicel, elliptic to wide-elliptic, 3.3-4.5 mm long, 1.6-1.7 mm wide, apex acute, with one colleter below each sinus, abaxial surface with dense short, glandular, and uncinate long tri- chomes, adaxial surface glabrous. Corolla narrowly campanulate, base to sinus length 5.5-6.5 mm limb sharply reflexed at each sinus; lobes 4-5 mm ng, apex rounded, apparently erect, margins re- volute; tube glabrous within, limb and lobes with stiff, erect, glassy, apparently unicellular trichomes m long, outer surface of corolla with eia short, glandular, and uncinate long tri- chomes; tube apparently suffused with deep purple, becoming purple-reticulate on limb and lobes. Co- rona lobes 3-3.5 mm long, linear-spathulate with tip deeply convoluted, glistening purple-black, sep- arate to base, base (below style apex) yellow and glistening. Gynostegium ca. 2 mm high and 1.5 mm wide at apex, short-stipitate, apex nearly plane with corpuscula forming high points and center slightly convex, terminal anther appendages cov- ering about half of apex. Corpusculum ca. 0.25 mm long, 0.17 mm wide, pollinia ca. 0.53 mm long, 0.37 mm wide. Fruits and seeds unknown. o Known only from the type collection, from the south side of the Isthmus of eee (Fig. 13), probably below 100 m elevat This new species has some ie innovations rus educ fits easily into the subgenus and, ctive a , is probably closely spit rr related to the two species from the Yucatan. The exaggerated development of the corolla limb at the sinuses, which even at the bud stage is obvious as the sinuses have the appearance of recurved horns, causes the modified tips of the corona lobes to be presented in a unique fashion, at the distal margin of the limb, with which they contrast in color, between the erect and revolute corolla lobes. The hairs on the inside of the corolla, the occasional subsidiary inflorescences, and the roundish (rather than pentagonal) style apices appear to be unique in the subgenus but not in the genus as a whole. The highly modified tips of the corona lobes suggest a link between this species and the two Yucatan species, as does similarity in indumentum. 12. Matelea aenea (Woodson) W. D. Stevens, comb. nov. Dictyanthus aeneus Woodson, Amer. J. Bot. 22: 691, pl. 1, fig. 4. 1935. TYPE: México. Yucatan: Progreso, 11-15 Aug. 1932 (fl), Steere 3005 (holotype, MO; isotype, MICH). Figure 16. Plants twining vines. Stems woody below, with thin to thick corky bark, herbaceous stems with Volume 75, Number 4 1988 Stevens 1559 Matelea subg. Dictyanthus FIGURE 15. D. Bipollinium. dense short and glandular trichomes and dense uncinate long trichomes to 2.5 mm. Leaf blade wide-ovate to very wide-ovate or occasionally ovate, 35-98 mm long, 26-70 mm wide, indumentum of uncinate long trichomes and of T ganar tri- h, apex acute chomes on veins below, surf to acuminate, base lobate, lobes convergent to widely divergent, with 4-11 acropetiolar colleters, often crispate; petiole 21-62 mm long, with indumentum 0.2mm Matelea suffruticosa (Wunderlin et al. 800) .— 4. Section of flowering stem.—B, C. Flowers.— of stem. Peduncle 4-11 mm long, with indumen- tum of stem; bracts linear to lanceolate, 3-4 mm long, with indumentum of stem or occasionally glabrous on adaxial surface; pedicel ca. 4 mm long, with indumentum of stem. Calyx lobes lanceolate to narrow-ovate, 6-9 mm long, 2- apex acute to attenuate, with 1(2) colleter(s) below each sinus, abaxial surface with dense short, glan- dular, and uncinate long trichomes, adaxial surface 1560 Annals of the Missouri Botanical Garden ju URE l6. Matelea aenea (4-D from Stevens C-157, a cultivated specimen | nun 2e E from Stevens 1145).— 4. Section of flowering stem.—B, C. Flowers.—D. Bipollinium.—E. Base of ste Volume 75, Number 4 1988 Stevens Matelea subg. Dictyanthus 1561 glabrous. Corolla campanulate, base to sinus length (6-)8-12 mm, limb ascending to nearly patent, not revolute; lobes 5-9 mm long, length to width (sinus-sinus) ratio 0.67-0.78, apex acute or rounded, ascending to slightly reflexed at tip, mar- gin not revolute; glabrous within except with dense short trichomes on limb and lobes, indumentum on outside of short, glandular, and uncinate long tri- chomes; tube convoluted with raised parts opposite corona lobes and forming deep sacs between them; pale yellow-green, sometimes with faint reticula- tions, these often drying darker. Corona lobes 6- 8 mm long, linear-spathulate in outline with. tip deeply rugose and glistening purple-black, other- wise yellow-green or tinted purple, connate at base, adnate by a thin septum to gynostegium, this sep- tum continuing as a narrow ridge nearly the length of the lobe. Gynostegium ca. 3 mm high and ca. 2.5 mm wide at apex, stipitate, apex broadly and shallowly concave with corpuscula as high points and occasionally also slightly convex in center, terminal anther appendages hardly covering mar- gin of apex. Corpusculum 0. 8 mm long, 0.12-0.15 mm wide, pollinia 1.08-1.18 mm long, 0.35-0.37 mm wide. Follicles ca. 95 mm long, ca. 20 mm wide, finely mottled, probably green and white, with scattered short and glandular tri- chomes, with ca. 40 thick projections to 2 mm long. Seeds ca. 4.5 mm long and 3 mm wide, with a raised, radially grooved margin, this essentially entire, inside this margin essentially flat on one side and convex on the other, both sides deeply verrucate to deeply rugose, dark brown; coma ca. 40 mm long. Collected only in the vicinity of Progreso, on the tip of the Yucatan Peninsula (Fig. 10), at near sea level. Growing in low scrub vegetation in thin, limestone-derived soils, apparently where not par- ticularly saline. Collected flowering June-August and in December. In describing Dictyanthus aeneus, Woodson (1935) summarized the differences as follows: “T. aeneus differs from T. yucatanensis Standl. [sic ] superficially in the smaller and more shortly peti- olate leaves with paler color and hispidulous or strigillose surfaces, and smaller, paler corollas with a more pronounced campanulate tube. Structural differences of the corolla and corona are conspic- uous as well." Later, Woodson (1941) considered Dictyanthus aeneus to be a synonym of Matelea yucatanensis (Standley) Woodson and so anno- tated the type specimen. I concur with his original recognition of Dictyanthus aeneus, but unfortu- nately his characters are not very diagnostic and, in fact, his drawing of the flower (Woodson, 1935; pl. 1, fig. 4a) has the shape of Matelea yucata- nensis and the size of M. aenea. The best char- acters for separating the two species are most easily observable in fresh flowers. The corolla limb and lobes of Matelea yucatanensis form essentially a right angle with the tube and have revolute margins, while those of M. aenea are ascending and do not have revolute margins. This difference can often still be seen in dried specimens when not carefully pressed. The corolla color is also strikingly different in fresh flowers, being yellow-green with faint re- ticulations in M. aenea and densely gray-purple- reticulate in M. yucatanensis. Unfortunately, M. aenea sometimes darkens in drying and this dif- ference is partially obscured. The most dependable floral characters in pressed specimens are the size and shape of the corolla lobes, but even with these the flowers often require boiling to be measured accurately. As noted in the descriptions, M. aenea has shorter and proportionately wider corolla lobes. There seem to be certain vegetative differences as well, but the variation within each species is large and there are too few specimens to make possible any meaningful conclusions. Matelea aenea tends to have denser long trichomes on the vegetative parts and smaller, wider, more crispate, and less urple-pigmented leaves. Although obviously very closely related and the only two species of the subgenus Dictyanthus occurring on the Yucatan Peninsula, M. aenea appears to be restricted to the coastal area immediately around Progreso, while M. yucatanensis is found at scattered, mostly in- land, localities. Both species are too poorly col- lected, however, to support much conjecture on their relative distributions. 13. Matelea yucatanensis (Standley) Wood- son, Ann. Missouri Bot. Gard. 28: 237. 1941. Dictyanthus yucatanensis Standley, Publ. Field Columbian Mus., Bot. Ser. 8: 37. 1930. TYPE: México. Yucatan: without precise lo- cality or date (fl), Gaumer 933 (holotype, F; fragment of holotype, C). Figure 17. Plants twining vines. Stems woody below, with thin to thick corky bark, herbaceous stems wit dense short and glandular trichomes and dense uncinate long trichomes to 2.5 mm long. Leaf blade ovate to wide-ovate or occasionally very wide-ovate, (39-)45-95 mm long, 24-81 mm wide, indumen- tum of uncinate long trichomes and also glandular trichomes on veins below, surface smooth, apex acuminate to attenuate, base lobate, lobes mostly 1562 Annals of the Missouri Botanical Garden AAA FIGURE 17. Matelea yucatanensis (Stevens C-158, a cultivated specimen of Stevens 1168).—A. Section of flowering stem.— Flowers. —D. Bipollinium.—E. Base of stem.—F. Base of adaxial surface of leaf blade, showing acropetiolar colleters. Volume 75, Number 4 1988 Stevens 1563 Matelea subg. Dictyanthus convergent to descending, with 4-10 acropetiolar colleters, margin often crispate; petiole (22-)42- 57(-82) mm long, with indumentum of stem. Pe- duncle 2-9 mm long, with indumentum of stem; bracts linear to lanceolate, 3-5 mm long, with indumentum of stem or sometimes with long tri- chomes on margin only and glabrous on adaxial surface; pedicel 3-5(-7) mm long, with indumen- tum of stem. Calyx lobes lanceolate to narrow- ovate, 7-10 mm long, 2-3.5 mm wide, apex acute to attenuate, with 1(2) colleter(s) below each sinus, abaxial surface with scattered short trichomes, dense glandular trichomes, and scattered uncinate long trichomes or occasionally nearly glabrous, adaxial surface glabrous. Corolla campanulate, base to sin- us length (7-)10-11 mm, limb patent, revolute; lobes 7-12 mm long, length to width (sinus-sinus) ratio 0.83-1.20, apex acute, patent or slightly reflexed at tip, margin revolute; glabrous within except with sparse to dense short trichomes on lim and lobes, indumentum on outside of short, glan- dular, and uncinate long trichomes; tube convo- luted with raised parts opposite corona lobes and forming deep sacs between them; densely gray- purple- depo reticulations wider in tube. Co- rona lobes -7 mm long, linear-spathulate in outline with Hi deeply rugose and glistening purple- black, otherwise deep purple, connate at base, ad- nate by a thin septum to gynostegium, this septum continuing as a narrow ridge nearly the length of lobe. Gynostegium 3-3.5 mm high and 2.5-3 mm wide at apex, stipitate, apex broadly and shallowly concave with corpuscula as high points and oc- casionally also slightly convex in center, terminal anther appendages hardly covering margin of apex. Corpusculum 0.24-0.26 mm long, 0.12-0.15 mm wide, pollinia 1.11-1.26 mm long, 0.28-0.35 mm wide. Follicles ca. 95 mm long, ca. 15 mm wide, finely mottled green and white, with scattered short and glandular trichomes, with ca. 55 thick projec- tions to 4 mm long. Seeds ca. 4.5 mm long an 3.5 mm wide, with a raised, faintly radially grooved margin, this entire, inside this margin essentially flat on one side and convex on the other, both sides deeply verrucate to deeply rugose, dark brown; coma ca. 35 mm lon The identifiable collection localities are scattered in the state of Yucatan (Fig. 10), at elevations probably well below 200 m. Almost certainly to be expected in the adjacent areas of Campeche and Quintana Roo. Growing in low forests and second growth and probably always in limestone-derived soils. Collected flowering in June, July, and October and fruiting in October. Most closely related to Matelea aenea; for com- parison see discussion under that species. These two species form a distinct unit morphologically and are geographically isolated from the other species of subgenus Dictyanthus. They are ob- viously related to M. suffruticosa and the several species grouped with M. pavonii but have clear affinities only with the former. Their most con- spicuous innovation, besides occupying a unique region and a unique environment (karst iP oe is the highly modified tip of the corona lobe. tips glisten as if they are wet but apparently pro- duce no secretion. They may function as ““pseudo- nectaries." It should be noted, however, that the sides of the corona lobes tory in apparently the same manner as the species grouped with M. pavonii. These two species, along with M. suffru- ticosa, are also unique in having predominantly uncinate long trichomes on the internodes, the oth- er species considered here having either entirely straight or only occasionally a few uncinate long trichomes on the internodes. LITERATURE CITED ANONYMOUS. 1852. Dictyanthus O Jor- an. pei Companion Florists’ Guide 1: 20-21. 53. Dictyanthus pavonii. Bot. Mas 79: t 4250. [E. O.]. 1857. Dictyanthus ee daro, ci Rchb. n D. Pavonii Decne. Gartenflora 6: 66, t. 18 nta: Dictyanthus campanulatus (Bell-flow- ered dictyanth). J. Hort. Pract. Gard., n.s. 3: 414- [B. J .] & D. BEATON. 1852. Bell-flowered dic- tyanth. ee campanulatus). Cottage Gar- dener 8: 236-237 BENTHAM, G. & J. D. Hooker. 1876. Genera Planta- rum ..., Asclepiadeae. 2: 728-785 Davis, H. B. 1936. Life and Work of Cyrus Guernsey Pringle. Univ. of Vermont, Burlington, Vermont. DECAISNE, J. 1844. Asclepiadeae. In: A. P. de Candolle, Prodromus Systematis Naturalis Regni Vegetabilis 8: -665. grad W.B. 1882. Asclepiadaceae. /n: F. D. God n & O. Salvin (editors), Biologia Centrali-Ameri- ana. Botany 2: 318-338, t. 5 m L. J. 1973. Classification of the architecture i J. Bot. 60: 17-33. Norman MoRREN, C. 1852. Le Tympananthe suberosa, mag- nifique vag igo - peus terre pour l'été. Bel- gique và 73, PLANCHON, j E.& L. V4 riis 1852-1853. Dic- tyanthus pavonii. Fl. Serres Jard. Eur. 8: 55-56, pl. 6, 770 1564 Annals of the Missouri Botanical Garden SCHUMANN, K. 1895. Asclepiadaceae. /n: A. Engler & K. Prantl, Die Natürlichen Pflanzenfamilien 4: 189- 6. Sessé, M. DE & J. M. Mociño. 1887-1890. Plantae Novae Hispaniae. Mexico. [Published in 9 parts as cr to Naturaleza (Mexico City), Ser. 2, Vol- ERE P. C. Trees and shrubs of Mexico. Mr Contr. U.S. Natl. Herb. 23: 1166- 1194, 1680 . WILLIAMS. 1969. Flora of Guatemala. Asclepiadaceae Fieldiana, Bot. 24: 407-472. STEARN, W. T. 1966. Botanical Latin. Hafner, New York TURCZANINOW, S. 1852. Asclepiadeae quaedam hu- cusque indescriptae. Bull. Soc. Imp. Naturalistes Moscou 25: 310-325. Watson, S. 1887. Contributions to American botany. XXI. r. Edward Palmer in the state of. Jalisco, Mexico, in 1886. Daedalus 22: 396-465. Woopson, R. E., JR. 1935. New Apocynaceae and Asclepiadaceae, Amer. J. Bot. 22: 684-693. orth American Asclepiadaceae. I. i of ili genera. Ann. Missouri Bot. Gard. 28: 193-244 FLORA OF THE VENEZUELAN Julian A. Steyermark* GUAYANA—VI' ABSTRACT Studies an the genus Sloanea (flaansapecsae) Jr the Flora of the Venezuelan TM have yielded changes ronren arensis, S. ca taniapensis, idsei, H longiaristata, Š. merevariensis, 3 parvifructa, S. “sipapoana, S. | icit subsp. autanae, S. s S dav steyermarkii su Alnbacaceas, and Theaceae are inclu jauaensis, S. subpsilocarpa, and S. wurdackii. Notes pertaining to the families Rapateaceae, ed. RAPATEACEAE In preparing the treatment of Rapateaceae for the Flora of the Venezuelan Guayana, a number of discrepancies have been noted in the previous treatments by Maguire (1958, 1982). In the in- terest of placing these observations on record be- fore the publication of the flora, the following data are presented. AMPHIPHYLLUM Amphiphyllum rigidum Gl. Maguire (1982: 105) gave the length of the involucral bracts as 3-4 cm long. On the three specimens in VEN they measure 2-2.5 cm long. CEPHALOSTEMON Cephalostemon vs. Duckea Maguire (1958) separated Duckea on the basis of the exappendiculate seeds. I am unable to sep- arate Duckea from Cephalostemon on gross mor- phological characters. They merge as noted in such species as Duckea squarrosa and Cephalostemon gracilis and should be combined under the earlier- published Cephalostemon. Duckea flava becomes Cephalostemon flavus (Link) Steyerm., a new combination. Cephalostemon flavus (Link) Steyerm., comb. nov. Duckea flava (Link) Mag., Mem. New York Bot. Gard. 10(1): 43. 1958. Cephalostemon junciformis (Mag.) Steyerm., comb. nov. Duckea junciformis Mag., Mem New York Bot. Gard. 10(1): 43, fig. 1. 1958. Cephalostemon flavus vs. Cephalostemon junciformis The bracteole apex is acute in C. junciformis and obtuse to rounded in the commoner C. flavus. Maguire (1958) distinguished them on the basis of the globose inflorescence 1-1.4 cm long of C. junciformis contrasted with the oblong inflores- cence 2-4 cm long of C. flavus. This difference does not hold true: several collections of C. flavus have short globose inflorescences 1 cm long. Mea- surements show the following intergradation: C. junciformis: heads 0.8-1.5 cm high, broadly hemispherical or subhemispherical. C. flavus: heads 1-4 cm high, depressed sub- globose to cylindric-oblong. The depressed or subglobose short inflorescences of C. flavus are represented by the collections of Maguire et al. 37662, uber 4860, 3115, 3129; and Vareschi & Magdefrau 6718, 6590. Cephalostemon cyperaceoides vs. Cephalostemon squarrosus In general, heads in C. cyperaceoides are 1- 1.5 cm broad and smaller than in C. squarrosus with heads 2-2.5 cm broad. Yet heads in C. cy- peraceoides vary in size and may reach 2 cm wide in such collections as Huber & Medina 5905. In others, such as Maguire 29276 (identified as C. squarrosus), the heads are 2.5 cm wide, while the mature bracteoles are nerved and long attenuate or subulate but not squarrose. Although the brac- teoles of Steyermark 75323 are strongly nerved, they show varying degrees of squarroseness, some ! This work was supported by National Science Foundation Grant BSR-85 152085. 2 Deceased, October 15, 1988. ANN. MISSOURI Bor. GARD. 75: 1565-1586. 1988. 1566 Annals of the Missouri Botanical Garden scarcely or not at all squarrose, others only slightly so. Also in C. squarrosus, specimens are found with strongly nerved squarrose bracteoles, but with small heads only 1.3 cm broad, as in Vareschi & Mag- defrau 6611 (VEN), so that overlapping appears in the sizes of the heads and in the degree of squarroseness. In Maguire et al. 30799 (identified as C. squarrosus), the bracteoles are strongly nerved but are mainly ascending at maturity. Moreover, C. squarrosus and C. cyperaceoides do not differ in the type of apex of the bracteole between as described by Maguire in his keys (1958: 42): "bracteoles ..., apex strongly aristate" in Duckea cyperaceoides vs. *bracteoles . . . , apex merely acuminate" in Duckea squarrosa. Both have attenuate apices, and no fine distinction can be drawn: the two taxa may have to be merged eventually. KUNHAR DTIA Kunhardtia rhodantha Mag., Mem. New York Bot. Gard. 10(1): 32, figs. 5, 6. 1958. Schoenocephalium sipapoanum Mag., Acta Bot. Venez. 14(3): 17. 1984. Schoenocephalium sipapoanum was based on a collection (Steyermark et al. 124534) from the summit of Cerro Sipapo. Although Maguire (1984— see above) noted that it had a general similar aspect to that of Kunhardtia, he placed the specimen in Schoenocephalium on the basis primarily of its anthers opening by four apical pores instead of being two-celled as in Kunhardtia. Other differences thought by Maguire to distinguish the two genera were the conspicuous exsertion of the porrect corollas and depressed-subglobose heads of Kunhardtia contrasted with the included corollas and sphaeroidal heads with the mature flowers ra- diate of Schoenocephalium. These gross morphological characters were ob- served by the author at the time the type collection was made, and having already seen thousands of individuals of Kunhardtia rhodantha on the sum- mit of the nearby Cerro Autana (Steyermark, 1974), the Sipapo collection was noted to be in all respects the same as the common Kunhardtia rhodantha. The depressed-subglobose, deep red heads with red, conspicuously exserted porrect flowers, and strong- ly imbricated leaf sheaths abruptly narrowed at their summits of the Sipapo collection match per- fectly the numerous specimens in the colonies of Kunhardtia rhodantha seen on the summit of Cerro Autana. Reexamination and careful comparison of the type collection of Schoenocephalium sipapoanum and Kunhardtia rhodantha does not reveal any further difference between them. Reexamination of the anthers in the type material of Schoeno- cephalium sipapoanum and Kunhardtia rhodan- tha shows no difference between the two. More- over, the other characters pertaining to Kunhardtia as emphasized by Maguire, together with an ex- amination of additional herbarium specimens, em- phasizes the conspecifity of the two taxa. MONOTREMA Monotrema aemulans vs. M. affine The key difference separating M. aemulans from M. affine in Maguire (1958: 46) is that in the former the primary bracts of the inflorescence do not exceed the head or do so only inconspicuously, whereas in M. affine they conspicuously exceed the head. However, specimens cited by Maguire, such as Maguire et al. 30491 from Yapacana savanna, have the bracts somewhat exceeding the heads and are 13 mm long. Thus, Maguire et al. 30491 could be placed in M. affine instead of M. aemulans. Monotrema affine may have the pri- mary bracts only 10-15 mm long. The isotype of M. affine, from Yapacana at VEN, has the longest bracts only 15-16 mm long and barely exceeding the head. The leaves of M. affine vary in width from 5-9 mm up to 16 mm in Huber 5939, those of M. aemulans from 6 to 10(-15) mm. In M. bracteatum var. bracteatum the leaves are only 2-4(-5) mm wide except in var. major, where they are 6-12 mm wide. Some specimens of M. aemulans have bracts usually 8-12 mm long, but in early stages pre- ceding anthesis, as in Steyermark et al. 130334 from Cerro Vinilla (VEN), they are only 1-5.5 mm long and equal the head, or they are slightly shorter to only slightly longer than the head. The two taxa are otherwise similar vegetatively and occur in the same Yapacana savanna The collections Huber & Tillett 2965 (identified by Huber as M. xyridoides) and Huber & Medina 5939 (identified by Maguire as “M. affine with broad leaf blades vel valde aff") are forms with leaves 1.4-1.5 cm wide and leaf sheaths 14-15 cm long, whereas other collections of M. affine vary in leaf width from only 0.5 to 0.9 cm. The broadly ovoid heads are also longer than broad as in M. xyridoides and are 17-21 mm long x 12 mm broad at the middle. The outer involucral bract is 21 mm long and barely longer than the head. Perhaps these represent introgressive collections Volume 75, Number 4 1988 Steyermark 1567 Venezuelan Guayana Flora— VI of M. xyridoides and M. affine. Huber 1640, identified as M. bracteatum, has the shorter broad- er outer bracts of M. affine and the shorter sec- ondary bracts of M. affine. A third species, M. bracteatum, with all the bracts elongated, occurs in the Yapacana savanna and occasionally is confused with M. affine which has shorter, broader outer bracts. A fourth species, M. xyridoides, with heads usually longer than broad having bracteoles round- ed or obtuse, likewise occurs in the Yapacana sa- vanna. Monotrema bracteatum may be confused with Cephalostemon cyperacoides on account of the outer bracts, which in C. cyperacoides are re- flexed, whereas those of M. bracteatum and M. affine are spreading to ascending. Monotrema bracteatum subsp. bracteatum Huber 3268 from W of Serrania El Tigre (iden- tified as M. affine by Maguire) represents M. brac- teatum subsp. bracteatum with the bracts and bracteoles long pointed. Monotrema arthrophyllum (Seub.) Maguire, Mem. New York Bot. Gard. 10(3): 47. 1958. y de Sii Fag abr a Seub. in Mart., Fl. Bras. 3(1): 131. 1847. This species was originally described from Ar- ara-Coara, Rio Caquetá, Colombia. Maguire et al. 44109 from scrub savanna, Araracuara, identified by Maguire as M. aemulans Kornicke, is a topo- type of Schoenocephalium arthrophyllum. In his English key (1958, p. 46) Maguire gave “5-10- flowered" for M. arthrophyllum, as contrasted with “numerous” and “50-75-flowered” for M. aemulans in both his English (1958) and Spanish (1982) treatments. This collection matches the photo of the type of M. arthrophyllum in shape and width of leaf, abrupt contracted leaf sheath summit at the base of the leaf blade, and much shorter leaves compared with the peduncle length. The leaf blades are strongly 1 1-nerved on the lower side, 13-14 mm wide, 14-22 cm long, and show the leaf sheath strongly 7-8 nerved. This compares well with the type photo of M. arthrophyllum. Although Maguire et al. 44109 has strongly nerved leaf sheaths, they do appear, although not very sharply, on the type photo of Schoenocephalium arthrophyllum. A specimen from Yavita, Territorio Federal Amazonas (Williams 14086, VEN), originally identified by Maguire in 1950 as M. arthrophyl- lum, was later identified by him as M. aemulans. Although never cited by Maguire (1958, 1982), this specimen partly agrees with the description and key characters assigned by him (1958) to M. arthrophyllum in having fewer spikelets with heads which are not subdidymous. However, the linear leaves are only 3.5-7 mm wide, whereas in the photo of the type of Schoenocephalium arthro- phyllum they are broader. The Williams 14086 specimen from Yavita does not match either the photo or the Maguire et al. collection from Co- lombia. Its much narrower leaves 3.5-7 mm wide merge, but not abruptly, at the base into the leaf sheath, which is 6 cm long and 4 mm wide. Also, the leaf blade and sheath are not strongly nerved as in the Colombian collections and are 21-25 cm long. The head is 8 mm wide and hemispheric as in M. aemulans, but the head is too small and few- flowered to be placed in M. aemulans. Probably, as Huber concluded, the specimen represents a depauperate inflorescence form of M. xyridoides rather than a form of M. aemulans, as M. ar- throphyllum was interpreted by Maguire. PHELPSIELLA Phelpsiella ptericaulis Maguire The genus is described in Maguire's (1958) ge- neric key as having yellow petals, but the Hoyos & Morillo 518 collection from Laguna Asisa, Cerro Asisa, Serrania Part indicates that the flowers are white. The winged strongly compressed stem and narrow leaf blades are strongly reminiscent of Ste- golepis breweri Mag. SCHOENOCEPHALIUM Schoenocephalium cucullatum vs. coriaceum The distinctions made between S. cucullatum and S. coriaceum (in Maguire's key, 1958: 37) were based on the relative distance separating the upper bracteoles and the tip of the sepals (in 5. coriaceum the sepals exceed ““the upper bracteoles by (3)4-5 mm in length," whereas in S. cucul- latum the sepals exceed the upper bracteoles **2(3) mm or less in length." Also, the width of 14-16 mm is given for the leaf blades of S. coriaceum contrasted with that of “2.0-3.5 cm” for S. cu cullatum. 'These differences have been found in- constant. It should be noted that although Maguire described the leaves of S. cucullatum in his key as '*2.0-3.5 cm broad," none of the specimens from VEN, including the paratype (Maguire et al. 37631) and the isotype (Maguire et al. 30486) of S. cucullatum have leaf blades exceeding 1 cm. 1568 Annals of the Missouri Botanical Garden Moreover, of all the specimens examined at VEN of S. cucullatum determined by Maguire, the leaf blades vary from 0.8 to 1.7 cm broad, those in S. coriaceum varying from 0.5 to 1.3 cm. With the material at hand, the specimens cannot be sepa- rated on either of these characters and, in com- bining them as conspecific, I am employing the binomial Schoenocephalium cucullatum. It should also be noted that in 5. teretifolium the sepals exceed the upper bracteoles by distances of 6-10 mm instead of, as stated, by *“(3-)4-5 mm” (1958). The same error occurs in S. cucul- latum, where the sepals exceed the upper brac- teoles by 2-4 mm and in 5. coriaceum by 2-6 mm. Schoenocephalium martianum Seub. It is probable that S. martianum Seub. may eventually be considered synonymous with S. cu- cullatum, in which case S. cucullatum has priority. However, differences in sepal length in relation to the bracteoles, leaf sheaths, and leaf blades appear to justify their separation for the present. The leaf sheaths in S. cucullatum are shorter and terminate more abruptly at the base of the leaf blade, whereas in S. martianum the leaf sheaths are more elon- gated and merge at their summit more directly with the leaf blade. In S. martianum the sepals exceed the upper bracteoles by 7-10 mm. Also, the leaves are longer in S. martianum than in S. cucullatum and S. coriaceum. | have seen a to- potype of 5. martianum from Colombia (Maguire et al. 44179 VEN, Vaupés). scrub savanna Araracuara, SAXOFRIDERICIA Saxofridericia duidae vs. S. grandis The separation of S. grandis from S. duidae on the basis of the presence of a petiole between the base of the leaf blade and the summit of the leaf sheath is not evident on some specimens. In his (1958) key, Maguire differentiated them as follows: Petioles absent, the leaf passing directly from s sum- mit of the leaf sheath to the leaf blade .. S. duidae Petioles 10-15 cm long, uniting the summit of "i leaf sheath with the leaf blade proper 5. grandis However, in Huber 4426, a topotype of S. gran- dis from Serranía Parú, the length of the petiolar portion varies from only 1 to 2 cm long to as much as 8.5 cm long in Huber 4347, Although the description of S. grandis gives 10— 15 cm as the length of the petiolar portion, the also from Paru. isotype (Cowan & Wurdack 31115) at VEN has the petiolar portion only 4 cm long. Maguire (1958) stated that S. grandis is the only species of sub- genus Saxofridericia to have developed a distinct petiole. In Huber's specimens the summit of the peduncle beneath the head is sharply angled, flat- tened, glaucous, and broadened to 2-2.7 cm. The leaf blades in S. grandis vary from 3.5 to 4.5 cm wide (4—5.5 cm wide according to Maguire). The sheath in 5. grandis is strongly indurated and 4- 7 cm wide. Saxofridericia duidae, confused with S. gran- dis, does not possess a petiolar portion between the summit of the sheath and the base of the leaf blade, the sheath is narrower (2-3 cm wide), less indurated, conspicuously carinate dorsally, the pe- duncle below the head narrower (1.2-1.7 cm wide), less conspicuously ampliated, with stronger ribbing, and the leaf blades mainly narrower (2.5-3.5-4) cm, but overlapping the measurements of S. gran- dis. However, although the other specimens of 5. duidae lack petiolar interruptions between the leaf sheath and the leaf blade, the VEN paratype of S. duidae (Maguire & Maguire, Jr. 29122) has an elongated petiolar portion 10-18 cm long, al- though it has the narrower leaf blades, narrower leaf sheath with dorsal keel, and narrower apical part of the peduncle characteristic of 5. duidae. The base of the leaf blade in this specimen is so narrowly attenuate as to appear petiolate and could be interpreted as sufficiently petiolate as the ma- terial seen of S. grandis. An additional character, however, for differen- tiating the two taxa is that in S. duidae the brac- teoles surrounding each flower are more conspic- uously pungent and more conspicuously imbricate with looser, longer, narrower apices, whereas in S. grandis the bracteoles are more appressed, with shorter, less pungent tips from a more broadly shaped contour. This paratype thus breaks down the distinctions between the two taxa, which are otherwise separated by the differences enumerated above. Saxofridericia compressa vs. S. spongiosa Maguire’s dimensions given for the diameter of the heads, peduncles below the summit of the head, and the leaf width do not apply to collections from VEN not seen or not annotated by him. Thus, the width of leaf blades separating S. compressa and 5. spongiosa (4-5 cm against 7-10 cm broad respectively) must be changed to 3-8 cm and 4.5- cm, respectively. Volume 75, Number 4 1988 Steyermark 1569 Venezuelan Guayana Flora—VI Saxofridericia petiolata vs. S. inermis The fibrous marcescent character of the leaf sheaths of S. petiolata versus the nonmarcescent, nonfibrous character of S. inermis is not apparent on specimens identified by Maguire. Although used as one of the separating key characters (1958), it does not serve to identify most extant herbarium material. STEGOLEPIS Stegolepis microcephala Maguire This is keyed out by Maguire as having the "mature sheaths hardened and nerveless." How- ever, the type specimen has the sheath clearly ed. Stegolepis membranacea Maguire The bracteoles are described as somewhat ob- tuse, but in Steyermark et al. 129658 from Ma- rahuaca, the bracteoles are subacute, and the sheath and the summit of the auricle have thin brown margins. The specimen is referred to 5. mem- branacea because of the nerved sheath (in the upper part) and lack of broad white scarious au- ricle. Stegolepis neblinensis Maguire This species was placed in the section having spikelets 5-50, but in S. neblinensis the spikelets are only 1-3 (peduncles 2-4). In the Spanish key (Fl. Ven.) there is confusion in the couplet **10" which gives ““sepals”” not reflexed. If the indurated sepals are absent, and only the bracteoles remain, then the length measurement is only 2 cm instead of the “2.5-3.5 cm” given for this species. Stegolepis parvipetala Steyerm. On the basis of the number of spikelets, supposed to be **70-100," the isotype (VEN) cannot be properly disposed, since the spikelets are less than 70 (actually there are 50 or fewer), thus relegating it to another part of the key. The key on p. 10 (1982, Fl. Ven.) gives the length of the spikelets as “10-16” mm. However the text (p. 120) gives “14-16” mm. In the key to the subspecies, the spikelets are “14-15” in subsp. parvipetala and * 10-12" in subsp. chimantensis. Stegolepis parvipetala is characterized by the rounded or broadly obtuse apices of the lower and middle bracteoles, blunt apex of the spikelet, thick No] diameter of the peduncle, conspicuously swollen summit of the trigonous summit of the peduncle, and small petals. Stegolepis pauciflora Gleason Although this species is placed by Maguire in that part of the key (1982, Fl. Ven., p. 111) having "auricles of the sheath not ligulate," actually they may attain a length of 25 mm, whereas in S. hitchcockii they are ca. 10 mm long. Stegolepis ptaritepuiensis Steyerm. The key (1958) gives leaf blades “2-2.5 cm" wide, but the text states “2-4.5 cm" wide. Stegolepis pungens Gleason The key in Fl. Ven. (1982: 110) gives spikelet measurement “3-8.8” cm. It should be “3-3.8” cm as given in the text. Actually some spikelets are less than 3 cm (2.5-2. Stegolepis steyermarkii Maguire This species is poorly segregated in the key on the basis of the nervation of the leaf sheaths and their relative thickness or induration. Although nerves may be present along the margin, they are not always present on the rest of the sheath as in some other species. The sheaths may be indurated, as in Steyermark et al. 92498, making it difficult to determine into which part of the key to place the specimen. Stegolepis steyermarkii vs. S. ferruginea Although these two species are very similar, they may be distinguished by the key characters given by Maguire (1965: 71), with S. steyermarkii hav- ing narrower and longer leaf sheaths, glabrous vs. minutely puberulent petals, narrower leaf blades (not given by Maguire), a more conspicuous, broad- er, and more elevated midrib below, and a more attenuate narrower leaf apex. Although the two species are apparently distinct, it should be noted that the length of some leaf sheaths of S. steyermarkii is not as long as that originally stated, i.e., 15-20 x 4.5 cm in text (18-20 x 3.5-4.5 cm in key). The type specimen shows the basal part of the plant with the sheath length complete. However, one of the paratypes (Steyermark & Nilsson 296 NY) has a lone sheath only 12.5 cm long, taken from an inner relatively shorter sheath. Other collections, such as Croat 1570 Annals of the Missouri Botanical Garden 53998, have sheaths 16 cm long. The latter spec- imen has the sheathing portion cut off above the base so that the length shown is not complete. Stegolepis vivipara Examination of the isotype (Steyermark & Wur- dack 332) at VEN shows only 15-20 spikelets and not 25-35 as stated in the original description. The heads are 3.5 cm in diameter. The bracteoles are + 20-25 and not **ca.16" as given by Maguire (1965). They are orbicular or orbicular-ovate, as stated, and pale brown with scarious margins. The midnerve is prominent below with many fine nerves, but much less manifest than in 5. parvipetala and subsp. chimantensis. The upper leaf surface has many fine nerves with a sulcate midrib as in S. parvipetala. The leaf sheath is soft, papyraceous, or submembranaceous and has many fine, closely crowded nerves. The leaf apex is subacute or nar- rowed to an obtuse or subobtuse apex. The broadly obovate petals, described as 22-24 mm long by Maguire (1965), are only 15 mm long and 12 mm wide in the dried state. The sepals are many-costate and 3.5-5 mm wide (3-4 mm as given by Ma- guire). All the bracts of a spikelet are more or less uniformly broadly suborbicular-ovate, obtuse, or subobtuse, whereas in 5. angustata the upper ones are more narrowed to a subobtuse or subacutely obtuse apex and are relatively longer than broad. In S. vivipara the bracts are more uniform and only slightly longer than broad. Also, the bracts of S. vivipara are paler and have more scarious mar- gins and a brown apex when contrasted with S. angustata, in which the bracts are darker brown and lack scarious margins and a brown apex. LITERATURE CITED MAGUIRE, B. 1958. The Botany of the Guayana High- land —III. Mem. New York Bot. Gard. 10: 19-49. The Botany of the Guayana High- land =1V, Mem. New York Bot. Gard. 12(3): 69- 1982. Rapateaceae. In: Fl. Venez. 11: 85- STEYERMARK, J. 1974. The summit vegetation of Cerro Autana. Biotropica 6: 7-13. NOTES ON SLOANEA (ELAEOCARPACEAE) IN THE VENEZUELAN GUAYANA The following notes are based on a treatment of the genus Sloanea in preparation for a Flora of the Venezuelan Guayana. This has necessitated a study of newly collected material obtained from recently completed expeditions to previously unex- plored areas, as well as a restudy of specimens identified by previous workers. I am grateful to the curators of F, GH, NY, US, and VEN for the loan of critical material. As a result of the present study, nine new species are recognized, and critical comments on other species are given. The revision by C. E. Smith (1954) on the New World species of Sloanea elab- orated 62 species, of which nine included taxa from the Venezuelan Guayana and eight species from other parts of Venezuela. An additional species, S. floribunda Spruce ex Benth., collected at San Car- los de Rio Negro, was erroneously ascribed to Brazil instead of Venezuela. Two additional species from the Venezelan Guayana, S. crassifolia and S. stey- ermarkii, were published by Smith in 1962 and 1967, respectively. Steyermark and Marcano-Ber- ti described S. megacarpa from the Guayana in 1966, and additional taxa originating from the Venezuelan Guayana were described by Steyer- mark in 1976 and 1978. Species not originally cited by Smith are now known to occur in the Venezuelan Guayana as a result of recent expe- ditions into that territory. They include, in addition to those published since Smith's revision and the nine described below, S. caribaea, S. guianensis, S. pubescens, S. multiflora, S. robusta, S. syn- andra, and S. terniflora. Sloanea pitttieriana and S. ptariana, previously included by Smith as syn- onyms of other taxa, have been newly studied and found to merit specific recognition. Sloanea floribunda Spruce ex Benth., J. Linn. oc., Bot. 5: suppl. 66. 1861. Sloanea maroana Steyerm., Pittieria 7: 14. 1978. A reexamination of 5. maroana Steyerm. shows that it cannot be maintained apart from S. flori- bunda. Sloanea laurifolia (Benth.) Benth., J. Linn. Soc., Bot. 5: suppl. 70. 1861. This species was treated by Smith (1954) as highly variable and included S. oppositifolia Spruce ex Benth. as well as the collection Cardona 1951 from the Venezuelan Guayana along the Rio Me- revari, which was assigned an herbarium name by Pittier but never published. Schomburgk 936, upon which Dasynema laurifolium Benth. (basionym of Sloanea laurifolia) was based, has opposite, acu- minate leaf blades identical with that of Spruce 3689, the type of Sloanea oppositifolia Spruce ex Benth. The leaf blades in both collections are glabrous throughout and ovate- or oblong-lanceo- Volume 75, Number 4 1988 Steyermark 1571 Venezuelan Guayana Flora—VI late. Cardona 1951, on the other hand, is distinct in having alternate, obovate, rounded leaf blades that are densely tomentose on the upper and lower midnerves. It further differs from the collections of both S. oppositifolia and Dasynema laurifo- lium in having acute instead of obtuse awns on the anthers, and styles free to the base instead of shortly parted only at the apex. Sloanea macrophylla Benth. ex Turcz., Bull. Soc. Imp. Naturalistes Moscow 31(1): 224. 1858. The dimensions given by C. E. Smith (1954) for the stamens of this species were taken from flowers in a very early stage of anthesis. He de- scribed the stamens as “4-6 mm 0.5-1 mm long, usually flattened laterally, mi- nutely puberulent; anthers 2-4 mm long, lanceo- ong; filaments late, minutely puberulent, connective prolonged into a glabrous awn 0.3-0.5 mm long." However, in other collections referred to this species (Liesner et al. 20919) it was noted that as the flowers mature the filaments and appendages of the anthers became more elongated. Sloanea caudata Stey- erm., placed by Smith (1954) in the synonymy of S. macrophylla also has much longer staminal dimensions than those described for that species. Sloanea pittieriana Steyerm., Fieldiana, Bot. 28: 359. 1952. This species was reduced to synonymy by Smith (1954) under Sloanea fendleriana Benth. Sloanea fendleriana was based originally on a specimen collected by Fendler (2489) from the Coastal Cor- dillera of northern Venezuela. Sloanea pittieriana was described from a collection originating in the Venezuelan Guayana Highland, some 700 kilome- ters distant. Reexamination of the type material of these two taxa and study of additional collections shows them to be quite distinct. Although vege- tative differences occur, the chief distinctions are found in the termination of the anther and in the spines of the capsule. In S. pittieriana the apex of the anther is obtuse and has an obsolete, minute, scarcely discernable knob. By contrast, the anther of S. fendleriana terminates in a shortly acuminate summit, originally described by Bentham (1861) as “‘breviter acuminatae,”” or as “prolonged into a small puberulent knob or short awn," as described by Smith (1954). Fruiting material of 5. fendleri- ana collected near the type locality in Estado Ar- agua (Pittier National Park, Pittier & Nakichen- ovich E 5425 VEN), identified by Smith, has merely lacking spines, whereas capsules granu P obtained within the distributional area of 5. pit- tieriana and referred to that species have slender, rigid, glabrous spines. Vegetative differences are likewise evident. In Sloanea fendleriana the lateral secondary nerves terminate dichotomously 3-8 mm from the leaf margins, and the veinlets of the lower surface are prominently reticulate. As opposed to this, the lat- eral nerves of S. pittieriana, best observed on the lower leaf surface, usually extend to the margins without branching, although some branching may be present 3-5 mm from the margin. Furthermore, the tertiary venation of the lower surface is less conspicuous, less elevated, and more impressed. Finally, the lateral nerves of S. pittieriana are more equidistant than in S. fendleriana, the latter with the nerves irregularly spaced. The following specimens, all from the Venezue- lan Guayana, pertain to Sloanea pittieriana: maya Ptari-tepui, Steyermark 59984 (holo- type, F; isotype, VEN); Ptari-tepui, Steyermark 60261 (F); between Eldorado and Luepa, plateau of Cerro Venamo, Steyermark & Nilsson 799 (VEN); Amaruay-tepui, Holst & Liesner 2840 (MO, VEN). Sloanea ptariana Steyerm., Fieldiana, Bot. 28: 360. 1952 This taxon, described from Ptari-tepui in the Venezuelan Guayana, was reduced to synonymy by C. E. Smith (1954) under Sloanea picapica Standley, a species based on fruiting material from Honduras. The holotype of S. picapica (C. & W. von Hagen 1390 NY), together with additional paratype specimens from Central America, iden- tified as 5. picapica by Smith, have been restudied by the author. No flowering material of S. picapica has been collected up to the present, but Damon Andrew Smith stated in his unpublished thesis (1985) on the Costa Rican species of Sloanea, that, although no flowers of 5. picapica had been seen, stamens were found adhering to very young fruits. He described them as being short-tomentose. His detailed description of the stamens is as follows: “filaments at least 1.7 mm long, 0.04 mm in di- ameter; anthers 0.6-0.9 mm long, about 0.2 mm wide, basally slightly cordate; anther sacs opening along entire length, but most widely at apex; awn .04-0.08 mm long, obtuse." Although no flowers of Central American spec- imens of Sloanea picapica were available to C. E. Smith for study, he based his detailed description of the flowers upon the flowering collection of the about holotype of S. ptariana. However, the stamens of 1572 Annals of the Missouri Botanical Garden S. ptariana are quite unlike those found by Damon Smith in S. picapica. Although the stamens of S. ptariana were originally described as pubescent, reexamination, as verified by C. E. Smith (1954), showed that they were glabrous throughout. In addition to the glabrity of the stamens in S. pta- riana, the filaments of that taxon are longer (3- mm) than those of S. picapica. Additional differences between these two taxa may be observed in their leaves. In Sloanea pta- riana the lateral nerves extend nearly unbranched to the margin and anastomose 1-3 mm from it. In the Central American specimens of 5. picapica the lateral nerves branch farther from the margin (in Little 25400, 4-6 mm from the margin before anastomosing). The apex of the leaf in 5. ptariana, moreover, is also longer and narrower throughout its length, and 3-6 mm wide at the base; in the Central American specimens of S. picapica the apex averages shorter and broader at the base, where it is 7-9 mm wide. Based upon the staminal and vegetative differ- ences noted between Sloanea ptariana and S. picapica, I am reinstating the former as a distinct species. This eliminates the previously held erro- neous concept of a disjunct geographical range for S. picapica, a disjunct range not normally found between Central American and Guayana Highland taxa. Sloanea bolivarensis Steyerm., sp. nov. TYPE: Venezuela. Bolivar: 7 km NE of Ciudad Piar, 7°28'N, 63*14'W, 350-500 m, 10 Apr. 1981, Ronald Liesner & Angel Gonzalez 11479 (hoiotype, VEN; isotypes, MO, NY). Figure 1. m, foli iorum laminis obovatis apice obtuse s 17-20 cm m 8-11 cm ando-denta tis longis hispidulis connectivo in acumen hispidulum 1. 5 mm longum producto; filamentis 1.2-1.5 mm longis dense hirsutulis; ovario hispidulo, stylo torto glabro. Tree, 25 m tall, the branches hirtellous. Leaves alternate or subalternate. Petioles 1.7-2.5 cm long, densely hirtellous; leaf blades obovate, obtusely acute at the abruptly prolonged apex, narrowed to an acute base, the larger ones 17-20 cm long, 8- 11 cm wide, the margins each side 11-18 repand- dentate or coarsely obtusely dentate; lower midrib and lateral nerves hirsutulous, upper midrib mi- nutely hirtellous, glabrous elsewhere on lower and upper leaf surfaces; lateral nerves 12-16 each side, ascending at 25—30°; tertiary venation finely ele- vated, minutely reticulate both sides. Inflorescence lateral, near summit of branch, umbellately 2-3- flowered; peduncle 1.5-2 cm long, densely gray- hirtellous; pedicels 10-15 mm long, densely gray- hirtellous. Sepals 4, equal, ovate, acute, 7-9 mm long, 3-5 mm wide, both sides minutely puberu- lous. Anthers lanceolate, 1.8-2.1 mm long, his- pidulous, pei laterally, the connective pro- onged into a hispidulous awn ong, somewhat Pss dc the body if the anther; filaments 1.2-1.5 mm long, densely hirsutulous. Ovary hispidulous; style 2.5-2.8 mm long, twisted, glabrous. Capsule not seen. The twisted style and umbellately few-flowered inflorescence relate this new taxon to Sloanea garckeana Schum., from which it differs in having anther awns uniformly pubescent; densely pubes- cent, shorter pedicels and peduncles; and wider repand-dentate to coarsely dentate leaf blades. Specimens of Sloanea garckeana identified by C. E. Smith and D. Alforo Castaneda show a range of variation. The species, originally based on the collection Riedel 888 from the province of Rio de Janeiro, has entire or subentire, oblong or oblong- lanceolate, acuminate leaf blades 2-5 cm wide, 2- 3-flowered inflorescences on elongated peduncles 5-6 cm long, sepals 8-9 x 4-4.5 mm, anthers (including the apical appendage) 5 mm long with the appendage described as glabrous, and tomen- tose filaments 3-4 mm long. Although Smith (1954) allowed for a range in variation of the anther ap- endage from sparsely pubescent to glabrous, the material I have examined has glabrous awns prin- cipally. Smith (1954) also allowed for a wide range of variation in glabrity and length of peduncles and pedicels. Sloanea cataniapensis Steyerm., sp. nov. TYPE: Venezuela. T. F. Amazonas: Dept. Atures, N side of Rio Cataniapo, 48 km SE of Puerto Ayacucho, 5%35'N, 67?15'W, 200-300 m, 10 May 1980, Julian A. Steyermark, Gerrit Davidse & Francisco Guanchez 122215 (ho- lotype, VEN; isotype, MO). Figure 2. Arbor 15 m, foliorum laminis elliptico-ovatis vel ob- longo-ellipticis apice abrupte acutis basi acutis vel subo umbellatim D | hear axibus lateralibus 1 floris; pedi- cellis 2-4 c g Volume 75, Number 4 Steyermark 1573 1988 Venezuelan Guayana Flora—VI A A 5cm uh shall M4) hd Bub / Y v ne FIGURE 1. Sloanea bolivarensis.— 4. Habit.—B. Flower in anthesis.—C. Pistil and receptacle.—D. Twisted style. OR Leaf blade, detail of apical abaxial side.—F. Stamen, laterally dehiscent 1574 Annals of the Missouri Botanical Garden — OS WAS SS AS pee 5.5mm — = ons = TIR BSS Paco == LE PRA pas = NN t mL = — ux CS TERY << SES CEE, MR NSS a O FIGURE 2. Sloanea cataniapensis.— A. Habit.—B. Flower bud.—C. Stamen.—D. Pistil.—E. One of the hir- as pe ees of the pistil. —F. Base of leaf blade and portion of petiole, from abaxial side. Volume 75, Number 4 Steyermark 1575 1988 Venezuelan Guayana Flora—-VI aequalibus 10 mm longis 7.5 mm lat heris (alabastro) 1%53'-1%27'N, 66*35'-66%32'W, 80 m, 23- pen s 5.5 mm longis dense ashes connectivo in 25 July 1984, Gerrit Davidse 27733 (holo- acumen rotundatum glabrum mm um y k pre dictis Mies (alabastro) 0.5- 0. 7 mm longis cibos. Tree 15 m tall, upper portion of branchlets densely puberulent. Leaves alternate or opposite. Petioles 1-2 cm long, densely fulvous puberulent; leaf blades chartaceous, elliptic-ovate or oblong- elliptic, abruptly acute at apex, acute to subobtuse at base, 7-12.5 cm long, 3.5-6.5 cm wide, entire, glabrous except for the densely minutely puberu- lent midrib on the lower side; lateral nerves 5-6 each side, elevated below; subsulcate above, as- cending at a 45? angle, terminating and anasto- mosing 1-2 mm from the margin; tertiary venation subelevated reticulate below, subrugulose above. Inflorescence umbellate to paniculate, 3-7-flow- ered; primary peduncles elongate, 5.5-9 cm long, 2.5-3 mm wide; secondary axes unbranched, l -flowered, densely fulvous-puberulent; pedicels 2- 4 cm long, 1.5-2 mm wide, densely fulvous-pu- berulent. Flower bud suborbicular, slightly acute at apex, 1 cm long, 1 cm wide basally. Sepals 4, covering the rest of the flowering parts in bud, equal, thickish, with thickened margins, broadly ovate, obtusely acute at apex, 10 mm long, 7.5 mm wide, densely fulvous-puberulent without, se- riceous within. Stamens ca. 60; anthers (in bud) linear, 5.5 mm long (including apical appendage), 0.7-0.8 mm wide, densely hirsutulous upward, dehiscent by an apical pore, the connective pro- longed into a rounded glabrous awn 0.7-1.1 m long; filaments (in bud) 0.5-0.7 mm long, glabrous. Ovary broadly ovoid, 3.5 mm long, minutely his- pidulous; style conic-linear, 2 mm long, densely hispidulous-appressed in the basal portion, else- where glabrous; stamens 4, glabrous. Related taxa, such as Sloanea laxiflora Spruce ex Benth. and S. synandra Spruce ex Benth., have branched lateral or secondary axes of the primary peduncles, whereas in S. cataniapensis the pri- mary peduncles branch only at their summits into three pedicellate flowers. Other differences are shown by the longer pedicels, shorter filaments, more densely pubescent anthers with longer as- cending pubescence, sepals obtusely acute, and more densely puberulous lower midrib of the leaf blade Sloanea davidsei Steyermark, sp. nov. TYPE: Venezuela. T. mazonas: Depto. Rio Ne- . gro, Rio Pacimoni, between its mouth and its junction with the Rio Baria and Rio Yatua, type, VEN; isotype, MO). Figure 3. Arbor 4- 15 m, ramulis superno dense patenti- "pilosis; foliorum basi acutis obtusis vel rotundatis 7- Th cm longis (3.5-) 4-9 cm latis integerrimis, costa media nervis laterali- e pilis p s dense munitis, superficie inferiore e tomentosa; Lear pam simplicibus llat I 5 cm longis dense pilosulis; Pedicelli (6-)10-20 mm longis dense P deli; sepalis quattuor sepamos late ies 5-8 mm longis; i -2 mm longis; connectivo in appendicem hir- sutulum 0.7-1 mm longum producto; riers 1.2-2.8 mm longis; stylis liberis glabris; capsulis ovoideis 2.2-3 m longis, spinis € acicularibus inaequalibus ma- eru 12-50 mm Tree 4-15 m tall, upper part of branches mi- nutely and densely pilosulous with spreading hairs. Leaves alternate or opposite. Petioles 1.2-3 cm long, densely tomentose; leaf blades oblong-ob- ovate, rounded at apex, acute, obtuse, or rounded at base, 7-14 cm long, (3.5-)4-9 cm wide, entire; lateral nerves 10-14 each side, elevated below, sulcate above, ascending at a 45? angle, anasto- mosing at the margins; midrib and lateral nerves below densely pilosulous with spreading hairs; mid- rib above sulcate, densely tomentellose; lateral nerves above slightly pubescent. Inflorescence lat- eral or terminal, simply 3-4-umbellately flowered, (2-)4-5.5 cm long; peduncle 0.5-3.5 cm long, densely pilosulous with spreading hairs; bracts at base of pedicels linear-lanceolate, acute, 3.5 mm long, densely tomentose without, sparsely ap- pressed within; pedicels (6-)10- mm long, densely pilosulous. Sepals 4, equal, enclosing the flower in preanthesis, ovate, acute, 5-8 mm long, 2.5-5 mm wide, densely cinereous-pubescent with- out, more sparsely appressed pubescent within, the margins thickened, densely pubescent. Anthers ob- long-elliptic or oblong-lanceolate, including the awn -2 mm long, densely hirsutulous, laterally de- hiscent, the connective prolonged into a hirsutulous subobtuse awn 0.7-1 mm long; filaments 1.2-2.8 mm long, equaling or longer than the anther. Ovary ovoid or suborbicular-ovoid, 3.5-4 mm long, 3 mm wide, brown-hispidulous, 4-angled; styles 4, free, divergent, glabrous, 1.5-2 mm long. Capsule ovoid, 2.2-3 cm long, densely covered with usually red, acicular, unequal, straight spines, the longer ones 12-50 mm long with appressed setulose projections along their length, the shorter bristles 3-4 mm long, overlying surface of dense, pale hispidulous hairs; mesocarp reddish. Seed golden, reddish proximally. 1576 Annals of the Missouri Botanical Garden Iz Z ^ E : NES JAEN URE Sloanea davidsei.—A. Habit. —B. Flower and pedicel, in anthesis with bracts.—C. Portion of infructescence. — D. Portion of lower leaf surface. —E. Stamen, with lateral dehiscence, ventral view. —F. Stamen, i istil. lateral view.—G. Pisti Volume 75, Number 4 1988 Steyermark 1577 Venezuelan Guayana Flora—VI Paratypes. | VENEZUELA. T. F. AMAZONAS: Rio Guainía Merca San Miguel and Maroa, 100-400 m, 30 June 1959, Wurdack & Adderley 43260 (MO, NY); DEPT. RÍO NEGRO: selva pluvial por las orillas del Medio Pacimoni, 1?40'N, 66°35'W, 3 Dec. 1984, S Clark 9140 (MO, VEN); igapó forests along Cano Cuweje, of San Carlos, 1%56'N, 67%03'W, 119 m, 4 Apr. 1980, Clark 7480 (MO); laja on right bank of Cano Cupueni, opposite mouth of Río Atabapo, 120-130 m, 12 Nov 3, Maguire, Wurdack & Bunting 36219 (MO, NY). Common names. |. Uruch, onoto rebalsero; ono- tillo, urucurana. The common names are derived from the su- perficial resemblance to the spinose fruits of Bixa orellana and B. urucurana. The species is related to Sloanea kuhlmanii Ducke of Amazonian Brazil but differs in the absence of a corrugated recep- tacle, in the shorter stamens with densely hirsu- tulous anthers and their appendages, in the entire leaf margins nonemarginate at the apex, in the densely pubescent peduncles and pedicels, and in the densely pubescent lower leaf surfaces, espe- cially on the nerves. The younger leaves of Davidse 27733 and W uardaek & Adderley 43260 have a more al tum on the lower leaf surface. The pubescence of the older, more mature leaf blades tends to be more sparse and shortly as- cending or spreading along the midrib and/or nerves, whereas the lower surface itself, including the tertiary veins, remains glabrous or essentially so. Sloanea longiaristata Steyerm., sp. nov. TYPE: Venezuela. T. F. Amazonas: Depto. Atabapo, forested slopes, Cerro Marahuaca, 1-2 km N of Sima Camp, 3?43'N, 65°31'W, 1,100 m, 8-9 Mar. 1985, Ronald Liesner 18455 (ho- lotype, VEN; isotype, MO). Figure 4. Arbor E 15 m, an juvenilibus tomentosis; petiolis 0.7-2 c e tomentosis; foliorum laminis late | uocis a 'llipico -oblongis apice rotundatis vel breviter iin asi acutis vel ira as 10-15 cm longis -11 cm a s praeter utrinque costam mediam nervos ad dios iae vel subundulatis; ¡Morencontil lateralibus racemosis 1.5-4.5 cm ntib in aristam led be 1.3-2 mm longam pro- ducto; filamentis 1.5-2.1 mm longis dense primis on stylis quadripartitis vel basi vix connatis 2.8-4 mm longis prae- ter basim adpresso-hispidulam glabris; capsulis ignotis. Tree 5-15 m high, the young branches tomen- tose. Leaves alternate, crowded at the summit of the branches. Petioles 0.7-2 cm long, densely to- mentose; leaf blades coriaceous, broadly oblong- obovate or elliptic-oblong, rounded at summit, or with a short broadly obtuse projection, cuneately acute or subacute at base, 10-15 cm long, 5-11 cm wide, the leaf surfaces glabrous, but the upper and lower midribs tomentose and the lower lateral nerves slightly pubescent, the margins entire to subundulate; lateral nerves 6-8 each side, elevated below, slightly impressed above, ascending at 45- °, terminating mainly at the margins or un- branched before reaching margins; tertiary vena- tion forming obliquely parallel connecting veins with the lateral nerves of the lower side, slightly elevated or impressed below. Inflorescence lateral, racemose, 1.5-4.5 cm long, 2.5 cm wide, 3-8- flowered, rachis moderately hirtellous with spread- ing hairs. Peduncles 5-7 mm long, densely hir- sutulous. Bracts subtending pedicels, alternate, linear, 2 mm long, densely hirtellous. Pedicels 4- 15 mm long, densely hirsutulous with spreading hairs. Sepals 4, reflexed, lanceolate, acute, 2.5-4 mm long, densely hirsutulous without, less pubes- cent within. Anthers elliptic or ovoid-oblong, 0.7— 1 mm long, densely hirsutulous, laterally dehiscent, the connective prolonged into a conspicuously elon- gated glabrous awn 1.3-2 mm long; filaments 1.5- 2.1 mm long, densely hirsutulous, much exceeding the anther body. Ovary ovoid-subglobose or ovoid- oblong, 3-3.5 mm long, 2-2.5 mm wide, densely hispidulous; styles 4, deeply divided or connate at the base, 2.8-4 mm long, glabrous except in the appressed-pubescent basal portion. Young capsule 4-celled Paratypes. | VENEZUELA. T. F. AMAZONAS: 1-2 km SE and E of San Carlos, 20 km S of confluence of Rio Negro and Brazo Casiquiare, 1?56'N, 67°3’W, 120 m, 22 Apr. 1979, Liesner 6875 (MO, VEN). DEPT. ATABAPO: Cerro slopes along E branch of Cano Negro, 3?43'N, 65° 1,140 m, 21-22 Feb. 1985, Steyermark & Holst 130507 (MO, VEN This species is most closely related to Sloanea duckei C. E. Smith and S. rufa Planch. ex Benth. From S. rufa it differs in the longer anthers and longer awns, more elongated inflorescences and pedicels, broadly rounded leaf apices, glabrous low- er leaf surface, and styles more divided. From S. duckei it differs in the longer awns; longer fila- ments; pubescent petioles, midribs, and secondary nerves of the leaf blades; broadly obovate or oblong- obovate leaf blades acute or subacute at base; short- er peduncles; and 4 instead of 5-7 sepals. From 1578 Annals of the Missouri Botanical Garden 7 B c E 4. Sloanea longiaristata.—A. Habit. —B. Sepal, exterior view.—C. Pistil, Miti free styles. —D. Pistil, showing coherent styles. —E. Stamen.—F. Portion of bristly hairs of young frui Volume 75, Number 4 1988 Steyermark 1579 Venezuelan Guayana Flora—VI other related species belonging to section Brevi- spicae C. mith, such as S. stipitata Spruce ex Benth. ad S. robusta Uittien, it is distinguished by the elongate awns longer than the anther body, shorter petioles, and differences in the leaf blades. The collection Liesner 6875 from the region of San Carlos is doubtfully assigned to this taxon. It differs in having the styles more united and smaller, glabrous leaf blades with the apices more obtusely prolonged rather than rounded and with more- undulate margins. Sloanea merevariensis Pittier ex Steyerm., sp. nov. TYPE: Venezuela. Bolivar: Alto Rio Guaña (Merevari) near Brazil frontier, without date, Felix Cardona 1051 (also numbered 1053, 1055) (holotype, F; isotypes, US, VEN). Fig- ure 5 Arbor; ramulis minute pubescentibus; petiolis 2.5-3.3 m longis, minute a adpresso- puberulentibus; foliorum la- basi acutis -13 cm subtus pilosula pilis laxis munita; nervis u puberulis pilis patentibus praeditis; nervis lateralibus ut- florescentiis umbellatis v moderate adpresso- -puberulentibus; sepalis quattuor re- flexis ovatis acutis 4-4.5 mm longis; antheris 1.5-1.6 mm longis hispidulis lateraliter dehiscentibus, connectivo in acumen late lanceolatum acutum 0.3-0.4 mm longum producto; filamentis 2 mm longis pilosulis; stylis quadri- partitis erectis supra medium glabris infra medium minute puberulis. Capsula ignota Tree with minutely pubescent branches. Leaves alternate. Petioles 2.5-3.3 cm long, minutely ap- pressed-puberulent except on the more densely pu- bescent canaliculate upper side; leaf blades ob- ovate, rounded at apex, acute at base, 8.5-13 cm long, 4-6.5 cm wide, glabrous on the upper and lower surfaces except densely tomentose above on the slightly sulcate midrib, this moderately laxly pilose below and laxly puberulous on lower lateral nerves, the nerves 9-10 each side, ascending at 45-50° and anastomosing 3-5 mm from margins; tertiary venation minutely reticulate, elevated both sides, with subparallel obliquely transverse con- nections with the lateral nerves. Inflorescence lat- eral and terminal, umbellate or corymbose-race- mose with 3-5 flowers. Peduncles 6-7 mm long, moderately appressed-puberulent; pedicels 7-8 mm long, moderately puberulent. Sepals 4, reflexed, dark in drying, subequal, 4-4.5 mm long, 3-3.5 mm wide at base, moderately pubescent without, sparsely appressed pubescent within, the margins pubescent. Anthers lance-elliptic, 1.5-1.6 mm long, hispidulous, laterally dehiscent, the connective pro- longed into a short lanceolate, subacute, minutely puberulent awn 0.3-0.4 mm long; filaments 2 mm long, pilosulous. Ovary subglobose, 0.7 mm long; styles 4-parted, erect, 1.5 mm long, minutely pu- berulous in lower half, glabrous above. Capsule unknown. This species was distributed by C. E. Smith as Sloanea laurifolia (Benth.) Benth. As indicated in another part of the text, S. laurifolia is treated by the present author as having less variability than that allowed by Smith. Sloanea merevariensis may be differentiated by the obovate, rounded leaf blades having the upper midrib densely tomentose. Sloanea parvifructa Steyerm., sp. nov. TYPE: Brazil (near Venezuelan border). Serra da Ne- blina, Rio Negro, Rio Cauaburi, Rio Maturacá, between Missào Salesiana and Serra Pirapucú, 800-1,000 m, 23 Jan. 1966, Nilo T. Silva & Umbelino Brazáo 60865 (holotype, MO; isotype, NY). Arbor 10 m, ramulis prope aa dense minuteque pubescentibus; petiolis 1.5-2.7 cm longis minute pube- rulentibus; foliorum lamin nis s obovato- -spathulatis, apice ro- tenuibus 1.5-3 mm longis antrorse pubescentibus. Tree 10 m tall, the branches minutely pubescent toward the summit. Leaves alternate. Petioles 1.5- 2.7 cm long, densely minutely puberulent; leaf blades obovate-spathulate, rounded or broadly ob- tuse at apex, narrowed at the acute base, 13-19 cm long, 4.5-7 cm wide, obscurely repand or en- tire, except for the sparsely pubescent upper and lower midribs, glabrous on both surfaces; lateral nerves 7-8 on each side, ascending at 50—60°. Flowers not seen. Infructescence aay racemose, bearing 4-5 fruits, the fruiting axes 3-3.5 cm ong, 1.5 mm wide, widely T E 88 from stem, aal puberulous with spreading-ascending tri- chomes; peduncle 5-9 mm long; fruiting pedicels slender, 8-10 mm long, densely puberulous with spreading-ascending hairs. Capsules relatively small, the valves 9-10 mm long, 7 mm wide; bristles slender, purple, 1.5-3 mm long, antrorsely pu- bescent. Seeds oblong, 7 mm long, 4.5 mm wide. — 1580 Annals of the Missouri Botanical Garden n ( MAI. fit f^ M h RAE fell (t ci Ww Pee he Mat Am "n 17 k, Sloanea merevariensis.— 4. Habit A ny —B. Flower, at anthesis.—C. Pistil.—D. Stamen.—E. Base of leaf blade, adaxial view, with petiole. —F. Lower portion of leaf blade. From the related Sloanea duckei C. E. Smith of Amazonian Brazil, the new species differs in having densely pubsadent petioles, a puberulous lower midrib, and lateral nerves ascending at a greater angle. From S. spathulata C. E. Smith of Territorio Acre, Brazil, it differs by having longer infructescences with longer pedicels, much shorter bristles, shorter petioles, more strongly ascending lateral nerves, and obliquely transversely connect- ing tertiary nerves straighter, less branched, and less prominent. Sloanea sipapoana Steyerm., sp. nov. TYPE: Venezuela. T. F. Amazonas: Cerro Sipapo (Pa- Volume 75, Number 4 1988 Steyermark 1581 Venezuelan Guayana Flora—VI ráque), Camp Savanna, 1,500 m, 15 Dec. 1948, Bassett Maguire & Louis Politi 27674 (holotype, MO; isotype, NY). Figure 6. ad 15 m alta ubique plerumque spud foliis amis: petiolis 0.9-1.5 cm longis; foliorum lamini riaceis lanceolatis vel lanceolato-ellipticis apice Ad acuminatis basi obtusis vel subacutis majoribus 8.5-13.5 elevata; nervis lateralibus supra vix manifestis subtus paul- lo elevatis utroque latere 11-13; venatione tertiaria subtus uer reticulata; infructescentia pu pedunculo 2.5- 3.5 cm longo glabro; pedicellis Po I vel sparsim puberulentibus; capsulis Pra e longis minute puberulentibus. Tree to 15 m tall with glabrous branches. Leaves alternate, glabrous throughout. Petioles 0.9-1.5 cm long; leaf blades coria eolate or lance- elliptic, slenderly acutely acuminate at apex, obtuse to subacute at base, the large ones 8.5-13.5 cm long, 2.5-5 cm wide, entire; midrib elevated above and below; lateral nerves 11- each side, as- cending at 25-35", not reaching margin, scarcely evident above, slightly elevated below, anastomos- ing with tertiary venation 2-7 mm from margin; tertiary venation finely reticulate on lower surface, scarcely evident on upper surface. Infructescence lateral. Peduncle 2.5-3.5 cm long, 1-1.5 mm iam., glabrous. Pedicels 1.5-2 cm long, glabrous to sparsely puberulent. Capsules unarmed, 4-valved, 2-2.5 cm long, the surface densely and minutely puberulent. Seed oblong, 1.1 cm long, 0.8 cm wide. Paratype. | VENEZUELA. T. F. AMAZONAS: Cerro Sipapo (Paráque), upper Cano Negro and nn branch (north) of Caño Profundo, 1,455 m, 10 Jan. 1949, Maguire & Politi 28266-A (MO, NY). Apparently a member of section 4, Corymbo- Racemi C. E. Smith, most closely related to Sloanea oppositifolia Spruce ex Benth. (= S. laurifolia (Benth.) Benth.), from which the new species differs in the alternate, coriaceous leaf blades with an elevated midrib, inconspicuous lateral nerves, and finely reticulate tertiary venation of the lower sur- ace. Sloanea steyermarkii C. E. Smith Sloanea steyermarkii C. E. Smith, described from a collection taken from the summit of Auyan- tepui, a massive sandstone table mountain of the Venezualan Guayana, appeared to be a distinct, isolated endemic species, characterized by the dense, congested, subsessile inflorescence; coriaceous, ru- gose leaf blades rounded at the apex; densely buff- or ferruginous-tawny tomentose lower leaf surfaces with the tomentum completely covering the midrib, nerves, and lower surface; and small, densely brist- ly fruits with bristles 2-4 mm long. Additional collections obtained from the summits of various other sandstone table mountains throughout the Venezuelan Guayana have similar patterns of leaf size, shape, rounded apex, con- gested inflorescences, and small fruits with short bristles but differ chiefly in the density of the pu- bescence on the lower leaf surface. In typical Sloanea steyermarkii, known from the mountains situated within the eastern drainage of the Rio Caroni and its tributaries (Ptari-tepui, Chimanta- tepui, Auyan-tepui, and Uaipan-tepui of Estado Bolivar), most or all of the leaves retain dense abaxial tomentum. However, a specimen collected from Uaipan-tepui (Koyama & Agostini 7185) has only the youngest leaves with a dense buff tomentum completely covering the lower surface, whereas the older ones have lost most of the dense tomentum and retain only mere traces on the lower surface and along the midrib and nerves. Westward on the summits of the sandstone mountains, beginning with the western drainage of the Rio Paragua of Cerro Guaiquinima and the Rio Caura of Cerro Guanacoco and the Meseta del Jaua, the indument of the lower surface is less prominent and is manifested only by pale sparse puberulence along the midrib and some of the lateral nerves. This tendency toward glabrity con- tinues westward on the summits of the sandstone mountains of the Territorio Federal Amazonas (Cerro Huachamacari, Yapacana, Aracamuni, Au- tana, and Sipapo). The taxa that have previously been treated as species, Sloanea autanae from subspecies with glabrous or glabrate leaf variations a more widely ranging S. steyermarkii geo- graphically isolated in the western portion of the Guayana Highland. KEY TO THE SUBSPECIES OF SLOANEA STEYERMARKII la. Lower leaf surface chiefly densely tomentose beneath, at least on the younger leaves, com- pletely covering the surface and nerves with a dense buff- or ferruginous-tawny tomentum; plas of Ptari- ep westward to Chimantá Mas- sif and Auyan-t ve does Bolivar ......................... aD" ISP ermarkii ei steyermarkii Lower leaf "dde shoe or scattered pubescence; lower m nd/or lat- eral nerves more or less Bip nene to glabrous a 2a. Leaf blades entire, mainly rounded or 1582 Annals of the Missouri Botanical Garden FiGURE 6. Sloanea sipapoana.— A. Habit. D. Leaf base, with portion of petiole, abaxial view.—E. Leaf blade showing apex, adaxial v broadly obtuse at Sede el e subsp. a 2b. Leaf blades. aa -repand, often wed to a less obtuse a aS. s steyermarkii ‘subsp. ¢ autanae Sloanea steyermarkii C. E. Smith subsp. jauaensis (Steyerm.) Steyerm., stat. nov. Sloanea jauaensis Steyerm., Bol. Soc. Venez. Ci. Nat. 33(132-133): 353. 1976. Sloanea Aen Steyerm. var. minor, Bol. Soc. Venez. Ci. Nat. 33(132-133): 354. 1976. Sloanea i Steyerm., Pittieria No. 7: 15. 1978. —B. Capsule, lateral view, dehiscing.—C. M Ae ventral view.— pecimens examined. VENEZUELA. BOLIVAR: Meseta Mi lows, Cerro Jaua, 4?48'50"N, 64?34'10"W, a sur-oeste, 1,810-1,880 m, Steyermark et al. 109841 (type of Sloanea jauaensis, VEN); Meseta del og Cerro s ies 50"N, 64*34'10"W, porción sur-oeste, 1,810- 0 m, Steyermark et al. 109695 (type of S. jauaensis ; Cerro Guaiquinima, Río Paragua, be m, B mu et al. 109760 (NY, 2 M F. AMAZONAS: Cerro Huachamacari, Summit Camp, 1 m, Maguire et al. 30088, 30097 (MO, NY); Cerro ees 3°45'N, 66°45'W, 1,000-1,200 m, Steyermar 103186 (type of S. yapacanae, By Cer 1,200 m, Maguire et al. 30737 (NY); Cae Sipapo, S basin, 1,835 m, Maguire & Politi 28684 (NY). xs d ES CE Volume 75, Number 4 1988 Steyermark 1583 Venezuelan Guayana Flora—VI A sterile collection from the summit of Cerro Aracamuni (Liesner & Carnevali 22722) probably is to be referred to this subspecies. Sloanea steyermarkii C. E. Smith subsp. au- tanae (Steyerm.) Steyerm., stat. nov. Sloanea autanae Steyerm., Pittieria 7: 13. 1978. Specimens examined. VENEZUELA. T. F. AMAZONAS: Cerro Autana, 4%52'N, 67°27'W, 1,230-1,270 m, Stey- ermark 105222 (type of S. autanae, VEN). Sloanea subpsilocarpa Steyerm., sp. nov. TYPE: Venezuela. T. F. Delta Amacuro: bosque plu- vial, E de Rio Grande, ENE El Palmar, near limits of Edo. Bolivar, 29 Nov.-18 Dec. 1964, Luis Marcano-Berti 447 (holotype, VEN; iso- types, MO, NY). Figure 7. Arbor 28-30 m; petiolis 0.6-3 cm longis; foliorum laminis oblongis vel oblongo-ellipticis apice obtuse acutis obtusis vel raro apc m hes obtusis vel rotundatis 8- ongis 4-7 c Mon aar ubique glabris; nervis lateralibus fusis wass -6(-8) venatione ter- tiaria supra prominente sin ia inflorescentia (post an- thesim) subpaniculata 2-5-flora; pedunculis 1.5-2 cm longis tomentellis; pedicellis post anthesin 1-2 cm longis fructiferis 2-2.5 cm longis tomentellis; sepalis quattuor coriaceis late ovatis acutis dense fulvo- ag ee 10- En cm longis 9 mm latis; antheris 2 mm longis adpres pubescentibus prope poros sub apicales ats Pres centibus, connectivo in acumen —— obtusum ga brum 1.5 mm longum producto; filamentis 0.5-1 m qe sparsim adpresso- Ea irom vel glabris; ovario m longo dense fulvo-tomentoso; stylis 2 mm longis Sve. aN 5 ; capsulis inermibus vel sparsim breviter tuberculatis subglobosis 4.5 x 4.5 cm. Trees 28-30 m tall. Leaves alternate. Petioles 0.6-3 cm long; leaf blades oblong or oblong-elliptic, acute or obtuse, rarely rounded at apex, obtuse or rounded at base, 8-16 cm long, 4-7 cm wide, entire, glabrous both sides; midrib sulcate above, elevated below; lateral nerves 4—6(—8) each side, impressed above, elevated below, branching and anastomosing 4-9 mm from the margins, ascend- ing at 50-60; tertiary venation prominently re- ticulate and subelevated above, reticulate and sub- elevated below. Inflorescence (past flowering) subpaniculate, terminal or lateral, 2—5-flowered; peduncles (past flowering) 1.5-2 cm long, 1.5 mm wide, 2-2.5 cm long (fruiting), brown-tomentose; bracts ovate, acute, 4-5 mm long, tomentellous on both sides; pedicels (past flowering) 1-2 cm long, 2-2.5 cm long, 3-3.5 mm wide (fruiting); sepals 4, coriaceous, persistent, broadly ovate, acute, 10-13 mm long, 9 mm wide at base, inner surface blackish, sparsely pubescent basally, dense- ly brown-tomentellous without. Anthers 2 mm long, 0.7-0.8 mm wide, appressed-pubescent, dehiscing laterally by subapical pores, the connective pro- longed into a glabrous, lanceolate, obtuse awn 1.5 mm long; filaments shorter than and about width of anther, blackish, 0.5-1 mm long, 0.5 mm wide, sparsely appressed to glabrous. Ovary subglobose- ovoid, 5 mm long, 5 mm wide, obtusely angled, densely brown-tomentose; styles 2 mm long, dense- ly brown strigillose. Fruit unarmed or with a few scattered short, obtuse, glabrous tubercles 0.5-1 mm long on a surface covered with dense brown tomentum intermixed with minute white hairs 0.1- 0.2 mm long, subglobose, 4.5 X 4.5 cm; mesocarp ligneous, 6-8 mm thick. atypes. VENEZUELA. T. F. DELTA AMACURO: E de Rio Cana. ENE of El Palmar, near limits of Edo. Bolivar, Upata (sobre la carretera nueva Upata-San Felix), 9 Sep. 1966, Blanco 578 (VEN). Almidon. Common name. This species has been confused with Sloanea laurifolia (Benth.) Benth., from which it differs in the larger, thicker sepals, pubescent anthers with a longer, lanceolate awn, subpaniculate inflores- cence, larger bracts, and stouter peduncle and ped- icels. From S. latifolia (Rich.) Schum. (S. inermis Ducke) S. psilocarpa is distinguished by the short- er stamens with pubescent anthers; larger fruits with shorter, obtuse, glabrous tubercles; broadly ovate sepals broader at the base; and usually obtuse to rounded leaf bases. Sloanea wurdackii Steyerm., Venezuela. Bolivar: Rio Parguaza, just below sp. nov. TYPE: mouth), 115 m , dack & Joseph Monachino 41017 (holotype, MO; isotype, NY). Figure 8. r 6-20 m; petiolis 3.5-10 cm longis apana pu- erer admodum glabris; foliorum laminis late ob- longo-ovatis apice rotundatis hasi TP cutis vel ratindatis m latis integerrimis r rhachidi glabra vel sparsim puberulenti; pedicellis elon- atis 1 longis minute puberulen- oduct pubescentibus fructu immaturo oblongo 1 go praeter partem basalem 3-4 mm longam aede spinosis, spinis rigidis 2-3 mm longis. 1584 Annals of the Missouri Botanical Garden DAN ^ ITI ! SAND = ARAS A 4.5cm FIGURE 7. fruiting branch. — D. Tree 6-20 m tall. Branches glabrous. Leaves alternate. Petioles 3.5—10 cm long, essentially gla- brous but sparsely microscopically puberulent; leaf blades broadly oblong-ovate, rounded at the some- times mucronate summit, subacute to rounded at Sloanea EIS NM —A. Habit. —B. Flower, post anthesis with reflexed sepals. —C. Portion of Sta base, the larger ones 18-25 cm long, 9.5-15 cm wide, entire or subentire, glabrous both sides; mid- rib essentially glabrous below but with traces of minute puberulence, prominently elevated; lateral nerves 10-12 each side, ascending at 60-70", Volume 75, Number 4 1988 Steyermark Venezuelan Guayana Flora—VI 1585 FIGURE 8. connective of anther. anastomosing 2-7 mm from the margin, elevated below, impressed above; tertiary venation minutely reticulate, slightly elevated above. Inflorescences elongated, subpaniculate or laxly racemose, 2- together, 4-8.5 cm long, 2-5-flowered, the rachis glabrous or minutely sparsely puberulent; pedicels Sloanea wurdackii.—A. Habit.—B. Capsule with pedicel. —C. Pistil. —D. Stamen.—E. Apical elongated, recurved at apex, 1-4 cm long, the lowest 3.5-4 cm long, minutely puberulent. Sepals strongly reflexed after flowering, lanceolate, 2.5 mm long, 0.7-1 mm wide at base, minutely ap- pressed-pubescent both sides. Anthers linear-lan- ceolate, 0.9-1 mm long, densely appressed-pu- 1586 Annals of the Missouri Botanical Garden bescent, dehiscent near the apex, connective 1962. Bol. Soc. Venez. Ci. Nat. 23(101): 70, prolonged into a short acutely obtuse pubescent knob 0.1-0.2 mm long; filaments 2 mm long, densely appressed-pubescent. Ovary 3 x 3 mm, ensely short pubescent with tubercular processes 0.5-2 mm long; style stout, 2 mm long, appressed pubescent from base to apex. Young fruit oblong, rounded at apex, 1.5-1.7 cm long, 1 cm wide, unequally spinose-tuberculate with rigid stout spines broad at base, 2-3 mm long except absent in the basal 3-4 mm, the tubercles minutely appressed- pubescent; more mature fruit wine purple, 2 cm long, 1.5 cm wide, the spines slender-tipped. Paratype. | VENEZUELA. T. F. AMAZONAS: Dept. Ata- bapo, Cano Negro, rio arriba desde la confluencia con el Rio Cunucunuma, 3?40'N, 65%45'W, 200-210 m, 8 Feb. 1982, Steyermark et al. 126231 (NY, VEN). This species is characterized by elongated gla- brescent pedicels recurved at the apex; elongated, irregularly laxly subpaniculate-racemose inflores- cence; short-awned, densely appressed pubescent anthers much shorter than the densely pubescent filaments; stout appressed pubescent style; short, rigid, tubercular spines on a surface lacking tu- bercules in the basal portion; and general glabrity of stems and leaves. The fruiting structures show similarity to some members of sect. Paniculi C. E. Smith, such as Sloanea caribaea King & Urban ex Duss. However, the absence of material in full anthesis makes the placing of S. wurdackii un- certain. The large leaf blades and elongate petioles somewhat resemble those of S. schomburgkii Benth., but that species has longer anthers, which are longer than the filaments. Although no stamens are present on the para- type, it is referred to the new species because the paratype has the characteristics shortly spinose fruit with rigid slender spines, which are absent in the basal area; the same type of elongated slender, recurved pedicels in an elongated inflorescence; similarly elongated petioles; similar leaf size, shape, and nervation; and general glabrity of the stems, petioles, pedicels, and foliage. The main differences noted in the paratype are the fewer lateral nerves which are farther apart, the lowest ones more spreading at a lesser angle; the leaf bases more rounded or broadly obtuse; and the peculiarly pitted upper surface of the leaf blade LITERATURE CITED iin G. 1861. On oe Sloanea. J. Proc. n. Soc., Bot. 5: 62-7 Sur. c. E. 1954. The New World species of Sloanea (Elaeocarpaceae). Contr. Gray Herb. 175: 1-114. n: J. Steyermark, Flora del oe tepui. Acta i Venez, 2(5-8): 244, fig. 14 SmrTH, D. A. 1985. Pp. 90-92 in The Costa “Rican Species of Sloanea. Unpublished Thesis. Duke Uni- versity, Durham, North Carolina. STEYERMARK, J. A Novedades venezolanas del género Sivaned (Elaeocarpaceae). Pittieria 7: 13- 17. L. Marcano-BERTI. 1966. Una especie nueva de Sloanea. Bol. Soc. Venez. Ci. Nat. 26(110): 467, fig. 1 & C. BREWER-CaRIAs. 1976. La vegetación de la cima del Meseta de Jaua. Bol. Soc. Venez. Ci. Nat. 32(132-133): 349-354, figs. 16-17. BOMBACACEAE POCHOTA The following name was inadvertently omitted from the recent publication of this genus by Stey- ermark & G. W. Stevens (1988): Pochota robynsii Steyerm. & W. D. Stevens, om. nov Bombacopsis rep Robyns, Mem. New York Bot. : . 1967, not Jig ide coriacea (Mart. & Zucc.) oan & W. D. Ste LITERATURE CITED STEYERMARK, J. & W. D. STEVENS. 1988. Notes on Rhodognaphalopsis and Bombacopsis (Bombaca- ceae) in the Guayanas. Ann. Missouri Bot. Gard. 75: 396-398. 'THEACEAE In the Flora of the Venezuelan Guayana —1ll (Steyermark, 1987b), Bonnetia bolivarensis Stey- erm., B. guaiquinimae Steyerm., B. ptariensis Steyerm., and B. tristyla subsp. nervosa Steyerm. were inadvertently republished. They were pub- lished earlier in the same volume and year (Stey- ermark, 1987a). The holotype specimen remains the same in B. bolivarensis, B. guaiquinimae, and B. ptariensis. In B. tristyla subsp. nervosa, although the citation of the holotype was changed from Steyermark & Bunting 103153 to Maguire et al. 30632, the original holotype citation of Stey- ermark & Bunting 103153 must be retained. LITERATURE CITED STEYERMARK, J. 1987a. Flora of the Venezuelan Guay- Ann. Missouri Bot. Gard. 74: 102-103. lora of the Venezuelan Guayana — H. Ann, Missouri Bot. Card. 74: 647-650 TAXONOMIC REVISION OF THE CENTRAL AMERICAN LISIANTHIUS SKINNERI SPECIES COMPLEX (GENTIANACEAE)! Kenneth J. Sytsma? ABSTRACT e Lisianthius skinneri alas aug d species complex consists of six closely related species in Pana ri. The last three species are described as and contrasted with genetic due (DNA and isozymes) í presented. suma is compared ong Spe Noto z sal divergence within the species complex is not correlated with molecular divergenc The Lisianthius skinneri (Gentianaceae) species complex is a small, geographically restricted, and interrelated assemblage of taxa. Lisianthius skin- neri ranges widely throughout Central America but is patchily distributed and shows much ecological and morphological variability. Five endemic species in isolated central Panamanian forests and humid coastal sites exhibit divergence from L. skinneri in habit, morphology, breeding system, and eco- logical tolerance. A taxonomic revision of the Lisi- anthius skinneri species complex based on floral and vegetative morphology is presented here. A biosystematic and evolutionary analysis of the Lisi- anthius skinneri species complex using breeding T E and cladistic s ci (Syts- ma, in , and results of isozyme (Sytsma & Schaal, 0854) and DNA (Sytsma n Schaal, 1985b) studies is presented elsewhere. Lisianthius P. Browne is an exclusively neo- tropical genus in the Gentianaceae. Lisianthius and a number of related neotropical shrubby genera form a distinctive but rather diverse group collec- tively known as the “‘lisianthioid gentians." Lisian- thius has been broadly interpreted in the past to include all these lisianthioid genera, usually as Li- sianthus Linnaeus (1767) or Lisyanthus Aublet (1775), orthographic variants of the accepted name Lisianthius P. Br. (1756) (Taxon 3: 242. 1954). he most comprehensive and recent taxonomic treatment of the Gentianaceae (Cilg, 1895) places Lisianthius and Macrocarpaea in the tribe Gen- tianeae, subtribe Tachiinae. The other lisianthioid genera were relegated to the Helieae. The taxon- omy of these lisianthioid gentians is in a state of chaos, with only Lisianthius (Weaver, 1972) and Macrocarpaea (Ewan, 1948; Nilsson, 1968) ad- equately monographed. A multidisciplinary study is now beginning on these lisianthioid genera (Nils- son, 1970; Maas et al., 1984; Maas, 1985). Many of these genera are confined to poorly accessible high-elevation peaks, thus explaining the small number of available specimens and the poor or incomplete nature of the few existing ones (Sytsma, 87) The genus Lisianthius consists of 30 species of woody or semiwoody Gentianaceae almost totally confined to Central America and the Greater An- tilles. One species, L. seemannii (Griseb.) O. Kuntze, ranges into northwestern Colombia. The center of diversity for the genus is in Guatemala and Mexico, which together have 12 species. The genus exhibits a high degree of endemism. Jamaica has eight species, all endemic. Panama has seven species, five of which are endemic. Lisianthius species found ! I thank Peter H. Raven and Barbara A. Schaal for their continued support for this study; Robert Dressler, Sandra Kna Sytsma and Lucy aylor for artwor app, Thomas Antonio, Clement Hamilton, and Bru k. The assistance of staff at the following herbaria in providing loan al is gratefully sa DUKE, F, MO, NY, US, and WIS. Special thanks goes to Willia eld; and Jackie Stein for assistance in the am D e who m possible the fieldwork in Panama through the Flora of Panama Project (NSF a BSR-830542 ? Botany Department, University of Wisconsin, Madison, Wisconsin 53706, U. S.A ANN. Missouni Bor. GARD. 75: 1587-1602. 1988 1588 Annals of the Missouri Botanical Garden in Cuba (two species), Hispaniola (one species), and Puerto Rico (one species) are all endemic to each island. Weaver (1972) monographed Lisianthius in detail and gave a much needed intrageneric classification scheme. Chromosome numbers for all Lisianthius species previously examined (20 out of 30) are identical. Two sections were recognized: Omphalostigma and Lisianthius. Section Om- phalostigma (Griseb.) Weaver is composed of only two species, L. saponarioides Cham. & Schlecht. and L. meianthus Donn. Sm., both exhibiting dis- tinctive salverform corollas and equal filaments in- serted near the apex of the corolla tube. Section Lisianthius possesses basically tubular or funnelform corollas and unequal filaments in- serted in the lower half of the corolla tube. This section is composed of subsect. Herbacei Weaver, annual or perennial suffrutescent herbs with de- terminate main axes, and subsect. Fruticosi Weav- er, perennial subshrubs or shrubs with indetermi- nate main axes. Subsection Fruticosi is further divided into three series (Longifolii Weaver, Exserti Weaver, Umbellati Weaver) based on inflores- cence architecture, exsertion of stamens and styles, and pollen grain reticulation. With its 18 species, series Longifolii is the largest species group in the genus and contains the Lisianthius skinneri species complex of Panama. 'THE LISIANTHIUS SKINNERI SPECIES COMPLEX BIOGEOGRAPHY AND ECOLOGY Seven species of Lisianthius occur in Panama. Four of these already had been known (Elias & Robyns, 1975) and three are described as new in this paper. Of the seven Panamanian species of Lisianthius, six (excluding L. seemannii (Griseb.) Kuntze) form an unusual and distinctive interre- lated species assemblage, hereafter referred to as the Lisianthius skinneri species complex. Lisian- thius seemannii, ranging from Costa Rica to north- western Colombia, is not especially closely related to the L. skinneri species complex, although it is likewise placed in series Longifolii of subsect. Fru- ticosi. Lisianthius seemannii has much smaller, usually ovate leaves and vegetatively most closely resembles the Jamaican L. longifolius rather than members of the L. skinneri species complex. The PORE Co ER in L. seemannii, a multi- unique in series Longifolii. The long narrow ale tube and lobes, paler inside than out, also clearly separate L. seemannii from the L. skinneri species complex. Lisianthius skinneri (Hemsley) O. Kuntze is the widest-ranging species in the genus, but with a marked patchy distribution from southeastern Gua- temala to the Darién province of Panama. It occurs from sea level to elevations of near 1,500 m in moist to wet tropical forests. The Atlantic Coastal Plain populations in Guatemala, Honduras, and Panama are frequently situated at sea level, while the populations in Costa Rica and the interior of Panama are found at mid elevations or occasionally higher elevations. Lisianthius skinneri has not been collected on the Atlantic Coastal region of Nica- ragua or Costa Rica. It is not known whether this absence is due to poor collecting in these areas or whether it reflects an unusual geographic disjunc- tion. The distribution of Lisianthius skinneri is clear- ly correlated with moisture availability. The mid- elevation forests situated on both the Atlantic and Pacific flanks of the Cordillera de Talamanca ex- tending southeastward from Costa Rica into west- ern Panama and the Cordillera de San Blas and Serrania del Darién in eastern Panama receive abundant rainfall. Only on the more humid and wetter Atlantic side has L. skinneri been able to occupy lower elevation sites near or at sea level. e distribution of Lisianthius skinneri throughout its range is markedly patchy, with pop- ulations usually separated by many kilometers. This is especially evident in central Panama where large portions of the lower to mid elevations have been extensively explored and collected. Lisianthius skinneri is very local in occurrence, as seen in the distribution map of all known populations in Pan- ama (Fig. 1). Some gaps in this distribution will be eliminated with more collecting, especially towards high elevations and on the Atlantic watershed, but undoubtedly the patchy distribution of this species will remain Most populations occur in disturbed or partially open habitats. Forest edge sites along roads and ths are now more common because of increasing utilization of the lower elevation forest by people. Often populations are seen in more closed habitats such as tree fall gaps. Scattered groups of individ- uals are also occasionally encountered in undis- turbed forests but do not form the larger clumped populations typically seen in the more disturbed habitats. Population size varies from few to about a hundred individuals. Most populations, however, cover less than 50 square meters and are composed of about 50 flowering shrubs. In contrast to the wide-ranging distribution of Lisianthius skinneri, the other members of the L. skinneri species complex are all endemic to central Panama. The physiography of the central Panama mountain system is unusual because it is relatively Volume 75, Number 4 Sytsma 1589 y Lisianthius skinneri ------ 82°00' w ------ 78°00'-W L. skinneri jefensis weaveri habuensis . peduncularis r rr Fr e * * "C a o FIGURE 1. low in relief, sinuous in nature, and discontinuous in extent. The central Panama region is thus marked by isolated, cloud forest peaks (600-1,500 m) interrupted by more extensive low- to mid-elevation forests. Four of the additional five species in the L. skinneri species complex are restricted to sep- arate cloud forest peaks in this region (Fig. 1). These species include L. jefensis Robyns & Elias, L. peduncularis L. O. Williams, L. aurantiacus L forest-edge sites near beaches on the Atlantic side. Like the forest populations, these populations still occupy a high-rainfall environment. All species except L. aurantiacus and L. weaveri are comprised of few populations restricted to a small area. Lisianthius aurantiacus and L. weaveri are more widespread in distribution but are markedly patchy in occur- rence. cloud- MORPHOLOGY Morphological characters used to delimit species in the Lisianthius skinneri species complex are almost exclusively confined to features of the in- florescence and flower. Additional vegetative char- acters are occasionally divergent in the group and L. aurantiacus Distribution of the Lisianthius skinneri species complex in Panama. will also be presented. Only characters that are variable among members of the species complex, and hence of possible phylogenetic interest, will be presented here. A more formal description of all characters will be given in the next section. Mea- surements were made primarily on dried speci- P ined for populations of L. skinneri occurring out- side Panama. Habit. tiacus are true shrubs. in open sites is typically arboreal and can reach heights over 6 m. These trees are slender-stemmed and evenly branched to the top. In more closed forest environments, L. aurantiacus is a 2-3 m branched shrub, as are the remaining taxa. Oc- casionally, taller individuals of L. peduncularis are seen in the elfin forest atop Cerro Carocoral. Like most species of sect. Lisianthius subsect. Fruti- cosi, these species are not distinctly woody except at the base, even though they persist for years. They usually have a single main stem that is branched above, but totally unbranched flowering plants are seen in L. skinneri, L. weaveri, and L. jefensis. The main shoot has an indeterminate All species except Lisianthius auran- isianthius aurantiacus 1590 Annals of the Missouri Botanical Garden growth pattern, and the lateral shoots ("flowering branches") have a determinate growth pattern. 8 P Leaves. Two types of leaves are evident in the Lisianthius skinneri species complex, true fo- liage leaves and lateral shoot leaves, which can be a problem if leaf characteristics are used in a sys- tematic analysis. This is especially so because many herbarium collections contain only lateral shoots without true foliage leaves (Weaver, 1972). The foliage leaves of all species except L. jefensis are obovate, membranaceous, glossy green above, and paler beneath. Leaves of L. habuensis have some- what undulating margins. Lisianthius jefensis leaves are subcoriaceous and usually smaller than those of the other five species. It is interesting that a number of other endemic species on Cerro Jefe also have subcoriaceous to fully coriaceous leaves (pers. obs.) Leaves on the determinate lateral "flowering branches" are smaller, with the tran- sition from foliage leaves to lateral shoot leaves abrupt. Two collections of higher elevation L. skin- neri from Volcán Arenal, Costa Rica (Wilbur & Stone 10257, Lent et al. 3321) appear to have thickened leaves as well, although Weaver (1972) did not mention it in his monograph. Lateral shoot leaves are ovate to lanceolate and merge into the foliaceous inflorescence bracts. Lateral shoot leaves are absent from L. aurantiacus because flowering branches are reduced to single or paired axillary flowers. Small floral bracts are present, however. Inflorescences. The inflorescence of Lisian- thius is very difficult to interpret but is an important characteristic in determining phylogenetic rela- tionships within the genus. Architecture of the in- florescence varies considerably in Lisianthius, and Weaver (1972) set forth a probable evolutionary . Fruticosi is characterized almost entirely by mew branching with each division terminated by dichasia. The dichasia are axillary or, more often, both terminal and axillary on de- terminate lateral shoots. Foliaceous leaves are lo- cated below the lowest trichotomy and the upper divisions are subtended by bractlike appendages. The various arrangements of inflorescences in the L. skinneri species complex are depicted sche- matically in Figure 2. Lisianthius skinneri, L. inside and L. jefen- sis are characterized by a ternate compound di- chasium in each axillary d (Fig. 2a). This inflorescence type comes closest in the L. skinneri complex to the primitive multi-compound dichasium that Weaver (1972) postulated for the genus. Simple reduction to a twice-compound di- chasium gives the inflorescence type seen in L. peduncularis (Fig. 2b). A further reduction of two lateral dichasia (Fig. 2c), or, alternatively, reduc- tion in the terminal dichasium (Fig. 2d) generates the two reduced inflorescences commonly exhibited y L. habuensis. Reductions of the L. peduncu- laris type to a single flower, but retaining the three sets of bracts, gives the three-bracted nodes sub- tending a one-flowered axillary inflorescence some- times evident in L. aurantiacus (Fig. 2e). Lisian- thius aurantiacus, however, usually has two levels of bracts and either one or two flowers in each axillary inflorescence (Fig. 2f, g). Calyx. The calyx is variable not only among the species but also within Lisianthius skinneri. Populations of L. skinneri near sea level in Gua- temala and Honduras exhibit the largest calyces in the species. They range from 6 to 8 mm, with the calyx lobes ranging from 3 to 6 mm long. The smallest calyx is seen in populations in the interior of Costa Rica and Panama. Here the calyx ranges from 5 to 6 mm, with the calyx lobes 2 to 4 mm, and is generally appressed and carinate. The ca- lyces in L. weaveri are similar to those in L. skin- neri but are stouter, spreading, and only somewhat ridged. . peduncularis and L. aurantiacus are very long: 10-13 mm and 13- 15 mm, respectively. The calyx lobes of these two species are exceptionally long, averaging about 34- 76 of the total length. Lisianthius jefensis also has a longer, more attentuate calyx than L. skinneri. The calyx in L. habuensis is similar in size to L. calyces in skinneri (6-8 mm long) but with longer dark green lobes (4-5 mm long) that are strongly carinate and scarious on the margins Corolla. The structure of the corolla in all species is basically uniform (Fig. portion of the corolla enclosing the ovary is con- stricted into a narrow tube. The stamens are in- serted on the inside of the esla tube at the distal end of the constriction. Immediately above the insertion of the stamens the corolla flares out, often abruptly. Just below the lobes the corolla tube is again constricted, but usually not to the degree as in the basal region. Lisianthius skinneri shows extreme variation in the proportions of the corolla. The Guatemalan plants (including the type speci- men) have long (5-5.4 cm), narrow, and mem- branaceous-textured yellow idus The corolla lobes are greenish, broad, and short acuminate. Plants from Honduras, Costa Rica, and the interior of Panama closely resemble the Guatemalan plants but have a broader, more inflated corolla tube, e basal Volume 75, Number 4 1988 Sytsma 1591 Lisianthius skinneri A - k. AA AA O Ñ @ FIGURE 2. the nd spec e-g. L. aurantiacus markedly so in the El Llano—Carti road populations in Panama. The two previously cited Costa Rican collections from Volcán Arenal (Wilbur & Stone 10257, Lent et al. 3321) are aberrant in having waxy, thick-textured corollas. The five endemic species in Panama are all characterized by thick- textured, waxy corollas. The correlation of this unique corolla form to wetter higher elevations or to the humid Atlantic Coastal Plain (Lisianthius weaveri in part) suggests that this floral character might provide protection against extremely damp conditions and subsequent floral destruction. The high incidence of damaged and rotting, thin-tex- tured flowers on higher-elevation L. skinneri plants on Cerro Jefe and the continental divide near Cas- cajal supports this idea. Lisianthius jefensis and L. weaveri exhibit the smallest corollas. The corolla of L. jefensis is 2.5-4 cm long, with spreading and recurved lobes 4-6.5 mm long, while the corolla of L. weaveri is 2.6-4.5 cm long, waxy yellow and green tipped, and surrounds long.exserted styles. The corolla of L. habuensis is broad and oe b ba —ÀÓ O 0 © Schematic arrangement of the inflorescences in the ys skinneri species complex. Arrows represent probable e pathways for inflorescence types and do no s.—a. L. skinneri, L. jefensis, and L. w imply phylogenetic pcm of L. habue weaveri.— b. w: ode =E€. d. ensis 4.9-6.1 cm long with the corolla lobes triangular, smallish (3.5-5 mm long and wide at the base), dark green with cream margins, and slightly re- curved. The flowers of L. peduncularis and L. aurantiacus are very similar, both having very long corollas (4.8-6 cm and 5.5-7.8 cm, respec- tively) and long attentuate lobes (8-10 mm and 9-11 mm, respectively). The pumpkin orange to red corolla color and dark green corolla lobes of L. aurantiacus are strikingly different from any Panamanian Lisianthius. Androecium. Weaver (1972) used the place- ment of stamens as a diagnostic character in sep- arating Lisianthius skinneri from L. jefensis and L. peduncularis. In L. skinneri the stamens are inserted well above the apex of the calyx lobes, whereas in the other two species the stamens are inserted at or just above the apex of the lobes. This latter condition also applies to the new cloud-forest species, L. habuensis and L. aurantiacus. The use of this character as diagnostic is misleading because d. L. habuensis.—e. L. peduncularis.—f. L. aurantiacus. it does not reflect a difference in the placement of the stamen, but rather in the length of the calyx. All species in the L. skinneri species complex have the stamens attached to the inner corolla tube about the level of the apex of the ovary. Stamen length varies depending on corolla tube length, with most species having stamens protruding slightly from the corolla lobes. The stamens in L. jefensis are pro- portionally longer than in the other species and are conspicuously exserted. Anthers and pollen are 1592 Annals of the Missouri Botanical Garden FicurE 3. Flowers in the Lisianthius skinneri species complex.—a. L. skinneri. —b. L. jefensis—c. L. weaveri— whitish yellow in most L. skinneri individuals, greenish yellow in some individuals of L. skinneri and in the other species, except for L. aurantiacus, in which they are bluish green. Gynoecium. Style, ovary, capsule, and seed size, like the stamens, are correlated with flower size in general. Exceptions include Lisianthius je- fensis and L. weaveri, which both have longer exserted styles. Stigma shape and size are diag- Volume 75, Number 4 1988 Sytsma 1593 Lisianthius skinneri nostic characters separating out L. habuensis. All other species in the complex exhibit smooth, small stigmas (1-1.5 mm broad). The stigma in L. ha- buensis is unique and divergent in being large (2.5— 3 mm broad), conical, and rough textured. SYSTEMATIC TREATMENT The grouping of populations into formal taxo- nomic categories has two underlying purposes. The first and traditionally the primary purpose of such an endeavor is the construction of an information retrieval system that permits effective communi- cation about the populations (Raven, 1979). The taxonomic system constructed should preserve in- formation that enters the system and should min- imize its loss once taxonomic decisions have been made. The second purpose is the construction of a taxonomic system that best reflects the phylo- genetic or historical relationships of the popula- tions. Often, both goals of taxonomic decision-mak- ing cannot be simultaneously satisfie In delimiting species in the Lisianthius skinneri species complex, problems were encountered in grouping. populations into formal categories of "species." As previously indicated, DNA tech- niques indicated the paraphyletic nature of L. skin- neri (Sytsma & Schaal, 1985b). Populations of L. skinneri examined at the DNA level could be placed in two separate lineages that have been involved in separate speciation events. Cladistic analysis strictly following Hennig’s (1966) principles would raise a western Panamanian L. skinneri population KEY TO THE LISIANTHIUS SKINNERI SPECIES COMPLEX to the species level to preserve monophyly and to reflect better the apparent phylogeny of the species complex. This separation is unsatisfactory because it violates the first purpose of an effective infor- mation retrieval system. Although clearly this west- ern population is phylogenetically divergent by DNA analysis, it also clearly falls within the range of morphological variation of the widespread L. skinneri as predefined. Because of the broad dis- tribution and ancestral nature of L. skinneri, it is very likely that many clusters of populations in L. skinneri will be shown by DNA analysis to rep- resent divergent and independent lineages and by similar reasoning would also have to be accorded species status. In assigning formal names to clusters of popu- lations in the Lisianthius skinneri species complex, therefore, additional sources of data were used in addition to genealogical or phylogenetic interpre- tation. For example, L. jefensis is identical to an eastern population of L. skinneri by cladistic DNA analysis (Sytsma & Schaal, 1985b) but is main- tained as a distinct species derived from L. skinneri based on its peculiar morphology, lack of hybrid- ization, and endemic nature. The western Pana- manian population of Lisianthius skinneri, ap- parently a member of a lineage that has given rise to other species, is maintained as L. skinneri be- cause no changes in morphology, reproductive bar- riers, or habitat preference are outside the range of variation found in typical members of L. skin- nert. la. Leaves subcoriaceous; asia and styles conspicuously exserted, and long surpassing the corolla lobes f 3. Li endemic to Cerro Jefe, Pana isianthius poem = c the corolla lobes, or only the styles conspicuously e 2a. Corolla lobes = 7 mm long; calyx lobe . Leaves membranaceous; m and styles included, or slightly exserted and not conspicuously surpassing xserted. s > 6 mm lon nflorescence of 1-2 axillary flowers; corolla pumpkin orange to red; plant often treelike 6. Lis ianthius aurantiacus 3b. Inflorescence of twice compound dichasia; corolla yellow; plants shrubby . Lisianthius peduncularis 2b. Corolla lobes < 5 mm long; calyx lobes < 6 mm lo 4 long; inflorescence of simple dichasia, never pee k aw as tigm luted Lis a. Calyx — 4-5 mm .5-3 mm in diameter, conical, convolute 4b. Calyx lobes 4 (rarely 5) m 1-1.5 mm in diameter, rounded, smo 5a. Corollas 5-5.4 cm long, thin- iu ELTE styles slightly exserted past anthers anthius hobaensis m long; eee of twice or ternately aaa ae stigma l. Lisianthius skinneri 5b. Corollas 2.6-4.5 cm long, thick-textured; styles long exserted past ee . Lisianthius weaveri m . Lisianthius skinneri (Hemsley) O. Kuntze, Rev. Gen. Pl. 2: 429. 1891. Leianthus skin- i Hemsley, Biol. Centr. Amer. 2: 345. . L. skinneri (Hemsley) Perkins, Bot. inia (Syst.) 31: 492. 1902. TYPE: Guate- mala. Locality not given: Skinner s.n. (lec- totype, K, photographs, F, Lisianthus arcuatus Perkins, Bot. Jahrb. Syst. 31: 492. 902 PE: none designate Lisianthus scopulinus Robyns & Elias, Ann. Missouri 1594 Annals of the Missouri Botanical Garden Bot. Gard. 55: 62, fig. 2. 1968. TYPE: Panama. Veraguas: mouth of Rio Concepción, Lewis, Croat & Hawker 2799 (MO) Shrub or subshrub, 0.5-3 m tall; stem terete, erect, green. Leaves membranaceous, deep to light green, paler below; blades 4-24 cm long and to 7.7 cm wide, obovate to obovate-elliptic, sharply acuminate; the costa prominent, lateral veins 2- 3, ascending; petiole amplexicaul, winged. Inflo- rescences terminal or axillary, composed of ter- nately compound, laxly flowered dichasia; folia- ceous leaved below first division, bracted above; bracts opposite, lanceolate to linear, 2 mm long or longer. Flowers horizontal to nodding; calyx 5-7 mm long, the lobes 3-5 mm long, carinate dorsally, + scariously margined, stout or attenuate, ap- pressed or spreading; corolla tubular-funnelform, constricted apically and distally, the tube 5-5. cm long and to 1 cm broad, bright yellow, occa- sionally greenish-tipped, membranaceous, the lobes ovate, 2.5-5.5 mm long, 2.5-4 mm wide, yellow- ish green to dark green, often with cream margins, often recurved at tip, short to long acuminate. tamens inserted within corolla at apex of ovary; filaments 4.2—4.7 cm long, unequal in length, long- est one at least equaling or surpassing corolla lobes; anthers 2-3 mm long and ca. 1 mm wide, bilobate at the base; pollen whitish yellow. Style 4.7-5.1 cm long, usually slightly exserted past anthers; stigma peltate, 1-1.5 mm broad. Capsules fusi- form, 1.8-2.2 cm long, beaked by persistent style to 9 mm long; seeds asymmetrical, surface cor- rugated, to 0.75 mm long. Chromosome number, 8. Distribution. From sea level to 1,300 m in moist to wet tropical forests or roadsides, at middle elevations in Costa Rica and the interior of Panama, and along the Atlantic Coastal Plain of Guatemala, Honduras, and Panama. Lisianthius skinneri as delimited here encom- passes all populations of the L. skinneri species complex outside Panama and most populations at low to mid elevations in Panama. Based on distri- bution and DNA analysis, L. skinneri is the an- cestral species in the complex (Sytsma & Schaal, 1985b). The four endemic cloud-forest species that were examined by DNA analysis arose from at least two lineages within L. skinneri. The corolla is longer and thinner in texture (membranaceous) ened and waxy corollas like other species in this complex. Lisianthius skinneri as defined here is composed of populations showing moderate vari- ability (Weaver, 1972). This is not surprising, con- sidering both the widespread distribution and an- cestral nature of the species. This variation, however, presents considerable problems when species circumscription is attempted. Lisianthus scopulinus Robyns & Elias from Veraguas Prov- ince, Panama, has somewhat large calyx lobes (4— 5 mm long) for L. skinneri. It resembles L. skinneri in all other traits and is here merged with L. skin- neri as another geographically variable population. One population of L. skinneri from Costa Rica is unique in having thickened flowers and in occurring at higher elevations on Volcan Arenal (Wilbur & Stone 10257, Lent et al. 3321). This population was not visited during the study, so its status as a distinct element in the L. skinneri species complex is only tentative. Representative specimens examined. GUATEMALA. IZABAL: south shore of Lake Izabal, Proctor et 5 3049 Paz & G). HONDURAS. ATLANTIDA: a to La Ceiba, Davidse & Pohl 2194 (MO); San Alejo, Standley 7829 (F); Cuyamel, Carleton 564 (F). Costa RICA. ALAJUELA: 17-22 km beyond San Ramón to Cataratas, Almeda et al. 4301 (F, MO); middle slopes N side of Volcán Arenal, Lent 3868 (F, MO); Lent et al. 3321 (F); Villa ege near San Carlos, Smith 1884 (F, MO); 13.5 mi. E of Arenal, 6.5 mi. W of Fortuna, Wilbur & Stone 10257 (DUKE, F, MO, NY, US). GUANACASTE: Volcán Miravalles, Burger & Gentry 9126 (F); Hacienda Santamaria, Dodge et al. 6320 (F, MO); lower slopes Cerro La Giganta [Cerro Miravalles], 2 km W of Rio Naranjo, Utley & Utley 1899 (DUKE); Tilarán, Valerio 115 (F); SE lower slopes Volcán Mira- valles, Wilbur & Almeda 16623 (DUKE, F, MO, US). HEREDIA: Finca Hnos. Vargas, Puerto Viejo de Sarapiqui, Jiménez 3577 (MO); Rio La Paz & idu nco de ee rapiqui, Pittier 14159 (US). PANAMA. COCLÉ: La Mes N of El Valle de Antón, Allen 2369 (MO); continen divide past Llano Grande, Dressler 5627 (MO); Rio S Juan below junction with Rio Tife, Hammel xs (MO) 3 (MO). COMARCA DE SAN BLAS: El Llano-Carti Rd., " ae N of Panamerican Hwy., Folsom 2622 (MO); Nusagandi, along El Llano-Carti Rd., below (N of) Punta Mama, 350 m, de Nevers & Nuñez 3565 (MO); Rio Nergala, 100-300 m, de Nevers & Herrera 4548 (MO); Rio Irgandi and Río Cartí Senni, de Nevers & Herrera 6610 (MO, WIS); Rio Cangandí & Río Titamibe, 50- 150 m, de Nevers 4677 (MO); Rio Sidro, base of Cerro Habu, Sytsma et al. 2622 (MO); Cerro Habu, 400-800 ft., Sytsma et al. 2799 (MO). DARIÉN: upper Río Mem- brillo on construction road to San Blas, Duke 10891(MO). PANAMÁ: (M . 58 ; 3955 (MO); Dwyer 2649, 9445 (MO); Knapp 927 (MO); Lewis et al. 2314 (MO); Maas et al. 1540 (MO); pe & D'Arcy 3672, 3673 ( ; Sytsma & Antonio 3828 (MO); Sytsma & Knapp 4795 (MO); hein et st 5006 Volume 75, Number 4 1988 ma 1595 Lisianthius skinneri Lisianthius weaveri FIGURE 4. (MO); Tyson 5320 (DUKE, MO); Weaver & Foster 1482 (DUKE, MO, NY); Weaver & Wilbur 2242 (DUKE, NY), = (DUKE, MO, NY); Webster et al. 16475 (DUKE, MO); Wilbur et al. 11338, 15540 (DUKE); Wilbur & Teeri 13606 (DUKE, MO, NY); El Llano- Carti region, 6-8 mi. from Pan American Highway, An- Lisianthius weaveri (Antonio 3737 (MO) ).—a. Habit.—b. Flower.—c. Dehisced fruit. tonio 1690 (MO); Croat 49099 (MO); D'Arcy UR (DUKE, MO); Folsom 1439 (MO); Gentry 5788 (M ; Hamilton & Stockwell 1102 pe 0); Knapp 5565 (MO) Nee & Dressler 9327 (MO); Sytsma & 1596 Annals of the Missouri Botanical Garden Sytsma 3099 (MO); Sytsma 3989 (MO); Sytsma & Andersson 4432 (MO); Wilbur & Luteyn 19490 (DUKE, F, MO, NY); 11 km N of Gamboa, Croat 32937 (MO); Río Boquerón, trail to Rio Pequeni, Dressler 6001 (MO), 10 km N of Magarita on road to Madrono, Hammel & D'Arcy 5117 (MO); Hammel 6007 (MO); headwaters Rio Arenal, Johnston ees (MO); Gorgas Memorial Labs yellow fever research camp, i ] lunki 3364 (MO); Pipeline road, 9 km N of Gamboa, Nee 7681 (MO, WIS); se road, 8 a N of Gamboa, Nee 9580 (MO); Canal Zone, Rio Indio, Steyermark & Allen 17427 (MO). VERAGUAS: mouth of Rio Concepción, Lewis et al. 2799 (DUKE, MO). E 2. Lisianthius weaveri K. J. Sytsma, sp. nov. TYPE: Panama. Colón: Santa Rita ridge trail along Rio Piedras, 8,000 ft., 9 Mar. 1979, Hammel 6357 (holotype, MO; isotype, WIS). Figure 4 Frutex ad 2.5 m altus. Folia obovata-elliptica, ad 23 cm longa et 8.5 cm lata. Inflorescentiae axillares et ter- minales, dichasio biternato vel triternato . Calyx 5 In. . m longi et lati. Stamina filamentis + exsertis. Stilus antherae superans 5-10 mm; stigma peltatum, 1- 1.5 mm latum. Capsulae fusiformes ad fusiformes late, 1.5-2 cm longa. Shrub or subshrub, to 2.5 m tall; stem terete, erect, green. Leaves membranaceous, deep green above, paler below; blades to 23 cm long and to 8.5 cm wide; obovate to obovate-elliptic, + strong- ly acuminate tipped; the costa prominent, with 2- 3 sets of lateral veins, these ascending and becom- ing parallel with margin; petiole amplexicaul. In- florescences terminal or axillary in upper nodes; ternately compound dichasia (or only twice com- pound) foliaceous leaved below, bracted above. Flowers horizontal or nodding; calyx 5-6 mm long, tube to 2 mm long, the lobes 3-4.5 mm long, carinate dorsally, scariously margined, triangular- attenuate, + appressed; corolla tubular-funnel- form, constricted distally and + apically, the tube 2.2-3.3 cm long and to 8 mm wide, waxy and fleshy, bright yellow; lobes 2-3 mm long and wide, broadly triangular, green but with cream or yellow margins. Stamens inserted within corolla at apex of ovary; filaments to 2.5 cm long, equaling or barely exceeding corolla lobes; anthers 2-2.5 mm long and 0.75-1 mm wide, bilobate at base, pollen yellowish. Style to 3.5 cm long, far exserted (5- 10 mm) past anthers; stigma peltate, 1-1.5 mm broad. Capsules fusiform to broadly fusiform, 1.5- 2 cm long, beaked by persistent style to 10 mm long. Distribution. | Lisianthius weaveri is known from mid- to high-elevation cloud forests in Coclé and Colón provinces and at sea level on the Atlantic Coastal Plain in Colón Province, Panama Lisianthius weaveri is the only cloud forest species of Lisianthius found on both sides of the the Panama Canal. This species has been collected from a number of sites in the higher ridges leading to Cerro Bruja, so far an inaccessible peak domi- nating the ridge of mountains paralleling the At- lantic Ocean in eastern Colón Province. Popula- tions of L. skinneri collected in the lower reaches of this region are quite distinct from those of L. weaveri. Lisianthius weaveri is also found near the continental divide above El Cope, Coclé Prov- ince, the southeastern edge of the Cordillera de Talamanca extending northwestard towards Costa Rica. Three additional populations occupy high- rainfall, forest-edge sites near the beach on the humid Atlantic side of Panama. Lisianthius weav- eri resembles L. jefensis, also with short and waxy corollas, but lacks the exserted stamens and co- riaceous leaves of the latter. The short, fleshy, and waxy corolla and the exserted style readily distin- guish L. weaveri from lower-elevation L. skinneri populations. Weaver (1972) cited the Weaver & Wilbur 2249 collection as distinctive with its short, fleshy corolla and long exserted style. Although L. skinneri is certainly variable in floral features, the collections of L. weaveri surpass this level of vari- ability and can be readily distinguishable from all other collections of L. skinneri. A possible but as yet untested origin of L. weaveri might involve hybridization between L. skinneri and L. jefensis. In many of its floral characters, L. weaveri is strikingly intermediate between these geographi- cally adjacent species. Lisianthius weaveri is named in honor for Richard E. Weaver, Jr., monographer of the genus. Representative specimens examined. PANAMA COCLÉ: Alto de Calvario, Folsom & Jaslon 2680 (MO); E e, W of sawmill, Hammel 3545 (MO). COLON: Cerro Pilón, Loften s.n. (MO); Cerro Santa Rita, Allen & Allen 5104 (MO); Santa Rita trail to Rio Piedras, Antonio 3737, 3869 (MO, WIS); Rio Miguel de La Borda near Guasimo, Croat 9919 (MO); W of Portobello, D'Arcy 698 (F, MO); Maria Chiquita, Dwyer & Y - E e tobelo, Weaver & Wilbur 2249 (DUKE, F. MO, NY). a Rita ridge, Wilbur et al. 15020 (DUKE). 3. Lisianthius jefensis Robyns & Elias, Ann. Missouri Bot. Gard. 55: 60, fig. 1. 1968 (as “Lisianthus”). TYPE: Panama. Panamá: Cer- ro Jefe, Elias & Hayden 1798 (MO). Volume 75, Number 4 1988 Sytsma 1597 Lisianthius skinneri Slender shrub or subshrub, 1-2.5 m tall; stem terete, green. Leaves glossy dark green above, paler below; blades to 12 cm long and to 4.5 cm wide, subcoriaceous, oblanceolate to narrowly ob- ovate, apically acuminate, the lateral veins prom- inent; petioles amplexicaul, winged. Inflorescences terminal or axillary, composed of twice to ternately compound dichasia, rarely reduced, usually loosely arranged; bracts opposite, 2-5 mm long. Flowers horizontal, or more often nodding; calyx 7-13 mm long, lobes long acuminate, 5-9 mm long, scari- ously margined, weakly carinate; corolla tubular, thickened and waxy, constricted apically and dis- tally, tube 2.1-3.5 cm long, sometimes greenish tipped, lobes 4-6 mm long, triangular, to 3 mm wide. Stamens inserted on corolla at apex of ovary; filaments 1.5-3 cm long, the longest ones long exserted past corolla lobes; anthers 2.5-3.5 mm long, yellow, bilobate at base. Style to 3.5 cm long, conspicuously exserted past anthers; stigma pel- tate, to 1 mm broad. Capsule shortly fusiform, 1- 1.6 cm long, beaked; seeds asymmetrical, corru- gated. Chromosome number, n 3 Distribution. | Lisianthius jefensis is known only from the Cerro Jefe region of the Province of Panamá, elevation 800-900 m. This species is widespread near the rounded peak but can be found scattered at lower elevations. Cerro Jefe is sub- jected to alternating periods of wet cloud cover and intense sunlight due to an unusual combination of local topography and climate. Lisianthius je- fensis is a conspicuous and locally widespread member of the floristically diverse and predomi- nantly shrubby flora of Cerro Jefe that contains numerous other local endemics. This forest is dom- inated by species of Clusia and Calopothrinax cookii. Two populations of Lisianthius skinneri are found immediately below the cloud forest zone. The transition between the two life zones is abrupt, with the change readily perceived. In one locality in- dividuals of L. skinneri and L. jefensis are only meters apart, with no hybrids reported or seen (see also Weaver, 1972). DNA analysis clearly indi- cates that L. jefensis has been derived recently from eastern populations of L. skinneri. Lisian- thius jefensis is morphologically similar to L. weav- eri in floral characters and might have been in- volved in the origin of the latter species by hybridization with L. skinneri. Representative specimens examined. PANAMA. PANAMÁ: Cerro Jefe, 6-8 mi. past Goofy Lake, 3-4 mi. past Cerro Azul, 800-900 m elevation; Almeda & Nakai 3452 (F, MO); Antonio et al. 3402 (MO); Antonio 4699 (MO); Busey 798 (MO); Correa & Dressler 1154 (MO, NY); Correa et al. 1617 (DUKE, MO); Croat 14435, 17341 (MO); D’Arcy & Hamilton 14801 (MQ); D’Arcy et al. 15516 (MO); Duke 8010 (MO), 9413 (MO, US); Dwyer et al. 5035 (MO); Dwyer & Hayden 8087 (MO); Folsom et al. 5655 (MQ); Foster 1164 (DUKE); Gentry 2115 (DUKE, F, MO), 6771 (F, MO); Hammel 3716, 4817 (MO); Hamilton & D’Arcy 602 (MO); Hayden 1008 (DUKE, MO); Kirkbride & Crebbs 16 (F, MO); app 925, 2225, 3509, 3538, 5202 (MO); Luteyn 3200 (DUKE, F, MO); Luteyn & Kennedy 3959 (DUKE); McPherson 6878 (MO); dad ad ua WIS); O, US); 7129 (MO); Nee (DUKE, MO); Wilbur & Tari 13604 (DUKE, MO, NY. Wilbur et al. 15524 (DUKE, F, MO, NY, US); Wilbur 24126 (DUKE, F); Witherspoon & Witherspoon 8484 4. Lisianthius habuensis K. J. Sytsma, sp. nov. TYPE: Panama. Comarca de San Blas: Cerro Habu, vicinity of peak, cloud forest, 800 m, 78°49'W, 9°23'N, Sytsma, Antonio & Dress- ler 2685 (holotype, MO). Figure 5. Frutex 1-4 m altus. Folia obovata-elliptica, 13-19.5 m longa, 3.5-5.8 cm la E iade. dicha us; pedunculis elongatis. Calyx viridis, tubularis, 6-8 mm n lobi 4- 5 mm longi. Corolla aurea, cerea, 4.9- 2 m longa; lobi triangulares, Mens cerescens ad margineum, 3.5-5 mm longi et lati. Stamina filamentis + exsertis. . Stilus antherae superans 1-3 mm; pesi grande, conicum, corrugatum, ad 3-3.5 mm longa. Capsulae Polos, ad 2 cm longa. Shrub or subshrub, 1-4 m tall; stem terete, erect. Leaves membranaceous, usually pale green to green; petiole 1.3-2.5 cm long, amplexicaul; blade obovate to obovate-elliptic, basally attenuate, apically long acuminate to attenuate, 13- cm long and 3.5-5.8 cm wide, with 2-4 ascending lateral veins, the costa prominent beneath. Inflo- rescences of reduced dichasia, usually once com- pound, or reduced, on long axillary or terminal peduncles, 14-30 cm long; bracts opposite, over 2 mm long. Flowers nodding; calyx tubular, green, 6-8 mm long, the lobes lanceolate, acuminate, scariously margined, strongly carinate, 4-5 mm long; corolla bright yellow, waxy, the tube funnel- orm, inflated, 4.9-6.1 cm long, the lobes dark green with yellow border, broadly deltoid or tri- angular ovate, 3.5-5 mm long and broad, recurved slightly. Stamens inserted within corolla tube at apex of ovary; filaments exserted just past corolla lobes; anthers 2-3 mm long, bilobate at base; pollen 1598 Annals of the Missouri Botanical Garden FIGURE 5. d. Stigma. yellowish. Style surpassing anthers; stigma large, conical, to 3.5 mm long, contorted, almost cor- rugated. Capsule fusiform, to 2 cm long, with short beak; seeds asymmetrical, seed texture corrugated. Distribution. | Lisianthius habuensis occurs near the eastern range of the species complex in the province of Panamá and in the Comarca de Lisianthius habuensis (Sytsma et al. 2685 Lisianthius habuensis (MO) ) .—a. Habit. —b. Flower. —c. Dehisced fruit. — San Blas. A large population was discovered at the very top of Cerro Habu (800 m), Comarca de San Blas. This peak is located on the Cordillera de San Blas adjacent to the Atlantic coast and receives extremely abundant rainfall. Lisianthius habuen- sis dominates the shrub layer on the very tip of Cerro Habu but is not found more than 50 m below Volume 75, Number 4 1988 Sytsma 1599 Lisianthius skinneri the summit. A second population was found near the Continental Divide on the the road from El Llano to Carti, Province of Panamá, approximately 20 km from Cerro Habu. A large portion of this population grows on the roadside, with a few scat- tered individuals in the forest interior. Several pop- ulations of L. skinneri are located 3-5 km south of (below) L. habuensis along the El Llano—Carti road. A third population was discovered at the headwaters of three rivers in the Province of Pa- nama at elevations of 100-400 m. Lisianthius habuensis is distinct from all other species of Lisianthius by having an unusual stigma. The large ovoid stigma is obvious in the field, although not as noti ed herbarium spec- imens. The unique corolla and nie further distin- guish it from all other species. DNA evidence in- dicates that this species is most closely related to a lineage giving rise to L. peduncularis and aurantiacus. Additional specimens examined. PANAMA. PANAMA: headwaters of Rio Chagres, Río Esperanza and Río Pie dras, 79°20'W, 9°20'N, de Nevers 4086 (MO); along El Llano-Carti road from Pan American Highwa 300-400 m, Sytsma 4002 (MO, WIS); Sytsma et ai 5003 (MO, WIS). 9. Lisianthius peduncularis L. O. Williams, Fieldiana, Bot. 31: 408, fig. 1. 1968 (as “ Lisi- anthus"). TYPE: Panama. Coclé: El Valle de Anton, Allen 3410 (MO) Shrub or subshrub, occasionally large, to 3.5 m tall; stems terete, distinctly woody below, herba- Leaves petiolate, the petiole amplex- icaul; blades dark green above, paler below, 6- cm long, to 7.5 cm broad, broadly ovate, abruptly ceous above. acuminate to acute; the lateral veins prominent, strongly ascending. Inflorescences longly pedun- culate, loose and open, once compound dichasia, sometimes reduced; bracts opposite, lanceolate to sublinear. Flowers nodding, the pedicels 6-13 mm long. Calyx 7.5-13 mm long, the lobes lanceolate, carinate, scariously margined, long acuminate at the apex, 5.5-10 mm long. Corolla tube funnel- form, 4.5-6 cm long, bright yellow, inflated in the middle; the lobes dark green, 6-10 mm long, long acuminate, usually spreading. Stamens inserted in the corolla tube at the apex of the ovary; filaments 3.2-4 cm long, just surpassing the corolla tube but not the lobes; anthers 2-3 mm long, yellow, bilobed at the base. Style to 5 cm long, just exceeding the corolla lobes, always surpassing the anthers; stigma peltate. Capsule fusiform, to 1.5 cm long, sharply beaked; seeds irregular in shape, i aL in texture. Chromosome number, n Distribution. | Lisianthius peduncularis, en- demic to the north rim and adjacent ridges of El Valle de Antón, Coclé Province, is now known to be composed of three small populations. Two pop- ulations are restricted to exposed elfin forest ridges (900-1,000 m) and usually are found associated with Symbolanthus pulcherrimus Gilg, a lisian- thioid shrub characteristic of such habitats. A third population is found on the northern lower flanks of the El Valle crater (800 m). This large population of approximately 80 individuals grows on a soft porous rhyolite bedrock in association with a low Clusia-dominated scrubby open forest similar to the vegetation type on Cerro Jefe. Lisianthius skinneri has been collected on the road from El Valle leading up to these L. peduncularis sites. P. Allen (2369) collected it in 1941, but the species has not been collected since from the region despite extensive searches and collecting through the Flora of Panama project. Morphologically, Lisianthius peduncularis most closely resembles the new L. aurantiacus with which it shares long corolla tubes and lobes, and reduced inflorescences. They differ strikingly in habit and corolla color. Both occur at the western edge of the species complex in Panama. DNA anal- ysis clearly indicates that these two species form a close pair of “‘sister species.” Representative specimens examined. PANAMA. COCLÉ: N rim, El Valle de Antón, Allen 1793 (MO, US); La Mesa, N of El Valle, Allen 2369 (US); Cerro Pajita, Allen & Allen 4187 (MO); El Valle, Club Campestre, ae 14288 (F, MO); Cerro Pilon, Croat 22945 (DUKE, O, WIS); Cerro Pilon, Duke 12192 (MO); Cerro Darocora), Duke & Dwyer 15094 (MO); Cerro Carocoral, Kirkbride 1094 (MO); trail past La Mesa, Clusia forest, Luteyn 4082 (DUKE); Cerro Pilón, Mori 6631 (MO); Cerro Gaital, Reveal & Balogh 4945 (MO); Divide SW dde et al. 2247 ( S Valle, Wilbur & Luteyn 11696 (DUKE, MO, NY); trail past La Clusia thicket, Wilbur et al. 15622 (DUKE). 6. Lisianthius aurantiacus K. J. Sytsma, sp. nov. TYPE: Panama. Coclé: Mountains between La Pintada and Cascajal, Dressler 5625 (ho- lotype, MO; isotype, WIS). Figure 6. Frutex vel arbor, ad 6.5 m alta; truncus ad 7 cm latis, ramosis aequaliter pee versus. F ad 25 cm longa et 6.5 c dichasio ge oet] 1-2-foribus; pedunculis elongatis, ad 12 cm longis. Calyx viridis, tubularis, 10-16 mm longis; lobi 8-13 mm longi, acuminati longe. Corolla aurea, au- 1600 Annals of the Missouri Botanical Garden Lisianthius aurantiacus FiGURE 6. Lisianthius aurantiacus (Hammel 2508 (MO)).—a. Habit.—b. Flower.—c. Dehisced fruit. Volume 75, Number 4 1988 Sytsma 1601 Lisianthius skinneri rantiacus, 5.5-7.8 cm longa; lobi triangulares, virides, 10-14(-17) mm longi, 3-5 mm lati. Stamina filamentis + o” antherae -5 mm longae. Stilus antherae superans 1-3 mm; stigma po aquamarinum. sulae usiformes, 1.8-2.5 cm longa. Shrub or slender-trunked tree, to 6.5 m tall. Stem to 7 cm wide, terete, evenly branched to the top. Leaves petiolate, the petiole 5-15 mm long, amplexicaul; blade glossy dark green above, slightly paler below, 2—3 lateral veins conspicuous, strongly ascending, the costa prominent below, membra- naceous; to 25 cm long and 6.5 cm broad, obovate to obovate-elliptic, basally cuneate to slightly at- tenuate, apically acuminate. Inflorescence axillary, opposite, consisting of 1 or 2 flowers; the peduncles to 12 cm long, containing 1-3 sets of foliaceous to linear bracts, the larger bracts to 15 mm long; the pedicels to 2 cm long. Flowers strongly nodding. Calyx tubular, dark green, 10-16 mm long, the lobes lanceolate, long acuminate, scariously mar- gined, carinate at the base only, 8-13 mm long. Corolla 5.5 -7.8 cm long, tubular-funnelform, in- flated, the tube bright pumpkin orange, the lower ¥ narrowly constricted, the lobes dark green, tri- angular, acuminate and spreading, 10-14(-17) long and 3-5 mm wide at the base. Stamens 4.5- 6.3 cm long, exserted to the midpoint of the lobes; filaments filiform, inserted on the corolla tube at the apex of the ovary; anthers 2.5-5 mm long, slightly sagittate at base, yellow. Style 4.7-6.7 mm long, slightly exserted past the anthers; stigma blue- green, capitate, slightly bilobed at apex. Capsule fusiform, m long, 5-7 mm diam., with a beak 4 mm long. Distribution. | Lisianthius aurantiacus has the most widespread distribution of the cloud-forest species. It has been collected in three localities: on the continental divide near Cascajal, Coclé Prov- ince (650 m); below the continental divide on the Atlantic watershed north of El Cope, Coclé Prov- ince (800-900 m); and the Cerro Tife region 15 km west of El Cope (400-450 m). Lisianthius aurantiacus usually occurs sporadically in closed forests, with only a few individuals seen together. ations in the more disturbed Cascajal area are large and more treelike (to 6 m), effectively forming a canopy. Lisianthius aurantiacus is found at lower elevations than the other cloud-forest species. The cloud forests in this region of Coclé Province are lower in elevation than in other areas of central Panama because of local climatic con- ditions. The forests are floristically more similar to mid-elevation moist forests where L. skinneri thrives. Indeed, a population of L. skinneri was discovered growing sympatrically with L. auran- tiacus in the Cascajal area. Lisianthius aurantiacus is undoubtedly the most spectacular member of the genus. Its arboreal hab- it, very large pumpkin orange corolla (thus the specific epithet), and highly reduced inflorescence distinguish it from all other Lisianthius species. A more northern species, L. axillaris, is strikingly similar to L. aurantiacus. The only species with red or orange flowers known prior to L. aurantia- cus was Lisianthius axillaris, a common roadside plant in Belize and surrounding regions. It exhibits not only a reddish corolla, but also an axillary inflorescence of a single flower as well. This is a clear case of floral convergence. Lisianthius au- rantiacus is most closely related to L. peduncu- laris, with which it shares several other floral char- acters. Representative specimens examined. PANAMA. COCLÉ: trail from Cano Sucio to Cerro Tife, base of wa- diam Antonio 3687 (MO, WIS); area between Cano Blanco del Norte, Cario Sucio and Chorro del Rio Tife, Dav ids se & Hamilton 23581 (MO, WIS); Caribbean side of divide at El Cope, Hamilton & Davidse 2680, 2693 (M S); 1 ; 7 km lano Grande, road to Coclesito, ivide e, roa a; 8509 a Knapp 1954 (M waterfall of Rio bud Knapp 3704 (MO, M Los Pe edregales, ridge between Rio Blanco > Norte and Río Caño Suc cio, Knapp & Dr essler 3788 (M 0, WIS): Caio Rd., Continental Divide, 500 rs et al. 6 4 mi. past Llano Grande to Cascajal, Sytsma 3981 (MO, WIS); Sytsma et al. 4379, 5005 (MO, WIS). LITERATURE CITED AUBLET, J. C. B. F. 1775. Histoire des Plantes de la Guiane pc d ee London & Paris. [Lisy- anthus 1: 201-203.] Eras, T. S. & A. s 1975. Family 160. Gen tianaceae. /n: Flora of Panama. Ann. Missouri Bot. Gard. 62: 61-101. EwaN, J. 1948. A revision of Macrocarpaea, a neo- tropical genus of shrubby Ec Contr. U.S. Natl. 09-24 Herb. 29: Guc, E 89 Gentianaceae. /n: Die Natürlichen Pflanzenfamilien 4: peru. ie 1966. Phylogenetic Systematics. Univ. Il- linois Press, Chicag E C. 1767. Mantissa Plantarum [prima] Stock- olm. [Lisianthus, 43.] 1 Maas, P. J. . Nomenclatural notes on neo- tropical Lisyantheae Pipe d Proc. Kon. Ned. ensch. C 88: 405-412 — — —, S. NILSSON, . C. HoLLAN7TS, B. J. H. TER wet. J. G. M. PERSOON & E. C. H. vaN HEUSDEN. 1602 Annals of the Missouri Botanical Garden 1984. WORST snis in uoo edd ceae —the Lisianthiu a Bot. Neerl. 3 71l- NILSSON, S. 68. Pollen morphology in the genus Macrocarpaea (Gentianaceae) and its taxonomic sig- nificance. Svensk Bot. Tidskr. 62: 338-364. —— 1970. Pollen morphological contributions to the taxonomy of Lisianthus L. s. lat. (Gentianaceae). Svensk Bot. Tidskr. 64: 1-43. Raven, P. H. 1979. Future nro in plant popu- lation biology. Pp. 461-481 i brig et al. (editors), Topics in Plant bte Biology. Colum- bia Univ. Press, New York. SytsMa, K. J. 1987. The shrubby gentian genus Macro- carpaea in Panama. Ann. Missouri Bot. Gard. 74: -313. SCHAAL. 1985a. Genetic variation, differentiation, and evolution in a species complex o tropical shrubs based on isozymic data. Evolution 39: 93. 1985b. Phylogenetics of the Lisi- anthius skinneri (Gentianaceae) species complex in Panama utilizing DNA restriction fragment analysis. Evolution 39: 594- 608. Weaver, R. E., JR. 1972. A revision of the neotropical genus Lisianthius (Gentianaceae). J. Arnold Arbor. 53: 76-100, 234-311. CHROMOSOME NUMBERS IN BACE CROTALARIEAE)! Ben-Erik Van Wyk and Anne Lise Schutte? ABSTRACT Original chromosome counts for Buchenroedera (new s ed an Lotononis (44 new specific reports) are ee The most common somatic number Buchenroedera and nine species of Loto in Loton nonis have 2n = 28. In Lotononis section Krebsia 2n = 28, 4 nis, 16, 84 were found in a closely related species group. This is the first report of a polyploid series in the pedi ios. and includes the hig hest numbers recorded in the tribe. ome numbers indicate anomalies in the The chrom existing qu classification of Lotononis and may provide evidence for a more natural generic and infrageneric classificati The genera Lotononis (DC.) Eckl. & Zeyh. and Buchenroedera Eckl. & Zeyh. are poorly known cytologically, with only six species of the former and none of the latter having been investigated previously. As part of an ongoing taxonomic study of these genera, chromosome counts were made for 47 species, representing almost the full range of variation in Lotononis (ca. 120 species centered in southern Africa, with a few extending into Asia) and Buchenroedera (ca. 16 species restricted to the eastern parts of southern Africa). The results are presented here, and their systematic signifi- cance in terms of an improved generic and infra- generic classification is discussed. MATERIALS AND METHODS Mitotic counts were made from root tips of ger- minated seeds. Standard methods of pretreatment in hydroxyquinoline (0.02% mass/volume) and staining in lacto-propionic orcein were used. The duration of hydrolysis (1-8 minutes) and the con- centration of HCl (0.2-0.5 N) proved to be im- portant. The chromosomes are small (ca. 1-3 um long). Voucher specimens (listed in Table 1) are housed at the Rand Afrikaans University Herbar- ium (JRAU). A list of the species studied and vouch- er specimen details are given in the Appendix. Our efforts to collect seeds have been rewarded by numerous rediscove provided a fairly representative sample of the two genera. ries of rare species and have RESULTS AND DISCUSSION The results listed in Table 1 are arranged ac- cording to Duemmer's (1913) sectional classifi- cation. Where morphologically heterogenous sec- tions of Lotononis have been subdivided into two or more groups, or where species have been moved to more appropriate positions, the reasons for doing so are given in the footnotes. The arrangement of species in Table 1 is aimed at facilitating the dis- cussion that follows and is not intended as a formal infrageneric classification, but it nevertheless re- flects major discontinuities and shows basic affini- ties. Several morphological characters provide links among the species of Lotononis with 2n — 28 and among those with 2n — 18. The latter are presently placed in various sections, indicating that Duem- mer's infrageneric treatment is artificial; that the same chromosome number has evolved indepen- dently in several different groups seems unlikely. Section Krebsia, for example, presently comprises three distinct groups, two of which have obvious ! We thank Dr. Johan Spies (Botanical Research Institute, Pretoria) and Dr. Gerrit Davidse (Missouri D. project at the rch on Lotononis and Buchenroedera by the senior University of Cape To ? Department of Botany, Rand Afrikaans University, P.O. Box 524, Johannesburg 2000, South Africa. ANN. Missouni Bor. GARD. 75: 1603-1607. 1988. 1604 Annals of the Missouri Botanical Garden TABLE l. Chromosome numbers in Lotononis and Buchenroedera. Species are arranged in sections following the treatment of Duemmer (1913) , with some minor modifications that are explained in the footnotes. All known counts are included—those taken from the literature are preceded by an asterisk (*). Chromosome Voucher or Genera, Groups, and Species Number (2n) Reference Buchenroedera Eckl. & Zeyh. B. lotononoides Scott Elliot 28 BVW 1966 B. meyeri Presl 28 BVW 1765 B. tenuifolia Eckl. & Zeyh. var. tenuifolia 28 BVW 1675 Lotononis (DC) Eckl. & Zeyh. Lotononis section Aulacinthus (E. Mey.) Benth. L. leucoclada (Schltr.) Duemmer 28 BVW 2430 L. gracilis (E. Mey.) Benth. 28 BVW 2250 Lotononis section Krebsia (Eckl. & Zeyh.) Benth. Part 1: Krebsia sensu stricto L. biflora (H. Bol.) Duemmer + 84 BVW 1952 L. carnosa (Eckl. & Zeyh.) Benth. 84 BVW 1663 L. caerulescens (E. Mey.) B-E. van Wyk' 56 BVW 2483 L. cytisoides (E. Mey.) Benth. 28 BVW 1721 L. cytisoides (E. Mey.) Benth. aff. 56 BVW 1761 L. divaricata (Eckl. & Zeyh.) Benth. 56 BVW 2484 L. divaricata (Eckl. & Zeyh.) Benth. aff. 42 BVW 1666 L. trisegmentata Phill. var. robusta Phill. forma robusta 28 BVW 1917 L. trisegmentata Phill. var. robusta Phill. forma sericea Phill. 28 BVW 1956, 1958 Part 2: L. digitata group? L. digitata Harv. 18 BVW 2341 L. benthamiana Duemmer 18 BVW 2538 "L. magnifica" B-E. van Wyk ined. 18 BVW 2549 Part 3: L. transvaalensis group? L. transvaalensis Duemmer 18 BVW 1860 L. procumbens H. Bol.* 18 BVW 2504 Lotononis section Polylobium (Eckl. & Zeyh.) Benth. Part 1: Polylobium sensu stricto L. exstipulata L. Bol. 28 BVW 2280 *[. involucrata (Berg.) Benth. 28 (Dahlgren, 1967) *L. serpens (E. Mey.) Dahlgr.* 18 (Goldblatt, 1981b) Part 2: L. angolensis group^ L. angolensis Bak. 18 (Byth, 1964) *[. listii Polhill 18 (Byth, 1964) *[. bainesii Bak. 36 (Byth, 1964) Lotononis section Telina (E. Mey.) Benth. L. acuminata Eckl. & Zeyh. 28 BVW 2581 “L. repens" B-E. van Wyk ined. 28 BVW 2573 L. pungens Eckl. & Zeyh.’ 28 BVW 1725 L. versicolor (E. Mey.) Benth.’ 28 BVW 1386 Lotononis section Oxydium Benth." L. rostrata Benth.? 18 BVW 2324 L. rostrata aff. 18 BVW 2429 L. acutiflora Benth. 18 BVW 2544 L. oxyptera (E. Mey.) Benth. 18 BVW 2318 L. lenticula (E. Mey.) Benth. 18 BVW 2018 L. rabenaviana Dinter & Harms 18 BVW 2057 Volume 75, Number 4 1988 Van Wyk & Schutte Lotononis, Buchenroedera Chromosome Counts 1605 TABLE l. Continued. Chromosome Voucher or Genera, Groups, and Species Number (27) Reference Lotononis section Lipozygis (E. Mey.) Benth. Part 1: L. polycephala group" L. polycephala "a: e ) Benth. 18 BVW 2408 L. bolusii Duem 18 BVW 2443 “L. E B. E. van Wyk ined. 18 BVW 2241 Part 2: L. eriantha group" L. eriantha Bent 18 ALS 383 L. foliosa H. Bol. 18 BVW 2607 L. lanceolata (E. Mey.) Benth. 18 BVW 1884 Lotononis section Leobordea (Del.) Benth. *L. platycarpa (Viv.) Pic.-Serm. 18 (Goldblatt, 1981b) Lotononis section Leptis (Eckl. & Zeyh.) Benth. Part 1: L. laxa group"? L. laxa Eckl. & Zeyh. 18 BVW 2015 L. woodii H. Bol. 18 BVW 2608 L. macrosepala Conrath 18 BVW 2622 Part 2: L. brachyloba group'* L. brachyloba (E. Mey.) Benth. 18 BVW 2244 “L. fruticoides” B-E. van Wyk ined. 18 BVW 2020 L. leptoloba H. Bol. 18 ALS 276 L. maximilianii Schltr. EL 18 ALS 271 L. maximilianii (chasmogamous) 18 ALS 282 Part 3: L. calycina group'* L. calycina (E. Mey.) Benth. 18 BVW 2621 L. "i unida ser 18 BVW 1899 L. humifusa 18 BVW 1700 L. mucronata x aff. 18 BVW 2619 "L. curvicarpa" B-E. van Wyk ined. 18 BVW 2725 ' Be tter known as s Lebeckia ai A de die ' Species added to Kre by Dun - & * Position in section Poly w n anoma ° Species section Polylobium by Bake ded 1 ? Superficially similar to L. la a by Harvey (1 (1862) and Duemmer (1913). 3). on a superficial characterization. 8 This section was referred to the dua Crotalaria by Duemmer (1913) ? Better known as L. micrantha (E. Mey.) Bent 10 A distinct etia of Lipozygis with indehiscent, ‘wind-dispersed fruit. u" A distinct group of pyrophytes from grassland area s of the eastern parts of southern Africa. 12 Perennial herbs with acute keel petals as in section Oxydium. " Annuals with acute keel petals as in section Oxydium. Annuals and perennials with obtuse keel petals as in the L. eriantha group of section Lipozygis. affinities elsewhere in the genus. The woody habit of L. digitata and L. transvaalensis was used to place them in Krebsia, but both are morphologi- cally very similar to various species of section Lep- tis. Another example is section Polylobium; Lo- tononis umbellata and its allies are closely related to section Aulacinthus and perhaps not distinct from it at the sectional level. The L. angolensis group is quit rom other species of section Polylobium and its position in this section is un- satisfactorily artificial. Two separate phylogenetic lines with base num- bers of x = 9 and x = 7 are suggested, and further research will show if other evidence supports such 1606 Annals of the Missouri Botanical Garden a dichotomy in the genus. Not a single count of 2n = 16 or 32 has been made, so that a base number of 8, which is common in some of the other genera, so far appears to be totally absent in Lo- tononis. At the generic level, the data also give some indications of affinity. Buchenroedera is so closely related to Lotononis (especially to section Krebsia that its generic status has been questioned (Polhill, 1976, 1981). The shared chromosome number of 2n — 28 (and presumably a base number of 7) agrees with chemical evidence (Van Wyk & Ver- doorn, 1988) that Buchenroedera is perhaps best considered a section of Lotononis. The remarkable similarities between species of Crotalaria and Lotononis have caused confusion in past taxonomic treatments. For example, most species of Lotononis section Oxydium were trans- ferred to Crotalaria by Duemmer (1913). The presence of macrocyclic pyrrolizidine alkaloids in both genera (Van Wyk & Verdoorn, in prep.) indeed indicates that Lotononis is more closely related to Crotalaria than to other genera of the tribe, all of which seem to contain only quinolizidine alkaloids. Crotalaria, however, have 2n — 16, or rarely 14 (Goldblatt, 198 la), while thos species of Lotononis that closely resemble C tion Oxydium and some groups of hina, all have 2n = 18. The morphological distinction between Lotononis and Crotalaria (Polhill, 1968) is there- fore strongly supported by the data at hand. Some of the woody species of Lotononis (sec- tions Aulacinthus and Krebsia) are very similar to species of Lebeckia. Lotononis caerulescens (E. Mey.) B-E. van Wyk, for example, has until re- cently been known as Lebeckia microphylla E. Mey., but morphological and chemical evidence (Van Wyk, 1988; Van Wyk & Verdoorn, 1988) clearly showed it to be misplaced in Lebeckia. The sections Aulacinthus and Krebsia sensu stricto have 2n — 28, 42, 56, and 84, while four counts of 2n = 18 are known for Lebeckia (Dahlgren, 1967). Here again, the cytological data agree with the morphological distinction between Lotononis and Lebeckia. Lotononis angolensis and related species (section Polylobium) are chemically similar to Lebeckia and also have the same chromosome number. Morphological characters such as the zy- gomorphic calyx and dimorphic stipules, however, — ( OULGUGUT CC ( are typical of Lotononis. Not a single count of 2n = 14 is known for Lotononis; so it seems to be cytologically different from the genus Pearsonia. The only available count for the latter genus was by Frahm-Leliveld (1969), who reported 2n — 14 for P. flava (Bak. f.) Polhill. The species of Pearsonia are similar to Lotononis except for their highly modified flowers (Polhill, 1973), and the shared chromosome base number of x — 7 may indeed indicate a common ancestry. From a phylogenetic point of view, the different base numbers in Lotononis suggest interesting questions about generic relationships in the Cro- talarieae. The base number of the tribe is almost certainly x = 9 (Goldblatt, 1981a), and 2n = 18 in some species of Lotononis is presumably the ancestral condition. The only way to achieve 2n — 28 (if Lotononis is monophyletic) is to postulate descending aneuploidy from n = 9 to 8 and 7 and subsequent polyploidy. Since 2n = 16 and 14 appear to be totally absent in Lotononis, it may be argued that Crotalaria and Pearsonia form part of the lineage that gave rise to the group of species with 2n — 28, 42, 56, and 84. If Lotononis proves to be polyphyletic, this possibility can be seriously considered, but the generic characters of the current concept of Lotononis are present in at least some species of each major group. Although there are marked phenetic similarities linking all the major groups, Lotononis as a whole is not monothetic. It is defined by combinations of apo- morphic tendencies, such as single stipules, suffru- tescent or herbaceous habit, absence of bracteoles, fusion of the lateral calyx lobes, verrucose upper suture of the fruit, tuberculate testa, elongated funicles, flower dimorphism associated with cleis- togamy, ability to produce HCN, and presence of macrocyclic pyrrolizidine alkaloids. There is not a single apomorphy known to us that would unam- biguously support monophyly. A possible solution would be to separate the lineage with 2n — 28 from the one with 2n — 18 and to split the latter into several smaller groups. Despite conflicting character information, there are some indications from the morphology that the geographically wide- spread and generally herbaceous 2n — 18 lineage is more primitive than the predominantly woody and essentially southern African 2n — 28 lineage. In a tribal context, the occurrence of polyploidy in Lotononis (section Krebsia) is of some interest. Polyploidy and high chromosome numbers are typ- ical of the Genisteae but have never been reported from any genus of the Crotalarieae (Goldblatt, 198 1a). It is also noteworthy that polyploidy should occur in an essentially woody group (previously considered to be one of the basal groups of Loto- nonis) and not in the supposedly more derived herbaceous groups. Unlike the situation in the other large genera of the Crotalarieae (Aspalathus and Volume 75, Number 4 1988 Van Wyk & Schutte 1607 Lotononis, Buchenroedera Chromosome Counts to some extent Crotalaria), there is no direct evi- dence of aneuploidy, although it must have played a significant role in the phylogeny of Lotononis. LITERATURE CITED BAKER, J. G 71. Suborder I. Papilionaceae. In: D. Oliver (editor), Flora of Tropical Africa. Volume 2. London. Byth, D. E. 1964. Breeding system and chromosome number in Lotononis bainesii Baker. Nature 202: 30. DAHLGREN, R. 1964. The genus Euchlora Eckl. & Zeyh. as a from Lotononis Eckl. & Zeyh. Bot. Not -388 p omosome numbers in some South African pe d the € Genisteae s. lat. (Legu- . 120: 149-160. 1 A sasa of the species of yin om 8 Zeyh. and Pleiospora Harv. Trans. Roy. Soc. South Africa 3: 275-335. FRAHM- DUE J. A. A s; Sp aa qr notes on African a Acta Bot. Neerl. 18: 67-73. GOLDBLATT, 1981a. Cytology and the »hylogeny of Duns In: R. olhill & P. H. Raven (ed- itors), Advances in Legume Systematics. Royal Bo- tanic Gardens, Kew. 1981b. Chromosome numbers in pue II. Ann. Missouri Bot. Gard. 68: 551-557 Harvey, W. H. 62. Leguminosae. Pp. 47-66 in H. Harvey & O. W. ps —€— Flora Capensis. Volume 2. Hodges & Smi PoLHiLL, R. M. 1968. Misal " notes on African species of Crotalaria L. Kew . 22: 169-348. 1973. A revision of aa i as sae — Papilionoideae). Kew Bull. 29: 383 . 1976. Genisteae (Adans.) Benth. B) NE tribes — Ua Bot. Syst. 1: 143-348 19 Tribe 29. Crotalarieae (Benth.) Hutch. E R. M. Polhill < & P. H. Raven (editors), Advances S. African E E ——— & G ERDOORN. 1988. The ws end onomic a ance of integerrimine in Buc dera and Nu ud section Krebsia. Biochem. cim Ecol. 16: 287-289 APPENDIX List of species, collection data, and voucher specimen details of the material used for chromosome counts. Voucher specimen numbers refer to our own collections (abbreviated as BVW and ALS) and are all housed in the Rand Afrikaans University a (JRAU). Authori- ties for names are given in Buchenroedera ide Loteni, Natal, BVW 1966. B. meyeri: Mhlahlane, Transkei, BVW 1765. B. dense ga var. tenuifolia: Queenstown, E Cape, BVW 16 r acuminata: Humansdorp district, S Cape, BVW 2581. L. dpi. id Khamiesberg, Cape, BVW 2 L. benthamiana: Springbok D Cape, BVW 8. L. biflora: per Natal, BYW 1952. L. bolusii: ee ys BVW 2443. L. ECT ree Ceres, ape, 4. L. caerulescens: Cradoc A 5 fÑ BIIP 2483. E. kin n Bethal, Transvaal, BVW 26 sa: Queenstown, E Cape, BVW 1663. “L. ined.): Devon, Transvaal, BVW 2725. L. 2607. i früitcoides" (ne) Graa Reinet district, Cape, BVW 2 š r Crahamstown district, L. lenticula: Colesberg, Cape, BVW 2 toloba: Nieuwoudtville, Cape, ALS 276. L A a da: Clanwilliam, Cape, BVW 2430. “L. longicephala” (ined.): Touw's River, Cape, BVW 2241 sepala: Bethal district, Transvaal, magnifica” (ined.): Khamiesberg, Cape, BYW 2549 L. maximiliani: Nieuwoudtville, Cape, ALS 27 1 (cleis- togamous form), ALS 282 (chasmogamous form). L. mu- cronata aff.: Ermelo district, Transvaal, BVW 2619. L. oxyptera: Citrusdal, Cape, B rostrata Aia Cape, sui 2429. L. ora: transvaalens L. t jae vs val aes rens, Orange Free State, BVW 1917. L. trisegmentata var. robusta forma sericea: Loteni, Natal, BVW 1956; Sani Pass, Natal BVW 1958. L. versicolor: Beaufort West district, Cape, BVW 1386. L. woodii: Wakker- stroom district, Natal, BVW/ 2608. GEORGE ENGELMANN TYPE SPECIMENS IN THE HERBARIUM OF THE MISSOURI BOTANICAL GARDEN' Steven J. Wolf? ABSTRACT Eight hundred ninety-two type specimens (in 62 families) representing 589 taxa described by George Engelmann have been located in the herbarium of the Missouri Botanical Garden. For each specimen the following are given: literature oni kind of type, locality and date of collection, collector, and name it is currently filed under in the herbari Asa Gray once remarked that George Engel- mann had the potential to become the “‘gatekeeper for all scientists going into the wilderness" (Dupree, 1959). Engelmann, a well-trained botanist and phy- sician, on the edge of the American frontier in mid- 19th century St. Louis, the gateway to the West, was indeed the right man in the right place at the right time. Some of the first botanical expeditions to the American West were organized, outfitted, and coordinated by Engelmann, who then received and processed the incoming plant specimens and sent them on to Gray (Timberlake, 1984). Among these early expeditions were Geyer’s to Illinois, Missouri, and Iowa; Lindheimer’s to Texas; Wis- lizenus’s to the New Mexico Territory and northern Mexico, and Fendler’s to the southern Rocky Mountains. By the 1850s most expeditions to the West were government supported; however, the plant specimens from them continued to be fun- neled through Engelmann in St. Louis (Timberlake, 1984). As a result, his herbarium, which consisted of about 97,000 specimens at his death, contained collections from more than 30 collecting expedi- tions. Although largely from the American West and Mexico, collections also came from South America and "He ade a "dita and industrious scientist, was much more than a gatekeeper. In addition to maintaining a full-time medical practice, he pub- lished more than 100 botanical papers (Sargent, 1884; Timberlake, 1984), described more than 600 new taxa of plants (Trelease & Gray, 1887), founded the Academy of Science of St. Louis, and was largely responsible for the foundation of what later became the herbarium of the Missouri Bo- tanical Garden. It was Engelmann who convinced Henry Shaw, a wealthy St. Louis businessman plan- ning a botanical garden, that to be scientifically credible his garden must have a library and her- barium (Timberlake, 1984). In fact it was Engel- mann who purchased the Bernhardi Herbarium for Shaw, which comprised the first specimens of this herbarium. After Engelmann's death in 1884, his entire herbarium was added to Shaw's herbarium which, upon Shaw's death, became known as the herbarium of the Missouri Botanical Garden. On the 100th anniversary of the death of George Engelmann, it seemed appropriate to re-examine his herbarium, particularly with respect to possible type specimens it might contain. Complete lists of Engelmann's publications, including his contribu- tions to the works of others, are provided in Sargent (1884) and Timberlake (1984), and references to and reprints of most of his newly described taxa are included in Trelease & Gray (1887). The only major omissions in the latter are Engelmann's con- tributions to Coulter's (1894, 1896) revision of North American cacti. In the present study, a search for type specimens of Engelmann's more than 600 described taxa was conducted. An un- edged. I thank Nan Michaels for her tone ble word processing ass ! Financial anon from the National Museum Act through a postdoctoral fellowship is gratefully acknowl- orin for suggesting ded pies and locating additonal types a M es, and Deann n iology Department, University of Missouri, ae on Missouri 64110, U.S.A. ANN. Missouni Bor. GARD. 75: 1608-1636. 1988. Volume 75, Number 4 1988 Wolf 1609 Engelmann Type Specimens derstanding of some of Engelmann's habits was most helpful in locating types. He sometimes wrote “n. sp." on newly named specimens or wrote the descriptions on the sheet itself. Although he some- times did not mention a specific collection or col- lector, it is often apparent which ones he had seen and used for his descriptions. For example, if his protologue stated “‘Battlefields of Buena Vista, near Saltillo, Flowers May," there is often a specimen with this exact wording on the label, collected in May by one of the collectors of that particular expedition. Additionally, if in describing a new tax- on he had stated that it resembled or was related to taxon “X,” the type would often have been initially labeled taxon “X,” but this epithet would later have been crossed out and the new epithet penciled in by Engelmann. The purpose of the present paper is to provide a list of types or possible type material contained in Engelmann's herbarium. Since many of the col- lections examined by Engelmann were widely dis- tributed, and since only specialists should designate lectotypes, I have adopted a conservative type terminology. The only exception is that I have adopted Rollins's (1972) view that “the existence of a holotype in the institution where the author worked is assumed until proven otherwise." If En- gelmann cited only one specimen and that specimen has been located in his herbarium, it is considered a holotype and supersedes previously designated lectotypes. Annotations by specialists on the sheets themselves, as well as known lectotypifications in the monographs listed below, have been accepted unless a holotype has been located. I have adopted the term “type material” in cases in which it is not clear what kind of type the specimen is but in which it is apparent that Engelmann did see it and may have used it in making his descriptions. This term is especially useful in dealing with Lindhei- mer's collections from Texas because they were apparently widely distributed and the ue numbers often consisted of mixed in different locales, in different years, and some- times even after the date of publication of Engel. mann & Gray (1845). It is therefore suggested that specialists examine these specimens, some of which bear Lindheimer's original collection label and number, in order to determine if they are heterogeneous material and/or what kind of types they are. In the present paper a total of 892 types, rep- resenting 589 taxa in 62 families, described by or attributed to George Engelmann are presented. In the following list the name of the taxon, its au- thorship, and place of publication are given, fol- lowed by the kind of type, location and date of collection, and the collector(s) and collection num- ber if known. Also given is the current name under which it is filed in the herbarium. Taxa for which works containing numerous lectotypifications have been consulted, and for which synonymies have been followed, include: 4gave (Gentry, 1982), Ar- ceuthobium (Hawksworth & Weins, 1972), As- clepias (Woodson, 1954), Cactaceae (Benson, 1982), Cuscuta (Yuncker, 1932), Euphorbia (Wheeler, 1941), Isoetes (Pfeiffer, 1922), Phora- dendron (Trelease, 1916), Quercus (Trelease, 1924), Sagittaria (Bogin, 1955), Vitis (Bailey, 1934), and Yucca (McKelvey, 1938, 1947). The taxa are arranged by family according to Kartesz & Kartesz (1980). ACANTHACEAE Dianthera humilis Engelm. & Gray, Boston J. Nat. Hist. 5: 55. 1845. SYNTYPE: west of Houston, Texas, 1845, F. Lindheimer 159. — Justicia ovata (Walt.) Lindau. Dipteracanthus nudiflorus Engelm. & Gray, Bos- J. Nat. Hist. 5: 229. 1845. HOLOTYPE: Sim's Bayou, near Houston, Texas, 1842, F. Lindheimer s.n. AGAVACEAE Agave angustissima Engelm., Trans. Acad. Sci. St. Louis 3: 306. 1876. HOLOTYPE: Octillo, Mexico, 1849, J. Gregg 959. Agave couessii Engelm. in Trelease, Annual Rep. Missouri Bot. Gard. 22: 94. 1911. HOLOTYPE: Ft. Whipple, Arizona, 1865, E. Coues & F. Palmer 253, “couessi” added by Engelm. = A. Parryi Engelm. var. couessii (Engelm. ex Trel.) Kearny & Peables. Agave desertii Engelm., Trans. Acad. Sci. St. Louis 3: 310. 1875. HOLOTYPE?: San Felipe, Cali- fornia, 1875, E. Palmer & G. Hitchcock s.n., annotated “‘n. sp." by Engelm. Agave falcata Engelm., Trans. Acad. Sci. St. Louis 3: 304. 1875. LECTOTYPE: Saltillo, xA 1847, A. Wislizenus 312 (2 sheets). — striata Zucc. subsp. falcata (Engelm.) e try. Agave maculata Engelm. in Torr., Bot. Mex. ound. Surv. 214. 1859. HOLOTYPE: Rio Grande, 1847, 4. Wislizenus 373 (3 sheets). = Manfreda maculosa (Hook.) Rose. Agave maculosa Hook. var. brevituba Engelm., Trans. Acad. Sci. St. Louis 3: 301. 1875. LECTOTYPE: below El Paso, Texas, 1851-1852, 1610 Annals of the Missouri Botanical Garden C. Wright 1905. — (Hook.) Rose. Agave newberryi Engelm., Trans. Acad. Sci. St. Louis 3: 309. 1879. HOLOTYPE: northwestern Arizona, 1858, J. S. Newberry s.n. = A. utahensis Engelm. Agave palmeri Engelm., Trans. Acad. Sci. St. Louis 3: 319. 1875. SYNTYPE: southern Ari- zona, 1869, E. Palmer s.n. (3 sheets). Agave parryi Engelm., Trans. Acad. Sci. St. Louis 3: 312. 1875. SYNTYPES: Arizona Territory, 1871, F. Bischoff s.n.; 1846, Emory s.n.; 1874, J. T. Rothrock 274 (3 sheets). Agave schottii Engelm., Trans. Acad. Sci. St. Louis 06. 1875. HOLOTYPE: Sierra del Pajarito, Arizona, 1855, 4. Schott s.n. Agave shawii Engelm., Trans. Acad. Sci. St. Louis 3: 314. 1875. SYNTYPE: San Diego, California, s.d., G. N. Hitchcock s.n. (2 sheets). Agave utahensis Engelm. in Watson, Bot. King's Expl. 497. 1871. LECTOTYPE: St. George, Utah, 1870, E. Palmer s.n. (2 sheets). Syntype: J. E. Johnson s.n. Agave virginica L. var. tigrina Engelm., Trans. Acad. Sci. St. Louis 3: 302. 1875. HOLOTYPE: South Carolina, 1873, Mellichamp s.n. — Manfreda virginica (L.) Rose. Agave wislizenii Engelm., Trans. Acad. Sci. St. Louis 3: 320. 1875. HOLOTYPE: San Sebastino, Mexico, 1847, A. Wislizenus 280 (2 sheets). = A. parrasana Berger Asyliron cs dle Engelm. ex Trelease, Proc. . Soc. 50: 433. 1911. HOLOTYPE: Presidio del Norte, Texas, 1880, D. Harvard s.n. Manfreda maculosa Yucca angustifolia Pursh var. radiosa Engelm. in Watson, Bot. King's Expl. 496. 1871. LECTOTYPE: Arizona, 1867, E. Palmer 201. = Y. elata Engelm Yucca angustissima Engelm. ex Trelease, Annual Rep. Missouri Bot. Gard. 13: 58. 1902. LECTOTYPE: deserts of the Colorado River, 1853 & 1854, J. M. Bigelow s.n. (2 sheets). Yucca brevifolia Engelm. in Watson, Bot. King's Expl. 496. 1871. LECTOTYPE: Date Creek, s.d., E. Palmer?, MO Nos. 135643 & 135646. Yucca elata Engelm., Bot. Gaz. 7: 17. 1882. LECTOTYPE: Camp Grant, Arizona, s.d., J. T. Rothrock 382. Yucca macrocarpa Engelm., Bot. Gaz. 6: 244. 1881. LECTOTYPE: Santa Rita Mts., 1880, G. Engelmann s.n. — gelm. Yucca schottii Engelm., Trans. Acad. Sci. St. Louis 3: 46. 1873. LECTOTYPE: Santa Cruz River, Arizona, 1855, 4. Schott s.n. Arizona, Y. schottii En- Yucca yucatana Engelm., Trans. Acad. Sci. St. Louis 3: 37. 1873. HOLOTYPE: Nohpat, Yu- catán, Mexico, 1865, A. Schott s.n. = Y. aloifolia L ALISMATACEAE Sagittaria calycina Engelm. in Torr., Bot. Mex. ound. Surv. 212. 1859. LECTOTYPE: Alex- andria, Louisiana, s.d., J. Hale s.n. Syntypes: banks of Missouri, 1856, G. Engelmann s.n.; Kimms Salt Pond, Missouri, 1856, G. En- gelmann s.n. — S. montevidensis Cham. & Schlecht. subsp. calycina (Engelm.) Bogin. Sagittaria calycina Engelm. var. fluitans Engelm. in Torr., Bot. Mex. Bound. Surv. 212. 1859. HOLOTYPE: western Texas, 1851-1852, C. Wright 1899. — S. montevidensis Cham. & Schlecht. subsp. calycina (Engelm.) Bogin. Sagittaria as saq Engelm. var. spongiosa En- gelm. in Gray, Man. Bot. 5: 493. 1868. HOLOTYPE: Wilmington, Delaware, 1860, Tat- nall s.n. — S. montevidensis Cham. & Schlecht. subsp. spongiosa (Engelm.) Bogin. Sagittaria cristata Engelm. in Arthur, Contr. Flor. Iowa 5: 3. 1882. LECTOTYPE & 2 ISOLECTOTYPES: Armstrong, lowa, 1881, Cratty s.n. = S. gra- minea Michx. subsp. cristata (Englem.) Bog- in. Sagittaria graminea Michx. var. platyphylla En- gelm. in Gray, Man. Bot. 5: 494. 1868. LECTOTYPE & ISOLECTOTYPE: Texas, s.d., F. Lindheimer 7 13. Syntype?: Mississippi, 1860, "ex herb A. Wood.” Sagittaria heterophylla Pursh var. p We Engelm. in Gray, Man. Bot. 5: 494. 1868. TYPE MATERIAL?: Lakes, American E (in Illinois, near St. Louis), 1846, N. Riehl s.n. — S. rigida Pursh. Sagittaria recurva Engelm. ex Patterson, Check- list 130. 1887. LECTOTYPE & ISOLECTOTYPE: Texas, s.d., F. Lindheimer 713. — S. gra- minea Engelm. var. platyphylla Engelm. Sagittaria trachysepala Engelm. ex Michelin in C., Monogr. Phan. 3: 74. 1881. ISOTYPE: Texas, s.d., Drummond 423. Sagittaria variabilis Engelm. in Gray, Man. Bot. 1: 461. 1848. LECTOTYPE: St. Louis, Missouri, 1846, G. Engelmann s.n. = S. latifolia Willd. ANACARDIACEAE Rhus microphylla Engelm. ex Gray, Pl. Wright- iana 1: 31. 1852. SYNTYPE: Texas, 1850, F. Lindheimer 734. Volume 75, Number 4 1988 Wolf 1611 Engelmann Type Specimens APIACEAE Apium butleri Engelm. in Watson, Proc. Amer. Acad. Arts 21: 453. 1887. ISOLECTOTYPE: eastern Texas, s.d., E. Hall 244. = Ammo- selinum butleri (Engelm.) Coult. & Rose. Cynosciadium pinnatum var. pumilum Engelm. & Gray, Boston J. Nat. Hist. 5: 218. 1845. TYPE MATERIAL: no locale but collected by F. Lindheimer s.n., 1843; prairie west of the Brazos, Texas, F. Lindheimer s.n. Annotated "n. sp." by Engelmann. Daucosma laciniata Englem. & Gray in Gray, Boston J. Nat. Hist. 6: 211. 1850. TYPE MAT- ERIAL: upper Guadalupe, Texas, 1846, F. Lindheimer s.n. Eryngium heterophyllum Engelm. in Wisliz., Mem. . No. Mex. 107. 1848. HOLOTYPE: Co- sihuiriachi, Mexico, 1846, A. Wislizenus 176. ARISTOLOCHIACEAE Aristolochia longiflora Engelm. & Gray, Boston J. Nat. Hist. 5: 259. 1845. HOLOTYPE: Texas, 1844, F. Lindheimer s.n. ASCLEPIADACEAE Asclepias brachystephana Engelm. in Torr., Bot. ound. Surv. 163. 1859. LECTOTYPE: New Mexico, 1851-1852, C. Wright 1692. Asclepias euphorbiifolia Engelm. ex Gray, Proc Amer. Acad. Arts 16: 104. 1881. LECTOTYPE: San Luis Potosi, Mexico, s.d., Schaffner 55. Asclepias involucrata Englem. in Torr., Bot. Mex. Bound. Surv. 163. 1859. LECTOTYPE: Copper Mines, New Mexico, 1851-1852, C. Wright 1690. Asclepias leucophylla Engelm. in Parry, Amer. Naturalist 9: 348. 1875. HOLOTYPE: southern Utah, 1874, C. C. Parry 207. — A. erosa Torr. Asclepias lindheimeri Engelm. & Gray, Boston J. t. Hist. 5: 250. 1845. SYNTYPE: Industry, Texas, 1844, F. Lindheimer 272. = A. oe- notheroides Cham. & Schlecht. Asclepias sullivantii Engelm. ex Gray, Man. Bot. 1: 366. 1848. SYNTYPE: St. Louis, Missouri, 1834, G. Engelmann 451. Astephanus utahensis Englem in Parry, Amer. Naturalist 9: 347. 1875. PROBABLE HOLOTYPE: southern Utah, 1874, C. C. Parry 209. En- gelmann wrote " on it. Gonolobus curtisii Engelm. in im apparently not published, but annotated “‘n. ` by En- gelm. TYPE MATERIAL: North ie a 1844, Matelea gonocarpa (Walt.) "spec. nov. Curtis s.n. — Shinners. Gonolobus cynanchoides Engelm. & Gray, Boston J. ist. 5: 251. 1845. TYPE MATERIAL: Industry, Texas, 1844, F. Lindheimer 273. — Matelea cynanchoides (Engelm.) Woods. Gonolobus reticulatus Engelm. ex Gray, Proc. Amer. Acad. Arts 12: 75. 1877. TYPE Ma- TERIAL: Guadalupe, Texas, 1846, F. Lind- heimer 461. — Matelea reticulatus (Engelm.) Woods. Mellichampia filifolia Engelm., apparently not published, but annotated “‘n. sp." by Engelm. TYPE MATERIAL: St. George, Utah, July 1874, Parry s.n. Sarcostemma heterophyllum Engelm. in Torr. t. Mex. Bound. Surv. 362. 1859. HOLOTYPE: Nn Mexico, 1851-1852, C. Wright 1681. ASTERACEAE Agassiza suavis Gray & Engelm., Proc. Amer. Arts 1: 46. 1847. SYNTYPE: San An- tonio, Texas, 1845, F. Lindheimer 351 dis- tributed as No. 437 for 1846 (2 sheets). — Gaillardia suavis (Gray & Englem.) Britton & Rusby Aster anomalus Engelm. in Torrey & Gray, Fl. Amer. 2: 503. 1843. TYPE MATERIAL: Rockspring near Prairie du Pont, Sept. 1842, G. Engelmann s.n. The only prepublication dated specimen. Aster azureus Lindl. ex Hook. var. vernalis En- gelm. ex Small, Fl. S.E. U.S. 1215. 1903. HOLOTYPE: prairies west of Houston, Texas, 1842, F. Lindheimer s.n (2 sheets). Keerlia bellidifolia Gray & Engelm., Proc. Amer. Acad. Arts 1: 47. 1846. SYNTYPE: Guada- lupe, Texas, 1845, F. Lindheimer 415. Liatris acidota Engelm. & Gray, Boston J. Nat. Hist. 5: 218. 1845. ISOTYPES: Texas, 1843, F. Lindheimer 72, 73. Lindheimera texana Gray & Engelm., Proc. Amer. Acad. Arts 1: 47. 1847. SYNTYPES: New Braunfels, Texas, 1846, F. Lindheimer 424 (2 sheets); New Braunfels, Texas, 1847, F. Lindheimer Porophyllum amplexicaule Engelm. ex Gray, PI. Wrightiana 1: 120. 1852. HOLOTYPE?: near Messillas, Mexico, s.d., J. Gregg s.n., anno- tated “n. sp." by Engelm. Rudbeckia missouriensis Engelm. ex Boynton & Beadle, Biltmore Bot. Stud. 1: 17. 1901. TYPE MATERIAL: Meramec Station, Missouri, July 1879, H. Eggert s.n.; Allentown, Missouri, Sept. 1879, G. Letterman s.n; Allentown, Missouri, July 1879, G. Letterman s.n. Sanvitalia angustifolia Engelm. ex Gray, Pl. 1612 Annals of the Missouri Botanical Garden Wrightiana 1: 112. 1852. HOLOTYPE?: Buena Vista, Mexico, s.d., J. Gregg 274, annotated "n. sp." by Engelm. Vernonia lettermanii Engelm. ex Gray, Proc. Amer. Acad. Arts 16: 78. 1881. LECTOTYPE: Washita, Arkansas, 1879, G. W. Letterman s.n. Syntypes: Arkansas, 1879 & 1880, G. W. Letterman s.n. (2 sheets). Vernonia lindheimeri Gray & Engelm., Proc. Amer. Acad. Arts 1: 46. 1848. HOLOTYPE: New Braunfels, Texas, 1846, F. Lindheimer 408. Finnia intermedia Engelm. in Wisliz., Mem. Tour. ex. 107. 1848. HOLOTYPE: Cosihuiri- achi, Mexico, 1846, 4. Wislizenus 182. = Z. peruviana (L.) L BORAGINACEAE Lithospermum breviflorum Engelm., Trans. Amer. Philos. Soc. 12: 203. 1861. HOLOTYPE?: 200 miles above Fort Pierre, 1853-1854, F. V. Hayden s.n., only Hayden specimen and “breviflorum” added by Engelm. Myosotis inflexa Engelm., Amer. J. Sci. 46: 98. 1844. HOLOTYPE?: no locale, 1842, C. A. Gey- er s.n., annotated “n. sp." by Engelm. Myosotis macrosperma Engelm., Amer. J. Sci. 46: 98. 1844. HOLOTYPE: Prairies, Texas, 1839, F. Lindheimer s.n., annotated “n. sp." by Engelm. BRASSICACEAE Dithyrea wislizenii Englem. in Wisliz., Mem. Tour. o. Mex. 97. 1848. TYPE MATERIAL: Mexico, 1846, A. Wislizenus 63 (2 sheets). Nasturtium calycinum Engelm., Trans. Amer. Philos. Soc. 12: 184. 1859. LECTOTYPE: Yel- lowstone River, 1853-1854, F. V. Hayden 93. Syntype: Ft. Union, mouth of Yellow- stone, 1855?, F. V. Hayden s.n. = Rorippa calycinum (Engelm.) Rydb. Sisymbrium incisum Engelm. ex Gray, Mem. Amer. Acad. Arts. II. 4: 8. 1849. SYNTYPES: Santa Fe Creek, New Mexico, 1847, A. Fend- ler 30 & 31; Moro, New Mexico, 1847, A. Fendler 29. = Descuriana richardsonii (Sweet) O. E. Schulz. subsp. incisa (Engelm.) Delting. Vesicaria auriculata Englem. & Gray, Boston J. Nat. Hist. 5: 240. 1845. ISOTYPE: Brazos, Texas, 1844, F. Lindheimer 271 (2 sheets). — Lesquerella auriculata (Englem. & Gray) S. Wats. Vesicaria recurvata Englem. ex Gray, Boston J. Nat. Hist. 6: 240. 1850. ISOLECTOTYPE: New Braunfels, Texas, 1846, F. Lindheimer 8. Syntype: New Braunfels, Texas, 1846, F. Lindheimer 330. — Lesquerella recurvata (Englem. ex Gray) S. Wats. CACTACEAE Cactus gabbii Englem. in Coult., Contr. U.S. Natl. Herb. 9. 1894. HOLOTYPE: erase of California, Mexico, 1867, W. Gab Mammillaria brandegeei (Coult.) i Bran- degee. Cereus berlandieri Englem., Proc. Amer. Acad. Arts 3: 286. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 38. 1859. LECTOTYPE: southern Texas, 1834, Berlandier 2433. = Echinocereus berlan- dieri (Englem.) Engelm. ex Rumpler. Cereus caespitosus Engelm. in Engelm. & Gray, Boston J. Nat. Hist. 5: 247. 1845. LECTOTYPE: cult. from Industry, Texas, 1845, F. Lind- heimer s.n. = Echinocereus reichenbachii (Tenscheck) Haage, f., ex Britt. & Rose. Cereus caespitosus Engelm. var. castaneus En- gelm. in Gray, Boston J. Nat. Hist. 6: 202. 1850. LECTOTYPE: Liano, Texas, 1847, F. Lindheimer s.n. = Echinocereus reichenba- chii (Tenscheck) Haage, f., ex Britt. & Rose. Cereus caespitosus Engelm. var. major Engelm., Proc. Amer. Acad. Arts 3: 280. 1856. LECTOTYPE: Texas, 1851, F. Lindheimer s.n. = Echinocereus reichenbachii (Tenscheck) Haage, f., ex Britt. & Rose. Cereus calvus Engelm. in Coult., Contr. U.S. Natl. Herb. 3: 409. 1896. HOLOTYPE: Peninsula of California, Mexico, 1867, IV. Gabb 2 Cereus chloranthus Engelm., Proc. Amer. Acad. Arts 3: 278. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 29. 1859. LECTOTYPE: El Paso and Stony Hills at Frontera, Texas, 1852, C. Wright 95 (3 sheets, 1 box). = Echinocereus chloranthus Engelm Cereus coccineus Engelm. in Gray, Mem. Amer. Acad. Arts. II. 4: 51. 1849. LECTOTYPE: Wolf Creek, Mexico, 1846, A. Wislizenus s.n. (2 sheets). Syntype: Santa Fe, New Mexico, s.d., A. Fendler 272. — Echinocereus triglochi- diatus Engelm. melanacanthus (En- gelm.) L. Benson. Cereus coccineus Engelm. var. cylindricus En- gelm. in Gray, Mem. Amer. Acad. Arts. Il. var. Volume 75, Number 4 1988 Wolf 1613 Engelmann Type Specimens 4: 51. 1849. LECTOTYPE: Santa Fe, New Mex- ico, 1846, A. Fendler s.n. ^ Echinocereus triglochidiatus Engelm. var. melanacanthus (Engelm.) L. Benson. Cereus coccineus Englem. var. melanacanthus Engelm. in Gray, Mem. Amer. Acad. Arts. II. 4: 51. 1849. LECTOTYPE: Santa Fe, New Mex- ico, 1846, A. Fendler 4. = Echinocereus triglochidiatus Engelm. var. melanacanthus (Engelm.) L. Benson. Cereus conoides Engelm. & Bigelow, Proc. Amer. Acad. Arts 3: 284. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 35. 1857. LECTOTYPE: Anton Chico, New Mexico, 1853, J. M. Big- elow s.n. Syntype: 1853, J. M. Bigelow s.n. = Echinocereus triglochidiatus Englem. var. melanacanthus (Engelm.) L. Benson. Cereus ctenoides Engelm., Proc. Amer. Acad. Arts 3: 279. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 31. 1859. LECTOTYPE: Santa Rosa, Mexico, 1853, J. M. Bigelow s.n. (2 sheets). Syntypes: Pecos, 1851, C. Wright s.n.; Rio Grande, 1853, J. M. Bigelow s.n. = Echinocereus pectinatus (Scheidw.) Engelm. Cereus dasycanthus Engelm. in Gray, Mem. Amer. Acad. Arts. IT. 4: 50. 1849. NEOTYPE: between San Antonio and El Paso, Texas, 1849, C. Wright s.n. = Echinocereus pectinatus (Scheidw.) Engelm. (Standl.) L. Benson. Cereus dasycanthus Engelm. var. minor Engelm., Proc. Amer. Acad. Arts 3: 279. 1850. LECTOTYPE: Frontera, Texas, 1851, C. Wright s.n. (2 sheets). Syntype: Chihuahua, Mexico, 1852, J. M. Bigelow s.n. — Echinocereus pectinatus (Scheidw.) Englem. var. (Engelm.) L. Benson. Cereus dubius Engelm., Proc. Amer. Acad. Arts 3: 282. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 36. 1859. LECTOTYPE: Rio Grande, Texas, 1852, C. Wright 410. — Echinocereus en- neancanthus Engelm. var. dubia (Engelm.) L. Benson. Cereus emoryi Engelm., Amer. J. Sci. 14: 338. 1852. LECTOTYPE: San Diego, California, 1850, C. C. Parry s.n. Cereus engelmannii Parry var. chrysocentrus En- & Bigelow, in Engelm., Proc. Amer. Acad. Arts 3: 283. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 35. 1857. LECTOTYPE: var. neomexicanus minor Bill Williams River, California, 1854, J. M. Bigelow s.n. — Parry var. chrysocentrus (Engelm. & Big- elow) Engelm. ex Rumpler. Cereus engelmannii Parry var. variegatus En- gelm. & Bigelow, in Engelm., Proc. Amer. Acad. Arts 3: 283. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 35. 1857. LECTOTYPE: Bill Williams Fork, California, 1854, J. M. Bigelow s.n. = Parry var. variegatus (Engelm. & Bigelow) Engelm. ex Rumpler. Cereus fendleri Engelm. in Gray, Mem. Amer. Acad. Arts. II. 4: 51. 1849. HOLOTYPE: Santa Fe, New Mexico, 1846, 4. Fendler 3. = Echi- nocereus fendleri (Engelm.) Engelm. ex Rum- pler. Cereus fendleri Engelm. var. pauperculus En- gelm. in Gray, Mem. Amer. Acad. Arts. II. 4: 51. 1849. HOLOTYPE: Santa Fe, New Mex- ico, 1846, A. Fendler s.n. = Echinocereus fendleri (Engelm.) Engelm. ex Rumpler. Cereus flaviflorus Engelm. ex Coult., Contr. U.S. Natl. Herb. 3: 391. 1896. HOLOTYPE: Santa Borga, Mexico, 1867, W. Gabb 10. = Echi- nocereus maritimus (Jones) Schumann. Cereus flexuosus Engelm. ex Coult., Contr. U.S. Natl. Herb. 3: 411. 1896. HOLOTYPE: Pen- Echinocereus engelmannii Echinocereus engelmannii insula of California, Mexico, 1867, W. Gabb 5. = Machaerocereus gummosus (Engelm.) Britt. & Ros Cereus asian ri Engelm. & Bigelow, in En- gelm., Proc. Amer. Acad. Arts 3: 285. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 33. 1857. LECTOTYPE: Cedar Woods, west of Zuni, New Mexico, 1853, J. M. Bigelou s.n. = Echinocereus triglochidiatus Engelm. var. gonacanthus (Engelm. & Bigelow) L. Benson. Cereus greggii Engelm. var. cismontanus Engelm. n Wisliz., Mem. Tour. No. Mex. 102. 1848. LECTOTYPE: Cadena, Mexico, 1847, J. Gregg s.n. Syntype: Paso del Norte, Mexico, 1846, A. Wislizenus 222. Cereus hexaedrus Engelm., Proc. Amer. Acad. Arts 3: 285. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 34. 1857. LECTOTYPE: near Zuni, New Mexico, 1853, J. M. Bigelow s.n. — Echinocereus triglochidiatus Engelm. var. mojavensis (Engelm. & Bigelow) L. Benson. Cereus longisetus Engelm., Proc. Amer. Acad. Arts 3: 280. 1856. HOLOTYPE: Santa Rosa, 1614 Annals of the Missouri Botanical Garden Mexico, 1853, J. M. Bigelow s.n. = Echi- nocereus longisetus (Engelm.) Engelm. Cereus mamillatus Engelm. ex Coult., Contr. U. S. Natl. Herb. 3: 405. 1896. HOLOTYPE: Peninsula of California, Mexico, 1867, Gabb 16. = Echinocereus mamillatus (En- gelm. ex Coult.) Britt. & Rose. Cereus mojavensis Engelm. & Bigelow in Engelm., Proc. Amer. Acad. Arts 3: 281. 1857 (pre- print, 1956); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 33. 1857. HOLOTYPE: Mojave Creek, California?, 1854, J. M. Bigelow s.n. = Echinocereus triglo- chidiatus Engelm. var. mojavensis (Engelm. & Bigelow) L. Benson. Cereus mojavensis Engelm. var. gelm. & Bigelow in Engelm., Proc. Amer. Acad. Arts 3: 281. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 33. 1857. LECTOTYPE: Colorado Chiquito, Arizona, 1853, J. M. Big- elow s.n. = Echinocereus triglochidiatus En- gelm. var. melanacanthus (Engelm. & Big- elow) L. Benson. Cereus Engelm., Proc. Amer. Acad. Arts 3: 285. 1857 (preprint, 1856); in Emory, Rept. U.S p Mex. Bound. Surv. 2: Cactaceae 37. 1859. LECTOTYPE: cultivated from C Wright & J. M. Bigelow s.n. = triglochidiatus Engelm. var. paucispinus (Engelm.) Engelm. ex W. T. Marshall. Cereus pectinatus Scheidw. var. rigidissimus En- gelm., Proc. Amer. Acad. Arts 3: 279. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 31. 1859. LECTOTYPE: Sonora, Mexico, 1855, 4. Schott s.n. = Echinocereus pectinatus (Scheidw.) Engelm. var. rigidissimus (Engelm.) Engelm. ex Rumpler. Cereus procumbens Engelm. in Gray, Mem. Amer. Acad. Arts. II. 4: 50. 1849. LECTOTYPE: near the mouth of the Rio Grande below Matamoras by the Missouri Volunteers, 1846, and cult. St. Louis May 1848. talophus (DC.) Rumpler. Cereus roemeri Engelm. in Gray, Mem. Amer. Acad. Arts. II. 4: 51. 1849. LECTOTYPE: Liano River, Texas, 1847, F. Lindheimer s.n. (2 sheets). = Echinocereus triglochidiatus En- gelm. var. melanacanthus (Engelm. & Big- elow) Cereus schottii Engelm., Proc. Amer. Acad. Arts 8. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 44. 1859. HOLOTYPE: Sierra de Sonoyita, Mex- ico, s.d., A. Schott 3. zuniensis En- Echinocereus = Echinocereus pen- — .. Benson. Cereus stramineus Engelm., Proc. Amer. Acad. Arts 3: 282. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 35: 1859. LECTOTYPE: near El Paso, Texas, 1851, C. Wright s.n. Syntype: J. M. Bigelow s.n. = Echinocereus enneacanthus Engelm. Cereus thurberi Engelm., Amer. J. Sci. II. 17: 234. 1854. LECTOTYPE: Sonora, Mexico, 1851, G. Thurber s.n. Cereus titan Engelm. in Coult., Contr. U. S. Natl. Herb. 3: 409. 1896. HOLOTYPE: Cape San Lucas to San Quentin, Mexico, 1867, IV. Gabb 1. = Pachycereus pringlei (S. Wats.) Britt. & Rose. Cereus viridiflorus Engelm. var. cylindricus En- gelm., Proc. Amer. Acad. Arts 3: 278. 1856 LECTOTYPE: New Mexico, 1851, C. Wright s.n. Echinocactus bicolor Galeotti var. schottii En- gelm., Proc. Amer. Acad. Arts 3: 277. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 27. 1859. HOLOTYPE: Cretaceous hills, near Mier, Mex- ico, 1853, A. Schott s.n. = Thelocactus bi- color Galeotti var. schottii (Engelm.) Krainz. Echinocactus dasycanthus Engelm., Proc. Amer. Acad. Arts 3: 277. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 28. 1859. LECTOTYPE: El Paso, Texas, 1852, C. Wright s.n. (2 sheets). Syn- type: Frontera, 1852, C. Wright 867 = Neolloydia intertexta (Engelm.) L. Benson var. dasycanthus (Engelm.) L. Benson. Echinocactus emoryi Engelm. var. rectispinus Engelm. ex Coult., Contr. U.S. Natl. Herb. 3 362. 1896. HoLoTYPE: Peninsula of California, Mexico, 1867, IV. Gabb 12. = Ferocactus wislizenii Engelm. Echinocactus flexispinus Engelm. in Wisliz., Mem. our. No. Mex. 112. 1848. HOLOTYPE: Pe- layo, between Chihuahua and Parras, Mexico, 1847, A. Wislizenus s.n. matacanthus abis E dU var. crassispinus (Engelm.) L. Benson. On same sheet as F. hamatacanthus (Muklenpfordt) Britt. & Rose var. crassispinus Engelm. Echinocactus hamatacanthus Muhlenpfordt var. crassispinus Engelm., Proc. Amer. Acad. Arts 3: 273. 1856. HOLOTYPE: Pelayo, between Chihuahua and Parras, Mexico, 1847, Wislizenus s.n. On same sheet as E. flexi- spinus Engelm. = Ferocactus hamatacan- thus (Muhlenpfordt) Britt. & Rose var. cras- sispinus Engelm Echinocactus horizonthalonius Lem. var. cen- trispinus Engelm., Proc. Amer. Acad. Arts 3: — Ferocactus ha- Volume 75, Number 4 1988 Wolf Engelmann Type Specimens 1615 276. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 26. 1859. LECTOTYPE: Frontera, New Mexico Ter- ritory, 1851-1852, C. Wright s.n. (2 sheets) = Echinocactus horizonthalonius Lem. Echinocactus intertextus Engelm., Proc. Amer. Acad. Arts 3: 277. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 27. 1859. LECTOTYPE: El Paso, Texas, s.d., C. Wright or J. M. Bigelow (no coll.), s.n. (2 sheets). Syntype: Chihuahua, 1846, A. Wislizenus s.n. = tertextus (Engelm.) L. Benson. Echinocactus lecontei Engelm., Proc. Amer. Acad. Arts 3: 274. 1857 (preprint, 1856); U.S. Senate Rept. Expl. € Surv. R.R. Route Pacific Ocean. Botany 4: 29. 1857. LECTOTYPE: Bill Williams Fork, Arizona, 1854, J. M. Bigelow s.n. Syntype: Gilia, Leconte s.n. = Ferocactus acanthodes (Lemaire) Britt. & Rose var. le- contei (Engelm.) Lindsay. Echinocactus limitus Engelm. in Coult., Contr. atl. Herb. 3: 374. 1896. LECTOTYPE: south of San Diego, 1878, Parker & Hitch- cock s.n. (2 sheets). = Ferocactus viridescens (Nutt.) Britt. & Rose. Echinocactus longehamatus Galeotti var. brev- ispinus Engelm., Proc. Amer. Acad. Arts 3: 273. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 22. 1859. LECTOTYPE: rocky mountains of the Limpia, Texas, 1852, J. M. Bigelow s.n. = Ferocactus hamatacanthus (Muhlenpfordt) Britt. & Rose. Echinocactus orcuttii Engelm., West. Amer. Sci. 2: 46. 1886. HOLOTYPE: Palm Valley, northern Lower California, 1883, C. R. Orcutt 641 (2 sheets). = Ferocactus orcuttii (Engelm.) Britt. & Rose. Echinocactus peninsulae Engelm. ex Coult., Contr. U.S. Natl. Herb. 3: 361. 1896. HOLOTYPE: Cape San Lucas to San Diego, 1867, IV. Gabb 11 Neolloydia in- ulasan polyancistrus Engelm. & Bigelow in Engelm., Proc. Amer. Acad. Arts 3: 272. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 29. 1857. HOLOTYPE: head of the Mojave, California?, 1854, J. M. Bigelow s.n. = Sclerocactus polyancistrus (Engelm. & Big- elow) Britt. & Rose. Echinocactus polycephalus Engelm. & Bigelow in Engelm., Proc. Amer. Acad. Arts 3: 276. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 31. 1857. LECTOTYPE: on Mojave River and Mojave Valley, California?, 1854, J. M. Big- elow s.n. (3 sheets Echinocactus polycephalus Engelm. & Bigelow var. xeranthemoides Engelm. in Coult., Contr. U.S. Natl. Herb. 3: 358. 1859. LECTOTYPE: northern Arizona, 1881, A. L. Siler s.n. Echinocactus pubispinus Engelm., Trans. Acad. Sci. St. Louis 2: 199. 1863. HOLOTYPE: Utah, 1851, H. Engelmann 1. = Sclerocactus pu- bispinus (Engelm.) L. Benson. Echinocactus setispinus Engelm. in Engelm. & Gray, Boston J. Nat. Hist. 5: 246. 1845. LECTOTYPE: Colorado River, 1844, F. Lind- Ferocactus setispinus (En- gelm.) L. Benson. Echinocactus sileri Engelm. ex Coult., Contr. U.S. Natl. Herb. 3: 376. 1896. HOLOTYPE: Cotton- wood Springs and Pipe Springs, Arizona, 1883, A. L. Siler s.n. (2 sheets). = sileri (Engelm.) L. Benson. Echinocactus simpsoni; Engelm., Trans. Acad. Sci. St. Louis 2: 197. 1863. LECTOTYPE: Kobe Valley, Nevada, 1859, H. Engelmann s.n. Api Utah Desert, s.d., H. Engelmann — Pediocactus simpsonii (Engelm.) Britt. " Hase. Echinocactus simpsonii Engelm. var. minor En- gelm., Trans. Acad. Sci. St. Louis. 2: 197. 1863. LECTOTYPE: Mt. Vernon, Parry, Hall & Harbour s.n. = Pediocactus simpsonii En- elm. var. minor (Engelm.) Cockerell. Echinocactus uncinatus Hopf. ex Foerst. var. wrightii Engelm., Proc. Amer. Acad. Arts 3: 272. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 20. 1859. LECTOTYPE: Frontera, Texas, 1852, C. Wright 88. Syntype: Frontera, Texas, 1852, J. M. Bigelow s.n. = Ancistrocactus unci- natus (Galeotti) Benson var. wrightii (En- gelm.) L. Benson. Echinocactus unquispinus Engelm. in Wisliz Mem. Tour. No. Mex. 12. 1848. LECTOTYPE: Pelayo, Mexico, s.d., A. Wislizenus s.n. = Echinomastus unquispinus (Engelm.) Britt. & Rose. Echinocactus viridescens Nutt. var. cylindraceus Engelm., Amer. J. Sci. II. 14: 338. 1852. HOLOTYPE: San Felipe, California, 1850, C. C. Parry s.n. — Ferocactus acanthodes (Le- maire) Britt. & Rose. Echinocactus whipplei Engelm. & Bigelow in En- gelm., Proc. Amer. Acad. Arts 3: 271. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & rv. R.R. Route Pacific Ocean. Botany 4: 28. 1857. HOLOTYPE: Colorado Chiquito, Ar- izona, 1853, J. M. Bigelow s.n. — Sclero- heimer s.n. — Pediocactus 1616 Annals of the Missouri Botanical Garden cactus whipplei (Engelm. & Bigelow) Britt. Rose. Echinocactus whipplei Engelm. var. spinosior Engelm., Trans. Acad. Sci. St. Louis 2: 199. 1863. LECTOTYPE: Camp Floyd, Utah, 1859, H. Engelmann s.n. (2 sheets). = Sclerocactus spinosior (Engelm.) Woodruff & L. Benson. Echinocactus wislizenii Engelm. in Wisliz., Mem. Tour. No. Mex. 96. 1848. LECTOTYPE: Do- nana, Mexico, 1846, A. Wislizenus s.n. (3 sheets). = Ferocactus wislizenii Engelm. Echinocactus wislizenii Engelm. var. decipiens Engelm. in Rothr., Bot. Wheeler's Surv. 128. 1878. HOLOTYPE: Camp Bowie, Arizona, 1874, J. T. Rothrock 492. = Ferocactus wislizenii Engelm. Echinocereus adustus Engelm. in Wisliz., Mem. Tour. No. Mex. 104. 1848. HOLOTYPE: Co- sihuiriachi, Mexico, 1846, A. Wislizenus s.n. (2 sheets). Echinocereus chloranthus Engelm. ex Rumpler in Forster, Handb. Cact. 2nd edition: 812. 1885. SYNTYPE: Texas, 1852, J. M. Bigelow s.n. Echinocereus coccineus Engelm. in Wisliz., Mem. Tour. No. Mex. 100. 1848. HOLOTYPE: Wolf Creek, Santa Fe, New Mexico, 1846, 4. Wis- lizenus s.n. (2 sheets). = F. triglochidiatus Engelm. var. melanacanthus (Engelm.) L. Benson. Echinocereus dasyacanthus Engelm. in Wisliz., em. Tour. No. Mex. 100. 1848. NEOTYPE: San Antonio to El Paso, Texas, 1849, C. Wright s.n. = E. pectinatus Scheidw. var. neomexicanus (Coult.) L. Benson. Echinocereus enneacanthus Engelm. in Wisliz., Mem. Tour. No. Mex. 112. 1848. HOLOTYPE: south of Chihuahua, Mexico, 1847, A. Wis- lizenus 244. Echinocereus polyacanthus Engelm. in Wisliz., em. Tour. No. Mex. 104. 1848. LECTOTYPE: Cosihuiriachi, Mexico, 1847, A. Wislizenus s.n. (2 sheets + 1 box). = E. triglochidiatus Engelm. var. neomexicanus (Standley) Ben- son. Echinocereus radians Engelm. in Wisliz., Mem. Tour. No. Mex. 104. 1848. HOLOTYPE: Co- sihuiriachi, Mexico, 1846, A. Wislizenus s.n. = E. adustus Engelm. Echinocereus rufispinus Engelm. in Wisliz., Mem. Tour. No. Mex. 106. 1848. HOLOTYPE: Co- sihuiriachi, Mexico, 1846, A. Wislizenus s.n. Echinocereus triglochidiatus Engelm. in Wisliz., em. Tour. No. Mex. 93. 1848. LECTOTYPE: Wolf Creek, New Mexico, 1846, 4. Wisli- zenus s.n. (2 sheets + 1 box). Echinocereus viridiflorus Engelm. in Wisliz., Mem. o. Mex. 91. 1848. LECTOTYPE: Wolf Creek, New Mexico, 1846, A. Wislizenus 514 (2 sheets). Mammillaria applanata Engelm. in Wisliz., Mem. Tour. No. Mex. 105. 1848. LECTOTYPE: west- ern Texas, 1845, F. Lindheimer s.n. = M. . var. applanata (En- Mammillaria arizonica Engelm. in Brewer & Watson, Bot. Calif. 1: 244. 1876. LECTOTYPE: Arizona, s.d., Coues & E. Palmer s.n. Syn- type: Arizona, 1871, Bischoff s.n. = Cory- phantha vivipara (Nutt.) Britt. & Rose var. arizonica (Engelm.) W. T. Marshall. Mammillaria barbata Engelm. in Wisliz., Mem. Tour. No. Mex. 105. 1848. HOLOTYPE: Co- sihuiriachi, Mexico, s.d., A. Wislizenus s.n. = Neomammillaria barbata (Engelm.) Britt. & Rose. Mammillaria chlorantha Engelm. in Rothr., Bot. Wheeler's Surv. 128. 1878. LECTOTYPE: St. George, Utah, 1874, C. C. Parry s.n. = Cor- yphantha vivipara (Nutt.) Britt. & Rose var. desertii (Engelm.) W. T. Marshall. Mammillaria compacta Engelm. in Wisliz., Mem. Tour. No. Mex. 105. 1848. Cosihuiriachi, Mexico, 1846, 4. Wislizenus s.n. Mammillaria dasyacantha Engelm., Proc. Amer. Acad. Arts 3: 269. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 15. 1859. LECTOTYPE & PROBABLE ISOLECTOTYPE: El Paso, Texas, C. Wright s.n. — Coryphantha dasyacantha (Engelm.) Or- cutt. Mammillaria desertii Engelm. in Watson, Bot. Calif. 2: 449. 1880. LECTOTYPE: “Ivanpah,” 1880, S. B. & W. F. Parish 455 (3 sheets). — Coryphantha vivipara (Nutt.) Britt. & Rose var. desertii (Engelm.) W. T. Marshall. Mammillaria echinus Engelm., Proc. Amer. Acad. Arts 3: 267. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 13. 1859. LECTOTYPE: western Texas, 1849, C. Wright s.n. (2 sheets). Syntypes: Presidio del Norte, 1852, J. M. Bigelow s.n. (2 sheets); western Texas, 1852, C. Wright s.n.; east of El Paso, Texas, s.d., C. Wright s.n. = Cor- yphantha cornifera DC. var. echinus (En- gelm.) L. Benson. Mammillaria fissurata Engelm., Proc. Amer. Acad. Arts 3: 270. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 18. 1859. LECTOTYPE: Rio Bravo del Norte, Texas, 1852, A. Schott s.n. Syntype: Pecos, Volume 75, Number 4 1988 Wolf Engelmann Type Specimens 1617 1852, J. M. Bigelow s.n. — Ariocarpus fis- surata (Engelm.) K. Schumann. Mammillaria grahami Engelm., Proc. Amer. Acad. Arts 3: 262. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 7. 1859. LECTOTYPE: El Paso, Tex- as, 1852, C. Wright s.n. (2 sheets). Syntypes: Colorado Basin, 1855, A. Schott s.n.; Gila, C. C. Parry s.n Mammillaria gummifera Engelm. in Wisliz., Mem. Tour. No. Mex. 106. 1848. LECTOTYPE: Co- sihuiriachi, Mexico, 1848, A. Wislizenus s.n. — M. heyderi Muhlenpfordt var. gummifera (Engelm.) L. Benson. Mammillaria hemisphaerica Engelm. in Wisliz., Mem. Tour. No. Mex. 106. 1848. LECTOTYPE: cult. in St. Louis, Missouri from Matamoras, 1846, St. Louis Volunteers. = M. heyderi Muhlenpfordt var. hemisphaerica Engelm. Mammillaria lasiacantha Engelm. var. denuta Engelm., Proc. Amer. Acad. Arts 3: 261. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 5. 1859. LECTOTYPE: west of the Pecos, 1852, C. Wright .n. = M. lasiacantha Engelm. Mammillaria lasiacantha Engelm. var. minor En- gelm., Proc. Amer. Acad. Arts 3: 261. 1857 (preprint, 1856); in Emory, Rept. U.S. € ex. Bound. Surv. 2: Cactaceae 5. 1859. LECTOTYPE: west of the Pecos, s.d., C. Wright s.n. = M. lasiacantha Engelm. Mammillaria macromeris Engelm. in Wisliz., Mem. Tour. No. Mex. 98. 1848. LECTOTYPE: Donana, New Mexico, 1846, A. Wislizenus — Coryphantha macromeris Mammillaria meiacantha Engelm., Proc. Amer. Acad. Arts 3: 263. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 27. 1857. LECTOTYPE: east of the Pecos, 1853, J. M. Bigelow s.n. — M. gummifera Engelm. var. meiacantha (Engelm.) L. Benson. Mammillaria micromeris Engelm., Proc. Amer. Acad. Arts 3: 260. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 3. 1859. LECTOTYPE: San Filipe to the Pecos, Texas, 1851, C. Wright s.n. Syn- types: western Texas, 1852, C. Wright s.n. (2 sheets) & 1849, C. Wright s.n. = Epi- thelantha micromeris (Engelm.) Weber. Mammillaria micromeris Engelm. var. greggii Engelm., Proc. Amer. Acad. Arts 3: 261. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 4. 1859. LECTOTYPE: Saltillo, Mexico, 1848, J. Gregg 8. = Epithelantha micromeris L. Benson. Mammillaria nuttallii Engelm. in Gray, Mem. Amer. Acad. Arts. II. 4: 49. 1849. LECTOTYPE & PROBABLE ISOLECTOTYPE: Ft. Pierre, South Dakota, 1847, F. V. Hayden s.n. = Cory- phantha missouriensis (Sweet) Britt. & Rose. Mammillaria papyracanthus Engelm. in Gray, Mem. Amer. Acad. Arts. II. 4: 49. 1849 LECTOTYPE: Santa Fe, New Mexico, 1857, A. Fendler s.n. = Pediocactus papyracanthus (Engelm.) L. Benson. Mammillaria pectinata Engelm., Proc. Amer. Acad. Arts 3: 266. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 12. 1859. LECTOTYPE: on the Pe cos, Texas, 1849, C. Wright (3 sheets). — Coryphantha cornifera (DC.) L. Benson. Mammillaria phellosperma Engelm., Proc. Amer. Acad. Arts 3: 262. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 6. 1859. LECTOTYPE: Mojave Creek, 1854, J. M. Bigelow s.n. Syntypes: Cila, s.d., LeConti s.n.; Bill Williams Fork, 1854, J. M. Bigelow s.n. = M. tetrancistra Engelm. Mammillaria pusilla Sweet var. texana Engelm., Proc. Amer. Acad. Arts 3: 261. 1857 (pre- print, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 5. 1859. SYNTYPES: aad Pass to Santa Rosa, s.d., J. M. Bigelow s.n.; Rio Grande, s.d., Poselger s.n. — M. per (Miller) Haw. var. texana (Engelm.) Mammillaria radiosa Engelm. in Gray, Boston J. Nat. Hist. 6: 196. 1850. LECTOTYPE: western Texas, 1846, cult. St. Louis, Missouri, F. Lindheimer s.n. (2 sheets). = Coryphantha vivipara (Nutt.) Britt. & Rose var. radiosa (Engelm.) Backeberg. Mammillaria recurvispina Engelm., Proc. Amer. Acad. Arts 3: 266. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 12. 1859. LECTOTYPE: Sierra del Pajarito, Mexico, 1855, 4. Schott s.n. = Cor- yphantha recurvata (Engelm.) Britt. & Rose. Mammillaria scheeri Muhlenpfordt var. valida Engelm., Proc. Amer. Acad. Arts 3: 265. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 10. 1859. LECTOTYPE: Limpio, Mexico, 1852, C. Wright s.n. Syntype: El Paso, s.d., C. Wright s.n. (2 sheets). = Coryphantha scheeri Muhlenpf- ordt. Mammillaria similis Engelm. & Gray. var. caes- pitosa Engelm. in Engelm. & Gray, Boston 1618 Annals of the Missouri Botanical Garden J. Nat. Hist. 5: 246. 1845. LECTOTYPE: cul- tivated St. Louis, from Industry, Texas, 1846, F. Lindheimer s.n. (2 sheets). = Coryphan- tha missouriensis (Sweet) Britt. & Rose var. similis (Engelm.) L. Benson. Mammillaria similis Engelm. & Gray var. ro- bustior Engelm. in Gray, Boston J. Nat. Hist. 6: 200. 1850. LECTOTYPE: cult. St. Louis from Pierdenales River, Texas, F. Lindheimer s.n. — Coryphantha missouriensis (Sweet) Britt. & Rose var. robustior (Engelm.) L. Benson. Mammillaria strobiliformis Engelm. in Wisliz., Mem. Tour. No. Mex. 115. 1848. HOLOTYPE: Rinconda, on rocks, 1847, A. Wislizenus s.n. — Neolloydia conoidea (DC.) Britt. & Rose. Mammillaria sulcata Engelm. in Engelm. & Gray, Boston J. Nat. Hist. 5: 246. 1845. LECTOTYPE: Industry, Texas, 1844, F. Lindheimer s.n. — Coryphantha sulcata (Engelm.) Britt. & Rose. Mammillaria tuberculosa Engelm., Proc. Amer. Acad. Arts 3: 268. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 14. 1859. LECTOTYPE: Flounce Mts., Chihuahua, Mexico, 1852, J. M. Bigelow s.n. (2 sheets). Syntypes: El Paso, 1851 & 1852, C. Wright & J. M. Bigelow s.n. (2 sheets). — Coryphantha strobiliformis (Poselger) Moran. Mammillaria vivipara Haw. var. radiosa Engelm. subvar. borealis Engelm., Proc. Amer. Acad. Arts 3: 269. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 15. 1859. LECTOTYPE: Santa Fe, New Mexico, 1847, A. Fendler 271 (2 sheets). Syntype: New Mexico, 1846, 4. Wislizenus s.n. = Coryphantha vivipara (Nutt.) Britt. & Rose. Mammillaria vivipara Haw. var. radiosa Engelm. subvar. neomexicana Engelm., Proc. Amer. Acad. Arts 3: 269. 1857 (preprint 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 15. 1859. LECTOTYPE: southern New Mexico, 1849, C. Wright s.n. Syntypes: San Pedro, on the Pecos, s.d., C. Wright & J. M. Bigelow s.n. (5 sheets); Sonora, 1855?, 4. Schott s.n. — Coryphantha vivipara (Nutt.) Britt. & Rose. var. radiosa (Engelm.) Backe- berg. Mammillaria vivipara Haw. var. radiosa Engelm. subvar. texana Engelm., Proc. Amer. Acad. Arts 3: 269. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 15. 1859. LECTOTYPE: Pierdenales, in sterile sandy plains in western Texas, cult. St. Louis, 1846, F. Lindheimer s.n. — Coryphantha vivipara (Nutt.) Britt. & Rose var. radiosa (Engelm.) Backeberg. Mammillaria wrightii Engelm., Proc. Amer. Acad. Arts 3: 262. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 27. 1857. LECTOTYPE: Cop- per Mines, New Mexico, 1851, C. Wright — M. barbata Engelm. Opuntia angustata Engelm. & Bigelow in En- gelm., Proc. Amer. Acad. Arts 3: 292. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 39. 1857. LECTOTYPE: Bill Williams Fork, 1854, J. M. Bigelow s.n. Syntypes: Inscrip- tion Rock, 1853, J. M. Bigelow s.n.; Cajon Pass, 1854, J. M. Bigelow s.n. = O. phaea- cantha Engelm. var. major Engelm. Opuntia arborescens Engelm. in Wisliz., Mem. Tour. No. Mex. 90. 1848. LECTOTYPE: Santa Fe, New Mexico, 1847, A. Fendler 277. Syn- type: no locale, A. Wislizenus s.n. = O. im- bricata Haw Opuntia arbuscula Engelm., Proc. Amer. Acad. Arts 3: 309. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 60. 1859. LECTOTYPE: Gilia, Arizona, s.d., A. Schott s.n. Opuntia arenaria Engelm. & Bigelow, Proc. Amer. Acad. Arts 3: 301. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 52. 1859. LECTOTYPE: Frontera, near El Paso, Texas, 1852, C. Wright 311. Opuntia basilaris Engelm. & Bigelow in Engelm., Proc. Amer. Acad. Arts 3: 298. 1857 (pre- print, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 43. 1857. LECTOTYPE: Cactus Pass, Bill Williams Fork, 1854, J. M. Bigelow s.n. Opuntia bernardina Engelm. ex Parish, Bull. Tor- rey Bot. Club 19: 92. 1892. LECTOTYPE: San Bernardino, California, 1880, G. Engelmann s.n. — O. parryi Engelm. Opuntia brachyathra Engelm. & Bigelow in En- gelm., Proc. Amer. Acad. Arts 3: 302. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 47. 1857. HOLOTYPE: Zuni, New Mexico, 1853, J. M. Bigelow s.n. = O. fragilis (Nutt.) Haw. var. brachyathra (Engelm. & Bigelow) Coult. Opuntia bulbispina Engelm., Proc. Amer. Acad. Arts 3: 304. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 55. 1859. HOLOTYPE: near Perros Bravos, Volume 75, Number 4 1988 Wolf Engelmann Type Specimens 1619 Mexico, 1848, J. Gregg probably 669. Photo only, type has not been located. Opuntia californica Engelm. in Emory, Notes Mil. Reconn. Ft. Leavenworth to San Diego, App. 2: 158. 1848. LECTOTYPE: Arizona, 18467, Emory? s.n. — O. kleiniae Englem. var. tet- racantha (Toumey) W. T. Marshall. Opuntia camanchica Engelm. & Bigelow in En- gelm., Proc. Amer. Acad. Arts 3: 293. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 40. 1857. LECTOTYPE: plains, Tucumcari, New Mexico, 1853, J. M. Bigelow s.n. — O. phaeacantha Engelm. var. camanchica (En- gelm. & Bigelow) L. Benson. Opuntia chlorotica Engelm. & Bigelow in En- gelm., Proc. Amer. Acad. Arts 3: 291. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 38. 1857. LECTOTYPE: Bill Williams Mt., Ar- izona, 1853, J. M. Bigelow s.n. Syntype: Mojave Creek, 1854, J. M. Bigelow s.n. Opuntia clavata Engelm. in Wisliz., Mem. Tour. Mex. 95. 1848. LECTOTYPE: Santa Fe, New Mexico, 1846, A. Fendler s.n. Opuntia clavellina Engelm. in Coult., Contr. U.S. Natl. Herb. 3: 444. 1896. LECTOTYPE: Pen- insula of California, 1867, W. Gabb 23. Opuntia cymochila Engelm. & Bigelow in En- gelm., Proc. Amer. Acad. Arts 3: 295. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & Route Pacific Ocean. Botany 4: 42. 1857. LECTOTYPE: Tucumcari Hills, New Mexico, 1853, J. M. Bigelow s.n. Syntype: Camanche Springs, 1853, J. M. Bigelow s.n. = O. macrorhiza Engelm. Opuntia cymochila Engelm. & Bigelow var. mon- tana Engelm. & Bigelow in Engelm., Proc. Amer. Acad. Arts 3: 296. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 42. 1857. LECTOTYPE: Sandia Mts., New Mexico, 1853, J. M. Bigelow s.n. = O. macrorhiza Engelm. Opuntia dulcis Engelm., Proc. Amer. Acad. Arts 3: 291. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 48. 1859. LECTOTYPE: El Paso, Texas, 1852, C. Wright s.n. = O. phaeacantha Engelm. Opuntia emoryi Engelm., Proc. Amer. Acad. Arts 3: 303. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 53. 1859. LECTOTYPE: seeds from Chihuahua, 1852, A. Schott s.n. — O. stanlyi Engelm. Opuntia engelmannii Salm-Dyck var. cyclodes Engelm. & Bigelow in Engelm., Proc. Amer. Acad. Arts 3: 291. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 37. 1857. LECTOTYPE: Hunah Creek near the Pecos, 1853, J. M. Bigelow s.n. = O. phaeacantha Engelm. var. major Engelm. Opuntia engelmannii Salm-Dyck var. littoralis Engelm. in Brewer & Watson, Bot. Calif. 1: 248. 1876. LECTOTYPE: Santa Barbara, Cali- fornia, 1874, O. Tittmann s.n. Syntype: Santa Cruz Island, California, 1874, O. Tittman s.n. = Q. littoralis Engelm. Opuntia erinacea Engelm. & Bigelow in Engelm., Proc. Amer. Acad. Arts 3: 301. 1857 (pre- print, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 47. 1857. HOLOTYPE: Mojave Creek, 1854, J. M. Big- elow s.n. Opuntia filipendula Engelm., Proc. Amer. Acad. Arts 3: 294. 1857. LECTOTYPE: near San Eli- zario, 1852, C. Wright s.n. = O. macrorhiza Engelm. var. pottsii (Salm-Dyck) L. Benson. Opuntia fragilis Haw. var. frutescens Engelm. in Engelm. & Gray, Boston J. Nat. Hist. 5: 245. 1845. LECTOTYPE: Colorado bottom prairie, 1844, F. Lindheimer 244. — O. leptocaulis DC Opuntia frutescens Engelm. in Engelm. & Gray, Boston J. Nat. Hist. 5: 245. 1845. LECTOTYPE: Colorado bottom prairie, 1844, F. Lindheimer 244. = O. leptocaulis DC. Opuntia frutescens Engelm. var. brevispina En- gelm., Proc. Amer. Acad. Arts 3: 309. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 53. 1857. LECTOTYPE: Texas, 1845, F. Lind- heimer s.n. = O. leptocaulis DC. Opuntia frutescens Engelm. var. longispina En- gelm., Proc. Amer. Acad. Arts 3: 309. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 53. 1857. LECTOTYPE: Laguna, Colorado, 1853, J. M. Bigelow s.n. = O. leptocaulis DC. Opuntia fulgida Engelm., Proc. Amer. Acad. Arts : 306. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 57. 1859. LECTOTYPE: western Sonora, Mex- ico, s.d., A. Schott Opuntia fusco-atra Engelm., Proc. Amer. Acad. Arts 3: 297. 1856. LECTOTYPE: prairies west of Houston, Texas, 1842, F. Lindheimer 33. Opuntia fusiformis Engelm. & Bigelow in Engelm., 1620 Annals of the Missouri Botanical Garden Proc. Amer. Acad. Arts 3: 297. 1857 (pre- print, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 43. 1857. LECTOTYPE: Deer Creek, Missouri?, 1853, J. M. Bigelow s.n. Syntype: Cow Creek, 1846, A. Wislizenus 417. = O. macrorhiza Engelm. Opuntia grahamii Engelm., Proc. Amer. Acad. Arts 3: 304. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 55. 1859. LECTOTYPE: Rio Grande, near El Paso, Texas, 1851, C. Wright s.n. = 0. schottii Engelm. var. grahamii (Engelm.) L. Benson. Opuntia grandiflora Engelm., Proc. Amer. Acad. 1857 (preprint, 1856). LECTOTYPE: cultivated St. Louis, Missouri from Industry, Texas, 1847, F. Lindheimer s.n. = O. mac- rorhiza Engelm. Opuntia hystricina Engelm. & Bigelow in En- gelm., Proc. Amer. Acad. Arts 3: 299. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 44. 1857. LECTOTYPE: Colorado Chiquito, Ar- izona, 1853, J. M. Bigelow s.n. = O. erinacea Engelm. & Bigelow var. Aystricina (Engelm. & Bigelow) L. Benson. Opuntia lindheimeri Engelm. in Gray, Boston J. Nat. Hist. 6: 207. 1850. LECTOTYPE: New Braunfels, Texas, 1845, F. Lindheimer s.n. Opuntia macrocentra Engelm., Proc. Amer. Acad. Arts 3: 292. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 49. 1859. LECTOTYPE: Rio Grande, near El Paso, Texas, 1852, C. Wright s.n. (2 sheets). = 0. violacea Engelm. var. macrocentra (En- gelm.) L. Benson. Opuntia macrorhiza Engelm. in Gray, Boston J. Nat. Hist. 6: 206. 1850. LECTOTYPE: between the Picardinales € Guadalupe, 1847, F. Lindheimer 1251. Opuntia missouriensis DC. var. albispina En- gelm. & Bigelow in Engelm., Proc. Amer. Acad. Arts 3: 300. 1857 (preprint, 1856); U.S. Senate Rept. Expl. € Surv. R.R. Route Pacific Ocean. Botany 4: 46. 1857. LECTOTYPE: Canadian River, Texas?, 1853, J. M. Bigelow s.n. = O. polyacantha Haw. Opuntia missouriensis DC. var. microsperma En- gelm., Proc. Amer. Acad. Arts 3: 300. 1857 (preprint, 1856); U.S. Senate Rept. Expl. € urv. R.R. Route Pacific Ocean. Botany 4: 46. 1857. LECTOTYPE: specimens from Ft. Pierre, South Dakota, cult. St. Louis, Missouri in 1854, no collector (2 sheets). = acantha Haw. Opuntia missouriensis DC. var. platycarpa En- O. poly- gelm., Proc. Amer. Acad. Arts 3: 300. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & R.R. Route Pacific Ocean. Botany 4: 45. 1857. LECTOTYPE: Yellowstone River, 1854, F. V. Hayden s.n. = O. polyacantha Haw. Opuntia missouriensis DC. var. rufispina Engelm. Bigelow in Engelm., Proc. Amer. Acad. Arts 3: 300. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 45. 1857. LECTOTYPE: Pe- cos, New Mexico, 1853, J. M. Bigelow s.n. = 0. polyacantha Haw. var. rufispina (En- gelm. & Bigelow) L. Benson. Opuntia missouriensis DC. var. subinermis En- gelm., Proc. Amer. Acad. Arts 3: 300. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 46. 1857. LECTOTYPE: Ft. Pierre, South Da- kota, 1853, F. V. Hayden s.n. = O. poly- acantha Haw. Opuntia missouriensis DC. var. trichophora En- gelm. & Bigelow in Engelm., Proc. Amer. Acad. Arts 3: 300. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 46. 1857. LECTOTYPE: Santa Fe Creek, New Mexico, 1853, J. M. Bigelow s.n. = O. polyacantha Haw. var. trichophora (Engelm. & Bigelow) Coult. Opuntia mojavensis Engelm., Proc. Amer. Acad. Arts 3: 293. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 40. 1857. LECTOTYPE: Mo- jave Creek, 1854, J. M. Bigelow s.n. = O. phaeacantha Haw. var. mojavensis (En- gelm.) Fosberg. Opuntia occidentalis Engelm. & Bigelow in En- gelm., Proc. Amer. Acad. Arts 3: 279. 1857 (preprint, 1856); U.S. Senate Rept. Expl & urv. R.R. Route Pacific Ocean. Botany 4: 38. 1857. LECTOTYPE: Los Angeles, California, 1852, J. M. Bigelow s.n. a de Eom California, 1855, A. Schott s.n. = O. ficu indica (L.) Miller. Opuntia parkeri Engelm. in Coult., Contr. U.S. Natl. Herb. 3: 446. 1896. HOLOTYPE: San Diego, California, 1879, C. F. Parker s.n. — O. parryi Engelm. Opuntia phaeacantha Engelm. var. brunnea En- gelm., Proc. Amer. Acad. Arts 3: 293. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 50. 1859. LECTOTYPE: Rio Grande, near El Paso, Texas, 1852, C. Wright s.n. = O. phaeacantha En- gelm. var. major Engelm. Opuntia phaeacantha Engelm. var. major En- Volume 75, Number 4 1988 Wolf Engelmann Type Specimens 1621 gelm., Proc. Amer. Acad. Arts 3: 293. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & urv. R.R. Route Pacific Ocean. Botany 4: 50. 1857. LECTOTYPE: Santa Fe, New Mexico, 1846, A. Fendler s.n. Opuntia phaeacantha Engelm. var. nigricans En- gelm. in Gray, Mem. Amer. Acad. Arts. II. 4: 52. 1849. LECTOTYPE: Santa Fe, New Mex- ico, A. Fendler 8. = O. phaeacantha Engelm. Opuntia procumbens Engelm. & Bigelow in En- gelm., Proc. Amer. Acad. Arts 3: 292. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & rv. R.R. Route Pacific Ocean. Botany 4: 39. 1857. LECTOTYPE: Aztec Pass, Arizona, 1853, J. M. Bigelow s.n. Syntypes: San Fran- cisco Mts., Arizona, 1853, J. M. Bigelow s.n.; Cactus Pass, 1854, J. M. Bigelow s.n. — O. phaeacantha Engelm. var. discata (Griffiths) L. Benson & Walkington. Opuntia prolifera Engelm., Amer. J. Sci. II. 14: 338. 1852. NEOTYPE: Mission Hills, San Di- ego, California, 1903, Leroy Abrams 3394. Opuntia pulchella Engelm., Trans. Acad. Sci. St. Louis 2: 201. 1863. LEcroTYPE: Walker Riv- er, Nevada, 1859, H. Engelmann s.n. Opuntia rafinesquei Engelm., Proc. Amer. Acad. Arts 3: 295. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 41. 1857. LECTOTYPE: St. Louis, Missouri, no collector. = O. humifusa (Raf.) Raf. Opuntia rafinesquei Engelm. var. arkansana En- gelm. ex Rumpler in Forster, Handb. Cact. 2nd edition: 922. 1885. LECTOTYPE: Fort Smith, 1853, J. M. Bigelow s.n. = O. hum- ifusa (Raf.) Raf. Opuntia rafinesquei Engelm. var. microsperma Engelm. & Bigelow in Engelm., Proc. Amer. Acad. Arts 3: 295. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 41. 1857. LECTOTYPE: cultivated St. Louis, Missouri, April 1854, no collector. = O. humifusa (Raf.) Raf. Opuntia rafinesquei Engelm. var. minor Engelm., U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany. 4: 55. 1857. LECTOTYPE: naked sandstone ledges at Mine la Motte, Mis- souri, 1845, G. Engelmann s.n. — Opuntia humifusa (Raf.) Raf. Opuntia ramosissima Engelm., Amer. J. Sci. II. 14: 339. 1852. NEOTYPE: Sonora, Mexico, 1855, A. Schott 2. Opuntia rufida Engelm., Proc. Amer. Acad. Arts 3: 298. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 51. 1859. LECTOTYPE: Presidio del Norte, Mexico, 1852, J. M. Bigelow s.n. Opuntia schottii Engelm., Proc. Amer. Acad. Arts 3: 304. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 54. 1859. LECTOTYPE: near mouth of Pecos and San Pedro, 1853, A. Schott s.n. Opuntia setispina Engelm. in Salm-Dyck, Cact. Hort. Dyck. 1849: 239. 1850. LECTOTYPE & SYNTYPE: Cosihuiriachi, 1846, A. Wislizenus s.n. = O. macrorhiza Engelm. var. pottsii (Salm-Dyck) L. Benson. Mexico, Opuntia sphaerocarpa Engelm. & Bigelow in En- gelm., Proc. Amer. Acad. Arts 3: 300. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & R. Route Pacific Ocean. Botany 4: 47. 1857. LECTOTYPE: near Albuquerque, New Mexico, 1853, J. M. Bigelow s.n. = O. poly- acantha Haw. var. junipera (Engelm. & Big- elow) L. Benson. Opuntia sphaerocarpa Engelm. var. utahensis Engelm., Trans. Acad. Sci. St. Louis 11: 199. 1863. HOLOTYPE: Steptoe Valley, Utah, 1859, H. Engelmann s.n. & Bigelow var. utahensis (Engelm. & Big- elow) L. Benson. Opuntia stenochila Engelm. & Bigelow in En- gelm., Proc. Amer. Acad. Arts 3: 296. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & R.R. Route Pacific Ocean. Botany 4: 43. 1857. LECTOTYPE: Canon de Zuni, 1853, J. M. Bigelow s.n. = O. macrorhiza Engelm. = O. erinacea Engelm. Opuntia stenopetala Engelm., Proc. Amer. Acad. Arts 3: 289. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 46. 1859. HOLOTYPE: Buena Vista, Mexico, 1848, J. Gregg 295 (2 sheets). Opuntia strigil Engelm., Proc. Amer. Acad. Arts 3: 290. 1857 loe, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 47. 1859. LECTOTYPE: 6 mi. west of the Pecos, Texas, 1851, C. Wright s.n. Opuntia tapona Engelm. in Coult., Contr. U.S. Natl. Herb. 3: 423. 1896. HOLOTYPE: Pen- insula of California, 1867, W. Gabb 20a. Opuntia tenuispina Engelm. & Bigelow in En- gelm., Proc. Amer. Acad. Arts 3: 294. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 50. 1859. LECTOTYPE: below El Paso, 1852, C. Wright 332. = O. macrorhiza Engelm. var. pottsii (Salm-Dyck) L. Benson. Opuntia thurberi Engelm., Proc. Amer. Acad. Arts 3: 308. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 1622 Annals of the Missouri Botanical Garden 59. 1859. noroTYPE: near Bacuachi, Sonora, 1851, G. Thurber s.n. Opuntia tortispina Engelm. & Bigelow in En- gelm., Proc. Amer. Acad. Arts 3: 293. 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 41. 1859. LECTOTYPE: Camanche Plains, 1853, J. M. Bigelow s.n. Opuntia vaginata Engelm. in Wisliz., Mem. Tour. No. Mex. 100. 1848. LECTOTYPE: mountains near El Paso, 1846, A. Wislizenus s.n. Syn- type: between Albuquerque and El Paso, 1846, A. Wislizenus s.n. = O. leptocaulis DC. Opuntia versicolor Engelm. in Coult, Contr. U.S. Natl. Herb. 3: 452. 1896. LECTOTYPE: mesas and foothills, near Tucson, Arizona, 1881, C. G. Pringle s.n. Syntype: same locale, 1881, C. G. Pringle 13712. Opuntia whipplei Engelm. & Bigelow in Engelm., Proc. Amer. Acad. Arts 3: 307. 1857 (pre- print, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 50. 1857. LECTOTYPE: Zuni, New Mexico, 1853, J. M. Bigelow s.n. Opuntia whipplei Engelm. var. spinosior Engelm. Bigelow in Engelm., Proc. Amer. Acad. Arts 3: 307. 1857 (preprint, 1856); U.S. Senate Rept. Expl. & Surv. R.R. Route Pacific Ocean. Botany 4: 51. 1857. LECTOTYPE: Santa Cruz River Valley, s.d., 4. Schott 5. = O. spinosior (Engelm.) Toumey. Opuntia wrightii Engelm., Proc. Amer. Acad. Arts : 1857 (preprint, 1856); in Emory, Rept. U.S. & Mex. Bound. Surv. 2: Cactaceae 59. 1859. LECTOTYPE: Presidio del Norte, Tex- as, 1851-1852, C. Wright s.n. Syntypes: Presidio del Norte, 1852, C. C. Parry s.n.; mountain sides of the Limpia, 1852, C. Wright 490. — O. kleiniae DC. CALLITRICHACEAE Callitriche antarctica Engelm. ex Hegelm., Verh. Bot. Ver. Brandenburg 9: 20. 1867. SYNTYPE: Kerguelen's Land, 1839-1843, J. D. Hooker s.n. Callitriche austinii Engelm. in Gray, Man. Bot. 5: 428. 1868. TYPE MATERIAL: New Jersey, s.d., Austin s.n. — C. terrestris Raf. Callitriche heteropoda Engelm. ex Hegelm., Verh. Bot. Ver. Brandenburg 9: 40. 1867. SYNTYPES: Bolivia, s.d., G. Mandon 1456 & 1496. Callitriche japonica Engelm. ex Hegelm., Verh. Bot. Ver. Brandenburg 10: 113. 1868. TYPE MATERIAL: Hakodadi, Japan, 1853-1856, C. Whipple s.n. CAMPANULACEAE d ce planifolia Engelm. in Coult., Bot. Gaz. FÉ: 18 SYNTYPES: Empire, Colorado, d G. Engelmann s.n. (6 sheets); Clear Creek, Colorado, 1881, G. Engelmann s.n. (2 sheets); Middle Park, Colorado, 1881, G. Engelmann s.n. Campanula scabella Engelm. in Coult., Bot. Gaz. 6: 237. 1881. LECTOTYPE: Scotts Mountain, California, 1880, G. Engelmann s.n. Lobelia mucronata Engelm. in Wisliz., Mem. Tour. No. Mex. 108. 1848. HOLOTYPE & ISOTYPE: Cosihuiriachi, Mexico, 1846, 4. Wislizenus 177. — L. fulgens Willd. Lobelia pectinata Engelm. in Wisliz., Mem. Tour. No. Mex. 108. 1848. HOLOTYPE & ISOTYPE: Cosihuiriachi, Mexico, 1846, 4. Wislizenus 192. — L. fenestralis Car. Lobelia phyllostachya Engelm. in Wisliz., Mem. Tour. No. Mex. 108. 1848. HOLOTYPE & ISOTYPE: between Monterey and Cerralbo, Mexico, 1847, A. Wislizenus 337. = L. ful- gens Willd. CAPPARACEAE Wislizenia refracta Engelm. in Wisliz., Mem. Tour. No. Mex. 99. 1848. HOLOTYPE: Rio Grande, near El Paso, Texas, 1846, C. Wright 88. CAPRIFOLIACEAE Wu spicatus Engelm. ex Gray, Bos- . Nat. Hist. 6: 215. 1850. SYNTYPE: mE 1849, F. Lindheimer 846. — S. or- biculatus Moench. CARYOPHYLLACEAE Cerastium nutans Raf. var. brachypodum En- gelm., Proc. Amer. Acad. Arts 29: 277. 1894. SYNTYPE: banks of Chouteair's Pond, s.d., G. Engelmann s.n. Drymaria nodosa Engelm. in Gray, Mem. Amer. Acad. Arts. II. 4: 12. 1849. HOLOTYPE: cult. St. Louis, Missouri from Cosihuiriachi, Mex- ico, 1848. — D. leptophylla (Cham. & Schlecht.) Fenzl ex Rohrb. var. nodosa (En- gelm.) J. Duke. Paronychia lindheimeri Engelm. in Gray, Boston J. Nat. Hist. 6: 152. 1850. TYPE MATERIAL: Volume 75, Number 4 1988 Wolf Engelmann Type Specimens 1623 Texas, 1845 and 1846, F. Lindheimer 335 (2 sheets). CLUSIACEAE Hypericum gymnanthum Engelm. & Gray, Bos- ton J. Nat. Hist. 5: 212. 1845. TYPE MATERIAL: Texas, 1844, F. Lindheimer 17. = H. mul- tilum L. CONVOLVULACEAE Cuscuta acuta Engelm., Trans. Acad. Sci. St. Louis 1: 497. 1859. HOLOTYPE: Chatham Island, Galapagos, Ecuador, 1852, Andersson s.n. Cuscuta angulata Engelm., Trans. Acad. Sci. St. Louis 1: 474. 1859. LECTOTYPE: Dutuitskloff, South Africa, Drege s.n. Cuscuta applanata Engelm., Trans. Acad. Sci. St. Louis 1: 479. 1859. LECTOTYPE: Arizona Ter- ritory, s.d., C. Wright Mexican Boundary Survey 1623-541. Cuscuta arvensis Bayrich ex Hook. var. pubescens Engelm., Trans. Acad. Sci. St. Louis 1: 495. 1859. LECTOTYPE: western Texas, 1847, F. Lindheimer s.n. = C. glabrior Engelm. var. pubescens (Engelm.) Yuncker. Cuscuta bracteata Engelm., Trans. Acad. Sci. St. Louis 1: 509. 1859. ISOTYPE: Goyaz, Brazil, s.d., Gardner 3348. Cuscuta californica Choisy var. apiculata En- gelm., Trans. Acad. Sci. St. Louis 1: 499. 1859. HOLOTYPE: Colorado River, California, 1854, J. M. Bigelow s.n. Cuscuta californica Choisy var. squamigera En- gelm., Trans. Acad. Sci. St. Louis 1: 499. 1859. HOLOTYPE: Rio Virgen, Utah, s.d., Remy s.n. Cuscuta cephalanthi Engelm., Amer. J. Sci. 43: 336. 1842. LECTOTYPE: near St. Louis, Mis- souri, s.d., G. Engelmann s.n. Cuscuta chinensis Lam. var. ciliaris Engelm., Trans. Acad. Sci. St. Louis 1: 480. 1859. LECTOTYPE: Mosul, Kurdistan, s.d., Kotschy 431 (3 sheets). Cuscuta corniculata Engelm. var. sphaerocyma Engelm., Trans. Acad. Sci. St. Louis 1: 504. 1859. LECTOTYPE: Goyaz Province, s.d., Bra- zil, Weddell s.n. Syntype: Rio Meta, s.d., Kar- sten s.n. Cuscuta coryli Engelm., Amer. J. Sci. 43: 337. 1842. LECTOTYPE: American Bottom, St. Louis, Missouri, 1841, G. Engelmann s.n. Cuscuta corymbosa Choisy var. grandiflora En- gelm., Trans. Acad. Sci. St. Louis 1: 483. 1859. LECTOTYPE: Popayán, Columbia, s.d., Humboldt s.n. Cuscuta corymbosa Choisy var. stylosa Engelm., Trans. Acad. Sci. St. Louis 1: 484. 1859. ISOLECTOTYPE: Mexico, 1846, Berlandier 822. Cuscuta cristata Engelm., Trans. Acad. Sci. St. Louis 1: 507. 1859. ISOTYPE: Argentina, s.d., Tweedy 1191. Cuscuta cuspidata Engelm. & Gray var. pratensis Engelm. in Engelm. & Gray, Boston J. Nat. Hist. 5: 224. 1845. LECTOTYPE: west of the Brazos, Texas, 1843, F. Lindheimer 125. Cuscuta decora Choisy ex Engelm. var. integrius- cula Engelm., Trans. Acad. Sci. St. Louis 1: 502. 1859. HOLOTYPE: Mendoza, Argentina, Gilles s.n. = C. indecora Choisy var. inte- griuscula (Engelm.) Yuncker. Cuscuta denticulata Engelm., Amer. Naturalist 9: 348. 1875. HOLOTYPE: St. George, Utah, C. C. Parry 205. Cuscuta epithymum Murr. var. angustata En- gelm. subvar. angustissima Engelm., Trans. Acad. Sci. St. Louis 1: 463. 1859. HOLOTYPE: Padua, Italy, s.d., Visiani s.n. = C. epithy- mum Murr. var. angustissima Engelm. Cuscuta epithymum Murr. var. kotschyi Engelm. subvar. scabrella Engelm., Trans. Acad. Sci. St. Louis 1: 464. 1859. LECTOTYPE: Sicily, s.d., Gussone s.n. = C. epithymum Murr. var. scabrella (Engelm.) Yuncker. Cuscuta epithymum Murr. var. obtusata Engelm. subvar. apoda Engelm., Trans. Acad. Sci. St. Louis 1: 462. 1859. HOLOTYPE: Koniah, Asia Minor, 1845, Heldreich s.n. = C. obtusata (Engelm.) Traub. Cuscuta epithymum Murr. var. obtusata Engelm., subvar. macropoda Engelm., Trans. Acad. Sci. St. Louis 1: 462. 1859. HOLOTYPE: Sierra Nevada, Spain, s.d., Funk s.n. = C. trium- virati Lange. Cuscuta epithymum Murr. var. sagittanthera En- gelm., Trans. Acad. Sci. St. Louis 1: 462. 1859. HOLOTYPE: Tunis, s.d., Kralik s.n. Cuscuta exaltata Engelm., Trans. Acad. Sci. St. Louis 1: 513. 1859. LECTOTYPE: New Braun- fels, Texas, s.d., F. Lindheimer 472 (3 sheets). Cuscuta globiflora Engelm., Trans. Acad. Sci. St. Louis 1: 520. 1859. ISOTYPE: Cuzo, Brazil, s.d., Pentland s.n. Cuscuta gracillima Engelm., Trans. Acad. Sci. St. Louis 1: 488. 1859. ISOLECTOTYPE: Mexico, Cuscuta gracillima Engelm. var. saccharata En- 1624 Annals of th Missouri Eos Garden gelm., Trans. Acad. Sci. St. Louis 1: 489. 1859. ISOLECTOTYPE: Oaxaca, Mexico, s.d., Liebman s.n. = C. saccharata (Engelm.) Yuncker. Cuscuta gronovii Willd. var. calyptrata Engelm., Trans. Acad. Sci. St. Louis 1: 508. 1859. LECTOTYPE: western Louisiana, s.d., J. Gregg s.n. Cuscuta gronovii Willd. var. curta Engelm., Trans. Acad. Sci. St. Louis 1: 508. 1859. ISOLECTOTYPE: northwest America, s.d., Douglas s.n. = C. curta (Engelm.) Rydb. Cuscuta japonica Choisy var. fissistyla Engelm., Trans. Acad. Sci. St. Louis 1: 517. 1859. HOLOTYPE: Hong Kong, China, s.d., Wright Cuscuta japonica Choisy var. paniculata En- gelm., Trans. Acad. Sci. St. Louis 1: 517. 1859. LECTOTYPE: Pekin (sic), China, s.d., Ki- rilow s.n. Cuscuta japonica Choisy var. thyrsoidea En- gelm., Trans. Acad. Sci. St. Louis 1: 517. 1859. ISOTYPE: Japan, s.d., Zollinger 355. Cuscuta kurdica Engelm., Trans. Acad. Sci. St. Louis 1: 470. 1859. HOLOTYPE?: Gara Mts., Kurdistan, Kotschy 388b. Cuscuta lehmanniana Bunge var. esquamata En- gelm., Trans. Acad. Sci. St. Louis 1: 515. 1859. HOLOTYPE: Mont Sipyle, Persia, s.d., Balansa 411. Cuscuta leptantha Engelm., Trans. Acad. Sci. St. Louis 1: 489. 1859. LECTOTYPE: western Tex- as to El Paso, Texas, 1849, C. Wright 522. Cuscuta lupuliformis Krock var. asiatica En- gelm., Trans. Acad. Sci. St. Louis 1: 516. 1859. LECTOTYPE: Volga River, Russia, s.d., Fischer s.n. Cuscuta micrantha Choisy var. latiflora Engelm., ans. Acad. Sci. St. Louis 1: 501. 1859. LECTOTYPE: Concon, Chile, s.d., Poeppig 89. Syntype: 1827, Poeppig 159. Cuscuta mitraeformis Engelm. in Hemsley, Diag. Pl. Nov. 54. 1880. HOLOTYPE: San Luis Potosi to Tampico, Mexico, 1878-1879, E. Palmer s.n. Cuscuta neuropetala Engelm., Amer. J. Sci. 45: 75. 1843. LECTOTYPE: near Houston, Texas, 1843, F. Lindheimer 124. — C. indecora Choisy var. neuropetala (Engelm.) Hitchcock. Cuscuta obtusiflora H.B.K. var. australis En- gelm., Trans. Acad. Sci. St. Louis 1: 492. 1859. LECTOTYPE: New Holland, Pi s.d., own s.n. — C. (usanta obtusiflora H.B.K. var. cesatiana En- australis R. gelm., Trans. Acad. Sci. St. Louis 1: 493. 1859. LECTOTYPE: Vercelli, Italy, s.d., Cesati 83?. — C. australis R. Br. var. cesatiana (Bertoloni) Yuncker. Cuscuta obtusiflora H.B.K. var. cordofana En- gelm., Trans. Acad. Sci. St. Louis 1: 493. 1859. HOLOTYPE: Kordofan, Africa, 1844, Fi- gari s.n. = C. cordofana (Engelm.) Yuncker. Cuscuta obtusiflora H.B.K. var. glandulosa En- gelm., Trans. Acad. Sci. St. Louis 1: 492. 1859. LECTOTYPE: Columbus, Georgia, 1838, Boykin s.n Cuscuta obtusiflora H.B.K. var. vera Engelm., Trans. Acad. Sci. St. Louis 1: 493. 1859. LECTOTYPE: Andes, Peru, s.d., Humboldt s.n. — C. obtusiflora H.B.K. Cuscuta odontolepis Engelm., Trans. Acad. Sci. St. Louis 1: 486. 1859. HOLOTYPE: Arizona, 1851-1852, C. Wright 1642. Cuscuta parviflora Engelm., Trans. Acad. Sci. St. Louis 1: 506. 1859. LECTOTYPE: Minas Ge- raes, Brazil, s.d., Ackerman s.n. Syntype: Vil- la Rica, Brazil, s.d., Pohl 5726. Cuscuta parviflora Engelm. var. elongata En- gelm., Trans. Acad. Sci. St. Louis 1: 506. 1859. HoLoTYPE: Goyaz, Brazil, s.d., Weddell 2125. Cuscuta pentagona Engelm., Amer. J. Sci. 43: 340. 1842. HOLOTYPE: Norfolk, Virginia, 1841, Rugel s.n. Cuscuta pentagona Engelm. var. calycina En- gelm., Amer. J. Sci. 45: 76. 1845. SYNTYPE: Texas, F. Lindheimer 126. = C. campestris Yuncker. Cuscuta planiflora Tenore var. tenorii Engelm., Trans. Acad. Sci. St. Louis 1: 466. 1859. LECTOTYPE: Naples, Italy, s.d., Tenore s.n. — C. planiflora Tenore. Cuscuta planiflora Tenore var. papillosa En- gelm., Trans. Acad. Sci. St. Louis 1: 467. 1859. LECTOTYPE: Djebel Zaghouan, Tunis, 1854, Kralik 410a Cuscuta polygonorum Engelm., Amer. J. Sci. 43: 342. LECTOTYPE: west of St. Louis, Missouri, s.d., G. Engelmann s.n. Syntype: west of St. Louis, Missouri, F. Lindheimer s.n. Cuscuta racemosa Mart. var. brasiliana Engelm., Trans. Acad. Sci. St. Louis 1: 505. 1859, SYNTYPE: Rio de Janeiro, Brazil, 1817, Mar- Lius s.n. Cuscuta racemosa Mart. var. miniata Engelm., Trans. Acad. Sci. St. Louis 1: 505. 1859. SYNTYPE: Brazil, 1857, Martius 1292. Cuscuta racemosa Mart. var. nuda Engelm., Trans. Volume 75, Number 4 1988 Wolf 1625 Engelmann Type Specimens Acad. Sci. St. Louis 1: 505. 1859. SYNTYPE: near Rio, Brazil, s.d., Sellow s.n. Cuscuta reflexa Roxb. var. brachystigma En- gelm., Trans. Acad. Sci. St. Louis 1: 519. 1859. LECTOTYPE: Calcutta, India, s.d., Gau- dichaud 129. — C. reflexa Wallich var. an- guina (Edgeworth) Yuncker. Cuscuta saururi Engelm., Amer. J. Sci. 43: 339. 1842. LECTOTYPE: St. Louis, Missouri, 1841, C. A. Geyer s.n. — saururi (Engelm.) MacMillan. Cuscuta squamata Engelm., Trans. Acad. Sci. St. Louis 1: 510. 1859. LECTOTYPE: western Tex- as to El Paso, Texas, 1849, C. Wright 518. Cuscuta stenolepis Engelm., Trans. Acad. Sci. St. Louis 1: 500. 1859. HOLOTYPE: Mare Island, San Francisco, California, 1855, C. Wright C. gronovii Willd. var. s.n. Cuscuta trichostyla Engelm., Trans. Acad. Sci. St. Louis 1: 495. 1859. ISOTYPE: Parana, s.d., Tweedie s.n. Syntype: Santatem, Brazil, s.d., Spruce 854. Cuscuta umbellata H.B.K. var. desertorum En- gelm., Trans. Acad. Sci. St. Louis 1: 488. 1859. LECTOTYPE: Piauhy Province, Brazil, 1847, Martius s.n. Cuscuta verrucosa Engelm. var. glabrior En- gelm., Amer. J. Sci. 43: 341. 1842. LECTOTYPE: west of Houston, Texas, s.d., F. Lindheimer s.n. Syntype: Texas, 1833, Drummond 247. — C. glabrior (Engelm.) Yuncker. pidanke compositarum Engelm., Amer. J. Sci. 44. 1842. LECTOTYPE: American Bottom, St. Louis, Missouri, 1841, G. Engelmann s.n. CUPRESSACEAE Juniperus occidentalis Hook. var. monosperma Engelm., Trans. Acad. Sci. St. Louis 3: 590. 1877. SYNTYPES: Cañon City & Manitou, Col- orado, 1874, G. Engelmann s.n. (5 sheets). CYPERACEAE Cyperus refractus Engelm. in Boeckl., Linnaea 36: 69. 1870. HOLOTYPE: Merimac, s.d., N. Riel Herbarium 496. EPHEDRACEAE Ephedra aspera Engelm. ex Watson, Proc. Amer. Acad. Arts 18: 157. 1883. TYPE MATERIAL: Sierra Madre, west of Saltillo, Mexico, 1880, E. Palmer 1288. Ephedra pedunculata Engelm. ex Watson, Proc. Amer. Acad. Arts 18: 157. 1883. TYPE MA- TERIAL: Nvalde, west of San Antonio, Texas, 1880, E. Palmer 1291. EQUISETACEAE Equisetum laevigatum A. Braun var. elatum En- gelm. in A. Braun, Amer. J. Sci. 46: 87. 1843. TYPE MATERIAL: banks of Mississippi, below Jefferson Barracks, St. Louis, Missouri, 1843, G. Engelmann s.n. (7 sheets). Equisetum laevigatum A. Braun var. scabrellum Engelm. in A. Braun, Amer. J. Sci. 46: 87. 1843. LECTOTYPE: Jefferson Barracks, St. Louis, Missouri, 1843, G. Engelmann s.n. Equisetum robustum A. Braun var. affine Engelm. in A. Braun, Amer. J. Sci. 46: 88. 1843. LECTOTYPE: below Jefferson Barracks, St. Louis, Missouri, 1843, G. Engelmann s.n. = E. hye- male L. var. affine (Engelm.) A. A. Eaton. Equisetum robustum A. Braun var. minus Engelm. in A. Braun, Amer. J. Sci. 46: 88. 1843. LECTOTYPE: banks of Mississippi, below St. Louis, Missouri, 1843, G. Engelmann s.n. = E. hyemale L. var. affine (Engelm.) A. A. Eaton. EUPHORBIACEAE Aphora humilis Engelm. & Gray, Boston J. Nat. Hist. 5: 54. 1845. ISOTYPE: west of Brazos, Texas, 1844, F. Lindheimer 306. = Argy- thamnia humilis (Engelm. & Gray) Muell.- Arg. Croton corymbulosus Engelm. in Rothr., Bot. Wheeler's Surv. 242. 1878. SYNTYPES: Camp Bowie, New Mexico, 1874, J. T. Rothrock 506; Buena Vista, Mexico, 1848, J. Gregg 71 & 288; west Texas, 1849, C. Wright 641, and 1851-1852, C. Wright 1805. — C. pottsii (Klotzsch) Muell.-Arg. Croton fruticulosus Engelm. in Torr., Bot. Mex. Bound. Surv. 194. 1859. SYNTYPES: Texas, 1849, F. Lindheimer 134 (2 sheets), 176, 177, 297a. Published as C. fruticulosum. Euphorbia acuta Engelm. in Torr., Bot. Mex. Bound. Surv. 189. 1859. LECTOTYPE: “N. Mex.," 1851, C. Wright 1839. Euphorbia angusta Engelm. in Torr., Bot. Mex. Bound. Surv. 189. 1859. LECTOTYPE: San Pe- dro River, Texas, 1851, C. Wright 1828. Euphorbia arizonica Engelm. in Torr., Bot. Mex. Bound. Surv. 186. 1859. LECTOTYPE: Arizona, 1856, A. Schott s.n. Euphorbia arkansas Engelm. & Gray, Boston J. Nat. Hist. 5: 261. 1845. SYNTYPES: near 1626 Annals of the Missouri Botanical Garden Houston, Texas, 1843 and 1845, F. Lind- heimer s.n. (2 sheets). Euphorbia astyla Engelm. ex Boiss. in DC., Prod. 15(2) 40. 1862. HOLOTYPE: Nazas River, Mexico, 1847, J. Gregg 457. = Chamaesyce albomarginata (Torr. & Gray) Small. Euphorbia barbellata Engelm. in Torr., Bot. Mex. Bound. Surv. 190. 1859. LECTOTYPE: Rio Grande, 1848, C. Wright s.n. Euphorbia baueri Engelm. ex Boiss. in DC., Prod. 15(2): 27. 1862. TYPE MATERIAL: “Nov. Hol- land," s.d., Bauer s.n. Euphorbia bicolor var. bicolor Engelm. & Gray, Boston J. Nat. Hist. 5: 233. 1845. SYNTYPE: west of Houston, Texas, 1841, F. Lindheimer 174 Euphorbia bicolor Engelm. & Gray var. concolor Engelm. & Gray, Boston J. Nat. Hist. 5: 233. 1845. TYPE MATERIAL: Lynchburg, Texas, 1842, F. Lindheimer s.n. Fuphorbia bifurcata Engelm. in Torr., Bot. Mex. Bound. Surv. 190. 1859. HOLOTYPE OR ISOTYPE: valley of the Limpio, 1852, J. M. Bigelow s.n. Euphorbia bilobata Engelm. in Torr., Bot. Mex. Bound. Surv. 190. 1859. LECTOTYPE: Santa Cruz, Sonora, Mexico, 1851-1852, C. Wright 1831 Euphorbia blodgettii Engelm. ex Hitchcock, An- Rep. Missouri Bot. Gard. 4: 126. 1893. LECTOTYPE: Key West, Florida, s.d., Blodgett .n. = Chamaesye blodgettii (Engelm. ex Hitchcock) Small. Euphorbia brachycera Engelm. in Torr., Bot. Mex. Bound. Surv. 192. 1859. HOLOTYPE: Donana, above El Paso, Texas, 1851, C. Wright 1821. Euphorbia capitellata Engelm. in Torr., Bot. Mex. Bound. Surv. 188. 1859. HOLOTYPE: San Ber- nardino, Arizona, 1851, C. Wright 1849. Euphorbia cinerascens Engelm. in Torr., Bot. Mex. Bound. Surv. 186. 1859. LECTOTYPE: Mon- terrey, Mexico, 1847, J. Gregg 215. Syn- type: western Texas to El Paso, 1849. C. "right 559. Euphorbia cinerascens Engelm. var. appendi- culata Engelm. in Torr., Bot. Mex. Bound. Surv. 186. 1859. LECTOTYPE: San Felipe, Cal- ifornia, 1852, G. Thurber 628. = E. melan- adenia Torr. Euphorbia commutata Engelm. in Gray, Man. Bot. 2: 389. 1856. TYPE MATERIAL: “Gasco- nade," May 1835, number 35, illegible but "commutata" added to label by Engelm. Euphorbia crenulata Engelm. in Torr., Bot. Mex. Bound. Surv. 192. 1859. TYPE MATERIAL: no locale, s.d., Hartweg 1950. curtisii Engelm. in Chapman, Fl. S. .S. 401. 1860. LECTOTYPE: North Carolina, s.d., s s.n. Euphorbia deltoidea Engelm. in Chapman, Fl. S. U.S. 2nd edition. 647. 1883. LECTOTYPE: Flor- ida, 1880, Curtiss 162. Euphorbia dentata Michx. var. cuphosperma En- gelm. in Torr., Bot. Mex. Bound. Surv. 190. 1859. LECTOTYPE: Copper Mines, New Mex- ico, 1851-1852, C. Wright 1834. Euphorbia dentata Michx. var. rigida Engelm. in Torr., Bot. Mex. Bound. Surv. 190. 1859. LECTOTYPE: San Pedro River, New Mexico, 1851-1852, C. Wright 1837. Euphorbia dictyosperma Fisch. & Mey. var. leio- Bot. Mex. Bound. 1835, cocca Engelm. in Torr., Surv. 191. 1859. LECTOTYPE: Texas, Drummond 327. Fuphorbia dictyosperma Fisch. & Mey. var. mex- icana Engelm. in Torr., Bot. Mex. Bound. Surv. 191. 1859. LECTOTYPE: Balson de Ma- pimi, 1847, J. Gregg 456. = E. mexicana (Norton) Engelm. Euphorbia dioeca H.B.K. var. indivisa Engelm. in Torr., Bot. Mex. Bound. Surv. 187. 1859. ISOTYPE: near the Copper Mines, New Mexico, 1851, C. Wright 1845. — E. indivisa (En- gelm.) Tidstrom. Euphorbia exstipulata Engelm. in Torr., Bot. Mex. Bound. Surv. 189. 1859. LECTOTYPE: Santa Fe, New Mexico, 1847, A. Fendler 790. Syn- types: eastern Sonora, C. Wright 1833 & 1838 Euphorbia floribunda Engelm. ex Boiss. in DC., Prod. 15(2): 39. 1862. LECTOTYPE: east of Guadalahara, Mexico, 1849, J. Gregg 856a. Euphorbia florida Engelm. in Torr., Bot. Mex. Bound. Surv. 189. 1859. LECTOTYPE: west of Chiricahua Mountains, Arizona, 1851, Wright 1829. Euphorbia fruticulosa Englem. ex Boiss. in DC., Proc. 15(2): 38. 1862. ISOTYPE: Coahuila, Mexico, s.d., J. Gregg 500. Fuphorbia garberi Engelm. ex Chapman, Fl. S. U.S. 2nd edition. 646. 1883. LECTOTYPE: coast of Florida, s.d., Garber s.n. Euphorbia geyeri Engelm. in Engelm. & Gray, Boston J. Nat. Hist. 5: 260. 1845. LECTOTYPE: Beardstown, Illinois, 1842, C. A. Geyer s.n. Euphorbia glyptosperma Engelm. in Torr., Bot. Mex. Bound. Surv. 187. 1859. LECTOTYPE: Fort Kearney, Nebraska, 1856, H. Engel. n s.n. Euphorbia glyptosperma Engelm. var. tenerrima ngelm. in Torr., Bot. Mex. Bound. Surv. 187. 1859. LECTOTYPE: Nueces River, Texas, Volume 75, Number 4 1988 Wolf 1627 Engelmann Type Specimens 1851, C. Wright 1853. = E. glyptosperma Engelm. Euphorbia grisea Engelm. ex Boiss. in DC., Prod. 15(2): 41. 1862. LECTOTYPE: Cerralro, Nuevo Leon, Mexico, 1847, J. Gregg 837. Euphorbia hirtula Engelm. in Watson, Bot. Calif. 2: 74. 1880. LECTOTYPE: southern California, 1880, E. Palmer 140 Euphorbia humistrata Engelm. ex Gray, Man. Bot. 2: 386. 1856. LECTOTYPE: St. Louis, Mis- souri, 1833, G. Engelmann 1139. Euphorbia lata Engelm. in Torr., Bot. Mex. Bound. Surv. 189. 1859. LECTOTYPE: western Texas, 1851, C. Wright 1841. Syntype: the Cimar- ron to Sand Creek 1847, 4. Fendler s.n. Euphorbia leptocera Engelm. ex Torr., Pacif. Rail. Rep. 4: 135. 1857. TYPE MATERIAL: Grass Valley, California, s.d., J. M. Bigelow s.n. Euphorbia lurida Engelm. in Ives, Rep. Colo. Riv- 26. 1861. LECTOTYPE: Bill Williams Mt., Arizona, 1858, Newberry s.n. Euphorbia marginata Pursh var. uloleuca En- gelm. & Gray, Boston J. Nat. Hist 5: 261. 1845. LECTOTYPE: prairies in the Colorado bot- tom, 1844, F. Lindheimer 303. Euphorbia montana Engelm. var. gracilor En- gelm. in Torr., Bot. Mex. Bound. Surv. 192. 1859. LECTOTYPE: Santa Fe Creek, New Mex- ico, 1847, A. Fendler 786. Euphorbia multicaulis Engelm. in Torr., Bot. Mex. Bound. Surv. 191. 1859. HOLOTYPE: Las Play- as, Sonora, 1851, G. Thurber 381. Euphorbia palmeri Engelm. in Watson, Bot. Calif. 2: 75. 1880. LECTOTYPE: at Tally’s, California, 1875, E. Palmer s.n. Euphorbia parryi Engelm., Amer. Naturalist 9: 350. 1875. LECTOTYPE: St. George, Utah, 1874, C. C. Parry 247. Euphorbia pediculifera Engelm. in Torr., Bot. ex. Bound. Surv. 186. 1859. LECTOTYPE: Sonoita Creek, Arizona, 1851, C. Wright 1848. Euphorbia peplidion Engelm. in Torr., Bot. Mex. Bound. Surv. 191. 1859. HOLOTYPE: San Pe- dro, Texas, 1851, C. Wright 1823. Euphorbia petaloidea Engelm. in Torr., Bot. Mex. Bound. Surv. 185. 1859. LECTOTYPE: Platte River, Nebraska Territory, 1858, H. Engel- mann s.n. = E. missurica Raf. var. inter- media (Engelm.) L. C. Wheeler. Euphorbia petaloidea Engelm. var. Sonic aqa Engelm. in Torr., Bot. Mex. nd rv. 185. 1859. HOLOTYPE: near hein S iil 1851, C. Wright 1826. — E. parryi Engelm. Euphorbia petaloidea Engelm. var. intermedia Bot. Mex. Bound. Surv. Engelm. in Torr., 185. 1859. LECTOTYPE: Ft. Pierre, South Da- kota, 1853, F. V. Hayden s.n. = E. missurica Raf. var. intermedia (Engelm.) L. C. Wheel- er. Euphorbia petaloidea Engelm. var. nicollettii En- gelm. in Torr., Bot. Mex. Bound. Surv. 185. 1859. LECTOTYPE: Powder River, Montana, 1854. F. V. Hayden s.n. = E. missurica Raf. var. intermedia (Engelm.) L. C. Wheeler. Euphorbia pilulifera L. var. discolor Engelm. in Torr., Bot. Mex. Bound. Surv. 188. 1859. LECTOTYPE: Sonoita Creek, Arizona, 1842, C. Wright s.n. — E. hirta L. Euphorbia platysperma Engelm. in Watson, Bot. Calif. 2: 482. 1880. LECTOTYPE: mouth of Colorado River, 1869, E. Palmer s.n. Euphorbia pringlei Engelm. in Patterson, Check list 115. 1887. LECTOTYPE: Canyon of Santa Rita, Arizona, 1881. C. G. Pringle 138. Euphorbia pycnanthema Engelm. in Torr., Bot. Mex. Bound. Surv. 188. 1859. HOLOTYPE: Lake Santa Maria, Chihuahua, Mexico, 1852, C. Wright 186. = E. capitellata Engelm. Euphorbia revoluta Engelm. in Torr., Bot. Mex. Bound. Surv. 186. 1859. LECTOTYPE: between Santa Fe and Moro River, New Mexico, 1847, A. Fendler 789. Euphorbia schizoloba Engelm. in Ives, Rep. Col- orado River Bot. 4: 27. 1861. HOLOTYPE: west- ern Arizona, 1858, Newberry s.n. Euphorbia serrula Engelm. in Torr., Bot. Mex. Bound. Surv. 188. 1859. LECTOTYPE: Gua- dalupe Pass, Sonora, Mexico, 1851, C. Wright 1843. Euphorbia setiloba Engelm. in Torr., Pacif. Rail. Rep. 5: 364. 1857. ISOTYPE: Fort Yuma, Cal- ifornia, G. H. Thomas s.n. Euphorbia stictospora Engelm. in Torr., Bot. Mex. Bound. Surv. 187. 1859. LECTOTYPE: Pawnee River, Kansas, 1847, A. Fendler 798 Euphorbia subpubens Engelm. in Watson, Bot. Calif. 2: 76. 1880. ISOTYPE & FRAGMENT OF HOLOTYPE FROM GH: Prescott, Arizona, 1876, E. Palmer 512. Euphorbia tetrapora Engelm. in Torr., Bot. Mex. Bound. Surv. 191. 1859. SYNTYPE: west of Brazos, Texas, 1839, F. Lindheimer s.n. Euphorbia tomentella Engelm. ex Boiss. in DC., Prod. 15(2): 32. 1862. LECTOTYPE: San Luis Potosi, Mexico, 1827, Berlandier 1358. Euphorbia trachysperma Engelm. in Torr., Bot. ound. Surv. 189. 1859. HOLOTYPE: San Pedro River, Arizona, 1851, C. Wright 1832. Euphorbia umbellata Engelm. ex Boiss. in DC., 1628 Annals of the Missouri Botanical Garden Prod. 15(2): 127. 1862. LECTOTYPE: Guada- lahara, Mexico, 1849, J. Gregg 850. Euphorbia villifera Scheele var. nuda Engelm. ex Boiss. in DC., Prod. 15(2): 45. 1862. LECTOTYPE: New Braunfels, Texas, 1850, C. Wright s.n. Euphorbia xantii Engelm. ex Boiss. in DC., Prod. 15(2): 62. 1862. LECTOTYPE: Cape St. Lucas, California, 1859-1860, L. T. Xantus s.n. Jatropha macrorhiza Benth. var. septemfida En- gelm. in Rothr., Bot. Wheeler's Surv. 244. 1878. SYNTYPE: Sulphur Springs, Arizona, 1874, J. T. Rothrock 546. Pilinophytum lindheimeri Engelm. & Gray, Bos- . Nat. Hist. 5: 232. 1845. SYNTYPE: Texas, 1843, F. Lindheimer 171. Stillingia angustifolia Engelm. ex Watson, Proc. Amer. Acad. Arts 18: 154. 1883. HOLOTYPE: Comfort, 1879, E. Palmer 1256. — S. texana I. M. Johnston. Tetracoccus dioicus Engelm. ex Parry, West Am. Sci. 1: 13. 1885. PARATYPE: Table Mountain, Northern Lower California, 1883, C. C. Parry & H. C. Orcutt 313 Tragia brevispica Engelm. & Gray, Boston J. Nat. Hist. 5: 226. 1845. SYNTYPE: Texas, 1848, F. Lindheimer 307. Texas, FABACEAE Amorpha canescens Nutt. var. leptostachya En- gelm. in Gray, Mem. Amer. Acad. Arts. II. . 1849. HOLOTYPE?: in woods between San Miguel and Vegas, 1847, A. Fendler 125. Amorpha incana Engelm., apparently not pub- lished. HOLOTYPE?: Texas, 1842, F. Lind- heimer 153, annotated “n. sp.” by Engelm. Astragalus lindheimeri Engelm. in Gray, Pl. Wrightiana 1: 52. 1852. LECTOTYPE: Texas, s.d., F. Lindheimer 746. Baptisia sulphurea Engelm. Bot. Gaz. 3: 65. 1878. HOLOTYPE: Oklahoma, 1877, Butler s.n. (2 sheets). = B. lactea (Raf.) Thieret x B. sphaerocarpa Nutt. Calliandra chamaedrys Engelm. ex Gray, Mem. Amer. Acad. Arts. II. 4: 39. 1849. SYNTYPES: Chihuahua, Mexico, 1847, J. Gregg 529, A. Wislizenus 251. Calliandra herbacea Engelm. ex Gray, Mem. Amer. Acad. Arts. IT. 4: 39. 1849. LECTOTYPE: New Mexico, s.d., A. Fendler 180. Cercis reniformis Engelm. ex Gray, Boston J. Nat. Hist. 6: 177. 1850. LECTOTYPE: deci Texas, 1845, F. Lindheimer 366. cidentalis Torr. Desmodium wislizenii Engelm. ex Gray, Pl. Wrightiana 1: 53. 1852. HOLOTYPE: Cosihui- riachi, Mexico, 1846, 4. Wislizenus 183. = D. retinens Schlecht. Eysenhardtia spinosa Engelm. ex Gray, Boston t. Hist. 6: 174. 1850. ISOTYPE: Chihua- hua, Mexico, 1846, 4. Wislizenus 133. Lespedeza leptostachya Engelm. in Gray, Proc. Amer. Acad. Arts 12: 57. 1877. LECTOTYPE: Iowa, 1871, C. E. Bessey s.n. FAGACEAE Quercus chrysolepis Liebm. var. palmeri En- gelm., Trans. Acad. Sci. St. Louis 3: 393. 1877. HOLOTYPE: 80 mi. E of San Diego, Cal- ifornia, 1875, E. Palmer 30. — Q. palmeri Engelm. Quercus dumosa Nutt. var. bullata Engelm., Trans. Acad. Sci. St. Louis 3: 393. 1877. SYNTYPES: Santa Lucia Mts., California, s.d., Brewer 473; New Idria, California, s.d., Brewer 776; Pope Valley, California, 1863, H. Bolander s.n. — Q. durata Jeps. Quercus falcata Michx. var. subintegra Engelm., Trans. Acad. Sci. St. Louis 3: 543. 1877. LECTOTYPE: South Carolina, 1875, Melli- champ s.n. = Q. subintegra Trel. Quercus hypoleuca Engelm., Trans. Acad. Sci. St. Louis 3: 384. 1876. LECTOTYPE: Copper Mines, and Arizona, 1851-1852, C. Wright 1869. Quercus lobata Nee var. fruticosa Engelm., Trans. Acad. Sci. St. Louis 3: 389. 1877. SYNTYPES: west of Shasta, California, 1862, H. Brewer 1336; near Tuolumne River, California, Lem- s.n. = Q. oerstediana R. Br. Quercus muehlenbergii Engelm., Trans. Acad. Sci. St. Louis 3: 391. 1877. TYPE MATERIAL: St. Louis, Missouri, Sept. 1838, G. Engelmann s.n. Annotated Q. muhlenbergii by Engel- mann, but no annotation date. The only pre- publication dated specimen Quercus tomentella Engelm., Trans. Acad. Sci. St. Louis 3: 393. 1877. HOLOTYPE: Guadalupe Island, California, E. Palmer 89. Quercus undulata Torr. var. jamesii Engelm., Trans. Acad. Sci. St.Louis 3: 382. 1876. TYPE MATERIAL: fragment of “‘original species ex Hb Torrey," Rocky Mountains, s.d., James s.n. FOUQUIERIACEAE Fouquieria splendens Engelm. in Wisliz., Mem. ur. No. Mex. 98. 1848. LECTOTYPE: south Volume 75, Number 4 1988 Wolf Engelmann Type Specimens 1629 of Chihuahua, Mexico, 1847, A. Wislizenus FUMARIACEAE Corydalis curvisiliqua Engelm. ex Gray., Pl. Wrightiana 2: 10. 1853. TYPE MATERIAL: Tex- as, Apr. 1848, F. Lindheimer s.n.; May 1851, F. Lindheimer s.n. GENTIANACEAE Erythraea nudicaulis Engelm. in Gray, Proc. Amer. Acad. Arts. 17: 222. 1882. HOLOTYPE: base of Santa Catalina Mountains, Arizona, C. G. Pringle s.n. Geniostemon coulteri Engelm. & Gray, Proc. Amer. Acad. Arts 16: 104. 1881. TYPE Ma- TERIAL: Mexico, s.d., Coulter, s.n., appears to be fragment of holotype from Gray’s herbar- Geniostemon schafferi Engelm. & Gray, Proc. Amer. Acad. Arts 16: 104. 1881. TYPE MA- TERIAL: San Luis Potosi, Mexico, 1879, Schaf- fer 80. Gentiana acuta Michx. var. nana Engelm., Trans. Acad. Sci. St. Louis 2: 214. 1862. HOLOTYPE: Colorado Territory, 1861, C. C. Parry 309. = Gentianella amarella (L.) Borner subsp. acuta (Michx.) J. M. Gillett. Gentiana barbellata Engelm., Trans. Acad. Sci. St. Louis 2: 216. 1862. HOLOTYPE: Snowy Range, Colorado, 1862, C. C. Parry s.n. = Gentianella barbellata (Engelm.) J. M. Gil- lett. Gentiana heterosepala Engelm., Trans. Acad. Sci. St. Louis 2: 215. 1862. HOLOTYPE: Uintah Mts., Utah, 1859, H. Engelmann s.n. = Gen- tianella amarella (L.) Borner. Gentiana oregana Engelm. in Gray, Syn. Fl. N. Am. II. 1: 122. 1884. LECTOTYPE: Blue Mts., Oregon, 1874, R. D. Nevius s.n. Gentiana parryi Engelm., Trans. Acad. Sci. St. Louis 2: 218. 1863. LECTOTYPE: Colorado Ter- ritory, 1861, C. C. Parry 304. Syntype: “Rocky Mountain Flora,” 1862, E. Hall $ J. P. Harbour 470. Gentiana prostrata Haenke var. americana En- gelm., Trans. Acad. Sci. St. Louis 2: 217. 1863. SYNTYPES: Rocky Mountains, 1862, E. Hall £ J. P. Harbour 475; Snowy Range, s.d., C. C. Parry 306. Gentiana wislizenii Engelm., Trans. Acad. Sci. St. Louis 2: 215. 1863. HOLOTYPE: Llanos, Chihuahua, Mexico, 1846, 4. Wislizenus 206. = Gentianella wislizenii (Engelm.) J. M. Gil- lett. GERANIACEAE Geranium fremontii Torr. ex Gray var. parryi Engelm. in Gray, Amer. J. Sci. II. 33: 405. 1862. LECTOTYPE: Clear Creek, Colorado, 1861, C. C. Parry 113? Geranium gracile Engelm. in Gray, Mem. Amer. cad. Arts. IT. 4: 27. 1849. HOLOTYPE: Co- sihuiriachi, Mexico, 1846, A. Wislizenus 173 (2 sheets). = G. atropurpureum Heller. Geranium pentagynum Engelm. in Wisliz., Mem. ur. No. Mex. 90. 1848. HOLOTYPE: Wolf Creek, New Mexico, 1846, A. Wislizenus 508 (2 sheets). = G. richardsonii Fisch. & Trautv. HYDROPHYLLACEAE Eutoca patuliflora Engelm. & Gray. Boston J. Nat. Hist. 5: 253. 1845. ISOTYPE: Texas, s.d., F. Lindheimer 280. — Phacelia patuliflora (Engelm. & Gray) Gray. Eutoca strictiflora Engelm. & Gray, Boston J. Nat. Hist. 5: 253. 1845. ISOTYPE: Texas, s.d., F. Lindheimer 279 (2 sheets). = Phacelia stric- tiflora (Engelm. & Gray) Gray. ISOETACEAE Isoetes bolanderi Engelm., Amer. Naturalist 8: 14. 1874. LECTOTYPE: Tuolumne River, Cal- ifornia, 1866, H. Bolander 5091. Syntypes: Mary’s Lake, California, 1870, H. Bolander 5080; Mono Trail, California, 1870, H. Bo- lander 5093. Isoetes butleri Engelm. var. immaculata Engelm., Trans. Acad. Sci. St. Louis 4: 388. 1882. HOLOTYPE: near Nashville, Tennessee, 1880, A. Gattinger 3812. = I. butleri Engelm. Isoetes cubana Engelm., Trans. Acad. Sci. St. Louis 4: 389.1882. HOLOTYPE: eastern Cuba, 1860, C. Wright s.n. Isoetes echinospora Durieu var. robusta Engelm., Trans. Acad. Sci. St. Louis 4: 380. 1882. HOLOTYPE: Lake Champlain, New York?, 1879, C. G. Pringle s.n. Isoetes engelmannii A. Braun var. georgiana En- gelm., Trans. Acad. Sci. St. Louis 4: 385. 1882. HOLOTYPE: Georgia, 1873, A. W. Chap- man s.n. = I. engelmannii A. Braun. Isoetes flaccida Shuttlew. ex A. Braun var. chap- 1630 Annals of the Missouri Botanical Garden manii Engelm., Trans. Acad. Sci. St. Louis 4: 386. 1882. HOLOTYPE: near Mariana, Flor- ida, 1850, Chapman s.n. = l. flaccida Shut- tlew. Isoetes flaccida Shuttlew. ex A. Braun var. rigida Engelm., Trans. Acad. Sci. St. Louis 4: 386. 1882. HOLOTYPE: Lake Flint, Florida, 1878, A. Garber s.n. = l. flaccida Shuttlew. Isoetes howellii Engelm., Trans. Acad. Sci. St. Louis 4: 385. 1882. HOLOTYPE: The Dalles, Oregon, J. & T. J. Howell 1880 Isoetes lacustris L. var. paupercula Engelm., Trans. Acad. Sci. St. Louis 4: 377. 1882. LECTOTYPE: Middle Park, Colorado, 1881, G. Engelmann s.n. = l. occidentalis Hend. Isoetes melanospora Engelm., Trans. Acad. Sci. St. Louis 3: 395. 1877.HOLOTYPE: Stone Mt., Georgia, 1869, Canby s.n. (2 sheets). Isoetes nuda Engelm., Trans. Acad. Sci. St. Louis 4: 385. 1882. HOLOTYPE: The Dalles, Oregon, 1882, T. Howell 26. — I. howellii Engelm. Isoetes pygmaea Engelm., Amer. Naturalist 8: 214. 1874. HOLOTYPE: Mono Pass, California, 1866, H. Bolander 6025 (2 sheets). = 1. bolanderi Engelm. var. pygmaea (Engelm.) lute. Isoetes riparia Engelm. in A. Braun, Flora (Re- gensb. Bot. Zeit.) 29: 178. 1846. HOLOTYPE: om Pennsylvania, 1844, Zantziger TUA riparia Engelm. var. canadensis Engelm., Trans. Acad. Sci. St. Louis 4: 383. 1884. TYPE MATERIAL: Crow River, Canada, s.d., J. Macoun s.n. Isoetes saccharata Engelm. in Gray, Man. Bot. 5: 676. 1868. HOLOTYPE?: Salisbury, Maryland, 1866, Canby s.n. JUGLANDACEAE Juglans rupestris Engelm. in Torr., Bot. Sitgr. Rep. 171. 1853. TYPE MATERIAL: Texas, 1849, F. Lindheimer 20 (1178) JUNCACEAE Juncus asper Engelm., Trans. Acad. Sci. St. Louis 2: 478. 1868. SYNTYPES: Quaker Bridge, New Jersey, 1864, 1865, 1866, Smith s.n., 1866, 1867; C. F. Parker s.n. (2 sheets). Juncus balticus Willd. var. littoralis Engelm., ans. Acad. Sci. St. Louis 2: 442. LECTOTYPE & ISOLECTOTYPE: Ipswich, 1842, Oakes s.n. — J. arcticus Willd. subsp. littoralis Engelm. Juncus balticus Willd. var. montanus Engelm., Trans. Acad. Sci. St. Louis 2: 442. LECTOTYPE & ISOLECTOTYPE: Rocky Mountains, 1862, K. Hall & Harbour 567. = J. arcticus Willd. subsp. ater (Rydb.) Hult. Juncus bolanderi Engelm., Trans. Acad. Sci. St. Louis 2: 470. 1868. HOLOTYPE: Mission Do- lore, California, 1865, H. Bolander s.n. Juncus brachycarpus Engelm., Trans. Acad. Sci. St. Louis 2: 467. 1868. SYNTYPES: Charleston, South Carolina, Beyrich s.n.; Louisiana, no date, J. Hale s.n.; Detroit, Michigan, July 1867, J. M. Bigelow s.n.; Michigan, no date, Folwell s.n. Juncus breweri Engelm., Trans. Acad. Sci. St. Louis 2: 440. 1866. HOLOTYPE: Monterey, California, May 1866, W. H. Brewer 651. Isotype: Monterey, California, May 1861, W. H. Brewer 651. — J. arcticus Willd. subsp. mexicanus (Willd.) Lint x subsp. pacificus (Engelm.) Lint. Juncus canaliculatus Engelm., Bot. Gaz. 7: 6. 1882. HOLOTYPE: San Bernardino, California, 1881, S. B. & W. F. Parish 1091. = J. macrophyllus Coville. Juncus chlorocephalus Engelm., Trans. Acad. Sci. St. Louis 2: 485. 1868. SYNTYPES: Sierras, California, 1866, Hillebrand 2338 (2 sheets); W. H. Brewer 1804 (2 sheets); H. Bolander Cal. St. 6033. Juncus drummondii E. Mey. var. humilus En- gelm., Trans. Acad. Sci. St. Louis 2: 445. 1868. HOLOTYPE: Mt. Shasta, California, 1862, Brewer 1383. Juncus dubius Engelm., Trans. Acad. Sci. St. Louis 2: 459. 1868. SYNTYPE: Mariposa, California, s.d., H. Bolander Cal. St. Surv. 60 Juncus hallii Engelm., Trans. Acad. Sci. St. Louis 2: 446. 1868. HOLOTYPE: Lake Ranch, Col- orado, 1862, E. Hall & J. P. Harbour 562. Juncus kelloggii Engelm., Trans. Acad. Sci. St. Louis 2: 494. 1868. HOLOTYPE: San Francisco, California, 1866, 4. Kellogg s.n. Juncus mertensianus Bong. var. paniculatus En- e rans. Acad. Sci. St. Louis 2: 480. 1868. ISOTYPE: Rocky Mts., 1858, Bourgeau s.n. Juncus obtusatus Engelm., Trans. Acad. Sci. St. Louis 2: 495. 1868. HOLOTYPE: Big Tree Grove, Mariposa, California, 1866, H. Bo- lander Cal. St. Surv. 6028. Juncus oxymeris Engelm., Trans. Acad. Sci. St. Louis 2: 483. 1868. sYNTYPES: Big Tree Grove, Mariposa, California, 1866, H. Bolander 95 (4 sheets). Juncus parryi Engelm., Trans. Acad. Sci. St. Louis 46. 1866. HOLOTYPE: Colorado, 1861, C. C. Parry 360. Juncus phaeocephalus Engelm. var. glomeratus Surv. Volume 75, Number 4 1988 Wolf 1631 Engelmann Type Specimens Engelm., Trans. Acad. Sci. St. Louis 2: 484. 1868. SYNTYPES: San Francisco, California, 1866, A. Kellogg 96 (6 sheets). Juncus phaeocephalus Engelm. var. gracilis En- gelm., Trans. Acad. Sci. St. Louis 2: 473. 1868. SYNTYPES: Sierra Nevadas, California, 1865, H. Bolander Cal. St. Surv. 2339 (2 sheets), H. Bolander 98 (6 sheets); 1865, Brewer 1709. = J. mertensianus Bong. Juncus phaeocephalus Engelm. var. paniculatus Engelm., Trans. Acad. Sci. St. Louis 2: 484. 1868. SYNTYPES: Napa Valley, California, s.d., J. M. Bigelow s.n.; San Francisco, California, A. Kellogg 97 (4 sheets). Juncus rugulosis Engelm. in Coult., Bot. Gaz. 6: : 1881. HOLOTYPE: San Bernardino Moun- tains, California, 9 Nov. 1880, W. G. Wright ios: re Engelm., Trans. Acad. Sci. St. Louis 2: . 1868. HOLOTYPE: Broad Mt., Potts- 1 T. 1865, C. E. Smith s.n. Juncus supiniformis Engelm., Trans. Acad. Sci. St. Louis 2: 461. 1868. HOLOTYPE: Mendo- cino, California, 1866, H. Bolander Cal. St. Surv. Juncus ibunt Engelm., Trans. Acad. Sci. St. Louis 2: 492. 1868. LECTOTYPE: De Long's Ranch, Yosemite Valley, California, 1866, H. Bolander 30. Juncus triformis Engelm. var. brachystylus En- gelm., Trans. Acad. Sci. St. Louis 2: 492. 1868. ISOLECTOTYPE: Ukiah, California, 1866, H. Bolander 31 (3 sheets) & Cal. St. Surv. 8457. Syntype: Ukiah, California, 1866, H. Bolander s.n. Juncus triformis Engelm. var. stylosus Engelm., Trans. Acad. Sci. St. Louis 2: 492. 1868. ISOLECTOTYPE: Yosemite Valley, California, 1866, H. Bolander 30. = J. triformis En- gelm. Juncus triformis Engelm. var. uniflorus Engelm., Trans. Acad. Sci. St. Louis 2: 493. 1868. SYNTYPES: Sierra Nevada, California, 1866, Hillebrand Cal. St. Surv. 2333; Long Valley, California, 1866, H. Bolander 32 (Cal. St. Surv. 4691) (2 sheets). = J. hemiendytus F. Herman. Juncus vaseyi Engelm., Trans. Acad. Sci. St. Louis : 448. 1868. SYNTYPES: Saskatchewan, Can- ada, 1852 and 1858, E. Bourgeau s.n.; Col- orado, 1862, E. Hall s.n. LAMIACEAE Brazoria scutellarioides Engelm. & Gray, Boston J. Nat. Hist. 5: 257. 1845. PARATYPES: on paths and slopes of the prairie, in heavy soil, Texas, F. Lindheimer 286, 287. Monarda lindheimeri Engelm. & Gray, Boston J. Nat. Hist. 5: 228. 1845. SYNTYPE: Houston, Texas, 1843, F. Lindheimer 151 (2 sheets). Scutellaria cardiophylla Engelm. & Gray, Boston J. Nat. Hist. 5: 227. 1845. ISOTYPES: Texas, 1843, F. Lindheimer 144 (2 sheets). LINACEAE Linum aristatum Engelm. in Wisliz., Mem. Tour. No. Mex. 101. 1848. HOLOTYPE: near Carizal, south of El Paso, Mexico, 1846, A. Wisli- zenus 101. Linum greggii Engelm. ex Gray, Pl. Wrightiana 26. 1852. HOLOTYPE: San Antonio near Sal. tillo, Mexico, 31 Aug. 1848, J. Gregg s.n. Linum rigidum Pursh var. puberulum Engelm. ex Gray, Pl. Wrightiana 25. 1852. ISOTYPE: New Mexico, 1847, A. Fendler 85. = L. puber- ulum (Engelm.) Heller. Linum a Engelm. in Gray, Boston J. Nat. t. 6: 232. 1850. ISOTYPE: New Braunfels, Sonn 1846, F. Lindheimer 337. Linum rupestre Engelm. var. cymulosum Engelm. in Gray, Pl. Wrightiana 26. 1852. HOLOTYPE: battlefield of Buena Vista, Mexico, 1848, J. Gregg s.n. LOASACEAE Mentzelia chrysantha Engelm. ex Brandegee, Fl. S. W. Col. 237. 1878. noLoTYPE: Canon City, Colorado, 1874. G. Engelmann s.n. LORANTHACEAE Arceuthobium abietinum Engelm. in Gray, Proc. Amer. Acad. Arts 8: 401. 1872. TYPE MA- TERIAL: Oregon, 1871, E. Hall 457 & 458 (2 sheets each); California, 1875, J. Muir Arceuthobium campylopodium Engelm. in Gray, Boston J. Nat. Hist. 6: 214. 1845. ISOTYPE: Oregon, label says Rocky Mountains, ex herb Hooker, 1843, C. A. Geyer 577 (2 sheets). Arceuthobium campylopodium Engelm. var. macrarthron Engelm. in Gray, Boston J. Nat. Hist. 6: 214. 1845. SYNTYPES: Santa Fe, New Mexico, A. Fendler 282 (2 sheets); California, 1848, Douglas s.n.; 1848, C. A. Geyer s.n. Arceuthobium cryptopodum Engelm., in Gray, Boston J. Nat. Hist. 6: 214. 1850. HOLOTYPE: Santa Fe, New Mexico, 1847, A. Fendler 283. = A. vaginata (Willd.) Presl. subsp. 1632 Annals of the Missouri Botanical Garden cryptopodum (Engelm.) Hawksworth & Wiens. Arceuthobium divaricatum Engelm. in Rothr., Bot. Wheeler's Surv. 253. 1878. LECTOTYPE: Salt River Valley, Arizona, 1873, Gilbert 116. Syntype: Camp Apache, Arizona, 1874, J. B. Girard s.n. Arceuthobium douglasii Engelm. in Rothr., Bot. Wheeler's Surv. 253. 1878. LECTOTYPE: Santa Fe River, New Mexico, 1874, J. T. Rothrock 69. Syntype: Camp Apache, Arizona, 1873, G. K. Gilbert 109. Arceuthobium douglasii Engelm. var. abietinum Engelm. in Watson, Bot. Calif. 2: 106. 1880. LECTOTYPE: Sierra Valley, California, 1875, Lemmon s.n.. = A. abietinum Engelm. Arceuthobium occidentale Engelm. in Rothr., Bot. Wheeler's Surv. 375. 1878. LECTOTYPE: Walker’s Basin, California, 1875, J. T. Roth- rock 429. Arceuthobium robustum Engelm. in Gray, Mem. Amer. Acad. Arts. II. 4: 59. 1849. SYNTYPE: Camp Apache, Arizona, 1873, G. Gilbert 108. — 4. vaginatum (W eins) Presl. subsp. cryp- topodum (Engelm.) Hawksworth & Weins. Arceuthobium verticilliflorum Engelm. in Wat- son, Bot. Calif. 2: 107. 1880. HOLOTYPE: Sier- ra Madre, Durango, Mexico, 1852, Seeman 2138 (2 sheets). Phoradendron flavescens Nutt. var. macrophyl- um Engelm. in Rothr., Bot. Wheeler’s Surv. 252. 1878. SYNTYPES: Gila River, Arizona, 1873, Gilbert 104; Camp Grant, 1874, J. T. Rothrock 362. Phoradendron juniperinum Engelm. ex Gray, Mem. Amer. Acad. Arts. II. 4: 58. 1849. HOLOTYPE: Santa Fe, New Mexico, 1847, 4. Fendler 281. Phoradendron lanceolatum Engelm. ex Gray, Mem. Amer. Acad. Arts. Il. 4: 59. 1849. HOLOTYPE: Rinconada, Mexico, 1847, J. Gregg 5. i onn orbiculatum Engelm. in Gray, Mem r. Acad. Arts. II. 4: 59. 1849. LECTOTYPE: Little Rock, Arkansas, 1837, G. Engelmann 707. Syntype: same data but G. Engelmann s.n. = P. flavescens Nutt. var. orbiculatum Engelm. LYTHRACEAE Lythrum ovalifolium Engelm. ex Gray, Boston J. Nat. Hist. 6: 187. 1850. SYNTYPE: Pierden- ales, Texas, 1845, F. Lindheimer 450. MALVACEAE Sida heterocarpa Engelm. ex Gray, Boston J. Nat. Hist. 6: 163. 1850. TYPE MATERIAL: Industry, Texas, 1844, F. Lindheimer 189. — S. spi- nosa L. MARSILEACEAE Marsilea macropoda Engelm. in A. Braun, Amer. Sci. II. 3: 56. 1847. HOLOTYPE: Matagorda Bay, Texas, 1845, F. Lindheimer s.n. Marsilea tenuifolia Engelm. in A. Braun, Amer. J. Sci. IT. 6: 89. 1848. HOLOTYPE: Guadalupe River, Texas, 1837, F. Lindheimer s.n. MARTYNIACEAE Martynia arenaria Engelm. in Wisliz., Mem. Tour. No. Mex. 100. 1848. HOLOTYPE: sandhills be- low El Paso, 1846, A. Wislizenus 92. = Pro- boscidea altheaefolia (Benth.) Dene. Martynia violacea Engelm. in Wisliz., Mem. Tour. Mex. 111. 1848. SYNTYPE: Lake Enci- nilla, north of Chihuahua, Mexico, 1846, 4. Wislizenus 145. = Proboscidea louisianica (Mill.) Thell subsp. fragrans (Lindl.) Bretting. NYMPHAEACEAE Nuphar polysepalum Engelm. var. pictum En- gelm., Trans. Acad. Sci. St. Louis 2: 282. 1866. LECTOTYPE: Colorado Territory, 1864, C. C. Parry s.n. OLEACEAE Menodora scoparia Engelm. ex Gray, Bot. Calif. 1: 471. 1876. SYNTYPES: Saltillo, Mexico, 1848, J. Gregg 198 & 527 ONAGRACEAE Gaura lindheimeri Engelm. & Gray, Boston J. Nat. History 5: 217. 1845. TYPE MATERIAL: prairie near Houston, Texas, 1843, F. Lind- heimer s.n. Gaura suffulta Engelm. in Gray, Boston J. Nat. Hist. 6: 190. 1850. ISOTYPE: New Braunfels, Texas, 1847, F. Lindheimer 611 (4 sheets). PAPAVERACEAE Corydalis aurea Willd. var. micrantha Engelm. in Gray, Man. Bot. 5: 62. 1868. TYPE Ma- TERIAL: **459"' no additional collection infor- mation but label has complete description Volume 75, Number 4 1988 Wolf 1633 Engelmann Type Specimens probably used in publication. — C. micrantha (Engelm.) Gray. Corydalis crystallina Engelm. in Gray, Man. Bot. 5: 62. 1868. TYPE MATERIAL: Washington City, ea 1835, no collector. Corydalis montana Engelm. in Gray, Man. Bot. 5: 62. 1868. TYPE MATERIAL: Santa Fe, New Mexico, 1847, A. Fendler 17. = Willd. Dicentra ochroleuca Engelm. in Coult., Bot. Gaz. 3. 1881. HOLOTYPE: Santa Monica, Cal- ‘fornia, 1880, G. Engelmann s.n. (2 sheets). C. aurea PASSIFLORACEAE Passiflora affinis Engelm. ex Gray, Boston J. Nat. Hist. 6: 233. 1850. SYNTYPE: Texas, 1849, F. Lindheimer 817. Passiflora tenuiloba Engelm. in Gray, Boston J. Nat. Hist. 6: 192. 1850. SYNTYPE: on the Liano, 1848 and 1849, F. Lindheimer s.n. (2 sheets). PINACEAE Abies grandis Lindl. var. densiflora Engelm., Trans. Acad. Sci. St. Louis 3: 599. 1878. TYPE MATERIAL: Mt. Hood, Oregon, 1871, E. Hall s.n.; British North America, 1862, D. Lyall s.n.; British Columbia, Canada, s.d., D. Lyall s.n. = A. amabilis (Dougl.) Forbes. Picea pungens Engelm., Gard. Chron. 11: 334. 1879. TYPE MATERIAL: upper Arkansas, Col- orado, 1878, M. E. Jones 364; Lake County, Colorado, 1876, G. Vasey s.n. Both speci- mens prepublication date and annotated by Engelmann. Pinus aristata Engelm., Trans. Acad. Sci. St. Louis 2: 205. 1863. TYPE MATERIAL: Rocky Moun- tains, 1862, C. C. Parry s.n.; Rocky Moun- tains, s.d., C. C. Parry s.n. (2 sheets); the latter two specimens probably used for illus- trations in publication. Pinus arizonica Engelm. in Rothr., Bot. Wheeler's Surv. 260. 1878. HOLOTYPE: Santa Rita Mts., Arizona, 1874, J. T. Rothrock 652. Pinus brachyptera Engelm. in Rothr., Bot. Wheel- er's Surv. 89. 1878. LECTOTYPE: Rock Creek to Santa Fe, 1846, A. Wislizenus 534 (2 sheets). = P. ponderosa Dougl. ex P. & C. Lawson. Pinus chihuahuana Engelm. in Rothr., Bot. Wheeler’s Surv. 103. 1878. HOLOTYPE: Co- sihuiriachi, Mexico, 1846, A. Wislizenus 232 (2 sheets). Pinus edulis Engelm. in Rothr., Bot. Wheeler’s Surv. 88. 1878. SYNTYPES: Cimarron to Santa Fe, 1846, A. Wislizenus s.n. & 535 (2 sheets); New Mexico, 1846, J. Gregg s.n. Pinus elliottii Engelm., Trans. Acad. Sci. St. Louis 4: 186. 1880. SYNTYPES: South Carolina, 1871 (2 sheets), 1872, 1873, Mellichamp s.n. — P. cubensis Griseb. Pinus flexilis James var. macrocarpa Engelm. in Rothr., Bot. Wheeler's Surv. 258. 1878. HOLOTYPE: San Francisco Mts., Arizona, 1871, F. Bishoff s.n. Pinus flexilis James var. reflexa Engelm. in Rothr., Bot. Wheeler's Surv. 258. 1878. SYNTYPES: southern Arizona, s.d., J. T. Rothrock 654 & 1001 Pinus flexilis James var. serrulata Engelm. in othr., Bot. Wheeler’s Surv. 258. 1878. HOLOTYPE: Mt. Graham, Arizona, 1874, J. T. Rothrock 783. Pinus greggii Engelm. ex Parl. in DC., Prod. 16(2): 396. 1868. HOLOTYPE: Saltillo, Mexico, 1848, J. Gregg 402. Pinus latisquama Engelm., Gard. Chron. 18: 712. 1882. HOLOTYPE: south of Saltillo, Mexico, 1880, E. Palmer 1299. = P. pinaceana Gord. Pinus macrophylla Engelm. in Wisliz., Mem. Tour. No. Mex. 103. 1848. HOLOTYPE: Cosihuiri- achi, Mexico, 1846, A. Wislizenus 233 (2 sheets & 1 cone box). = P. engelmannii Carr. Pinus osteosperma Engelm. in Wisliz., Mem. Tour. No. Mex. 89. 1848. LECTOTYPE: Buena Vista, Mexico, 1847, J. Gregg 321 (2 sheets). — P. cembroides Zucc. Pinus parryana Engelm., Amer. J. Sci. II. 34: 332.1862. HOLOTYPE: southeast of San Diego, California, in Mexico, 1850, C. C. Parry s.n. (2 sheets). = P. quadrifolia Parry. Pinus strobiformis Engelm. in Wisliz., Mem. Tour. ex. 102. 1848. LECTOTYPE: Cosihuiri- achi, Mexico, 1846, A. Wislizenus 231 (2 sheets & 1 cone box). Syntype: Cosihuiriachi, Mexico, 1846, 4. Wislizenus 155. = P. ay- acahuite Ehrenb. ex Schlecht. Pinus wrightii Engelm., Trans. Acad. Sci. St. Louis 4: 185. 1880. SYNTYPES: eastern Cuba, 1860- 1864, C. Wright 3190 & 1462b (1 sheet & 1 cone box); 1859, C. Wright 1462 (a?). Tsuga caroliniana Engelm. in Coult., Bot. Gaz. 6: 223. 1881. svNTYPEs: South Carolina, 1851, L. R. Gibbs s.n., A. H. Curtiss s.n.; North Carolina, 1841, A. Gray & Carey s.n. 1634 Annals of the Missouri Botanical Garden PLANTAGINACEAE 1881. SYNTYPES: Copper Mines, 1851, C. Plantago pusilla Nutt. var. major Engelm., Bot. Gaz. 8: 175. 1883. HOLOTYPE: Indian Terri- tory, 1874, G. D. Butler s.n. — P. elongata Pursh. POACEAE Aristida basiramea Engelm. ex Vasey, Coult. Bot. Gaz. 9: 76. 1884. SYNTYPE: Minneapolis, Min- nesota, 1883, W. Upham s.n. Aristida ramosissima Engelm. ex Gray, Man. Bot. 5: 618. 1868. TYPE MATERIAL: east of Belle- ville, Illinois, Aug. 1845, G. Engelmann s.n.; St. Clair County, Illinois, Aug. 1858, G. En- gelmann s.n.; southern Illinois, 1861, G. En- gelmann s.n. e rd ui littoralis Engelm., Trans. Acad. i. St. Louis 1: 436. 1859. SYNTYPES: Man- a s.d., Berlandier 3227 (2 sheets); Texas, 1841, 1842, 1843, F. Lindheimer s.n. (5 sheets). Spartina junciformis Engelm. & Gray, Boston J. Nat. Hist. 5: 238. 1845. SYNTYPE: Galveston, Texas, 1843, F. Lindheimer 207 (3 sheets). POLYGONACEAE Eriogonum alpinum Engelm., Bot. Gaz. 7: 6. 1882. HOLOTYPE: Scotts Mt., California, 1880, G. Engelmann s.n. Eriogonum atrorubens Engelm. in Wisliz., Mem. Tour. No. Mex. 108. 1848. HOLOTYPE & ISOTYPE: Cosihuiriachi, Mexico, 1846, 4. Wis- lizenus 172 (2 sheets). Polygonum cristatum Engelm. & Gray, Boston J. Nat. Hist. 5: 259. 1845. HOLOTYPE: Texas, 1844, F. Lindheimer 237 (exsiccatae no. 296), “n. sp." added by Engelm. PORTULACACEAE Lewisia brachycalyx Engelm. ex Gray, Proc. A Acad. Arts 7: 400. 1868. SYNTYPES: Fort Whipple, Arizona, 1865, E. Coues « E. Palmer 211 (2 sheets); Utah, 1859, Brewer s.n. Portulaca lanceolata Engelm. in Gray, Boston J. Nat. Hist. 6: 154. 1850. SYNTYPE: Liano, Texas, 1848, F. Lindheimer s.n. = P. um- braticola H.B.K. Portulaca retusa Engelm., Boston J. Nat. Hist. 6: 154. 1850. HOLOTYPE: “ex sem. texanis cult. St. Louis,” from Llano in western Texas, Aug. 1849, G. Engelmann s.n Portulaca suffrutescens Engsln Bot. Gaz. 6: 236. Wright 874; Fort Whipple, Arizona, 1865, E. Coues & E. Palmer 547; Santa Rita Mts., Arizona, 1880, G. Engelmann s.n. Talinum aurantiacum Engelm. in Gray, Boston J. Nat. Hist. 6: 153. 1850. SYNTYPES: on the Llano, Texas, 1848, F. Lindheimer 579; on the Sabinas, Texas, 1847, F. Lindheimer s.n. Talinum calycium Engelm. in Wisliz., Mem. Tour. No. Mex. 88. 1848. HOLOTYPE: on the Ci- marron, 1846, 4. Wislizenus 475 (2 sheets); one is annotated *'n. sp." by Engelm. Talinum sarmentosum Engelm. in Gray, Boston J. Nat. Hist. 6: 153. 1850. SYNTYPE: Texas, 1847, F. Lindheimer 580. Talinum spathulatum Engelm. in Gray, Pl. Wrightiana 14. 1852. LECTOTYPE: western Texas to El Paso, Texas, 1849, C. Wright 35. — T. paniculatum (Jacq.) Gaertn. POTAMOGETONACEAE Potamogeton diversifolius Raf. var. spicatus En- gelm., Amer. J. Sci. 46: 102. 1843. HOLOTYPE: Cosihuiriachi, Mexico, 1846-1847, A. Wis- lizenus 159. RANUNCULACEAE Ranunculus texensis Engelm. in Engelm. & Gray, Boston J. Nat. Hist. 5: 210. 1845. SYNTYPE: near Houston, Texas, 1843, F. Lindheimer 1. = R. oblongifolius Ell. Ranunculus trachyspermus Engelm. in Engelm. & Gray, Boston J. Nat. Hist. 5: 211. 1845. SYNTYPE: Texas, 1843, F. Lindheimer 2. — R. pusillus Poir. var. angustifolius (Engelm.) L. Benson. Ranunculus trachyspermus Engelm. var. angus- tifolius Engelm. in Engelm. & Gray, Boston J. Nat. Hist. 5: 211. 1845. SYNTYPE: Galves- ton Island, Texas, 1843, F. Lindheimer 3. — R. pusillus Poir. var. angustifolius (Engelm.) L. Benson. Ranunculus trachyspermus Engelm. var. lind- heimeri Engelm. in Engelm. & Gray, Boston J. Nat. Hist. 5: 211. 1845. SYNTYPE: Galves- ton, Texas, 1843, F. Lindheimer s.n. Thalictrum fendleri Engelm. ex Gray, Mem. Amer. Acad. Arts. II. 4: 5. 1849. ISOTYPE: Santa Fe, New Mexico, 1847, A. Fendler 13 (2 sheets). ROSACEAE Greggia rupestris Engelm. in Wisliz., Mem. Tour. x. 114. 1848. SYNTYPE: Cosihuiriachi, Volume 75, Number 4 1988 Wolf Engelmann Type Specimens 1635 Mexico, 1846, A. Wislizenus 228 (2 sheets); Saltillo, Mexico, A. Wislizenus s.n.; Buena Vista, Mexico, 1847, A. Wislizenus 324. = Cowania mexicana D. Don. Prunus gracilis Engelm. & Gray, Boston J. Nat. Hist. 5: 243. 1845. ISOTYPES: west of the Brazos, Texas, March, 1844, F. Lindheimer 237; “Post Oak Plum ...," Texas, March, 1844, F. Lindheimer s.n. Prunus minutiflora Engelm. in Gray, Boston J. Nat. Hist. 6: 185. 1850. LECTOTYPE: between San Antonio and New Braunfels, Texas, s.d., F. Lindheimer 388. Rosa minutifolia Engelm., Bull. Torrey Bot. Club 9: 97. 1882. HOLOTYPE: All Saint's Bay, Cal- ifornia, 1882, C. C. Parry & M. E. Jones s.n. Rosa spithamea A. Gray var. subinermis Engelm., Bot. Gaz. 6: 236. 1881. HOLOTYPE: Big Trees, Sierra Nevada, California, 1880, G. Engel- mann. s.n. RUBIACEAE Bouvardia glaberrima Engelm. in Wisliz., Mem. Tour. No. Mex. 107. 1848. LECTOTYPE: Co- sihuiriachi, Mexico, 1846, 4. Wislizenus 161 (2 sheets). SAXIFRAGACEAE Fendlera rupicola A. Gray var. lindheimeri En- gelm. & Gray in Gray, Pl. Wrightiana 1: 77. 1852. HOLOTYPE?: Texas, 1850, F. Lindhei- mer 792 (2 sheets). Heuchera sanguinea Engelm. in Wisliz., Mem. our. No. Mex. 107. 1848. HOLOTYPE: Llanos, 1846, A. Wislizenus 210 (2 iud Saxifraga debilis Engelm. ex Gray, Pro Acad. Arts l5: . 1863. HOLOTYPE OR ISOTYPE: Colorado, 1861, C. C. Parry 167. SCROPHULARIACEAE Castilleja indivisia Engelm. in Engelm. & Gray, Boston J. Nat. Hist. 5: 47. 1845. SYNTYPE: from Houston to the Colorado, 1844 & 1846, F. Lindheimer 284 (3 sheets). Maurandya wislizenii Engelm. ex Gray in Torr., Bot. Mex. Bound. Surv. 111. 1859. LECTOTYPE: near Val Verde, New Mexico, 1846, A. Wis- lizenus 45. Penstemon coccineus Engelm. in Wisliz., Mem. our. No. Mex. 107. 1848. HOLOTYPE: Llanos, 1846, A. Wislizenus 207. = P. wislizenii (Gray) Straw. SPARGANIACEAE sparganium eurycarpum Engelm. in Gray, Man. Bot 1. 1868. TYPE MATERIAL: American Bottom, 13 Aug. 1860, G. Engelmann s.n. No specimens cited but collected prior to pub- lication date. Sparganium simplex Huds. var. androcladum Engelm. in Gray, Man. Bot. 5: 481. 1868. LECTOTYPE: near Boston, 1856, G. Engel- mann s.n. Syntypes: American Bottom, near St. Louis, Missouri, 1860, G. Engelmann s.n.; St. Clair County, 1854, G. Engelmann s.n.; St. Louis, Missouri, s.d., G. Engelmann s.n., annotated “n. sp." by Engelm. = S. andro- cladum (Engelm.) Morong. STYRACACEAE Styrax platanifolia Engelm. ex Torr., Smithsonian Contrib. Knowl. 6: 4. 1854. SYNTYPE: Texas, 1851, F. Lindheimer s.n. V ALERIANACEAE Fedia stenocarpa Engelm. ex Gray, Boston J. Nat. Hist. 6: 216. 1850. TYPE MATERIAL: New Braunfels, Texas, 1848, F. Lindheimer 407. = Valerianella stenocarpa (Engelm. ex Gray) Krok. VITACEAE Vitis arizonica Engelm., Amer. Naturalist 2: 321. 1869. SYNTYPES: Santa Cruz, Sonora, Mexico, C. Wright 919; New Mexico, s.d., C. Wright s.n.; Sonora, Mexico, 1851, G. Thurber 703; Rio Verde, Arizona, 1865, E. Coues & C. C Parry 551 & 553; southern Arizona, 1866, C. C. Parry s.n. = V. monticola Buckley subsp. arizonica (Engelm.) Rogers. Vitis pi Engelm. in Gray, Boston J. Nat. . 6: 166. 1850. SYNTYPE: Texas, 1844, F pe e s.n. ZYGOPHYLLACEAE Guaiacum angustifolium Engelm. in Wisliz., Mem. Tour. No. Mex. 113. 1848. LECTOTYPE: about "is and Saltillo, Mexico, 1847, A. Wisli- zenus 293. — Porlieria angustifolia A. Gray. Larrea glutinosa Engelm. in Wisliz., Mem. Tour. No. Mex. 93. 1848. HOLOTYPE: Saltillo, Mex- ico, 1846, J. Gregg 10, annotated “n. sp.” by Engelm. Syntype: Saltillo, Mexico, 1847, J. Gregg 369. 1636 Annals of the Missouri Botanical Garden LITERATURE CITED Bailey, L. H. 1934. The species of dur peculiar to North America. Gentes Herb. 3: 150-244. BENSON, L. 1982. The Cacti of the ade States and Canada. Stanford Univ. Press, Stanford, California. BocIN, n . Revision of the genus SAP m. New York Bot. Gard. 9: 179-233 UN J. M. 1894. Preliminary revision of the North Anhalonium and Lo- Preliminary | revision of the North Actions species of Echinocactus, Cereus, and a. Contr. U.S. Natl. Herb. 3: 355-462. 959. Asa Gray, 1810-1888. Harvard Uni niv. Press, Cambridge, Massachusetts. ENGELMANN, G. & A. Penn 1845. Plantae Lindhei- merianae, an enumeration of the plants collected in Texas, and distributed is ird ie Boston J. Nat. Hist. 5: 210-264. GENTRY, H. S. 1982. Agaves of Psi" North America. Univ. of Arizona Press, HawkswonrTH, F. G. € D. WEINS. 1972. Classification of Dwarf Mistletoes (Arceuthobium). Forest Service, U.S. Dept. Agriculture, Washington, KARTESZ, J. T. & R. KanrEsz. 1980. A Synonymized eR of the Vascular Flora i: the United States, anada, and Greenland, Volum Carolina Press, Chapel McKELvEy, S. D. 1938. vocab of the Southwestern United States. I. Arnold Arboretum, Jamacia Plain, Massachusetts. 19 Yuccas of the Southwestern United hate: II. Arola Arboretum, Jamaica Plain, Mas- sachusetts. PFEIFFER, N. E. 1 Monograph of the Isoetaceae. Ann. Missouri Bot. fom d. 9: 79-233. RoLLINs, R. C. 1972. The need for care in choosing lec apa b 21: -637. SARGENT, 884 Botanical papers of ie En- ge v pr Moti cud 9: 69-74. diestra, P. P. eorge Engelmann: Scientist TRELEASE, W. 1916. e Genus Phoradendron. Univ. of Illinois Press, Urbana, Illinois. he American oaks. Mem. Natl. Acad. Sci. 22: 1-255. —— — — & A. Gray (editors). 1887. The Botanical Works of the Late George Engelmann. John Wilson n, Cambridge, Massachusetts. WHEELER, L. C. 1941. Euphorbia subgenus Chamae- syce in Canada and the United States exclusive of southern Florida. Rhodora 43: 97-154, 168-205, 223-286. Woopson, R. E. 1954. The North American m x nep L. Ann. Missouri Bot. Gard. Yunes KER, T. G. 1932. The genus Cuscuta. Mem. Tor- rey Bot. Club 18: 113-331. SYNOPSIS OF DICHANTHELIUM (POACEAE) IN FLORIDA! Bruce F. Hansen and Richard P. Wunderlin? ABSTRACT to species and varieties and a iiie taxa i bein ptr are recognized for Florida, with synonymy and typification compiled. Keys in o names is provided. The n ew combinations D. ensifolium var. breve, D. ensifolium var. unciphyllum, wid D. portoricense are proposed. Recent papers concerning Dichanthelium that have dealt with either the southeastern United States (Gould & Clark, 1978; Lelong, 1984, 1986) or with only a small part of the genus (Freckmann, 198 la) have been very influential in redefining the limits of the taxa. The present paper expands on the treatment of the Florida species in Wunderlin (1982) and Clewell (1985) in order to modernize the concepts and to rectify some errors. At the same time, it is an attempt to classify the Florida specimens better, recognizing that the taxa ac- cepted here might well be inadequately separable in other geographic areas, which seems the rule for this genus. However, we feel that practicality alone would mandate such a classification for this very difficult genus to facilitate assignment of names to specimens. In Dichanthelium, reticulate evo- lution caused by hybridization and autogamy is extremely common (Spellenberg, 1975a, b; Lelong, 1984, 1986). Thus, no claim is made here that a “natural” classification has been constructed. The merits of accepting the segregate genus Dichanthelium versus the inclusive Panicum will not be argued here except to point out that Di- chanthelium is as “good” a grass genus as many others (e.g., Brachiaria, Sacciolepis, and many more in different tribes). We will not detail here which authors accept one or the other genus; those KEY TO THE FLORIDA SPECIES OF DICHANTHELIUM la. Leaves at base of pla that wish to use Panicum will not be swayed by our arguments. Gould & Clark (1978) provided 24 characteristics separating Dichanthelium from Panicum; none of these alone separates the genera, but the total we believe defines Dichanthelium as a natural segregate from the large and variable Panicum. It is our belief that the acceptance of Dichanthelium provides a more consistent generic classification of the grasses of Florida. e species and varietal concepts used are roughly those of Gould & Clark (1978) as modified by Lelong (1984, 1986). Because the synonymy is so extensive and in many cases confusing, all known synonyms are compiled and included in the hope that future workers will find it easier to follow the nomenclature. The types are listed where known, but it must be emphasized that little original re- search in typification is involved here; the citation of types is mostly from published material, prin- cipally Hitchcock & Chase (1910), Gould & Clark (1978), and Lelong (1984). It seems inevitable that several new combinations are necessary, this in a group already overburdened with surplus names. The key fairly reflects the variation in Dichan- thelium species, such that many taxa will key out in two or more places. When a character falls between an overlapping lead, either choice should lead to the correct species. numerous, relatively long and soft, similar to and only slightly shorter, if at all, -10 than the soft, deren green lower culm blades; plants branching only at base; blades mostly 4 ide. ' The authors wish to thank Gerrit Davidse of the Missouri Botanical Garden for carefully reviewing the manuscript and for many helpful comments. Thanks are also due to Dan Nicolson of the Smithsonian Institution for help with a knotty nomenclatural ee and to Jimmy Massey of North Carolina State University for help As in locating certain W. W. Ashe collec * Department of Biology, JOE. of Sol Florida, Tampa, Florida 33620, USA. ANN. MISSOURI Bor. GARD. 75: 1637-1657. 1988. 1638 Annals of th Missouri Eon Garden Sheaths retrorsely pilose; uppermost culm blade at least % as long as those of basal tuft .... D. oe 2a. 2b. Sheaths glabrous or ascending pilose; uppermost culm blade less than 34 as long as those of basa rigosum Leaves at base of pan typically forming a basal rosette of short, relatively broad blades, these ua lb. Leav conspicuously shorter than the culm blades; plants normally branching above the base, especially in the autumnal phase; blades 3-25 mm wide or wider. 3a. Spikelets 0.8-3.2 mm long. 4a. Spikelets 0.8-2 mm long. D. acuminatum Ligule hairs, at least some, 2-5 mm long 5b. Ligule hairs under 1.5 mm long or absent . Lower culm internodes (and usually sheaths) pubescent, puberulent, strigose, or villous. 7a. Blades of midculm leaves mostly 2-5 mm wide; plants usually at least 10 cm tall D. aciculare 7b. Blades of midculm leaves less than 2 mm wide; plants less than 10 cm tall D. ensifolium (var. breve) 6b. e culm internodes (and usually sheaths) glabrous Blades of midculm leaves (vernal phase) 5-15 cm dar mostly 2-5 mm wide ..... . vid 8b. Blades of midculm leaves less than 6 cm long, or if longer, then wider than 6 m B of lower culm leaves aig or subcordate; blades 7-30 mm wide. lon D. erectifolium 10a. Spikelets 1-1.1 10b. Spikelets 1.4-2.2 mm m sphaerocarpon Ob. Bases of lower culm leaves Ke than cordate or subcordate; blades usually less than 7 mm wide. lla. Spikelets 1.6-2.5 mm long; plants usually diffuse l. D llb. ee 0.8- 1.5 mm long; plants usually sati forming ... 4b. Ui 2.1-3.2 mm lor 12a. Blades of primary MM at least some, (10-)13-25 mm wide. 13a. Culm nodes conspicuously bearded, the lower internodes with long, spreading hairs scoparium . dichotomum D. ensifolium 13b. us nodes glabrous or slightly hairy, the internodes typically glabrous. Ligule a fringed or entire membrane. a. Plants less than 75 cm tall; leaf bases cordate „n D. comm 15b. Plants mostly more = 75 cm tall; leaf bases rounded ... D. P ipu ms 14b. Ligule a ring of hairs or abser l6a. Spikelets 1.4-2.2 mm long. broadly elliptic to suborbicular 0. . sphaerocarpon D. clandestinum tum 16b. Spikelets 2.4-3.5 mm long, narrowly ovate or elliptic ......... 12b. Blades d iind culms up to 12 mm wide. 17a bases cordate or subcordat 182. Spikelets 1.4-2.2 mm n broadly elliptic to ovate _.... 18b. Spikelets 2.4-3.2 mm long, elliptic to narrowly ovate oo. 17b. Pent Pri other than cordate or subcor à te. gule hairs, at least some, 2-5 m ves of the upper glume ud lemma of E iil floret broad, distinct; sheaths glabrous or pubescent with short hairs „a D. oligosanthes 20b. Nerves of the upper glume and lemma of the lon ea narrow, relatively .. D. sphaerocarpon ). commutatum indistinct; a spreading pubescent with lon 1 inch .2-3.1 mm long; ligules usually xi a distinct ring of airs in front of long hairs D. ovale 21b. Spikelets 1.4-2 mm long; ligules lacking a distinct ring of shorter D. acuminatum 19b. Ligule hairs bs bona mm long. 22a. C M nodes, at least lower ones, conspicuously bearded. Glabrous- pes sa ee just below the nodes; leaves usually vety-pubescent scoparium velv : 23b. i nau band not present; leaves glabrous or variously pubescent below, y. 4a. Culm ia le at vee lower ones, strigose or villous; blades pubesce w . 0 nt belo 24b. Culm internodes glabrous; blade surfaces glabrous ................... . dichotomum vale 22b. Culm nodes not bearded, lower ones sometimes sparsely hairy or puber- ulent 25a. Ligule membranous; plants typically 75-150 cm tall aa scabriusculum tall. 25b. Ligule a ring of hairs or absent; plants mostly less than 70 cm t Hansen & Wunderlin 1639 Volume 75, Number 4 1988 Dichanthelium in Florida 26a. Blades of midculm leaves (vernal phase) linear and stiff, often becoming involute, mostly 2-5 mm wide, 6-15 cm long ..... 26b. Blades of midculm leaves lanceolate, usually soft, romaine flattened, mostly more than 5 mm wide when 8 cm long or longer 27a. Upper glume and lemma of the lower floret with broad, ounded nerves; spikelets rounded at the apex .............. D. oligosanthes 27b. Upper glume and lemma of the lower floret with narrow, faint or srs indistinct nerves; spikelets usually pointed at the apex. 28a. Lower Bs internodes pubescent ... D. portoricense h 28b. Lower culm internodes glabrous ...... D. dichotomum 3b. Spikelets 3.3-5.2 mm lon 9a. Blades of at least some culm leaves 13-35 mm wide. 30a. Spikelets broadly elliptic to obovate, feda with diens ^mi nerves n D. oligosanthes 30b. Spikelets narrowly elliptic to obovate, not strongly n la. Culm nodes glabrous or slightly pubescent; spikelets 3.3-3.8 mm long ......................... D. clandestinum 31b. Culm nodes bearded; spikelets 3.8-5.2 m m lon 2a. Ligules 2.5-4 mm long; leaf blades eae puberulent-tomentose on one or both sufaces . ravenelii 32b. Ligules 1.5 mm long; leaf blades glabrous to puberulent ou... 29b. Blades of culm leaves up to 12 mm wide. D. boscii 33a. Midculm blades narrow, stiffly ascending, often becoming involute, mostly 6-15 cm ph D. aciculare 33b. Midculm blades beieder, spreading, 3-8 cm long, 3-8 mm wide ooo... D. Mte ccn Dichanthelium aciculare (Desvaux ex Poiret) Gould & Clark, Ann. Missouri Bot. Gard. 65: 1116. 1978. Panicum aciculare Desvaux ex Poiret in Lamarck, Encycl. Suppl. 4: 274. July 1816. TYPE: United States. Herb. s.n. (holotype, P; isotype, US, fragm.). Although the protologue gives the locality as “Habitat in india orientale," Hitchcock $ Chase (Contr. U.S. Natl. Herb. 15: 166. 1910) stated that the type collection “is without doubt rom the southeastern United States." Desvaux Panicum angustifolium Elliott, Sketch Bot. S. ricas 1: 129. Dec 1816. Panicum nitidum jene rck va angustifolium (Elliott) A ray, r. Gr E 112. 1835 asea angustifolia (Elo) Nieuw- SA land, Amer. Midl. Naturalist 2: 64. . Dichan — oe ' (Elliott) ou rn 26: 59. 1974. E: United States. South Carolina: El- s.n. (lectotype, PH-M). Lectotypified by Hitchcock & Chase (Contr. U.S. Natl. Herb. 15: 166. 1910). Panicum NCC ane Bosc ex Sprengel, Syst. Veg. 1: 312. 1825. TYPE: United States. "Carolina": Bosc s.n. (lectotype, B-W; isotype, G). Lectotypified by Hitchcock & Chase (Contr. U.S. Natl. Herb. 15 166. 1910). Panicum curtisii Steudel, Syn. Pl. Glumac. 1: 66. 1853. TYPE: beg States. South Carolina: Curtis s.n. (holot ). Ms neuramhum Grisebach, Cat. Pl. Cub. 232. 1866 : Cuba. Oriente: 1860, Wright 3453 (lecto- type, GOET). Lectotypified by D & Chase (Contr. U.S. Natl. Herb. 15: 17 PONE neuranthum Grisebach var. ramosum Grise- h, Cat. Pl. Cub. 232. 1866. Panicum fusiforme A. p ed Contr. U.S. Natl. Herb. 12: 222. 1909. TYPE: Cuba. 1863, Wright 3454 (holotype, GOET). Panicum ee Ashe, J. Elisha Mitchell Sci. Soc. 15: 42. 1898. TYPE: United States. North Carolina: 2 Co., Penitentiary Woods, Raleigh, Apr. 1895, Ashe s.n. (lectotype, US). Lectotypified by Hitch- cock & Chase (Contr. U.S. Natl. Herb. 15: 177. 1910 P iru arenicola Ashe, J. Elisha Mitchell Sci. Soc. 15: 56. 1898. TYPE: United States. North Carolina: Or- ange is Chapel Hill, June 1898, Ashe s.n. (lec- totype, US). Lectotypified by Hitchcock & Chase (Contr. U.S. Natl. n 15: 166. 1910). Panicum ovinum Lamson-Scribner he J. U.S. mith, D.A. Div. Año d ten p Panicum aciculare Desvaux ex Poi m (La Scribner & J. G. Smith) Bete, Phytologia 48: - 192. 1981. TYPE: United States. Texas: Waller Co., 25 lotype, US). Panicum arenicoloides Ashe, J. Elisha Mitchell Sci. Soc. 16: 89. Feb. 1900. Pontus aciculare Desvaux ex Poiret var. arenicoloides (Ashe) Beetle, Phytologia 48: 192. 1981. TYPE: United States. North Carolina: ew Hanover Co., near Wilmington, 7 June 1899, Ashe s.n. (holotype, NCU?; isotype, US, fragm.). Specimen at NCU not loc aed, Panicum oe Ashe, J. Elisha Mitchell Sci. Soc. 16: eb. 1900. TYPE: United States. North is ew Hanover ur rune 1899, Ashe s.n. (holotype, NCU; iso Panicum orthophyllum Ashe, I. Elisha Mitchell Sci. Soc. 1640 Annals of the Missouri Botanical Garden 16: 90. Feb. 1900. TYPE: United States. North Carolina: New Hanover Co., June 1899, Ashe s.n. (holotype, ied isotype, US, fragm.). Specimen at NCU not located. Panicum PORA Ashe, North Carolina Agric. Exp. ull. 175: 116. Aug 1900. TYPE: United States. Delaware: Newcastle Co., near Centerville, 6 July 8, Commons s.n. (holotyp e, NCU). Panicum pinetorum Swallen, Proc. Biol. Soc. Wash. 55: 93. July 1942. Type: United States. Florida: Lee near Bonita Springs, 14 Oct. 1940, Silveus 6604 (holotype, NA). Panicum oo W. Brown, Bull. Torrey Bot. Club 69: 539, f. 1. Oct. 1942. TYPE: United States. North Sii Durham Co., Bennett Memorial, 5 mi. W of Durham, Brown 2492 (holotype, DUKE). The vernal phase is easily recognized by the long, narrow, rather stiff upright leaves. In the autumnal phase, D. aciculare can be confused with several other species, making determinations very cult. E his species usually occurs in drier sites through- out Florida. Dichanthelium acuminatum (Swartz) Gould & Clark, Ann. Missouri Bot. Gard. 65: 1123 1978. Panicum acuminatum Swartz, Prodr. 23. 1788. TYPE: Jamaica: Swartz s.n. (ho- lotype, S; isotype, US, fragm.). As treated here, specimens with long ligule hairs define the Dichanthelium acuminatum complex, which consists of three species characterized by spikelet length: D. leucothrix (0.8-1.2 mm), D. acuminatum (1.4-2 mm), and D. ovale (2.2-3 mm). Dichanthelium ovale was separated from D. acuminatum by Gould & Clark (1978) by its sup- posedly doubled ligule, which forced specimens with midsized spikelets into D. ovale. While it can at times be quite pronounced, this character is often difficult to see and actually becomes undetectable on many specimens. When segregated by spikelet length, however, formerly confusing taxa such as D. ovale var. addisonii (Gould & Clark, 1978) and D. ovale vars. villosum and pseudopubescens (Lelong, 1984, 1986) fall readily into D. acumi- natum. Within D. acuminatum, many of the pubes- cence and habit phases have received varietal or specific status in the past. Lelong (1984, 1986) recognized six varieties. His varieties acuminatum, "unciphyllum," and columbianum rely on wheth- er the culm and sheath pubescence is appressed or spreading or both; this completely intergrades in a large suite of specimens. His variety fascic- ulatum is based on larger leaves, again a com- pletely intergrading character. On the other hand, in the glabrous varieties lindheimeri and densiflor- um, the lack of pubescence seems to be correlated with smaller spikelet size and generally robust habit for the former, and with an elongate inflorescence for the latter. Therefore, we are recognizing only three varieties in Florida la. Culms and sheaths pubescent ..... var. acuminatum lb. Culms and sheaths Shaan or sat 2a. a narrow, common Pis cm m leaves dark green oo r. hpc ee r. lindheimeri 2b. Panicle broad, 5- 8 cm long; leaves yellow green Dichanthelium acuminatum var. acumina- tum Panicum lanuginosum Elliott, Sketch Bot. S. Carolina 1: 123. 1816. Dichanthelium lanuginosum (Elliott) Gould, Brittonia 20: 60. 1974. TYPE: United States. Georgia: Baldwin s.n. (holotype, CHARL; isotype, N. Middle United States 145. 1824. Pani- cum huachucae Ashe var. ETEA (Torrey) F. Hubbard, Rhodora 14: 171. 1912. Panicum lind- heimeri Nash var. fasciculatum ll pala Rhodora 23: 228. 1922 (“1921”). P m lan ginosum Elliott var. um Torra i Fernald, 1934. Dichanthelium lanugino- Piina dichotomum Linnaeus var. fasciculatum Tor- rey icum acum latum Ln Lelong, Barone ^36: 16 TYPE: United States. New Jersey: Torrey s s.n. (ho- *, ) t Panicum nitidum Lamarck var. roges Torrey, Fl. N. iddle United c 146. 1824. TYPE: United States. New Jersey: Torrey s.n. ae NY; isotype, US, ragm. Panicum nitidum Lamarck var. pilosum Torrey, Fl. N. iddle United States 146. 1824. TYPE: United States. New York: Torrey s.n. (holotype, NY; isotype, US, fragm.). Panicum ir ood ovis bed villosum A. Gray, N A am. 2: 11 . Dichanthelium acu- minatum (Swartz) e ^ pa rk var. villosum (A. n wartz var. villosum (A. Gray) Beetle, Phytologia 48: 192. 1981. Panicum ovale Elliott ep villosum (A. Gray) Le- long, Brittonia 36: 272. 1984. TYPE: United States. New York: Ontario Co., en s.n. (holotype, GH? Panicum diana Linnaeus var. villosum Vasey, U.S.D.A. Div. Agrost. Bull. 8: 31. 1889. TYPE: United States. District of Columbia: near Pierce's Mill, Rock Creek, 1 July 1883, Vasey s.n. (lecto- type, US). Lectotypified by pow & Chase (Contr. U.S. Natl. Herb. 15: 233. ° Panicum villosissimum Nash, Bull. a B Club 23: 896. Dichanthelium lanuginosum (Elliott) Gould var. villosissimum (Nash) Gould, Brittonia 26: ; chanthelium villosissimum (Nash) Freckmann, Phytologia 39: 270. 1978. TYPE: United Ne) ~J] ae Volume 75, Number 4 1988 Hansen & Wunderlin 1641 Dichanthelium in Florida States. re Bibb Co., Ocmulgee River swamp below Macon, 18-24 May 1895, Small s.n. (ho- type, NY: pde NY, US). Panicum atlanticum Nash, Bull. Torrey Bot. Club 24: 346. 1897. TYPE: United States. New York: Bronx, New York Botanical Garden, ES June 1897, Nash Lamson-Scribner v nu in McNeill & Doc Naturaliste Canad. 103: 562. 1976. Dichanthelium commonsianum (Ashe) Freckmann, Phytologia 39: 271. 1978. TYPE: United States. New Jersey: Cape May Co., Cape May, June 1898, Commons 34.1 (holotype, NCU; isotypes, NY, S). U Panicum filiculme Ashe, J. Elisha Mitchell Sci. Soc. 15: 59. Feb. 1898, non Hackel 1895. TYPE: United States. Georgia: DeKalb Co., Stone Mountain, Aug. 1895, Ashe s.n. (lectotype, US). Lectotypified by PD & Chase (Contr. U.S. Natl. Herb. 15 210. Panicum pos TER Ashe, J. Elisha Mitchell Sci. Soc. : 55. Feb. 1898. TYPE: United States. District of Columbia: 1897, Kearney s.n. (lectotype, NCU; iso- lectotypes, NY, US). Lectotypified by Hichcock and Chase (Contr. U.S. Natl. Herb. 15: 233. 1910). Although Lelong (1984) cited NY as the location of the lectotype, Hitchcock & Chase (1910) clearly stated that the type is in Ashe's herbarium, which is now at NCU. Panicum d bag Ashe, J. Elisha Mitchell Sci. Soc. 1 . Feb. 1898. Panicum lanuginosum Elliott var. iM on (Ashe) A. Hitchcock, Rhodora 8: 208. 1906. TYPE: United States. Arizona: Huachuca Mountains, 1882, Lemmon s.n. (lectotype, US). Lectotypified by HAR & Chase (Contr. U.S. Natl. Herb. 15: 215 Panicum meridionale rin ‘4 ae Mitchell Sci. Soc. 15: 59 8. Panicum unciphyllum Trinius var. meridionale (Ashe) Lamson- d 2 Merrill, Rhodora 3: 123. 1901. Panicum lin eri Nash subvar. meridionale (Ashe) Farwell, ATE Midl. Nui iq 11: 45. Panicum lanuginosum Elli A yp pd ed Farwell, Pap. d. Sci. 26: 41. Dichanthelium he eee Pardos 39: 270. i : United States. North Carolina: Burke Co. Ton Ridge, June 1893, Ashe s.n. (lectotype, us) Pa el by Popp & Chase (Contr. atl. Herb. 15: 210. Eu bonis co Ashe, J. Elisha Mitchell Sci. Soc. 15: eb. 1898. TYPE: United States. North Carolina: Orange Co., Chapel Hill, June 1898, a s.n. (holotype, not located) Hitchcock & Chas (Contr. U.S. Natl. Herb. 15: 210. 1910) reli this name to the synonymy of Panicum meridionale based on the description only. Panicum scoparioides Ashe, J. Elisha Mitchell Sci. Soc. 15: 53. Feb. 1898. Panicum villosissimum Nas var. E (Ashe) Fernald, Rhodora 36: 79. 1934. United States. Delaware: Newcastle Co., ile 25 June 1873, Commons 283 (lec- totype, US; isotype, NY). Lectotypified by Hitchcock & Chase (Contr. U.S. Natl. Herb. 15: 238. 1910). Panicum tennesseense Ashe, J. Elisha Mitchell Sci. Soc. 15: 52. Feb. 1898. Panicum lindheimeri Nash var. tennesseense (Ashe) Farwell, Amer. Midl. Naturalist : 4 8. TYPE: United States. Tennessee: La Vergne be ., 7 Aug. 1897, Biltmore Herb. 7087 (holotyp Panicum dua Nash, Bull. Torrey Bot. Club 25: 83. 12 Feb. 1898. Panicum commonsianum Ashe subsp. addisonii (Nash) W. Stone, New Jersey State Mus. Annual Rep. 1910: 205. 1911. Panicum common- um Ashe var. addisonii (Nash) R. Pohl, Amer. . Naturalist 38: 582 1978. TYPE: United States. ay Jersey: Cape May Co., Wildwood, 30-31 Ma Bicknell S.R. (holotype, NY; isotype, US, fra iil Panicum implicatum Lamson- Scribner, in Britton & Brown, Ill. Fl. N. U.S. 3: . 20 June T Panicum unciphyllum ae var. implica (Lamson-Scribner) Lamson-Scribner & Merrill, Rho- dora 3: 123. 1901. Panicum lindheimeri Nash var. implicatum (Lamson-Scribner) Fernald, Rhodora 23: 228. 1922 (“1921”). Panicum lanuginosum Elliott - Mains: Cu Scribner s.n (holotype, US; isotype, NY) Panicum thurowii Lamso & G. Smith, U.S.D.A. Div. Agrost. Circ. 16: 5. 1 July 1899. & J. G. Smith) Gould ig es ciliosum Nash, Bull. Torrey Bot. Club 26: 568. 1899. TYPE: United States. Mississippi: Har- rison ‘Co. Biloxi, 1 Sept. 1898, Tracy 4580 (ho- lotype, NY; isotypes, NCU, US). Dium Aura p ene ae Bull. Torrey Bot. Club 26: 577. Nov. 1899. Panicum villosissimum Nash var. pseudopubescens (Nash) Fernald, Rhodora 36: 79. 1934. Panicum ovale Elliott var. Bitte e bescens (Nash) Lelong, E s nó TYPE: United States. Alabama: a bur May 1898, Earle & Baker 1537 (halo, NY: isotype, US) Je occidentale Lamson-Scribner, Annual Rep. Mis- ri Bot. Gard. 10: 48. 1899. TYPE: Canada. British Columbia: Vancouver Island, “Hab. in Nootka-Sund,”” Haenke s.n. (holotype, MO; isotypes, PR, US, fragm.). The choice of the PR specimen as lectotype by Hitchcock & Chase (Contr. U.S. Natl. Herb. 15: 228. 1910) is invalid; Lamson-Scribner worked from the specimen at MO and never saw the PR material. Panicum eie Ashe, J. Elisha Mitchell Sci. Soc. 15: 113. 1899. TYPE: United States. North Carolina: Orange Co., Chapel Hill, 29 June 1898, Ashe s.n. y pee. Us) Mr by Hitchcock & Chase . U.S . Herb. 15: 220. 1910). bi benar Ashe, J. Elisha Mitchell Sci. Soc. 16: 84. Feb. 1900. Panicum meridionale Ashe var. e cie Gas Fernald, Rhodora 36: 76. 1934. : United States. North Carolina: Beauford or 1642 Annals of the Missouri Botanical Garden yde Co., near Scranton, 26 May 1899, Ashe s.n. (lectotype, US). pii ees by Hitchcock & Chase (Contr. U.S . Herb. 15: 212. 1910). Panicum S Ashe, J. Elisha Mitchell Sci. Soc. 16: 86. Feb. 1900. TYPE: United States. Minnesota: Carlton Co., Carlton, Aug., Ashe s.n. (holotype, Panicum wilmingtonense Ashe, J. Elisha Mitchell Sci. Soc. 16: 86. Feb. a: TYPE: United States. North Carolini: New Hanover Co., near Wilmington, 17 May 1899, Ashe s.n. jor NCU; isotype, US). Panicum alabamense Ashe, North Carolina Agric. Exp. Sta. Bull. 164: 116. Aug. 1900, non Trinius 1854. TYPE: United States. Alabama: Lee Co., Auburn, May 1898, Alabama Biol. Surv. 1530 (holotype, NCU; isotypes, NY, US). Panicum auburne Ashe, as ion Agric. Exp. Š Bull. 175: 115. Aug. 1900. TYPE: United B utes. Alabama: Lee Co., jmd 7 May 1898, Earle & Baker 1527 (holotype, NCU; am. US). Panic um une iphyllum Trinius forma pilosum Lamson- 7 July 1891, Fernald 501 (holotype, GH). Panicum unciphyllum Trinius forma prostratum Lam- on-Scribner & Merrill, Rhodora 3: 124. 20 May 901. Panicum languidum A. Hitchcock & Chase, Contr. U.S, Natl. Herb. 15: 232. 1910. TYPE: United States. Maine: York Co., South Berwick, 26 S 1897, Fernald s.n. (holotype, GH). sla xanthospermum Lamson-Scribner & C. Mohr n Mohr, Contr. U.S. Natl. Herb. 6: 348. 31 July 1901. TYPE: United States. Alabama: Butler Co., Mohr s.n. (holo li US). r 30: orida: n Co., Lake Jackson, 12 May Curtiss s aja NY). ipie lanuginosum Elliott var. siccanum A. Hitch- & Chase, Rhodora 8: 207, 1906. TYPE: United "iir Illinois: LaSalle Co., Starved Rock, 1 July 1901, Chase 1602 (holotype, US). Panicum oricola A. Hitchcock a Chase, Rhodora 8: 208. anicum columbianum Lamson-Scribner var. s a (A. Hitchcock & Chase) Fernald, Rhodora 6: 79. end TYPE: United States. Delaware: Sus- sex Co., Lewes, 18 June 1905, Hitchcock 47 (ho- lotype, US; isotypes, MO, Panicum eee ocius A. Hitchcock & Chase, Rhodora 206. 1906. Panicum lanuginosum Elliott var. ivi cocius hi Hitchcock & Chase) Dore ie McNeill & Dore, Naturaliste Canad. 103: 576. 1976. oa thelium villosissimum (Nash) E nn var. pra cocius (A. Hitchcock & Chase) Freckmann, Phy. tologia 39: 270. 1978. TYPE: United States. Illinois: Stark Co., near Wady Petra, 30 June 1900, Chase 9 (holotype, US). Panicum unciphyllum Trinius var. thinium A. Hitchcock thase, Rhodora 8: 209. 1906. Panicum columbi- anum Lamson-Scribner var. thinium (A. Hitchcock & Chase) A. Hitchcock & Chase in Robinson, Rho- dora 10: 64. 190 1886, — 28 & Chase) F. Hubbard, Rhodora 14: 172. 1912. Fhb sà shoals be Cl abulorum d & Clark var. thinium (A. "Hitchcock : bie ie & Clark, Ann. Missouri Bot. Gard. 65: 1113. 1978. TYPE: United States. New Jersey: Ocean Co., Toms River, 28 July 1906, Chase 3577 (holotype, US). Panicum oweniae Bicknell, Bull. Torrey Bot. Club 35: l Apr. 1908. TYPE: United States. hute Nantucket Co., 1907, Bicknell s.n. (holotype, NY; isotypes, NY, JS). Panicum — ae Ashe var. .. ola A. erani & Chase in Robinson, Rhodor 6 May 1908. TYPE: United States. District of Columbia: 28 June 1904, Chase 2400 (holotype, US). Panicum olivaceum A. Hitchc Natl. Herb. 15: 225. Verapaz: Coban, Feb. 1888, Tire kheim 428 (ho- lot US). PUN ificum A. Hitchcock & Chase, Contr. U.S. Natl. Herb. 15: 229. 1910. TYPE: United States. California: Shasta Co., Castle Crag, % mi. E of hotel, 3 Aug. 1908, Hitchcock 3070 (holotype, US). Panicum lindheimeri Nash var. septentrionale Fernald, Rhodora 23: 227. 1922 (“1921”). Panicum lanu- os Elliott var. septentrionale (Fernald) Fer- nald, Rhodora 36: 77. 1934. TYPE: Canada. New Brunswick: St. John River, Woodstock, 14 July 1916, Fernald & Long 12527 (holotype, GH). Panicum deamii A. Hitchcock & Chase in Deam, Indiana Conserv. Dept. Publ. 82: 284, Aa 7518). ue TYPE: United es Indiana: e Co., 44 mi Pine, 21 June 1926, Deam 43287 (holotype, US). Panicum mundum Fern ald, Rhodora 38: 392, pl. 443(1- 5). 1936. TYPE: United States. Virginia: Sussex Co., l. of Homeville, 25 Aug. 1936, Fernald & Long 6499 (holotype, GH; isotypes, MO, N US Panicum brodiei H. Saint John, Fl. S.-e. Washington 51. TYPE: United States. Washington: Whitman Co., Wawel, Snake River, June 1893, Brodie s.n. Panicum lassenianum Schmoll, Madrono = 95. 1939. E: United States. California: Plumas Co., Devil's RR Hot Spring Valley, pese 4082 (holotype, US). Panicum glutinoscabrum Fernald, Rhodora 49: 122. ~ TYPE: United States. Virginia: Nansemond , ca. Vó mi. W of Kilby, 8- 12 Sep. 1946, Fernald et Pa 15186 (holotype, GH; isotype, PH). This taxon is common in north Florida south to the central peninsula, occurring primarily in pine flatwoods. Dichanthelium acuminatum (Swartz) Gould & Clark var. densiflorum (Rand & Redfield) Gould & Clark, Ann. Missouri Bot. Gard. 65: 1127. 1978. Panicum nitidum Lamarck var. vaio Rand & Redfield, Fl. Mt. Desert 174. 1894. Panicum acuminatum Swartz var. "ue sas (Rand & Redfield) Lelong, Brit- tonia 36: 270. 1984. TYPE: United States. Maine: Hancock Co., Mt. Desert, Ripples Pond, 28 July 1892, Rand s.n. (holotype, GH). Volume 75, Number 4 1988 Hansen & Wunderlin 1643 Dichanthelium in Florida Panicum spretum Schultes, Mant. 2: 248. 1824. Di- Mensa m (Schultes) Freckmann, Phy- tologia 48: 1981. TYPE: United States: “N. Anglica," Muhlenberg Herb. Panicum no. 37 (ho- lotype, Panicum eatonii i Nash, Bull. Torrey Bot. Club 25: 84. 898. TYPE: United States. New Hampshire: Rock- ngham Co., Seabrook, 1897, Eaton s.n. (holotype, NY: isotype, US). Panicum octonodum J. G. Smith, U.S.D.A. Div. Agrost. Il. 17: 73. 1899. Panicum nitidum Lamarck var. oc tonodum (J. G. Smith) Lamson- uir & Merrill, iv. Agrost. Bull. 24: 34. 1901. TYPE: United States. Texas: Waller tu. 5 May 1898, Thurow 6 (holotype, US). Panicum paucipilum Nash, Bull. Torrey Bot. Club 26: 573. Nov. 1899. TYPE: United States. New Jersey: ape May Co., Wildwood, 30-31 May 1897, Bick- nell s.n. (holotype, NY; isotypes, NY, US). This weak variety is apparently rather uncom- mon in Florida, found only along the northern edge of the state. Dichanthelium acuminatum var. lindheim- eri (Nash) Gould & Clark, Ann. Missouri Bot. Gard. 65: 1127. 1978. Panicum lind- heimeri Nash, Bull. Torrey Bot. Club 24: 196. 1897. Panicum lindheimeri ae var. typ- icum Fernald, Rhodora 23: . 1922 (^1921"), nom. inadmiss. cond lanugi- nosum Elliott var. lindheimeri (Nash) Fer- nald, Rhodora 36: 77. 1934. Dichanthelium lindheimeri (Nash) Gould, Brittonia 26: 60. 1974. Dichanthelium lanuginosum (Elliott) Gould var. lindheimeri (Nash) Freckmann, Phytologia 39: 270. 1978. Panicum acu- minatum Swartz var. lindheimeri (Nash) Bee- tle, Phytologia 48: 193. 1981. TYPE: United tates. Texas: Comal Co., banks of the Guade- lupe River, Near New Braunfels, 1846, Lind- heimer 565 (holotype, NY; isotypes, MO, NY). Panicum funstonii Lamson-Scribner & Merrill, U.S.D.A. Div. Agrost. Circ. 35: 4. 1901. TYPE: dicke: States. i ank of eah River at Three Rivers, 26 July 1891, Coville & Funston 1286 (holotype, US). In Florida, this variety is confined to the north- ern counties. Variety lindheimeri represents the more glabrous extremes of D. acuminatum. Dichanthelium boscii (Poiret) Gould & Clark, Ann. Missouri Bot. Gard. 65: 1101. 1978. Panicum boscii Poiret in Lamarck, Encycl. Suppl. 4: 278. 1816. TYPE: United States: Bosc s.n. (lectotype, P; isotype, US, fragm.). Lectotypified by Hitchcock & Chase (Contr. U.S. Natl. Herb. 15: 317. 1910). Panicum waltheri Poiret i in Lamarck, Encycl. Suppl. 4: Michaux s.n. (lectot 788. Not Linnaeus, 1753. Pan icum eil ive Poir. in Lam. Encycl. Suppl. 4: 282. . Not Pursh, 1814. Panicum e lium var. ene Vasey, Bull. Bot. Div., U.S. Dept. Agric. 8: 33 [error for 34] 1889." If, as interpreted here, Panicum wa typified by the type of P other hand, P. waltheri Poir. is considered a ty- pographic error for walteri, and is therefore a later homonym of P Nash. This was effectively done by Hitch- cock & Chase (Contr. U.S. Natl. Herb. 15: 317. 1910), who cited P. porterianum as a no be valid or not, P. porter- ianum is always disposed as its homotypic synonym. Panicum Vir | iis " australe Vasey, U.S.D.A. Div. Bull. 34. 1889. TYPE: 2 States. Ko Clark ie Thomasville, 16 888, Mohr s.n. (holotype, US) Panic um rao Linnaeus var. molle Vasey, U.S.D.A. Agrost. Bull. 8: 34. 1889. Panicum waltheri Poiret var. molle (Vasey) Le Bull. Torrey Bot. Club 20: 194. 1893. Panicum boscii Poiret var. molle (Vasey) A. Hitcheock E Chase in Robinson, Rhodora 10: 64. States. District of Columbia: Ward s.n. (holotype, US) This species is rather uncommon in Florida, occurring in the panhandle and south to Levy County. m: " 1 1 (T; A h Ll Brittonia 26: 59. 1974. Panic Ind en . Chasea clandestina (Lin- naeus) Nieuwland, Amer. Midl. Naturalist 2: 64. 1911. TYPE: United States. Pennsylvania: Kalm s.n. (lectotype, LINN 80.57). Lecto- typified by Hitchcock & Chase (Contr. U.S. Natl. Herb. 15: 312. 1910). Panicum pedunculatum Torrey, Fl. N. Middle United States 141. 1824. Panicum clandestinum Linnaeus var. pedunculatum (Torrey) Torrey, Fl. New York 1644 Annals of the Missouri Botanical Garden 2: 426. 1841. TYPE: United States. New York: “On the island of New-York," Aug., Torrey s.n. (holo- type, NY; isotype, US, fragm.). Panicum decoloratum Nash, Bull. Torrey Bot. Club 26: 570. 1899, TYPE: United States. Pennsylvania: Bucks Co., Tullytown, 30 May 1899, Bicknell s.n. (ho- lotype, NY; isotype, US). An uncommon species in Florida, found only in the western panhandle. It occurs in moist to wet sandy soil in woods. Dichanthelium commutatum (Schultes) Gould, Brittonia 26: 59. 1974. Panicum nervosum Muhlenberg ex Elliott, Sketch Bot. S. Carolina 1: 122. 1816, non Lamarck 1797. Panicum commutatum Schultes, Mant. 2: 242. 1824 Panicum polyneuron Steudel, Syn. Pl. Glu- mac. 1: 91. 1854, nom. illegit. TYPE: United States. “Car. et Geor.,” Elliott Herb. s.n. (lectotype, CHARL). Lectotypified by Hitch- cock & Chase (Contr. U.S. Natl. Herb. 15: 303. 1910) Panicum lng Lamarck var. majus Pursh, Fl. Amer. Sept. 1814. TYPE: United States: Pursh s.n. muss K). Lectotypified gt Hitchcock & Chase (Contr. U.S. Natl. Herb. 15: Panicum umbrosum LeConte ex Torrey, Cat. Pl. New York 91. 1819, non Retzius 1786. TYPE: United Essex Co., Bloomingdale, Le- Conte s.n. A gi uie NY). Lectotypified by Hitch- as & Chase (Contr. U.S. Natl. Herb. 15: 301. TN UTEM pi Panic. 230. 1826. TYPE: United States. “Am. ` Enslin s.n. (lectotype, LE). Lectotypified by Hitchcock & Chase (Contr. U.S. Natl. Herb. 15: 304. 0). aS leiophyllum Fournier, Mexic. Pl. 2: 20: 1886, Nees von Esenbeck 1829. TYPE: c e Cordovensi," Jan., Bourgeau minor Vase ra commutatum Schultes var. ^ . 1889, as “mi- A. Div. Agrost. Bull. 8: 34 norus." Panicum ashei G. Pearson ex Ashe, J. Eli- sha Mitchell Sci. Soc. 15: 35. 1898. Panicum com- mutatum Schultes var. ashei (C. Pearson ex Ashe) Fernald, Rhodora 36: 83. 1934. TYPE: United States. South Carolina: Aiken Co., Aiken, 1867, Ravenel s.n. (lectotype, NY). Lectotypified by Hitchcock & Chase (Contr. U o Natl. Herb. 15: 304. 1910). Panicum ashei is a new name for Vasey's variety and should be uut as a homotypic synonym, not retypified with another element as did Hitchcock & Chase (1910). anro es Vasey, U.S.D.A. Div. Agrost. Bull. 8: 31. 1889, TYPE: United States. Louisiana: East Baton aton Rouge, 1 Oct. 1885, Joor lub 20: Mou Aug e le US). l due b Hitchcock & Chase =° (Contr. U.S. Natl. Herb. 15: 304. 1910). The lec- totype is a sheet without data; collection data come from the protologue. Panicum manatense Nash, Bull. Torrey Bot. Club 24: 897. TYPE: United States. Florida: Manatee Co., NE of Palmetto, 21 Aug. 1895, Nash 2428a (holotype, NY). Panicum egutiateraie Lamson-Scribner, U.S.D.A. Div. 2 1674 (lectotype, US). Lectotypified by Hitch- cock & Chase (Contr. U.S. Natl. Herb. 15: 3 1910). Panicum commelinifolium Ashe, J. dp Pig wa Sci. Soc 9. 1898, non Rudge . Panicum curranii Ashe, J. Elisha Mitchell A cnn 15: 113. le TYPE: United States. Georgia: DeKalb Co., ear ican US). Lectotypified Y Hitchcock & Chase ontr. U.S. Natl. Herb. 15: Panicum Ae n Nash, Bull. Torrey Bot. Club 26: 571. 1899, TYPE: United States. Florida: Lake Co., Eustis, 12-31 Mar. 1894, Vash 45 (lectotype, NY). Lectotypified by Hitchcock & Chase (Contr. U.S. Natl. Herb. 15: 310. 1910). Panicum subsimplex Ashe, North gg ze: Exp. a. Bull. 175: 115. 1900. TYP el States. ME Newcastle Co., near Wilmi ngton, 16 Aug. , Commons s.n. (holotype, NCU). hus ise Lamson-Scribner & J. G. Smith ex sh in Small, Fl. S.E. U.S. 103, 1327. 1903. TYPE: United States. Mississippi: Harrison Co., Biloxi, 1896, Tracy 3074 (holotype, Panicum hintonii Swallen, Contr. U.S. Natl. Herb. 29: 419. 1950. exico. México: Bejucos, Te mascaltepec, 8 Nov. 1932, Hinton 2527 (holotype, US). Dichanthelium commutatum is very common throughout Florida. It is found primarily in moist, shaded areas. Dichanthelium dichotomum (Linnaeus) Gould, Brittonia 26: 59. 197 mum Linnaeus, Sp. Pl. 58. 1753. TYPE: United States. Virginia: “Habitat in Virginia,” Clay- ton 458 (lectotype, BM; isotype, US, fragm.). Lectotypified by cre & Chase (Contr. U.S. Natl. Herb. 15: 190. 1910). Lelong (1984) erred by citing the iin at LINN. 4. Panicum dichoto- Panicum nitidum Lamarck, Tabl. Encycl. 1: 172. 1791. ichotomum Linnaeus var. nitidum (La- TYPE: d n “E. Carolina," Fraser s.n. (ho- o , Panicum "odorum a Encycl. 4: 744. 1798. Panicum dichotomum Linnaeus var. nodiflorum ar eiie Cat, Pl. Cub. 234. 1866. TYPE: United States. bea xai Fraser s.n. (holotype, -LA; isotype, US, E Panicum barbulatum Michaux, Fl. Bor.-Amer. 1: 49. . Panicum dichotomum Linnaeus var. bar- bulatum (Michaux) Alph. Wood, Class-Book Bot., ed. 1861. 786. 1861. Panicum pubescens La- Volume 75, Number 4 1988 Hansen & Wunderlin 1645 Dichanthelium in Florida marck var. barbulatum (Michaux) Britton, Cat. Pl. New Jersey 280. 1889. Panicum nitidum Lamarck var. barbulatum (Michaux) Chapman, Fl. South. U.S., ed. 3. 586. 1897. TYPE: Canada: “Hab. in Canada P. capillari affine. Ad ripas amnis: Riviere a Jacques Cartier dicti od Michaux s.n. (lecto- type, P-M; isotype, US, fragm.). Lectotypified d Hitchcock & Chase (Con. U.S. Natl. Herb. 0). Panicum microcarpon Muhlenberg ex Elliott, Sketch Bot. S. Carolina 1: 127. 1816. TYPE: United States. Georgia: Baldwin s.n. (lectotype, CHARL). Lecto- typified r EET & Chase (Contr. U.S. Natl. Herb. 15: Panicum UE LeConte ex Torrey, Cat. Pl. New 91. 1818, non Elliott 1816. TYPE: unknown. Panicum tremulum Sprengel, Neue Entd. 2: 103. 1821. TYPE: United States: Muhlenberg Herb. s.n. (lec- totype, B [destroyed]; isotype, US, fragm.). Lecto- typified by Hitchcock & Chase (Contr. U.S. Natl. Herb. 15: 190. 1910). Panicum nitidum Lamarck var. barbatum Torrey, Fl. . Middle United States 146. 1824. TYPE: unknown. Panicum nitidum Lamarck var. ramulosum Torrey, Fl. Middle United States 146. 1824. Panicum di- long, Same 36: 265. New Jersey: near Quaker Bridge, June 1818, Torrey s.n. rr Panicum dumus Deane: Opusc. Sci. Phys. Nat. 88. 831. TYPE: United States? e hera calidii Desvaux Herb. s.n. (holotype, P). Panicum ia um Linnaeus var. oe Vasey, U.S.D. . Agrost. Bull. 8: 30. 1889. TYPE: United hatos Mississippi: ane Co., Lake, Tracy 127 (lectotype, US). Lectotypified by es & Chase (Contr. Rs . Natl. Herb. 15: . 1910). Panicum dichoto Linnaeus var. y U.S.D.A. Div. rel. Bull. 8: 30. 1889. Panicum nitidum Lamarck var. viride (Vasey) Britton, Trans. New York Acad. Sci. 9: . Panicum ra- mulosum Michaux m voe (Vasey) Porter, Bull. Torrey Bot. Club 2 1893. TYPE: United ene District of i “Woodley Park, 1881, Ward s.n. Auge ype, US). Lectotypified by Hitch- cock & Chase (Contr. U.S. Natl. Herb. 15: 191. Panicum nudicaule Vasey, U.S.D.A. Div. Agrost. Bull. 8: 31. 1889. TYPE: United States. Florida: Santa Rosa Co., May 1886, Curtiss 3583 (lectotype, US; isotypes, NY, TAES). Lectotypified by Hitchcock & Chase (Contr. U.S. Natl. Herb. 15: 179, 1910). Panicum dichotomum Linnaeus var. commune S. Watson Coulter in A. Gray, Manual, ed. 6. 633. 1890. TYPE: unknown Panicum pericia? Nash, Bull. Torrey Bot. Club 22: 422. 5. TYPE: United States. Florida: Columbia Co., Lake City, 29-31 Aug. 1895, Nash 2500 (lectotype, NY; isotypes, NY). Lao ai by Hitchcock & Chase (Contr. U.S. Natl. Herb. 15: 199. 1910). Panicum annulum Ashe, J. Elisha Mitchell Sci. Soc. 15: 898. Panicum bogueanum Ashe, J. Elisha Mitchell Sci. Soc. 16: 85. 1900, nom. illegit. TYPE: United States. District of Columbia: 1882, Ward s.n. (lectotype, US). Lectotypified by Hitchcock & Chase (Contr. U.S. Natl. Herb. 15: 185. 1910). Panicum — Ashe, J. Elisha Mitchell Sci. Soc. 15: 7. 1898. Panicum dichotomum Linnaeus var. lu- bla (Ashe) Lelong, Brittonia 36: 265. 1984. TYPE: bordering Lake Mattamuskeet, ] une 1898, Ashe s.n. e 2: bi xar) y^ Pap & Chase (Co erb. 15: Panicum. maculatum p J. EB Satu Sei Soc. 1898, non Aublet 1775. Panicum Jadki- nense S uie J. Elisha Mitchell Sci. Soc. 16: 85. Panicum dichotomum Linnaeus var yod- kinense (Ashe) Lelong, Brittonia 36: 266. 1984. PE: United States h Carolina: Wake Co., Raleigh, May 1895, totypified by Hitchcock & Ch Herb. 15: 195. 1910). Panicum mattamuskeetense Ashe, J. Elisha Mitchell Sci. 15: 45. 1898. Panicum dichotomum Linnaeus var. mattamuskeetense (Ashe) Lelong, Brittonia 36: 265. 1984 United States. North Carolina: ake Mattamuskeet, 10 June-6 July 1898, Ashe s.n. (lectotype, US; isotype, NY). Lec- totypified e Hitchcock & Chase (Contr. U.S. Natl. He 6. 1910). Ashe s.n. (lectotype, US). Lec- ase (Contr. U.S. Natl. rb. 15: Panicum nes Ashe, J. Elisha Mitchell Sci. Soc. : 1898. Panicum dichotomum Linnaeus var. dedu (Ashe) Lelong, Brittonia 36: 265. 1984. T Carolina: Dare Co., s Lectotypified by Hitchcock & Chase (Contr. U.S. Natl. Herb. 15: 196. 1910). Panicum clutei Nash, Bull. Torrey Bot. Club 26: 569. 18 (Nash) Fernald, Rhodora 39: 386. 1 United States. New Jersey: Tuckerton to Atsion, 3- 6 July 1899, Clute s.n. «bens. NY) Panicum curtivaginum Ashe, J. Elisha Mitchell Sci. Soc. 16: 85. 1900. TYPE: United States. Mississippi: Jack- son Co., Petit Bois Island, 8 May 1898, Tracy 4584 (lectotype, NCU?; isotype, US). Lectotypified by Hitchcock & Chase (Contr. U.S. Natl. Herb. 15: 196. 1910). Specimen at NCU not located. Panicum taxodiorum Ashe, J. Elisha Mitchell Sci. Soc. 1900. TYPE: United States. vd i casieu Par., Lake Charles, Sep. 1898, Macken 460 (lectotype, NCU; Et NY). Lectotype by Hitchcock & Chase (Contr. U.S. Natl. Herb. 1 198. 1910). Both Lelong (1984) und Gould & dE (1978) cited the “‘holotype” as residing at NY. Hitchcock & Chase, by contrast, stated that the type was in Ashe’s herbarium, which is now at NCU. Panicum multirameum Lamson-Scribner, U.S.D. x e Agrost. Circ. 19: 2. 1900. TYPE: Mexico. Ve Near Jalapa, 1889, Pringle 7882 fects. Us, isotype, MO). Lectotypified by Hitchcock & Chas (Contr. U.S. Natl. Herb. 15: 185. 1910) Panicum wai digna Lamson-Scribner & Merrill, Div. ost. Circ. 29: 9. . TYPE: United Scien Elbow Herb. s.n. (holotype, CHARL). Panicum gravius A. Hitchcock & Chase, Rhodora 8: 1906. TYPE: United States. Delaware: New- ¿asilo Co., on the old Commons farm, between Centreville and Mt. Cuba, 30 n 1906, Chase 3620 (holotype, US). Panicum caerulescens Hackel ex A. Hitchcock, Contr. U.S. Natl. Herb. 12: 219. 1909. TYPE: United States. 1646 Annals of the Missouri Botanical Garden Florida: Dade Co., Miami, 3 Apr. 1906, Hitchcock 706 (holotype, US; isotype, Panicum lucidum Ashe var. opacum Fe rnald, Rhodora 39: 386. 1937. TYPE: United States. Virginia: Prince George Co., N of Gary Church, 25 Aug. 1936, Fernald & Long 6484 (holotype, GH; isotypes, MO, US = Gould & Clark (1978) included the generally cushion-forming plants with small spikelets as va- rieties of D. dichotomum. We agree with Lelong (1984, 1986) that these should be separated from D. dichotomum, but instead of two species each with two varieties, as one species with three vari- eties: D. ensifolium vars. ensifolium, unciphyl- lum, and breve. This leaves D. dichotomum con- sisting of the larger, more diffuse plants with spikelets longer than 1.5 mm. Lelong (1984, 1986) recognized seven varieties of P. dichotomum in this group, based on leaf size and pubescence forms. The Florida material shows too much overlap in these characters for satisfactory separation. This species is common throughout Florida, oc- curring in both dry and moist habitats. Dichanthelium ensifolium (Baldwin ex Elliott) sould, Brittonia 26: anicum en- sifolium Baldwin ex Elliott, Sketch Bot. S. Carolina 1: 126. 1816. Panicum nitidum La- marck var. ensifolium (Baldwin ex Elliott) Va- sey, U.S.D.A. Div. Agrost. Bull. 8: 29. 1889. Dichanthelium dichotomum (Linnaeus) Gould var. ensifolium (Baldwin ex Elliott) Gould Clark, Ann. Missouri Bot. Gard. 65: 1119. 1978. TYPE: United States. Georgia: Baldwin s.n. (holotype, CHARL; isotypes, PH, US, fragm.) e recognize three varieties of Dichanthelium ensifolium in Florida la. Lower internodes strigose; leaves less e 2 N AETAT . breve lb. Lower internodes glabrous to slightly beni lent; most leaves wider than 2 r 2a. ee pubescent; leaves often s white margin - unciphyllum 2b. Spikelets glabrous to ‘sparingly bee ent; leaves without white margins .. var. ensifolium Dichanthelium ensifolium (Baldwin ex Elliott) Gould var. ensifolium Panicum chamaelonc ‘he Trinius, m Panic. 242. 1826. : United States. “Am. ` Enslin s.n. (ho- lotype, LE; isotype, US, iiem) Panicum nitidum Lamarck var. minus Vasey, Tont; S I Florida: St. Johns Co., . Augu Dine: Apr. 1860. Canby s.n. (lectotype, p adum by E cock & Chase (Contr. U.S. Natl. Herb. 15: 1910). dixeris baldwinii Nuttall ex Chapman, Fl. South. U.S., 3. 586. 1897. TYPE: United States. Florida: Baldwin s.n. (lectotype, PH). Lectotypified by Hitchcock & Chase (Contr. U.S. Natl. Herb. 15: Panicum brittonii Nash, Bull. Torrey Bot. Club 24: 194. 1897. TYPE: United States. New Jersey: Ocean Co., Forked River, 29 May-2 June 1896, Britton s.n. (holotype, NY). Panicum glabrifolium Nash, Bull. Torrey Bot. Club 24: 196. 1897. Dichanthelium dichotomum (Linnaeus) Gould var. glabrifolium (Nash) Gould & Clark, Ann. Missouri Bot. Gard. 65: 1120. 1978. TYPE: United States. Florida: Hillsborough Co., Tampa, 20 Aug 1895, Nash 2415a (holotype, NY). Panicum cuthbertii Ashe, J. Elisha Mitchell Sci. Soc. 15: 48. 1898. us United States. South Carolina: Helena Island, xad Cuthbert n. (holoty UE isotype, US, fra Panici 'um glabrissimum Ashe, J. Elisha Mitchell Sei. Soc. 15: 62. 1898. Panic Fá shallotte Ashe, J. Elisha Mitchell d S. 16: 84. 1900, nom. illegit. TYPE: United States. North ene Dare Co., Manteo June 1 Ashe s.n. (lectotype, US; mo NY, US). Lectotypified by Hitchcock & Chase (Contr. U.S. Natl. Herb. 15: 265. 1910). Panicum curtifolium Nash, Bull. Torrey Bot. Club 2 569. Panicum ensifolium Baldwin ex Elliott var. qui pi (Nash) 1984. TYPE: United States. Mississippi: Jackson Co., E i Springs, 2 May 1898, Nash 4598 (holotype, NY; isotype, US). Panicum an Nas h, Bull. Torrey Bot. Club 26: 571. 1899, TYPE: United States. Alabama: Lee Co., Au- burn, 7 May 1898, Earle & Baker 1532 (holotype, NY; isotype, US). Panicum flavovirens Nash, Bull. Torrey Bot. Club 26: 572. 9. TYPE: United States. Florida: Lake Co., "along the edge of road leading to the ford near the T. & K. . bridge across the Wekiva river," near Eustis, 26 30 June 1895, Nash 2061 (holo- e, Panicum austromontanum Ashe, J. Elisha Mitchell Sel. Soc. 1900. TYPE: United States. Alabama Jackson Co., Sand Mountain, June 1899, Ashe s.n. Co US). Lectotypified by Hitchcock & Chase (Contr. U.S Natl. p n 267 ud um | iculatu e, J. Elisha Mitchell I: . 1900. TYPE: S United States. North C E of Jacksonville, 20 = sra a c. . 16: 87. ui. Onslow Co., chcock & € atl. Herb. 15: 265. 1910) Panicum vernale A. Hitchcock & Chase, Contr. U.S Natl. Herb. 15: 266. 1910. TYPE: United States, Florida: Columbia Co., Lake City, 16 Apr. 1906, Hitchcock 1020 (holotype, US; isotype, ISC). ase (Contr. U.S This variety is very common throughout Florida, occurring in seeps, bogs, and wet pinelands. Dichanthelium ensifolium (Baldwin ex Elliott) Gould var. breve (A. Hitchcock & Chase) B. F. Hansen & Wunderlin, comb. nov. BASIONYM: Panicum breve A. Hitchcock & Chase, Contr. U.S. Natl. Herb. 15: 271. 1910. Dichan- Volume 75, Number 4 1988 Hansen & Wunderlin 1647 Dichanthelium in Florida thelium dichotomum (Linnaeus) Gould var. breve (A. Hitchcock & Chase) Gould & Clark, Ann. Missouri Bot. Gard. 65: 1120. 1978. Panicum chamaelonche 'Trinius var. breve (A. Hitchcock & Chase) Lelong, Brittonia 36: 267. 1984. TYPE: United States. Florida: Mar- tin Co., “Jensen,” 5 1906, Hitchcock 4 (holotype, US). This variety is endemic to the white sand scrub vegetation of central Florida. Dichanthelium ensifolium (Baldwin ex Elliott) Gould var. unciphyllum (Trinius) B. F. Han- sen & Wunderlin, comb. nov. BASIONYM: Pan- icum unciphyllum Trinius, Gram. Panic. 242. 1826; and the autonym created by P. unci- phyllum Trinius var. implicatum (Lamson- Scribner) Lamson-Scribner & Merrill 1901. Panicum acuminatum Swartz var. unciphyl- lum (Trinius) Lelong, Brittonia 36: 269. 1984. TYPE: North America, without data (lectotype, LE). Lectotypified by Hitchcock & Chase (Contr. U.S. Natl. Herb. 15: 259. 1910). For comments concerning the mistypification by Lelong (1984), see under D. acuminatum var. implicatum. Panicum tenue Muhlenberg, Descr. ee 118. 1817. anicum liton Schultes, Mant. 2: 250. 1824, nom. illegit. Panicum macrum Kunth, dm eee 1: . 1829, nom. illegit. Dichanthelium dichotomum Éionseus) Gould var. tenue no Gould & uri Bot. Gard. 6 — P O fragm.). A by es & Chase (Contr. U.S. Natl. Herb. 15 259. Panicum alhomarginatun Nash, Bull. Torrey Bot. Club 24: 4 897. TYPE: ern States. Florida: Lake Co., near fes June 894, Nash 925 (holotype, NY; isotypes, MO, "U S). Panicum trium Nash, Ball Torrey Bot. Club 26: gos 1899. TYPE: United States. Georgia: Bibb Co., E River Swamp below Macon, 18-24 us 1895, Small s.n. (holotype, NY; isotype, US). Panicum gracilicaule Nash in Small, Fl. S.E. U.S. 98, 27. 1903, non Rendle 1899. Panicum concin- nius A. Hitchcock & Chase, Contr. U.S. Natl. Herb. 15: 263. 1910. TYPE: United States. Alabama: Jack- son Co., Sand Mountain, 1900, Harbison 2415 (holotype, NY). Lelong (1984, 1986) applied the epithet unci- phyllum to a variety of unciphyllum is certainly the earliest epithet for Lelong's variety, there is a good deal of doubt that it belongs with D. acuminatum. Hitchcock & Chase (1910) lectotypified Panicum unciphyllum with material now at LE. The LE material apparently has two collections on one sheet. According to . acuminatum. ile Hitchcock & Chase, the collection chosen as lec- totype is conspecific with Panicum tenue, while the other collection (by Enslin) matches Panicum columbianum, neither of which belongs in D. acu- minatum. Without having seen the specimen, Le- long (1984) cited as holotype (not lectotype) of Panicum unciphyllum the Enslin collection at LE, this without explanation of the rejection of the earlier Hitchcock & Chase lectotypification. With- out such explanation and barring further discov- eries concerning the type, the lectotypification of Hitchcock € Chase must be followed. Therefore, Panicum unciphyllum is once again considered conspecific with Panicum tenue, which in turn is placed in Dichanthelium ensifolium. It is unfor- tunate that unciphyllum also turns out to be the earliest varietal epithet for what has been called Panicum tenue, this adding to the confusion. This variety often approaches D. dichotomum in general aspect but is better classified with D. ensifolium on the basis of spikelet size. It is com- mon throughout Florida, occurring in wet pinelands and woods. Dichanthelium erectifolium (Nash) Gould & ‘lark, Ann. Missouri Bot. Gard. 65: 1105. 1978. Panicum sphaerocarpon Elliott var. floridanum Vasey, U.S.D.A. Div. Agrost. Bull. : . 1889. Panicum erectifolium Nash, Bull. Torrey Bot. Club 23: 148. 1896. Pan icum floridanum (Vasey) Chapman, Fl. South. U.S., ed. 3. 585. 1897, non Trinius 1835. TYPE: United States. Florida: “‘Moist pine bar- rens, Mosquito Inlet,” May 1879, Curtiss 3599 (lectotype, US; isotype, MO). Lectotyp- ified by Hitchcock & Chase (Contr. U.S. Natl. Herb. 15: 256. 1910). This species is closely related to D. sphaero- carpon but readily separable on the basis of spikelet size. It is quite common throughout most of Florida, except for southern Florida and the Keys, in moist flatwoods and meadows. Dichanthelium laxiflorum (Lamarck) Gould, Brittonia 26: 60. 1974. Panicum laxiflorum Lamarck, Encycl. 4: 748. 1798. Panicum dichotomum Linnaeus var. laxiflorum (La- marck) Beal, Grasses N. Amer. 2: 139. 1896. TYPE: United States?: (holotype, P-L). Lamarck Herb. s.n. Panicum xalapense Kunth, a herr Bonpland & nth, Nov. Gen. Sp. | . 1816. TYPE: Mexico. Veracruz: near Jalapa, . & Bonpland s.n. (holotype, P). Panicum ruprechtii Fournier, Mexic. Pl. 2: 21. 1886, 1648 Annals of the Missouri Botanical Garden non Fenzl 1854. TYPE: Mexico. Veracruz: Jalapa, Galeotti 5733 (holotype, BR). Panicum pyriforme Nash, Bull. Torrey Bot. Club 26: 579. 1899. TYPE: PA prt Florida: Lake Co., 1894, Nash 239 (ho- lotype, NY; isotypes, MO, NY, US Panicum xalapense Kunth var. strictirameum A. Hitch- Chase, Contr. U.S. Natl. Herb. 15: 161. 1910. Panicum laxiflorum Lamarck var. strictira- meum (A. Hitchcock & Chase) Fernald, Rhodora c 34. TYPE: United States. Mississippi: Hinds Co., Jackson, 28 Apr. 1906, Hitchcock 1311 (ho- lotypes, US). Dichanthelium laxiflorum is common through- out Florida except for Dade and Monroe counties. It is usually found in moist woods. Dichanthelium leucothrix (Nash) Freckmann, Phytologia 58: 101. 1981. Panicum leuco- thrix Nash, Bull. Torrey Bot. Club 24: 41. 1897. Panicum acuminatum Swartz var. leu- cothrix (Nash) Lelong, Brittonia 36: 271. 1984. TYPE: United States. Florida: Lake Co., near Eustis, July 1894, Nash 1338 (holotype, NY; isotypes, NCU, NY, TAES, US). Panicum os Bosc ex Roemer & Schultes, Syst. Veg. 2 TYPE: United States. uie (lectotype, isotype, B-W). Lec by Hitchcock & Chas (Contr. U.S. Natl. Herb. 15: Panicum minutulum Desvaux, oe usc. E Phys. Nat. 87. 1831, non Gaudichaud-Beaupre 1826. TYPE: United States?: Desvaux Herb. s.n. (holotype, P). Panicum parvispiculum Nash, Bull. T Zub 24: 3 897. TYPE: United St oo sh Co., Darien Junction, 25-27 Ju all s.n. (holotype, NY; isotype, US, fragm.). ae w dc a dr -Scribner, U.S.D.A. Div Agrost. B , f. 4. 1898. d acuminatum rds Gould & Clark var t- ianum (Lamson- Poer Gould & Clark, i Mis- souri Bot : 1126. 1978. Dichanthelium w o (Lamson Scribner) Freckmann, Phy- tologia 48: 1981. TYPE: Cuba: Wright 3463 (holotype, De Panicum Pi e Nash, Bull. Torrey Bot. Club 26: 574. 2 1899. Dichanthelium acuminatum DARE Gould & Clark var. cs eed ips Gould & Clark Missouri Bot. Gard. 6 1978. Dic hanthelium iongiligulatum m Freckmann, Phytologia 48: 981. Panicum acuminatum Swartz var. Dcos (Nash) Le- 4. TYPE: United States. : Franklin Co., ‘Apalachicola, 1892, Vasey s.n. ace NY). Dichanthelium leucothrix is the small-fruited member of the D. acuminatum complex. Robust, glabrous specimens may be difficult to separate rom D. acuminatum var. lindheimeri, which has larger, fewer-fruited inflorescences and a more northerly distribution. There may be some justifi- cation for the separation of the glabrous (tradi- uonally Panicum longiligulatum) and pubescent (P. leucothrix) specimens at the varietal level; most previous authors have indeed split this taxon into at least two taxa based on forms of pubescence. The pubescence, however, intergrades completely from the extremes, while otherwise the taxon is quite homogeneous. Therefore, we have chosen not to recognize the segregates. This species is common throughout Florida, mostly in pine flatwoods. Dichanthelium oligosanthes (Schultes) Gould, Brittonia 26: l anicum pauciflo- rum Elliott, Sketch Bot. S. Carolina 1: 120. 1816, non R. Brown 1810. Panicum oligo- santhes Schultes, Mant. 2: 256. 1824. Pan- icum scoparium Lamarck var. pauciflorum Lamson-Scribner, Tennessee Agric. Exp. T cock & Chase (Contr. U.S. Natl. Herb. 15: 285. 1910). Panicum macrocarpon Torrey, Fl. N. Middle United States 143. 1823, non LeConte 1819. TYPE: United States. : Franklin Co., banks of the Connecticut River, near Deerfield, Cooley s.n. (ho- otype, NY). Panicum scoparium Lamarck var. angustifolium Vasey, U.S.D.A. Div. Agrost. Bull. 8: 32. 1889. TYPE: United States. South Carolina: Ravenel s.n. (lecto- type, US). Lectotypified by Tee & Chase E U.S. Natl. Herb. 15: 286 Panicum scribnerianum Nas sh, Bull. Duda Bot. Club 22: 421. 1895. Panicum oligosanthes Schultes var. Rhodora 36: 80. Schultes) Gould scribnerianum (Nash) Fernald, ichanthelium oligosanthes A Co., Wysox, July Side A SES & - Natl. Herb. 3. 1910). Gould & Clark (1978) erroneously pus that n name is based on Pan- icum macrocarpon Panicum helleri Nash, Bull. Torrey Bot. Club 26 1899. Panicum oligosanthes Schultes var. heller (Nash) Fernald, Rhodora 36: 80. 1934. TYPE: United States. Texas: Kerr Co., Kerrville, 14-21 May 1894, Heller 1759 (holotype, NY). Panicum pernervosum Nash, Bull. Torrey Bot. Club 26: 576. 9. TYPE: United States. Texas: Harris Co., Món: 16 Apr. 1872, Hall 830 (holotype, NY). Most recent authors have followed Fernald in recognizing P. scribnerianum at least at the vari- etal level. The characteristics of pubescence used to separate the taxa, however, are very unreliable and seem not to be correlated with any other con- Volume 75, Number 4 1988 Hansen & Wunderlin Dichanthelium in Florida 1649 crete evidence that differentiation has occurred in this species. There seems to us little to be gained by the formal recognition of these pubescence phas- es. This species is rather uncommon in the northern tier of counties, occurring in dry pine-oak-hickory woods Dichanthelium ovale (Elliott) ied : Clark, Ann. Missouri Bot. Gard. 65: 1978. Panicum ovale Elliott, Sketch Be * Carolina 1: 123. 1816. TYPE: United States. Georgia: Camden Co., St. Marys, Baldwin s.n. (holo- type, CHARL) Panic P Pia secon Sketch Bot. S. Carolina 1: 124. 1 n Lam rck 1791. Panicum consangui- thelium consanguineum (Kunth) Gould & Clark, Ann. Missouri Bot. Gard. 65: 1115. 1978. TYPE: United States: Elliott Herb. s.n. (holotype, CHARL; isotype, US, fragm.). Panicum ciliiferum Nash, Bull. Torrey Bot. Club 24: 195. 1897. TYPE: United States. Florida: Lake Co., Eustis, 12-31 Mar. 1894, Nash 147 (holotype, NY; isotypes, MO, US Panicum malacon Nash, Bull. Torrey Bot. Club 24: 197. 7. TYPE: United States. at a Lake Co., Eustis, NY; isotypes, Panicum georgianum Ashe, J. Elisha Mitchell Sci. Soc. 15: 36. 1898. Panicum cahoonianum Ashe, J. Eli- sha Mitchell Sci. Soc. 15: 113. 1899, nom. illegit. TYPE: United States. Georgia: McIntosh Co., Darien Junction, 27 June 1895, Small s.n. (lectotype, US). Lectotypified by Hitchcock & Chase (Contr. U.S. Natl. Herb. 15: 169. 1910). Panicum strictifolium Nash, Bull. Torrey Bot. Club 26: Mee 899, TYPE: United States. Florida: Lake Co., Eustis, 3 May 1894, Nash 603 (holotype, NY; Sieb NY, U Panicum erythrocarpon Ashe, J. Elisha Mitchell Sci. Soc 9 0. TYPE: United States. North Carolina: New Hanover Co., 1 mi. N of Wilmingto on, 19 May 1899, Ashe s.n. (holotype, NCU; isotype, US). (Shinners) Freckmann, pee ES 271. 1978. o b E tates. Wisconsi s Co., 12 SE of Adams, Shinners & Shaw 4415 poda WIS nid GH, MIL, MIN, US). As defined here, Dichanthelium ovale consists of the large-fruited specimens of the D. acumi- natum complex. If smaller-fruited plants are in- cluded in this taxon, as in past treatments, problems in naming the many overlapping specimens become insurmountable. By delimiting the species as we ave, a more homogeneous and practical system results, one where almost every specimen can be classified. Gould & Clark (1978) recognized Dichanthe- lium consanguineum as a separate species char- acterized by pilose upper leaf surfaces. Again, the pubescence character seems insufficient for the separation of species Dichanthelium ovale is common in sandhills, pinelands, and disturbed habitats throughout Flor- ida. Dichanthelium portoricense (Desvaux ex Hamilton) B. F. Hansen & Wunderlin, comb. nov. BASIONYM: Panicum portoricense Des- vaux ex Hamilton, Prodr. 11. 1825. TYPE: Puerto Rico: Desvaux Herb. s.n. (holotype, P). — lancearium E Gram. Panic. 223. 1826. : United States: “Am. e ” Enslin s.n. (ho- A dd LE; beca US, fra dico um ied. yllum Bosc. ex e. Nees von Esenbeck ı Martius, Fl. Bras. Enum. 2(1): 227. Snel non Sprengel 1822. TYPE: a States: Bosc s (holotype Panicum w ATAR : Nash, Bull. Torrey Bot. Club 23: 4 6. TYPE: United States. Florida: Lake Co. Eustis, 16-31 May 1894, Nash 781 (holotype, NY: isotypes, MO, NY, US). Pa a um E Lamson-Scribner, U.S.D.A. Div. . Bull. 7: 18, f. 60. 1897. Dic :hanthelium Scribner s.n. (lectotype, US). Lectotypified by Hitch- ase (Contr. U.S. Natl. Herb. 15: 247. iir o nashianum Lamson-Scribner, U.S.D.A. Div. ost. Bull. 7: 79, f. 61. 1897. Panicum porto- Er ense Desvaux ex Hamilton var. nashianum (Lam son-Scribner) Lelong, Brittonia 36: 267. 1984. TYPE: United States. Florida: Lake Co., vicinity of Eustis, 15-30 Apr. 1894, Nash 466 (lectotype, US; iso- type, MO). Lectotypified by Hitchcoc Chase (Contr. U.S. Natl. Herb. 15: 273. 1910). Panicum tsugetorum Nash, Bull. Torrey Bot. Club 25: 86. York: Bronx Eustis, 12-31 Mar. 1894, Nash 72 (holotype, NY: isotypes, MO, N Panicum psammophilum Nash, Bull. Torrey Bot. Club 26: 576. Nov. 1899, non Welwitsch July 1899. TYPE: United States. New Jersey: Ocean Co., Tom's River, 25-31 July 1898, Clute 175 (holotype, NY; isot ype Panicum ditinat Ashe, J. Elisha Mitchell Sci. Soc. 1650 Annals of th Missouri an Garden 16: 88. Feb. 1900. TYPE: United States. North Carolina: Onslow Co., near Ward's Mill, 11 mi. E of Jacksonville, 19-21 May 1899, Ashe s.n. (lec 1910). eas (1984) cited the EE as ome at UNC, but Hitchcoc clearly stat E that the type is in Ashe's hei now at Panicum pauciciliatum Ashe, J. Elisha Mitchell Sei. Soc. 16: 87. Fe d TYPE: api States. North Carolina: New Har ar Wilmington, 20 May 1899, Ashe s.n. lado NC U; isotype, — by Hitchcock & Chase (Contr. U. S. Herb. 15: 272. 1910). Specimen at NCU not iso um Ke is ime Lamson-Scribner var. patulum on-Scribner & Merrill, U.S.D.A. Div. Agrost. : 9. Dec. 1900. Panicum dod ue Scribner & Mes ill) A. Hitchcock, iin 8: 209. 1906. Panicum lancearium Trini ar. piod (Lamson-Scribner & Merrill) Fernald, "Rhodora 36: 80. 1934. Dichanthelium sabulorum (Lamarck) (Lamson-Scribner & ¿Florida Man- atee uui 3 Sep. 1898, ifs 1296 Ced US). a O pa As pointed out by Lelong (1984), Panicum sa- bulorum Lamarck, whose type is from U Ly ruguay, is not conspecific with our material, contrary to the treatment by Gould & Clark (1978). This leaves as the earliest name in the taxon Panicum por- Desvaux ex Hamilton, which unfortu- nately necessitates a new combination in Dichan- thelium. Gould & Clark (1978) and Lelong (1984, 1986) recognized two varieties in this species, based on toricense spikelet length, inflorescence size, and leaf length. In the Florida material, however, spikelet size is but poorly correlated with inflorescence size, both of which seem unrelated to leaf size. Infraspecific taxa have therefore not been applied to this species. Panicum columbianum was classified by Lelong (1984) as a variety of P. acuminatum. According to the description of Hitchcock & Chase (1910), P. columbianum and its synonym (fide Lelong) P. tsugetorum have ligules up to 1.5 mm long. The short ligules take P. columbianum out of the P. acuminatum complex. Dichanthelium portoricense is probably the most commonly collected member of the genus in Flor- ida. It occurs in a variety of habitats throughout the state. Dichanthelium ravenelii (Lamson-Scribner & Merrill) Gould, Brittonia 26: 60. 1974. Pan- icum ravenelii Lamson-Scribner & Merrill, U.S .D.A. Div. Agrost. Bull. 24: 36. 1901. TYPE: United States. South Carolina: Elliott s.n. (lectotype, ARL). Lectotypified by Hitchcock & Chase (Contr. U.S. Natl. Herb. 15: 287. 1910). Panic um sc n mx var. major Vasey, U. S. D.A. . Bull. 8: 1889, as "majus." TYPE: Unied States South ae Ravenel s.n. (holo- pe, US). SEE , This large grass, related to D. scoparium, is known in Florida only from the panhandle region, occurring in dry hammocks. Dichanthelium scabriusculum (Elliott) Gould & Clark, Ann. Missouri Bot. Gard. 65: 1110. 1978. Panicum scabriusculum Elliott, Sketch Bot. S. Carolina 1: 121. 1816. Panicum vis- cidum Elliott var. scabriusculum (Elliott) Beal, Grasses N. Amer. 2: 143. 1896. TYPE: United States. Georgia: Chatham Co., Baldwin s.n. (holotype, CHARL) Savannah, Panicum lanuginosum ient ex Sprengel, Syst. Veg. 1: 319. 1825, non Elliott 1816. Panicum eriophorum wes es ex Kunth, E Enum: Pl. 1: 128. 1833. TYPE: ted States. co. Bosc s.n. (holotype, ques Panicum n nealleyi asey, Bull. Torrey Bot. Club 13: 2 2 TYPE: United States. Texas: May 1885, Nea! s.n. (holotype, Panicum dichotomum Linnaeus var. elatum Vasey, U.S.D.A. Div. Agrost. Bull. 8: 31. 1889. TYPE: United States. Alabama: Mobile Co., 18 June 1888, Mohr sn. d US). Lectotypified T che cock ase (Contr. U.S. Natl. Herb. 15: 298. Panicum CEPIT Ashe, N. Carolina Agric. Exp. S : 115. 1900. TYPE: United States. North Carolina: Johnston Co., Wilsons Mill, 15 July 1897, Ashe s.n. (lectotype, NCU; isotypes, NCU). Lectotypified by Hitchcock & Chase (Contr. U.S. Natl. Herb. 15: 299. 1910). Panicum aculeatum A. Hitchcock & Chase, Rhodora 8: 209. 1906. TYPE: United States. District of Colum- bia: bari Park, 27 July 1904, Chase 2520 (ho- lotype, US). Banus rec cognitum de E Rhodora 40: 331, pl. 497- 498. 19 =: United States. N ` , Lo e poi PH). Gould & Clark (1978) erroneously ted the location of the holotype as US. This species is quite common in northern Flor- ida, ranging southward to Orange County. It is primarily found in bogs, cypress swamps, and wet ods. Dichanthelium scoparium (Lamarck) Gould, Brittonia 26: 60. 1974. Panicum scoparium Lamarck, Encycl. 4: 744. 1798. Panicum scoparium Lamarck var. genuinum Lamson- Volume 75, Number 4 1988 Hansen & Wunderlin Dichanthelium in Florida 1651 Scribner, Tennessee Agric. Exp. Sta. Bull. 7 48. 1894, nom. inadmiss. TYPE: United States: “Caroline,” Michaux s.n. (holotype, P-M). Hitchcock & Chase (Contr. U.S. Natl. Herb. 15: 287. 1910) mistypified var. genuinum with Elliott’s material of P. ravenelii. Lamson- Scribner’s citation of P. scoparium Lamarck necessitates the obvious disposal of this in- admissible varietal epithet here. j deos que cens Lamarck, Encycl. 4: 748. 1798. s cum laxiflorum Lamarck var. pubescens (La- rck) Cha apman, Fl. South. U.S., ed. 3. 586. 1897. lia pubescens (Lamarck) Nin land, Amer. Midl. Naturalist 2: 64. 1911. TYPE: United States. South Carolina: Michaux s.n. (lectotype, P-M). Lec- totypified by Hitchcock & Chase (Contr. U.S. Natl. Herb. 15: Panicum s. Elliott, Sketch Bot. S. Carolina 1: 123, P. Ae ). 1816. TYPE: United States: Elliott Her n. (lectotype, CHARL). Lectotypified by hase (Contr. U.S. Natl. Herb. 15: igo nitidum Lamarck var. velutinum Doll in Mar n Bras. 2(2): 247. 1877. TYPE: United States. Although this species is usually locally abundant, it is confined in Florida to the northern tier of counties. It occurs in marshes, moist woods, and roadsides. Dichanthelium sphaerocarpon (Elliott) Gould, Brittonia 26: 60. 1974. Panicum sphaero- carpon Elliott, Sketch Bot. S. Carolina 1: 125. 1816. Panicum dichotomum Linnaeus var. diee deat ira eae "e Bi. Class- Book Bot., 1861 : microcarpon Mie var. pon (Elliott) Vasey, Grass. U.S. TYPE: United States. Georgia: Baldwin s.n. (holotype, CHARL). . Panicum anus ar- 12. 1883 Panicum kalmii Swartz, Adnot. Bot. 1829. TYPE: Un ited States. Pennsylvania?: Kalm s.n. (holotype, Panic 2 nitidum Lamarck var. crassifolium A. eh ex Doll in Martius, Fl. Bras. 2(2): 247. 1877. D States. New Je KR?; isotype, pura j ages vicarium Fournier, Mexic. Pl. 2: Mexico. eh Cordis: s.d., 285 (holotype, P). Panicum inflatum Lamson-Scribner € J. G. Smith, U.S.D.A. Div. Agrost. Circ. 16: 5. 1899. Panicum sphaerocarpon Elliott var. inflatum (Lamson-Scrib- ner & J. G. Smith) A. Pipe in Hitchcock & edd ET US. Herb. 15: 253. 1910. Apr States. Misciss sippi: Harrison Co., Bi- loxi Oct. 898, Tracy 4622 (holotype, US). Panicum missippense Ashe, J. Elisha Mitchell Sci. Soc. 16: 91. 1900. TYPE: United States. Louisiana: ersey: Gray? 30 bee pore Orleans Par., “Banks of the Mississippi River below New Orleans in October," Ashe s.n. (holotype, not located). This species occurs in bogs and seeps in the panhandle area of Florida. Dichanthelium strigosum (Muhlenberg ex Elliott) Freckman, Brittonia 33: 457. 1981. Panicum strigosum Muhlenberg ex Elliott, Sketch Bot. S. Carolina 1: 126. 1816 TYPE: United States. “Hab. in humidis. Car: & Georg:," Elliott Herb. s.n. (lectotype, CHARL; isotype, US, fragm.). Lectotypified by Hitchcock & Chase (Contr. U.S. Natl. Herb. 15: 164. 1910). Freckmann (1981b) recognized that strigosum was the earliest specific epithet for this species and made the necessary combinations for the three rather well marked (for this genus) varieties. KEY TO THE VARIETIES OF DICHANTHELIUM STRIGOSUM la. Spikelets puberulent or dnce 1.5-2.1 mm long r. leucoblepharis lb. Spikelets glabrous, 1.1-1.8 mm bn eaf blade surfaces glabrous; nen 1.2- on at x da 2b. Leaf blade surfaces pilose; spikelets : 5 mm long var. bis Dichanthelium strigosum (Muhlenberg ex Elliott) Freckmann var. strigosum Panicum Mr Yes Lamarck var. pubescens Vasey, Contr . Natl. Herb. 3: 30. 1892. Dichanthe- lium ine a UR (Trinius) Gould & Clark var. pubescens (Vasey) Gould & Clark, Ann. Missouri Bot. Gard. 65: 1101. 1978. Panicum ciliatum Elliott var. pubescens (Vasey) Freckmann ex R. Pohl, Fieldiana, Bot., n.s. 4: 356. 1980. Panicum leucoblepharis Trinius var. pubescens (Vasey) Beetle, Phytologia 48: 192. 1981. TYPE: United States. Florida: Duval Co., Curtiss H (lectotype, n egi dns: by Hitchcock & Chase (Contr. atl. Herb. 15: 164. 1910). Panicum pebei te Lamson-Scribner, Tennes- see Agric. Exp. Sta. Bull. 7: 53, pl. 16(61). 1894. TYPE: de nited Sue Tennessee: White Cliff Springs, July , Lamson-Scribner s.n. (lectotype, US). in dca, by ps & Chase (Contr. U.S. Natl. Herb. 15: 164. 0). Uncommon in Florida, this variety ranges from northern Florida south to the central peninsula. Dichanthelium strigosum (Muhlenberg ex Elliott) Freckmann var. glabrescens (Grise- bach) Freckmann, Brittonia 33: 457. 1981. Panicum dichotomum Linnaeus var. gla- brescens Grisebach, Fl. Brit. W. 1.553. 1864. 1652 Annals of the Missouri Botanical Garden Dichanthelium leucoblepharis (Trinius) Gould & Clark var. glabrescens (Grisebach) Gould & Clark, Ann. Missouri Bot. Gard. 65: 1100. 1978. TYPE: Jamaica: Purdie s.n. (holotype, K). ee pore pas iue Bull. Torrey Bot. Club 24: : United gro Florida: Hillsbor- e o. 20 Au 895, Nash 2420a o NY; isotype, US, E This variety is common throughout Florida in moist pinelands, bogs, and coastal swales. Dichanthelium strigosum (Muhlenberg ex Elliott) Freckmann var. leucoblepharis (Tri- nius) Freckmann, Brittonia 33: 457. 1981. Panicum leucoblepharis Trinius, Clav. Agrostogr. Antiq. 234. 1822. Dichanthel- ium leucoblepharis (Trinius) Gould & Clark, Ann. Missouri Bot. Gar : 1099, 1978 TYPE: United States. “Am. bor ba s.n. (lectotype, LE). Lectotypified by Hitchcock Chase (Contr. U.S. Natl. Herb. 15: 162. 1910). Panicum dedu Elliott, Sketch Bot. S. Carolina 1: 126. n Marcklin 1792. Panicum ciliatifolium "Révis. Gramin. 1: 36. 1829. : United . Elliott Herb. s.n. (holotype, CHARD). Baa um m ciliatifolium Desvaux, "wasa Sci, Phys. Nat. 88. 1831, non Kunth 1829. "America bo- reali" (specimen unknown) Kunth, When making the new combination Panicum ciliatum Elliott var. pubescens (Vasey) Freckmann ex R. Pohl (1980), based on P. laxiflorum var. pubescens Vasey (1892), Pohl created the auto- nym Panicum ciliatum Elliott var. ciliatum. At first glance, this autonym would seem to be the earliest varietal name for this taxon. However, Panicum ciliatum Elhott, 1816, is illegitimate be- cause it is a later homonym of Panicum ciliatum Marcklin, 1792, as pointed out by Veldkamp (1976). Since an autonym can only be established on a legitimate name (ICBN Art. 26), as verified by Dan Nicolson (pers. comm.), the next epithet in line, leucoblepharis, must be chose This variety is rare in Florida, found prm in the northern counties, occurring in pinelands. EXCLUDED SPECIES Dichanthelium linearifolium (Lamson-Scribner) ould This northern species was reported from the Florida panhandle by Clewell (1985), but no sup- porting specimens have been found. LITERATURE CITED CLEWELL, A. F. 1985. Guide to the Vascular Plants of the Florida Panhandle. University Presses of Florida / Florida State Univ. Press, Tallahassee, Florida FRECKMANN, R. W. 1981a. Realignments in the Di- chanthelium acuminatum complex (Poaceae). Phy- i p 99-110. 81b. The correct name for Dichanthelium ment (Poaceae) and its varieties. Brittonia -458. GouLD, F. W. € C. A. CLARK. 1978. y ne vet (Poaceae) in the United States and Canada. Missouri Bot. Gard. 65: 1088-1132. HrrcHCOCK, A. S. & A. CHASE. 1910. The North Amer- ican species of Panicum. Contr. U.S. Natl. Herb. 15: 1-396. LELONG, M. G. 1984. New combinations for Panicum subgenus Panicum and subgenus Dichanthelium (Poaceae) of the southeastern United States. Brittonia 36: 262-273. l A taxonomic treatment of the genus Panicum (Poaceae) in Mississippi. Phytologia 61: 251-269. SPELLENBERG, R. W. 1975a. Autogamy and hybridiza- tion as evolutionar chanisms in Panicum sub- genus Dichanthelium (Gramineae). Brittonia 27: 87- 95. 1975b. Synthetic hybridization and taxonomy of w western North American Dichanthe un group Lanuginosa (Poaceae). Madrono 23: 134-153. VELDKAMP, J. F. 1976. Panicum ciliatum B (Gramin- eae) ve to be salid P. leucoblepharis Trin. Taxon 25: C R. P. Guide to the Vascular Plants of Central Florida. U niv. South Florida Press/Uni- versity Presses of Florida, Gainesville, Florida. INDEX Chasea angustifolia (Elliott) Nieuwland, p. 1639. = D. aciculare clandest ina en Nieuwland, p. 1643. = D. clan cana Esqui) Nieuwland, p. 1651. = D. sco- parium Dichanthelium aciculare (Desvaux ex Poiret) Gould & Clark, p. 1639. acuminatum (Swartz) Gould & Clark, p. 1640. var. densiflorum ( (Rand & Redfield) Gould & Clark, p. 16 var. r. fasciculatum ey a p. 1640. = D. var. t implicatum (Lamson: Seribner) Siri u ban P. 164 acumin var. dud: (Nash) Could & Clark. 1643. var. t (Nash) Gould & Clark, p. 1648. — D. leucothri var. Turow (Lamson Scribner & J. G. i a n: & di p. 1641. — D. acuminatum var. na var. villosum (A. Gray) shai be Pq p. 1640, = minatum var. acu var. urightianum nien Scribner) Gould & Clark, p. 1648. = D. leucothri Volume 75, Number 4 1988 Hansen & Wunderlin 1653 Dichanthelium in Florida angustifolium (Elliott) Gould, p. 1639. — D. aciculare boscii (Poiret) Gould & Clark, p. 1643. clandestinum (Linnaeus) Gould, 3. columbianum (Lamson-Scribner) Freila p. 1649. ortoricense commonsianum di ed p. 1641. = D. acu- minatum var. ac atum var. cuchlamo eun (Shane) Freckmann, p. 1649. = D. o Bed “(Schultes) Gould, p. consanguineum (Kunth) Gould & ead p. 1649. = D. ovale dichotomum (Linnaeus) Gould, p. var. breve (A. deci ida & Pee Gout & Clark, p. 164 6. = D. ensifolium var. breu var. vides aa (Baldwin ex Elliott) Gould & Clark, p. 1646 . ensifolium var. ensifolium var. eslabriolium Ni now & Clark, p. 1646. sifolium v var. tenue (Muhlenberg) Could E Clark, p. 1647. nsifolium var. uncip aan. (Baldwin ex Elliot) Could, p. 1646. ock & Chase) B. F. Hansen & u var. Wenn ae (Trinius) B. F. Hansen & Wunderlin, 47. erectifolium (Nash) Gould & Clark, p. 1647 la ij acit raga Gould, p. 1640. = D. — var. t fasciculatum (Torrey) pute p. 1640. = D. acuminatum var. acumina var. pup Maru Freckmam, p. 1643. = D. minatum var. lindhei acu var. villosissimum (Nash) Could, p. 1640. — D. acu- minatum var. d iu uh laxiflorum (Lamarck) Gould, p. leucoblepharis (Trinius) Gould & Clack, p. 1652. = D. strigosum var. leucoblepharis var. glabrescens (Grisebach) Gould & Clark, p. 1652. = D. strigosum var. rescens var. Hp (Vasey) Gould & Clark, p. 1651. — D. trigosum NOn (Nash) Freckmann, p. 1648. lindheimeri (Nash) Gould, p. 1643. = D. acuminatum v indheimeri longiligulatum (Nash) Freckmann, p. 1648. — D. leu- cothrix po (Ashe) MUS p. 1641. = D. acumi- Rh (Schultes) Gould, p var. scribnerianum (Nash) Baa s 1648. = D. oli- gosanthes ovale (Elliott) Gould & Clark, p. 1649. var. addisonii (Nash) Gould & Cul, p. 1641. = D. acuminatum var. acuminatu portoricense (Desvaux ex Hamilton) B. F. Hansen & Wunderlin, p. 1649. ravenelii (Lamson-Scribner & Merrill) Gould, p. 1650. sabulorum (Lamarck) Gould & Clark var. patulum ra — = e Gould & Clark, p. 1650. — D. portor var. thinium (A. Hitclicack & Chase) Gould & Clark, . 1642. = D. acuminatum var. acuminatum scabriusculum (Elliott) Gould & Clark, p. 1650. scoparium (Lamarck) Gould, sphaerocarpon (Elliott) Gould, » 1651. spretum i — p. 1643. = D. acumi- natum var. dens — (Mubilenbesg-« ex x Elliott) Freckmann, p. 1651. ar. glabrescens (Grisebach) Freckmann, p. 1651. r. leucoblepharis (Trinius) Freckmann, p. 1652. villosissimum (Nash) Freckmann, p 40. = D. acu- atum var. acuminatum var. apo (A. Hitchcock & d Freckmann, 1642. — D. acuminatum var. acuminatum E P TO (Lamson-Scribner) Freckmann. p. 1648. = D. leucothrix Milium clandestinum (Linnaeus) Moench, p. 1643. = D. clan- destinum Panicum ird Desvaux ex Poiret, — D. aciculare aci . 1639. ar. arenicoloides (Ashe) Beetle, p. 1639. = D. a ulare var. 2 ew (Lamson- pee & J. G. Smith) Beetle, P. — D. acic aculeatum A. Ta iran & “Chase, p. 1650. = D. sca- culu acuminatum Swarts, p. 1640. = D. acuminatum var. acuminatum var. OPI (Lamson-Scribner) Lelong, p. 1649. = D. p var. t densiforum and E er. e p- nsifloru var. fasciculum. Corey) Lelong, z 1640. = D. uminatu var. r, implicatum Lamson Serine) Beetle, p. 1641. 1642. var. encor (Nash) bou » "1648. — D. leuco- var. lindheimeri (Nash) Beetle, p. 1643. — D. acu- minatum var. lindheimeri var. M dup (Nash) Lelong, p. 1648. = D. leucothrix Var. t wnciphyllum oe ccs p. 1647. = D. nsifolium var. uncip var. r. villosum (A. Gra 5) Beetle, p. 1640. = D. acu- min var. acuminatu edison Nash, p. 1641. = D. acuminatum var. acu- E eaa yen p. 1642. = D. acuminatum var. acu- minatum albemarlense Ashe, p. 1641. = D. acuminatum var. acuminatum albomarginatum Nash, p. 1647. = D. ensifolium var. unciphyllum angustifolium Elliott, p. 1639. = D. aciculare T LeConte ex Torrey, p. 1645. — D. di- otomum "n Ashe, p. 1645. — D. dichotomum end ola Ashe, p. 1 lare ashei G. Pearson ex Ashe, p. 1644. = D. commutatum atlanticum Nash, p. 1641. = D. acisibestum var. acu- minatum auburne Ashe, p. 1642. — D. acuminatum var. acu- minatum austromontanum Ashe, p. 1646. — D. ensifolium var. baldwini “Nuttall ex SECURE p. 1646. — D. ensifolium r. ensifoliu 1654 Annals of the Missouri Botanical Garden barbulatum Michaux, p. 1644. — D. dichotomum ar. molle (Vasey) A. Hitchcock & Chase, p. 1643. boscii dies E ee & Chase, p. 1646. — D. ensifolium . bre TM Nas h. P- =D. idol var. gis es brodiei I hee Jn. p. 1642. umina ar. caerulescens Hackel ex A. Hitchcock, p. 1645. = D. oto S ri rim . 1649. = D. ovale icio Trinius, P. 1646. = D. ensifolium var. ensifo var. breve (A. Hitchcock & Chase) Lelong, p. 1647. - nsifolium var. breve chrysopsidifolium. A p. 1642. — D. acuminatum ciliatifolium. Desvaux, p. 1652. = D. strigosum var. leucoblepharis rn 2 p. 1652. = D. strigosum var. leu- blephar diliaiusm Elis, p. 1652. = D. strigosum var. leuco- blephar var. pubescens (Vasey) Freckmann ex R. Pohl, p. 1651. =D. s gee Hie strigosum ciliiferum Nash, p. = D. ovale ciliosum Nash, p. 164 ^n — D. acuminatum var. acu- minatum clandestinum Linnaeus, p. = D. clandestinum var. rd He) Torrey, p. 1643. andesti coni Nash, "i "1645. = D. dichotomum columbianum Lamson-Scribner, p. 1649. = D. portori- cense var. commonsianum (Ashe) cd p. 1641. = D. acu- minatum var. acuminatu var. r. oricola (A. Hitchcock g Chase) Fernald, p. 1642. acuminatum var. acuminatum var. bno (A. Hitchcock & Chase) A. Hitchcock & ase, p. 1642. = D. UND var. acuminatum mutatu commelinifolium Ashe, p. 1644. commonsianum Ashe, p. 1641. — D acuminatum var. acuminatum ipi addisenii je W. Stone, p. 1641. — D. minatum var. acuminatum cu r. var. addisonii (Nash) R. ay p. 1641. = D. acu- minatum var. acumina var. eee (Shinners) R. Pohl, p. 1649. — comitatum Schultes, p. 1644. = D. commutatum G. Pearson ex Ashe) Fernald, p. 1644. — . commutatum var. consanguineum (Kunth) Beal, p. 1649. — D. je var. latifolium Lamson-Scribner, p. 1644. — D. co mutatum r. minor Vasey, p. 1644. — D. commuta comophyllun Nash, E 1642. — D. URL var. ac concinnius ing "Hitchcock & Chase, p. 1647. = D. en- cryptan curranii Ashe, p = p. commutatum curtifolium Nash, p. 1646. = D. ensifolium var. ensi- Pablo Steudel, p. 1639. = D. acicula urtivaginum As ha p. 1645. = D. dichat mum hier Ashe, p. 1646. = D. deor var. v. ensifolium eamii chcock & Chase, p. 1642. mi- at ar. acuminatum de m Nash, p. 1644. — D. clandestinum delawarense Ashe, P € — D. aciculare ] i dichotomum eus, p. 4. = D. m (Niehaus) Alph. Wood, p. 1644 var. commune S. Watson & Coulter, p. 1645. = D. dichotomum var. divaricatum Vasey, p. 1645. = D. dichotomum elatum Vasey, p. 1650. = D. scabriusculum var. ar. fasciculatum m p. 1640. — D. acuminatum var. acumin var. Preis Cue: p. 1651. = D. strigosum var. glabrescens var. laxiflorum (Lamarck) Beal, p. 1647. = D. laxi- florum var. lucidum (Ashe) Lelong, p. 1645. = D. dichoto- mum var. mattamuskeetense (Ashe) Lelong, p. 1645. = D. dichotomum var. nitidum (Lamarck) Alph. Wood, p. 1644. = D. dichotomum var. nodiflorum (Lamarck) Grisebach, p. 1644. = D. dichotomum var. ramulosum (Torrey) Lelong, p. 1645. = D. di- chotomum var. roanokense (Ashe) Lelong, p. 1645. = D. dicho- tomum u var. sphaerocarpon (Elliott) Alph. Wood, p. 1651. — D. sphaerocarpon var. um HN p. 1640. — D. acuminatum var. acumin var. sin Vasey, p . 1645. = D. dichotomum var. E REA (Ashe) Lelong, p. 1645. — D. dicho- dum iiec dum p. 1645. — D. dichot ied Nash, p. 1646. — D. ensifolium var. re eatonii Nash, p. 1643. = D. acuminatum var. densiflo- iia Baldwin ex Elliott, = D. Lir: pin ar. Fit egi (Nash) os j^ "1646. m var. ensi a ium vu. Trinius, p. — D. commutatum epilifolium Nash, p. 1644 = D. commutatum equilaterale Lamson- Scribner, p. 1644. — D. commu- tatum erectifolium Nash, p. 1647 eriophorum Schultes ex Kunth, p. culum erythrocarpon Ashe, p. 1649. — D. ovale euchlamydeum Shinners, p. 1649. — D. ovale filiculme Ashe, p. 1641. = D. acuminatum var. acu- = D. n 650. = D. scabrius- minatum filirameum Ashe, p. 1639. = D. aciculare flavovirens Nash, p. 1646. — D. ensifolium var. ensi- li olium floridanum (Vasey) n p. 1647. = D. Wb pg peu Lamson- one er & Merrill, p. 1643. = D. acuminatum var. lindhein meri fusiforme A. Hitchcock, p. 1639. = D. aciculare Volume 75, Number 4 988 Hansen & Wunderlin Dichanthelium in Florida 1655 georgianum Ashe, p. 1649. — D. o pons Nash, p. 1646. — D. | eae var. en- glabrissimum Ashe, p. 1646. sifoliu glutinoscabrum Fernald, p. 1642. var. acuminatum gracilicaule Dis p. 1647. — D. ensifolium var. en- — D. acuminatum — D. ensifolium var. un- gravius À. Hitchcock & Chase, p. 1645. = omum pore m p. 1641. D. dicho- — D. acuminatum var. helleri Nash, p. 16 = D. oligosanthes Beterophyllum Bose e ex xC. Nees von Esenbeck, p. 1649. = D. portoricense var. thinium (A. Hitchcock & Chase) F. Hubbard, p. 2. = D. acuminatum var. acuminatum hintonii Swallen, p. 1644 huac end jocis p. 1641. uminatum da pacas DN (Torrey) F. Hubbard, p. 1640. acuminatum var. acuminatum var. Seco A. Hitchcock » a p. 1642. = D. a minatum var. acumina = D. commutatum = D. acuminatum var. ac- implicatum Lamson-Scribner, p. m — D. acumina- tum var. acuminatum pu Lamson-Scribner & J. G. Smith, p. 1651. = D. sphaerocarpon joorii Vasey, p. 1644. = D. commutatum kalmii Swartz, p. 1651. — D. SUE lancearium Trinius, p. 9. var. patulum (Lamson- Shar 8 Merrill) Fernald, p. 1650. = D. portoricense pun A. Hitchcock & Chase, p. 1642. — D. acu- natum var. acuminatum lanuginosum Bosc ex Sprengel, p. 1650. = D. scabrius- culum lanuginosum "ud p. 1640. — D. acuminatum var. var. pgs m (Torrey) Fernald, p. 1640. — D. acuminatum var. acuminatum var. huachucae (Ashe) A. eM p. 1641. = D. acuminatum var. acuminatu var. E plata (Lamson- Scribe Fernald, p. cuminatum var. acuminatum var. lindheimeri (Nash) Fernald, p. 1643. — D. acu- minatum var. lindheimeri subvar. € (Ashe) inia p. 1641. = D. acuminatum var. praecocius (A. Hitchcock & Chase) Dore, p. 1642. =D.a 1641. mina var. iun dre (Fernald) Fernald, p. 1642. = D. acuminatum var. acum var. siccanum À. duet E Chase, p. 1642. = D. acuminatum var. acuminatu lassenianum Schmoll, p. 1642. = "D. acuminatum var. acuminatu latifolium Linnae australe s ey, p. 1643. = D. boscii var. clandestinum (Linnaeus) Pursh, p. 1643. — D. clandestinum var. molle Vasey, p. 1643. — D. bos laxiflorum Lamarck, p. 1647. = D. ve var. pubescens (Lamarck) Chapman, p. 1651. = D. scoparium var. pubescens Vasey, p. 1651. = D. strigosum var. strigosum var. strictirameum (A. Hitchcock & Chase) Fernald, . 1648. = D. laxifloru leiophyllum Fournier, P- 1644. = D. commutatum p r = p. 1652. = D. strigosum var. s oblepha ar. pubescens (Vasey) Beetle, p. 1651. s m var. strigosum ms Nash, p. 1648. — D. leucothri. — =e Nash, p. 1643. = D. Sees var. lind- = D. stri- var. ‘fasciculatum PRA iid p. 1640. — D. acuminatum v var. ie a i (Lamson: Series) ens p. 1641. um var. acu nies: meridional Bend Farwell, ii 1641. = D. acuminatum var. atum var. septentrionale Fernald, p. 1642. = D. acumi- natum var. var. eset (Ase) Farwell p. 1641. = D. acu- minatum var. M 1 Fernald, p. "1643. = = D. acuminatum var. lindhei liton hule x 1647. — D. ensifolium var. unciphyl- um longiligulatum Nash, p. 1648. — D. leucothrix longipedunculatum Lamson-Scribner, p. 1651. — D. stri- gosum var. strigosum lucidum Ashe, p. 1645. — D. dichotomu = D. cee es 8 = D. ensifolium var. unci- um . = D. dichotomum p. 1649. — D. ovale 4. — D. commu , p. 1645. = D. dichos omum var. clutei (Nash) Ferdi hp 1645. = D. iiie qug dieras pA p. 1641. = D. acuminatum va var. .albemarlese on ae p. 1641. = D. acu- microcarpon ' Muhlenberg ex - Elliot, p. 1645. = D. di- var. ua (Elliott) Vasey, p. 1651. = D. sphaerocarpon iade ini in p. 1641. = D. acuminatum var. PN din us ue p. 1648. — D. leucothrix mississippiense Ashe, p. 1651. — D. iphaerocathon iios ameum Lamson-Scribner, p. 1645. — D. dicho- P Fernald, p. 1642. = D. acuminatum var. acu- minatum mutabile Lamson-Scribner & J. G. Smith ex Nash, p. 1644. = D. commutatum nashianum Lamson- Scribner, p. 1649. = D. portoricense var. pa Ta Lamson-Scribner & Merrill, p. 1650. = D. portoricense iile Veer p. 1650. = D. scabriusculum nemopanthum Ashe, 9. = D. aciculare nervosum Muhlenberg ex Elliott, p. 1644. = D. com- muta 1656 Annals of the Missouri Botanical Garden Md Grisebach, p. 1639. — D. acicular var. osum Grisebach, p. 1639. — D. MATE nitidum pes rck, p. 1644. = D. dichotomum t. r. angustifolium (Elliott) A. Gray, p. 1639. = D. acic var. ba =s E Torrey, p. 1645. = D. dichotomum var. ne (Michaux) Chapman, p. 1645. — D. dichotomu var. ciliatum Torrey, p. 1640. = D. acuminatum var. acuminatum var. a A. Gray ex Doll, p. 1651. = D. sphaerocarp var. densiflorum ias & oo p. 1642. = D. acuminatum . densifloru var. ensifolium (Baldwin ex Elliott) Vasey, p. 1646. D = D. commutatum var. minus Vasey, p. i = D. fla var. ensifolium var. octonodum (J. G. Smith) Lamson-Scribner & Mer- rill, p. 1643. = D. acuminatum var. densiflorum var. pilosum Torrey, p. 1640. = D. acuminatum var. acuminatum mulosum Torrey, E. eds = p. Rn E var. iens Doll, p. = p. scopar var. villosum A. Gray, s 3200. — D. acuminatum var. acuminatum var. viride (Vasey) aris 1645. = D. paa = D. dichotom nudicaule Vasey, p. 1645. = e dichotomum occidentale Lamson- iion p. 1641. = D. acumina- tum var. acumin octonodum J. - Smith. n "1643, = D. acuminatum var. densifloru oligosanthes Schultes, p. 1648. — D. oligosanthes var. helleri (Nash) Fernald, p. 1648. — D. oligosanthes var. scribnerianum (Nash) Fernald, p. 1648. D oligosanthes EUN dà A. Hitchcock & Chase, p. 1642. = D. acu- minatum var. acuminatum onslowense Ashe, p. 1649. — D. portoricense orangense iugi p. 1641. = D. acuminatum var. acu- mina — m ‘Hitchcock & Chase, p. 1642. | var. acuminatu or Monitum a P. 1639, = E aciculare ovale Elliott, p. 1 í var. rpseudopubescens (Nash) Lelong p. 1641. = D. var. acum var. pesci (A. Gray) Telong, p. 1640. minatum var. acumin ovinum Lamson-Scribner & 1 "G. Smith, p. 1639. = D. — D. acumi- — D. acu- aciculare oweniae Bicknell, p. 1642. — D. acuminatum var. acu- minatum pacificum A. Hitchcock & m p. 1642. — D. acu- minatum var. acuminatu parvipaniculatum Ashe, p. 1646. — D. ensifolium var. ensifolium parvispiculum Nash, p. 1648. — D. anaku as s gu Nash, p. 1649. = D. ricense son-Scribner & Merrill. ^i "Hitchcock, p. = D portoricense e Ashe, p. 1650. = D. portoricense pauciflorum Elliott, p. 1648. = D. oligosanthes 1643 pa Sii p. = D. acuminatum var. den- sifloru Ont Torrey, p. 1643. — D. clandestinum pernervosum Nash, p. 1648. — D. oligosanthes pinetorum Swallen, p. 1640. — D. aciculare polycaulon Nash, p. 1652. — D. strigosum var. gla- brescens polyneuron Steudel, p. 1644. — D. commutatum porterianum Nash, p. 1643. — D. boscii portoricense Desvaux ex Hamilton, p. 1649. — D. por- toricense var. nashianum (Lamson-Scribner) Lelong, p. 1649. = ortoricense praecocius A. Hitchcock & Chase, p. 1642. = D. acu- ar. acuminatum psammophilum Nash, p. 1649. = D. portoricense pseudopubescens Nash, p. 1641. = D. acuminatum var. pou Lamarck T aas scoparium um id dm p. 1645. = D. oto nubium Nash. p. 1643. — D. bos pyriforme Nash, p. 1648. — D. b m ramulosum Michaux var. viride (Vasey) Porter, p. 1645. — D. dichotomum ravenelii Lamson-Scribner & Merrill, p. 1650. = D. ravenelii recognitum Fernald, p. 1650. — D. nem r scabriusculum Elliott, p. 1650. — D. scabriusculum wp Ashe, p. 1641. — D. acuminatum var. natum parias Lamarck, p. 1650. = D. scoparium var. m s 1648. = D. d um Lamson-Scribner, p. 1650. = D. s p var. ipti Vasey, p. 1650. — D. ravenelii var. pauciflorum Lamson-Scribner, p. 1648. — D. oli- hes scribnerianum Nash, p. 1648. — D. oligosanthes setaceum Muhlenberg, p. 1639. — D. aciculare 6. = D. dol var. ensifolium s sphaerocarpon Elliott, p. 1651 phaerocarpon r. floridanum Vasey, p. 1647. = D. m var. inflatum Eee Scribner & J. mith) A Hitchcock, p. 1651. = D. E m sphagnicola Nash, p. 1645, = D. dichotomum ure Me p. 1643. — D. acuminatum var. den- "rietifplium Nash, p. 1649. = D. ovale trictum ex Roemer & Schultes, p. 1648. = D. eucothrix strigosum Muhlenberg ex Elliott, p. 1651. gosum var. strigosum s EI Lamson-Scribner & Merrill, p. 1645. . dichotomum subsimplex Ashe, p. 1644. — D. commutatum subuniflorum Bosc ex Sprengel p. 1639. = D. aciculare m m P- = D. acuminatum var. acu- = D. stri- odo "Ashe, p. 1645. = D. dichotomum tennesseense — p. 1641. = D. acuminatum var. acuminatu tenue Muklenbees: p. 1647. — D. ensifolium var. un- ciphyllum dion Lamson-Scribner & J. G. Smith, p. 1641. — D. minatum var. acuminatum la: Sprengel, p. kag = D. dichotom trifolium Nash, p. = D. inns var. unci- phyllum Volume 75, Number 4 1988 Hansen & Wunderlin 1657 Dichanthelium in Florida tsugetorum Nash, p. 1649. — D. portoricense umbrosum LeConte ex Torrey, p. 1644. — D. commu- tatum unciphyllum Trinius, p. 1647. — D. ensifolium var. unciphyllum var. implicatum (Lamson- Scribner) Lamson-Scribner & Merrill, p. 1641. = D. acuminatum var. acu minatum var. meridionale (Ashe) Lamson-Scribner & zu . 1641. = D. acuminatum var minatu forma pu Edd Scribner & ae P. 1642. =D.a m var. acuminatu forma prostratum yer paier & Merrill p. 1642. = D.a um var. acuminatum var. nium. ^ Hitchcock & "Chase, p. 1642. = D. natum ES. ' Hitchcock & CM. p. 1646. — D. ensifolium var. ensifolium vicarium Fournier, p. 1651. = D. sphaerocarpon B illosissimum Nash, p. 1640. acuminatum — D. acuminatum var. var. pseudopubescens (Nash) TM p. 1641. — D. acuminatum var. acuminatu var. scoparioides (Ashe) Fernald, p. 1641. = D. acu- minatum var. acuminatum villosum Elliott, p. 1649. = D. ovale — Elliott, = D. scoparium r. brad (Elliott) Beal, p. 1650. = D. sca- al um waltheri Poiret, p. 1643. = D. boscii var. molle (Vasey) Porter, p. 1643. webberianum Nash, p. 1649. = D. portoricense lili ia ic Ashe, p. 1642. = D. acuminatum var. acuminatu enanas Lo mom p. 1648. bia leucothrix xalapense Kunth, p. — D. laxi var. strictirameum A “Hitchcock & Chase, p. 1648. = D. boscii = D. laxifloru xanthospermum. jecit Scribner & C. Mohr, p. 1642. = natum var. acuminatum ia jid p. 1645. — D. dichotomum NOTES PASSIFLORA EGLANDULOSA, A NEW SPECIES IN SECTION CIECA (MEDIKUS) DC. FORMERLY INCLUDED WITH P. TRINIFOLIA MASTERS Passiflora trinifolia Masters *'is very common in the forests of the Occidente" of Guatemala ac- cording to Standley & Williams (1961). Indeed, this name has been applied to a locally abundant apetalous passionflower of the wet montane forests of northern Central America. However, while pre- paring a revision of Passiflora L. section Cieca (Medikus) DC., vegetative characters was noted among the spec- imens circulated as P. trinifolia. Consequently, f two distinct species: the actually very rare P. trinifolia, and the more common but previously unnamed species described below. extreme and bimodal variation of closer analysis revealed th Passiflora eglandulosa MacDougal, sp. nov. TYPE. Guatemala. San Marcos: wet mountain forest at Aldea Fraternidad, W-facing slope of Sierra Madre between San Rafael Pie de La Cuesta and Palo Gordo [ca. 14%56'N, 91%52'W], 1,800-2,400 m, 10-18 Dec. 1963, Williams, Molina & Williams 25997 (holotype, F; isotypes, ENCB, G, NY, S, US, W —2 sheets). Figures 1, 2B, 3 Exe ad sectio Cieca pertinens, scandens; stipulae e 2.5-9 mm latae; petioli eglandulosi; folia eglan- dulosa trilobata haud peltata, lobis acuminatis vel caudatis, pedunculi ebractenti petala nulla; coronze filamenta bise- riata, filam exterior ied 3- m longis, usque ad .3 mm Ara filamentis interioribus ad 1.5 mm Wee (niea i api ovarium glabrum; semina re- ticulato-fovea Slender, climbing, perennial herb 2-4 m long, minutely puberulent and sparsely to lightly pubes- cent with trichomes of 2 size classes: smaller (mi- croscopic) trichomes 0.05-0. clavate and antrorsely bent or appressed, present throughout; larger trichomes (0.2-)0.4-0.6(-0.8) O mm, unicellular, ANN. MISSOURI Bor. Garb. 75: 1658-1662. mm, unicellular, cylindrical and pointed, slightly bent antrorsely. Stem perennial with little second- ary growth, terete or subterete but drying strongly obtusely sulcate, glabrescent below, sparsely to lightly pubescent above with trichomes 0.25-0.6 (70.7) mm, pubescence often restricted to one side of stem; posture of shoot apex straight, negatively geotropic. Stipules (3.5-)5-14(-20) x (2.5-)3- = i mm, ovate, with ca. 5-7 veins departing stem, the midvein only slightly off center pss only slightly oblique), apex acute, often apiculate, margins ciliolate-setose. Petioles (0.6-) 1-3.5(-4.6) em, eglandular, slightly canaliculate and adaxially pubescent (at least distally) with tri- chomes (0.2-)0.3-0.5(-0.6) mm, abaxially gla- brescent with only microscopic trichomes. Laminas 2.8-10(-12) x 4.0-15(-17) cm at fertile nodes, with 5 primary veins, 3-lobed ca. /,—¥, the distance to the shallowly cordate base, the lobes triangular to deltate or ovate, long-acuminate to caudate, the angle between the lateral lobes (120-)125-160 (-170Y, the central lobe longest, with the ratio of lateral to central lobe lengths (0.68-)0.75-0.90 (70.95), the ratio of laminar width to length 1.30- 1.65(-1.75), adaxially nearly glabrous or glabres- cent with a few (0.2-)0.3-0.6(-0.8) mm trichomes restricted to the primary veins, abaxially sparsely to lightly puberulent with microscopic trichomes 05-0.06 mm; laminas not variegated (except trace of pericostal whiteness seen on very few leaves at distal flowering nodes of MacDougal 316); lam- inar nectaries absent; seedling and juvenile laminas depressed obovate or narrowly transversely rhom- bic/elliptic, more shallowly 2-3-lobed, the angle between the long-acuminate lateral lobes 105-120? in seedlings and 160-170? in juveniles, central lobe shortest (or reduced to a cusp), with ratio of lateral to central lobe lengths 1.25-2.8, ratio of laminar width to length 2.0-5.7. Tendrils straight 1988. Volume 75, Number 4 1988 Notes 1659 during development at shoot apex, often suppressed on fertile determinate axillary branches. Prophyll of vegetative ramifying bud 1, narrowly ovate, acute. Peduncles (5-)8-19(-2 occasionally solitary, uniflorous, ebracteate. Flow- ers ca. 1.5-2 cm diam.; hypanthium 4-5.5 mm diam. with 5 retrorse spurs 0.4-0.8 mm long be- tween bases of sepals, or sometimes spurs obsolete; stipe (2.0-)3.0-5.0(-8.0) tween horizontally and erect, pale to light yellow- green except as noted, nearly inodorous; sepals (6.5-)7.5-9.0 x 2.3-3.9 mm, broadly lanceolate, rounded at apex, the 2-3 outermost cucullate an with a (0.5-)0.8-1.2 mm blunt subapical cornus, often abaxially flushed with very deep red to pur- plish red; petals absent; filamentous corona in 2 series, the outer filaments ca. 29-31, 3.0-4.0 mm long, 0.2-0.3 mm diam., filiform, widest at base, slightly attenuate distally, reflexed above the mid- dle and the tips often slightly incurved, yellowish green at base, light yellow distally; inner series (0.7-)1.0-1.5 mm, capillary to subclavate, sub- mm, membranous, pli- 3) mm, geminate or mm; flowers borne be- erect; operculum 1.5- cate, sometimes with an inconspicuous narrow pur- plish band near middle, apex white-papillose; nectary trough RIOT pu annulus; limen (disk) not — ] d 3.0-3.6(— 4. 0) mm along androgynophore, the free portions ca. 3 mm, spreading but not perpendicular to androgynophore; anthers 2.2-2.7 mm, oriented perpendicular or nearly so to their filaments at anthesis; ovary 1.5-1.9 x 1.0-1.4 mm, ellipsoid to widely ellipsoid, glabrous; styles ca. 4-5 mm, filiform, typically geniculate above middle; stigma capitellate, 0.7-0.9 mm diam. Fruit 10-13(-16) x 9-13(-14) mm, widely ellipsoid to subglobose or slightly obovoid, pericarp bluish black, glaucous, insipid; arils only half the length of the seed, firm, whitish or grayish, insipid; seeds 4.6-5.0 mm long, (2.9-)3.1-3.5 mm wide, (2.0 -)2.2-2.5 mm thick, reticulate-foveate with (15-)18-21(-24) foveae per side, obovate to widely obovate or subpyriform, the chalazal beak obtuse or obsolete, the micropylar beak obtuse or often somewhat oblong and blunt. Additional specimens examined. EL SALVADOR. AHUACHAPAN: nebelwald region, Cerro Grande de Apa top of Cerro Verde, 30 July 1977, Croat ange (MO); cloud forest, Mountain Cerro Verde, 1,800 0 Fe 1968 (fr), Molina & Montalvo 21514 (F, NY). Hoe SANTA paie ca 88?05' ow. ane m San Miguel and summit of mountain, near upper limits of Finca Caieta [ca. 14°59’N, 89*54'W], 1,600-2,300 m, 10 Feb. 1942, Steyermark 43787 (BR, F). GUATEMALA: Finca Nacional "La Aurora," 1938-1939, Aguilar 89 (F); cerca encino y Pinus maximinoi, Choacorran, km 20 a San Juan Sacatepéquez, 2,000 m, 17 Sep. 1982 (fl), J. Castillo et San Ándimeills. 1,700 m, 26 Sep. 1972 (fl, & Molina 27543 (ENCB, F, US); ravine near e Canales, 1,900 m, 25 Jan. 1947 vas 2 & Molina 11822 (F). JALAPA: Volcán N of Jalapa, 1,300-2,200 CG); moist forest at and above Aguas Amargas, slopes of Volcán de Zunil, 2,430-2,850 m, 17 Feb. 1939, Standley 65404 (F); wet hillside forest, Aguas Amargas, western slope of Volcán de Zunil, 2,450 m, 14 Jan. 1941 (8), Standley 83336 (F); damp thicket along road above Santa Maria de Jesüs, ca. 1,680 m, 25 Jan. 1941 (fr), Standley 84846 (F, US); densely forested white sand quebrada, El Pocito, S of San Martin Chile Verde on road to Colomba, 2,200 m, 27 Jan. 1941, Standley 84997 (F, G— 2 sheets); damp dense mixed forest on white sand slopes above Mujuliá, between San Martin Chile Verde and Colomba, 1,800 m, 1 Feb. 1941, Standley 85571 F); thickets on slopes and ridges between Quebrada Chi- charro and Montana Chicharro, on SE-facing slopes of Volcán María, 1,300-1,400 m, 18 Jan. 1940 (fl, fr), Steyermark 34360 (F); western slopes of Volcán Zunil, opposite Santa María de Jesús, 1,500 m, 21 Jan. 1940 (fr), Steyermark 35094 (F). SAN MARCOS: wet cin quebrada, Barranco Eminencia, road between San Mar- cos and San Rafael Pie de la Cuesta, in upper part of the barranca between Finca La Lucha and Buena Vista, 2,500- 2,700 m, 6 Feb. 1941, Standley 86368, 86379 (F); thickets in pine woods in flat below cliffs along Rio Ma- & W of town of Tajumulco, NW slopes of Volcan Tajumulco, 2,300-2,380 m, 26 Feb. 1940, Steyermark 36663 (F, US); montane um forest on outer slopes of Tajumulco Volcano, n Marcos, ca. 2,300 m, 31 Deo E. 1964 n Williams et al. 26864 (F, GH, NY). SUCHI- TEPEQUEZ: Volcán Santa Clara, between Finca El Naranjo and upper slopes, 1,250-2,650 m, 23 May 1942, Stey- ermark 46628, 46692 (F, US). zacaPa: cloud forest in ravine bordering Quebrada Alejandria, summit of Sierra de las Minas, vicinity of Finca Alejandria, 2,500 m, 13 Oct. 1939 (seedling), Steyermark 29859 (F) — The following common names are recorded from herbarium specimens: “granadilla de culebra” Guatemala); “granadilla” “Hoja de murciélago,” (Quezaltenan- Guatemala, “flor de murciélago,” go, Guatemala). Specimens of P. eglandulosa were not collected until after Killips 1938 monograph, so the de- scription there of P. trinifolia applies strictly to Masters's species. The description of P. trinifolia in Standley & Williams (1961), however, is a com- posite drawn from Killip and their observations of P. eglandulosa. Passiflora eglandulosa is super- ficially similar to the poorly known P. trinifolia by 1660 Annals of the Missouri Botanical Garden FIGURE l. Passiflora eglandulosa (from MacDougal 316 except as noted) .— A. Habit. —B. Leaf from juvenile plant. "e Stipule (Clewell & Hazlett 3858) .— D. Stipule.—E. Seed (Standley 84846) . having similarly three-lobed leaves and unusually broad stipules but is easily distinguished even in the herbarium by the absence of petiolar and lam- inar nectaries. Additionally, at fertile nodes the leaves of P. eglandulosa always have the central lobe longest; the laminas resemble those of the sympatric Oreopanax sanderianus. The leaves of P. trinifolia commonly have the central lobe short- est at lower fertile nodes. Leaves of juvenile plants are transversely bilobed in both species but are occasionally peltate in P. trinifolia, resembling miniature leaves of P. coriacea A. L. Juss. Th leaves of juveniles are never at all peltate in P. eglandulosa. Volume 75, Number 4 1988 Notes 1661 A ea R laI at "mE AD A. Flower of Passiflora trinifolia from type locality (MacDougal 637GR) .—B. Flower of Passiflora 6). eglandulosa ( MacDougal 31 Living material of both species was collected in the field by the author and grown at Duke Uni- versity, allowing detailed comparison of the flowers (Fig. 2). Passiflora eglandulosa differs notably in having flowers oriented above rather than near or below the horizontal plane; buds slightly horned at the apex; sepals proportionally narrower; outer co- ronal filaments much finer and broadest at the base; inner coronal filaments not broadly capitate; limen smaller and unspotted; more gracile androecium d gynoecium; and anthers that present pollen distally to laterally instead of subproximally. This anther orientation is unusual in the section and genus as a whole, and may be associated with a mode of pollination different than that of the other species in the section. The habits and habitats of the two species are remarkably different, and they are not sympatric (Fig. 3). Passiflora eglandulosa climbs to around four meters in shady ravines and at the edges of wet premontane to montane broad-leaved forest on the volcanic cones of southern Guatemala to central Honduras. The chartaceous leaves are bright green adaxially and usually exhibit drip tips. In contrast, P. trinifolia is known only from Baja Verapaz, Guatemala, from three stations within 12 km of each other. The habitat is open, strongly seasonally dry pine with oak forest, associated with grasses and agave. Although perennial, the species has annual shoots that are only up to 1 m long, and some fertile shoots may be but 0.25 m long. The leaves are dark green, without drip tips, and are very stiff and rigid. Passiflora trinifolia is apparently self-incom- patible, since 33 attempts to self-pollinate it in the greenhouse failed to yield fruit. Fully mature fruits are unknown in this species. Passiflora eglandu- losa, on the other hand, proved to be significantly self-compatible but not autogamous in cultivation. No unpollinated flowers set fruit over several years of cultivation, but 10 of 18 self-pollinated flowers produced (1-)3-9 seeds per fruit. The fruits turned purple 40-44 days after pollination. Nine or 10 *P. eglandulosa WP. trinifolia FicuRE 3. Map of Guatemala and neighboring countries showing distributions of Passiflora eglandulosa and P. trinifolia. 1662 Annals of the Missouri Botanical Garden seeds per fruit appears to be the normal maximum for this species, judging from the several fruiting collections examined. Passiflora eglandulosa (misidentified as P. tri- nifolia) has been included in three other studies of passionflowers. The development and physiology of the floral nectary of a clone of MacDougal 316 was examined by Durkee et al. (1981), who found it to be similar to the others in their study. Dried leaf samples of MacDougal 316 were chromato- graphically screened by McCormick (1982) for flavonoids. Concentrations were so low that no com- pounds could be verified, but according to Mc- Cormick (pers. comm.), traces of 3-O-glycosylfla- vonols but not C-glycosylflavones were detected. This is similar to several other species in section Cieca. Benson et al. (1975) reported the passion- flower butterfly Heliconius hortense to be an her- bivore of this species, a report that I confirm from field observations (insect voucher identified by J. Mallet) I am grateful to D. E. Stone of Duke University, who directed the fieldwork and establishment of the living collections. The initial research was ac- complished during graduate studies at Duke Uni- versity under a National Science Foundation Fel- lowship, further supported by NSF grant DEB- 7912607. Postdoctoral support has been gener- ously provided by the Jessie Smith Noyes Foun- dation. John Myers prepared the drawing. LITERATURE CITED Benson, W. W., OWN, JR. & L. E. GILBERT. 1975. Coesalntion x: plants and herbivores: passion flower butterflies. dii tion 29: 659-6 DuRKEE, L. T., D. L & W. H. REISNER. 1981. he floral and extra- orl nectaries of s. I. The floral nectary. 8: KiLLip, E. P. 1938. floraceae. Publ. Field Mus. Nat. Hist., e in merican aie of ders Bot. Ser. 1-613. McCormick, S. P. 1982. Flavonoid Chemistry of Pas- siflora Subgenus Plectostemma. Ph.D. Dissertation. Department of Botany, The University of Texas at Austin, Austin, Texas. SrANDLEY, P. C. & L. O. WILLIAMS. 1961. Passiflo- raceae. In: Flora of Guatemala, Fieldiana Bot. 24(7): 5-146. —John M. MacDougal, Missouri Botanical Gar- den, P.O. Box 299, St. Louis, Missouri 63166- 0299, U.S.A. MOLLINEDIA (MONIMIACEAE), A NEW GENUS FOR PARAGUAY Two genera of Monimiaceae have been reported previously for Paraguay—the monotypic Henne- cartia (H. omphalandra Poisson, endemic to Southern Brazil, eastern Paraguay, and northeast- ern Argentina) and Siparuna (S. guyanensis Aubl., cited by Hassler, 1917: 19). A third genus, Mol- linedia (M. clavigera Tulasne), has recently been collected in eastern Paraguay. This is a predictable occurrence, since the species grows in the nearby states of Sào Paulo, Paraná, and Santa Catarina in southern Brazil (Peixoto, 1979). Paraguayan material of Mollinedia clavigera exam- ined. PARAGUAY. CANENDIYU: 15 km SE of Katueté, isolated pond in middle of cultivated field dominated by grasses with zone of agricultural weeds next to soybean field, 24?15'S, 65%40'W, Hahn 2087, 15 Feb. 84 (PY). The three genera of Monimiaceae found in Para- guay can be distinguished as follows: la. Drupes free on a flat receptacle. Ovule dulous. Anthers sessile, langitudinally deluscent Mollinedia lb. Drupes surrounded by a concave receptacle. 2a. Drupes 1-2. Receptacle opening in seg ments. Anthers sessile, transversely dehis- cent. ities pendulous 0... Hennecartia . Drupes many. Receptacle not opening. An- thers pr! valvately dehiscent. Ovule erect Siparuna bo c LITERATURE CITED HassLER, E. 1917. Addenda ad Plantas Hasslerianas. Kundig, Genéve. PEIXOTO, A. L. 1979. Contribugáo ao conhecimento da secao Exappendiculatae Perkins do género Molli- nedia Ruiz et Pavón. Rodriguesia 31(50). — Lidia Molas, Inventario Biológico Nacional, Ministerio de Agricultura y Ganadería, Asun- ción, Paraguay; Elsa Zardini, Missouri Botan- ical Garden, St. Louis, Missouri 63166, U.S.A.; and Bruce A. Stein, The Nature Conservancy, Washington, D.C. 20036, U.S.A. ANN. Missouni Bor. Garb. 75: 1663. 1988. A NEW SPECIES OF RUELLIA (ACANTHACEAE) FROM WESTERN MEXICO Ruellia L., of tribe Ruellieae Nees emend. Bre- mekamp (1944), is a tropical and subtropical genus of 200 species (Long, 1964). Its variation is im- pressive and its taxonomy complex. Lindau's (1895) circumscription is adopted here, and its unifying characters are contorted aestivation of the corolla lobes; didynamous stamens with muticous, equal or nearly equal two-celled anther lobes; and sphe- roid or ellipsoid, three- or more-porate, reticulate, and spinulose or banded pollen grains. The primary centers of diversity of the genus are Indo-Malaya, Brazil, Africa, Mexico, and Cen- tral America (Long, 1970). In Mexico there are about 40 species of Ruellia, many of which occur locally, which suggests evolution by fragmentation (Ramamoorthy & Lorence, 1987). The variation in shape, size, and color of the corolla is very pronounced, indicating strong adaptive radiation to pollen vectors, which has led to definable natural constellations of infrageneric species groups. These deserve sectional ranke, but their recognition awaits li thy, in prep.). Amon revisionar these is the chiropterophilous group: Ruellia bour- gaei, R. coulteri, R. palmeri, R. pulcherrima, R. jaliscana, and the new species described below. The bats pollinating some of these are Leptonyc- teris nivalis and Anoura geoffreyi, both of the subfamily Glossophaginae. Ruellia sarukhaniana Ramamoorthy, sp. nov. TYPE: Mexico. Michoacán: Coalcomán, S of Naranjillo, 1,200 m, in woods, 24 Nov. 1938, G. Hinton et al. 12659 (holotype, GH). Figure 1. Ruellia ae rena Standley affinis a qua foliis oblan- ceolatis differ Suffrutescent herbs to 1 m high. Stem distinctly 4-angled, with numerous cystoliths; angles reddish brown. The leaves 15-22 cm long, 2-3.8 cm wide, oblanceolate, acuminate at tip, narrowed and de- current onto the 1-cm-long petiole, sinuately den- tate along margin, chartaceous, venation (12-14 pairs) actinodromus, pilose above with short, mul- ticelled, white hairs, pubescent along nerves below, cystoliths numerous. Inflorescence terminal and ANN. MISSOURI Bor. GARD. 75: 1664-1665. axillary, cymose-panicles, the peduncle and its branches densely tomentose with white hairs, in- terspersed with glandular hairs. Leaves subtending inflorescences reduced, bractlike, to 8 cm long, to 0.05 cm wide, oblanceolate. Bracts to 4.2 cm long, narrowly oblanceolate, tomentose with white hairs; bracteoles linear to subulate, glandular hairy. Ped- icel to 2 cm long, tomentose glandular hairy. Calyx 3.2 cm long, deeply subequally 5-lobed with 2 lobes longer than the rest, the lobes to 2.2 cm long, 0.3 cm wide, linear, villous with glandular hairs. Corolla to 7 cm long (the tube 3.5 cm long, 4 mm wide), yellowish, the throat 2.5 cm long, 1 cm wide, the lobes 1 cm long, 1 cm wide. Stamens 4 in pairs, the filaments 1.5 cm long; anthers 0.7 cm long. Ovary 6 mm high, glabrous, the nectariferous disc surrounding ovary 2 mm high, fleshy, the style 5.7 cm long, the stigma of 2 flat lobes. Seeds not known. The species is named after Dr. José Sarukhan Kermez, formerly Director of the Instituto de Bio- logía and currently coordinator of scientific re- search in the National University of Mexico. Ruellia sarukhaniana is similar to R. jaliscana with which it shares the glandular hairy, linear- lanceolate calyx lobes and its corolla form and branched inflorescence. The new species differs from the broadly ovate leaved R. jaliscana by having oblanceolate leaves. The two are allopatric: Ruellia jaliscana is re- stricted to northwestern parts of the transmexican volcanic chain in the state of Jalisco, and R. sa- rukhaniana is found in the Coalcomán area, the Their similarities and affinities suggest a possible common origin for the two species, with allopatry contributing to their differentiation. I thank Drs. D. Wasshausen (US) and T. Daniel (CAS) for comments on an earlier version of the typescript, and Fernando Chiang (MEXU) for the Latin diagnosis. LITERATURE CITED m C. E. B. 1944. — sp as a e Strobilanthinae (Acanthaceae). Verh. Ned. Aka d. barea Afd. Nate. Tweede Sct 41: 1-306, pls. 1988. Volume 75, 1988 Number 4 Notes 1665 FIGURE LINDAU, G. Die Natiirlichen Pflanzenfamilien Ab) 2 D D HERBARIO NACIONAL DEL INSTITUTO DE BIOLOGIA (MEXU) Ruela sarukhaniana sarukhaniana leon “aw. HOLOTYPE Det. T. P. Ramamoorthy Fecha 198 & a —— M^ HARVARD } e M. P eue uer wak w > M RR AEN $" pl w LU M MU wipe "P w K WE BE m2 Law dE AM WM QE te, Family Úcauth. Name Bat pu ined ln E z MM ghe file 1200 M. sco. E istric comm, ti € cee «ec Mee ' Mobi tas pues Description 1 m high. Fl, yellowish, 1. Ruellia sarukhaniana. Photograph of holotype. 1895. Acanthaceae. In: Eng ler : Prantl, 4-354. Lonc, R. W. 1964. Bios cic investi, cal s in sou Florida p of Ruellia (Acanthaceae). Amer o a E United States. J. Arnold Arbor. 51(3): 9. as T. P. & D. H. LoRENCE. t -852. he genera of Acanthaceae in the 1987. Species vicariance in the Mexican flora. Adansonia 2: 167- 175 —T. P. Ramamoorthy, Instituto de Biología, Apartado Postal 70-233, Ciudad Universitaria, Delegación Coyoacán, México D.F. 04510, Mex- ico. A NEW SPECIES OF STILLINGIA (EUPHORBIACEAE) FROM NORTHERN PERU The predominantly neotropical genus Stillingia Garden ex L. was last revised by Rogers (1951), who recognized 23 American species. He did not treat the three or four paleotropical species that are distributed from Madagascar to Fiji. One ad- ditional species has since been described from Mi- nas Gerais, Brazil (S. argutedentata Jabl., Phy- tologia 14: 451. 1967). A recently discovered species in northern Peru brings the total of Amer- ican species to 25. Stillingia parvifolia Sánchez Vega, Sagást. & uft, sp. nov. TYPE: Peru. Dept. Cajamarca: Prov. Cajamarca, Distr. Namora, en la que- brada del Rio Llallumayo, 2,775 m, 18 June 1984, I. Sánchez Vega & W. Ruiz Vigo 3618 (holotype, CPUN, F neg. 62127; iso type, HUT). Figure 1. Frutex glaber plus minusve ramosissimus, ramulis te retibus; folia brevipetiolata, alterna vel in brachyblastis 2-4 mm longis portata, laminae coriaceae, ovato-ellipti- cae, 6-11(-13) mm longae, margine callosae aequaliter serratae, dentibus glanduliformibus. Inflorescentiae ter- minales, bicularia, navicularia, persistentia; capsulae globosae, profunde trilobae; semina (immatura laevia, carunculata. Shrub, glabrous, + highly branched, the branchlets terete, + maroon, the bark irregularly sulcate; short shoots 2-4 mm long. Leaves alter- nate or borne on short shoots, short-petiolate; pet- ioles canaliculate, 1-1.5 mm long; blades coria- ceous, ovate-elliptic, 6-1 1(-13) mm long, 5-6 mm wide, 1.25-2.2 times as long as wide, broadly obtuse to rounded at base, obtuse to rounded at apex; margin callose, minutely and remotely ser- rulate, the teeth glanduliform, 4-19 per side, the base eglandular; midrib conspicuous, prominent be- low, the secondary veins obscure. /nflorescences terminal, 1-2 cm long, bisexual. Staminate cy- mules single-flowered, the bracts obovate, acute, navicular, ca. 1.2 mm long, persistent, biglandular at base, the glands oblong or suborbicular, 0.8-1 mm long, patelliform; calyx 2-lobed, ca. 1 mm ANN. Missouni Bor. GARD. 75: 1666-1668. long; stamens 2, the filaments 1-1.2 mm long, the anthers 2, ca. 0.8 mm long, longitudinally dehis- cent. Pistillate cymules 1-3 at base of inflores- cence, single-flowered, the bracts as in the stami- nate cymules; sepals 3, orbicular, navicular, ca. 12 mm long, the apex truncate or obtuse, slightly erose; ovary sessile, 3-carpellate; styles 3, free, -1.7 mm long, recurved at tip. Capsule glo- jon glabrous, deeply 3-lobed, ca. 5 mm long; lobes of the persistent gynobase 3-4 mm long; seeds (only immature seen) ca. 4 mm long, ca. 1.5 mm wide, smooth, prominently carunculate. Rogers (1951) distinguished two subgenera in Stillingia. Subgenus Stillingia is characterized by staminate cymules with 3-13 flowers, pistillate flowers with two or three well-developed sepals, and carunculate seeds, whereas subgenus Gym- nostillingia (Muell. Arg.) D. Rogers is character- ized by staminate cymules with a single flower, pistillate flowers with obsolete or minute and fu- gacious sepals, and usually ecarunculate seeds. The single-flowered staminate cymules of S. parvifolia would seem at first sight to place that species in subg. Gymnostillingia, but the ample and persis- tent sepals of the pistillate flowers and the carun- culate seeds argue against an easy accceptance of that disposition. Furthermore, the restriction of subg. Gymnostillingia to Mexico and adjacent areas of Guatemala and the United States makes such a conclusion even more unlikely. In addition, S. par- vifolia is not particularly close morphologically to any of the species that comprise subg. Gymno- stillingia. Stillingia parvifolia shares several specialized characters with two South American species of subg. Stillingia, S. peruviana D. Rogers (the only other species of the genus in Peru), and S. boden- benderi (Kuntze) D. Rogers of Brazil and Argen- tina. All three species are shrubs, have small short- petiolate leaves that are puen borne on smal peg-like short-shoots serrate, often callose, and with peculiar glandull- form serrations. he most reasonable interpretation of S. par- 1988. Volume 75, Number 4 Notes 1667 FIGURE 1. Stillingia parvifolia (Sánchez Vega & Ruiz Vigo 3618, CPUN) .—A. Flowering branch.—B. Leaf.— C. Lower portion of inflorescence.—D. Staminate flower.—E. Pistillate flower.—F. Capsule. 1668 Annals of the Missouri Botanical Garden vifolia would seem to be that it belongs to subg. Stillingia, where it is particularly closely related to S. bodenbenderi and S. peruviana, and that it has developed a reduced staminate cymule inde- pendently from the species in subg. Gymnostillin- Stillingia parvifolia may be distinguished from its closest relatives by means of the following key: KEY TO STILLINGIA PARVIFOLIA AND RELATIVES la. Staminate cymules 5-7(-9)-flowered; branch- tapering to an acute (rarely rounded) apex; sta- minate bracts plane. 2a. Petioles 0-3 mm long; leaf margins prom- = callose, the serrations ei glanduliform; staminate bracts r rounded, n . bo aik e m long; leaf margins o scurely callose, the serrations raay glanduliform; staminate bracts deltate, mu- cronate . peruviana ] Sanet cymules 1-flowered; branchlets n no wandlike, 2-3 mm thick; petioles 1-1. long; blades 6-11(-13) mm b. not cae to apex, this rounded or obtuse; staminate bracts navicular . parvifolia bo c All three species in this group have highly re- stricted distributions and are poorly known. Stil- lingia peruviana is known from three collections in Huancavelica Province, Peru, while S. boden- benderi is known from only two collections in Cór- doba Province, Argentina, and a third from the state of Sao Paulo, Brazil. The distinctions between S. bodenbenderi and S. peruviana are not entirely convincing; the relative status of these two species will depend on the future availability of more ample collections. Stillingia parvifolia is known only from the type collection from dry forest in a rocky ravine with clayey soil, where its associates include the shrubs Myrica sp., Prunus sp., Piper sp., and Gynoxys sp.; the herbaceous perennials Calceo- laria phaceliifolia Edwin and Dalea sp.; as well as several annuals. We are grateful to Segundo Leiva González for preparing the drawing and to Dr. Michael O. Dillon of the Field Museum for facilitating communica- tions between the authors. LITERATURE CITED Rocers, D. J. 1951. A revision of ded in the New World. Ann. Missouri Bot. Gard. 38: 207-259. — Isidoro M. Sánchez Vega, Herbario, Univer- sidad Nacional de Cajamarca, Cajamarca, Peru; Abundio Sagástegui Alva, Herbarium Truxi- llense, Universidad Nacional de Trujillo, Trujillo, Peru; and Michael J. Huft, Missouri Botanical Garden. Mailing address of M.J.H.: Department of Botany, Field Museum of Natural History, Chicago, lllinois 60605-2496, U.S.A. (Direct correspondence and reprint requests to M.J.H.) KUBITZKIA VAN DER WERFF, A SUPERFLUOUS NAME OR NOT? Kostermans (1988) discussed the need for a new generic name to replace Systemonodaphne Mez and came to the conclusion that the new nam Kubitzkia van der Werff (Taxon 35: 164. 1986) is superfluous. Kostermans's argument is that Meissner (1864) cited Laurus geminiflora Desv. as the basionym of his Goeppertia geminiflora with a question mark; his opinion is that citation of a doubtful synonym in the description of a new taxon rules out this doubtful synonym as the basi- onym of the new taxon and that, in the case of Kubitzkia, Laurus geminiflora Ham. is not the basionym of Goeppertia geminiflora, and there- fore also not the type species of Systemonodaphne Mez. The difficulty is that here one is led to spec- ulate whether Meissner, when he cited Laurus geminiflora Ham. as a basionym of his Goeppertia geminiflora with a question mark, accepted what he published or published what he did not accept (see Articles 34.1 and 34.2 of the Code). In his particular case, the choice is an easy one for the following reasons: A) Meissner (1864, p. 175), under his description of Goeppertia (?) gemini- flora, cited the earlier references as follows: ** Lau- rus geminiflora Desv. in Hamilt. Prodr. Fl. Ind. occ. p. 37. Walp. Ann. 1 p. 578? (non Reinw.).” In Walpers (1848), this species is cited as follows, “L. ? geminiflora Dsvx. in Hamilt. Prodr. 37 (nec Reinw. mss.)." It seems very likely that Meissner, in citing the Walpers reference, cited him correctly and included Walpers's question mark. This ques- tion mark reflected the doubt Walpers expressed about the correct generic placement of Laurus geminiflora Ham. and not any doubt from Meiss- ner as to L. geminiflora being the basionym of Goeppertia geminiflora. B) Meissner (1864) also included lists of excluded species at the end of each genus and gave the generic placement he accepted. On p. 236 he lists “L. geminiflora Desv. = Goep- pertia (?) geminiflora." When Meissner had doubt about the placement of certain species, he clearly expressed this; see p. 239: “L. surinamensis Sw. = Species obscura, cfr. Oreodaphne guianensis N.," or p. 236: “L. dubia Wall. = Cinnamomum iners Rw. 6 ?" In my opinion, Meissner's treatment of Goeppertia geminiflora shows that Meissner accepted without doubt Laurus geminiflora Ham. as the basionym for his Goeppertia geminiflora and that for reasons explained earlier (van der Werff, 1986), Kubitzkia is the valid name for Systemonodaphne sensu Mez non sensu typi. LITERATURE CITED KosTERMANS, A. J. G. H. 1988. What is the value of an in presen) mark in plant descriptions? Taxon 33. MEISSNER, A F. 1864. Lauraceae. In: D.C., Prodr. 15: -260. — a A. 1848. Annales Botanices Systemati- 1127. ca WERFF, Ed VAN DER. 1986. Kubitzkia van der Werff, a new name for a genus of neotropical Lauraceae. Taxon 35: 164-166. —Henk van der Werff, Missouri Botanical Gar- den, P.O. Box 299, St. Louis, Missouri 63166, U.S.A. ANN. Missouni Bor. Garb. 75: 1669. 1988. EDITOR'S NOTE THANKS - van der Werff for a year of willing, cheerful, and The editors wish to close the year with an expres- — able help in reviewing and polishing Latin descrip- sion of gratitude to Dr. John Dwyer and Dr. Henk tions. Instructions for Authors, Annals of the Missouri Botanical Garden The Annals publishes i m i s in systematic botany and related fields. nglish or Spanish. When in Spanish, an English Bad in i édition to a Spanish abstract is required. Short papers, especially nomenclatural matters and new taxa, are treated as “Notes,” which have their own format as described below. Authors are urged to contact the editor prior to submitting manuscripts. Manuscripts sent already reviewed and re- vised are generally subject to the full normal review pro- cess initiated here. Manuscripts not properly prepared may be returned for revision prior to review. This includes manuscripts with improper abbreviations or with diacrit- ical marks not indicated. Page costs are $70 per page, although charges may be reduced under certain circumstances. Corrections in proof will be billed nonnegotiably to the author at $2.50 per line. Manuscripts must be typed (output from dot matrix printers must be entirely clear) on one side of nonglossy 814 x 11 inch paper. Leave at least one inch as margin all around, except on the first page. There three inches should be left blank at the top. The text should be double-spaced or triple-spaced throughout, in- cluding abstract, notes, legends, tables, specimen P 1 oria Cited, and octal tes. Do not jus sti ify Tics. boldface, n other pane typefaces should not be used—they will appear later as communicated to the printer by the editor. Words to be set in italics should be underlined. Accents, umlauts, and other diacritical marks are the responsibility of the author. Footnotes should be typed double-spaced as paragraphs o contains acknowledgments, including granting agencies, and herbaria that loaned specimens. The second footnote is the author's address. Addresses for additional authors are each separate additional foot- notes. No other footnotes are permitted, except footnotes in tables, and these are best avoided. In papers formatted as “Notes,” there are no footnotes: the acknowledgments p the paragraph preceding the Literature Cited section, and the author's name and address make up the last paragraph of the article. In Notes, the names and addresses of multiple authors are all placed in the same paragraph. Long, complex tables are discouraged. Tables not neat and easily understood will be returned for improvement. Leave out vertical lines, and avoid horizontal lines. Lines are added as needed by the editor and/or printer. Type table captions double-spaced as paragraphs at the tops of the tables. A concise abstract should accompany each paper other than “Notes. Periods are used after all abbreviations (which should metric measures, compass direc- tions, and herbarium designations. States are not abbre- viated, and cities are usually spelled out, although “St.” is AEA pi Abbreviate periodicals according to BPH (Botanico-Periodicum-Huntianum), au- thors according to The Draft Index of Author Abbreviations Compiled at the Herbarium, Roy- abbreviations are to be capitalized, and that books listed in the Literature Cited section should have their titles spelled out when feasible. Note further that names of taxonomic authorities may be spelled out in major taxo- nomic citations such as the formal citation at the head of a taxonomic treatment. Except by prior arrangement with authors not having access to crucial references, the editor does not provide the service of looking up abbreviations. In the Literature Cited, periodicals should be listed as follows: author's last name, initial(s). Year. Full title of article. Journal abbreviated as in BPH. Volume: pages. Do not give parenthetical part numbers after volume numbers unless kn part is paginated separately, which is unusual. For more than one author, follow this style: author's last name, initia), second author's oen -— name & third author's initial(s), last name. Citat books should appear as follows: inde s last name, ini- tial(s). Year. Full Title (edited by Editor), 3rd edition, Volume 2. Publisher, Place. Citations of work “‘in prep.” and similar references to inaccessible sources are strongly discouraged. Except by prior arrangement when neces- sary, the editor does not perform library research to fill in yes bibliographic information. A brief Latin diagnosis for each new taxon is preferred to a ponies pa Latin description. se one paragraph per basionym as follows: Taxon author, literature citation, type citation, e.g., Bauhinia TYPE: Mexico. Veracruz: Valle de Cordoba, Bourgeau 1713 (holotype, K; isotypes, F, GH, P, US). Exclamation points are not used for specimens examined. Indicate unseen types as : Keys should be dichotomous, indented, and consistent with descriptions. LI should be numbered as la, lb, 2a, and so fort Cite specimens as foi llows: Additional specimens —€— (or Selected speci- mens examined or Paratypes). MEXICO. OAXACA: Sierra San Pedro Nolesco, Talea, 3 Feb. 1987 (fl), Jergensen 865 (BM, G, K, US). Dates and reproductive status are optional Send copies of illustrations; eer) will be requested if the manuscript is accepte will be returned after publication. All illustrative catia should be mounted on stiff cardboard. The maximum size for printed illus- page width (5% inches or ca. 140 mm). numbered in Arabic numerals in the order of their citation in the text. Parts of figures are labeled with capital letters. Photographs should be black-and-white prints. Transpar- encies and negatives are not accepted. Group figures into composite plates when possible, and abut edges of pho- e edges are squared. include telerik to latitude and longitude. Scale bars should appear on à illustrations, on p ee ca dam and on maps. unwelcome to the point of dels not being accepted. Scanning electron micrographs should be free of con- spicuous charging. Axes on graphs are all labeled. No explanatory text is to appear in figures. Rather, provide clear captions in sequential paragraphs apart from the other text, one paragraph for each group of figures, and following the style in current issues of the Annals. For other aspects of style, consult a recent issue of the Annals; the latest edition of The Chicago Manual of Style; or write the editor. Communication with the editor at all points of the publication process is encouraged. Manuscripts and Woo qnc should be addressed to the Editor, Annals of the Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166, U.S.A in cantions are ANNALS OF THE MISSOURI BOTANICAL GARDEN VOLUME 75 1988 The ANNALS, published quarterly, contains papers, primarily in systematic botany, contributed from the Missouri Botanical Garden, St. Louis. Papers origi- nating outside the Garden will also be accepted. Authors should write the Editor for information concerning arrangements for publishing in the ANNALS. EDITORIAL COMMITTEE E K. Rocers, Editor EORG Missouri Botanical Garden MarsHaLL R. CRosBY Missouri Botanical Garden GERRIT DAVIDSE Missouri Botanical Garden Jonn D. Dwyer Missouri Botanical Garden & St. Louis University PETER GOLDBLATT Missouri Botanical Garden HENK VAN DER WERFF Missouri Botanical Garden Colophon This volume of the ANNALS of the Missouri Botanical Garden has been set in APS Times Roman. The text is set in 9 point type while the figure legends and literature cited sections are set in 8 point type. The volume has been printed on 70# Centura Gloss, an acid-free paper Vn idi to have a shelf-life of over 100 years. Centura Gloss is manufactured by the Consolidated Paper Company. Photographs used in the ANNALS are reproduced using 300 line screen halftones. The AR used in the production of the ANNALS is a proprietary method known as Permanent Binding The ANNALS is printed and distributed by Allen Press, Inc. of Lawrence, Kansas 66044, U.S.A. (O Missouri Botanical Garden 1989 ISSN 0026-6493 De Nevers, GREGORY C. (See Andrew Henderson & Greg de Nevers) ......... DIAMOND, JARED. Factors Controlling Species Diversity: Overview and Syn- thesis DiLCHER, Davip L. (See Greg J. Retallack & David L. Dilcher) ....................... DUELLMAN, WILLIAM E. Patterns of Species Diversity in Anuran Amphibians in the American Tropics FonERo, ENRIQUE. Book Review FRITZE, KAREN J. & Norris H. WILLIAMS. The Taxonomic Significance of Pollen Morphology in the Columnea Alliance (Gesneriaceae: Gesnerioi- deae) GBILE, Z. O. & S. K. ADESINA. Nigerian Solanum Species of Economic Importance GENTRY, ALWYN H. Changes in Plant Community Diversity and Floristic Composition on Environmental and Geographical Gradients ........................ GENTRY, ALWYN H. New Species and a New Combination for Plants from Trans-Andean South America GEREAU, R. E. Book Review GOLDBLATT, PETER & MURIEL E. Poston. Observations on the Chromosome Cytology of Velloziaceae GOLDBLATT, PETER & BRUCE A. STEIN. Pollen Morphology of Pillansia L. Bolus (Iridaceae) GOTTLIEB, L. D. Towards Molecular Genetics in Clarkia: Gene Duplications and Molecular Characterization of PGI Genes GRAHAM, ALAN. Studies in Neotropical Paleobotany. V. The Lower Miocene Communities of Panama— The Culebra Formation GRAHAM, ALAN. Studies in Neotropical Paleobotany. VI. The Lower Miocene Communities of Panama— The Cucaracha Formation HAHN, WILLIAM JAMES. A New Species of Ilex (Aquifoliaceae) from Central merica HANSEN, Bruce F. & RicHARD P. WUNDERLIN. Synopsis of Dichanthelium in Florida Haynes, ROBERT R. Reproductive Biology of Selected Aquatic Plants ......... HENDERSON, ANDREW & GREG DE NEVvERS. Prestoea (Palmae) in Central America HosuiNo, TAKUJI & PauL E. BERRY. Chromosomal Observations on Fuchsia Species and Artificial Hybrids HosHino, TAKUJI & GERRIT DAVIDSE. Chromosome Numbers of Grasses (Poaceae) from Southern Africa. I Hurt, MicHaEL J. A New Species of Strychnos (Loganiaceae) from Nica- ragua Hurt, MIcHAEL J. (See Grady L. Webster & Michael J. Huft) ......................... Hurt, MicHaEL J. (See Isidoro M. Sanchez Vega, Abundio Sagastegui Alva & Michael J. Huft) 1168 168 1169 VOLUME 75 ADESINA, S. K. (See Z. O. Gbile & S. K. Adesina) ANDERSON, GREGORY J. & DAVID SYMON. Insect Foragers on Solanum Flow- ers in Australia ARMESTO, JUAN J. (See Mary T. Kalin Arroyo, Francisco A. Squeo, Juan . Armesto & Carolina Villagrán) ARROYO, Mary T. KaLin, FRANCIS A. SQUEO, JUAN J. ARMESTO & CAROLINA VILLAGRÁN. Effects of Aridity on Plant Diversity in the Northern Chilean Andes: Results of a Natural Experiment BarstEv, HENRIK. Two New Rushes (Juncus, Juncaceae) from Chiapas, exico BARRETT, SPENCER C. H. Evolution of Breeding Systems m Eichhornia (Pontederiaceae): A Review BARRINGER, KERRY. (See Lorin I. Nevling, Jr. & Kerry Barringer) .............. BEHNKE, H.-DIETMAR. Sieve-Element Plastids and Systematic Relationships of Rhizophoraceae, Anisophylleaceae, and Allied Groups BERRY, PauL E. Nomenclatural Changes in the Genus Fuchsia (Onagraceae) BERRY, PauL E. (See Takuji Hoshino & Paul E. Berry) BRIDSON, DIANE M. (See Jon C. Lovett, Diane M. Bridson & Duncan W. Thomas) CANESSA A., EDWIN. (See Nelson Zamora V., Louis J. Poveda A. & Edwin Canessa A.) CHASE, MARK W. (See Jeffrey D. Palmer, Robert K. Jansen, Helen J. 1 Michaels, Mark W. Chase & James R. Manhart) CLUSTER, PAUL D. (See Richard A. Jorgensen & Paul D. Cluster) ................. Cook, CHRISTOPHER D. K. Wind Pollination in Aquatic Angiosperms ........... Cox, PAUL ALAN & R. Bruce KNox. Pollination Postulates and Two-Di- mensional Pollination in Hydrophilous Monocotyledons DAHLGREN, Rorr M. T. Rhizophoraceae and Anisophylleaceae: Summary Statement, Relationships Dar, G. H. (See A. R. Naqshi, G. H. Dar, G. N. Javeid & P. Kachroo) .. D'ARcY, WILLIAM G. (See Ronald L. Liesner & William G. D'Arcy) ............ DAVIDSE, GERRIT. (See Takuji Hoshino & Gerrit Davidse) DAVIDSE, GERRIT € ROoBERT KRAL. Two New Species of Calyptrocarya (Cyperaceae: Sclerieae) from Venezuela and Observations on the Inflo- rescence Morphology of the Genus DE Lima, Harotpo C. A New Species of Hymenolobium (Leguminosae— Papilionoideae) from Central America De Nevers, GREGORY C. Notes on Bactris divisicupula and Bactris fus- cospina. Reexamined 1988 862 842 55 55 1387 1150 1153 853 1145 1151 JANSEN, RoBERT K. (See Jeffrey D. Palmer, Robert K. Jansen, Helen J. Michaels, Mark W. Chase & James R. Manhart) JANZEN, DANIEL H. Management of Habitat Fragments in a Tropical Dry Forest: Growth Javeip, G. N. (See A. R. Naqshi, G. H. Dar, G. N. Javeid & P. Kachroo) JORGENSEN, RICHARD A. & PAUL D. CLusTER. Modes and Tempos in the Evolution of Nuclear Ribosomal DNA: New Characters for Evolutionary Studies and New Markers for Genetic and Population Studies .................. Juncosa, ADRIAN M. AND Hinosui Tose. Embryology of Tribe Gynotrocheae (Rhizophoraceae) and Its Developmental and Systematic Implications Juncosa, ADRIAN M. & P. Barry ToMLINSON. A Historical and Taxonomic Synopsis of Rhizophoraceae and Anisophylleaceae Juncosa, ADRIAN M. & P. BARRY TOMLINSON. Systematic Comparison and Some Biological Characteristics of Rhizophoraceae and Anisophylleaceae Kacmroo, P. (See A. R. Naqshi, G. H. Dar, C. N. Javeid € P. Kachroo) Kaur, ROBERT B. Cupular Structure in Paleotropical Castanopsis (Faga- ceae) KeaTING, RICHARD C. & VOARA RANDRIANASOLO. The Contribution of Leaf Architecture and Wood Anatomy to Classification of the Rhizophoraceae and Anisophylleaceae Knox, R. Bruce. (See Paul Alan Cox & R. Bruce Knox) KraL, ROBERT. The Genus Xyris (Xyridaceae) in Venezuela and Contiguous Northern South America KRaL, ROBERT. (See Gerrit Davidse & Robert Kral) KraL, R. & MARIA DAS GRACAS DE Lapa WANDERLEY. Ten Novelties in Xyris (Xyridaceae) from the Planalto of Brazil LEARN, GERALD H., Jr. (See Barbara A. Schaal & Gerald H. Learn, Jr.) Les, DonaLD H. Breeding Systems, Population Structure, and Evolution in Hydrophilous Angiosperms LiESNER, RoNALD L. & WILLIAM G. D'Ancv. Two New Species of Inga (Leguminosae) from Panama Lovett, Jon C., Diane M. Bripson & Duncan W. Tuomas. A Preliminary List of the Moist Forest Angiosperm Flora of Mwanihana Forest Reserve, Tanzania Lowry, Porter P., II. Notes on the Fijian Endemic Meryta tenuifolia (Araliaceae) MacDoucaL, Jonn M. Passiflora pusilla (Passifloraceae), a New Species from Central America MacDoucaL, Jonn M. Notes on Passiflora eglandulosa, a New Species in Section Cieca (Medikus) DC. Formerly Included with P. trinifolia Mas- ters McPHERSON, GORDON. New and Noteworthy Taxa from Panama .................... 1499 1238 1410 1278 1296 1499 1480 1658 McPHERSON, GORDON. A New Species of Vantanea (Humiriaceae) from anama MANHART, JAMES R. (See Jeffrey D. Palmer, Robert K. Jansen, Helen J. Michaels, Mark W. Chase & James R. Manhart) MAXWELL, RICHARD H. A New Species of Dioclea Kunth (Diocleinae, Fa- baceae) from the Venezuelan Guayana MICHAELS, HELEN J. (See Jeffrey D. Palmer, Robert K. Jansen, Helen J. Michaels, Mark W. Chase & James R. Manhart) MILLER, JAMES S. A Revised Treatment of Boraginaceae for Panama ..... Mo as, LIDIA, ELSA ZARDINI & BRUCE A STEIN. Mollinedia (Monimiaceae), a New Genus for Paraguay NaosHI, A. R., G. H. Dan, G. N. Javeip & P. KAcHRoo. Malvaceae of Jammu and Kashmir State, India NEILL, Davip A. Experimental Studies on Species Relationships in Erythrina (Leguminosae: Papilionoideae) NEVLING, Lorin I., JR. & KERRY BARRINGER. A New and Endangered Species of Daphnopsis (Thymelaeaceae) from Ecuador NIKLAS, KARL J. Patterns of Vascular Plant Diversification in the Fossil ecord: Proof and Conjecture ORNDUFF, ROBERT. Distyly and Monomorphism in Villarsia (Menyantha- ceae): Some Evolutionary Considerations OSBORN, JEFFREY M. & EDWARD L. SCHNEIDER. Morphological Studies of the Nymphaeaceae Sensu Lato. XVI. The Floral Biology of Brasenia schreberi OvEWOLE, S. O. Chromosome Counts and Karyomorphology of Some West Tropical African Scilleae (Liliaceae) OvEWOLE, S. O. Karyotype Variation in Pancratium hirtum A. Chev. (Amaryllidaceae) PALMER, JEFFREY D., ROBERT K. JANSEN, HELEN J. MICHAELS, MARK W. CHASE & JAMES R. MANHART. Chloroplast DNA Variation and Plant Phylog- eny PENG, CHING-I. The Biosystematics of Ludwigia Sect. Microcarpium (On- agraceae) PHILBRICK, C. THomMas. Reproductive Biology of Freshwater Aquatic An- giosperms: An Introduction PHILBRICK, C. Thomas. Evolution of Underwater Outcrossing from Aerial Pollination Systems: A Hypothesis Poston, MURIEL E. (See Peter Goldblatt € Muriel E. Poston) ........................... PovEDA A., Luis J. (See Nelson Zamora V. € Luis J. Poveda A.) ................... PovEDA A., Luis J. (See Nelson Zamora V., Luis J. Poveda A. & Edwin Canessa A.) Price, H. James. DNA Content Variation Among Higher Plants .................... 1148 1180 730 1180 456 1663 1499 1180 QUIÑONES, Luz Mita. Una Nueva Combinación in Bauhinia (Fabaceae- Caesalpiniodeae) Raj, BHoj & HENK VAN DER WERFF. A Contribution to the Pollen Mor- phology of Neotropical Lauraceae RAMAMOORTHY, T. P. A New Species of Ruellia (Acanthaceae) from West- ern Mexico RANDRIANASOLO, VOARA. (See Richard C. Keating & Voara Randrianasolo) RAVEN, PETER H. (See Hiroshi Tobe & Peter H. Raven) RAVEN, PETER H. (See Hiroshi Tobe & Peter H. Raven) RAVEN, PETER H. (See Edward L. Vezey, Varsha P. Shah, John J. Skvarla l eter H. Raven RAVEN, PETER H. & P. Barry TOMLINSON. Rhizophoraceae- Anisophylle- aceae: À Symposium RETALLACK, GREG J. 8 Davip L. DILCHER. Reconstructions of Selected Seed Ferns RoHwER, JENS C. Three New Species of Nectandra from the Venezuelan uayana SAGÁSTEGUI ALVA, ABUNDIO. (See Isidoro M. Sánchez Vega, Abundio Sa- gastegui Alva & Michael J. Huft) SÁNCHEZ VEGA, IsipoRo M., ABUNDIO SAGÁSTEGUI ALVA & MicHAEL J. Hurt. A New Species of Stillingia (Euphorbiaceae) from Northern Peru ............ ScHAAL, BARBARA A. & GERALD H. LEARN, JR. Ribosomal DNA Variation Within and Among Plant Populations SCHNEIDER, EDWARD L. (See Jeffrey M. Osborn € Edward L. Schneider) SCHUTTE, ANNE Lise. (See Ben-Erik Van Wyk & Anne Lise Schutte) ......... SENDULSKY, TATIANA. (See Fernando O. Zuloaga & Tatiana Sendulsky) .... SHAH, VARSHA P. (See Edward L. Vezey, Varsha P. Shah, John J. Skvarla 1 & Peter H. Raven) SkvaARLa, JoHN J. (See Edward L. Vezey, Varsha P. Shah, John J. Skvarla 1 & Peter H. Raven) SMITH, PauL D. (See Kenneth J. Sytsma & Paul D. Smith) Squeo, Francisco A. (See Mary T. Kalin Arroyo, Francisco A. Squeo, Juan rmesto & Carolina Villagrán) STEIN, BRUCE A. (See Peter Goldblatt & Bruce A. Stein) STEIN, BRUCE A. (See Lidia Molas, Elsa Zardini & Bruce A. Stein) ................ STEVENS, WARREN DOUGLAS. A Synopsis of Matelea subg. Dictyanthus (Apocynaceae: Asclepiadoideae) STEVENS, WARREN Doucras. (See Julian A. Steyermark & Warren Douglas Stevens) STEYERMARK, JULIAN A. Flora of the Venezuelan Guayana— IV ... 1155 STEYERMARK, JULIAN A. Notes on Oldenlandia filicaulis and Oldenlandia tenuis (Rubiaceae) STEYERMARK, JULIAN A. Flora of the Venezuelan Guayana—V cc STEYERMARK, JULIAN A. Flora of the Venezuelan Guayana— VI „u STEYERMARK, JULIAN A. & WaRREN DOUGLAS STEVENS. Notes on Rhodog- naphalopsis and Bombacopsis (Bombacaceae) in the Guayanas ............ Symon, Davip. (See Gregory J. Anderson & David Symon) SYTSMA, KENNETH J. Taxonomic Revision of the Central American Lisian- thius skinneri Species Complex (Gentianaceae) SYTSMA, KENNETH J. & PauL D. SmitH. DNA and Morphology: Comparisons in the Onagraceae THomas, DUNCAN W. (See Jon C. Lovett, Diane M. Bridson € Duncan W. omas Tose, HiRosHi. (See Andrian M. Juncosa & Hiroshi Tobe) Tope, HirosHt & PETER H. Raven. Seed Morphology and Anatomy of Rhizophoraceae, Inter- and Infrafamilial Relationships Tose, HirosHI & PETER H. Raven. Additional Notes on the Embryology of Polygonanthus (Anisophylleaceae) and Relationships of the Family TOMLINSON, P. BARRY. (See Adrian M. Juncosa & P. Barry Tomlinson) ... TOMLINSON, P. BARRY. (See Adrian M. Juncosa & P. Barry Tomlinson) ... TOMLINSON, P. Barry. (See Peter H. Raven & P. Barry Tomlinson) ........... VAN DER WERFF, HENK. Eight New Species and One New Combination of Neotropical Lauraceae VAN DER WERFF, HENK. A New Species of Ocotea (Lauraceae) from Nic- aragua and a Note on Ocotea jorge-escobarii VAN DER WERFF, HENK. Notes on Kubitzkia van der Werff, a Superfluous Name or Not? VAN DER WERFF, HENK (See Bhoj Raj & Henk van der Werff) ...................... VAN Wyk, BEN-ERIK & ANNE LISE SCHUTTE. Chromosome Numbers in Lotononis and Buchenroedera (Fabaceae—Crotalarieae) VEZEY, EDWARD L., VARSHA P. SHAH, JOHN J. SKVARLA & PETER H. Rav- Morphology and Phenetics of Rhizophoraceae Pollen ......................... VILLAGRÁN, CAROLINA. (See Mary T. Kalin Arroyo, Francisco A. Squeo, Juan J. Armesto & Carolina Villagran) WANDERLEY, MARIA DAS GRAGAS DE Lapa. (See R. Kral & Maria das Graças de Lapa Wanderley) WEBERLING, Focko. The Architecture of Inflorescences in the Myrtales ... WEBSTER, GRADY L. & MICHAEL J. Hurt. Revised Synopsis of Panamanian Euphorbiaceae WIERSEMA, Jonn H. Reproductive Biology of Nymphaea (Nymphaeaceae) WILLIAMS, Norris H. (See Karen J. Fritze & Norris H. Williams) ................ Worr, STEVEN J. George Engelmann Type Specimens in the Herbarium of the Missouri Botanical Garden WUNDERLIN, RICHARD P. (See Bruce F. Hansen & Richard P. Wunderlin) ZAMORA V., NELSON & Luis J. PovEbA A. Una Nueva Especie de Guettarda L. (Rubiaceae, Guettardeae) para Costa Rica ZAMORA V., NELSON, Luis J. Povepa A. & EpwiN Canessa A. Una Nueva Especie de Caryodaphnopsis Airy Shaw (Lauraceae) para la Region Neotropical ZARDINI, ELSA. (See Lidia Molas, Elsa Zardini & Bruce A. Stein) .................. ZULOAGA, FERNANDO O. & TATIANA SENDULSKY. A Revision of Panicum Subgenus Phanopyrum Section Stolonifera (Poaceae: Paniceae) ............ 1608 1637 1157 Volume 75, Number 3, pp. 739-1168 of the ANNALS OF THE MISSOURI BOTANICAL GARDEN was published on October 19, 1988. Three New Species of Nectandra from the Venezuelan Guayana Jens G. Rohwer ....... A Synopsis of Matelea subg. Dictyanthus (Apocynaceae: Asclepiadoideae) Warren Douglas Stevens Flora of the Venezuelan Guayana — VI Julian A. Steyermark Taxonomic Revision of the Central American Lisianthius skinneri Species Complex (Gen- tianaceae) Kenneth L SYMA doe feli Chromosome Numbers in. Lotononis and Buchenroedera (Fabaceae— Crotalarieae) Ben-Erik Van Wyk & Anne Lise Schutte George Engelmann Type Specimens in the Herbarium of the Missouri Botanical Garden Steven J. Wolf Synopsis of Dichanthelium (Poaceae) in Florida Bruce F. Hansen & Richard P. Wunderlin NOTES Passiflora eglandulosa, a New Species in Section Cieca (Medikus) DC. Formerly Included 1 with P. trinifolia Masters John M. MacDougal ... Mollinedia (Monimiaceae), a New Genus for Paraguay Lidia Molas, Elsa Zardini & Bruce A. Stein .. A New Species of Ruellia (Acanthaceae) from Western Mexico T. P. Rama- moorthy ; A New Species of Stillingia (Euphorbiaceae) from Northern Peru Isidoro M. Sánchez 1 Vega, Abundio Sagástegui Alva & Michael J. Huft ..........— Kubitzkia van der Werff, a Superfluous Name or Not? Henk van der Werff i- Editor's Note ————À À re ere QR 1587 603 1608 1637 658 1663 1664 1670 CONTENTS Towards Molecular Genetics in Clarkia: Gene Duplications and Molecular Characterization f PGI Genes L. D. Gottlieb ` Chloroplast DNA Variation and Plant Phylogeny Jeffrey D. Palmer, Robert K. Jansen, 1 Helen J. Michaels, Mark W. Chase & James R. Manhart Ribosomal DNA Variation Within and Among Plant Populations Barbara A. Schaal -£ Gerald H. Learn, Jr. DNA and Morphology: Campa in the Onagraceae Kenneth J. Sytsma & James des Smith Modes and dni in the Evolution of Nuclear Ribosomal DNA: New Characters for Evo- lutionary Studies and New Markers for Genetic and Population Studies Richard A. Jorgensen & Paul D. Cluster DNA Content Variation Among Higher Plants H. James Price Rhizophoraceae- Anisophylleaceae: A Symposium Peter H. Raven & P. Barry Tom- 1 linson kikin and Anisophylleaceae: Summary Statement, Relationships Rolf : M. T. Dahlgren A Historical | and Taxonomic Synopsis of Riissphoracese and Aniso- phylleaceae - Adrian M. Juncosa & P. Barry Tomlinson ............. Lie ; Systematic Comparison and Some Biological Characteristics of TENSON and | ophylleaceae Adrian Juncosa & P. Barry Tomlinson ; Seed Morphology and Anatomy of Rhizophoraceae, Inter- and Infrafamilial Feliu ships Hiroshi Tobe & Peter H. Raven ' ‘The Contribution of Leaf Architecture and Wood Anatomy to Classification of the Rhi- l zophoraceae and des Richard C. Keating & Voara Randriana- solo .... Morphology and Phéuctice of Rhizophoraceae Pollen Opere E Vezey, Varsha : : P. Shah, John J. Skvarla & Peter H. Raven ; SP Sieve-Element Plastids and Systematic Relationships of Rhizophoraceae, Ache 169 180 1207 1217 1259 1278 1296 1319 . 1343 1369 254887 - aceae, and Allied Groups. H.-Dietmar Behnke Embryology. of Tribe Gynotrocheae (Rhizophoraceae) | and Its Developmental and Sys tematic Laphentions ; DEM M. Juncosa & Hiroshi Tobe Additional Notes on the Embryology of Polygonanthus (Anisophylleaceae) a and 1 Rela- E tionships of the Family - Hiroshi Tobe & Peter H. Raven ....... New Speci nd | Nes Coml jin f r Plants from Trans-Andean South America dm | V. The L Lowe