ANNALS OF THE MISSOURI BOTANICAL GARDEN VOLUME 80 1993 Colophon This volume of the ANNALS of the Missouri Botanical Garden has been set in APS Bodoni. 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# Vintage Gloss. This is an acid-free paper designed to have a shelf-life of over 100 years. Vintage Gloss is manufactured by the Potlatch Paper Company. Photographs used in the ANNALS are reproduced using 300 line screen halftones. The binding 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 1993 ISSN 0026-6493 Number 1 Annals of the Missouri Botanical Garden Volume 80, Number 4 Winter 1993 The Annals, published quarterly, contains papers, primarily in systematic hotany, € tributed from the Missouri Botanical Garden, St. Louis. Papers originating outside th Garden will also be accepted. Authors should write the Managing Editor for information. concerning arrangements for publishing in the ANNALS. Instructions to Authors are d n we back of the last issue of each volume. Editorial Committee Henk van der Werff Editor, Missouri i GE Garden à ‘Amy Scheuler | · Managing Editor, Миздин Botanical Garden | Dos SEE Editorial Assistant, > үе Missouri Botanical Garden T | "Masi Dues Cirri Moran —— Publications СТА . Missouri Botanical Garden Ihsan A. Al-Shehbaz Missouri Botanical Garden Gerrit Davidse Missouri Botanical боа ; Roy E. Gereau Peter Goldblatt aus Missouri Botanical Сан MES Gordon мерено = Missouri Botanical Garden P. Mick Richardson Missouri Botanical Garden For subscription information contact Department Eleven, P.O. Box 299, St. Louis, MO 63166-0299. - - Subscription price is $100 per volume U.S., $105 Canada and Mexico, $125 all other countries. Four _ issues per volume. The journal Novon is included in the subscription price of the ANNALS. < © Missouri Botanical Garden 1993 The ÁNNALS OF THE MISSOURI BOTANICAL. GA _ Postmaster: Send address changes to ANNA THE Missouri BOTANICAL GARDEN, Depar . Eleven, P.O. Box 299, St. Louis, MO 63 166-0 (ISSN 0026-6493) is published quarterly by | Missouri Botanical Garden, 2345 Tower Grove / enue, St. Louis, MO 63110. Second class posta paid at St. Louis, MO and additional mailing 0 Volume 80 Annals Number 1 of the 1993 Missouri Botanical Garden MONOGRAPH OF THE Noriaki Murakami? and NEOTROPICAL SPECIES OF "00" C. Moran’ ASPLENIUM SECT. HYMENASPLENIUM (ASPLENIACEAE)! ABSTRACT Asplenium sect. Hymenasplenium is one of the best defined groups of Asplenium, being св by creeping rhizomes, dorsiventrally symmetrical steles, swollen petiole bases, unique rachis-costae structure, and chromosome base numbers of x = 38 ог 39. In the Neotropics, the section has ten species and three hybrids. The species are A. delitescens, A. hoffmannii, A. laetum, A. obtusifolium, A. ortegae, A. purpurascens, A. repandulum, A. riparium, A. triquetrum, and A. volubile. The hybrids are A. ш (of unknown parentage), А. delitescens Х А. laetum, and A. xincisoserratum (= A. hoffmannii Х А. laetum). All the aforementioned species and hybrids are endemic to the Neotropics. Central America and the Andes harbor the most species and endemics. A cladistic analysis was not done because the Lm species of the section apparently do not form a monophyletic group separate from the paleotropical one Section Hymenasplenium is one of the best Asplenium species have erect or ascending rhi- defined groups within Asplenium, distinguished by — zomes, radially symmetrical steles, nonswollen pet- the following synapomorphies: creeping rhizomes, iole bases, and n = 36 or multiples thereof (rare dorsiventrally symmetrical steles, swollen petiole exceptions differ in only one of these character- bases, unique rachis-costae structure and chro- istics). mosome base numbers of x — 38 or 39. All other Hymenasplenium was first described by Hayata ' We thank the curators of the following е for making loans available: А, AAU, B, BM, BR, COL, СВ, F, G, GH, K, L, M, MICH, NY, PORT, QCNE, S, U, UC, 2. We also thank the following botanists who helped one or both of us with fieldwork in Latin America: pales Anger, Gerardo Aymard, Mireya Correa, Lisa and Larry Dorr, Michael Grayum, Hiroshi Kidono, Maria Morrello, David Neill, Benjamin Vllgaard, Francisco Ortega, and Pablo Sanchez. We also thank the Smithsonian Tropical Research Institute for the use of its Fortuna Dam facilities when we worked in Panama, and UNELLEZ in Guanare, Venezuela, for the use of its facilities. The research for this paper was done while the first author worked at the Missouri Botanical Garden supported by a grant from The Japanese Society for the Promotion of Science. He thanks Peter H. Raven, who served as his official host. Finally, we thank our wives, Misaki Murakami, who did the illustrations, and Cirri K. R. Moran, who helped with the computer-related aspects of the research. ? Botanical Gardens, Nikko, University of Tokyo, 1842 Hanaishicho, Nikko, pv 321-14, Japan. * Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166-0299, U.S.A ANN. MISSOURI Bor. Ganp. 80: 1-38. 1993 Annals of the Missouri Botanical Garden FIGURE 1. Creeping rhizome, dorsiventral stele, and swollen leaf bases— characteristics of Asplenium sect. Hymenasplenium. (1927) as a monotypic genus based on 4. unilater- ale Lam. and distinguished entirely by the stelar structure of the long-creeping rhizome (Fig. 1). Hayata was greatly influenced by stelar structure and described several genera based solely upon this character (e.g., Diploblechnum, Monachosorella, Boniniella, Pentarhizidium, and Hymenaspleni- um). Iwatsuki (1975) revised the species of Hy- menasplenium, which were then thought to occur primarily in the Old World, and reduced the genus to a section of Asplenium. This ranking has been followed by nearly all pteridologists, including us. The present monograph treats 10 neotropical species that clearly belong to section Hymenas- plenium alongside the Old World species. (The Old World species of the section are 4. apogamum Murakami & Hatanaka, 4. cardiophyllum (Hance) Baker, A. cataractarum Rosenstock, 4. cheilo- sorum Kunze, A. excisum C. Presl, 4. hondoense Murakami & Hatanaka, 4. obliquissimum (Haya- ta) Sugimoto & Kurata, 4. obscurum Blume, 4. subnormale Copel., and A. unilaterale Lam.) We have not done a cladistic study of the neotropical species because they are not a monophyletic group distinct from the paleotropical species. The first author plans to do a cladistic study of all the species in the section worldwide using evidence from mor- phology, anatomy, chloroplast DNA, and nuclear gene sequencing of the small subunit of ribulose- bis phosphate carboxylase. The only difference we have seen between the Old and New World species is that the New World ones have proliferous roots, whereas the Old World ones do not. Proliferous roots do not correlate with any other character (besides geography) and therefore cannot be used to define a natural group. Previous workers, studying only the Old World species of section Hymenasplenium, have sur- veyed anatomy (Iwatsuki & Kato, 1975; Misuta et al., 1980), reproductive patterns and morpho- logical variation (Murakami & Iwatsuki, 1983), phytochemistry and chemotaxonomy (Murakami & Hatanaka, 1983, 1985, 1988a; Murakami et al., 1985), ecology and morphological variation (Kato & Iwatsuki, 1985, 1986), genetic diversity (Watano & Iwatsuki, 1988), taxonomy (Murakami & Hatanaka, 1988b), and chromosome numbers (Mitui et al., 1989). The New World species are much more poorly studied. For a long time, workers in the Old World thought that the only New World species that belonged to section Hymenasplenium was А. ob- tusifolium. Smith (1976) was the first to point out A. melanopus (= А. purpurascens), А. laetum, А. that the New World species 4. delitescens, hoffmannit, and А. repandulum also belonged to the section. He did not, however, study the group in detail. SPORES We examined the spores of all the neotropical species of section //ymenasplenium and found that perispore morphology was useful in defining two groups. The first group has five species, which we informally refer to as the “repandulum group.” It comprises 4. obtusifolium, А. repandulum, А. riparium, А. triquetrum, and А. volubile. All have spiny spores (Fig. 2A-C) or, obtusifolium, papillate spores (Fig. 2C, H). The in the case of 4. second group comprises the rest of the neotropical species, which all have cristate spores (Fig. 2D- F). The eight paleotropical species of section Ну- menasplenium examined by Kato et al. (1990) FIGURE 2. Тһе spores of various neotropical species Mexico, Copeland 15940 (MICH). — sporangium, Panama, Antonio 3530 Steyermark et al. 121451 (UC). B. 4. volubile, Colom Brazil, Sperling 6035 (MO). —H. A. obtusifolium, race with 64 spores per sporangium, Venezuela, of Asplenium, sect. Hymenasplenium. — А. A. riparium, bia, Forero et al. 7237 (F).—C. А. repandulum, Ecuador, E. А. ortegae, Steyermark et al. ).—G. A. obtusifolium, race with 32 spores per ~ e ~ 3 € AS © = bo e со Volume 80, Number 1 Murakami & Moran 1993 Asplenium sect. Hymenasplenium Monograph Annals of the Missouri Botanical Garden 20 = " NY ht Pes • 0 о МАР 1. had cristate spores and would therefore belong to this second group. Because most species of As- plenium outside of section Hymenasplenium have cristate spores (Tryon & Lugardon, 1991), the spiny and papillate character states should be con- sidered apomorphic. GEOGRAPHY The neotropical species of section Hymenas- plenium are all endemic to the Neotropics. They range from southern Mexico to Panama, the An- tilles, and South America from Venezuela to south- eastern Brazil, forming a wide arc around most of Amazonian Brazil (Map "he Andes from Venezuela to Bolivia harbor the most species (8) and this is the only region with endemics (i.e., Asplenium ortegae, А. repandu- lum, A. volubile). Costa Rica and Panama are also species-rich, containing six species. The Antilles The distribution of Asplenium sect. Hymenasplenium in the New World. and extreme western Amazonian Brazil both have two species, and the Guianas, southeastern Brazil, and Paraguay all have one species. The Serra do Mar region of southeastern Brazil, which is a center of species richness and endemism for ferns (Tryon, 1972), has played a minor role in the diversification of the section. Only one species (4. triquetrum) occurs there and it is nearly endemic (Map 9). Asplenium obtusifolium is notable for its nearly circum-Caribbean distribution (Map 5) and its cor- relation with geography of the 32- and 64-spored races. PRINCIPAL COMPONENTS ANALYSIS Our recognition of 10 neotropical species of Asplenium sect. Hymenasplenium is based pri- marily upon qualitative characters as shown in the key. In order to check the validity of our species circumscriptions, we decided to do a Principal Com- Volume 80, Number 1 1993 Murakami & Moran 5 Asplenium sect. Hymenasplenium Monograph ata URE И Parts of the plants pucr for the Prin- nents Analyses (Figs. 4, 5). pan length d petiole length (PL), petiole u^ (PW), number of pinna pairs ), pinna length (PNL), pinna ce length (PBL), pinna width at middle (PML), pinna excised portion length (PEL), number of veins (NV), depth at pinna base (DPB), auricle length (OD), serration depth at tip (TD), number of sori (NS), sorus length (SL), distance to sorus istance from costa (DC), seg- W), number of lobes (NL), rhizome width (RW), petiole base width (PBW), distance between two petioles (DBP), root width (RT). ~ ponents Analysis (РСА) of certain quantitative characters (Fig. 3) measured on herbarium spec- imens throughout the range of each species. We separated the species into two groups according to their separation in the first couplet of the key. The first group, which is identified under lead lb in the A. laetum, А. de- litescens, A. purpurascens, and А. ortegae. The key, contained А. hoffmannii, second group, which we believe to comprise a close- ly related group of species and which is identified under lead la of the key, contained А. obtusifoli- um, А. repandulum, А. riparium, А. volubile, and 4. triquetrum. Then, we did a PCA on each group using computer program NTSYS version 1.40 (Rohlf, 1988). The data set for each her- barium specimen was projected on three principal component axes for the first group (Fig. 4) and two axes for the second group (Fig. 5; the third axis accounted for little variation in the second group). Tables 1 and 2 show the Eigen values and Eigen vectors of the Principal Components. The result for the first group was that all species were separated from each other (Fig. 4). This result supports our circumscriptions of the species. The result for the second group was not as clear Fig. 5). Although А. obtusifolium, А. riparium, and 4. volubile were separated from each other, А. repandulum completely overlapped А. trique- ~ trum and А. volubile. Asplenium repandulum, FIGURE 4. Principal Components Analysis of 23 characters of А. delitescens (D), А. hoffmannii (Н), A. laetum (1), A. ortegae (R), and А. purpurascens (P). The three Principal Component axes account for 62% of the measured variation VU y Y 6 Annals of the Missouri Botanical Garden 10 0.97 ГУ 0.0) DL —1.0 т T xd <=. si. me © rar FIGURE 5. race), А. repandulum (E), А. riparium (1), А. volubile c *, axes account for 54% of the measured variation. however, can be distinguished from 4. triquetrum by petiole length relative to the lamina, habitat, and range (see key, couplet 4), and from 4. vol- ubile by the carinate rachis and rachidial wings parallel to the plane of the lamina. These char- acteristics were not included in the measurements used for the PCA. Another result was that the two spore-races of 4. obtusifolium could not be sep- arated by the morphological characteristics mea- sured (Fig. 5) RELATIONSHIP TO OTHER GROUPS OF ASPLENIUM We are not sure what group of species within Asplenium is most closely related to section Hy- menasplenium. Several species of Asplenium (A. abscissum Wil . argentinum Hieron., А. host- mannii Hieron., an) A. otites Link) greatly resem- ble certain species in section Hymenasplenium in leaf form. Consequently, they are often misiden- tified as a species in section Hymenasplenium, especially 4. laetum (which see for comparison). These species, which can be immediately distin- guished from section. Hymenasplenium by their erect rhizomes and rachis-costa architecture, may be the closest group in Asplenium related to section 0.0 0.5 10 15 PC1 Principal Components Analysis of 20 characters of А. obtusifolium (А, 32-spored race; 5, 64-spored V), and А. triquetrum (T). The two Principal Component Hymenasplenium. This suggestion is based only on similarities in leaf form and is therefore tenta- tive. ‘TAXONOMIC TREATMENT Asplenium sect. Hymenasplenium (Hayata) Iwatsuki, Acta Phytotax. Geobot. : 1975. Basionym: Hymenasplenium Hayata, Bot. Mag. (Tokyo) 41: 712. 1927. TYPE: As- plenium unilaterale Lam. rhi- zome creeping, green to blackish, with two rows Plants terrestrial, epipetric, or epiphytic; of alternately arranged petioles on the dorsal sur- face, scaly near the apex; rhizome stele dorsiven- tral, composed of two unequal meristeles connected by lateral strands, the ventral meristele wider and the dorsal one narrower, bearing roots from either the ventral, dorsal, or connecting meristeles; petiole terete, scaly at base, glabrous distally, greenish to atropurpureous, the base swollen and often per- sisting after the leaf has fallen and decayed; lamina usually 1-pinnate or (in 4. cardiophyllum) simple and cordate; rachis not or very shallowly grooved, with or without perpendicular or flat green wings, lacking buds; costae bordered by a flange of green Volume 80, Number 1 1993 ABLE l. The Eigen vector values for each PC in the analyses of А. delitescens, А. hoffmannii, A. laetum, A. ortegae, and A. purpurascens (Fig. 4). Char Principal component acter (Fig. 3) | 2 3 4 LL 0.579 | —0.699 0.200 0.043 PL 0.896 | —0.233 0.086 0.064 PW 0.833 —0.151 0.110 —0.055 NP =0112 -=0.811 0.388 0.172 PNL 0.936 0.032 —0.045 —0.143 PBL 0.579 —0.531 —0.393 0.072 PML 0.912 0.098 —0.237 —0.030 PEL —0.044 —0.720 —0.497 —0.163 NV 0.796 0.432 0.17 0.077 DPB 0.263 0.165 — 0.370 0.696 ор —0.157 – 0.304 0.395 0.157 тр – 0.245 0.306 — 0.617 0.128 NS 0.737 0.093 0.018 0.428 SL 0.667 —0.028 -0.098 —0.129 DE 0.272 0.130 —0.225 —0.469 DC 0.058 —0.542 —0.614 —0.113 SGL 0.732 —0.092 Á —0.101 —0.336 SGW 0.838 0.241 —0.096 —0.018 NL 0.757 0.403 0.074 —0.176 RW 0.483 —0.435 0.274 0.054 PBW 0.752 —0.159 0.051 0.161 DBP 0.614 0.172 —0.075 0.481 RT 0.782 0.223 0.279 —0.303 tissue that is decurrent on the adaxial surface of the rachis; decurrent margins of the pinnae not thickened; veins free or (in 4. cardiophyllum) anastomosing; spores bilateral, ellipsoid to subglo- bose, the perispore commonly cristate, but occa- sionally spiny or papillose; x = 38 or 39 1. Asplenium delitescens (Maxon) L. D. Go- mez, Brenesia 8: 52. 1976. Diplazium de- Murakami & Moran 7 Asplenium sect. Hymenasplenium Monograph TABLE 2. The Eigen vector values for each principal component in the analyses of 4. obtusifolium, A. repan- dulum, A. riparium, A. volubile, and A. triquetrum (Fig. 5). Character Principal component (Fig. 3) 1 2 LL 0.866 —0.343 PL 0.721 0.337 PW 0.646 0.218 NP 0.696 —0.561 PNL 0.887 0.089 PBL 0.649 0.291 PML 0.659 0.452 PEL 0.010 0.722 DPB —0.602 0.491 OD 0.511 —0.355 TD —0.392 0.560 NS 0.798 —0.214 SL 0.599 0.309 DE 0.476 0.413 DC 0.024 0.523 SGL 0.489 0.363 SGW 0.314 0.590 RW 0.645 0.089 PBW 0.770 —0.041 DBP 0.264 =0.702 litescens Maxon, Contr. U.S. Natl. Herb. 10: 497, t. 56, fig. 1. 1908. туРЕ: Cuba. Oriente: vic. of San Luis, 18 Feb. 1902, Pollard & Palmer 348 (holotype, US; isotypes, F, MO). Figure 6A; Map 2. Plants terrestrial; roots 1-1.5 mm wide; rhizome 3.5-7 mm wide, nearly naked; scales 2-2.5 mm х 0.25-0.35 mm, black-brown, narrowly lanceo- late, toothed; petiole bases 2.8—4.5 mm wide, swol- len, 1-2 mm distant from each other on the same row; petiole 15-25 x 0.15-0.2 cm, ca. % the KEY TO THE NEOTROPICAL SPECIES OF ASPLENIUM SECT. HYMENASPLENIUM la. Leaf apex conform, subconform, or hastate; spores spiny or papillose; rhizome greenish when living (sometimes turning blackish upon drying 2a. Pinna apices obtuse or rarely acute; pinna pairs 3-10, usually with a free basal acroscopic lobe; spores Ant papillose; plants rheophytic, usually under or near waterfalls; Costa Rica, Panama, Colo = Venezuela, illes obtusifolium L. N = Pinna apices acute or rarely obtuse; pinna pair spores spiny; a epiphytic or epilithic and (usually) on rocks = strea 3a. strongly carinate abaxially, triangular in cross section, with Rachis s 8-18, usually without a sim ee Mess lobe adaxial green wings in the same ca as bo lamina (these wings are difficult to distinguish i in dried material); Ecuador, Peru, Bolivia, Paraguay, southern Brazil. 4a. Plants epilithic; petiole ca. 4 as long as the lamina length; Bolivia, Paraguay, southern Brazil C. Moran 4b. Plants epiphytic; petiole 4-' as long as the lamina; М жи Ecuador and Per 7. A. . triquetrum Murakami in R. не Kunze о c rep . Rachis rounded abaxially, terete or semi-terete in cross section, with adaxial green “= регреп Annals of the Missouri Botanical Garden dicular to the plane of the lamina (the wings difficult to see in dried specimens); southern Mexico d to western Venezuela an cuador. ба. Petioles 14-14 as long as the lamina; plants twining around small trees or saplings; ша internodes 1-2.5 cm long; Costa Rica, Panama, wester n Colombia, western Ecuado 4. volubile Murakami & R С. Moran n Venezuela and Ecuador 5b. Petioles ү to ipai du is lamina; E usually on boulders along streams; rhizome internodes 2-0 a A. ong; co to west lb. Leaf ig nonco 6a. Fre riparium Liebm. nform, prism ps cristate; Tinie brown or blackish when liv e pinna pairs 2-6(-8); rachis green; Mexico to Panama, Trinidad, Venezuela, Colombia 2. A. hoffmannii Hieron. 6b. Free pinna pairs 6-25; rachis atropurpureous to castaneous. Ta. Veins (in the middle of the pinnae) 2-4-forked; rachis grooved, with an adaxial green wing perpendicular to the plane of the lamina; western Venezuela, and Ecuador and Peru on the eastern 5 side of the Andes 7b. Veins (in the middle of the pinnae) 1 R. C. Moran А. ortegae Murakami & -2-forked; rachis grooved but without adaxial green wings. Ва. Pinnae pinnatifid throughout their length, usually lobed 14-14 to the costae; Ecuador ex Kuhn 8b. a entire to serrate, sometimes with a basal auricle on the acroscopie side and/or basiscopic da. Pinnae 7-11 x 1.2 outhern ia Es Venezuela and Peru 9b. Prae? 2-6 x southern и urpurascens Мен. -2 cm, 6-9 pairs; petiole and ice stramineous to greenish brown; A. d elitescens (Maxon) L. D. Gómez 2 cm, 10-25 pairs; petiole 4 rachis име ve to . laetum Sw. leaf length, stramineous to green-brown, occasion- ally partly purplish, glabrous; lamina 20-30 cm long, green, l-pinnate, deltate with an abruptly reduced, pinnatifid apex; rachis like the petiole, grooved and with herbaceous green wings; lateral pinnae 7-11 x 1.2-2 cm in the middle of the leaf, 6—9 pairs, lanceolate to linear-lanceolate, long- acuminate, the margin slightly toothed, the teeth .5-1 mm deep, excavate on the basal basiscopic side /,—/, of the total length, acroscopic base trun- cate, auricle 1-4 mm long; veins 2-forked; sori 5— 10 mm long, occasionally diplazioid in the abruptly reduced pinnae near the apex; spores 33-38 um long, 64 per sporangium, perispore cristate. Additional specimens rs ined. MEXICO. CHIAPAS: f Yaxchilan on the banks of И 29935 (Е, МО, NY); Mpio. de Las A W side of Laguna Miramar E of San Quintin, 350 m, Breedlove er (F, MICH, he confluence 34140 (MO); Palenque, in rocky streamside forest by the ruins, weg Chater et al. 128(BM, MO); Mpio. Ocosin- S de Frontera Corozal, sobre la orilla del Río (ошоо. 120 m, Martínez S. 8008 de Ejido Benemérito de las mbo de Cacao, 160 m, Martínez S. 8136 (MO, "NY. 2 al SE de Crucero Corozal, camino Palenque-Boca cantum, 200 m, Mc 5. 6402 (MO); Mpio. Ocosingo, Boca Lacantum a Palenque, 220 m, Martínez S. 16399 (MO); Cerro de Macuspana, Rovirosa 1038 (GH), 3135 (MO); Ruina Palenque, 100 m, Saiki М234 (F, Z). ОАХАСА: Mpio. Sta. Магла Chimalapa, a pie de la Piedra Siniguichi en la cabecera del arroyo Huahuagtza, cerca de Piedra Blanca, 7 km E de Sta. Maria, 270 m, 16%55'N, 94°37'W, Hernández С. 734 (MO, NY); filo del cerro al E del Arroyo Monte Rico, ca. 20 km al E de Sta. Maria, 400-500 m, 16%55'N, 94°42’W, Hernández G. 1574 (MO, NY); ca. 25 km al E de Sta. Maria, cerro al N de la unión del Río Blanco (del S. a con el Río del Corte, al N del Rio del prs 400-5 16%55'N, 94°42'W, Hernández G. 1 ou ca. i m e Sta. Maria por la Pics al Rio Pinal, Chimalapilla, 300 m, 16%57'N, 94°36'W, Hernández С. & González L. 1787 (МО); Dtto. Tuxtepec, 9 km 5 of Tuxtepec and 4 km W of Rte. 175, behind Fabricas de Papel Tuxtepec, 100 m, Mickel 5784 (NY, UC); between Tuxtepec and Playa Vicente, 28 km SE of Rte. 147, 100 m, Mickel 7212 (NY, UC); Dtto. Juchitán, 1 km N of Palomares, 1 km E of highway on road to La Pedrera "Paso de Buques," 100 m, Mickel & Pardue 6843 (NY, UC) 5 km S of Palomares, 150 m, Mickel & Pardue 6866 (NY). TABASCO: Teapa, 4 km SE of Teapa on road to Tacotalpa, base of Cerro de Madrugal, 300 m, Croat 47908 (CR, MO). VERACRUZ: Córdoba, Finck 159 (UC); Mpio. Hidalgotitlan, near Campamento La Laguna, 100 m, 17°17'N, 94°30'W, Nee 29974 (F); Mpio. Hidalgo- titlan, entre Hermanos Cedillo y La Escuadra, Rio Solo- suchil, 150 m, Vasquez 977 (NY); 2 km al N del Poblado no. 2, zona de Peppe Mpio. Jesús Carranza, 100 m Vásquez T. 2485 (NY). BELIZE. STANN CREEK: Stann Creek railway, 17 mi., Gentle 27 14 (F, GH, MICH, NY); caves and railway, 33 m, Schipp 8-275 (B). TOLEDO: Columbia Forest Reserve, ca. 1-2 mi. N of entrance, Croat 24173 (MO); Condemn Branch Hills, Gentle 5139 , MO, 5) in cohune ridge, Chavarrias road, Rese- mederes, across Columbia pies Gentle 626 1 (F, С, NY, S, UC). ATEMALA. PROV. UNKNOWN: Los Amates, 30 m, Deam 470 A Smith: 5675 (NY). PETEN: ca. 4 S of Lacand Contreras 3432 (MICH, S); Chinchilá, Sebol Road, un 10500 (F); along Río P Diego, between Finca Yalpemech and San Diego, 50-150 m, Steyermark 45308 (F). HONDURAS. ATLANTIDA: d eu Valley, near Tela, 20-600 m, Standley 54153 (F). COPAN: — со 1 ar Volume 80, Number 1 Murakami & Moran 1993 Asplenium sect. Hymenasplenium Monograph Sh NX EN y 0 f MAJ, у Y NA 7) y = A (И | ==> WD dA || B “ss MAL. / M M YY ) IS, Д4 А [| ч A ДИ Nos WE 7 "VEN EAN 6 SE ЈИ V west ИИ Й Sy, 52. SS СР S ә, New SS = NX. Ws. NI eS Nes SN Ws - > d 7 SS SK 1227 а" EE SS DNS E Y a ii b Y iN NS. NS NN SS AN NN Q Ws s al esses ОГ WESS e 74 SY = SS | (= = Ws NS, UN М Possible hybrid between 4. delitescens S 2 SAE OSEAN а 27 саа ' ‘ W LJ pr а TE E FIGURE 6.—A. Asplenium delitescens, Panama, Seibert 583 (MO) —B. x laetum, Ecuador, Luteyn et al. 8502 (NY). Annals of the Missouri Botanical Garden MaP 2. The distribution of Asplenium delitescens. 20 km NW de Santa ue de Copán, hacienda Ocoteseco, 1,300 m, ipo R. 65 (МО); Fort of La Cumbre, near creek, Thieme 4 (UC y SANTA BARBARA: San died Sula, 200 m, Thieme 5675 (B, Lr NICARAGUA. MATAGALPA: El Carmen, 2 a Waslala, 600- 700 m, 13?13'N, 85*37'W, MUR 1906. 3 (MO). ZELaYa: Dept. же fields, 3. s km SE Cerr d Isidro, 65 m, d rs. Proctor et al. 2 4 (F). Costa RICA. ALAJUE- cate, И í SUB. SAN JOSE: Carrillo. 300 m, Wercklé 17432 (B). PUNTARENAS: Palmar Norte, Gra yum et al. 9152 (MO, UC). PANAMA. BOCAS DEL TORO: vic. of Laguna de Chiriqui, Hart 42 (UC). CANAL ARE near Madden Dam, Alston 8868 (BM); „Жон Colorado Island, Aviles 16 (F, b hill at Canal Zone bour i lett & Lasser 16580 (GH, MIC и МО); no locality, 'Cornman 51. 21 i POT to Em- Croat 6516 (MO); Barro Colorado Island, W igi Trail 3, Croat 8053 (МО); short cut to Lutz Trail behind animal house, Croat 8515 (F ); Ae "Trail 200, Croat 8597 (MO); Armour Trail 2170, Croat 8643 (F, МО); Barro Colorado Island, Armour Trail 850, Croat 9425 (MO); Barro Colorado Island, Armour Trail 1040, Croat 11639 (MO, NY); а Colorado Island, Drayton rail 100, Croat 12662 (MO); Barro Colorado Island, ribus Trail 500, Croat 14056 (MO); vic. of Madden Lake along Boy Scout road, 100 m, Croat 38327 (МО); Rio Providencia 3 km SE of Achiote near W border of Canal Zone, 5-100 m, Gentry 8679 (AAU, MO); Hun- newell 16375 (MICH); Pipeline road, Limbo Hunt Club, Kennedy & Andrews 1876 ; Barro Colorado Island, Kenoyer 29 (GH); Valley of Masambi on the road to Las ascadas Plantation, 20-100 m, Maxon 4677 (BM, GH, NY): Barro Colorado Island, Lake Gatun, Maxon et al. 6807 (GH); trail along Rio Petipie from road to Fort Sherman from Gatún Locks, Mori & Kallunki 2672 6 km E of Gamboa, headwaters of Río Casaya, Nee a 26 (UC); Madden Dam, near Río Chagres, 50- 75 m, Seibert 583 (GH, МО); 4 mi. NE of Gamboa, vic. of Gold Creek, Seibert 587 Barro Colorado Island, Shattuck 189 (F, MO); oe Коле trail, Wet- more & Woodworth 105 (GH); Madden Lake, Wither- spoon & Witherspoon 8797 (MO). CHIRIQUI: Burica Pen- sula, 8 mi. W of Puerto Armuelles, 200 m, Croat 22493 F, MO). COLON: ca. 6 km N of Chilibre, along shores of Madden Lake, 25- rus sd 79°36'W, Knapp 2716 О). DARIEN: nea uth of Rio Yape, 20 m, Allen 7 (MO); Rio Chico, ye Yavisa at junction with Rio = = Б' Volume 80, Number 1 1993 Murakami & Moran 11 Asplenium sect. Hymenasplenium Monograph Chucunaque to ca. 1 hour by outboard from junction, Burch et al. 1095 (МО); E slope of Cerro Sapo, 500 m, Hammel 1266 (MO); Paya to Pucra, Stern et al. 235 MO), Tyson et al. 4835 (MO); 2 mi. E of nta Fé, Tyson et al. 4835 (UC); 2 km E of Juan Diaz, Cornman 512 Aer MO, UC); Piriati, S of Pan Amer- ican Hwy 00 m, 900'N, 78°30'W, Hamilton 529 (МО. ^M 1859-1861, Hayes 57 (BM, F, GH); Bajia Sta., Hayes 355 (BM, K); ing ri Topia River, 50 m, Killip 2708 (B, MICH, S), 5 , Killip 2806 (MO); Gatuncillo bi ies & a е 415 (NY) Panama Viejo?, Seema 6 (BM); Marraganti and vic., m, Williams 1026 dS PANAMA: Campana Hill, 800 m, Alston 8925 (BM); Isla de Bayano, trocha 2 Correa А. et al. 2973 (F, MO); near Juan Diaz, 0-7 m, Killip 2512 (S); Serrania de Маје, ridges 5 of Chocó village of Lm Rio Tine drainage system, 500-600 m, 8°47'N, 2334 (AAU, CR, MO); Chilibre, Martines 9 (MO) S is ey, Ekman 2760 (S).VENEZUELA. ARAGUA: Colonia Tovar, Fendler 143 (С, NY). BARINAS: along Rio Caparo, 2-4 km up river from dam site, 100-200 m, 7941'М, 71°28'W, Liesner € González 9473 (MO). FALCON: Parque Nacional Quebrada el Toro, 600 m, van der Werff 402 (UC). LARA: border area à between Edo. Lara/ Yaracuy, Sierra de Aroa, 10-13 mi. of Ur- achiche along dirt road leading NW din Urachiche to Duaca, 1,400 m, 10°14’N, 69904", Smith et al. 1340 (MO). SUCRE: Dtto. Sucre, along Quebrada el Tigre, 5 of Fila La Baqueta (in the future basin of Represa Neveri), 350 m, 10%07'N, 64°19'W, Davidse & González 19242 (MO, NY, UC). TERRITORIO FEDERAL AMAZONAS: upper Orinoco region, Ugueto, Croziat 794 (BR). COLOMBIA. ANTIOQUIA: Guapa, 53 km S of Turbo, 60 m, Haught 4628 (COL); 53 km S of Turbo, Guapá, 80 m, Haught 4684 (NY, S). BOLIVAR: Boca Verde, on Rio Sinu, 100- 300 m, Pennell 4213 (GH); 12 km S of Carraipia, Com. Guajira, 450 m, Haught 4286 (COL, NY). cHoco: near Caserio La Teresita, 0.5-2.5 km N of the INDERENA camp on the Rio Truando, 50-100 m, Lellinger & de la Sota 232 (COL) Nercua and Divide, Schott 57 (Е); Katios, sector gei fL Jagua, Sororia Creek, 200 m, Haught 3613 (COL); ien Marta, H. H. Smith 2693 (NY). SANTANDER: vic. O Puerto 2d between Carare and Magdalena rivers, a ue m, Haught 1715 (COL, GH); SE of Puerto Berr ae Haught 2865 (COL, GH, UC). ECUADOR. LOS RIOS: yi nton Vinces, Jauneche Forest, k Mn and Palenque, on Estero dii al. 7089 (AAU, МО, ОСМЕ). у бе edo to Calceta, Haciend 2976 (GH, S, UC). PERU. above river on stee PARA: Serra dos Carajas, ca. 6 km NW of AMZA camp 3-Alfa, on the road to camp 4-Alfa, 200-250 m, 5?47'S, 50°34'W, Sperling et al. 6035 (GH, NY, UC); 12 km W of camp ECB on the ferrovia, ca. 57 km W of road BR 150, 150 m, 5%35'S, 49*15'W, Sperling et al. 6381 (BM, F, GH, NY). Asplenium delitescens occurs from Mexico to Peru and east-central Brazil, and Cuba (Map 2). It grows terrestrially in wet forests from О to 800(— 1,400) m. Rarely, it grows on rocks. The species is characterized by stramineous or greenish petioles, few (6—9) pinna pairs, and deltate laminae with abruptly reduced apices (Fig. 6A). Where the lamina abruptly contracts to the nar- rowed apical segment, the pinnae and lobes are often retuse. Occasional specimens are interme- diate between А. delitescens and А. laetum (Fig. 6B). These are treated in the hybrid section of this monograph. Asplenium delitescens closely resembles А. ађ- scissum Willd. (which is not a member of section Hymenasplenium) in leaf cutting and outline. To the naked eye, А. abscissum differs from А. de- litescens by its erect (rather than creeping) rhizome and the obovate, evenly serrate (rather than qua- drangular-truncate, and irregularly serrate) pinnae that are immediately beneath the apical segment. Two microscopic characters also distinguish the species. Both species are sparsely scaly abaxially where the pinnae join the rachis. In 4. abscissum, the scales are 0.2-0.6 mm long, reddish, and uni- seriate. In A. delitescens, the scales are 0.5-1.5 mm long, blackish, and uni- to triseriate. The struc- ture of the pinna and rachis is also diagnostic. Asplenium abscissum has a slightly raised costa and a thickened pinna margin decurrent on the side of the rachis (Fig. 16C). In contrast, Æ. de- litescens has a slightly sunken costa bordered by a green flange of tissue that is decurrent on the adaxial surface of the rachis (Fig. 16D). The de- current margin of the pinna is not thickened (this form of the rachis and costa juncture is constant throughout section /7ymenasplenium ). Smith (1976) first pointed out that 4. delites- cens and several other neotropical Asplenia be- longed to section Hymenasplenium. He also re- alized that this species belonged to Asplenium, rather than Diplazium where it was originally placed, and gave an excellent discussion on the characteristics that distinguish the two genera. 2. Asplenium hoffmannii Hieron., Hedwigia 60: 258. 1919. TYPE: Costa Rica. Alajuela: Aguacate, Aug. 1857, Hoffmann 836 (holo- Annals of the Missouri Botanical Garden ЖУУ ree (Toe || AN \ vi | / | же NEN | \ ) | " || л CM FicunE 7. Asplenium hoffmannii. — А. Panama, Johnston 131 (US). — B. Mexico, Breedlove 33893 (MO). type, B; isotype, NY fragment, photo BM ex mm wide; rhizome 1.8-3.5 mm wide, almost na- B). Figure 7; Map 3. ked; scales 0.8-1.3 x 0.2-0.3 mm, black-brown, Asplenium membranifolium Maxon, Amer. Fern J. 24: ovate-lanceolate; petiole base 1.4-2.4 mm wide, 72. 1934. TYPE: Panama. Panamá: ca. 10 mi. E of swollen, 0.3-2.7 mm distant from each other on Panama City, Killip 2567 (holotype, US; isotype, һе same row; petiole 4-9 x 0.05-0.1 cm, 14-16 NY fragment). the leaf length, greenish, glabrous; lamina 15-40 Plants epilithic, rarely terrestrial; roots 0.4-0.6 ст long, light green, membranaceous, 1-pinnate, Volume 80, Number 1 1993 Murakami & Moran Asplenium sect. Hymenasplenium Monograph МАР 3. The distribution of Asplenium hoffmannii. lanceolate to oblong with attenuate apex; rachis green, grooved, with perpendicular wings; lateral pinnae 2-3.5 x 0.7-1.1 cm in the middle of the leaf, 2-6(-8) pairs, subquadrangular to lanceolate, the margin deeply toothed, 1—2 mm deep, excavate for ca. % the length of the basiscopic side, obtuse to acute at apex, shortly stalked at lower pinnae, auricle absent; veins 1-2-forked; long; spores 33-38 um, 64 per sporangium, peri- sori 3-5 mm spore cristate. Additional specimens examined. MEXICO. CHIAPAS: Mpio. of Ocosingo, ruins of Yaxchilan on the banks of the Rio Usumacinta, 300 m, Breedlove 33893 (MO); ruins of Yaxchilán on banks of the Rio Usumacinta, 300 m, Breedlove 42851 (MO); Tuxtla- San Andrés, Sc hmid Corte, 200 m, 16%56'N, 94*41"W, (NY); Dtto. Pochutla, 29.6 km NE of Pochutls, Oax Pr road, 300 m, oci 1304 (NY); Dtto. i20 5 km W of Rte , road from near Palomares Re Tuxtepec, 100 m, Mickel 7411 (NY, UC); Dtto. Pochutla, 15 km N of Pochutla, 300 m, Mickel & Leon- ard 5138 (NY); Juquila, 34 km N of Puerto Escondido, O m, Mickel 6077 (NY). GUATEMALA. SANTA ROSA: near Oratorio, 1,200 m, Standley 60656 (F). HONDURAS. COPAN: Ruinas de Copán, Bernoulli 834 (B), Bernoulli & Cario 290 (B, K). EL SALVADOR. PROV. UNKNOWN: Salto near Huizuca, Rio Las Lajas, 300 m, Seiler 718 (NY). NICARAGUA. RIO SAN JUAN: San Bartolo, Seymour 6200 (MO). Costa Rica. GUANACASTE: 4 km W Bagaces, Opler s.n. (CR). GUANACASTE: Rincón de la Vieja National Park, slopes of Volcán Santa María, 700-900 m, 10?46'N, 85°18'W, A. R. Smith et al. 1961 (CR, MO, UC). PUNTARENAS: Palmar Norte, Grayum et al. 9159 (MO, UC). PANAMA. CANAL AREA: Juan Mina, Flat Rock, Chagres River, Bartlett & Lasser 16843 (GH, MICH, МО); Ala- juela, Cornman 548 (MICH, UC); Fort Sherman, Dwyer 6823 (MO); trail along Rio Petitpie, from road to Fort Sherman from Gatun Locks, Mori & Kallunki 2670 (MO); near Сага, hills W of the Canal, Standley 27218 (L, MICH, NY, UC); 1 mi. N of Summit Garden, on rock embankment of bridge just above small creek, Sytsma 2314 (МО); Old Fort San Lorenzo, from rocks near fort in shade, 10 m, Tyson 2237 (MO); Madden Forest Pre- serve, across from George W. Green Park, Welch 19856 (MO); vic. of bans Hydrographic Station, Río Pe- queni, 80 m, Woodson et al. 1601 (GH, MO). coLow: 6 km N of Chilibre, along shores of Madden Lake (Lago Alajuela), 25-50 m, 9*12'N, 79*36'W, Knapp 2686 (MO) Panamá Viejo, Seemann 72 (BM; photos GH, UC) Taboga Island, Cornman 597 (MICH, UC); San José Island, Perlas archipelago, ca. 55 mi. SSE of Balboa, Johnston 131 (ВМ, GH, U), 258 (BM, GH), 385 (BM, GH); ca. 55 mi. SSE of Balboa, Johnston 444 (GH); Panamá Viejo, Seemann 369 ( PANAMA: Madden , forest near spillway, at bottom of dam, 50 m, 79°37'W, Churchill 3851 (MO, UC); 5 mi. W of bridge over ra ied Lake, 200 m, Hammel 5106 (MO); Rio Tecumen, Standley 29372 (CH). 1 DAD: Blue Basin, Hombersley 257 (BM). VENEZUELA. ARAGUA: prope coloniam е O Л рго il aut (NY, mixed with Asplen п). Annals of the Missouri Botanical Garden PORTUGUESA: Dtto. Araure, a Hijitos, entre Agua Blanca y San Rafael, 160 m, Ortega & Aymard 2089 (UC); Dtto. Guanare, near Bie ы 400 m, 4. R. Smith et al. 856 (MO, UC); Dtto Seo along road to San José de La Montana, 500 m, 920'N, 69°43'W, Smith et al. 1095 (NY, UC); Ta P Boca de Monto; 28- 32 km NNE de Guanare, arriba del puente sobre el Rio María, 400-500 m, 9?18-19'N, 69942-43W, Steyer- mark et al. 127064 (MO, UC); 7 km NE de Boca de Monte, 22 km NE del vado del Rio Suruguapo, 45 km NE por la autopista Guanare-Ospino, en el sitio Las 69?41'30"W., Steyermark et al. 127236 (MO, UC). COLOMBIA. BOLIVAR: Torrecilla, near Turbaco, 150-300 m, Killip & Smith 14658 (K, NY). Xj Asplenium hoffmannii occurs from southern Mexico to Colombia, Venezuela, and Trinidad (Map 4). It grows on rocks along deeply shaded streams in wet forests. Rarely, it grows on old walls or terrestrially. The elevation ranges from 0 to 500(- The species is characterized by membranaceous laminae, few (1—3) leaves per rhizome, deeply and evenly serrate pinnae, and few (usually 2-6) pinna pairs (Fig. 7). It hybridizes with 4. laetum to pro- duce 4. Xincisoserratum (which see; Fig. 16). The hybrid differs from A. hoffmannii by having more pinna pairs (6-14), more than three leaves ~ per rhizome, basally darkened petioles (especially in large leaves), and aborted spores. 3. Asplenium laetum Sw., Syn. Fil. 79, 271. 1806. NEOTYPE (designated by Proctor, 1985): Schkuhr, Krypt. Gew. 1: 65, t. 70. 1809. Figure 8A; Map 4 a ian Paris 6: 273 “America,” n uia GH, US ex Asplenium virens Desv., 1827, not C. Presl, 1825. lector unknown (P-Herb. они ). a | gern а Presl, Tent. Pt gs 10 836, r 4. laetum on r p Distrito Federal: Car Bredemeyer s.n. (W- Herb. Jacquin). This is pro me bly the b aei used by Schkuhr (1809) in drawing his t. hich has been designated as the neotype of 4. obse (see above). dx lugubre Liebm., Mexic. bregn. 243 (seors. 91). 1849. LECTOTYPE (designated by А. R. Smith, 981): Mexico. Veracruz: Baranca de Mirador Liebmann [Fl. Mex. 322] (lectotype, С not seen; isolectotypes, B, K). Plants terrestrial or epilithic; roots 0.5-0.8 mm wide; rhizome mm wide, sparsely scaly; scales 2-3.3 x 0.23-0.33 mm, black-brown, lin- ear-attenuate; petiole base 1.5-3.5 mm wide, swol- len, 0.4-2.5 mm distant from each other on the same row; petiole 4-20 x 0.09-0.15 cm, 6-Y the leaf length, blackish reddish purple, sometimes green, grooved, sparsely scaly at base; lamina 15- 40 cm attenuate apex; rachis like petiole, grooved and long, 1-pinnate, lanceolate to oblong with with green, perpendicular, herbaceous wings; lat- eral pinnae 2- 0 cm in the middle of the leaf, 10-25 pairs, subquadrangular, the margin slightly to strongly toothed, the teeth 0.5-2 mm deep, the basiscopic side excavate basally for ca. % the pinna length, the apex acute, the base broad- ly cuneate and almost sessile to stalked, auricle 0— 3 mm long; veins 1-forked; sori 2.5-7.5 mm long; spores about 27-38 um, 64 per sporangium, peri- spore cristate. ач specimens examined. МЕХІСО. CHIAPAS: e Guatemala border, Mpio. of Las Margaritas, low н at the о tate), 300 m, Breedlove 34197 (MICH, MO); Mpio. of Frontera Comalapa, 6-8 km E of Frontera Com d along road to Ciudad Cuauhtémoc, 0 m, Breedlove 39059 (NY); 13 km N of Berriozabal, Mpio. ны a logging road to Colonia Figaroa, 1,600 m love & А. R. Smith 31319 (МУ); los alrededores de la zona Arquelogica de Palenque, Cabrera et i 1938 (BM, А i , Chater et 1864- 1870, Ghiesbregh 426 (K); O, UC), 5 e £e = — (c e a E. ‘ss O ~ dad ~ ч Y ~ Y SA ©з ac ш o un 4 nandez G. 1008 (NY); Chimalapilla, ca. de 18 km al NE ү Sta. Maria por la vereda al Rio Pinal, 300 m, 16%57'N, 36'W, Hernández G. & González 1787 (NY); Dtto. Тамер 4-9 km 5 of Valle Nacional on Rte. 175, 200- 700 m, Mickel 5881 (NY, UC); Dtto. Ixtlán, 29 km S of vale Dex 80 km N of Ixtlán de Juárez, trail E of Rte. 175 at Campamento Vista Hermosa toward Ladu, 00-650 m, Mickel 6461 (NY, rs Dtto. Ixtlán, 76 km N ij Ixtlán ка o on Rte. 175, 5 km S of Vista o La Esperanza, Yo 188 (NY, i - Morton & Ja la = Lis Diaz, 330 m, Hallberg 1468 NY) 400 m "Hallberg 1471 (NY). TABASCO: cerro arriba del Ejido Za unu, talpa, 3 km al E del ejido Lázaro Cárdenas, 50 m, Cowan 2072 (NY); Teapa, 3 km E of Teapa, slope of Ce rro las mpanas, ca. 50 km S of Vila netas, 50-100 m, Conrad et al. 2889 (MO); Cerro de Cocorrá, near Teapa, pee 828 (NY). VERACRUZ: prés Cordova, Chiqui- ite, Bourgeau 2151 (K, L, MO, NY); Mpio. Hidal- un 6 km E de Cedillo camino a La Laguna, 110 m Dorantes et al. 2555 T A Orizaba, 1855. m, Müller 1835 (МУ); in Dec. diris 174 (7); Mirador, Barranca de San E 1,000 m, Ross 688 (BM). BELIZE. STANN CREEK: caves and Ра 30 m, Schipp Volume 80, Number 1 1993 Murakami & Moran Asplenium sect. Hymenasplenium h Monogra У Ду La т а = "a ut == ZO = ==; i Gh UN ZAM IN W E^ NUN К\ AN ОА > [77 27. / {| › UU SS ү pá ce PN Wa SSY УМЕ 7 Zu „ NE, A 4 SSAI A < AA ra P Zaz LEI SRS == л We 2 SS а Ye ZZ SSE SN > 4 ЧУ ООЛ С ый = a «а. ва Ер. UL Ss Hisaki, I! AN V (2 bs = = N ) "diae | —В, C. Asplenium triquetrum, Brazil E 8.— A. Asplenium laetum, Panama, Mori & Kallunki 2826 (МО). — in C. Fic l Dusén psy (MO). Cross section of rachis shown i Annals of the Missouri Botanical Garden MaP 4. The distribution of Asplenium laetum. S-277 (B). TOLEDO: Edwards road beyond Colombia, on hilltop beyond Central camp, Gentle 7378 Be EN MICH, NY, S, UC); Colombia Forest Reserve, forest camp ca. 6 mi. S of Cabro, in upper Rio Grande drainage area, 300 m, Proctor 3607 1 (BM, MO). GUAT- EMALA. ALTA VERAPAZ: Finca Mercedes, Teleman, Panzós, faldas da la Sierra de las Minas, quebrada Mercedes, 100 m, Martínez S. et al. 22711 (MO); along Rio Icvolaty, N and NW of Finca Cubilquitz to Quebrada Diablo, 300- 350 m, Steyermark 44752 (F), 1 mi. SW of Sibicte, vic. of laguana Sapala (Chajvovuch), 280 m, Steyermark 44894 (F, UC). ESCUINTLA: SE of Esc aie “along or near Rio Michatoya, 250-300 m, Standley 89091 (F); be tween Panzós and Senahü, vic. of Cacao, 275 m, 6 (МУ); Nazatenango, orillas del Rio Grande, 347 (B); Cubilquitz, von Türckheim 8048 (B, NY, S). HUEHUETENANGO: Sierra de los Cuchumatanes, between Ixcan and Finca San Rafael, 200-800 m, Steyermark 49389 (F). PETEN: W of Chinajá, along Rio San Román, 50 m, Steyermark 45507 (NY); along Rio San Roman, 50 m, Steyermark 54407 (F) ranco Hondo, 1,100 m, Standley 88974 ATLANTIDA: near Tela, Lancetilla Valley, 20- -600 m, Stati- dley 52815 (F), 53554 (F); Mts. back of Puerto Sierra, Wilson 161 (NY), NICARAGUA. Barber Bernoulli Wilson 57; CHINANDEGA: vic. of Chichigalpa, 90 m, Standley 11234 (UC). cHONTALEs: Castillo, Mar. 1893, Shimek s.n. (F, MO). Masaya: Sierra de Managua, región de Las Nubes Standley 8744 (F). Rivas: Isla de Ometepe, ми slope of Volcán Maderas, 800-1,000 m, 11?27'N, LW, Stevens el al. ог sear ZELAYA: Cerro E. e km 00- 320 m, 1340" М, 84°30' W, Pi- ~ m, Maxon 623 (NY), Maxon 641 (NY): | À a, Las Animas RR station, 500 m, Chrysler 5134 (MO. "UC; pes Río Naranjo, 2 km W of Orosi, 1,400 m, 9?48'N, 83%52'W, Lent 4061 D CR, F, МО); Tapanti, ca. E km S of Paraiso, 1,150 m, Mickel 1862 (NY), 2317 (NY); Carpintera, Polakowsky 15 1 (B); Tur- rialba, slope of the Rio Reventazón, behind the Instituto Interamericano de Ciencias Agrícolas, 600 m, Mickel 2061 (NY), 3359 (NY). GUANACASTE: along Río La as Flo- res, between Quebrada | and Q. Sangui- suela, Hacienda pa 450 m, 10°40'N, 83*04/5^W, Grayum et al. 4906 (MO). LIMON: near Rio ings у mi. SW of на ca. 3 mi. NE of Bratsi, 15 m, Croat 43251 (MO, UC); lago sin nombre al pie de Fila Lleskila— Talamanca, 1,160 m, Gómez et al. 23094 (AAU, BM, Volume 80, Number 1 1993 Murakami & Moran 17 Asplenium sect. Hymenasplenium Monograph MO, UC); base of hills on coastal side, between Punta Cocles and Punta Uva (E of Puerto Viejo de Talamanca), 20-60 m, 938 °43'W, Grayum et al. 4419 (BM, MO, UC); Cordillera de Poss ridge between Rio Dantas (opposite mouth o по Seco) and Rio Barbilla, 250-419 m, 10900'5"М, 83:26'W. Grayum et al. 8943 T d = © 3 2 з Yo e, Talamanca, 300 m, 7.5-8.5 km by road W of Ciudad Colón, ridge between Rio Virilla and Que- brada Micos, along road between Finca Micos and Llano Limón, 550-650 m, a 84?18'W Smith et al. 1620 (MO, UC); 5 mi. 5 о Sta. Мага, 2,000 m Stork 1764 (MICH). Ent NAL AREA: near Frijoles, (UC) Barro Colorado Island, Shannon 8475 (F, NY); Frijoles, 30 m, Killip 2810 (B, S), 2829 (S); along Rio Indio Gatün, 0-10 m, Maxon 4868 (BM, 5); Pipeline road, 8 km from main gate, Mori & Kallunki 2826 (AAU, NY); 8 km NW of Gamboa, along Rio Mendoza and small о 0.5-1 km upstream from Pipeline Road bridge, ,300 m, Maxon 5055 (BM COCLE: El Valle de Antón, 1,000 m, Alston 8707 (BM). COLON: N of Diamante, ridge NW of abandoned mine on Quebrada de la Mina, 600-700 m, 9?24'N, 79*35'W, Churchill & de Nevers 4242 (MO); near Porto Bello, 5- 20 0 m, Maxon 5733 (NY). DARIEN: Parque Nacional del 77917 М, Cuadros V. et al. 3895 (UC); near Cerro Pirre amp, on Rio Parasenico, 200-500 m, Gentry & Clewell 7064 (MO); on Rio Parasenico, 200-500 m, Gentry & Clewell 7069 (MO); RENARE hut in Папеп Natl. Park, 20 m, McDonagh et al. 446 (BM); bank above Rio Paca, Stern et al. 705 (BM, NY, UC). PANAMA: at Rio Tito, Peninsula de Panamá, 275 m, Ekman 15271 (BM, NY, 5); Cordillera Central, Los Haitises, La Llanada, Ekman 15485 (F, G, NY, 5); Sta. Bárbara de Panamá, Peninsula de Panamá, Ekman 15836 (K, S); Tapia e vic. of Juan Diaz, Killip 2710 (S). СОВА. LAS VILL Trinidad Mts., Santa Clara, El Porvenir, 650-75 50 m Britton & Wie: 5300 (NY); San Blas, La Sierra, Prov. Sta. Clara, 200-300 m, Jack 7093 (MO, NY). ORIENTE: Loma del Gato, о 757 (К, MICH, S, О, ОС); Prov. directly S of Jag (NY, S), Otto 5176 (K); Underwood & Earle 989 (NY); E villam Monte Ver- , MO, NY, 5, de, Wright 1026 (6, K, C), Wright 1027 (G, K, L, UC); in 1859, 1860, Wright 1086 (BM, 30 m с Dollwood, Harris А 4 (K) T о 4 5 Б ~ ~ в ч же, | ~. a Maxon 254 from Hoper Bay, Moore 7 (NY); Chepstow Ridge, St. Georges, Moore s.n. (NY); Moy Hall, Orcutt 6835 (UC). TRELAWNY PARISH: 5 mi. N of Quick Step, above Aberdeen P.O., Cockpit country, Proctor 4069 (МО); Tue "à SW of Ecclesdown, E slope of the John Crow Mts., 600 m, Proctor 22690 (BM, MICH, U); Doll е пеаг Silver Hill Gap, 1,000 m, Underwood 2304 (NY); Mans- field, near Bath, Underwood 2802 (NY). DOMINICAN REPUBLIC. Liali, 100-500 m, Abbott 2627 (GH); ravine 0.2 mi. SE Д ари MO, Ekman 4430 (5); Massif рө Cahos, Hinche, Mons Vail- lecite, 600 m, Ekman 6106 (S); Massif de la Tloffe, eastern group, "Mt. жк 1,200 m, Ekman 7332 (S); vic. of St. Louis du Nord, Leonard & Leonard 14363 (NY) Port Margot to Correil, 400-500 m, Nash 199 (NY). PUERTO Rico. Luquillo Mts., Wilson 280 A c MONTSERRAT. In Oct. 1879, Holme s.n. (К). Sr. KITTS. 1889, Berkeley s.n. (NY). GUADELOUPE. In 1868, p we 338 (K), L 'Herminier 18 (K); bois du Maine Gommer bod du Galion, 400-600 m, Duss 4199 (NY). Dominica. Imray 120 (K); Rosalie, Lloyd 699 (NY); Trafalgar Falls, Aa big iul b» ind eig r Hautenis boisies de la Riviere vis 5 (NY). Sr. Lucia. Upper Roseau a Ry ‚ 30 СЕ “Feb. 1936, B (BM); Nov. 1883, if зена 26 (K). GRENADA. ST. PARISH: хы s of id > Catherine, 500-60 Proctor 17250 (U); near мА d D no » ocali, pas 87 (K, NY). Sr. wie wrt, 400 m, Cooley 8393 ', UNKNOWN cen п. ( mith 1362 (BM); in 1890, Smith 358 (С); 1890, H. H. Smith & W. Smith 1121 (K, NY); forest above S end of the inland 300 m, H. H. Smith & G. W. Smith 961 (K). ТОВАСО. Campbelton woods, Charlonevile, Broadway 3026 (B, BM, G, MO, NY, S, 4); Castara woods, Broadway 4212 (BM, G, K), 7180 (BM, F K, MO, S, Z); in 1889, Hart 4166 (К); near Parrot Hall, ca. 2 mi. SW of Parlatuvier, 50 m, Jermy 11370( Johnston 161 (BM); forest reserve of i limestone hill, Fay о — Falls, Johnson 74 (BM); S tio mi. N of Arima, 250 m, Mickel 9516 (NY); without locality, Prestoe 714 (BM, MO). FRENCH GUIANA. ININI: Inini River, en aval de degrad Fourmi Camp de э М les Monts de l'Inini, Cremers et al. 8780 (B, BR, F, Z); Montagne de l'Inini, extremité NW, 550 m, 3?30' 40", 53?36'W, Cremers et al. 8900 (Е, p Sy M de l'Inini, zone central, extremité dd , 3°31'N, 53°35'W, Cremers et al. 9314 с 4 7); Мом Galbao, сас central, Haute Ton Makouali, , 3°35'N, 53718" М, de Granville et al. 8947 (BM, M I VENEZUELA. ANZOATEGUI: NE of Bergantin, Quebr psy Ere of Rio Zumbador, 500 m poca 611 F, MO). ARAGUA: prope Coli Tovar, Fendler m (B, G, GH, MO, NY). BoLIVar: limites con el ан т Delta Amacuro, 36 km al NW de El Palmar, 8°25'N, 62°00'W, Aymard C. к (UC), 5297 (мо) DISTRITO FEDERAL: Caracas, Сћа- , Otto 609 (B, С, K). FALCON: Dtto. Silva, Hacienda Moor ca. 10 km NW de Tucacas, Cerro mo сл Annals of the Missouri Botanical Garden Sanare, 100 m, Burandt 715 (UC); Dtto. Federación, Parque Nacional “Cuevas de la Quebrada El Toro," 500- 600 m, Quijada 103 (NY); Sierra de San Luis, arriba (UC). MERIDA: Mérida, Morita 365 (B, BM). MIRANDA: Cerros del Bachiller, near E end, Quebrada Corozal, S of Santa Cruz, 20-65 m, 10?9'N, 65°48'W, Steyermark & Davidse 116306 (MO, UC). PORTUGUESA: Dtto. Apure, & Aymard 1818 (UC); Dtto. Сите: entre Chabasquen y Córdoba, 1,250 m, Ortega & Stergios 1467 (UC); "Рио. Sucre, Los Paramitos, а 20 km por aire al SO de Biscucuy, 1,000-1,500 m, 9?20'N, 69905 МУ, Ortega et al. 1886 (MO, NY, UC). sucnE: Ortega et al. 1845 (MO) Dtto. Cagigal, near border with Dtto. Arismendi, 850 m, 10939–40"М, 62?43'W, Steyermark et al. 121514 (MO, NY, UC); in 1846, Moritz 23 (C). zuLIA: Dtto. Mara, cuenca del Rio Guasare, alrededores del Desta- camento Guasare no. 1 (La Yolanda) 200-250 m, 10°52'N, 72°29'W, Bunting et al. 12036 (UC); SW de Machiques, a lo largo de la Quebrada Perayra, afluente del Rio Tokuku, SE de la Misión de Los Angeles de Tokuku, 450-475 m, о need (MO). COLOMBIA. ANTIOQUIA: near Guapá, 53 km S of Turbo, 60 m, Haught 4627 (МУ, 5); ia Kalbreyer 1916 (B); Titiribi, Kalbreyer 1499 (B); Urrao, Vereda Calles, arque Nacional de Las Orquideas, cabana del INDERE- NA en Rio erp quebrada La Agudelo, 1,450 m, Ca- llejas et al. 284 ); Tocarema, 2,200 m, Lindig 236 BM). CHOCO: e San José del Palmar- Noria. entre La Italia (pueblo nuevo) y Curundo, 430-450 m, Forero 134 (MO); near Caserío La Teresita, 0.5- 2.5 km N of the INDERENA Camp on Río Truando, 50-100 m, Lellinger & de la Sota 550 (COL); us Schott 51 (F, MO), 755 (MO); йк Nacional Natural Los Katios, sector Tilupo camino del Salto, {ку R. 332 (COL). MAGDALENA: Santa Marta, Finca El Re cuerdo, , 800 m, Bennett 20 (F); Arroyitos in icon N of house, 1,800-1,900 m, 10%56'N, 73°58'W, Kirkbride 2368 (COL); 1,000 m, H. H. Smith 2450 (BM, F, MICH), 958 ur MICH, S). CUNDINAMARCA: Pueblo Nuevo, quebrada "La Guay acana,' m os m, Murillo 363 (COL. NORTE DE SANTANDER: Oca 2,000 m о (В, К); 1846-1852, Schlim 397 (B, "BM, m ANDER: Es- 7 (B) 1 TOLIMA: Ca- lh M: ‘ is | 79°33'W, Croat 55623 (F, МО). Los RIOS: Río Palenque Biological Station, km 56 Quevedo-Sto. Domingo, 150- 220 m, Evoy 61 (NY), Lojtnant 15757 (AAU). MANABI: El Recreo, Eggers 15118 (BM, F, L, M), 15175 (F). MORONA-SANTIAGO: cerca la Parroquia de Bomboiza, en la carretera Gualaquiza- Zamora, alrededores del puente sobre el Rio Bombioza, 800 m, Baker & Zaruma 6476 (MO, NY). NAPO: Parque Nacional Yasuni, Anangu, 260-350 m, 76°23'W, 0°31-32'5, Mllgaard et al. 39205 (AAU, NY). PICHINCHA: Loc. 170-175, Sto. Domingo- Quinindé, 300 m, Acosta S. 13783 (F). PERU 12 km E HUANUCO: Leoncio Prado. La Cueva de las Pavas, 5 Ып 5 de Tingo Maria, Dtto. Кира Кира, 672 m, Schunke V. 3261 (F); Tingo Maria, on steep rocky slope above Rio Huallaga, Croat 21082 (MO); hills above river on steep limestone forested slope, 650 m, Moran 3683 (MO), m, Tryon & Tryon 5228 (BM, F). JUNIN: Cochero, m, Pavón 186 (С); Chanchamayo, 1,500 m, Se Пе 20 (S), 4909 (BM); area of Pichita Caluga, Walden 8 84 (BM). Pasco: Oxapampa, 50 km from Оха- pampa, Rio El Tunqui, 1,620 m, 75°30'W, 10%15'S, Smith et al 1710 (MO). SAN MARTIN: Juan Jui, Alto Rio Huallaga, 400- a Klug 4253 (BM, F, МО, NY, 5). BOLIVIA. LA PAZ: Prov. Iturralde, Tumupasa, 500 m, Williams 107 6 (NY, UC); 600 m, Williams 1076a(NY). COCHABAMBA: Prov. Chapare, vic. o a Tunari, aoe Rio Espirito e on trail to Baja Copacabana 16°57 512: 65?25'W, Croa (MO). LA РА М прав, Polo-Polo bei oca 1,100 m, Hüchiien 3334 (S) RAZIL. STATE UNKNOWN: no locality given, Mendonca 1208 (B). РАКА: eie airstrip, 1.5 hours "ape from Lageira airstrip, Rio Maicuru, 0%55'S, 54?26'W, Strudwick et Be 357 74 (NY). RIO GRANDE DO SUL: Cruz, Jurgen Glaziou 5318 | . ARGENTINA. SALTA: Parque Nacional El Rey, Brown 1 338 (МО). Asplenium laetum grows terrestrially in deeply shaded forests or on boulders along streams. It is the most common and widely distributed species of section Hymenasplenium in the W orld (Map 4), ranging from Mexico to northern Argen- tina, but absent from most of Amazonia. The spe- cies has sometimes been reported from Africa and Madagascar, but thes imens of A. inaequilaterale Willd., a species that e reports are based on spec- does not belong to section Hymenasplenium. The species is characterized by dark petioles and rachises, short-creeping rhizomes, and atten- uate lamina apices (Fig. 8A). In other characters it is extremely variable. Some single-character vari- ation is correlated with geography. For example, plants from the Canal Area in Panama usually have obtuse or retuse pinnae with nearly entire or shal- lowly serrate margins (e.g., Croat 7008, Kennedy 2673, Béliz 150). Plants from Bolivia and Peru have attenuate pinna apices with sharply and obliquely serrate margins (e.g., Croat 51271, Klug 1253, Schunke V. 3261). We have not, however, subdivided the species because no other charac- teristics correlate with this variation, nor are there any suites of quantitative characteristics that cor- relate with geography (Fig. 9). Future studies, es- pecially of isozymes and chromosomes, may show that 4. laetum as circumscribed here actually con- sists of several species. Occasional specimens will key to А. /aetum that are actually intermediate between А. delitescens and А. laetum (Fig. 6B). Such intermediate spec- imens are treated in the hybrid section of this monograph. Several species of Asplenium that do not belong Volume 80, Number 1 Murakami & Moran 19 1993 Asplenium sect. Hymenasplenium Monograph B FIGURE 9. Principal Components Analysis of 30 characters of Asplenium laetum. BE, Belize. BR, Brazil. BO, Bolivia. CO, Colombia. CR, Costa Rica. EC, Ecuador. GU, Guatemala. NI, Nicaragua. PA, Panama. PR, Peru. VE, WI, West Indies. Venezuela. to section Hymenasplenium are frequently mis- identified as A. laetum because of their similarity in leaf cutting and outline (1.е., А. argentinum Hieron., 4. hostmannii Hieron., and А. otites Link). These species can be distinguished from A. laetum by their erect rhizomes and greenish petioles. Be- sides these characteristics, they also differ from .4. laetum in the form of the rachis and costa junc- tures. Asplenium laetum has the form typical of section Hymenasplenium: the costa is slightly sunken and bordered by a flange of green tissue that is decurrent on the adaxial surface of the rachis, and the decurrent margin of the pinna is not thickened (Fig. 16D). In contrast, the other Asplenium species mentioned above have prominu- lous costae not bordered by a flap of green tissue, and the decurrent margins of the pinnae are thick- ened and often a lighter color than the rest of the laminar tissue (Fig. 160). 4. Asplenium obtusifolium L., Sp. Pl. 1080. 753. Asplenium riparium Liebm. var. ob- i ae (Klotzsch & Karsten) Hook., Sp. . 1860. TYPE: Petiver, Pter. Amer. ae 5 T fig. 14 (erroneously cited by Lin- naeus as f. 4). [= Plumier, Traité Foug. Amér. t. 67. 1705.] Based on a plant from Morne de la Calebasse, Martinique. Figure 10; Map 5 Asplenium aquaticum Klotzs ch & Karsten, Linnaea 20: 354. 18 In locis palu- Karsten s.n. n. (holotype, B; photo. GH ex B). dosis," Plants epilithic; roots 0.6-1 mm wide; rhizome 2.5-4.5 mm wide, nearly mg ш» projecting apex sparsely scaly; scales 0.9- 1. -0.4 mm, brown, deltate-acuminate; он we Lo 3.5 mm wide, swollen each other o mm distant from me row; Sinon 4-10. ¿10,25 oem, У ў the leaf length, terete, green- ish, fleshy and dull, glabrous; ae 8-20 cm dark green, translucent, 1— long, 2-pinnate (lowest pair of pinnae usually lobed or divided basally), deltate to oblong with subconform apex; rachis like petiole, narrowly marginate adaxially; lateral pinnae 2.5- 4.5 х 0.8-1.5 cm in the middle of the leaf, 3- 10 pairs, slightly to strongly ascending, obliquely oblong to oblong-lanceolate, often laciniate basally into 1-3 segments, simply toothed, the teeth 0.5- 1.5 mm long, the basiscopic margin excavate ca. 14 the pinna length, the apex usually obtuse, rarely acute, auricle 1-2.5 mm long; veins l-forked at middle part of pinnae; sori 4.5-9 mm long; spores papillate, 27—33 um long for race with 64 spores per sporangium, 42-47 um long for race with 32 spores per sporangium. 20 Annals of the Missouri Botanical Garden ЖА LZ LAN pS WAS AN XQ N Xx FIGURE 10. H. H. Smith 1126 (NY). —B. Venezuela, van der Werff 702 (MO). — C. Grenada, Sherring 131 (K). Additional specimens examined. PANAMA. BOCAS DEL TORO: 6.3 mi. N of bridge over Fortuna Lake, between Fortuna and Chiriqui Grande, 2.2 mi. N of Continental Divide, 820 m, 8?45'N, 82?16'W, Croat & Grayum 60385 (AAU, MO, UC); Fortuna Dam region, along road to Chiriquí Grande, 950-1,000 m, 8?*48'N, 82°10'W, McPherson 8666 (MO). cocLE: El Cope, forest on Con- Asplenium obtusifolium. — А. Colombia, K) tinental Divide above town, 700-900 m, 8?38'N, 80°38'W, Hammel 13645 (MO, UC). VERAGUAS: vic. of Escuela Agricola, Santa Fé, near trail to top of Cerro Tute, along stream, 800 m, Antonio 3530 (MO); 11-13 km beyond Agricultural School at Santa Fé, valley of the Río Dos Bocas, 350-500 m, Croat 25732 (MO); 6.5 km from Santa Fé, base of Cerro Tute, Folsom 3062 (MO); Volume 80, Number 1 1993 Murakami & Moran 21 Asplenium sect. Hymenasplenium Monograph The distribution of е obtusifolium. The 32-spored race occurs west of the dashed line; the 64- lin spored race occurs east of the dashed NW of Santa Fé, 8.8 km from Escuela Agricola Alto de Piedra, Mori & Kallunki 3964 (MO). PUERTO Rico. Toro Negro, 1936, Quick s. n. (MICH); prope Utuado, Sintenis 627 1 (B, BM, G, S), 6443 (B, BM, G, K, L, NY), 6533 (B, BM, C, GH, K, L, 5); Toro Negro, 50 km WSW of San Juan, Tryon 7399 (BM, GH, MO, NY). GUADELOUPE. No locality, L'Herminier s.n. (B, BM, GH, L, MO, NY) in 1862, L'Herminier 79 (B, BR, С, К), 769 (L). Domi- NICA. Rosalie, 1903, Lloyd 698 AM Mt. eed Lloyd 871 (NY); path from Fresh Water Lake to Boe Lake, 800 m, е & Adams 98 (MO): Sylvania Estate, gorge between the estate house and orange tation, 488 m, Hadar 1092 (GH); NE slopes of Morne Diablotin, 800 m, Hodge & Hodge 2793 (GH), Imray 13 E David Parish, along Boeri Lake Trail near Fresh Water Lake r oad, 1,200 m 571 (G oeri La e, Whiteford 417 4 (BM): AW slopes of Morne Diablotin. Whitefoord 4549 (BM); slopes of Mi- cotrin, along path NW of Fresh Water Lake toward Boeri 0 m, Wilbur et al. dg St. dba Morne Gonivo, 900 m, Eggers 608 (B, BR, С, L, MICH, 5, UC); no locality, in 1843, ps s.n. (L). [eund Jan. 1868, Hahn 68 (B, BM, G, GH, K, S); Calebaye, Belanger 801 (B, G); no locality, Duss 1636b (F, G, GH, МО); bois des Barris-Taunes, 450-800 m, Duss 4205 (NY); no locality, Husnot 334 (B, BM, BR, F, C), Pitard s.n. (С). Sr. Lucia. e Edmund District, SE of Piton Troumassée, 600-650 m, Proctor 17703 (Uy; Morne Calebasse, N of Morne “Rouse, 650-760 m, Proc- tor 21691 (GH, U); no locality, Sieber 363 (BM, M); in 1833, Sieber s.n. ape С); source of Canaries River, 800 m, Box 456 (BM); Upper Roseau River, 300 m, Box ч 3 (BM); St. Georges, Annandale, Broadway s.n. , NY). Sr. VINCENT. Majorca estate, 400 m, Barron Des 74F (BM); Lisdara, Hades 2410 (GH, NY); upper Lellinger valley of Richmond River, 330-540 m, Morton 6170 (GH); without locality, Smith & Smith 7 10 (BM). GRENA- DA. Sherring 131 (K, NY); NW slope of Mt. St. Cath- erine, 500-700 m, Proctor 17254 (GH). Tobaco. Vil- el-Karst, main ridge, Broadway s.n. (BR, G, GH, MO, Britton et al. 1656 (GH, NY); Maracas, Broadway 5353 (BM, BR, F, MO, UC); Maracas Falls, 200 m, Fay 673 (K); in nah 1880, Fendler 139 (B, BM, G, GH, K, MO, NY, UC), Han 241 (S), Holdridge 50 (K), Monterey 68 (8 hee Wright Nature Centre, 400 m, Fre) 9446 NY) VENEZUELA. ANZOATEGUI: NE of Ber gantin, Quebrada Negra, tributary of Rio Zumbador, 500 m, ad 61148 (F, С, MICH, MO, NY, 5); Cerro Los Pajaritos, between Cerro San José and Cerro Peonia, along headwaters of Quebrada La Tigra, 1,500-1,600 m, Steyermark 6 1569 (C). ARAGUA: Henri Pittier National 3k of summit, 970 m, 10%20'N, 67934 МУ, Croat 60574 (MO); prope Coloniam Tovar, Fendler 131 (BR, G, GH, K, MO, NY); Colonia Tovar, Morit BR, P, photo GH, MICH m, Pannier 132 (M). BARINAS: Dtto. Bolivar, entre Sto. Domingo y Altamira, en el sitio denominado bees 1,300-1,500 m, Ortega & van der Werff 2358 UC), Ortega & van der Werff 2348 (М). CARABOBO: es cabeceras del Rio San Gian, 750-900 m, Steyermark 95249 (NY). DISTRITO FEDERAL: Depto. Li- bertador, entre La n y Chichiriviche, a lo largo del Río Chichiriviche, 450 10°31'N, 67*14'30"W, Stey- ermark & Espinoza 112683 (MO, NY). FALCON: Sierra 22 Annals of the Missouri Botanical Garden e M ш” - ENN 3 ze . К а" a CM | Wig Za Soy y = NS = КОО с сша al d “хх NS у Se NY Es E Ss PEW Y f Е ИРА о ч, fà 4 а 42272 27272277) EEES 00277295 Зд RODS > SS [| SQ | et al. 2720 (AAU). —B, С. А. ortegae, Holm-Nielsen RE ll. Asplenium purpurascens, Ecuador, (MO). Ficu =A, Venezuela, van der Werf & González 4599 Volume 80, Number 1 1993 Murakami & Moran 23 Asplenium sect. Hymenasplenium Monograph de San Luis, arriba de La Chapa, 1,200 m, van der Werff m, Ortega & R. F. Smith 2440 (00). MIRANDA: Dtto. Page Fila La Tigra, Hda. San Juan, 18 km SW de Cüpira, 05'N, 64?45'W, Ortega & González . UC); Cerros del Bachiller, above Quebrada Сакош, 5 of Sta. Cruz, 10 km W of Cüpira, 600 m, 10?9'N, 65?48'W, Steyermark & Davidse 116896 (MO, UC). sucre: Pen- insula de Paria, Cerro de Humo, ca. 14 km N del pueblo de Rio Grande Arriba, arriba de Boca de Cumaná y Punto Siparo, МЕ de Irapa, 1,060 m, Steyermark 94822 (NY); Peninsula de Paria, Cerro de Rio Arriba, W de Cerro de Humo, a lo largo de js Кая Isabel, arriba de Santa Isabel, 600- 700 m, Steyermark & Rabe 96232 (NY); Dtto. Cagigal, юнга de "Paria ria, near border with Dtto. Arismendi, low trail between El Pauiil and El Brasil, 850- 890 m, 10°39-40'N, 62°43’ teyermark et al. 121451 (MO, NY, UC); Dtto. нео. Peninsula de Paria, between Пије and Ко i umo, , 10°41 -42'N, 62°3 'W et al. 2 (МО, ы Conci. PROV. Karsten s.n. (B). MAGDELENA: Sierra Neva Marta, Alto Rio Dern Alto de Mira, 900-1,10 Madriñán & Barbos vri Pins МО); Sierra Nevada de Santa Marta, edge of a La Sirena, 1,100 m 10%59'N, 73*59'W, Kirkbride 2188 (COL, NY, UC): NORTE DE SANTANDER: Ocana, 1,600 m, Schlim 653 (B, BR, G, K, L). SANTANDER: Tequendame, 1,200 m, Kal- breyer 951b (B). Asplenium obtusifolium has a circum-Carib- bean distribution (Map 5). It grows in the spray of, or underneath, waterfalls and on wet rocks along shaded streams. Из elevational range is 200 to 17 т. This species is characterized by greenish peti- oles, fleshy leaves, small (8-20-cm-long) laminae, and few (3-10) pinna pairs. The pinnae are usually obtuse and the basal pairs lobed or divided basally (Fig. 10). Asplenium obtusifolium is remarkable by hav- ing two races: one with 32 spores per sporangium, the other with 64 spores. The sporophytes of the two races appear indistinguishable and could not be distinguished by a Principal Component Analysis of 20 quantitative morphological characters (Fig. 5). The two spore races differ geographically, with the 32-spored race occurring only in the western portion of the range and the 64-spored race oc- curring only in the eastern portion (Map 5). The 32-spored race is presumably apogamous, but this idea needs to be investigated. In the Old World, 4. apogamum Murakami & Hatanaka, 4. cheilosorum Kunze, and А. hondoense Murakami & Hatanaka have been shown to have both a sexual race with 64 spores per sporangium and an apog- amous race with 32 spores per sporangium (Mitui P6. The LE of Asplenium purpurascens "n к and А. ortegae (open triangles). et al., 1989). Therefore, it seems likely that the 32-spored race of 4. obtusifolium is apogamous. 5. Asplenium ortegae Murakami & R. C. Mo- TYPE: Venezuela. Apure: Dtto. Paez, along Rio Arauca at Colombian border, van der Werff & González 4599 (holotype, MO; isotype, NY). Figure 1 1B, C; Map 6 ran, sp. nov. Plantae terrestres. Rhizoma 3.5-5.2 mm latum, in- ternodiis approximatis, squamis 0.5-1.9 mm longis, ca. 0.2 mm latis, lanceolatis, clathratis, brunneis vel nigris- centibus, 3- 6- cellulas latis. Petioli 6-21 cm longi, atro- жө sparse pee vel glabrata. Venae (2- furcatae. Sori 3-7 mm longi. Annals of the Missouri Botanical Garden Plants terrestrial; roots 1.1-1.7 mm wide; rhi- zome 3.5-5.2 mm wide, nearly naked; scales 0.5- 1.9 x 0.23-0.29 mm, black-brown, lanceolate; petiole base 2.2-5.5 mm wide, swollen, 0.5-1.1 mm distant from each other on the same row; petiole 6-21 0.09-0.15 cm, Vs-V4 the leaf length, atropurpureous, sparsely scaly at base; lam- ina 11-25 ight green, herbaceous, l -pinnate, deltate to oblong with acuminate apex, apex pinnatifid, abruptly reduced or evenly ta- pered; rachis green to dark reddish purple, grooved cm long, with green herbaceous wings perpendicular to the plane of the lamina; lateral pinnae 5-9 x 0.9- 1.7 cm in the middle of the leaf, 6-10 pairs, lanceolate to linear-lanceolate, the margin slightly double-toothed, 0.5-1.3 mm deep, the basiscopic side excavate for /,- у, the length of the pinnae, stalked at lower pinnae, auricle absent; veins 3-4 ~ times forked; sori 3-7 mm long; spores about 30 um, 64 per sporangium, perispore cristate. Paratypes. | VENEZUELA. PORTUGUESA: Рио. Ospino, Quebrada Honda, en el caserio La Cristobera, Ortega et al. 3093 (BM); Dtto. Guanare, Mpio. izquierda del cano Iguez, Pérez 1 (UC); terrenos del la UNELLEZ, una quebrada en la Mesa Alta, 9%4'N, 69°49'W, Ortega 1912 (UC); рио. Araure, еп los Rios Bocoy y Riecito, limite entre los Estados Lara y Portuguesa, 800-1,200 m, 9%37'N, 69°25'W, Ortega & Aymard 1817 (UC); entre Agua Blanca y Los Hijitos, 160 m, Ortega & Aymard 2087 (NY); entre Agua Blanca y San Rafael, Quebrada Los Hijitos, 160 m, Ortega & Aymard 2088 (UC); Dtto. Guanare, along road to San José de La Montana, 500 m, 9?20'N, 69°43'W, A. R. Smith et al. 1062 (UC); Boca de Monte, 23 km N del vado del Rio Suruguapo, 38 km N por la drei Guana- re-Ospino, en sitio Las Marias, 400 m, 9?18'N, 69437, Steyermark et al. 127 134 (MO, UC). ZULIA: Duo Perijá, W of the intersection of the Rio Aricuaisá and the Maracaibo-La Епа Hwy. (Hwy. 6), 10 m, Davidse et al. 18205 (MO, UC); Dtto. Colón, Пао рй of Rio ui Maracaibo Hwy. (Hwy. 16), 30 m, Davidse et al. 18826 (MO, UC); Boca e Monte, 23 km el vado s Rio Suruguapo, 3 p | n n о Le Маг s, 400 m, Steyerm al. 127131 (MO, UC). ы. CUNDINAMARC A: (adis Medina desviación a la izquierda, Rio пе 500-550 dl (МО). МЕТА: i Guatiquia, Lehmann Ó Little үз Ti ttle 8427 cone DOR. PA i m upriver on Rio Curar жое Military Camp, 200 m, 1928'5, 75*58'W, Brandbyge & Asanza C. 31225 (AAU); Cei- lán, Pica from Ceilán to Rio Cononaco, on the N side of UC). PERU. JUNIN: Colonia puse 680 m, Killip & Smith 24924 (F); Puerto Ber- múdez, 375 m, Killip & Smith 26633 (F). LORETO: Yurimaguas, lower Rio Huallaga, 135 m, Killip & Smith 29078 (NY). BRAZIL. ACRE: Cruzeiro do Sul, Rio Juruá and Rio Moa, between Mundurucus and Tatajuba, Rio Juruá, Maas et al. P12902 (F) Asplenium ortegae is named for Francisco Or- tega, prominent Venezuelan pteridologist, who has given us much hospitality and assistance during fieldwork in Venezuela. From Colombia to Bolivia, the species occurs on the eastern side of the Andes; in Venezuela, it occurs on both sides (Map 6). It grows on shaded forest floors in wet forests from O to = m. Asplenium ortegae is characterized by short- creeping rhizomes, atropurpureous petioles, trun- cately serrate-lobed pinna margins, and usually 3 or 4-forked veins (Fig. 1 1B, C). Rarely, the basal pinnae become pinnatifid, and the plants then re- semble 4. purpurascens (which also has atropur- pureous petioles). The two species may be sepa- rated by the characteristics given in the key. Despite the resemblance to А. purpurascens, evidence from chloroplast DNA supports a close relationship of 4. ortegae with А. delitescens (this evidence will be presented in a future paper by the first author). Morphological characteristics also support this relationship. Both species have deltate laminae, relatively few (6-10) pinna pairs, and similarly shaped laminar apices. Asplenium deli- tescens, however, differs by having stramineous petioles, less serrate pinna margins, and twice- forked veins (Fig. 6A). It may eventually be shown that А. ortegae is of hybrid origin involving 4. delitescens as one of its parents. 6. Asplenium purpurascens Mett. ex Kuhn, Linnaea 36: 1869. TYPE: Ecuador. Chimborazo: “ad pedem montis Chimborazo,” Spruce 5697 (holotype, B not seen; fragment. B). Figure 11A; Map 6 Asplenium melanopus Sodiro, Anal. Univ. Quito 9: 88. 1893 [= Crypt. Vasc. Quit. 189. 1897]. Diplazium . (isosyntype, Pallatanga cerca de Puente de Chimbo," (isosyntype, K) Sodiro s.n. Plants terrestrial; roots 1.5-2 mm wide; rhizome 4.2–6.5 mm wide, sparsely scaly; scales 1.6–2.3 x 0.35-0.55 mm, black to dark brown, lanceolate; petiole base 2-5 mm wide, swollen, 2—4 mm distant from each other on the same row; petiole 17-27 х 0.15-0.2 em, about !4 the leaf length, dark reddish purple, scaly at base; lamina 20-30 cm long, dark green, firm-herbaceous, 1-pinnate, del- tate, the apex attenuate, pinnatifid; rachis like the Volume 80, Number 1 1993 Murakami & Moran 25 Asplenium sect. Hymenasplenium Monograph petiole, grooved, with herbaceous green wings; lat- eral pinnae 6.5-12 x 1 m in the middle of the leaf, 8-11 pairs, lanceolate to linear-lan- ceolate, long-acuminate, the basiscopic side exca- vate for /—'/ the length of the pinna, stalked at the lower pinnae, auricle absent, the margin lobed to pinnatifid, the tip of each lobe with rounded teeth, about 1 mm deep; veins 3-4 times forked; sori 4-8 mm long, often diplazioid; spores 27-33 um, 64 per sporangium, perispore cristate. Additional specimens examined. ECUADOR. CHIM- BORAZO: vic. of Sacramento, on E side of Rio Sacramento 1,200-2,000 m, Wiggins 11095 (NY, UC). COTOPAXI: Tenefuerste, Rio Pilalo, km 52-53, Quevedo- Latacunga, 750-1,300 m, Dodson & Gentry 12275 G QCNE); 20 km NW of Corazón, Rio Guapara, ‚ Sparre 17173 (S); in confiano Columbiae, Du 43 (UC); an Mateo, road to Esmeraldas airport, gravel road 8.6 km beyond bridge over Rio ооо yond Univ. Luis Vargas Torres Est. MO, UC). GUAYAS: near Bucay, western cordillera, junc- tion of the provinces of Guayas, Canar, оа и Bolivar, 300-400 m, Сатр 3785 (Е, GH, NY); L Alta, Cerro de Vias Muertas, Cordillera de Соот 250 m, Valverde 510 (МО). Loja: W of El Limo, Alam Cazaderos Road, 1,400 m, Harling & Andersson 22310 (F). Los RIOS: Rio Palenque Biological Station, km 56 Rd. о ђа roundings of Монака, 100- 200 m, 1%47'S, 79°17'W, 2,720 m, Holm-Nielsen et al. 2720 (F, M 100-200 m, 1?47'S, 79*17'W, Holm-Ni (AAU, B). MANABI: 11 km E of San P d Portoviejo-Pichincha, 400 m, Harling & Andersson 24978 (Е). PICHINCHA: Km 170-175, via Sto. Domingo- Quinindo, 300 m, Acosta S. 13815 (F); Reserva EN- DESA, Carretera Quito-Puerto Quito, km 113, 00%05'N, 79?02'W, Arguello 498 (AAU); | ENDESA, Car- retera Quito- Puerto Quito, km 113, 800 m, 00*05'N, 79?02'W, Ayala 58 (AAU); San ЈЕ de Toachi, 100 km W of Quito, 1,000 m, Bell 200 (BM, GH, 5); by Río Toachi, above confluence with Río Pilaton, 1,000 m, Bell 238 (BM, GH, 5); 20 km W of Sto. Domingo de los Colorados, 300 m, Cazalet & Pennington 5199 (B, K, y 20 km W of Sto. Domingo de los Colorados, 300 m, Cazalet & Pennington 5224 (UC); Sto. uA de los Colorados, Fagerlind & Wibom 1593 (5); Sto. Do- mingo de los p dne 500 m, Holdridge 1617 (GH); Finca Carlita, m 13 on road to Sto. Domingo, 550 m, 00915'5, To" W. Holm-Nielsen et "m 7066 (AAU); valle Nanegal, Aug. 1874, Sodiro 8.74 (K); prope Sto. Domingo, Aug. 1875, Sodiro 8.75 (K). Asplenium purpurascens is largely restricted to the western side of the Andes of Ecuador (Map 6). It grows in wet forests from 80 to 1,400 m. The species is characterized by short-creeping rhizomes, atropurpureous petioles and rachises, and pinnatifid pinnae (Fig. 11A). No other neotropical species of section Hymenasplenium has regularly pinnatifid pinnae. The holotype of А. purpurascens was not among the specimens we received on loan from B. There were two very small fragments on a sheet at B, but these were not enough to identify the species. We feel confident that we are applying the name correctly because it is possible to distinguish the species based on Mettenius's original description. According to his description, 4. purpurascens has creeping rhizomes, lustrous purple petioles, and pinnatisect pinnae. also agrees with applying the name to the above- cited plants e provenance of the type 7. Asplenium repandulum Kunze, Linnaea 9: 65. 1834. TYPE: Peru. Huánuco: Pampayaco, in sylvis montosis ad arborum truncos, July 1829, Poeppig s.n. (holotype, B not seen; isotype, NY (fragment) not seen). Figure 12; Map 7 Plants epiphytic; roots 0.6–1.5 mm wide; rhi- ome 2.5-7 mm wide, nearly naked; scales 0.5- 2.5 x 0.2-0.5 mm, black, ovate-lanceolate, clath- rate; petiole base 2.5-4 mm wide, swollen, 1-2 N cm distant from each other on the same row; petiole 4-15 x 0.15-0.3 cm, about /—Y, the leaf length, dark green-brown, fleshy and dull, terete, glabrous; lamina 22-40 cm long, dull green, 1-pinnate, lan- ceolate to oblong with subconform apex; rachis like petiole, with adaxial green wings in the same plane as the lamina, strongly carinate abaxially; lateral pinnae 4-6 x 1.1-1.8 cm in the middle of the leaf, 12-17 pairs, obliquely oblong to oblong-lan- ceolate, margin almost entire to only slightly toothed, the basal basiscopic side excavate for /—14 the pinna length, acute at apex, broadly cuneate and almost sessile at base, auricle 2-6 mm long; veins 1-forked; sori 4-10 mm long; spores 40-48 um, 64 per sporangium, spiny. Additional examined. ECUADOR. MORONA-SANTIAGO: Macuma, on the Macuma River, S of Chiquaza, 700 m, van der We erf 660 (GH, MO). PASTAZA: anos— aut Jivaria de Pintuc, excurs. nach Canelos, 1,200-1,300 m, Stübel 878 (B). PICHINCHA: prope Sto. M de а E Sodiro s.n. (K). P AMAZONAS: of eca, Serrania de cada 1,650- E m, ко ен et al. 23073 (MO, UC). HUANUCO: Leoncio Prado, Dtto. Hermilio Valdizan, road from Puma- La КА 1; : Chanchama о, (F), 907 (F); о San Ramón, Schunke Ha 0 1,700 m, Killip & Smith 24547 (NY), 24804 (NY); Pichita Caluga, 1,800 m, Walden 41(BM); Tarma, Chan- chamayo Valley, above La Merced at Cumbre Yacunay 26 Annals of the Missouri Botanical Garden ES Q P, А ° C Len GF А nA И СА y M MM pA A AA REX di d we 0 ==> ss Va pe e 7 > = RZN gr э? QUE e N EN SSNS ы, | ss #7 / A И, y ESOS sy JE ЕД а A EET AN VA Ё mS B = =з С) NUS T ELLA p— ZE, ZN WEF - < У SVE, 7 SN SN О р“ ass o CAS o к ZZ S| = KS SK] i) ЕЩ SS Sr А ae Im | Nyy < 22225 | Сг FIGURE 12. Asplenium repandulum іп А. — С. Peru, Hutchison 1196 (С) A, B. Ecuador, van der Werff 660 (MO). Cross section of rachis shown Volume 80, Number 1 1993 Murakami & Moran 27 Asplenium sect. Hymenasplenium Monograph near Ren 2,000 кыены у i 1196(F, G, M, MICH, Y, S). MADRE DE DIOS: M (F); reise von Moyabomba zum Huallaga Stübel 1099a (B). Pasco: Oxapampa, vic. of Chequitavo, Gran Pajonal, 1,250 т, 10°45’S, 74°23'%, D. N. Smith 5168 (MO). Asplenium repandulum is characterized by long- creeping rhizomes, prominently carinate rachises, and orientation of the rachidial wings in the same plane as the lamina (Fig. 12). It occurs east of the Andes in Ecuador and Peru (Map 7). It grows in wet forests from 700 to 2,000 m as an epiphyte on large tree trunks. One specimen from French Guiana, disjunct from the Andean populations, has a carinate rachis and may represent 4. repan- dulum but was collected on wet soil (Sommet Ta- bulaire, ca. 50 km SE of Saul, 750 m, de Granville 3508 (F)). In the carinate rachis, it also resembles A. triquetrum, but the main range of that species is even farther away than that of A. repandulum. It is not typical of either species to grow on the soil. The identity of the plant from French Guiana needs more research. As outlined in the key, the rachis cross section and orientation of the adaxial wings on the rachis are important taxonomic characteristics distin- guishing А. repandulum, А. triquetrum, А. ri- parium, and A. volubile. The characteristics, how- ever, can be difficult to interpret because of changes during pressing and drying. Although the carinate rachis of А. repandulum and А. triquetrum is a prominent and striking feature of living plants, in pressed plants the keeled edge of the rachis is pushed to one side, so that only one side of the 3-angled rachis faces up. This gives the impression when viewed from above that the rachis is smooth or rounded. When this distortion occurs, it can usually be detected because pinna bases on one side of the rachis are hidden beneath the keel, but the attachments of the pinna bases on the other side are visible. he orientation of the rachidial wings may be difficult to determine because the wings sometimes shrivel upon drying and (in 4. riparium and 4. volubile) the wings may be pressed flat into the same plane of the lamina. Usually, the best place to observe the orientation of the wings is at the pinna junctures. Asplenium repandulum is most closely related to A. triquetrum based on rachidial wing orien- tation and the strongly carinate rachis. However, A. repandulum resembles А. volubile by the long- creeping rhizomes and epiphytic habit (see А. volu- bile for comparison). Map 7. The distribution of Asplenium repandulum. Although we have not seen the holotype of 4. repandulum, we feel confident that we are applying the name correctly because this is the only epi- phytic species of section Hymenasplenium that occurs in Amazonian Peru. 8. Asplenium riparium Liebm., Mexic. bregn. 244 (seors. 92). 1849. [= Kongel. Danske Vidensk. Selsk. Skr., Naturvidensk Afd., s 5, 1: 244. 1849]. Asplenium obtusifolium L. var. riparium (Liebm.) Domin, Pteridophy- ta Isl. Dominica in Rozpr. Král. Ceské Spo- lecn. Nauk, Tr. Mat.-Prir., Nov. Rad. 2: 175. 1929. LECTOTYPE (designated by A. R. Smith, 1981): Mexico. Veracruz: Hacienda de Java, Liebmann [Fl. Mex. 310] (lectotype, C not seen; isolectotypes, B, K). Figure 13B; Map 8. Plants epilithic or terrestrial; roots 0.7—1.3 mm wide; rhizome 3-6 mm wide, nearly naked; scales 0.8-1.5 x 0.2-0.4 mm, brown, ovate-lanceolate; petiole base 2.5-4.5 mm wide, swollen, 1-2.5 mm distant from each other on the same row; petiole 10-20 x 0.1-0.25 cm, 44-14 the leaf length, dark green-brown, fleshy and dull, broadly and shallowly grooved, glabrous; lamina 15-30 cm long, dark green, 1(-2)-pinnate, lanceolate with conform 28 Annals of the Missouri Botanical Garden SSSA E => pe 3 c" ===> SS = LD, AS NS ESOS SS = NR \ NS \ ES SN p NSERM W NE SSN \ Ne f 22 auia 77 d pp == SO ay а LEG «М. ip | y B A 7777 GE É \ P а 1 WA yoo PS LEE oy 1 чыш Ё Й AY 22 ES WY eS — ZG, De 2 22 AZ 7 > YW 27 SSN у“. ; = |. | x Z PE — |, E a NN = = у © ИУ МИ 5 Si ( m \ 6 "p V М Ket о мод us AT FicURE 13. A-D. A. riparium. — A. Cross section of rachis. — B. Fertile leaf, Colombia, H. H. Smith 968(MO).— C. Senescent leaf with plantlets, Murakami s.n. (MO). — D. Plantlets proliferating from receptacular tissue of senesced leaf. —E. Asplenium volubile, Ecuador, Asplund 5550 (F). Volume 80, Number 1 1993 Murakami & Moran 29 Asplenium sect. Hymenasplenium Monograph terminal pinna, herbaceous; rachis like petiole, with green wings perpendicular to the plane of the lam- ina; lateral pinnae 4-8 х 1-2 cm in the middle of the leaf, 8-14 pairs, oblong-lanceolate, the mar- gin toothed, the teeth about 0.5 mm deep, the basiscopic margin excavate for /-/, the pinna length, broadly cuneate and almost sessile at base, auricle 1-4 mm long; veins 1—3 times forked; sori 5.5–10 mm long; spores 38-43 um, 64 per spo- rangium, spiny. n — 39 or 40 (Smith & Mickel, 1977) Additional specimens examined. MEXICO. CHIAPAS: Mpio. of Rayón, in the Selva Negra 10 km above Rayón Mezcalapa along road to Jitotol, 1,700 m, Breedlove 26082 (F, MICH, MO, NY), 29819 (F, MICH, NY), 32617 (MICH, NY); Mpio. of Rayón, above Rayón Mez- calapa along road to Jitotol, 1,700 m, Breedlove 35410 (MO); Mpio. of Cintalapa, 3 km E of Francisco Маши, NE of Cintalapa, 1,250 m, Breedlove 38081 (MO); 2 mi. N of Colonia Toquian, on S slope of Volcán Tacanó, O m, Croat 47151 (MO); in 1864-1870, Ghies- breght 416 (K); 11.6 mi. N of Linda Vista Tropical Biol. Sta., Pueblo Nuevo Solistahuacán, Gittins 4220 (NY); Siltepec. Matuda 221 (MICH, MO); Bourgeau 362 (B). OAXACA: Dtto. Villa Alta, from Yetzelalag toward Lovani, 1,200 m, Hallberg 1498 (NY); Dtto. Tehuantepec, be- tween La Chiguzhe and Guevea de Humboldt, along Con- tinental Divide, 1,100-1,500 m, Hallberg 1744 (NY, UC); Mpio. Sta. Maria Chimalapa, ca. 7 km S de Sta. Maria, por la vereda a la cabecera del Rio Escolapa, 400 m, 16%51'N, 94?*41'W, Hernández С. 2457 (NY); ca 15 km al ESE de Sta. María, filo entre Arroyo Majipana (al О) y Rio Blanco (al E), 810 m, 16%52'N, 94?34'30"W, Hernández G. 2523 (NY); Mpio. Sta. Maria Chimalapa, ca. 8 km al N de Sta. Maria, canada al O de la vereda x paso Jugcuypac del Rio Verde, 300 m, 16?57'N, 41'W, Hernández G. & González L. 1749 (МО, к 24 km S of Valle Nacional (km 85), road from Ixtlan to Tuxtepec, 900 m, Mickel 1434 (MICH, АЫ, Dtto. Ixtlán, 79 km W of Ixtlán de Juárez on Rte. 175 1 km S of Campamento Vista Hermosa, Va m, Mickel 5650 (NY, UC) Dtto. Tuxtepec, 4-9 km S of Valle Nacional on Rte. 175, 300-700 m, Mickel 5893 (UC); 200-700 m, Mickel 5918 (NY, UC); Dtto. Ixtlán, 29 500-700 m, Mickel 6360 (NY, UC); km S of Valle Nacional, 80 km N of бијен de Juarez, trail E of Rte. 175 at Campamento Vista Hermosa toward Ladu, 500 m, Mickel 6446 (NY); Dtto. Ixtlan, 7 km S of Vista Hermosa, 71 km N of Ixtlan de Juárez on Rte. 175, 1,800 m, Mickel 6715 (NY, UC); Dtto. Ixtlan, 2- 3 km S of Vista Hermosa, 75-76 km N of Ixtlàn de Juárez on Rte. 175, 1,400 m, Mickel & Pardue 6548 (AAU, NY); Оно. Ixtlán, 2-3 km S of Vista Hermosa, 75-76 km N of Ixtlan de Juárez on Rte. 175, 1,700 m, Mickel & € 6580 (UC); Ixtlan, 76 km N of Ixtlan de Juárez on Rte. 175, Campamento La Esperanza, Mick- el 7172 (2); Villa ris ca. 20 mi. NE of Villa Alta, valley of the Yelagago River, 1,200 m, 17?25'N, 96%05'W, Mickel 1087 (MICH). PUEBLA: Curva de los Millones, 10 km S of Misantla, 1,350 m, Bohs et al. 1701 (GH); 15 km 5 of Misantla, Paz de Enriquez cloud forest, Bohs et Map 8. The distribution of Asplenium riparium. al. 1832 (GH); vallée de Córdoba, Bourgeau 2014 (B, GH, K, L, MICH, MO, NY); 8 km S of Misantla, 750 m, Conant 700 (GH); in 1889-1891, Finck 35 (МУ); Mirador, Galeotti 6274 (BR, K), 6275 (ph BR), Linden 31 (MICH); Sta. Rita m Misantla, Hahn 362 (K); 12 km S of Misantla, 1,350 m, Kress et al 703 (GH); Barranca de Mirador, d ы man (B, К); s.n. Mirador, in 1839, Linden 68 (К); Orizaba, in 1835, Muller 1772 (NY); 7.2 km E of Tebanca, 7.2 km E o E side of cs Catemaco, 2.6 km W of Bastonal lumber camp, 910 Nee & Schatz 19943 (F); Barranca of ein near ee 1,100 m, Pringle 7889 (GH); Zac- iapan, Purpus 1983 (B, UC); Barranca de Tenampa, BRA 2846 (B, BM, F, GH, MO), Purpus 2946 (M), Purpus 15420 (F); environs of Xalapa, road from Xalapa Misantla, Paz de Enriquez, N of Naolinco, 1,600 m, Sperling 4945 (GH); below Sta. Rita, 12 km ntla, 1,350 m, ~ 4995 (GH); Mpio. Yecuatla, Loma Santa Rita m, Ventura A. 3267 (NY); Mpio. de Atzalan, Rane a E "Caballo, 1,000 m, Ventura A. 14368 (G); Mpio. Totutla, El Mirador, 1,000 m, Ventura A. 16002 (MO). GUATEMALA. ALTA VERAPAZ: trail between Sepacuite and о E. m, (BM, F, NY; Eo 1,650 m, d 71321 (F); road 1 Tactic a s the divide on road to Tamahú, 1,500-1 m, Standley 90658 (Е); between Tactic and the а " Татаћи, 1,500-1,600 m, Standley 90667 (F, UC); и. 1,200 m, von Türckheim 630 (B, GH, K, NY, UC). BAJA VERAPAZ: San Rafael Chilasco, Salama, Guzmán V. 553 (F) Los Andes, Finca Panama, Brenckle 47-201 (UC). IZABAL: Cerro San Gil, 300-900 m, Steyermark 41908 (F). QUEZALTENANGO: Volcán Zunil, 1,500 m, Skutch 976 (GH); Finca Pirineos, below Santa Maria de Jesús, 1,350- 1,380 m, Standley 68216 (F); road between Finca Pi- rineos and Patzulin, 1,200-1,400 m, Standley 86710 (F); between Finca Pirineos and Patzulin, 1,200- 1,400 m, Standley 86744 (F, UC). SAN MARCOS: Barranco Emi- a and Buena Vista, 2,500-2,700 m, Standley 86332 (F). HONDURAS. COMAYAGUA: ca. 8 km SW of em road to Jesus de Otoro, 1,200 m, Burch 4 (NY) near El Rincón, Clewell Mus (NY); Cerro i Meambar, OLANCHO: 8.5 km W d del Real, 900 m, Ortega U. 295 (MO). PROV. UNKNOWN: Mr. Elliot's plantation, Wilson 37 (NY). NICARAGUA. ZELAYA: Costado sur del Cerro La Pimienta, y N del Cerro Annals of the Missouri Botanical Garden El Hormiguero, a orilla del Cano El Hormiguero, 800- 900 m, 13?44'45"N, 84%59'45"W, Grijalva a (UC); Costado SW de Cerro El Hormiguero, 900-1,000 m, 13?44'10"N, 84559'50"W, Grijalva 471 (UC); Cerro Saslaya, 1,100 m, Меш 3820 (MO); Cano El Hormiguero, 750-800 m, 13?46'N, 84°59'W, Pipoly 5929 (МО); trail toward Cerro El Inocente to Cerro Saslaya, 1,050- 1,150 m, 13%46'N, 85°01'W, Stevens 6677 (MO). Costa RiCA. ALAJUELA: near е along road to Colonia Virgen del Socorro, 800 m r et al. 11916 (F); forest along Rio Sarapiquí а рып crossing of road ; Reserva Forestal San Ramón, Hernández 8608 (CR); Receive Forestal de San Ramon, Rio San Lorenzito, agua arriba a partir de la estacion, 800-1,000 m, 10?12'53"N, 84?36'28"W, Herrera Ch. et al. 339 (MO, UC), Moran 4119 (CR, F, MO, UC), Moran 4136 (CR, MO, UC); ca. 23 km NE of San Ramón, along Río La Balsa, 850 m, Taylor 17819 (NY). CARTAGO: hills at N side of Rio Navarro, between Rio Sombrero and UC) E of dep road to s between Sabillas and Chitaria, ca. E of Finca Mata de Cana, 900 m, Lellinger & White 1431 (F, МО); 3 km SE of Cachi, 1,420 m, Lent 2 , F); 22 km E of Turrialba, leh ridge above ТА 1.200-1,450 m, Micke ; ista, 20 km NE of | Ко Chitaria, 750 m, Saiki CR. 59 (F); El Muñeco, x 2678 (UC). a dpi) а ~ ácimo, N slopes of sean 1,000 m, Rossbach 3860 (GH); Siquirres, 60 m, Kuppe 545 (M); Suerre, uis de Santa Clara, 300. m, J. D. Smith 6885 (B, BM, F, G, GH, M, NY); Talamanca, forets de Tsaki, Жен m, Топаиг 9473 (CR); 1901-1905, Wercklé 1713 (NY, S). PUNTARENAS: Cordillera de Tala- manca, area around Rio Canasta, 9.5 km NW of Agu Caliente, between Cerro Frantzius and Cerro Pittier, 1,500-1,600 m, 9?902'N, 82°59'W, Davidse et al. 28440 (AAU, CR, MO, UC), 28518 (UC); 5 km S of San Vito de Java, vic. of biological field station at Finca Wilson, 1,100-1,200 m, Mickel 3143 (NY); Monteverde, 1,100 w (M vic. ok El General 975 m, Shite h 2715 (GH, MICH, О, NY, 5); 35 mi. 5 of Cartago, 5 mi. 5 of Santa Maria de Dota, 1,300 m, Stork 1767 (GH, MICH, UC). Pan- AMA. CHIRIQUI: near Fortuna Dam, 1,200 m, Hampshire & Whitefoord 212 (BM); ca. 3.5 km NW of Bajo Mono, along Río Caldera (Boquete eu and on slope to the east, 1,600 m, 8%50'N, 82° , Зтић et al. 2455 (UC); no locality given, Wagner 809 (M); Dtto. Boquete, Fortuna Dam site, Continental Divide, 1,100 m, van der ы rf & - van Hardeveld 6766 (MO, UC). DARIEN: Ser- nia del Darién, Cerro Tacarcuna, 1,550-1,650 m, Gen- i» 14069 (MO). PANAMA: Above Sta. Fé, beyond Escuela Agricola, 1.8 mi. beyond fork in тар on Pacific slope, Croat 34167 (MO). VERAGUAs: Valley io Dos Bocas, road between Alto Piedra (above Sta. Fé) and Calovebora, 350-400 m, Croat 27417 (MO). VENEZUELA. BA ARINAS: Dtto. Bolivar, entre Sto. Darin y Altamira, en el s denominado Que bradón, en el camino que lleva al aa (B). PORTUGUESA: Dtto. Sucre, Los Paramitos, 20 km por d 1,000-1,500 m, 9?20'N, 1824 (MO, NY, UC); Dtto. Paraiso de Chabasquén, along road to Córdoba, ca. 27 min. from Chabasquen, jede below summit and below road, 1,500 m, 9°23'N, 69*54'W, А. R. Smith et al. 1037 (MO, UC, Z). TRUJILLO: bu Boran. 33.5 km SE of Восопо, road to Guaramacál, 1,300 m, 9°12'N, 70°06'W, A. R. Smith et al. 1545 (MO, UC). COLOMBIA. CUNDINAMARCA: Salto de Tequendamama, 2,300 m, Cua- trecasas 194 (F). HUILA: Guadalupe, carretera via a Flo- cia, a 2 km, 1,100 m, Osorio 112 (COL). MAGDALENA: Santa Marta, above Onaca, 1,100 m, H. H. Smith 968 F, GH, L, MICH, MO, S, UC). макіхо; Mpio. Ricuarte, along Rio Imbi, 2-3 km above Ecopetrol Cam- pamento Palmar, 3 km W of Ricuarte, 1,150 m, Croat 7155 1(МО); Ricuarte, entre Arrayán y Las Vegas, 1,400 m, Mora 4087 (COL). SANTANDER: Kalbreyer 857a (B). TOLIMA: Casas de las Varones, Stubel 16a (B). EcuADOR. NAPO: Cascada San Rafael, km 101 via Lago Agrio, falls on Río Quijos, 1,275-1,375 m, Foster 85-150 (UC); Archidona, Carretera Hollin- Loreto, Rio Huataraco, 800— 1,000 m, Cerón & Factos 7450 (MO, QCNE). PASTAZA: Colonia Játiva ca. 7 km N of Mera, 1,200-1,300 m, Harling & Andersson 16908 (AAU, F, МО); 2 km al NE de Mera, Hacienda San Antonio del Barón von Hum- boldt, 1,300 m, Baker et al. 5797 (MO Asplenium riparium grows on boulders along streams in wet forests from 200 to 1,900(-2,300) m. It ranges from Mexico to western Venezuela, Colombia, and Ecuador (Map 8). e species is characterized by short-creeping rhizomes, petioles 43-14 the length of the lamina, and abaxially rounded (not carinate) rachises (Fig. Since Liebmann's publication of Asplenium ri- parium, almost all pteridologists have considered it synonymous with 4. repandulum. The two spe- cies clearly differ, however, by the rachidial char- acteristics used to separate them in the key. In addition, they differ in habitat: 4. riparium grows on boulders along streams, while А. repandulum grows on tree trunks. The two species overlap in range only in eastern Ecuador (Maps 7, 8). Asplenium riparium is most closely related to A. volubile because both have strongly carinate Volume 80, Number 1 Murakami & Moran 31 1993 Asplenium sect. Hymenasplenium Monograph = Li 5 & К c c 5 5 = Ф Ф TP o О = = Ф Ф ~ zd ~ 5 5 © 5 - Ф D @ & 2B ao ao o 100 0.1 T T PUT TTTTTT T ^. 100 Internode length / Petiole diameter FIGURE 14. Quantitative differences between А. ri- FIGURE 15. parium (open squares) and 4. volubile (X). The mea- surements were taken from herbarium specimens through- out the entire range of both species rachises. The two species are easily distinguished by the features given in the n Petiole denen relative to the lamina and inter istance clearl distinguish the two species (Fig. a Figure 15 shows that the two species maintain their distinctive characteristics even when growing in the same population. e most remarkable thing about 4. riparium is that it produces plantlets from the soral recep- tacles on senescent leaves that have drooped and touched the soil (Fig. 13C, D). Iwatsuki € Kato (1985) and Kato & Iwatsuki (1985) that several Old World species of section Нутеп- mentioned asplenium produce plantlets on the pinnae, but they did not specify whether the plantlets prolif- erated from the soral receptacles. е Asplenium triquetrum Murakami & R. С. Moran, sp. nov. TYPE: Bolivia. La Paz: Prov. Nor Yungas, Polo-Polo bei Coroico, 1,100 m, Buchtien 625 (holotype, MO; isotypes, BM, K, NY, Z). Figure 8B, C; Ma Plantae terrestres; rhizoma 2-6 mm latum, fere nu- dum, breviter reptans, internodis 0.4-2 cm longis; squa mae 1-2.3 x 0.5- 1. m, ovatae, ы laminae d 40 cm longae, 1-pinnatae, 12-17 jugatae, lanceolatae el oblongae, apice subconforme; rhachis triquetra, alis viridibus, ad laminam parallelis; pinnae 5-9 x 1.1-1.8 cm; venae l-3 furcatae; sori 5-9 mm longi; sporae spi- nosae Plants terrestrial; roots 0.4-1.2 mm wide; rhi- zome 2-6 mm wide, nearly naked; scales 1-2.3 x 0.5-1.2 mm, dark brown, ovate, clathrate; pet- Variation in quantitative characteristics of Asplenium riparium (open squares) and А. volubile (black squares) from the Fortuna Dam area, Panama. The measurements were taken from living plants. iole base 2-4 mm wide, swollen, 0.4—2 cm distant from each other on the same row; petiole 10-25 0.15-0.25 cm, 43-14 the leaf length, dark green- brown, fleshy and dull, glabrous, shallowly grooved; lamina 22-40 cm long, dark green, lanceolate to oblong with subconform apex; rachis | -pinnate, with abaxial green wings in the same plane as the lamina, strongly carinate abaxially; lateral pinnae 5-9 x 1.1-1.8 cm in the middle of the leaf, 12— 17 pairs, obliquely oblong to oblong-lanceolate, the margin occasionally lobed, almost entire to only slightly toothed, the basiscopic margin excavate for /.—V5 the pinna length, acute at apex, broadly cu- neate and almost sessile at base, auricle 2-6 mm long; veins 1—3 times forked; sori 5-9 mm long; spores 40-52 um, 64 per sporangium, spiny. МАР 9. The distribution of Asplenium triquetrum. Annals of the Missouri Botanical Garden Paratypes. BOLIVIA. LA PAZ: Prov. Lareca ја, Маргі, Н, O m, Zardini & Aguayo 9501 (MO). PARAGUARI: Parque Nacional ои vecino al Salto Guar rani, desk 268 (MO). BRAZIL. PROV. UNKNOWN 2 X Glaziou 7954 (В); Glaz iou 12283 S). MINAS GERAIS: Caldas, Mosen 2111 (GH). PARANA: Serra do Mar, inter Ypiranga ab Volta Grande, 600 m, Dusén 3351 (MO, S); Ypiranga, Dusén 3551 (S); Serra o Mar, Desvio Ypiranga, 700 m, Dusén 6774 (С, 5); Guaratuba, Dusén 13730 (5); Serra do Mar, Desvio Ypi- ranga, 19 Sep. 1908, Dusén s.n. (BM, BR, F, G, GH, K, MO, NY, 5, 7); Mpio. Morretes, Rio Маз Сайга, 100 m, Hatschbach 23237 (UC); Mpio. Morretes, Estr. Gra- ciesa, Greta Funda, 500 m, Hatschbach 24738 (MO, 5, UC); Mpio. Morretes, Jurape, Hatschbach 41961 (UC, Z). RIO DE JANEIRO: Organ Mts., 37, Gardener 168 К); environs de Rio de Janeiro, Glaziou 12289 (К); Sebastianipol, ad rivulos Mandioccae, Martius s.n. (B, K, М); Serra d'Estrella, Aug. 1817, Martius s.n. (MJ Serra dos Orgàos, 1817, gie s.n. (M); near Rio de Janeiro and Bahia, Webb 66 (NY); Serra Estrella, Luetzelburg 21(M); Serra dos Orgáos, Morro Ass burg 6209 (M, S, UC), vnu PA (М); Serra Estrella bei Petropolis, wasserfall, O m, Luetzelburg 6301 (M, MICH, UC); Serra ndr Luetzelburg 6888 (F, K, M, MICH, UC) Serra Estrella bei do Los zelburg 7396 (М); Corcovado, waterfall, Luetzelburg 13007 (M); A de Janeiro, un E (В, К, S); Mandioca, Sello: O GRANDE DO SUL: 27 ~ А Melo, Jürgens 391 (S). SANTA CATARINA: Joinville, ME 186 (NY); Araranguá, Reitz 729 (K, MO); Florianopolis, Sertão da Lagoa, Rohr 1054 E L, NY); erar Schmalz 81 (5); Joinville, 800 Schmalz 186 (Е); Lages, Spannagel 143 (NY, S, 00 Curitiba, ord 1 (M). s40 PAULO: Iguape, 10 Sep. , Brade s.n. (GH); Morro das Pedras, Brade 5242 Б. ом да Canteneita, prope Sào Paulo, 1,100 m, Brade 6528 (GH); Cajuva, Brade 8240 (GH, NY); Alto da Serra, Edwall & Puttemans 5004 (Sy; Luederwaldt Niederlein 1922 (В). Asplenium triquetrum is the southernmost spe- cies of the section (Map 9). It grows from 400 to 1,100 m on wet rocks along deeply shaded stream- banks, around cave entrances, and near small wa- terfalls. Asplenium triquetrum is characterized by its epipetric habitat, strongly carinate rachis, and ori- entation of the rachidial wings in the same plane as the lamina (Fig. 8B, C). It is most closely related to A. repandulum (which see). 10. Asplenium volubile Murakami & К. С. Moran, sp. nov. TYPE: Ecuador. Cotopaxi: idi es road, km 46 from Que- vedo, 600 m, 0955'S, 79?11'W, Holm-Niel- sen a al. 2905 (holotype, a isotypes, AAU, JC). Figure 13E; Map Plantae epiphyticae; rhizoma 2.5-3.5 mm latum, fere nudum, longe reptans, internodiis 2-4 cm longis; squamae conforme; rhachis abaxialiter rotundata, alis viridibus ad laminas perpendicularibus; pinnae 4-5.5 x 0.9-1.7 cm 11- pares; venae 2-furcatae; sori 5- p mm longi; sporae spinosae. Plants epiphytic; roots 0.5-1.2 mm wide; rhi- zome 2.5-3.5 mm wide, nearly naked; scales 1- 1.3 x 0.2-0.4 mm, brown, ovate-lanceolate; pet- 4 cm dis- tant from each other on the same row; petiole 4.5- 11 x 0.1-0.25 ст, У-У the leaf length, dark green-brown, fleshy and dull, shallowly grooved; 1 -pinnate, oblong-lanceolate to ovate with a conform to sub- iole base 2.2-3.8 mm wide, swollen, 2- lamina 22-35 cm long, dark green, conform apex, thin-herbaceous; rachis like petiole, with green wings perpendicular to the plane of the lamina; lateral pinnae 4-5.5 x 0.9-1.7 cm in the middle of the leaf, 1 1—18 pairs, oblong-lanceolate, the margin only slightly toothed, excavate !6 ~ portion of the basal basiscopic side absent, acute- obtuse at apex, broadly cuneate and almost sessile at base, auricle 1.5-5 mm long; veins 2-forked; sori 5-10 mm long; spores 48-50 um, 64 per sporangium, spiny. Paratypes. Costa RICA. ALAJUELA: oe Bijagua, El Pilon, Rio Celeste, 700 m, 10%49'N, 84°27 a NY, UC). GUANACASTE: border with Alajuela, above Bi- jagua, slopes of Miravalles, 1, 500 m, Gómez et al. 19195 (MO, UC). LIMON: 5 km SW de Guápiles, bosque cerca o Toro Amarillo, 300-400 m, THEN M. 2816 along Сашка! го (MO, UC); 5.3 mi. N of bridge over Per" Dam, betw Fortuna and Chiriqui Grande, 1.2 mi. N of Con Divide, 910 m, 8?44'N, 82? TWO Cae & Grayum ee (MO). COLOMBIA. cHoco: San José del Palmar, del Rio Torito (afluente del Rio Habita), declive sa на Finca “Los Gua | 684 camino а Miralindo, Cordillera Haden, vertiente ori- ental, Hoya del Rio Cali, vertiente derecha, Quebrada- honda, 2,100-2,250 m, Cuatrecasas 16441 (COL, F, y El Silencio, Yanaconas, 1,900-2,200 m, Killip & García 33811 (COL, GH). ECUADOR. COTOPAXI: “Andes Quitensis,” Chimborazo, Spruce 5689 (G, K). Los RIOS: Volume 80, Number 1 1993 Murakami & Moran 33 Asplenium sect. Hymenasplenium Monograph Cerro Mombe, Hacienda Clementina on Rio Pita, 660 m, у 5, 77°45'W, Holm-Nielsen et al. 26835 (AAU). PICHINCHA: Toachi ad San Miguel, Sodiro 66 (UC). Asplenium volubile occurs in Costa Rica, Pan- ama, and coastal Colombia and Ecuador (Map 10). It grows from 300 to 2,200 m in wet forests as an epiphyte on small trees or saplings. Asplenium volubile has been previously iden- tified as A. repandulum. In addition to the rachis characteristics given in the key, it differs from А. (4.5-11 cm long), which is only У to У, the leaf length (petiole 10-25 cm long and about Уз the leaf length in 4. repandulum). In both species the rhizome is long- creeping and epiphytic, but field observations by both of us have shown that the rhizomes differ in repandulum by its shorter petiole habit. Asplenium volubile has a twining rhizome that grows spirally around the trunks of young trees or saplings that are usually less than 5 cm wide. This characteristic can usually be ascertained in herbarium specimens because the rhizomes ap- pear sinuous or contorted. In contrast, 4. repan- dulum grows on large trees and the rhizome is more or less straight, not spiraling upward around the trunk. Asplenium volubile has a longer-creep- ing rhizome than 4. repandulum. The distance between petiole bases on the same side of the rhi- zome (orthostichy) is 2.2-3.8 cm in 4. volubile, and 0.4-2 cm in 4. repandulum. In addition to the morphological characteristics, the two species differ completely in range: А. volubile is restricted to the western side of the Andes (Map 10) and 4. repandulum is restricted to the eastern side (Map See A. riparium for comparison to that species. HYBRIDS 11. Asplenium Х раругасешт (Jermy $ T. Walker) Murakami & R. С. Moran, comb. nov. Diplazium X papyraceum Jermy & T. Walker, Bull. Brit. Mus. (Nat. Hist.), Bot. 13: 264, fig. 10. 1985. TYPE: Trinidad. Charuma Ward: Central Range Forest Reserve, Brick- fields Teak Plantation, 3 mi. S of Forest Rest- house, 60 m, in wet ground by stream, 9 July 1963, Walker T6178 (holotype, BM not seen). Plants terrestrial; rhizome 2-3 mm wide, sparse- ly scaly; scales 1-1.5 mm long, narrowly lanceo- late, blackish, entire; petiole base 2-3 mm wide, ca. 1 mm distant on the same row; petiole 15-22 cm long, brown, slightly shorter than the lamina, gray-green with narrow black wings, glabrous; lam- МАР 10. The distribution of Asplenium volubile. ina 25-30 х 12-17 cm, l-pinnate, ovate-lanceo- late, the apex abruptly to gradually reduced, pin- natifid; rachis brown, sparsely scaly with minute ~ 0.5-1 mm long) filiform, unicostate scales and slightly longer, narrowly triangular scales; pinnae -2.2 cm, short-petiolulate, 9—12 pairs, ~ the basal TA, men margin slightly excavate, the acroscopic margin coarsely serrate, the serrations 1-2 mm deep; veins 2 or З times forked; sori 3- 11 mm long; spores abortive. cimen examined. TRINIDAD. CHARUMA WARD: Central Range Forest Reserve, Brickfields Teak Planta- tion, 3 mi. S of Forest Resthouse, 60 m, in wet ground by stream, 9 July 1963, Jermy 2178 (BM). Asplenium X papyraceum is known only from Trinidad. Jermy 4 Walker (1985) have an excellent photograph of the paratype (Jermy 2178). The plants resemble the above-cited locality in A. delitescens but differ by their shorter, relatively wider, and more numerous pinnae. Asplenium X papyraceum is believed to be a hybrid based on its irregular meiosis, two size class- es of chromosomes, triploid chromosome number, and abortive spores (Jermy & Walker, 1985). The asplenioid characteristics of 4. X papyraceum are clathrate scales and ап X-shaped petiolar strand that unites distally in the petiole. It belongs to section H ymenasplenium as evidenced by its short- creeping rhizome with a typical broad ventral meri- stele and by its rachis-costa structure. 34 Annals of the Missouri Botanical Garden Jermy & Walker (1985) discussed the evidence relating the hybrid to Asplenium, but concluded that it belonged in Diplazium based on the chro- mosome count of n — 123, a multiple of 41, which is the base number for Diplazium. The count of n = 123, however, does not exclude Asplenium. Although a base number of 41 is not known in Asplenium, it is not far off from the numbers known in section Hymenasplenium, which have n — 38 or 39 (Mitui et al., 1989). The most likely 123, given the morphological evidence that clearly indicates the plant is an As- explanation of n — plenium, is that the parents had the unusually high chromosome number of n = 41. The resemblance to A. delitescens suggests that that species is one of its parents; however, А. delitescens is not known from Trinidad (Map 2). 12. Asplenium delitescens (Maxon) L. D. Go- mez X A. laetum Sw. Figure 6B. The specimens listed below are presumably hy- brids based on their intermediate morphology be- tween А. delitescens and А. laetum Specimens examined. ELIZE. STANN CREEK: Stann Creek Railway, 50 m, Schipp 49 (BM, MICH, MO, UC). Costa RICA. ALAJUELA: Rio Chiquito, ca. 40 km on road to Upala, 0 m, Gómez 18620 (MO, UC). SAN JOSE: Zona Proteotora El Rodeo, along Fila Diamante, S from Alto Gracias a Dios, 4 km (by air) SW of Ciudad Colón, 900 m, Grayum & nnd 966 1 (MO, UC). VENEZUELA. ARAGUA: near Guamitas, Parque Nacional, 760 m, Alston 5828 (MO). FALCON: Parque Nacional Quebrada de la ,N , 600 m, Zarucchi & Cuadros 4040 A: 7 km E of Codazzi, 300 m, Haught 3761 (COL, GH). META: La Macarena, Río Güejar ca. 10 km below junction with Rio Zanza, 470 m, S. Smith & Idrobo 1523 (COL, GH, UC). ECUADOR. MORONA-SANTIAGO: Cordillera de Cutucü, road Méndez- m, van der Werff & Palacios 10403 (F, al. 8502 (NY, UC). PERU. HUANUCO: ca. pueblo de Puerto Inca, a unos 85 km de la confluencia con el Río Ucayali, 400-500 m, Schunke V. 2967 (NY). LORETO: above Pongo de Manseriche, 250 m, Mexia 6223 (BM). MADRE DE DIOS. Manu, Parque d gm каце Pakitsa Station, Tachigali Trail to 8 km mp, 350 m, 11°56’S, 71?16'W, Foster & Baldeón ree (F). BOLIVIA. BENI: Prov. Ballivian, Rio Colorado, Collegio Técnico Agrope- cuario de Rio Colorado, 235 m, 15°00’S, 67°10'W, Fay 207 ) LA PAZ. Prov. Larecaja, Isupuri, 500 m, ‘illiams 1091 (GH, NY); Prov. Nor Yungas, puente obre el Rio Beni, 550 m, Beck 13367 (F). ~ The above specimens would key to 4. laetum because of their number of pinna pairs (10-21) and atropurpureous petioles and rachises. These intermediates differ from A. laetum by their larger laminae and often abruptly contracted apices. The above specimens differ from 4. delitescens in the shape of the lamina apices (especially the shape of the pinnae below the apical segment), greater num- ber of pinna pairs, and dark petioles (Fig. 6B). The shape of the lamina apex is perhaps the best char- acteristic to detect the intermediates, but it is dif- ficult to characterize the plants because of their inherent variability in features that distinguish the parents. Most of the specimens have aborted spores, but some appear to have normal spores. The specimens treated here need to be investigated cytologically and enzymatically to confirm their hybridity and determine their parentage. А few may eventually be shown to have 4. ortegae or 4. purpurascens as the other parent. 13. Asplenium Xincisoserratum (Rosen- stock) Murakami & R. С Asplenium laetum Sw. var. incisoserratum Rosenstock, Repert. Spec. Nov. Regni Veg. 22: 8. 1925. TYPE: Costa Rica. San Jose: Grenadilla, 1,200 m, 26 Dec. 1910, Brade 425 (holotype, S; isotypes, B, M, UC). Figure 16. . Moran, stat. nov. Plants epilithic; roots 0.1-0.5 mm wide; rhizome -4 mm wide, scaly at the apex; scales 2-2. 0.25-0.35 mm, blackish, narrowly triangular to linear; petiole base 0.5-2 1.5 mm distant from each other on the same row; petiole 3-8 x 0.03-0.1 cm, ca. 16 -V4 the leaf length, mostly stramineous to greenish, sometimes mm wide, swollen, 1— brown basally (especially in large leaves), glabrous; lamina 8-20 cm long, green, 1 -pinnate, lanceolate, gradually to abruptly tapered to a pinnatifid apex; rachis greenish; lateral pinnae 1.4 x 0. 6-14 pairs, lanceolate, acute to obtuse, coarsely dentate, the teeth 1-2 mm deep, the basal basi- scopic side excavate for 15-14 the length of the pinna; veins l-forked; sori 3-5 mm long; spores irregular, aborted. 5-1 cm, Additional specimens examined. MEXICO. OAXACA: Tuxtepec, above Jalapa de Díaz, 330 m, Hallberg 1467 (NY). Costa Rica. CARTAGO: Rio Reventazón, below Tur- rialba, 540-600 m, Skutch 4681 (CR, F, GH, MO). san JOSE: Rio Surubres, 300 m, Brade 430 (B, GH, M, S, UC); vic. of El General, 1,100 m, Skutch 2531 (MICH, MO, S); basin of El General, 675-900 m, Skutch 4790 (CR, MICH, MO). Ткімірар. No locality, Fendler 61 pro иа СН, MO). COLOMBIA. VALLE: valley of Río Digua, O m, Alston 7837 (MO). Asplenium X incisoserratum grows on wet cliffs and boulders in streams from 300 to 1,200 m. Murakami & Moran Asplenium sect. Hymenasplenium Volume 80, Number 1 1993 Monograph wi NS SN © NUN NY Qw DN ` V) IP g y AVS: X NX K ix >z ZZ < RS RN Y==S" <= f D 227 E FR TIS AE) 199) Asplenium X incisoserratum s dde 430, СН). — А. Large fertile leaf. — B. Small fertile leaf. — aised, ungrooved rachis and costa, and thickened decurrent d FIGURE 16. A, B. cissum, juncture of rachis a pinna, note the со кше of гасыз and pinnae, note the grooved rachis an 399, MO). . A. absc pinna margins (Skutch 2339, MO).— the narrow wing that borders the = and the rachidial groove (Martínez S. 16 36 Annals of the Missouri Botanical Garden Based on morphology (Fig. 16), it appears to be a hybrid between 4. hoffmannii and А. laetum. From A. hoffmannii, it differs by having more pinna pairs, more than two or three leaves per rhizome, and a basally darkened petiole (especially in large leaves). From 4. laetum, it differs by its usually greenish petiole, greenish rachis, fewer number of pinna pairs, and more deeply and thoroughly in- cised-serrate pinnae. All of the specimens cited above have aborted spores. Two miscellaneous notes: The specimen cited from Colombia appears to be this hybrid, but 4. hoffmannii is not known from western Colombia where the specimen was collected. Apparently, Ro- senstock (1925) did not publish a new name when referring to Brade 430 as representing a "form minor, die vorliegenden Exemplare fruktifizieren schon bei 6 cm Höhe.” This diagnosis is in German, contrary to Rosenstock's varieties in the same pa- per, which are in Latin. Also, the typography of “forma minor” is in roman, not boldface or italic as he usually used when publishing new names. LITERATURE CITED Hayata, B. 1927. On the systematic importance of the stelar system in the Filicales, 1. Bot. Mag. (Tokyo) 41: 697-718 IwaATSUKI, К. 1975. Taxonomic Hus of P DE . X. Acta Phytotx, s M. Kar : a structure of Ás- plenium cae E and. its allied species. Kalikasan 4: 165-174 1985. Diversity of vegetative re- production in the ји with reference to leaf-borne proliferation. Pp. 124-131 in H. Hara (editor), Or- igin and Evolution ot Diversity in Plants and Plant ommunities. Academia Scientifica, Tokyo JERMY, А. C. & T. G. WALKER. 1985. Cytotaxonomic studies of the Ser A es Bull. Brit. Mus. (Nat Hist.), Bot. 13: 276. М gaea кі. 1985. An unusual sub merged aquatic oo a Asplenium unilaterale. Amer. Fern. J. 75 & poe in ecology, mor- phology and reproduction of Asplenium sect. Hy епа ита а ан in Seram, Indonesia. J. Fac. Sci. Univ. Tokyo, Sect. 3, Bot. 14: 37-48. NAKATO, 5. AKIYAMA & K. IWATSUKI. 199 0. The systematic position of Asplenium car- diophyllum (Aspleniaceae). Bot. Mag. (Tokyo) 103: 61-468 MisuTA, S., M. Kato & K. IwarsuKi. 1980. Stelar structure of ш Bot. Mag. (Tokyo) 93: -289. Мато, K., N. MURAKAMI & K. DwarsuKi. 1989. Chro- mosomes and systematics of Asplenium sect. Hy- menasplenium (Aspleniaceae). Amer. J. Bot. 76: 1689-1697. MonTON, C. V. & D. B. LELLINGER. 1966. The Poly- podiaceae subfamily поета in Venezuela. Mem. ; 1-49, New York Bot. Gard. — MuRAKAMI, М. & S.I. НАТАМАКА. 1983. D-2-amino- pimelic acid and trans-3,4-dehydro-D-2-aminopi- melic acid from Asplenium unilaterale. Phytochem- istry 22: 2735-2737. & ————. 1985. Chemotaxonomic studies non-protein amino acids in Aspleniaceae (Pter- 48 in H. Нага (editor), Origin Co munities. Academia Scientifica, Tokyo 988a. Chemota xonomic pepa (Asplen Asplenium sect. Hyme Pe iari aceae). or Mag. Sree 101: -372 88b. A ann taxonbiny of the И None complex in Japan i c. Sci Taiwan. . Univ. Tokyo, Sect. 3, Bot. 83- WATSUKI. 83. Observation on the variation of Asplenium unilaterale in Japan with cial reference to apogamy. J. Jap. Bot. 58: 257- 262. ——, J. Furukawa, S. Oxupa, & 5.-1. HATANAKA. 1985. Stereochemistry of 2- -aminopimelic acid and related o acids in three species of Asplenium. 4 Ferns of шшш British Mu- iy (Natural History), Londor Roane F. J. 19 Numerical кш and multi- variate analysis system. Version 1.40. Exeter Pub- ishing, New Yor Rosenstock, E. 1925. Filices novae a cll. Alfred et rt Brade in Costarica collectae. Repert. Spec. Nov egni V -23. SMITH, А. К. 1976. Diplazium Pic and the neotropical species of Asplenium sect. Hymenas- enium. Amer. Fern. J. 66: 1981. Part 2, BI EE Pp. 1-370 in D. E. Breedlove (editor), Flora of Chiapas. California Academy of Sciences, San Francisco. ICKEL. 1 Chromosome counts for Mexican ferns. Brittonia 29: 391-3 Tryon, A. К. € B. LUGARDON. e Spores of the о Springer- Verlag, N w York. Ткүом, R. M. 1972. Endemic are 22d geographic speciation in tropical American ferns. Biotropica 4: 12 T Y. e ña x ПЛО 1988. Genetic variation in an apog ern: Japanese apogamous form of Asplenium. indo Bot. Mag (Tokyo) 101: 213- INDEX TO COLLECTORS? NAMES AND NUMBERS This monograph i is dur on the specimens cited below, which consist of abou boldfac of the species and hybrids in this monograph, given below for easy reference SPECIES 1. А. delitescens (Maxon) L. D. Gómez 2. A. hoffmannii Hieron. 3. A. laetum Sw. 4. A. obtusifolium L 5. A. ortegae Murakami & R. C. Moran 6. A. purpurascens Mett. ex Kuhn Volume 80, Number 1 1993 Murakami & Moran 37 Asplenium sect. Hymenasplenium Monograph 7. А. repandulum Kunze 8. A. riparium Liebm. 9. A. triquetrum Murakami & R. C. Moran 10. A. volubile Murakami & R. C. Moran HYBRIDS 11. A. косе (Jermy & T. Walker) Murakami & R. C. Mor 12. A. ни X А. laetum 13. A. xincisoserratum (Rosenstock) Murakami & R. C. ran Abbott 2627 (3). Acosta S. 13783 (3); 13815 (6). Aguayo 268 (9). Allen 347 (1). Alston 5828 (12); 7716 (3); 7837 (13); 8707 (3); 8868 (1); 8925 (1). Antonio 3530 (4). Argüello 498 (6). Asplund 5550 (10). Avilés 16 (1); 66 (3). Ayala 58 (6). Aymard C. 5274 (3); 5297 3). aker & Zaruma 6476 (3). Baker et al. 5797 (8). Barber 6 (3); 2500 (3). Barrington 404 (8). Bartlett & Lasser 16580 (1); 16843 (2). Beck 13367 (12). iios 1 (4). Béliz 150 (3). Bell 200 (6); 238 (6). pr 20 (3). Bernoulli те (3); 834 (2). Bernoulli & Cario (2). Bohs et al. 1701 (8); 1832 (8). Bourgeau 362 j^ 2014 (8); 2151 (3). Box 456 (4); 464 (3); 473 (4). Brade 425 (13); 430 (13); 765 (8); 5242 (9); 6528 (9); 8240 (9). Brandbyge & Asanza C. 31225 (5); 31649 (5). Breed- love 26082 (8); 29819 (8); 29935 (1); 32617 (8); 33123 (1); 33850 (1); 33853 (1); 33893 (2); 34197 (3); 35410 (8); 38081 (8); 39059 (3); 42851 (2). Breedlove & McClintock 34140 (1). Breedlove & Smith 21683 (3); 31319 (3). Brenckle 47-201 (8). Brenes 21520 (3). Brit- ton & Wilson 5300 (3). Britton et al. 909 (3); 1656 (4). Broadway 3026 (3); 4212 (3); 5353 (4); 7180 (3). Brown 338 (3). Buchtien 625 (9); 3334 (3); 3348 (9). Buch- wald 43 (6). Bunting et al. 12036 (3). Burandt 715 (3 Burch 6114 (8). Burch et al. 1095 (1). Burger et al. 1916 (8). Cabrera et al. 1938 (3). Callejas E al. 2841 (3). Camp 3785 (6). Cazalet & у а oni ~ Chater et al. 128 (1); 131 3). Corda 5134 (3). Chur- chill 3851 (2). Churchill & de Nevers 4242 (3). Clement 7 (3). Clewell 3097 (8). Conant 700 (8). Conrad et al. 2889 (3). Contreras 3432 (1); 10500 (1). Cooley 8393 (3). Cornman 512 (1); (1). Correa A. et al. 2973 (1 & Марапа 3139 (3). Crawford 554 (1). 8780 (3); 8900 (3); 9314 (3). Croat 5192 (1); 6: 8053 (1); 8475 (3); 8515 (1); 8597 (1) 8643 (1); 9425 (1); © 12662 (1); 14056 (1 ~ Cowan roat 10); piae (4); (10). Croziat 794 (1). Cruger 44 (3). Cuadros V. et al. 3895 (3). Cuatrecasas 194 (8); 8607 (10) 18441 (10). Davidse & González 19242 (1). Davidse et al. 18205 (5); 18826 (5); 28440 (8) 28518 (8). Deam 470 (1). 3508 (8). d ille et al. 8947 (3). Dodson ntry 10051 A 12275 (6). Dodson et al. 7089 (1). orantes et al. 2555 (3). Dusen 3351 (9); 3551 (9); 774 (9); 13730 e Duss 1635 (3); 1636b (4); 4199 (3); 4205 (4). Dwyer 6823 (2) dwall & Puttemans 5004 (9). Eggers 608 (4); 15118 (3); 15175 (3); 15214 (6). Ekman 2533 (1); 2760 (1); 4430 (3); 6106 (3); 7332 (3); 15485 (3); 15836 (3). Evoy 61 Fagerlind & Wibom 1593 (6). Fay 665 (3); 673 (4); 2078 (12). Fendler 61 (2 & 13); 131 (4); 136 (3); 139 (4); 143 (0, Finck 159 (1); 35 (8). Folsom 3062 (4); 735 (8). Forero & Jaramillo 2134 (3). Forero et al. 6840 (10); 7237 (10). Foster 85-150 (8). Foster & Baldeon 12806 (12). Foster et al. 3449 (7 Galeotti 6274 (8); 6275 (8). Gardener 168 (9). Gentle 2714 (1); 5139 (1); 6261 (1); 7378 (3). Gentry 8679 (1) 14069 (8). Gentry & Clewell 7064 (3); 7069 (3). Gentry et al. 23073 (7). Ghiesbreght 416 (8); 426 (3). Gilbert 54 (3). Gittins 4220 (8). Glaziou 5318 (3); 7247 (9); 7954 (9); 12283 (9); 12289 (9). Gómez 2168 (8); 18620 (12). Gómez et al. 19195 (10); 23094 (3). Grayum 8506 (8). Grayum & Hammel 5530 (8); 8466 (8). Gra- yum & Schatz 3140 (8). Grayum & Zamora 9661 (12). Grayum et al. 4419 (3); 4906 (3); 5060 (8); 5808 (8); 8943 (3); 9152 ү 9159 (2). Grijalva 299 (8); 471 (8). Guzmán V. 553 Hahn 68 (4); s (8). Hallberg se Hes pad ); 1471 n 1498 (8); 1744 (8). Hamilton 529 (1). Hammel 14338 (3); 15271 (3); & ror 16908 (8); 22310 (6); 24978 (6). 7374 (3). Harrison SVT174F (4). Hart 42 (1); 205 yes 57 (1); € 1008 (3); 1574 (1); 1920 (1 ); 2021 (2); 2457 (8); 2523 (8). Hernandez G. & Gonzalez 1787 (3); 1787 (1); (8). Неттега Vo. 3311 (8). Herrera Ch. et al. 339 (8). Hodge 4). Hoffmann 759 y 836 (2 . Holdridge 50 (4); 1617 6). Holm-Nielsen et al. 2720 (6); 2905 (10); 7066 (6); 26835 (10). irgend 68 (4); 257 (2); 390 (4). Hor- 8). Hunnewell 16375 (1). Husnot 334 093 (3). Jamat 24 (3). Jermy 2178 A AS 11370 (3 | Jiménez M. 2816 (10). Johnston 74 (3); ; 258 (2); 385 (2); 444 (2). Jürgens 27 re 372 (9); d 9). Kalbreyer 317 (3); 857a (8); 951b (4); 1499 (3); 1916 (3). Kennedy & Andrews 1876 (1). Kenoyer 29 (1). Killip 2512 (1) 2567 (2); 2708 (1); 2710 (3); 2806 (1); 2810 (3); 2829 (3); 34888 (3). Kilip & García 33811 (10). Killip & Smith 14658 (2); 24547 (7); 24804 (7); 24924 (5); 26633 (5); 29078 (5). Kirkbride 2188 (4); 2368 (3). Klug 4253 (3). Knapp 2686 (2); 2716 (1). Knapp & cy 2334 (1). Kress et al. 703 (8). Kupper 528 (8) 5 Vs N = -— ao m= = = T 5 (8). и 18 (3); 79 (4); 169 (4). Lehmann 8830 (5). Lellinger 517 (3); 571 (4). Lellinger & White 1431 (8). Lellinger & de la Sota 532 (1); 550 (3). Lent 330 (8); 1411 (7); 2063 (8); 4061 (3). Leon ard & Leonard 14363 (3). Liebmann 310 i zalez 9473 (1). Liesner et al. 68 (8). Lindig 236 (3). Little & Little 8427 (5). Lloyd 698 (4); 699 (3); 871 (4). Lojtnant 15757 (3). Losch 770 (3). Luederwaldt 21480 (9). Luetzelburg 21 (9); 6209 (9); 6296a (9); 6301 (9); 6888 (9); 7150b (9); 7396 (9); 13007 (9). Luteyn et al. 8502 (12) 5). Madrinán & Barbosa 258 (4). Martinez 9 (1). Martinez S. 8008 (1); 8136 (1); 8402 (1); 16399 (1). Martinez S. et al. 22711 (3). Martius 38 Annals of the Missouri Botanical Garden 342 (9). Matuda 221 (8). Maxon 623 (3); 641 (3); 2545 (3); 4158 (3); 4677 (1); 4868 (3); 5055 (3); 5733 (3); 7 (3). Maxon & Harvey 6674 (1). Maxon & Hay 3263 (8). Maxon & Killip 244 (3); 772 (3). Maxon et al. 6807 (1). McDonagh et al. 446 (3). McPherson 8666 (4). Mendonca 1208 (3). Mexia 6223 (12). Mickel 1087 (8); 1304 (2); 1434 (8); 1862 (3); 2061 (3); 2317 (3); 3143 (8); 3359 (3); 3401 (8); 5650 (8); 5784 (1); 5881 (3); 5893 (8); 5918 (8); 6077 (2); 6360 (8); 6446 (8); 6461 (3); 6715 (8); 7172 (8); 7188 (3); 7212 (1); T i e (4); 9516 (3). Mickel & Leonard 5138 (2 7 (2). Mickel & Pardue 6548 (8); 6580 (8); 6843 n 6866 (1). Moore 7 (3). Mora 4087 (8). Morales R. 5 (1). Moran 3092 (3); 3121 (3); 3676 (1); 3683 (3 | 19 (8); 4136 (8); 5329 (6). Moreno 19063 (1). Mori & Kallunki 2670 (2); 2672 (1); 2826 (3); 3964 (4 Moritz 23 (3); 365 (3); 384 (4). Morton 6170 (4). Morton & Makrinius 238 (3). Mosen 2111 (9); 2665 (9); 3736 EA L| 186 (9); 1772 (8); 1835 (3). Murillo 363 (3); 7 (5). Murray 26 (3). bu 9 (3). Nee 7722 (3); 9026 (1); 9456 (3); 29974 (1). Nee & Schatz 19943 (8). Neill 3820 (8). Niederlein 1922 (9 Ocampo S. 1637 (3). Allgaard et al. 39205 (3). Orcutt 6835 (3). Ortega 1912 (5). Ortega & он 1817 (5); 1818 (3); 2087 (5); 2088 (5); 2089 (2 = T "n T П 5 (8). Ortega et al. 1824 (8); 1845 2 1886 (3); 3093 (6). Osorio 112 (8) Otto 176 (3); 609 (3). Palmer 117 e Pannier 132 (4). Pavón 186 (3). Pennell 4213 (1). Pérez 1 E ТЫ 4861 (3); 5929 (8). Plowman & Schunke V 3 (7). Polakowsky 151 (3). Pollard & Palmer 348 (1 | 714 (3). уе 7889 (8). Proctor 4069 (3); 17250 (3); 17254 (4); 17703 (4); 19252 (3); niens ); 22690 (3); 36071 (3). Кош et al. 27114 (1). Purpus 174 (3); 1983 (8); 2846 (8); 2946 (8); 6735 (3); 6765 (3 y 7241 (3); 15420 (8). Quijada 103 (3). Reitz 729 (9). Riba et al. 318 (8). Rivera 23 (1). Rohr 1054 (9). Ross 688 (3). Rossbach 3860 (8). Rovirosa 828 (3); 1038 (1); 3135 (1). Rowlee & Rowlee 415 (1). Saiki CR-59 (8); M234 (1); M230 (3). Sandwith 1790 (4); 1833 (3). Schipp 8-275 (1); 49 (12); S-277 (3). Schlim 397 (3); 653 (4). Schmalz 81 (9); 186 (9). Schott 51 (3); 57 (1); 755 (3). Schunke 20 (3); 4909 (3). C. Schunke 155 (7); 897 (7); 907 (7). Schunke V. 2967 (12); Бо 3261 (3). Seaverns 13 (3). Seemann (1); 72 (2); 369 (2). Sehnem 2977 (9); 6501 (9). Seibert 583 (1); 587 (1). Seifert 42 (3). Seiler 718 (2). Seymour 0 (2). Shattuck 189 (1); 283 (3). Sherring 87 (3); МУ 131 (4). Shillingford & Adams 98 (4). Sieber 363 (4). Sintenis 6271 (4); 6443 (4); 6533 (4). Skutch 976 (8); 2531 (3); 2715 (8); 4681 (13); 4790 (3). Smith 358 (3); 1121 (3); 1362 (3); 2449 (3); 5168 (7); 5675 (1). Smith et al. 1340 (1); 1710 (3); 2455 (8); 1620 (3). A. R. Smith et al. 856 (2); 1037 (8); 1062 (5); 1095 (2); 1545 (8); 1620 (3); 1875 (10); 1961 (2); 2455 (8). H. H. Smith 958 (3); 968 (8); 1126 (4); 2693 (1). H. i Smith & С. W. Smith 710 (4); 961 (3); 2450 (3). J Smith 6885 (8). mith & Idrobo 1523 (12); TUS 1). Sodiro 66 (1 0): 8. 74 (6); 8.75 (6). Spannagel 143 5 3 (6). Sperling 4945 (8); 4995 (8). Sperling et al. 6035 (1); 6381 (1). Spruce 5689 (10); 5697 (6). Standley 8744 (3); 11234 (3); 26184 (1); 27218 (2); 29372 (2); 52815 (3); 53554 (3); 54153 (1); 60656 (2); 68216 (8); 71321 (8); 86332 (8); 86710 (8); 86744 (8); 88974 (3); 89091 (3); 90658 (8); 90667 (8). Stern et al. 235 (1); 705 (3); 824 (1). Stevens 6677 an ro et al. 6582 (3). Steyermark 41908 (8); 2 (3); 44894 (3); 45308 (1); 45507 (3); 49389 з | Sn EN 61143 (3); 61148 (4); 61569 (4); 94822 mea ey 95249 (4). Steyermark et al. 121451 (4); 121514 (3); 121590 (4); 127064 (2); 127131 (5) 127134 (5); 127236 (2). Stork 1764 (3); 1767 (8); 2678 (8). wick et al. 3574 (3). Stübel 16a (8); 878 (7); 1099a (7). Sytsma 2314 (2). Tate 401 (9). Taylor 17819 (8). Thieme 4 (1); 5675 (1). Tonduz 9473 (8). Tryon 7399 (4). 5228 (3). Tyson 2237 (2). Tyson et al. 4835 Ule 70 (9). Wc 2304 (3); 2802 (3). Underwood & Earle 615 (3); 9 Valverde 510 | уап der Werff 402 (1); 660 (7); 3403 (3). van der Werff & González 4599 (5). van der Werff & Palacios 10403 (12). van der Werff & van Hardeveld 6766 (8). van der Werff & TR 702 (4). Vásquez 977 (1); 2485 (1). Ventura A. 3267 (8); 14368 (8); 16002 (8). Viereck 1 (9). von Türckheim 630 (8); 2824 (3); 8048 (3). Wacket 162a (9). Wagner 809 (8). Walden 41 (7); 84 (3). Walker T6178 (11). Webb 66 (9). Welch 19856 (2). Wercklé 1713 (8); 17432 (1). Wetmore & Wood- worth 105 (1). Whitefoord 4174 (4); 4549 (4); 6164 (3). Wiggins 11095 (6). Wilbur et al. 7473 (4). Williams 1026 (1); 1076 (3); 1076a (3); 1091 (12). Wilson dl (8); 161 (3); 280 (3); 577 (3). Witherspoon & With spoon 8797 (1). Woodson et al. 1601 (2). Wright 1026 3); 1027 (3); 1086 (3). Zardini 10966 (9). Zardini & Aguayo 9501 (9). Zu- luaga R. 986 (1); 332 (3). — ~ A REVISION OF THE GENUS PLEUROTHYRIUM (LAURACEAE)! Henk van der Werf” ABSTRACT Pleurothyrium, a genus of Lauraceae mostly consisting of trees, ranges from Guatemala to Bolivia and possibly ew. Taxonomic history, characters, and phylogenetic classification are discussed. Data on phenology, habitat preference, and altitudinal range are presented when available. In the course of identifying neotropical Laura- ceae accumulated in MO or received on loan from other institutions, Ї found several undescribed spe- cies of Pleurothyrium. Four of those were pub- lished in earlier papers (van der Werff, 1987, 1988), but because the number of novelties kept increasing, I decided that writing a revision of the genus was the best way to deal with the many new aa Notable among those are the collections by J. Cuatrecasas from Choco and Valle, Colombia, which include three striking, undescribed species. Recent collections by Missouri Botanical Garden staff members have also added significantly to the number of species. In this paper 19 species of Pleurothyrium are described, which brings the number of accepted species to 39. Because so many undescribed species were collected recently, I an- ticipate that the number of Pleurothyrium species will continue to grow. Another indication of this future growth is that 12 species are known from the type collection only, and an additional seven are known from two collections. Only three species have been collected more than 20 times. The large number of undescribed species found during this study shows clearly that our knowledge of neo- tropical Lauraceae is still incomplete. It is not co- incidence that the largest percentage of unde- scribed species comes from Ecuador, a country where an active tree collecting program has been in place during the past seven years. Seven of the 12 Pleurothyrium species known from Ecuador are new to science. There is little information available about eco- nomic uses of Pleurothyrium species. In central Peru the wood is used as timber and is considered to be of reasonably good quality. The wood of several Colombian species is used for making boards or canoes. It is likely that the wood of other species is also used locally, but because most of these species are rare and several do not grow into tall trees, their use is economically not very important. MATERIALS This revision is based on about 300 collections, some with many duplicates, found in many her- baria. A general difficulty with a generic revision in the Lauraceae is that many specimens are not identified to genus and therefore will not be sent when a loan of material of a particular genus is requested. One can solve this problem by personally selecting loans or by having a knowledgeable friend do so. In this case, Jens Rohwer kindly annotated all Pleurothyrium specimens in F, NY, and US to genus for me. In addition to these collections, I received loans from B, BM, BR, C, E, INPA, K, L. LE, P, S, U, UC, and VEN. I also had the opportunity to study the specimens of COL, IAN, 35 , and SP and thank the curators of these herbaria for loans and assistance during my visits. Recent collections by Missouri Botanical Garden botanists (notably B. Hammel, Costa Rica and Pan- ama; G. McPherson, Panama and Colombia; A. Gentry, Colombia and Peru; C. Dodson, Ecuador; ' This study would not have been к without the collecting efforts of many collectors, especially J. Cuatrecasas and Missouri Botanical Garden staff members, as well as the often large loans I received from several herbaria. I gratefully acknowledge their MU J. Crisci’s assistance with the cladistic analysis is gratefully acknowledged. I thank several botanists (C. Berg, G. McPherson, R. Moran) who tried out the key and suggested improvements. The illustrations, both older and new ones, were made by J. Myers. The project was supported by National Science Foundation grant BSR-8918096. 2 Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166-0299, U.S.A. ANN. MISSOURI Bot. GARD. 80: 39-118. 1993 40 Annals of the Missouri Botanical Garden D. Neill, Ecuador; C. Cerón, Ecuador; W. Pala- cios, Ecuador; R. Vásquez and J. Jaramillo, Peru) have contributed a great deal to this study. TAXONOMIC HISTORY AND GENERIC RELATIONSHIPS The genus Pleurothyrium was published by Nees in Lindley's Natural System of Botany, ed. 2, which appeared in July 1836 fide Taxonomic Lit- erature 2, followed soon (October 1836) by a more detailed treatment in Nees's Systema Laurinarum. Initially, Nees included two species, P. сћгузо- phyllum and P. bifidum, in his genus but added in his Systema Laurinarum P. poeppigii and P. cuneifolium. He described the genus as having six staminal glands and characterized it by the large glands and the lateral anther cells, which open back-to-back. Nees did not designate a type spe- cies. Вешћат & Hooker (1880) expressed the opinion that P. chrysophyllum Nees fits the ge- neric description best, and this species has been generally accepted as the lectotype species (Pax, 1889; Kostermans, 1952; Коћмег, 1986). 1848, Nees described a fifth species of Pleurothy- rium. Meissner (1864) added three species to Pleu- rothyrium, all of which are now placed in other — = - genera. [n his generic description, he mentioned the presence of a hypogynous disk forming a nar- row ring on which stamens and glands are inserted. Meissner also stated that only six glands were pres- ent at the base of stamens of whorl III and that the staminodia do not possess glands at their base, this in contrast to Nees (1836a, b), who reported glands at the base of the staminodia. Baillon (1870) essentially accepted Meissner's circumscription of the genus; he stated that the disk was situated in the receptacle and that its apex was curved inward. he number of staminal glands remained six. Ben- tham & Hooker (1880) described Pleurothyrium as having a fleshy disk with six lobes alternating with six stamens, this in contrast to earlier authors, who described the disk as a narrow ring. Bentham & Hooker (1880) also mentioned the presence of glands at the base of the stamens of whorl III but did not describe how the glands are placed in re- lation to the disk. Actually, their description of the disk applies to the glands; the later confusion be- tween disk and glands has its beginning here. Pax's description (1889) is brief and essentially the same as Bentham & Hooker's. Mez (1889) did not men- tion the presence of a disk in his description of Pleurothyrium but stated that all nine stamens had two glands at their base, which were almost always fused. In comparison to Nees's description, two other differences are noteworthy: Mez did not men- tion staminodia in his generic description, but Nees included biglandular staminodia in his description; according to Mez, the outer six anthers have two introrse and two extrorse, rarely lateral dehiscing locelli, whereas Nees described the locelli as lateral. Mez was the last person to revise Pleurothyrium; he excluded the three species Meissner (1864) described, but added one species described by Meissner (1864) as a Nectandra. Mez recognized five species in the genus. In the 1930s, several species of Pleurothyrium were described (Ducke, 1930; Smith, 1935; Schmidt, 1928, 1933) without any change in generic circumscription. MacBride (1938) essentially accepted Mez's description of ibed two Venezuelan the genus. Lasser (1942) d species of Pleurothyrium, but did not discuss the generic concept. Kostermans (1952, 1957) looked critically at the validity of Pleurothyrium as a genus. He found that Pleurothyrium only differed from Ocotea in the position of the anther cells and the presence of glands at the base of all stamens. In his opinion, these differences were not sufficient for the recognition of Pleurothyrium as a genus, and he proposed to treat it as a section (1952) or subgenus (1957) of Ocotea. In 1952, he formally transferred all Pleurothyrium species to Ocotea. Bernardi (1962) followed Kostermans and treated Pleurothyrium as a subgenus of Ocotea. His de- scription, however, partly followed Meissner: he mentioned a disk in the form of a narrow ring with the stamens inserted on the ring, but differed from Meissner in crediting all stamens with glands at their bases. Hutchinson (1964) fell back on Ben- tham & Hooker’s concept of Pleurothyrium, with fleshy disk, this with six teeth alternating with the stamens as well as glands at the base of the rat three stamens. He accepted five species, half number Kostermans (1952) Еа Allen ORE discussed the concept of Pleurothyrium in detail and concluded that it deserved generic status, char- acterized by the presence of glands at the base of all stamens and the position of the anther cells. It is worth noting that Kostermans (1952) did not accept that these characters warranted recognition of Pleurothyrium as a genus. Kubitzki (1981) ex- pressed the opinion that Pleurothyrium is a het- erogenous genus and that its species actually belong to Ocotea and Nectandra. In contrast to this view, Коћмег & Kubitzki (1985) stated (based on a detailed analysis of Ocotea sens. lat.) that Pleu- rothyrium is а valid genus characterized by the lateral position of the locelli and the greatly en- larged glands. They pointed out that in Pleuroth y- rium only the inner three stamens (not all nine Volume 80, Number 1 van der Werff 41 1993 Revision of Pleurothyrium TABLE 1. Distribution records of Pleurothyrium species. Bolivia P. intermedium, P. poeppigii Brazil P. acuminatum, nobile, P. panurense, P. parvi amapaense, P. ampli ed P. cuneifolium, P. insigne, P. intermedium, P P P orum, P. ncei, P. undulatum, P. vasquezii igii, P. pra е P. glabritepalum, P. grandiflorum, P. margin- , P. tomiu ipod e triana ciflorum, Fit . giganthum, P. glabrum, P. insigne, P. ји P. parvi- lia ати ilosum, P. racemos , P. cinereum, "P. cuneifolium, P. insigne LP. ile, P. panurense, P. parviflorum, P. poeppigii, P. vasquez- Colombia Р, bracteatum, P. crass | = ale, P. poeppigti, P. synan Costa Rica Р. golfodulcense, Р, hexaglandulosium, P. palmanum, P. pauc Ecuador P. cinereum, P. cuneifoliu florum, P. poeppigii, P. A P. tomiwahlii, P. trianae, P. wil Guatemala P. westphalii Honduras P. trianae Nicaragua P. trianae Panama Р. pi ie P. palmanum, Р. p Peru P. acuminatu Ve P. brochidodrom um жолы ок Р. maximum, P. : ii, P. williamsii, P. ыз о. P. tri Venezuela P. amapaense, P. costanense, P. M e narium P. trianae stamens) have glands at their base, thus accepting Nees's description and rejecting that of Mez and subsequent authors. Rohwer (1986) further dis- cussed and illustrated this concept and accepted eight species groups in Pleurothyrium, based on a survey of the types of the published species. Van der Werff (1987, 1988) accepted the circumscrip- tion of Pleurothyrium given by Rohwer & Kubitzki (1985), as did Burger & van der Werff (1990); in these en an additional five species are described. However, in recent British literature (Willis, 1973; Mabberley, 1987) only one species is attributed to Pleurothyrium; in all likelihood, these authors follow Bentham & Hooker (1880), who wrote that only one species matched Nees's description. In this study I accept Pleurothyrium as a dis- tinct genus, most closely related to Ocotea and Nectandra. It differs from these two genera in two easy to observe androecial characters. In Pleu- rothyrium the outer six stamens have at least two anther cells in a lateral position (and often all four are lateral), and the six staminal glands are strongly enlarged and always grow outward between the six outer stamens and separate these stamens. In many species the glands become confluent and form a large, pillowlike mass with the anthers embedded in the surface. Other characters that are uncom- mon or lacking in Ocotea and Nectandra, but frequently present in Pleurothyrium, are the po- sition of the inflorescences in the axils of cataphylls (not or rarely in axils of normal leaves) and the presence of a marginal vein in the leaves. Whether Ocotea or Nectandra is the closest relative of Pleu- rothyrium is still an open question. This study has not pointed to a particular genus or species group as the closest relative. Van der Werff (1991) published a key to the New World genera of Lauraceae, including Pleu- rothyrium. Although the greatly enlarged and fused glands are unique among neotropical Lauraceae, they also occur in the Asian-Australian genus Endiandra. Several Australian species are illustrated in Hyland (1989). only character Pleurothyrium and Endiandra have The enlarged, fused glands are about the in common. Endiandra has only three two-celled stamens and does not have a cupule subtending the fruit; it is closely related to Beilschmeidia. Thus, the character of the glands is considered to be an example of convergence and does not indicate a close relationship. DISTRIBUTION AND ECOLOGY Pleurothyrium species have been collected from Guatemala, the northernmost extension, to Bolivia and southern Brazil, the southernmost extension. It has been collected in all countries between these extremes with exception of El Salvador and the three Guianas. The genus is absent from the West Indies. Table 1 shows the species present in each country. The greatest concentration of species oc- curs in Peru, Brazil, Ecuador, and Colombia, while Pleurothyrium is represented by only a few species in the other countries. The Brazilian species are, with exception of P. amplifolium, Amazonian, and most occur in the upper part of the Amazon basin; only one or two species have been found down- stream from Manaus. Most species are only known neifolium and P. poeppigii occur at 1,800 m near Oxapampa, but these species are better represented The Ecuadorian P. obovatum 1s only known from three collections at 1,200-1,800 m. In Venezuela, P. costanense has been collected in the lowlands. Annals of the Missouri Botanical Garden at 2,200 m, but is more common around 1,000 m; the poorly known P. steyermarkianum has been collected at 1,400 and 2,000 m. Our knowledge about the habitat requirements of Pleurothyrium species is very poor. er col- lections usually do not have habitat information other than "forest" and besides, many species are known from very few collections. For a reliable estimate of the habitat requirements, | would want at least ten collections with the same habitat in- formation. This is only available for P. parviflo- rum, à species restricted to seasonally flooded for- ests in Peru, Brazil, and Ecuador. It is to be expected in the Amazonian forests of Colombia as well, but it has not yet been collected there. Pleurothyrium panurense is known from fewer collections, but all of these were made in flooded forest. Most of the new or recently described species are from Colombia and Ecuador. For instance, six of the ten species from Colombia are new; for Ecuador these figures are seven out of twelve, but for Brazil it is two out of eleven species. Because of the Projeto Flora Amazonica, Amazonian Brazil has been reasonably well collected, probably better than Colombia, and it seems unlikely that many additional Pleurothyrium species will be collected in Brazil. On the other hand, Colombia has at least three additional, but incompletely known species. Taking this into consideration, I expect that the ond diversity will be in Peru, Ecuador, and Colom MORPHOLOGY AND TAXONOMIC CHARACTERS At the beginning of this survey of the taxonomic characters of Pleurothyrium species, it should emphasized that about half of the species are known from only one or two collections. This is not enough for a good understanding of the morphological vari- ation within the species, and there are difficulties interpreting differences in degree of pubescence, leaf size, and color of pubescence. Therefore, | decided not to recognize taxa based solely on these characters, even though, for instance, the Peruvian collection of P. insigne and the Costa Rican col- lection of P. hexaglandulosum look different from the types. Additional collections of many species are needed for a better understanding and, when available, may well lead to changes in the classi- fication here proposed. abit. Pleurothyrium species vary from small to tall trees. There is no information about different types of tree architecture represented in the genus, but it is likely that species with clustered leaves have a different growth type than species with evenly spaced leaves. Habitat. habitat is poor. The great majority of the species are restricted to the wet lowlands (below 500 m). Pleurothyrium cuneifolium and P. poeppigii have As with habit, label information on been found from the lowlands up to 1,800 m, an unusually wide range of altitude. Pleurothyrium costanense 18 restricted to cloud forest in Vene- zuela, between 800 and 2,200 m. Pleurothyrium trianae occurs occasionally up to 1,200 m, but is most common at elevations between 300 and 600 m. Pleurothyrium palmanum is known from four collections in Costa Rica and Panama, between 1,000 and 1,600 m. Finally, there are three mon- tane species known from only three or fewer col- lections: P. bracteatum at 1,000 m in Colombia, P. obovatum at 1,200-1,800 m in southern Ec- uador, and P. racemosum at 900 m in Panama. More detailed habitat preferences are only known for P. panurense and P. parviflorum: both are restricted to flooded forente i in the upper Amazon basin. In the vicinity of Iquitos, Peru, I collected P. parviflorum in flooded forests on sandy and clay soils, but never found it outside flooded forests. Fistulose twigs. Four species, Pleurothyrium cuneifolium, P. obovatum, P. parviflorum, and P. poeppigii, have consistently fistulose twigs, while his is a reliable and readily visible character mentioned by Nees (1836b) in his descriptions. Mez (1889), cit- ing Poeppig (In ramulis revera fistulosis degunt P. trianae has rarely fistulose twigs. formicarum agmina pessime pungentia), referred to the ants living in the hollow twigs. On several, but certainly not all, labels of these species, the presence of aggressive ants is mentioned; I foun P. parviflorum sometimes inhabited by aggressive ants and sometimes by timid ones, although each tree was inhabited by one kind only. In Lauraceae fistulose twigs inhabited by ants are not restricted to Pleurothyrium, but also occur in Ocotea javi- tensis (HBK) Pitt., O. dendrodaphne Mez, the species related to 0. atirrensis Mez & J. D. Smith, including Aiouea vextrix van der Werff, a Licaria species from Costa Rica, and an Aniba species from Iquitos, Peru. Most Lauraceae with hollow from northern Ecuador, collected at 2,400 m el- evation, also had hollow twigs inhabited by ants, as did an undescribed Cinnamomum species from central Ecuador, collected at 2,200 m elevation. Both high-elevation species were inhabited by non- aggressive ants, which did not defend the twigs and scarcely left the hollow core when the twigs were broken. In Pleurothyrium cuneifolium and P. poeppi- Volume 80, Number 1 1993 van der Werff 43 Revision of Pleurothyrium gii, the main axis of the inflorescences is not rarely fistulose. А good example is the type collection of P. poeppigii. These fistulose inflorescences tend to have very short lateral branchlets with the flow- ers rather densely fasciculate. In the past, all col- lections with this striking inflorescence shape were annotated as P. poeppigii, but floral characters show that this inflorescence shape also occurs in P. cuneifolium. In inflorescence can be reduced and carry a dense cotea javitensis the entire cluster of flowers; such specimens have also been identified as Pleurothyrium (for instance, Klug 2908 as P. densiflorum Smith). Exactly how the activity of ants induces this peculiar inflorescence shape is not known to me. Leaf position. Most species have alternate leaves, the predominant arrangement in Laura- ceae. About ten species have the leaves clustered at the tips of the branches, potentially a useful character to separate species. However, | found this character not as useful as I hoped, because in some species the leaves are more or less clustered, and in some others the leaves are so large that an herbarium specimen consists of one leaf and an inflorescence and does not show the arrangement of additional leaves. Thus, I used leaf position in the key only when separating relatively small-leaved Leaf shape ranges from elliptic vate. [n most species it is rather stable, but I „е it difficult to express leaf shape accurately in words and have not used it very much in the key. A few striking leaf shapes should be mentioned: in Pleurothyrium maximum, P. insig- ne, and P. williamsii the leaves are gradually narrowed toward the base and at the base are abruptly rounded to cordate; in P. racemosum and . pilosum the leaves are obtuse or rounded both at the base and apex. Pleurothyrium tomiwahlii has a similar leaf shape, but the base is sometimes acute, not obtuse. Leaf venation. Leaf venation includes both camptodrome and brochidodrome types as well as their intermediates. All species with small flowers and free staminal glands (the first eleven species in the key) have predominantly camptodrome ve- nation, while the species with larger, rotate flowers have camptodrome or brochidodrome venation. Leaf size or shape does not seem to affect venation type strongly; some species with large, obovate leaves have a strongly developed marginal vein (Pleu- rothyrium insigne, P. maximum), but others (P. tomiwahlii, P. giganthum) do not. Among the species with small leaves, P. marginale has a strong marginal vein, but P. golfodulcense, P. glabrite- palum, and P. westphalii do not have a marginal vein. Although venation type seems characteristic for each species, the various types are hard to describe accurately and have not been used much in the key. A detailed study of the venation types of Pleurothyrium is in preparation and will be published separately. ndument. The type of indument is a very useful character for identification. Because indu- ment on leaves and twigs can be different, I have mostly used the indument on the lower leaf surface, in which I recognize three types. A little more than half the species have glabrous leaves or carry a has erect straight or tortuous (in Pleurothyrium racemosum) hairs, which never cover the entire surface and which, in some species, can be sparse. The third group has a dense (mostly tomentose) pubescence, which completely covers the lower leaf surface. With the exception of three species (P. crassitepalum, P. panurense, and P. tomiwahlii), these pubescence types can easily be recognized. In the three exceptions, it is sometimes difficult to choose between group one (glabrous or sparse, appressed pubescence, the surface visible) or group three (dense pubescence, covering the entire sur- face). These species are therefore included twice in the key. The three groups based on pubescence types are not natural; sometimes closely related species (for instance, P. insigne-P. maximum) are placed in different groups. These groups are formed solely to allow easy identification. Inflorescences. Inflorescences of Pleurothyri- um are as a rule thyrso-paniculate in the sense of Weberling (1985). They consist of a central axis with a number of alternately positioned lateral axes. The lateral axes are once to several times cymosely branched; the central flower of each cyme flowers before the lateral ones and is usually absent in all but the terminal cymes. Of each cyme, the lateral elements may again be cymosely divided, but the central element always develops into a single flow- er. Sometimes the lateral elements of the lateral branches fail to develop and then the inflorescence is racemose (as in P. racemosum). In most species inflorescence bracts are early deciduous and absent at anthesis. Exceptions are Pleurothyrium bracteatum and P. golfodulcense, both of which have rather large bracts at anthesis. The thyrso- paniculate inflorescence type is very in all larger genera, such as Ocotea, Persea, Aniba, Licaria, Aiouea, Endlicheria, and Cinnamomum. In general, the inflorescences are found in the axils of cataphylls just below the terminal bud. Annals of the Missouri Botanical Garden Although they appear terminal, I have always found them to be lateral. In the species group charac- terized by erect tepals, inflorescences occur oc- casionally in the axils of regular leaves, but this is uncommon. In Pleurothyrium giganthum, the in- florescences are inserted in the axils of cataphylls below the leaves. However, many species are rep- resented by very few collections and | am not certain that the position of the inflorescences, below or above the leaves, is of any significance. The nearly consistent presence of inflorescences in the axils of cataphylls is a helpful character for the identification of Pleurothyrium. The great ma- jority of species of Ocotea and Nectandra have inflorescences in the axils of normal leaves, but such species as O. fasciculata (Nees) Mez and N. megapotamica (Spreng.) Mez are exceptions. Flowers. The flowers offer a number of dis- criminating and/or interesting characters. Eight species have more or less erect tepals at anthesis, this in. contrast to the remaining species, which have a rotate flower shape. Six of these eight (the exceptions are Pleurothyrium acuminatum and P. amapaense) also have reflexed margins of the tepals. These reflexed margins are not a result of aging of the flowers; I have seen P. parviflorum and P. trianae in the field and even young flowers have reflexed margins of the tepals. On herbarium sheets one can see that sometimes buds have al- ready reflexed margins of the tepals. I consider the species with erect tepals, together with P. cunei- folium and P. obovatum, a closely related group, characterized by flower shape, absence of a mar- ginal vein, similar indument on the pistil, occur- rence of fistulose stems, tepals. and the rather slender Most other species have rotate flowers and broadly ovate to roundish tepals. Species with flow- ers more than 10 mm in diameter show a tendency toward unequal tepals, with the inner tepals slightly shorter and narrower than the outer ones. Th group of species with erect tepals always has equal tepals. In one species, Pleurothyrium tomiwahlii, the tepals are strongly reflexed at anthesis and become parallel to the pedicel in older flowers. The pubescence on the outside of the tepals is frequently much denser or longer on the areas exposed to the outside in bud than on areas pro- tected from the outside. Thus, tepals and a basal, triangular patch on the inner the outer three tepals have a similar pubescence. This is an unusual character in Lauraceae, but is fairly common in the group of species with rotate flowers. It does not occur in the group of species with erect tepals. The pubescence on the inside of the tepals is variable and is a useful character in separating the taxa. A few species (Pleurothyrium cuneifolium, P. glabrifolium, P. glabritepalum, P. panurense, and P. prancei) have a glabrous inner surface of the tepals (in P. prancei the outer three are gla- brous inside, the inner three papillose). Occasion- ally, some scattered papillae or hairs may be pres- ent, especially near the base of the tepals, but a homogeneous cover of papillae or hairs is lacking. Тће group of species with pubescent inner surfaces of the tepals is also small; it includes P. bracteatum, P. giganthum, P. crassitepalum, P. tomiwahlii, P. grandiflorum, P. marginale, P. pilosum, and P. westphalii. In P. racemosum the inner tepals are pubescent and the outer three have a large, triangular, pubescent patch at the base and are otherwise papillose. The remaining species all have a papillose i inner surface. e different androecial configurations are the most interesting feature of Pleurothyrium. In the group of species with erect tepals the enlarged glands grow outward, and the six outer stamens are separated from each other by the glands (Fig. 1). In this flower type the glands do not grow around the outer anthers, and the glands do not touch each other at the periphery of the flower. The stamens have relatively long filaments and the anther cells occupy nearly the entire anther (Fig. 2). The anther is somewhat curved inward, but not horizontal; thus, the anther cells are arranged in an upper and lower pair. The lower pair is always toward the outside in relation to the upper pair. Because of the long filaments, the anthers are clearly raised above the glands. In the next stage, the glands begin to enclose the outer stamens, while the an- thers become horizontal, have lateral anther cells, and are not raised above the glands. An example is P. vasquezii (Fig. 3). In other species the glands completely surround the outer stamens and appear The contact lines between the glands are sometimes clearly visible (Fig. 4) or almost disappear (Fig. 5). In the final stage all stamens are crowded in the center to form a wall around the stamens. of the flower, surrounded by a broad wall of com- pletely fused glands (Fig. 6). The enlarged glands do not confine the stamens to the center of the flower in all species. In some species the glands grow between the stamens and separate them, es- Fig. 7). This configuration is not common; it occurs in P. golfodulcense, P. tomiwahlii, P. crassitepalum, and P. racemosum. The species pecially the stamens of the second whorl cinereum, P. with this flower type also tend to have stamens and Volume 80, Number 1 1993 van der Werff 45 Revision of Pleurothyrium FicunEs 1-4.— right.) Stamen (Whorls I-II) of P 1. (Top left.) Flower of Pleurothyrium poeppigti, tepals removed (Palacios 2775 t.) . acuminatum (van der —2. (Top pw Werff 9967). — 3. (Bottom left.) Flower of P. vasquezii, tepals removed (Vasquez 7889). — 4. (Bottom right.) Flower of P. tomentellum, seri removed (Gudino 12). glands forming a taller ““dome””; this dome сап ђе as high as its width. In P. tomiwahlii the dome is also seated on a short column, thus further raising the androecium. All species with horizontally bent anthers and glands that are (nearly) enclosing the outer stamens have rotate flowers. These species have, on the average, distinctly larger flowers than the species with erect tepals. Unfortunately, data on pollination are available, but it seems likely that the species with large, rotate flowers are pol- linated by different organisms than those with small flowers with erect tepals. The indument of the pistil shows considerable variation. Sixteen species were found to have the pistil more or less covered with papillae. Often the lower part of the ovary was glabrous and the density 46 Annals of the Missouri Botanical Garden S 5-8. Р. о I: remo (Jimenez 648).— (Juncosa 1675). of the papillae decreased also on the style, but papillae were easily seen on the upper part of the ovary. This condition is present in all species with erect tepals, as well as in Pleurothyrium bifidum, P. brochidodromum, P. nobile, P. palmanum, P. panurense, P. vasquezii, and P. williamsii. Ten species have a glabrous pistil: P. tomiwahlii, P. crassitepalum, P. glabrifolium, P. glabritepa- — 5. (Top ci vod of P. insigne, _tepals removed (Palacios 4388). 6. (Top right.) Flower о . (Bottom left.) Flower of P. golfodulcense, tepals removed 8. (Bottom s ) Flower of P. broc Pide aim. tepals removed (Vasquez 7865). lum, P. grandiflorum, P. marginale, P. pilosum, P. racemosum, P. giganthum, and P. westphalii. The remaining species have a (partly) pubescent pistil. In a few species one finds six bald spots on the upper part of the pubescent ovary correspond- ing to the contact points between the six glands and the ovary. I have not used the indument of the pistil frequently in the key, because the pu- Volume 80, Number 1 1993 van der Werff 47 Revision of Pleurothyrium bescence is only readily visible on the style in a few species. Fruits and cupules are unknown for many spe- cies. When known, the cupules are relatively deep- ly cup-shaped and have nearly always conspicuous lenticels or warts. Because of the deep cupules, fruiting collections have sometimes been misiden- tified as Aniba sp., but venation (Pleurothyrium has in general more lateral veins and sometimes Ф distinct marginal vein) апа indument (ferruginous tomentum is a good indication for Pleurothyrium) help separate the two genera. Frequently, the tep- als are persistent on the young cupules. The tepals eventually become damaged and break off or the growing cupule causes the tepals to fall off. In this character, Pleurothyrium species differ from Nec- tandra, where the tepals tend to be basally united and fall off as a unit, together with the stamens. Unusual characters are the presence of six large lobes on the cupules of highland populations of P. poeppigii in the Selva Central in Peru; the com- mon name *'roble zapallo" alludes to this (zapallo — squash). Young cupules of P. giganthum are sharply hexangular, a feature not found in other species. In P. cinereum the tepals are persistent on the mature cupule, which may be double-rimmed. Pollen. In a survey of pollen types of neo- tropical Lauraceae, Raj & van der Werff (1988) found that, with the exceptions of Cryptocarya and Cassytha, all investigated species had inaper- turate, spinulose pollen grains. Pollen of Pleurothy- rium was found to be very similar to that of Ocotea, the main differences being size and ultrastructural details of the sporoderm. Pollen of Nectandra is more distinct. Three species of Pleurothyrium (P. parviflorum, P. trianae, and P. cuneifolium) were included in this study. Because these three species belong to the group with erect tepals and erect or curved anthers, | checked pollen of two species with rotate flowers and bent anthers (P. insigne and P. racemosum). These two species have the same inaperturate, spinulose pollen that is present in nearly all neotropical Lauraceae. Because all five investigated Pleurothyrium species have the same pollen type, it seems unlikely the pollen will be useful in the classification of Pleurothyrium species, and pollen of the other species has not been studied. CLADISTIC ANALYSIS A cladistic analysis, based on the principles of phylogenetic systematics as developed by Hennig 6), was made of Pleurothyrium, using the 39 species recognized in this revision as terminal taxa. Because several species of Pleurothyrium are still poorly known, the choice of characters to be used in the analysis was limited to those characters vis- ible on a herbarium sheet. Ideally, one wishes for characters from other fields (such as embryology, anatomy, chemistry) as well as more characters. The characters and the character states used are listed in Table 2; the data matrix is presented in able 3. The choice of outgroup, and therefore also the polarity of the character states, initially posed considerable problems. Of all genera of Lau- raceae, /Vectandra and Ocotea are the closest to Pleurothyrium, based on overall morphology. However, neither of the two seems closer to Pleu- rothyrium than the other. Because Ocotea and Nectandra are very similar in the character states used in this analysis, which of them was chosen as the outgroup seemed relatively unimportant, thus a preliminary analysis was made with the generic concept of Nectandra as outgroup. This analysis resulted in 1,988 trees; after successive weighting 293 trees were generated. А strict consensus tree showed as the most derived group of species those species with erect tepals and free, weakly enlarged glands. This topology ran completely counter to my ideas of trends within Pleurothyrium. Based on knowledge gained from my nine years of study- ing Lauraceae, I consider the following character states in Pleurothyrium as primitive: free, weakly enlarged glands, stamens with only two of the four anther cells lateral and the anthers raised well above the glands, the occasional presence of inflo- rescences in the axils of normal leaves, the absence of a marginal vein, and erect tepals. As derived character states, | regard fused, strongly enlarged glands, stamens with four lateral anther cells, an- thers that are not or scarcely raised above the glands, inflorescences strictly in the axils of cata- phylls, presence of a marginal vein and spreading, rotate (or even reflexed) tepals. These ideas on primitive and derived characters in Pleurothyrium are based on my extensive experience with neo- tropical Lauraceae. Because Nectandra and sev- eral hermaphroditic species of Ocotea (including most South American Ocotea species with perfect flowers) have flowers with spreading tepals, the Pleurothyrium species with erect tepals ended up as derived in the phylogenetic tree. There are some Ocotea species with perfect flowers and erect tepals (for instance, the O. insularis group), but various reasons (entirely different indument of anthers, presence of flattened branchlets of inflorescence) make it highly unlikely that O. insularis can be considered as outgroup for Pleurothyrium. 48 Annals MESES GEN Garden TaBLE 2. Characters and character states used in cladistic analysis of Pleurothyrium. 1. Position of tepals — spreading — erect 2. Margin of tepals plane — inrolled 3. Marginal vein = lacking — present 4. Indument on inner surface of tepals — glabrous — papillose — pubescent 5. Indument of pistil — glabro — papillose — pubescent 6. Indument of floral tube — glabrous — papillose — pube: 1. Glands 8. Stamens 9. Twigs 10. Number of lateral veins 11. Tepals 12. Position of leaves 13. Pubescence on lower leaf surface 14. Position of inflorescence 15. Position of anther cells 16. Pubescence of bracts Character states of 4, 5, 6, and 10 are unordered. scent — [ree, not surrounding stamens = surrounding stamens, + fused adjacen = separated from each other by glands = solid = fistulose 5-9 = glabrous or appressed erect = in axils of leaves = in axils of bracts = 2 lateral = 4 lateral = only on outer surface — on inner and outer surface - > –—">о >" “> о>о–о>—•>=—-о=—оо=—оом=—еом=—оом=—о=о»=-о»-со || A second analysis was made with a primitive species of Pleurothyrium, P. acuminatum, as out- group. This species was chosen because it lacks the derived characters listed earlier, and it also lacks fistulose twigs and has plane, not inrolled, tepal margins. Analysis of the data matrix using the MHennig* and bb* Wagner parsimony options of Hennig86 (Farris, 1988) produced 540 most parsimonious trees (length 79 steps, consistency index 0.25). A strict consensus tree based on these 540 trees showed an alphabetical listing of 32 species, with only seven species slightly better re- solved. The successive weighting procedure on the 540 trees resulted in at least 1,149 trees (overflow) with a length of 110 steps and a consistency index of 0.50. Note that the high value for the length is a function of the weight being scaled up to a value of 10. Although most characters were weighted after the successive weighting procedures, four of the sixteen characters (10, 11, 13, and 14) weight- ed zero. The analysis was repeated without those characters and resulted in 230 most parsimonious trees with a length of 44 steps and a consistency index of 0.34. After successive weighting, the anal- ysis produced 118 trees with a length of 98 steps and a consistency index of 0.58. As a consequence of deleting the four ““weightless”” characters, the analysis resulted in a smaller number of trees with fewer steps and a higher consistency index: these four characters must have contributed a great deal of homoplasy to the first analysis. А strict consensus tree based on the 118 trees of the second analysis is presented in Figure 9a, while one of the 118 equally parsimonious trees, selected at random, is presented in Figure 9b. The following comments apply to both cladograms. The changes in character states throughout the clado- gram are added to the tree. In both cladograms, the species group I consider primitive forms the basal group. The second, largest group is not well Volume 80, Number 1 1993 van der Werff 49 Revision of Pleurothyrium TABLE 3. Data matrix for cladistic analysis. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Nectandra ? 0 0 | 2 2 0 0 0 1 0 0 ? 0 1 ? acuminatum 1 0 0 1 2 0 0 0 0 0 0 0 0 1 0 0 amapaense 1 0 0 1 1 0 0 0 0 0 0 0 0 1 0 0 amplifolium 1 1 0 1 0 0 0 0 0 0 0 0 0 1 0 ? bifidum 0 0 0 1 1 1 1 0 0 1 2 0 1 1 1 0 bracteatum 0 0 | 2 2 2 | 0 0 2 1 0 1 1 1 1 brochidodromum 0 0 1 1 1 0 1 0 0 1 0 1 1 1 1 1 cinereum 0 0 0 1 2 2 1 1 0 1 1 0 0 1 1 0 costanense 0 0 | 1 2 0 0 1 0 1 0 0 1 0 0 0 crassitepalum 0 0 1 2 0 0 1 0 0 1 1 1 0 1 1 ? cuneifolium 1 1 0 0 1 0 0 0 1 ? 0 0 ? 0 0 1 giganthu 0 0 1 2 0 0 1 0 0 2 1 1 1 1 1 1 glabrifoli 0 0 0 0 0 0 1 0 0 1 0 0 ? 1 1 0 glabritepalum 0 0 0 0 0 0 1 0 0 0 1 0 1 1 1 2 golfodulcense 0 0 | 1 2 2 1 0 0 0 0 0 0 1 1 1 grandiflorum 0 0 0 2 0 2 1 0 0 2 1 0 0 1 1 ? hexaglandulosum 0 0 1 0 2 2 0 0 0 2 0 0 0 0 0 ? insigne 0 0 1 1 2 2 1 0 0 2 1 0 1 1 1 0 intermedium 1 1 0 1 0 0 0 0 0 1 0 0 0 1 0 0 marginale 0 0 1 2 0 0 1 0 0 ? 1 0 0 1 1 ? maximum 0 0 1 0 2 2 0 0 0 2 0 0 ? 1 0 0 obile 0 0 0 1 l 0 1 0 0 0 0 0 0 1 1 0 obovatum 0 0 1 1 0 0 0 0 1 2 0 0 ? 1 0 ? palmanum 0 0 0 1 0 0 0 0 0 0 0 0 1 1 1 | panurense 0 0 1 0 0 0 1 0 0 1 0 1 0 1 1 0 parviflorum 1 1 0 1 0 0 0 0 1 1 0 0 0 0 0 0 pauciflorum 0 0 1 1 2 2 1 0 0 2 0 0 1 1 1 1 losum 0 0 0 2 0 2 1 0 0 0 ? | 1 1 1 ? peoppigil 1 1 0 1 0 0 0 0 1 2 0 0 0 1 0 1 prancei 0 0 1 0 2 2 ? 0 0 1 1 0 ? 1 1 0 racemosum 0 0 1 2 0 2 1 0 0 0 0 | | 1 1 0 steyermarkianum 0 0 0 1 0 0 0 0 0 | 0 0 0 1 0 0 synandru 0 0 0 | 2 0 0 0 0 2 0 0 1 1 1 0 tomentellum 0 0 0 1 0 2 1 0 0 1 0 0 1 1 1 0 tomiwahlii 0 0 1 2 0 0 1 0 0 2 1 1 1 1 1 1 trianae 1 | 0 1 0 0 0 0 0 | 0 0 0 0 0 0 undulatum 1 | 0 1 0 0 0 0 0 1 0 0 0 0 0 0 vasquezii 0 0 1 1 1 1 0 0 0 1 0 0 1 0 1 0 westphalii 0 0 0 2 0 0 1 0 0 0 0 0 0 0 0 ? williamsii 0 0 1 1 | 1 1 0 0 2 1 1 0 1 1 1 resolved; the large number of trees before con- structing the consensus tree reflects variation in topology in this group. The cladograms include only four synapomorphies (for characters 1, 2, 7, and 10), indictaing that all other changes are changes in unordered states or homoplasies. Although near- ly all cladograms include some amount of homo- plasy, to have a cladogram for 39 terminal taxa with only four synapomorphies suggests that the data needed for a cladistic analysis were not avail- a Reviewing this attempt at making a phylogenetic classification as a byproduct of a taxonomic revision of Pleurothyrium, one may note several problems. First, the number of characters is much lower than the number of terminal taxa. Second, the poor understanding of relationships within Lauraceae makes the choice of an outgroup somewhat arbi- trary, and therefore the choice of polarity of the character states cannot be defended very well. Because the choice of outgroup influences the clas- sification strongly, the cladogram cannot be any more reliable than the choice of outgroup. Third, the phylogenetic classification is entirely based on gross morphological characters of herbarium spec- imens. These characters were not selected because 50 Annals of the Missouri Botanical Garden 5 rE ACUMINATUM I— — —AMAPAENSE AMPL I FOLIUM f— INTERMEDI UM 2 5 2 | RI ANAE К — —UNDULATUM 5 B poppe 6 соон XIMUN E ETE 11 12 E oe ГГ eee 11 ү pm 1 3 1 1—— ——"^sautz 11 6 8 A A PI NERE LUM Wu Y о Д ТИС ГАС ЧРИ ; ODBIFETALUN n GRANDIFLORUM 5 s A A TNSTGNE +— INSIGNE aa TOMENTELLUN + —WESTPHALII 3 5 7 11 B 4 R Hn. u || u AAA OBILE 4 RGINALE 4 3 rf} P ANURENSE 4 10 RASSITEPALUM ACEMOSUM TOMIWAHLII ILOSUM FIGURE 9а. Strict consensus tree of the 118 equally parsimonious trees. they were regarded as good indicators of phylo- genetic relationships, but because they were all that was available. Ideally, one wishes for characters that are regarded as good indicators of phylogenetic relationships. In reality, one does not have much choice about quantity and quality of characters when 19 of the 39 terminal taxa are known from only one or two collections. In general, a taxonomic classification of a group requires the data needed for the circumscription and identification of the taxa involved. For a phy- logenetic classification one needs, in addition to the data for a taxonomic classification, also data show- ing the evolution within the group over time. Or, to phrase it differently, a phylogenetic classification contains more information than a taxonomic clas- Volume 80, Number 1 3 van der Werff 51 Revision of Pleurothyrium jowan ——31L.— — —^MAPAENSE —————AMPL I FOLIUM INTERMEDIUM 2 |-__-ТАТАНАЕ H——————UNDULATUM lid у лини | — Y NANDRUM 5 RACTEATUM LFODULCENSE AUCIFLORUM rf} P ANURENSE IGANTHUM 5 ROCHIDODROMUN 11 — Пе LABRIPETALUM 7 4 a FIGURE 9b. One of the 118 equally parsimonious trees. sification, and consequently requires a larger input of information. In the case of Pleurothyrium, suf- ficient information is available for taxonomic re- vision, but not enough for a phylogenetic classifi- cation. Obtaining sufficient information for a proper phylogenetic analysis is a project unto itself and making a phylogenetic analysis cannot be merely tacked on to making a taxonomic revision. Relationships between Pleurothyrium species as expressed in the cladogram and as based on mor- phological similarities show some congruence. For instance, I consider P. racemosum, P. tomiwahlii, and P. pilosum as closely related, a relationship expressed in all cladograms inspected. Likewise, the species with erect tepals, usually with inrolled margin of the tepals, are closely related, as shown Annals of the Missouri Botanical Garden in the cladograms; | would also include P. obor- atum in this group, a species placed in the more advanced group in the cladograms. Other pairs of closely related species (at least in my opinion) placed well apart in the consensus tree are P. insigne and P. maximum and, to a lesser degree, P. costanense and P. synandrum, as well as P. westphalii and P. grandiflorum. SPECIES CONCEPT In preparing this treatment, it was necessary to deal with the number of published species whose distinctiveness was less than convincing. In some cases ample recent collections made it clear that certain taxa (for instance, P. reflexum and P. zu- lianense) had to be synonymized. In other cases such collections were not available. In these in- stances I let myself be guided by the principle to only change the status quo if there was a strong reason for doing so. This means that some species whose distinctiveness | doubt аге being maintained (examples are P. amplifolium, P. undulatum, P. amapaense, and P. intermedium) and that other taxa, represented by, in my opinion, inadequate collections, are not being described. My reasons for not describing these taxa are twofold: (a) in- complete descriptions make identification of future, complete collections very difficult; (b) the absolute certainty that specimen X (sterile or in fruit) rep- resents an undescribed species of genus A can turn a few weeks later into nagging doubt. TYPIFICATION Botanists who described species of Pleurothyri- um in the nineteenth century never designated holotypes. Usually they cited only a type collection, often indicating in which herbaria they had seen duplicates of the type collection. When only one specimen was cited, 1 have regarded this as the holotype, even though it was not explicitly anno- tated as such. If several duplicates exist of the type collection, I have listed them as syntypes. In all cases the syntypes I have seen belonged to the same species, and there cannot be any confusion about the application of the names in question. For this reason it seemed unnecessary to designate lectotypes, and I have therefore not selected lec- totypes. Another reason not to select lectotypes is that I very likely have not seen all the existing syntypes, as required by the Code. For instance, I have not seen the specimens at P, and 1 have no doubt that syntypes of some species are present in other herbaria I did not consult. Lectotypification would be advisable if two or more collections are cited as types of a given species, but this is not the case for any Pleurothyrium species. 'TAXONOMIC SECTION Pleurothyrium Nees, in Lindley, Nat. Syst. Bot.: 422. 1836. Ocotea sect. Pleurothyrium (Nees in Lindley) Kostermans, J. Sci. Res. (Jakarta) 1:-122. (Nees in Lindley) engines, Comm. es. Inst. 57: 40. 1957 Gig aha MA Nees (le totype, selected by 1952) 1952. Ocotea subg. Pleurothyrium orest YPE: Pleurothyrium Kostermans, Trees. Twigs solid or in a few specimens fistulose and inhabited by ants. Leaves alternate or clus- tered near the tips of branches, mostly elliptic or obovate; venation camptodrome or brochidodrome. Inflorescences in axils of cataphylls, rarely in axils of normal leaves, thyrso-paniculate, rarely by re- duction racemose, the ultimate divisions cymose. Flowers rotate at anthesis or in some species with erect tepals. Tepals 6, equal or subequal, then with the inner three a little narrower and shorter. Fertile stamens 9, all 4-celled, the outer 6 with the anthers curved inward or sharply bent inward, the cells mostly lateral, sometimes 2 cells extrorse or in- trorse; inner three anthers mostly curved outward, with lateral cells. Staminodia frequently present, but small and inconspicuous. Filaments of the inner three stamens each with two glands near the base, these strongly enlarged and growing outward, sometimes fused and enclosing the outer stamens and the anthers embedded in the glandular tissue. Pistil glabrous, papillose or pubescent, largely im- mersed in the receptacle. Cupule usually deeply cup-shaped, conspicuously warty. Fruit a one-seed- ed drupe KEY TO THE SPECIES OF Pleurothyrium 1 Twigs fistulose sl Twigs solid . Flowers rotate, 7 Dn -14 mm diam.; tepals gla- erect at anthesis, flowers ca. 5 mm diam., tepals papillose on inner surface 4 Terminal buds, inflorescences, and flowers ferruginous-tomentellous; leaves glabrous below . P. obovatum 3. Indument, when present, brown; leaves mostly окны or tomentellous below rarely glabrou | cuneifolium Twigs, inflorescences, and flowers E Or minutely puberulen P. parv m 4. Twigs, inflorescences, or flowers papis or tomentellous ~ = N Volume 80, Number 1 van der Werff 53 Revision of Pleurothyrium 5(4) Lateral veins 14-18 on each side; leaves 15(13). Bracts to 1 cm long present on inflorescence mly chartaceous to pene always during anthesis; marginal vein present; pu- with fistulose twigs ooo . poeppigii bescence on diis twigs and es 5. ateral v -12; leaves chartaceous; reddish brow P. bracteatum rarely with fistulose twigs 0. P. trianae 15. Bracts gene rally lacking on inflorescences 6(1) Tepals at anthesis erect and sometime at anthe MN small); marginal veins isted, the margins usually reflexed; sta lacking or spe. ea on inflores- minal glands free, never coalesced; dia cence bro yellow-brown о 16 eter of 5 mm or less; marginal vein 16(15). e veins 18. EL ane generally wid- lacking, lower leaf surface never with erect t below the middle ________ . synandrum hairs ( 16. lateri veins to 15- 16 leaves did at or 6. Tepals at anthesis rotate, plane, the margins above the middle not reflexed; staminal glands free or co- (16). Tepals subequal, the inner three shorter and alesced; dia r of flowe qual Or narrower than the outer three, both i inner greater than 8 mm (in P. glabrifolium, P an E outer surface pubescent; marginal ve vasquezii, and P. tomentellum flowers 5- pre unn 8 mm diam.); marginal vein present or a 17. Tepals equal, the inner surface papillose o sent, in some species lower leaf surface with puberulous; marginal vein weakly devel- erect hairs ed or lackin 7(6). Leaves + clustered at tips of branches; 18(17). Flowers 10-12 mm diam pubescence on inflorescences and young 8. Flowers to 8 mm diam. _____ omentellum twigs ferruginous-tomentellous .. P. amapaense 19(18). Style gray-pubescent; receptacle deti 7. eaves alternate, not clustered; pubescence inside; indument of lower leaf surface + on infloresce and young twigs gray or appressed, the hairs eit short ... ci ereum (light) brown, never ferruginou 19. Style papillose; receptacle glabrous inside; 9(7) Margin of tepals not reflexed; tepals + indument lo leaf surface tom twisted at anthesis; leaf apices acuminate lous, the hairs + erect, curled . almanum cuminatum 20(12). Lower leaf surface with erect bain. this 8. Margin of tepals reflexed; tepals not Moor indument sometimes sparse cc leaf apices acute 20. Lower leaf surface glabrous or with ap- 9(8). Twigs and pes axes sparsely p pressed hairs bescent; flowers and terminal bud m 21(20). Leaves obovate or oblong-obovate, 20-60 more densely pubescent ............ Р. шл dun cm long, the marginal vein strongly devel- 9. Degree of pubescence on flowers and te oped from near the base to the apex ........ minal bud similar as on inflorescence axes P. insigne and twigs 21. elliptic (in P. гасетозит 10(9). Flowers subsessile, the pedicels shorter than rnm but ie than 15 cm long), » ma ihe flowers ПОЕ ТРЕЕ ginal vein lacking or only present in the 10. Flowers with pedicels as long as or longer distal half of the leaf nde flower 22(21). Leaves 30-60 x 15-24 cm, obovate, clus- 11(10). Petioles 8-14 mm long; filaments of sta- cis marginal vein lacking or weakly de- mens longer than anthers, clearly visible in eloped; flowers large, ca. 2 cm diam. old flowers; leaves 7-10 cm wide; known iganthum nly from Rio de Janeiro .......... P. amplifolium — 22. Leaves smaller and narrower, alternate or 11. Petioles 10-20 mm long; filaments as long clustered, marginal vein sometimes present; as or shorter than anthers, scarcely ex flowers less than 15 mm diam. „u ceeding glands and not readily visible in en 23(22). Tepals reflexed at anthesis; androecium sit- owers; leaves gs narrower than 8 cm uated on top of a short, pubescent column; northern South America ooo... E trianae pubescence on lower leaf surface sparse; 12(6). Lower leaf surface tomento leaves obovate, clustered _____ P. tomiwahlii lous, ү ан surface completar hidden by 23. Tepals spreading at anthesis; androecium the in ке never raised on a short column; pubescence 12. Lower leaf surface glabrous or sparsely pu- on lower leaf surface visible; leaves elliptic, qe nt, the leaf surface never completely long, or obovate, alternate or clustered 24 13(12). bon of lower leaf каран whitish, very пена Toar RISE aod окоо œ ipm shért. the i ndividual hairs scarcely visible 14 2% Leaf base and apex not both obtuse or y.y 13. dar of lower leaf aut ce brown or пена usually at least apex acute or acu- rider very short or longer and then ас 26 al fuss sable 25(24). pend 15-25 cm long, densely pilose be- (13). Inner surface of tepals glabrous; inflores- low P. pilosum cences 5-10 cm long; leaves 15-30 cm 25. Leaves 6-14 cm long, glabrescent with age, lon . panurense indument persisting longest near the base 14. €— ries of tepals а а indores- along margin and veins ............... P. racemoseum cen 5-25 cm long; leave 26(24). Inflorescences few-flowered (+ 7 flowers wa P. bifidum per inflorescence), 2-5 cm long Annals of the Missouri Botanical Garden 28(26). 28. 29(28). Inflorescences with (many) more than 7 , tepals half-erect; eaves 16-35 cm long; ovary and recep- tacle pubescent pauciflorum Flowers ca. 13 mm diam.; tepals spreading ds 10-18 cm long; ovary and recep- acle glabrous . glabritepalum Pis pubescent; ‘pubescent style easily vis- e above the androecium : costanense Pistil d style not ias en 29 Flowers 5-6 mm diam.; leaves uen P. vasquezii leaves + clus- re P. br а ii E de surface of tepals glabrous (in P. pran- cei only larger, outer three tepals; а smaller ones are papillose) or with a few small lines of papillate hairs = Inner surface of tepals uniformly pubescent or papillose-pubescent . Leaves 30-50 cm long, the base abruptly rounded or cordate; inflorescences 40-65 P. maximum Leaves 10-30 cm long, the base acute or tuse; inflorescences to 20 cm lon 32 obt . Leaves clustered at tips of branches; lower le ud 9-10 mm diam.; d markedly discolorous; inundated forests ... ?, panurense dn not clustered; upper and lower leaf urface not or scarcely discolorous ... m ee tepals papillose inside, shorter and. narrower than the outer ones, these gla- brous inside; inflorescence tomentose | prancei All tepals equal in shape and glabrous in- side; inflorescence tomentellous P. ин . Leaves io ея at tips of branches ... Leaves not clust > . Leaves at base јата cordate, the пр ong; inner . P. williamsii Leaves at base attenuate to вена пр obtuse; petioles to 17 mm inner sur face o tepals tomentellous 6 papi P. crassitepalum Tepals ear at anthesis, equal; androe- acute; lower leaf surface glabrous or with few scattered, erect hairs, never with a cover of papillae tomiwahlii . Inner surface of tepals pubescent; pistil gla- b rous Inner surface of tepals papillose; pistil pa- nt (in P. steyermarkian- ou | Marginal vein prominent on lower leaf sur- fac P. marginale Ma ina vein lacking on lower leaf surface 39 Lateral veins 15-20 on each side; flower 15-17 mm ш, inner tepals narrower than outer o E 8 mm lon rounded leaf bases, has petioles 20-30 mm long) ............................ i leaves narrowly obovate or oblanceolate; Amazonian Peru and Ecuador Р. w oe Pedicels 1.5-2 cm long; leaves obova Panama and Costa Rica . P. ey ош . Bracts of inflorescences present at ies sis; leaf tips sharply acute to Wii leaf surface dull ecl dL Hess Bracts of inflorescences ned at t anthesis: leaf tips obtuse to acute; upper leaf surface somewhat shin . Style densely gray-pubescent; receptacle pubescent inside P. cinereum Style glabrous ог brown-papillose; recep- tacle glabrous i inside . Lateral veins 7-10 on each side; glands forming a flat, pillowlike mass ______ Lateral veins 11-16 on each a dinde forming ridges surrounding stamens nobile pns SMS Pleurothyrium acuminatum van der Werff, sp. nov. TYPE: Peru. Loreto: Prov. Requena, Arboretum Jenaro Herrera, van der Werff, Vasquez & Jaramillo 9967 (holotype, MO; isotypes, AAU, AMAZ, BR, C, F, G, HBG, K, MEXU, NY, QRS, S, U, US, USM). Fig- ures 10, 11. . Ramuli teretes, solidi, dense, minute adpresse emm min adpresse pubes- x 3- anati, ion mina 9, 5 Јосећаа, locellis lateralibus et entibu ntis antheris aequantibus; glandulis incrassatis, liberis. "Pistillum ca. 1.5 mm longum, ovario ellipsoideo, ongo, dimidio basale e dimidio distale parum pubescente, stylo pubescente. Fruc- tus ignoti. Volume 80, Number 1 93 van der Revision of Pleurothyrium Werff 55 3 ne Ü Fare E Tr ART DN P. acuminatum. —A. Habit. FIGURE 10. —B. Flower showing abscission line of tepals. —C. Floral detail. — D. Young fruit, — E. Stamens— Whorl I on right, Whorl II in center, Whorl III on left.—F. Branching pattern of inflorescence. Tree, 15 m. Twigs terete, solid, densely, but minutely appressed gray pubescent, 3 mm diam. ca. 5 cm below the tip. Terminal bud somewhat shiny, densely appressed-pubescent. Leaves alter- nate, thinly chartaceous, 10-20 x 3-6 cm, ellip- tic, the base gradually narrowed in the petiole, the tip acuminate, the acumen 1-1.5 cm long, the upper surface glabrous and gland-dotted, this best visible on young leaves, the lower surface glabrous or with some minute appressed hairs, especially along the veins and near the base; lateral veins 6– 10 on each side, arching upward near the margin and weakly loop-connected in the upper half of the lamina; venation immersed on upper surface, mid- rib raised on lower surface, the lateral veins and tertiary venation less so. Petioles 1-2 cm long, adaxially flattened, somewhat swollen near the base, with a similar indument as the twigs. Inflorescences Annals of the Missouri Botanical Garden о 200 400 600 800 10001m \ у . о Prepared by Hendrik А. Ауркета x FIGURE 11. in axils of cataphylls, 8-12 cm long, slender, laxly branched, the branchlets once or twice cymosely branched, minutely tomentose, bracts deciduous at anthesis. Pedicels (4-)5(-8) mm long, minutely tomentellous. Flowers white. Tepals 6, equal, most- ly erect at anthesis, oblong, 3 ] mm, often wrinkled, twisted, but the margins not consistently inrolled, the outer 3 minutely puberulous to pap- illose outside, the inner 3 with a papillose, tri- angular base, otherwise glabrous, the inside slightly papillose to glabrous. Tepals united at the very base and falling from old flowers as a unit. Recep- tacle constricted at the tip, the constriction line turning into a dehisc Stamens 9, 4-celled, the cells lateral and opening upward, filaments about as long as anthers, gla- brous; glands enlarged, protruding between the out- er stamens, free. Pistil ca. 1.5 mm long, ovary 1 mm long, the lower half glabrous, the upper half slightly pubescent, style pubescent. Receptacle gla- brous inside. Fruit unknown. Flowers: August-Sep- tember. Elevation range: 100-200 m Distribution of P. acuminatum (ж), P. amapaense (8), P. amplifolium (О), and P. bifidum (B). Collections studied. BRAZIL. AMAZONAS: Sao Paulo de kk venca, Ducke RB25676 (RB). PERU. LoRETO: Re- quena, Arboretum of Jenaro Herrera, ? (МО), ? s.n. (MO), ‘Cols 50 (F, МО), van der Werff 9967 (AAU, AMAZ, BR, C, F, G, HBG, K, MEXU, MO, NY, QRS, S, U, US, USM). Pleurothyrium acuminatum is only known from a few collections from the Arboretum Jenaro Her- rera in Amazonian Peru and one collection from Amazonian Brazil. It is included and illustrated in Spichiger et al. (1989) as Ocotea undulata (Meiss- ner) Mez, a species that I place in Pleurothyrium and which is closely related to P. acuminatum. The erect tepals at anthesis, the relatively small (for Pleurothyrium) glands, the shape of the pistil and tepals, and the weakly loop-connected lateral veins all point to the group of species with reflexed margins of tepals as the closest relatives of P. acuminatum, even though the margins of the tepals are plane in this species. The other main group of Pleurothyrium, characterized among others by ro- tate flowers, has much broader tepals and usually Volume 80, Number 1 1993 van der Werff 57 Revision of Pleurothyrium larger, fused glands. Within the group of Pleu- rothyrium species with erect tepals, P. acumina- tum stands apart by its rather long tepals with a plane margin, long, slender pedicels, and the very short, almost papillose indument of the flowers. The acuminate leaves are also a useful character. Pleurothyrium amapaense Allen, Mem. New k Bot. Gard. 15: 92. 1966. TYPE: Brazil. Amapá: Rio Araguari, Регез et al. 51430 (holotype, NY not seen; isotypes, K, S, US). Figure 11. Tree, 15 m tall. Twigs terete, solid, the young parts ferruginous- tomentellous, branches, chartaceous, elliptic-obovate to elliptic, 10-20 x 4-6 cm, the base cuneate, the tip shortly acuminate, glabrous, lateral veins 8-10 on each side, curving upward near the margin and loop- connected in the upper half of the lamina; venation immersed on upper surface, costa and lateral veins raised on lower surface. Petioles ca. 1 cm long, narrowly canaliculate, with similar indument as twigs. Inflorescences in axils of deciduous bracts, ferruginous-tomentellous, to 12 cm long, panicu- late, the branches once or twice cymosely branched, bracts lacking at anthesis. Pedicels slender, 4—5 mm long, ferruginous-tomentellous. Flowers green- ish brown fide collectors, ca. 5 mm diam. Tepals 6, equal, minutely tomentellous outside, papillose inside, ovate, 2-2.5 mm long, often twisted, spreading or reflexed in old flowers, finally breaking off along a dehiscence line at the base, the margin plane. Stamens 9, 4-celled, the cells lateral, open- ing upward, anthers glabrous, filaments about as long as anthers; glands enlarged, free, falling off in old flowers. Pistil ca. 2 mm long, ovary ellipsoid, glabrous near base, becoming progressively more papillose toward the tip, style densely papillose, 0.5 mm long, stigma small. Floral tube deeper than wide, glabrous inside. Fruits unknown. Flowers: October. Elevation range: ca. res 51430 VENEZUELA. AMAZONAS: Rio Negro, upper Rio Baria, Davidse 27524 (MO, ү EN). Collections studied. BRAZIL. AMAPA: Rio bis nn Pi- IE, NT, 5. U5) Pleurothyrium amapaense is the only Pleu- rothyrium species reported east of Manaus, Brazil. It is only known from the type collections, but as Allen (1966a) already stated, there can be no doubt about its generic position. paense has, for the genus, rather small flowers; it Pleurothyrium ama- differs from the other small-flowered species in having tepals with plane, not reflexed, margins, in having clustered leaves, and in its ferruginous in- dument. Provisionally included here is Davidse 27524 (MO), collected from Depto. Río Negro, Territorio Federal Amazonas, Venezuela. This collection dif- fers in its thinner leaves, shorter pubescence, more elliptic leaves with a less well-developed uon vein, and its habit (label states it is a liana). T floral characteristics agree reasonably well bei those of Pleurothyrium amapaense, but it is pos- sible that, once more collections are at hand, it is better recognized as a distinct species. Pleurothyrium amplifolium (Mez) Rohwer, Mitt. Inst. Allg. Bot. Hamburg 20: 43. 1986. Nectandra amplifolia Mez, Arbeiten Konigl. Bot. Gart. Breslau 1: 131. 1892. TYPE: Brazil. Rio de Janeiro: Alto Macahe, Glaziou 17731 (holotype, B; isotypes, B, G, K, NY, P). Fig- ure 11. Tree, size unknown. Twigs terete, solid, brown- tomentellous, glabrescent, ca. iam. 5 ст below apex. Terminal bud brown-tomentellous. Leaves alternate, chartaceous, elliptic to broadly 15-20 x slightly acute, the apex rounded or shortly acute, elliptic, 7-10 cm, the base obtuse or upper surface glabrous and with venation im- mersed, the lower surface minutely puberulous when young, soon glabrescent, costa and lateral veins raised, tertiary venation immersed; lateral veins 8— 1l, arching upward near the margin, not loop- connected. Petioles 8-14 mm long, the upper side flat, with similar pubescence as twigs. Inflores- cences in axils of cataphylls, brown-tomentellous, 8-12 cm long, paniculate, the branchlets 1—2 times cymosely branched, flowers more than 30, bracts absent at anthesis. Flowers ca. 4 mm diam. Tepals 6, equal, + erect at anthesis, the margin reflexed, brown-tomentellous outside, but inner three tepals with papillose margin and tip, all tepals papillose inside. Stamens 9, 4-celled, glabrous, the outer six with one pair of cells lateral and one pair introrse, the inner three with one pair extrose and one pair lateral; filaments clearly visible between glands, са. 0.5 mm long, longer than anthers; glands enlarged, visible as triangular lobes between the outer sta- mens, but very low in front of the stamens and not fused. Pistil ellipsoid, ca. 1.5 mm long, ovary gla- brous, style papillose. Receptacle glabrous except for a few hairs at the very base. Fruit and cupule unknown. Flowers: October. Elevation range un- known. Annals of the Missouri Botanical Garden Collection Mis cahe, Glaziou 17 BRAZIL. RIO DE JANEIRO: Alto Ma- 1 (B, G, K, NY, P). Pleurothyrium amplifolium is only known from the type and is also, if the label data are correct, the only Pleurothyrium species from southern Bra- zil. It belongs in the group of species with reflexed margins of tepals, small flowers, often with erect tepals and a poorly developed (or without a) mar- ginal vein. It stands apart from the other species in this group by its distribution, broad leaves, and indumentum. Three isotypes were annotated as Persea cor- data Mez var. glabra Mez, but the holotype does not bear this annotation. Persea cordata var. gla- bra is a numen nudum included in Glaziou's list of determinations, while Persea cordata (Vell.) Mez, the correct citation, is a later homonym of Persea cordata Meissner. Pleurothyrium bifidum Nees, Syst. Laur.: 351. . TYPE: Peru. Loreto: Yurimaguas, Poep- pig 2398 (GZU, LE, W destroyed?, holotype not designated). Figure 11. Tree, to 10 m. Twigs terete, glabrous or nearly so, solid. Terminal bud brown-tomentulose. Leaves alternate, said to be clustered near tips of branches, subcoriaceous, elliptic to obovate-elliptic, 30—40 x m, the base obtuse or acute, the tip acuminate, acumen to 3 cm long, the upper surface glabrous, lower surface covered with a dense, mi- nute, gray-whitish tomentellous indument; venation immersed on upper surface, midrib and lateral veins raised on lower surface, tertiary venation not vis- ible; lateral veins 12-16 on each side, weakly loop- connected in the distal half of leaf. Petioles 3-5 cm long, glabrous. Inflorescences in axils of cata- phylls, ferruginous-tomentellous, 15-25 cm long, paniculate, the upper branchlets once or twice cy- mosely branched, the lower ones racemosely branched, but ultimate divisions cymose; bracts subpersistent at anthesis; brown-tomentellous out- side, glabrous inside. Flowers white or creamy- white, 10-12 mm diam. Tepals 6, the inner three with a slightly narrower base than the outer ones, broadly elliptic, 4-5 mm long, the outer three ferruginous-tomentellous outside, the inner three with a tomentellous, triangular basal patch, oth- erwise papillose; inside papillose. Stamens 9, 4-celled, the cells lateral or one pair lateral-ex- trorse, the valves back-to-back, glands large, com- pletely surrounding stamens, but not fully fused. Ovary ellipsoid, ca. 1 mm long, densely papillose, with 6 glabrous patches near the tip, style short, ca. 0.5 mm long, glabrous, stigma platelike. Floral tube papillose inside. Cupule a rather shallow cup, ca. 2 cm wide and 1 cm high, verrucose, the pedicel swollen; fruit ellipsoid, 2.5 x tem 1.5 ст. Fruits: Sep- r. Flowers: June-September. Elevation range: Collections studied. PERU. AMAZONAS: Rio Cenepa, Quebrada Chigka Entsa, Ancuash 602 (G, MO); Bagua, Cascadas de Mayasi, mg Quebrada Tambillo, Wurdack 2047 (F, , US). LORETO: Yurimaguas, Poeppig 2398 (GZU, ГЕ) Pleurothyrium bifidum is an uncommon spe- cies, only known from three collections in Ama- zonian Peru. The type specimens (GZU, LE) are in bud, thus I have not been able to study flowers of the type. A type photograph (A) of the type in W shows that this specimen, probably destroyed in World War II, was also largely in bud. My floral description is based on the two recent collections and agrees with earlier floral descriptions based on the type collection. Nees stated in the original description that the leaves were clustered, but the type specimens at hand have a very short piece of the twig, which does not show if the leaves were clustered or not; the W type had clustered leaves, as can be seen on the type photo. The recent collections have the leaves more or less clustered. However, the recent material differs from the types in having thicker twigs and narrower leaves; in other characters it agrees with the type. Pleurothyrium bifidum is closely related to P. panurense (Meissner) Mez, as is shown by the, for Pleurothyrium, very unusual, whitish indument on the lower leaf surface. Pleurothyrium bifidum has larger leaves and longer inflorescences and petioles than P. panurense; it also has a papillose ovary and floral tube (glabrous in P. panurense), and the flowers are slightly larger and have the inner sur- face of the tepals brown-papillose (glabrous in P. panurense). Тће size differences do not allow easy identification. For instance, Krukoff 4800 is iden- tified as P. panurense based on floral characters; its leaves can be interpreted equally well as large P. panurense or small P. bifidum. Kostermans (1952) made combinations for a number of Pleurothyrium species in Ocotea; he accepted for Pleurothyrium bifidum the na Ocotea bifidum Poeppig ex Nees. However, Oco- tea bifidum is a herbarium name cited by Nees as a synonym in the description of P. bifidum and is, therefore, not validly published. Pleurothyrium bracteatum van der Werff, sp. nov. TYPE: Colombia. Boyaca: El Humbo, ca. 130 mi. N of Bogotá, elev. 3,000 ft., Lawr- ance 619 (holotype, MO; isotypes, E, F, UC, US). Figures 12, Volume 80, Number 1 1993 van der Werff 59 Revision of Pleurothyrium acteatum. — А. Habit through flower, showing pistil and stamens embedded in glands. — D. Pis FicuRE 12. P. br surface. — H. Detail of lower leaf surfa Arbor, 20-25 m alta. Ramuli solidi, teretes vel paullo angulati, Tufo- tomentosi. Gemma terminalis К mm сгаз- sa, rufo-tomentosa. Folia alterna, subcoriacea, 15-30 x 8-12 cm, ан vel oblongo-elliptica, же rotundata, apice paullo acuta, supra glabra, subtus rufo-tomentosa, nervis lateralibus 14-20 utr g Ф = a o ч 4 e А ФЕ d et ET E O З m tione su m е lateralibus elevatis, venatione иса paullo ele- etioli rufo-tomentosi, ,10-15 mm . Inflorescentiae ex ox hou duas tomentosae 10- uplo cymae more ramosis; bracteis sub anthesi praesen- tibus, кути -tomentosis, eis ad ramu ulorum се basim m terminalium 6. sübaequalia late ovata m lon tomen- tella. Stamina 9, glabra, ve locellata, locellis не glandulis permagnis, coalitis, stamina cingentibus. Оуаг- . — B. Detail of inflorescence showing bracts and flower. — C. Cross section til. — E, F. Stamens. — C. Detail of upper leaf ium ellipsoideum, ca. 1.3 mm longum, basi glabrum, parte 6 mac Tubus floralis intus i tementellas. Fructus ignotus. Tree, 20-25 m tall. Twigs solid, terete or slightly май" reddish-brown-tomentose, 5— diam. 5 em below the tip. Terminal buds to 7 mm thick, rufous-tomentose. Leaves alternate, subcoriaceous, 15-30 x 8-12 cm, oblong to oblong-elliptic, the base rounded, the tip slightly acute, glabrous above, rufous-tomentose below, venation immersed on up- per surface, midrib and lateral veins raised on lower surface, the tertiary venation slightly raised; lateral veins 14-20 on each si near the margin and united with the superior vein, form- side, curving upwar ing a marginal vein in the upper half of the lamina. Annals of th Missouri AD Garden о 200 400 600 #00 1000km о 100 200 300 400 500 800 miles 979 b о h FiGURE 13. Distribution of P ш) Petioles rufous-tomentose, strongly canaliculate, 10-15 mm long phylls, rufous-tomentose, 10—15 cm long, panic- ulate, . Inflorescences in axils of cata- the branchlets usually twice cymosely branched; bracts present at anthesis, those at the base of the lower branchlets 12-15 mm long, ovate, rufous-tomentose, gradually reduced in size toward the ultimate divisions, the bracts at the base of the ultimate cymes ca. 5 mm long. Pedicels of open flowers ca. 8 mm long, rufous-tomentose. Flowers creamy white inside, at least 13 mm diam. (none of the flowers had all tepals spreading). Tepals 6, + equal, broadly ovate, ca. 5 mm long, the outer the inner three with a tomentose, basal triangular patch, otherwise three tomentose on the outside, tomentellous; all tepals tomentellous inside. Sta- the cells lateral and opening back-to-back, the glands strongly enlarged mens 9, glabrous, 4-celled, and forming a thick wall surrounding the stamens, fused. Ovary ellipsoid, ca. 1.3 mm long, the very base glabrous, the middle part brown-tomentellous, the upper part tomentellous with 6 bare spots where . bracteatum (X), P. brochidodromum (O), P. cinereum (O), and P. costanense glands touched the ovary; style ca. 1 mm long, glabrous. Floral tube rather deep, brown-tomen- tellous inside. Fruit unknown. Flowers: February. Elevation range: ca. 1,000 m. ollection studied. COLOMBIA. BOYACA: El Humbo, JS) С Lawrance 619 (E, Е, MO, UC, Pleurothyrium bracteatum is only known from the type collection from the El Humbo region in Boyaca, Colombia; this area also yielded Aiouea angulata Kostermans, another Lauraceae only known from the type collected by Lawra This new species can be readily identified by the large bracts of the inflorescence, which persist dur- ing anthesis. Other useful characters are the rufous pubescence (described by the collector as burnt sienna), the rounded leaf bases, and large flowers. It can be confused with Pleurothyrium costanense from similar elevation in Venezuela, but the latter species lacks bracts at anthesis, has smaller flowers, a distinctly pubescent style, and an almost glabrous ovary. Volume 80, Number 1 van der Werff 61 Revision of Pleurothyrium 'The collector noted that the stamens were in the form of a on. The wood is said to be soft and has no economic uses. Pleurothyrium brochidodromum van der rff, sp. nov. TYPE: Peru. Loreto: Maynas, km 32 of Carretera Iquitos- Nauta, Vasquez & Jaramillo 7865 (holotype, MO; isotypes, AMAZ, F, С, НВС, NY, USM). Figures 13, 14 16 m alta. Ramuli teretes, solidi, ferrugineo- tomentosi, glabrescentes. Gemma terminalis fusco-tomen- na, ad 5 mm diametro. Folia ad ramulorum glabra sed costa d p is dias € pilis e erectis prae edita (secus cost indum 18- ua nga, o eo- papillosa, extus ferrugineo-tomentella eos i interiora margine et apice papillosa). Stamina 9, 4-locellata, locellis lateralibus, minute papillosa; glandulae magnae, stamina cingentes, coalitae. Ovarium parvum, ovoideum, in stylum attenuatum, stigmate antheris obtecto. Fructus ignotus. Tree, 16 m tall. Twigs terete, solid, ferruginous- tomentose, but soon glabrescent, 3-4 mm diam. 5 cm below tip. Terminal bud large, 5 mm thick, fuscous tomentose. Leaves clustered near tips of branches, chartaceous, 12-22 X 4-7 cm, elliptic to obovate-elliptic, the base acute, the tip acumi- nate, the upper surface glabrous except for some short hairs on midrib, the lower surface with scat- tered, brown, erect hairs, these denser along main veins and midrib; venation immersed on upper sur- face, midrib, lateral veins, and to a lesser degree tertiary venation raised on lower surface; lateral veins 12-15 on each side, the lower 2-3 arching upward, the other ones + straight and ending in a well-developed marginal vein. Petioles 1.5-2 cm long, terete, with similar indument as twig, ca. 2 mm diam. Inflorescences in axils of cataphylls, 8– 14 cm long, ferruginous-tomentose, paniculate, the branchlets 1—2 times cymosely branched, ca. 40- flowered, bracts often persisting at anthesis, fer- ruginous-tomentose on both surfaces, to 4 mm long (but bracts subtending flowers smaller), elliptic. Flowers green fide collectors, Tepals 6, equal, 3-4 mm long, oblong, spreading to slightly reflexed at anthesis, the inner surface ferruginous papillose, the outer tepals ferruginous- tomentellous on the outside, the inner three with 9-10 mm diam. papillose margin and tip, otherwise tomentellous. Stamens 9, all 4-celled, finely papillose, the cells lateral; stamens ca. 1 mm long, the filaments short and nearly as wide as the anthers; glands conspic- uously enlarged, completely surrounding the sta- mens, fused. Ovary small, ovoid, the very base glabrous, situated in the glabrous, shallow recep- tacle, with a band of coarse papillae near the mid- dle, the upper part gradually narrowed into the style, finely papillose, pistil ca. 1.4 mm long; stigma in all flowers seen not exposed, but covered by the anthers. Floral tube glabrous. Fruits unknown. Flowers: August. Elevation range: са. 1: Collection studied. PERU. LORETO: Maynas, Carre- tera Iquitos- Маша, km 32, Vásquez 7865 (AMAZ, F, G, HBG, MO, NY, USM). Pleurothyrium brochidodromum, only known from the type collection made near Iquitos, Peru, is easily identified by its sparse, erect pubescence on the lower leaf surface, clustered leaves, and floral size. Other Pleurothyrium species from Am- azonian Peru with an erect. pubescence on lower leaf surface include P. insigne and P. vasquezii. Pleurothyrium insigne differs in its much larger, obovate leaves, while P. vasquezii differs in its alternate (not clustered) leaves and smaller flowers (5-6 mm vs. 9-10 mm in P. brochidodromum ). A brochidodromous venation type is relatively common in Pleurothyrium, but is particularly clearly visible in P. brochidodromum. However in such species as P. maximum and P. williamsii the brochidodromous venation is equally well de- veloped. Although the type collection has good young larged glands the stigma is large, almost platelike, and easily visible. Pleurothyrium cinereum van der Werff, sp. nov. TYPE: Ecuador. Esmeraldas: Panadero, 5 km from the river, Játiva & Epling 2039 (holotype, MO; isotypes, NY, US). Figures 13, 15 Arbor ad 32 m. Ramuli angulares, fusco-tomentelli, я solidi. Folia alterna, firme chartacea, 13- 20 m, m, elliptica, basi obtusa, apica acuta, raro obtusa, ds venatione impressa, subtus costa nervisque (10-14 utroque costae latere) elevatis, reticulatione paullo elevata, venatione in dimidio distale brochi phyllorum ortae, fusco-tomentellae, 10-15 cm longi, ra- mulis semel vel bis cymae more divisis. Pedicelli ca. 5 62 Annals of the Missouri Botanical Garden LA E 22-4 А РОИ M. A SN “=> ея - ey А: E ot FIGURE 14. P. brochidodromum. "s Habit. — B. Flower. — C. Cross section through flower. — D, E. Lateral and frontal views of outer stamen. — Е, eral and frontal views of inner stamen. — Н. Pistil. — I. Bud, showing indument on inner and outer tepals.—J : Detail of ids leaf surface. ps basi cinereo-pubescenti, omnia intus papillosa basi nero- pubescenti. Stamina 9, A eem a os Мене а bus: t ingentibus, connatis. Pistillum ad 2 mm vnd cinereo- неона mm longi. Flores rotati, n diametro. Tepala 6, subaequalia, 3 exteriora s elliptica vel setuidat ta, mm, extus cinereo-pubescentia, 3 interiora elliptica, 4 X 3 mm, secus basim angustata, extus cinereo-puberuli Volume 80, Number 1 van der Werff 63 Revision of Pleurothyrium = Flower. — D. Pistil. — E. Detail of flower, showing embedded с. "В from Gentry & Smith 45117, others from Neill 8954. FIGURE 15. P. cinereum. — А. Habit, flowering. — B. Habit, fruiting. — C. 64 Annals of the Missouri Botanical Garden ovario ellipsoideo, sensim in stylum attenuato. Recepta- culum intus cinereo-pubescens. Cupula profunda, lenti- cellata, tepalis persistentibus. Fructus ellipsoideus, ad 6 cm longus. Tree, to 32 m tall, with buttresses. Twigs an- gular, brown-tomentellous, becoming glabrous with age, solid, 4 mm diam. ca. 5 cm below apex. Terminal bud brown-tomentellous, about as wide as the twig. Leaves alternate, firmly chartaceous, 13-20 x tip acute, rarely obtuse, venation impressed on upper surface, midrib and lateral veins (10—14 on each side) raised on lower surface, tertiary venation -10 cm, elliptic, the base obtuse, the slightly raised, lateral veins weakly loop-connected in the distal half of lamina, upper surface glabrous, the lower surface with varying amounts of minute, 2.5 cm long, with similar indument as twigs. Inflorescences appressed hairs. Petioles canaliculate, 1.5- in axils of cataphylls, brown-tomentellous, 10-15 cm long, the branchlets once or twice cymosely branched, bracts deciduous at anthesis. Pedicels ca. 5 mm long. Flowers rotate, 11 mm diam. de- scribed as creamy-white, yellowish-ferruginous or brick-red. Tepals 6, subequal, the outer three bescent outside, the inner three elliptic, 4 х 3 mm, narrowed toward the base, gray-puberulent outside except for a basal, triangular, gray-pubes- cent patch; all tepals papillose inside except for the gray-pubescent base. Stamens 9, all 4-celled, the cells dorsal-lateral and opening back-to-back; glands strongly enlarged, almost completely surrounding the outer stamens, fused. Pistil ca. 2 mm long, gray-pubescent, ovary ellipsoid and gradually nar- rowed into the style, the gray-pubescent style in dried flowers visible above the androecium. Floral tube gray-pubescent inside. Cupule deeply cup- shaped, lenticellate, 2 x 2.5 cm, tepals persisting on the margin of the cupule. Fruit an ellipsoid berry, to 6 cm long. Fruits: February- May. Flowers: May, July. Elevation range: sea level to 1,600 m Med vius ECUADOR. ESMERALDAS: Rio On- 4 (NY) Panadero f San Lorenzo, lo NY, US); Рики во; 5 km from river, Jativa gor (MO, NY, 05); Panadero. Little 21359 ( ME). маро: Archidona maco, Neill 8954 (MO). PERU: LORETO: | тошћ of Rio pue in Rio Putumayo, Klug 2122 (A, F, 6, GH, K, MO 5, 05); Pongo de Manseriche, шр. of Rio Santiago in Maran non, Tessmann 4634 (G). SAN MARTIN: A. Gentry & D. N. Smith 45117 (MO). Брин ти cinereum is known from several rather widely separated localities. The type collec- tion is from the coast near Esmeraldas, Ecuador. Little et al. 21359 was collected the same day at the same locality and likely came from the same tree. Janse 280, a fruiting collection, is also from Esmeraldas. Neill et al. 8954 was collected on the slopes of volcano Sumaco 1,150 m elevation, Napo province, Ecuador. This collection, with flowers and a fruit, has slightly smaller and more pubescent leaves than the Esmeraldas specimens and has the pleasant odor commonly encountered in Aniba. The third locality is along the Rio Putumayo, Dept. Loreto, Peru, at 180 m elevation. The specimens from Esmeraldas have creamy-white flowers, from volcano Sumaco yellowish-ferruginous, and from Peru brick-red flowers. The collections all possess a slightly shiny upper leaf surface, a pubescent style, very slight difference in pubescence on outer surface of tepals between inner and outer tepals (much less than common in the genus), and fre- quently an obtuse leaf base. Visibly pubescent styles are very uncommon in Pleurothyrium and are an important reason for placing these collections in one species. Pleurothyrium cinereum is included in Little & Dixon (1969) as Pleurothyrium sp. Vegetatively P. cinereum strongly resembles P. steyermarkianum. The latter species has quite dif- ferent flowers (with glands forming ridges, not a central dome, and a glabrous style), and its pedicels are strongly triangular in cross section. e common name used by the Huitot Indians, cited by Klug, is Naime-ey. The epithet cinereum refers to the gray-pubes- cent style, one of the most distinctive features of this species. Provisionally placed here is a fruiting collection from the San Martin Province, 20 from Rioja, elevation 1,000 m (Gentry & Smith 45117, MO). This collection has the denser indument also found in the Neill collection from Ecuador. In leaf characters it is a good match for P. cinereum. The fruit is large (4 x 2-2.5 cm) and the cupule has a double margin. The outer margin is formed by the persistent tepals and the inner margin is erect, ca. £ cm taller than the outer one. Such a pro- nounced double margin has not been seen in other collections of Pleurothyrium. Remnants of sta- mens found on the cupule clearly point to Pleu- rothyrium; they have four lateral anther cells. The Neill collection includes one weakly double-mar- gined cupule; cupules from Esmeraldas are less mature and do not show a double margin. If the Gentry collection is not an aberration, P. cinereum is the only Pleurothyrium species with а double- rimmed cupule. Pleurothyrium costanense van der Werff, Ann Missouri Bot. Gard. 74: 408. 1987. TYPE Volume 80, Number 1 1993 van der Werff 65 Revision of Pleurothyrium FIGURE 16. Venzuela. l'alcón: Sierra de San Luis, above La Chapa, van der Werff 3654 (holotype, U; isotype, CORO). Figures 13, 16 Tree, to 20 m tall. Twigs somewhat angular, ferruginous tomentose, but glabrescent with age, ca. 6 mm diam. 5 cm below apex, solid. Terminal bud ferruginous tomentose. Leaves alternate, char- taceous, 20-35 8-17 cm, elliptic to broadly elliptic, the base obtuse or acute, the apex rounded P. costanense. — А. Habit. — B. Flower. — C. Cupule. to shortly acuminate; upper surface with variable amounts of tomentose pubescence on midrib and lateral veins, othewise glabrous; lower surface with + dense cover of erect, ferruginous hairs, this becoming tomentose on midrib and lateral veins. Venation immersed on lower surface, but midrib, lateral veins, and tertiary venation raised on lower surface; lateral veins 10-17 on each side, arching near the margin toward the tip and becoming loop- connected in the distal half of the lamina. Petioles Annals of the Missouri Botanical Garden 1.7-3 cm long, ferruginous tomentose. Inflores- cences in the axils of foliage leaves or cataphylls, ferruginous tomentose, 10-20 cm long, the branchlets 2-3 times cymosely branched. Flowers white or buff, 6-9 mm diam. Tepals 6, equal, spreading at anthesis, the outer 3 ferruginous pu- bescent outside, inner 3 puberulous outside except for a basal, triangular pubescent patch; all tepals papillose inside, elliptic, to 4 mm long. Stamens 9, all 4-celled, the outer 6 with the lower pair lateral- extrorse, the upper pair lateral-introrse, the inner 3 with extrorse or extrorse-lateral (upper pair) cells; filaments short, glabrous. Glands conspicuous, part- ly enclosing the outer stamens, but not fused. Ova- ry ellipsoid, with 6 longitudinal lines of hairs, oth- erwise glabrous, gradually narrowed in the conspicuously pubescent style; pistil ca. 2.8 mm long. Receptacle glabrous inside. Cupule large, warty. Fruits: March-August. Flowers: July-Sep- tember. Elevation range: 700-2,200 m Collections studied. | VENEZUELA. Bernardi 6879 (С, MER, NY). ANZOATEGUI: Cerro Peonia, of Bergantin, Steyermark 6 1393 (F). DISTRITO FEDERAL: mountain slope uerto Cruz, Whitfor (A). FALCON: Sierra U). MERIDA: Andres Bello, La о 1123 (MER, ); Andres Bello, San Eusebio, Quintero 11 (MER). MIRANDA: Paez, Fila La Tigra, m SO de upira, González Опера 1215 (С, MO). MONAGAS: ~ Pueblo de Yucucual, Lau 10 (MO); Montana de Aguacate, Steyermark 62230 (F); Caripe, а Bosque J-03 (MO), Bosque Ј-ОЗА (MO), Bosqu у О). SUCRE: Cerro Espejo, peninsula de Paria. Bona & Rabe 96072 (G, K, NY, US). Pleurothyrium costanense is known from the Cordillera de La Costa, the Sierra de San Luis in Edo. Falcón, and the Venezuelan Andes, where it occurs between 700 and 2,200 m elevation. Most collections are from about 1,000 m. Only one other Pleurothyrium species, P. steyermarkianum Al- len, is known from the Venezuelan Andes, but this species lacks the erect-tomentose pubescence on the lower leaf surface and the young twigs. Pleu- rothyrium trianae is not uncommon at the base of the Andes and Cordillera de La Costa, but this species differs greatly from P. costanense in the reflexed margins of the tepals and the dense, but short tomentellous indument on the lower leaf sur- face, which completely covers the leaf surface. There is a good deal of variation in the color of the indument in P. costanense. Especially the type and Steyermark 96072 have a light color of in- dument, whereas some of the other collections are dark ferruginous. In other characters, such as leaf shape, type of indument, raised tertiary venation, and the short, densely pubescent style, these col- lections agree with each other, and consequently all are placed in the same species. Pleurothyrium crassitepalum van der Werff, Sp. nov. TYPE: Puerto Merizalde, elev. 5-20 m, Cuatrecasas 13986 (holotype, F; isotype, US). Figures 17, 8 olombia. Valle: Rio Naya near Arbor magna. Ramuli teretes vel subangulares, folio- rum cicatribus magnis praediti, hornoti ni dense, minute, ње fusco-pubescentes, annoti b pneum entes. Gem ma bein dense, Е ои o-pubescens. Fo lia ad apices ramulorum congesta, chartacea, ier ena vel п -obovata, 10- 57 x 6- i m, basi attenuata vel cuneata, apice rotundata vel obtusa, supra more divisis. Pedicelli 5-11 m Flores extus rufescentes, intus albidi fide coll., ca diametro. Tepala 6, манира late ovata, exteriora ca. 5 mm += interiora ca. m longa, dense fus tomentella in sicco, crassa "Stamina 9, 4-locellata, locellis lateralibus, enter: papillosis, e magnis connatis stamina cingentibus. Ovarium um, ellipsoideu mm longum, style glabro, ca. L5 mm longo, dawns non superante. Tubus floralis glaber. Fructus ignotu Large tree, Twigs solid, terete or somewhat an- gular, with large scars of old leaves, the tips dense- ly, but minutely, brown appressed pubescent, gla- brescent on older parts, 5-6 mm diam. 5 cm below apex. Terminal bud with a dense, appressed tan pubescence. Leaves clustered near the tips o branches, firmly chartaceous, elliptic to elliptic- obovate, 10-17 x 6-9 cm, the base attenuate ог cuneate, the tip rounded or obtuse, the upper sur- face glabrous, the lower surface covered with a short, minute, almost papillose indument; lateral veins 13-16 on each side, arching upwards near the margin and connected with the superior vein in the upper half of the lamina, the marginal vein well-developed; venation immersed on the upper surface, but midrib, lateral veins and tertiary ve- nation raised on lower surface. Petioles 11-17 mm long, flattened on the upper side, ca. 3 mm in i Inflo- rescences in axils of cataphylls near the tips of 7-14 cm long, mostly racemose, but occasionally with branchlets once cymosely branched, with 10-20 flowers, bracts not present at anthesis. Pedicels 5- diameter, with a similar indument as twigs. branches, densely brown-tomentulose, 11 mm long, densely brown tomentulose. Flowers dark reddish brown outside, white inside fide col- lector, ca. 13 mm diam. Tepals 6, subequal, all Volume 80, Number 1 1993 van der Werff 67 Revision of Pleurothyrium FIGURE 17. stamen P. crassitepalum broadly ovate, the outer 3 ca. 5 mm long, the inner 3 ca. 4 mm long, all on both surfaces densely brown tomentellous, thick. Stamens 9, all 4-celled, the cells lateral and opening back-to-back, the an- thers papillose; glands strongly enlarged, forming a central dome, enclosing the outer anthers and fused. Ovary glabrous, ellipsoid, ca. 1 mm long, style glabrous, ca. 1.5 mm long, not surpassing stamens; stigma not or hardly enlarged. Floral tube bit. — B, C. Flowers. — D. Pist A. Ha . —G, H. Lateral and oaks view of inner stamen. — I. Detail of une leaf surface. —J. Base D —E, К. Lateral and frontal view of outer of leaf. glabrous. Fruit unknown. Flowers: February. El- evation range: 5-20 m. Collection studied. | COLOMBIA: VALLE: Puerto Meri- zalde, Rio Naya, Costa del Pacifico, Cuatrecasas 13986 US). tm Pleurothyrium crassitepalum is one of several Pleurothyrium species known only from the type. Distinctive characters are the thick tepals, which Annals of the Missouri Botanical Garden о 200 400 600 800 1000km ¡A e =_==—— о 100 200 300 400 500 600 т Р RR FicunE 18. ) appear almost double-margined, the very short and dense pubescence on the leaves, which is unusual in the group of species with large flowers and rel- atively few-flowered inflorescences, and the raised tertiary venation on the lower leaf surfaces. The clustered leaves with rounded tips are also useful characteristics. Common names for this species are nadde, nande, and palo blanco. The wood is used for making boards. Pleurothyrium cuneifolium Nees, Syst. Laur.: 352. 1836. Ocotea triangulata Kostermans, Ј. Sci. Res. (Jakarta) 1: 2. . ТҮРЕ: Brazil. Amazonas: Tefé (Ega), eens 2908 E not designated; С— 2 sh LE). Figures 19, 20. eets, ie ее chrysophyllum Nees, Syst. Laur.: 351. TYPES: Peru. Cuchero, Poeppig 17 18 (В, as no. 1719, GZU, LE, US); Peru. Chachapoyas, Yambrasbamba, Matthews 1432 (E, K). Pleurothyrium macranthum Nees, Linnaea 21: 512. Distribution of P. crassitepalum (A), P. giganthum (8), P. glabrifolium (O), and P. golfodulcense 848. SYNTYPES: Peru. Poeppig 1790 (not seen); Poeppig 2125 (G, LE). Tree, to 20-25 m. Twigs angular or ridged when young, becoming + terete with age, often fistulose, often lenticellate, the young parts minutely to- mentellous, glabrescent with age; diameter of young twigs 4-8 mm, depending on how strongly fistulose the twigs are. Terminal bud densely brown tomen- tellous. Leaves alternate, chartaceous, (narrowly) elliptic to (narrowly) elliptic-obovate, 15-30 x 7- cm, the base cuneate, acute or abruptly round- ed, the tip obtuse or acute, lateral veins 12-18 on each side, arching upward near the margin and weakly loop-connected in the distal half of the lamina, immersed on the upper surface, midribs and lateral veins raised on lower surface, upper surface glabrous or nearly so, lower surface gla- brous, papillose or tomentellous, the indument wearing off with age. Petioles 1.3-2.0 cm long, canaliculate, with similar indument as twigs. Inflo- rescences in axils of cataphylls, or infrequently in Volume 80, Number 1 van der Werff 69 1993 Revision of Pleurothyrium FIGURE 19. P. cuneifolium. — А. Habit. — B. Detail of normal inflorescence. — C. Stunted inflorescence with main axis inhabited by ants. — D. Androecium and glands. — E. Pistil. — Е. Young fruits. —G. Hollow twig, showing holes used by ants. 70 Annals of the Missouri Botanical Garden о 200 400 всо 800 1000Km AAA AA So a] Ə 100 200 JCO 400 $00 800 mies р, Hendrik А А FIGURE 20. axils of normal leaves, brown-tomentellous, 10-27 cm long, the lateral branchlets to four times cy- mosely branched but sometimes (when inflores- cence fistulose) reduced to short spurs with clus- tered flowers, many-flowered, bracts often persistent during anthesis, ovate, to ca. 5 mm long, tomen- tellous on both surfaces. Pedicels of variable length, the lateral flowers of a cyme often nearly sessile, but flowers on fistulose inflorescences with pedicels of up to 7 mm long, with same indument as inflo- rescence. Flowers white, yellow to yellowish red, from 7 to 14 mm diam. Tepals 6, equal, + erect, elliptic or elliptic-ovate, 3-7 mm long, the outer 3 (minutely) tomentellous outside, inner 3 with a basal, triangular, (minutely) tomentellous patch, otherwise glabrous, all tepals glabrous inside (rarely papillose), the margins plane or inrolled, not re- exed. Stamens 9, 4-celled, the lower 2 cells lat- eral-extrorse, the upper 2 lateral-introrse; glands strongly enlarged, surrounding the base of the outer stamens, fused. Pistil 2-3 mm long, the ovary about as long as the style or slightly shorter, some- times well defined, sometimes gradually narrowed Distribution of P. cuneifolium (Ж) and P. glabritepalum (8). into the style, the lower 14-24 glabrous, the re- mainder (and the style) papillose, the stigma often platelike. Floral tube glabrous inside, rarely pu- bescent. Cupule cup-shaped, (sometimes?) with 6 thick lobes, warty, ca. 2 cm wide, 1 cm high. Fruit ovoid, ca. 2 cm long. Fruits: February-March. Flowers: nearly year-round. Elevation range: 100— Collections studied. BRAZIL. AMAZONAS: ad Ега (= cia, Shepherd 537 (MO); Mpio. de San Carlos, near ISA hydro-electric dam, McPherson 13365 (MO); Mpio. de San Luis, sector Río Samana- Rio Claro, vereda Tulipan, Cogollo 170 (MO). Boyaca: El Humbo, 130 mi. N of Bogotá, Lawrance 780 (А, F, С, UC, US). HUILA: Little 8488 (A). r Werff 11157 (MO). Naro: 8 km Rio abajo de Misahualli, Neill 7019 (F, K, MO), Neill 7128 (МО). Peru: CUCHERO: Poeppig 1718 (US); Pe- ruvia subandina, Poeppig 1845 (C), Poeppig 2125 (C). AMAZONAS: Quebrada de Bashuchunuk, Kayap 145 (F, C, MO); Quebrada Kayamas, Río Cenepa, Kayap 648 (МО). cHACHAPOYAS: Yambrasbamba, Matthews 1432 (Е, K). cuzco: Hda. Santa Rosa (Convención), Soukop 910 Volume 80, Number 1 1993 van der Werff TA Revision of Pleurothyrium (F). JUNIN: Chanchamayo, Mina Pichita, above San Ra- mon, van der Werff 8673 (MO); Oxapampa, Chontabam- ba, Smith 306 1 (MO). LORETO: а Ауаіа 574 (MO); Fortaleza, nl Yurimaguas, Klug 2779 (A, F, C, GH, K, MO, S, US); б ырыл». Klug 2931 (A, F, С, GH, К, S); Cachipuerto, Río Cachiyacu, Klug 3116 (A, F, G, GH, K, MO, S, US); Pumayacu, between Balsa- puerto and Moyabamba, Klug 3195 (A, F, G, GH, K, MO, NY, S), Klug 3228 (A, F, С, GH, K, МО, NY, 5, US); La Victoria on the Amazon River, Williams 2715 (F). MADRE DE DIOS: Tambopata, Gentry 45760 (MO); Tambopata Wakwe pe Gentry 57817 (MO); Tam- bopata, Cuzco Amazonico Lodge, Nuñez 12207 (MO). Pasco: Palcazu Valley, M 2631 (MO); Palcazu Valley, Buenos Aires, Hartshorn 2937 (MO); Oxapampa, Huancabamba, Reynel 102 (MO); Oxapampa, Chonta- bamba valley, 2-8 km W of Oxapampa, Gentry 39902 (MO); Oxapampa, Distr. Iscozacin, Palcazu valley, Pario- na 63 (MO); Oxapampa, Chontabamba, 2 de Mayo, Smith 1743 (G, MO), Smith 1756 (G, MO); Oxapampa, Villa Rica, van der Werff 8266 (MO), van der Werff 8287 MO); Oxapampa, Villa Rica, Santa Polonia Alta, van der Werff 8346 (MO). SAN MARTIN: Zepalacio, near Moya- bamba, Klug 3567 (F, С, GH, МО, NY, S, US). UCAYALI: Distrito Inahuaya, Comunidad de Santa Rosa, Campos 364 (MO); 65 km NE of Pucallpa, Gentry 58574 (MO). As accepted here, P. cuneifolium is а wide- ranging species, known from Colombia, Ecuador, Peru, and Brazil, and occurs in the Amazonian lowlands and up to 1,800 m in the Peruvian Andes. Given its wide distribution and presence in different habitats, it is not surprising to find that it is a variable species. The two species placed in syn- onymy under P. cuneifolium, P. chrysophyllum and P. macranthum, were not published at the same time by Nees and he did not discuss the characters that separate the three species. Meiss- ner (1864) regarded P. chrysophyllum as scarcely more than a variety of P. cuneifolium, differing in a denser indument, slightly smaller flowers, shorter pedicels, more branched inflorescences, and flowers with a linear bract near their base. Pleu- rothyrium macranthum differed, i 1864), in its larger flowers. Mez (1889) placed P. macranthum under P. cuneifolium, but kept P. chrysophyllum apart; he separated the two on the ~ relative length of the style and ovary: the style longer than the ovary in P. cuneifolium and shorter in P. chrysophyllum. 1 found this not a reliable character: in most flowers I dissected the style was about as long as the ovary, and it seemed that often in old flowers the apical part of the style and the stigma had fallen off, which makes it unsuitable as a character to separate two taxa. MacBride (1938) did not mention P. macranthum and sep- arated P. cuneifolium and P. chrysophyllum by the density of their indument on the leaves. The denser indument occurs mostly on specimens from higher elevations, such as the syntypes of P. chrys- ophyllum and several collections from the Oxa- 800 m), but Ayala 574 (Quistacochas, near Iquitos, 140 m) is equally to- pampa area А mentellous. Because minutely tomentellous speci- mens have been collected near Oxapampa and in the lowlands, and no other characters are restricted to the collections with denser indument, I reduce P. chrysophyllum to synonymy. The larger num- ber of collections now available show that there are no discontinuities in flower size, and thus P. mac- ranthum cannot be recognized as a distinct taxon. The largest flowers, with tepals to 7 mm long, are found on three Klug collections (3195, 3228, and 3567), which have strongly fistulose branches and fistulose inflorescences. The branchlets of the in- florescences have, in these collections, been re- duced to short spurs bearing a cluster of flowers. This inflorescence shape is quite similar to that of the type of P. poeppigii and, indeed, the Klug collections had in some herbaria been annotated as P. poeppigii. However, the flowers of P. cu- neifolium are clearly larger, lack reflexed margins of their tepals, and are glabrous on the inner sur- face of the tepals. Included in P. cuneifolium are a few collections with entirely glabrous leaves, for instance, Law- rance 780. Neill 7019 provides the link between the collections with glabrous leaves and tomentel- lous leaves; it has young, tomentellous and older, glabrous leaves Unfortunately, fruiting material of this species it not well known. A common name in the Oxa- pampa region is roble zapallo, because the cupule has six thick ribs, like the ribs on a squash. This is both unusual for Lauraceae and easy to see on fresh material, but disappears largely on dried spec- imens. It seems that this cupule type also occurs in lowland populations (Palcazu valley). Flowering material of the roble zapallo fits very well in Р. cuneifolium. Infrequently found in Lauraceae are the strik- ingly lenticellate twigs of many collections of P. cuneifolium; other useful characters for P. cunei- folium are the fistulose twigs, tomentellous inflo- rescences, (usually) glabrous inner face of tepals, and the more or less erect tepals (especially the inner ones) at anthesis. Although P. cuneifolium lacks reflexed margins of the tepals, I do not think its closest relatives are among the other species with plane tepals. Leaf shape, venation, the more or less erect tepals, and indument of the pistil all suggest a relationship with the species with reflexed margins of tepals, such as P. trianae and P. poep- pigii. Probably also closely related is P. obovatum, 72 Annals of the Missouri Botanical Garden a species with fistulose twigs (and sometimes fis- tulose inflorescences), more or less erect tepals at anthesis, and some lenticels on the small piece of twig 1 have seen. A few specimens of Pleurothyrium cuneifolium lacked a fistulose stem and would key to P. trianae, while very few collections of P. trianae had a fistulose stem and would key to P. cuneifolium. Apart from the characters used in the key, these two species can be separated as follows: P. cunei- folium has longer tepals (3-7 mm vs. ca. 2 mm), larger and more persistent bracts protecting the flower buds, and has the inner surface of the tepals usually glabrous (papillose in P. trianae). Pleurothyrium giganthum van der Werff, s nov. TYPE: Ecuador. Los Rios: Rio Palenque Science Center, km 56 Quevedo-Sto. Domin- go, Dodson 18217 (holotype, QCNE). Figures 18, 21. ipod ad 30 ea nue Ramuli anguiares, ош etd tomentosi, cicatri i ec клейи. Cl 1-2.5 ст longa. [um terminalis fusco-tomentosa, ad 7 mm longa. Folia ad api- ces ramulorum aggregata, chartacea, 30-60 x (10-)15- obovata vel obovato- elliptica, sensim versus basim vel rotundata; nervis lateralibus 15- 20 u utroque costae latere, s supra im mersis, subtus costa minus elevata; venatione in dimidio distale brochidodrom ma. Pet- s imma )u mmatura hexangularis, jenticellats, sen- sim in dat attenuata. Tree, to 30 m tall. Twigs roundly angular, solid, brown-tomentose, diameter 7-9 mm ca. 5 cm low the tip with conspicuous scars of fallen leaves, often with cataphylls at the base of the most recent growth, these cataphylls appressed and 1-2.5 cm long. Terminal bud brown-tomentose, ca. 7 mm long, narrower than the twig. Leaves clustered near the tips of the branches, chartaceous, 30-60 x (10-)15-24 cm, obovate or obovate-elliptic, grad- ually narrowed toward the base, the base abruptly rounded, the tip shortly acute to rounded, lateral veins 15-20 on each side, immersed on the upper surface, the midrib strongly raised on the lower surface, lateral veins and tertiary venation also raised on lower surface; lateral veins arching up- ward near the margin and becoming loop-con- nected in the distal half; upper surface drying dark green, with brown-tomentose pubescence along the midrib, less so along the lateral veins, the lamina with scattered, erect hairs, which wear off with age; lower surface with brown-tomentose midrib and lateral veins, the lamina with erect hairs, which leave the surface readily visible. Petioles brown- tomentose, 5-10 mm long and 4-6 mm thick. Inflorescences racemose, in the axils of cataphylls below the leaves, ca. 6 cm long, much shorter than the leaves, brown-tomentose, the flowers subtended by brown-tomentose bracts ca. 5 mm long. Pedicels ca. 1.2 cm long, brown-tomentose, gradually wid- ened toward the buds, 3-angled, the ridges con- tinuing as a low keel on the 3 outer tepals. Flowers ca. 2 cm diam., light green with whitish margins. Tepals 6, subequal, the outer 3 ca. 10 x the inner 3 ca. 7 X 9 mm, 7 mm, more constricted at their base than the outer 3, broadly ovate to broadly elliptic, pubescent on both surfaces. Stamens 9, the valves back to back, glands strongly enlarged, completely enclosing the stamens and fused. Ovary (in young fruit) subglobose, glabrous; floral tube glabrous inside. Young fruits fully enclosed in cu- pule, this lenticellate, gradually narrowed into ped- icel, ca. cm long, and strongly 6-angled. Fruits: May. Flowers: January, September. Collections studied. ECUADOR. LOS RIOS: Rio Palen- que Science Center, Dodson 10181 (МО), van der Werff 12362 (MO) Rio Palenque Science Center, Dodson 13632, 18217 (both QCNE), Dodson 6302, van der Werff 9471 (both MO). Pleurothyrium giganthum is only known from one or two trees at the Rio Palenque Science Cen- ter. Characteristic for this species are the large, clustered leaves, the conspicuous cataphylls, and the erect pubescence on the lower leaf surface. This species was first collected flowering in January 1983, but the collected flowers were later lost. Immature fruits were collected in May 1987, also indicating January- February as the flowering sea- son. Since 1987, C. Dodson regularly checked the only known tree for flowers and finally, in Septem- ber 1990, the tree became deciduous and flowered briefly when the new foliage developed. The specific epithet is based on the Greek words gigas (giant) and anthos (flower) and refers to the very large flowers of this species. This species was included in the Flora of Río Palenque (Dodson & Gentry, 1978) as Ocotea sp. nov. Pleurothyrium glabrifolium van der Werff, sp. nov. TYPE: Ecuador. Napo: Rio Payamina, 2-10 km 5 of the road Coca-Loreto, Cerón Volume 80, Number 1 993 van der Werff 73 Revision of Pleurothyrium FIGURE 21. & Palacios 3017 (holotype, MO; isotypes, AAU, C, COL, C, HBG, K, LE, NY, QAME QRS, U, US). Figures 18, 22. Arbor, ad 15 m. Катић teretes, solidi, ad apicem brunneo- See glabrescentes. Ge e brunneo-tomentosa. Folia alterna, aonack то єк о . . : ) ае equalia, ovata, m lo b 3 riora extus pisse intus glabra, | interiora extus basi tomentella, demum glabra, intus glabra. Stamina 9, 4-locellata, locellis шеге йи, filamentis quam antheris brevioribus; glandulis magnis, connatis, stamina cingen- P. giganthum. — А. Habit. — B. Inflorescence. tibus. Pistillum glabrum, 1 mm longum, stylo breve, stig- mate peltata. Receptaculum glabrum. Fructus ignoti. Tree, to 15 m tall. Twigs terete, solid, the very tip and terminal bud brown-tomentose, older parts becoming glabrous, ca. 4 mm diam. 5 cm below apex. Leaves alternate, subcoriaceous, elliptic, 1 4— 5 x cm, the base acute, tip acuminate, glabrous on both surfaces, the upper surface gland- dotted, venation immersed on upper surface, costa and lateral veins raised on lower surface, the ter- tiary venation less so; lateral veins 12-16 on each side, curving upward near the margin and becom- ing loop-connected in the distal half of the lamina. Petioles 1.5-3 cm long, canaliculate, with similar tomentum as the twig. Inflorescences in axils of cataphylls, 7—16 cm long, brown-tomentellous, pa- 74 Annals of the Missouri Botanical Garden FIGURE 22. P. glabrifolium. — A. Habit. — B. Flower. — C. Cross section of flower. — D. Detail of inflorescence. — . stamen. — Е. Pistil. Volume 80, Number 1 1993 van der Werff 75 Revision of Pleurothyrium niculate, many-flowered, the branchlets 1—2 times cymosely branched, bracts mostly deciduous at anthesis. Pedicels 2-3 mm long, brown-tomentel- lous. Flowers 6-7 mm diam., whitish. Tepals 6, equal, ovate, ca. 3 mm long, tomentellous outside, but inner 3 with glabrous margin and tip; glabrous inside. Stamens 9, glabrous, 4-celled, the cells lat- eral; filaments shorter than the anthers, ca. 0.2 mm long; glands enlarged, fused, completely en- ies the stamens. Pistil glabrous or nearly so, . 1 mm long, the style very short, stigma plate- like. Floral tube shallow, glabrous; ovary largely covered by the enlarged glands. Fruits unknown. Flowers: December. Elevation range: 250-300 m. dide суе ECUADOR. NAPO: Rio ayomini; km RS, U, 05) El Chuncho, 6 km from Rio Payamino, Palacios 2288 (MO). Pleurothyrium glabrifolium is only known from two collections, made in the same area in the low- lands of eastern Ecuador. The combination of gla- brous inner surface of tepals, glabrous and alternate leaves is characteristic for this species. The only other Amazonian species with the inner surface of the tepals glabrous is P. panurense, but that species has clustered leaves with a minute, whitish, and dense indument on the lower leaf surface. The мам нулы glabrous stamens and pistil of P. gla- brifolium are an uncommon character in Pleu- rothyrium. Pleurothyrium glabritepalum van der Werff, sp. nov. TYPE: Colombia. Choco: hills behind Bahia Solano (Puerto Mutis), Gentry & Forero 7201 (holotype, MO; isotypes, HBC, U). Fig- ures 20, 23 r, ad 8 m alta. Катић solidi, teretes, porphyreo- Miis glabrescentes. Gemma terminalis parva, por- phyreo-tomentosa. ps pales випас be its ea Pd 3.5-6 cm, basi acuta, acumina cu- ne 1-2 cm lo bue. supra zn duy idea et pili di is- persis prac sed secus | costam nervosque laterales i in- indumento paginam non tegente, sed secus costam nervos- 4 cm lon ae, 1-3- dem abra, intus glabra. Stamina 9, glabra, 4- locellata, locellis a glandulis incrassatis stamina cingenti- 2E coalitis, ovarium depresse globosum, glabrum, ca. 1 m diametro, tubus floralis glaber. Fructus ignotus. Tree, at least 8 m tall fide collectors. Twigs solid, terete, brown-tomentose when young, gla- brescent, 2-3 mm diam. ca. 5 cm below the tip. Terminal bud small, brown-tomentose. Leaves al- ternate, thinly chartaceous, elliptic, 10-18 x 3.5- 6 cm, the base acute, the tip acuminate, the acu- men 1-2 cm long, the upper surface with gland dots and scattered brown hairs, these much denser along midrib and lateral veins, the lower surface with a rather sparse, erect pubescence, the hairs somewhat wrinkled and not covering the leaf sur- face; indument brown-tomentose along midrib and lateral veins; lateral veins 5-8 on each side, ve- nation immersed on upper surface, midrib and lat- eral veins raised on lower surface, the tertiary venation slightly so, lateral veins arching upward near the margin and weakly loop-connected in the distal half of the lamina. Petioles 10—15 mm long, terete, with similar pubescence as twigs. Inflores- cences in axils of cataphylls, 1-3-flowered, 2-4 cm long, brown-tomentose, bracts lacking at an- thesis. Pedicels 4-5 mm long, tomentose. Flowers green with tan center (fide collectors), ca. 13 mm diam. Tepals 6, subequal, the outer three 5.5 mm long, the inner three 4.5 mm long, the outer three tomentulose outside, inner three with a basal, tri- angular tomentulose patch, otherwise glabrous; all tepals with a gl 4-celled, glabrous, the cells lateral and opening brous inner surface. Stamens 9, back-to-back, glands strongly enlarged, surround- ing the stamens, fused; ovary depressed globose, glabrous, ca. 1 mm iam., receptacle glabrous in- side. Fruit unknown. Flowers: January. Elevation range: below 250 m Collection studied. COLOMBIA. CHOCO: hills behind Bahia Solano (Puerto Mutis), Gentry & Forero 7201 HBG, MO, U) ~ Pleurothyrium glabritepalum, named after the glabrous inner surface of the tepals, is only known from the type collection. Useful characters for its identification are the following: the more or less erect pubescence, which is also present on the upper leaf surface, the gland dots on the upper leaf surface, the alternate, acuminate thin leaves, the few-flowered inflorescences, and the glabrous inner surface of the tepals. The leaves have, for the genus, few lateral veins and a marginal vein is weakly developed. Pleurothyrium golfodulcense Burger & Za- mora, Fieldiana Botany, New Series, 23: 115. Annals of the Missouri Botanical Garden ML А; { 1374 FIGURE 23. 1990. TYPE: Costa Rica. Puntarenas: Alto de las Mogas, Sanchez, Zamora & Brenes 1228 (holotype, CR not seen; isotype, F not seen). Figures 18, 24. Tree, to 15 m tall. Twigs terete, solid, gray pubescent, 2-3 mm diam. ca. cm below tip. Terminal bud densely gray-pubescent. Leaves al- ternate, thinly chartaceous to coriaceous, 10-2 4-7 cm, elliptic to elliptic-obovate, base acute or obtuse, apex acuminate, acumen to 2 cm long, upper surface glabrous, except for the + pubescent midrib, gland-dotted, lower surface glabrous or with minute appressed hairs, but midrib and to a lesser degree lateral veins pubescent; lateral veins 6-9; loop-connected in distal half of lamina, venation P. glabritepalum. — А. Habit. — B. Flower. — C. Leaf base, showing indument. — D. Leaf apex. immersed on upper surface, midrib and lateral veins raised on lower surface. Petioles 6-22 mm long. Inflorescences in axils of cataphylls or normal leaves, to 10 cm long, racemose, ferruginous pubescent, the lateral branches often subtended by an elliptic bract, to 6 mm long, pubescent on both surfaces; pedicels to 14 mm long, often with a bracteole halfway between inflorescence axis and flower. Flowers white, rotate, 9-12 mm diam. Tepals 6, equal, elliptic, 4-6 mm long, ca. 3 mm wide, gray to ferruginous pubescent outside, papillose inside. Stamens 9, 4-celled, the cells lateral to apical, opening back-to-back, the glands strongly enlarged and, together with the stamens, forming a dome 2.5 mm wide and 1 mm high, the glands enclosing the stamens, but not fully fused. Pistil 2.2 mm Volume 80, Number 1 van der Werff 77 1993 Revision of Pleurothyrium FicunE 24. P. golfodulcense. — А. Habit. — B. Part of inflorescence. — C. Cross section of flower. — D, E. Lower and upper surface of leaf base. 78 Annals of the Missouri Botanical Garden long, ovary ca. 1 mm long, pubescent, style sparse- ly pubescent. Receptacle pubescent inside. Cupule 2 cm tall, Fruit unknown. Flowers: January-February. Elevation range: 200-800 m with coarse warts, 2 cm wide. Collections studied. Costa RICA. PUNTARENAS: for- O), Kernan 187 ( Osa, Ra MO). ncho Quemado, p eun 648 ( Pleurothyrium golfodulcense is restricted to the Osa peninsula and adjoining area in Costa Rica. I have accepted a rather wide concept of this species and include here, for instance, Kernan 184, which differs from typical P. golfodulcense in its thinner leaves, shorter petioles and gray vs. ferruginous pubescence. In leaf size and shape, venation, pu- bescent pistil and receptacle, gland-dotted leaves, petiole length, and bracts pubescent on both sur- faces it fits in Р. golfodulcense, and because it was collected together with typical P. golfodulcense (Kernan 187) and young leaves of Allen 5885 are as thin as those of Kernan 164, I prefer to include Kernan 184 in P. golfodulcense. The closest relatives of Pleurothyrium golfodul- cense are probably the Panamanian species P. racemosum and P. pilosum and the Ecuadorian- Colombian P. tomiwahlii (see discussion under P. tomiwahlii). Pleurothyrium grandiflorum van der Werff, i 1 Gard. 74: 408. 1987. TYPE: Colombia. Choco: Rio Mecana, Juncosa 1675 (holotype, MO). Figures 25, 26. nn. Missouri Bot. Tree, 15 m. Twigs ridged, becoming terete, sol- id, very finely appressed puberulent, ca. 3 mm diam. 5 ст below the tip. Terminal bud appressed puberulent, shiny. Leaves alternate, thinly char- 15-24 x elliptic, the base sharply acute or slightly decurrent taceous, 4-7 cm, elliptic or narrowly on petiole, the tip acute, the upper surface nearly glabrous, but with some fine appressed hairs, es- pecially on the midrib; the lower surface with scat- tered, minute, appressed hairs, these more nu- merous along lateral veins and midrib; venation on upper surface immersed, but slightly raised on low- er surface; lateral veins 15-20 on each side, arch- ing upward near the margin and becoming loop- connected in the upper half of the lamina. Petioles 1-1.5 ст long, with same kind and density of pubescence as twigs. Inflorescences to 10 cm long, with same kind and density of pubescence as twigs, racemose, few-flowered (usually 2 flowers present on each inflorescence), bracts not present at an- thesis. Pedicels to 6 mm long, appressed puberu- lous. Flowers yellow, 1.5-1.7 cm diam., fragrant (same odor as in some euglossine bee-pollinated flowers, fide collector). Tepals 6, subequal, the in- ner three narrower than the outer three, the inner surface with a dense, short pubescence. Stamens , 4-celled, the outer six with lateral cells, the inner three with extrorse cells, anthers glabrous, filaments not discernible in the glandular mass. Pistil 1.3 mm long, glabrous, the ovary partly sunken in the pubescent floral tube and covered by the glabrous glandular mass. Staminal glands strongly enlarged, fused, forming a thick wall out- side the stamens. Fruit unknown. Flowers: January. Elevation range: below 10 Collection studied. COLOMBIA. CHOCO: Rio Mecana, Juncosa 1675 (MO) Pleurothyrium grandiflorum represents the ex- treme in the tendency toward large flowers and few-flowered inflorescences in the genus. Its ve- nation type, with a rather weakly developed mar- ginal vein, is not specialized and its pubescence type is also rather nondescript, quite unlike the brown-tomentose pubescence which is common in Pleurothyrium, but rarely found in the closely related Ocotea and Nectandra. Pleurothyrium grandiflorum is only known from the type collec- tion in the Chocó, an area known to be rich in endemic species. Pleurothyrium hexaglandulosum van der Werff, Ann. Missouri Bot. Gard. 75: 417. 1988. TYPE: Panama. Colón: Rio Guanche, 5 km upstream from road to Portobelo, Hammel & Trainer 14781 (holotype, MO; isotypes, BR, PMA). Figures 26, 27 mall tree, to 5 m tall. Twigs terete, densely cm, obovate, ч ле narrowed toward the base, the base abruptly rounded to subcordate, the tip acuminate, the upper surface glabrous, the lower surface with some small appressed hairs, especially along midrib and lateral veins; lateral veins 14— 18 on each side, immersed on upper surface, raised on lower surface, the veins arching toward the tip near the margin and prominently loop-connected in the upper two-thirds of the lamina. Petioles tomentellous, ca. mm long, swollen, to 5 mm thick. Inflorescences in axils of cataphylls, to 65 cm long, brown puberulous, paniculate, the branch- lets 3-4 times cymosely branched, the basal Volume 80, Number 1 1993 van der Werff 79 Revision of Pleurothyrium FicunE 25. P. grandiflorum. Habit and flower. branchlets to 25 cm long, the upper ones gradually shorter. Bracts absent at anthesis. Pedicels 1.5 (-2) em long, densely gray-pubescent. Flowers ca. 10 mm diam., green, turning yellow. Tepals 6, equal, ca. 4 mm long, glabrous inside except where glands and stamens were not pressed against tepals, thus showing a narrow line of hairs in the center of the lower part of the tepal, this expended in a diamond-shaped pattern in the upper part; margin of tepals flat. Stamens 9, 4-celled, the outer six with introrse-lateral cells, the inner three with ex- trorse-lateral cells; in young flowers the anthers bent down and not raised above the glandular mass, but stamens becoming straight in older flowers with anthers clearly raised above the glands; filaments pubescent on the back. Glands conspicuously en- larged, but not fused, although enclosing the outer stamens. Ovary broadly ovoid, 1 x 1.5 mm, with short, gray pubescence, partly enclosed in the pu- bescent floral tube. Style ca. 0.5 mm long, with some gray pubescence. Glands and stamens decid- uous in older flowers, leaving the ovary fully ex- 80 Annals of the Missouri Botanical Garden | | | ке \ — O 200 400 600 800 1000km i њиви) == O 100 200 300 400 $00 600 miles k R. урок "t „~ А | > \ < zu : 4 979 b FIGURE 26. ш) posed. Fruits unknown. Flowers: March. Elevation range: 150-200 m Collections studied. Costa RICA. PUNTARENAS: Osa Peninsula, Croat & Grayum 59792 (MO). PANAMA. COLON: Rio Guanche, ca. 5 km upstream from road to Portobelo, Hammel & nou 14781 (BR, MO, PMA) Pleurothyrium hexaglandulosum is only known from the type collection. The collection Croat & Grayum 59792, from the Osa Peninsula, Costa Rica, is tentatively placed here, although it differs in a number of minor characters. The twigs are almost glabrous, the inflorescences much shorter (to 15 cm long) and less pubescent, the tepals papillose, not pubescent, and the pedicels are short- er. On the other hand, the Costa Rican collection has the short, swollen petioles, the leaf shape, the pattern on the inner face of the tepals (though not as sharply outlined), and the androecial structure of the type of P. hexaglandulosum. Because the differences are quantitative and the leaf shape, P the Netherlands D tS B Distribution of P. grandiflorum (X), P. hexaglandulosum (8), P. insigne (O), and P. intermedium thick petioles, and lax inflorescences are un- matched in Pleurothyrium, I place the Costa Rican collection in P. hexaglandulosum. However, if future collections res that the differences between the Panamanian and Costa Rican populations are consistent and not individual variation, it might be best to describe the Costa Rican material as a new taxon. Pleurothyrium insigne van der Werff, sp. nov. TYPE: Ecuador. Napo: Reserva Biológica Jatun y, NY, QAME, “ORS, U). Figures 26, 28. Arbor, ad 12 m alta. Ramuli teretes, solidi, da if tomentosi, vetustiores glabrescentes. Folia alterna, firm chartacea, 20-60 x 7-15 cm, obovata vel anguste obo. vata, apice breviter acuminata vel rotundata, basi (acuta) Volume 80, Number 1 van der Werff 81 Revision of Pleurothyrium £A Рае FIGURE 27. fallen off. —E. Sta obtusa vel abrupte rotundata (cordata), supra praeter cos- tae basem pu serene ere, imis pilis oracin parvis prae ita, cas taneoque, venatione supra immersa, а, subtus costa nervis- que elevatis; nervis (14- )18-26(-28) utroque costae la- tere, ad marginem in venam conspicuam marginalem terminantibus. Pet ioli 1.5-4 cm longi, н в -tomen- tosi. Inflorescentiae ad а ramulorum ex axillis brac- tearum ortae, ferrugineo-tomentosae, ad 40 cm longae, сар ои i. Pedicelli ad | cm longi. Flores rotati, 12- 1.5mm A а — A. Нађи. —В. Вазе of leaf. —C. Flower. — D. Old flower with androecium m diametro. bora 6, subaequalia, tres exteriora ut "elliptica , 5-6 x 4-5 mm, extus la, intus ute papillosa; tres interiora 4-5 X m, extus basi tomentella, demum papillosa, intus papillosa. Stamina 9, 4- b locellis lateralibus, glandulis permagnis, sta- mina cingentibus, staminodia parva, 3. Ovarium (de- presse) globosum, ca. 1.5 mm diametro, basi glabrum, demum ferrugineo- ние ва stylo ca. 1 mm longo. m intus tomentellum. rears verrucosa, 1.5 2.5 cm; fructus ellipsoideus, 2.5 x 2 cm. 82 Annals of th Missouri Botanical Garden FIGURE 28. P. insigne. — А. Нађи. — B, С. ша D. E. Outer and inner stamen.— F. Pistil. —G. Fruit. — H. anis on den leaf surface. —I. Leaf bas Volume 80, Number 1 1993 van der Werff 83 Revision of Pleurothyrium Small tree, to 12 m tall. Twigs solid, terete, 5- 9 mm diam. 5 cm below the tip, ferruginous to- mentose, but glabrescent with age. Leaves alter- nate, firmly chartaceous, 20-60 x 7-15 cm (nar- rowly) obovate, the base (acute) obtuse, abruptly rounded or cordate, the tip rounded or shortly acuminate; upper surface with some pubescence at the base of the midrib, otherwise glabrous, lower surface with an indument of erect hairs, this much denser on midrib and lateral veins; venation im- mersed on upper surface, midrib, lateral veins and marginal vein prominently raised on lower surface; )18-26(-28), connected by a conspicuous marginal vein. Petioles 1.5-4.0 cm lateral veins (14—- long, to 5 mm diam., castaneous tomentose. Inflo- rescences clustered near tips of branches in axils of cataphylls, rarely along leafless short-shoots, 15-40 cm long, paniculate, ferruginous tomentose. Bracts of inflorescences, if present, to 4 mm long, ovate, tomentose outside, nearly glabrous inside. Pedicels to 1 cm long. Flowers reddish outside, greenish white inside, the tepals spreading, 12-15 mm diam. Tepals 6, subequal, outer 3 elliptic, 5- 6 x 4-5 inner 3 elliptic, ca. 5 X mm, tomentose outside, papillose inside, —3 mm, outside with a basal, triangular, t , otherwise pap- illose; inside papillose. Siemens, 4-celled, the cells lateral; staminodia 3, minute, linear, ca. 0.3 mm long, hidden by stamens and glands. Glands of ae stamens strongly enlarged and enclosing the outer stamens, forming a large mass, but not fully fused. Ovary (depressed) globose, ca. 1.5 mm diam., the base glabrous, upper part pubescent. Style ca. | mm long. Floral tube shallow, pubescent within. Cupule 1.5 x Fruit ellipsoid, ca. 2.5 cm long, well exserted from cu- pule. Fruits: August- February. Flowers: June-No- vember. Elevation range: 100- 5 cm or smaller, warty. Collections studied. BRAZIL. AMAZONAS: Humaita, М 070 Р za" de Tenneco, 35 km NE of Motalvo, Zak 4511 (MO). NAPO: Jatun Sacha, 8 km downstream from Misahualli, Palacios 1332 (AAU, C, COL, С, HBG, K, KUN, LE, MO, NY, QCNE, QRS, U, US); Jatun Sacha Biological Station, Neill & Alvaredo 9033 (MO, QAME), Palacios 1518 (MO), Palacios 4388 (MO). PERU. LORETO: Varade- ro de Mazan, Croat 19439 (MO, NY); Iquitos, Santa Maria del Ojeal on Rio Amazonas, Davidson 5336 (MO, ; Maynas, jos Suerte (Rio Itaya), millo 8444 (M quena, Sinchic xad Oxapampa, Iscozacin, Arboretum, Pariona 49 (MO). SAN MARTIN: San Martin, Knapp & Mallet 8205 (MO). Pleurothyrium insigne is a striking species, eas- ily recognized by its large, obovate leaves with a pronounced marginal vein, the castaneous-tomen- tose indument of twigs, and the erect hairs on the lower leaf surface. Initially, I divided the specimens included in P. insigne between two species, one with abruptly rounded or cordate leaf base and short petioles (to 2 cm long), the other with an obtuse leaf base and longer (3-4 cm) petioles. Because some collections (for instance, Kruko 6232) include specimens with obtuse and abruptly rounded leaf bases and the difference in petiole length is not consistent, I decided to place all spec- imens in one species, even though the extremes look quite different. The specimens with an abrupt- ly rounded to cordate leaf base are very similar to P. maximum; the latter differs only in its glabrous lower leaf surface, smaller flowers, equal tepals, and relatively smaller glands. Pleurothyrium max- imum is largely recognized because of the floral differences. I consider such differences as funda- mental in Pleurothyrium and expect that additional collections of P. maximum will show that these differences will be consistent. Pleurothyrium insigne is known from several widely scattered localities in Ecuador, Peru, and Brazil. The best collections are from the Jatun Sacha Biological Station, where the species is lo- cally common. All collections from that area have leaves with an obtuse base; Peruvian collections and the sole Brazilian collection can have obtuse or abruptly rounded/cordate leaf bases. Pleurothyrium intermedium (Mez) Rohwer, Mitt. Inst. Allg. Bot. Hamburg 20: 43. 1986. Nectandra intermedia Mez, Repert. Spec Nov. Regni Veg. 16: 308. 1920. TYPE: Brazil. Acre: Seringal Sao Francisco, Ule 9408 (ho- lotype, B; isotype, K). Figure 26. Tree or shrub, 5-20 m tall. Twigs terete, solid, sparsely pubescent to glabrous, diameter 5 cm below the apex ca. 3 mm. Terminal buds densely appressed yellowish pubescent, the bud contrasting with the dark twig. Leaves alternate, chartaceous, elliptic to obovate-elliptic, 9-1 -5 cm, the base acute, the tip obtuse, upper surface glabrous, the venation immersed, the lower surface glabrous or nearly so, the midrib raised, lateral veins weakly raised; lateral veins 10-12 on each side, arching upward near the margin, not or very weakly con- nected with the superior vein. Petioles ca. 1.5 cm long, with similar tomentum as twigs. Inflores- cences in axils of deciduous bracts, ca. 12 cm long, paniculate, the branchlets with racemosely ar- ranged cymes, often the two lateral flowers of a cyme missing; main axis sparsely pubescent, the Annals of the Missouri Botanical Garden FIGURE 29. P. marginale. — А. Habit. — B. Flower. — C. Base of leaf underside, showing indument. branchlets denser pubescent, pedicels almost white pubescent, жы ) mm long. Flowers yellowish en, ca. m diam., the tepals = erect at anthesis. Tepals 6. equal, ovate, ca. 2 mm e puberulous outside, papillose inside, the margin reflexed. Stamens 9, 4-celled, the cells lateral fil- aments glabrous, about as long as the anthers, glands enlarged, but not fused and not surrounding the outer stamens, but visible as large lobes sep- arating the outer stamens; anthers raised above the level of the glands. Ovary glabrous, ovoid, gradually narrowed into the short style, stigma platelike; pistil 1.8 mm long. Floral tube glabrous inside. Fruit unknown. Flowers: August. Elevation range: 100-200 m "о a Collections e Jaa, SANTA CRUZ: Ichilo, Reserva Forestal , Cam royo Perdido, Neill 9435 (MO, ОСМЕ) Neill 9439 o ОСМЕ). BraziL. AMAZONAS: Seringal Sáo Francisco, Ule 9408 (B, K). PERU. MADRE DE DIOS: Tambopata, Hermosa Chica, Pesha 10 (MO). ~ Pleurothyrium intermedium is only known from the type collection, two recent collections from Bolivia, and one recent collection from Peru. It belongs to the group of species with erect tepals at anthesis and is most closely related to two other, equally poorly known, species from Brazil (P. am- plifolium and P. undulatum). More collections are needed for a better understanding of the status of those three species. Pleurothyrium marginale van der Werff, sp. ev. Cuatrecasas 17196 (holotype, US; isotype, F). Figures 29, 30 Volume 80, Number 1 van der Werff 85 Revision of Pleurothyrium о 200 асо 800 400 1000am —M———À = о 100 200 360 «20 $00 400 ~ Р RR FiGURE 30. Arbor, ad 20 m. Ramuli чык solidi, minute fusco- puberuli, glabrescentes. Folia a odroma, supra ersa, subtus s lateralibus et venatione tertia elevata. Petioli 15-22 a n ngi. ipu i нй ex axillis ctearum ortae, fusco-tomentulos 4-6 cm longae, mosae, uni- vel biflorae, bracteis a anthesi absen- tibus. Pedi velli ca. 4 mm longi, pedunculi ca. 9 mm longi, ч —. et breviter stipitata cc nq fusco-tomentella. CHEN 9, 4- Isi ај locellis edi ingentibus. Ovarium ellipsoideum, gla А 0.8 longum; lum um, ca. 1 mm longum, stigma pel- tum. Tubus floralis glaber. Fructus perjuvenalis in tubo florale inclusus; fructus adultus ignotus. Tree, 20 m tall. Twigs terete, solid, minutely brown-puberulous, glabrescent, ca. 3 mm cm below the tip. Leaves alternate, not clustered, firmly chartaceous, elliptic, 8-14 x 3-5 cm, the base acute to attenuate, the tip acuminate, the diam. 5 Distribution of P. marginale (X), P. maximum (8), P. nobile (O), and P. obovatum (I). upper surface glabrous or with some minute hairs, especially near the base of the midrib, the lower surface with scattered, minute, appressed hairs, these denser along midrib and lateral veins; ve- nation brochidodromous, immersed on upper sur- face, midrib and lateral veins raised on lower sur- face, a marginal vein well developed; lateral veins 8-12 on each side. Petioles 15-22 mm long, flat- tened on upper surface, pubescence similar to that of twigs. Inflorescences in axils of cataphylls, brown- tomentulose, 4—6 cm long, racemose, 1-2-flowered (but probably several flowers fallen off), bracts ab- sent at anthesis. Flowers yellow fide collector, ca. 12 mm diam. Pedicel ca. 4 mm long, peduncle ca. 9 mm long, brown-tomentulose. Tepals 6, subequal, the outer three broadly ovate, ca. 5 mm long, the inner three mostly rhombic with a short, broad stalk, ca. 4 mm long; both inside and outside brown tomentellous in sicco. Stamens 9, 4-celled, the cells lateral, opening back-to-back, anthers glabrous; glands enlarged, fused, enclosing the stamens. Ova- ry ellipsoid, ca. 0.8 mm long, glabrous, style gla- brous, ca. 1 mm long, stigma platelike. Floral tube Annals of the Missouri Botanical Garden glabrous inside. Very young fruit fully enclosed in the enlarged floral tube; mature fruit unknown. Flowers: April. Elevation range: below 100 m. Collection studied. | COLOMBIA. VALLE: Costa del Pa- cifico, Rio Cajambre, Cuatrecasas 17196 (F, US). A species only known from the type collection, Pleurothyrium marginale derives its name from its well-developed marginal vein. This species can be readily identified also by the sparse, appressed indument on the lower leaf surface and the few- flowered inflorescences. The collector indicated that the tree had tall buttresses (as in Cecropia) and canary-yellow wood. Buttresses are rare among neotropical Lauraceae: I have only seen them in a Caryodaphnopsis species from Peru. Yellow wood is a common feature in Aniba, but is infrequent outside this genus. The wood is used for making boards and canoes; the common name is palo blan- со. Pleurothyrium maximum О. C. Schmidt, No- tizbl. Bot. Gart. Berlin-Dahlem 10: 235. 1928. Ocotea maxima (O. C. Schmidt) Kostermans, J. Sci. Res. (Jakarta) 1: 122. 1952. TYPE Peru. Amazonas: mouth of Rio Santiago, Pon- go de Manseriche, terra firma forest, 160 m elev., Tessman 4040 (holotype, B; isotypes, G, NY). Figure 30. Small tree, 4-6 m tall. Twigs terete, solid, fer- ruginous tomentose, glabrescent in age, 5-6 mm diam. a few cm below the tip. Leaves alternate, chartaceous or firmly chartaceous, 30-50 x 7- cm, obovate to narrowly obovate, gradually narrowed toward the base, the base abruptly round- ed to cordate, the tip acute to acuminate, glabrous above, the lower surface with a few scattered, minute, appressed hairs, otherwise glabrous; midrib and lateral veins immersed on upper surface, prom- inently raised on lower surface, the tertiary ve- nation weakly raised. Lateral veins 20-30 on each side, near the margin connected by a conspicuous marginal vein. Petioles 1-2 cm long, 4-5 mm diam., with similar pubescence as the twigs. Inflo- rescences slender, lax, 40-65 cm long, ferruginous tomentose, the branchlets once or twice cymosely branched, in axils of cataphylls at tips of twigs, bracts mostly lacking at anthesis. Pedicels 5-9 mm long, ferruginous tomentose. Flowers rotate, ca. 8 mm diam. dark green-red-brown outside, dark green inside, the glands bright green, stamens yel- lowish. Tepals 6, equal in shape and size, 2.5 x 1.5 mm, the margin flat, the outer three ferrugi- nous tomentellous outside, the inner three with a basal, triangular, tomentellous patch, otherwise glabrous; all tepals glabrous inside except for a narrow line of short hairs in the center of the lower part, this expanded into a diamond-shaped outline in the upper part. Stamens 9, 4-celled, the outer six with 2 cells lateral and 2 cells introrse, the inner three with one pair lateral and one extrorse; glands enlarged, protruding between and + en- closing the outer stamens, but definitely free. Im- mature ovary globose to ovoid, 1-1.5 mm long, covered with short brown hairs; style glabrous. Floral tube inside covered with short reddish hairs. Fruit and cupule unknown. Flowers: September- November. Elevation range: below 200 m Collections studied. PERU. AMAZONAS: mouth of Rio Santiago, Pongo de Manseriche, Tessmann 4040 (B, G, NY), Tessmann 4529 (B, G, NY). Pleurothyrium maximum is only known from two collections from the area where the Rio San- tiago joins the Rio Maranon, in northern Peru. It is vegetatively similar to P. insigne, which see for further discussion. Pleurothyrium maximum also which differs in its nar- rower leaves (5 cm above the base, 2-3.5 cm wide resembles P. williamsii, vs. 9-6 cm wide), shorter indument on twigs (or glabrous) and inflorescences, thinner leaves that dry dark green, and shorter inflorescences (but few In P. glands of the stamens are free and do not form a wall surrounding the stamens, while in P. williams- ii the glands do form a wall, with only a faint line inflorescences аге known). maximum the indicating the boundaries of the individual glands. Pleurothyrium nobile A. C. Smith, Phytologia 1: 120. 1935. Ocotea nobilis (A. C. Smith) а. J. Sci. Bes (Jakarta) 1: 122. 1952. TYPE: Brazil. Amazonas: near mouth of Rio Embira, Krukoff 5121 (holotype, NY not seen; isotypes, A, B, C, K, MO, S, US). Figure 30 Tree, 25 m tall. Twigs solid, terete or ridged, glabrous, 6-7 mm diam. 5 cm below the apex. Terminal bud brown-tomentellous. Leaves alter- nate, stifly chartaceous, elliptic, 15-30 x 6-12 cm, the base obtuse to rounded, the tip acute, glabrous on both surfaces, venation immersed on upper surface, midrib and lateral veins raised on lower surface, tertiary venation weakly raised on lower surface, lateral veins 7-10 on each side, arching upward near the margin, but not or very weakly connected with the superior vein. Petioles 2-3 cm long, glabrous, canaliculate. Inflorescences in axils of cataphylls, to 12 cm long, minutely Volume 80, Number 1 1993 van der Werff 87 Revision of Pleurothyrium puberulous, glabrescent, paniculate, the branchlets 3-4 times cymosely branched, few-flowered, bracts not present at anthesis. Pedicels 3-4 mm long. Flowers ca. 10 mm diam. Tepals 6, equal, brown- puberulous on outside, minutely brown-papillose to glabrous inside, ovate, ca. 4 mm long. Stamens 9, 4-celled, the cells lateral, anthers and upper part of filaments brown-papillose, glands large, com- pletely surrounding the stamens but not fully fused. Ovary ovoid, ca. 2 mm long, the lower part gla- brous, the upper part brown-papillose, except for six large glabrous spots, where glands were in con- tact with ovary; style poorly differentiated, brown- papillose, 0.2 mm long. Floral tube glabrous inside. Cupule and fruits unknown. Flowers: June-July. Elevation range: 200-300 m Collections studied. vid AMAZONAS: mouth of Rio =, чај ан ан G, K, МО, NY, S, US). PERU. HUA : Puerto Eus. ia Llullapichis, Kroll 285 ПА ‘rol 422 (MO). LORETO: Coronel Portillo, m 98 c era Pucallpa-Huanuco, M. Castillo 12 (Е, K, MO, NY. US, WIS), M. Castillo 15 (F, K, US). Pleurothyrium nobile is rarely collected and is known only from five collections. The type collec- tion was made in varzea forest, two later collections (both from the same tree) in dry tropical forest, and two from wet forest. The recent collections have smaller, less-branched inflorescences, but the collections agree well in leaf shape, type of indu- mentum, and floral characteristics. Smith (1935) cited Krukoff 4800 as a paratype of P. nobile. Because this collection has а whitish, very short appressed indument on lower leaf surface, acu- minate leaf tips, and slender inflorescences, I in- clude it in P. panurense (Meissner) Mez. The com- mon name for the species in Peru is lobo moena. Pleurothyrium obovatum van der Werff, sp. nov. TYPE: Ecuador. El Oro: 15 km NW of Paccha on the road to Pasaje, 1,800 m elev., Brandbyge 42366 (holotype, AAU). Figures 30, 31 Arbor parva, ad 10 m . Ramuli teretes, fistulos Gemma terminalis унео omen Folia alt perobovata, 40-45 16-20 cm, fi acuta, apice мар vel breviter acuminata, g gineo-tomentelli, 20— 30 cm longae, paniculatae. Flores ca. 11 mm diametro. Tepala 6, aequalia, elliptica, dense tomentella vel papil- losa. Stamina 9, 4-locellata, locellis interioribus extrorsis, superioribus lateralibus, antheris erectis, super glandulas prominentibus. Glandulae liberae. Ovarium glabrum, stylo ca. 1 mm longo, papillis ÓN prae- dito. Tubus floralis glaber. Fructus igno Small tree, to 10 m tall. Twigs terete, fistulose, with scattered small, erect ferruginous hairs, di- ameter of twigs 5 cm below tip ca. 9 mm. Terminal bud minutely, but densely, ferruginous tomentel. lous. Leaves alternate, strongly obovate, 40-45 x 16-20 cm, firmly chartaceous, gradually narrowed to the acute base, the tip blunt or with a very short acumen, glabrous on both surfaces, midrib slightly raised or immersed, otherwise venation immersed on upper surface, midrib prominently raised on lower surface, lateral veins and tertiary venation progressively less raised. Lateral veins ca. 20 on each side, near the margin arching upward and in the distal half of the lamina connected with the superior vein, but a marginal vein not well devel- oped. Petioles ca. 2 cm long, glabrous or with some scattered ferruginous hairs, 5 mm diam. Inflores- cences in axils of cataphylls, sometimes fistulose, ferruginous-tomentellous, 20-30 cm long, panic- ulate, the lateral branches 2-3 times cymosely branched, many-flowered, bracts deciduous at an- i m long, gradually widened in floral tube. Flowers outside ferruginous brown, inside whitish, ca. equal, 4 mm long, elliptic, densely tomentellous or thesis. Pedicels ca. 3 mm mm diam., rotate. Tepals 6, papillose on both surfaces, the margins plane. Sta- mens 9, 4-celled, the lower pair extrorse, the upper pair lateral, the cells large; anthers erect and raised well above the glands. Glands not completely en- closing the outer stamens, not fused. Ovary ellip- .5 mm long, glabrous, gradually паг- ong and with some ferruginous papillae; stigma platelike. Floral soid, ca. rowed into the style, this ca. 1 mm tube glabrous inside. Very young fruit fully en- closed in floral tube. Fruits and cupules unknown. Flowers: Elevation 1,800 m September. range: 1,000- Collections studied. ECUADOR. AZUAY: Cuenca, E Sillado, Parque Molleturo, Ortiz-Proyecto Molleturo 192 (MO). EL око: Pasaje, 15 km NW of Paccha on road to Pasaje, Brandbyge 42366 (AAU). тол: 15 km E of Alamor on road to Celica, Brandbyge 42343 (AAU). Pleurothyrium obovatum is only known from three recent collections in southern Ecuador. The fistulose (lenticellate?) branches, relatively small glands of the stamens, and floral shape indicate a close relationship to the P. cuneifolium group. Pleurothyrium obovatum is, however, easily rec- ognized by the ferruginous indument, which be- comes floccose on older twigs, and obovate, gla- brous leaves. The large, erect anthers are another character of this species. The leaves of the two Brandbyge collections are more narrowed toward the base than the Ortiz collection, and it is likely 88 Annals of the Missouri Botanical Garden eO ges а C FicuRE 31. P. obovatum. — А, Habit. — B. Flowers. —C. Stamen (Whorl I). — D. Stamens s I and IIT) with fused filaments. — E. Pistil. — Е. Detail of inflorescence axis, showing hole used as passage by an Volume 80, Number 1 1993 van der Werff Revision of Pleurothyrium | | NE | = ed | [| | ty | | | | A | gamm пау | = pne NDS. > — LN MEN | x^ Li H —| Pp | - | | | | | A | | A ÉL Á " G £c > | deeem | | у. La pum | | || | | | PS” S тен | _| \ | m | | о 20 00 600 800 !000km || [Ce e ыыы) [nm am A EE] O 100 200 300 400 500 600 miles \ \ 100 9c 30 FiGURE 32. Distribution of P. palmanum (X), P. panurense (O), and P. prancei (О that the few collections do not show the full range of morphological variation. Pleurothyrium palmanum (Mez & J. D. Smith) Rohwer, Mitt. Inst. Allg. Bot. Hamburg 20: 41. 1986. Ocotea palmana Mez & J. D. Smith, Bot. Gaz. (Crawfordsville) 33: 258. 1902. TYPE: Costa Rica. San José: La Palma, Tonduz CR 12652 — J. D. Smith Herb. 7374 (B, GH, US: also reported in K, M; holotype not designated). Figure 32. Trees, to 20 m tall. Twigs solid, ridged or terete, tomentulose with brown or yellow-brown hairs. Ter- minal bud tomentulose. Leaves alternate, clustered near the tips of branches, chartaceous, 15- x 7-15 cm, obovate or broadly obovate, the base cuneate or acute, the tip rounded or with a very short, blunt acumen, upper surface with short, brown pubescence along midrib, otherwise gla- brous, lower surface densely yellow-brown to brown- tomentulose, sometimes with taller hairs mixed in, — lateral veins 5—9 on each side, these arching toward the tip and becoming weakly loop-connected in the distal half of the lamina; venation immersed on the upper surface or the midrib slightly raised, midrib and lateral veins elevated on lower surface, tertiary venation slightly raised. Petioles 1-2 cm long, te- rete, with similar pubescence as twigs. Inflores- cences in axils of cataphylls, 4-10 cm long, pa- niculate, the branchlets once cymosely branched, rarely depauperate and racemose, with up to 20 flowers, brown-tomentellous, bracts subtending flowers sometimes persisting, these elliptic, to 4 mm long, tomentellous on both surfaces. Pedicels ca. 3 mm long, tomentellous. Flowers ca. 10-12 mm diam. Tepals 6, equal in size, ca. 5 mm long, elliptic, the outer three tomentellous outside, inner three papillose except for the tomentellous base; all tepals papillose inside. Stamens 9, 4-celled, the cells lateral, the valves nearly back-to-back; fila- ments and anthers each ca. 1 mm long, papillose; glands large, partially fused, but not enclosing all outer stamens. Style ca. 1.5 mm long, papillose; 90 Annals of th Missouri Botanical Garden ovary globose, glabrous, ca. 1 mm long, enclosed in the glabrous floral tube. Cupule and fruit un- known. Flowers: November. Elevation 1,000-1,600 m range: Collections studied. COSTA RICA. LA PALMA: Rowlee 233 (F, US), CARTAGO: Barbour 1011 (F, MO). PUNTARE- NAS: Reserva ои Montev pes Powell is Pacific 6 (MO). SAN La Palma, Tonduz d PANAMA. E Palo Alto: just E of Boquete, Stern et al. 1064 (MO). Pleurothyrium palmanum is a rarely collected species. The concept of this species accepted here is different from that in Burger & van der Werff (1990); one of the two flowering collections cited in that publication (Barbour 1012)is here excluded from P. palmanum. It differs in its indument (lower leaf surface is nearly glabrous, not entirely covered y a brown-tomentellous indument) and flower structure (glands are fused and enclosing stamens, not only partly fused and not enclosing stamens). The only flowering collections I have seen are the type and Haber 9526. The glands in P. palmanum are not strongly enlarged and the stamens are erect, with the lower pair of locelli lateral and the upper introrse. The available collections are quite 8-28- cm-long)leaves; Barbour 1011, а sterile collection, variable in leaf size. The type has large (1 has leaves to 15 cm long, while Stern et al. 1064 (from Chiriqui, Panama) and Haber 9526 have even smaller leaves, to 10 cm long. Because Haber 9526 and the type have the same flower type, leaf shape (+ obovate), and indument, taxonomic value to the striking variation in leaf I attach no size. The collection Barbour 1012 remains un- placed. As mentioned earlier, | exclude it from P. palmanum, but the specimens (a MO and a US sheet) have few flowers, some of which had been dissected by earlier botanists and were unrecog- nizable, while others were covered with fungus or were otherwise damaged. Vegetatively, Barbour 1012 somewhat resembles P. golfodulcense. Pleurothyrium panurense (Meissner) Mez, Jahrb. Bot. Gart. Berlin 5: 468. 1889. tandra panurensis Meissner, D.C. Prodr. 15(1): 157. 1864. Ocotea panurensis (Meiss- ner) Kostermans, J. Sci. Res. (Jakarta) 1: 122. 1952. TYPE: Brazil. Amazonas: near Panure along the Rio Uaupés, Spruce 2449 (B, BM, BR, E, C, K; holotype not designated). Figures 32, 33. Tree, to 20 m tall. Twigs solid, terete, minutely E brown-tomentellous, ca. 4 mm diam. 5 cm below the apex. Terminal bud brown-tomentellous. Leaves clustered near tips of branches, chartaceous, 15- Ох cm, elliptic to elliptic-obovate, the base obtuse or rounded, the tip acuminate; upper sur- face glabrous, lower surface covered with a very fine, almost appressed whitish tomentum, this par- tially wearing off on older leaves; lateral veins 12- 16, arching upward near the margin and loop- connected in the upper half of the laminae, ve- nation immersed on upper surface, costa and lateral veins raised on lower surface. Petioles 5-14(-24) mm long, with similar indument as twigs. Inflores- cences in the axils of cataphylls, brown-tomentel- lous, 5-10 cm long, paniculate, the branchlets 1- 2 times cymosely branched, 15-30-flowered, bracts deciduous at anthesis. Pedicels to 5 mm long. Flow- ers white, ca. 8 mm diam. Tepals 6, equal, outer three gray-tomentellous outside, inner three with a basal triangular tomentellous patch, otherwise glabrous, elliptic, ca. 3 mm long. Stamens 9, 4-celled, all with 2 cells extrorse, the other pair lateral; stamens scarcely raised above the glandular mass that completely surrounds the stamens; ovary glabrous, globose, ca. 1.3 mm long, enclosed most- ly in the rather deep, glabrous floral tube, the upper part covered by the glands; style ca. 0.5 mm long, papillose; stigma platelike. Cupule deeply cup- shaped, 2-2.5 gradually narrowing into the pedicel. Fruit ellipsoid, cm wide, ca. 1.5 cm long, warty, ca. 2.3 cm long, 1.4 cm wide. Fruits: December- April. Flowers: May-June. Elevation range: 100- 200 m Collections studied. BRAZIL. AMAZONAS: near mouth of Rio Embira, Krukoff 4783 (G, NY); mouth of Rio Embira, tributary of River Tarauaca, Krukoff 4800 (A, ga Uaupes, Spruce 2449 (B, BM, BR, E, G LORETO: Maynas, Mishana, Río Nanay, Ayala 5734 (MO); Maynas, vicinity of Iquitos, Sides 3687 (MO); Maynas, Pto. Almendras, Vasquez 10233 (MO); San Antonio, Rio Pintuyacu, Vásquez et al. 7468 (MO). ucAYALI: Coronel Portillo, Leoncio Prado (Yarina Cocha), Vasquez 4970 i Pleurothyrium panurense is a rarely collected species, known with certainty from six collections: the type collection, four recent collections from Peru, and a collection from Brazil. All collections, including the type (indicated on the K sheet), come from inundated forest. Allen (1964) listed two ad- ditional collections of P. panurense, Froes 20526 and 21295 (both NY), and presented a description of cupule and fruit based on Froes 20526. 1 assign these collections to P. vasquezii because the pu- bescence on the lower leaf surface consists of scat- tered, erect hairs (not a finely appressed, whitish indument) and because the leaves are not clustered. Volume 80, Number 1 1993 van der Werff 91 Revision of Pleurothyrium FicunE 33. Also, the marginal vein is more strongly developed, and the pubescence on the young twigs is ferru- ginous-tomentose rather than light-brown tomen- tellous. Thus, Allen's description of cupules and fruits of P. panurense should be ignored. Actually, Pleurothyrium panurense can be readily identified by the following combination of P. panurense. — А. Habit. — B. Flower. — C. Detail of lower leaf surface. characters: short (to 10-cm-long) inflorescences, inner face of tepals glabrous, leaves clustered, leaves with an obtuse or rounded base, and leaves with a whitish indument on lower leaf surface. The flower size is normal for the genus and the glands are well developed. It is likely that Prance et al. 13927 (NY) be- 92 Annals of the Missouri Botanical Garden о 200 400 600 800 1000km \ O 100 200 300 400 $00 600 miles \ RR E FIGURE 34. longs here, too. However, the leaves are rather narrow and not clustered; the indument agrees with Pleurothyrium | panurense. specimen is sterile. Krukoff 4800 and Vasquez 4970 are provi- sionally placed here. They differ from P. rense in their acute leaf bases and longer petioles Unfortunately, the (18-24 mm vs. 5-13 mmin typical P. panurense), but agree well in inflorescence and floral charac- ters. Both collections were made in flooded forest. Although it is possible that future collections will show these collections to be different from P. panu- rense, 1 prefer to include them for the time being in P. panurense. Krukoff 4800 was cited by A. C. Smith as a paratype of P. nobile, but it differs from that species in its indument. Pleurothyrium parviflorum Ducke, Arch. Jard. io de Janeiro 5: 114. 1930. Ocotea parviflora (Ducke) Kostermans, J. Sci. Res. (Jakarta) 1: 122. 1952. TYPE: Brazil. Ama- zonas: Igarapé do Lago de Serpa prope Ita- coatiara, ad ripas inundatas, Ducke RB 19935 Distribution of P. parviflorum (X), P. pauciflorum (O), and P. pilosum (О). (holotype, RB; isotypes, B, U). Figures 34, 35 Pleurothyrium densiflorum A. C. Smith, Bull. Torrey Bot. Cl ~ Kostermans, 3; кы Res. (Jakarta) 1: 122. 1952. : Peru. Loreto: Mishuyacu, Klug 1372 (holo- dn NY not seen; isotype, US). Tree, to 30 m tall, but usually not exceeding 20 m. “Twigs slightly angular or terete, fistulose, glabrous or nearly so, ca. 5 mm diam. a few cm below the tip. Terminal bud covered with gray, fine appressed hairs. Leaves alternate, chartaceous, 10-20 x 3.5-9 cm, elliptic, the base acute to truncate, apex acute, glabrous or nearly so on both surfaces. Lateral veins 10-15 on each side, curving toward the tip near the margin, but not uniting with the superior veins; veins immersed on upper surface, midrib and lateral veins raised on lower surface. Petioles 10-15 mm long, glabrous, some- what canaliculate. Inflorescences in axils of leaves or cataphylls, glabrous or with some minute ap- pressed pubescence, 5-15 cm long, the branches up to 4 times cymosely branched, frequently flat- Volume 80, Number 1 1993 van der Werff 93 Revision of Pleurothyrium FIGURE 35. inflorescence. tened, many-flowered; bracts absent at anthesis, but present in younger parts of inflorescences, lin- ear, to 2.5 mm long, glabrous or with some minute appressed pubescence, the margin ciliate. Pedicels 3-5(-7) mm long. Flowers green or greenish white, ca. 5 mm diam. Tepals 6, equal, narrowly elliptic, ca. 2 mm long, the margins reflexed, glabrous or P. parviflorum. — А. Habit. —B. Flower. Fruit. — E. Part of — C. Cross section of flowers. — D. minutely papillose outside, papillose on inner sur- face. Stamens 9, ca. 0.9 cells lateral, filaments glabrous, as long as anthers. mm long, 4-celled, the ands almost completely encircling the outer sta- mens, but not fused. Ovary glabrous, ellipsoid, ca. 0.7 mm long, gradually narrowed into the 0.4-mm- long, glabrous style. Floral tube deep, fully en- 94 Annals of the Missouri Botanical Garden closing the ovary, glabrous. Cupule with large len- ticels, completely enclosing the fruit when young, at maturity bowl-shaped, 2.5 cm wide, 1 cm high, sometimes smaller and shallower; fruit. ellipsoid, 3.6 x 2.7 cm, well exserted. Fruits: December- April. Flowers: April- August. Elevation range: 100— 300 m Collections studied. BRAZIL. ACRE: Rio Moa, 10 km from Maita, Prance et al. 11970 (K, NY). AMAZONAS: uc aii de Serpa, Ducke RB 19935 (B, RB, Uy Rio Jutahy, Froes 21036 (F, NY); Rio Asapinus, Tapajos, Miranda Bastos 17 (RB); delta of Rio Jauaperi, Mori 20470 (MO, NY); mouth of Rio Ica along Rio Solimoes, Mori et di 9074 (K, MO, NY); Mun. 2 "e Ecologica Jutai-Solimoes, Cid Ferreira et al. 72 O, NY). ECUADOR. A. Rio Cuyabeno, Berg & Ak kerman 1050(F, K, MO, U); Aguarico, Parque Nacional Yasuni, Garza Cacha, ven 4982 (MO). PERU. Soledad, nacocha, Nueva Esperanza de Panaillo, Vasquez 10500 (MO). LonETO: Manfinfa on upper Rio Nanay, Llewelyn Williams 1090 (F); Caballo-Cocha, Llewelyn Williams 2273 (К, С); Maynas, Caballo-Cocha, Ayala 3386 (MO, NY); Maynas, Croat 17593 (F, MO, NY); Maynas, Puer- Nanay, G ada с с Portugal nas, Mishuyacu, Klug 1301 (F, NY, y: a 1372 (NY, US); Maynas, shana, López 8634 (NY); Maynas, Rio Nanay below Bellavista, McDaniel 15243 (F, NY); Maynas, Río Momon above Bellavista, o 20716 (MO, NY), “изу 20753 s, Rio Atacuari, Mori et al. (F, M NY); Mut Río Yaguasyaco, affluent of Rio Ampiyaco, Plowman 6908 (F); Maynas, vio Isla Rondina, Plow man et al. 6410 (F, GH); Maynas, Rio Nanay, cerca a Santa Clara, Revilla & ju 2532 (F. MO); Maynas, Rio beer ade V us 742(F, G, yu Maynas, Alber- que Selva Tou Momon, Ruiz 1 MO); Maynas, Yanamono Ex i ama Tourist зно ав van кзн Werff 9956 MO); Maynas, Yanamon ПР ma Lodge, Vasquez & Jaramillo 9138 (MO); Maynas, Iquitos, Caseria Nueva Jerusalem, Isla Iquitos, Vásquez 11507 (MO); Requena, Jenaro Herrera, d 55 (MO); Requena, Rio Tapiche, Gentry et al. 21284 (MO); Requena, Jenaro Herrera, Lamotte IMG. Lamotte 085 (MO); боа. Jins Tapiche), Vásquez et al. 4858 (MO, NY); Ucayali, Yari- nacocha (near Pucallpa), Vásquez & Jaramillo 1520 (MO); Ucayali, Jenaro Herrera, Quebrada Supay, Vás- quez 12284 (MO). Pleurothyrium parviflorum is one of the most frequently collected Pleurothyrium species. It is apparently a common tree in seasonally flooded forest (tahuampa) in Amazonian Peru and has been reported also from Brazil (Estado Acre and Ama- zonas) and Ecuador. This species can be easily identified by its habitat preference, and the com- bination of hollow stems (sometimes inhabited by ants), nearly glabrous inflorescence, glabrous leaves, and the reflexed margin of the tepals. There are a few other Pleurothyrium species with hollow twigs and tepals with reflexed margin and of those, Pleurothyrium poeppigii Nees can be confused with P. parviflorum. However, P. poeppigii has a dense, minute cover of hairs on the lower leaf surface, stiffer leaves, pubescent flowers and inflo- rescences, and smaller glands that do not enclose the outer stamens. One collection, Nunez 5901 (MO), is intermediate between the two. It resembles P. parviflorum in the lack of pubescence and the texture of the leaves and size of the glands, but has the stiffer, narrower infloresences and pubes- cence of the flowers found in P. poeppigii. Com- mon name: Yacu Moena (= Moena del agua). Pleurothyrium pauciflorum van der Werff & ammel, sp. nov. TYPE: Costa Rica. Puntare- nas: Cantón de Osa, ca. 2 km NE of the union of Rio Pavón and Rio Rincón, tree, 30 m, fl., Hammel et al. 17909 (holotype, MO; iso- types, BM, CR, MEXU, USJ). Figures 34, 36 Arbor, ad 35 m. Ramuli teretes, solidi, ferrugineo- tomentelli. Gemma apicalis ferru e. -tomentella. Folia alterna, firme chartacea, 16-35 12 cm, oblonga vel elliptica, basi acuta vel obtusa, apice acumine ad 2 cm longo praedita, nervis lateralibus 14-20 utroque cos- tae latere, brochidodroma in dimidio distale, supra glabra vel basi sparse tomentella, subtus pilis plus minusve erectis semi-erectis. Tepala 6, aequalia, crassa, 4-5 mm ca. 4 mm lata, intus paullo papillosa, tres exteriora extus tomentella, tres interiora basi tomentella, demum papil- losa. Stamina 9, 4-locellata, locellis lateralibus, glandulis perauctis, staminibus cingentibus. Pistilum ca. 3 mm i im i m glabrum at- uc 3 ст m profundae, tepalis eee persis- des peu, ellipsoideus, 3 x 1.5 с Tree, to 35 m tall. Twigs terete, solid, ferru. ginous-tomentellous, 4—6 mm diam. ca 5 cm below apex. Terminal buds ferruginous-tomentellous. Leaves alternate, grouped near tips of branches, firmly chartaceous, 16-35 x 7-12 cm, oblong or elliptic, the base acute or obtuse, tip acuminate, acumen to 2 cm long, but frequently damaged, Volume 80, Number 1 van der W 95 1993 erff Revision of Pleurothyrium \ FIGURE 36. P. pauciflorum. — А. Habit. — B. Inflorescence with flower. — C. Fruit. 96 Annals of the Missouri Botanical Garden lateral veins 14-20 on each side, curving strongly upward near the margin and becoming loop-con- nected in upper half of the leaf, upper leaf surface glabrous or with traces of a tomentellous indument along midrib, the lower surface with + erect hairs, this indument denser along veins and becoming tomentellous along major veins; veins immersed on upper surface, midrib and lateral veins prominently raised on lower surface, the tertiary venation slight- ly raised, but readily visible because of its indu- ment. Petioles 1.8-3.2 cm long, upper surface flat, with similar indument as on twigs. Inflorescences in axils of (tardily) deciduous bracts, to 3 em long, with up to 7 flowers (usually 3 or 1), ferruginous- tomentellous, bracts at base of inflorescences fre- quently present at anthesis, ovate or elliptic, to 5 mm long, pubescent on both surfaces. Pedicels ca 2 mm long, ferruginous-tomentellous. Flowers ca. 8 mm the tepa spreading, never fully rotate; in old flowers tepals erect. Tepals 6, equal in size and shape, thick, 4— 5 mm long, ca. 4 mm wide, all tepals inside slightly diam., s half-erect or somewhat papillose, the outer 3 outside ferruginous-tomen- tellous, inner 3 with a basal triangular tomentellous area, otherwise slightly papillose. Stamens 9, all 4-celled, the cells lateral, glands strongly enlarged, enclosing stamens and fused. Pistil ca. 3 mm long, the pubescent ovary gradually narrowed into the glabrous style; receptacle + pubescent. tescences 2.5 Infruc- -4.5 cm long, usually with 1 fruit; cupule thick, warty, ca. 2 cm tall, 3 cm wide and 1.5 em deep, the tepals often persisting on the margin; fruit ellipsoid, ca. 3 х 1.5 June. Flowers: October. 200 m. cm. Fruits: 50- Elevation range: Collections studied. COSTA RICA. PUNTARENAS: Osa, Hammel 17009 (MO); Osa, Reserva Indigena Guaimi, Hammel 17909 (BM, CR, MEXU, MO, US). Pleurothyrium pauciflorum can be readily iden- tified by the ferruginous-tomentellous branchlets and the short, few-flowered inflorescences. Depau- perate specimens of P. golfodulcense may also have short inflorescences, but this species has twigs covered with more or less appressed, ascending hairs, lacks erect hairs on the lower leaf surface, and has rotate flowers. It is uncommon to find a Pleurothyrium species with strongly enlarged glands that has half-erect tepals instead of rotate flowers. The androecium was, in all flowers seen, more or less covered by the tepals. Also, the tepals were, in comparison to other Pleurothyrium species, quite thick and fleshy. It would be interesting to discover what insect might pollinate this species; the ar- rangement of tepals suggests a beetle. кеш pilosum van der Werff, sp. nov. PE: Panama. San Blas: Cangandi, 9?24'N, ог. hills near village, de Nevers, Her- a & Charnely 7520 (holotype, MO; iso- M CAS, PMA). Figure 34 or. Ramuli angulares, lateritio-tomentosi, cicatri- cibus conspicuis praediti. Gemma terminalis brevis, crassa, lateritio-tomentosa. Folia ad apices ramulorum congesta, chartacea, 15-25 x 7-13 cm, obovata, basi obtusa vel rotundata, арген rotundata vel а ala nervis lateralibus 6 t acutiore quam venis distalibus abeuntibus, supra costa nervisque immersis, subtus elevatis, reticulatione paullo elevata; supra glabra, subtus dense day ; Me hir se- cus costam nervosque tomentoso. Petioli 1.5-3 cm longi, lateritio-tomentosi. Inflorescentiae axillis Mira Ni ortae, ad 15 cm longae, lateritio-tomentosae. Flores non liberis. Pistillum glabrum, receptaculum pubescens. Fruc- tus immaturus in cupula profunda, hexangulare inclusus. Tree. Twigs angular, reddish brown tomentose, with conspicuous scars of old leaves, solid, 6 mm diam. ca. 5 cm below the tip. Terminal bud short, thick, reddish brown tomentose. Leaves clustered near tips of branches, chartaceous, 15-25 x 7- 3 cm, obovate, the base obtuse to rounded, tip rounded or very shortly acuminate, lateral veins — 6—9 on each side, the lowest pair leaving the midrib at the very base of the lamina and under a smaller angle than the more distal lateral veins, lateral veins approaching the leaf margin very closely and strongly arching upward, but only the upper 2 or 3 veins loop-connected, midrib and lateral veins immersed on upper leaf surface, but prominently raised on lower surface, the tertiary venation slight- ly raised; upper surface glabrous or with very few scattered reddish hairs, lower surface densely cov- ered with erect, reddish hairs, these not obscuring the surface, the indument becoming tomentose along lateral veins and midrib. Petioles flattened above, 1.5-3 ст long, reddish brown tomentose. Pubes- cence on twigs and veins on lower leaf surface described by collectors as orange. Inflorescences in axils of cataphylls, at tips of branches, to 15 cm long, reddish brown tomentose, branchlets once cymosely branched or flowers racemosely ar- ranged. Open flowers not seen. Tepals (based on remnants on young fruits) broadly elliptic, densely pubescent on both surfaces, 3 mm long. Stamens pubescent, 9, 4-celled, cells lateral; glands en- Volume 80, Number 1 1993 van der Werff 97 Revision of Pleurothyrium larged, forming a dome and largely enclosing sta- mens, free (the glandular mass is broken up in fruiting stage). Ovary and style glabrous; recep- tacle pubescent inside. Young fruits enclosed in cupule, cupule with 6 ribs, to 1.5 cm long. Young fruits: March. Elevation range: ca. 3 Collection studied. PANAMA. SAN BLAS: Cangandi, de Nevers et al. 7520 (CAS, MO, PMA). Pleurothyrium pilosum, only known from the type collection from San Blas, Panama, is closely related to P. racemosum, another Panamanian spe- cies only known from the type. These two species have in common a glabrous ovary and style, pu- bescent receptacle, clustered leaves, the basal lat- eral vein leaving the midrib under an acute angle at the very base of the lamina, and lateral veins that approach the leaf margin very closely. Al- though some of these characters occur also in other Pleurothyrium species, the ascending basal lateral veins are unique to these two species. Pleuroth y- rium pilosum differs from P. racemosum in its larger leaves (to 25 vs. 14 cm), in its leaf shape (obovate vs. oblong), indument of lower leaf surface (pilose vs. with scattered curled hairs or subgla- brous) and indument of twigs (longer, red-brown vs. shorter plain brown). Unfortunately, the only collection of Pleurothy- rium pilosum is in young fruit. The description of floral parts is based on dried remnants on the young fruits, and I am not certain that the dimensions are correct. Because the type collection was made from a marked tree in a permanent quadrat, it is likely that the tree will be recollected in the future. Pleurothyrium poeppigii Nees, Syst. laur.: 349. 183 oo poeppigii (Nees) Kostermans, J. Sci. Res. (Jakarta) 1: 122. 1952. TYPE: Peru. s Poeppig addenda 301 (B, C, GZU, L, holotype not designated). Figure 37. ad ca krukovii A. C. Smith, S оя ls 121; Ocotea krukovii (A. C. Smith) т | ch Res. (Jakarta) 1: : Brazil. Acre: near mouth of Rio Macas: Krukoff 5563 (holotype, NY not seen; isotypes A, G, K, MO) Tree to 30 m tall. Twigs terete or subangular, fistulose, minutely brown-tomentellous or glabrous, 4-7 mm diam. са. 5 cm below the tip. Terminal bud minutely brown-tomentellous. Leaves alter- nate, firmly chartaceous or subcoriaceous, 20–: -10 cm, (narrowly) elliptic to slightly (narrow- ly) obovate, the base acute, tip rounded or shortly acute, the upper surface glabrous, lower surface minutely tomentellous, papillose or rarely glabrous, lateral veins 14—18 on each side, arching upward near the margin, not or very weakly loop-con- nected, venation immersed on the upper surface, midrib and lateral veins raised on the lower surface. Petioles 1.5-2 cm long, with similar indument as on twigs, canaliculate. Inflorescences in axils of deciduous bracts, or infrequently in axils of regular leaves, to 18 cm long, minutely tomentellous or subglabrous, the branchlets complanate, to two times cymosely branched, the flowers often appearing densely clustered; bracts deciduous at anthesis. Pedicels 2-3 mm long. Tepals 6, equal, + erect at anthesis, the margin reflexed, elliptic, ca. 2 mm long, the outer three minutely tomentellous outside, the inner three with a minutely tomentellous, tri- angular, basal patch, otherwise papillose; all tepals papillose on inner surface. Flower color described as white, green, and dull orangish; odor of melon. Stamens 9, all 4-celled, glabrous, 2 cells + ex- trorse, the other 2 + introrse, filaments about as long as anther, outer stamens separated by the enlarged glands, but glands not surrounding sta- mens and remaining free. Ovary ellipsoid, mostly included in the glabrous receptacle, glabrous or papillose near the tip, the style about as long as ovary, papillose; pistil ca. 2 mm long. Cupule cup- shaped, = warty, ca. 1.5 cm high, 1.8 cm wide. Fruits: January, August, and November. Flowers: March, April, August, and December. Elevation range: 400-1,800 m У Ie BOLIVIA. COCHABAMBA: Totor Yungas of mayu, Cardenas 3966 (F, US). BRAZIL. АСКЕ: ШИШ of Rio Macauhan (tributary of Rio Yaco), Krukoff 5255 (A, F, G, MO, NY, S, US); near mouth of Rio Macauhan (tributary of Rio Yaco), Krukoff 5263 (S); mouth of Rio Macauhan (tributary of Rio Yaco), Krukoff 5563 (A, G, K, we ), Krukoff 5722 (A , NY). COLOMBIA. PUTUMAYO: alon miyaco River, shove confluence with Rio Mocoa, alo 1268 (GH). ECUADOR. NAPO: road to Puerto Napo-Tena, Lugo 320 С 7-16 km E of Puerto Napo, Neill 6555 (F, МО, NY); along road to Tena, 2 nd om Misahualli, Palacios 2775 (MO); Jatun Sacha, m abajo de Puerto Misahualli, Palacios 982 (МО); Plon Neill 7603 (MO). Peru. Ruíz & Pavón s.n. (F). LORETO: Wr des Yurimaguas, Poeppig Addenda 301 (B, G, GZU, LE). MADRE DE DIOS: Cocha Cashu РИТ Station, Manu National Park, Gentry 43560 MO, NY); Manu, Rio Manu, Cocha Cashu Station, Da- vidson 85-19 (F); Manu, Cerro de Pantiacolla, Rio Раја- toa, Foster 10642 (MO); Manu, Rio Manu, Foster 3153 (F, MO); Manu, Cocha Cashu uplands, Manu Park, Nuñez 5901 (MO). Pasco: around Oxapampa, van der E 8343 (MO), van der Werff 8345 (MO); Oaxapampa, 5 km SE of Oxapampa, Smith 2793 (MO); Oxapampa, around Villa Rica, van der Werff 8288 (MO). ~ Annals of the Missouri Botanical Garden о 200 400 500 800 !000«m ЯНЫШ ыны ыйа quát. O 100 200 300 400 500 800 miles Prepared by Hendrik R. Rypkema FIGURE 37. ) As accepted here, P. poeppigii is a wide-ranging rrino n Amazonian Brazil, Bolivia, Peru, гае and Colombia. Its nearest relative is P. parviflorum Ducke, from which it differs in the tomentellous flowers and inflorescences, the nar- rower inflorescences, the tomentellous lower leaf surfaces, and the more coriaceous and generally larger leaves. The type specimens of P. poeppigii and P. krukovii are both somewhat atypical for this species. Poeppig addenda 301, the type of P. poeppigii, has strongly fistulose twigs and even some of the inflorescences are fistulose. These in- florescences are shorter and more compact than nonfistulose inflorescences on other specimens. The flowers of Poeppig addenda 301 are slightly larger than on the other specimens, and the pedicels are slightly thicker. However, I feel that these differ- ences do not warrant the recognition of two species, and І am inclined to accept most of these differ- ences as a consequence of the strongly fistulose character of twigs and inflorescences. Oddly enough, none of the labels indicate that the twigs were inhabited by ants, although the hollow twigs and Distribution of P. poeppigii (X), P. racemosum (O), P. steyermarkianum (O), and P. synandrum the exit holes along the twigs strongly suggest that ants colonize the twigs. See also discussion under P. cuneifolium. The type rather thin and glabrous leaves, characters found e of P. krukovii has, for this species, in P. parviflorum. The inflorescence and the flower buds have, however, the indumentum of P. poep- pigii. In spite of the fact that A. C. Smith described flowers of P. krukovii, none of the isotypes I have seen had flowers, and the microfiche photo of the NY holotype does not show flowers either. Krukoff 5722, a paratype, also has glabrous leaves, but has tomentellous flowers. Krukoff 5255 has stiffer it fits my concept of P. poeppigii very well. These ob- leaves and tomentellous leaves and flowers; servations allow two approaches. One is to define three species narrowly: P. poeppigii, only known from its type; P. krukovii, with very few collections (excluding, for instance, Krukoff 5255) and defined by its glabrous, chartaceous leaves and subglabrous inflorescence axes; and a third, undescribed species occurring in Brazil, Peru, Ecuador, and Colombia with subcoriaceous, tomentellous leaves and rather Volume 80, Number 1 1993 van der Werff 99 Revision of Pleurothyrium lax inflorescences. The other approach is to accept P. poeppigii in a wider sense, defined by its fis- tulose stems, tomentellous flowers, narrow inflo- rescences, and similar floral structure. I take this approach rather than recognizing three poorly de- fined and hard to identify taxa. For the same reason, the collection Cardenas 3966 (F, US) is included in Pleurothyrium poep- pigii, although its obovate leaves are aberrant for this species. In other characters, such as flower size and structure, fistulose and tomentellous twigs, it fits in P. poeppigii, and 1 am not willing to describe a new taxon with only one collection based solely on a different leaf shape. This is the only collection of P. poeppigii from Bolivia. In addition to Pleurothyrium parviflorum Ducke and P. poeppigii, P. cuneifolium Nees can also have fistulose twigs. This latter species can be rec- ognized by its angular, lenticellate twigs, larger flowers, and tepals that are glabrous on the inner surface. Although most collections of Pleurothyrium poeppigii are from lowland sites, I include her Ф three collections from the Oxapampa area in cen- tral Peru from ca. 1,800 m elevation. Pleurothyrium prancei van der Werff, sp. nov. TYPE: Brazil. Acre: vicinity of Serra de Moa, varzea forest, stilt-rooted tree, 8 m tall, Prance et al. 12392 (holotype, F; isotypes, К, MO, , 5). Figure 32 Arbor, ad 8 m. Ramuli parum айып, cinnamomeo- tomentelli; glabrescentes. Gemma t tomentosa sete alterna, — ea, ouod anguste о acumin 4.5-9 cm, C псе ata, acumine a ui cm longo, кке glabra, | elevata; nervis 9 sum arcu . Petioli 2-3 cm lor E cataphyllorum ortae, in apici ibus ra- mulorum congestae, der TAE 10-20 cm longae, ramulis semel, raro bis cymae more furcatis, brac- teis sub = кади "Pedicelli 5-10 v longi. Flores intus virides, m diametro. Tepala 6, subaequalia, ка sone dien , tres exteriora late elliptica 4 x 3 mm, extus cinnamomeo-tomentella, intus maxime basi Е centri A pes demum glabra; tres interiora prx 3 x extus parte нка tomentella, dem rn ntus kde ense a 9, 4-locellata, locelli к, fila- florescentiae axillis vix stamina cingentibus, liberis. Pistillum ca. 2 r gum, pubescens, ovario sensim in stylum brevem atten- uato, stigmate incrassato. Fructus ignotus Tree, to 8 m tall, with stilt roots fide collectors. Twigs slightly angular, brown-tomentellous, gla- brescent, 4 mm diam. ca. 5 mm below tip. Terminal bud brown-tomentellous. Leaves alternate, char- taceous, 15-30 x 4.5-9 cm, narrowly obovate to narrowly elliptic, the base acute, the apex acu- minate, acumen to 1.5 cm long, but usually broken off, glabrous above, midrib and to a lesser degree lateral veins tomentellous below, the lamina gla- brous or with a few scattered hairs near the base, venation immersed on upper surface, midrib and lateral veins raised on lower surface, tertiary ve- nation slightly raised; lateral veins 9-12 on each side, arching upward near the margin and loop- connected in the distal half. Petioles 2-3 cm long, + terete, with a similar indument as on twigs. Inflorescences in axils of cataphylls near tip of twigs, brown-tomentellous, 10-20 cm long, the branchlets once or rarely twice cymosely branched, bracts deciduous at anthesis, those subtending buds ca. 2 mm long, tomentellous outside, glabrous in- side. Pedicels 5-10 mm long, brown- tomentellous. Flowers green inside, 8-9 mm diam. Tepals 6, subequal, spreading, the margin plane, the outer 3 broadly elliptic, 4 x 3 outside, glabrous inside, except for some hairs near mm, brown-tomentellous the very base and a narrow, median, papillose strip; inner З elliptic, 3 х 2 mm, the outside with а basal, triangular, glabrous, the inside densely papillose. Stamens 9, 4-celled, the cells lateral, opening back-to-back, filaments about as long as anther, glabrous; glands tomentellous patch, otherwise enlarged, protruding between the outer stamens and touching adjoining gland in front of outer sta- mens, but free. Pistil ca. 2 mm long, the ovary pubescent, gradually narrowed into the short style, stigma enlarged. Floral tube pubescent inside. Fruits unknown. Flowers: April. 100 m Elevation range: ca. Collection studied. BRAZIL. ACRE: Cruzeiro do Sul, vicinity of Serra da Moa, Prance 12392 (F, MO, NY, 5) Pleurothyrium prancei, only known from the type collection, occurs in varzea (flooded forest) and reportedly has stilt roots, very unusual for a Lauraceae. It is characterized by the combination of glabrous, alternate leaves, a pubescent pistil, and flowers with the outer tepals glabrous inside, and inner tepals papillose inside. Pleurothyrium prancei does not seem to have close relatives. Campbell et al. 8057 in Acre, Brazil is possibly this species; the specimen ‚ from the upper Rio Moa is, however, sterile and hence the identification tentative. Itis a pleasure to name this species after Ghillean Prance, who has contributed so widely to the knowl- edge of the Amazonian flora. 100 Annals of the Missouri Botanical Garden Pleurothyrium racemosum van der Werff, sp. nov. TYPE: Panama. Panama: Cerro Jefe, along summit road and trail into Chagres Valley, elev. 900 m, McPherson 12120 (holotype, MO; isotypes, BM, F, НВС, MEXU, PMA) Figures 37, Arbor parva, ad 7 m alta. Ramuli teretes, hornotini Veri aah ton Tepala 6, aequalia, us ca. 4.5 mm lon- ga, rotata vel iu reflexa. ae: bescentibus. G teriora cingentes, liberae. Ovarium anguste a glabrum; stylus glaber, stigmate pe eltato. Tubus floralis profundus, pubescens. Fructus ignoti. Small tree, to 7 m tall. Twigs terete, solid, brown- tomentose when young, the tomentum turning gray on older twigs, the demarcation between brown and gray tomentum sharp, diameter of the young twigs ca. 3 mm. Terminal bud brown-tomentose. Leaves clustered near tips of branches, firmly chartaceous, 6-14 x the base truncate or acute, the tip obtuse or slightly 2.5-4 cm, oblong or somewhat obovate, acute, the upper surface glabrous or with some scattered brown hairs, these concentrated along the base of the midrib and lateral veins, the lower surface with more brown hairs becoming tomentose along the base of the midrib, lateral veins and margin. Venation, including midrib, immersed on upper surface, raised on lower surface; lateral veins 6-9 on each side, curving toward the tip near the margin and venation becoming brochidodromous in the distal half of the laminae. Basal veins leaving midrib under a more acute angle than more distal veins, and situated at the very base of the lamina. Petioles brown tomentose, 9-12 mm long. Inflo- rescences in the axils of cataphylls at the tip of the branches, brown tomentose, 3-7 cm long, race- mose, 4-10 flowers per inflorescence; individual flowers representing a dichasium with the lateral flowers undeveloped, this indicated by 2 bracts at the transition from peduncle to pedicel. Peduncle to 10 mm long, pedicel to 3 mm long, both brown tomentose. Flowers brown outside, creamy-white inside, 9 mm diam. Tepals 6, equal in size, 4.5 mm long, elliptic, spreading to reflexed at anthesis; outer tepals brown-tomentose outside, with a tri- angular, gray-tomentose patch inside, otherwise papillose; inner tepals with a triangular, brown- tomentellous patch, otherwise glabrous outside, gray-tomentellous inside. Stamens 9, all 4-celled, the anthers curved inward, the cells lateral; fila- ments pubescent. Bases of filaments united, form- ing a ridge carrying the free parts of the stamens and the glands. Staminal glands enclosing the outer stamens, but not fused. Ovary narrowly ellipsoid, glabrous, 1 mm long; style glabrous, 1 mm long; stigma peltate. Floral tube deep, densely pubescent inside. Fruits and cupule unknown. Flowers: Feb- ruary. Collection studied. | PANAMA. PANAMA: Cerro Jefe, McPherson 12120 (BM, F, HBG, MEXU, MO, PMA). The type (and only) collection of Pleurothyrium racemosum was made on Cerro Jefe in Panama, a low mountain about 800 m high, which has yielded many endemics. The indument on the lower leaf surface can be rather sparse, especially on older leaves, but remnants of the characteristic curled hairs are always present near the base of the leaf and along the major veins. This species is closely related to P. pilosum, where differences between these species are discussed. Pleurothyrium steyermarkianum Allen, Mem. ew Yor Venezuela. Mérida: La Azulita, forest above (8), Steyermark 56077 (holotye, F not seen; isotype, VEN). Figures 37, Tree, to about 20 m tall. Twigs angular, solid, minutely tomentellous when young, becoming gla- brous with age, 4-5 mm diam. about 4 cm below apex. Leaves rnate, coriaceous, 12-25 X 5- 10 cm, elliptic to broadly elliptic, base acute or obtuse, apex acute or acuminate, glabrous and somewhat shiny above, with scattered, appressed hairs or glabrous below, lateral veins 11-1 each side, curving upward near the margin and weakly loop-connected in the distal half of lamina, midrib and lateral veins immersed on upper sur- face, raised on lower surface, tertiary venation scarcely visible on upper surface, slightly raised on lower surface. Petioles canaliculate, 1.1-2.5 cm long. Inflorescences in axils of deciduous bracts, to 16 cm tellous, long, minutely and laxly brown-tomen- the branchlets 2-3 times cymosely branched, bracts subtending the flower buds broad- ly ovate, boat-shaped, tomentellous outside and glabrous inside. Pedicels to 2 mm long, tomentel- lous, thick, acutely triangular in cross section, the three ribs continuing as keels on the outer 3 tepals. Volume 80, Number 1 van der Werff 101 1993 Revision of Pleurothyrium D FIGURE 38. P. racemosum. — A. Habit. — B. Flower. — C. Flower, side view. — D. Pistil. — E. Stamens I and III. — F. Stamen П. —С. Detail of upper leaf surface. — Н. Detail of lower leaf surface. 102 Annals of the Missouri Botanical Garden FIGURE 39. P. steyermarkianum. — А. Habit. ticulation pattern on leaf. — Н. Part of inflorescence. — Flowers ca. 8 mm diam., pale yellow-green and fragrant fide collectors. Tepals 6, elliptic, 3-3.5 mm long, papillose inside, the outer 3 tomentellous outside, the inner 3 with a central tomentellous strip, otherwise glabrous outside. Stamens 9, 4-celled, the cells lateral; glands enlarged, forming ridges surrounding the stamens, fused. Pistil and receptacle glabrous, ovary ellipsoid. Cupule cup- shaped, 2 cm wide, 1 ст high; fruit ellipsoid, ca. 2 x April. cm. Flowers: 1,300-2,000 m Elevation range: — В. Flower. — C, D. Pistil. I. Underside of leaf base. “3 Detail of lower leaf surface. 23 Ems rd, "m ; үз ШҮ, < ( m | ФА TAIN А mre Je Ra ee : (DS Ц Stamens. — E. —F. Petiole.—G. Re- Collections studied. VENEZUELA. MERIDA: Selva nu- blada de la Zona La Trampa, Bernardi 2169 (С, MER, NY); La Azulita, Steyermark 56077 (VEN). Pleurothyrium steyermarkianum is only known from two collections made in the Andes of Vene- zuela. The floral structure, especially the way in which the enlarged glands form ridges surrounding the stamens, the triangular or three-ribbed pedicels, and broad bracts of the inflorescence suggest a relationship with P. cuneifolium and its allies. How- ever, P. steyermarkianum has a characteristic leaf Volume 80, Number 1 1993 van der Werff 103 Revision of Pleurothyrium indument, lacks fistulose twigs, and differs in floral dimensions from P. cuneifolium. When publishing Pleurothyrium steyermar- kianum (as P. steyermarkiana), Allen indicated that the holotype was deposited in NY. This is an error. The holotype was at F and no isotype was found at NY. The F holotype was sent on loan from F to MO, but unfortunately never arrived and presumably is lost. The only yita of the type collection studied is the VEN iso Allen also indicated the collection ан 2169 had flowers and fruits. All sheets I have seen, however, have floral buds and fruits; thus, the only specimen of P. steyermarkianum with flowers is the VEN isotype. Pleurothyrium synandrum van der Werff, sp. TYPE: Colombia. Valle: Rio Calima, La Trojita, Cuatrecasas 16656 (holotype, F; iso- types, U, US). Figures 37, 40. nov. Arbor, ad 30 m. Ramuli teretes, ferrugineo-tomentu losi, glabrescentes. Gemma terminalis dense imaginado: pubescens. Folia alterna, coriacea, elliptico- ovata vel el- iptica, -8 cm, basi et apice acuta, su liptica, 15-25 x 6-8 t ta, supra glabra, subtus ferrugineo-tomentosa vel Melle aucis g P pilis longioribus secus costam et venas laterales praedita; marginem versus arcuatis, sed non connexis. Petioli l longi, ferrugineo-tomentulosi. Inflorescentiae ex a illis bractearum ortae, 8-15 longae, ferrugineo-to- mentulosae culatae, ramulis 2-3 plo cymae more divisis. Flores pallide flavi, 11-12 mm diametro. Tepala 6, aequalia, 5 mm longa, intus papillosa, exteriora extus — interiora minute puberula praeter basim pm bescentem. Stamina 9, 4-locellata, 6 exteriora duobus locellis extrorsis, duobus lateralibus, 3 interiora locellis iorum dorsali tia. Glandulae dilatatae sed stamina exteriora v bientes. Cupula 4-5 cm late, 2.5 cm alta, verrucosa; fructus ovoideus, 4 x 2.5 cm. Tree, 30 m tall. Twigs terete, solid, the tip with brown slender spreading hairs, brown-tomentulose on older parts, the indument wearing off on older parts, diameter of twig 6-8 mm 5 cm below tip. Terminal bud densely brown pubescent. Leaves alternate, coriaceous, elliptic-ovate to elliptic, 15- 25 х cm, base and apex acute, the upper surface glabrous, lower surface ferruginous-tomen- tose or tomentulose with some longer hairs along midrib and lateral veins; venation immersed on upper surface, midrib and lateral veins raised on lower surface; lateral veins 18-24, closely spaced, arching toward apex near the margin, but not con- nected with their superior vein. Petioles ca. 1 ст d brown-tomentulose, broadly canaliculate, ca. mm diam. Inflorescences in the axils of cata ea 5-15 cm long, paniculate, the branchlets 2-3 times cymosely branched, brown-tomentel- lous, 40-60 flowers per inflorescence, bracts most- ly deciduous at anthesis, ca. 3 mm long, elliptic, brown-pubescent outside, glabrous inside. Flowers i Tepals 6, equal, elliptic, 5 mm long, papillose on inside, outer tepals pubescent outside, inner ones minutely puberulous except for the pubescent base. Stamens 9, 4-celled, mm diam. pale yellow, 11-12 the outer 6 with one pair of cells extrorse, the other pair introrse; inner stamens with extrorse cells; filaments of outer stamens dorsally pubescent; filaments and anthers of inner 3 stamens poorly differentiated, the stamens forming a tight cylinder around the brown-pubescent style. Glands en- larged, protruding between the outer stamens, but these scarcely enclosing, free. Receptacle pubes- cent inside. Cupule, when pressed, 4-5 cm diam., ca. 2.5 cm tall, warty fruit ovoid, 4 x 2.5 cm when dry. Flowers and fruits: March. Elevation range: ca. 50 m. Collection studied. COLOMBIA. VALLE: Rio Calima, La Trojita, Cuatrecasas 16656 (F, U, US). Pleurothyrium synandrum is only known from the type collection from near the coast in Depto. Valle, Colombia. This species can immediately be recognized by its ovate-elliptic leaves with the nu- merous lateral veins close together and its unusual floral structure. In other Pleurothyrium species with strongly enlarged glands, not only do the glands grow outward, but also sideways and somewhat inward and the inner three stamens are therefore separated from each other by glandular tissue. Frequently, the glandular mass covers the upper part of the ovary and surrounds the style, while the lower part of the ovary is sunk in the floral tube. In P. s completely surround the style and the glandular tissue does not take part in protecting the pistil. synandrum the inner three stamens The androecium of this species contains much mu- parts become soft and slimy; I could therefore not make sure whether the inner stamens were actually (partly) fused or merely tightly pressed together. Another unusual feature of P. synandrum is that the basal pair of locelli on the outer stamens has virtually become extrorse. The glandular mass in this species shrinks strongly upon drying, and the outer stamens are raised much higher above the glands in dried flowers than in boiled flowers. Pleurothyrium tomentellum van der Werff, sp. nov. TYPE: Ecuador. Pastaza: Pozo petro- lero **Golondrina" de Petro-Canada, 30 km NW of Curaray, elev. 400 m, tree, 15 m, 10 104 Annals of the Missouri Botanical Garden FIGURE 40. P. synandrum. — A. Habit. July 1989, Gudiño 12 (holotype, МО; types, AAU, COL, C, НВС, K, LE, M NY, QCNE, QRS, US). Figure 41. XU. Arbor, ad 30 m alta. Ramuli teretes, tomentelli. Gemma ranas tomentella. Folia alterna, coriacea -35 6.5-13 cm, (late) elliptica vel oblonga, apice et basi acuta vel obtusa, supra glabra, subtus ferrugineo-tomentella; — B. Flower. lower leaf surface. — С. Detail of upper leaf surface. C. Androecium. —D. Fruit. — E. Cupule. —F. Detail of nervis lateralibus 9- latere, costa nervis- que supra immersis, cin а ма reticulatione subtus paullo elevata, supra non evidente. Petioli 1.7-3.6 cm lo ongi, canaliculati. Inforescentiae in axillis bractearum (raro foliorum) ortae, 7-15 cm longae, tomen albi, 6-8 mm ко Tepala 6, intus papillosa, oblonga, a rma 4- docela, jbaellis lateralibus, flamenta ca. 0.8 mm Volume 80, Number 1 1993 van der Werff Revision of Pleurothyrium 105 | TN | I | | жс uc C 6. ; AAT re с | я | |" el i | == FIGURE 41. Distribution of P. tomentellum (ж), P. tomiwahlii (O), P. undulatum (O), and Р. westphalii (B). gis, glandulis incrassatis, stamina cingentibus, contiguis, non connatis. Ovarium glabrum, in stylo attenuatum; tubo florale intus tomentello. Fructus ellipsoideus, 3 x 1.5 cm, cupula verrucata, 2 cm lata, 1 cm profunda. Tree, to 30 m tall. Twigs terete, smooth or finely ridged, brown-tomentellous, solid, 4-7 mm diam. ca. 5 cm below tip. Terminal bud brown-tomen- tellous. Leaves alternate, coriaceous, 13-35 x .5-13 cm, elliptic, broadly elliptic or oblong, the tip and base acute or obtuse, upper surface gla- brous, opaque, lower surface densely rufous-to- mentellous or coarsely papillose, the surface com- pletely covered by the indument; lateral veins 9- 14 on each side, arching upward near the margin, but not loop-connected; midrib and lateral veins immersed on upper surface, raised on lower sur- face; tertiary venation (very) weakly raised on low- er surface, not visible on upper surface. Petioles 1.7-3.6 cm long, canaliculate to deeply canalic- ulate, with similar indument as twigs. Inflores- cences in axils of deciduous bracts, rarely in axils of leaves, to 7-15 cm long, the branchlets to 4 times dichotomously branched, rufous-tomentel- lous, bracts mostly deciduous at anthesis, if present tomentellous outside and subglabrous or tomentel- lous inside. Pedicels 2-6 mm long, rufous-tomen- tellous. Flowers 6-8 mm diam. Tepals 6, white, papillose-puberulous on inside, oblong, 2.5-3 mm long, equal in size and shape. Stamens 9, 4-celled, the cells lateral, the filaments ventrally pubescent, filaments ca. 0.8 mm long. Glands strongly en- larged, completely surrounding the stamens, but not fused. Pistil glabrous, 1.5 mm long, ovary gradually narrowed into the style, style and ovary of the same length; floral tube densely tomentellous within. Fruit ellipsoid, 3 x 1.5 cm, cupule cup- shaped, мапу, 2 cm wide, 1 cm tall. Flowers: January-July. Fruits: January. ommon name (Huaorani): Ontumo. Collections studied. ECUADOR. MORONA-SANTIAGO: 35 km NE from Montalvo, Zak 4459 (MO, QCNE). PASTAZA: 106 Annals of the Missouri Botanical Garden 30 km NWE of Curaray, Gudino 12 (AAU, COL, C, HBG, LE, K, MEXU, , NY, QCNE, QRS, US); 30 km NE from Curaray, Gudiño 17 (MO, ОСМЕ); 115 km S of Coca, near Rio Tiguino, Hurtado 1464 (MO, ОСМЕ); Via Auca, 115 km 5 of Coca, Neill 8741 (МО, QCNE); 115 km 5 of Coca, near Rio Tiguino, Neill & Hurtado 8783 (MO, QCNE), Rubio 56 (MO, QCNE); 110 km 5 of Coca, 10 km from Rio rud Palacios 3416 (MO, QCNE). PERU. HUANUCO: Pachitea, Pucallpa region, 24 km SE of Puerto Inca, Wallnoefer 18-29788 (MO). Pleurothyrium tomentellum is a species with a limited distribution, known only from the Amazo- nian lowlands of Ecuador, from areas that recently became accessible due to oil exploration, and one collection from Amazonian Peru. According to the label data of Neill & Hurtado 8783, the species is locally very common. Pleurothyrium tomentel- lum can be readily identified by the combination of its tomentellous leaves and small (6-8-mm-diam.) flowers. Noteworthy are also the glabrous pistil, pubescent receptacle, and rather small cupule. In leaf characters (texture, indument, lack of marginal vein) and its short inflorescence it resembles P. palmanum from Costa Rica, but that species has larger flowers, broader, more or less obovate leaves, and a glabrous receptacle. Most collections have tomentellous lower leaf surfaces and inflorescence The type collection has a shorter indument on the leaves, bracts that are tomentellous on both surfaces. approaching the papillose condition, and the inner surface of the inflorescence bracts is subglabrous. In floral characters the collections are identical (only the sole Peruvian collection has a sparser indument on inner surface of tepals), and the dif- ferences mentioned are too weak for recognition of two taxa. Moreover, the type was collected to- gether with a specimen with the slightly longer indument (Gudino 12, 17). Pleurothyrium tomiwahlii van der Werff, sp. nov. TYPE: Ecuador. Los Rios: Centinela Ridge, elev. 600 m, 16 July 1991, et al. 12365 (holotype, MO; isotypes, AAU, B, BR, C, COL, E, F, G, GH, HBG, К, KUN, LE, MEXU, MY, QCNE, QRS, S, SEL, U, US). Figures 41, 42. , 20 m. Ra muli crassi, teretes, juvenales ferru- з. Gemma ter- 1 rotundatis, obtusis vel breviter acuminatis, (late) obovatis vel (late) ellipticis, шше glabris vel pau- cis pilis erectis praeditis; nervis ostae latere. Petioli 2 Ф w "E о Ф e panicu mosis, 7-12 cm longae, b Tepala 6, aequalia, extus pubescentia, 3 interiora intus van der Werff pubescentia, 3 exteriora intus praeter partem triangula- hesi reflexa, ca. sensim in attenuatum; е аан et tubo filamentorum intus pu- bescente. Frutus ignotu Tree, to 20 m tall. Twigs + terete, solid, with pale bark, the young growth ferruginous-tomen- tose, soon glabrescent, ca. 8 mm diam. 5 cm below tip. Terminal bud brown tomentose. Leaves clus- tered near tips of branches, alternate, chartaceous, laminae 11-40 x cm, the base obtuse or acute, the tip rounded, obtuse or shortly acuminate, (broadly) obovate, or (broadly) elliptic; glabrous on both surfaces or with scattered erect hairs, these mostly along the major veins or near base of the lamina; lateral veins 8-10 on each side; venation immersed on upper surface or weakly raised, midrib and lateral veins raised on lower surface. Petioles 2.5-6 cm long, strongly canaliculate, the base swollen, with similar indument as on the twig. In- florescences in the axils of cataphylls near the tips pn paniculate, the lateral of branches, t y branched, 7-12 cmlong, hranchle bracts EE ER at кина. Flowers cream-col- ored, fragrant. Tepals 6, equal in shape, densely pubescent outside, the inner three densely pubes- cent inside, outer three with a basal, triangular, pubescent patch, otherwise glabrous or finely pa- pillose; at anthesis reflexed, ca. 4 mm long, elliptic. Stamens 9, all 4-celled, the cells lateral, filaments of the outer 6 stamens dorsally pubescent; glands at base of inner stamens strongly enlarged and separating the outer stamens from each other; bas- es of filaments fused into a pubescent column, ca. | mm long. Ovary slender, glabrous, gradually narrowed into the style; inside of the floral tube and column formed by the stamens densely pu- bescent. Fruit unknown. Flowers: July and Decem- ber. Elevation range: 600-1,400 m. Collections studied. COLOMBIA. ANTIOQUIA: Amalfi, Salazary Marengo, Callejas 9143 (MO). ECUADOR. LOS RIOS: о В, BR, С, MO, NY, ОСМЕ ORE < S, SEL, U, US, ander Werf et al. 12367 (AAU, QRS, МО, ОСМЕ). PICHINCHa: Cima de las NE de Па, Dodson 14531 (MO). Pleurothyrium tomiwahlii is readily identified by its clustered, obovate leaves, long petioles with swollen base, its flowers with reflexed tepals, and the base of the filaments united in a column, giving к flower the appearance of a gonoloboid Ascle- . Its closest relatives are the Costa Шош F. ДОИШ ense, in which the androecium Volume 80, Number 1 van der Werff 107 1993 Revision of Pleurothyrium E x FIGURE 42. Р. tomiwahlii. — A. Habit. — B, C. Flower sideways and seen from above. — D. Part of inflorescence. — E. Leaf base. — Е. Shade leaf, showing less obovate shape and more acute leaf base. 108 Annals of the Missouri Botanical Garden forms a dome (without a column) and in which the glands separate the outer stamens just as in P. tomiwahlii; and the Panamanian species P. ra- cemosum and P. pilosum, with similar floral struc- ture as in P. golfodulcense, and with relatively long petioles, obtuse leaf bases and apices, and clustered leaves as in P. Details of indument, leaf shape and size as mentioned in the tomiwahlii. species descriptions, as well as the reflexed tepals, characterize P. tomiwahlii. There is considerable variation in leaf size among the few available collections, probably a result of exposure to sunlight. The type tree, with rather small leaves, was free-standing in a pasture, while van der Werff et al. 12367 was ап understory tree in a forest patch. The Dodson collection has the largest leaves and was growing in forest. Pleu- rothyrium tomiwahlii is a rare species; during a day-long search at Centinela only three individuals were seen. Although the single Colombian collec- tion represents a great range extension, wahlii is such a distinctive species that I have no doubt about its identification. The specific epithet honors the Wahl family, in P. tomi- recognition of their keen interest in conservation of tropical forests. Pleurothyrium trianae (Mez) Rohwer, Mitt. Inst. Bot. Hamburg 20: 43. 1986. Nectandra trianae Mez, Jahrb. Bot. Gart. Berlin 5: 439. 1889. туРЕ: Colombia. 7riana 1037 (ћојо- type, P not seen; isotypes, BM, NY). Figure 43 Pleurothyrium zulianense Lasser, Bol. Tecn. Minist. Agric. a 18. 1942. Octoea | (Lasser) к ans, J. Sci. Res. (Jakarta) 1: 122. 1952. ld ulia: en selvas dein del Rio Lora. Pittier 10947 (holotype, VEN not seen; isotypes, 5). me i Plewrothyrium Кл Lasser, Bol. Tecn. Minist. Agric. ЗНА | a reflexa (Lasser) Kostermans, J. а Res WO 122 2 TYPE: Venezuela. e San Isidro rico, Tamayo 1094 re VEN not seen; nares F, US). Large trees, becoming more than 20 m tall. Twigs roundly angular or terete, solid (rarely fis- tulose), glabrous, but the tips minutely brown-to- mentellous, ca. 3 mm diam. у ст REGN the apex. Terminal buds minutely brown 5. Leaves alternate, chartaceous, elliptic, 9-25 x 3- 10 em, the upper surface glabrous, the lower surface mi- nutely papillose to glabrous, the base acute, the tip shortly acuminate. Lateral veins 8-12 on each side, arching upward near the margin, not loop- connected, venation immersed on upper surface, midrib and lateral veins raised on lower surface. Petioles 10-20 mm long, shallowly canaliculate. Inflorescences in axils of cataphylls, minutely brown- tomentellous, paniculate, the twice cymosely branched, bracts deciduous at an- ranchlets once or thesis. Pedicels 2.5-4.5 mm long, minutely brown- the tepals at anthesis erect or halfway spreading. Tepals tomentellous. Flowers white, ca. 4 mm diam., 6, equal, ca. 2 mm long, the margin reflexed, minutely tomentellous outside, papillose inside. Sta- mens 9, glabrous, 4-celled, the cells mostly lateral and opening upward; filaments about as long as the anther, glands enlarged, visible as large lobes between the outer stamens, free. Ovary ellipsoid, glabrous or slightly papillose near the tip, style papillose, pistil 1.5-2.0 mm brous inside. Cupule warty, t mm tall, the pedicel not d oct. fruit. ellipsoid, o х 1 collected in every month except June-July. Ele- vation range: 10-1,400 m. — mm. Fruits: February-June. Flowers: Collections studied. | COLOMBIA: Prana 1037 (BM, Y). ANTIOQUIA: Frontino, Murri, 20-28 km from Nu- tibara, McPherson 12999 (MO). сносо: ui ebad Ta- paral, hoya del Rio San Juan, Forero 4135 (МО); Ta- paralito, Oüebrada Taparal, N of Palestina, Gentry 53812 NA Gentry 53833 (MO); Palo Gordo, Triana 2042 (K); basin of Rio San Juan, tributary Rio Taparal, van Rooden 650 (MO, U), van Rooden 681 (MO, U); hoya del Rio San Juan, Rio Bicordo, arriba de Noanama, Forero 4693 (MO); idee Rio Cabi, Prance 28012 (NY). МОВТЕ DE SANTANDER: re, Rio Cu bugon, junto a la quebrada de Gibraltar, aa 13227 (F, US). Calima, La Trojita, a: 16321 trecasas 163214 (F, U, US), Cuatrecasas 16781 (F, Gentry 53758 (MO (F), Cuatrecasas 16864 (F, US). Costa Rica. ALAJUELA: iles, Zamora Palenque Biological Center, MORONA-SANTIAGO: Pozo Petrolero * CO, Zak 4608 (MO). NAPO: Reserva Florestica El Ahuano, Estación INIAP, Palacios 2094 (MO); 3 km from en- trance to Jatun Sacha toward Rio Arajuno, Palacios 2804 (MO); Archidona, S sl , Comunidad EI Pacto, Palacios 4727 (MO): Tena, Estac ción Biológica Jatun Sacha, Cero ), Palacios 4273 (MO), Palacios M И US S). 1 ve, Vásquez 1 О). Eee est t Reserve, Breteler n (NY, U, US) El Vigia, Мос querys 1002 (K, NY, x m iA banks of Rio Bum dau d. 1584 (N ); pos Ciudad Bolivia [= raza), Aristeguieta pi cela VEN); Reserva Forestal de Ticoporo, Aristeguieta 6978 (VEN); sinas La Vieja, io Pedraza, Bernardi 1137 (С, МО); Res serve, bank o . Pedraza, Dorr 4720 (MO, NY, PORT); Reserva Forestal de Caparo, Jimenez 1289 (NY), Marcano 2913 Volume 80, Number 1 1993 van der Werff 109 Revision of Pleurothyrium T O 200 400 600 800 1000km ==) ppp O 100 200 300 400 $00 800 miles Е \ СЯ 44 AR T--..----..-.2..d....... FiGURE 43. Distribution of P. trianae (X) and P. williamsii (O). (BR, С, MO); Caserio Barragan, Salcedo 225 (С, US); Cacao, Barinitas, Valverde 15 (MO). LARA: Tamayo 1094 (F, US); Iribarren, Parque Nacional Terepaima, Fila San Esteban, Smith 7463 (MO); Iribarren, Parque Nacional Terepaima, Represa Rio Claro, Smith 9021 (МО); Iri- barren, Loma de los Naranjos, Montana de Macanillal, Steyermark 111580 (С, NY, U, VEN). MERIDA: Little 15822 (NY), Marcano 1177 (MO). PORTUGUESA: Guan- are, 17 km NW of Tucupido, N slope of Fila Las Palmas, Davidse 21459 (MO, NY), Davidse 21467 (MO, NY). PORTUGUESA / BARINAS: Represa de Bocono, Aymard 1747 O). TA ukos, Perija, 2085 (NY); road Machiques- Colon, near Rio Caratumbo, de Bruijn 1422 (MO, NY, US, VEN); along Rio Lora, above Camp 2, Perija, Pittier 10947 (G, GH, US). Pleurothyrium trianae is, as treated here, a wide-ranging and variable species. It has been re- corded from Honduras to Ecuador and Venezuela, but the records are not evenly distributed over this area. Only two collections are known from Central America. Nearly all South American collections come from the following three areas — (a) Pacific lowlands of Colombia (type locality of P. trianae), (b) Andean foothills in Venezuela (type localities of P. zulianense and P. reflexum), and (c) the Oriente of Ecuador, while almost no collections are known from the intervening areas. It is therefore not sur- prising to find local differentiation. The Colombian collections tend to have almost glabrous leaves, a rather sparse indument on twigs, and pedicels that are 3.5-4.5 mm long. The Venezuelan plants have more indument on twigs and leaves and have ped- icels 2-2.5 mm long. The Ecuadorian plants agree with the Colombian specimens in their indument, have variable length of pedicels (2-4 mm), and have smaller leaves (9-14 x 3-5 cm) than the other populations (Palacios 2094 has larger leaves). These differences are nearly all quantitative and form an insufficient basis for the recognition of separate taxa. The length of pedicels cannot be relied on for the separation of taxa because the lateral flowers of a cyme nearly always have shorter pedicels than the central flower and young flowers tend to have slightly shorter pedicels than old flow- 110 Annals of the Missouri Botanical Garden ers. I expect that once collections of P. trianae become available from other areas, the differences between populations as described above will dis- appear. For example, a collection from Antioquia at 1,050 m elevation (McPherson 12999) has the denser indument and shorter pedicels commonly found in Venezuelan collections; it also has stiffer leaves than usual for P. trianae. The relationships of P. trianae are with a small group of poorly defined species that is characterized by small flowers with erect, more or less inrolled This roup consists of P. acuminatum, P. amapaense, tepals, solid twigs, and chartaceous leaves. P. amplifolium, P. intermedium, and P. undu- latum. Pleurothyrium acuminatum differs in its relatively long pedicels ((4-)5(-8) mm), and long tepals (3 mm long); P. amplifolium has broad leaves, while P. intermedium, P. undulatum, and P. amapaense have a different indument. It is possible that, once more material is available, some of these species will be placed in synonymy. More distantly related are P. parviflorum and P. poeppigii, which both have fistulose twigs. Ad- ditionally, P. parviflorum differs in its glabrous flowers and P. poeppigii in its coriaceous leaves. A few collections of P. trianae have fistulose stems, and the differences cited above help separate these specimens from P. parviflorum and P. poeppigii. Differences between P. trianae and P. cuneifolium are discussed under the latter species. Pleurothyrium undulatum (Meissner) Rohwer, Mitt. Inst. Allg. Bot. Hamburg 20: 44. 1986. Fari undulata Meissner, DC. Prodr. 15(1): 864. Ocotea undulata (Meiss- ner) Ms prs Bot. Gart. 5:218. 1889. TYPE: Brazil. Amazonas: Prov. Rio Ne- gro, Riedel 1412 (holotype, LE not seen; iso- types, B, G, K, L, NY). Figure 41 Berlin Tree of unknown size. Twigs terete, solid, mi- age. Leaves alter- utely brown-tomentellous, glabrescent wit Terminal bud brown-tomentellous. nate, chartaceous, elliptic to narrowly elliptic, 10— тх cm, base and tip acute, glabrous on upper surface, glabrous or minutely papillose on lower surface, venation immersed on upper sur- ће lateral veins (12-15 on each side) poorly visible and less face, midrib raised on lower surface, elevated; marginal vein poorly developed or lack- ing. Petioles ca. 1.5 cm long, with similar pubes- cence as twigs. Inflorescences in axils of cataphylls or leaves, brown-tomentellous, especially the ulti- mate branches, 8-12 cm | the branchlets 2—3 times cymosely branched, the flow- ong, paniculate, ers close together; bracts present, ovate, to 2 mm long, tomentellous outside, minutely so inside. Ped- icels to 1 mm long, tomentellous. Flowers ca. 3 mm diam., the tepals + erect at anthesis. Tepals 6, equal, ca. 2 mm long, tomentellous outside, papillose inside, the margin reflexed. Stamens 9, 4-celled, the cells lateral, glands enlarged and pro- truding between the outer stamens, but not en- closing them and not fused. Ovary globose, gla- brous, enclosed by the glabrous receptacle; style papillose. Fruits unknown. Collection studied. BRAZIL. Riedel 1412 (В, С, К, NY Pleurothyrium undulatum is only known from the type collection made by Riedel in 1824-1825 Meissner (1864) gave as locality only Amazonas, Prov. Rio Negro; the isotype in NY gives as locality Barra, an old name for Manaus. Pleurothyrium undulatum was placed by Mez (1889) in Ocotea; e considered that the presence of glands at the base of all stamens was typical for Pleurothyrium, and although he noted that in respect of anther shape this species resembled Pleurothyrium, he treated it in Ocotea because of the presence of only six glands. Pleurothyrium undulatum belongs to a small group of species characterized by tepals with reflexed margins, often erect tepals at anthe- sis, marginal vein lacking or weakly developed, small flowers (less than 5 mm diam.) and inflores- cences sometimes present in axils of leaves and not restricted to axils of cataphylls. Within this group, the narrow, almost glabrous leaves and the to- mentellous pubescence of the inflorescences char- acterize this species. Pleurothyrium vasquezii van der Werff, sp. nov. TYPE: Peru. Loreto: Maynas, km 32 along carretera Iquitos-Nauta, elev. 150 m, primary forest, Vasquez & Jaramillo 7889 а МО; isotypes, AAU, AMAZ, Е, С, NY). Fig ures 44, 45. , 20 m alta. Ramuli teretes, solidi, fusco- tomen- a vel late elliptica, 15-30 x 5-10 (rotundata), apice acuminata, supra glabra, subtus pilis erectis praedita, venatione supra immersa, subtus elevata, in dimidio bois Mi eru venis ea 9-13 ra la costae latere. Petioli 1 m longi, canaliculati. Inflorescentiae ex bractearum "axillis ortae (raro ex folio- rum axillis). fusco. tomentosae, 10-17 cm longae, panicu latae. Pedicelli 3-4 m ia rotati. Te 2 mm long и plano, extus tomentella, intus papillosa. Tubus oralis ad ostium constrictus. Stamina 9, 4-locellata, lo- и lateralibus vel extrorso-lateralibus, minute papillosa; glandulis magnis, staminibus exterioribus cingentibus, libe- Volume 80, Number 1 1993 van der Werff 111 Revision of Pleurothyrium NV. м; [ | | А о 200 400 600 800 1000кт SS Se о 100 200 300 400 500 600 miles \ 100 9 FIGURE 44. Distribution of P. vasquezii (X). ris. Ovarium ellipticum, papillosum, ca. 1 mm longum, in receptaculo profundo, papilloso inclusum. Cupula ver- rucosa, 1 cm alta, 1.3 cm lata. Tree, to 20 m tall. Twigs solid, terete, fuscous- tomentose, glabrescent, 3-4 mm diam. 5 cm below the tip. Terminal bud fuscous-tomentose. Leaves alternate, stifly chartaceous to coriaceous, elliptic to broadly elliptic, 15-30 x 5-10 cm, the base acute to obtuse, rarely rounded, the tip shortly acuminate, the upper surface glabrous and with immersed midrib and lateral veins, the lower sur- face with erect hairs, the surface well visible, the elevated midrib and, to a lesser degree, lateral veins tomentose, tertiary venation also raised, lateral veins 9—13 on each side, curving upward near the mar- gin, a marginal vein quite well developed in the distal half. Petioles 1.5-3 cm long, canaliculate, with same kind of tomentum as twig. Inflorescences in axils of cataphylls, rarely in axils of leaves, fuscous-tomentose, 10—17 cm long, paniculate, the branchlets 2-3 times cymosely branched, many- flowered, bracts lacking at anthesis. Pedicels 3-4 mm long, tomentose. Flowers yellow or green, 5- 6 mm diam., the tepals spreading at anthesis. Tepals 6, equal, oblong, ca. 2 mm long, the margin plane, tomentellous outside, papillose inside. Floral tube constricted just below tepals. Stamens 9, 4-celled, the cells lateral or one pair lateral-extrorse, opening back-to-back, filaments and anthers minutely pap- illose; glands enlarged, almost completely enclos- ing the outer stamens, but not fused. Ovary elliptic, papillose, ca. 1 mm long, enclosed in the papillose, deep receptacle, the upper part covered by the enlarged glands, style ca. 0.5 mm long, papillose; stigma not enlarged, partly hidden by the longer stamens. Cupule cup-shaped, warty, ca. 1 cm tall, 1.3 cm wide (but probably not fully mature). Fruits: February. Flowers: August-October. Elevation range: 100-600 m Collections studied. BRAZIL. ACRE: Rio Jurua, 1 km upstream from Colonia Rodrigues Alvez, Campbell 10852 (MO, NY); Reserva INCRA Santa Luzia, km 40 BR-364, 112 Annals of the Missouri Botanical Garden 3 ote T seeps: 7 ELA PADO A SES » ле? EET, X МА re “д \ AA IDA DW OA P UIS ARR e с? FIGURE 45. suríace. P. vasquezii. — А. Habit. — B. Part of inflorescence with flowers. — C. Fruit. — D. Detail of lower leaf Volume 80, Number 1 1993 van der Werff 113 Revision of Pleurothyrium Campbell 6887 (MO, NY). AMAZONAS: Mun. de Пар: tos- Naua, km 32, Vásquez & Jaramillo 7889 (AAU, AMAZ, F, G, Herrera, without data (MO). of Puerto Maldonado, Smith & Jaramillo 384 (MO, US); Manu, Parque Nacional Manu, Rio Sotileja, Foster 11719 (MO); Tambopata, Tambopata Nature Reserve, Gentry 58103 (MO). SAN MARTIN: Mariscal Caceres, Tocache Nuevo, Quebrada de Mantencion, Schunke 13643 (MO); Nacional Aléxauder von Humboldt, Oliveira 21 (MO) Pleurothyrium vasquezii is known from several collections in Amazonian Peru and adjacent Brazil, as far downstream as Manaus. It is the only Pleu- rothyrium species in this area with a sparse, erect pubescence on the lower leaf surface and alternate leaves. Pleurothyrium brochidodromum has a very similar indument on the lower leaf surface, but differs in having larger flowers (9-10 vs. 5-6 mm diam.) and in its clustered leaves. A few collections placed in P. vasquezii were earlier identified as P. panurense (Allen, 1964), but this species has clus- tered leaves, a very short, appressed indument on the lower leaf surface that completely covers the surface, and flowers ca. 8 mm in diameter Two Froes collections (20526, 21235) from Brazil have somewhat more coriaceous leaves and roundish leaf bases. However, these collections are fruiting, which may explain their stiffer leaf tex- ture. Because Ramirez 61 from Peru also has roundish leaf bases and agrees in floral characters with P. vasquezii, I placed the Brazilian collections here. Spichiger et al. (1989) reported this species as Nectandra amplifolia Мег (= P. amplifolium (Mez) Rohwer) from the Arboretum at Jenaro Her- rera, Peru (tree 159 in parcela 1). This latter species differs in its puberulous indument on the lower leaf surface, and in its more or less erect tepals with reflexed margins. The same tree is also cited under Ocotea undulata (Meissner) Мег (= P. undulatum (Meissner) Rohwer), which also has a different indument on the lower leaf surface, and more or less erect tepals with reflexed margins at anthesis. Pleurothyrium vasquezii is named after Ro- dolfo Vasquez, whose fine collections have added much to our knowledge of the trees in Amazonian Peru. Pleurothyrium westphalii van der Werff, Ann. Missouri Bot. Gard. 74: 410. 1987. TYPE: Guatemala. Alta Verapaz: Sacté, Kunkel 9 (holotype, BR). Figures 41, 46. Tree, to 20 m tall. Twigs solid, terete, young ones with brown, dense appressed pubescence, old- er ones glabrous. Terminal bud densely brown- pubescent. Leaves alternate, membraneous, 15- X 4–7 cm, elliptic or narrowly elliptic, the base acute, the margin somewhat decurrent, apex acute, glabrous and gland dotted on the upper surface (except for some appressed hairs near the base of the midrib), lower surface with few appressed hairs, more so near the base; lateral veins 5-9 on each side, arching upward near the margin, but not loop- connected; venation immersed on upper surface, midrib, lateral veins and tertiary venation weakly raised on lower surface. Petioles to 1 cm long, with similar pubescence as twigs. Inflorescences in axils of deciduous bracts, attached below the leaves, ca. 8 cm long, densely gray-brown pubescent, the low- est branches once cymosely branched, otherwise inflorescence racemose. Flowers ca. 8 mm diam. Pedicels ca. 4 mm long. Tepals 6, equal, ovate, 3-4 mm long, densely pubescent on both surfaces. Stamens 9, 4-celled, the cells lateral, anthers gla- brous, filaments brown-pubescent; glands strongly the glabrous receptacle; stigma platelike. Fruits unknown. Flowers: April. Elevation range: » 1,100 m. Collections studied. | GUATEMALA. Sacte, Kunkel 9 (BR), Kunkel 17 (MO) ALTA VERAPAZ: Pleurothyrium westphalii, known only from two collections in Guatemala, is the northernmost rep- resentative of the genus. The species can be readily identified by its appressed, gray-brown pubescence, the pubescent tepals (also inner surface is pubes- cent), pubescent filaments, the few-flowered inflo- rescences, the membraneous leaves, and the ab- sence of a marginal vein. Pleurothyrium williamsii O. C. Schmidt, Re- pert, Spec. Nov. Regni Veg. 31: 189. 1933. Ocotea williamsii (О. C. Schmidt) Koster- mans, J. Sci. Res. (Jakarta) 1: 122. 1952. TYPE: Peru. Loreto: Pebas on the Amazon River, Llewelyn Williams 1766 (holotype, F probably lost; isotype, B, fragment С). Figure 43 Shrub or small tree, to 8 m tall. Twigs angular, solid, becoming terete in age, brown-tomentellous to glabrous, 4-5 mm diam. 5 cm below apex. 114 Annals of the Missouri Botanical Garden FIGURE 46. P. westphalii. Habit and flower. Leaves clustered, rarely alternate, chartaceous, 30— 1 ст, oblanceolate or narrowly obovate, the base cordate, the tip acuminate, acumen to 3 cm long, both surfaces glabrous, but with varying amounts of minute hairs on midrib and lateral veins; venation immersed on upper surface, midrib and lateral veins raised on lower surface, the tertiary prominently loop-connected in the upper 25 of the lamina. Petioles to 8 mm long, with similar indu- fos me ASA aces ARRIT ~ 5 AM о 3 3 AU EN ment as twig. Inflorescences in axils of deciduous bracts, pendent, to 20 cm long, rufous-tomentel- lous, the branchlets 2-5 times cymosely branched, bracts present at anthesis, ovate, tomentellous, to 4 mm long. Flowers white, 8-9 mm diam. Pedicels 3 mm long. Tepals 6, subequal, the outer three slightly wider than the inner three, the outer three tomentellous outside, inner three with a basal tri- angular tomentellous patch, otherwise papillose, inside of all tepals papillose. Stamens 9, 4-celled, the cells lateral, anthers minutely papillose, glands Volume 80, Number 1 1993 van der Werff 115 Revision of Pleurothyrium of stamens greatly enlarged, surrounding the sta- mens, fused. Ovary and style densely brown-pa- pillose, floral tube brown-papillose inside; ovary globose, ca. 0.7 mm long, the style distinct, ca. 0.5 mm long. Cupule of young fruit cup-shaped, ca. 2 cm wide and 1 cm tall; young fruit ellipsoid, ca. 1.5 ст long. Fruits: July-October. Flowers: June-July. Collections studied. ECUADOR. NAPO: Aguarico, Re- serva Faunistica Cuyabeno, Palacios 7667 (МО). PERU. 1766 (B); Maynas, Iquitos, Asociación Agraria es Vásquez 10877 (MO); Maynas, Explornapo Camp, Sucursari, Vasquez 8119 (MO), Vasquez 1307 8 (MO). Pleurothyrium williamsii is only known from the type collection, four collections, all from Peru in the area north of the Rio Napo-Rio Amazonas, and one collection from Ecuador. Unfortunately, the holotype, which was requested from the Field Museum, disappeared, together with other Pleu- rothyrium types, while being sent to St. Louis. Of the type collection, only a duplicate in B (with buds) and some inflorescence fragments in G (with a few flowers) exist. The recent collections are a good match as far as floral characters and leaf shape are concerned, but there are some differ- ences. The B specimen has alternate leaves, while the recent specimens and the photo of the holotype (in NY and F) show clustered leaves. The photo- types have the inflorescences alternate along a leafless twig, the recent specimens have the inflo- rescences near the tip of the stem, while the B specimen has a detached inflorescence. The B spec- imen is also more tomentulose. However, the sim- ilarities in leaf shape and flowers outweigh these differences, and I do not hesitate to assign the two recent collections to P. williamsii. A close relative of this species is Pleurothyrium panurense (Meissner) Mez, a species with elliptic- obovate leaves, an obtuse leaf base and similar flowers. It differs, however, in its smaller leaves, the whitish indument on the lower leaf surface, its glabrous ovary, and longer petioles. Pleurothyrium insigne differs from P. williamsii in из larger leaves, the leaves not so gradually narrowed toward the base, erect indument on lower leaf surface, and its larger flowers. All three species occur in Am- azonian Peru and/or adjacent Brazil. Another close relative is P. maximum, which see for further discussion. IMPERFECTLY KNOWN SPECIES Among the collections were found a number of specimens that do not belong to any of the treated species and which very likely represent undescribed species. Because these specimens are incomplete (sterile or fruiting), they are not formally described, but only listed below so as to call attention to their existence. 1 hope that in the near future material adequate for their description will become avail- able. G. Proctor Cooper 539, Panama, Bocas del Toro, region of Almirante (F, NY, A fruiting collection with large (to 35 cm), el- liptic to elliptic-oblong, acuminate leaves. Leaves are glabrous below and have a strongly developed marginal vein; the twigs are solid, glabrous or near- ly so. A distinct species, included in Burger & van der Werff (1990) as Pleurothyrium sp. A. Gentry 57004, Colombia, Valle, Bajo Calima (MO). A sterile specimen with gigantic leaves, accord- ing to the label ca. 1 m long. The specimen has leaves 60-70 cm long and 30-35 cm wide, densely rusty-tomentose below. Monsalve 1651, Colombia, Valle, Bajo Calima (MO) A species with clustered leaves, dark ferrugi- nous-tomentose below, ca. 15 X 6 cm. Young inflorescences and infructescences are very short, ca. 1 ст long, and carry only one bud or fruit. There are five collections of this species, but none with flowers. López & H. Triana 24, Colombia, Antioquia, Par- que Nacional de las Orquideas (MO). Characterized by its obovate to obovate-elliptic leaves, glabrous below, with 15—20 pairs of lateral veins and an obtuse to rounded leaf base. The young cupules are covered with many small len- ticels. Vasquez 3220, Peru, Loreto, Requena (MO). Leaves glabrous, whorled, narrowly oblong with abruptly rounded base. Related to P. williamsii, but with narrower, oblong leaves and much smaller inflorescences. EXCLUDED SPECIES Pleurothyrium bahiense (Meissner) Barroso = Ur- anodendron bahiense (Meissner) Rohwer Pleurothyrium chrysothyrsus Meissner = Rho- ostemonodaphne Pleurothyrium cowanianum C. K. Allen = Rho- dostemonodaphne kunthiana (Nees) Rohwer 116 Annals of the Missouri Botanical Garden Pleurothyrium ferrugineum Meissner = Ocotea arnottiana (Nees) van Pleurothyrium glandulosum (Lundell) Lundell = | ndra Pleurothyrium velutinum Meissner = Ocotea cal- ophylla Mez LITERATURE CITED йш. Ck, . Botany of the Guayana Pow . Lauraceae. Mem. New York Bot. Gard. 10: 4 MER Contributions to the botany a a ee aioe Mem. New York Bot. Gard. . 1966b. Notes on the Lauraceae of tropical America. I. The generic status of Vectandra, Ocotea and Pleurothyrium. Phytologia 13: 221-231. i a des Lauracées. Li- brairie Hachette et Cie., BENTHAM, G. & J. D. Hooker. rum 3: 146-164. London d С. а. Andes, Mérida, Venezuela. BURGER, W. € H. vaN DER Wrnrr. 1990. Flora dd taricensis: Lauraceae. Fieldiana Bot. 23: 1-12 Dopson, C. Н. & А. H. GENTRY. 1978. Flora A iis Rio Palenque Science Center. Selbyana 4: 1-628 DUCKE, A. Plantes nouvelles ou peu connues de la région amazonienne IV. Arch Jard. Bot. Rio de Janeiro 5: 101- : Farris, J. S. 1988. Hennig86. Version 1.5. Documen- tation Hennic, W. 1966. Phylogenetic Systematics. U Illinois Press, Urbana. T vu. J. 1964. The Genera of Flowering Plants, . Clarendon Press, Oxford. Неа. к Р.М. 19 А revision of Lauraceae in Australia (excluding Cassytha). Austral. Syst. Bot. 2: 135-367. к A. J. G. Н. 1952. A historical ird of ceae. J. Sci. Res. (Jakarta) 1: 113-127. 1997 Lauraceae. Commun. “е Кез. 1пзї. 1880. Genera Planta- Universidad de Los — niv. 57: 1-64. б К. 1981. A note оп /Vectandra (Lauraceae), with the description of a new species. Acta Amazonica i 307-308 Tecn. Minist. ela 3: 7 = .L.&R 1969. Arboles communes de la Provincia de Esmeraldas. FAO, Ror dcn D. J. 1987. The Plant-book. Солы ress, Cambridge. Mac Биг, Е. 1938. Lauraceae. In: pu of Peru. Field at. a te Ser. 13(2): 8 Meissen. C. uraceae. [a ES Candolle, Prodromus Systema Naturalis Regni Vegetabilis ISP 1-2 Mrz, C. T pm pue Jahrb. Kónigl. Bot. Garten Berlin 5: NEES VON ESENBECK, C. C. D. RM Pleurothyrium. In: J. Lindley, A Natural System of Botany. London 36b. Systema Laurinarum. Berlin. 1848. Laurinae. In: J. F. Klotzsch, Beiträge zu einer Flora der Aequinoctial-Gegenden der neuen Welt. Linnaea 21: 487-526. Pax, F. 1889. Lauraceae. /n: Engler & Prantl, Na- türlichen Pflanzenfamilien 3(2); 1 126. Ras, B. & Н. VAN DER WERFF. 88. А contribution to ба pollen morphology of у ан Lauraceae . Missouri Bot. Gar 67 homer J. С. o einer Monographie der Gattung Ocotea Aubl. (Laura eee). sensu lato Mitt. Inst. Allg. Bot. Hamburg 20 7 & K. Kupirzki. 1985. а. im ка -Komplex (Lauraceae). Bot. Jahrb. 107: 12 Em О. C. 1928. Lauraceae. In: J. Mildbread, Plantae Tessmannianae о VII. Notizbl. Bot. Gart. Berlin-Dahlem 10: 225-236 1933. Beitrage zur Kenntnis gi Andinen sidamerikanischen Lauraceen I. Report. Spec. Nov. Regni Veg. 31: ur A. c 35. Plantas krukovianas IV. Phytologia ч. HIGER, jy J. Меког, P. Loizeau & L. STUTZ DE ORTEGA. 1989. Contribución a la Flora de la Ama- zonia Peruana. Boissiera 43: 180-218 WEBERLING, F. 1985. Zur br _____ der Lauraceae. Bot. Jahrb. Syst WERFF, Н. VAN DER. 1987. Six new м of neo- nae oo Ann. Missouri Bot. Gard. 74: 0l- | n . Eight new species and one new со bination erar Lauraceae. Ann. Missouri Bot. Gard. 75: 402-4 А E to the genera of Lauraceae in 1991. ne New World. Ann. Missouri Bot. Gard. 78: 377- Wiuus, J C. 1973. A Dictionary of the Flowering Plants and Ferns, ed. 8, revised 7 К. Airy-Shaw. Cambridge Univ. Press, Cambridge List of species recognized. New species are in boldface. Pleurothyrium acuminatum van der Werff Pleurothyrium amapaense Allen Pleurothyrium amplifolium (Mez) Rohwer Pleurothyrium bifidum Nees Pleurothyrium bracteatum van der Werff Pleurothyrium brochidodromum van der Werff —_— — — . Pleurothyrium golfodulcense Burger & Zamora . Pleurothyrium grandiflorum van der Werff : Pleurothyrium a E van der Werff . Pleurothyriu an der Werff 18. Pleurothy rium niermedium (Mez) Rohwer ale van der Werff — = = — — гор оо О о оо оо о о x — Е [и | 22. Pleurothyrium obovatum van der 23. а palmanum (Мег & J. D. үле Roh- 24. Pleurdih 'rium panurense (Meissner) Mez 29. АА parviflorum Ducke 26. Pleu Enn rium pauciflorum van der Werff & Ham Volume 80, Number 1 1993 van der Werff 117 Revision of Pleurothyrium 27. Pleurothyrium pilosum van der Werff 28. Pleurothyrium poeppigii Nee 29. Pleurothyrium prancei van der Werff 30. Pleurothyrium racemosum van der Werff rff 39. Pleurothyrium williamsii O. C. Schmidt Specimens studied. If collected by a team, only last name of the first collector on the label is listed. Thus, Berg & Akkermans becomes Berg Allen 5885 (14); pcne 602 (4); Aristeguieta 1584 (35); Aristeguieta 164 Rire ө 2085 (35); ] 10); Ayala 3386 5). s 25); Bernardi 1137 (35); Bernardi 2169 (31); Seana 6879 (8); Bosque J-03 (8); Bosque Ј-ОЗА (8); Bosque Ј- 12 (8); Brandbyge 42343 (22); Brandbyge 42366 (22); Breteler 3517 (35); Breteler 3667 (35); Bunting 872 (35); Burger 4690 (14). ыо о 9143 (34); Campbell 6887 (37), Campbell 37); Campos 364 (10); Cardenas 3966 (28); с 12 (21); Castillo 15 (21); Castillo 50 (1); Сегоп 017 (12); Ceron 4982 (25); Cid 859 (37); Cid Ferreira 7292 (25); Cogollo 170 (10); Croat 17593 (25); Croat 19439 (17); Croat 59792 (16); nuc rn 13227 (35); Cuatrecasas 13986 (9); Cuatrecasas 163214 (35); Cua- trecasas 16656 (32); Cuatrecasas 16675 Te Cuatre- casas 16781(35); Cuatrecasas 16864 (35); Cuatrecasas 17196 (19). Davidse 21459 (35); || 21467 (35); Davidse 27524 (2); Davidson 5336 (17); de Bruijn 1422 (35); de Nevers 7520 (27); Diaz 241 (25); Dodson 6302 (11); Dodson 6419 (35); Dodson 10181 (11); Dodson 13632 (11); Dodson 14531 (34); Dorr 4720 (35) Ducke RB19935 (25); Ducke RB25676 (1). Forero 4135 (35); Forero 4693 (35) Foster 3153 (28); Foster 10642 (28); Foster 11719 (37); Freitas 55 E Froes 20526 (31); Froes 21036 (25); Froes 21235 (37 ~ Gentry 7201 (1 2 veni 16627 (25); i eti 18317 (25); Gentry 21284 (25) Gentry 25807 (25); Gentry 25914 (25); Gentry 28959 (25); Gentry 39902 (10); Gentry 43560 (28); Gentry 45117 (7), Gentry 45760 5); Gentry Gentry 58574 D rear 17731 (3); Grandez 1732 (25); Dudino 23); a 14781 (16); Hammel 9 (26); d 17909 (26), Hammel 18152 (35); Hartshorn 2631 (10); Hartshorn 2937 (10); Huber 1557 (10); Hurtado 1464 (33). Janse 280 (7); Jativa 1104 (7); Jativa 2039 (7); da 648 (14); Jimenez 1289 (35); Juncosa 1675 p 145 (10); Kayap 648 (10); Kernan 184 (14); Kernan 187 (14); Klug 1301 (25); Klug 1372 (25); Klug 2122 (7); Klug 2779 (10); Klug 2931 (10); Klug 3116 (10); Klug 3195 (10); Klug 3228 (10); Klug 3567 (10); Knapp 8205 (17); Kroll 285 (21); Kroll 422 (21); Krukoff 4783 (24); Krukoff 4800 (24); Krukoff 5121 (21); Krukoff 5255 (28); Krukoff 5263 (28); Krukoff 5722 и Krukoff 6232 (17); Kunkel 9 (38); Kunkel РА 085 (25); Lamotte 0200 (25); Lau 10 (8); Lawrance 619 (5); Lawrance 780 (10); Little 8488 (10); Little 15822 (35); pos 16121 (35); Little 21359 (7); Llewelyn Williams 1090 (25); Llewelyn Williams 2273 (25); np 8634 d Lugo 320 (28). Marcano Berti 1 (8); Marcano Berti 1177 (35); Marcano ud Nodes 1432 (10); 25); McDaniel 20753 (25); т. 0(30); (35); McPherson 13365 (10); Miranda Bastos 1 Mocquerys 1002 (35); Mori 9074 (25); Mori Ж, (25). Neill 6555 (28); Neill 7019 (10); Neill 7128 (10); Neill 7603 (28); Neill 8741 (33); Neill 8783 (33); Neill 8954 (7); Neill 9033 (17); Neill 9435 те Меш 9439 Palacios 982 (28); Palacios 1332 (17% Palacios 1518 17); Palacios 2094 (35); Palacios 2288 (12); Palacios 2775 (28), Palacios 2804 (35); Palacios 3416 (33); Palacios 4273 (35); Palacios 4334 (35); Palacios 4388 17); Palacios 4436 (35); Palacios 4727 (35); Palacios 7667 (39); Pariona 49 (17); Pariona 63 (10); Peters 52/84 (25); Pires 51430 (2); Рипег 10947 (35); Plow- man 6410 (25); Plowman 6908 (25); Poeppig Addenda 301 (28); Poeppig 1718(10); Poeppig 1845 (10); Poep- pig 2125 (10); Poeppig 2398 (4); Poeppig 2908 (10); Prance 11970 (25); Prance 12392 (29); Ргапсе 13927 (24); Prance 28012 (35). ~ (8). Ramírez 61 (37); Riedel 1412 (36); Revilla 742 (25); Revilla 2532 (25); Revilla 3687 (24); e 102 (10); van Rooden 650 (35); van Rooden 6 35); Rowlee de (23); Rubio 45 (33); Ruíz n ua Ruíz s.n 28). Salcedo 225 (35); Santos 78 (37); Schunke 3475 37); Schunke 13643 (37); 384 (37); Smith 1743 (10); Smith 1756 (10): Smith 2793 (28); Smith 3061 (10); Smith 7463 (35); Smith 9021 (35); Soejarto 1268 (28); Soto 613 (35); Soukop 910 (10); Spruce 2449 (24); Stern ae (23); Steyer- mark 56077 (31); Steyermark 613 ; Steyermark 62230 (8); Steyermark 96072 (8); oo 111580 35). Tamayo 1094 (35); Tessmann 4040 (20); Tessmann 4529 (20); Tessmann 4634 (7); Tessmann 5175 (25); Thomas 6476 (37); Tonduz CR 12652 (23) Triana 1037 (35); dpi. 2042 (35). Ule 9408 (18); Unknown s.n. AR Valverde 15 (35); Vásquez 896 (35); Vásquez 1520 25); Vasquez 4858 (25); oan 4970 (24); Vásquez 5469 (37); Vásquez 7468 (24); Vásquez 7823 (17); Vásquez 7865 (6); Vásquez 7889 (37); Vásquez 8119 39); Vásquez 8444 (17); Vásquez 9138 (25); Vásquez 9365 (17); derum 10233 (24); f ed 10500 (25); Vásquez 108 39); Vásquez 11507 (25); Vásquez 12284 (25); 2 ur 12921 (35); Vasques 13078 (39). Wallnoefer 18-29788 (33); van der Werff 3654 (8); van der Werff 8266 (10); van der Werff 8287 (10); van der Werff 8291 (10); van der Werff 8343 (28); van der /ег 8345 (28); van der Werff 8346 (10); van der (ye ~ 118 Annals of the Missouri Botanical Garden Werff 8673 (10); van der Werff 8887 (35); van der Werff 9471 (11); van der Werff 9956 (25); van der Werff 9967 (1); van der Werff 10050 (25); van der Werff 10066 (25); van der Werff 11157 (10); van der Werff 12362 (11); van der Werff 12365 934); van der Werff 12367 (34); Whitford 49 (8); Williams 1090 (25) Williams 1766 (39); = 2273 (25); Williams 2715 (10); Wurdack 2047 (4). Zak 4459 (33); in 4511 (17); Zak 4608 (35); Za- mora 1363 (35). — Note added in proof At the very last moment, the collection Herrera & Fallas 4638 was received for identification. This collec- a sus a discussion of its affinities are presented in this n Pleurothyrium immersum van der Werff, sp. nov PE: Costa Rica. Punarenas: Cantón de Golfito, Dos Brazos de Rio Tigre, Fila Puriscal Rico (85), Herrera & Fallas 4638 (holotype, MO; isotypes, BM, CR, HBC, INB, MEXU, QRS, US, USJ). Haec species P. golfodulcense proxima, sed foliis flo- ribusque minoribus, reticulatione immersa, tepalis intus pubescentibus et inflorescentia paniculato-cymosa differt. Tree, 20 m tall. Tus terete, B finely gray ap- pressed-pubescent, 1-2 mm below the tip. Terminal bud densely cue ‘pubescent Leaves x -2.5 ст, elliptic, acute, dee surface gla- Pos with scattered ap- pressed hairs, these more dense along midrib, or glabrous; midrib slightly raised on both surfaces, otherwise venation ences in axils phylls, paniculate-cymose, the lateral branchlets once or twice cymosely branched, densely gray- -pubescent, bracts mm the tepals co to bent downw Pedicels m long. Tepals 6, equal or nearly so, pubescent on а surfaces, s margin plane, elliptic, ca. 3c restricted to the center of the glandular mass. 1.2 mm long, the moderately pubescent ovary gradually narrowed into the glabrous style. Receptacle pubescent inside. Cupule and fruit unknown. Flowers: November. Elevation range: ca. 700 m. Pleurothyrium immersum is only known from the type position of the stamens in the glandular mass (not crowded together in the center, but stamens of whorl II separated from the others by the glandular tissue), position of inflo- rescences (not infrequently in axils of normal leaves), the vs. 10-20 cm), the lateral veins are immersed and not raised on lower leaf surface, flowers are much smaller (4-5 vs. 9-12 mm diam.), tepals are pubescent (not papillose) on inner dero, з the inflorescences are paniculate- -cymose, not гасето In the T Pleurothyrium à immersum would run to couplet 37, et mbination in P. immersum of pubescent inner а of tepals ee a pubescent pistil will cause a problem. However, P. immersum differs from all other Pleurothyrium species with pressed-pubescent lower with immersed venation. The glandular dots on the upper leaf surface are also a good indicator for this species. A REVISION OF PANICUM SUBG. DICHANTHELIUM SECT. DICHANTHELIUM (POACEAE: PANICOIDEAE: PANICEAE) IN MESOAMERICA, THE WEST INDIES, AND SOUTH AMERICA! Fernando O. Zuloaga,? Roger P. Ellis,’ and Osvaldo Morrone? ABSTRACT A revision of Panicum subg. Dichanthelium sect. Dichanthelium (Poaceae: Panicoideae: Paniceae) for Mesoameri- a, the West Indies, and South America is presented, considering exomorphological, anatomical data and caryological his as duda seven species and seven varieties area. Panicum caparaoense is described as a new species, and Panicum dichotomum var. An anatomical т of section Dichanthelium, with morphological descriptions, and distribution maps. a separate genus. The taxonomic position of Dichanthelium within together with a key, is not sufficiently dud to be r Panicum is discussed. regarded as of section Dichanthelium of Pan nicum are recognized from t tenue as a new combination. photomicrographs of representative species, is given It is concluded that section Dichanthelium Dichanthelium was established by Hitchcock & Chase (1910) as a subgenus of Panicum in their treatment of North American species of the genus. Dichanthelium was delimited due to the presence of an overwintering rosette of short, broad leaves, which produce simple culms with terminal, chas- mogamous inflorescences in the spring. Later in the season, branch culms arise at the nodes of the vernal culms, forming loose to rather dense fas- cicles of reduced leaves and branches together with cleistogamous inflorescences. Hitchcock & Chase (1910) recognized 17 in- formal groups in Dichanthelium, 110 species and 9 subspecies. The taxa were distinguished princi- ally by characters of the spikelet and leaf blade. Gould (1974) raised Dichanthelium to generic rank. This decision was followed by Clark & Gould (1975), Brown & Smith (1975), Gould & Clark (1978), Clark (1977), Gould (1980), Freckmann (198 la, b), and Hansen & Wunderlin (1988) based on the foliar and floral dimorphism, ploidy level, Kranz syndrome, and ornamentation of the upper anthe- cium as diagnostic characters for the genus. Pohl (1980), Lelong (1984), Zuloaga (1987), Webster (1988), and Morrone & Zuloaga (1991), however, retained Dichanthelium as a subgenus of Pani- cum, due to a lack of adequate characters to sup- port the segregation of these two taxa. Variation in several characters, such as the presence or ab- sence of the basal rosette, and foliar and floral dimorphism in tropical species were not considered by Нисћсоск & Chase (1910), Gould (1974), and subsequent authors in their decision to separate Dichanthelium from Panicum. The present treatment includes a comprehensive taxonomic and anatomical analysis of the 37 spe- cies of section Dichanthelium occurring in Me- soamerica, the West Indies, and South America, which comprise all tropical species of the section. MATERIALS AND METHODS Scanning electron micrographs (SEMs) were prepared of the upper anthecia of most of the species, utilizing the procedures described by So- derstrom & Zuloaga (1989). The vouchers for this ! We tha Simon Guggenheim Memorial Foundation, to whic nk Vladimiro Dudas for the illustrations. Zuloaga acknowledges support of a fellowship from the John e expresses appreciation. Zuloaga's fieldwork in Colombia and Venezuela was carried out under grant #3964-88 from the National Geographic Society, to whom Zuloaga is also grateful. ? Instituto de Botánica Darwinion, Casilla de Correo 22, San Isidro, 1642, Е * Grassland Research Center, Private Bag X05, Lynn East, 0039, South Afric ANN. Missouni Bor. Garb. 80: 119-190. 1993. 120 Annals of the Missouri Botanical Garden study are marked with an asterisk in the specimens examined section of the taxonomic treatment. Field procedures included collection and drying of her- barium specimens, and immediately fixing leaves, spikelets, and other structures in FAA (formalin- ethanol-acetic acid) (Johansen, 1940) and there- after transferring this material to 70% ethanol in the laboratory for storage. Transverse sections of leaf blade material 10 шт thick were prepared after desilicification in 30% hydrofluoric acid (Breakwell, 1914), dehydration in a methyl cel- losolve series (Feder & O'Brien, 1968), and em- bedding in Tissue Prep (Fisher Scientific); the sec- tions were stained in safranin and fast green. Abaxial epidermal scrapes of leaf blades were prepared by removing the mesophyll and vascular tissue with a scalpel and camel hair brush following the method of Metcalfe (1960). The epidermis was stained in safra The adie terminology of Ellis (1976, 1979) was used to describe the anatomical struc- ture of the leaf blades. The following abbreviations are used in the anatomical description: vb/s—vascular bundle/s l'vb/s—frst-order vascular bundle/s 2'vb/s—second-order vascular bundle/s 3'vb/s—third-order vascular bundle/s ibs—inner bundle sheath; i.e., mestome sheath obs—outer bundle sheath; i.e., parenchyma sheath MORPHOLOGY Basal rosette. The rosette is typical of North American species of section Dichanthelium and is also sometimes present in some West Indian spe- cies, P. scoparium Lam., P. aciculare Desv., and P. acuminatum Sw. However, this feature is absent from species from Mesoamerica and South Amer- ica, as well as from some species that are wide- spread in both North and South America, such as P. aciculare, P. acuminatum, and P. sphaero- carpon Elliott. Foliar dimorphism. Leaf size is variable in representatives of section Dichanthelium. Leaf blades are usually broad and long on the main culms and smaller on the axillary culms in species from North America. However, this dimorphism is absent in species from Mesoamerica and Sout America. Main and axillary culms are persistent in Mesoamerican and South American species (Morrone & Zuloaga, 1991), while in North Amer- ican species only the rosette usually persists. The inflorescence in section Dichanthelium is typically a lax, diffuse panicle Inflorescences. borne on the uppermost node of the culm; it is short-exserted to exserted. The axes of the branch- es are pilose or glabrous; when glabrous they usu- ally have multicellular glands. Each axis ends in a developed spikelet; these spikelets are solitary on the branches, or occasionally geminate in speci- mens of P. hebotes Trin. usually open, is contracted in P. superatum Hackel The panicle, although and P. aciculare. Glands. istic feature of several species of section Dichan- Multicellular glands are a character- thelium. 'These glands are present in P. acumi- natum, P. sabulorum var. polycladum (E. Ekman) Palacios, P. sabulorum var. cordatum Zuloaga & Morrone, P. stigmosum Trin., P. adenorachis Zu- loaga & Morrone, P. sphaerocarpon, P. strigo- sum Muhlenb., P. surrectum Chase ex Zuloaga & Morrone, P. ensifolium Elliott, P. scoparium, P. cucaense Zuloaga & Morrone, P. divergens HBK, and P. laxiflorum Lam. These glands occur on the axes of the inflorescence branches, peduncles, ped- icels, culms, sheaths, and blades. Spikelets. void, with two glumes, two florets, and two flowers. They disarticulate at the base of the lower glume. Lower glume length varies from У, to Y the length of the spikelet. The upper glume and lower lemma are isomorphic and 7-11(-15)-nerved, or occa- The spikelets are ellipsoid to obo- sionally 5-nerved in P. sciurotis Trin. and P. en- sifolium. The lower floret is either staminate or neuter; the lower palea may be as long as the lemma or reduced to absent in P. aciculare and P. penicillatum Nees ex Trin. The upper anthe- cium is almost as long as the spikelet; the upper lemma and palea are indurate, smooth and shining, and pale (but dark in P. hebotes) with simple pa- pillae regularly distributed over the surface. The upper lemma is shortly apiculate or crested and pilose or glabrous, with the margins involute and covering 23 of the upper palea. The upper palea is 2-nerved. The upper floret is bisexual, with two conduplicately truncate lodicules that embrace the lower margins of the palea, three stamens, and a gynoecium with two free styles and a plumose stigma. Floral dimorphism. | Hitchcock € Chase (1910) and Gould & Clark (1978) described the presence of both chasmogamous and cleistogamous flowers in the species of Dichanthelium from Nort America. Study of herbarium specimens and native and cultivated Mesoamerican and South American plants, including widespread species from both North America and South America, demonstrated that the production of chasmogamous and cleistoga- mous flowers is extremely variable. Climatic fac- Volume 80, Number 1 1993 Zuloaga et al. Revision of Panicum subg. Dichanthelium sect. Dichanthelium tors, such as humidity and temperature, appear to determine the proportion of cleistogamous flowers produced. Cleistogamous flowers also occasionally occur on species of section Dichanthelium from Mesoameri- ca and South America. CHROMOSOME NUMBERS Section Dichanthelium is characterized by a basic chromosome number of x — 9. Gould & Soderstrom (1967) mentioned one deviation in this character, a count of 2n = 20, x = 10 for P. missionum Mez, non Ekman (= Panicum surrec- tum Chase ex Zuloaga & Morrone). Most of the species are diploid, with 2n = 18. In addition, six tetraploid counts, one hexaploid count, and one octoploid count are also known. The following list summarizes chromosome counts determined for species of section Dichanthelium treated in the present contribution: Panicum aciculare Desv. var. aciculare: 2n = 18 (Brown, 1948; Dubcovsky & Zuloaga, 1992). P. aciculare Desv. var. arenicoloides (Ashe) Beetle: n = 18 (Brown, 1948; Gould, 1958; Burton, 1942, as P. arenicoloides Ashe; Brown, 1948, as P. angustifolium Elliot). acuminatum Sw. var. acuminatum: n — 9 (Davidse & Pohl, 1972a, 1978, as P. oli- vaceum A. Hitchc. & Chase; Spellenberg, 1970, as P. olivaceum); 2n = 18 (Dubcovsky & Zuloaga, 1992). acuminatum Sw. var. longiligulatum (Nash) Lelong: 2n — 18 (Brown, 1948, as P. lon- giligulatum Nash and P. wrightianum Scrib- ner). P. aequivaginatum Swallen: 2n — 18 (Dubcovsky & Zuloaga, 1992). . davidsei Zuloaga & Morrone: 2n — 72 (Dub- covsky & Zuloaga, 1992). . dichotomum L.: 2n = 18 (Brown, 1948; Clark, 1977; Sherif et al., 1983; Brown, 1948, as P. barbulatum Michaux, P. yadkinense Ashe, and P. caerulescens Hackel ex A. Нисћс.). dichotomum L. var. unciphyllum (Trin.) Zu- loaga & Morrone: 2n = 18 (Brown, 1948, as P. albomarginatum Nash). P. divergens HBK: 2n — 18 (Burton, 1942; Brown, 1948; Sherif et al., 1983, as P. commutatum Schult.). ensifolium Baldwin ex Elliott: 2n — 1948; Brown, 1948, as P. SN > вә! ~ TU S © 18 (Brown, chamaelonche Trin. ). P. laxiflorum Lam.: 2n = 18 (Brown, 1948; Gould, 1958; Pohl & Davidse, 1971; Bowden, 1960, as P. xalapense Kunt . portoricense Desv. ex Ham.: 2n — 18 (Brown, 1946; Brown, 1948, as P. lancearium Trin.). P. pycnoclados Tutin: 2n — 18 (Dubcovsky & Zuloaga, 1992). . sabulorum Lam. var. polycladum (E. Ekman) Palacios: 2n = 36 (Parodi, 1946; Nunez, 1952, as P. sabulorum; Dubcovsky & Zu- loaga, 1992). P. sabulorum Lam. var. sabulorum: 2n = 54 (Dubcovsky & Zuloaga, 1992; Parodi, 1946; Nünez, 1952, as P. fultum Hackel). P. sciurotoides Zuloaga & Morrone: 2n — 18 (Dubcovsky & Zong. 1992). P. scoparium Lam.: = 9 (Davidse & Pohl, 1972b); 2n = 18 Brow, 1948) P. sphaerocarpon Elliott: n = 9 (Spellenberg, 1970; Davidse & Pohl, 1972a, 1978; Reveal & Spellenberg, 1976); 2n = 18 (Church, 1929; Brown, 1948; Gould, 1958; Pohl & Davidse, 1971; Reveal & Spellenberg, 1976; Dubcov- sky & Zuloaga, 1992; Brown, 1948, as P. sphaerocarpon Eliott var. inflatum (Scribner & Smith) A. Hitchc.) P. strigosum Muhlenb.: 2n = 18 (Brown, 1946; Brown, 1948, as P. ciliatum Eliott). viscidellum Scribner: n = 18 (Davidse & Pohl, 1972a, 1974); 2n = 36 (Pohl & Davidse, 1971; Dubcovsky & Zuloaga, 1992). v Tv D LEAF ANATOMY Anatomical data for 14 species of section Di- chanthelium are presented by means of photo- micrographs (Figs. 1-23). The following species were studied: P. aciculare, P. acuminatum, P. aequivaginatum, P. cumbucana Renvoize, P. davidsei, P. hebotes, P. peristypum Zuloaga & Morrone, P. pycnoclados, P. sabulorum, P. sci- urotoides, P. sphaerocarpon, P. stigmosum, P. superatum, and P. viscidellum. For comparative purposes the North American species P. boscii Poir., P. clandestinum L., and P. oligosanthes Schultes are also included in this study. Panicum acuminatum is the only species studied anatomi- cally with specimen samples from both American continents. For this species a comparison between the leaf anatomy of the basal rosette leaves and the cauline leaves is also made. e leaf anatomy of section Dichanthelium is variable, with few unifying or diagnostic characters being evident; the vascular bundle sheaths and their extensions are the most consistent and valuable anatomical character for distinguishing the section 122 Annals of the Missouri Botanical Garden B ~ Oi un AA у ~ Abro SSA Aten’ M Aou cS . Е За EFE m ecran Peut O БҮ a Volume 80, Number 1 1993 Zuloaga et al. 123 Revision of Panicum subg. Dichanthelium sect. Dichanthelium FIGURE y nato [erentiated x o vascular bundles (arrowed). — Leaf blade anatomy of South American specimens of Panicum acuminatum. A-C. Transectional . Outline of half of lamina showing rounded margin with well-developed sclerenchyma cap and undif- Detail of semiradiate chlorenchyma, colorless outer bundle sheath extension cells, uniseriate adaxial and abaxial bundle sheath extensions, and small sclerenchyma strands associated of the third-order еа the —— rs, which are and fur „р, E. Abaxial pre structure Ж, the niis zones; microhairs also re iii aaa commc undles and slight adaxial ribs and furrows; note the raised cushion cells associated with n in the abaxial ови! zones. — C. Specimen with thinner leaf without ribs with few macrohairs, th LE. Sp two inflated epidermal TP the central files of the н а zones with bulliformlike long cells (based on Zuloaga 50). 4202; А х 100; B-E x and subgenus Dichanthelium from most of the other C, subgenera of Panicum. A detailed de- scription of section. Dichanthelium based on the studied species is presented below LEAF BLADE IN TRANSVERSE SECTION Outline: expanded, either flat or very broadly V-shaped; arms of lamina either straight or out- wardly bowed; two halves of lamina symmetrical about the median vascular bundle; leaf width vari- ~ able and leaf blade section includes between 17 and 117 vbs; P. aciculare (Fig. 1) is an exception with narrow (only 11 vbs), inrolled leaf blades. Ribs and furrows: variable, from flat adaxial surfaces without ribs or furrows to medium furrows (about a quarter of the leaf thickness); furrows wide and open, occurring between all vbs; adaxial ribs, when present, кене оуег ше vbs, with rounded apices, and all y uniform; abaxial ribs usually absent but slight e may be developed; in P. aciculare abaxial ribs are clearly present (Fig. 1 B). Midrib: variable, from undifferentiated median vb to definite keel; median vbs, structurally indistin- FIGURE 1. blade containing only three | prol with the a median bundle (arrowed) stru : ES ata, the cost by tall and narrow cork cells, the intercostal one Leaf blade anatomy of Panicum aciculare. — A. Blade outline showing на паггом, коч pc urally i i silica bodies separated cells are elongated-rectangular with very sinuous walls, and small, narrow microhairs occur in the intercostal pe macrohairs are absent (based on Zuloaga & Londono 4201; A х 100; B, С x 250) 124 Annals of the Missouri Botanical Garden FIGURE. = blade secti bundles; note кш based macrohair; adaxial rib with a tendency to adaxial palisade and abaxial spongy mesophyll tissue; note adaxial and sheath extensions associated with the third-order bundle Leaf blade anatomy of Panicum acuminatum specimens from North America. A-D. Transverse leaf edian part of blade with midrib (arrowed) structurally barely distinct from lateral first-order s and furrows not present. — B. Semiradiate chlorenchyma tissue abaxial uniseriate bundle the extensions linked to small sclerenchyma strands. — Specimen with diffuse chlorenchyma tissue with very large intercellular air spaces and very thin cuticle. — D. Specimen with compact mesophyll tissue; no adaxial r macrohair with associated cushion the stomata), intercostal long cells and numerous stomata in files adjacent to the of specimen with thin cuticle and diffuse mesophyll; no ES macrohai , based on oe & P and narrow costal zones, only three cells wide (A, B, & Ellis 30807; D, based on Zuloaga 2492; A x 1s D, F x25 guishable from other l'vbs, most common (Figs. А, 12B, 15C, 16A, 17A, 18A, C, 23A); панка (Figs. ЗА, 11A, 14A, 21A) and keels (Figs. А, С, 10A, 22A) much less common; keels include either 1 or 3 abaxially positioned vbs and adaxially located parenchyma ground tissue; no lacunae preser Va scular bundle arrangement: number of vbs in entire blade; P. aciculare, with З (Fig. 1), ‹ P. sabulorum, which may have only 5, also have the narrowest leaf blades; 2 а астешаге апа Р. sabulorum, Fig. 12D), 3, 4, 5, 6, 7, and rarely , 3'vbs are located dd consecutive lateral ]'vbs; the wider the leaf the more 1'vbs and the more minor bundles are present between the 1'vbs; 2'vbs absent. Vbs usually displaced slightly toward de thickness although they may be centrally situated; all vbs positioned the abaxial side of the bla ibs or furrows developed. base cells (arrowed), elongated fingerlike microhairs (at least twice the length of . Epi . Abaxial epidermis. — E. Detail of small narrow costal zones ermis rs, sinuous walls of long cells, яни ы omata 30825; C, F, based on Dune CILE. at the same level. Vascular bundle description: 3'vbs angular with phloem distinguishable. elliptical in outline; phloem adjoins ibs; lysigenous cavities and protoxylem present; circular metaxy- lem vessels narrow with diameters less than half those of the obs cells. Vascular bundle sheaths: obs of 1' and 3'vbs conspicuous, entire, round to elliptical, consisting of 5-8 (3'vbs) or 12-18 (1'vbs) inflated, rounded cells; chloroplasts absent except А iphdéros arpon, Fig. viscidellum in a few specimens (e.g., 15A-D, as well as specimens of with silicified macrohair bases, Fig. 20). Adaxial and abaxial bundle sheath extensions in all species; adaxial 3'vb extensions usually uniseriate (but may be biseriate) and consisting of 1-5 cells, whereas abaxial extensions are shorter (usually only 1 cell deep) and wider; the extensions consist of con- spicuous, thin-walled, colorless cells very similar to the obs cells; the extensions extend and join small Volume 80, Number 1 Zuloaga et al. 125 1993 Revision of Panicum subg. Dichanthelium sect. Dichanthelium rs у се" МӨТ E yet a КАИРУ |) + a 3 у», Мә 47 A gas $. Sete DRE pak AN TW РРА Ба US ^ 17 1] ГЕРЕ LN LJ Sa Arr Sia ES = ера dde MB Vus. ar vaen FIGURE 4. A comparison of the leaf blade anatomy of the upper culm leaves and the basal rosette leaves of m acuminatum. A, B. Upper culm leaves.— A. Transverse section showing slight adaxial ribs and furrows, distinct bulliform cell groups, and very short bundle sheath extensions; the mesophyll is unusual in that the abaxially located cells are also palisadelike in their arrangement, and the mesophyll tissue is dense, without large intercellular air spaces. —B. Abaxial epidermis with short long cells, short microhairs (less than twice the length of the stomatal complexes) and dumbbell-shaped silica bodies alternating with cross-shaped cork cells on the central files of the costal stomata, no macrohairs, and with costal zones to five cells wide; the silica bodies are irregular in shape, often resembling the dumbbell type (A-D, based on Zuloaga 2490; A-D x250). ar pori ie E T gah ~ > CM E жа: s Me hs OP. Lone vira гр = Эме dag у is A i — — —— IL "*—— Dd A o st FIGURE 5. Leaf blade anatomy of Panicum aequivaginatum. A, B. Transectional anatomy. — A. Center of lamina (e.g., B) lack these central intercostal files with microhairs and macrohairs (A, B, based on Zuloaga et al. 4470; C, based on Zuloaga et al. 2469; А x100; B x250; C x400). 126 Annals of the Missouri Botanical Garden FIGURE 6. Leaf blade anatomy of Panicum cumbuca , B. Transverse leaf sections. — A. Gently tapering margin with evenly spaced vascular bundles. — B. Detail еы ‘bundle sheath extensions, small adaxial and abaxial sclerenchyma strands and rather dense chlorenchyma tissue with a tendency toward the radiate type of cellular arrangement. — C. Abaxial ш structure with narrow costal zones only three to five cells wide; silica bodies are dumbbell-shaped to nodular and very narrow; the a intercostal zones composed of short, hexagonal long cells and microhairs (arrowed) and macrohairs are common, the latter being rather short and pointed and inserted between only a few (two to three) m eed epidermal cells one of which appears to be inflated (A-C, based on Calderón 2425; A x100; B, C x 250). SSS SSS EC E = = = ОВЕ 7. Le ques eral of Panicum davidsei. A-C. Transectional anatomy. — A. Gently tapering margin with prod ichyma xial ribs and furrows well developed, slight abaxial furrows present. — В. Long (four or more cells), uniseriate bundle sheath extensions associated with thinds order adios onl first-order bundles with similar anatomical structure except first-order adaxial bundle sheath extension uniseriate and three cells long (ar- rowed). — D, E. Abaxial epidermal structure. — не Interstomatal cells, stomata, and mic ee (arrowed) common but no hpi iidem appendages present. — E. No appendages except for microhairs; two stomatal files in each intercostal each separate a short interstomatal cell puer files of three cell files, the la with alternating dumbbell- ike silica bodies and cork cells (A, B, D, based on Zuloaga et al. 4427; C, E, based on Zuloaga et al. 4406; A x100; B-E x250). Volume 80, Number 1 1993 Zuloaga et al. Revision of Panicum subg. sect. Dichanthelium Dichanthelium FIGURE 8. cells large and inflated (A, B, based on Zuloaga et al. 2 et al. 2390; А x100; C x 250; B, D x400) sclerenchyma strands on both adaxial and abaxial surfaces. Ibs entire, of small cells with uniformly thickened walls. Sclerenchyma: chyma strands associated with all vbs, both adax- ially and abaxially; these strands linked to bundle sheath extensions; fibers lignified; small, rounded sclerenchyma caps located in leaf margins (Figs. 2A, ТА, 15A, Mesophyll: chlorenchyma semiradiate with ad- axial cells tending to a palisade-type of arrange- ment; abaxially located chlorenchyma cells irreg- ular in shape, diffusely arranged with large intercellular air spaces, resembling spongy meso- phyll; walls not invaginated; mesophyll varies from small, inconspicuous scleren- dense to very diffuse, irregular tissue; always more than 5 chlorenchyma cells between consecutive vbs. No colorless tissue apart from the bundle sheath extensions. Adaxial epidermal cells: bulliform cells present in adaxial furrows between all vbs; usually in re- stricted groups with larger, fan-shaped central cells; size of bulliform cells very variable. Epidermal cells small, regular in size; no papillae or prickles, but macrohairs may be present. Abaxial epidermis cells: very narrow costal zones; wide intercostal zones, often with larger, inflated, bulliformlike cells in the central files (e.g., P. hebotes, Fig. 8B, D); cuticle often very thin; chyma. urrows, and diffuse mesophyll tissue with large е air о ~ ДГ KI E + $ =, 9 + ЕТ hs af bl ade anatomy of Panicum hebotes as seen in transverse section. — А. Expanded leaf blade with tail of very short “| sheath adaxially e iiid these ed epidermal cells. — Edid f keel structure 373; C, based on Zuloaga et al. 4534; D, pas on a no papillae or prickles, but macrohairs either com- mon or absent, either with raised cushion-bases or inserted between unmodified epidermal cells as seen in section. ABAXIAL EPIDERMIS AS SEEN IN SURFACE VIEW Intercostal long cells: variable, short-rectan- gular (Figs. 13F, 16E) to hexagonal, resembling bulliform cells (Fig. 10C, D); side walls parallel or angled outward, end walls vertical, and anticlinal walls either unthickened or slightly thickened; rect- angular cells with moderately undulating anticlinal walls but hexagonal cells usually with straight walls; intercostal cell shape variable across intercostal zones, with lateral cells being elongate and rect- angular and central files being hexagonal and bul- iformlike; no short cells between successive long cells except in P. aciculare (Fig. 1 absent in P. aciculare (Fig. 1C) es otherwise always present; low-triangular to low dome-shaped; two rows of stomata in each intercostal zone, these restricted to the edges adjacent to the costal zones; a single short interstomatal cell present between . Stomata: successive stomata in a file. Intercostal short cells: absent except in P. acic- ulare (Fig. 1C), where they are tall and narrow. apillae: absent. Prickles: absent throughout sam- 128 Annals of the Missouri Botanical Garden <-> А [e 3 ae ED A A ч ay х - ad D sap X - . NC) TI x D ta FicURE 9. Abaxial leaf blade epidermis of Panicum hebotes. — A. Epidermal pattern showing marked zonation very narrow costal zones and wide intercostal zones with numerous macr shaped acrohair bases associated wi microhairs are three-celled (arrowed)—a stoma Zuloaga et al. 2390; A x 160; C x 250; B, D x 400). ple. Microhairs: bicellular, fingerlike, with basal cell slightly shorter than distal cell; elongate, equal in length to the stomatal complexes to more than twice their length; distal cell deciduous with very thin walls; common in the central files of the in- tercostal zones; very rarely absent as in P. sphaer- ocarpon (Fig. 15E, F); Panicum hebotes always with unique tricellular microhairs (Fig. УВ, D — Macrohairs: either absent ог present, variable in structure; two types can be recognized: long, thick hairs inserted into raised cushion-bases (Figs. 1 1C, 19A- D), specialized epidermal cells associated with the base Figs. 6C, 9A-D, 10C, D); all intermediates occur between these extremes; recurved hairs may occur and thin, needlelike hairs with only two (Figs. 2D, 14D, E); macrohairs always restricted to intercostal zones; size and frequency very vari- able. Silica bodies: costal bodies irregular in shape, cross-shaped to equidimensional dumbbell-shaped to elongate dumbbell-shaped or even nodular (the latter type only overlying 3'vbs); confined to costal zones where they alternate regularly with similar- shaped or shorter cork cells; only 3 (rarely 5) files ohairs. — B. Detail of irregular dumbbell- costal silica bodies alternating with short, rectangular cork cells; intercostal long cells short with sinuous walls; 1 1 ith two specialized epidermal cells, one of which appears inflated; note that all the a very unusual configuration. — C. this species; cells associated with the macrohair bases stain dar | a; the microhairs (arrowed) all consist of three cells (A, B, based on Zuloaga et al. 2373; C, D, based on ypical epidermal cellular arrangement of y. — D. Short intercostal long cells, macrohairs, and of cells per costal zone, with 1 file with silica bodies. No intercostal silica bodies. DISCUSSION OF LEAF ANATOMY OF SECTION DICHANTHELIUM Leaf blades in section Dichanthelium are typ- ically flat, open, expanded, and symmetrical about the midrib. Leaf width is variable, including be- tween 11 and 117 vascular bundles. Panicum acic- ulare differs from all other representatives of the section in having narrow, inrolled blades with only 11 bundles (Fig. 1A, B). This species also differs in several other respects, which may be linked to these narrow leaf blades he degree of development of adaxial ribs and furrows varies in section Dichanthelium, from complete absence to well-developed medium ribs and furrows. The full range of variation found in the section can occur in a single species, as is seen in P. acuminatum (Fig. 2B, C). Midrib structure is also variable, ranging from median vascular bun- dles structurally indistinguishable from the lateral Volume 80, Number 1 1993 Zuloaga et al. Revision of Panicum subg. Dichanthelium sect. Dichanthelium A meos vd RIA Ре VES on eai FIGURE 10. Leaf blade anatomy of Panicum peristypum. А, В. Ana Well-developed keel containing three vascular bundles and adaxial parenchyma ground tissu atomy as seen in transverse section e. — B. Detail i is mesophyll and bundle sheaths and their extensions. C, D. Abaxial | — C. Epidermal pattern with very narrow costal zones and wide intercostal zones containing numerous macrohai Detail of inflated basal cells ed s.—D. with the macrohairs, bicellular microhairs, stomata, and short, а shaped long cells (A-D, based on Zuloaga ). et al. 2398; A x100; B, С x250; D x400 first-order bundles to well-developed keels including up to three vascular bundles and extensive paren- chyma ground tissue. Generally, each species is consistent in midrib structure, having either a me- dian bundle, a midrib, or a keel. This is not always consistent, however; for example, all midrib types um pycnoclados. FIGURE 11. Anat of Pan A, B. Transectional anatomy. were apparent on different leaves of the same plant of P. peristypum The vascular hule sheaths and their extensions are the most consistent and valuable anatomical character distinguishing Dichanthelium from the other C, taxa in Panicum. Colorless parenchyma ОРТ и HEUS — A. Midrib MERDA comprising a single vascular bond авио from other first-order bundles by the large adaxial and abaxial sclerenchyma girders. — B. Rather com pact m order bundles vith- long uniseriate extensions; note macrohair rowed). — C. 4404; A x100; B, C x ophyll of the intermediate type with palisade and spongy tissue differentiated; third- with base inserted between raised cushion cells (ar- axial ке with thick macrohairs associated with cushion bases (A-C, based оп Zuloaga et al. 50). 130 Annals of the Missouri Botanical Garden FIGURE 12. Leaf blade foro of iei um sabulorum in transverse sec — А. Expanded blade showing "s eer base of thick, cushion-based macrohair owed). — B. Cen heus na with median vascular bundle (arrowed) indistinguishable in lateral ud -order bundles. — C. Detail of diffuse mesophyll tending to the radiate type of arrangement; note short bundle sheath extensions and macrohairs without cushion bases (arrowed). — D. Denser mesophyll tissue with conspicuous outer bundle sheath and extension cells and small but discrete bulliform fans located in the troughs of the shallow adaxial furrows. — E. Typical anatomy of this species except that no macrohairs are present. — Е. Well-developed bundle sheath extensions and very diffuse mesophyll (A, C, based on Zuloaga 2060; B, based on Zuloaga 1972; D, based on Zuloaga 2128; E, based on Zuloaga 2380; F, based on Zuloaga 2352; A, B x100; C-F x250). sheath and extension cells are present in all taxa of specialized chloroplasts in the bundle sheath studied. Panicum sphaerocarpon (Fig. 15A-D)is cells, indicates that section Dichanthelium is non- an exception, as unspecialized chloroplasts appear ranz throughout. More than five chlorenchy- to occur in the outer sheath and extension cells. matous mesophyll cells are always found between In a few specimens of P. viscidellum (Fig. 20A successive vascular bundles, implying that many D), plastidlike structures are also evident in these chlorenchyma cells are more than two cells re- tissues, but these appear to be found primarily in moved from the nearest bundle sheath cell. These older, senescing leaves. Nevertheless, all specimens parameters have been shown to predict reliably the of section Dichanthelium examined had bundle photosynthetic pathway in grasses (Hattersley & sheath extensions, the length correlated with the Watson, 1975), and section Dichanthelium is con- thickness of the leaf blade. Section Cordovensia sequently considered to be entirely C,. This has of subgenus Dichanthelium also has prominent been confirmed by Brown (1977) for 73 species adaxial bundle sheath extensions. Extensions аге of Dichanthelium. thus diagnostic for subgenus Dichanthelium, but The mesophyll tissue of section Dichanthelium occur sporadically in the other C, subgenus of is also of interest in that the chlorenchyma tissue Panicum, Phanopyrum (Raf.) Pilger, where they does not display the irregular pattern of cellular are found in section Lorea Zuloaga (P. chnoodes | arrangement that is normally associated with non- Trin. and P. trinii Kunth), and often in species of | Kranz leaf anatomy. Instead, it is semiradiate, with sections Parvifolia (A. Hitchc. & Chase) Pilger, the adaxially located cells palisadelike and the ab- Stolonifera (A. Hitchc. & Chase) Pilger, and Me- — axial cells resembling spongy mesophyll tissue. This gista Pilger. Sections Laxa (A. Hitchc. & Chase) anatomy is clearly distinguishable from that of the Pilger, Parviglumia (A. Hitchc. & Chase) Pilger, _ pooid C, grasses, but most C, panicoids also have and Phanopyrum of subgenus Phanopyrum are mesophyll of this type. Panicum sphaerocarpon without bundle sheath extensions, as is the inter- (Fig. 15B) is unusual in that the adaxial chloren- mediate C,/C, subgenus Steinchisma (Raf.) Zu- chyma cells are also interspaced with intercellular loaga. air spaces, although a semiradiate arrangement is The mesophyll tissue, together with the absence — still evident. Volume 80, Number 1 Zuloaga et al. 1993 Revision of Panicum subg. Dichanthelium sect. Dichanthelium м 3 m Бы ott Boyett MAR FIGURE 13. Abaxial epidermis of Panicum sabulorum. — А. Epidermal pattern with short macrohairs restricted to the center of the intercostal zones; note numerous stomata in two files immediately adjacent to the costal zones. — B. Detail of macrohair bases and few associated epidermal cells, irregular arrangement of dumbbell-like silica bodies in costal zones, and short bicellular microhairs. — C. Specimen without abaxial macrohairs and somewhat wider costal zones (but still consisting of three cell files). — D. Specimen without macrohairs, but note pairs of inflated epidermal cells indicative of deciduous macrohair bases (arrowed); the microhairs are clearly visible, particularly the basal cell. — E. Specimen in which the intercostal zones can be subdivided into two areas: those adjacent to the costal zones with sinuous-walled long cells, few stomata, and few microhairs, and the centrally located files which are bulliformlike; a few small macrohairs were observed in these central files. — Е. Intercostal zones with rectangular, sinuous-walled long cells throughout, stomata in two separate files near the costal zones and bicellular microhairs common along the central cell files (A, B, based on Zuloaga 2060; C, based on Zuloaga 1972; D, based on Zuloaga 2128; E, based on Zuloaga 2352; К, based on Zuloaga 2380; A, C x 160; E x 250; B, D, Е x 400). Panicum hebotes is unique in section Dichan- and Panicum validum Мег (Zuloaga et al., 1989) thelium in having three-celled microhairs (Fig. 9B, of the Panicoideae D). Multicellular microhairs are rare in the Po- Multicellular glands are present in the abaxial aceae, where they have only been reported for the intercostal zones of the epidermis of P. adenora- Guaduelleae, a primitive bambusoid tribe (Soder- chis (Morrone & Zuloaga, 1991). strom & Ellis, 1987), lodicules of Streptogyna crinita (Soderstrom & Judziewicz, 1988), a few species of Pentachistis from the Arundinoideae This study of species of Dichanthelium from (Ellis & Linder, 1991), Paspalum L. (Türpe, 1966) Mexico and the West Indies to Argentina dem- CONCLUSIONS 132 Annals of th Missouri Botanical Garden FIGURE 14. lamina showing keel with additional sclerenchyma and Thin blade with palisade and spongy me esophyll tissue distinguishable; bundle sheath ERN very short. — C. Thin eat th extensions visible in s without cushion bases ES p cells with sinuous walls; elongate, fingerlike microhairs occ absen Detail of macrohairs and their bases, microhairs, and long cells, which are very sho Leaf blade anatomy of Panicum sciurotoides. A-C arenchyma development (arrowe Anatomy in transverse section. — À. Central ; note macrohairs. — B. е са these files (arrowed); stomata rt and irregular in shape (A, В, D, based on Zuloaga et al. 2349; C, E, based on Zuloaga et al. 2354; А x100; B, C, D х250; E 00). onstrates that there are insufficient differences to justify separation of Dichanthelium from Panicum as a distinct genus. The data presented also show a significant variation in the characters traditionally used to separate these taxa at the generic level. The rosettes and foliar and floral dimorphism are traditionally used to delimit Dichanthelium as a genus, but are only characteristic of species with a boreal distribution, and are absent in the rest of the species of the subgenus (Morrone & Zuloaga, 1991). This even applies to widespread species of Dichanthelium distributed from the United States to northern South America. Morrone & Zuloaga (1991) suggested that these characters could rep- resent an evolutionary adaptation to harsh winters. Gould & Clark (1978) noted that Dichantheli- um includes mostly diploid species, with only three known polyploids. They contrasted this with Pan- icum, in which polyploid species predominate. Dub- covsky & Zuloaga (1992) confirmed the presence of additional polyploid species, including tetraploids and octoploids, in Dichanthelium representatives from South America. Consequently, differences in the ploidy level between Dichanthelium and other subgenera of Panicum are considered to be tax- wn & Smith (1975) and Gould & Clark (1978) justified the separation of Dichanthelium from Panicum on the basis of the Kranz syndrome. These authors characterized Dichanthelium as С, or non-Kranz and Panicum as C, or Kranz. How- ever, many non-Kranz species of Panicum are pa in dini ит Phanopyrum (Zuloaga, 1987), 50 this does not seem useful in separating Dichanthelium. There are also no consistent dif- ferences in the ornamentation of the upper anthe- cium between Dichanthelium and other groups of Panicum (Zuloaga, 1987). This difference was used ; & Clark (1978) to distinguish Dichan- thelium from Panicum ићи Panicum, Dichanthelium can be distin- guished at the subgeneric level by the following set by Gould Volume 80, Number 1 1993 Zuloaga et al. Revision of Panicum subg. Dichanthelium sect. Dichanthelium aA. "e. KC ISI deese > У - ^ Ў 23 > ?M E о intercostal long cells, and two files of stomata per intercostal zone on Zuloaga 4249; А, C x100; E x160; B, D F, G x2 of characters: lax inflorescences; ellipsoid to obo- void spikelets; upper glume and lower lemma usu- ally 7-11-nerved; upper anthecium apiculate or shortly crested, and simple papillae on the lemma and palea. Anatomically, all species are non-Kranz or C,, with the outer parenchymatous sheath lack- ing specialized chloroplasts; a complete inner mes- tome sheath surrounds first-order vascular bundles; mesophyll irregularly radiating, and palisadelike toward the adaxial surface, with 5-8(-10) meso- д» => 0 ophyll characteristic in that spongy and palisade Epidermal pattern showing absence of macrohairs, short . — К. Costal zones of five cell files, two with dumbbell- shaped silica bodies; intercostal long cells with thickened periclinal cell walls, particularly the bulliformlike centrally located cells; lateral long cells more rectangular in shape with files of stomata; no microhairs evident. — С. Bicellular microhairs present in the central files of the intercostal zones (A, B, G, based on Zuloaga 4035; C, D, E, F, based 50) phyll cells between consecutive vascular bundles; extensions of the outer bundle sheath extending toward both epidermides and of 1-4 cells deep; dumbbell-shaped or nodular silica bodies; bicellular microhairs; and the stomatal complex dome-shaped or triangular. Section Cordovensia (A. Нисћс. & Chase) L. Parodi differs from section Dichanthelium, as cir- cumscribed here, in the membranous ligules, spike- lets with the upper glume and lower lemma usually 134 Annals of the Missouri Botanical Garden FIGURE 16. Leaf anatomy of Panicum stigmosum. A-C. Leaf blade in transverse section. — А. Central part of lamina showing median vascular bundle.— B. Third-order vascular bundles with uniseriate sheath extensions; note abaxial stomata (arrowed). — C. Detail showing diffuse mesophyll with large intercellular air spaces in the spongy tissue. D, E. Abaxial epidermis. — D. Epidermal zonation with wide intercostal zones and very narrow intercostal zones.— E. Detail of stomata with triangular subsidiary cells, rectangular long cells with sinuous walls, bicellular microhairs, and costal zones three cells wide with central file with dumbbell-shaped silica bodies alternating with short cork cells (A-E, based on Zuloaga et al. 2351; А x100; C x160; B x250; D, E x400). 5-nerved, and the lower flower and palea absent. inflated bulliform cells that lie parallel to the epi- Anatomically, species of section Cordovensia have dermis; the abaxial epidermis has homogeneous midribs with rounded adaxial and abaxial projec- intercostal zones; long cells have very deeply un- tions; they have restricted groups of 5-7 large, — dulating cell walls. In section Cordovensia the ex- FIGURE 17. Leaf anatomy of Panicum superatum. А, B. Transectional anatomy. — А. Median part of blade showing median vascular bundle (arrowed) identical to other first-order vascular bundles. —B. Anatomical detail of conspicuous, translucent outer bundle sheath and extension cells and diffuse mesophyll tissue. — C. Abaxial epidermis without macrohairs and with inflated, short intercostal long cells (A-C, based on Zuloaga et al. 2374; A x100; B С x 250). Volume 80, Number 1 1993 Zuloaga et al. 135 Revision of Panicum subg. Dichanthelium sect. Dichanthelium FicunE 18. showing sclerenchyma c extensions associated with third-order bundles and bundles located dist: order vascular bundles. — Transectional leaf blade anatomy of Panicum viscidellum. — А. Lateral and median part of blade cap in margin and median vascular bundle (arrowed). — B. much shorter, biseriate adaxial extensions from t — C. Midrib bundle (arrowed) with ao larger sclerenchyma girders than are associated with ie laterally argin showing sclerenchyma cap and cushion-based macro Detail of uniseriate bundle sheath e first-order air. — E. Detai of mesophyll tissue tending to radiate type of arrangement; note intercellular air spaces and uniseriate bundle sheath extensions. — F. Chlor whereas the abaxial tissue is irregular, resembling spongy mesophyll; note uniseriate sheat adaxial and abaxial sclerenchyma strands associated with all enchyma cell arrangement clearly evident with the adaxial cells somewhat radiately arranged extensions and sma vascular bundles (A, B, based on Zuloaga et al. 4532; C, D, based on Zuloaga et al. 4530; E, based on Davidse 32336; F, based on Stevens 25420; A, C x100; B, D- F x250 tensions of the outer parenchymatous sheath are composed of 1(-2) cells, or are occasionally absent. TAXONOMIC TREATMENT Panicum subgenus oo A. Hitche. ase, Contr. U.S . Herb. 15: 142. 1910. TYPE SPECIES: aud dichotomum L. Panicum section Dichanthelium Panicum section Dichotoma (A. Hitchc. & Chase) Hsu, . Fac. Sci. Univ. Tokyo, sect. 3, Bot. 9: 119. 1965. Gro up Dichotoma A. Hitchc. & Chase, Contr. U.S. Natl. Herb. 15: 179. 1910, nom. inval. Perennials with or without a basal rosette of shorter, broader leaves than those of the culms. Culms erect or decumbent-erect, usually freely branching at the upper nodes; internodes hollow, with or without glands. Sheaths striate, pilose or glabrous, with or without conspicuous glands. Lig- ules membranous-ciliate, the membranous portion small, the cilia variable in length. Blades linear to ovate-lanceolate, flat or inrolled, densely pilose to glabrous; foliar dimorphism present or absent. !n- florescence terminal and axillary, lax, open, oc- casionally contracted, axis of the branches with or without glands, pilose or glabrous. Spikelets ellip- soid to obovoid, pilose or glabrous, cleistogamous flowers present or absent. Lower glume of variable length, usually !4-!^ the in of the spikelet, up to % in several species, 1-7-nerved, acute or trun- cate, a short internode и the lower and upper glume present or absent. Upper glume acute to obtuse, usually not covering the apex of the upper anthecium, (5-)7-15-nerved. Lower lemma glumi- form, (5-)7-15-nerved. Lower palea conspicuous, as long as its lemma to small, absent in P. aciculare and P. penicillatum; lower flower staminate or absent. Upper anthecium indurate, smooth, shiny, with simple papillae all over its surface, pilose to- ward the apex of the lemma or glabrous, shortly 136 Annals of the Missouri Botanical Garden FIGURE 19. — Abaxial leaf blade ipe of Panicum viscidellum. — A. Prominent macrohairs common in the central files of the intercostal zones; these hairs have small cushions associated with their bases. — B. Long, slender macrohairs inserted between a few specialized epidermal cells. —C. Epidermal pattern showing narrow costal zones with files of stomata on eit her side; the central files of the wide intercostal zones with numerous long, slender macrohairs. —D. Detail of macrohair bases that are inserted between three or four inflated epidermal cells; numerous а = microhairs (А, based on Stevens 25420; B, based on Zuloaga et al. 4537; С, D, based on Zuloaga 4532; A, C x160; B, D x250). apiculate or crested; lodicules 2, conduplicate, 3(2). Inflorescences tracted; spikelets obovoid truncate; stamens 3; styles 2, free, the stigmas 2, 33. P. superatum plumose. Caryopsis with hilum punctiform, embryo 3. Inflorescences lax, a spikelets narrowly ellipsoid to ellipso 4(3) Spikelets 1.5-1. 9 mm long % or less the length of the caryopsis. Basic chro- mosome number: x = 9. Anatomy: С,, non-Kranz. A section with 37 species in Mesoamerica, the 4. Spikelets 3-6.8 mm long „i West Indies, and South America, inhabiting edges 9(4). Culms erect, 15-30(-60) cm tall; spikelets hirsute with papillose-pilose hairs; Mexico 19. P. pedicellatum of forest in west habitats or in open places, on moist or dry sandy soils. 5. Culms leaning on vegetation, to 2.2 m tall; rie glabrous; Brazil n : 6(5). Spikelets 3-3.3 mm long; lower palea 2.1- KEY TO THE SPECIES 2. mm que lower ae 3- Сет leaf E Upper glume and lower lemma 10-15- blades with scabrous и ried toward nerve 2 the base; the sheaths s hirs a M Upper glume and lower lemma (5-)7-9- i ace and cid. Ў Кога пегуе 8 Р: pu 2(1). Lower palea absent; culms scandent, lean- 6. Seibel 4.6-6.8 mm long; dd Hals 4- ing on vegetation, to 10 m tall; internodes 5.5 mm long; lower glume 5-nerved; leaf solid toward the base; leaf blades pseudo- blades with ciliate margins; the sheaths and petiolate; lower glume as long as or a little nodes glabrous; Brazil, Rio de Janeiro and shorter than the upper glume om Espirito Santo P. penicillatum 7(6). Spikelets 4.6-5 mm long; upper anthecium 2. Lower palea present; m Vd in to 4-4.2 mm long; leaf blades 10-12 cm lon erect or scandent, 0.15-0.60(-2.2) m tall... 17. P. itatiaiae internodes hollow; leaf blades not pseudo- Ta Spikelets 5.9-6.8 mm long; upper anthe- petiolate; lower glume 2-2 the length of cium 5.2 mm long; leaf blades 14-21 cm the spikelet 3 long 2 7. P. caparaoense Volume 80, Number 1 Zuloaga et al. 137 1993 Revision of Panicum subg. Dichanthelium sect. Dichanthelium FIGURE 20. Panicum viscidellum specimens with silicified cushion bas ses of the Ere pa D. Tra sectional leaf anatomy. — А. Leaf section showing margin and median vascular bundle (arrowed). — B. Detail of ba airs with cushion bases (arrowed): these cushion cells do not appear to be silicified; note the outer bundle sheath cells, which contain chloroplastlike bodies. —C. Section with chloroplasts in bundle sheath and extension cells; mesophyll relatively compact. — D. Section through silicified cushion-base cells (arrowed). E, F. Abaxial epidermi . Prominent macrohairs inserted into cushion bases, the cells of which are refractive, indicating silicification T fue with small, indistinct macrohairs, but with conspicuous cushion bases that contain crystalline material resembling opaline $ silica (A, B, E, based on Zuloaga & Londoño 4203; C, D, F, based on Zuloaga 4184; A x100; B-F x250 8(1). Spikelets usually geminate, at least in por- Denn: and spikelets not congested on the tions of the inflorescence; lower glume di- morphic; upper anthecium black at matu- 12(11). ee ellipsoid, 3-3.2 mm long, mi- rity; blades asymmetrical at the base ......... nutely puberulent; с ао short-ex- 15. P. hebotes serted or included on the upper leaves, 3- 8. Spikelets not geminate; lower dp iso- 5 cm long; Serra do Sol, Venezuela ........... morphic; upper anthecium whitish to pale, e Á 35. P. telmatum not black at maturity; blades adl sym- 12. Spikelets obovoid, 1.9-3(-3.2) mm long, metrical at the base ......... 9 glabrous; terminal inflorescences exserted, 9(8). Lower palea conspicuous, as long as or 3-20 cm long; eastern Brazil, Bolivia, Par- nearly as long as the lower lemma; lower aguay, Uruguay, Chile, and Argentina .......... 13 ower Pier. or absent . 10 13012). Robust plants leaning on vegetation; эн 9. Lower palea small or absent, ‘shorter than blades 10-22 cm long, 2c the lower lemma; lower flower absent ............. 14 spikelets glabrous, (1.9— је. 4-3.2 mm js 10(9) Lower glume not embracing the upper inflorescence 9-21 cm long, glandular; fo- g ume; stipe inconspicuous between the low- liar dimorphism absent .......... . P. stigmosum r and upper glume; spikelets narrowly el- 13. Caespitose or decumbent perennials no lipsoid, 1.8-2.2 mm long ..... 34. P. surrectum leaning on vegetation; leaf blades 1 -9(-15) 10. Lower glume DRE the upper glume; cm long, 0.2-1.4(-2) cm wide; spikelets stipe present between the lower and upper hirsute to glabrous, when glabrous 1.9- glume; spikelets obovoid to ellipsoid, 2.3(-2.6) mm long; inflorescences 2-10 (1.9-)2-3.2 mm lon . ll cm long, the axis glandular or eglandular; 11(10). Inflorescences contracte ed, spiciform or with foliar dimorphism present .. 25. P. sabulorum the first-order branches slightly divergent, 14(9). Plants branching only at the base, with the spikelets congested on the branches . — internodes and basal leaves numerous ane superatum in osette 11. Inflorescences lax, first- order branches di- 14. Plants normally branching above the base, 138 Annals of the Missouri Botanical Garden FIGURE 21. keel development due to presence of additional нт cells associated with the median vascular bundle. Widely spaced vascular bundles and very diffuse chlorenchyma tissue. wing narrow costal zones, wide intercostal zones, stomatal files, and macrohairs. — D. Leaf blade anatomy of Panicum boscii. B. Transectional anatomy. — А. Midrib showing slight =, C, D. Abaxial epidermis. — C. Epidermal pattern Detail of macrohairs and their associated basal cells, long, fingerlike microhairs, low-triangular subsidiary cells, p rectangular intercostal long 00). cells with sinuous walls (A-D, based on Davidse 30822; А x250; C x160; B, D x 15(14). 15. ate 16(15). with internodes elongated and basal leaves not in a rosette ї Blades linear, length : width ratio 10-30:1, with scabrous margins; MA black .... . P. ensifolium Blades lanceolate to linear- е length : width ratio 3-10:1, with nd cili- о and blades glabrous to sparsely pi- 32. P. strigosum Spikelets 2-2.7 mm long, short-hirsute; up- per anthecium 1.8-2.3 mm long; sheaths and blades densely pilose with long papil- lose-pilose hairs . P. laxiflorum . Spikelets obovoid; lower glume y- /(- У) the length of the spikelet 18 Spikelets ellipsoid: lower glume (/,-)%- et 2. P. acuminatum 19 Я оа linear, length : width ratio 10-20:1; er palea absent; inflorescences contract- E few-flowered 1. P. aciculare Blades ovate-lanceolate to lanceolate or lin- ear-lanceolate, length : width ratio 3- 10:1; lower palea present; inflorescences lax, few- to many-flowere 20 20(19). Mie glandular band present just below 26. P 20. 21(20). 2]. 22(2 22. 23(21). Spikelets 1-1.8 mm long; love er the 23. 1). . scoparium Glandular band absent ps the nodes ...... zi Blades 1-7 cm lon 5(-1.2) mm wide; inflorescences Es ы 1-8 сш long, 1-3 cm wide 22 Blades 2.5-10(-16) ст long, 0.5-1.7 cm hri inflorescences multiflowered, (2-)5- m long, 3.5-10 cm wide Blades narrowed at base; sheaths, nodes, and internodes glabrous to short-pilose; spikelets 0.6-1 mm wide; lower glume /- /, the length of the spikelet 12. P. dichotomum Blades subcordate; sheaths, nodes, and in- ternodes pilose; spikelets 1.1-1.3 mm wide; lower glume /,-/, the еп, of the spikelet P. portoricense glume Р, sphaeroca pm Spikelets (2.1-)2.3-3.2 mm long; glume ',-'4 the length of the KE axis glands; blades with papil- P. divergens ai 24(17). Blades clasping the culm, cordate, the (05 length : width ratio 3-5(- Zuloaga et al. 139 Revision of Panicum subg. Dichanthelium sect. Dichanthelium Volume 80, Number 1 1993 FIGURE irregularly dumbbell- shaped to nodular. ly E MES RINT а» ТЕЕ Ma «eg a E ims 22. Leaf anatomy of Panicum clandestin num. A-C. A a single vascular bundle. — B. Detail of diffuse mesophyll and hen pube bundle sheath extensions. — tomy in cross section. — А. Central keel Ewa —E. Thicker cuticle with less undulating long cell walls and wider costal zones; note microhairs, which are at least twice as long as the stomata (A, B, C, based on Davidse 30816, MO; D, E, based on Davidse 30818, MO; A х 100; B x 250; С, D, E x 400 24. Blades not clasping the culm, narrowed to ] mm ong с axis of the inflorescence subcordate, the length: width ratio 10-30:1 densely hirs . sciurotoides 30. Spikelets enel 1.7-2.5 mm longs inflo- (24). Upper glume and lower lemma 5(-7)-nerved, rescences 10-15 cm long; main axis gla- the nerves inconspicuous; lower lemma in- brous or short, sparsely pilose . 3l flated at base 26. P. sciurotis 31(30). Blades amplexicaulous, ovate- lanceolate, 1.5- 25. Upper glume and lower lemma 7-9(-11)- 3 cm wide, velutinous; upper anthecium 2.2 nerved, with T nerves; lower lemma mm lon 21. P. peristypum not inflated at bas 31. Blades not amplexicaulous, linear-lanceolate, (25). Inflorescences ie: flowered, 1-2.5(-3) cm .5-1.3 em wide, hispid or din upper long; blades overlapping, 1- 2: ст е, 0.3- anthecium 1.6-1.9 mm lon 0.6 cm wide .. . P. cumbucana | aequiv aginatum 26. Inflorescences mult "puma E 9-12 cm 32(29). Blades with margins cartilaginous and cov- long; usi not overlapping, 2-12 cm long, ered with manifest сШа ____ 31. P. stipiflorum 0.5-2 cm wide 27 32 Blades without cartilaginous margins an 27(26). Pzeudoligule manifest with long whitish hairs 28 dd cilia on the upper portions of the zi Pseudoligule absent 29 arg 2827). Spikelets 2.2-2.8 mm long 1... 33(32). Lu ome %-% the length of the spikelet; 36. P. umbonulatum inflorescences 3-6(-10) cm long; Ecuador, 28. Spikelets 1.5-2(-2.2) mm os , Peru, Venezuela, Brazil __ 24. P. pycnoclados P. visc scidellum 33. Lower glume /— the length of the spikelet; (27). сове 1. = 2.5 mm long, n am inflorescences (5-)10-22 cm long; United the lower and upper glume States, Mexico, Honduras _ Р: divergens 34(24). Lower glume ys У the length of the spikelet 29. Spikelets 2.2-3.3 mm long, with a id between the lower and upper glume | 34. Lower glume ¥—‘/ the length of the spikelet (29). Spikelets short-hispid, 1.5-1.9 mm long; i in- florescences 2.5-9 cm long; lower palea 0.7- (34). Pseudoligule present _____ 2. P. acuminatum 140 Annals of the Missouri Botanical Garden FicunE 23. expanded lamina showing median vascular bundle (arrowed). — sheath extensions. — C. Abax and bicellular microhairs; ti iu absent (A-C, based on Davidse 30819, МО; А x 35. 36(35). 38(37). 39(37). 40(39). 40. 41(40). 41. 42(34). 42. 43(42). Pseudoligule absent 36 Lower palea absent; inflorescences соп- tracted, few-flowered; blades linear ............ - = aciculare L palea present; inflorescences lax, od: blades окоро to unc lan e 1 . Plants densely hispid; spikelets hirsute ____ 38 Plants sparsely pilose or glabrous; spikelets short-hispid or glabrous Blades lanceolate, 3-15 cm long, 0.2-0.8(- m wide, flat, the margins ciliate; spike- lets 2-2.7 mm long; lower glume 14-14 the length of the spikelet; eastern United States to Costa Rica and the West Indies . P. laxiflorum m m 4-6 cm long, 0.2 wide, the margins involute, scabrous; Ein 2-23 mm long; r more the length of the ud Brazil . P. cabrerae Blades 4-6 cm long, 0.3-0. EN cm wide; plants with manifest glands on sheaths, gis sis axis of the e panicles 5- m lon P. adenorhachis ae 3. a 12- 16) cm ud 0.2-1.3(- 1.5) em wide; plants without glands, oc- casionally present on the axis of the pani- cles; panicles 3-22 cm long сл Spikelets glabrous, 1.7-2.5 mm long; Ven- ezuela, Guyana, and Brazil 0. aequivaginatum Spikelets pilose, (2.1-)2. 3- 4.1 mm long; United States, Mexico, and iade NM 41 Spikelets papillose-pilose, 3.3-4.1 m . ео dM —)2. 3- 3.2 mm ong 13. P. divergens Spikelets with a stipe between the lower and er glume Spikelets without a stipe between the lower Vin sparsely pilose, (2.1 and upper glume Spikelets glabrous, 2.2- 2.7 mm long; lower glume 1(-3)-nerved; panicles lax, blades regularly spaced on the culms 9. P. cucaense Leaf blade anatomy of Panicum oligosanthes. A, B. Transectional anatomy. — A. 43. 44(42). — Central part of —B. Detail of diffuse mesophyll and biseriate bundle ial epidermis with dumbbell-shaped costal silica bodies, low triangular subsidiary cells, 00). 100; B, C x Spikelets hirsute, 2.8-3.2 mm long; lower lume 5-7-nerved; panicles contracted, blades congested on the culms _______- 8. P. congestum seas 10- = cm tall; blades 2-5 cm long, wide; panicles 2-4 cm long, des “эне 2.4- x Р. - petropolitanum Culms 50-1,000 cm y es 4-32 long, 0.3-3.2 cm wide; panicles 4-45 cm long, us pilose or glabrous, 2.6-4.2 45 Love palea absent; blades 18-32 cm long, Е? m wide, pseudopetiolate; panicles 30- 45 em dong. ————2 20. P е оа Lower sips present; blades 4-12 cm lon 3-1 wide; panicles 4-16 cm long ....... 46 | Spikelets | hispid, 2.6-3.1 mm long; upper glume and lower lemma 7-nerved; blades m lanceolate, 4-5(-12) cm long, 0. m 1) cem wide, peduncle and e x hirsut > heliophilum Spikelets glabrous to ue pilose be- ves, 3.3 mm long; upper glume lanceolate, 8-12 cm long, 0.7-1 cm wide, peduncle and main axis glabrous о 5. P. assurgens . Panicum aciculare Desv. ex Poiret, Encycl. Suppl. 4: 274. 1816. Dichanthelium acicu- lare (Desv. ex Poiret) Gould & Clark, Ann. Missouri Bot. Gard. 65: 1116. 1978. TYPE: "Habitat in India Orientali" (holotype, P; frag- ment, US 2808908). Plants perennial, with or without a basal rosette of broader leaves. Culms erect to decumbent and geniculate at base, then erect, freely branching, 10-30(-65) cm tall, internodes hollow, terete, densely covered with short hairs, greenish to pur- plish, nodes compressed, brownish, pilose to gla- brous. Sheaths shorter or longer than the inter- Volume 80, Number 1 1993 Zuloaga et al. 141 Revision of Panicum subg. Dichanthelium sect. Dichanthelium FIGURE 24. | жеш micrographs of upper Ашын of E ene of Panicum. — А. Panicum assurgens x 50 (from Harley et al. 15229). —B. Panicum a х 75 (based on жек y et al. 17383).—C. И cumbucana X100 ету оп i Hatley et al. 16869 D. Panicum surrectum 0 (based on Chase 9380). E Panicum divergens. —E. Upper portion of the lemma x 50. nodes, greenish to purplish, with whitish appressed hairs or glabrous, striate, one margin ciliate, the other glabrous, the upper margins long-ciliate. Lig- ules 0.5-1.2 mm long, short-membranous at base and long-ciliate at the apex; collar densely pilose. Blades linear, 2.8-7(-13) ст long, 0.2-0.4 ст wide, flat or with the margins involute, long-ciliate to scaberulous, attenuate at the base, adaxial sur- face short-pilose or with scattered papillose-pilose hairs, abaxial surface scaberulous to glabrous. /n- . Apex of the на x 250 (based on Joor ps florescence terminal, exserted, peduncle cylindric, or short-exserted; 1.5-4.5 cm short- pilose or scabrous, pulvini short-pilose, first- and glabrous, up to 25 cm long, panicles few-flowered, contrac ted, long, 2-2.5 cm wide; main axis flexuous, second-order branches flexuous, short-pilose or scabrous, pedicels claviform, pilose. Axillary pan- icles similar to the terminal one. Spikelets solitary, obovoid to ellipsoid, 1.9-2.8(-4.2) mm long, 1.3 mm wide, greenish or tinged with purple, densely 142 Annals of the Missouri Botanical Garden ^ Panicum aciculare var. aciculare O P. congestum np assurgens * P. aequivaginatum * P. adenorhachis * P.aciculare var. arenicoloides & P. penicillatum о 200 400 600 800 1000km =] O 100 200 300 400 500 600 miles Prepared by Hendrik А. Rypkema 40 m ho m [20 7—40 e NB * — ho И " wc * “RS — — Lo d 40 FIGURE 25. Distribution of Panicum aciculare var. aciculare, P. congestum, P. assurgens, P. aequivaginatum, P. adenorhachis, P. aciculare var. arenicoloides, and P. penicillatum. pilose or glabrescent. Lower glume ovate, 1.3 mm long, 43-14 the length of the spikelet, acute to truncate, nerveless to 3-nerved, with a small in- ternode between the lower and upper glume. Upper lume ovate, obtuse, 7-nerved, hispid on the ab- axial surface. Lower lemma 7-nerved, acute. Low- er palea absent; lower flower absent. Upper an- thecium obovoid, 1.8-2.1 mm long, 1.2 mm wide, pale, shiny, pilose toward the apex of the lemma, short-apiculate, lemma 7-nerved. Caryopsis obo- void, brownish, 1.1-1.3 mm long, 1.1 mm wide, hilum punctiform, embryo less than half the length of the caryopsis. Distribution and ecology: southern United States, Mexico, Mesoamerica, the West Indies, Colombia, and Venezuela, at edge of forests or in open, wet, and sandy places between sea level and : m. Although Poiret mentioned in the original de- scription that the species was from “India Orien- tali," it was obviously collected in the West Indies. KEY TO THE VARIETIES 1. Spikelets 1.9-2.8 mm long, without a manifest internode between the lower and upper g la. l. Spikelets 2.8-3.3(-4.2) mm long, w ternode 0.3-0.6 mm long between the lowe upper glume cc lb. var. Mint la. Panicum aciculare var. aciculare. Fig- ures 1, Panicum neuranthum Griseb., Cat. Pl. Cuba: 232. 1866. TYPE: Cuba. Oriente: without locality, anno 1860, Wright 3453 (lectotype, GOET not seen; isolecto- types, G, K, NY). Panicum arenicolum Ashe, J. Elisha Mitchell Sci. Soc. l 6. 1898. TYPE: United States. North Carolina: in is vicinity of Chapel Hill, Ashe s.n. ps NY). Volume 80, Number 1 1993 Zuloaga et al. Revision of Panicum subg. Dichanthelium sect. Dichanthelium P. acuminatum var. longiligulatum P. acuminatum var. villosum P. cabrerae Dr pope nr — нек | 00 miles | $200 400 вао 800 1000km EH] 40 10, ' rik А. Rypkema Lo | FIGURE 26. acuminatum var. villosum, and P. cabrer Panicum МР не qr e in Small, Fl. Southeast. U.S.: 100, 1327. 1903. TYPE: United States. Flor- ida: Leon Co., Lake Ja ш. 12 May 1886, Curtiss n. (holotype, NY; isotype, US 2808891) Selected specimens cit | ВАНАМА5. New Provi- dence, swamp near Tea House, Britton et al. 599 (K). Chico Valley, 6 km МЕ of San Pedro, Wood 53 0 m, -Diaz et al А: anó, in poids „кара; Ekman 12581 (US). ISLA DE LA АТУУ ENTUD: abana de la G S). DoMiNICAN REPUBLIC. LA н ее gis 16089 (US). GUATEMALA. ZACAPA: Sier e las Minas, Steyermark 29707 (05). Hartı. "Massif de la Selle, De 6855 (G, z E i P un x. o У о a n Yus 1,500 m, Standley 29366 (F, GH). FRANCISCO MORAZAN: Distribution of Perens acuminatum var. acuminatum, P. acuminatum var. longiligulatum, P. between Cuesta de Los Muertos and Monte Obscuro, near La Montañita, Molina 14707 (US). Jamaica. Mason River Savanna, 2.75 mi. due NW of Kellits, 2,100-2,300 ft., Proctor 26565 (US); Pick Lam, by oed prd Нета 12812 (US). NICARAGUA. MATAGALPA: 5 km № of Santa Maria de Ostuma, between Matagalpa and ie Wil- liams et al. 23906 (G). Gen Rico. Las Marias, Sin- tenis 5985 (G, US, W); Mont t vi сину of Ма pd güez, Chase 6273 (US). Олко STATES. FLORIDA County, vicinity of Eustis, Nash 1243 (P у NORTH CARO- LINA: E of Wil Gr. Hb. raa sieh 1,300 m, Pardon 151 (US). DISTRITO FEDERAL: 5.5 km down road to Carayaca, between nia Tovar, Davidse & Tillett 4070 (MO). FALCON: sierra de San Luis, Montana de Paraguariba, entre el hotel Parador y Curimagua, 1,300 m, Steyermark 85871 (VEN). LARA: selva siempreverde a lo largo de la quebrada sobre rocas calcareas, 7 km de Barbacoas, entre Barbacoas у La Pena, Steyermark 85862 (VEN). MIRANDA: Sebastopol, Badillo s.n. (US 1760675). PORTUGUESA: Dtto. Sucre, Fila del Helechal, 1,500-1,600 m, Ramírez-Reyes 2513 RT) El Junquito and Co lon Panicum aciculare is anatomically distinct from the rest of species of section Dichanthelium in having a narrow, inrolled leaf blade, stomata ab- sent, and intercostal short cells occurring between 144 Annals of the Missouri Botanical Garden Un oe A Panicum cucaense * P.cumbucana * P. davidsei Л P. ensifolium | С P. hebotes | V P. heliophilum | e P.laxiflorum | * P. pedicellatum 4 P. portoricense о 200 400 600 800 1000km [A is tl a air |u mg, ur ems es, cos | O 100 200 300 400 500 600 miles E 27. Distribution of Panicum cucaense, P. cumbucana, P. davidsei, P. dichotomum, P. divergens, P. ensifolium, P. hebotes, P. heliophilum, P. laxiflorum, P. pedicellatum, and P. portoricense. the rectangular long cells. The mesophyll tissue also exhibits no pattern of cellular arrangement lb. Panicum aciculare Desv. ex Poiret var. arenicoloides (Ashe) Beetle, Phytologia 48: 192. 1981. Panicum arenicoloides Ashe, J. Elisha Mitchell Sci. Soc. 16: 89. 1900. TYPE: United States. North Carolina: New Hanover Co., near Wilmington, 7 June 1899, Ashe s.n. (fragment and photo, US). Figure 25. Panicum angustifolium Elliott, Sketch Bot. S. Carolina 1: 129. 1816. Panicum nitidum € var. angus- _ (Elliott) A. Gray, М. Amer „2:112. . Chasea angustifolia (Elliott) Nieuw., Amer. Midi Naturalist 2: 64. Dichanthelium an- с а (Elliott) Gould, Brittonia 26: 59. 1974. Carolina: without locality, Elliott s.n. oe CHARL not seen; fragment and photo, US). Panicum neuranthum Griseb. var. ramosum Griseb., Cat. Pl. Cuba: 232. 1866. Dichanthelium aciculare (Desv. ex Poiret) Gould & C. A. Clark var. ramosum (Griseb.) Davidse, Novon 2: 104. 1992. Panicum fusiforme А. Hitchc., Contr. U.S. Natl. Herb. 12: 222. 1909. туРЕ: Cuba. Without locality, anno 1863, к 3454 (holotype, GOET not seen; isotypes, , , P, US, W, fragment and photo, US P REN This variety has a similar habitat and distribution to that of variety aciculare. Selected specimens cit ELIZE. EL CAYO: Moun- tain Pine Ridge, ue n Agustin, ‘Lundell 656 1 (US). ORANGE : savanna ca. 5 km August Pine Ridge on the road to Trin idad, Davidse & Brant 32806 (MO). TOLEDO: Swasey Branch, Monkey River, Gentle 3903 (MO). CuBa. ISLA DE LA JUVENTUD: San Pedro, Britton & Wilson 14305 (US). ORIENTE: Sierra Maestra, Pinar de Bayamita, Ek- man 10329 (US). PINAR DEL RIO: 17 km 5 of z: del Rio, Hitchcock 23257 (US). SANTA CLARA: saban Manacas, León € Cazañas 5840 (US). GUATEMALA. Volume 80, Number 1 1993 Zuloaga et al. Revision of Panicum subg. Dichanthelium sect. Dichanthelium Panicum itatiaiae A P. peristypum | Ш P. petropolitanum | „= * P. pycnoclados | * P. sciurotis * P. sciurotoides | Ж P. scoparium | = | m— | | | | SS ee а | — | | Ed | \ | \ | | о 200 400 600 800 !000km | | L M vas so | \ \ = о 100 200 300 400 500 600 miles \ \ \ \ Е 28. анн of P. itatiaiae, P. peristypum, P. petropolitanum, P. pycnoclados, P. sciurotis, P. um. FIGURE sciurotoides, and P. scopar CHIQUIMULA: on moist pine forest bank between Guate- mala-Honduras border and Atulapa, 900 m, Molina & Molina 25283 (MO, US). HONDURAS. COMAYAGUA: Co- conita pass above Siguatepeque, sand db Williams 18842 (US). coPAN: Hacienda Espiritu o to Quebrada Mojanales, Blake 7458 (US). FRANCISCO MORAZAN: faldas de la Mt. Uyuca, entre Las Flores y Quebrada El Granadi- llo, Williams & Molina 14813 (US). ocoTEPEQUE: rocky hillsides of Machuca, 900 m, Molina & Molina 27868 (МО, US). Jamaica. Halliss Savanna, Upper Clarendon, Harris 12234 (US); Bull Head Mountain and vicinity, Pa 9552 (US). MEXICO. CHIAPAS: Mun. Palenque, of Palenque on road to Catazaja, Breedlove & аА 55333 (МО). NICARAGUA. San Rafael del Norte, 1,200-1,350 m, Miller & Griscom 132 (US). VENEZUELA. MIRANDA: eae Badillo, Aug. 1938 (US). 2. Panicum acuminatum Sw., Prodr. 23. 1788. Dichanthelium acuminatum | (Sw.) Gould & Clark, Ann. Missouri Bot. Gard. 65: 1123. 1978. TYPE: Jamaica. Without locality, Swartz s.n. (holotype, S not seen; fragment and photo, US 2808906; isotype, M). 10-35(-60) ст tall; culms decumbent and geniculate, then erect, branching at the middle and upper nodes; internodes 2-5 cm long, hollow, striate, densely hirsute or glabrous, nodes densely pilose with whitish retrorse hairs or glabrous. Sheaths 1—5 cm long, striate, greenish to purplish, short-pilose, hirsute, the margins large- Plants perennial, ly ciliate or glabrous. Ligules membranous-ciliate, 0.2-1 mm long, with a pseudoligule formed by long, whitish hairs up to 4 mm; collar puberulous to densely pilose. Blades linear-lanceolate to lan- ceolate, 2-6.5(-9) cm long, 0.2-0.8(-1) cm wide, subcordate, acute, flat, long-hispid to glabrous on the adaxial surface and pilose on the abaxial one, the margins scabrous to ciliate, the basal ones long- ciliate. Inflorescence terminal, long-exserted, pe- duncle hispid, up to 30 cm long; panicles lax, 146 Annals of the Missouri Botanical Garden P» Р. stipiflorum t Р. strigosum O 200 400 воо воо 1000 m 400 500 800 mii Prepared by Hendrik А Rypkema FIGURE 29. Р. и P. stipiflorum, and P. strigos diffuse, multiflowered, 2-4.5(-9) cm long, 1-4(- ) ст wide; main axis flexuous, long hispid to nearly glabrous, glands occasionally present, first- and second-order branches flexuous, glabrous to sparsely pilose, glands occasionally present, pulvini densely pilose, pedicels terete, sparsely pilose or glabrous. Axillary panicles numerous, smaller than the terminal one, short-exserted. Spikelets obovoid or ellipsoid, (0.9-)1.5-2(-2.4) mm long, 0.7-1 mm wide, hispid, greenish to purplish, without an internode between the lower and upper glume, up- per glume and lower lemma subequal. Lower glume ovate-acuminate, 0.6-0.8 mm long, 44-14 the length of the spikelet, obtuse to truncate, 1(-3)- nerved. Upper glume ovate, 1.5 mm long, 9-nerved. Lower lemma glumiform, 7-9-nerved. Lower palea lanceolate, small, 0.6-0.9 mm long, 0.3 mm wide, hyaline, glabrous, lower flower ab- sent. Upper anthecium ovoid, 1.3-1.5 mm long, 0.9 mm wide, globose, pale, shiny. Caryopsis el- lipsoid, 1 mm long, 0.7 mm wide, hilum puncti- form, embryo У, the length of the caryopsis. hie eed | m Bu 0 100 200 | uL | 9 9 "m | EN Aart HM жые | 7 = A СА ( | | Distribution of Panicum оми var. floridanum, P. sphaerocarpon var. sphaerocarpon, KEv TO THE VARIETIES 1. Spikelets 0. 9-1.5 mm 2 ои pera Or sparsely pilose with hairs mm lon pcne cu ЖӨ 1. Spikelets 1.5-2.4 mm long, blades usually densely pilose 2. Spikelets 1.5-1.9(-2.4) mm e the leaves not densely pilose, the hairs only 0 d: m long 2a. var. acuminatum 2. Spikelets 1.9-2.4 mm LT leaves dene pilose with hairs 2-4 mm lon . 2c. var. villosum 2a. Panicum acuminatum var. acumina- tum. Figures 2, 3, 4, 26 Panicum tennesseense Ashe, J. Elisha Mitchell Sci. = 92792. 98. Panicum lindheimeri Nash v tennesseense (Ashe) Farwell, Amer. Midl. Кашы 11: 45. 1928. TYPE: United States. Tennessee: La Vergne Co., 7 Aug. 1897, Biltmore Herb. 7087 (holotype, US). Panicum | Nash, Bull. Torrey Bot. Club . 1903. TYPE: Puerto Rico. Santurce, 9 Jan. а. Holas Hilo 12 (holotype, NY; fragment, US 80593) Volume 80, Number 1 1993 Zuloaga et al. Revision of Panicum subg. Dichanthelium sect. Dichanthelium 10 100 90 30| N • |--- > pud и ^^ ss T ЖЫ B M и у = . 4 - x— . Фф n (| өө. . „Ре А >; ss 7e e. | _ _ === .. po | | | ¿AA e d | | | | A Panicum surrectum e P.viscidellum Е \ | | | ee \ \ \ \ === | о 200 400 600 800 1000km \ | O 100 200 300 400 500 600 miles | 100 90 FIGURE 30. Panicum olivaceum A. Hitchc. & Chase, Contr. U. Natl. Herb. 15: T e TYPE: Guatemala. Alta Verapaz: Сођап, 1,400 m, Feb. 1888, von Tuerck- heim 428 o s fragment, US 823309; isotype, US). Distribution and ecology: United States, Mex- ico, the West Indies, Mesoamerica, and northern South America in Colombia, Ecuador, and Vene- zuela, growing in forests, moist banks, and open ground, from sea level to 2,800 m. Selected specimens cited. BELIZE. BELIZE: Belize In- ternational Airpo . CA YO: Mountain ant Gentle 8551 (US). TOLEDO: iiw Ri Creek, Gentle 4122 ( US). Cocos ANTIOQUI S). ACA: entre Chinavita y Tibaná, 1,700 m, Zulo 4120 (COL, MO, SI). cauca: de Balbo de Balboa, 2,180 m, Zuloaga & Гопаопо 4202* (MO, Distribution of Panicum surrectum and P. viscidellum. SI). CUNDINAMARCA: subida a Alto del Tigra, 1,700 m r la ca К. Londoño 7281 (COL). VALLE de CAUCA: Hoya del Rio Cali, Pichinde, Cuatrecasas 18559 (US). COSTA RICA. CARTAGO: La Estrella, Standley 39320 (US). GUANACASTE: Cordillera de Guanacaste, Volcán Rincón de la Vieja, Hacienda Guachipelin, Pohl & Davidse 11676 (US), MO). SAN JOSE: between Aserri and Tarbaca, 900 m, Standley 34048 (US), 1,675 m, Davidse & pe 972 (MO). СОВА. ISLA | LA JUVENTUD: near Nueva Gerona, Curtiss 307 (G, US). ORIENTE: jn Maestra, Pinar de лы Ekman 10330 (G, US). NAR DEL RIO: environs de Sum 3472 (P, US). SANTA CLARA: at the mi Ekman 16841 (US). DoMINICAN REPUBLIC. Loma de la Sal, 1,300-1,400 m, Liogier 13365 (NY). LA VEGA: vicinity of Jarabacoa, 500-1,200 m, Allard 14508 (US). SAN JUAN: Sabana Nueva, Cordillera Central, N of Rio Arriba, 6,200 ft., Howard 9162 (P, US). ECUADOR. IM- 148 Annals of the Missouri Botanical Garden no 100 90 ~ 30 ~ 30 iy 4^] _ — > y A У 20 „к - ER m-— 9 ~ = © d XT | | ~ | Eu Е » пор | | —— g | Sut Emm „| — -————— Pone * Panicum superatum * P. telmatum © P.umbonulatum 0 NT — 1n 207 ——~30 ыйы rari \ \ 100 200 300 400 500 6 З | | \ \ \ A 100 90 FIGURE 31. Distribution of Panicum superatum, P. telmatum, and P. umbonulatum. BABURA: Collapi, 840 m, Acosta-Solis 12834 (F, US). PICHINCHA: without aur d Padilla 1122 (MO). Er Sat- ADOR. CHALATENANGO: along Hwy. 4, 4 km SSE of La Palma, Pohl & Davidse 11894 (MO). “Gu JATEMALA. ALTA VERAPAZ: along Rio Сасћа, 1,360 m, Standley 89929 (US). EL QUICHE: 2 km NW of Nebaj, Metzler 21 (MO). Steyermark 49025 (US). IZABAL: trail from Los Amantes to Izabal, Blake 7739, 7762, 7780 (US). PETEN: 1 km N of Роршп, Harmon & Dwyer 2708 (MO). SOLOLA: pine woods bordering Rio Bravo, in vicinity of Finca Mocá, S-facing slopes of Volcán Atitlán, 1,000 m, Steyermark 47958 (US) Haiti. Massif de la d Grand-Cosier, Ekman 6856 (G). HONDURAS. COMAYAGUA: Zacate, La Libertad, Caballero 124 (MO). COPAN: acenda Espiritu Santo, Blake 7436 (US). EL PARAISO: Las Casitas, Swallen 11075 (US). FRANCISCO MORAZAN: Mt pelas, near Za- m Hitchcock s.n., Amer. Gr. Hb. 133 (P, US, W). MEXICO. CHIAPAS: Mun. Las Margaritas, pes albergue Tziscao, along W side of Lago Tizscao, 1,450 m, Davidse et al. 29845 (MO). TABASCO: near the | km post W of Hui manguillo on the Huimanguillo- Francisco Rueda road, Davidse & Davidse 9368 (MO). NICARAGUA. CHONTALES: along road from Jutigalpa NE toward La Libertad, ca. 17.4 km NE of Rio Mayales, at ford of Rio El Bizcocho, Stevens 4118 (МО). ZELAYA: Puerto Isabel, Atwood 2941 MO). PANAMA. CHIRIQUI: El Boquete, Killip 4525a (US), Davidson 766 (MO). Puerto Rico. Maricao, Sintenis 355 E M, US, W). UNITED STATES. ARKANSAS: Clark Count „Барат. 66629 (MO). MARYLAND: Without lo- cality, {кш 2490* (SI). MISSOURI: Franklin County, Shaw Arboretum of the Missouri Botanical Garden, near Gray's Summit, Davidse & Ellis 30807*, 30825* (MO). VIRGINIA: Alexandria, Zuloaga 2492" (SI). VENEZUELA. ARAGUA: Colonia Tovar, 1,900 m, Pittier 9962 (VEN). TRITO FEDERAL: Sabanas de El Junquito, 2,000 m, Рипег 13773 (VEN). MIRANDA/DISTRITO FEDERAL: Cam- pamento Las Rocas, below Topo Galindo, W of Pico Naiguata, Cordillera de la costa, Tillett et al. 27 (VEN). MERIDA: 7 km SE of Santo Domingo along Mérida- Barinas highway, 1,700 m, Davidse 3201 (MO). TRUJILLO: La Cristina, Tamayo 1861 (US, VEN). ~ Volume 80, Number 1 1993 Zuloaga et al. 149 Revision of Panicum subg. Dichanthelium sect. Dichanthelium * Panicum sabulorum var, cordatum . P. sabulorum var. polycladum A Р sabulorum var. sabulorum O 200 400 600 800 1000km i) r O 100 200 300 400 500 600 miles | FIGURE 32. var. sabulorum A sample of five specimens from both North America and South America was studied anatom- ically. The anatomical variation exhibited by the specimens originating from either continent is no greater than the variation observed between spec- imens from the same continent. The transectional anatomy clearly shows this trend. Only two of the specimens from the United States have a slightly differentiated midrib (Fig. 3A), whereas the other two North American specimens, as well as the South American specimens, all only have median vascular bundles (Fig. 2A). The semiradiate nature of the chlorenchyma tissue arrangement and bun- dle sheath extensions are consistent features of all the specimens. Panicum acuminatum is the only species stud- ied in which the anatomy of the basal rosette leaves was compared with that of the upper culm leaves. This comparison indicates that the rosette leaves Distribution of Panicum sabulorum var. cordatum, P. sabulorum var. polycladum, and P. sabulorum lack adaxial ribs and furrows and have a denser mesophyll than the upper leaves (Fig. 4). 2b. Panicum acuminatum var. longiligu- latum (Nash) Lelong, Brittonia 36: 270. 1984. Panicum gc aere Nash, Bull. Torrey Bot. 6: 574 99. Dichan- thelium acuminatum (Sw.) с & Clark var. longiligulatum (Nash) Gould & Clark, Ann. Missouri Bot. Gard. 65: 1127. 1978. Dichanthelium longiligulatum (Nash) Freck- mann, Phytologia 48: 102. 1981. TYPE: Unit- ed States. Florida: Franklin Co., Apalachicola, anno 1892, Vasey s.n. (holotype, NY; photo and fragment, US 78339). Figure 26. Panicum = Nash, Bull. Torrey Bot. Club 24: 4 . Pan = m acuminatum “ leuco- Ke E long, Brittonia 36: 27 84. Di- chanthelium aie pel (Nash) "ыч Phy- 150 Annals of the Missouri Botanical Garden tologia 58: 101. 1981. TYPE: United States. Florida: sis Co., near Eustis, July 1894, Nash 1338 (ho- lot e, NY; isotypes, US 208336, 742828, 742830). Bun. rightianum Me U.S.D.A. Div. Agrost. Circ. Dichanthelium acuminatum (Sw.) Gould & n var. wrightianum Sues Gould & Clark, Ann. Missouri Bot. Gard. 65: 1126. Freckmann, Phytologia 48: 101. 1 Without locality, Wright 3463 (holotype, US 2808947; isotype, С). Panicum pilatum Swallen, Fieldiana, Bot. 28: 26. 1951. TY d теша, rud i Turimiquire, rocky ,500 m, 6 May 194 5. ане. 62606 Do. US 1911673; iso- types, F, NY). Distribution and ecology: United States, Mex- ico, the West Indies, Mesoamerica, Colombia, and Venezuela, 0-2,500 m, in open, wet savannas on sandy soils and forest edges. Selected specimens cited. BELIZE. BELIZE: ca. 6 km N of Sand Hill along the old Northern Highway to Maskall, Davidse & Brant 32844 (MO). cayo: Mountain Pine Ridge, Baldy Beacon and vicinity, Davidse & Brant 33069 MO, SI). STANN CREEK: in pine ridge, Commerce Bight Pine Ridge, Gentle 8170 (US). TOLEDO: savanna between Chun Bank and ‚ Killip & Smith 15193 (MO, US). CUBA. ISLA DE LA IVENTUD: Santa Bárbara, white-sand savannas, Alain & Killip 2108 (US). PINAR DEL RIO: Arroyo Mantua, Ekman Ab (G, ee DOMINICAN REPUBLIC. Cordillera Central, v. de S a, Sabana de la Маг, Ekman 15636 (C, US). Шол RAS. CRAC IAS A DIOS: alrededores de Puerto vin vd lal Миа MEXICO. CHIAPAS: Mun. Oco- sin 6 km oca Lacantum, camino a Pale aen rud p 348 (MO). NICARAGU A. Comarca del Cabo, Puente Pozo Azul, Seymour 4576, 4741, 4743 (MO). zELAYA: Puerto Ca id ezas, € by (MO, Vee ZUELA. ANZOATEGUI: Distrito Libertad, ridges a tops of Montañas Negras, "along He Sucre and Anzoátegui border, ela iss of Ber jen , NE of Bue Aires, Serrania de Turimiquire, Davidse & González 19537, 19611 (MO, SD. LARA: Dtto. pond hacia las "filas de las vacas," caserio La Pena a 16.5 m de km de Humocaro Alto, Burandt Jr. V0383 (МО). 2c. Panicum acuminatum var. villosum (А. Gray) Beetle, Phytologia 48: 192. 1981. Pan- icum nitidum Lam. var. villosum А. Gray, N. Amer. Gram. & Сур. 2: 111. 1835. Di- chanthelium acuminatum (Sw.) Gould & Clark var. villosum (A. Gray) Gould & Clark, Ann. Missouri Bot. Gard. 65: 1124. 1978. Pani- cum ovale Elliott var. villosum (A. Gray) Le- long, Brittonia 36: 272. 1984. TYPE: United States. New York: Ontario Co., without lo- cality, Sartwell s.n. (isotype, MO). Panicum villosissimum Nash, Bull. Torrev Bot. Club 23: 149 1896. TYPE: United States. Georgia: Bibb Co., Ocmulgee River, swamp below Macon, 18-24 May 1895, Small s.n. (holotype, NY; isotypes, NY, US). Panicum pseudopubescens Nash, Bull. Torrey Bot. Club : 577. 1899, туРЕ: United States. Alabama: Lee о., Au bii n, 7 May 1898, Earle & Baker 1537 (holotype, NY; isotype, US). Distribution and ecology: widespread in the eastern United States; also in Mexico, Honduras, Guatemala, 2,500 m and Nicaragua, in forests up to Selected specimens cited. GUATEMALA. CHIMAL- TENANGO: near Rio Pixcayó, between Chimaltenango and San Martín Jilotepeque, Standley 64360 (F). HUEHUETE- NANGO: around Laguna de Ocubila, E of Huehuetenango, 1,900 m, Standley 82760 (F, US). HONDURAS. FRANCISCO MORAZAN: vicinity of Suyapa, 1,200 m, Swallen 11287 MO). MEXICO. CHIAPAS: Mun. Teopisca, marsh near Teo- m, Breedlove & Davidse 54820 (MO). mi. SW of Sola de Vega along the road to Puerto Escondido, pine-oak forest on mountain slope, 2,080 m, Davidse & Davidse 9688 (MO). SAN LUIS POTOSI: San Luis Potosi, Schaffner 146 (NY). NICARAGUA. ESTELI: along new road from Hwy. 1 (at km 135.5 and ca. 10.6 km W of bridge at La Trinidad) to San Nicolás, ca. 9.5 km from Hwy. 1, 1,200- 1,400 m; pine-oak forest, Stevens & Montiel 14796, 17972 (MO). NUEVO SEGOVIA: 3 km N of Dipilto, pine forest on steep hills, 900 m, Pohl & Davidse 12199 (MO). ~ 3. Panicum adenorhachis Zuloaga & Mor- rone, Ann. Missouri Bot. Gard. 78: 154. 1991. TYPE: Brazil. Bahia: Mun. Rio de Contas, 6- 10 km ao NO de Rio de Contas, na estrada para o Pico das Almas, 13°32’S, 41?53'W, 1,000 m, 21 July 1979, Mori, King, dos Santos & Hage 12451a (holotype, CEPEC; isotypes, MO, 05). Figure 25. Plants perennial; culms trailing, leaning over vegetation, many-noded, sparingly branched, in- ternodes cylindric, hollow, glabrous, nodes brown- ish, short-pilose or glabrous. Sheaths striate, ca- ducous, glabrous, covered with small glands, the margins long-ciliate; collar short-pilose. Ligule 0.2— 0.6 mm long, a ciliate membrane. Blades narrowly lanceolate, 4-6 cm long, 0.3-0.5 cm wide, flat, subcordate, d adaxial surface scabrous, the ab- axia gl nd dira nn small glands; margins long cil ciliate with tuberc airs near base, otherwise scabrous. /nflorescence terminal, a lax, diffuse panicle 5-7 cm long, 4-5 cm wide, few- flowered; main axis flexuous and glandular, the nodes distant; first-order branches widely diver- gent, solitary at each node with scabrous and glan- dular axes and long-pilose axils; spikelets solitary and widely spaced; pedicels claviform, short-pilose. Axillary panicles similar to the terminal one. Spike- lets narrowly ellipsoid, 2.7-3 mm long, 0.9 mm Volume 80, Number 1 1993 Zuloaga et al. 151 Revision of Panicum subg. Dichanthelium sect. Dichanthelium wide, upper glume and lower lemma short-hispid and subequal, without a stipe between the lower and upper glume. Lower glume 0.7-0.8 mm long, less than / the length of the spikelet, acuminate, nerveless or l-nerved, hyaline. Upper glume 9-nerved. Lower lemma 9-nerved. Lower palea lanceolate, 1.8 mm long, 0.4 mm wide, hyaline, ciliolate on the margins; lower flower absent. Upper anthecium narrowly ellipsoid, 2.3 mm long, 0.8 mm wide, stramineous, brownish at maturity, papil- lose with simple papillae in longitudinal rows; lem- ma short-mucronate and with small microhairs at the apex; rachilla prolonged beyond the upper an- thecium as a short mucro. Caryopsis ellipsoid, brownish, 1.5 mm long, 0.7 mm wide; hilum punc- tiform, embryo less than half the length of the caryopsis. Distribution and ecology: known only from Pico das Almas in Bahia, Brazil, rupestres on rocky, humid soils, 1,000-1,850 m. growing in campos Additional specimens examined. BRAZIL. BAHIA: Alto do Pico das Almas, 1,850 m, 20 Feb. 1987, Harley et al. 24460 (K); Mun. Rio de Contas, Pico das Almas, Vertente leste, subida do pico do Campo do Queiros, 13°32’S, 41*58'W, Harley et al. 26437 (K, MO, SI). Affinities of P. adenorhachis are discussed in Zuloaga & Morrone (1991). 4. Panicum aequivaginatum Swallen, Contr. U.S. Natl. Herb. 29: 271. 1949. TYPE: Brazil. о Маг. 1943, Fróes 19950 (holotype, US 1910768; photo of the holotype, K; isotype, US 2140782). Figures 5, 24B, 25 Panicum appressifolium Swallen, Mem. New York Bot. Gard. 9: 258. 1957. TYPE: Guyana. Samwarakna- tipu (Holi- -tipu), Kamarang River, Wenamu Trail, 1,100 m, 10 Nov. 1951, Maguire & Fanshawe 32564 (holotype, F 1449017; isotypes, K, NY, P, RB). Panicum belmonte Renvoize, Kew Bull. 37: 325. 1982. TYPE: Brazil. Bahia: 24 km SW of Belmonte, on un to Itapebi, 24 Mar. 1974, Harley et al. 17 Eo eee es CEPEC 10051; isotypes, K, MO, 2955 Panicum а ilum Renvoize, Kew Bull. 39: 180. 1984 PE: Brazil. Bahia: Mun. Salvador, Dunas . 1976, Araujo et al. 96 (holo- type, CEPEC; isotypes, CEN, K). Apparently annual, 20-100 cm tall. Culms de- cumbent, rooting and branching at the lower nodes, then erect; internodes hollow, 3.5-10.5 cm long, short-hispid or glabrous, nodes compressed, pilose with appressed whitish hairs. Sheaths 3-8.5 cm long, usually shorter than the internodes, hispid or glabrous, striate, the margins ciliate. Ligules 0.3- 0.6 mm long, membranous, the apex ciliate; collar pilose with whitish hairs. Blades linear-lanceolate, 3.5-12 cm long, 0.5-1.3 cm wide, ascendent, hispid or glabrous, long-pilose toward the base on the adaxial surface, cordate to subcordate, acute, the margins scaberulous and ciliate, the midnerve conspicuous. /nflorescences exserted, a terminal, lax and diffuse panicle 10-15 cm long, 3-7 cm wide; main axis glabrous, flexuous, smooth, the pulvini short-pilose or glabrous, first-order branch- es alternate, divergent from the main axis and flexuous, pedicels smooth and glabrous, flexuous. Spikelets paired or solitary, ellipsoid, 1.7-2.5 mm .7 mm wide, glabrous, with a short, incon- spicuous internode between the lower and upper long, glume. Lower glume ovate-acuminate, 0.9-1.3 mm long, 42 or less the length of the spikelet, glabrous, not embracing the upper glume, 3-nerved, the nerves either anastomosing or not toward the apex. Upper glume usually not covering the apex of the upper anthecium, 9-nerved, the nerves anasto- mosing or not toward the apex. Lower lemma glumiform, 9-nerved. Lower palea linear-lanceo- late, small, hyaline, glabrous; lower flower absent. Upper anthecium ellipsoid, 1.6-1.9 mm long, 0.6 mm wide, with simple papillae, apiculate; stigmas whitish. Caryopsis ellipsoid, brownish, hilum punc- tiform, embryo Уз or less the length of the cary- opsis. Distribution and ecology: occurring in Vene- zuela, in the states of Bolivar and Amazonas, Guy- ana, and northern Brazil, from 900 to 2,000 m. Frequent in savannas or mountain slopes, in sandy soils in Venezuela; in Brazil it grows in sandy soils of campos rupestres and restingas. Additional specimens examined. BRAZIL. BAHIA: El Salvador, coastal dunes 2 km N of town of Itapua, near sea level, Plowman & Almeida 10045 (CEPEC, MO, US); Estrada de Bom (US); Lencois, gu BR 242, Leng at km 225, Mori & Boom 14257 amie MO); camino de Marat a крш. 15k 1 2469* (МО, ВВ, е en da Rod Canavieiras, 32 E. e Canavieiras, Belém 1 PEC, IAN, NY, SI, en US); rodovia Brasilia- Fortaleza, BR-242, 8 km da estrada para Lengois, campo rupestre, — et al. 6530 (CEN, K, MO, SP). RIO DE JANEIRO: estinga de Jacarepaguá, do lado NW da Pedra de Itaüna, Seen 5349 (RB, SI); Guanabara, estrada BR-6, Hoehne 599] (SI, SP); Rio de Janeiro, próxi Bandeirantes, Hoehne 5789 (SI, SP). RoRAIMA: Mt. Ro raima, Philipps Camp, 5,200-6,000 ft., Tate 281 (NY, US). Guyana. Pakaraima Mountains, Mt. Aymatoi, m" et al. 5769 (K, MO, NY, US). VENEZUELA. AMAZONA Cerro de la a Rio Yatua, along Canón Grande E of Cumbre Camp, 1,100 m, Maguire et al. 42193 (NY, P, US, VEN). BOLIV AR: la Gran Sabana, km 167, S of El Dorado along highway to Santa Elena, 24 km S of La 152 Annals of the Missouri Botanical Garden simia Davidse 4762 (MO, SI, VEN); la Gran Sabana, 2 km al sur de La Ciudadela, 5?47'N, 61?25'W, 1,300 m, um et al. 4408 (MO, SI, VEN); a 3 km al N de Kamoiran, Gran Sabana, 5?35'N, 61?20'W, 1,100 m, Zuloaga et al. 4470* (MO, SI, Ma U A сро пеаг t 5 foot of the peak of Uaipán, Koyama & Agostini 7 f 7391( (VEN); MU Gran Sabana, inmediaciones del Monumento al Soldado Pione- ro, 38.5 km al S del sitio “Piedra de la Virgen," Huber et al. 12908 (MYF, SI). This species is related to P. pycnoclados and P. sciurotoides, but has spikelets intermediate in size between both species. It also differs from P. sciurotoides in that the latter species has cordate and amplexicaulous leaves, the main inflorescence axis hirsute, and usually pilose spikelets. Panicum pycnoclados differs from P. aequivaginatum by its cordate and amplexicaulous, asymmetrical leaf blades, smaller inflorescences, 3-6(-10) cm long, spikelets 2.2-3(-3.3) mm long with a conspicuous internode between the lower and upper glume; low- er glume 12-% the length of the spikelet. Panicum thinophilum only represents a depau- 2 perate variant o aequivaginatum with pro- fusely branching culms and axillary panicles. 5. Panicum assurgens Renvoize, Kew Bull. 37: 5. 1982. TYPE: Brazil. Bahia: on road to Abaira, ca. 8 km to N of the town of Rio de Contas, 13933'5, 41947'М, 1,000 m, 18 Jan. 1972, Harley et al. 15229 (holotype, CE- PEC; isotypes, K, MO, NY, US 2955120). Figures 24A, 25. Probably perennial. Culms leaning on adjacent vegetation, to 2 m long, internodes 4-10 cm long, cylindric, glabrous; nodes glabrous, compressed. Sheaths 3-7 cm long, shorter than the internodes, striate, sparsely hispid, with one margin ciliate, the other glabrous. Ligules membranous, the mem- branous portion 0.2 mm long, short-pilose at the apex with hairs nearly 0.1 mm long; collar shortly pilose. Blades linear-lanceolate, 8-12 cm long, 0.7-1 ст wide, flat, sparsely hispid, attenuate at base and with the apex acute, the margins finely scabrous. Inflorescences terminal, 12-15 order branches whorled toward the base of the long-exserted, lax, diffuse, cm long, 7-10 cm wide, first- inflorescence, otherwise subopposite or alternate, axis of the branches flexuous, glabrous, eglandular, pulvini glabrous, pedicels glabrous. Spikelets ellip- i ‚ 1.2 sparsely pilose between the nerves, greenish, with- soid, 3.3 mm long mm wide, glabrous or out a conspicuous internode between the lower and upper glume, upper glume and lower lemma sub- 2 "7 equal. Lower glume ovate-lanceolate, 2.1-2.7 mm long, 34-7, the length of the spikelet, not embracing the upper glume at its base, glabrous, 3-nerved, the nerves anastomosing toward the apex. Upper glume 2.7 mm long, as long as the lower lemma or a little shorter and not covering the apex of the upper anthecium, 9-nerved. 9-nerved, 2.7-2.9 mm long. Lower palea lanceo- late, 2 mm long, 0.3 mm wide, hyaline, sparsely Lower lemm a pilose; lower flower absent. Upper anthecium el- lipsoid, 2.7 mm long, | mm wide, papillose, upper lemma slightly apiculate, the upper margins in- durate, sparsely pilose at the apex. Caryopsis el- lipsoid, 2.1 mm long, 0.9 mm wide; hilum punc- tiform, embryo У, or less the length of the caryopsis. Distribution and ecology: known only from the type collection, where it grows at edges of forests in the state of Bahia, Brazil, 1,000 m at approximately 6. Panicum cabrerae Zuloaga & Morrone, Ann. Missouri Bot. Gard. 78: 156. 1991. TYPE: Brazil. Bahia: Mun. Rio de Contas, Pico das Almas, a 18 km ao SNW de Rio de Contas, 13°33'5, 41%57'W, 1,600-1,850 m, 22 July 1979, Mori, King, dos Santos & Hage 12475 (holotype, CEPEC; isotypes, MO, US). Figure 26 50 cm tall, vegetation, sparingly branching, internodes cylin- Perennial. Culms ca. leaning on dric, hollow, sparsely pilose with whitish hairs, nodes many, densely pilose with long whitish hairs. Sheaths striate, 1.5-2 cm long, longer than the internodes, long-hispid, the margins pilose with whitish hairs, more so toward the upper portion; auricles small, pilose; collar pilose. Ligules ciliate-membranous 4-6 cm long, 2 mm wide, attenuate at base, acuminate ca. 0.2 mm long. Blades linear-lanceolate, at apex, the margins involute, densely pilose and scabrous, with long, whitish hairs on both surfaces, deciduous at maturity. /nflorescence terminal, a lax, diffuse panicle 4.5 6 cm wide, 6 cm long, 3.5 peduncles hispid, 5-7 cm long; main axis cylin- dric, densely hispid on the lower portion, otherwise glabrous, the nodes distant; first-order branches alternate, divergent and reflexed, hispid basally, glabrous distally, pulvini pilose to glabrous; pedicels claviform, smooth, glabrous, 2-9 mm long. Spike- -2.3 mm long, 0.8-0.9 mm wide, plano-convex, hirsute, with whitish hairs on the glumes and lower lemma. Lower glume 1.1-1.3 mm long, М or more the length of the spikelet, 3-nerved, acute, a small stipe present lets solitary, ellipsoid, 2 between the lower and upper glume. Upper glume 9-nerved, nerves anastomosing toward the apex. Volume 80, Number 1 1993 Zuloaga et a Revision of Panicum subg. Dichanthelium sect. Dichanthelium Lower lemma 9-nerved. Lower palea lanceolate, 1.5-1.8 mm long, 0.3 mm wide, glabrous, trun- cate, hyaline; lower flower absent. Upper anthe- cium ellipsoid, 1.9 mm long, 0.7 mm wide, papil- lose, with simple papillae all over its surface, glabrous or with a few microhairs toward the apiculate apex; lodicules truncate, ca. 0.2 mm long, anthers 3, 1.2 mm long; rachilla prolonged or not beyond the upper anthecium as a short mucro. Caryopsis ob- ovoid, 1.1 mm long; hilum oblong, embryo less than У, the length of the caryopsis. Distribution and ecology: growing in campos rupestres, in the state of Bahia, Brazil. Additional specimens examined. BRAZIL. BAHIA: Mun. Rio de Contas, Pico das Almas, vertente leste, subida do pico do campo N d iroz, 13%32'S, 41%58'W, 1,600 m, Harley. Giulietti, Stannard & Hind 26320 (K). Panicum cabrerae shares with P. aciculare a similar vegetative pattern, with linear, pilose blades, small panicles and spikelets of more or less the same size. Panicum cabrerae differs from P. acicu- lare in that the latter species has obovoid spikelets with the upper glume and lower lemma 7-nerved, also, P. found from the southern United States to Colombia and Venezuela. and the lower palea absent; aciculare is 7. Panicum caparaoense Zuloaga & Morrone, da nov. TYPE: Brazil. Espírito Santo: Mun. Muniz Freire, Rodovia BR-262, 1,000 m, 21 July 1982, Hatschbach & Guimaraes 15170 (holotype, K; isotype, MBM not seen). Figure 36 Ligula membranacea 0.8 mm longa; lamina lanceolata 14-21 cm longa, 1 cm lata, гие paniculne laxae, diffusae, 23 cm longae; spiculae modi ae, 5.9- 6.8 mm longae, glabrae: gluma иы 25 spiculae lon- gitudine, acuminata, 5-nervia; gluma superior 13-15 ner- via; lemma inferius 13-15-nervium, palea inferior lan- ceolata, 5.5 mm longa, 1.2 mm lata, margine longe pilosa; anthoecium longe ellipsoideum, 5.2 mm longum, 1.6 mm latum, papillis simplicibus praeditum, necnon lemmate marginibus superioribus cristato ac piloso Plants perennial?, the base not seen; internodes dark, pressed, glabrous. Sheaths striate, one margin cil- hollow, cylindric, glabrous, nodes com- iate, otherwise glabrous. Ligules membranous at the base and laciniate at the apex, 0.8 mm long, brownish, surmounted by long whitish hairs on the adaxial surface of the blade; collar densely and shortly pilose. Blades lanceolate, 14-21 cm long, 1 ст wide, flat, attenuate at the base and apex, the adaxial surface with scattered, appressed hairs, the abaxial surface glabrous, commonly purplish, the margins ciliate. Inflorescence terminal, ex- serted, peduncle terete, ca. 22 cm long, glabrous; panicles lax, diffuse, 23 cm long, ca. 20 cm wide; main axis terete, glabrous, the pulvini brownish, glabrous; first-order branches alternate, divergent, the axis of the branches and pedicels smooth, gla- brous. Spikelets solitary, narrowly ellipsoid, 5.9— .8 mm long, glabrous. Lower glume 4.1 mm long, ¥, the length of the spikelet, acuminate, 5-nerved, separated from the upper glume by a conspicuous internode. Upper glume scaberulous toward the apex, 13-15-nerved, the nerves anastomosing to- ward the apex. Lower lemma scaberulous toward the apex, 13-15-nerved, the nerves anastomosing toward the apex. Lower palea lanceolate, 5.5 mm long, 1.2 mm wide, hyaline, the margins long-pilose with whitish hairs; lower flower male, anthers ca. 3 mm long. Upper anthecium narrowly ellipsoid, 5.2 mm long, 1.6 mm wide, smooth, shiny, in- durate, with simple papillae regularly distributed over the lemma and palea, lemma 9-nerved, crest- ed and pilose on the upper margins, greenish at the apex, palea pilose on the upper margins; lod- icules 2, conduplicate, 0.4 mm long; stamens 3, the anthers 3 mm long; styles free, stigma plumose; rachilla prolonged beyond the upper anthecium as a short mucro. Caryopsis unknown. Distribution and ecology: known from the state of Espirito Santo in Brazil, where it grows in edge of forests. Paratype. BRAZIL. ESPIRITO SANTO: Castelo, Braco do Sul, «ӨР 19172 (US). Panicum caparaoense is related to P. itatiaiae and P. davidsei. 4.6-5 mm long, the lower palea glabrous or sparse- Panicum itatiaiae has spikelets ly pilose, and the leaf blades subcordate. Panicum davidsei has spikelets 3-3.3 mm, lower flower absent, and upper anthecium glabrous or shortly pilose. Тће specific epithet makes reference to the type locality of this new species, the Serra do Caparaó in Espirito Santo, Brazil. 8. Panicum congestum Renvoize, Kew Bull. 37: 329. 1984. TYPE: Brazil. Bahia: 22 km NW of Lagoinha, which is 5.5 km SW of Delfino, on side road to Minas do Mimoso, m, 6 1974, Harley, Renvoize, Erskine, Brighton & Pinheiro 16869 (ho- lotype, CEPEC 10082; isotypes, K, MO, NY, RB, US 2955113). Figure 25. Caespitose perennial. Culms erect, freely branching at the upper nodes, 30-50 cm tall, rigid: 154 Annals of the Missouri Botanical Garden internodes elongated toward the base of the plant, terete, hollow, densely hirsute, nodes brownish, pilose. Sheaths persistent or deciduous, striate, densely hirsute, the margins membranous. Ligules membranous-ciliate, ca. | mm long, with whitish hairs; collar pilose. Blades linear-lanceolate, 1.5- 5 cm long, 0.2-0.4 cm wide, congested on the culms, the margins involute, narrowed at base and acuminate at the apex, densely hirsute on both surfaces with rigid whitish hairs. /nflorescences with base included within the upper leaves, with a peduncle ca. 0.5 cm long; panicles 2 cm long or smaller, contracted, with only 2-3(-9) spikelets per panicle; main axis hirsute, flexuous, first-order branches short, the axis of the branches and ped- icels densely hirsute. Spikelets solitary, ellipsoid, 2.8-3.2 mm long, | mm wide, hirsute, with short- papillose hairs, upper glume and lower lemma sub- equal or the upper glume a little shorter. Lower glume 2.2-2.6 mm long, ?4 or more the length of the spikelet, acute, 5-7-nerved, the nerves anas- tomosing toward the apex, with a conspicuous in- ternode between the lower and upper glume. Upper glume 2.6 mm long, 9-nerved. Lower lemma 9-nerved, glumiform. Lower palea lanceolate, 1.7 mm long, 0.4 mm wide, hyaline, the margins cil- iolate; lower flower absent. Upper anthecium el- lipsoid, 2.4 mm long, 0.8 mm wide, papillose, with simple papillae regularly distributed over the lem- ma and palea, apiculate and pilose at the tip of the lemma, lemma 5-nerved; internodes conspicuous between the upper glume and the upper anthecium. Caryopsis ellipsoid, brownish, hilum oblong, em- bryo Уз the length of the caryopsis. Distribution and ecology: known only from campos rupestres in the state of Bahia, Brazil, where it is found on sandstone rocks at approxi- mately 1,000 m. This species is distinguished by its densely hir- sute and profusely branching culms, hirsute sheaths and blades, and few-flowered inflorescences that are short-exserted with the base included in the upper leaves. 9. Panicum cucaense Zuloaga & Morrone, Ann. Missouri Bot. Gard. 78: 158. 1991. TYPE: Brazil. Rio de Janeiro: Mun. Petrópolis, Morro do Cuca, entre Vale dos Videiras e Araras, campo de altitude, 1,600 m, 27 Jan. 1983, Martinelli & Simonis 9011 (holotype, RB; isotypes, MO, SI). Figure 27. Caespitose, shortly rhizomatous perennial. Culms 20-30 cm ascending and rooting at the lower nodes, freely tall, many-noded, erect to geniculate- branching; internodes 1—4 cm long, cylindric, stri- ate, glabrous; nodes compressed, glabrous, brown- ish. Sheaths striate, ca. 1.2 cm long, longer than the internodes, strongly distichous, long-pilose on the distal portion, otherwise glabrous, one margin pilose with long, whitish hairs, the other one mem- branous. Ligule ciliate-membranous, the membra- nous portion ca. 0.1 mm long, the ciliate portion 0.4 mm long. Blades linear, 1.5-5.5 cm long, 0.1-0.2 cm wide, folded, attenuate toward the apex, pilose on the adaxial surface toward the lig- ule, otherwise glabrous, the borders scabrous. Pri- mary inflorescences lax, diffuse, few-flowered, 2- 3.5 em long, 1.5-3.5 cm wide; peduncles to 6 cm long, glabrous; main axis striate, flexuous, gla- brous, somewhat glandular, the branches alternate, divergent and widely spaced, glabrous, the axils of the branches pilose; pedicels glabrous, 1-6 mm long. Secondary inflorescences similar to the ter- minal one. Spikelets ellipsoid, 2.2-2.7 mm long, 0.8-0.9 mm wide, gaping at maturity, glabrous. Lower glume lanceolate, 1.5-1.9 mm long, 23-3 the length of the spikelet, acuminate, 1(-3)-nerved, a small stipe present between the lower and upper glume. Upper glume and lower lemma subequal, 7 —9-nerved, acute. Lower palea lanceolate, 1.7— 1.9 mm long, 0.4–0.7 mm wide, hyaline, glabrous; lower flower absent. Upper anthecium narrowly ellipsoid, 2.1-2.3 mm long, 0.7-0.8 mm wide, papillose, with simple papillae evenly distributed over the lemma and palea, apiculate, the apex scabrous, otherwise glabrous; lodicules ca. 0.2 mm long; stamens 3, anthers 1.2 mm long. Caryopsis ellipsoid, 1.4 mm long, 0.7 mm wide, brownish; hilum aid embryo less than half the length of the caryopsi Distribution and ecology: inhabiting open and dry habitats, in rocky, granitic soils of high-altitude campos of mountains in eastern Brazil, at 2,000- 2,500 m. Additional specimens examined. BRAZIL. RIO DE JANEIRO: wp Serra dos Orgaos, Pedra do Sino, 2,100-2,170 m, Feb. 1953, J. Vidal 11-6467, 11-6485 (R, SD; Ms ое ке das Videiras, Morro do Cuca, 1,800 m, 17 Apr. 1976, Martinelli 814 (RB); without locality, Glaziou BN ). Affinities of P. adenorhachis are discussed in Zuloaga & Morrone (1991). 10. Panicum cumbucana Renvoize, Kew Bull. 37: 332. 1982. TYPE: Brazil. Bahia: by Rio Cumbuca, 3 km S of Мисиге, near site of small dam on road to Cascavel, 850 m, 4 Feb. 1974, Harley, Renvoize, Erskine, Brighton Volume 80, Number 1 1993 Zuloaga et al. Revision of Panicum subg. Dichanthelium sect. Dichanthelium & Pinheiro 15930 (holotype, CEPEC 10042; isotypes, K, MO, NY, US 2955114). Figures 24C, 27 Short-rhizomatous perennial. Culms long de- cumbent, then erect, to 35 cm tall, freely branch- ing at the upper nodes, internodes 1-2.8 cm long, hispid, hollow, nodes hispid. Sheaths 1-2 cm long, usually shorter than the internodes, hispid or gla- brous, one margin long-ciliate, the other short- pilose. Ligules 0.4-0.6 mm long, shortly mem- branous at the base, ciliate at the apex, arcuate; collar shortly pilose. B/ades lanceolate, 1-3 cm long, 0.3-0.6 cm wide, rigid, ascending and slightly divergent from the culms, cordate and amplexi- caulous, acute, glabrous or shortly and densely hispid on both surfaces, with or without long hairs toward the base on the adaxial surface, the margins scaberulous, long-ciliate toward the base of the blades, the midnerve inconspicuous. /nflorescences exserted or with the base included within the upper leaves, panicles lax, few-flowered, 1-2.5(-3) cm long, 1-2 cm wide; main axis hispid to sparsely pilose, first- and second-order branches shortly pi- lose, divergent, eglandular, the pulvini pilose; ped- icels sparsely pilose. Spikelets solitary or paired, ellipsoid, 1.9-2.4 mm long, 0.6-0.8 mm wide, shortly hispid or glabrous, upper glume and lower . Lower glume ovate, l- .9 mm wide, У; the length of the шы, pilose or glabrous, not embracing the upper glume at its base, (1-)3-nerved, with a conspicuous in- ternode ca. 0.3 mm long between the lower and upper glume. Upper glume 1.5-1.9 mm long, pilose, 7–9-пегуед, usually not covering the apex of the upper anthecium. Lower lemma 1.5-1.9 mm long, glumiform. Lower palea linear-lanceo- late, 1.2-1.5 mm long, 0.2-0.3 mm wide, hyaline, glabrous; lower flower absent. Upper anthecium ellipsoid, 1.5-1.8 mm long, 0.5- mm wide, indurate, papillose, and pilose; 5-пегуед, palea 2-nerved, anthers 0.9 mm long, purplish. Caryopsis ellipsoid, 1.2-1.5 mm long, 0.2 m wide; hilum punctiform, embryo less than У the length of the caryopsis. Distribution and ecology: found in campos ru- pestres in the state of Bahia, Brazil, in sandy, rocky soils from 900 to 1,400 m Additional specimens examined. BRAZIL. BAHIA: 10 km NW of Mucugé, Mori et al. 12698 (CEPEC, MO, SI, US); Serra do Sincorá, 2-3 km approximately SW of Mucugé on the road to Cascavel, 950 m, Harley et al. 18829 (CEPEC, MO, NY, US); Mucuge, Serra do Sincorá, 7 km N of Mucugé on road to Andarai, growing in small patches under Velloziaceae and other plants, Calderón et al. 2425* (SI, US); Mucugé, Serra do Sincorá, 26 km S of Andarai, 1,000 m, on rocks, between Ре Баја hills of marble rock on N side of Jacobina, 600- m, Webster et al. 25725 (МО); estrada entre An- darai-Mucugé, Noblick & Pinto 2870 (К). Panicum cumbucana is closely allied to, and perhaps conspecific with P. stipiflorum, which only differs by having leaf blades 2-4.5 cm long, usually not overlapping, and inflorescences 3-4.5 cm long. e culms and leaf blades are usually purplish in P. cumbucana, and the blades are, in many cases, overlapping. Anatomically, the abaxial epidermis of P. cum- bucana closely resembles that of P. hebotes, P. peristypum, and P. sciurotoides due to the nu- merous, needlelike macrohairs without cushion bas- es and because of the short, angular, intercostal cells, which resemble bulliform cells in surface view. However, in transverse section this resemblance is not so evident because P. cumbucana does not have well-developed adaxial ribs and furrows and lacks a keel (Fig. 6A, B) 11. Panicum davidsei Zuloaga & Morrone, Ann. Missouri Bot. Gard. 78: 158. 1991 TYPE: Venezuela. Bolivar: Cabanayén, La Gran Sabana, wet inundated savanna, 1,300, 3 Dec. 1973, Davidse, Ramia & Montes 4796 (ho- lotype, MO). Figures 7, 27. Perennial. Culms decumbent, rooting and branching at the lower nodes, then becoming erect, leaning or not on vegetation, to 2.2 m long, in- ternodes 8-22 cm long, cylindric, striate, Mena. compressed, nodes hirsute. Sheaths 5-8 cm long, sparsely hirsute with long-tuberculate hairs, one of the margins long-ciliate, the other one membra- nous; collar long-pilose with whitish hairs. Ligules membranous-ciliate, ca. 0.2 mm long. Blades nar- rowly lanceolate, 9-13 cm long, 1-1.3 cm wide, flat, subcordate, the apex acuminate, the adaxial surfaces long-pilose toward the base, otherwise gla- brous, the abaxial surfaces sparsely pilose, the mar- gins scabrous and ciliate toward the base, the mid- nerve conspicuous. /nflorescence terminal, long-exserted, a lax, diffuse panicle 12-20 cm long and 7-12 cm wide, peduncle 10-40 cm long; main axis cylindric, striate, pilose near the branches, eglandular, otherwise glabrous, lower branches 156 Annals of the Missouri Botanical Garden whorled, the upper ones subopposite to alternate the axis of the branches flexuous, glabrous, eglan- dular, the pulvini pilose; pedicels glabrous, flexu- ous. Spikelets ellipsoid, 3-3.3 mm long, 1 mm wide, scaberulous, attenuate toward the base and with a small stipe between the lower and upper glume, upper glume and lower lemma subequal with manifest nerves. Lower glume ovate-acumi- nate, 1.5-1.8 the spikelet, 3-nerved. Upper glume 2.4-2.7 mm long, not covering the apex of the upper anthecium 11-14-nerved, the nerves anastomosing toward the apex. Lower lemma ca. 2.7 mm long, 10-12- nerved, the nerves anastomosing toward the apex. Lower palea lanceolate, 2.1-2.4 mm long, 0.3 mm wide, glabrous, hyaline; lower flower absent. Upper anthecium ellipsoid, 2.4-2.7 mm long, 1 mm wide, papillose, apiculate, the lemma with a green, scabrous crest on the apex and with the upper margins membranous, prolonged toward the apex as small wings; rachilla prolonged beyond the upper anthecium as a small mucro; e ca. 0.4 mm long; stamens 3, anthers ca. 1.2 mm long. Caryopsis ellipsoid, 1.8 mm long, 0.9 mm wide, hilum punctiform, embryo Уз the length of the caryopsis. Distribution and ecology: found in Bolivar Ven- ezuela, and Roraima, Brazil, on forest edges (where it leans on the vegetation), or in open places in 1,200-1,300 m sandy soils, BRAZIL. RORAIMA: 1927 (RB 110786). Additional ПАР ехатипеа. Aldeia do Tuchar a, Rondon s. s.n., July 1,300 m, Zuloaga et al. 4406* (MO. SI, VEN); a 1 km al S del Puente Sakaika, Gran Sabana, 1,200 m, Zuloaga et al. 4427* (MO, SI, VEN). Affinities of P. adenorhachis are discussed in Zuloaga & Morrone (1991). The large number of nerves in the upper glume and lower lemma of this new species suggests a relationship to P. itatiaiae and P. caparaoense, species from eastern Brazil with spikelets 4.6-6.8 mm long. Panicum davidsei has a thick leaf blade without a midrib, well-developed adaxial ribs and furrows, and conspicuous, long bundle sheath extensions (due to the thickness of the lamina) (Fig. 7A-C). 12. Panicum dichotomum L., Sp. Pl: 58. 753. Dichanthelium dichotomum (L.) Gould, Brittonia 26: 59. 1974. TYPE: United States. "Habitat in Virginia," Clayton 456 (lectotype, BM not seen; photo and fragment, US 2808912). Virginia: mm long, nearly 2 the length of Perennial. Culms decumbent, 10-25(-80) ст long, fasciculate and densely branching at the mid- dle and upper nodes, internodes glabrous, cylindric, hollow, nodes glabrous or short-pilose. Sheaths gla- brous, glandular or eglandular, the margins mem- branous, one margin long-ciliate. Ligules mem- branous-ciliate, 0.1 —0.6 mm long; collar glabrous. Blades linear-lanceolate, 2-7 cm long, 0.2-0.5( 1.2) ст wide, flat or slightly involute, glabrous to sparsely pilose, narrowed at the base, attenuate at the apex, the margins scaberulous, cartilaginous or not, usually ciliate toward the base, otherwise gla- brous. /nflorescences terminal, short- to long-ex- serted, peduncle glabrous, 3-16 cm long; panicles 1-5 cm long, 1-3 ст wide, lax, few-flowered; main axis scabrous or glabrous, glandular or eglandular, the pulvini glabrous, first-order branches alternate, divergent, glandular or eglandular; pedicels flex- uous, 2-9 mm long, scabrous or glabrous. Axillary panicles similar to the awe one. Spikelets sol- itary, obovoid or ellipsoid, 1.4-2.3 mm long, 0.6- 1 mm wide, glabrous or сре hispid, greenish or tinged with purple, the upper glume and lower lemma subequal. Lower glume ovate, 0.3-0.7 mm long, М to УЗ the length of the spikelet, obtuse to truncate, glabrous, nerveless or |-пегуед. Upper glume 1.2-1.8 mm long, not covering the apex of the upper anthecium, 7-nerved, separated from the lower lemma by a short internode. Lower lem- ma 1.2-1.7 mm long, glumiform, 7-nerved. Lower palea lanceolate, small, 0.8-1 mm long, 0.3 mm wide, glabrous, hyaline; lower flower absent. Upper ].1- wide, pale, indurate, acute, papillose, scaberulous toward the apex. Caryopsis obovoid, dark, 1.2 mm long, 0.8 mm wide; hilum punctiform, embryo less than 14 the length of the caryopsis. anthecium obovoid, .7 mm long, 0.7-1 mm KEY TO THE VARIETIES l. 8. ун cartilaginous Le spikelet 1.8- 12 2.3 mm long А dichotomum 1. Blades with cartlaginous margins; I 1.4- 2b 1.6 mm long var. tenue 12a. Panicum dichotomum var. dichoto- mum. Figure 27. Panicum nitidum Lam., Tabl. Encycl. 1: 172. 1791. Panicum Wi M Ar L. var. а TEN ) Alph. Wood, Cl ook Bot ed. 3: TYP United S “E Ca tolin мө er s.n. yos. P-LAM; fragment n photo US 80880). Panicum nodiflorum ah Encycl. 4: 744. 1798. Pan- icum dichotomum var. тойот (Lam.) Gri- seb., Cat. Pl. Cuba a 1866. TYPE: United States. South Carolina: without loc ality, Fraser s.n. (holo- type, P-LAM; fragment and photo, US 2808963). Volume 80, Number 1 1993 Zuloaga et al. 157 Revision of Panicum subg. Dichanthelium sect. Dichanthelium Panicum barbulatum n Fl. E ж 803. otomum L. barbulatum vonage Alph. Wood. Class-Book Bor., ed. : 6l. Panicum pubescens Lam. var. Variarum (Michaux) көш, Cat. Pl. New Jersey 280. ] m Lam. var. белпа Ману . 1897. TYPE: x s.n. (lectotype, P-MICH; fragment, US). Punta um maculatum Ashe, J. Elisha Mitchell Sci. Soc. 15: 44. 1898, non Aublet 1775. um dread nense Ashe, J. Elisha Mitchell Sci. 16: 900. Panicum dichotomum L. var. pir (Ashe) Lelong, Brittonia 36: 266. 1984. TYPE: Unit- ed States. North Carolina: Wake Co., Raleigh, May 1895, Ashe s.n. (lectotype, US). Panicum ИЛЫ Ashe, J. Elisha Mitchell Sci. Soc. 1898. Pani tium Којо! mum L. var. roano- dt eA Lelong, Brittonia 36: 265. 1984. TYPE: United States. North Carolina: Dare Co. ; Roanoke Island, June 1898, 4 Panicum multirameum Scribner, U.S.D.A. D Circ. i 2. 1900. TYPE: Mexico. Veracruz: near , 1889, Pringle 7882 (lectotype, US 743295; ite MO). Panicum caerulescens Haske ex A. Hitche., Contr. U.S. erb. 12: 219. 1909. TYPE: United States. Florida: ‘Dade ‚ Miami, 3 Apr. 1906, Hitchcock 706 (holotype, US; isotype, NY) Distribution and ecology: common in North America, from Canada to Mexico, in moist ground, open swampy wood and wet meadows. In South America it is known from the Cerro Santa Ana, Falcón, Venezuela. Selected specimens cited. CUBA. HABANA: Batabano, León 6152 (US). ISLA DE LA JUVENTUD: along road to Santa Isabel near SE base of Cerro Daguilla, sabanas, Killip 44853 P ORIENTE: Sierra de Nipe, Loma Men- sura, León 19828 (US). SANTA CLARA: Sabana de San Marcos, León 9203 (US). DoMiNICAN REPUBLIC. AZUA: Sierra de Ocoa, San José de Ocoa, 1,550 m, Ekman 11942 (US). LA VEGA: vicinity uf терасна, 500- 1,200 m, Allard 14509a (US). MONTE CRISTI: oup Central, on the ridge between Rio Cenobi and Rio de la Cidra, 00 m, Ekman 12697 (US). GUATEMALA. ALTA VERAPAZ: њи swamp just E of Tactic, 1,300 m, Сеча 43942 (Е, US). 7АСАРА: between Loma El Picacho 2 ap de Monos, 2,000-2,600 m, Steyermark 42 US). Jamaica. Bull Head Mountain, Hitchcock 0532 Miguel ce _ 500 ft., 4690 (US). УЕХЕЛИЕ1. A. FALCON: Рага npe Santa Ana, arriba de Santa Ana, 600 m, Wingfield 7124 (VEN); Cerro Santa Ana, arriba de Santa Maria, 1,200 m, Wing- field 6844 (MO). 12b. Panicum dichotomum var. tenue (Muhlenb.) Zuloaga & Morrone, comb. nov. P. tenue Muhlenb., Descr. Gram. 118. 1817. Dichanthelium dichotomum (L.) Gould var. tenue (Muhlenb.) Gould & C.A. Clark, Ann. Missouri Bot. Gard. 65: 1119. 1978. TYPE: Muhlenberg Herb. 192 (lectotype PH-M not seen; fragment and photo of lectotype, US not seen). дш е кран Trin., Gram. Panic. 242. 1826. um acuminatum Sw. var. unciphyllum (Trin.) Lelon Be Brittonia 36: 269. 1984. Dichanthelium ensifolium (Elliott) Gould var. n nt Hansen & Wunderlin, Ann. Missouri Bot. Gar 1647. 1988. Dichanthelium d AU (L.) Gould var. unciphyllum (Trin.) Davidse, Novon 2: 1992. TYPE: North America: without locality, Trut: tinick s.n. (lectotype, LE not seen сали о marginatum Nash, Bull. Torrey Bot. Club 1897. TYPE: United States. Florida: Lake үө June 1894, Nash 925 (holotype, US 208344). Distribution and ecology: eastern United States, Texas, Mexico, Belize, and the moist white-sand savannas, moist sandy woods, and ‘est Indies, in marshy areas. Selected specimens cited. BELIZE. EL CAYO: Moun- ee Pine Ridge, 550-700 m, Davidse & Brant 33037 UBA. ISLA DE LA JUVENTUD: Howard Estate, along Rio Callejón, Killip 44805 (US). PINAR DEL RIO: Arroyo ntua, savannas, Ekman 10994 (US); S of Guane, ae de үң Julián, León & Roca 7009 (US). MEXICO. CHIAPAS: a Trinitaria, montane rainforest at Lagos de Colores, e de Montebello National Park, Breedlove : (MO). avidse 55038 13. Panicum divergens HBK, Nov. Gen. & Sp. 1: 102. 1816. TYPE: Ecuador. Pichincha: Sangolqui, Humboldt & Bonpland s.n. (ho- lotype, P; fragment, US 80645). Figures 24E, Е, 27 Panicum commutatum Schultes, Mantissa 2: 242. 1824. Dic nani ketam commutatum (Schultes) Gould, Brit- OI 26: 59. 1974. Panicum nervosum Muhlenb. . Sketch Bot. S. Carolina 1: 122. 1816, non Lam., 1797. iv um polyneuron Steudel, Syn. PI. Glumac. 1: 91. 4. TYPE: United States. **Car. et Geor.," Elliott v rb. s.n. Msg CHARL not en; photo and fragment, US 80874). Paican leiophyllum F exic. PL 2: 20. non Nees, 1829. TYPE: Mexico. Veracruz: * Cordobensi,”” Cir. 19: 2. 1900. План аи МЕ РА Рини Sida Gould, Brittonia 32: 357. 1980. TYPE: Mexico. Michoacán: Patzcuaro, 10 Nov. Pringle 5203 (holotype, US 743918). Panicum joorii Vasey, U.S.D.A. Div. Agrost. Bull. 8: 31. 89. Panicum commutatum Schultes var. joorii (Vasey) Fernald, jan 39: 388. 1937. TYPE: а States. Louisiana: near Бис Rouge, 1 Oct. 885, Joor 39 (holotype, "US 2808925). mE TAA Swallen, Contr. U.S. Natl. Herb. 29: 422. 1950. TYPE: Guatemala. Jalapa: collected in oak onde around the top of Cerro Alcoba, just 158 Annals of the Missouri Botanical Garden E of Jalapa, 1,300-1,700 m, 2 Dec. 1939, Stey- ermark 32513 (holotype, F; isotype, US 2236475). Panicum hintoni Swallen, Contr. U.S. Natl. Herb. 29: 950. TYPE: Mexico. México: collected at Be- jucos, Temascaltepec, 610 m, 8 Nov. 1932, Hinton 2527 (holotype, US 1867697; isotype, US 1865776 — Short-rhizomatous perennial without basal ro- settes of broader leaves, or rosettes occasionally present. Culms geniculate, decumbent and branch- ing at the lower nodes to erect, freely branching at the middle and upper nodes, (15-)40-120 cm tall, internodes hollow, terete, short-pilose or gla- brous, nodes dark, pilose. Sheaths striate, short papillose-pilose with caducous hairs or glabrous, one margin long-ciliate, the other glabrous. Ligules short-membranous at base and short- to long-ciliate at the apex, 0.2-0.4(-1) mm long; collar densely pilose. Blades lanceolate, 4—10(—16) cm long, 0.7 — 1.2(-1.7) cm wide, flat, clasping, asymmetric and cordate or subcordate at base, the apex attenuate, densely papillose-pilose with caducous hairs or gla- brous, densely pilose toward the base on the adaxial surface, the margins ciliate toward the base, oth- erwise scabrous. /nflorescences terminal, exserted, peduncle glabrous, 20-30 cm long; panicles lax, diffuse, (3-)10-22 cm long, 6-10 cm wide; main axis flexuous, greenish to purplish, glabrous or short-hispid, glandular or eglandular, the pulvini brownish or purplish, glabrous, first-order branches alternate or subopposite, divergent, the axis of the branches and pedicels smooth, flexuous, scaberu- lous, glandular or eglandular. Axillary panicles present or absent, similar to the terminal one. Spikelets solitary, narrowly ellipsoid to obovoid, (2.1-)2.3-3.2 mm long, 1.2-1.3 mm wide, plano- convex, greenish to purplish, shortly and sparsely hispid. Lower glume 0.5-1.5 mm long, М to less than / the length of the spikelet, ovate, 1-nerved, separated from the upper glume by a short inter- node. Upper glume covering the upper anthecium or not, acute, 7—9-nerved, the nerves anastomosing toward the apex. Lower lemma acute, 7—9-nerved, the nerves anastomosing toward the apex. Lower palea lanceolate, 1.3-2 mm long, 0.4 mm wide, hyaline, glabrous; lower flower absent. Upper an- thecium ellipsoid or obovoid, 2.3-3 mm long, 1- 1.1 mm wide, smooth, whitish, indurate, shiny, papillose, the apex of the lemma shortly apiculate and with unicellular macrohairs. Caryopsis ob- ovoid, 1.3 mm long, 1 mm wide; hilum punctiform, embryo less than % the length of the caryopsis. Distribution and ecology: eastern United States to Mexico and Guatemala and the West Indies (see discussion of the possible presence in Ecuador); it grows at the margins of pine or oak woods or in open field borders, to 2,700 m. Cub BA. aa Bari 1 (US). Selected specimens cited. d, HIM TENANGO: Chichavac, Skutch 524 (US). Hartı. Massif du Nord, Ekman 6189 (US). HONDURAS. OCOTEPEQUE: Aldea de Belén Gualcho y alrededores; 40 km al E de Nueva Ocotepeque, Nelson et al. 3800 (ОМАН). MEXICO. CHIA- PAS: Mun. Jitotol, near Colonia El Laurel, ca. 5 km N of Jitotol, M d et is 29630 (MO). JALISCO: Zapotlan, Hitchcoc r. Gr. Herb. 193 (MO, P). MICHOA- CAN: Duas. "Hitchcock s.n., Amer. Gr. Herb. 226 (F, MO, P). MORELOS: Huitzilac, Lyonnet 623 (MO). NAYARIT: 8 mi. W of Tepic, McVaugh 18881 (US). NUEVO LEON: 2 mi. 5 of Monterrey, Mueller 414 (US). VERACRUZ: Jalapa, Hitchcock s.n., Amer. Gr. Hb. 225 (MO The type locality of P. divergens is, according to Humboldt, Bonpland & Kunth, Sangolgui in Pichincha province, Ecuador. This is probably : abel error since Humboldt also made collections in Mexico, where this species is commonly found. Another possible explanation is that the species has, as in the case of P. umbonulatum, a disjunct distribution. 14. Panicum ensifolium Baldwin ex Elliott, Sketch Bot. S. Carolina 1: 126. 1816. Pan- icum nitidum Lam. var. ensifolium (Piian Vasey, U.S.D.A. Div. Agrost. Bull. 9. 1889. Dichanthelium ensifolium (Elliott) Gould, Brittonia 26: 59. 1974. Dichantheli- um dichotomum (L.) Gould var. ensifolium (Elliott) Gould & Clark, Ann. Missouri Bot. Gard. 65: 1119. 1978. TYPE: United States. Georgia: without locality, Baldwin s.n. (ho- lotype, CHARL not seen; fragment, US). Fig- ure 27. Panicum ЧЕ праве Trin., Gram. Panic. 242. 1826. TYPE: S Am. bor dun ttinick ex coll. En- үз н LE not seen). slini)," 8-30 cm tall, frequently unbranched above the base; internodes terete, greenish to purplish, pilose, nodes Densely tufted perennial. Culms erect, shortly pilose. Sheaths striate, greenish or purplish, glabrous, the margins membranous, ciliate or only pilose near the ligule. Ligules 0.4 mm long, mem- branous-ciliate, the membranous portion reduced. Blades linear, 2-6.5 cm long, 0.1-0.3 cm wide, rounded at base, flat to slightly involute, greenish or purplish, sparsely pilose on the adaxial surface or glabrous, the margins scabrous. /nflorescences terminal and axillary from the uppermost nodes, Volume 80, Number 1 1993 Zuloaga et al. 159 Revision of Panicum subg. Dichanthelium sect. Dichanthelium short-exserted, peduncles up to 5 cm long, cylin- dric, glabrous; panicles 1-4.5 cm long, main axis purplish, flexuous, glandular or eglandular, sca- brous or glabrous; pulvini glabrous, the branches alternate, divergent, axis of the branches and ped- icels flexuous, scabrous or glabrous, greenish or purplish, glandular or eglandular. Axillary panicles numerous, similar to the terminal one. Spikelets obovoid to ellipsoid, 1-1.4 mm long, О wide, greenish to purplish, shortly hispid or gla- mm long, Уз the length of the spikelet, truncate to acute, nerveless, hya- 6 mm rous. Lower glume line. Upper glume obtuse, shorter than the lower lemma and not covering the upper anthecium, 5- 7-nerved. Lower lemma as long as the spikelet, 5-пегуед. Lower palea lanceolate, 0.5-0.7 mm long, 0.3 mm wide, hyaline, glabrous; lower flower absent. Upper anthecium ellipsoid, 0.9-1.2 mm long, 0.4–0.6 mm wide, indurate, smooth and shin- ing, pale, papillose; lemma 5-nerved. Caryopsis obovoid, black, 0.6-0.8 mm long, 0.5 mm wide; hilum punctiform, embryo 4—% the length of the caryopsis. Distribution and ecology: United States, Belize, and Cuba, in pine white-sand savannas, 0-100 m Selected specimens cited. BELIZE. 3 пи. W of Boo town, O'Neill 8500 (MO, US); pine ridge N of aviation field, Bartlett 11236 (MO, US). CUBA. ISLA DE LA JUVENTUD: vicinity of Los Indios, Britton et al. 14218 (US) 14221 (MO, US). PINAR DEL RIO: km 13 on La Coloma road, León & Alain 19474 (US); Раши}: аї Rincón del Prado, Ekman 11047 (05); Laguna Santa Bárbara, Ekman 11465 (US) Panicum ensifolium has many fascicled culms branching from the base, spikelets 1-1.4 mm long, and a black caryopsis. 15. Panicum hebotes Trin., Mém. Acad. Imp. St.-Petersb., Ser. 6, Sci. Math 1: 301. 1834. Panicum Roni Trin. var. genuinum Doell l 7 т Шер. TYPE: Brazil. “У. sp. Bras” (holotype, LE; fragment, US 974693). Figures 8, 27, 33 Panicum mirandum Luces, J. Wash. Acad. Sci. 32: 163, fig. 8. 1942. TYPE: Venezuela. Miranda: Guinand Estate (Cárdenas), Siquire Valley, 500-1,000 m 19-24 Mar. 1913, Pittier 6483 (holotype, US 602176; isotype, NY). Panicum infuscum Swallen, Phytologia 14: 82. 1966. TYPE: Brazil. Rio de Janeiro: Tijuca, 700 m, 28 Apr. 1930, Chase 12145 (holotype, US 1448475). Panicum subtiliracemosum Renvoize, Kew Bull. 42: 922. 1 р, ое Renvoize, Hatschbach's Paraná Grasses: 39. 1988, error for P. subtilira- cemosum. TYPE: Brazil. Paraná: Cachoeira dos Tur- cos, Hatschbach 46020 (holotype, МВМ; isotype, K). Plants perennial. Culms decumbent, rooting and branching at the lower nodes, then becoming erect, -90 cm tall, leaning on vegetation, internodes 3.5-15 cm long, hispid, hollow, nodes compressed, pilose, brownish. Sheaths 2.5-7.5 cm long, shorter than the internodes, hispid, one margin ciliate, the other membranous. Ligules 0.3 mm long, mem- branous and laciniate at the apex, collar hispid. Blades lanceolate, 4.5-15 cm long, wide, flat, hispid or with the adaxial surface gla- brous, narrowed at the base, occasionally cordate, the apex attenuate, the margins scabrous. /n/flo- rescences terminal, usually long-exserted, peduncle 12-32 cm long, hirsute; panicles lax, diffuse, 7— 22 cm long, 4.5-11 cm wide; main axis flexuous, hispid, eglandular, first-order branches alternate to subopposite, second- and third-order branches gla- brous or sparsely pilose, eglandular; pedicels sca- brous, terete. Spikelets ellipsoid, 1.5-1.9 mm long, 0.6-0.9 mm wide, scaberulous toward the distal portion of the upper glume and lower lemma, oth- erwise puberulous or glabrous, greenish, the upper glume and lower lemma subequal with manifest nerves; internodes inconspicuous between the lower glume and upper anthecium. Lower glume 0.7- 1.5 mm long, ovate, acute to obtuse, 3- 7-nerved, glabrous, not embracing the upper glume, dimor- phic, some small, Y the length of the spikelet, 1— 3-nerved, other longer, ?4 the length of the spikelet, 7-nerved. Upper glume 1.4-1.7 mm long, 9-пегуед, not covering the apex of the upper an- thecium. Lower lemma glumiform, 9-nerved. Low- er palea lanceolate, 0.7-1.2 mm long, 0.3 mm wide, hyaline, glabrous; lower flower absent. Upper anthecium ellipsoid, 1.3-1.7 mm long, 0.7-0.8 mm wide, indurate, smooth, shiny, pale to brownish or black at maturity, puberulous at the apex, pa- pillose, shortly apiculate, lodicules 0.2 mm long, stamens 3, the anthers 0.7 mm long; stigma whit- ish. Caryopsis ellipsoid, 1 mm long, 0.6 mm wide, brownish, hilum punctiform, embryo less than Уз the length of the caryopsis. Distribution and ecology: forest edges in Ven- ezuela, Ecuador, Bolivia, and Brazil, from 500 to 1,800 m Additional specimens examined. BOLIVIA. LA PAZ: Yungas, Bang s.n., 1890 (C, K, MO, NY, US 823922, W). BRAZIL. CEARA: Baturité, Eugenio 285 (RB, US), 2649 (RB). EsPIRITO SANTO: Santa Teresa, Reserva Bio- lógica de Nova Lombardia, Picada da Cachoeira, Zuloaga et al. 2423, 2425 (RB, SI, US). MINAS GERAIS: Itacolumi, E of Ouro Preto, 1,300 m, Chase 9408 (MO, NY, US); 160 Annals of the Missouri Botanical Garden FIGURE 33. Panicum hebotes. — А. Habit, with panicle included. — B. Detail of ligule and lower portion of виа blade. С. Spikelet, lateral view. —D. Spikelet, lower glume view.—E. 5 Sid upper glume view. —F. Low lea. —G. Upper anthecium, lemma view. —H. Upper anthecium, palea ds w. ryopsis, embryo cd. Caryopsis, hilum view. (A, based on Brade 20119, B -J; based on Chase 9640.) Volume 80, Number 1 1993 Zuloaga et al. 161 Revision of Panicum subg. Dichanthelium sect. Dichanthelium Serra do Caparaó, Chase 9640 (F, MO, NY, US). PARANA: Marumbi, vedi a 3943 (K, PKDC). RIO DE JANEIRO: Tijuca, 1,000 m, Chase 12162, 12165 (US); Corcovado, Chase a 9745 (MO, NY, US); Floresta sob regime de Preservagào Permanente do IBDF, entre V Princesas y Rocio, Zuloaga et al. 2390* (MO, RB, SI, US); Parque Nac. Itatiaia, Picada Três Picos, Zuloaga et al. 2373* (RB, SI, US); S of Petrópolis, 530 m Davidse et al. 11411 (MO, 51); vicinity of Rio de Tanaro, Pico da Tijuca, Chase 8484 (MO, NY, 05); Vista Chinesa, Brade 20119 (US). RIO GRANDE DO SUL: without locality, Orth 1938 (US). SANTA CATARINA: Trés Barras, Garuva, eitz & Klein 5561 (US). são PAULO: 38 km SW of Jacupiranga along Hw to Curitiba, Davidse et al. y.1 10949 (MO, SI, SP); Cananeia Island, 1 km FLA 7 о Q, Q 2 alo SP, UB, 05); Parque do Estado, Clayton & Епеп 4190 (K, NY, SP); 19 km SW of Juquitiba, along Hwy. 11 to Curitiba, Davidse et al. 10933 (SP). COLOMBIA. NARISO: Mun. Ricaurte, vicinity of Ricaurte, along Rio Imbi, ca. 2-3 km above Ecopetrol Campamento Palmar, 3 km NW of Ricaurte, along trail to Ramos, 1,150 m, Croat 71502 DOR. TUNGURAHUA: valley of Pastaza River, cock 21862 (NY, US). VENEZUELA. ARAGUA Choroni, Parque Nacional Henry Due carretera de Maracay a Choroni, 1,560 m, Zuloag al. 4534* (MO, SI, VEN). BOLIVAR: E of Cerro Él Picacho, Las Nieves and Las Chicharras, 45 km N of Tumeremo, vicinity of Deborah, Altiplanicie de Nuria, 600-650 m Steyermark 89073 (NY, US, W). DISTRITO FEDERAL: El Junquito, en la sombra de la selva, 1,900 m, Shnee 484 (MY). sucRE: Distrito Nariño and Distrito Arismendi, Pe- ninsula de Paria, trail between crossing of Río Tacarigua to summit of slopes E of Cerro Humo, descending to Las Melenas, Steyermark et p. 121762 (NY, US). vaRACUY: Distrito Nirguá, 5 km Nirguá by road, 10?12'N, 68*34'W, 1,200 m, Paridis et al. 20904 (MO, VEN). The type material of P. hebotes and P. miran- dum, together with the other specimens examined, indicate that there are no differences between these taxa. Geminate spikelets, characteristic of Vene- zuelan specimens, are also occasionally present in other specimens from Bolivia and Brazil. anicum subtiliracemosum is also considered The Brazilian specimens cited by Renvoize (1987) under P. sub- tiliracemosum include Dombrowski 2899 and Hatschbach 38052, which are included under Р. schawckeanum Mez, a species of section Parvi- folia (A. Hitchc. & Chase) Pilger. Hatschbach et al. 13731 (K), considered by Renvoize as an in- termediate specimen between P. stigmosum and to be a synonym of P. hebotes. P. subtiliracemosum, is indeed P. surrectum. Chase 9408 has peculiar cordate blades; however, it shares similar spikelets, dark anthecia, and asymmetrical blades with other specimens of P. hebotes Anatomically, P. hebotes has a distinctive keel incorporating three vascular bundles (Fig. Adaxial ribs and wide, shallow furrows are present, and the mesophyll is diffuse but nevertheless of the semiradiate type (Fig. 8B, D). The three-celled . 9B, D) are unique to this species in the section and occur on all three specimens microhairs (Fig examined anatomically. 16. Panicum heliophilum Chase ex Zuloaga & Morrone, Ann. Missouri Bot. Gard. 78: 152. 1991. TYPE: Brazil. Minas Gerais: Cha- péu do Sol, Serra do Cipó, 110 km NE of Belo Horizonte, 900 m, 28 Mar.-1 Арг. 1925, Chase 9147 (holotype, US; isotypes, F, GH, MO, NY). Figure 27. Culms 50–70(– 100) cm tall, tangled, leaning on vegetation, the Short-rhizomatous perennial. basal portion decumbent and geniculate, the upper portion erect, freely branching; internodes cylin- dric, pilose, solid toward the base, otherwise hollow; nodes many, villous. Sheaths striate, 0.5-4 cm long, strongly distichous, hispid with short hairs, the margins long-ciliate; auricles small, pilose; col- lar brownish or purplish, covered with short whitish hairs. Ligule a ciliate membrane, 0.2-0.7 mm long, the cilia 0.1–0.6 mm long. Blades linear- lanceolate, 4-5(-12) cm long, 0.3(-1) cm wide, flat or folded, subcordate, densely hirsute on both surfaces, the margins cartilaginous, scabrous and short-ciliate. Primary inflorescences terminal, ex- serted, peduncles hispid, to 8 cm long; panicles lax, diffuse, 4-16 cm axis flexuous, sparsely hirsute, eglandular, the pul- long, 3-13 cm wide; main vini brownish, hispid; first-order branches opposite or alternate, the lower ones whorled or not, di- verging from the axis; axis of the branches short- hirsute or glabrous; pedicels smooth, glabrous, 2- 13 mm long. Axillary panicles similar to the ter- minal one, smaller. Spikelets solitary, ellipsoid, 2.6-3.1 mm long, 1-1.1 mm wide, hirsute, green- ish, without a stipe between the lower and upper glume. Lower glume 2-2.7 mm long, % the length of the spikelet, acuminate, not embracing the upper glume at its base, 3-nerved, the nerves anasto- mosing toward the distal portions Upper glume -nerved. Lower pa- lea elliptic, 2 mm long, 0.3 mm wide, hyaline; and lower lemma subequal, 7 lower flower absent. Upper anthecium narrowly ellipsoid, 2.4-2.7 mm long, 0.8-0.9 stramineous, brownish at maturity, papillose, with mm wide, simple papillae evenly distributed over the lemma and palea, densely pilose, with appressed, long hairs toward the apex of the lemma and palea; lemma 5-7-nerved, rachilla prolonged or not into a short mucro above the upper anthecium; lodicules ca. 0.6 mm long; stamens 3, the anthers ca. 1.3 mm long, purplish. Caryopsis ellipsoid, 1.4 mm long, 162 Annals of the Missouri Botanical Garden 0.8 mm wide, brownish; hilum oblong, embryo Уз the length of the caryopsis. Distribution and ecology: campos rupestres in Minas Gerais, Brazil, in sandy soils between 900 and 1,400 m Additional specimens examined. BRAZIL. MINAS GERAIS: Diamantina, summit of Serra de San Antonio, 1,400 m, Chase 10354 (US); Mun. Jaboticatubas, Lagoa Santa a Conceigáo do Mato Dentro, Senduls urman 1920 (SI, SP); km 114 ao largo da rodovia Lagoa Santa- Conceição do Mato Dentro, 1,160 m, Sendulsky 1609 (SP); Saramenha, Gomes 3955 (US); Serra de Santo An- tonio, 2-5 km e Diamantina, 1,050-1,100 m, Burman & Sendulsky 696, 697, 709 (SI, SP); Serra do Cipó, ao longo da rodovia Lagoa Santa—Conceigao, Sen- dulsky 446 (SP); Serra do Cipo, Burman 525 диын 485 (SP); Serra de Itatiaia e Serra de Lavras as, Burman 384 (SP); Serra de Ouro Branco, Cas- UE 25628 (HBy; Serra de Cipó, Mun. Jaboticatubas, km 116 ao longo da rodovia Lagoa Santa, 6 June 1970, Joly et al. 100 (SP); Metallúrgica, Serra de Ouro Branco, 1,250-1,600 m, Chase 10289 (MO = Affinities of P. adenorhachis are discussed in Zuloaga & Morrone (1991). 17. Panicum itatiaiae Swallen, Phytologia 14: 82. 1966. TYPE: Brazil. Rio de Janeiro: Parque Nacional Itatiaia, Picada Macieiras, 1,700- 1,800 m, 18 Jan. 1925, Chase 8327 (holo- type, US 1255836; isotypes, F 561591, GH, NY, US 1258426). Figures 28, 30. Culms geniculate, decumbent and rooting at the lower nodes, ascendent and branch- Perennial. ing at the upper nodes, more than | m tall, inter- nodes 3-14 cm long, terete, rigid, hollow, glabrous, nodes purplish, glabrous. Sheaths 3-8 cm long, usually shorter than the internodes, glabrous with only one margin ciliate. Ligules 0.3-0.4 mm long, membranous at the base and short-ciliate or lacin- iate at the distal portion; collar purplish, short- pilose. Blades lanceolate or ovate-lanceolate, 10- 12 cm long, 1-1.4 cm wide, flat, subcordate, gla- brous, with long hairs close to the ligular region, the midnerve manifest, the margins ciliate and sca- brous. /nflorescences terminal, exserted, peduncles cylindric, to 40 cm long, glabrous; panicles lax, diffuse, 13-18 cm long, 10-20 axis flexuous, smooth, glabrous, pulvini brownish, cm wide; main glabrous, first-order branches divergent, alternate or opposite, the axis of the branches triquetrous, purplish, scabrous; pedicels smooth, glabrous, pur- lish, 3-8 mm long. Spikelets solitary, narrowly ellipsoid, 4.6-5 mm long, 1.6-1.7 mm wide, gla- brous, greenish, tinged with purple, the upper glume and lower lemma subequal. Lower glume 2.6-3.: mm long, / or more the length of the Е acuminate, 5-nerved, glabrous, separated from the upper glume by a short internode. Upper glume 4—4.3 mm long, usually not covering the apex of the upper anthecium, obtuse, (11—)13-пегуед, the nerves anastomosing toward the apex, the outer surface scaberulous, herbaceous. Lower lemma 4.3-4.6 mm long, 13-nerved, the nerves anasto- mosing toward the apex, acute, scaberulous. Lower long, 1- 1.1 mm wide, membranous, the margins short-ciliate; lower palea ovate-lanceolate, 4-4.1 mm flower male, the stamens 3, the anthers 2.2 mm long. Upper anthecium lanceolate, 4—4.2 mm long, 1.3-1.6 mm wide, apiculate, scabrous and sparsely pilose toward the apex, otherwise glabrous, papil- lose, with small simple papillae regularly distributed over its entire surface; lemma 7-nerved, the palea scabrous at its apex. Caryopsis obovoid, 2.5 mm long, 1.5 mm wide, brownish; hilum punctiform, embryo !6 the length of the caryopsis. Distribution and ecology: known only from the type collection made in the Sierra de Itatiaia, in Ко de Janeiro, Brazil, in forests at 1,700- 1,800 m; collected under bamboo, with the base long- decumbent. When describing this species, Swallen indicated that spikelets were 4-4.5 mm long, but none of the types examined had any spikelets within this range. 18. Panicum laxiflorum Lam., Encycl. 4: 748. 1798. Dichanthelium laxiflorum (Lam.) Gould, Brittonia 26: 60. 1974. Panicum di- chotomum L. var. laxiflorum (Lam.) Beal, Grasses N. Amer. 2: 139. 1896. TYPE: United States. Without locality and collector (holo- type, P-LAM; fragment and photo, US 2808927). Figure 27. Panicum a e HBK, Nov. Gen. & Sp. 1: 103. 1816. TYPE: Mexico. Veracruz: near Jalapa, Hum- boldt & Bonpland s.n. oo P). Panicum ruprechtii Fourn c. Pl. 2: 21. 1886, non enzl, 1854. TYPE: V Veracruz Jalapa, Gale- otti 5733 (holotype, BR not seen). Panicum ап ra ex ' Ashe yd: те raum Sci. Soc E: United States. Mar land: а Fall рана p type, Panicum pyrifor me Nash, Bull. Torrey Bot. Club 26: 579. 1899. туРЕ: United States. Florida: Lake Coun- 12-31 Маг. 1894, Nash 239 (holotype, NY; isotype, a 221672). Panicum xalapense HBK var уы гатеит А. Нисһс. & Chas kj eile U.S. Natl. erb. 15: 161. 1910. TYPE: United States. M in small close bunches, Jackson, 28 Apr. 1906, Hitchcock 1311 (holotype, US 558449) 23 May 1897 (iso. ~ a Eustis, Volume 80, Number 1 1993 Zuloaga et al. 163 Revision of Panicum subg. Dichanthelium sect. Dichanthelium Perennial. Culms densely caespitose, fascicu- late, internodes very short at base of the plants, culms decumbent, profusely branching at the lower -30(-40) cm tall; internodes terete, glabrous, hollow; nodes brownish, densely pilose with long whitish hairs. Sheaths stri- ate, hirsute with rigid papillose-pilose hairs, the nodes, then erect, few-noded, margins membranous. Ligules membranous-cili- ate, 0.3-0.9 mm long, the pseudoligule absent; collar pale, densely pilose. Blades lanceolate, 3— 15 cm long, 0.2-0.8(-1) cm wide, flat, narrowed at base, acuminate, the margins long-ciliate, papil- lose-pilose on both surfaces or only on the upper portion of the adaxial surface or glabrous, the mid- nerve not manifest. Terminal and axillary inflo- rescences similar, short-exserted or included on the upper leaves; peduncle terete, hirsute or glabrous, to 15 cm long; panicles lax, 2-9 cm long, 2-7 cm glabrous, glandular or eglandular; pulvini long-pi- lose, first-order branches alternate, divergent, the wide; main axis flexuous, densely hirsute or axis of the branches smooth, sparsely hirsute or glabrous, glandular or eglandular; pedicels long, sparsely hispid. Spikelets solitary, plano-convex, ellipsoid to more commonly obovoid, 2-2.7 mm long, 1-1.4 mm wide, short-hirsute, greenish, up- per glume and lower lemma subequal. Lower glume the length of the spikelet, 1-3-nerved. Upper glume ovate-acuminate, 0.8-1.3 mm long, 4-2 acute, 9-nerved, covering the apex of the upper anthecium or not. Lower lemma glumiform, 9-nerved. Lower palea elliptical, 1–1.4 mm long, 0.5-0.7 mm wide, hyaline, glabrous; lower flower absent. Upper anthecium broadly ellipsoid, 1.8- 2.3 mm long, 1-1.4 mm wide, indurate, smooth, shiny, papillose, lemma apiculate, the apex green- ish. Caryopsis broadly obovoid, brownish, 1.3 mm long, 1.2 mm wide; hilum punctiform, embryo a little less than 14 the length of the caryopsis. Distribution and ecology: eastern United States and Mexico to Guatemala, El Salvador, Honduras, Nicaragua, Costa Rica, and the West Indies, in Cuba, the Dominican Republic, and Haiti. Inhabits moist banks and edge of pine and oak woods or mixed hardwood forest, between 700 and 2,700 m. Selected specimens cited. Costa RICA. CARTAGO: 3 m up mountain from Tejar, dea Hwy., Taylor & Taylor 11842 (MO, US). SAN JOSE NW of San Pablo, Pohl & Pinette 13123 (MO). "Сива. ORIENTE: Sierra Maestra, La Gran Piedra above Daiquiri, Ekman 1604 (С, US). PINAR DEL RIO: Consolación del Sur, pineland hills at Juan Moreno, Ekman 10860 (G, NY, US). DOMINICAN a AZUA: Sierra de Ocoa, San José de Ocoa, Ekman 11855 (US). La VEGA: 5 km N of Tireo, NE of un. Gould & Jiménez 14112 (MO). SAN JUAN: Sabana Nueva, 6,200 ft., N of Rio Arriba, Howard & Howard 9079 (US). SANTIAGO: Jiromé, 700 m, Valeur 293 (MO, US). GUATEMALA. ALTA VERA- PAZ: cut-over steep slopes 3 mi. E of Tactic on Route 7E toward Tamahu, Wilbur 14916 (MO). km N of El Chol, Harmon & Dwyer 3168 (MO). Hartı. . des Commissaires, Holdridge 1729 (MO); vicinity of St. Louis du Nord, mountain SW & mem 14503 (US). olón, Izaguirre 53 (MO) Cristóbal, La Esperanza, Ordonez 84 (MO) MORAZAN: Parque La Tigra, Cruz 70 (МО). OCOTEPEQUE: Belén Gualcho, 4 km from town on road from Corquin, Blackmore & Chorley 3838 (MO). OLANCHO: Montana Los Zapotes, 10 km NO de Campamento, Soto 44 (MO). MEX XICO. Chinantlá, Liebmann 328 (MO). cuiAPAs: Mun. m N of fio Dore et al. 29621 (MO). н HIDALGO: a Hwy. 5 between Pachuca and Tampico, along road to San Cristóbal which leaves main highway 100.8 mi. NE of Pachuca vicinity of turnoff, 1. from San Cristóbal, Croat & Hannon 65910 (MO). OAXACA: 3.5 km al S de San Andrés Yaa, en la desviación a Oaxaca, Torres 2020 (MO). PUEBLA: 0.8 mi. 5 of Zacapoaxtla on road from Zaragoza, Brunken & Perino 265 (MO); Trinidad Iron Works, Pringle 13250 (MO). QUERET ^RO: off road from Jalpán to Xilitla, 5.8 mi. of San s Potosi state line, Thomas 2794 (MO). VERACRUZ: 3 pe NE of Jilotepec on road to Naolinco, Nee & Taylor 26237 (MO); Córdova, Bourgeau 2162 (MO). NicARAGUA. ESTELI: lado S del Cerro Tomabü, por el Valle Las Cuevas, Moreno 18455 (MO); N slope of Cerro ЕЈ Fraile, ca. 13°25'N, 86?16'W, Stevens & Mon- tiel 18103 (MO). JINOTEGA: along H km al S de Matagalpa, Moreno 17656, 17694 (MO). 19. Panicum pedicellatum Vasey, U.S.D.A. pedic ellatum (Vasey) Gould, Brittonia 26: 60. TYPE: United States. Co., rocky woods, June 1885, Reverchon 1620 (holotype, US 2383607; isotypes, MO, NY). Figure 27. Panicum transiens Swallen, J. Wash. Acad. Sci. 21: 436. 1 . exico. Tamaulipas: Sierra de San Carlos, A Mesa de Tierra, vicinity of San Jose, 1,000 m, 19 July 1930, Bartlett 10454 (holotype, US 1501526 Tufted perennial. Culms erect to ascending, 15- 30(–60) cm tall, simple at the base, freely branch- ing at the middle and upper nodes; internodes cy- lindric, hollow, shortly pilose with tuberculate hairs, the margins membranous, glabrous or shortly pi- lose. Ligules membranous-ciliate, 0.3-0.6 mm long; collar shortly pilose. Blades lanceolate, 5-8.5 cm long, 0.2-0.8(-1.3) cm wide, flat, subcordate, the 164 Annals of the Missouri Botanical Garden lower margins long-ciliate, the adaxial surface pi- lose with appressed, whitish hairs, the abaxial sur- face shortly pilose or glabrous. Inflorescence ter- minal and axillary from the uppermost nodes, short to long-exserted, peduncle cylindric, 5-20 cm long, shortly pilose; panicles few to multiflowered, 3— 8(-10 the pulvini shortly pilose, branches alternate, di- vergent, the axis of the i paired, unequal, 2-12 mm long, triquetrous, pilose ranches pilose; pedicels or scaberulous. Spikelets narrowly ellipsoid, 3.3— 4.1 mm long, 1.3-1.5 mm wide, short-papillose- ilose, greenish to purplish. Lower glume ovate- acuminate, 1.8-2.2 mm long, У; the length of the spikelet or less, 3—5-nerved, separated from the upper glume by a distinct internode, not embracing the upper glume. Upper glume obovate or ovate, 3.2 mm long, obtuse, not covering the apex of the upper anthecium, 9-11-nerved, the nerves anas- tomosing toward the apex. Lower lemma 9-nerved, pilose. Lower palea 1.2-1.9 mm long, 0.4 mm wide, lanceolate, hyaline, glabrous; lower flower absent. Upper anthecium narrowly ellipsoid, 3— 3.5 mm long, 1.2 mm wide, indurate, smooth, shining, papillose and with prickle hairs at the apex. Caryopsis unknown United States, in Texas, Mexico, and occasionally in Guatemala, Distribution and ecology: growing in moist, shaded places, on rocky slopes, bordering rivers or in open pine or oak woods, to 2,500 m Additional specimens examined. | GUATEMALA. EL PROGRESO: hills between Finca Piamonte and slopes SE of Finca Piamonte, САА 43463 (Е). MEXICO. C OAHUI- LA: middle and u s of Canón de la Hac 102°25'W, Chiang et al. 9452 (МО), Johnston et E 10974 (MO). NUEVO LEON: Dulces Nombres, 1,310 m yer & Rogers 2663 (MO). UNITED STATES. TEXAS: Kerrville, Heller 1736 (MO), 2,000 ft.; Comanche Springs, Lindheimer 1265 (МО); along banks of Guadalupe River, near Kerrville, Palmer 33820 (MO). 20. Panicum penicillatum Nees ex Trin., Gram. Panic. 196. 1826. TyPE: Brazil. With- out locality, Langsdorff s.n. (holotype, LE not seen). Figures 25, 144. Panicum discolor Trin. ex Nees, Agrost. Brasil.: 5 © 5 un "о = o 8 да dE — со © dnd : Brazil. Without locality, Sellow s.n. (holotype, B; Wien BAA, US; isotype, K). Perennial. Culms decumbent, branching at the lower nodes, leaning on vegetation rooting and ) cm long; main axis flexuous, shortly pilose, and scandent, reaching up to 10 m, internodes 8- 18 cm long, solid on the lower portion, rigid, cy- lindric and pilose; nodes dark, thickened, densely pilose with long whitish hairs or glabrous. Sheaths 10-14 cm long, the lower ones shorter and the upper ones longer than the internodes, papillose- pilose with long tuberculate, caducous hairs or gla- brous, one margin ciliate, the other membranous, auricles membranous. Ligules ca. 0.5 mm long, membranous, lacerate on the upper portion; collar brownish, shortly pilose or glabrous. Blades lan- flat, pseudopetiolate, the pseudopetiole brownish, pu- berulous; blades narrowed to subcordate, slightly asymmetrical, the apex long-attenuate, sparsely ceolate, 18-32 cm long, 1.2-3.2 cm wide, pilose or glabrous, the abaxial surface usually glau- cous, the margins scabrous. /nflorescences termi- nal, exserted, peduncle cylindric, glabrous; pani- ered, 30-45 20—40 cm wide; main axis flexuous, scabrous or cles lax, diffuse, multiflow cm long, glabrous, the pulvini brownish, glabrous, first-order branches divergent, alternate or opposite, occa- sionally whorled toward the base, the axis of the branches scabrous; pedicels scabrous, triquetrous. Spikelets solitary, narrowly ellipsoid, 3-4.2 mm long, 1.1-1.4 mm wide, greenish to purplish, the upper glume and lower lemma subequal, acute, with strong nerves. Lower glume 2.8-3. long, as long as or a little shorter than the upper glume, 3-5-nerved, acuminate to subulate, pilose toward the apex, scabrous on the inner surface. Upper glume 3-4 mm long, acute, (7-)9-11- nerved, shortly pilose toward the apex, pilose on the inner surface. Lower lemma glumiform, 3-4.1 mm long, 9—1 l -nerved, pilose on the inner surface. Lower palea absent; lower flower absent. али id, 3.2-3.7 mm long, 1.3 mm wide, stramineous, with simple papillae anthecium narrowly OVOI regularly distributed over its entire surface, lemma apiculate, the apex and upper margins pilose. Сагу- opsis narrowly ellipsoid, 2.2-2.4 mm long, 1.2 mm wide, brownish; hilum punctiform, embryo a little less than ! the length of the caryopsis. Distribution and ecology: Brazil, in the Distrito Federal and the states of Goiás, Minas Gerais, and Rio de Janeiro, in dense colonies at forest edges, where it grows leaning on vegetation, from 450 to 1,500 m. К specimens examined. BRAZIL. DISTRITO FEDERAL: a do Rio Sào Bartoloméu, Equipe IBGE 3572 (IBG EX próximo a Sobradinho, Filgueiras 1964 (IBGE, MO). Goias: capoeira and gallery margin, са. 12 km 5 de Corumbá de Goiás, 1,000 m, pos et al. 10848 (MO, NY, US); Santo Antonio do Descoberto, fazenda de Agricultura Natural, 13 Dec. 1989, Alvarenga & Oli- veira 525 (IBGE, SI). MINAS GERAIS: Vigosa, Chase 9465, Volume 80, Number 1 1993 Zuloaga et al. 165 Revision of Panicum subg. Dichanthelium sect. Dichanthelium (F, US), 640 m, between Sào Geraldo and Vigosa, Chase 10200, 10217 (US), Kuhlmann 1942 (IAN, RB), Pires & Black 2844 (IAN, US); Fazenda do Aguada, Cabeceira MO, NY, US); Serra do Espinhago, Serra do Itabirito, 50 a SE of Belo Horizonte, Irwin et al. 19727 (MO, UB, US); Barbacena, Serra Mantiqueira, Chase 8716 (F, GH, NY, US); Anna Florencia, E of Ponte Nova, Chase ponga, Chase 9608 (F, GH, MO, US); Lagoa Santa, Warming s.n. (NY). RIO DE JANEIRO: Antonio, Glaziou 5718 (K, P, 05); Corcovado, Chase 9752 (MO); Serra dos Orgáos, Luetzelburg 6051 (NY). Without locality, Riedel 2151 (K, US), Glaziou 16626, 18633 (US), 16628 (P, US), 22612 (G). Panicum penicillatum has a peculiar, bamboo- like habit, with rigid culms leaning on the vegetation and reaching up to 10 m high (T. Filgueiras, pers. comm.). The spikelet also lacks a lower palea. 21. Panicum peristypum Zuloaga & Mor- rone, Novon 1: 113. 1991. TYPE: Brazil. Es- pirito Santo: Mun. Domingos Martins, Domin- gos Martins, selva en cerro del Dr. Kautsky, 600-850 m, 7 May 1985, Zuloaga, Mar- tinelli & Vázquez Avila 2398* (holotype, RB; isotypes, MO, SI, US). Figures 10, 28. Short-rhizomatous perennial. Culms decumbent and rooting at the lower nodes, then erect, ca. 80 cm tall, internodes 7-12 cm long, compressed, striate, glabrous; brownish. Sheaths striate, 5-9(-18) cm long, usu- ally shorter than the internodes, glabrous, shiny, the margins membranous, ciliate toward the distal nodes compressed, glabrous, portion. Ligules 0.3 mm long, membranous at the base, then shortly ciliate, collar shortly pubescent. Blades ovate-lanceolate, 7— wide, cordate, amplexicaulous, acute, velutinous, cm long, 1.5— the lower margins long-ciliate, otherwise ciliate. Inflorescences terminal, exserted, panicles lax, dif- fuse, 15 cm long, 8 cm wide; main axis sparsely pilose, eglandular, the pulvini pilose, first-order branches alternate or subopposite, the axis of the branches scabrous, flexuous, eglandular; pedicels claviform, flexuous, scaberulous. Spikelets narrow- ly ellipsoid, 2.4-2.5 mm stipitate, glabrous, greenish or tinged with purple, long, 1 mm wide, non- the upper glume and lower lemma subequal. Lower glume ovate, 1.2-1.4 mm long, Y or a little more the length of the spikelet, glabrous, 1-3-nerved, not embracing the upper glume. Upper glume 1.9— 2.3 mm long, (7—)9-nerved, with manifest nerves, not covering the tip of the upper anthecium. Lower lemma glumiform, 2—2.3 mm long, 9-nerved. Low- er palea lanceolate, 1.7 mm long, 0.4 mm wide, hyaline, glabrous; lower flower absent. Upper an- thecium ellipsoid, 2.2 mm long, 0.8 mm wide, indurate, papillose, lemma crestate, scaberulous; the stamens 3, anthers 1.4 mm long, stigma pink. Caryopsis unknown. Distribution and ecology: known only from the type collection in edge of forest in Espirito Santo, Brazil, at 650-850 m. Related to P. sciurotoides, P. peristypum has amplexicaulous, i 2.4-2.5 mm long, and lower palea 1.7 mm long. 1.5-3-cm-wide leaves, spikelets It differs from P. sabulorum and P. stigmosum by having narrowly ellipsoid spikelets (obovoid in P. sabulorum and P. stigmosum), lower glume not embracing the upper glume at its base, and without a short stipe between the lower and upper glume. Panicum peristypum also lacks the foliar dimor- phism present in P. sabulorum and has, as another difference with P. stigmosum, amplexicaulous leaves. The keel structure of P. peristypum deserves comment. Two different leaves of Zuloaga et al. 2398 were examined, and the leaf illustrated (Fig. 10A) clearly has a keel containing three vascular bundles as well as extensive adaxial parenchyma tissue. However, the other leaf from the same plant is without a keel or even a midrib, having a median vascular bundle distinguishable only by its central location in the lamina 22. Panicum petropolitanum Zuloaga & Morrone, Ann. Missouri Bot. Gard. 78: 161. 1991. TYPE: Brazil. Rio de Janeiro: Mun. Pe- tropolis, Araras, base de Pedra Maria Com- prida, saxicola, heliofila, crescendo em beira de Rio, 23 Mar. 1968, Sucre & Braga 2553 (holotype, RB). Figure 28. Stoloniferous perennial? Culms geniculate, root- ing and branching at the lower nodes, then becom- ing erect, freely branching, 10-25 cm tall, few- noded; 1-4 cm long, cylindric to compressed, hollow, glabrous; nodes compressed, internodes brownish, covered with long hairs to glabrous. Sheaths striate, 1.5-5.5 cm long, longer than the internodes, one of the margins long-ciliate with whitish hairs, the other one membranous or with long hairs in the upper portion, the rest of the surface glabrous. Ligules membranous-ciliate, ca. 0.2 mm long; collar brownish, glabrous. Blades linear-lanceolate, 2-5 cm long, 0.1—0.3 cm wide, flat, glabrous, the margins scaberulous, the apex attenuate. Inflorescences terminal, exserted, pe- duncles to 16 cm long; panicles lax, diffuse, 2-4 cm long, 1–3.5 cm wide; main axis flexuous, cy- lindric, glabrous, the pulvini glabrous, first-order 166 Annals of the Missouri Botanical Garden branches alternate, divergent, and distant, the axis of the branches glabrous, eglandular; pedicels tri- quetrous, scaberulous, 1-7 mm long. Spikelets solitary, narrowly ellipsoid, 2.4-2.6 mm long, 0.9 wide, glabrous, the upper glume and lower lemma subequal, without a stipe between the lower mm and upper glume. Lower glume 1.8-2 mm long, 34 the length of the spikelet, lanceolate, not em- bracing the upper glume, 1(-3)-nerved, the mi nerve scabrous toward the apex. Upper к! (7—)9-nerved, acute, embracing the lower lemma. Lower lemma 9-nerved, embracing the upper an- thecium. Lower palea lanceolate, 1.5 mm long, 0.3-0.4 mm wide, hyaline, glabrous; lower flower absent. Upper anthecium narrowly ellipsoid, 2.3 .7 mm wide, glabrous, shiny, with simple papillae evenly distributed over its entire surface; mm long, lemma apiculate, apicule ca. 0.2 mm long, pilose; lodicules ca. 0.3 mm long, embracing the lower margins of the palea; the stamens 3, anthers 1 mm long. Caryopsis ellipsoid, 1.2 mm long, 0.6 mm wide; hilum punctiform, embryo less than Уз the length of the caryopsis. Distribution and ecology: Brazil, in mountains of the state of Rio de Janeiro, growing in open and humid places on borders of streams, on rocky soils. Panicum petropolitanum differs from P. cu- caense in its stoloniferous habit; culms geniculate, few-noded, rooting and branching at the lower nodes and then becoming erect; linear-lanceolate, flat leaf blades; and inflorescence with a peduncle to 16 cm long. 23. Panicum portoricense Desv. ex Ham., Ргодг.: 11. 1825. Dichanthelium portori- cense (Ham.) Hansen & Wunderlin, Ann. Mis- Wet Bot. Gard. 75: 1649. 1988. TYPE: Puer- o Rico. Without locality, Desvaux Herb. s.n. (holotype, P). Figure 27. 223. 1826. Panicum lancearium Trin., Gram. Panic.: TYPE: Uni b Circ. 7: 79. 1897. Panicum port nashianum (Scribner) Lelong, Brittonia 30: 267. 1984. TYPE Fs States Florida : Lake Cou Eustis, 30 . 1894, Nash 466 (holotype, "Us 208 Panicum о Nash, Bull. Torrey Bot. Club 574. 1899. TYPE: а States. Florida: Lake Cou 1894, Nash 72 буе: US). Pam па Зоре: уаг. рн Scribner & ., U.S.D.A. Div. Agrost. Circ. 27: 9. Dec. 1900. Pan anicum patulum (Scribner & Merr.) A. Hitchc., Rhodora 8: 209. 1906. Panicum lancearium Trin. var. patulum (Scribner & Merr.) Fernald, Rhodora 36: 80. 1934. Dichanthelium sabulorum (Lam.) Gould & Clark var. patulum (Scribner & Merr.) Gould & Clark, Ann. Missouri Bot. Gard. 65: 1113. 8 : Manatee Coun- ty, Bradenton, 3 Sep. 1898, Combs 1296 (holotype, US 2808982). P; pak eg Ashe, J. Elisha Mitchell Sci. Soc. 16: . 1900. TYPE: United States. North Carolina: New P. pauciciliatum red Е 7 00. TYPE: United States. North Carolina: New Hanover County, near Wilmington, 20 May 1899, Ashe s.n. (lectotype, NCU not seen; isolectotype, ). Caespitose perennials, with а basal rosette of broader leaves present or absent. Culms decumbent to erect, 4-30 cm tall, freely branching at the upper nodes, internodes cylindric, greenish to pur- plish, hollow, shortly and densely pilose; nodes shortly pilose. Sheaths striate, greenish to purplish, shortly pilose, the margins ciliate. Ligules mem- branous-ciliate, 0.3-0.7 mm long. Blades lanceo- late, 1-5 cm long, 0.2-0.6 cm wide, flat, subcor- date, the apex acuminate, greenish or occasionally purplish, shortly and densely or sparsely pilose on both surfaces, the lower margins long-ciliate or completely ciliate. /nflorescences terminal, exsert- ed or partially included in the upper leaves, pe- duncles cylindric, shortly pilose, to 4 cm long; panicles lax, few to multiflowered, 1-4 cm long; main axis flexuous, shortly pilose, first-order branches alternate or subopposite, divergent, the axis of the branches and pedicels shortly pilose. Axillary panicles similar to the terminal one. Spike- 1.6-2.4 mm long, 1-1.3 mm wide, shortly hispid, the upper glume and lower lets solitary, obovoid, lemma subequal, or the upper glume shorter and not covering the upper anthecium. Lower glume 0.7-1.1 mm long, М ог more the length of the spikelet, ovate, obtuse to truncate or acute, 1–3- nerved. Upper glume 7—9-nerved, obtuse. Lower lemma 7-nerved, obtuse. Lower palea lanceolate, 0.6-0.8 mm long, 0.4 mm wide, hyaline, glabrous; lower flower absent. Upper anthecium obovoid, 1.2-1.5 mm long, 0.6-0.8 mm wide, pale, smooth, shiny, indurate, with simple papillae, the lem 7-nerved, shortly apiculate, bn at the apex. Caryopsis obovoid, 0.7— .9-] mm wide; hilum punctiform, pace less than % the length of the caryopsis. Distribution and ecology: United States, Mex- nm long, ico to Belize and the West Indies, in moist sandy pinelands, to 1,400 m. Additional specimens examined. BELIZE. Pine Ridge N of aviation field, Bartlett 11225 (US). CUBA. ISLA DE LA JUVENTUD: vicinity of Siguanea, Britton et al. 15379 (US). ORIENTE: Sierra Maestra, La Gran Piedra, 1,200 Volume 80, Number 1 1993 Zuloaga et al. 167 Revision of Panicum subg. Dichanthelium sect. Dichanthelium m, Ekman 8151 (US); Sierra de Nipe, in pineland, be- n еа se Loma Mensura, Ekman 15290 (C). DEL RIO: o Mantua, Damují, in white sand at Pr ifia бома 11022 (US); Sábalo, іп ан on white sand, Ekman 11427 (С, US); sandy Sabana de los Remates, León & Victorin 18702 (US). DOMINICAN REPUBLIC. LA VEGA: La Vega, Cordillera Central, Con stanza, Loma del Medio, Ekma 14094 (G, US); prope Constanza, von Tuerckheim 3321 (G, US); Montecristi, . Valeur 4 (US). PUERTO Rico. San 0588 (P); track from Manati to Vega Baja, Underwood & Griggs 955 (US); Santurce, Heller 6442 (MO, US); vicinity of San Juan, between Catano and Guainabo, pis 6631 (US); Laguna Tor- 53 (MO, US); W end of Laguna — UNITED STATES. FLORIDA: Lake County, Eustis, Vash 1 337, 24. Panicum pycnoclados Tutin, J. Bot. 72: 340. 1934. TYPE: Guyana. Kaieteur savanna, Potaro River, 1,100 ft., 20 Aug. 1933, Tutin 508 (holotype, BM not seen; isotypes, K, US). Figures 11, 28, Panicum ае иа Swallen, Brittonia 3: 149. 1939. TY enezuela. Bolivar: Mount Auyan-tepui, 2,200 m, Tate 1286 (holotype, US 1723625; fragment, VEN 222501; isotype, NY). Panicum tiricaense Swallen, Mem. New York Bot. Gard. 9: 400. 1957. TYPE: Venezuela. Bolivar: upper falls of Rio Tirica above summit camp, alt. 1,940- 1,950 m, central section, Chimantá Massif, 7 Feb. 1955, Steyermark & Wurdack 535 holotype, US 2182126; isotypes, F 1480635, K, iin gt. Swallen, Mem. New al Bot . Gard. 1957. TYPE: Venezuela. Bolivar: frequent in | у along Rio Tirica, below summit camp, central section, Chimantá Massif, 1,925 m, 5 Feb. 1955, Steyermark & Wurdack 475 (holotype, US 2182124; isotypes, G, K, NY). Perennial. Culms leaning on vegetation, decum- bent, rooting and branching at the lower nodes, then ascendent to erect, freely branching at the upper nodes, 12-60 cm tall; internodes 3-15 cm long, terete, hollow, glabrous or hirsute, nodes com- pressed, pubescent with retrorse hairs. Sheaths 2— 5 cm long, glabrous or hirsute, delicate, the mar- gins long-ciliate. Ligules membranous at the base and short to long-ciliate at the apex, (0.2-0.3-)1— 1.5 mm long; collar pilose with whitish hairs. Blades lanceolate or ovate-lanceolate, 2—-4.5(- 12) cm long, 0.5-1.8 ст wide, herbaceous, flat, divergent, gla- brous or sparsely pilose to densely hispid, cordate, the base amplexicaulous and asymmetric, the apex acute, the margins scabrous, the lower margins long-pilose. /nflorescences terminal, exserted, pe- duncle hispid to glabrous, 6-20 cm long; panicle lax, 3-6(-10) cm long, 1.5-6 cm wide; main axis glabrous or hirsute with long papillose-pilose hairs, eglandular, the pulvini pilose or glabrous, first- order branches flexuous, ascendent, alternate or subopposite, occasionally whorled toward the base, glabrescent, eglandular; pedicels pilose toward the distal pode or glabrous, smooth. Spikelets ellip- soid, = .9 mm wide, glabrous or in acute and attenuate at the base, .3) mm long, 0.8-0 with a stipe between the lower and upper glume, upper glume and lower lemma subequal. Lower glume ovate-acuminate, (0.7-)1.2-1.8 mm long, 14-39 the length of the spikelet, 3(—5)-nerved, em- bracing the upper glume. Upper glume 1.9-2.7 mm long, 9-nerved, not covering the apex of the upper anthecium. Lower lemma glumiform, 7-9- nerved. Lower palea linear-lanceolate, 1.5-1.8 mm long, 0.3 mm wide, hyaline, glabrous; lower flower absent. Upper anthecium ellipsoid, 1.8-2.1 mm long, 0.8 mm wide, indurate, pale, lemma slightly crested at the apex, scabrous and pilose; stamens 3, anthers 0.8 mm long. Caryopsis ellip- soid, 1.5 mm long, 0.8 mm wide, hilum punctiform, embryo / the length of the caryopsis. Distribution and ecology: Ecuador, Peru, Ven- ezuela, Guyana, Surinam, and Brazil, on forest edges, often leaning on vegetation, or in open plac- es, between 500 and 2,600 m Additional specimens examined. BRAZIL. BAHIA: Rui Barbosa, Almeida 22 (US); Serra do Tombador, NW of Jacobina, on BR-234 to Ouro Branco, 950 m, Harley et al. 16634 (K); Serra do Sincorá, Lagoa Encantada, 19 km NE of Ibicoara, Harley et al. 15783 (CEPEC, K, US). CEaRa: Campo Grande, Swallen 4531 (US). RORAI- MA: without locality, Rondon s.n. (RB 110790). Ecua- DOR. LOJA: between La Toma and Loja, 1,800-2,600 m, Hitchcock 21420 (US). Guyana. Kaieteur Plateau, Ma- 1 К, NY); vicinity of Kaieteur n of Potaro Ec ca. 1,400 ft MARCA: dob a 7k Socota, Sánchez-Vega 2295 (SI). SuRINAM. 5 Lucie Rivier, Irwin et al. 55684 (K, SI, US); 14 km N of Lucie Rivier, Maguire et al. 54280 (US). VENEZUELA. AMAZONAS: Dpto. Atabapo, Cerro Huachamacari, base o main wall hey slope below it, E side, 800- 1,300 m, forest, 3?40'N, 65?43'W, Liesner 25867 (MO). ANZOATEGUI: Distr. Freites, burro trail between San Durnal and Pajaritos, Davic : González 19792 2 (VEN): бео Peonia, above Los Pajaritos, 31 km NE of Bergantin and N of Mundo Nuev ин nia of Turimiquire, 1, - ‚700 Сог 19914. 4 (MO, V EN) Distr. Libertad: ridges and tops of Montañas Negras, along the Sucre and Anzoátegui border, 20 airline km NE of Bergantin, NE of Buenos Aires, Serranía de Turimiquire, 2,000 m, Davidse & González 19563 (MO). BOLIVAR: Gran Sabana, formación Roraima, 6 al 5 de m, Davidse et al. 4749 (MO, US, VEN), 4783a ne. al VEN); la Gran Sabana, ca. km 167, S of El Dorado along Hwy. to Santa Elena, 24 km S of La 168 Annals of the Missouri Botanical Garden E NA. BE XG FIGURE 34. Panicum pycnoclados. — A. Habit. — B. Detail of clasping base of the blade. — C. Spikelet, lower glume view. — D. Spikelet, lateral view. — E. Spikelet upper glume view. — Е. did palea.—G. Upper anthecium, lemma view. —H. Upper anthecium, palea view.— I. Detail of apex of the lem = Upper palea with lodicules, мини Je stigma. — К. Caryopsis, embryo view. —L. Caryopsis, hilum view. (Al based on mal 87974. Volume 80, Number 1 1993 Zuloaga et al. 169 Revision of Panicum subg. Dichanthelium sect. Dichanthelium Ciudadela, Davidse 4763 (MO, SI, VEN); a 12 km al N de La Ciudadela, Gran Sabana, 1,400 m, 5?50'N, 61927 У, Zuloaga et al. 4474 (MO, SI, VEN); entre Piedra de la Virgen y la parte alta de la Escalera, carretera a la Gran Sabana, 500-1,400 m, 6?0'N, 61?25'W, Zu- loaga et al. 4404* (MO, SI, VEN); Distr. Piar, Chimantá Massif, 1,600-1,700 m, Steyermark 75642 (NY, US), 128196 (SI); Sierra Тиш, Rio Toro, between Rio La Reforma and Puerto Rico, N of El Palmar, 200 m, Stey- S, V 109377 (NY); Jaua, astor centro-meri del Río Marajano, 4°48'N, 64?32'W, Huber 13043, 13046 (MO). Zuloaga & Judziewicz (1991) mentioned that the specimen Tutin 651 is marked as type of P. pycnoclados in BM. This is one of the species of Panicum found at higher elevations, being present in Ecuador to 2,600 m. Swallen (1939, 1957) described three species from Venezuela: P. albociliatum, P. tiricaense, and P. tiricaoides. No morphological differences were observed between these species and P. pycno- clados. Panicum pycnoclados is characterized by hav- ing asymmetrical, lanceolate or ovate-lanceolate leaves and spikelets 2.2-3(-3.3) mm long with a conspicuous internode between the lower and upper glume. While most of the specimens of P. pycno- clados examined have glabrous spikelets, some populations have spikelets with short hairs. Panicum pycnoclados shows close affinities to P. stipiflorum, and the latter species may represent a variant of P. pycnoclados. Panicum pycnocla- dos has spikelets with the lower glume 12-% the length of the spikelet, and blades with only the lower margins pilose. Panicum stipiflorum has spikelets with the lower glume usually 14 the length of the spikelet and blades with long-ciliate margins. Anatomically, the specimen studied has a defi- nite keel that is structurally distinguishable from the other first-order vascular bundles due to the presence of additional sclerenchyma tissue in the form of adaxial and abaxial girders (Fig. 11A). The cushion-based macrohairs of P. pycnoclados are of interest, as this hair type is not common in the section. The hair bases are clearly sunken between numerous raised and modified epidermal cells which form the cushion (Fig. 1 1B). The hairs themselves are also much thicker than the needlelike hairs that occur on other species such as P. cumbucana (Fig. 6C), P. hebotes (Fig. 9A-D), and P. peris- typum (Fig. 10C, D). The thick, cushion-based macrohair type is best developed in P. руспос!а- dos, although similar hairs do occur in P. visci- dellum (Fig. 19), some specimens of P. acumi- natum (Fig. 3E, F), and near the leaf blade margin in P. davidsei and P. sabulorum (Fig. 12A) 25. Panicum sabulorum Lam., Encycl. 4: 744. 1798. Dichanthelium sabulorum (Lam.) Gould & Clark, Ann. Missouri Bot. Gard. 65: 1112. 1978. TYPE: Uruguay. Montevideo: Montevideo, Commerson s.n. (holotype, Р; fragment, BAA; isotypes, R, 05). Figure 12. Short-rhizomatous perennials, with two different kinds of culms: young culms simple, erect, with broad, cordate, lanceolate leaves; older culms erect to decumbent, occasionally rooting and profusely branching at the lower and upper nodes, to ] m long, with leaves smaller than those of the primary culms; internodes hollow, cylindric, hirsute, nodes ths 2.5-6 cm long, hirsute to glabrous, the margins ciliate, densely pilose to glabrous, dark. Sheat lands present or absent, more conspicuous toward the distal portion. Ligules membranous-ciliate, 0.4— 1.1 mm long. Leaves lanceolate, 1-9(-15) cm long, 0.2-1.4(-2) cm wide, flat, the base subcor- date or cordate, amplexicaulous, the apex acumi- nate, rigid to herbaceous, hispid, with short, stiff hairs or with papillose-pilose hairs, or glabrous, the lower margins long-ciliate, otherwise scabrous, the Primary midnerve inconspicuous. inflorescences ax, diffuse, few- to multiflowered, panicles 2-10 — cm long, 2-8 cm wide; main axis hirsute or gla- brous, glandular or eglandular, first-order branches alternate, divergent, the axis of the branches long- hirsute to scabrous and glandular or eglandular; pedicels claviform, hirsute, glandular or eglandu- lar; axillary panicles similar to the terminal one, smaller. Spikelets obovoid, biconvex, 1.9-3 mm long, 0.9-1.4 mm wide, paired or solitary, shortly stipitate, greenish and tinged with purple, hirsute to glabrous, glumes and lower lemma with manifest nerves. Lower glume 2—% the length of the spike- let, 3-5-nerved, the nerves anastomosing toward the apex, ovate, obtuse to acute and embracing the upper glume on its base. Upper glume 7-9- nerved, obtuse, not covering the apex of the upper anthecium. Lower lemma glumiform, 7—9-nerved. Lower palea ovate, 1.6 mm long, 0.8 mm wide, hyaline, the margins short-ciliate; lower flower male or absent. Upper anthecium broadly ellipsoid, 1.8— 2.5 mm long, 0.7-1.2 mm wide, stramineous, shiny, shortly apiculate, with simple papillae in regular rows; lemma 5-nerved. Caryopsis broadly ellipsoid, 1.5 mm long, 1 mm wide; hilum punctiform, em- bryo less than half the length of the caryopsis. Кеу TO THE VARIETIES 1. Leaf blades clasping the culm 25c. var. cordatum 1. Leaf blades not clasping the culm 02 170 Annals of the Missouri Botanical Garden 2. Spikelets (2.2-)2.4-3 mm long, pilose; blades pilose; plants growing in open, Rd places ..... var. sabulorum 2. Spikelets 1.9-2.3(-2.6) mm bo glabrous or sparsely pilose; blades glabrous; plants growing on forest edges in humid places 25b. var. polycladum 25a. Panicum sabulorum var. sabulorum. Figure 32. P. acutatum Steudel, Syn. Pl. Glumac. 1: 86. 1853. ile. Concepción: Concepción, D'Urville s.n. са ES fragments BAA, US 8044 Panicum latiglume Doell in C. Martius, Fl. Pras, 2(2): 257. 18 di ee latiglume var. villosum nie ЕС Martius, Fl. Bras. 2(2): 257. 1877. Pan i че Del f. Pun (Doell) E. Ekman, pus Вог. 1300): 2 913. Panicum latiglume Doell К к D E. Ekman, Ark. Bot. 13(10): 1913, nom. Шер. TYPE: Brazil. Santa Catarina: Ida de Santa Catarina, Gaudichaud 90 (isotype, Р; fragment, US 80869). Panicum latiglume Doell var. decanum Doell in С. Martius, Fl. Bras. 2(2): 257. 1877. Panicum lati- glume Doell f. decalvatum (Doell) E. Ekman, Ark. Bot. 13(10): 28. 1913. TYPE: Brazil. Santa Catarina: Isla de Santa Catarina, Gaudichaud 91 (fragment, US 80869). D um q ese бац Spec. Nov. Regni Veg. 1909. YPES: uay. Montevideo: doi и deed de la Barra de Santa Lucia, Arechavaleta s.n. (syntypes, W, С, BAA). Para- guay. Cordillera: in campis prope Caraguatay, Hass- ler 3124 (syntypes, G, NY, W, fragment, BAA). Distribution and ecology: southern Brazil, Ar- gentina, Paraguay, and Chile; found in open places on sandy soils. Selected ч “Үт? ARGENTINA. BUENOS AIRES: Balcarce, Hunziker 2166 (SI). coRRIENTES: La Cruz, Виг- kart 7943 (SI), Tew 12364 (US); ruta nac. 12, 20 km E de Ituzaingó, Zuloaga et al. 3231 (SI). ENTRE RIOS: arroyo El Palmar, Burkart & Crespo 22927 (SI), Zu- loaga & Morrone 3857 (MO, SI); Colonia Ayuí, Zuloaga & Morrone 3852 (MO, SI); Colón, Palmar, Burkart et al. 23267 (SI); ruta 12, desvío a Holt, arroyo El Cuartillo, Zuloaga 3864 (MO, SI). снасо: Isla del Cerrito, Kra- povickas & Cristóbal 20048 (CTES). BRAZIL. PARANA: Colónia Orleàs, Curitiba, Dombrowski 1958 (CTES); Mun. i r. Graciosa, Hatschbach & Dionisio 417 (US). RIO GRANDE DO SUL: Taimbesinho, prope Sào Francisco de Paula, Rambo 54532 (US); Rio Grande, Swallen 9252 (US); entre Sao Lourenço e Ca- maqua, Pereira 6811 (RB); Bagé, Hulha Negra, Allem & Vieira 1809 (ICN, 51); Sta. Maria, Est. Experimental, 23 Nov. 1955, Camargo s.n. (SI); Osório, estrada Pas nhos a Osório, 7 km antes do entroncamento com a BR 101, margem da Lagoa dos Barros, Valls et al. 2308 (CTES); Lages, Swallen 8140 (US). SANTA CATARINA: Sombrio, Reitz 1978 (NY); Serra dos Ilheos, Smith & Klein 15464 (NY, SI); Ubatuba, Hans 317 (SI); without locality, D'Urville s.n. (W). s&o PAULO: Cananeia, Ilha do Cardoso, Praia do Marujá, da Silva 249 (MO); Sào Paulo, Parque do Estado, Sendulsky 768 (SI), 842 (SI, SP); Santos, Sendulsky 739 (SI, SP). CHILE. CONCEPCION: Concepción, D'Urville 1821 (P), ere s.n. (P). MAULE: without locality, Fuentes s.n., Dec. 1911 (W). PARAGUAY. PARAGUARI: Paraguari, Balansa "ps K, P); i icai de Piribebuy, au desus de Mbatobi, 1 Nov. 1883, Balan 4202 (P). URUGUAY. CANELONES: Ruta Бө Бареа ти Km ico, _ 834, 835, 836, 837 82 (81), Osten 6917 (W), , 51). MoNTEVIDEO: Montevideo, Arechavaleta 144 (W), Montoro 2667 (P); prés Mon- виги B-6217 (SI). km de Tacuarembó, Cabrera & Zuloaga 32424, 32425 (SI). In this variety the spikelets are pilose, (2.2-)2.4-3 mm long, the axis of the inflorescence is hirsute, and the blades are rigid and hirsute. 25b. Panicum sabulorum var. polycladum (E. Ekman) Palacios, in Burkart, Fl. Ilustr. Entre Rios, Colec. Сет. INTA 6(2): 316. 1969. Panicum polycladum E. Ekman, Ark. ot. 11(4): 24. 1912. TYPE: Argentina. Mi- siones: Bonpland, 27 Jan. 1908, Ekman 651 (holotype, 5; fragment, US 80923; isotype, CORD). Figure 32. PANE demissum Trin., Spec. Gram. 3: pl. 319. 1832. :: Brazil. Rio de Janeiro: Rio de Janeiro, Martius s.n. i (Барока LE; fragment, US 557430). Panicum ramosissimum Trin., Mem. Acad. St. Petersb. VI. Sci. Nat. 1: 312. 1834. TYPE: Brazil. Without locality, Sellow s.n. (holotype, LE; fragment, US 31). Panicum pencanum Philippi, Anal. Univ. Chile 93: 713. 1896. TYPE: Chile. Concepción: Concepción, 1888, Philippi s.n. (isotype, CORD; fragment, BAA, US). j delta ; 12: 291. 1968. TYPE: Argentina. Buenos Aires: Delia del Paraná inferior bonaerense, arroyo Chaná Mini, 14 Jan. 1923, Parodi 4928 (holotype, SI; isotypes, BAA, US 1160950). Distribution and ecology: found from the states of Minas Gerais and Rio de Janeiro to southern Brazil; also in Paraguay, Uruguay, Argentina, Bo- livia, and Chile. Plants grow in shady and wet areas, on forest edges, to 1,500 m Selected specimens cited. | ARGENTINA. BUENOS AIRES: Punta Lara, Zuloaga & Deginani 1891, 3085 (SI), 3859, 3861 (SI); Cerro La Peregrina, 20 km N of Mar del Plata, Eyerdam et al. 23678 (MO, SI, US); Escobar, Paraná de Las Palmas, Zuloaga & Morrone 3079, 3860 (SD, 4543, 4544 (MO, SI). conpoBa: W of Tanti at Rancho Alegre, Renvoize 3743 (SI); between Cerro Blan- co and La Hollada, ruta nac. 20, km 767, Solomon 4091 (MO). CORRIENTES: Puente Pesoa, Quarín 1744 (CTES, km E de Ituzaingó, Zuloaga et al. 3228 (51); Cuay Grande, Zuloaga & Deginani 554 (SI); Estancia Santa Teresa, Pedersen 2997 (P). ENTRE RIOS: ueva Escocia, Bacigalupo et al. 871 (SI); ruta 14 y Río Gualeguaychü, 4 km al N, Bacigalupo et al. 579 Volume 80, Number 1 1993 Zuloaga et al. 171 Revision of Panicum subg. Dichanthelium sect. Dichanthelium (SI); s Jose de Feliciano, Arroyo Feliciano, Bacigalupo et al. 696 (SI). FoRMOsa: without locality, Kermes 346 Єр. JUJUY: Sierra de Zapla, Mina 9 de б Cabrera et al. 32004 (SI); Santa Bárbara, Venturi 9545 (SI, US, pudo 9691 (US). MISIONES: Villa Venecia, Renvoize et al. 4 (P, SI); casa de Horacio Quiroga, Zuloaga et al. e (MO, SI); La Granja, Posadas, Ekman 645 (G, US); al. 19715(MO, y Barbacena, Serra Mantiqueira, Chase 8646 (NY, US), 8674 (NY, US); Mun. Itamonte, PARANA: Passa ы Santa Vitoria do Palmar, Portela, Valls et al. 1780 (CTES, US); Nova. Petr opolis, km 100 da "p? 116, ao N da ponte sobre o Rio Cai, Falls et al. 1664 (CEN, ICN, SI), 1665 (ICN, SI). RIO DE JANEIRO: Corcovado, е 8159 (MO, NY); Pati do Alferes, Paid et s 87 (RB); deis Nacional Itatiaia, picada Macie oaga et al. B. 51, US a 8207 (ВВ. о 8268 (US); pou Nac. drum Rio Campo Belho, Zuloaga et al. 2361* (RB, SI, US); Serra dos Orgáos, 16 km 5 of Itaipara, Davidse et al. 11422 (MO). SANTA CATARINA: 4 km S of Хапхеге, Smith & Klein 13066 (K, NY, SI); W of Chapeco on road to Guatambü, Smith & Reitz 12534 (K, R, US); Mun. Ponte Serrada, 2 km S of Paraná line, fazenda, Smith et ut Joaquim, dank & Reitz 14305 (K, NY, SI, do Jordao, Serra Mantiqueira, Chase se 9894 (US); city of Gio Paulo, Agua Funda, Епеп & Fiten 6236 (MO). CHILE. Taleamavida, Barros 291 (SI). MAULE: ad rivulos pr. Constitución, Reiche s.n. (W). PARAGUAY. Encarnación, Pavetti & n 10855 (US). CAAGUAZU: in viciniis Caaguazü, Balansa 4a (K, P), Hassler 9212 (С, К, , US). CORDILLERA: Eusebio Ayala, Schinini 2423 (CTES), 2583 (CTES, SI). Guaira: Villa Rica, Ba- poene) (С), За (P). URUGUAY. no Toledo, Osten 3 (US). FLORIDA: Río Yi y Arroyo Mansavillagra, К Та Palma, Me 1153 (U JS). eS S TACUAREMBO: Gruta de los Cuervos, Cabrera & Zuloaga 32393 (SI) Panicum sabulorum var. polycladum usually grows in humid and shady places, on forest edges or streams. It differs from variety sabulorum by having smaller spikelets (1.9-2.3(-2.6) mm long) that are glabrous or with scattered hairs toward the apex; herbaceous and usually glabrous blades; and panicles with the main axis and branches com- monly glandular. Glaziou 4305, from Rio de Janeiro, Brazil, was cited by Doell (1877) as P. nodiflorum Lam. (= P. dichotomum L. var. dichotomum). 25c. Panicum sabulorum var. cordatum Zu- loaga & Morrone, Novon 1: 117. 1991. TYPE: Brazil. Paraná: Carvalho, Dusén 13336 (ho- lotype, US; isotype, W). Figure 32. Additional specimens examined. BRAZIL. PARANA: Serrinha, Jonson 1076a (US), Dusén 15596 (NY, SI, US); Trancheira, Јопзоп 1099a (US). SANTA CATARINA: Piloes, 200 m, Reitz & Klein 3225 (NY, US); Mun. Praia ‘Grande, Valls 10087 (CEN, SI). In this variety the spikelets are glabrous, 2-2.2 mm long; manifest glands are present on the gla- brous sheaths and axis of the inflorescences; and the leaf blades are amplexicaulous, cordate, asym- metrical at base, glabrous and up to 15 cm long and to 2 cm wide. 26. Panicum sciurotis Trin., Gram. Panic. 228. 26. Dichanthelium sciurotis (Trin.) Da- 4. 1992. TYPE: Brazil: “V. sp. Brasil," Chamisso s.n. (holotype, LE). Fig- vidse, Novon 2: 10 ure 28. Panicum rostellatum Trin., Mém. Acad. Imp. Sci. Saint- Sci. P. T Steudel, Syn. Pl. Glumac. 1: 85. 1853, Desv. 1831. Panic iurotis var. - не Doell in C. Martius, Fl. Bras. 2(2): 250. 1877. ТУРЕ: Paraguay. Without locality, Rengger s.n. (holotype, P). ай trichopiptum Steudel, Syn. Pl. Glumac. 1: 85. 1853. azil. Bahia: without ped Salzmann s.n. (lectoty id P; fragment, US 29035 Р: Е Salzm. ex Steudel, Syn. Pl. . TYPE: Brazil. Bahia: without oni in sa- vi tein Salzmann 684 (holotype, P; fra ments, US 967933, 974705, 2903510; m G, K, MO, | P. sciurotis Trin. var. а Doell in C. Martius, as. 2(2): 250. 1877. TYPE: Brazil. “In Insula S. Catharinae," ro i s.n. (lectotype, here des- ignated, LE). Annual. Culms geniculate, decumbent, rooting and branching at the lower nodes, then becoming erect, 5-25 cm tall; pressed, hirsute, 2-8 cm long; nodes dark, densely pilose with retrorse whitish hairs. Sheaths striate, 1-5 ст long, hirsute, with long whitish hairs, the margins ciliate. Ligules membranous-ciliate, ca. 0.4 mm long; collar pilose, brownish. Blades ovate- internodes hollow, com- 172 Annals of the Missouri Botanical Garden lanceolate, 1.5-9 cm long, 0.8-2.8 cm wide, her- baceous, flat, hirsute, papillose-pilose with short, rigid hairs, the base amplexicaulous, the margins ciliate, the midnerve inconspicuous. /nflorescence terminal, exserted; peduncle 5-7.5 cm long, flex- uous, hirsute; panicles lax, diffuse, 2-7 cm long, 1-5 cm wide; main axis flexuous, papillose-pilose with long hairs; pulvini pilose; first-order branches divergent, alternate or opposite, the axis of the branches flexuous, smooth, glabrous or with scat- tered long hairs; pedicels smooth, glabrous; axillary panicles present, similar to the terminal one. Spike- lets solitary, ellipsoid, 1.5-1.7 mm long, 0.7-0.8 mm wide, hirsute, with short hairs, the upper glume and lower lemma subequal or the upper glume shorter, the nerves inconspicuous. Lower glume 0.3-0.4 mm long, М or less the length of the spikelet, nerveless, truncate, not embracing the upper glume at the base. Upper glume obtuse, not covering the apex of the anthecium, 5(- 7)-nerved, the nerves anastomosing, obtuse. Lower lemma glumiform, obtuse, slightly inflated at the base, 5-nerved, the nerves anastomosing toward the dis- tal portion. Lower palea lanceolate, 0.9 mm long, 0.2 mm wide, hyaline, glabrous; lower flower ab- sent. Upper anthecium ellipsoid, 1.4-1.5 mm long, 0.5-0.7 mm wide, glabrous, smooth and shiny, short-apiculate, with simple papillae all over its surface. Caryopsis obovoid; hilum punctiform; em- bryo Уз the length of the caryopsis. Distribution and ecology: restricted to north- eastern Brazil, where it grows in sandy soils, in open places or forest edges. Additional specimens examined. BRAZIL. BAHIA: Ba- hia, Chase 7901 (F, RB, SI, US, W); Cruz das Almas, Pinto 941 (US). cEARA: Campo Grande, Swallen 4539 (RB, SI, SP, US); Campo Salles to Crato, Swallen 4303 (K, R, SP); Serra do Araripé, Luetzelburg 26101 (M, US); Serra do Araripe, Novo Exü, Luetzelburg 26141 (M). PERNAMBUCO: Carvarü, Tenorio 66-197 (US); Ga ránhuns, Chase 7791 (US) vicinity of Recife, Chase 7732 1/2 (USy Recife, Tavares 788 (US). RIO GRANDE DO NORTE: Natal, Swallen 4677 (SP, US); Nova Cruz to rie Swallen 4823 (RB, SP, US); entre Parna- e Sao Jose de Mipibú, e 1663 (R, 05); bourn locality, Riedel s.n. (K, M, P, US 974731). The type locality of the species, Santa Catarina, Brazil, is probably erroneous, since P. sciurotis has only been found in northeastern Brazil, in Ceará, Pernambuco, Bahia, and Rio Grande do Norte. Тће same is probably true for the type material of one of its synonyms, P. cordifolium, which was originally cited from Paraguay. Panicum sciurotis differs from P. sciurotoides in having spikelets with the lower glume nerveless and truncate; upper glume and lower lemma 5(- 7)-nerved with the nerves not manifest; upper an- thecium inserted laterally in relation to the axis of the rachilla; and lower lemma inflated at the base. Panicum litigosum Steudel, a species often as- sociated with P. sciurotis, is a synonym of P. brevifolium L., an African species in Bahia, Brazil. 27. Panicum sciurotoides Zuloaga & Mor- rone, Novon !: l. . Dichanthelium sci- urotoides (Zuloaga & Morrone) Davidse, No- 4. 1992. TYPE: Brazil. Minas Gerais: Distr. Diamantina, Christais, near Corriga Duas Pontes, 1,160 m, 13 May 1931, Mexía 5819 a d isotypes, С, K, M, P, R). Fig- ures 14, von 2: 10 Annual. Culms decumbent, geniculate, rooting and branching at the lower nodes, erect portion 10-85 cm tall, branching at the upper nodes; internodes 2.5-11.5 cm long, terete, hollow, hispid to glabrous; nodes dark, compressed, densely pilose Sheaths striate, 3–6.5 cm long, sparsely papillose-pilose to glabrous, the with retrorse whitish hairs. margins ciliate or one margin ciliate and the other one membranous. small, surmounted by long hairs at the base of the blade; collar pilose. Blades ovate-lanceolate, 3-10 cm long, 0.8- clasping and asymmetric, either densely hirsute with short to long appressed tuberculate hairs or short-hispid on the abaxial surface, or glabrous, the margins ciliate toward the base, otherwise sca- Ligules membranous-ciliate, cm wide, flat, the base cordate, brous. /nflorescence terminal, lax, diffuse, pyra- midal, 2.5-9 cm long, 1.5-8.5 ст wide; main axis flexuous, hirsute, with long papillose-pilose hairs; branches alternate, flexuous, the axis of the branches smooth, glabrous; pedicels glabrous. Spikelets solitary, narrowly ellipsoid, 1.5-1.9 mm long, 0.5-0. m wide, greenish, short-hispid, the upper glume and lower lemma subequal. Lower glume ovate, 0.4-0.9 mm long, usually 14-14 or more the length of the spikelet, acute to truncate, 1-3(-7)-nerved. Upper glume 1-9(-11)-nerved, not embracing the lower lemma. Lower lemma glumiform, 7—9-nerved, acute, not inflated at the base. Lower palea lanceolate to truncate, 0.7-1 .2-0.3 mm wide, hyal lower dower absent. Upper anthecium ellipsoid, 1.2-1.7 mm long, 0.4—0.6 mm wide, smooth, shiny, short-apiculate, with simple papillae over its entire mm long, ine, glabrous; surface. Caryopsis 1.1 mm long, 0 widely distributed from Belize and Panama to Bolivia and Brazil; Distribution and ecology: common on forest edges, with the culms straggling in low vegetation, often in disturbed and open plac- es in sandy soils; from 100 to 1,400 m. Volume 80, Number 1 1993 Zuloaga et al. 173 Revision of Panicum subg. Dichanthelium sect. Dichanthelium Selected specimens cited. BELIZE. EL CAYO: ca. mi. S of Grano de Oro on road between Millionario ond La Flor, Croat 23391, 23398A (MO). BOLIVIA. LA PAZ Hacienda Canana, sobre el camino a Tipuani, Buchan 7115 (BAA, G, US), 7116 (MO); Hacienda Simaco sobre el camino a Tipuani, Buchtien 5319 (M, МО, US). BrazIL. ALAGOAS: Porto de Pedras, Fazenda Canada, Campelo 1444 (CTES). BAHIA: Bom Jesus, eq 135a (М); Sào Bento das Lages, Luetzelburg 133 (K, M); 12 km N of Cachoeira, 225 m, Chase 8082 (MO, US); Serra do Tombador, NW of Jacobina, on BR 234 highway to Ouro Branco, Harley et al. 16636 (CEPEC, K, US); Serra da Agua de Rega, ca. 27 km N of Seabra, road to Agua de Rega, 1,000 m, Irwin et al. 31023 (MO). CEARA: Barra do Cocó, Black 55-18336 (IAN); Praia Náutica de Fortaleza, Black 55- 18347 (ТАМ). ESPIRITO SANTO: T al. 2427 (RB, SI, US). MINAS GERAIS: Diamantina, Serra de San Antonio, Chase 10373 (US); ca. 1 km S of Sào Pedro do Suaqui along Hwy. MG-3, 520 m, Davidse et al. 11495 (МО); Juiz de Forá, Morro do us Chase 8571 (US); abandoned diamond mines, steep slo of Diamantina, /rwin et a N Irwin et al. 28151 (F, K, MO, pinhaco, ed 8 km E of Diamantina, Irwin et al. 27 666 (K, MO, UB, US). PARANA: Ман ombrowski E (K, US): Рука. Mun. Antonina, S I T 14 (K, US). PERNAMBUCO: Dois Irmaos, vicinity of Penn buco, Chase 7732 (MO, US); 5 Garanhuns, 300- Serra da Itatiaia, Chase 8246 (IA nck 229 (W). SAO PAULO: a bw Portovelo, S River, А mi. above Кену Falls, Cot ап : , US). FRENCH GUIANA: 0 mi de Granville e et al. add (CAY, P, US ). Pu A COCLE: al sawmill e El Cope, Mun 3822 (SI). VENEZUELA. BOLIVAR: m al N de La Ciudadela, Gran Sabana, 5?50'N, ndi. Zuloaga et al. 4475* (MO, SI, VEN); Salto El Dante, 35 km N of La Ciudadela on way down from La Gran Sabana, along highway to El Dorado, Davidse 4951 (MO, SI, VEN); entre Piedra de la Virgen y la parte alta de la Escalera, carretera a la Gran Sabana, 60%01'N, 61?25'W, Zuloaga et al. 4399 (MO, SI, VEN). FALCON: cerca de Maraparari, Lasser & Foldats 3005 EN). YARACUY: 7.5 km N of Salom, cloud forest, 10?15'N, 68°29'W, Liesner & Steyermark 12373 (МО); 5 km N of Salom, transition between savanna and forest, Davidse et al. 20674 (MO) Panicum sciurotoides is related to P. sciurotis Trin., from which it differs in having the lower glume Y3-14 or more the length of the spikelet and 1-3(-7)-nerved; the upper glume 7-9(-1 1)-nerved; the lower lemma 7-9-nerved and not inflated at base; and the upper anthecium inserted basally. Panicum viscidellum differs from Panicum sci- urotoides in its pseudoligule composed of long hairs. Anatomically, all specimens examined are ver similar in leaf anatomy (Fig. 14). The keel of this species is distinctive, = only a single vas- ДА). The asymmetrical, S-like nature of the lamina on either side of the keel is cular bundle (Fig. 1 of particular interest, as this structure does not occur in any other taxon of section Dichanthelium, yet is com subgenus Dichanthelium. mon in section Cordovensia, also of Large-s = form of P. sciurotoides: BRAZIL. RIO DE JANEIRO: between Alta Boa Vista and Silvestre, vicinity of Rio de Janeiro, 450-520 (F, NY, 05); camino Dos ciem J Alta Bóa Vista, vicinity of Rio de Janeiro, 100-200 m, 12 Feb. 1925, Chase 844 6 (US): Corcovado, Sep. 1920, Kuhlmann & Ducke 16233 (SI); Corcovado, vicinity of Rio de Janeiro, 700 m, 11 Jan. 1925, Chase 8169 (F, GH, MO, NY); Corcovado, 6 Mar. 1924, Bailey & Bai- ley 742 (US); Tijuca, 1,000 m, 28 Арг. 1930, Chase 12158 (US). These specimens, included provisionally in P. sciurotoides, have spikelets 1.9-2.2 mm long and panicles 3.5-8 cm long. They were collected in io de Janeiro, Brazil, in mountains of the Serra do Mar, scandent on the forest margins. 28. Panicum scoparium Lam., Encycl. 4: 744. 1798. Dichanthelium scoparium (Lam.) Gould, Brittonia 26: 60. 1974. TYPE: United States. South Carolina: without locality, Mi- chaux s.n. (holotype, P-MICH; photo and fragment, US 2808935). Figure 28. Plants perennial with a rosette of winter leaves. Culms ascending, slightly decumbent at the base, to erect, to 1.6 m tall, simple and then freely branching at the upper nodes; internodes cylindric, hollow, villous, covered with appressed whitish hairs, purplish to greenish, nodes densely pilose, covered with retrorse whitish hairs, with a glandular zone below. Sheaths striate, densely papillose-pilose and covered with conspicuous glands, the margins shortly pilose. Ligules membranous-ciliate, 1.5— 2 m long, membranous portion inconspicuous or to 0.5 mm long. Blades lanceolate, (2-)10-19 cm long, (0.3-)1.5-1.8 cm wide, flat, the base subcordate, the apex attenuate, hispid on both sur- faces, the margins shortly pilose, the midnerve conspicuous. /nflorescences terminal and axillary from the uppermost nodes, terminal inflorescence exserted, peduncle hispid, ca. 25 cm long; panicles lax, open, diffuse, 10—18 cm long; main axis flex- uous, hispid and covered with conspicuous glands, the branches alternate to opposite, divergent, the 174 Annals of the Missouri Botanical Garden axis of the branches hispid toward the base, oth- erwise glabrous, glandular; pedicels solitary, 1-15 mm long, glabrous and covered with glands; ter- minal panicles with chasmogamous flowers; axillary panicles ca. 4 cm long, similar to the terminal one. Spikelets solitary, obovoid, 2.5-2.8 mm long, 1.3— 1.6 mm wide, greenish to purplish. Lower glume 8(-1.5) mm long, 7-'Á or occasionally slightly longer than ! the length truncate to ovate, 0.4-0. of the spikelet, nerveless to 1 -пегуеа. Upper glume and lower lemma subequal or the upper glume shorter and not covering the upper anthecium, obtuse, 9-nerved. Lower palea lanceolate, hyaline, 1.3 mm long, 0.6 mm wide; lower flower absent. Upper anthecium ellipsoid, 2-2.2 mm long, 0.9- 1.4 mm wide, smooth, shining, pale, glabrous, pap- illose, shortly apiculate. Caryopsis obovoid, 1.5 mm long, mm wide, olivaceous and with dark spots; hilum punctiform, embryo less than 2 the length of the caryopsis. Distribution and ecology: southeastern United States, Mexico, and the West Indies, in marshes, moist pinewoods, and roadsides. Selected specimens cited. CUBA. ORIENTE: Chanascal de la Cueva, Sierra de Nipe, León 19856 (MO); crest of Sierra Nipe, Morton & Acuria 3082 (05); pine woods, iron mine, 5 of Guaro, Hitchcock 234 14 (US). DOMINICAN . Cordillera Central, near La Cumbre, Ekman 11432 (С, US). 1А УЕСА: vicinity of Constanza, 1,000- 1,600 E Allard 17 7485 (US); 16 km E of El Rio along hwy., 1,220 m, Davidse 2657 (MO); 18 km 5 of Jara- bacoa ions the highway, 1,160 m, Davidse 2651 (MO). MONTE CRISTI: Lagunas de Cenobi, Ekman 12943 (US). SAN JUAN: Piedra del Aguacate to Rio del Oro, Howard & Howard 9331 (US). saNTIAGO: declives del Pico de 3431 (US). Puerto Rico. Sierra de Naguabo, Rio Icaca Shafer 3481 (US). UNITED STATES. ARKANSAS: Miller County, collected near Texarkana, 4.4. & E. Geller 4236 (Р). DISTRICT OF COLUMBIA: Brookland, Holm s.n. (P). SOUTH CAROLINA: о moist sandy soil, Hitchcock S-n., r. Gr. Hb. 178 (P). VIRGINIA: Cape Henry, moist sand sre bank of dich: Chase s.n., Amer. Gr. Hb. 17 P). 29. Panicum sphaerocarpon Elliott, Sketch Bot. S. Carolina 1: 125. 1816. Dichantheli- um sphaerocarpon (Elliott) Gould, Brittonia 26: 60. 1974. Panicum dichotomum L. var. sphaerocarpon (Elliott) Alph. Wood, Class- Book Bot., ed. 1861: 786. 1861. microcarpon Мићјепђ. var. sphaerocarpon (Elliott) Vasey, Grass. U.S.: 12. 1883. ТУРЕ: United States. Georgia: without locality, Bald- win s.n. (holotype, CHARL and fragment, US 81121) Panicum not seen; photo Caespitose perennial, with or without a basal rosette of short, broad leaves. Culms decumbent and geniculate, branching at the base, to erect, branching at the upper nodes, 15-40(-60) cm tall; internodes short toward the base, compressed, hol- low, glabrous; nodes short-pilose or glabrous. Sheaths striate, greenish to purplish, shortly ciliate on the margins, glabrous in the rest of the surface. Ligules absent or to 0.8 mm long, short-membra- nous at base, then long-ciliate, or absent; collar brownish, glabrous. Blades lanceolate to ovate- lanceolate, 2.5-9 cm long, 0.5-1.5 cm wide, flat, rigid, glabrous, the base subcordate to cordate, the apex acuminate, the margins scabrous, short-ciliate to ciliate toward the base. Inflorescences terminal, exserted, peduncles 6-25 cm long, glabrous; pan- icles lax, diffuse to contracted, (2-)5-11 cm long, 3.5-7 cm wide; main axis flexuous, glabrous, glan- dular; first-order branches alternate, divergent or appressed, distant, the axis of the branches smooth, glabrous, glandular; pedicels long, greenish to pur- plish, glabrous, smooth, glandular; axillary panicles similar to the terminal one. Spikelets solitary, ob- ovoid to broadly ellipsoid, 1.1-1.8 mm long, 0.9- 1 mm wide, short-hirsute, the upper glume and lower lemma subequal. Lower glume broadly ovate, obtuse, 0.2-0.4(-0.9) mm long, У, the length of the spikelet or occasionally to 43-14 the length of the spikelet, nerveless or 1(-3)-nerved separated Upper glume 1.3-1.7 mm long, obtuse, shiny on the inner from the upper glume by a short stipe. surface, short-hirsute, 7-nerved, the nerves anas- tomosing toward the 1.2-1.7 anastomosing toward the apex. Lower palea ovate- ase. Lower lemma glumi- form, m long, 7-nerved, the nerves lanceolate, 0.6 mm long, 0.3 mm wide, hyaline, the margins sparsely pilose or glabrous, lower flow- er absent. Upper anthecium broadly ellipsoid, 1— 1.3 mm long, 0.6–0.9 mm wide, glabrous, smooth, shiny, indurate, papillose with simple papillae; cleis- togamous flowers present or absent. Caryopsis ob- ovoid, 0.7 mm long, 0.5 mm wide, brownish; hilum punctiform, embryo Уз the length of the caryopsis. Panicum sphaerocarpon is a species with a distinctive anatomy. The absence of differentiation between the spongy and palisade mesophyll sepa- rates this species from all others in the section. The chlorenchyma cells are also characteristic in that their cytoplasm is centrally concentrated and adaxial air spaces are present (Fig. 15B, D). In addition, the adaxial bulliform cells are in discrete fans consisting of only 3-4 cells. Кеу TO THE VARIETIES l. Spikelets 1.1-1.4 mm long l. Spikelets 1.4-1.8 mm long .. ar. floridanum . var. sphaerocarpon Volume 80, Number 1 1993 Zuloaga et al. Revision of Panicum subg. Dichanthelium sect. Dichanthelium 29a. Panicum sphaerocarpon var. sphaero- carpon Panicum inflatum Scribner & Smith, U.S.D.A. Div. Agrostol. Circ. 16: 5. 1899. Panicum sphaerocar- pon Elliott var. nd (Scribner & Smith) A. Hitchc., in je Hitchc. & Chase, Contr. U.S. Natl. Herb. 15: 253. 1910. туРЕ: United States. Missis- RUN oen County, Biloxi, Tracy 4622 (holo- Panicum vicarium Fourn., Mex. Pl. 2: 20. 1886. TYPE: Mexico. Veracruz: Córdoba, Schaffner 285 (holo- na P. Panicum auburne Ashe, North Carolina Agric. Exp. Sta. Bull. 175: 115. 1900. TYPE: United States. Ala- bama: Lee County, Auburn, 1898, Earle & Baker 1527 (isotype, NY; fragment, US 2383621) Distribution and ecology: United States, Mex- ico, Mesoamerica, and northern South America (Colombia and Venezuela). Grows in forest edges, gravelly banks, or in open sites, in partially dis- turbed areas, between 600 and 2,700 m ted specimens cited. BELIZE. BELIZE: 4 km of Zuloaga & Londoño 4249* (COL. SI). ANTIOQUIA: Rio Negro, 2,110 m, Archer 240 (US); Santa Elena, 1,500- А т, pens 1167 (US). OCENDINAMARE A: Subida a Alto а Трге, poys 4035* (COL, SI). Cosra RICA. 30: 3 k of Agua Caliente, 1,400 m, Pohl & Davidse 11377 US. HEREDIA: Volcán Barba, Cordillera Central, Pohl & Davidse 10535 (MO). SAN JOSE: ca. 2 km SE of Tarbaca on road to San Gabriel, 1,600 m, Pohl & Davidse 11046 (US). ECUADOR. TUNGURAHUA: Banos al Topo, Acosta-Solis 20832 (US). EL SALVADOR. CHALA- TENANGO: E slope of Los Esemiles, Tucker 1056 (P). SANTA INTIBUCA: Cerro San Cristóbal, la ш. Са 5 (МО). FRANCISCO MORAZAN: 2.5 km N of Zambrano, on road to San Francisco de Soroguara, Pohl & Davidse 12131 (MO). OCOTEPEQUE: common in pine forest and rocky hillsides of Machuca, Honduras and Guatemala border, Molina & Molina 27870 (MO). OLANCHO: Rio W, Campamento, Amador 89 (M О) no 98 (MO). MEXICO. CHIAPA 7 near Teopisca, Breedlove & Dune i (MO); Mun. La Trinitaria, 9 km S of Comitan on Mexican Hwy. 190, Breedlove & Davidse 54954 (MO). GUERRERO: 3.5 km al N de Zoyotepec, brecha maderera, Martine z-Salas & Torres 2562 (MO). JaLisco: E of San Se ey “Hacienda del Ototal, Mexia 1812 (MO). MICHOACAN: са. 10 mi. W of Ciudad Hidalgo, near km Post 160 on Hwy. 15 to Morelia, Davidse & Davidse 9842 (MO). VERACRUZ: ] k s.n., Amer. Gr. Hb. 155 (MO, P); . Pringle 8344 (MO, Hwy. 1 (at km 135.5 and ca. 10.6 km W of bridge at La Trinidad) to San Nicolás, Stevens 14802, 18000 (MO). PANAMA. CHIRIQUI: E slope of Volcán de Chiriqui (Bart), WNW of Bo үзеш, Davidse & D'Arcy 10153 (MO). VENEZUELA. ARAGUA: Distr. Ricaurte, ca. Colonia Tovar, a 600 m del desvio n Jarillo, via Colonia Tovar, 2,170 m, Zuloaga & Ortíz 4280* (MO, SI, VEN). DISTRITO FEDERAL: moun- red from ү нб with main road between El Junquito Pi Colonia Tovar, Davidse & Tillett 4079 (MO). 29b. Panicum sphaerocarpon var. florida- num Vasey, U.S.D.A. Div. Agrost. Bot. Bull. 8: 33. 1889. Panicum floridanum (Vasey) Chapman, Fl. South U.S., ed. 3: 585. 1897, non Trin., 1835. Panicum erectifolium Nash, Bull. Torrey Bot. Club 23: 148. 1896. Di- chanthelium erectifolium (Nash) Gould & Clark, Ann. Missouri Bot. Gard. 65: 1105. 1978. Dichanthelium sphaerocarpon (El- liott) Gould var. floridanum (Vasey) Davidse, Novon 2: 104. 1992. TYPE: United States. Florida: **Moist pine barrens, Mosquito Inlet,” May 1879, Curtiss 3599 (lectotype, US 81122; isolectotype, MO). Figure 29. Distribution and ecology: open white-sand sa- vannas and moist pine woods of southeastern Unit- ed States, Belize, and Cuba, 0-100 m e specimens examined. BELIZE. BELIZE: 12.5 mi of Belize along Northern Hwy., Croat 23259 P Wein 10716 (MO); savanna near Ridge Lagoon Plantation, 11 mi. from Belize on Northern Hwy., Gentry 7860 (MO). km S of Santa Birbiri, Blanco s.n. (US 2183813), Killip 44825 (US); along road to San Francisco de las Piedras, Killip 43723, 14193 (US) ORIENTE: Moa, sag d’aviation, Victorin et al. 21448 (US). PINAR DEL RIO Herradura, at a small laguna, аы 11599 (05); ta Grifa, in open places, Ekman 11242 (US); between Guane and Remates, near sea level, edge of Laguna de Cebo, Killip 32328a (US). SANTA CLARA: Motembo, León et al. сл Variety floridanum differs from variety sphae- rocarpon by its smaller spikelets 1-1.5 mm long, erect leaves that are appressed to the culms, gla- brous nodes, and ligules 0.3-0.4 mm long. 30. Panicum stigmosum Trin., Gram. Panic. 194. 1826. TYPE: Minas Gerais: “ad rip. rivuli Pibanha, in Serra dos Orgaos, 1. (holotype, LE not US Brazil. mis. mense Aprili Riedel” seen; photo of the type, K, fragment, 974621). Figures 16, 29. P. mollicomum Kunth, Revis. Gramin. 1: Suppl. 9. 1830. F k. 1: 306. > without locality, Haenke s.n. (PR, holotype not seen; frag- ment of the type, US 974627). 176 Annals of the Missouri Botanical Garden Perennial, (30-)50-100 cm tall. Culms decum- bent, branching and rooting at the lower nodes, then ascendent, leaning on vegetation, branching at the upper nodes; internodes 4-19 cm long, compressed, glabrous, with glands toward the distal portion; nodes compressed, brownish, pilose or gla- brous. Sheaths 3-11 cm long, shorter than the internodes, hispid or glabrous, glandular, one mar- gin ciliate, the other one ciliate toward the apex but otherwise glabrous. Ligules (0.3-)1-1.6 mm long, shortly membranous at the base, then ciliate; collar villous or puberulous, brownish. Blades lin- ear-lanceolate, 10-22 cm long, 1-2.2 cm wide, the base subcordate or cordate, slightly asymmet- rical, the apex attenuate, sparsely pilose or glabrous on the adaxial surface or hispid on both surfaces, the margins shortly ciliate, scabrous, the abaxial surface glandular or eglandular. /nflorescences ex- serted, peduncle (6-)9-34 cm long, cylindric, gla- brous or sparsely pilose, glandular; panicles lax, diffuse, 9-21 cm long, 6-15 cm wide; main axis flexuous, glabrous or sparsely pilose, glandular, the pulvini pilose, first-order branches divergent, al- ternate or subopposite, flexuous, the axis of the branches flexuous, scaberulous, glandular; pedicels claviform, shortly hirsute or scabrous. Spikelets ellipsoid or obovoid, (1.9-)2.4-3.2 mm long, (0.9-)1.1-1.3 mm wide, greenish to purplish, gla- brous, shortly stipitate, the upper glume and lower lemma subequal. Lower glume ovate, 1.1-1.8 mm long, 42-% the length of the spikelet, embracing the upper glume, (1-)3-nerved, glabrous. Upper glume (1.6-)2-2.4 mm long, not covering the apex of the upper anthecium, glabrous, 7—9-nerved, the nerves manifest. Lower lemma glumiform, (1.6-)2- 2.4 mm long, 7-9-nerved. Lower palea lanceolate, (1.6-)1.8-2.1 mm long, 0.6–0.7 mm wide, hya- line, glabrous, the margins ciliate; lower flower present or absent, male when present; stamens 3, anthers to 0.9 mm long. Upper anthecium ellip- soid, 1.9-2.1 mm long, 0.9-1 mm wide, acumi- nate, stramineous, indurate, with simple papillae all over its surface; lodicules ca. 0.3 mm long; stamens 3, the anthers 0.7 mm long. Caryopsis ellipsoid, 1.3 mm long, 1 mm wide; hilum punc- tiform, embryo !^ the length of the caryopsis. Distribution and ecology: Brazil, in the states of Espirito Santo, Minas Gerais, Rio de Janeiro, Santa Catarina, and Зао Paulo, on forest edges in tangled colonies, 1,000-2,200 m Additional specimens examined. BRAZIL. ESPIRITO SANTO: ‘Serr ra do Caparaó, Chase 10106, 10112 (US). Nac. Itatiaia, camino para las Agul- has Negras, Zuloaga et al. 2382 (RB, SI, US); Serra da Gramma, 1,100 m, Chase 9543 (US), 1,700 m, Chase 9556 (NY); Serra do Caparaó, 2,100-2,200 m, Chase 9662 (F, NY, US i Nes E of Ouro Preto, Chase 9381 (NY, US), 1 (US). RIO DE JANEIRO: Floresta Regime de EC. Permanente do IBDF, entre Vale dos Princesas y Rocio, 1,000- 2389 (SI); Macieiras, Serra de It | NY, US), 8322 (US), Zuloaga et al. 2351“, 2358" (SD, cieiras, Serra de bow 1,200-1,800 m, Chase 8322 ; Ri eiro, Glaziou 18629 (US, W). SANTA m S of Campo Alegre, on the road to Jaragua do Sul, ‘Smith & Klein 7334 (US). SAO PAULO: Salesópolis, Estagao Biologica de Вогасела, Mattos 14248 (P, US). Without state and locality, Riedel y G Panicum stigmosum is closely related and per- haps conspecific with P. sabulorum, differing only in: the plants being bigger, scandent; foliar dimor- phism is absent; the inflorescence is bigger ((6-)9- 34 cm long); and leaf blades and axis of the inflo- rescence are glandular. There are several inter- mediate specimens between both species, however, 351, 2382, Glaziou 9381, such as Zuloaga et al. 23. 17946, 18629, Chase 9361, Mattos & Mattos 14248. Macrohairs are lacking in the two specimens examined in the anatomical study (Fig. 16D, E). The mesophyll is characteristically very diffuse with large intercellular air spaces. 9421, and Panicum stigmosum var. parviflorum Doell is a dubious name. Doell included in his original di- agnosis two specimens: Sellow 4906 and Regnell 11-1360. There is a fragment of the latter speci- men at US which is P. pantrichum Наске! (sect. Cordovensia). Small-spikelet form of P. stigmosum: BRAZIL. ESPIRITO SANTO: Serra do Caparaó, Chase 10105 (MO, US). MINAS GERAIS: Serra da Gramma, E of Araponga, 1,200 m Chase 9531 (F, GH, MO, NY, US). RIO DE JANEIRO: Вега de Itatiaia, Macieiras, 1,700-1,800 m, Chase 832 1 (GH, US). , These specimens have spikelets 1.9-2.1 mm long, but are otherwise similar in all characters to other material P. stigmosum. The rachilla is pro- longed as a short mucro above the upper anthecium in Chase 10105. 31. Panicum stipiflorum Renvoize, Kew Bull. 37: 329. 1982. TYPE: Brazil. Bahia: 16 km 950-1,000 m, 8 Mar. 1 3 voize, Erskine, Brighton & Pinheiro 17019 (holotype, CEPEC; isotypes, K, NY, US 2955108). Figure 29. Volume 80, Number 1 1993 Zuloaga et al. 177 Revision of Panicum subg. Dichanthelium sect. Dichanthelium Annual? Culms decumbent at the base, then ascendent, 20-30(-80) cm tall, freely branching at the upper nodes, many-noded, internodes terete, hollow, short-pilose, nodes compressed, densely pi- lose with short, appressed, whitish hairs. Sheaths striate, sparsely or densely hirsute with short hairs, or long-pilose in the upper portion, the margins long-ciliate, more so in the upper portion. Ligules 0.3 mm long, membranous-ciliate. Blades ovate- lanceolate, 2-6 cm long, 0.6-1.5 cm wide, cor- date, amplexicaulous, sparsely to densely hirsute with short, rigid hairs, or glabrescent, the margins cartilaginous, scabrous and long-ciliate with papil- lose-pilose hairs. Inflorescences terminal, exserted, peduncles to 20 cm long, cylindric, shortly hirsute; panicles lax, diffuse, 3-8 cm long, 2-5 cm wide; main axis flexuous, hirsute; pulvini densely pilose; first-order branches alternate, divergent, the axis of the branches hirsute, smooth, eglandular; ped- icels hirsute, 0.6-2 mm long. Spikelets paired ог solitary, ellipsoid, 2.2-2.6 mm long, 0.9 mm wide, sparsely hirsute with short hairs on the glumes and lower lemma, greenish or tinged with purple. Lower lume acute, 44 or less the length of the spikelet, occasionally larger, not embracing the upper glume at its base, 3–5-пегуед, separated from the upper glume by a conspicuous stipe. Upper glume 1.8- 2.1 mm long, occasionally a little smaller and not covering the apex of the upper anthecium, 7-9- nerved, the nerves anastomosing. Lower lemma glumiform, 7-nerved, the nerves anastomosing to- ward the apex. Lower palea elliptical, 1.5 mm long, 0.3 mm wide, hyaline, glabrous; lower flower absent. Upper anthecium ellipsoid, 1.7-1.8 mm long, 0.7 mm wide, indurate, smooth, shiny, papil- lose with simple papillae regularly distributed all shortly apiculate and pilose toward the apex. Caryopsis over its entire surface; lemma 5-nerved, ellipsoid, 1.5 mm long, 0.9 mm wide; hilum punc- tiform, embryo Уз the length of the caryopsis. Distribution and ecology: Bahia, Brazil, in ce- rrados in rocky, sandy soils, between 900 and 1,100 m, in flower between February and April. Additional specimens examined. BRAZIL. BAHIA: 34 >й Doudo, са. 21 km of morr a 17004 (paratypes, K, NY, US); Fazenda Mundo Novo, Botafogo, Zehntner 248, 263 3 (RB) Serra do Tombador, ca. 18 km E of Morro de Chapéu, Irwin et al. 30702 (MO, UB); Serra Tinguá, рие 2017 (М); Sentacé, Zehntner 2048 (M). Panicum stipiflorum is characterized by its cor- date, amplexicaulous leaves with conspicuous cil- iate margins, spikelets with a stipe between the lower and upper glume, and lower glume usually less than У the length of the spikelet. 32. Panicum strigosum Muhlenb. in Elliott, Sketch Bot. S. Carolina 1: 126. 1816. Di- chanthelium strigosum (Muhlenb.) Freck- . 1981. TYPE: United Elliott Herb. s.n. (lectotype, CHARL not seen; isolectotype, US; photo and fragment, US 2808953). Figure 29. Panicum ciliatum Elliott, Sketch Bot. S. Carolina 1: 126. , non Panicum gaps Maercklin, 1792. Panicum dichotomum L. glabrescens Griseb., Fl. Brit. W.I. 553. 1864. ЕЭ петина leucoblepha- ris (Trin.) Gould & Clark var. glabrescens (Griseb.) Gould & Clark, Ann. Missouri Bot. Gard. 65 mann var. кын (Griseb.) Freckmann, Brit- 1. TYPE: Jamaica. Without lo- Purda. s.n. (holotype, K) Lam. var. pubescens Vasey, Con 65: 1101. 1978. Panicum ciliatum Elliott var. pu- bescens (Vasey) Freckmann ex R. W. Pohl, Field- iana, Bot., n.s. 4: 356. 1980. Panicum leucobleph- aris Trin. var. pubescens (Vasey) Beetle, Phytologia 48: 192. 1981. TYPE: United States. Florida: Duval County, without locality, Curtiss s.n. (lectotype, US). P polyenulon 2 Bull. Torrey Bot. Club 24: 200. Side : United States. Florida: Hillaborougk mpa, 20 Aug. 1895, Nash 2420a (ћо- bb e, "NY; isotype, US). F: е Scribner, Tennessee Agric. Exp. iie .B 1894. TYPE: United States. Ten- essee: White Cliff Pd July 1890, Scribner s.n. (lectotype, US 742594 Short-rhizomatous perennials, caespitose and with many fasciculate culms. Culms erect, 3-20 cm tall, few-noded, unbranched below the inflores- cence; internodes short, cylindric, hollow, short- pilose or glabrous; nodes pilose, compressed. Sheaths 1 cm long, longer than the internodes, striate, glabrous, one margin ciliate, the other membranous. Ligules 0.2-0.4 mm long, membra- nous-ciliate; collar glabrous. Blades lanceolate, 1— 5(-7) em long, 0.3-0.5(-0.7) cm wide, flat, gla- brous or with long sparse hairs, the base narrowed or subcordate, the apex acute, the margins long- ciliate with rigid papillose-pilose hairs or occasion- ally glabrous. /nflorescences terminal, exserted, peduncle 1-14 cm long, flexuous, glabrous or sometimes hirsute near the junction with the pan- icle; panicles lax, diffuse, few-flowered, 2-5 cm 178 Annals of the Missouri Botanical Garden long, 1.5-4.5 cm wide; main axis flexuous, dense- ly hirsute with long whitish hairs, occasionally gla- brous, glandular, the pulvini pilose; first-order branches alternate, the axis of the branches hirsute, flexuous and glandular; pedicels glabrous, clavi- form, glandular; similar to the terminal one. Spikelets solitary, ob- 0.6-0.9 mm wide, puberulous or glabrous, greenish to purplish, the upper glume and lower lemma subequal. Lower glume ovate, 0.5-0.8 mm long, /-% the length of the spikelet, nerveless or 1-nerved, embracing axillary inflorescences present, ovoid or ellipsoid, 1–1.7 mm long, the upper glume at its base, acute or truncate, with a small internode between the lower and upper glume. Upper glume obtuse, 7-nerved. Lower lem- та 7-nerved, glumiform, obtuse. Lower palea 0.3- 0.7 mm long, 0.1-0.2 mm wide, hyaline, glabrous; lower flower absent. Upper anthecium ovoid, 1- 1.2 mm long, 0.6-0.9 mm wide, biconvex, whitish, indurate, smooth, glabrous, papillose; lemma short- ly apiculate; stamens 3, the anthers 0.5 mm long. Caryopsis unknow Distribution си ecology: States, Mexico, Mesoamerica, and the West Indies, in South America in Colombia and Venezuela. Com- monly found in wet meadows or swampy areas and southern United savannas, between sea level and 2,100 m Selected specimens cited. BELIZE. BELIZE: 4 km W of Hattieville along the western hwy. to Belmopan, 80 m, Davidse & Brant 32982 (MO, SI). EL cAvo: Mountain Pine Bu along Coona Cairn Road, Davidse & Brant 33037 А (MO). ORANGE WALK: savanna ca. 5 km N of August Pine Ridge on the road to Trinidad, Davidse & idge, Mullins Branch, Monkey River, Gentle 3748 (MO). COLOMBIA. CAUCA: Cuatro Esquinas, 1,700-1, & Killip 6362 (NY, US); Munchique, 2,1 Alston 8205 (COL, US). MAGDALENA: Santa ба, Smith 163 (MO, NY, P, US). Costa Rica. GUANACASTE: región del Volcán Cacao, Cerro Pelado, Chacon 2366 (МО). Cuba. Nueva Gerona, Palmer & Riley 990 (05); botweati San Francisco de Las Piedras and Cerro La Canada, moist soil along stream, Killip 44614 (US). ORIENTE: Sierra de Ni nas at foot of Loma Men PINAR DEL RIO: NW of Pinar del Rio, Hitchcock 23299 (05); Мапша, Ekman 11088 (G); Sierra де los Organos, Ekman 11068 (С, 05); Маи. de Santa Rosa, 1 Allard 15870 (US). GUATEMALA. ALTA VE (US). Haiti. Massif du Nord, Marmelade, Ekman 8206 (G, US). HONDURAS. GRACIAS A DIOS: Tuas, campamento maderero al W de Brus Laguna, Velson & Hernández 1005 (MO). OLANCHO: Маша del Rio de la población de Culmi, Nelson & Romero 4708 (MO). Jamaica. Halliss Savanna, Upper Clarendon, Harris 12225 (MO, US); James Hills, Savanna, Harris 12852 (MO, US); Bull Head Mountain and vicinity, Hitchcock 9550, 9551 (US). Mexico. CHIAPAS: Municipio of Palenque, 8-12 km N of Palenque on road to Catazaja, Breedlove 55356 (MO). NICARAGUA. ZELAYA: ca. 3 km S of Bilwaskarma, Pohl & Davidse 12262 (MO); along Rio Tuapi from bridge (on road between Puerto Cabezas and Tuapi) to coast, Stevens 7816 (MO). PANAMA. BOCAS DEL TORO: Santa Catalina, Blackwell et al. 2696 (MO). состе: Caribbean side of divide at El Соре, Hamilton & Davidse 2634 (MO). HERRERA: 10 km W of Las Minas on road to El Toro, roadside and cafetal, Sytsma & D'Arcy 3221 (MO). PANAMA: Distrito Capira, Cerro Campana, van der Werff & Herrera 6167 (MO); hills NE of Hacienda La Joya, Dodge et al. 16884 (MO). VERAGUAS: along beach, Con- cepción, Hammel 5237 (MO). PuERTO Ric 0. Monte Mesa, vicinity of Mayagüez, Chase 6276 (US); Campo Alegre, between Manati and Laguna del Tortuguero, Chase 6621 (US). 33. Panicum superatum Hackel, Oesterr. Bot. Z. 51: 427. 1901. TYPE: Brazil. Rio de Ja- neiro: Serra dos Orgaos, 7 Mar. 1889, Gla- ziou 17904 (holotype, W; isotype, P, frag- ment, P, fragments US 1280065, 1715312). Figures 17, 31, Caespitose, short-rhizomatous perennial. Culms erect to ascendent, -)0 m tall, branch- ing at the base or Dum the upper nodes; internodes 4-6 cm long, hispid or glabrous, terete, hollow; nodes compressed, long-pilose, or glabrous. Sheaths 4-10 cm long, striate, glabrous or hirsute with long papillose-pilose hairs, one margin long-ciliate, the other membranous. Ligules membranous-cili- ate, 0.6-1.5 mm long; collar glabrous or densely pilose. Blades lanceolate to linear-lanceolate, 4.5- 16 cm long, 0.6-1.5(-2.5) cm wide, hispid or glabrous, densely pilose on the adaxial surface near the subcordate or cordate base, the apex attenuate, the margins стаје. /nflorescence terminal, ex- serted, peduncle hispid; panicles contracted to slightly open, 5-8(-17) ст long, 0.8-5(-10) cm wide; main axis hispid, the pulvini pilose; first- order branches appressed or slightly divergent from the main axis, alternate or subopposite, the axis of the branches hispid or short-pubescent and eglan- dular; pedicels flexuous, scabrous, hispid with long hairs. Spikelets congested, obovoid, (2.1-)2.4-3.3 mm long, 0.9-1.2 mm wide, sparsely pilose or glabrous, the upper glume and lower lemma sub- equal, shortly stipitate. Lower glume ovate-lan- ceolate, 1.6-2.4 mm long, 4—% the length of the spikelet, clasping, 3-5-nerved, with a short stipe between ha: lower and upper glume. Upper glume 1.8-)2. 7-)9-11-nerved, covering PE apex of the upper anthecium. Lower lemma (1.9—)2.2-2.5 mm long, 7-9-nerved. Low- not “~ 2.5 mm long, ( Volume 80, Number 1 Zuloaga et al. 1993 Revision of Panicum subg. Dichanthelium sect. Dichanthelium AS T Ф 2 р. Ж» p 2 => VE E АЕ ER c e c QW Sas FIGURE 35. Panicum superatum. — A. Habit. — B. Spikelet, lower E view. — C. Spikelet, upper glume view. — D. Spikelet, lateral view. — E. Lower palea. — F. Upper anthecium, lemma v H. Detail of apex of the lemma. — І. Caryopsis, embryo view.— J. Cade hilum view. (AI based on | Chase 9350.) er anthecium, palea view 180 Annals of the Missouri Botanical Garden FIGURE 36. A-C. Panicum penicillatum. — А. Habit. — B. Spikelet, lower glume view. — С. Spikelet, upper p view. D, E. Panicum caparaoense. —D. Sp n lower glume view. — E. Spikelet, upper glume view. F, G. Panicum itatiaiae. —F. Spikelet, lower glume view. n ikelet, upper glume view. (A-C, based on Alvarenga & Oliveira 5; D, E, based on Hatschbach & emn: 4. 5170. F, G, based on Chase 8327.) Volume 80, Number 1 1993 Zuloaga et al. 181 Revision of Panicum subg. Dichanthelium sect. Dichanthelium er palea lanceolate, (1.5-)2.1 mm long, 0.6 mm wide, membranous, the margins ciliate; lower flow- er absent or present, male when present, with 3 anthers each 1 mm long. Upper anthecium ellip- soid, 1.9-2.5 mm long, 0.9-1.2 mm wide, indu- rate, smooth, the apex of the lemma T scaberulous; ap ^ the anthers 0. long. Caryop .7 mm long, 1 mm en hilum ie pies уз the length of the caryopsis. Distribution and ecology: Brazil, on mountain slopes from Espirito Santo to Rio Grande do Sul, 900-2,650 m Additional specimens examined. BRAZIL. ESPIRITO SANTO: Serra do a rocky open campo, 2, m, Mexía 4014 (NY, P, 05); Serra do Сарагао, 2,280- 2,400 m, Chase 10084 (IAN, RB, US). MINAS GERAIS: Barbacena, Serra Mantiqueira, Chase 8667 (F, US); Ouro Preto, Villa Rica, 1,100 m, Chase 9350 (F, GH, MO, Y, 05); cerrado on middle slopes of Pico de Itacolumi, ca. 3 km S of Ouro Preto, 1,750 m, Irwin et al. 29483 (МО, UB); Mun. Itamonte, Parque Nacional de Itatiaia, camino para las Agulhas Negras, 1,550-1,800 m, Zu loaga et al. 2374* (MO, RB, SI, US). PARANA: 4 km E of Guarapuava along highway BR-277 to Curitiba, 1,050 , Davidse et al. 11319 (МО); Bocaiuva do Sul, Clayton SUL: Cambara do Sul-Itaimbezinho-Bela Vista, Valls et al. 1870, 2903 (CEN); Serra da Rocinha, prope Bom Jesus, in dumetosis, Rambo 5384 1 (US). SANTA CATARINA: Mun. W of Cacador, Smith & Klein 10899 ardim da Serra, Alto, 20 km S of 5 km 5 of Ponte Alta along highway BR-116 to Lajes, Davidse et al. 11104 (MO, SP); Ponte Serrado, 94 k W of Joaçaba, 700-900 m, Smith & Klein 14008 (НВ, K, 51; Mun. Urubici, 19 km N of Perico, Smith & Klein 15889 (HB, US); Mun. Lajes, Serra do Ilhéos, Smith & Klein 15456 (К, US). s&o PAULO: Jardin Botánico e Parque do Estado, Sendulsky 1063 (51; Campos do Jordao, Serra Mantiqueira, sandy campo, 1,600 m, Chase 9822 (NY, US). Related to Panicum sabulorum and P. stig- mosum, P. superatum can be distinguished by its contracted panicles and appressed spikelets. here are two indurate florets in Chase 9403. Sendulsky 1063 has anthers only 0.3 mm long. 34. Panicum surrectum Chase ex Zuloaga & Morrone, Novon 1: 111. 1991. туРЕ: Brazil. Minas Gerais: Barbacena, long and tangled in moist brushy base of higher slope, Chase 8664 (holotype, US; isotypes, F, NY). Figures 24D, 30 Short-rhizomatous perennial. Culms decumbent to geniculate, then erect, branching at the upper nodes, scandent, 45-120 cm tall; internodes com- pressed or cylindric, 7-13 cm long, glabrous; nodes compressed, glabrous, brownish. Sheaths 4— cm long, shorter than the internodes, glabrous to papillose-pilose toward the base, shiny, one margin ciliate, the other one ciliate toward the base, oth- erwise glabrous. Ligules 0.2 mm long, membra- nous-ciliate; collar pilose. Blades linear- Tessa 5-12 cm long, 0.6-1 cm wide, glabrous to short- hispid, attenuate at the base and apex, the margins scabrous, long-ciliate toward the base; midnerve conspicuous. /nflorescences terminal, exserted; panicles lax, diffuse, 3.5-15 cm long, 3-10 cm ide; main axis glandular or eglandular, flexuous, glabrous, the pulvini glabrous; first-order branches ascendent, whorled toward the base, then subop- posite or alternate, the axis of the branches gla- brous, flexuous, glandular or eglandular; pedicels triquetrous, glabrous, glandular or eglandular. Spikelets narrowly ellipsoid, 1.8-2.2 mm long, 0.8 mm wide, glabrous, greenish, nonstipitate, the up- per glume and lower lemma subequal, the nerves manifest. Lower glume ovate, 0.9-1.3 mm long, LY) the length of the spikelet, not embracing the upper glume, (1—)3-nerved, the lateral nerves inconspicuous. Upper glume 1.6-2 mm long, not covering the apex of the upper anthecium, 9-nerved. Lower lemma glumiform, 1.8-2.1 mm long, 9.nerved. Lower palea lanceolate, 1.5-1.8 mm long, 0.4 mm wide, shortly pilose near the apex, otherwise glabrous, hyaline; lower flower male or sterile, stamens 3, the anthers 1 mm long. Upper anthecium ellipsoid, 1.6-1.8 mm long, 0.7 mm wide, pale, indurate, papillose, the apex of the lemma shortly crested and pilose; stamens 3, the anthers 1.2 mm long, purplish. Caryopsis ovoid, 1.3 mm long, 0.7 mm wide; hilum punctiform, embryo less than ?4 the length of the caryopsis. Distribution and ecology: Brazil, occasionally present in Paraguay, found at forest edges, between 600 and 1,800 m Additional specimens examined. BRAZIL. DISTRITO FEDERAL: ca. m E of Brasilia, /rwin & Soderstrom 5419 (NY, UB, US). colas: Luziania, 15 km ao 5 da km N NE of the intersection of Hwys. 381 a 12 km SW of São Gonçalo do Sapucai, 870 1 m, Davidse & Ramamoorthy 10602 (MO, SI); Itacolumy, E of Ouro Preto, 1,400 m, Chase 9380 (NY, US), 9407 (US); Juiz de Fora, Krieger SVD 1227 (RB, US); Pogos de Caldas, 1,100-1,500 m, Chase 10677 (NY, SP, US); Serra da Gramma, 1,700 m, slope above jungle, Chase 9555 (F, GH, NY, US). PARANA: Mun. Gral. Carneiro, Rio Lageado, Hatschbach et al. е (К); Trés Barras, Dusén 17531 (F, С, GH, NY, US). RIO DE JANEIRO: below Macieiras, Serra de Itatiaia, 1,700-1,800 m, Chase 8324 (GH, NY, 182 Annals of the Missouri Botanical Garden US). RIO GRANDE DO SUL: Taimbezinho, pántano, Boechat s.n. (ICN 41067). SANTA CATARINA: Rio Capinzal, Dusén 17876 (US); Santa Cecilia, 1,100 m, Reitz & Klein ени ys SAO PAULO: Serra ded Campos do Jord 1,580 m, open sandy campo, Chase 9815 (F, NY, US) 9839(NY, US), ГАИ 1737 (US), Leite 3510 (GH). PARAGUAY. CAAGUAZU: Caaguazú, Balansa 7 (G) Panicum surrectum shows affinities with P. sa- bulorum, P. stigmosum, and P. superatum. Pan- icum surrectum is characterized by its long-ellip- soid spikelets, embracing the upper glume. Panicum stigmosum, and smaller lower glume, not P. sabulorum, and P. superatum have a lower glume 14-34 the length of the spikelet, covering the base of the upper glume, and ellipsoid to ob- ovoid, globose spikelets. Additionally, in P. supera- tum the panicles are contracted, with the spikelets congested on the branches, and P. stigmosum has bigger spikelets, (2-)2.4-3.2 mm long, and leaves 10-22 cm long. Panicum surrectum has, as other species of subgenus Dichanthelium (Morrone & Zuloaga, 1991), conspicuous glands on the axis and pedicels of some specimens Cleistogamous ног were observed in Chase 9815, with anthers 0.6 mm long. 35. Panicum telmatum Swallen, Phytologia | 4: 81. 1966. ТҮРЕ: Frontier between Roraima, Brazil and Bolivar, Venezuela, Serra do Sol, 28 Dec. 1954, 2,100 m, Maguire & Maguire 40400 (holotype, US 2182190; isotype, NY). Figure 31. Culms elongated, arching, hollow, densely nodes pilose. Sheaths Tufted perennial. branching, internodes cylindric, hispid with appressed hairs; papillose-pilose to short-pilose, the margins long- ciliate. Ligules membranous-ciliate, membranous portion 0.2 mm long, cilia 0.8-1 mm long, collar pilose with whitish hairs. Blades lanceolate, 3.5— 5.5 cm long, 4-9 mm wide, flat, the base rounded, the apex acuminate, the adaxial surface sparingly hispid and with long, whitish hairs near the ligule, the abaxial surface scabrous, the margins long- ciliate. Inflorescence with base included in upper- most sheath or short-exserted, 3-5 cm long, py- ramidal, few-branched, the branches divergent from the axis; main axis, the branches and pedicels hirsute with whitish hairs. Spikelets solitary, ellip- soid, 3-3.2 mm long, 1.2-1.4 mm wide, minutely appressed-puberulent; lower glume separated from the upper glume by a short internode. Lower glume 2-2.3 mm long, broadly ovate, blunt, 3—5-nerved, about ?4 the length of the spikelet, the nerves anastomosing. Upper glume and lower lemma as long as spikelet, broadly rounded on back, 7-9- nerved, the nerves anastomosing toward the apex. Lower palea ovate, 2 mm long, 0.6-0.8 mm wide, ca. % as long as the lemma, hyaline, glabrous, the margins short-ciliate, the lower flower absent. per anthecium ellipsoid, 2.3-2.5 mm long, 1.1 mm wide, nearly as long as spikelet, glabrous, short-pilose at the apex of the lemma; lemma 7-nerved, with simple papillae over its entire sur- face. Caryopsis unknown. Distribution and ecology: endemic to Uei-tepui (Cerro del Sol), Venezuela, about 25 km SE of Mt. Roraima, where growing in "burnt part of boggy central area." ede specimens examined. VENEZUELA. BOLI- : Distr. Roscio, altiplanicie que rodea la cumbre del Uei. -tepui (Cerro del Sol), sector occidental por encima del valle del Rio Arabopo, Huber 10008 (МО, MYF, SI). Panicum telmatum is related to P. sabulorum; the latter species has smaller, biconvex obovoid spikelets 1.9-3 mm long and is known from central and southern Brazil, Paraguay, Uruguay, Bolivia, Chile, and Argentina 36. Panicum umbonulatum Swallen, Contr. U.S. Natl. Herb. 29: 420. 1950. Dichan- thelium umbonulatum (Swallen) Davidse, No- von 2: 105. 1992. TYPE: Guatemala. Zacapa: upper slopes of Sierra de Las Minas, along Rio Repollal, 2,100-2,400 m, 12-13 Jan. 1942, Steyermark 42469 (holotype, US 1935002; isotypes, F, NY). Figure 31. Jalapa, 1,700 m, 2 Dec. 193 (holotype, F 1040005; fragment and photo, US 2236414) Panicum ramiparum Swallen, Contr. U.S. Natl. Herb. 23. 1950. TYPE: Guatemala. Quezaltenango: collected in mountains near Santa Maria, S of Que- zaltenango, 25 Маг. 1932, Weatherwax 160 (ћо- lotype, US 1914983). Perennial without a basal rosette of broader leaves. Culms decumbent, rooting and branching at the lower nodes and trailing, to erect, 40-100 cm tall, internodes terete, hispid, greenish to pur- plish; nodes compressed, purplish, densely papil- lose-pilose with whitish retrorse hairs. Sheaths stri- ate, papillose-pilose with deciduous tuberculate hairs, the margins ciliate. Ligules membranous- ciliate, 0.6-1 mm long; pseudoligule present, with long hairs at the adaxial surface of the blade; collar Volume 80, Number 1 1993 Zuloaga et al. 183 Revision of Panicum subg. Dichanthelium sect. Dichanthelium densely pilose. Blades lanceolate, 5-12 cm long, 0.4-1.4 cm wide, flat, the base cordate to sub- cordate, the apex acuminate, short-hispid on the adaxial surface or glabrous and long-pilose toward the base, the lower margins ciliate, otherwise gla- brous. Inflorescences terminal, exserted, peduncle terete, ca. 20 cm long, hispid, densely pilose toward the distal portion near the junction with the panicle; panicles lax, diffuse, multiflowered, 4—12 cm long, 3.5-10 cm wide; main axis flexuous, hispid, the pulvini pilose; first-order branches alternate, di- vergent, the axis of the branches triquetrous, greenish to purplish, scabrous and sparsely hispid, spikelets solitary on second- or third-order branch- es; pedicels triquetrous, scabrous. Spikelets ellip- soid, (2.2-)2.4- ong, 1-1.1 mm wide greenish to purplish, plano-convex, sparsely hispid or glabrous, the upper glume and lower lemma subequal, as long as or shorter than the upper anthecium. Lower glume 1.2-1.4 mm long, Y5- V4 the length of the spikelet, obtuse, 1-nerved, not embracing the upper glume. Upper glume 7-9- nerved. Lower lemma glumiform, 7-nerved. Lower palea lanceolate, 1.5 mm long, 0.4 mm wide, hyaline, glabrous; lower flower absent. Upper an- thecium narrowly ellipsoid, 2.4-2.5 mm long, 0.9— 1.1 mm wide, whitish, smooth, shiny, indurate, papillose, apiculate and short-pilose and with a green spot on the apex of the lemma. Caryopsis ellipsoid, 1.6 mm long, 1 mm wide; hilum punc- tiform, embryo less than / the length of the cary- opsis. Distribution and ecology: Mexico to El Sal- vador, and Ecuador, on forest edges, in moist plac- es, between 550 and 2,400 m Additional specimens examined. ECUADOR. LOJA: Argelia, 2,300 m, Espinosa 547 (US). EL SALVADOR. SANTA ANA: Hacienda San Jose, Rohweder 2111 (MO). . 14 mi. be- Tactic, Croat le (MO). BAJA VERAPAZ: along highway ca. 14 to Cobán, . S of turnoff to Salama, Croat 41154 (MO). na tul m e-oak forest in canyon of Río Chixoy near Malacatancito about 20 km SW of Huehuetenango, 1.600 m, Williams et al. 22170 (US), 22171 (F). QUEZALTENANGO: Chiquival, 2,360 m, de Koninck 117 (US). Honpuras. 17 km of Nueva Ocotepeque, Harmon & Dwyer 377 1 (MO). COMAYAGUA: E de Comayagua, Hernández 198 (MO). EL PARAISO: 31 km S of El Zamorano, on road between Guinope and Mandasta, Pohl & Davidse 12152 (MO); edge of bog in mixed oak-pine forest near Piedra iibi: 1,500 m, Williams 15983 (US). FRANCISCO MORAZA Universidad, Campo abierto, Lainez 91 (MO; W rade s Flores toward of San анай, Pohl & Davidse 11984 (MO); Jutiapa, ca. 7 km NE of Tegucigalpa, Pohl & Davidse i 7 МО); La Tigra, 35 km МЕ de Tegucigalpa, Ordoriez 42 (MO). OCOTEPEQUE: alrededores de Belén Gualcho, Nelson et al. 4026 (MO). Mexico. CHIAPAS: Mun. Siltepec, on the ridge above Siltepec on the road to Huixtla, Breedlove & Almeda 58273 (MO). ОАХАСА: 56 km de Tlaxiaco rumbo a Putla, Beetle M-4807 (MO) This species is related to P. viscidellum, differ- ing in its larger spikelets (2.2-)2.4-2.8 mm long. It is also related to P. divergens; the latter species lacks a conspicuous pseudoligule and usually has sparsely hispid spikelets. 37. Panicum viscidellum Scribner, U.S.D.A. Div. Agrost. Circ. 19: 2. 1900. Dichanthe- lium viscidellum (Scribner) Gould, Brittonia 32(3) 357. 1980. TYPE: Mexico. Veracruz: near Jalapa, 1,250 m, 1899, Pringle 8089 (lectotype, US 354526; isolectotypes, M, MO, P). Figures 18, 19, 20, 30. Panicum tremulosum m. Notizbl. Bot. Gart. Berlin- ahlem 7: 76. 1917. TvPE: Venezuela. Distrito Fed- eral: prope сак Moritz s.n. (holotype, В; frag- ment, ). Panicum PU p Swallen, J. Wash. Acad. Sci. 21: 15.1 . TY olombia. Antioquia: Quebrada del Alto, in mountains above ver 17 June 1930, Ar- chet 160 (holotype, US 144 Panicum blakei Swallen, а 422 1950. rvPE: Guatemala. Izabal: along trail om Los Amates to Izabal, Cerro de Las Minas, bon! юч, Blake 7817 (holotype, US 1163068) . Natl. Herb. 2 mala. H een Quetzal and Barillas, Sierra de los 0 m, 18 ШУ 1942, Steyer- mark 49123 (holotype, US 1935043). A Short-rhizomatous perennial. С rooting and branching at the lower nodes, den erect, branching at As upper nodes, 15-100 cm tall; internodes 3.5-18 cm long, hollow, striate, compressed, dien with long, rigid hairs; nodes compressed, densely pilose with whitish, retrorse hairs. Sheaths 2.5-13.5 cm long, papillose-pilose with tuberculate, caducous hairs, the margins cil- iate. Ligules short-membranous at the base, then long-ciliate, 0.3-1(-2) mm long, with a conspic- uous pseudoligule with hairs 2-3 mm long; collar pilose. Blades ovate-lanceolate to lanceolate, 3— 14 cm long, 0.4-2 cm wide, glabrous or short- pilose on both surfaces, the base cordate and clasp- ing, the apex acute, the margins long-ciliate and scabrous. Inflorescences terminal, exserted or par- tially included within the upper leaves, peduncle flexuous, 10-23 cm long, glabrous or sparsely or densely hispid; panicles diffuse, 3-12 cm long, 2- 184 Annals of the Missouri Botanical Garden 10 cm wide; main axis glabrous or densely hirsute; pulvini pilose and scaberulous, first-order branches alternate to subopposite, ascendent and divergent, the axis of the branches glabrous or hispid, eglan- dular. Spikelets solitary, ellipsoid, (1.5-)1.8-2(- 2.2) mm long, 0.7 mm wide, sparsely pilose, oc- casionally glabrous, the upper glume and lower lemma subequal, the stipe absent. Lower glume ovate, 0.6-1 mm long, 4%—' the length of the spikelet, sparsely pilose, acute, nerveless to 1–3- nerved, not embracing the upper glume. Upper glume 1.4-1.8 mm long, 7-9-nerved. Lower lem- ma glumiform, 7-nerved. Lower palea 0.7 mm long, 0.2 mm wide, hyaline, glabrous; lower flower absent. Upper anthecium ellipsoid, 1.5-1.8 mm long, 0.6 mm wide, indurate, papillose, shortly apiculate, apicule puberulous; stamens 3, the an- thers mm long. Caryopsis ellipsoid, 1.2 mm long, 0.6 mm wide, hilum punctiform, embryo Y the length of the caryopsis. Distribution and ecology: from Mexico and the West Indies to northern South America, in Colom- bia, Ecuador, and Venezuela. Commonly found on forest edges, roadsides, and riverbanks, in moist places among shrubs, between 100 and 2,800 m Selected specimens cited. BELIZE. BELIZE: Ridge La- goon Plantation ca. 12 mi. NW of Belize, Croat 24077 А (MO). EL cayo: Mountain Pine Ridge, along Coona Cairn Road, Davidse & Brant 33022 (MO). STANN CREEK: Pine Ridge, Gentle 8491 (US). TOLEDO: lower part of Rich- ardson Creek, affluent of Bladen Branch, 100-250 m, Davidse & Brant 32336* (MO, SI). COLOMBIA. ANTIOQUIA: Mun. Salgar, km 15 of road Salgar—El Dauro, 5 76907 МУ, Zarucchi et al. 5957 (MO). yb Acueducto Villa de Leyva, 2,100 m, Wood 442 K); entre Chinavita y Tibana, 1,700 m, .. 4117 nas, l, 700- i, 850 m, Killip & García 33718 (COL, US. MAGDAL- ENA: Sierra Nevada de Santa Marta, región del Campano, 1,300 m, appen Gutiérrez 1918 (COL, US). NARIÑO: Mun. Junin, 1,100 m, Pinto et al. 1889 (COL); La Florida, рта и 3314 (COL); Reserva Га Planada, Que- brada El Mar-La Calladita, Olga de Benavides 9533 (MO). NORTE DE SANTANDER: road from Pamplona to To- ledo, crossing the divide between Rio La Teja and Rio Mesme, 2,500-2,800 m, Killip & Smith 19812 (MO, US). SANTANDER: Mesa de los Santos, 1,500 m, Killip & Smith 15106 (US). TOLIMA: “La Trinidad,” Libano, Pen- nell 3323 (US). VALLE DEL CAUCA: La Cumbre, 1,600 m, Pennell & Killip 567 1 (NY), Pennell 5020 (NY). Costa Rica. ALAJUELA: 15 km an Ramon by road, 2.5 km N of Balsa on road to San Lorenzo, Liesner Judziewicz 14878 (MO, SI). CARTAGO: ca. 8 km Cartago by air, Finca El Chaparral, 4.5 km S of bridge on Agua Caliente at Lourdes, Liesner & Judziewicz 14619 МО); 3 km S of Agua Caliente, 1,400 m, PoAl & Davidse 11376 (US). GUANACASTE: Parque Nacional Rincón de la Vieja, SSE slopes of Volcán Santa Maria above Hacienda Santa Maria, 900-1,200 m, Davidse et al. 23439 (MO, SI). PUNTARENAs: Las Alturas along Rio Coton, Davidse 24132 (MO, SI). SAN JOSE: Cordillera de Talamanca, 14 ~ m, Pohl & Davidse 11618 (US). С LA JUVENTUD: Columbia, Ekman 12406 (US). EcUADOR. TUNGURAHUA: Valley of Pastaza River, between Canos and Cashurco, 8 hr. E of Banos, 1,300-1,800 m, Hitchcock 21809 (05); valley of Rio Pastaza, Tintin Rio Verde Grande, Asplund 7839 (6, K, d Agoyan, Asplund ud (MO). EL SALVADOR. CHALATENANGO: pass on Hwy. 7 km SSE of La Palma, Pohl & Davidse 11889 (MO). с. JATEMALA. ALTA VERAPAZ: Coban, 1,350 m, von Tuerck- heim 3836 (М); savanna on S side of Cerro Chinaja, between Sachaj and Sacacac, 150-180 m, Steyermark 45168 (US). BAJA VERAPAZ: Biotopo del Quetzal, WNW of Purulha, Stevens et al. 25420* (MO); 4 km N of El Chol, Harmon & Dwyer 3127 (MO). HoNDURas. AT- LANTIDA: valley of Rio Cangrejal, 17 km SSE of La Ceiba, ohl & Davidse 12084 (MO). COMAYAGUA: 14 km SE of Taulabe by road, Pohl & Davidse 12112 (MO). EL PARAISO: Cerro Monserrat, cerca de Yuscarán, Nelson & Romero 4267, 4279 (MO); 32 km W of Danli along Hwy. 4, Pohl & Davidse 11928 (MO). GRACIAS A DIOS: Caserío de Rus-Rus, vaguada del Rio Rus-Rus, Velson & Romero 4097 (MO). L4 PAZ: Agua Blanca River, between Chinacla and Planes de Mulle, Molina & Molina 24346 (MO). FRANCISCO MORAZAN: Rio del Gallo, near El Jicarito, 2 km from El Zamorano, Swallen 10992 (MO); 2.5 km N of Zambrano on road to San Francisco de Soroguara, Pohl & Davidse 12136 (MO); 14 km S of El Zamorano on road to Guinope, Pohl & Davidse 12147 (MO). OLANCHO: Montana Los Zapotes, 10 km noroeste cam pamento, Martínez 37 (MO). SANTA BARBARA: Trinidad, Finca Las Colmenas, Ra pde 15 (МО). JAMAICA. — Ground, Adams 10098 (MO). Mexico. CHIAPAS: Lag Pojoj, Lagos de Montebello National Park, Breedlove ү Davidse 55022 (MO); Mun. La Trinitaria, Lago Tziscao, Lagos de Montebello National Park, Breedlove & Davidse 55031 (MO). ОАХАСА: Mun. San Pedro Yaneri, Distr. Ixtlán, 17.3 km S of La Esperanza and 40.3 km SE of Valle Nacional, 1,900 m, Davidse et al. 30257 (MO). VERACRUZ: about 7 mi. SSE of Jalapa in area of rolling hills, Reeder & Reeder 6000 (MO); Orizaba, Hitchcock s.n., Amer. Gr. Hb. 179 (MO). NICARAGUA. ESTELI: Cerro Tisey, faldas del lado 5, 1,300- 1, e m, аы 9670 (MO, SI). GRANADA: Volcán Mombacho, lado N, arriba de ү El Progreso, Neill 868 (MO). in A: along wy. 3 са. 1.9 Lu NW of Aranjuez road entrance, Stevens 5576, 5591 (MO). MATAGALPA: Santa 1 Сога ега Central de Nicaragua, between War om and Jinotega, 1,300-1,500 m, ы iams et um 3591 (F, US). NUEVA SEGOVIA: El Cine p de El Jicaro, Moreno 19510 (MO). RIV AS: Isla Ds, Volcán Concepción, costado NW del v e referencia a “Los Hatillos ~ ATEGUI E Du Fr above Los Pajaritos, 31 airline km NE of Bergantin and Volume 80, Number 1 1993 Zuloaga et al. Revision of Panicum subg. Dichanthelium sect. Dichanthelium N of Mundo Nuevo, Serrania de Turimiquire, 1,400- 1,700 m, Davidse & оре 19914 (МО, VEN). АКА- cua: Distr. Ricaurte, cerca de Colonia Tovar, a 0.6 km del е bo pine via Colonia Tovar, Zuloaga & Ortíz 4279*, 1 (MO, SI, VEN); Parque Nacional Henry Pittier, A aga & Ortíz 4530*, 4532*, 4537* (MO, ЈЕ ); Distr. — 0.5 km del desvio a Buenos Aires, Zuloa al. 4288 (MO, SI, VEN). COJEDES Cerro Azul, Fila i Banques al NE de La Sierra, 1,100- 1,300 m, Delascio 7535 (VEN). ои а Сог- dillera Costanera, El Junquito, Chase 12427 (K, VEN), 12428 (US), 12431, 12440 (US, E ge (05); Piso de El Avila, Alston 5561а (MO); Parque M n Pico m ; m, Steyermark 99 (M, US, VEN). L LARA: ien ban en, El Cortijo «bod " V0570 (MO, US). M A: Capellanía (Bailadores), 1,600 m, Tamayo 2362 (У EN). MIRANDA: 34 km of Santa Lucia by road, Davidse 2912 (MO). TACHIRA: between Las Delicias and Paraguita, bordering Rio Tachi- ra, Steyermark 57148 (US). Panicum viscidellum is a common species with some degree of variation in the size of leaves and pilosity, characters that were used for describing P. tremulosum, P. blakei, and P. furtivum. Pan- icum viscidellum is related to P. sciurotoides and P. umbonulatum, differing from the first species by the presence of a conspicuous pseudoligule and its usually larger spikelets, and from P. umbonu- latum by its smaller spikelets. Pohl (1980) noted that the correct name fo: this species might be Panicum reflexopilum Steu- del, Syn. Pl. Glumac. 1: 84. 1853 (described from Oaxaca, Mexico). However, neither he nor we could find the type specimen of P. reflexopilum, so the name remains dubious. All specimens examined for anatomical purposes have a leaf anatomy typical of section Dichan- thelium, including numerous macrohairs, either cushion-based (Fig. 18D) or with few modified epi- dermal cells associated with the hair bases (Fig. 19A- D). Several of the specimens exhibit ап un- usual silicification of the basal cells associated with the macrohairs (Fig. 20). The hairs themselves may be lost but the modified, then clearly seen to contain opaline silica deposits (Fig. 20F). swollen basal cells are LITERATURE CITED Возурем, W. M. 1960. Chromosome numbers and tax- onomic notes on oder =, ^ Twenty-five genera. Canad. J. Bot. 38: 54 ien ELL, E. 1914. A iuis s dc le ai anatomy of some native species of the genus Andropogon N.O. Gramineae. Proc. Linn. Soc. New South Wales, ser. 2, 39: 385-394 Brown, W. V. 1946. A Pr и study in the Pan- iceae. Amer. J. Bot. 33: A hun m in the gramineae. me 1. Bot. 35: 382-395. The Kranz syndrome and its pies in grass systematics. Mem. Torrey Bot. Club. -9 & B. N. SmITH. 1975. The genus Dichan- Teri Misaia). Bull. Torrey Bot. Club. 102: кы € W. 1942. A cytological study of some species in the tribe Paniceae. Amer. J. Bot. 29: 355- 359. Сноксн, C. L. 1929. Meiotic DM in certain Gramineae. II. Paniceae -— ia затим е. Вог. Gaz. (Crawfordsville) 88: ~ C.F. A. 1977. osten Studies of Selected xas Species of Dichanthelium (Poaceae). Duce а Texas A & M “Univ. College Suou, T exas. Ы Е. У. Gourp. acteristics of Panicum, Dichanthelium nochloa. Amer. J. Bot. 62: 743-748 DavipsE, С. & R. W. Pour. 1972a. Chromosome num- bers and notes on some Central American grasses. A J. Bot. 50: 273-283. 72b. poia numbers, meiotic behavior, and notes o e grasses from Central America and the West Indies. Canad. J. Bot. 50: 1441-1452. 1975. Some epidermal char- ) and Echi- 1974. Chromosome numbers, mei- otic behavior LM notes on tropical American grasses (Gramineae). gy" J. Bot. 52: 317-328. 97 Chromosome numbers of tropical American p (Gramineae). Ann. Mis souri Bot. Gard. 6 49 DoELL, J. C. 1877. Tr ibe 3. а. In: C. Е. P. von Martius (editor), Flora Brasiliensis 2(2): 33-342. Mu- nich, Vienna, Leipzi Duscovsky, J & F. 0. ZULOAGA. 1992. Recuentos cromosómicos en especies sudamericanas del género Panicum (Poaceae: Panicoideae: Paniceae). Bol. Soc. Argent. Bot. 27: 201-206. ELuis, R. P. 6. in for standardizing com- parative leaf blade anatomy in the Poaceae. I. leaf blade as viewed in transverse section. Bothalia 109 12: 65- 1979. А procedure for п com- parative leaf blade anatomy in the Poaceae. II. The epidermis as seen in surface view. Bothalia 1. 641- 672. P. Шрек. 1992. Atlas of the leaf anatomy in Pentachistis (Arundineas: Poaceae). Mem. Bot. Su outh Africa 60: 14 FEDER, М. & О.Р 1968. Plant RN nique: some yc and new methods. Amer Bot. 55: 123-142. оаа К. m 1981a. The correct name for Di- chanthelium унији и and its va- rieties. Brittonia 33: 457 19 1 ii sci c in the Dichanthelium oaceae). Phytologia 48: 99- 10. Соор, F. W. 1958. Chromosome numbers in south- western grasses. Amer. J. Bot. 45: 757-7 186 Annals of the Missouri Botanical Garden ——. 1974. Nomenclatorial changes in Dichan- thelium (Poaceae). Phytologia 39: 268-271. ——. 1980. The Mexican species of Dichanthelium — 87 ie зе 32: 353-364. & C CLARK. 1978. Dic кыш он (Pontes). in le United pore and Canada. Ann. Missouri Bot. Gard. 65: 1088-1132 —— а Т. ава а 1967. ован numbers of tropical American grasses. Amer. J. Bot. 54: 676-683 Hansen, В. F. & R. P. WuNDERLIN. 1988. p of Dichanthelium (Poaceae) in Florida. Ann. Mis souri Bot. Gard. 75: 1637-1657. HarrERSLEY, P. W. & L. WATSON. Anatomical parameters for predicting no pathways of grass leaves: the “maximum lateral cell count’ and the “maximum cells distant count.” Phytomor- phology 25: 325-333. Hircucock, А. S. & A. СНАЗЕ. 1910. The North Amer- ican species of Panicum. Contr. U.S. Natl. Herb. JOHANSEN, D. A. 1940. Plant Microtechnique. McGraw- Hill, New York. 1984. New combinations for Panicum sub- m and Dichanthelium ар of Southeastern United States. Brittonia 62-213. METCALFE, Е К. . Anatomy of the Monocotyle- dons. 1. Gramineae. Clarendon Press, Oxford. Mois. o & F. О. ZuLoaca. 1991. Estudios mor- fologicos en el subgénero Dichanthelium de Pani- cum (Poaceae: Panicoideae: Paniceae), con especial referencia a Panicum sabulorum. Ann. Missouri Bot. Gard. 78: 915-927. Investigaciones cariosistemáticas en tribu Paniceae. Rev о. Plata 28: 209 255. Aires. Gramineae. /n: W. Burger (editor), ora Сонын. Fieldiana: Botany, 4: 1-608. Chromosome numbers of Costa Rica grasses. Brittonia 23: 293-324. RENVOIZE, S. A. 1987. New grasses from Paraná, Bra- zil. Kew Bull. 42: 921-925. . W. SPELLENBERG. 1976. Miscel- nts of Western American 2. — 86 & К. L. HORNBERGER. 1983 IOPB Chromosome Number 08. ш А. S., . B. SMITH : À. Lave (editor), Rep orts LXXX. Tax NUS aid T. R. & R. P. ELLIS. 1987. The position of bamboo genera and allies in a system of grass classification. Pp. 225-238 in T. R. Soderstrom, K. . Hilu, C. S. Campbell & M. E. Barkworth (editors), Grass Systematics and Evolution. Smithsonian Inst tution Press, Washington, D.C. E. J. Jupziewicz. 1988. Systematics of the amphi- atlantic bambusoid genus Streptogyna (Po- aceae). Ann. Missouri Bot. Gard. 74: 871-888. . ZULOAGA. 1989. A revision of the genus Olyra and the new segregate genus Parodi olyra (Poaceae: Bambusoidae: Olyreae). Smithsonian Contr. Bot. 69: 1-79. SPELLENBERG, R. W. In: А. Lóve (editor), IOPB Chromosome Number Reports XXV. Taxon 19: 112- 113. SWALLEN, J. R. 1939. Gramineae. /n: Gleason & Killip (editors), The Flora of Mount Auyan-Tepui, Vene- zuela. Brittonia i йч did 152. . In: Maguire, ire mark & Wurdack (editors) dou of Те Chimantá Mas- sif I. Gran Sabana, Venezuela. Mem. New York. Bot. Gard. 9: 393-408 TÜüRPE, А. M. 1966. Histotaxonomía de las он i 9 Paniceae (Poaceae: Panicoideae). Syst. Bot. 13: 5 60 ZULOAGA, ay O. 1987. Systematics of the New World species of deri (Poaceae: Paniceae) Pp. 287- n T. R. Soders Hil M. E. A (editors): Grass Systematics and Evolution. Smithsonian Institution Press, Washing- ton, D.C & E. J. Jupziewicz. 1990 [1991]. Panicum. Pp. 370-436 in Gorts-Van Rijn (editor), Poaceae, Flora of the Guianas. Koeltz Scientific Books, Koenig- stein. , J. Duscovsky & О. Morrone. 1989. Exo- morphological, anatomical, and cytological studies in Panicum validum (Poaceae: Panicoideae: Paniceae): its systematic position within the genus. Syst. Bot. 14: 220-230 INDEX TO CITED SPECIMENS. Each specimen is listed by the first collector, even when other collectors participated in the collecting. Vouchers utilized for anatomical studies are marked with an asterisk. Acosta Solis 12834 (2a); 20832 (29a). Ahumada 4829 (25b). Alain 2108 (2b); 8796 (28). Adams 8258 (32); 10098 (37). Allard 14508 (2a); 14509a (12a); 14893 (Ла); 14910 (18); 15870 (32); 15873 (1a); 16029 (1a); 16089 (la); 16408 (28); 16494 (la); 17402 (18); 17407 (28); 17485 (28); 17490 (18); 17516 (28). Allen 772 (32); 1809 (25a); 1868 (25a). Almeida 22 (24). Alston 5561a (37); 7978 (37); 8205 (32). Alvarado 72 (18) эы ag (20 Amer. Gr. Hb 84 (18); 88 (1а); 133 (2a); (28); 178 (28); 179 (37); 193 (13); 225 (13) 2 Anderson 35542 (27); 35583; (33) 35729 (27 (33); 36155 (20); 36397 (34). Andino 76 (18). Аеш 2194 (27). Агсһег 240 (29а); 1077 (2а); 1167 (29а); 1182 (2a); 1306 (2); 3228 (37); 3314 (37); 3318 (2а); 4440 (25b). Arechavaleta 144 (25a); 267 (25b). Arnason 17200 (37). Asplund 7627 (37) 7839 (37). Atwood 2941 (2a). Bacigalupo 579 (25b) 696 (255); 847 (25a); 871 (25b) 1031 (25a). Badillo s.n. (la); s.n. (1b). (25b); 4a (25b); 5 (25b); За (25b); 7 (34 (25a). Balls 4690 (12a). Bang s.n.; 1890 (15 73; 119 (25b). Barbosa 18 (27). Barreto 2003 (25b); 2008 (255); 10510 (20); s.n. (25а); s.n 29] (25b). Barkley 1918 (37). Bartlett 11225 (23); 11236 (14); 11602 (12b); 11621 (2a); 11780 (1b). Beetle 1911; 1972 (25a); М-2193 (18); M-4807 (36). Belém 1692 (4). Berrio 6350 (25b). Berrios 57 (37). Bertero s.n. (2a). Black 2153b (33); 55-18336 (27); 55-18347 (27). Black- more 3838 (18). Blackwell 2696 (32). Blake 7436 (2a); 7457 (2a); 7739 (2a); 7762 (2a); 7780 (2a); 7458 (1b). Blanchet 193; s.n. (26). Blanco s.n. (29b). Blauner 231 (2a). Boechat s.n. (34). Boelcke 788 (25b); 6795 (25b). Volume 80, Number 1 Zuloaga et al. Revision of Panicum subg. Dichanthelium sect. Dichanthelium Bonilla 167 (37). Bourgeau 2162 (18). Box 3962 (37). Bovin 193 (25a). Вгаде 19172 (7); 7857 (15); 8108 (15); 20119 (15); 20226 (33); 7837 (27); 20368 (27); 20524 (255); 20757 (25b); s.n. (30); s.n. (33); 10878 (12a). Breedlove 13960 (37); 21946 (13); 27062 (2c); 28188 (18); 38851 (37); 40893 (18); 41291 (2c); 53885 (2c); 53886 (29a); 54707 (2c); 54741 (2c); 54814 (29a); 54820 (2c); 54954 (29a); 54998 (2c); 55020 (18); 55022 (37); 55031 (37); 55034 (2b); 55037 (32); 55038 (12b); 55104 (18); 55114 (12a) 55115 (18) 55141 (13); 55154 (29a); 55179 (29a); 55196 (2c); 55208 (29a); 55211 (18); 55333 (1b); 55356 (32) B55497 (2c); 55518 (18); 56009 (29a); 55649 (29a); 55687 (12b); 58273 (36). Bourdeth 17 (36). Brito 26 (27). Britton 599 (la); 3583 (23); 4090 (2a); 6651 (23); 6831 (23); 6952 (2b); 14150 (29b); 14218 (14); 14221 (14); 14224 (2a); 14226 (23); 14283 (12b); 14305 (1b); 14320 (2a); 14357 (37); 14411 (32); 14705 (295); 15379 (23); 15383 (12b); 15674 (2a); 15706 (29b). Brunken 265 (18); 336 (18). Buchtien 26 (27); 46 (27); 414 (27); 5319 (27); 7115 (27); 7116 (27). Bueno 1773 (18). Burandt Jr. V0175 (37); V0327 (37); V0383 (2b); V0546 (37); V0570 (37). Burch 6057 (37). Burchell 1111 (25b); 4246-2 (15); 4445 (25b). Burkart 275 (25b); 3922 (25b); 7943 (25a); 8886 (25b); 8936 (25b); 8937 (25b); 11186 (25b); 11280 (25b); 11332 (25b); 17540 (25b); 18833 (25b); 18851 (25b); 19392 (25b); 19982 (25a); 20801 (25b); 20978 (25b); 21662 (25a); 22864 (25b); 22927 (25a); 23267 (25a); 24080 (25a); 26208 (25b); 28480 (25а); 29692 (255); 30224 (25a) Burman 201 (27); 208 (27); 211 (27); 384 (16); 485 (16); = ы 6); 696 (16); 697 (16); 709 (16); 754 (27); 778 Caballero 124 (2a). Cabrera 23574 (2 on ae (25b); 28777 (25b); 28898 (25b); 32004 (25b); 32393 (25b); 32424 (25a); _ 5a). Calderón 23 (37); : 2414 (10); 2425* (10). ~ 1444 (27). Canisio 365 (25 Carauta 2340 (25b). Же йн 25628 (16). 148 (29а). СҺасоп 2266 (32). adn 151 (1a). € 6249 (2a); 6273 (la); 6276 (32); 6475 un 6621 (32); as + с ios as io 2 / (30); 9555 (34); 9556 (30); 9608 (20); 9640 (15 (20); 9662 (30); 9680 (25b); 9690 (33); 9730 (25) 9731 (25b); 9745 (15); 9752 (20); 9769 (25b); (15); 9815 (34); 9822 (33); 9839 (34); 9894 (25b): 10081 (15); 10084 (33); 10106 (30); 10112 (30); 10187 (27); 10200 (20); 10217 (20); 10251 (27); 10253 (25b); 10289 (16); 10297 (20); 10354 (16); 10373 (27); 10387 (34); 10677 (34); 11023 (34); 12161 (33); 12162 (15); 12165 (15); 12170 (15); 12427 (37); 12428 (37); 12431 (37); 12436 (la); 12440 (37); 12441 (37); 12458 (37). Chiang 9452 (19). Claussen 228 (20). Clayton 4190 (15); 4240 (25b); 4269 (25b); 4280 (25b); 4285 (33); 4305 (34); 4371 (25b); 4711 (25b); 4741 (15); 4743 (27). Сопсајуев 2211 (33). Conrad 3270 (18). Contreras 528 (32); 2922 (2a). Cook 37 (2a). Coradin 6530 (4). Costa Sacco 182 (25b) 202 (25b) 364 (25a). Cowan 1891 (24); 2109 (27). Croat 12066 (37); 21272 (1a); 21315A (37); 21521 (37); 23398A (27); 23259 (29b); 23391 (27); 24077A (37); 24777A (29b); 41154 (36); 41422 (36); 46634 (18); 54243 (27); 63947 | Mea epe 65910 (18); 71502 (15). Cruz 70 (18); 173 (18); 1 (36). Cuatrecasas 185 2a). Curtiss 307 a 06 т Ў да Silva тА (15); 180 (15); 249 y 284 (27) 2a). Davidse 972 (2a); (29a); 26 0 3481 (2a); 3495 (2a); 3532 (2a); 3572 (2а); 4017 (37); 4070 (1a); 4074 (37); 4079 Lt 4749 (24); 4762 (4); ); 4795 (24); 4951 (27); 9368 4 (37); 19914A (37); 29621 (18); 29625 (12a 29630 ( 30257 Т 30807* s N en we * ~ (37); 32844 (2b); 32911 (2a); 32927 (la); 32981 (2b); i 3022 (3 Davidson 766 (2a). bu s.n. (27). Da Я 36). del Mazo s.n. (25b). Delascio 7535 : aree 66629 (2a). Diogo 599 (15). Dionisio 891 = Díaz z 983 (la). Diaz-Piedrahita 1077 (37); 3815 ү. раје 16884 (32). pri 277 (27), 1958 2s 2200 (33); 3943 p de Dorsett 152b a). Dusén 646 (25a); 1099a (15); 2039 e 3659 (33); 3661 (33); 14522 (27); 15565 (25b); 5596 (25c); 15766 (25b); 15767 (25b); 16127 (25b); Tie a 17520 (25b); 17531 (34); 17876 (34). 0 (2a); 10082 (32); 10127 (32); 10716 (29b); 12711 ant 15016 (2a); 15207 (29a) D'Orbigny 8 (25a). D'Urville s.n. (25a); s.n. (27); 1821 (25a). Ebinger 908 (32). Edwards 438 (19). Eggers 2021 (2a). Eggler 9 (13). Eiten 6236 (25b); 6929 (27). Ekman 645 (25b); 1095 (18); 1312 (1a); 1604 (18); 1946 (1a); (2a); 6857 (32); 8117 (18); 8151 (23); 8206 (32); 9101b (28); 9701 (1b); 9704 (aal 10297 (13); 10329 (1b); 10802 (12b) 10803 (32) 10860 (18); 10937 10994. (12b); 11047 (14); 11058 (2b); 11068 (32); 11088 (32); 11138 (2b); е? i B (2b); 11242 (29b); 11261 (23); la) 11388 (1а); 11421 (23); 11430 11630 (1a); 11651 (37 " 11654 (1b) 117 12 (37) 11713 (2b); 11855 (18); 11942 (12a); 11946 (Ла); 11995 (1b); 11996 (la); 12026 (23); 12098 (14); 12123 (32); 12152 (14); 12402 (1а); 12406 (37); 12581 (1а); 12697 (1 2a); 12794 (1а); 12943 (28); 12969 (32); 13808 (23); 13932 (12a); 14094 (23); 14674 (23); 15290 (23); 15636 (2b); 15896 (28); 16841 (2a); 17078 (la); 17891 (la); 18395 (1a). Emygdio 1663 (26). Enamorado 37 (18). Erazo 38 (36). Eskuche 1803 (25b). Espinosa 547 (36). Eugenio 285 (15); 2649 (15). Eyerdam 23132 (25b); 23678 (25b). 188 Annals of the Missouri Botanical Garden Felippone 2706 (25b); 2786 (25a). Fernández 84 (1a Figueroa ( Filgueiras 1964 . Flora Falcón 949a (1a). Folsom 2534 (37). Fosberg 42730 (32). Frey- гав 3285 (36); 3288 (36). Fróes 19950 (4). Fuentes s.n. — 5a). pom 98 (29a). Galland 55 (25a). eun 241 ra Gamerro 66 (25b). Garcia-Barriga 4434 (2a); 4435 (1a Gardner 5653 (15). Gau јер " "n (15); 12 (27); 7 25a) Gay s.n. (25a). Geller 4236 (28). Gentle p (14); 3665 (2a); 3748 (32); 5905 (b 4122 (2a); s (2b; 8491 (37); 8550 (32); 8551 (За). Gentry (29b). Gereau 2107 (18). Giberti P 5b). Claziou ee 25b); 5718 (20); 9091 (20); 1863 = ~ 4 © * 25b). Gorgun 18 (2b). 9262 (12a); 134110620 y е, 15705 (23). Gomes 3955 (16). G ille 8966 (27). Gutiérrez 138 (2a); 1083 (2a) Hage 1691 (27). Hall 7602 (29a). Hamilton 1760 es 2094 эт, 5237 (32); T ed. 3634 s Hans 7 (25a). Harley s.n. (10); 0 (10); 15783 ei Sor 16636 ЭТ; 170 en 18829 (10); 24460 43 . Harmon 2708 (2a 2a); 11597 (2a); 12225 (32); 2 12234 2 (32). Hashimoto - РЈ сл m КҖ = 3 27 12293 (25by 13731 (34); 147 58 (27); 32663 (27); 35451 (25b); 35525 (25a); 38037 (25b); 43562 (25b); 47961 (10); ba (10). Heller 1736 (19); 6442 (23). Heringer 18124 (34). Hernández 89 (37) 198 (36). Herter s.n. (25a); 324 (25a); 1153 (25b). Hitchcock s.n. (2a); 117 (1b); 8178 (37); 8252 (2a); 9532 (12a); 9535 (Ла); 9550 (32); 9551 (32); 9552 (1b); 9561 (32); 9697 (За); 21211 (27); 21420 (24); 21809 (37); 21862 (15); 23257 (1b); 23271 (2a); 23298 (1a); 23299 (32); 2205 (12а); 23411 (32); Hitchcock 23414 (28); 23428 (la 23433 (12b); 23434 (1b). Hodge 6663 (2a); 6954 (31). Hoehne 4741 (27); 5789 (4); 5991 (4). Holdridge 1729 (18). Holm s.n. (28). Holway 1737 (34); 6444 (1a); 9079 (18); 9162 (2a); 9331 (28). Huber 10008 (35); 12905 (24); 12908 (4); 13043 (24); 13046 (24). Hunt 451 (37). Hunziker 2166 (25a Imaguire 2762 (25b). Irgang s.n. (25a). Irwin 2311 (27); 5419 (34); 10848 (20); 19715 (25b); 19727 (20); 27543 (27); 27666 (27); 27933 (27); 28151 (27); 28937 (33); 28942 (34); 29483 (33); 30702 (31); 31023 (27). Izaguirre 53 (18). Johnson 18C762 (2a). Jiménez 1236 wo 2436 (2a); 4427 (18). Johnston 8919 (19); 10974 (19 (16); 513 (2 Juncker 5715 (1 а, Jürgens 61b (25b); 109 (25а); G-198 (25b); G-327 (25b); 9327 (25a). Kappel s.n. (25a). Kermes 346 (25b). Killip 4525a (2a); 4560 (2a); 5671 (37); 15106 (37); 15193 (2b); 19812 (37); 20046 (37); 20544 (37); 32328a (29b); 32376 (2b); 33718 (37); 42682 (2a); 42848 (37); 42942 (la); 43723 (29b); 44058 (29b); 44193 (29b); 44614 (32); 44805 (125); 44825 (29b) 44832 (lb) 44853 (12a); 44881 (2b). Мет 2646 (27); 5936 (25b); 11718 (27); 12031 (33). Knapp 3822 (27). Knoth 4007 (37). Koyama 7391 (4). Kral 69308 (32). Krapovickas 20048 (25a); 11627 (25b); 20360 (25b); 22801 (25b); 22848 (25a); 23817 (25b). Krieger SVD 1227 (34). Kuhlmann = s.n. (20); 52 (30); 276 (27); 1942 (20); 1943 (15); 1948 (27); 2044 (25b) 15017 (25b). Kummrov 829 (25b); Láinez 91 (36). Lasser 3005 (27 (13). Lehmann 974 (2a); 985 (37); 7000 (37). Leite 487 (25a); 3510 (34); 4262 (20). Leonard 3821 (2a); 3852 (18); 4331 (la); 4346 (18); 4501 (18); 4701 (2a); 7838 (2a); 7920 (18); 8199 (2a); (12b) 3472 (а); 4371 (а); 4829 (32); 5177 (1a); 5840 (1b); 6152 (12a); 6422 (1a); 6457 (13); 6902 (2a); 7007 (1b); 7009 (125); 7451 (1b); 8615 (29b); 9203 (12a); 9245 (la); 14201 (13); 14821 (2a); 17885 (12b); 17979 (28); 18702 (23); 18789 (2b); 18882 (2a); 19193 (1a); 19414 (2b); 19474 (14); 19828 (12a); 19856 (28); 19981 (12a); 20007 (23). Liebmann 328 (18). Liesner 13514 (37); 12373 (27); 14466 (37); 14592 (37); (37); 14878 (37); 23580 (27); 25867 (24). Lindheimer 1265 (19). Liogier 10532 (1a); 11713 (18); 13365 (2a); 14162 (2a); 15872a (1а); 15974 (28); 16106 (28). Lour- teig 2884 (25a). Luces 35 (37). Lundell 6559 (32); 6561 (1b); 6563 (12b). Luteyn 8262 (37); 8294 (37). Lützel- burg 133 (27); 135a || 6051 (20); 26101 (26); 26141 (26). Гуоппе! 623 Maas 5769 (4). Ми s.n. (25a); s.n. (25b). M galhàes s.n. (27); 15584 (27). Maguire 23446 (24); 42193 4); 54280 (24). Malme 422 (25a); 435 (25a). Malmierca 2006 (25b). Manara s.n. (1a); s.n. (37). Marcano 4867 (18). Martinelli 814 (9). Martin 65 (37). Martinez 37 37); 59 (2a); 16348 (2b). Martinez Salas 1494 (37); 2562 (29a). Mattos 11991 (25a); 12720 (33); 14248 (30). Maxon 594 (2a). Márquez 115 (29a). McDaniel 27158 (28). McKee 10588 (23); 11288 (29a); 11293 (2a); 11381 (15); 11388 (1b); 11390 (12b). McVaugh 18881 (13); 19612 (13). Mendes Magalhaes s.n. (20). Metzler 6 (29a); 21 (2a). Mexia 1812 (29a); 4014 (33); 4738 (27); 5368 (20). Meyer 2663 (19). Miller 132 (1b). Molina 1272 (36); 7550 (37); 8034 (36); 14707 (1а); 24346 (37); 25283 (lb); 26112 (36); 27489 (29a); 27868 (1b); 27870 (29a). Montes 10815 oa 25b). Montoro 2667 (25a). Moreira-Filho 416 reno 9473 (18); 9670 (37); 17694 (18); 18455 (18); 19510 (37 14257 (4). Morillo 1433 (la); 1445 (37); 1447 (la); 8150 (27). Morton 3081 (1a); 3082 (28); 3100 (32); 3123 (12a). Mroginski 435 (25b); 437 (25b); 441 (25b). Mueller 408 ка 414 (13); 2057 (19). Muniz s.n. (27). Murillo 204 (2a Nash 1243 (1 ak 1337 (23); 2076 (23). Nee 10694 (37); 16805 (37); 23182 (29a); 26237 (18); 26325 (2a). Neill 868 (37). Nelson 634 (2a); 656 (2a); 1005 (32); 3618 (36); 3800 (13); 3842 (32); 4026 (36); 4097 (37); 4267 (37); 4279 (37); 4708 (32). Nicora 3067 (25a); 4170 (25a); 5903 (25b); 6027 (25b); 6530 (25b). Nob- lick 2472 (27); 2735 (26); 2870 (10). Olga de Benavides 8680 (37); 8825 (37); 9213 (37); . Leavenworth 621 = as — 85 (29a). Orth 5 (25a); 1938 (15). Osorto 44 (37). Osten 6511 (25b); 6917 (25а); 20083 (25b). O'Neill 8500 14 ). Pabst 6558 (25b). Pachano 228 (27). Paciornik 117 27). Padilla 1122 (2a). Palmer 213 (32); 990 (32); 1065 37); 1083 (2a); 33820 (19). Parodi 3922 (25a); 12032 (25b); 12364 (25a); 12397 (25b). Partch 69-96 (2a). Pastore 1227 (25b). Pavetti 10855 (25b). Pedersen 2997 25b). Peixoto s.n. (15); s.n. (20). Pennell 3323 (37); 5020 (37); 5671 (37); 6362 (32); 7280 (2a); 7300 (2a); mm — Volume 80, Number 1 1993 Zuloaga et al. 189 Revision of Panicum subg. Dichanthelium sect. Dichanthelium 8106 (1a). Pereira 1489 (33); 6732 (25b); 6811 (25a); 7056 (33). Peterson 7362 (2a). Pickel 5054 (25a). Pi- neda 82 (37). Pinto 117 (27); 941 (26) 1739 (2a); 1803 (32); 946 (2a); 982a (2a); 5889 (la); 6042 (la); 9382 (37); 9962 (2a); 13773 (2a); 13788 (37); 13792 (1a). Plowman 10045 (4); 10139 (25b). Pohl 12572 (29a); 12685 (32); 12749 (37); 12756 (36); 12839 (18); 10535 (29a); 11046 (29a); 11074 (18); 11249 (37); 11376 (37); 11377 (29a); 11379 (18); 11618 (37); 11676 (2a); 11677 (2a); 11889 (37); 11894 (2a); 11928 (37); 11977 (36); 11984 (36); 12084 (37); 12112 (37); 12131 (29a); 12136 (37); 12141 (2a); 12147 (37); 12152 (36); 12199 (2c); 12262 (32); 13123 (18). Ponte-Baez 22892 (37). Porter 4104 (37). Pringle 8083 (18); 8344 (29a); 13250 DADA Proctor 7334 (2a); 26565 (1a); 26567 (32). Purpus 9 (2a). Paes s.n ues 1593 (25b); 1653 (255); 1744 (25b) 2872 (25by 3171 (25b); 3200 (25b); 3205 (25b). Rambo 29194 (255); 29573 (25а); 34202 (25b) 38181 (25a); 44279 (25a); 44638 (25b); 44677 (25b); 53841 (33); 54088 (25b); 54532 (25a). Ramia 140 (37). Ramirez Reyes 2513 (la). Reeder 6000 (37); 6079A (29a). Regnell III-1358 (20). Reiche s.n. (25b). Reitz 1978 (25a); 2595 (25b); 3225 (25c); 3626 (27); (25b); 5159 (25b); 5278 (25b); 5279 (33); 5449 (33); 5561 (15); Ll (33); 10291 (33); 10362 (33); 21 . Renvoize 2929 (25a); 3034 (25b); 3109 (25b); ane (25b); 3743 s 4072 (25a); 4307 (27). Riedel s.n. (25a); s n. (26); 226 (30); 228 (27); 2151 (20); s.n. (15) Robleto 214 (37). Roca 7944 (la). Roe 2079 (13). Rohweder 2053 (36); 2111 (36); 2264 (36). ч 2205 (13). Roldán 289 (2a). Rom- ap s.n. id ). Romero-Castaneda 1443 (2a). Rondon n. (11); s 24). Rose 21134 (27). Rosengurtt s.n. (2 Sb): B- 241 (25b) B-247 (25b); B-739 (25b); B-2722 (25b) B-5077 (25b) B-6217 (25a); 6930 (25b); 9001 с 9253 с беры Rotman 61 (25b). Rubio 1 (29a); 77 (29 8 (29a). Rzedowski 1158 (13). Salguero 15 (3 2 alzmann 697 (27). Sampaio 1608 (25b); 4699 (27). Sandwith 1359 (24). Santos 2687 (25b). — ее s Sánchez Vega 2295 (24). Scala s.n. (25a) 6 (2c). Schenck 229 (27). Schinini 2423 (25b) 2583 pes 8775 (255); 11374 (25b). Schnem 1605 (25b). Schwacke 13153 (27); 14409 (25b). Schwarz 4979 om 6756 (25b). Scolnik s.n. (2a). Se gadas-Vianna 650 (30). Sellow 523 (34). Sendulsky 321 (15); 424 (33); 446 (16); 593 (25b); 686 (255); 693 (15); 711 (25b); 712 (25by 723 (15); 739 (25а); 768 (25a); 842 (25a); 947 (15); 1058 (25b); 1063 (33); 1166 5 ~ N сл 3531 (1a); 10484 (1b); 10639 (1а); 10873 (1а); 11021 (29b); 11049 (1b). Sharp 45928 (18). Shnee 484 (15). dv 355 (2a); 1224 (la); 5985 (la). Skutch 524 . Skvortzov 19 (25b). Smith 163 (32); 441 (25b); NEA (2а); 6107 (25b); 7314 (33); 7334 (30); 7640 (33); 7825 (33); 8059 (25b); 8664 (25b); 8710 (34); 10899 (33); 12534 (25b); 12712 (255); 12792 (25b); 13066 (25b); 13484 (255); 14008 (33); 14305 (25b); 15456 (33); 15464 (25a); 15478 (25b); 15491 (280); : 15523 (33); 15549 (33); 15558 (25b); 15629 (2 15680 (25b); 15737 (25b) 15809 (33); 15889 E 15896 (25b). Soderstrom 960 (37); 960a (1a); 968 (37 w = Solomon 1300 (27); 4091 (25b) 9523 (27). Soto 44 (18). Souza Santos 88 (27). Sp ). 22811 (29a); 22813 (36); 28745 (1b); 29366 (1a); 34048 (2a); 39320 (2 а); peer (2a); 56009 (1a); 56261 (2a); 64360 (2c); 65681 (2c); 69059 (12a); 71468 (а); 82760 (2c); 89929 (2a). Stern 1180 (2a). Stevens 4118 (2a); 5576 (37); 5578 (18); 5591 (37); 7649 (32); 7650 (37); 7816 (32); 12809 (32); 14796 (2c); 14802 (29a); 17972 (2c); 18103 (18); 18000 (29a); 18105 (37); 25420* (37); 25467 (18). Stevenson 2783 (23). Steyermark 29707 (la); 42418 (la); 42770 (12a); 43463 (19); 43942 (12a); 44005 18); 45168 (37); 47958 (2a); 49025 (2a); 56978 (1a); 57148 (37); 59906 (24); 75642 (24); 85862 (1a); 85871 (Ла); 87974 (24); 88596 (27); 89073 (15); 93883 (24); 94668 (12a); 97638 (27); 99143 (37); 99396 (37); 99468 (37); 104200 (24); 105858 (1a); 106199 (27); 109377 (24); 117547 (27); 121762 (15); 127736 (37); 127739 (37); 128196 (24). Strang 833 (33). St. Hilaire 296 (34); 323 (27). Sucre 1738 (15); 1955 (15); 2089 (15); 2317 (33); 3116 (25b) 4017 (25b) 9087 (25b). Sullivan 185 (37). Svenson 4486 (2b len 4303 (26); 4531 (24); 4539 (26); 4677 (26); 4823 (26); 7034 (25b); 7211 (25b); 7406 (25b); 8064 (25b); 8125 (25b); 8140 (25a); 8419 (25b); 8571 (25b); 9076 25a); 9252 (25a); 10824 (37); 10895 (36); 10952 (37); 10992 (37); 11012 (2a); 11018 (36); 11021 (36); 11034 (36); 11047 (36); 11048a (36); 11070 (37); 11075 (2a); 11287 (2c); 11299 (2a); 11324 (37); 11356 (2a); 11357 (37); 11376 (37); 11397 (37); 11411 (37). Sytsma 1462 (37); 32 E Tamayo 114 (1a); 1625 (37); 1861 (2a); 2362 (37); 49885 (37). Tate 281 (4). Tavares 788 (26). Taylor 127 (18); 11842 (18); 11868 (37). Tenorio 66-197 (26). Tharp 43164 (19). Thomas 2794 (18). Tillett 27 (2a). Toro 290 (2a); 353 (2a). Torres 2020 (18). Tracy 9074 10); 9078 (2a). Trejos 190 (1а). Troncoso 2020 (25a); 2670 (25a); 2707 (25a); 2709 (25a); 2802 (25a); 3253 (25a). Trujillo 346 (37); 826 (37). Tucker 1056 (29a). Tyson 4436 (37). Ule s.n. (33); 664 (33). Underwood 955 (23). Valeur 4 (23); 293 (18). Valls s.n. (25b); 1114 (25b); 1178 (25b); 1398 (25b); 1651 (25b); 1664 (25b); 1665 (25b) 1780 (25b); 1809 (25b); 1870 (33); 2308 (25а); 2361 (255); 2903 (33); 10087 eds van der Wer 6167 (32). Vasconcellas s.n. (27). Vázquez Avila 189 (25b). Venturi 8428 (255); 9545 (25b) 9691 (25b). Vianna 650 (30). Victorin 21448 (29b). Vidal П-6467 (9); II-6485 (9). Villela 48 (29a). von Sneidern 1305 37). von Tuerckheim 11-1322 (12а); 3059 (18); 3321 (23); 3836 (37). rming s.n. (20). Wawra 970 (15). Weberling 719 (37). Webster 25725 (10). Whitefoord 2407 (2b). Wilbur 14916 (18). Wiley 464 (1b). Williams 12076 (36); 14813 (1b); 15602A (2a); 15983 (36); 17130 (36); 18842 (1b); 22170 (36); 22171 (36); 23591 (37); 23906 (1a); 31265 (2a); 41919 (37). Wingfield 5415 (37); 6471 (1a); 6483 (37); 6707 (12a); 6844 (12a); 7124 (12a); 8518 (32). Wood 3660 (37); 3861 (37); 4426 (37) 5352 (la). Woodbury s.n. (23). Woolston G-117 (25b). Wright 3467 (28); 3874 (2a). Zarucchi 5957 (37); 5958 (2a). Zehntner 248 (31); 263 (31); 2017 (31); 2048 (31). Zuloaga 444 (25b); 463 (25b); 486 (25b); 554 (25b); 765 (25b); 1081 (25a); 1579 (25b); 1891 (25b); 1972 (25b); 2060 (25b); 2128 (25b); 2349 (27); 2351* (30); 2352* (25b); 2354* (27); ~ ~ ~ ~ — 190 Annals of the Missouri Botanical Garden 2358* (30); 2361* (25b); 2363 (30); 2373* (15); 2374* (33); 2380 (25b); 2382 (30); 2389 (30); 2390* (15); 2409 (27); 2423 (15); 2425 (15); 2427 (27); 2469* (4); 2490 (2a); 2492 (2a); 3079 (25b); S085 25D) 3191 (25b); 3228 (256) 3231 (25a); 3409 (25b); 3846 (25b); 3852 (25a); 3857 (25a); 3859 (25b); 3860 ud 3861 (25b); 3864 m 3962 (37); 4025 (2a); 4035* (29a); m oat | 0 ио err: 4120 (2a); 4184* 4474 (24); 4475* (27); 4530* (37); 4532* (37); 4534* (15); 4537* (37); 4543 (25b); 4544 (25b). A REVISION OF MONNINA SUBG. PTEROCARYA (POLYGALACEAE) IN NORTHWESTERN SOUTH AMERICA! Bente Eriksen? ABSTRACT Three subgenera are currently recognized in the genus Monnina, which consists of approximately 175 species. This paper treats the systematics of Monnina subg. Pterocarya in northwestern South America. The subgenus is a naturally coherent taxon characterized by dry, samaroid fruits. Jt consists of ca. 25 species, 11 of which occur in d. the area concerne The genus Monnina was established by Ruiz & Pavón (1798) and was dedicated to José Monino, the count of Floridablanca. It comprises fruticose species with drupaceous fruits, as well as herba- ceous ones with more or less winged samaras. De Candolle (1824) described the section Pterocarya to accommodate the species that have the fruit margin expanded into a membranaceous wing. This division, however, is artificial because it neglects a number of species with s is small or completely lacking. Chodat (1896) re- alized this and emended section Pterocarya DC. amaras, where the wing to comprise all annual or perennial herbs with sa- maras and gave the taxon subgeneric rank. The genus Monnina comprises ca. 175 species, ca. 25 of which belong to the subgenus Pterocarya. Species of Monnina subg. Pterocarya have nev- er been revised as a group. The works of Chodat (1894, 1896, 1909, 1915, 1934), Chodat & Wil- czek (1902), and Ferreyra (1990) mostly included various descriptions of new species, and those of Ferreyra (1946, 1953) and Grondona (1945) were entirely floristic. The aim of the present paper is to give an extensive treatment of the subgenus in . Here, the southern limit of is defined as the Peruvian border with Chile and Bolivia. This geographical boundary coincides with a natural phytogeograph- ical boundary (Molau, 1988). A revision of the remainder of the species is in preparation. ECOLOGY AND GEOGRAPHICAL DISTRIBUTION Plants of Monnina subg. Pterocarya are con- fined to semiarid habitats. The distribution is dis- junct, with a western and eastern territory in South America and an exclave in Mexico and the south- western United States. The western territory is found along the coast of the Pacific Ocean and in the mountains from southern Ecuador to central Chile, extending into western Bolivia and Argen- tina. The eastern territory covers northeastern Ar- gentina, southern Paraguay, Uruguay, and the dry parts of southeastern Brazil, principally east of Rio Paraná (Fig. 1). The map shows the narrow gap in the South American distribution area directed north-south through Argentina. The gap coincides with the extensive plain occupied by the Larrea Desert of Gran Chaco and adjacent areas (Walter, 1979) and is probably too dry to be favorable for Monnina. Grondona (1945) noticed the absence of Monnina in this particular area and established, as a consequence, their preference for mountainous habitats. He did not exclude the possibility that specimens might be collected there in the future. Since Grondona wrote the Flora, however, no spec- ' This paper is based upon research supported by Anna Ahrenbergs rüttningen, and Adlerbertska forskningsfonden, which are gratefully acknowledged. I deeply appreciate the comme Fond, Kungl. och Hvitfeldtska stipendienin- ents and criticisms of Ulf Molau, Lennart Andersson, and Gunnar Harling. I thank the curators of the following herbaria ;H, GOET of types), MO, Museum of Natural History in Góteborg for determining the insect found in the flowers the Flora RES base map no. | prepared by Н. R. American maps are Botany, State University of Utr LE, LY, M, MA (photos h Rypkema, Institute for Systematic ? Department of Systematic B University of Goteborg, Carl Skottsbergs Gata 22, S-413 19 Goteborg, Sweden. ANN. Missouni Bor. Garb. 80: 191-207. 1993. 192 Annals of the Missouri Botanical Garden © 200 400 600 800 1000km \ n " айыы \ HA A O 100 200 300 400 500 600 miles Prepared by Hendrik R. Rypkema — — loo. ad af [d » Рен = ё / M FIGURE 1. imen has been found in that area. The fact that different species occur on either side of the gap supports the assumption that the plain serves as a barrier. To the north, the two territories are delim- ited by areas with higher humidity or, in the Plan- alto do Mato Grosso, by periodical flooding. The coastal desert of Peru is, like the Gran Chaco, very dry and should be unsuitable for Monnina species, but here the “garua,” or fog blanket, caused by the cold sea current off the Peruvian coast, plays an important role. Even though the precipitation is not measurable, a considerable amount of water is yielded through fog (Walter, 1979). Only one species occurs north of the equator, viz M. wrightii in Mexico and the southwestern United States, but the same species is abundant in Bolivia and Argentina and has probably been in- troduced to Mexico. The semiarid habitats to which the species are confined are usually devoid of dense and tall veg- Distribution of Monnina subg. Pterocarya. etation. In some areas of southern Ecuador the natural vegetation is mainly dense shrubland, and competition for light does seem to be a limiting factor. The Monnina species here are always con- fined to open grassland, natural or artificial. Flowering begins near the end of the rainy sea- son and continues 3-9 months. In habitats with sporadic rainfall, the flowering season is variable and flowering occurs whenever humidity allows it. The altitudinal range of Monnina is rather wide, sometimes even within species. In most cases, how- ever, two species sharing latitude are usually sep- arated altitudinally with little or no overlap. the different species are rarely sympatric, it is not surprising that no specimens have been found that could be interpreted as hybrids. Besides, the pollination mechanism in Monnina is likely to pre- vent interspecific pollination. The flowers are pol- linated by bumblebees. On landing, the insect press- es the released pollen through an opening near the Volume 80, Number 1 1993 Eriksen 193 Monnina subg. Pterocarya keel ^ FIGURE 2. the various parts were measure Floral structures showing the way in which d. Outer sepal, wing, and keel given as 1 x 2. Style length 1. Petal-stamen sheath given as 1/2 wide, 3/4 high. asl = anterior stylar lobe. apex of the keel and at the same time foreign pollen is deposited on the stigma. Because the spe- cies differ greatly in flower size and stigma position, only a flower of the same species gets pollinated, even if the insect visits another species during a foraging bout. MATERIALS AND METHODS This study is based on herbarium specimens, pickled material, and fieldwork. Field studies were carried out in 1987-1988 in Ecuador and northern Peru. Five ipd of Monnina subg. Pterocarya were seen in the MEASUREMENTS Measurements were taken on the leaves of the main axis (leaves on the branches are generally smaller). The measurements are given in the de- scriptions as length X width. For the inflorescence, the main one was measured (inflorescences on the branches are shorter). Inflorescences that have started to fruit were selected, since the inflores- cence axis elongates during flowering. e pedicel was measured when carrying an open flower. The flower parts measured were taken from the bud just above the youngest unfolded flower. This choice ensures intact flower parts and still gives a relatively good picture of the flower size. The way in which the various flower parts were measured is shown in Figure 2. In the descriptions, mea- surements for the petal-stamen sheath are given as “width at base" /**width across petal lobes" and "length of filament sheath" /“total length including petal lobes." Fruit size is given for mature fruits. Vegetative parts of dry material were measured. My experiments have shown that leaves of Mon- nina shrink approximately 10% in the drying pro- cess. The descriptions have not been corrected for shrinkage. Flowers are measured on rehydrated material, except for a few specimens of which I have pickled material. TERMS The nodal glands present in some species of Monnina are traditionally, and so also here, called stipules, although they are not homologous with stipules. The construction of the flower in Monnina 15 quite extraordinary, and this type of papilionaceous ower is unique for the Polygalaceae. Figure 2 shows the floral parts and the terms used for each. SYSTEMATICS Monnina subg. но (DC.) Chodat, Bull. erb. Boissier 4: 251. 1896. Monnina sect. Pterocarya DC., RN 1: 340. 1824. T Monnina pterocarpa Ruiz & Pavon, E type here designated. Annual or perennial herbs. Leaves alternate, short-petiolate to subsessile, with or without stip- les. Stems and branches terete, covered by uni- cellular hairs, either white, conical and curved or compressed and straight, or yellowish with knoblike thickenings on their outer side and = straight. Laminae simple, pinnately veined, entire or rarely undulate. Flowers arranged in simple racemes, each subtended by a bract and two bracteoles, irregular; outer sepals 3, small, herbaceous, free or rarely the lower two shortly connate at base; inner sepals (wings) 2, large, lateral, petaloid; petals 3, the lower (keel) carinate, the upper two united in their lower part with the filament sheath (see below), but free from each other, forming a conduplicate petal- stamen sheath; stamens diadelphous, (4-)6-8, the filaments fused for most or all of their length into a filament sheath to which lateral margins the upper petals are adnate; style compressed, geniculate to gently curved, apically unequally cleft, the pos- terior lobe with a sessile or = stalked capitate or discoid stigma, the anterior lobe sterile, often form- ing a tooth. Ovary l- or 2-locular, subtended by a glandular disc enlarged toward the adaxial side into a knoblike process. Fruit a regularly or irreg- ularly winged, or occasionally wingless samara. 194 Annals of the Missouri Botanical Garden KEv TO THE SPECIES OF MONNINA SUBC. la. e absen Sta 3a. Fruit villous. 4a Perennial, the leaves linear, ericoid PTEROCARYA IN NORTHWESTERN SOUTH AMERICA amens 8, the filaments free in the upper part, or anthers seemingly sessile. l. M. fosbergii 4b. Annual, most leaves rhomboid, not er 5a. Stem and branches erect, wings a icd bright blue, the keel yellow apically, anterior stylar dde a blunt tooth .............. . leptostachya 5b. MR and branches ascending, wings pale blue, the keel pale yellow throughout, anterior oth 3. М. tylar lobe а p 3b. Fruit glabrous or pubesc ments free in d = polygonoides M. ramosa 4. 6b. а adnate throughout, p anthers seemingly sessile on the petal-stamen sheath. Та. Leaves linear, style straight or bent 30°, posterior stylar lobe with a spherical stigma 9. M. apica . macbridei picaity Tb. Leaves narrowly obovate to narrowly rhomboid, the upper leaves sometimes linear, style bent 80—90°, posterior stylar lobe carrying a bilobed, discoid stigma at base ... 6. M. bue пече hf sal nal f. lth 2b. Stamens 6, sometimes 4 in flowers on branches; wo outer stamens of each fascicle connate throughout, the third Te in upper part or reduced (Fig. 8B), the fruits crescent-shaped filaments of the t А и 8b. The filaments of the three (two) stamens of each fascicle all connate throughout (Fig. ВЕ), th fruits rotund. 9a. Upper leaves m lanceolate to linear, 0.8-3.7 cm long, obtuse, the inflorescence d 7. M. EN triangular in outline us ae n outline m long, acute, the inflorescence many-flowered, ^. | ee „М. ‘filifolia M. pterocarpa M. herbacea lb. Stipules pres 10a. Tall e usually > 0.5 m, the older parts of stem and branches glaucous, the sc robust, the stigma a sessile or short-stalked, bilobed disc, the fruit always with a broad, entire 10b. Small plant, usually < 0.5 m, the older parts of stem and branches green or straw yello ow, flowers delicate, the stigma resembling a hammer-head (Figs. 9K, 10J), the fruit usually irregularly winged, occasionally wingless or with a narrow, entire wing Distribution. Dry lowland and mountainous ovate, 1.8-2.0 x 0.8-1.0 mm, arched, the lower areas from central Ecuador to central Chile, and from central Brazil to Uruguay; Mexico and the southwestern United States (Fig. 1). — Monnina fosbergii Ferreyra, Phytologia 69: 356. 1990. TYPE: Ecuador. Loja: S of Ona, above Tablón de Ona, 2,740 m, 21 Feb. 1945, Fosberg 23204 (holotype, US; isotypes, NY, P). Figure 3 Perennial herb, decumbent, to 70 cm across, 18-28(-50 fide Barclay & Juajibioy) cm tall, the ascendent branches to 40 cm long crowning the short, woody stem. Branches rooting at nodes, te- rete, 1 mm diam., furrowed by decurrent leaves, white-pubescent, the hairs curved. Leaves exstip- ulate; petiole very short or absent; blade narrowly lanceolate to linear, ericoid, coriaceous, 4-13 x 1—1.5 mm, acute at apex, cuneate at base, white- pubescent throughout. Inflorescence few-flowered, narrowly triangular in outline, 11-24 cm long; pedicel 0.5 mm, hirsute; bracts lanceolate, acute, 2.0-3.0 x 0.7-0.8 mm, abaxially hirsute, blue with a green, ciliate margin. Flowers all chasmoga- mous; outer sepals slightly unequal, the upper one narrowly ovate, 1.2-2.0 x 0.7-0.8 mm, all ciliate and blue; wings obovate, 3 X 2 mm, truncate at base, the upper margin ciliate, blue; keel З х mm, truncate at base, glabrous, blue with a yellow, emarginate and 2-lobed apex; petal-stamen sheath 2-3/4 mm wide, 1.5-2/2-3 mm high, the petals hirsute adaxially and abaxially on the lower half, the lobes blue, otherwise greenish as the filament sheath; stamens 8, the upper part of the filaments free, glabrous, 1.2-1.5 mm long including the an- ther, the anther to filament ratio 0.3-0.5; style 1.5-2 mm long, glabrous, bent 45? approximately V4 from base; anterior stylar lobe a blunt tooth not exceeding 0.1-0.2 mm in length, the posterior lobe with a sessile or very short-stalked, spherical stigma (Fig. 10A). Fruit 2-locular, narrowly obpyriform with a basal tooth dorsally, 2 x 4 mm when mature, , probably green. Distribution. Probably endemic to the Ona area in Ecuador, ca. 2,800 m (Fig. 6B). villous, wingless ecimens ii ide CUADOR. AZUAY/LOJA: be- tween Ona and Rancho Ovejero, 2,800 m, 1-2 Aug. 1959, Barclay & Juajibioy 8437 (MO). LoJA: road Ona- Volume 80, Number 1 1993 Eriksen 195 Monnina subg. Pterocarya FIGURE 3. vidual. —B. Leaf. — Keel. — К. Petal-stamen diris nh. Fruit & Juajibioy 8437 (MO). Scales: A = 2 cm C-H = 1 mm. r sepal. =G, B E —H. Saraguro, ca. km 9, 2,800 m, 13 Feb. 1991, Mllgaard 98803 (AAU, GB, оса, ОСМЕ) Monnina fosbergii closely resembles the Chil- ean species M. linearifolia, but the floral parts are different. Monnina fosbergii has eight stamens, a sessile stigma, and a 2-locular, villous, wingless samara, whereas M. linearifolia has six stamens, the stigma at the tip of a very long stalk, and a l -locular, glabrous, (1990), in the original description, compared M. fosbergii with M. filifolia, to which species it is only distantly related. Monnina filifolia has, as the epithet suggests, very narrow leaves, but they are not of the ericoid type. The two species differ winged samara. Ferreyra in habit and in all essential characters. Ferreyra (1990) characterized M. fosbergii as an erect herb. It is actually decumbent and woody at the base. Perennial species of subgenus Ptero- carya are common from southern Peru (M. ra- mosa) and southward and in the eastern distribution area, whereas in the north the only perennial rep- resentative is M. fosbergii. Ferreyra also stated вере lik —A. Wines indi- —D. Wi E. . A-G, Fosberg 23204 (isotype, NY). H, Sp ‚В = that the ovary is pubescent, which is true only in the broad sense of the word. More precisely, the ovary, and later the fruit, is villous. 2. Monnina yy notas Benth., Pl. Hartw. 125. 1845. TYPE: Ecuador, Loja: near Loja, Hartweg “е (lectotype, selected here, К; isolectotypes, NY, P, W). Figure 4. Annual herb, 15-50 cm, rarely taller. Stem erect, 0.5-3(-4) mm diam., terete or slightly furrowed, unbranched or branched distally, white-pubescent, the hairs curved. Leaves exstipulate; petiole 1-2(- 3) mm long, pubescent; laminae on the main axis changing in shape and size from base to apex of the stem; lowermost blades obovate to narrowly obovate, 0x0 middle ones rhomboid to narrowly rhomboid, 1.5- 6.5 х and uppermost blades lanceolate to narrowly lan- ceolate, 0.8–5.0 0.2-1.0 cm, apex; all laminae chartaceous, cuneate at base, cm, acute at apex, 0.5-2.5 cm, acute to acuminate at apex, acuminate at glabrous except on veins and margin where they are pubescent (in young leaves pubescent through- out). Inflorescence many-flowered, very narrowly triangular in outline, 8-40 cm long; pedicel 0.5 mm long; bracts lanceolate, acute, 1.0-2.0 x 0.3- 0.5 mm, less than twice the length of the bud, green, blue, or both colors, ciliate. Flowers all chas- mogamous; outer sepals ovate, equal in size or the upper one slightly larger, 1.0-2.0 mm long, ciliate, blue along the margin, otherwise green; wings ob- ovate, 2.5-4 x 2-3 mm, truncate and ciliate at base, bright blue; keel 2.5-4 x 1.5-2 mm, trun- cate at base, glabrous, bright blue with a yellow, 3-dentate apex; petal-stamen sheath 2-4/ 4-6 mm wide, 1.5-3/1.5-3 mm tall, the petals hirsute adaxially and at the base abaxially, the lobes bright blue, otherwise greenish as the filament sheath; stamens 8, the upper part of filaments free, gla- brous, 0.75-1.25 mm long including the anther, the anther to filament ratio 0:3-0:5; style 2-3 mm long, glabrous, bent 60—90° just above the middle; anterior stylar lobe a blunt tooth, not exceeding 0.1 mm in length, the posterior lobe with a sessile, elongated stigma (Fig. 10B). Fruit 2-locular, ovoid 1.5-2.5 x 2-3.5 mm, wingless, green, often with purple spots at base or or obpyriform, villous, a purple line dorsally. Distribution. Central Ecuador to the Eastern Cordillera of northern Peru. Scattered to locally frequent as a weed in dry scrub, on roadbanks, and among grass in cultivated areas, 1,200-3,000 m (Fig. 6A). It appears after the rainy season and flowers from January to September. 196 Annals of the Missouri Botanical Garden In Peru, M. leptostachya has only been col- lected in Amazonas in the Eastern Cordillera. A similar distribution is shown by Calceolaria comosa Pennell subsp. comosa (Molau, 1988). The area between the Peruvian and Ecuadorean localities is poorly collected, and it is uncertain whether the distribution is continuous or whether the Peruvian population constitutes an exclave of a disjunct dis- tribution. Specimens examined. | ECUADOR. AZUAY: along the Río Tarqui, 4-18 km 5 of Cuenca, 2,500-2, 700 m, 26 Feb. 1945, Camp E-1836 (NY); near the village of Girón, 99 n Girón and Santa isabel. 7 June 1979, Løjtnant & Molau 14174 (AAU); 0 ч 5 of Gualaceo оп road to = 5 ees КТЕ [е Mihuir, EN of UC). CHIMBORAZO: canyon of Río Chanchán near Huigra, 1,200-1,350 m, 7- 14 May 1945, Camp E-2995 (NY), aed E-3058 (NY); © Riobamba El Truinfo, between n Juan and Llimbe, 00 m, 25 Feb. 1987, Freire end 598 (А AU); de 3 000 m, 28 May 1942, a 32 i 650 (MO, WIS); La Arachis, 2, 080 m, 28 Маг. 1946, Espinosa 14 (NY); W slopes of Cerro Villonaco, 1,800- 2,100 m, 9 Apr. 1974, Harling & Andersson 13302 (GB); Vilceabamba- Yangana road, -1,800 m, 15 Apr. 1974, Harling & Andersson 135; 50 (CB), 1,900 m, 21 Apr. 1980, Harling & Andersson 18446 (GB); road Loja-Zaruma, between the San Pedro and Las Chin- chas, 1,900-2,200 m, 30 Apr. 1974, Harling & An- dersson 14094 (GB); 5 km E of Celica on road to 18181 (GB); road Catamayo (La T 10 km N of Gonzanamá, 2,100 m, 24 Apr. 1980, Harling & Andersson 18625 (GB); 6 km S of Vilcabamba, 2,100 m, 9 Feb. 1982, Harling et al. 20454 (GB); between Chuquiribamba and Taquil, 2,200 m, 14 Feb. 1982, Har- ling et al. 20719 (GB); 4-6 km N of Yangana on road to Vilcabamba, 1,800-1,900 m, 5 Feb. 1985, Harling & Andersson 21636 (GB); between Loja and San Lucas, 600 m, 6 Sep. 1923, Hitchcock 21450 (US); km 66 on Pan American Highway N of Loja, 2,400 m З May 1973, Holm-Nielsen et al. 4751 (AAU, NY, US ); 27-29 km from Porto Velo on road to ree Е it Cam- pamento “Santa Ana," 1,200 , Molau А , СВ, ОСА, oc No PERU AMAZONAS: Prov. Bongara, 11 km E of Pomacocha, 1,950 m, 9 Feb. 1984, Gentry & Smith 45209 (MO); between Rio Utcubamba and Pomacocha, 1,700 m, 29 Jan. 1964, Hutchison & Wright 3857 ( Chachapoyas, Tingo, near Chachapoyas, 1,900 m, 12 Apr. 1950, Ferreyra 7095 (Е); W slopes of Cerro Calla- FIGURE 4. Monnina leptostachya. — А. m individual. —B. Habit. — Lower outer sepal. —E. bis ne. —F. stamen sheath. —H. Gyn —]. Fru Eriksen et al. 2941 (СВ). В. 1 Molau = Eriksen m .1= t. А, Molau, (СВ). Scales: A = 3 cm. С-Н = 2 1 Calla, 27 km E of Balsas on road to Leimebamba, 2,250 m, 23 Mar. 1988, Molau & Eriksen 3385 (GB); d a E of Chachapoyas, 2,400-2,500 m, 21 June 1948, nell 15582 (PH); Utcubamba S of Chachapoyas, 1'700-. 1,750 т, 26 Јипе 1948, Гене 15590 (РН); 1 km SW of Chachapoyas, 2,300 m, 22 May 1962, Wurdack 473 (F, GH, UC, US) Monnina leptostachya is closely allied to M. polygonoides but is easily distinguished by its erect habit and blunt anterior stylar lobe (see Fig. 10B and C and discussion under M. polygonoides). Ferreyra (1953) synonymized M. leptostachya under M. herbacea, based on type photographs and fragments. The two species have, in fact, very little in common, and the mistake would not have been made with the original material at hand. ne record states that the vernacular name 15 ro deed which r means small (female) "Igui- In Quechua, is the common name for all species of Monnina. 3. Monnina polygonoides Chodat, Bull. Soc. Bot. Genéve sér. 2, 25: 200. 1934. TYPE: Pour Huánuco: Prov. Huánuco, Ambo, 2,100 m, 5 Apr. 1923, Macbride 3181 (lectotype, Volume 80, Number 1 1993 Eriksen 197 Monnina subg. Pterocarya FIGURE 5. flowering individual. — B. 3379 (СВ). Scales: А = 1 с Monnina d e e — А. Habit and dee канш & Eriksen selected by Ferreyra (1946), F; isolectotypes, GH, US). Figure 5. Annual herb, ascendent, 5-30 cm tall, the branches to 40 cm or more. Stem 0.5-3 mm diam., usually much branched from the base, the stem and branches terete, furrowed by decurrent leaves, white-pubescent, the hairs curved. Leaves exstip- ulate; petiole 1-1.5 mm long, pubescent; laminae changing in shape and size from base to apex of the branches; lowermost blades narrowly obovate, 0.3-1.5 x 0.1-0.5 cm, obtuse at apex, cuneate at base, middle ones narrowly rhomboid, 0.7-3.2 X 0.3-1.4 cm, acute at apex, cuneate at base, the uppermost leaves lanceolate to linear, 0.9-2.3 7 ст, acute at apex, shortly attenuate at base; all laminae chartaceous, minutely mucro- nate, pubescent on veins and margin, otherwise glabrous (in young leaves pubescent throughout). Inflorescence few-flowered, triangular in outline, 1.5-20(-30) cm; pedicel 0.5 mm; bracts narrowly lanceolate to linear, 1.5 x 0.1-0.2 mm, acute, never exceeding twice the length of the bud, bluish green, ciliate, deciduous. Flowers all chasmoga- mous; outer sepals usually slightly unequal, the upper one narrowly ovate to lanceolate, 1.5 mm long, 3-nerved, the lower ones ovate, 1.0-1.5 mm long, 1-nerved, all of them ciliate, greenish blue 2-2.5 mm, truncate at base, glabrous, pale blue; keel 2.5-3.5 2 mm, truncate at base, glabrous, pale yellow, 3-dentate at apex; petal-stamen sheat -3/4- 5 mm wide, 1.5-2/2-3 mm long, the Баје hirsute adaxially and at base abaxially, the lobes pale blue, otherwise yellowish as the filament sheath; stamens 8, the upper part of filaments free, glabrous, 1- 1.5 mm long including the anther, the anther to filament ratio 0.3-0.5; style 1.5-2 mm long, gla- brous, bent 45? just above the middle; anterior stylar lobe a pointed tooth, 0.2-0.25 mm long, the posterior lobe with a sessile, rounded stigma (Fig. 100). Fruit 2-locular, narrowly obpyriform with a dorsal gibba, 2 х 4 mm, villous, wingless, green, the gibba purple Distribution. rocky slopes in dry, mountainous areas, 2,000- 2,850 m (Fig. 6C). It has been collected flowering from February to May, i.e., from the end of the rainy season and a couple of months onward. to pale blue; wings obovate, 2.5-3.5 x Northern to central Peru. Steep, Specimens examined. PERU. AMAZONAS: Prov. Bon- p road Moyobamba-Chachapoyas, 2,250 m, 13 Apr. 984, Croat 58232 (МО); Prov. Chachapoyas, 9 km Er. and W of Chachapoyas on road to Caclic, 2,000 m, 25 Mar. 1964, Hutchison & Bennett 4516 (F, UC). ANCASH: Prov. Bolognesi, Conay, 6 km below Chiquián, 2,600 m, 12 May 1950, Ferreyra 7381 (UC, mixed coll.); Prov. er "T hillsides above Rio Santo е from Mancos, 10 Apr. 1 970, Earle Smith & Blas 4 olau & Eriksen món E of Celendin, | 00 m, 15 Арг. 1948, Pennell 15181 (GH, PH). HUANUCO: Prov. ; 198 14551 (F, b Prov Huánuco, o, 2,100 n 22 Арг. 1946. Woy iius 34234 (F, LIBERTAD: a antiago thuco, near Santiago de Chuco, 2,8 , 13 June 1984, dude m et al. 11773 (MO). РАЗСО Prov asco, Huar 900 m, 3 Apr. 23, Mac 3116 E "GH. MICH, US). De- partment шыш sine loco, Weberbauer 6235 (F) Monnina polygonoides is closely related to M. leptostachya. characterized by 2-locular, villous fruits. However, Monnina polygonoides is ascendent, has pale blue These two species are primarily flowers, and a pointed anterior stylar lobe (Fig. 198 Annals of the Missouri Botanical Garden 10C), whereas M. leptostachya is erect, has dark blue flowers, and a blunt anterior stylar lobe. The two species may be confused when dried, but they are easily distinguished in the field because of habit and flower color. Ferreyra (1946) synonymized M. polygonoides under M. herbacea and thereby also related it to M. leptostachya (see above under this species). Monnina polygonoides is, as mentioned above, closely related to M. leptostachya, but it has little in common with M. herbacea. 4. Monnina ramosa I. M. Johnston, Contr. Gray Herb. 70: 77. 1924. TYPE: Peru. Are- quipa: Prov. Arequipa, S slope of Chachani Mountain N of Arequipa, 3,355 m, Mar. 1920, Hinkley & Hinkley 13 (holotype, GH; is type, Perennial herb, 15-25 cm tall. Stem 1.5-2.5 mm diam., the branches spreading, white-pubescent, the hairs curved. Leaves exstipulate; petiole 0.5-1 mm long, pubescent; blade lorate to linear, 8-30 x 1-2 mm, chartaceous, acute at apex, cuneate at base, terete, much branched from the base, white-pubescent throughout. Inflorescence many- flowered, narrowly triangular to very narrowly tri- angular in outline, 5-18 cm long; pedicel 0.5 mm; bracts narrowly ovate to ovate, acute 2 x 0.7-1 mm, bluish, pilose. Flowers all chasmoga- mous; outer sepals equal, narrowly to broadly 1.2-2.0 x 0.6-1.0 mm, bluish, ciliate; wings obovate, 3.5 X 2 mm, cuneate at base, blue, ovate, glabrous; keel 3 x 1.5-2.5 mm, truncate at base, yellow at least apically, glabrous, 3-lobed at apex, the central lobe emarginate; petal-stamen sheath - mm wide, 2/3 mm tall, hirsute adax- ially at sinus between the petal lobe and the sta- minal sheath and sparsely on the lower half of the petal; stamens 8, the upper part of the filaments free, glabrous, 1 mm long including the anther, the anther to filament ratio 1.0; style 1.5 mm long, glabrous, bent 80? 24 from base; anterior stylar lobe a long (0.2 mm), pointed tooth, the posterior lobe a 0.5-mm-long stalk with a rounded stigma at the tip (Fig. 10D). Fruit unilocular, compresed, + elliptic, 4 х З mm, densely strigose with a glabrous wing. Distribution. Only known from southernmost Peru, where it occurs in sandy habitats above 2,3 m (Fig. 6E). It may be endemic to the Arequipa area or it may be present further south reaching into the adjacent mountains of northern Chile (al- though it has not yet been collected there). FIGURE 6. Distributions of the species of Monnina — C. Monnina polygonoides (dots) and M. wrightii (star). —D. Monnina macrostac un (dots) and M. macbridei (star). — ] a (dots), M. ramosa (star), and M. filifolia (circle). Specimen examined. PERU. AREQUIPA: Prov. a Volcán Misti, 2,300 m, July-Aug. 1921, Pies 1019 (U). Monnina ramosa has, like M. fosbergii from Ecuador, a general appearance similar to that of M. linearifolia from Chile, i.e., а decumbent, woody perennial with small, linear leaves. The most ob- vious distinction between M. linearifolia and M. ramosa is that the former has six stamens not eight. Monnina fosbergii, like M. ramosa, has eight stamens with the upper part of the filaments free, but all other flower parts are different. The fruit of M. fosbergii is 2-locular while that of M. ramosa is 1-locular. Monnina ramosa grows sym- patrically with M. macbridei, but the latter species is easily recognized because it is annual and has seemingly sessile anthers. The exact colors of the flower parts are uncer- tain. The colors fade in the drying process, and the label data are imperfect. Volume 80, Number 1 1993 Eriksen 199 Monnina subg. Pterocarya — А. Habit and abit of Loma type. —C. —D. Lower outer sepal. —E. Wing. Petal-stamen sheath. ve B, = ] mm IGURE 7. Monnina macrostachya. flowering individuals. Upper outer sepal. — С. Gynoecium. —H. . A, C-I, Molau & Pond 3515 (G Grant тало (СН). Scales: А = . C- m Monnina macbridei Chodat, Bull. Soc. Bot. Genève sér. 2, 25: 199. 1934. TYPE: Peru. Arequipa: Prov. Arequipa, Arequipa, 2,500- 2,600 m, 7-16 Apr. 1925, Pennell 13167 (lectotype, selected by Ferreyra (1946), F; fragments, USM; isolectotypes, GH, LE, NY, PH, S, U Annual herb, 5-35 cm tall. Stem erect, 0.5-2 mm diam., terete or slightly furrowed by decurrent leaves, unbranched or sparsely branched distally, white-pubescent, the hairs curved. Leaves exstip- ulate; petiole 0.5-1 mm long, pubescent; laminae on the main axis changing in shape and size from base to apex of the stem; lowermost blades spathu- late, 1.4-3.0 x 0.2-0.5 cm, upper ones lorate to linear, 1.7-4.2 x «0.1-0.4 mm, acute at apex, cuneate at base; all laminae chartaceous, white- pubescent throughout. Inflorescence many-flow- ered, very narrowly triangular in outline, 2-25 cm long; pedicel 0.5 mm long; bracts lanceolate, 1.5- 2.0 x 0.5 mm, acuminate, probably green with bluish tinge, glabrous. Flowers all chasmogamous; outer sepals unequal in shape and size, pale blue, glabrous; the upper sepal narrowly ovate, 1.5 x 0.7 mm, the lower ones narrowly ovate to lanceo- late, 1.0-1.5 x 0.5 mm, all acute at apex; wings obovate to narrowly obovate, 1.5- 3 mm, cuneate at base, pale blue, glabrous; keel 1.5 x 2.5-3.5 mm, cuneate at base, glabrous, profoundly emarginate and 2-lobed at apex, probably yellow throughout; petal-stamen sheath 1-2/2-3 mm wide, 2-2.5 x 2.5-4 mm tall, the petal sheath hirsute adaxially, the lobes probably pale blue, oth- erwise greenish yellow as the filament sheath; sta- mens 8, the anther 0.1–0.2 mm long, seemingly sessile or the two central anthers occasionally with а 0.1-mm-long free filament; style 1-1.5 mm long, glabrous, straight or bent 30? 25 up; anterior stylar lobe an upright pointed tooth, 0.1 mm long, the posterior lobe a 0.4—0.6-mm-long stalk, bent 90° in relation to the style, with a rounded stigma at the tip (Fig. 1OE). Fruit unilocular, broadly elliptic to circular in outline, 3- -4 mm, glabrous or pubescent, winged, green, possibly purple-spot- Distribution. From the present collections, M. macbridei seems endemic to the Arequipa area in southern Peru, where it inhabits sandy sites on dry slopes, ca. 2,000-3,400 m (Fig. 6D). As argued earlier under M. ramosa, which occurs sympat- rically with this species, it is plausible that the distribution extends into the mountains of northern Chile. Specimens examined. PERU. AREQUIPA: Prov. Are- quipa, Arequipa, 21 Aug. 1925, Cockerell s.n. (US); Chachani mountains N of Arequipa, 3,355 m, Mar. 1920, ien & Hinkley 18 (GH); Tingo, 2,100-2,300 m, 8 Apr. 1925, Pennell 13110 (syntypes of M. macbridei, GH, LE, US); SW side of Rio Chili, 4 km N of Arequipa, 2,500 m, 8 May 1977, Solomon 2780 (MO). Monnina macbridei is close to M. macro- stachya in some aspects. Both have a petal-stamen sheath with eight seemingly sessile anthers, a char- acter that is unique for these two species. The habit of M. macbridei is very similar to that of M. macrostachya. 'The two species differ in other characters, e.g., the posterior stylar lobe, the stig- ma, and the apex of the keel 6. Monnina macrostachya Ruiz & Pavon, Syst. Veg. Fl. Peruv. Chil. 1: 173. 1798. TYPE: Peru. Lima: Provinces of Cercado (Lima) and Chancay, Ruíz & Pavón s.n. (lectotype, selected here, MA). Figure 7. Me lanceolata (Poir.) DC., Prodr. 1: 339. 1824. onym: Polygala lanceolata Poir., Encycl. 5: 498. 1804. TYPE: Peru. Sine loco, asis s.n. (ho- lotype, P). 200 Annals of the Missouri Botanical Garden Monnina macrostachya var. pumila А. Gray, U.S. Ex- ploring gd 15, Botany, Phanerogamia part I: 107. 1854. TYPE: Peru. Lima: Prov. Canta, above Obrejilo Wilkes expedition s.n. (lectotype, selected here, GH; isolectotype, Monnina darc Chodat, Bot. ин Mae 42: 102. 1909. TYPE: Peru. Arequip v. Islay, Mollendo, 20-100. m, 1906, Weberbauer. 1505 (lectotype, selected here, Monnina weberbaueri var. elongata Chodat, Bot. Jahrb. 42: 102. 1909 E: Sub Ancash: Prov. Cajatambo, below Ocros, 2,300-2,400 m, Weber- bauer 2724 (B probably dde NEOTYPE: Peru. Ancash: Prov. Bolognesi, between Colcas and Ca- jacay, 2,400 m, 8 Apr. 1988, Molau & Eriksen 3512 Моптта weberba aueri var. pachyantha Chodat, Bot. Jahrb. Syst. 42: 102. 1909. TYPE: Peru. Lima: Prov Huarochiri, railroad between Lima a Matucana, 3,000 m, Weberbauer 185 (B probably destroyed). NEOTYPE: Peru. Lima: Prov. Huarochiri, Matucana, 2,900 m, 9 Apr. 1939, Goodspeed 11332 (GH; isoneotypes, NA, UC). Monnina weberbaueri var. maxima Chodat, Bot. Jahrb. 9. TYPE: а с : Prov. Huaylas, between Samanc Caraz, 3,100 m, Weberbauer 3125 lectotype, selected Dos MOL). Monnina arenicola Ferreyra, J. A Arbor. 27: 137. 1 TYPE: Peru. Arequipa: qa slay, S of Mo- ll lendo, 17 Nov. 1935, Mexia 4175 (holotype, GH; fragments, USM; isotypes, M, UC). Annual herb, 3.5-46 cm tall. Stem erect, 0.5- 3 mm diam., terete, slightly furrowed by decurrent leaves, unbranched to much branched all along the main axis, the branches erect or ascending, white- pubescent, the hairs curved. Leaves exstipulate; petiole 1-2 mm long, pubescent; laminae on the main axis changing in shape from base to apex of the stem; lowermost blades narrowly obovate, 0.4— 3.6 x 0.2-1.8 cm, obtuse to acute at apex, middle ones narrowly rhomboid, 1.0-4.2 x 0.4-2.1 ст, acute at apex, and uppermost blades lanceolate to linear, 0.8-4.2 x 0.1-1.2 cm, acute at apex; all laminae chartaceous, minutely mucronate, cuneate at base, pubescent on veins, margin, and sometimes the upper surface, otherwise glabrous (in young leaves pubescent throughout). Inflorescence many- flowered, 1.5-31 cm long, dense at apex, rather abruptly becoming looser and more expanded; ped- icel 1 mm long; bracts lanceolate, acuminate, 1.0— 2.0 mm long, ciliate, blue. Flowers all chasmoga- mous; outer sepals usually unequal in shape and size, blue; upper sepal ovate to narrowly ovate, -3 mm long, boat-shaped, acuminate, glabrous or ciliate along margin and midrib abaxially, lower 1-2.5 mm long, acuminate, glabrous or ciliate along the mar- sepals narrowly ovate to lanceolate, gin; wings obovate, 1.5-3 x 3-4.5 mm, cuneate at base, glabrous or ciliate basally along the abaxial margin, blue; keel 1.5-2 x 3-5 mm, 3-dentate at apex, cuneate and ciliate at base, usually sparse- ly pilose along the midrib adaxially, blue, the apex or only the central tooth yellow; petal-stamen sheath 1.5-3/3-5 mm wide, 2-4.5/2.5-5 mm long, the etals hirsute adaxially and at least at base abax- ially, the lobes blue, otherwise greenish as the fil- ament sheath; stamens 8, the anther 0.2-0.7 mm long, seemingly sessile; style 1.5-3 mm long, gla- brous, bent 80—90°, rarely less, ca. уз from apex; anterior stylar lobe a short (ca. 0.1 mm) pointed tooth, the posterior lobe 0.5-0.8 mm long, at base carrying a bilobed, discoid stigma, the apex a point- ed, hirsute tooth (Fig. 102). Fruit unilocular, 2 х m, glabrous or pubescent, in dry condition lacunate, usually winged but often irregularly so, green, often suffused with purple. Reddish flowers occur in some specimens. In the population I saw, ca. 10% of the individuals had red flowers. Distribution. mas near the coast or on rocky slopes in the moun- tains, 0-3,500 m (Fig. 6D). Peru. In sand dunes and in Lo- PERU. ANCASH: Prov. Bolo- ecimens examined 1950, Ferreyra 7316 (US); Cony : km below Chiquián, 2,600 m, 12 May 1950, Pieds ra 7381 (UC, mixed coll.); below Chiquian, 3,200-3,3 m, 22 May 1948, Pennell 15412, 15413 (PH); rov. Carhuaz, N of Chancos, 3,000-3,100 m, 11 May 1948, Га! 15314 (РН); pie Huaylas, trail to cave across io Santo from Man 9 Apr. 1970, Earle Smith & Blas 4902 vs mixed E ). AREQUIPA: Prov. Caraveli, N of Pta. Chala, 18 1957, па, 2648 (0); Prov. Islay, 6 km E of с 580 m, 17 Nov. 1986, Dillon et al 4830 (US); 7 km NE of Matarani, 350 m 1974, Evaristo López 51 (MO); Mollendo, 16 Oct. 1925, ад 3551 (GH), 600 m, 6 July 1946, sie & Pear. 5 (PH), 175 m, 30 Sep. 1938, rth & Mons 15762 (GH, UC, US). AYACUCHO: de rov. Lucanas, between Nazca and Puquio, 1,500-1,600 m, 19 Mar. 1949, Ferreyra 5452 (F). Par Prov. Contumazá, La Herilla, Guzmango, 3,300 m Apr. 1967, Sagástegui et al. 6426 (GH, MO, US); El a 2,800 m, 14 June 1983, Sagdstegui & Lépez 10657 MO, US); around Guzmango, 2,500 m, 22 May 1978, d & Mostacero 9100 (F, MO); 2,300 m, 29 1982, iE AC et al. 10331 (MO). HUANUCO: d Huánuco, Mito, 2,750 m, 8-22 July 1922, Macbride & D DUE 1548 (F). LA LIBERTAD: Prov. Otuzco, Salpo-Chanchacap, 2,800 m, 24 May 1984, Sagástegui et al. 11653 (MO, US). Lima: Prov. Caja- tambo, Ambar, 2,010 m, 16 Apr. 1939, Stork 11451 GH, UC); Piar Canta, near Yaso on the Lima- Canta road, 1,600-1,800 m, 8 May 1950, Ferreyra 7256 (US); summit between Huamantanga and Puruchuco, jra thews 461 (GH); Canta, 2,700-2,900 m, 11-19 Jur 1925, Pennell 14343 (F, GH, PH); along Rio Chillón. above Obrajillo, 2,700-3,000 m, 13-23 June 1925, Pennell 14372 (F, GH, PH), 14373 (PH), 2,300-2,400 m, 15-17 June 1925, Pennell 14436 (F, GH, LE, PH); ~ = Volume 80, Number 1 1993 Eriksen Monnina subg. Pterocarya Prov. epo Lomas de pone o, S of Chancay, 300 8, Stork & Vargas 9351 (GH, MO, NA, Sep. 1938, Stork & Horton 9143 (GH, UC); Rio Blanco, 3,500 m, 25 Mar. 1952, Hutchison 567 (UC); above Sta. Eulalia, 1,300 m, 3 Mar. 1942, Goodspeed 33016 (UC); Prov. Lima, S of Lima, 31 July 1941, Grant 7440 (GH, MICH, MO, UC, US); 70 km E of Lima, just W of Surco, 1,700 m, 24 Mar. 1952, Hutchison 553 (UC); Surco, 2,034 m, Feb. 1948, Soukup 3722 (F). MoQUEGUA: Prov. Ilo, 9 km E of По, 250 m, 20 Aug. 1957, Ellenberg 2732 (0); Prov. Moquegua, Carumas, 3,200 m, 2 21 Feb.- 6 Mar. 1925, Veherbaue 7301 (F), Mt. vu mde NW of Moquegua, 2,000-2,100 m, Маг. 1925, Weberbauer 7451 ud ja NE. "e: PASCO: Prov. Da niel Carrión, Yanahua 3,050 m, 16-22 June 1922, Macbride & аннын а ies (F). Department and Province un- known: sine loco, Fischer s.n. (LE). In his systematic overview of the genus Mon- nina, Chodat (1896) classified M. macrostachya among the stipulate species. In 1909, he described the taxon M. weberbaueri and its three varieties to accommodate some exstipulate specimens from central and southern Peru, otherwise essentially ike M. macrostachya. The type of M. macro- stachya is, however, exstipulate, and thus I have here synonymized M. weberbaueri and its varieties under M. macrostachya. Monnina macrostachya is a very variable spe- cies. In Lomas and sand dunes, the plants are often lower than they are when growing in the mountains, but the leaves are not proportionally smaller, which results in a very peculiar appearance (Fig. 7B). Monnina arenicola and M. macrostachya var. pumila were described mainly on the basis of small- er plant size, but I have found no evidence in other characters supporting the exclusion of small plants into a separate taxon. The degree of reduction of the fruit wing varies considerably. There is a con- tinuous series of forms— from fruits with entire wings through very irregularly winged ones to com- pletely wingless fruits. Furthermore, the fruits range from glabrous to densely pubescent. “з . Monnina amarella Chodat, Bull. Soc. Bot. Genéve sér. 2, 25: 200. 1934. TYPE: Peru. Cuzco: Prov. Cuzco, San Sebastián, 3,300- 3,400 m, 25 Apr. 1925, Pennell 13629 (lec- totype, selected by Ferreyra (1946), F; iso- lectotypes, GH, NY, PH, US). Figure 80-Е. Annual herb, 3-27 cm tall. Stem erect, 0.5-2 mm diam., terete, furrowed by decurrent leaves, usually branched all along the main axis, the branches erect, white-pubescent, the hairs curved. Leaves exstipulate; petiole 0.5-1 mm long; laminae on the main axis changing in shape and size from base to apex of the stem, the basal ones oblanceo- late, 0.5-1.6 x 0.2-0.6 cm, obtuse and emar- ginate at apex, the upper ones narrowly lanceolate to linear, 0.8-3.7 x 0.1-0.9 cm, obtuse and either minutely mucronate or rarely emarginate or acute at apex; all laminae chartaceous, cuneate to at- tenuate at base, glabrous except on the midrib. Inflorescence few-flowered, "torpedo -shaped in outline, 1.5-18 cm long; pedicel 0.5 mm long; bracts lanceolate to narrowly lanceolate, acumi- nate, ca. 0.5-1.5 mm long, dark blue, glabrous. Flowers of the main inflorescence chasmogamous, those of the branches usually cleistogamous (mea- surements for these given in parentheses); outer sepals unequal in shape and size, dark blue, gla- brous; upper sepal ovate to narrowly ovate, 1.3- 1.5 x 0.6-1.0 mm (0.8-1.5 x 0.6-1.0 boat-shaped, acute, lower sepals narrowly ovate to lanceolate, 1.0-1.2 x 0.3-0.5 mm (0.8-1.0 x 0.3 mm); wings narrowly obovate, 2-2.5 x l- 1.5 mm (2-2.5 x 1-1.5 mm), cuneate at base, glabrous, dark blue; keel 2.5 x 1.5 mm (2-2.5 x 1–1.5 mm), 3-lobed at apex, the central lobe mm), emarginate, truncate at base, glabrous, yellow; pet- al-stamen sheath 1.5-2.5/4-4.5 mm (1-2/3-4 mm) wide, 1-2/1-2.5 mm (1-2 x 1-2 mm)long, the petals pilose to hirsute adaxially on sheath, the lobes dark blue, otherwise greenish like the filament sheath; stamens 6, in flowers on branches often 4, the filaments of the 3 (2) anthers of each fascicle fused throughout (Fig. 8F), glabrous, 0.8-1 mm (0.8-1.0 filament ratio 0.25-0. (1-1.5 mm) long, glabrous, evenly curved, mm) including the anthers, the anther to 5 (0.3); style 1.5-2.0 mm the central part dorsiventrally expanded building a cav- ity, to 0.8 mm wide; anterior stylar lobe a short pointed tooth, 0.1—0.2 mm long, the posterior lobe with a sessile, rounded stigma (Fig. 10G). Fruit 1- locular, compressed, rotund in outline, 2.5-4. 3.5-5 green, often partly covered by the remaining, with- mm when mature, glabrous or puberulent, ered keel, the wing ca. 0.8 mm wide, bluish-tinged. Distribution. Open slopes among grass and rocks in the moun- tains, ca. 2,000-4,000 m (Fig. 6A). Central and southeastern Peru. Specimens examined. PERU. ANCASH: ipd Huaraz, Cerro Shaurema SE of Huaraz, 3,250 m, 9-13 Ma 1948, Pennell 15357 (PH). AYACUCHO: sine loco, ca. 2,400 m, 1951, Soukup 4018 (F). с : m, 18 Apr. 1973, Brunel 906 (MO); Pampa de Anta, 202 Annals of the Missouri Botanical Garden vicinity of Huarocondo, 3,200 m, Herrera 3638 (F); ov. Canchis, Checacupe, 3,400 m, 5 Apr. 1954, Rauh is [e h P707 (Е); Prov. Calca, Intihuatana Pisac, 3,300 0 Mar. 1945, largas 506 Ci ESE of Cuzco, 3,400 m, 14 Apr. = 1029 (0); near Cuzco, 3,300-3, s m, s 1925, Herrera 681 (F, US syntype), 3,600 m, Mar. 1929, Herrera 2393 (F); road between Cuzco and Urcos, km 34 at Huaro, 3,100 m, 25 May 1977, Solomon 3086 (MO); E of Cuzco on road to Pisac, 3,700-3,900 m, 11 Арг. 1985, Stein et al. 25324 (МО); Saxaihuaman, 3,600 m, Mar. 1929, Herrera 2388 (syntype, F), 3,500- 3,600 m, 24 Apr. 1925, Pennell 13589 (PH); Prov. Urubamba, Machupicchu, 2,000 m, 10 Apr. 1957, El- lenberg 1184 (U); Chincheros, 3,000-3,300 m, 16 Feb. 1982, King et al. 300 (ECON, F); Ollantaytambo, 3,000- n, 26 Apr. 1925, Pennell 13678 (PH), 3,100 1939, Vargas 11061 (F, UC). HUANCAVELICA: Prov. Huancavelica, 8 km E of Mejorada, 2,400 m, 13 Mar. 1939, Stork & Horton 10909 (F, UC); vicinity of Mejorada, 2,800- ua a 946, Soukup 2950 (F, GH, MO), 3,270-3,300 m, 1 Apr. 1951, Tovar 331 (US). Monnina amarella possesses two kinds of flow- ers: normal chasmogamous flowers with six stamens on the main inflorescence, and smaller, cleistoga- mous flowers with four stamens and more or less reduced petal-stamen sheath on the branches. Cleistogamous flowers are found in M. wrightii as Vernacular name: “bolsa bolsa." 8. Monnina filifolia Chodat, Bull. Soc. Bot. 8. 1934. TYPE: Peru. Tayacaja, valley of the Mantaro River, below Colcabamba, 1,900 m, Mar. 1913, Weberbauer 6454 (holotype, К: isotypes, GH, NY, S, US, USM). Genéve ser. 2, 25: 19 Huancavelica: Prov. Annual herb, 43-65 cm tall. Stem erect, 1.5- 3 mm diam., terete, furrowed by decurrent leaves, unbranched or branched distally, the branches erect, white-pubescent, the hairs curved. Leaves exstip- ulate; petiole 1-2 mm long; laminae changing in shape and size from base to apex of the stem, the basal ones oblanceolate, 1.3-1.9 x 0.2-0.3 cm, obtuse or sometimes obtuse with a point at apex, 0.1-0.3 ст, acute at apex; all laminae chartaceous, cuneate at the upper ones linear, 2.5-7.0 x base, glabrous. Inflorescence many-flowered, very narrowly triangular in outline, 21-46 cm long; pedicel 1 mm long; bracts linear, acute, ca. 2 mm long, glabrous. Flowers all chasmogamous; outer sepals in size, narrowly ovate, ciliate; upper sepal 1.5 x 0.8 mm, boat-shaped, the lower ones 1.0 x 0.5 mm; wings obovate, 2.5 х 1.5 mm, cuneate at base, glabrous, lilac (fide Weberbauer); keel 2.5 x 1 mm, 3-lobed at apex, the central lobe deeply emarginate, truncate at base, glabrous, yellow (fide Weberbauer); petal-stamen sheath 1.5/ 3 mm wide, 1/1.5 mm long, the petals hirsute adaxially at base, the lobes lilac (fide Weberbauer); stamens 6, the filaments of the three stamens in each fascicle fused throughout, glabrous, 0.8 mm long, the anther to filament ratio 1.0; style 1 mm long, glabrous, evenly curved, the central part dor- siventrally expanded building a cavity; anterior sty- ar lobe a short blunt tooth, the posterior lobe with a sessile, elongate stigma (Fig. 10G). Fruit 1-locular compressed, rotund in outline, 5 X 4 mm when — mature, densely pubescent, the wing ca. 0.8 mm wide. Distribution. Central Peru, ca. 2,000 m (Fig. 6E). Known only from the type collection Monnina filifolia is very similar to M. amarella. Both lack stipules and have six anthers with the three filaments of each fascicle united throughout. The most striking differences between the two spe- cies are the growth habit (branches only developed distally vs. all along the main axis), the leaf size and shape (upper leaves very long, narrow, and acute vs. shorter, wider, and obtuse), the number of flowers (many vs. few), and the outline of the inflorescence (very narrowly triangular vs. "tor- pedo” shaped). The leaf size and shape vary within many other species, whereas growth habit, flower number, and inflorescence outline are generally constant characters within a species. The collection of M. filifolia comes from a lower altitude (1,90 m) than the majority of the M. amarella collections (2,900-3,900 m). That fact indicates either that M. filifolia is a separate taxon with a distinct habitat preference, or that M. filifolia should be lumped into M. amarella as a lowland ecotype. It is important to note that M. amarella collections from approximately the same geographical area and comparatively low altitude as M. filifolia (Stork & Horton 10909 and Soukup 4018, both 2,400 m) have a tendency toward longer and more narrow leaves. However, more ample material is needed to determine whether M. filifolia should be con- sidered a separate taxon, as done here, or whether it should be included in M. amarella. 9. Monnina wrightii A. Gray, Pl. Wright. 2: 31. 1853. ТУРЕ: U.S.A. New Mexico: near the copper mines, 1851, Wright 938 (lec- totype, selected here, GH; isolectotypes, GH, NY, PH, US). Figure ВА-С Monnina brachystachya Griseb., Abh. Kónigl. Ges. Wiss. Gottingen 19: 75. 1874. TYPE: Argentina. Tucu- Volume 80, Number 1 1993 Eriksen 203 Monnina subg. Pterocarya D A D е Monnina wrightii (&-C) and M. amarella (D-F). Habit. —B, F. Detail of petal-stamen sheath тент а fascicle of stamens with connate fila- ments. —C, E. Fruits. A, C, Cabrera et al. 17038 (GH). B, Arséne 554 . D-F, Stein et al. o _ Scales: А, D = 5 cm. В, F = 1m : Catamarca, Маг.-Арг. 1872, Lorentz 105 resi selected here, GOET). Monnina macrostachya var. stenophylla Kuntze, Revis. Gen. Pl. 3(2): 10. 1898. TYPE: Bolivia. Larecaja (La Seja?), 1892, Kuntze s.n. (lectotype, selected here NY Monnina doe: Chodat, Meded. Rijks-Herb. 27: 29. 1915. TYPE: Bolivia. Santa кез i Achiras С. 1.300 m, , Herzog 1726 онны selected here, С; еј L, S, W). Nomenclatural note: Monnina brachystachya was based on two incongruent elements, viz Lo- rentz 105 and Lorentz 414. The original descrip- tion seems to be based on the flowers of #414 and the fruits of #105. Lorentz 414, however, belongs to a species different from M. wrightii, probably M. dictyocarpa Griseb. Monnina brachystachya is here typified on Lorentz 105, because this col- lection is the most complete, showing both fruits and flowers (cleistogamous) as well as the habit of the plant. The name M. brachystachya has tra- ditionally been applied to the species represented 05. It was probably the confused original description of M. brachystachya that caused Cho- at to describe M. eriocarpa. Annual herb, 14-86 cm tall. Stem erect, 1—5 mm diam., terete, usually branched distally, white- pubescent, the hairs curved. Leaves exstipulate; petiole 1-3 mm long, pubescent; laminae on the main axis changing from narrowly ovate at base to very narrowly lanceolate or linear near inflo- 1.4-8.5 x 0.2-4.2 acute to acuminate at apex, cuneate at base, the rescence, cm, chartaceous, margin and midrib pubescent, otherwise glabrous, the young leaves pubescent throughout. Inflores- cence many-flowered, narrowly to very narrowly triangular in outline, 3.5-54 cm long, sometimes very densely pubescent; pedicel 0.5 mm in open flower; bracts very narrowly lanceolate, acuminate to caudate, 1.5-2 mm long, ciliate, undulate, yel- lowish green mixed with blue. Flowers of the main inflorescence chasmogamous, those of the branches usually cleistogamous (measurements for these giv- en in parentheses); outer sepals unequal, the upper sepal ovate to narrowly ovate, 1-1.5 x 0.5-1 mm (1-1.5 x 0.4-0.6 mm), boat-shaped, the lower ones lanceolate, 1-1.5 x 0.3-0.5 mm (1-1.5 x 0.2–0.5 mm), all acute at apex, ciliate, yellowish mixed with blue; wings obovate to narrowly ob- ovate, 1.5-3 1-2 n -2.9 X 0.8-1.5 mm), cuneate to attenuate at base, glabrous, blue; keel 1.5-3 x 0.7-1.5 mm (1.5-2 x 0.4-1.2 mm), emarginate at apex, truncate at base, gla- brous, possibly yellow throughout; petal-stamen sheath 1-3/1.5-4.5 mm (0.3-1.5/2.5-3 mm) wide, 1-2/1-2.5 mm (1-15/1.2-2 mm) tall, a tuft of hairs present at the upper margin between the two filament fascicles, the petals hirsute adax- ially, the lobes blue, otherwise greenish as the fil- ament sheath; stamens 6, ог two of them + reduced or totally absent, the upper part of the filaments of the outer two stamens of each fascicle fused throughout, the third stamen, when present, single, glabrous, 0.7-1.2 mm (0.5-1 mm) long ое the anther, the ма to filament ratio 0.3- 1 (0. 1); style 0.8-2 mm (0.5-1.2 mm) long, Poen bent 45-90? Y, to Y, from the base, with or without a cavity at the middle; anterior stylar lobe absent or a very short, blunt tooth, the posterior lobe with a sessile, crested stigma (Fig. 10H). Fruit 1-locular, crescent-shaped in outline, 2-5 X 3-6 mm when mature, pubescent, some fruits rarely uni- or bi- laterally glabrous, green, narrowly winged (ca. 1 se. Distribution. Southern Peru to central Ar- gentina; Uruguay; Mexico and the southwestern United States. Gravelly slopes in the mountains, 1,000-3,000 m (Fig. 6C). In the north it flowers in September- October, in the south between No- vember and April. Monnina wrightii shows an extreme disjunction in its distribution. The type, in appearance, belongs to the homogeneous Mexican-North American population, while the main population of the spe- cies, in which variation among individuals is ex- tensive, is found in Bolivia and Argentina. Since the rest of subgenus Pterocarya occurs south of the equator, and the largest variation within the species is found there as well, I presume that M. wrightii has been introduced to Mexico and the southwestern United States from the Bolivian—Ar- gentinian population. Grondona (1945) was aware of the wide distribution of this species, considered by him to be M. brachystachya, and wrote: “Se extiende desde México por la Región Cordillerana 204 Annals of the Missouri Botanical Garden hasta La Rioja, en el noroeste de la Argentina." He did not realize that an earlier, valid name, Monnina wrightii, had been published based on a type from New Mexico. The species seems rather uncommon in Peru and Uruguay; only a single collection is known from each of these countries. Specimen examined. PERU. CUZCO: Prov. Urubam- ba, Ollantaytambo, 2,850 m, 26 Apr. 1925, Pennell 13683 (PH). Additional material studied (representative collec- ARGENTINA. CATAMA (GH). TUCUMAN: Schreiter 3317 (GH); Baile 150(GH ). BOLIVIA. COCHABAMBA: Eyerdam 249. LA PAZ: Mandon 836 (GH). PANDO: Buchtien 250 (GH). TARIJA: Fiebrig 2054 (LY). MEXICO. CHIHUAHUA: Pringle 302 (B). MORELIA: Arsène s.n. (LY). URUGUAY. SACUAREMBO: Osten 6537 (GH). This species possesses two kinds of flowers, a fact already noted by Grondona (1945). Normal chasmogamous flowers occur on the main inflo- rescence axis, while small and very reduced cleis- togamous flowers occur on the branches. In the cleistogamous flowers the petal-stamen sheath is usually reduced to the filament sheath carrying two fascicles each with two anthers, which in turn are adherent to the stigma. The reduction is not always complete, и The style is usually C-shaped and very s Тће fend of M. brachystachya (Lorentz 105) lacks the main inflorescence axis and con- sequently has only cleistogamous flowers. Gron- dona's illustration is slightly misleading because chasmogamous and cleistogamous flowers are shown on different individuals. Тће field notes on colors of the flower are rather sparse. | have not seen this species alive, so the description of the colors is given with reservation. 10. Monnina pterocarpa Ruiz & Рауоп, Syst. Fl. Peruv. Chil. 1: 174. 1798. TYPE: Peru. Lima: Provinces Lima and Chancay, Ruíz & Pavón s.n. (lectotype, selected here, MA, photo of type seen). Figure 9A-H. Monnina angustifolia DC., Prodr. 1: 340. 1824. TYPE: Per Monnina chanduyensis Chodat, Bull. H 16 94. TYPE: Ecuador. Guayas, C the coast of the Pacific Ocean, 1865, Spruce 6398 (holot me G; isotypes, E, P, W). Monnina pterocarpa var. exauriculata Chodat, Bull. c. Bot {нган ser. 2, 25: 202. 1934. TYPE: Peru. Tumbes: Prov. Tumbes, Cancas, 1 Mar. 1927, Weberbauer 7757 (lectotype, selected here F; isolectotype, US). Monnina piurensis Ferreyra, Phytologia 69: 359. 1990. RE 9. Monnina pterocarpa (A-H) and M. her- — A. Habit and detail of flowering branch. GU bacea (I-K). — B. Upper outer sepal. — C. Lowe K —J. Petal-stamen sheath. cium. A, H, Molau & Eriksen 3466 (GB). B- С, Dod 7393 (GH). 1, Molau & Eriksen т iua J, K, к: ер С). Scales: А = ‚ В- F = 2 mm. H = 2 mm. J, К = 1 mm. E: Peru. Piur v. Piura, near Piura, Ferreyra 5882 e p isotype, US). Annual (or perennial?) herb, ca. 0.5-2.5 m tall. ca. 3-10 mm diam., below, loosely branched above, the stem and older Stem erect, terete, simple branches often glabrous, glaucous, the younger branches sparsely white-pubescent, the hairs curved. Leaves stipulate; petiole 1—3 mm long, pubescent; laminae on the main axis lanceolate to narrowly lanceolate in some plants, narrowly ovate to lan- ceolate in others, 2-6.5 x 0.2-3.2 cm, charta- ceous, acute at apex, cuneate at base, + pubescent throughout, the branches always with gradually shorter and more narrow laminae toward the in- florescence. Inflorescence many-flowered, very narrowly triangular in outline, ca. 15-40 cm long; pedice mm; bracts lanceolate, acuminate, l- mm long, + densely white-hirsute. Flowers all chas- mogamous; outer sepals equal in size or the upper Volume 80, Number 1 1993 Eriksen 205 Monnina subg. Pterocarya sepal slightly larger, narrowly ovate to lanceolate, 2-2.2 mm long, glabrous or rarely sparsely pilose abaxially; wings obovate to broadly obovate, 2- 3.5 x 1.5-3 mm, cuneate to truncate at base, glabrous, pink to violet; keel 2.5-4 x 2-2.5 mm, truncate at base, glabrous, emarginate with two lateral lobes at apex, yellow, at least the tip and two lateral spots; petal-stamen sheath 1-2.5/3-7 mm wide, 0.7-3/1-3.5 mm tall, the petals pu- bescent at the basal margin, the staminal sheath pilose along the midline abaxially (a tuft of these hairs is seen between the filament bundles), the fold marking the line between the petal and filament sheath fringed by long hairs, the petal lobes pink to violet, otherwise greenish as the filament sheath; stamens 8, the upper part of the filaments free, glabrous, 1-1.5 mm long including the anther, the anther to filament ratio 1.0; style 1.5-3.5 mm long, glabrous, = sigmoid; anterior stylar lobe ab- sent or a very short blunt tooth, the stigma a bilobed disc, sessile or at the tip of a short posterior lobe (Fig. 101). Fruit 1-locular, compressed, orbicular to rotund, 5-7 x 4-7 or pubescent, sometimes varying on mm when mature, glabrous the same in- dividual, green, occasionally purple-spotted, broad- winged, the wing purple. Ecuador, Peru, and Chile. In sand and gravel in dry areas along the coast, or on steep, rocky slopes in the coastal mountains, 0-2,200 m (Fig. 6B). Distribution. pecimens examined. ECUADOR. GUAYAS: near ui Libertad, 23 Feb. 1972, Beck Breuninger 156 (GB); 1 near Salinas, 5 Apr. 194 Centinela, 6 Apr. 1941, Svenson 11390 (GH, NY, UC, US). РЕКЏ. ANCASH: Prov. Bolognesi, km 46 on road from Pativilca to Recuay, 580 m, 27 Jan. 1983, Dillon et al. 3074 (F, MO), 1,400 m, 9 July 1982, о et al 37308 (МО); Colcas, 2,200 m, 19 May 15377 (PH) Prov. Huaylas, Huallanca, ы 5 on road to Caraz, 1,700 m, 8 Mar. 1973, Beck 7892 (CB). CAJAMARCA: Prov. Cajamarca, 50 km SW of Cajamarca, between Magdalena and San Juan, 1,840 m, 5 June 1963, Ugent & Ugent 5394 (WIS); Prov. Contumaza, a Ж. from Pacasmayo on road to Cajamarca, са. 500 Apr. 1988, Molau & oe 3466 (GB); El к. de San Antonio, 400 m, 14 May 1983, eai o et al. 9184 (MO); Ascope- ros obal, 150 т 1983, Sagástegui & Mostacero 11336 (MO, US); ): Que. ien 1984, Sánchez Vega 3400 ( 600 m, 22 Nov. 1964, Hut eas P is 7116 (Е, UC); Puquio Valley, 1,300 m, 10 Маг. 1954, Rauh & Hirsch P405 (F). L4 LIBERTAD: Prov. Trujillo, El Porta- d Ascope-San Benito, 600 m, 11 Aug. 1957, López 4 (UC); Cerro о Salaverry, Il m, 24 E 1983, López 9156 (МО); Cerro Cabezón, 700 m Feb. 1983, Sagástegui & Mostacero 10453 (MO, us MA: Prov. Huarochiri, Chosica, 900 m, 1 Feb. 1976, Cou? 16385 (MO), 28 Apr.-2 May 1922, Macbride & Featherstone 494 (F), 11-13 Mar. 1923, Macbride & Featherstone 2874 (F, US); near Sta. Eulalia, above Chosica, 1,100-1,400 m, 2 Apr. 1939, Goodspeed 11308 (F, GH, NA, UC), 1,100-1,200 m, 10 Apr. 1977, Gentry et al. 19152 (F, MO), ca. 900 m, 5 Oct. 1935, Mexia 4007 (GH, MO, UC), 22 May 1963, Ugent & Ugent 5283 (UC, WIS), 30 July 1941, Grant 7393 (F, GH, UC), 800 m, Soukup 2049 (F, GH); above the village Sta. Eulalia, 1,500 m, 15 May 1942, Goodspeed 33143 GH, MO, UC); Santa Eulalia-San Juan de Iris road, 1,200 m, 24 Apr. 1982, Smith & León 1346 (MO), 1,300 m, 24 Apr. 1982, Smith & León 1356 (MO); Prov. Lima, 20 km NE of Trapiche on road to Huaral, Canta Valley, 1,000 m, 5 Aug. 1957, Hutchison 1024 F, UC); Canta ен | km from Lima, 1,100 m, 7 Oct. 1956, Keie X Quives, 800-1,000 m, 9 June 26 Pennell зане СН, PH); Lima, 1838-42, sine — ~ Ferreyra 5938 (Е); Paita, 0- 20 m, 4 July 1925, Penneli 14812 (F, GH, LE, PH, US); Prov. Talara, Talara, 1925, dde d 7 (F), 11 Oct. 1925, Johnston 3512 (F, GH), Om, 1 Feb. 1954, Rauh & Hirsch P22 (F); Negritos, 23 Dec. 1928, Haught F91 (Е); road Зшапа-Тајага, 14 Oct. 1983, Sagástegui 10900 (MO). TUMBES: Prov. Contralmirante Villar, Zorritos, 100 Ellenberg 1327 (U). Department unknown: sine loco, Martinet s.n. (F). Because of the size of the plant, the basal parts are usually not present in herbarium specimens. It is therefore difficult to see whether the plant is annual or perennial. It is hard to believe that a 2.5-m-tall plant with a more or less woody base could possibly sprout and mature in one short sea- son in a dry area. On the other hand, I have seen several withering plants devoid of leaves in a nat- ural population, and the root system seems too tiny for a perennial. Collectors disagree on the subject; some noted “annual,” others "perennial" on the abels. It seems as if there is a gradient from very broad-leaved specimens in the north to almost lin- ear-leaved ones in the south of the distribution area. That is, however, an artifact. The Ecuadorean specimens are generally smaller than the Peruvian ones, and the whole plant, instead of a branch, is usually collected. Since the leaves on the main axis are the largest of the plant, the specimens from the north seem extraordinarily broad-leaved. Bracts vary in length from 1 to 3 mm and tend to be longest in bip from the north. The transition is gradual, how Ferreyra (1990) described the species M. pi- urensis on the basis of larger leaves and longer bracts relative to M. pterocarpa. Monnina pi- urensis is, for the reasons mentioned above, in- cluded here in M. pterocarpa. 206 Annals of the Missouri Botanical Garden The color of the flowers fades after drying and is mostly very poorly described on the labels. Ac- cording to the notes I have seen, pink flowers with a brownish yellow keel are common. It is, however, my experience that the brownish tinge of the yellow keel is due to post-pollination color change. The plants I saw had more violet flowers, the keel with a yellow tip and two lateral, yellow spots. Variation in flower color is also seen in M. macrostachya. In a few specimens, unhatched insects (Hyme- noptera-Chalcidoidea, determined by T. Hag- stróm) were found inside the flowers. This group of insects is richly represented in the tropics and, according to Hagstróm, taxonomically very diffi- cult. Аз it is known to contain both parasitizing and pollinating species, it is not known whether the presence of the insects is a matter of pure para- sitism or part of the pollination syndrome. 11. Monnina herbacea DC., Prodr. 1: 340. 1824. TYPE: Peru. Lima: near Lima, Lagasca 53 (holotype, G-DC). Figure 9I-K. Monnina idee Chodat, Bull. Herb. Boissier 2: 167. 1894. : Ecuador. Ecuadorean Andes, 1857- 1859, S race 5977 (holotype, G; isotypes, E, GH, GOET, LE, NY, S, US, W). Monnina siena Chodat, ain Jahrb. 2 42: кын 909. ТҮРЕ: Peru. Cajamarca: Prov. Hualgay San Miguel 2,200 m, Wi T uem 3919 (lectotype, selected here, MOL; isolectotype, USM). Annual herb, ca. 20-100 cm. Stem erect, 1— 4 mm diam., terete, slightly furrowed by decurrent leaves, unbranched or branched distally, white- pubescent, the hairs curved. Leaves stipulate; pet- iole usually 2 mm long, pubescent; laminae lan- ceolate to narrowly lanceolate, 0.1-2.0 x 1.2- 5.2 cm, chartaceous, acute to acuminate at apex, cuneate at base, glabrous except at the margin and the midrib below, the young leaves pubescent throughout. Inflorescence many-flowered, narrow- ly triangular in outline, ca. 15-50 cm long; pedicel 1 mm long; bracts narrowly lanceolate, 1-2 mm long, acuminate, dark bluish-tinged, ciliate and of- ten sparsely pilose abaxially. Flowers all chasmoga- mous; outer sepals narrowly ovate to lanceolate, slightly unequal in size, the upper sepal 1.5 x 0.5- 0.7 mm, the lower ones 1-1.5 x 0.5 mm, glabrous or rarely sparsely ciliate, dark purplish blue; wings obovate, 2.5-3 x 1.5- at base, glabrous, dark purplish blue; keel 2.5-3 1.5-2 mm, cuneate at base, glabrous, 3-lobed at apex, the central lobe emarginate, bright yellow, the keel otherwise dark purplish blue; petal-stamen sheath 2-2.5/3-4 mm wide, 2/2-3 mm tall, the 2 mm, cuneate to attenuate Bea N, ов FIGURE 10. Styles Pterocarya in northwestern | i. —B. M. and stigmas of и. subg. South Americ ЈЕ С. poly- es. —D. M. ramosa . M. macbridei. —F. M. macrostachya. —G. M. amarella and i ни . M. wrig vs —I. M. pterocarpa. —J. bacea. Scale — m. petals pubescent at lower margin, the lobes hirsute adaxially, dark purplish blue, otherwise yellowish green as the filament sheath, the fold marking the line between the petal and filament sheath fringed by long hairs, the upper margin of the filament sheath provided with long, curled hairs; stamens 8, the upper part of the filaments free, glabrous, 0.8-1.0 mm long including the anther, the anther to filament ratio 0.3-0.5; style 1.5-2 mm long, glabrous, bent 90? at the middle; anterior stylar obe absent or a very short blunt tooth, the posterior lobe resembling a hammer-head, i.e., a short sta with a shorter and narrower stalk perpendicular to it, at the tips of which the stigma is found (Fig. 10J). Fruit 1-locular, + crescent-shaped in outline, 3-5 х 2-4 mm when mature, glabrous or pubes- cent, sometimes so on the same inflorescence, wing- less, or irregularly or regularly winged, green with purple lines or spots, the wing, when entire, purple- margine Distribution. Central Ecuador to central Peru. Gravelly slopes in the mountains, ca. 1,200-2,800 m (Fig. 6E) Specimens examined. ECUADOR. CHIMBORAZO: Nariz del Diablo, 2, 000 m, 2 June 1939, Asplund 6848 (US); Camp E-2996 (NY); Simbambe, 2,500 m. 1 Tune 1947, diui 27655 (US), 2,000 m, 28 May 1942, Haught 3 (BR mixed coll., F, GH, US); Huigra, 1,200 m, ae 1923, Hitchcock 20612 (GH, NY, US); vicinity of Huigra, Hda. Licay, 19 Aug. 1918, Rose & Rose 22213 (GH, NY, US), 7 Sep. 1918, Rose & Rose 23853 (GH, NY, US). Loja: Road Amaluza-Machay-Jimbura, near Machay, ca. 2,000 m, 29 Feb. 1988, Freire Fierro 1084 (AAU, GB). PERU. ANCASH: Prov. Huaylas, trail to cave across Rio Santo from Mancos, 9 Apr. 1970, Earle Smith Volume 80, Number 1 1993 Eriksen 207 Monnina subg. Pterocarya & Blas 4902 (US mixed coll.). CAJAMARCA: d ja- bamba, Chicdén below Sunchubamba, 2,800 m, 10 Ju pa 1957, Ellenberg 1920 (U); Prov. EH between Magdalena and San Juan, 1,600-1,700 m, 10 Apr. 1950, (F) on the road to Chilete, 1 km below 5069 (Е, MICH, MO, UC, US), 5070 (Е); La Pena пе 1,730 m, 21 May 1975, Sagástegui et al. 7923 (МО); Prov. Celendin, above Balsas, 30 km E of Celendin on road to Chachapoyas, 2,300 m, 23 Mar. 1988, Molau & Eriksen 3383 (GB); Prov. Contumazá, road to Guaya- bo, 1 km SE of Magdalena, 1,300-1,500 m, 18 Mar. 1988, Molau, Eriksen et al. 3334 (GB); 3-4 km S of Contumaza along road to Cascas, 2,850-2,900 m, 4 Apr. 1985, Molau et al. 1798 (GB); between Chilete and San Juan, 16 May 1952, Ochoa 1500 (F, GH); La Montana, m, 3 Apr. 1981, Sagástegui et al. 9720 (МО); arawd Contumazá, 2,600 m, 25 May 1981, p et al. 9851 (МО); San Martín, 2, 700 m, 31 May 1988, Sagástegui et al. 14026 (GB). HUANUCO: Prov. Huanuco, 10 km N of Huanuco on road to Tingo Maria, 1,900 13 Apr 1977, Gentry et al. 19255 (MO). LAMBAYEQUE: Prov. 1, 10 Prov. Canta, along Rio Chillón, near Viscas, 1,800- 2,000 m, 10-15 June 1925, Pennell 14467 (F, GH, PH, US Province Er Dombey 625 (F, P). Pasco: Prov Paucartambo, between Corhuamayo and Paücartämbó, Ochoa 1035 (F, GH). Monnina herbacea is, together with M. ptero- carpa, unique among the species of subgenus Ptero- carya in northwestern South America because of having stipulate leaves. Stipulate leaves are more common among the species in the eastern distri- bution area. The main difference between the two species mentioned is the construction of the anterior stylar lobe and the position of the stigma. In M. herbacea the anterior stylar lobe resembles a hammer-head with rounded er at the tips, while the stigma pterocarpa is a sessile or short-stalked, bilobed disc (Fig. 101, J). Furthermore, M. her- bacea is in all parts smaller, the flowers are of more delicate texture with differences in proportion and shape of many details, and the stipules tend to be longer. They are also separated altitudinally, with P. pterocarpa primarily occurring from the coast up to 1,500 m (rarely higher) and M. her- bacea above 1,500 m (occasionally lower). Overlap exists, and the two species are even sympatric in some areas. The name M. herbacea has, since the species was described by de Candolle in 1824, been as- signed to almost every small, annual species in the subgenus except the taxon of the type. Chodat (1896) placed it among the species lacking stipules, but my studies of the type at G-DC revealed that stipules are present. In 1894, Chodat described M. spruceana from Ecuador, and later (1909) M. graminea on material from northern Peru, both stipulate. Together with the type of M. herbacea from the Lima area they constitute a single, uni- form taxon. Ferreyra (1946, 1953), based on Cho- dat's erroneous note on lack of stipules, and the wingless fruits mentioned in the diagnosis, applied the name mostly to specimens of M. leptostachya and M. polygonoides. "palomilla. La d Vernacular name: LITERATURE CITED ed A. P. DE. 1824. Prodromus 1: 338-340. Gaon: n Н. 1894. Polygalaceae novae vel parum cognitae. Bull. Herb. Boissier 2: 167-174. 1896. Conspectus Mere generis Mon- о Herb. Boissier 4: -253. Plantae novae ar imprimis We- пейте p Polygalaceae andinae. Bot. Jahrb. iin ninae. та von Dr. Th. Herzog auf seiner zweiten Reise durch Bolivien in den Jahren 1910 und sir gesammelten Pflanzen. Meded. Rijks-Herb. == 27: . e е и Bull. Soc. Bot. балача ser. 2, 25: . WILCZEK. "ud Contributions à la flor de la république Argentine. Bull. Herb. Boissier sér. 2, FERREYRA, R. of Monnina. J. Arnold 1953. A revision of the РЕС species of Monnina е Lloydia 16: -226. . 1990 ; taxa of Monnina Сао for South Ал Phytologia 69: 354-360. GRONDONA, E. M. Las especies argentinas на Бе! nero Monnina (Polygalaceae). Darwiniana 7: uc А revision of the Peruvian species 167 Arbor. 27: 123- Mon U. 1988. pia Part I. Tribe Cal- ceolarieae. Flora Neotropica 47: 1-326. New York Bot. Gard., New York. Ruíz, H. & J. A. Pavón. 1798. Systema vegetabilium florae Peruvianae et Chilensis. 1: 169-174. Madrid. WALTER, Н. 79. Vegetation of the Earth, 2nd edition. Springer-Verlag, New Yor THE MISCONSTRUED AND RARE SPECIES OF COMMELINA (COMMELINACEAE) IN THE EASTERN UNITED STATES' Robert B. Faden? ABSTRACT Five species of Commelina (С ommelinaceae), all introduced in the eastern United States, are reviewed and discussed. % e о 5 А іе) e e. o ES S r. gigas is gir 2 d is bere with the more common C. diffu eir seeds melina cda is determined to be a polyploid of C. diffusa, and sa, and the species are y gigas is considered likely to have been separately t Va introduced. Сотта benghalensis probably became naturalised in Florida early this century. Of the two recently C. Пи skaolii is known from a single (2n — 56), all first reports from the United States. e is be ecoming well спори вон Миа roadsides in Florida. Maps of the easter . be "nghale nsis, C. d and C. ica var. gambiae benghalensis (2n — 22) rida population that may have been к, and С. or ), C. diffusa var. diffusa (2n = 30), ¿Es pa 230) and C. nigritana var. gambiae The purpose of this paper is to supply back- ground information about five misunderstood or rare species of Commelina in the eastern United States. All of them will be fully described in my treatment for the Flora of North America, so detailed descriptions have not been included. In- stead, data are provided and discussed that cannot be contained in the flora because of format and length restrictions. All five species treated here were introduced in the eastern United States. Among other naturalized species, only C. communis L. is not considered, because it is widespread and well known. Commelina diffusa Burm. f., the second most common and widespread introduction, is dis- cussed only in the context of its separation from the often overlooked C. caroliniana Walter and its relationship with the enigmatic Florida plant C. igas Small. Among the five taxa dealt with in detail herein, three were either misinterpreted (C. caroliniana), or ignored (C. gigas), or overlooked (C. benghalensis L.) in the most recent monograph of the U.S. species (Brashier, 1966). The other two, C. forskaolii M. Vahl and C. nigritana Benth.. had not yet been collected by 1966 DISTINCTIONS BETWEEN COMMELINA CAROLINIANA AND C. DIFFUSA Commelina caroliniana was recently shown to be a distinct species in the eastern United States and an old introduction from India (Faden, 1989). It is a widespread weedy plant (Fig. 1), yet it remains poorly understood. Undoubtedly this is because of its close resemblance to the even more widespread C. diffusa. Both species are usually annual, diffusely spreading plants that root at the nodes. Both have spathes that are solitary (not clustered), simply folded (i.e., and usually acuminate at the apex; small flowers with three blue petals (as contrasted with C. com- munis L., margins not fused), which has a white lower petal); and typically five-seeded capsules with the dorsal locule one-seeded and indehiscent and ventral locules two- "Т thank R. P. Wunderlin, J. Beckner, and J. Popenoe for assistance with xu fieldwork in Florida in 1983; R. D. Thomas and R. P. Wu A. R. Tangerini for the maps and line drawings; R. nderlin for supplying living material of C. caroliniana and C. diffusa var. gi igas, respectively; э Oliver for technical assistance with the chromosome counts; the curators of the following herbaria for assistance during visits to their institutions or for supplying loans of specimens CHARL, CLEMS, DUKE, F, FLAS, FSU, FTG, GH, LSU, LTU, MO, ae Faden for heroically caring for the living plants for and VPI; and A. J. NY, PH, TENN, UARK, USF, m em of Botany, National Museum of Natural History, ош а Washington, D.C. 20560, US ANN. MISSOURI Вот. Garb. 80: 208-218. 1993. Volume 80, Number 1 1993 Faden 209 Commelina in the Eastern United States mA ЈАКА СУА TY HT Од, VPE A Кал FiGURE 1. seeded and dehiscent. Because authors have been concerned mainly with separating C. diffusa from C. communis, it is hardly surprising that cryptic C. caroliniana has been overlooked. It may be noted that reviewer Austin questioned whether C. diffusa truly is annual in southern Flor- ida. In my experience, this species is usually pe- rennial in the tropics, so potentially it could persist year-round in the deep South. To determine wheth- er it does indeed survive mild winters in this region, one would need to make long-term observations of populations. Further north, both C. diffusa and C. caroliniana clearly are annual because they lack subterranean, perenniating organs. The only modern floras in which C. caroliniana has been correctly circumscribed are Gleason (1952, 1963), Gleason & Cronquist (1991), and Steyermark (1963). Radford (1968) considered it a doubtful species. In all of these works, except Gleason & Cronquist (1991), C. caroliniana and C. diffusa were distinguished by supposed differ- ences in habit and spathe length, characters that do not work. Brashier (1966) considered C. car- oliniana to be a synonym of C. diffusa. Distribution of Commelina caroliniana in the United States. The distinguishing features that may be used to separate C. caroliniana from C. diffusa are sum- marized in Table 1. The most absolute difference between them is the testa (Fig. 2). In addition to being nearly smooth (vs. reticulate), the seeds of C. caroliniana are also larger and darker brown than those of C. diffusa. The spathes of both species are distinctive, and thus flowering specimens almost always can be identified. The spathes of C. caroliniana charac- teristically lack an exserted upper cincinnus (heli- coid cyme) and have a lower edge (i.e., the spathe midrib) that is nearly straight to only slightly curved (Fig. 3A-C). The spathes are not strongly falcate. In contrast, the spathes of C. diffusa typically have an exserted upper cincinnus bearing one to several flowers. The lower edge of the spathe is moderately curved, making the spathes distinctly falcate (Fig. 3D-F). Specimens of C. diffusa with very small spathes that lack an exserted upper cincinnus nor- mally also bear larger, more typical spathes. The spathe margin is commonly ciliate basally in C. caroliniana (at least a few long hairs present) and usually ciliolate to glabrous (rarely ciliate) in 210 Annals of the Missouri Botanical Garden FIGURE 2. Seeds of Commelina species in the United States. — A-C. 86831, US).—D-G. ЎА d var. diffusa (Faden & Fader 76/1, US gigas (Faden et al. 83/35, US). All to the same scale C. diffusa. The hairs on the peduncles of the spathes are longer in C. caroliniana (to 0.5 mm long) than in C. diffusa (to ca. 0.1 mm). The floral differences between the two species, listed Table 1, though the floral morphology of C. diffusa has been need further verification. Al- recorded from numerous populations in the Paleo- tropics, it has been observed in only a few, scattered populations in the United States. Similarly, living Commelina caroliniana (Thomas et al. S). — H- K. Commelina diffusa var. material from only three populations of C. caro- liniana (all from Louisiana, supplied by R. Dale Thomas) has been studied. Populations of C. diffusa typically have some or all flowers with the medial staminode vestigial or lacking. The antherodes (staminode anthers) in this species never have a central maroon spot. The medial stamen anther normally has a violet to maroon connective (but see C. gigas below). Volume 80, Number 1 1993 Faden 211 Commelina in the Eastern United States 1cm 7 FIGURE 3. US; B, Mellichamp s.n., US; C, C jurtiss 5177, Plowman 13267, US; F. Anderson 10893, US). All to the same bs. In C. populations observed had three equal staminodes with fully developed, 4—6-lobed antherodes. In two caroliniana the flowers from all three of the populations the antherodes had a central maroon spot; in the third population the antherode lobes were entirely yellow. In Garber s.n. (US) from Florida, the yellow antherodes of pressed flow- ers show a central blue spot. In all three living collections, the medial stamen anther consistently had a white connective. It is probable that at least some floral differences noted between the two spe- cies will be confirmed from further observations. Cytological differences between the two species have also been found. Although C. diffusa has been reported to have many different chromosome num- bers in various parts of its pantropical and warm temperate range (see Faden & Suda, 1980, for summary), the only count for typical material from the United States is 2n — 30 (two populations from Spathes of Commelina species in the United States. — A-C. Co —D-F esc caroliniana (A, Garber s.n., . Commelina diffusa var. diffusa (D, Smith 486, US; E, —G-I. Commelina diffusa var. uote. (Faden et al. 83/35, US). Florida: Faden et al. 83/32, J. Thomas s.n., vouchers at US). An exact count was not obtained for C. for all three Louisiana populations (R. D. Thomas 86722, R. D. Thomas et al. 86830, R. D. Thomas et al. 86831, all US). This agrees in ploidy (but not exact number) with two reports from India of n = 45 for this species (as C. hasskarlii) (see Као et al., 1968, for summary). Based on preliminary data, typical C. diffusa is diploid in the United States while C. caroliniana is hexaploid and prob- caroliniana, but 2n — ca. 86 was secured ably also aneuploid. Only one specimen that is probably C. carolinia- na cannot be identified with certainty. Unfortu- nately, Toler 45 (UARK) would represent the sole record from Arkansas (see Fig. 1). The single spathe of this flowering specimen is typical for C. caro- liniana, but the spathe base is broken, and the marginal hairs cannot be checked. Better material 212 Annals of the Missouri Botanical Garden TABLE 1. Differences between Commelina diffusa and C. caroliniana. COMMELINA CAROLINIANA COMMELINA DIFFUSA Spathes not at all to slightly falcate, margins usually ciliate basally Upper cyme usually vestigial (rarely well developed and l-flowered) All 3 staminodes with well-developed antherodes, an- therodes yellow, usually with a central maroon spot Middle anther connective w Capsules (5-)6-8 Ventral locule seeds кй to faintly alveolate, 2.4— 4.3(-4.6) mm long Spathes usually distinctly falcate, margins glabrous or ciliolate (rarely ciliate) Upper cyme in larger spathes well developed and 1- several-flowered Middle staminode commonly lacking or without anther- ode, antherodes entirely yellow Middle anther connective maroon to violet Capsules 4-6.3 mm lon Ventral locule seeds deeply reticulate, 2-2.8(-3.2) mm long is needed to confirm this species from the state which would appear to be a reasonable extension of its confirmed range. The record from Jackson County, Missouri, is based on two collections (one sterile and thus un- certain) from Sheffield (Bush 3307, 3332, both GH, NY), which the collector (in 1905) noted as introduced. In Flora of Missouri, Steyermark (1963) treated this as an introduced species. It has not been re-collected in the state. Curiously, there PH) of C. but without further is a very old specimen (Nuttall s.n., caroliniana labeled “Missouri” locality information. The state of Missouri would seem to be a very unlikely location for such an early collection of this species. The specimen 7racy 5122 (NY) is annotated as having been used for the illustration of C. car- oliniana in Gleason (1952, 1963). This is the only previously published illustration of this species based on North American material. Commelina caroliniana shows a limited amount of morphological variation throughout its North American range (in comparison with that shown in India). This suggests a single introduction. If the — 86, which would be a unique number in the genus, can be confirmed and is found to be general among Amer- ican populations of this species, that would provide further evidence for a single introduction. highly unusual chromosome number Plants of C. caroliniana occur in many different habitats, such as fields, roadsides, yards, waste places, especially in moist situations, and as a weed in crops, partcularly rice, sugarcane, and corn. Flowering occurs mainly from July to November. THE STATUS OF COMMELINA GIGAS Commelina gigas was described by Small (1933) from hammocks of the Lake Okeechobee region of Florida. The plant, which has never been re- ported outside of Florida, has remained shrouded in obscurity since its description, being either ig- nored entirely (e.g., Brashier, 1966; Shetler & Skog, 1978) or listed without comment (e.g., Fa- den, 1982; Kartesz & Kartesz, 1980; Wunderlin, 1982). When I began work on the Commelinaceae of the southeastern United States, 1 recognized that Small's species required a critical reevaluation. Surveys of endangered species of Florida had called attention to C. gigas as possibly endangered 1980; Ward, 1978). This culminated in a brief paper by Austin (1985) in which C. gigas ~ Austin et al., was claimed to be no more than a giant form of the introduced weed C. diffusa. While I agree with Austin that C. gigas is conspecific with C. diffusa, I find it more distinctive than he did. This is par- ticularly true when specimens of C. caroliniana, which Austin did not recognize, are removed from consideration, and only those belonging to C. dif fusa and C. gigas are compared. Commelina giga parently ener herb that sometimes scrambles s is a robust, spreading, ap- in shrubs. Its spathes are solitary, simply folded, falcate aad very long and narrow (Fig. 3G-I). The upper cincinnus is usually exserted and one- to three-flowered. The flowers have three blue petals. The medial staminode may be lacking, or present and represented only by a filament, or with a re- duced antherode (relative to the other two), or fully developed and similar to the lateral antherodes. he medial stamen anther connective is white. The ovaries are 5-ovulate, but the capsules typically have only one or two seeds in them. The testa has a strongly raised reticulum (Fig. 2). The descrip- tions of the floral parts in C. gigas are based on a small number of flowers from two collections of the same population (Faden et al. 83/35 and Volume 80, Number 1 1993 Faden 213 Commelina in the Eastern United States TABLE 2. Differences between Commelina gigas and typical C. diffusa. COMMELINA GIGAS COMMELINA DIFFUSA (TYPICAL) Plants sometimes scrambling in shrubs Leaves 6-14 cm long, 1-3.3 ст wide Spathes 2.3-4 cm long, 2.5-6X as long as high Middle anther connective white Capsules typically 1-2-seeded (through abortion) Plants diffusely spreading, not scrambling in shrubs Leaves 1.5-5(-8) cm long, 0.5-1(-2.2) cm wide Spathes (0.5-)0.8-2.5(-3.7) cm long, 1.6-2.7(-3.3)x as long as high Middle anther connective maroon to violet Capsules 5-seeded (occasionally less, by abortion) Wunderlin & Beckner 9118), so the account of the stamens and staminodes should be considered provisional. The morphological differences between C. gigas and typical C. diffusa are summarized in Table 2. To these may be added ploidy. A plant of C. gigas (Faden et al. 83/35, voucher US) has been found to be a euploid hexaploid, with 2n = 90 сћго- mosomes, while Florida plants of C. diffusa, as noted above, were found to be diploid (2n = 30). Except for the color of the middle anther con- nective, all of the differences between C. gigas and C. diffusa are quantitative, and all can be explained by the difference in ploidy. The hexaploid is more robust and has larger leaves and spathes. The small number of seeds in the capsules of C. gigas could be due to reduced pollen fertility. Pollen from two Florida populations of C. diffusa was found to average 94% fertile (range 72—100%, N = 8); pollen from one population of C. gigas averaged 64% fertile (range 53-78%, N = 8). In cultivated plants of C. gigas few capsules are pro- duced, and those show the same low seed numbers as field-collected specimens. The seeds of C. gigas and C. diffusa are indis- tinguishable (Fig. 2), although Small (1933: 264) had erroneously described those of the former as smooth. Typically, seeds of related species of Com- melina differ in size or testa pattern, or both. This identity of C. gigas and C. diffusa seeds, and the types of magnitude of the differences between these plants demonstrate that C. gigas cannot stand as a species. However, it does have a distinct range and it is recognizable, both in the field and in the herbarium. I think that C. gigas is best treated as a variety of C. diffusa. Commelina diffusa Burm. f. var. gigas (Small) Faden, comb. nov. Commelina gigas Small, Manual Southeastern Flora, 264. 1933. TYPE (cited on p. 1503 of the same book): Florida. Palm Beach Co.: E shore of Lake Okeechobee, 9-10 May 1917, Small 8247 (holotype, NY). Commelina diffusa var. gigas is restricted to south-central peninsular Florida (Fig. 4). A map of its distribution was published by Ward (1978), showing it in all counties surrounding Lake Okee- chobee and extending southeastward into Broward County. Ward's map shows a number of county records that I was unable to confirm from speci- mens present in 1983 in the three largest Florida herbaria (FLAS, FSU, USF). Commelina diffusa var. diffusa is distributed from Virginia to eastern Kansas south to Texas, the Gulf Coast and southern Florida. Commelina diffusa var. gigas was first collected in 1913 and has been collected rarely since. Only 13 collections have been seen. Its undercollection may be due in part to its apparently sparse flow- ering—even the type specimen is sterile—as well as to its preference for wet habitats, such as cypress swamps, riverbanks and lakeshores. John Beckner (pers. comm.) stated that C. diffusa var. gigas easily could have achieved its present distribution by dispersal along waterways. Flowering specimens have been seen from March to May, August, No- vember, and December. The origins of C. diffusa var. gigas have not been determined. Either it arose in situ. through polyploidy, or else it was introduced and thus not derived from local C. diffusa. 1 believe that it is more likely to have been introduced, because direct origin of a hexaploid from a diploid is implausible, and neither tetraploids nor plants of intermediate morphology are known. However, I have not been able to find any specimens from elsewhere that exactly match the Florida plants. This is why I have recognized variety gigas as distinct. Austin (1985) wrote that he observed plants in Hawaii that were identical to C. gigas, but Wagner et al. (1990) recorded the spathes of Hawaiian C. diffusa 214 Annals of the Missouri Botanical Garden ra" < е | adn S «ww Y - N| T ] го] 2 [масом kK N lo USSELL Hens Sw [y P а 9. “+ и Sy | DOOLY c i ~ 7 te CRISP , (URN a ^ IRWIN WAYNE * z ~ ш 4 Ad a 1% Ф 2 a ya a¥ COLQUITI Au want b ~ o9 |ә = А 3 M al > CAMDEN 9 Y HENRY «y CLINCH Cuanto. ~ + је = K 2 о < [4 > «9 4 т о ~ ECHOLS ~ HA a ON ЈАСК5ОМ e g Т | e MADISON ~ “ 26, ~ ALACHUA (ТЕКУ OSCLOLA ГА , We o” 25 У zat ' FIGURE 4. Distributions of Commelina species in the southeastern United States. Commelina diffusa var. gigas (solid triangles). Commelina benghalensis (circles), Commelina forskaolii (open triangle), and Commelina nigritana ). var. gambiae (squares Volume 80, Number 1 1993 Faden 215 Commelina in the Eastern United States as 1.5-2 cm long, which is too small for variety gigas. The origin of variety gigas is likely to have been the Paleotropics, but it cannot be localized further. COMMELINA BENGHALENSIS: A WELL-ESTABLISHED WEED IN THE SOUTHEAST Commelina benghalensis, a common paleo- tropical weed, was reported by Duncan (1967) as a new species for the United States. In 1983 it was designated a “‘noxious weed," subject to strin- gent import regulations, by the U.S. Department of Agriculture (Lasseigne, 1983). Despite this fan- fare and concern, C. benghalensis was first col- lected in the continental United States in 1928 (Ruprecht s.n., FLAS), and by the mid 1930s it appears already to have become established in Flor- ida. It was first cited from Florida in an unpublished manuscript (McFarlin, 1935), and its floral mor- phology and floral biology were studied by Uphof (1934), although the plant was misidentified as C. virginica L. The earliest record of C. benghalensis from the present United States was a pre-statehood, 1990). Commelina benghalensis has been collected in at least 12 counties in Florida and three in Georgia 1909 collection from Hawaii (Wagner et al., (Fig. 4). It grows in fields, yards, other cultivated areas and disturbed sites, and especially in orange groves. This annual species can be recognized by: its funnelform spathes that are often clustered; relatively broad leaves that frequently have red hairs at the summit of the sheath; and cleistoga- mous flowers that are borne at the base of the plant and are usually subterranean (in addition to normal, aerial, chasmogamous flowers). The plants show a great deal of variation in leaf and spathe size and presence or absence of red hairs on the summit of the sheath and of an exserted upper cincinnus in the spathe. John Beckner (pers. comm.) stated that he thought there were two forms of C. benghalen- sis in Florida, differing in robustness and leaf size. A study of 25 collections of C. benghalensis from Florida and Georgia revealed two extremes of variation: plants with small, obtuse leaves, red hairs on the sheaths, and small spathes lacking an exserted cincinnus; and plants with larger leaves (at least some acute), no red hairs, spathes that frequently contain an exserted cin- and larger cinnus. Although plants exhibiting these extremes are quite different in aspect, most plants аге in- termediate. The small-leaved type occurs only at the northern limits of distribution of the species; the larger-leaved type and intermediates are more randomly distributed but are confined to Florida. It is possible that the extremes of morphological variation represent two separate introductions from the Old World that have since interbred, losing much of their distinctness. However, the morpho- logical plasticity of this weed growing under dif- ferent conditions has not been investigated, and it is possible that all of the variation observed may be due to environmental factors. А chromosome count has been obtained for a plant of Faden et al. 83/33 (US) from De Soto County, Florida, 2n — 22. This agrees with other published counts for diploids of this species (see Faden & Suda, 1980, for summary). Commelina benghalensis was recently found to be adventive in southern California (Faden & San- ders, in prep.). It is the only introduced species of Commelina in the western United States. Two RECENTLY INTRODUCED SPECIES: COMMELINA FORSKAOLII AND C. NIGRITANA Commelina forskaolii is a common species of the drier parts of Africa and southwestern Asia. It is characterized by: small, solitary spathes with fused margins; blue flowers with lyrate, winged, lateral stamen filaments; and capsules in. which usually only the dorsal locule seed develops and is shed with the dorsal capsule valve. Many popu- lations also produce subterranean, cleistogamous flowers from the nodes of decumbent shoots. Commelina forskaolii was first collected in the United States from a Dade County (Florida) San- itary Department landfill in 1980. I first learned of its existence when I was sent a specimen of Correll & Popenoe 51166, the first collection, for determination. In November 1983, Popenoe showed us the single known population, from which we obtained specimens and living material. The pop- ulation was quite extensive. It appeared to have been derived mainly through vegetative reproduc- tion by diffusely spreading stems, although some seeds were present. The plants were very typical of the species, which I know well from Africa. Cleistogamous flowers could not be found. Because this species is considered a significant weed in many parts of its range, I reported it to the U.S. De- partment of Agriculture. In 1984, it was sprayed with herbicide. Whether it is still extant in the United States is unknown. A chromosome count was obtained from culti- vated material from this population (Faden et al. 83/26, US), 2n = 30, which is usual for this species. I first became aware of another unusual species of Commelina in Florida in 1978 when I was asked what an orange-flowered species in Florida might 216 Annals of the Missouri Botanical Garden be. No specimens were forthcoming, and it was not until several years later that 1 was reminded of this incident when І received a draft of the Commeli- naceae for Wunderlin (1982). The name Com- melina umbellata enth., caught my um. a 1983, accompanied by Wunderlin and John Beckner, I collected three populations of C. nigritana from roadsides in Hardee and Polk counties, Florida. готте та. nigritana is a widespread African Thonn., а synon of C. ni- gritana savanna species and occasional weed. It belongs to a group of annual species with fused spathes and buff- to orange- or apricot-colored flowers. These species also share the unusual basic chromosome number x — 14. Two varieties of C. nigritiana are recognized by Brenan (1968), tana, with one-seeded capsule locules and pitted variety nigri- seeds, and variety gambiae (C. B. Clarke) Brenan, with two-seeded ventral locules and reticulate seeds. All Florida plants belong to variety gambiae, the more narrowly distributed form, which occurs in west Africa from Senegal to northern Cameroon. Commelina nigritana var. gambiae was first collected in Florida near Bradenton, Manatee d in 1976 (Wunderlin 5721, FLAS, TENN, SF). Two years later it was collected in Broward uus on the opposite side of the peninsula. It is now known from at least six counties (Fig. and appears to be spreading rapidly. It has been collected almost exclusively along roadsides, and it is likely to be dispersed by mowing. So far it has not been collected in crops, so its potential to be- come an important weed is unknown. Commelina Ккү TO THE UNITED STATES SPECIES OF COMMELINA la. Spathes with margins free to the bas 2a. Perennials with erect to мге: чы shoots; roots tuberous; leaves linear to linear-lanceolate .... li nigritana var. gambiae flowers from September to December. A Florida plant of C. nigritana var. gambiae (Faden et al. 83/27, US) was found to be tetra- ploid, with 2n — 56 chromosomes. The only pre- vious counts for this variety, based on African material, were 2n — , with occasional root tip cells in one population having 56 chromosomes (Монок. 1967) DISCUSSION Among the nine species of Commelina in the United States, six are not indigenous: C. bengha- lensis, C. caroliniana, C. communis, C. diffusa, C. forskaolii, and C. nigritana. Only C. dianthi- folia, C. erecta, and C. virginica are native. The recent discoveries of C. forskaolii and C. nigritana in Florida and of C. benghalensis in California should alert botanists, especially in the South, to collect any unusual-looking plants of this genus. Commelina, with 170 species, is the largest genus of Commelinaceae and its species are often dificult to identify. Capsules and seeds are fre- quently critical, so they should always be looked for when making collections of this genus. Flowers should be pressed separately between absorbent paper or cellophane. Floral features that should be particularly noted are: lateral sepal fusion (when present); paired petal color; lower petal degree of development and color; number of staminodes and color of the antherodes; and color of the anther connectives and pollen. dia ianthifolia 2b. Annuals or perennials usually with decumbent to scandent shoots; leaves narrowly EPEE to ova 3a. Spathes generally whitish or pale green towards the peduncle with contrasting, dark green veins; middle petal white or paler than the others; capsules bilocular; seeds rugose E v s ae ommunis 3b. Spathes lacking contrasting oi middle жа + concolorous with the others; capsules lala seeds reticulate or smooth to intly alveolate 4a. Spathes not at all to slightly falcate; upper cyme usually vestigial (rarely well developed and 1-flowered); seed s smooth to faintly alveolat caroliniana 4b. Spathes usually distinctly ai upper cyme in larger spathes usually well developed an 1-several-flowered; seeds reticulate diffusa lb, Spathes with margins fused ‘basally y. 5a. Flowers apricot- or peach-colored 5b. Flowers blue (rarely lilac to lavender or white 6a. Leaves with auricles at summits of sheaths; roots stout; middle petal white, inconspicuous; locules all 1-seede 5. C. nigritana erecta 6b. Leaves not auriculate; roots thin; middle petal blue to lilac or lavender, conspicuous; some NUM usually 2-seede Та. Perennials with erect to ascending shoots; sheaths with red hairs at the summit; lamina 6- cm long; spathes 1.5-3 cm long; underground, cleistogamous flowers lacking ....................... 7. C. virginica Volume 80, Number 1 1993 Faden 217 Commelina in the Eastern United States Tb. Annuals or perennials with ascending to Wii ww repent or scrambling shoots; sheaths tho with or wit 2-7(-9) cm long; spathes 0.5-2 cm long; rs sometimes phia amina ovate to lanceolate—elliptic; sheaths often with red hairs at summit; lateral stamen filaments pon winged; capsules usually 5-seeded, the ventral locule seeds develope . benghalensis со c . Lamina oblong to lanceolate-oblong or oblong-elliptic; sheaths lacking red hairs; lateral 8 tamen filaments winged; capsules usually 1-seeded, only the dorsal locule seed developed C. forskaolii LITERATURE CITED AUSTIN, D. F. 198 = gigas rediscovered and lost. Palmetto 5(4): 1 . N. Jones, B. E. us & C. E. NAUMAN. 1980. Endangered Plant Species Survey of Southern Florida. Fish and Mas Service, Office of Endan- ered Species, Atlant BRASHIER, C. K. 1966. А revision of Commelina (Plum.) L. in the U.S.A. Bull. Torrey Bot. Club. 93: 1-19. BRENAN, J. P. M. 1968. 0 in F. N. а" ки. Flora of West Tropical Africa, ed. 2, 3, ondon. DuNCAN, W. H. 1957. species new in the United States. UE 19: 2 FADEN, R. B. 1 Commelinaceae. Pp. m National List of Scientific Plant Names, vol. E List of Plant Names. Soil Conservation Service, United States Department of Agriculture, U.S. Government ap Office, Washington 1989. Commelina posu. (Commelina- pee A misunderstood species in the United States is an old introduction. Taxon 38: 43-53. UDA. 1980. Cytotaxonomy of Com- elias: Chromosome numbers of some African and Asiatic species. Bot. J. Linn. Soc. 81: 301- ни GLEASON, H. A. The New Britton and Brow Illustrated Flora KS the Northeastern United у and Adjacent Canada. New York Bot. Gard., New York. . 1963. The New Britton and Brown Illustrated Flora of the Northeastern United States and Adjacent Canada, third printing, slightly revised. Hafner Publ., . CRONQU ist. 1991. Manual of Vascular Plants a Northeastern United States and Adjacent Canada, second edition. New York Bot. Gard., New York. KanrEsZ, J. Т R. Kartesz. 1980. A Synonymized Checklist of the Vascular Flora of the United aun Canada, and Greenland. Univ. of North Carolin Press, Chapel Hill. LASSEIGNE, А. 1983. Noxious bien of the езе Noxious Weed Act, No. 21: Commelina p hs States Department of үка мча bene ng С. ntral jn of Mc —— J. B. Flora of the e Lake Region of Florida. Prepa T but not sub rdum as a Ph.D. Dissertation, nee of Mich- igan. Manuscript at the University of Florida, Gaines- ville. Morton, J. K. 1967. Africa: A oo survey. 60: 167-221. S A. H. 1968. Commelinaceae. Pp. 2 і n А. Н. Radford, Н. E. Ahles & С. К. Bell (editors), Маша! of the Vascular Flora of the Carolinas. Univ. of North Carolina Press, Chapel Hill. The Commelinaceae of West Ј. Linn. Soc., Bot. Rao, R. S., R. V. KAMMATHY & R. S. RAGHAVAN. 1968. Cytotaxonomic studies on Indian Commelinaceae: A review. J. Linn. Soc., Bot. 60: 357-380. SHETLER, S. С. & L. E. Skoc (editors). 1978. А Pro- visional Checklist for Flora North a (revised). Monogr. Syst. Bot. Missouri Bot. Gard. SMALL, J. К. 1933. Manual of the Renee de Flora. Univ. of North Carolina Press, Chapel Hill. STEYERMARK, J. A. 63. Flora of Missouri. Iowa State Univ. Press, Ames Орноғ, J. C. 1934. ` Vergleichende blütenmorpholo- gische und iii i qi Studien an Commelina virginica. Ber. Deu wi Bot. Ges. 52: 173-180. Wacner, W. L., D. R. Her T & S. Н. SOHMER. 1990. Manual of the M eris Plants of Hawai'i, vol. 2. Univ. of Hawaii Press, Honolulu. Warp, D. B. (editor). 1978. Rare and Endangered Biota of Florida, vol. 5. Plants. Univ. Presses of Florida, Gainesville. WUNDERLIN, В. P. 1982. Guide to the Vascular Plants of Central Florida. Univ. Presses of Florida, Tampa, St. Petersburg, Fort Meyers, Sarasota. APPENDIX. Specimens of Commelina caroliniana seen from the United States. Because of the difficulty in identifying Commelina car- oliniana with certainty, all American specimens seen of sheet have the additional number(s) in brackets. Allen 12665 (NLU), Allen s.n. (NLU), Anderson s.n. (MO#1967952), Ahles 36000 (NCU), Ahles & Bell 18355 (NCU), 21053 (NCU), Ashbey 18 (GH, MO), 38 О € Bell 16222 (NCU), 16223 (TENN), Brown 1158 (LSU, NLU), 5946 (LSU), 20180 (LSU), Bush 3307 (GH, NY), 3332 (GH, NY). Claycomb s.n. (LSU), Cormier 446 (NLU), Correll 36514 (GH), Cory 11038 (GH), 11461 (GH), Curry 310 (LSU), Curtiss 2992 (F, PH), 4144 (MO, NY, US), 5177 F, FSU, GH, MO, NY, US), 6273 (FLAS, NCU), s.n. (MO#1749041), s.n. (US#223467), Cuthbert s.n. (FLASH5472). Duke 228 (NCU), Duncan 30556 (CLEMS), 30601 (NCU), Dutton & Taylor 217 [5334a] (NLU). Elliott s.n. (CHARL). Faircloth 6250 (MO), US# 2216458), s.n. (GH). Garber s.n. (US#33708), Genelle € Fleming 2207 USF), Gibbes s.n. (NY), Ginn 16 (LSU), Graves 1938 (MO). Hoffman 12 (FLAS). ~ (FLAS#6661, Felix s.n. ~ 218 Annals of the Missouri Botanical Garden Langlois s.n. (GH, PH). Mellichamp s.n. (US#936363), Mellinger s.n. (GH), Miles 147 (FLAS), Montz 122 (LSU), 3646 (LSU). Nuttall s.n. (PH). Pias & Thomas 4482 (LSU), Pias et al. 4417 (LSU). Ravenel s.n. (NY), Reed 41364 (US). Shinners 21414 (FSU, GH, руг Small 241 (LSU). Tharp 1526 (F), s. (LSU), 132 (LSU), 202 (LSU), р (LSU), Thieret 9588 (FSU), 16367a (VDB), Pera 80000 (NLU), 86722 (05), Thomas & Allen 47784 (NLU), 79277 (NLU), 87174 (NLU), 98115 (MO, NLU), 98208 (NLU), 98330 (NLU), 98444 (NLU), Thomas & Briley 69679 [900] (NLU), Thomas & Laird 50479 [2826] (NLU), Thomas et al. 26933 (NLU), 56572 [2689] (NLU), 61035 [1570] 6252] (NLU, US), 91099 [3944] (NLU), 94030 [13236] (NLU), Thompson s.n. (US#1537440), Thompson & Nishida 3045 (US), Toler 45 (UARK), Tracy 5122 (F, alter (or Fraser) s.n. (BM- Walter, lectotype of Faden, 1989), Warnock et al. 46372 (NCU), Williams 56 (NLU), no collector or number (PH, Muhlenberg Herbarium No. 92, pro parte). TRADESCANTIA CRASSIFOLIA (COMMELINACEAE), AN OVERLOOKED SPECIES IN THE SOUTHWESTERN UNITED STATES! Robert B. Faden? ABSTRACT Tradescantia a (Commelinaceae), a widespread species in Mexico, is reported from the United States for the first time, on two nineteenth-century that from New Mexico i is deemed al collections. The specimen from Texas is considered reliably localized; Tradescantia crassifolia Cav., a common spe- cies known from Mexico to Guatemala, has not been reported from the United States. In the course of preparing the Flora of North America treatment of Commelinaceae, 1 discovered two old specimens that document this species in the Southwest. 1 call attention to this record and encourage local bot- anists to seek this species in order to determine whether it is still extant in the United States. The specimens seen are: UNITED STATES. TEXAS: Chinati [Chenate] Mountains, V. Havard, October 1880 (US sheet #225303) (Fig. 1). NEW MEXICO: Without specific locality, Edwards, 1848 (NY). It is probable that the locality of the Texas specimen is correctly stated. Valéry (often written without the accent mark of the original French) Havard (1846-1927), an Army physician with a background in botany, collected plants in western Texas from 1880 to 1884 (Lanjouw & Stafleu, 1957; McVaugh, 1957). McVaugh, quoting monthly reports filed by Havard (originals in the National Archives), noted that in Havard's report of 30 September 1880 he mentioned a trip made to the Chinati Mountains, and that in a later report he recorded being in these mountains 20-24 De- cember the same year. McVaugh concluded that Havard probably gained most of his knowledge of the Chinati Mountains between September and De- cember 1880, when he was Post Surgeon at Camp Eagle Nest in Presidio. Havard’s (1885) “Report on the flora of western and southern Texas” in- cluded a brief account of the flora of the Chinati Mountains, demonstrating that he clearly visited the area and could have collected plants there (publication of this in the Proceedings of the U.S. National Museum was a direct quid pro quo for Havard's donation of his collections to the Smith- sonian Institution, according to correspondence in the Smithsonian Institution archives). There is no evidence that Havard ever collected in Mexico, the only other possible origin of this specimen. Hunt (in litt., 4 Oct. 1990) stated that a range extension for 7. crassifolia into Texas and New Mexico would not be unreasonable. In the U.S. National Herbarium there are two Mexican col- lections of T. crassifolia from the Sierra del Car- men, Coahuila, a mountain range very close to Big Bend National Park, Texas, and six collections ican localities are in the same U.S.D.A. diness Zone (Zone 8) (Cathey, 1990) as the Chinati Mountains, which could thereby be expected to support many of the same plant species. Johnston (in litt.) has noted that several Mexican species are known in the United States only from the Chinati Mountains, including Calliandra coulteri S. Wat- son (Fabaceae) and Argythamnia astroplethes J. Ingram (Euphorbiaceae). The New Mexican collection is more problem- atical. The collector Edwards is Lewis Allison Ed- wards (fide R. Barneby via M. Nee) (1823-1877), an Army surgeon who was stationed near Monter- 1846- 1848, and sent plant collections to John Torrey. ican War, rey, Mexico, during the Mex ' I thank reviewer Marshall Johnston and an anonymous reviewer for helpful suggestion en of Botany, National Museum of Natural History, Smithsonian Institution, -— D.C. 20560, .S.A ANN. MISSOURI Bor. GARD. 80: 219-222. 1993. 220 Annals of the Missouri Botanical Garden FicunE 1. SITED BY TI Locality : Collector : a v. DL = S). Tradescantia crassifolia Cav. (Havard s.n., 1 LASA. UNITED STATES NATIONAL HERBARIUM. Y^" d. € ef, Volume 80, Number 1 1993 Faden 221 Tradescantia crassifolia in the Southwest- ern United States He is commemorated by Cracca edwardsii A. Gray (fide R. Barneby). Edwards was based in Santa Fe after the Mexican War. On 29 June 1848 he wrote to John Torrey (archives New York Botanical Gar- den) from Monterrey, Mexico, that “We have just received orders to proceed to California & Santa Fe... . We start in about 14 days, passing through Parras, the great vine growing country, Chihuahua, Paso del Norte, &c." Edwards wrote that he hoped to “make amends” during the next 10 months for not having collected many specimens recently. Ed- wards's actual itinerary was detailed in a 20 Oc- tober 1848 report (original in National Archives) sent to Thomas Lawson, Surgeon General, U.S. Army, from Santa Fe: “The command left Mon- terey [sic] on the 25th of July, arriving at the village of Castanuelas on the 31st, a distance of 115 miles. On the 26th of August we reached Chihuahua, remaining at that city until the lst ultimo. On the 30th of September we encamped at the settlement of Sa Parida, distant from Chi- huahua 376 miles. On the 10th inst. at this place, making in all a distance of about 1050 miles." He was stationed at Santa Fe until September 1850 (Woodward, 1879). ether Edwards actually collected plants in we arrived New Mexico is uncertain (e.g., he is not included in McKelvey, 1955). Rodgers (1942: 219), stated collectors who collected in the vicinity of Santa Fe, these collec- that Edwards was one of several tions forming the basis of Plantae fendlerianae (Gray, 1849), but a perusal of this work revealed that the only Edwards's collections cited are from Mexico. The only other evidence is the label on the specimen of 7. crassifolia cited above. The problems with that information are: (1) “New Mex- ico" (as of February 1848) was the newly ceded ew Mexico Territory, consisting of most of the present states of New Mexico and Arizona, as well as a small portion of Colorado, and some of western Texas (Wooton, 1906) (the locality is thus rather vague); and (2) the Santa Fe area, where Edwards was based and would have had time to collect plants, appears to be climatically unsuitable for this plant (R. Spellenberg, pers. comm.). The specimen most likely came from further south and was col- lected en route to Santa Fe. Edwards passed through appropriate habitats for this species (and at the right time of year to find it) around Cd. Chihuahua, where the plant is known to occur. Thus it could have been collected in Mexico and later mislabeled. Hence, the New Mexico record is unproven. Tradescantia crassifolia Cav. (Fig. 1) is easily recognized by a number of features: tuberous roots; distichous, lanceolate to ovate leaves; cobwebby, white pubescence on the vegetative parts (especial- ly the leaf sheaths); sessile cyme-pairs in the axils of the upper 1—5(-7) leaves; pedicels and sepals pilose with eglandular hairs. Although this species is very variable in Mexico, especially in shoot length, leaf shape, and amount of pubescence on the veg- etative parts, it is unlikely to be confused with any of the five other species of Tradescantia that occur in Trans-Pecos Texas and/or New Mexico. All of these species have some or all of the cyme-pairs stipitate and subtended by foliaceous or spatha- ceous bracts. In addition, three of them, T. wrightii Rose & Bush (Texas and New Mexico), 7. pine- torum Greene (New Mexico), and T. occidentalis (Britton) Smyth (Texas and New Mexico), have linear-lanceolate, glabrous (to scabridulous) leaves, and glabrous to glandular-puberulent sepals. The other two west Texas species, both formerly placed in Setcreasea (see Hunt, 1975), T. brevifolia (Tor- rey) Rose and 7. leiandra Torrey, have terminal or leaf-opposed inflorescences, clawed petals that are gamepetalous basally, and glabrous filaments. In addition, 7. leiandra has sepals with glandular or eglandular hairs, and T. brevifolia has glabrous sepals. Tradescantia crassifolia is the only species of section Mandonia D. R. Hunt (Hunt, 1980) to occur in the United States. After this manuscript was completed and re- viewed I had the opportunity to reexamine Ed- wards's collection at the New York Botanical Gar- den. The sepals and pedicels of the specimen have numerous glandular hairs, a feature that does not agree with 7. crassifolia or with any other species of section Mandonia (Hunt, 1980). Thus, this pu- tative record of 7. crassifolia from New Mexico is even more doubtful. In view of this new infor- mation, I subsequently restudied Havard’s speci- men and reconfirmed that it is indeed T. crassifolia. Although the specimen has only sparsely pubescent leaves (unlike many Mexican collections), it does not match 7. crassifolia var. glabrata C. B. Clarke, which has the lamina glabrous except for a marginal band of dense, white, woolly hairs. LITERATURE CITED САТНЕҮ, Н. M. 1990. USDA Plant Hardiness Zone .5.р.А., Agric. Res. Service, Misc. Publ. GRAY, А. . Plantae fendlerianae. Mem. Amer. Acad. Arts, ser. 2, 1: 1-116. Havanp, V. 1885. Report on the flora of пон а and southern Texas. Proc. U.S. Natl. Mus. 8: 4 939. Hunt, D. R. . The reunion of НА and Separotheca with тт" American Com- melinaceae: I. Kew Bull. 30: 443-458. 19 Sections and series in Tradescantia. 222 Annals of the Missouri Botanical Garden American Commelinaceae: IX. Kew Bull. 35: 437- 442. Lanjouw, J. . А. STAFLEU. 1957. Index оа riorum, dg 2(2), Collectors E-H. Regnum Veg. 175-295. McKELvEY, S. D. . Botanical Exploration of the Trans-Mississippi West к Arnold Arbo- E. 5 e tp D £ о = Personal Pli ad and military record of Colonel Valery Havard, Medical Corps, U.S. Army, with a brief evaluation of Havard's bo- tanical work. Asa Gray Bull., n. s., 3: 177-191 ADI 1942. John Torrey, a Story of orth American Botany. Princeton, New Jerse Woopwarp, J. J. 1879. Lenis A. iip ^ men Princeton Univ. Press, cud ир ie ae by Charles ‘Wright, Bull. Torrey Bot. Club 33: 566. EL GENERO /NGA Mario Sousa S.? (LEGUMINOSAE: MIMOSOIDEAE) DEL SUR DE MEXICO Y CENTROAMERICA, ESTUDIO PREVIO PARA LA FLORA MESOAMERICANA?} RESUMEN n la preparacion del género Inga para la Flora Mesoamericana se discuten someramente la nomenclatura y taxonomia de cada especie, su variación, asi como la posible influencia э hombre. Fue necesario lectotipificar siete nombres: /nga alba, 1. densiflora, I. multijuga, 1. oerstediana, I. paterno, 1. ruiziana, e 1. vera. También se cita material representativo de todas las áreas donde habita cada especie, эы а las de fuera del área de la Flora. Se describen trece especies nuevas para la ciencia: /nga bella, 1. cabrerae, 1. calcicola, 1. ү І. chiapensis, І. dasycarpa, 1. davidsei, I. dwyeri, 1. ismaelis, 1. lacustris, 1 pseudoinvolucrata, І. sinacae, e І. tenella. También se hizo un cambio nomenclatural de /nga cocleensis var. megantha a 1. cocleensis subsp. se 9н ABSTRACT recursor to the treatment of /nga for the Flora Mesoamericana, this summary of the genus is a including the nomenclature, taxonomy, and variation of all the species, as well as the possible influence of hum in altering patterns of distribution and variation. Seven names are lectotypified: Inga alba, I. densiflora, І. multi sped cuspidata, I. Poncii ad dasycarpa, 1. davidsei, 1. dwyeri, 1. ismaelis, I. lacustris, L pseudoinvolucrata, Е sinacae, and Í. tenella. А nomenclatural change was made for Inga cocleensis var. megantha to І. cocleensis subsp. megantha Para la preparación del tratamiento del género El género ha sido tratado en forma integral por Inga (Leguminosae: Mimosoideae) para la Flora Bentham (1845, 1875) y para Mesoamerica por Mesoamericana, que abarca del Sur de México a Pittier (1916, 1929) y León (1966), y en floras Panama, ha sido necesario realizar este articulo locales como la de Costa Rica por Zamora (en que formaliza taxa y da información complemen- prensa) y la de Panamá por Schery (1950). En el taria y aclaratoria, que las normas editoriales de último tratamiento del área, León (1966) incluye la Flora no permiten. 49 especies, de las cuales nosotros estamos reco- E agradece al asi Botanical Garden que por medio de su beca, financiada por la Fundación Jessie Smith Noyes, asi como de la ayuda de la Dirección General | de Personal Académico, UNAM, que hicieron posible, durante mi ano sabático, ku la estancia para esta investigación. En el Missouri Botanical Garden mi trabajo fue facilitado grandemente por Enrique Forero y los Davidie, Gerrit y Jeany, tanto en aspectos académicos como logisticos. Agradezco a James Zarucchi shes sus Aperira sugerencias al manuscrito; a Roy Gereau por mejorar el manuscrito, agregando información útil; a Nelson Zamora que hizo asequible su investigación inédita en Inga; a Bruno Manara de Inga vera; a Terence D. Pennington por facilitar mi acceso a varios prestamos de Inga en Kew; y a los curadores de A, BM, BR, CAS, СМ, CR, CHAPA, F, EAP, ENCB, С, GH, IEB, К, LL, MEXU, MICH, MO, NY, 5, TEFH, TEX, U, US, W, WIS y XAL por facilitar el material botánico para su estudio * Herbario Nacional, айни de Biología, UNAM, Apartado Postal 70- 367, Coyoacán, 04510, México, D.F., Mexico ANN. MISSOURI Bor. Garb. 80: 223-269. 1993. 224 Annals of the Missouri Botanical Garden nociendo 47, aunque no necesariamente con los mismos nombres; de las 81 especies que aqui se tratan, tal crecimiento en el nümero de especies, fundamentalmente, por el gran acümulo de nuevas colectas que se han hecho en el área en los ültimos 25 anos, que han incorporado tanto nuevas especies para la ciencia como varias otras solo conocidas para Sudamérica. Este nuevo ma- terial también ha permitido interpretar mejor la variación de las especies antes sólo conocidas por escaso material, lo que en algunos casos permite sinonimizar binomios de uso local en nuestra área. Sin embargo, lo anterior no significa que ya el tema esté agotado, por el contrario, como hay varios ejemplares que podrian ser nuevos taxa, que no creo conveniente formalizar por lo incompleto del material, y además algunas especies ya descritas sólo son conocidas por el tipo. Respecto al material representativo se decidió dar mayor información de la que permiten las nor- mas editoriales de la Flora Mesoamericana; ésto es debido a que la Flora sólo autoriza usar una colecta por cada una de las diez regiones en las cuales está dividida la flora, además de una refe- rencia muy corta de sus limites geográficos extra mesoamericanos. Asi se decidió ampliar esta in- formación por considerarse ütil tanto en cuanto al concepto de especie como por sus obvias impli- caciones de uso y distribución regional. También se cita el material de fuera de la región meso- mericana por considerase de interés dado que no existe actualmente un tratamiento integral equi- valente en América del Sur, sólo trabajos regionales como de Bentham (1876) para Brasil, Poncy (1985) para la Guayana Francesa, y de Macbride (1943) para el Perü. Asi se cita un ejemplar para cada uno de los estados de México y departamentos, distritos, estados o provincias de cada pais cen- troamericano, asi como de los de Sudamérica, ha- ciendo excepción de sólo una colecta para cada una de las Guayanas y paises de las Islas del Caribe. Brasil se dividió en regiones dado la amplitud del pais. El formato de las descripciones sigue al de la Flora Mesoamericana; sin em nae en general los caracteres usados para el género son los em- pleados por Pittier (1916, 1929), León (1966), y Poncy (1985), aunque no necesariamente su ter- minologia. Asi de Poncy adoptamos el concepto de las brácteas homomorfas y heteromorfas que ella maneja en sus descripciones e ilustra en sus là- minas; además aqui estamos empleando en forma extensiva el uso de botones florales que parece ser un carácter ütil en la taxonomia de /nga. Los botones florales sólo son citados ocasionalmente como еп /. acuminata (León, 1966), pero en gene- ral no son descritos, ni menos usados como un carácter diagnóstico. En la descripción de las es- pecies nuevas y de las del tratamiento de la Flora este carácter es usado extensivamente, cuando co- nocido. Su uso es referido como "botones con el "abierto" o cáliz cubriendo" y las opciones son "cerrado"; aqui hay que aclarar que se refiere al estado de desarrollo del botón floral cuando el ta- mano del cáliz es aproximadamente un Уз menor al del tamano normal de una flor en antesis, siendo ésto importante ya que, todos los botones realmente incipientes cuentan con un cáliz cerrado. Asi se pueden categorizar los botones florales (con algunos ejemplos claros) de la siguiente forma: Cáliz cerrado, botones casi esféricos: Inga gold- manii, 1. rubiginosa. 2. Caliz cerrado, botones ovoides а obovoides a tubulares: /. calcicola (véase Fig. 1D), /. exa- ata. 3. Cáliz cerrado, botones rostrados, los lóbulos uni- dos en uno: /. acuminata, 1. urabensis, 1. ve- nusta. 4. Cáliz abierto en un poro, botones tubulares: /. leonis, 1. sierrae. 5. Cáliz abierto, tubular, los lóbulos aplicados a la corola, la que crece con el cáliz: /. sinacae, 1. thibaudiana. aliz abierto, tubular, los lóbulos conniventes en sus ápices, el botón obtuso en el ápice, la protocorola más pequena que el cáliz: /. fas- tuosa, l. mucuna. 7. Cáliz abierto, márgenes, el botón rostrado, la protocorola más los lóbulos conniventes en sus па que el cáliz: /. pavoniana, І. tonduzii, I. ve 8. Caliz p RN los lóbulos erectos no conniventes, la protocorola más pequeña que el cáliz: /. cook- ii, 1. ismaelis (véase Fig. ДА, E), I. setosa. Respecto a los agrupamientos supraespecificos en /nga, el sistema de Bentham (1845, 1875, 1876) modificado por León (1966), necesita más refinamientos, sobre todo en los limites de la sección Leptinga con la sección Bourgonia (ver discusio- nes de /nga acrocephala, 1. alba), basados en el tipo de inflorescencia, la cual puede variar de ca- pituliforme a racimos cortos. Por otro lado en la sección Leptinga hay varios grupos naturales como él de alrededor de /. cordistipula incluyendo: /. bella, I. portobellensis, І. pseudoinvolucrata, e l. spiralis. Tambien кен a I айне о зе puede agrupar а otr g ip У COMO: Volume 80, Number 1 Sousa S. 225 El Género Inga I. brevipes, I. polita, I. sciadion, e І. umbratica. Con 1. saffordiana quedaria I. filiformis, este úl- timo grupo muy natural que ha divergido gran- demente del resto de la sección. Como sugirió Leon (1966: 344) todos estos grupos no parecen tener parentesco entre ellos y el tipo de inflorescencia parece más bien un paralelismo. De los agrupamientos existentes, quizá uno de los más naturales es la serie /nga, definida fun- damentalmente por la legumbre, pero que a la vez cuenta con las especies más variables y panmixicas del género. La sinonimia la hemos tratado cuando en forma original estamos agregando información nueva no tratada por Schery (1950), León (1966), o Poncy (1985); asi estamos excluyendo la citación de los cambios que Kuntze (Revis. Gen. Pl. 1: 182-189. 1891) hizo a Feuilleea. En algunos casos se dan sinónimos ya reportados por otros autores, pero se toma esta libertad para aclarar problemas. También siempre se apunta el basiónimo cuando éste existe, o el homónimo tardío en el cual se basa el nombre en uso. Las claves fueron elaboradas tratando de agru- par especies afines y por ello su uso se circunscribe a la identificación de especies y en forma limitada a ejemplares, a este respecto hay que apuntar que en una gran proporción los ejemplares sólo cuentan con un estadio, es decir en flor, fruto o, desde luego, estériles. Sin embargo, la elaboración de las claves de sólo flores o frutos nos llevó frecuente- mente a por un lado, ubicar a una misma especie en varias entradas, y por otro lado, agrupar varias especies al final de una dicotomía. Dado el nivel de conocimiento del género Inga en Mesoamérica no estamos formalizando cate- gorias infraespecificas, con una excepción; sin em- bargo cuando hemos creido que deberia hacerse, lo mencionamos en cada caso, dejando la decisión final para quien revise al género en conjunto y asi lo uniformice. La fitogeografia de /nga en Mesoamérica mues- tra un importante elemento endémico de 33 es- pecies (41.25%) incluyendo al área fitogeográfica del norte, o sea el sur de Veracruz y la zona de Tuxtepec en Oaxaca; si a ella agregamos la otra área natural al Sur, es decir la zona del Chocó colombiano y ecuatoriano, sumariamos 11 especies más o sea un total de 44 especies que representan el 55% de las Ingas del área. Los otros elementos importantes son las 12 es- pecies amazonicas (15%); las cinco de la flora de Mexico (6.25%); brazo Colombia-Guayanas y el otro Colombia-Bo- las siete del arco que forma un livia (8.7%), generalmente en areas montanosas; las cuatro Colombia-Venezuela y en ocasiones Trin- idad (5%); y otros modelos de distribucion mas amplia que sólo representan a ocho especies o sea el 10% de las Ingas. Para este análisis estamos considerando a 80 especies, ya que /nga edulis muy probablemente se trata de una especie introducida al área meso- americana. (26.47%) son del area de bajas altitudes de la vertiente del Del elemento endémico, Golfo de México; de las áreas montanosas, cinco especies (14.596) corresponden a Chiapas-Hon- duras y siete especies (20.58%) a la de Nicaragua— Panamá, estando las siete presentes en Costa Rica; de las áreas de baja altitud, hay dos especies (5.88%) de Veracruz-Oaxaca a Costa Rica-Panamá, dos especies (5.88%) endémicas a Costa Rica, tres especies (8.82%) a Costa Rica-Panamá, y seis especies (17.65%) sólo de Panamá. Inga Miller Gard. Dict. Abr., ed. 4, 2. 1754. Arboles inermes; ramas teretes a tetragonales, glabras a pelosas. Hojas alternas, estipuladas, | -pinnadas, paripinnadas, 1—14-yugadas, los folio- los opuestos; raquis teretes a alados, con glándulas interfoliolares, frecuentemente rematando en un apéndice. Inflorescencias generalmente axilares en fascículos o paniculadas en ramas afilas, las flores uniformes, en racimos, espigas, capítulos o um- belas; brácteas florales homomorfas a heteromor- fas, fértiles a estériles. Flores sésiles a largamente pediceladas; cáliz sinsépalo, campanulado a tubu- lar, con o sin escotaduras, 5-lobado, en botón abier- to a cerrado; corola simpétala, tubular, subturbi- 5-lobada, estambres monadelfos, con el tubo exerto a in- nada a turbinada, pelosa; numerosos, serto, las anteras eglandulares; el polen en polia- dras; gineceo |-carpelar, el ovario multiovular; es- tilo más largo que los estambres, el estigma simple o lobado. Legumbre aplanada, tetragonal o sub- terete, derecha, curveada, helicoidal a circinado- espiralada, coriácea a lenosa, glabra a pelosa, las valvas en ocasiones reducidas y entonces los már- genes hipertrofiados; irregularmente dehiscente; semillas carnosas, oblongas, desnudas o en ocasio- nes cubiertas por una delgada testa papirácea que a su vez está cubierta por un tejido endocárpico suculento que ha sido interpretado como arilo, sin linea fisural; cotiledones bien desarrollados, la ger- minación frecuentemente ocurre en el fruto. Aprox. 350 spp. Trópicos y subtrópicos húmedos de Ате- rica. 226 Annals of the Missouri Botanical Garden GRUPOS la. Inflorescencias racemosas o espigadas. 2a. Cáliz campanulado a brevemente tubular. За. Cáliz 1-1.5 mm; legumbre glabrescente a glabra. GRUPO 1 3b. Cáliz 3-6 mm; MS BM pelosa. GRUPO 5 2b. Cáliz tubular a ciatifor a. Legumbres aplan das a tetragonales. 5a. Lóbulos del cáliz triangulares a atop 6a. india maduras glabras a pelosas, pero no tan densamente que impida ver el color as valvas, con indumento más o menos rigido. та. "Cá " Ра А 2 n. generalmente sin escotaduras, en botón obtus . GRUPO 2 7b. = da m, generalmente con escotaduras, en botón dis о > apiculado a a GRUPO 3 6b. Centres maduras con pelosidad densa que da el color al fruto, generalmente hirsutas. UPO 4 Lobulos del caliz cuspidados a caudados. GRUPO 6 4b. Legumbres subteretes, acor ds я СКОРО 7 lb. Inflorescencias capituliformes o um 8a. In ores cencias capituliformes, con ieee sésiles o pediceladas, los pedicelos hasta 3 veces la longitud G 08 el cáliz. 8b. Plena umbeliformes, los pedicelos de más de 3 veces la longitud del cáliz. 000 GRUPO 9 GRUPO 1 Flores pequeñas, cáliz 1-1.5 mm, менше а brevemente tubular; legumbres aplanadas, glabrescentes а glabras. la. Каши жн аыр más de 2a. e la legumbre nadas За. Corola enne infundibuliforme, pelosa, tubo estaminal inserto o cortamente exerto. 1. longispica 3b. ка Jerar campanulada, generalmente glabra, tubo estaminal largamente exerto (en el Con- tin ericano, inserto en las Antillas). 1. laurina 2b. Valvas EN la leg umbre abolladas o con proyecciones cónicas a la altura de las semillas. 4a. Legumbre fisa, abollada a la altura de las semillas. 5a. oe iolos hasta 12 cm de ancho; raquis foliar terete a marginado; bracteas florales triangular- tu І. coruscans 5b. Foliolos hasta 5 cm de ancho; raquis foliar marginado a alado; brácteas florales aleznadas. semialata тые 1. riopalenquensis 4b. Legumbre dn p con proyecciones cónicas a la altura de las semillas. lb. Raquis floral corto, hasta 2.5 c a. Flores ioa о casi ы. a. Pedünculos florales más Е 1.5 ст; tubo estaminal menos de 1.5 veces más largo que Іа согоја; le generalmente menos de 20 m. 8a. Glandulas енор аго urceoladas, el cuerpo glandular 0.7-1.0 x 0.7-1.0 mm; legumbre asta 2.3 cm de ancho, las valvas aplanadas a la altura de las semillas. _________. 1. belizensis 8b. Meis pl а cupuliformes, el cuerpo glandular 1.5-2.1 x 1.5-2.1 mm; legumbre has m de ancho, las valvas abolladas a la altura de las semillas. ........................ I. pezizifera Tb. Pedimos iuis bab 1.1 cm; tubo estaminal 2 veces o más largo que la corola; la generalmente más de 25 m. 1. alba 6b. Flores не pediceladas. І. heterophylla Grupo 2 Caliz 2-5 mm, generalmente sin escotaduras, en botón obtuso. la. Brácteas florales heteromorfas, las distales cuculado-estipitadas, las proximales ovadas a obovadas, oblongas o triangulares. 2: dns foliar terete a к» pelosidad de las ramas esparcida. Hojas (4-)6(-8)-folioladas І. acrocephala M Hojas l0- 12-folioladas 4a. анк simétricos a ligeramente asimétricos; valvas de la legumbre glabras o casi glabras, camas . ruiziana 4b. Foliolos иша не ts asimétricos en la base; valvas de la legumbre escuamosas. o... 1. samanensis 2b. Raquis foliar alado; pelosidad de las ramas densa. I. hintonii lb. Brácteas florales homomorfas, lanceoladas, linear-lanceoladas o triangulares. Flores proximales sésiles. Volume 80, Number 1 Sousa S. 227 El Género Inga 6a. Raquis foliar terete. I. punctata 6b. Raquis foliar marginado a 7a. Corola de 3-4.5 veces más "usn que el cáliz. 1. stenophylla 7b. Corola hasta 2.5 veces más larga que el cáliz. Ва. Cáliz 1.6-2.2 mm; legumbre hasta 1.6 ст de ancho. coco 1. ийй. Vorne in Cáliz 2.5-5 mm; legumbre hasta 4.8 cm de ancho. nsiflora сл c . Flores proximales pedice 9a. Raquis foliar glabro, las quails interfoliolares ca. 3 mm de diametro, escuteladas; caliz glabrescente; legumbre hasta 4 cm de ancho, glabra. 1. latipes 9b. Raquis foliar velutino con tricomas pardo-amarillentos, las glándulas interfoliolares ca. 1 mm de diámetro, cupuliformes; cáliz peloso; legumbre hasta 2.6 cm de ancho, pelosa. . /. pinetorum GRUPO 3 Cáliz 5-15 mm, tubular, generalmente con escotaduras, en botón obtuso o apiculado a rostrado. la. rag 2—6-folioladas . Legumbres obras. За. Legumbres hasta 2.5 ст de ancho; cáliz en botón largamente rostrado. ____________ 1. acuminata 3b. Lu: es 3.5-6 cm de ancho; cáliz en botón apiculado, obtuso a largamente rostrado. 4a. Raquis foliar generalmente terete; ramas jóvenes teretes. I. leiocalycina 4b. Raquis foliar angulado a cuadrangular o marginado a alado; ramas jóvenes anguladas a cuadrangulare 5a. Hojas 6- folioladas; foliolos acuminados; flores amarillas. 6a. Ramas jóvenes y nervaduras primarias tomentosas con tricomas pardo-amarillentos; seed Dee 11 cm de ancho; glándula en el acumen foliolar ocasionalment pres . urabensis бб. hamasi jóvenes y nervaduras primarias glabras; foliolos hasta 7 cm de ancho; glándula acumen foliolar frecuentemente presente. 1. venusta Hojas 4-folioladas; foliolos obtusos a apiculados; flores blancas. coo. 1. spectabilis 2b. ои pelos as, con tricomas más o menos rigidos . Inflorescencias largas y laxas, las flores espaciadas en el raquis I. dwyeri ' m cortas y compactas, las flores congestas en el raquis. І. hayesii lb. Hojas 8-18-folioladas 8a. Raquis floral hast : € 1. skutchii 8b. Raquis floral has 9a. Glándulas а | еар I. ciliata 9b. Glándulas interfoliolarees sésiles a ésiles 10a. Valvas de las legumbres see abras, lla. Hojas (10-)1 2-22-folioladas; corola 20-25 mm; legumbre hasta 4.6 cm de ancho. 1. multijuga llb. Hojas 8-12(-14)-folioladas; corola 12-18 mm; legumbre hasta 2.5 ст de ancho. . 1. aestuarorium 10b. Valvas de las legumbres maduras pelosa 12a. Corola 16-29 mm; legumbres Eius 13a. пи ка кен a ligeramente ds oloros; ¡e interfoliolares más pe- e el raquis foliar; flores robustas; cáliz 11-15 mm. . cuspidata 13b. Foliolos nudi te discoloros; glándulas wer буз tan anchas o más nchas que el raquis foliar; flores esbeltas; cáliz 3-5 mm. 000000... I. thibaudiana 12b. Corola. TD 11 mm; legumbres hispid xe 14a. Corola 9-11 mm; brácteas 1-1.7 ; botones florales obtusos. ___ 1. micheliana 14b. Corola 7-8 mm; brácteas 3-9 mm; бин florales rostrados a © apiculados. . flexuosa GRUPO 4 Legumbres maduras con pelosidad densa que da el color al fruto, generalmente hirsutas. la. Glándulas interfoliolares generalmente largamente estipitadas. 2 quis foliar terete. І. exalata 2b. Raquis foliar alado a. Flores esbeltas. 4a. Cáliz más de 10 mm; brácteas tan largas o más largas que el cáliz, persistentes. 5a. Pelosidad de tallos, ejes foliares y florales hirsuta; estipulas ovadas a triangular-cordatas, persistentes; brácteas generalmente lanceoladas. 6a. Hojas 2-4-folioladas; corola 15-18 mm. І. jimenezii 6b. Hojas (4-)6-10-folioladas; corola 20-27 mm. I. tonduzii 228 Annals of the Missouri Botanical Garden 5b. Pelosidad de tallos, ejes foliares y florales tomentosa; estipulas filiformes a triangular- 1 atenuadas, pronto аан cas; brácteas filiformes a oblanceoladas . calderonii 4b. Cáliz menos de 10 mm; brácteas generalmente menos del и де largo del cáliz, caducas. Та. Cáliz sain anie estriado o totalmente cubierto de indumento; brácteas ovadas a triangulares. im 6-folioladas; foliolos cartaceos. s 8- 12-folioladas; foliolos coriá Tb. Cáliz Su кыны estriado; brácteas on a lineares. 3b. Flores care 9a. Estipulas cordatas, tan anchas como largas; glandulas sobre las nervaduras medias de los foliolos Шш шты presentes; inflorescencias multifloras (10-22); corola 24- mm. oldmannii I. litoralis 1. herrerae 1. tenella 9b. Estipulas no cordatas, más largas que anchas; glándulas sobre -> nervaduras medias de los foliolos ausentes; inflorescencias paucifloras (3-4); corola 28-32 mm. io... /. davidsei lb. Glandulas intertodoleres sésiles a casi sésiles. 10a. Glándulas sobre las nervaduras medias de los foliolos frecuentemente presentes; foliolos cocleados. la. Raquis foliar marginado a generalmente alado; brácteas 6-9 mm. . sierrae uis foliar terete; brácteas ca. 2 mm. 10b. o sobre las nervaduras medias de los foliolos ausentes; foliolos más o menos aplanados. cáliz con spicuamente estria En s Cáliz glabro excepto el ápice; corola 35-55 m 1. mucuna 13b. Cáliz aa en um su о. corola hasta 25 m o 4.5-8.5 cm; legumbre hasta 18.5 cm, ен con tricomas ferrugineos; Ll a i a. . leonis l 4b. Peciolo 1-3.5 cm; legumbre hasta 12 cm, hirsuta con tricomas pardo-amarillentos; corola verdos 1. chiapensis 1 2b. сав sin estriaciones o si las tiene cubiertas por el indumento. a 14-15 mm, subturbinada; raquis foliar terete 1. cabrerae 15b. Corola 2] mm o más, tubular; raquis foliar terete o marginado a alado. 6a. Flores proximales sésiles; legumbre hasta 5 cm de ancho. 1. sinacae 16b. Flores proximales pediceladas; legumbre hasta 2.7 ст de ancho. .................. І. calcic ola GRUPO 5 І. rubiginosa Cáliz 3-6 mm, robusto, campanulado. GRUPO 6 Raquis foliar alado; lóbulos del cáliz cuspidados a caudados, cáliz abierto en botón; legumbres aplanadas a tetragonales. la. Corola 18-3 n; legumbres cm. 2a. Flores E певања cáliz 16-21 mm; legumbre aplanada; tallos jóvenes con tricomas hasta mm. I. ismaelis 1 2b. ind proximales sésiles; cáliz 9-17 mm; legumbre tetragonal; tallos jóvenes con tricomas hasta І. pavoniana lb. Gia. 10: 15 mm; legumbres 3.9-7.6 х cm. 3a. Tallos y ejes foliares glabros; estipulas ЗНА ВА subuladas; glándulas interfoliolares sésiles. __. 1. lacustris 3b. Tallos y ejes foliares hispidos; не setiformes, arqueadas; glándulas interfoliolares largame nte Lc ookii estipitadas GRUPO 7 Legumbres subteretes, acordonadas. la. Raquis toliar tere 2a. Hojas (4-)6- 8 10) fohioladas. I. oerstediana 2b. Hojas o 16-folioladas 1. cocleensis lb. Raquis foliar 3a. Glándulas interfoliolares urceoladas a pateniformes, el ósculo circular, generalmente hasta 1.2 mm de diámetro. 4a. Legumbres con los márgenes parcialmente hipertrofiados, las valvas reducidas pero aún claramente se distinguen. 5a. Flores a sésiles; cáliz 6-10 mm, tubular, esbelto; corola hasta 20 mm, generalmente . affinis el doble de la longitud del cáliz 5b. Flores proximales frecuentemente pediceladas; cáliz 10-18 mm, infundibuliforme, robusto; corola hasta 25 mm, menos del doble de la longitud del cáliz. vera 4b. Legumbres con los márgenes totalmente hipertrofiados, las valvás dificilmente distinguibles de lòs márgenes; flores proximales sésiles. Volume 80, Number 1 Sousa S. 229 1993 El Género Inga 6a. Hojas (4-)6-8-folioladas; brácteas hasta 2.3 mm; legumbre hasta 2.2 cm de diámetro. І. pauciflora b. Hojas 4-folioladas; brácteas hasta 11 mm; legumbre hasta 3.5 cm de diámetro. ... 1. chocoensis 3b. Glándulas interfoliolares patenifo rmes, el ósculo наши oval и oval alargado a frecuentemente lunular, generalmente más de 2 mm de diámetr Та. Hojas (4-)6-8(-10)-folioladas; Dráeten as o legumbre 8-32 x 1.2-1.5 cm, los márgenes _ кө hipertrofiados, las valvas о das pero distinguibles. я cteas -4 mm, reniformes a ovales: pce 1. oerstediana . Brácteas 7-15 mm, oval- oblongas, РЕА ue I. latibracteata Tb. Hojas 8- 12. folioladas; brácteas más o menos per stentes; ; legumbre 40-150 x 2-3.5 cm márgenes totalmente hipertrofiados, las valvas indistinguibles de los márgenes. u- . edulis Grupo 8 Inflorescencias capituliformes, con flores sésiles a pediceladas, los pedicelos hasta 3 veces la longitud del cáliz. la. pne foliar marginado a alado. 2a. Organos vegetativos y legumbres pelosas a Hojas 4-6-folioladas; flores densamente pelosas; legumbre 4-6.5 cm de ancho. ............. I. dasycarpa 3b. Hojas 6-10-folioladas; flores glabras, excepto los lóbulos de la corola; es des 1.8-2 cm de A E НИКС БИРИРНЕЗСРЕ 1. и N c ancho. . Organos vegetativos y legumbres glabrescentes a glabras. 4a. Ra jm s foliar marginado, las glándulas interfoliolares piriformes, largamente estipitadas; corola 5 OS L xim Ab. А foliar alado, las glándulas interfoliolares pateniformes a cupuliformes, sésiles a casi sésiles; corola 20-50 mm. Sa. Bo 13-14 mm; corola 20-23 mm, densamente setosa con tricomas pardo-amarillentos e 14 superior; tubo estaminal inserto a cortamente exerto. ..... . I. por HN Бый 5b. саћа 24-33 mm, corola 38-50 mm, glabra e меке. еб sericea con tricomas canescentes en el ápice de los lóbulos; tubo estaminal largamente exerto. ле 1. bella = Raquis foliar terete 6a. Estipulas pronto caduc 7 rola pelosa en ius su lon mas e ба шм сн tomentosas. .... I quaternata 8b. Ramas e inflorescencias gla brescentes a glabras 9a. Cáliz 1.5-3 mm; corola 6-9 mm, la cea 2.2-3.5 veces la longitud del cáliz. NINE ow Ob. Cáliz 5-7 mm; corola 9-12 mm, la corola hasta 2 veces la longitud del cáliz I. mortoniana 7b. Corola к excepto pelosa sobre los lóbulos. Oa. nculo 0.5-1.5 cm; cáliz 1.2-1.5 mm; corola 6-7 mm; legumbre ca. 7 x 2.5-3 c клр а A нн ан ШӨН дыны А Ны айын. q 10b. Pelincul (1 -)3-6 cm; pads 1.5-2 mm; corola 4-6 mm; кнын 15-50 x 3. 52 5 ст linear- UAE ПУН = | кы . FL. pum. 6b. е persisten ntes | ipulas nbn a obovadas Cáliz aplicado a la corola; corola 4—7.5 mm. ttt I. paterno 12b. Cáliz amplio no aplicado a la corola; corola 17-25 mm. ojas 2-4-folioladas; legumbre linear-oblonga. -i ak i ees 3b. Hojas 6-8-folioladas; legumbre UU вина gg Sat пл A I. spiralis 11b. PS aciculares; lóbulos del cáliz filiformes a lanceolado-aleznados Tallos y ejes foliares glabrescentes; glándulas elias sésiles; legumbre аш - Qo seu n E — H—h A M I. filiformis 14b. Tallos y ejes foliares 5 setosos; glándulas interfoliolares largamente estipitadas; p oves largamente pelosa, sin lenticelas. e saffordiana GRUPO 9 Inflorescencias umbeliformes, los pedicelos más de 3 veces la longitud del cáliz. la. Hojas 8-12-folioladas. nnn 1. flagelliformis lb. Hojas generalmente 2-6-folioladas. га. Estipulas aleznadas; brácteas 1.2-1.4 mm. ooo _ L. sertulifera 2b. Estipulas lanceoladas u oblanceoladas a linear-lanceoladas; brácteas 2-9 m За. Hojas (2-)4-6-folioladas; pedúnculo floral 2-3 cm; brácteas di 5 mm, oblanceoladas. 1. umbellifera 3b. Hojas. A 4) folioladas: “pedúnculo foral. hasta 1 cm; 2 brácteas hasta 91 mm, , lanceoladas a linear- Та MA MN" — oo I. polita 230 Annals of the Missouri Botanical Garden Inga acrocephala Steudel, Flora 26: 759. 1843. TIPO: Surinam: Hostmann 1067 (isotipos, BM, F, K, MO). Inga brev rige aa Repert, Spec. Nov. Regni Ve 62. o: J. А. Purpus 282 (ho- о 4 а Inga brevipedicellata se habia mantenido como una especie del sur de México y del norte de Centroamérica, pero han empezado a aparecer ejemplares de Costa Rica y Panamá que extendie- ron su distribución (Zamora, anotación de ejem- plares, com. pers.). También con las nuevas co- lectas la variación se ha ampliado y asi tenemos ejemplares de Veracruz francamente pedicelados y los de Chiapas, hacia el sur, sésiles. En Sudaméri- ca en Colombia y Ecuador las hojas se mantienen con seis foliolos, pero en el área de las Guayanas tenemos hojas 8-folioladas, Unico carácter que po- dria mantener separada a /. brevipedicellata de I. acrocephala. Sobre el tipo de /nga brevipedicellata que de- signo Harms (1923), J. А. Purpus 282 del Her- bario de Berlin (B), fue destruido y el único ара- rente duplicado que he visto está en el Smithsonian (US, 1202084): se trata de un ejemplar en flor ue muestra claramente las flores sésiles а casi sésiles, las brácteas homomorfas linear-lanceoladas además de hojas 4—6-folioladas caracteristicas de Inga punctata Willd.; además, existe el ejemplar de J. A. Purpus 281 (G), el cual está correctamente identificado como /. leptoloba Schldl. [1. punctata Ша. ], por lo que debe haber habido una confusión de etiquetas. Por otro lado, por el momento, mantenemos separada a /nga acrocephala де 1. acreana Harms (véase tambien /. allenii), no por los razonamien- tos, descripción, y ejemplares citados por Poncy (1985), ni por las observaciones de Macbride (1943) respecto al raquis foliar, sino por el cáliz de mayor tamano de /. acreana Harms. Otro aspecto de variacion en /. acrocephala es la transición de tipos de inflorescencias (que en ocasiones ocurren en el mismo ejemplar (/barra M. 470, MEXU) de racimos cortos a capitulos, y de hecho debilita ~ enormemente la taxonomia actual de secciones de- finidas por tipos de inflorescencias. Material representativo. | MEXICO. CHIAPAS: Téllez V. 646 (MEXU). VERACRUZ: Sousa 1 e GUA TEMALA. EL PETEN: e 0474 ABAL: Tenor et al. 14518 (MEXU). HONDURAS. ATLANTIDA: Мойла R. 10493 (F). Costa RICA. ALAJUELA: Jiménez 1768 (F). HEREDIA: Schubert & Holdridge 1353 (A). LIMON: Ham- mel & л 14340 (NY). PUNTARENAS: Kennedy 2009 (MO). PANAMA. COLON: Crosby & Denslow 6480 (MO). DARIEN: Croat 68964 (MO). PANAMA: Dressler 3272 5x (MEXU). vERAGU UAS: Croat 27 696 (MO). COLOMBIA. AN- NAM: Hostmann 1067 (M : Wac henheim 312 (US et al. 465 (MO). BRASIL. (NORTE) AMAPA: Maguire et al. Inga acuminata Benth., London J. Bot. 4: 600. 1845. TIPO: Trinidad: Lockhart 334 (holotipo, Esta especie hasta hace poco ha sido colectada en Mesoamerica. Muy característica por sus bo- tones florales con el caliz cerrado largamente ros- rado — н ен o. COSTA RICA. HEREDIA: Opler 93 (F). PANAMA. BOCAS DEL TORO: McPherson & Aran- "i 10102 (MO). DARIEN: Duke 4934 (MO). ZONA uet CANAL: Foster 2976 (MO). TRINIDAD: Britton et al. 257. (GH). COLOMBIA. CHOCO: Gentry 9336 (MO). ај ZULIA: 27 & Gutiérrez 1556 (MO). EcUADOR: H. Eggers 1582: | Inga aestuariorum Pittier, Contr. U.S. Natl. r 8: 183, t. 89. 1916. TIPO: Costa Rica: Tonduz 6793 (holotipo, US; isotipos, BR, C). León (1966) incluye a /nga aestuariorum como sinónimo de /. multijuga Benth. (véase ésta), adop- tando el concepto de Bentham (1875) y recono- ciendo que ésta "podria quizá ser una variante sugerencia que Bentham tambien consideró al incluir el ejemplar de Oersted de Costa subespecifica,” Rica (“perhaps a variety with rather smaller flow- rs”). Sin embargo, en el norte del Istmo de Pa- nama las dos especies son simpátricas, y así Fendler en enero 24 de 1850 las colectó (en flor) en Cha- gres, única área que él visitó en su único viaje a Panamá (véase Stieber & Lange, 1986). Ambas especies fueron colectadas bajo el mismo nümero (51) de colecta, estando depositada /. aestuariorum en el GH e /. multijuga en K, Inga aestuariorum la mantengo separada considerando su menor nümero de foliolos por hoja, la flor más pequena y el fruto que sólo llega hasta 2.5 cm de ancho. Es interesante agregar que el material de la vertiente del o de México es muy homo- geneo, distinguiendose por sus glándulas interfolio- lares del raquis foliar caracteristicamente con el osculo muy cerrado, siendo transversalmente lin- ear-oblongo; en cambio en el material de Nicaragua a Panamá, el ósculo es más o menos circular. Material representativo. MEXICO. CHIAPAS: Breed- love 34962 (MO). OAXACA: Sousa eta VERACRUZ: Sousa 3085 (MEX PAZ: Standley 90278 (F). EL PETEN: Tün Ortiz 1084 Volume 80, Number 1 1993 Sousa S. EI Género /nga (BM). NICARAGUA. RIO SAN JUAN: Grijalva & Almanza 3606 (MO). Costa RICA. PUNTARENAS: Burger & Liesner 6702 (F). PANAMA. CHIRIQUI: Croat 2 1906 (MO). HERRERA: Carrasquilla & Lao 347 (MO). VERAGUAS: Allen 154 (MO). Inga affinis DC., Prodr. 2: 433, 1825. TIPO: Brasil: Raddi s.n. (holotipo, G-DC, foto MO). Especie muy relacionada con /nga vera Willd. e 1. uraguensis Hook. et Arn. [“uruguensis” |; de la primera difiere /. affinis por contar con flores sésiles o casi sésiles, un cáliz más pequeno (6-10 mm), tubular y la corola generalmente el doble del cáliz. De /. uruguensis, las diferencias son más tenues, ya que ambas cuentan con flores del mismo tamaño y la diferencia fundamental está en la ma- yor relación de la longitud del cáliz con respecto a la corola; en /. uruguensis la corola es solamente 1.5 vez más larga que el cáliz. Por lo anterior sin grandes problemas /. uruguensis podría pasar como sinónimo de /. affinis. En Mesoamérica esta especie había sido inter- pretada como Inga edulis C. Martius o bien /. vera Willd.; de la een a pesar de contar con flores de similar tamano y proporcion, difiere por las glandulas meee mas pequenas y el fruto de menor tamano con las valvas angostas pero aun visibles, caracteres que la acercan mas a /. vera. Material E nai ir MEXICO. CAMPECHE: Ка- mírez R. & Flores C. 97 (MEXU). cHiAPAS: Martinez ¿ Aguilar 12368 (MEXU). OAXACA: poise et al. 13104 (MEXU). SAN LUIS POTOSI: Fryxe Anderson 3425 (F). TABASCO: Orozco S. & González 26 (MEXU). VERACRUZ: Palmer 362 (MO). BELICE: Dwyer HO: pies U. 210 (MO). ALES: cell 5 (MEXU). ZELAYA: Stevens 10460 (МО). COSTA hus. LIMON: Davidse & Herrera 30907 (MO). PANAMA. Р : Munoz 47 (MO). COLOMBIA. CASANARE: Uribe U. "5684 (F). VENEZUELA. APURE: Davidse & Gonz sin 4354 (MO). BOLIVAR: Ll. Williams 11615 (MO). ECUADOR. GUAYAS: Harling & Anderson 19342 (MO). PERU. CUZCO: C. Vargas C. 1. (MO). MADRE DE DIOS: Gentry et " о BRASIL. : Plowman et al. 4 (MO). (NORESTE) 593 (MO). PERNAMBUCO: Gardne 38 (US). (CENTRO OESTE) DISTRITO FEDERAL: /rwin et al. 11421 (MO). coras: Prance & Silva 59533 (MO). МАТО G ROSSO: Pirani 1163 MR (SUDESTE) SÃO PAULO: Епеп & Clay- ton 6078 (MO). (: suL: Hagelud 12270 (F). ARGENTINA. CORRIENTES: Seigler et al. DS- 10147 (MO). снасо: Charpin & Eskuche AC20146 (MO). PARAGUAY: Morong 528 (MO). Urucuay: Bernardi 18680 (MO). Inga alba (Sw.) Willd., Sp. Pl. ed. 4, 4: 1013. 1806. Basiónimo: Mimosa alba Sw., Prodr. 85. 1788, non M. Vahl, 1807. TIPO: Guayana Francesa: von Rohr s.n. (fl) (lectotipo, desig- nado aqui, B Sobre la lectotipificación de /nga y su basiónimo Mimosa alba Sw., el mismo O. Swartz (1788) nos informa que se trata del material de **Cayenna," y al estar marcada la descripción de la especie por un pequerio sol similar a un asterisco, el material lo consulto en el Herb. de J. Banks (BM) como el mismo Swartz refiere en su prefacio, sin anotar al colector. Por otro lado Poncy (1985) refiere al ejemplar de von Rohr como el tipo de este nombre de Cayenne en el BM, sin lectotipificarlo, asi que aqui solo se está formalizando lo que Poncy apuntó, y yo concuerdo. Existe un ejemplar en flor, anotado como Mimosa alba Sw. de Cayenne y colectado por von Rohr s.n. (BM). Especie que alcanza las mayores tallas en el genero (hasta 40 m), con flores e inflorescencias inconspicuas, lo que ha contribuido a su escasa representación en herbarios. Sin embargo, existen varios ejemplares estériles, que hacen suponer su aparente frecuencia en el dosel de las selvas altas perennifolias de América cálido ћитеда. Inga alba está estrechamente relacionada a /. lateriflora Miq. de difiriendo fundamentalmente por el tipo de inflo- rescencia; esta afinidad debilita los limites de las la seccion Leptinga Benth. secciones Bourgonia Benth. y Leptinga Benth. Esta especie en Мезоатепса ha sido confundida con /nga fagifolia (L.) Willd. ex Benth. (/. laurina (Sw.) Willd.). Material representativo. | MEXICO. OAXACA: Hernán- dez G. 2495 € VERACRUZ: Wendt et al. s.n. (MEXU). NICARAGUA. ZELAYA: Proctor et al. 27 кзы (МҮ). Costa RICA. HEREDIA: ariel 11662 (MO). NAS: ПА 5275 (F). PANAMA. SAN BLAS: de Nevers et al. 3 (MO). ZONA DEL CANAL: Wetmore & vie 129(F). бина VAUPES: 4. S. Barclay & К. - Sch tes 509 (K). VENEZUELA. AMAZONAS: Colchest „242 К). BoLIVAR: Steyermark 60417 (MO). DELTA AMACURO: Blanco 171 (MO). MERIDA: Breteler 4756 (U). GUYANA: Maas & Westra 3901 (MO). SURINAM: B. W. [Sur. For. Dept.] 6174 (MO). GUAYANA FRANCESA: Prévost & Gre- nand 973 (MO). ECUADOR. NAPO: Neill et al. 8292 (MO). PERU. LORETO: Vasquez Livia: Bang Lectae 1439 (M : Robelo et al. 2280 (MO). AMAZONIA: Krukoff 6603 (MO). PARA: Silva & Souza 2355 ( ONDONIA: Cid et al. 4942 (MO). RORAIMa: Milliken & К. Miller 777 (К). ~ ~ Inga allenii León, Ann. Missouri Bot. Gard. 53: 346. 1966. TIPO: Panamá: Allen 2687 (ho- lotipo, US). muy caracteristica por sus glándulas interfoliolares piriformes largamente estipitadas. Inga allenii en general está relacionada a /. mor- 232 Annals of the Missouri Botanical Garden toniana León, como ya León (1966) lo había apun- tado, pero la encuentro más cercana а 1. acreana Harms (incluyendo a /. myriocephala Pittier) del área suroeste del Amazonas, de la cual es dificil separar, posiblemente siendo /. allenii una varie- dad de /. acreana. Material representativ o. Costa RICA. HEREDIA: Gó- mez P. et al. 21111 (MO). LIMON: Grayum et al. 3502 (MO). PUNTARENAS: Utley & Utley 1201 (F). PANAMA. COCLE: Ebinger 965 (GH). Inga belizensis Standley, Publ. Field Columbian us., Bot. Ser. 4: 307. 1929. Tipo: Belice: Schipp 24 (holotipo, F; isotipos, BM, K, MO, Como ya apunto Leon (1966) esta especie esta cercanamente relacionada a Inga fagifolia (L.) Willd. ex Benth. (/. lauriana (Sw.) Willd.); sin embargo se trata de dos especies aparte, no sólo por la inflorescencia sino también por el fruto (véase clave Material representativo. MEXICO. CHIAPAS: Martí- nez S. 8517 (MEXU). BELICE. EL CAYO: Lundell 6599 (NY). STANN CREEK: Schipp 24 (MO). GUATEMALA. EL PETEN: Lundell & Contreras 20841 (MEXU). Inga bella M. Sousa, sp. nov. TIPO: Costa Rica. Puntarenas: Reserva Forestal Golfo Dulce, Osa Peninsula, Rancho Quemado, ca. 15 km of Rincon, i in bottom of S end of valley along . Chavarría & A. Solis 16962 (holotipo, MO; isotipos, CM, CR, ). Inga | auct. ra Beurling. Ilustración: Pit- tier, Contr. U.S. Natl. Herb. 18: t. 83-84. 19 Similis /ngae portobelle nsi Beurling, a qua tamen in- florescentia pauciflora, calyce ac corolla duplo vel ultra longioribus, calyce laevi, corolla ex sparse canescenti- seric кан, nec non tubo stamineo longe exserto differ Arboles 6-10 m; ramas subteretes, glabras. Ho- jas 4-folioladas; estipulas 7-24 mm, foliáceas, ova- das a ovado-lanceoladas, cordatas en la base, per- sistentes; peciolo 1-4.5 cm, alado; foliolos concoloros, cartáceos a subcoriáceos, algo abolla- dos, la base ligeramente asimétrica, cuneada, el ápice acuminado a cuspidado, el haz opaco, glabro, el envés opaco, glabro excepto esparcidamente se- riceo sobre la nervadura primaria y secundarias, nervadura primaria eglandular; par basal de ied 6-16 x (3-)5-7.5 ceolados a oblanceolados, el par apical 17-21 x -8.5 ст, elipticos a obovados; raquis foliar 4.5- 7.5 ст, alado, el ala hasta 18 mm de ancho, cm, elipticos, o lan- oblanceolada a obtriangular; glándulas interfolio- lares sésiles o casi sésiles, pateniformes a cupuli- formes, el cuerpo glandular 2.5-3 x 2.5-3 mm; apéndice hasta 16 mm, caduco. Inflorescencias capituliformes, solitarias; pedünculo 0.5- aplanado, sulcado, glabro; raquis floral muy re- ducido, las brácteas gradualmente heteromorfas, 4—12 mm, foliaceas, ovadas a oblongas, las de la periferia de mayor talla, persistentes. Flores sésiles о brevipediceladas, los pedicelos hasta 5 mm, ro- bustos; botones florales con el cáliz cubriendo, des- conocidos; cáliz 24—33 mm, tubular-infundibulifor- me, inflado, liso, glabro, sin escotaduras, los lóbulos 3-4 mm, ampliamente triangulares; corola 38-50 mm, tubular a subturbinada, blanco-verdosa, gla- bra excepto sericea con tricomas canescentes en el ápice de los lóbulos; tubo estaminal largamente exerto, blanco. Legumbre 18.5-22.5 x 2.6-3 x .4-0.7 cm, aplanada, curveada, linear-oblonga, sesil, rostrada en el ápice, esparcidamente pube- rulenta a glabra, las valvas transversal e inconspi- cuamente nervadas, las suturas, marginadas, los márgenes de las suturas aplanados. Distribución y hábitat. De la vertiente pa- cifica de Costa Rica y Panama; en Costa Rica solamente conocida en la Peninsula de Osa y en Panamá, en el Darién. En altitudes de 5-250 m. Florece de abril a principios de junio, fructifica de abril a junio. Paratipos. | COSTA RICA. PUNTARENAS: Parque Nac inp la Sirena Station, 8?29'N, 83°34'0, В. P & C. Kernan 16656 (F, MO); Pa gd Nacional Cor- cova Sirena, Monkey Woods, 8?27'-30'N, 83°33'- ; Kernan 363 (МО); Santo Domingo, Golfo Dulce, с 9879 [colección base de la ilustración de Pittier ] (BM, BR, Е, GH, К, MEXU, MO, NY, 05); Playa Blanca, Golfo Dulce, M. Valerio 468 (F). PANAMA. DARIEN: Cana- Cuasi trail along Rio Cuasi, R. L. Hartman 12486 (МО); main stream of Rio Cuasi, 0- 2.5 mi. S of Tres Bocas, J. H. Kirkbride & J. Duke 1132 (MO). Inga bella está muy relacionada con /. porto- bellensis Beurling (sus diferencias son explicadas en la diagnosis, se relaciona con /. inflata Ducke pero /. bella difiere de esta ültima por las hojas de véase |. portobellensis); tambien el Amazonas, mayor talla y menos coriáceas, el raquis foliar alado, las estipulas bien desarrolladas y persisten- tes, las flores sésiles a casi sésiles, y la legumbre que es aplanada a la altura de las semillas. El epíteto hace resaltar la belleza de las hojas y flores de esta Inga, que potencialmente podría usarse como or- namental. Inga cabrerae M. Sousa, sp. nov. TIPO: Guate- mala. Baja Verapaz: Sierra de las Minas, 3 km SE of Purulhá, 2,000 m, 6 Jan. 1974, Volume 80, Number 1 1993 Sousa S ; 233 EI Género /nga L. O. Williams, A. Molina R., T. P. Williams & A. R. de Molina 43415 (holotipo, F; iso- tipo, US). Frutex vel pi Lapi ramis juvenilibus dense ferru- gineo-tomentos a 3-jugata; stipulis 2-3 mm longis, ex d oru weiss irt foliolis coriaceis, nervo medio ulas interdum terentibus; rhachidi follar tereti, glan- fulis song ame sessilibus с Inflorescentiae spiciformes; rhachidi florali 2-3.5 cm longa; bracteis ho- momorphis ca. 2 mm longis, late Sarii Flores sessiles; calyce ca. 6 mm longo, tubulari; corolla 14-15 mm longa, ах brunneo-lutescenti-villosa. Legumen usque a cm latum, lineari- а ex complanatum sub- tetragonale, ferrugineo-hirsutu Arbustos a árboles 2-6 m; ramas angulosas, ferrugineo tomentosas cuando jóvenes, posterior- mente glabrescentes. Hojas 6-folioladas; estipulas 2-3 mm, liguladas a suborbiculares, caducas; pe- ciolo 2-3 cm, terete; foliolos ligeramente discolo- ros, coriáceos, cocleados, la base simétrica, obtusa a aguda, el ápice generalmente apiculado a acu- minado, en ocasiones obtuso, el haz brillante, es- parcidamente piloso, tomentuloso sobre la nerva- dura primaria, el envés opaco, esparcidamente tomentuloso por la superficie, más densamente so- bre las nervaduras, la nervadura primaria en oca- siones glandular proximalmente; par basal de fo- liolos 5.5-6.5 x 3.5- 12-14 x 7-8.5 cm, anchamente elipticos; raquis foliar 6.5-8.5, terete; glándulas interfoliolares se- siles, urceoladas, el cuerpo glandular 1.5-2 x 1.5- 2 mm, apéndice ca. 2 mm, grueso, caduco. Inflo- rescencias en espigas 1-2 fasciculadas; pedunculo 1.5-3.5 cm, terete, on асар sulcado, fe- rrugineo tomentoso; raquis floral 5 em, ovados, el par apical m, las flores algo espaciadas proximalmente, pince distalmente, las brácteas aparentemente homo- morfas ca. 2 mm, anchamente ligualadas, caducas. Flores sésiles; botones florales con el cáliz abierto los lobulos erectos, obtusos; cáliz ca. 6 mm, tubular, sin estrias aparentes, tomentoso con tricomas par- do-amarillentos, sin escotaduras, los lobulos ca. 1 anchamente triangulares; corola 14—15 mm, са ил blanca, vellosa con tricomas pardo- amarillentos; tubo estaminal inserto, blanco. Le- gumbre ca. 22 x 3 x 1 ст, aplanada, subtetrago- nal, curveada, linear-oblonga, atenuada en la base, apiculada en el ápice, ferrugineo hirsuta, las valvas aplanadas, las suturas marginadas a acostilladas, las costillas hasta 4 mm de ancho, los márgenes de las suturas nervados. Distribución y hábitat. Conocida del área de las Lagunas de Montebello, Chiapas, México, y de la Sierra de las Minas en Guatemala. En vegetación de transición entre selvas altas y bosques caduci- folios, en bosques caducifolios con Liquidambar y en bosques de neblina. En altitudes entre los 1,600 y 2,000 m. Florece a principios de enero y fruc- ићса en marzo. Paratipos. MEXICO. CHIAPAS: Mpio. La Trinitaria, a 3 km al O de Tziscao, dentro de los límites del Parqu Natural Lagos de Monte Bello, E. Cabrera C Ho de Cabrera 6063 (MEXU); a 15 km al E de las Lagunas de Monte Bello, Cabrera C. et al. 1861 (MEXU, МО). Especie muy cercana a /nga dasycarpa M. Sousa e /. tenella M. Sousa de las cuales se dis- tingue fundamentalmente por su "equus (ше te- rete, sus glándulas i d , con un cuerpo glandular de mayores dimensiones y la presencia de glandulas en la nervadura pri- maria. En cuanto a las inflorescencias /. cabrerae posee espigas, lo que la acerca más a /. tenella, pero el cáliz de este ültima es tubular-cónico, es- triado y esparcidamente peloso. Es interesante ano- tar que estas tres especies se encuentran simpa- tricas en el area de las Lagunas de Montebello en Chiapas, estando aparentemente aisladas estacio- nalmente por floraciones en diferentes periodos; asi Inga cabrerae florece en enero (aunque hay que aclarar que el dato es de Guatemala), 1. tenella en marzo-abril, e /. dasycarpa en octubre. Se dedica esta especie a Edgar Francisco Ca- brera Cano (1957-), colector botánico del sur de Mexico. Inga calcicola M. Sousa, sp. nov. TIPO: Mexico. Oaxaca. Curva del Diablo, a 4 km al ESE de Temazcal, en el camino al vertedero, Mpio. San Miguel Soyoltepec, Distr. Tuxtepec, 50 987, M. Sousa S., G. Andrade, s C. & L. Cortés A. 13090 (holotipo, EXU; isotipos, MEXU, MO). Figura 1. Arbores фиш folia (3—)4—5-jugata; rhachis foliaris ndulis interfoliolaribus diminutis, sessi- libus vel fere о. nflorescentiae racemosae corym- понеке pauciflorae; flores 21-27 mm longi, pedicellati vel sessiles; calyx ac corolla tubulares, tubo sta minali inserto. Legumen „шш oblongum angustum, brunneo- lutescenti-hirsutulum Arboles 3-12(-20) m; ramas teretes, velutinas con tricomas pardo-amarillentos cuando jóvenes, posteriormente glabrescentes. Hojas (6-)8-10-fo- lioladas; estipulas 5 mm, liguladas a lanceoladas, caducas; peciolo 1-2.5 cm, terete a marginado; foliolos ligeramente discoloros, cartáceos, la base ligeramente asimétrica, cuneada a obutsa, el ápice acuminado a cuspidado, el haz opaco, esparcida- mente tomentoso, después glabrescente, las nervaduras tomentosas con tricomas pardo-ama- rillentos, el envés pálido con la misma pelosidad pero ésta más densa, la nervadura primaria eglan- Annals of the Missouri Botanical Garden Па" А l. Inga calcicola M. Sousa. A. Rama con inflorescencias. — B. —H UN: AN. N =, pem E, 444 | Detalle de estipulas.—C. Glandula interfoliolar. —D. Botón floral. — Е. Flor.—F. Cáliz. —G. Corola.—H. Tubo estaminal.— I. Gineceo. (Tomado de Sousa 13090.)—J. Fruto. (Tomado de Sousa 10273.) dular; par basal de foliolos 3.5-8.5 x 2.1-4.2 cm, eliptico-lanceolados a ovados, el par apical 12.5-17(-19.5) x 4-6.5(-8.4) cm, elipticos en ocasiones obovados; raquis foliar 7-11(-20) cm, alado entre todos los pares de foliolos, el ala hasta 15 mm de ancho, oblanceolada; glándulas inter- foliolares sésiles a subsésiles, urceoladas, el cuerpo glandular 0.8-1 x 0.8-1 mm; apéndice ausente. Inflorescencias racimos corimboides, 1-2 fasci- culadas; pedúnculo 3.5-7 cm, subterete, sulcado, tomentoso con tricomas pardo-amarillentos; raquis floral 2-3 cm, las flores espaciadas proximalmente, congestas distalmente, las brácteas algo hetero- morfas 3-4 mm, concavas, las proximales fre- Volume 80, Number 1 1993 Sousa S. 235 EI Género /nga cuentemente estériles, ovadas en el Уз superior del pedünculo, persistentes, las distales fértiles, espa- tuladas, persistentes. Flores proximales pedicela- das, el pedicelo hasta 7 mm, las distales sésiles; botones florales con el cáliz cerrado, obtusos; cáliz 11-21 mm, tubular, estriado, tomentoso con tri- comas pardo-amarillentos, sin escotaduras, los ló- bulos 1-1.5 mm, triangular-obtusos; corola 21- 27 mm, tubular, verde pálido, canescente sericea; tubo estaminal inserto, blanco. Legumbre 19-23.5 x 0.4-0.5 cm, aplanada, subespiralada a espiralada, linear-oblonga, sésil en la base, api- culada a rostrada en el ápice, hirsütula con tri- comas pardo-amarillentos, las valvas aplanadas, las suturas derechas, los márgenes de las suturas apla- nados. Distribución y hábitat. Especie solamente conocida del norte de Oaxaca, en el distrito. de Tuxtepec, en suelos de litosoles negros calizos. En selvas altas subperennifolias con Brosimum (Mo- raceae), como un elemento del estrato bajo агђогео. En altitudes entre 30 y finales de mayo a la primera quincena de junio, la O m. Florece desde fructificación de septiembre a diciembre, con frutos en el suelo en febrero. Nombre vernáculo. | Guatope. Usos. “Los frutos maduros son comestibles.” XICO. OAXACA: Distr. Tuxtepec, Mpio. San Miguel Жошо: Presa de Temazcal, al NO sobre la cortina de la presa, L. Cortés & К. Torres C. 3(MEXU, MO); Isla Isabel Maria, cerca de San Miguel Soyaltepec, Cortés & Torres 841 (MEXU); Temazcal, cerros cerca de los vertederos, a 7 km al SE de la vii de la presa Miguel Alemán, Martínez S. & L. Rico А. 6132, 6785 (MEXU); cortina de la Presa Alemán, jm al, M. Sousa & L. Rico A. 10273 (MEXU); a 4 km al SE de Temazcal, camino a los vertederos de la presa, Sousa e Г. Calzada 13028, 13048, 13066 (MEXU, МО); enfrente de la caseta de vigilancia de los vertederos de la pr Alemán, Temazcal, Torres ыеп (MEXU); Mpio. Santa Maria Jaontepet, El Aguila, La Joya del n Martínez S. & C. H. ee: 24021 (MEXU, MO). Inga calcicola podria incluirse en la serie Lon- giflorae de Bentham, y se parece а I. longiflora Spruce ex Benth. de Brasil, por contar con flores pediceladas, pero no es muy cercana, difiriendo de ella en un gran nümero de caracteres. Es interesante anotar, para la interpretación del uso de los caracteres en /nga, que la corola de esta especie continua creciendo aün después de la antesis, la cual ocurre al inicio del dia; asi en la manana (Sousa 13048, 13066) la flor ya abrió y los estambres están erectos, pero la corola no so- brepasa el cáliz, mientras que en la tarde (Sousa 13090) la corola ya creció y mide casi el doble del largo del cáliz (sólo en el segundo caso se revisaron las flores manana y tarde; las primeras colectas ocurrieron en la manana). Por otro lado cuando en botón, la protocorola es más pequena que el cáliz. Inga calderonii Standley, J. Wash. Acad. Sci. 13: 352. 1923 (como *calderoni"). TIPO: El Salvador: Calderón 1392 (holotipo, US; iso- tipo, Especie muy interesante la cual en flor se parece mucho a y de hecho ha sido confundida con /nga vera Willd., pero se distingue de ésta por las glán- dulas interfoliolares largamente estipitadas. En fru- to sus relaciones se inclinan hacia el grupo de /. villosissima Benth. y quizá podria haber tenido su origen еп la cruza de /. vera e I. tenella M. Sousa. Se mantiene como especie válida ya que sus ca- racteres parecen estables. MEXICO. CHIAPAS: Sousa et al. 11820 (MEXU). GUATEMALA. пи Harmon 1934 (MO). SACATEPEQUEZ: Roe 786 (WIS). SUCHITE- UEZ: Standley 88950 (F). EL SALVADOR. AHUACHAPAN: Croat 42101 (MO). LA LIBERTAD: Calderón 1513 (GH). SONSONATE: Tucker 1336 (F) Material representativo. Inga ciliata C. Presl, Symb. Bot. 2: 11, t. 58. 1834. TIPO: Brasil: Lhotky s.n. (holotipo, PR; isotipo, С). Para sinónimos véase a Bentham (1876) agre- gando а Í. leptantha Benth. Especie hasta hace poco reportada sólo de la costa este de Brasil, sin embargo han empezado a obtenerse colectas de Ecuador y Perú asi como de Mesoamérica en Costa Rica y Panamá, estas úl- timas reconocidas por Zamora (en prensa). Inga ciliata C. Presl es una especie muy varia- ble en cuanto al número de sus foliolos y sus glándulas interfoliolares largamente estipitadas a sésiles. Este ültimo carácter varia aun en la misma rama (Martius 1093, MO), y siendo el carácter que empleó Bentham (1845) para distinguir a /. ciliata (elevada) de 1. leptantha (largamente es- tipitada), no es posible mantenerlas aparte; además de que la lámina (t. 58) de la descripción original de 1. ciliata claramente ilustra a las glándulas estipitadas. Dentro de la variación de /nga ciliata, con foliolos un poco más grandes y raquis floral màs robustos, disticos y nudosos, podria entrar muy bien 1. disticha Benth. de las Guayanas y del norte де Brasil; sin embargo, por el momento las mante- nemos separadas n la nueva balón de Inga ciliata, tam- 236 Annals of the Missouri Botanical Garden bién se agrega una serie más para Mesoámerica, la serie Leptanthae Benth. Material црна Costa Rica. LIMON: Mori 348 ales PAN CLE: Folsom 3147 (MO). Ecua- Neill et p 7960 (MO). PERU. AMAZONAS: Mi 1446 (MO). BRASIL. (NORDESTE) BAHIA: Salz- mann s.n. (MO); (SUDESTE) RIO DE JANEIRO: ae hnath Inga cocleensis Pittier, Contr. U.S. от НегЬ. 18: 211. 1916. TIPO: Panama: R. S. Williams 405 (holotipo, NY). De esta especie Elias (1967) nombró una va- riedad de flor más grande para Sudamérica, /nga cocleensis var. megantha T. S. Elias, la cual debe ser una subespecie ya que se trata una variación geográfica. Inga cocleensis subsp. cocleensis aterial representativo. BELICE. EL CAYO: Hoa i Howe 1420 MS Ls Epig 5214 (LL). G 6 (F). HONDURAS. ATLAN- ^ o COLON: Saunders 627 (BM). CORTES: 9 (NY). NICARAGUA. RIO SAN JUAN: Moreno 23008 MO. ZELAYA: Moreno 24986 (МО). Costa in . HEREDIA: McDowell 718 (LL). PA AMA. COCLE: . Williams 405 (NY). COLON: a Pompa et al. E (MEXU). PANAMA: Correa etal. 5 (F). ZONA DEL CANAL: Croat 7941 (MO). Inga cocleensis subsp. megantha (T. S. Elias) . Sousa, stat. nov. /. cocleensis var. me- zona T. S. Elias, Phytologia 14: 211. 1967. TIPO: Colombia: Pérez-Arbelaez & Cuatre- casas 6458 (holotipo, US). Material MOL кока )MBIA. CHOCO: Gentry & d 30366 (MO). vA Mazuera 30 (MO). VENEZUELA. ZULIA: Davidse et a 18297 (MO). ~ Inga cookii Pittier, Contr. U.S. Natl. Herb. 18: 203. 1916. TIPO: Guatemala: Cook & Griggs 505 (holotipo, US). Inga undi ne _ jam ous Mus. Nat. Hist., Bot 22 : Guatemala: C. L. эе 315 ) D. 2 Especie cuya legumbre no se conocia hasta su colecta (Contreras 10337) en 1970, la cual tiene dimensiones de 4-7.5 x 1.8-2.5 x 0.3 aplanada, oblonga, derecha, sésil, apiculada y to- cm, es mentosa con tricomas pardo-amarillentos. Sin em- bargo, se le han adjudicado frutos de otras especies: asi primero Standley y Steyermark (1946) le asig- nan el fruto de /nga pavoniana Don, tanto por su descripción, nombre vulgar de ““machetón,” lo- calidad y colecta (véase Standley 92857, F). Des- pues León (1966) si cita dos colecciones de Hon- duras (Yuncker et al. 60292, 6678), tratandose de Í. tenella M. Sousa. En el caso de León (1966) es interesante comentar que en efecto existen ejem- plares (como Molina R. 24244, NY; 25431, BM) que en gran medida son intermedios entre /. cookii e Í. tenella. Material representativo. GUATEMALA. ALTA VERA- PAZ: Stey rermark 45 159a (Е). EL PETEN: Lundell & Con- treras 20655 (МО). IZABAL: erg 10337 (LL). HONDURAS. ATLANTIDA: Standley 5425 Inga cordistipula C. Martius, Flora 20(2): Beibl. 37. Mimosa plana Vell. Conc., FI. Flumin. 11: t. 10, non 28. 1831. TIPO: Brasil: Vellozo, Fl. Flumin. 11: t. 10. 1831. Inga urceolata Zamora, Brenesia 33: 117. 1991. TIPO: ата: McPherson 10496 (holotipo, MO; isotipos, BM, CR, MEXU, US Martius (1837) transfiere de género y crea un nuevo nombre especifico para Mimosa plana Vell. onc., lámina 10, ésto debido a que la lámina 28 de Vellozo corresponde a una especie de Acacia que lleva el mismo nombre. Inga cordistipula es una especie hasta ahora conocida sólo del sureste de Brasil. Еп Panamá, su variación es reducida, pero en Brasil varia gran- demente, estando representados todos los carac- teres que la podrian distinguir en Mesoamerica; asi su raquis foliar terete, está ilustrado por Barroso (1965) para Guanabara [=Rio de Janeiro] y los foliolos más anchos, se aprecian en el sinónimo /. fluvii-novi Harms (Schwacke 8811, B, foto MO). Sin embargo, el material de Panamá tiende a contar con estipulas un poco más angostas, pero tan largas (véase de Nevers et al. 7382, MO) como su análogo de Brasil. Mat PANAMA. COLON: McPher- son 84: 51 (M Nee et al. 8787 (MO). SAN BLAS: de ines et e 7382 (MO). Шей. drea MINAS GERAIS: Schwacke 8811 (B, foto MO). DE JANEIRO: Glaziou 2977 (BR). erial ик о. МАМ Inga coruscans Humb. et Bonpl. ex Willd., Sp. Pl, , 2: 1017. 1806. TIPO: Colombia: Bonpland 1574 (holotipo, B-W, foto MO). Inga E DA et Killip, Ann. New York Acad. 1936. TIPO: Colombia: Mutis 3539 ibolotipe, E Esta especie se caracteriza, dentro de la sección Bourgonia Benth., por contar con foliolos relati- vamente grandes y hojas brevipecioladas. Inga coruscans es de amplia distribución, pero Volume 80, Number 1 1993 Sousa S. 237 El Género Inga muy mal colectada o quizá de distribución espo- rádica, tanto en Mesoamérica como en Зидате- rica. Material representativo. Costa RICA. CARTAGO: Tonduz 12991 (BM). PUNTARENAS: Liesner 1829 (MO). PANAMA. CHIRIQUI: Croat 22161 (WIS). DARIEN: Duke 13004 (NY). COLOMBIA. ANTIOQUIA: Soejarto et al. 4136 (A). BOYACA: Lawrance 784 (MO). GUYANA: de la Cruz 3599 (MO). ECUADOR. LOS RIOS: Dadon 5227 (MO). PERU. LORETO: Encarnación E-1087 (MO). pasco: Smith 3928 (MO). BOLIVIA. COCHABAMBA: SIDA 448 (MO). BRASIL. (NORTE) AMAZONIA: Krukoff 4769 x~ Inga cuspidata M. Sousa, sp. nov. TIPO: Panamá. cas del Toro: Rio San Pedro, Jan. 1978, B. L. Gordon 3 (holotipo, MO). Specie Ingae multijugae Bentham propinqua, ab ea tamen foliolis ex anguste lanceolatis ellipticis, apice cus- pidato, inflorescentiis paucifloris, rhachide florifera brevi, necnon legumine angustiore valvis strigulosis praedito satis diversa. Arboles 10-12 m; ramas teretes, ferrugineo to- mentulosas cuando jóvenes, posteriormente gla- brescentes. Hojas 12-18-folioladas; estipulas 4-6 mm, liguladas, triangulares a filiformes, más o menos persistentes; peciolo 0.7-2.5 cm, terete; foliolos tenuemente discoloros, cartáceos, la base cuneada, el ápice cuspidado, el haz opaco, glabro excepto tomentoso sobre la nervadura primaria, el envés rojizo, esparcidamente sericeo con tricomas canescentes a pardo-amarillentos, la nervadura pri- maria eglandular; par basal de foliolos 3-4 x 0.9- 1.6 cm, lanceolados a elipticos, el par apical 10.5- 12.5 x 2.6-3.6 cm, lanceolados a angostamente elipticos; raquis foliar 8.5-20 cm, generalmente terete, rara vez angostamente alado, el ala hasta mm ancho, oblanceolada; glándulas interfolio- lares casi sésiles a brevistipitadas, el estipite ro- busto, cupuliformes, el cuerpo glandular 1.8-2 x mm; apéndice 3-4 mm, filiforme. Inflores- cencias en espigas solitarias; pedúnculo 4-6 cm, terete, inconspicuamente estriado, tomentoso a ve- lutino con tricomas ferrugineos; raquis floral 1-2 cm, paucifloro, las flores congestas, las brácteas aparentemente homomorfas ca. 2 mm, cuculadas, pronto caducas. Flores sésiles; botones florales con el cáliz abierto, apiculados; cáliz 11-15 mm, tu- bular, estriado, sericeo con tricomas pardo-ama- rillentos, con escotaduras, los lóbulos 2.5-3 mm, triangulares; corola 25-29 mm, tubular a subtur- inada, amarillenta, vellosa con tricomas pardo- amarillentos; tubo estaminal inserto. Legumbre 10- 17.5 x 1.5-2.1 x 0.8-1.1 cm, algo aplanada a más bien türgida, derecha a curveada, linear-oblon- ga, sésil o casi sésil, apiculada en el ápice, estri- gulosa (más densamente sobre las suturas), las val- las suturas derechas, multisulcadas, los márgenes de las su- vas inconspicuamente nervadas, turas redondeados. Distribución y hábitat. tiente Caribe de Panamá, en las provincias de Bocas Endémica de la ver- del Toro y San Blas. Esta especie habita bajo con- diciones riparias y de vegetación secundaria, en altitudes hasta de 200 m. Florece en diciembre y enero y fructifica a partir de enero a abril. Nombres vernáculos. Guamo de monte, bung- guagua (en Guaymi). Usos. Los Kuna usan la pulpa dulce alrededor de las semillas como comestible. Paratipos. PANAMA. BOCAS DEL TORO: Rio San Pedro, 3 992]'N, 79903", С. (MO); Río Congandi from village of Congandi to just above its confluence with Rio Titamibe, 9%27'N, 79°8’W, de Nevers et al. 4644 (MO); El Llano-Carti road, km 26.5, along Río Саги Chico, 9°19’N, 78°55'W, de Nevers et al. 5336 (MO); along the Rio Sidro at base of Cerro Нађа, 923'N, 78°49'W, К. Sytsma et al. 2613 (MO). Inga cuspidata está emparentada con /. mul- tijuga Benth., de la cual difiere por varios carac- teres (véase diagnosis), además de que en ocasiones los raquis foliares de /. cuspidata son alados y el fruto (maduro) es cilíndrico en vez de aplanado. El epíteto hace énfasis en el ápice cuspidado de los foliolos. Inga chiapensis Miranda ex M. Sousa, sp. nov. TIPO: México. Chiapas: en canada hacia El Aguacate, al N de La Chacona [NO de Tuxtla Gutierrez], 800 m, en selva alta subdecidua, 19 mar. 1950, F. Miranda G. 6127 (holotipo, MEXU Arbores e mediano procerae, ramis juvenilibus brunneo latescent: tomentosis. Folia 2-4-jugata; stipulae 3-4 mm ciduae; foliola tenuiter | tescenti-villosa. Legumen breve, latum, incrassatum, com- карм жез lutescenti-hirsutum. Arboles 5-18 m; ramas teretes, tomentosas con tricomas pardo-amarillentos cuando jóvenes, pos- teriormente glabrescentes. Hojas 4-8-folioladas; estipulas 3-4 mm, ovadas a triangulares, caducas; peciolo 1-3.5 ст, terete a alado; foliolos ligera- mente discoloros, cartáceos, la base ligeramente asimétrica, cuneada a obtusa, el ápice acuminad a cuspidado, el haz ligeramente brillante, piloso después glabrescente, el envés opaco, esparcida- 238 Annals of the Missouri Botanical Garden mente tomentoso, las nervaduras tomentosas con tricomas pardo-amarillentos; la nervadura primaria eglandular; par basal de foliolos 3.7-7 x 2.2-3.5 cm, generalmente elipticos en ocasiones ovados, el par apical 9.5-16 x 4-7.5 cm, lanceolados a eliptico-lanceolados; raquis foliar (2.8-)3.5-9(-15) 6 mm de ancho, oblan- ceolada, raramente obtriangular; glándulas inter- cm, alado, el ala hasta foliolares sésiles а brevistipitadas, urceoladas, el cuerpo glandular 0.6- apendice 9-11 mm, setiforme, pronto caduco. In- mm; » florescencias en espigas, generalmente solitarias; pedünculo (1.5-)3-6.5 cm, subterete, inconspi- cuamente sulcado, tomentoso con tricomas pardo- amarillentos; raquis floral 2.5-4 cm, las flores es- parcidas proximalmente, congestas distalmente, las brácteas homomorfas 1.3-1.8 mm, cóncavas, oblanceolado-cocleadas a ovado-cocleadas, cadu- cas, las bases cortamente decurrentes. Flores sé- siles; botones florales con el cáliz abierto, obtusos; cáliz 8- 10 mm, tubular, estriado, piloso, con es- cotaduras, los lobulos 1.5-2 mm, triangulares; co- rola 15-18 mm, tubular-subturbinada, verdosa, moderadamente vellosa con tricomas pardo-ama- rillentos; tubo estaminal inserto, blanco. Legumbre 8-12.5 x 3.2-3.6 recha, oblonga, sésil en la base, apiculada a obtusa 1.5-2 cm, aplanada, de- en el ápice, hirsuta con tricomas pardo-amarillen- tos, las valvas aplanadas, las suturas derechas, redondeadas, los márgenes de las suturas 1-sul- cados. Distribución y hábitat. Conocida solamente de dos áreas bien definidas, el margen norte de la Depresión Central de Chiapas en vegetación de selvas bajas a altas subcaducifolias a caducifolias y en la zona de Uxpanapa, Veracruz, en selvas altas perennifolias con Dialium, Ormosia (Legu- minosae), Terminalia (Combretaceae), Licania (Chrysobalanaceae), Calophyllum, Symphonia (Guttiferae), Guarea (Meliaceae), Sloanea (Elaeo- carpaceae), Sterculia (Sterculiaceae), y Tapirira (Anacardiaceae). En altitudes de 200 m y de 850- 1,100 m. Florece de mediados de marzo a prin- cipios de mayo; fructifica de junio a septiembre. atipos. MEXICO. VERACRUZ: Mpio. Jesús Carran- Za, е al S de iud 2 (+3-6 km al S del entronque e terracería La Laguna-Sarabia con camino a Ejido La Paz), 17°12’N, 94238 30"O, T. Wendt et al. 5304, 5633 (CHAPA, MEXU). Cutapas: Mpio. Ocozocoautla, а 1 km al NO del entronque А у Ренеа México, carretera 190, 4. es G. 555 (MEXU, MO); Mpio. San Fernando, San ondo. F. Miranda G. 7219 (MEXU); Mpio. Tuxtla em Cerro Hueco, al SSE aie н Vi es Miranda 5570, 6878 (MEXU); en пада а El Aguacate, cerca y al N de La Сћасопа, Miranda 6413 (MEXU); adelante Mirador La С | midero, Miranda 7662, 7718 (MEXU); entre los km 18-22 de la carretera de Tuxtla Gutiérez hacia e Canon del Sumidero, O. Téllez V. et al. 6690, 757 XU). Inga chiapensis muestra en Mesoamérica si- militudes con /. mucuna Walp. et Duchass. e /. leonis Zamora; con esta ültima son más estrechas, pero no suficientes para unirlas. Su variación se puede agrupar en la de la Depresión Central de Chiapas y la de Uxpanapa, Veracruz: en la primera area 1. chiapensis muestra el cáliz ligeramente mayor, la corola más chica y las hojas que tienden a poseer más foliolos que en el área de Uxpanapa, además que varian en cuanto a hábitats y altitud. Sin embargo, por el momento las mantengo bajo un sólo epiteto con la posible distinción posterior en categorias infraespecificas, dependiendo de más co- lectas, sobre todo del área baja hümeda de Ux- panapa. Esta especie fue reconocida como entidad nueva por Faustino Miranda Gonzalez hace más de 40 anos y en su honor se emplea el epiteto que el designó, a pesar de que ya se le conoce hoy en dia fuera de Chiapas. Inga chocoensis Killip ex T. S. Elias, Phytologia 14: 210. 1967. TIPO: Goles Cuatrecasas 16372 (isotipo, F). Inga d arpa Zamora et Poveda, Brenesia 33: 101. TIPO: : Costa Rica: Herrera & Rivera 842 пе CR; isotipos, MEXU, МО). Esta interesante especie está cercanamente em- parentada con /nga longipes Benth. del alto Ama- zonas de Реги e /. duquei Harms del Valle de Cauca en Colombia; sin embargo, hay varios ca- racteres que las separan. Material representativo. NICARAGUA. MATAGALPA: Veill 1746 (MO). Costa Rica. ALAJUELA: Utley & Utley 3213 (F). HEREDIA: Bawa 544 pe d “AS DEL TORO: Proctor Cooper 527 (К). coc (MO). PANAMA: Porter et al. 4185 (MO). nou BIA. VA LLE: Cuatrecasas 16098 (F). ECUADOR: ESMERALDAS: Lit tle & Dixon 21023 (MO) Inga dasycarpa M. Sousa, sp. nov. TIPO: México. Chiapas: 12 km al E de Tziscao, sobre el camino a Santa Elena, en el Parque Nacional Lagos de Monte Bello, 1 oct. 1983, Е. Са- brera & Н. de Cabrera 5885 (holotipo, MEX U; isotipo, MO). Figura 2. rbores medianae; rami juveniles dense ferrugineo- tomentosi. Folia 2-3-jug u glandulalis ex a stipitatis praedita. d rescentiae capituliformes; pedunculus 1- cm rhachis floralis subglobosa; bes ligu TM | siles; calyx 7-9 mm longus, tubularis, inconspicue stria- Volume 80, Number 1 1993 Sousa S. 239 EI Género /nga 7 a FIGURA 2. Inga dasycarpa M. Sousa.— А. Rama con inflorescencia. — B. Detalle de estípulas. —C. Glándula interfoliolar. — D. Flor. — E. Cáliz. — F. Corola. — С. Tubo estaminal. — H. Gineceo. (Tomado de Е. Cabrera 5885.) — F I. Fruto. (Tomado de О. Téllez 642.) tus; corolla 17-19 mm longa. Legumen usque ad 6.5 cm latum, complanatum, oblongum, ferrugineo-hirsutum. Arboles 4-15 m; ramas teretes, densamente ferrugineo tomentosas cuando jóvenes, posterior- mente glabrescentes. Hojas 4-6 folioladas; esti- pulas 1.5-2 mm, liguladas, caducas; peciolo 1-2 cm, terete a alado; foliolos generalmente conco- loros, en ocasiones ligeramente discoloros, carta- 240 Annals of the Missouri Botanical Garden ceos a coriaceos, la base simétrica a ligeramente asimétrica, obtusa a cuneada, el ápice agudo a acuminado, el haz ligeramente brillante, modera- damente a esparcidamente piloso, el envés opaco, moderadamente piloso, la nervadura primaria eglandular; par basal de foliolos 3.3-5.1(-7) x 1.1-2.5(-4.5) ст, ovados a lanceolados, el par apical 7- 10(- 12.5) x 3.2-5.2(-7) cm, elipticos a obovados; raquis foliar 2.5-4(-7) cm, alado, el ala 5-7(-13) mm de ancho, eliptica, generalmente oblanceolada, rara vez obtriangular; glándulas in- terfoliolares subsésiles a estipitadas, clavuliformes, el cuerpo glandular 0.7-0.9 x dice, ausente. Inflorescencias capituliformes, 1-2 0.7—0.9 mm; apén- fasciculadas; pedünculo 1. cm, subterete, in- conspicuamente sulcado, ferrugineo tomentoso; ra- quis floral ca. 4 mm, subgloboso, las flores con- brácteas homomorfas 1.5-2 mm, concavas, liguladas, caducas. Flores sésiles; boto- gestas, las nes florales con el cáliz cubriendo, desconocidos; cáliz 7-9 mm, tubular, inconspicuamente estriado, tomentoso con tricomas pardo-amarillentos, con escotaduras, los lóbulos 1.5-2 mm, ligulados; co- rola 17-19 mm, infundibuliforme, blanca, canes- cente vellosa; tubo estaminal inserto, blanco. Le- gumbre 15-21 x 4-6.5 x 0 derecha, oblonga, sésil, apiculada a rostrada en el ápice, ferrugineo hirsuta, las valvas aplanadas, las 5 cm, aplanada, suturas más o menos derechas, marginadas, los márgenes de las suturas 1-sulcados. De Chiapas, México a Cortés, Honduras en regiones montañosas, en Distribución y hábitat. vegetación de bosques caducifolios con Liquid- ambar (Hamamelidaceae), Magnolia (Magnolia- ceae), Clethra (Clethraceae), Vochysia (Vochysia- ceae), también en bosques de Pinus y Quercus. En altitudes entre los 1,350 a 1,600 m. Florece en octubre y fructifica de diciembre a abril. Paratipos. MEXICO. CHIAPAS: Mpio. La dl 15 km E of La Trinitaria on road to = По, E. Breedlove 42062 (MEXU, МО); E of Laguna Taeke Monte eve National Park, M. СЕЕ et al. 11876 E MEXU, MO); a 10 km al E de Montebello, a 1 km . et al. 642 (МЕХО), Montaña San Idalfonso entre Baña- deros y unico: A. Molina R. 11459 (LL, NY, US). Inga dasycarpa se encuentra muy relacionada con /. villosissima Benth. de las montañas de Vene- zuela, pero en /. dasycarpa las inflorescencias tienden a ser capituliformes, las brácteas más pe- quenas y pronto caducas, las flores más chicas y las estipulas de „лепог tamaño y pronto deciduas. ambién está mu; relacionada con /. tenella M. Sousa, difiriendo de ella por tener /nga dasycarpa inflorescencias capituliformes, flores y legumbres de mayor talla, y las hojas coriaceas que en general poseen menos foliolos. Inga dasycarpa e I. tenella son dos especies que en gran medida son simpá- tricas, tanto en áreas geográficas generales como en los mismos hábitats, sin que hayamos encontrado posibles hibridos. Estas especies aparentemente re- fuerzan su aislamiento con periodos de floración diferentes. El epiteto hace referencia a la pelosidad de la legumbre. Inga davidsei M. Sousa, sp. nov. TIPO: Guate- mala. Izabal: El Estor, 23 Mar. 1972, E. Contreras 11483 (holotipo, MEXU; isotipos, LL, MO). Figura 3. Species Ingae goldmanii Pittier proxima, a qua ta foliis foliola numerosiora habentibus, stipulis longioribus quam latioribus, foliolis coriaceis et supra nitentibus, glan- dulis per nervum mediu sentibus, praesente, glandulis nena longe stipitatis, flo- ribusque maioribus recedit. Arboles 7-20 m; ramas subteretes, ferrugineo hirsutulas cuando jóvenes, posteriormente glabres- centes. Hojas (6-)8(-10)-folioladas; estipulas -)7-14 mm, liguladas a triangulares, caducas; peciolo (1.5-)2.5-5 cm, terete; foliolos discoloros subcoriáceos, la base simétrica a casi simétrica, cuneada, el ápice obtuso a acuminado, el haz bri- llante, glabro excepto ferrugineo tomentoso en la nervadura primaria, el envés opaco, piloso con la nervadura primaria tomentosa como el haz, la ner- vadura primaria eglandular; par basal de foliolos 7-9.5-13) x 3.2-4.3(-9) cm, elipticos a oblon- go-elipticos, el par apical 15-18(-23) x 5.5-9 ( cm, generalmente elipticos, en ocasiones obovados; raquis foliar 9.5-15 cm, ferrugineo to- mentoso, terete, marginado a angostamente alado principalmente entre los 3 pares distales de foliolos, el ala hasta 9 mm de ancho, eliptica; glándulas interfoliolares generalmente largamente estipita- das, clavuliformes, el cuerpo glandular 0.5- 5-0.9 mm; apéndice 10-25 mm, setiforme, ar- queado, caduco. Inflorescencias en espigas solita- rias; pedunculo 6–8.5 cm, terete, aparentemente liso, ferrugineo tomentoso; raquis floral 0.6-1 cm, las flores congestas, las brácteas homomorfas 6– mm, gruesas, ligeramente concavas triangular- liguladas, una estéril en el 14-14 superior del pe- dúnculo, persistentes. Flores sésiles; botones flora- les con el cáliz cubriendo desconocidos; cáliz 17— 19 mm, ciatiforme, lanoso con tricomas pardo a pardo-amarillentos, sin escotaduras, los lóbulos 3— 4 mm, triangular-ligulados; corola 28-32 mm, sub- turbinada, pardo-amarillento lanosa; tubo estami- nal inserto, blanco. Legumbre 24-30 x 3.3-6.5 Volume 80, Number 1 Sousa S. 241 EI Género /nga Y 2 а КА 3. = davidsei M. Sousa.—A. Rama con inflorescencia. — B. Detalle de estipulas.—C. Glandula жаз — D. Flor. — E. Cáliz. — F. Tubs estaminal. —G. Gineceo. (Tomado de E. Contreras 11483.)—H. Fruto. (Tomado de G. Davidse & А. Brant 32330.) X 0.9-1.5 cm, aplanda, ligeramente espiralada, Distribución y hábitat. Al sur de la Peninsula oblonga sésil en la base, apiculada en el ápice, de Yucatan, en el norte de Guatemala, Belice, y ferrugineo hirsuta, las valvas aplanadas, las suturas Honduras en selvas bajas a medianas caducifolias, derechas, redondeadas, los márgenes de las suturas en sitios riparios y en la orilla de un pinar, sobre aplanados а 1-sulcados. suelos someros arcillosos, en altitudes entre los 100 Annals of the Missouri Botanical Garden a 900 m. Florece de mediados de febrero a marzo y fructifica de marzo a abril. Nombres vernáculos. Palo de paterno sil- vestre, cola de mico. sos. El fruto es comestible y tiene sabor a frijoles. Farat ti 05. BELICE. EL CAYO: Dent's Drive, ras . Davidse & A. E. Brant 32330 (MEXU, MO). ;UATEMALA. ALTA VERAPAZ: Cook & Griggs 380 (US). HUEHUETENANGO: Cerro Jolomtac, above Finca San Ra- fael, Sierra de los аана Ј. А. add 49146 (A, : Lancetilla, Mama Fes Isidro, R. Cruz 275 (ЕЕН), L as Mangas road La Ceiba- Yaruca, above Las Mangas, Pennington & P. House 13404 (K). ga davidsei está emparentada con /. gold- manii Pittier del sur de Мезоатепса al norte de Sudamérica; sin embargo, difiere de ella tanto en caracteres vegetativos como en cuanto al tamano de la flor (véase diagnosis). Se dedica esta especie al Gerrit Davidse (1942-), botánico distinguido de Jardin Botánico de Missouri y colega del autor en el proyecto Flora Mesoam- ericana, quien colectó esta especie bajo el desar- rollo del proyecto. Inga densiflora Benth., Trans. Linn. Soc. Lon- о 7. 1875. TIPO: Perú. San Martin: Tarapoto, Spruc e 4504 (fl, fr) (lectotipo, de- signado aqui, K herb. Benth.; GH, K (3 ej.)). isolectotipos, Para sinónimos véase León (1966). León (1966) designó a Spruce 4504 como el tipo de /nga densiflora sin mencionar el herbario en que está depositado ni que ejemplar, por lo que se considera una lectotipificación incompleta. Aqui se lectotipifica designando a K en el herb. Bentham, al ejemplar con flores y frutos. nga densiflora Benth., como ya lo apuntó León (1966), es una especie muy variable, particular- mente en la pelosidad y en la forma de sus glándulas interfoliolares. También a esta variación, segura- mente, el hombre ha contribuido, asi como a am- pliar su distribución, por su uso como sombra de café. Es muy probable que en el área cafetalera del Soconusco en Chiapas, México, el hombre sea el responsable de la presencia de esta especie, aunque en la actualidad se ај también en estado silvestre “escapada Marti- ULEU: MEXICO. CHIAPAS: GUATEMALA. RETALH Material representativo. nez S. 19847 (MEXU). MacQueen & Hughes a NICARAGUA. CHONTA- LES: Bunting & Licht 1173 (F). MATAGALPA: J. B. Salas & B. Taylor 2485 Ce. ZELAYA: ла 3447 (МО). COSTA RICA. ALAJUELA: Croat 46799 (MO). CAR- TAGO: сола и 388 D d M Sousa et al. (MO). HERRERA: ipsc nd a 285 3 (MO). per 65 1 (NY). VERAGUAS: ned 6232 ANTIOQUIA: Archer 658 (US). CAUCA: Lehmann 904 (US). vALLE: Langlassé 63 (US). A. DISTRIT DERAL: Pittier 13028 (MO). TACHIRA: Steyermark et al. 119675 Boer а. Davidse et al. 20916 (MO). “¿CUADOR n & Cerón 3056 (MO). PERU. MAD- RE DE DIOS: n p Stratton. 234 (MO). SAN MARTIN: Spruce 4504 (GH). Inga dwyeri M. Sousa, sp. nov. TIPO: Panamá. Panama: Cerro Jefe to 2,900 ft., 5 Jan. 1972, J. D. Dwyer & А. Gentry 9494 (holotipo, MO; isotipo, MEXU) Arbores АЕ ramis juvenilibus ferrugineo-hirsutu- lum. Folia (2 sh qud run 2-3.5(-5) mm longae, persistentes; foliola chartacea bullata; rhachis foliaris teres, interdum anguste alata, glandulis ex subsessilibus stipi- i i ie orescentiae spicatae, an- floribus distantibus, bracteis ' yx tubu ferrugineo-villosus, emarginatus, corolla tubulosa, canes- centi luteo-villosa. Legumen parvum, complanatum, ob- longum, brunneo-luteo-hirsutum. Arboles 3-10 m; ramas teretes, ferrugineo hir- sütulas cuando jóvenes, posteriormente glabres- centes. Hojas (4—)6-folioladas; estipulas 2-3.5(-5) mm, ovado-cordatas a triangular-subuladas, per- sistentes; peciolo 1—1.8(-3) ст, terete а margi- nado; foliolos ligeramente discoloros, cartáceos, abollados, la base asimétrica a ligeramente asi- métrica, cuneada a truncada, el apice mucronulato a acuminado, el haz opaco, casi glabro excepto ferrugineo tomentoso en la nervadura primaria, el envés opaco con pelosidad similar al haz, la ner- vadura primaria eglandular; par basal de foliolos —T(-11) x 2.1-3.7(-4.8) ст, ovados a elip- ticos, el par apical 7-9 x 2.8-4.2 cm, elipticos a subobovados; raquis foliar 1.5—4.1(- 7.5) cm, generalmente terete, en ocasiones alado, el ala hasta 4 mm de an glándulas interfoliolares subsésiles a robustamente estipitadas, claviformes, el cuerpo glandular 0.6– 8 mm; apéndice 2-3 mm, pronto caduco. Inflorescencias en espigas, 1-3 fascicu- ladas; pedúnculo (0.5-)2.5-4.5 cm, delgado, más o menos aplanado, sulcado, ferrugineo hirsútulo; raquis floral 1.3-3 cm, las flores espaciadas, las brácteas homomorfas, 2-3 mm, cóncavas, ovadas, cho, angostamente oblanceolada; Volume 80, Number 1 1993 Sousa S. 243 EI Género /nga persistentes (en fruto en ocasiones lignifican). Flo- res sesiles; botones florales con el cáliz cerrado, atenuado-apiculado; cáliz 6-9 mm, tubular, estria- do, esparcidamente ferrugineo velloso, con esco- taduras, los lóbulos 1.2-1.8 mm, triangulares; co- tubular, canescentes a amarillos; tubo estaminal exerto, blanco. Legumbre 9-10.5 x (1.9-)2.6-2.9 x 0.3- 0.7 ст, aplanada, derecha, oblonga, sésil, apiculada rola - mm, vellosa con tricomas a rostrada en el ápice, moderadamente hirsuta con tricomas pardo-amarillentos, las suturas derechas a constrictas, nervadas, los márgenes de las suturas aplanados a sulcados. Distribución y hábitat. Sólo conocida de Panama, entre 225 y 950 m. Florece en febrero, abril, y septiembre y fructifica de noviembre a enero Paratipos. PANAMA. DARIEN: Rio Pierre, 10 mi. S of El Real near Dos Bocas, R. Foster & T. Lowenbach 2272 (MO). PANAMA: Goofy Lake to ca. 8 mi. S of Goofy Lake toward Cerro Jefe, J. D. Dwyer 70694 (МО); Cerro Azul, A. Gentry 6132 (F, МО); along El Llano to Carti road, about 8 mi. from Pan-American highway, G. McPherson 10487 (MO). san BLAS: El Llano-Cartí Road, 9%19'N 78955 М, Н. Herrera & S. Kaza 7334 (MO). Especie muy variable, cuyas hojas de brotes vegetativos además de ser de mayor talla, tienden a tener el raquis foliar alado. Muy caracteristica por sus foliolos abollados y sus inflorescencias es- beltas y laxas con brácteas persistentes. Sus re- laciones en Mesoamérica parecen ser con /nga pinetorum Pittier e 1. hayesii Benth. Se dedica esta especie a John D. Dwyer (1915-), quien ha contribuido grandemente al conocimiento de la floristica mesoamericana. ver РЕГ C. Martius, Flora 20: ВеіЫ. 113. 837. EM ynga Vell. Conc., Fl. Flumin. 1 t. 3. 1831, non L. ПРО: Brasil: Vellozo, ЕІ. E dra TRES 1 Especie muy caracteristica por su robustez tanto de hojas y flores como de fruto. Bajo este concepto Inga edulis no se le ha colectado o es escasa en las Guayanas, donde /. scabriuscula Benth. habita. Inga edulis es sin duda una especie de la Ama- zonia, que como propone León (1966) posiblemente fue introducida por el hombre a Mesoamérica y quiza también en Colombia: Antioquia y Chocó, ya que siempre está asociada con asentamientos hu- manos y en plantaciones de café y cacao. Material representativo. HONDURAS. MORAZAN: Standley 16086 (F). NICARAGUA. ZELAYA: Bunting « Licht 679 (F). Costa RICA. LIMON: Grayum & Sleeper 1357 (MO). PANAMA. BOCAS DEL TORO: Тода 2560 (TEX). CHIRIQUI: Schmalzel 1332 (MO). DARIEN: Croat 27311 (MO). PANAMA: bush 2145 (MO). COLOMBIA. AMAZONAS: Gillett 16511 (MO). se Shepherd s.n. (MO). CAQUETA: n et al. О). CHOCO: Forero et al. 3070 (MO). vAUPES: Sc А et al. 24042 MO). VENEZUELA. AMAZONAS: Ll. Williams 14452 (MO). ELTA AMACURO: Marcano Berti 382 (F). ECUADOR. NAPO: Neill 63 18 (MO). PASTAZA: Neill & Palacios 6558 (MO). PERU. AMAZONAS: Huashikat 1612 (MO). LORETO: Croat 17522 (MO). pasco: D. N. Smith 3818 (MO). SAN MAR- TIN: Knapp et al. 7270 (MO). UCAYALI: Gentry & Díaz 58361 (MO). BOLIVIA. BENI: Boom 4563 (NY). (NORTE) AMAZONIA: Prance et al. 10793 (MO). P Austin & Cavalcante 4139 (MO). os Inga exalata T. S. Elias, Phytologia 14: 208. 1967. TIPO: Costa Rica: Brenes 3634 (holo- tipo, F; isotipo, NY) Esta especie sólo hasta hace poco se colectó en Panamá. Sus relaciones no son muy claras, sin embargo se le puede colocar tentativamente en el complejo de /nga villosissima Benth., su pelosidad general y glándulas interfoliolares como tanto por por su legumbre. Material а Costa RICA. ALAJUELA: Brenes 17057 (F). CARTAGO: Poveda 941 (F). PANAMA. COCLE: Me Mee 136. 32 (MO). PANAMA: McPherson 12347 (MO). Inga filiformis Zamora, Brenesia 33: 103. 1991. TIPO: Panamá: Mori et al. 6824 (holotipo, МО; isotipo; CR). Esta especie tan caracteristica se relaciona, al menos en parte, con /nga saffordiana Pittier, en cuanto a la forma de sus estipulas, el raquis foliar terete, y los lobulos de cáliz filiformes, pero по casi libres. Esta especie presenta una variante en el Cerro Jefe con foliolos velutinos en el envés (Systma & Anderson 1984, MO), la cual se encuentra en un piso altitudinal mayor. Mat PANAMA. COLON: Correa & Dressler 686 ANAMA: Sen a & Anderson 4442 dp SAN BLAS: “Herre a& yee 400 (MO). ZONA DEL CANAL: Tyson et ка 4536 (МО). erial Epis que ME Inga flagelliformis (Vell. Conc.) C. Martius, Flo- ra 20(2): Beibl. 112. 1837. тоза Ме Conc., Basiónimo: Mi- Fl. Flumin. Vellozo, Fl. : Brasil: "iuda үч Ё pe a Inga Јан Liesner et D'Arcy, Ann. esi Bot. Gard. fig. 1. 1988. TIPO: má: Foster y Eod 1828 (isotipos, F, MO, US). 244 Annals of the Missouri Botanical Garden Inga flagelliformis es una especie muy variable, la cual tiende a tener la inflorescencia sobre tallos gruesos ya sin hojas. En Panamá y Colombia y en la mayoria del material de Brasil, las umbelas son axilares y solitarias a fasciculadas; sin embargo, en un ejemplar de Minas Gerais (Duarte et al. 2765, MO) las inflorescencias se presentan en tallos grue- sos. La legumbre se presenta de casi recta (Foster y Kennedy 1828) a helicoidal (Monsalve B. 988). Dentro de esta variación Inga jefensis Liesner et D'Arcy encaja muy bien, por lo que la incluimos bajo 1. flagelliformis. sta especie está cercanamente emparentada al complejo de /nga tenuistipula Ducke, 1. lallensis Spruce ex Benth., e /. tessmannii Harms, todas ellas con una flor muy similar y pedicelos más cortos o faltantes. ~ Material тергени: PANAMA. COCLE: Porter е al. 4400 (MO). COLON: Gentry et al. 8854 (МО). DARIEN: Antonio & Нани 4325 (MO). PANAMA: Churchill & de Vevers ( OLOMBIA. CHOCO: Duke 9580 (NY) VALLE: Monsalve B. 988 (MO). BRASIL. (NORTE) AMAPA S. Mori et al. 157 . AMAZONIA: Cid et al. 77 (GH). PARA: Duc “ Ryu (US). (SUDESTE) MINAS GERAIS: Duar- te et al. 5 (MO). RIO DE JANEIRO: Luschnath s.n. (MO) Inga flexuosa Schldl., Linnaea 12: 559. 1838. TIPO: Mexico. Veracruz: Schiede 674 (isoti- ров, BR, С, GH). Inga BS. Harms, Repert. Spec. Nov. Regni Veg. 13: 526. 1915. TIPO: México. Veracruz: das 8125 зи А, BM, F, MEXU, MO, US Steudel creó un nuevo nombre, /nga schie- deana, para Inga flexuosa Schldl. aduciendo la ya existente /. flexuosa de Graham, sin embargo, este ültimo nombre carece de descripción (nom. nud.) y sólo se encuentra en una lista en manuscrito (copia en MO), y por lo tanto no se considera válidamente publicado y homónimo tardío. La identidad de Inga flexuosa ha permanecido en duda desde Bentham (1875) y Standley (1922), ésto debido a la pobre descripción original de la no se puede crear un parte fértil, así como lo estéril de los duplicados vistos; sin embargo, Britton & Rose (1928) la equi- paran como /. pringlei Harms, lo cual parece posible por eliminación, considerando las pocas es- pecies que hay en el área de la localidad del tipo, en los alrededores de Jalapa, Veracruz. En Inga flexuosa las bracteas lineares muy lar- gas que en ocasiones sobrepasan al cáliz, los botones florales con el cáliz abierto cuyos lóbulos son con- niventes a todo lo largo formando un ápice rostrado a apiculado, asi como la согоја más pequena la separan de /. micheliana Harms, especie que le es muy afin. Material representativo. | MEXICO. OAXACA: Sousa et al. 9355 (MEXU). vERACRUZ: Castilleja 66 (ХАТ). Inga glomeriflora Ducke, Arch. Jard. Bot. Rio de Janeiro 3: 1922. TIPO: Brasil. Para: A. Ducke RB16609 (isotipos, BM, С, US Inga Li capi Zamora, Brenesia 33: 107. 1991. јозга Rica: Hammel et al. 16963 (holotipo, CR; isotipos, MEXU, MO). Esta especie se ubica en la seccion Bourgonia, teniendo relaciones estrechas, pero es diferente a nga fanchoniana Poncy de la Guayana Francesa, con la cual comparte varios caracteres, asi ambas especies tienden frecuentemente a carecer de glan- dulas interfoliolares en los ültimos pares de foliolos, sus hojas son glabras con ejes teretes, y las le- gumbres son circinado-espiraladas. Material representativo. Costa RICA. PUNTARENAS: Hammel d E 16963 (MO). PANAMA. PANAMA: Dwyer 2 (US) COLOMBIA. META: Callejas 7013 (MEXU). on (NORTE) AMAZONIA: Ducke 1207 (US). PARA: Ducke RB 16609 (US). Inga goldmanii Pittier, Contr. U.S. Natl. Herb. 18: 198. 1916. про: Panamá: Goldman 1866 (holotipo, US). Esta vistosa especie muestra grandes afinidades con /nga davidsei M. Sousa en Mesoamérica (véase esta), de ella Leon (1966) nombro la serie Gold- manianae León basandose en las ооо florales, particularmente su robusto cáliz. En especie también son notables las glándulas clie supernumerarias (presentes tambien en /. sierrae Britton et Killip); este ultimo caracter parece haber aparecido mas de una vez en el género Inga ya que también esta presente en /. adenophylla Pit- tier e 1. chocoensis Killip ex T. S. Elias, especies claramente de la sección /nga que cuentan con la legumbre subterete y acordonada que es caracter- istica de la sección. Material representativo. NICARAGUA. MATAGALPA: Grijalva & Grijalva 1382 (MO). zELAYA: Ortiz 1255 MO). Costa RICA. ALAJUELA: ini а o 95 1 (A). LIMON: Lent 3690 (MO). PANAMA. BOCAS ORO: Dwyer 4464 (MO). согом: McPherson & Merello 8218 (MO). Los SANTOS: Duke 11904 (MO). PANAMA: Knapp & Mallet 3932 (MO). VERAGUAS: Dwyer 4310 (MO). ZONA DEL CANAL: /Vee 8924 (MO). COLOMBIA. VALLE: Gentry 62901 (MO) = Volume 80, Number 1 1993 Sousa S. 245 EI Género /nga Inga hayesii Benth., Trans. Linn. Soc. London 30: 617. 1875. TIPO: Panama: Sutton Hayes 62 (holotipo, K). Especie conocida hasta hace poco sólo de Pana- má, pero con las colectas recientes ya se cuenta con material de Costa Rica y Colombia. Material representativo. COSTA RICA. PUNTARENAS: Poveda 658 (MO). cn ПР Stern et al. 964 (MO). PANAMA: Dwyer 14844 (MO). ZONA DEL CANAL: Croat 14912 (MO). COLOMBIA. ANTIOQUIA: Zarucchi et al. 4950 (MO). cuoco: Gentry & Brand 36969 (MO). Inga herrerae Zamora, Brenesia 33: 107. 1991. TIPO: Costa Rica: Herrera 1524 (holotipo, CR; isotipos, MEXU, ). Especie endémica de las montanas de Guana- caste, Costa Rica, caracteristica por sus foliolos subcoriáceos a coriáceos y numerosos, y como Za- mora (en prensa) ya ha apuntado sus peciólulos muy cortos. Es muy probable que el material en fruto de Wilbur & Stone 10703 pertenezca a esta especie. Material representativo. COSTA RICA. GUANACASTE: Wilbur & Stone 10703 (NY) Inga heterophylla Willd., Sp. Pl. ed. 4, 4: 1020. 06. TIPO: Brasil: Hoffmansegg s.n. (holo- tipo, B-W, foto MO) Para sinónimos véase Poncy (1985). Especie que ha sido confundida en Mesoamérica con Inga sertulifera DC. Inga heterophylla Willd. se trata de una especie fundamentalmente de la Amazonia que marginalmente entra а Мезоате- rica, en el Darién, en Panamá. Su inflorescencia es muy variable y va de un racimo corto corimboide hasta una umbela, lo que ha creado problemas en su identificación, ya que se le ha colocado general- mente en la sección Leptinga Benth., sección que en general varios autores la caracterizan por contar inflorescencias en umbelas. Sin embargo Ben- tham (1845) claramente la define como de racimos umbeliformes. erial representativo. PANA DARIEN: Knapp & Mallet 3091 (MO). TRINIDAD: Broadway 5839 (MO). SuRINAM: B For. Dept.] 36. 30 (MO). AYACUCHO: Weberbauer 7 : uds ). JUNIN: Stein & Rana 2395 (MO). MADRE DE 0105: R. hey rs et al. 12409 (MO). LORETO: Gentry et aL 39333 (MO). SAN MARTIN: Klug 3521 (US). BOLIVIA: Solomon 9489 (MO). 4534 (MO). (sUR) PARANA: Hatschbach 17964 (F Inga as Sandw., Bull. Misc. Inform. 1937: ‚1 como “hintoni””). TIPO: México. na Mexico: Hinton 7617 (holotipo, K; isoti- po, Inga davidsoniae Standley, Publ. Field Mus. Nat. Hist., Bot. Ser. 2 . 1940. TIPO: Panamá: Пауел 943 (holotipo, F; isotipos, MO, US). Inga nubigena А. R. Molina, Ceiba 18: 99. 1971. про: onduras: Molina R. 10656 (holotipo, F; isotipo, Inga hintonii es una especie muy variable, que se localiza en México y Мезоатепса de Nayarit, México a Chiriqui, Panamá; siempre en bosques hümedos de montana, en ocasiones a más de 2,100 m de altitud. Su pariente más cercano es, sin duda, І. micheliana Harms, especie endémica a la ver- tiente del Golfo de Mexico de Veracruz a Guate- mala; de ella se distingue fundamentalmente por la flor más grande de 1. micheliana. Material representativo. | MEXICO. CHIAPAS: Sousa et al. 13178 (MEXU). jalisco: Magallanes 4445 (MEXU). EDO. MEXICO: pee R. 462 U). MiCHOACAN: Chá- zaro B. et al. 5 MEXU). NAYARIT: Téllez V. 10021 (MEXU). OAXACA: "ares T. 1820 (MEX U). GUATEMALA. SOLOLA: Steyermark 47063 (F). HONDURAS. CORTES: Mo lina R. 7251 (F). INTIBUCA: Molina R. & Molina 24438 (F). LA PAZ: е ы MORAZAN: Cruz NA: Molina R. & Montalvo 21494 (МО). s SONSONATE: t: Molina R. & | d (NY). NICARAGUA. МАТАСАТР. ZELAYA: /Veill 3858 (MO). CE Hn ALAJUELA: on 6805 (Е). PUNTARENAS: Koptur SK-115 (MO). SAN JOSE: Stevens 13671 (MO). PANAMA. CHIRIQUI: Stern et al. 1046 (MO). Inga ismaelis M. Sousa, sp. nov. TIPO: Mexico. Veracruz: km 5 camino Plan de Arroyo- Ar- royo Alegre, Mpio. Hidalgotitlan, 17?15'N, 94?35'O, 140 m, 24 abr. 1974, J. Dorantes et al. 3011 (ћојопро, MEXU; isotipo, NY). Figura 4. Arbores ex mediano procerae; rami juveniles ex ане dos -ferrugineo-setosi hispidi. Folia (3-)4-5 stipulae 10-14 mm longae, triangulari-subu йе foliola ere ses sae co rymboideae, dcinde florali brevi, bracteis elongatis per- sistentibus. Flores prosimales pedicellati, distales fere ses- siles; alabastra floralia calyce aperto; calyx tubulosus, lobulis e iE -subulatis; corolla tubulosa, canescenti-villosa. Legumen usque ad 27 cm longum, complanatum, crassum, oblongum, sparse hispidum. Arboles 6–20 m; ramas teretes, moderadamente setosas a hispidas con tricomas amarillos a ferru- gineos cuando jóve pos mente glabres- centes. Hojas (6–)8– "10. tdlioladas; estipulas 10- 4 mm, triangular-subuladas, persistentes; peciolo 1-2 cm, angostamente alado; foliolos tenuemente 246 Annals of the Missouri Botanical Garden WCU а S "м, ae 2 NAAA SY 207 "с, ES J FIGURA 4. Inga ismaelis M. Sousa. — A. Rama con inflorescencias. — B. Detalle de estipulas.— C. Еоћојо. — D. Detalle de glándula interfoliolar y ápendice. (Tomado de J. Grimes et al. 2817.) —E. Botón floral. —F. Flor.—G. Cáliz. — Н. Corola. —1. Tubo estaminal. —J. Gineceo. (Tomado de J. Dorantes et al. 3011.)—K. Fruto. (Tomado de J. Dorantes 3493.) Volume 80, Number 1 Sousa S. 247 EI Género /nga discoloros, cartáceos, la base ligeramente asime- trica, cuneada a obtusa, el ápice agudo a cuspidado, el haz opaco, moderadamente a esparcidamente piloso, el envés pálido, moderadamente piloso, la nervadura primaria eglandular; par basal de foliolos 4.5-7.5 x 2.5-4.3 apical 15-20 x 5.5-8 cm, lanceolados o elipticos cm, ovados a elipticos, el par a oblanceolados; raquis foliar 5.5-16 cm, alado, el ala 5-11(-18) mm de ancho, generalmente elip- tica, en ocasiones oblanceolada; glándulas inter- foliolares en ocasiones ausentes, sésiles о casi sés- iles, urceoladas cuando presentes, el cuerpo glandular 0.8-1 x 0.8-1 mm; apéndice 6-12 mm, setiforme. Inflorescencias en racimos corim- boides, 1-2 fasciculadas; pedúnculo 5-8 cm, más o menos aplanado, sulcado, setoso; raquis floral 1.5-3 cm, las flores espaciadas proximalmente, congestas distalmente, las brácteas homomorfas 15 mm, triangular-subuladas а lanceolado-su- buladas, generalmente con una estéril en el М superior del pedünculo, persistentes. Flores pedi- celadas las proximales a casi sésiles las distales, el pedicelo hasta 9 mm, delgado; botones florales con el cáliz abierto, los lóbulos erectos no conniventes; cáliz 16-21 mm, tubular, estriado, piloso, sin es- cotaduras, los lóbulos 5-7 mm, triangular-subu- lados; corola 23-27 mm, blanca, canescente vellosa; tubo estaminal inserto tubular a subturbinada, a cortamente exerto, blanco. Legumbre ca. 27 3.5 x 0.6 linear-oblonga, sésil, rostrada en el ápice, espar- cm, aplanada, más o menos derecha, cidamente hispida, las valvas aplanadas, inconspi- cuamente nervadas, las suturas prominentes, los márgenes de las suturas l-sulcados. Distribución y hábitat. Del área de Uxpana- en el norte de Oaxaca, México, en sitios riparios en selvas altas perennifolias con Dialium, Ormosia (Leguminosae), Licania (Chrysobalanaceae), y Brosimum (Moraceae), en suelos profundos, ar- cillosos, amarillentos, en afloramientos de rocas calizas en altitudes entre los 100 a 150 m. Florece en abril y fructifica en septiembre. Nombres vernáculos. Chacaruite, guagene- cuil de la montana, vaina, jinicuil. Usos. Los frutos maduros son comestibles. Paratipos. MEXIC :0. OAXACA: Distr. epu os s Jia S del ae 2, 3- CRUZ: Mpio. ü 4 km S del ind "i la terraceria La rag na Boca del Monte en el cam al N del Poblado 2, Ӯ? Grimes 2817 (MO, NY, TEX). Mpio. Hidalgotitlán, ven а а Laguna, Avendano 40 (MEXU, XAL); km 0 2 del camino Plan de Arr icr ee ы. 1791 15'N, 94°40'0, таас et al. 2806 (NY, a km 4-5 caminito que cr la carrete Dorantes a La Lagun 3493 (MEXU, XAL): l kmal N ace Hermanos Cedillo, 1 7?16'N, 94?37'O, M. Vázquez et al. 385 (ENCB, NY, WIS, XAL) Inga ismaelis está emparentada con /. setosa Don e 1. pavoniana Don, siendo más cercana a /. setosa Don del Perú, de ésta difiere /. ismaelis por tener las glándulas del raquis foliar sésiles o casi sesiles, las flores mas grandes, las estipulas mucho más largas que anchas así como las brácteas alargadas en vez de cordatas. El largo de los lóbulos del cáliz sólo se equipara al de los de /. velutina Willd. del Amazonas. Esta especie se dedica a Juan Ismael Calzada (1950-), colector de las selvas de México, quien trabajó para el proyecto Flora de Veracruz, cuyo personal es el que fundamentalmente ha colectado a esta especie. Inga jinicuil Schldl., Linnaea 12: 559-560. 1838. TIPO: Costa Rica: Jiménez & Züniga 815 (holotipo, CR; isotipo, MO). Especie conocida de la vertiente Pacifica, San José, Costa Rica, y del Cerro Pierre en el Darien en Panamá. Sus relaciones еп Мезоатепса ра- difiriendo de ella en varios caracteres; tambien muestra, aunque no recen ser con /nga calderonii, relaciones con la sección Vul- – ап cercanamente, ртае Benth. De este ültimo grupo Inga jimenezii difiere fundamentalmente por los lóbulos cortos del caliz. Material re eprese ntativo. COSTA RICA. SAN JOSE ménez & Zuniga 815 (MO). PANAMA. DARIEN: Gentry 1612 (BM). Inga jinicuil Sehldl., Linnaea 12: 559-560. 1838. TIPO: México. Veracruz: Schiede 675 (isoti- pos, CAS, G, GH, Inga jinicuil, a pesar de disponerse a partir del trabajo de León (1966) con nuevas y numerosas colectas, se mantiene aparte, sin que pueda in- cluirse con /. paterno Harms; sus diferencias son tanto en caracteres vegetativos, florales, como en los habitats. /nga jinicuil se encuentra en un piso altitudinal mayor que el де /. paterno (véase clave de especies). Material representativo. | MEXICO. CHIAPAS: Marti- 22146 (MEXU). MICHOACAN: vore 36 (F). Mo- relos: Cordero C. 2 (MEXU). ОАХАСА: Rivera R 91 (MEXU). PUEBLA: Miranda G. 1435 (MEXU). VERACRUZ: Pes ada 1969 (MEXU). GUATEMALA. BAJA VERAPAZ: Lun- dell & Contreras ?0885 (LL). nez 5. Inga lacustris M. Sousa, sp. nov. TIPO: Mexico. Veracruz: costa N de la Punta Levisa, en la Laguna de Sontecomapan, Mpio. Catemaco, 1832'N, 95°11'0, 15 abr. 1973, F. Мепеп- 248 Annals of the Missouri Botanical Garden dez L. 144 (holotipo, MEXU; isotipos, MEXU, MO, XAL). Figura 5 Arbores parvae; rami juveniles de tomentosi cito glabrescentes. Folka 2-jugata; stipu ongae, Mops uda subulatae; foliola ex lanceolata ‘aliptioa apice alat ta dun sessilibus о Inflorescentiae racemosae floribus dis ibus; bracteae sensim heteromorphae. Flo- res pedicellati; alabastra floralia att obtusa, calyce ap l bees laevis, marginatus; corolla tub- 1 centi-sericea; tubus stam egumen immaturum) parvum, complanatum, oblongum, sparse hispidulurr Arboles 3-7 m; anguladas, esparcidamente tomentosas cuando јо- ramas teretes a tenuemente venes, pronto glabras y rojizo brillante. Hojas 4-folioladas; estipulas 9-12 mm, triangular-subu- ladas, persistentes; peciolo 0.7-2.2 cm, terete a alado; foliolos tenuemente discoloros, cartáceos a subcoriáceos, la base simétrica cuneada a ligera- mente cordata, el ápice acuminado a cuspidado, el haz brillante a opaco, glabro, el enves algo brillante a opaco, glabrescente, la nervadura primaria eglan- dular; par basal de foliolos 3.5-9.5 1.8- 4(-6) cm, lanceolados a elipticos, el par apical (8-)18-23(-27.5) x (3-)5-6.5(-8) cm, elipticos; raquis foliar 2.5-5 cm, marginando a alado, el ala 5-9 mm de ancho, oblanceolada; glándulas inter- foliolares sésiles, pateniformes, el cuerpo glandular, ca. 1 X 1 mm; apéndice 4 mm, setiforme, pronto caduco. Inflorescencias en racimos solitarios; pe- dünculo 4.5-9 cm, estriado, esparcidamente to- mentoso a glabrescente; raquis floral 5-7.5 cm, las flores esparcidas, particularmente las proxi- males, las brácteas gradualmente heteromorfas, 1— 4 mm, las proximales cóncavas ovado-lanceoladas, las distales estipitado-cuculadas, en ocasiones con una estéril del 4 al ápice del pedúnculo, persis- tentes. Flores proximales pediceladas, el pedicelo esbelto hasta 3.5 mm, las distales casi sésiles; bo- tones florales con el cáliz abierto y los lóbulos imbricados, atenuado-obtusos; cáliz 3-9 mm, tu- bular, liso, esparcidamente tomentoso, sericeo a glabrescente, con escotaduras, los lóbulos ca. 1 mm, triangular-cuspidados; corola 13-15 mm, tu- bular, blanca, esparcidamente canescente sericea; tubo estaminal largamente exerto, blanco. Leg- umbre (inmadura) 4.5- 2 x 0.2 cm, aplanda, derecha, oblonga, truncada en la base, apiculada en el ápice, esparcidamente hispidula, las valvas aplanadas, las suturas onduladas, mar- ginadas. Distribución y hábitat. Sólo conocida de la orilla N de la Laguna de Sontecomapan, en la región volcánica de Los Tuxtlas en Veracruz, México, en una selva perennifolia que está delimitada por manglares de Laguncularia racemosa, sobre sue- los arenosos; al nivel del mar. Esta especie está muy restringida en área y se limita a unos cuantos árboles. Florece en abril y junio, inicia la fructifi- cación en abril y aún tiene frutos inmaduros a fines de julio. MEXICO. VERACRUZ: Mpio. Catemaco, Punta Levisa, Lag pis e Sontecomapan, /. Calzada 987 (MEXU), s.n. (153 r. 1973] (MEXU), 1805 (MEXU), рин D 12004 (MEXU, MO, XAL); Ran- ho Levisa, en Pants Levisa, orilla N Laguna de Sonte- ERAN M. Sousa « 1. Cal: ada 13021 (MEXU, MO). Paratipos. Inga lacustris esta emparentada con /. cookii Pittier; sin embargo, difiere de ella, particularmente en su pelosidad, estipulas y glándulas interfolio- lares, asi como brácteas florales y pelosidad de la legumbre. El epiteto se emplea para hacer enfasis en el tan particular hábitat, de una especie en in- minente peligro de extinción. Inga latibracteata Harms, Repert. Spec. Nov. Regni Veg. 19: 64. 1923. про: México. Ve- racruz: Pringle 8159 (isotipos, Е, MEXU, US) Inga borealis T. S. Elias, Phytologia 14: 208. 1967. IPO: México. San Luis Potosi: King 4275 (holotipo, NY; isotipos, F, MICH, TEX, US). Para demás sinónimos, excepto /. zacuapanica Harms, véase León ( Especie muy variable que se caracteriza por tener flores con una bráctea subyacente muy bien desarrollada que llega a proporciones foliáceas en Inga borealis que representa el extremo de va- riación de este carácter, asi como el extremo norte de distribución de la especie. El otro extremo de variación seria /. oerstediana Benth., de la cual I. latibracteata representa una variación geo- gráfica; sin embargo, creo deben mantenerse se- paradas por el momento. Existen varias formas intermedias entre esta especie e /. vera Willd. que interpretamos como hibridos; el nombre más an- tiguo que corresponde a este hibrido propuesto es I. xalapensis Benth. (véase a éste para mayores detalles). e НЕ ан MEXICO. CHIAPAS: Топ 5576 (MEXU). oaxaca: Torres C. & Ramirez 8461 (MEXU). PUEBLA: Ca ampos V. «e P А. 204 (MEXU). s Fernández N. 3891 (ENCB). SAN LUIS POTOSI: King 427. TEX). VERACRUZ: Velázquez L. 171 (MO). Inga latipes Pittier, Contr. U.S. Natl. Herb. 18: 183. 1916. TIPO: Costa Rica: Tonduz 13056 (holotipo, US; isotipos, BM, K, ra Esta especie es sólo conocida de las áreas mon- tanosas de la Provincia de Cartago en Costa Rica. Volume 80, Number 1 Sousa S. 249 1993 El Género Inga FIGURA 5. Inga lacustris M. Sousa.—A. Rama con inflorescencia. — B. Detalle de estipulas.—C. Glándula interfoliolar. — D. Botón floral. — E. Flor. — Е. Cáliz. —G. Согоја. — H. Tubo estaminal.—I. Gineceo. F. Menéndez L. 144. —J. Fruto. (Tomado de Sousa 13021.) A pesar de estar muy cerca de /. punctata Willd., y florales (flores pediceladas y el cáliz casi glabro) no creo que se pueda sinonimizar con ella, como сото en el fruto (mucho más grande). Por su cáliz sugirió pero no hizo León (1966). Inga latipes casi glabro y su fruto de grandes dimensiones se difiere de |. punctata tanto en caracteres vege- parece más a Í. leiocalycina Benth., pero 1. latipes tativos (glándulas interfoliolares muy promimentes) tiene hojas de menor tamaño, 6-folioladas, las glán- 250 Annals of the Missouri Botanical Garden dulas interfoliolares más grandes, además de las flores pediceladas. Material representativo. COSTA RICA. CARTAGO: Lent 2111 (F) Inga laurina (Sw.) Willd., Sp. Pl. ed. 4, 4: 1018. 1806. Basiónimo: Mimosa laurina Sw., Prodr. 85. 1788. TIPO: St. Kitts: Masson s.n. (ho- lotipo, BM, no localizado). Inga Mast alae que ex Benth., Trans. Linn. Soc. 1875 С Basin: Mimosa fagi- rs L., Sp. J 510, : 753. non Inga fagifolia I Don, Gen. Hist. 2: 391. 2. [= I. ruiziana Don A pesar de que el basiónimo más antiguo de esta especie es el de Linneo no se le puede emplear como lo hizo León (1966), ya que Willdenow (véase descripción de Inga semialata) no formalizó el cambio a Inga, siendo hasta Bentham (1875) que lo hizo, pero antes G. Don (1832) lo empleó para otro taxon. Sobre si el material de las Antillas difiere sig- nificativamente de él del continente americano, León (1966) da una buena discusión del problema y concuerdo con él, ya que en términos generales la diferencia se reduce a un sólo carácter: el tubo estaminal exerto o inserto. MEXICO. CHIAPAS: Sousa et al. 12908 MEXU ‹ GUERRERO: Soto N. et al. 5180 EXU). JALISCO: Santana & Guzmán 3330 (WIS). NAYARIT: Flores F. et al 919 (MEXU). OAXACA: Torres C Martínez S. 4817 (MEXU). GUATEMALA. ALTA А CHIQUIMULILLA: Clarke 19918. . ESCUINTLA: Hayes s.n. (GH). RETALHULEU: Standley 87254 (F). HONDURAS. LEMPIRA: Molina R. 24147 (F). EL SALVADOR. AHUACHAPAN: Stan- dley & Padilla B. 2836 (GH). ТА PAz: Flores 283 МО SAN VICENTE: Standley 21744 (F). NICARAGUA. ZELAYA Little 25397 (F). CosrA RICA. GUANACASTE: Herrera 1460 MO). HEREDIA: Hartshorn 1325 (MO). PANAMA. COCLE: СА 2045 (МО). HERRERA: Allen 4037 (MO). Los TOS: Croat 9753 2 PANAMA: пече 817 (ВМ). VERAGUAS: Tyson 60 CANAL: Croat : У Valeur 716 (F). у . Sr. THOMAS: pipe & 6 (F). TORTOLA: Britton & Shafer 716 (F). 7 (F Marble 1226 А NICA: Stehle 5690 IVAR: . et al. 7228 (MO). PASTAZA: Palacios Neill 627 (MO). BOLIVAR: Little 67 37 (MO). PERU. D Кой 11688 (MO). (CENTRO OESTE) DISTRITO FEDERAL: Irwin & Soderstrom 6168 (MO). сога5: Heringer 14770 (MO). (SUDESTE) RIO DE JANEIRO: Pereira & Duarte 4498 (MO). (SUR) PARANA: Hatschbach 43201 (MO). Ак- GENTINA. CORRIENTES: Schinini & Vanni 15506 (MO). Inga leiocalycina Benth., London J. Bot. 4: 598. 1845. TIPO: Guyana: Schomburgk 829 (lec- totipo, designado por Poncy (1985), P: iso- lectotipos, BM, F, G, K Inga yunckeri Standley, Field Mus. Nat. Hist., Bot. Ser. 9: 296. 1940. TIPO: Honduras: Yuncker et al. 8805 (holotipo, F; isotipos, BM, С, K, MO, US). Especie hasta hace poco sólo conocida de Brasil y las Guayanas (Poncy, 1985), pero en realidad se extiende hasta el sur de México. En Meso- américa ya tenia un nombre: /nga yunckeri Stand- ley como endémica de Honduras (León, 1966). Inga leiocalycina es en general una especie bas- tante ћоторепеа, la cual debido a su similitud con I. punctata Willd. (véase ésta) se le ha identificado como tal (véanse también comentarios de Poncy, 1985). Poncy (1985) sinonimiza bajo esta especie a Inga strigillosa Spruce ex Benth. (Trans. Linn. Soc. London 30: 612. 1875) e 1. dumosa Benth. (loc. cit., 30: 612. 1875), lo cual no me parece justificable, ya que estas dos presentan los cálices pelosos que se parecen más a /. punctata Willd. Inga leiocalycina es un elemento amazonico de selvas altas perennifolias que se extiende en las areas bajas cálido hamedas de Norte y Sudamérica. Material representativo. | MEXICO. OAXACA: Wendt 4424 (MEXU). VERACRUZ: Wendt et al. 377 1 (MEXU). ATEM EL PETEN: Contreras 6008 (LL). Marshall et al. 445 (NY). HONDURAS. ATLANTIDA: Yunck- er et al. 8805 (MO). CORTES: García 50 DOE 53 (MO). NICARAGUA. RIO SAN 25549 (MO). ZELAYA: Standley 19361 (F). Costa Rica. HEREDIA: Hartshorn 1 А 22813 (МО). Sanders 14717 (MO). COLON: Nee 72 ). SAN BLAS McPherson 12733 (MO). COLOMBIA. VALLE: Hilty J-39 (US). VENEZUELA. AMAZONAS: Lister olc 2504 Marcano-Berti et al. 38- 1-77 (G). : Hitchcock 17158 (US). Teu. Prance et aL 55726 (MO). GUAYANA FRANCESA: Poncy 133 (MO). PERU. AMAZONAS: 7906 (MO Кауар 277 (MO). LORETO: К. Vásquez et al. 7 ) PASCO: D. Smith 1905 (MO). BOLIVIA. PANDO: Prance et al. 6470 (NY). BRASIL. (NORTE) AMAZONIA: Prance et al. 13707 (F). PARA: Fróes 32210 (US). 0 (M (CENTRO OESTE) MATO GROSSO: Berg et al. P18440 (MO). Inga leonis Zamora, Brenesia 33: 111. 199]. TIPO: Costa Rica: Lent 3541 (holotipo, CR; isotipos, К, MEXU) Esta especie está en el complejo de /nga vil- losissima Benth., I. dasycarpa M. Sousa, Г. te- Volume 80, Number 1 1993 EI Género /nga nella M. Sousa, 1. chiapensis Miranda ex M. Sousa, e 1. sierrae Britton et Killip. Es más cercana a /. chiapensis, de la cual difiere en sus hojas, en sus frutos, asi como el color de la corola. Material representativo. COSTA RICA. ALAJUELA: Lent 3541 (MEXU). HEREDIA: Utley & Utley 2825 (F). LIMON: Hammel et al. 17061 (MO). PANAMA. BOCAS DEL TORO soho 9855 (MO). cocLE: Sytsma 1883 (NY). CHI- 1: McPherson 12830 (MO). Inga litoralis Zamora, Brenesia 33: 115. 1991. TIPO: Costa Rica: Burger et al. 10572 (holo- tipo, CR; isotipos, BM, F, MEXU, NY). Esta especie, sólo conocida del tipo, está cer- canamente emparentada a /nga tenella M. Sousa, de la cual difiere por la pelosidad del cáliz, el pedünculo floral, y la forma de las brácteas, además de que desconocemos su fruto. Material representativo. COSTA RICA. PUNTARENAS: Burger et al. 10572 (MEXU) Inga ed Standley, Publ. Field Mus. Nat. , Bot. Ser. 18: 497. 1937. TIPO: Costa NER Brenes 6371 (holotipo, F). Especie poco conocida que hasta hace poco se colectó en Рапата y Colombia. ial representativo. COSTA RICA. ALAJUELA Shute 3705 (A). PUNTARENAS: Koptur & о sK. 7 (MO). SAN JOSE: Utley 5752 (MEXU). PAN BOCAS DEL TORO: McPherson 8408 (MO). COLOMBIA. VALLE DE CAUCA: Monsalve B. 1730 (MEXU). Inga micheliana Harms, Repert. Spec. Nov. Regni Veg. 13: 525. 1915. TIPO: Guatemala: Heyde & Lux 3319 (isotipos, GH, K, MO, US). Esta especie tiene como parientes cercanos por un lado а Inga flexuosa Schldl. y por otro a /. hintonii Sandw. (véanse discusiones de éstas). n /nga micheliana el raquis foliar es más frecuentemente terete, pero en ocasiones se pre- senta alado, como lo es en el tipo. Material representativo. MEXICO. CHIAPAS: Miran- da 6. 7116 (MEXU). GUATEMALA. CHIQUIMULA: Hughes 1461 (MEXU). QUICHE: Heyde & Lux 3319 (GH) Inga mortoniana Léon, Ann. Missouri Bot. Gard. 351, fig. 6. 1966. про: Costa Rica: А. Smith 1230 (holotipo, NY; isotipos, F, US). Esta especie está muy relacionada a /nga ac- rocephala Steudel, particularmente con la variante de inflorescencias capituliformes, pero con flores más grandes: cáliz 5-7 mm de largo en /. mor- toniana (versus 1.5-3 mm en /. acrocephala) y corola (7-)9-12 mm de largo (versus 6-9 mm), la corola más grande de menos de 2 veces el largo del cáliz (versus. la corola de 2.2-3.5 veces el largo que el cáliz); /. mortoniana tiende tambien a tener un cáliz glabrescente а esparcidamente canescente sericeo. Ambas especies caracteristi- camente tienen campos glandulares en los lóbulos del cáliz, que en cálices maduros adquieren un aspecto calloso. En fruto ambas especies, hasta donde puedo ver, son indiferenciables, y de hecho, Inga mortoniana podria pasar a una categoria infraespecifica de /. acrocephala, pero debido a las antedichas diferencias florales creo prudente mantenerlas aparte. En general /nga mortoniana habita en un piso altitudinal mayor que /. acro- cephala. El material de Venezuela (Steyermark 91992) tiene la corola más pequena pero el cáliz grande y glabrescente. terial representativo. COSTA RICA. ALAJUELA: Skutch 3780 (MO). PUNTARENAS: Hariahorn 1790 (NY). VENEZUELA. DISTRITO FEDERAL: Steyermark 91992 (NY). ign s mucuna Walp. et Duchass., Ann. Bot. Syst. 52. про: Panamá: Duchassaing "T Келна: GH). Coincidimos con León (1966) en que /nga mu- сипа esta estrechamente emparentada a /. fastuo- sa (Jacq.) Willd., pero a pesar de las nuevas co- lectas hechas posteriormente al trabajo de León (1966), estos taxa, aün se sostienen aparte. Por otro lado /nga vera Griseb. ex Benth., que León (1966) apuntó como especie endémica de Trinidad relacionada con /. mucuna, se puede considerar sinónimo de /. fastuosa. Asi tenemos /. fastuosa de Trinidad, Venezuela, y las Guayanas e 1. mu- cuna de Panamá y Colombia. Material representativo. | PANAMA. COLON: Nee 9101 (MO). DARIEN: Kennedy 2899 (MO). PANAMA: Duchas- saing 81 (GH). SAN BLAS: Dwyer 6647 (MO). ZONA DEL CANAL: Croat 12234 (MO), COLOMBIA. ANTIOQUIA: Brand & Cogollo 214 (MO). cHoco: Juncosa 2508 (MO). Inga multijuga Benth., Trans. Linn. Soc. London . 1875. TIPO: Panama: Sutton Hayes 645 (lectotipo, designado aqui, K, el colector y colecta lo dió Pittier en 1929, pero no el herbario; isolectotipos, BM, BR, С, P). Bajo este nombre se ha aglutinado desde la de- scripción original, una amplia variación que el mis- mo Bentham (1875) reconoció; posteriormente Pit- 252 Annals of the Missouri Botanical Garden tier (1916) segrego id aestuariorum Pittier, la cual León (1966) volvió a unir pero que aqui con- sidero aparte (véase discusión де 1. aestuariorum). También se ha determinado bajo /. multijuga, ma- terial recientemente colectado que al estudiarlo lo estoy describiendo como /. cuspidata M. Sousa (véase ésta). Material representativo. COSTA RICA. LIMON: Ко- bles 1889 (MO). PUNTARENAS: Khan et al. 501 (BM). PANAMA. COLON: Holdridge 6512 (MO). DARIEN: Gar- wood et al. 20484 (BM). SAN BLAS: Lewis et al. 194 (MO). zoNA DEL CANAL: Croat 5403 (MO). VENEZUELA. TACHIRA: Liesner et al. 9621 (MO) — Inga oerstediana Benth., Bot. Herald 117. 1853. TIPO: Panamá: Seemann 1688 (lectotipo, de- signado aqui, BM, el colector lo dió Pittier en 1929, pero no el herbario ni la colecta). Inga endlicheri (Kuntze) J. P Macbr., Field Mus. Nat. B 943. pre І. fasci- culata Poeppig, Nov. ae Sp. Pl. 3: Willd. 1806. riPo: Реги: Poeppig a (holotipo, ist., Bot. Se Inga minutula (Schery) T. S. Elias, Phytologia 14: 211. 1967. Basionimo: /. edulis C, Martius var. Pd Schery, Ann. Missouri Bot. Gard. 37: 209, fi 1950. TIPO: Panamá: Allen 968 (holotipo, MO; isoti- pos, F, US). Para demás sinónimos véase León (1966). Inga oerstediana es una especie muy variable, fundamentalmente de las montanas mesoameri- canas, donde frecuentemente es empleada como sombra de café. Sin embargo existen poblaciones de baja altitud en las que en ocasiones varia la pelosidad de sus flores hacia tonos cinéreos en vez de ferrugineos. Esta variación ha sido elevada a especie (/. minutula (Schery) T. S. Elias), lo que no parece justificable mantener; podria consider- arse como variedad, pero bajo /. oerstediana en vez de /. edulis C. Martius. noreste montarioso de su distribución, adquiere una Este complejo, a bráctea de grandes proporciones y asi tenemos /. latibracteata Harms, especie que mantenemos por a combinación de caracteres que presenta (véase discusión de /. latibracteata). En Sudamérica /. oerstediana se parece mucho, particularmente en sus flores y frutos, a /. scabriuscula Benth. de las Guayanas, el Territorio Delta Amacuro de Vene- zuela, y la Amazonia de Brasil, pero difiere por las hojas (la forma de los foliolos, nervaduras, y pe- losidad). A su vez 1. scabriuscula es similar a /. edulis (véase discusión bajo /. edulis Las especies de este grupo de /nga de la sección Inga con glándulas interfoliolares prominentes for- man una serie interconectada teniendo en un ex- tremo /. edulis y en otro |. latibracteata, pero creo que deben mantenerse separadas para su me- jor entendimiento. Material representativo. | MEXICO. CHIAPAS: Sousa et al. 13179 (MEXU). GUERRERO: Soto N. 5226 (MEXU). OAXACA: Breedlove 25129 (MO). TABASCO: Hernández X. & ене 239 (MEXU). vERACRUZ: Dorantes et al. 2579 (MO). BELICE. STANN CREEK: Gentle 3020 (A). GUATEMALA. ALTA VERAPAZ: Standley 89677 (F). BAJA VERAPAZ: Harmon & Dwyer 3151 (MO). CHIMAL TENANGO: Standley 64478 (F). ESCUINTLA: Standley 63873 (F). QUEZALTENANGO: Standley 87910 (F). SA- CATEPEQUEZ: Standley 63336 (F). SUCHITEPEQUES: Hughes et al. 1381 HONDURAS. ATLANTIDA: Yuncker 4922 (MO). COLON: Bangham 250 (F). COMAYAGUA: Mo- lina R. 8067 (F). CORTES: Allen nd (F). EL PARAISO: O. Williams - Molina R. 11993 (MO). MORAZAN: Rodr riquez M. MO). OLANCHO: а Mejía 163 (MO). YORO: Hagen & Hagen 1117 (F). EL SaLv у APAN: Standley & Padilla B. 2733 (F). Carlson 206 (F). SAN SALVADOR: Calderón 1564 (GH). SANTA ANA: L. O. Williams et al. 15168 (F). NICARAGUA CARAZO: Neill 246 (МО). CHONTALES: Stevens 22601 (MO). GRANADA: Sandino & McKearin 4 16 (MO). ESTELI: Sousa et al. 12993 (MEXU). JINOTEGA: Sandino 82 (MO). MATAGALPA: Vincelli 336 (MO). RIO SAN JUAN: Riviere yee: ZELAYA: Stevens 12335 (MO). Costa RICA. ALA : Brenes 5495 (CR). CARTAGO: Hammel & А ит gros (MO). HEREDIA: Hartshorn 943 (MO). LIMON: Gómez P. et al. 22763 (MO). PUNTARENAS: Lies- ner 2208 (MO). SAN JOSE: Barringer & Christenson 3286 F). PANAMA. COCLE: Lao 282 (MQ). COLON: McPherson 7930 Мо, cHIRIQUI: Huft 1845 (MO). DARIEN: Stern MO). PANAMA: Dwyer 1778 (MO). SAN BLAS: lar 380 (MO). ZONA DEL CANAL: Allen 1972 (MO). COLOMBIA. ANTIOQUIA: Uribe-Uribe 2075 (US). CHOCO: Gentry & Fallen 17712 (MO). MAGDALENA: White & Alverson 653 (MO). NORTE DE SANTANDER: i asas et al. 12215 (MO). VALLE: Devia 257 (МО). EZULA. CARACHE: Benítez de Rojas 1929 (MO). LARA: "Davidse & González 20963 (MO). MERIDA: Breteler 3215 (M MIRANDA: Gonzdlez pee 1360 ). ECUADOR. EL ORO: Steyermark 539. )). ESMERALDAS: Litte & on 21074 (MO). a AMAZONAS: а 442 (MO). CUZCO: Nunez et al. 8249 (MO). M DE DIOS: Gentry 43696 (MO). PASCO: Smith et al. eae » (MO) Inga paterno Harms, Repert. Spec. Nov. Regni Veg. 13: 419. 1914. про: Guatemala. Santa Rosa: Barberena, 1,000 m, July 1893, Heyde & Lux 3280 (lectotipo, redesignado aqui, US; ). isolectotipos, С, Inga е Pittier, Contr. . Natl. Herb. 1916. TIPO: México. пе Cook 805 bin US). Harms (1914) al describir а Inga paterno по designo holotipo y la basó en varias colectas, sin- tipos; de ellos León (1966) escogió como tipo a Preuss 1387, ejemplar que ya para esas fechas Volume 80, Number 1 1993 Sousa S. 253 EI Género /nga habia sido destruido en B, y en vista que no se han encontrado duplicados en otros herbarios, consi- dero la lectotipificación de León fue inapropiada y designo aqui el lectotipo en base a un isosintipo existente, en vista que los otros sintipos de B tam- bien fueron destruidos. Especie muy afin a /nga cinnamomea Spruce ex Benth. de la Amazonia, pero lo suficientemente diferenciada para considerarla en una sola entidad con Í. cinnamomea. Se cultiva ampliamente en México y Centroamé- rica por lo apetecido de la pulpa de su fruta, la cual es muy popular desde épocas precolombinas que se vende ampliamente en mercados muy distantes a su área de cultivo, también se usa grandemente como sombra de café y cacao. Asi sabemos que se ha introducido en la Peninsula de Yucatán, y también en Nicaragua donde se le cono- ce como “guava extranjera," mente en Costa Rica. y muy probable- Material representativo. | MEXICO. CHIAPAS: Martí- nez S. et al. 20718 (MEXU). GUERRERO: Gonz ález L. & Castaneda 143 (MEXU). OAXACA: 37 : Basurto & Duran Md (MEXU). ores 10177 (XAL). TABASCO: Cowan 1991 (MEXU). VERACRUZ: Ibarra M. 21 56 (MEXU). GUATEMALA. EL PETEN: Aguilar 377 (MO). EL PROGRESO: Steyermark 43746 (F). GUATEMALA: Harmon 201 1 (GH). IZABAL: Тип Ortiz 2436 (MO). QUEZALTENANGO: Ste dicii 33398 (F). QUICHE: Heyde & Lux 3309 (US). ТА ROSA: Heyde & Lux 3280 (US). HONDURAS. AT- LANTIDA: o 35 (МЕХ О). COMAYAGUA: Standley Chacón P. 66 F). EL PARAISO: L. O. Williams & Molina R. 9032 а LEMPIRA: Hughes & A ea 1429 (MEXU). oer Molina R. 1062. EQUE: Molina R. 24156 (F). OLANCHO: с Y (MO). EL SALVADOR. SAN SALVADOR: Standley 21756 (GH). SANTA ANA: Allen & Armour 6815 (F). NICARAGUA LEON: Grijalva et al. 4184 (MO). MANAGUA: Araquistain 3522 (MO). MATAGALPA: Sousa et al. 12944 (MEXU). ZELAYA: Laguna 145 (MO). Costa RICA. ALAJUELA: Smith 6490 (GH). CARTAGO: R. Torres 97 (F). HEREDIA: Hammel 11571 (МО). LIMON: Pennington & Zamora 13344 (K). SAN JOSE: siepe 1481 (M Inga pauciflora Duchass. et Walp., Linnaea 23: 746. 1850. про: Panamá: Duchassaing s.n. (holotipo, B; isotipo, GH). Inga pauciflora Duchass. et Walp., endémica de Panamá, está muy estrechamente relacionada a 1. meissneriana Miq. de las Guayanas y noreste de Brasil; en fruto no es posible diferenciarlas y en flor son similares, excepto que la flor de /. pauciflora es totalmente sésil mientras que en /. meissneriana las flores proximales muestran pedi- celos. Afortunadamente /. meissneriana fue publi- cada un ano después que /. pauciflora, por lo que dejamos pendiente la decisión de unirlas Material representativo. PANAMA. CHIRIQUI: Croat 26800 (MO). DARIEN: Stern et al. 968 (MO). VERAGUAS: Allen 4417 (MO). ZONA DEL CANAL: Croat 8756 (MO). Inga pavoniana Don, Gen. Hist. 2: 388. 1832. TIPO: Peru: Ruíz & Pavon s.n. (holotipo, OXF herb Lamb.). Inga еа Benth., London J. Bot. 4: 608. 1845. xico. Tabasco: Linden 726 (holotipo, K; tipos, С (2 ej.)). Inga donacana Y F Macbr., Field ra Nat. Hist., Bot. - 13: 1943. TIPO: Perú. San Martin: Тако Spruce 4503 (holotipo, K; E NY, OXF). Para demás sinónimos véase León (1966) bajo Inga sapindoides Willd Jsamos el nombre de С. Don en vez de /nga sapindoides Willd., que fue él que escogió León (1966), уа que /. sapindoides se basa en un ejem- plar estéril y fragmentado de Venezuela (Brede- meyer 4, microficha IDC. 7440. 19024 MO ex B-W) que no muestra ni la inserción de la hoja con el tallo ni las estipulas. Además, en la descrip- ción de Willdenow (Sp. Pl. ed. 4, 4: 1012. 1806), la corola es blanca y glabra, mientras que, en la especie generalmente referida como /. sapindo- ides, la corola es verde pálido y canescente vellosa, siendo blancos los estambre. Bentham (1875) ya habia discutido este problema agregando que Willd- enow nunca vio las flores, sino que Bredemeyer le dio el dato. Por lo anterior empleamos a /. pavon- iana Don, basada en un ejemplar en flor del herb. Lambert, colectado por Ruiz y Pavón, del cual no ay duda de su identidad, y dejamos a /. sapin- doides como un nombre de aplicación dudosa. Material representativo. | MEXICO. CHIAPAS: Sousa et al. i (MEXU). GUERRERO: Soto N. 7897 (MEXU). OAXACA: Torres C. 9441 (MEXU). PUEBLA: Basurto & Durán 400 (MEXU). TABASCO: Sousa et al. 11740 MEXU). VERACRUZ: Wendt 3254 (MEXU). BELICE. EL CAYO: Gentle 8663 (LL). STANN CREEK: Dwyer et al. 557 (MO). TOLEDO: Davidse & Brant 32424 (MO). edd LA. ALTA VERAPAZ: Contreras 4650 (MEXU). EL PETEN: C ontre ras 345 7 (MO). QUEZALTENANGO: Croat x Налага TacQueen & (MEXU). SAN MARCOS: Dwyer ihe 4(M Heyde & Lux 6095 (MO). NDURAS. ATLANTIDA ^g 155 (MO). COLON: а 1062 (M Nelson et al. 2864 (MO). CHOLUTECA: E. Repulski ! us (TEFH). LEPIRA: Hughes & Pips i pou 1430 (MEXU). MoRAZAN: Molina 390 (GH). OLANCHO: Blackmore & bur 2069 (BM). Er, SALVADOR. 5 ANA: Villacort » espe 27. Standley 21803 (GH). Nica UA. CHIN va е Grijalv a e (MO). CHONTALES: ve 2831 6 (MO). GRANADA: Moreno 4088 (MO). JINOTEGA: Araquistain & Castro rd (МО). MANAGUA: ея 45 34 (MO). Ma- TAG : Moreno 23594 (MO). NUEVA SEGOVIA: Atwood et eral 6864 (MO). RIO SAN JUAN: Araquistain 3255 (MO). AYA: Pipoly 4476 (MO). Costa RICA. ALAJUELA: Utley 254 Annals of the Missouri Botanical Garden & Utley 3986 (MO). canTAGO: Davidse & Pohl 1493 MO). HEREDIA: Neill 5117 (MO). LIMON: Herrera & Martínez 2229 (MO). PUNTARENAS: Liesner 1939 (MO). AMA. BOCAS DEL TORO: von Wedel 672 (MO). CHIRIQUI: Liesner 447 (MO). DARIEN: Hammel 1326 (MO). PANAMA: Croat 34431 (M · TOBAGO: о 4355 . EcU О: МО). PERU. SAN MARTIN: Spruce 4503 (С - LIVIA. BENI: 7. D. fees ы et al. 13493 (K). CHILE. TARAPACA: R. S. Shepard 300 (GH). Inga pezizifera Benth., London J. Bot. 4: 587. 1845. TIPO: Guyana: Schomburgk 124 (holo- tipo, K; isotipo, BM) Inga microstachya ч et Killip, Ann. New York Ас . Sci. 35: 115. 1936. TIPO: Colombia: Mutis 3633 (holotipo, D isotipos, G, K). Especie cercanamente relacionada a /nga rio- palenquensis A. H. Gentry (véase ésta), de la cual difiere tanto en caracteres florales como del fruto; ambas especies tienen una distribución que en gran medida se sobrepone. Material d Costa Rica. LIMON: Cha- cón 292 (MO). PANAMA. BOCAS DEL TORO: Gordon 8 (MO). COCLE: Knapp 37 32 (MO). coLoN: е 3675 (МО). ZONA DEL CANAL: Schmalzel 585 (МО). COLOMBIA. ANTIOQUIA: Soejarto 2912 (MO). CHOCO: León 514 (MO). VENEZUELA. AMAZONAS: Wurdack & Adderley 43298 (US). DELTA AMACURO: Marcano-Berti 272 (NY). YARACUY: Steyermark et al. 106750 (MO). Guyana: Forest Dept. British Guiana 3439 (NY). SURINAME: Stahel 230 (MO). GUAYANA FRANCESA: de Granville B5330 (MO). Ecua- DOR. ESMERALDAS: Little & Dixon 21080 (US). BRASIL. (NORTE) AMAZONIA: Ducke 1249 (MO). PARA: Ducke 16709 (US). RORAIMA: Prance et al. 10195 (MO). Inga pinetorum Pittier, Contr. U.S. Natl. Herb. 18: 185, pl. 90. 1916. про: Belice: Peck 343 (holotipo, GH; isotipo, K). Esta especie seria de esperarse que estuviera tambien en el norte de Guatemala, pero a pesar de las multiples colectas que se han hecho de esta región, aün no hay representación de /nga pine- torum. erial representativo. | MEXICO. CHIAPAS: Rama- moorthy A (MEXU). oAXACA: Chavelas P. & Pérez EXU). rABasco: Téllez V. & Martínez S. 910 (MEXU). M ee ndis Tenorio 713 (MO). BELICE. 1 (M AUGUSTINE: bos CAYO: Dwyer 11649 (MO). T CREEK: Ud de 36217 (GH). TOLEDO: Gentle 4149 Tu Inga polita Killip, Ann. New York Acad. Sci. 35: 92. 1936. про: Colombia: Lehmann 3824 (holotipo, K). Figura 6. Esta especie muestra claras relaciones con el complejo de /nga umbellifera, de la cual difiere por los pedünculos cortos, por las brácteas florales hasta 9 mm de largo, las cuales cubren totalmente a los botones florales, y por las hojas fundamen- talmente 2-folioladas. Inga polita varia en sus extremos de distribu- ción; asi en Costa Rica, en el Golfo Dulce, sus hojas en la misma rama van de 2-4-folioladas mientras que, en Colombia, los foliolos son de mayor tamano que el material mesoamericano. Material representativo. Costa RICA. PUNTARENAS: Jiménez & Poveda 446 (MO). PANAMA. COCLE: Davidse & ión 23683 (MO). PanaMa: McPherson 11895 MO). sa AS: de Nevers 4651 (MO). COLOMBIA. BOLI- VAR: Ge ns p Cuadros 57430 (MO). cauca: Monsalve B. 1652 (MO). vaLLE: Cuatrecasas 16574 (F). ~ Inga portobellensis Beurling, Kongl. Vetensk. 4. 122. 1856. про: Рапа- сад. Нап má: Billberg s.n. (holotipo, S). Especie endémica al área más ћитеда de Рапа- má, en los alrededores de Portobelo en la provincia de Colón y norte de la Zona del Canal. Pittier (1916) interpretó a esta especie con ma- terial de Golfo Dulce, en Costa Rica; Schery (1950) y León (1966) la han aceptado, aunque con dudas, ya que la descripción de Beurling (1856) no con- cuerda bien con el material de Costa Rica. Las diferencias entre ambas especies no sólo radican en el tamano de la flor, sino tambien en la forma del cáliz y pelosidad de la corola (véase /. bella M. Sousa). Material representativo. | PANAMA. COLON: Knapp & Schmalzel 3591 (MO). ZONA DEL CANAL: Gentry 3565 MO) ~ Inga pseudoinvolucrata M. Sousa, sp. nov. TIPO: ama. Panama: Gorgas Memorial Labs yel- low гам research camp, ““Campamento Qua- tro," 5-10 km NE of Altos de Pacora, ca. 600 m, on ridge top, 21-24 Nov. 1974, S. Mori « J. Kallunki 3406 (holotipo, MO). res e puc procerae; rami juveniles ferrugineo- hirsuti. Folia 3-5-jugata; stipulae 4-5 mm longae, trian- gulares; hes lanceolato-elliptica, ex ovata tenuiter obo- ta, apice acuminata; rhachis foliaris alata, glandulis ex subsessilibus longe stipitatis, claviformibus. Inflorescentiae capituliformes, solitariae; bracteae oblanceolatae, florum periphericorum interdum in pseudoinvolucrum coaduna- Volume 80, Number 1 Sousa S. 255 1993 El Género Inga , \ [ = t VE FIGURA 6. Inga polita Кур. — A. Rama con inflorescencia. — B. Detalle de estipulas. — C. Glándula interfoliolar. (Tomado de McPherson 11915.)—D. Inflorescencia en botón. (Tomado de С. de Nevers et al. 7062.) — Е. Flor.— F. Cáliz. —G. Corola. — H. Tubo estaminal. — I. Gineceo. —J. Legumbre. (Tomado de G. Mori et al. 6360.) tae; calyx tubulosus paullo inflatus, striatus, glaber; corolla tubuloso-infundibuliformis, fere glabra; tubus stamineus insertus. Legumen 18-21 cm longum, 1.8-2 cm latum, = . . lineari-oblongum, complanatum, hirsutum, valvari-ner- do jóvenes, posteriormente glabrescentes. Hojas o- um. 10-folioladas; estipulas 2-7 mm, liguladas, estria- Arboles 5-7(-18) m; ramas teretes, tomentosas a hirsútulas con tricomas pardo-amarillentos, cuan- со 256 Annals of the Missouri Botanical Garden das, persistentes; peciolo 0.5-1.2 cm, marginado a angostamente alado; foliolos discoloros, cartá- ceos, la base en ocasiones asimétrica, cuneada, rara vez obtusa, el ápice acuminado-mucronato, el haz opaco, glabro excepto sericeo a tomentoso con tricomas pardo-amarillentos sobre las nervaduras y los márgenes, el envés palido, esparcidamente tomentoso a hirsutulo, la nervadura primaria eglan- dular; par basal de foliolos 2.5-3.5 1.5-2.2 cm, lanceolados a ovados, el par apical 4.5-7 x cm, lanceolados elipticos en ocasiones obovados; raquis foliar 4-8 cm, alado, el ala 3-5 mm de ancho, angostamente eliptica; glándulas interfoliolares subsésiles a largamente estipitadas, clavuliformes, el cuerpo glandular 0.3-1 l mm; apéndice 4-8 mm, setiforme, pronto cad- uco. Inflorescencia capituliforme, solitaria; pedún- culo 2-3 cm, aplanado, sulcado, hirsútulo; raquis floral muy reducido, las bracteas algo heteromor- fas, las de la periferia 3 mm, frecuentemente un- iendose y formando un seudoinvolucro, cordatas a oblanceoladas, las centrales 3-4 mm, lanceoladas a oblanceoladas, persistentes. Flores sésiles o casi sésiles; botones florales con el cáliz cubriendo, se desconocen; cáliz 5-7 mm, tubular, algo inflado, estriado, glabro, sin escotaduras, los lóbulos 1-1.5 mm triangulares; corola 16-18 mm, tubular, blan- ca, glabra excepto pilosa con tricomas pardo-amar- illentos sobre los lóbulos; tubo estaminal inserto a cortamente exerto, blanco. Legumbre 17.5-21 x 1.8-2 x amente túrgida, derecha, linear, brevistipitada en 0.3 cm, más o menos aplanada a liger- la base, rostrada en el ápice, hispida con tricomas pardo-amarillentos, las valvas inconspicuamente nervadas transversalmente, las suturas margina- das, los márgenes de las suturas aplanados. Distribución y hábitat. Solo conocida de dos áreas en los alrededores de El Llano en la Prov. Panamá, Panamá, y en el Bajo Calima en Valle, Colombia. En selvas húmedas asociada a Catoblas- tus y Hedyosmum scaberrimum. En altitudes entre los 100 a 600 m. Florece de finales de septiembre a noviembre (con flores marchitas en marzo) y fructifica a partir de finales de marzo. tipos. re apa along El Llano—Carti— Tupile road, 12 mi. above Pan-American Hwy. rad 1177 (MO, TEX, їч WIS); EL anes Cart 19.1, 9°19'N, 78°15'W, С. de Nevers 1. a 5134 (MO). COLOMBIA. VALLE: c 2 x Concesión Pulpapel/Buenaventura, 3?55'N, Monsalve B. 1827 (MEXU, MO). Especie relacionada al grupo de /ngae con flores glabrescentes en las cuales el cáliz es amplio (in- flado) y no se aplica a la corola como en /nga bella, 1. cordistipula, 1. portobellensis, e 1. spira- lis; además, es muy interesante por la combinación de caracteres tales como, glándulas interfoliolares variando de subsésiles a largamente estipitadas y clavuliformes con las inflorescencias capituli- formes, combinación que sólo conocemos en /nga saffordiana Pittier, de la cual difiere grandemente. Un carácter muy distintivo de la nueva especie es su tendencia a formar un involucro incompleto en la periferia de capitulo y de cual toma su nombre especifico. Inga punctata Willd., Sp. Pl. ed. 4, 4: 1016. 1806. про: Venezuela: Bredemeyer 7 (ho- lotipo, B-W, foto MO). Inga VB MER Linnaea 12: 560. 1838. TIPO: Mex o. Vera : Schiede s.n. (isotipos, BM, F, G, Para demás sinónimos véanse Schery (1950) y León (1966). Con /nga punctata Willdenow creó un nuevo nombre para Mimosa fagifolia sensu Jacq. (Select. E Amer. Hist. 264, t. 164. 1763) non L. (Sp. 6. 1753); sin embargo, Jacquin claramente, en su publicación, se refiere a la M. fagifolia de Linneo (que corresponde a /nga laurina (Sw.) Willd.; véase esta), además de que todos los demas elementos concuerdan con ella. Por lo anterior, Willdenow interpreto la descripción y lamina де Jacquin como otra especie de la cual él tenia un ejemplar de Caracas, Venezuela, con la que ђазо su descripción. Inga punctata Willd. está basada en elementos heterogéneos, de los cuales, afortu- nadamente los subsecuentes autores a DeCandolle (Prodr. 2: 435 la especie con el material de Venezuela y no con el de La Martinica. Asi Mimosa fagifolia en la obra de Jacquin no debe ser interpretada como un partir de 5) han interpretado homónimo tardío, y por tanto no se trata de un nombre ilegitimo. El concepto de esta especie para Mesoamerica, ue muy bien interpretado, por Schery (1950) y León (1966); sin embargo, debido a sus similitudes, pero discernibles diferencias con /nga latipes Pit- tier (véase ésta y discusión de León, 1966 leiocalycina Benth. (véase ésta y discusión de Pon- cy, 1985) se les ha querido incluir como extremos de variación de /. punctata. En Sudamérica el problema es mas complejo; asi tenemos a /. stri- gillosa Spruce ex Benth., la cual Poncy (1985) sugiere que con más información deberia ser si- nonimizada con /. punctata pero, por lo pronto la coloca de sinónimo де /. leiocalycina Benth., tanto el Ed e beo BM, С, К (2 ej.)) como de ela Amazonia (Peru. Amazonas: Volume 80, Number 1 1993 Sousa S. El Género Inga Ferreyra 5037, MO. Bolivia. Pando: Sperling & King 6534, NY. Brasil (Norte). Amazonas: Berg et al. 197 14, NY; Магапћао: Daly et al. D.669, MO; Rondónia: Prance et al. 5467, NY). Aqui considero a /. strigillosa como una especie aparte, que se caracteriza por un raquis floral muy corto, paucifloro y las brácteas filiformes que sobrepasan a los botones florales, dando un aspecto de inflo- rescencia parecido a Dale En cuanto a los taxa infraespecíficos descritos para Inga punctata, no los estamos considerando aqui. Asi 1. punctata var. elongata J. F. Macbr. p Ecuador y Реги (holotipo, Killip & Smith 06, F), con flores cuya corola llega hasta 14 mm de largo y raquis foliar angostamente alado, debe corresponder a otra especie. Por otro lado la subespecie chagrensis Pittier (tipo, ЈУ“. R. Maxon 47 88; isotipo, BM) y la variedad panamensis Benth. (sintipos, Sutton Hayes 357, BM, K; Seemann 406, K (2 ej.)) no parecen diferentes al resto de I, punctata. Material representativo. | MEXICO. CHIAPAS: Ton 7238 (MEXU). HIDALGO: Tenorio 2515 (MEXU). ОАХАСА: Tor- res C. et al. 2517 (MEXU). PUEBLA: Basurto & Durán 12792 (MEXU). . BELICE. BELICE: Liesner & Dwyer 1487 (MO). EL e Dwyer 12724 (MO). STANN CREEK: Dwyer et al. 3 (MO). TOLEDO: Whitefoord 3191 (BM). GUATEMALA. ALTA VERAPAZ: Con- treras 7868 (LL). IZABAL: дан 234 (LL). EL PETEN: Lundell 18335 (MEXU). ALHULEU: Hughes et al. 1383 (MEXU). SAN MARCOS: "White 5274 (MO). SANTA 5А: Téllez | 619 wes bi HONDURAS. ATLANTIDA Standley 5 A). со Qe 598 (MO) COMAYAGUA: Edwards Р-4. 36 (а Blackmore & steed on (BM). CORTES: ај 133 Мо! R. & Molina 27906 (Е). MORAZAN: Tuc ker 761 (LL). MOSQUITIA: Gentry et al. 7508 (MO). CHO: Croat et al. 64451 (MO). EL SALVADOR. COMA- SAGUA: Calderón 1354 (GH). LA LIBERTAD: Rohweder 2943 (MO). SANTA ANA: Кишон & S. Martínez 273 (К). NICARAGUA. BOA LES: Мее 2834 (MO). MATAGALPA: Sousa et al. 12925 (MEXU). NUEVA SEGOVIA: Moreno et al. 24765 (MO). RIVAS: Sandino 938 (MO). zELAYA: Stevens 7221 (MO). Costa RICA. ALAJUE- LA: Taylor 37 18 (MO). GUANACASTE: Sousa et al. 12753 (MEXU). HEREDIA: D. Smith 1187 (MO). LIMON: Stevens et al. ae (MO). PUNTARENAS: Лә et al. 10573 (F). SAN JOSE: Tonduz 7020 ел TORO: "Mc Pherson 11627 (MO). с COLON: Hamilton & Stockwell "3699 . CHIRIQUI: Hampshire & Whitefoord 743 (BM). DARIEN: Stern et al. 548 (MO). Los santos: McPherson 13507 (MO). PANAMA: Tyson 6745 (MO). SAN BLAS: Gentry éd aes VERAGUAS: Antonio 2404 (MO). ZONA DEL CANAL: 6248 (MO). ToB ANTIOQUIA: Rentería et al. dern 4879 (MO). SANTANDER: god & Forero 15487 (МО); VALLE: Пела 260 (MO). VENEZUELA. MIRANDA: 6 (MO). j JINOTEGA: ba & Quezada 188 t AGO: ery) 3909 (F). CAE MO). CHOCO: von Snei- Croat 21768 (MO). TACHIRA: Steyermark et al. 120093 (MO). к Benítez de i 1879 (MO). Var UY: Rutkis 470 € Ecua . NAPO: Cerón 358 MO). PERU. argas C. 7318 (MO). B SANTA CRUZ: Steinbach 7396 (MO). Inga quaternata Poeppig, Nov. Gen. Sp. Pl. 3: 79. 1845. TIPO: Brasil: Poeppig 121 (holo- tipo, W; isotipo, F). Para sinónimos véase León (1966). Especie muy variable que León (1966) carac- terizó muy ien y seguimos su concepto. Sólo agregamos que en su extremo norte en Veracruz, México, la inflorescencia en ocasiones pasa de cap- ituliforme a una espiga corta corimbosa, dandole aspecto de /nga nobilis Willd. erial representativo. | MEXICO. CHIAPAS: Meave et al. B-65 1 (МЕХ О). VERACRUZ: = C. 992 (MEXU). 7 сали TOLEDO: Dwyer : Contreras 6209 dide Yuncker et al. 8337 (ВМ), СОРАМ: е TES: Nelson et pe 3072 (MO). NICARA- reno 23265 (MO). ZELAYA: Ortíz r & Judziewicz 14597 r« Bello 2668 (MO). LIMON: pe to PUNTARENAS: na 683 (MO). MA. BOCAS DEL TORO: Cooper & Slater 13 (Е). COLON: ue 1326 (MO). д Рант 5270 (09). DARIEN: : McPherson 10325 Little 13699 (Е). COLOMBIA. EL META: Killip 34459 (MO). Killip & Smith 15477 ( | MIRANDA: Pittier 6012 (US). N Cerón et al. 486 (MO). маРО: Neill & Manni O sTAZA: MacBryde 1517 (MO). cia pe ыа Wurdack 2217 (MO). cuzco: Wasshaus Encar- nación 753 (MO). LORETO: Croat 19389 (MO). MADRE E DIOS: Barbour 5737 (MO). ucAYALI: Gentry & Díaz 58596 (MO). BoLivia. BENI: Rusby 990 (MO). BRASIL. (NORTE) AMAZONIA: Cid & Souza 3005 (MO). Inga riopalenquensis A. H. Gentry, Selbyana 2: 39, pl. 12A. 1977. про: Ecuador: Dodson 5463 (holotipo, MO; isotipo, US). Especie cercana a /nga pezizifera Benth., de la cual difiere fundamentalmente por el tan par- ticular fruto lenoso con proyecciones en las valvas, la pelosidad de la corola, y el cáliz de mayor ta- mano. En el material de Costa Rica, sus legumbres son más verrugosas que las de Ecuador. Aqui se da una nueva distribución a una especie descrita recientemente como endémica del Chocó ecuatoriano (Rio Palenque). Gentry (1982) tabula las relaciones fitogeográficas de Rio Palenque con Мезоатепса y encuentra que el 18% de las es- 258 Annals of the Missouri Botanical Garden pecies tienen el patron de "América Central a Ecuador occidental (sin incluir el Amazonia)," lo que representa una alta proporción de una área que él considera natural. Material representativo. COSTA RICA. HEREDIA: Hartshorn 184 eee PUNTARENAS: U ley & Utley 4867 (MO). COLOMBIA. BOYACA: Lawrance 783 (MO). Ecua- DOR. LOS RIOS: ets 31009 (MO Inga rubiginosa (Rich.) DC., Prodr. 2: 434. 1825. Basiónimo: Mimosa rubiginosa Rich., Actes Soc. Hist. Nat. Paris 1: 113 TIPO: Guayana Francesa: Leblond s.n. (holo- tipo, G-DEL, foto TEX). Inga standleyana Pittier, Contr. U.S. Natl. Herb. 204. 1 TIPO: Panamá: Pittier 5496 [он US; isotipo, US). Dentro de la variabilidad de /nga rubiginosa, I. standleyana Pittier parece representar un ex- tremo con flores más robustas y en general un poco más cortas, con distribución en Panamá a Vene- zuela, mientras que /. rubiginosa sensu stricto es de las Guayanas y norte de Brasil. Sin embargo, a diferencia de León (1966), las unimos en base a su débil diferencia, proponiendo que posterior- mente podría formalizarse a nivel infraespecifico. Material оа PANAMA. DARIEN: Pittier 5496 (US). COLOMBIA. ANTIOQUIA: Brant & Escobar 1275 Men ). VENEZUELA. BOLIVAR: Blanco 450 (NY). URO: Wurdack & adir 39650 (NY). MERIDA: C. Smith 2909 (MO). Su У For. pues 6074 (MO). GUAYANA pone. 6121 (MO). . (NORTE) AMAPA: Pires | я m 50274 (MO). РАКА: ‘Maciel etal 468 (TEX). (NORESTE) MARANHAO: Fróes 1749 (MO) Inga ruiziana Don, Gen. Hist. 2: 391. 1832. TIPO: Perú: Ruíz & Pavón s.n. (lectotipo, designado aqui, OXF herb. Lamb., y flores; isolectotipos, BM, OXF) con hoja Inga. fagifolia Don, Gen. Hist. 2: 391. 1832. TIPO: Perú: z & Pavón s.n. (holotipo, OXF). Para demás sinónimos véase Schery (1950). León (1966: 301) cita como tipo de /nga rui- ziana Don a Ruíz 5, de la cual he visto un ejemplar en el Smithsonian (US); por otro lado Macbride (1943: 40) citó como tipo una colecta de Ruiz y Pavón, sin informar sobre el herbario ni alguna colecta determinada. Sin embargo, el material que George Don vió y citó en la descripción original fue el del Herbario de H. B. Lambert que fue comprado por Fielding, ahora Herbario Fielding en la Universidad de Oxford, donde existen tres ejem- plares de /. ruiziana colectados por Ruiz y Pavón: uno está anotado como “ Mimosa fagifolia sp. nov. del Perú” (se trata del tipo де /. fagifolia эе! у los otros dos como ** Mimosa sp. nov. del Per De estas dos se escogió el ejemplar con hoja y flores como lectotipo. Especie poco variable de la Amazonia que llega hasta las selvas de la vertiente Caribe de Nicaragua. Material representativo. NICARAGUA. CHONTALES: Stevens 2842 (MO). JINOTEGA: Moreno 946 (MO). МА- TAGALPA: Araquistain & Moreno 2700 (MO). RIO SAN JUAN: Stevens 4917 (MO). ZELAYA: кк 368 (МО). Costa RICA. ALAJUELA: Herrera 18 O). CARTAGO: Tonduz 13054 (GH). HEREDIA: Sc E ji Bockbrader 2 (MO). LIMON: Pittier 16646 (US). PUNTARENAS: e en 10454 (MO). SAN JOSE: Skutch & Barrantes 110 (GH). PANAMA. BOCAS DEL TORO: Blum 1346 (MO). COLON: Nee 9114 (MO). CHIRIQUE: Liesner 456 (MO). DARIEN: Hartman 12227 (MO). SAN BLAS: Mori et al. 6848 (MO). ZONA DEL CANAL: Dwyer 1455 (MO). COLOMBIA. CAUCA: Cuatrecasas 142 12 (US). choco: Duke 9892 (MO). VALLE: Cuatrecasas 16522 (US). ECUADOR. NAPO: Neill & Palacios 7120 (MO). PASTAZA: Cerón et al. 4356 (MO). PERU. AMAZONAS: Huashikat 1424 (MO). 1200: Núñez 6524 (MO). HUANUCO: Seidenschwarz 16211 (MO). LORETO: Mexia 6176 (MO). има: Medina 142 (MO). MADRE DE DIOS: Núñez 10592 (MO). PASCO: Gentry et al. 63235 (MO). SAN MARTIN: Ferreyra 7804 (MO). UCAYALE: Begazo 23 (MO). BRASIL. (NORTE) AMAZONIA: Riedel s.n. (K). Inga saff ordiana Pittier, Contr. U.S. Natl. Herb. 18: 176. 1916. TIPO: (holotipo, US). Panamá: Pittier 5676 De esta notable especie de /nga lo que más ha llamado la atención de los botanicos es su inflo- rescencia caulinar, carácter que ahora sabemos se presenta en al menos cinco especies de /nga, tres de las cuales se encuentran en Мезоатепса. Sin embargo, el carácter de esta especie que me parece ünico, es su cáliz, el cual es casi totalmente partido en cinco lóbulos lanceolado-aleznados, que persis- ten en el fruto Material representativo. | PANAMA. DARIEN: Hammel et al. 16128 (MO). COLOMBIA. ANTIOQUIA: Rentería et al. 3790 (MO). cuoco: Fernández 267 (US) Jribe, Mutisia 1: 5. 1952. 1903 (isotipo, Inga samanensis L. | TIPO: Colombia: Uribe-Uribe US) Inga i age Leon, Ann. Missouri Bot. G 53: 1966. TIPO: Costa Rica: León 4869 ca. DN Inga paludicola A. R. Molina, Ceiba 18: 101. 1974. TIPO: Nicaragua: Molina R. 2443 (holotipo, F). Especie muy caracteristica por tener legumbres con indumento escuamoso, carácter poco comün en las especies mesoamericanas. El material de Volume 80, Number 1 Sousa S. 259 EI Género /nga Colombia no difiere del de Мезоатепса, por lo que adoptamos el nombre más antiguo de Uribe. Material representativo. NICARAGUA. MATAGALPA: Sandino 2802 (MO). ZELAYA: yc 5086 (MO). Costa RICA. - ALAJUELA: Hartshorn 983 (F). HEREDIA: Hammel : Gómez P. et » 22768 (MO). PAN- . 487 (MO). coLoN: McPherson 8517 (MO). MM ANTIOQUIA: Cogollo 1274 ( Inga semialata (Vell. Conc.) C. Martius, Flora 20(2): Beibl. 111. 1837. Basiónimo: Mimosa semialata Vell. Conc., Fl. Flumin. 1: t. 5. 1831. TIPO: Velloso, Fl. Flumin. 11: t. 5. 1831. Figura 7. Inga eo quss Willd., nom illeg., Sp. Pl. ed. 4, 4: 1015. 1806. TIPO: Venezuela: Bredemeyer s.n. (holotipo, W, e Inga = T. 9. Elia 5, 377. 1966. TIPO: Pisa Tyson y Blum байр МО). . Missouri Bot. Gard. 53: 3803 Aqui estoy usando el nombre de Martius ya que él de Inga marginata Willd. es superfluo, porque Willdenow no habia hecho el cambio que corres- pondia de Mimosa fagifolia L. (Sp. Pl. 516. 1753) a [nga fagifolia, sino creó un nuevo nombre que resulta ilegitimo. La descripción de /. marginata illd., y el ejemplar citado (Bredemeyer s.n. de Caracas, Venezuela, B-W) corresponden a otra especie diferente a la de Linneo; además se cita como sinónimo a Mimosa bourgonii Aublet (Hist. Pl. Guiane 2: 941, t. 358. 1775), la cual corres- ponde a otra tercera especie. El nombre de Willde- now ha persistido en uso debido a que Bentham (1875) lo citó como, “Inga marginata Willd. excl. syn.," excluyendo los En la lista de sinónimos de /. marginata desde Bentham (1875), I. sapida Kunth (in Humb., Bonpl. et Kunth, Nov. Gen. Sp. 6: 286 a escoger; sin ambat; se tra basado en un ejemplar estéril de dificil determi- $4 5 а -- 5 9 sinonimos. . 1823) es el seri más antiguo a de un nombre nación, y la descripción del fruto es muy general ("legumina qunique-aut sex pollicaria subarcuata”) para permitir su identificación a nivel de especie; además como el mismo Kunth aclaró, él no vio el fruto sino la información se la proporcionó Hum- boldt, asi 1. sapida queda incertae sedis. Aqui también incluyo como sinónimo a Inga tysonii T. S. Elias, la cual parece no sostenerse como especie aparte debido a la gran variabilidad de 1. semialata. Los caracteres diagnósticos usados por Elias (1966) son la inflorescencia congesta y el pedünculo corto. Respecto a la gran variación de esta especie es necesario apuntar que la población del Cerro Jefe, en Panamá, entre los 800 a 1,800 m de altitud ha sido identificada como /nga fagifolia (L.) Willd. ex Benth. [/. laurina (Sw.) Willd.], debido quizá a que sus raquis foliares son teretes a submargi- nados. Lo interesante de esta población del Cerro Jefe es que fácilmente se diferencia de las del resto de Mesoamérica: asi de ellas difiere por tener es- tipulas más cortas, el raquis foliar terete a sub- marginado, las láminas foliares coriáceas, la corola más larga, esparcidamente sericea, y el fruto más ancho con las valvas verrugoso-lenticeladas; pero si se revisa el material de esta extensa especie se ve que en Sudamérica cada uno de los caracteres de la población del Cerro Jefe aparece. Así, por ejemplo, las poblaciones del Pert (Апсиазћ 263, MO) cuentan con los frutos tan o más anchos. Por estas razones parece que la llegada de 1. semialata a Mesoamérica ocurrió al menos en dos ocasiones y que la población de Cerro Jefe es una muestra de caracteres diferentes a la de las áreas mas bajas de Costa Rica y Panamá. En el limite norte de esta especie (Chiapas y Veracruz), existen otras poblaciones que muestran el raquis foliar terete a submarginado, pero con otras combinaciones de caracteres a las de Cerro Jefe, incluyendo: foliolos de mayor tamano, car- táceos а subcoriaceos, y un fruto grande, pero con las valvas aplanadas; este ültimo carácter no ha sido reportado para la especie. Esta ültima variante aparentemente (desconocemos el fruto) está tam- bién representada en Costa Rica (Cartago, Lent 1122, GH; Guanacaste, Herrera 753, MO) y Pa- namá (Chiriqui, McPherson 9886, MO), pero es más escasa que la forma de raquis foliar alado. La ilustración que aqui se muestra (Fig. 7) cor- responde a la población del Cerro Jefe. Material representativo. MEXICO. CHIAPAS: Ton 4947 (MEXU). VERACRUZ: Sinaca C. 1024 (MEXU). Costa RICA. ALAJUELA: Haber et al. 1817 (MO). CARTAGO: Utley & Utley 5063 (MO). GUANACASTE: Herrera 753 (MO). HEREDIA: D. Smith 354 (MO). LIMON: Grayum et al. 7 (MO). PUNTARENAS: O. Jiménez & R. Züniga ds (MO). SAN JOSE: Skutch 4288 (GH). PANAMA. BOCAS DE : von Wedel 1807 (MO). COCLE: Hammel 2651 (МО). COLON: Nee 9078 (MO). CHIRIQUI: Croat 22570 DARIEN: McPherson 12257 (MO). Los SANTOS: Mc- dijon. 9244 (MO). VERAGUAS: Croat 34174 (MO). ZONA DEL CANAL: Blum «& Tyson 1997 (MO). de AMAZONAS: Galeano et al. 1009 (MO). ATLANTICO: Rom ro C. i — CHOCO: sd & Fallen 17744 (MO). VENEZUE ; mark & Carreno E. 107587 (MO). ZULIA: Bunte d 11086 (MO). Guyana: Tutin 460 (US). SURINAME: B. W. 3223 (MO). 7 (MO). LORETO: Croat 20245 (MO). MADRE DE DIOS: Terborgh 6206 (MO). PASCO: 260 Annals of the Missouri Botanical Garden BO ТОЛЩ ODER m D „ RO ES 0) "y « Ly: = > H Mill Inga semialata (Vell Conc.) C. Martius. —A. Rama con inflorescencia. —B. yis ЈЕ ен Botón floral y brácteas. (Tomado de orrea et al 550.)— Fic : (Tomado de McPherson 7443 огоја. — С. Tub áliz.— McPherson 10623.) Ge n et al. 42055 (MO). PUNO: Núñez & Muñoz 5268 N MARTIN: ed 2669 (MO). UCAYALI: Begazo Solomon 44522 (MO). 1 (MO). BOLIVIA. AZ: Gentry & (MO). PANDO Sperling & King 6532 (MO). SANTA CRUZ: Neill et al. 9292 (MO). BRASIL. (NORTE) AMAZONIA: Kru koff 6437 (MO). PARA: Silva & Souza 2227 (NY). RONDO- — С. А о езшш, — Н. Gineceo. (Tomado de McPherson 7443.)—I. Fruto. alo dé NIA: Prance et al. 6050 (NY). (NORESTE) BAHIA: Lus- DE JANEIRO: Duarte MO). SAO PAULO: Gentry & Silva 58732 (MO). (SUR) PARANA: Hatschbach 445 10 (MO). SANTA CATARINA: Conrad & Dietrich 2098 Volume 80, Number 1 1993 Sousa S. 261 EI Género /nga (MO). ARGENTINA. MISIONES: Eskuche 379 (MO). PARANA: Brunner et al. 895 (MO). oblación de Cerro e quilla 2186 (MO), Correa A. et al. 550 ( (MO), Folsom & Harp 1359 (MEXU), 2007 (MEXU), McPherson 7443 (MO), 9796 (MO), 10632C (MO), 12333 (MO Inga sertulifera DC., Prodr. 2: 436. 1825. TIPO: Guayana Francesa: Stoupy s.n. (holotipo, P-JU 14523 [IDC. microficha 6206. 1067: III. 3 )). Mimosa coriacea Pers., Syn. Pl. 2: 262. 1806. [nga iacea (Pers.) Desv., J. Bot. Ag a 3: 71. 1814, non Humb. et Bonpl. ex Willd. 18 Para demás sinónimos véase Poncy (1985). Esta especie ha sido identificada en Mesoameri- ca como /nga heterophylla Willd., y por ello no se habia reportado para el área. Es una especie amazonica bastante variable cuyos limites aun по me están claros y por ello no estoy considerando las variedades descritas. En nuestra área los raquis foliares son teretes. Material representativo. COSTA RICA. ALAJUELA: Molina R. et al. 17747 (MO). GUANACASTE: Hammel 17501 (MO). LIMON: Davidse & Herrera 31315 (MO PUNTARENAS: Sánchez & Poveda 1277 (MO). PANAMA. COCLE: Antonio 3569 (MO). DARIEN: Gentry & Mori 14035 (MO). HERRERA: McPherson 10934 (MO). PANA- MA: Nee et al. 8784 (MO). ZONA DEL CANAL: Gentry 1897 (WIS). COLOMBIA. CHOCO: Forero & Jaramillo 4333 (MO). VALLE: Cuatrecasas 17422 (US). VENEZUELA. AMAZONAS: Level 76 (MO). sucRE: Steyermark et al. 121563 (MO). GUYANA: de la Cruz 2311 (US). SURINAM: Amshoff 2024 (MO). GUAYANA FRANCESA: Wachenheim 299 (MO). ECUADOR. NAPO: Palacios 1856 (MO). PASTAZA: Holguer 4655 (MO). s AMAZONAS: Tunqui 837 (MO). LoRETO: Rimachi Y. 713 5: Nünez 5469 (MO). BOLIVIA. LA PAZ: (NORTE) AMAPA: Irwin et al. 47498 (F). "e Vieira et al. 255 . PARA: Prance et al. P25414 (MO). RONDONIA: /Vee 34810 (F). (CENTRO OESTE) MATO GROSSO: Berg et al. P19865 (MO). — Inga sierrae Britton et Killip, Ann. New York Acad. Sci. 35: 123. 1936. про: Colombia: Archer 1366 (holotipo, US). Inga бгепези poer i Publ. Field Mus. Nat. Hist., Bot. Ser. 18: 4 . TIPO: Costa Rica: Brenes 5121 (holotipo, D Esta especie es muy caracteristica por sus hojas cocleadas, sus нши enon supernumera- a espiga muy corta; se ubica en el ия de н villosissima Benth., I. tenella M. Sousa, I. dasycarpa M. Sousa, e І. leonis Zamora. El material de Colombia y el de Мезоатепса no parecen diferir, рог lo que las unimos bajo el nombre más antiguo de /. sierrae. Material representativo. NICARAGUA. ZELAYA: Mo- lina R. 15086 (F). Costa RICA. ALAJUELA: Brenes 4988 (F). GUANACASTE: Haber & Bello C. 2864 (MO). PUNTARE- NAS: Téllez V. 4369 (MEXU). SAN JOSE: Jiménez M. 2620 (NY). PANAMA. CHIRIQUI: Him 297 (MO). COLOMBIA. ANTIOQUIA: Archer 1366 (US). HUILA: Little 9040 (NY). Inga sinacae M. Sousa et Ibarra-Manriquez, sp. nov. TIPO: México. Veracruz: Estación de Bio- logia Tropical Los Tuxtlas, Mpio. San Andrés Tuxtla, 18°34’—36'N, 95?04'-09'O, 4 mayo 1984, С. Ibarra, В. С. Gómez V. & S. Sinaca C. 1547 (holotipo, MEXU; isotipos, MEXU, MO). Figura 8. Arbores medianae, ramis juvenilibus brunneo-luteo-to- mentosis. Folia 2-4(-5)-jugata; stipulae 2.5-4 mm lon- gae, ovatae; foliola chartacea; rhachis foliaris ex teres Еи alata glandulis sessilibus urceolatis. Inflorescen- tiae racemosae; rachis floralis 2.5-5(-8) cm longa; brac- teae cras ex ovat sessiles; ETARA floralia calice aperto; calyx laevis, cinereo- e corolla tubulosa, e canescenti- villoso lanata. Legumen usque 35 cm longum, latum, incrassatum, n, complanatum oblongum, reu e lutescen- ti-hirs Arboles 6-22 m; ramas teretes, tomentosas con tricomas pardo-amarillentos. Hojas 4—8(— 1 0)-folio- ladas; estipulas 2.5-4 mm, ovadas, caducas; pe- ciolo 1.5-3.5(-7) ст, terete; foliolos discoloros, cartáceos, la base de simétrica a ligeramente asi- métrica, generalmente cuneada en ocasiones ob- tusa, el ápice obtuso a acuminado, el haz opaco, posteriormente brillante, esparcidamente canes- cente sericeo después glabro, el envés opaco, es- parcidamente canescente velutino, las nervaduras primaria y secundarias canescentes a tomentosas con tricomas amarillo palido, la nervadura primaria eglandular, el par basal de foliolos 6-9 x 3-4.5 cm, generalmente elipticos, en ocasiones ovados, el par apical 11-17 x 4.5-6 cm, lanceolados a elipticos; raquis foliar 4-12 cm, tomentoso con tricomas pardo-amarillentos, terete, marginado a angostamente alado principalmente entre los pares distales de foliolos, el ala hasta 6 mm de ancho, eliptica a oblanceolada; glándulas interfoliolares sésiles, urceoladas, el cuerpo glandular -] mm; apéndice muy reducido o ausente. In- florescencias en racimos solitarios; pedünculo 2- 4(-8) cm, terete, aparentemente liso, tomentoso con tricomas pardo-amarillentos; raquis floral 2.5— 5(-8) cm, las flores esparcidas particularmente las proximales, las brácteas homomorfas 1.5-2 mm, gruesas, ligeramente cóncavas, ovadas a suborbi- culares, caducas. Flores sésiles; botones florales con el cáliz abierto, obtusos; cáliz 9-13 mm, tubular, liso, cinéreo tomentoso, sin escotaduras, los lóbulos triangulares; corola 2.3-3.2 cm, tubular, blanco- 262 Annals of the Missouri Botanical Garden FIGURA 8. Inga sinacae M. Sousa et Ibarra-Manriquez. — А. Rama con inflorescencia. (Tomado de G. Ibarra M. 1547.) — B. Detalle de yema. (Tomado de S. Sinaca C. 184.) — C. Glándula interfoliolar. (Tomado de G. [barra M. 1547.) —D. Flor. — E. Cáliz. —F. Согоја. — G. Tubo estaminal. — H. Gineceo. (Tomado de R. Cedillo T. 3600.)— I. Legumbre. (Tomado de С. Ibarra & $. Sinaca C. 1836.) Volume 80, Number 1 Sousa S. 263 EI Género /nga verdosa, densamente vellosa a lanosa con tricomas canescentes; tubo estaminal inserto, blanco, los es- tambres erectos en el crepúsculo. Legumbre 13- 35 x (2.3-)3-5 x 0.8-1.5 cm, aplanada, derecha a curveada, oblonga, sésil en la base, apiculada a rostrada en el ápice, hirsuta con tricomas pardo- amarillentos, las valvas aplanadas, las suturas de- rechas, aplanadas a 1-sulcadas. Distribución y hábitat. Hasta ahora sólo co- nocida de la región volcánica de Los Tuxtlas en el sur de Veracruz, México, en el estrato medio de selvas altas perennifolias y en encinares. En alti- tudes entre fructifica de junio a agosto. m. Florece de abril a junio, Nombre vernáculo. Vaina peluda. Usos. Sombra de cafetos. Paratipos. MEXICO. VERACRUZ: Mpio, Catemaco, Rio (MEXU, MO), 3602 (MEXU, MO), 3624(MEXU), 3626 (MEXU, МО); ues San Andrés Tuxtla, Estación de Biologia Tropical “Тоз Tu . : (MEXU), С. eM M. 217 (MEXU), 1836 (MEXU, MO); Laguna Escondida, 3 km NO d id Biologia Tropical “Los Tuxtlas,” S. Sinaca C. 184 (М MO), 893 (MEXU, МО); sin Mpio., Arriba de San Bee: nando, 4. Gómez-Pompa 4861 (MEXU). Inga sinacae no parece tener un pariente tan cercano que permita comparación; sin embargo, tiene fuertes relaciones con la serie Longiflorae de Bentham. Con esta especie se ilustra muy bien lo variable del raquis foliar, el cual, aun en el mismo árbol, varia de terete a alado pasando por margi- nado. Esta especie se dedica a Santiago Sinaca Colin (1959-), colector nativo de la region de Los Tux- tlas, quien ha contribuido grandemente al cono- cimiento floristico de la Estación de Biologia Trop- ical Los Tuxtlas Veracruz, del Instituto de Biologia, UNAM Inga skutchii Standley, Vw Field Mus. Nat. Hist., Bot. Ser. 23: 1943. TIPO: Costa Rica: Skutch 4823 DR. F; isotipos, A, US Especie muy poco colectada que se relaciona estrechamente con Inga stenophylla Standley, pero esta última cuenta con una flor más pequeña. Material representativo. COSTA RICA. PUNTARENAS: Liesner 1989 (MO). SAN JOSE: Skutch 4823 (MO) Inga pe Меи Willd., Sp. Pl. ed. 4, 4: 1017. 1806. Basiónimo: Mimosa specta- bilis M. en Skr. Notis Selsk. 2: 219, von Rohr s.n. t. 10. 1792. TIPO: Colombia: (holotipo, C no localizado). El tipo de Vahl fue colectado en Santa Marta, Colombia, por J. von Rohr; sin embargo, en las microfichas del Herbario de Vahl en Copenhague (C), no se ha localizado, por lo que de no encon- trarse otro duplicado de von Rohr, se debe lecto- tipificar con la lámina de Va Especie muy caracteristica de la cual León le creó su serie propia: Spectabiles León (1966). Se trata de una especie que muestra cierta variabilidad sobre todo del raquis foliar que va de terete a francamente alado. Respecto a la variedad schimpfi Little basada en Inga schimpfii Harms (Repert. Sp. Nov. Regni Veg. 43: 112. 1938) del Ecuador y Peru, los isotipos en MO y US, asi como otros ejemplares, parecen sostener a esta variedad geográfica, fun- damentalmente discernible por caracteres de la in- florescencia y de las flores. Inga spectabilis var. spectabilis Material representativo. MEXICO. OAXACA: Wil- liams 9386 (F). TABASCO: Chavelas P. et al. ES-2197 (MEXU). NICARAGUA. RIO SAN JUAN: Moreno 23018 (MO). UE ШЕ PUNTARENAS: Davidse & Herrera Ch. 26. 301 (MO). SAN JOSE: Dayton 3114 (MO). PANAMA. BOCAS DEL TORO: Peterson & а 6665 (МО). COLON: Holdridge & Maasola 6331 (MO). DARIEN: Sytsma & D'Arcy 3471 (MO). CHIRIQUI: Allen 1944 (MO). HERRERA: Dw yer 2818 M (МО). Los SANTOS: Dwyer i270 (MO). SAN BLAS: de /Ve- vers 5976 (MO). ZONA DEL CANAL: Croat 15044 (MO). COLOMBIA. ANTIOQUIA: pem 1428 (US). cuoco: Forero S Jaramillo 5304 (MO). HUILA: Little 7924 (US). META: па et al. 1198 (US). VALLE: Alston 8638 (BM). i cand MERIDA: Ruiz T. 601 (MO). Inga spectabilis var. schimpfii (Harms) Little, Phytologia 19: 268. 1970. Material representativo. | ECUADOR. ESMERALDAS: Little & Dixon 21191 (US). Peru. HUANUCO: D. Smith 1301 (MO) Inga spiralis Liesner et D'Arcy, Ann. Missouri Bot. Gard. 75: 385, fig. 2. 1988. про: Рапа- та: Mori & Kallunki 5577 (isotipo, МО). Esta magna especie de /nga tiene su simil en /. bullata Benth. sensu stricto (sin incluir a /. cam- panulata Benth., como Bentham sugirió en 1876 y Taubert hizo en 1892), de la provincia de Rio de Janeiro en Brasil, pero la mantengo aparte, fundamentalmente porque ignoramos la morfologia del fruto de 1. bullata, además del enorme espacio 264 Annals of the Missouri Botanical Garden geográfico entre ambas entidades. También mues- tra afinidades con /. jaunechensis A. H. Gentry del Ecuador, pero difiere porque esta ültima cuenta con el peciolo alado, las flores en espigas cortas, y el fruto curvo, pero no lenoso y circinado-espira- lado. Material representativo. | PANAMA. PANAMA: Dress- ) ler 4325 (MO). SAN BLAS: Liesner 1314 (MO Inga dip em Standley, Publ. Field Mus. Nat. , Bot. Ser. 18: . 1937. TIPO: Costa Re. Brenes 6841 (olstigo; F; isotipo, CR). Inga жле ein jus Field Mus. Nat. Hist., А 223: . TIPO: Costa Rica: Tabon 1008 elano F: e. MO, US) Especie muy relacionada a /nga skutchii Stan- dley (véase ésta), también tiene grandes semejanzas con /. tenuipedunculata León, con la que se le ha confundido. Con esta ultima, /. stenophylla difiere por su corola en general más larga, el cáliz más equeno y más peloso, el mayor nümero en pro- medio de foliolos por hoja y el hecho que habita en un piso altitudinal superior. El ejemplar de Nicaragua está estéril, por lo que su identificación es tentativa. Material representativo. NICARAGUA. CHONTALES: Gentry et al. 43944 (MO). Созта RICA. ALAJUELA: Brenes 11574 (NY). PANAMA. COCLE: Knapp 1153 (MO). Inga tenella M. Sousa, sp. nov. TIPO: México. Chiapas: km 18 Colonia Cuauhtemoc, Mpio. La Trinitaria, 15 abr. 1985, Ашзћ S. Топ 8185 (holotipo, MEXU; isotipo, MO). Arbores iid o ramis juvenilibus dense ое tomentosis praeditae. Folis 3-4-jugata; stipulae -5 ; foliola ex өңе а acea; rhachis foliaris FEN glandulis e ssilibus longe stipitatis. urlorescentiae spiciformes; ics ulus 1.8-4 cm longus; rachis floralis brevis; bracteae ex ob- longae lineares. Flores sessiles; calyx 5-7 mm longus, tubularis-conicus, striatus; corolla 8-14 mm longa. L "umen usque ed 4 cm latum, ex complanatum subtetra- gonale, oblongum, ferrugineo-hirsutum. Arboles (3-)5-20 m; ramas subanguladas a ter- etes, densamente ferrugineo tomentosas cuando jóvenes, posteriormente glabrescentes. Hojas 6-8- folioladas; estipulas 2.5-5 mm, liguladas а subu- adas, caducas; peciolo 1-2(-4) cm, terete a marginado; foliolos concoloros, en ocasiones lig- eramente discoloros, membranáceos a cartáceos, la base simétrica a ligeramente asimétrica, obtusa a cuneada, el ápice agudo a acuminado, en oca- siones cuspidado, el haz ligeramente brillante, mod- eradamente a esparcidamente piloso, el envés opa- co, moderadamente piloso, la nervadura primaria eglandular; par basal de foliolos (2.3-)4.5-7 x 1.3-)2.1-3.3 ст, ovados a lanceolados, el par apical 7-14 x 3.1–6.5 cm, ovados a obovados, elipticos; raquis foliar 5.5-7.5(-9) ст, alado, el ala hasta 7 mm de ancho, eliptica a oblanceolada; glándulas interfoliolares sésiles a largamente esti- | el cuerpo glandular 0.5- ~ pitadas, clavuliformes, 0.8 x 0.5-0.8 cencias en espigas, 1.8-4 cm, ferrugineo tomentoso; raquis floral 1— 1.3 cm, las flores congestas, las brácteas homo- morfas, 1.5-5 mm, concavas, oblongo-lineares a oblanceoladas, caducas. Flores sésiles; botones flor- tubular- mm; apéndice ausente. Inflores- 1-2 fasciculadas; pedünculo ales con el cáliz abierto; cáliz 5-7 mm, cónico, estriado, esparcidamente estrigoso a to- mentoso, con escotaduras, los lóbulos 1.5-2 mm, triangulares; corola 8-14 mm, infundibuliforme, blanca, canescente vellosa; tubo estaminal inserto, blanco. Legumbre 7-14 x 3-4 x 0.8-1.5(-1.8) cm, aplanada a ocasionalmente subtetragonal, de- recha a subespiralada, oblonga, sésil, apiculada en el ápice, ferrugineo hirsuta, las valvas aplanadas, las suturas más o menos derechas, marginadas, los margenes de las suturas aplanados а |-sulcados. Distribución y hábitat. De Chiapas, México a Comayagua, Honduras, en áreas montanosas, en bosques caducifolios con Liquidambar (Hama- melidaceae), bosques de neblina y pinares. En al- titudes entre 1,170 y 2, m. Florece en marzo y abril e inicia la ,800 m, en ocasiones hasta fructificacion en mayo y junio, los frutos persis- tiendo en al suelo hasta abril del siguiente ano. Nombre vernáculo. | Sakil kok (Chiapas); co- sós, cosoz (Honduras). Paratipos. | MEXICO. CHIAPAS: Mpio. La Trinitaria, 10 km ENE of Dos Lagos above Santa Elena, D. E. Breedlove 56196 (MEXU). GUATEMALA. ALTA VERAPAZ: along Rio C. chá, between Cobán and San deos Carchá, Stanley 89779 (A). aa VERAPAZ: SE of Porabi, У inas, L. O. к. et "E 434195 (F). YAGUA: s ravine pus El Achote, D. Hazlett 1563 (MO); a 18 km al N de Otoro camino La Esperanza - Rie nn [Siguatepeque], Martinez S S. & Té- llez V. 12816 (MEXU, ps | Cedral, Cordillera УДА a a Molina R. 7206 m Cascada El Chor- rito, 16 km al SO de ооо, Molina К. 10907 (NY) Quebrada El Rincón, between El Portillo and El Porvenir, 10 km of Sigua atepeque, d R. & Mo- lina 25431 (BM); in a wet ravine near El Achote, ii above the plains Е Siguatepeque, T. С. unc ker et с 6292 (СН ORTES: Cofradia, montana El Cusuco. D. Ruiz 218 aer. INTIBUCA: El Cedral · 20 km al SO de i eee сап a Jesús a Otoro, Molina 6153 (F, ; Ca laveras, Cordillera de Opalaca, Molina R. 6423 b 'GH, US); El Huise, entre km 10 y 13 Volume 80, Number 1 1993 Sousa S. 265 El Género Inga carretera a Marcala, Molina R. 13860 (NY, US); Los Banos de La Esperanza, Molina R. 13928 (NY), 25533 (F, US); La Quebrada, El Huise, 9 km al e La Es- peranza, Molina R. 14006 (NY, US); Cerro San Cris- tóbal, O de La Esperanza, D. Ruiz 112 (NY, TEFH). La PAZ: Quebrada El Cerrón, 8 km to Marcala, Molina R. & А. R. Molina 24244 (NY), Catarata La Chorrera, 10 km N de Belén Gualcho, C. Nelson et al. 3981 (MO). Inga tenella M. Sousa pertenece al complejo de especies relacionadas con /. villosissima Benth. e I. dasycarpa M. Sousa (véase discusión de ésta); tambien se integra, en parte, con /. cookii Pittier (véase esta). El epiteto se refiere a que dentro del grupo de especies a las que se relaciona, /. tenella es la que tiene la flor más chica, esbelta y frágil. Inga tenuipedunculata León, Ann. Missouri Bot. Gard. 53: 308, fig. 3. 1966. TIPO: Mexico. Veracruz: Ll. Williams 9029 (holotipo, NY; isotipo, F). Tanto en la clave como en la descripción de Inga tenuipedunculata León, la corola es citada como de 6–7 mm de largo; sin embargo, tanto еп el tipo como en otras colecciones vistas por mi, la corola es más pequena, de 4-6 mm de largo. Esto aunado a que en la clave León le da a la corola де /. barbourii Standley [= 1. stenophylla Stan- dley, véase ésta] como medida menos de 4 mm de largo, a pesar de que en el texto de la descripción él apunta como de 5-6 mm de largo, lo cual ha creado gran confusión en la determinación de estas especies. Inga tenuipedunculata es una especie de las selvas altas perennifolias del sur de México a Costa ica. Material i Manan te MEXICO. VERACRUZ: Do- rantes 5. al. 2956 (МЕХ 0). NICARAGUA. MATAGALPA: Neill 7258 (MO). Costa RICA. ALAJUELA: Herrera 1847 (MO) Inga thibaudiana DC., Prodr. 2: 434. 1825. TIPO: Guayana Francesa: Thibaud s.n. (holoti- po, Inga na е Verh. Bot. Vereins Prov. Bran e . 1907. TIPO: Perú: Ule 645 1 (holo- tipo, B, fot Inga rec ordii Britton et Rose in Standley, Trop. Woods : 5. 1929. TIPO: Belice: Record s.n. (holotipo, US No. 1209905). Una de las especies de /nga con mayor amplitud de distribución en selvas altas perennifolias. Aqui incluimos tambien a /. peltadenia Harms del Peru. Material б MEXICO. CHIAPAS: Martí- nez S. 18333 (MEXU). BELICE. 9197 (MEXU). TOLEDO: Dwyer 13023 (MO). GUATEMA- STANN CREEK: Gentle LA. ALTA VERAPAZ: Standley 90278 (F). IZABAL: Marshall et al. 391 (MO). HONDURAS. CORTES: Molina К. 1( (F). GRACIAS A DIOS: Nelson 849 (MO). OLANCHO: Black- re & Heath 1861 (BM). NICARAGUA. RIO SAN JUAN: Moreno 231284 (MO). ZELAYA: Stevens 12745 (MO). Costa RICA. HEREDIA: Hammel & Robles 16693 (MO). LIMON: Stevens et al. 24965 (MO). PUNTARENAS: Utley &U еу 4869 (MO). SAN JOSE: Skutch 4232 (A). PANAMA. ONA DEL CANAL: /Vee 9693 (MO). TRINIDAD: Schultes 18629 (US). COLOMBIA. AMAZONAS: Gillett 165 14 (US). ANTIOQUIA: Shepherd s.n. (MO). CHOCO: Gentry & Uribe-Uribe 861 (US). SANTANDER: Gentry 15391 (MO). VALLE: Cuatrecasas 14884 (US). bear pese Valverde et al. 1214 (MO). BOLIVAR: ен 46 (МО). cARABOBO: Pittier У ANDA: ко» мени & Ortega 1366 (МО). & Liesner 120680 (MO). TACHIRA: 14297 (MO). GUYANA: de la Cruz J ESMERALDAS: Pe, P. 302 APO: Palacios 2505 ЕРЕ ZONAS: Berlin 971 (MO). HUANUCO: Schunke V. 1508 (US). MADRE DE DIOS: Seibert 2166 (MO). LonETO: Croat 19788 eg MARTIN: Schunke V. 8145 (MO). BOLIVIA: PA g & King 6624 (US). BRASIL. (NORTE) ACRE: Nelson Lu AMAZONIA: Pachec o et al. 92 (MO). PARA: Sperling et a 20 DONIA: /Vee 34909 (NY) n ORAIMA: Prance et al. 9998 (NY). (NORDESTE) BAHIA: Salzmann s.n. (MO). PERNAMBUCO: Gentry et al. 50189 (MO). (SUDESTE) ES- PIRITO SANTO: Hatschbach 46783 ( MINAS GERAIS: Heringer & Eiten 15188 (US). сав 2 о Inga tonduzii J. D. Smith, Bot. Gaz. (Crawfords- ville) 44: 112. 1907. TIPO: Costa Rica: Топ- duz 12928 (holotipo, US; isotipos, BM, K). Inga tonduzii es cercana pero distinta de /. barbata Benth. de Brasil, la primera con brácteas, cáliz, y corola de mayor tamano. Esta especie hasta hace poco era sólo conocida de Costa Rica, pero el ejemplar estéril de Stevens 6886 de Nicaragua parece pertenecer a esta especie. terial representativo. NICARAGUA. ZELAYA: Ste- vens 6886 (MO). Costa RI ALAJUELA: Murphy 1363 (MO). CARTAGO: Lent 1022 (MO). HEREDIA: Chacón С. 986 (F). PUNTARENAS: Utley & Utley 4913 (MO). SAN JOSE: Burger & Stolze 5344 (NY). Inga umbellifera (M. Vahl) Steudel, Nomencl. Bot. 1: 431. 1821. Basionimo: Mimosa um- bellifera M. Vahl, Ecolog. Amer. 3: 30. 1807. TIPO: América austral, van Rohr s.n. (holotipo, C, foto MO). Inga de alae eal Poeppig, Nov. Gen. Sp. Pl. 3: 77, t 289. . TIPO: Peru: Poeppig 097 (holotipo, ). Inga gracilipes Standley, J. Wash. Acad. Sci. 15: 101. 266 Annals of the Missouri Botanical Garden 1925. про: Panamá: Standley 30353 (holotipo, Inga nd 'eana Britton et Killip, Phytologia 1: 193 0: Colombia: Lavranze 260 (hebtipo, l Te Ria T. MO). Aqui estoy considerando а Inga umbellifera en un sentido estricto sin incluir a /. sciadion Steudel, Г. umbratica Poeppig, ni a 1. brevipes Benth. (esta ültima no incluida en la sinonimia de León, 1966, ni por Poncy, 1985, para la Guayana Francesa), todas ellas con pedünculos breves y otras combi- naciones de caracteres que amplian el concepto de I. umbellifera a un punto de inmanejable; asi que su delimitación requiere que se estudie con más cuidado el material sudamericano. Material а Costa RICA. ALAJUELA: TRE 9052 (MO). HEREDIA: Schubert & palus и 1302 (GH). LIMON: Robles 2649 (MO). PUNTARENAS Gómez P. 22032 (MO). SAN JOSE: Skutch & Bartantes 5059 (GH). PANAMA. COLON: Croat 13178 (MO). DARIEN Knapp & Mallet 3056 (MO). PANAMA: Miller et al. 853 (MO). SAN BLAS: de Nevers & Marcus 3883 (MO). ZONA DEL CANAL: Weaver & Foster 1597 (GH). COLOMBIA. AMAZONAS: Schultes & e 16468 (US). ANTIOQUIA: Shepherd 773 (MO). ZUELA. BOLIVAR: Cardenas 870 (US) GUYANA: Robinson 85-0106 (US). SURINAME: Lanjouw & Lindeman 3019 (US). GUAYANA FRANCESA: ort & Pennington 17946 (MO). PERU. AMAZONAS: Knapp & Alcorn 7701 (MO). JUNIN: Stein & Todzia 2366 (MO). LORETO: ЛАДА Y. 3134 (MO). MADRE DE DIOS: E et dias 19629 (MO). SAN MARTIN: Klug 3692. UCAYALI: Gentry & e 58403 (MO). BRASIL. (NORTE) AMAZO a Hil. ^ al. 13043 (MO). PARA: Silva & Bahia 3575 (MO). Inga urabensis L. Uribe, Caldasia 4: 406. 1947. TIPO: Colombia: O. Haught 4716 (isotipo, K). Inga herr Ducke, Arch. Jard. Bot. Rio de Janeiro . 1925, non /. c me arpa (Jacq.) Willd., Sp. B el 4, 4: 1026 carpum (Jacq.) i ris E |. 810031 palo RB; Sr ч oto Inga enc , Brenesia 3 199]. TIPO: a Rica: ano 13081 (holotipo, CR). m dfi на BA sil: Duck Esta especie fue primero descrita por Ducke (1925) pero debido a que desafortunadamente usó de L. Uribe lo reemplaza. Aunque él no tuvo tal intención, y con- un homónimo tardio, el nombre secuentemente designó tipo. Su afinidad es muy estrecha con /nga venusta Standley, de la cual, no es fácil de separar; ésto es más critico en Colombia donde /. venusta ad- quiere foliolos de gran talla. En Mesoamérica las diferencias son un poco más claras, y aqui /. venus- ta se encuentra en un piso altitudinal mayor que Г. urabensis. Material representativo. COSTA RICA. HEREDIA: Hammel 11658 (DUKE). LIMON: Stevens 23736 (MO). PANAMA. BOCAS DEL TORO: McPherson & iie da 10159 (MO). SAN BLAS: Sugden 6 19 (MO). CoLom Hoyos & Hernández 705 (MO). BOYACA: Lawrence 779 (К). choco: Forero & Jaramillo 4571 ( PASTAZA: Palacios & Neill 700 (MO). nr (NORTE) PARA: Ducke RB10031 (B, foto, MO). Inga venusta Standley, Publ. Field Mus. Nat. , Bot. Ser. 18: . 1937. TIPO: Costa cu Brenes 6256 nd F; isotipo, CR). Inga venusta está ubicada en el complejo de especies de flores amarillas, elevado a serie por Leon (1966): serie Pilosulae basada en 1. pilosula (Rich.) J. F. Macbride; aqui agregamos la reciente- Cárdenas et G. De Martino. También muestra estrechas similitudes con /. urabensis L. Uribe (véase ésta). mente descrita /. neblinensis L. Material representativo. Costa RICA. Herrera 1608 5 GUANACASTE: Haber 5577 (MO). PANAMA. COCLE: McPherson 11263 (MO). COLO Mori ни 4895b (LL). CHIRIQUI: Mc dinis 9055 enis AGUAS: McPherson 12814 OMBIA. HOCO: Ge entry & Brand 36940 (MO). VALLE: C et al 48361 (MO). ALAJUELA: et al. 5577 T Inga vera Willd., Sp. Pl. ed. 4, 4: 1010. 1806. Mimosa inga L., Sp. Pl. 516. 1753, non. ell. Conc. /nga inga (L.) Britton, Fl. Ber- muda 170. 1918, nom. illeg. TIPO: Jamaica (lectotipo, Sloane, Voy. Jamaica 2: 58, t. 183, 1725, designado aqui). Para sinónimos véase León (1966) exceptuando a Inga eriocarpa Benth. e І. xalapensis Benth. Con /nga vera, Willdenow creó un nombre den- tro de /nga evitando el tautónomo que resultaria al usar el epiteto de Linneo, como lo hizo Britton posteriormente. e Mimosa inga L. se ha reconocido desde Bentham (1845) que Linneo incluyó bajo este nombre a todas las especies relacionadas [serie Inga] que él conocía, incluyendo tres citas: Ray (Hist. Pl. 2: 1762. 1688), Sloane (Cat. Pl. Jamaica 153. 1696; Voy. Jamaica 2: 58, t. 1 1725). Willdenow (1806) es el primero que la interpretó como lo que nosotros identificamos como Inga edulis C. Martius con material (P-H y B herb., B- W., IDC microficha no. 7440 microfoto 19020) y especie. Posteriormente Kunth (1824) adoptó esta determinación, la cual Bentham (1875) reconoció. El elemento de Linneo que podria caer bajo esta interpretación es la descripción de Ray (1688) por tratarse de una /nga del Brasil; sin embargo, en la descripción de la legumbre madura (la flor es descripción que corresponden a dicha desconocida) ésta es "compresa" y de 6.8-13.6 Volume 80, Number 1 1993 Sousa S. 267 El Género Inga cm de largo, caracteres que no corresponden a Inga edulis. Por otro lado posteriormente Bentham (1875) estrechó los limites de Mimosa inga L. y de hecho pero no de forma la lectotipificó, apun- tando que la “forma tipica" es de Jamaica; asi se reduce el problema a la segunda cita de Sloane. Linneo también citó al Hort. Cliff. 209, pero según F. Barrie (com. pers.) no existen ejemplares de herbario. Asi, aqui se lectotipifica a /. vera con la lámina de Sloane, a pesar de que existen dos frutos en el herbario de Sloane, material del cual Linneo no tuvo acceso. Inga vera es una especie muy polimorfa, que, muy problablemente agrupe a varias microespe- cies, que se han cruzado y ahora forman un com- plejo de híbridos, de los cuales no es posible delimi- tar taxa ni siquiera a nivel infraespecifico. Sin embargo, los extremos de variación son reconoci- bles y en no pocos casos se presentan en las mismas áreas geográficas, haciendo dificil su separación en variedades geográficas o subespecies como León (1966) hizo. Asi a guisa de ejemplo la muy caracte- ristica variante de /nga fissicalyx Pittier, de cáliz grande con lóbulos largos y atenuados y corola cortamente exerta, es conocida principalmente del área de la vertiente del Golfo de México, de Ve- racruz al norte de Guatemala (area де /. vera mn spuria), pero aparece de nuevo (aunque con brác- teas no tan largas) en el corazón de /nga vera subsp. eriocarpa (Benth.) León en Temascaltepec, éxico (Hinton 5572, F) y en el área de I. vera Willd. subsp. vera en la Republica Domini- cana (Mejía & Zanoni 9062, MO) Es muy posible que el hombre haya ayudado о. de grandemente a romper barreras entre las variantes originales, en vista que hoy en dia se trata de un complejo con claras tendencias hacia “árboles ma- leza" hábitats alterados. de amplia distribución, particularmente en La variación es muy notable en las flores asi como las brácteas subyacentes, tanto en forma como en dimensiones y pelosidad; Es otro lado las legumbres son bastante homogenea Respecto a Inga eriocarpa ba (holotipo, Coulter s.n., K) que León (1966) pasó a subespecie de 1. vera Willd., el material más caracteristico representativo (MEXICO. Colima: Magallanes et al. 2815, MEXU; Jalisco: Mexia 1842, BM, MO; Edo. México: Hinton 5911, F; Michoacán: Emrick 22, F; Morelos: Muskus 20, Е; Nayarit: Flores |. 1953, MEXU) se parece más a /. oerstediana Benth. que a /. vera, tanto en hojas y glándulas interfoliolares como en pelosidad, pero con flores más robustas. Es muy posible que /. eriocarpa sea un híbrido entre /. oerstediana e 1. vera. Ya Ben- tham (1875) consideró que /. eriocarpa podria ser la misma especie que /. xalapensis Benth. Sin embargo, /nga eriocarpa parece ser una buena especie que ha desarrollado una caracteristica den- sa pelosidad lanuginosa en su cáliz y corola, dife- rente a la de sus postulados padres especificos, pero además /nga eriocarpa se está cruzando con I. vera produciendo todas las gradaciones tanto de glándulas interfoliolares como de tamanos de flores y formas de cáliz; un ejemplo de ello se muestra en los ejemplares del tipo de /. oophylla Riley (tipo, González Ortega 250, holotipo, K; isotipo, 5) = С, Sobre la presencia де /. eriocarpa en Мезоате- rica existen colecciones de Chiapas (Goldman 858, US; Sousa et al. 11541, MEXU) que tienen la pelosidad de /. eriocarpa, pero los demás carac- teres se inclinan fuertemente hacia /. vera. Re- conociendo que son posibles hibridos en donde do- minan los caracteres de /. vera, por lo que no la estamos considerando para nuestra área. nga vera agrupa una gran variación no min por la cruza de sus variantes, sino tambien por cruzarse con /. oerstediana e I. latibracteata (véase Inga X xalapensis Benth.), aumentando en- ormemente su variabilidad. Con Inga affinis DC. (véase ésta) las similitudes tambien son grandes y sus diferencias son más bien de promedios que de calidad; sin embargo por el momento las mantengo separadas. Material representativo. | MEXICO. CAMPECHE: Flores 10297 (XAL). CHIAPAS: Martínez S. et al. 20162(MEXU). COAHUILA: Palmer 290 (GH). DURANGO: H. Gentry 5317 e GUERRERO: Martínez S. & Morales eee (MEXU). HIDALGO: Tenorio L. & Romero de T. 2517 (MEXU). JALISCO: - Bullock 1582 (MEXU). EDO. MEXICO: Hinton e (F). MICHOACAN: Soto N. et al. 8589 (MEXU). NAYARIT: Flores F. et al. 958 (MEXU). peus Sousa et us 6268 (MEXU). QUERETARO: López Ch. 306 (IEB). QUINTANA ROO: Sousa 12346 (MEXU). SAN LUIS POTOSI: Seler 765 (GH). SINALOA: Tenorio omero de T. 13698 (MEXU). TABASCO: Sousa et al. 11740 (MEXU). TAMAULIPAS: Valiente et al. 145 (MEXU). VERACRUZ: dede 1917 (F). BELICE. AUGUSTINE: Hunt 465 (MO). : Liesner Ф Dwyer 1650 (MO) EL CAYO: Ba lick et eg 1921 (MO). ( et al. 14480 (MEXU). 3023 (MO). CHIQUIMULA: Téllez V. & Martínez 5. 8879 (MEXU). EL PETEN: Tún Ortiz 1750 (F). ESCUINTLA: Ton- UEZALTENANGO: | D. Smith 2822 (US). RETALHULEU: Standley 88784 (MO). SACATEPEQUEZ: dia as 348 (US SANTA ROSA: Standley 79177 (F). SOLO ковање 5883 (US) SUCHITIPEQUEZ: Ste ноћ 47803 (Е). ZACAPA: Steyermark 42136 (F). HONDURAS. ATLANTIDA: Yuncker et al. 8333 on COLON: _ s 879 о м YAGUA: Duenas 5 О). TES: /Velson et al. 879 (MO). CHOLUTECA: Espinal o 196 (MO). GRACIAS рт у рат 268 Annals of the Missouri Botanical Garden : DIOS: Gentry et al. 7504 (MO). EL PARAISO: Martinez . & Téllez V. 12813 (MEXU). MORAZAN: Pineda 167 d OCOTEPEQUE: Nelson et al. 1494 (MO). OLANCHO: Sok ez 170 (MO). SANTA BARBARA: Padilla 34 (MO). EL . AHUACHAPAN: Croat 42167 (MO). CABANAS: ponens 1587 (US). LA LIBERTAD: Standley 23236 (GH). LA UNION: Standley 20935 (US GUEL: Standley 21142 (GH). SAN SALVADOR: Calderón. 1565 s VICENTE: Standley 2 1674 (GH). SANTA ANA: Carlson 765 (F). SONSONATE: Pittier 1928 (US). Nini CARAZO: Grijalva & Almanza 3572 (MO). CHINANDEGA: Robleto 1818 (МО). CHONTALES: Stevens & Montiel 19137 (MO). & Araquistain 717 (MO). ESTELI: Moreno 8141 (MO). JINOTEGA: Moreno 16419 (MO). LEON: Grijalva 2381 (MO). марк17: MacQueen & Styles 1 (MEXU). MANAGUA: Sandino 256 1 (MO). Masaya: Cal- deron et al. 70 (MO). MATAGALPA: Neill 1997 (MO). NUEVA SEGOVIA: Neill 1660 (MO). RIO SAN po WHO 3007 (MO). Rivas: Moreno yp (MO). z : Neill 4087 (MO). Costa Ri UELA: D 2047 1 (F). CARTAGO: Stork 2111 (Е). бене ASTE: Sousa et al. 12754 (MEXU). HEREDIA: Pittier & Tonduz 2515 (US). PUNTA- RENAS: 7 (BM). SAN JOSE: Skutch 4695 ЈАМАМА. COCLE: Folson & Kauke 2773 (MO). DARIEN: ed = Correa 3402 (MO). HERRERA: Allen 4069 (MO). Los SANTOS: Stern et al. 33678 (MO). PAN AMA: Dressler 3352 (MO). VERAGUAS: Folsom et al. 2478 (MO). ZONA DEL CANAL: Nee & Tyson 10894 (MO). Сива: Shafer 12420 (MO). JAMAICA: Proctor 37 = (МО). Puerto Rico: Duke 7346 (MO). Haiti: Nash 337 (F). REPUBLICA DOMINICANA: Mejía & Zanoni e (MO). COLOMBIA. CHOCO: Duke 11014 (MO). GUAJIRA: Arboleda et al. 420 (MO). 5АМТА MARTA: H. H. Smith 106 (MO). VALLE: Alston 7950 (F). VENEZUELA. ARAGUA: Pittier 12299 (F). DELTA AMACURO: Rusby & Squires ab Dip GUARICO: Aristeguieta 7032 (MO). PORTUGUESA: Ayma 94 (MO). MERIDA: Liesner & González 10611 MO ZULIA: Davidse et al. 18638 (MO). — Inga Xxalapensis Benth. (pro sp.), London J. Bot. 4: 616. 1845. TIPO: Linden 671 (holotipo, K). México. Veracruz: Inga X zacuapanica Harms (pro sp.), Repert. Spec. Nov. Regni Veg. 19: 63. 1923. TIPO: México. Veracruz: Purpus 3684 (isotipos, F, MO, US). Aqui se está postulando que /nga xalapensis Benth. y su sinónimo /. zacuapanica Harms son vera Wild. e /. lati- se trata de una hibridización de origen hibrido entre /. bracteata Harms; muy comun que aun no se estabiliza, y por tanto hay ejemplares intermedios y otros más cercanos а uno u otro progenitor; sin embargo, los mas comunes son con hojas similares a /. vera, pero con la bráctea floral grande y ancha де /. lati- bracteata y la flor de tamano intermedio a chica y de las proporciones de /. latibracteata. La glan- dula interfoliolar varia de una a otra, predominando la de cavidad grande de /. latibracteata. Estos hibridos se presentan a casi todo lo largo de la distribución de /nga latibracteata, exten- diendose /. X xalapensis mas al sur hasta el oeste de Honduras. МЕН Ю. CHIAPAS: Ton 1794 indez M. Material de ap экы (F). HIDALGO: Herr ME AXAC ТА VER 12113 (F). HONDURAS. OCOTEPEQUE: Molina R. 2418 BM). LITERATURA CITADA Barroso, G. M. 1965. Leguminosas da Guanabara. Arch. Jard. Bot. Rio de Janeiro 18: 10 BENTHAM, C. Notes on Mimoseae, with a synopsis of dE onda T: Bot. 4: 577-622. 1875. Revision of the Suborder Mimoseae. vun Soc. London 30: 335-668. 76. Inga Willd. In: C. F. P. Martius (ed- fao. P Brasiliensis 15(3): 458-500. BEURLING, P. J. 1856. Primitae florae portobellensis. Kongl. Vetensk. Acad. Handl. 1854: 107-148. Britton, N. L. & J. N. Козе. 1928. Inga. In: №. Amer. Fl. 23(1): 2-16 1832. A Conen History of the — ш . G. & F. sin in Londor Pla ntes peg Trans. eu connues de : A new species of /nga (Leguminosae) rom Panama. Ann. Missouri Bot. Gard. 53: 377. Erias, Т. 1967. 14: 205-212. GENTRY, А. Н. 1982. Phytogeographic patterns as ev- idence for a Chocó refuge. Pp. 112-136 in G. T. Prance (editor), Biological Diversification in the Trop- ics. Columbia Univ. Press, New Yor Harms, Н. 1914. Über einige von P. Preus е Arten der ran io EE Repert. Spec. Nov Regni v Veg. 13: —— —. 1923. музы americanae novae. Re- Spec. Nov. Regni Veg. 19: 61-70. >. Inga Willd. In: Humboldt, pland y Kunth, Nova Genera et Species Bein 6: 283-305, ши ed.] Par LEON, J. 1966 entral м. and West Indian species of haa. (Leguminosae). Ann. Missouri Bot. Gard. 53: 265-359. MAcBRIDE, J. F. 1943. Inga Willd. In: J. F. Macbride, aps of Pe Field Mus. Nat. Hist., Bot. Ser. 13(3): Notes on the genus /nga. Phytologia мами С ud = florae brasiliensis. Flora 28. : p ir of 2. genus Inga. Pons U.S. Natl. Herb. 18: 173-223. 1929. The Middle а species of the genus Inga. J. Dept. Agric. Porto Rico 13: 117 17 Poncy, 0. Le genre /nga (Legumineuses, Mi- mosideae) en e ad Mure Мет. Mus. Natl. ist. Nat., Sér. B., 31: 1-124. 1688. “Historia s ... Vol. 2. Mariae SCHERY, R. W. 1950. Inga. In: R. E. Woodson & R. Schery oe Flora of Panama, Ann. Missouri 25 Bot. Gard. Volume 80, Number 1 1993 Sousa S. 269 EI Género /nga SLOANE, H. 1696. Catalogus Plantarum quae in Insula maica Pe Proveniunt. . . . D. Brown, Londini. A Voyage to the Islands Madera, Bar- bados, Nieves, S. Christophers and Jamaica. . . . Vol. 2. B.M., London. STANDLEY, P. C. Inga. In: P. C. Standley, Trees and shrubs of Mies. Contr. U.S. Natl. Herb. 23(2): . STEYERMARK. 1946. Inga Willd. In: P. C. Sendo, & J. A. voie Flora of Gua- temala. Fieldiana, Bot. 24(5): 34- STIEBER, M. T. € C. LANGE. hi ustus Fendler (1813-1883), professional plant collector: selected correspondence pi т Engelmann. Ann. Mis- 31. souri p Gard. ане 0 788. S a et ч жк Plantarum 5 es wederi, Upsalia тант, - 1892. ue novae v. minus cog- nitae austro-americanae. Flora 75(n.s. 50): 68-86. SIE. C. L. 1805- "ud Caroli a Linné на Plantarum. ... Vol. auk, Berolin ZAMORA, N. 1991. eee especies de ae Miller кинни para Mesoamérica. Brenesia 33: 99- . Tratamiento de la familia Mimosaceae (Faba- ceae) para Costa Rica. Brenesia (en prensa). REVISION DEL GENERO Mario Sousa S.? y Velva E. Rudd? STYPHNOLOBIUM (LEGUMINOSAE: PAPILIONOIDEAE: SOPHOREAE) RESUMEN Se confirma la segregación de а de Зорћога en base a su hábitat, estipelas, bractéolas, legumbre y numero cromosómico. Se amplian sus limites de cuatro a nueve especies, al describirse cinco nuevas especies para la ciencia: Styphnolobium burseroides, S. ide. 5. parv iflorum, 5. protantherum, y 5. sporadicum. Se reconocen dos secciones, haciendose una nueva combinación: sección Oresbios. El género se centra en el area mexicano-mesoamericana. Se muestra su cercana afinidad con Саба en base a el hábito, bractéolas, y dehiscencia de la legumbre ABSTRACT The genus Styphnolobium is recognized as a segregate from Sophora on the basis of habit, stipels, bracteoles, legume, and chromosome number. It is enlarged from four to nine species through the description of five species new to science: Styphnolobium наи 5. кар 5. parviflorum, S. раке, and 5. sporadicum. Two sections are recognized, one requi e new combination section Oresbios. The genus is centered in the Mexican- | агеа. Through similarities | in habit bracteoles, and НЕ of > legume, it is clearly most closely related to Ca El género Styphnolobium (Leguminosae: Pa- Ма (1981). Después Yakovlev (1967) cambió de pilionoideae: Sophoreae) fue creado por Schott — opinión y lo reconoció como género aparte, revi- (1830) a partir de Sophora japonica L., pero ha viendo también a Сайа Berlandier. sido considerado frecuentemente como parte de El concepto de este grupo como unidad, bajo Sophora, como en la ultima revision del género las diferentes categorias que ya apuntamos, еп por Tsoong & Ma (1981). Sin embargo Bentham realidad ha incluido solamente a dos especies So- (1837, 1839) y Burkart (1952) lo reconocen, aun- phora japonica L. | japonicum) que el primero posteriormente (1865) modificó ra- del sureste de Asia y 5. affinis Torrey et A. Gray dicalmente su criterio, o bien, existe el consenso (Styphnolobium affine) del sureste de los Estados de asignarle una categoria propia, asi Yakovlev ^ Unidos, pero en los últimos años han empezado а (1964) lo colocó como sección, que reconoce Rudd aparecer nuevas especies para la ciencia en México (1972) y como subgénero Tsoong en Tsoong & y Centro América que corresponden a este grupo ' El trabajo de campo de esta revisión y una corta estancia del primer autor en el New York Botanical Garden fueron financiados en parte por la Organización de Estados Americanos, Proyecto: Estudios dui NAE en algunos géneros de У Leguminosas, асе, y Palmas de México: Спорепепса, bienio 88-89, PRDCyT- . La finalización e este t | primer autor disfrutó de una beca del Missouri Botanical CE con financiamiento de la Fundación Jessie Smith Noyer, así como de la ayuda de la Dirección General de Personal Académico, UNAM, traducción de las diagnosis al latin y revisión del manuscrito; a Linda Ellis por sus magnificas láminas botánicas; a mi esposa Gloria Andrade de Sousa por su gran ayuda, particularmente el haber pasado el manuscrito a un procesador de textos; y a los ae de CR, HNMN, K, LINN, LL, MEXU, MO, NY, TEX, US por facilitar el material botanic о para su estu erbario Маана, Tne de Biologia, UNAM, Apartado Postal 70-367, Coyoacan, 04510, Mexico, D.F., Mexi co. ене А iology Department, California State University, 18111 Nordhoff Street, Northridge, California 91330 ANN. MISSOURI Bor. Савр. 80: 270-283. 1993. Volume 80, Number 1 1993 Sousa S. & Rudd 271 Revision del Género Styphnolobium (véase Sousa, 1990). Se totalizan hasta el momento nueve especies, que en conjunto permiten ahora evaluar mejor el problema al aumentar en varias veces la variabilidad conocida. Asi decidimos es- tudiar el grupo desde diferentes puntos de vista, invitando a G. Palomino y a sus alumnas Patricia Martinez y Caos Bernal (Bernal & Martinez, 1989), para ver el problema de los cariotipos y números cromosómicos, que a la vez ella logró que E. Martinez O. viera el polen, sus resultados son dados en el artículo que sigue a éste (Palomino et al., 1993 Ya sea que la posición, sea dentro de Sophora o fuera, Styphnolobium mantiene varios carac- teres propios distintivos y reconocibles, también es claro que está cercanamente relacionado a Sopho- ra sensu lato particularmente a Calia como pro- ponen en diferentes grados Yakovlev (1967, 1968), Tsoong & Ma (1981), y Polhill (1981). Asi podemos resumir sus diferencias con la si- guiente clave: la. Herbaceas perennes, arbustos o rem flores ebracteoladas; legumbre dehisc Sophora Arboles; flores ls тйк, indehis- cente. 2a. Hojas sin estipelas; cáliz bilabiado, el labio vexilar más largo que el carinal; legumbre con pericarpio seco; 2n = 18... Саба Hojas estipeladas; cáliz trunco o casi trun- co, si bilabiado, el labio vexilar más corto que el carinal; legumbre con pee. со- riáceo, mesocarpio у endocarpio carno n = 28. Sty а hu T M d Concordando en principio con Yakovlev (1967, 1968), separandolas en tres grupos, estando muy cercanamente relacionado a Calia, aunque no bajo la misma categoria como proponen Tsoong & Ma (1981), los que las incluyen bajo el subgénero Styphnolobium, incluyendo ademas а Sophora pachycapa Schrenk ex C. A. Meyer del Asia Me- nor, la cual es totalmente diferente en hábito ade- más de que carece de estipelas y bractéolas, ca- racteres ces de Sophora sensu stricto. e los caracteres morfológicos, los nu- meros cromosomicos de seis de las nueve espec лез conocidas de Styphnolobium presentan un nümero de 2n — 28 (Palomino et al., sigue); contrastando con 2n = 18 tanto para 50- 1993, articulo que phora como para Calia. Resulta muy sugestivo el hecho que en las Sophoreae de Norte América existan dos géneros más con nümero diploide de 8, es decir Cladrastis y Pickeringia (véase Goldblatt, 1981) éste último considerado de posi- ción incierta y frecuentemente colocado, aunque con reticencia, en las Thermopsidae por su similitud en las hojas (Turner, 1981) Por otro lado, si comparamos morfológicamente a Styphnolobium, Cladrastis, y Pickeringia no resultan tan diferentes en cuanto a su tipo básico de inflorescencia (Fig. 1A-G). También muestran claras cercanias con Саба, asi tanto Pickeringia y varias especies de Calia presentan gradaciones de flores francamente solitarias en las axilas de las hojas a racimos, esto ya había sido observado por Nuttall (1840) en Pickeringia. La gradación va de hojas 3-folioladas cuyos foliolos laterales se re- ducen grandemente y constituyen las bracteolas y el foliolo terminal o central que funge como la bráctea floral subyacente al pedicelo. En cuanto a Cladrastis debido a lo efimero de sus diminutas bractéolas, no son descritas y en general se inter- pretan como carentes; su legumbre dehiscente, aplanada y delgada lo diferencia. Dentro de las Sophoreae pero ya no en el grupo Sophora (Polhill, 1981), sino en el grupo Dussia, Dussia, en el único conteo reportado por Goldblatt (1981), presenta tambien 2n = 28. Aqui se trata de un género con francas relaciones fitogeográficas hacia Sudamérica, que sin embargo comparte más caracteres con el grupo Styphnolobium que con Sophora sensu stricto. Al aumentar dramáticamente el conocimiento de la variabilidad de Styphnolobium es posible sugerir tendencias evolutivas en el género (véase Tabla 1); así de su análisis Styphnolobium affine resultó la ünica especie que cuenta con todas las fases primitivas de los nueve caracteres tratados. Por otro lado ninguna posee todas las fases avan- zadas de los caracteres usados: la que conjuntó más fue S. burseroides y en base a estos caracteres es que arreglamos las especies en el tratamiento sis- temático, aunque con limitaciones, ya que en al- gunas especies faltó el conocimiento de estructuras o estadios como son las plántulas Sobre su fitogeografia Sousa y Delgado (en pren- sa) dan una relación que aqui modificamos en vista de la nueva información obtenida. Sus especies tienden a ser endémicos restringidos (Fig. 2), que actualmente habitan en climas templados a calidos y de subdesérticos a húmedos. Existen, afortunadamente, varios registros fó- siles de frutos y foliolos asociados de la Formación Claiborne е n & Dilcher, 1986; Heren- deen, 1990, 1992), a del Eoceno Medio del ceo de Kentucky y Tennessee y asociados a bosques que contaban con: Podocarpaceae, Thea- ceae, Anacardiaceae, Magnoliaceae, Hamameli- daceae, Moraceae, Sapotaceae, Annonaceae, Nys- 272 Annals of the Missouri Botanical Garden FIGURA Botones florales mostrando bractea floral y bractéolas.—A. Pickeringia montana. (Tomado de McMurphy 80, MO.)—B. Sophora А &ypsophyta о (Tomado de Burgess 1764, МО.) — С. о burseroides. (Tom e Guízar EXU.)— D. EUIS afifne. (Tomado de Palmer 6264, MO.) —E. Hu ase c conzattii. a omado к^ Е Тоггеѕ С. 5231, 0.) — Е. Styphnolobium оо (Tomado de Slama 43, MO.)—G. Cladrastis kentukea. (Tomado de iris pom MO.) Volume 80, Number 1 1993 Sousa S. & Rudd 273 Revision del Género Styphnolobium TABLA 1. Tendencias evolutivas en Styphnolobium. Primitivos Avanzados Estipelas bien desarrolla- Estipelas muy reducidas a das faltantes Inflorescencia axilar, ra- Inflorescencia terminal, cemosa 1 Cáliz no ктөр Cáliz 5-Чеп Polen con a exina de re- Cáliz giboso Cáliz trunco o casi trunco Polen con la exina de re- tículo cerrado Ovario glabro Legumbre homogenea Legumbre con marcas Germinación con los coti- Germinación con los coti- ledones epigeos ledones hipogeos Eofilos bien desarrollados Eófilos reducidos a catáfi- los saceae, Lauraceae, Fagaceae, Juglandaceae, Myricaceae, Rhamnaceae, Euphorbiaceae, y otras. Esta composición florística de la vegetación se acer- ca mucho a los bosques caducifolios o mesófilos de montaña que actualmente habitan las vertientes húmedas de las sierras del sur de México al norte de Nicaragua. Sin embargo, dada la abundancia, diversidad, y grupos de leguminosas conocidas (He- rendeen, 1990, 1992) y considerando lo pobre en leguminosas que actualmente este tipo de vegeta- ción posee, es posible que también habian otros tipos de vegetación menos húmedos y más cálidos, como selvas medianas a altas subperennifolias y aún más estacionales, francamente caducifolias con abundantes mimosoideas como los depositos ates- tiguan, y no necesariamente confinado a áreas tem- pladas como Herendeen (1992) sugiere. Por lo anterior, la naturaleza de endemismo en este género es de paleoendémicos, en los que el representante asiatico muestra ya caracteres avan- zados y únicos en el género, como el polen y el tipo de inflorescencia; sus parientes americanos parecen ser sus predecesores, siendo Styphnolo- bium affine el posible puente entre la rama asiatica y la rama americana del género. diera Schott, Wiener Z. Kunst. 3: 844. 1 . Sinónimo: Sophora secc. rums Hom (Schott) Yakovlev, Trudy Lenin- : 56. 1964. Ho- monimo tardio: Tsoong, Acta Phytotax. Sin. 19(2): 161. 1981. ESPECIE TIPO: Styphnolo- bium japonicum (L.) Schott. him.-Farm. Inst gradsk. Arboles; corteza lisa, fisurada a escamosa pla- coide irregular, en ocasiones exfoliante, con tonos metálicos; el crecimiento en brotes cortos y largos, frecuentemente ramifica en los cortos y las ramas tienden a parecer casi verticiladas. Hojas alternas, estipuladas, 1 -ріппадаѕ, imparipinnadas, los folio- los generalmente opuestos proximalmente y mar- cadamente alternos distalmente o todos opuestos o alternos, estipelados. Inflorescencias racemosas, axilares, raramente terminales y paniculadas; flores pediceladas, los pedicelos articulados basalmente, con una bráctea subyacente subamplexicaule en la base; bractéolas presentes, 2; hipantio desarrollado; cáliz lobado a truncado, giboso o no giboso; corola papilionoidea, polipétala, rosado pálido a blanque- cina, glabra, pétalos unguiculados; lámina del es- tandarte suborbicular, reflexa, maculada; alas si- métricas a asimétricas, sin rugosidades; pétalos de la quilla libres y ampliamente traslapados; estam- bres libres; anteras uniformes; ovario glabro a pe- loso, estilo glabro; estigma terminal, atenuado. Le- gumbre indehiscente, suculenta, moniliforme a torulosa, péndula, frecuentemente estipitada por aborción de las semillas proximales, homogénea o con marcas lisas en el margen vexilar, exocarpo coriáceo, mesocarpo y endocarpo carnoso y mu- cilaginoso; semillas oblongo-elipsoideas, lisas, ne- gruzcas a negras, con el hilo lateral, con el en- dospermo cubriendo al embrión y la radicula incurvada. Plántulas con los cotiledones epigeos o raramente hipogeos, las primeras hojas por encima de los cotiledones (eófilos) alternas en ocasiones subopuestas, imparipinnadas o raramente reduci- das a catáfilos. 2n = CLAVE PARA LAS SECCIONES DE STYPHNOLOBIUM la. Inflorescencia terminal, pen ada. o NT Sección 1. Styphnolobium lb. Inflorescencia : SUAE. racemosa, en ocasiones tenuemente se ramifica Sección 2. Oresbios SECCION 1. Styphnolobium. Arboles; inflorescencia terminal, paniculada; po- len con la exina de reticulo cerrado. Una especie del sureste de Asia. Styphnolobium japonicum (L.) Schott, Wiener Z. Kunst. 3: 844. 1830. Basiónimo: Sophora јаротса L., Mant. Pl. 1: 68. 1767. TIPO: Japón (aparentemente introducida), Kleynhoff s.n. Para sinónimos véase Rudd, 1972. Figura Arboles hasta 25 m; ramas glabras, esparcida- Hojas 7-17-folioladas; pulas 3-4 mm, linear-atenuadas, pronto caducas; mente lenticeladas. esti- peciolo 1.4-3.5 cm; raquis foliar 5.5-25 cm; fo- 274 Annals of the Missouri Botanical Garden o 800 kms 1 n 1 1 n FiGURA 2. Ma vertido, 5 gus pad — Cuadrados, dium. — Cuadrado lleno, S. conzattii. — Estrella, S. liolos alternos a subopuestos, estipelados, las esti- mm, aciculares, subpersistentes; lami- nas 1-6 x 0.5-2.5 cm, lanceoladas, la base obtusa pelas ca. en ocasiones aguda, el ápice acuminado a agudo, el haz glabrescente, el envés esparcidamente ca- nescente sericeo, en ocasiones glauco. Inflorescen- cias terminales, en paniculas; floración tardia; ejes tri- secundarios 3.5-17 cm; bráctea 1-1.5 mm, angular-atenuada, caduca; pedicelo 3 mm, arti- culado pesima: e diminutamen- eolas te canescente sericeo; brac ca. mm - triangular-atenuadas, cercanas e cáliz, pronto ca- ducas; flores 12-15 mm; cáliz 3-4 mm, ciatiforme, no giboso, glabrescente excepto moderada a den- samente canescente sericeo sobre los lóbulos y mar- gen, 5-lobado, los lóbulos hasta 1.2 mm, redon- deados; pétalos blancos a amarillo pálido; estandarte largamente unguiculado, la una hasta 4 mm; ovario esparcidamente amarillento sericeo. Legumbre 3- 12 x 0.7-1 cm, moniliforme, homogénea, gla- brescente, excepto esparcidamente canescente se- ricea en el estipete, constricciones y rostro esti- pitada, el cg hasta cm; semillas О 0 , 1-8 por fruto. Plantulas con los co- ieee: ети los eófilos imparipinnados, en ocasiones l-foliolados o 3-foliolados. 2n = 28 a de distribución de las especies mexicano-mesoamericanas de Styphnolobium. — Trian S. burseroides. — Circulos, caudatum. — Asterisco, S. gulo in- S. parviflorum. —Circulos llenos, S. spora- monteviridis Distribución y hábitat. Esta especie aparen- temente es sólo nativa de China, pero desde hace mucho se ha cultivado y naturalizado en Japón, de donde originalmente se describió, en la actualidad se le cultiva ampliamente en el mundo. Se trata de un árbol caducifolio que florece en agosto- septiembre y fructifica en octubre-noviembre en China. CHINA. ANHUI: K. Yao 8472 (MO). CHENGJIAN o & Kan 79360 (MO). HUBRI: Е. H. Wilson 2558 (M (0). JIANGRI: K. Yao 9113 (MO). XIZANG: G. Forrest 24754 (MO). YUNNAN: A. Henry 9871 (US). Material ПЕН El ejemplar anotado como Зорћога јаротса en el Herbario de Linneo LINN 522.9 no es S. Japonica, у si es Sophora se parece más a 8. tomentosa L. Para fines de esta revisión no se estan consi- derando los taxa infraespecificos descritos para esta especie, para ello véase el trabajo de Sophora de Tsoong & Ma (1981) Styphnolobium japonicum muestra el carácter ünico en el género de inflorescencia terminal en panicula y sus relaciones más cercanas parecen ser con 5. affine. Volume 80, Number 1 1993 Sousa S. & Rudd 275 Revision del Género Styphnolobium SECCION 2. Oresbios dne M. Sousa et Rudd, comb. nov. Basionimo: Sophora secc. Oresbios Rudd, Phytologia 21(5): 327. 1971. ESPECIE TIPO: ii (Standley) M. Sousa et Rudd Arboles; inflorescencia axilar, racemosa; polen con la exina de retículo abierto. Ocho especies de América del Norte y Mesoamérica, extendiéndose una a Colombia. CLAVE PARA LAS ESPECIES DE STYPHNOLOBIUM SECC. ORESBIOS la. Legumbre moniliforme, 0.7-1 cm de diámetro, con l- -8 semi llas. (Texas, Oklahoma, негу = m O Qa [=] s > = o [nd © ч E. ~ 5 5 u- £g 3 © = m: = x: [e] | М о > e Sudamérica.) 2a. Legumbre у әд velutina ‚О canescen- te-sericea За. Legumbre homogénea sin marcas en el margen vexilar. 4a. Foliolos coriáceos; estipelas bien desarrolladas; fo lio los redondea- a, — — . S. conzattii 4b. Folíolos cartáceos; estipelas poco desarrolladas; foliolos largamente ce; infl iin con cáliz 3.5- 41 mm. (Nica 3. 5. дага ЗЬ. Же, a con marcas ocres aa al secarse) en el margen vexila 5a. Hojas 15-21 folladas, СЕХ -3 х 0.8- te tomentoso en el. envés; cáliz $i x 2-3.3 cm, gla n el envés; caliz casi trunco (Costa Fa Jalisco, México.) ____ 6. 5. protantherum 2b. Legumbre glabra; ovario glabro 6a. Fs liz 5-lobado; hojas 13- 17-foliola- s. (Monteverde, Costa Rica.) S. monteviridis 6b. Cáliz trunco; hojas 5-13-folioladas. Flores 11-13 mm; hojas 5-7- fololadas (Sur de Veracruz, México.) .. O us ia Tb. Flores 15- 20 1 mm; hojas 9-1: folioladas. (S. ме : Color bia.) Lu . о ит 1. Styphnolobium affine ue et A. Gray) : 807. 1843. Ba- siónimo: Ta o айыз Torrey et A. Gray, mer. 13} 3 O. 1840. TIPO: Estados a Louisiana, & River, Leavenworth s.n. (lectotipo, designado por Rudd (1972), NY). Figura 1D alp., Repert. Bot. Syst. 1 Arbustos a árboles 2-5(-10) m; ramas espar- cidamente canescente sericeas a tomentulosas cuando jóvenes, pronto glabras, esparcidamente lenticeladas. Hojas (9-)13-—19-folioladas; estipulas ca. 2 mm, lineares, pronto caducas; peciolo 1.2- 3.5 cm; raquis foliar 4.5-13.5 cm; foliolos alter- nos, frecuentemente opuestos, estipelados, las es- tipelas 1 mm, lineares, pronto caducas; láminas 1— 4.5 X 0.5-2.5 cm, elipticas en ocasiones ovadas u obovadas, la base aguda en ocasiones algo cu- neada a obtusa, el ápice obtuso a agudo, el haz glabro a glabrescente, el envés esparcidamente ca- nescente sericeo. Inflorescencias axilares; floración tardía; pedúnculo 0.5-2 cm; raquis floral 2-8 cm; bracteolas 0.2-0.5 mm, lineares, frecuentemente distantes del cáliz, pronto caducas; flores 13-16 mm; cáliz 3-4 mm, ciatiforme, no giboso, espar- cidamente canescente sericeo, 5-lobado, los lóbulos hasta 1 mm; pétalos blancos a lila, en ocasiones matizados de pürpura, glabros; estandarte larga- mente unguiculado, la una hasta 3.5 mm; ovario densamente canescente sericeo. Legumbre 3-6 (-15) x 0.7-1 ст, moniliforme, ћотовепеа, gla- bra excepto canescente sericea en el estipite, cons- tricciones y rostro, i qe el estipite hasta 1.5 cm; semillas O T 0.4 cm, 1-4 por ruto. Plántulas con ms cotiledones epigeos, los eófilos imparipinnados. 2n = 28. Distribución y hábitat. Conocida del sureste de los Estados Unidos en los estados de Texas, Oklahoma, Arkansas, y Louisiana, en bosques ca- ducifolios con Ulmus, Bumelia, Celtis, Crataegus, Diospyros, y Fraxinus; en bosques de Quercus y en praderas; frecuentemente en suelos derivados de calizas o alcalinos. Florece de finales de marzo a mayo y fructifica de agosto a noviembre. Nombres vernáculos. | Eve's necklace, neck- lace tree. Material representativo. ESTADOS UNIDOS. ARKANSAS: B. F. Bush 2487 (MO), G. W. Letterman s.n. (MO), J. Palmer 9396, (MO), 26719 (MO). LOUISIANA: J. W. и 21142 (LL). OKLAHOMA: 1. Emig 797 up G. J. Goodman 5460 (TEX), S. a s.n. (MO), Hopkins & M. Rogers 3976 (MO, TEX), Е. L. une 36246 (MO), E. J. Palmer 1047 3 (мо) 24047 (МО). TEXAS: D. с Correll 15919 (LL), L. Cory 43021 (TEX), 54769 (LL), D. Chesser s.n. и C. L. Lundell 276 Annals of the Missouri Botanical Garden p Lundell 10285 (LL), D. Lynch s.n. (TEX), A. Ruth 5 (US), 626 (US), L. D. Smith 385 (US), H. H. York D (TEX), 46257 (TEX). 2. Lor eae conzattii (Standley) M. Sou- sa et Rudd, . Missouri Bot. Gard. 77(3): 575. 1990. Eis ease Sophora conzattii iu Contr. U.S. Natl. Herb. 23(2): 436. 22. Calia conzattii ciate Yakovlev, г. Leningradsk. m.-Farm. Inst. 21(4): 45. 1967. TIPO: México. ( Oaxaca: Conzatti et al. 3171 (holotipo, US; isotipo, MEXU her- bario Conzatti). Figuras 1E, 2, y 3 Arboles 22-25 m; corteza placoide; ramas ca- nescente velutinas a tomentulosas cuando jóvenes, pronto glabrescentes, esparcidamente lenticeladas. Hojas 5—9-folioladas; estipulas ca. 0.8 mm, linear- aleznadas, pronto caducas; peciolo 2-5 cm; raquis foliar 4–10.5 cm; foliolos opuestos a casi opuestos en ocasiones los distales alternos, estipelados, las estipelas 1—2 mm, recurvadas, crasas, persistentes; làminas 3.5-5.5(-7.5) x 2.2-3.5(-4.5) cm, co- riaceas, ovadas, eliptico-ovadas a anchamente elip- ticas, la base obtusa a aguda, el apice obtuso a agudo, el haz y enves glabros. Inflorescencias axi- lares; floración tardía; pedúnculo muy reducido hasta 0.8 cm, raquis floral 2.5-5.5 cm; bráctea са. 1 triangular-atenuada, pronto caduca; bracteolas ca. 0.6 mm, oblongas, pronto caducas; mm, flores 20-25 mm; cáliz 5-6 mm, infundibuliforme, ligeramente giboso, ferrugineo sericeo; casi trunco; pétalos violeta, glabros; estandarte largamente un- guiculado, la una hasta 4 mm; ovario canescente sericeo. Legumbre 6.5-14 x 2.2-2.5 cm, toru- losa, homogénea, canescente sericea, densamente en el estipite, constricciones y rostro, esparcida- mente a la altura de las semillas, estipitada, el estipite hasta 2.5 cm; semillas 2 x 1.4 ст, 1-3 semillas por fruto. 2n — 28. Distribución y hábitat. | Especie endémica del Cerro Espino en la Sierra Madre del Sur, Oaxaca, Mexico, en altitudes entre 1,150 y 1,200 m, poco comün en selvas medianas perennifolias, en suelos rocosos. Florece de abril a mayo y cuenta con frutos maduros en abril del ano siguiente. Nombres vernáculos. — Frijolillo, granadillo. MEXICO. OAXACA: 4901 (MEXU, MO), Material representativo. 598 (MEXU), Torres C. MEXU , MO). Sousa 5231 Esta especie es muy caracteristica por sus hojas coriáceas; estipelas persistentes, crasas y recur- vadas; y las flores grandes. 5mm IGURA 3. Styphnolobium conzattii. — А. Base de foliolo terminal, mostrando e iip prominente: y pul- vinulo. (Tomado de R. Torres C. 1.) 3. Styphnolobium caudatum M. Sousa et Rudd, nn. Missouri Bot. Gard. 77: 573, fig. 1990. TIPO: Nicaragua: Sousa et al. 12985 (holotipo, MEXU; isotipos, BM, HNMN, MO). Figura 2. Arboles de más de 30 m; corteza escamosa pla- coide irregular, con tonos metálicos, con abundan- tes lenticelas prominentes; ramas diminutamente canescente sericeas cuando jóvenes, pronto espar- cidamente, esparcida a moderadamente lenticela- Volume 80, Number 1 1993 Sousa S. & Rudd 277 Revision del Género Styphnolobium das. Hojas 9-13-folioladas; estipulas desconocidas; pecíolo 2.5-5.5 cm; raquis foliar 7-15 cm; foliolos alternos, estipelados, las estipelas hasta 0.8 mm, oblongas, pronto caducas; laminas 4-9 x (1.3-)2- 2.7 cm, lanceoladas, la base redondeada ligera- mente asimétrica, en ocasiones cuneada, el ápice largamente acuminado a caudado, el haz glabro, el envés esparcidamente sericeo. Inflorescencias axilares, en ocasiones algo paniculadas, 12-17 cm; pedicelo 5-9 mm; cáliz (en frutos inmaduros) 3.5— 4 mm, giboso, pardo velutino, casi trunco. Legum- bre 5-14 x densamente ferrugineo velutina, estipitada, el es- tipite hasta 4 cm; 1-2(-4) semillas por fruto (se- 2.3-2.7 cm, torulosa, homogénea, millas maduras no fueron vistas). Distribución y hábitat. Solo se conoce de la localidad tipo en Esteli, Nicaragua, de un sólo árbol, en una canada hümeda, en la vega de un rio, a 900 m de altitud. Con frutos inmaduros durante mayo. laterial representativo. NICARAGUA. ESTELI: P. P. M Moreno 25655 (MEXU, MO) 4. DE monteviridis M. Sousa et , sp. nov. TIPO: Costa Rica. Puntarenas: Mone. Pacific slope, 10?18'N, 84%48'0, 1,450 m, W. Haber 9118 (holotipo, MEX U; isotipo, MO). Figura 2. Arbores magni, folia proterantha, 13-17 foliolata, fo- liolis lanceolatis vel ellipticis, glabris. Inflorescentiae a llares, pedunculi ad 1.2 cm longi, rhachides florales (3.5 —)5- 11 ст longae; bracteoli diminuti, decurrentes, caducei; flores mediani; calyx infundibuliformis, gibbosus, 5-lobatus; vexillum unguiculatum; ovarium rum. Legumina ho- mogenea, glabra, 1 -2(-4) seminibus onusta. Arboles (15-)30-40 m; ramas esparcidamente canescente pilosas cuando jovenes, pronto glabras, moderado a densamente lenticelado-verrugosas. Hojas 13-17 -folioladas; estipulas ca. З mm, linear- atenuadas, pronto caducas; peciolo 2.5-6 cm; ra- quis foliar 9-15 cm; foliolos alternos a subopuestos proximalmente, estipelados, las estipelas ca. 1 mm, linear-liguladas, pronto caducas; láminas 3.2-6 x 2.2- a aguda, el ápice acuminado, el haz y епуез glabros. 2.5 cm, lanceoladas a elipticas, la base obtusa Inflorescencias axilares; floración tardia; pedünculo muy reducido hasta 1.2 cm; raquis floral (3.5—)1 1 cm; bráctea 3-4 mm, linear-aleznada, pronto ca- duca; pedicelo 4-6 mm, pardo velutino; bractéolas ca. 0.2 mm, lineares, decurrentes por 1 mm, pron- to caducas; flores ca. 20 mm; cáliz 7 mm, ancha- mente infundibuliforme, giboso, pardo sericeo а velutino, 5-lobado, los lóbulos ca. 1 mm, apicula- dos; pétalos lavanda con nervaciones oscuras; es- tandarte largamente unguiculado, la una hasta 6 mm; ovario glabro. Legumbre (inmadura) 5.5-9(- 13) x 2.2 cm, torulosa, homogénea, glabra, es- tipitada, el estípite hasta 2.5 cm, glabro; semillas (inmaduras) 2 x 1.7 cm, 1-2(-4) por fruto. Plán- tulas se desconocen. Distribución y hábitat. Esta especie solo se conoce del área de Monteverde, Costa Rica, en altitudes entre 1,400 y 1,550 diados de febrero y a finales de julio, el periodo de m. Florece a me- fructificacion aun se desconoce, pero a finales d agosto ya cuenta con frutos bastante desarrollados pero aün inmaduros. Paratipos. Costa RICA. PUNTARENAS: Monteverde, 10°47'N, 84%50'0, А. Gentry & W. Haber 48728 (MO); 10°18'N, 84?43'O, B. Hammel 17037 (MO); along road to Eston Rockwell farm, G. S. Hartshorn 1898 (MO NY) Esta especie pertenece al grupo de Styphno- lobium con ovario glabro, legumbre homogénea. El epiteto se refiere a Monteverde, área que es muy rica en endemismos. 5. Styphnolobium parviflorum M. Sousa et Rudd, sp. nov. TIPO: tacion de Biologia Tropical Los Tuxtlas, U.N.A.M., Mpio. San Andrés Tuxtla, 150 m altitud, en selva alta perennifolia, suelo arci- lloso, 9 enero 1974, M. Sousa 4299 (holotipo, MEXU; isotipo, MO). Figuras 2 y 4. éxico. Veracruz: Es- Arbores magni; folia proterantha 5-7 foliolata, foliolis ovatis, oblongis n axillares; pedunculi ad 1 cm long 4 cm longae; bracteoli parvi, пени pats flores parvi; calyx infundibuliformis, agibbosu atus; = atum; ovarium gla ri precem hom 1(-2) seminibus onusta. Cotyledones epi- longe piden genea, glabr "aeae. Arboles 22-35 m; cuando jóvenes, pronto glabras, moderadamente ramas canescente sericeas lenticelado-verrugosas. Hojas 5- 7-folioladas; esti- pulas ca. 1.5 mm, lineares, pronto caducas; peciolo 2.5-4 cm, raquis foliar 2.5-6.5 ст, foliolos opues- tos a alternos, estipelados, las estipelas ca. 1.2 mm, lineares, 1.7-)2.2 elipticas, la base obtusa, el ápice obtuso a apiculado pronto caducas; laminas 3.5-8.5 х -4 cm, ovadas, oblongas a anchamente ~ en ocasiones acuminado, el haz y епуез glabros. Inflorescencias axilares; floración tardia; pedünculo muy reducido hasta 1 cm; raquis floral 1-4 cm; bráctea ca. 2.5 mm, angostamente triangular-ate- nuada, pronto caduca; pedicelo 2- mm, pardo sericeo, ђгастеојав ca. 0.5 mm, liguladas, pronto caducas; flores 11-13 mm; cáliz 3-4 mm, infun- dibuliforme, no giboso, pardo sericeo, trunco a casi trunco; pétalos blancos a rosado pálido; estandarte 278 Annals of the Missouri Botanical Garden FIGURA 4. Styphnolobium parviflorum M. Sousa et Rudd. — A. Rama con inflorescencias. — B. Cáliz pedicelado 2 ^ con bráctea y bractéolas. — C. Estandarte. — D. AI Petalo de la quilla. —F. Gineceo y androceo. (Tomados de a.— E. Ре Ponce y Cedillo T. 6.) —G. Legumbre. (Tomado de Calzada 994.)—H. Semilla. (Tomada de Sousa 4299.)— I. 94.) Plántula. (Тотада de Calzada 99 largamente unguiculado, la una hasta 4 mm; ovario glabro. Legumbre 4.5-11 x 3.5-4 cm, torulosa, homogenea, glabra, sésil a estipitada, el estipite hasta 3 cm; semillas 2.8 x 1.8-2 cm, 1(-2) por fruto. Plantulas con los cotilédones epigeos, los eófilos imparipinnados; 5-9(-10)-foliolados; 2n — 28 Distribución y hábitat. Conocida solamente de dos áreas del estado de Veracruz, la región de Los Tuxtlas y la zona de Uxpanapa, en vegetación Volume 80, Number 1 1993 Sousa S. & Rudd 279 Revision del Género Styphnolobium de selvas altas perennifolias a subperennifolias, for- mando parte del dosel de la selva con Brosimum alicastrum Sw., Simira salvadorensis (Standley) Steyerm. y Pseudobombax sp. en litosoles calizos y en suelos negros pedregosos de origen volcánico. En altitudes entre 100 y 250 m. Florece de prin- cipios de diciembre a la primera quincena de enero, la fructificación se inicia en febrero, habiendo fru- tos maduros hasta mayo. Nombres vernáculos. Мејаза, melcocha, palo de Eun a. Paratipos. | MEXICO. VERACRUZ: Mpio. Jesús Carra za: 0.5 km al N del Poblado 2, 17?15'30"N, 94940' О. T. Wendt et al. 5282 (MEXU); Mpio. San 2 dura m = Biología E Los Tuxtlas, 5 236'N, 95909'0, J. I. Calz ada и al Mo us (MEXU), p 52 (MEXU), С. Ibarra M. 615 (MEXU), 778 (MEXU), /099 (MEXU 1299 (MEXU, MO), F. Ponce & R. Cedillo "ee 6 (MEXU), S. Sinaca C. & F. Chigo S. 729 (MEXU, MO); Laguna Escondida, a 3 km al NO d e la ы de Biologia Tropical Los Sinaca C. 708 (MEXU, MO). Tuxtlas, Especie que cuenta con varios caracteres avan- zados en el género, tales como cáliz trunco o casi trunco, ovario y legumbre glabras, reducción del nümero de foliolos, los cuales en plántula son más numerosos que en las hojas de los ejemplares ar- bóreos y flores pequeñas, por este último carácter se nombra a esta especie. 6. Styphnolobium protantherum M. Sousa et Rudd, sp. nov. TIPO: México. Jalisco: entrada a Cuenca 3, Estación de Investigación, Ex- perimentación y Difusión Chamela, U.N.A.M., Mpio. La Huerta, 24 febrero 1984, J. Arturo S. Magallanes 4159 (holotipo, MEXU). Fi- guras 2 y 5. Arbores mediani vel O folia 9-11 агыр Шо entibus vel glabris. Inflore тез, mediani; pedunculi 1.4-3.5 ст longi; rhachides florales 3 cm longae; bracteoli parvi, inent attenuati; calyx infundibuliformis, dun gibbosus, subtruncatus, ovarium riceum. Legumina maculis ochraceis, laevibus үн 2(-4) seminibus onusta. Cotyledones epigaea Arboles 12-25 m; corteza longitudinalmente fi- surada que forma placas; ramas canescente seri- ceas cuando jóvenes, pronto glabras, esparcida- mente lenticeladas. Hojas 9-1 l-folioladas; estipulas ca. 1.6 mm, linear-aleznadas, pronto caducas; pe- ciolo 1.8-4 cm; raquis foliar 8-11(-13) cm; fo- liolos opuestos a alternos, estipelados, las estipelas 4-0.6 mm, triangular-atenudas, pronto caducas; láminas 3-6.5 x 1.8-3.2 cm, ovadas, anchamente elipticas, en ocasiones elipticas u obovadas, la base obtusa, el ápice obtuso, apiculado a cortamente acuminado, el haz y envés esparcidamente canes- cente sericeos, pronto glabros. Inflorescencias axi- lares en ocasiones paniculadas; floración precoz; pedúnculo 1.4-3.5 cm, raquis floral 7-13 cm; bráctea ca. 1.5 mm, angostamente triangular-ate- nuada, pronto caduca; pedicelo 3-3.5 mm, pardo sericeo; bractéolas ca. 1 mm, linear-atenuadas, pronto caducas; flores 14—16 mm; cáliz 4-6 mm, infundibuliforme, algo giboso, pardo sericeo, casi trunco; petalos rosados; estandarte largamente un- guiculado, la una hasta 5 mm; alas algo asimétricas; ovario densamente canescente sericeo. Legumbre 4–12.5(-15.5) x 2-2.5 cm, torulosa, con marcas ocres y lisas en el margen vexilar, glabra excepto sericea en el estipite, estipitada, el estipite hasta 6 cm; semillas 2.5 х 1.5 cm, 1-2(-4) por fruto. Plántulas con los cotiledones epigeos, los eófilos N imparipinnados, 7—9-foliolados. 2n = Distribución y hábitat. Sólo Mc de la costa de Jalisco en selvas medianas subperennifolias a subcaducifolias, en suelos metamorficos. En al- titudes alrededor de los 800 m. En el periodo de sequia esta especie es caducifolia y florece sin hojas a finales de febrero a principios de marzo, fructifica a partir de marzo contando con frutos maduros en mayo. Paratipos. MEXICO. JALISCO: Mpio. La Huerta, Ез- tación de Biología, Chamela, U.N.A.M., E of Eje C o Bullock 1495 (MEXU, MO); Arroyo el Larco, к 990 (MEXU, МО); Estación de и Сћате При куне 2177 (MEXU, МО); entrada a la Cuenca " 9 (MEXU); Vereda Buho, 4237 (MEXU, 10). >: Styphnolobium protantherum muestra ciertas afinidades con S. burseroides, pero es marcada- mente diferente (véase clave); su epiteto hace én- fasis en su floración previa a la producción de hojas en la estación seca. 7. Styphnolobium sporadicum M. Sousa et Rudd, sp. nov. TIPO: Colombia. Antioquia: Mpio. de Venecia: 4.2 km E of Bolombolo on road to Venecia, Hacienda La Plata, 06?01'N, 15?48' W, 920 m, J. L. Zarucchi & B. Eche- verry 1666 ae MO; isotipos, CR, HUA no vistos, K, MEXU, US). Figura 2. Arbores mediani; vel magni. Folia proterantha, 9- foliolata, foliolis ovatis, ellipticis vel ovato- ellipticis. ^N res, dracone ad 1 cm lon ngi, rhachides florales 1.5-8 m longae; bracteoli parvi; flores me- diani; calyx банний, paullo "ibbosus, fere trun- ium gla P 1. Legumina homogenea, glabra, 1-2 seminibus onus rescenciae axillar E catus; ovar Arboles 18-30 m; ramas esparcidamente ca- nescente sericeas cuando jóvenes, pronto glabras, 280 Annals of the Missouri Botanical Garden A NO — FIGURA 5. Styphnolobium protantherum M. Sousa et Rudd. — A. Ramas con hojas y flores. (Con hojas tomado G de Magallanes 2177, con flores de Bullock 1495.)— B. Flor sin pétalos mostrando pedicelo, bractéolas, cáliz, gineceo, y androceo. — С. Estandarte. — D. Ala. — E. Pétalo de la ашћа. —F. Legumbre mostrando las marcas vexilares. — C. Plántula. (Flor, frutos, y plántula tomados de Bullock 1495.) esparcida a densamente lenticelado-verrugosas. alternos, estipelados, las estipelas 0.6-0.8 mm, Hojas 9-13-folioladas; estipulas ca. 1 mm, trian- liguladas, pronto caducas; láminas (4-)6-9.5(- gular-atenuadas a liguladas, pronto caducas; pe- 11.5) x 3.5-4(-5) cm, ovadas, elipticas a ovado- ciolo 3.5-5.5 cm; raquis foliar 9-13 cm; foliolos lanceoladas, la base obtusa a aguda en ocasiones Volume 80, Number 1 Sousa S. & Rudd Revision del Género Styphnolobium cuneada, el ápice acuminado a cortamente acu- minado, el haz y envés glabros. Inflorescencias axilares; floración tardia; pedunculo muy reducido hasta 1 cm; raquis floral 1.5-8(-11) cm; bractea 1-1.5 mm, triangular-aleznada a linear-aleznada; bractéolas muy reducidas, pronto caducas; flores 15-16 mm; cáliz 4-6 mm, infundibuliforme, algo giboso, pardo a cinéreo velutino, casi trunco; pe- talos rosado pálido a purpüreos; estandarte larga- mente unguiculado, la una hasta mm; ovario x 2.2-2.7 cm, torulosa, homogénea, glabra, sésil a estipitada, el estipite glabro. Legumbre 4.2-10 hasta 2.5 cm; semillas (maduras se desconocen), 1-2 por fruto. Plántulas se desconocen. Distribución y hábitat. temente tiene una distribución amplia y esporádica, Esta especie aparen- en selvas altas subperennifolias con Brosimum ali- castrum Sw. sobre litosoles calizos en altitudes en- tre 50 y 500 m, en el norte de Oaxaca, México; en altitudes entre 300 y 600 m en la reserva de El Imposible en El Salvador; en selvas medianas caducifolias a 920 m, en el Valle del Rio Cauca, Antioquia, Colombia. Florece en marzo y fructifica en el norte de Oaxaca a finales de mayo y en El Salvador a finales de noviembre. ombres vernáculos. Almendrillo, almendri- llo de montana (en El Salvador). Hacienda El Imposible, D. A. Cáceres 39 (MEXU), J. A. Tablas s.n. (MEXU). Esta especie es la que tiene la distribución más amplia del género, y por lo incompleto del material (se desconocen sus flores) en su limite norte en Oaxaca, Mexico, es dificil precisar si en realidad estamos tratando solamente con una sola entidad 'ахопописа o dos; sin embargo, рог los caracteres disponibles no nos es posible separarlas convincen- temente. El epiteto hace referencia a la distribución esporádica de esta especie. 8. Styphnolobium burseroides M. Sousa, Rudd selva baja caducifolia sobre ladera caliza, 9 junio 1984, F. González Medrano 14126 (holotipo, MEXU; isotipos, MEXU, MO). Fi- guras 1C, 2, y 6 Arbores parvi; folia 15-21 foliolata, foliolis ovalis, mins vel late ellipticis, subtus arachnoidei-ca- nescentibus. Inflorescentiae axillares; flores coaeta- praedita, subra semina glabescentia. Plantulae hy- pogaeae, eophyllis cataphyllis reductis. Arboles 4—10 m; corteza externa escamosa, ex- foliante; ramas densamente canescente velutinas a tomentosas cuando jóvenes, pronto glabrescentes. Hojas 15-21 -folioladas; estipulas 2-3 mm, angos- tamente triangular-atenuadas, pronto caducas; pe- ciolo 0.2- 1 cm; raquis 6–10.5 cm; foliolos alternos a opuestos, en ocasiones estipelados, las estipelas 0.4-0.6 mm, liguladas, frecuentemente faltan; la- minas 1.4-2.2 x 0.7-1.2 cm, ovadas, oblongas a anchamente elipticas, la base obtusa, el ápice obtuso, apiculado a mucronulato, el haz canescente velutino a piloso, el envés canescente aracnoideo. Inflorescencias axilares; floración coetánea; pedün- culo muy reducido a nulo; raquis floral 1.2-4 cm; bráctea ca. 2 mm, linear-oblanceolada; pedicelo 2—3 mm, densamente canescente velutino a piloso; bractéolas 0.5-0.6 mm, lineares, pronto caducas; flores 14—15 mm; cáliz 5-6 mm, ciatiforme, mar- cadamente giboso, esparcidamente canescente pi- loso, más densamente sobre los lóbulos, 5-lobado, los lóbulos hasta 1 mm, triangulares; pétalos ro- sados, excepto el estandarte que es blanco con máculas verde-amarillentas; estandarte largamente unguiculado, la una hasta 4.5 mm; ovario densa- mente canescente sericeo. Legumbre 4-10.5 x 1.6–1.7 cm, torulosa, con marcas ocres y lisas en el margen vexilar, esparcidamente pilosa a gla- brescente excepto sericea en el estipite, constric- ciones y rostro, estipitada, el estipite hasta 2 cm; semillas 2.3— 1.4 cm, 1-2(-3) por fruto. Plántulas con los cotiledones hipogeos, los eófilos reducidos a catáfilos. 2n — Distribución y hábitat. gion sur del Desierto de Tehuacán, en los estados de Oaxaca y Puebla en México, en vegetación de selvas bajas caducifolias con Cephalocereus sp., Bursera spp., . Actinocheita fili- cina (Sessé et Mociño) Barkley, Leucaena, y Wim- Endémico de la re- cacia coulteri, meria, en matorrales esclerófilos con Actinocheita, Rhus, y Brahea siempre sobre suelos someros, negros, de origen calizo, en altitudes entre 1,520 y 1,900 m. Florece en marzo y junio y fructifica en julio y marzo. Paratipo. MEXICO. OAXACA: Distr. Huajuapan de León, Yucumduchi, 10 km al límites de Puebla-Oaxaca, 4. Garcí Зара dia EXU); Distr. Teposcclala, a 3 km al NO de Tamazulapán, carr. Tamazulapán- А de León, Е. Maria 28424 (MEXU, MO). PUEBLA: Mpio. Hue huetlán, P. Tenorio L. & А. Salinas 11539 (MEXU, 282 Annals of the Missouri Botanical Garden FIGURA 6. de estipula. Medrano 14126.)— D. Estandarte. N. 1343.) —H. Legumbre mostrando | zdlez mados de Guizar marcas vexilares. (Tomado de Sousa 11952.)—I. Semilla. (Tomada de Guizar N. 1343.) —Ј. Plántula. (Tomada de Sousa 11959. Volume 80, Number 1 1993 Sousa S. & Rudd Revision del Género Styphnolobium 283 MO); Mpio. Izúcar de SEHE igi Infiernillo, 12 km al SE de Raboso, dent e la cuenca Vistahermosa, E. Guizar N. 1343 MEN} Mpio. cas , 8-19 km al S de Molcaxac, carr. hacia San Jua ui ad. F. Chiang & A. Valiente 2005 (MEXU, MO), Chiang et al. 2149 (MEXU), F. González Medrano et al. 11727 (MEXU), M. Sousa et al. 11952 (MEXU), 11959 (MEXU). Especie muy caracteristica tanto por la pelosi- dad en sus hojas, su cáliz fuertemente giboso y dentado, las contrastantes marcas de sus frutos, y las plántulas con cotiledones hipogeos y los eófilos reducidos. El nombre especifico aduce a su super- ficial semejanza con el género Bursera tanto en sus hojas, como en su corteza que se exfolia. LITERATURA CITADA BENTHAM, G. 1837. Commentationes de leguminosarum nger, Vienna neribus commen- tationes. Ann. Wiener Mus. Nh 2: 161) 142. BERNAL, G. . P. MARTINEZ А. 1989. Determinación del cariotipo de algunas especies del género Sophora L. sens. lat. = Leguminosae). Tesis UNAM (Zaragoza), Mexic a e a 1952. Las уе инан Argentinas. ACME Agency, Buenos Aires. башы gag Р. 1981. Cytology a, the phylogeny of the Leguminosae. Pp. 427-463 in R. M. Polhill & P. H. Raven (editors), MR in Legume System- ; . Royal Botanic Gardens, Kew HERENDEEN, P. s. 1990. Fossil History of Leguminosae from Eocene of Southeastern North America. Ph.D. Dissertation. Indiana University, e 1992. The fossil history of the Leguminosae from the Eocene of Southeastern North America. Pp 85-160 in P. S. Herendeen & D. L. Dilcher (editors), Advances in Legume Systematics, Part 4. The Fossil Record. Royal Botanic Gardens, Kew. & . DILCHER. 1986. Fossil Leguminosae from Келеш Kentucky and Tennessee. Second In- ternational Legume Conference, “Biology of the Le- guminosae,”” Missouri Bot. Gard. Poster Session (véa- se Abstracts p. 47 NUTTALL, T. 1840. Pickeringia. In: J. Torrey & A. Gray. Fl. N. Amer. 1(3): 388. Wiley & Putnam, New York, London. Palomino H., G., M. P. MARTINEZ, M. C. BERNAL & M. БОЗА 5. 1993. Diferencias cromosómicas entre e los géneros Sophora L. & Styph- nolobium Schott. Ann. Missouri Bot. Gard. 80: 284- үш к. M. 1981. Tribe 2. Sophoreae Sprengel (1818). Pp. 213-230 in R. M. Polhill & P. H. Raven (editors), Advances in Legume Systematics, Part 1. Royal Botanic Gardens, Kew. Fen 2 Корр, V. Leguminosae- Faboideae-Sopho- reae. N. mer. Fl., ser. 2, 7: 1-53. SCHOTT, КА 1830. Styphnolobium. Wiener Z. Kunst. 2: SOUSA 5 "M. 1990. Adiciones a las Papilionadas de la flora de Nicaragua y una nueva combinación para Oaxaca, México. Ann. Missouri Bot. Gard. 77: 573- DELGADO. Mexican Leguminosae: phy- togeography, endemism e origins. /n: T. P. Ra- mamoorthy, R. Bye, A. Lot & J. Fa (editors), Bio- logical Diversity of Mexico: Origins and Distribution. Oxford Univ. Press, New York. (In quen TsooNc, P. С. 1981. A study on the genus e ise (new taxa). Acta Phytotax. Sin. 19(1): 1-22; 19(2): 143-167 TURNER, * 1981. Tribe 31. Thermopsideae Ya- kovlev in 2). Pp. 403-407 in R. M. Polhill & P. H. Raven (editors), Advances in Legume Systematics, Part 1. Royal Botanic Gardens, Kew. YAKOVLEV, С. P. 1964. A new system for the genu Sophora, and its po Trudy Leningradsk Khim-Farm. Inst. 17: 50-77. [In Russian.] 967. Systematical Е geographical studies of the genus Sophora L. and allied genera. Trudy Leningradsk Khim-Farm. Inst. 21(4) 42-62 (In Russian.] 1968. The genus Сайа Тегап et Berl. (Sopho- eae) in America. Trudy Leningradsk Khim-Farm. Inst, 26: 104-112. [In Russian. ] DIFERENCIAS CROMOSOMICAS ENTRE ALGUNAS ESPECIES DE LOS GENEROS ЗОРНОКА L. Y STYPHNOLOBIUM SCHOTT' Guadalupe Palomino,? Patricia Martinez,” Carmen Bernal,? y Mario Sousa S.* RESUMEN En este estudio se determinó el número cromosómico (2n) y cariotipo de | aes sec и 5. velutina var. пни. 5. tomentosa, Styphnolobium japonicum, 5. а. 5. protantherum, S. burseroides, y S. conzattil. Las tres primeras presentaron un Zn = 18, un núme o básico x = , peque ñas variaciones en sus cari iotipos, un par de cromosomas con constricción i ed Las cinco e restantes, mostraron 2 8, x = 14; sus cariotipos fueron muy similares y presentaron res de cromosomas con constricción secundaria. Los resultados tre apoyan la propuesta de Sousa y Rudd de incluir al ue de Lo dd con 2n — 28 en el género Styphnolobium. ABSTRACT ' first three species presented 25 — In this study chromosome numbers (27) and karyotypes of Sophora secundiflora, S. velutina var. zimbabweensis, tomentosa, Styphnolobium japonicum, S. parviflorum 18, with 5. conzattii were These herum, S. burseroides, and th a basic chromosome number of x = 9. species had small variations in their karyotypes and one chromosome patr with secondary constriction. The other five species 2n presente — 28, x — 14 and their karyotypes wer e very s These results agree ‘with Sousa and Rudd’s proposal to incide species with 2 imilar, three iu showing secondary constrictions. = 28 i in the genus Styphnolobium. El género Sophora L. pertenece a la familia Leguminosae (— Fabaceae) y a la tribu Sophoreae. Se encuentra distribuido en las regiones tropicales y tempiadas de ambos hemisferios, y es un género grande y variable. Rudd (1972) senala que ciertas especies podian tratarse en géneros diferentes. Pol- hill (1981) senala entre 45 a 50 especies para el género, Rudd (1968) senala 75, mientras que Ya- kovlev (1967) enlista 40. Rudd (1968) coloca a Styphnolobium Schott en sinonimia bajo Sophora. А Styphnolobium, sin embargo, se le ha dado reconocimiento como ca- tegoria infragenérica, asi Yakovlev (1964) y Rudd (1972) lo reconocen como sección en el género. Hasta hace poco sólo se habia hecho la transfe- rencia a Styphnolobium de Sophora japonica y 5. affinis y recientemente Sousa y Rudd 1990) hacen lo propio con Sophora conzattii y en Sousa, describen una nueva especie para Nicaragua: Styphnolobium caudatum. Entre 984 personal del Herbario Na- cional de México (MEXU) colectó ejemplares aün no registrados para la ciencia y cuyas caracteris- ticas hacian pensar en plantas pertenecientes al genero Sophora. Comparando estas nuevas espe- cies con material de Sophora japonica, 5. affinis, y 5. conzattii, Sousa & Rudd (1993), observan diferencias entre estas especies con especies del género Sophora, ubicando estos autores a las pri- meras en el género Styphnolobium. ‘Este trabajo fue financiado en parte por la ОЕА, Proyecto: Estudios Biosistemáticos en Algunos Géneros de Leguminosas, PAN y Palmas de México: Citogenética, Bienio 88-89, PRDCyT-OEA y por el CONACyT CLAVE PCCNCNA-02 У » Nacional MEXU de la Universidad Autónoma de México el haber РРА las Se semillas de Sophora бшек on Sophora ane Styphnolobium protantherum, Styphnolobium burseroides, y Styphnolobium conzattii; al Herbario Naci var. zimbabweensis p Universidad de Lis Ag radecemos a Javi artinez Ramón la obter al-Jardin Botánico de Zimbabw оа, is de Se ag hen japonicum, para realizar esta investigación nción de las fotogr e, las semillas de Sophora а s de este traba ajo. PS ub de Biologia, Universidad Nacional Autónoma de México, Jardin Botánico, Ciudad Universitaria, C.P. 04510, D.F., México. : dei de ee Universidad Nacional Autonoma de México, Herbario Nacional, Ciudad Universitaria, C.P. 04510, D.F., Méxi ANN. Missouni Вот. Garp. 80: 284-290. 1993. Volume 80, Number 1 1993 Palomino et al. 285 Nümero Cromosómico en Sophora y Styphnolobium Nümeros cromosómicos (2n) y cariotipos de TABLA 1. is AES Schott tres especies del género Sophora L. y cinco del género Nümero cromo sómico órmula stricción Taxon Localidad (2n) cariotipica secundaria Sophora secundiflora México, Tamaulipas: sg de 18 6m + 3m — sm lm Maquihuana, ej. La Pena, Mal- da s.n. México, Tamaulipas: Mpio. Bustamente, ej. Felipe An- geles, Mart — velutina var. Zimbabwe: ш proporciona- 18 6m + 3m — sm lm mbabween das por el Herbario Nacional- Jardin Botanico de Zimbabwe Sophora tomentosa México, Quintana Roo: Pto. Mo- 18 4m + 4m — sm + Ism lm — sm relos, Sousa 10995 Styphnolobium japo- Cultivada: semillas proporciona- 28 13m + 1m — sm 3m icum das por ardin Botánico de la Universidad de Lisboa, Portugal Styphnolobium parvi- México, Veracruz: Mpio. Sn. 28 12m + 2m — sm 3m florum Andres Tuxtla, /barra s.n. Styphnolobium pro- México, Jalisco: Mpio. La Huer- 28 llm + 3m — sm Зт tantherum ta, Chamela, Bullock 1495 Styphnolobium burse- México, Puebla: Mpio. Molca 28 12m + 2m — sm 3m roides jac, Sousa 11952, Sousa 11959 México, ice Distrito de Po- 28 13m + lm = т Зт Styphnolobium con- айй бенен Cerro Espino, Torres 49 En un estudio palinológico de cuatro especies del género Sophora sensu stricto y seis del grupo Styphnolobium, Martinez (com. pers.) observó di- ferencias en relación a el tipo de reticulo, tamano y forma del grano de polen, estableciendo dos gru- pos. El primer grupo está formado por: Sophora secundiflora, S. tomentosa, S. arizonica, y S. macrocarpa y el segundo grupo esta integrado por Styphnolobium parviflorum, S. protantherum, S. burseroides, y S. conzattit. Revisando los estudios quimiotaxonómicos rea- lizados por Izaddoost (1975, 1979) y Bailey (1974), se observa que 5. јаротса y S. affinis carecen de alcaloides (Izaddoost, 1975), ácido pipecólico y ácido 4-hidroxipipecolico (Izaddoost, 1979), a di- ferencia de las otras especies del género Sophora estudiadas por Izaddoost (1975, 1979) emás se aprecia que ambas especies contienen el poli- mero galactomanosa con composición similar (An- derson, 1949; Kooiman, 1971; Bailey, 1974), pero las demás especies del género Зорћога estudiadas por Bailey (1974) no lo contienen. Reid & Meier (1970) y Bailey (1971), en base a sus investiga- ciones en la presencia o ausencia de galactomanosa en las semillas de leguminosas, suponen que estos compuestos pueden ser buenos indicadores репе- ricos. En relación al género Sophora, Bailey (1974) indica que aunque él sólo estudió ocho taxa en total, sus resultados parecen ser de considerable interés relativo a la revision que hace Yakovlev (1967) del género, al colocar a 5. japonica y S. affinis en el género Styphnolobium, ya que son las ünicas especies que contienen galactomanosa. Para las especies del género Sophora se han reportado nümeros cromosómicos n — 19 y 11 y 2n — 16, 18, 22, 28, y 36 (Federov, 1974; Gold- blatt, 1981b, 1984, y 1988). Goldblatt (198 la) consideró que la tribu Sopho- reae tuvo un origen poliploide con x — 14, posteriormente en su evolución ha habido una des- cendencia aneuploide con la diferenciación de li- neas gapo en la tribu. Sólo se tiene conocimiento e 8 para Sophora ene S. affinis, y a lutea (Goldblatt’ 1 a) l objetivo de este trabajo fue obtener el número cromosómico 2n y el cariotipo de tres especies de a Sophora pertenecientes a varias secciones y com- pararlas con el género Styphnolobium del cual se obtuvo información de cinco especies de las diez conocidas (ver Sousa € Rudd, 1993). Asi se estudió Annals of the Missouri Botanical Garden Volume 80, Number 1 1993 Brad et al. 287 ero Cromosómico en Sophora y ли им Sophora secundiflora [sección Сайа (Berlandier) Rudd], S. velutina var. zimbabweensis [sección Disemaea (Lindl.) Yakovlev], y S. tomentosa (sec- ción Aigialodes Rudd). En Styphnolobium se de- terminaron cariotipos para Styphnolobium japo- nicum, 5. conzattii, y tres especies nuevas para la ciencia: Styphnolobium parviflorum, S. pro- tantherum, y S. burseroides (Sousa & Rudd, 1993). MATERIALES Y METODOS Las plantulas y las semillas utilizadas en esta investigación fueron proporcionadas por el MEXU, Index Seminum del Jardin Botánico del Instituto de Biologia UNAM y Herbario Nacional-Jardin Bo- tánico Nacional de Zimbabwe, Africa (Tabla 1). Todo el material fue determinado por el ültimo autor y los ejemplares de respaldo se depositaron en el MEXU Las semillas de las distintas especies fueron so- metidas a tratamientos previos a la germinación. Las semillas de Sophora secundiflora (Everitt, 3), S. velutina, y Styphnolobium japonicum germinaron al ser tratadas con ácido sulfürico con- centrado durante 5-15 minutos, las де Зорћога tomentosa al remojarlas en agua destilada durante 48 horas y finalmente las de Styphnolobium pro- tantherum, S. burseroides, y 5. conzattii con es- carificacion mecánica. Después de estos trata- mientos, se sembraron en cajas Petri con algodón y papel filtro esterilizadas y mantenidas en una estufa a una temperatura de 30 + 2°C. Las semillas que germinaron se colocaron en vermiculita para su enraizamiento, regándose con solución nutritiva Hoogland. Posteriormente las plántulas se transplantaron a tierra preparada en una proporción de tres partes de tierra negra por una de tierra de hoja, y actualmente se mantienen vivas en los invernaderos del Jardin Botánico. Para Styphnolobium parviflorum se obtuvieron plán- tulas traidas de la Estación Biológica de Los Tuxt- las, del Instituto de Biologia de la UNAM, las cuales ueron mantenidas en los invernaderos del Jardin Botánico ( os meristemos radiculares seleccionados para la observación de las células en metafase se ob- tuvieron de raices secundarias en división celular activa. Estas raices tenian una longitud de 1-2 ст y fueron cortadas de 7:30 a 8:00 A.M., pretratadas con el mitostático 8-hidroxiquinoleina a una con- centración 0.002 M, durante cinco horas a una temperatura de 18?C y en la obscuridad. | término de este tiempo, las raices fueron lavadas con agua destilada y despues se fijaron en solucion Farmer (alcohol absoluto y ácido acético glacial, en una proporción de 3:1); en seguida fueron almacenadas en el refrigerador a 4?C, donde pudieron conservarse por tiempo indefinido. Para realizar la tinción de los cromosomas, las raices se hidrolizaron con ácido clorhídrico 1 Ма 60?C durante 10 minutos y posteriormente se in- trodujeron en una solución de Feulgen preparada a base de fücsina básica de acuerdo con Garcia ~ 1988) durante una hora y a temperature ambien- æ Ф El meristemo radicular ya tenido fue cortado y colocado en un portaobjetos limpio, agregando una gota de aceto-orceina al 2%. Después se colocó el cubreobjetos dándole un ligero aplastamiento y gol- peándolo con la punta de un lápiz para separar las células y poder observarse al microscopio. Ense- guida se fijó la preparación por el método del hielo seco (Conger & Fairchield, 1953) y finalmente se montaron los preparados en bálsamo de Canadá. Para la elaboración de los cariotipos se selec- cionaron 20 células de cada individuo, consideran- do cinco plantas por población. Las mejores células fueron fotografiadas en un fotomicroscopio П Zeiss, con un objetivo 100x y un optobar 1.25. Los cromosomas de las células seleccionadas se dibujaron en una camara lúcida Zeiss. A partir de estos dibujos y de las fotografias obtenidas se ela- boraron los cariotipos. La clasificación de los cro- mosomas se realizó de acuerdo a Naranjo et al. (1986) basado en Levan et al. (1964). RESULTADOS En la Tabla 1 y Figuras 1A-H y 2А-Н, se indican los nümeros cromosómicos (2n), las for- mulas cariotipicas, y los cromosomas con doble los géneros constriccion de las ocho especies de Sophora y Styphnolobium analizadas. DISCUSION En Sophora secundiflora se observó un 2n — 18, el cual concuerda con los ya publicados por Atchison (1949), Darlington & Wylie (1955), y Fict Cromosomas somátic os de: Es жа tomentosa. урп protantherum. — a, ub burseroic пеон а ит. Styphnolobium conzattii . Sophora secundiflora. — В. Sophora velutina var. zimbabweensis. hnolobium pare iflorum. —F. Styphnolobium chas y los números 1, 2, y 2. Згур 3 indican los cromosomas con doble constricción. Си escala corresponde а 5 y Missouri Botanical Garden Annals of the 288 ere ee n5 NAY NE m ns хи RM их XA RR 21% NM NIY SAR ON GK ли ни пи ми их ag ИК ИИ КИ H nw © ал e $9 xe es NREN RNRY HE ARAN ANNE d 3 а HS HY RY NN МИЙ ия NN KU ТИЈ ин ни KK ИН BE WK HH NYAS KH икки НИ RM ANOS КК ЛА BH МИ ХИ УМ ми ни vs onc vv 2 UHR AA SR RA NM se we g MM XN хиих onc v ws ХИ ХИ v Volume 80, Number 1 Palomino et al. 289 Nümero Cromosómico en Sophora y Styphnolobium Goldblatt (1981b, 1984). Un 2n = 18 se determina para 5. velutina var. zimbabweensis y es la pri- mera vez que este dato se informa, al igual que su cariotipo. El 2n = 18 determinado para 5. tomen- tosa en esta investigación confirma varios conteos previos informados para esta especie por Atchison (1951), Darlington € Wylie (1955), Federov (1974), y Goldblatt & Davidse (1977). En Styphnolobium japonicum se observó un п = , el cual está de acuerdo con los ya pu- blicados por Darlington & Wylie (1955), Berger et al. (1958), Federov (1974), Fernandes et al. (1977), y Goldblatt (1981b, c, 1984). Para S. parviflorum, S. protantherum, S. burseroides, y S. conzattii se determinó por primera vez en este trabajo un 2n — Sobre la base de los nümeros cromosómicos de las ocho especies estudiadas, se encuentran dos grupos claramente definidos, un primer grupo con un nümero básico de x — 9 y un segundo grupo con un x — 14. El observar diferentes nümeros básicos en especies relacionadas puede significar, según Love € Love (1974), que las mismas per- tenecen a géneros distintos, siempre y cuando esta informacion se relacione con algunas caracteris- ticas morfológicas (Moore, 1968), tal como ocurrió en Hauya y Xylonagra (Raven & Lewis, 1960). Esta situación es la presente en estos dos grupos de especies y los resultados citológicas son apo- yados por los caracteristicas morfologicas diferen- ciales mencionadas por Sousa & Rudd (1993), donde proponen la ubicación de las especies con x — ]4 en el género Styphnolobium. En Sophora secundiflora y S. velutina var. zimbabweensis se observó un cariotipo con seis pares de cromosomas metacéntricos y tres pares de cromosomos metacentricos-submetacentricos, un par de cromosomas metacentricos en constric- ción secundaria. El cariotipo de S. secundiflora concuerda con el dibujo de los cromosomas de esta especie encontrado por Atchison (1949). Kawakami (1930) reportó un n = 9 para Sopho- ra angustifolia y el dibujo de los cromosomas de esta especie es semejante al cariotipo de S. secun- diflora y S. velutina var. zimbabweensis. La única diferencia es que S. angustifolia no tiene un par de cromosomas con constricción secundaria. Los cariotipos obtenidos para Sophora secun- diflora y S. velutina var. zimbabweensis son li- geramente diferentes en relación a S. tomentosa en la proporción de metacéntricos, metacéntricos- submetacéntricos y submetacéntricos. Estas varia- ciones en los cariotipos pudieron haber sido causadas por inversiones pericéntricas y translo- caciones desiguales (Stebbins, 1971). Atchison (1949) obtiene para Sophora tetrap- tera y S. microphylla un 2n = 18 y un par de cromosomas con constricción secundaria, concor- dande con lo que se encontró en las especies con un 2n = 18 estudiadas en este trabajo. Los cariotipos de Styphnolobium japonicum, S. parviflorum, S. protantherum, S. burseroides, y 5. conzattii fueron muy parecidos, con tres pares de cromosomas metacéntricos con constricción se- cundaria. Se pudieron apreciar variaciones en la proporción de metacéntricos y metacéntricos-sub- metacentricos, las que pudieron originarse por rea- С ди cromosómicos llevados a ae durante el r lutivo de esta 1971). 5 Las relaciones de Lees. dextro de los cromo- somas que comprenden el cariotipo son usualmente similares en poblaciones y especies relacionadas, asi como el nümero básico de grupos de ligamiento genico (Kenton, 1986). Estas relaciones se man- tienen en las especies analizadas en este estudio, el primer grupo con x = 9 y cariotipos similares entre si, y son diferentes a las especies que mues- tran x — 14 donde los cariotipos de las cinco especies estudiadas son semejantes entre si. Considerando que la tribu Sophoreae probable- mente tuvo un origen poliploide con un x — 14 y en su evolución сенне generó descendencia соп nümeros básicos menores, x — 8, 9, 11, por aneu- ploidias Goldblatt, 19814), se puede pensar que el género que se estableció primero fue Styphno- lobium el que tiene un x — y posteriormente por aneuploidias se estableció el género Sophora con un x — 9. каа btenidos en este estudio En base a los se encontraron dos grupos citogenéticamente di- ferenciables, un primer grupo con un x — 9 y 2n — |8 formado por: Sophora secundiflora, S. velu- tina var. zimbabweensis, y S. tomentosa, y un segundo grupo con un x — 14 y 2n — 28 integrado — FIGURA 2. A. о secundiflora 2n = 18, 6m + 3m — sm.—C. Зорћога ш. 2n = 18, 4m + 13m + 1m — sm.— E. Styphnolobium parvifloru , Пт + 3m — sm.—G. s — sm + lsm.— = 28, 12m + 2m — sm.— = Bou. ree 2n = 18, 12m + 2m — Condens 2n = 28, 13m + lm — sm. La escala corresponde a 5 um. Cariotipos - tres especies del genero piis L. y 5 especies del género e ена Schott. — B. Sophora velutina v var. zimbabweensis 2n yphnolobium 290 Annals of the Missouri Botanical Garden por: Styphnolobium japonicum, S. parviflorum, S. protantherum, S. burseroides, y S. conzattii. LITERATURA СТТАРА ANDERSON, E. 19 Endosperm mucilages of legumes. Industr. Engin. Chem. 41: 2887-2890. ATCHISON, E. 1949. -Studies in the Leguminosae. TV. Ch and of miscellaneous Leguminosae J. Elisha Mitchell Sci. Soc. 65: 118-12 1951. па in the Leguminosae. VI. Chro- mbers among tropical woody species. -546. . Polysaccharides in ip Legumi nosae. Pp. 503-541 in J. B. Harborne, D. Bou & Turner (editors), teo um of js Leguminosae: и о London 19 ctomannans and öther soluble e »saccharides in (е bs rue Zealand J. B 12: 131-136. BERGER, C. A., E. К. Wirkus & R. M. McManoN. 1958. Cytotaxonomic studies in the Leguminosae. Bull. Tor- rey Bot. Club 85(6): 405-415. Concer, A. D. € L. M. FAIRCHIELD. 1953. quic freeze method for p smear slides permanent. Stain Technol. 28: DARLINGTON, C. D. & A. 3 "o 1955. Chromosome Atlas of e Plants 2nd edition. George Allen & Unwin, Londor Everitt, J. Н. 083. ón of mescal bean (So- Е secundiflora) seeds. Southw. Naturalist 28(4): 7-444. FEDEROV, A. (editor). 1974. Chromosome Numbers of Flowering Plants. NE by Otto Koeltz Science Publishers, Germa FERNANDES, A., M. F. p & M. Queiros. 1977. Contribution à à la connaissance cytotaxinomique des spermatophyta du ji al. киыр (supp. 2). 86. Bol. Soc. Brot., а 51: Garcia, V. А Téc cnicas y Procedimientos de Citogenética Vegetal. Universidad nU. de Cha- pingo, México. GOLDBLATT, P. diu C divi d and the phylogeny of Leguminosae. Pp. 427-463 in R. M. Polhill & P H. Raven (editors), аи іп Legume Systematics, Part 2. Royal Botanic Gardens, Kew. (editor). bu. to plant эсш numbers. pora D. Monogr. Syst. Bot. Misso Bot. Gard. 198 ое р. " ponen IT. . Index to plant ЕТ numbers. 1979- 1981. Monogr. Syst. Bot. Misso Bot. Gard. 8 194. (editar). TA Index to plant а numbers. pow 1985. Monogr. Syst. Bot. Missour Bot. Gard. 23: 1-122. G. Dav VIDSE. 1977. Chromosome numbers in legumes. Ann. Missouri Bot. Gard. 64: 121-128. Izappoosr, M. 1975. Alkaloid chemotaxonomy of the genus = Phytochemistry 14(1): 203-204. — . Synergistic effects of alkaloids and pi- с е оп Ба toxicity | the seeds of Sop secundiflora. Tex Sci. 31(4): 319-323. KAWAKAMI, J. : no numbers in Legu- · (Tokyo) и ind 328. ora conocimiento biologico de los recursos vegetales en éxico. Jardi hos Mm Biol. UNAM: 11-36. Kooman, P. 19 Structures of the galoctomannans m seeds of Annona muricata, Arenga saccha rifera, Cocos nucifera, Convolvulus o and Sophora d ae) a. deu Res. 20: 329-3 LEVAN, А., . А. SANDBERG. 1964. No А уен for Pe position on. chromo- somes. Hereditas 52: 201-220 Love, А. & D. Love. Tow labial Atlas of the Slovenian Flora, Volume 1. Cramer, Germany. MOORE, D. M. The karyotype in taxonomy. Pp. 61-75 in W. H. Heywood (editor), Modern Methods in Plant Tonino: Academic Press, London. NARANJO, C. A., L. Poccio & P. E. BRANDHAM. 1986. А new template for ses dun icq c | lassification of chromosomes. Darwiniana 27: 39-4 des R. M. 81. Ske: eae Pp. 3-230 in R. M. Polhill & P. H. Raven (editors), re e in Legume Systematics, Part 1. Royal Bo tanic Gardens, Kew Raven, P. H. & H . LEWIS. 1960. Observations on the chromosomes and relationships of Hauya and Xy- lonagra. Aliso 4: 483-484. Rel, J. S. С. H. MEIER. Chemotaxonomy aspects of the reserve galactomannans in leguminous seeds. Zeitschrift für Pflanzenphysiologie 62: 89-92. Корр, V. E. Leguminosae of Mexico-Faboideae. Sophoreae and Podalyrieae. Rhodora 70: 492- 32. = ? 72. Leguminosae-Faboideae-Sophoreae. North American Flora, Series II, Part 7. Botanical Garden, New York. Sousa S., M. 1990. Adiciones a las Papilionadas de la flora de Nicaragua y una nueva combinación para oer ке Ann. Missouri Bot. Gard. 77(3): 573-5 & | Кор 1993. Revisión del género Styph- nolobium (Leguminosae). Ann. Missouri Bot. Gard. STEBBINS, G. L "197 1. Chromosome Evolution in Higher Plants. Edward o London. YAKOVLEV, G. 964. A new system of s genus 'ophora L. id its phylogeny: s a ] Proc. Len- ingr. Chem.-Pharm. Inst. 1967. Notes on ik чуча and geogra- phy of the T ine Sophora L. and related dui [En ruso.] Proc. Leningr. "Chem. Puno. Inst. 42-62 Volume 79, Number 4, pp. 737-972 of the ANNALS OF THE Missouni BorANICAL GARDEN was published on October 14, 1992. Volume 80, Number 1, pp. 1-290 of the ANNALS OF THE MISSOURI BOTANICAL GARDEN was published on January 28, 1993. Monographs in Systematic Botany from the Missouri Botanical Garden Number 43 Comparative External Pollen Ultrastructure of the Araceae and Putatively Related Taxa Michael H. Grayum | 167 рр., illustrated, softcover, $30.00, plus postage. : This monograph presents the first extensive scanning electron microscope (SEM) survey of pollen of Araceae. Pollen of approximately 380 species of Araceae was studied, representing 99 of the 105 currently recognized genera. An effort was made to obtain material from all subgenera and sections distinguished in the most recent available taxonomic revisions (98 of 114 infrageneric taxa are represented in the study). Standardized, synoptic pollen descriptions - are provided for all genera studied, and both whole-grain and high-magnification SEM micrographs are assembled in an atlas of 63 plates and 397 figures. An additional 12 plates and 92 figures summarizing the data accompany a detailed analysis of the polarity and evolution of pollen character states in Araceae. А discussion of the relationships of Araceae to other flowering plants features 15 plates and 104 figures depicting pollen of potential aroid relatives, у Se | dada CU c pde to аа 7 То place an order, send check or money order in U.S. funds, payable through a U.S. bank; U.S. shipments: add 22.00 for one book and $.75 for each additional book; non-U.S. shipments: add $3.00 for one book, and $.75 for AE de until payment is received. Mail form with your check or money order, payable to Missouri Botanical Garden, Department Eleven, Missouri Botanical Garden О. Box 299 Please send copy(ies) of Monograph No. 43. 3t. Louis, MO 63166-0299, U.S.A. Send ойе. P" О Payment enclosed. O Send invoice ($1.00 fee will be Name added to total). Address — Postal Code Country 80(1) Prices ARE ЅОВЈЕСТ To CHANGE WITHOUT NOTICE CONTENTS Monograph of the Neotropical Species of Asplenium Sect. Hymenasplenium (Aspleni- aceae) Noriaki Murakami & Robbin C. Moran A Revision of the Genus Pleurothyrium (Lauraceae) Henk van der Рег ________-. A Revision of Panicum Subg. Dichanthelium Sect. Dichanthelium (Poaceae: Ра Paniceae) іп Mesoamerica, the West Indies, and South America Fernando O. Zuloaga, Roger P. Ellis & Osvaldo Morrone A Revision of Monnina Subg. Pterocarya (Polygalaceae) in Northwestern South America Bente Eriksen The Misconstrued and Rare Species of Commelina (Commelinaceae) in the Eastern United - States Robert B. Faden Tradescantia crassifolia (Commelinaceae), an Overlooked Species in the Southwestern . United States Robert B. Faden - El INC Inga (Leguminosae: Mimosoideae) del Sur de México y Citra meas Estudio d 223 Previo para la Flora Mesoamericana Mario Sousa S. | = | : Revisión del Género wur (Leguminosae: Papilionoideae: Sophoreae) ^ Mario - TOI SMS C OLS A tru kale i ы е ахсыс ен | - Diferencias Cromosómicas entre aise Especies de ka Сер Зорћога L. y Зрурћа lobium Schott - Guadalupe duros Patricia Parne Carmen Bernal y ; ; Mario Sousa S. | еи _ Cover illustration. | Pleurothyrium giganthum van der Werf, by J. Myers. a 191 Убе 80 К | Number 2 ` - Pm Volume 80, Number 2 Spring 1993 Annals of the Missouri Botanical Garden The Annals, published quarterly, contains papers, primarily in systematic botany, соп- tributed from the Missouri Botanical Garden, St. Louis. Papers originating outside the Garden will also be accepted. Authors should write the Managing 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 Henk van der Werff Editor, Missouri Botanical Garden Amy Scheuler Managing Editor, Missouri Botanical Garden Diana Gunter Editorial Assistant, Missouri Botanical Garden Ma gdalen Lampe irri Moran Publications Staff Ihsan A. Al-Shehbaz Missouri Botanical Garden Gerrit Davidse Missouri Botanical Garden Roy E. Gereau Missouri Botanical Garden Peter Goldblatt Missouri Botanical Garden Gordon McPherson | Missouri Botanical Garden P. Mick Richardson Missouri Botanical Garden _ The ANNALS OF THE MISSOURI Bii 2 (ISSN 0026-6493) is published quarterly e Missouri Botanical Garden, 2345 Tower Grove AY enue, St. Louis, MO 63110. Second class poi paid at St. Louis, MO and additional m annals OF For subscription information contact Department Eleven, P.O. Box 299, St. Louis, MO 63166-0299. Subscription price is $100 per volume U.S., $105 Canada and Mexico, $125 all other countries. Four issues per volume. The journal Novon is included in the subscription price of the ANNALS. PosrMASTER: Send address changes to ; | THE Missouri BOTANICAL GARDEN, Departmen 029. Eleven, Р.О. Box 299, St. Louis, MO 6316 у 25 © Missouri Botanical Carden 1993 [ THIS PUBLICATION IS PRINTED ON ACID-FREE PAPER. | AE inia Volume 80 Number 2 1993 Annals of the Missouri Botanical Garden KNOWLEDGE BROKERING: THE MECHANICS OF SYNTHESIS. INTRODUCTION! Nancy Morin? and Theodore M. Barkley? Two items prompted us to propose this topic for the 38th Annual Systematics Symposium held at the Missouri Botanical Garden 4-5 October 1991. One is that scientific libraries have been growing rapidly; with the subsequent reduction in any one person's ability to remain c urrent with his/her field. The other is that we systematis element of synthesis in our "ab and are sensitive ts have an important to the need to account for increasing amounts of information. Systematists fall into two intergrading camps, monographers and students of the flora and fauna. The former study natural groups through a variety of techniques to provide new insights and to elu- cidate evolutionary relationships; the latter assess and synthesize information, particularly the works of monographers, and they prepare floras and man- uals to account for all of the relevant organisms of a geographic region. These floras and manuals make the results of systematic research readily useable to a wider community, for they are a link between those who generate basic knowledge and the consumers of that knowledge. Floras and man- uals are ever more important for wise stewardship as mankind becomes more obsessive at impacting the landscape. The Missouri Botanical Garden is engaged in several sizeable floristic projects, so it seemed appropriate to direct this symposium at the nature of information resources and their utiliza- tion. A further factor in planning this symposium is the development of widely available computer ca- pabilities for handling large amounts of information. /e are the first generation of systematists to have the technical tools to evaluate the great amounts of information stored in our libraries and natural history collections. The speakers were selected to present infor- mation and views of which systematists are not traditionally aware. We believe that we were for- tunate to bring their unique expertise and breadth to this symposium The evening — K. W. Bridges of the University of Hawaii, presented a dynamic, multi- media program that impressed those in attendance with innovative ways to consider information. Un- fortunately, the program is not amenable to the printed page, and so it is not included here. 1 5 Missouri Botanical Garden, l at the Missouri Botanical Garden n St. Missou ouis, and the six articles a follow it are the proceedings of the 38th Annual Systematics Symposium of the edge e The ies s of Synthesis. The symposium was held 4-5 Oct. ,U.S upon work aa by the National [o Foundation under grant number DEB-8918138. ӨЕ аа Missouri 63166, ANN. Missouni Bor. Савр. 80: 291. 1993. SYNTHESIS: A HISTORICAL PERSPECTIVE! Theodore M. Barkley? ABSTRACT Biology and other scientific enterprises have generated an astonishing amount of information through the traditional techniques information, the em provide a mo ofe experiment an and there is both ne hesis of the accumulated о 50. Some segment erging technologies for assessing literature and ayosin information. The techniques of historiography may del. In 1955, I began graduate school, and in the course of events an advisory committee was formed, a thesis topic selected, the research methods agreed upon, and other routine matters were settled. Shortly thereafter, someone asked, “Have you exhausted all of the literature yet?” It was a perfectly rea- sonable question, and I set about finding and read- ing everything written on the topic. If difficult then, it would be all but impossible today. In the inter- vening years, the literature has "exploded," as they say, and access to it is both complex and incom- plete. 1 cannot imagine asking a beginning graduate student in 1991 if he/she has exhausted the lit- erature and then expecting an honest "yes" response. n We all can agree that the enterprise of scientists has been productive, for libraries are bulging with the products of our efforts. The stunning success of the Soviets with their launch of the first space satellite, “Sputnik I,” in 1957, prompted the U.S government to increase greatly the amounts of federal support for basic research. Shortly after Sputnik I, generous support began for the whole spectrum of the sciences and technology, through the National Science Foundation, the National In- stitutes of Health, and various other agencies. We may be justly proud of our science, as we point to the libraries and note the long shelves of data and discussions and conclusions. However, we presume that the products of our science have a utility and significance greater than those we have been find- ing in them, if we could just synthesize the knowl- edge that is in hand. LITERATURE RESOURCES The libraries of North American academic in- stitutions are diverse; they were reviewed and ranked recently in the Chronicle of Higher Edu- cation (1991). There are a few giants, e.g., Har- vard, University of California at Los Angeles and at Berkeley, University of Texas, and Yale, which have huge holdings and acquisitions programs. Harvard has nearly 12 million volumes and added 261,846 volumes in the academic year 1989- 1990. Most of the major university libraries fall in the range of 1.5-3 million volumes and ad about 40—60 thousand volumes per year. My home institution, Kansas State University (KSU), has a bit smaller library, with some 1.2 million volumes and an annual i iR approaching 40,000 vol- umes. The K$ rary has been growing at 4— 6% per year, as lese other academic libraries. The implication is clear. The KSU library is currently adding 564,000 inches of new shelf space (— 14,450 m) just to make the holdings accessible, and other libraries can tell of similar additions. The factors affecting the inescapable growth of libraries are treated by Okerson & Stibbs (1991) in a paper titled “The Library ‘Doomsday Machine." Li- brarians worry about budgetary problems and shelf space. The former can be handled, at least in theory, by appropriating more money; the latter is not so easy. The point is: How can we continue to store and access the accumulating literature? We systematists are as committed to scribbling as anyone, and we have produced a wonderful trail of literature. Tod Stuessy (1990) has written a fine textbook on the topic, Plant Taxonomy, the Sys- tematic Evaluation of Comparative Data, where- in he summarizes the practices and goals of tax- опоту in its broadest sense, as seen by one of this generation's notable practitioners of the science. ' This paper is contribution No. 92-231-A from the Kansas Agricultural Experiment Station, Manhattan. * Herbarium, Division of Biology, Kansas State University, Manhattan, Kansas 66506, U.S.A ANN. Missouni Вот. Савр. 80: 292-296. 1993. Volume 80, Number 2 1993 Barkley 293 Synthesis It is a fine text for beginning graduate students to "state of the art, 99 view the and it is deemed a worthy successor to Principles of Angiosperm Taxonomy by P. H. Davis & V. Heywood 63). Both Stuessy and Davis & Heywood draw on wide bases of references, and both have exten- sive bibliographies. Davis & Heywood is a tome of 532 pages, of which 44 pages are references, i.e., 8% or 1 page out of 12. Davis & Heywood cite 1,286 references, all published before 1963. On the other hand, Stuessy has 514 pages, of which 83 are references, i.e., 16% or l page ош of 6. Stuessy cites 2,862 references, of which 1,919, or % of them, аге from between 1970 and 1 July 1988 (the cut-off date in the preparation of the book), or a period of 18V? years. There are some interesting com- parisons. Like his predecessors, Stuessy was selec- tive. He gathered information and cited sources to illustrate how taxonomy is practiced, and he was not attempting to be comprehensive. Moreover, some authors and topics were given different de- grees of importance by Stuessy than by his pre- decessors, e.g., am, a theoretician of floral structure, has one work cited by Stuessy and six cited by Davis & Heywood; I. W. Bailey, the comparative anatomist, has one work cited by Stuessy and ten cited by Davis & Heywood; H. G. Baker, the ecological systematist, has three works cited by Stuessy and thirteen cited by Davis & Heywood. Even when we allow for the differences in the predilections of the authors and the different times when they were writing, it is evident that a notable portion of what is published has a certain "shelf life,” i.e., time after it is published, and eventually falls into respectable neglect. Note that the older literature is not wrong merel it is most relevant for a period of because it is old, rather, it is н politely forgotten. I wonder how often we have einvented the wheel" because of our inability to grasp what is hidden i in the literature. E. S. Deevy, a geologist of Yale University, reviewed the rele- vant literature for his 1949 landmark paper “Bio- geography of the Pleistocene," date he regarded the literature resources as and even at that ter- rifying.” SOURCES OF OUR LITERATURE The learning of antiquity has been admired by historians since the Renaissance, and much has been written about the great stores of information in the ancient world. The library at Alexandria surely was a justly fabled treasure of fact and philosophy. However, the literature from antiquity must be read with a different mindset than we moderns have, for books that were handwritten by scribes (rather than printed on a printing press) were written with the notion that they were merely permanent records and were not especially con- cerned with making a point, i.e., ancient books largely reflected conventional wisdom and were not polemics. Students of Hellenic scholarship note that Theophrastus wrote some rather obscure and el- liptic sentences. Probably what has survived are his lecture notes; he had no idea that posterity would hang on his every written word. The advent of printing from moveable type in the mid- 1 400s changed forever the ways we record and disseminate information. These changes and their impact are related most interestingly in From Script to Print, by aytor author was a man of letters at Cambridge Univer- sity, and he detailed how literate mankind of Eu- rope was apprehensive about the impact of printing. Medieval schooling included a prodigious memory as a component, which printing later obviated. The ability to read had meant previously that a person could read aloud from manuscript texts. Printed books meant that the literate public would learn to read silently and rapidly, and books could be written to argue a case or to be read for personal enter- tainment. Chaytor was not aware of the irony in his book when he discussed the degradation of the medieval values of learning that were brought about by a new technology. Today, we are uneasy about the revolution of electronic information transfer, and we complain that kids spend too much time watching television. In 1500, people worried that the younger generation was getting hooked on read- E g. About 50 years passed before the impact of printing and the realities of mass communication settled onto the literate communities of Europe. It happened nearly concurrently with the start of commercial exploitation of the New World, the rise of nationalism, and the great theological upheavals. Printing meant that Everyman could have a piece of learning and could write about it with the ex- pectation that someone would read what he had written. Printing was essential to put learning into the public domain and, therefore, to the making of modern science. Written publication was early done in books and in the Latin language of medieval scholarship. By the late 1600s, scientific publication was usually through journal articles in the colloquial languages. The development of scientific publication is detailed by Daniel Boorstin (1983) in The Discoverers. The shift of scientific pre-eminence from Italy to north- 294 Annals of the Missouri Botanical Garden western Europe through the 1600s saw, among other things, the rise of the Royal Society of Lon- don. Henry Oldenburg, a German who had im- migrated to England, became the Secretary of the Royal Society and, in effect, was the general editor of publications. In this capacity, he designed the notion of the journal paper and initiated the Philo- sophical Transactions, which is the type specii imen denburg, for he was the most influential editor in the history of science Journals are published by scientific. societies, commercial firms, governmental agencies, and rarely today but formerly even by individuals. Jour- nals have two purposes: (1) to communicate infor- mation among those who are interested and (2) to record information (i.e., to communicate to later generations). | have no idea how many journals there are, or have been, in bot entries in the B-P-H (Botanico-Periodicum-Hunti- anum) or in the Union List of Serials, used by any. Counting the librarians, would reveal many thousands. THE USES OF THE LITERATURE The advent of available modalities (e.g., inex- pensive laboratory equipment, easy travel, rela- tively easy publication) moved the natural sciences from contemplative activity to increased observa- tion and experiment. The 20th century has been the preeminent time for sophisticated observational and experimental methods. The graduate students of this century are trained with the presumption that good science means the addition of new data, derived from observation or experiment. Critical analyses and syntheses of previously published in- formation have rarely been stressed in graduate training. Thus, review papers and state-of-the-art works have not been esteemed quite so highly as publications derived from fresh experiment and/ or new observa Synthesis del КЕРНЕ of information al- ready at hand, however, have always been a part of scholarship. Synthesizers may be recognized as four species, and like many species with which 1 am familiar, the boundaries among them are not sharp. Compilers and catalogers write dictionaries, bibliographies, nomenclatural compendia, etc., with e goal of accounting for everything available on a particular topic. The approach and mindset of a dedicated compiler was described by K. M. Elis- abeth Murray (1977) in Caught in the Web of Words, a biography of her grandfather, James Murray, whose passion was words and who orga- nized and compiled the Oxford English Dictio- nary. 2. Integraters produce expanded catalogs that are edited to conform to a point of view, e.g., encyclopedia that carries a particular view of m kind, or a regional or continental flora that implies a certain uniformity of taxonomic interpretation. Integration involves scientific judgment on the part of the integraters. A lucid but lengthy apologia for integration was presented by Jean Le Rond d'Alem- bert in 1751 as an introductory essay to the mag- num opus of the French Philosophes, the Ency- clopédie of Denis Diderot (d'Alembert, 1963). 3. Digesters selectively seek data and ideas from the literature and derive a comprehensive treat- ment to tell a story or instruct novices. Digesters write textbooks and state-of-the-art compendia. The aforementioned works of Stuessy and Davis & Hey- wood are examples of the works of digesters. 4. Assimilators are those who compile, inte- grate, and digest the works at hand. Assimilators produce expanded insights and new conceptual schemes. Among the great examples are the works of James Hutton and Charles Lyell, who assimilated the available information into the theory of Uni- formitarianism and created modern historical ge- ology. The Russian chemist D. Mendeleev likewise created the periodic table of the atoms through And so, too, with Charles Darwin and Alfred Russell Wallace, inter alia, who insightful assimilation. made the theory of evolution scientifically respect- able. А notable current venture of assimilation in systematic botany is the preparation of a refined general system of classification to account for the present level of RU deam in a justifiable phy- logenetic scheme. Those include Armen Takhtajan (1980. 1991) in St. Pe- tersburg, the late Arthur Cronquist (1981, 1988) in New York, and Robert Thorne (1976, 1983) in Claremont. Another contemporary assimilator engaged in this project of considerable influence is Lynn Margulis, who has developed the notion that the eukaryotic cell is derived from a prokaryotic ancestor that co- opted other prokaryotes, which, in turn, evolved into the intra-cellular organelles that are typical of the eukaryotes (Margulis, 1981). The elucidation of the genetic code rested upon some elegant ex- perimental research, but the “central dogma" of molecular biology is an example of assimilation of data from disparate sources (Watson, Some great works have elements of all four of these items, e.g., Die natürlichen Pflanzenfami- lien of A. Engler & K. Prantl (1887-1915) is Volume 80, Number 2 1993 Barkley Synthesis 295 largely an integration and digest of what was known, but the approach had an element of assimilation in that the editorial policy sought to use the work as a platform to support a particular understanding of plant evolution. In recent years, our colleagues in ecology have found it necessary to synthesize their information into a whole that exceeds the sum of the parts, and they have done so by engaging mathematical concepts and computer technology. The resultant efforts аге a subspecies of assimilation called **mod- епа," by which means it is hoped to create a precise depiction of the dynamics of the ecosystem or some part of it. A MODEL FOR SYNTHESIS Rather little is written on the methods of syn- thesizing great amounts of information, or at least little that is current among us biologists. Our col- leagues the historians, however, might give us a clue from their methodology. Historians work on a subject for which there is neither experimental nor observational research, by our reckoning. Rather, historians are in the business of organizing and synthesizing diverse sorts of information about people, places, and events, and then composing possible conclusions. Historians note what has hap- pened and then suggest interpretations as to how and why. Histories are written with different ends in mind, and distinctions among historical. ap- proaches are as old as written history. Herodo- tus, “the father of history," told the story of the Athenians' rebuff of the Persians so as to make it a timeless event of cosmic significance, but in the telling, he allowed inaccuracies of fact and some fanciful interpretations. Thucydides, on the other hand, recounted the Peloponnesian Wars with pre- cise accuracy, but with ggestion of a relevance that transcended the affairs of Greece. Needless to note, both historians have been read and respected through the ages The craft of writing history is called "histori- ography" and is often taught in history depart- ments as a class for beginning graduate students. There are textbooks and workbooks in historiog- raphy, often with titles suggesting methods or tech- niques (e.g., Stoffle & Karter, 1979). Much effort is given to locating and evaluating sources of his- torical information, and some effort to the matter of integration of information. Historiographers usu- ally assume an "angle" or a point of view about what is being studied, e.g., a Marxist historian could interpret the Reformation as a struggle among the social classes, whereas a theological worldview woul offer a different explanation of events. Алаган of the point of view is termed * historians. ‘revisionism’ e biologists already share some procedures with historians, for we draw up hypotheses and test them, and we have our own ideologies and methods that are subject to re-evaluation. The point, how- ever, is that historiographers have made a profes- sion of learning to compile and evaluate information from the written record, and we biologists may certainly look to our colleagues in history for guid- ance and methodologies. A MODEST PROPOSAL The time has clearly come to seek the “Broader Synthesis” from the information that is accumu- lating in the libraries. Most of us are established as practitioners in our professions; we are biologists or librarians or information manipulators, etc., and it is unlikely that many of us are seeking new avenues for our lives. However, please consider the following: 1. As noted earlier, we biologists believe that the products of our research have a value that transcends the miserly accumulation of facts in a library, i.e., our research collectively "means something.” We do not often ask to whom it means something, or how it is meaningful. We simply assume that our research is part of a grander scheme, if we could but make the connections. By extension, the proverbial “answer to cancer" is likely to be elucidated by a synthesizer of infor- mation sitting at a computer terminal, rather than by a wet-bench researcher. 2. We are the first generation to see the op- portunities of widely available electronic syntheses of information. The advent of printing changed the ways we handled information, and now the advent о => r technology is assuring new е^ Вата. bags in the handling of infor- mation. The techniques and mindsets of compilation and synthesis are hardly new— historians have been compiling and synthesizing for centuries. us, we have the motive, the means, and the opportunity to tackle the treasure trove of infor- mation in our libraries. І propose Let us prepare a new subspecies of biologist, whose primary domain is not the field or laboratory (we have done very well at training field-and-lab researchers), but rather whose activity is mining the data in hand, and who is versed in the tech- niques of historiography as well as the approaches of traditional, observational, and experimental bi- 296 Annals of the Missouri Botanical Garden 5 ology. The new biologist would not be a "historian," i.e., not a student of the history of science, but rather a biologist who addresses the Big Picture, asking new questions and testing hypotheses with the abundant and diverse information in hand. Think of the kinds of topics that such biologists could address, e.g., (a) changes in the structures of the biotic communities of North America since the melting of the Wisconsin ice sheet, (b) the impact of vegetation dynamics on air pollution and climate change, (c) an integration of paleobotanical evi- dence into the general system of angiosperm tax- onomy. | Approaching these synthetic studies would ге- quire being well versed in all of the resources, plus having linguistic ability in more than English. Like- wise, our newly trained synthesizer would require competence in electronic data management, where, | am told, the capabilities of the hardware far exceed the uses to which it is presently put. Can we train a generation of biologists to regard the libraries as their sources of information? We will surely do so. But how long will we still have stick-in-the-muds, who assume that measuring a plant in the field or pouring from a test tube in the lab is worth more than asking new questions of data already in hand? As we plan for our new subspecies of biologists, we would do well to co-opt the librarians early ir the scheming. I imagine that they should be рге- pared for an eager group of biologists who have recently discovered The Library and who will treat it as we now treat our laboratories. What will make Synthetic Biology a viable field for the future? Certainly two things: (1) some tan- gible resources to entice academia to engage these people and (2) thoughtful consciousness-raising among biologists to convince our colleagues that The Time Has Come. But the ultimate and com- pelling reason to study the accumulated body of information is simply because we have built it, and it is there. NOTE arles Darwin, Herodotus, James Hutton, Charles Lyell D. Mendeleev, Thucydides, and Alfred Russel Wal- lace are noted in the text, but specific citations are not given here. All are oe figures, whose works are i whom the abundant secondary literature is most о ай The connections among Dar- win, Hutton, Lyell, and Wallace are well noted by Eiseley (1958). LITERATURE CITED Boorstin, D. 1983. The Discoverers. Random House, New York CHAYTOR, H. J. 1945. From Script to Print. W. Heffer & Sons, Cambridge. [Reprint, 1950. CHRONICLE OF HIGHER EDUCATION. March 2 . Fact File: Seay? of research TIME in U. s. and Canada, 1989-1990. CRONQUIST, А. Ha Integrated System of Classi- fication of Flowering Plants. Columbia Univ. Press, New York. . 1988. The Evolution and Classification. of Flowering Plants. New York Botanical Garden, New ork. D'ALEMBERT, J. L. 1963. Preliminary Discourse to the ышара of Diderot, translated by R. N. Schwab. Library of the Liberal Arts, Bobbs-Merrill, Indianap- olls. Davis, P. H. & V. H. HEvwoop. е Тахопошу. 1963. Principles of Van Nostrand, Princeton, = ee E 949 Biogeography of the Pleistocene. Bull. € Cul Soc. America 60: 1315-1416 uo L. m 5 o Doubleday, New = x c е9 [es e © ж e zm ip by A ENGLER, А. PRANTL (editore) natürlichen Pflanzenfamilien. Wilhelm Engelmann, Leipzig MARGULIS, L. 1981. Symbiosis in Cell Evolution. W. Freeman, San Francisco. [Margulis has an ex- tensive bibliography on the topic; this book is an introduction.] Murray, K. M. E. 1977. Caught in the Web of Words. Yale, New Haven. [Paperback by Oxford Univ. Press, 1979 OKERSON, А. & К. STIBBS. 199]. The library “Dooms- day A ” Publisher's Weekly, February 8, pp. 36- STOFFLE, | & S. KARTER. 1979. Materials and Methods for History Research. Libraryworks, Neal-Schuman, w York. STUESSY, T. F. Plant Taxonomy, the Systematic Evaluation of Comparative Data. Columbia Univ. Press, New Yor TAKHTAJAN, А. Outline of the n of o Plants (Magnoliophyta). Bot. Rev. 46: 225 1991. Evolutionary Trends in Flowering Plants. Columbia Univ. = New York. THORNE, R. F. 1976 o been c em of the эле. Evol. Biol. 9: 35-106. 19 Proposed a ТЯ in the An- giosperms. pee J. Bot. Watson, J. 968. The Double Helix. New et Atheneum, THE FUTURE OF Arlene E. Luchsinger? SCIENTIFIC COMMUNICATIONS: MORE QUESTIONS THAN ANSWERS! ABSTRACT The paradigm of scientific literature has been the paper journal for the past 300 years. An economical design is needed for future information acquisition and storage, and the electronic journal may be the model for the future. Four issues seem rive the production and disse ee of future scientific literature: the development of an se to acceptable standardized баша the copyright of intellectual property; the use and preservation of the literature; and i i s electronic journals are characterized as being from the not-for-profit sector, originating with faculty members in universities. They are text only, they are not peer and they are relatively unknown. Electronic Е will not supplant paper journals in the near future, but we are in the early stages of enun. communica For more than 300 years, scientific journals From ancient clay tablets to manuscripts to have been the medium for communicating new printed books and journals, microforms, and com- ideas and a stable means to chronicle scientific puter files, libraries over the centuries have ob- achievement. Before the first journal, correspon- tained, preserved, and organized these sources of dence to other intellectuals provided a method to information. Before the end of this century, another distribute speculations, theories, and philosophies. format will be common in most libraries: infor- The exchange of these letters was encouraged by mation of an electronic design. the formation of official scientific academies, and Library subscriptions to journals involve a com- the letters were read to the members who attended mitment to continue their purchase and preser- the meetings. The secretary of the society often vation on an annual basis. Debate and concern has assumed the role of soliciting, receiving, and re- been growing over the escalating cost of scientific sponding to the letters of the correspondents. The — journals and the increasing number of new journals. secretary was also responsible for advising those This expansion of journal literature has provided who could not attend the meetings. The monoto- not only more titles than ever before, but more nous hand copying of the news and ideas for each specialized titles in every discipline. The prolifer- of the absent correspondents was laborious and ation of journals (the number doubles every 10 to time-consuming (Kronick, 1976). 15 years) and the shrinking acquisitions budgets The printing press, Gutenberg's contribution in of most academic libraries will force the focus of the mid-15th century, was one of the greatest academic libraries from acquiring resources for innovations of history. The printing press had been teaching and research to providing access to in- most successful in producing books and was used formation (Broad, 1988; Smith, 1991). in 1665 to produce the first scientific. journal, the Librarians need to make decisions about how to Philosophical Transactions of the Royal Society utilize their resources. Should they acquire paper of London. As scholarly correspondence increased, copies of the information, or provide access to the character of the scientific journal changed to information using electronic means? Will the ability include original research, reviews of research, book — to access other collections be more expensive than reviews, and news reports. The letters have con- continuing to subscribe to the paper copies? Will tinued as letters to the editor (Chappell, 1970; libraries rely on cooperative borrowing, or shift to Kronick, 1976). purchasing journal articles from commercial ' Discussions with William Gray Potter contributed to my омо of and interest in this evolving field. I thank my colleague, Virginia Benjamin, who read the manuscript a s elpful comments. cience Library, University of Georgia, Athens, Georgia 30602, ANN. Missouni Bor. Garp. 80: 297-303. 1993. 298 Annals of the Missouri Botanical Garden sources? Several commercial providers exist now, and more are projecting entry into the marketplace (Jackson, 1992; Lenzini & Shaw, 1992). What level of staff interface will be needed for the various modes of access? The cost for delivery of an article may depend on the urgency of the requestor and the mode of delivery. That is, articles may be sent by mail, overnight courier, electronic mail, or fax. Who will decide whether a library user needs in- formation immediately or if documents can be de- livered after a brief wait by conventional interli- brary loan? Some librarians believe that the institutions that can provide ready access to nec- essary materials will recruit the brightest faculty and students; those that c Кон "ud this access | & Saunders, will lose the competiti 1991). Four issues seem to be driving the production and dissemination of future scientific. literature: format, the copyright of intellectual property, how the literature is used and preserved, and funding for the access to the literature. NEW APPROACH TO PACKAGING INFORMATION The current paradigm of scientific literature is the paper journal. To change this paradigm, we must approach the current publication dilemmas with innovative mental restructuring, fresh ap- proaches, and renewed imaginations. Because one impetus for change is the increasing cost of paper journals, the change to electronic formats will not happen if the costs are more than the users are willing to pay Electronic publishing may be defined as pub- lishing that encompasses all computer-assisted methods for the production, manipulation, and dis- semination of information (Brailsford & Beach, 1989). Electronic publishing encompasses on-de- mand publishing, bibliographic databases, desktop publishing, documents on optical disks, etc., but the electronic scientific journal is a major concern to the academic community. What is the future of the electronic journal? What are the technical problems of production? What are the implications for libraries? How will access to the electronic journal be provided? The characteristics of today's electronic journal are outlined by Okerson (1990). These journals are from the not-for-profit sector; they originate with faculty members in universities; they are text only; they are free or inexpensive; they have gen- erous copying statements; the circulation or sub- scription lists are rapidly growing; and they are relatively unknown. A recent publication, Direc- tory of Electronic Journals and Newsletters (Strangelove & Kovacs, 1991), lists 30 journals and more than 60 newsletters that are available over electronic networks. More typical of electronic information imple- mentations are electronic indexes such as Biolog- ical Abstracts and Agricola, which correspond to their paper counterparts. Most college and uni- versity libraries have or are planning an on-line catalog of their holdings with author, title, subject, and keyword searching (Burke, 1991; Sitts, 1988). Some libraries have installed journal indexes such as Medline and Current Contents that are accessed through the on-line catalog. These electronic on- line indexes are available not only to users in the library but also to remote users through dial-up access. On-line journals exist now that are versions of the same title published in paper. The American Chemical Society in the early 1980s made its pri- mary journals available on-line. They are accessible through the on-line database service, STN. Elec- tronic journals could conceivably be archived on a mainframe computer and linked to on-line cat- gs. Rather than subscribing to an entire journal, a profile of a consumer’s subject interest could be matched electronically against articles in a wide range of journals. Much more information would be made available than if one subscribed only to those journals that could be supported financially; the cost would be for the articles that were selected. The consumer could be the individual scientist, or a library that serves a scientific community (Lan- caster, | ere are some obstacles to the use of electronic media or formats. Illustrations are not a part of current electronic communications. Many scien- tists lack or do not want to use electronic com- munications. The medium may appear to be dif- ficult and inaccessible. Most people are accustomed to a paper format for journals, and paper products are readily available through conventional and un- complicated means (Loken et al., 1991; Okerson, 1990 Standards are lacking for the format used to convey electronic data. These standards include, but are not limited to, the coding for specific in- structions to the computer such as what designates underlining, italics, or paragraph spacing. Stan- dards are absent for encoding and distributing texts and accompanying images. Standardization of sym- bols and graphics and the rules for their use must Volume 80, Number 2 Luchsinger g 299 Future of Scientific Communications be developed, and there must be interfaces for text and images (Gould & Pearce, 1991; Weber, 1990). he problem of moving textual material among different types of computers with incompatible op- erating systems and differing applications software was solved by the American Standard Code for Information Interchange (ASCII). ASCII is a binary code used in all personal computers. However, ASCII lacks the codes for such things as editorial add worth to the idea of electronic publications. ASCII text files prevent the use of different fonts such as foreign character sets, mathematical sym- bols and scientific notation, illustrations, and color. There are other text files t i problem, but may not be practical for users who at might solve the do not have the necessary software to use those files (Bailey, 1991; Basch, 1991; Kaufman & LeClercq, 1990). Today's scientific paper publications have high- quality color illustrations. We do not know the limitations of monitor and printer displays, but to be successful, microcomputer hardware must be inexpensive and improved so that the quality of the data reception on screens and printers is high and is as acceptable as paper copy (Bailey, 1991; Weber, 1990). Another challenge is the establishment of high- speed research networks and their integration into a national networking infrastructure. These net- works must have the speed and bandwidth to ac- commodate an increasing data transmission re- quirement. Legislation to establish the National Research and Education Network (NREN) was first intro- duced by Senator Albert Gore, Jr., in 1988. year since then, revised versions of this act have been introduced in both the U.S. House and Senate. Each The proposal is to connect supercomputer centers and to accommodate the massive amounts of data produced by high-performance computer projects. REN would expand and improve the existing interconnected group of research networks such as NSFNET, MIDNET, and SURANET that are col- lectively known as Internet. It is expected to be completed by the year 2000. Federal agencies will be expected to work with state and local agencies, libraries, educational institutions, and private net- work services to provide access to electronic in- formation resources held by publishers, research facilities, and libraries. The creation of the EN will expedite the publication of electronic journals (Henderson, 1991; Brownrigg, 1990) COPYRIGHT With the invention of the photocopier came the concern that these machines would put some jour- nals out of business. The photocopier removed a section of readers who might have subscribed to a journal if that was the only way they could have acquired the papers they needed. The publishers normally do not receive revenue from these readers (Brailsford & Beach, 1989 Copyright is not intended as a permanent mo- nopoly, but software vendors have been known to use technical strategies to thwart attempts to copy a program into a user's computer memory (Pool, 1985) Photocopies may be made of today's paper jour- nals and mailed to a colleague or borrowing library. Fax machines may be used to transport that same copy faster. An article in digital form may be sent from a computer workstation and arrive at one or more distant workstations almost instantaneously. What will keep the receivers of the document from redistributing the material (Weber, 1990)? Sharing or even reselling books has never been prohibited, and such prohibition would not be en- forceable. Colleagues may read information from your computer terminal or ask for a copy of a copyrighted electronic publication you have or- dered, paid for, read, and stored in your computer's memory. It is easy to do this; press a couple of keys, and the information is sent to an electronic mailbox. If you comply, you may be disregarding the copyright laws (Pool, 1985; Solomon, 1985). ¿nforcement of copyright laws from the com- mercial publisher's view is ап economic, not an intellectual, issue. If copyright cannot be enforced, that is, if the reader does not have to pay to read an electronic document, there would be little in- centive for a commercial publisher to distribute merchandise in an electronic format. Will publish- ers permit storage of their electronic publications in a central site? If copyright is not necessary for commercial publishers to collect revenues, and if a charge is collected for each use of the electronic publications, will copyright be obsolete (Smith, 1991)? UsING AND PRESERVING ELECTRONIC LITERATURE According to Dillon (1991), scholarly journals are read for several reasons: for work purposes, to keep up with the literature, for reference, or for learning. Her study indicated that readers of ac- ademic journals scanned the tables of contents first, 300 Annals of the Missouri Botanical Garden and then rapidly browsed and turned pages quickly back and forth through an article. Electronic ver- sions must have the capability to scroll through pages as well as skip forward and backward through a document. Users of electronic journals will want to perform tasks other than full reading of a text. Paper is more adaptable than a video screen; the white paper and black ink format will not transfer easily to screens. There is a rush by the scientific community to get information into print. Today the lag time be- tween submitting а paper and seeing it in print is usually from 4 to 18 months. Full-text electronic journals could be printed in any location and for- warded to users, thus dramatically shortening the time lag. Informal use of electronic mail (e-mail) systems is increasing in academia (Loken et al., 91; Smith, 1991). Id munication form has met resistance when it threat- eas are exchanged, but this com- ens the more formal channels, that is, the refereed journal (Piternick, 1989). The role of the scientific journal today is to provide a mechanism to notify others of the results of research. Publishing is also done for scholarly prestige and the academic re- ward system. Will electronic publications be ac- ceptable for promotion, tenure, academic honors, author recognition, and increases in salary? If ar- ticles published in peer-reviewed electronic journals are not accepted for promotion or tenure, will re- searchers be kept from fully exploring the possi- bility of publication in the format? Who will be on the editorial enough to handle the manipulation of electronic oard—only those who are skilled text? Will the part of the publication process that is not automated be the editorial work by referees and editors? Will the contributions of scientists who are uneasy about the integrity of scientific data be jeopardized (Eisner, 1991; Gould & Pearce, 1991; Smith, 1991)? Invasion of privacy may be an issue if publishers of the electronic information systems collect data about subscribers’ proprietary interests and sell this information to a third party. Some publishers today sell the subscription lists of their paper publications. Haywood (1988) described the “electronic ar- istocracy.” They consume, enjoy, and digest elec- tronic information in private without having to share with others. They work in an information-rich en- vironment and can afford the up-front connect and search charges, or they work for an organization that pays these charges. These users could also be termed the “information elite." A significant portion of the world's population will be the "information poor." These people will not have access to electronic publications. This will be true not only of third-world scholars, but also those in some developed countries. Until language translation systems are available, English will con- tinue to dominate the databases, and those who ee cannot read English will be "information poor." Inefficient communication and transportation net- works, inadequate telephone and mailing services, and a dearth of technical expertise will challenge the development of electronic information delivery in the third world (Haywood, Libraries subscribe to and maintain journals that only a few people read or consult when first pub- lished. ership in later years as certain papers begin to be Those journals, however, may build read- cited. Preservation of printed volumes, films, pho- tographs, and slides has been a task of librarians. 'The endured through the centuries. paper format has been successful and has The librarians of the future dealing with electronic publications may not have the preservation problem—there simply may not be anything to save. There are several nontrivial obstacles to the preserving of electronic data files. Magnetic media degenerate due to heat, sunlight, humidity, dry- ness, magnetism, dust, and unintentional erasure. How long will an electronic media exist. without self-destructing? Data tapes, hard and floppy disks, and digital optical storage have a lifetime of a few 1990). Who is and who will be doing the archiving of intellectual decades (Kaufman & LeClercq, content of electronic publications? Soon the technology of today may be obsolete. Will we be able to continuously move stored data from one technology to another? Electronic stan- dards will be more important than ever. It will also be essential that archives of electronic serials be maintained by libraries or other organizations (Bai- ey, THE ECONOMICS OF ELECTRONIC PUBLISHING Information acquisition and publishing is becom- ing increasingly centralized. Users of inexpensive packages of information (e.g., books) should not be disenfranchised by the sudden disappearance of information if that information becomes accessible only by electronic means. Users must be able to access the electronic networks at a reasonable cost. Government funding will be needed to bring mod- ern, fast, and relevant information networks to the majority of the population (Bailey, 1991; Hay- wood, 1988). The level of university support for networked Volume 80, Number 2 1993 Luchsinger Future of Scientific Communications publishing ventures is a concern. Endorsement by scholars and publishers is essential. А charge for use could accommodate those for-profit publishers. The noncommercial presses, such as universities and societies, might not be threatened as long as the editorial and review processes remain within their purview (Okerson, 1990; Smith, 1991). Most revenue will come from subscriptions, fees for con- nect time, royalties for viewing and printing pages, charges from users downloading information, and rental of compatible equipment. There will be less revenue, and nearly all of it will be delayed in coming because most use will be “оп demand.’ Advertising income will be lost. Publishers worry about the revenues they will expend and what the profits will be. They need to know how to make money on this technology (Standera, 1987; Eisner, 1991) The publishing process for scholarly print jour- nals is financially underwritten by the scholarly community either from page charges or library subscriptions. Government grants may provide funding for the research for new data. This new information is forwarded to a journal publisher, and frequently grants provide the resources for page charges. State and federal governments provide the money for libraries to purchase the information that was formulated by the researchers Rogers & Hurt (1990) described a scenario for a Scholarly Communication System. Briefly, schol- ars could publish and read other publications on this electronic network. The system could also be used as a communication device to seek data from current and archival files. Those with valid pass- words could leave signed statements related to the articles they have read in a comments field. An article that had been in the system for six months would be flagged for review. The author would have several weeks to use the responses from the com- ments section and prepare a final, revised version he boards would perform the reviews now undertaken by editors of to submit to a review board. scholarly journals. If the article was approved by the board, it would become a permanent part of the communication system. Grants from the participating universities, foun- dations, and government agencies would provide the capital for establishing the system. Annual fees paid by each academic institution would pay for selection and training of reviewers, honoraria for reviewers, and administrative costs. Operation costs and hourly usage charges would be paid by a mem- bership fee from each institution. Authors whose articles were printed out by the system would be provided with royalty payments. Libraries would have access to the system and could print files using sophisticated printers (Rogers & Hurt, 1990). This new communication system would provide numerous advantages to colleges and universities. Тће amount of money spent for subscriptions to publications would be significantly decreased. The need for additional library space would be reduced. The hours spent by library personnel processing, shelving, and reshelving journals could be better spent on helping users find the information they want. The system would establish and enforce the format standards to be used for storing information. Universities would regain control over the review and publication of research. That is, rather than having publishers and their referees decide who gets published, the universities would have a role in choosing the members of review and manage- ment boards. The system thus could provide better information for universities to make hiring, pro- motion, tenure, and salary decisions. In addition universities would regain copyright control (Oker- son, 1991; Rogers & Hurt, 1990) ONE SOCIETY’S VISION In 1988 the president of the American Physical Society (APS) formed a six-person Task Force on Electronic Information Systems. The extensive re- port of the Task Force was published in the Bulletin of the American Physical Society (Loken et al., 1991). While the report concerns the physics com- munity, the findings and visions may apply to all disciplines. Some of the findings were: (1) those who have otherwise avoided the use of computers for re- search are utilizing word-processing on desktop personal computers; (2) many have personal com- puters that are linked to electronic services and on-line databases; (3) the use of electronic mail (e- mail) through BITNET or Internet has increased in frequency and scope; and (4) e-mail is used for short communications, nonlocal pp and submission of papers (Loken et al., ) The APS vision is for the year ds but it may be reached by 2010 or even 2000. Workstations with color-graphic resolution comparable with to- day's best on-paper printing may be used by the majority of scientists. These workstations may be connected and have access to local or remote serv- ers within seconds. A single, worldwide, fiber-optic network with a structure that is analogous to a highway system with superhighways, state and re- and driveways is gional highways, local roads, planned. Storage of information will be 10* less 302 Annals of the Missouri Botanical Garden than the cost of storing documents on paper. Soft- ware standards will be in place to deal with text, equations, graphics, illustrations, and multimedia output (Loken et al., 1991 Computerized scientific literature delivered to graphic terminals may have output that is three- dimensional, dynamic, and that can be viewed in arbitrary orientation, with arbitrary magnification and at variable speeds—all selected by the indi- vidual viewer. The user may be able to request documents, sections of documents, or texts to which the original document refers. The user may be able to make comments on the document. The capacity of computers to perform translation of natural lan- guages will be developed. This will permit an author to submit a paper in one language and have it read in the languages of the readers’ choice. Authors may submit their manuscripts by electronic. sub- mission to publishers using a form that follows an international standard. Messages between referees and editors and between editors and authors may be electronic (Loken et al., 1991). The APS Task Force felt that there could be separate databases supporting broad subjects. The individual databases could form a central database that would be accessible. Managers of the databases would be responsible for the physical operations such as maintaining the hardware, software, and networking, as well as making and implementing the policy decisions of the supporting organization. his organization could be a government or private enterprise (Loken et al., 1991) Search strategies will be efficient and easy to use. Entire documents may be indexed by biblio- graphic information, descriptors, symbols, and per- haps diagrams. Searches will be more interactive; computers will be used as expert systems to lead an investigator through a search. Today, the results of on-line searches are downloaded to paper or floppy diskette. In 2020, downloading will be over the network into computer files (Loken et al., 1991). Most scientists have a set group of journals they read and use as vehicles for publishing their papers. What impact will the electronic publication and retrieval of documents have on the journal in which it is published? Will journals as we know them exist? Today, many scientific journals are only sold to libraries. The cost to the libraries is the same whether the journal has a large or small readership. In 2020, the use of the electronic publication could be monitored. These observations could be used to determine the number, kind, and extent of any use (Loken et al., 1991) CONCLUSION Print journals will probably continue as the pri- mary method of scholarly communication in the near future. However, we are in the beginning stages of the development of electronic commu- nication. Paper will not disappear, and the new media will not threaten the future of the printed book. Electronic journals may be read on-line, but most scientists will want a printed copy. Publishing in both print and electronic formats may be nec- essary, and this will increase the cost to the pub- lisher. Long-term storage will be in electronic for- mats rather than microfilm and microfiche. The promise of successful technology alone will not force the acceptance of electronic publishing. The real concerns may be fiscal, legal, regulatory, social, and political factors rather than technolog- ical issues. However, as computer technology be- comes a central force in scientific dialogue, elec- tronic publishing may supplant the printing press as a comparable force in communication among readers. LITERATURE CITED BAILEY, Electronic (online) publishing in action. Online 150): 28-35. Basch, В. Books online: visions, plans and per- spectives for poems text. Online 15(4): 13-23. BRAILSFORD, D. F. & R. J. Beach. 1989. Flies publishing a journal and its production. Compute J. 32: 482-493. Broap, W. J. E Science can't keep up with flood of new journals. c York Times Feb. 16: C1, C11. BROWNRIGG, E. 990. Developing the information superhighway. Pp. 55-63 in C. A. Parkhurst (editor), OPAC: content TT form. 8 in N. van Pulis (editor), Think Tank on the Present and Future of the Online Catalog: Pro ceedings. Reference and Adult те Division, merican Library Association, Chic CHAPPELL, W. Short History Pt the Printed Word. A. A. Knopf, New York. wae A. 1991. New technology and the reading ocess. A aad i in Libraries 11(6); 23-26. ise R. Publishers work toward startin utable ү science journals. The Scientist 5(5): Соор, C. C. & К. Pearce. 1991. Information Needs in Science: Ап Assessment. The Research Libraries Group, Mountain View, California. Havwoop, T. p eae а the with- ering of public access. Pp. 195-206 in H. Dyer & . Tseng (editors), New А for the Information Profession. Taylor Graham, London. HENDERSON, C. С. 1991. National Research and Ed- ee : си 5.1067 апа НК 3131. Pp. n C. A. Parkhurst (editor), Library Per- Volume 80, Number 2 1993 Luchsinger 303 g Future of Scientific Communications spectives on NREN. Library and Information Tech- nology Association, Chicago. 99 Stand and deliver. Wilson Li- 1990. Archiving он їгопіс journals: who's responsible for what. In mation Issues (University of Tennessee) 201): 1 Квомск, D. A. 1976. A History of Scientific Be Technical Periodicals, 2nd edition. Scarecrow Press, : 316- 325 А 1992. UnCover and nCover2: an article citation database and service featuring scr de m Interlending and Doc- ument о 12- LoKEN, S. C., ET AL. о APS Tack Force on ve mer. Physica : & L. M. асе 1991. The virtual 1990s. Computers in Li- 1991. The report of the nic Information Systems. | ull. A MITCHELL, M. library: an agenda for the braries 11(4): 8-11. OKERSON, A. 1990. Symposium held on electronic jour- nals. ARL 153: 7. 199]. With feathers: effects of copyright and ownership on pee) ponam College and Re- rch Libraries 52: -4 фий, A. B. 1989. eed to find alternatives to the scientific journal: a brief review. J. Academic Librarianship 15: 260-266. Poor, I. S. Whither electronic copyright. Pp. 217-230 in M. Greenberger (editor), Electronic Pub. lishing Plus. Knowledge Industry Publications, White Y g S. HURT. communication should work in the 21st cen College and Research Libraries 51: 5-8. Sitts, M. K. (editor). 1988. The Automation Inventory E н search Libraries. Association of Research Li- ries, Washington бийи, ЈЕ 99 аа the acquisitions dilemma: into the electronic information Gcr ment. College and Research Libraries 52: 23 SOLOMON, R. J. a eh ibi concept of Бани property. Pp. 231-240 in M. Greenber- ger (editor), Electronic Publishing Plus. Knowledge ГА Publications, White Plains, New York. STANDERA, O. The Electronic Era of Publishing. Elsevier, New York. STRANGELOVE, M. 4 D. Kov ос 1991. Di- rectory of Electronic Journals, Newsletters and Ac- ademic Discussion Lists. Association of Research Li- braries, Washington, WEBER 1 Тһе ен на future of oe publishing. Publishers Weekly 237(25): 76-8 1990. How scholarly tury. OWNERSHIP AND OUTREACH: A MODEL FOR ADMINISTRATION OF SHARED DATA Jean E. Nash ABSTRACT own to each other. I however, means for promoting multiple use of data that accommodate researchers’ concerns can be f costs associated with retentio issues include which data should be retained, the the 1 Sharing data is a common practice among researchers in all fields. In most, pubs sharing is limited to data rs kno n this paper, issues ra e ata and a access to data gene n and maintenance, stem are generated. Finally, throughout the paper an yst ument is made for institutional consideration of data FRA ER issues at the policy level prior to development of formal data sharing mechanisms The concept of multiple-use data in scientific research is not a new one; in fact, sharing data is common among researchers. However, the context of sharing data for most researchers is that of Researcher A sending a data disk or a copy of lab results to Researcher B. This form of data sharing appears to raise no eyebrows within our current research culture. It is when a larger scale is pro- posed that support for the concept of shared, mul- tiple-use data falters. At that point, promotion of data sharing seems to run into some perceptual roadblocks that, impede the development of multiple-use data on a large scale. The evolution of the Utah Resource for Genetic and Epidemiologic Research (RGE) has if not addressed, will continue to included the development of a certain focus, albeit fuzzy, to the apprehensions that the idea of multiple use of data generally raises. It is my purpose here to delineate the issues involved and then use the experience of RGE to demonstrate how those ap- prehensions may be addressed. That in turn will serve as the basis from which to generate a model for management of effective data sharing. I begin with two caveats: first, as the experience of RGE is solely with human subjects data, the issues raised and the model outlined will at times be unique to research with human subjects. How- ever, many of the principles involved should be applicable to other areas of biological research. econd, in the interests of simplicity, 1 use the {айл definition of data, which summarizes the lack of consensus on a single definition among researchers and administrators considering data sharing: Data are whatever the researcher desig- nates as such (derived from remarks, U.S. ) Health Service, 1990 OWNERSHIP Perhaps the pivotal concern raised in any dis- cussion of data sharing is that of data ownership. Who owns the data generated in the course of scientific research? As demonstrated by the results of a workshop on “Data Management in Biomedical Research” convened by the Public Health Service in April 1990, consensus on this issue does not exist. Investigators who have келеп data in ` The IS awards a ‘chapel their research regard those data as "theirs PHS stated that “When grant, the data developed by the grantee institution are owned by the grantee institution ...." Insti- tutions that receive grants and support investiga- tors (many of whom wrote the grant proposals in the first place) are caught in the middle. Ownership, in a legal sense, deals with property and one’s rights to use of that property. Ownership ' RGE, School of Medicine, University of Utah, Salt Lake City, Utah 84132, U.S.A ANN. MISSOURI Bor. GARD. 80: 304-308. 1993. Volume 80, Number 2 1993 Nash 305 Administration of Shared Data is important, as Fishbein (1991) pointed out, be- cause “... whoever owns also controls dissemi- nation, especially the timing of dissemination, pub- lication and proprietary rights. Moreover, whoever owns data also has the ultimate ability to control their. preservation and destruction." Basing argument about data ownership on the principle of "works-for-hire," Fishbein agrees with the position taken by the Public Health Service that institutions are the owners of data generated in the course of federally supported research. It is her assertion that “this basic principle regarding ownership of data is [not] open to debate as a legal matter." In her opinion, the problems raised by the question of ownership are, in fact, ethical problems of ap- propriate conduct, rather than legal ones. In legal terms, the Public Health Service and Fishbein are correct about ownership of data. differ, however, with Fishbein on the roots of the actual issues with which we deal. — Issues of what is, or is not, ethical conduct in the sciences are secondary consequences of the important aspect of human behavior just noted: scientists who have generated data regard them as "theirs." It may be a logical argument that, since most university policies and procedures manuals exempt certain products of scholarly work from university own- ership, those products not exempted are indeed owned by the university. Viewing the exempted scholarly works from another perspective, how- ever, a different logic appears. The book or article or review is owned by the author, who generated it. On this, we all can agree. Moving backward, the book or article or review is usually based on data generated by the author. If the product is owned by the person who gen- erated it, how can it be that some of the pieces of the product, generated by the same person, are not "owned" by that person also? In behavioral terms, although not perhaps in strictly legal terms, data and product seem inextricably intertwined. Ownership of one appears to confer ownership of the other. must accommodate this perspective. ny resolution of the issue of ownership OWNERSHIP: RESPONSIBILITY AND STEWARDSHIP Resolution of this disagreement between human and legal realities will take many forms. The co cept of stewardship mentioned by Fishbein is one that maximizes the utility of the data, while min- imizing friction over ownership. No matter who owns data in a strict legal sense, all parties benefit from appropriate management of those data. In policies regarding data being developed at the Uni- versity of Utah, it is proposed that a fundamental relationship to data be that of responsibility. Dif- fering parties in the university community have differing responsibilities to each other and to the data generated there. As examples: The university has a responsibility to protect the research interests of all members of its community. To do so, it may be necessary to require that an investigator who moves to another university take responsibility to ensure that data generated by his/her team are available to all members of the original research team. Or: For a researcher to enjoy the benefit of the university's responsibility to deal appropriately with allegations of research misconduct, the re- searcher must take responsibility to maintain data in a form that facilitates resolution of those ale- gations. Focus on the various responsibilities for data acknowledges their value for both individual and collaborative work and encourages productive use of data while protecting the interests of all parties. Final resolution of the issues of ownership will probably occur with the evolution of the sci- entific culture. During that process, however, spe- cific institutional policies regarding data provide mechanisms by which work can proceed without undue constraint. However, as noted in the 1989 [114 Institute of Medicine report, “The Responsible Conduct of Research in the Health Sciences," “few academic institutions have established explicit stan- dards for responsible research practices .. Without such policies, Fishbein's conclusion ibas 7... af the academy does not come to a consen- sus— preferably in some formal way—about the ethical and professional obligations of researchers in stewardship of their research data, the courts or the government may do it for them" is unfor- tunately logical and predictable. ACCESS Тће other major category of concerns usually raised about data sharing is access. In order for data to be shared (that is, used by someone other than the generating researcher) they must be re- tained, maintained and accessible. The volume of research data, enormous in the biomedical arena alone, limits the feasibility of this. Even in large, well-funded programs that rely on sharing of data, such as the Human Genome Project, problems of size and access have arisen (Aldous, 1991). We will have to make some difficult decisions as to which data are kept for shared use. Prospective utility of data will have to be evaluated relative to the costs of retention, maintenance, and accessi- 306 Annals of the Missouri Botanical Garden bility, preferably on an institutional level, so that resources may be allocated in the most effective manner. What is effective will vary from institution to institution, and from one college or department to another within a given institution. If we wish to encourage sharing of data, problems associated with retention, maintenance, and accessability must be identified so that practical solutions can be found. ACCESS: RETENTION Which data justify retention? At the 1990 PHS data management workshop, substantial agreement was reached on retention of certain types of data. Тће first type of data were those that were expen- sive to generate. Epidemiological data were cited in this category, as were data from longitudinal studies. The sheer scale of most such research makes the data generated very expensive. Making such data available for multiple use reduces the cost of future work in those areas, as well as in areas not yet foreseen. The second kind of data to be retained were those that are no longer replicable (or that would be extremely difficult to replicate). The major example noted here was studies on spe- cies that are no longer available for research pur- poses. The final category of data to be kept, and the one for which there was much less agreement than the above two, was data that had significant public policy implications. Keeping these data en- sures that the foundations of the policies are as open to review as the policies themselves. Within a given institution, other types of data may be deemed worthy of retention; these are mentioned here only as examples of some data retention rec- ommendations. The benefits of retaining certain data for mul- tiple use can be agreed upon. There are, however, other factors in the decision to make data available for multiple use. Chief among these are the costs of retaining / maintaining the data and costs asso- ciated with providing access to them. ACCESS: STORAGE COSTS The first set of costs associated with data sharing are those incurred in storing data. Data may be stored as hard copy (such as lab notebooks, X-rays, photographs, printouts) or electronically. Either form of storage entails space costs. Space, on com- puters or in filing cabinets, is a valuable commodity in most institutions. Its costs may be onetime, such as purchase of filing cabinets, or ongoing, as in monthly charges for electronic data storage. In the case of most epidemiological data, for example, size is a significant factor in space calculations. Furthermore, how data are stored may be a factor in accessibility. For example, data on a tape in archives are not as accessible as data present on- line and may thus not be as valuable to another researcher. Minimizing storage costs may also min- imize utility, thus subverting the purpose of storing the ACCESS: CONFIDENTIALITY A second element in decisions about the cost of some data retention is confidentiality. Maintenance of human subjects data requires strict protections for the privacy of the subjects and the data about them. Locking file cabinets, appropriate computer security procedures, and staff training in appro- priate data handling methods are necessary. Per- sonnel and personnel training are necessary ap- propriate maintenance of confidential data. Costs are both onetime and ongoing. ACCESS: CATALOGING COSTS Another major factor in the costs of maintaining data for multiple use is cataloging. А prospective data user needs to know exactly what the data are, where they reside, in what format they are avail- able, and what the access procedures are. While oversight by the generating researcher is absolutely necessary to the cataloging process, creation and upkeep of data descriptions is not an appropriate use of researchers! time. Personnel becomes a sig- nificant cost determinant here. ACCESS: MANAGEMENT The final area of questions raised about access is who will have access to the data and for what purposes. At this point, the human logic about Researchers have con- their" data. ownership is significant. cerns about what others will do with * Will it be used to further scientific inquiry or per- haps for purposes of discrediting the generating researcher? Will it be used to do work the re- searcher has planned to do? Will it be used in research that is not sound? These are legitimate concerns. Allowing the generating researcher to control access to his/her data resolves them. How- ever, requiring a scientist to manage use of the types of data identified as valuable for multiple use would add an intolerable burden of administrative work. Appropriate management of access requires development and administration of mechanisms that allow reasonable control by generating researchers yet promote maximum use by others. Аз in reten- tion/ maintenance, costs here are associated with personnel and the media in which data are stored. Volume 80, Number 2 1993 Nash 307 Administration of Shared Data Costs of data retention, maintenance and access for multiple use clearly exceed the capacity of single researchers or even of most research groups. Тћиз, in order to promote data sharing in the biological sciences, the costs must be accepted on an institutional level. RGE is an example of one such institutionally supported data sharing pro- gram. History or RGE RGE is a data administration program supported at the University of Utah. Its primary responsibility is a large electronic database of records on past and present residents of the state, which is used in genetic, epidemiologic, and demographic re- search. It began the the mid-1970s when Univer- sity of Utah researchers interested in genetic ep- idemiology obtained family history records from the Genealogical Society of Utah and received per- mission from the Church of Jesus Christ of Latter- Day Saints (LDS or Mormon) to compile family history records (three-generation genealogies) into a computerized genealogical database. At the same time, the growing group of researchers began to computerize death certificates from the Utah De- partment of Health and cancer records from the Utah Cancer Registry. Both of those data sets, along with the 1880 U.S. census of Utah, were linked to the genealogical records. By the early 1980s, research uses of these data were extremely varied: cancer epidemiology, historical demogra- phy, cancer genetics, genetics of autism, epide- miological research into the effects of nuclear fall- out, and genetics of early coronary disease were the major areas of research. The management needs occasioned by the diversity of research projects using an aggregate of private, public and research data were the impetus behind the creation of RGE. RGE OPERATIONS Perhaps the single most important facet of RGE is that it functions as an intermediary between data ** contributors and data users. RGE “owns” neither data nor research. Data contributors own their data and negotiate agreements with RGE regarding how those data may be used. Data users own their research projects and work with RGE to complete projects within the constraints of the data agree- ments. RGE has custody of the data and the re- sponsibility to ensure that the agreements are up- eld. With this structure, many of the ownership is- sues that impede data sharing fade from impor- issues of tance. Dissemination and publication, ownership identified by Fishbein, are but insignif- icantly different from other research practices. ile publications using RGE-held data are re- viewed, the reviews are limited to ensuring that certain agreed-upon protocols are used. Control over preservation and/or destruction of new in- formation (data) created by the data user is retained by the user. Control over preservation and/or de- struction of data created by the data contributor is, as defined in the agreements described above, retained by the contributor. The issue of the constraints on the use of data placed by the data contributor is a more thorny one. Where data ownership is absolutely clear, as with the LDS Church family history records, the data contributor may impose whatever restraints on usage desired. Publicly held records, such as those from the Utah Health Department, are sub- ject to state law and regulation, with which few would argue. Our experience with data generated solely by researchers and then given to RGE cus- tody is limited. That limited experience leads me to propose, however, that if agreement can be reached on the need for retention of certain data, agreement can be reached on appropriate use, as well as on protections for the data contributor. However, such agreements are considerably easier to reach if the institution has already defined its policies regarding management of data. In those circumstances, the outlines of a culture of multiple- use data have already been traced. The data within RGE’s custody fit the criteria for retention described above—they are data that would be very difficult and expensive to duplicate. Support for storage and maintenance of these data includes an administrator, a database manager, and a dedicated computer with both hardware and software support. User fees are charged and used to augment institutional support. Past experience лав shown that reliance solely on research grants, without providing a "floor" of institutional support, allows data to deteriorate. The funding fluctuations inherent to American research grants result in sup- port that can be insufficient to maintain data qual- In addition, institutional support, institutional memory about ity, particularly over time. without the data can easily be lost. In both cases, the utility of the data diminishes Confidentiality of data and management of ac- cess are the responsibilities of the RGE adminis- trator. Confidentiality protections are central to the access procedures; both electronic and personnel protection measures must be defined. Extensive guidelines about use requirements are provided to all researchers prior to application for access. Be- fore access to data is granted, a research protocol specifying confidentiality protections must be ap- 308 Annals of the Missouri Botanical Garden proved. All users must sign confidentiality pledges accepting their responsibility to protect the confi- dentiality of the data. These protocols are required in addition to the more general protocols required by Institutional Review Boards. IRB approval is required for any projects requiring access to in- dividually identifying information. Applications for access are reviewed by a com- mittee that includes representatives of data con- tributors, computer personnel, and other scientists in the field. The review is not a peer review of the proposed science, but a review the procedures for management of confidentiality and specific data. As most users are familiar with the data and the procedures, the review is not a tortuous process. The RGE administrator is responsible for working with prospective users to acquaint them sufficiently with the smoothly. The data manager also works with pro- spective users to ensure that proposals that are submitted can be accomplished with RGE-held data. procedures so that reviews proceec COMPONENTS OF A MULTIPLE-USE DATA MANAGEMENT PROGRAM Over the past decade or so, RGE has developed to provide a group of users with a shared data management program. Although the user group is not large, it is diverse. Applying RGE's experience to the apprehensions outlined earlier gives a picture of a multiple-use data management system ground- ed in the following components: 1. Some characteristic of the data makes them valuable yet not easily available elsewhere. 2. Ownership is managed in such a way as to protect the interests of all parties. Written agreements, review authority, and appeal pro- ceedings are integral parts of the system. 3. Detailed guidelines, procedures, etc., are de- veloped by data contributors and data users in concert. These are provided to any researcher prior to application, along with consultation with staff who are familiar with the data and the guidelines. 4. Administration is handled by a non-research third party, whose function is defined as ac- commodating the constraints of data contrib- utors while promoting scientific use of the data. 5. An adequate “floor” of support is guaranteed so that data do not deteriorate, information about the data is retained and available, and there is sufficient staff support to data contrib- utors and data users. While development of a multiple-use data man- agement program will likely proceed more smoothly in an institution with already defined data man- agement policies and practices, such definition is not a requirement in the creation of a data sharing program. SUMMARY The particular constellation into which these components are grouped will be determined by the data, its contributors, users and research environ- ment. RGE is one such constellation that works at the University of Utah. Evolving from the specific to the general, operation of RGE has served as a basis from which institutional policies regarding data are being developed at Utah. Creating a locally effective constellation of the above components will proceed more smoothly if such policies are devel- oped prior to designing any shared data manage- ment program. The combination of institutional ata policies and a well-designed data management program provides a strong basis for the promotion of sharing of data in the biological sciences. LITERATURE CITED ALDOUS, P. Human genome databases at the crossroads. Mans 352(11 July): 94. FISHBEIN, 9 Piu cd E research data. Aca- demic Medicine 66(3): 12 33. INSTITUTE OF MEDICINE. lesa. a Responsible Con- duct of Research in the Health Sciences. National . 1990. Data b in Biomedical Research. Report of a W EMERGING TECHNOLOGY FOR BIOLOGICAL DATA COLLECTION AND ANALYSIS! о Stanley А. Morain? ABSTRACT Increasing concern for Earth's biological resources — their inventory, ра ад ө изе ures for botanical data collection, storage, and г gram, the International Се eosphere/Biosphere Program, to modernize practices and proce Research Pro multinational efforts are already requiri rements for automatin ng massive species databases and бакчы ses of plan of which i is either not readily available, or locked in herbaria on specim e to se a new and entering technologies that can reduce t —has prompted efforts val. America's Global Change the ted Space Year, and related nt species information, much n sheets. One of the first orders of business e time, expense, massive, computerized ior ier тш sets. А p ‘of these prospects are described here in conte on the flora of the Rocky Mounta timizing field data hens and analyzing of my research It is estimated that Earth’s flora and fauna rep- resent somewhere between 10 and 30 million life- forms, of which a scant 1.7 million have been named scientifically (Wilson, 1989). Human im- pact on the global environment is rapidly altering and destroying whole ecosystems, and untold num- bers of species are becoming extinct before they have been discovered, named, and assessed for their beneficial properties. There is an on-going demand to complete a worldwide species inventory, but this cannot be realized without using new and emerging technologies. In the early 1980s it was hoped that new computer technology would permit scientists access to data and images from herbarium specimens, allowing them to compare and identify the large number of new collections being made. Our vision of what that technology can do for botanical and taxonomic research has vastly broad- ened as the technology has become more sophis- ticated and more generally available. In this report, I review the potential for new taxonomic data collection and processing tech- niques, ranging from electronic notepads and mi- crominiature, digitally compatible cameras to da- tabase development and management systems applicable to herbarium inventories and research laboratories. For example, an on-line image data- base that permitted botanists to retrieve pictures of type specimens, rotate them in 3-D, and zoom in on key structural elements would significantly reduce the time and personnel required to manage herbarium collections. Field camps and herbaria equipped with a portable imaging system could communicate with a centralized image database to compare specimens located in other collections, or to those freshly collected in the field. Finally, the availability of interactive databases designed for specimen attributes, landscape analysis, and hu- man imprints contributing to landscape changes would permit a new genre of research hypotheses and modeling. Already in the 1990s, significant strides have been made in imaging, database man- agement, telecommunications, miniaturization, global positioning, and computing technology that will allow rapid modernization in field and herbar- ium practices. Moreover, there are major new ca- pabilities evolving in the general tool kit of Earth- observing technologies that will enhance not only the collection of taxonomic, ecophysiological, and distributional data about species, but will also re- duce the time and expense of analyzing the data and disseminating the information. The gional laboratory for adopting new data collection Rocky Mountains represent an exciting re- and management technology for plant species in- formation, and will be used here to illustrate the potential. The origin and development of the Rocky Mountain flora throughout the Tertiary Period, the shifting and sifting of elements and communities during the glacial Quaternary Period, and the man- ifest changes and imprints superimposed by human cultural systems in the past millennium (especially in the past half-century) all contribute to our un- derstanding of the area as an ancient but constantly ' This research was sponsored in part by the National Aeronautics and Space Administration, Office of Commercial 19 Programs, Technology Utilization Division, under contract ASw-4 * Technology Application Center, University of New Mexico, Albuquerque, New Mexico 87131-6031, U.S.A. ANN. Missourt Bor. Garp. 80: 309-316. 1993. 310 Annals of the Missouri Botanical Garden changing and increasingly vulnerable set of envi- ronments. The region stretches for more than 25? latitude, from northern New Mexico to the Laird River in British Columbia, and rises at its highest peaks to over 4,000 m from a general elevational base of 1,600 m Rocky MOUNTAIN ЮАТА SETS Activities are underway that enhance the Rock- ies as a technological test bed. The Flora of North America (FNA) Project (Morin et al., 1989), the revision of the Flora of the Rocky Mountains, the Greater Yellowstone Ecosystems Project, and the Global Environmental Monitoring System/ ‘Global Resources Information Database (GEMS/GRID) are all efforts aimed at the development of plant species databases. Likewise, they have all shown a compelling need to address global change issues on a regional basis. e ЕМА project will result in a 13-volume synoptic flora of the plants of North America and a computerized database derived from this infor- mation. In one design (Morin & Tomlinson, 1988), 13 data files were envisioned: taxonomic place- ment, scientific and vernacular name(s), authori- ties, synonyms, type specimen(s), literature, mor- phology, breeding system(s), chromosome numbers, flowering time, elevation, habitat, and distribution. At the Floristics for the 21st Century workshop (Morin et al., 1989) it became clear that the da- tabase would also need to contain an image file, data on relative abundance or rarity, status as threatened or endangered, economic significance, plant-animal associations, ethnobotanical uses, and accurate locality data. Such data sets, when merged with other biophysical parameters, define а geo- graphic information system. In sum, the data uni- verse should consist of both taxon data and spec- tral/spatial data from satellite and/or aircraft remote sensors, and geographic information sys- tems (GIS). These data sets need not be co-housed, but they must be accessible through networks, and their files must be compatible to address manage- ment questions being raised by environmentalists and global change scientists. The revision of the Нога of the Rocky Moun- tains began in earnest in 1989, spearheaded by onald L. Hartman, curator of the Rocky tain herbarium at the University of Wyoming. The chapter describing the origin and development of the flora of the Rocky Mountains (Morain, in re- view) to be included in this publication contains a oun- section on human impacts since European settle- ment. It is in this context that the development of a species information system compatible with a GIS seems most urgent. Habitat fragmentation (and outright loss of some habitats), environmental changes induced by acid rain and the rise in con- centration of greenhouse gases, fire management practices, economic activities (both extractive and recreational), and alterations due to plant intro- ductions are all human imprints that affect the pace and direction of floristic and vegetational change. Today, only a few areas of pristine wil- derness remain in the Rocky Mountains in an ex- panse of logged, mined, grazed, and farmed land. There is an extensive network of national forests, national parks, state forests, wilderness areas, na- 193 tional monuments, and other “protected” or mul- tiple-use areas, but even these suffer from overuse and general environmental degradation. The spe- cies they harbor already represent isolated, and in some cases peripheral, populations about which we know very little. Worse, the boundaries on these civil and political units become evident on the land- scape as economic development and land use changes outside the units evolve, thus fragmenting any avenues for plant migration that may someday be necessar The oo Yellowstone Ecosystem has emerged in these concerns as a specific area of interest. The range of local climates, elevations, lithologies, and soils results in a natural fragmentation of the land- scape and striking vegetation dynamics. It is the largest ecosystem remaining in the Rockies that has been relatively untouched, and as a conse- quence there are many questions about the eco- system that should be addressed on behalf of the There ја огта, Great Basin, region's general history and development. are disjunct elements of and boreal vegetation that contribute to the eco- system's uniqueness. It has long been recognized that management questions of such intricacy re- quire joint and collaborative research and the use of modern technologies to coordinate the programs The Greater Yellowstone Ecosystem Coalition is a step in that of many state and federal agencies. direction. Agencies meet periodically with the Na- ture Conservancy to discuss problems within the National Park area and to coordinate their various management programs. А geographic information system is being created to catalog and explain the interactions between plant and animal species, es- pecially the rare or endangered forms stranded by their environmental history The United Nations Environment Programme ОМЕР) has established а Global Environmental Monitoring System (GEMS) as а means for as- ~ sessing environmental conditions, the quality of life, Volume 80, Number 2 Morain 311 Emerging Technology and the changing conditions of life around the world. The United Nations Environment Pro- gramme GEMS/GRID database has been devel- oped to assist these efforts. As of 1990, GRID consisted of some 15 global data sets (such as soil, topography, climate, vegetation) representing the first-order physical environmental parameters af- fecting human welfare. They are largely compiled from sets created at national and smaller scales, but private sector efforts are also gaining impor- buted by institutional (e.g., tance, as are th environmental, university, and other not-for-profit) groups. At the national and regional levels, mission- oriented government agencies have contributed a wide array of information coverages. What is clear, however, is that many GRID coverages are at too coarse a resolution for regional application. Their utility is greatest at continental, hemispherical, and global scales. Nevertheless, they provide a stimulus for refinement and the basis for initial model build- dertaken through the International Geosphere / Biosphere Program (IGBP), which is designed to collect transect data across the region. GLOBAL CHANGE MODELING Global change modeling will impact the course of research in the Rocky Mountains and could serve as a driver for modernizing taxonomic data collec- tion. Morain (1990) has summarized the main ar- guments. If the concentration of CO, and other greenhouse gases doubles by the middle of the 2 151 century, some climate models show that by the mid-2000s the Earth's temperature could rise be- tween 1.5?C and 4.5°C (UNEP/GEMS, 1987). ; 3*C increase would emulate the global temperature of 13 million years ago (Mathews, 1987), roughly analogous to the Middle Miocene. While the actual rate, cause, and magnitude of increase is debatable, it is generally agreed that there will be an increase, that as a result, there will be northward and Model redistributions of species, plant com- munities, and whole vegetational zones toward cool- „> er regimes. Concomitant with global warming, a change in precipitation regimes up to 546 is predicted by climate models (Mathews, 1987). Some areas would gain, while others would become more xeric. Sev- eral models suggest that there would be a decrease in total annual precipitation in the Rockies. Clearly, the flora will respond to these temperature and moisture stimuli differently. Predictive modeling and visualization based on a modern species infor- mation database, and an environmentally oriented image-based GIS that contains detailed ecophysi- ological data on a species-by-species basis, will be required to: (1) know which new assemblages could possibly populate habitats of the future; (2) predict how long it would take them to adjust by natural means; and (3) determine whether there is sufficient land area at higher elevations to sustain them. At present, there is almost no evidence by which to assess the rates and directions of biotic adjust- ment in the Rocky Mountains, but we do know that existing communities and distributions are in disequilibrium with today's environments (Habeck, 1987). Becker (1961) suggested that the Oligocene Florissant flora of Colorado expanded into Montana at a rate of 1 m every 30 years. If the anticipated rate of global warming actually occurs, some stud- ies have suggested that vegetation in the Rockies will have to migrate 300 km poleward, which at Becker's rate would take 9 million years, or hun- dreds of meters upslope (perhaps as many as 50 Although ecophysiological tolerance to adjust to these mi- grations, others with slower rates of migration, or those that lack access to new habitats will either require human intervention to ensure the intro- duction of suitable genetic stocks into new localities or they will necessarily become rare, endangered, or regionally extinct. In any the need to migrate upslope will be ecologically impossible for many species of plants might possess the case, alpine tundra forms, and may be very restrictive to numerous other species, since land area will physically diminish. As Lewin (1989) indicated, the depth of biological knowledge required to manip- ulate species’ geography is extraordinarily com- plex. But such manipulation will undoubtedly be required because natural migrations will be stymied by existing and evolving land use restrictions. To achieve this in-depth knowledge will require anal- ysis using GIS technology, an ability to model changes through time, and to then animate the rates and directions of predicted change for better long-term planning. MECHANICS OF SYNTHESIS Existing data sets are so diverse in their origin ite exists to assemble and structures that a and convert these assets ints a body of regional data. The compatibility of GIS databases operating in a multitude of computer hardware/software con- figurations is essential if economy of operation is to be achieved and duplication avoided. In addition to the GEMS/GRID coverages that are available in both raster and vector format, there 312 Annals of the Missouri Botanical Garden are several databases under development at the federal level. NPFLORA is a module in the National Park Service database called COMMON goner, pers. comm.). It contains almost 80,000 records on some 14,300 taxa in the National Park system, including 12 areas in the Rocky Moun- tains. Although the module is not specimen based, it is a documented floristic list of species present in each of the park areas. It currently contains information to the subspecies level including re- gional distribution, status, habitat descriptors, and citations. It will soon include synonyms as set forth in the National List of Scientific Plant Names (NLSPL) prepared by the U.S. Department of Ag- riculture-SCS (Soil Conservation Service, 1982). Yellowstone National Park has been developing a GIS for management purposes. At present, the system consists of two vegetation layers: habitat types, as defined by Daubenmire (1966); and cover types, as defined by dominant emergents. In ad- dition, there are coverages (some only partial for the 2.2 million acres) for bedrock geology. precip- itation, and hydrology. Digital line graph (DLG and elevation models (DEM) are also included. The system is running on a Masscomp computer using Earth Resources Laboratory Analysis (ELAS) and Geographic Resources Analysis and Statistical Sys- — tem (GRASS) software and operates in both raster RA is available for Yel- lowstone but has not yet been merged into the GIS. and vector modes. A specimen-based electronic catalog is also being created by the Park museum curators, but its re- lation to the Yellowstone GIS has not been ad- dressed. Тћеге are several other databases under соп- struction, or already operating, for management areas within the Rockies. Many of these are housed in universities or federal agencies (e.g., DEM and DLG files) and contain coverages captured for dif- ferent reasons. For example, the Technology Ap- plication Center (TAC) at the University of New Mexico is compiling a checklist of Rocky Mountain species, as developed by John T. Kartesz at the University of North Carolina. This represents the first time that a modern checklist has been con- structed for the region and from which we can research the phytogeography of species. In a com- panion effort the TAC team is retrieving the lists of rare and endangered species from the Nature Conservancy's Heritage Programs in each state for each of the mountain ranges within the Rockies. We are plotting these data onto 1: 250,000 scale maps and loading the results into ELAS image processing software to create a region-wide data- base. The data are then plotted onto Digital Ele- vation Models and will eventually be superimposed onto NOAA Advanced Very High Resolution Ra- diometer and Landsat Thematic Mapper (TM) data. ENABLING TECHNOLOGIES As the elements of an initial operating system are being configured, it clearly will be necessary to develop the means for rapid and efficient species data collection in the field, and for rapid infor- mation retrieval and dissemination in herbaria. Among the obvious candidates for this moderniza- tion are global positioning systems (GPS), pen- equipped computers, microminiature cameras, new satellite data collection systems, and advanced vi- sualization technology. GLOBAL POSITIONING SYSTEMS Global positioning system (GPS) technology, like GIS, has already reached commercial status and has numerous technical and scientific applications. Compact and low-cost, hand-held GPS devices are already available. Magellan Systems Corporation, for example, has produced a battery-powered unit called the МАМ 1000 PRO, which weighs only 0.85 kg and measures 21.5 X 5.0 cm. It has an operating temperature between — 40°C and 70°C and operates on six AA alkaline batteries. It pro- vides two-dimensional, single-fix, autonomous po- sitioning accuracies of 25 m, or three-dimensional accuracies of 30 m in 2.5-3 minutes. Using two receivers, accuracies improve to 5 m in 2-D, and 10 m in 3-D. Data can be downloaded directly into a data logger or personal computer. Its display is a 4-line, 16-character alphanumeric liquid crystal display (LCD). Gilman et al. (1989) have also described a hand- held system, called TRIMPACK, that has been designed to address applications where the GPS equipment must function independently of external power, mechanical support, or antenna. By using LCD technology, they keep power consumption under 3 watts and capture 20 hours of continuous use. Industry leaders are expecting the price per unit to decrease rapidly from about $3,000 to perhaps $500. Other references on hand-held GPS technology include McDonald (1989), Smith et al. (1989), Baker (1987), and Smith (1985). egarding taxonomic data collection specifical- ly, Mlle (1989) has described his use of a portable GPS in the Ituri rainforest of Zaire. He used a 12- volt gel cell battery-powered Magnavox 4400 backpack receiver with a detachable 0.6 m alu- minum antenna. The system was solar-recharged with an ARCO C100 solar panel. With it, he was Volume 80, Number 2 9 Morain 313 Emerging Technology able to obtain 3 hours of receiver time per recharge, during which he captured three-dimensional geo- graphic locations in forest openings as small as 0.125 ha Hand-held GPS technology represents an essen- tial link in creating a species information system for the Rocky Mountains. When in place, the full complement of 21 NAVSTAR satellites will permit continuous data collection in remote areas, thus freeing taxonomists from interpolating or, worse, guessing at the location of their collection sites. Environmental attributes such as aspect and ele- vation will also be much more accurate and con- sistent, especially if field crews are correlating data from several receivers. PEN-EQUIPPED COMPUTERS An exciting new development in computing 15 the pen-equipped electronic clipboard. This genre of devices is expected by the industry to sweep the marketplace for applications where traditional clip- boards are already used to record data. One com- pany, Grid Systems, has produced a 2-kg “Grid- pad" containing the power of a hand-held (not laptop) personal computer, It looks essentially like tablet, but one that derives its electronic utility from recognizing block characters and storing them in battery-backed RAM cards. It quickly recognizes characters in nine different lan- an "etch-a-sketc guages and has a “fill-in-the-blanks” graphics form that could be ideal for recording specimen data in the field. Gridpad is a DOS machine, but other manufacturers are planning to use ROM-based DOS an indows environments. An essential ingredi- is to PCs and workstations—the objective being to avoid data ent in this evolving capability, of course, ensure that data are transferrable to office reentry and transcribing errors. Pen-equipped technology is so new, and potentially so large, that several American and Japanese manufacturers have entered the market and are currently posturing for software companies to develop applications. In an article titled “Computers Without Keys,” Normile & Johnson (1989) described Canon's "Ал Note” and Sony's “PalmTop.” These, too, use an electronic pen to allow users to take notes, do word processing, calculate ape formulas, and create graphics. The PalmTop is x 16 x 5 ст with a 10 x 15 ст LCD vum and weighs 1.5 kg e included). It uses a 68HC000 micropro- cessor and has 2MB ROM for programs and 320K B RAM for temporary data storage. In support of modernizing field taxonomy, now is the time to ensure that adequately robust pen- equipped computers and software are developed for botanists’ needs. Those elements in the design of the global species information system that re- quire accurate field data could form the basis for the requisite. software, and could even include 2-dimensional or 3-dimensional positional data ob- tained from hand-held GPS observations. For man- agement of herbarium specimen data, an ability to hand-scan information quickly and download it simultaneously into a database is essential. MICROMINIATURE CAMERAS A new generation of video imaging cameras is available for field and herbarium use. These are described in product literature as microminiature color charge-coupled device (CCD) cameras that record images in digital format. The cameras are about the size of a person's thumb, the imaging element of which is about one-half inch consisting of as many as 420,000 CCD sensing elements. Together with their accessories, these cameras could be ideal tools for rec EAE field specimens using ] power from a field v e (current technology would permit ranging as up to 30 m from the vehicle), or in converting herbarium specimens to CD-ROM formats. Resolution of these systems also permits close-range imaging of plant morphologies. Among the other specifications attractive to fiel and laboratory taxonomy are: (1) electronic shutter speeds ranging from 1/60th to 1/10,000th second; (2) low (10 lux) subject illumination re- of a quirements; (3) wide range of operating tempera- tures (— 10°С to 4090); (4) 12-volt DC or 120-volt AC power source; and (5) an array of interchange- able lenses from 3 mm to 24 mm. SATELLITE DATA COLLECTION SYSTEMS The Earth Observing System (EOS) is a major element in the National Aeronautics and Space but it is also a major international cooperative element Administration’s “Mission to Planet Earth,” in the International Geosphere/Biosphere Pro- gram, the U. obal Change Research кайа and the World Caste Research Prog 1991; Thompson, 1991). It has a о теби = tential for impacting the ways and means for col- — ecting, storing, disseminating, and analyzing bio- logical data of the future. Currently it is designed as a 15-year program to commence with a series of EOS-A satellites in 1998. components include a Space Measurement System Its three primary (EOSSMS,, consisting of a wide variety of sensors; the Data and Information System (EOSDIS), de- signed to archive and disseminate data, and to 314 Annals of the Missouri Botanical Garden supply products to the science and research com- munities; and the Scientific Research Program (EOSSRP) consisting of institutionally supported individuals and collective research efforts Space Measurement System (EOSSMS) The needs of the Earth-observing and Earth System Science research communities encompass the ionosphere, atmosphere, hydrosphere, litho- sphere, and biosphere. Past and present satellites are already collecting data from these realms using imaging and nonimaging (sounding and profiling) systems, and many of these data sets will be readied for general access as part of a pre-EOS dissemi- nation system. The EOSSMS sensors are divided into two categories representing a plethora of ex- periments and applications in the earth sciences. The first category consists of facility instruments, so called because they represent the core NASA instruments for EOS, and will be managed mainly y SA facilities or other closely related non- NASA laboratories. The second category is referred to as Principal Investigator Instruments because they represent more narrowly defined, special pur- pose experiments. Data captured by both sets of instruments will be divided into three categories to ensure their widest use and analysis: (1) facility investigations; (2) principal investigator investiga- tions; and (3) interdisciplinary investigations. Each of the facility sensors has numerous mea- surement and integration functions as well as sci- entific applications. The Moderate Resolution Im- aging Spectrometer (MODIS), for example, will be collecting data for at least two dozen experiments, including: (1) the dynamics of snow and ice over large areas; (2) studies of primary production in the world’s oceans; (3) species variability and im- proved carbon and nitrogen cycling determina- tions; (4) monitoring global vegetation dynamics; (5) global aerosols monitoring; and (6) land surface DIS-N (nadir) will have 36 channels for data capture ranging from 0.4 to 14.2 um in 10- to 360-nm bandwidths. MODIS-T (tilt) will have 32 channels between 0.4 and 0.88 um in 10- to 15-nm bandwidths. Such fine resolution spectral data are expected to provide temperature measurements. insights into ecosystem dynamics, global change research models, and general vegetation monitor- ing. Data and Information System ( EOSDIS) EOSDIS represents the means by which data processing and analysis are to be made available to the scientific community. At the heart of all EOS activities is the EOS Operations Center located at Goddard Space Flight Center in Greenbelt, Mary- land. It will serve as a communication link between EOSSMS and EOSDIS to ensure sensor operations and raw data capture. Working together with the System Management Center, the Instrument Con- trol Facility, and the International Partner Oper- ations Center, raw data and level-O data will be distributed to a network of centers called Distrib- uted Active Archive Centers (DAACs). Level-O data will be released to these centers to be further refined by their internal Product Generation Systems into level 1 or higher products, or stored for later use in their Data Archive and Distribution Systems. This initial operating structure and management plan for EOSDIS cause ultimately several versions will succeed this is referred to as Version 0, be- design. Version 0's focus is on science data pro- cessing, but is designed to be flexible and evolu- tionary as subsequent technologies, expertise, ex- perience, and infrastructure come into play (Taylor et al., 91) An integral part of EOSDIS is represented by data that have been, and are now being, collected. These pre-EOS data sets are divided into a “Path- finder" series having global extent (e.g., NO AVHRR/Global Area Coverage, Total Ozone Ver- tical Sounder, and Geostationary Operational En- vironmental Satellite—GOES); a “Precursor” se- ries of important, but not necessarily global, data A AVHRR/Local Area Coverage, Airborne Visible and Infrared Imaging Spectrom- sets (e.g., eter, and Thermal Infrared Multispectral Scanner); and “non-EOS” data (e.g., Landsat Thematic Map- per and Multispectral Scanner). The DAAC at EROS Data Center has recently inaugurated the Global Land Information System as a prototype in Version О to provide access to some of these data: namely, AVHRR, Landsat, 1-degree USGS Digital Eleva- tion Models, Matthews Global Vegetation Data Set, and Major World Ecosystems, among others. ‘These are accessible for searching, browsing, and inter- rogating through Internet and direct-dial using IBM- compatible terminals. ADVANCED VISUALIZATION TECHNOLOGY Three-dimensional image processing technology and 3-dimensional GIS modeling techniques are in the forefront of new earth science applications, especially those in meteorology, oceanography, hy- drology, and geology, where depth sounding and profiling have been long-term 2-dimensional tools. In terrain analysis, the availability of digital ele- vational models (DEMs) has enabled ecologists to Volume 80, Number 2 1993 Morain 315 Emerging Technology superimpose computer classified land cover types onto a static representation of relief, and to even quantify the elevational ranges, aspects, and total area occupied by each type. At a higher level of sophistication, and requiring massive computer processing, techniques are available for moving the terrain past an observer. Several videos have been produced by the Visualization Laboratory at the Jet Propulsion Laboratory in Pasadena, California, to illustrate how static terrain and weather con- ditions might a Е and through the landscape." pressive and instructive, but they do not animate appear to someone “flying around These are indeed im- “change.” Visualizing temporal changes in land- scape ecology requires a static observation point while animating a time series as viewed from that perspective. Research is now beginning that will add the time dimension (in the form of multiple- image dates) to this capability. What has not been attempted yet is to adapt the technology for 3-D modeling of vegetation cov- er—that is, to assess the structure of plant com- munities in the spatial domain. D^ta sets that would permit this kind of analysis are not available either. Unlike geologic, atmospheric, and oceanographic applications where vertical profiles are standard measurement attributes (ie., well logs, depth soundings, balloon soundings), vertical structural profiles are not a common measurement in botan- ical and ecological studies. Certainly they are not acquired systematically, although one economic means for getting them would be to use low-altitude aerial mapping photography. The needs of the glob- al species information system, and the GISs that will interface with it, will surely impact future data types and their collection strategies. But, as Ellef- son & Kaiser (1990) have suggested, it is not just for the species information system or GIS modeling They cite global change issues and improved forest planning that such developments are emerging. and management as driving forces behind the need to develop new methods in forest research. SUMMARY Accurate locational and ecophysiological data for most dominant species may be relatively easy to incorporate into image-based and object-oriented databases, even today. Similar data for rare and endangered species and even dominant species at the periphery of their ranges, however, still require data collection, and this must be done by using less time-consuming and less costly means. It is not unreasonable to imagine that in our drive to collect massive amounts of environmental data and make measurements of current earth resources, we will design “new-age” botanists who are equipped with more than plant press, hand lens, ruler, and knap- sack. New-age botanists will have voice, data, and video links to their home institutions, as well as to the world's major herbaria and centralized data repositories. In the future, they will be able to predict and visualize the impact of environmental changes on flora and vegetation, assess the poten- tial impacts of varying economic development sce- narios, and better manage ecosystems and com- munities at risk. Emerging technologies that will MB pro- cessing chips; seamless integration of spatial and foster these improvements include 256 nonspatial data sets; pedabyte storage capabilities; advanced computer networks, local area networks, and client servers; advanced visualization capabil- ities; advanced telecommunications systems; and miniaturization of electronic field and herbarium equipment. LITERATURE CITED Baker, P. J. 1987. GPS in the year 2000. J. Navigation + 175-183. Becker, H. F. 1961. Oligocene plants from the Upper Ruby River Basin, southwestern Montana. Geol. Soc. 82. Daur BENMIRE, R. 1966. Vegetation: ет а of typ- al communities. Science 151 -298. ELLEFSON, P. V. & H. F. KAISER. "1990. Forest and natural resource economics and policy research: strategies and Map n for the coming decade. Re- newable Resources J. 8: 16-22. mM D., J. LovELL, А. STAVROS, J. TANG & L. WEBER. 989. Design considerations for a self- ge GPS receiver. Pp. 397-402 in Proc. ION. Satellite Division's Шыны! Technical Meeting. ти of Navigation, Washington, D.C. Навеск, J. R. 198 Present-day vegetation in the northern Rocky Mountains. Ann. Missouri Bot. Gard Lewin, R. 1989. How to get plants into the conser- vationists” ark. Science 244: 32-¢ MarHEWs, J. Т. 1987. Global climate change: toward a greenhouse policy. Issues Sci. Techn. 3: 57-68. McDosarp, K. D. 198 PS receiver ара, in tech- nology, mo and performanc 57-359, 61-385 in Proc. ION. Satellite о. Inter- national Technical Meeting. Washington, D.C. MORAIN, S. А. rolegomena on a Species Infor- mation System for the Flora of the Rocky Mountains. Designs for a снн lant Species Information Sys- tem, p Gre ——. Global Earth seen: coordinating sets in US pro- grams for global monitoring. Pp. 2-19 in 1. Down- man (editor), die Spatial Data 2000. Univ. College London, Londo Morin, N. R. & K. i TOMLINSON (editors). N 1: 1-5. mer. Newslett. 1988. Fl. Annals of the Missouri Botanical Garden Morin, N. R., R. D. WhHETSTONE, D. WILKEN & K. L. TOMLINSON. 1989. Floristics for the 21st century. onogr. Syst. Bot. Missouri Bot. Gard. 28. NonMiLE, D. & J. T. JOHNSON. 1989. oo with- out keys. Popular Sci. August: 66-69 9 civil uses envisioned for satellite navigation system. Commercial Space 1: 27-29 SMITH, M.S S., L. Neat, R. R. Thomas & С. Н. WALKER. 1989. Low apre no and arrays for satellite navigation. Pp. 34 in Proc. Int. Conf. Anten- nas & Рина ПСАР 89). Inst. Electrical Engin., London. SoIL CONSERVATION SERVICE. 1982. UE List of Scientific Plant Names, 2 volumes. U.S.D.A., U.S. Gov. Pu Office, Soil о нв Serv., Techn. Publ. #1 TAYLOR, T. D., RAMAPRIYAN & J. С. Dozier. 1991. The development of the EOS ns and In- big uw System. Proc Sci. Meet- ne Toso, R. J. nter for Earth Observing System Land Кураш Data. Proc. 29th Aerospace Sci. Meeting, Amer. Inst. Aeronautics & Astronautics (AIAA). UNEP/GEMS. Тће Greenhouse Gases. Envi- ronm. Libr. #1. UNEP. Nairobi. WILKIE, D. S Parlamant of a backpack GPS in a tropic al rain forest. oo Engin. & Remote Sensing 55: 1747-1 WiLsoN, E. О. 1989. Threats to Tc Sci. Amer. Sep.: 108-116 DATA BANKING AND THE ROLE OF NATURAL HISTORY COLLECTIONS! Nancy R. Morin? and Janet Gomon? ABSTRACT riety of strategies is being used to capture information from natural history specimens and to develop new Mun databases containing information on taxonomic i making them available through centralized ‹ or dispersed databases. Geographical d instructional pro relationships of taxa are being linked to other data sources, jn both collections and non-collections databases groups. Consortia of institutions are being formed for the are underway to develop data standards and data-sharing protocols in order to facilitate sharing of these data among institutions and across disciplines. Natural history collections provide a rich infor- mation resource about our natural world. The past several years have seen administrators of system- atics collections and systematists placing greater emphasis on providing improved access to collec- tions data and supporting the synthesis of infor- mation from collections. New directions in re- search, a broadening of users of systematics information (from environmentalists and natural resource managers to science educators), coupled with relatively recent advances in computer tech- nology, have been the driving forces. The result is a significant shift in purpose and expansion of col- lections-related databases, from the traditional da- tabases serving primarily as indexes to the physical collections, to ones that are capable of being used as true research and educational tools. The traditional catalog or database (for example, specimens in a herbarium) contained a relatively small core of text information about a single mu- seum's collections. Within the botanical commu- nity, core information entities are usually contained in the specimen "label" and have included speci- men identity, locality and date of collection, col- lector, field number, and for some organisms, sub- strate. When only parts of a plant are collected, some morphological features (e.g., overall size) have been noted. Additional data, such as the habitat in which the organism occurred, or other organisms with which it occurred, have often been included. Other information may also be placed on the label, e.g., chromosome counts or ethnological infor- mation, or a reference to a publication or reposi- The place and method of storage for each item have also been recorded. tory. Unless computerized, or otherwise cross-indexed in some way, none of this information is readily available from the stored collections themselves. To find specimens from a specific locality, the entire collection must be searched manually. Similarly, to re-create an itinerary for a collector whose fie book is not extant, it is necessary to comb the holdings for collections made by that collector. These traditional catalogs or databases, if au- tomated at all, often cover only a small portion of an institution's collection, making them minimally useful as research tools. So, although the collections were considered an extremely rich information re- source, their utility was largely a potential yet to be realized. The adage, “data rich but information poor," seemed to apply. CURRENT DATA BANKING STRATEGIES AND PROJECTS Today, the picture looks quite different with regard to data banking. The systematics commu- nity has undertaken multiple strategies to expand specimen-based databases and develop new re- search databases containing derived, rather than primary, collections data (e.g., taxon-based data- bases). In addition, collections data have been in- ! We tha ? Missouri Botanical Garden, P.O. Box 299, St. nk T. Duncan, R. Haynes, J. me ie W. Brigham for им information for this paper. . Missouri 63166, U.S ' National Museum of Natural History, Smithsonian а. Washineion, no 20560, U.S.A. ANN. Missouni Bor. Ganp. 80: 317-322. 1993. Annals of the Missouri Botanical Garden tegrated with non-collections data sources to pro- vide a richer summarize the community's information. resource. Опе can current, primary strat- egies in data banking efforts as follows: 1. Rapid automation strategies for specimen-based data in order to bring entire collections into an automated database. N . Coalescing of a single discipline's specimen data from multiple institutions into a centralized, eas- ily accessed database. о Employing a variety of new tools, from ana- lytical tools to authoring (software) tools. Incorporation of new data types (e.g., and sound samples) with traditional text data. = image P Expansion of collections-related databases by including new data elements, both primary and derived, or by linking these databases with non- collections-related data sources (e.g., chemical databases). Integration of collections data across disci- ^ plines. -] Integration of system architectures across de- partments within an institution and across in- stitutions. Three current projects exemplify the first two strategies of rapid computerization of collections data and coalescing of that data into centralized databases. The Southeastern Regional Floral Information System (SERFIS) has as its objectives: (1) to de- velop a centralized database composed of label data of specimens in herbaria of the southeastern United States; (2) to provide data management and storage for these data; (3) to provide technical assistance and support to these herbaria; and (4) to participate in national and international attempts to establish computerized access to specimen label data (Ma- crander & Haynes, 1990). Label images are cap- tured at each herbarium, stored in digital form, then entered into the database at the University of Alabama. Each herbarium receives its data back in computerized form, and each has access to the About 0 herbaria are involved, covering са. ш specimens, approximately 10% of the bo- tanical specimens held in U.S. and Canadian in- stitutions. The Specimen Management System for Califor- nia Herbaria (SMASCH) has objectives similar to those of SERFIS, but its strategy for data collection is different (Duncan, 1991). The Association of California Herbaria has devised a plan whereby its three largest members, the California Academy of Sciences, the University of California, Berkeley, ete database via a network connection. 10 and Rancho Santa Ana Botanical Garden, each have responsibility for entering data from about one-third of the largest California families. Once this phase is completed, coordinated data entry of the remaining families for each institution and for the remaining institutions will be undertaken. About 53 institutions and 6.5 million specimens (10.8% of the total U.S will be involved. Data entry and access will take . and Canadian holdings) place over NSFNet between institutions and local area networks within institutions. Images primarily from a collection of photographic slides will begin to be scanned as part of the project. The Inter-Institutional Database of Fish Biodi- versity in the Neotropics is a major effort to build an international database bringing together infor- mation from all major collections of neotropical fishes, the Academy of Natural Sci- ences of Philadelphia, the University of Michigan initiated b Museum ет and Cornell University. It will make tic and geographic data from specimens ГИ in collections in North, Cen- tral, and South America and in Europe. It includes a Geographic Information Systems interface, an electronic atlas of neotropical type localities, and remote access features. This effort uses the soft- ware MUSE for its central database. MUSE has been developed for and used by the ichthyological MUSE recom- data set and .. community at many institutions. mends adherence to a "primary" structure but is flexible in how the "secondary" data, which are less likely to be exchanged, are handled. These three examples all provide access to col- lections data from more than one institution, and thus they each require integration of data across institutions. he third and fourth strategies, employing new tools and expanding the traditional catalog with new types of data such as images and sound sam- ples, are particularly well-exemplified by the fol- lowing projects. Geographical Information Systems (CIS) when linked with collections databases offer powerful an- alytical capabilities. They provide new information about the locality from which a specimen was col- lected, and they can be used to analyze complex relationships between the distributions of taxa, their characteristics, and their relationships with other taxa. Various current projects analyze precise lo- cality information from systematics collections us- ing GIS (1) to manage and conserve individual taxa and (2) to manage geographical areas and habitats. The Illinois Natural History Survey combines data on natural features, man-made features (roads, Volume 80, Number 2 1993 Morin & Gomon Data Banking and Natural History Collections 319 pipelines, etc.), and distributions of individual col- lections, to make recommendations about planned land use in light of its impact on the environment. The survey can use the same data to determine the physical and ecological requirements of indi- vidual taxa for either research or management applications. Information on the location of organ- isms was used to suggest locations for the Super- conducting Super Collider (SSC), а high-energy physics research facility under the auspices of the U.S. Department of Energy. The extensive data- base maintained by the Natural History Survey on the wildlife and plants of Illinois, including their distribution and abundance, their habitats, and their life histories and status within the state, plemented with computerized databases lands, natural areas, soils, property boundaries, and was sup- on wet- other information necessary for siting the SSC and evaluating its potential environmental impacts (Jo- selyn, 1988). The Survey uses similar data to pre- dict distributions of taxa. For example, county rec- ords for a water beetle, Haliplus tortilipenis, were compared with such information as average Jan- uary temperature, annual precipitation, and per- centage forest cover. GIS was used to determine which other counties have ecological parameters similar to those of counties within which the beetle is known to occur. That is, to determine which counties have the same combination of tempera- ture, precipitation, and forest cover as the counties where collections of the beetle have been made. Based on this information, а hypothetical distri- bution range can be mapped showing areas where the species might be expected to occur. The Illinois Natural History Survey also uses knowledge of habitats and collecting patterns to elp “clean example, fish can be assumed to have been col- = up" locality and mapping data. Fo lected in water, so collection sites that do not fall within a water body can be "snapped" to the near- est stream. Unfortunately, when more than one stream is nearby, the correct choice may not be made. Collections are often made at bridges or near roads, and that information can be used to influence which stream site is chosen by the com- puter. The best solution, according to Warren Brig- ham at the Survey, is to sit down with the collector in front of a terminal on which streams, bridges, and roads can be displayed and ask the collector to show where the collection was actually made. Similar, but even more sophisticated GIS meth- ods are used by the Environmental Resources In- formation Network (ERIN) in Australia to increase the accuracy of computerized collections infor- mation (Slater & Noble, 1991). Australian natural history institutions are collaborating among them- selves and with the ERIN Unit, a part of the Australian National Parks and Wildlife Service, to computerize their collections following community- adopted standards and protocols. Data are trans- ferred from the institutions into the central ERIN database and can be used in GIS and other appli- cations. ERIN uses a climatic file, called BioClim, that incorporates data from thousands of data- gathering locations for 16 climatic parameters such as wet season rainfall maximum, wet season rainfall minimum, and wet season maximum temperature. By mapping reported localities for a taxon, a sta- tistical analysis can be used to identify specimens that fall outside the typical climatic range for the taxon in any one or a combination of parameters. Those specimens can then be checked to see that they have been correctly identified and that the locality was recorded correctly. BioClim data can be used as a modeling tool by adjusting any one of the 16 parameters, which results in associated dependent changes in the rest of the parameters. The current climatic requirements of species can be inferred from the species’ current distributions; thus, if the distribution of a given climatic type is changed the concomitant change in species distri- bution can be predicted. This information can be used to determine whether natural area preserves are available in the predicted future climatic zones to provide refuge for sensitive taxa. A new tool of the medical profession, and one that a number of researchers at the Washington University School of Medicine (St. Louis) have employed and studied, is X-ray computed tomog- raphy combined with three-dimensional image pro- cessing (Hildebolt et al., 1990 enables scientists to conduct research without phys- his technology ically damaging specimens. It also improves the ability to make new measurements and preserves images of objects that must be returned to their places of origins. Physical anthropologists at the National Museum of Natural History (Smithsonian Institution) recently completed a prototype in which two-dimensional image slices were produced, then used to produce three-dimensional reconstructions that show both the exterior and interior surfaces of specimens and objects. In the future, one may find these types of images commonly associated with museum collection databases and available by network to other researchers. The relatively new authoring tools that enable creation of interactive monographs and tutorials have been used in several recent projects aimed at science education and the identification of spec- imens. As part of the Biological Curriculum In- 320 Annals of the Missouri Botanical Garden novation Study (BioCIS) led by Chad McDaniel of the University of Maryland, Smithsonian Institu- tion researchers have been working on the devel- opment of introductory college-level biology courses through the production of hypertext multimedia databases that document aspects of biological di- versity. These applications combine text, image, and sound from a variety of museum information sources. Projects being undertaken include a da- tabase of information on mammal species of the world, an application documenting collection and processing of botanical specimens, an identification manual for crocodiles and their products, and an application that allows the user to “deforest” the Amazonian rainforest while learning what species can or cannot live in the remaining forest frag- ments. A botanical project similar to BioCIS has been envisioned by Arturo Gómez-Pompa of the Uni- versity of California, Riverside. His prototype, Q-Taxa, incorporates images of specimens, living plants, drawings, maps, plus specimen information and taxonomic information in an environment that invites exploration by both layman and researcher (Gomez-Pompa & Plummer, 1990 Programs for producing descriptions and inter- active keys from sets of characters and character states, DELTA, have been written by Michael Dall- witz and Les Watson of CSIRO (Australia). An interactive key to cultivated plants in North Amer- ica using hypertext is being developed by Larry Ragan at Pennsylvania State University. Interac- tive keys to families or to taxa within individual families are also available, e.g., a family key from T. Duncan, a key to Australian rainforest trees by B. Hyland, and a key to the orchids of Great Britain by R. Pankhurst. Although such files are not de- rived directly from specimen data, they are a prod- uct of the expert staff associated with natural his- tory collections and are important tools in curation and education. The fifth strategy supports information synthesis in two ways: (1) by broadening collections-related databases beyond the traditional specimen “label” database to include both primary and derived data, and (2) by linking collections-related databases with other, non-collections-derived sources. The first of these two ways is exemplified by the Hymenoptera Database System (HDS) and Flora of North America (FNA). They are taxon-based online databases. HDS is maintained by system- atists at the U.S. Department of Agriculture and National Museum of Natural History, and FNA is maintained by the Flora of North America project at the Missouri Botanical Garden. The Hymenoptera Database System contains systematic and biological information оп about 18,000 species of saw flies, parasitic wasps, true 1974, for a description of earlier phases of this project). wasps, ants, and bees (see Krombein et al., The project has integrated data from both ento- mological and botanical sources, one botanical source being Index Nominum Genericorum, a da- tabase of all validly published generic names for 1979, 1986). The HDS contains literature references on taxonomy, biology, and plants (Farr et al., morphology for described taxa. Information is pro- vided on geographic range, ecology, whether in- troduced or adventive (accidently introduced), plant or animal hosts, parasites, predators and prey, pol- len preferences, and flowers visited. А wide variety of queries can be made from this database, e.g., a biological control specialist could ascertain. the names of parasitic Hymenoptera introduced from other parts of the world and successfully established in the United States, and obtain information on their known distributions. Flora of North America is a binational, multi- institutional project to produce a floristic treatment of the vascular plants and bryophytes growing out- side of cultivation north of Mexico. In addition to 14 published volumes, the project will produce a taxonomic database containing information on the relationships, characteristics, and distributions of the plants. А key with most floristic and faunistic projects, is that it feature of the FNA project, as provides authoritative data on the plants. Flora of North America, while accumulating morphological data in a traditional manner, captures that mor- 'TROPICOS, devel- oped at the Missouri Botanical Garden) that allows phological data in a database (1 analysis of correlations among taxa, and among characteristics. Moreover, Flora of North America, by integrating morphological data with collections- based, mappable distribution data, can provide a tool for studying the relationship between morpho- logical characteristics and the physical parameters within which the plants occur (Morin, 1991). Furthermore, taxonomic databases like HDS and FNA can serve as connecting points to link dis- parate data using a taxonomic framework. An in- dependent query might ask for the chemicals pro- duced by plants that occur on a certain soil type. Specimen information mapped using GIS against soils gives the names of plants that occur on the soil type in question, and those names can be used in turn to query a remote phytochemical database to get a list of the chemicals they are reported to produce. The number of databases that can be linked through the taxonomic framework is unlim- Volume 80, Number 2 Morin & Gomon 321 Data Banking and Natural History Collections ited but includes ethnological, reproductive, ana- tomical, genetic, specimen-based, and bibliographic data. Researchers in the Evolution of Terrestrial Eco- systems Program (ETE) (University of California, Santa Barbara, and National Museum of Natural History) have initiated the ETE database to allow broad-scale comparisons of the paleobiology and paleoecology of terrestrial ecosystems, plus their plant and animal communities. The ETE database will document and compare plant and animal as- semblages from the Silurian to Recent times and will establish patterns of similarity and novelty in terrestrial ecosystems. Researchers can then de- termine the degree to which ecological roles of organisms and structural characteristics of eco- systems have changed. These patterns will then be the basis for assessing the impact of environmental conditions on ecosystem structure, and the rela- tionship of such phenomena to evolution within specific groups of organisms. ETE database incorporates both primary and derived information about fossil localities and the plant and animal species collected from them (Damuth, 1991). For each locality, in addition to its name, age, and location, there are fields for sedimentological, taphonomic, and paleoecological data. Each locality has a species list (vertebrate, plant, or invertebrate). For each species there is a taxonomic classification and a set of morphological and ecological descriptors that vary with the type of species. A locality-species table associates lo- calities and species and records relative abundance of species at a given locality. The bibliographic table contains publications that were sources of information for the database. The museums table records which museums have material and from which localities. The time units table names geo- chronological units and their associated maximum and minimum ages (in years). Last, the sedimentary structure and taphonomic table records observa- tions relevant to depositional environment. n example of linking collections-related data to other data sources is the Environmental Pro- tection Agency- National Oceanographic Data Center (EPA-NODC) Taxonomy System modern- ization effort. The EPA recently turned to system- atists at the National Museum of Natural History, U.S. Department of Agriculture, National Marine Fisheries Service, and the U.S. Fish and Wildlife Service to help improve an existing taxonomic sys- tem by developing an underlying logical data model for taxonomy. The database includes valid scientific names, synonyms, and common names. Each vali scientific name is associated with a single taxon of a higher rank, and each synonym is associated with a valid name. The database provides all taxa con- tained within a higher taxon, scientific names cor- responding to common names, and common names applied to any scientific name. The system is seen as critical to supporting EPA’s Environmental Monitoring and Assessment Program and Office of Water Quality. The database enables field biologists to associate data recorded in the field or laboratory ~ water chemistry, locations, numbers of organisms, and biomass) with a taxon. Interactive access be- tween the taxonomy system, the Ocean Data Eval- uation System, and a biological survey system called BIOS are planned. The last two strategies are essential to laying the groundwork for broader-scale information syn- thesis. These projects may well support the next major step in providing access to natural history collections. These efforts involve standardization of "data architectures" rameters) and development of protocols to allow The International (e.g., definition of data pa- data sharing across systems. Working Group on Taxonomic Databases in the Plant Sciences (TDWG) is an organization that has attempted to establish standards, if none exist, for various data elements in plant taxonomic work, expressly for the purpose of facilitating data ex- change, and it is one of several organizations that have efforts underway in these areas. The Asso- ciation of Systematics Collections Computerization and Networking Committee has also initiated an important effort to develop these standards across all systematics collections. A systems architecture undertaking is being led by a group from the largest free-standing (non- university associated) natural history institutions and botanical gardens in the United States. The member institutions have agreed to participate as a consortium to plan modernizations of their col- lection and research information management en- vironments. The purpose is to install at each in- stitution new equipment and software to support research needs, enhance collections management, and improve interconnectivity within and among the consortium institutions. An example of a similar undertaking within а major institution is the new project at the University of California, Berkeley, to integrate data from the various museum collec- tions throughout its campus. This would allow, for example, linking archeological, ethnobotanical, and taxonomic information together on an institutional network. Other institutions are planning to connect integrated online library systems to their research databases to allow direct linkages of library infor- mation to research-derived data and vice versa. 322 Annals of the Missouri Botanical Garden CONCLUSION These new approaches all support а primary goal of the nation’s systematics collections insti- tutions: to improve information availability to our systematics researchers and to a growing com- munity of information users, including environ- mentalists, conservationists, and science educators. To achieve this goal, new working relationships have been formed within and among organizations, across taxonomic disciplines, and between system- atists and information science and engineering pro- fessionals. The new technology makes available entirely new categories of information that now can be derived from collections, and it provides new tools with which to analyze and use both traditional and new kinds of information for research and resource management. LrrERATURE CITED Damutn, J. D. Electronic research collections: perspectives and an example from the Evolution of Te erre estrial Ecosystems C onsortium. In: E. C. Dudley and Evolutionary Biology IV. Dioscorides Press, Port- land, сеје os Duncan, T. 199 . Association of California Herbaria, Inc. Working Paper Number 1 FARR, e R., J. A. e & F. A. STAFLEU (editors). dex um genericorum (plantarum). Regn Were bie ae 100-102. — == 6 ————. 1986. Supplement |, Regnum Vegetabile 113 GÓMEZ- em A. & O. E. PLUMMER. 1990. Video oras: a new tool for systematic botany. Taxon 39: 576-585. HILDEBOLT, C. VANNIER & R. H. Кмарр, 1990 Validation study of skull three-dimensional computerized tomography measurements. Ar Physical Anthropology 82: 283-294. JosELYN, M. 19 On siting the SSC. /n: Illinois Geo- E Information сед. The Illinois Natural His- tory Survey Reports 281-282. KROMBEIN, K. V. КЕ, төлү J. Скоскетт. 1974. The North American Hymenoptera Catalog: a pio- neering effort in computerized publication. Bull. En- tomological Soc. Amer. 20(1): 24-29. MACRANDER, А. M. & R. R. Haynes. 1990. SERFIS: a methodology for making ч herbaria specimen data bases a reality. Taxon 39: -44]. Morin, N. R. 1991. Beyond the du ds databasing flora of North America information. /n: E. C. Dudley (editor), The Unity of Evolutionary Biology, Pro- ceedings of the International Congress for E and Evolutionary Biology IV. Dioscorides Press, Port- land, Oregon. SLATER, W. R. € S. J. NOBLE (editors). 1991. ERIN Program Brief, June 1991. Mn National Parks and Wildlife Service, Canber THE NATIONAL SCIENCE FOUNDATION: FINANCIAL, INTELLECTUAL, AND KNOWLEDGE BROKER FOR SYSTEMATIC BIOLOGY' James R. Estes? ABSTRACT The National Science Foundation (NSF) is the e financial broker for a pey with NSF funding rch in the United dd view by pe most basic systematic resea Foundation, which are bd to merit re Systematists generate ideas in the в. This review process has n реј to enhance the science n of proposals to the and, therefore, constitutes one facet of intellectual Шога Program officers, in turn, react to the proposals and their reviews in ma (PIs) to provide intellectual а NSF also PR a proactive role in the intellec symposia, and conferences; (2) development of initiatives; (3) funding of (1) sponsorship of study grou orkshops, Small Grants for Exploratory { шл (SGER) proposals; existing NSF awards. Whether reactive or proactive Foundation is not a direct player in knowledge e cs and analys and results from the d ini it funds. Rather, it relies on the scientifi journals and indices. However, NSF does fund innov collection-label data; and to exchange Ass data, nowledge and/or data iine Infor directly into society-at- educational programs. Systematic biolo ing recommendations for awards, and in the process interact with ideas and ideas gained from large by co i proposals for scientific studies with proposals for formal and/or informal gists are in an ideal position to take advantage of any such funding opportunities the Principal Investigators tual brokering of systematics throug and (4) fu nding Special Creativit ty Extensions (SCE) to because of the d extensive interactions that occur between natural history museums and members of the systematic commun President Harry S. Truman signed the National Science Foundation Act of 1950 (Public Law 81- 507) to culminate a decade of debate about the structure and function of the agency. This law with its various amendments provides the National Sci- ence Foundation (NSF) authority to promote re- search in the United States, and in that role, NSF has been a benefactor of basic scientific research (England, 1982). This is especially true for sys- tematic biology, because NSF provides more than 90% of the federal funding awarded to U.S. col- leges and universities for studies of phylogeny, classification, and biodiversity (BBS Task Force, 1991 NSF as FINANCIAL BROKER FOR SYSTEMATIC BIOLOGY During the half-century of NSF’s existence, new comparative techniques and methods of analysis have emerged to transform systematic biology, in general, and plant systematics, in particular. The most compelling element in this transformation has been the coupling of molecular and computational biology. The field of plant systematics has also been aided and abetted by the conservative nature of the chloroplast genome. The emergent field of plant molecular phylogenetics has developed so rapidly that a rigorously determined and testable phylo- genetic framework for the green plants is already 1988). Phy- logenetic relationships within large and complex well within our reach (Palmer et al., families, such as Asteraceae, Fabaceae, and Po- aceae, are also yielding to molecular analysis with high levels of resolution (Jansen et al., 1991; Lavin & Doyle, 1991; Lavin et al., 1990; Soreng et al., 1990). It is even possible to probe evolutionary divergence deep within the hierarchy of life using data from molecular sequence analyses of extant taxa and microfossils from deep strata, embedded "І sincerely appreciate the assistance of W. F. Harr Schindel, K. Estrada, P. Olmert, С. Stuck, D. —_ апа М. аги speak on a subject about which I was and г Yates, F. Griffo, D. Barkley for inviting me to much greater appreciation, eu Peter Raven and the staff gram Director, Systematic & Population Biology Cluster, National Science Foundation, y Present address: Robert Bebb Herbarium, Department of Botany and Microbiology, (несш ? Associate Pro D.C. 20550, U.S. of Oklahoma, Norman, Oklahoma 73019, U.S.A. ANN. is, J. Edwards , P. Werner, L. Krishtalka, J. Rodman, T. tosh. I also thank T. H. main a novice, but for which I now have of the Missouri Botanical Cadei for their hospitality. Washingto Missouni Bor. GARD. 80: 323-332. 1993. 324 Annals of the Missouri Botanical Garden as always within the modern evolutionary paradigm (a small sampling of papers includes Buchheim et а!., 1990; Gajadhar et al., 1991; Knoll, 1992; Lake, 1991; Palmer et al., 1988; Schlegel, 1991; Woese, 1987). The resulting phylogenetic infer- ences can then be used to interpret patterns and derive syntheses in fields as diverse as biogeogra- phy, behavior, and development (Brooks & 1991). scarcely seemed possible even a decade ago. 4 McLennan, These advances would have Systematists have also sounded a clear signal that the world faces a crisis over the loss of bio- logical diversity (National Science Board, 1989; Prance, 1984; Solbrig, 1991). This realization has resulted in a renewed interest in surveying and cataloging the world's biota and the consequent writing of floras and faunas (Heywood, 1984b). In this effort, systematists are also beginning to em- ploy more sophisticated systems for data storage and retrieval to ensure that the locality data for each specimen and the descriptions and nomen- clature for each taxon can be readily retrieved; thus the flora or fauna can be efficiently and ef- fectively revised (Bisby, 1984; Dallwitz, 1984; Cibbs Russell, 1989; Pankhurst, 1974). Watson et al. (1986) have effectively used this approach for the grasses of Canada, and a consortium of African botanists are accomplishing the same for the entire Flora of Southern Africa (Gibbs Russell, 1989, 1990). It is possible that floristic work is at an all- time high, with several major surveys of plant di- versity in the temperate and tropical zones either recently completed or well under way (Heywood, 1984a). The printed products of the most modern of these endeavors may be reminiscent of floras and faunas of the past two centuries, but they differ in having a foundation based on detailed databases, and the most forward-looking floristic projects also include equipment for the highly precise determi- nation of map coordinates and involve the potential for interacting with Geographic Information Sys- s (Rodman, 9 This is surely one of the most exciting periods in the history of systematic biology, in part because so much seems within our grasp, but also because we realize the magnitude and importance of the tasks before us. It is arguable that these, and other, major de- velopments would have transpired in the absence of . However, it is undeniable that the Sys- tematic Biology Program (Table 1) provided fund- ing for much of the development of the theory and techniques of systematic biology: the collection of specimens and data and the interpretation of results that has fueled this burgeoning of systematic bi- ology. In addition, officers of the Systematic Bi- ology Program established a special competition for survey and inventory research (NSF, 19904), by allocating approximately $2.3 million. exclu- sively to fund floras and faunas. This competition now forms the basis of the newly erected Biotic Surveys and Inventories (BS&I), which with Sys- tematic Biology is in the Division of Environmental Biology (DEB) (Table 1), attesting to the Division's commitment to the importance of cataloging the plants, animals, fungi, and microbes of the world. Research Collections in Systematics & Ecology RCS&E), also in DEB, provides funds for the maintenance and computerization of those research collections (NSF, 19852). It is difficult to envision this flowering of the two ~ cardinal dimensions of plant systematics—phylo- genetics and floristics— without the monetary sup- port of NSF (Table 2 erated by these two approaches and their use in ). With the excitement gen- other contexts, the number of proposals received in Systematic Biology appears to be on the rise (Table 3). Additionally, both molecular and diver- sity studies are costly—the former requires a large budget for supplies and materials and the latter for travel and fieldwork; both are labor intensive. As a consequence, this past year also saw a dramatic increase in the level of requested funding (Table 3). This placed tremendous stress on a budget that has grown by much smaller increments (Table 2). Regardless, the importance of the roles of System- atic & Population Biology (S& PB) and Biotic Sur- veys & Inventories (BS&I) as financial brokers of systematic biology is clear and secure. NSF as INTELLECTUAL BROKER FOR SYSTEMATIC BIOLOGY PROPOSAL PREPARATION AND REVIEW AS ELEMENTS OF INTELLECTUAL BROKERING In general, funding of awards by NSF is made in response to proposals submitted by Principal Investigators (PIs), who are members of the re- search community. Typically this funding is based on the recommendations of program officers (POs), who are at the base of NSF’s administrative hi- erarchy. Funding is, therefore, a grass-roots phe- nomenon both in the community and at NSF, and the decisions concerning funding or declination of proposals at NSF is primarily reactive. That is, POs in S&PB and BS&I respond to proposals that result from the pool of systematic biologists. Thus, as noted above, POs are important players in the financial brokering of the science, but decisions concerning funding are also at the intersection of Volume 80, Number 2 1993 Estes 325 NSF and Systematic Biology TaBLE 1. Reorganization of the programs in environmental biology at NSF. The core of the Systematic Biology Program (represented in greater detail than the remaining programs) was recently united with elements of Population Biology to form the Systematic Systematic Biology and Population Biology co merged to form the Ec m any of several divisions and prog ology Studies Cluster. A third cluster, Long-Te formed from the Biological Research Resources Program and the lon & Population pend Cluster i represented in greater detail than the other clusters). nue to func on as indep endent units for budgetary and proposal managed ograms have also cts in Environ € eof was m Ecos s. PIs sho grams. The Ecology and Ecosystems pr erm Projec ng-term projects unit uld contact programs officers concerning the submission of proposals in particular research еш. “Only applicable programs are shown in the Division of Integrative Biology and Neurosciences. Previous organization Present organization Directorate for Biological, Behavioral & Social Sciences (BBS) Division of Biotic |" & Resources (BSR) Ecology Behavioral hn Com cology ава нее Ecosystems Studies Program & Physiological Ecology Program Behavioral, morphological & molecular adaptation Life-history studies Molecular evolution Physiological ecology Population genetics Systematic Biology Program Bio geography Classification theory Functional morphology Phylogenetic analysis Speciation Surveys and inventories xonomic revision Biological Research УРИ Program Collection support Monograph support Directorate for Biological Sciences (BIO) Division of Environmental Biology (DEB) Ecological Studies Cluster Community Ecology Program Ecosystems Program Integrative & Theoretical Min Program Population Interactions Prog Systematic & Population Biology ules Morphological & Molecular жна Ргоргап! Taxonomy & Classification Progr Classification theory evisions НЕ Coevolution & Paleobiology Pro- speciation Molecular Population Biology Population & усне Genetics Program Life IDA & Popula ynamics Long-Term Projects in E mei Biology Cluster Research Collections in Systematics & Ecology Pro- gram Biotic Surveys & Inventories Program Division of Integrative Biology & Neurosciences (IBN) кезе: d & Functional Ecology Program nimal Behavior Program financial and intellectual brokering. This results because the continuing dialog between the PO and PI has substantial room for discussions of the sci- ence itself. These discussions range through time from initial deliberations concerning an idea for a research project through discussions of the reviews of the proposal. At the critical first step in the process, scientists interested in submitting a proposal may contact the relevant PO, with telephone or electronic mail being the most effective means of communication at this time. POs can often provide counsel re- garding the potential viability of a project in the process of merit review. Frequently, yu qu can be offered that might direct the PI t ocus or protocol that has the potential to ee ma- terially the research proposal, and more impor- tantly, the proposed research. The following hy- pothetical cases provide examples of the advice POs might offer prospective РІ: (1) an indication that a proposed flora has a lessened chance of being recommended for funding by the Biotic Sur- vey and Inventory Advisory Panel unless a mean- ingful and useful database is included; (2) a com- ment that reviewers react more favorably to proposals in molecular systematics if the inferred phylogeny is then used to comprehend other ra- diation patterns in the taxon; and (3) a suggestion that a monograph with a rigorous analysis of the phylogeny of the taxa would be a more meaningful contribution than only a set of keys and descrip- tions. Of course, the advice carries no guarantees, and the PIs are free to accept, decline, or modify the suggestions, but whether or not these proposals 326 Annals of the Missouri Botanical Garden TABLE 2. Funding history in Systematic Biology Pro- жы fiscal year 1985 to fiscal year 1992, in millions of dollars. All figures are from NSF Budget Requests to ongress except for bracketed figures, which indicate Current Plan or Estimates. FY'86 bu by Gramm-Rudmann-Hollings sequestrations. B i- ; DEB— Div rectorate ied Biological Sciences udgets were reduced vision of En- vironmental Biology; SB — temente Biology Program. Fiscal year NSF BIO DEB SB 1985 1,50 245 58 12.2 1986 1,523 262 > 239 0061 58 12.2 1987 1,628 259 58 2.7 1988 1,718 266 59 12.4 1989 1,88 281 63 12.7 1990 2,026 294 67 13.7 1991 [2,316] [315] [1] [14.4] 1992 [2,722] [356] [78] [15.9] are funded, the modified research will likely ђе more significant and have a broader impact on systematics if the recommendations are given care- ful consideration. Тће Foundation is deeply rooted in the academic tradition of merit review. Accordingly, most pro- grams in DEB employ the advice of practicing scientists by (1) mail-solicitation of reviews from specialists in some aspect of the proposed research (2) convening advisory panels composed of scientists with broad expertise prior to making their own recommendations for awards. This practice provides for a much more thorough review. Per- haps just as importantly, a thorough and critical review elicits suggestions that sharpen the focus, clarify the goals, and improve the protocol of the proposed research. In brief, should have a salutary scientific impact on each proposed project and the PI, and ultimately then on the entire field. Peer review, including review the review system by the POs, is the most important element in the intellectual brokerage practiced by NSF because it can most directly influence individual PIs, albeit the reviews are available only to the individual PI, and then only after the proposal has been funded or declined. The “Panel Summary" important document in this phase of intellectual is the most brokering because it fixes the proposal under con- sideration within the context of modern systematic biology, and because it identifies both the flaws and strengths of the proposed research. The individual reviews, however, generally are more detailed and cover specific issues concerning taxon and char- acter selection, methods of data collection and anal- ysis, and the potential systematic and taxonomic BL Number of proposals submitted and total unding requested in Systematic Biology Program, fiscal year 1985 to fiscal year 1991. Requests in millions of dollars. umber of Funds Fiscal year proposals requested 1985 348 $52.4 1986 330 $50.5 1987 348 $55.1 1988 335 $54.0 1989 326 $60.0 1990 347 $64.8 1991 361 $79.2 impact of the work. There is, therefore, consid- erable responsibility thrust upon a scientist selected to review a proposal, and that responsibility is par- titioned between the responsibility of being fair to the PI and being rigorous for the science. Ке- gardless, reviewers should always attempt to im- prove the proposed work and hence the science. For this aspect of the system of intellectual brokering to function effectively, it is essential that reviewers write substantive, even if succinct, re- views in which they provide a detailed analysis of both the strengths and the weaknesses of the pro- posed scientific problem and the research protocol. In some instances, reviewers may have knowledge of only one or a few aspects of the proposal; how- ever, it is incumbent on these reviewers to criticize that aspect of the proposal in which they are expert. Only then does the PI receive the intellectual as- sistance necessary for development and the Panel and the Program obtain sufficient data to make an informed recommendation. The routine assignment of unqualified ratings of “excellent” or "very goo for all proposals to study а. taxa ог to investigate certain processes seldom, if ever, helps the PIs during the Panel's discussion of proposals, and it deprives PIs of the reviewer's expertise. If a proposal is truly excellent, it is vital that the reviewers explicitly and precisely explain in what ways it excels, because panels and programs almost exclusively use the narrative, rather than the rating scale, of the Proposal Evaluation form. Funds allocated to the programs in systematics, that is S&PB and BS&I, probably always will be insufficient to fund all of the proposed systematic research that is worthy of accomplishment. (It is arguable that such levels of funding would actually have a dampening effect on innovation and ad- vances such as those described for plant molecular phylogenetics.) Rather, because of its limited bud- Volume 80, Number 2 Estes 327 NSF and Systematic Biology get, the Systematic Biology Program must, and should, focus on research that will advance the field, that is thorough and in depth, and that will have an impact beyond the boundaries of the taxon under investigation. It is precisely in this funding environment that reviewers, panels, and program officers can be most creative influences in the dis- cipline. For instance, it seems obvious that POs in the Systematic Biology Program over the period from 1980 to 1989 responded to and encouraged the development of molecular systematics while maintaining the legitimacy of the review process. This creative management of the program was instrumental in the rapid development of molecular systematics and molecular evolution (Goodman, ; Goodman et al., 19 the breakthrough in understanding phylogeny and evolution of the Plant Kingdom. All the while the Program continued to fund meritorious proposals that were based on other comparative techniques, both traditional and new (Committee of Visitors for the Systematic Biology Program, 1991). At the end of the last decade, the Program also made a bold decision to establish the Survey and Inventory competition, thus enhancing our knowledge of the world's biodiversity.* As the new decade opens, the Program will undoubtedly identify other promising new vistas, perhaps including new approaches in systematic biology to understand phenotypic evo- lution, biogeography, and ecology based on and in the context of phylogenetics (Brooks & McLennan, 1991); the development of algorithms that permit phylogenetic analysis of hybrid complexes; and the refinement of methods of estimating phylogeny us- ing developmental pathways. All the while, the Program must continue to fund the most innovative molecularly and morphologically based phyloge- netic analyses and taxonomic revisions. Neverthe- less, this phase of intellectual brokering can only ther av- and made possible be reactive, and somewhat conservative. enues do exist, however, that offer a more active role for NSF program officers. NEW DIRECTIONS IN SYSTEMATIC RESEARCH In addition to funding research by reacting to proposals, the Foundation is directed to initiate basic scientific research (Public Law 81-507); therefore, NSF has the authority to function as an ' This decision is worthy of recognition. The program officers in Systematic Biology who responded so positively were Division Director and Deputy Division Director, re- spectively, and provided support for the decision. intellectual broker beyond what is possible in the standard DM grant programs. Appropri- t that permit a more proactive ately, p role for program officers. This permits flexibility and the capacity to respond to new findings and techniques. Fortunately, NSF is governed and ad- ministered by working scientists. The National Sci- ence Board (Public Law 81-507) governs and sets policy for the Foundation, and many of its members are active scientists. In addition, many of the ad- ministrators in the directorates, divisions, and dis- ciplinary programs are Visiting Scientists (often termed grams after one or two years in Washington. For instance, currently (1.е., June 1992) in the Division of Environmental Biology (DEB) 75% of the pro- gram officers are rotators. In addition, all of the permanent officers are research biologists, and all have taken advantage of NSF's sabbatical program by taking a year's leave in a university setting. The result is that the programs concerned with the biology of species, populations, communities, and "rotators"), who return to research pro- ecosystems are steered by peers who are keenly interested in the development of their own sciences and capable of discerning trends within the disci- plines. Program officers may stimulate the launching of new fields of systematic inquiry or the re-chart- ing of traditional areas in at least four ways, and examples will be drawn from the field of biodiversity similar scenarios could have been described for ~ cladistics, morphometrics, molecular systematics, or molecular evolution): 1. Funding of study groups, workshops, con- ferences, and symposia. This avenue typically in- volves a dialog between a PO and individual sci- entists, who may be the representative(s) of scientific societies. These venues provide opportunities for planning research agendas, transferring technolo- gies, discussing common research topics, and pre- senting research results for a common field of study, respectively. Two examples are the following. (1) Systematics Agenda 2000 had its origins in a study group organized by the Smithsonian Institution and the Association for Systematics Collections to seek a way for the systematics community to express its views and to take action concerning the decline in global biodiversity. A broader charter evolved, and officials at NSF indicated that they would be willing to entertain a proposal to fund additional workshops and conferences. The proposal received favorable reviews and was funded. (2) The work- shop on the development of a protocol for botanical surveys held in 1989 resulted from an extensive set of discussions between POs in Systematic Bi- 328 Annals of the Missouri Botanical Garden ology and the editors of Flora of North America. One result was publication of a guide for the de- velopment, management, and conduct of floristic projects, Floristics for the 21st Century (Morin et al., which provides counsel to PIs who intend to subiit biotic survey (plant, as well as animal and microbial) proposals to BS&I. 1er- ence to the elements of this publication should lead to more mature and better developed proposals to conduct floristic— and faunistic— research and, not coincidentally, more innovative and useful surveys. с evelopment of initiatives. Initiatives pro- vide a mechanism for obtaining additional, or new, funds for the Foundation, divisions, and programs, as well as in the federal budget as a whole. The idea for an initiative may originate at any level in the agency, but often occurs at the program officer level. Initiatives may also grow out of interactions with the scientific community, as products of work- shops, recognition of proposal trends, awareness of major unresolved scientific questions, congressional action, directives from the Director of NSF, the National Science Board, or all of these. Once again using the development of the current NSF programs in biotic surveys and inventories as an example: The над d wide (Direc torate for Biological Sciences— BI petition in search in Conservation and Restoration. Biology (NSF, 19902) resulted from all of the above mech- anisms, as did the Biotic Survey and Inventory asic Re- competition, which arose from the reallocation of internal funds in the Systematic Biology (cf. National Science Board, ' Program А This last-men- tioned competition is now the basis for a new pro- gram (NSF, 1990d; Yates & Estes, 1992). T funds have since been augmented by funds from hese the U.S. Agency for International Development (A.I.D.) to support biodiversity research in devel- oping countries and by A.I.D. and the National Institutes for Health (NIH) for a program in drug development, biodiversity, and economic devel- opment in developing countries (NIH et al., 1992; Schweitzer et al., 1991). The second of these aug- mentations was рце with the assistance of а workshop jointly funded by These agencies and others—such as Department of State, De- partment of Defense, Department of Agriculture, Department of Energy, and Department of Inte- rior—have responded to to a widespread recog- nition of the need for more biodiversity research. This combined effort therefore had а compounding effect. 3. Small Grants for Exploratory Research (SGER). for investigations that are urgent or risky or that This instrument for funding is designed have the potential to result in new directions in 989). They may be initiated by an investigator or, rarely, through the PO con- tacting а worker with appropriate expertise. Be- basic research (NS cause they do not require an external review, there is a relatively rapid turn-around in funding. Appropriately, relatively few are granted each year (five in Systematic Biology in fiscal year 1991). Of these, one provided funding to develop tech- niques for the study of biodiversity of free-living soil nematodes. As the Science Board (NSB, 1989) report recommended, diversity in the rhizosphere needs special attention. 4. Special Creativity Extensions (SCE). Programs have the capacity to grant up to two- year extensions for awards when creativity and productivity have been extraordinarily high. SCEs are made without any action on the part of the PIs; in fact, PIs cannot institute a request for such an extension. Not surprisingly, these extensions have been made sparingly. our SCEs were made by the Systematic Biology Program over the past three years, and one was awarded to continue a highly creative faunistic survey. Intellectual brokering at NSF, therefore, has tended to function early in the research enter- prise—it functions through initiatives and propos- als. In this light, it is not surprising that the NSF direct-access database provides entry to the ab- stracts (equivalent to the Program Summaries that are submitted with proposals) of funded proposals. This new service, (STIS) The Science and Tech- nology Information Service (NSF, 1991b), per- mits anyone with a modem to scan these abstracts using keyword or topical searches. Thus, it is rel- atively easy to ascertain if the Foundation has funded research in a particular area and to discern research trends. Other information is also acces- sible through the use of STIS, including the text of most NSF publications. The Systematic Biology Program also publishes an annual Awards List (available by writing the Program), which provides an indication of funding trends, and constitutes the Program’s primary effort in direct knowledge bro- kering. NSF 4s KNOWLEDGE BROKER FOR SYSTEMATIC BIOLOGY Perhaps surprisingly, NSF does not a play a significant role in the formal release or indexing of data that results from funded research projects. There are traditional and historic, as well as eco- nomic, reasons for this seeming paradox. Volume 80, Number 2 1993 Estes 329 NSF and Systematic Biology NSF came into being after the basic structure of the U.S. scientific enterprise was well estab- lished. Although the basic research enterprise in was, and remains, centered on colleges and universities and other similar non-profit institutions, in a real sense it consisted of a diffuse, but well- organized, suite of scientific societies with members scattered at institutions throughout the country. One service of these societies has been the dissem- ination of scientific knowledge among their mem- bers and the scientific community at large; as a consequence, most of the societies publish. peer- reviewed journals. Most scientific societies also hold or sponsor annual meetings that encompass both independent research reports and symposia. Sym- posia provide fora for defining new research di- rections as well as summarizing important research findings. In addition, indexing services were also available in 1950; their computerized descendants provide easy, if sometimes expensive, accessibility to the literature. e Foundation apparently accepted and placed its trust in the ability of these scientific communities to recognize and to publish the results of worthy scientific research, rather than imposing another layer of bureaucracy and reports. (This philosophy continues to be reflected in the criteria for funding of systematic monographs within Research Collec- tions in Systematics and Ecology (RCS&E) (pre- viously Biological Research Resources Program or R) (NSF, 1985b).) The decision to require rel- atively few and brief reports and to trust in the scientific integrity of the publication system of the societies freed PIs to concentrate on accomplish- ment of the proposed research, rather than on the writing of extensive, detailed reports. Because the editors, editorial boards, and reviewers work for scientific journals without real compensation, the task is accomplished efficiently, with integrity, and less expensively than if the government managed the system. Thus, NSF depends upon non-govern- mental agencies to monitor publication of the re- sults of its awards. (Albeit, early in its history, NSF provided a strong foundation for continued funding of Biological Abstracts (England, 1982).) In my view this system works effectively and efficiently. It would be quite erroneous to believe from this lack of direct involvement, that NSF is disinterested or that it fails to provide funding for the important task of knowledge brokering. The Foundation fos- ters and promotes publication of research results. For instance, publication costs are routinely in- cluded in proposal and award budgets, with the NSF budget form even including a specific line item “Publication Costs/ Documentation / Dissemi- nation" on the NSF Summary Proposal Budget (NSF Form 1030). NSF can also support publi- cation of meritorious research beyond this basic level of support through competition for mono- graph publication costs within RCS&E as men- tioned above. Pertinent for plant systematics, this program subvented the publication of Systematic Botany Monographs through its foundling stages, promoting the spectacular success the journal has enjoyed. NSF requires listings of publications and reprints for the PI’s final project report— required of all funded projects ane it requires the inclusion of the section “Results from Prior NSF Support” (NSF, 1990c) for all grant applications where the PI or co-PI has received NSF support in the pre- vious four years. Clearly both these exercises are for the convenience of the review system and are not intended as mechanisms for the dissemination of research syntheses. The development and maintenance of electronic databases should follow pathways similar to those for the review and publication of research journals. That is, the society or community that collects and uses the data should provide the organizing ele- ments and expertise for the database. NSF already provides seed funds for those database projects that are considered meritorious (NSF, 1990b). The Hu- man Genome database is a prime example (Cin- kosky et al., 1991; Roberts, 1991; Salton, 1991). Other databasing needs that involve the systematic community include ongoing attempts to make the data in museums and herbaria more accessible and useful (Bisby, lud. Gibbs Russell, 1989; NSF, 1990b; Stuessy & Thomson, 1981). RCS&E cur- rently provides funding for collection management, specimen databases, and electronic networking in museums and herbaria (NSF, 1985a). There may also be a need for a central facility for plant mo- lecular sequences and their alignment for the de- velopment of phylogenies and another for elec- tronically stored phenotypic data for automated h of 1990 (NSF, 1991a; Public Law 101-606 have with respect to databases, but clearly it calls for data collection and management of information that is highly relevant to systematic biology, and there will be clear opportunities for the discipline and the research collections. Perhaps NSF's most influential role in the syn- thesis of research results is through the funding of symposia, either those in association with the an- nual meetings of societies or symposia that are free-standing. Once again, NSF is dependent upon 330 Annals of the Missouri Botanical Garden TABLE 4. Funding by Country for Biodiversity in De- veloping Countries Joint NSF/A.I.D. Competition, fiscal years 1990 and 1991. Total awards — $7,166,220 (NSF funding = $4,677,220; A.I.D. funding = $2,489,000). Number Number о о Country awards Country awards Argentina | Madagascar 2 Bolivia 2 Malaysia 1 Brazi 8 Mexico 2 Chile 2 Namibia 1 Colom 1 Nicaragua 1 Costa Rica 7 Panama l Dominican Philippines 2 Republic | Thailand | Ecuador | Venezuela 1 Gabon l ot: 37 Indonesia | the research community to recognize the need for syntheses and to write creative proposals for fund- ing. The Annual Systematics Symposium at Mis- souri Botanical Garden provides one example of NSF funding for meetings that focus entire fields of research, such as the methods of phylogenetic analysis within the Plant Kingdom, the subject of the 1 symposium. The previously cited Flo- ristics for the 21st Century (Morin et al., 1989) accomplished the same goal for floristic research. One final example is the recognition by the eco- logical community of the need for a center for ecological synthesis and analysis. Accordingly, a proposal was submitted to the Foundation by James H. Brown and Stephen R. Carpenter on behalf of that community for a workshop to complete a needs assessment and task analysis for the establishment of such a national center. Although program ofh- cers can stimulate such applications, it seems clear that the ideas are most likely to emerge from the scientists themselves. OPPORTUNITIES FOR EXPANDING THE IMPACT OF SYSTEMATICS I believe that Systematic Biology is a viable scientific discipline with profound implications for humankind. That viewpoint is not, however, uni- versally understood, or perhaps even accepted. There are two great arenas where systematists need to act with vigor to change that perception: (1) developing countries and (2) the American public. Members of our community are already actively engaged in the former domain, although there is TABLE 5. Criteria for funding in the joint NSF/A.I.D. program Biodiversity in Developing Countries. 1. Scientific merit. . Impact on host country and its biological diversity: a. Conservation of biological diversity. b. Contribution to sustainable с: 7 to development of biodiversity infra- N use of biotic resources. structu E ^ [m ect on economy or on the economic role of studied species or habitat on the economy much yet to be done. The latter arena promises a more truculent foe, ignorance. There may be а general lack of public understanding of such basic issues as cause and effect, adaptation, evolution, food webs, the role of natural systems in sustaining human populations, humans as organisms, et cet- era, et cetera. INTERNATIONAL DEVELOPMENT AND SYSTEMATIC BIOLOGY The community of U.S. Cheering working NSF a the U. has = a long and in concert with Agency for International Development, important presence in the Neotropics. These gov- ernment agencies and members of the U.S. botan- ical community have played an important role in the development and shaping of systematic ге- search and systematic institutions in developing countries. However, the relationships between bot- anists and developing countries are changing, and now should, and typically do, involve a partnership arrangement. All the core programs in DEB fund research in foreign locales, but these may or may not involve foreign scientists and scientific infrastructural de- velopment of the host nation. Two programs at the Foundation, however, explicitly encourage this par- ticular level of international involvement. Biotic Surveys & Inventories (NSF, 1990d) urges such an interaction between the U.S. and the foreign scientists, if the proposal involves research in a developing country. Most reviewers of BS&I pro- posals hold the PI accountable for the degree and level of involvement for such projects. Although expected in most such proposals, this activity is required for the research to be considered for fund- ing through Biodiversity in Developing Countries (BDC) that is jointly supported by NSF and A.I.D. In its first two years, BDC funded 37 projects in 19 countries (Table 4). A.I.D. funds can be used only to support host country scientists and infra- Volume 80, Number 2 1993 Estes NSF and Systematic Biology structure, title to any purchased equipment must remain in-country at the close of the award period, and all awards are subject to the principles that guide A.I.D. (Table 5). Counterpart NSF funds support the U.S. scientists! participation in the projects. A similar situation occurs for the program concerned with conservation of biodiversity, de- velopment of pharmaceuticals, and economic con- dition in third-world countries. This program is jointly funded by NSF, A.I.D., and the National Institutes of Health (NIH et al., 1992; 1991). International Programs at NSF also promotes Schweitzer et al., international interactions. In my judgment, this program is under-utilized by the community of Systematic Biologists as a possible funding source. PRESENTATION OF SYSTEMATIC BIOLOGY TO THE GENERAL PUBLIC NSF funds meritorious proposals for the transfer of scientific knowledge and ideas to society-at-large. One means of accomplishing such a transfer 15 to couple scientific studies with formal and/or infor- mal educational programs. This combination of ba- sic science and education may be proposed in sin- gle, successive, or concurrent proposals (NSF, 1988a, b). Thus, systematists might propose ex- hibits or lecture-series that explain systematic bi- ology as a scientific discipline and that use the results of a funded project. Systematic biology is in an ideal position to take advantage of any such funding opportunities be- cause of the extensive interactions between natural history museums and herbaria and systematists. Museums and herbaria have received NSF support for development of instructional materials for class- room use and participation in teacher training ef- forts. Too often, however, these efforts have been conceived and proposed by non-scientists from pub- lic education offices without direct input from evo- lutionary biologists or systematists. SUMMARY NSF funds most basic U.S. research in system- atic biology and acts as an intellectual broker at various junctures in the chain of events that lead from the germinal idea for a research effort to the formulation of formal and rigorous inferences. However, the Foundation is not a direct player in the formal release of the data and conclusions that the Foundation can act as a knowledge broker in the result from funded projects. Nevertheless, funding of international and educational grants. LITERATURE CITED BBS Task Force. 1991. Adapting to the Future: Report of the BBS Task Force Looking to the 21st Century. NSF 91-69, Washington, D.C. Вазву, К. А. 1984. Automated taxonomic information , 301-822 in V. Н. Heywoo od & D. M Moore (editor) Current Concepts in Plant Taxon- c Press, New York. A. MCLENNAN. 1991. ову Ecology, and Вл A Research Program in Com parative Biology. Univ. Chicago Press, Chicago. BucHHEIM, M. A., M. TunMEL, E. A. ZIMMER & R. L CHAPMAN. 1990. Phylogeny of Chlamydomonas (Chlorophyta) based on cladistic-analysis of nuclear 18s ribosomal-RNA sequence data. J. Phycol 274-282. eee. M. J., J. W. FickETT, P. Сима & C. Bur 1991. Electronic data publishing and GenBank. Sci. ence 252: 1273-1277. COMMITTEE OF VISITORS FOR THE SYSTEMATIC BIOLOGY ROGRAM. 1991. Report of the Committee of Vis- itors for the Systematic Biology Program. Systematic Biology de Washington, D.C. DaLLwrrz, M. J. User s Guide to the DELTA System: A a System for Coding Taxonomic erar CSIRO Division of Entomology Report No. 13. CSIRO, Canberra. ENGLAND, M. A Patron for Pure Science: The Natural Science Кошке s Formative Years. NSF, Washington, GAJADHAR, A. A. W. C. Макоџакрт, R. Наш, J. GUNDERSON, E. V. ARIZTICA-CARMONA & M. К SOGIN. Ribosomal-RNA sequences of Sarcocystis muria, Theileria annulata, and оа cohnii reveal evolutionary relationships among api- complexans, Mr pos and ciliates. Molec. Bio- chem. Parasitol. 45: 147-154 С̧іввѕ RusseLL, G. Е T The PRECIS uicina systems and its implications for the flora of souther Africa. Pp. 11-22 in N. R. Morin, R. D. Whetstone, D. Wilken & K. L. Tomlinson (editora), Floristics for the 21st Century: Proceedings of the Workshop. Monogr. Syst. Bot. Missouri Bot. Gard. 28. 1990. Grasses of Southern Africa. National Botanic Gardens, Botanical Research Institute, Pre- toria, South Africa. GooDMAN, M. 1992. volution 1: 1-2. ‚ M. CLEGG, J. FELSENSTEIN, W. FrrCH, D. HILLIS, M. RirEy, Е. RUDDLE, D. SANKOFF, P. ARZBERGER, M. COURTNEY, P. o үл C. LvNcH, J. PLESSET, M. Weiss & T. Yates. 1992 Worksho on Mo- lecular Ee Molec. Phylogenetics Evolution. 83-85 Editorial. Molec. Phylogenetics Heywoop, V. Н. 1984а. The current scene in plant taxonomy. Pp. 3-21 in V. H. M m ciui) Current bep dg in Plant Taxon- cademic Press, New А 1984 4b. Designing Hr for the future. Pp. 397-410 in V. H. Heywood & D. M. Moore (edi- tors), uh Concept in Plant Taxonomy. Aca- demic a New JANSEN, R cJ Mes, | Aer grs E J. Kim, S. C. eu LEY, L. E. МАТ & J. D. PALMER. 1991. Chloroplast DNA variation in the Asteraceae, 332 Annals of the Missouri Botanical Garden phylogenetic and implications. Pp. 2 Component of pe U.S. Global Change Research Рго- 9 in D. Soltis, P. Soltis & J. Doyle (editors), Bed SF 91- Э, Washington n & Hall, New York. Molecular Systematics. Chapman KNOLL, A. H. 92. The early evolution of eukaryotes: a ddr perspective. Science 256: 622-627 LAKE, J. 19 Tracing origins with molecular se- uo. es: metazoan and eukaryotic beginnings. Trends Biochem. Sci. 16: 46-50. Lavin, M. & J. J. Doyle. 1991. Tribal relationships of Sphinctospermum (Leguminosae): integration o traditional and chloroplast DNA data. Syst. Bot. 16 J. D. Palmer. 1990. Evolutionary implications of the loss of the chloroplast DNA in- verted repeat in the Leguminosae subfamily Papili- onoideae. Evolution 44: 390-402. Morin, N. R., R. D. WutrsTONE, D. WILKEN & K. L. TOMLINSON (editorek 1989. Floristics + the 21st E Proceedings of the ee Monogr. Syst. . Missouri Bot. Gar Nanos INSTITUTES OF HEAL TH, NATIONAL INSTITUTE OF MENTAL HEALTH, NATIONAL SCIENCE FOUNDATION & U.S. AGENCY FOR INTERNATIONAL DEVELOPMENT. . International л Biodiversity Groups: Request for Proposals. John E. Fogarty International Center, National Institutes of Health, Bethesda, Maryland. NATIONAL SCIENCE BOARD. 1989. Loss of Biological Diversity: A Global Crisis Requiring International Solutions. NSF, Washington, D.C. NATIONAL SCIENCE FOUNDATION. 1985a. Guidelines for Аже for Support of Systematic Collections. NSF , Washington, D.C. Publication Support for Major Tax- onomic Revisions and Systematic Monographs. NSF D.C. 85-13, Washington, 1988a. Material nodus diu Research and Informal Scienc on: Program Announce- ment. ee 88- 29, Wachingt on, D.C. ——— b. Projects to din: the Effective Use g of 5 and Math- -103, Wash- of е. in the Tea ematics: Program Solicitation, NSF 8 . 1989. Small Grants for Exploratory Research (SGER). NSF 89-85, Hur i 1990a. esearch in Conservation and Restoration ive Spec ial Competition. NSF 90- 66, ш hington, D.C. 0b. Database Activities in Biological, Be- те jer Social Sciences: Program Announce- ment. NSF 90-70, Washington, D.C. c. Grants for оен and Education in Science and Engineering: An Application Guide. NSF 90-77, Washington, D.C. NSF’s при Biology Ме pnm Axnaunces January 199] Panel for Survey and ventory Proposals. NSF, Washington, D.C. 19914. Global Change Research Program: A © © © gram. N 9 : The Science and Technology / D.C. ‚ R. К. Jansen, H. J. MicHaELs, М. W. CHAS & J.R. MANHARDT. 1988. Chloroplast DNA variation 2 шү phylogeny. Ann. Missouri Bot rd ar PANKHURST, R. 3 Тота Automated identification in systematics. Taxon 23: 45-51 m G. T. 1984. Completing the inventory. Pp. 65-396 in V. H. Heywood & D. M. Moore (edi- en Current Concepts in Plant Taxonomy. Aca- demic Press, New York. 1950. National Science Foun- 1950 PUBLIC TM 101-606. 1990. The Global Change Act of 1 ROBERTS, Eu 199]. dici DNA sequencing errors. Science L^ 1255-1256. Ropman, J. 1989. Workshop goals. Pp. 1-2 i N. , R. D. Whetstone, D. Wilken & K. L Tomlinson tedora, Floristics for the 21st Century: rkshop. Monogr. Syst. Bot. SALTON, G. . Developments in automatic text re- trieval. Science 253: 974-980. ва М . Protist evolution and phylogeny as discerned from small subunit ribosomal-RNA se- quence comparisons. European J. Protist. 27: 207- 219 Ne) o = SCHWEITZER, J., Е. С. Намргеу, J. Epwarps, W. Е. Harris, M. R. GREVER, S. A. SCHEPARTZ, G. CRAGG, 91. Commentary: sum- mary of the Workshop on Drug Development, Bio- Bini ~ | and Economic Growth. J. Natl. Can- . 83: 1294-1298. 0. T. (editor). 1991. From Genes to Ecosys- : A Research Agenda for Biodiversity. IUBS, Cambridge, Massachusetts. SORENG, R. J., J. I. Davis & J. J. Роу. 1990. phylogenetic rin of chloroplast DNA restriction site variation in Poaceae subfam. Pooideae. Pl. Sys Evol. 172 83-97 STUESSY, T. К. € K. S SOLBRIC, ке 1981. viro Pri- | Systematic Biology. The As- sociation for ia Collections, pe Kan- sas. Watson, L., S. С. AIKEN, M. J. DaLLwrrz, L. P. Ler- комтсн & M. Dusk. 1986. Canadian grass gen- ега: keys and descriptions in English and French from an automated data bank. Canad. J. Bot. 64: 53-70. ХУОЕЗЕ, C. R. Bacterial evolution. Microbiol. . 5 & J. F Estes 1992. Support for bio- div ersity resear the ed Science Founda- tion. Canad. Biodiversity 1: REVISED CIRCUMSCRIPTIONS AND SUBGENERIC TAXONOMIES OF CALANDRINIA AND MONTIOPSIS (PORTULACACEAE) WITH NOTES ON PHYLOGENY OF THE PORTULACACEOUS ALLIANCE! Mark А. Hershkovitz? ABSTRACT Cladistic analysis of Portulacaceae supports a rev ised taxonomy of Calandrinia sensu lato. Calandrinia, as here n American Calandrinia and Monocosmia are probably cladistically nested in section inia, Acaules smia. , and Monoco sister groups of either Rumicastrum or Montia, as proposed in a previous cladistic analysis. Montiopsis Kuntze, traditionally included in Calandrinia, comprises ca. 15 western South American species in two subgenera, Montiopsis and Dianthoideae. At present, cladistic analysis provides only weak evidence linking Calandrinia and Montiopsis resolves the relationship between the two which were not included in t the analysis, belong to the eastern American/ African group. T a framework for additional studies of phylogeny of the portulacaceous alliance and of Centrospermae in general Calandrinia Kunth, in its broadest circumscrip- tion, comprises more than 100 perennial and an- nual herbaceous species in western North America, western South America, and Australia, and rep- resents the most diverse genus in the family (cf. Carolin, 1987, in press; Hershkovitz, 19904, 1991a, b; Kelley, 1973; McNeill, 1974; Nyany- ano, 1986a, 1990; Pax & Hoffmann, 1934; von Poellnitz, 1934a). Considerable evidence has emerged that Calandrinia sensu lato is an unnat- ural genus (Carolin, 1987; Hershkovitz, 1990a, 19913), although differences in opinion remain re- garding its taxonomy (Table 1) and the phyloge- netic relationships of its constituent taxa to one another and to other Portulacaceae. As part of a continuing effort to clarify phylogeny in Portula- caceae, the primary aim of the present paper is to help establish natural circumscriptions of Calan- drinia and its segregates. Because Calandrinia comprises such a phylogenetically divergent assem- blage, however, the scope of this study encom- passes all of Portulacaceae as well as the closely allied families Basellaceae, Cactaceae, Didierea- ceae, and Hectorellaceae. TAXONOMIC HISTORY OF CALANDRINIA The traditional subgeneric and sectional tax- onomy of Calandrinia was developed by Reiche (1897, 1898) and von Poellnitz (1934a) for the s manuscript is revised from a chapter submitted as part of a doctoral d at the University of ' Thi California, Davis. I gratefully acknowledge assistance from Jim Doyle, the late Richar Joan Nowicke, George Russell, Rudy ral John Strother, Grady Hershkovitz, Walt Kelley, June McCaskill, , US. This research re Fellowship, a айаш Pre: doctoral Fellow ous a qoe ced Short-Term Visitor m the е of California, Davis, two travel awards from the Department of Botany, University of California, Davis, a and P. ovitz. Eyde, Donna Ford, Philip ia: B, CAS, DAV, * Laboratory of Molecular a. Museum Support Center, Smithsonian Institution, Washington, D.C. 20560, S.A. ANN. MISSOURI Bor. Ganp. 80: 333-365. 1993. Annals of the Missouri Botanical Garden 334 B . Classification history of members of Calandrinia sensu lato and related taxa. Listed below are the generic and sectional assignments of members of Calandrinia sensu lato and other taxa regarded here as cladistically related (see also Carolin, 1987; Ford, 1992; а 1990a, 1991а, 1992). The table compares traditional classifications with those adopted by Nyanyano (1986a, 1990), Carolin (1987, in press), and the present paper. The traditional classification synthesizes the taxonomy propos by Anón (1953), Kelley (1973), Kelley & Swanson (1986), w (1989), McNeill (1974), Pax & Hoffmann (1934), Peralta (1988), von Poellnitz (1934a), a iche (1897, 1898). This table lists only explicitly proposed synonymy. For example, Nyanyano included Calandrinia axilliflora illares as a synonym of the i.e., the 6a) treatment. Except 19 a species of Calandrinia sect. Cistanthe, but did not include Calandrinia sect. 1898). Th 1987, in press; Kelley, 1973; Reiche, 1897, r species varies among the ds ssifications, particularly in Nyanyano's (19 latter (cf. Carolin, зе synonymy is not perfect, assignment of particular for Rumicastrum sect. Basales, the sectional assignments depart markedly from those of von Poellnitz (1934; see also Syeda of the sections of Rumicastrum are accepted here uncritically. Reiche’s (1897 of Calandrinia are indicated in parentheses. The nomenclatural os is listed the first time each taxon is listed > Australian calandrinias in a 8 ( 7) work & An 1989). The more recent circumscriptions ) subgeneric assignments of the sections in the ta Traditional Nyanyano Carolin Hershkovitz шл sis Kuntze subg. Моптпор е (subg. Hir e Reiche) 8 Calandrinia $ Calandrini- Baitaria § Hirsutae opsi (R sis Nyanyano, ined. eiche) Carolin, ined. Calandrinia Kunth § Cal- andrinia Montiopsis subg. Mon- Baitaria § Condensatae (Reiche) Carolin, ined. bun ae Reiche Calandrinia (Hirsutae) 8 Condensatae tiopsis eiche DENM (Hirsutae) Calandrinia 8 Cistanthe Baitaria 8 Condensatae Montiopsis subg. Mon- arviflorae Reiche tiopsis Calandrinia dns Calandrinia $ Calandrinia Baitaria § Dianthoideae Montiopsis subg. Dian- eiche) $ (Reiche) Carolin, ined. thoideae (Reiche) D. D I. Ford Baitaria 8 Acaules Calandrinia $ Acaules (Reiche) Carolin, ined. (= Baitaria Ruiz a- vón $ Baitaria) Reiche Calandrinia (Glabrae) Calandrinia $ Calandrinia $ Acaules Reiche Calandrinia (Glabrae) Calandrinia $ Calandrinia Calandrinia sens. str. Calandrinia $ Calandrin- andrinia) Calandrinia (Glabrae) $ Axillares Reiche C 'alandrinia (Glabrae) the Reiche Calandrinia (Glabrae) § Arenariae Reiche Calandrinia (Glabrae) § Rosulatae Reiche Calandrinia (Glabrae) Andinae Reiche Calandrinia (Glabrae) Amarantoideae Calandrinia § Basales Calandrinia § Tubero- sa Calandrinia $ Pseudo- dianthoideae Poelln. Calandrinia $ Cistanthe Calandrinia $ Cistanthe Calandrinia $ Calandrinia Calandrinia $ Calandrinia Calandrinia $ Calandrini- Opsis Calandrinia sens. str. Cistanthe $ Cistanthe Cistanthe $ Cistanthe Cistanthe $ Cistanthe Cistanthe $ Cistanthe Cistanthe $ Amarantoi- deae (Reiche) Carolin, ine ита astrum § Basales n.) Carolin, ined. Rumicastrum Poelln ? Carolin, ined. Rumicastrum § Pseudo- Facies (Poelln.) Carolin, ined. (= Rumi- castrum Ulbr. $ Rumi- castrum) Tuberosae Calandrinia 8 Calandrin- га Cistanthe 8 Cistanthe Cistanthe $ Cistanthe Cistanthe $ Cistanthe Cistanthe $ Cistanthe Cistanthe $ Amarantoi- deae (Reiche) Carolin ex Hershkovitz Rumicastrum § Basales Rumicastrum $ Tubero- Rumicastrum $ Rumi- castrum Volume 80, Number 2 Hershkovitz 335 1993 Calandrinia and Montiopsis TABLE 1. Continued. Traditional Nyanyano Carolin Hershkovitz Calandrinia $ Api- — cales Poelln. Calandrinia $ Uniflo- — ae Poelln. Calandrinia 8 Pachy- — podeae Kelley & Swanson Calandrinia § Partitae — Calandrinia $ Cistanthe oelln. Calandrinia $ Macro- car] Anon Calandriniopsis E. Calandrinia § Calandrini- Franz FIT Montiopsis Kuntze Montiopsis Calyptridium Nutt. Calyptridium Spraguea Torr. Calyptridium Philippiamra Kuntze Philippiamra (Silvaea Phil. non Hook. f. & Arn. ex Baillon) Monocosmia Fenzl Calyptridium Lewisia 8 Strophiolum — B. Mathew Rumicastrum spp. Rumicastrum spp. Rumicastrum spp. Rumicastrum spp. = Talinum Adans. spp. — Anacampseros $ Tubero- sae Schreiteria Carolin Schreiteria Baitaria § Condensatae Montiopsis subg. Mon- tiopsis Baitaria § Condensatae Montiopsis subg. Mon- tiopsts Cistanthe 8 Calyptri- dium (Nutt.) Hersh- Calyptridium KOVItZ Calyptridium Cistanthe § Calyptri- dium Cistanthe § Amarantot- Cistanthe § Philippiam- deae ra (Kuntze) Hersh- Monocosmia Calandrinia § Monocos- mi 1) Hers Fenzl) Hersh- ovitz Cistanthe § Strophiolum (B. Mathew) Hersh- kovitz American and Australian species, respectively. Reiche (1897) recognized two subgenera (Glabrae and Hirsutae) and 12 sections of Calandrinia in Chile (Table 1). Reiche (1902; cf. 1897, 1898) assigned the type species, C. caulescens Kunth (= C. ciliata (Ruiz & Pavón) DC), to C. (Glabrae) sect. Compressae (thus, sect. Calandrinia ). Reiche's system was later extended, implicitly or explicitly, to cover extra-Chilean American species by Anón (1953), Kelley (1973), MacBride (1937), and Peralta (1988). Two additional American sec- tions of Calandrinia, sections Macrocarpae (Anón, 1953) and Pachypodeae (Kelley & Swanson, 1986), were later proposed to accommodate spe- cies that did not correspond to any of Reiche's sections. These new sections were not assigned to subgenera. Also, Montiopsis Kuntze and Calan- driniopsis Franz were established on the basis of species referable to subgenus Hirsutae (Carolin, in press; Ford, 1992). Montiopsis was accepted by Pax & Hoffman (1934) and Nyanyano (1986a, 1990) but was not mentioned by Carolin. Calan- driniopsis was accepted by Pax & Hoffmann (1934), but not by Anón (1953), Nyanyano (1986a, 1990), or Carolin (in press). The Australian calandrinias were classified in six sections by von Poellnitz (1934a; see Table 1). One of the sections (sect. Partitae) is actually synon- ymous with Anacampseros sect. Tuberosae (Caro- lin, 1987: 410; not mentioned by Gerbaulet, 1992). The other sections remained in taxonomic limbo for many years: they were subsequently referred to merely as the Australian calandrinias (Kelley, 1973; Pax & Hoffmann, 1934) and were never fully integrated into Reiche's (1897) subgeneric classification. Thus, a later “cladistic analysis of Calandrinia" (Syeda & Ashton, 1989; see also Syeda & Carolin, 1990) included only the Austra- lian species, while a revision of the sectional tax- onomy of Calandrinia (Nyanyano, 1986a; see also below) referred only to the American sections. Within the last six years, three workers have proposed revised supraspecific classifications of all or part of Calandrinia sensu lato (Table 1), in- cluding Nyanyano (1986a, 1990), Carolin (1987, in press), and Hershkovitz (1990a, b, 1991a-c). 336 Annals of th Missouri Botanical Garden Syeda & Ashton (1989) and Syeda & Carolin (1990) did not formally propose a revised sectional taxonomy of the Australian species based on their cladistic analysis. Their analysis will not be con- sidered in detail here except in reference to their data, much of which was incorporated in Carolin's (1987) work. Nyanyano's (1986a, 1990) treatment most closely follows the traditional classification but is unsatisfactory especially because each of the three sections of Calandrinia sensu lato he circum- scribed includes species with traits contrary to the sectional diagnoses. For example, Nyanyano 1986a) characterized section Calandriniopsis as pubescent, yet included therein the glabrous Cal- andrinia sect. Amarantoideae (cf. Hershkovitz, 199 1b; Kelley, 1973); Cistanthe as glabrous, yet included therein the characterized section pubescent Calandrinia sect. Parviflorae (cf. Ford, 1992; Hershkovitz, 1993; Kelley, 1973). Also, contrary to the states found in nuni of the included Nyanyano's (1 generic key diag- -10 sta- mens and no tricolpate pollen (cf. Kelley, 1973), species, nosed Calandrinia sensu liio as having 5 and his tribal key to Portulacaceae (Nyanyano, 1990) diagnosed Calandrineae as pubescent, pe- rennial herbs or shrubs, and thus misrepresents the glabrous and/or annual members (cf. Kelley, 1973; Pax & Hoffmann, 1934; Reiche, 1898). Finally, Nyanyano (1986a) failed to mention the Australian sections of Calandrinia sensu lato, as well as sev- eral American taxa, in his treatment. In his cladistic analysis of Portulacaceae, Carolin (1987) assigned the 18 traditionally recognized sections of Calandrinia sensu lato (excluding sects. Partitae and Pachypodae, which belong to other Table 1; Hershkovitz, 1991c) among 11 s that, in his analysis, showed cladistic genera see relationships with several different genera of Por- tulacaceae. Carolin concluded that Calandrinia sen- su lato was not monophyletic and proposed re- stricting its generic circumscription to include only section Calandrinia (including sect. Axillares; Ta- ble 1). In his shortest cladogram (1987, fig. 7), Calandrinia sensu stricto is a sister group to the monotypic genus Monocosmia: these are, in turn, a sister group to the Australian calandrinias. Caro- lin preferred a slightly longer cladogram in which Calandrinia sensu stricto + Monocosmia is a sis- ter group to his OTU Naiocrene (Montia sensu McNeill, 1975); he attributed the additional length of his preferred cladogram to character state changes that might be morphologically correlated. He concluded that the relationships of the Cal- andrinia sensu stricto + Monocosmia clade were unresolvable without additional study. Carolin (1987, members of Calandrinia sensu lato variously to Cistanthe, Baitaria, Rumicastrum, and Schrei- teria (Table 1). In previous papers, 1 discussed Carolin's (1987) in press) recommended assigning other work in the light of additional evidence from leaf morphology (Hershkovitz, 1990a, 199 1a, b, 1992, 1993). I corroborated Carolin's opinion that Cal- andrinia sensu lato is unnatural and should be segregated into several genera. | adopted Carolin's proposal to classify four of Reiche's (1897) Cal- andrinia sections and the genus Philippiamra (syn. Silvaea) in Cistanthe (Table 1). M of Cistanthe departed from Carolin's in segregating y classification Cistanthe sect. Philippiamra from section Ата- rantoideae and in recognizing Calyptridium (in- cluding Spraguea) and the monotypic Lewisia sect. Strophiolum as sections of Cistanthe (Table 1). I tentatively accepted Carolin’s proposed circum- scriptions of Rumicastrum and Schreiteria, but | presented evidence that the Calandrinia sensu stricto + Monocosmia clade is most closely related to and possibly nested within Calandrinia sect. Acaules, which Carolin referred to Baitaria (Table 1). Hence, I advocated including Monocosmia (see 1991c andrinia, although I found no unequivocal evi- also Hershkovitz, and Baitaria in Cal- dence linking section Acaules with the other mem- bers of Baitaria sensu Carolin (i.e., Calandrinia sect. Dianthoideae and subg. Hirsutae; Table 1). In the present paper, I treat the last two as Mon- tiopsis subgenera Dianthoideae and Montiopsis, respectively (Table 1). PHYLOGENETIC REANALYSIS OF THE RELATIONSHIPS AMONG MEMBERS OF CALANDRINIA SENSU LATO My earlier taxonomic conclusions on Calandrin- ia sensu lato employed cladistic concepts but no formal numerical analysis other than preliminary McClade (Maddison & Maddison, 1987) manipu- lations of data modified from Carolin's (1987). I provide here a more extensive numerical analysis of Portulacaceae that aims primarily to clarify the circumscription and relationships of members of Calandrinia sensu lato. This work synthesizes my earlier reinterpretation (Hershkovitz, 1990a) of several characters included in Carolin’s (1987) analysis, new evidence from leaf morphological studies (Hershkovitz, 1990a, 1991b, 1993), and additional character analyses as noted in Appendix Volume 80, Number 2 1993 Hershkovitz 337 Calandrinia and Montiopsis 1. This analysis falls far short of a comprehensive revised phylogeny of Portulacaceae, especially be- cause this family is probably paraphyletic with re- spect to Basellaceae, Didiereaceae, Hectorellaceae, and Cactaceae (see discussion). Along with Por- tulacaceae, these families constitute what 1 call the ortulacaceous alliance (Hershkovitz, 19914; "portulaca alliance" of Gibson & Nobel, 1986; Portulacineae of Thorne, 1976, 1983). Also, data for the poorly known genera Talinella (see Nyany- ano, 1986b) and Amphipetalum (see Spichiger, 1988) remain to be incorporated, and a new genus, Xenia, was segregated from Anacampseros (Ger- baulet, 1992; see also below) while this manuscript was in review. Except for Hectorellaceae (see Skip- worth, 1961), however, there is no evidence that the excluded taxa are especially close relatives of Calandrinia. The excluded taxa appear to be re- lated to eastern American/ African members of Portulacaceae, whereas members of Calandrinia sensu lato except for section Macrocarpae (Car- olin, 1987) and the Australian endemics, seem to be related to western American Portulacaceae (Hershkovitz, 1990a, 1991a, d; cf. Carolin, 1987; see also discussion). My assumption, at this point, is that neither the eastern American/ African nor western Атепсап group is polyphyletic, and that one group effectively provides an outgroup or set of outgroups for the other. Excluding ант. distinctive members of the eastern Атепсап/ rican group will not prevent at least partial recov- ery of phylogenetic information pertinent to Cal- andrinia sensu lato, although ultimately a more thorough analysis will be necessary. Another prob- lem with the present analysis is that while Portu- lacaceae and allies are clearly members of the order Centrospermae, their position in the order, hence the polarity of many of their variable characters, remains uncertain (Hershkovitz, 1989, 19914; cf. 1984; Carolin, 1987). Rodman et al., MATERIALS AND METHODS 1. Taxa. The 33 OTUs (excluding a hypo- thetical ancestor) recognized in the present analysis (Table 2) are substantially the same as Carolin's (1987) but with changes and precautions described below. I treated Rumicastrum, Claytonia, and Lew- isia as single OTUs based on Carolin’s (1987) finding that each is monophyletic. Even if mono- phyletic, Lewisia is problematic because of its poly- morphism and poorly correlated character states among the species (Hershkovitz, 1992). I will re- TABLE 2. Taxa for cladistic m Listed below are the full names of the OTUs r analysis. The оа of ike taxa are as proposed by Carolin (1987, in press) except where modified by Hershkovitz (see Table 1). Four taxa (Montiopsis subg. Montiopsis, Claytonia, Rumicastrum, and Lewisia) split for analytical purposes by Carolin pud are here lumped ~ because Carolin determined that cognized in the present ach is monophyletic. Traditional synonyms of iE are listed. Anacampseros L. sect. Anacampseros (= А. § Tele- phiastrum (Medik.) Fenzl) 4nacampseros sect. Avonia E. Mey. ex Fenzl mpseros sect. Tuberosae von Poelln. Kunth s pressae Reiche) Ana бана а ect. Calandrinia (= C. sect. Сот- Calandrinia (Acaules) acaulis Kunth Calandrinia (Acaules) affinis Gillies ex it Calandrinia (Acaules) ie у Gillies ex Arn. Calandrinia (Acaules) caro i Hershkovit Fi Ford Calandrinia (Acaules) compacta Barnéoud Calandrinia sect. Monocosmia (Fenzl) Ho Calyptrothec a Gilg Ceraria Pearson & Stephens Cistanthe Spach sect. Cistanthe Cistanthe sect. Amarantoideae (Reiche) Carolin ex Hershkovitz Cistanthe sect. Calyptridium (Nutt.) Hershkovitz Cistanthe sect. Philippiamra (Kuntze) Hershkovitz Cistanthe sect. Strophiolum (B. Mathew) Hershkovitz Claytonia L a Gillies ex Hook. & A:n. Lenzia Phi ., except section Natocrene Montia sect. Naiocrene (Torrey & A. Gray) Pax & К. offm. Montiopsis Kuntze subg. Montiopsis пи и subg. Dianthoideae (Reiche) D. I. Ford Portulaca L. нра Јас Rumicastrum Ulbr. Schreiteria Carolin Talinaria Brandegee Talinopsis A. Gra y Talinum Adans. sect. Talinum (= T. sect. Talinastrum ~ Talinum sect. Phemeranthus (Raf.) DC. consider this problem in my discussion of the re- sults. Although Carolin (1987) determined that Mon- tia is monophyletic, I treated section /Vaiocrene (1-2 spp.; McNeill, 1975) as a distinct OTU. Be- cause it includes the only rosettiform perennial species of Montia, section 'Vaiocrene possibly rep- resents the basalmost clade in the genus. This sec- tion possesses traits that are otherwise discordant 338 Annals of the Missouri Botanical Garden in Montia, however, including brachytetracytic stomata and a chromosome base number of 11 (see Appendix 1). Thus, treating Montia as a single OTU scored for the character states of section Naiocrene is problematic. As noted above, 1 assigned Calandrinia sect. Dianthoideae and subgenus Hirsutae to Montiop- sis subgenera Dianthoideae and Montiopsis, re- spectively. My choice of subgeneric тапк for these s (1897 1898) sectional divisions of subgenus оле entities permits recognition of Heic Reiche divided this subgenus into Calandrinia sects. Hirsutae (perennials), Condensatae (robust an- nuals), and Parviflorae (diminutive annuals); these circumscriptions are at least partially artificial (Ford, The OTUs for Cistanthe are the sections I have recognized elsewhere (Table 1). Except for the addition of C. sect. Strophiolum, these correspond to Carolin's OTUs for the equivalent taxa. My previous analysis (Hershkovitz, 19902) in- dicated that the apparent cladistic affinities of Cal- andrinia sects. might vary depending on how Calandrinia sect. Calandrinia and Monocosmia Acaules was scored for its variable traits. No spe- cies of section 4caules has the entire suite of traits that I had earlier cited as evidence linking this section with section Calandrinia. Hence, I treated as independent OTUs five putative species of sec- tion caules (Hershkovitz, 1993) possessing dif- ferent combinations of character states. I have retained Carolin's two OTUs for Talinum, but they analysis. Elsewhere, three may be unsatisfactory for phylogenetic Hoffmann, 1934 or four (Nyanyano, 1986a) subdivisions of Talinum ~ have been recognized, and it is not clear how Саго- lin would have assigned these to his two sections. Specifically, Carolin (in press) divided Talinum into section Phemeranthus (enveloping seed aril pres- ent; ca. 25 spp.) and section Talinum (enveloping seed aril absent; ca. 6 spp.). Talinum group “ of Pax & Hoffmann (1934: ee 249), more or less equivalent to “sect. Concentricum" of Nyanyano (19862), includes up to 16 species that apparently lack the enveloping aril (Nyanyano, 19862) but, in other respects, span the morphological extremes between the broad-leaved, diffusely branched spe- cies typifying section Talinum and the terete-leaved, rosettiform species typifying section. Phemeran- thus. 1 cannot account for these species in the context of Carolin's classification, but 1 leave this problem unresolved. In addition to the above, I have previously dis- cussed possible paraphyly among the sections of Cistanthe (Hershkovitz, 1991a). Other candidates for paraphyletic OTUs include Calandrinia sect. Calandrinia and Montiopsis subg. Dianthoideae (Hershkovitz, 1990a), in which other OTUs may be nested. For the present analysis, I retained these phenetically distinct but possibly paraphyletic OTUs but will reconsider the specific consequences in the discussion. The potential pitfalls of paraphyly would best be avoided by species-level analysis of the family, which is beyond the scope of the present study. 2. Outgroup. The outgroups of Portulaca- ceae or, more appropriately, the portulacaceous alliance, are unknown (Hershkovitz, 1989, a, 199 1a, in prep.; cf. Carolin, 1987; Rodman, 1990; Rodman et al., 1984); hence most character state polarities cannot be established by the two-out- group method proposed by Maddison et al. (1984). In the present analysis, I designated a hypothetical ancestor scored as “0” for 28 of the 46 total characters with the balance scored as unknown. The character state polarities were subjectively determined on the basis of state distributions among other Centrospermae, which presumes that the por- tulacaceous alliance is derived therein (but see Hershkovitz, 1989). 3. Characters. Appendix 1 lists 46 mostly binary characters and data sources, and Table 3 lists the OTU scores. multi-state/unordered. A “0” is entered for all А few of the characters are states designated as ancestral a priori and, for convenience, states determined to be primitive a posteriori. Otherwise, the state designations are arbitrary. I did not follow infrageneric analyses published elsewhere to derive scores for Rumicastrum or the OTUs of Anacampseros. Contrary to OTU scores | adopt here, Syeda & Ashton (1989) and Syeda & Carolin (1990) presumed that such traits as inaperturate pollen and numerous carpels are prim- itive in Rumicastrum (see Appendix 1). Some scores indicated in Gerbaulet's 92) Anacampseros analysis, published while this manuscript was in review, are contrary to those adopted here. 1 have attempted to note all discrepancies in Appendix 1. Because Gerbaulet’s results are generally consis- tent with those here and also because Anacamp- seros is phylogenetically relatively distant from Calandrinia and Montiopsis, I did not reanalyze using Gerbaulet’s scores. Infrataxon variation was problematic in scoring many OTUs (Table 3). I attempted to score variable OTUs for the presumed primitive state. In some Volume 80, Number 2 1993 Hershk shkovitz 339 Calandrinia and Montiopsis cases, I scored variable OTUs for the derived state of a polarized character, especially where I suspect that the primitive" ample, I scored Rumicastrum for having tuberous derived state in Portulaca- ceae—even though most species are not tuberous. state is secondary. For ex- Tubers in Rumicastrum occur in perennial species (Pate & Dixon, 1982; West, 1986), however, which I presume to be primitive with respect to the an- nuals, although this has not been demonstrated, specifically, in Rumicastrum (see, e.g., Syeda Carolin, 1990). I used an approach suggested by ixon & Davis (1991), viz, dividing polymorphic entities into monomorphic OTUs, to alleviate vari- ation problems in Calandrinia sect. Acaules and, to a lesser extent, in Montia (see above). Never- theless, infra-OTU variation remains for one or more characters (nine in Portulaca) in 18 of 33 OTUs, and character variation in other Centro- spermae is also problematic in my consideration of ancestral states (see Appendix 1). I hope to be able to solve these problems in follow-up analyses. At this point, I just emphasize that such variation exists and note where it might affect the results. Many workers (e.g., Doyle & Donoghue, 1986; Nixon & Davis, 1991; Sanderson & Donoghue, 1989) have recognized that cladistic analyses in- corporating missing data scores can be misleading because the data, if available, might yield alter- native results. Nevertheless, Table 3 includes three sorts of missing data scores, all scored as unknown (^?")in the parsimony program (see below). These include instances where the character state pre- sumably exists in the OTU but has not been ade- quately or at all investigated (^?" in Table 3): where infrataxon variation prohibits a ерее scoring decision for the entire OTU (“ in Table 3); and where “X” characters (Doyle € НЕ 1986) are employed. Doyle & Donoghue (1986) devised "X" acters in order to reduce biases in inter-OTU dis- tances where some OTUs lack a particular char- acter that has two or more unpolarizable states. char- For example, they used “X” scoring for seed shape because radially and bilaterally symmetrical seeds are plausibly derivable from one another, as well as from the seedless condition, by a single evolu- tionary step. А three-state ordered transformation series was not justifiable, and simply using three unordered states (1.е., seeds absent/radiospermic / platyspermic) would bias against the homology of seed presence per se. Treating the different seed types as independent binary characters (i.e., seeds radiospermic vs. not and seeds platyspermic vs. not) would necessitate two steps in the derivation of one type from the other. Doyle & Donoghue treated seed presence/absence as one character and seed shape as another, with seedless taxa scored as unknown (*X") for the latter. In the present analysis, I adopted “X” scoring for leaf fimbrial vein anatomy (char. 14), stomatal subtype (char. 16), trichome ribbing (char. 21), the derivation of solitary flowers (chars. 22, 23), bract morphology (char. 26), capsule valve morphology after dehis- cence (char. 41), and chromosome base numbers when other than nine (char. 46). In addition to uncertain polarities and missing scores, continuous, rather than discrete, character state variation poses a problem for the present analysis (see Stevens, 1991). Indeed, some of the character states here recognized as different, e.g., petal numbers five, rarely six, versus usually more than five, blatantly overlap. Other states merely intergrade, but the scores reflect consistent dis- tinctions between taxa. For example, in Portula- caceae as a whole, festooned brochidodromous leaf venation (char. 12) intergrades with brochidodro- mous, which intergrades with other venation pat- terns (see Hershkovitz, 1990a, 1991a, 1992, 1993). The distribution of this trait in the clado- grams must be evaluated with the variation in mind. | emphasize that the scores employed here rep- resent approximations or, in effect, hypotheses of actual character state differences that must even- tually be validated by the sort of analyses proposed by Stevens (1991). 1. Analysis. I analyzed the Table 3 data ma- trix using PAUP 3.0r (Swofford, 1991) on a Mac- intosh Classic П computer with 6 megabytes of memory (RAM) plus “Virtual RAM" (УКАМ). I allotted 4.5 and 8.5 kilobytes of memory to PAUP for use without and with VRAM, respectiv ish he set MAXTREES at 9,000-10, ae and 17,0 22,000 for the lower and higher memory я ments, respectively. I found that, with my hard- ware, increasing the number of trees in memory beyond 18,000 slows PAUP beyond practical func- tionality. For all searches, | used MULPARS, Lundberg rooting, and the designated ancestral states (ANC- STATES) options. By including the ancestor in the search parameter, | generated rooted trees. As recommended by Swofford (1991), I used a variety of PAUP’s customizable search modes to facilitate parsimony analysis. The options include taxon ad- dition sequences that vary the configuration and length of the initial tree and “heuristic” branch- 340 Annals of the Missouri Botanical Garden b ¿0000 OOTOT 00001 10100 0001X 00001 OXOOT £0000 01000 D142]121[2S à £0000 00000 00000 T0000 OOTTX 00000 ¿XOOT ¿0000 OTOTO UN IISDIVUN Y I 1000X 00000 T0000 00000 0000X 00000 TX000 TOOTO OTOOT Dvovod X 00000 OOTOT 00010 TOOTO 0000X 00000 TXOTO TOOAT т1000 рорјт лов 0 10000 00000 00001 10000 00210 TUTTO 0X000 OTTOO 000TO sisdoijuopy $ sisdonuopy 0 10000 00000 00000 00000 00210 TUTTO 0X000 OTTOO 00010 apapioujuviq $ sisdonuopy 1 TOOTT 00000 00000 1000Х TOTTX 00001 OTTOO OTTOO 00010 2u242010A| $ DNUOJY A TOOTT 00000 00000 TOOOX IOIIX 00000 OTTOO OTTOO 00010 (2u212010N/ $ хә) Dnuo A 10000 00000 OTOTO OTTOO 0001X 00000 0X000 OTTOO 00010 0151727] b ¿0000 00000 00000 0000 ¿TXTX 00000 0X000 àll00 00010 1227 X OTTOO OOTTT 000TO T0000 OTXTX 00000 TXOOT ¿0001 11001 DIUIDYDAL) A TOOTT 00000 00000 0000X TOTTX 00000 OTTOO OTTOO 00010 DIUOJÁD]) b T0000 00000 00000 01001 OOTTX 00000 охото 21100 00010 unjondo4jg $ 2uuisi) à ¿000X 00000 TOTOO 00001 OOTTX 00000 OXOTT ¿TTOO OOTTO piupiddipiyd $ owmisir) 1 10000 00000 00000 00001 OOTTX 00000 OXOTI 11100 00010 9YIUDISI) $ 9YIUDISI) I 10000 00000 00100 00001 OOTTX 00000 OXOIT ¿TTOO OOATO iumptajdAq:) $ ои) 0 10000 00000 00000 00001 OOTTX 00000 OXOTT 21100 00110 apapiopupapurp. $ 2YPUDISI) & 1000X 00000 10000 00000 0000X 00000 TXOTO TOOTO OTOOT D14DA9) à £0000 00001 OTO¿T 15000 0000X 00000 1Х022 ¿0000 00001 19341014 d AD) e T0001 00000 00001 1000Х LXI 00011 00100 оттоо 00110 тимрирјо) $ DIMAPUD]D) 1 T000X 00000 OOTOT 1000X XII 00011 00100 01100 OOTTO опихозоиор $ DIVILPUDID) à £0000 00000 0000T TOOOX ITXIT 00010 0X000 оттоо 00010 Djonduioo тимрирјо) | ¿0000 00000 00001 11000 OTXTT OOOTT 00100 01100 00010 nuio402 тимрио0) Z T0000 00000 00001 TTOOX TIXTT 00017 0X000 01100 00010 nsopidsopo отимриђуо;) à 10000 00000 00001 1100Х TIXII OOOTT 00100 OTTOO 00010 тию viuipurv) à £0000 00000 00001 10000 OTXTX 00001 00100 01100 00010 SINDID тимриојо) X OTTOO 11100 00000 TOOTO OOTTX 00010 1X004 10000 TTOTO әр$оләфт], $ so12sdumopup X OTTOO IU 00000 TOOTO OOTTX 00000 1X004 10001 10000 тиоар $ коло ашрорију X OTTOO їїїїї 00000 TOOTO OOTTO 00010 тхооо 10001 10000 золовзашрорир $ so1osdumopuy X 00000 00055 0000 00000 000¿X 55550 àX040 5500 0005 101500uy 9P Ф 0? ge 08 са oz ST OT SO ALO "(1хој 295) gyep Jussu uey} јацјеј 'лајовлецо 2швїш в војопор у, "езер o[qe1o1djejutun 10 *ajo[duroour ‘Bussu ajouap sx1eur uonsan() 1 xipuaddy ш pequosop әле sojejs pu? злајовлецо) 'аваововјто 4 јој xujeur eq “e TAYI "војеј5 олош JO OM} Zursseduroouo Апраемел uoxeje1jut 9jeorput 521005 paurprapur) uy 'гопарциоо YIM 21028 0] UOX?] оцу ш ојдемел 00] SI јајовлецо ay} jeu] зодопор А V Volume 80, Number 2 1993 Hershkovitz 341 Calandrinia and Montiopsis Continued. TABLE 3. 45 40 OTU 10001 000X1 00000 X1100 01001 00000 00011 Talinaria Talinopsis 01001 01001 X1000 00000 Talinum $ Phemeranthus 11100 00001 2 00000 Talinum § Talinum 40 20 swapping algorithms that vary the extent to which existing trees are modified in the search for new trees. The fastest algorithm, **nearest neighbor in- terchange" (NNI), rearranges branches locally, and ence it may not find shorter trees with radically different topologies. The slowest and most thorough heuristic algorithm, “tree bisection-reconnection" (TBR), rearranges trees radically in its search for shorter trees. Intermediate between these is "sub- tree pruning-regrafting” (SPR). I experimented with all three modes according to the objective of the particular search, but I found that SPR was dis- pensable for my purposes Other PAUP options used here include strict, semistrict, and majority-rule consensus tree com- putation, the last of which also produces a table of clades and their frequency among all of the trees in memory; character change and apomorphy lists that output the character state changes oc- curring for all characters and at all internodes, respectively; the tree filter, which finds among the trees in memory those of particular lengths or topologies; search constraints, which search for trees of particular predefined оре овна in any of the search modes; and “‘starting trees," which per- mits PAUP to swap on one or more specified trees. 'The specific utility of each of these options in the present study is described below in the results. RESULTS From 100 replicates of rapid searches initiated using a random taxon addition sequence, variously TBR or NNI, and 30-tree limits, I found minimal length trees of 107 steps. The consistency index (CI) of 107-step trees is 0.449; the rescaled con- sistency index (RC; Farris, 1989) is 0.346. Be- "missing data" scores and infra-OTU character variation, the actual CI and RC may be lower (Nixon & Davis, 1991; Sanderson & Don- oghue, 1989). Using the present data, 1 filtered from ca. 29,000 previously saved trees over 11,000 107-step trees and none shorter. The ca. 29,000 trees had been generated using a slightly different data matrix. Entering the ca. 11,000 trees as start- ing trees, I used TBR to increase their number to over 15,000 107-step trees with over 13,000 re- maining to swap. I then switched to NNI, which is faster but finds fewer of the possible equally parsimonious trees, and increased the number of trees to over 19,000, with over 14,000 remaining to swap. At this point, I aborted the search because, cause of the from previous experience using similar data sets, I found that such a ratio of trees saved/remaining is an indication that the total number of equally parsimonious trees is at least an order of magnitude 342 Annals of the Missouri Botanical Garden higher, in this case exceeding the memory capacity of both the hardware and the program. Moreover, with this large tree buffer, PAUP was swapping on less than two trees per minute. In order to generate a reliable consensus tree given the memory limitation (i.e., to avoid an over- representation of particular topologies among the total number of trees), I initiated additional search- es using TBR to completion constraining for two particular Ree that were rare among the ca. 19,000 as determine d by the clade frequency table: a constraint for trees swith Claytonia + Montia (2 OTUs) + Lewisia as outgroups to the remaining OTUs (in 43 of 19,000 trees), and one for trees with Calandrinia acaulis + C. caespitosa + C. carolinii forming а clade (38/19,000). These searches yielded a total of over 5,000 additional 107-step trees. Тће strict consensus tree of the total ca. 24,000 107-step trees (saved in three files) is shown in Figure 1. The black, shaded, and diagonally hatched branches delineate the western American, eastern American/ African, and Australian taxa, respec- tively (Hershkovitz, 1991a, fig. 3). Putative posi- tions of Cactaceae, Basellaceae, Didiereaceae, and Talinella (see below) are indicated by dashed branches. The semistrict consensus tree topology 1990) generated from the 17,000-tree treefile was identical to the strict consensus. The consensus tree is not a phylogeny Ен & Tehler, 1990; Maddison, 1989): "length" 1 14— seven steps longer than that a its x cen (Bremer, bifurcate resolutions. It does show, however, which groups are monophyletic in all 24,000 rearrange- ments and all unambiguous consensus character state changes (i.e., excluding those dependent on the particular resolution). The transitions indicated in Figure 1 for characters 15 and 24 appear equiv- оса! but are not: by using constrained searches such as those described below, I determined that resolutions consistent with alternative positions for these transitions do not occur among 107-step trees. For reference, all character state changes de- termined using the accelerated transition (ACC- TRAN) optimization are listed in Appendix 2. Col- lectively, Appendix 2 presents a distorted picture of character state evolution in Portulacaceae in that seven superfluous changes are indicated and many changes would appear differently using al- ternative optimizations. In one or more optimiza- tions, characters not indicated on the consensus tree undoubtedly provide synapomorphies for one or more clades in any of the possible resolutions, e.g., fimbrial vein presence (char. 13) must support at least one subclade in any resolution within the Calandrinia clade. Because of the large number of trees, however, it is not practical to evaluate the character evidence for each permutation of the consensus. The various alternatives for ambiguous or equivocal character state changes not shown in Figure | are not all mutually compatible, i.e., the topology required to assign a particular character state change to a given node of the consensus tree effectively prohibits other changes from simulta- neously occupying that node. I did not attempt to identify a permutation of the consensus tree that would resolve the largest number of ambiguities. From the clade frequency matrices (not shown), | found a wide variety of resolutions of the polyto- mous nodes, but not all possible sister group re- lationships were evident. To determine which of the possible sister group relationships among the unresolved basal 5-tomy were compatible with 107- step trees, I initiated additional searches using ap- propriate constraints, TBR and/or NNI, a random addition sequence with 50 or more replicates, and 30-tree limits. Table 4 shows all compatible relationships and the length of the shortest trees found for incompatible groupings. Table 4 also shows minimal tree lengths generated when con- straining for alternative sister group relationships of the Calandrinia clade, the Montiopsis clade, the Calandrinia + Montiopsis clade, the eastern American/ African clade (see above; cf. Fig. 1), and the Calandrinia + Montiopsis + eastern group clade. I used constrained topology searches in order to evaluate additional phylogenetic hypotheses, e.g., Carolin's (1987) results that showed Calandrinia sections Calandrinia and Monocosmia linked with Rumicastrum or Montia (see above). From more than 50 replicates using a random addition se- quence and NNI, I found minimal length trees of 116 апа 111 steps, respectively, for the alternative relationships. Likewise, searches initiated using ap- propriate constraints to evaluate Carolin's (1987) result showing Baitaria as monophyletic generated minimal length trees of 1 10 steps. I have previously suggested that Calandrinia sect. Acaules is para- phyletic with respect to sections Calandrinia and Monocosmia. Constraining for the monophyly of section Acaules generated 108-step trees. Con- straining for a clade comprising all OTUs formerly assigned to Calandrinia sensu lato excluding those sections that were based on taxa belonging to other genera (e.g., sect. Pachypodeae) generated 122- step trees. Constraining for the monophyly of Ta- linum sects. Talinum and Phemeranthus (cf. Caro- lin, 1987) generated 109-step trees. Finally, be- Volume 80, Number 2 Hershkovitz 343 1993 Calandrinia and Montiopsis Talinaria (1) Anacampseros S Anacampseros Anacampseros 8 Avonia Be & @ d f» Anacampseros § Tuberosae КӘ > wmm Grahamia OF хг E aw Nvv Talinopsis M ww мн cc Cactaceae Ra Portulaca Basellaceae? Talinum $ Talinum Schreiteria ® Talinum $ Phemeranthus E RRA Calyptrotheca - акои Basellaceae? eS es press: Didiereaceae ye «ww: Talinella Io sees Ceraria — a вызна Portulacaria 1 Y Q лез © om Rumicastrum Calandrinia compacta —— Calandrinia acaulis > Calandrinia caespitosa Calandrinia carolinii Calandrinia affinis EE m Calandrinia $ Calandrinia = Calandrinia § Monocosmia AAT Tue Montiopsis $ Dianthoideae әр Montiopsis 8 Montiopsis р би Cistanthe 8 Amarantoideae M —г— Cistanthe 8 Calyptridium = G m Cistanthe 8 Philippiamra 44 Cistanthe $ Cistanthe Cistanthe 8 Strophiolum Claytonia "55o mu Montia S Naiocrene Montia excl. 8 Naiocrene BERS Lewisia = Lenzia FiGURE 1. Consensus tree for pa Shown here is the consensus for 24,000 107-step cladograms found in the present phylogenetic o rn American/ African, ralian, and western American taxa are delineated by, respectively, gra ur ол. hatched, and black | secl Taxa delineated by hyphenated branches were not included in ile computer analysis, but were added in manually (see €— Unambiguous character state changes are shown on the m iy е 1) Polarized characters are shown in black, unpolarized characters are еее in white, and “х” ' denote characters scored as панка (^X" character or missing value) for an OTU on an adjacent branch (cf. Table 3). The positions = the character 15 and 24 transitions appear equivocal but are ы with all known 107-step resolutions of the t Botanical Garden issouri Annals of the M 344 DUO] + 01и0100]7) IYJUDISI) 801 801 + + + + 5 DIZUIT + 801 801 + ... DISINIT 801 801 601 801 801 dno18 “y - B + B + дпола ^q + sisdoiuopy + Diui4punn-) 80I 60 1 60 1 801 80 1 + sisdoiuopy + Diui4punnp-) 801 801 801 801 801 801 pa ia ge sisdonuoyy 801 801 OTI Ill 601 801 oes + srt тимрирјр) UOX?] Duo IYIUDISI) pizua' pisinə] dnois ^q dnoi3 “ӯ sisdo sisdo отпар ѕпшш + 0100] + sisdo -пиоју пио -UD]D/) dno13u] ој) -пиоју + тимр + тимр -и 0107) -UD]D!) (T “BLY ә 51x91 325) dno13 ueoueury иләјѕәм 991 Sursuduroo ape[» e ‘Ajaanoaye ‘st дполаш ayy ‘dno ^q eui Jo ose» oy) UJ 'зезововјп)ло4] JO 1epureura1 әці pu? uoxej paieorput əy} пзомјад drysuonejes dnosZ 191815 e 0] 5лојол | uoxej ѕпшш дполвиј,, (Т ‘Big “yo 1x9} 99s) огоововјпулод jo dnoj3 ueoujy ;/ueousury ulajsea Əy} 01 злојол „Чпола “y,, peyeusisap uoxej әчү ‘с APL вмо[[о} Ашопохеј оџопов ay] 'peieorpur osje st sdiysuonejas dno13 saysts o[queduroour qua sees) Jo qigue| umunu ayy (1 7214 ә) sisÁpeue juasald əy} ш рипој ѕәәц Чә15-/()] әці Зиоше 1n550 zey; sdrysuonejes dnois 29181 зле + .. > . Áq мојад ројопо(ј 's92J1 dajs-) 0T ЧИМ ојаце шоа sdiysuoy eyes dno.13 2915165 `p ATAV], Volume 80, Number 2 1993 Hershkovitz 345 Calandrinia and Montiopsis cause Figure 1 shows two unambiguous reversals to tricolpate from panaperturate pollen (cf. Hersh- kovitz, 1989, 19913), I constrained for trees show- ing the reversal to be at least equivocal. This search generated 108.step trees that are included among those showing the eastern American/ African and western American groups as monophyletic (Table To summarize, for purposes of the present study, the most important results shown in Figure 1 in- clude, first, evidence consistent with my earlier conclusions (Hershkovitz, 1990a, 19912) that Cal- апатита sects. Calandrinia, Acaules, and Mono- cosmia form a monophyletic group. Second, a sin- gle trait (trichome presence, char. 17) supports the sister group relationship between Calandrinia and Montiopsis, although the trichomes in the two gen- era are structurally distinct (Hershkovitz, 1993). In trees only one step longer, Montiopsis has at least six possible alternative sister groups (Table 4). Third, the consensus is consistent with Carolin's (1987) results indicating that Calandrinia sensu lato is unnatural and that Cistanthe is monophy- letic. Fourth, the tree supports my proposal (Hersh- kovitz, 1991a, d) that the western American and eastern American/ African Portulacaceae are phy- logenetically set apart from one another. In this analysis, the latter appear to be monophyletic and the former paraphyletic but equivocally топорћу- letic in trees one step longer (Table : DISCUSSION Although the highly polytomous Figure | con- sensus tree might be dissatisfying for its lack of resolution, it serves the primary purpose of this study, which is to determine the relationships of Carolin's (1987) Calandrinia sensu stricto + Monocosmia clade and, in general, to advance taxonomic understanding of Calandrinia sensu lato. The results of this analysis also provide a frame- work for additional study of more precise relation- ships within the segregates of this genus, as wel as among other Portulacaceae and Centrospermae. A. TAXONOMY AND RELATIONSHIPS OF CALANDRINIA AND MONTIOPSIS The consensus tree supports my earlier conten- tion (Hershkovitz, 1990a) that Calandrinia sects. Calandrinia and Monocosmia are closely related to section Acaules and my hypothesis that the last is paraphyletic with respect to the first two. The paraphyly of section Acaules is evidenced by a single unambiguous character state change (pres- ence of tetracytic/staurocytic stomata, char. 16) and, ambiguously, by another (fimbrial vein pres- ence or subtype (chars. 13, 14). I also emphasize that section Calandrinia is possibly paraphyletic with respect to section Monocosmia. igible autapomorphy for section Calandrinia (cap- sule-valve inrolling, char. 41) is nonuniversal (Ap- e only el- pendix 1) and not readily evaluated in section Monocosmia. While I believe that the results of the present analysis adequately evidence the monophyly of Calandrinia, | do not believe that a cladistic sec- tional taxonomy can be justified in the absence of a species-level analysis, especially because the functional OTUs employed here for section Acaules form a clade in trees only one step longer than the shortest. Because I am unable to offer anything cladistically more defensible, Carolin's (in press) circumscriptions of sections Acaules and Calan- drinia өы sect. Axillares) are here retained. nsistent with my earlier. conclusions tiles 1990a), the present analysis shows Montiopsis as monophyletic and linked only weak- ly with Calandrinia. The monophyly of Montiop- sis Is evidenced by its unique inflorescence (char. 23), assorted multicellular trichomes (chars. 18- 20), and chromosome base number (char. 46). According to Figure 1, the reversal from panaper- turate to tricolpate pollen provides an autapomor- phy for subgenus Dianthoideae. 1 have previously noted (Hershkovitz, 1991a) that reversals from panaperturate to tricolpate pollen indicated in Ca- rolin's (1987, fig. 7) cladogram (for subg. Dian- thoideae and the Portulacaria + Ceraria clade) are actually equivocal. In the present analysis, both are unambiguous but become equivocal by allowing an additional step or, as 1 determined fortuitously, by scoring Calandrinia caespitosa for lacking tri- chomes. For this reason, and because there appears to be no precedent for such a reversal elsewhere among angiosperms, 1 hesitate to accept this trait as an autapomorphy or, without additional analysis, a phylogeny that requires such a reversal. Oth- erwise, | cannot document an autapomorphy for subgenus Dianthoideae, so the possibility that it is paraphyletic with respect to subgenus Montiop- sis must be considered (see also below). Subgenus Montiopsis can be diagnosed on the basis of a unique trichome type that consists of elongated, intertwined cells (Ford, 1992; Hershkovitz, 1993; Kelly, 1973) and also, perhaps, on the basis of its pantoporate pollen. ted above, the sister group relationship between Montiopsis and Calandrinia is supported only by the shared presence of trichomes (char. 17), although the trichomes are structurally dis- 346 Annals of the Missouri Botanical Garden tinct. I have also noted leaf venation and epidermal similarities between Montiopsis and Calandrinia (Hershkovitz, terns in Montiopsis superficially resemble those in 1993). Specifically, venation pat- some species of C. sect. Acaules (although the similarity is in the shared lack of distinctive fea- tures) and tetracytic /staurocytic stomata occur in M. subg. Dianthoideae. Against the weak evidence favoring a Montiopsis/ Calandrinia sister group relationship, however, there is also weak evidence to the contrary. In previous PAUP runs, I inad- vertently scored Calandrinia caespitosa for lack- ing trichomes (although it is variable for this trait, and I scored it correctly for trichome morphology, char. 21). These searches generated the ca. 29,000 (108-step) trees from which the ca. 11,000 107- step trees were filtered using the corrected data (see above). In fact, the 29,000 trees in the earlier search probably represent only a small fraction of the total number of 108-step trees given that par- ticular data file. In any case, with the single dif- ferent score, all of the possible sister group rela- tionships for Montiopsis indicated in Table occurred among minimal length trees. This result renders the trichome evidence for a Calandrinia + Montiopsis sister group relationship even more tenuous because of its dependence on precise re- lationships and character state distributions in Cal- andrinia that cannot, at present, be adequately evaluated. I had previously been inclined to retain Montiopsis in Calandrinia, but given the relative homogeneity of each genus and the possibility that they are not closely related, their segregation seems warranted. Perhaps the only justification for re- taining Montiopsis in Calandrinia would be if M. subg. Dianthoideae proved to be paraphyletic with respect to both Calandrinia and M. subg. Mon- tiopsis. 1 have considered this possibility (Hersh- kovitz, 1990a). but it remains unscrutinized. igure 1 shows Calandrinia and Montiopsis together forming the sister group to the eastern American/ African Portulacaceae evidenced by the shared derivation of panaperturate pollen. As I noted above, this sister group relationship is also contingent upon the homology of trichomes in Cal- andrinia and Montiopsis, and the relationship be- comes equivocal in trees one step longer than min- imal (Table 4). Among the 108-step trees are those showing a sister group relationship between Cal- andrinia and Claytonia + Montia (Montieae), reminiscent of Carolin’s (1987) suggestion of a close relationship between Calandrinia (minus sect. Acaules) and Montia. The latter seems unlikely not only on parsimony grounds, but because either Calandrinia (sensu Hershkovitz) or Montieae must then be artificial. Nevertheless, a Calandrinia/ Montieae relationship deserves additional scrutiny, especially because my data matrix presumes that putative similarities in inflorescence and fruit mor- phology in C. section Calandrinia and Montieae are not homologous (see Appendix See Table 5 for a synopsis of a revised sectional taxonomy of Calandrinia and Montiopsis. B. TAXONOMY AND RELATIONSHIPS OF OTHER MEMBERS OF CALANDRINIA SENSU LATO Figure 1 confirms Carolin's (1987) general re- sult that the remaining members of Calandrinia sensu lato are phylogenetically interspersed among several other traditional genera of Portulacaceae. My earlier proposed circumscription of Cistanthe, which includes members of Calandrinia sensu lato and three to four other traditional genera (Table 1), is upheld. The consensus of the infrasectional relationships differs from, but is not inconsistent with, what I suggested earlier (Hershkovitz, 199 1a, fig. 3). Specifically, my earlier figure shows either section Strophiolum or Cistanthe as possible sister groups of the remainder of the genus. As I noted before, however, section Cistanthe is highly poly- morphic and section Amarantoideae lacks an unequivocal autapomorphy; hence resolution of phylogeny within Cistanthe will necessitate a spe- cies-level analysis. According to Figure 1, the remaining two mem- bers of Calandrinia sensu lato, Rumicastrum and Schreiteria, are members of the eastern Ameri- can/ African Portulacaceae and are not closely re- lated to other members of Calandrinia sensu lato. Rumicastrum is rosettiform, however, and indi- vidual species possess leaf morphological features characteristic of Cistanthe (Hershkovitz, 1991b). Hence, I hesitate to conclude that its position in the consensus tree is correct. My results for Rumi- castrum are difficult to compare with Carolin’s ‚ fig. 7) because of overall topological dif- ferences in our respective trees. Nevertheless, ig- noring the position of Calandrinia sects. Calan- drinia and Monocosmia in Carolin's tree, the closest relatives of Rumicastrum and Schreiteria аге members of the eastern American/ African group. This result is not inconsistent with Figure 1. The possibility that Rumicastrum (or Rumi- which occurs in some trees) is basal in the eastern Amer- castrum + Talinum sect. Phemeranthus, ican/ African group may prove to be significant in resolving the problem of pollen evolution in Por- tulacaceae. To avoid the reversal to tricolpate pol- Hershkovitz 347 Volume 80, Number 2 1993 Calandrinia and Montiopsis TABLE 5. Key to and synopsis of the Calandrinia TABLE 5. Continued. sections and Montiopsis subgenera. The diagnoses are based on data presented in Appendix 1 and Table 3 А. Calandrinia la. Plants perennial, rosettiform, without leafy es or with unicel- stems arising fr apparently racemose or art oe 2-3; seeds several in a 3-valvate capsule 1-2 па Dude ee adt fruit. 2a. Leaves probably always with (sometimes papillate) unicellular trichomes; flowers apparently racemose; cdd 3; seeds b. Leaves with papillate trichomes or N d puis paniculate; stig 1-2 in a 2-valvate, “incompletely debis cent SUA like fru АЗ. C. sect. Мопосозтїа monandra (Ruiz & — DC. One species in Chile and Argen B. деер мык Kuntze, Revin, Gen. Pl. 3(2): 1 1898. M. boliviana Kuntze (= M. cu ingii (Hook. & Arn.) D. I. For B1.M. subg. Montiopsis. About 15 spp. of temperate western South America. B2. M. E ао наев (Reiche E 1. totype, ihe by Ford) ib М ѕрр. of temperate western South Americ len in trees as long as 108 steps, Portulacaria + Ceraria (+ Calyptrotheca) must be the basal branch in the eastern American/ African clade or Portulacaria + Ceraria must occupy the second most basal branch. Thus, with Rumicastrum oc- cupying the basal branch, tricolpy is, at best, equiv- ocally primitive in this group, and the number of topologies that permit tricolpy to be primitive is restricted. C. PHYLOGENETIC RELATIONSHIPS AMONG OTHER GENERA OF PORTULACACEAE A thorough consideration of the relationships among Portulacaceae or, more appropriately, the portulacaceous alliance, is ultimately critical for B. њи resolving the precise relationships among members a. Plants perennial; multicellular trichomes уаг- of Calandrinia sensu lato but is beyond the scope iously uniseriate or multiseriate and glandular of the present paper. Nevertheless, the results of or eglandular (or rarely absent), but never - | formed ‘by. elongate, intertwined һайт cells: the present analysis are at least sufficient to com- pollen tricolpate . ubg. Dianthoideae раге with Carolin's (1987) results and to offer lb. Plants annual or perennial; аткы tri- general conclusions. usly uniseriate or multiseriate igure 1 supports my earlier notion (Hershko- and glandular or eglandular, but trichomes te ТООЛ ut a op Це пите poh formed by elongate, intertwined _ ea ? uas Ру ов ра! i barbed) hair cells always present; poller tween the fundamentally eastern American/ Afri- pantoporate ......... B2. M. subg. lont can members of Portulacaceae and the western American and places the Australian endemic Ru- A. Calandrinia Kunth in HBK, ds «шетт 6: micastrum in the former group. In trees one step . CAULescens 77. 1823. nom. conserv. TYPE: Kunth и conserv.; = C. ciliata (Ruiz & P von) D AI. c sect. Calandrinia. About f temperate western America Ls montane southern jns с emala, and temper- ate South Am A2.C. sect. меи к Reiche, Ber. Deutsch. Bot. Ges. 15: 500. 1897. ТУРЕ: C. acaulis Kunth (lectotype, here designated). About 8 spp. of the Andes region, with one disjunct to 2 Мехісо. АЗ. С. Monocosmia (Fenzl) Hershko- vitz, Phytologia 70: 223. 1991. TYPE: С. longer, the eastern American/ African and western American groups are each equivocally monophy- letic (Table 4). Figure 1 can be compared with Carolin's (1987, fig. 8) area cladogram in which all of the minor clades he recognized except Ки- micastrum can be assigned to eastern America/ Africa or western America. Figure | also герго- duces, albeit with less resolution, Carolin's clade "H," comprising Portulaca, Talinopsis, Graham- ia, Anacampseros, and Тайпага. Gerbaulet (1992) discussed in detail monophyly and inter- relationships among the last four genera (“Ana- 348 Annals of the Missouri Botanical Garden campseros group"; but see discussion of Cactaceae below). My results with respect to relationships within and proximal to the Anacampseros group would have differed using Gerbaulet's (1992) char- acter data, but I suspect that the fundamental conclusions presented here (viz, the circumscrip- поп of Calandrinia sensu stricto, the eastern American/ African vs. western American distinc- tion, and even the relative position of the Ала- campseros group) would remain. Figure 1 shows Portulacaria and Ceraria as sister groups, but this relationship has not been controversial (cf. Carolin, 987, in press; McNeill, 1974; Nyanyano, 1989; Pax & Hoffmann, 1934; Rauh & Schólch, 1965). Because of the pu problems in OTU de- limitation, infra-OTU aracter variation, and character analysis, Figure 1 must be interpreted cautiously. For example, Lewisia comprises spe- cies with character combinations suggestive of oth- er Portulacaceae (Hershkovitz, 1992), including (Kelley & hylogenetic resolution. within species of Calandrinia sect. Acaules Swanson, 1985). Lewisia will establish its primitive character states, and hence clarify its relationships to other genera, Also, in the present analysis | found Talinum to be poly- including Calandrinia and Montiopsis. phyletic and 7. sect. Phemeranthus as a possible The first result is consistent with Carolin's (1987). Nevertheless, both results should be regarded as highly suspect until sister group of Rumicastrum. the putative distinctions between the Talinum OTUs are better articulated. Also questionable is an es- pecially close relationship between Calyptrotheca and Portulacaria + Ceraria. Calyptrotheca in- cludes two arborescent species that otherwise re- semble Talinum sect. Talinum species and differ from Portulacaria and Ceraria in having broad, not especially succulent leaves and showy flowers with numerous stamens and ovules. Figure 1 shows opposite leaves as a synapomorphy of Calyptro- theca with Portulacaria + Ceraria, but subop- posite leaves characterize Talinum paniculatum (Jacq.) Gaertn. (pers. obs.). Consistent with Carolin’s (1987) findings, the present analysis does not support any existing su- prageneric taxonomy of Portulacaceae (e.g., McNeill, 1974; Nyanyano, 1990; Pax & Hoff- mann, 1934), including Carolin’s (in press). Each 1s contains а polyphyletic taxon, of the earlier syste variously Talineae or Calandrinieae, that comprises one-third to one-half of all Portulacaceae genera, including 7alinum and Calandrinia, but usually excludes such taxa as Calyptrotheca and Calan- Whatever its name or drinia sect. Monocosmia. precise composition, this large taxon is misinfor- mative and should be abandoned. Other established suprageneric categories, such as the tribe Montieae (excluding Wangerinia; Carolin, 1987), remain defensible cladistically. Unfortunately, the present results do not suggest an alternative comprehensive classification. Con- ceivably, the eastern American/ African and west- ern American groups could constitute subfamilies, but I hesitate to formalize this because of the lack of additional evidence on their precise interrela- tionship and because the eastern American/ Afri- can group is probably paraphyletic with respect to Basellaceae, Cactaceae, and Didiereaceae (see be- low). Likewise, the unresolved position of Cacta- ceae may confound Carolin's (1987, in press; see also Gerbaulet, 1992) otherwise defensible circum- scription of tribe Portulaceae. Each of the branches from the polytomous nodes in Figure 1 might be As | have noted, however, some of the bifurcately resolved nodes are poorly supported or disputable. By col- lapsing such nodes, the suprageneric taxonomy recognized as a suprageneric taxon. would be nearly redundant with the generic tax- onomy. Even in 107-step trees, а suprageneric classification of Talinum is problematic. Thus, for now, I recommend listing the genera as a single linear series with established monophyletic supra- generic categories (e.g., Montieae and Portulaceae sensu Carolin) referred to only informally. Ulti- mately, I believe that phylogenetic evidence will warrant segregating most or a the western American taxa into a distinct family, expanding the existing circumscriptions of Cactaceae, Bas- ellaceae, and Didiereaceae to include their portu- lacaceous sister taxa, and retaining Portulacaceae only for the monophyletic residue. D. PHYLOGENETIC RELATIONSHIPS OF UNANALYZED MEMBERS OF THE PORTULACACEOUS ALLIANCE I did not include Basellaceae, Didiereaceae, Cac- taceae, Hectorellaceae, Talinella, or Amphipeta- lum in this analysis because I have not yet studied these adequately and because I do not believe that their inclusion would affect the present conclusions regarding the circumscriptions of Calandrinia and Montiopsis. I also did not consider Xenia, which was segregated from Anacampseros (Gerbaulet, 1992) while this manuscript was in review. All but Hectorellaceae probably belong to the eastern American/ African clade as evidenced by their non- rosettiform habit, nonclasping, constricted leaf bas- es, parallelocytic/nonbrachyparacytic stomata, and (although unknown for Amphipetalum) copious mucilage (Cronquist, 1981; Eggli, 1984; Gibson Volume 80, Number 2 1993 Hershkovitz Calandrinia and Montiopsis & Nobel, 1986; Hershkovitz, unpublished; Paliwal, 1965; Rauh, 1956; Bacigalupo in Spichiger, 1988). These taxa will be considered in future work, but a summary of existing information can be included ere. 1. Basellaceae. Basellaceae have nearly al- ways been linked with Portulacaceae (reviewed in Sperling, 1987), especially with Ceraria and Por- tulacaria because of their single-seeded indehis- cent fruits (Pax & Hoffmann, 1934; Sperling, 1987). Basellaceae agree better with Talinum sect. Talinum in leaf venation pattern (Hershkovitz, un- published), more herbaceous habit (at least the leaves and leafy stems), and panaperturate pollen (vs. tricolpate in Ceraria and Portulacaria; cf. Sperling, 1987). Unlike Ceraria and Portulacaria, Basellaceae and Talinum lack epidermal wax crys- tals (Engel & Barthlott, 1988), but this may be a symplesiomorphy. Evidence for a close Basella- ceae- Didiereaceae sister group relationship (Rod- man et al., 1984) is problematic (Hershkovitz, 1989). Actually, Basellaceae and Didiereaceae do share traits, e.g., single-seeded, indehiscent fruits, but in the context of Portulacaceae, these would be, at best, symplesiomorphies (i.e., with Ceraria and Portulacaria). 2. Didiereaceae and Talinella. Didiereaceae resemble Ceraria and Portulacaria in gross form, leaf venation pattern (Hershkovitz, unpublished), floral and inflorescence morphology (Rauh & Scholch, 1965), and epidermal micromorphology (Engel & Barthlott, 1988). Didiereaceae have also been linked with Cactaceae on the basis of sero- logical evidence (Jensen, 1965), vegetative anat- . and graft compatibility. (see Cronquist, 1981; Gibson & Nobel, 1986; Rauh, 1983). Existing arguments are moot if both Cac- taceae and Didiereaceae are interrelated via Por- tulacaceae and their similarities attributed to sym- plesiomorphy. Important differences between Didiereaceae and Cactaceae include perianth mor- phology, the absence of axillary hairs/scales in the former, and their biogeographic remoteness. Di- diereaceae also resemble Talinella in their woody habit, much-branched inflorescences, flowers (Hershkovitz, 19914; cf. Вашоп, 1886; & Hoffmann, 1934). The affiliation of Talinella, a poorly known Madagas- car-endemic genus of 2-4 spp., with Portulacaceae has been questioned (Carolin, in press; Nyanyano, 1986b; Pax & Hoffman, 1934). Talinella is dis- tinctive in having axile rather than free-central placentation, but its combination of betalain pig- and small Carolin, in press; Pax ments, globular sieve-tube plastid inclusions, dich- lamydous perianth, bipartite calyx, and nonbra- chyparacytic or parallelocytic stomata (Behnke et al., 1975; Nyanyano, 1986b) along with normal secondary growth and copious mucilage (pers. obs.) does not occur outside of the eastern American/ African group of the portulacaceous alliance. With- in this group, the balance of its features suggests that 7alinella is closest to Didiereaceae, Ceraria, and Portulacaria. 3. Cactaceae. А close relationship of Cacta- ceae to Portulacaceae has been advocated previ- ously (Chorinsky, 1931; Gibson & Nobel, 1986; Schumann, 1899; Thorne, 1976, 1983) and per- haps indirectly via Didiereaceae (Jensen, 1965; Rauh, 1983; Rauh & Dittmar, 1970), but the position of Cactaceae among Centrospermae (in- deed, among angiosperms) has been at least as controversial as Centrospermae origins and phy- logeny as a whole (see, e.g., Benson, 1982; Bittrich & Hartmann, 1988; Boke, 1964; Buxbaum, 1949; Croizat, 1960; Cronquist, 1981, 1988; Engler, 1925; Erbar, 1986, 1988; Leins & Schwitalla, 1986, 1988; Leuenberger, 1986; Moeliono, 1970; Takhtajan, 1980, 1986; Zimmerman, 1985). Traits of Cactaceae that, compared to other Centrosper- mae, support a relationship with Portulacaceae in- clude the combination of betalains, normal sec- ondary growth, and petals (see Cronquist, 1981). There is general agreement that the leafy genus Pereskia is the most primitive among cacti, but opinions differ as to which species is the most primitive Pereskia (compare, e.g., Gibson & Nobel, 1986; Leuenberger, 1986). Both Leuenberger (1986) and Gibson & Nobel (1986) presumed that one or another arborescent, tropical, northern South American Pereskia species was primitive in the genus, although arborescence in the portulaca- ceous alliance is otherwise restricted to African taxa and only a few otherwise widely distributed species of Portulacaceae occur in northern South America. Leuenberger (1986) presumed that a shrubby (vs. arborescent) habit, tuberous roots, and stem stomata are derived traits within Pereskia, yet all occur among eastern American/ African Portulacaceae (Carolin, in press; pers. obs. for stem stomata, e.g., in Talinum spp.). The presence of axillary hairs and scales in Cactaceae (Chorinsky, 1931; Schumann, 1899) suggests a relationship to the clade comprising Por- tulaca, Anacampseros, Talinaria, Talinopsis, and Grahamia (Fig. 1). Among these, the poorly stud- ied monotypic, Patagonian genus Grahamia re- sembles Pereskia most closely. Grahamia is a 350 Annals of the Missouri Botanical Garden woody scrambler with proleptic shoots that ter- minate in a single showy flower subtended by а whorl of somewhat spinescent sepaloid bracts (Ca- rolin, in press; Pax & Hoffmann, 1934; cf. Leuen- 1986). A close Grahamia- Pereskia re- ced by traits Оа оповіши the monophyly of Gerbauler s (1992) “Anacampseros group,’ berger, lationship might be countereviden as well as by the chro- mosome base number 11 in Cactaceae versus nine in Carolin's Portulaceae. A more detailed evalua- tion of the evidence on cactus origins, however, is beyond the scope of the present paper. 4. Hectorellaceae. At present, | am unable to discern any known features of the geophytic, New Zealand endemic Hectorellaceae that would allow its unequivocal assignment to either the east- ern American / African or western American groups 1986a; Skip- worth, 1961). Hectorellaceae are copiously mu- 1975) with apparently anomocytic (Skipworth, 1961), brachyparacytic, of Portulacaceae (cf. Nyanyano, cilaginous (Yoong et al., or even cylclocytic stomata (Hershkovitz, unpub- lished). The leaves are highly reduced and spe- cialized, however, so it is difficult to compare their morphology with leaves of other Portulacaceae. 5. Amphipetalum. | Bacigalupo (in Spichiger, 1988) proposed a close relationship between 4m- phipetalum and Calyptrotheca because they share 6-valvate acropetally dehiscent capsules. This trait has been reported to occur in Talinum (Talinum) rivae Chiov. (von Poellnitz, 1934b; cf. Carolin, in press), but perhaps this Somalian endemic actually belongs to Calyptrotheca. Acropetally dehiscent 3-valvate capsules occur in Talinum paniculatum however (Galati, 1986). The her- baceous habit, small size (< 0.5 (Jacq.) Gaertn., m high), sparsely or unbranched inflorescence, numerous ovules (> 10), and = 40-porate pollen of Amphipetalum are contrary to the traits of Calyptrotheca and suggest a closer relationship to Talinum section Talinum spp. (Carolin, 1987, in press; Nyanyano, 1986a; Pax & Hoffmann, 1934), although the pollen pore number of Amphipetalum exceeds that reported in any species of Talinum. Amphipetalum shares with Montiopsis an indumentum of uniseriate, mul- ticellular trichomes. E. WHAT IS PRIMITIVE IN THE PORTULACACEOUS ALLIANCE 2 The consensus tree (Fig. 1) is disappointing in its failure to identify which taxon or taxa form the sister group of the remainder. Table 4 shows that several resolutions are possible among the 107- step trees, and many more emerge when an ad- ditional step is allowed. But this result is possibly as informative as it is uninformative, 1.e., it shows that the distributions of putatively primitive traits in the family are poorly correlated. For example, tricolpate pollen does not co-occur with the chro- mosome base number nine (Appendix 1; cf. Table 3). The paradoxical discordance of character state distributions can be appreciated in view of the following speculative model of portulacaceous evo- lution based on biogeography. Biogeographic evidence might suggest that the eastern American/ African group is more ancient than the western American. The former group, with or without Rumicastrum, is distributed in three Gondwanan continents and Madagascar. Bas- ellaceae (aside from cultivated and/or weedy spe- cies; see Sperling, 1987), Cactaceae, Talinum, Ап- acampseros / Talinaria, Portulaca, and possibly Amphipetalum/Calyptrotheca are disjunct be- tween America and Africa. Portulaca and Ana- campseros are also disjunct to Australia (Carolin, 1987; Geesink, 1969; Pax & Hoffmann, 1934). Except for the African species of Rhipsalis, the disjunct taxa do not appear to be specialized for long-distance dispersal, which suggests that the eastern American/ African group may have differ- entiated during a period when the flora was more easily interchanged, i.e., ca. 50 million years ago or earlier (cf. Gibson & Nobel, 1986; Leroy, 1978; Raven, 1983; Raven & Axelrod, 1974). Endemism in the western American group is almost entirely limited to generally montane and/or desert regions of temperate western America. The exceptions in- clude a distinctive species of Montia in Australia/ New Zealand (Nilsson, 1966a) and other eastern American or Eurasian species of Montieae that are otherwise closely related to western American spe- cies (cf. McNeill, 1975). That Calandrinia (Cal- andrinia) ciliata is native to Australia (cf. von Poellnitz, 1934a) has not been confirmed (Kelley, 1973; cf. West, 1986). Considering that the geo- physical events responsible for creating the pres- ent-day ecological attributes of temperate western America are believed to have occurred within the last 15 million years, one might conclude that the western American group differentiated much more recently than the eastern American/ African group (cf. Raven & Axelrod, 1978; Solbrig, 1976). More- over, the North American taxa of this group are believed to have been derived from the South American (Raven & Axelrod, 1978; but see also Hershkovitz, 19914). Contrary to the biogeographic scenario, the cla- distic results suggest that the western American Volume 80, Number 2 1993 Hershkovitz 351 Calandrinia and Montiopsis group is as old as if not older than the eastern American/ African group (Fig. 1). The positions of individual taxa in the consensus tree are sometimes no less counterintuitive. For example, Portulaca might seem to be relatively primitive among Por- tulacaceae: It occurs on three southern continents, most species have the typical centrospermous chro- mosome base number nine, and some species have the putatively primitive inflorescence type for the family (Carolin, 1987; Geesink, 1969, 1987). Por- tulaca is epigynous and has C, photosynthesis, but these are likely autapomorphies. Yet, in Figure 1, Portulaca emerges as derived among Portulaca- ceae because of its panaperturate pollen and ax- illary hairs and scales. The only tricolpate members of the eastern American/ African group are Cera- ria and Portulacaria, which otherwise appear de- rived in their growth form and fruit morphology. Many western American taxa have tricolpate pollen and, among these, species of Lewisia have di- chasial inflorescences. Indeed, the present analysis showed Lewisia as a possible outgroup of the re- mainder of the analyzed taxa. Yet, Lewisia is char- acterized by a derived fruit type, and the species with dichasia have such derived traits as glandular- dentate sepals, laterocytic stomata, and chromo- the dichasial species may be derived within Lewisia; Hershko- vitz, 1992; cf. Mathew, 1989). Moreover, the ge- nus is restricted to western North America. some base numbers of 12, 14, 15 (i.e., rom the above comparison, it is clear that Portulacaceae exhibit a high degree of what Takh- tajan (1991 and citations therein) calls hetero- bathmy, i.e., a disharmonious distribution of prim- itive and derived traits. Takhtajan's observation that high degrees of heterobathmy characterize grade groups is perhaps applicable here in view of the probable paraphyly of Portulacaceae. Alter- natively, the perception of high heterobathmy may reflect conceptual constraints as to which traits are necessarily primitive. For example, if enough рћу- logenetic evidence accumulates that the chromo- some base number nine and perhaps even tricolpy are advanced conditions in various Portulacaceae, the degree of perceived heterobathmy in the group would be reduced. F. THE PORTULACACEOUS ALLIANCE AND CENTROSPERMAE PHYLOGENETIC RESOLUTION The results of the present analysis accord with my earlier admonition that progress in resolving Centrospermae phylogeny is constrained by poor understanding of phylogeny within the included traditional families (Hershkovitz, 1989, 1991f). Higher-level phylogenetic analysis requires com- parison of demonstrably or at least plausibly prim- itive character states of polymorphic taxa. For Portulacaceae, a set of putatively primitive char- acter states cannot at present be established be- cause different polarities for several characters re- main equally defensible. Moreover, the results presented here indicate that some traits present among Portulacaceae and typical of possible cen- trospermous outgroups, such as the chromosome base number nine, cannot be presumed primitive in the former. Finally, the likelihood that other Cacta- ceae, Basellaceae, Didiereaceae, and Hectorella- ceae, are phylogenetically nested within Portula- caceae suggests that phylogenetic analysis among the last must precede advances in the understand- ing of the first three. traditional centrospermous families, i.e., The hindsight provided by formal cladistic anal- ysis is not required to appreciate these precautions. Irrespective of the taxonomic system followed, even the most rudimentary analysis of character vari- ation among Portulacaceae should reveal the extent of polymorphism at lower taxonomic levels and indicate that the primitive character states for the family are not readily discernable. Likewise, close relationships of Cactaceae, Basellaceae, Didierea- ceae, and Hectorellaceae to specific members of Portulacaceae rather than to the family as a whole were explicitly proposed in classical literature. The portulacaceous alliance is not unique in its degree of phylogenetic irresolution and possible paraphyly of familial and lower level taxa: similar taxonomic problems characterize other Centrospermae as well (Hershkovitz, 1991e, This emphasis on problems of polymorphism and paraphyly should not imply that portulacaceous and/or Centrospermae phylogeny are hopelessly complex and unresolvable. Rather, inroads are pos- sible by preliminary resolution of such problems at the lower taxonomic levels, such as that presented here for Calandrinia sensu lato. Additional prog- ress in resolving portulacaceous phylogeny will be facilitated by similar analysis of other polymorphic taxa, e.g., Talinum and Lewisia. Collectively, such studies will help consolidate the portulacaceous OTUs, thereby facilitating analysis of Centrosper- mae as a whole. Naturally, scrutiny of lower level taxa in other Centrospermae will also be required. CONCLUSIONS Although phylogeny of Portulacaceae remains poorly resolved, the present cladistic analysis sup- ports the segregation of Montiopsis, Cistanthe, 352 Annals of the Missouri Botanical Garden Rumicastrum, and Schreiteria from Calandrinia. Calandrinia should be restricted to sections Cal- andrinia, Monocosmia, and Acaules. The first two may be cladistically nested in the last, and section Monocosmia may be nested in section Calandrin- ia. Montiopsis is a monophyletic genus with two subgenera, Montiopsis and Dianthoideae, and the latter may be paraphyletic. Calandrinia and Mon- tiopsis are possibly sister groups but uncertainty to this effect and the distinctions between the two genera warrant their separation. The present anal- ysis also supports my earlier circumscription of Cistanthe (Hershkovitz, 1991a) and hypothesized distinction between the eastern American/ African and western American members of Portulacaceae. Otherwise, relationships among Portulacaceae and members of this family to Basellaceae, Cactaceae, Didiereaceae, and Hectorellaceae remain poorly resolved. The number and magnitude of unresolved problems preempt, for now, the reliability of fa- milial-level ру овепоне analyses of Centrosper- e (cf. Rodman, ) N at present, more approachable than others, h ome of the problems ar nalyses r of the more fe polymorph genera. LITERATURE CITED ANDERBERG, A. & A. TEHLER. 1990. Consensus trees, a necessity in taxonomic practice. Cladistics 6: 399- 402 ARÓN, D. C. (SUAREZ DE CULLEN). 1953. Las | Argentinas e Lows Calandrinia. Bol. . Аг- ч 9. 1986. Histoire des Plantes, Volume 9. . Hachette, Paris. BECKER, C. 1895. Beitrag zur vergleichenden Anatomie der Portulacaceen. Doctora oo Universitat Erlangen, Erlangen, m BEHNKE, H.-D., . МАВ E P- -type sieve nen plastids and betalains in Portulacaceae (including Ceraria, Portulacaria, and Talinella). Canad. J. Bot. 53: 2103-2109. Benson, L. 1982. The Cacti of the United States and Canada. Stanford Univ. Press, Stanford, California. BERNARDELLO, L. M. . The chromosomes of Gra- hamia (Portulacaceae). Pl. Syst. Evol. 163: 127- 1 I Em T & L. Por 131. BrrrricH, V. & Н. E. K. HARTMANN. 1988. The Ai- oaceae— a new approach. J. Linn. Soc., Bot. 97: 239-254. & RUCK. 1989. What is primitive in Mesembryanthemaceae? An analysis of И polarity of character states. S. African J. Bot Г. Boke, №. Н. 1964. The с m a new in- terpretation. Amer. J. Bot. 51: 598-610. BREMER, К. 1 Cea E i consensus. Cladistics 6: 369-372. BuxbBaum, F. 1949. Vorläufer des Kakteen- о bei den Phtyolaccaceen. Oesterr. Bot. Z. 96: 5-14. CAROLIN, R. C. A review of the family Portu- lacaceae. Austral. J. Bot. 35: 383-412. Portulacaceae. In: K. Kubitzki ‘editor), Fam- a of Vascular Plants, Volume 2 tersu chung der Haar- gebilde bei Portulacaceen ond Cactaceen. er Bot. Z. 80: 308-327. CORNER, E. J. H. 1976. The Seeds of Dicotyledons. Cambridge Univ. Press, Cambridge, England. a L: . Principia Botanica. Salland De- Venter, The Netherlands. Gon пт, A. An Integrated System of Classi- n of Flowering Plants. Columbia Univ. Press, New Yor — o Я 1988. The Evolution and Classification. of Flow 8 Plants, 2nd edition. New York Botanical G , New Pe rk. eJ. us. HUE. 1986. Seed plant eny i the origin of angiosperms: an exper- imental cladistic approach. Bot. Rev. (Lancaster) 52: 321 ECKARDT, T. 1976. Classical Pug Тетир of rgo и plants. Pl. Syst. Evol. Eccu, U. 84. Stomatal types of ao PL t ph 197-214. ur NDORFER, F. 1 romosome numbers and dif- ntiation of centrospermous families. Pl. Syst. Evol. 21-30. ENcEL, T. & W. BARTHLOTT. 1988. Micromorphology of epicuticular waxes in centrosperms. Pl. Syst. Evol. 161: 71-85 ENGLER, А. 1925. Reihe Opuntiales: Historische Ent- wicklung der Ansichten über die systematische Stel- lung der Reihe. Pp. 592-594 in A. Engler & K Aes (editors), Die Natúrlichen Pflanzenfamilien. .À . Engelmann, Leipzig. г. С. 1986. Untersuchungen zur Entwicklung der spiraligen Blüte von Stewartia Ио (Theaceae). Bot. Jahrb. Syst. 106: Early developmental е in flowers and their value for systematics. Pp. 7-23 in P. Leins, S. C. Tucker & P. K. Endress (editors), Aspects of Floral Development. Cramer, Berlin Farris, J. S. 1989. The retention index and the rescaled consistency index. Cladistics 5: 417-419. Fonp, D. I. 1992. Systematics and Evolution of Mon- tiopsis subgenus Montiopsis (Portulacaceae). Ph.D. Dissertation. Washington’ University, St. Louis, Mis- souri. 993. New combinations in Montiopsis. Phy- tologia 74: 273-278. СагАТІ, B. С. 1986. Ontogenia del fruto de Talinum paniculatum (Jacq.) Gaertn. (Portulacaceae). Paro- diana 4: 123-131. GEESINK, R. 1969. An account of the genus Portulaca in Indo-Australia and the Pacific (Portulacaceae). Blumea 17: 275-307. 1971. Portulacaeae. Pp. 121-133 in C. С. G. J. Van Steenis (editor), Flora Malesiana, Series 1, Volume 7. Wolters-Noordhoff, Groningen, The Netherlands. 1987. Theory of classification of organisms. Pp. 91-126 in E. F. de Vogel (editor), Manual of Herbarium Taxonomy: Theory and Practice. UNESCO/MAB, Jakarta GERBAULET, M. 1992. Die Gattung Апасатрзегоз L. Volume 80, Number 2 1993 Hershkovitz Calandrinia and Montiopsis 353 (Portulacaceae). I. deis un zur Systematik. ot. Jahrb. Syst. 113: 477-564. GIBSON, A C. & P. S. NoBEL. 1986. The Cactus bias Harvard Univ. ^er Cambridge, Massachusett ен) М. 19 Phylogenetic i. in Ce мање a brief appraisal. Taxon 38: 602- 1990a. Phylogenetic and Morphological Stud- ies in Portulacaceae. Ph.D. Dissertation. University of California, Davis. se le in Portula- caceae. Phytologia 68: ———— cone aa and revised circumscription of Cistanthe RUM (Portulacaceae). 021. Ann. bes issouri Bot. Gard. 78: 9]b. Leaf iie idle pa Cistanthe Spach (Portul ). Ann. Missouri Bot. Gard. 78: 1022- 60. 1991c. Taxonomic notes on Cistanthe, Cal- andrinia, and Talinum (Portulacaceae). Phytologia 70: 209-225 More Centrospermae I: the portula- caceous alliance. Amer. J. Bot. 68(6, pt. 2): 191. [Abstract.] 1991e. More нерее П: the phyto- laccaceous ECN and the chen-ams. Amer. J. Bot 78(6, cd : 191. [Abstract.] a More Centrospe: mae as the cary- тш alliance, Aizoaceae, odd taxa, and sum- mary. Amer. J. Bot. 78(6, pt. 2): 192. ez 1 Leaf morphology and taxonomic anal. ysis of Cistanthe tweedyi (nee Lewisia tweedyi; Portulacaceae). Syst. Bot. 17: 220-238 . 1993. Leaf morphology of Calandrinia and Montiopsis (Portulacaceae). Ann. Missouri Bot. Gard 80: 366-396. JENSEN, U. 1965. Seroligische Untersuchungen zur Frage der оаа са der Didierea- rb. Syst. 84: 233-2 1973. Pollen ea m and Rela- tionships i in the Genus Calandrinia H.B.K. (Portu- lacaceae). M.S. Thesis. California State University, Northridge. & J. R. Swanson. 1985. Oreobroma megarhi- zum is a calandrinia of Section Acaules (Portula- caceae). Brittonia 37: 56-57. & . 19 new section and subspe cies of Calandrinia (Portulacaceae). Phytologia 60: 171-17 79. y P. & 5. SCHWITALLA. 1986. Studien in Cacta- twicklung von Pereskia. Beitr. Biol. Pflanzen. 60: 23. & . 88. Placentation in Cactaceae. Pp. 57-68 in P. Leins, S. C. Tucker K. Endress (editors), Aspects of Floral Development. Cramer, Berlin. Leroy, J.-F. 1978. Composition, origin, and affinities of the Madagascan vascular flora. Ann. Missouri Bot. Gard. 65: 535-589. LEUENBERGER, B. E. 1986. due uc Mem. New York Bot. Gard. 41: МАСВЕШЕ, J. F. 1937. Po с ня Pp. 562-573 in ]. Е. Macbride (editor), Flora of Peru, Volume 13, Part 2, [Publ. Field Mus. Nat. Hist. Chicago, ot. Ser. Mappison, W. P. 1989. Reconstructing character evo- lution on polytomous cladograms. Cladistics 5: 365- aT, . R. Mappison. 1987. MacClade: а phy- logenetics program distributed by the authors. Pri- vately distributed. ‚ M. J. DONOGHUE & D. К. Mappison. 1984. Outgroup analysis and parsimony. Syst. Zool. 33: 103. MarHEW, B. 1989. The genus Lewisia. The Royal Botanic Gardens, Kew, England. MCNEILL, J. 1974. Synopsis of a revised classification of Portulacaceae. Taxon 23: 725-728. A generic revision of Portulacaceae tribe Montieae using techniques of numerical tax- onomy. Canad. J. Bot. 53: 789- | TT: The genus Talinaria. Southwest Nat. 22: -397. 394 METCALFE, C. R. & L. CHALK. 1950. Anatomy of the п Clarendon du Oxford. Мивү, Т. Н. 1980. Studies in the floral anatomy of Cayan l Amer. J. Bot. 67: 1046- 0. nu B. M. Cauline or Carpellary Placen- tation among Dicotyledons, Koninklijke Van Goreum Co. N. V., Assen, The Netherla FEN NiLSSON, O. 1966a. Studies in Montia L. and Claytonia d allied genera. I. Two new genera, Mona and Paxia. Bot. Not. 119: 265-285. 1966b. Studies in Montia L. II. Some chro- mosome numbers. Bot. Not. 119: 464-468. 1970. Studies in Mob L. and Claytonia L. and allied genera. IV. The genus Crunocallis Rydb. Bot. Not. 123: 119-148. 71а. Studies in Montia L. and Claytonia Е. and alied genera. V. The ee Montiastrum (Gray) els Bot. Not. 124: 87-1 ——. 1971b. Studies in Montia E p Claytonia E A allied genera. VI. The genera Limnalsine 8 са and Maxia О. Nilss. Bot. Not. 124: 187- TE K. C. € J. I. Davis. 1991. Polymorphic taxa, missing values, and cladistic analysis. Cladistics 7: 233-241. NYANYANO, B. L. . Pollen morphology in Talineae (Portulacaceae) sensu McNeill, 1974. Biologia Af- ricana 2: 41-47 1986a. Tribal and Generic Relationship and Classification of the Portulacaceae (Centrospermae). Ph.D. Dissertation. University of Reading, Reading, England. 1986b. The taxonomic position of Talinella iPostülusmoena). Feddes Repert. 97: 771-773. 1986c. Taxonomic significance of the sto mata aa in the Portulacaceae. Feddes Repert. 97: 989. Тахопоту of the endemic African gen- era a the Portulacacea [sic] (Centrospermae). Feddes. Repert. 100: 235-239. . Tribal and generic relationships in the Pastulacecess (Centrospermae). Feddes Repert. 101: 41. PaLiwaL, С. S. 1965. The development of stomata in Basella rubra Linn. Phytomorphology 15: 50-53. РАТЕ, J. S. & К. W. Dixon. 1982. Tuberous, Cormous, and Bulbous Plants. Univ. Western Australia Press, Nedlands, Western Australia. Pax, F. € K. HorFMANN. 1934. Portulacaceae. Pp. 354 Annals of the Missouri Botanical Garden 34-262 in A. Engler & K. Prantl (editors), Die Natürlichen Pflanzenfamilien. Engelmann, Berlin. PERALTA, I. E. Sinopsis de las especies де Cal- је им Meu ај de los Andes Mendocinos. Bol. 5 dee лије Bot. 25: 511-537 POELLNITZ, К. уох. 1933. ы чен L. prs einer Monographie Bot. Jahrb. Syst. 65: 382-4 Die о Агїеп i der Fees d 35: 161-173. 4b. Monographie = Gattung Talinum Adans. “Fedde Repert. 3 RAUH, отуде е schichtliche, histogenetische und anatomische Un tersuchungen an den Sprossen der тенни iae mov nos Akad. Wiss., Math.-Naturwiss. КІ. = The morphology and systematic роз- tition of the Didiereaceae of Madagascar. Bothalia 43. TIMAR. 1970. Weitere Untersuchun- ische Untersuchungen an den SproBachsen und « Dornen der Didiereaceen. Sitzungsber. Heidelberger Akad. Wiss., Math.-Naturwiss. Kl. 1969/1970(4): 5-88 Н. Reznik. 1961. Zur Frage der syste- matischen Stellung der Didiereaceen. Bot. Jahrb. Syst. 4-105. H.-F. Scuórcu. 1965. Weitere Untersuch- ungen an Didiereaceen. 2. Teil. Infloreszenz-, blü- tenmorphologische und embryologische Untersu- chungen mit Ausblick auf die systematische Stellung der Didiereaceen. Sitzungsber. Heidelberger Akad. Wiss., Math.-Naturwiss. Kl. 1965(3): 6-218. Raven, P. Н. e migration and evolution of floras in the Southern Hemisphere. Bothalia 14: 325 22 & D. A. AXELROD. 1974. Angiosperm bioge- ography and the past о movements. Апп Missouri Bot. Gard. 61: 539-673. ps rigin and PE NE of California Publ. Bot the California flora. Uns. 1-134. REICHE, К. 1897. Zur systematik der | Ar- n der аши Calandrinia. Ber. Deutsch. Bot Ces. 15 03 398. Flora de Chile, Volume 2. University of Chile, Santiago 1902 Flora de Chile, Volume 3. University of Chile. Santiago. Marc да E. 1990. Centrospermae revisited, Part I. n 39: 383-393. . K. Ouivrn, R. R. Nakamura, J. U. CL AMMER, JR. & A. Н. BLEDSOE. 1984. А taxo- nomic analysis and ae шыш of Centro- spermae. Syst. P 9:297- о. M. J. & M. J. a HUE 89. Patterns variation in levels of homoplasy. Evolution 43: 1781-1795. SCHUMANN, K. M. 1899, Die V ler Cactaceae im Verhaltniss zu ihrer systematischen Gliederung. Kónigl. Akademie der Wissenschaften, Berlin 1899. [Abh. Konig]. des "ra Wiss., Phys.-Math. Kl. 1899 (Anhang): 1 4.] SHARMA, H. P. 1954. Studies in the order Centrosper- males. I. Vascular anatomy of the flower in certain species of the Portulacaceae. J. Indian Bot. Soc. 33 8-111. SkipworTH, J. P. 1961. The taxonomic ae o Hectorella caespitosa Hook. s. Roy. S New Zealand, Bot. 1: 17-30. SOETIARTO, S. R. & E 1969. Ontogenetical айд shoot apex into the floral apex. Canad. J. Bot. 47 133-140. SoLBRIG, O. T. 1976. The origin and floristic pua of the South American e desert and s esert regions. p. 7-50 in . Goodall ud. Evolution of Desert Biota. Univ. a Press, Austin. SPERLING, С. 1987 y Ph.D. Dissertation. Harvard University, Cambridge, Massachusetts. pu m R. 1988. Notulae ad floram paraquaiensem, . Candollea 43: 409-420. STEBBINS, "s L. 1974. Flowering dedi Evolution above of Harvard Univ se Character states, morphological variation, and en analysis: a review. Syst | 553-583. Bot. 16: SWOFFORD, D. L. Ten PAUP: | Analysis Using Parsimony, Version 3.0[r]. Computer program distributed by ме ишн Natural His story Survey, UN 1989. A cladistic а of бана (Portulacaceae). Pakistan J. Bot. 158-169 C. CaROLIN. 1990. Phylogeny and origin of the Nore Calandrinia (Portulacaceae). Pa- 22: 75-7 Outline of the classification of flowering plants aoe Bot. Rev. (Lan- caster) 46: 225-3 198 ee Magnoliophyta. Soviet Sci- Leningrad. [In Russian. Evolutionary Trends in Plants. Colum- bia Univ. Press, New York RE R. F. 1976. А DEED oo of ag o ae ne Evol. Biol. 9: 35 1983. Proposed new a in the an- iosperms. Nordic J. Bot. 3: 85-117 Wene RLING, P. 1 Morphology of Flowers and In- P 9. florescences [English translation by R. J. Pankhurst]. ences Press, 19 ulacaceae. P. Jessop & H. R. Toelken (editors), Flora of South gend 4th edition, Volur South Australia Government Printing Division, M RE Yoonc, N. S., LIPSON & J. L. WALKER. сыа Hectorellaceae — a member of the Centro- . New Zealand J. Bot. 13: 567-57 sper ia. A. D. 1985. Systematics of Seguieria (Phytolaccaceae?) and Pereskia (Cactaceae). Amer. Bot J: . 6: 977. [Abstract.] Volume 80, Number 2 1993 Hershkovitz 355 Calandrinia and Montiopsis APPENDIX 1. Characters and character states for ролине, ние ык аге me characters, analyses, and states l alysis e scores for the taxa are provided ; in Table Where и effective polarities, 1.е., characters hen "0" hypothetical ancestor, are indicated. The “or data also denotes a posteriori determined primitive states. Oth- erwise, the state designations are arbitrary HABIT (CHARS. l- Portulacaceae are sorted here according to their wood- iness, presence of a basal leaf rosette, and phenology. Other Centrospermae are variously woody or herbaceous; therefore, the polarity of this character is proe, al- ondary niii. (Cronquist, 1981). Portulacaceae can be largely sorted according to those that produce a prom- inent bod or suprabasal rosette in which axillary, T eral aerial flowering stems arise versus those in w aerial stems proliferate vegetatively. Even these two cat- , e.g., in Montiopsis. Studies of чеш © development may permit better beam nt of character states and/or evaluation of the validity of this eae. Both growth forms occur I. in other Centrosper- mae, hence I designate no polarity. 1 presume perennials persisting via the caudex or main stem to be primitive with respect to annuals. I scored Naiocrene as perennial and Моппа as annual. Perennials occur among the latter, but they apparently persist via stolons rather than a cau- ex or main crown (Nilsson, 19664). Carolin (1987) and ene (1992) employed alter- native character state d У in their analyses. Carolin (1987) categorized Portulacaceae variously as phanero- phytes, passive chamaephytes, active chamaephytes (in- cluding protohemicryptophytes and rosette hemicrypto- phytes) and t three categories intergrade, i.e., erophytes. For Portulacaceae, the first some taxa scored by Carolin as passive chamaephytes (e.g., Talinum sect. Ta- linum) perennate via distal buds on erect branches rather than from rosette axils (e.g., in Нин spp.). former condition seems more akin to d Jaana и: habit, and the latter to rosette Пе йм hytes. Cal- andrinia sect. Acaules and most Lewisia spp. must be regarded as rosette Каа rather than passive chamaephytes as Carolin scored them. The active cham- aephyte category is ignored here because so characterized Portulacaceae seem to be derived members of otherwise rosette-forming perennial taxa, e.g., Claytonia species. Gerbaulet (1992) recognized ihres size classes of cham- aephytes and four types of habit (including rosettiform). Her data indicate that my scoring of Anacamsperos sect. Tuberosae as rosettiform is incorrect. 1. Stems herbaceous or woody n at the base, 0; stems becoming woody throughou 2. Plants not rosettiform, 0; des rosettiform, 1. 3. Plants perennial, 0; plants annual, 1 Roots (CHAR. 4) Carolin’s (1987) characters and scoring are adopted here. Tuberous roots are uncommon outside of the por- tulacaceous ap and I presume them to be derived. da & Ash n (1989) Msi tuberous roots to be a derived trait Ps Rumicastrum, but tuberous roots oc in the perennial species (Pate & Dixon, 1982), so I ied Rumicastrum as tuberous. 4. Roots not tuber-forming, 0; roots tuber-forming, 1. STEMS (CHARS. 5, 7) Carolin's (1987) interpretation and scoring are adopted here. Unless axillary hairs and scales/bristles are consid- Gerbaulet (1992) о several scale types within Anacampseros, and she scored scale d in Gra- hamia, Talinaria, and Ta alinopsis as unknown. Her re- || sults also indicate that, contrary to my score, axillary hairs are primitively absent in Anacampseros sect. Avo- ia ~ ~ ~ 5. Stems without axillary hairs, 0; stems with axillary hairs, l. 6. Stems without axillary scales/bristles, 0; stems with axillary scales/ bristles, LEAVES (CHARS. 7-21) The leaf ls and data are from Hershkovitz (1990a, 1991b, , 1993, unpublished), including re- iss m н я от nd r (1895), Carolin (1987), Kel- ley (1973), Metcalfe & Chalk (1950), Nilsson (1966a, 1970, 1971a, b), and references cited in these. I exclude several characters that I have reported on elsewhere be- cause of excessive infra-OTU variation, lack of variation at the level of the present OTUs, and/or inadequate data. These include petiole ip лыны, primary vein anatomy, vein density, presence of three-dimensional venation, the ard-cell ycle of bla like cells (Hershkovitz, 1993), | nie and stomatal in- dex. number of epidermal cl in contact with the pair, the presence of an o . Leaf position. 1 tentatively presume alternate leaves to be primitive, but opposite leaves occur in man oaceae. pole acter for Calyptrotheca. Opposite bae are rare in Portulac QA ее Molluginaceae, and Aizo missco this char a but characterize the puta- tively most primitive species (Geesink, 1969, 1987). I scored the genus as “? 7. Leaves alternate, 0; leaves opposite, 1. Leaf base morphology. Nos and nonjointed leaf bases occur in othe Clasping, nonclasping, er Centro- spermae, hence I designate no polarity for the leaf base characters. I previously interpreted the leaf bases of Ки- micastrum as clasping (Hershkovitz, 19902), but accord- ing to West (1986), they are not. My int based on herbarium, rather than live specimens, so I here erpretation was 356 Annals of the Missouri Botanical Garden FIGURES 2-4. Inflorescence morphology in Calandrinia. ко m of Calandrinia (Calandrinia) ciliata [U.S.A., California: Hershkovitz s.n. (spirit collection)]. Note the lack of dense coiling and the leaf (L) and bract (B) associated with each flower. — 3. Inflorescence of a (Calandrinia) axilliflora [Chile, Region O'Higgins: Volume 80, Number 2 1993 Hershkovitz Calandrinia and Montiopsis 357 adopt West's opinion. It is possible that Rumicastrum is intermediate for this character. 8. Leaf bases jointed, 0; leaf bases not jointed, 1. 9. Leaf bases clasping, 0; leaf bases not clasping, 1 af mucilage cells. Copious mucilage produc- tion ds pi does large numbers of foliar mucilage cells does not appear to be common among other Centrosper- mae; hence I presume it to be derived. Copious mucilage production characterizes at least some species of Cis- tanthe and Rumicastrum, but I have not adequately surveyed either genus; hence I scored these as unknown. Mucilage cells occur in Calandrinia sect. Calandrinia and species of Lewisia (Becker, 1895), but the mucilage content of the leaves appears to be relatively slight (pers. obs.). 10. Leaf mucilage cells none or few and not conspicuous, 0; leaf mucilage cells conspicuous and relatively nu- merous, 1. . Leaf vein anatomy. 1 have not found e Hershkovitz, 1991b) to occur among o e I presume them to be ded a = B. AMENS Ré ribbonlike, вени Тћеге тау ђе по of some taxa are almost alwa taxa they are only weakly man distinct boundary Lae the d bonlike veins may prove to be at least infrequently present r states, and rib- in taxa scored as absent. The scoring adopted here for variable taxa biases against homoplasy. 11. Ribbonlike veins absent, ent, 1 0; ribbonlike veins pres- Leaf venation pattern. Festooned brochidro- mous venation occurs among other Centrospermae, al- though ena brochidodromous is more common. With- out more confide o which Centrospermae are „бе to the portulacaceous alliance, I leave the polar ned brochidodromy negra with Pes nceast certain. Festo llo among Portu ados and, Us scored as present, is absen or narrowest leaves. The scoring aed here for variable taxa biases against homoplasy ong t in the bs and/ 12. Leaf venation not festooned brochidodromous, 0; leaf venation festooned brochidodromous, 1 F. Leaf fimbrial vein. Fimbrial veins occur in other Centrospermae but only in otherwise apparently derived taxa (e.g., among Caryophyllaceae subfamilies Alsinoide- ae and Caryophylloideae). Thus, I presume fimbrial in Po veins m pendicular to the primary vein at the leaf apex, and those that become more parallel (Hershkovitz, 1990a, 1993). The transformation series adopted here permits fimbrial Calandrini the species 15 и for the trait, and I am not certain nis as present for fimbrial veins, although that my specimens are correctly determined (cf. Hersh- kovitz, 1993). 13. Fimbrial vein absent, 0; fimbrial vein present, 1. 14. Fimbrial vein becoming perpendicular to the primary at the apex, 0; fimbrial vein parallel to the primary vein at the apex, 1; fimbrial vein absent, X. vein С. Stomata. Parallelo- and/or н stomata раш Hershkovitz, 1 990а, 1991) do not appear o be mmon outside of the а аШапсе; 5 hence Lo am inclined to presume that this state is derived. most common stomatal type outside of the ~ anomocytic (Hershkovitz, unpublished; Metcalfe & Chal 1950), is rare and probably derived within the d Hence I leave the stomatal type polarity undetermined. Brachytetra- and staurocytic stomata are presumably de- rived with respect to brachyparacytic. I presume that laterocytic stomata, which are apparently restricted to Lewisia spp., are derived from brachyparacytic (Hersh- kovitz, 1992). Likewise, I presume that anomocytic sto- куне Nilsson, 1966a, 1970, 1971a, b; Nyanyano, 1986a, c) are derived from brachyparacytic, which characterize other species of these genera. Brachytetra-/staurocytic sis sect. Dianthoideae, but I scored this taxon for their absence. 8 stomata occur in Montiops 15. Stomata predominantly ade ones sim- ilar types (see Hershkovitz, 1990a, lb), 0; sto- mata diii d је pep non- brachyparacytic, 1. . Stomata шы тј brachyparacytic, nonbrachy- paracytic, or parallelocytic, 0; stomata commonly brachytetracytic and/or staurocytic, 1. — с Н. Foliar trichomes. a limited distribution among Portulacaceae. Trichomes Several trichome types have occur commonly among Centrospermae; hence no polar- ities are presumed within Portulacaceae. I have adopted treats unicellular trichomes ndular, multi- 19 a transformation series that t as an intermediate in deriving ribbed, glan cellular and/or multiseriate trichomes (see Kelley, 1973). Unicellular foliar trichomes occur in several eastern Amer ican/African of these as lackin chomes as presen bs u tionably derived therein. Gerbaulet’s (1992) results in- dicate that I should have scored Табпата and possibly Ceraria as possessing trichomes, and Anacampseros sect. <— Сау s.n. (P)]. Note the lack of dense coiling and the two leaves (L1 and L2) А with each flower. — 4. Terminal ы of highly ramified inflorescence of Calandrinia (Monocosmia) mon e flowers are solitary and axillary (1), some are associated with bracts (2), and some are uec ud Il. 7 (US)]. Som id (3). ane by Leslie Randa dra [Argentina, Rio Negro: Buchtien 358 Annals of the Missouri Botanical Garden IGURE 5. Cleared dev eloping inflorescence of Clay- tonia sibrica L. [U.S.A., California: near upa, ИЕ UC)] showing dense coiling. Scale bar = ca. 3 mn Tuberosae should have been scored as lacking them. Tri- chomes do not occur in Montia (contra Carolin, 1987, appendix 2), and if unicellular trichomes in Calandrinia sect. Acaules spp. are indeed glandular, they are probably not homologous to multicellular glandular trichomes of Montiopsis (cf. Carolin, 1987). I have not been able to sei the presence of trichomes in Lenzia (cf. Carolin 987). l Inribbed unicellular trichomes have been reported ined. (Kelley, ve been unable to verify the d of з а id ави Е. Muell., 1973), bu the idi apse imen (Hershkovitz, 1993). In any case, strum as absent. Epidermal papillae oc- cur in several other Portulacaceae, гоні Montia and Cistanthe (Hershkovitz, 1990a, Ilb). Here, I distin- uish between papillae, which are ena swollen epider- mal cells, and trichomes, in which the center of the emer- gent cell wall has a distinct U- or V-shaped A odia e. Gerbaulet (1992) u I here consider trichomes. Finally, ©“ used the term “papillae I have Pon X зеі trichome character so that the interderivation of ribbed, unribbed, and no trichomes re- scoring for the ri quires a single step. Foliar trichomes absent, 0; foliar trichomes pres- ent, 18. Foliar ЕЕ ари unicellular, 0; foliar trichomes mul- ticellular, 1 19. Foliar trichomes uniseriate, 0; foliar trichomes mul- tiseriate, |. 20. Foliar ай nonglandular, О; foliar trichomes glandular 21. js tric don nonribbed, 0; foliar trichomes ribbed, ; foliar trichomes absent, X. INFLORESCENCE (CHARS, 22-26) A. Inflorescence type. I tentatively concur with Ca- rolin (1987) in ро that monochasia and solitary flowers are derived in Portulacaceae. That metaxytriads are primitive cannot be established a priori, however n Carolin, 87): this type was unequivocally derived i Carolin’s cladogram. Carolin (1987) interpreted the inflorescence in C. sect. Calandrinia as monochasial, but I regard the “inflores- cence" to be a vegetative sympodium with solitary, ax- (1989) def- inition of an inflorescence, my inte rpretation is proble matic illary flowers. In the context of Weberling's but, nevertheless, emphasizes the evident distinction in infloresc ence morphology betwe еп section Calandrinia and Portulacaceae with unquestionably monochasial in- florescences. In Cistanthe and Montieae, for example, including Claytonia sibirica L. (Fig. 5), the developing inflorescence is strongly coiled, and the inflorescence bracts are much smaller than foliage leaves or, in some species, abse C (Ruiz 2), the deve vd inflorescence is nt. In ndrinia & Pavon) DC. (Fig. cincinnoid but leafy throughout. Each (Calandrinia) ciliata er is associated with one leaf and one small bract, and de cauline leaves are gradually reduced distally but otherwise identical to the basal leaves. Axillary buds occur in both the leaf and oract axils (pers. obs.). In Calandrinia axilliflora Bar- néoud (Fig 3), flowers arise from the axils of otherwise normal foliage leaves that alternate with leaves having vegetative axillary buds. In some specimens, the inflores- cence has very short internodes and thus resembles a whole rosettiform plant of Calandrinia sect. Acaules. The character scoring I adopt for section Calandrinia is obviously quas -laden; hence the results must be judged accordingly. Calandrinia (Monocosmia) monan- dra (Ruiz & Pavón) DC. (Fig. 4) has a unique inflorescence structure that I cannot classify. It is highly branched, the main stem is leafy toward the base, and both single flowers and multiflowered branches may arise from leafy, brac- teate, or naked nodes. I have also observed cauliflory in this species загћег work (Hershkovitz, 1990a), I ha so reexamined the inflorescence of Montiopsis and bnc concluded that it conform . Since my E s to either a botryoid or a bostryx (cf. Weberling, 1989)— either way, the in- florescence is similar in all members of the genus and My new interpretation is based only on studies of herbarium distinct from other Portulacaceae (cf. Carolin, 1987). — material and requires developme ental analysis to verify. I tentatively scored Lewis ia for having dichasia, but monoc Ћазла and solitary flowers also occur in this genus 1992; Mathew, 1989). that dichasia are secondary in Lewisia be they occur in is possible species with a presumably derived chromosome base num- Volume 80, Number 2 1993 Hershkov shkovitz 359 Calandrinia and Montiopsis ber (Hershkovitz, 1992). Solitary flowers also occur in Anacampseros sect. Avonia, Talinum sects. Talinum and Phemeranthus, and Rumicastrum (von Poellnitz, 1933, 1934a, b), but I presume that this is condition. In any not the primitive case, the transformation series I adopt ere requires only one step between dichasia, cincinni, botryoids/ bostryces, im solitary flowers, while metaxy- triads are scored as one step from dichasia and two steps from the other ai assuming that dichasia would con- stitute an intermediate state. 22. Flowers in metaxytriads, 0; flowers not in metaxy- triads, Flowers nonleafy cin 23. о in d or dichasia, 0; flowers in nni, 1; flowers in мнн чер чып, 2; flowers hae or iiie: in cincinni with a least one normal foliage leaf per 24. Flowers not solitary, 0; flowers a or — in cincinni with at least one normal foliage leaf per flo B. Inflorescence bracts. Кот inflorescence bract po- larities, I concur with Carolin (1987) that two is bracts ен ог subequal brac d unequal. The . I have bes pes the Cistanthe sections for d bracts, but the degree of q other taxa (e.g., d be scrutinized quantitatively. While are primitive, and that spect to differential is also problemat s are prim- itive with re act size intergradation with subequal bracts in Montiopsis) shou this manuscript was in review, I reevaluated bract mor- кка with a pair of foliaceous bracts pedicel. Calandrinia affinis has unjointed peduncles, usu- ally with a pair of small, scalelike bracts near the base. Rarely, one of these is foliaceous and more proximal to the flower. (Peralta (1988, fig. 2B) illustrated three bracts in C. affinis, two basal and scalelike and one foliaceous, but I have not observed this condition.) Based on prelim- inary observations, I had scored C. affinis as having one bract or none. Moreover, I now question my interpretation of the bracts in section Calandrinia. 1 had scored these taxa for one bract or none, but it is possible that the leaf pair in C. axilliflora or the leaf/bract pair in other species are homologous with the foliaceous bract pair in C. acaulis and C. carolinii. I leave this problem unresolved pending additional analysis. 25. Inflorescence bracts two per flowering node, 0; in- florescence bracts one or none, 1 N an . Bracts of a pair + equal, 0; bracts of a pair markedly unequal, 1; bracts one or none, X PERIANTH (CHARS. 27-29) The apparently biseriate perianth of Portulacaceae has been variously interpreted as homologous to a bim involucre plus tepals (Cronquist, 1981, 1988; Eckardt, 1976; Gerbaulet, 1992; Stebbins, 1974) or as a tru biseriate perianth jene 1987; Geesink, 1960, Milby, 1980; Sha , 1954; Soetiarto & Ball, vibe For practical pues, I adopt the latter interpretation. ера! morphology. Тһе degree of sepal persis- tence dn fruit жү varies within the taxa cir- cumscribed here (e.g., in Rumicastrum spp. а 4 Ashton, 1989; in Talinum spp., Ma ae 1937), but early deciduous sepals are probably primitive in some taxa. Like Carolin (1987), I scored Talinum for deciduous sepals, but those of 7. paraguayensis Speg. seem per- That persistent sepals are primitive overall is assumed, but deciduous sepals occur among other Centrospermae fectly persistent (pers. obs.). Glandular-toothed bracts and sepals occur in Lewisia and Schreiteria, and I presume them to be derived. Glan- dular and dentate sepals that problematic. The glands are associated with trichomes, hence I do not consider them ho occur in Montiopsis are mologous to glandular teeth. The гаса оор in this genus appear to be as- sociated with the base and flanks of the sepals and bracts, whereas those of Lewisia and Schreiteria are concen- trated toward the apex. Thus, I scored Montiopsis as absent for glandular-dentate bracts and sepals. Some spe- cies of Montiopsis sect. Montiopsis do have lobed (den- tate?) sepals, however (Ford, 1992; Peralta, 1988, figs. 61, 7G). I interpret the sepals of some Cistanthe taxa, especially those of section Calyptridium, as erose rather than dentate. I suspect that dentate bracts in most Lewisia spp. ar ived, but I scored the genus as present to test the remote > blas that Lewisia and Schreiteria are somehow related. 27. Sepals persistent, 0; sepals deciduous, 28. Sepals and bracts not glandular-dentate, 0; sepals and bracts glandular-dentate, 1 B. Petal numbers. because they vary within and intergrade among taxa Some taxa, however, seem to have consistently higher petal numbers than others. Carolin (1987) presumed that higher petal numbers were primitive in Portulacaceae (as did Syeda & Aston, 1989, and Syeda & Carolin, 1990, for Rumicastrum), but I Petal numbers are problematic scored fewer petals as primitive ab a petal (or bind E are, among Cen- t common outside of tulacaceous dinee. Rie ish they are in ме derived mong Aizoaceae; Bittrich & Struck, 1989). Irr spective of Carolin's (1987) polarity determination for this trait, his scores for this character are sometimes incorrect; e.g., nacampseros and Talinaria (Pax & Hoffmann, lone McNeill, 1977; von Poellnitz, 1933). As with the preceding character, the score for the variable taxon Lewisia is experimental. 29. Petals typically five to six or fewer, 0; petals typically e than five, 1 STAMENS Carolin (1987) scored Grahamia and Portulaca for having connate stamens adnate to the petals. The con- dition in these species is possibly extreme, but similarly fused stamens are probably the general condition among Portulacaceae (see, e.g., Perala. 1988, for Calandrinia sensu lato and Mathew, 1989, for Lewisia). I prefer not 360 Annals of the Missouri Botanical Garden FiGURE Capsule morphology in Montia and Calandrinia. —6. De [U.S.A., California: Thomas 3543 (US)]. A. Wh hisced capsule of Montia parvifolia ole fruit showing capsule-valve inrolling, the narrow teardrop-shaped grooves near the margins of the capsule valve base, and the na ribs. A sepal has been folded down to show Volume 80, Number 2 1993 Hershko vitz 361 Calandrinia and Montiopsis to recognize stamen fusion as a character in this analysis. I agree with Carolin (1987) that stamen number is too highly variable lio OTUs (sometimes within i to be useful at the present level of ndividuals) analysis, especially because the нена. condition is equivocal. POLLEN (CHARS. 30, 31) The pollen character transformation series proposed by Carolin (1987) whether narrow- and broad-pantocolpate pollen can al- ways be distinguished, Carolin's transformation series re- is problematic because, irrespective of Possibly interme- diate states occur: several taxa reported by Carolin (1987) to be pantocolpate were reported by Nyanyano (1985 1986a) to be pantoporate. Gerbaulet (1992) reported, however, that contrary to mine and Carolin's scores, Ап quires two steps between these states. acampseros sect campseros has pantoporate pollen. The transformation series adopted here presumes only that three apertures are primitive with respect to several, and that colpi are primitive with respect to pores. 30. Pollen tricolpate, 0; pollen panaperturate, 1. 31. Pollen colpate, 0; pollen porate, 1. STIGMAS (CHARS. 32, 33) The comments for petal number are applicable here: stigma numbers intergrade, the polarity is uncertain, and some ps Carolin's (1987) scores are not in agreement with the literature, e.g., for 4nacampseros (cf. Pax & Hoff- 1934; von Poellnitz, 1933). Two зе di are unusual among Portulacaceae, howev sumed to be derived. The s possibility that the genus is eed to Portulaca or Gra- mia. mann, ver, and are pre- ore for Lewisia tests the 32 Stigmas two or three, gen four, 0; stigmas typ- ically four or more, 33. Stigmas three or more, 0; stigmas two, 1. FRUIT (CHARS. 34-42) fruit characters, polarities, and scores proposed acters are left uncertain. These characters should be better scrutinized because I believe they may intergrade more than Carolin’s scoring would indicate, especially in Talinum sects. Talinum and Phemeranthus. The fruit of Talinum (Talinum) rivae Chiov. is acropetally dehiscent (Carolin, in press; von Poellnitz, 1934b), but I suspect that this Somalian species actually belongs to Calyptro- theca. Acropetal dehiscence has been described for Ta- (Jacq.) Gaertn. Doo 1986). Ac cording to Carolin 3 press), the exocarp and endocarp in Talinum sect. inum sometimes E (cf. Carolin, 1987). Carolin (1987) scored Calandrinia sect. Mono- cosmia as indehiscent, but elsewhere (Carolin, in press) reported occasional valvate dehiscence. Thus, I scored this section as dehiscent. Contrary to mine and Carolin's scores, Gerbaulet (1992) scored Grahamia for having a pericarp not differentiated into two layers. I have disagreed (Hershkovitz, 1990a) with Carolin's (1987) interpretation of the pericarp of Calandrinia sect. Calandrinia: the distinctive *grooves and ribs" that char- acterize the capsule base of Montieae (Fig. 6) do not occur in the capsule base of any members of C. section Cal- andrinia. The capsule in rs latter may, as Carolin (1987) ciliata; Fig. 7), ut the capsule in C. axilliflora Barnéoud does not mark- edly inroll (Fig. 8). I scored Calandrinia as present for inum paniculatum noted, inroll after dehiscence (e.g., in capsule valve inrolling and absent for capsule valve grooves and ribs. I scored indehiscent taxa and Monocosmia as unknown (**X") for character 41. 34. Fruit without acropetal, valvate dehiscence, 0; fruit with acropetal, valvate dehiscence, 1. 35. Fruit dehiscent, 0; fruit indehiscent, 1. 36. Fruit epidermal cells with sinuous radial walls, 0; fruit epidermal cells with straight walls, 37. Fruit Ln cells elongate, 0; fruit epidermal cells isodiametric 38. Pericarp not onde differentiated into two layers, 0; pericarp strongly differentiated into two layers, 1. 39. Exocarp and endocar and endocarp separating, 1. 40. Exocarp not deciduous, 0: exocarp deciduous, 1. 41. Capsule valves not inrolling after dehiscence, 0; cap- sule be inrolling after dehiscence, 1; fruit hiscent, X arp not separating, 0; exocarp inde- 42. Capsule valves not grooved and ribbed at the base, 0; Capsule valves grooved and ribbed at the base, 1. SEEDS (CHARS. 43-44) I have retained Carolin’s (1987) characters, polarities, and scores for seed traits the homology between the “ of Talinum sect. Phem eranthus and that fy eres has not been ub lished (see also Gerbaulet, 1992). I also note that many Portulacaceae have (usually less conspicuous) micropylar appendages (Corner, 1976; Hershkovitz, 1992), but I have not yet analyzed the variation in this feature. As Carolin pointed out, however, aril” the capsule bas is not pre Leslie Randa . B. Magnified view of capsule base showing grooves and ribs. d ciliata [U.S.A., California: Wolf 3542 (US)] showing capsule-valve inrolling. al has been folded down to show di capsule ba Г (С аа аш) axillifiora [Chile, Region Aconcagua: Steudel s.n. ribbed capsule base is evident. Following dehiscence, the E valves may ret irregularly. A sepal has been folded down to show the capsule base ll. — 1. Dehisced capsule of Calandrinia A grooved, ribbed base e. — 8. Partially dehisced capsule of Calandrinia pl. Neither capsule- -valve inrolling nor a grooved, ain their shape or twist and fold . С = groove, R = rib, 5 = sepal. Drawings by 362 Annals of the Missouri Botanical Garden /--- Talinaria *--- Апасатрѕегоѕ $ Anacampseros *--- Anacampseros § Avonia | | | \--- Anacampseros 6 Tuberosae 5 /----3 | +-------- Grahamia | | /---36 \-------- Talinopsis | | | \ Portulaca /---37 | + Talinum § Talinum | | | + Schreiteria | + Talinum 8 Phemeranthus | /---40 foem Calyptrotheca | | | | | 39 /---- Ceraria | | \----38 | | \---- Portulacaria | | | N Rumicastrum | | / Calandrinia compacta /---46 | | | | /[---------- Calandrinia acaulis | | | | | | /---43 + Calandrinia caespitosa | | | | | | | | | + Calandrinia carolinii | | | \---42 | | | + Calandrinia affinis | | | | | \------- 45 | /---- Calandrinia § Calandrinia | | \----41 | | X---- Calandrinia $ Monocosmia | | | | /---- Montiopsis $ Dianthoideae | \ 44 | \---- Montiopsis $ Montiopsis | | |-------- Cistanthe $ Amarantoideae ---52 | | /---48 /---- Cistanthe $ Philippiamra | | \--47 | /---49 \---- Cistanthe 8 Calyptridium | | | + 50 N Cistanthe 8 Cistanthe | | | \ Cistanthe § Strophiolum | | /---- Claytonia | | + 51---- Montia § Naiocrene | | | \---- Montia exc. $ Naiocrene | (9) + Lewisia | \ Lenzia FIGURE 9. Consensus tree of Portulacaceae as in Figure 1 but with nodes labeled for referencing Appendix 2. Volume 80, Number 2 Hershkovitz 363 1993 Calandrinia and Montiopsis 43. Seeds without enveloping aril, 0; seeds with aril, 1. tonia as unknown. No ordering for base numbers of 10, 44. Embryo encircling the perisperm, 0; embryo not 11, or 12 is supposed— these can all be derived from 9 encircling the perisperm, 1. via one step. My character definitions are faulty in treating nonequivalent states (i.e., 10, 11, and 12) as bentikupds- CHROMOSOME NUMBER (CHARS. 45, 46) I resort to this treatment because the primitive base num- ber in some well-surveyed taxa, e.g., Lewisia and Montia, is uncertain but clearly not 9. I scored Cistanthe sect. Strophiolum as unknown but not 9, but my conclusion hat its base number is 11 or 12 is speculative (Hersh- kovitz, 1992). Gerbaulet (1992) reported x = 9 in Tali- naria, which I scored as unknown. My consensus tree (Fig. 1), however, predicted x — 91 in this genus. This character is added here based on data compiled and/or reported by Bernardello (1989), Hershkovitz (1992), Nilsson (1966b), Nyanyano (1986a), and For (1992; unpublished). The data are incomplete for most of the larger taxa. ae base ки 9 is tentatively presumed to be iti s the most widespread among other невя СЕ 1976), base numbers of 8 also occur among otherwise primitive 45. Chromosome base number 9, 0; chromosome base Aizoaceae (Bittrich & Struck, ion Тће base number um not 9, 1 8 occurs in derived species of Portulaca but is possibly 46. Chromosome base number 10, 0; chromosome base primitive in Claytonia (see Gerbaulet, 1992, for discus- number 11, 1; chromosome base number 12, 2; sion and references). Because of variability, I scored Clay- chromosome base number 9, APPENDIX 2. Character state changes in the consensus tree. Listed below, by character number (cf. Appendix |) and tree node (Fig. 9; cf. Fig. 1) are the character state changes of the consensus cladogram for Portulacaceae as output using PAUP's "list changes" option and "accelerated transition" optimization in the "describe trees" command. The number of changes listed is 114; hence the list includes excess homoplasious changes with respect to any 107-step resolution of the consensus tree. Unambiguous changes are shown in Figure 1. "CI" = consistency index. Character CI Steps Changes 1 0.500 | поде 35 0 => 1 bien 1 node 40 0 => 1 node 3 2 0.333 1 node 40 1 = 0 node | і 1 node 34 0 = 1 Anacampseros $ Tuberosae 1 node 40 1 => 0 node 39 3 0.333 1 node 42 0 = 1 node 41 | поде 49 : => 1 node 48 | node 51 0 1 d excl. § Naiocrene 4 0.250 l node 46 D => | node 40 | node 34 1 => 0 "ncmpe $ Апас е" 1 node 34 1 = 0 Anacampseros $ Ave 1 node 39 1 = 0 С орттен 5 1.000 1 node 37 0 = 1 node 6 0.500 l node 37 0 => 1 node З o | node 34 1 => 0 Anacampseros $ Tuberosae 7 0.500 1 поде 35 0 => 1 m | node 39 0 => 1 node 38 8 1.000 1 node 46 1 — 0 es 40 9 1.000 | node 46 1 => 0 node 40 10 1.000 1 node 46 0 => 1 node 40 11 0.200 | node 46 0 => 1 node 40 1 node 35 1 => 0 node 34 1 node 36 1 => 0 Portulaca 1 node 39 1 — 0 node 38 | node 50 0 => node 49 12 0.250 1 node 36 О= 1 Portulaca 1 node 37 0 = 1 Talinum $ Talinum || node 38 0 => 1 Ceraria 1 node 52 0 = 1 node 50 364 Annals of the Missouri Botanical Garden APPENDIX 2. Continued. Character CI Steps Changes 13 0.333 1 node 43 О= 1 node 42 1 node 42 ] — 0 Calandrinia acaulis 1 node 52 0 => 1 node 51 14 1.000 1 node 43 1 => 0 node 42 15 0.500 1 node 46 0 => 1 node 40 1 node 37 1 = 0 Schreiteria 16 0.333 1 оде 37 0 = 1 Schreiteria 1 node 43 0 => | node 42 1 node 51 0 = 1 Montia $ Naiocrene 17 0.200 1 node 34 0 => 1 Anacampseros $ Anacampseros 1 node 34 0 => 1 4nacampseros $ Tuberosae | node 37 0 => 1 Talinum $ Talinum | node 46 0 => | node 45 l node 42 1 => 0 Calandrinia acaulis 18 1.000 | node 45 0 => | node 4 19 1.000 1 node 45 0 — | node 44 20 1.000 1 node 45 0 = ] node 44 21 1.000 | node 45 0 => | node 43 22 0.500 1 node 36 1 => 0 Portulaca | n 40 ] => 0 node 39 23 0.400 1 node 35 0 => ] node 3 | node 40 0 => 1 Rumicastrum 1 поде 45 0 => 2 под l node 52 0 => 1 node 50 1 node 52 0 = ] node 5 24 0.333 І node 35 0 = 1 Grahamia 1 node 45 0 => | node 43 1 node 52 0 => | Lenzia 25 0.250 | node 45 0 => ] node 43 1 node 42 1 => 0 Calandrinia acaulis 1 node 42 1 = 0 Calandrinia carolinti l node 52 0 = ] nod 20 1.000 l node 51 0 = | node 50 27 0.250 1 node 35 0 = | node 34 1 node 36 0 => 1 Portulaca l node 37 0 = 1 Talinum $ Talinum 1 node 40 0 — 1 Talinum $ Phemeranthus 28 0.500 1 node 37 0 = 1 Se hreite ria 1 node 52 0 = 1 Lewi 29 0.200 1 node 42 О = 1 Calandrinia acaulis l node 42 0 => 1 Calandrinia carolinit 1 node 42 0 = 1 Calandrinia affinis 1 node 50 0 — 1 Cistanthe 8 Strophiolum | node 52 0 = 1 Lewisia 30 0.200 | node 52 0 = | node 46 | node 39 1 => 0 node 38 | node 44 1 = 0 Montiopsis $ Dianthoideae 1 node 51 0=> 1 Montia $ Naiocrene | node 51 0=> 1 Montia excl. $ Naiocrene 3] 0.250 | node 37 0 = 1 Schreiteria 1 node 39 0 => 1 EE 1 node 45 0 = 1 node 4 1 node 44 0 => 1 ria om § Montiopsis 32 0.333 1 node 35 0 = 1 Grahamia 1 ode 36 0 => 1 Portulaca 1 node 52 0 => | Lewisia Volume 80, Number 2 Hershkovitz Calandrinia and Montiopsis 365 APPENDIX 2. Continued. Character CI Steps Changes 33 0.500 1 node 41 0 => 1 Calandrinia $ Monocosmia | node 48 0 = 1 node 47 34 0.500 1 node 39 0 => 1 Calyptrotheca 1 node 52 0 => | Lewisia 25 0.500 1 node 39 0 => 1 node 38 1 node 47 0 — 1 Cistanthe 8 Philippiamra 36 0.333 1 node 40 0 => 1 node 37 1 node 34 1 => 0 Anac ae $ Tuberosae 1 node 39 0 => 1 Calyptrotheca 37 0.333 1 node 36 0 => | node 3 | node 34 1 => 0 Guan $ Tuberosae 1 node 37 0 = 1 Talinum $ Talinum 38 1.000 1 node 40 0 => 1 node 37 39 0.500 | поде 35 0 => 1 поде 34 1 node 35 0 => 1 Talinopsis 40 1.000 ] node 35 0 => 1 node 34 41 0.500 l node 41 0 = 1 Calandrinia $ Calandrinia l node 52 0 => | node 42 1.000 | node 52 0 = | node 51 43 0.500 || node 36 0 => ] node 35 1 node 40 0 => 1 Talinum $ Phemeranthus 44 1.000 | node 36 0 => ] node 35 45 0.500 1 гоо! О= | node 52 l node 37 1 = 0 node 36 46 0.400 l node 52 ] — 2 node 46 | node 38 2 => 1 Portulacaria | node 41 2 = 1 Calandrinia $ Monocosmia 1 node 45 2 => () node 4 | node 48 1 => 0 вија 5 4marantoideae LEAF MORPHOLOGY OF CALANDRINIA AND MONTIOPSIS (PORTULACACEAE)! Mark А. Hershkovitz? ABSTRACT Calandrinia sect. Calandrinia, C. sec blade and petiolar morphology, fimbrial veir Acaules have irregularly brochidodromous venation (ali ho е c logenetic analysis of foliar characters of Portulacaceae: (1) supports a close relationship between C. sects. (2) indicates that C. and Monocosmia and apani ies of C. sect. Acaules; to the preceding taxa; (3) p may be nested within Calandrinia; (4) provides weak e that foliar characters may not of foliar morphology, such as that p between leaf form, function, and on TO and some species of C. , predominantly brachytetracytic ЖЕ о stomata surrounded by a cycle of subsidiary-like cells, and а ribbed trichomes ог рарШаг trichomes. Othe vides weak ps | by nonfoliar evidence that Cla more |“ i SH Phylogenetically analyz Acaules share identical leaf sect. Acaules may be paraphyletic with respect tonia and Montia ce linking Montiopsis with Calandrinia; and (5) indicates ed systematic accoun ts sented here, may provide a basis for advancing theories on the relationship geny. In previous papers (Hershkovitz, 1991a, 1992), I demonstrated the utility of foliar morphology in clarifying phylogenetic relationships among mem- bers of Portulacaceae. The present paper presents and evaluates foliar morphological data for Cal- andrinia Kunth and Montiopsis Kuntze e circumscriptions of Calandrinia end Mon- tiopsis accepted here are based on a cladistic re- analysis of Carolin's (1987) data for Portulacaceae that incorporates the leaf morphological evidence presented below (Hershkovitz, 1993). Calandrinia and Montiopsis include ca. 29 of the 100-150 species formerly classified in Calandrinia sensu lato. I had previously advocated retaining. Mon- tiopsis in Calandrinia (Hershkovitz, 1990; see also 1991b, fig. 3), but cladistic analysis does not ЕИ support this relationship. (Hershkovitz, 1993). Calandrinia, as here circumscribed, in- cludes ca. 14 speci ies in three sections: (1) C. sect. Calandrinia (= C. sect. e bd Һе; in- Axillares Reiche; c 5 spp. of western America); (2) C. sect. unb Reiche (ca. 8 spp. of western South America and Mexico); and ) C. sect. Monocosmia (Fenzl) Hershkovitz (1 sp. in Chile and Argentina). Montiopsis is restricted to temperate South America and includes two sub- genera, M. subg. Montiopsis (2 Calandrinia subg Hirsutae; ca. 15 spp.) and M. subg. Dianthoideae (Reiche) D. I. Ford (ca. 3 spp.). cluding С. sect. — Uy MATERIALS AND METHODS Gross leaf morphology, leaf den pattern, and epidermal morpholog y were examined and pho- tographed according to the protocols I described previously (Hershkovitz, 1991a). The morphology This manuscript is revised from California, Davis. sue submitted as ои of a d dissertation at d University of gratefully deben A assistance from Hershkovitz, Walt Kelley, June McCaskill, Joan Моне ca. Ru e late Richard Eyde, Donna Ford, Philip | бе Schmid, John Бый, Grady V, n Doyle, t Webster, rap Yankowski, two anonymous reviewers, and the curators of the owing herbaria: B, CAS, , RSA, UC, US. This research rec eived financial support from a Smithsonian Graduate Student Fellowship, a Smithsonian Pre-doctoral Fellowship, a Smithsonian Short-Term Visitor Award, two Jastro-Shields Scholarships from the University of California, Davis, two travel awards from the Department of Botany, University P. Hershkovitz. of California, Davis, and U.S.A * Laboratory of Molecular Systematics, Museum Support Center, Smithsonian Institution, Washington, D.C. 20560, ANN. MISSOURI Bor. Garb. 80: 366-396. 1993. Volume 80, Number 2 1993 Hershkovitz 367 Leaf Morphology of Calandrinia and Montiopsis TABLE 1. for venation and epidermal features. Listed by genus, section, species, locality, collector, number, and herbar- Species, citations, and specimens examined ium are specimens of Calandrinia and о рге- vared for venation ("у") and epidermal (“e”) stu ( imens of Calandrinia sect. Acaules are identified acc Каи to the key in Table 2. Determinations of species for Montiopsis are fide Ford (1992). Spe and Rumicastrum illustrated for foliar features are also cited. Locality information is limited to the largest political < 'cimens of Montia division within a country. om Kunth in HBK, Nov. Gen. Spec. 6: (= учи и sect. . Deutsch. Bot. Colandrini sect. Calandrinia. ;.ompressae Reiche, Ber 502. 1897.) C. alba (Ruiz & Pavon) DC., Prodr. 3: 359. 1828. ERU. LA LIBERTAD: Lopez 1223 (US), e. C.a ese Barnéoud in Gay, Fl. Chile =. Fisica Politica de Chile, Botanica] 2: 1847 (18467 ). CHILE. METROPOLITANA: pisi 2786 (US), e, v; Philippi 697 (US), v; VALDIVIA: Joseph 2728 (US C. ж 5. "Watson, Proc. Amer. Acad. Arts 11: 24. 1876. U.S.A. CALIFORNIA: Hoover 6697 (UC ), С im (Ruiz & Pavón) DC., Prodr. 3: 359. 1828. COLOMBIA. SANTANDER: Araque & Barkley 18010 (US), v. GUATEMALA. HUEHUETENANGO: Beaman 3102 оча е. PERU. HUANCAVELICA: Tovar 12 (US), v. U.S.A. CALIFORNIA: Wolf 3453 (US), e, v; OREGON: Maguire & Holmgren 26606 (US), v C. compressa Schrad. ex DC., WITHOUT LOCALIT ACONCAGUA: Joseph 4032 3720 (US), v. Calandrinia sect. Acaules Reiche, Prodr. 3: 359. 1828. Y. Gerhard s.n. (B), v. CHILE. 32 (US), v; MAULE: Joseph Ber. Deutsch. Bot. Ges. 15:500. 1897. C. acaulis Kunth in HBK, Nov. 1823. C. acaulis var. acaulis. ARGENTINA. SALTA: lenturi 7017 (US), v. BOLIVIA. LA PAZ: Asplund 6289 (US), v. ECUADOR. PICHINCHA: Heilborn 231 (US) v. MEXICO. MEXICO: Мун 1881 (F), v 4256 (US), v; Balls 7 (US), e. PERU. LIMA: Cerrate 995 (US), v. C. acaulis var. magna Macbride, Candollea 5: 350. 1934. PERU. LIMA: Killip & Smith 21648 (US), Gen. Spec. 6: 78. 5 Pringle M C. affinis Gillies ex Arn. in Cheek, Edinburgh J. Nat. Geogr. Sci. 1: 355. 1831 LOCALITY: Reiche s.n. (B), e; MAULE: Werdermann 606 (F), v; METROPOLITANA: Hastings 497 (US), ; Mahu 10998 (UC), v C. о Gillies ex ree in Cheek, Edinburgh J. Nat. Geogr. Sci. 1: 356. 1831. ARGENTINA. CHUBUT: Beetle 35 1 (US), v; NEUQUEN: Rossow 1084 (NA), HILE. WITHOUT TABLE 1. Continued. e; Schjovskoy s.n. (M), e, v; SANTA CRUZ: Sleumer 653 (US) v; Sleumer 967 ~ v. CHILE. ACONCAGUA: Buchtien s.n. (US Ca pian Hershkovitz & Ford, Elo У 74: 282. 199 TINA. CATAMARCA: Jorgensen 1200 Mud e; JUJUY: Venturi ре ey (US), e, v; SALTA: 7018 (US), e: Venturi 8538 (US), v; TUCUMAN: Venturi 4824 (U У: е, У. C colc соны Barnéoud in Gay, Fl. pes oe a y Politica de Chile, etn 2 (“1846 ") CHILE. ARAUCAN (UC, US), e, v; BIO-BIO: Werdermann 1325 (US), e C. compacta Barnéoud in Gay, Fl. Chile [Hist. Fisica y Politica de Chile, Botanica] 2: 481. 1847 (“1846”). ARGENTINA. SALTA: gus 9239 (US), v. d ANTOFAGASTA: ll'erdermann 1035 (US), : Werdermann 952 (US), v C. nens Phil., Linnaea 28: 690. 1856. AR- GENTINA. NEUQUEN: Arr oe Leuenberger 3438 (B), e, v. CHILE. MAULE: - Philippi 291 (B), e Calandrinia sect. Monocosmia (Fenzl in Endl. & 7 Venturi A: Ey in am ene US на s.n. ( Hershkovitz, Phytologia 70: 223. 1991. C. monandra (Ruiz & Pavón) DC., Prodr. 3: 359. 828. ARGENTINA. RIO NEGRO: Buchtien 37 (US), v. CHILE. ACONCAGUA: Joseph 3719 (US), Rev. Gen. Pl. 302): 14. 1898 Montiopsis subg. Montiopsis (= Calandrinia subg. ен Reiche, Ber. Deutsch. Bot. 17.). ^ т (Hook. & Arn.) D. I. Ford, dn 4:275. 1993. BoLivia. La PAZ: Asplunc e, v. Montiopsis Kuntze, Ges. 15: 502 (US), e M. capitata (Hook. & Arn.) D. I. Ford, Phytologia . ARGENTINA. RIO NEGRO: Fischer e, у; SANTA CRUZ: Sleumer 1166 (US), А ATACAMA: Joseph 5123 (US), v; COQ "un 4505 (US), M. copiapina (Phil.) D. I. Ford, Рајне 7 (4: 275. НИЕ. ATACAMA: Johnston 6069 (US), i ee M. P sil uin & Arn.) D. 1. Ford, Phytologia 3. CHILE. ACONCAGUA: Buchtien 1149 ( пр M. parviflora PW) D. I. Ford, Phytologia 74: 276. CHILE. ATACAMA: Johnston 5882 (US), M. перат (Barnéoud) D. I. Ford, Phytologia 1993. CHILE. COQUIMBO: Looser 2176 " S) v M. 21 (Hook. & Arn.) D. I. Ford, Phytologia би . CHILE. ANTOFAGASTA: ша 359 m ; COQUIMBO: ee 4817 (US), v; Rose н Rose 19254 (US M. с (Ruiz & Pavó on) D. I. Ford, ge 1993. CHILE. BIO-BIO: Joseph 3856 (US 368 Annals of the Missouri Botanical Garden TABLE |. Continued. e, v; MAULE: Grau & Grau 1516 (M), v; Wer- dermann ms (US), v; METROPOLITANA: Kuntze s.n. (US), v. Montiopsis subg. ше чы (Reiche) D. 1. Ford, Phytologia 74: 277. 19 M. ee i ey ex an ) D. 1. Ford, Phytologia . 1993. CHILE. ME TROPOLITANA: Hastings js 0 dm $), v; Hastings 468 (UC, US), e, v; Has- tings 489 (US), v ELA 698 (US), v; o^ Huic- GINS: Aravena 3335 › (UC), v; Bastin 17 (US), v. M. gayana (Barnéoud) D. 1. Ford, Phytologia 74: 277. 1993. ARGENTINA. RIO NEGRO: Buchtien 36 (US), v; SANTA CRUZ: Sleumer 1248 (US), e, v. CHILE. ARAUCANIA: Jl'erdermann 1219 (US), e, y BIO-BIO: Joseph 3496 (US), e, v; MAULE: Joseph 5265 (US), v. M. «йога | ex Arn.) D. I. Ford, Phytologia : 271. 1993. CHILE. MAULE: P dd 56 15 (US), у; соне Philippi s.n. (B), « Montia L., Sp. Pl. 87. 1753; Gen. PL, ч 754. M. parvifolia (Mog. ex DC.) Greene, FI. ку 181. 1891. CANADA. BRITISH COLUMBIA: Calder et al. 14865 (US), e. U.S.A. CALIFORNIA: Condit s.n. (UC), v; Randall s.n. (UC), e, v; Thomas 3542 (US), v. Rumicastrum Ulbr. in Engler v Prantl, Nat. Pflanzen- 2 Aufl., 16c: 519. 19 Rumicastrum sp. WITHOUT DATA: OW alter Kelley cleared leaf collection, no. K-242 fam., is described according to the terminology suggested by Dilcher (1974) for leaf size and shape, Hickey (1979) for venation pattern, Wilkinson (1979) for stomatal configuration, and Kelley (1973) for tri- chomes. The data presented here are supplemented with data from Becker (1895), Carolin (1987), Ford (1992), Kelley (1973), Nyanyano (19804, b), and Peralta (1988). Table 1 lists the species, their full citations, and representative specimens. All specimens examined for epidermal and venation features are listed. Also listed are species, citations, and specimens of other Portulacaceae here illus- trated Since submitting the morphological data pre- sented here as part of a dissertation (Hershkovitz, 1990), I have substantially revised my taxonomic opinions on the species of Calandrinia sects. Cal- andrinia and Acaules. | previously treated Cal- andrinia compressa Schrad. ex DC. as a synonym of C. ciliata (Ruiz & Pavón) DC., but the former, restricted to Chile, is distinguished by its strongly compressed calyx of deltoid sepals. Calandrinia Provisional . to the species of Calan- che pon drinia sect. Acaules Rei — Бы Peduncles jointed; inflorescence bracts 2, foli- aceous, attached above the middle of the pe- duncle. 2a. Stamens 5-10; capsule usually membra- nous at maturity; leaves without trichomes or papillar trichomes (Ecuador, Colombia, exico, Venezuela) or occasionally with рарШаг tric es Eo the leaf margin (Argentina, Boliv eru). 3a. Flowers zr em long .............. . acaulis var. acaulis 3b. Flowers ca. 2 em long TREES acaulis var. шш 2b. Stamens 12-20; pian "usually indura at V ded leaves often with um Or apillar trichome C. carolinii Peduncles not Mcr nord ence bracts none or 2, scalelike and at the peduncle, or 2, — — the base of 'alelike, basal and one foliaceous, attached more арша ally 4a. Inflores Е tached near one s scence bracts 2 Z ^ £e Inflorescence - half as long as leaves, stamens ca. 8......... ¿cole oe 5b. Inflorescence > half as long as leaves dom oblanceolate, flat, more than twice as broad as thick, glabrous or with trichomes or pap- illar trichomes along the ч! margin. .. affinis Ob. Leaves linear- edm ca. as e as thick, with extra-marginal tri- chomes or papillar tric Pop " > graminifolia 4b. Inflorescence bracts non Та. Peduncles longer than leave a. Stamens 6-7, petals ка a WHITE ае анат е 8b. Stamens more than 7, ud ro inge, ог yellow C. cae i 7b. Pedune les shorter than leaves ...... ciliata, apparently largely extra-Chilean, has a more campanulate calyx and more ovate sepals. Both species, however, are highly polymorphic, and sev- eral segregates have been described for each (cf. Reiche, 1898; Rydberg, 1932: 291, spp. 2-10). The taxonomy of Calandrinia sect. Acaules is more complex and less satisfactory. Table 2 pre- sents a revised, but still tentative, key to the spe- cies. Part of the confusion in my earlier work (Hershkovitz, 1990) stems from the “disappear- ance" of Baitaria acaulis Ruiz & Pavón from the floristic literature and/or the assumption that this name was synonymous with Calandrinia acaulis Kunth. Syntypes of Вапата acaulis reportedly Volume 80, Number 2 1993 Hershkov Leaf M of Calandrinia and Montiopsis 369 include some plants of Calandrinia acaulis. The former species is now Calandrinia carolinii Hersh- kovitz & Ford (Hershkovitz & Ford, 1993). Plants of C. carolinii have usually been misidentified as C. acaulis. popsa this error, I had mistakenly referred them C. colchaguensis Barnéoud (Hershkovitz, To 242). Likewise, I referred bona fide material of C. Gillies ex Arn. The last two species are very similar, colchaguensis to C. affinis however, and intergrade in their overlapping range (D. I. Ford, pers. comm.). I also referred to Cal- andrinia compacta Barnéoud diminutive Bolivian plants of C. acaulis. 1 previously recognized, pri- marily on geographical bases, C. megarhiza Hem- sley (Central America) and C. saltensis Hauman (Argentina), as distinct from C. acaulis (South America). I now believe such distinctions are not warranted. Calandrinia megarhiza and C. sal- tensis, which represent the range extremes e C. acaulis, are distinguishable from one another by the presence of trichomes and slightly higher sta- men number of C. saltensis. These traits are vari- able, however, in C. acaulis from Peru and Bolivia. Finally, I previously included Calandrinia fue- giana Gand. in C. Peralta, distinct, although I have no leaf anatomical data caespitosa Gillies ex Arn. (cf. 1988). I now recognize the former as for the species. Identification of specimens of Montiopsis subg. Dianthoideae follows Peralta (1988), except that I refer M. hirtella Philippi to M. gayana (D. I Ford, pers. comm.). Identification of specimens of Montiopsis subg. Montiopsis follows Ford (1 In order to evaluate the systematic significance of the foliar morphological data presented here, | reanalyzed my earlier PAUP analysis of Portula- 1993), generated by variously weighting and excluding caceae (Hershkovitz, comparing results foliar and nonfoliar characters. The data and char- acter analyses are from Hershkovitz (1993, table 3, appendix 1). RESULTS A. LEAF MORPHOLOGY 1. Gross leaf morphology. morphology in Calandrinia and Montiopsis can be discerned from the illustrations of cleared leaves (Figs. 1-30, see below). Leaves of C. 70 mm long and 2-15 mm wide (nanophyllous to microphyllous; Dilcher, 1974); the basal leaves are usually larger than the cauline leaves. Within these sect. Calandrinia are mostly 10- The gross leaf size parameters, leaf shape varies from linear to oblanceolate or, when the lower portion of the leaf is more abruptly constricted into a winged petiole, often rhombic, ovate, or deltoid above the point of constriction and attenuated below (e.g., C. axilli- flora, Fig. 1). Leaves in C. (Monocosmia) monandra (Fig. 5) are ca. 25 mm long and 8 mm wide, generally wide-petiolate, rhombic above the point of con- striction, and grossly indistinguishable from simi- larly sized and shaped leaves in C. sect. Calan- drinia. Leaves in C. sect. Асаш!ез (Figs. 6, 9, 12, 15, 18) are all basal and exhibit a similar morphological range to those in C. sect. Calandrinia, but they are indistinctly to not at all petiolate, perhaps more succulent (although not readily evaluated from her- barium specimens), and/or much longer (up to 150 mm in C. acaulis var. magna). Leaves in Montiopsis subgenera f Figs. 19, 20, 23, 24) and Montiopsis (Figs. 27, 28) are usually 3-20(-40) mm long and usually less than 3 mm wide (leptophyllous to nanophyllous; Dilcher, 1974), usually oblanceolate to linear, and sessile or with а weakly defined petiolar region The leaf base margins in some specimens of M. andi- cola (Fig. 19) and M. gayana (subg. Dianthoi- deae) are irregularly dentate. The dentations in- ~ between the clasping leaf base and lamina. tergrade with marginal trichomes toward the lea apex (see below). Slight degrees of dentation also occur in conjunction with trichomes in M. subg. Montiopsis. Тће leaf bases in all members of Calandrinia and Montiopsis are flattened, somewhat clasping to nearly sheathing, and not obviously jointed at the abscission zone. Figures 1-30 il- lustrate gross leaf venation patterns of Calandrinia 2. Leaf venation pattern. and Montiopsis; for comparative purposes, Figure 31 illustrates venation in Montia parvifolia. Fig- ures 32-37 illustrate fine venation of C. and Acaules. Figures (Cal- (Monocosmia) monan- sects. Calandrinia, Monocosmia, 38-41 illustrate leaf apical venation in C. andrinia) axilliflora, C. dra, C. (Acaules) carolinii, and Montia parvifo- lia. The descriptions presented here are based on studies of over 150 cleared leaves from 65 indi- viduals of ca. 23 species that represent the gross morphological extremes in Calandrinia and Mon- tiopsis as evident from herbarium specimens. My comparisons of the data are based on examination of ca. 1,500 cleared leaves representing nearly all 370 Annals of the Missouri Botanical Garden species or species groups of Portulacaceae and ca. 500 additional cleared leaves from ca. 250 selected species throughout Centrospermae enation in C. sect. Calandrinia (Figs. 1—5, 32, 33, 38) is characterized by a primary vein that is prominent basally but much diminished to- ward the leaf apex (Fig. 3) or distinct but ramified at the extreme apex (Fig. 38); secondary veins and prominent distal ramifications thereof that form indistinct loops ("arches"; cf. Hickey, 1979) near the leaf margin (Figs. 2, 4) or merge with the fimbrial vein (Fig. 1; see also below); and tertiary veins (sometimes constituting the highest distinct vein order, e.g., in Figs. 1, ) that reticulate irregularly and/or give rise to freely ending veinlets Fig. 32) or, in larger leaves, to a fourth order of reticulating veins (Fig. 32 Areolation is imperfect; the areoles are variably sized and irregularly shaped (Figs. 32, 33) but, in contrast to Hickey’s (1979) characterization of imperfect areolation, there are relatively few freely ending veinlets: the ultimate vein-enclosures usu- ally contain no veinlets or one, usually unbranched, veinlet. A statistical study of the number and dis- tribution of freely ending veinlets per ultimate vein enclosure, a potentially useful taxonomic character in Calandrinia, was not undertaken. The marginal venation in leaves of C. sect. Cal- andrinia is characterized by a fimbrial vein (= a contiguous vein following the leaf margin; Figs. 1-4, 38), which is only weakly consolidated in the basal portion of the leaf but exceedingly prominent toward the leaf apex, where it becomes perpendic- ular to and overtops the primary vein (Fig. 37). Leaf venation in C. sect. Monocosmia (Figs. 5, 34, 39) is, in all respects, indistinguishable from that in C. Some species of C. sect. Acaules possess a fim- sect. Calandrinia. brial vem similar to that in C. sect. Calandrinia, affinis 6-8, 35), C. ac an var. magna (Figs. 9-11, 36), and C. caro- linii (Figs. 12-14, 37, 40). Other taxa lack a fimbrial vein, e.g., C. compacta (Figs. 15, 16), C. caespitosa (Figs. 17, 18), and some specimens of (Figs. acaulis, C. 2 C. affinis. А fimbrial vein is weakly evident іп small- leaved specimens of C. acaulis from Bolivia and Peru and the single examined specimen from Ar- gentina (“C. saltensis”). In general, the fimbrial vein is less well consolidated in C. sect. Acaules carolinii; Figs. 12-14, 37, 40) Calandrinia and Monocosmia (especially in C. than in C. but, as in the last sections, it becomes perpendicular sects. to and overtops the primary vein at the leaf apex (Fig. 40). The primary vein in C. sect. Acaules is more or less distinct at the leaf apex (Fig. 10) or ramified at the extreme apex as in C. sects. Cal- andrinia and Monocosmia ~ 7, 13, 16, 175 especially evident in Fig. Otherwise, venation in C. sect. Acaules appears less well organized than in C. sect. Calandrinia. The secondary veins, especially toward the leaf apex, may merge with the fimbrial vein (Figs. 7, 10); toward the leaf base, the secondary veins form distinct secondary loops within the confines of the fimbrial vein (Figs. 8, 11, 14, 35-37). In leaves lacking a fimbrial vein, the secondary veins form FIGURES 1-5.* the constriction between leaf blade and petiole. — 1 . C. (Calandrinia) breweri (Hoover 6697). - . Whole leaf е у. FIGURES 6-14.* blade.— 7. Leaf apex. — 8. Midlaminal e portio 9. Leaf blade. — 10. Leaf apex.— 11. blade. — 13. Leaf арех. — and 0.2, respectively 3 FIGURES 15-22.* (Acaules) np (Werdermann Ке я (Schjovskoy s.n.). — 17. Leaf blade. Note dentations near the leaf base, two of which are Leaf blade.— 21. Le 1.4, 0.6, ud 0.6, respectively. Cleared leaves of Calandrinia sect. Acaules. aminal/ no portion. 12-14 14. Midlaminal portion. Scale bars are (mm) ca. 1.8, 0.8, 0.8, 3.2, 1.1, 1.1, Cleared leaves of Calandrinia ie Pires and Monocosmia. X, approximate See of 1ole leaf o 2 2: Ты bla де. — E me of C. (Monocosmia) monandra (Buchtie n 37). Seale bu. are ке са. 1.6, 1.9, 1.2, 1.2, апа 1.0, и ахі ога (Joseph 2 idlaminal/ и portion. — 6-8. C. affinis (Werdermann 606). —6. Leaf C. acaulis var. magna (Killip & Smith 21648). — C. carolinii (Venturi 4824). —12. Leaf 1.4, 0.7, Cleared leaves of Calandrinia sect. 4caules and Montiopsis subg. Dianthoideae. 15, 5. Leaf blade.— 16. Lea 18. Leaf apex. — 19 авиона (X). 20-22. af apex. — 22. Midlaminal/marginal portion. Scale bars are (mm) ca. 1.2, 0. f apex. 17, C. (Acaules) урша . Leaf blade of M. (При ува) andicola (Bastin 17). M. andicola (Aravena 33356 2 20. 7, 0.9, 1.8, 0.8, ! FIGURES 1-37. 1. 1, primary vein; 2, secondary vein; etc.; Cleared leaves of Calandrinia, Montiopsis, and Montia 1" i a Vouchers are listed in Table al v L^, secondary loop; F, fimbri Volume 80, Number 2 Hershkovitz Leaf Morphology of Calandrinia and Montiopsis Annals of the Missouri Botanical Garden EAG qu. e rv ket A ur и ERE ДӨ. Volume 80, Number 2 Hershkovitz 1993 Leaf Morphology of Calandrinia and Montiopsis m. A o Annals of the Missouri Botanical Garden ap mmt —— À m À— + 27 A i Sn A ои ; Se e SS = Sse) шыл. " = uU us m = == - ан Аа Е E ~ = —— M = $ Єл? 4 des, y ме Sx hate а P hk У Volume 80, Number 2 1993 Hershkovitz Leaf Morphology of Calandrinia and Montiopsis 375 irregular loops or recurve and ramify freely (Figs. 15-18). The higher-order venation in C. sect. Acaules (Figs. 7, 8, 10, 11, 18, 14, 16, 17, 35- 37) is more open than in C. sects. Calandrinia and Monocosmia: the tertiary veins are more free- ly ramified; the areolation is less developed; more, and more highly branched, freely ending veinlets occur per ultimate vein-enclosure; and the mar- ginal venation (in those taxa lacking a fimbrial vein) consists of irregular loops and freely ending vein- lets. A fourth order of veins is sometimes distinct Acaules (Fig. 11 in larger leaves of C. sect. T sects. Calandrinia and [he fimbrial vein in C. Monocosmia and species of C. sect. Acaules su- perficially resembles that found in Montieae, e.g., in Montia parvifolia (Figs. 31, 41). In Montieae, however, the fimbrial vein merges with the prom- inently dilated primary vein at the leaf apex and remains (or becomes) parallel to it; this morphology characterizes fimbrial veins in certain Caryophyl- laceae (Hershkovitz, unpublished). The perpendic- ular orientation of the fimbrial vein at the leaf apex in Calandrinia is unique. Venation in Montiopsis subgenera Dianthoi- deae (Figs. 19-25) and Montiopsis (Figs. 26-31) resembles that in species of C. sect. lack prominent and distinct at the leaf apex or becomes Acaules that mbrial veins. The primary vein remains ramified at the extreme apex. The secondary veins may be irregularly ramified/reticulate (e.g., M. andicola, Fig. 22), ramified/recurved (e.g., M. gayana, Fig. 24), or form irregular loops (e B- M. gayana, Fig. 23; M. umbellata, Figs. 26, 27; and M. trifida, Fig. 30). highest distinct order and are irregularly ramified The tertiary veins form the to reticulate. Areolation is imperfect; the ultimate vein enclo- often branched, freely ending veinlets; and the marginal venation sures often have three or more, is finely looped to open. In some specimens of M. (Dianthoideae) andicola, a single veinlet pro- trudes into some of the larger basal leaf dentations (Fig little organization in venation pattern: one or two . 19). The smallest leaves in these sections show pairs of weakly distinct, ramified (rarely looped), secondary veins give rise to a few reticulating veins or freely ending veinlets. 1. Cleared le: Randall s.n.). Voucher listed in Table 1. X, approximate position of the constriction between leaf blade and petiole. e ca. 1.7 mm. af of Montia parvifolia 3. Stomatal morphology. | Nyanyano (1986a, b) reported that, among the 22 species of Cal- andrinia sensu lato investigated, 15 had paracytic stomata and 7 had (presumably brachy-)tetracytic stomata. Nyanyano (1986a) did not identify the species of Calandrinia sensu lato examined other than the few he illustrated, so the correlation be- tween his data and the taxonomic circumscriptions employed in the present paper cannot generally be Nyanyano (1986a) did, however, il- lustrate brachytetracytic stomata in C. (Acaules) ascertained. acaulis and also reported tetracytic stomata in C. (Monocosmia) monandra. He did not report vari- ation in stomatal morphology within species, al- though variation in stomatal type is the rule in — FIGURES 23-30.* 24, 25. M. V de (Joseph 3 & Grau 1516). — Cleared ye es of Montiopsis. .— 24. Leaf Leaf nn — 27. Leaf blade. 28-3( half of leaf blade. 5 Leaf арех. — 30. Midlaminal dic Scale bars are (mm) ca. 0.8, 0.5, and 0.6, respectively. 23. Leaf blade of M. (Dianthoideae) blade. — "d Buchtien 36). Leaf ape ‚ 21. Montiopsis) umbellata (Grau M. но trifida wi 3592). — ical , 0.7, 0.8, 1.6, y 1; gayana ( pex. 26 376 Annals of the Missouri Botanical Garden Volume 80, Number 2 1993 Hershkovitz 377 Leaf Morphology of Calandrinia and ti Montiopsis individual leaves of Calandrinia and many other members of Portulacaceae (Hershkovitz, 1991a, 1992, unpublished). Figures 42-54 illustrate stomatal morphology in species of Calandrinia and Montiopsis; for com- parison, Figure 55 illustrates stomata of Montia parvifolia. | examined stomatal morphology in 33 specimens of ca. 25 species of Calandrinia and Montiopsis. For comparison, I examined stomata in selected taxa throughout Portulacaceae. Stomata of C. sect. Calandrinia, including C. alba (Fig. 42), C. breweri (Fig. 43), and C. ax- illiflora (Fig. 44) are predominantly staurocytic (Figs. 42-1, -3, -4; 43-2; 44-1, -2) to brachytetra- cytic (Figs. 42-3, 43-1). The contact cells (i.e., the cells bordering the guard cells) are often surround- ed by an additional cycle of "subsidiary-like" epi- dermal cells morphologically intermediate between subsidiary cells and ordinary epidermal cells (Fig. 42-2, -3). The inner tangential walls of these cells, which abut the subsidiary cells, form smooth arcs, but the outer tangential walls are generally more sinuous, like ordinary epidermal cells. Other sto- matal types occur infrequently in C. sect. Calan- drinia, including brachyparacytic, intermediates between brachyparacytic and brachytetrac ytic (e.g., brachyparacytic with one polar subsidiary cell, Fig. 44-3), hemi-amphibrachyparacytic (Fig. 43-3; cf. Hershkovitz, 1991a), amphibrachyhexacytic, an- omocytic, and other minor morphological variants not readily classified. The number of contact cells in C. sect. Calandrinia is usually four or five. Calandrinia (Monocosmia) monandra (Fig. 45) has stomatal morphology very similar to that of C. sect. Calandrinia. Staurocytic and brachytetracytic stomata also occur commonly to predominantly in most of the examined taxa of C. sect. carolinii (Fig. 46), C. and C. subsidiary-like cells is also evident (Figs. 46-1; 47- 2: 48-1, defined. Brachyparacytic and related stomatal types (е.в. Acaules, including C. colchaguensis (Fig. 47), caespitosa (Fig. 48). An outer cycle of -2) but may be incomplete or poorly eod hemi-amphibrachy- ic) 3 paracy e more common in specimens of C. desk p? aules (especially in C. caespitosa; Fig. 48) than in C. sects. Calandrinia and Mono- cosmia; such stomata predominate only in C. com- pacta (Fig. 49). Irrespective of stomatal type, the number of contact cells in C. sect. Acaules is usually four or five, rarely three or six. The predominance of brachytetracytic and stau- rocytic stomata found in C. sects. Calandrinia, Monocosmia, and Acaules resembles the condition found in Montia parvifolia (Fig. 55). In this spe- cies, however, an outer cycle of cells is less dis- cernable. Brachy- to amphibrachyparacytic stomata E dominate in M. subg. Dianthoideae, e.g., gayana (Fig. 50) and M. andicola (Fig. 51). uan rocytic stomata sometimes occur commonly but M. gayana, Fig. 50). These staurocytic stomata are often surrounded by never predominantly (e.g., an irregular outer cycle of subsidiary-like cells sim- ilar to that found in Calandrinia. The number of contact cells in M. subg. Dianthoideae is most commonly four or five. Brachy- to amphibrachyparacytic stomata also predominate in M. subg. Montiopsis, e.g., in copiapina (Fig. 52), M. gilliesii (Fig. 53), capitata, and M. umbellata. Staurocytic stomata occur infrequently in M. copiapina, and an outer cycle of subsidiary-like cells is sometimes distin- guishable. Hemibrachyparacytic апі anomoc ytic stomata predominate in M. parviflora (Fig. 54). The number of contact cells in M. subg. Montiopsis is usually four but may be as many as six (Fig. 54) Kelley (1973) re- ported unicellular ribbed trichomes in all species 1. Trichome morphology. of C. sect. Calandrinia; no trichomes in C. sect. Acaules; glandular or eglandular, unbranched or branched multiseriate-multicellular trichomes in all species of Montiopsis; and, additionally, barbellate multiseriate-multicellular trichomes in all species of Montiopsis subg. Montiopsis. Carolin (1987) scored C. sects. Calandrinia and Monocosmia for having simple trichomes, the OTUs of Montiopsis for having both multicellular and glandular tri- chomes, and C. sect. 4caules for having glandular trichomes (but neither unicellular nor multicellular trichomes). Peralta's (1988) descriptions of tri- chomes in Mendozan species of Montiopsis accord — FIGURES 32-37. 33€, Calanda breweri (Hoover 6697). — affinis (Werdermann 606). — 36 carolinii (Venturi 4824). All MER bars — ca. Higher order venation in dicus — 32. С. С. (Monocosmia) monandra (Buchtien 37).— 35. C. (Acaule У ас aulis var. magna (Killip & Smith 21648). — 37. C (Acaules) (Calandrinia) axilliflora ME 2786). — 35. С. (Acaules) 378 Annals of the Missouri Botanical Garden Volume 80, Number 2 Hershkovitz 379 Leaf Morphology of Calandrinia and Montiopsis FIGURES 42-45.‘ Epidermal preparations of Calandrinia sects. Calandrinia and Мана 'osmia. — 42. C. (Cal- andrinia) alba н 1223). 1, stauro-; three contact cells; outer cycle of three cells. 2, brachytetra-; outer cycle of cells. 3, stauro-; four contact cells; partial outer cycle of cells. —43. C. (Calandrinia) y ee eri Hoop er 6697). 1, brachytetra-; "boot distinct cycle of outer cells. 2, stauro-; five contact cells; weakly differentiated outer cycle of five cells. 3, more or less hemi-amphibrachypara-. — 44. C. (Calandrinia) axilliflora (Jose ph 2786). 1, stauro-; three contact cells; partial outer cycle of cells. 2, stauro-; four contact cells; partial outer cycle of cells. 3, brachypara-; one polar subsidiary cell; without distinct outer cycle of кие б С. Manse osmia) monandra (Joseph 3719). 1, brachytetra-; without distinct cycle of outer cells. 2, stauro-; five contact cells and nearly enclosing outer cycle of five cells. Scale bars are (um) ca. 100, 100, 100, and 50. лан о 'FicunES 42-69. Epidermal preparations of Calandrinia, AK Montia, and Rumicastrum. The suffix -cytic is omitted from stomatal types. Specimens are vouchered, Table €— FIGURES 38-41. Cleared leaf apices of Calandrinia and у > ifolia. 1l, primary vein; К, fimbrial vein. — 38. apres (Calandrinia) axilliflora (Joseph 2 ae . Calandrinia (Monocosmia) monandra (Buchtien 37).—40. Calandrinia carolinii (Venturi 4824).— Шы. parvifolia (Randall s.n.). All scale bars O um. 380 Annals of the Missouri Botanical Garden FIGURES 46-49.‘ Epidermal preparations d ес sect. Acaules. — 46. C. carolinii (Venturi 8786). 1, brachytetra-; partial outer cycle of cells. 2, stauro-; three contact cells and weak evidence of outer cycle of се 18. — 47. C. colchaguensis (Werdermann 1325). 1, i ytetra-; without a 7 outer diri e of cells. 2, irregularly brachytetra-; outer cycle of cells. 3, stauro-; Ls contact cells; without outer cycle ells. 4, more or less hemi- amphibrachypara-; one polar aci cell. 5, brachypara-. — 48. C. caespitosa (Sc hjovskoy s.n.). 1, brachytetra-; weak evidence of outer cycle of cells. eee five contact cells; partial outer cycle of cells. 3, via named apie , ћеха-. — 49. C. compacta ата 1035). 1, hemi- amphibrachypara- 2, tl brachypara-; two su iue cells adjacent to one of the guard cells. All scale bars — ca. 100 um —-— FiGURES 50-55.* Epidermal preparations of Mannes and Montia d 50. Montiopsis (Dianthoideae) gayana (Werdermann 1219). 1, amphibrachypara-. 2, hemi-amphibrach . 3, stauro-; four contact cells; nearly complete outer cycle of cells. ү. Montiopsis (Dianthoideae) andicola (Hastings 488). 1, hemi-amphibrachypara-. 5 IL; 2, brachypara-.— 52. s (Montiopsis) copiapina (Johnston 6 l, amphibrachypa ra-. 2, stauro-; four contact cells; weak evidence of an outer cycle of cells. — 53. Montiopsis (Montiopsis) gilliesii (Buchtien 1149). 1, hemi- ИРА 2 Ha sl У 54. en p i i parviflora (Johnston 5882). 1, anomo-; our contact cells. 2, anomo-; six contact cells. 3, anomo-; five cells. 4, hemibrachypara-. — 55. Montia ie, ска е et = 14865). 1, stauro-; four я = 2: eae 3, stauro-; three contact cells. All 00 y Volume 80, Number 2 Hershkovitz 1993 Leaf Morphology of Calanarinia and Montiopsis Annals of the Missouri Botanical Garden n C. Пе. к (Hoover с ‘arolinii (у enturi 4824). — Trichomes in M. o UM andicola Шаш and uniseriate towards ti seriate-multic Gla bars — ca. 100 um. with Kelley's data, but she reported papillae in the species of Calandrinia sect. Acaules. Figures 56-67 illustrate trichome types in Cal- andrinia and Montiopsis; for comparison, Figures 68 and 69 illustrate trichomes/papillae in Rumi- castrum and Montia. Trichome presence and gross morphology was evaluated for all examined her- barium specimens and studied in detail in epidermal preparations (see Table 1). || examined specimens of C. sect. Calandrinia um preparations of Calandrinia sects. Calandrinia and Monocosmia and кин 15 D eL 57. Ribbed unicellular papillae in C. (Acaules) nia) breweri (Hoover 6697). — . Ribbed unicellular hairs and . Ribbed unicellular pollo. in C. (Monocosmia) monandra (Joseph 3719). 60-62. 488). — 60. оне trichome, multiseriate towards base 1. Multi ellular trichome. —62. Bra ndular multiseriate-multicellular trichomes in M. TU abr. gayana (Sleumer 1219). АП scale nched multiseriate-multicellular tri- possess unicellular ribbed trichomes (Fig. 56), as reported by Kelley (1973). The trichomes may be up to ca. 600 um long but are sometimes merely papillar (Fig. 57). Unlike papillae in other genera (e.g., Cistanthe, Hershkovitz, 1991a; and Montia, Fig. 69), the outer tangential cell wall has a distinct U- ог V-shaped process rather than a more gen- eralized swelling. Thus, I distinguish between tri- chomes (process as long as or longer than broad), papillar trichomes (process broader than long), and Volume 80, Number 2 1993 Hershkovitz 383 Leaf Morphology of Calandrinia and Montiopsis Ficu vi = 69. Ере vifolia. —64. Multiseriate barbed tric tiser trichome s Montic opsis (Montiopsis) m (Тозері З ntiopsis (Monti 67. Glandular Тами multicellular trichomes (arro bed papillae Base of multiseriate barbed trichome ton 6069). — 68. Unicellular unrib — ome in Montiopsis е capitata (Fischer 169).— rmal preparations of Montiopsis subg. Montiopsis, Rumicastrum sp., = Montia par- . Macerated 865) showing barbed tips of individual a cells. — 66 iopsis) parviflora (Johnston 5882) showing elongate, ws) in Montiopsis (Montiopsis) copia- in Montia pisc (Calder et a 5) Unicellular энше к in Rumicastrum sp. (Unvouchered specimen K-242, courtesy of Walter А. Kelley.) All a. scale bars = c papillae (without a distinct process). I regard as papillar trichomes papillae sensu Peralta (1988; re: Calandrinia) and Gerbaulet (1992; re: Апа- campseros, etc.). The trichomes and papillar tri- chomes of Calandrinia may sparsely or densely populate the leaf margin and may occur severally adjacent (Fig. 57). eaves in members of C. sect. brous or bear unicellular ribbed trichomes and/or Acaules are gla- papillar trichomes identical to those found in C. sect. Calandrinia, carolinii (Fig. 58). Whether these trichomes are glandular, as inter- preted by Carolin (1987; see also Peralta, 1988), could not be determined in the present investiga- the unicellular tri- du С. tion. Structurally, however, chomes of C. multicellular and obviously glandular trichomes of Montiopsis (see below). Trichome presence varies affinis, C. carolinii, C. sect. Acaules are distinct from the within species C. tosa, and C. compacta. They may occur through- caespi- Annals of the 384 Missouri Botanical Garden (оцАоошоџе ‘-eredAyorsq) = = = V/N = = MES оцАоолоје] Y/N — = F + (+) V/N 01579'1 — = = V/N — = с-р оцАовлеф А цовла — V/N — B " + V/N 02иә7 (оцАоошоцв) E Е E V/N = = S-T оцАовле 4 Ацовла — V/N = + + - + ayunisi7) (on &2ouneis /-21191 -Ацовла) on&oouroue = = = V/N = = С-Ф 'оцАовлеа Ацовла = + = 4 + + опиоју = = = V/N = = G-p onÁ&oe1edáuoeuq m + = + + + DIUOJAD]’) (2n&2o1ne1s /ецәіќцэвлд) IDIPIOYIUDA] + 4. E = = (+)— C-p оцАовледАцовла Y/N — = + + V/N '3qns sisdoiopy (оцАоошопв) sisdonuopy + =e + = = Е c-b оцАовледАцовла V/N = — + + V/N 'Sqns sisdonuo[y оцАоолтеј5 отихоооиој - — — + (6—)+ + 6-} / епа цовла + + + + + + у pare (оцАовлеаАцовла) on&oounejs војпоојру = = = (V/N) + (-)+ -/- S-t /-ещәуќцэвїа + –—/+ —/+ + + V/N $ пимриојр) 9n&ooune]s DIUILPUD]D) = = = + + + с-ф /-ejayAyor sq + $ + + р + $ тимритјо) вошоцо зошоцо зошоцо peqqu ѕәшоцә s[peo зао addy xade je шәл xode је pajutol Suidse|o (quasaid uox? | ап ад ап зәшоцә ER ax] 198] [918 o]s ie[noip puq yeəm — oseq aseq j) әұешәѕ emj? еә ur renos -Álempis -uoo jo peoidÁ p -uadied -ung шәл pə, A родшм пи -nmu - OMA -un -qns jo Jequinu шәл Алеши sajonag лејпр ојрАо [eotdÁT үкүш] -ueg 19]n() ML (#1661 'zu^oxus1ag 295) оцАоолпеје /on4ov11o1Aqovaq 1dooxo (оцАовлед Ацовлацаше ‘-3-9) sjueuea [е орпјош 01 papuajut slay әле v]PUIOJS | IA (1 хгриодде “£661) 21лоуалон ut рәдѕц әле 1x9} уозола Əy} шолу рәлиәр JOU едер JO} заолпос ‘aided ои 42661 ZHAOAYSI9H aas) овоововјтлод UBILIJY /иеэпәшүұ иләјѕвә eulos ш juasa1d souroqoui Алејихе jou pue Á[uo “зәшоцощ ¡ended Surpn[our *seuroqotj лецој 01 19591 Vep euroqot r oeredAyorsg,, `Кцшв} sty] 10} езер рорпјош jou олец зопоц pue зваовјелојзон payesyseaut Ajayenbape jou әлец [ ‘ехе рорпјохе әѕәці jo syen əy; “UMOUA зелдор әч oi *j29ga1 exei UBILIJY /ивомошту ulojseo 991 лој етер әчү 'sso[pouiouoN (2661 "ZitAoXusroH gas) stsA[pue onstpe[o оца ш рәрпүәш jou әләм mq dno18 ueoujy /пвомешу uiajsea əy} Jo ѕләдшәш “A peousipepo ‘ere штутодашр pue 'зваоволорцј ‘283981989 'eveov[[oseg "Клио ә[#ш e se роја aie овоовоејпјлод чвомју /ивомешу uiojseo əy} pue snuad Aq pasy ore штијѕротшпу pue (2661 '#1661 'zu^oxqsioH ә) оезововјпо4 чеомешу илојзем 94, ‘sasayjuaied ш pojeorpur зле (uoxej e ut satoads OM] JO auo JO оцаодовлецо sojejs *a'l) ѕәде1ѕ әлеу "eeaoPov[niioq лешо ш uonnquistp ey) pue sisdoiuopy pue DIUIIPUD]D:) jo SUBA peo] СЄ 3TIHV[] Volume 80, Number 2 1993 Hershkovitz 385 Leaf Morphology of Calandrinia and Montiopsis Continued. TABLE 3. Glan- Outer dular Fimbrial Multi- multi- Multi- vein perpen- Primary Petioles Tri- Typical Leaf Leaf vein Fim- winged cellular seriate cellular cellular ~ ER per tri- chomes chomes E |] ~ stomatal chomes ribbed chomes cells present) clasping jointed at apex vein at apex type cells anisocytic? N/A (+) Taxon = (+2) N / A Rumicastrum (brachyparacytic) — (+ in 3 (4) +/— —, N/A Eastern American/ African group in Schreiteria) out the abaxial leaf surface of C. graminifolia. Trichomes are apparently absent in C. acaulis from Ecuador northwards, occur occasionally in Bolivian and Peruvian specimens, and are present in all three examined Argentinian herbarium specimens (*C. saltensis”). In effect, trichomes occur in С. acaulis only in its range overlap with C. carolinii. My observations of variable trichom section Acaules are consistent with Peralta (1988) but not Kelley (1973), the latter reporting the absence of trichomes in all species. me presence in e specimens of C. sect. Monocosmia (Fig. 59) possessed only unicellular ribbed papillar tri- chomes, which have a distinct U-shaped emer- gence. The unicellular trichomes and papillar trichomes of C. sects. Calandrinia, Acaules, and Monocos- mia are distinct from those of other Portulacaceae, e.g., Rumicastrum sp. (Fig. 68), in bearing ribs. As I have noted, I regard unicellular epidermal emergences in some Portulacaceae, e.g., Montia parvifolia (Fig. 69) and Cistanthe spp. (Hersh- kovitz, 1991a), as papillae rather than trichomes. Foliar trichomes in M. subg. Dianthoideae are multicellular and range from uniseriate (Fig. 60; only the basal cell is biseriate) to multiseriate (Fig. 61) to branched multiseriate (Fig. . If present, branched multiseriate trichomes typically occur along the leaf base margins and intergrade with vascularized leaf dentations (see above). The uni- seriate trichomes tend to occur along the leaf mar- gin, although in some specimens they occur oc- casionally on the lamina. Glandular trichomes are present or absent in M. andicola and M. gayana (Fig. 63). Some specimens of M. subg. Dianthoi- deae appear to be glabrous. While my observations on trichome structure in subgenus Dianthoideae agree with earlier reports (Kelley, 1973; Peralta, 8), the material | examined includes what I presume to be a disproportionate number of gla- brous or nearly glabrous specimens of otherwise pubescent species. Leaves (and sepals) of all examined species of M. subg. Montiopsis bear distinctive multiseriate trichomes composed of long intertwined hair cells. Ford (1992) noted that, in some species, the tips of the individual hair cells are bent outward so that 5). According to Kelley (1973), the multiseriate trichomes are several cells long, i.e., the individual hair cells are transversely divided. In the present study, individual hair cells traced to their point of origin at the leaf surface (Fig. 66) did not show evidence of transverse walls. Apparently the hair 386 Annals of the Missouri Botanical Garden TABLE 4. Foliar characters, character states, and TABLE 4. Continued. scores for Portulacaceae used in the present cladistic exercise. The character numbers accord with, and the Ceraria 1001010X100000X character analyses are provided in, Hershkovitz (1993: Cistanthe $ Amarantoideae 011?110X000000X appendix 1). Polarities, where ic ure are scored as Cistanthe $ Calyptridium 011?110X000000X "0" in the hypothetical ancestor. The ' designation — Cistanthe $ Cistanthe 011V110X000000X also iud ан ни riori Fd ‘termined ini state s. Oth- Сое КИБ др 011?110X000000X bin "ii s A pn Cistanthe 8 Strophiolum 011?010X000000X states are вз ad зна Doyle & осе (1986) | | and are analyzed as missing data in PAUP. Other missing Claytonia 01100011000000Х data scores are denoted as 72" (unknown) or "V" (too Grahamia 000?100X100000X variable within the OTU to determine). Character states Lenzia 011?000X000000X variable within an OTU but scored for the poene Lewista 0110000X000000X primitive state are underlined in the data matrix Montia (excl. $ Vaiocrene) 01100011000000X Montia $ Naiocrene 1100011010000X 7. Leaves alternate, 0; leaves opposite, | Montiopsis 8 Dianthoideae 0110000X0011110 8. Leaf bases jointed, О; leaf bases not jointed, l. Montiopsis § Montiopsis 0110000X0011110 9. Leaf bases clasping, 0; leaf bases not clasping, к Portulaca V001010X100000X 10. Leaf mucilage cells conspicuous and relatively nu- Portulacaria 1001000X100000X merous, 0; leaf binge d cells none or few and not | = Rumicastrum 000?100X?00000X conspicuous and few, 1. М Pe = 11. Ribbonlike veins eet 0; ribbonlike veins pres- Schreiteria 000?100X010000X ent, 1 Talinaria 0001000X100000X 12. Leaf venation not festooned brochidodromous, 0; leaf Talinopsis 100?100X100000X venation festooned brochidodromous, 1. Talinum $ Phemeranthus 0001100X100000X 13. Fimbrial vein absent, 0; fimbrial vein present, Talinum $ Talinum 0001110X1010000 14. Fimbrial vein bec ese perpendicular to the primary 7 Е = 21 vein at the КА ; fimbrial vein parallel to the at the apex, 1; fin 15. Stomata mci brachyparacytic and/or and lershkovitz, 1991a), 0; predominantly parallelocytic and/ or nonbrachypara- primary vei ibrial vein absent, X. similar types (see ; stomata cytic, 1. 16. Stomata predominantly brachyparacytic, nonbrachy- stomata commonly paracytic, or parallelocytic, 0; brachytetracytic and/or staurocytic, 17. Foliar trichomes absent, О; foliar trichomes pres- en 18. Tric cine unicellular, 0; trichomes multicellular, 19. Trichomes uniseriate, 0; trichomes multiseriate, | 20. Trichomes SS 0; trichomes eal 2. — 21. Trichomes md d, 0; trichomes ribbed, QX chomes abser OTU 7 21 Ancestor 0???0?0X?0????X Anacampseros 8 Anacampseros 0001000X1010000 Anacampseros § Avonia 0001?00X100000X Anacampseros 8 Tuberosae 0001?00X101000X Calandrinia acaulis 01100010010000X Calandrinia affinis 011000100110001 Calandrinia caespitosa 0110000X0110001 Calandrinia carolinii 011000100110001 Calandrinia compacta 0110000X0010001 Calandrinia $ Monocosmia 011000100110001 Calandrinia $ Calandrinia 011000100110001 Calyptrotheca 000???0X100000X cells divide infrequently, if at all (cf. Becker, 1895). In any case, these trichomes sparsely populate or, in some species, densely cover the leaf surface and may be as long as ca. З mm. I found glandular trichomes resembling those of M. subg. Dianthoi- deae in only one specimen of M. subg. Montiopsis (M. copiapina, Fig. 67), but they have been re- ported in several other species (Ford, 1992; Kelley, 1973). The glandular trichomes ranged from few- celled and uniseriate to multiseriate. Unlike the eglandular trichomes, the elongate cells of the mul- tiseriate-glandular trichomes are clearly trans- versely divide B. PHYLOGENETIC ANALYSIS Table 3 compares the foliar traits of Calandrin- ia and Montiopsis with those found in other Por- tulacaceae. Only those traits of Calandrinia and Montiopsis that are differentially distributed among other Portulacaceae are listed, e.g., linear leaves occur in most genera of the family and hence are not listed. Figure 70A and B compares consensus trees derived from cladistic analysis of all (from Hershkovitz, 1993) versus only nonfoliar charac- ters, respectively. Figure 71 is a consensus derived from trees generated using weighted foliar char- acters. T ] definitions, data, and character analyses are presented elsewhere (Hershkovitz, Volume 80, Number 2 1993 Hershkovitz 387 Leaf Morphology of Calandrinia and Montiopsis TABLE 5 rescaled consistency indices n Figures ТОА, B a generate the consensuse ) for all ‚= foliar characters, and nonfoliar characters Tree statistics for Figures 70A, B and 71. Listed below are the lengths, — indices (CI), and dices (RC r the trees used to e statistics vary among the different trees used to generate Figure 70A when ранени only the chris ee Note that CI and RC vary inversely with respect to tree g length for the character subsets for Figure 70A. Length Length Length (non- (all (foliar foliar Consensus char- char- char- tree acters) CI RC acters) CI RC acters) CI RC Figure 70A 107 0.449 0.346 29-32 0. "S 7-0.469 0.441- x 385 75-78 0. et 0.423 0.326-0.306 i 7 0B 108 0.444 0.341 34 0.441 35 74 446 33: Figure 71 121 0.397 0.284 25 0.600 dem 96 a 0.210 1993), but, for convenience, the foliar data scores are listed in Table 4. Figures 70 and 71 do not show the speculative positions of Basellaceae, Cac- taceae, Didiereaceae, or Talinella (cf. Hershkovitz, 1993, fig. 1). Table 5 lists the lengths, consistency indices (CI) and rescaled consistency indices (RC) of the trees on which the consensuses are based. For each treefile, the tree statistics for all, foliar, and just nonfoliar characters are compared. Not all of the characters listed in Table 3 were included in the cladistic analysis (see Hershkovitz, 1993, appendix 1): petiolar morphology, primary vein morphology at the leaf apex, and the presence just of an outer cycle of subsidiary-like cells were ex- cluded. Figures 70B and 71 were each independently derived by two procedures: (1) searching under various PAUP modes (Swofford, 1991; see Hersh- kovitz, 1993) using weights of 100 for the nonfoliar and foliar character sets, respectively, followed by weight equalization to derive comparable tree sta- tistics; and (2) searching using only the desired character set, followed by restoration of the ex- cluded characters and screening for the shortest trees. The aim in the latter procedure was to re- cover the shortest trees based on the included data that were otherwise most consistent with the ex- cluded data. The dashed lines in Figures 70B and 71 denote nodes not supported by at least one nonfoliar or foliar character state change, respec- tively. These branches would be collapsed in a consensus of all trees generated using nonfoliar or foliar characters exclusively. I aborted the searches using weighted nonfoliar characters when 9,000 trees were saved. Inasmuch as the consensus of these trees (Fig. 70B) is similar to that of trees generated using all ch ters equal ly weighted (Fig. 70A), I suspect that the number of equally parsimonious trees is actually more than an order of magnitude greater (cf. Hershkovitz, 1993). Despite their similarity, the consensuses compared in Figure 70 are based on completely different sets of trees: those yielding Figure 70A are 107 steps, and those yielding Figure 70B are, over all characters, 108 steps. Nevertheless, the unambiguous character state changes defining the nodes in Figure 70B are essentially those defining the nodes in Figure 70A (see Hershkovitz, 1993) minus those that become ambiguous when some nodes are collapsed. Hence, 1 do not show the unambiguous character state changes on the Figure 70B tree. The consensus in Figure 71 is based on 490 trees found using weighted foliar characters and AUP's tree bisection/reconnection algorithm to completion. The unambiguous foliar character state changes are indicated and numbered according to Hershkovitz (1993, appendix 1). DISCUSSION 1. TAXONOMIC SIGNIFICANCE OF FOLIAR MORPHOLOGY IN CALANDRINIA AND MONTIOPSIS A. Phenetic comparison of foliar traits of Cal- andrinia and Montiopsis. Table 3 emphasizes the degree of homogeneity of foliar morphology within Calandrinia and Montiopsis, the distinctions in foliar morphology between these genera and other western American Portulacaceae, and the addi- tional distinctions between these genera and eastern American/ African Portulacaceae (cf. Hershkovitz, Calandrinia is distinguished by the combination of the fimbrial vein that overtops and becomes perpendicular to the primary vein at the leaf apex, the predominance of brachytetracytic/staurocytic stomata commonly with an outer cycle of subsid- iary-like cells, and unicellular-ribbed trichomes. These features are absent in some species of Cal- 388 Annals of the Missouri Botanical Garden ——— Talinaria Anacampseros 8 Anacampseros -— Anacampseros S Avonia — - Anacampseros S Tuberosae - — ———— Grahamia -—-—-—-— Talinopsis Portulaca —— Talinum S Talinum —— Schreiteria ———-- — Talinum 8 Phemeranthus — -——— Calyptrotheca Ceraria ——-—- Portulacaria ———— WNNNNNNNNNNNNNNNNNNNM -——— Rumicastrum = ————— АМАМ — Calandrinia compacta — -—- Calandrinia acaulis —— аы (УУУУУУУУ» —- Calandrinia caespitosa -— -—- Calandrinia carolinii -—- -—— Calandrinia affinis —— - Calandrinia § Calandrinia - - Calandrinia § Monocosmia - - Montiopsis § Dianthoideae - -- Montiopsis § Montiopsis — — —— | - Cistanthe S Amarantoideae - - Cistanthe S Calyptridium - - Cistanthe SPhilippiamra - = |- — Cistanthe S Cistanthe —- E -- Cistanthe S Strophiolum — ——— Claytonia —— —- Montia $ Naiocrene — J—— — Montia excl. S Naiocrene -- — ______ т лз аы RSS ANE RU а Lewisia тла —————- КЕ 70. Juxtaposed consensuses of cladograms of Portul ted using all characters equally weighted versus ae nonfoliar characters. Table 5 lists length and consistency statistics. Shaded branches delineate the eastern American/African taxa (Hershkovitz, 1993), and the diagonally hatched branch denotes Australian ende- mism.— 70А. Consensus of cladograms of Portulacaceae den using Squaly weighted foliar and nonfoliar characters. This consensus and its unambiguous character state changes are discussed and illustrated elsewhere (Hershkovitz, 1993: fig. 1). The Recte positions of Po. Ges e, Didie ereaceae, and Talinella, however, using nonfoliar characters exclusively. All except the dashed branch are at least equivocally supported by at least one nonfoliar character state change. The dashed branch is supported only by a foliar character state change. The ME character state changes in Figure 70B include those indicated н for Figure 70A (Hershkovitz, 1993: fig. 1) minus those ne become ambiguous because of decreased resolutio Volume 80, Number 2 Hershkovitz 389 1993 Leaf Morphology of Calandrinia and Montiopsis Talinaria Anacampseros S Anacampseros TE Anacampseros S Tuberosae e Anacampseros § Avonia о“ Grahamia о 12 meses: Talinum S Talinum Talinum § Phemeranthus Portulaca PEL " T [ — MR 7t Portulacaria nien Talinopsis Calyptrotheca *8,9,10 16 Schreiteria hhh Rumicastrum Calandrinia compacta Calandrinia acaulis ТИТ Claytonia Montia excl. § Naiocrene Montia $ Naiocrene Calandrinia carolinii Calandrinia affinis 117 Calandrinia $ Calandrinia Calandrinia $ Monocosmia Calandrinia caespitosa TTITIIT Montiopsis $ Montiopsis Montiopsis § Dianthoideae Cistanthe S Amarantoideae Cistanthe $ Calyptridium Cistanthe 8 Philippiamra Cistanthe $ Cistanthe Cistanthe $ Strophiolum (71) Lewisia Lenzia FicURE 71. Consensus of trees generated using weighted foliar characters. The foliar character definitions, states, and OTU scores are reproduced in Table 4. Table 5 lists length and consistency statistics. Shaded branc hes delineate the eastern CAST / African taxa арену 1993), and the ван hatched branch denotes на endemism. The trees were generated by 100-fold weighting of 15 foliar characters from Hershkovitz (1993: table 3 and appendix 1, chars. 7-21) and, independently, by using foliar таны powan followed by reinstatement of nonfoliar characters and filtering for the shortest trees. Unam iguous foliar character state changes are indicated. Polarized characters are shown in black and unpolarized in white. “X” and “t” indicate that OTUs at adjacent branches were scored as ps data (**?”) because of “X” character definitions or uncertainty (see Table 4). All except the dashed branches are at least equivocally supported by at кай one foliar character state change. The dashed branches are i ed cu by nonfoliar character state changes 390 Annals of the Missouri Botanical Garden andrinia sect. Acaules, however. Fimbrial veins, although morphologically distinct, characterize Claytonia and Montia, and brachytetracytic sto- mata characterize one species of the latter. Brachy- gularly occur in Mon- and tetracytic stomata also re Dianthoideae, Schreiteria, ong other stomatal types in Cis- tanthe (Hershkovitz, 19914). other Portulacaceae but are not ribbed and, in Montiopsis and Атрћре ит Bacigalupo (in Spi- chiger, 1988), Glandular or eglandular multiseriate trichomes are restricted to Trichomes occur in are multicellular. Montiopsis and represent the sole foliar distinction between this genus and other Portulacaceae. Both Calandrinia and Montiopsis share with other western American Portulacaceae their leaf base and petiolar morphology and stomatal contact cell number. These features are generally not pres- ent in eastern Атегісап/ African Portulacaceae. Well-differentiated petioles are absent in C. sect. Acaules, however. One species of Rumicastrum (“Calandrinia pumila F. Muell.") has both winged petioles and brachyparacytic stomata with four contact cells (Hershkovitz, unpublished). The relatively pronounced diminution and/or distal ramification of the leaf primary vein in Cal- andrinia sects. Calandrinia, Monocosmia, and species of section Acaules may provide a synapo- morphy for all or part of the genus. In some mem- bers of Portulacaceae, e.g., Lewisia, the primary vein is always prominent and/or unramified at the leaf apex, whereas other taxa, e.g., Cistanthe and various eastern American/ African Portulacaceae, always have apically weak to indistinct primary veins (Hershkovitz, 1991a, unpublished). Some taxa, however, are intermediate for this feature, and evaluation of primary vein diminution is some- what subjective. Traits not compared in Table 3 include dentate leaf margins, the number of distinct vein orders, and the number of veinlets in the ultimate enclo- sures. The dentate leaf margins that characterize Montiopsis subg. Dianthoideae are here regarded as unique. Dentate leaf margins in species of Lew- isia occur toward the apex of the leaf, rather than, as in M. subg. Dianthoideae, at the extreme base. I have not scrutinized veinlet numbers throughout Portulacaceae. Nevertheless, the paucity of freely ending veinlets might reinforce the relationship be- tween Calandrinia sects. Calandrinia and Mono- cosmia, because veinlet numbers are higher among members of C. sect. Acaules. The poor differen- tiation and/or low number of distinct vein orders of Calandrinia characterize similar-sized leaves in species of several genera throughout Portulaca- ceae. B. Phylogenetic analysis of foliar traits of Calandrinia and Montiopsis. Тһе purpose of the cladistic exercise in this study was to evaluate the degree of taxonomic utility of foliar characters, in particular for clarifying the systematics of Cal- The degree of agree- ment between phylogenetic estimates based on fo- andrinia and Montiopsis. liar versus nonfoliar characters might be taken as a reliability index of each character set, and the ability of the foliar set to provide additional reso- lution or “break ties” in analyses based on the nonfoliar set might be taken as a measure of com- plementarity. If the character sets suggest conflict- ing phylogenetic resolutions, the data set providing the greatest degree of congruence with all char- acters combined might be regarded as superior. Evaluation of the utility of foliar versus nonfoliar characters requires additional considerations as well, however, including the size of the respective data sets and the degree to which each taxon has been investigat Superficially, the cladistic analysis suggests that the foliar data tend to misinform. The Figure 71 consensus tree derived from analysis of weighted foliar characters shows significant similarities but also departures from the consensuses of trees gen- erated using all characters (Fig. 70A) or weighted nonfoliar characters (Fig. B). When measured with all characters equally weighted, the trees gen- erated with weighted foliar characters are 13% longer than those generated with equal weightings (Table 5). In contrast, the consensuses of the Figure 70A and 70B are nearly identical even though the sets of trees used to generate each consensus are necessarily mutually exclusive. Thus, the analysis suggests that the foliar characters are somewhat dispensable. These results must be evaluated in the context of results for individual subclades and char- acters. The foliar versus nonfoliar data suggest rather different relationships for Calandrinia and Mon- tiopsis. The nonfoliar data (Fig. 70B) indicate that Calandrinia is monophyletic and possibly (but not necessarily) the sister group of Montiopsis, and t Calandrinia sects. Calandrinia and Mono- cosmia are sister groups. The foliar data (Fig. 71) indicate that Calandrinia—with Claytonia and Montia nested therein—is a sister group to Mon- tiopsis, and that sections Calandrinia and Mono- cosmia may not be sister groups. In both consen- suses, Montiopsis is monophyletic. The combined Volume 80, Number 2 1993 Hershkov Leaf MEO) of Calandrinia and Montiopsis data (Fig. 70A) agree more with the nonfoliar data, conserving only the Calandrinia + Montiopsis clade suggested by the foliar data. Scrutiny of Figure 71 reveals that both the Calandrinia + Montiopsis clade and the inclusion of Claytonia and Montia therein are not strongly supported even by the foliar data. The sole synapo- morphy of the Calandrinia + Montiopsis clade is the derivation of trichomes (char. 17), which, as I have noted, are structurally distinct in the two groups. My character definition and scoring in this case was intended to permit trichomes to be ho- mologous in the two groups unless contradicted by other evidence. In the case of Figure 70A (see Hershkovitz, 1993) and Figure 71, contradicted nor independently supported. Like- it is neither wise, the inclusion of Claytonia ana Montia re- flects my effort, via the use of “X” characters (Doyle & Donoghue, 1986; Hershkovitz, 1993), to permit the structurally distinct fimbrial veins (char. 13) in the two groups to be homologous and also, via recognizing Montia sect. Naiocrene as a distinct OTU, to allow the homology of brach ytetra- cytic/tetracytic stomata (char. 16). These char- acters provide the sole evidence for the inclusion of Claytonia and Montia within Calandrinia, and brachytetracytic/staurocytic stomata must then be lost in Claytonia and most Montia species. While the foliar evidence provides at least some support for the inclusion of Claytonia and Montia within Calandrinia, excluded foliar evidence might refute it. I did not include petiole presence in my analyses because all OTUs with petiolate taxa in- clude taxa without petioles. Petioles characterize Calandrinia sect. Monocosmia, all species of sec- tion Calandrinia (although variable within species of the latter), Montia sect. Naiocrene, and several species of Claytonia and the remaining Montia sections. Petioles are lacking (or hardly distinct) in Calandrinia sect. Acaules. In the context of Fig- ure 71, this evidence could provide a synapomor- phy for Calandrinia sects. Calandrinia and Monocosmia and would appear to cause Claytonia and Montia to become, at least equivocally, a sister group of the first two or fall out of the Calandrinia clade altogether. The only robust clades that would remain would be those supporting a close relation- ship between the sections of Calandrinia and the monophyly of Claytonia + Montia (cf. Fig. 70B). For other Portulacaceae, cladistic analysis of foliar versus nonfoliar characters also yields only partial congruence. Figures 70B and 71 concur in diagnosing the monophyly of Claytonia + Montia, Montiopsis, the eastern American/ African group + Rumicastrum, and Anacampseros + Talinaria (see also Gerbaulet, 1992). For Montiopsis and the eastern American/ African group + Rumicas- trum, the foliar evidence provides substantial cor- roboration of nonfoliar evidence (Fig. 71; cf. Hershkovitz, 1993, fig. 1) agree regarding the precise relationships within the . The consensuses dis- eastern American/ African clade. Reconciliation of these disagreements is beyond the scope of the present paper: I have not adequately scrutinized either the taxonomy or leaf morphology of the eastern American/ African group for this purpose. The apparent concurrence between Figures 70B and 71 regarding the monophyly of Cistanthe is artificial. Analysis with weighted foliar characters yields the result shown in the Figure 71 consensus in which brochidodromous leaf venation is a syn- apomorphy of Cistanthe. This evidence becomes equivocal using foliar characters only (i.e., by col- lapsing the dashed branches in Fig. 71). Thus, the foliar evidence complements the nonfoliar for di- agnosing monophyly of Cistanthe (cf. Hershkovitz, 1993, fig. 1). A peculiar result of the present analysis is that the Calandrinia + Montiopsis + eastern Amer- ican/ African group + Rumicastrum clade is sup- ported in all trees generated using all (Fig. 70A) or weighted foliar characters (Fig. 71). This clade is equivocal in the nonfoliar consensus (Fig. 70B); hence one might surmise that the clade must be supported by a foliar character. In fact, the only character supporting this topology in Fig. 70A is the derivation of panaperturate pollen (see Hersh- kovitz, 1993). The reason that the clade appears unequivocal in the foliar weighted trees is because no evidence contradicts the homologization of tri- chomes in Calandrinia + Montiopsis. As a result, panaperturate pollen becomes equivocally primi- tive in Montiopsis, which then forces the homology with equivocally primitive panaperturate pollen in the eastern American/ African group + Rumicas- trum. | fortuitously discovered that scoring Cal- andrinia caespitosa as absent for trichomes is sufficient to cause, among other things, the Cal- andrinia + Montiopsis + eastern American/ Af- rican group + Rumicastrum clade to become equivocal. That a minimal data modification can have such major ramifications should elicit caution in interpreting cladistic results, especially in this case because trichomes are only variably present in species of Calandrinia sect. Acaules. It is possible that differences between the con- sensuses in Figures 70 and 71 reflect artifacts that might be expected given the structure of the data. 392 Annals of the Missouri Botanical Garden The foliar and nonfoliar data comprise, respec- tively, 15 and 31 characters for 33 taxa. A larger character set should yield a more reliable result, although the result should not necessarily differ from that using a smaller set. Perhaps consistent with this principle, the 31-character nonfoliar data consensus is more similar to the 46-character com- bined data set consensus than is the 15-character foliar data consensus. Yet, obviously contradicting the principle, the 31- and 46-character data sets yield virtually identical consensuses, despite. the substantial difference in data set size. Another consideration is the claim that, for a given number of taxa, the perceived consistency index (СЇ; = the data divided by the length of the cladogram; the number of apomorphic states in i.e., the degree of perceived homoplasy), should vary inversely relative to the number of characters in the data (Archie, 1989, 1990). Because the C is often interpreted as a measure of the reliability — of the data and accuracy of the tree, factors that artificially inflate the CI (in this case, a small char- acter data base) should also distort the cladistic results. Alternatively, Sanderson & Donoghue (1989) found no decisive evidence of a relationship between the number of characters and the СТ, although they agreed that the СТ may not ђе a The data in Table 5 superficially appear to support the view of Archie (1989, 1990): using the smaller foliar data set is much higher good indicator of cladogram accuracy. the CI of trees generated than that of trees using the larger nonfoliar data set. The CI of trees generated using all 46 char- acters, however, is slightly greater than that of trees generated using 31 nonfoliar characters. Moreover, the Cls for individual foliar characters (i.e., the number of apomorphic states of the char- acter divided by the number of state changes in the tree) are, in general, higher than for nonfoliar, even in trees generated using the combined data set (Table 5). A possible artifact in the CI data arises when some characters vary in only a few OTUs, e T the trichome traits that are restricted to the two OTUs of Montiopsis. The CI for these three char- acters cannot possibly be lower than 0.5. In other words, irrespective of phylogeny, the maximum number of state changes for these characters is two. The CI of a binary character with ca. half of the OTUs like-scored can approach (but not equal) zero, e.g., if each apomorphic state is independently derived. Farris (1989) proposed a rescaled consis- tency index (RC), which adjusts for biases in char- acter state frequencies, although Archie (1990 Because PAUP calculates — criticized. this statistic. RC and not Archie's (1989, 1990) proposed “‘ho- moplasy excess ratio," I list RC in Table 5. For the present data, the RCs are lower than but more or less parallel to the Cls, and the RCs of the nonfoliar rather than the foliar characters appear to be more depressed relative to the CIs. Perhaps this is not surprising: while some of the foliar char- acter states vary among relatively few taxa, others, e.g., leaf base morphology (chars. 8, 9), have ро- tentially very low but actually very high CIs. More rigorous and sophisticated analyses are feasible for assessing the cladistic behavior of foliar versus nonfoliar characters in Portulacaceae, but the significance of foliar data should be clear from the evidence presented here. For example, the foliar evidence generally supports a close relation- ship among the sections of Calandrinia (the spu- rious results for Claytonia and Montia notwith- standing), the monophyly of Montiopsis, and, weakly, a sister group relationship between the last two genera (Fig. 71). consistent with those based on nonfoliar data (Fig. These results are at least ТОВ) and corroborate my earlier propositions Hershkovitz, 1990, 1991b, 1993) regarding the systematics of Calandrinia and Montiopsis. Spe- ~ cifically, the foliar data support the close relation- ship of Calandrinia sect. Monocosmia to Cal- andrinia as proposed by Carolin (1987) and refute earlier taxonomic opinions on a close relationship between section Monocosmia and Cistanthe sect. Calyptridium (McNeill, 1974; Nyanyano, 1986a, 1990; Pax & Hoffmann, 1934). In supporting the coherence of Calandrinia, the leaf data also refute Carolin’s (1987) results linking Calandrinia sects. Calandrinia and Monocosmia with Rumicastrum. The leaf data might be interpreted as supporting (or at least not refuting) Carolin’s alternative pro- posal that Calandrinia sects. Calandrinia and Monocosmia are closely related to Montia. A close relationship between Calandrinia and Claytonia + Montia remains a possibility based on all char- acters (cf. Hershkovitz, 1993, table 4) although the precise relationship proposed by Carolin is prob- ably incorrect. Although not the focus of the pres- ent paper, the leaf data also provide strong support for the distinction between the eastern American/ African and western American Portulacaceae. Fi- nally, analysis of the foliar versus nonfoliar data or Portulacaceae yields results consistent with 1986) that foliar characters are more conserved evolu- opinions put forth elsewhere (e.g., Levin, tionarily than implicit in traditional taxonomies em- phasizing reproductive characters. Although the taxonomic utility of foliar characters is now widely 1988: 176-184; appreciated (e.g., Cronquist, Volume 80, Number 2 993 Hershkov Leaf ооо of Calandrinia and Montiopsis 393 Takhtajan, 1991, chapter 1), the opinion that foliar and other vegetative organs “аге characterized by more reversibility than reproductive organs" (Takhtajan, 1991: 21) persists. In fact, in the present study, the foliar characters proved to be more conserved than the nonfoliar in cladograms derived using all characters and about equally con- served as nonfoliar in cladograms in which the nonfoliar characters are exceedingly weighted. Of course, some foliar characters, e.g., petiole pres- ence, were excluded from the analysis because they appear to be inherently homoplasious, but the same is true of some reproductive characters, e.g., sta- men and ovule number. 2. DEVELOPMENTAL, FUNCTIONAL, AND EVOLUTIONARY CONSIDERATIONS OF FOLIAR MORPHOLOGY IN CALANDRINIA AND MONTIOPSIS In my consideration of foliar morphology of Cis- tanthe (Hershkovitz, 199 la), ideas put forth on the adaptive value of leaf shape, I briefly discussed leaf venation pattern, and stomatal complex mor- phology. Those comments are applicable here. For Calandrinia and Montiopsis, the functional sig- nificance of trichomes must also be considered. In addition, models attempting to explain venation and epidermal patterning as a consequence of cellular- level physiological phenomena during early tissue differentiation must be reconciled in the light of detailed systematic studies of foliar morphology such as that presented here and elsewhere. Givnish (1979) supposed that leaves not re- quiring much physical support should tend to have more parallel rather than pinnate venation in order to optimize vascular supply efficiency. As I noted previously (Hershkovitz, 1991a), venation in Cis- tanthe does not seem to conform to this expec- tation, nor does that of Calandrinia and Mon- tiopsis. The largest/broadest leaves among these two genera are the basal leaves of Calandrinia sects. Calandrinia and Monocosmia—these li o more or less prostrate on the ground, hence require little internal support. Leaves of Montiopsis and alandrinia sect. Acaules also seem to lack the need for much internal support because they tend to be succulent (at least relative to their dimensions) and may also lie prostrate on the ground. Yet, venation is basically pinnate in both genera, even in bun with a more or less grasslike habit. The tion of the primary vein and well-developed fimbrial vein in Calandrinia sects. Calandrinia and Monocosmia and some species of section Acaules might be perceived as a sort of parallel venation pattern, but, in contrast to grasses and other taxa with parallel veins, the fimbrial vein is best developed toward the apex and is poorly to not at all distinct basally. The functional signifi- cance of the fimbrial vein itself is not clear. This structure could conceivably be associated with some type of leaf apical secretory function, but no ob- vious glandular structure was identified at the apex of these leaves. Whatever the functional signifi- cance of the fimbrial vein, some species of Cal. andrinia sect. Acaules appear to survive well with- out it. As I noted previously (Hershkovitz, 19914), the functional significance of stomatal complex mor- phology — specifically, the number, shape and ar- rangement of subsidiary cells—remains unknown; hence I can offer no explanation for the predom- inance of brachytetracytic / staurocytic stomata and partial to complete outer cycle of subsidiary- -like cells in Calandrinia sects. Calandrinia and Mono- cosmia and species of section Acaules. Possibly the more or less radial rather than bilateral sym- metry of the stomatal complex in these taxa un- derlies its functional significance. Again, the lack of brachytetracytic /staurocytic stomata in C. t least one species of Montiopsis subg. Dienthouisas must (Acaules) compacta and their presence 1 not be overlooked when considering function. In the absence of ultrastructural and/or phys- iological evidence, the functional significance of trichomes in Calandrinia and Montiopsis remains conjectural. For example, the multiseriate tri- chomes composed of intertwined hair cells of Mon- tiopsis subg. Montiopsis, because of their size, density, and collective wall thickness, might have a protective function (see, e.g., Ehleringer, 1984). The pubescence in most of the species appears sufficiently dense to substantially modify the leaf boundary layer parameters and hence possibly offer protection in the desert and xeric montane habitats where the species occur. When viewed with po- larized light, these trichomes produce intricate, col- orful diffraction patterns and hence must also mod- ify the quality of solar radiation reaching the photosynthetic tissue. Such speculations on tri- chome function would require comparative ultra- structural, chemical, and ecophysiological study to verify (e.g., Behnke, 1984; Kelsey et al., 1984). Foliar venation and stomatal patterning have also been considered on purely developmental grounds (Lyndon, 1990; Meinhardt, 1982; Sachs, 1989, 1991). It has been supposed that the reg- ularity of foliar morphological patterns is a function histogenetic autocatalysis that leads to the devel- opment of veins or stomata in one locus and passive or active inhibition of such development in adjacent 394 Annals of the Missouri Botanical Garden loci (Meinhardt, 1982; Sachs, 1989, 1991). Com- puter simulations of growth in which one or two parameters corresponding to the concentration of physiological regulatory substances are modified have generated intricate patterns such as those ws leaf venation (Meinhardt, 1982; Mitchison, Rh ibe inductive effect of auxin on vascular development, Sachs’s (1989, 1991) “can- alization" hypothesis of vein patterning maintains that, in leaf primordia, the veins are initiated as preferred channels for draining endogenously pro- duced auxin. As the vascular strand develops, its conductive efficiency increases both absolutely and relative to the conductive efficiency of the sur- rounding cells, which are then passively inhibited from differentiating as vascular tissue. As the leaf expands, and local drainage patterns induce new veins to connect to existing veins, the venation pattern is elaborated. “The difference between vein systems of various leaves would be due to different parameters of auxin formation and transport and to differences in the details of early leaf growth" (Sachs, 1989: nisms underlying stomatal complex patterning have also been elaborated (Lyndon, 1990; Meinhardt, 1982; Sachs, 1991) Developmental models of foliar morphological 1). Ontogenetic causal mecha- pattern formation provide the reductionist toe-hold necessary to facilitate experimental analysis, but systematic characterization of patterns, especially when augmented by phylogenetic analysis, provides the phenomena to be explained, a means for as- sessing the general applicability of a particular model, and a source of natural evolutionary ex- periments in which developmental differences pre- dicted from models can be studied with the lowest possible degree of interference from evolutionary artifact. The fimbrial vein in Calandrinia provides a case in point. Sac hs’ s (1989, 1991) hypothesis implies that vein | should be proportional to the importance of the vein for auxin transport during the course of development, and that the auxin source-sink relationship between leaves and roots, respectively, should most strongly influence the attenuation pattern of at least the major veins. It is not obvious why the fimbrial vein in Calan- drinia—often more prominent than the primary vein in the distal portion of the lamina—should be the preferred channel for auxin drainage. More- over, the fimbrial vein is exceedingly prominent apically and indistinct basally, i.e., toward the roots. Thus, it might be appropriate to test whether this atypical structure contradicts Sachs’s model or is somehow an “exception that proves the rule." The model also predicts that biochemical developmental distinctions should be detectable among the species of Calandrinia sect. Acaules according to whether or not a fimbrial vein is present in the mature tissue. unfortunate aspect of the developmental models is that they appear to uncouple mature tissue structure from mature tissue function. The emphasis on the independence of specific physio- logical events at different developmental stages is beneficial in dispelling notions of predetermination but may obscure the interdependence of these same events in phylogenetic space. Aspects of the adap- tive value of leaf form such as those discussed by Givnish (1979) are undoubtedly significant in de- termining the survival of a phyletic line. Thus, the role of selection on the mature phenotype as a feedback on the evolution of biochemical ontogeny must be appreciated. Sachs (1991) discussed the potential role of epigenetic selection upon the evo- lution of pattern formation but emphasized selec- tion during ontogeny rather than selection for func- tionality after differentiation is completed. Neither the mature foliar morphological patterns, nor on- togenetic studies of relatively few and not closely related species can, by themselves, reveal if and how developmental programs have been rewritten during phyletic evolution. Syntheses of information on developmental mechanisms with knowledge of the diversity of form and phylogenetic reconstruc- tions of the relationships between different forms may thus provide a means through which general theories of development can be advanced and eval- uated. CONCLUSIONS Foliar morphological evidence corroborates re- vision of the generic limits of Calandrinia and Montiopsis. Calandrinia is characterized by leaves with a fimbrial vein that overtops and becomes mau. to the primary vein at the leaf apex, brachytetracytic/staurocytic stomata commonly with at least a partial outer cycle of subsidiary-like cells, and unicellular ribbed trichomes along the leaf margin. These traits are not universal in Cal- andrinia sect. Acaules. Structurally distinct fim- brial veins characterize Claytonia and Montia, and brachytetracytic stomata characterize one species of the latter. The foliar evidence that Claytonia and Montia are nested within Calandrinia is con- tradicted by nonfoliar and perhaps additional foliar characters. unique trichome types. Only the presence of tri- chomes per se supports a close relationship between Calandrinia and Montiopsis. Phylogenetic anal- ysis of foliar characters alone strongly supports a Montiopsis is characterized by its Volume 80, Number 2 Hershkovitz Leaf Morphology of Calandrinia and Montiopsis phylogenetic distinction between the eastern Amer- ican/ African and western American Portulaca- ceae, with the Australian endemic Rumicastrum belonging to the former. The analysis also indicates that foliar characters may be as well conserved evolutionarily as nonfoliar. Existing models relating leaf form to function and ontogeny are generally too rudimentary for reconciliation with the data presented here for Calandrinia and Montiopsis. Nevertheless, the existence of at least one foliar feature described here, the fimbrial vein in Cal- andrinia, would not appear to have been predicted by current models. The availability of detailed, phylogenetically analyzed systematic accounts of foliar morphology such as that presented here might be exploitable in advancing theories relating mature leaf form to function and development. LiTERATURE CITED AncHIE, J. W. 1989. Homoplasy excess ratios: new indices for measuring levels of homoplasy in phylo- genetic systematics and a critique of the consistency index. Syst. Zool. 38: 253-269. —. 1990. Homoplasy excess statistics and reten- tion indices: a reply to Farris. Syst. Zool. 39: 169- Be ars zur vergleichenden Anatomie BECKER, C. der n = Universitat 1895. Portulacaceen. Doc Erlangen, Erlangen, En BEHNKE, H.-D. 1984. Pla pus аш аы апа ultrastructure: general terminology, taxonomic a plications, ri aspects of trichome-bacteria interac- tion in leaf tips of Dioscorea. Pp. in E. Rodri- uez, P. L. Healy & I. Mehta (editora) Biology and Се of Plant Trichomes. Plenum Press, New Yor Ma R. C. 1987. A с of the id Portu- lacaceae. Ени J. Вог. 35: 383-4 PRU А. 1988. The OR oy, NM of Flowering Plants, 2nd edition. New York Botanical arden, New York. бйлу, D. L. 1974. Approaches to the identification of angiosperm leaf remains. Bot. Rev. (Lancaster) 40: 1-157. роте, J. A. & M. J. DONOGHUE. e Seed plant phylogeny and thes rigin of angiosperms: an exper- imental cladistic approach. Bot. Rev. ia 52: EHLERINGER, J. 84. Ecology and о of leaf nang in North ud rican desert plan 113-132 in E. Rodriguez, P. L. Healy & I. Мема (editors), Biology un: Chemistry of Plant Trichomes. Plenum Press, Farris, J. S. 1989. The retention index and the rescaled consistency index. Cladistics 5: 417-419. Кокр, D. I. 1992. Sy s tematics and Evolution of Mon- tiopsis Subgenus Montio a Washington d. St. Louis, Mis- Ce M. 1992. Die Gattung Anacampseros L. ortulacaceae). I. Untersuchungen zur Systematik. 564. . 1979. On the adaptive significance of leaf form. Pp. 375-407 in O. T. Solbrig, S. Jain, . Johnson & P. H. Raven (editors), Topics in Plant Population Biology. Columbia Univ. Press, New York. Некзнком1т2, M. А. 1990. Phylogenetic and Mor logical Studies in Portulacaceae. Ph.D гало University of California, Davis, California. 1991a. Leaf morphology of Cistanthe Spach (Portülaeacens Ann. Missouri Bot. Gard. 78: 1022- 1060. Phylogenetic assessment and revised circumscription of org ~ A: Ann. Missouri Bot. Gard. 78: 21, 1 Leaf о. Ms taxonomic anal- ysis of Cistanthe tweedyi (пее Lewisia tweedyi; a Syst. Bot. 17: 220-238. 1993. Revised circumscriptions and subge- ric taxonomies of Calandrinia and Montiopsis oni uar with notes on phylogeny of the por- tulacaceous alliance. Ann. Missouri Bot. Gard. 80: 365 & D. I. Кокр. 1993. Calandrinia carolinü: new name for Baitaria acaulis Ruiz & Pavon. Phy- tologia 74: 279-2 Hickey, L. J. 1979. x revised кезш dpt of the architecture of dicotyledonous leav . 25-39 in . R. Metcalfe & L. Chalk петину d а of the Dicotyledons, 2nd edition, Volume 1. Clarendon Press, Oxford. KELLEY, W. A. 1973. Pollen Morphology and Rela- tionships in the Genus Calandrinia H.B.K. (Portu- lacaceae). M.S. Thesis. California State University, Northridge. Ketsey, R. G., С. W. REYNOLDS & E. RODRIGUEZ. 1984. The chemistry of biologically active opo a se- n in plant glandular trichomes. Pp. n E. Rodriguez, P. L. Healy & I. Mehta ter Biology and e E of Plant Trichomes. Plenu Press, New LEVIN, G. A. . Systematic foliar morphology Phyllanthoideae (Euphorbiaceae). III. Cladistic ш. sis. Syst. Bot. 11: 515-530. 1990. Plant Development. Unwin Hy- an, London. MCNEILL, J. 1974. Synopsis of a ipe classification of Portulacaceae. Taxon 23: 7 MEINHARDT, 1982. Models of piu Pattern Formation. Academic Press, New York. MiTCHISON, C. J. 1980. A model for vein formation in higher pos qeu Roy. Soc. London, Ser. B, Biol. Sci. 207 09 NYANYANO, В. |: ae Tribal and Generic Relation- ship and Classification of the Portulacaceae (Centro- spermae). Ph.D. Dissertation. University of Reading, enr ак —. Taxonomic significance of the sto- ap complex i in the Portulacaceae. Feddes Repert. 97: E Tribal and generic relationships in the Portulacaceae (Centrospermae). Feddes Repert. 101: -24]. cann 1934. Portulacaceae. Pr andrinia (Portulacaceae) de los Andes Mendocinos. Bol. Soc. Argent. Bot. 25: 511-537 REICHE, К. 1898. Flora de Chile, Volume 2. University of Chile, Santiago. 396 Annals of the Missouri Botanical Garden P. A. 1932. Portulacaceae. Pp. 279-336 . Britton et al. (editors), North American Flora, Volum me 21. New York Botanical Garden, New Yor Sacus, T. 1989. The development of di Maine during leaf eh Curr. Topics Pl. Biochem. Physiol. 8: 168-183. Pattern. Formation in Plant Tissues. Cambridge Univ. par Cambridge, England. SANDERSON, M. J. & M. J. DoNoGHUE. 1989. Patterns variation in levels of homoplasy. Evolution 43: 1781-1795. es ER, R. 88. Notulae ad 2m paraquaiensem, 17. Candollea 43: 409-4 dn SworronD, D. L. 1991. PAUP: Phylogenetic Analysis Using Parsimony, Version 3.0[r]. Computer program distributed by the Illinois Natural History Survey, Champaign, Illinois TAKHTAJAN, А. 1991. Evolutionary Trends in Plants. Columbia Univ. Press, New York. Witkinson, Н. P. 1979. The plant surface (mainly leaf). Pp. 97-165 in C. R. Metcalf L. Chalk ‘ara нн of the Dicotyledons, 2nd edition, Volume 1. Clarendon Press, Oxford. A REVISION OF CHIONOLAENA (COMPOSITAE, GNAPHALIEAE) Susana E. Freire? ABSTRACT The neotropical genus Chionolaena is accepted as including the species of Leucopholis, which has been considered a. Chion chrysocoma, and key to all taxa are presented. A cladistic analys Guatemala. Seventeen species of Chionolaena are rec bi hi C. seemannii. Descriptions, typification, synonymy, tribution maps, illustrations, indice s perfor sis of Chio ered capitula. Chionolaena consists of small, more or less bia, = x n Brazil, with disjunct n Venezuela). The most closely iur ognized, one o Chionolaena у ар c and a med using morphological л өлө new combinations are made: nolaena was Polarity of characters was based on outgroup comparison ith the кн isse diras анин: Two monophyletic din groups were resolved: 1) the “concinna species grou m „ ех ellae, С. jeffreyi, € phylicoides, C. pore and кг wittigiana), Within each group, an taxa, i.e., a closer historical relationship is indicated between the Colombian and Brazilian areas than between bus and the Mexican area sister group of the South Amer (including C. "| C. arbus . latifolia, C. lavandulifolia, С. lye (including C. concinna and Sore and (2) the cula, C. capitata, C. cc h this resolution place Chionolaena DC. (including Leucopholis) is a genus of small shrubs found in high elevations within the neotropical region (Cabrera & Willink, 1973) from Mexico to southern Brazil. laena comprises 17 species characterized by di- Chiono- chotomously branched stems, reflexed leaves with revolute margins, central florets commonly func- tionally staminate, and pappus bristles with clavate cells at their tips. It is closely related to the Mexican genus Gnaphaliothamnus Kirp. The last complete taxonomic treatment of this J. G. Baker (1882), who recognized nine species. Seven species were published by var- genus was by ious authors under Chionolaena since the appear- ance of Martius's Flora Brasiliensis. In addition to problems with delimitation of species of Chiono- laena, the generic limits became unclear. The aims of this revision are to clarify generic and specific limits, to interpret species interrela- tionships using cladistic methods, and to compare the results of cladistic analysis with the geographic distribution of the genus. MATERIALS AND METHODS This study was based on specimens from the following herbaria: BM, C, COL, E, С, GH, HBG, „БР. My КРЕЗ S. U; W (see List of Exsiccate at end of article). and vegetative parts were dissected and observed after boiling in water. Cross sections of leaves and made by free hand and stained with 2% safranin. All drawings were made by the author stems were from herbarium material. Information on flowering periods, habitat, and color of corollas was taken from herbarium labels. Species descriptions are presented alphabetically. The applied cladistics methods were summarized in Nelson & Platnick (1981) and Wiley (1981). Identification of apomorphic characters was based on the outgroup comparison (Platnick, 1979; Wa- trous & Wheeler, 1981) with Gnaphaliothamnus as the outgroup. Table 1 shows the characters and character states used in the cladistic analysis. Five characters have more than two states (4, 6, 8, 9, 10). These characters were treated as nonadditive. ! I thank Arne Anderberg, Angel L. Cabrera, Jorge V. description of Chionolaena mexicana who kindly loaned ges The present study was completed at the Swedish Museum of Natural Hi: the Consejo Nacional de uper кашне Científicas y Técnicas de Argentina. I am grateful t "of these institutions for support of m Crisci, Juan J. Morrone, J. Strother, and two anonymous I also thank Angel L. Cabre epartamento Cientifico de Plantas Vasculares, Museo Fs La Plata, 1900 La Plata, Argentina. ANN. Missouni Bor. Garb. 80: 397-438. 1993. 398 Annals of the Missouri Botanical Garden TABLE 1. Characters and character states. Characters Character states l. Adaxial leaf surface tomentose (0); glabrous (1) 2. Leaf ma slightly revolute (0); conspicuously revolute (1) 3. Stem "У dichotomous (0); subdichotomous (1) 4. Type of саришезсепсе — corymbs (0); umbels (a); solitary (b) 5. Corymbophore absent (0); present (1) 6. Type of capitula аи 7. Inner bracts obtuse (0); acute (1) 8. Involucre shape obdelta ate (0% ро oblong (1 9. Relation of female to staminate florets in slightly more numerous than fur | staminate florets, 1.1:1-2. heterogamous (0); homogamous (with perfect florets) (a); nearly unisexual (b); ho- (with unisexual florets) (c) 4 mm wide) (a); broadly campanulate (12 mm pistillate florets outnumbering the functionally staminate, 5:1 (0); pistillate florets 5.1 (a); pistil- heterogamous capit- late florets slightly more numerous than functionally staminate or functionally sta ula minate florets slightly more numerous than pistillate (bx СОХ staminate florets slightly more numerous than or equaling pistillate, 1:1-2:1 10. Achene pubescence shortly pa (0); glabrous (a); villous (b) 11. Achenial hairs 12. Pappus bristles free 13. Apical cells of pappus linear (0); clavate (1) bristles obtuse at the tip (0); acute at the tip (1) (0); fused at base into a ring (1) In the data matrix (Table 2), plesiomorphic char- acters were coded with 0, apomorphic ones with 1 (multistate characters with a, b, and c), and unknown characters with question marks. The data matrix was analyzed using the phylogenetic pro- А . Data matrix. The character numbers cor respond to the numbers used in the text and in the clado- grams. Question marks indicate missing information. Character 1111 Taxon 123456789 0123 Gnaphaliothamnus 000000000 0000 С: idiocephala 000a0b100 0011 C. arbuscula 100b0blac blll C. capitata 000a0blac b011 C. columbiana 000a1000a а?11 C. concinna 00100b00b 0011 C. chrysocoma 110a001aa а?11 C. eleagnoides 00000000c a?11 C. isabellae 00001a00? blll C. jeffreyi 000b0010c b011 C. latifolia 100a000ac b011 C. lavandulifolia 000a0b10b a?11 C. lychnophorioides 100b001ba blll C. mexicana 00000000a 0011 C. phylicoides 110a0cla? b011 C. запоги 00000010a 0011 C. зеетаппи 1010000b0 0011 C. wittigiana 10000010c blll gram ipis d version 1.5 (Farris, 1988). The options used were “mhennig,”” for constructing multiple вы followed by “bb” (branch breaker). Furthermore, the successive weighting procedure was applied, which calculates weights from the best fits to the most parsimonious trees, using rescaled consistencies (rc), which are the products of the character consistency (c) and the character retention index (r). The product is scaled to lie in the range 0-10. The weighting procedure is repeated on successively produced trees until the trees no longer change (Farris, 1989) HISTORICAL SURVEY Chionolaena was first described by de Candolle (1836) on the basis of the Brazilian species C. arbuscula from a Vauthier collection. The genus was characterized by its solitary, heterogamous, many-flowered heads in which the central florets were functionally staminate. In 1843 Gardner created the genus Leucopholis on the basis of the Brazilian species L. phylicoides with few-flowered, congested, homogamous capit- la In 1882 Baker placed Leucopholis in a section of Chionolaena, which is characterized by having capitula with 8-10 florets, in contrast to section Euchionolaena with its many-flowered capitula. Six species were listed under section Fuchiono- laena: C. arbuscula, C. lychnophorioides, C. gla- Volume 80, Number 2 1993 Freire 399 Revision of Chionolaena ziovii, C. glomerata, C. isabellae, and C. witti- giana, of which the last four were new. The section Leucopholis included three species: C. latifolia, C. phylicoides, and C. longifolia, of which the last was new. Since 1882, seven species have been added to Chionolaena. The first two, C. eleagnoides and C. sartorii, were described by Klatt (1887: 88) based on Liebman collections from Mexico. The next year, Wawra (1888, 2: 30) described C. innovans (presently placed in synonymy of C. is- abellae). In 1893, B. D. Jackson transferred the Mexican species Elichrysum lavandulaefolium Kunth to Chionolaena. Blake (1935: 312) de- scribed C. columbiana from Mexico. It was ap- proximately 50 years before the next species of Chionolaena, C. jeffreyi H. Robinson (1984: 121) from Brazil and C. breweri (Steyermark (1984: 26) (presently placed in synonymy of C. latifolia) from Venezuela, were described. Cabrera (1961) maintained the generic status for Leucopholis. He noted that members of this genus are distinguishable by their few-flowered, homogamous heads. In the present study, Leucopholis is not rec- ognized. The species of Chionolaena show varia- tion in the number of the flowers and in the pro- portion of pistillate to functionally staminate florets (including homogamous heads). It is evident that distinction between these genera is impossible based on sexuality alone. Dillon & Sagástegui (1990) proposed Pseudoli- gandra for the Colombian species Oligandra chrysocoma W edd. According to the cladistic anal- ysis, this new genus cannot be accepted, i.e., Oli- gandra chrysocoma is a derived species with its closest relatives within Chionolaena. If Pseudoli- gandra is treated as a separate genus, Chiono- laena becomes paraphyletic. DELIMITATION AND GENERIC RELATIONSHIPS Chionolaena belongs to Gnaphalieae (sensu An- derberg, 1989; Bremer, 1987) of the Compositae. Merxmüller et al. (1977) considered Chionolaena to belong within the “Lucilia group” (1.е., Lucilia, Belloa (= Lucilia, Freire, 1986), Berroa, Chev- reulia, Facelis, Gnaphaliothamnus, Leucopholis, Luciliopsis, Mniodes, and Raouliopsis) princi- pally on the basis of its style branches with dorsal sweeping hairs. Chionolaena is distinguished from other genera of Gnaphalieae by its dichotomously branched stems, deflexed leaves (at maturity) with revolute margins, absence of stem phloem fibers, undivided stereome (i.e., lower part uniformly thickened) and white opaque lamina of the inner involucral bracts, often purple upper corollas, functionally staminate her- maphroditic florets, and basally fused pappus bris- tles with clavate apical cells. Leucopholis consis- tently differs from Chionolaena only in having few-flowered capitula. The genus Leucopholis is therefore not recognized (see Historical Survey). Preliminary results from cladistic analysis of Gnaphalieae indicate that Gnaphaliothamnus is the sister group of Chionolaena (Anderberg, 1989; Anderberg & Freire, 1989). It differs from Chiono- laena in having free pappus bristles with linear apical cells. MORPHOLOGY HABIT The stems are dichotomously branched, i.e., at the apex of each stem there is a terminal aborted primordium or, eventually, a terminal capitulesc- ence. When the capitulescence matures, axillary buds develop, producing a dichotomous growth pat- tern. In а few species (C. concinna and C. see- mannii), however, only a single bud develops at the base of the capitulescence, and the resulting growth is subdichotomous, leaving the capitules- cence conspicuously lateral. TRICHOMES Throughout Chionolaena two types of tri- chomes occur together on the leaves and stems: nonglandular uniseriate hairs (Fig. 1D) and glan- dular hairs (Fig. 1C). The former were described as "aseptate flagellate hairs" by Ramayya (1962) and consist of 1—3 small, basal cells and a long, basally swollen terminal cell. Clandular hairs con- sist of a unicellular basal foot and a uniseriate or biseriate row of cells. Only glandular hairs were found on the abaxial surfaces of corolla apices. CAPITULESCENCE The arrangement of the capitula in Chionolaena varies (Fig. 1). Heads may be solitary and terminal or densely clustered in umbelliform cymes (i.e., "umbels"). Sometimes these umbelliform clusters are arranged in “corymbs.” In most species of Chionolaena the upper flow- ering branches are leafy. In some species, e.g., C. columbiana and C. isabellae (Fig. la, b), the capitulescence is supported by a leafless slender stem, i.e., a peduncle, termed **corymbophore" by Lundgren (1972) for Anaxeton. Missouri Botanical Garden Annals of the 400 — I ОЗтт ] EEG Oimm Ll o ———D ate mom CAPITULESCENCE o о а EL ) ) ) С JJ). E > = © : is : Volume 80, Number 2 993 Freire Revision of Chionolaena CAPITULA The capitula are either heterogamous with func- tionally staminate florets and pistillate florets, ho- mogamous with perfect florets, or homogamous- unisexual with functionally pistillate or staminate hermaphroditic florets. The heterogamous condition. predominates in Chionolaena. Only two species have homogamous capitula: C. isabellae, with all perfect florets, and C. phylicoides, with unisexual capitula. Four species of Chionolaena are subdioecious: C. arbuscula, C. concinna, and C. lavandulifolia have nearly unisexual capitula in different plants; Chionolaena aecidiocephala has two types of ca- pitula, homogamous capitula with all flowers func- tionally male and heterogamous capitula with pis- tillate, female flowers as well as male central flowers. FLORETS Most species have functionally staminate central florets and filiform, pistillate, peripheral florets. Only one species, C. aecidiocephala, shows reduction in the fertility of the staminate florets, i.e., anthers without pollen (neuter florets). All species have fertile pistillate florets. The proportion of pistillate to functionally sta- minate florets differs slightly or sometimes mark- edly. In six species (C. arbuscula, C. capitata, C. eleagnoides, C. jeffreyi, C. latifolia, and C. wit- tigiana) the functionally staminate florets com- monly equal or slightly exceed the pistillate in num- ber (ca. 1 : 1-2: 1). In other species (C. chrysocoma, C. columbiana, C. concinna, C. lavandulifolia, C. lychnophorioides, C. mexicana, and C. sar- torii) the pistillate florets are slightly more nu- merous than the functionally staminate florets (ca. Li: .5:1). In only two species, C. aecidio- n and C. seemannii, the female florets out- number the functionally staminate florets (ca. 5 : 1). ACHENES In Chionolaena only the pistillate florets are capable of producing achenes, except in C. isa- bellae, which has perfect florets. The sterile ovaries of the functionally staminate florets are usually flattened, without an embryo, and sometimes con- tracted toward the middle or strongly reduced. Differences between the fertile achene and the ster- ile ovary are further addressed in the systematic treatment below. The achenes are usually oblong or oblong ellip- tic. Within the subdioecious species, only C. ae- cidiocephala has slightly dimorphic achenes, i.e., the central-floret ovary of heterogamous capitula is longer than the central-floret ovary from ho- mogamous capitula. Within the genus, the achenes may be pubescent or glabrous. Chionolaena columbiana, C. eleag- noides, C. chrysocoma, and, usually, C. lavan- dulifolia have glabrous achenes. The hairs are always 3-cellular, with two parallel elongated cells and a shorter, Itera. basal cell with a thickened ir" (Hess, 1938) is commonly present in many pl. genera of the outer wall. is "duplex h Compositae. In contrast with all other Chionolaena species having pubescent achenes, C. arbuscula, C. isabellae, C. lychnophorioides, and C. witti- giana have achenial hairs acute at the tip (Fig. 1F). In the remaining specics these hairs are ob- use. Furthermore, there is variation in achenial hair length; the four species listed above and C. phylicoides, C. latifolia, C. jeffreyi, and C. cap- itata have sericeous, villous achenes with hairs ca. 0.3 mm long, whereas C. aecidiocephala, C. la- vandulifolia (when hairy), C. and C. sartorti have hairs ca. 0.1 mm long (Fig. 1G). seemannii, CHROMOSOME NUMBER Gametic chromosome number has been inves- C. la- tigated in one species of Chionolaena, i.e., — FIGURE 1. Chionolaena. — А and differentiated а ари (end). — the abaxial face. C, D. С. apical cell hsc at the as sE, C. isabellae, elongated tricellular hair with acu pappus bristles fused t with sweeping hairs on 6001 (S); D, Cabrera 25329 (LP); E, Hemmendorff 64 umbel (C. aecidiocephala, C. capitata, C. C. concinna, C. eleagnoides, C. mexicana, C. sartorit, corymbophore (a, C. columbiana; b, C. isabellae). urbuscula, cross section of leaf w abellae, = hara. — C. Glandular hairs biseriate беди С. А abaxial т with cyclocytic stomata. Е, together near the base dct a Pra ring below.— I. C. e clavate apica al cells.—J. C. wittigiana, style branches with two p stigmatic areas o G, H, Mac КЕ 412 Diagrams of capitulescence types in С m ae arbuscula, C. jeffreyi chrysocoma, C. i, C. seemannii, C. witt C. aec Pd 'ephala, cross section of stem pad continuous кро xylem (X2) s densely lanate on “р. ун ШЫЛА зы with G. Achene hai F. t, clavate hair with obtuse ар. —H. C. о eagnoides, pappus bristles with y F, s 940 (LP); 1, Liebman 316 (GH); J, Dusén atifolia, C. lavandulifolium, hylicoi igiana); capitulescence Meum rted by Annals of the Missouri Botanical Garden vandulifolia (= Gnaphalium lavandulifolium), which was found to have n — y De Jong & Longpre (1963). This number suggests tetraploidy, based on x = 7 in Gnaphaliinae (Merxmiller et al., 1977). CLADISTICS Chionolaena and its sister group, Gnaphalio- thamnus, form a larger monophyletic group that is characterized by having dichotomously branched stems without phloem fibers, reflexed leaves (at maturity) with revolute margins, an undivided ster- eome in the inner bracts, and commonly function- ally staminate central florets. Chionolaena is mor- phologically distinct because its pappus bristles are fused at the base into a ring and have clavate cells at their tips. Thirteen apomorphic character states were found (Table 1) by outgroup comparison (Hennig, 1966; Platnick, 1979; Watrous & Wheeler, 1981) using the monotypic genus Gnaphaliothamnus as out- group. Thirty-five equally parsimonious clado- grams were generated from the data matrix. All have 39 steps, a consistency index of 0.51, and a retention index of 0.62. If the synapomorphies for the whole genus (characters 12 an ) are ex- cluded, the consistency index is 0.48. When the successive weighting procedure was applied, 11 minimum length trees resulted after the second round of weighting, with length 118, a consistency index of 0.68, and a retention index of 0.75. Again, if the synapomorphies for the whole genus are ignored, the consistency index is 0. The strict consensus tree of the 11 dessus (Fig. 2) shows that there are three monophyletic groups: (1) the * ing C. concinna and C. seemannii); (2) the "eleag- concinna species group" (includ- noides species group" (including C. C. columbiana, and C. isabellae) and (3) the * eleagnoides, sar- C. la- Jeffreyi, С. capitata, C. wittigiana, C. lychnophorioides, C. latifolia, € phylicoides). Chionolaena mexicana appears to be isolated. torii species group" (including C. sartorii, vandulifolia, C. aecidiocephala, C. arbuscula, C. . chrysocoma, and C. s an example of how characters may be dis- tributed, 1 have presented one of the 11 equally parsimonious cladograms (Fig. 3). The cladogram chosen shows two monophyletic groups: (1) the “concinna species group" (including C. concinna and C. seemannii) and (2) the “mexicana species group" (including C. mexicana, C. C. columbiana, C. eleagnoides, isabellae C. sartorii, C. la- vandulifolia, C. aecidiocephala, C. jeffreyi, C. capitata, C. latifolia, C. chrysocoma, C. phyli- coides, C. wittigiana, C. arbuscula, and C. lych- nophorioides). The Mexican species C. concinna and C. see- mannii have stems subdichotomously branched (character 3). Fifteen species comprise the *‘mex- icana species group," which is defined by the slightly more numerous pistillate florets than functionally staminate florets (character 9а). groupings appear Mun this last group: the ' (including C. eleagnoides, isabellae””) and the “sar- C. la- aecidiocephala, C. jeffreyi, С noides species group' C. columbiana, and C. torii species group" (including C. sartorii, vandulifolia, C. capitata, C. latifolia, C. chrysocoma, C. phyli- coides, C. wittigiana, C. arbuscula, and C. (усћ- from nophorioides). Chionolaena eleagnoides Mexico, C. columbiana from Colombia, and C. isabellae from Brazil have glabrous achenes (char- acter 10a). E abellae appear together united by capitula sup- Chionolaena columbiana and C. is- ported by a corymbophore (character 5). The second large monophyletic group, **sartorii species group," is justified by the acute inner bracts (character 7). Chionolaena sartorii is the sister species of the other “sartori group” taxa. The latter species are united by umbelliform clusters of capitula (char- acter 4a). There are two monophyletic groups with- in the last group. The Mexican species C. lavan- dulifolia and C. aecidiocephala share capitula nearly unisexual (character 6b). The second mono- phyletic group is formed by the mostly Brazilian species C. jeffreyi, C. capitata, C. latifolia, C chrysocoma, C. phylicoides C. wittigiana, C. ar- buscula, and С. lychnophorioides. They are unit- ed by villous achenes (character 1 Ob) and the func- tionally staminate florets slightly more numerous than or equaling pistillate florets (character 9c). The relationships among C. capitata, C. jeffreyi, and C. latifolia are unclear. Chionolaena arbus- cula and C. lychnophorioides appear together united by the solitary capitula (character 4b). In some solutions, C. arbuscula and C. wittigiana appear together, whereas in other solutions C. ar- buscula is placed as the sister species of C. lych- nophorioides and C. w ittigiana. Chionolaena wit- tigiana, C. arbuscula, and C. lychnophoriodes have achenial hairs acute at ihe tip (character 11). In all cases C. chrysocoma and C. phylicoides appear together, sharing conspicuously revolute leaf margins (character 2). The sections distinguished by Baker (1882), sec- tion Euchionolaena (C. arbuscula, C. isabellae, Volume 80, Number 2 Freire Revision of C tionolaena 403 Gnaphaliothamnus mexicana concinna E seemannii eleagnoides columbiana isabellae sartorii lavandulifolia aecidiocephala jeffreyi capitata FIGURE 2 trees. C. lychnophorioides, C. wittigiana) and section Leucopholis (C. phylicoides, C. latifolia) do not represent monophyletic groups. The character by which these sections have been separated (number of florets per capitulum) is variable intragenerically and also sometimes infraspecifically. BIOGEOGRAPHY The total range of Chionolaena (Fig. 4A) com- prises the montane habitats of America within the neotropical region (Caribbean, Amazonian, and haqueno dominions) as distinguished by Cabrera & Willink (1973). The strict consensus tree of relationships among species of ( hionolaena for all equally parsimonious wittigiana lychnophorioides EN arbuscula latifolia chrysochoma Lf phylicoides Chionolaena has a discontinuous distribution; it is present in central Mexico, Central America (Cos- ta Rica), northern Colombia, southern Brazil, and Plant species with similar ranges are frequent, e.g., Agar- ista (Judd, 1984), Heliconia (Andersson, 1985), Maranta (Andersson, 1986), and ferns (Тгуоп, 1962, 1970; Conant, 1983). tionships among the highlands of Mexico and Ama- northern Amazonia (southern Venezuela). The floristic rela- zonas and the Andean region have been expressed by Hemsley (1879-1880, vol. 4: 228) and Epling (1940) in Rzedowski (1965). The monotypic Gnaphaliothamnus, the sister nm өө өө Missouri Botanical Garden Xx Annals of the EI E: GUAJIRA AMAZONIAN CERRADO PARANENSE ATLANTICA CAATINGA |MESOAMER L |СНАОЏЕКО | El САРШВВЕАМ DE MONTANA 404 за ргогЈОоцдОШ ЗА | ч 2 © eynasnqie H= ч еџе!Б дим Ф sepiooiAud El zy сч о ewososAiyd ~ е!ј0]!је| —5 о eye yides 5+ 5, tAaijjal — P o 2 ejeudeooipiose 2 pt ra eijojiynpueaAe| H 11031es зејјадез! eu eiquin|oo 10a = гер!оџбеаје еџеоз!х•еш m9a P пџџешог 5 euui2uoo GNAPHA LIOTHAMNUS Volume 80, Number 2 Freire 405 1993 Revision of Chionolaena group of Chionolaena, has a Mexican and Central In the “eleagnoides species group," two prin- American (Guatemala) distribution. e 17 species of Chionolaena are distributed k (1973): seven species (C. aecidiocephala, C. concinna, C. eleagnoides, C. lavandulifolia, C. sartorii, and C. seemannii) are narrowly endemic as follows. according to Cabrera & Willin mexicana, C. to the mountainous areas of central Mexico (most being known from a single locality) in the “Me- soamerican Mountain” Province within the Carib- bean Dominion. Chionolaena lavandulifolia has disjunct populations in central Costa Rica as well. Two species (C. columbiana and C. chrysocoma) are confined to the northern Andes of Colombia, on the Sierra de Santa Marta and the Sierra de Perijá, in the “Guajira” Province within the Ca- ribbean Dominion. The remaining species are en- demic to the mountains of southern Brazil in the Chaqueño Dominion (**Caatinga" Province: C. jef- freyi) and Amazonian Dominion (*Amazonian" Province: C. latifolia; “Cerrado” lychnophorioides; Province: C. ar- buscula and C. “Paranense” Province: C. isabellae, C. capitata, C. diei and C. arbuscula; “Atlantica” Provi itata, C. latifolia, C. phyllicoiodes). Only C. lati- folia has a disjunct distribution, being also recorded from Cerro Marahuaca in Venezuelan Guayana. In order to formulate biogeographic hypotheses for the central Mexican, Colombian Andean, and vince: C. cap- southern Brazilian areas in the neotropical region, cladistic analysis of Chionolaena was used. Addi- tional cladistic analyses of other taxa in this region will allow testing the following results. In the area cladograms (Fig. 4B, C) based on the consensus tree (Fig. 2), the name of each taxon was replaced by an abbreviation of the area in which it occurs. The first events leading to the separation of the ancestral population of Chiono- laena are unclear. The three monophyletic groups, roup," the "sartorii species group,” the “concinna species "eleagnoides species group," and the identified in the consensus tree (Fig. 2) can, how- ever, be used to analyze the relationships between the geographical areas of each taxon. Two of these Е ' and the are today sympatric and groups, the “eleagnoides species group” "sartorii species group" present similar vicariant patterns, which can be presumed to have developed simultaneously. cipal vicariant events can be postulated. The first event could have split a large and continuous pop- ulation into northern (evolving into a Mexican tax- on, i.e., C. eleagnoides) and southern (evolving into South American taxa) populations. The second event could have separated a small population in the Colombian area (evolving into C. columbiana) from the more southern population, which evolved into the Brazilian taxon (C. isabellae) (Fig. 4D). In the “sartorii species group," a Mexican pop- ulation (evolving into C. sartorii, C. lavandulifo- lia, and C. aecidiocephala) may have first split off from the main population. Subsequently, a large Brazilian population (evolving into C. jeffreyi, C. capitata, C. latifolia, C. wittigiana, C. arbuscula, and C. lychnophorioides) could have become iso- lated from the rest of the main population. The remaining population may have then extended its range toward the southeast and split off another Brazilian population (evolving into C. phylicoides) leaving behind a Colombian population (evolving into C. chrysocoma o summarize, although the sequential events involved in this group are unclear because the phylogeny is unresolved, it is clear that a closer biogeographical relationship exists between the Co- lombian Andean and southern Brazilian areas than between either and the central Mexican area. SYSTEMATIC TREATMENT CHIONOLAENA Chionolaena DC., Prodr. 5: 397. Chionolaena arbuscula DC. 1836. TYPE: m Gardner, London J. Bot. 2: 10. 1843. TYPE: copholis о. м dne P seudoligandra Dillon & & "Taxon 39: 125. 1990. TYPE шы шм је socoma (Wedd.) Dillon & Sagast. Shrubs or subshrubs. Stems erect or ascendent, dichotomously or subdichotomously branched, woolly tomentose and leafy when young but later glabrous and leafless with marked leaf scars in the lower parts. Vascular cylinder without phloem fi- bers, surrounded by a distinct endodermis (Fig. 1A). Leaves sessile, closely set or spreading, sub- <— FIGURE 3. spon rphies, “X” indicates Sis sha s listed in Table 1. provinces as outlined in Cabrera 8 Willink (1973). Numbers refer to One of the 16 equally parsimonious cladograrís, Huang m ter distribution. Black bars indicate м 8 e, “=” indicates parallelism. raphic distributions are summarized by Vivien heb regions, dominions, and Annals of the Missouri Botanical Garden Tropic: of Capricorn dioe x < a ж >< < о c ш ul w O c w с = 20m z a om | Sr | ЕЕС B E Y MEX MEX MEX COL BRA MEX BRA COL BRA | МЕ " cor Volume 80, Number 2 1993 Freire 407 Revision of Chionolaena decussate, in four distinct rows; erect in the upper part of the stem, becoming reflexed with age (only one species has appressed leaves); midrib promi- nt; margins entire, slightly or conspicuously rev- d (Fig. 1B); mucronate; adaxially glabrous or tomentose, abaxially densely lanate and whitish, with glandular and nonglandular hairs (Fig. 1C, D). Epidermis with cyclocitic stomata surrounded by 4-6 subsidiary cells scattered on the abaxial surface (Fig. 1 E); cuticle thick. Capitula shortly pedunculate in dense umbels or corymbose capitu- lescences, sometimes solitary, terminal on the main stems, sometimes supported by a slender peduncle (Fig. 1). Heterogamous or homogamous. /nvolu- cres narrowly oblong, campanulate or broadly cam- panulate; bracts imbricate, membranaceous, gen- erally yellowish brown, entire; outer ones broadly ovate to linear-ovate, acute, obtuse or acuminate at the apex, pubescent adaxially; innermost bracts narrow, linear or obovate, obtuse or acute at the apex, glabrous with white opaque lamina and lower part uniformly thickened (i.e., stereome undivided). Receptacle glabrous, flat or somewhat convex, epaleate. Florets 5 to ca. 100; corollas white-cream, often red-tipped; pistillate florets (when present) fewer than, equaling, or more numerous than the hermaphroditic ones. Pistillate florets fertile; co- rollas filiform, minutely bilabiate, lobes with biseri- ate glandular hairs. Styles bifid; style branches erect to spreading, linear, glabrous, rounded; sty- lopodium indistinct. Achenes narrowly oblong or narrowly elliptic, glabrous or pubescent with long or short duplex hairs (Fig. 1F, G). Pappus bristles as in hermaphroditic florets but with apical cells slightly inflated, not clavate. Hermaphroditic flo- KEY TO THE SPECIES OF CHIONOLAENA rets functionally staminate (occasionally function- ally pistillate, neuter, or fully fertile); corollas tu- bular, slightly expanded at top, 5-lobed, the lobes papillose with biseriate glandular hairs. Styles shortly bifid or sometimes undivided; the branches rounded truncate or acute; with apically obtuse sweeping hairs, commonly the sweeping hairs occurring al- most to the stylar partition; abaxial faces with two parallel stigmatic areas constituted by small and rounded papillae (Fig. 1J); stylopodium and sty- appendages, thickenings in end theropodium (collar) distinct, tails slightly shorter than, equaling, or longer than the antheropodium. Ovaries flattened, often strongly contracted with embryo not developed. Pappus bristles + thick, barbellate with clavate (occasionally slightly inflat- ed) apical cells (Fig. 11); fused at the base into a single ring and connate in groups of different lengths (Fig. 1H); equaling the corolla. The name of the genus seems to refer to the snow white and cottony tomentum on the lower surface of the leaves (from Greek, “chionos” = snow, “laenos”” = cloak or mantle). Distribution. Seven species from the central region of Mexico; two species confined to the north- ern area of Colombia; and eight species in southern Brazil. Local name. The name “Arnica” is applied to some Brazilian species (Glaziou, 1910), viz, C. arbuscula var. hololeuca (— C. capitata), C. in- novans (— C. isabellae), C. wittigiana, and glomerata (— C. latifolia). la. Capitula solitary (occasionally arranged in terminal, small, loose umbels). 2a. S a with as obtus scending stems; leaves with lamina conspicuously narrowed at the base; dii = C. jeffreyi 2b. Shrubs « erect, rigid, woody; leaves not notably narrowed at the base; achene hairs acute at ds apex. 3a. Leaves narrowly linear, 0.7-1 mm m wide, margins conspicuously revolute ooo... . arbuscula ЗЬ. Leaves broadly linear, 1.2-3 mm wide, margins slightly revolute ................... I2. G. lychnophorivides lb. Capitula arranged in clusters or corymbs. Capitula arranged in dense umbels. mbels supported by a peduncle at maturity 5b. Umbels sessile at maturity. 6a. Involucres narrowly oblon 7 . 4. C. columbiana g a. Leaves with lower lamina completely concealed by the revolute margins. E 4. — A. Distribution of Chionolaena. B,C. Area cladograms. — B. Complete area cladogram with the same ms as the consensus tree in Figure 2. (Nam C. Reduced area cladogram. Groups of taxa occurrin g in the sam cladogram.— D. Postulated aah we of the ancestral population after the two principal vicariant events. (MEX = olombian area, BRA = Mexican area, COL = C Brazilian area. 408 Annals of the Missouri Botanical Garden 8a. Leaves reflexed; achenes pubescent; capitula homogamous (unisexual) 8b. Leaves closely appressed to stem; achenes glabrous; capitula heterogamous 6. C. phylicoides C. c hrysoc oma - di Leaves id 5- 6b. ос obdeltat 10a. Erect, rigid sedis leaves suborbic 10b. Sree shrub; leaves linear-elliptic 11. 4b. | arranged in . Leaves broadly presen 11-12 chenes sericeous; achen ne mm wide; achenes glabrous re a Leaves obovate, linear, or elliptic (when obovate 3-7 mm wide); achenes pubescent. airs acute or subac 3a. Leaves narrowly obovate; corymbs lax, ned by a peduncle at maturity Leaves with lower surface visible (leaf margins ed revolute). 9a. Leaves elliptic, 1.5-4 cm long, adaxial fac 13 mm long, adaxial face tomentose 10. C. latifolia . C. capitata ular 1. C. aecidiocephala v lavandulifolia C. eleagnoides 8. C. isabellae 13b. Leaves linear; corymbs dense, sessile 17. C. wittigiana 12b. Achenes with short obtuse hairs. l4a. n linear to elliptic. ое weakly woody, caespitose; leaves narrowly linear, 1-2 cm long, 1 ide; corymbs of 3-4 capitula 15. C. sartorii 15b. Shrub woody, rigid, ascendent; leaves linear-elliptic, 3.5-4.5 cm long, 4-5 mm wide; corymbs of many capitula (up to 10)... 16. C. seemannii 14b. Leaves obovate. 16a. leaves slightly Vini er branched, rigid shrub; capitulescence terminal at maturity; ttenuate at the b ase C. mexicana 16b. Subdichotomously branched subshrub; capitulescence lateral at maturity; leaves olat 5. € subpeti 1. Chionolaena aecidiocephala (Grierson) Anderberg & Freire, Notes Roy. Bot. Gard. Edinburgh 46(1): : aecidiocephala Grierson, | ard. Edinburgh 31: 389, fig. 1972. Gna- phalium aecidiocephalum (Grierson) L. О. Williams., Phytologia 25: 459. 1973. TYPE: Mexico. Oaxaca: Ixtlán, Comaltepec, Cerro de Humo Chico, in grass in sun, 3,050 m, 2 Mar. 1968, MacDougall 4129 (holotype, E; isotype, 5). Figure 5. Erect, multibranched, rigid shrub, 10-20 cm high. Leaves suborbicular, obtuse, 2.5-5 mm long, mm wide with slightly revolute margins, tomentose adaxially, densely whitish lanate abax- ау. Capitula homogamous and heterogamous, arranged in dense umbels sessile and terminal on the upper branches. /nvolucre obdeltate. Involu- cral bracts ca. 32; outer bracts ovate, obtuse, 3- 4 mm long, 2-2.5 mm wide, purple at the tip; inner bracts linear, slightly obtuse, 6-8 mm long, 0.8-1.2 mm wide, purplish above stereome. Flo- rets 29-35, pistillate florets more numerous than the functionally staminate or all functionally sta- 2 concinna minate in homogamous capitula. Pistillate florets 0-)25-26; corollas 4 mm long; achenes elliptic, 1-1.5 mm long, 0.25-0.40 mm wide, pubescent with short, apically obtuse, duplex hairs. Func- tionally staminate florets 5 (and functionally neu- ter in heterogamous capitula), 29-35 (in homog- amous capitula); corollas 3.5-4 mm long; style branches rounded, 0.2 mm long; fertile anthers 2 mm long with tails slightly shorter than anthero- podium; sterile anthers 1.5 mm long, tails slightly exceeding antheropodium and without pollen grains; ovaries sterile, flattened, 0.5-0.8 mm long, 0.25- 0.30 mm wide, shortly pubescent as in pistillate florets. ~ Distribution. Known only from the type lo- cality, i.e., mountains of Oaxaca in Mexico, grow- ing [am 2,700 and 3,050 m. Flowering col- lection was made in March. Additional specimen examined. Ixtlan, Comaltepec 10,000 ft., MEXICO. OAXACA: , in grass in sun, Cerro de Humo Chico, T. MacDougall, 6 Jan. 1961 (E). Chionolaena aecidiocephala is distinguished from other Mexican species of the genus by its FIGURE 5. Chionolaena aecidiocephala leaf adaxial surface ud leaf abaxial surface (right) staminate floret. — J. Style of functionally staminate "iita pr E Pied berg & Freire . — А. Portion of plant. — . Leav . Involucral bracts. — H. Pistillate floret. F ^ Sd - K. Style of Шш floret. —L. Stamen of functionally Volume 80, Number 2 Freire 409 1993 Revision of Chionolaena 110 105 100 55 90 85 "m x 4247 х | EN : \ 4 \ M B | | Bo | dS | | 25 А _ А Зст ____________ В-а 4mm Е Esq! | = H-12mm 0 | NN L О | J-M 1mm | co N-PO.5 тт staminate floret from homogamous capitulum. — M. Stamen of functionally neuter floret from heterogamous capitu- ит. — №. Top of pistillate floret. —O. Achene of pistillate floret. . Ovary of functionally staminate floret from homogamous capitulum. Below: Mexican distribution. A-P, MacDougall 4129 (5). 410 Annals of the Missouri Botanical Garden suborbicular leaves and purplish bracts. It is closely related to C. lavandulifolia: both species have obtuse outer bracts and the capitula arranged in dense umbels. N . Chionolaena arbuscula DC., Prodr. 5: 397. 1836. TYPE: Brazil. Minas Gerais: Marianne, 1833, M. Vauthier 298 (lectotype, selected here, G; isolectotype, GH, photograph at LP). Figure ~ Erect, rigid shrubs, with thin woody stems, 50 cm high. Leaves narrowly linear, 1-2 cm long, 0.7-1 mm wide with conspicuously revolute mar- gins, glabrous on adaxial side, abaxial surface densely woolly. Capitula heterogamous, solitary on the upper branches, sometimes arranged in small, loose, terminal umbels at the ends of the leafy stems. Involucre narrowly oblong. Involucral bracts 12-52; outer bracts linear-ovate, acute, 5 mm long, 1 mm wide, with few long woolly hairs, inner bracts linear elliptic, acute, 3-6 mm long, О 0.5 mm wide. Florets 32-65, functionally stami- nate florets commonly more numerous than the pistillate, occasionally equaling or fewer. Pistillate florets 10-32; corollas 1.8-2.4 mm long; achenes oblong, 0.7-1.2 mm long, 0.2-0.5 mm wide, se- riceous with long, apically acute, duplex hairs. Functionally staminate florets 21-47; 1.8 mm long; style branches acute, 0.15- corollas 0.40 mm long; anthers 1-1.5 mm long with tails shorter than the antheropodium; ovaries sterile, pubescence as in pistillate florets. Restricted to the state of Minas Gerais in southeastern Brazil. It occurs in the plan- Distribution. alto, which has a temperate climate and a well- marked dry season. Chionolaena arbuscula usu- ally grows adjacent to waterfalls on gray, sandy soils at 1,5 in May, August, January, and February. m. Flowering collections were made Additional specimens examined. BRAZIL. МІМА GERAIS: Serra de Itatiaia, Claussen, 1839 (GH); Serra 0 crevices adjacent to waterfall, 28 Jan. 293 10 (NY) Serra do Espinhaco, riacho margin, cerrado, sedge meadow (brejo) sandstone outcrops and gallery for- est, ca. 10 km W of Вагао de Cocais gray sandy soil, ca. 1,500 m, 24 Jan. 1971, Irwin et al. 29023 ate Environs de Rio de Janeiro et d'Ouro-Preto-Brésil, 6 Aug 1885, Glaziou 14959 (NY); Serra de Ouro Preto, A : 1892, Ule 2602 (HBG); prope Itambe do Matto dentro et in Serra de Piedade, Martius 843 (M); Serra do Сагаса, 880, Glaziou 11036 (G), 25 Mar. 1957, F. Pereira 2637, Pabst 3473 (both RB); Serra de Ibitipoca, 15 May 1970, Krieger & Urbano 8649 (RB); Campos de Ouro Branco, 26 Nov. 1922, Campos Porto 1218 (RB); pie do Curral, Belo Horizonte, 20 May 1952, P. L. Roth 1401 (RB). Sine loco. BRASILIA: n s.n. (NY), Mar- tius 829 (М); in 1903, L. Damazio 1159 (RB). Chionolaena arbuscula is closely related to C. lychnophorioides: both species are erect, rigid, woody shrubs with solitary capitula and achene hairs acute at the apex. Chionolaena arbuscula is easily distinguished by its narrowly linear leaves with conspicuous revolute margins and capitula with male florets outnumbering female florets. 3. Chionolaena capitata (Baker in Martius) Freire, comb. nov. Achyrocline capitata Ba- ker in Martius, Fl. Bras. 6: 117, tab. 39. 1882. Leucopholis capitata (Baker) Cufod., Feddes Repert. 31: 329. 1933. SYNTYPES: razil. Minas Gerais: Serra de Itatiaia, Sello 859 (not seen); Rio de Janeiro: Glaziou 4843 (K, S, US), Glaziou 5899 (lectotype, selected here, S), Glaziou 6593 (US), Glaziou 7723 (С). Figure 7. Chionolaena arbuscula DC. var. hololeuca Baker ex az., em. Soc. Bot. France 6. 1910. SYNTYPES: Brazil. Rio de Janeiro: Serra de Itatiaia, US), Glaziou 6593 (US), Glaziou 8768 (С, K). Weakly woody shrub with multibranched, erect stems, 20-60 cm high. Leaves linear, 5-13 mm long, 0.6-1 mm wide with slightly revolute mar- gins, tomentose adaxially, densely lanate abaxially. Capitula heterogamous (occasionally һотора- mous), arranged in dense umbels sessile and ter- minal on the upper branches. /nvolucre narrowly oblong. Involucral bracts 16-25; outer bracts ovate to linear-ovate, acute, 1.5-3.5 mm long, 0.5-1.5 mm wide; inner bracts linear, acute, 4-8 mm long, 0.4-1 mm wide. Florets 5-11, functionally sta- minate florets slightly more numerous than the pistillate, sometimes equaling, or all functionally FIGURE 6. | Chionolaena arbuscula DC.— А. Habit. D- 6. Involue ral bracts. =й Pistillate floret. floret. - Style of pistillat —B, ( l. Funcionally staminate Ovary of AER staminate flor Stamen. i. Below: Brazilian distribution. А, Clausen s.n. (GH); B :. Leaves, abaxial surface ah adaxial surface (right). — floret of functionally staminate pos of pistillate floret. — N. -N, Irwin et а 29310 (NY). Freire 411 Revision of Chionolaena Volume 80, Number 2 1993 L-N 1mm H-K 2mm A 3cm Annals of the 412 Missouri Botanical Garden Volume 80, Number 2 1993 Freire 413 Revision of Chionolaena staminate. Pistillate florets (0—)2—4; corollas 1.5- 4 mm long; achenes oblong-elliptic, 0.5-1.2 mm long, 0.25-0.50 apically obtuse, duplex hairs. Functionally sta- mm wide, pubescent with long, minate florets 2-8; corollas 1.8-3.5 mm long; style branches truncate or rounded, rarely sub- acute, 0-0.4 mm long; anthers 1-1.8 mm long with tails equaling antheropodium; ovaries sterile, contracted toward middle, mm long, 0.15- 0.40 mm wide, pubescence as in pistillate florets. Restricted to the states of Minas Gerais, Rio de Janeiro, and Paraná of southeastern Distribution. Brazil. Chionolaena capitata is found in the plan- alto, which has a temperate climate and a slightly to well-marked dry season. It often grows in rocky places, between 2,300 and 2,850 m. Flowering collections were made from April to October. Additional m imens examined. BRAZIL. MINAS GERAIS: bue aparaó, Pico de Calçado, 2,850 m, 25 Sep. 1 A. C. Brade 17013 (NY); Pico Luis Ignasio, 2,300 m, i3 Sep. 1941, А. C. Brade 16898 (NY); slope of Pontào Crystal?, in rocks, 30 Apr.-4 May 1925, Chase 9708 А Subida Agulhas Negras, ca. 2,400 m, 16 Aug. 1969, Sucre 5765 (COL). PARANA: Dusén s.n. 8). 1933, Brade É 23 Aug. 1960, O. Handro 9423 3 (NY): 2; 350 n m, terreno pedregoso, 29 Apr. 1977, Joly C. A. et al. 6790 (NY); 2,400 m, 4/10 June VH F. Tamandaré 6384 (S); Agulhas Negras, 3,000 m, 4/10 June 1913, F. Tamandaré & А. С. а DEAN 19/ 27 July 1901, е 33(5 ); Municipio Rezende, "hn daas. de Itatiaia, жы proximo do abrigo Re Martinelli 7762 (RB); Pedra do Altar, 2,400 m, 1937, A. C. Brade 15599 (NY); b rie Аи Kegels," (Wa- wra 396 (NY); planalto, km 16 de la carretera, ca. 2,300 m, 25/26 May 1961, сагаа & Fabris 14212 (GH, , Retiro, 5 June 1902, ana, 4 July 1978, Lua & de pea 30 (RB); Serra do rdc 26 Dec. 1895, Ule 3913 (НВС). Chionolaena capitata is recognized by its adax- ially and abaxially pubescent, linear leaves and dense umbellate clusters of capitula. Glaziou (1910) mentioned the variety hololeuca ex Baker and cited the following specimens 4842, 4843, 6593, and 8768. Analysis of these speci- mens, except 7842, shows that they belong to C. capitata because they have small leaves and few- flowered capitula arranged in dense umbels. Fur- thermore, the specimens Glaziou 4843 and 6593 form part of the type material cited by Baker (1882) for Achyrocline capitata (— capitata). Chionolaena 3 Chionolaena columbiana S. F. Blake, J. ash. Acad. Sci. 25: 312. 1935. TYPE: Co- lombia. Paramos of the Sierra Nevada de Santa Marta, about 30 mi. inland from Dibulla, ca. 4,390 m, July 1932, William Seifriz 494 (holotype, US). Figure 8. Caespitose shrub with weakly woody stems, 15— 1.3-4.5 cm long, 3-4.5 mm wide with slightly revolute mar- 30 ст high. Leaves linear-obovate gins, tomentose adaxially, densely lanate abaxially. Capitula heterogamous, arranged in dense umbels surrounded by leaves of the same shape as the stem leaves, umbels supported at maturity by a woolly, thin peduncle with a few scattered, gen- erally rather small leaves. Involucre obdeltate. In- volucral bracts 25-27; outer bracts ovate, acu- minate, 4 mm long, 1.5 mm wide; inner bracts linear elliptic, obtuse, 3-5 mm long, 0.5-1 mm wide. Florets 13-20, pistillate florets slightly more numerous than the functionally staminate. Pistil- late florets 7-9; corollas 1-2.8 mm long; achenes oblong, 0.3-0.8 mm long, 0.15-0.20 mm wide, glabrous. Functionally staminate florets 6-8; co- rollas 1.5-3 mm long; style branches rounded, 0.3-0.4 mm long; anthers 1 mm long with tails equaling antheropodium; ovaries sterile, flattened, 0.3-0.7 mm long, 0.15-0.20 mm wide, glabrous. Distribution. | Restricted to northern Colombia on the Sierra de Santa Marta and the Sierra de Perija. It grows in rocky places, between 3,000 and 3,900 m from April to November. . Flowering collections were made Additional specimens examined. COLOMBIA. МАС- DALENA: Cordillera Oriental, Sierra de Perijá, 12 km E E of Manaure, E of Valledupar, 1 km f s Venezuelan border, 3,000 m, 5 Feb. 1945, M. L p 10864 (NY, US); Sierra de Perijá, plain between Cerro Venado and Cerro Avión, paramo 3,270-3,350 m, 8 Nov. 1959, Cuatrecasas & Castaneda 25115 (US); x erra Perijá, Tres Tetas (south peak, 3,000 m, 28 Apr.- May 1942, Carriker 15 (05); Mun. La Paz, э colombo- venezolana, Serrania de Perijá, entre los cerros El oon do y El Avión, 3,200-3,600 m, 3 Mar. 1959, Romero & Castañeda 7356 (MO); Sierra Nevada de Santa Marta, 1 km S of laguna Rio Frío, 3,800-3,900, FIGURE 7. Chionolaena Ma cea (Baker in Martius) Freire. — A. dione of plant. — B, € ): Н. Pistillate flor um of functionally staminate flore -N. Stamen. Below: Brazilian distributian. A- N, Hemmendorff 6. 33 ( (S). (left), adaxial surface (right . Involucral bracts. - Style of pistillate floret. — K. functionally staminate floret. - 'aves, abaxial surface I. Functionally d floret. — J. keen of pistillate flor Ovary of Annals of the Missouri Botanical Garden Ии Ve CERA "an TRY | SU АП Volume 80, Number 2 93 Freire 415 Revision of Chionolaena ca. 10%55'N, 73*53'W, 30 July n J. H. Kirkbride Jr. & E. Forero 1803 (MO, NY, RB, US); Sierra Nevada de Santa Marta, Transecto del Burünca (Filo La Cumbre), 3,850 m, 19 Aug. 1977, Rangel et al. 988 (U); Sierra Nevada de Santa Marta, alrededores de cabeceras de Rio Ancho, Paramo de Macotama, on roc outcrop of high ridge SE of campsite at joining of two streams, 3,840 m, 13 Feb. 1959, Barclay & Juajibioy 6927 (US). Chionolaena columbiana is closely related to C. isabellae from southern Brazil: both species have obovate leaves and pedunculate capitules- cences, but the former differs by its densely um- bellate clusters of capitula and acuminate outer bracts. л · Chionolaena сопсїппа (A. Gray) Anderberg & Freire, Notes Roy. Bot. Gard. Edinburgh 46(1) 37-41. 1989. Gnaphalium concin- num А. Gray, Proc. Amer. Acad. Arts 15: 34. 1879. Anaphalis concinna (A. Gray) Grierson, Notes Roy. Bot. Gard. Edinburgh 31: 392. 1972. TYPE: Mexico. San Luis Po- tosi: 22?N, alt. 6,000-8,000 ft., in 1878, Parry & Palmer 423 (lectotype, designated by Anderberg & Freire (1989), GH; isolec- totypes, BM, E). Figure 9. Subshrub, with thin, subdichotomously branched stems, 20-40 cm high. Leaves obovate, subpe- tiolate, obtuse, 1.5-2.5 cm long, 4-7 mm wide with slightly revolute margins, tomentose adaxially, densely whitish lanate abaxially. Capitula heter- ogamous, arranged in sessile and terminal corymbs that later become deeply lateral. /nvolucre obdel- tate. Involucral bracts 30-34; outer bracts ovate, acute, 4 mm long, 1.5 mm wide; inner bracts stipitate, obtuse, 6 mm long, 0.5 mm wide. Florets 36-42, pistillate florets slightly more numerous than, = equaling, or slightly fewer than the func- tionally staminate. Pistillate florets ca. 20; corollas -2.9 mm long; achenes elliptic, 0.6–0.8 mm long, 0.3 mm wide, pubescent with short, apically obtuse, duplex hairs. Functionally staminate flo- rets 16-22; corollas 1.5-3 mm long; style branch- es truncate, 0.4 mm long; anthers 1.5 mm long with tails equaling antheropodium; ovaries sterile, strongly contracted, 0.2-0.6 mm long, 0.10-0.15 mm wide, pubescence as in pistillate florets. Distribution. Restricted to central Mexico. Known only from the original site, which is at 22°N between 1,800 and 2,400 m. Flowering collections were made in September. Additional specimen examined. MEXICO. SAN LUIS POTOSI: Sep. 1879, Schaffner 222 (BM, GH). Chionolaena concinna is distinguished from other Mexican species of the genus by its slender, subdichotomously branched stems and obovate leaves. 6. Chionolaena chrysocoma (Wedd.) Freire, comb. nov. Oligandra chrysocoma Wedd., Chlor. Andina I: 158. 1855. Pseudoligandra chrysocoma (W edd.) Dillon & Sagast., Taxon 39: 125. 1990. TYPE: Colombia. Nouvelle Grenade, prov. de Rio Hacha, Sierra Nevada, 3,800-4,300 m, 1852, Schlim 809 (holo- type, P). Figure 10. Erect, rigid, multibranched shrub, 30-40 cm high. Leaves closely appressed to stem, linear, acute, 4—6 mm long, 0.8-1 mm wide, glabrous adaxially, densely whitish lanate and completely concealed by the revolute margins abaxially. Capitula het- erogamous, arranged in dense umbels sessile and terminal on the upper branches. /nvolucre nar- rowly oblong. Involucral bracts 21—31; outer bracts linear-ovate, acute, 4-4.5 mm long, 0.5-0.8 mm wide; inner bracts linear elliptic, slightly acuminate, Florets 9-13, pistillate florets slightly more numerous than the functionally staminate. Pistillate florets 4-8; co- rollas 2.2-3.2 mm long; achenes oblong, 0.6–0.8 mm long, 0.25-0.30 mm wide, glabrous. Func- tionally staminate florets 3-6; corollas 2-4 mm long; style branches rounded to obtuse, 0.2-0.4 mm long; anthers 1.5 mm long with tails equaling 0.7 5-6 mm long, 0.6-0.2 mm wide. antheropodium; ovaries sterile, flattened, 0.4— mm long, 0.2 mm wide. Distribution. Restricted to northern Colombia on the Sierra de Santa Marta. This species grows at the highest elevations of any species in the genus (between 3,300 and 4,300 m). Flowering collec- tions were made in March and August. Additional specimen examined. COLOMBIA. MAG- DALENA: Sierra Nevada de Santa Marta, Transecto del Buritaca, Filo La С epe 3,300 m, Aug. 1977, O. Ran- gel & А. Cleef 934 (L — E 8. Chionolaena columbiana S. F. Blake. volucral bracts. — C. dre floret. — H. Functionally staminate floret. — I. e of pistillate floret. — M. Style of functionally staminate er used staminate floret. — K. Stamen.—L. St — А. Portion of m — B. Lea adaxial surface. т. i In- Achene of ae floret. —J. O of Below: Colombian distribution. A—M, Kirkbride & Forero 1603 (MO). 416 Annals of the Missouri Botanical Garden Chionolaena chrysocoma is distinguished from other species of the genus by its appressed leaves. It is similar to C. phylicoides: both species exhibit narrowly oblong involucres, capitula arranged in dense umbels, and lower leaf surfaces completely concealed by the revolute margins. 7. Chionolaena eleagnoides Klatt, Leopoldina 23: 88. 1887. Gnaphalium eleagnoides (Klatt) S. F. Blake, Contr. U.S. Natl. Herb. 23: 1511. 1926. [Gnaphalium hypochaio- naeum Sch. Bip. ex Klatt, Leopoldina 23: 88 1887, C. eleagnoides, as synonym, no- men nudum.] TYPE: Pelado, Aug. 841, Liebman 316 (holotype, C; isotypes (fragment), GH; photograph ex € in 05). Fig- sub exico. ure 11 Moderately branched shrub with weak, woody stems. Leaves broadly obovate, narrowed at the base, 3-4.5 ст long, 11-12 mm wide with slightly revolute margins, adaxially arachnoid pubescent, densely whitish lanate abaxially. Capitula. heter- ogamous, arranged in corymbs sessile and terminal to the stems. /nvolucre obdeltate. Involucral bracts ca. 42; outer bracts ovate, obtuse, 2.5 mm long, 1.5 mm wide; inner bracts oblong obovate, 3 mm long, 1 mm wide. Florets 30-34, functionally sta- minate florets slightly more numerous than pistil- late or equaling them. Pistillate florets 15-16; corollas 1.5-2.5 mm long; achenes oblong, 0.5 mm long, 0.25 mm wide, glabrous. Functionally staminate florets 15-18; corollas 1.2-2.5 mm long; style branches rounded, 0.2 mm long; anthers -2 mm long with tails equaling antheropodium; 9 ovaries sterile, flattened, 0.3 mm long, 0.2 mm wide, glabrous. Distribution. Known only from the type lo- cality at Pelado, Mexico, at 3,000 m. Flowering collections were made in August. Chionolaena eleagnoides is recognized by its broad, obovate leaves with slightly revolute mar- gins. Chionolaena isabellae Baker in Martius, Fl. Bras. 6: 130, tab. 46. 1882. SYNTYPES: Brazil. Rio de Janeiro: Serra de Itatiaia, 22 Jan. 1873, Glaziou 6601 (lectotype, selected here, 5; isolectotypes, С, UC); Serra de Itatiaia, 10 Jan. 1876, Glaziou 1810 (NY, S). Figure 12. 8. Chionolaena put iovii Baker in Martius, Fl. Bras. 6: 130. 1 TYPE: Brazil. Minas el. sommet de la Sierra de Tirapitinoja, 2,120 m, 20 Nov. 1876, sid iou 876 69 (lectotype, selected here, P; isolec- — totypes, G, Chionolae па innovans Wawra, Itin. Princ. S. Coburgi : 30. 1888. TYPE: Brazil. анаја, Gipfel des Kegels, To Wawra 420 (holotype, W). Erect, rigid, multibranched shrub, 15-40 cm high. Leaves narrowly obovate, 1.3-2.2 ст long, 3-5 mm wide with slightly revolute margins, to- mentose adaxially, densely lanate and whitish abax- ially. Capitula homogamous, arranged in dense umbels, these umbels aggregated in spreading cor- ymbs terminal to the stems supported at maturity by а peduncle, 1.5-2.5 ст long, wholly devoid of leaves. Involucre obdeltate. Involucral bracts 18- 28, yellow; outer bracts broadly ovate, obtuse, 2.5-3 mm long, 1.5-2 mm wide; inner bracts linear oblong, obtuse, 4-5 mm long, 0.4-1 mm wide. Florets 12-31, all perfect (apparently with embryo developed); corollas 2-3.2 mm long; style branches rounded, 0.3-0.5 mm long; anthers l- 1.3 mm long with tails equaling or slightly ex- ceeding antheropodium; achenes oblong, 0.5-0.9 mm long, 0.2-0.4 mm wide, pubescent with long, apically acute, duplex hairs. Distribution. Restricted to the states of Minas Gerais and Rio de Janeiro of southeastern Brazil. This species occurs between 2,000 and 3,000 m. FIGURE 9. Ми и br: functionally staminate flore B an d s. G. Pist aa floret — K. Stame L. Ova A, Parry € Palmar 123 E acts. Jelow: Mexica FIGURE 10. Chionolaena chrysoc а (Wedd.) Freire. Involucral bra G. Pistillate floret K. Style of functionally staminate lora TL Top functionally staminate floret. Dt :olomibian pact den FIGURE 11. е па iis пен Klatt . Pistillate flore nally е Ye -M, Lislman 316 (C of functio Mexican distribution. Chionolaena concinna (А. Gray) Anderberg & Freire. H. de isti] staminate )vary of | staminate Неле | staminate floret. — I. f pis de . Portion of plant. о staminate floret. — I. Sta of functionally staminate floret. 2). — А. ps — B. Leaf abaxial surface. — C- F. floret. pd of pistillate floret. — Ј. Style of Achene of pistillate floret. B-M, M). A. Po rry «e Palmer 423 (B Leaf abaxial surface. Style of pistillate floret. vary of rtion of plant. — B. Stamen. — J. $ oret. — M. Achene of pistillate floret. — N. Ov -N, Schlim 809 (P). B.L men. eaf abaxial surfac C-F. Involucral -J. Style of ка floret. — К. Style -М. Achene of pistillate floret. Below: Volume 80, Number 2 1993 Freire Revision of Chionolaena (_______АЗст A 3 В 1 с т — СЕ 4mm w GH2mm г !-М атт ~ ` ЛА | ~ Ј шы ox ~ X 9 s ig M ^d HEN y et ~ 80 А ен а Missouri Botanical Garden Annals of the 418 d ger Tee i egre gue AT di fes rn же | he ¡E > $42 Nc | y = were >" > p у à. же”? e E A 4 мере a ee P oí LS A 3 ст > C-F2mm co век 1mm СМ 0.5 тт Volume 80, Number 2 Freire 419 1993 Revision of Chionolaena ZAS кад — EM — L—JA3cm A 1cm A bf 2mm L— G-I 1mm LLL LJ J-M 0.5 mm 420 Annals of the Missouri Botanical Garden “ТУДО. UA OR «Ne а ~ S QE c ч КУТУ Ss Dy ИЩ жч en? рр Ee i o AC Le AI у, AN — Volume 80, Number 2 1993 Freire Revision of Chionolaena Flowering collections were made from February to December, except in July and August. Addition él e examined. BRAZIL. MINAS GERAIS- RIO D EIRO: Municipio de Itamonte, Parque Nacional do Datis prawi para opel у 2,450 m, 2 May 1 rtir ds F. Zuloc Vaz- re Avila 10843 (NY) arque Nac. do Itatiaia, 12 Mar. 1947, Cuchi 940 (LP); Itatiaia, Ped " Айат, n m, Mar. 1937, 4. C. Brade 15596 (NY) almost on the summit of v Mt. Itatiaia, 3,000 m, 26 . 1915, Oakes Ames s.n. (GH); Itatiaia, 2,450 m, E iis 1968, rid & s 25605 (M) a do Itatiaia, He lhas Negras, 3,000 m, 4-10 June i 3. Brade 6385 (S), 27 7 July 1901, pod ae 632 E E June 1902, pd 542 (S); Itatiaia, 2,400 m, 9 933, Brade 12 21260 (both RB); 2 500 m, 5 Te Brade 20 He 2 "is Serra do Насаа; colhina no pico das Agulhas Negras, 3,000 m, May 1913, Brade ; Itatiaia, Prateleiras, 2,200 m, 8 Feb. 1945, Brade 17421 (RB), 12 Apr. 1975, Camerik 29 (U); Campo Itatiaia, 13 May 1906, Luederwaldt 16733 pro parte (US), 3 Dec. 1964, deg 197 : ; Parque Nac. do d awe vicinity 2 Abrigo Rebouças “gregarious,” 2,350 m, 1 Feb. Linde eman "У Низ 4134 (Uy сы from si do E e пјене do Itatiaia to planalto, pd valley near broadcasting station, 2,200 n А leas x Haws $197 (U); Sitio ЈЕ = T eR 2,300 m, 899, Gounelle s.n. med Felson region. зу 2.300. m, d do Itatiaia, 27 Feb. 1895, Ule 3374 (HBC). Chionolaena isabellae is closely related to C. columbiana: both species have obovate leaves and a capitulescence supported by a slender peduncle. Chionolaena isabellae is distinguished from C. co- lumbiana by its corymbosely arranged, homoga- mous capitula and its villous achenes with apically acute, duplex hairs. 9. Chionolaena jeffreyi H. Robinson, Phyto- logia 55: 121. 1984. TYPE: Brazil. Municipio de Rio Contas, Pico das Almas a 18 km NW de Rio de Contas, 1,600-1,850 m, 24 July 1979, R. M. King et al. 8141 (holotype, RB not seen; yim СЕРЕС not seen, MO not seen, US). F ге 1: Subshrub, dns woody, multibranched, with slender ascendent stems, 10-20 linear-elliptic, narrowed at the base, 3-8 mm long, 1-2 mm wide with slightly revolute margins, to- cm high. Leaves mentose adaxially, densely lanate abaxially. Ca- pitula heterogamous, solitary on the upper branch- es. Involucre obdeltate. Involucral bracts 20-25; outer bracts 3.5 mm long, 1.2 mm wide; inner bracts 4.5 mm long, 1 mm wide. Florets 23-26, pistillate florets slightly fewer than the functionally staminate. Pistillate florets 8-11; corollas 2-2.5 mm long; achenes oblong elliptic, 0.7-0.8 mm long, 0.3 mm wide, pubescent with long, apically obtuse, duplex hairs. Functionally staminate flo- rets 15-25; corollas 2-2.8 mm long; style branch- es rounded, 0.2 mm long; anthers 0.7—1 mm long with tails equaling or slightly exceeding anthero- podium; ovaries sterile, 0.15 mm long, 0.20 mm wide, pubescence as in pistillate florets. Distribution. species endemic to the state of Bahia in Brazil, where the climate is temperate with long dry pe- riods. This species is found growing in damp, shady places, between 1,600 and 1,850 m. Flowering collections were made in March and July. Chionolaena jeffreyi is the only Additional specimen examined. BRAZIL. BAHIA: middle and upper NE slopes of the Pico das Almas ca. 25 kn WNW of the Vila do Rio de Contas, E conglomerate metamorphic and quartzite rock outcrop with associated scrubby А wis damp flushes and grassland and marsh in some areas zt јао ; 1,600-1,850 m, 19 Mar. 197 . M. Harley et al 19677 (paratypes, NY, КВ— 2 as U, US). Chionolaena jeffreyi is distinguished from other species of the genus because of its short petiolate, elliptic leaves, solitary capitula, and weakly woody, ascendent stems. 10. Chionolaena latifolia (Benth.) Baker in Martius, Fl. Bras. 6: 132, tab. 43, fig. 2. 1882. Leucopholis latifolia Benth., Hooker's Icon. Pl. vol. 12: 14. 1876. TYPE: Brazil. Rio de Janeiro: Itatiaia, 2,400 m, 9 July 1872, Glaziou 1841 (holotype, P; isotype, 5). Figure 14 Chionolaena Siocon ud in iier Fl. Bras. 6: 130. 1882 : Brazil. Rio de Janeiro: m tiaia, 6 July ni Glaziou 4851 ae selected ere, Р; о. h ex P in СН); Itatiaia, 9 July 1872, Glaziou 5896 (e excluded here, P е na breweri ол & Maguire, Acta Bot 14(3): 26, fig. 17. 1984. TYPE: Venezuela. Territorio Federal Amazonas: рки ен Cerro Marahuaca, Cumbre, dades del zanjón, 2.685 1 m, Jan. 1981, B. Maguire et B 65587 cedi. VEN not seen; isotype, — FIGURE 12. Chionolaena eran Baker in A ect floret. — Achene. — I. Stamen. — J. Martinelli: et al. 10843 (NY). Involucral A-K, .—B. Leaf adaxial surface. —C-F. —A. Habit P K. Style. Below: Pradhan distribution. appus. 422 Annals of the Missouri Botanical Garden —3 A 3cm LL 1 B-C 5mm LL j D-G3mm і 1 H-1 2mm La 1 J-N 1mm Volume 80, Number 2 1993 Freire 423 Revision of Chionolaena Erect, rigid shrub, devia branched with densely leafy stems, 15- m high. Leaves ob- long-elliptic or linear-elliptic, acute, 1.5-4 cm long, 3-5 mm wide with slightly revolute margins, gla- brous and bright adaxially, densely lanate and whit- ish abaxially. Capitula heterogamous, arranged in dense umbels, sessile and terminal on the upper branches, these umbels sometimes aggregated cor- ymbosely. /nvolucre narrowly oblong. Involucral bracts 15-21; outer bracts ovate, acuminate, 3— 4 mm long, 1-1.5 mm wide; inner bracts linear, subacute, 4–6.5 mm long, 0.5-1 mm wide. Florets 5-14, pistillate slightly fewer than the functionally staminate. Pistillate florets 2-6; corollas 2-3 mm long; achenes oblong to oblong-elliptic, 0.6- 1.2 mm long, 0.2-0.4 mm wide, pubescent with long, apically obtuse, duplex hairs. Functionally sta- minate florets 3-9, corollas 2-3 mm long; style branches truncate, 0.2—0.4 mm long; anthers 0.8— 1.3 mm long with tails slightly exceeding or equal- ing antheropodium; ovaries sterile, oblong, flat- tened, 0.6-1 mm long, 0.2-0.4 mm wide, pubes- cence as in pistillate florets. Distribution. open rocky slopes, between 2,300 and 3,000 m. This species has a discontinuous distribution, con- Chionolaena latifolia grows on fined to northern Amazonia and southeastern Bra- zil. Similar distributions are frequent among other 1 1958; Мааз, 1972; Ап- . Flowering collections were made plant species (Harling, dersson, 1 between October and Marc Additional qu үзаң BRAZIL. MINAS GERAIS: Папаја, 2,400 m, 10 Jan. 1876, Glaziou 5903 (cited by Baker for Chionolaena iilo NY, 5); near Air of Mt. Itatiaia, Campo Bello, 3,000 m, 26 Dec 6, O. Ames s.n. (GH). RIO DE JANEIRO: Itatiaia, Awl: Negras, 2,600 m, 22-28 Nov. 1938, Markgraf & Brade 3702 (RB—2 sheets); Agulhas Negras, Mt. Па- пала, Estação Biologica, 2,300-: ~ 22924'S. 44°41'W, lower slopes, 5 Jan. 1929, Smith 1682 (GH, US); Serra do Itatiaia, ca. 3, Ld m, 27 July 1901, Hem- mendorff 648 (5); ca. 2,200 m, in campo saxoso, 17 а 1903, Dusén 2007 (S) Sera, "da Pedra Asentada, rupitens, 2,300 m, 30 May 1 а Dusén 395 (S- sheets); debe Pedra do Alta O m, Mar. ea Brade 155 —2 sheets); pa vede 1906, deris 16733 pro parte Resende, Itatiaia National dom rige: oi Mt Itatiaia, 2,400 m, 22°25-26'5, 44°4 2'W, 16-1 km N of shelter house, **Abrigo Rebougas” in valley dari to the May ; Municipio rock dub “Pedra do Altar," 3 Nov. 1965, G. Eiten . T. Eiten 6553 (US). VENEZUELA. TERRITORIO FE- DERAL AMAZONAS: Depto. Atabapo, Cerro Marahuaca, cumbre, porción suroccidental (Atahua-Shiho), vegetación no arbolada en terreno pendiente, 2,450-2,480 m, 3*30'N, 65°20'W, 9-10 Feb. 1982, Steyermark, Guariglia, Holmgren & Mori 126322 (paratype of Chionolaena breweri, ay na pang Aq 65?28' 2,600 m, 30 Mar.- 983, Steyermark & F. De lascio 129265 5 та "ем id ME FHUIF, ra zona boscosa en la falda este del riachuelo, 2,480-2,500 m, 3°35'N, 65?20'W, 1-2 Feb. 1982, Steyermark et al. 126074 (Uy Cerro Marahuaca, summit on undulating plateau with islands of shrubs and Marahuaca with stream branch leading to Salto de Monos and SE of summit Camp, 3°37'N, 65?23'W, 2,520-2,650 m, 26-27 Feb. 1985, Steyermark & Holst 130822 (MO); open rocky plateau of Cerro de Marahuaca above Salto Los Monos on trib- шагу of headwaters of Rio Iguapo, 3%37'N, 65?23'W, 2,555 m, 26 Feb. 1985, R. L. Liesner 17987 (U). Chionolaena latifolia is an easily recognized species with characteristic foliage, i.e., elliptic leaves with glabrous, bright adaxial surfaces and slightly revolute margins. 1984) described C. breweri from Venezuela. This species was said to differ from C. latifolia in the width of the capitulescence, size of the capitula, and habit. Steyermark and Maguire (Steyermark, Because this variation is continuous and not cor- related with distribution, these species are treated here as conspecific. І have excluded the sterile syntype of Chiono- laena glomerata, Glaziou 5896, from this species and placed it under C. wittigiana because of its foliage. This specimen has linear leaves with adax- ially tomentose surfaces, whereas the leaves of C. glomerata (= C. latifolia) are elliptic with gla- brous, bright adaxial surfaces. 11. Chionolaena lavandulifolia (Kunth) Benth. & Hook. f. ex B. D. Jackson, Index Kew., fasc. I: 516. 1893. Elichrysum lavan- p те Kunth, Nov. Gen. Sp. ed. fol. IV: 86. 8. Gnaphalium lavandulaceum DC., Prodr. | 227. 1837, nom. nov. phalium lavandulaefolium Willd. Chiono- laena lavandulaceum (DC.) Hemsley, Prodr. Biol. Centr. Amer., Bot. 2: 134. 1881, nom. illeg., oldest epithet not used. Gnaphalium lavandulaefolium (Kunth) S. F. Blake, Contr. ue to Gna- — FIGURE 13. (right). —D-G. Inv Chionolaena јеђ оу. i H. Robinson. - olucral bracts. — H. Pist staminate floret. — K. Style of pistillate floret. — L. Stamen. — M. A. Habit. illate floret. —1. Functionally staminate floret. —J. Style — B, C. Leaves, adaxial surface (left), abaxial surface of functionally Achene of pistillate floret. — N. Ovary of functionally staminate floret. Below: Brazilian distribution. A-N, Harley et al. 19677 (RB). 424 Annals of the Missouri Botanical Garden Мара Ny v. #55 (1 > LA na ты ZN AY ES ARS RES E ЈУ 5 PE Ed PES > m — MY > САМ EN LA Xx => s 2 Km. 77 == == РА) ri ~ => à _ == LL j 0-1 3mm L l J-K 2mm L J LN 1mm Volume 80, Number 2 1993 Freire 425 Revision of Chionolaena Gray Herb. n.s. 52: 21. 1917, hom. illeg., non МИРИНА "Meroe A Willd., . 3: 1868, 4, which is ÉL Ci alan pede on., Mem rn. Nat. Hist. Soc. 5: 550, 1826. TYPE: Mexico. Crescit in frigidis Andium Mexica- norum in nobilissimo monte Nauhcampatepe- tle juxta urbem Perote, alt. 1720 hex. h. Floret Martio Herb. HBK (holotype, P not en; photograph at CH, LP-neg.no.37976, distributed by F). Figure 15. Caespitose, multibranched shrub with weakly woody stems, 10— igh. Leaves linear-ellip- tic, acute, 7-18 mm long, Dew mm wide with margins sometimes slightly revolute, tomentose adaxially, densely lanate and whitish abaxially. Ca- pitula heterogamous, commonly arranged in dense umbels sessile and terminal on the upper branches, sometimes in a compact corymb. /nvolucre ob- deltate. Involucral bracts 25-44; outer bracts broadly ovate, obtuse, 3-4 mm long, 1-2 mm wide, dark brown at the tip; inner bracts linear-oblong, obtuse, 6.5-9 mm long, 0.5-1 mm wide, dark brown above the stereome. Florets 11–48, pistil- late more numerous or slightly more numerous than the functionally staminate, sometimes slightly few- er than the functionally staminate. Pistillate florets 6-24; corollas 3-4.5 mm long; achenes oblong elliptical, 0.7-1.9 mm long, 0.3-0.6 mm wide, glabrous, occasionally shortly pubescent with api- cally obtuse, duplex hairs. Functionally staminate florets 5-13; corollas 3-5 mm long; style branches rounded or subacute, 0.2-0.4 mm long; anthers 1.5-2.2 mm long with tails shorter than anther- opodium; ovaries sterile, flattened, 0.25-0.70 mm long, 0.1-0.4 mm wide, glabrous, occasionally with short pubescence as in pistillate florets. Pappus bristles with apical cells slightly swollen and acute at the apex. Distribution. | Chionolaena lavandulifolia has a highly disjunct distribution, i.e., it occurs in the central mountains of Mexico and in the central parts of Costa Rica. This species is found among rocks in sandy, gravelly soil and volcanic sand on open exposures, commonly on summits between 3,150 and 4,200 m. Flowering collections were made between July and March, except in Novem- ber and December. Additional specimens examined. MEXICO. MEXICO: summit ledges, Nevado de To m, 27 Mar. 1954, Troll 721 (М); уйнен S side of mtn. at La Joya, са. 4,015 m, “‘in crevices at base of cliff; occasional," 8 Aug. 1960, пали 3011 (UG); 4,000-4,050 m, “under rock ledge," 30 July 1958, Beaman 1961 (UC); La Joya de Alcalican, extremo SW del Ixtaccihuatl, municipio de Amecameca, “‘pradera al- pina," 3,900 m, 13 Nov. 1977, Rzedowski 3555 1 (UC); Ixtaccihuatl, 12,000- 13,000 ft., Sep. —€ 1905, Pur- pus 1523 (UC). PUEBLA: Mt. Orizaba, rocks, alpine region, Mar. 1908, Purpus 2990 (UC); Orizaba, 3,800-4,200 m, Oct. 1906, Ross 1265 (HBG, M); Mount, 13,300 ft., rocks and volcanic sand open exposures, 29 July 1928, Balls B5139 (UC). a s Cofre de Perote, summit of mtn., 4,192 m, 6 Mar. 1958, Beaman 2145 (UC). Costa Rica. On rocky а on lower slope of Cerro bu Buenavista massif, 3,400 m, 14-16 Mar. 1969, Weston 5877 (UC); Rochers de la region super- ieure du Cerro de Buena Vista, 19 Jan. 1891, Pittier 3430 (G); Cerro de La Muerte region, Cordillera de Tala- manca, near summit of Cerro Sakira, 3,415 m, 8 Jan. 1985, Horn 38 (UC); Chirripo your 3,450 m, 29 Apr. Mund ig d (M), 27 Apr. 1932, Kupper 1301 o 2, Kupper 1148 (M); below Sabana iud А c 3,150 m, 24 Feb. 1976, Wes- ton 10122 (UC); Cerro de Chirriposillo (Cuericil) within ca. 200 m of main summit, 3,400 m, 2 Mar. 1964, Weston 1551 (UC); Prov. of Cartago, shrub-paramo (Chusquea, Hypericum, Paccinium), Cerro de La Muerte, Pan-American highway, 5 km above Millsville (about 8 km above Nivel), С ordillera de Talamanca, 3,400-3,500 m, crevice in rocks, common at summit, characteristic together with caespitose Hypericum of rocky high pa- ramo, 25 July 1949, К. W. Holm & H. H. Iltis 573 (M); in summo monte Irazu, Jan. 1847, Oersted 10570 (GH, MO, UC). ~ Chionolaena lavandulifolia is closely related to C. aecidiocephala. In both species the outer bracts are obtuse and the capitula are arranged in dense umbels, but the former is easily distinguished by its adaxially densely lanate, linear-elliptic leaves, and commonly glabrous achenes. 12. Chionolaena lychnophorioides Sch.-Bip., Pollichia 20-21: 391. 1863. SYNTYPES: Bra- zil. Sellow s.n. (in B presumably lost); Riedel — FIGURE 14. abaxial surface (right ж pistillate floret. — K. of functionally HA floret. GH). = Loa c = Chionolaena онад (Benth.) Baker in Martius —D- ac e of functionally staminate floret. — Below: northern Amazonian (Venezuela) and Brazilian distribution. A-N, Sind 1682 5. — А. Habit. — B, C. Leaves, adaxial surface (left), L. Stamen. — M. Achene Annals of the Missouri Botanical Garden Ld A 3cm LL ] B-C 1cm LLL] D-G 3mm L ЈН 3mm L ] J-N 2mm M wW M x =, E zi Volume 80, Number 2 1993 Freire 427 Revision of Chionolaena s.n. (in rocky area on Itacolumy mountain, Aug. 1824, LE not located); Claussen 184 (Minas Gerais, Crarassa, lectotype, selected here, P; negative at LP). Figure 16 Erect, rigid shrub, moderately branched, 15- 35 cm high. Leaves linear, acute, 1.3-3.3 cm long, 1.2-3 mm wide with slightly revolute mar- gins, adaxially arachnoid pubescent, densely lanate and whitish abaxially. Capitula heterogamous, sol- itary on the upper branches, occasionally arranged in terminal, small, loose umbels at the ends of the leafy stems. Involucre broadly campanulate. In- volucral bracts 28-29; outer bracts linear ovate, acute, 5 mm long, 1–1.5 mm wide; inner bracts linear, 5-6 mm long, 0.9 mm wide. Florets 48- 54, pistillate more numerous than the functionally staminate. Pistillate florets ca. 35; corollas 2— mm long; achenes oblong, 1-3 mm long, 0.3- 1.2 mm wide, pubescent with long, apically acute, du- plex hairs. Functionally staminate florets ca. 13; corollas 3.5-4 mm long; style branches acute or rounded, 0-0.4 mm long; anthers 1.5-1.7 mm long with tails equaling antheropodium; ovaries sterile. Distribution. is restricted to the state of eastern Brazil. It occurs in the planalto, which has a temperate climate and well-marked dry season. Chionolaena lychnophorioides inas Gerais in south- This species grows on stony and rocky mountain slopes, between 1,350 and 1,450 m. Flowering collections were made in April, May, and Decem- ber. Additional P d ern Mes MINAS GERAIS: y Itacolumy Our , Glaziou 14960a MO, US); Itacolumy (E of o 8 Apr. 1925, Chase 9389 (GH, MO); Itacolumy, Feb. 1892, Ule 2601 чоң. һасойшшу, Ouro Preto ‘“сатреѕ a," 28 a БЕРЕ Dec , A. Macedo 2801 (NY, m US» Pico ve depressao, defen (valo) no alto Serra, 5 May 1 M. c oap 1696 (COL); Serra de Ibitipoca, ds м Pio: 1,350-1,450 m, formações de arenito de Serie Lauras, 14 May 1970, Sucre & Krieger 6636 (RB). Chionolaena lychnophorioides is distinguished by its usually solitary capitula with numerous florets and its linear leaves. It is related to C. arbuscula, from which it is readily distinguished by its broadly linear leaves with slightly revolute margins and capitula with pistillate florets outnumbering the functionally staminate florets. 13. Chionolaena mexicana Freire, TYPE: Mexico. Hidalgo: Sierra de Pachuca, 10,000 ft., 20 Feb. 1899, Pringle 7700 (ho- lotype, K). Figure 17. sp. nov. Frutex erectus dichotome ramosus, usque ad ca. 0.3 m altus. Caules ascendente-erecti, tomentosi, foliati, ve- tustiores glabri nudi, cicatricibus foliorum notati. Folia alterna, suberecto-patentia, inferiora deflexa, obovata, ob- mm longa, 3-7 mm lata, apice siii voluta, supra tomentosa, subtus den t per- bi terminales. Capitula o € tusa, 15 и mm longae. Stilus bifidus. Achenia oblonga, ca. 1 mm longa, ca. 0.3 mm lata, minute puberula. Flores disci ca. 11, hermaphroditi, steriles et masculae fungentes. Corol- lae tubulosae, 5-lobatae, ca. 4 mm longae. Stylus bifidus, ramis apice truncatis et dorso pubescentibus. Antherae ca. 1.8 mm longae, caudae an Јани са. 0.3 mm longae. Achenia sterilia, anguste oblonga. Pappus albus, ooo setis nai corolla parum ke ы ђаг- bellatis, basi ir , florum tubulo- sarum apice incrassatis. Erect, dichotomously branched, rigid shrub, 30 cm high. Leaves narrowly obovate, slightly atten- uate at the base, obtuse, 1.5-2.5 cm long, 3-7 mm wide with slightly revolute margins, tomentose adaxially, densely whitish lanate abaxially. Capit- ula heterogamous, 10-18, arranged in corymbs sessile and terminal on the upper branches. /n- volucre obdeltate. Involucral bracts 34—36; outer bracts ovate, slightly acuminate, 2.2-2.5 mm long, 1 mm wide; inner bracts linear, obtuse, 5-6.5 mm long, 0.5 mm wide, dark above the stereome. Flo- rets 25-26, pistillate slightly more numerous than the functionally staminate. Pistillate florets 14- 15; corollas 3.8 mm long; achenes oblong, 1 mm long, 0.3 mm wide, pubescent with short, apically obtuse, duplex hairs. Functionally staminate flo- rets ca. 11; corollas 4 mm long; style branches truncate, 0.4 mm long; anthers 1.8 mm long with tails slightly shorter than antheropodium; ovary sterile, flattened, 0.7 mm long, 0.2 mm wide, pu- bescence as in pistillate florets. FIGURE 15. зел surface (left), adaxial m (rig floret. —J. Stamen. — К. Achene of pistillate bu -L Ova le of functionally staminate floret. Below: Mexican and Central American (Costa Rica) distcibulum. bee —N. St N, Gonzalo Flores M. PV78120 (UC). Chionolaena lavandulifolia oo Benth. & Hook. f. ex B. D. Jackson.— A. Habit ). >. Involucral bracts .— B, C. Lea ctionally staminate floret. — I. Pistil late Style of pauiat .—H. Fun ary of molle staminate floret. — M. Annals of the Missouri Botanical Garden Distribution. Known only from the type col- lection, the state of Hidalgo, Sierra de Pachuca at 2,700 m. A flowering collection was made in Feb- ruary. This species is readily distinguished from all other species of Chionolaena by its obovate leaves, corymbosely arranged heterogamous capitula, and achenes with obtuse short hairs. Its closest relative appears to be C. concinna, the only other Mexican species with obovate leaves, capitula arranged in corymbs, and achenes with obtuse short hairs. Chionolaena mexicana differs from C. concinna by its erect, rigid habit, and capitulescence terminal at maturity. In addition, the leaves of C. concinna are narrowed into a petiolariform base. In C. mex- icana, the leaves are basally attenuate. 14. Chionolaena phylicoides ebd Ba- ker in Martius, Fl. Bras. 6: 131. 1882. Leu- о phylicoides Gardner, aren Bot. 2: 11. 1843. TYPE: Brazil. Rio de Janeiro: 841, Gardner 5772 (lectotype, selected here, GH, photograph at LP; isolectotypes, G— 3 sheets, NY, S). Fig- Organ Mountains, 1836-1 Erect, rigid shrub, moderately branched with stems densely leafy, 15-45 cm high. Leaves ob- long, acute, 5-10 mm long, 1-2 mm wide, gla- brous and bright adaxially, densely woolly and com- pletely concealed by the revolute margins abaxially. Capitula homogamous, unisexual, arranged in ense umbels sessile and terminal on the upper branches. /nvolucre narrowly oblong. Involucral bracts 20-30; outer bracts linear ovate, acumi- nate, 4—5 mm long, 1 mm wide; inner bracts linear, obtuse, 4-5 mm long, 0.7-0.8 mm wide. Florets 7-12, all hermaphroditic functionally staminate florets with ovary not developed, functionally pis- tillate florets without anthers. Functionally pistil- late florets ca. 10; corollas 3-3.8 mm long.; style branches truncate, 0.4-1 mm long; anthers 0; achenes oblong, 0.8-0.9 mm long, 0.2-0.5 mm wide, pubescent with long, duplex hairs obtuse at the tip. Functionally staminate florets 7-12; co- rollas 2-3.5 mm long; style branches truncate, 0— 0.5 mm long; anthers 5, 1.2-1.8 mm long with tails equaling antheropodium; ovaries sterile, flat- mm long, 0.15-0.40 mm wide, pubescence as in pistillate florets. tened, Distribution. | Restricted to the states of Minas Gerais and Rio de Janeiro in southeastern Brazil This species grows commonly in rocky places on hillsides beween 1,950 and were made between June and November where a slightly marked dry season occurs. 2,850 m. Flowering collections Additional specimens examined. BRAZIL. MINAS GERAIS: . 1887, Glaziou 16157 (G, NY); Serra do Caparaó, Pico do Calgado, 2,850 m, 25 Sep. 1941, Brade 17012 NY); Pico ia Cristal, 2,600 m, 24 Oct. 1941, Brade 17004 (N RIO DE JANEIRO: 958, Кари ll 8432 (RB); Campo и Antas, 1940, Brade 16517 (GH, NY), 30 Jan. 1942, Pereira 208 (NY, RB); 2,000 m, Rizzini, 5 Ded 1951, (RB 73702 pro parte); Felsen, 1,950 m, Oct. 1952, Markgraf 10102 (NY, RB pro parte); Caminho para o Campo das Antas, 1 Nov. 1970, eer do Projecto Serra dos Orgáos (RB); Centralstock, 2,400-2,600 m, Aug. 1915, Luetzelburg 6252a, b, 6386 (both M); Municipio de Teresópolis, Parque Nacional de Serra dos Orgàos, Campo das Antas, t 2,000 m, 1 Feb. 1983, Martinelli & Simonis 90. 51 (US), Dionisio & Octavio, 28 June 1942 deira, 2,000- 2,100 m, crescendo em € formadas com Melastomataceae, Prepusa sp., Gramineas, Cyperaceas, Farney, E. C. Dalcin ES Pe na 800 (RB) p des Orgues, 1885, Glaz E. 14961 G). WITHOUT STATE: “Minas Cla- parao,” O. Voll., 2 Jan. 1937 (RB); Pedra do Sino, 1,750 m, rara no Edda 17 Sep. 1948, Rizzini 368 (RB). WITHOUT LOCATION: “‘turberas con Sphagnum,” Cabrera 12227 (LP). >= Chionolaena phylicoides is closely related to C. chrysocoma. In both species the capitula are arranged in umbels and the abaxial leaf surfaces FIGURE 16. Chionolaena lychnophorioides Sch.-Bip. —A. Habit. — D, surface (right). —D-G. Involucral bracts. floret. — K. Style of d staminate flore Bae tillate floret. — I. mu staminate flore :hene of pistillate floret. — C. Leaves, ршн surface es abaxial Style of pistillate M. Summit of pistillate corolla.— N. Stamen. Below: Brazilian distribution. A N, "Chase 9389 (MO). FIGURE 17. G. Pistillate 2 staminate floret. — L. Achene of pistillate floret. A-M, Pringle 7700 (K). Chionolaena mexicana Freire. FIGURE 18. surface (left), abaxial surface (right . Involucral bra capitulum. — I. Style. А. Habit. I Leaf abaxial surface -H. Functionally staminate floret. — I. Stam Chionolaena phylic Mid | yardner) Baker in Martius. -C-F. Involucral brac n. — J. Style of pistillate floret. — K. Style of func sionally M. Ovary of dnd staminate floret. Below: Mexican distribution. -A. Portion of plant. — B, C. Leaves, adaxial ts. — H. бкл» pistillate floret from homogamous (NY) —J. Асћепе. Ens File (које A-J, Glaziou 16157 Volume 80, Number 2 1993 Freire Revision of Chionolaena 429 DN eS D í ~ с ~ у Ñ B AD м DA em m XT 2 У ae dM > ВЕРУ. s Bp Rr S n Т Е S бе 46 У P) NS 7 е. NES #4 у T c a s. t / g ЯР РА Ws AR Z 5 Jj à SV М Y - 2 = A y sere SY a Se Жс A x N > :: о EL — ғ z SSS 222 2 E яу) ae ES e — BFF Missouri Botanical Garden Annals of the 430 S ( er А 2 \ SN (Ка D 4 \ i N > DY” MS \ \} Sar = i у sd >>> Е 2 X d c XS A "4 e) A. c Eta EOS | / / Volume 80, Number 2 1993 Freire Revision of Chionolaena NN der PO RA E SN N T NS. NN == = i \ ЖУЛА Y = АД AN Lod "а МЕД 7 MAA = = e ZI NS ZA) N/ NS MAN AN UN wy AM 2 ASIN AR YL NAN AN { AM SEA A IN pia, і AAN ПАД) = => oA RN EN > МКК ZR S 2; IS E LAN À 222 À) 227 SAS 25 238 cj > SS A YR. Oy D SSN IN > DA MT = NN AA И WAI WS. ES DS PEIES РРА OL W X = 3 2 SS Ca Fi р N = РУР WAS AS Gi, NW ER. 22 — <2 A Annals of the Missouri Botanical Garden are completely concealed by the revolute margins. Chionolaena phylicoides however, distin- guished by its homogamous capitula and reflexed is, leaves. 15. Chionolaena sartorii Klatt, Leopoldina 23: 89. 1887. Gnaphalium sartorii Sch.-Bip. ex Klatt, Leopoldina 23: 89. 1887, prosyn. sub. C. sartorii, nomen nudum. Gnaphalium sar- torii (Klatt) F. J. Espinosa-Garcia, Bol. Soc. México 45: 21. 1984. TYPE: Mexico. о. June 1942, Liebman 308 (ho- lotype, C; isotype (fragment), GH). Figure 19. Caespitose subshrub with weakly woody stems, 8—10 cm high. Leaves narrowly linear, acute, 1- 2 cm long, 1 mm wide with slightly revolute mar- gins, tomentose adaxially, densely whitish lanate abaxially. Capitula heterogamous, 3-4, arranged in corymbs sessile and terminal on the upper branches. /nvolucre obdeltate. Involucral bracts 20-25; outer bracts narrowly ovate, acuminate, 3 mm long, 1-1.2 mm wide; inner bracts linear, subacute, 5.5-6 mm long, 0.5 mm wide. Florets ca. 20-2], pistillate more numerous than func- tionally staminate. Pistillate florets 14—15; co- rollas 4 mm long; achenes oblong 0.8 mm long, 0.3 mm wide, pubescent with short, apically obtuse, duplex hairs. Functionally staminate florets ca. 6; corollas 4 mm long; style branches rounded, 0.2 mm long; anthers 1.8 mm long with tails equal- ing or slightly exceeding antheropodium; ovaries sterile, 0.8 mm long, 0.3 mm wide, pubescence as in pistillate florets. Known only from the type lo- cality in Sempoaltepec, Mexico, at about 3,300 m. The flowering collection was made in June. Chionolaena sartorii is recognized by its caes- Distribution. pitose habit, linear leaves, acute inner bracts, and 16. Chionolaena seemannii (Sch.-Bip. in See- mann) Freire, comb. nov. Gnaphalium see- mannii Sch.-Bip. in Seemann, Bot. "Herald" 7-8: 309. 1856. TYPE: Mexico. Sierra Madre, NW Mexico, Seemann 1994 (holotype, K). Figure 20. Chionolae na corymbosa Hemsley, Diagn. Pl. Nov. 2: 32. l Nomen superfluum. TYPE: Mexico. Sierra Maire, NW Mexico, Seemann 1994. Woody, rigid shrub with ascendent, erect and densely leafy branches, 15-20 cm high. Stems subdichotomously branched. Leaves linear elliptic, acute, 3.5-4.5 cm long, 4-5 mm wide with slightly revolute margins, glabrous adaxially, densely whit- ish lanate abaxially. Capitula heterogamous, up to 10, arranged in primary terminal, sessile corymbs becoming lateral later in development. /nvolucre broadly campanulate. 38, dark toward middle; outer bracts ovate, obtuse, 4 Involucral bracts ca. mm long, 2 mm wide; inner bracts 6 mm long, 0.5 mm wide. Florets ca. 84, pistillate much more numerous than the functionally staminate. Pistil- late florets ca. 70; corollas 3.5 mm long; achenes elliptic, 0.7 mm long, 0.3 mm wide, pubescent with short, apically obtuse, duplex hairs. Functionally staminate florets ca. 5-3.8 mm long; style branches truncate, 0.4 mm long; anthers 2 mm long with tails slightly exceeding anthero- podium; ovary sterile, oblong, flattened, 0.5 mm 4; corollas 3. long, 0.2 mm wide, pubescence as in pistillate florets. Distribution. | Known only from the type lo- cality, Sierra Madre, Mexico. No flowering data are available. Chionolaena seemannii is recognized by its cor- ymbosely arranged, broadly campanulate capitula, long linear leaves, and subdichotomously branched few, corymbosely arranged capitula. stems. as FIGURE 19. Chionolaena sartorti Klatt. — А. Habit. — B, C. Lea | surface (left), abaxial ight) D-G. — bracts. . Functionally staminate pn = ЊЕ istillate floret. — J. Stamen. — К. St ub of dur L. Style of pistillate floret. A-N, Реван акани ган seemannii (Sch. Bip. urfa i . Involucral . Pistillate floret. is stillat floret. — K. Style ab incl staminate flore staminate floret. Below: Mexican distribution. A FIGURE 21. (left), abaxial surface ior Stamen. Style of pistillate N. Ovary of fare бапа staminate floret. E132 (S). staminate flore . Below: Mexican distribution. ; 20. . Involucral brac "ers L. St le of nn Achene of pistillate floret. 308 (C in Seemann) Freire Chionolaena P ipis Baker in а t). —D-( N. Ovary of functionally staminate ). A. Portion of plant. — B. Leaf abaxial ; Punctionally staminate floret. — I. Stamen.—J. Style of Achene of pistillate floret. — M. Ovary of functionally ) M, ln 1994 (K Portion of plant. — B, C. Leaves, adaxial surface H. Pistillate floret. —I. Functionally staminate floret. — J. tionally staminate floret e of pistillate floret. O. Summit of pistillate corolla. Below: Brazilian distribution. A-O, Dusén Volume 80, Number 2 Freire 1993 Revision of Chionolaena J A3cm L J B-C 1cm o J D-J 2 mm О I K-N 1mm 434 Annals of the Missouri Botanical Garden I OAM - SS a eee гы ra ТО E ар M ee a, — ~ G-H 2mm |-Ј 1mm J L : J Volume 80, Number 2 Freire 1993 Revision of Chionolaena 435 EGE D | | Annals of the Missouri Botanical Garden 17. p ape wittigiana Baker in Martius, . 6: 129. 1882. TYPE: Minas erais: in cacumine Serra de ltatiaia, Glaziou 590 (lectotype, selected here, K; isolecto- d C). Figure 21. Brazil. Woody shrub with ascendent, erect, and densely leafy stems, 15-30 cm high. Leaves linear or linear elliptic, stiff, 2-2.5 cm long, 1.5- i slightly revolute margins, adaxially arachnoid pu- З mm wide with bescent, densely lanate and whitish abaxially. Ca- pitula heterogamous, numerous, arranged in dense corymbs, sessile and terminal on the upper branch- es. Involucre obdeltate. Involucral bracts 20-24; outer bracts linear ovate, acuminate, 4—5 mm long, | mm wide; inner bracts 5-6 mm long, 0.5-0.8 mm wide. Florets 12-24, functionally staminate florets more numerous than pistillate florets. Pis- tillate florets 8—9; corollas 1.8-3 mm long; achenes oblong elliptic, 0.5-0.8 mm long, -0.4 mm wide, pubescent with long, apically acute, duplex hairs. Functionally staminate florets 12—15; со- rollas 1.5-2.2 mm long; style branches truncate, 0.25-0.30 mm long; anthers 1-1.8 mm long with tails equaling antheropodium; ovaries sterile, flat- tened or contracted toward middle, 0.4—0.6 mm long, 0.3 mm wide, pubescence as in pistillate florets. Distribution. | Restricted to the state of Minas Gerais in southeastern Brazil. This species is found on stony and rocky mountain slopes between 2,400 and 2,600 m. Flowering collections were made in October and November. Chionolaena wittigiana is recognized by its lin- ear, somewhat stiff leaves and the heterogamous capitula arranged in compact corymbs. Additional specimens examined. BRAZIL. MINAS GERAIS: Itatiaia, 9 July 1872, Glaziou 5896 (paratype с а laena E >) Serra do Itatiaia, ca. 2, ‘in api ibus,” 19 Oct. 1903, Dusén S: 32 (5); p. pen has Negras, 2, #00: т, ee m . 1938, Markgraf & cas 5704 (herb. No. 6, NY: Agulhas = a do Itatiaia, 30 jua in Ule 3733 (HBG DUBIOUS SPECIES Chionolaena longifolia Baker in Martius, Fl. Bras. 6(3): 131. 1882. I have not found specimens of C. longifolia in any herbaria where I have studied. Baker 131) cited only one collection of C. longifolia (Sellow 2114). The type was probably destroyed. There is a photograph distributed by the Field Museum of Natural History (No. 15146) as the type for C. longifolia. The photograph agrees with Baker's description; however, the label on the photo reads: “Sello 853" instead of Sello 21 14. It there- fore must be excluded from the original material. LITERATURE CITED АМРЕКВЕКС, А. 1989. Phylogeny and rec care of the ше Inuleae (Asteraceae). Canad. J. Bot. 2211-22 29 А ———— & S. Freire. 1989. A transfer of two species from Anaphalis DC. to Chionolaena DC. (Astera- ceae, Mr o ae). Notes Roy. Bot. Gard. Edinburgh 46( ANDERSSON, L Es 7. The genus /schnosiphon (Mar- pa Opera Bot. 43: 1-114. 985. Revision of Heliconia subgen. Sten- ос Hams saa Heliconioideae). Opera Bot. 82: 1-124. 1986. Revision of aiu um Maranta (Marantac 7 Nordic J. 129-156. BAKER, J. С. 1882. ia ae. 5 ле In: C. Е. Martius, Monachii. Flora Brasiliensis 6(3): 117, 128- ә BLAKE, S. К. 1935. New Asteraceae from the United States, Mexico, and South America. J. Wash. Acad. Sci. 25(7): 311-325. BREMER, K. 1987. Tribal Ts pon of the As- teraceae. Cladistics 3: CABRERA, A. L. 1961. bbe: aciones sobre las Inuleae- Gnaphalineae (Compositae) de América del Sur. Bol. So и Bot. 9: 359-386. . WILLINK. 1973. Biogeografia de América Latina. Programa Regional de Desarrollo Cientifico y Tecnológico. OFA, ү NA D.C. Serie Biolo- gia, Monografia No. 13 dr ee A. P. DE es In: Prodromus X uralis Regni Vegetabilis, Volume CONANT, "DS 83. A revision of the genus Alsophila КО in the Americas. J. Arnold Arbor. 78 382 iion. M. О. & A. SAGASTEGUI ALVA. 1990. Chaar dra Less. revisited and the need for a new genus Pseudoligandra (Asteraceae: Inuleae). Taxon 39: 2 125-128. EPLING, C. 1940. The distribution of the American = = М Proc. 571- Sixth Pacific. Sci. Congr. 4: Pss if S. 1988. Hennig86. Version 1.5. Documen- tatio The retention index and the rescaled consistency index. Cladistics 5: 417-419 FREIRE, S. E. 1986. Revision del genero Luc dú (Com- positae, Inuleae). Darwiniana 27: 431-490. GARDNER, G. 18 Descriptions of four new genera of plants from the Organ Mountains. London. J. Bot. : 10 GLAZIOU, A. F. 1910. Plantae сера Sex a Glaziou Y Mém. Soc. Bot. Fra 40 HaRLING, C. 5 ни of de Суи еае. Acta Horti Berg. s : HemsLey, W. B. 1879-1888. poule X F. D. Godwin & O. Salvin, Biologia Centrali-Americana, 5 volumes R. H. Porter, London Hennic, W. 1966. Phylogenetic Systematics. Illinois Press, Urbana. HESs, Vergleichende Untersuchungen uber die 2 oa der Compositen. Bot. Jahrb. Syst. 68: 435-496, tab. Univ. Volume 80, Number 2 1993 Freire 437 Revision of Chionolaena Jackson, B. D. 1893. oe Kewensis. An Enumeration I Genera and Species of Flowering Plants. Fasc. 516. звани Press, Oxford. T: D. C. D. DE & E. LoNcPRE. 1963. Chromo- пе Studies in Mexican Compositae. Rhodora 65: 0. D A 3 A taxonomic revision of the Amer- ican species of Agarista (Ericaceae). J. Arnold Arbor. 65: 255-342. Krarr, F. W. Beitrage zur Kenntniss der Com- positen. Leopoldina 23: 88-91. LUNDGREN, J. 1972. Revision of the genus oo Gaertn. (Compositae). Opera Bot. 31: 1- Maas, P. J. M. 1972. Costoideae. Flora жо 8. New York. MERXMÜLLER, H., P. LeiNs & H. RokssteR. 1977. In- uleae — systematic review. Chapter 21 in V. H. He wood et al. (editors), The Biology and Chemistry of he Compositae. Academic Press, New Yor у N. PLATNIC унс апа Biogeography, о and Vicariance. Columbia Univ. Press, New PLATNICK, М. 1979 Philosophy and the transformation of cladistics. Syst. Zool. 37-546 Ramayya, N. 1962. unti on ‘the tric ones of some Senin I. General structure. Bull. Bot. Surv. ndia i 77-18 ROBINSON, H. 984. New species of Chionolae de cm Stenoc "ine Ран Brazil (Inuleae: Asteraceae). 1965. Relaciones geográficas y posibles origenes de la Flora de México. Bol. Soc. Bot. México 29: 121-177. STEYERMARK, J. 1984. In: Flora de la Guayana Ve nezolana. I. Nuevos taxa Е la а Жыланды: Асїа Bot. Venez. 14(3): 26 TRYON, А A d | the fern Pul Ja- mesonia. Contr. Gray. Herb. 191: 109-20 197 А Wir жы of o Е genus Er- sorus. Contr. Gray . 200: 74. Каток WHEELER. Dur “The aed ethod of phylogeny reconstruction. Syst. Zoo 2 & ©: Sans. Н. 1888. Jn: Itinera Principium S. Coburgi. Die botanische Ausbeute von den Reisen ihrer Ho- heiten der Prinzen von Sachsen-Coburg-Gotha. I. Reise der Prinzen phun und August um die Welt (1872- 73). II. Reise der Prinzen salen und Ferdinand nach Brasilien (1879). v. Fernsee und G. von Beck. Druck = аш уоп Carl Gerold's Sohn, Wi WiLeY, E. О. 1981. lts The Theory and Practice of Phylogenetic Systematics. Wiley, Ne rk. INDEX OF SCIENTIFIC NAMES Synonyms are in italic New names and combinations are in \ boldface. Achyrocline capitata Baker Anaphalis aecidiocephala Grierson Chionolaena DC. Bécidiocepruna (Grierson) Anderberg & Freire arbuscula DC. мао DC. var. hololeuca Baker ex Glaz. breweri Steyerm capitata (Baker in un Freire columbiana S. F. concinna (A. Gray) vo В & Freire corymbosa Hemsle chrysocoma = ) Freire enth. ook. f. ex т lavandulifolia (Kunth) ob & Hook. f. ex B. D. Jackson longifolia Baker || ychnophorioides Sch.-Bip. mexicana Freire phylicoides (Gardner) Baker sartorii Klatt seemannii (Sch.-Bip. in Seemann) Freire wittigiana Baker Elichrysum |" Kunth lavandulaefolium Gnaphalium aecidioc ae (Grierson) L. O. Williams concinnum А. Gray eleagnoides (Klatt) S. F. Blake hypochaionaeum Seh, -Bip. ex Klatt lavandulaceum lavandulaefolium (Kunth) S sartorii Sch.-Bip. ex Kla sartorii (Klatt) F. J. Espinosa Garcia seemannii Sch.-Bip. . F. Blake phylic oides Gardne Oligandra и Wedd. NUMERICAL LisT OF TAXA Chionolaena (1) C. aecidioc ephala (Grierson) Anderberg & Freire (2) C. arbuscula (3) C. capitata (Haker j in oe Freire (4) C. columbiana S. F. Blake (5) C. concinna (A. a Freire (6) C. chrysocoma (Wedd.) Freire (7) С. eleagnoides Klatt (8) C. vir dida Baker (9) С. jeflreyi Н. dicen (10) C. latifolia (B . (11) C. lavandulifolia ( ‚к Benth. & Hook. f. ex B. D. Jacksor (12) C. iychnophorioides Sch.-Bip. (13) C. mexicana Freire (14) C. phylicoides (Gardner) Baker (15) C. запоги Klatt (16) C. seemannii (Sch.-Bip.) Freire (17) C. wittigiana Baker List oF EXSICCATAE The figures in parentheses refer to the species number. Numbers correspond to those given in the numerical list of taxa 438 Annals of the Missouri Botanical Garden Balls, B5139 (11). Barclay & Juajibioy, 6927 (4). Nogueira, 30 (3). Luederwaldt, 16733 pro parte (8); Beaman, 1961 (11); 2145 (11); 3511 (11). Brade, A C., 757 (8); 6385 (8); 12659 (3); 12733 (8); 15596 (8); 15597 (10); ПР 16517 (14); 16898 (3); 17004 (14); 17012 (14); 17013 (3); 17421 (8); 20282 (8); 21260 (8) A. L., 12227 pe Cabrera, A. L. & H. Fabris, 14212 (3). Camerik, 29 (8). Campos Porto, 1218 (2). Carriker, 15 (4). Chase, 9389 (12); 9708 (3). Claus- sen, year 1839 (2). Cuatrecasas & Castaneda, 25115 (4) c = Damazio, L., 9 (2). Dionisio & Octavio, 28 June аЙ 0); 468 (3); 542 (8); 2007 (10); 2132 (1 Dx s.n. year 19..5870128 (3 Eiten, С. & L. T. Eiten, 6553 (10). Equipe do Projecto Serra dos Orgàos, 1 Nov. 1970 (14). Farney et al., 800 (14). 4840 (8); 4841 (10); ); 5896 (17); 5899 (3); 5 3); 6601 (8); 7723 (3); 87 68 (3): 8769 (8): 11036 (2); 14959 (2). 14960 (12); 14960a им 14961 О, n (14). Gounelle, Feb. 1899 (8). Grant, M. L., 4 (4). Guchioni, 940 (8). Handro, O., 9423 (3) Harley, R. M., et al., 19677 (9). потеза 632 (8); 633 (3); 648 (10). Holm, R. W. & H. H. Шв, 573 (11). Horn, 38 Шш ). rwin et al., 29023 Ars d . Å., et En 90 Kin ОВ. М., є 814 1 (9). Kirkbride, Н. & E EE. 1803 (4). Krieger & Urbano, 8649 (2). Kupper, Ga tenet 5772 (14). Glaziou, ( : 4851 (1 0 ( 29310 (2). Пи 1148 (11); 1301 (11), 1301a (11 Liebman, 308 (15); 316 (7). Шезпег, R. L., S 987 (10). Lindeman & Haas, 4134 (8); 5197 (8). Lua & 16733 pro parte (10). Luetzelburg, 6252a, b (14); 6386 4). ~ -Ó MacDougall, 4129 (1), 6 Jan. 1961 (1). Macedo, A., 2801 (12). Magalhaes, 1696 (12). 65587 (10). Markgraf, 10102 (14). 3702 (10); 5704 (17). Martii, 829 (2). Martinelli, (3). Martinelli & Simonis, 9051 (14). Martinelli, al., 10843 (8). Martius, 843 (2). Merxmüller & Hei. 25605 (8). kes Ames, 26 Dec. 1915 (8); Oersted, 10570 (11). Pabst, 3473 (2). Parry & Palmer, 423 (5). Pereira, E., 208 (14); 2637 (2). Pittier, 3430 (11). ш 4223 (11); 7700 (13). Purpus, 1523 (11); 2990 Rangel, O . Cleef, 934 (6). Rangel et on (4). Rizzini, 5 June 1951 (14); 368 (14). isn y Castañeda, 7356 (4). Ross, 1265 (11). Roth, P. L., 1401 26 Dec. 1916 (10). , 809 (6). Schaffner, 222 (5). Shenell, 8432 (14), ERN ij "E "Brasilia" (2). Seemann, 1994 (16). Seifriz, W., 494 (4). Bye 1682 (10). pq J. ¿ Е. De ee 129265 (10). Steyermark, J. & B. Holst, 130822 (10). Steyermark, J., et al., 126074 (10); 126322 (10). Sucre, 5765 pur a & Krieger, 6836 d A. Tamandaré, F., e Татапдаге, F. Brade, 6383 (3). d l (11 Ule, E., 2601 (12); os (2); 3374 (8); 3733 (17); 3913 (3) as M., m 1937 (14). 298 (2). Vianna, 197 (8). Voll, O., 2 Weston, 155 — H., 396 (3); 420 (8). (11) mun 10122 (11). A REVISION OF MALVAVISCUS (MALVACEAE)! Billie L. Meghan С. Mendenhall’ Turner? and ABSTRACT o reevaluation of herbarium srg of the genus Malvaviscus yields two widespread species, M. arboreus of North America ап . concinnus о America, two loc alized species, M. achanioides of Mexico and M. williamsii of Peru and Colombia, and a cien! cultivar of unknown origin, M. penduliflorus. In spite of recent attempts to delimit and classify its many specific and subspecific taxa, Malvaviscus (Malvaceae) remains enigmatic. The genus is highly variable morphologically and given to populational forms. Furthermore, these forms intergrade pro- ducing an array of character combinations. Ап extreme taxonomic treatment of the genus would recognize a horde of intergrading forms. Indeed, this extreme infrageneric variation has led to the proposal of over 50 specific names within Mal- vaviscus, though Schery (1942) recognized only three in his monograph of the genus. Because of the common intergradation and re- combination of character states, few characters within Malvaviscus have proven useful for specific recognition. Modern interpretations of the genus have generally recognized about a dozen overlap- ping taxa. However, we believe that the over- whelming majority of the species proposed under Malvaviscus represent only two biologically sig- nificant taxa, M. arboreus in North America and M. concinnus in South America. HISTORY OF THE GENUS The genus Malvaviscus was established almost 200 years prior to its monograph by Schery (1942). Among the first generic names to include Mal- vaviscus were Alcea, Althaea, and Malva. Lin- naeus did not recognize the genus and placed its species in Hibiscus in 1753. In 1759, Fabricius distinguished Malvaviscus based on a single spe- cies, i.e., Hibiscus malvaviscus L. (A discussion of the typification of Malvaviscus can be found in Taxon 17: 87, 1968.) In 1788, Swartz proposed to rename the genus as Асћата. In 1824, A. P. de Candolle divided Malvaviscus into two sections: Achania and Anotea. Anotea was raised to generic rank by Kunth in 1846, and Achania is now recognized as synonymous with Malvaviscus. Generic treatments have not been in agreement. Schery's monograph (1942) included a complex of 11 varieties of M. arboreus as well as two additional species that have since been transferred to Pa- vonia. Other treatments, which have recognized up to a dozen species, occur in floras throughout the range of Malvaviscus (Robyns, 1966; Stand- ey, 1923; Standley & Steyermark, 1949). How- ever, the only significant treatment of the genus since Schery has been th plary work of Fryxell (1988) in his monograph of the Malvaceae of Mex- ico. In his discussion of the taxonomic problems within Malvaviscus, Fryxell acknowledged the dif- “There is a certain sameness of morphology that runs through ficulty of clearly delimiting species: the genus, and clear-cut differentiating characters are lacking." Fryxell recognized six species of Mal- vaviscus in Mexico, three of which are maintained here. CHROMOSOME NUMBERS The first report of a chromosome count for Malvaviscus was presented by Skovsted in 1935 (Table 1). He reported a count of 2n — ca. 84 for an unidentified cultivated species in Kew Gardens, presumably obtained originally “from gardens on the Gold Coast." We take the species to be penduliflorus of the present treatment since Skovsted noted: Morphologically, the type under observation is distinctly different from the wild growing M. arboreus Cav. Its ' We are grateful for the encouragement and idis comments our esteemed colleague, Paul Fryxell, throughout the course of this study. Our к, is based u at the following institutions: BRG (3), F NY (492), RB (6), SP (9), TEX (170), US (381). ? Plant Resources Center, Department of Bot ANN. of approximately 3,200 specimens on deposit n the (701), herk. Toni (2), cH (431), HUA (7), LL (200) МС (6), МО (783), ancy Webber ug the illustration. tany, Universi * Department of Botany, "University of Texas, Austin, Texas 78713, U. 5. ity of Texas, Austin, Texas 78713, U.S.A. А. Missouni Bor. Garb. 80: 439-457. 1993. 440 Annals of the Missouri Botanical Garden TABLE 1. Chromosome numbers in Malvaviscus. Taxon 2n number Voucher /reference M. arboreus var. arboreus re- ca. 56 Mexico. Veracruz: Fryxell & Bates 940; Bates (1976) ported as var. mexicanus M. arboreus var. arboreus re- ca. 56 Mexico. Veracruz: Fryxell et al. 1676; Bates (1976) ported as var. mexicanus M. arboreus var. arboreus re- ca. 28 Mexico. Jalisco: Fryxell, Bates & Blanchard 1575; Bates ported as M. penduliflorus (1976 M. arboreus var. drummondii ca. 28 U.S.A. Texas: Travis Co., Mendenhall 485; reported here M. penduliflorus reported as ca. 84 Kew Gardens, where cultivated; Skovsted (1935) | 5р M. penduliflorus 86 India. Karnataka: where cultivated; Krishnappa & Munirajappa cytological behaviour indicates that it is probab ly a hybrid although its origin appears unknown. It is com- en cultivated in Trinidad where it i = vegetatively reproduced as it is apparently il sterile. SPECIES CONCEPT In this treatment, species concepts in Malvavis- cus are largely those of the senior author, the junior Krishnappa & Munirajappa (1982), however, ^ author being new at the herbarium bench. Because reported M. penduliflorus to have a count of 2n fieldwork on the genus was limited, species concepts = 86, presumably а miscount of 2n = 84, ог аге based on morphogeographical considerations possibly an aneuploid clonal derivative of what and our experience with species recognition in oth- would seem to be its original number, 2n — 84, er groups (e.g., Asteraceae, Fabaceae), both in the since other taxa in Malvaviscus appear to have field and as hypothesized by herbarium evaluations. an ancestral base number of x = 14, if not x = We believe our specific delimitations in Malvav- 7. Bates (1976) listed a count of 2n = ca. 28 for iscus are populational in nature and are compa- M. penduliflorus, but we believe the voucher con- rable to what most workers refer to as good species. cerned is more likely а morphological variant of | We do not claim this to be true for M. penduli- the widespread, highly variable M. arboreus var. — florus, for we suspect that the plants referable to arboreus, as conceived of here. this species are relatively uniform, mostly sterile Our report of 2n = ca. 14 pairs for M. arboreus cultivars, developed very early on by accidental or var. drummondii is to some extent biased because synthetic means. We have recognized M. pendu- of the report of 2n = 28 by previous workers. In liflorus as a truth, the count would have been any number naturally in the wild except as it has persisted or “ee synthetic species," not occurring between 12 and 15 pairs, the chromosomes being escaped cultivation by vegetative means, althoug very small and not forming neat bivalents. it is possible that occasional hybrids between M. In summary, only two taxa of Malvaviscus have penduliflorus and M. arboreus might occur, as been counted with reasonable certainty to date: M. noted under the discussion that follows each. arboreus var. arboreus with 2n — ca. 28 and ca. Our species concepts are similar to those of 56, and the cultivated M. penduliflorus with counts previous revisional workers on Malvaviscus, name- of 2n = ca. 84 and 86 (Table 1). The base number ly Schery (1942) and Fryxell (1988). In spite of of the genus appears to be x = 14, if not 7. The the plethora of pressed specimens available for latter base would agree with that found in the large, study, extraordinary variation, and widespread dis- closely related genus Pavonia, where some 35 ог tribution of Malvaviscus, Schery recognized a sin- more species have been counted, all on a base of gle species with eleven varieties while Fryxell rec- x = 7, ranging from n = 14 to n = 56 (although ognized six species in Mexico, which suggests that some authors have reported occasional counts of Һе would likely have recognized less than ten taxa both n = 13 for a species that otherwise has been for the remainder of its distribution. In short, our counted as n = 14 (e.g., Pavonia zeylanica (L.) recognition of five species for Malvaviscus (in- Cav. with 2n = 52 (Krishnappa & Munirajappa, cluding M. penduliflorus) is conservative, standing 1980) vs. 2n = 56 (Dasgupta & Bhatt, 1976, between the evaluations rendered by Schery and 1981, 1982)). Fryxell. Except for M. penduliflorus, we believe Volume 80, Number 2 1993 Turner & Mendenhall 441 Revision of Malvaviscus the taxa that we have recognized are comparable in nature and constitute populational or biological units that can be verified by future field workers. TAXONOMIC TREATMENT Malvaviscus Fabr., Enum. 155. 1759. ТҮРЕ: Hi- iscus malvaviscus L. [= Malvaviscus ar- boreus Cav.]. See bau (1966) for typifi- cation of Malvaviscus. Achania pilosa Achania Sw., Prodr. 102. 1788. TYPE: Sw Erect or clambering (vinelike) shrubs or small trees, pubescent or glabrous. Leaves petiolate, the blades linear-lanceolate to broadly ovate-cordate, unlobed to 3—5-lobed, serrate to subentire; stipules subulate, deciduous. Flowers solitary or several in the leaf axils or sometimes in apical cymes; bract- lets of the involucel linear to obovate (rarely broad- ly ovate), usually (5-)8-9; calyx campanulate or tubular, 5-lobed; petals 5, red (rarely white), asym- metrically obovate-cuneate, auriculate toward the base, forming a tubular corolla; androecium usually exserted, the staminal column with 5 apical teeth, the filaments short and + retrorse; styles 10 with capitate stigmas. Fruit a fleshy dique н red (rarely white), with 5 carpels, each Base chromosome number, x = 14. Diagnosis and relationships. Malvaviscus is a member of the tribe Malvavisceae C. Presl, Reliq. Haenk. 2: 135, 1835 (as “Malvaviscaceae””) by virtue of its five uniovulate carpels, schizocarpic fruit, ten free styles, and apically 5-toothed sta- minal column. Distinguishing generic characters of Malvaviscus are its auriculate petal and red or rarely white baccate fruit. The petal auricle is the basal lobe that gives each petal its characteristic mitten shape. Apical lobes are called simply lobes. The genera thought to be closest to Malvaviscus are Pavonia, Lopimia, and Anotea. The largest of these, Pavonia, has dry fruits. Lopimia and Anotea are small genera that have blue-black fruits (dry in Lopimia) and lack the auriculate petal. Although Malvaviscus can be found in gardens throughout the tropical and subtropical world, it is native to the Western Hemisphere. Its range ex- tends from Peru and northern Brazil to the southern United States and the West Indies. Although in- dividuals and populations are highly variable, es- pecially in Mexico and Central America, we rec- ognize two major taxa, M. arboreus in the north and M. concinnus in the south. A third, relatively localized species of northwesternmost South Amer- ica, M. williamsii, is recognized but clearly belongs to the M. concinnus complex. Were the characters that mark it not so easily identified, and if it had not already been treated at the specific level, we would probably have recognized this as an inter- grading variety of the latter. Finally, for practi- cality, we maintain two species that would otherwise be absorbed by M. arboreus. These are M. pen- duliflorus, a distinctive hybrid cultivar, and M. achanioides, an intermediate form between M. arboreus and M. concinnus. KEY TO SPECIES OF MALVAVISCUS la. Leaves mostly 15-35 cm long, the blades usu- ally strongly cordate; vestiture of peduncles 5- mm across; South Amer ama, Costa Rica, and Gulf slopes of Honduras and Mexico (Chiapas, ат Verac 2a. Calyces mostly 18-20 mm кч vestiture of а d petioles densely pilose ijs hai - m long; Honduras and Xico . 3. M. oe 2b. Calyces mostly 20- 30n m long; vestitu of velutinous hairs ан 0.2-1.5 mm long; South America, Panama, and Costa ica 3a. Involucellar еа broadly elliptical- ovate, wi at or near the middle, the margins strongly — Peru and Co M. williamsii 3b. Bibi. [hee fos to linear lanceolate, broadest at or near the base; South Ali Panama, and Costa Rica . .4. M. concinnus lb. Leaves mostly 5-20 cm n long, if longer the 5-70 mm long; calyces mostly 8-18(-20) mm long; fruits, when present, mostly 8-14 mm across; North America and West Indies or else cultivated. 4a. Cultivated, mostly A POE producing fruits; flowers pendulou ly 42-70 mm long; Вала ovate and sub- glabrous; widespread in gardens, and often escaping or persisting following cultivation 2. M. penduliflorus 4b. Wild and cultivated, usually fertile and white fruits; flowers usu- and rarely coastal areas of northern South America, OMNE cultivated and per- sisting elsew 5a. Leaves ies uniformly 3-lobed and form a velvety vestiture only rarely exceeding 0.25 mm high; 442 Annals of the Missouri Botanical Garden mostly United States, Ln о ех- tending into Mexico the Gulf ЫШЫ regions of Tamau Шз. cc. arboreus var. dru i л c p oO E < o un < 5 = А о = EZ = = Е Lm = & = & gt Ф un = un = DE iind овом ћете 3 in cultivation Em arboreus var. arboreus 1. Malvaviscus arboreus Cav., Diss. 3: 131. T t. 48 la. Malvaviscus arboreus var. arboreus Malvaviscus arboreus Cav., Diss. 3: 131. t. 48. f 1787. Hibiscus malvaviscus L., Sp. Pl. 694. 1753. Malvaviscus coccineus Medikus, Malvenfam. 49. 1 787, nom. superfl. Achania malvaviscus (L.) Sw., Prodr. 102. Achania coccinea Salisb. Prodr. 385. 1796, nom. superfl. Malvaviscus malvaviscus (L.) Millsp., Publ. Field Columbian Mus., Bot. Ser. 2: 73. 1900. TYPE: “Hibiscus M em us" (holotype, LINN-875.22). Achania mollis Aiton, Hort Kew. 2: 459. 1789. Mal- vaviscus mollis (Aiton) DC., Prodr. 1: 445. 1824. TYPE Mexico (holotype not seen, eren ad- ional information according to Fryxell, 1 Prodr. 102. 17 E Prodr. 1: 445. pilosus (Sw. Fane ett & нй Fl. жек 5: : 37. 1926. ТҮРЕ: Jamaica. Swartz s.n. (holotype, S not seen, according to m een 1988 у. е spiralis Cav., Icon. 5: 20. t. 434. 1799. Mal- viscus ciliatus DC., Prodr. 1: 445. 1824, nom. ей . Achania ciliata Spreng. Syst, Veg. 3: 100. 1826, nom. superfl. TYPE у Par зата. Taboga ‚ bland, Nee s: n, о МА пої зееп). nch, Meth. Supa, 208. 1802. TYPE: no ids t pre un to Stafleu & Cow an (1981 Malvaviscus МИ РЕН Kunth, Nov. Gen. Sp. 5: 288 [folio ed. p. 224]. 1822. TYPE: Mexico. Guerrero: near Acapulco, Humboldt & Bonpland, 1803 (ho- lotype, P-HBK not seen, according to Fryxell, 1988). Malvaviscus grandiflorus Kunth, Nov. Gen. Sp. 5: 286 hs a ed. p. 223]. 1822. TYPE: Mexico. Guanajuato: pe Guanajuato, Humboldt £ Bonpland, 1803 hobbie P-HBK not seen; fragment holotype, F). Malvaviscus pentacarpus DC., Prod 1824. cones Florae Mexicana no. 124 (To Collection acc. no. 631.1093, Hunt Institute). (Cf. photo F-30507 of G a (not the type), according to Fryxell, 1988.) Malvaviscus arboreus var. cubensis Schldl., Linnaea 11: 360. 1837. Malvaviscus arboreus subsp. cubensis o ise Folia Geobot. Phytotax. Praha 5 TYPE: Cu ) dus dendi “LZ des ee isotype, W? not seen, ac- cording to Fryxell, 1988). Pavonia urtic aefolia 1835. Malvaviscus urticifoli oliu esl, Relig. Haenk. 12 s (C. Presl) Frysell Syst. Bot. 12: 279. 1987. TYPE: Mexico. “In terris occidentalibus" (probably in the state of Guerrero or diano | uring Haenke's 8 dius ME to Mexico and return durin 1), ains s.n. reed, designated by Fryxell ioe ind 902A not seen; isolectotype, PR-197902B 101 seen n). Malvao: iscus sepium Schldl., Linnaea 11: 361. 1837. Malvaviscus arboreus var. p ME p Ann. dod od Gard. 226 . TYPE Mexico. Ver near Кыры, June ү коле, 480 (е Ban troyed; photoholotype, F, E Malvaviscus arboreus var. mexicanus Schldl., Linnaea 11: 359. 1837. i Ve- racruz, Schiede s.n. etapa, designated by Fryxell (1988), HAL not seen). а brevi ipes Bend. Bot. Voy. Sulphur 68. 1844. YPE: ica. Nicoya, Barclay s.n. (holotype, 88). no Malvaviscus sagraeanus А. Rich., Hist. Phys. Cuba, PI. Vasc. 131. t. 14. 1845 [1841]. Malvaviscus ar- boreus E а (A. Rich.) Е. С. Baker, Ј. Bot. 37: 899, TYPE: Cuba, de la Sagra s.n. (holotype, E not seen; isotype, K not seen, ac- cording to Fryxell, 1988). Mia o A. Rich., Hist. Phys. Cuba, Pl. c. 1: 845 [1841]. TYPE: Cuba. de la Sagra s.n. (specimen unknown, according to Fryxell, 1988). Malvaviscus arboreus var. p Griseb., Fl. Brit. . 83. 1859. ТҮРЕ: Jamai ‚ Without locality, without date, Wullsc beber 768 (holotype, GOE not seen, according to Fryxell, 1988). Malvaviscus nus Pittier & J. D. Smith, Bot. Gaz. (Cra an devi ille) 23: 238. 1897. Malvaviscus ar- e var. palmanus (Pittier & J. D. Smith) pai . Missouri Bot. Gard. 29: 222. жеш Cos sta Rica. San Jo "s forests of La Pal `5, 100 ft., July 1895, Tonduz 97 12 (holotype, CR: not seen; isotype, US; photoisotype, MO). a ato arboreus var. sloanei E. G. Baker, J . 1899. TYPE: Jamaica. Sl v. Е. ч (holotype, BM not seen, according to Fryxell, 1 Malvaviscus и E. С. Baker, J. Bot. 37: 347. : Belize. e ed 27 Dec. d J. ^o udi 34, „т ВМ пої зееп, according to Fryxell, 1 И a lanceolatus Rose, Gon US uae о 5: 175. 1899. TYPE: Chia r Ci- ains 12-15 Feb. 1896, “Nelson 3807 ee sotypes, GH—2 sheets). Wiican arboreus var. grisebachii E. G. Baker, J. Bot . 1899. TYPE: Jamaica. (Based on M. arboreus var. a Grisebach, which i get! applies to a vestiture form of M. arboreus var. arboreus.) Grisebach cited no specimen; reiha did Baker. Any еы, if designation is needed, would presumably reside at BM or K. Malvaviscus "polus E. G. Baker, J. Bot. 37: 34 899. : Costa Rica. Without specific loc ality, Volume 80, Number 2 1993 Turner & Mendenhall Revision of Malvaviscus 443 е date, Polakowsky 197 (holotype, BM not йлы се ко mottii Millsp., Publ. Field Columbian us t. Ser. 2: 73. 1900. TYPE: Cayman Islands. Dus Brae. above Spot Bay, 8 Feb. 1899, Mills- pue 1166 (holotype, F). Malvaviscus rivularis Brandegee, Zoe 5: 211. YPE: Mexico. Sinaloa: near Culiacán, banks of the Cofradia River, en 1904 (holotype, UC not seen; isotype, US). Malvaviscus conzattii Greenman, Field Mus. Nat. eis Bot. Ser. 2: 333. 1912. TYPE: ip Oaxaca: Pablo, Huitzo, 1,600 m, 25 quce ps 1981 (holotype, F; isotype, K no Malvaviscus oaxacan us Standley, Du U ©. Natl. Herb. кы TYPE: Mexico. Oaxaca: N of Tux- . 1894, Nelson 348 (holotype, US). Malvaviscus hintonii Bullock, Kew Bull. 1937: 291. 937. Malvaviscus arboreus var. hintonii (Bullock) Schery, Ann. Missouri Bot. Gard. 29: 217. 1942. TYPE: Mexico. Mexico State: Distr. Temascaltepec, Tejupilco, 15 May 1933, Hinton 3928 (holotype, K not seen; isotypes, A, NY). Malvaviscus arboreus var. brihondus Schery, Ann. Mis- souri Bot. Gard. 29: 213. 1942. TYPE: Belize. Honey Camp, Sep. 1929, Lundell 480 (holotype, MO; iso- types, F, US). Malvaviscus arboreus var. lobatus Robyns, Ann. Mis- souri Bot. Gard. 52: 572. 1966. TYPE: Panama. Chiriqui: Fred Collins’ dip at edge of coffee plan- tation, 6,000 ft., 2 Aug. 1960, J. Ebinger 692 (holotype, MO; isotype, F). 1905. Erect or clambering shrubs 1-10 m high. Stems variously stellate-pubescent to glabrate, rarely gla- rous. Leaves mostly cm long, 3- wide, petioles mostly 1-12 cm long, the blades variously ovate to cordate, less often elliptical, mostly unlobed, but occasionally with shallow sub- terminal lobes, variously irregularly serrate, or rarely entire, pubescent with mostly stellate hairs, often densely so, but sometimes glabrous or nearly so. Calyces mostly 8-15(-18) mm long, variously pubescent, the hairs simple or stellate, rarely gla- brous, the subtending bracts mostly 5-8(-11), lin- ear to linear-oblanceolate. Corollas red, rarely white, mostly (15-)20-42(-50) mm long, the petals most- ly imbricate at anthesis, not normally flaring. Sta- minal column usually exserted for ca. 1⁄4 of its length, rarely included. Fruit usually bright red but Chro- mosome number, 2n = ca. 14 and 28 pairs. sometimes white, 8-14(-16) mm across. We interpret this widespread, highly variable species as having a single populationally meaningful variety, variety drummondii, as noted below. We include in variety arboreus most other material previously placed in M. arboreus from North America north of Panama, and that from the West Indies. This includes Malvaviscus arboreus var. mexicanus, recognized by Fryxell (1988), and oth- er taxa accepted by him as distinct, as noted in the synonymy above. Malvaviscus arboreus var. arboreus is perhaps the most variable species that the senior author has had the displeasure to work with. Detailed mapping of character states in Mex- ico and other regions reveals a hodge-podge of vari- ation that must relate to localized genetic parti- tioning, as well as environmental influences. Thus, while any given population is relatively uniform with respect to, say, vestiture or leaf shape, closely contiguous as well as distant populations may show markedly different vestiture types and leaf shapes. The same may be said with respect to corolla and calyx size, and their vestiture. A similar type of variation to that mentioned above is found throughout the West Indies and in Central America; all of this must be evident from the considerable synonymy and varietal names that have been proposed for M. arboreus. We also believe that at least some of the vari- ation found in M. arboreus must have arisen rel- atively recently by local hybridization with M. pen- duliflorus, a cultivar which, for convenience and because of its wide киин in gardens through- out the world, we have treated as a species. Evi- dence for this is mostly speculative (based largely on herbarium collections from areas where M. ar- boreus grows with or near M. penduliflorus) for, as noted by Fryxell (1988), M. penduliflorus ap- pears to be largely sterile (at least it does not readily set fruit). Nevertheless, occasional plants do de- velop fruits; indeed, in parts of southern Mexico and Central America, fruiting specimens assignable to that species seem to occur. The latter plants might be viewed as ancestral fruiting stocks of M. arboreus (we have treated these as such), from which the remarkably uniform M. penduliflorus developed. This is discussed in more detail under the latter Distribution (Figs. 1, 2). Mexico, West In- dies, and throughout most of Central America (oc- casionally cultivated elsewhere, as indicated below), 0-2,00 0 m; flowering all seasons. We include M. hintonii in synonymy under M. arboreus var. ar- boreus, although Fryxell (1988) maintained the species (as M. urticifolius), while Schery reduced it to varietal rank under M. arboreus. The name is applied to specimens with white corollas, which otherwise appear to have the characters of M. Diagnosis and relationships. arboreus. Indeed, it might reside within our con- cept of M. penduliflorus, except for its shorter corollas and pilose vestiture along the upper stems. One of the syntypes of M. hintonii (Hinton 7912, NY, US) has a notation by the collector that “the Annals of the Missouri Botanical Garden 7 m LIT] ТП [71 16 "5 3T E Jj T 1 11 TT C 33 C c (€ C C E +e * 4 C e + Л „“ 22 | j * 50 ha A 5 4" 222 ө т n 0 А Malvaviscus achanioides 2 Malvaviscus arborcus var. arboreus T NES Malvaviscus arboreus var. drummondii € Malvaviscus arboreus var. drummondii, cultivated : m 1; Malvaviscus concinnus @ Malvaviscus penduliflorus > > a FIGURE 1. arboreus var. drummondii, are marked with cultivation, or thought to sius escaped from саан fruit is very rare," suggesting that this might be a color form of M. penduliflorus, or perhaps a hybrid segregate between the latter and M. ar- boreus. For additional discussion of this problem see M. penduliflorus. Malvaviscus conzattii, because of its relatively large corollas and lanceolate leaves, was relegated to synonymy under M. penduliflorus by Schery (1942). Corolla size would position it in our concept of the latter, but its leaf blade shape (having cordate bases) suggests that it belongs to the M. arboreus complex, although it is possible that the type con- cerned is of hybrid origin, as noted above Malvaviscus oaxacanus and M. lanceolatus were both retained by Fryxell (1988), but we be- lieve these to be localized populational “calyx- forms" or “leaf-forms” of M. arboreus var. ar- boreus. Malvaviscus oaxacanus possesses a rather P dense, short, furfuraceous vestiture on the calyces, y Mal- vaviscus lanceolatus has lanceolate-elliptic, nearly otherwise it is very similar to M. arboreus. glabrous leaves (similar to those of M. penduliflo- rus) and nearly glabrous involucels, otherwise it is very similar to typical M. arboreus. Schery (1942) Natural кокту, of Malvaviscus in үө and Central America. Isolated collections of Malvaviscus e known to be cultivated, thought to have persisted after also included both of these names in synonymy under M. arboreus. S of the entire M. arboreus complex, we thought it possible to rec- ognize M. lanceolatus as distinct, but restricted In an initial “run-through” the name to those individuals and/or populations with strictly ovate-elliptic, nearly glabrous leaves and glabrous or nearly glabrous calyces. So con- strued, the "species" would have comprised (along with the type) the following specimens: MEXICO. CHIAPAS: Mpio. Union Juarez, between Union Jua- rez and Santo Domingo, 1,060 m, 6 Sep. 1980, Fryxell 3203 (MO, NY). ОАХАСА. ca. 8.4 mi. S of Putla de Guerrero, ca. 1,000 m, 16 Jan. 1979, Croat 45810 (MO); Putla, Vicente, ca. 1,200 m, MacDougall, 1970 (NY); Juquila, Lachao, Santa Rosa, ca. 1,200 m, MacDougall, 1971 (NY); Ya- eo, Arroyo San Pedro, 440 m, 23 Mar. 1938, Mexia 9204 (GH, MO, NY, 05); 5 slope of Sierra S of Lachao, 1,700 m, 5 Dec. 1962, Moran 10114 (US). Fryxell (1988), however, cited specimens from Nayarit, Jalisco, Morelos, and Puebla that we intended to include within our concept of M. ar- boreus var. arboreus, most of these having ovate Volume 80, Number 2 1993 Turner & Mendenhall 445 Revision of Malvaviscus 0 | 84^ 809 767 Ры FIGURE 2. closed circles) from the West Indies leaf blades and a vestiture reminiscent of the latter. Indeed, all of M. lanceolatus, sensu Fryxell, can be submerged within M. arboreus without creating much of a morphological ripple, for even as we intended to circumscribe M. lanceolatus, the taxon would have differed from M. arboreus by a com- bination of weakly differentiating characters, main- ly leaf shape (lanceolate-elliptic), essentially gla- brous stems and foliage, relatively large glabrous calyces, and longer corollas (mostly 40—50 mm). said **culti- The same can be with respect to the 9 var, largely sterile derivative from the M. M. penduliflorus, for it appears to be a arboreus complex centering about M. lanceolatus (sensu Fryxell). Indeed, all of the characters that pur- portedly distinguish M. penduliflorus and M. lan- ceolatus from M. arboreus crop up in one or more combinations throughout the range of M. arboreus, as treated here. For example, M. palmanus has been applied to populations from Central America having ovate-elliptic blades and other characters resembling those of M. lanceolatus. In eastern Panama and parts of Costa Rica, M. arboreus shows variation in the direction of M. Collections of Malvaviscus arboreus var. arboreus (small closed circles); and M. penduliflorus (large concinnus, especially in the size of calyx and fruit and in leaf shape and vestiture, and occasional specimens from this region have been somewhat arbitrarily positioned in one or the other species. A case might be made for the treatment of the mostly South American M. concinnus as a broad regional variety of M. arboreus, such as was done by Schery (1942) under the name Malvaviscus arboreus var. longifolius. Strangely, Robyns (1966), in his treatment of Malvaviscus for Pan- ama, mentioned none of this variation, relegating all of the Panamanian material that he examined to M. arboreus var. mexicanus, including the cul- tivated M. penduliflorus. scribed a new variety, M. arboreus var. lobatus, Nevertheless, he de- which is no more than a form of his M. arboreus var. mexicanus with apically lobed petals. Representative specimens s UNITED STATES. FLORIDA: Monroe Co., Bi e Key, hammock borders, 26 July 1981, back 97 08 (GH, MO, cul- tivated?). HAWAII: Maui ala, persistent along road- en 8 Aug. 1926, Degener 9840 (NY, cultivated?). OUISIANA. Orleans Parish, Orleans, 1 Nov. 1971, Ewan 23082 (NY, Er NORTH CAROLINA: New 446 Annals of the Missouri Botanical Garden Hanover Co. Imington, along railroad cut, 3 Sep. ley 75248 (F, MO). PETEN: Santa Elena, entre carretera Wi 1970, Leonard 3714 (GH, NY). MEXICO. CAMPECHE: Nov. 1931, Lundell 970 (А, Е, MO 27 Sep. O, US). HIDALGO: Valley of Tula, 8,3 , 10 Sep. 1899, Pringle 8232 (A, F, А . US). JALISCO: barranca near Gua- dalajara, 5,000 ft., 11 May 1901, Pringle 8498 (F, GH, MO, $); mountains N of Autlán, 1,500-1,650 m, 5 Oct. 1960, McVaugh 19917 (GH, 5 n c US). MEXICO: Temascaltepec, бо һагг 1934, Hinton 6719 (А, F, GH, LL, US). мс HOAC AN: пе са 1 Arsène m, l ug. 1 5494 (A, NY, n MORELOS: 8 km NW a uesta de Матије 4 Apr. 1962, valley, 1 Oct. 1894, е ее Сег 914, Агзепе 10058 (Е, MO). ho 3 of Tihosuco, 4 Aug. 1972, Н, MO). SAN LUIS POTOSI: Tamazun- y 1937, pas i | (F, MO, NY, TEX). on , Lamb 399 (GH, MO, Т: Маг. 1975, Novelo 74 О km NW of El Progreso, 21 Aug. 1941, Stanford poe (GH, MO, NY). vERACRUZ: al N de Orizaba, Cerro de Esca mela, 15 Oct. 1966, Rosas 66 = о > a = ~ ч ~ ~ N с со м. ~ . 1917 3. TOLEDO DISTRICT: River, 24 . 1941, Gentle 3812 (А, LL, MO, US). N CREEK DISTRICT: АП Pines, 5 ft., 23 Jan. 1930, А . WEST INDIES. BAHAMAS: C 1904, Brace 1527 (F, BA: without яи without date, Wright 2068 (MO, NY, US); Sierra de Anafe, Pinar del Rio, 14 Dec. 1911, Wilson 11297 (NY). JAMAICA: Walderston, Manchester, 2,600 ft., 1 Jan. 1918, Harris 12863 (F, GH, MO, NY, US). TRINIDAD: Caroni River Pea 1917, Curran 1340 (US). VIRGIN ISLANDS: St. Thom Dec. 1928, Nelthrop 3 (NY). GUATEMALA. ALTA VERAPAZ: 8 km below Tactic, along the Rio Frio, 1,400 m, 1 Apr. 1941, Standley 90542 (F, MO). BAJA VERAPAZ: Nino Perdido, bordering Río San José, 30 May 1977, Lundell 21022 (LL). CHIMALTENANGO: Tecpam, 2,100 m, 3 Aug. 1933, Skutch 541 (A, F, NY, US). CHIQUIMULA: 3-5 m N of Jocotan, Cerro Tixixi, 10 Nov. 1939, nt 31635 (F); 3-15 mi. NW of Chiquimula along Rio Taco, between Chiquimula and Montaña Barriol, 26 Oct. 1939, Steyermark 30615 (F). ESCUINTLA: NE of Escuintla, wooded barranca of Río Burrion, 720 m, 16 Mar. 1941, Standley 8957 1 (F, MO). GUATEMALA: Guatemala, 1,480 m, 30 May 1923, Ruano 405 (US). HUEHUETENANGO: San Miguel Acatan, 6,400 ft., 19 Aug. , Skutch 1021 (A, F, NY, US). IZABAL: vicinity of Quir aus , 15- 31 May 1922, Standley 23857 (СН, NY, US). JUTIAPA: vicinity of Jutiapa, 850 m, 24 Oct.-5 Nov. 1940, Stand- of the Monkey Turicentro y San Francisco, a 12 km, ca. 20 m lado sureste де la carretera, 9 Feb. 1972, Ortiz 2244 мо, NY). QUEZALTENANGO: Santa Maria de Jesüs, Finca Pi- reneos, 1,350-1,380 m, 11 Mar. 1939, Standley 68236 F, NY). QUICHE: San Miguel Uspantan, Арг. 1892, Smith 2920 (GH, MO, NY, US). RETALHULEU: E of Ajaxa, 330 m, 23 Feb. 1941, Standley 8824 - CATEPEQUEZ: W of Alotenango, wooded Чо је Escuintla, 3,200 ft., 3 Aug. 1965, Breedlove 11413 (Е, LL). SAN MARCOS: Tajumulco, 2 slopes of Volcán Ta- jumulco, barrancas, 2,300-2 m, 25 Feb. 1940, СА 36553 (Е). SANTA ROSA: Malpais, 4,000 pp., Nov. 1893, Smith 6071 (GH, US). soLoLa: San Pedro, babe slopes of Volcán Santa Clara, 1,900-2,100 m, 6 June 1942, Steyermark 47120 (A, F). SUCHITEPEQUEZ: near Patulul, 330-600 m, 5 Jan. 1939, Standley 62146 (A, Е). zacapa: Zacapa, l Jan. 1908, Kellerman 7019 F, NY, US). EL SALVADOR. AHUACHAPAN: p Mri 1921, Padilla 197 (А, MO). LA UNION: vicinity of La Unión, 150 m or less, 13-21 Feb. 1922, Standle) 20809 (CH). MORAZAN: Montecristo, 140 m, 8 Dec. 1941, Tuck- er 490 (US). SAN MIGUEL: Laguna de ету 75 m, 20 Feb. 1922, Standley 21021 (GH, US). SAN SALVADOR: without locality, Jan. 1922, Calderón 121(GH, MO, NY, S). SAN VICENTE: vicinity of San Vicente, 400-500 m, 7-14 Feb. 1947, wir dd 3573(F). па vicinity of Sonsonate, 220-300 m, 18-27 Mar. 1922, Standley 22300 (NY, US). Hon. ATLANTIDA: vicinity of Tela, O m, 14 Dec. 1927-15 Mar. 1928, МА 53743 A, F, . CHOLUTECA: San Marcos de Colón, 6 June 1970, Barkley 40510 (F, GH, MO). comayacua: 20 а al N de Siguatepeque, Barranco Trincheras, 1,400 п 18 July 1962, Molina 10840 (F, LL, NY, US). COPAN: 8 km de Santa Rosa de Copán, entre El Portillo y San ~ ~ Smith 5153 (GH, NY, US). DISTRITO CENTRAL: S 10 July 1969, Barkley 29429 (GH). EL PARAISO: road from Danli to El Paraiso, 22 Feb. 1952, Carlson 2517 F). GRACIAS A DIOS: Río Lisiksa, 13 Nov. 1976, Fryxell 2807 (F, NY). ISLAS DE LA BAHIA: Roatan Island, 4 km E of Coxenhole, 5-20 m, 21 Apr. 1967, Molina 20701 (F, US). MORAZAN: Jicarito River, 2,600 ft., 18 June 1948, Glassman 1652 (F, GH, TEX, NY). oc Aie i. NW of Ocotepeque, Mt. Сога] de Cordillera Meren 25 Aug. 1968, Molina 22080 (F, NY). OLANCHO: 20 km ~ 1927, Standley 53926 (А, BOACA: toward Camoapa, 200-5 Do Seymour 5426 (MO). cARAZO: N bank of Rio Escalante, km upstream from mouth, 24 Aug. 1977, Neill 2438 (MO). CHINANDEGA: El Viejo, near Rio Chiquito, 0-100 m, 27 Dec. 1969, үе 267 1 (MO). | 2.8 km N of Cuapa, 400-500 m, 21 Jan. 1978, Stevens 6043 (МО); vicinity of Juigalpa, wet la t along Río Paigua, 160 m June 1947, Standley 9226 (F). COMARCA DEL CABO: Bilwaskarma, Thaler Memorial Hos- pital, pine woods, 9 July 1972, Robbins 5777 (MO, NY, cultivated?). ESTELI: 6 km de Pueblo Nuevo, carretera a Limay cerro San Ramón, 810 m, 27 Sep. 1980, Moreno 3094 (MO). cRANADA: Granada, Lake Nicaragua, 24 Dec. 1968, Hamblett 1112 (F, GH, NY). JINOTEGA: 10 km NE of Jinotega, La Bastilla, 11 Jan. 1969, Zelaya 2145 F, NY). LEON: Momotombo, 27 July 1972, Robbins 6110 (F). MADRIZ: Cerro Volcán de Somoto (Tepesomoto), Volume 80, Number 2 1993 Turner & Mendenhall 447 Revision of Malvaviscus 1,500-1,600 m, 16 Apr. 1980, Araquistain 2122 (МО). MANAGUA: 12 km E of Managua, vicinity of Escuela Nacional de Agricultura y Ganaderia, Коше 1, 16 Jan 1969, Seymour 2235 (GH, MO, NY, cultivated?). MASAYA: without locality, 13 Feb. 1903, Baker 163 (GH, NY, US). MATAGALPA: Cordillera Central de Nicaragua between Matagalpa а) and Ji Mon Santa Maria de Ostuma, 1,300- 1,500 m 1963, Williams 23401 (F, LL, US). NUEVA SEG OVIA: , 7 Aug. 1977, Stevens 3039 MO). RIO SAN JUAN: near Cano ‘Chotaleno. 20 km NE of El Castillo, 200 m, 18-21 Apr. dine M 3582 (MO). RIVAS: Belén, K97, Route 2, 7 Jan. 1969, Moore 1913 ava, ipi sias on Mt. Liveco , Nelson 5048 (MO); Punta del Mico, 0-100 m, P Маг. 1971, Nelson 4248 (MO). Costa RICA. ALAJUELA: Santiago de San Ramón, 26 Jan. 1937, Brenes 21982 (F, NY). CARTAGO: 3 km SE of Cartago, 1,200 m, 10 Aug. 1967, Taylor 4231 (MO, NY). GUANACASTE: along road from Santa Cruz to Playa amarindo near Rio La Lima, 40-80 m, 28 dg 1966, NY); Comelco m, 8 Jan. 1967, Burger 4328 (F, MO, NY). Limon: Los Diamantes on to Guapiles, de Santa Clara, 6 и 1957, Carlson 3452 (F, US). PUNTARENAS: 4 mi. W of Rincón de Osa, forested area near Pun airfield, 4-7 June 1968, Burger 5425 (F, NY); Monteverde, Oct. 1977, Dryer 1662 (F, MO); Cabo Blanco Nature Reserve, 1- 7 Dec. 1969, Burger 6576 (F, LL). SAN JOSE: vicinity of El General, 1,040 m, Jan. 1936, Skutch 2368 (A, GH, MO, NY, US); La Palma, 1,540 m, Aug. 1898, Smith 7393 (F, GH, MO, NY). PANAMA. BOCAS DEL TORO: Water Valley, 9 Sep. 1940, von бле 657 (GH, MO, US). CHIRIQUI: Burica Peninsula, 11 m slope of La Popa, 5,400 ft., 5 Aug. 1972, D'Arcy 6408 (LL, MO, NY). cocLE: vicinity of El Valle, 800-1,000 m, 22 Dec. 1936, Allen 91 (А, GH, MO). COLON: Sal- 972, Gentry , NY). DARIEN: usns Chepigana, 5 Маг. 1940, Terry 1386 (A, F, MO). HERRERA: hill above Chepo de las Minas, 700 m, 19 Dec. 1977, Folsom 6979 (MO). LOS SANTOS: Las Tablas, 10 Aug. 1962, Dwyer 2480 (MO, US). PANAMA: N of El Llano, 500-800 m, 25 July 1972, Gentry 5573 (LL, NY). SAN BLAS: Comarca de San Blas, trail along Continental Divide, 25 July 1986, McDonagh 368 (MO). ЕТ Sona, 500 m, 24 Nov. gd Allen 1045 (F, O, US). Mie AT- NTICO: Barranquilla е vicinity, Tulara, Jan. Elia; 428 (US). BOLIVAR: vicinity of Turbaco, SU 300 m, 6-22 Nov. 1926, Killip 14187 (СН, MO, NY). GUAJIRA: Maicao, Arroyo Tabaco near the Intercor Coal campamento at Tabaco, 9 Mar. 1981, Bunch 0437 (HUA). SUCRE: trail from Coloso to Reserva de Primatas, 300-350 m, 17 Nov. 1981, Gentry 34793A (MO). GUYANA, Cultivated in British Guiana Botanic Gardens, Apr. 1907, се ids (BRG, cultivated). VENEZU JELA. DISTRITO FEDE und мо d ч 7121 (F, GH, ‘US. cultivated) м los Altos, 1,300 m, 24 June SOCIETY ISLANDS: Raia 692 (MO, cultivated). INDIA. WEST BENGAL: Calcutta, Sib- pur, without date, Raizada 29153 (MO, cultivated). Kenya. Nairobi, Keren, Feb. 1963, Gardner EAH, 12669 (US, cultivated). MALAYSIA. SARAWAK: Kapit, 1929, Clemens 21029 (NY, cultivated). THAILAND. Bangkok, 1899, Zimmermann 51 (MO, US, cultivated). VIETNAM. Da Nang, “ M: Tourane," May-July 1927, Clem- ens 4275 (MO, US, cultivated). lb. Malvaviscus arboreus var. drummondii (Torrey & A. Gray) Schery, Ann. Missouri Bot. Gard. 29: 215. 1942. Malvaviscus drummondii Torrey & A. Gray, Fl. N. Amer. 1: 230. 1838. Pavonia drummondii (Torrey & A. Gray) Torrey & A. Gray, Fl. N. Amer. 1: 682. 1840. Hibiscus drummondii (Torrey & A. Gray) M. J. Young, Familiar Lessons in Botany with Flora of Texas 186. 1873. TYPE: United States. Texas: Austin Co., San Felipe de Tejas, T. Drummond, 1835 (Texas Drum- mond Coll. III, no. 1) (lectotype, selected here, NY [Torrey herbarium]; isolectotype, GH; probable isolectotype, ) Selection of a NY specimen from the Torrey collection as lectotype was necessitated by the ab- sence of material at GH that might have been examined by either Gray or Torrey; the isotypes cited above were accessioned by GH well after M. drummondii was publishe t is unclear whether Young was describing Hi- biscus as a new species, as a new combination after Torrey & A. Gray, or asa previously treated taxon. It is unlikely that Young intended to present a new species. Elsewhere in the flora Young's new species bear the notation “п. sp." The specimen described was certainly of Torrey & Gray's previously de- scribed Malvaviscus drummondii (red flowers, red fruit), and the identical specific epithet suggests Young was transferring the species to Hibiscus. However, Young did not cite Torrey & Gray as basionym authors and did not specifically state that Hibiscus drummondii was a new combination. Young did have at least indirect access to the Flora of North America; Torrey & Gray were cited elsewhere in the flora. However, many species that Torrey & Gray had named did not include an author citation. It seems possible that Young was simply unable to associate an author with what she thought was a legitimate name, Hibiscus drum- mondi. Malvaviscus arboreus var. drummondii closely resembles variety arboreus, but the former plants are erect suffruticose, clone-forming, herbs or shrublets 0.5-1.5 m high; blades having mostly obtuse or rounded apices and abruptly cordate bases, the vestiture of petioles and peduncles dense- ly and uniformly stellate-pubescent, the hairs 448 Annals of the Missouri Botanical Garden scarcely exceeding 0.25 mm. Chromosome num- ber, 2n = ca. 14 pairs. Distribution (Fig. 1). northeastern Mexico and southern Texas along the Gulf Coast, but more widespread (albeit patchy) and numerous throughout central and eastern Tex- Sparsely occurring in as, especially in rich soils along streams in shaded areas, cultivated or escaped from cultivation east- ward to Florida; flowering all seasons, depending upon time, severity, and duration of frost condi- tions. Diagnosis and relationships. According to Schery (1942), the variety drummondii “is one of the most distinct varieties of M. which is an overstatement if one considers that he included within his concept of M. arboreus several taxa that others, including the present workers, would treat as distinct species. Actually, variety drummondii is a weakly differentiated, relatively uniform, populational complex of M. arboreus, which is largely restricted to central Texas. South- ward along the Gulf coastal region of northeastern Mexico it appears to grade into variety arboreus, especially in the vicinity of Tampico, Mexico. Southward from Tampico variety arboreus be- comes increasingly shrubby or clambering, the leaves less uniformly lobed, with mostly acute api- ces and a vestiture on the petioles and stems that is sparsely distributed, usually in lines, or some- times glabrate or nearly so. Fryxell (1988) did. not account for M. arboreus var. drummondii in his treatment of Mexico; the several specimens from Mexico that we cite here as belonging to variety drummondii are superficially similar to variety ar- boreus and, as indicated, the two taxa appear to intergrade in northeastern Mexico. It is questionable whether variety drummondii is truly native to the states east of Texas, although it is certainly native to Texas, as attested to by the numerous and remarkably uniform populations that grow in central Texas. The taxon apparently was, early on, taken into cultivation in the more coastal southeastern United States. Collections in this area have been relatively isolated and are represented in Figure 1 as cultivated, whether known to be cultivated, thought to be persisting after cultiva- tion, or thought to be escape Representative specimens examined. UNITED STATES. ALABAMA: Tuscaloosa Co., University of Alabama us, near Smith Woods, 25 July 1965, Dermus 444 (GH, cultivated?). FLoRIDA: Columbia Co., ca. 3 mi. N of 27 along Ichetucknee River, Will, 1961 (GH); Duval O., r Jacksonville, without date, Curtis s.n. (GH, cultivated?) Escambia Co., Pensacola, Brinker, 1941 arboreus,” (MO, cultivated); Hillsborough Co., without locality, Fred- holm, 1904 (GH, cultivated?); Leon Co., Tallahassee, live oak woods, 3 Oct. 1957, Godfry 56118 ien GEORGIA: diei vacant lot by railr yards, 5 : cultivated?) LOUISIANA: a Parish, ia without date, Hale s.n. (NY, cultivated from Texas). MISSISSIPPI: Adams Co. Natchez, without date, Gale ), Jackson Co., near Ocean Springs, 14 July 1889, Fredholm 2176 (US); Lincoln Co., 11 mi. W of Brookhaven, 11 July 1950, Webster 3281 (NY, US). TEXAS: Angelina Co., Neches River, Boon, 1934 (TEX); Aransas Co., Goose Island, N shore of Copano Bay, 10 June 1953, Johnston 53175.27 (TEX); Bastrop Co., Bastrop State Park, 10 June 1953, Gentry 1467 (NY); Bell Co., along Leon River channel, 5 July 1954, York 54566 (TEX); Bexar Co., without locality, Jermy, 1904 (MO, NY); Brazoria Co., 4 mi. S of Angleton, 22 Oct. 1948, Rogers 6579 (TEX); Brazos Co., College Station, 19 Sep. 1916, Palmer 10756 (MO, i. E of Brownwood, along U.S. Hwy. at 10 (LL, cultivated); y Co., Marble Falls, Vanderbilt ihe (US); Caldwell Co., Co- lumbia, along streams, 1899, Bush 312 (MO, NY); Calhoun Co., 4 mi. NE и. Tivoli, Guadalupe River bottoms, 24 Nov. 1945, Cory 51153 (GH, NY, TEX); јатегоп Co., near Brownsville, banks of the Rio Grande, Aug. 1888, Pringle 1959 (F, GH, NY, MO, USy Colorado Co., Eagle Lake, 21 Aug. 1946, Warnock 46369 (TEX); Comal Co., Comanche Spring, Aug. 1849, Lind- heimer 685 (F, GH, MO, NY, TEX); Dallas Co., Dallas, Reverchon, 1879 (F); DeWitt Co., without aei Rie- del, 1941 (TEX); Fayette Co., Muldoon, 3 Oct. 1950, Ripple 51-773 (TEX); Galveston Co., without decus Nelson, 1941 (GH, TEX); Gillespie Co., Fredericksburg, without date, Jermy 707 (MO); Goliad Co., Goliad, 22 Sep. 1926, Williams 70 (F, MO); Gonzales Co., Ottine, woodland near bog on the Soefje farm, 3 Oct. 1943, табса 13856 (F, GH, MO, NY, TEX); Grimes Co., Navasota, 1897, Turner 2 (NY); Harris Co., Houston, 1 July 1872, Hall 53 (F, GH, NY); Hays Co., San Marcos, Tharp, 1939 (F, GH, NY, TEX); Jackson Co., Navidad River, 1 July 1915, Drushel 2844 (MO); Karnes Co., Green, Media Creek bottoms, 2 Sep. 1952, d 1004 (TEX); Kendall Co., Spanish Pass, 5 July 1911, Clemens 609 (MO, NY); Kennedy Co., King Ranch, headquarters, 23 Sep. 1958, Lundell 15136 (LL, МО, NY) Гауаса Со., 18 mi. SE of Yoakum, along Hwy. e 16 July 1949, Tharp 49167 (F, TEX); Limestone , N of Groesbeck, 18 Aug. 1968, Fryxell 706 (F, MO. Madison pe near Trinity River, 13-14 July 1909, Dixon 441 (F, GH, NY); McLennan Co., Waco, without date, Pace 221 (ИО Montgomery Co., Willis, Aug., Warner s.n. (MO); Nacogdoches Co., ой locality, Barrett, 1944 (TEX); Nueces Co., Bishop, 20 June 1925, Eifrig 19 (F); Refugio Co., Gulf Coast, “probably near Tivoli," July 1976, Ил ата 454 (GH, MO); Robertson Co., 13 mi. E of Benchley, 28 July 1950, Gould 5765 (TEX); San Patricio Co., Mathis, 26 Sep. 1958, Correll 20414 (LL, NY); Travis Co., Harthaven, 6 Oct. 1944, Warnock W1021 (F, NY, TEX); Tyler Co., 5 mi. NW of Woodville, roadside park, 28 Sep. 1948, Cory 54840 (LL); Victoria Co., 8 mi. from Victoria, in field along Coletto Creek, 3 Oct. 1952, Correll 14819 (LL); Walker Co., near Lake Livingston, 10 Aug. 1975, Fryxell 2538 ashington Co., without locality, June 1938, Brac kett 175 (TEX); Wharton Co., Pierce, 14 Sep. 1901, Tracy 7476 (F, GH, MO, NY, TEX); Wilson Co., Flo- Volume 80, Number 2 1993 Turner & Mendenhall Revision of Malvaviscus 449 Adag: June Ls Age a 2 (TEX). MEXICO. COAHUILA: O (TEX, possibly cultivated). Ta- 2 18 Feb. 1939, LeSueur 283 (F, TEX); vicinity of Tampico, 3-6 June 1910, Palmer 525 (GH, US). 2. Malvaviscus penduliflorus DC., Prodr. 1: 445. 1824. Malvaviscus arboreus var. pen- duliflorus (DC.) Schery, Ann. Missouri Bot. Gard. 29: 233. 1942. Malvaviscus arboreus subsp. penduliflorus (DC.) Hadac, Folia Geo- bot. Phytotax. 5: 432. 1970. TYPE: Mexico. Without locality, date, or collector, Icones Florae Mexicanae no. 100 (Torner Coll. acc. no. 6331.1712, Hunt Institute, according to Fryxell, 1988). Malvaviscus penduliflorus, except for its large corollas (mostly 5-6 cm long) and somewhat larger calyces (mostly 15-20 mm long), falls well within the descriptive parameters of M. arboreus var. arboreus. Chromosome number, 2n = Distribution (Figs. 1—3). duliflorus is a widespread cultivar of unknown or- Malvaviscus pen- igin. It is widely planted in gardens throughout the New World and escapes cultivation, to judge from information on herbarium labels. In addition, it is believed to hybridize occasionally with locally na- tive taxa, mostly M. arboreus; flowering all sea- sons. Diagnosis and relationships. Malvaviscus penduliflorus is remarkably uniform throughout its artificial range and is distinguished from M. arboreus var. arboreus by its larger corollas (most- ly 50-60 mm long vs. 20-50 mm) and somewhat larger calyces (mostly 15-20 mm long vs. 10-15 mm). In addition, the leaves are nearly always ovate and subglabrous, and the flowers tend to be single and axillary with a pendulous nature. Its original habitat (or the area from which the “cul- tivar" might have arisen) is unknown, but it was first described by de Candolle from drawings made for the /cones Florae Mexicanae (cf. Fryxell, 1988), perhaps from garden material. As noted by Fryxell (1988), the species only rarely sets fruit, and this is verified by our own observations. Nev- ertheless, the area of origin of this species is prob- ably south-central Mexico, for numerous collec- tions from this region might reside within the descriptive parameters of M. penduliflorus except that the corollas tend to be shorter. It is likely that M. urticifolius (7 M. hintonii) is a white-flowered corolla form of M. penduliflorus with somewhat shorter petals, for as noted by Fryxell (1988), M. hintonii appears to be largely sterile and appears to occur in cultivation. We have, nevertheless, placed M. hintonii within the syn- onymy of M. arboreus var. arboreus because its leaf shape and vestiture more closely resemble that of M. arboreus. Certainly, it is not typical of ma- terial we include under M. penduliflorus. The co- rolla of M. penduliflorus is nearly always crimson or red; occasional specimens will have corollas that are pale pink (e.g., Hu 123534, from Hong Kong, US), or *pure white" (Plowman 11215, Peru, F). It should also be noted that while the androecium is normally included within, or barely exceeding the corolla, as noted by Fryxell (1988), it is not yond the corolla for 1-3 cm (e. TEX; Hinton 13529, LL; Taylor 36, TEX); a least this is true for our concept of this ub E cultivated taxon. We accept the likelihood that M. hintonii is a synonym of the earlier M. urticifolius (as treated by Fryxell, 1988), the latter having been first col- lected by Haenke, probably on his trip from Aca- pulco to Mexico City and return during the year 1791. As noted under M. arboreus var. arboreus, it is likely that occasional hybridization between M. arboreus and M. penduliflorus occurs, at least to judge from seemingly intermediate specimens, most of these occurring in regions where both are known to occur near each other. This is all discussed here to emphasize that the probable region of origin of the morphologically uniform cultivar M. penduliflorus is south-central Mexico, perhaps from the region south of Mexico City. Representative specimens examined. UNITED STATES. FLORIDA: Dade Co., Miami, 24 Dec. 1927, Mol- denke 3581 (NY, cultivated); Hendry Co., Clewiston, 4 May 1958, Cooley 6231 (NY); Hillsborough Co., Auburn Highlands, July 1970, Burch 3755 (MO, cultivated); Lee Co., East Fort Myers, apparently escaped, Moldenke 993 (MO, NY, US); Manatee Co., Palmetto, 31 July 1970, Burch 3757 (MO, cultivated); Palm Beach Co., Delray Beach, edge of coastal sand dune thickets, 29 Dec. 1966, Moldenke 24176 (LL); Seminole Co., 1.4 mi. W of San- ford city limit, grassy thicket near Florida Hwy. 46, p. 1960, W ard 2240 (GH, US); Volusia Co., Ponce Inlet, gr edge of salt marsh, 4 Oct. 1 HAWAII: Oahu, Mokuleia, 30 July 1937, iic 11402 (GH, MO, NY, cultivated). TEXAS: Bosque Co., Clifton 16 Jane c ee 15 (F); Brazos Co., College Station, owerbed, e Hart Hall, (Texas A&M Unive aia 22 Nov. 1948, Trew 162 (TEX, an , Brownsville, 21 Dec. 1919, Han (GH, MO, NY, cultivated); Harris Co., Hou Nursery, 5 Dec. 1933, without collector (TEX, culti- vated); Webb Co., Laredo Jr. College Campus, 11 Nov. 450 Annals of the Missouri Botanical Garden BO? TOS 609 10° 3 h S 10° Ки = 6 VENEZUELA aii x A „лнн C ? М = i > = LA ба У 2 ~> COLOMBIA 5 P ==> i += NE. > => М Га == > s. Vd * pp зз о 1.594 0 3 A i 0° 7 w ` ; ) фи ~; o. 4 ' LE Ан @ ө 5 __..- А | BRAZIL € 30 ^ о0о г ? = = ; A • 5 г - = = ~ MEX ndn 10° == E: == 10° е = Т 7 Т ъ 2 = 62: 5 i Е 3 PE cd у * =е = I— == ~ ~ MATE Бы ) М... E =5t= == NIS HS ЩИ pem Se SE Se") ` a теза у 1 a = = ул { 2 гт, БОМА 4 = 7 ==. тени MAT = a, B - © О 600km 2 == = } EE un / 20? т аә ~ 20° а mA / 4 === Ze ==; Н = A T ` P { e E | 80° 709 60? FIGURE 3. Distribution of Malvaviscus concinnus E closed circles), M. penduliflorus (large closed circles) and M. williamsii (small open circles) in South Ame 1962, Garcia 28 (TEX, iip. Willac Raymondville, 18 Apr tivated). MEXICO. CHIAPAS: okolum, E of Sibanilha, 5, 200 O ft., Breedlove 1104 8 (Е, Ter enejapa , Rios 273 AJUA Co., 5 , near the school house 15 July 1965, LL). CHIHUAHUA: Kou Grande, 19 May 1960, Pennington 151 (TEX). najuato, 1891, Duges 280 (GH, о GUERRERO: mi. (LL, TO: Gua- Atoyac de Alvarez, 4923 (TEX). MEXICO: Temascaltepec, Tenayac m, 8 June Zitácuaro, La 13529 ro LL, NY, US). MORELOS: near Cuernavaca, 28 Aug. 1935, Bailey 305 (F). NAYARIT: 6. mi. E of Jaleocotás 800-1,200 m, 9 Sep. 1960, McVaugh 5,500 f 1,000 m, Mora , 1,300 19 Dec. m, 12 Dec. 1984, Cowan , 1,450 1933, Hinton 4014 (A, NY, US). MICHOACAN: 1938, Hinton Volume 80, Number 2 1993 Turner & Mendenhall 451 Revision of Malvaviscus 18932 (NY, TEX, US). NUEVO LEON: on I. T. E.S.M. cam- pus, 1,700 ft., 28 July 1970, Taylor 36 (TEX, culti- vated). OAXACA: Santo Domingo, 1,600 m, 22 Dec. 1906, Conzatti 1683 (F). PUEBLA: Дена 940 m, 4 Маг. 1976, Baez 275 (Е). SAN LUIS POTOSI: Mpio. San Antonio, El Lejem, 15 Dec. 1978, Alcorn 2260 (TEX, arene SINALOA: vicinity of Mazatlan, 7 Apr. 1910, Rose 1415 (NY, US, cultivated). VERACRUZ: 2 km S of Tampico, Td 1910, и 391 (MO); near Córdoba, 20 July 8 (MO). BELIZE. St. Johns College EA Dieckman 237 (МО, aa West INDIES. Кон tivated). CUBA: Sierra Maestra, 30 km S of Bayamo, 400 m, 17 Aug. 1951, Webster 4123 (GH, US). Curacao: at pool St. Martha, Jan. 1970, Arnold-Broeders 3888 (A). DOMINICAN REPUBLIC: Peravia, El Taton, 13.7 San José de Ocoa en la carretera a La Laguna, 20 Oct. 1982, Mejia 23795 (NY). GUADELOUPE: Basse Ter re, 25 Mar 82, Howard 19775 (NY, cultivated). JAMAICA: St. Andrew, U.C.W.I. campus, 550 ft., 30 Oct. 1957, Yuncker 17241 (NY, cultivated). PuERTO Rico: Barranquitas, along Hwy. 156, 6.0 mi. SW of Сотепо, 4.7 mi. SW of junction with Hwy. 776, 25 Nov. 1981, Hansen 9335 (MO). Sr. Lu er Canaries River approach- ing Morne Gimie, 28 Jan. 1985, Howard 19935 (A NY). TonTOLA: Treasure A Hotel grounds, 10 m, 15 Dec. "Arcy 387A (MO, cultivated). TRINIDAD AND Tosaco: I.C.T.A. Savanna, 23 Feb. 1959, Richards 1378 (NY). HONDURAS. ATLANTIDA: vicinity of La Ceiba, near Puente Alto stop on S.F. Co. RR, 800 ft., 19 July 1938, siena оо СН, MO). COMAYAGUA: vicinity of Siguatepequ m, 25 Mar.-5 Apr. 1947, Stan- dley 6658 (F, MR MORAZAN: Zamorano, 800 m, Feb. 1945, Rodriguez 2264 (F). NICARAGUA. ZELAYA: epe "| Puerto Cabezas, 11 Feb. 1977, van Stelle 18 и ТА larg ^ JOSE nal Museu , 1,1 + канан апа old citrus grove a nol "a 1,390 m, 5 Jan. 1975, Nee 14143 (MO). COCLE: near El Cope, 27 ct. 1967, Garner (A). COLOMBIA. A Ula: Me- Я п dellin, Ciudad Universitaria U. de A., 6 Aug. 1981, Alzate J үче CUNDINAMARCA: valley abov e Colegio, 800 ft., 1968, Barkley 38852 (TEX). SANTANDER: 17 En SE, of San Vicente de Chucuri on road to Zapatoca, 1,250 m, 25 Jul i oo 15438 (MO, NY). VALLE DEL CAUCA: Palm 00 m, 2 Jan. 1972, Maas 576 (MO); Cali, | o Valle Zona de residencias, 26 Nov. 1982, Paz 12 (MO). VENEZUELA. ARAGUA: Maracay, Agronomia, Universidad Central de Venezuela, 24 June 1963, Trujillo 5709 (F, cultivated). DISTRITO FEDERAL: Caracas, Colinas de El Paraiso, Jan. 1944, Lasser 1025 sco 25 Oct. 1944, Vasconcellos 177 (RB). RIO DE JANEIRO: Rio de Janeiro, Estrada da Vista Chinesa km 2, Alto da Boa Vista, 3 Sep. 1979, Carauta 3183 (F). SAO PAULO: Limeira, Hoehne, 1946 (SP). ECUADOR. COTOPAXI: 1.9 km NW of El Corazon, road between Quevedo and El Corazon, 1,225 m, 5 Apr. 1983, Croat 55836 ma NY); near La Mana, islet in Río San Pablo, 500 ft., July s Webster 22724 (TEX). GUAYAS: calde 0-20 964, Valverde 680 (US); Capeira, km 21 Gc to Daule, 17 Sep. 1982, Dodson 11280 (MO). LOS RIOS: km 56 Quevedo-Santo Domingo, Rio Palenque Biological Station, 150-220 m, 31 Mar. 1971, Dodson 4290 (F, MO). маро: Puerto Napo, 7 km S of Tena, 8 Aug. 1988, Mendenhall 15 (TEX, cultivated). PERU. CUZCO: La Convención, Quillabamba, Salaspampa, road to Kiteni, chacra del Sr. Alejo Daniel Caceres, Rio Uru bamba, 1,110 m, 28 Oct. 1986, Nuñez 6320 (MO). HUANUCO: Tingo Maria, Rio Huallaga, 12 Mar. 1977, Boeke 1229 (MO, NY); Jardin Botánico de Tingo Maria, O m, 8 Dec. 1981, Plowman 11215 (F, cultivated). JUNIN: N of La Merced, 12 km N of Puente Paucartambo, Rio Paucartambo, chacra Schuler, 960 m, 6 Feb. 197 Teppner 79/318 (05). OLD WonLp: EcvPr. Cairo, Cairo en, = Aug. 1974, without collector (LL, NESIA. Moorea: Faatoai Valley, usually ү килче but beginning to escape in areas of deserted dwellings, 19 July 1967, Smith 103 (MO, NY, US). INDIA. UTTAR sag en Dehra Dun, Forest Research Institute, Raizada, 1954 (NY, cultivated); Lucknow, Central Drug Research Institute, 113 m, 26 Apr. 1982, Parsad 11769 (F, cultivated). PAKISTAN. SIND: Karachi, National College, 6 Aug. 1969, Abedin 3777 (NY, US, cultivated). PAPUA Мек GUINEA. MANUS ISLAND: Pelikawa Village, 25 June 1971, Stone 10494 (A, cultivated). PHILIPPINES. LUZON: Manila, Jan. 1954, Steiner 332 (US, ma MINDANAO: Davao, valley near Santa Cruz, 900 ft., АМО. 1587 (A). SOCIETY ISLANDS. TAHITI: Punaauia, Outumaoro, Hotel Maeva, 3 m, 2 s не uer € (NY, cultivated). Tat- : |. Pingtu 29 Sep. 1976, Ching-en Chang 17 (MO, Т Twain cm 14. Nov. 1970, с; 710 о cultivated). WESTERN SAMOA. UPOLU ISLAND: Moto'otua in the Apia area, -2 Jan. 1975, Whistler 172006 (US, culóvated) 3. Malvaviscus achanioides (Turczaninow) Fryxell, Syst. Bot. 4: 253. 1979. Abelmos- chus achanioides Turczaninow, Bull. Soc. Imp. Nat. Moscou 31: 196. 1858. Hibiscus acha- nioides (Turczaninow) Hemsl., Biol. Amer., Bot. 1: 121. 107%, Ta hen: ‘in sylvis Teapae," 2,000 ft., Lin- den 838 [938] (holotype, not located; pho- toisotype, Malvaviscus cutteri Standley, Publ. Field Columbian Mus., Bot. Se . TYPE: Honduras. Atlantida: Lancetilla Valley near Tela, Dec. 1927-Mar. 1928, Standley 54127 (holotype, F; isotypes, A, US). Shrub or small tree 1-4 m high. Stem densely pilose, the hairs 1-2 mm long. Leaves mostly 15- 35 ст long, 5-20 cm wide, scarcely reduced up- ward; petioles 3-15 cm long, pilose like the stem; blades broadly ovate to cordate, rarely weakly lobed, pubescent on both surfaces, especially along the major veins, the margins irregularly serrate, un- dulate, to nearly entire. Flowers mostly 1-7 in short terminal subfasciculate clusters, or rarely 20— 30 in an abbreviated panicle, the peduncles mostly 0.5-3.0 cm long. Calyces mostly 16-22 mm long, pilose, yellowish, the subtending bracts 8-11 and 452 Annals of the Missouri Botanical Garden about Уг as long at the calyx. Corollas 4-5 cm long, red, not flaring, petal apices occasionally lo- bate. Fruit ca. 15 mm across, fleshy, and red. Distribution (Fig. 1). | Eastern Honduras to Mexico, along the lower montane slopes and ad- jacent flood plains of the Gulf coastal regions, most- ly evergreen rainforests, 100—1,000 m; flowering October to January. Diagnosis and relationships. The large leaves, which are scarcely reduc d upward along the stem, pilose vestiture, and large calyces distin- guish this species from the more typical elements of M. arboreus. In vegetative characters, and be- cause of its large yellowish calyces, it approaches M. concinnus of South America, but in most other traits it is like M. arboreus. Speci which have been called M. cutteri, have somewhat larger, thinner, mens from Honduras, more cordate, leaves than specimens from Mexico; otherwise these are quite similar. Occasional spec- imens from the more montane Pacific slopes of western Chiapas (e.g., Matuda 5357, Boquerón, ca. Motozintla, 2,540 m, LL), and perhaps some of those examined by Fryxell (1988) but not ex- amined by us, resemble M. achanioides in vestiture and leaf shape, but in other characters they more nearly approach M. arboreus (e.g., the corollas are quite short, ca. 30 mm long, and tend to flare) and we have placed such collections in the latter, believing the pubescence and larger leaves to be convergent characters in this area. An alternative treatment might include all of M. achanioides within an expanded M. arboreus, but that would also collapse the more artificially main- tained M. penduliflorus. In short, we have main- tained a nomenclature that is both traditional and pragmatic, though imperfect if one considers the character variations found in the occasional pop- ulation, which tend to blur the specific lines pro- ed. Finally, except for its somewhat smaller calyces and less flaring, somewhat shorter corollas, M. achanioides is similar to the more southern M. concinnus, and some future workers might opt to treat the former as part of the latter complex. Additional specimens examined. MEXICO. CHIAPAS: Mpio. Ixtacomitán, stream above Ixtacomitán, 455 m, Breedlove 56796 (LL, NY). raBASCO: Mpio. Teapa, Cerro del Cocona, 3 km from Teapa, 26 Oct. 1980, Cowan 3318 (F, МО, NY, TEX); Grutas del Cocona, ca. Teapa, 18 Nov. 1979, Ramos 2696 (NY). VERACRUZ: 5 km E of Tebanca (5 km E of E side of Lago Catemaco), m, 15 Jan. 1981, Nee 1997 1 (F). GUATEMALA. Martinez et al. 23588 (MO). HONDURAS. ATLANTIDA: Lancetilla — Valley near Tela, 6 Dec.-20 Mar. 1928, Standley 52756 (A, F, US). 4. Malvaviscus concinnus Kunth, Nov. Gen. Sp. 22. Achania concinna (Kunth) Sprengel, Syst. Veg. ed. 16, 3: 100. 1826. TYPE: Ecuador. Loja: "Crescit prope Loxam Peruvianorum," Humboldt, 1802 (ho- lotype, P-HBK not seen; photoisotype (from Berlin), F, СН) i nn balbisii DC., Prodr. 1: 445. 1824. TYPE: azil е to In dex Kewensis). Without lo- cality, without collector (holotype, G-DC, microfiche (single leaf only)). Malvaviscus cordatus Balbis ex DC., Pro 9. ds nomen nudum (merely ге һу de Candolle nonymous with M. balbis 35 Malvaviscus ee a dien Reli Haenk. 2: 135. 853 Mexici," with- out die б N s.n. [Rm PR not seen; pho- ” as from Mexico, as also suggested by Fryxell (pers. omm.). Malvaviscus —— Linden & Planchon, Pl. Columb. . TYPE: Venezuela. Trujillo: La Pena, 1,675 m, ca. 1846, Schlim 751 (holotype, BR not seen; photoisotype, F). This tome (Plantae Columbianae) was not sani tively published until ca. 1875, according to Spragu 927). The name, however, was only que estionably published at that tim e, there being only five copies ht give the latter date as the point of effective publication iaa ae pen lumb eanus Linden & Plan chon, Pl. Co- 3. TYP enezuela. Distrito Federal: Ca- racas, La a а date, Funck 372 (holo- type, BM not seen). See discussion after M. elegans, above. Guerke (1892) took up this name for his Flora Brasiliensis treatment. Malvaviscus speciosus Linden & Planchon, Pl. Columb. . TYPE: Venezuela. Mérida: forests, 1,950 m, dur. 1842, Linden 354 (holotype, BM not seen; photoisotype, F). See discussion after M. e elegans, above Malvaviscus spathulatus Garcke in Otto & Dietr. ., Allg. Gartenzeitung 21: . 1853. TYPE: Costa Rica. Oersted 407 (holotype, B destroyed?; photoholo- туре, F-9429; isotype, C not seen; photoisotype, F-21598 Matas longifolius Garcke in Otto & Dietr., Allg. Gartenz. 22: 321. 1854. TYPE: cultivated in Erfurt, Ger dade from je collected in northern Peru, Volume 80, Number 2 1993 Turner & Mendenhall 453 Revision of Malvaviscus Warszewicz s.n. (holotype, B destroyed?; photo- 427 ., Bjull. Moskovsk. Obsc. Prir., Otd. Biol. 31: 190. 1858. SYNTYPES: umbre, Funck 350 (spec- imen not located); Caracas, es pag de Funck [Ga- leotti] 372 (syntypes, G not , P not seen; pho- tosyntype, F-23715, на | to Fryxell, 1988). Malvaviscus ee pr. Turcz., Bjull. Moskovsk. Obsc. Я 31: 190. 1858. Malvaviscus glabrescens Planchon & Linden ex Triana & Plan- chon, Ann. ir owe E I zur nom. ше ТҮРЕ: Co lom r: Ocana, 850, L. Schlim p сз. E not seen; photo- SERA J Майдан leucocarpus Planchon & Linden ex Triana Planchon, Ann. Sci. Nat. (Paris) T 169. 1862. TYPE: Colombia. Save, Cordillera Oriental, Triana 5278/3 (lectotype, not seen; isolectotype, COL : Bol. 960. Морава velutinus Planchon & Linden ex Triana & Planchon, Ann. Sci. Nat. (Paris) 17: 168. 1862. TYPE: Colombia. Tolima: Maraquita, “Entre la Mesa et El Espinal, bassin du Magdalena," 400-1,200 m, Jan. 1854, Triana 3132 (holotype, P not seen; cdam, NY not seen, US not seen; according to xell, pers. comm.). Mice: guerkeanus Hieron., Bot. Jahrb. Syst. 21: 320. 1895. TYPE: Colombia. Low wet areas alon us Magdalena, Маг. 1868, А. Stuebel 106b (lec- otype, B destroyed?; isolectotypes, F, GH). ма, maynensis Huber, Bol. Mus. Paraense Hist. ied TYPE: Peru. “Cerro de Can- chahuaya," Quebra ado, 27 Oct. 1898, J. Huber 1383 к. MG; photoholotype, Е; o m "o e 2 ч gs c 5 m ivl o Lond + 00 — о isotype, R). Malvaviscus integrifolius Ulbr., Verh. Bot. Vereins Prov. Brandenburg (1908) 88. 1909. TYPE: Brazil. Ama- тоша: “bei Humaytha am oberen Juruá,” 22 Apr. 1901, E. Ule 5444 (holotype, B destroyed?; frag- ment holotype, R; photoholotype, F, GH; photoiso- type, US). Malvaviscus ulei Ulbr., Notizbl. Königl. Bot. Gart. m 1915. TYPE: Brazil. Amazonia: “Alto Acr ES з с Auristella," June 1911, Ule 9591 P lotype, B destroyed?; photoholotype, GH). Shrubs, small trees, or clambering vines, mostly 1-8 m high. L densely pubescent to subglabrous, the vestiture uni- eaves broadly ovate to cordate, formly short and variously spaced to long and close- ly packed so as to appear velutinous. Flowers single and axillary, or in terminal aggregations, the pe- duncles 1-10 cm long. Calyces mostly broadly campanulate at maturity, 18-40 mm long, 10- 30 mm wide, variously pubescent, as noted in the discussion below, the subtending bracteoles mostly 10-20, filiform to linear-lanceolate (1-3 mm wide), if the latter, nearly always broadest at or near the base. Corollas pale pink to dark red, usually broadly flaring at anthesis, mostly 40-70 mm long, rarely less, broadly to sometimes narrowly oblanceolate, usually at least a few petals with some degree of apical lobing, rarely not. Style and staminal column mostly protruding from the corolla for 10–40 mm. Fruits large and fleshy, red or white, mostly 15- mm wide at maturity, the seeds large, ovoid, up to 10 mm long and 4 mm wide. Distribution (Figs. 1, 3). | Northwestern South America along the upper tributaries of the Amazon River in Peru, Ecuador, Colombia, Venezuela, Bra- zil, Panama, and Costa Rica, from 100 to 2,000 m, often in lower montane cloud forests; flowering all seasons. Diagnosis and relationships. Malvaviscus concinnus is an extremely variable species throughout its range, often over very short dis- tances, the variation being much like that found in M. arboreus in North America. Thus, collections from San Martin, Peru, may possess a minutely stellate-pubescent vestiture (0.2 mm high or less) upon the calyces and associated bracts (Klug 3919, F, GH, MO, NY, US), or the vestiture may be quite pronounced or hirsute-stellate (0.5-1.5 mm high) on these same organs (Schunke-Vigo 7767, MO, US), and intermediates between these ex- tremes may also occur (Schunke-Vigo 6898, MO, JS). The corolla length and degree of apical flaring are quite variable: among the specimens just cited, and others from San Martin, the length varies from ca. 42 mm (Schunke-Vigo 6898, MO, US) and not so widely flaring (4 cm across), to ca. 65 mm (Allard 22074, US) and markedly flaring (to 9 cm across). Variation from population to population is es- pecially apparent along the upper, more western, tributaries of the Amazon River basin from Ven- ezuela, Colombia, Ecuador, and Peru eastward to 70*W longitude. Thus, the names M. elegans, M. oligotrichus, M. longifolius, and M. maynensis have been applied to forms with broadly lanceolate leaves; M. velutinus to forms with a vestiture of densely packed, long-stellate hairs; M. integrifoli- us and M. ulei to forms with nearly entire leaves, the latter having somewhat shorter petioles and longer calyces; and M. speciosus with trilobed, cordate leaves. In all of the aforementioned type material, as well as among the representative spec- imens cited, there is considerable variation in leaf shape, corolla size, vestiture, and fruit dimensions, so much so that one is forced to the conclusion that there exists in South America a single wide- spread, highly variable species, M. concinnus, and a more localized, more readily recognized, rela- tively uniform species, M. williamsii. The latter might have been treated, with equal validity, as a variety of M. concinnus. But M. williamsii is much Annals of the Missouri Botanical Garden better delimited than variety drummondii, the only infraspecific taxon maintained under M. arboreus, and it is certainly better delimited than M. achani- oides. Some comments on the large white fruits oc- casionally found in M. concinnus seem in order uch fruits in North America are mostly confined to Panama and Costa Rica, where they are usually associated with plants having large cordate leaves; however, similar fruits also occur on elliptical-leaved specimens from Costa Rica that have been called M. palmanus (e.g., Burger 5425, F, NY; Burger 5492, F, NY; Burger 7228, F, GH), but which we include in M. arboreus. White fruits, though smaller than those found in Central America, oc- casionally occur on Mexican plants referable to typical M. arboreus. Thus, fruit color (the basis for the name M. leucocarpus), like leaf shape and vestiture, is variable and cannot be used alone for delimitation. Representative specimens examined. ANAMA. COCLE: El Cope, E of sawmill above El Cope, 2,300 ft., 27 July 1978, Hammel 4096 (MO). DARIEN: lower slopes along Rio Coasi, 26 Dec. 1980, 8 (MO). PANAMA: N of El Llano, 500- 800 m, 25 july 1972, Gentry 5573 (F, MO). vERAGUAS: near Sante Fé, road at base of Cerro Tute, 3,000 . 1979, Antonio 1899, 1901 (MO). Costa Ric A. ALEJUELA: between Canas and Upala, 4 km NNE of Bi- j 1976, Croat 36309 (MO); San Carlos, margen del Rio Penas Blancas, 29 June 1985, Haber 1756 (MO). HEREDIA: Finca La Selva, OTS Field Station along the Rio Puerto Viejo, just E of к junction with the Río Savapiqui, ca. 100 m, 29 Nov. 1982, Mc- Dowell 995 (LL, ed 3 Dec. 1982, a 1028 - , 1 June 1971, Proctor 32220 mazon River 1969, Croat 7566 (MO). ANTIOQUIA: Cordillera Central Boquerón, ca. 16 km NW of Medellin on road to Turbo, 6°20'N, 75?40'W, 2,500- 2,600 m, 10 Jan. 1986, Stein 3160 (MO). ATLANTICO: region de Barranquilla, llanada de Juanmina, 10 m, 15 Jan. 1961, Dugand 5499 (NY). BOLIVAR: near Cartagena, La Popa, 50-175 m, 2 Nov. 1926, Killip 14052 (A, СН, NY). ВОХАСА: Valle de la Uvita, near Uvita, 2,490- 2,560 m, 16 Sep. 1938, Cuatrecasas 1850 (F, US). CESAR: La Jagua, Magdalena Valley, 5 Sep. 1924, Allen 565 (F, MO). CUNDINAMARCA: Mpio. La Mesa, carretera a a Anapoima, 1,300 m, 14 May 1952, Fer- nández 1336 (NY, US). GUAJIRA: N pra of Cerro del Espejo Serranía de Perija, Venezuela border 72950'W, 2,550 m mith 734 (several individuals and еш localities under this number) (A, , MO, ) l Mico Airstrip, 400 m, 7 Nov. : Umbria, 1930, Klug 1712 (A, F, MO). SANTANDER: vicinity of Las Vegas, 2,600- 3,000 m, 21-23 Dec uei cares 16087 (A, GH, MO, NY ‚ US). SUR DE SANTA о 1935, Haught e i бз, Valle del Rio Magdalena, 400 m 3 кили. 10496 (F, MO, NY, US). VENEZUELA du 1947, Pittier 15 Que brada Quintero, dd , 1976 NY) DISTRITO FEDERAL: Camino de ronda Las Flores a Papelón s Aug. 1939, Delgado 271 (F, US); near en 1 Cotiza de Los Venados, Oct. 1924, 1.000 m, 94 Le 19 ultu d 3943 (F). MERIDA: Rio Gonza, La Isla Jaji, 1 Nov. 1968, López-Palacios 1918 (NY). MIRANDA: E of ша Serraduro, E of Hacienda Garate, 16 km NE of Caucaguito, 30 km NE of Petare, between Petare and Guarenas, 1,200-1,500 m, 6 Oct. 1963, Steyer- mark 91616 (GH). PoRTUGUESA: Pozo Blanco, estación lluviosa, 2 Oct. 1979, Ortega 804 (MO). TACHIRA: Que- brada Agua Azul, S of El Reposo, 14 km SE of Delicias, 7931'N, 72?24'W, 2,150-2,300 m, 22-23 July 1979, Steyermark 11874 (MO). ECUADOR. CHIMBORAZO: Si- bambe, Hacienda "Та Carmela," 100-1,600 m, 16 Aug. 1943, Solís 5349 (F). ESMERALDAS: between Tonchigue and Galera, Esmeraldas- Muisne Road, 23 Nov. 1980, Harling 16693 (F). GuaYas: 12 km W of Guayaquil, Hacienda Barcelona Trasil, 4 Apr. 1962, Gilmartin 667 (US). MANABI: above Noboa, 200 m, 19 July 1942, Haught 3413 (GH, NY, US). MORONA-SANTIAGO: Parroquia Cu- manda, Rio Pastaza, ca. 4 km W of Mera, 23 Aug. 1968, Lugo 355 (F, NY). маро: Rio Gueppi (tributary of Rio Putumayo), veni Peruvian border post of Puerto Peru, 20 1978, Gentry 21871 (F, MO). EL ono: Sitio ин репјато, 100 ft., 23 Nov.-16 Dec 1978, Escobar 824 (HUA, TEX); Тукан Santa Коза and La Chorita, 0-100 m, 27 Aug. 1923, Hitchcock 21141 (NY, US). pastaza: Puyo- Ter ena road, 4.5 km from Puyo, ca. 900 m, 8 July 1980, Sobel 2452 (NY). SANTIAGO-ZAMORA: E slope of the cordillera, valley of the Rios Negro and Chupianza on the trail from Sevilla de Oro to Méndez, 1 Nov. 1944, Camp E-829 (US). Peru. AYACUCHO: Estrella, between Huanta and Río Apurimac, 500 m, 8 and 14 May 1929, Killip 23065 (F, NY, US). HUANUCO: Leoncio Prado, near border with Ucayali, 1,620-1,760 m, 75?48'W, 9°05'S, 10 Aug. 1980, Gentry 29567 (F, MO); Pachitea, Honoria, Bosque Nacional de Iparia, a lo largo del Rio Pachitea cerca del campamento Miel de Abeja, 300-400 m, 5 May 1967, Schunke-Vigo 925 (F, NY, US). JUNIN: Schunke Hacienda, above San Ramón, 1,400-1,700 m, 8-12 June 1929, Killip 24714 (F, NY, US). LoRETO: along Rio Amazonas, 5 of Iquitos, 18 Aug. 1972, Croat 19309 (F, GH, MO, NY); Maynas, Moena Cano between Iquitos and Rio Itaya, 7 Jan. 1976, Gentry 15654 (F, MO, ); Balsapuerto, 220 m, Арг. 1933, Klug 3015 (A, F, GH, MO, NY, US); Coronel Portillo, Bosque Na- cional de Iparia, a lo largo del Rio Ucayali cerca del pueblo de Iparia, 18 Aug. 1968, Schunke-Vigo 2617 (F, GH, МО). МАРНЕ DE DIOS: Tambopata, Tambopata Nature Reserve, ca. 30 air km SSW Puerto Maldonado at efflu- ence Rio La Torre/Rio Tambopata (SE bank), 12?49'S, 69°17'W, 260 m, 17 May 1980, Barbour 5332 (F, MO, Volume 80, Number 2 Turner & Mendenhall 455 1993 Revision of Malvaviscus NY). SAN MARTIN: Juan Jui Alto Rio Huallaga, 400 m, margins mostly strongly imbricate. Fruits red Oct. 1934, Klug 3919 (F, GH, MO, NY, 05); near Boquerón, Boquerón Pass, 92 km from Tingo Мапа on highway to Pucallpa, 400 m, 16 Dec. 1949-5 Jan. 1950, Allard 22074 (US); Mariscal Caceres, Uchiza, E of Pu- ente del Rio Uchiza, 400 m, 25 July 1974, Schunke- Vigo 7767 (MO, US); Marsical Caceres, Tocache N Puerto Pizana, Río Huallaga, 350 m, 4 June 1974, Schunke-Vigo 6898 (MO, US). TUMBES: Zarumilla, Bos- q Nacional de Tumbes, cerca de Campo Verde, 600- 800 m, 17 Dec. 1967, Simpson 380 (F, NY). UCAYALI: Bosque von Humboldt, entrance to Carretera Marginal, km 88 of Pucallpa- Tingo María Road, 75°02'W, 08%4.5'S, 220 m, 14 June 1987, Gentry 58344 (F, MO). BRAZIL. AMAZONAS: Manariao, basin of Rio Juruá, 27 May 1933, Krukoff 4589 (A, F, MO, NY, US); basin of Rio Juruá, near mouth of Rio Embira (tributary of Rio Tarauaca), 7930'S, 70915", 1 July 1933, Krukoff 5150 (A, F MO, NY, US) 5. Malvaviscus williamsii Ulbr., Notizbl. Bot. Gart. Berlin-Dahlem 11: 545. 1932. Mal- vaviscus arboreus var. williamsii (Ulbr.) Schery, Ann. Missouri Bot. Gard. 29: 226. 1942. TYPE: Peru. Loreto: Lower Rio Nanay, forest between Rio Nanay and Rio Napo, 6 June 1929, L. Williams 706 (holotype, B destroyed?; isotype, F not located). Ulbrich cited only two specimens, one at B the other at F, neither of which we could locate. From tha A inti and location there is little doubt as to the identity of the material concerned. In lieu of extant types we have proposed the following: NEOTYPE: Peru. Loreto: Maynas, "trail between Rio Amazonas above Indiana and Mazan (Rio Napo) between Mazan and halfway," mature upland rainforest over clay, 100-130 m, 5 July 1971, Sidney McDaniel 15218 (neotype, MO; isoneotype, F). Fig- ure 4 ~“ Erect ог clambering (lianalike) shrub to 7 m mostly 10-30 cm long, 6-22 cm 3-16 cm long, pubescent like the stems; blades wide; petioles mostly cordate, rarely subcordate, the margins ir- regularly dentate to nearly entire. Flowers mostly axillary and single along the upper stems, erect or occasionally pendulous; corollas pink to bright car- mine, 48—70 mm long, at least some of them lobed, the apical portion usually flaring at anthesis and the staminal column usually exserted for 10-30 mm. Calyces 18-30 mm long in fruit, usually densely pubescent with long, stellate, yellowish hairs, the associated bracts 5 or 10, 10-18 mm long, 4-8 mm wide, broadly elliptical-ovate, widest at or near the middle, sparsely stellate pubescent, the (white), 8 mm high, 15-30 mm diam., fleshy at maturity. Distribution (Fig. 3). Northwestern South America, along the Pacific Coast tributaries of west- ern Colombia, and also in a restricted area on the eastern slopes of the Andes, along the Napo River in Peru; flowering all seasons. Diagnosis and relationships. Malvaviscus williamsii, because of the very broad imbricate bracts (Fig. 4), is readily distinguished from the seemingly sympatric M. concinnus. The former, however, occurs mostly along the Pacific Coast tributaries, although the type is from the eastern slopes of the Andes, mainly along the Napo River in Peru (Fig. 3). Material from the latter region is essentially indistinguishable from that of the Pacific coastal region. Populations of M. concinnus are not known to occur with or especially close to M. williamsii, nor does the principal character (bract shape), which distinguishes between the two, ir- reverentially crop up hither and yon, as occurs with most other “diagnostic” characters in the genus Malvaviscus. Most specimens of M. con- cinnus have linear or linear-lanceolate bracts (mostly 1-3 mm wide), which are broadest at or near the base; occasional specimens, however, have somewhat broader bracts (3-5 mm) and may be said to approach the character state found at the type locality of M. williamsii. Because of this, some workers might prefer to accord the taxon only varietal rank, as was done by Schery (1942). But among the species of Malvaviscus, it is as distinct and probably more so than M. achanioides evaluation and treatment of the M. arboreus com- plex of North America. Additional d examined. COLOMBIA. ANTIOQUIA: Mpio. de Turbo, carretera Tapon del ее 10-20 m, Brand y Narváez 620 (MO, HUA); Chigorodo, ca. 15 km W of Chigorodo, ca. 100 m, 14 Mar. 1962, Feddema 1893 (NY); Mpio. Chigorodo, Vere- da Malagon, ca. 20 m, 11 Feb. 1986, Renteria 4505 (MO); Mpio. Uraba, ca. al Río Chado, 70-90 m, Feb. 1950, Uribe 2024 (US— 2 sheets). cHoco: Mpio. de Rio Sucio, orillas del Rio Truando, entre la confluencia de los Rios Chintado y Salado, Romero-Castaneda 6121 (МО); orillas del Rio Truando, 24 Oct. 1956, Romero- Castaneda 4074 i 1967, Duke 9992 (MO); area of Baudo on Rio Baudo, ca. 5 m, 3 Feb. 1967, Fuchs et al. 21725 (MO, US); area of Baudo, 11 Feb.-29 Mar. 1967, Fuchs & Zanella 22225 (US— 2 sheets); 8-10 km E of Tutunendo, 150 456 Annals of the Missouri Botanical Garden FIGURE 4. m, 14 June 1982, Gentry & Brand 36872 (MO); between Bolivar and Quibdo, 290 m, 8 Nov. 1983, Juncosa 1338 (MO). PERU. LORETO: Gamitanacocha, Rio Mazan, 100- 1935, Schunke-Vigo 293 (A, F, NY, 1 1 i San Francisco, 130 m, 8 Mar. 1978, Díaz & Jaramillo 129 ‚ MO). NARIÑO: between Tumaco and El Diviso, along Rio Mojada, 150 m, 8 Jan. 1956, Vogel 32 (US). Malvaviscus williamsii (Feddema 1893, US). EXCLUDED NAMES Malvaviscus palmatus Ulbr., Verh. Bot. Vereins Prov. Brandenburg 1908: 89. 1909. TYPE: Brazil. Minas Gerais: “am oberen [Rio] Ju- rua,” 25 Apr. 1901, Е. Ule 5443 (holotype, B destroyed?; isotype, MG). Volume 80, Number 2 1993 Turner & Mendenhall 457 Revision of Malvaviscus Schery (1942), not having examined type ma- terial, accepted this species in Malvaviscus with reservation, noting, “from description only, it is difficult to say with surety that this species may not belong in the genus Pavonia; yet Ulbrich's description of the fruit and his excellent illustration of the type specimen indicate that this plant prob- ably is a Malvaviscus. On the other hand, the original description does not state that the petals are auriculate, from which the inference is that the plant may be Pavonia." Examination of the type shows the fruit, which was described by Ulbrich as "niger baccatus," to be rather rigidly, or not clear- ly, bacculate, having five well-defined fused carpels each with a small, smooth, rounded ridge extending down its length, the whole appearing brownish red. In short, the fruit is not clearly black and fleshy as described, although this might have been so in living material; there is no indication that Ulbrich observed fresh fruits, and our observations leave the matter moot. Regardless, dissection of the co- rolla did not reveal any clearly defined auricles on the petals, and because of this and the seeming absence of stellate pubescence and the markedly palmately lobed leaves, we here transfer the species to Pavonia. Pavonia palmata (Ulbr.) B. Turner & M. Men- denhall, comb. nov. Malvaviscus palmatus Vereins Prov. Brandenburg 9. TYPE: Brazil. Minas Gerais: *am oberen [Rio] Juruá," 25 Apr. 1901, E. Ule 5443 (holotype, B destroyed?; isotype, MC). LITERATURE CITED BAKER, E. С. 1899. Notes on Malvaviscus. J. Bot. 37: 44-348 76. Chromosome numbers in the Mal- 1966. Proposals on Hibiscus and Mal. vaviscus. Taxon 15: 163-164. DascuPTA, А. € R. P. Внатт. 1976. In: A. Lóve (editor), IOPB chromosome number reports LIII. Taxon 25: у 1. Cytotaxonomy of Мајуа сеае П. Clüomotome numbers and karyotype mal yses of Thespesia, Hibiscus, UN Pavonia a Malachia: Cytologia 46: 149- & El of Malva ceae III. Meiotic studies of Hibiscus, Abelmosc hus Azanza, Thespesia, secum Urena, and Pav nia. ч 47: 109-11 FRYXELL, P. А 88. ты аа of Mexico. Syst. Bot. : 1892. Malvaceae II (Pavonieae, Hibisceae). In: C. F. i von Martius (editor), Flora Brasiliensis 12(3): 45 5. KRISHNAPPA, D. G. & Me 1980. In: A. Lóve (editor), IOPB chromosome number а LXVIII. Taxon 29: 535. 82. In: A. Lóve (editor), IOPB chromosome number m LXXVI. Taxon 31: 582. RoBvNs, A. 1966. Malvaceae. In: R. E. Woodson & . W. Schery (editors), Flora of Panama. Ann. Mis- souri Bot. Gard. 52: 497-578. SCHERY, R. W. 1942. yo e У аи Апп. Missouri Bot. Gard. 29: Е 35. т ани ы in the Mal- vaceae I. J. Genet. 31: 263-296. SPRAGUE, T. А. 1927. Sessé and Mocino’s Plantae Novae е and Flora Mexicana. Kew Bull. 1926: 417 STAFLEU, F. ч R. Cow AN. 1981. ues liter- ature П, vahine 3. Regnum Veg. 105: 0 STANDLEY, P. C. 1923. Malvaceae. In: Анан me shrubs of Mexico. Contr. U.S. Natl. Herb. 23: 746-786. . STEYERMARK. 1949. Malvaceae. Flora of Custemala; Fieldiana, Bot. 24(6): 324-380. THE RISE AND FALL OF PSEUDOLUDOVIA ANDREANA (CYCLANTHACEAE)! 2 Roger Eriksson? ABSTRACT eudoludovia pipe described in a monotypic genus, is ae on a mixed collection. The element here eed as lectotype is conspecific with Sphaeradenia lauchea indreana is reduced to a synonym of S. P laucheana, and Pseudoludovia i is placed into synonymy under a Sphaeradenia laucheana is neotypified. In the most recent monograph of the Cyclan- thaceae, Harling (1958) recognized 11 genera. Since then, two additional genera have been de- scribed, Chorigyne (Eriksson, 1989) and Dian- thoveus (Hammel & Wilder, 1989). The existence of the monotypic genus Pseudoludovia Harling has been questioned lately, however, and the genus has not been accounted for in recent investigations of the Cyclanthaceae (Wilder, 1978, 1981; Eriksson, 1989; Hammel & Wilder, 1989). An André collection (no. 2298) from rather high altitudes in central Colombia, with a fertile speci- men deposited at K and a sterile duplicate at NY (Harling, 1958), gave rise to Pseudoludovia an- dreana Harling. A sterile specimen from southern Colombia was also cited in the original description. The genus combined characteristics of both dovia and Sphaeradenia. The leaves of these col- lections are entire, the habit is climbing as in Lu- dovia, and the young infructescence has united pistillate flowers with one apical placenta, con- crescent styles, well-developed tepals, and asym- metrical staminate flowers as in Sphaeradenia. Superficially, the vegetative and fruiting parts o the material in K seem to be connected (Fig. 1). In 1974 in southern Colombia, Wilder and Her- nandez collected live, sterile material of a species with entire leaves, which resembled those described for Pseudoludovia andreana (Wilder, 1978). In cultivation, part of this material developed inflo- rescences that clearly showed it to belong to the genus Ludovia. At that time Wilder (1978) ques- tioned the existence of Pseudoludovia, due to the similarity of the leaves. Furthermore, he suggested that if Pseudoludovia andreana was based on a mixed collection, the infructescence should be se- lected as type material. Consequently, he described his material as the new species Ludovia bierhorstii G. Wilder. Later he had the characteristics of the type of P. andreana in K described to him, which supported his earlier view (Wilder, 1981). During a visit to the herbarium at Kew, I ex- amined the type of Pseudoludovia andreana and concluded that this taxon is based on a mixed collection. This conclusion is based on the following: (1) The leaf is very similar to those of the later described Ludovia bierhorstii. (2) The young fruit- ing spadix has the same morphology as those of Sphaeradenia laucheana (Mast.) Harling. (3) The infructescence is not connected to the vegetative parts, but has been mounted to simulate connec- tion. The fractures of the pieces do not match, an the epidermal cells differ somewhat between them. (4) The part interpreted by Harling as the lower part of the peduncle (Fig. 1, below arrow) is most certainly a stem fragment, which would be consis- tent with а lianalike habit. This being a peduncle, the plant would have a terminal inflorescence. This condition is unknown in the Sphaeradenia group, to which Sphaeradenia and Ludovia belong and to which Pseudoludovia was assigned. (5) No fur- ther fertile material of Pseudoludovia has ever been found, although it has been looked for. All fertile collections with leaves like those described or P. andreana have turned out to be L. bier- horstii. Furthermore, 5. laucheana has recently been collected in the vicinity of the type locality of P. andreana. (6) A line on the label of André 2298 reads “fol. . . . ad арс. bifid.” [leaves ... towards apex bifid], which is not the case with the leaf on the sheet. However, the statement would fit a Sphaeradenia speci I think it is most probable that the mixture has been made after the material had been sent from ' I thank Lennart Andersson for reading the палаги and suggesting improvements, and Gunnar Harling and George Wilder for sharing their of view o points o the subject. * Department of ош Botany, Universo of Goteborg, Carl Skottsbergs Gata 22, 5-413 19 Goteborg, Sweden ANN. Missour! Вот. Garb. 80: 458-460. 1993. Volume 80, Number 2 1993 Eriksson 459 Pseudoludovia andreana South America, since L. bierhorstii is known only from the lowlands of southern Colombia and north- ern Ecuador and not from the area stated on the label. According to the /nternational Code of Botan- ical Nomenclature (ICBN) articles 7.4 and 9.2 1988), one of the elements of this mixed collection must be selected as lectotype. (Greuter et al., Harling's description is based on both elements and gives no indication to which the name should be attached. Wilder has typified the taxon to which the leaf belongs with other material (Wilder, 1978), and his name Ludovia bierhorstii has been ac- cepted and used. Designating the leaf as type would reduce this name to a synonym. I therefore select the infructescence of André 2298 (K) as lectotype for Pseudoludovia andreana. This is in accor- (Greuter 1988). Accordingly, the specimen labeled as André 2298 at NY, consisting of only vegetative dance with ICBN recommendation 78.5 et al., parts, should not be regarded as a duplicate of the type collection. Pseudoludovia andreana, typified with the in- 2298 (K), clearly belongs to the earlier i genus Sphaeradenia (Har- ling, 1954). lobes of the staminate flowers, the slightly elongate fructescence of An The well-developed adaxial perianth anthers without secretion globule, the free and cus- pidate tepals that are rather thick at base and have an apical glandule, and the stigmas that are lan- ceolate seen from above and do not surpass the tepals in height fall well within the variation range of, and show that it is conspecific with, Sphaerade- nia laucheana. The latter name is earlier and has priority (ICBN 57.1, Greuter et al., 1988). Con- sequently I reduce P. andreana to a synonym of S. laucheana, leaving the genus Pseudoludovia to fall into synonymy under Sphaeradenia. Sphaeradenia laucheana (Mast.) Harling, Acta Horti Berg. 17: 3. 1954. Salmia laucheana Hort. Sander. ex Mast. Carludovica Gard. Chron., ser. 3, laucheana , Curtis's Bot. g. 1 . ТҮРЕ: 27 May 1897, Hort. Kew s.n. (neotype, selected here, K). Pseudoludovia andreana Harling, Acta Horti Berg. 18: 40. Y E 5 Beg о: © © — ec Risaralda: La s Cru | es, 2298 a (lectotype, selected here: K). Sphaeradenia laucheana (Mast.) Harling was imported from Antioquia, Colombia, by Sander & Co. of St. Albans (Hooker, 1899) and was pre- n отр Pra доб Ir Ce tm абл. Identified Ьу С. Hanse and cited by him in his Monograph of the Family Cyclanthacea. de Cote HERBARIUM KEWENSE. FIGURE 1. The mixed type specimen of Pseudolu- dovia andreana in K. The arrow indicates where the different parts have been mounted together. sented at the Ghent exhibition in Belgium in 1893. It did not receive any prize, however, as did the cyclanth Ludovia lancifolia Brongn. Gardeners’ Chronicle reported on the exhibition and included an unsigned description of this plant under the name Salmia laucheana Hort. Sander (Masters, 1893). This description must be ascribed to Masters, the editor of the magazine (Stafleu & Cowan, 1981). The beginning of Masters's de- scription reads ““Salmia laucheana, Hort. Sander— A cyclanth of the genus Carludovica ...” Ac- E NE to ICBN 34.2 (Greuter et al., 1988), the assignment to Carludovica does not invalidate the name Salmia laucheana. A clone of this plant was received by Kew in 1895; it flowered for the first time in May 1897, when a specimen was pressed. Hooker (1899) re- described this cultivated material as Carludovica laucheana and provided an excellent illustration. He also cited Salmia laucheana from Gardeners’ Chronicle. 460 Annals of the Missouri Botanical Garden Harling (1954) transferred C. laucheana to the new genus Sphaeradenia. As basionym he cited the name from Salmia laucheana as a Hooker's description, regarding "nomen seminudum’ (Harling, 1958). However, the description in Gar- deners’ Chronicle is rather exhaustive with respect to vegetaüve structures, and the name is thus valid- ly published. Harling's incorrect reference to the author of the basionym does not invalidate the combination made by him; it should be regarded as a bibliographic error (ICBN 33.2, Greuter et al., 1988) There is no original material of 5. laucheana, and, accordingly, a neotype must be selected. Har- ling accredited Hooker with the basionym, and therefore he и Hr in the pressed spec- imen, Hort. Kew May 1897 (K), as the holotype. Housses а this specimen was рге- pared from a clone of the original material, I select it as neotype for Sphaeradenia laucheana. LITERATURE CITED ERIKSSON, R. Chorigyne, a new genus of the Cy P nd from Central America. Nordic J. Bot. Tais 1988. AS alan bei of Bo- -328. Mori a new aa | ү eem Ann. Missouri Bot. Gard. 16: HARLING, ri eH Leine n a new genus of the Cyclanthaceae. Acta Horti Berg. -6. 1958. Monograph of the Cyclanthaceae. Acta Horti Berg. 18: 1-428, pls. 1 Hooker, J. D. 1899. A [M Bot. ag., ser. 3, vol. 55, tab. 7683 Masters, M. T. 1893. New plants [to be shown at Ghent]. Gard. Chron., ser. 3, 13: 433-464. STAFLEU, F. А. & S. Cowan. 1981. Taxonomic literature, 2nd edition. Regnum Veg. 105: 1-980 WILDER, G. 78. Two new species and a new subgenus of Cyclanthaceae. J. Arnold Arbor. 59: 74-102. Morphology of adult leaves in the Cy- clanthaceae (Monocotyledoneae). Bot. Gaz. 142: 564— 588. GREUTER, € ET AL. CHROMOSOME CYTOLOGY IN Peter Goldblatt, Masahiro Takei,’ TROPICAL AFRICAN GLADIOLUS (IRIDACEAE)! d Z. А. Razzaq' ABSTRACT Original chromosome counts been found to depart from this pattern, the only С. atropurpureus has x = 12 (2n = 24 + х=11 (28 = 22); G. unguiculatus has x = 13 (2п = has 2n = 28. Because on morpho оа С. bellus, 2 of 2 39 in two соме of С. no of d irrespective of base number species, G. melleri (2n = 30) and material, ca. flowering early in the growing season for 16 species of tropical African Gladiolus, 5 southern African, and | Eurasian is pro ably the ancestral basic number for known examples of dysploidy 1 in the genus. In subgenus Gladiolus, 0-5B, 36); G. serapiflorus, G 26 + 0 ological grounds most of these species do not se > gregarius, and G. p 0-2B, also possibly 2n = 24); а seudospicatus have nd G. hos thus em to be immediately related, we = 60, indicates polyploidy in only one more tropical African species of Gladiolus. Counts decoratus suggest triploidy based on x = 13, m ysploidy for the genus. Although the chromosomes of Gladiolus are ПА small, some details provide additional data relating to species relationships. Measurements o aking this one more possible tal Бака iia length sho most of the diploid species have a comparable amount of chromosome material. Two G. decoratus (2n = 39 33-4095 above expected levels, a condition possibly related to their characteristic rapid growth an wever, have significantly higher amounts of chromosome Ever since the first chromosome counts (Vil- morin & Simonet, 1927; Ernst-Schwarzenbach, 1931; Bamford, 1935) were made in Gladiolus, the largest and most taxonomically complex genus of Iridaceae subfamily Ixioideae (Goldblatt, 1991), x = 15 has been considered the probable ancestral base number for the genus (Goldblatt, 1971). Ex cept for the occasional occurrence of B chromo- somes, the several Eurasian, southern African, and Madagascan species counted all have diploid num- bers based on x — 15 and the only significant cytological variation reported in Gladiolus has been polyploidy. Eurasian species are consistently poly- ploid (van Raamsdonk & de Vries, 1989) while most southern African and all Madagascan species examined are diploid. Until now, only two endemic tropical African species have been studied cyto- logically (Table 1) and these have corresponded to the Afro-Madagascan pattern in being diploid, 2л — 30. A third, C. dalenii, which also occurs in Arabia, South Africa, and Madagascar, is polyploid in Africa, 2n — 45, 60, 75, and 90. Counts here for 16 tropical species of Gladiolus, 12 never before counted, at least for tropical Africa, and an additional five southern African species, confirm that while x = 15 is the probable ancestral base number, important variation occurs in some trop- ical species. Descending dysploidy, probably in four separate lineages, has led to the establishment of secondary base numbers, x = 14, 13, 12, and 11. Polyploidy is confirmed in G. dalenii in tropical Africa (the species is diploid in Madagascar). One more tropical species, G. bellus, appears to be polyploid. Gladiolus, a genus of some 215-220 species, remains poorly sampled cytologically, and in view of the variation recorde investigated in more detail. here, needs to be MATERIALS AND METHODS Seeds or corms of wild collected species (Table 1) were sprouted in the greenhouse, and emergent leaf and root tips were harvested for analysis of mitosis in one of two ways. For rapid determination ANN. MISSOURI Bor. САВР. 80: 461-470. 1993. 462 Annals of the Missouri Botanical Garden TABLE 1. Chromosome numbers in sub-Saharan African and Madagascan Gladiolus. For comparative purposes, some Eurasian species were also examined, but past counts for the area are not reviewed. Original counts are indicated in bold type and provided with collection data; bd published counts are referenced, and nomenclature adjusted o reflect current taxonomy (Lewis et al., 1972; Goldblatt, 1989). Unless ds indicated, voucher specimens are ned at MO Diploid Species number Collection data or reference Southeastern and Tropical Africa and Madagascar C. actinomorphanthus 28 Zaire, Shaba, Schaijes 5090, Schaijes 5152-64 Duvign. & Bokstael G. aequinoctialis Herbert 30 Harvey, 1966 (as Acidanthera) 30 Cameroon, Mt. ме Boussard s.r G. atropurpureus Baker 24 + 0-5B Zaire, Schaijes 5178-81, Schaijes 108. Malawi, Mt. Malosa, 36 np no voucher G. Е Goldblatt 30 Goldblatt, 989 G. bellus C. H. Wright 60 Malaw ‘Mt EU Oliver s.n. G. bojeri (Baker) Goldblatt 30 Goldblatt, G. callianthus Marais 30 Sharma & le 1960; Bannerjee & Sharma, 1971; Gold- blatt, 1971; Ohri & Khoshoo, 1985 (all as Acidanthera bi- olor G. crassifolius Baker 30 Bamford, 1935; Goldblatt, 1971 30 Malawi, Mulanje, Chapman 8364; Tanzania, Mufindi, Lovett 3276 30(-31) Zimbabwe, Piso Goldblatt 9077 G. dalenii Geel 60 Vilmorin & Simonet, 1927; Ernst-Schwarzenbach, 1931; Bam- ford, 1935; Van 1939; Ohri & Khoshoo, 1985 (all as G. primulinus), Bamford, 1935 (as С. coccineus 60 Bamford, 1935 (as G. platyphyllus) 75 Bamford, 1935 (as G. quartinianus) 75, 90 Bamford, 1935 (as G. dracocephalus) 75, 90 Bamford, 1935 (as G. psittacinus) 45, 60, 75 Goldblatt, 1971 (as G. natalensis var. natalensis and var. coo- rt) 30 Goldblatt, 1989 90 Paiva & Leitao, 198 60 Zimbabwe, e си s.n., no voucher 90 Ethiopia, purs Ryding 1488 (UPS) G. decaryi Goldblatt 30 + 0-2B Goldblatt 30 Madagascar, M ИК Puff 800803-1/5 G. decoratus Baker 39 ie Zomba, Goldblatt 9098; Luchenza, la Croix s.n., no voucher G. erectiflorus Baker 30 Malawi, Dedza riage Goldblatt 9091, Goldblatt 9105; Zaire, Shaba, Scha G. gregarius Baker 22 Central African ie Harris & Fay 2241; Tanzania, Mu- bes Lovett 4622; Tanzania, Njombe, Hann 600; Zaire, n et al. les G. horombensis Goldblatt 30 E )89 G. laxiflorus Baker 30 Tanzania, Bon Goldblatt et al. 8050 G. longicollis var. platype- 30 Goldblatt, 1971 (as G. praelongitubus) talus (Baker) Oberm. G. luteus Lam. 30 Goldblatt, 1989 G. melleri Baker 30 Tanzania, Ufipa, Goldblatt et al. 8135; Zaire, Shaba, Schaijes 24 123 G. papilio Baker hybrid 75 Bamford, G. pseudospicatus Cor- 22 aire, us | et al. 199 dova G. saundersii Hook. 30, 45 Bamford, 1935 30 South Africa, Naudes Nek, Batten s.n. G. serapiflorus Goldblatt 22 Zaire, Shaba, Schaijes 5137-42 Volume 80, Number 2 Goldblatt et al. 463 1993 Chromosomes in African Gladiolus TABLE l. Continued. Diploid Species number Collection data or reference G. sericeo-villosus Hook. sp. sericeovillosus 30 South Africa, Natal, Goldblatt & оа 8430 su subsp. calvatus (Baker ) 30 Zimbabwe, Lomagundi, Goldblatt Goldblatt G. varius Bolus f. 30 Goldblatt, 1971 G. unguiculatus Baker 26 Zaire, Shaba, Schaijes 5143, 5146, 5148 24 Zaire, Shaba, Schaijes 5150 26 + 2B Zaire, Shaba, Schayes 5123 24, 26 Zaire, Shaba, Schaijes 5124 G. watsonioides Baker 30 Kenya, Mt. Kenya, Миа G. zambesiacus Baker 30 Malawi, Mt. Zomba, Goldblait 2092, Goldblatt 9099 G. zimbabweensis Gold- 30 Zimbabwe, Nyanga, Goldblatt blatt Cape species (winter rainfall area of southern Africa) G. abbreviatus Andrews 30 Goldblatt, 1971; de Vos, 1976 (as Homoglossum) G. alatus L. 30 Bamford, 1935 G. Mene s L. 30 Bamford, 1935; Verryn, 1988 С. эче spei Goldblatt & 30 Goldblatt, 1971; de Vos, 1976 (as Homoglossum merianellum) de G. i ifolius Ja Jac 30 Bamford, spen, Goldblatt, 1971 G. buckerveldii "| Bolus) 30 Goldblatt, 1971 Goldblatt G. cardinalis Curt. 30 Ernst-Schwarzenbach, 1931; Brittingham, 1934; Bamford, 1935; Nakajima, 1936 G. carmineus C. H. 30 Ва 1935 Wright G. carneus Delaroche 30 Bamford, 1935 (as G. blandus, G. pappei, and G. callistus); Jerryn, 1988 С. caryophyllaceus 30 Bamford, 1935 (as G. hirsutus) (Burm. f.) Poir G. ceresianus L. Bolus 30 Goldblatt, 1971 G. cunonius (L.) Gaertn. 30 Goldblatt, 1971 (as .4nomalesia cunonia) G. debilis Ker 30 Bamford, 1935 C. equitans Thunb. 30 Bamford, 1941 (as G. namaquensis) С. a Jacq. 30 Verryn, 1988 G. gracilis Jacq. 30 Bamford, 1935 G. gueinz ii Kunze 30 Goldblatt, 1971 30 South Africa, eastern Cape, Goldblatt 9052 G. huttonii (N. E. Br.) 30 de Vos, 1976 (as Homoglossum) Goldblatt & de Vos G. lapeirousioides Gold- 30 Goldblatt, 1974 blatt C. leptosiphon Bolus f. 60 South Africa, Cape, Cockscomb, Bongers s.n. (МВС) G. liliaceus Houtt. 30 Bamford, 1935 (as G. grandis) G. maculatus Sweet 30 Bamford, 1941 G. martleyi L. Bolus 30 Bamford, 1941 G. odoratus L. Bolus 30 Bamford, 1935 C. oppositiflorus Herbert 30 Bamford, 1935 46 Ernst- ra h, 1931 G. orchidiflorus Andrews 45 Bamford, 19 G. Aegean Delaroche subsp. permeabilis 30 Bamford, 1935 o ulis (Burch. ex 30 Nordenstam, 1969; Goldblatt, 1971 (as G. edulis) Ker) Oberm. 464 Annals of the Missouri Botanical Garden TABLE |. Continued. Diploid Species number Collection data or reference G. pritzellii Diels 30 Goldblatt, 1971 G. priori М. E. Br. 30 Goldblatt, 1971 (as Homoglossum) G. punctulatus Schrank 30 Bamford, 1935 (as G. villosus and G. hirsutus) G. quadrangulus Dela- 30 Verryn, 1988 roche 30 South em Ls Flats, Goldblatt 525 (BOL) G. recurvus L. 30 Bamford, C. saccatus (Klatt) Gold- 30 Goldblatt, Da (as Anomalesia) blatt & de Vos G. scullyi Baker 45 Bamford, 1935 (as G. formosus) G. sempervirens G. Lewis 30 Bamford, s (as G. splendens) G. tenellus Jacq. 30 G. teretifolius Goldblatt & 30 datus); Goldblatt, 198 e South Africa, Cape, Bokkeveld ie mE Goldblatt 4984 Bamford, 935 (as G. trichonemifolius) de Vos, p 6 (as Homoglossum muirii) Ernst-Schwarzenbach, 1931; Brittingham, 1934; Bamford, ; Mensinkai, 1939; Ohri & eri 1985 1935; Ernst-Schwarzenbach, 1971; Verr l (also as G. cuspi- 1941 1941; Goldblatt, 1971; de Vos, 1976 (as Homoglos- Iran, Qushi Pass, Goldblatt 850 cult. uri MM Garden, Goldblatt s.n. Кане de Vos G. tristis L. 30 1935 G. undulatus L. 30 Шашы, G. uysiae L. Bolus ex С. 30 Lewis С. watermeyeri L. Bolus 30 Bamford, G. watsonius Thunb. 30! Bamford, sum) G. atroviolaceus Boiss. 60 G. byzantinus L. 90 Morocco, G. kotschyanus Boiss. 60 Iran, Baneh, of 2n = 66 for G. watsonius (as Homoglossum) by Sharma & Talukdar (1960) are almost certainly based on misidentified plants. of chromosome number, root tips were pretreated in saturated aqueous m-bromonaphthalene for 4 hours at room temperature, then fixed in 3:1 ab- solute ethanol-glacial acetic acid for 5 minutes. After hydrolysis in 10% HCI at 60°C for 6 minutes, tips were squashed in FLP orcein (Jackson, 1973). For more critical analysis, leaf and root tips were pretreated in 0.002 M aqueous 8-hydroxyquinoline for 4-5 hours at room temperature, then fixed in modified Carnoy's solution (absolute ethanol-glacial acetic acid-chloroform 6: 3: 1) at 8°C for about 24 hours. Both root and leaf tips were stained in Feul- gen reagent and squashed in 1% aceto-orcein. Pho- tographed preparations were all made using the latter technique. OBSERVATIONS Chromosomes of Gladiolus are relatively small, 0.7-2.9 um n 1-21), and diploid numbers are based on x = 15 (Table 1). Several tropical African species differ in base number. The common tropical African species n the majority of species examined G. gregarius has 2n = 22, based on counts from four populations as far apart as Central African Republic, southern Zaire, and Tanzania; closely related G. pseudospicatus, centered in southern Zaire and most common on soils with high con- centrations of heavy metals, also has 2n — 22 (Figs. 1, 2). Gladiolus serapiflorus has 2n = 22, based on several counts from a single population from Zaire (Fig. 3). The widespread G. atropur- pureus appears to have x — 12: two populations rom Zaire have 2n — 24, sometimes with B chro- mosomes ranging in number from 1 to 5 (Figs. 4, 5); and three individuals in a population from Ma- lawi are triploid, 2n — 36 (Fig. 6 The widespread and common Gladiolus unguic- ulatus has proven cytologically puzzling. While most plants sampled from two populations in south- western Zaire have 2n = 26 (or 2n = 26 + 2B), some individuals have 2n — 24 (Figs. 7-9). The Zairean endemic, G. actinomorphanthus, has 2n — 28 (Fig. 10) in two populations examined. Karyotypes of most species comprise metacen- tric to submetacentric chromosomes (symmetrical Volume 80, Number 2 Goldblatt et al. 465 1993 Chromosomes in African Gladiolus FIGURES 1-9. Mitotic metaphase in Gladiolus. —1. С. p (2n = 22).—2. G. gregarius (2n = 22).— 3. С. serapiflorus (2n = 22). — 4. G. atropurpureus (2n = 24).—5. С. atropurpureus (2n = 24 + 2B).— 6. С. atropurpureus (2n = 39). — 7. G. unguiculatus (2n = 24).—8. С. unguiculatus (2n = 26). —9. G. unguiculatus 2n = 26 + 2B). B ко. are indicated by the arrows. Scale bar = 5 um. ~ 466 Annals of the Missouri Botanical Garden | | | E A ~ | e “e 1 ,* v y" ô Д + LO Ld 11. С. к! FicunES 10-18. Mitotic тим ва. in Gladiolus. — 10. G. actinomorphanthus (2n = 28). — (2n = 30).—12. С. e = 31).— 13. 6. decaryi (2n = 30).—14. G. zambesiacus (2n = mbabweensis (2n — -— l6. G. melleri (2n = 30).—17. G. gueinzii (2n = 30). — 18. С. pni cma ini calvatus (2n — m | bar = 5 um Volume 80, Number 2 1993 Goldblatt et al. Chromosomes in African Gladiolus 467 FIGURES 19-21. 21. G. byzantinus (2n — 0). Scale bar = 5 in arm ratio) of fairly uniform size. All chromo- somes have a large heterochromatic segment that stains darkly at early prophase. In Gladiolus gre- garius, G. pseudospicatus, and G. decoratus, each with the longer chromosomes 2.1-2.7 шт long, thus exceeding the average, the size range is con- tinuous. However, bimodal karyotypes with one or two conspicuously longer chromosome pairs 2.4— 2.9 um long, characterize G. atropurpureus, G. serapiflorus, and the 2n = 2 unguiculatus (Figs. 3-5). The Mediterranean Gladiolus byzantinus, 2n — 90, included in this study for comparative purposes, also has a bimodal karyotype (Fig. 21), but only one pair of longer chromosomes despite its hexaploid karyotype. cytotypes of G. otal chromosome length, a comparative mea- sure of genome size, is similar in most of the diploid species (Table 2), irrespective of base number, and most species have between 31.4 and 39.5 um of chromosome material. This supports our view that dysploidy alone has been responsible for changes in base from x — 15 to 11. Two species, G. melleri and 6. decoratus, have significantly higher amounts of chromosome material, 33-40% above expected levels. We suggest that this may be related to their comparatively rapid growth patterns early in the growing season. Both complete their flowering by the end of December, or earlier, and within the first few weeks of the rainy season. Other species Mitotic metaphase in Gladiolus. —19. G. decoratus (2n = 39).— 20. G. bellus (2n = 60).— um with similar growth patterns such as G. unguicu- latus and G. atropurpureus do not have compa- rably enlarged genomes (Table 2). There is usually one pair of satellites per chro- mosome complement in diploid species (2n = 22- 30). In G. atropurpureus, however, there are con- sistently two pairs of satellites (satellite number in the presumed triploid plants is uncertain). In the population of G. decoratus that we examined, also presumably triploid, the karyotype includes three satellite chromosomes. Polyploidy is reconfirmed for tropical African Gladiolus dalenii and for the Eurasian G. byzan- tinus. Ап Ethiopian population of С. dalenii has 2n = 90, and one from Zimbabwe has 2n = 60. The Malawian endemic G. bellus, not before count- ed, has 2n — 60 (Fig. 20) and must be assumed to be a tetraploid species. Counts of Zn — 39 in G. decoratus (Fig. 19), obtained from corms of three individuals of one population from southern Malawi, and from one corm of a second population, suggest triploidy and a base number of x = 1 the species. This awaits confirmation when more material becomes available. Counts for both Asian species examined, G. kotschyanus and G. atro- violaceus, 2n = 60, are consistent with polyploidy in other Eurasian species of Gladiolus, but the only other count for the latter, 2n — 45 (van Raamsdonk & de Vries, 1989), is puzzling. 468 Annals of the Missouri Botanical Garden TABLE 2. Total length (um) of metaphase chromosome complement of selected species of Gladiolus, mostly from tropical Africa. Bimodal karyotypes are indicated by X. Figures in parentheses for G. unguiculatus refer to voucher numbers in Table 1 iploi number 27 Total chromo- some length Length of Taxon Size range um um asic set um Bimodality G. actinomorphanthus 28 1.8-0.9 34.7 17.4 — С. atropurpureus 24 2.3-1.3 41.4 20.7 — 36 2.9-1.1 56.9 18.9 X G. bellus 60 1.4-0.7 67.3 16.8 — G. crassifolius 30 1.4-0.7 31.3 15:7 — G. decaryi 30 1.8-0.8 34.6 17.3 X G. decoratus 39 2.6-1.6 77.3 26.4 = G. gregarius 22 2.4-1.4 37.6 18.8 — G. gueinzii 30 2.2-1.0 39.5 19.8 — G. melleri 30 2.2-1.2 48.9 24.5 E G. + 22 2.1-1.1 34.4 17.2 — G. serapiflor 22 2.5-1.2 33.3 16.7 X G. iia subsp. calvatus 30 1.5-0.8 31.4 15.7 — 6. unguiculatus (5124) 26 2.4-1.1 41.0 20.5 X (5148) 26 2.5-1.1 39.7 19.9 X (5150) 24 2,0-1.0 34.0 17.0 — G. zambesiacus 30 1.4-0.9 32.7 16.4 X B chromosomes, reported in Gladiolus previ- ously only in the Madagascan G. decaryi (Gold- blatt, 1989), seem common in G. atropurpureus and G. unguiculatus. Additional G. decaryi, ex- amined here, lacked B chromosomes. As is usual ith B chromosomes, they are small, ca. 0.8 um long, often vary in number within individuals, and can be recognized at prophase by their precocious and uniformly dark staining (Figs. 5, 9). Whether the variation in chromosome number in G. ип- guiculatus, either 2n = 24 or 26, the presence of a type of B or supernumerary chromosome has not been established. In plants with 2n — 26, the number is consistent in several individuals examined, and no precocious staining was noted except when the number exceede à when typical B chromosomes could be identified. We suspect that plants with 2n = 24 may hybrids, perhaps with G. serapiflorus, 2n — ee which grows at the same site. Study material was collected in fruit, and counts were obtained from germinating seeds. This makes it impossible to con- firm hybridity in the vouchers, at least until seed- lings flower in 3-4 years. Provisionally we assume that G. unguiculatus, at least in southern Zaire, is dysploid, normally with 2n = e are uncertain whether the additions] chro- mosome in one plant of a Zimbabwean population of G. crassifolius is a B chromosome or not, but is also due to other individuals examined have a normal comple- ment of 2n — DISCUSSION Including the original counts in this paper, and after bringing the taxonomy up to date, some 75 species of Gladiolus are known cytologically (Table 1). Although this seems substantial, it is less than 39% of the estimated 220 species in the genus (Goldblatt, in prep.). The first chromosome counts made in Gladiolus (Vilmorin & Simonet, 1927; Ernst-Schwarzenbach, 1931; Bamford, 1935) es- tablished x = 15 as the likely ancestral base number for the genus. Although most Cape and eastern southern African species counted were diploid, 2л — 30, these early studies also suggested that poly- ploidy was common in Gladiolus. Studies by Lewis et al. (1972) and Geerinck (1972) indicate that the several tropical and eastern southern African polyploid species recognized in these early cyto- logical studies are conspecific (Table 1), and the earliest name for the species has been shown to be G. dalenii (Hilliard & Burtt, 1979). It is now evident that polyploidy is uncommon in Gladiolus outside Eurasia, where species are tetraploid to dodecaploid, except for a puzzling count of 2n = 5 in G. atroviolaceus (van Raamsdonk & de Vries, 1989). Outside Eurasia Gladiolus is pre- Volume 80, Number 2 1993 Goldblatt et al. Chromosomes in African Gladiolus 469 dominantly diploid, and dysploidy is more frequent than polyploidy. The ancestral basic chromosome number in Gladiolus is probably x = 15, as assumed in the past (Goldblatt, 1971). The species in which dys- ploidy occurs all seem specialized, although in dif- ferent ways. Thus none seems a suitable candidate for an ancestral position in the genus in which several broad foliage leaves, broadly winged seeds, ase number of x 15 appear symplesiomorphic (Goldblatt, 1991; prep.). The species in which dysploidy occurs are branched stems, and a all small-flowered and small in stature, both features probably plesiomorphic in Gladiolus. Except for G. actinomorphanthus, they also all have clawed tepals grooved toward the base and united for a short distance, a probable synapomorphy uniting about half the species of the genus in subgenus Gladiolus. No other characters unite all the dysploid spe- cies, and we conclude that except for G. atropur- pureus and G. serapiflorus, which appear closely related, descending dysploidy has occurred inde- pendently in four lineages. This conclusion is partly supported by the karyotypes. Those of G. atro- purpureus and G. serapiflorus are bimodal with a single comparatively long chromosome pair. Mor- phologically, С. atropurpureus and С. serapiflorus seem united in their very short foliage leaves, often reduced to sheaths with vestigial blades, and coarsely fibrous corm tun Gladiolus gregarius and G. pseudospicatus have three synapomorphies, two unique in the ge- nus, long imbricate bracts and a short style with short branches, and a third, unbranched stems. Gladiolus unguiculatus also has unbranched stems but is remarkable in having flowering and vege- tative cycles at different seasons. Lastly, G. acti- nomorphanthus, which seems primitive in having actinomorphic flowers, belongs to an alliance of species adapted to unusual soil conditions that have terete 4-grooved leaves and seeds in which the wings are reduced to vestigial ridges. Polyploidy has evidently played a minor role in the evolution of Gladiolus, except in Eurasia and North Africa where a single polyploid event in the founding population, ancestral to all the species in the region, could account for the pattern here. Polyploidy elsewhere has been sporadic. Except for reports of triploidy in G. scullyi and G. orchidi- florus (Table 1) that need additional study, only one species in the Cape region, tentatively assigned to G. leptosiphon, is polyploid. Excluding G. dalen- ii and G. bellus, only two tropical and eastern southern African species have been reported as polyploid, G. papilio and G. saundersii, and both the latter counts are for so-called hybrids (Bamford, 1935). At least wild G. saundersii appears to be diploid. The significance of polyploidy in G. dalenii is difficult to gauge. It is one of the most successful species of the genus, and certainly the one wit the widest geographical distribution (Goldblatt, 1989; in prep.). It extends from the eastern Cape, South Africa, to Senegal in West Africa and from Madagascar to Ethiopia and Yemen. Perhaps poly- ploidy has played a role in enabling the species to thrive under the wide range of conditions that occur across its range. The role of humans in its dispersal may also be important. The species is cultivated in West Africa, and as an important medicinal plant it may have been spread over long distances by human agency. The presumed ancestral basic chromosome number in Gladiolus, x = 15, is shared in subfamily Ixioideae only with the mono- or ditypic Radi- nosiphon (Goldblatt, 1971), and the two genera are evidently palaeotetraploid. This base number is apomorphic in Ixioideae, and it is the only syn- apomorphy uniting Gladiolus and Radinosiphon. Most other genera of Ixioideae have lower base numbers (Goldblatt, 1971), x = 11 and 10 being most frequent. Other high base numbers in Ixioi- deae are x — 13 or 14 in Romulea, and 13 in Geissorhiza and Hesperan- 6 in Tritoniopsis, tha, the two genera possibly the most closely re- lated to Gladiolus after Radinosiphon. LITERATURE CITED BAMFORD, R. 1935. The chromosome number in Gladi- olus 5 Agric. Res. 51: 945-950. 41. Chromosome ipis and hybridisation in iran J. Hered. 32: 2. BANNERJEE, M. & A. K. dee 1971. A cytotax- onomical analysis of several genera of the family Iridaceae. Pl. Sci. 3: 14-29. BRirTINGHAM, W. E. 1934. Cytological studies on some genera of the Iridaceae. Amer. J. Bot. 21: 77-83. ERNST-SCHWARZENBACH, M. 1931. Contribution à l'étude des POP ai chez le | Gladiolus L. Ann Sci. Nat. Bot., sér. 10, 13: 5 GEERINCK, D. 972. Revision du genre ‘Gladiolus L. (Iridaceae) au Zaire, au Rwanda et au и Bull. Jard. Bot. Nat. Belgique 42: 269-28 GOLDBLATT, P. 1 Cytological and LAT studies in the southern African Iridaceae. J. S. Af- rican Bot. 3 i 1974. In: ee some numbers of phaner- ogams. Ann. Miss t. Gard. 61: 901-902. 989. Systematic P Gladiolus L. (Iridaceae- Ixioideae) i in Madagasc r. Bull. Mus. Nat. Hist. Nat., ér. 4, Ada пае 11: 235- 255 199]. An overview of the systematics, phy- же Bir peut p a African Iridaceae. Contrib. Bolus Herb. 470 Annals of the Missouri Botanical Garden Harvey, M. J. 1966. In: IOPB chromosome number garden Gladiolus П. Variation in chromosome com- reports 7. Taxon 15: 155-163. plement and meiotic system. Cytologia 50: 213-231. HILLIARD, O. M B. L. Bunrr. 1979. Notes on some Paiva, J. & M. T. Lerrao. Nümeros chromos- plants of southern Africa chiefly from Natal: VIII. otes Roy. Bot. Gard. Edinburgh 37: 285-325. JACKSON, R. 1973. Chromosome evolution in Haplo- pappus gracilis: a centric transposition race. Evo- 43-256. A. OBERMEYER & T. T. BARNARD. 1972. A revision of the South African dica of Gladiolus. . African Bot. Suppl. 10: MENSINKAI, S. W End studies in the genus т" Cytologia 10: 51-58. NAKAJIMA, 6. hromosome numbers in some crops and wild angiosperms. Jap. J. Genet. 12: 211- 9 NORDE NSTAM, B. 1969. Chromosome studies on South African vascular plants. Bot. Not. 122: 398-408. . N. Oum, D. € T Кнозноо. 1985. Cytogenetics of sómicos para alguns taxa da Africa tropical — П. Bol. Soc. Brot., sér. 2, 62: 117-130. RAAMSDONK, L. W. D. van & T. DE Vries. 1989. Bio- systematic studies in European прве of Gladiolus (Iridaceae). Pl. Syst. Evol. 165: 198. SHARMA, А. К. & C. TALUKDAR. 1960. romosome studies in members of Iridaceae and their mechanism of speciation. Genetica 31: 340-384. VERRYN, S. D. 1988. /n: Chromosome number reports 99. Taxon 37: 396-399. ViLMORIN, В. DE & M. SIMONET. 1927. Nombre des chro DE Vos, 976. Die Suid. Afrikaanse species van Ho omoglossum, J. S. African Bot. 42: 301-359. Constance I. Millar? IMPACT OF THE EOCENE ON THE EVOLUTION OF PINUS L.’ ABSTRACT Pinus evolved in middle latitudes of the Northern Hemisphere in the middle Mesozoic. By the late Cretaceous "is had spread east and west throughout Laurasia, attaining high diversity in eastern Asia, the eastern United States, and western Europe, but having little representation at high northern latitudes. Changing climates in the early Теш established warm and humid tropical/subtropical conditions in a broad zone to 70°N throughout middle latitudes. Pines and their relatives disappeared from many middle-latitude areas during this time and were replaced by diverse angiosperm taxa of the boreotropical flora, which were adapted to the equable, tropical climate. The effect of this climate change and spread of boreotropical flora was to displace pines from their former habitats. A hypothesis is defended that pines shifted, during the three warm periods of the Eocene, into three major refugial areas in the Northern Hemisphere: high latitudes, latitudes, and upland regions of middle latitudes, especially in western North America. Some of these refugial are Ponderosae, Contortae, and Strobi were distributed over several refugia; subsections Leiophyllae, Australes, and Cembroides evolved in southern refugia in North America; and Can in southern refugia along the Tethys seaway in Eurasia. Following the cooling and drying of the climate at the end of the Eocene, many angiosperm taxa of the boreotropical flora became extinct colonized middle latitudes, a zone they have occupied to the prese isolated lineages expanded and met and pines re nt. Migration out of refugia provided additional opportunities for hybridization and introgression, as formerly The past two decades have seen an explosion of information on the paleohistory of the Earth. Ev- idence on plate tectonics has clarified the position of continents in different ages, continental geo- morphology, and the dynamics of inland seaways and changing coastlines. Physical and biological evidence has been used to infer paleoclimates with finer resolution in time and space. New fossil dis- coveries have added to the record of past vege- tation, and new diagnostics for identifying. taxa have led to systematic revisions of many fossil floras. The widespread use of radioisotope dating has added precision to determ floras. ining the ages of fossil This information, together with phylogenetic analyses of extant taxa, has contributed new in- sights and a revised understanding of evolution for many plant groups. In pines (family Pinaceae, ge- nus Pinus L.), major syntheses have focused on two time periods in the history of the genus. Studies on the Mesozoic history of the pine family, and especially Pinus, have significantly changed our understanding of the origin of the genus (Miller, 977, 1982, 1988; Robison, 1977; Black- well, 1984; Stockey & Ueda, 1986; Stockey & Nishida, 1986). Similarly, studies on the Quater- nary history of pines have led to new interpretations about the impact of recent paleohistoric events on the genetic structure and evolutionary relationships of extant species (Critchfield, 1984, 1985) The broad-scale events that influenced the evo- lution of the genus between its origins in Mesozoic (Table 1) and its present diversity remain obscure. How did important secondary centers of pine di- versity in Mexico, western North America, and eastern Asia originate? How do these areas relate to the primary centers of origin for the genus? What events triggered the diversifications of taxa within the genus, and how have historical events influenced current and fossil distribution? Although there have been important contributions to under- standing regional biogeography and evolution of pines in the Tertiary (Eguiluz Piedra, 1985, 1988; Axelrod, 1986; Lauria, 1991), the impact of Pa- ' Т especially thank B. B. Kinloch for Meo discussion and review of the manuscript. I also thank D. Axelrod, L. Loveless, C. Miller, S. Strauss, E. Zavar nd an anonymous rev iewer for critical comments on the manuscript. I dedicate this paper to the late W. B. Critchfield (1923-1989), whose studies on the impacts of the Мечиев оп conifers demonstrated that genetic structure of e xtant species cannot be understood without looking to the ? Institute of Forest Genetics, Pacific Southwest Research Station, U.S.D.A. Forest Service, Berkeley, California 94701, U.S.A ANN. Missouri Bor. Garb. 80: 471—498. 1993. 472 Annals of the Missouri Botanical Garden TABLE 1. Approximate ages and durations of geological eras from the Mesozoic to present. uration (millions Millions of Era Period Epoch of years years ago Cenozoic Quaternary Holocene Approximately the last 10,000 years Pleistocene 2.4 2.5 Tertiary Neogene Pliocene 4.5 7 Miocene 19 26 Paleogene Oligocene 12 34) Eocene 16 54 Paleocene 1] 65 Mesozoic Cretaceous 71 136 Jurassic 54 190 Triassic 35 225 ' The Oligocene-Eocene boundary is accepted to be З 34 Ma, coinciding with the terminal Eocene event. Authors 38 Ma. publishing before the middle 1970s and some current ones accept the boundary as : leogene (Paleocene through Oligocene) events, es- pecially the Eocene, on the evolution of the genus as a whole has not been analyzed. In this paper, I attempt to synthesize recent information on plate tectonics, climate, fossils, and biogeography of pines and other dominant plant groups as they affected pine evolution. From this synthesis, I argue that the Eocene was one of the most important phases in pine evolution. SYSTEMATICS AND CURRENT BIOGEOGRAPHY OF PINUS The genus Pinus is one of the most widely dis- tributed genera of trees in the Northern Hemi- sphere. Pines occur predominantly at middle lati- tudes (30°-55°N), but important centers of pine species also exist at high (> 55°N) and low latitudes (< 30°N) (Figs. 1, 2; Critchfield & Little, 1966). Pines are abundantly represented in North Amer- ica, Central America, Europe, and Asia, with some taxa extending into northern Africa. Within their range, pines occur in diverse habitats, extend from sea level to 3,700 m, and dominate natural veg- etation in many regions. They are absent from hot, wet, tropical environments, where they are poor competitors with other taxa (Mirov, 1967; Bond, 1989). At tropical latitudes, pines occur only in uplands or semi-arid regions. Pinus contains more species than any other genus of conifers, although Podocarpus may rival it. Pines have been recognized since Classical times, and more than 40 classification systems have been proposed (Critchfield & Little, 1966; Mirov, 1967; Little & Critchfield, 1969; Price, 1989; Millar & Kinloch, 1991). The most widely accepted is the system of Little & Critchfield (1969), which built upon and modified the classification of Shaw (1914, 1924). Little & Critchfield (1969) divided Pinus into 3 subgenera, 5 sections, 15 subsections, and 94 species (Fig. 3). They updated the classification by incorporating new types of information based on genetic data, especially from contemporary studies on hybridization and biochemical variation. Thus, their classification implicitly suggests phy- logenetic relationships and common origins of spe- cies and groups of species. Since the time of Little and Critchfield’s classification, several new species have been described, especially from species-rich and as yet still incompletely known regions such as Mexico. In this paper, I accept Little and Critch- field’s authority for species, and cite authors of taxonomic names only for those taxa outside of their system. I also accept most aspects of their 473 Evolution of Pinus Millar Volume 80, Number 2 1993 "SILISIA], ÁS = AS ‘IDUDIUIQDG = BS “IDSOLIPUO] = = оа “әрән = та гора 19200) = () гр] ] Adoro = 71 “9D]10]UO7) = 0) "гагид мир) = 8) “SI]DLISNY = Y "snua3qns əy}; uyum suotnoosqns jo suoneoo| [219u93 SUIMOUS 'snuid snuesqns snuij jo чоцпа еј "I 380514 Annals of the 474 Missouri Botanical Garden ar) јориромпојјрд = -(snutdodureonq) тируа шоам = M 994115 = $ “IDUDIPADLIL) = £) 'горголдшо) = 87) 'орлашо) а 'snuogqns 941 ШҶИМ suonoasqns jo suomeo»o| [елапо8 SUIMOUS ‘snuidodwupong snuesqns pue s$nqo4lg snua3qns snuid jo uounqugjswTq "с 3809104 Ov Millar 475 Volume 80, Number 2 1993 Evolution of Pinus 'suonoesqns pue 'suonoes *e1euosqns ozur payissejo sarads əy} Яшмоцѕ (6961) PPYYNUJ 29 гр] o1 Suipioooe smuij snuas əy, jo Ашопохеј сє ANNI пиоѕме] 810208] sisueqno Pueisejbnop siejuepiooo — 5пдол5орпеза eeequeo PBUPOBOYO/W оте ибамуиву гејбиџа suebund еешпгајиош РШРООО BunoJes uedooo 166916 ерби sisueDueinp emed esnep euqe/6 ниирщәбиә РиЕЛӘЛО! gjeounui вивушблл eJeuJyos Кәшә! ueynoo ejenueye вџојисо epee} SISUBOYSEM BUEIUIQES B]EIDEJ EUEISXUEq sujsnjed Psojapuod @eUBIUIGeS — eedie2o(Q) евџиојџод səjensny ePsoJapuod Iu2i64pj8q Еби зуеодод PSOUIS8! S8J]jSaAKS gaud ӘРӘШД Iizyouunj ey Aydo¡a eejAudoie] ичбтахог SISUGUPUES S@SUBUBURD p T ДЕ MET M Зав ШЕ. зпша зпша рјецуојио pue еру РӘШӘ пиозјеи PubP8ouid 1 ггешившхеш ролш шп gesue епојирепб Eneebuo] euesbunq ѕупрә Bueunoyeq риюруеәб sepio1queo ӘРИРШПОДЕД әғивірлезәәя X SOPIOJQUED а M [28727] sngous SNNId Snuet 941 пбием виеђегивј ened [570020262 SIULJOJIQOJIS Byund зухеџ SISUBIBJOx виешедше) РОЛІ вүодциош РЈдШӘо 5п9015 syneoiqie шашем seeds 19015 serqus) шеуашәлгу :uonoesqns а | snqons snuidodweong X :uoyes snuidodweong :snuabqns 476 Annals of the Missouri Botanical Garden classification as the best presently available hy- pothesis of phylogenetic relationships among the groups of extant taxa. MESOZOIC BIOGEOGRAPHY ORIGIN OF PINES The prevailing hypothesis until the mid-1970s on the origin of the genus relied on the contem- porary interpretation of Mesozoic fossil flora and the prevailing theories of the origin of cool tem- perate vegetation (Chaney, 1940; Mirov, 1967; Mirov & Hasbrouck, 1976). Fossil pines had been described from Triassic, Jurassic, and abundant Cretaceous locales, with pines especially abundant and diverse at high northern paleolatitudes. Mirov's widely cited interpretation dated the genus to the late Paleozoic or earliest Mesozoic, with its origin centered in a far-northern circumpolar continent known as Beringia. According to Mirov, the sub- sequent evolution of pines unfolded in a steady and progressive migration southward during the Me- sozoic and Tertiary, culminating in a final south- ward thrust toward the equator during the Pleis- tocene. This interpretation was cast into doubt by sys- tematic revisions of Mesozoic coniferous fossils. Alvin, Creber, Miller, Stockey, and others com- pared internal anatomy of fossil and extant pina- ceous cones and found four diagnostic traits that characterize Pinus (summarized in Miller, 1976). When previously described fossils were reanalyzed, many that were originally ascribed to Pinus were reclassified in the extinct pinaceous genera Pityo- strobus and Pseudoaraucaria. 'This applied to all the known pinaceous remains from the Triassic and Jurassic, and many from the Cretaceous. In particular, all of the high-latitude macrofossils, pri- marily in the collections of Heer, 1868-1883, originally treated as Pinus were reclassified, in some cases, even as angiosperms. Тће revisions, combined with new fossil discov- eries in the last two decades, result in a Cretaceous fossil record of about 25 species of ire from eight Northern Hemisphere regions (Table 2; Fig. : About half are known from petrified cones whose internal anatomy has been confirmed as Pinus. Others are needle, wood, or pollen fossils, generally accepted to be Pinus. Of special note is the dis- tribution of these fossil pines. Although a geograph- ic bias may be expected due to proximity of fossil locations to active paleobotanists, fossil pines occur at middle and a few high latitudes, widely spread east and west, with apparent centers in the north- eastern United States, Japan, and western Europe. The earliest known pine, P. belgica, from the early Cretaceous (about 130 Ma) was found in Belgium. Pollen of an early Cretaceous pine was also found in amber deposits from Alaska (Langenheim et al., 1960). All other pine fossils are known from middle to late Cretaceous deposits. The taxonomic diversity of fossil pines in the Cretaceous record is broad; two major subgenera and five subsections are represented. The origin of the genus is thought to be early-middle Mesozoic, although probably not the Paleozoic as suggested by Mirov (Miller, 1976; Eguiluz Piedra, 1985; Axelrod, 1986; Millar & Kinloch, 1991). A major change from Mirov's thinking concerns the location of the center of origin of the genus and the paths of subsequent radiation. At the beginning of the Mesozoic, there was one land mass, Pangaea (Smith 1981). By the early Jurassic, a northern supercontinent, Laurasia, separated and began to et al., drift from a southern continent, Gondwanaland. Although there may sea in northern latitudes than at present (Wolfe, ave been more land above 1985) little evidence exists for a circumpolar con- tinent, Beringia, that would have supported the origin of pines (Hickey et al., 1983; Eguiluz Piedra, 1985; Wolfe, 1985). Most importantly, no fossil evidence exists for a high-latitude Mesozoic center of origin for pines. Mesozoic pine fossils occur between 31°N and 50°N latitude, with only two records from higher latitudes (Table 2). Similarly, fossils of six species of Pseu- doaraucaria and over 20 species of Pityostrobus, genera considered most closely related to Pinus and most likely to have been the ancestral gene pool to Pinus, also occurred exclusively at middle latitudes, concentrated in eastern North America and western Europe (Miller, 1976, 1988). Hence, a circumpolar origin for Pinus is unsupported, and pine origins in middle latitudes are more likely. The regions of the northeastern United States and west- ern Europe, which would have been contiguous in the early and middle Mesozoic (Smith et al., 1981), are the current candidates for the center of origin of the genus (Miller, 1976; Eguiluz Piedra, 1985; Axelrod, 1986; Millar & Kinloch, natively, the diversity of Cretaceous pines and Ри- lter- yostrobus in Japan suggests that pines might have evolved in eastern Asia. DISTRIBUTION OF PINES AND CLIMATE OF THE LATE CRETACEOUS To understand the impact of the early Tertiary on pines, it is important to stress the Late Creta- ceous distribution of pines and the prevailing cli- Millar 477 Volume 80, Number 2 1993 Evolution of Pinus од ош plom ee 09104 210204 и "порама bu Аојтн "uius Шол] пәхе] * 1861 $5244 Амвлалш у одрџаше) © deur aseg ‘pom зподовјол“) 342] oui Jo Чеш uo зојешрлоозозјед ројешцез Aq роддеш “susodap споаовјало)у шолу S[ISSOJ autd jo поцпаше(ј ‘p 382514 | T t T T + + 409 Я + + + Es + + - - + + H + 40 5подовј810) 8427 * зподовјејо Aye Y b^ + d эт . + + + vil e a y e - + е + 4 09 x: У 1 n L L 1 1 4 1 081 021 09 0 09 021 081 Annals of the 478 issouri Botanical Garden M IDSOLIP uod (9861) JUTEN Y JTW 3187 Mo8€ Месо Лора, NoOF Kujoge]q *Aos19[ мом /saqpagsnp sisuapoom[jo 'а әр} udo (9861) град 29 &exoois эте A9ET №06 Act I Меру oprexxog *uedef эозоләрид | idad seis de bal MO d (9861) ер!Ҷѕ! 79 4949015 эте 09€ I No0S Sect I Мору оргеуҳоң *uedef Apo | „АНУ d (2861) *PUSIN Y BPP) me] 09€ I NoOS Aoc? 1 Net T орге оң ‘иеде snuiq "8405 nijofijjoqpyopnosd 'а (9861) &PIUSIN m Kexoois (2661) 980 "EI HASET Мобу Aot 1 No£ T oprexxog ‘иеде әр$оләрио д pyofiqosssopnasd “q (2861) *PUSIN Y *per] aw AoS£ I Ne8T Мост! Noc oprexxo ‘чеде упија "8408 pipusapryovd d (9861) BPIYSIN 79 4249015 AT JoS£ T Мову sect I Noc oprexxog *uedef (6061) &o1ge[ Y оон 2] №26 Not E МР2 Nol? puss] uejeiS ‘зод MƏN (2261) uosiqoy (+61) Auuag 3187] MoS€ №66 Moll Nol? s1195NYIBSSPIA IDSOLIPUO] ођојопђитђ `q (£881) xnalanbsa7] 3187 Mol9 Мере №66 No8£ sesuey IDSOLIPUOJ upaisuanb `q (9861) *PUSIN 19 Kexoois (2261) #1180 em] ASE I No6T Зоест! № oprexxo ‘иеде SƏSUILVUD) voyofiaqoy q (9861) BPIYSIN 25 Aexpois a] 3561 Моб? Зоест! No£ t oprexoH *uedef sasisaiás vyíydviopnəsd ‘q (9861) BPIYSIN 1) 4239015 ҸЕ] ASET Моб? Hot I Мер oprexxoH ‘uedef зада) ÁS voyofiq d (9861) PPUSIN Y 4249018 3187] сет No8? "octlNect оргехлон *uedef sa41saa]Ag Dj] dp4121 ^d (9861) PPIUSIN 2 4320S aT обе [ Ne8T oct D Noch oprexxoH ‘чедер | (6061) 421491 Y JPH "T №26 Nol € МР2 Nol t pues] ueieiS "од MƏN appAuydoia'] | (LL61) uosiqoy 3187] MoS£ МЄЄ Moll Nel? S1195NYIBSSBIA /sa41saa]&g ораш qd (;%61) Auueg 3187] AOT. № MoS L Мебе олемвјо(] sa41saa]Aq ‘ds smug (ре61) &oueq) ned! MoSS Мору MoS6 Мору €josauurq s241s2a]Ag nsu] d (2661) 921etd aeq Арея MoSS МО? MoS6 Мору 9josauurpg S341859 ÁÇ чә[о snug SƏD Snp (+861) erg әдет Арея MoS Nol€ Mo88 Мере 09 sshusig 10411851 /грволориоа sapioso1apuod `q (0961) чију Apre 9187 3681 NoSt Ach Ne8T це snuig `9дпз тутшогуцот `q (1961) Te 19 10019 Apre 9187 Mo6€ Nol E MoS L Мобе STEMA] ѕпиәд uapod стила (£961) 19uua1g Кре әзе Mol? NZE Mo6L Мебе рие тер snuas uaqod ти] (0961) T? 19 штоцизвив] Арея №001 №08 №091 №02 JeAng лим "exsepy snuas uapod snu1g (0961) чү 019! Моб No6€ Чоў №06 чәрүеә A, ‘шп!8әң sa4]s2a]Ag 1913199 "d 2991919 (ер) озјеД шәт) иоце207] Áputgy uoneognuuop ony apnitZuo[ /зрпуце] (1861) "те 19 чиш uro1j гојешрлоозовје ] (#ипоА oi ро) a3e jo лорло ajeurxoadde ш paisi sysodap спозовјало) шолу зоша [resoj jo saurugje pue поцпаыцз] `Z 318V[ Volume 80, Number 2 1993 Millar 479 Evolution of Pinus Continued. TABLE 2. Age (Ma) Latitude / longitude Reference Paleo Location Current Affinity Identification Jeffrey (1908) Late 34°N 37°W 31°N 38°W 49°N 135°E 41°N 74?W 39°N 75°W New York, Staten Island Delaware subg. Strobus Strobi Pinus sp. P. magothensis ) Stopes & Kershaw (1910) Japan, Hokkaido subg. Strobus oensis P. yez Pinus pollen Frederiksen et al. (1988) Late 75°N 88°W TON 151°W Alaska, Colville River genus mate at the end of the Mesozoic. By the Late Cretaceous, pines had reached eastern and western edges of Laurasia and occurred at middle latitudes in many locations between these extremes (Table 2; Fig. 4). The widespread distribution of pines in Laurasia by this time indicates that wherever within middle latitudes they originated, their main route of migration was east and west, and not predom- inately southward, as Mirov suggested. Migration from eastern North America to western Europe was not impeded until late in the Mesozic, by which time Laurasia had begun to split into North Amer- ica and Europe. Laurasia severed first in the south and last in the north. High-latitude connections in the North Atlantic became increasingly reduced toward the end of the Mesozoic, and low seas may have covered the land (Ziegler et al., 1983; Tiff- ney, 1985a; Parrish, 1987). This region would have provided only minor corridors for pine mi- gration. Evidence also exists for land connections at high latitudes in the Bering Sea region between Siberia and Alaska being used as corridors for temperate-adapted flora. Within the new continents, continuing east-west migration in the late Cretaceous must have been hindered by seaways that extended the full north— south length of the continents (Kurten, 1966; Tiff- ney, 1985a). These seaways divided the continents into separate phytogeographic provinces, creating greater floristic affinities between eastern North America and western Europe, and western North America and eastern Asia, than between the east- west parts of each continent (Wolfe, 1975; Tiffney, 1985a). The Late Cretaceous was a time of climatic quiescence and equability (Parrish, 1987; Up- church & Wolfe, 1987; McGowran, 1990). Sea levels were high, and tectonic activity low, creating stable global climates. Although the breakup of Pangaea had commenced, paleocontinents were still relatively undispersed, resulting in average tem- peratures in the middle and high latitudes about °-20°C warmer than the present (Savin, 1977; Shackleton & Boersma, 1981; Parrish, 1987; Up- church & Wolfe, 1987). Evidence on rainfall in the Cretaceous inferred from foliar physiognomy of angiosperms indicates that rainfall patterns were zonal. The Northern Hemisphere had a humid re- gion around the paleoequator, a dry zone at low- middle latitudes, and a zone of higher rainfall above 5°N (Parrish, 1987). In general, however, lati- tudinal gradients were shallower, and changes in temperature with latitude were about 12-13 of pres- ent gradients (Parrish, 1987; Upchurch & Wolfe, 1987). Temperature and rainfall appear to have 480 Annals of the Missouri Botanical Garden been stable annually, with little seasonality at lat- itudes below 45°N (Upchurch & Wolfe, 1987). At higher latitudes, day lengths and precipitation ap- parently varied seasonally. The major cordillera of the Northern Hemisphere were not developed, or existed only at low elevations. Volcanic activity was minor, so there were few orographic effects on climate and little regional diversity in climate. EARLY TERTIARY BIOGEOGRAPHY CHANGING CLIMATES Major changes in climate and vegetation char- 1990). Al- though these events have long been discussed by acterized the early Tertiary (Wolfe, paleobotanists interested in angiosperm evolution (Wolfe, 1975; Tiffney, 1985a, b; Friis et al., 1987), their impact on conifer evolution has not been analyzed. In general, average temperatures rose and rainfall increased in the early Tertiary. The trends toward increasing temperature and humidity started in the early Paleocene and continued into the Eocene, reaching maxima in the early Eocene, about 52 Ma. By this time, average temperatures had increased 5?-7?C above the Late Cretaceous (Savin, 1977), and tropical/subtropical conditions apparently extended at many middle and high lat- itudes to 70°-80°N (Wolfe, 1985; 1990). High temperatures and humidity, however, did not persist stably throughout the Eocene. Al- though the late Paleocene/early Eocene (54-52 Ma) was the warmest and wettest period, there McGowran, were at least two other warm periods, from about 46 to 42 Ma, and about 36 to 34 Ma, separated by cooler intervals that were approximately equal in duration to the warm periods (Fig. 6; Wolfe, 1978, 1985; McGowran, 1990). Average annual temperatures may have fluctuated as much as 7?- 10°C between warm and cool periods of the Eocene (Wolfe, 1978). Several causes for these climatic developments have been suggested, including ma- jor tectonic events, changes in sea level, and sub- marine volcanism resulting in accumulation of at- mospheric carbon dioxide and greenhouse heating (Wolfe, 1978; Parrish, 1987; McGowran, 1990; Kerr, 1991). Alternatively, large amounts of car- bon dioxide may have been produced as a result of ocean-atmosphere interactions following a major extraterrestrial impact at the Cretaceous/ Tertiary boundary (O’Keefe & Ahrens, 1989), which led to greenhouse warming (Wolfe, 1990). Although latitudinal gradation was not great, the pattern in the Eocene differed from both the Cre- taceous and the later Tertiary and Quaternary: in general, many low-latitude locations were relatively warm and seasonally dry, middle latitudes were warm and wet, and high latitudes were cool and dry (Parrish, 1987; Wolfe, 1978). Truly arid zones apparently did not exist; there is no evidence for Paleogene arid deserts or tundra (Axelrod, 1979). Although these general trends in temperature and humidity existed throughout a broad latitudinal zone worldwide, there was geographic heteroge- neity in the intensity of conditions. The warm hu- mid zone was widest in North America and western Europe, and narrower in central and eastern Asia (Chaney, 1940; Parrish, 1987; Hsu, 1983). In general, continental elevations were low throughout the Paleogene, and upland areas apparently existed primarily in one middle-latitude region of western North America and in Antarctica (Axelrod, 1966; Wolfe, 1985, 1987; Wolfe & Wehr, 1987; Wing, 1987). In the upland area of western North Amer- ica, the climate was anomalously temperate com- pared to other middle-latitude areas. Volcanism and mountain-building in this area during the Eocene also created heterogeneity in local climates and habitats. THE ANGIOSPERM BOREOTROPICAL FLORA The changes in climate during the early Tertiary drastically affected global floristics (Friis et al., 1987; Wolfe, 1975, 1978, 1985). Diverse tropical and subtropical angiosperm floras appeared with increasing geographic representation during the Paleocene and the warm intervals of the Eocene throughout broad zones at middle latitudes in both Northern and Southern Hemispheres. Originally identified from the London Clay formations of En- gland (Reid & Chandler, 1933), similar angiosperm floras have been described from many deposits elsewhere, including western and eastern North America (from the Pacific Coast to Nebraska and Texas; Vermont, Alabama), western and eastern Europe (including England, France, Belgium, Ger- many, Bulgaria, Ukraine, and Russia), northern Egypt, China, and Japan (Mai, 1970; Graham, 1972; Wolfe, 1975, 1985; Tiffney, 1985a, b). Subtropical assemblages occurred north as far as 0° in Alaska, and at other high-latitude locations in Canada, Greenland, and Siberia (Wolfe, 1977, 1985). In North America, the average zone ex- tended from 30°N to 50°N (Wolfe, 1985). The origin and major radiation of many angiosperm taxa is documented in these diverse floras. Plant communities in these Paleogene floras were adapted to warm, humid, and equable conditions. These taxa were similar to those found in modern rainforest vegetation of Malaysia and other extant tropical rainforest regions and show comparable adaptations. Common genera include Engelhard- Volume 80, Number 2 1993 Millar Evolution of Pinus tia, Ficus, Pterocarya, Nypa, Platycarya, Weth- erellia, Alangium, and Nyssa. А few gymno- sperms such as Glyptostrobus, Taxodium, and occasionally Sequoia occurred in these floras, but they, too, were apparently adapted to warm, humid conditions. Recognizing their northern locations and their adaptations to warm, humid conditions, Wolfe (1975, 1977) referred to these widespread angio- sperm assemblages as the “boreotropical flora." While this assemblage in no way suggests an or- ganic, indivisible unit, it does imply that similarly adapted individual taxa migrated east and west rapidly and unimpeded, at middle latitudes throughout North America and Eurasia in the early Tertiary. The boreotropical flora reached its great- est development in the warm periods of the Eocene. During cool intervals, its latitudinal extent shrank. EOCENE PINES The early Tertiary radiation of many angio- sperm lineages and the migration of angiosperm boreotropical taxa east and west at middle latitudes has many parallels with the late Mesozoic radiation and migration patterns of pines. Boreotropical flo- ras occur in the same locations worldwide during the early Tertiary as pines did in the late Mesozoic. With very few exceptions, pines are not found in boreotropical fossil floras. This prompts the ques- tion, where did the pines go* The Tertiary record of pines begins in the Eo- cene; no pines are known from the Paleocene. Pines of the earliest Eocene occur primarily in high (65% 80°N) and low (2°N) latitude deposits in North America and Eurasia (Table 3; Fig. 5). High-lati- tude locations include central Alaska (Wahrhaftig et al., 1969; Dickinson et al., 1987; Fredericksen et al., 1988), Ellsmere Island, Greenland, Iceland, and Spitsbergen (Manum, 1962; Schweitzer, 1974). Low-latitude pines occurred in Borneo (Muller, 1966) and from one low-middle latitude location of the late early Eocene, near San Diego, California (Axelrod, 1986). Pines in these deposits were as- sociated with other pinaceous conifers and with cool temperate angiosperms such as Betula, Alnus, and Ulmus. Ages of these fossils correspond to the first warm humid period of the Eocene (Fig. 6). Fossil pines from the middle Eocene continue to be represented at high and low latitudes, but appeared for the first time in the Tertiary in middle- high latitude locations in North America and Eur- asia. Most of the known fossils from the cool period of the earlier middle Eocene are in western North America, from British Columbia (Miller, 1973; Stockey, 1983, 1984), Washington (Wolfe & Wehr, 1987), Idaho (Axelrod, 1986), and Colo- rado (Wodehouse, 1933; МасСтне, 1953) (Table 3; Fig. 5). Western North America in general contains some of the richest plant-bearing deposits from the Eocene in the world, and dozens of fossil floras have been described. Only a few of these contain pines and other temperate taxa, and these are all concentrated in northern Idaho, central W yoming, north to central Idaho, and British Co- lumbia. This is the region that has been identified as an upland area with average elevations of 1,200– 1,500 m (Axelrod, 1965; Axelrod & Raven, 1985; Wolfe, 1987). Pine deposits in some of these areas are associated with active volcanos and high ele- vations (e.g., Bull Run, Thunder Mountain, Axel- rod, 1965, 1986; Creede, 2,500 m, Wolfe & Schorn, 1989). The warm humid period of the middle Eocene is represented by few pine fossils (Table 3). Pine was present at high latitudes in the Mackenzie Delta of Alaska (80°N, Norris, 1982), at low-middle lat- itudes in Borneo (2°N, Muller, 1966) and southern Alabama (36°N, Gray, 1960), and also at a few middle latitudes in Nevada (50?N, Axelrod, 1966, 1968), within the upland plateau of western North America. Pine fossils from the subsequent cool period of the later Eocene occurred at middle latitudes in Washington (Miller, 1974), Nevada (Axelrod, 1966, 1968), Colorado (Leopold & MacGinite, 1972; Axelrod, 1986), New Mexico (Leopold & MacGinite, 1972), and Japan (Huzioka & Taka- hashi, 1970; Tanai, 1970, interpreted by Wolfe, 1985 to be early-late Eocene), Fushun, China (Hsu, 1983), as well as a continuing presence in Borneo (Muller, 1966) and at high latitudes (Norris, 1982) Table 3 The latest Eocene marks the final widespread period of tropical conditions at middle latitudes in the Paleogene (Fig. 6). Pine fossils from the late Eocene occur primarily at high latitudes in western Siberia (Dorofeev, 1963), Alaska (Norris, 1982), British. Columbia (Hopkins et al., 1972), at low latitudes in Borneo (Muller, 1966), and along the Tethys seaway in southeastern Europe (Chiguriae- va, 1952). Pines from the southeast coast of China, in the provinces of Jiang-su, Zhejian, and Fujian (Hsu, 1983), may also be from this period (Table 3). THE TERMINAL EOCENE EVENT AND PINE EXPANSIONS The end of the Eocene was marked by the most rofound climatic event of the Tertiary (Burchardt, 1978; Wolfe, 1978; Parrish, 1987; McGowran, c Ф о pus с (5 © © c 25 c-— Lo ос 5 20 Ф co с = < > сч co я (086) родаху (2¿£61) чмо (2861) PA Y злом (#861) 4942015 (#861) 4932015 (2261) IEW (7861) Áox»ois (2761) JN (2861) 4242015 (£61) moig (££61) әѕпоцәро дү (2861) 17-9? Ly-9y [А 21%-9# LvY-9tv Lb-OF Mov8 NoSS №28 №09 №06 №19 №06 №19 №06 №19 №06 №19 №26 №99 №08 №065 Мор 11 Мосту Лов 11 Мобу A06 1 NoOS Луо0е | №06 Mo07 1 NoOS Mo07 1 NoOS MolG [ NoSS №601 Могу urejuno[q 1epunuT "оцерј ayqnday *uojgurqse A појазима "eiquin[or) чепи uoio2uugd "eiqum[o?) qsntg uojaoutiq equino) qenug uojia2utugd "eiquinqor) чеши SLY Wg *eiquin[o?) Чел IAY U39319 "орелојоју unysn y apnununofpg snuaS IDSOLIPUOJ sa1jsaa]Ag snqo41S "qns sa11saa]Xq $9]D1ISNP IDSOLIPUOJ SILISIAÁAS Sapio4q шә”) sap101n0f]0q d uajod smund NUOSLIPUD `q sisuauojaquud 'а s1SUIJULDYA NUIS а прјоил) га sisuapoom pap а 1199 а nsH (9/61) nr y Sung әле рие PIJA 35811 NoSt Дера 1 Noct ‘guyo UJ9ISPIYLION snuas uopod snuid (6861) чолед X poupxy 8t-LT MoZ6 Nol? MoLIT No££ eruop[e?) илацупос IVSOLIPUO Y џајјод ути] (S861) чәлеу X родаху 8t MoZ6 Nol? Mol TU No££ ?rulojier) илоцупос saplosquia’) ualjod snuiq (9861) родеху 8b-L? A66 Noct Mol 11 No££ PruJOJI[eo) илоцулос упфол; “3qns sisua4Duap ^d (9961) лепи Ае 318] FOOT Noo ОТТ Моб ози1од илојзомуи мом snuas пзјјод snuld (£161) nuige W pjodoeT Apre Mo02 Мед Mof01 Nol? иип *ejoxe(T YHON snuas uopod snuig (#261) 1огломцоб (2961) unue Арея 90 №02 deb l Моб; џовлоде и 46 snuas џајјод snuid (2961) unue Apre] Яе8 No0S №15 NoS9 pue[o] snuas uəjjod snuig (2961) umue]y Арея Мед №09 Mec S №0/ pue[uaa^) snuas uod smug (2961) unue Ае Mo8T NoSL Мор8 №22 pue[s[ злеш у "греше snuad uod smuid (sute18 3]e»oestq) (2861) “12 19 uosunpiq Хине ua №08 №91 №69 ÁajeA yq 'exsery ѕпиәЯ ua[od ; snuid (8861) 1 19 uesxuopo1j] Арея 21 №08 MolST №02 Əta *exsery snuas uopjod snug (6961) ‘Te 19 Зпјеціце v, pS 1 МАЛ AMe8t D Nob9 eueuaN *exse[y snuas uayjod ути] 92u219]JaM (еу) оә[ед jua.Lm-) UuOon?2o7 ÁAputgy uoneognuop] у apniiZuo| /opnine'T (1861) че 19 чиш Wo зојешрлоооогјед -(8ипоќ ој pjo) әде jo 1opio ојештхоладе ur pois зизодор ausoog шолу ѕәша [Isso], Jo sourugje pue поцпа еј '£ 318v] Millar 483 Volume 80, Number 2 1993 Evolution of Pinus (2861) "SH ae] Лов 11 NotE-87 Мо06 I Мобе— 566 Buy) u1ojseaqinog snuos uo[jod snuid (9861) родаху (2661) MNULDIPA Se-Tt Mo9L No8T ANoSO T Мобе JUBSSILIO| Y "орелојој) 190115 119]994M `d (9861) | роцәхү “(£561) ULN S€£-T€ MOL No8T MoSOT No6€ juessulo[4 *opeJo[o7) 199114 njunssi410jj са (9861) роцәхү (2061) зишоову St-T€ №92 No8T MoSOT No6€ juessuo|j *opeJo[o7) 190416 nsizanis са (2561) PA9eLInalu) 31e] Мобе Мобе Жобр Nol? ио JeATy PUIN snuad uojod пи IDSOLIPUOJ (£961) ^99]o10(q 318] 19 Nec S 62 Медо JoAly yosÁ11] воде sa4]saa]Ag pnsouids "I (£961) ^99]o10(T 318] TEL NoSS Mec 8 №09 ДОЛ] WAT uA snuos ‘ds пи] (£961) 499jo10(q 91871 952, NeT S 468 №26 IAY 40 ‘PUIS snuad “ds пи (£961) ^99]010(q 318] Чеде Nec S 469 No8S ләл эрлер, LUIS snuas “ds стила JoATy (2261) T? 19 surydoy St Mo06 №99 MoGGI NoSS drus1eg “eiqunpo) ysuug snuas ua[[od snuid (261) INUIOIBIA Y pjodoo L£-9€ №92 No6T МСОТ Мобе ə yey) 'орелојојђ) snuad uəjjod snuid (2161) eue] 3187] Ае] осе 1 NoOS MocP FT Мер onysny “uedef snuas ‘ds snuig (0261) üjseqexe] Y xong 318] Ајде JOET МОЁ ЧОЕТ Море aq) “uedef UJ9ISIMYINOS snuagd ‘ds snurd (Z261) auge y pjodoeT 8t Mo08 Nott №201 NoS€ оцечзәя *OMXIJA мом snuas usjod snuig (8961) родаху 8t Mol6 №06 Mol [1] Nol E ung [ng 'ерелом 20140]u0/) Sisuapaoa]n `d 301101407) (F261) WW ajeT Агея Mot6 No09 MoE | Nol? Seq әпит *uojurqse ҳу /sa41s2a]&g тајјот d (9961) родаху YY-OT M068 NolS MoS 11 Мост urseg saddory "грелом IVUDIMOS] DY 1158012 "d (8961) Родэху ct M006 NoOS MITT Nol t ома 'PpeA9N 100418 119]99ут "а (0961) £219 9[PP'IN Mo69 No9€ Mo88 Nol € spnig 9euioqrep) *eureqepy ѕпиәд uo[pd спијај (2861) SHION 91e T PUN №08 No08 МЕТ №69 цә AZUL *ереивэ) snuaa uojod пи] аопелојон (EA) o31td quadiny поцв2071 Кушу поцеоцциорј зву apnitZuo| /apniue'] 'penuguo?) СЄ IYL Annals of the 484 Missouri Botanical Garden а ош pjiom jojuauiuo202]pd o102012uniq *uopug ә Хорти *чишс шошу uoxe1 ‘1861 ssaig Аизләлгигү гдрмашео) © Чеш aseg pom euaooo[eq әці jo deur uo зајешрлоозогјед рәјешцѕә Aq роддеш ‘sysodap опазод uroy srssoj әша jo uonnqujs([ сс INNI T T Т Т с> F & + + + + + + + + + ^ MEL F * P t t T + 40 spoued Áp ‘joop e spoued ршпу “ште v = + x -— d Е „У 22014 ER DEEP У 08! Oct 09 Volume 80, Number 2 1993 Millar 485 Evolution of Pinus Percent of entire-margined species Inferred mean annual temperature (^C) x j 0 0 X 10° yrs GURE 6. of entire leaves), indicatin ng t e warm and cool episodes i Eocene (terminal Eocene event). Sigma Xi, The Scientific Research Society 1990). The fluctuations from warm to cool periods during the Eocene were minor compared to the drop in regional temperatures at the end of the Eocene. Average annual temperatures dropped ap- proximately 10°-13°C, in some areas over only one million years (Fig. 6). The decline in average temperatures was accompanied by a large decrease in rainfall and increase in seasonality. Whereas in the Eocene the mean annual range of temperatures (seasonality) is estimated as опу 3°-5°C, in the early Oligocene it is estimated as 25°C—about twice the range of the present (Wolfe, 1971). In many parts of the world, complex continental cli- mate patterns developed, possibly for the first time, and continental ice sheets first occurred in the Oligocene (McGowran, 1990). Extensive volca- nism and mountain-building in the Rocky Moun- tains, Himalayas, and Mexican ranges created local climatic diversity. Major tectonic events have been associated with the climatic cooling at the end of the Eocene with two schools of thought on their effect. One line of evidence suggests that changing positions of the Earth's continents and oceans had direct effects on the Earth's climate, leading to Estimated temperatures from four _ in North America during the EE (based on jio. the Eocene and the drop in tempera ature at the end of From Wolfe (1978). Вере by permission of American Scientist, ond of major cooling (Barron, 1985; McGowran, 1990). Another possibility is that the formation of giant uplifted plateaus in southern Asia and the American West led to accelerated chemical weathering, decline in atmospheric carbon dioxide, and a "greenhouse cooling" effect (Ruddiman & Kutz- back, 1989, 1). 'This terminal Eocene event (Wolfe, 1978) was marked by widespread extirpation of boreotropical angiosperm taxa from many middle latitudes world- wide, leaving only remnants in a few areas where the climate presumably remained mild. The ex- tinctions of boreotropical floras were mirrored by great expansion of pines and other cool-adapted taxa into many middle-latitude locations. Many of the same fossil-bearing sites that contained boreo- tropical taxa in their late Eocene horizons were dominated by cool-temperate conifer taxa in early Oligocene horizons. In some localities, this change occurred in only one million years (Axelrod, 1965). The existing record suggests that pines made their first Tertiary appearance at many middle- latitude locations during the Oligocene (Table 4; Fig. 7), recolonizing areas where they had occurred Annals of the 486 Missouri Botanical Garden O deur əseg "рџом әпәзогр\ Ајлез 941 јо deur uo вајешрлоооовјеа рәівшцѕә Aq раддеш 'susodap 93u320t]q рәэ1әэ[эѕ pue әиә2о8цО) шо} 511550] әша jo uonunqujsuT ‘sdpu рот јрјизипиозотра. >1020194DY g *uopug x» Хорти ‘rug шолу позе) '1861 55914 Áis1oAmu(] oSpuqure:) T T ] 1 T 1 T T T T Џ L di + Е > + T + + аџезоуј e гџезобуо v F + 4 + y” L | 1 4 | 1 081 021 09 0 09 Oct 08! `2 IANI Millar 487 Volume 80, Number 2 inus f P Evolution o 1993 2D]101u07) (S£61) шацед 49187 ToS Мосе TOS NoOF eiseoneosuel] ulojse] /SILISIAÁS anidoqipu qd (р261) 11022227) 91271 ЧО NoS€ "lect Noct PISPINBISUBI] илајзо M sa11saaj&g psn&pndoad `q nSUP£) плојзому ом (8661) Bung "I 3.06 Nott 496 Noct “BUIYD) UI9JS9MQON snuas uayjod ути] шерту (2861) SH 29187 Зосб Мору 496 Месту ‘euyy UI9JS9MQHON snuas џојјод snuiq (ures 1) (2961) Те 19 шпоцизвиг] 9127] Mo€8 Not I Луосб №91 sedey ‘onxa snuaa чә[ой snuig (6861) (sisuopaoa]p ә) HOM члоцос (9861) родоху LG Mo£6 Мобе MoSOT Меве арзело 'орвлојо)) 99110740.) ‘ds smuig (6861) OM 29 члоцос (9861) родеху LG Mo£6 Мобе MoSOT No8€ әрәәлгу орелојо) АЕ, sisuoupnfups `q (6861) 99A 79 u1ogog LZ №6 No6€ MoSOT No8£ apeal’y ‘operojn IDSOLIPUO Y apunagon `q (6861) злом Y 202 LG Mo£6 Мобе MoSOT No8€ apssr) орело[огу apununofpg 1155049 а (2261) әишсуәер] y pjodoeT "El №66 NoOT MoSOT No6€ Sager) 'орело[огу snuas ua[od стил (2261) посту № рјодоо 1] (1961) 193998 оде] MoZOT Nol? Мост! Меср швея резцлолвод 'euejuo]q snuad uogod зпила (9861) родеху 6c №01 Nod? Aet 11 Месту 1931) [80 "oqep] IVUDILNO/]1g sopio4nofjoq `d (9861) родеху бс AVo£O T Nol? Mot 11 NoST 19919 [POD оцерј 199-115 snaumuup qq (6961) 21W 49187 №001 NoOS Mo£ TT NoZ t полу “BUBJUOJA] IDSOLIPUO] sisuauoan "d (086 D зә 79. poomepuy) 91971 MoOTT Noc S Mot G [D No8T ordurAq() *uoigutqse A, дрзолориоа d а а d (2261 '2261) HOM 9187 Ло8с | №89 MoSt 1 №09 aah J? HS snuəs uapod snurg (9961) MW әв Ајде "190 LT Мос JOTT Noc 091106] илојзому OK] snua3 uagod snug pue[s e1ue[eosq (1861) Те 19 squrg әле Аре MoSTT №965 №21 №06 "eiqum[or) qsyug мешеу а Уа, | (6261) X991) USUE SNY SMSYNE 2) əsnoy (1261) 194 Арея Лу01 1 №09 Mo9G I №96 ‘erqumnjog qsuug snuas чә[о smuigq (9261) Азреа 9 родеху GE Mo86 Море Mo80T No£ € o1loqs[rH *OMXAA мом apupunofpg 1155042 "d (2261) HOM (1961) 194998 9E-ZE А001 No9b = Moa II №6 шея Aqny *eueiuo] snuog uopod snuid (2161) зишоовуј y рјодоо 1 Адел №98 No8b МРОТ Мор Ájunor) иип *ejoxe(] Ччмом snuog uagod стила (1861) злом Ie ANO TI No6? Mo£Z IL NeS suoÁ'- 'uodaJ() snuas uagod стила (2361) чогеу (1861) HOM 16 №801 Nol? Mo811 Меру 199.1) әриң ‘uo3ə10 snuas чә[ой snuid a2ua19Jay (EN) оәјед jualin’) поцеоо] Anuy поцеоуупорј "y apnitZuo[/apnineT (1861) Те 19 Чишс шол зојетрлоозозјва '(Suno& ој рјо) әде jo лорло ојештхоладе ut роја зизодор suavo шолј ѕәша [ој Jo saniuyye pue uognquis(| "p злауј, 488 Annals of the Missouri Botanical Garden in the Mesozoic. Pines are known from Oligocene deposits that range in North America from the Gulf of Alaska (Wolfe, 1972), British Columbia (Banks et al., 1981), Washington (Underwood & Miller, 1980), Montana (Miller, 1969; Leopold & MacGinite, 1972), Idaho (Axelrod, 1986), Oregon (Wolfe, 1981), Colorado (Leopold & MacGinite, 1972; Schorn & Wolfe, imen from Chiapas, Mexico (Langenheim et al., 1967) (Table 4). In Asia, Oligocene deposits are in eastern and western Transcaucasia (Palibin, 1935; Czeczott, 1954), and in northwest China, in the Quidam Basin, and in the Jinguan Basin of western Gansu (Sung, 1958; Hsu, 1983). In Japan, pine was present as a minor component of a few floras (Tanai, 1972). P Oligocene strata of Borneo (Muller, 1989), to a single spec- ine was abundant in the though some of the pine-bearing deposits were as- sociated with areas of volcanism and mountain- building, others were in lowland and coastal areas where boreotropical flora had flourished in previous epochs. The climatic deterioration of the early Oligocene was followed by an ameliorating and warming trend during the late Oligocene and into the Miocene. Rainfall, however, stayed moderately low and cli- matic conditions remained temperate at middle lat- itudes. The record indicates that pines rose in abun- dance throughout middle latitudes in North America (Axelrod, 1986), Europe (van der Burgh, 1973; Klaus, 1989), and Asia (Hsu, 1983) in the Mio- cene; the direct ancestors of many modern pine species can be traced to Miocene pines. The warm conditions of the Miocene supported expanded pine forests at high latitudes, such as Banks Islan Hills et al., 1974), MacKenzie Delta (72°- , 1984), and Wrangell Mountains, Alaska (68°N, Wolfe, 1969) (Table 5). Pine fossils in low-latitude deposits, such as Veracruz, Mexico (17°N, Graham, 1976) indicate the continued pres- ence of pines in mountainous low latitudes, despite the fact that lowlands in these latitudes were dom- inated by boreotropical flora. IMPACT OF EARLY TERTIARY CLIMATE ON PINE EVOLUTION EOCENE PINE REFUGIA The preceding overview of biogeography pro- vides evidence for the hypothesis that pine distri- butions shifted latitudinally and that pines expand- ed and contracted in elevational extent several times from the late Mesozoic to the middle Tertiary. In general, the hypothesis states that pine popu- lations (1) occurred throughout middle latitudes in the Mesozoic, (2) were fragmented and displaced to high and low latitudes and to middle-latitude uplands in the Eocene, and then (3) reappeared widely throughout middle latitudes in the Oligocene and Miocene, where they remained for the rest of the Tertiary. The Mesozoic origin and spread of pines east and west throughout Laurasia apparently occurred under a warm temperate climate that was equable and had little latitudinal gradation. Early migration and radiation of pines was favored in the Mesozoic not only by climatic conditions, but by the absence at high-middle latitudes of competing angiosperm taxa. Although Wo Mae | originated about the same time e pines in the Early Cretaceous (Taylor & Hickey, 1990), their rise to dominance in middle latitudes did not occur until the late Cretaceous (Wolfe & Upchurch, 1986). The shift in global climates toward warm humid conditions of middle latitudes in the early Tertiary appears to have favored the migration and radiation of angiosperms at the expense of pines. Although pines have broad tolerance of climatic and edaphic conditions, they do not survive or grow well in hot and humid climates (Mirov, 1967). More impor- tantly, in these conditions they are poor competi- tors with angiosperms in seedling establishment, height growth, and reproduction. Conifers domi- nate in areas where angiosperm competition is re- duced, for example, by fire, cold, or nutrient short- ages (Bond, 1989). Such a situation apparently initiated widespread extirpations of pines over the lowlands at middle latitu itat acted as refugia for pines during the warm, humid periods of the Paleogene. n general, the fossil evidence indicates three important refugial zones. The circumpolar high- latitude zone originally thought by Mirov to be the cradle of Pinus emerges not as a primary Mesozoic es. The remaining areas of suitable pine hab- center of origin but as an important Eocene re- fugium. Although many Mesozoic fossils from the polar region that had been identified as pine were later discredited, credible new records, mostly pol- len, confirm araras pine at many high-latitude locations in America and Eurasia during the warm periods pe the Eocene (Table 3). Pines may have been even more widespread during the early Tertiary in polar regions than indicated by the geography of present land masses. Much of the Arctic Sea is shallow, and tectonic evidence sug- gests that there may have been more land above sea level in polar regions in the early Tertiary than at present (Smith et al., 1981; Wolfe, 1985). Land Volume 80, Number 2 1993 Millar Evolution of Pinus 489 connections between North America and Eurasia in the North Atlantic (Tiffney, 1985a, b) would have provided high-latitude corridors for east-west migration of pines during the Eocene (Tiffney, l a, The major limiting condition for plant growth poleward now is light. The suggestion that the Earth may have had a lower angle of inclination in the early Tertiary (Wolfe, 1978), which would allow more light at high latitudes, has been discredited (Donn, 1982 983; Barron, 1984) Alternatively, geophysical and climatic conditions ; McKenna, of the early Tertiary may have increased avail- ability of carbon dioxide and permitted plants to photosynthesize under light regimes that were pro- hibitive under conditions of lower carbon dioxide concentrations that apparently followed the end of the Eocene (Berner et al., 1983; Creber & Chalo- ner, 1984; Ruddiman & Kutzbach, 1989, 1991; Kerr, 1991). Low latitudes in North America and Eurasia also appear to have been refugia for pines during the warm humid periods of the Eocene. There is ample evidence that conditions in general were warmer and drier at low latitudes than at middle latitudes in the Eocene. These conditions have been docu- mented for the southeastern United States, Central America, Taiwan (Wolfe, 1975), southern big and southeastern Asia (Guo, 1980; Hsu, aL and Borneo (Muller, 1966). Dunne the аа the southern boundary of Eurasia was the Tethys seaway. Since the Indian subcontinent had not collided with Asia, the area of the present Hima- layas marked the Tethys coastline in Asia. Unlike circumpolar regions, the low-latitude dry areas were disjunct and disconnected, even between the south- eastern United States and Central America, and refugial populations were unlikely to have been connected by broad migration routes (Wolfe, 1975). There is less abundant fossil evidence for pines at low latitudes than at high latitudes during the Eocene. There are few early Tertiary plant-bearing deposits in the areas of special interest, such as Mexico, Central America, and the Mediterranean, primarily due to the lack of depositional sediments (Martin & Harrell, 1957; Eguiluz Piedra, 1985). Nevertheless, Eocene pine fossils have been found areas of southeastern Europe that were at low latitudes during the Eocene (Table the single specimen of pine at Chiapas (14°N) from the Oligocene (one of the earliest Tertiary plant records in Mexico), and the abundance of pines in Distribution and affinities of select high- and low-latitude fossil pines from Miocene deposits listed in approximate order of age (old to young). Paleocoordinates 'TABLE 5. from Smith et al. (1981). Reference Hills et al. (19 Age (Ma) Paleo дом 92°W Latitude/longitude Current 2N 1229W (= Location Affinity Identification 74) 7: 7 Early Canada, Banks Island Strobi Pinus sp. Hills et al. (1974) Early 74°N 122°W Canada genus Pinus sp. 72-85°N 94-122?W -80*N 130-150°W 2°N 110°E 70?N 160°W 62°N 140°W 7 MacKenzie Delta/Alaska Northwestern Borneo genus en 78N 142°W 68°N 122°W ] 7?N 85°W not available genus Alaska, Wrangell Mountains genus en Graham (1976) Middle ] 7?N 94°W Pinus pol Pinus pol Mexico, Veracruz Klaus (1989 Klaus (1989 Klaus (1989) not available Pinus sp. not available Yugoslavia Sylvestres not available Romania Sylvestres Pinus sp. 490 Annals of the Missouri Botanical Garden Veracruz (17?N) from the Miocene suggest that pines may have been present in Mexico and Central America earlier (Table 4). Similarly, pines were present in the Oligocene and abundant in Miocene deposits from southern Europe in the region of the Tethys (Palibin, 1935; Czeczott, 1954; Klipper, 1968; van der Burgh, 1973; Klaus, 1989). Pines were not absent from middle latitudes dur- ing the Eocene. Their presence in select fossil de- posits from interior western North America testifies to the presence of mid-latitude refugia. Pines in these regions more commonly date to the cool periods of the Eocene. During these times pine populations apparently expanded, whereas during the warm humid periods they contracted into nar- row refugial areas where conditions were favorable. During warm humid periods, pines in interior west- ern North America may have been fragmented into small local refugia and, therefore, poorly repre- sented in fossil deposits. The interior of western North America was anomalous for the Eocene world in having upland areas and centers of volcanism. Several of the pine- bearing deposits occurred at high elevations in Eo- cene calderas (Thunder Mountain, Idaho; Creede, Colorado), where climates were more temperate than elsewhere at middle latitudes. This region has been documented for angiosperm flora as being one of the first areas at middle latitudes in North Amer- ica where climatic conditions became temperate during the Paleogene. Widespread volcanism along the cordillera added many cubic meters to the landscape in this region, building a highland that stretched from Arizona to Canada (Leopold $ MacGinite, 1972; Wing, 1987; Wolfe, 1987; Ruddiman & Kutzbach, 1991). Other middle-latitude areas such as eastern Asia may have supported pines during the warm humid periods of the Eocene, although only western North America is known to have had important upland regions. Although the early Tertiary fossil record of eastern Asia is not as complete as in western North America, pines have been found at a few middle-latitude locations in the Eocene. In Japan, the first Tertiary pines date to the last cool period of the Eocene and are not documented for the warm period at the end of the Eocene (Huzioka & Takahashi, 1970; Wolfe, 1985). These Eocene records may represent marginal populations of spe- cies that were centered farther north and expanded southward during the cool periods. Boreotropical angiosperms occurred throughout Asia in the warm humid periods of the Eocene, although they do not appear to have extended as far north as in North America (Chaney, 1940; Hsu, 1983). SECONDARY CENTERS OF PINE DIVERSITY The widespread extirpation of pine populations from middle latitudes at the end of the Mesozoic is hypothesized to have led to extinction of many pine species and to have greatly depleted genetic diversity in others. Many Cretaceous fossil pines had combinations of traits not known in extant species and represent lineages that went extinct. Of two closely related pinaceous genera of the Mesozoic, Pseudoaraucaria fossils have not been found in rocks younger than the Late Cretaceous, and the youngest Pityostrobus fossils date to the early Eocene (Miller, 1976). The disappearance in these two genera of many diverse lineages closely related to Pinus, represents major extinctions in the pine family during the early Tertiary. Despite this depletion of diversity in pines and taxa related to pines, the tectonic events and con- sequent migrations of the early Tertiary appear to have culminated in the creation of several new centers of pine diversity and the evolution of new pine lineages. This change appears to have been due in large part to increased tectonic activity. Although the period from the Mesozoic through the Paleocene had been relatively quiescent and land elevations were generally low across the con- tinents, by the late Eocene several regions were active tectonically. In subsequent epochs, volca- nism and mountain-building became increasingly important locally and globally. e uplift of new mountain ranges and volcanic activity created environmental heterogeneity. Ar- eas where active mountain-building coincided with Paleogene pine refugia are hypothesized to have become centers of pine radiation. Local climatic diversity was created by elevational differences, and rainshadow and other orographic effects de- veloped. Diversity of soils evolved, with many areas having newly disturbed sites following volcanic ac- tivity. Mountain-building created new barriers to migration and gene flow, causing lineages to be fragmented and isolated. All these conditions must have favored divergence and speciation in pines. Conversely, refugial areas that did not undergo major early Tertiary mountain-building, such as the high-latitude zones, were centers from which pines migrated in the Neogene, but did not become important centers of pine speciation. Although pines expanded and contracted from refugial areas during the fluctuating warm and cool periods of the Eocene, it was the terminal Eocene event that seemed to initiate major migrations out of refugia and to coincide with a time when pines appeared more widespread at middle latitudes. These Volume 80, Number 2 1993 Millar 491 Evolution of Pinus migrations provided further opportunities for di- vergence. As new environments were encountered, genetic isolation may have occurred, and possibil- ities for genetic drift by founder effects arose, as well as for hybridization with formerly isolated lin- eages. EVOLUTION WITHIN PINUS IN RELATION TO EOCENE REFUGIA By the end of the Paleogene, all subsections of Pinus, with the possible exception of Cembrae, had evolved (Axelrod, 1986; Millar & Kinloch, 1991). The events of the early Tertiary probably gave rise to at least two subsections, speciation of lineages within several subsections, and the current bioge- ography of many subsections and groups within several subsections. The hypothesis that pines were concentrated into refugial regions during the Еосепе explains the current bimodal distribution and pattern of diversity of pines at low and high latitudes (see Axelrod, 1986, for maps of pine subsections). In North America three subsections, Cembroides, Leiophyllae, and Australes, appear to have been concentrated in southern refugia during the Eocene and radiated from them subsequently (Axelrod, 1986). Subsections Cembroides and Leiophyllae seem to have been limited to western North Amer- іса and Central American refugia whereas Aus- trales may have had refugial areas with a broader southern distribution, including the Gulf Coast (Та- bles 3, 4). The current distribution of Australes in both the southeastern United States and in Central America/Caribbea suggests that Australes may have had refugia in both regions. Although not identifiable to subsection, the abundant pine fossils — the Eocene from southern Alabama (Table 3 are from a site that is within the range of several extant species of Australes. Small genetic distances between taxa of Australes versus Leiophyllae (Strauss & Doerksen, 1991) suggest that these subsections were in contact or did not diverge from each other until after the Mesozoic. Four subsections in North America appear to have been fragmented by Eocene events into sev- eral refugial regions. A division into northern and southern refugia in subsection Sylvestres is indi- cated by the extant lineages of P. resinosa (40% 52°N) and P. cubensis (restricted to Cuba, 22°N). Eocene fossils with affinities to Sylvestres that sup- port a northern refugium have been found in British Columbia and northern Washington (Table 3). Subsection Contortae similarly seems to have been divided into northern and southern refugia: the lineages leading to P. contorta and P. banksi- ana clearly had northern origins and boreal ad- aptations, whereas those leading to P. clausa and P. virginiana had southeastern affinities. Eocene fossils with affinities to Contortae have been found in northern Washington, and Oligocene fossils in British Columbia, corroborating a northern refug- ium (Tables 3, 4). Subsection Ponderosae also may have had both northern and southern refugia. There are 10-17 species endemic or nearly endemic to Mexico and entral America, whereas wide-ranging Pinus pon- derosa has a northern distribution and northern ecological affinities in из habitat preference and vegetation associates. Early Tertiary pines most closely allied to Ponderosae have been found in British Columbia (Eocene, Table 3), Washington, and Montana (Oligocene, Table 4). northern Lineages of Ponderosae may have been concen- trated also in Rocky Mountain refugia, as suggested by Oligocene fossils from Colorado with affinities to this subsection (Table 3). A recent re-evaluation of evolutionary patterns in Ponderosae also indi- cates northern and southern division in the sub- section (Lauria, 1991). Pinus ponderosa is con- sidered to originate in northern latitudes, while other distinct phylads had southern origins in Mex- ico and Central America. Subsection Strobi has northern and southern lineages in North America. Ancestral lineages of Pinus monticola and P. strobus would have mi- grated from northern refugia, and those of P. stro- biformis, P. ayacahuite, and P. chiapensis (Mar- tinez) Andresen (distinct from P. strobus, of which it was once classed as a variety; Andresen, 1964, 1966) would have migrated from southern refugia. Northern refugia are corroborated by Eocene fossil pines allied to subsection Strobi from British Co- lumbia (Table 3). Lineages of subsection Strobi may also have been fragmented into middle-latitude refugia in the Rocky Mountain region, represented currently by P. flexilis and possibly P. strobifor- mis, and by Oligocene fossils from Idaho and Col- orado (Table 4). e small subsection Balfourianae is an ancient lineage (Kossack, 1989; Strauss & Doerksen, 1991; Millar & Kinloch, 1991) that appears to have been entirely concentrated in middle-latitude Rocky Mountain refugia during the early Tertiary. Pres- ent distribution of the three closely related species is in the Rocky Mountain/Great Basin/Sierra— Cascades Ranges of the western United States be- tween 35°N and 41°N latitude. Early Tertiary fos- sils with probable affinities to this subsection were found in Idaho (Eocene) and Colorado (Oligocene) (Tables 3, 4). 492 Annals of the Missouri Botanical Garden In Eurasia, sequences of conifer-bearing depos- its from the early Tertiary are not as widely dis- tributed as in North America, but evidence for pine evolution in relation to refugia exists. Three ancient subsections, Canarienses, Pineae, and Krempfi- ani, have extant species restricted to regions that were along the Tethys seaway. Fossil evidence ex- ists for a Tethys refugial area of pines in southeast Asia, in the region where P. Кгетрји now occurs. Pine was recorded in abundance throughout early Tertiary strata and in coastal southeast China and in northwest Borneo (Tables 3, 4). Other fossil floras for southeast Asia include temperate taxa, suggesting that pines would have found favorable habitats In western Europe, the boreotropical angio- sperm flora was recorded abundantly from 45°N to 56?N (Chaney, 1940). Although there are no fossil records from currently low latitudes in Eu- rope, pine fossils are known from areas of southeast Europe that were along the Tethys seaway (Chi- guriaeva, 1952; Mirov, 1967). Furthermore, there is indication of dry zones in the middle Eocene along the European and north African Tethys, in- directly supporting the occurrence of pines there (Parrish et al., 1982; Parrish, 1987). Fossil pines allied to Canariensis and Pineae were widespread along the Tethys seaway in western Europe during the Miocene (reviewed in Klaus, 1989), suggesting more widespread distributions for both subsections in this region in the early Tertiary. in Eurasia several sub- sections appear to have been divided by Еосепе events into northern and southern refugia. In Syl- As in North America, vestres, lineages represented by extant P. sylves- tris most likely migrated from northern refugia. Eocene pine fossils from high latitudes in western Siberia allied to Sylvestres corroborate a northern refugium in Eurasia. The remaining 14 extant spe- cies in Sylvestres, excluding P. densiflora, and P. thunbergiana, and possibly P. yunnanensis, rep- resent lineages from southern refugia. Although few Eocene deposits from southern Europe are plant-bearing, abundant fossils allied to Sylvestres are known from Oligocene and Miocene deposits in the Tethys region (Table 5; Palibin, 1935; Czec- zott, 1954; Klipper, 1968; van der Burgh, 1973; Klaus, 1989). ' represent lineages from middle-latitude refugia, as documented by fossil pollen of unknown affinity from the Eocene and Oligocene of Japan and north- ern China (Tables 3, 4). Subsection Gerardianae may have been divided between a southern Tethys 'he exceptional three species may refugium, as represented by extant P. gerardiana, and a middle-latitude refugium in northern China, as represented by extant P. bungeana, and by early Tertiary deposits containing pine pollen in China (Tables 3, 4). Refugia for subsection Strobi in Eurasia were mostly along the Tethys. Of the eight extant Eur- asian species in the subsection, only P. parviflora seems clearly allied to lineages derived from more northerly refugia. The remaining seven modern species are all distributed along the former Tethys region. No early Tertiary fossils from Eurasia have been specifically allied with subsection Strobi, al- though Miocene fossils of the subsection testify to their presence in that region (Klipper, 1968). The consequence of certain Eocene refugia be- coming secondary centers of pine diversity is best described for Mexico and Central America. Al- though parts of the Mexican and Central American isthmuses were transiently under water during the Mesozoic (Kellum, 1944; Eguiluz Piedra, 1985), they were elevated during the Еосепе and later Tertiary. Tectonic activity was especially great in Mexico during the late Eocene and Oligocene/ Miocene. Tertiary volcanic activity significantly reshaped the Sierra Madre Occidental and created the Sierra Madre Oriental and the Transversal Neovolcanic Axis. Uplift in the early Eocene rebuilt almost the entire ranges of the Sierra Madre del Sur and Sierra Madre de Chiapas (Eguiluz Piedra, 1985). Mexico and Central America are home to as many as 83 extant taxa of pines, including 48 species, 21 varieties, and 14 forms (Eguiluz Piedra, 1985) ation and speciation are ongoing, which has been attributed to active mountain-building that began in the Eocene (Equiluz Piedra, 1985; Axelrod, 1986; Karamangala & Nickrent, 1989; Lauria, 1991). Active radiation is evidenced by closely related species clusters, with subsections Ponde- rosae and Cembroides the best examples. There are 10—17 species of Ponderosae (Eguiluz Piedra, 1985) endemic to Mexico and Central America and 10-16 species in Cembroides (Zavarin, 1988). The difficulty of separating the taxa of Cembroides, . Within many subsections, extensive radi- implying close genetic relationship, has led some botanists to the conclusion that Cembroides is a young subsection. Phylogenies based on DNA-se- quence divergence indicate, however, that section Parrya, including subsection Cembroides, is an- cient (Kossack, 1989; Strauss & Doerksen, 1991). Ancestral Cembroides lineages would have been concentrated in Mexican/ Central American refu- gia during the early Tertiary, and the major pulse of radiations now occurring in the subsection may have begun in the early Tertiary. Volume 80, Number 2 1993 Millar 493 Evolution of Pinus Two subsections may have originated in Mexi- can/Central American refugia (Axelrod, 1980, 1986). Unlike other subsections, Oocarpae and Sabinianae have no Mesozoic fossil record and are not clearly documented until the Miocene. On the basis of specialized and apparently derived mor- phological adaptations (Shaw, 924; Little & Critchfield, 1969; Klaus, 1980; van der Burgh, 1984), these subsections have long been considered to have originated recently. Phylogenetic analysis of DNA divergence also confirms their relative youth, especially of Oocarpae (Strauss & Doerk- sen, 1991). Using fossil and floristic evidence, Ax- elrod (1980) traced the origins of the northern (California /Baja California) elements of Oocarpae to mainland Mexico/ Central America prior to the Miocene. The northern lineages appear to have diverged by the time they reached California (Millar et al., 1988), indicating that the radiation events occurred farther south in the early Tertiary. Al- though the systematic coherence of the extant Lat- in American taxa of Oocarpae remains uncertain, с relationships of some Latin American spe- cies link Oocarpae with Australes and Ponderosae (Critch&eld, 1967), both of which have Mesozoic fossil records, appear to have had Mexican/ Central American Paleogene refugia, and may have been ancestral to Oocarpae. Subsection Sabinianae has a limited fossil rec- ord consisting of a single taxon allied to P. sabi- niana and confined to southern California (Axelrod, 1986). Fossils are known only from the late Mio- cene through Quaternary, and Axelrod (1981) traced the origin of the extant species to Mexican species of Ponderosae. Genetic relationships of ex- tant taxa link Sabinianae with Ponderosae on the basis of terpene affinities (Zavarin et al., 1967 and crossing evidence (Critchfield, 1966; Conkle & Critchfield, 1988). Together this evidence points to а Mexican/Central American origin of Sabi- — nianae from early Tertiary Ponderosae lines. PLEISTOCENE VERSUS EOCENE IMPACTS In this paper I document that tectonic, climatic, and biogeographic events of the Eocene had a major impact on pine distributions and evolution. The question may arise whether the effects of the Eocene were so confounded by subsequent events of the Tertiary and especially the Quaternary as to be indecipherable. I will address only the Pleis- tocene, since its potential was probably the great- est. The Pleistocene was a time of profound change unprecedented in the history of the Earth. In some ways the Pleistocene was analogous to the Eocene. The Pleistocene was also an epoch with fluctuating climates, but the deviation from temperate climates in the Pleistocene was toward glacial conditions, whereas in the Eocene it was toward tropical ep- isodes. The amplitude in average temperature be- tween glacial and interglacial periods of the Pleis- tocene (5°-10°C, Bowen, 1979) was about the same as estimated for the cool and warm periods of the Eocene. The Pleistocene differed in lasting only 2 million years (cf. 20 million years for the Eocene), in having many more cycles (16-18, Bowen, 1979), and in having alternating periods of unequal duration with glacials longer than in- terglacials. The events of the Pleistocene had enormous effects on vegetation, including pines. In northern latitudes, pine distributions were displaced by con- tinental ice sheets (e.g., Pinus contorta/ P. banks- iana, Critchfield, 1985); in mountainous regions elsewhere, species migrated up or down in elevation Miki, 1957; Van Devender & Spaulding, 1979). Along coasts and in other lowlands, pine popula- ~ tions shifted north and south in response to the climate cycles (e.g., Oocarpae, Axelrod, 1980; Millar, 1983). Concomitant to the shifts in distri- bution of pines were major changes in the genetic structure of species. The flux of population expan- sion and contraction, coupled with drastically changing selection regimes, affected the structure of genetic variation within species and allowed some species to hybridize (Critchfield, 1984, 1985; Kin- loch et al., 1986; Millar, 1989). In general, however, the Pleistocene does not appear to have completely reshuffled the genus in the way the Eocene did, and many of the Tertiary patterns and the evolutionary events that date to that period have been maintained. Notwithstanding the existence of local refugia, Pleistocene events primarily affected Pinus in a gradient from north to south, with the effect that species and popula- tions shifted south then north (or down then up in elevation) following the cycle of the glacial and interglacial periods. The impact of the Eocene, by contrast, was greatest in the latitudinal center of the genus and had the effect of dissecting the genus and concen- trating pines into widely disjunct regions. Further- more, during the early Tertiary, and unlike the Pleistocene, almost no upland regions (except in interior western North America) could offer local refuge from unfavorable climates. Whereas many pine species appear to have gone extinct in the early Tertiary, no pine extinctions in North Amer- ica are attributed to the Pleistocene (Critchfield, 494 Annals of the Missouri Botanical Garden 1984), although western Europe suffered a signif- icant impoverishment of pine flora (Klipper, 1968). ome speciation, for example, in Balfourianae, Cembroides, and Oocarpae, seems to have been triggered by the Pleistocene, but no major new trends have emerged. Insufficient time has elapsed since the close of the Pleistocene for its full impact to be felt on evolution in Pinus. Patterns initiated by the Pleis- tocene appear minor compared to the effects of the Eocene and are insufficient to erase the evo- lutionary i impacts of the early Tertiary. Тћиз, many of the major evolutionary patterns of the early Tertiary can still be traced in the biogeography and relationships of extant pines. VALIDATION OF THE HYPOTHESIS The arguments developed in this paper result in a working hypothesis about the effect of the Eocene on pine evolution. The hypothesis is a reconstruc- tion of pine history based on available information from the fossil record, climatic and tectonic evi- dence, and biogeographic record of angiosperms and conifers. Many gaps in information exist that, if filled, would corroborate or negate this hypoth- esis. These gaps fall into several categories. Ex- panded fossil records are urgently needed, es- pecially in regions hypothesized as Eocene pine refugia that currently have meager fossil docu- mentation. These include Paleogene records for entral America and Mexico, low latitudes along the European Tethys, and high latitudes in eastern Asia. Technologies that allow more accurate iden- tification of fossil affinities for macro- and especially micro-fossils at intrageneric levels would help in tracking the biogeography and evolution of these roups. These technologies need to be applied to published fossil floras, with revised taxonomic lists. Certain pine lineages have especially meager fossil records, including Cretaceous records of sections Parrya, especially Gerardianae and Balfouria- nae, and Cretaceous/early Tertiary records of sec- tion Pinea. Accurate dating of fossils and corre- lation with paleoclimatic and paleogeological events will also help to track the fine scale paths of these pines. Finally, existing and emerging genetic and molecular technologies should be applied to esti- mate genetic distances and divergence times among subgeneric groups. Such genetic techniques will allow tests of the hypothesized times of geographic isolation. LITERATURE CITED ALvIN, K. L. 1960. Further conifers of the Pinaceae from the Wealden Formation of c Mem. Inst. Roy. Sci. Nat. Belgique. 146: 1-39. ANDRESEN, J. W. The taxonomic status of Pinus F Phytologia 10: 417-421. A multivariate analysis of Pinus chia- mie кө ы strobus phylad. Rhodora 68: 1- 24. AXELROD, D. 1. 1965. А method for determining the altitudes of Tertiary floras. Palaeobotanist 14: 144- 171. The Eocene Copper Basin Flora of оаа Nevada. Univ. Calif. Publ. Geol. Sci. 1968. Tertiary floras and topographic о of the Snake River basin, Idaho. Bull. Geol. Amer. ps 113-73 Я Desert и its age and origin. Pp. 1- d in J. R. Goodin & D. K. Northington (editors), Arid Land Pus Resources. International Center for Arid and Semi-Arid Land Studies, Lub- — = [5] OZ fo un А History of the maritime closed-cone n Alta and ЊУ о“ Univ. Calif. Publ. Geol. Sci. 120: 1 юс climatic changes in relation to vegetation disjunction and speciation. Amer. Nat- uralist 117: 847-870. 86. Cenozoic history of some western Amer- ican | pines. Ann. Missouri Bot. Gard. 73: 565-641. & H. P. ВАПЕҮ. 1976. Tertiary vegetation, climate ма altitude of the Rio Grande depression, New Mexico-Colorado. Paleobiology 2: 235-254. & . RAVEN. Auc of the Cordil- leran flora. J. Biogeogr. 12 Banks, H. . ORTIZ- им & C. M. HARTMAN. 1981. Pinus escalantensis, sp. nov. a new permi- neralized cone from the Oligocene of British Colum- bia. Bot. Gaz. (Crawfordsville) 142: 286-2 Barron, E. J. Climatic e " the vari- able obliquity explanation of T ceous- Paleogene high latitude floras. Geology 1 505. 597. 19 Estimations of fis Tertiary global cool- ing trend. Palaeogeogr. Palaeoclimatol. Palaeoecol. 50: 45-61. BECKER, E. W. . Oligocene plants from the upper Ruby River аш southwest Montana. Mem. Geo. er. BERNER, R. A., s LASAGA & R. M. CARRELS. 1983. The carbonate-silicate geochemical cycle and its ef- fect on atmospheric carbon-dioxide over the past 100 million years. Amer. J. Sci. 283: 641-683 BLACKWELL, 1984. Fossil Ponderosa-like pine wood from the Upper Cretaceous of northeast Mis- 1979. Geographical hup ie on the _ Quaternary. Een Phys. Geogr. 6. : | e spores eee ds of the dE јен nd + Maryland. Dept. Geol. Mines Wa- ter . Bull. BROWN, R. `W. T The recognizable species of the Green SM flora. U.S. Geol. Surv. Prof. Paper 185- C: 45 Additions to some fossil floras of the western a United States. U.S. Geol. Surv. Prof. Paper 206. 186-Ј: 163- BuRCHARDT, B. 1978. Oxygen isotope palaeotempera- Volume 80, Number 2 1993 Millar 495 Evolution of Pinus tures from the Tertiary period in the North Sea area. Nature 275: 121-123. CHANEY, R. W. 1940. apud кек and continental history. Bull. Geol. Soc. Amer. 51: 469-488. —————. 1954. A new pine Cretaceous of om th Minnesota and its edet significance. Ecol- ogy 35: 145-151. CHIGURIAEVA, А. А. . Materials on Eocene vege- tation of the right bank of the River Don. Saratov Univ. Sci. Proc. 35: 197-200. [In Russian.] CoNkLE, M. T. € W. B. CRITCHFIELD. 1988. Genetic variation and hybridization of ponderosa pine. Pp. 21-43 in Ponderosa Pine: The Species and its Man- agement. Washington State Univ. Coop. Ext. Pull- man, Washington. CREBER, С. T. & W. С. CHALONER. indications from tree rings in fossil woods. Pp. 4 in P. T. Brenchley (editor), Fossils and Climate. Wiley & Sons, New York. CRITCHFIELD, W. В. 1966. Crossability and у cia of the California big-cone pines. U.S.D.A. For Serv. Res. Pap. NC-6: 36-44 1967. Crossability and relationships of Ass California closed cone pines. Silvae Genet. 16: 97. 1984. Climatic 984. Impact of the Pleistocene on the genetic structure of North American conifers. Pp. 70-118 in R. M. Lanner (editor), Proceedings of the 8th North America Forest Biology Workshop; Logan, Utah, 30-August 1, 1984 The jn Quaternary S d E Pod pole and jae pine. Canad. J. Forest Res. А E. L. LITTLE. 1966. Geographic disiribunon of pines of the world. U.S.D.A. Forest Serv. Mis Publ. 991. CzEczorr, H. 1954 of Pinus halepensis and P. brutia Bot. Congr. Paris & Nice 8: 19 | Dickinson, К. A., К. D. CUNNINGHAM А Т. А. AGER. 1987. Geology and origin of the Death Valley ura- nium deposit, Seward Peninsula, Alaska. Economic Geology 82: 1558-1574. Donn, У. L. The enigma of high-latitude pa- leoclimate. | m Palaeoclimatol. Palaeoecol. -212 The past and present distribution x M Internatl. Рокот, Р. I. Tertiary floras of western Si- = a. Dokl. Akad. Nauk. S.S.S.R., Moscow. [In Rus- iu peur 17 PIEDRa, T. о Origen y evolución del gé- o Pinus. Dasonomia Mexicana 3(6): 5-31. 1988. Соо де Іоѕ Ріпоѕ Ріпопегоѕ Мех- icanos. Pp. 83-92 in М. Е. Passini, D. C. Tovar & T. Eg ба cun ыо. Proceedings of II Sim- posio Nacional Sobre Pinos Pinoneros. Centro de EDWARDS. Palynology of Мааа орнар. and Paleocene , Lower Colville River, Northslope of Alaska. ‚ J. Earth Sci. 25: 512-527. , W. G. CHALONER & P. R. CRANE (editors). "The Origin of Angiosperms and Their Bio- logical Consequences. Cambridge Univ. Press, Cam- Outline of the origin and historical between Asia and in А. Graham ridge. GRAHAM, А. 1972. recognition of floristic affinities eastern North America. Pp. 1-16 (editor), Floristics and Paleofloristics of = and East- ern North America. Elsevier, Amster 197 Studies i in neotropical пт ions II. The Miocene communities of Puerto Rico. Ann. Mis- souri Bot. Gard. 56: 308-357 Gray, J. 1960. Temperate pollen genera in the Eocene ле pm Alabama. Science 132: 808-810. Скоот, J. J., ENNY & C. R. Скоот. 1961. Plant о E the age of the Raritan, Tuscaloosa, and Magothy formations of eastern USA. Palaeon- tographica 108B (3-6): 121-140 Guo, S. 1980. Late Cretaceous and Eocene floral prov- inces. Inst. Geol. Palaeontol., Academia Sinica, Nan- jing. 5 О. 1868-1883. Flora Fossilis Arctica, 6 vol- ‚М. R. Dawson & D. К. HOI. 1983. Arctic od biota: Paleomagnetic evidence of age disparity with mid-northern latitudes during the late Cretaceous and early Tertiary. Sci- HiLis, L. V., J . R. SWEET. 1974. Juglans eocinera, nov. sp., Beaufort Formation (Ter- tiary), southwest Banks Island, Arctic Canada. Can ad. -90. HorLICK, А. & E. C. JEFFREY. 1909. Studies on Cre- taceous coniferous remains from Pire New ork. Mem. 3: Hopkins, М. S., . RUTTER & , Maus dd paleoecology, and palynology of some Oli- sint ocky im Pert ~ British anad. 7. Earth Sci 460 : 3. hs Cretaceous id a enozoic ene in China, emphasizing their pine ae ps or ica. i -508. Huzioka, K. & E. TAKAHASHI. 1970. The er flora of the Ube coal-field, southwest yim em J: Min. Coll. Akita. Univ., Ser. A., 4(5): 1 90 JEFFREY, E. C. On the structure a leaf in Cretaceous pines. Ann. Bot. (London) 22: 207-220. & D. L. NICKRENT. 1989. An KARAMANGALA, R. R electrophoretic study of representatives of subgenus B Diploxylon of Pinus. Canad. J 07: 1750- 1759. KELLUM, L. 1944. Geologic history of northern i ous and its bearing on petroleum cM Bull. Am Assoc. Petrol. Geol. 28: 301-325. Kerr, R. A. 1991. Did a bon of volcanism overheat ancient earth? Science 251: 746-747. KiNLocH, B. B., R. D. WESTFALL & С. I. Forrest. 1986. Caledonian Scots p pine: origins and genetic structure. New Phytol. 104: 703-729. Kraus, W. Neus Beobachtungen zur Morphol- apis des Zapfens von Pinus und ihre Bedeutung für die Systematik, Fossilbestimmung, Arealgestaltung, und Evolution der Gattung. Pl. Syst. Evol. 134: 137- 171 19 на оу pines and their history. PI. HA Evol. 162: =163 KLIPPER, К. аиа я aus den Tertiaren Deckschichten des Niederrheinischen Hauptflózes. 1 & 2. Palaeontographica Abt. B. 121: 159-168; 123: 213-220, Kossack, D. S . The IFG copia-like element. Characterization of a transposable element present at high copy number in the genus Pinus and a history 496 Annals of the Missouri Botanical Garden of the pines using IFG as a За ли Ph.D. Disser- c land е ои in the arly Tertiary. Commenta Biol. 29(5): 1-5. tene J. H., B. HACKNER & A. Eu RTLETT. 1967. angrove pollen at the Rosin site of Oligo- iocene amber from Chiapas, Mexico. Bot. Mus. 9-324. , €. J. SMILEY & J. Gray from the coastal plain of Alaska. Bull. Geol. ie ‘eas 71: 1345-1366. Lauria, F. 199 nomy, systematics, and phylog eny of ide eee Ponderosae London (Pina. ceae). Alternative concepts. Linzer Biol. Beitr. 23(1) 129- giji. E B. & H. D. Macc INITE. 1972. Develop- Машаа, Pp. 147-200 in А 1 Floristics and Paleoforisties of Asia and Eastern North America. Elsevier, Lon LESQUEREUX, L. 1883. Gant Diei to the fossil flora of the Western Territories III. The Cretaceous an S. Geol. Surv. of the Territories . & W. B. CRITCHFIELD. 1969. puri of the s M (Pines). U.S.D.A. Forest S Misc. Pub. МАССІМІТЕ, Н. ^n PE dium taris of the florrissant beds, Colorado. Publ. Ca e Inst. Wash. 599 Mal, D. Н. 1970. куены, Elemente i im rie d "ia en Paláontol. Abh. Abt. B, Paláobot. а 03. MANU d s. 1962. Studies in the Tertiary flora of Spits- bergen, with notes on Tertiary flora of Ellsmere Is- land, Greenland, and Iceland. Norsk Polarinst. Skr. 125, S. 1-127. MARTIN, P. & B. E HARRELL. 1957. The Pleistocene Fossil records of some west Amer- ican conifers, Publ. Carnegie Inst. Wash. 346: 139- 158. McGowran, В. 1990. Fifty million years ago. Amer. Scientist 78: 30-39 McKenna, M. С. 1983. EST land mass rearrange- ment, cosmic events, and Cenozoic terrestrial organ- isms. Ann. Missouri Bot. Gard. 70: 459-489. 5 with special refer- echn. Os dia City Univ., Ser. D, Biol. 8: 221-272 Милк, C. I. 1983. A steep cline i in Pinus muricata. Evolution 37: 311-319. Allozyme variation of bishop pine as- sociated жу pygmy forest pe in northern Califor- nia. Canad. J. Forest Res. 19: 870-879. — а B. K. KINLOCH. Dor Taxonomy, phylog- eny, and coevolution of pines and their stem rusts. ia 1-38 in Y. Hiratsuka, J. Samoil, P. Blenis, P. ane & B. Laishley (editors), Rusts of Pines. Pro- nie Hier Rusts of Pine W ence Sep. 18-22, Forestry Canada, Kanne. Alberta. Inf. Rej ‚ S. H. STRAUSS, M. T. CONKLE & R. D. Wes 1988. Allozyme differentiation and biosystenaties of the California closed cone pines (Pinus subsec Oocarpae). Syst. Bot. 13: 351-370. MiLLER, С. N. 1969. Pinus avonensis, a new speices of petrified cones from the Oligocene of Montana. Amer. J. Bot. 56: 972-978. 1973. Silicified cones and vegetative remains of Pinus from the Eocene of British Columbia. Con- trib. Univ. Michigan Mus. Paleontol. 24: 101-118. inus wolfei, a new petrified cone from the Eocene of Washington. Amer. J. Bot. 61: 777 (id. . 1976. Early dnd in the Pinaceae. Rev. Palaeobor. Palynol. 21: 101-117. —— —— —. 1977. Mesozoic чо ба) Bot. Rev. (Lancas- ter) 43: 217-280. — ——-. 1982. Current status of Paleozoic and Mesozo- ic па Rev. Palaeobot. Palynol. 37: 99-114. 1988. The origin E deals conifer families. Pp. 448-487 in C. B. k (editor), Origin and Evolution of Gy и "Columbia Univ. Press, New York & J. M. MALINKy. 1986. Seed cones of Pinus from the Late Cretaceous of New ми USA. Rev Palaeobot. Palynol. 46: 257-272 Mirov, N. T. 1967. The Genus Pinus. York. HAsBROUCK. 1976. The Story of Pines. Indiana Univ. Press, Bloomington. ше: n Montane pollen from the Tertiary rthwest Borneo. Blumea 14: 231-235 | Spore- pollen evidence for early Oli- pisode in northern Ronald Press, 32. On the structure “d affinities of some Cretaceous plants from Hokkaido. 2nd Contribution. = og Univ., Tokyo, Sec. 3, Bot. 2 (part 7 О” бо T D & T. J. AHRENS. 1989. Impact рго- duction of CO, by the Cretaceous/ Tertiary extinction olide bon the resultant heating of the Earth. Nature 338: 247-2 PALIBIN, 1. { 935. Stages of development of gs ora from the сте сеоце period. Sovetsk Bot. 87. Global palaeogeography and pa- laeoclimate of the late Cretaceous a nd early Tertiary. 51-73 in F. M. Fris, W. G. Chaloner & P. R. Crane (editors), The Origins of Angiosperms and their Biological Consequences. Cambridge Univ. Press, ambridge А. ER & С. К. ЅСОТЕЅЕ. 1982. Rain- fall patterns and the distribution of coals and evap- orites in the Mesozoic and Cenozoic. Palaeogeogr. Palaeoclimatol. Palaeoecol. 40: 67-101. Penny, J. S. 1947. Studies on the conifers of the Magoth y flora. Amer. J. Bot. 34: 281-296. 197 PIEL, K. M. Palynology of Oligocene out from ees British Columbia. Canad. J. . 49: 1885-1 Pierce, R. 78 dure Minnesota Cretaceous pine pollen. Science 125: 26. Prick, R. А. 1989. The genera of Pinaceae in the southeastern United States. J. Arnold Arbor. 70: 247-305. Кыр, E. M. € M. E. J. CHANDLER. 1933. The London РА ure British Museum Natural History. Volume e T C 1984. Past and Present Vegetation of Volume 80, Number 2 1993 Millar 497 Evolution of Pinus the Far Northwest of Canada. Univ. Toronto Press, Toronto. RoBisoN, C. R. 1977. Pinus triphylla and Pinus quin- ar е the Upper Cretaceous of Massachu- setts. r. J. Вог. 64: 726-732. Rouse, G. E P W. Н. MarruEws. 1979. Tertiary geology and palynology of the Quesnel area, British Columbia. Bull. Canad. Petrol. Geol. 27: 418-445. Ruppiman, W. К. & J. E. KurzBacH. Forcing of late Cenozoic Northern Hemisphere climate by iine E in southern Asia and the American LE Res. 91(D15): 18409-18427. оа uplift and climatic ange. Sci. a 264(3): 6 Savin, S. M. 1977. The history the earth’s suface temperature during the past 100 million years. An- nual Rev. Earth Planetary Sci. 5: 319-356. ScHoRN, Н. E. € J. A. Worrk. 1989. Taxonomic revision of the Spermatopsida of the Oligocene Creede flora, southern Colorado. U.S. Geol. Surv. Open-file Report 89-433, Washington, D.C. SCHWEITZER, H. J. Die “tertiáren” Spitsbergen. Palaeontographica Abt. B 1 а=» N. & A. BoersMa. 1981 The climate Зи > ocean. J. Geol. Soc. London 138: Пн 89 n E R. 1914. The genus Pinus. Publ. Arnold Arbor. 5. 24. ie on the genus Pinus. J. Arnold Ай. 5: ras SMITH, А. G., А. M. on & J. C. BRIDEN. Ped Pu ue World Maps. bridge Univ. Press, Cambridge. STOCKEY, R. A. ee Pinus driftwoodensis sp. n. from the early Tertiary = | Columbia. Bot. Gaz. (Crawfordsville) 144: 56. Middle M ene remains from British es Bot. Gaz. (Crawfordsville) 145: 262-274. M. NISHIDA. Pinus harborensis sp. nov. and the affinities of permineralized leaves p the red Cretaceous of Japan. Canad. J. Bot 1856-1866. 1981. Cam- Үү Џера. 1986. Permineralized pinaceous leaves from the m ace of Hokkaido. Amer. J. Bot. 73: 1157-1 STOPES, M. C. & E. M. B 1910. The anatomy of Cretaceous pine leaves. Ann. Bot. (London) 24: Strauss, S. H. & A. H. DoERKSEN. 1991. Restriction fragment analysis of pine phylogeny. Evolution 44: 1081-1096. Sunc, Z. 1958. Tertiary spore and pollen gar es from the red beds o iu-Chuan, Kansu, and t geological and botanical significance. Acta enin tol. Sinica 6: 159-167 Liv. 1976. The Paleogene spores and pollen grains from the Fushun coal by. northeast China. Acta Pulasontel. Sin. 15: 14 Tanal, T. 1970. The Oligocene floras sin ihe Kushiro coal field, Hokkaido, Jap. J. Fac. Sci., Hokkaido Univ. ser. 4, 14: 14 1972. Terr history of vegetation in Japan Pp. 235-256 i m (editor), Floristics iid Paleofloristics of Asia he Ea North America. Elsevier, London ТАЗТОК, D. W. & L. J: HICKEY. 1990. An Aptian plant with attached leaves and flowers: Оев for angiosperm origin. Science 247: 702-704. TIFFNEY, B. M. 1985a. The Eocene North Aan land bridge: its importance in Tertiary and modern phy- jer "E us Northern Hernies T dd Arbor. 66: ач on the origin of the flo- ristic ра between eastern Asia and eastern America. J. Arnold Arbor. 66: 73-94. UEpa, Y. & M. Мана. 1982. On petrified pine leaves from the Upper Cretaceous of Hokkaido. J. Jap. Bot 1: 133= UNDERWOOD, J. C. € C. a MILLER. 1980. Pinus bu- chananii, a new species based on a petrified cone from the Oligocene of. Washi ington. Amer. J. Bot. 67: 1 oe pi UrcHurcH, G. R. & J. A. WoLreE. 1987. Mid-Creta- ceous to ш: Tertiary vegetation and climate: evi- dence from fossil leaves 5 = Consequences. Cambridge Univ. Press, Cambridge VAN DER BURGH, 973. Holzer der Niederrheinischen Braunkohlen formation 2. Holzer der Braunkohlen- gruben * ah Theresia” z u He erzogenra n “Zu- W baud id “Victor” zu Zülpich. Nebst einer мав matisch anatomischen Bearbeitung der Gattung Pinus L. Rev. Palaeobot. Palynol. 15: 13-21 ai . 1984. Phylogeny and biogeography of the genus Pinus. Pp. 199-206 in A. Farjon (editor), Pines, Pravin s and Descriptions of the Genus Pinus. . W. Backhuys, Leiden W. G. Senne, 1979. De- velopment of vegetation and climate in the south- western United States. Science 204: 701-710. WAHRHAFTIG, C., J. A. WoL a | B. LEoPoLp. 1969. The coal- bearing grou e Nenana coal field, А е7 к Geol. Өш. Bull. 1270 1-30. WING, Eocene and Oligocene floras and vegetation ge the Rocky Mountains. Ann. Missouri . Gard. 74: 748-784. Кози. R. P. 1933. Tertiary pollen, II. The oil of the Eocene Green oi ver Formation. Bull. rrey Bot. Club 60: 479-5 "i FE, ais A. 1969. Neogene icm and vegetational history of the Pacific Northwest. Madroño 20: 83- 110. —. 1971. Tertiary climatic fluctuations and meth- ods E analysis of Tertiary floras. Palaeogeogr. Pa- laeoclimatol. Palaeoecol. 9: 27-57 1972. An interpretation of Alaskan Tertiary борай, Рр. 201-233 in А. Graham (editor), Floristics and Paleofloristics of Asia and Eastern North Amer- ica. Elsevier, Amsterdam. Some apadt of plant geography of the Шам H g the oo = las Ann. Missouri Bot. Gard 62: pa "E ~ а, 1977. Paleogene floras from the Gulf of Alas- ka region. U.S. Geol. Surv. Prof. Paper 997 1978. A paleobotanical interpretation of Ter- tiary climates in the Northern Hemisphere. Amer. Sci. 66: 694 we A chronologic framework for the Ce- nozoic megafossil floras of northwestern North Amer- 498 Annals of the Missouri Botanical Garden ica in relation to marine geochronology. Pp. in J. M. Armentrout (editor), Pacific Northwest и p Testen unnm Special Pap. Geol. Soc. Amer. 1 Distribution of major vegetational y По EE e Tertiary. Geophysical Monogr. 32: 375. А . Ап overview of the origins of the mod- ern vegetation and flora of the northern Rocky Moun- 03. А Palaeobotanical evidence for a marked temperature increase following the Cretaceous /Ter- tiary boundary. Nature 343: 153-156. . E. SCHORN. 1989. Paleoecologic, paleo- climatic, and evolutionary significance of the Oligo- cene Creede flora, Colorado. Paleobiology 15: 180- 198. & UPCHURCH. 1986. Vegetation, = matic, and flor al changes at the Cretaceous- Tertiar boundary. Nature 324: 52: & W. Жене. 1987. Middle Eocene dicoty- ledonous үч from Ma northeastern Wash- S. Geol. Soc. Bull. 1597. 988. Taxonomy s pinyon pines. Pp. 29- | assini, D. C. Tovar, & T. Equiluz Piedra (editors), и of the П Simposio Na- cional Sobre Pinos Pinoneros 6-8 Agosto 1987. Cen- tro de Genética Forestal, Chapingo, Mexico. . HATHAWAY, TH. REICHERT & Y. В 1967. LiNHART. hemotaxonomic study of Pinus torreyana Parry turpentine. Phytochemistry 6: 1019-1023. ZIEGLER, A. M., C. R. Scorese & S. R. BARRETT. 1983. эко and Cenozoic paleogeographic maps. Pp. 0-252 in P. Brosche . Sunderman (editors), Tidal Friction and the Earth’s Rotation, Volume II. Springer-Verlag, Berlin. PLANT LIFE-FORMS AND BIOGEOGRAPHIC RELATIONS OF THE FLORA OF LAGUNILLAS (3095) IN THE FOG-FREE PACIFIC COASTAL DESERT' J. J. Armesto and P. E. Vidiella? ABSTRACT The coastal desert of north-central Chile supports a rich but poorly known flora. Here, we document the importance of the annual and perennial habits i in the flora of Lagunillas (3095), at the southern transition from plants represent 41% of the total flora (191 species) and a most importan xcluding non-native weeds, however, decreases the importance of annuals to 33%. This value is higher than that for high-altitude deserts, but lower than for most other, less- -equitable, lowland deserts both in North America and eastern Asia. rubs, geophytes, and other perennial herbs comprise 33%, 12%, and 21% of the native flora, respectively. The Cosmopolitan element, including weed species, accounts for 41° ^o of the coastal desert flora. South American taxa include desert endemics (13%), Andean (12%), and Tropical American (10%) species. A large proportion (15%) of the taxa, especially shrubs, have disjunct distributions in the arid regions of North and South America. сом suggest that the present coastal desert flora derives primarily from recently div ersified coastal desert ша Апдеа m endemic to these areas, and secondarily from cosmopolitan weeds. resence of a number o ofa mphitropic xerophytes lends support to the hypothesis ce arid habitats were continuous through the lowland tropics xn the Glacial period. The Pacific coastal desert of South America ranges from southern Ecuador at са. 5°S, to about 3295 in north-central Chile (Rauh, 1985). Three major climatic and vegetational zones have been recognized within this extensive geographic region (Rauh, 1985): the narrow “tree-desert” zone of Ecuador and northern Peru, dominated by savan- nas and drought-deciduous forests; the ““fog-loma”” desert of Peru and northern Chile, with isolated patches of perennial and ephemeral vegetation maintained by permanent fog on the heights of the coastal range; and the fog-free northern Chilean coastal desert, with a predominantly ephemeral vegetation that develops on lowland areas in re- sponse to irregular rainfall events. The modern climate of the Pacific coastal desert is subjected to the desiccating influence of the Humboldt cold cur- rent, which also produces a negative thermal anom- aly, resulting in temperatures 3-5?C lower than expected for the corresponding latitude, as well as highly uniform temperatures throughout the region (Ochsenius, 1 Arid environments developed very recently in South America (Ochsenius, 1982; Arroyo et al., 1988), in close association with geoclimatic events such as the Andean orogenesis and the onset of the Antarctic glaciation, both of which became more intense during Pleistocene- Holocene times. Tertiary climates were generally warmer and wet- ter than the present (Solbrig, 1976), and tropical forests covered most of the mountainless South American landscape. Climatic changes began in the mid-Eocene with a gradual decrease in tem- perature and an increase in the drying effect of the Andean range. Because of the rapid inception of the arid climate (Simpson, 1975; Ochsenius, 1982), the Pacific coastal desert provides a unique scenario to examine how plant and animal com- munities become modified by sudden environmental 1988). changes (e.g., Arroyo et al., ! We are grateful to G. Arancio, R. Osorio, and F. Squeo of the Herbarium of Universidad de La Serena (015), who played a major role in assembling and identifying the collection of plan anonymous reviewer provided valuable criticism of an earlier version nts on which this study is based. An of the manuscript. Partial funding for this research was provided by a grant from the OAS (1987-1988), and by the National Geographic Society (1989). This is a contribution to the Program of the Unidad de Investigación de Zonas Aridas of the Universidad de La Serena. ? Laboratorio de Sistemática y Ecologia Vegetal, Departamento de Biología, Facultad de Ciencias, Universidad de Chile, Casilla 653, Santiago, Chile ANN. MISSOURI Bor. Ganp. 80: 499-511. 1993. 500 Annals of the Missouri Botanical Garden Although partial floristic accounts are available for certain areas, especially in the ““fog-loma” re- ion of southern Peru and northern Chile (e.g., Johnston, 1929; Ellenberg, 1959; Rundel & Mahu, 1976; Péfaur, 1982; Torres & López-Ocana, 1982; Dillon, 1985; Rundel et al., 1991), complete sys- tematic surveys of the local floral or fauna and their relation to other world biotas are lacking for many areas in the fog-free southern Atacama des- ert. This information can be relevant for refining and enhancing current paleoecological, geological, and climatic models about the Pacific coastal des- ert. In this paper we present the results of a survey of the vascular flora of Lagunillas (3095), located at the southern margin of the fog-free coastal desert of north-central Chile (Fig. 1). Our goals were to assess the importance of the annual and perennial habits in the flora of Lagunillas in relation to other deserts of the world, and to examine the biogeo- graphic relationships of the taxa in order to ad- vance our understanding of the origins and history of the coastal desert flora. STUDY SITE Fieldwork was conducted at viis (Fig. 1), 15 km south of Coquimbo (309 , 71°21'W), Chile, a field station managed М d U niversidad e La Serena. Floral surveys encompassed ap- proximately 25 km? of a gently sloping, extensive coastal terrace, traversed from east to west by an intermittent ravine up to 10 m deep and 10-20 m wide, which remains dry during most of the year. Extensive coastal terraces are a frequent feature of the coastal desert landscape between 27 and 30°S (Börgel, 1983). Altitude ranged from sea level up to 400 m on the summits of the nearby coastal hills. The terrace was covered by modern and fossil dunes, deposited over a calcareous hard- pan (caliche) layer, which extends throughout the area. The calcareous hard-pan is found at variable depths, surfacing in some places, especially on the slopes of the ravine. Further inland, closer to the coastal hills, alluvial deposits become more impor- tant, and a rocky substrate prevails on the sur- rounding hillsides. The regional climate is semiarid with an average annual rainfall of 130 mm (di Castri & Hajek, 1976). Eight years of climatic records for Faro Punta Tortuga in Coquimbo (30°S) (E. Hajek, pers. comm.) indicate that annual precipitation is highly variable, with a maximum of 200 mm and a min- imum of 30 mm. Records for 15 years from La Serena (29%54'S) indicate yearly totals varying be- tween 0 and 177.5 mm (Muñoz, 1985). Rains fall mainly between April and July (winter). Air tem- perature rarely exceeds 22°C or falls below 10°C. For La Serena, the annual average is 14.7°C, and the mean minimum and maximum temperatures аге 11.7°C and 18.4°C, respectively (Fuenzalida, 1965). Due to the accentuated influence of the ocean, air humidity remains over 80% during most of the year. The vegetation is physiognomically dominated by evergreen and summer-dormant shrubs, except for a few months after the rains when herbs become abundant. Encelia tomentosa, Haplopappus cerberoanus, and Flourensia thurifera. Average shrub cover during the dry season varies between 8 and 25% J. Gutiérrez, unpublished data). After the rainy season, plant cover increases up to 90% due to The most common shrub species are ~ the growth of shrubs and the development of a dense cover of annuals and geophytes (Armesto, Vidiella € Gutiérrez, in press). Human impact through dry farming, and cattle and goat grazing has been important in this area as in most of the coastal desert. Before 1962, portions of the study area were cultivated in wet years mainly for wheat and potatoes, and they have been intermittently grazed since then. Nevertheless, the flora is rep- resentative of that of other less-disturbed areas of the coastal desert (unpublished data of the authors). METHODS Collections of plants used in this survey were conducted in Lagunillas during the growing seasons (September- December) of 1987, 1988, and 1989. No previous collections for the area were available. The number of species present in the area varies from year to year depending on the amount of rainfall. In years with low rainfall (20 mm or less), plant cover is composed mainly of shrubs and exotic weed species (Armesto, Vidiella & Gutiérrez, in press). The period of study included two years of abundant rainfall (1987: 168 mm; 1989: 80 mm), and hence we are reasonably certain that all plants potentially growing in the area have been collected. Voucher specimens of all the species collected were stored at the Herbarium of the Universidad de La Serena (ULS). Nomenclature of species follows Marticorena & Quezada (1985), except for the Nolanaceae, which follows Mesa (1981). Plants were Classified by life form (annual herbs, perennial herbs, geophytes, and shrubs) according to field observations and the literature. For each genus, the number of species present in Chile (Marticorena Volume 80, Number 2 1993 Armesto & Vidiella 501 Flora of the Chilean Coastal Desert & Quezada, 1985), and the estimated total number of species known from all the world (Mabberley, 1987) were also compiled. According to the modern geographic distribu- tions of the genera, obtained primarily from Willis (1985) and Mabberley (1987), we assigned the species to the following biogeographic elements: (1) Cosmopolitan: species of genera distributed in sev- eral continents, both in temperate and tropical regions; (2) Pantropical: genera distributed within the tropics of the Old and New World; (3) Tropical American: genera primarily distributed in the low- land tropics of Central and South America; (4) Amphitropical (sensu Raven, 1963): genera occur primarily in warm deserts of North and South America, but are absent from the tropical regions in between; (5) Andean: genera mostly restricted to the tropical and/or temperate Andes of South America; and (6) Endemic: genera confined to the Pacific coastal desert of northern Chile and south- ern Peru. For some genera, assignments were made based on systematic revisions. For example, the genus Calandrinia occurs in desert regions of North and South America and Australia. However, current evidence indicates that the genus spread and di- versified in the Andes of South America (Kelley, 1973). Hence we assigned the species of Calan- drinia to the Andean element. Although the genus Fuchsia has a broad distribution from North to South America, in addition to New Zealand and Tahiti, Fuchsia lycioides is the only member of the section Kierschlegeria that is restricted to the coast of north-central Chile (Berry, 1982) and re- lated to other South American sections. We as- signed this species to the Tropical American ele- ment. The genus 7risterix was assigned to the Andean element based on the work of Kuijt (1988). Finally, four genera (Astephanus, Lastarriaea, Mesembryanthemum, Muehlebechia) had un- known distributions, or their distribution did not fit the definitions given above. RESULTS FLORISTICS A total of 191 vascular plant species, belonging to 61 families, have been collected within an area of about 25 km? in Lagunillas (Table 1). The fam- ilies with the highest number of species in the flora are Compositae, Gramineae, Boraginaceae, Lili- aceae, Papilionaceae, Onagraceae, and Amarylli- Bolivia —20°S Lagunillas Santiago — 40°S Pacific Ocean 300 Km 72°W | FIGURE 1. Location of Lagunillas, the study site in north-central Chile. The stippled area marks the approx- imate extent of the fog-free Pacific coastal desert daceae (Table 2). Compared to the composition of the flora of a high-altitude desert in northern Chile 1982), the coastal desert is char- acterized by a lower diversity of Gramineae and by the presence of bulbous monocots (Liliaceae, Amaryllidaceae). The relevance of Liliaceae and Amaryllidaceae in the coastal desert is also con- (Arroyo et al., trasted with the lower relative frequency of these families in the flora of Chile (Table 2). Onagraceae and Boraginaceae also seem to be better repre- sented in the coastal desert than in the Chilean flora or in the high-Andean desert flora (Table 2). Annals of the 502 | Garden issouri Botanica M a1Áqdoo?) IHd WV 9 I Sesser] '205 (а 19 M) 22019 vissvuD) jenuuy dOWLNVd 00€ 98 овооемејпудолцос qud Pd p1:0]022]77) јешизлад МУЛАМУ OSI 02 әвәзвәв[пуло,{ рш vsoyipuvss pupun) jenuuy NVAGNY OSI L зезововјпј104 ‘ds pupuno) qniys dOWLNVd 001 G aeoovturd[esae") иоея (^y 19 ^H) огојпдир viuidjvsav’) jenuuy VASO) 0€ 22 овошшело) ‘ds snwoig jenuuy WSOD 0€ ZZ овошшело) RIOD зпитаојлод snwmosg qnays NANA I | аваоврш ес 'ssaquie?) xo “1198 n1]of1s12u1 nisagpiagl qnays *WSOO EEZ ey звизодшо) 7] »sopid suapig AS INSOO OST OS звазврџодлон "V 19 "Н D7D42u00]2 s1420128 qniys IHd WY ZI I aeytsoduio’y "Веј SIPIOISOLQUID род quas IHdNWV OSE С? эвиѕоішоз) ‘slag (d 19 Y) Spuy siuny2ong qnays IHd WY OSE c овизод шоу ‘ds suupyoovg qniys WSOD 001 CZ aeaoeipodouay’y qd vpundas xajdiupe jenuuy WSO 0007 OF araoeuorideg so[) snjmum SNAJDIDAIS Y jetuualag (1) og I зезоврепајооу 'euoe(] $пғорлитшә8 snupnudojsp [еїшиәлә{ dOM.ILNVd 00€ Z ILIMVIYIOJNSUY pur ха seSpug sisuojnyo nn[20]ojs14p- гешиалод аочјмуа 00€ c аваовтоојојашу ‘yreyong (4252101М) 1898р 01]20]01s14E qnays IHd WY LI + овооттаоц "BIY PNW (1Y9IYIS) DUDLIJLIQ DIUUDYIA TA jenuuy NVAGNV ZI 11 aRaoRIUOUSIG по ‘q (CT) PIVIPDA viJ&24p qnays NANA 7 © оеаовоовјоЈАЦа4 “boy (4 19 са) 517020701] омешохиј ayAydoasy NANA I I aeaoet[rT Jayeg xo ‘лең njnpidsno pniuna4oup. jenuuy IHd WY 0€ 2, зезовшвелод 191847) (SLOP) nui2A]D2 Dni32uisup јешпалод 500 tC + зеџгодшо) auaaly ("=ѕәгр) sruossiumi2 толдшјр jenuuy IHd WV I 1 aeyisodulor) y 19 y пп snddpdoXqqup aiqdoo?) WY dOH.L 05 66 әвәәррүАлешү zounjq 2D4421$ D142u1204]S]p aiqdoo?) WY dOH.L 0S 66 овооврујалешу quid пи оошо әзАцЧоәс) WV dOML 0€ 6c зезоврујАлешу qud риро оадшзолјујр puud WY dONL 9 1 оваогмејпуаолцос МО (3 7D sipuoipia2ur nosuo]p jenuuy NVIUNV 027 OTI aeaeuor ide g "Y 19 CH 92u21) nnusapp qnays NVIUNV oez OTI aeaoeuoyideg у 19 cH 9]¡4ydosonu тш«арр jenuuy МУЗОМУ 0с OTI aeaoeuoqideq so[) рор vusapp 4945 dOHINVd 00c'I 8 9999 ESOUII]AI TOW (ION) u2ano торојр шлој әу упошаје [210] ayy Á[rure y satoadg a1ydeiZ0a801g satoads вотоо де jo Jaquinyy Јо 1equmw "әәә рәэпроди] = + 'sjuaulg[s јо uontugap IO} 1X9] за "влопов 941 jo поцпа дар əy} uo paseq sjuaure[o отделвоз вола 0j sjuaurugtsse лоцу рие *j1osop [e1seo9 uPo[mq) aui ur sejpungeT jo воле ayi ut рајзајјоо зогзодс `1 318v] 503 Armesto & Vidiella Volume 80, Number 2 1993 Flora of the Chilean Coastal Desert jenuuy xI]INSOO 09 v овоовшелој) попу xo Е! ("D Sopi02D]puti unmpo45] jenuuy xINSOO 09 + овооешвела) иону xo “Wey T CT) tuno mar штрола jenuuy WSOD 00Z LI aeysodulo”) ‘ds uo4232147] qniys WSO?) ОР L овоовлроца 4 "Хор ^V. 70D) xa :ddaog ошрир vapoydy qniys IHd WV 6t I овизодшог) "ше зидозидиро ni]a2u7] [еїшпәлә,] dOM.INVd 009 cv овдово10050ц] "да зод psnfmuoy ролодгоц] јешизлод аон муа 6 1 овоовјплјолиог) ‘ds олриоуцоцј jenuuy WSOD cc CI 9e1oponi) рае CIW 19 “Yosty) nuniZguinuno DIUIDANISA(] puud 500 001 6 овооврепајосу "итцос (aura) un AydAyood шпузирих,) јепшолза INSOO OOI 6 aeo9eperda¡os y "y 19 Н ul nofi 1041904 un. UDUA. i$. jenuuy 500 ср 01 аваовупоепо) '[MPO-19N sisua]n[o DINISNY jenuuy WSOD ОР ОР аваовшаелод “suyof (ҷа) рај уутаошој а рујирзаАлђ) ¡enuuy 500 ОР ОР авооешелоц "Uyg 27042u10]2. рујирјаАл) jenuuy WSOD ОР ОР овооешаелод әцәәу Cd) о удоуцођцор ругирјаАл) jenuuy xINSOO é #1 авашштело) әлот (7T) шпимпш и015ә]5117) puud W30N3 ¿ 29 оваовлјејиј тв njykudoonvp]S DIADISILA) jenuuy WSO) 00€ L овооејпазел“) әцәәу (Ae) ритлојо рјпозрл“) jetuueled xINSOO 08 t aeysodulo”) | niofidouo402 017107) jeluualeg 500 01 € aeaoe[[Audoe-) “ds 00111107) qniys аон Уа 022 I овооешаелод "V 19 "H рарирогр трлођ) [e1uuaJa g WSOD ocz 9 овоовјплјолиог) ‘Slag sisuoqn[2 snjnajoauo;) a14ydoac) W3N4 9 9 әврәәвүцадоәәј, ‘puq (под "qp ојојпира шро рлоујирио [gnuuy IHd WV 9¢ I овооелдеи() SIMO] ә SIM9T (AVN) D]]9U9] D14401) qniys МУЛАМУ CZ 9 aeyisoduno”) под “q suomio vavimbny;) puud IHdNV 0€ CZ IBIIRUOBÁ]O Y qug usury 9YIUDZ110Y/) yenuuy xINSOO OSI Ра авзотродоизцу) I шп) штродоиәу?) jenuuy xINSOO OSI vc aevaortpodouay’) “Т 291p4nui umnipodou217) jenuuy WSO?) OST tc овоотродопоцђ) "] s2pi01504qui. wnipodouey’) jenuuy NVIUNV ТР oe aeyisoduno”) 'sse'] xa "ddooq гмрдиђу DI9YIUDIIDY/) jenuuy NVAGNV 167 oe aeysoduno”) "ssa (под `q) PIDQUA] DIIYIUDIIDY/) qniys WY dOML GLI 1 овооеивјос QUH, J тбара шпајгђ) qniys WSOD Oct 11 aejtsoduro") у дан sisuapnjo ролпојид) jetuueleg IHd NV 9 e овозе||АЦцдоА лег) "лао ји 19 "SN (ULA AA) штит вошра ршдиолтрло) jenuuy IHd WV се I овооелвеи() әцәәу (ABI) DIDJUIP nuossium:) шлој 331] уџошаје [2101 9m) Аше sei adc oiqde180oso01g sarads sardads jo Joquinyy Јо Jaquinyy “panuquo”') `1 119V], c o Y © С) © © c 25 — aW о = 5 20 Ф co c= < = ч © т; jenuuy 4 ¿ 1 аваовповАјод WY SISUIYI DIDILIDISD] jenuuy *WSOD I I овошшвело) yuso (7T) »24np ртуоірштрт апіс IHd WY GI 7 овоовџошелч у 19 7H D2p101s12 DILIUDLY jenuuy WSOD CZ e әкәшшр1с) “ds pisaja0y qnaus МУЛАМУ LS 9с оваовиздлод "PION (QUAY) зоргоиттојог nijjounf jenuuy WSOD OS oe зеизодшо) gjseq pununu siusoysodA y jenuuy NANA с с овлојцјодши | "V 19 CH зарлолугјтод гпалозојошоц әзАчцЧоәс) WV dOML 9; ez ərə vep euy 1омед (ddaog) asuo;jvad wnaijsvoddi]I од ово) WV dOML 91 gz аваоврцјАлешу Jayeg (QH) прјоигод umnaisvoddig] jenuuy *WSOD 02 1 аезов || Ацо ле“) Эа »242u12 pumua H qniys WSO) osz 95 әвәәрш82108] "V 19 CH штјацаоиојг штаолјоцон jenuuy IHd WV OF L зеизодшог) "лав (ҷа) 2072u2144n штигјон jenuuy IHd WV OF L aeyisoduno”) Áapeg (00H) шпопошоло unnajag] апіс IHdWY¥ 091 19 оеизодшо) auoray (&uray) smupo42q422 гпадоаојаоц qniys IHdWV 091 19 оеизодшог) Аел) “Y (90) snijofiaaod. гпадоаојаоц qniys IHd WV 091 19 aeyisoduio’) qug sa 9104 snddvdojdvy qniys NVJANV 9 с дела јод у sop wnisngos иојАуаоишА.) qniys IHd NV 61 9 овизодшог) MXPIA CV 19 ^H) 05001591 01224121107) jetuuasag WSOD OSI 1# aeysoduio’) Чор panna шпуруарио [etuueled WSOD OSI [# зеуизодшог) (qug un] 4ydo19194 штјуоуарио) гешиолод WSOD OSZ 9 оваовпоцло А SBAOT) 19 мовицос (uoq '() VesNYydyns n14p]npuv]^) aAydoay NAUNA tt Ф овзовуг] риту »2ununas падао) jenuuy IHd WY CZ t звзовшошајод "qasue) sisuaiaip]pa pipi) jenuuy „4500 OOF ZG ILIMIY "Т 2u1dp што puud „500 се Р аваовмешт Y ‘ds pupwny qnaug WV 4041 001 c аваовлвец) "upuy зортолок) DISYIN] [etuussed WSOD CZ Ф IPIIRIUINUBA Y ‘ds niuoqunaJ qnaug WSO?) 0€ c оеаов || АцдовА7 ү 19 "Н sisu2]n[o. Diuoz n4 jenuuy *WSOD 009'1 gz aevaoriqioydny “1 snjdad viquoydng jenuuy WSO 009'1 97 aeaoriqioydny qug sisuainbja тдлоцапа qnius WSOD ge z овизодшог) RIOD тлајру штаојрапа qniys WSOD 9e 2, овизодшо) "ава" wn Aydouoyoaps штаорапа jenuuy WSOD 0£c 91 дело једш q иеді Хә а ununpquimboo umiguxag jenuuy *WSOD 09 Ф әрәовїирләс) попу хә мен T CT) штоуохгош штрола шлој әт упошаја [210] arty) Аше 4 sardadg a1yde 13093018 satvads satoads jo Jaquinyy о Jaquinyy 'penunuo) "Т x18V[ 505 Armesto & Vidiella Volume 80, Number 2 1993 Flora of the Chilean Coastal Desert jenuuy WSO?) 00€ CII оваоврцех() TABS хә "мед рујирлопи S1]DX() a14ydoac) 500 00€ CII оваоерцех() иле оштарош вох) ques INSOO 00€ STI ista se bio Y "ug рәуир818 SDX() апче IHd WV 9e t деизод шоу ua Aa sivun snaodsoKaud() jenuuy xIHdlAV 08 LI o?ooP1ZPu() AUT] хә ‘әрә 01215 D421]0u9() jenuuy IHd WV 08 LI 9voov1deu() Are) sisuaquinboo D42110u9() jenuuy IHd WV 08 LI овоовлавец() ‘ds рлгузоио() jetuualag W30N3 81 eI овоовивјом "да вод vijofipas vunjoy jenuuy W30N3 91 eI аваовивјом pur] охоролра punJoN TS NAIUNA 81 el эви ом asuqof (V 19 ^H) 910/14 vunjoy qniys W30N3 81 eI ƏLIDLULJON "да зод v1ofiynssps9 DUDJON [peruua1oq МУЛАМУ é OI a?outure1«4) asa (гапу до uu) p2ojfignd. орәѕѕәду qniys WV dOML I I IBIDPINÁJA] орле) ја шпырциет (ueg) sisuaquimboo SIYJUDIDAL JA] puud NVAGNV 6€ ес aejrsoduro^) ‘ds misun qniys (€) Gl I aeaoPuosA[oq "suqof (WIG “| `f) DIDINISDY 0i1322qu2]y2n]y jenuuy W30N33 @ c a?irsoduro^) га 19 Y bpyiuuid ромруззојј ¡enuuy «(2) L 1 овоовотту vp umuipisid2. штиоујирАлдшгавгуј jenuuy W30N3 ¿TZ IZ ILII PONIN ‘ds рәјпаиоиәуү [ишү xINSOO 0с Y avaovuomideq ПУ CT) vorpui sniopijopy јешпалод WSOD ï FI овошшело) ‘ds Dona jenuuy xINSOO 9с 8 овооепош вд `1 рудлошлјоа ogpoipapy jenuuy xINSOOD ere 8 авзовпош два uospng (71) оодоо 0ZnD21po]q јешизлод «500 0€ 1 seqe] 7] 2402]na штаті jenuuy *WSOD 0€ € оваовлјејиј "] »40jftiaand. vapp jenuuy *WSOD og € әрәәелүерү “1 SISUIDIMU DAJDYA] jenuuy NANA It 81 a?aoviq1aqusoa[? ]N "да вод sijmuny v1q194S9]D W qniys dOWLNVd coe 8 әрәэр[пирЧшрг) у 19 cH DA y dx od D1]2q07] jenuuy NVAGNYV col Ch овоор5во] '[M?£-19M 200211) 0007 qniys IW3N4 с 1 оваоврш ес под 'q CV 19 ^H) »sojynpur]s. ооипгој']ј qniys WY 4081, € I овоотрлеовџу "V 19 CH (пој) 92nisnpo 09191] ә1АцЧоәс) NAIUANA ZI Il аваовцг] qud su22s22njoia 2u&40202n12] a14ydooc) W30N3 ZI Il овоовцг] Keg noundand зиХлозопој a1Áudoo-) WIJANA ZI 11 овооецг] qug “4 ssuaquanboo 2uX40202n2] jenuuy NVAGNV OP ср овизодшог) ‘ds vi19y9n97] шлој әу juouro[a [210] ary) rure y saradg oryde1309301g satoads satoads jo Jaquinyy јо Jaquinyy "рәпициог) "ү эту, Annals of the 506 Missouri Botanical Garden Y S02 « ОО? I £9 T WV 40841. 001 el ванае agotay хә YaseuAy штадујирлојоц umun]og кыга WY JOYL 001 мы кнн, ‘ddaog шпәрюалэѕ wimryoursdsic 1 koc. WY 4041 001 el sili "ваза ^W хә ‘Aap “9 шләәипГ unnpounitsig pes dOMH.LNVd CZ I жаш, ‘ddaog wnjnpidsna штщәиыХз bus WSOD OFZ 01 ES aia А suo snofnuoX4q SOLAS Зе 5 500 005'1 227 nodu) Адзилед ја M4] (Y 19 ^H) nunun рииәс̧ ud WSO) 00€'I GGG oesoduror) "за xe мәң smpofisyny 0199498 ae d 509 0081 ZZZ зеџводшог) Ашом SNUDOLIGIII O1DIUIC Pad WSOO 00€'I ZZZ paure “Y 19 "y snupiuounig 0199498 И V ШЕ ГКК! 9 9 Mon «ен. "V 19 cH Seplolyng отәәиәс̧ pes NANA €I > y J тилед шпирАр8 иојрјдаог 12 jenuuy el овооеивјос | 125 xINSOD S 7 qd +724011] snyjuvz1yo¢ qnaq é овошшвло) 4 1425 a14yd A WV JOHL L 6 аваовтрлвови SSON EU SAIS за пон WV 4081 L 9c ee: "аде (САРТ) smupzAjod snun[os pnuuy МУЛОАМУ Sc > | ү zung (әң) saproppooAyd юую1у4оро1 48146 6 овоовјејиес | [р1уаороц y плус NV4dN Y 891 L оваецашолу Бе ме а а ен. iis NVAUNV 891 L ovaoeqauroag qug pjsnuoa ning к NVadNV 891 L әвәэецәшо2 "OW 709749 Ding bann NVAGONV € R ito тор ритиолојлод DANY pis NVHGONV 9 | звана а под | »ofiouno рухпола binos аомлмуа 01 I aeaoeursequin[q nsuqof SESH “мотоа bouis PEE l 1 зеизодшођ JS CH торе» оеша iue WSOD OSG zi потат uo(] “q sn;n]oaaa smdap202]d jenuuy WSO 08 GG o?eoo?urdejue|d | ‘ds ойи] enuy INSOO OSS ZÉ oro Pude ue] [ug 224011] o2v1uv]d MR IHdWV 001 11 оваоешве од 9 19 y ујпр:а у обршитја | V xdOH.LNVd CT > р ";suqof Cyd) 5701302 s&auy10q0120 yenuu € 9P99PUOqIO9 A ii [1000197] V IHdINV 01 auaaly (7T) n4ojipou vÁ jenuuy b əLəsvLuBeIog HIF 1а IHd WY 01 Е | эа Са зә ^W) смозгиу рАл020199 әзАцЧоәс) у аваовшаелод 1 1994 ls WAAN I I ынк "jsuqof ('1sugof) гудлошлр 7440201224 шы INSOO 0€ с piu. ио а Cd 19 ^M) v3jn4202 раупура Su ИУ JOHL ср + asin гоа sisuaqn[o DIYIAUOIDG xINSOO 00€ ст] soon Кою) md snydoqAx() шлој әт уошаја [e101 — A boef pasos sippx() отцделдоо 801] зогоод5 Sods К | 4 вето 46 jo Joquinyy јо iequinN 'penunuo) "Т яну, 507 "PIensny pue 'ривјео7 мом 'воџешу ymos (£) "e[n pue ‘вету ‘ецецѕпу *spue[s Á1eue7) 'eoupy YINOS 'ешлојцео) *ueouello]po]N (2) "PIpu[ pue свомшју 'вомешу YNOG pue YON] (1) Flora of the Chilean Coastal Desert Armesto & Vidiella jenuuy WSOD oz © овгошшело) “ds тап 4 jenuuy 0800 00€ ZL оеаовјогА ‘ddaog vypisnd рот д puud WSOD 0Sc ср овоовпеџоје A у 19 cH пзодрмд DUDILI]D 4 jenuuy «#5600 сф 11 ILIMLI “Т SUSAN DINA) puud NVAGNW ¿ Ф әвәәвретдәјәѕұ әшүерү (aura) 040jf143Juo02 nipoam[ [еїшиәлә{ WV dOML 98 9I aveoor[ooedoa |, 199M8 40]0214) umjoandoa[ оџзгеле тон NVIUNV 11 € IPIIRYIUBIOT “sual у 19 Moeg хә "uSar[] UBA COA хә SIN) sa 4ydo хәр р ¿[enuuy WIJANA pI #1 овизод шог) ds мота, ә1АцЧоәс) NANA © z PIP] “IRA (са 19 ^W) wnssownjd unpjodou2ii[ [etuuasag WSOD 09-04 01 әвәәэволу ‘ds ртшоёрлјЈә р әзАцЧоәс) IWA3GN3 @ 7, әвәэвүцаоэә ү, вот) xo “1198 плоџтола рорјтаозој [et1uuaag 500 OSI bp овопшвло) шд], »sounjd ра) jenuuy «600 021 8 овоов || А доле) "PILA sippidsno DIAD)IIIS [e1uuaJag WSOD 00€ 11 aejeiqe'] "ujuag хә ozunw DIDIUNA] SAYIDIC peda WSOD 00€ 11 SEES] YA 510407] SAYIDIS [etuua1ag IHd WV 09 € IBIIRAJE]A под 'а (чоон) »q0]15niqo рдојолорцас qas WV JOHL SG € 1018] ‘bug (рит) оплајрз ajoonudg jenuuy *WSOD 79 € овизодшо у ЇН CTD) 22950 snyouog qnauc WSOD OOF t9 оваовивјос ‘легу штриила шпирјос jenuuy xINSOO OOF T #9 овооеивјос np шпати шпирјос шлој әу упашајо [2101 arty) Ayture ] satvadg a1ude1309301g вотоо 5 вогоо 5 јо roquinyy jo 19qUINN] Volume 80, Number 2 1993 гропициој "Т 119V 508 Annals of the Missouri Botanical Garden TABLE 2. Most important plant families in the flora of Lagunillas in the Chilean coastal desert and their relative representation in the flora of the high Andean desert (above 2,000 m) at 19% in northern Chile, and in the flora of continental Chile. Lagunillas High Andes* Flora of Chile** Family NT % МТ % Compositae 34 17.8 91 32.6 1,033 18.0 Gramineae 10 9:2 34 12.2 595 10.4 Boraginaceae 9 4.7 3 1.1 116 2.2 Liliaceae 8 4.2 0 0 50 0.9 Papilionaceae 7 3.7 12 4.3 321 5.6 Jnagraceae 6 3.1 1 0.4 44 0.8 Amaryllidaceae 6 3.1 0 92 1.6 p Total number of spe cies. Source: Arroyo et al. (1982). ** Source: Marticorena (1990). LIFE-FORMS Annual plants represent the highest proportion of the total flora (40.8%, Table 3), followed by shrub species (28.8%). Bulbous perennials repre- sent 10.5% of the species. All the species that are seen aboveground only after the winter rains, i.e., annuals, geophytes, and rhizomatous perennial herbs, comprise 70.7% of the total flora. NATIVE VERSUS INTRODUCED SPECIES Fifteen percent of the species collected are in- troduced (Table 3), most of them annuals. If these exotic species are excluded from the analysis, an- nuals decrease their representation to 33.1% (Ta- ble 3). This results from the fact that one-third of the annual species are introduced. Thus, annual and shrub species represent equivalent proportions of the native flora of Lagunillas (Table 3). There were no exotic species among geophytes, and only one shrub species was introduced. BIOGEOGRAPHIC ELEMENTS Cosmopolitan genera represent a high propor- tion (41.4%) of the total flora (Table 4), being especially common among annual (55.1% of all annuals) and perennial herbs (54.1% of perennial herbs). Some of these species belong to weedy families; probably most have life-history traits that enable them to invade recently open habitats. In contrast, only 27.3% of the shrub species belong to Cosmopolitan genera. Species that belong to disjunct genera in North and South American de- serts represent 15.2% of the flora (Table 4). This disjunct element includes 21.8% of the shrubs and 16.7% of the annual species. Endemic species amount to 13.1% of the total flora. Of these en- demic taxa, 18 (72.0% of all endemics and 9.4% of the total flora) are restricted to the Chilean desert. The highest proportion of endemics is found among the geophytes and the lowest among the perennial herbs. Other important biogeographic el- ements in the flora of Lagunillas are the Andean and Tropical American (Table 4). The Andean element is more frequent among the annuals (11.675) and shrub species (16.4%). Tropical- American elements are more represented among the geophytes (45%) than among species of any other life form. is of total, native, and introduced species by life form in the flora of Lagunillas in the Pacific LE 3. coastal desert of Chile. N — number of specie All species Native species Exotic species Life form N %* N %* N %* Annuals 78 40.8 54 33.1 24 85.7 Geophytes 20 10.5 20 12.3 = Perennial herbs 37 19.4 34 20.8 3 10.7 Shrubs 55 28.8 54 33:1 | 3.6 Hemiparasite l 0.5 1 0.6 Total 191 100% 163 85% * 28 15%** * Percentages with respect to the totals below each column. ** Percentages with respect to the total number of species. Volume 80, Number 2 Armesto & Vidiella Flora of the Chilean Coastal Desert 509 TABLE 4. N = number of species. Biogeographic elements by life form in the flora of Lagunillas. See text for definition of elements. Perennial Biog aphic Annuals Geophytes herbs Shrubs Total iogeographic element N % N % N % N % N y Cosmopolitan 43 55.1 | 5.0 20 54,1 15 27.3 79 41.4 Amphitropical 13 16.7 1 5.0 3 8.1 12 21.8 29 15:2 Endemic 8 10.3 9 45.0 2 5.4 6 10.9 25 13.1 Andean 9 11.6 - — 4 10.8 9 16.4 23 12.0 Tropical American — — 9 45.0 3 8.1 7 12.7 19 9.9 Pantropical 3 3.8 — 4 10.8 5 9.1 12 6.3 Unclassified 2 2.5 — 1 2.7 1 1.8 4 2.1 Total 78 100.0 20 100.0 37 100.0 55 100.0 191 100.0 DISCUSSION Most species in the flora of Lagunillas are an- nuals (Table 3), a large proportion of which (24 out of 78, or 30.8%) are non-native weeds. These weedy species are native to Eurasia, and their presence and abundance in the study area may reflect the recent history of farming and livestock raising in north-central Chile (Ваћге, 1979). All of them are also found further south in the mediter- ranean region of Chile (Navas, 1979), particularly in abandoned pastures or agricultural land. It seems likely that many weed species that came with the Europeans colonized the coastal desert via central Chile. Considering native species only, the percentage of annuals in the flora of Lagunillas in the Chilean coastal desert is higher than in high-altitude Chilean and North American deserts. Annuals represent between 10 and 29% of the flora above 2,000 m in the northern Chilean Andes (10-146 species), and 1995 of the flora at 1,200 m in the Great Basin, U.S.A. (322 species), where plants are sub- jected to low temperatures in addition to the con- straints in water availability (Arroyo et al., 1988). In contrast, the proportion of native annual species (33%) in the Pacific coastal desert site is lower than in other lowland deserts with less equitable temperature regimes. Annual species may repre- sent 40-70% of the flora in Middle-Eastern deserts (Orshan, 1986), and 52-55% of the flora in the Sonoran and Colorado deserts of California (Rich- erson & Lum, 1980). Although the number of species of geophytes is often lower than the species richness of other plant life forms, they represent a high proportion of the flora in the coastal desert. The relative importance of Liliaceae and Amaryllidaceae, the major bulbous taxa, in the flora of the coastal desert is higher than in high-altitude deserts in northern Chile (Ar- royo et al., 1982) or in the flora of continental Chile (Table 2). Geophytes are also important com- ponents of the flora in the deserts of Israel, Af- ghanistan, and Iran (Breckle, 1983). In contrast, geophytes are less important in North American deserts (Shreve, 1942). The representation of the various biogeographic elements in the flora of Lagunillas should provide insights into the history of the Pacific coastal desert vegetation. Based on the geological evidence about the origin of this desert (Ochsenius, 1982; Arroyo et al., 1988), it is reasonable to assume that the oldest elements of its flora originated from a tropical stock of woody taxa during the Eocene. Probable relicts from that period are arid-adapted Pantrop- ical elements such as Prosopis, Acacia, Caesal- pinea, and Cordia (Table 1), which are now wide- spread in dry subtropical environments (Sarmiento, 1972). The past and modern geographic ranges of these woody genera, which include arid to moist tropical and subtropical regions of America and Africa, suggest that they might have originated before the breakup of Gondwana was completed (Solbrig, 1976) in the early Tertiary. The low rep- resentation of Pantropical elements today (6.3%, Table 4) indicates, however, that many old tropical elements became extinct. It is likely that most of these Tertiary tropical elements disappeared due to the rapid and strong aridization that took place during the late Pleistocene-early Holocene (Och- senius, 1982; Simpson, 1975). Consequently, most of the modern desert flora originated more recently, in direct relation to the uplifting of the Andes and the increasing aridity of the Pacific coast of South America during the Quaternary. We propose that some of the modern desert genera (e.g., 4desmia, Chaethantera, Loasa, Mu- tisia, Porlieria, and Puya) diversified in the An- dean highlands and invaded the lowland desert subsequently. All of these South American genera ave a large number of species (Table 1), which are mostly restricted to the Andes. 510 Annals of the Missouri Botanical Garden Pacific coastal desert endemics represent the third most important element in the flora (Table wo subgroups can be recognized among the desert endemics. Some endemic genera appear to have diversified very recently in the coastal desert area (Mesa, 1981), exhibiting a high number of species per genus (e.g., Cristaria, Malesherbia, Nolana, Table 1), probably extending their ranges into the Andean environment (F. Squeo, unpub- lished manuscript on the genus Malesherbia). On the other hand, other low-diversity genera, which are often monospecific and have a geographic dis- tribution restricted to one or a few localities in the Pacific coastal desert (e.g., Ancrumia, Bridgesia, Llagunoa, Pasithaea, Phleocarpus, Reichea) may represent remnants of a pre-aridization subtropical flora. In the latter group we could also include some Tropical American genera (e.g., Alonsoa, Cestrum, Colliguaja, Oxybaphus, Sphacele) that have a lower specific diversity in Chile (Table 1) than in the rest of South America. A considerable proportion of the taxa belong to an element with a disjunct distribution in the deserts of South and North America. Because these species are presently restricted to arid lands, it is hard to explain how they may have reached their present distribution without assuming that arid habitats ex- tended through the lowland tropics during the last glacial period or earlier (e.g., Simpson & Neff, 1985). Raven (1963) argued that most of these disjunct species could have migrated from South to North America across the Tropics, taking ad- vantage of hypothetical subhumid sites in post- Miocene times. This hypothesis assumes that these species are capable of long-distance dispersal (Ra- ven, 1963); however, this dispersal ability is seldom found among xerophytic species (Ellner & Schmi- da, 1981). For a majority of the disjunct taxa in Lagunillas (e.g., Cardionema, Camassia, Encelia, Table 1), it is hard to argue in favor of the long- distance dispersal hypothesis. The present distri- bution of these disjunct elements provides support for the hypothesis that arid environments extended through the area of the lowland Tropics during the last glacial period (Simpson & Neff, 1985). For some of the annual species (e.g., Clarkia, Gilia, Camissonia), dispersal associated with humans in modern times seems a more likely explanation. An important component of the flora of the study site is the Cosmopolitan element. А major propor- tion of these Cosmopolitan taxa are annual and perennial herbs (Table 4). Because of their wide- spread distribution, it is likely that many of them could have been introduced by humans in recent times (e.g., Erodium, Galium, Schismus). Because deserts are relatively open environ- ments of recent origin (Arroyo et al., 1988), we expect that short-lived, opportunistic species, with high reproductive capacity, may contribute an im- portant proportion of their flora. LITERATURE CITED ARMESTO, J. J., P. E. ViDIELLA & J. R. GUTIÉRREZ. 1993. Plant communities of the fog-free coastal desert of Chile: Plant strategies in a fluctuating environment. Rev. Chilean Hist. Nat. 66: (in press). ARROYO, M. T. K., К. A. SQUEo, J. J. ARMESTO & C. VILLAGRAN. 1988. Effects of aridity on the plant diversity in the northern Chilean Andes: results of a ~ natural experiment. Ann. Missouri Bot. Gard. 75: 55-78. VILLAGRAN, C. MARTICORENA & J. J. AR- MESTO. 1982. Flora y relaciones biogeográficas en los Andes del norte de Chile. Pp. 71-92 in A. Veloso & E. Bustos (editors), El Ambiente Natural y las Poblaciones Humanas de los Andes del Norte Grande de Chile (Arica, lat. 18?28'S), Vol. 1. ROSTLAC, Montevideo, Uruguay. BAHRE, C. J. 1979. Destruction of the natural vege- tation of north-central Chile; Univ. California Pub- lications in Geography 23 BERRY, P. E. The systematics and evolution of Fuchsia section Fuchsia (Onagraceae). Ann. Mis- ri Bot. Gard. 69: 1-198. Bóncr EL, UR 1 Geografía de Chile, Vol. II. Geo- ram Instituto Geográfico Militar, Santiago, hile C BRECKLE, 5. W. 1983. Temperate deserts and semi- deserts of iue and Iran. Pp. 271-319 in N. West (editor), Ecosystems of the World 5. на Deserts and Semi-Deserts. Elsevier, Am- Gen i 1 & E. НајЕК. 1976. Bioclimatología de Chile. Ediciones de la Universidad Católica de Chile, DILLON, The silver lining of a very dark аш, MAP, studies in coastal Peru during the iue 3 El Nino events. Field Mus. Nat. Hist. Bull. : 6- о ues H. Über der wasserhaushalt tro- r nebeloasen in der küstenwüste Peru. Ber Deutsch. Bot. Ges. 73: 449-462 ELLNER, S. & A. SCHMIDA. 1981. Why are adaptations for long-range PE so rare in desert plants? Oecologia 51: | | Н. 1965. Orografia. In: Geografía Есопо- a de Chile. CORFO, Editorial Universitaria, San- o, Chile. ine I.M. 1929. I.P the flora of northern hile. Contr. Gray He rb. -1 KELLEY, У. A. 1973. Pollen paite and relation- ships in the genus Calandrinia HBK (Portulaca- ceae) M.S. Thesis. California State University, Northridge. Копт, J. 1988. Revision of Tristerix (Loranthaceae). Syst . Bot. Monogr. 19: ManBERLEY, D. J. 1987. The Plant- Book. Cambridge Univ. Press, Great Britain. MARTICORENA, С. 1990. Contribución a la estadística Volume 80, Number 2 1993 Armesto & Vidiella Flora of the Chilean Coastal Desert de la flora vascular de Chile. Gayana, Bot. 47: 85- 11 e M. Quezada. 1985. кыр Ba la flora hd de Chile. Gayana, Bot. Mesa, A. 1981. | M des poration Flora Neotropica 26: 1-197. Muñoz, M. 19 pride del Norte Chico. Dirección de Bibliotecas, Archivos y Museos. Coa Munici- palidad de La na, La Pu ed Navas, L. E. Flora de la MER Santiago de Chile, Vol. I, II and III. Pide de la Universidad de Chile, Editorial Universitaria, Santiago, Chile. Oc HSENIUS, C. 1982 tacama: t P South American Quaternary. Striae 17: 112-131. . The deserts of the Middle East. Pp. -26 in M. Evenari, I. Noy-Meir У. Goodall (editors), Ecosystems of the World 12B. чеч Deserts and Arid Shrublands. Elsevier, Amster nm J. E. 1982. Dynamics of iip communities the Lomas of southern Peru. Vegetatio 49: 163- Raun, W. 1985. The Peruvian-Chilean “> Рр. 239-267 in M. Evenari, I. Noy-Mei D. W. Goodall (editors), Ecosystems of the Wo rid 12A. Hot Deserts and Arid Shrublands. Elsevier, Amsterdam. Raven, P. H. 1963. Amphitropical relationships in the floras of North and South America. Quart. Rev. Biol. RicHERSON, P. J. & K. Lum. 1980. Patterns of plant species diversity in California: relation to weather and topography. Amer. Naturalist 116: 504-536 RUNDEL, P. W., M. O. DILLON, ы РАМА, Н. A. MooNEY, & E: GULMON & J. R. HLERINGER. 1991. The phytogeography and белөн of the coastal Atacama eruvian е Aliso 13: 1-49. Manu. 1976. Community structure and diversity in 2 conta fog desert in northern Chile. Flora 165: У SARMIENTO, G. PE. Ecological and floristic conver- gences between seasonal plant ше» of tropical and subtropical South America. J. Ecol. 60: 367- 410 SHREVE, F. 1942. The desert vegetation of North Amer- ica. Bot. Rev. (Lancaster) 8: -246 uM B. B. Pleistocene changes in the flora f the high tropical Andes. Paleobiology 1: 27 294. & J. L. Nerr. 1985. Plants, their pollinating pps and the great American interchange. Pp. 427- in F. G. Stehli & S. D. Webb (editors), The cue American Biotic Interchange. Plenum Pub- The origin and floristic affinities e South dosis n desert and semi- ptos regions. Pp. 7-49 in D. W. Goodwall (editor), Evolution of Dat Sk Univ. Texas Press, Austin. C. Lopez-OcaNa. 1982. Estudio bio- A Dictionary of the Flowering Plants and Ferns, student edition. Cambridge Univ. Press, Cambridge, Great Britain. EVOLUTION OF (ARUNDINEAE: GRAMINEAE) Н. E. Connor! and M. I. Dawson? ABSTRACT mprothyrsus, a ditypic arundinoid, South American grass genus, consists predominantly of populations of exclusively female plants in which seeds are set b some aposporous species of the related genus Cortaderia, and in some diplosporous species o Poa and и она It is another example of the association between apomixis and dioecism outlined by Gustafsson 45 years a Lamprothyrsus is an arundinoid genus erected by Pilger (1906) to accommodate South American grasses with long, erect, hyaline, lateral lemma lobes and even longer, stiffer, recurved to reflexed awns; two species are now recognized: L. peruvi- anus Hitche., primarily from Ecuador and Peru, and L. hieronymi (O. Kuntze) Pilger from Bolivia and northern Argentina. The genus is unique within the Gramineae in that it is the only one where almost all plants are female, and reproduction 15 solely from seeds set autonomously apomictically (Connor, 1979; Parodi, 1949). Because its affinity lies clearly with austral Cortaderia (Conert, 1961, 1987; Clayton & Renvoize, 1986), Lamprothyr- sus might be suggested as an end of one evolu- tionary line of the cortaderioid facies of the tribe Arundineae. Its chromosome numbers are, there- fore, of particular interest. Here we report a highly polyploid somatic number for L. peruvianus, and discuss its significance in the Arundineae and in the evolution of the reproductive biology of the genus. CHROMOSOME NUMBER AND REPRODUCTION THE PLANTS One of us (H. Connor) collected seeds of tall and tussock-forming L. peruvianus between Tar- ma and San Ramon, Departmento Junin, at 2,300 m in the Peruvian Andes in August 1975; 100 plants were grown in the gardens of Botany Di- vision, DSIR, Lincoln, New Zealand. All plants were monomorphic females—male-sterile—and of the floral form illustrated for Lamprothyrsus by Bernadello (1979) and Nicora & Rugolo de Agra- sar (1987), and have remained unchanged for 17 years in cultivation. All plants set seed abundantly by autonomous apospory where the embryo sac develops from the nucellus (Connor, 1979, 1987; terminology after Nogler, 1984). These seeds ger- minate freely CHROMOSOME NUMBER The count obtained for L. peruvianus is 2n — ca. 136 (Fig. 1); voucher CHR 449637. The chro- mosomes are small (0.7-2.4 иш). This is the highest reported number for the Arundineae, though it is very close to 2n = ca. 132 reported for Phragmites australis. High levels of polyploidy are already recognized in the tribe; among the highest numbers are: l. 2n = ca. 132 : Phragmites australis (Cav.) = ]llx, P Steudel, cited as P. communis Trin. in papers prior to Clayton (1968); Gorenflot, 1986) 2. 2n = 120 : Rytidosperma indutum (Vick- = 10x, ery) Connor & Edgar (Brock & Brown, 1961, as Dantho- nia) 3. 2n = 108 : Arundo donax L. (Christopher — (18x) & Abraham, 1971 = 12x : Cortaderia jubata (Lem.) Stapf (Connor & Edgar, 1974) ' Department of Geography, University of Canterbury, Christchurch, New Zealand. Ze ? Landcare Research New ANN. MISSOURI Bor. Garp. 80: aland, Lincoln, Canterbury, New Zealan 512-517. 1993. Volume 80, Number 2 1993 Connor & Dawson 513 Evolution of Reproduction in Lamprothyrsus FIGURE 1. 10x : Monachather : C. rudiuscula Stapf, in plants raised from seeds collected (by H. Connor) at the /ocus clas- sicus, Santa Rosa de Los An- des, Chile; voucher CHR 468718. This is the first report for this taxon. : Rytidosperma spp. (Brock & Brown, 1961) : Phragmites australis (Lenoir et al., 1975; Gorenflot et al., 1984) : Cortaderia spp. (Hair & Beu- zenberg, 1966) : Rytidosperma spp. (Brock & ) Brown, 1961 paradoxus Steudel (Brock & Brown, 1961, as Danthonia bipartita) : Phragmites australis (Goren- flot et al., 1984; Lenoir et al., 1975; Pohl & Davidse, 1971) : Arundo donax (Christopher & Abraham, 1971) : A. plinii Turra (Fernandes & Queiros, 1969) : Cortaderia spp. (Connor & Edgar, 1974) Metaphase of mitosis in L. peruvianus, X 3,000; Zn = ca. 136. Тће basic number for Arundo species is uncer- tain; the levels of ploidy shown in parentheses are based on the assumption that x — 6. The chromosome number sometimes varies among populations. Of the taxa listed above, those with variable chromosome numbers are: Arundo donax 2n = са. 60, 72, 108, 110 (Chris- topher & Abraham, 1971; Fernandes & Queri- os, 1969; Larsen, 1963) Phragmites australis 2n — 36, 42, 44, 46, 48, 49, 50, 51, 52, 54, 72, 84, 96, 132 (Gorenflot et al., 1984; Gorenflot, 1986) Rytidosperma indutum 2n = 72, 96, 120 (Brock & Brown, 1961) R. caespitosum (Gaudich.) Connor & Edgar 2n — 24, 48, 72 (Brock & Brown, 1961) R. laeve (Vickery) Connor & Edgar 2n — 24, 48, 72 (Brock & Brown, 1961) R. linkii (Kunth) Connor € Edgar Zn = 24, 48, 72 (Brock & Brown, 1961) Even allowing that the chromosome count for L. peruvianus is approximate, it does not fit any of the usual arundinoid range of basic numbers indicated above, viz x = 6, x = 9, x, = 12, nor does it fit with а cytologically isolated genus such as New World Gynerium, x = 11 (Pohl & Davidse, 1971), nor with austral, Old World Chionochloa, x = 7 (Dawson, 1989). 514 Annals of the Missouri Botanical Garden REPRODUCTION IN HIGH POLYPLOID ARUNDINOIDS Reproduction is sexual in high polyploid arun- dinoids; the exception is gametophytic apomixis in female plants in the two species of Lamprothyrsus, and in at least six species of Cortaderia (Berna- dello, 1979; Connor, 1965, 1973, 1979, 1983a, , 1987; Costas-Lippmann, 1979; Parodi, 1949; Pohl & Davidse, 1971; Philipson, 1978). Gender diversification is an uncommon evolutionary step in the tribe as a whole and, Lamprothyrsus and Cortaderia apart, is seen only in dioecism in mono- typic Gynerium, in gynodioecism in one species of Chionochloa, and in andromonoecism in Alloeo- chaete and Phragmites (Connor, Asexuality in females evolved in Lamprothyrsus and Cortaderia in the identical form of autonomous (nonpseudogamous) apospory after dioecism, sub- dioecism, or gynodioecism had become fixed as 1987). Further, megagametogenesis is similar in both; the embryo gender dimorphisms (Connor, sac, nucellar in origin, is commonly 6-nucleate, since one synergid and one antipodal are absent (Philipson, 1977, 1978; M. N. Philipson in Con- nor, 1987). Philipson & Connor (1984) examined the megagametophyte of eight genera of the Arun- dineae for haustorial synergids; they were present in Lamprothyrsus, Cortaderia, Danthonia, Sie- glingia, Купаозретта, and Pyrrhanthera, but absent from Arundo and Phragmites. There is no correlation between apomixis and haustorial syn- ergids, because sexual reproduction is recorded in all genera except Lamprothyrsus. Sexual reproduction in andromonoecious Phragmites australis is not very successful, as estimated from seed setting (Cartier & Lenoir, 1980; Gustafsson & Simak, 1963; Luther, 1950), even though megagametogenesis is normal (Cartier & Lenoir, 1980) and pollen fertility is high (Lenoir et al., 1975; Stoian et al., of single, self-incompatible genotypes may be one 1974). Large colonies cause, but the instability of chromosome number reported by Gorenflot et al. (1984) and Lenoir et al. (1975) could contribute in small part. Richards (1986) suggested that in northern English and Scottish populations *' mer temperatures are too low for seeds to ripen adequately." There is no suggestion of apomixis in Phragmites. Arundo donax is less variable in chromosome complement than P. australis. All flowers are per- fect. The only information we have on reproductive biology indicates that many megagametophytes fail to mature (Gustafsson, 1947; Philipson & Connor, 1984 Sexual reproduction is usually cleistogamous in Monachather, and either chasmogamous or cleis- togamous in A ytidosperma (Brock & Brown, 1961; Cashmore, 1932). Monachather is an extremely specialized monotypic genus of eremaean Australia; R ytidosperma is especially well developed in Aus- tralasia (Connor & Edgar, 1986). LAMPROTHYRSUS MALE-FERTILITY Although there are substantial reports that Lam- prothyrsus sets seeds exclusively on female plants in the absence of pollen, Bernadello (1979) re- corded two specimens of L. hieronymi with polli- niferous anthers: one from northern Argentina (Fa- bris 3191, 1962), and the other from Bolivia (Steinbach 3453, 1917). Foster (1966) had cited Steinbach 3453 in GH withou sex-form; it is male (E. A. Kellogg, in litt.). In the flowers of both, as illustrated by Bernadello (1979) and by Nicora & Rugolo de Agrasar (1987), the anthers are large and the gynoecia small; the flow- t reference to its ers cannot be interpreted as female flowers with male-fertile anthers. Both sex-forms are directly comparable with the dimorphic floral syndrome in Cortaderia (Connor, 1973). Pollen in the male flowers of L. hieronymi is viable (Bernadello, 1979); we assume that it could lead to fertilization in normally reduced embryo sacs. This would necessitate strict sexuality i in some plants, or the co-occurrence of sexual embryo sacs with aposporous sacs— this latter is frequent enough in gametophytic apomictic grasses. However, for the greater part plants will have produced unre- duced aposporous embryo sacs; to add to the dif- ficulty, at the time of anthesis precocious embryo- genesis and endosperm formation is already well advanced. Female plants of that type would need to re-regulate the maturation of the gynoecium to allow the sequence of pollination at anthesis, pollen tube growth, and fertilization. Gynoecia in polli- niferous plants are imperfectly developed as Ber- nadello (1979) reported, and as the Argentinian and Bolivian specimens of L. hieronymi show; it seems very unlikely that fertilization and seed set would occur there. Any of the benefits deriving from a generation of sexual reproduction that male plants could interpose into persistent asexuality (Lynch & Gabriel, 1983) would seem denied to Lamprothyrsus because the chances of males con- tributing genes to the next generation are very slight. The presence of two polliniferous plants in L. hieronymi suggests, of course, that there may be more male plants in the populations than the her- barium collections reveal, but this needs to be es- Volume 80, Number 2 1993 Connor & Dawson 515 Evolution of Reproduction in Lamprothyrsus tablished. Should there be other male plants, ex- plicit dioecism may be current, and asexual reproduction, though predominant, may not be ex- clusive. Synchronous maturation and anthesis of male and female elements would lead to sexually reproduced seeds. Nevertheless, genes for asexu- ality would be successfully invading the populations to the detriment of dioecism, so these two male plants could be among the last of the male remnants of that species. The count of one male plant in about 100 specimens listed by Bernadello (1979) for the whole of its Argentinian range may support this proposition. The two male plants may only represent a low frequency sporadic mutation for the restoration of male-fertility, which brought with it the correlative changes in the gynoecium characteristic of dioe- cism and gynodioecism in the Arundineae. Such a mutation would be simpler if male-sterility was ge- netically controlled as in Cortaderia (Connor & Charlesworth, 1989) than if the nuclear-cytoplas- mic system common to wild plants (Couvet et al., 1986) was operatin It seems improbable that the male plants are the residue of a pseudogamous system because, as Asker (1980) has pointed out, pseudogamy is an unlikely precursor to non-pseudogamy; further, the two states never co-occur in one agamic complex. The synchrony essential to sexual reproduction would also be prerequisite of pseudogamy. To date, polliniferous plants are unreported in L. peruvianus, but they are as possible in this species as in L. hieronymi. They would be accom- panied by the same benefits, limitations, or con- straints already outlined for L. hieronymi. EVOLUTION OF APOMIXIS Pathways to the formation of populations con- sisting exclusively of female plants usually involve hybridism at some level; another associate is dioe- cism (see Westergaard, 1958). There is also a close correlation between hybridism, high levels of poly- ploidy, and apomixis. Lamprothyrsus, with two oreophilous species, would be best interpreted as the contemporary, asexual residue of an old, austral, allopolyploid genus that had become dioecious, but which later substituted autonomous apospory in females for the sexual functions of dioecism. A mutation for apo- mixis spreading through females in ІМ : IF di- oecious populations can readily lead to fixation (Charlesworth, 1980; Crowe, 1988; Lloyd, 1980; Marshall & Brown, 1981; Maynard Smith, 1971). Clearly, this pathway has been followed in apo- sporous South American species of Cortaderia (Connor, 1973, 1983b). Some diplosporous species of Poa and Calamogrostis have similar histories (Kellogg, 1990). That hypothesis avoided any direct role for hy- bridism in the evolution of apomixis, although the high level of polyploidy indicates ап amphiploid origin for taxa in Lamprothyrsus. Amphidiploidy itself neither assists nor hinders the development of an agamospermous system, but amphiploids will more easily accommodate occasional restoration of male-fertility. What selection pressures would have favored the adoption of an autonomous apomixis in a di- oecious taxon? It would be idle to invoke attractive ideas such as climatic conditions becoming unfa- vorable to the maturation and dispersal of pollen so that males became of lowered fitness; or that the sex-forms became overdispersed and exceeded pollination distances; or that sexual reproduction became limited by the ecological separation of the sex-forms as recorded by Fox & Harrison (1981) for Festuca, and by Freeman et al. (1976) for Distichlis, but undetected in Buchloe dactyloides (Quinn, 1991; Shaw et al., 1987) or in any of four Australian dioecious grass genera (Connor & Ja- cobs, 1991). Speculation about those forces should remain speculation; nevertheless, any increase in pollen fitness that was expected with the evolution of dioecism has been foregone, and with it male heterogamety. But all plants produce seeds, not alf of them; and precocious embryogenesis allows the benefit of the development of seeds before both panicle emergence and anthesis. It is evident that the maternal fitness and het- erozygosity incorporated by autonomous apomixis remains high if measured in Lamprothyrsus by contemporary ecological versatility, and by mor- phological variability (Bernadello, 1979; Conert, 1961; Pilger, 1906), even though apomixis is gen- erally not thought to be selected to increase eco- logical or evolutionary potentialities. Gabriel (1988) showed that apomicts, throug mutation and selection, could maintain a high level of genetic variability, and that they could be ee to perform well even in ecologically unstable en ronments. However, as Kellogg (1990) о. observed, little is known of the patterns of vari- ability in Lamprothyrsus, and both Conert's (1961) and Pilger's (1906) taxonomic attempts at ac- counting for infraspecific variation in L. hieronymi were unacceptable to Bernadello (1979). Lamprothyrsus as a genus is predominantly au- tonomously apomictic in reproduction; such a state in the Gramineae is approached only by monotypic 516 Annals of the Missouri Botanical Garden Vardus, where apomixis may be autonomous; but there it is diplosporous, not aposporous, and is not exclusive (Rychlewski, 1961). All flowers are per- fect and protogynous. Nogler (1984) felt that М. stricta L. should be reinvestigated because some critical data are unavailable. Dioecious grasses are significantly a New World phenomenon; the greatest number of dioecious gen- era is in Mexico (Reeder, 1969). In none of them is there any known departure from sexual repro- duction save for Lamprothyrsus and Cortaderia. It seems unlikely that there is a common ecological element, other than mountain environments, as- sociated with the evolution of a specialized form of gametophytic apomixis, which supplanted a system The explanation for the selection of this pathway must already specialized in gender diversification. lie in reproductive efficiency, or in fitness. Lam- prothyrsus offers an opportunity for evolutionary study as unique as itself. It is yet another example of the association between dioecism and apomixis elaborated by Gustafsson (1947) LITERATURE CITED ASKER, S. 1980. Gametophytic apomixis: elements and genetic regulation. Hereditas 93: 277-293. BERNADELLO, L. M. 1979. Sobre el género Lamproth yr- sus (Poaceae) en Argentina. Kurtziana 12-13: 119- 132 BROCK, R. D. € J. A.M. Brown. 1961. С yiotexonamy of Australian Danthonia. Austral. J. Bot. 9: 62-9 CARTIER, D. & A. LENOIR. 1980. Üoniribution à al Étude du développement de l'ovule du Phragmites aus- tralis (Cav.) Trin. ex Steud. Rev. Gén. Bot. 87 295. CASHMORE, А. B. An investigation of the taxo- nomic and agricultural characters of the Danthonia group. Е d onwealth of Australia, Sci. Indust. Res Bull. -22. cun TH, B. The cost of sex in relation to mating system. J. Theoret. Biol. 84: 655-671. ёл. J. & A. ABRAHAM. 1971. Stüdies on the cy tology and io yd of South Indian grasses. Cy- tologia 36: 579-5 См, W.D. 1968. "The Eea name for the com- reed. rp 17:168-169. му S. А. Renvoize. e Pra Graminum: Grasses a the World. Kew Bull. Conert, Н. J. Die Systematik und Anatomie der Arundineae. Verlag von J. Cramer, Weinheim, Ger пе . Current concepts of the systematics of the Panic гема ist 239-250 in T. R. Soder strom, K. W. Hilu, C. 5. Campbell & M. E. Bark, worth (editors), Grass Бузган and Evolution. Smithsonian Institution Press, Washington, RE e E. 4 i grasses. V. depen spec NS Bot. a 25 еа aand ies of Cortaderia. New l bus systems in Cortaderia (Gra- us f edis 27: 663-678 1979. Breeding systems in the grasses: a survey. New Zealand J. Bot. 17: 547-574. 1983a. Names and types of Cortaderia Stapf (Gramineae) In Taxon 32: 633-634. —_— 1983b. Cortaderia (Gramineae): чк hybrids i: the breeding system. Heredity 51: 395 ——— . Reproductive biology in the grasses. Pp. Pr | in T. R. Soderstrom, К. W. Hilu, С. s” Cam . E. Barkworth (editors), Grass Systematic and Evolution. Smithsonian Institution Pre с топ, D.C 990. Breeding systems in New Zealand n. XI. Gynodioecism in Chionochloa bro- moides. New Zealand J. Bot. 29: 59-65 . Charlesworth. Ca of male- sterility in cra Cortaderia (Gramineae). Heredity 63: 373-382. 1974, Names and types in Cor- taderia Stapf (Gramineae). Taxon 23: 595-605. & 1986. Australasian e к= diversification ad specialization. Pp. 41: 4 in B. A. Barlow (editor), Flora and Fauna of Aline Aus- tralasia: Ages and Origins. CSIRO, Melbourne. S. W. L. Jacops. 1991. Sex ratios in di- oecious Austra lian grasses: a preliminary assessment. Cunninghamia 2: 385-390. COSTAS- pU ee M. 1979. Embryogeny of Cortade- ria selloana and C. jubata (Gramineae). Bot. Gaz Crawfordsville) 140: 393-397. Couver, D., F. BoNNEMAISON & P.-H. Gouyon. The maintenance of females among hermaphrodites: the importance of nuclear cytoplasmic interactions. eredity 57: 325-330. pm 1988. The importance of recombination. -13 in R. E. Michod & B. R. Le "d S s o m :ellaneous species. New Zealand J. Bot. 27: 163- 165 FERNANDES, A. & M. Querios. 1969. Contribution a la connaissance cytotaxinomique de Spermatophyta | ries Bol. Soc. Brot. (2nd series) 43: 20- e. ES C. 1966. Studies in the flora of Bolivia IV. Rhodora 68: 223-358 Fox, J. К. & A. T. Hannison, 1981. of sexes and water balance in a dioeciou Oecologia (Berlin) 49: 233-235 FREEMAN, D. C., L. С. KLIKOFF & К. T. HARPER. 1976. oe resource utilization by the sexes of di- ecious plants. Science 21: 72-77 буйы, W. 1988. еи genetic сые for parthenogenetic species. Pp. 73-82 . de Jong (editor), Population си апа e Springer- erlag, Berlin, Heide Qu R. 1986. Degr Habitat assortment us grass. ep. és et niveaux de la variation C . ex Steud. ETH Stiftung Rübel Zürich 87: ВАС, M. Jav, J. n а M. К. Viric EL. 1984. Le RARE polyploide de Phrag- mites australis (Cav.) Trin. ex Steud. dans la région теднеггапеепе. Webbia 38: 715-721. GUSTAFSSON, А. 1947. Apomixis in Higher Plants. П. Volume 80, Number 2 1993 Connor & Dawson Evolution of Reproduction in Lamprothyrsus 517 The Causal Aspects of Apomixis. С. W. K. Gleerup, Lund, Sweden & M. SIMAK. 1963. X-ray photography ado seed о in Phragmites communis Trin. Her ditas 49 -450. Hain, J. B. & E. I. BEUZENBERG. 1966. Contributions to a chromosome atlas of the New Zealand flora— 7. Miscellaneous families. New Zealand J. Bot. 4: 255-266. KeELLocc, E. A. 1990. Variation and species limits in agamospermous grasses. Syst. Bot. 15: 112-123. LARSEN, K. 1963. Studies in the flora of Thailand. 14. Cytological studies in vascular plants of Thailand. Dansk Bot. Ark. 20: 211-275. RTIER, R. GoRENFLOT & P. olyploidie et méiose du Phragmites australis (Cav.) Trin. ex Steud. Compt. Rend. Hebd. Séances Acad. Sci., Sér. D 280: 621-624. Benefits and handicaps of sexual M. K. Hecht, W. E & B. Wallace (еве), in Biol. 13: 69-111 LurHER, H. 19 achtung über de fructifikative Md ep PA von Phragmites communis Trin. Act. ot . BROWN. a The evolution of apomixis. Heredity 47: 1-15. 1971. The origin and maintenance ме Sum. $ of s -175 in . Williams PEN ae UN Aldine Atherton: Chicago Nicora, E. G. & Z. E. RUGOLO DE AGRASAR. Los Géneros de Gramineas sd а dd Austral. Editorial Hemisferio Sur S. Án nos Aire Noc p G. A. 1984. I mixis. Pp. 475- in B. M. Johri (а. Embryology of Angio- БЕА Springer- Verlag, Berlin. РАкорІ, L. 9. Los géneros de aveneas de la flora Argentina. Revista Argent. Agron. 16: 205-223. PHILIPSON, M. N. Haustorial dian in Cor- taderia Granini. New Zealand J. Bot. 15: 777- 778. 1 Apomixis in SA onu. ш (Сга- mineae). New Zealand J. Bot. 16: 9. & H. nnor. 1984. Haustorial synergids in danthoniid grasses. Bot. Gaz. (Crawfordsville) 145: 7 PILGER, E ` 1906. Lamprothyrsus, eine neue gattung - gräser und ihre verwandten. Bot. Jahrb. Syst. , Beiblatt 85: 58-67 йш. К W. & G. DAVIDSE. 1971. Chromosome num KS ж" Costa Rican grasses. Brittonia 23: 293 -324. QUIN 1 Evolution of dioecy in Buchloe ecc (Gramineae): tests for sex-specific vege- tative characters, yo al пи uer and sexual iche- TO er. J. Bot. Renan, J.R 69. bc gramineas ds de Мез: Bol. Soc. Bot. México 30: 121-126. RICHARDS, А. J. 1986. M Breeding Systems. George Allen & Unwin, Londor RYCHLEWSKI, J. 1961. Cus embryological studies in the apomictic species Nardus stricta L. Act. Biol. megs Ser. Bot. 4: 1-23. SHaw, R. B., C. M. Bern & С. L. WINKLER. 1987. Sex patios of Buchloe dactyloides (Nutt.) Engelm. da о, s on the shortgrass steppe. Bot. Gaz. -89. 14 STOIAN, Ac LENOIR, P. Raicu & R. GORENFLOT. 1974. a méiose et la taille polliniques d'individus octo- ploides du Phragmites australis (Cav.) Trin. ex Steud. ey Rend. Hebd. Séances Acad. Sci., Sér. D 278: -4 5 en AARD, , M. The mechanism of sex de- termination in flowering plants. Advances Genet. 9: 217-281. FIRST REPORT OF BALLISTOCHOR Y IN THE POACEAE! Tatiana Sendulsky? ABSTRACT newly described mode of fruit dispersal mechanism in bambusoid grasses of the genera Sucrea and Raddia (Poaceae: Bambusoideae: Olyreae), previously unknown, is here discussed. Fruit dispersal in grasses (Poaceae) occurs in many different ways. The dispersal units (diaspores) mostly consist of caryopses protected by their bracts, which may be equipped with a variety of straight or twisted awns, bristles, hooks, pointed calluses, harpoonlike hairs, tissues for buoyancy. Alternatively, they may have feathery plumes, and special none of these specialized structures. In this case the dispersal unit may be composed of the whole spikelet, the caryopsis with its lemma and palea, or the naked caryopsis alone, with no protecting bracts; all three types fall as they are to the ground. There are other modes of dispersal: wind, rain- wash for short distances, streams, rivers, and even the sea, termites, ants, birds, fish, and through а variety of means of вина intentional or unintentional, related to h der Pijl, 1982; Sendulsky, 1965, 1966; ELM 1987; Filgue- 1986; Silberbauer-Gottsberger, 1984; Sen- dulsky et al., 1987). T ese modes of dispersal in grassland grasses iras, usually depend on one principal condition— the dry state of the mature dispersal unit. Thus caryopses, enclosed in their dry bracts with dry hairs and plumes, are light and easily carried by the wind. The awns, sharp-pointed calluses, and hooks be- come harder when dry and adhere better to fur or clothing. Only dry bracts let the anthecium or the caryopsis loose so that it falls easily to the ground. A completely different condition for dispersal of diaspores is required by some genera of herbaceous bambusoid grasses. These grasses belong to the large subfamily Bambusoideae and "аге not grass- land grasses but are confined to warm and humid environments, within forests, usually under dense shade” (Soderstrom & Calderón, 1974). According to Soderstrom & Calderón (1980), in 1972 and 1976 they collected about 162 dif- ferent species of Bambusoideae in Brazil; of these, At the time about 12 species of Raddia and a few species 80 were undescribed or new to science of Sucrea and of Eremitis, most of them still un- described, were brought under cultivation in the Rio de Janeiro Botanical Garden by Dimitri Sucre, who accompanied Soderstrom in fieldwork in Es- pirito Santo and Rio de Janeiro. In 1972, living samples of each species of these bambusoid grasses were given to me by Sucre and I started a living collection, which I still have in my office at the Instituto de Botánica in Sào Paulo, Brazil. Among known species there аге Raddia R. distichophylla (Schrad.) and the recently described Sucrea macu- brasiliensis Bertol., Chase, lata Soderstrom. These species have revealed an unusual fruit dispersal mechanism that I could nev- er have discovered by simply collecting, preparing, and studying the dried plant specimens. At first I was surprised by the appearance of seedlings of different species of Raddia and Sucrea maculata in pots occupied by other plants, at a distance of 1.5-2 m from the mother plants that had produced the fruits. The mystery was explained one hot evening when I was working late in the office. Preceded by a quiet "click," a fruit flew ' I thank Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) for the grant that has made possible a discovery of this new dispersal phenomenon in grasses. I owe gra n of the text, to Tarciso S. Filgueiras (IBGE) for helpful advice and suggestions, for his valuable assistance and revisio titude to my colleague Alasdair G. Burman to L. С. Labouriau (UB) for critical review of the manuscript, and to Stanley Yankowsky, Department of Bota any, Smithsonian Institutio not illustrated specifically in this paper, was used on, for anatomical preparation of the spikelet section of Sucrea maculata, which, while this is for the study and for schematical representation of the glume anatomy. I especially acknowledge the interest displayed by the late T. R. Soderstrom in this unusual and unexpected grass dispersal mechanism, and I dedicate this work to his m ? [nstituto de Botánica de Sào Paulo, Caixa Postal 4005, 01061- .970, Sao Paulo, £ 1993. ANN. MISSOURI Вот. Garb. 80: 518-521. memor S.P., Brasil. Volume 80, Number 2 1993 Sendulsk 519 y Ballistochory in the Poaceae and fell on my desk. I went to the plant, the branches of which were in slight movement, to see how it had happened; my first attempts to catch a fruit failed; the fruits were much faster than I was. Finally I caught one within its spikelet and had to hold it firmly, feeling the strong movement of the bracts between my fingers as I carried it to the microscope. DISPERSAL Herbaceous bamboos live mainly in humid trop- ical forests, where there is little or no wind; ane- mochory, one of the common modes of dispersal in grasses, is substituted in the bambusoid grasses mentioned above by autochory, i.e., the dispersal of the diaspores by the plant itself, using ballistic movement, which depends upon the turgor or vol- ume pressure in its cells and on the different an- atomical structures of the two sides of the glume. The diaspore of the Raddia species and of Sucrea maculata is composed of the caryopsis with its lemma and palea, which wrap tightly around it and form the anthecium. At the base of the anthecium there are two green glumes, firm in texture, as long as a spikelet or longer; these glumes are re- sponsible for the rapid ejection of the anthecium. he irritability of the plant, and the response that produces the movement, depend on factors that may act as a stimulus. According to Stras- burger’s Textbook of Botany (Harder et al., 1967: 393), “This stimulus releases a complicated series of reactions in the living material, the energy for which comes from the organism itself and not from the initial stimulus." Thus, during an investigation of the catapult phenomenon in herbaceous bam- busoid grasses, two kinds of factors were distin- guished that may act as stimuli. EXTERNAL FACTORS High temperature. On hot summer days, the plant increases its ballistic activity (pers. obs.). Wind, drops of water, contact with insects or animals, or any mechanical touch. Тһе ripe diaspores are very sensitive to the slightest touch; a drop of water, an insect, or a slight breeze may trigger ballistic movement. AUTONOMOUS FACTORS THAT ARE WITHIN OR ARISE WITHIN THE PLANT ITSELF Accelerated ripening of the fruit. The ап- thecium, when being ejected, is fully formed but still bears white, fresh, plumose stigmas, which indicate the rapid development of the caryopsis from the moment of pollination to the phenomenon of ejection. The plant has to produce ripe diaspores before the glumes become dry and withered, as this will prevent them from producing the catapult process, which depends on the turgor pressure in the cells. At the moment of dispersal, the diaspore and glumes must be full of water and turgid. Here lies the main difference in conditions for dispersal in grassland grasses and in this group of herbaceous bambusoid grasses. Weight and surface of the anthecium. The mature anthecium, in order to be effective, has to be heavy, while its surface is very smooth. These combined factors permit more speed, as the an- thecium slips from the glumes, and increase its trajectory. Nyctinastic movement. The sleep movements of the leaves, which were observed in Sucrea macu- lata and in all species of Raddia, often cause ejection of the diaspores. When the dense, fernlike leaves start their upward and downward movement, they sometimes get caught against each other; continuous movement causes the leaves to jerk free and the plant, or a few of its branches, swings. This sudden jerk or swaying of the branches is enough to elicit dispersal. Тће dor- siventrality of the glumes of bambusoid grasses Difference in anatomical structure. involves inequality of the cells on the two sides of the glumes; this is necessary for nastic movement. The external (abaxial) side of the glume consists of large, uniform, narrow, uprightly arranged, thin- walled epidermal cells, twice as large as those of the internal (adaxial) side (Fig. 1). The cells of the external side tend to expand transversally because of their form, density, tightness, and turgor. The internal side consists of smaller flat cells. When the turgor in the cells of the external side reaches its maximum, the layer of the cells of the internal side resists the expansion of the external layer. But there comes a moment when, as the result of some metabolic process, or the response to a certain stimulus, the internal side cells suddenly lose ten- sion and cease to resist. The volume of the internal side suddenly decreases, while that of the external side increases. Abscission of the anthecium. The turgid ripe anthecium, clasped tightly by its turgid glumes, holds the state of tension. The moment of abscission is important; the anthecium separates from the pedicel at its base and above the insertion of the glumes. This is decisive for ejection, equal to the phenomenon of the abscission of the middle lamellae 520 Annals of the Missouri Botanical Garden FIGURES 1-3. of Sucrea maculata (glumes in lower right, anthecium in upper left). —3. Longitudinal section of the fruit wall of the Impatiens species. Schematical representation, based on Strasburger (Harder et al., 1967). Abbreviation: ab. s. — abaxial side. mellae in /mpatiens spp., Balsaminaceae (Harder et al., 1967). As a result, the anthecium starts its slipping movement out of the tight glumes, and simultaneously the sides of the glume start from the base to roll in upon themselves forming four springs, which fling the anthecium 1.5-2 m (Fig. 2 Ballistichory in /mpatiens basically depends on the interaction between the thin-walled parenchy- ma cells of the external layer (expansion tissue and the smaller collenchyma cells of the internal layer (resistance tissue) of the fruit wall (Harder et al., 1967; Fahn, 1974). Comparison of the longitudinal sections of the fruit of Impatiens species with the cross section of the glume of Sucrea maculata makes clear the similarity in anatomical structures of the fruit wall and the glume. A rather fleshy fruit of /mpatiens М Ballistic movement. — 1. Transversal section of а glume of Sucrea maculata. — 2. Flying diaspore species (Fig. 3) presents four layers of thin-walled parenchyma cells on the external side and two or three layers of collenchyma cells on the internal side; the glume, a thin organ in Sucrea maculata, has only one layer of large, thin-walled epidermal cells on the external side, one layer of smaller cells on the internal side, and two layers in between, consisting of the mesophyll cells with chloroplasts. In spite of the fact that the contraction of these dissimilar organs (fruit wall of /mpatiens, glumes of grass) proceeds in different directions (longitu- dinally in the fruit wall and transversely in the glume), the catapult mechanism is the same in these plants, which thus tend to escape the shade of the mother plant. The dispersal of diaspores by means of ballistic movements in the i oaceae was previously un- known; it is hoped that these observations on this Volume 80, Number 2 1993 Sendulsky Ballistochory in the Poaceae mode of dispersal in the herbaceous bambusoid grasses will serve as a guide for further research. LITERATURE CITED DavipsE, С. 1987. Fruit dispersal in the Poaceae. Pp. 143-155 in T. R. Soderstrom, K. W. Hilu, C. S. y tematics and Evolution. Smithsonian Institution Press, Washington, FAHN, ч 1974. Plant Anatomy, 2nd edition. Pergamon s, Oxford, Englan йиш. T. S. 1986. О Conceito de Fruto em Gra- mineas e seu uso na Taxonomia da Familia. Pesquisa Wo pee Brasileira 21(2): 93-100. Harper, R. 1 trasburger's Textbook of ны. NO English edition: lbs n ed from 28th n edition. Longman PIL, E VAN DER. 1982. [geni у of D oui in Higher Plants, 3rd edition. Springer-Verlag, York. SENDU SKY, oT 1965. Contribution to the study of fruits and associated structures of grasses from the “Cerra- dos" (Aristida, Chloris, Mesosetum, Sorgastrum, Tristachya, Wo ips and Paspalum) jen Biol. Venez. rud. 63. 19 au to the study of fruits and associated structures of grasses from the *Cerrados" II — Andropogon L. Anais Acad. Brasil. Ci. (Rio de me 38(Supl. ): 159-185. S. FILGUEIRAS & A. С. BURMAN. Ta embryos and impar Pp. 31-35 in Т. К. Soderstrom, K. ‚ С. S. Campbell & Bark- worth (editors), сш Кы, S. a Smithsonian Institution Press, Washin SILBERBAUER-GOTTSBERGER, І. 1984. Fruit пети and trypan ed in Brazilian Cerrado grasses. Pl. Sys Evol. 1-37, e T R. & C. E. CALDERON. 1974. Primitive forest grasses and accu of the Bambusoideae. "ous Biotropica 6(3): Е & , 1987. In search of the Аң bamboos. Res. Rep. i. Geogr. Soc. 12: 647-654 Volume 80, Number 2, pp. 291-521 of the ANNALS OF THE MISSOURI BOTANICAL GARDEN as published on May 13, 1993. Monographs in Systematic Botany from the Missouri Botanical Garden Number 43 Comparative External Pollen Ultrastructure of the Araceae and Putatively Related Taxa i; Michael H. Grayum 167 pp., illustrated, softcover, $30.00, plus postage. This monograph presents the first extensive scanning electron microscope (SEM) survey of pollen of Araceae. Pollen of approximately 380 species of Araceae was studied, representing 99 of the 105 currently recognized genera. An effort was made to obtain material from all subgenera and sections distinguished in the most recent available taxonomic revisions (98 of 114 infrageneric taxa are represented in the study). Standardized, synoptic pollen descriptions are provided for all genera studied, and both whole-grain and high-magnification SEM micrographs are assembled in an atlas of 63 plates and 397 figures. An additional 12 plates and 92 figures summarizing the data accompany a detailed analysis of the polarity and evolution of pollen character states in Araceae. A discussion of the relationships of Araceae to other flowering plants features 15 plates and 104 figures depicting pollen of potential aroid relatives. i each additional book. Orders should be prepaid; a $1.00 fee will be added to orders requiring invoices. No shipments are made until payment is received. Mail form with your check or money order, payable to Missouri Botanical Gar den, to: ^ ; Please send St. Louis, MO 63166-0299, U.S.A. Send book(s) to: - De : А t Tage Eleven, Missouri Botanical Garden руба) of Mopograph No; 43. [1 Payment enclosed. O Send invoice ($1.00 fee will be · mee added to total). Address Postal Code Country 80(2) Prices ARE ЗивјЕСТ To CHANGE WITHOUT NOTICE CONTENTS Knowledge Brokering: The Mechanics of Synthesis, the 38th Annual Systematics Symposium of the Missouri Botanical Garden Introduction to the Symposium Nancy Morin & Theodore M. Barkley ................ Synthesis: A Historical Perspective Theodore M. Barkley The Future of Scientific Communications: More Questions Than Answers Arlene E. Luchsinger Ownership and Outreach: А Model for Administration of Shared Data Jean E. Nash Emerging Technology for Biological Data Collection and Analysis Stanley A. Morain Data Banking and the Role of Natural History Collections. Nancy R. Morin & Janet Gomon pai The National Science Foundation: Financial, Intellectual, and Knowledge Broker for dps and Jomes A Ene s o лы rr рса ане Revised Ci | ic Taxonomies of Calandrinia and Montiopsis (Portu- lacaceae) with Notes on Phylogeny of the Portulacaceous Alliance Mark A Hershkovitz Sega uw. Leaf Morphology of Calandrinia and Montiopsis (Portulacaceae) Mark A. Hershkovitz ORS Se МО ИМ УА» њу MEE A ue ЊЕ ea 2 A Revision of Chionolaena (Compositae, Gnaphalieae) Susana E. Freire — —— — A Revision of Malvaviscus (Malvaceae) Billie L. Turner & Meghan С. Mendenhall — The Rise and Fall of Pseudoludovia andreana (Cyclanthaceae) Roger Eriksson ......... - Chromosome Cytology in see African Gladiolus (Iridaceae) Peter Goldblatt, Ma- _ sahiro Takei & Z. A. Razz РИ Impact of the Eocene on the nem of Pus L. Constance 1. Millar mn- | Plant Life-forms and Biogeographic Relations of the Flora of Lagunillas (30°) in the Fog: | free Pacific Coastal Desert J. J. Armesto & P. E. Vidiella AH Evolution of Reproduction i in Lamprothyrsus (Arundineae: Gramineae) Н. E. Connor $ & cendi ЛИРИ diti radit = сьш First Report of Ballistochory in the Poaceae Tatiana Sendulsky „ninem У Cover illustration. Pleurothyrium giganthum van der Werff, by J. Myers. У pet ene К t A Lt Tewe T LET PIE Volume 80, Number 3 Summer 1993 Annals of the 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 Garden will also be accepted. Authors should write the Managing 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 Henk van der Werff Editor, Missouri Botanical Garden Amy Scheuler Managing Editor, Missouri Botanical Garden Diana Gunter Editorial Assistant, Missouri Botanical Garden Magdalen Lampe Ciri Moran — Publications Staff _ Ihsan A. Al-Shehbaz Missouri Botanical Garden Gerrit Davidse Missouri Botanical Garden Roy E. Gereau Missouri Botanical Garden Peter Goldblatt Missouri Botanical Garden _ Gordon McPherson Missouri Botanical Garden P. Mick Richardson А Missouri Botanical Garden For subscription information contact Department Eleven, P.O. Box 299, St. Louis, MO 63166-0299. Subscription price is $100 per volume U.S., $105 Canada and Mexico, $125 all other countries. Four issues per volume. The journal Novon is included in the subscription price of the ANNALS. © Missouri Botanical Garden 1993 _. THE Missouri BOTANICAL GARDEN A UU ett e The ANNALS OF THE Missour! BOTANICAL rat (ISSN 0026-6493) is published quarterly by enue, St. Louis, MO 63110. Second class give paid at St. Louis, MO and additional mailing z hanges to ANNALS © ; i Department Eleven, P.O. Box 299, St. Louis, МО 631660299 THIS PUBLICATION IS PRINTED ON ACID-FREE PAPER. | е д^ Volume 80 Number 3 1993 Annals of the Missouri Botanical Garden NZ rbcL SEQUENCE DATA AND PHYLOGENETIC RECONSTRUCTION IN SEED PLANTS: FOREWORD Gerard Zurawski! and Michael T. Clegg’ When asked to write a brief foreword to this volume, we wondered what, if anything, needed to be said. Certainly the articles presented here speak for themselves about the current and future impact of comparative rbcL sequencing on plant phylo- genetics. In the end, we felt it might be useful to describe briefly our view of the history that has led to this large and highly cooperative effort in systematics. For us at least, the history is important because it illustrates the interlocking nature of bi- ological research. The key players in this story include geneticists, biochemists, molecular biologists, biometricians, systematists, private industry, and national funding agencies. This diverse group laid the framework for what must surely be regarded as the most exciting effort in plant systematics in recent de- cades. Where did it all begin? One starting point was the discovery that some cytoplasmic organelles have their own DNA molecule. Another starting point was the suggestion by Zuckerkandl and Pau- ling in 1965 that the topology of branching of evolutionary trees could be inferred from the ac- cumulation of amino acid substitutions in proteins. The phenomenon of cytoplasmic inheritance provided the first tantalizing hints that extranuclear DNA molecules might exist in plant and animal cells. Wildman showed that an abundant protein fraction in plant cells called the **Fraction 1 Pro- tein" was transmitted cytoplasmically in tobacco. Later, in the mid-1960s, it was established that the chloroplast and mitochondrial organelles con- tained DNA molecules. The Fraction 1 Protein was shown to be a component of the enzyme ribulose- 1,5-bisphosphate carboxylase, responsible for fix- ing CO, in photosynthesis. Most work during the early 1970s was directed toward establishing the physical characteristics and size of the chloroplast DNA (i.e., that it is a closed circular molecule with an average GC content of about 38%). Beginning in the mid to late 1970s the new tools of recom- binant DNA technology were aggressively applied to the study of the chloroplast genome. Because the chloroplast genome is relatively small and be- cause chloroplast DNA constitutes an abundant fraction of total cellular DNA, most workers in the embryonic field of plant molecular biology focused on the study of the chloroplast genome. A problem ' DNAX Research Institute, 902 California Avenue, Palo Alto, California 94304. > Department of Botany and Plant Sciences, University of California, Riverside, California 92521, U.S.A. ANN. MISSOURI Bor. GARD. 80: 523-525. 1993. 524 Annals of the Missouri Botanical Garden of nearly universal interest was the cloning and sequencing of the large subunit of ribulose-1,5- bisphosphate carboxylase (rbcL, the gene for Frac- tion 1 Protein). McIntosh, Poulsen, and Bogorad were the first to achieve this feat in 1980 by defining the rbcL sequence from maize. Our involvement in this story began in 1981. One of us (M. Clegg) was a Sabbatical Visitor in the laboratory of Whitfeld and Bottomley at the Division of Plant Industry, C.S.I. R.O. in Canberra, Australia. The proposed research was to utilize RFLP analysis of chloroplast DNA to define relat- edness between various modern and ancient barley cultivars. G. Zurawski was a Research Fellow in the same laboratory and had recently completed the rbcL sequence from spinach and was starting the pea rbcL sequence. A casual comment by G.Z. to М.С. expressed an opinion that “RFLP analysis is fine, but why wouldn't you want to comparatively sequence part of the barley chloroplast genomes— you would be dealing with many more base changes and they would be fully defined.” This conversation prompted a determination by M.C. to learn D cloning and sequencing and was the beginning of our enduring collaborative friendship. The next part of our story continued in the laboratory of G.Z. at the DNAX Research Institute of Cellular and Molecular Biology in Palo Alto, California. We spent a couple of hectic weeks to- gether (M.C. as a Sabbatical Visitor) completing the sequence of rbcL and atpf genes from three diverse lineages of barley. Comparisons of the maize and spinach rbcL sequences indicated a high de- gree of similarity between taxa that represented the two classes of flowering plants. However, the extent of this similarity did not prepare us for the complete absence of nucleotide differences between our hard-earned barley sequences. This result ap- peared remarkable in view of the moderate to high levels of nuclear gene polymorphism common in barley and most other plant taxa. Comparison of the maize and barley rbcL sequences established that the rate of substitution was on the order of 1 х 107? substitutions per synonymous site per year, a very low value. Surprisingly, our friendship sur- vived the hard work required to establish this low value. Soon after, comparisons of cyanobacterial and flowering plant rbcL sequences (with S. E. Curtis) confirmed the slow rate of evolution of this chloroplast-encoded gene over the whole sweep of plant evolution. It was clear from these initial re- sults that the DNA sequences from slowly evolving chloroplast genes could be used to reconstruct plant phylogeny at its deepest levels. A detailed statistical analysis of the limited num- ber of rbcL sequences available in 1985 (with K. Ritland) showed that rbcL gene sequences could indeed be used for the reconstruction of plant evo- lutionary history. Despite these encouraging re- sults, it was still a difficult and laborious task to sequence a gene from any given plant taxon. The Maxam-Gilbert method of chemical modification was the standard sequencing approach of the early 1980s. This method required detailed restriction mapping and the construction of elaborate se- quencing strategies for its implementation. The time from initial preparation of DNA through to cloning a chloroplast gene and to determining its sequence frequently required six months of labor In parallel with work on chloroplast gene se- quences, Palmer and collaborators were establish- ing the utility of chloroplast DNA restriction site analyses for the study of plant relationships. This work captured the attention and imagination of the systematics community and attracted a large num- ber of imaginative, young systematists to the use of molecular data for phylogenetic analysis. Three developments laid the foundations for the mass of coordinated data presented in this issue of the Annals. The first was the adoption of a se- quencing protocol that utilized Sanger dideoxy-se- quencing together with a standard set of rbcL prim- ers. The rbcL primers were based on sequences that were conserved between the two distantly re- lated taxa (maize and spinach). Preliminary work with D. E. Giannasi showed that this primer set worked well for sequencing other plant rbcL genes. The set of primers was made freely available to the scientific community courtesy of DNAX Re- search Institute of Molecular and Cellular Biology (supported by Schering-Plough Corporation). The second major development was the invention of the The PCR method eliminated the cumbersome and time-con- polymerase chain reaction method (PCR). suming operations of molecular cloning. These two developments have led to a rich and exciting da- tabase that is the subject of this issue. The third parallel development was the explosive growth in computing power experienced in the 1980s and the widespread dissemination of algorithms for phy- logenetic inference by Swofford and by Felsenstein. This issue is the culmination of these three devel- opments, and in a larger sense it is the culmination of independent research trends in disciplines rang- ing from the chemistry of nucleic acids to system- atic biology and even to computer science. The whole exercise has been great fun for the Volume 80, Number 3 1993 Zurawski & Clegg 525 Foreword two of us. We could hardly have dreamed where this effort would lead when we began to work to- gether in 1981. It has been exciting to watch a technical revolution sweep through an entire area of biology. It is also gratifying and amusing to be told by colleagues at various institutions that this is one area where the plant sciences are way ahead of the animal sciences. INTRODUCTION Douglas E. Soltis, Mark W. Chase, and Richard G. Olmstead’ During the past several years, while rbcL se- quences were being rapidly generated, it occurred to us that an excellent opportunity existed to con- duct the first in-depth molecular phylogenetic anal- ysis of seed plants in general, and angiosperms in particular. We therefore invited the many inves- tigators throughout the world involved in rbcL se- quencing studies to contribute their unpublished sequences for a broad phylogenetic analysis. The responses to our inquiries were uniformly positive. The large analysis (Chase et al.), as well as the numerous companion papers, therefore attest to the spirit of cooperation in the systematics com- munity. We stress to readers of the large analysis that the trees presented therein should not be viewed in some absolute way as depicting “the phyloge- netic relationships" among seed plants. To the con- trary, numerous most-parsimonious solutions exist, and the results presented must be viewed with this cautionary note in mind. Furthermore, we do not view this exercise as an end in itself, but rather as a stimulus for directing future research phylogenetics of seed plants. Undoubtedly, the large analysis could ben taxon density) as well as more sophisticated and lengthy data analyses. We also hope that the large analysis will not only help to direct future molecular into the efit from sequences (greater investigations, but also investigations of morpho- logical, anatomical, chemical, and cytological char- acters. In conducting a phylogenetic analysis of such magnitude as the large analysis, several important points were made clear. First, it became very dif- ficult for individual investigators to remain isolated as students of particular groups (e.g., rosids, as- terids, magnoliids, etc.). For example, in the large analysis some traditional members of Rosidae ap- peared within an expanded Asteridae clade, and traditional Hamamelidae appeared in lineages as- sociated with distinct groups of rosids. Thus, in- vestigators working on supposedly distantly related groups were frequently connected as a direct result of their rbcL sequencing efforts. Similarly, in many instances, the taxa under investigation in one lab became suitable outgroups for studies in other labs of other groups of plants. All of these events forged new collaborations and fostered cooperation among many, perhaps most, of the numerous co-authors of the large analysis. Second, as a byproduct of this massive collaboration, many sequences became widely available, allowing those working on com- panion papers to sample more thoroughly than could have been done in isolation. Third, it is pos- sible to generate sequences much more easily and rapidly than one can rigorously analyze them. That is, our molecular methodology seems to have sur- passed in speed our computer-assisted means of data analysis. Clearly, future efforts should be di- rected toward improved and more rapid means of data analysis. Appropriate levels of time and re- sources need to be dedicated to the analysis of sequence data, as well as their acquisition. To assist readers, we have put together a large appendix at the end of this issue that lists all of the taxa for which rbcL sequences are reported, both in the large analysis and in the companion papers. Readers should note that the companion papers often include rbcL sequences for additional taxa not used in the large analysis of Chase et al. It is our hope that this appendix will assist readers in rapidly determining those taxa for which rbcL sequences are already available. However, it is also important to keep in mind that this appendix is Also provided in the appendix is voucher infor- mation for the taxa for which rbcL was sequenced. Although the great majority of species analyzed has been properly vouchered, vouchers were un- fortunately not prepared for some of the taxa for which rbcL sequences are reported in the papers of this issue. This may, in part, be an unfortunate consequence of the rapidly advancing DNA tech- nology allowing amplification of genes such as rbcL from DNA extracted from minute amounts of leaf material. Hence, leaf material is often obtained from botanical gardens or from collectors, and suf- ' Department of Botany, Washington State University, Pullman, a 99164-4238, U.S. ? Laboratory of Molecular Systematics, Royal Botanic Gardens, Kew Kingdom A. mond, Surrey TW9 3AB, United i j а of E.P.O. Biology, University of Colorado, Boulder, Colorado 80309-0334, U.S.A. ANN. MISSOURI Bor. Garb. 80: 526- 527. 1903. Volume 80, Number 3 Soltis et al. 527 Introduction ficient material for the preparation of vouchers is often unavailable. The studies in this issue also illustrate, therefore, the critical role that botanical gardens now play as sources of living material for phylogenetic reconstruction. Unfortunately, plants grown at botanical gardens sometimes are not rep- resented by herbarium specimens, and we would therefore urge botanical gardens to make voucher specimens for their living collections. Thus, the studies in this issue reinforce the need and impor- tance of voucher information for all taxa used in molecular systematic studies. When vouchers are not available, the results obtained should be con- sidered carefully, particularly in those instances in which surprising or unexpected taxonomic place- ments were obtained for such samples. We hope that the large analysis and companion papers presented herein will not only act as a stimulus for future molecular and non-molecular systematic endeavors, but also will underscore what can be accomplished through the collaboration of numerous investigators. The large number of in- vestigators cooperating in the large analysis is, as far as we are aware, unprecedented in the history of systematics. The future holds many additional opportunities for broad collaborations of similar scope as rbcL sequences from additional taxa and sequences of additional genes are obtained. It is our sincere hope that this issue will serve as a model and stimulus for what can be accomplished through the cooperation of numerous scientists. We dedicate this issue to Arthur Cronquist, whose ideas and syntheses of systematic data provided inspiration, guidance, and stimulus to all botanists and whose classification provided much of the con- text for the papers presented. Plant systematics has benefited immeasurably not only from Arthur's hypotheses, but also from his personality and wit. PHYLOGENETICS OF SEED Mark W. Chase, Douglas E. Soltis,’ Richard G. Olmstead,* David Morgan,’ PLANTS: AN ANALYSIS OF Donald H. Les, Brent D. Mishler,° NUCLEOTIDE SEQUENCES Melvin R. Duvall,’ Robert A. Price," FROM THE PLASTID Harold G. Hills? Yin-Long Qiu,’ 1 Kathleen A. Kron,’ Jeffrey Н. Rettig,’ GENE rbcL Elena Conti," Jeffrey D. Palmer, James R. Manhart,? Kenneth J. Sytsma,” Helen J. Michaels," W. John Kress,” Kenneth С. Karol," W. Dennis Clark,'* Mikael Hedrén,'* Brandon S. Gaut, Robert К. Jansen, ^ Ki-Joong Kim, Charles F. Wimpee, James F. Smith," Glenn R. Furnier,^ Steven H. Strauss," Qiu-Yun Xiang,* Gregory M. Plunkett,’ Pamela S. Soltis? Susan M. Swensen,'® Stephen E. Williams,? Paul A. Gadek,” Christopher J. Quinn," Luis E. Eguiarte, Edward Golenberg,” Gerald H. Learn, Jr. Sean W. Graham,” Spencer C. H. Barrett,? Selvadurai Dayanandan,” and Victor A. Albert ABSTRACT We present the results of two exploratory parsimony analyses of DNA sequences from 475 and 499 species of seed plants, respectively, representing all major taxonomic groups. The data are exclusively from the chloroplast gene rbcL, which codes for the large subunit - inima 1,5- “bisphosphate pin не oxygenase (RuBisCO or RuBPCase). We used two different state-transforma mptions resulting in two sets of cladograms: (i) equal-weighting for the 499-taxon analysis; a мө (ii) a proc due viia сао weights transversions over transitions within characters and codon positions among characters for the 475-taxon analysis. The degree of congruence between these results and other molecular, as vell as morphological, cladistic Раг ри indicates that rbcL sequence variation contains historical evidence appropriate for phy en analysis at this taxonomic level of sampling. Because the topologies presented are necessarily approximate and cannot be evaluated adequately for internal support, these results should be assessed from the perspective of their ceca value and used to direct future studies, both molecular and morphological. In both analyses, the three genera of Gnetales are placed together as the sister group of the flowering plants, and the anomalous aquatic Ceratophyllum (Ceratophyllaceae) is sister to all other flowering plants. Several major lineages identified correspond well with at least some recent taxonomic schemes for angiosperms, particularly those of Dahlgren and Thorne. The basalmost clades within the angiosperms are orders of the apparently polyphyletic subclass Magnoliidae sensu Cronquist. The most conspicuous feature of the topology is that the major division is not monocot versus dicot, Asteridae and Dilleniidae. Subclass Caryophyllidae is monophyletic and derived from within Rosidae in the 475-taxon analysis but is sister to a group composed of broadly delineated Asteridae and Rosidae in the 499-taxon study. ' The authors acknowledge the following support: U.S. National Science Foundation c rant BSR- sigh З HL and С - -91 BSR-9107484 to BDM; BSR-9007293 and BSR-9020055 to KJS; BSR-8821264 to KAK; BSR 8717600 and BSR. о to JDP; BSR-9020171 to RKJ; BSR-8817953 to a ho RAP and BSR-8957023 to SHS; BSR-880193 . Alverson; a Sloan Postdoctoral Fellowship to KJK -8914635 to VAA; a grant from the American Philosophical = to SEW; А nien Research Council E ЖҮСҮП to РАС and CJQ; an operating grant graduate scholarshi G from the Natural Sciences & Engineering Research Council of Canada; a grant to MWC and doctoral fellowship to VAA | from the American Orchid Society; and a postdoctoral scholarship ANN. Missouni Bor. Ganp. 80: 528-580. 1993. Volume 80, Number 3 1993 Chase et al. 529 Phylogenetics of Seed Plants Current assessments of higher-level relationships in seed plants are based largely on informed judg- ments of the relative value of various reproductive and vegetative characters (including secondary chemistry) and to some extent on historical pre- cedent. Authors of recent taxonomic schemes (for example, Dahlgren, 1980; Takhtajan, 1980, 1987; Cronquist, 1981; Thorne, 1983, 1992) have syn- thesized an enormous amount of information. Nev- ertheless, their taxonomic decisions have been guid- ed by estimations of which characters are reliable indicators of relationships. These differing judg- ents are responsible for radical differences in delimitation and relative ranks of taxa in eac system of classification. Recently, a number of explicit, cladistic hypotheses have been developed at inclusive hierarchical levels (e.g., Crane, 1985, 1988; Dahlgren & Bremer, 1985; Dahlgren et al., 1985; Doyle & Donoghue, 1986, 1992; Bremer et al., 1987; Donoghue & Doyle, 1989; Loconte & Stevenson, 1991; Martin & Dowd, 1991; Ham- by & Zimmer, 1992; Hufford, 1992; Olmstead et al., 1992; Taylor & Hickey, 1992). Such cladistic studies have previously been limited in scope; some data matrices contain significant taxonomic gaps, and in others characters for some taxa are missing. Both of these factors may have unpredictably mis- leading effects (Nixon & Davis, 1991; Platnick et al., 1991). Despite a great deal of investigation and analysis, seed-plant phylogenetics is, at best, in a preliminary stage of investigation and knowl- e. Molecular data, specifically DNA sequences, have received a great deal of attention as a potential source of “phylogenetically informative" charac- ters that are putatively less ambiguous than non- molecular characters. Such pronouncements suffer from the limitation that, at higher taxonomic levels, no extensive sampling and phylogenetic description of DNA sequence variation has taken place. The most taxonomically comprehensive analysis of nu- cleic acid sequences published so far on plants (rRNA; Hamby & Zimmer, 1992) used only 60 taxa, and a number of these were partial sequences. Martin & Dowd (1991), using nucleic acid se- quences of the small subunit of RuBisCO (rbcS) inferred from amino acid sequences, studied 335 taxa from 135 families, still less than a third of angiosperm families. If insufficient sampling of taxa or characters (i.e., sequence length, acknowledged as a problem with the rbcS data studied by Martin & Dowd, 1991) are indeed factors, then no valid investigation of the gene sequence exists for seed plants. Thus we are left with d an unfounded assessment of sequence data as having the potential to aid in estimating higher-level dE 8. "informativeness" of a given P Suggestions that the chloroplast gene rbcL, which codes for the large subunit of ribulose-1,5-bis- phosphate carboxylase/oxygenase (RuBisCO or RuBPCase), was an appropriate locus to use in phylogenetic studies began with Ritland & Clegg (1987) and Zurawski & Clegg (1987). Initial at- tempts to evaluate relationships used only a dozen or so sequences representing all land plants (Palmer et al., 1988; Giannasi et al., 1992). Other recent studies have been restricted to single families (Doe- bley et al., 1990; Kim et al., 1992) or putatively closely related families (Soltis et al., 1990; Les al., 1991; Donoghue et al., 1992; Olmstead et al., 1992; Rettig et al., 1992). Most of the latter studies began the process of incorporating signifi- cantly greater sampling to enhance their phylo- genetic perspective. The use of rbcL was spurred . Zurawski, who generously made available a set of internal sequencing primers. The advent of temperature cyclers and high-temperature-resis- tant DNA polymerases (sometimes termed Poly- merase Chain Reaction or PCR) has greatly en- hanced rates at which gene sequence data are accumulating, so that effects on phylogenetic es- timates of more intensive sampling of sequence variation can now be investigated. We wish to examine here the degree to which a representative sampling of sequence variation for rbcL contains evidence of the evolutionary history of seed plants. In this study, we address the quality of evidence present in rbcL sequences for all major seed-plant lineages (roughly 265 families, the exact number depending on the taxonomic scheme fol- lowed). To a limited extent, we will compare our phylogenetic hypotheses with recent schemes, but such comparisons are difficult and must be consid- ered heuristic b t group relationships ex- from the DGAPA, Universidad Nacional Autónoma = Mexico, Mexico, to LEE. We also thank David L. Swofford er G. Martin, helpful suggestions; and the Royal Botanic Gardens, manuscript. Footnotes 2-23 follow the Literature Cited. Westin, Michael J. Donoghue, and several anonymous reviewers for their Kew, for support in the final stages of the preparation of the 530 Annals of the Missouri Botanical Garden pressed in cladograms are difficult to relate directly to diagrams and statements of progenitor /descen- dant relationships used in many taxonomic schemes. Dahlgren et al. (1985) and Dahlgren & Bremer (1985) have published analyses most similar to the ones presented here but they are not of a similar scope. Although some specific topological compo- nents can be compared to other cladistic studies (Conti et al., 1993; Rodman et al., 1993; Qiu et al., 1993; and several of the other studies in this issue), morphological phylogenetic studies at this level with similar, broad taxonomic sampling do not exist. The computational difficulties of evalu- ating internal support (e.g., the bootstrap, Felsen- stein ; decay analysis. Bremer, | о = Жерїй from matrix randomness (Archie, 1989; Faith € Cranston, 1991; Kaállersjó et al., 1992) for such large numbers of taxa likewise prevent us from addressing extensively these issues here. We do present an analysis of families traditionally re- ferred to Hamamelidae as an example of how in- ternal support for specific clades might be exam- ined. (See also other papers in this issue that address internal and external support for subsets of the general results.) Thus, the broad relationships de- scribed here can be used to focus more restricted (with. fewer taxa), and therefore more rigorous, investigations. We expect that patterns presented here will change somewhat as sequences of more species are added or if methodological improvements permit exact solutions (for a discussion of progress, see Penny et al., 1992). These results nonetheless have great value, both from heuristic and methodological perspectives, although the preliminary nature of these studies precludes a detailed examination of implications for seed-plant taxonomy and character evolution, as well as investigations of gene and protein evolution. We have generally followed the taxonomic cir- cumscriptions of Cronquist (1981) for dicots be- cause this system conforms closely to those used in most textbooks and floras. For monocots, we have adopted the system of Dahlgren et al. (1985 but have changed the superorder ending *‘-iflorae’ ә ~ ,* used by Dahlgren to the more appropriate **-anae (Thorne, 1992). MATERIALS AND METHODS Selection of taxa for this study was not guided by a specific plan. Close examination of genera included in the analysis will reveal an uneven tax- onomic distribution; some groups are well repre- sented (e.g., Asteraceae and Asteridae in general, Cornaceae, Ericales, Magnoliaceae, Zingiberanae), whereas others are poorly sampled (e.g., dilleniid orders, especially Violales and Theales). Despite all subclasses and orders have at least some representatives. All spe- lack of a coordinated effort, cies that are used in this issue of the Annals of the Missouri Botanical Garden are listed alpha- betically by family in a final Appendix along with other information concerning voucher status, se- quence gaps, literature citations for published se- quences, and figures of this taxon occurs. Some taxa dar were included in the paper in which eac 475-taxon data set were excluded from the 499- taxon matrix. Thus, in the second analysis certain families have fewer representatives, but the overall representation of lineages is greater. ue to the large number of laboratories that contributed unpublished sequences, no standard- ized procedure was used to produce the sequences analyzed here. A generalization would be the fol- lowing: a fragment containing rbcL was amplified from a total DNA extract using primers that flank or are near the ends of the coding region; this fragment was then directly sequenced using one of several different procedures or was cloned using standard recombinant DNA techniques; dideoxy sequencing generally included both strands for at least Y of the minimally 1428-bp gene. Some work- ers used more closely spaced primers to sequence only one strand of DNA; either strategy appears to provide reasonably error-free sequences. Most extraction protocols relied on fresh or freshly dried leaf samples, but some samples were amplified from DNA extracted from herbarium specimens as old as 20 years. Amplification of rbcL from some taxa produced two different products. Some of these products were different enough in size to observe on an agarose gel, whereas others were detected initially because multiple bands occurred at the same points in au- toradiograms, indicating that more than one tem- plate was present. Nearly all cycads produced two loci (Hills & Chase, unpublished), which were sep- arately cloned and sequenced to characterize both copies. In cycads examined, one copy contained deletions that disrupted the reading frame, an in- ee dication that this сору may represent a “‘pseudo- gene." In Convolvulaceae, Olmstead (unpublished) detected two copies of rbcL, one copy of which contained deletions. In Canella (Canellaceae; Qiu et al., 1993) and Galphimia (Malpighiaceae; M. W. Chase, H. G. Hils & W. R. published), two size-conserved copies of rbcL were Anderson, un- also encountered. In Galphimia, one copy clearly contained substitutions at numerous sites otherwise conserved among angiosperms, suggesting that this Volume 80, Number 3 1993 Chase et al. 531 Phylogenetics of Seed Plants copy was a pseudogene. In Canella, no such un- usual substitutions were observed in either copy. Two, reading-frame-intact copies of rbcL are also reported in Ulmaceae (E. Conti & K. J. Sytsma, unpublished). For Canella, we included both se- quences in all analyses, but because the two se- quences are always each other's sister we show only the position of **Canella" (in fact, there are two terminals here; the complete matrix used in Search II thus has 500 terminals but only 499 taxa). Sequences of rbcL were easily aligned by sight. Among land plants, the coding region contains little size variation through base position 1428 (num- bered from the first nucleotide of methionine in the start codon, AUG). Positions 1426, 1427, and 1428 are the most common stop-codon among land plants; longer reading frames, up to 1452 in some monocots and 1458 in the Asteraceae (Kim et al., 1992), appear to be due to insertions, often of a short repeating sequence. Most laboratories have sequenced past this codon, but for phylogenetic analysis we have terminated all sequences at po- sition 1428 to be confident that we have analyzed homologous regions (the portion of the gene in- cluded in this analysis for each species is also included in the Appendix). All sequences were en- tered into a text file in NEXUS format (used by PAUP 3.0; Swofford, 1991) as complet and then analyzed directly in nucleotide жый Ма- trices used in both searches are available from the first author upon receipt of request and a diskette for each matrix. TREE-SEARCH STRATEGY Parsimony-based methods permit direct exam- ination of hypothesized character-state changes on the reconstructed tree, and this information can be used in studies of molecular evolution. (These are, however, likely to be underestimates of se- quence change and could be misleading for this reason.) Numerous empirical studies have shown that not all classes of substitutions are equally like- ly, and this kind of information may be incorpo- rated into weightng schemes for nucleotide data (Swofford & Olsen, 1990; Albert & Mishler, 1992; among others). Various models of molecular evo- lution exist, and appropriately circumspect use of these may assist in the separation of historical signal from homoplasy. The character-state weighting model of Albert et al. (1993, this issue) uses prob- ability formulae to calculate weights for different classes of molecular change. We have used their method in the 475-taxon analysis presented here. Although the assumptions of the Albert et al. model are admittedly simplistic, we nonetheless support the investigation of weighting approaches to nu- cleotide data and view the model used here as a justifiable first approximation. It must be recog- nized that giving all categories of molecular change equal transformational weight is also an assump- tion, but one that the investigation of Albert et al. (1993) found to be adequate if sampling effects were not a factor. Actual weights applied in this study fall within such a narrow and minimally asymmetrical range that deviation of results from those using the “equal weighting" criterion of Fitch (1971) is likely to be slight (see Albert et al., 1993). Recently, concern has focused on the probability that “islands” of equally parsimonious trees exist, particularly in large data sets (Maddison, 1991). Because of their enormous size, our analyses do not use methods of multiple random taxon-addition, which have been suggested to uncover such dis- junct, equally optimal islands. This topic is ad- dressed in most empirical papers in this number. The search for parsimonious trees consisted of several separate but linked heuristic searches using either PAUP 3.0r (Search I; 475 taxa) or 3.0s (Search Il; 499 taxa). All searches included the full data matrix (all codon positions). Search I was performed on a SPARC II (Sun, Inc.) workstation (PAUP 3.0 for non-Macintosh computers is avail- able only by special arrangement with D. Swofford). For Search II, a Macintosh Quadra 800 with 20 MB RAM vas used. (Although slower than a Sun computer, the more interactive nature of a search on a PC is preferred by many workers.) In Search I, an initial heuristic search with character-state changes given equal weight (i.e., “unordered” status), SIMPLE data addition se- quence, STEEPEST DESCENT, and NNI (nearest- neighbor interchange) branch swapping algorithm was used to find a single tree (MULPARS option deactivated). The second phase used this single tree obtained from phase one as a starting tree for another heuristic search, this time using the SPR (subtree pruning-regrafting) branch swapping al- gorithm with MULPARS deactivated again. The third phase paralleled the first and second; the single SPR tree from phase two was used as the starting topology for a heuristic search using the TBR (tree bisection-reconnection) branch swapping algorithm, again with MULPARS deactivated. The fourth phase used the single TBR tree as the starting point for a heuristic search employing a character-state weighting criterion (with a dif- ferent step matrix for each codon position; cf. Swofford, 1991; Albert & Mishler, 1992; Albert et al., 1993). This time MULPARS was activated. Annals of the Missouri Botanical Garden Transversion substitutions were weighted over transitions differentially by codon position (Albert et al., 1993). The specific weights used were the following: for transitions, 5520 (first positions), 6368 (second positions), 4039 (third positions); for transversions, 6620 (first positions), 7470 (second positions), 5127 (third positions). These weights were calculated from empirically derived param- eters (see Albert et al., 1993, for a complete de- scription and justification). Because use of step matrices is CPU-intensive, this search was exe- cuted with the simplest branch-swapping algorithm, NNI. Additionally, because of dynamic RAM lim- itations (a tree of 475 terminals occupies a great deal of memory), we restricted our search to a maximum of 500 trees. Although the initial TBR tree was by default a member of a single island, we hoped to provide a bridge to shorter trees through use of the STEEPEST DESCENT option (see Mad- dison, 1991; Swofford, 1991). This final step yield- ed 500 equally parsimonious, weighted trees. Be- cause the maximum prespecified number of trees was found, many others probably exist at the same length. Search I required approximately 200 hours to complete. Search II was performed for three reasons. First, the likely existence of other islands of equal or greater parsimony prompted us to use a strategy that would be more likely to find shorter trees and perhaps a different topology. Second, we were con- cerned about the effects of the Albert et al. (1993) weighting scheme upon the resulting topology. Third, we wished to examine positions of additional taxa (and make use of updated sequences) that became available after Search 1 was completed; many of these belonged to previously unrepre- sented lineages. Differences between these two sets of trees could thus be due to different taxon sam- pling, corrections or completions of sequences after Search I was finished, or search strategy. We did not intend these two searches to be controlled, direct tests (i.e., with only one variable differing between them); we show them both, rather than simply the one that we judge to be better (Search П), because their similarities, despite variation in taxon composition and search strategy, are con- siderable. They each represent results of searches that in their own context are worthy of publication, and their differences should be viewed as reasons to be skeptical of both and as cause for future study with more rbcL data as well as other char- acters. In Search II, we were able to save more trees at the shortest length found: 3,900 rather than only 500 trees. The initial starting tree was pro- duced by using the CLOSEST addition sequence with the HOLD option set for five trees (this in effect permitted initial swapping on several differ- ent starting topologies). Approximately 120 hours were required merely to add all taxa in this manner. The initial search (i) used NNI swapping and STEEPEST DESCENT with MULPARS off. The shortest single tree found was then swapped on using (ii) TBR, which generally found a shorter tree, at which time (iii) NNI (with MULPARS) was used. When use of MULPARS resulted in 3,900 trees, which used up available RAM, a single tree was randomly selected (iv) to swap on with TBR (MULPARS off). If this resulted in a shorter tree being found, the search was then stopped and re- started (iii, again) using NNI and MULPARS and this shorter tree as starting point (iii and iv were repeated until no shorter trees were found). The shortest tree length found with this method was 16,305. Three randomly selected trees from the 3,900 saved at this length were swapped in suc- cession to completion with TBR (no MULPARS), and no shorter trees were found. A strict consensus tree was computed, and branch lengths for one randomly selected tree were calculated using the ACCTRAN optimization. Search II thus used no relative weighting; it required approximately four weeks to complete. CAVEATS Methodologically, these searches suffer from (i) uncertainty about maximum parsimony, (ii) un- questionable absence of many trees at the same level of optimality, (iii) identification of only a single topology, and (iv), in Search I, incomplete branch swapping on any of the shortest trees found. We would never recommend these search strategies for smaller data sets, but several options were seriously restricted by the number of taxa included (these are reputedly the largest PAUP analyses attempted to date). Specific sections of the general topologies, when analyzed in a more “localized”” manner, pro- vide different sets of relationships (see Michaels et al., 1993; Morgan & Soltis, 1993; both this issue). The broader taxon distribution of the general anal- ysis (thus with far greater outgroup information) may be assessing character-state change on the immediate branch leading to a specific ingroup differently than in more restricted analyses. Great- er outgroup information could “improve” the in- group analysis or add spuriousness to it; it is gen- erally best to implement tree searches both with and without outgroups to examine their effects. If more restricted analyses differ from those pre- Volume 80, Number 3 993 Chase et al. 533 Phylogenetics of Seed Plants sented here, we would certainly favor the former because of increased confidence in finding parsi- monious solutions. We nonetheless view these anal- yses as instructive about seed-plant relationships and the utility and limits of rbcL information (see Discussion). The major limiting factor in our studies is clearly matrix size. When confronted with the largest mo- lecular database relevant to seed plants, compu- tational trade-offs inevitably arose. The amount of time spent on these approximations may not be directly proportional to the time taken to generate all the rbcL sequences, but no method exists to predict how many trees at how many steps shorter might have been found after additional months or even years of continuous computation. We have opted for the approximations presented here rather than commit ourselves to an open-ended experi- ment. Potential effects of errors in autoradiogram read- ing and data entry should be considered. We de- tected а number of internal stop-codons in se- quences used in these analyses, and other workers have reported errors of various kinds. Most of these were corrected in the matrix used in Search П. Certainly the potential for errors is present, but the effects of such mistakes should not be extensive because they are likely to be random. Identification of some taxa in this analysis has been questioned due to peculiarity of placement. One example of this has been now identified among the species used in Search I. We have not expunged it from the illustrations; the results represent the outcome of real tree searches and are instructive for that reason. The sequence analyzed was an actual member of the clade into which it was placed, but it was not the species to which it had been attributed. The material sent by a botanical garden and labeled as Kirengeshoma was evidently mis- identified. It was almost certainly a member of the Parnassia group (Saxifragaceae sensu lato; the position it occupies in Fig. 11A) rather than a er of Hy rangeaceae. Sequencing of another ,»5 unpublished). Other "surprising" results also have been checked by obtaining another sample of the taxon in question and re-sequencing rbcL. The original sequence of purportedly saxifragaceous Montinia (which nested in Solanales) was checked and found to be accurate. Still others (such as Sargentodoxa among Fabaceae) are still being reassessed. i vouchers for each species in this study (see Ap- ost workers have tried to make pendix). Many tissue samples were provided by botanical gardens, and, if vouchers are not included with samples, investigators are dependent upon identification of these plants by the respective or- ganizations (see Goldblatt et al., 1992; it is critical that a voucher sample taken from the same plant used for DNA extraction be included at the time of collection; accession numbers or vouchers taken previously are subject to later events, such as lost labels or collector mistakes). Because a number of the samples used in these two analyses are un- vouchered (see Appendix), reproducibility is com- promised. In cases with multiple species from a family, we should be able to recognize grossly mis- identified samples (but this is true for only 37% of the 265 families represented). RESULTS For display purposes, we show here the com- binable component consensus (Bremer, 1990) of the 500 equally parsimonious trees found in Search I (Figs. 1-15, A series). Because these are char- acter-state weighted trees, tree lengths and tree statistics (e.g., consistency index, etc.) are not com- parable to those of Fitch trees and are not given here. For Fitch trees found in Search II, the length was 16,305 with a consistency index (C.L, ex- cluding unique substitutions and constant charac- ters) of 0.102 and a retention index (R.I.) of 0.632. The branch lengths, again with ACCTRAN opti- mization, are shown on one of the 3,900 trees selected at random (Figs. 1-15, series B). Branches that collapse in the strict consensus tree are indi- cated by arrows on the B series (Figs. 3-15B). First we will summarize the topology found in Search I (“A” series of figures). The results of Search II (**B" series) have been interdigitated with those of Search I to facilitate comparisons. After describing the results of Search I, we briefly ex- amine major differences between the two results. Note that all figures in the A series are from the combinable component consensus tree, whereas in the B series, Figures 1B and 2B are the strict consensus tree (branch lengths in Fig. 2B, however, were taken from a single tree) and the remainder (Figs. 3B-15B) are a single tree randomly selected from 3,900 equally parsimonious Fitch trees (this may be confusing; for example, whereas the po- lytomy among monosulcate clades in Figs. 1B and 2B is due to variation among the 3,900 trees for branches in this portion, the topology shown in 3B is resolved because it is resolved in the single tree selected). To indicate branches of the B series that are absent in the strict consensus tree of the 3,900, 534 Annals of the Missouri Botanical Garden we have placed arrows to their right. The branch lengths presented in the B series should under no circumstances be interpreted as meaningful mea- sures of support; thus, in the example of Hama- melidae provided, branches that decay at one step less parsimonious have lengths that range from 2 to 16 steps, whereas those that decay at four steps less parsimonious range from 7 to 14 steps (Fig. 16). The sole reason for providing branch lengths in the B series is to permit readers to estimate roughly relative degrees of divergence and to iden- tify cases in which long terminal branches are connected to short internal branches (a situation in which adding related taxa often radically alters hypothesized relationships). Search I used the taxonomic circumscriptions of Cronquist (1981) for dicots and Dahlgren et al. (1985) for monocots, although we acknowledge that other re- cent systems fit these results better. We arranged the unrooted trees of both searches with cycads sister to all other seed plants (Figs. 1–3) in accord with recent results of several non-molecular cla- distic studies (Crane, 1985, 1988; Doyle & Don- oghue, 1986, 1992). In Search I, conifers are paraphyletic, but some trees (not shown) found in Unless stated otherwise, we have Search II have a monophyletic conifer lineage; the strict consensus tree from Search II is unresolved regarding conifers (Figs. 1B, 2B; we have cited in this section the B series of figures along with the A series if they include the same general set of taxa). The three genera of Gnetales, Ephedra, Welwitschia, and Gnetum, are highly divergent from all other seed plants but were nonetheless identified as sister of the angiosperms (Fig. 3A, B), within which Ceratophyllum (Ceratophyllaceae) alone is sister to and highly divergent from the rest (Fig. 4A, B). The major feature of flowering plants (exclusive of Ceratophyllum) is their separation into two major groups; these correspond well with distributions of the two major angiospermous pollen types, uniaperturate (monosulcate and monosul- cate-derived) and triaperturate (tricolpate and tri- colpate-derived). Ceratophyllum has inaperturate pollen (Cronquist, 1981). The major exception to this split is the presence of tricolpate pollen in Illiciaceae and Schisandraceae, which fall among the monosulcate taxa (Fig. 4A, B; see Qiu et al., 1993, this issue). No morphological support for monophyly of the monosulcate clade has been rec- ognized in the literature (their pollen type exists among nonflowering seed plants and thus must be considered plesiomorphic). Three monophyletic lineages within uniapertur- ate magnolids were identified, and these corre- spond closely, although not exactly, to (i) Mag- noliales, (ii) Laurales, and (iii) “paleoherbs” (here efined as composed of Aristolochiales, Piperales, and Nymphaeales; Fig. 4A). Monocots (also with uniaperturate pollen and sometimes included in "paleoherbs"; Donoghue & Doyle, 1989) repre- sent a fourth member of this clade. Among the paleoherbs are also nested several problematic fam- Illiciaceae, Schisandraceae (both Illiciales), Amborellaceae (Laurales), and Austrobaileyaceae (Magnoliales). Chloranthaceae are also allied cla- distically with the paleoherbs, but Chloranthus does not form a monophyletic group with other Piper- ilies: ales. Monocots are a well supported monophyletic group (see Duvall et al., 1993, and Qiu et al., 1993, both this issue) and are derived from within monosulcate Magnoliidae; the paleoherbs are their immediate sister group (Fig. 4A). Within monocots (Fig. 5A, B), Aranae plus Pleea (of polyphyletic Melanthiaceae; Lilianae) are basal-most, followed by Alismatanae plus Burmannia (Burmanniaceae) and Aletris (Melanthiaceae). Lilianae form a para- phyletic series of three lineages that correspond well to Dioscoreales, Liliales, and Asparagales of Dahlgren et al. (1985), except in the placement of certain families (Iridaceae, Orchidaceae, and Smilacaceae) and genera (Chamaelirium of Melan- thiaceae: Melanthiales). Vellozia (Velloziaceae: Bromelianae), Freycinetia (Pandanaceae: Panda- nanae), and Sphaeradenia (Cyclanthaceae: Cy- clanthanae) together form a adn clade that is collectively sister of the Liliales. The “com- melinoid” group of monocots (Fig. 6A, B) incor- porates all of those that Harris & Hartley (1980) found to exhibit fluorescing cell-wall phenolics. In both searches, this commelinoid clade includes monophyletic Arecanae and Zingiberanae and polyphyletic Bromelianae and Commelinanae (Fig. 6A, B). Cyclanthaceae have the same phenolic biosynthetic pathway but do not accumulate end products; they are not members of the commelinoid assemblage (Clark et al., in prep.). he two orders of Magnoliidae with triaperturate pollen, Ranunculales and Papaverales, form a clade that is sister to the rest of “eudicots” (Fig. 7A, B). The term “eudicot”” has been variously defined in the literature, but we use it here to refer to all angiosperms with triaperturate or triaperturate-de- rived pollen (Donoghue & Doyle, 1989; Doyle & Holton, 1991). This is one of the best supported clades among angiosperms (Qiu et al., 1993, this issue). Two other basal clades within the eudicots (Fig. 7A, B) consist of some Hamamelidae (Troch- odendraceae and Tetracentraceae) and Platana- ceae, Sabiaceae, Nelumbonaceae, and Proteaceae. Within eudicots, two large sister clades are iden- Volume 80, Number 3 1993 Chase et al 535 Phylogenetics of Seed Plants tified, one that corresponds roughly to Asteridae and the other to Rosidae (Figs. 1, 2). Membership in both lineages is considerably expanded with re- spect to their circumscription by Cronquist (1981), although less so with respect to the circumscriptions of Dahlgren & Bremer (1985) and Thorne (1992). These two major clades (Fig. 2A, B) reflect the division of eudicots into two major groups (Young & Watson, 1970): sympetalous/tenuinucellate and polypetalous/not tenuinucellate (asterids and ros- ids, respectively, in our figures). The “crassinu- cellate" condition actually consists of several dif- ferent states. Exceptions to this generalization exist, but these traits appear much less homoplastic here than in most systems of classification. The basal- most lineage within Rosidae includes Saxifragaceae sensu stricto and Crassulaceae besides lower ha- mamelids such as Cercidiphyllum and Hamamelis (Fig. 8A; see Morgan & Soltis, 1993, this issue, for a discussion of Saxifragaceae sensu lato). The next-most-basal group contains Caryophyllidae (in- cluding Plumbaginaceae and Polygonaceae; Rettig et al., 1992) plus Droseraceae, Nepenthaceae (Ne- penthales, Dilleniidae; see Albert et al., 1992b), Dilleniaceae (Dilleniidae), and Vitaceae (Fig. 9A). The remaining Rosidae are split into two large sister groups (Figs. 2A, 10A, 11A). In one (rosid I) are several families of higher Hamamelidae, Eu- phorbiales, Fabales, Linales, Polygalales, and Ro- sales (Fig. 11A). Other members of this clade in- clude a number of dilleniid families: Ochnaceae (Theales), Datiscaceae, Passifloraceae, and Viola- ceae (all Violales). Ordinal boundaries of this group of Rosidae (sensu Cronquist, 1981) are largely unsupported; this assemblage is particularly het- erogeneous. The largest polytomy in the consensus tree from Search I occurs at the base of this group, and sampling of the families that potentially belong to this clade is the most sparse in this analysis. The other major lineage of Rosidae (rosid II; Fig. 10A) includes orders Myrtales (see Conti et al., 1993, this issue) and Sapindales, for which ordinal boundaries are reasonably intact. Malvaceae (Mal- vales; Dilleniidae) and all but one of the mustard- oil families (those in Capparales plus others in Dil- leniidae; see Rodman et al., 1993, this issue) are associated with Sapindales. Geraniaceae are also members of this clade, although other members of Geraniales appear elsewhere (Oxalidaceae with a group of families in Rosales, Fig. 11A; Balsami- naceae with Ebenales, Fig. 13A, B; and mustard- oil-producing Limnanthaceae and Tropaeolaceae with Capparales, Fig. 10A, B; Price & Palmer, 1993, this issue). Two members of Rosales, Greyia (Greyiaceae) and Francoa (Saxifragaceae), appear derived from within Geraniaceae, if Viviania is included in Geraniaceae (see also Price & Palmer, 1993, and Morgan & Soltis, 1993, this issue). The tenuinucellate/sympetalous clade that ter- minates in Asteridae sensu Cronquist (1981) was also identified by Olmstead et al. (1992) in their efforts to circumscribe subclass Asteridae using rbcL sequences. This study greatly expands upon their sampling and identifies as members of the asterid clade two lineages of often polypetalous Rosidae (Figs. 2A, 12A): (i) Santalales plus Paeoni- aceae and Gunneraceae and (ii) some families of Cornales plus Hydrangeaceae (see Xiang et al., 1993, this issue). The sister group of Asteridae is a clade (Figs. 2A, B, 13A, B; see Kron & Chase, 1993, this issue) that contains the dilleniid orders Ebenales, Ericales, Primulales, Diapensiales, plus some members of Theales (Actinidiaceae and Thea- ceae). Sarraceniaceae (Nepenthales; Dilleniidae) and Roridula (but not Byblis of Byblidaceae of Ro- sales: Rosidae) are also members of this lineage (Albert et al., 1992b). Polemoniaceae (Solanales: Asteridae) and Balsaminaceae (Geraniales: Rosi- dae) also belong to this ericalean/ebenalean group. (See also Olmstead et al., 1993, this issue, for a treatment of the Asteridae sensu lato.) Asteridae sensu Cronquist split into two major sister groups. In one of these (asterid II; Fig. 14A, B) are families of Asterales, Calycerales, Campan- ulales, Dipsacales, and some Solanales (Menyan- thaceae). Rosid taxa that are members of this clade include Apiales, Aquifoliaceae (Celastrales), some Cornaceae (Cornales), Pittosporaceae and Gros- sulariaceae (both Rosales). In the other major clade (Fig. 15A, B) fall orders Callitrichales, Gentianales, Lamiales, Rubiales, Scrophulariales, and most So- lanales, although these ordinal limits are not always supported (see Olmstead et al., 1993, this issue). In this clade (Fig. 15A, B) is a group that includes rosids Aucuba (Cornaceae: Cornales) and Garrya (Garryaceae: Cornales) and hamamelid Fucommia (Eucommiaceae: Eucommiales), all of which ac- cumulate aucubin (Cronquist, 1981) and share dis- tinctive anatomical wood characteristics (E. Whee- ler, pers. Thus the suite of floral characteristics that have been interpreted as sup- comm.). port for the monophyly of Asteridae sensu Cron- quist (1981) appears either (i) to have twice arisen independently from ancestors with rosalean and cornalean floral traits or (ii) to have undergone reversals in groups traditionally included in Rosidae (sensu Cronquist; Donoghue et al., 1992; Olmstead et al., 1992, 1993, this issue). Search II. А number of taxa from Search I were removed from Search II (all marked with a “+”? in the A series of figures) to accommodate the representatives of additional lineages in Search П. 536 Annals MD EL Garden Most of the removed taxa were from monophyletic families in which six or more species were present in Search I (for example, Asteraceae, Ericaceae sen- su lato, Magnoliaceae, and Poaceae). Two others, Burmannia (Burmanniaceae) and Hydrolea (Hy- drophyllaceae), were removed because they are highly sequence divergent from all other taxa; in separate, smaller analyses, these two appear to be involved in “branch attractions” and attach in rad- ically different positions as other taxa are added or removed (other similarly divergent genera, for example, Paeonia and Gunnera, do not cause these problems and were kept; we admit to being rela- tively arbitrary in removing only these two taxa). Burmannia presents an interesting case. Members of Burmanniaceae are often achlorophyllous; in spite of being green, Burmannia may still derive a great deal of its nutrition through its mycorrhizal associate. [n such cases in other families, a number of protein loci exhibit higher rates of sequence divergence (as measured by relative branch lengths when compared to completely autotrophic mem- bers of their lineages; C. dePamphilis, pers. comm.). Thus the high levels of sequence divergence, which make it difficult to place these taxa accurately, are a product of or at least associated with their partial heterotrophy. For Search II, additional taxa were available (these species are marked with asterisks in the B series of figures). The placements of these are described first, and then different arrangements of taxa included in both searches are identified (the clades involved with major shifts of position are marked with an “$” in Fig. 2B). This last section describes only the major shifts of position, but many shifts also occur within clades (for ex- ample, 12 “minor” shifts take place among the taxa in Fig. 15, a clade in which only a few new ee species were added). What constitutes a “major versus a ‘‘minor” alteration is, of course, a matter of personal perspective. We would therefore advise readers to examine carefully the trees from both searches for taxa of specific interest, which is one reason we presented and intercalated the results of both analyses. Positions of additional taxa. Ginkgo (Gink- goaceae) intercalates between cycads and conifers (Figs. 1-3B). Taxus (Taxaceae), Cephalotaxus (Cephalotaxaceae), and Sciadopitys (Taxodiaceae or in its own family) are members of the non- Pinaceae clade of conifers, with Sciadopitys sister to the rest of that clade (Fig. 3B). Within magnoliid angiosperms, Lactoris (Lactoridaceae) is sister to Aristolochia (Aristolochiaceae). Other new mag- noliid species added in Search II represent addi- tional members of families already present in Search I, and all of these form monophyletic units with other members of their respective families (Fig. B). Sister to all other monocots is Acorus (Araceae; see Duvall et al., 1993, this issue). Calochortus (Calochortaceae) is sister to Liliaceae, Нетего- callis (Hemerocallidaceae) is sister to Chlorophy- tum (Anthericaceae), and Lomandra (Dasypogona- EN ceae) is a member of the often arborescent clade composed of Agavaceae, Asphodelaceae, and Xan- thorrhoeacae within Lilianae (Fig. i members of families in Search I are all placed as ditional sister taxa to their respective family representa- tives. Among pa commelinoid clade identified in Search I (Fig. 6A, B) are the following additional families: Sparganium (Sparganiaceae) sister to Ty- pha (Typhaceae) and Lachnocaulon (Eriocaula- ceae) among the graminoids. Among eudicots (Fig. 7A, B), Buxaceae) is sister to Trochodendrales (Trocho- Pachysandra ~ dendron and Tetracentron). Composition of the various rosid clades is somewhat different in Search II (Fig. 2B; see below), but the following additional taxa are placed roughly among the rosid II (Fig. 10B) clade identified in Search I: Shorea (Dipter- ocarpaceae), Theobroma (Sterculiaceae), Tilia (Til- iaceae), and Bombax (Bombacaeae) are members of a malvalean clade (represented only by Gos- Akania (Akaniaceae) and Bretschneidera (Bretschneideraceae), both Sap- indales, are members of the mustard-oil clade. Three additional members of Capparaceae, Cleome, Koe- sypium in Search 1). berlinia, and Setchellanthus, along with Cappa- ris, create a polyphyletic Capparaceae (Fig. 10B; see also Rodman et al., 3, this issue). Hyp- seocharis (Oxalidaceae) is sister to a clade con- taining many Geraniaceae (Fig. 10B; see Price & Palmer, 1993, this issue). Among members of rosid I (Fig. 11B, C) are representatives of the following new families: Reinwardtia (Linaceae) is sister to Viola (Violaceae); Sargentodoxa (Sargentodoxa- ceae) is imbedded in Fabaceae; Humulus (Can- nabinaceae), 7rema (Ulmaceae), and Boehmeria (Urticaceae) fall into an urticalean clade with Mo- rus and Ficus (both Moraceae), but Ulmaceae are paraphyletic to Cannabaceae; Coriaria (Coriaria- caeae), Begonia (Begoniaceae), and three genera of Cucurbitaceae are placed with Datiscaceae; Bet- ula (Betulaceae) and Carya (Juglandaceae) are members of the clade containing Fagaceae and Casuarinaceae; Mouriri and Osbeckia (Melasto- mataceae), Punica (Punicaceae), Trapa (Trapa- Volume 80, Number 3 1993 Chase et al. 537 Phylogenetics of Seed Plants ceae), and Heteropyxis (Myrtaceae) are members of a clade of (mostly Myrtales) that includes On- agraceae and Combretaceae (Fig. 1 1C). In asterid lineages, two additional families have been added: Byblis (Byblidaceae) and Vahlia (Saxifragaceae) are placed among Schrophulariales (Fig. 15B) Changes in placements of major clades. At higher levels among the seed plants, Search II produced relatively few **major" differences in tax- on placements from the topology of Search I. The conifers in some, but not all, 3,900 trees are mono- phyletic. Piperales-Aristolochiales were sister to all monosulcate angiosperms (Figs. 2B, 4B). The "paleoherb" clade of Search I (Fig. 4A) was thus split in two with the portion containing Chloran- thaceae, Illiciales, Austrobaileyaceae, Amborella- ceae, and Nymphaeales situated as sister to Mag- noliales. Although these may seem to represent major shifts, extremely short branches separate these clades (Fig. 4B); thus neither topology has 1993, this issue). Within monocots, only shifts among the groups in much internal support (Qiu et al., Aranae, Alismatanae, and Lilianae occurred (Fig. 5B): Aranae (minus Acorus) are sister to Alisma- tanae; the clade containing Pandanaceae, Cyclan- thaceae, and Velloziaceae is isolated and no longer sister to Liliales, and Dioscoreales are sister to the commelinoid taxa (the position of Asparagales in the trees from Search I). Among commelinoid taxa (Fig. 6B), Stegolepis (Rapateaceae) is sister to Bromeliaceae, Typhales are sister to Juncaceae- Cyperaceae, and Flagellaria is sister to the other graminoid clade. Among eudicots, three major shifts of taxa pres- ent in Search I occurred. The first series of re- arrangements involves the heterogeneous asterid V clade (Fig. 12A). Gunnera (Gunneraceae) in- tercalates as an isolated lineage (Fig. 7B) between Trochodendrales (plus Pachysandra of Buxaceae) and higher eudicots (asterids, caryophyllids, and rosids). Santalales, Phoradendron (Viscaceae), Schoepfia (Olacaceae), and Osyris (Santalaceae) become sister to Caryophyllidae-Droseraceae- Nepenthaceae (Fig. 9B), and this larger caryophyl- lid clade is shifted from a position within the rosid clade (rosid III; Figs. 2A, 9A) to sister to the larger asterid-rosid clade (Figs. 2B, 9B). The remaining member of asterid V, Paeonia (Paeoniaceae), is deeply imbedded within rosid III clade (Fig. 8B) as sister to Crassulaceae. We attach little signifi- cance to these shifts of position; internal branches of these groups are among the shortest supporting positions of major clades. The second series of shifts occurs within rosids. Families of Myrtales (plus Qualea of Vochysiaceae, Polygalales) and the clade containing Viviania, Wendtia (both Geraniaceae), Greyia (Greyiaceae), and Francoa (Saxifragaceae) are shifted from rosid II to rosid I (Fig. 11B, C), and relationships of intermediate-level clades (containing several fam- etc.) are somewhat modified from their position in Search I. ilies, i.e., those of Urticales, Fagales, The third shift involves the two taxa that were sister to the expanded caryophyllid clade from Search I (Fig. 9A): Dillenia (Dilleniaceae) and Vitis (Vitaceae), which become sister to the rest of the larger asterid clade (Fig. 128; Vitis and Dillenia in a sense exchanged positions with Santalales). Again, these groups have short internal branches, and these shifts would require little change of over- all parsimony. In several instances, the topology of Search П (which we favor because it was a more complete search) is more similar to that found in investi- gations of restricted nature (for example, outgroup relationships of Asteraceae are identical with those found by Michaels et al., 1993, this issue, and Olmstead et al., 1992, whereas those of Search I deviated in several ways). DISCUSSION Although rbcL sequences for several groups of spore-bearing plants are available (true mosses, hornworts, liverworts, Equisetum, Isoetes, Lyco- podium, Psilotum, and both eu- and leptosporangi- ate ferns), their use as outgroups is complicated by extensive sequence divergence relative to that in the seed-plant ingroup. No other extant lineages of land plants are likely to have shared a common ancestor with seed plants for well over 350 million years, and a great deal of sequence change, much of it in the form of multiple, unrecoverable sub- stitutions, has eraron. biis of these other land plant t shown, but see Hamby & Zim. 1992, үн similar re- sults) that are radically different from all previous hypotheses of relationships (e.g., Bremer et al., 1987). In contrast, seed-plant relationships pre- sented here are at least congruent in gross aspect with comparable morphological studies (Crane, 1985, 1988; Doyle & Donoghue, 1986, 1992; Loconte & Stevenson, 1991). Addition of highly sequence-divergent outgroups could be expected to increase ingroup homoplasy with unpredictable 1978). analyses, we have chosen to use the more conser- topological results (Felsenstein, In these vative approach of an unrooted ingroup analysis 538 Annals of the Missouri Botanical Garden of seed plants, which are almost certainly mono- phyletic (Doyle & Donoghue, 1986, 1992) ikewise, effects of missing groups upon topol- ogies may be profound (Donoghue et al., 1989 and unpredictable, and we expect the absence of ~ the numerous extinct lineages of early land plants in molecular data matrices to pose potentially se- rious problems for elucidating relationships of ex- tant lineages. Perhaps more conservative genes sampled for more of the sequence variation present within extant groups may be able to “bridge” gaps caused by extinction. Features of genome organi- zation (such as gene and intron content and gene order; Downie & Palmer, 1992; Raubeson & Jan- sen, 1992) may offer more robust hypotheses than gene sequences for such questions, but these are likely to be too few to provide a fully resolved tree by themselves. A great deal more experimentation with combined data sets of morphological and mo- lecular characters is obviously needed, and a num- ber of these studies are underway (both with rbcL and rRNA/rDNA data). Consideration of these problems here is premature. Sister-group status of angiosperms and Gnetales (Figs. 2A, B and ЗА, B) is corroborated by other cladistic studies (Crane, 1985, 1988; Doyle & Donoghue, 1986, 1992; Loconte & Stevenson, 1991). The isolated position of Ceratophyllum as sister to all other angiosperms has been argued 1991). In studies this issue) in which non- previously (Les, 1988; Les et al., by Qiu et al. (1993, flowering seed plants were removed, this arrange- ment was made equivocal by the existence of an- other equally parsimonious island in which Cera- tophyllum occurred in a radically different position. mby & Zimmer (1992; rRNA) also found a yet different placement for Ceratophyllum (but we suspect that the sparser sampling of their study may be responsible for most of the differences from those found with rbcL). In instances in which a taxon's morphology and anatomy are as divergent and potentially modified as those of Ceratophyl- lum, its position becomes difficult to address ade- quately in cladistic studies. Ceratophyllum has been absent from many morphological cladistic studies, such as those of Doyle & Donoghue so corroboration is currently precluded. he general groupings of angiosperms (exclusive of Ceratophyllum) identified in these two analyses are highly similar to each other and to those of most recent taxonomic schemes, particularly those of Dahlgren (1980), Dahlgren et al. (1985), and Thorne (1992; this last has admittedly idque results of several molecular investigations). Fur- thermore, results of rbcS (Martin & Dowd, 1991) and ribosomal studies (Hamby & Zimmer, 1992) are quite similar to ours as well. How these larger groupings (clusters of families and in some cases orders) are inter-related is the point at which their similarity diverges. Although having quite different implications for angiosperm origins and evolution, the preferred hypothesis in one of these studies is not vastly different in relative parsimony from those favored in other investigations. For example, with mostly morphological data, Doyle & Donoghue 1986) discovered a “paleoherb rooting” step less parsimonious. Constraining a paleoherb rooting for angiosperms, at Nymphaeales (as in Hamby & Zimmer, 1992), was also only slightly less parsimonious in the subset of angiosperm rbcL sequences studied by Qiu et al. (1993, this issue). en examined to address basal angiosperm re- at one ~ lationships, all these data appear to lack a strong historical signal. A number of phenomena have been suggested to be capable of confounding molecular phyloge- netic studies. We consider below several of these factors and examine their potential to affect studies of rbcL sequence variation and then address some additional concerns about future directions of mo- lecular systematic study. EFFECTS OF PARALLEL NUCLEOTIDE SUBSTITUTIONS IN INDEPENDENT LINEAGES An effect of unrecoverable (due to extinction) or unobserved (due to insufficient sampling) char- acter-state changes is the introduction of spurious similarities, which may result in treatment of in- dependently derived nucleotides at a given base position as homologous (see Albert & Mishler, 1992). Such mistaken interpretations of indepen- dent events can lead to "branch attractions" if an analysis includes а great number of such assess- ments (Felsenstein, 1978). dequate taxon sam- pling, sometimes referred to as appropriate "taxon density," is one means of reducing potentially in- accurate assessments of similarity, but determining at what point sampling is sufficient has so far only been addressed in an a posteriori manner. e improvement afforded to assessments of character-state change by increased taxon sam- pling is counterbalanced by a decrease in com- putational speed and ability to ascertain how near results are to maximum parsimony. Intrafamilial studies are not as likely to be affected severely by these problems because, in general, numbers of taxa are not as great and evenness of sampling is better. At higher taxonomic levels within seed plants, а paradoxical impediment to progress arises: if taxon number is great enough to assess character- Volume 80, Number 3 1993 Chase et al 539 Phylogenetics of Seed Plants state changes accurately, then one reasonably can expect to reach only suboptimal phylogenetic so- lutions, but if taxon number is restricted enough to gain confidence of the maximum parsimony of the trees found, spurious assessments of character- state identity could obscure all but the closest re- lationships. To evaluate relationships of Plumbaginaceae and Polygonaceae to families of Caryophyllales, Gian- nasi et al. (1992) used a number of phenetic and "phylogenetic" methods, including a Fitch-Mar- goliash dendrogram based upon genetic distances (Kimura, 1981; Felsenstein, 1990), maximum par- simony (PHYLIP, Felsenstein, 1990; PAUP, Swof- ford, 1991), and maximum likelihood (ML; Fel- senstein, 1981). All three methods produced the same lack of resolution concerning relationships of these families; to state that Plumbaginaceae and Polygonaceae are “not closely related” (meaning "closely similar") to Caryophyllales does not pre- clude them nonetheless from being closest relatives. From the perspective of results presented here (Fig. 9A, B), spurious similarities in the data analyzed by Giannasi et al. (1992) appear to affect equally results of all three tree-building methods: Gossyp- ium simultaneously attracts higher asterids and magnoliids (in our trees Gossypium is well imbed- ded among rosids; rosid II, Fig. 10A, B). (See Olmstead et al., 1992, 1993, this issue, for an example and discussion of the effects of taxon sampling. Prospects for improvements of tree-building methods with greater numbers of taxa exist (Penny et al., 1992). In the example of Giannasi et al. (1992), none of the methods employed succeeded in eliminating what we interpret as branch attrac- tions due to the small number of taxa sampled and use of distantly related outgroups. Phylogenetic studies using morphological characters for a closely related group of rosid or asterid families would likely be affected adversely by outgroups of mag- noliids or monocots. This phenomenon is perhaps even more probable with nucleotide data in which homology is initially assessed only by nucleotide position and character states are restricted to the same four alternatives. Character “homology” for distantly related taxa can be easily determined with rbcL data (i.e., a given nucleotide position in this size-conserved gene; "primary homology," de Pinna, 1991), but this does not mean that assess- ments of character-state homology (synapomor- phy) are less subject to homoplasy than with other data. If phylogenetic analyses of nucleotide data can- not be expected to reveal relationships when sam- pled with manageable numbers of taxa (from the standpoint of computation of minimal trees) and if sampling with sufficient numbers of taxa precludes assessing the parsimony of results, then we have reached an impasse until improved methods of anal- ysis are developed. An appreciation of this problem has led us to be skeptical of the overall topologies presented (Figs. 1, 2), and competing ideas of re- lationships should not be overlooked when per- forming more restricted analyses of these and other gene sequences. INTERNAL SUPPORT FOR THE BROAD TOPOLOGY: AN AD HOC ANALYSIS OF SUBCLASS HAMAMELIDAE For purposes of suggesting an appropriate man- ner to examine internal support of families tradi- tionally recognized as a natural group, we selected one of the most controversial subclasses, Hama- melidae sensu Cronquist (1981), for which we have data from 18 of 24 families. Numerous phyloge- netic and systematic studies of the Hamamelidae have been completed (see various authors in Crane & Blackmore, 1989). The morphological features suggesting a close relationship among these families are largely those associated with the temperate amentiferous syndrome, and these clearly could be the result of parallel modification in unrelated lin- eages. In their developmental characteristics and wood anatomy, families of Hamamelidae are par- ticularly heterogeneous (Cronquist, 1981; Crane & Blackmore, 1989). In performing the broad analysis, we sought to avoid a priori ideas about what constituted monophyletic subgroupings of an- giosperms, but for this heuristic example we have accepted ad hoc the outgroup relationships found in the general study. We selected 72 species that included all mem- bers of Hamamelidae and their immediate sister taxa as identified in trees from both searches. No attempt was made to select species that would reproduce the particulars of the general topologies. A tree search under the Fitch (equal weights) cri- terion using 2,000 random sequence additions, ULPARS, STEEPEST DESCENT, and NNI branch-swapping (but permitting only 10 trees to be held at each step) found only one island at maximum parsimony (i.e., all trees could be found by single branch swaps using any one of them as a starting tree; cf. Maddison, 1991). Additional as well as shorter islands could still exist but are un- likely after 2,000 repetitions (this type of search required about 24 hours to complete on a Mac- intosh Quadra 950 with 20 MB of RAM). After random addition searches were completed, the trees 540 Annals of the Missouri Botanical Garden found were used as starting points in a single anal- ysis with MULPARS on and TBR swapping to completion; this process should find all equally par- simonious trees in the single island identified. Our search found 36 equally parsimonious Fitch trees, three of which were optimal under the weighting criterion of Albert et al. (1993, Materials and Methods). One of these was randomly selected and is shown with Fitch branch lengths (ACCTRAN optimization; Fig. 16). The tree length was 2,234 steps, the C. I. was 0.288 (excluding unique characters), and the R. I. was 0.532. The maximum parsimony trees also were used to search this issue; see for trees up to five steps less parsimonious under the Fitch criterion; the FILTER TREES option was used to identify trees at each length, and a strict consensus tree at each step was computed. The number of steps less parsimonious at which each topological component decayed was recorded (Fig. 16; “decay values” are shown below the branches: "dO" indicates that the branch is a polytomy in the strict consensus of the maximum parsimony trees, **d1" indicates that the branch is a polytomy in the consensus tree at one step less parsimonious, etc.). This analysis identified eight om into which Hamamelidae fell (Fig. 16): Еисот- mia, clade A, sister to Aucuba (Doracene) and Garrya (Garryaceae) and nested within an asterid clade; Hamamelidaceae, Cercidiphyllaceae, etc., members of clade B, in a series paraphyletic to Saxifragaceae- Grossulariaceae; Fagales, etc., clade C, sister to Fabaceae-Polygalaceae; Urticales, clade D, sister to Rosaceae; Leitneria, clade E, nested within fam- ilies of Sapindales; isolated Trochodendrales, clade F, sister to Buxaceae; Platanus, clade G, situated in a heterogeneous group; and the last, Kuptelea, clade H, situated among Ranunculales. Many of these clades are well supported internally, decaying at three or more steps less parsimonious (some, such as Trochodendrales-Buxaceae and the Plat- anus assemblage are weakly supported as mono- phyletic lineages but clearly are not members of other well supported groups, leaving them in iso- lated positions apart from other Hamamelidae). Results of this restricted analysis are congruent with the "РО found in the broad searches (Figs. combine morphological data with this molecular matrix and perform constraint experiments in which various rearrangements of these taxa are examined for their relati of parsimony. The general conclusion from B example is that the Hama- melidae do not form a monophyletic lineage; they are shown to be grossly polyphyletic. Besides the amentiferous syndrome, the major trait of Ham- amelidae is the presence of tannins, which is like- wise compatible with a relationship to the other tannin-containing families, Fabaceae, Rosaceae, Saxifragaceae sensu stricto, and Crassulaceae. Many authors (e.g., in Crane & Blackmore, 1989) have discussed these families in terms of which are "lower" and which are “higher” families. This distinction finds some support from these results; the “lower” groups either stand in an isolated po- sition near the base of the eudicots (Eupteleaceae, Platanaceae, and Tetracentraceae- Trochodendra- ceae) or basal within rosids (rosid IV, Fig. 2A; Daphniphyllaceae, Hamamelidaceae, and Cercidi- phyllaceae), whereas most “higher” hamamelids (Casuarinaceae, Fagaceae, Moracae, Ulmaceae, and Urticaceae) demonstrate a well supported relation- ship to Fabaceae or Козасеае (d > 5 on two branches at the base of the largest clade in Fig. . The position of Leitneria in Sapindales near Burseraceae is corroborated by a shared suite of secondary compounds and presence of intercellular resin canals. The example presented above is not intended to be more than a superficial phylogenetic treatment of families traditionally referred to Hamamelidae. It is meant to serve as an example of how the general topology, which itself is suspected of being suboptimal and presently cannot be examined by decay analysis because of its size, may identify a relevant analysis within which questions of opti- mality and relative support can be addressed. We are pleased that general relationships found in the broad analyses hold up well when addressed in this and other more restricted investigations, none of which have found vastly different topologies. ON THE INFORMATIVENESS OF ALL SUBSTITUTIONS The majority of character-state changes in pro- tein-coding genes have been demonstrated to occur at third positions within codons, and numerous empirical studies have shown third position substi- tutions to be more abundant in this and other data sets. Some workers have experimented with dis- carding third position substitutions from their anal- yses or analyzing nucleotide sequences inferred from amino acid data onymous substitutions; Martin & Dowd, 1991). In several studies (Conti et al., 1993, this issue; Don- oghue et al., 1992; Kim et al., 1992; Smith et al., 1993, this issue), all three codon positions have been found to exhibit similar levels of homoplasy which standardizes all syn- (and perhaps similar rates of change per site as Volume 80, Number 3 1993 Chase et al. 541 Phylogenetics of Seed Plants well). It has generally been assumed that, for most genes at some unspecified higher taxonomic level, third positions become saturated and tlierefore are more likely to be uninformative or even misleading (Swofford & Olsen, 1990), hence the “logic” for discarding third positions and using only nonsynon- ymous substitutions. In the empirical studies cited above, multiple substitutions apparently have oc- curred at certain first and second positions at low taxonomic levels (i.e., within families). Variable third positions appear as consistent on average as variable first or second positions, thus making them as reliable at reconstructing relationships (as a class perhaps they are better because they are more numerous; Donoghue et al., 1992). Whether factors such as codon usage (tRNA biases known to favor only a subset of the possible third position nucleotides for a given amino acid) might contribute to the relatively higher consisten- cies of some synonymous substitutions is a topic that should be investigated. Codon usage is usually studied in a pairwise fashion, but it would be better examined from a phylogenetic perspective. Most molecular models, including the one (a modified version of that of Kimura, 1980) that is the basis for the weighting scheme used here (Albert et al., 1993), are arguably too simplistic in their as- sumptions concerning patterns of nucleotide sub- stitutions. Ultimately, we could imagine that knowl- edge of processes, specific to rbcL, responsible for the higher consistencies of some synonymous sub- stitutions might permit construction of alternative weighting models that would extract more historical signal from rbcL (Albert et al., 1992a). Until we better understand evolutionary constraints on all positions within a coding sequence, it seems most prudent to use methods that do not eliminate any evidence of the underlying process. BRANCH LENGTH INEQUALITIES AND UNEQUAL RATES The fossil record suggests an early, and perhaps rapid, diversification of angiosperms (e.g., Doyle & Hickey, 1976). Several different categories of specialized flowers appeared nearly simultaneously in geological terms. These included strobiloid types characteristic of Magnoliales, simpler forms similar to those of Laurales, highly reduced sorts like those of Piperaceae and Chloranthaceae, and catkin- bearing specimens comparable to those of lower hamamelids (i.e., eudicots like Platanaceae; Schwarzwalder & Dilcher, 1991). Most of the seg- ments are short in the basal portion of the rbcL trees (but not the branches of monocots or eudicots; Fig. 2A, B), which could be interpreted as support for hypotheses that production of diverse flower types took place rapidly and early in angiosperm evolution. Nonetheless, other explanations could explain patterns of uneven distribution of character-state changes identified by this and many other analyses of molecular data (e.g., Jansen & Palmer, 198 These include episodic changes of rates of sequence divergence and effects resulting from an uneven distribution of sampled taxa that could cause cer- tain character-state changes to be assigned to sev- eral more terminal branches rather than to a single, more internal one. The latter effect could be ex- plained if subsequent substitutions over the long history of groups obscured or even eliminated ev- idence of synapomorphic character states for these groups. Greater sampling of variation within ter- minal groups may result in character states being optimized to a single, more basal branch, with re- versals and further changes in terminal lineages, as opposed to being optimized to appear as inde- pendently arising in several terminal groups. Sev- eral other factors, including the specific method of optimization and interpretation of the robustness of groups with “weak” character-state support, are considerations relevant to the distribution of ho- moplasy. This is another area in which future em- phasis should be placed because advances (sup- rted b theoretical considerations) in understanding the effects of character-state opti- mization could improve evaluation of critical inner- most branches. Lineage-specific rate asymmetry, a potential contributor to spurious branch attractions (Hendy & Penny, 1989; Albert et al., 1993, this issue), is significant for rbcL (Bousquet et al., 1992), yet may not be extensive enough between lineages to be problematic. Initial estimates of total sequence кеше rate per year produced a range of 5- r (i) panicoid versus pooid grasses (Doebley et al., 1990), (ii) Petunia versus Tobacco, and (iii) Colchicum versus Lilium (Wendel & Al- bert, 1992, from data presented in Albert et al., 19922). Subsequent studies of woody magnoliid taxa with austral disjunction patterns (and for which plate tectonics suggest appropriate di have indicated divergence за ae between 4-5 x 107" and 1.4-1.7 x 1079 (У. A. Albert, unpublished; note that all rates ел are for lineage pairs, and so the average rate for each lineage is half the figure shown). Only the lowest of these rates is approximately an order of mag- nitude different from the herbaceous taxon com- parisons. Wilson et al. (1990) indicated slightly slower rates for palms. Recent work on monocots times) 542 Annals of the Missouri Botanical Garden using relative rate tests (Gaut et al., 1992) dem- onstrated five-fold variation for rbcL, which still falls within the range cited above. Although 5-10- fold differences in rates would appear potentially to contribute enormous branch-length differences, the small overall rates involved mean that great divergence times would be a more important factor. In addition, differences in rates appear to be highly lineage-correlated (for example, in graminoids; Gaut 1992) rather than random, and if more extensive samplin possible in these “fast” clades, then effects of rate differences can be offset. et al., at lower taxonomic levels is Most if not all rbcL rates will probably exist within a relatively narrow “window,” perhaps ар- por the range illustrated above (107'"– 107", V. A. Albert, M. W. Chase & J. F. Wendel, vnpablishell. If that is correct, lineage-specific rate inequalities are unlikely to be a primary factor in branch attraction; rather, asymmetrical divergence times would be implicated because the product of rate and time is the central parameter in consid- erations of potential systematic errors (see Albert et al., 1993, this issue). Thus uneven sampling or extinction of lineages may present greater problems than do differences in rates. EFFECTS OF LATERAL GENOME TRANSFER, ANCESTRAL POLYMORPHISMS, AND DIFFERENT MODES OF INHERITANCE The trees presented here represent only infor- mation from a single gene, and factors peculiar to its evolution could lead to erroneous results. Several of these phenomena are discussed below, but we feel that their impact is likely minimal. Genome transfers would result in all or parts of a genome being phylogenetically coherent (transferred as a unit) at the time transit occurs (Doyle, 1992; Riese- berg & Soltis, 1991), but before and after move- ment most characters within genomes evolve in- dependently (although still linked if on the same chromosome) and ought to be expected to contain historical evidence. Hybridization is unlikely to in- fluence phylogenetic analyses except at lower tax- onomic levels. Even matings between divergent parents still occur within portions of families (and usually among closely related species in a genus rather than between families. Parental taxa at the time of an ancient hybridization also were likely closely related, and in these genomes highly con- served loci, such as rbcL, would have been similar ~ or even identical. Ancestral polymorphisms (Pamilo & Nei, 1988; Wu, 1991; Doyle, 1992) also affect only closely related taxa, and, like the effects of ancient hy- bridization, those of ancient polymorphisms would likely be minor relative to subsequent genetic di- vergence. If lineages that contain polymorphisms diverged in a closely spaced manner and uneven sorting did take place, it is unlikely that any evi- dence of such variation within a progenitor could be identified as such over the great amounts of time involved in this study. Furthermore, роју- morphisms are short-lived and are undocumented for conservative, single-copy loci, such as rbcL. At most these effects would be highly localized among groups of closely related terminal taxa and, with adequate taxon sampling, would not be ex- pected to perturb greatly the results. Lateral transfers not involving exchange of ga- metes (by unknown mechanisms) may have oc- curred between major lineages in the past (i.e., of rbcL from a purple bacterium to a red algal an- 1992). Such transfers pres- ently appear rare among land plants and seed plants cestor; Morden et al., in particular; furthermore, artificial transforma- tions are relatively difficult and often have a de- stabilizing or transient effect on transformed plants. While we must admit that this is an unknown area that could have played a role in certain anomalous placements in the rbcL trees (e.g., Montinia and Vahlia of Saxifragaceae sensu lato among asterids; Fig. 15), at the same time we are not prepared to advocate it as a scenario until trees based on other data demonstrate a pattern consistent with such hypotheses (as in the example of Morden et al., 1992). Using evidence from studies of secondary chemistry and development, Morgan & Soltis (1993, this issue) build strong cases for the highly dispersed groupings found in their study of Saxi- fragaceae sensu lato. Furthermore, many of their findings were also congruent with recent investi- gations of non-molecular characters. Rather than resorting to explanations involving lateral transfers of chloroplasts by mechanisms about which we can only speculate, we would prefer first to examine specific cases from a cladistic perspective rather than from that of current taxonomic schemes. n rbcL tree is not solely a maternal tree. Al- though chloroplast transmission is principally ma- ternal and uniparental, several groups exhibit a paternal (e.g., conifers) or biparental pattern (Neale et al., 1986; Szmidt et al., 1987; Whatley, 1982; Wagner et al., 1987; Corriveau & Coleman, 1988; White, 1990; Owens & Morris, 1991). Others that have been thought most probably maternal, such as Liriodendron and Magnolia (Corriveau & Cole- man, 1988), consistently exhibit 5-15% paternal 1993). The potential inheritance (Sewell et al., Volume 80, Number 3 1993 Chase et al. 543 ie of Seed Plants effects of a mixed pattern of inheritance on gene trees are also unlikely to affect studies at interfam- ilial levels and above. GENES VERSUS TAXA One of the more persistent controversies sur- rounding molecular phylogenetic studies has cen- tered on whether results, at a given taxonomic level, will be “improved” more by adding additional taxa sequenced for the same gene or by adding sequence analyses of additional genes for the same set of taxa (analyzed simultaneously or each per- formed independently for assessments of congru- ence; Pamilo & Nei, 1988; Wu, 1991). From the standpoint of corroboration, phylogenetic studies of other data sets are absolutely crucial. Never- theless, it seems quite clear from our work on these data sets that ideas of relationships have changed considerably as more taxa have been added. We suspect that no single gene sequence can provide reasonable hypotheses of relationships of seed plants when sampled uncritically and superficially, but perhaps erroneous results from one locus would be "corrected" by stronger signals present in the oth- ers. Data to evaluate this most critical question do not exist: would separately analyzing 20 gene se- quences for the same set of 25 to 30 taxa produce well supported relationships? We are convinced that adding representatives of the still numerous and diverse families absent from this analysis has obvious computational complications). A great deal of the variation present within the gene still remains to be sampled taxonomically. Studies in several groups of plants indicate that rbcL often can be valuable at rather low taxonomic levels (Albert, in press, in the slipper orchids; S. Graham, B. Morton & 5. Barrett, unpublished, in Eichhornia; R. Price & J. Palmer, unpublished, in Pelargonium; S. Williams & M. Chase, submitted, in Drosera; Xi- ang et al., 1993, this issue, in Cornus). What can be concluded from these studies is that although it is absolutely critical that more genes be studied to provide corroboration, it is unlikely that any of these studies will make a contribution to under- standing seed-plant relationships if the level of sam- pling is too sparse. Only empirical studies will ul- .. 9° timately resolve the versus taxa controversy. We estimate that by the time this paper is in print, more than 1,200 sequences of rbcL from seed plants will exist, and future studies will undoubtedly benefit from this enormous da- tabase. An additional implication of adding more taxa is that the results presented here may not be stable or reflective of those of an analysis of two, three, or four times as many sequences. We point to two regions of the general topology that are morpho- logically quite heterogeneous and that we believe are possibly generated by undersampling: hama- melid II (Figs. 2A, B, 7A, B) and rosid I (Figs. 2A, B, 11A, B, C). In the former, the critical taxa needed to provide more appropriate relationships could be some of the still-unsampled families, but these groups (represented by Lambertia, Nelumbo, Platanus, and Sabia) could just as likely represent the isolated relicts of now largely extinct lineages. In the case of rosid I, only a small percentage of the families has been sampled (for instance, only 7 of 24 families in the Violales: Dilleniidae), and it is in such a case that we might predict “curious” Topological instability often occurs in cases where long terminal branches sister-group relationships. are next to short internal branches, that is, hy- pothesized relationships are drastically altered by the addition of related taxa that "break up” long branches. Two surprising pairs of sister taxa from Search I, Erythroxylum-Viola and Ochna-Dry- petes, are quite divergent and connected to each other by relatively short branches (Fig. 1 1A; branch lengths shown only in 118). In Search II, the addition of Reinwardtia displaced Erythroxylum from Viola to Drypetes, leaving Ochna to stand isolated from any other taxon (Fig. 1 1B). In such situations, assessments of relationships are difficult, ut, as more closely related species are adde distinguishing synapomorphies for families і groups of families as distinct from autapomorphies for individual species will become more reliable. We would not argue that the relationships found for rosid I are “better” in Search I than in Search II (or vice versa), but rather point these out as areas of the trees that require additional sampling and in which many of the relationships suggested by these analyses of rbcL have little or no mor- phological support. VALUE OF THE BROAD ANALYSIS We view the relatively robust internal support found by evaluations of portions of the broad anal- ysis as an indication that rbcL sequences contain information relevant to the evolutionary history of angiosperms (Fig. 16, on Hamamelidae and most of the other papers in this issue; in particular see Conti et al., 1993, and Rodman et al., 1993, for which comparable cladistic analyses of non-molec- ular data are also available and compared). Well- 544 Annals of the Missouri Botanical Garden characterized families and groups of families (as evidenced in several more restricted studies of mor- phology, anatomy, and secondary chemistry; e.g., Rodman, 1991a, b; Hufford, 1992) are largely congruent with our results. Other evidence sup- ports the monophyly of the angiosperms (Doyle & Donoghue, 1986, 1992), monocots (Dahlgren et al., 1985), and eudicots (Donoghue & Doyle, 1989). Likewise, the position of Ceratophyllum as sister to the rest of the angiosperms compares favorably with the fossil record (Les, 1988; fossil fruits from 120 million years ago are identical to extant fruits, D. Dilcher, pers. comm.). The status of the families of the Papaverales-Ranunculales collectively as sister to the rest of the eudicots (Figs. 1, 2) also finds ample external support (Donoghue & Doyle, 1989, among others). The results reported here do find some sur- prising" relationships. Several families, such as Chenopodiaceae (Fig. 9A, B) and Berberidaceae with only two representatives each in our studies, are paraphyletic to other families. These results do not surprise us because sampling has been dem- onstrated repeatedly to be a major factor, and insufficient or uneven sampling can generate anom- alous relationships. Concomitantly, our results em- phasize the need for studies of familial limits to include much better taxon sampling than is gen- erally the case in these two analyses. Some of these instances of paraphyly may be accurate; many pairs of temperate, herbaceous/woody, tropical families have long been suspected of being unnat- ural (for example, Lamiaceae are derived within Verbenaceae and Brassicaceae within Capparaceae in our studies). ther “major” findings of this study, although discordant when viewed from the perspective of various taxonomic treatments, find support from For example, the placement of Ericales as sister to asterids (Figs. 1, 2) was also found by Hufford (1992), and the monophyly of most mustard-oil recent studies of non-molecular characters. families (Fig. 10A, B) was previously suggested by Rodman (1991b). Other sets of relationships are unique to this analysis and require more thorough morphological and molecular studies. These in- clude: (1) the position of several families with poly- petalous corollas and supposed affinities to Saxi- fragaceae among each major lineage of Asteridae (requiring a hypothesized reversal of the sympet- alous condition in Escallonia, Montinia, Phyllono- ma, and Vahlia (all Saxifragaceae sensu lato); (ii — nesting of аШеппа orders Capparales, Malvales, Theales, and Violales among rosid clades (the sys- tems of both Dahlgren, 1980, and Thorne, 1992, treated these groups in a manner somewhat similar to our topology); (iii) relationships of Nepenthaceae and Droseraceae to Caryophyllidae sensu lato (Fig. 9A, B); and (iv) specific associations of numerous problematic genera like Dillenia (among rosids near asterids, Fig. 12B; see Olmstead et al., issue), Impatiens (near members of Ebenales and 13A, B), and Nelumbo and Lam- Fig. ТА, В). Placement of most of these genera and families Ericales; Fig. bertia (among lower hamamelids; has varied substantially among recently proposed classifications (although no one has suggested the relationships found here), and their positions in this study will undoubtedly add to the controversies. Because additional lineages were present in Search II and different methods were used to con- struct the trees, it is impossible to evaluate whether the topology found in Search II represents a dif- ferent island of trees. Indeed, multiple islands of equally parsimonious trees were found in other studies in this issue: Morgan & Soltis (1993), Olm- stead et al. (1993), and Qiu et al. (1993). Certainly shifts of some taxa, especially Paeonia from a basal asterid to sister of Crassulaceae and Saxifragaceae sensu stricto, suggest a radically different expla- nation of the distribution of at least some char- acters. Paeonia is well supported internally in its new position (its sister status to Aibes does not decay even at five steps less parsimonious; Fig. 16). Shifts of Dillenia, Gunnera, Santalales, and Vitis seem, at first glance, to be major alterations of position, but branches are so short near the split between asterids and rosids that these could not involve many additional steps in either topology = ig. Although the trees of Search II are preferred to those of Search I because some of them were swapped on to completion and therefore are more likely to represent at least a local optimum, tax- onomic conclusions based on either search are un- timely. When faced with the fact that large num- bers of angiosperm families are still unrepresented in the rbcL data set, we would argue that a valid assessment of the most appropriate positions of many taxa, such as Dillenia, Gunnera, and Vitis, grossly premature. Some conclusions, polyphyly of Hamamelidae and Dilleniidae, for example, seem well supported now. This study is noteworthy not only for its scope but also for the large number of contributors whose unpublished sequences made up the bulk of the data analyzed. This wide collaboration was advan- tageous to all workers; many found that taxa se- quenced by other laboratories supposedly working Volume 80, Number 3 1993 Chase et al. Phylogenetics of Seed Plants 545 on distantly related groups (in recent taxonomic schemes) fell into or near their group of interest. Prime examples of this are the close phylogenetic relationship of Nepenthaceae and Droseraceae to families of Caryophyllidae, Pittosporaceae to Api- aceae-Araliaceae, Malvales and Capparales (both ““dilleniids””) to Sapindales, and Corokia (Corna- ceae) to Asteraceae. Broad analyses are thus important in providing evaluations of a priori assumptions about appro- priate sets of study taxa for more focused and rigorous studies; they should be formalized so that someone takes the initiative to perform them. Gov- ernmental funding agencies should facilitate stud- ies, such as the one presented here, that are well beyond the scope of individual laboratories. This study demonstrates the potential of this kind of analysis, but no single individual or laboratory could have received formal support for sampling this diverse set of taxa and performing the phylogenetic analysis; it would have been deemed by reviewers too broad and too unfocused. Although extramural funds supported most of the other studies in this issue, no support was received specifically for the broad analysis. It is the investment in individual studies that justifies a further expenditure to sup- port syntheses that supply an essential overall per- spective, even though they may be necessarily approximate. The benefits of performing this largest-yet phy- logenetic study of seed plants lie not only in support of specific relationships hypothesized by other stud- ies (e.g., the sister group status of the Ericales and higher asterids proposed by Hufford, 1992) and in identification of previously unhypothesized monophyletic groups. This study also presents a comprehensive, explicit hypothesis for higher level relationships, permitting and encouraging initiation of studies evaluating other character systems that may show congruence with the major lineages of seed plants described here. We have performed none of the essential experiments that these results suggest (topological constraints, removal of char- acters, combining morphological with molecular data, etc.) and have not developed implications these topologies may have for specific character transformations in seed plants or molecular evo- lution of rbcL or RuBisCO. Believing that serious consideration of the significance of these general topologies is best handled at the more manifest evel of other papers published here (and else- where), we have chosen instead to emphasize rea- sons for caution. We are content to present the findings of these studies as examples of potential benefits and pitfalls of such exercises. At the least, these analyses represent an attempt to improve both our understanding of seed-plant evolution and methods of phylogenetic inference. Corroboration by other data sets analyzed in a similar fashion is by far the most significant measure of relationships proposed here, and we hope this process of eval- uation will be innervated by our efforts. LITERATURE CITED ALBERT, V. A. Phylogeny of the slipper orchids (Cypri- pedioideae: Orchidaceae) from congruent morpho- logical and molecular data sets. Nordic J. Bot. (in press). & B. D. MISHLER. 1992. On the rationale and utility of weighting nucleotide sequence data. Cladis- tics 8: 73- _ —————, M. VW. СнАЗЕ & B. D. MisHLER. 1993. acter-state civi for cladistic analysis of protein- coding DNA sequences. Ann. Missouri Bot. Gard 80: 752-7 . MISHLER & M. W. CHase. 1992a. Char- acter- state weighting for restriction site data in phy- logenetic reconstruction, with an example from chlo- roplast DNA. Pp. | i Soltis & J. J. Doyle (editors), Molecular Systematics in Plants. Chapman & Hall, New York. WiLLiams & M. W. m 1992b. Carnivorous plants: Phylogeny and structural evo- lution. Science 257: 1491-1495. ARCHIE, J. W. A randomization test for phylo- к. information in systematic data. Syst. Zool. 38: 252. а т. ‚5. Н. Srrauss, А. D. DOERKSEN & R. PRICE. 1992. Extensive variation in е rate of rbcL gene sequences among seed plants. Proc. 8. BREMER, K. i The limits of amino acid sequence data in angiosperm phylogenetic reconstruction. Evo- lution 42: 795-803. 90. = aaa component consensus. Cla- disties 6: 369-3 CJ: кеМ B. D. MISHLER & S. P. CHUR- . 1987. On cladistic relationships in green plants. Taxon 36: 339-349. Cowri, E., A. FiscHBACH & К. J. Syrsma. 1993. Tribal relationships i in Onagraceae: Implications from rbcL sequence data. Ann. Missouri Bot. Gard. 80: 672- 85 CORRIVEAU, J. L. & A. W. Coleman. 1988. Rapi screening method to detect potential biparental in- heritance of plastid DNA and results for over 200 angiosperm species. Amer. J. Bot. 75: 1443-1458. CRANE, P. R. 1985. Phylogenetic analysis of seed plants Ann . Missouri Bot. F T 8. Major clades e relationships i in the “higher” gymnosperms. Pp. 218-272 in C. B. Beck (editor), Origin and Evolution ы Gymnosperms. Со- lumbia Univ. Press, New York. 1989. Evolution, pate m CRONQUIST, А. An Integrated Sreten of Classi- tion of Flowering Plants. Columbia Univ. Press, New York. 546 Annals of the Missouri Botanical Garden DAHLGREN, R. T. . A revised system of classifi- cation of the angiosperms. Bot. J. Linn. Soc. 80: 91-124. & K. Bre 1985. Ки clades of the angiosperms. Сайына, 1: 349- iH DP, o. 19 The Families of the Monoa am ан Сы, Бой and Taxonomy. Springer- e ork. Dou: 1, M. DunBiN, E. M. Gor LM T. CLEGG D. 1990. dbi is Hard sis of the large subunit of carboxylase (rbcL) nucleotide se- quence among the grasses (Gramineae). Evolution 108. 4: 10 DONOGHUE, M. J. & J. A. DOYLE. studies of seed plan morphological characters. Pp. 181-193 holm, K. Bremer & H. Jornvall (editors), The Hi- erarchy of Life: Molecules and Morphology in Phy- logenetic Analysis. Elsevier Science Publishers, Amsterdam , R. С. OLMSTEAD, J. F. SMITH & J. D. PALMER. 1992. Phylogenetic relationships of Dipsacales based on rbcL sequences. Ann. Missouri Bot. Gard. 79 333-345. ‚ J. A. DOYLE, J. GAUTHIER, А. C. KLUGE & T. Rowe. 1989. The importance of fossils in phylog- eny reconstruction. Ann. Rev. Ecol. Syst. 20: 431- 460 1 989. ое Бозум, S. R. € J. D. PALMER. 1992. Use of chlo- roplast DNA rearrangements in reconstructing plant phylogeny. Pp. 14-35 in P. S. Soltis, D. E. Soltis & J. J. Doyle (editors), Molecular Systematics in E 1986. Seed plant phylogeny and the o origin of the angiosperms: An experimental cladistic approach. Bot. Rev. 52: 321- 431. & —— ———. 1992. Fossils and seed plant phy- logeny reanalyzed. Brittonia 44: 89-106. . Hickey. 1976. Pollen and leaves from the mid Cretaceous Potomac Group and their bearing on early angiosperm evolution. Pp. 139-206 in B. Beck (editor), Origin and Early Evolution of An- giosperms. Columbia Uni ork. Рот, J. J. Gene trees and species trees: Mo- lecular systematics as one-character taxonomy. Syst. Bot. 17: 144-163. & Hotton. 1991. Diversification of ul an- dede pollen in a cladistic contex 169-195 n S. Blackmore & S. H. Barnes (editors) "Pollen and d. ко Clarendon Press, Oxfor T. DUVALL, LEGG, M. x Chase, W. D. CLARK, W. J. KRESS, Se M Hints, L. E. EGUIARTE, J. F.S н, B. S. Gat E A. Zimmer & G H LEARN. 1993. Phylogenetic pu е w nn. Missouri qus Gard. 80: a “él 19. Рата. D. & P. Cranston. 1991. d a cladogram oul this short s arisen by chance sone Cladistics 7: 1-28. FELSENSTEIN, J. . Cases in which parsimony o compatibility rem will be а dex aed Syst. Zool. 27: 410. — 81. нан trees from ОМА sequenc- es: А maximum likelihood approach. J. Molec. Evol. 17: 368-376. 1985. Confidence limits on phylogenies: An approach using the bootstrap. Evolution 39: 783- PHYLIP manual, version 3.3. Uni | California, University Herbarium, Bees Cal вене Ww M. 1971. Toward defining the course of evolution: Minimum i change for a specific tree to- 16. SE, W. D. CLARK & M. T. CLEGG. Relative rates of nucleotide substitution at the rbcL locus of monocotyledonous plants. J. Molec. Evol. 35: 292-303. GIANNASI, D. E., С. ZURAWSKI, С. LEARN & M. T. CLEGG. 1992. Evolutionary relationships of the Caryophyl- lidae cs on comparative rbcL sequences. Syst. Bot. 17: 1-15. ба и сн & L. M. McCook. 1992. Documenting scientific data: The need for voucher specimens. Ann. M . 79: 96 : 1992. Ribosomal RNA asa phylogenetic tool in plant systematics. Pp. 50- 9] in P. S. Soltis, D. E. Soltis & J. J. Doyle (editors), Molecular Systematics in Plants. Chapman & Hall, New York. Harris, Р. J. & R. D. HARTLEY. 1980. Phenolic con- stituents of cell walls of monocotyledons. Biochem Ecol. 8: 153-160 & D. PENNY. 1989. A framework for the quantitative study of evolutionary trees. Syst. Zool. 38: 297-309. Hurronp, L. . Rosidae and their relationships to ai er nonmagnoliid e A phylogenetic naly sis chemical data. Ann. Missouri Bot. Gard. 79: 218-248. JANSEN, R. K. & J. D. PALMER. 1988. Phylogenetic ee of chloro last DNA restriction site vari- atio mes un isieae шл. Amer. J. Bot. 15: 751. KALLERSIÓ, "i , А. С. KLUGE & C. Bur 1992. Skewness and иса ы ы Cladistics 8: 275- 287 Km, KJ. R. K. Jansen, R. S. WALLACE, H. J. MICHAELS & J. D. PALMER. 1992. Phylogenetic implications of rbcL sequence dro in be Asteraceae, Ann. Missouri Bot. Gard. Kimura, M. A mea met oe xl for estimating evolutionary rate e of base substitutions through com- parative studies of nucleotide sequences. J. Molec. Evol. 16: 111-120. . Estimation of evolutionary distances between homologous nucleotide sequences. Proc. Natl. Acad. U.S.A. 78: 454-458 Kron, К. A. & M. W. Cuase. 1993. Systematics of the Ericaceae, Empetraceae, Epacridaceae an lated taxa based upon rbcL sequence data. пен 1, The origin and affinities of the Cer- atophyllaceae. Taxon 37: 326-345 ‚ D. К. Garvin & C. F. WiMPEE. 1991. Mo- jecular evolutionary history of ancient A an- giosperms. Proc. Natl. Acad. U.S.A. 10119- 0123. Госомте, Н. & D. W. STEVENSON. 1991. the Magnoliidae. Cladistics 7: 267-296. MappisoN, D. R. 1991. Discovery and importance of multiple islands of most-parsimonious trees. Syst. Zoo 28. Cladistics of Volume 80, Number 3 1993 Chase Rod of Seed Plants 547 MARTIN, P. С. & J. M. Dowb. 1991. Studies of an- giosperm phylogenies using protein sequences. Ann Missouri Bot. Gar : 296-337. MICHAELS, H. J., К. M. Scorr, R. С. OLMSTEAD, T. SZAR¢ R. K. JANSEN & J. D. PALMER. 1993. Interfamilia relationships of the Asteraceae: Insights from rbcL sequence variation. Ann. Missouri Bot. Gar 742-765. MORDEN, C. W., C. F. DELWICHE, M. KunseL & J. D. PALMER. 1992. Gene phylogenies and the endo- symbiotic origin of pos т e ae 8: 75-90. MoncAN, D. & D. E. Sor 1993. Phylogenetic relationships among loa of Saxifragaceae sensu lato based on rbcL sequence data. Ann. Missouri Bot. Gard. 80: 631-660. NEaLE, D. B., N. C. WHEELER & К. W. ALLARD. 1986. Paternal inheritance of chloroplast DNA in Douglas- d. J. Forest Res. 16: 1152-1154. NIXON, K.C.8 IS Polymorphic taxa, missing values, and cladistic analyses. Cladistics 7: 233-241. OLMSTEAD, R. G., H. J. MichaELs, К. M. Scorr & J. D. PALMER. 1992. Monophyly of the Asteridae and identification of their major lineages inferred from A sequences of rbcL. Ann. Missouri Bot. Gar 79: 249-265. —— ——, B. Bremer, К. M. Scorr & J. D. PALMER. 1993. A parsimony analysis of the Asteridae sensu lato based on je ее Ann. Missouri Bot. Gard. 80: 700- 199]. Cytological basis for cytoplasmic inheritance in Pseudotsuga menzie- sii. II. о and proembryo development. Amer. J. Kd y? PALMER, J. D., P ee H. J. MicHaELs, M. W. CHASE & J. M. MaNHART. 1988. Analysis of chlo- roplast DNA variation. Ann. Missouri Bot. Gard. 7 1180-1206. Рампо, P. & M. Ме. 1988. Relationships between gene trees and species trees. Molec. Biol. Evol. 5 -583. PENNY, D., М. р. HENDY & M. A. STEEL. 1992. Prog- i trees. Trends Ecol. Evol. Biol. 7: -19. 19 DE Pinna, M. C. С. ie and tests of ho- mology in the cladistic жо Cladistics 7: 367- 394. PLATNICK, N. I., C. E. GRISWOLD & J. A. CODDINGTON. 1991. On missing entries in cladistic analyses. Cla- distics 7: 337-343. Price, R. & PALMER. 1993. Phylogenetic relationships of the Geraniaceae and Geraniales from rbcL s vili comparisons. Ann. Missouri Вог. Gard. 80: 66 | н ‚ D. Н. Les & С. R. Par 1993. Molecular phylogenetic of the Magnolia Cladistic analyses o e plas- b^ dene rbcL. Ann. Missouri Bot. Gard. "80: 587- L. A. & R. К. Jansen. 1992. Chloroplast А evidence оп the ancient evolutionary split in nar land plants. Science 255 eee 1699. Rerric, J. H., Н. D. Witson & J. M. MANI ART. 1992. Phylogeny of ше Caryophyllales— Gene sequence data. Taxon 41: 201-209. RIESEBERG, L. Н. 4 р. Е. em mu ЈВЕЗОМ, 1991. Phylogenetic consequences of түе gene flow in plants. Evol. Trends Pl. 5: 65- RITLAND, К. € M. T. с Тш Evolutionary anal- yses of plant DNA sequences. Amer. Naturalist 130: 574-5100. КОРМАХ, J. 1991a. А taxonomic analysis of glucosi- nolate-producing plants, Part 1: Phenetics. Syst. Bot. 16: 598-618. —— ——. 199]b. A taxonomic analysis of p 6: A producing plants, Part 2: Cladistics. Syst. Bot 61 , К. PRICE, К. KAROL, E. Conti, К. SYTSMA & J. D. insu 1993. Nucleotide sequences of the rbcL gene indicate monophyly of mustard oil plants. Ann. Missouri Bot. Gard. 80: 686-699. ScHWARZWALDER, R. SEWELL, M., M. W. CHASE & C. R. PARKS. 1993. Trace paternal inheritance of chloroplast DNA in Liriodendron and Magnolia (Magnoliaceae). Amer. J. Bot. 80 (in press). SMITH, J. F., W. J. Kress & E. A. ZIMMER. 1993. Phylogenetic analysis of the Zingiberales based on rbcL sequences. Ann. Missouri Bot. Gard. 80: 620- 30. 6 Sorris, D. E., P. S. Sorris, M. T. Стесс & M. Dur 1990 rbcL sequence d s and abes. genetic relationships in Saxifragaceae sensu lato. Proc. Natl. Acad. U.S.A. 87: 4640-4644 Sworronp, D. L. 1991. juro Phylogenetic Analysis Using Parsimony, Versio . Computer program distributed by the Illinois Natural History Survey, Champaign, Illinois. . J. OLSEN. 1990. Phylogeny reconstruc- tion. Pp. 411-501 in D. M. Hillis & C. Moritz (editors), Molecular eri Sinauer Associates, Sunderland, Massachus SZMIDT, A. E., ALDEN & К -E. HALLGREN. 1987 Paternal inheritance of ни DNA in Larix. Pl. Molec. Biol. 9: 59-6 TAKHTAJAN, A. ae. of the classification of flowering plants (Magnoliophyta). Bot. Rev. 46: 225- 359. 1987. Systema Magnoliophytorum. Nauka, Leningrad. 5 Da Tavrom, D. У. € L. J. Hic KEY. 1992. Phylogenetic evidence for iom rear origin of angiosperms. Pl. Syst. Evol. 180: 137-156. THORNE, R. F. Рина new realignments in the angiosperms. Nordic J. Bot -117 3 2. An updated caca of the flow- ering plants. Aliso 13: 365-38 WAGNER ., G. R. FURNIER, а ү SAG комен, S. M. Wiliams, D. В. DAN K & R. LLARD. 1987. Chloroplast DNA ae aes in [уез and jack pines and their hybrids. Proc. Natl. Acad. U.S.A. 84: 2097- А V. А. ALBERT. 1992. Phylogenetics m): Character-state weighted parsimony analysis of chloroplast-DNA re- ic implications. Syst. sll . 1982. ен рака ез ls i ез itance: Green algae to angiosperms. Bot. Rev 5 527- WHITE, E. E. 1990. Chloroplast DNA in Pinus mon- 548 Annals of the Missouri Botanical Garden ticola. 2. Survey of within-species variability and detection of heteroplasmic individuals. Theoret. Appl. = en. 79: 251-255. WiLson, M. A., В. Gaut & M. T. CLE 1990. Chlo- roplast DNA di slowly in the рап family. Molec. Biol. Evol. -314. Younc, D. J. er Sason. 1970. The А о dicotyledons: A study of the m levels of hi- erarchy. Austral. J. Bot. 18: 387- Wu, C.-L. 1991. Inference of iis 8 GM ny in relation to segregation of ancient polymorphisms. Genetics 127: 429- 435. XIANG, Q.-Y., D. E. Sorris, D. R. MORGAN & P. S. Sorris. 1993. Phylogenetic relationships of Cornus L. sensu lato and putative relatives inferred from rbcL se- тз data. Ann. Missouri Bot. Gard. 80: 723- 34. ZURAWSKI, G. T. Стесс. 1987. Evolution of higher-plant chloroplast DNA-coded genes: Implica- tions for structure-function and ин “ум Ann. Rev. Pl. Phys. 38: 391- > Department of Biology, University of North Carolina, Chapel Hill, North Carolina 27599-3280, U.S.A. Current address for MWC: Laboratory of Molecular Systematics, partment of Biology, Wake Forest University, Winston Salem, North Carolina 27109, U.S.A. Current address for VAA: Institutionen For me "n Uppsala a Uppsala S-751 21, Sweder artment of Botany, Washington State University, Pullman. Washington 99164-4238, U.S.A. ' Department of E. P. O. Biology, University of Col- orado, Boulder, Colorado eda 0334, U.S.A. i nces, Tverai ity of Wis consin, Milwaukee, Wisconsin 53201, U.S.A. боны ору and Evolution- ary Biology, EUER AN of Connecticut, Storrs, Connect- icut 06269-3042, U.S • Department of Es Duke University, Durham, North Carolina 27708-0338, U.S.A. Current address: Depa rtment of Integrative Biology, U "dies of Califor- nia, нард M 94720, U.S ' Depar of Botany and Plant = rag iced of California. Riverside. California 92521-0124, U.S.A. Current address for LEE: Centro de Ecología, Universidad acional Autonóma de México, Apartado Postal 70-275, University, Raleigh, North Carolina 27695, U.S.A. Cur- rent address for MRD: Office of Biotechnology, Iowa State University, Ames, Iowa 50011-3260, U.S.A. epartment of Biology, Indiana University, Bloo- mington, Indiana 47405, U.S.A. Current address for RAP: Department of Botany, University of Georgia, Athens, Georgia 30602, U.S.A ° Department of Bio logy, Texas A & M University, Coilege Station, Texas 77843, U.S.A. Current Address for JHR: Department of Biology, College of the Ozarks, Point Lookout, Missouri 65726, U.S.A. ment of Botany, University of Wisconsin, Madison, Wisconsin 53706, U.S.A '! Department of Biological Sciences, Bowling Green Ра University, Bowling Green, Ohio 43403, U.S.A. ? Department of Rigs Smithsonian Institution, Washington, D.C. 20560, U.S.A. Current address for JFS: Department of Biology, Boise State анод Boise, Idaho 83725, U.S.A '* Department of Botany, Arizona State University, Tempe, Arizona 85287, U.S.A. ‘Institutionen For Systematisk Botanik, Uppsala Universitet, Uppsala S-751 21, Sweden. ^ Department of Botany, University of Texas, Austin, Texas 78713, U. " Department of Forest Resources and Plant reese University of Minnesota, St. Paul, Minnesota 5510 eUS, x ' Department of Forest Science, FSL 020, Oregon State averse, Corvallis, Oregon 97331- 7505, U.S.A. ment of ss: o Santa Ana Botanic Garden, Claremont, си 91711. 3101, ЈА. epartment of Bio ology, Lebanon a? College, Annville, Pennsylvania 17003-0501, U.S.A. ? School of Biological Sciences, Waveney of New South Wales, P.O. B ет ox 1, Kensington, New South Wales, artment of Biological Sciences, Wayne State Uni- versity, Detroit, Michigan 48202, U.S.A * Department of Botany, = of Toronto, To- ronto, DUREE M58 382, Can * Department of Biology, са University, Boston, Ma dea 02215, U.S.A. 550 Annals of the Missouri Botanical Garden Asteridae -- including py Ericales, Primulales, < Ebenales, Santalales, Apiales, Cornales, and Rosidae -- including Violales, Malvales, Г mustard-oil families, | higher Hamamelidae, and Caryophyllidae hamamelids 4 ranunculids 4 paleoherbs 4 Laurales G Magnoliales pcb du fo gnetophytes Pinaceae other conifers ads FiGURE 1. Summaries of the major clades identified in: (A) the combinable component consensus tree of 500 equally parsimonious trees found for 475 taxa using the character-state weighting method of Albert et al. (1993, this issue); and (B) the strict consensus tree of 3,900 equally parsimonious trees for 499 taxa found using the Fitch (even weights) criterion. These are ingroup networks arranged arbitrarily with the cycads sister to all other seed plants. Volume 80, Number 3 1993 Chase et al Phylogenetics of Seed Plants 1B Asteridae -- including Ericales, Primulales, Ebenales, Santalales, Apiales, Cornales, and some Rosales Rosidae -- including Violales, Malvales, mustard-oil families, and higher Hamamelidae 4 caryophyllids Gunneraceae hamamelids 4 ranunculids paleoherbs II Magnoliales Laurales monocots paleoherbs I Ceratophyllaceae 552 Annals of the Missouri Botanical Garden за asterid I asterid II asterid III asterid IV asterid V rosid IV rosid III Е гоѕіа П rosid I Е — — — —— hamamelid П hamamelid I ranunculids paleoherbs qt monocots 7] Laurales Magnoliales Ceratophyllum Gnetales Pinaceae other conifers cycads FIGURE 2. Summaries of the same topologies as in Figure 1. In B, Fitch branch lengths are optimized from a single tree; optimization on consensus trees is like ly © особа оние branch lengths. Names of the specific clades identified do not conform to the composition of families used in most taxonomic schemes, but rather are designated Names of each clade correspond to groups shown in Figures 3-15. Clades marked with a “$” in B are those that i A. Volume 80, Number 3 Chase et al. 553 1993 Phylogenetics of Seed Plants 2B asterid I asterid II asterid III asterid IV asterid V 9 2. rosid I 5 qa rosid II 5 = rosid III caryophyllids$ Gunnera% hamamelid II hamamelid I ranunculids 22 sr paleoherb ms Magnoliales monocots m Laurales 39 paleoherb I | Ceratophyllum 18 el Gnetales other conifers L 22 Pinaceae l 23 Ginkgo’ cycads Jun 11 24 554 Annals of the Missouri Botanical Garden 3A | angiosperms Gnetum Gnetaceae Lr Welwitschia Welwitschiaceae Ephedra Ephedraceae [ Abies —— — Keteleeria Pseudotsuga Ш Larix E Picea sitchensis | Picea pungens E = Pinus griffithii Pinus radiata Cedrus E Pseudolarix Tsuga Callitris Widdringtonia __ Taxodium Metasequoia Taxodiaceae | Sequoiadendron | Pinaceae Cupressaceae Podocarpus Podocarpaceae Cycas . Cycadacae PT Microcycas} Zamia Chigua a Bowenia Macrozamia | ап Encephalartos си pes Y = Ceratozamia ros m Stangeria ros I [ Lepidozamiaf ман | — Dioon M pal Рај FIGURE 3. А portion о the overall analysis showing the gymnosperms." (Numbers above the branches in B are the numbers of substitutions iei onto one tree randomly selected from the 3,900 saved in Search II.) Note that A is the consensus tree of whereas B is a single tree with branches not present in the strict consensus of Search II marked by an arrow. с пега marked with a "t" in А were omitted from Search II; genera marked with an asterisk in B were not available for Search I. Volume 80, Number 3 Chase et al. 1993 Phylogenetics of Seed Plants 555 3B | angiosperms 2 ap und E Gnetaceae = c Welwitschia | Welwitschiaceae y - Ephedra Ephedraceae — Sciadopitys* Taxodiaceae 6 — Taxus* Taxaceae 28 Cephalotaxus* Cephalotaxaceae 26 Callitris | Widdringtonia Cupressaceae Metasequoia B е Taxodiaceae 8 Taxodium _| 65 Podocarpus Podocarpaceae l Abies E 5 15. Keteleeria 4 Pseudolarix Pinaceae METRI ESAE Ginkgoacae Cycadacae Zamiaceae 4A eudicots Euryale mend а Nymphaea Nymphaeaceae Nuphar Barclaya Barclayaceae* = ређа = Cabombaceae" > Amborella Amborellaceae“ © Schisandra Schisandraceae! = Illicium Illiciaceae" d. БЕ Austrobaileya Austrobaileyaceae"| Y Chloranthus Chloranthaceae* Dt = B Заги | Aristolochiaceae * Asa — E Houttuyn "T Saururaceae? Saurur Ex pad Piperaceae 4 = monocots | Hernandiaceae E Hedyearsa Monimiaceae“ iosperm Idiospermaceae“ * збив M = ydp fridust Calycanthaceae“ s Chimonant Ф Cinnamomum — | Lauraceae® | Persea mE ee Canella Canellaceae” Ly ince Winteraceae” ae Magnolia macrophylla? | Magnolia salicifolia Michelia Manglietia = Talauma singaporensis | Magnoliaceae" an Talauma ovatat 3 LI Magnolia tripetalat Р. Magnolia hypoleuca > Liriodendron TERI o Degeneria Degeneriaceae” = Це Аппопа b ME Annonaceae Asimina Galbulimima Himantandraceae” Eupomatia Eupomatia сеае” Кпета Myristicaceae” __ Ceratophyllum Cerat tophyllaceae* | ast I ast II ast III ast IV ast V ded "Magnoliidae, Nyphaeales еп "Magnoliidae, Magnoliales ham II "Magnoliidae, Laurales ran g Magnoliidae, Piperales aem "Magnoliidae, Aristolochiales = e ‘Magnoliidae, Illiciales FicURE 4. Basal portion of the overall analysis showing the positions of е wr армен poen ots єк нен pollen), eudicots (dicots with triaperturate pollen n), and the three cla ‘prim " dico espoir pollen). Note that, exclusive of Ceratophyllum, the angiosperms form two sister ps са by le general pollen aperture number (one versus three). (Numbers above the branches in B are the numbers of Substitutions ei ize 5 nto one tree randomly selected from the 3,900 saved in Search П.) Genera marked with a T" were omitted from Search II; genera marked with an asterisk in B were not available for Search I. The figure че the position of two сн amplified from a total cellular DNA template (see Materials & Methods). Note that А is the consensus tree of Search I, whereas B is a single tree with branches not present in the strict consensus of Search II marked by an arrow. Volume 80, Number 3 Chase et al. 557 1993 Phylogenetics of Seed Plants 4B — Nymphaeaceae * Barclayaceae” a Cabombaceae” 2 Amborellaceae“ 4 Schisandraceae" ~ Illiciaceae! c Austrobaileyaceae’| = Chloranthaceae? — Canellaceae” Winteraceae” п = dà Magnoliaceae^ - Talauma singaporens = esr Liriodendron tulipifera 57 T b= Liriodendron chinensis* | Б 1 ^ ae 7 Deg Degeneriaceae” a Hm 5= Galbulimima Himantandraceae^ ac qm Eupom Eupomatiaceae? 7 2 20—24 Annona 22 13 Asimina Annonaceae” 20 " 21 С ga " : ma Myristicaceae i 4) 13 27 Cinnamomum Lauraceae" 59 ersea == — 2 Hedycarya Monimiaceae* > 15 Idiospermum Idiospermaceae? | 2 6 | | 8 25 Calycanthus Cal th с [£g Chimonanthus punc ~ 33 li N u | » = Her = Hernandiaceae* 24 Gyrocarpus* monocots Lr, "о 11 39 Sar Aristolochiaceae* | 22 18 porum i ec 12 Y Lactoris* Lactoridaceae A e Saururus S d е z[ US Houttuynia aururaceae 39 18 Piper* Р а = 34 Рірегасеае — 44 Ререготїа — Ceratophyllum Ceratophyllaceae ast zh а ast V ros] , Magnoliidae, Nyphaeales ros Ш "Magnoliidae, Magnoliales fam m | “Magnoliidae, Laurales aln * Magnoliidae, Piperales mon E ° Magnoliidae, Aristolochiales u > ли f .. eos pal I Magnoliidae, Illiciales it one 558 Annals of the Missouri Botanical Garden 5A ns commelinoids Xanthorrhoea Xanthorrhoeaceae ! Kniphofi pp rthia Asphodelaceae ! Dracaenaceae! Danae "m uscaeae! Nolina A фи rina] Nolinacaceae! Bow yacinthaceae! Choo ie Anthericaceae J (iiv Amaryllidaceae! ris [ pr Anomatheca Iridaceae о Pain i Tecophilaeaceae! - ——[ Hypoxit- Hypoxidaceae ! — Onun SPRS Нит caceae* ium Medeola T ас Alstroemeria Паша Colchicum Burchardia Colchicaceae! Chamaelirium Melanthiaceae^ Vellozia Velloziaceaeh Freycinetia andanace роеви clanthaceaet | K— — losco Dioscoreaceaee Tacca Тассасеае“ | letris — Melanthiaceae? Burmanniat Burmanniaceae4 = Sagittaria pu Pi oin ifolia Alismataceae* мырын mogetonaceae* ve P Melamine b i À— [ ES Len iphyllum la : astI mnostachys гасеве ast II ast III ast M NIV ,Aran ros Ш Liliana, ! Рени FIGURE 5. the 3,900 saved in Search II.) Genera marked with a arch I. Note that А is the consensus tree of Search I, whereas B is a single an asterisk in B were not available for Se *Alism Lilianne E Burmanniales *Lilianae, Dioscoreales na ria asias 'Lilianae, Lilia JLilianae, ма e tree randomly tree with branches not present in the strict consensus of Search II marked by an arrow. The five basalmost lineages of the monocots, composed of the aroids, alismatids, and lilioid groups. Volume 80, Number 3 Chase et al. 1993 Phylogenetics of Seed Plants 559 8 ; a SB 19 Г Cypripedium* Р 6 = Neuwiedia Orchidaceae Oncidium ra* Dasypogonace Hemerocallis* Hemerocalidaceae Chlorophytum Anthericac | Xanthorrhoea шла ' iphofia | | Asphodelaceae' |Hyacinthaceae f cont yl Ruscaceae ' Nolinaceae i . Amaryllidaceae' Iridaceae? Tecophilaeaceae | Hypoxidaceae : 4 - commelinoids 27 они Dioscoreaccae" 3 y 2 Тассасеае“ ^ 13 14 16 Pandanus* Pandanaceae " i 10 Freycinetia " 20 Sphaeradenia Cyclanthaceae 30 Vellozia Velloziaceae‘ 7 Smilacaceae* а Calochortaceae 15 Liliaceae? 12 Colchicaceae? Alstroemeriaceae? EE Melanthiaceae“ 18 == 30 8 Sagittaria graminea " M Sagittaria latifolia | Alismataceae 44 20 Alisma 24 56 — b 1 13 Potamogeton Potamogetonaceae 23 41 Spathiphyllum aa? 15 1759 Pistia* 23 Lemna* Lemnaceae" 45 Gymnostachys | Araceae? Acorus* ast I ast II ast III ast IV a ast V Aranae ros I ren : Alismatanae gun ,Lilianae, Melanthiales ham I Lilianae, Liliales pal II *Lilianae, Dioscoreales mag f . . A Bromelianae, Velloziales y ECyclanthanae + on "Pandananae = 'Lilianae, Asparagales 560 Annals of the Missouri Botanical Garden 6A Avena Puccinella $ Triticum Hordeum} Aegilops Poaceae! Cenchrus Pennisetumt Neurachne Togi — Elegia Restionaceae! Juncus j Oxychloe Juncaceae C dk j Cyperaceae Prionium Juncaceae! Flagellaria Flagellariaceae! T ha . Shaceae? tegolepis E Кара(еасеае! Glomeropiteairnia t Bromeliaceae® Tillandsia Pu Hechtiat Ter duda pallida Tradescantia zebrina Commelinaceae! Tradescantia soconuscana| Pontederia Pontederiaceae* PRU um Philydraceae4 Anigozanthos aemodoraceaec Raven nd Strelitz | Strelitziaceae> | Phenakospermum Calathea Marantaceae^ Hed um =>] — Riedelia inol b cio hee Zingiberaceae Glo ч Со — T. apeinocheilos Costaceae? Mus ^ Musaceae’ Orchidantha Lowiaceae^ Heliconia Heliconiaceae> Caryotat иш Le xs DN AM Arecaceae* ast I Chamaedorea ast II Calamus ł ast III Serena st IV " m ast V Arecanae os IV D. ros III Zingiberanae y “Bromelianae, Haemodorales Damn a dBromelianae, Philydrales e а à Bromelianae, Pontederiales р. e Commelinanae, Commelinales li lau EBromelianae, Bromeliales cer "Bromelianae, Typhales ЕЕ iCommelinanae, Poales соп JCommelinanae, Cyperales FIGURE 6. The terminal lineages of the monocots, composed of the palms, gingers, and commelinoids. (Numbers above the branches in B are the numbers of substitutions optimized onto one tree randomly selected from the 3,900 saved in Search П.) Genera marked with a “t” in ere in B were not available for Search I. Note that A is the consensus tree of Search I, whereas B is a single tree with branches not present in the strict consensus of Search II marked by an arrow. Volume 80, Number 3 1993 Chase et al. Phylogenetics of Seed Plants 6B Avena Е Cenchrus Poaceae! Neurachne Oryza mbusa* L| Lachnocaulon* Eriocaulaceae“ gia Restionaceae ! Flagellaria Flagellariaceae’ Juncus j Oxychloe Juncaceae Cyperus Cyperaceae | ex ionium Juncaceae j Sparganium* Sparganiaceae” Typha Typhaceae” Stegolepis Rapateaceae" в Bromeliaceae Commelinaceae! Pontederiaceae? Philydraceae? Haemodoraceae* Strelitziaceae” Marantaceae” Zingiberaceae” Costaceae” Musaceae” Lowiaceae? Heliconiaceae” 5 7 Chamaedorea 13 12 ume Arecaceae? ath L 4 Serenoa 5 | ast Y y Arecanae ros п Zingiberanae ros . car ¿Bromelianae, Haemodorales ham H , Bromelianae, Philydrales ran 08 (В romelianae, Pontederiales mag Commelinanae, Commelinales нт ,Bromelianae, Bromeliales H cer , Bromelianae, Typhales пс ЈЕ Commelinanae, Poales Ј Соттећпапае, Cyperales 562 Annals of the Missouri Botanical Garden » higher eudicots Trochodendron Trochodendraceae" | Tetracentron Tetracentraceae“ | Sabia Sabiaceae* Lambertia Proteaceae“ Nelumbo Nelumbonaceae Platanus Platanaceae” Akebia Lardizabalaceae = Mahonia . А | Berberidaceae [wey Ш "шец Caulophyllum | Ranunculus Xanthorhiza |Ranunculaceae ‘ Caltha Cocculus Menispermaceae“ — Euptelea Eupteleaceae” Dicentra Fumariaceae’ Papave l1. 4 m Papaveraceae" mE Sanguinaria spi[nounue.1 ros IV “ Magnoliidae, Papaverales ros I , Hamamelidae, Hamamelidales " Magnoliidae, Ranunculales D , Magnoliidae, Nymphaeales ce " Rosidae, Proteales con Hamamelidae, Trochodendrales FIGURE 7. Тһе basalmost lineages of the eudicots, composed of Ranunculales-Papaverales, Trochodendrales, and a aeh айша lineage (plus Gunnera in B). (Numbers above the branches in B are the numbers of substitutions optimized onto one tree randomly selected from the 3,900 saved in Search II. ) Genera marked with an asterisk in B were not available for Search I. Note that A is the consensus tree of Search I, whereas B is a single tree with branches not present in the strict consensus of Search II marked by an arrow. Volume 80, Number 3 Chase et al. 1993 Phylogenetics of Seed Plants dn - higher eudicots — Gunnera Gunneraceae” — | = Trochodendron Trochodendraceae? | Z —Tetracentron Tetracentraceae® |3 - Pachysandra* Buxaceae‘ = ој Sabia Sabiaceae“ = s | E Lambertia Proteaceae‘ = «rac Nelumbo Nelumbonaceae? |F = Platanus Platanaceae — J = Akebia | Lardizabalaceae 18 n Caulophyllum [Berberidaceac : = Ranunculus = „| Y, = Xanthorhiza | |Ranunculaceae* E Ш Caltha . . | e | “—— Cocculus Menispermaceae? |= — - Euptelea Eupteleaceae" = or Dicentra Fumariaceaé n = Papaver P И Sanguinaria apaveraceae zn ‘Magnoliidae, Papaverales reel „Hamamelidae, Hamamelidales a ‘Magnoliidae, Ranunculales ES <= "Magnoliidae, Nymphaeales | ES ‘Rosidae Proteales pal I ‘Rosidae, Euphorbiales = : , Hamamelidae, Trochodendrales ~ Rosidae, Haloragales 564 Annals of the Missouri Botanical Garden 8A Boykinia ага Saxifragaceae“ Astilbe f Heuchera Tetracarpaea Grossulariaceae“ — Myriophyllum | Haloragaceae* Penthorum Saxifragaceae" 2 Sedum Е — i Dudieya Crassulaceae* Kalanchoe Crassula — RIDES Grossulariaceae* = Пеа — | Rhodoleia Hamamelidaceae" Hamamelis __г Cercidiphyllum Cercidiphyllaceae ’ à zd rosid IV па | , Hamamelidae, Daphniphyllales mor "Hamamelidae, Hamamelidales mag ‘Rosidae, Rosales pin “Rosidae, Haloragales FIGURE 8. The basalmost lineage of rosid dicots, which includes a number of lower hamamelids. (Numbers above the branches in B are the numbers of substitutions optimized onto one tree randomly selected from the 3,900 saved in Search II.) Genera marked with a “t” in A were omitted from Search II; genera marked with an asterisk in B were not available for Search I. Note that the positions of this clade and that of the Caryophyllidae (Fig. 9A, B) differ significantly i in the results of Searches I and II (see Figs. 1, 2). Note that A is the consensus tree of Search I, whereas B is a single tree with branches not present in the strict consensus of Search II marked by an arrow. Volume 80, Number 3 Chase et al. 1993 Phylogenetics of Seed Plants 8B > me | . " La He Grossulariaceae * y [- — Boykinia i | ә Ls Heuchera Saxifragaceae * Saxifraga = Tetracarpaea Grossulariaceae * yriophyllum Haloragaceae ° Penthorum Saxifragaceae * : Paeonia Paeoniaceae ? ш Sedum Dudleya Crassulaceae * Kalanchoe Crassula E >= Ribes Grossulariaceae * z Rhodoleia Hamamelidaceae ° .DE— Cercidiphyllum Cercidiphyllaceae * ТЕ Dept Daphniphyllaceae ^ - Hamamelis ги | у — Altingia* Hamamelidaceae a шу — Liquidambar* z^ E rosid III ham I "Hamamelidae, Hamamelidales — "Hamamelidae, Daphniphyllales aw “Rosidae, Rosales gne “Dilleniidae, Dilleniales | q “Rosidae, Haloragales 566 Annals of the Missouri Botanical Garden Dillenia Dilleniaceae' Dianthus Caryophyllaceae* | Е Atriplex | | | $рїпасїа Chenopodiaceae Amaranthus Amaranthaceae‘ Basella Basellaceae* EM Portulaca Portulacaceae* Alluaudia Didiereaceae* Mollugo Molluginaceae m Phytolacca Phytolaccaceae* — Mirabilis Nyctaginaceae* E Rivina Phytolaccaceae" _ Trianthema Aizoaceae* — Stegnosperma Phytolaccaceae“ Nepenthes Nepenthaceae” ZH Plumbago Plumbaginaceae? | Rheum Polygonaceae* Drosera Droseraceae” zi Vitis Vitaceae* zy rosid III p p MN ,Rosidae, Rhamnales X Dilleniidae, Nepenthales nos ‘Caryophyllidae, Polygonales mag Caryophyllidae, Plumbaginales M ‘Caryophyllidae, Caryophyllales eye Dilleniidae, Dilleniales FIGURE 9. The lineage that includes the Caryophyllidae. (Numbers above the branches in B are the numbers of substitutions optimized onto one tree randomly selected from the 3,900 saved in Search II.) Genera marked with an asterisk in B were not available for Search I. Note that the positions of this clade and that of rosid IV (Fig. 8A, B) differ significantly in the results of Searches I and II (see Figs. 1, 2). Note that A is the consensus tree of Search I, whereas B is a single tree with branches not present in the strict consensus of Search II marked by an arrow. Volume 80, Number 3 1993 Chase et al. 567 Phylogenetics of Seed Plants 9B T 7 Dianthus Caryophyllaceae * и Atriplex __|Сп di á gm inacia enopodiaceae ш Amaranthaceae * Basella Basellaceae * 2. Portulaca Portulacaceae * Alluaudia Didiereaceae* Mollugo Molluginaceae * Phytolacca ada accaceae* Mirabilis inaceae Rivina Plist с. Trianthema о . Stegnosperma Phytolaccaceae* „~ Plumbago Plumbaginaceae ° 31 14 Rheum Polygonaceae’ - *— Nepenthes Nepenthaceae” Б „ Drosera filiformis Drosera spathulata* 2 Drosera petiolaris* — OET ? — Dionaea = Phoradendron Viscaceae* 19 = Schoepfia Olacaceae’ sı —— Osyris Santalaceae’ = caryophyllids pu , Rosidae, Santalales Dilleniidae, Nepenthales “Caryophyllidae, Polygonales Caryophyllidae, Plumbaginales ‘Caryophyllidae, Caryophyllales 568 Annals of the Missouri Botanical Garden 10A Gossypium Malvaceae? Leitneria Leitneriaceae! " Ailanthus Simaroubaceae Poncirus Rutaceae^ Я Вигѕега Burseraceae Acer Aceraceae” . LJ . h P Sapindaceae в aes ] Anacardiaceae Tropaeolum Tropaeolaceae" Brassica Brassicaceae" Capparis Capparaceae* - Reseda Resedaceae* Tovaria Tovariaceae Batis Bataceae® С Limnanthes : pen L— Floerkea Limnanthaceae* [— Carica Caricaceae' LL Moringa Moringaceae’, alea Vochysiaceae larkia ^ | Oenothera Epilobium — Lopezia Onagraceae "E Hau po Fuchsiaf — Circaea Ludwigia — — Guisquai Ens aceae“ isqualis Combretaceae deed Crossosomataceae" Pelargonium | F = буп беис мит Monsonia ast II Greyia Greyiaceae” Ш Francoa Saxif ragaceae^ ast V Viviania Geraniaceae ros IV ros III c= posid II ham 1 Rosidae, Geraniales Гршеппазе, Violales P4 PRosidae, Rosales 8Dilleniidae, Batales mag “Rosidae, Myrtales ^ Rosidae, Sapindales gme Rosidae, Polygalales і Hamamelidae, Leitneriales *Dilleniidae, Capparales JDilleniidae, Malvales FIGURE 10. One of the two “higher” rosid lineages. (Numbers above the branches in B are the numbers of substitutions optimized onto one tree randomly selected from the Ride saved in Search II.) Genera marked wi “+” in A were o i isk i i У > = ЕЕ that A is the consensus tree of Search I, whereas B is a single tree oh branches not present in the strict consensus of Search II marked by an arrow. Volume 80, Number 3 Chase et al. 1993 Phylogenetics of Seed Plants 10B Leitneriaceae ! Simaroubaceae * Rutaceae < Burseraceae < Aceraceae © Sapindaceae © Anacardiaceae * Bombacaceae ћ Tiliaceae Sterculiaceae h Malvaceae ћ 40 Г Shorea zeylanica* Dipterocarpaceae $ CL Shorea stipularis* Moringa Moringaceae 4 s Carica Caricaceae f Stanleya* А а пунила Brassicaceae * Cleome Capparaceae d Capparis Reseda Resedaceae 9 2 Tovaria Tovariaceae 9 F> Koeberlinia* Capparaceae 4 Bataceae * Limnanthaceae ? Setchellanthus* Capparaceae d Bretschneidera* Bretschneideraceae © 10 1 Akania* Akaniaceae © 25 Tropaeolum Tropaeolaceae ? pes Crossosoma Crossosomataceae b : 20| 31 Pelargonium 261 L257 Geranium Geraniaceae ? ast I Monsonia ast II 33 " " ast Ш Hypseocharis* Oxalidaceae ? а У = ros T, < rosid II т " Rosidae, Geraniales ham II -Rosidae, Rosales 5 ,Rosidae, Sapindales ee “Dilleniidae, Capparales po "Dilleniidae, Batales pe l Dilleniidae, Violales gne "Dilleniidue, Theales pin "Dilleniidae, Malvales cyc ‘Hamamelidae, Leitneriales 570 Annals of the Missouri Botanical Garden 11A Celtis Ulmaceae ! Morus Moraceae! Rhamnus Rha ea Krameria Krameriaceae* Guaiacum Zygophyllaceae™ Spiraea Photinia R b Prunus pci Geu Myrica Myrica Fagus Fagaceae Casuarina Casuarinaceae i ania tad Fagaceae Datis Datiscaceae ! CEU " Trigonia © Ch, имин 3 lanus |Chrysobalanaceae р Erythroxylum ^ Erythroxylaceae " Viola Violaceae! Ochna Ochnaceae! Drypetes Euphorbiaceae® Humiri Humiriaceae^ [—— Euphorbia Euphorbiaceae® —— кажый. ашин .. Passifloraceae! Acridocarpus Galphimia Thryallis Bunchosia Malpighiaceae“ == Мазсарта Dicella Byrsonima n= uonymus Celastraceae * Brexia Grossulariaceae Lepuropetalon Saxifragaceaeb Kirengesho mat Hydrangeaceae> Parnassia Saxifragaceae > Pisum | | 5 e Fabaceae 4 Bauhinia | г Securidaca Polygal c пашу МЕНЕ Polygala orygalaceae Cephalotus Cephalotaceae " кон Тгетапагасеае © b | Ceratopetalu m | Cunoniaceae | Eucryphia Eucryphiaceae lis Oxali d ceae* ast ast III : ast ЈУ rosid I ros | | Ros idae, Geraniales | mel -——— ° Rosidae, Rosales j Dilleniidae, Theales ham Т , Rosidae, Polygalales : Hamamelidae, Fagales pal * Rosidae, Fabales | Hamamelidae, Urticales E , Rosidae, Celastrales „Hamamelidae, Casuarinales ШР s Dilleniidae, Violales “Rosidae, Sapindales Ба osidae, Euphorbiales” Rosidae, Rhamnales a сус "Rosidae Linales ° Hamamelidae, Myricales FIGURE 11. The second “higher” rosid lineage. (Numbers above the branches in B and C, see foldout, are the numbers of on optimized onto one tree randomly selected from the 3,900 saved in Search II.) Genera marked with a “4” i were omitted from Search II; genera marked with an asterisk in B and С were not available a for Search I. The taxon labeled “Kirengoshoma’ was, subsequent to Search I, discovered to be misidentified and was пио from Search II; из identity is unknown. Note that А is the consensus tree of Search I, whereas B and ; are a single tree with branches not present in the strict consensus of Search II marked by an arrow. Volume 80, Number 3 Chase et al. 571 1993 Phylogenetics of Seed Plants 11B hi 11C А Аааа ашы | шч rosid I гуе Malpighiaceae halotus Cephalotaceae > М 19 Mascagnia 10 5 Piatyiheca Tremandraceae 9 сена auera : uphorbia Euphorbiaceae" 3 Ceratopetal Cunoniaceae b Dun Paesifioreccae" i— Eucryphia . Eucryphiaceaeb umiria 15 Oxalis . 5 тома ж то ан 12 А» errh oa* — Oxalidaceae icania enothera hrysobalanus | rad Clarkia + ll inaceae° т: Violaceae* Ep ilobium аын Erythroxylaceae ° Circaea Onagraceae* rypetes Euphorbiaceae" pezia da Celastraceae™ auva Grossulariaceae * ya Saxifragaceae? Ludwigia — g | Lythrum Lythraceae* Sargentodoxaceae Trapa* Ча ipi: є | іса* Punicaceae < Fabaceae Heteropyxis* Myrtaceae. | UM Vochysiaceae d Polygalaceae Osbeckia* Melastomataceae € Cannabaceae? Quisqualis __| Чнпасеае * 1 p i ж Combretaceae* Urticaceae® Greyia Moraceae® Francoa Saxifragaceae b Rhamnaceae ! Viviania : = Kramer а i | 6_ Wendin Geraniaceae ygophyllaceae rosid II 7 Tetrameles* Datisca cannabina* latisca glomerata uffa* "ucurbita* is _ уљем balansae* Vothofagus dombeyi* rosid I (in part) Rosi Rosi Hamamelidae, Fagales Hamamelidae, Juglandales Hamamelidae, Casuarinales Hamamelidae, Urticales Dilleniidae, Violales Magnoliidae, Ranunculales , Rosidae, Rosales " Rosidae, Sapindales Rosi а b c d e f o 8585 — Fue - Rosaceae * asi zi as! Coriariaceae’ ast V ros I " ros III Datiscaceae * car am II hamI ran А е pal П Curcurbitaceae mag mon lau Begoniaceae* pal I Myricaceae one Betulaceae* " [ Сог Сазиагїпасеае рїп Juglandaceae cyc Fagaceae * idae, Polygalales dae, Rhamnales Hamamelidae, Myricales dae, Fabales Rosidae, Celastrales Rosidae, Euphorbiales Rosidae Linales PDilleniidae, Theales rosid I (in part) è Rosidae, Geraniales , Rosidae, Rosales „Rosidae, Myrtales “Rosidae, Polygalales 572 Annals of the Missouri Botanical Garden = higher asterids - Cornus kousa Cornus canadensis Cornus walteri Cornaceae? Cornus mast = Cornus florida — = Alangium Alangiaceae? |2 - Decumaria a M =F Hydrangeaceae ‘| fu = Philadelphus = 7 rig € Nys d Сево JN иин пе rg Araliaceae* Davidia Nyssaceae? др | Сиппега Gunneraceae* |2 Paeonia Paeoniaceae^ le Phorodendron Viscaceae’ m. Schoepfia Olacaceae” Qu .—— Osyris Santalaceae’ . | < zu ) — a re Rosidae, Santalales I _Dilleniidae, Dilleniales ur , Rosidae, Haloragales - , Rosidae, Cornales ce , Rosidae, Apiales : Rosidae, Rosales FIGURE 12. The two basalmost lineages of the general asterid clade. (Numbers above the branches in B are the numbers of substitutions optimized onto one tree randomly selected from the 3,900 saved in Search II.) Note the different composition of asterid V in Searches I and Il. Specie marked with a “t” in A were omitted from Search II. Note that A is the consensus tree of Search I, whereas B is a single tree with branches not present in the strict consensus of Search II marked by an arrow. Volume 80, Number 3 1993 Chase et al. Phylogenetics of Seed Plants 573 12B la Ie Cornus kousa Cornus canadensis Cornus walteri Cornus florida Alangium Diplopanax Camptotheca Decumaria Hydrangea Carpenteria Philadelphus higher asterids Согпасеае“ Alangiaceae? Nyssaceae? Araliaceae* _ Муззасеае“ Hydrangeaceae“ Dilleniaceae” Vitaceae ° , Rosidae, Rhamnales Dilleniidae, Dilleniales „Rosidae, Rosales , Rosidae, Cornales Rosidae, Apiales _ AI PHO9jSE А PLI9]Se 574 Annals of the Missouri Botanical Garden 13A Clavija | Theophrastageae! Anagallis Primula Ardisia qe Diospyros Ebenaceae ^ Diapensia Diapensiaceae Polemonium Polemoniaceae ћ C Symplocus Symplocaceae” Impatiens Balsaminaceae 8 Camellia Theaceae Callunat Erica | a bg Ericaceae* Elliottiat = Leioph yllum? | Сетапоја ^ |Empetraceae NAododendrom | Cassiope — ши E Zenobia А с А і Daboecia Ericaceae Геисотовт Gaultheria Vaccinium m Cyathodes? entachondra . E Leucopogonf |Epacridaceae* rhea a nmn t [^ Arbutus . |Ericaceae® Arctostaphylos? Гуго — Pyrolaceae’ nkianthus |Ericacea : Г Actinidia Andece a El e Sarraceniaceae“ ula Byblidaceae е c yril Cyrillaceae —— Styrax Styracaceae" C d Clethraceae® rL Manilkara b | Chrysophyllum Sapotaceae | Е le Fouquieriaceae ast I = z asterid III а. pow М * Dilleniidae, Violales па [у Dilleniidae, Ebenales c L] ee . ros П Dilleniidae, Ericales ros I d : ham П , Rosidae, Rosales ran. ° Dilleniidae, Nepenthales m. f Dilleniidae, Theales e „Rosidae, Geraniales = ^ Asteridae, Solanales ра i Dilleniidae, Diapensiales сус 3 Dilleniidae, Primulales FIGURE 13. The Lippe sister lineage to the clade composed of traditional asterids. (Numbers above the branches are the numbers of oo Жир on the general semi-strict consensus tree in А and one tree randomly selected from the 3,900 saved in B.) Genera marked Miss “+? in A were omitted from Search II; genera marked with an asterisk in B were not available for Search I. Note that A is the consensus tree of Search I, whereas B is a single tree with branches not present in the strict consensus is of Search II marked by an arrow. Volume 80, Number 3 Chase et al. 575 1993 Phylogenetics of Seed Plants 13B 31 js 17[ 3e Clavija Theophrastaceae ! 5 26750 Anagallis Primulaceae ! 24 Ardisia Myrsinaceae)? Los Styrax Styracaceae * 32 Clethra Clethraceae ‘ 3 [20 Symplocus Symplocaceae* = Diospyros Ebenaceae* Cassiope T Be farta Ericaceae“ Ceratiola __|Етреггасеае! 3 ica ш Rhododendron Chamaedaphne Ericaceae! 1 Gaultheria 15 Daboecia Vaccinium — . | Pentachondra : f 8 acris Epacridaceae P Pyrola — | Pyrolaceae ш = НА Епсасеае! EET Enkianthus | 21 та « |Sarraceniaceae! L 22 eliamphora* | 19 Roridula Byblidaceae” 10] тү Actinidia Actinidiaceae® 3 — [27 Camellia Theaceae* 4 ril Cyrillaceae f se [4 Матор Sapotaceae * 41 Chrysophyllum Р : ог 3s Impatiens Balsaminaceae 81756 Diapens Diapensiaceae © | 14 Polemonium Polemoniaceae ^ Fouquieria Fouquieriaceae ° ыт asterid III st III r .. • ast IV 7€ — “Dilleniidae, Violales r ir bAsteridae, Solanales ros III е ee . . car “Dilleniidae, Diapensiales ham П Rosidae, Geraniales pal 11 ‘Dilleniidae, Ebenales mon Dilleniidae, Ericales m PDilleniidae, Theales "Rosidae, Rosales "Dilleniidae, Nepenthales Јршепнаае, Primulales 576 Annals of the Missouri Botanical Garden 14A S mphoricarpos Caprifoliaceae! aleria. Valerianaceae! Dipsacus ipsacaceae Sambucus Caprifoliaceae! xa A Viburnum Caprifoliaceae! Е D: Calyceraceaeh elia Сатрапшасеае 2 Сатрапша Lo Scaevola Goodeniaceae £ Stokesiat dect Li Vernonia bea Tragopogon Cichorium Гастисат Е Сагапіаї Сасоѕтіа слон, Cartham = Dimorphotheca — Achillea Asteraceae f Chrysanthemum} Dd Делдеш E ш Tagetes Flaveriaf снн oreopsis — Eu pa sic Helianthus f r— Barnadesia = — Dasyphyllumt Е Villarsia |Menyanthaceae © Menyanthes ___| y Corokia Cornaceae* m T n |Araliaceae® ола Pittosporaceae b Coriandru БЕ ШЫ cn: dre + Apiaceaed | Sanicula Griselinia Cornaceae с Berzelia Bruniaceae b Escallonia Grossulariaceae b Æ ] огпасеае Г yllonoma Grossulariaceae b ex vomit | 2; Ч Пех et Aquifoliaceae? ast [| as ast III e MY asterid II ros Ш , Rosidae, Celastrales s „ Rosidae, Rosales ham I * Rosidae, Cornales bal Rosidae, Apiales реду "Asteridae, Solanales mag f Asteridae, Asterales e Asteridae, Campanulales con PAsteridae, Calycerales i Asteridae, Dipsacales Chase et al. 577 Volume 80, Number 3 1993 Phylogenetics of Seed Plants 14B Adoxaceae’ Caprifoliaceae a | | Valerianaceae! Dipsacaceae! Araliaceaeh 4 Apiaceaeh Pittosporaceae b ornaceae $n, Bruniaceaeb 8 [37 Boopis Calyceraceae 8 Ur Scaevo la |. . Goodeniaceaed Asteraceae ! 9 7 n E 6 Ed 8 Carthamnus Corokia Cornaceae* 14 н Vilarsia Menyanthaceae* 3 31 Lobelia d m 143. Campanula Campanulaceae ] Ст l iaceae” 7 Helwingia Cornaceae“ 13 om aceae Aquifoliaceac” ast I ып -— Рея П ast IV as a О m Rosidae, Celastrales ros е ros Ш P Rosidae, Rosales Bs * Rosidae, Cornales а d Asteridae, Campanulales meg * Asteridae, Solanales i f Asteridae, Asterales cer 5 Asteridae, Calycerales у ха . Rosidae, Apiales = ' Asteridae, Dipsacales FIGURE 14. One of the two clades of traditional Asteridae. (Numbers above the branches in B are the numbers with of substitutions optimized onto one tree randomly selected from the 3,900 saved in Search II.) Genera marke a were omitted from Search II. Note that A is the consensus tree of Search I, whereas B is a single tree with branches not present in the strict consensus of Search II marked by an arrow. 578 Annals of the Missouri Botanical Garden 15A Ni na on Solanaceae‘ Se Convulvulaceae © Montinia Grossulariaceae! ee op Boraginaceae Hydrophyllum Hydrophyllaceae * ойс ydrophy Callitriche Callitrichaceae * l [ Dista im Schrophulariacese! Nelsonia Nelsoniaceae Justicia americana usticia ‹ оёѕіеѕ = Ridtya Acanthaceae Рана jede A us Thunbergia Thunbergiaceae" " Phaulopsist ко. ellia Acanthaceae! Е Lepidagathis А ш Sesamum Pedaliaceae Verbena Verbenaceae® Teucrium Lamiaceae Clerodendrum Verbenaceae® Prostanthera — Scutellaria DM AR Lamiaceae® Physostegia Pogostemon ү, гр A — Verbenaceae? E arpagophytum | ; f Dubois гит | Pedaliaceae y” Buddleja Buddlejaceae! E Catalpa Bi потасеае! Streplocarpus Gesneriaceae C Hl Sen at Lentibulariaceae ! y” Jasminum Ol f = L- Ligustrum eaceae г Exacum T ocynum oc Asclep ias А ера iadaceae* = 5 pf pec NE Loganiaceae* en Rubiaceae? Hydroleat Hydrophyllaceae* rt ao Corna еае? аггуасеае — Eucómmia Fucoinmiaceae" T asterid I ast M ? Hamamelidae, Eucommiales ros I Rosidae, Cornales ros * Asteridae, Solanales pam M d Asteridae, Rubiales ran ales f Asteridae, Schrophulariales 8Asteridae, Lamiales hAsteridae, Callitrichales 'Rosidae, Rosales FIGURE 15. The second clade of traditional Asteridae. (Numbers above the branches іп B are the numbers of substitutions optimized onto one tree randomly a from the 3, oat saved in Search II.) Genera marked with a uo dk omitted from Search Il; genera mar with an asteri B were not available for Search I. Note that A is he consensus tree of Search I, whereas B is a single tree iin branches not present in the strict consensus of Sn II marked by an arrow. Volume 80, Number 3 1993 Chase et al. Phylogenetics of Seed Plants 579 15B la la Nicotiana . pi onere Solanaceae! etunia Convolvulus i Тротога Convulvulaceae Grossulariaceae * Boraginaceae 5 3% een Hy сое : 14 Callitriche ____ Callitrichaceae igitalis | | f 77 А ntirrh inum, | Scrophulariaceae Org Justicia = H ypoéstes = Ruttya == Barleria f 7 Lepidagathis Acanthaceae ЕЈ Ruellia 4 = Aphelandra nthus — Thunbergia Thunbergiaceae f Nelsonia Nelsoniaceae triculari. i У Pinguicula Lentibulariaceae f Streptocarpus__ Gesneriaceae E Lamium Physostegia i g Pogostemon Lamiaceae riu E аа ор Verbenaceae 5 Scu Prostanthera |Lamiaceae 8 оаа icarpa Verh а m Verbenaceae 5 esamum Еш уна Pedaliaceae f oboscidea | Buddleja Buddlejaceae f talpa Bignoniaceae Byblis * Byblidaceae ® кетиши Oleaceae f ustru 'ahlia* Saxifragaceae © стапа |Gentianaceae £ 16 Pen Rubiaceae Y з= Apocynum Apocynaceae * 3 9 іаѕ Asclepiadaceae * ige : c 19 m т, Loganiaceae 9r Aucuba Cornaceae a Piso Garryaceae P bo Eucommia Eucommiaceae * t • t ч" asterid I ži as " . ros], a Hamamelidae, Eucommiales a b Rosidae, Cornales fiam p € Asteridae, Gentianales ran y d Asteridae, Rubiales mag * Rosidae, Rosales LE f Asteridae, Schrophulariales E BAsteridae, Lamiales L pin h Asteridae, Callitrichales E ! Asteridae, Solanales Solanaceae JA 11 ace d5 rryaceae Ericacea Hydrangeaceae iai a rossulariaceae Г hyllum js траг] Tu bar formosana ы TEPES orientalis | B “ Rhodoleia E Daphniphyllum di Hamamelis Sargentodoxa Sargentodoxaceae um Medicazo Fabaceae ly Polygalaceae 4 tu En Casuarina a gus C Frigomobal onobalanus 10fagus balansae 1ofagus dombeyi riaria manos sca Datiscaceae 4 Cucurbitaceae En че опїасеае atiscaceae D : Rosaceae E Malpighiaceae _ЈЕ оозе , Acer pee \ Алга РИМ Си Сиппегасеае Т, rochodendron | Е Tetracentron Pachysandra Buxaceae Sabia — рар Lambertia Proteac Nelumbo Nelu cc 9 Platanu d3 Akebia a Lardizabalaceae on : Caulophyllum Berberidaceae Ranuncül Xanthorhiza Ranunculaceae Caltha Cocculus Menispermaceae г Euptelea | 10 Dicentra Fumariaceae 12 ras. Pap aver P "m Lo Saheuindria | apaveraceae ЕС! An example of using ix general analysis to focus a narrower study of internal support for a Ми Ваа о Numbers above the horizontal lines indicate the number of substitutions optimized (ACCT- RAN) onto one of the most parsimonious trees found using character-state weighting (i.e., these are Fitch steps, equa weighting). The numbers below the horizontal lines (preceded with a 74") are the number of steps less parsimonious А enhn | iria consensus of most-parsimonious Fitch trees are indicated by **d0," signifying that they “decay” at maximum parsim >roups of taxa often considered to be members of Hamamelidae are bracketed and lettered to designate sas independent lineages (for further explanation see text). AN ANALYSIS OF RELATIONSHIPS WITHIN THE CUPRESSACEAE SENSU STRICTO BASED ON rbcL SEQUENCES: Paul A. Gadek and Christopher J. Quinn? ABSTRACT The analysis of rbcL sequences of representative species of 13 of the 20 genera of the Cupressaceae sens has produced a cladogram that agrees with data derived from other sources goy for the recognition of a basically нон ћемо subfamily (Cupressoideae) and ut the boundary between the subfamilies needs to be realigned by o the Cupressoideae. The tribal arrangement of Li is largely artificial: the Libocedreae, The cladogram provides a southern subfamily (Callitro ideae), b Tetraclinis from the Callitroideae t upresseae, and Thujopsideae are clearly paraphyletic. . The monophyly of the family is кей moving While there is evidence from morphology, leaf and wood anatomy (Eckenwalder, 1976; Hart, 1987; Quinn & Gadek, 1988), and seed proteins (Price & Lowenstein, 1989) to support the mono- phyly of the Cupressaceae sensu stricto, the data for some of these characters, as well as for leaf biflavonoids (Gadek & Quinn, 1983, 1985), wood anatomy (Quinn et al., unpublished), and immu- nological comparisons of seed proteins (Price, pers. comm.), have not correlated well with the widely accepted intrafamilial classification of Li (1953; Table 1). Cladistic analyses of the database have shown a high level of homoplasy (Gadek & Quinn, unpublished). We therefore initiated an analysis of rbcL sequences in representative taxa in the hope of obtaining evidence of a more conservative nature that will permit estimates of relationships within the family. MATERIALS AND METHODS Species from which rbcL sequence data were obtained are listed in Tables 2 and 3 and are drawn from genera in both subfamilies and all six tribes (Table 1). Sequences will be lodged in Genbank on completion; they are also available from the au- hors. Total DNA was isolated using the СТАВ method (Doyle & Doyle, 1990). A series of synthetic oli- gonucleotide primers based on the rbcL sequences of Nicotiana tabacum and Zea mays, as well as one provided by M. W. Chase (designed by Helen Michaels and Jeff Palmer), were used for cyclic D amplification initially to produce a double stranded product containing the rbcL gene and subsequently to obtain single stranded products of both strands. The latter were sequenced by the standard dideoxy chain termination method. Se quences have been verified by sequencing both strands. The primer at the 5' end included the first 16 bases of the gene, while that at the 3' end was outside the coding portion of the gene. Hence, sequences start from the first complete codon (po- sition. 19) and terminate at the end of the stop codon (position 1428), а total length of 1,410 bp. Sequences of Glyptostrobus and Cryptomeria (unpublished, Soltis et al., pers. comm.) were in- cluded as representatives of the Taxodiaceae, the generally accepted sister group for the Cupressa- ceae sens. str. Phylogenetic analyses were per- formed with PAUP version 3.0g (Swofford, 1991) using both the default character-state weightings as well as the model of Albert et al. (1993). The default weightings give equal weighting to all char- acter states, while the Albert model invokes a tree- searching procedure that differentially weights ' We acknowledge the assistance of the following: W. D. Clark, Arizona State зы for advice on sequencin Crozier, Ching Crozier, and and S. B © [^7] ~ ~ © (еы = a m - a ~ ~ -— Š А о ЧЕ a Ox a un legg, University of California, uet for assistance with techniques in the initial stages of this work; and N. Skelton and D. Crayn for (echnical: assistance. This work is supported by Australian Res ouncil grant no. AD9031851 ? School of Biological Science, Üniversity of New South Wales, P.O. Box 1, Kensington, NSW, 2033, Australia. ANN. MISSOURI Bor. GARD. 80 : 981-580. 1998. 582 Annals of the Missouri Botanical Garden TABLE 1. Intrafamilial classification of Li (1953) for аге quite low levels of divergence between some the Cupressaceae sensu stricto. Representative species of genera indicated by * were included in this study (see Table 2) Cupressaceae sensu stricto Subfamily Mai Subfamily Callitroideae Tribe Tribe Cupres: ctinostrobeae бос Actinostrobus Chamaecyparis* Callitris keinia it Fitzroya Tribe Thujopsideae Tribe Libocedreae Platycladus* Libocedrus* Thujopsis* Papuacedrus ја» Pilgerodendron Microbiota* Austrocedrus Calocedrus* Diselma* Tribe Junipereae Widdringtonia* Juniperus* Neocallitropsis Tribe Tetraclineae Tetraclinis* transversions over transitions within characters, and codon positions 2>1>3 among characters. Output trees from PAUP were also transferred into MacClade (Maddison & Maddison, Harvard Uni- versity) and manipulated to test other topologies. RESULTS The 5' end of the sequences for several species are still incomplete, approximately 40 bases miss- ing from the start of Cupressus and Thuja, and 30 from Diselma. Percentage sequence divergence was calculated, using the available homologous se- quence data, for all taxon pairs (Table 2). There pairs of species: viz, Juniperus conferta and Cu- pressus sempervirens (0.7%), Thuja occidentalis and Thujopsis dolobrata (0.6%), Platycladus ori- entalis and Microbiota decussata (0.9%), and Di- selma archeri and Widdringtonia cedarbergensis (0.9%). Callitris rhomboidea is clearly the most divergent species, showing divergences ranging from 5.1% (with Calocedrus decurrens) to 3.6% (with Libocedrus bidwillii). The levels of sequence di- vergence among the taxa indicate that the gene has the potential to be informative about relation- ships within the family. Two trees were found by PAUP branch and bound analysis using the unweighted (default) char- acter assumption and TBR swapping option. These trees differed only in placing Calocedrus either basal to, or as part of an unresolved trichotomy involving Cupressus + Juniperus and Platycladus + Microbiota. The tree with the lowest f-value is shown in Figure 1. The Cupressaceae sens. str. are monophyletic, and there are two monophyletic clades representing northern and southern genera. Tetraclinis appears with Thuja + Thujopsis, in a clade basal to the remaining northern genera, while Chamaecyparis is basal to the unresolved tri- chotomy referred to above. Libocedrus is basal to a clade containing all the remaining southern gen- era. PAUP identifies in this data set 14 apomorphs that characterize Cupressaceae sens. str., of which 10 are unique synapomorphs. The clade nga ing the southern genera is supported by sev unique synapomorphs prp 459, 6 grs 1194, 1203, 1308, a 1363), and dica are potentially three more етос at positions Pairwise comparisons of rbcL sequences, showing percent divergence. Northern species are listed above the horizontal line, southern below 1 2 3 4 5 6 7 8 9 10 11 12 1. Microbiota decussata — 2. Tetraclinis articulata LA. = 3. Platycladus orientalis 0.9 1.8 4. Juniperus conferta 1.0 LO 12 >= 9. Cupressus sempervirens 1.0 2.1 15 07 — 6. Calocedrus decurrens 2.0 26 22 19 21 — 7. Chamaecyparis obtusa 14 22 16 13 16 23 = 8. Thuja occidentalis lo 17 15. L2 16 20 L6 — 9. Thujopsis dolobrata 11 15 13 09 14 18 14 06 — 10. Diselma archeri 2.4 2.9 2.7 25 26 31 28 25 23 — 11. Widdringtonia cedarbergensis 2.4 2.9 28 25 25 3.3 28 26 25 09 — 12. Callitris rhomboidea 4.2 3.9 45 45 48 51 45 41 40 36 36 — 13. Libocedrus bidwillii 27 31 31 27 29 33 30 26 25 20 19 58 Volume 80, Number 3 1993 Gadek & Quinn Relationships within Cupressaceae Sens. Str 583 BLE 3. Voucher specimens. All specimens are lodged in the John T. Waterhouse Herbarium (UNSW), University TAB of New South Wales Species Source/ voucher Collector / locality Callitris rhomboidea R. Br. ex L. C. Rich. UNSW21734 Gadek 13.ix.89, cult. alocedrus decurrens (Torrey) Florin UNSW 22326 Gadek 1.iv.92, cult. Chamaecyparis obtusa (Siebold & Zucc.) Endl. UNSW21735 Gadek 3.xii.91, cult. Cupressus sempervirens L. UNSW 14297 Debreczy & Racz x.80, Turkey; raised from seed. Diselma archeri Hook. f. UNSW 21742 Quinn 8.xii.91, Tasmania Juniperus conferta Parl. UNSW 14290 Gadek 13.viii.90, cult. Libocedrus plumosa (D. Don) Sarg. UNSW 21741 Quinn 6.xii.91, cult. Microbiota decussata Komar Gadek s.n. Gadek s.n. v.92, cult. Platycladus orientalis (L.) Franco UNSW21737 Gadek 3.xii.91, cult. cv. ‘Flagelliformis’ UNSW21733 Quinn 3.xii.91, cult. Tetraclinis articulata (Vahl) Mast. UNSW21730 Fernando et al. 23.viii.91, cult. Thuja occidentalis L. UNSW21732 Quinn 3.xii.91, cult. Thujopsis dolobrata (L. f.) Siebold & Zucc. UNSW21736 Gadek 3.xii.91, cult. Widdringtonia cedarbergensis Marsh UNSW17574 Quinn & Cunliffe 2 vi. 85, South Africa; raised from seed that are undetermined in a single species (33, 450, and 1227). Three additional apomorphs showing parallelisms elsewhere in the cladogram are also identified. The clade comprising the north- ern genera is supported by three apomorphs, of which two are unique synapomorphs (positions 237 and 339). There is only weak apomorphic support for other northern groupings within this clade, al- though Juniperus + Cupressus 15 characterized by three unique synapomorphs (168, 318, and 507) and Microbiota + Platycladus by one (1047). The data set was then subjected to bootstrapping (PAUP, 100 replicates), and this provided strong support for the monophyly of the southern taxa (95%) and within it the grouping of Diselma + Widdringtonia (97%). The northern taxa, includ- 75% level, and there was strong support for the clades comprising Thuja + Thujopsis (80%) and Juniperus + Cu- pressus (79%), while Microbiota clustered with Platycladus at the 63% level. The placement of Tetraclinis within this group was more equivocal: it clustered with Microbiota + Platycladus in 33% of trees and with Thuja + Thujopsis in 27%. When the Albert et al. model of character-state weightings was invoked, PAUP heuristic analysis found three trees, all one step longer than the most ing Tetraclinis, clustered at the parsimonious trees found using the equally weight- ed character option. All the Albert trees clustered Tetraclinis within the clade of Microbiota and Platycladus, and placed Calocedrus basal to Chamaecyparis (Fig. 2). The position of Chamae- cyparis is determined by a single nonsynonymous 2nd codon position change, at position 284, for which there is considerable homoplasy within the cladogram. DISCUSSION The molecular estimate of the phylogeny pre- sented here gives strong support for the monophyly of the Cupressaceae sens. str. Monophyly of the family is also supported by the opposite and de- cussate phyllotaxis, the absence of a germination papilla and the compound nature of the sexine in the pollen (Ueno, 1959; Hart, 1987), and appar- ently also in the basipetal sequence of protoxylem differentiation within the leaves (Quinn & Gadek, 1988). There are further characters cited by Hart (1987) that show reversals or parallelisms. Although the sampling of genera in this study is still quite limited, especially among the southern genera, some assessment of the support given to the present intrafamilial taxonomy (Table 1) by the sequence data is already possible. The southern genera form a strong clade but there is no support for the inclusion of the North African genus Tetra- clinis within the mainly southern Callitroideae. The subfamily distinction (Li, 1953) was based on a single character, ovulate cone scales imbricate or valvate, which has been inconsistently scored in several genera by different authors (cf. Buchholz, 1948; Janchen, 1950) and which de Laubenfels (1965) described as "spurious." Independent ev- 584 Annals of the Missouri Botanical Garden aida Eon 3 (1) | или B pancia 1 (0) 30) а 79% : — Mn 2 (1) Calocedrus 3 (2) Chamaecyparis 75% Tetraclinis 2 (0) Thuja 2 (1) 80% Thujopsis Diselma Widdringtonia 13 (7) Callitris 95% Libocedrus FIGURE 1. Опе of two shortest trees found by PAUP branch and bound analysis, arranged with Glyptostrobus and E Line (Taxodiaceae) as the outgroup. The tree shown is that with the lowest f-value. Apomorp ic suppor as identified by PAUP is indicated above the branches, with the number of unique synapomorphs in brackets. Frequency of occurrence in bootstrap replicates of groups of taxa to the right of the branch are indicated as a ch. percentage below the bran idence of an affinity between Tetraclinis and cer- (Gadek & Quinn, 1985), and transfusion tracheids tain northern genera, especially Cupressus and (Gadek & Quinn, 1988). Juniperus, has come from recent work on seed The pattern of relationships within the northern proteins (Price, pers. comm.), leaf biflavonoids and southern clades (Fig. 1) is strikingly at variance Volume 80, Number 3 Gadek & Quinn 585 1993 Relationships within Cupressaceae Sens. Str. Glyptostrobus — Шиш Сгурїотетїа Microbiota T'etraclinis Platycladus Juniperus Cupressus Chamaecyparis | Calocedrus Jo Thuja Th uJopsis Diselma Widdringtonia Callitris Libocedrus FIGURE 2. One of three shortest trees found using PAUP after invoking the character weighting assumptions of Albert et al. (1993), shown with relative branch lengths. The other two trees resolve the trichotomy by placing Chamaecyparis either basal to the other two clades, or basal within the Juniperus—Cupressus clade. In each case, PAUP assigns a relative branch length of zero to these internodes. The tree shown is one step longer than that in Figure 1. Annals of the Missouri Botanical Garden with the tribal groups of Li (1953). The placement of Callitris (Actinostrobeae) as sister group to Di- selma + Widdringtonia (Libocedreae) renders the latter tribe paraphyletic. The monophyly of neither the Cupresseae (Cupressus and Chamaecyparis) nor the Thujopsideae (Thuja, Thujopsis, Caloce- drus, Platycladus, and Microbiota) is supported by the sequence data. Indeed, all the trees derived from PAUP using either default or Albert et al. char- acter-state weightings, and from the bootstrap anal- ysis, contain the same three strong clades within the northern genera: Juniperus + Cupressus, Mi- crobiota + Platycladus, and Thuja Thujopsis. The position of the other northern genera is equiv- ocal on these sequence data, but there is some evidence from the Albert trees, and the bootstrap analysis, for the association of Tetraclinis with the first two of these clades. Such an affinity is sup- ported by both chemical and anatomical data: 7et- raclinis possesses the same leaf biflavonoid pattern that distinguishes Cupressus and Juniperus (main- ly parental biflavonoids with cupressuflavone a ma- jor constituent) and all five genera accumulate cu- pressuflavone (Gadek & Quinn, 1985; unpublished). Cupressuflavone derivatives also occur in Calo- cedrus decurrens, but are not found in the species of Chamaecyparis or Thujopsis sampled here. Tet- raclinis, Platycladus, Cupressus, and Juniperus all possess trabeculate pitting in their transfusion tracheids, a state not found in Chamaecyparis, Thuja, or Thujopsis, while the intermediate state of barred pits occurs in. Calocedrus (Gade Quinn, 1988). These same characters support the grouping of Cupressus with Juniperus, which Li (1953) assigned to a separate tribe (Table 1), rather than with Chamaecyparis to form the Cupresseae. They also support the separation of Platycladus from the Thuja + Thujopsis clade with which it has long been associated. The close association of Thuja and Thujopsis accords with other available data, and their divergence from the other northern genera finds some support from wood characters (Hart, 1987). The lack of a robust resolution of affinities between these clades and other northern genera, however, particularly with regard to the placement of Chamaecyparis, Calocedrus, and Tetraclinis, is a reflection of the low evolutionary rate of divergence of the rbcL gene within this group. Sequence data from a faster-evolving gene may help resolve the cladogram. CONCLUSIONS The analysis of rbcL sequences of representative species of 13 of the 20 genera of the Cupressaceae sens. str. has produced a cladogram that finds con- siderable support from data derived from other sources. The monophyly of the family is well sup- ported. The cladogram provides support for the recognition of a basically northern subfamily (Cu- pressoideae) and a southern subfamily (Callitroide- ae), but the boundary between these subfamilies needs to be realigned by moving Tetraclinis from the Callitroideae to the Cupressoideae. The tribal arrangement of Li (1953) is largely artificial: the Libocedreae, Cupresseae, and Thujopsideae are clearly paraphyletic. Although some robust clades are recognized (viz, Пета + Widdringtonia, Juniperus + Cupressus, and Thuja + Thujopsis), the present database will not provide a robust res- olution of relationships among the genera because of the low evolutionary rate of divergence of the rbcL sequences, especially within the northern sub- family. LITERATURE CITED ALBERT, V. A., M. W. Chase & B. D. MISHLER. 1993. Character-state weighting for cladistic analysis of pro- tein-coding DNA sequences. Ann. Missouri Bot. Gard. 80: 752-766. BucuHorz, J. T. 1 Generic and sub-generic dis- irem E Ж Coniferales. Bot. Gaz. (Crawfordsville) бей, 5 T & T L. DoyLE. 1990. DNA from m tissue. Focus 12: ECKENWALDER, J. 9 Re-evaluation of Cupres- saceae and тахо А proposed merger. Ma- drono rl m 00. 1:06 INN. Isolation of plant 13-15. GADEK, P. ae ie Cupressaceae. Phytoc istry 22: 969-972 & 1983. Biflavones of the )chem- 985. Biflavones of the subfamily cre ‘Cupressaceae. Phytochemistry 24: 7-27 & 1988. Pitting of transfusion tra- d in Cupressaceae Austral. J. Bot. 36: 81-92. Hart, J. 1987. cladistic analysis of conifers: bon results. P Arnold Arbor. 68: 269-307. JANCHEN, E. Ak LAUBENFELS, D. J. DE 1965. The relationships of des roya (Molina) Johnston and Diselma archeri J. Hooker based on morphological considerations. Ph. tomorphology 9. Li, H. 1953. A reclassification of ге Pri and Cu- 17- | 59. Ani immu- nological comparison of the Sciadopityaceae, Taxo- = ic bu depend Syst. Bot. 14: 141-149. QUIN A. Сарек. 1988. The sequence of pie ee ме in the leaves of Cupressaceae. : 1344-1351. SworrorD, D. 1991. PAUP, Phylogenetic analysis us- ing parsimony. ПІ. Illinois Natural History Survey, mpaign. UENO, i 1959. Studies on pollen grains of ке mae, concluding remarks to the relationships betwe Coniferae. J. Inst. de A echn. Osaka City Univ., Ser. D., Biol. 11: 109- MOLECULAR PHYLOGENETICS OF THE MAGNOLIIDAE: CLADISTIC ANALYSES OF NUCLEOTIDE SEQUENCES OF THE PLASTID СЕМЕ rbcL' Yin-Long Qiu; Mark W. Chase,?* Donald H. Les, and Clifford R. Parks ABSTRACT Nucleotide sequences of the plastid protein-coding gene rbcL from 64 species of 36 families in subclass Magnoliidae sensu Cronquist and representatives of all other major seed plant groups were analyzed by pp in a four analyses. ee (Ceratophyllaceae) was found to be sister to all other angiosperms. Other magnoliids formed five major groups, roughly corresponding to the Magnoliales, Laurales, vibes cuna їйїн, Nym- phaeales, and jon erales. Four magnoliid lineages, those with monosulcate or monosulcate-derived pollen (Magnoliales, Laurales, Aristolochiales/Piperales, and Nymphaeales), and the monocots (with the same type of pollen) formed a weakly supported monophyletic group. The Illiciales (Illiciaceae and Schisandraceae), with a unique type of tricolpate pollen, also fell into this monosulcate clade. Relationships among these major lineages were resolv ed h low levels of support. Magnoliidae with tricolpate and tricolpate-derived pollen, Ranunculales se ries о er-group relationship of По with the two general pollen types. They also deviated in the specific taxon us du of some major clad The dicotyledonous subclass Magnoliidae (Cron- quist, 1981) contains most of the putatively prim- itive angiosperms. A thorough phylogenetic un- derstanding of the complex is critical for study of the origin and reconstruction of the phylogeny of angiosperms. The Euanthial Theory of angiosperm evolution (Arber & Parkin, 1907) postulates that angiosperms with large strobiloid flowers such as Magnoliales are more primitive than those with diminutive, simple flowers (e.g., Piperales and Chlo- ranthaceae). Over the past two decades, however, paleobotanists have accumulated a large body of fossil evidence which suggests that, during their early evolution, angiosperms with small and simple flowers might have paralleled or preceded those with large and complex flowers (Dilcher, 1979; Upchurch, 1984; Crane et al., 1986; Friis et al., 1986; Crane, 1989; Taylor & Hickey, 1990; Pe- 1991). A similar challenge has come from studies of extant plants (Burger, 1977, 1981; Hamby & Zimmer, 1992; Taylor & Hickey, 1992). Furthermore, which particular groups of Magno- dersen et al., liidae are related to the monocots and eudicots (the dicots with tricolpate and tricolpate-derived pollen ' We thank the following people for help in obtaining plant material for this study: W. Barthlott, H.-D. dias D. J. Crawford, S. Dayanandan, W. C. Dickison, P. K. Er . Liu, D. 6. Taylor, T. Terrazas- Salgado, L F. Wimpee, D. K. for assisting with par individuals for 2s mars shed sequences. J. A. manuscript; their suggestions and in the . B. Thi idress, H. Forbes x M. Miller, J. C. Ralston, D. E. Stone, T. F. 1, J. L. Ward, K. J. Wurdack, and Y. To M. Deti jew 5. E. Williams for nc assistance, V. the data analysis, P. C. I>? and J. R. le, P. R. Crane, and J. W. sightful comments are gratefully acknowledged. This и Coker fellowship and a fellowship from the Graduate School, University of North Carolina , L. Goff, G. aan B. Ke ег Stuessy, E. Р. Xia. also thank H. i: Albert and B. = Midis Massey for helpful discussion and many other itically reviewed the NSF taia BSR-8906496 to M. Chase and BSR-8817992 to D. Les. Publication of _ paper is partially supported by the A. E. Radford and Laurie S. Radford Trust for Publication in Systematic Botar * Department of Biology, The University of Nor th Carolina, Chapel Hill, North Carolina 27599-3280, U. S.A ‘Current address: Кор of Molecular Systematics, The Royal Botanic Gardens, Kew, Richmond, pem TW9 ЗАВ, ОК. ‘ Department of Biological Sciences, The University of Wisconsin, Milwau ee, Wisconsin 53201, ^ Current е Department of Ecology and Evolutionary Biology, University of Connecticut, „м rk onnecticut 06269, U.S ANN. MISSOURI Вот. GARD. 80: 587 -606. 1993. 588 Annals of the Missouri Botanical Garden types, exclusive of Illiciales, cf. Walker & Doyle, 1975; Doyle & Hotton, 1991) remains a debatable issue (Dahlgren & Clifford, 1982; Dahlgren et al., 1985; Walker & Walker, 1984; Donoghue & Doyle, 1989; Doyle & Hotton, 1991). The Magnoliidae have been known as the ranali- an complex, Ranales, or Polycarpicae. In the clas- sification scheme of A. P. de Candolle (1824), their prototype first appeared; it included key families of Magnoliales (Magnoliaceae, with Illicium of Il- liciaceae and Drimys and Tasmannia of Winter- aceae; and Annonaceae, with Kadsura of Schisan- draceae), Nymphaeales (Nymphaeaceae, with Nelumbo of Nelumbonaceae), and Ranunculales/ Papaverales (Ranunculaceae; Menispermaceae, with Lardizabala of Lardizabalaceae; Berberida- ceae; Papaveraceae; and Fumariaceae). These taxa were placed sequentially at the beginning of the system, which covered only dicots and gymno- sperms, but no collective name was given to this aggregation of families. Lindley (1833) was the first one to erect two taxa, Ranales and Anonales, to cover this group of plants. In Genera Plantar- um, Bentham & Hooker (1862) followed de Can- dolle’s treatment for this complex, except for ad- dition of a lauralean family, the Calycanthaceae; they placed all these taxa together under Ranales. Situating these families at the beginning of the classification systems antedated formulation of the Euanthial Theory by Arber & Parkin (1907), which equated flowers of the Magnoliales to Bennettita- lean strobili and provided a theoretical explanation for recognizing Magnoliidae as the primitive an- giosperms. Eichler (1890) and Engler & Prantl (1891), despite advocating Amentiferae as the primitive angiosperms, also made a significant con- tribution to development of the ranalian concept by adding to the complex the major lauralean fam- ilies, Hernandiaceae, Lauraceae, and Monimi- aceae, and several other taxa such as Lactorida- ceae and Trochodendraceae (also including Cercidiphyllum of Cercidiphyllaceae and Eupte- lea of Eupteleaceae). In his 1915 paper, often regarded as the first truly phylogenetic classifica- tion, Bessey brought to the ranalian complex the piperalean families (Piperaceae, Saururaceae, and Chloranthaceae). By early this century, the Mag- noliidae were conceptually mature and comprised five major lineages: Magnoliales, Laurales, Piper- ales, Nymphaeales, and Ranunculales/ Papaver- ales. Modern systematists (Takhtajan, 1987; Cron- quist, 1988; Dahlgren, 1989; Thorne, 1992) have agreed upon the basal position of Magnoliidae in angiosperms and recognition of the five major lin- eages; nonetheless, their opinions differ widely on relationships among these lineages. Takhtajan (1987) divided the complex into two subclasses, placing monosulcate lineages in Magnoliidae and tricolpate lineages in Ranunculidae (but Illiciales and Nelumbonaceae were in the former). Dahl- gren's (1989) treatment of the complex resembled that of Takhtajan (1987), but no higher categories were given to the two assemblages, Magnolianae/ Nymphaeanae and Ranunculanae (cf. fig. 1 in Dahlgren, 1989). In the monosulcate assemblage, Dahlgren placed Piperales with Nymphaeales, rath- er than with Magnoliales and Laurales as in Takh- tajan's (1987) scheme. A major difference between Cronquist's (1981, 1988) and Takhtajan's (1980, 1987) systems is that Cronquist did not recognize Ranunculales/ Papaverales as a separate subclass. Thorne (1992) divided the complex into three su- luding most monosulcate perorders, Annonan (ir families plus о айва Paunvirales) Nym- phaeanae, and Rafflesianae (the latter two narrowly defined). Relationships of several isolated families, Aus- trobaileyaceae, Ceratophyllaceae, Chloranthaceae, Illiciaceae/Schisandraceae, Lactoridaceae, and Nelumbonaceae, have been extensively debated. These taxa are either included in one of the major lineages of Magnoliidae or treated as independent minor groups, but their affinities are typically stated to be uncertain. Recently, several cladistic studies of morpho- logical, phytochemical, and molecular data have attempted to answer some of the questions con- cerning the phylogenetics of Magnoliidae (Dahlgren & Bremer, 1985; Donoghue & Doyle, 1989; Lo- conte & Stevenson, 1991; Martin & Dowd, 1991; Hamby & Zimmer, 1992; Taylor & Hickey, 1992). These efforts have met with only partial success due to: (i) widespread homoplasy in primitive an- giosperms, (ii) problems of choosing appropriate ingroups and outgroups, (iii) difficulty of coding characters in morphological cladistic studies, or (iv) insufficient sampling (a major problem in the mo- lecular investigations). Protein-coding genes have been shown to reflect the genealogical history of organisms (Atchley & Fitch, 1991), though to what degree and at what taxonomic level this is true awaits further study. Evolutionarily conservative genes, such as rbcL, which codes for the large subunit of ribulose-1,5- bisphosphate carboxylase/oxygenase (RuBisCO), are considered suitable for study of relationships among taxa of higher rank (Ritland & Clegg, 1987; Zurawski & Clegg, 1987; Palmer et al., 1988). They provide new characters that permit an in- Volume 80, Number 3 1993 Qiu et al. Molecular Phylogenetics of Magnoliidae 589 dependent evaluation of phylogenetic relationships among plant groups uninfluenced by previous ideas of transformation of morphological characters and putative relationships among taxa. In the present study, we perform parsimony analyses of rbcL sequences from species representing almost all magnoliid families and endeavor to (1) evaluate the concepts of Magnoliidae, (2) identify major lineages in the subclass, investigate relationships within these major lineages, and assess affinities of several prob- lematic taxa, (3) determine the relative phyloge- netic positions of the major lineages within Mag- noliidae, and (4) examine relationships of the Magnoliidae to monocots and eudicots. MATERIAL AND METHODS Well aware of the pitfalls caused by inadequate sampling of sequence variation (Albert et al., 1993; Chase et al., 1993), we sampled at least one species from each family (this of course does not address adequately the putative monophyly of these fam- ilies). Thus 64 species from 36 families repre- senting all orders of the Magnoliidae were included (we follow the classification scheme of Cronquist, 1981, unless otherwise indicated). These taxa are listed in the Appendix at the end of this issue. Several sequences have been drap previously (Golenberg et al., 1990; Les et al., 1991). Missing families are Circ алаардан; Gomortegaceae, Tri- meniaceae, Hydnoraceae, and Rafflesiaceae (Takh- tajan, 1987; Dahlgren, 1989; Thorne, 1992). It is likely that the last two, which are achlorophyl- rbcL, as was found in other parasitic plants (dePamphilis & Palmer, 1990). Total cellular DNA was extracted using the mod- ified CTAB method (Doyle & Doyle, 1987) from 0.6-2.0 g fresh or silica gel-dried leaves (Chase & Hills, 1991) collected from a single plant (in the case of Hernandia ovigera L., the leaf was re- moved from a 22-year-old herbarium specimen). DNAs were purified by CsCl-ethidium bromide gradient centrifugation. A fragment containing the gene rbcL was amplified using a thermo-stable DNA polymerase and two synthetic 26-mer primers based on the rbcL sequences from Zea mays L. and Vicotiana tabacum L. The amplified fragment was ligated into Bluescript vector and cloned with stan- dard recombinant techniques. Sequencing was per- formed using Sequenase 2.0 (US Biochemical, Inc.) for dideoxynucleotide chain-termination reactions, and sequences were obtained from both strands. For several species, the gene was directly se- quenced from the amplified fragment without clon- ing. Sequence data were analyzed using PAUP 3.0s (Swofford, 1991). Several searches were conducted to approach problems at different scales and with different techniques. Search A. All 65 magnoliid sequences were analyzed together with 434 other seed plant rbeL sequences (see Chase et al., 1993) to identify major lineages in the Magnoliidae, determine relative po- sition of these lineages, and examine relationships Two sequences of Canella winterana (L.) Gaertner, both with intact reading frames, were obtained from the DNA sample extracted from a single plant. They differ by 57 base pairs; 24, 15, and 18 of these differences occur at the first-, of the subclass to monocots and eudicots. , and third- codon position, respectively. Both sequences were used in all analyses; one of them may be a "pseudo- gene." Due to the magnitude of the data set and the limitation of the software, the level of parsimony second- of this broad search was not certain (see Chase et al, 1993, for a more detailed explanation and discussion of the result of this larger analysis). This search helped us select a smaller number of taxa for a manageable" analysis to examine internal support. Search B. We performed another analysis with all monosulcate magnoliid sequences plus se- lected gymnosperms, monocots, and eudicots that served as place-holders for their respective groups identified in Search A. An analysis under the Fitch criterion SIM weights for all substitutions; Fitch, 1971) using 2,000 random sequence additions, STEEPEST DESC ENT, JLPARS (but permit- ting only 10 trees to be held at each step), and NNI (nearest-neighbor interchange) found only one island of trees. ~ All trees within this island were recovered by using the trees found in the random searches as starting points with MULPARS and TBR (tree bisection-reconnection) until swapping was completed. earch С. То examine the effect of distant outgroups such as Gnetales on the ingroup network, we conducted an ingroup-only analysis of flowering plants. Again, this analysis was performed under the Fitch criterion using 2,000 random sequence additions in the same manner as described above. In addition to the same set of monosulcate mag- noliids and monocots as were used in Search B, representative species of all families in the Ranun- interfamilial relationships within this group. T islands of equally parsimonious trees were found, and all of the trees in each of those islands were recovered by the same method as in Search B (swapping to completion with MULPARS and TBR). 590 Annals of the Missouri Botanical Garden To gain a measure of the robustness of the topology, decay of parsimony (Bremer, 1988) was examined. All maximum parsimony trees of both islands were used as starting trees to search for trees up to five steps less parsimonious; the FILTER TREES option was used to identify trees at each length, and a strict consensus tree was computed at each step. The number of steps less parsimonious at which a branch collapsed was recorded as the decay index. The larger the number, the more robust a branch. A decay index of 1 indicates that the branch is present in all maximum parsimony trees but collapses at maximum parsimony plus one step. Search D. There has been great interest in recent cladistic studies to determine the basalmost lineage of angiosperms. The Nymphaeales are often claimed as a sister group to all other angiosperms (Donoghue & Doyle, 1989; Hamby & Zimmer, 1992), but our results did not place them in such a position. To evaluate the hypothesis of Nym- phaeles being the basalmost angiosperms, we per- formed a topology constraint experiment using the same set of taxa as in Search B, but with the Nymphaeales constrained to be sister to all other | (using the CONSTRAINTS option of PAUP). The search was conducted under the Fitch criterion using 2,000 random sequence additions in the same manner as described in Search B. This is а more accurate means to determine the loss of parsimony associated with an alternate topology; merely constraining the topology based on the shortest trees found overall without performing branch swapping is likely to result in an inaccurate assessment of tree length because it does not permit character transformations to be optimized on dif- ferent branches than in the unconstrained trees. RESULTS Search A. puted from 3,900 equally parsimonious tree length of 16,305 steps (Chase et al., 1993). es trees have a consistency index (C.I.; for potentially synapomorphous characters) of 0.1 tention index (R.I.) of А strict consensus tree was com- and a re- .638; a summary of this tree is shown in Figure 1 (for the detailed version of the strict consensus tree, see Chase et al., 1993). Exclusive of Ceratophyllum, which is sister to all other flowering plants, the angiosperms are split into two clades that correspond to the two general pollen types, monosulcate and tricolpate (for the sake of convenience, we refer to angiosperms with monosulcate and monosulcate-derived pollen types as the “‘monosulcates” and those with tricolpate “eudi- the latter in accord with Doyle & Hotton's, and tricolpate-derived pollen types as the cots,” 1991, use of that term). Five major lineages are identified among the Magnoliidae; they are roughly equal to the Magnoliales, Laurales, Piperales (in- cluding Aristolochiaceae and Lactoridaceae), Nym- phaeales (excluding Nelumbonaceae but including Amborellaceae, Austrobaileyaceae, Chloranthace- ae, and llliciales), and Ranunculales (excluding Coriariaceae, Sargentodoxaceae, and Sabiaceae but including Eupteleaceae, Fumariaceae, and Papav- eraceae). (We will use these ordinal circumscrip- tions throughout the rest of this paper; i.e., under Piperales we will include Aristolochiaceae and Lac- toridaceae, unless stated otherwise.) The four mag- noliid lineages with monosulcate pollen types, (i) Magnoliales, (ii Nymphaeales, plus the monocots (with the same type of pollen), form a clade. The Illiciales, which have anomalous tricolpate pollen (see below), also — Laurales, (ш) Piperales, and (iv) fall into this monosulcate clade. The Piperales are sister to all other monosulcates; Magnoliales/Nym- phaeales, Laurales, and monocots form an unre- solved trichotomy in the strict consensus tree. The Ranunculales, Nelumbonaceae, and Sabiaceae, all of which have tricolpate pollen, together with other angiosperms possessing the same pollen type, form a monophyletic group. This large clade corresponds to Walker & Doyle's (1975) nonmagnoliid dicots or Doyle & Hotton’s (1991) “eudicots.” The Ra- nunculales are sister to the rest of the eudicots. In another basal lineage of eudicots are Nelumbona- ceae and Sabiaceae (with Platanaceae and Protea- ceae; hamamelid I, Fig. 1). The Coriariaceae and Sargentodoxaceae are not affiliated with the Mag- noliidae, the former being sister to the clade of Maid HO dep ri and the latter imbedded i A single island of 46 trees at a length of 3,414 steps with C.I. of 0.275 (unique substitutions excluded) and R.I. of 0.566 was found. One of these maximum parsimony trees favored by the weighting criterion (Albert et al., 1993) is shown with Fitch branch lengths optimized on it ACCTRAN optimization; Fig. 2). Even though far fewer taxa (82 species) were used in Search B, the same general topology as Search A was obtained. Search С. wo islands of 89 and 3 trees respectively, at a length of 2,674 steps with C.I. of 0.346 (unique substitutions excluded) and R.I. of 0.531 were found. For illustration, one tree from ~ each island was selected using the weighting cri- terion of Albert et al. (1993). The trees in the Volume 80, Number 3 1993 t al. 591 Molecular Phylogenetics of Magnoliidae 28 39 asterid I asterid II asterid III asterid IV asterid V rosid I rosid II rosid III caryophyllids Gunnera hamamelid II hamamelid I 22 Ranunculales sp- Nymphaeales ?- Magnoliales 11 24 monocots 8 Laurales Piperales 12 Ceratophyllum Gnetales other conifers Pinaceae Ginkgo cycads Annals of the Missouri Botanical Garden |в 72 44 Спешт ss 2 Welwitschia | Gnetales | U8— Ephedra 34 А 1 Houttuynia Piperales Saururus а roris | — Lactoridaceae gristo chia т Saruma Aristolochiaceae TTA Iro E Magnolia hypoleuca nola macrophylla ulimima Magnoliales Chloranthaceae Nymphaeales Amborellaceae Illiciales __. Austrobaileyaceae Laurales is tia gem А isma a Spathiphyllum Gymnostachys ea monocots ss eudicots — Ceratophyllaceae ciadop Taxus. - Ti ium | 21 5$ Metasequoia | 54 Callitris conifers 15 Podocarpus 5 pps Peeudotsuga us " ALO Тира __ inkgo Ginkgoaceae су ае В 7 = Ал cycads а 2 Encephalartos , Volume 80, Number 3 1993 Qiu et al. Molecular Phylogenetics of Magnoliidae 593 larger island (Fig. 3), though less resolved, are congruent in the general topology with the trees from Search B (Fig the smaller island (Fig. 4) are different from those . 2). In contrast, ihe trees of found in Search B; Ceratophyllum allies with Ca- nella in the Magnoliales rather than being situated in an isolated position. These trees were arranged to agree with the results of two previous searches (i.e., with the eudicots sister to the monosulcates). Many parts of this topology (Fig. 3) do not demonstrate strong internal support (decaying at two or fewer steps less parsimonious). At the level of the relationships among the major groups iden- tified, all are weakly supported (even though the monophyly of some of the orders themselves are strongly supported; see below). The shift of Cer- atophyllum, which is the only taxon with a dif- ferent position in the two islands of trees, dem- onstrates its low level of internal support. It would appear that the use of distant outgroups such as Gnetales may be introducing homoplasy such that the smaller island does not exist at maximum par- simony in Search B (we cannot evaluate the degree of parsimony of the trees or the existence of other islands of equally parsimonious trees in Search А because of the size of the data set). Groups with moderate to strong support (decaying at three or о ~ more steps less parsimonious; Fig. 3) are eudicots, monocots, Laurales, “core” Magnoliales (excluding Canellaceae and Winteraceae), Nymphaeales (in- cluding the associated woody families, but exclud- ing een ue Piperales, and Ranunculales. rch D. А single island of 186 trees at a a ы 3,425 steps with C.I. of 0.274 (unique substitutions excluded) and КЛ. of 0.565 was found. ne of the maximum parsimony trees chosen by the weighting criterion is shown in Figure 5. Com- pared with the trees from Search B (length of 3,414 steps; the same taxa are present in both), the trees recovered in this search are only slightly less par- simonious: 11 steps or only 0.32% longer. Although relationships among the four mono- sulcate magnoliid lineages, monocots, and eudicots are resolved with only low levels of support, these clades are consistently identified, and relationships within the clades remain unchanged in the series 1, 2, 3, 4). Even though the constraint experiment makes the group related to of analyses (Figs. the Nymphaeales paraphyletic to all other angio- sperms, it leaves them in near proximity to each other and otherwise identifies the same major lin- eages as in the previous searches. Ceratophyllum is the single taxon that shifts wildly with each treatment. DISCUSSION l. PHYLOGENETIC EVALUATION OF THE MAGNOLIIDAE From the review of historical development of the ranalian concept and the treatment of the com- plex in several modern classification systems pre- sented in the introduction, it is clear that five major lineages, Magnoliales, Laurales, Piperales, Nym- phaeales, and Ranunculales, were consistently rec- ognized. In this study, we obtained clades that are roughly equal to these traditionally recognized lin- eages, and they occupy basal positions among an- giosperms. Our analyses were conducted only un- der the criterion of parsimony, and we did not rely on any previous hypothesis concerning evolution of morphological characters or putative relation- ships among these taxa. The high degree of cor- respondence between the clades identified in this study and those traditionally recognized categories provides support for these lineages (even though support for several of these was weak in our study) and is an indication that both morphology and rbcL sequences are informative for phylogenetic recon- struction. The split of the angiosperms, exclusive of Cer- atophyllum, into a monosulcate and eudicot clade in our results was unexpected. Monophyly of the monosulcates has never been suggested, and at present we know of no unequivocal morphological or phytochemical characters to define the clade. Possession of monosulcate pollen is not a synapo- morphy because this pollen type is also found in nonflowering seed plants. The ethereal oil cells could be a potential synapomorphy for the mono- sulcates; they occur in all component lineages (at least in some basal members) of this clade and absence of these cells in Amborellaceae/ Nym- phaeales and most monocots (basal monocots such as Acorus have the ethereal oil cells, Cronquist, 1981) can be interpreted as due to secondary loss. A close relationship between the Magnoliidae and the monocots has been proposed by Huber (1977), Hegnauer (1977), and Kubitzki & Gottlieb (1984), —— FIGURE 2. 82 species of gymnosperms, monosulca One of the 46 maximum- бадан і (Fitch e magnoliids ) , monocots, an the weighting criterion of Albert et al. (1993) but branch lengths are Fitch bici а trees from the single qe found for rbcL sequences of eudicots. rticular tree was selected by (ACCTRAN 594 Annals of the Missouri Botanical Garden Piperales — Lactoridaceae Aristolochiaceae Laurales monocots 528 Spathiphyllum Magnoliales Chloranthaceae Е Nymphaeales Amborellaceae Illiciales __ Austrobaileyaceae eudicots ма PUTEM todoxa melis qe ой Pachysandra Sabia nus Plata Velumbo Ceratophyllum FIGURE 3. One of the 89 maximum-parsimony (Fitch) trees, chosen with the weighting criterion, from the larger of the two islands found for rbcL sequences of 74 species of angiosperms. The branch lengt 1s are Fitch optimizations (ACCTRAN). Decay indices are indicated below each branch and are preceeded with a “d” (“d0” indicates a branch present in the particular tree favored by the weighting criterion but not in all шла trees). Volume 80, Number 3 Qiu et al. 595 Molecular Phylogenetics of Magnoliidae chisandra cium jus trobaileya us Calycanthus Idiospermum 2 z Hedycarya 26 12 Cinnamomu m E 25. Persea 8 15 Оез 197 > ~ = SEES Б Sargentodoxa е а 6 = Tetracentron 122. Pachysandra Sabia 7 1 Platanus LP. Nelumbo dendron Magnolia hypoleuca alauma lagnolia macrophylla Piperales Lactoridaceae Aristolochiaceae Magnoliales — Ceratophyllaceae<@— Magnoliales Chloranthaceae Nymphaeales — Amborellaceae —Ouriciates Austrobaileyaceae monocots Laurales eudicots —— E 4. One of the three maximum-parsimony (Fitch) trees, chosen with the weighting criterion, from the e two islands found for rbcL sequences of 74 species of angiosperms. The branch lengths are Fitch optimizations (ACCTRAN). 596 Annals of the Missouri Botanical Garden n Wefwitschia iz ZEN = ID R odium etasequoia Galfuris Po pearpus о Bp а omia : шш упа Piperales 1 aururus ctoris ___Lactoridaceae gristolochia . | ита ¡Aristolochiaceae um — eudicots —À ___Ceratophyllaceae Magnoliales monocots Pp oenix кое СЗ, Ила, obw. agnolia macroph ylla albulimima | egeneria Magnoliales nona mina ananga Enema e edyosmum m EE Mind Chloranthaceae arani us = ocar gus Ви и с e Laurales ya Ginnamomum ersea d re Pirie вал бы Illiciales " ~ L20 Austrobaileya Austrobaileyaceae bo Y Amborellacéa pal Nym = 19 „6 Yasenia Nymphaeales и Cabomba 18 al Zamia x ncephalartos Volume 80, Number 3 1993 Qiu et al. Molecular Phylogenetics of Magnoliidae 597 but they relied on what are probably symplesiomor- phic features. Currently available fossil evidence indicates that monosulcate angiosperms antedate eudicots (Doyle, 1969; Muiler, 1970, 1981, 1984; Doyle & Hickey, 1976; Hickey & Doyle, 1977; Friis & Crepet, 1987; Hughes & McDougall, 1990; Doyle & Hotton, 1991), and it has been postulated that the latter were derived from the former (Walk- er & Walker, 1984). Crane & Lidgard (1990), after examining palynological diversity in the Cre- taceous, suggested that the monosulcates probably represent an evolutionary grade rather than a clade. Three different scenarios may explain the discrep- ancy between our results and these previous hy- potheses. First, the fossil evidence is most certainly incomplete, and future paleobotanical exploration may find older evidence for eudicots. Second, the weakly supported monosulcate clade is perhaps generated spuriously by high levels of sequence divergence among the monosulcates, eudicots, and gymnosperms (i.e., the monophyletic monosulcate clade is the result of “long branch attraction"; cf. 1978). lates that two lines, both of which contained mono- Felsenstein, The third explanation postu- sulcates, existed: one was paraphyletic to extant eudicots, and the monosulcate members of this line are now extinct; the other line included at the least all extant monosulcates. This third scenario implies the existence of yet unidentified lineages of mono- sulcates in the fossil record. It has long been recognized by phylogenists that Magnoliidae sensu Cronquist are paraphyletic to monocots and eudicots. A cladistic study of basal angiosperms using morphological characters has corroborated this viewpoint (Donoghue & Doyle, 989). Our study of rbcL sequences further dem- onstrates the paraphyly of the Magnoliidae. Takh- tajan (1987) and Dahlgren (1989) removed the Ranunculales from the Magnoliidae and placed them in a higher category equal to the rest of the sub- class. These plants lack several features used to define the Magnoliidae, e.g., cells in the parenchymatous tissues and monosul- spherical ethereal oil cate pollen. In our analyses, the Ranunculales con- sistently fall into the strongly supported eudicot clade apart from the rest of the Magnoliidae sensu Cronquist. The Nymphaeales are another member of the Magnoliidae that are often treated as an indepen- dent group of equal rank to the rest of the subclass (Thorne. 1992) or to all other monosulcate mem- bers of the subclass (i.e., excluding Ranunculales: Walker & Walker, 1984; Takhtajan, 1987). This study does not support such treatment. The Nym- phaeales are only one of the major lineages in monosulcate Magnoliidae, and furthermore, sev- eral other taxa, Amborellaceae, Austrobaileyaceae, and Illiciales, share а close, unique phylogenetic relationship with them. The enigmatic genus Ceratophyllum forms its own clade and is sister to all other angiosperms when outgroups to the angiosperms are included, but when only angiosperms are analyzed its highly divergent sequence is involved in two islands of equally parsimonious trees in which it occurs in radically different positions (Figs. 3, 4). The affin- ities of Ceratophyllum have been controversial, and most recent authors favored a relationship with Cabomba of the Cabombaceae (Cronquist, 1981; Takhtajan, 1980; Dahlgren, 1989). After a critical review of all the evidence, Les (1988) questioned this connection. The phylogenetic position of Cer- atophyllum shown in this study largely corrobo- rates Les's (1988; that the genus represents a vestige of the ancient Les et al., 1991) proposition angiosperms that diverged early from the line lead- ing to most other modern taxa. From Aptian sed- iments in Victoria, Australia, and Albian and Cen- manian sediments Kansas, North America, Dileher (1989) reported fossil fruits similar to those of extant Ceratophyllum species; fossil leaves re- sembling those of modern Ceratoph yllum were also found associated with the fruits. Possession of a suite of distinctive characters (partially sealed car- pels, unitegmic ovules, branching pollen tubes, and lack of exine, vessels, and perianth) makes the genus anomalous among angiosperms. o evidence, however, can be considered to give strong support for any of the placements of Cer- atophyllum in our rbcL trees. The inclusion of distantly related outgroups (in particular, the highly divergent Gnetales) may be responsible for the placement of Ceratophyllum as sister to the rest of the angiosperms by introducing spurious synapo- morphies. Nevertheless, in the search with non- flowering seed plants removed (Search С), Cera- — FIGURE 5. Nymphaeales were specified as the sis t parsimonious, пуан ен trees found w optimizations (ACCTRAN). One of the 186 maximum-parsimony (Fitch) trees found in a constraint ee in which the er of all other angiosperms . These trees are only steps les 5 parsimonious ith the same data matrix. The branch lengths are Fitch 598 Annals of the Missouri Botanical Garden tophyllum remains as an outlier to the monosulcates and eudicots in the trees of one island (Fig. 3). In the trees from the other island, the genus is allied with Canella in the Magnoliales (Fig. 4), which seems unsatisfactory, both from the perspective of the long branches of Ceratophyllum and Canella and lack of corroboration from other evidence. In the topology constraint experiment (Search D), Ceratophyllum again ends up in another position . Our overall conclusion is that the rbcL analyses do not defin- that is potentially spurious (Fig. 5) itively resolve the phylogenetic relationships of Cer- atophyllum, but they do suggest, in agreement with Les's hypothesis (1988, 1991), that it is dis- tantly related to all other extant angiosperms and may represent an ancient and highly modified taxon allied to the oldest angiosperms, which is consistent with its long fossil history. 2. PHYLOGENETICS OF THE MAGNOLIALES, LAURALES, PIPERALES, NYMPHAEALES, AND RANUNCULALES The isolated position of Winteraceae has been recognized for some time (Bailey & Nast, 1943; Ehrendorfer et al., 1968; Thorne, 1968; Walker, 1976a; Takhtajan, 1980), and the family is placed d tly in several classification systems (Wal ker & Walker, 1984; Takhtajan, 1987; Dahl- gren, 1989; Thorne, 1992). A morphological cla- distic study showed that Winteraceae fall outside of the “ 1989). not compatible with other members of Magnoliales, e.g., vesselless wood, poorly organized leaf vena- core" Magnoliales (Donoghue & Doyle, The family has several features that are tion, and ulcerate pollen shed in permanent tetrads. The Canellaceae are another family not securely placed in the Magnoliales (Donoghue & Doyle, 1989). A relationship between these two families emerges from this study as well as previous work. Leroy (1977) suggested that Winteraceae are allied to Canellaceae via the anomalous Madagascan ge- nus Takhtajania; both have a compound uniloc- ular ovary, a type of gynoecium rare in the Mag- noliales. The two families (plus the Illiciales) were found to be associated in some of the equally most parsimonious trees in Donoghue & Doyle's (1989) study. Gottlieb et al. (1989) found that among members of Magnolianae and Piperales sensu Dahl- gren, Winteraceae and Canellaceae share sesqui- terpenoids of the drimane and rearranged drimane type, and at least two representatives of the dri- mane type occur only in these two families. A recent study of 265 rDNA also showed that these 1992). The other families of the Magnoliales, Anno- two families are related (Suh et al., naceae, Degeneriaceae, Eupomatiaceae, Himan- tandraceae, Magnoliaceae, and Myristicaceae, are often recognized as "core" Magnoliales (Thorne, ; alker, 1976a). Donoghue & Doyle’s (1989) morphological cladistic study also identified this group. Our study of rbcL sequences offers moderate support for this group (Fig. 3). The My- risticaceae, traditionally regarded as an advanced member of Magnoliales (Walker, 1976b; Walker Walker, 1984), occupy a weakly supported basal position in this clade (Fig. 3). e anomalous magnoliaceous genus Lirioden- dron stands in an isolated position from the rest of the family, and its inclusion in the ies oia has been questioned by Barkley (1975). The lationship of Liriodendron to да magnoliaceous members is well supported in this study (Fig. : A reliable monophyly morphological character nmn of Magnoliaceae is their gynoecium, spirally arranged follicles on an elongate recepta- cle, which elsewhere is only found in Ranuncula- ceae. The mid-Cretaceous (Cenomanian) angio- spermous fructification Archaeanthus provides evidence for a long history of this type of fruit Dilcher & Crane, 1984). A fining the Magnoliaceae is their large stipules that enfold terminal buds. The rest of the Magnoliaceae are so homogenous that their taxonomy is noto- riously difficult (Dandy, 1964; Law, 1984; Noo- teboom, 1985), and rbcL sequences show little divergence among the genera (Qiu, Chase & Parks, unpublished). nother character de- ~ In Laurales, Hernandia and Gyrocarpus have been placed in two unrelated families, Hernandi- aceae and Gyrocarpaceae, by some (Walker & Walker, 1984; Takhtajan, 1987). Our study shows that the relationship between these two genera is strong even though their rbcL sequences are di- vergent (Fig. 3). Dahlgren's (1983, 1989) inclusion of these two genera in the Lauraceae is not sup- ported by this stud The Calycanthaceae and Idiospermaceae form a strongly supported clade, though resolution of their. relationships to other lauralean families is poor. Interfamilial relationships within the Laurales are not well resolved (Fig. 3). Lack of several important taxa, such as Trimenia, Gomortega, Atherosperma, and other monimiaceous genera, may be a problem. Inclusion of these taxa in future studies should help to improve the resolution. The relationship between Aristolochiaceae and Piperaceae/Saururaceae has not been consistently recognized. Takhtajan (1987) and Cronquist (1988) placed the two groups together, whereas Dahlgren (1989) and Thorne (1992) allied the Aristolochi- Volume 80, Number 3 1993 Qiu et al. 599 Molecular Phylogenetics of Magnoliidae aceae with the “core” Magnoliales rather than with the Piperales. Results from our study clearly favor association of the two groups. The affinity of Lactoridaceae is an extensively debated issue. In the past, this family has been placed in the Magnoliales (Cronquist, 1981), Lau- rales (Takhtajan, 1980), Piperales (Walker & Walker, 1984; Thorne, 1992), or in its own order Lactoridales between the Magnoliales and Winter- ales (Dahlgren, 1989). A study of the family with numerical approaches suggested a magnolialean affinity (Lammers et al., 1986), but the proposition has been criticized by Carlquist (1990) on the basis of wood anatomy. Recent cladistic investigations using morphological data indicated a relationship to the Piperales (Dahlgren & Bremer, 1985; Don- oghue & Doyle, 1989). A detailed comparative study of wood anatomy revealed that the Chloran- thaceae, Piperaceae, and Lactoridaceae are similar in wood anatomy (Carlquist, 1990). In our anal- yses, support for a relationship of Lactoris to Aris- tolochia is weak, but alliance of the genus within Piperales is moderately supported (Fig. 3). Aris- tolochia has a rather divergent sequence, and more sampling in future studies may help to characterize better the affinities of the genus and Lactoris. The Nymphaeales are distinct among primitive angiosperms in that they occupy a peculiar eco- logical niche and exhibit many characteristics of aquatic adaptation. How they are related to the rest of the Magnoliidae has been a long-standing issue in the phylogenetics of primitive angiosperms. In this study, we found the Nymphaeales are allied with several isolated taxa, Amborellaceae, Austro- baileyaceae, and Illiciales, whose relationships have also been widely debated. This topology is rather novel, but it is not drastically different from some previous ideas. Takhtajan (1980) and Cronquist (1981) suggested the Nymphaeales are derivatives of a woody ancestor with magnolialean features. The Austrobaileyaceae and Illiciales share many floral features with the Magnoliales and have been treated as members of that order in the past. The Amborellaceae are generally regarded as a basal member of the Laurales and have been placed near the Magnoliales (Cronquist, 1981). Hence, it is not unreasonable that these taxa form a group para- phyletic to Nymphaeales. The Amborellaceae are a poorly studied mono- typic family endemic to New Caledonia and have always been treated as a member of the Laurales. Bailey & Swamy (1948), however, found that the only species of the family, Amborella trichopoda Baill., lacks vessels and then questioned its place- ment in the Monimiaceae. Whether the angio- а are primitively vesselless is an unresolved ssue (Young, 1981; 1989; Carlquist, 1992). Nym- phaeales are secondarily vesselless (Cronquist, Donoghue & Doyle, It is possible that the 1981). If vessellessness in primitive angiosperms is indeed due to secondary loss, this character may be of significant phylogenetic value in associating Amborella with the Nymphaeales. Another line of evidence supporting the relationship between Am- borella and Nymphaeales is their lack of ethereal oil cells in parenchymatous tissues (Money et al., 1950; Cronquist, 1981). The ethereal oil cells are a characteristic feature of monosulcate Magnoli- idae, and are otherwise found in all members. Fi- nally, both Amborellaceae and Nymphaeales have granular pollen that has no or reduced endexine in the nonaperture regions of the exine, features that elsewhere are found only in the "core" Mag- noliales, Trimeniaceae and some Chloranthaceae (Walker, 1976b). Although these “negative” acters need to be investigated further, they support char- a relationship between Nymphaeales and Amborel- — aceae. The Illiciales stand as the only exception of a tricolpate pollen-bearing taxon among the mono- sulcate angiosperms. Previously, Walker & Doyle (1975) have highlighted their problematic aperture type. Huynh (1976, cited in Doyle & Hotton, 1991) revealed that the tricolpate condition in Il- liciales is of a different type from that of eudicots; the three colpi are oriented according to Garside's rule, not Fischer's rule as in the latter. This implies that the tricolpate condition evolved independently in these two lineages. Our results support such interpretation. Leaf architecture (Hickey & Wolfe, 5) and morphological cladistics (Donoghue & Doyle, 1989) also favor this hypothesis, though neither of these studies associates Illiciales with the particular group of monosulcate magnoliids as iden- tified in this study. A relationship between Illiciales and Winteraceae has been proposed by some work- ers (Walker, 1976a; Donoghue & Doyle, 1989; Doyle et al., 1990) who relied partly on palyno- logical evidence, yet such a relationship was not observed in investigations of chromosome number, leaf epidermis, and secondary compounds (Ehren- dorfer, 1968; 1912: 19 . The one pollen character, coarsely retic- Baranova, Gottlieb et al., that unites the two groups in Donoghue & Doyle's (1989) cladistic study has not been observed in Halkeripollis, the ulate and semitectate exine, Early Cretaceous fossil pollen of Pa winter- aceous affinity (Doyle & Hotton, 19€ The Austrobaileyaceae are yet mi poorly studied family. They have been allied to Monimi- 600 Annals of the Missouri Botanical Garden aceae in Laurales (Bailey & Swamy, 1949) or Annonaceae of Magnoliales (Endress, 1980, 1983; Endress & Honegger, 1980). As noted by Cron- quist (1981), the family has a combination of char- acters that make its placement in either order difficult: unilacunar nodes, opposite leaves, climb- ing habit, laminar stamens, hypogynous flowers, monosulcate pollen, and pluriovulate carpels. The relationship between Austrobaileyaceae and Illi- ciales identified in this study is supported by their similar sieve-element plastids (Behnke, 1988). iven the extensive fossil record of Chlorantha- ceae (Couper, 1958; Kuprianova, 1967; Hughes et al., 1979; Muller, 1981; Upchurch, 1984; Walker & Walker, 1984; Friis et al., 1986; Chlon- ova & Surova, 1988), it may be safe to suggest that the family was more diverse in the past. A recent report of chloranthaceous fruits with anat- ropous ovules, instead of orthotropous ovules as in extant species, from Cenomanian sediments in east- ern North America corroborates such a hypothesis (Pedersen et al., 1991). The modern family stands in a rather isolated position and has been placed in the Laurales (Thorne, 1992), Piperales (Cron- quist, 1988), in its own order Chloranthales be- tween Lactoridales and Piperales (Takhtajan, 1987), or between Winterales and Illiciales (Dahlgren, 1989). In this study, we found the Chloranthaceae are allied with Austrobaileyaceae, Illiciales, Am- borellaceae, and Nymphaeales, but this position is weakly supported (Fig. 3). It has been suggested the Amborellaceae are related to the Chlorantha- ceae on the basis of floral morphology and anatomy (Endress, 1986a, 1987). Behnke (1988) also found Chloranthaceae share similar sieve-element plastids with Austrobaileyaceae and Illiciales. Assignment of Ranunculales and other ranalian taxa such as Eupteleaceae, Nelumbonaceae, and Trochodendrales (Trochodendraceae and Tetra- centracaeae) into two subclasses, the Magnoliidae (or Ranunculidae) and Hamamelidae by Takhtajan (1987) and Cronquist (1988) may have led to a belief that these taxa are distantly related. In our analyses, they form, together with several other taxa (Buxaceae, Gunneraceae, Platanaceae, Pro- teaceae, and Sabiaceae), a paraphyletic group at the base of the eudicots. In comparing vegetative and reproductive morphology, Nast & Bailey (1946) dissociated Fuptelea from Trochodendron and Tet- racentron, to which they have been frequently allied (Endress, 1986b). Examinations of pollen morphology and leaf architecture corroborated the isolated position of Euptelea (Praglowski, 1974; Wolfe, 1989). Nast & Bailey (1946) pointed out that the genus is of general ranalian affinity because it possesses several features of “the syndrome of primitive characters, cium, incompletely sealed carpels, and numerous ,* such as apocarpous gynoe- stamens originating in centripetal sequence. They suggested that Kuptelea is obviously not related to the ranalian families with monosulcate pollen and ethereal oil cells, but rather to those with tricolpate pollen and no ethereal oil cells (1.е., Ranunculaceae, Berberidaceae, Lardizabalaceae, Menispermaceae, and Trochodendraceae). The Sabiaceae were tentatively placed in the Ranunculales by Cronquist (1981). This study shows that the family is not a member of the order, but occupies a basal position in the eudicots. The exact relationships of Nelumbonaceae have not been certain, although their affinity to the Mag- noliidae is generally recognized. Tricolpate pollen, roots with vessels, and production of benzyl-iso- quinoline and aporphine alkaloids are among the characters that warrant removal of this family from Nymphaeales. Serological and cytological evidence shows the family does not belong to the Nym- phaeales (Simon, 1971; Raven, 1975). These char- acters plus absence of ethereal oil cells suggest a relationship to basal eudicots rather than the mono- sulcate magnoliids. Both Donoghue «€ Doyle’s (1989) morphological cladistic study and phyto- chemical evidence (Gottlieb et al., 1989) support such a position. 3. THE BASALMOST LINEAGE OF ANGIOSPERMS Although the Ranalian Hypothesis that Mag- noliidae represent the primitive angiosperms has been generally accepted, which lineage in the sub- class is the most primitive remains an extensively debated issue (Friis et al., 1987). Woody Mag- noliales with large strobiloid flowers and herbaceous Piperales with diminutive and simple flowers both have their proponents. Reports of chloranthoid and other non-magnolialean reproductive and vegeta- tive structures in the Early and Middle Cretaceous (Couper, 1958; Dilcher, 1979; Krassilov et al., 1983; Upchurch, 1984; Walker & Walker, 1984; Friis et al., 1986; Crane et al., 1986; Crane, 1989; Taylor & Hickey, 1990; Pedersen et al., 1991) challenge the status of Magnoliales as the most primitive angiosperms. Cladistic studies finding that Gnetales and Bennettitales are equally closely re- lated to angiosperms (Crane, 1985; Doyle & Don- oghue, 1986) add further controversy to the de- ate. Paleobotany has played an instrumental role in Volume 80, Number 3 1993 Qiu et al. 601 Molecular Phylogenetics of Magnoliidae elucidating the early evolution of angiosperms. A large body of fossil evidence of early angiosperms has been gathered in the last several decades, doc- umenting presence of several major lineages of primitive angiosperms in the Early and Middle Cre- taceous: Magnoliales, Laurales, Piperales, Nym- phaeales, Ranunculales, and “lower” Hamameli- dae (Couper, 1958; Doyle, 1969; Muller, 1970, 1981, 1984; Doyle & Hickey, 1976; Doyle et al., 1977, 1990; Hickey & Doyle, 1977; Dilcher, 1979; Hughes et al., 1979; Krassilov et al., 1983; Walker et al., 1983; Dilcher & Crane, 1984; Upchurch, 1984; Walker & Walker, 1984; Crane et al., 1986; Friis et al., 1986; Ward et al., 1989; Drinnan et al., 1990, 1991; Hughes & McDougall, 1990; Taylor & Hickey, 1990; Herendeen, 1991; Pedersen et al., 1991). These studies have greatly expanded our knowledge of early angiosperm evo- lution but have not answered the question of the origin of angiosperms. Further exploration of sed- iments of the preceding Early Cretaceous and Late Jurassic ages will be necessary before firm conclu- sions regarding the origin of the angiosperms are reached (Hughes & McDougall, 1987, 0 In a cladistic study of mostly morphological data, Donoghue & Doyle (1989) identified the Magno- liales as the basalmost lineage among angiosperms. However, they found that Nymphaeales are basal in trees only slightly less parsimonious (1 step lon- ger or 0.56% less parsimonious; the length of their shortest trees was 178 steps). Our constraint ex- periment (Fig. 5) demonstrated a similar small loss of parsimony for this alternate hypothesis (0.32% less parsimonious; see Results). Hamby & Zimmer (1992) reported that Nymphaeales and Piperales are paraphyletic to other angiosperms in their in- vestigation using ribosomal RNA sequences, but many important taxa were absent from their study. Martin & Dowd (1991) found Illiciales to be sister to all other angiosperms in their study of rbcS ~ amino acid sequences. In our analyses, we found only weak support for relationships at the base of angiosperms (Fig. З). Thus, the rbcL results, in spite of being resolved in favor of Ceratophyllum if outgroups were included, do not strongly support or refute any hypothesis of which group of Mag- noliidae is sister to all other angiosperms. 4. RELATIONSHIPS OF THE MAGNOLIIDAE TO MONOCOTS AND EUDICOTS The relationship between Magnoliidae and monocots has been discussed extensively (Dahlgren & Clifford, 1982; Dahlgren et al., 1985). It is widely believed that the two groups are closely related. Hegnauer (1977) reported that the Mag- noliidae (including Ranunculales) and monocots ex- clusively use the tyrosine-pathway for synthesizing cyanogenic compounds, whereas other dicots ex- ploit phenylalanine for the same purpose. In study- ing the distribution and biosynthesis of shikimate- derived compounds, Kubitzki & Gottlieb (1984) found that angiosperms are essentially composed of two groups, the * (mag- noliids, monocots, and caryophyllids) and the “ros- iflorean block" (the rest of the dicots). Previously, different pairs of the magnoliids and monocots have been proposed to account for the origin of the monocots: Ranunculales/ Piperales vs. Alismatales (Bessey, 1915), Nymphaeales vs. Alismatales (Takhtajan, 1969; Cronquist, 1968), Piperales vs. Arales (Burger, 1977), and Annonales vs. Dios- coreales (Dahlgren et al., 1985). Considering the heterogeneity of the Magnoliidae and a great num- ber of similarities between the subclass and mono- cots (Dahlgren € Clifford, 1982; Dahlgren & Bremer, 1985), it is not surprising to see diverse opinions about their relationship. Most, if not all, of these similarities are based on character states that are potentially symplesiomorphic or due ‘magnolialean block” such to parallel evolution. Although resolution of relationships at the base of angiosperms is weak in our study, several groups may be removed from the list of possible close allies to the monocots. The Ranunculales are a member of a well-supported eudicot clade and do not show any relationship to the monocots. The Nym- phaeales alone are unlikely to be sister to thes mono- cots because they show a strongly supported re- lationship to Amborellaceae, Austrobaileyaceae, and Illiciales. Any other major lineage of the Magno- liidae could be placed as a sister group to the monocots with only an insignificant loss of parsi- ony. Monophyly and origin(s) of the eudicots have been subject to a remarkable diversity of opinions. Interpretation of traditional phylogenetic diagrams such as those of Takhtajan (1959, 1969, 1980, 1987) would lead to a conclusion that + оїз аге olyphyletic and composed of several lineages in- dependently derived from various monosulcate an- giosperms with magnoliid features. Under the as- sumption that eudicots are derived from the monosulcates, Walker & Walker (1984) suggested that the monosulcate magnoliid families Wintera- ceae, Illiciaceae, Schisandraceae, and Aristolochi- aceae have ties with the subclasses Ranunculidae and Caryophyllidae and that the Lactoridaceae, 602 Annals of the Missouri Botanical Garden Piperaceae, Saururaceae, and especially Chloran- thaceae are linked to the Hamamelidae and through it to remaining dicots. In a cladistic investigation of basal angiosperms using morphological char- acters, Donoghue & Doyle (1989) concluded that Ranunculidae, Nelumbonaceae, and paleoherbs (in- cluding Lactoridaceae, Aristolochiaceae, Pipera- ceae, Saururaceae, Nymphaeaceae, Cabombaceae, and monocots) are more closely related to the Ham- higher" dicots than the Mag- noliales and Chloranthaceae. Furthermore, they amelidae and other “* found that in some of their most parsimonious trees ranunculids and hamamelids form a monophyletic group sister to the paleoherbs, whereas in others the ranunculids are dissociated from the hama- melids and united with the paleoherbs. Doyle & Hotton (1991) reported that the oldest tricolpate pollen from the late Barremian-early Aptian of Gabon exhibits two features of typical paleoherb pollen: the sculpture consists of intermixed large and small lumina (heterobrochate) and varies from coarser at the equator to finer at the poles (graded). The monophyly of eudicots (dicots with tricol- pate and derived pollen types, including Nelum- bonaceae and Takhtajan’s Ranunculidae but ex- cluding Illiciales) is strongly supported by this study. Fossil evidence from the Potomac group in eastern North America shows that all tricolpate pollen and derived types can be traced back to a morpholog- ically generalized complex of tectate-reticulate tri- colpate pollen, thus corroborating the hypothesis i 1975) This type of tectate-reticulate tricolpate pollen is of single origin of the eudicots (Wolfe et al., found in the extant Rananculales, Platanaceae, and Trochodendrales (Muller, 1984), which is consis- tent with placement of these taxa at the base of eudicots in the rbcL trees (Figs. 1—4; also see Chase 1993). Many basal eudicots are wind-pollinated, for ex- et al., ample, Buxaceae, Eupteleaceae, Menispermaceae, Platanaceae, Tetracentraceae, and possibly Gun- neraceae and Trochodendraceae (Crane et al., 1991). Environmental factors such as the increas- ing aridity in a period immediately after the Bar- remian (Doyle et al., 1977; Hickey & Doyle, 1977), when tricolpate pollen first appeared (Hughes & McDougall, 1990; Doyle & Hotton, 1991), might have played a critical role in the rise of eudicots (Walker & Walker, 1984). The preponderance of wind-pollination in basal eudicots suggests that the group could be primitively wind-pollinated. Further investigation of other wind-pollinated members of the Ranunculales and "lower" Hamamelidae is needed before we have adequate information to address this question. CONCLUSIONS This study of nucleotide sequences of the plastid protein-coding gene rbcL has identified the same major lineages as have many classification systems and previous cladistic studies of the Magnoliidae: (1) Magnoliales, (i) Laurales, (ui) Piperales, (iv) Nymphaeales, and (v) Ranunculales. The exact familial composition of some of these clades is sig- nificantly different from any previous hypothesis. The most conspicuous difference is that Amborella- ceae, Austrobaileyaceae, and Illiciales are strongly supported as members of the Nymphaeales clade. The relationship of these anomalous woody taxa to the Nymphaeales has only weak corroboration of shared absent features. Another novel feature of our results is that angiosperms fall into two mono- phyletic groups corresponding to the general pollen types of monosulcate and tricolpate, and they show a sister-group relationship. Our finding of the mono- phyletic eudicots lends strong support to pollen characters as major synapomorphies for these an- giosperms. In contrast, we cannot identify any non- molecular character to define the weakly supported monosulcate clade. Resolution of relationships at the base of angiosperms is low. The position of Ceratophyllaceae as a sister group to all other angiosperms is likewise controversial, but it is sup- ported, though not unequivocally, by the fossil ev- idence and the distinctive morphology of these aquatic plants. 'e conclude that our molecular survey of pu- tatively basal angiosperms gains great support from the congruence of the groups identified in this study with those recognized in classification systems and recent cladistic analyses of nonmolecular data. Our analyses of the rbcL sequence data were conducted merely under the criterion of parsimony, and did not rely on any previous hypothesis of evolution of morphological characters or relationships among basal angiosperms. We take this to be an indication that the molecular character systems contain his- torically relevant information. The general rela- tionships among basal angiosperms, as well as com- position of certain clades (e.g., Nymphaeales) identified in our rbcL trees, are novel and intrigu- ing, and we suggest that they be considered as alternatives to the traditional ideas about the phy- logeny of the early angiosperms. lronically, this molecular investigation adds more controversy in- stead of providing solutions to the questions sur- rounding early angiosperm evolution. А complete understanding of the origin of angiosperms cer- tainly requires multidisciplinary studies of both fos- sil and extant plants with morphological and mo- lecular approaches. Volume 80, Number 3 1993 Qiu et al. 603 Molecular Phylogenetics of Magnoliidae LITERATURE CITED ALBERT, V. A., M. W. Сназе & B. D. Менги. 1993. сан state weighting for cladisti f pro tein-coding DNA sequences. Ann. Missouri Bot. Gard. 80: 752-7 ARBER, E. A. N. & J. PARKIN. 1907. On a вш of i , Bot. 38: 29 W. M. Етен, Tos [^ trees 254: 554-558. Вапех, IL. W. € C. С. Nast. 1943. The p e md: morphology of the Winteraceae VII. Summary and conclusions. J. Arnold Arbor. 24: 37-47. B. С. L. Swamy. 1948. Amborella tricho- poda Ba il, a new morphological type vesselless 54. dicotyledon. J. Шш Arbor. 29: 245- & The dic and re- lationships of poesie J. Arnold Arbor. 30: 11220. Baranova, M. 1972. Systematic anatomy of the leaf Vu rus in the Vi bor M and some related fam- s. Taxon 21: 44 байы F. А. 1975. ү m fam. n., order Magnoliales. Phytologia 32: 304. -D. 1988 BEHNKE, Н. Sieve-element plastids, dera protein, and evolution of flowering plants: III. Mag noliidae. Taxon 37: 699-732. Bessey, C. E. 1915. зш plants. Ann. Missouri Bot. Са 64. The phylogenetic taxonomy of rd. 2: 1096 Е. К. 1988. The limits of amino acid sequence data in angiosperm phylogenetic reconstruction. Evo- lution 42: 795-803. BURGE 77. The Piperales and the monocots. Alte rnative e hypotheses for the origin of monocoty- ledonous flowers. Bot. Rev. (Lancaster) 43: 345- Piin Heresy revived: The monocot theory of a ques origin. Evol. Theory 5: 189-225 бий А. P. DE. 1824. Prodromus Systematis Na- tur E» Regni Vegetabilis Pars I. Treuttel & Würtz, Cabos. S. 1990. Wood anatomy and relationships of ap o Amer. J. Bot. 77 5 Мо а; anatomy and stem of Chlora wood anatomy of XUNG FON with comments on dicens vessellessness, and the origin of monocotyledons. IAWA Bull. 13: 3- 16. Chase, M. У. € Н. С. Hiis. 1991. Silica gel: A ideal material for field preservation of leaf emus for DNA studies. Taxon 40: 215-220 LTIS, R. С. OLMSTEAD, D. MORGAN, D. ALBERT. 71993. Phyloge- netics of seed plants: An analysis of nucleotide se- quences from the plastid gene rbcL. Ann. Missouri Bot. Gard. 80: 528-580. CHLONOVA, А. F. & T. D. Surova. 1988. Pollen wall ultrastructure of C lavatipolle nites incisus Chlonova and two modern species of Ascarina (Chlorantha- ceae). Pollen & Spores 30: 29-44. Couper, В. A. 1958. British Mesozoic microspores and ad grains. ие Abt. B, Palaophy- tol. 103: 75-1 S Р.В. 1985. the oed of angios Сага, 2 793. i. Paleobatamem нијеси: s es early radiation of st. Evol. 162: 165- & S. Lipcarp. 1990. Angiosperm radiation and patterns of Cretaceous palynological diversity. Pp. 377-407 in P. D. Taylor Larwood (editors), Major Evolutionary Radiations, Clarendon p analysis of seed plants erms. Апп. Missouri Bot. o 191. , E. M. FRIIS & К. R. PEDERSEN. 1986. Lower бое ous angiosperm flowers: Fossil evidence on early ана of dicotyledons. Science 232: 8 854. 9. MANCHESTER & D. L. DILCHER. 1991. Reproductive and vegetative structure of Norden- skioldia (Trochodendraceae), a vesselless dicotyle- ph CRONQUIST, А. 1968. The Evolution and d of Flowering jue mea Mifflin, Bosto 1981. An Integrated System of nee а of Flowering Bus Columbia Univ. Press, New York. 1988. T volution and Classification of кыса Plants, 2nd. ed. The New York Botanical Garden, New York. DAHLGREN, G. 1989. The last Dahlgrenogram System of classification of the dicotyledons. Pp. 249-260 in K. Tan (editor), Plant Taxonomy, Phytogeography and Related Subjects. Edinburgh Univ. Press, inb inburgh. DAHLGREN, R. General aspects of angiosperm evolution and macrosystematics. Nordic J. Bot. 3: 49. :MER. 1985. Major clades of the angiosperms. Cladistics 1: 349-368. & H. T. Слекокр. 1982. The ле dons: А Comparative Study. Academic Press, don. —— — — а P. Е. Yeo. 1985. The Families of the Monocotyledons. Structure, Evolution, and 0-57 in J. Hutchinson, The Genera of Flowering Plants, Vol. I. Dicotyledons. Clarendon Press, Oxford. DEPAMPHILIS, C. W. & J. D. PALMER. 1990. Loss of photosynthetic and chlororespiratory genes from the plastid genome of a parasitic flowering plant. Nature 348: 337-339 DILCHER, D. L. 19. Early angiosperm reproduction: An introductory report. Rev. Palaeobot. Palynol. 27: 1-328. The occurrence of fruits with affinities ceous sed- Archaeanthus: Ап early angiosperm from the Cenomanian of the west- 604 Annals of the Missouri Botanical Garden ern interior of North America. Ann. Missouri Bot. : 351-383. Dinos HUE, M. J. & J. А. роті. analysis of iode an am- amelidae. Pp. 17-45 in P. R. Crane & S. Blackmore (editors), Eu с ж: Fossil History of the Hamamelidae, 1: Introduction and ‘Lower’ Hamamelidae. Clarendon Press, Oxford. 1 1 ea Phylogenet DovrE, J. A. Cretaceous angiosperm pollen of the Atlantic Coastal Plain and its evolutionary sig- J. Arnold Arbor. 50: 1-35. nificance. & M. J. DoNocHUE. 1 а plant phy- logeny and the origin of angiosperms: An experi- mental cladistic approach. Bot. Rev. (Lancaster) 52: 431 32 CKEY. 1976. Pollen and leaves from the mid-Cretaceous Potomac group and their bearing on early angiosperm evolution. Pp. 139-206 in € B. Beck (editor), Origin = Early Evolution of An- giosperms. Columbia Un ress, New York. & C. L. Horrow. 1991. Diversification of 2 v esr 5 in a cladistic context. Pp . Blackmore & S arnes (editors), Pe ed cin Clarendon Press, Oxford. —, & . Warp. 1990. Early d taceous tetrads, zonasdlculate pollen, and Winter e. I. Taxonomy, pa diu and Шоке Д “Cladistic analysis implications. Amer. J. Bot : 1544-1557, 1558-1568 L P BIENS, A. DOERENKAMP & S. JARDINE. 1977 ieu de у pollen from the pre-Albian lower Ce. ous of equatorial Africa. Bull. Centr. Rech. Ex- 73. “Bull -] DRINNAN, A. N. P. R. oen E FRIS € K. R. 1990. C lowers ш the Po- tomac Group (mid-Cretaceo astern North America. Bot. Gaz. 151: 370 во она 91. Angio- sperm "Tm and tricolpate pollen wit oa. af- i d from the Potomac Group (mid-Cretaceous) of n North пика Amer. ot. 78: 153-17 Pa ERE. F., Е. KRENDL, E. HABEL ER & W. SAUER. Viam Сонону тшше and evolution in prim- e angiosperms. Taxor : 337-3 EICHLER, А. W. 1890. Syllabus der Vorlesungen über specielle und der inisch-pharmaceutische Botanik. Fünfte Auflage. Gebrüder Borntraeger, Berlin. ки Р. К. 1980. The reproductive structures and tematic Lys of en Austrobaileyaceae. Bot i hrb. Syst. 101: 393-4 983. The early "um d of Aus- trobaileya. Bot. Jahrb. Syst. 103: 97. 86a. Reproductive structures and phylo- genetic вон Бајо | of extant primitive angiosperms. PL Е Evol. 152: 1-2 986b. Floral structure, systematics, and phy- ub in Trochodendrales. Ann. Missouri Bot. Gar 73: 297-324. . 1987. The Chloranthaceae: Reproductive structures and eee. position. Bot. Jahrb. Syst. 226 R. HONEGGER. 1980. The pattern of the Austrobaileyaceae and its phylogenetic significance. Grana 19: 177-182. ENGLER, А. & K. PRANTL. 1891. Die Natiirlichen Pflan- бен Teil 3, Abteilung 2. Wilhelm Engel- eipzi er nm J; 1978. Cases in which parsimony and compatibility methods will be positively misleading. Syst. Zool. 27: 401-410. Етен, У. M. 1971. Toward defining the course of evolution: Minimal change for a specific tree topol- ogy. Syst. Zool. 20: 406-416. Frus, E. M. € W. L. СВЕРЕТ. | Time of арреаг- ance of floral features. Pp. 145-179 in E. M. Friis, W. С. Chaloner & P. R. жо (editors), The Ürigins of Angiosperms and Their Biological Consequences. Cambridge Univ. Press, Cambridge , W. G. CHALONER & P. R. бим. troduction to angiosperms. Pp. 1-15 in E. W. С. Chaloner & P. R. Crane не Тће Origins of Angiosperms and Their Biological Consequences. Cambridge Univ. Press, Cambridge. CRANE & K. R. PEDERSEN. 1986. Floral ev idence for Cretaceous chloranthoid angiosperms. Nature (London) 320: 163-164. a E. M., - E. Grannasi, M. T. Cece, С. J. AILEY, M. DURBIN, D. HENDERSON & G. ZURAWSKI. 1990, Chloroplast DNA sequence from a Miocene | nolia species. Nature (London) 344: 656-658. байы О. R., М. А. C. KAPLAN, К. Ковик! & J. К. Тогеро Barros. . Chemical dichotomies in the Magnolialean complex. Nordic J. Bot. 8: 437- Намву, R. К. & E. A. ZIMMER. 1992. Ribosomal RNA as a phylogenetic tool in plant systematics. Pp. 9] in P. S. Soltis, D. E. Soltis & J. J. Doyle (editors), Molecular Systematics of Plants. Chapman & Hall, New York. HEGNAUER, В. 1977. Cyanogenic compounds as sys- tematic markers in Tracheophyta. Pp. 191-209 in K. Kubitzki (editor), Evolution and Classification of Кына Categories. Pl. Syst. Evol. Suppl. 1. Springer, Vienna. HERENDEEN, P. S. 1991. Lauraceous wood from the mid-Cretaceous Potomac Group of eastern North America: Paraphyllanthoxylon marylandense sp. nov. Rev. Palaeobot. Palynol. 69: 277-290. Hickey, L. J. & J. A. Поуге. 1977. Early Cretaceous fossil evidence for To E evolution. Bot. Rev. red 43: 3-10 J. A. WOLFE. Tues The bases of angio- sperm к Vegetative morphology. Ann. Mis- souri Bot. Gard. 62: 538-589. Новев, Н. 1977. The treatment of the monocotyledons in an evolutionary system of classification. Pp. 285- 298 in K. Kubitzki (editor), Evolution and Classifi- nn of Higher Categories. Pl. Syst. Evol. Suppl. 1. Springer, Vienna. Hucues, N. К. € A. B. McDoucaLL. 1987. Records of angiospermid pollen entry into the English ihi Cretaceous succession. Rev. Palaeobot. Palynol. 50 255-272 Barremian-Aptian angio- spermid pollen records from southern England. Rev Palaeobot. Palynol. 65: 145-151. , С. E. Drewry & J. F. Laine remian earliest angiosperm pollen. Palaeontology 22: 13-535. Ноумн, K.-L. 76. L'arrangement du pollen du genre Schisandra (Schisandraceae) et sa signification phy- Volume 80, Number 3 1993 Qiu et al. 605 Molecular Phylogenetics of Magnoliidae cake chez les angiospermes. Beitrage zur Biol- e der Pflanzen 52: 227-2 бше V. A., P. V. SHILIN & V. А. VACHRAMEEV. 1 duci flowers from Kazakhstan. Rev. Palaeobot. Palynol. 40: 91-113 KUBITZKI, К. & O. R. GOTTLIEB. 1984. Micromolecular patterns and the evolution and major classification of angiosperms. Taxon B 375-391. KUPRIANOVA, L. 7 nological data n the history of the elas Pollen € Spores 9: 95-100 LAMMERS, T. G F. SrUESSY & O. M. Зима. 1986. Systematic relationships of the Lactoridaceae, an en demic family of the Juan а Islands, Chile. 6. |. Syst. Evol. 152: 243- Law, Y- W. 1984. A вале study on id taxonomy of the family Magnoliaceae. a Phytotax. Sin. 22 89-109. [In Chinese, English summary. Leroy, J.-F. 1977. A compound ovary with open car pels in Winteraceae е Ev Busen im- plications. Science 196: 977-978. Les, H. D. 19 The origin and 2 m of the Cer- atophyllaceae. а 37: 326-435. , D. К. GanviN & C. F. эзы а 1991. Mo- pondas о. history of ancient aa an- giosperms. Proc. Natl. Acad. U.S.A. 88: 19e 23. LINDLEY, J. 1833. Nixus Plantarum. Ridgway et Filios, ondon. LoconTE, Н. & D. W. STEVENSON. ы з of the Magnoliidae. Cladistics 7: 267- MarTIN, Р. С. & J. M. Dowp. 1991. um of an- giosperm phylogeny bu Uae sequences. Ann Missouri Сань: 78: 337 Money, L. L., . BAILEY & B С. L. Swamy. The E and cipere of the Monimi- aceae. J. Arnold Arbor. 31: 372-403 MULLER, J. 1970. Palynological evidence on early dif- ferentiation of angiosperms. Biol. Rev. 45: 417-450 1981 ossil pollen Mir ‘of extant angio- Rev. o 47: 1-142 Significance of fossil pollen for angio- Ann. Missouri Bot. Gard. 71: 419- Bot. sperms sperm + 443. Nast, С. G. & I. W. Вапеу. 1946. Morphology of А and comparison with 7rochodendron. Ј. dd Arbor. 27: 186-192. Nori, H. P. 1985. Notes on Magnoliaceae, with a revision of Pachylarnax and Elmerrillia and the Мање 2 of Manglietia and Michelia. Blu- mea 31: 65-121. Parmer, J. D., R. К. JANSEN СНАЗЕ & J. R. МАХНАКТ. variation and plant phylogeny. Gard. 75 206 PEDERSEN, К. R., P. R. CRANE & E. M. Friis. 1991. Fruits from the mid-Cretaceous of North America ith pollen grains of the Calv ides ds type. Grana ‚ H. J. Michaels, M. W. 1988. Chloroplast DNA Ann. Missouri Bot. 30: = PRAGLOWSKI, J. 1974. ы pollen morphology of the rochodendraceae, Tetracentraceae, Cercidiphylla- ceae, and Eupteleaceae, with reference to taxonomy. Pollen & Spores 16: 449-467 Raven, P. H. 1975 bases of кы RR pny log- eny: Cytology. Ann. Missouri Bot. Gard. 24- 1987. Evolutionary anal- м К. € M. T. CLEGG. ysis of pes DNA sequences. Amer. Naturalist 130: -8100. SIMON, J. T 1971. Comparative зегојову of the order Nymphaeales II. Relationships of the Ny аннан еае and Nelumbonaceae. Aliso 7: 325-350 SuH, Y., L. B. THIEN & E. A. ZIMMER. 1992. Phylogeny of the basal flowering plants: Evidence from the se- quences of 265 ribosomal DNA. 5th Intl. Symp. International Bi арт of ied Biosystematics, St. Louis, Missouri, p. SWOFFORD, D. L. 1 History Survey, Champa 1959. Die Esel der Angiospermen. Gustav Fischer Verlag, Jena. —— — —. 1969. Flowering M Origin and Dispersal. Olives kA d Edinburg ———— Outline i classification of flowering m sd Bot. Rev. (Lancaster) 46: 225-359. "llinois Natural Abst Pat Р 3.0s. 1987. Systema Magnoliophytorum. Nauka, Leninopoli. [In Russian. TAYLOR, dee & L. J. Hickey. 1990. An Aptian plant villi а ached leaves and Сона: ae for Mode origin. Science 247 04. & ————. 1992. Phylogenetic evidence for the ча eous origin of angiosperms. Pl. Syst. Evol. 180: -136. THORNE, E e 1968. Synopsis of a putatively phylo- genetic classification of the flowering plants. Aliso 6: 5776 197 6. А pary eee = the Angiospermae. Pp. 35 n M. К. Hec mF Steere & B. Wallace editors so ne Vol. 9. Plenum Press, Yor 992. An d phylogenetic classification of the p d plants. Aliso 13: 365-389. UrcuuncH, С. R. 1984. Cuticle evolution in early taceous пасти from the Potomac ics ‘of Virginia and Maryland. Ann. Missouri Bot. Gard. 71: 522-550. WALKER, J. W. 76a. Comparative pollen morphology lution of Angiosperma, Columbia Univ. Press, New York. 1976b. Evolutionary significance of the exine in the pollen of primitive p pie Pp. 251-308 in I. K. Ferguson & J. ller (editors), The Evo- lutionary Significance of "ie Exine. Academic Press, London —— — & J. A. DoYLE. 1975. The bases of angiosperm Er Palynology. Ann. Missouri Bot. Gard. 62: 3. WALKER. Ultrastructure of lower Terea angiosperm pollen and the origin and T evolution of oe plants. Ann. Missouri Bot. Gar 464–. RENNER & re ;. WALKER. 1983. Win- ТОИ ; pollen in the m Cretaceous of Israel Early evidence of a magnolialean angiosperm family. Science E "а 13-1279: Warp, J. V., Doyle € C. L. Horron. 1989. Probable tol — ш pollen from the early Cretaceous. Pollen & Spores 31: 113-132. WoLFE, J. A. Leaf. arc Fri da gos sis of the Hamamelidae. Pp. 75-104 in P е € 3 Blackmore (editors), Evolution, зала 'and Fos. 606 Annals Mou TES Garden sil History of the Hamamelidae, Vol. 1. Introduction and ‘Lower’ Hamamelidae. Clarendon Press, d. & ————, J. А. ПозтЕ € V. M. РАСЕ. 1975. The bases of angiosperm JR ‚ар Ann. Mis- souri Bot. Gard. 62: YOUN . A. 1981. | c aes primitively dito Syst. Bot. 6: 313-330. ZURAWSKI, C. & M. T. Стесс. 1987. Evolution of higher-plant chloroplast DNA-encoded genes: Impli- cations for structure-function and Wr stud- s. Ann. Rev. Plant Physiol. 38: 391-418. PHYLOGENETIC Melvin R. Duvall, Michael T. Clegg,’ Mark W. Chase,** W. Dennis Clark,’ HYPOTHESES W. John Kress, Harold G. Hills, FOR THE MONOCOTYLEDONS Luis E. Eguiarte, James F. Smith, CONSTRUCTED FROM Brandon S. Gaut; Elizabeth A. Zimmer, Foes SEQUENCE DATA! and Gerald H. Learn, Jr? ABSTRACT A sequences for the plastid locus that encodes the large subunit of ribulose 1,5-bisphosphate carboxylase oxygenase (rbcL) were determined for 18 spe 'cies of monocotyledons in 15 families. den ata were analyzed together ith sequences for 60 other monocot species in a total of 52 families by the maximum likelihood method "iae ing one, presumably optimal, topology. An additional 26 species were ас ded (104 зна monocot nt. and analyzed by the pour. method with an outgroup of 18 dicot species nu 109 trees of 3,932 steps. The rbcL data show at least moderate support for seven lineages corresponding to the following ps superorders, or кык Агесапае; Asparagales (excluding Hypoxidaceae) plus deis Cyclanthanae plus Pandananae; Dioscoreales; Or- chidales; Typhales; and Zingiberanae. Six clades corresponding to families or genera are well supported, anu Agavaceae, Asphodelaceae, Bromeliaceae, Hypoxidaceae, Poaceae, and Tradescantia. The two, earliest diverging multispecies clades in our rbcL phylogenies, Alismatanae and Aranae, are only weakly supported, and НАН Commelinanae, and Lilianae are paraphyletic. In all Sue Acorus calamus is Wc e ur isolated as the sister species to the remaining species of monocotyledons Innovations for the manipulation of nucleic acids lineages of т tyledons based on rbcL se- and advances in computer technology now permit quence data hou "a attempted previously for: phylogenetic analysis of homologous sequences of (1) Arecaceae (Wilson et al., 1990); (2) Bromeli- DNA from large numbers of organisms. These base- aceae (Clark et al., in prep.); (3) Poaceae (Doebley to-base comparisons of nucleotides afford the high- — et al., 1990); and (4) Zingiberales (Smith et al., est possible resolution of inherited mutations in 1993). These initial attempts met with a disap- DNA molecules and can be applied to questions of pointing lack of resolution, in some cases because higher-order plant systematics. A locus that has of the low substitution rate of rbcL, but strongly been selected for such studies by molecular sys- — suggested that these data would have greater phy- tematists is the plastid gene that encodes the large logenetic utility when applied to more divergent subunit of ribulose 1,5-bisphosphate carboxylase’ lineages. Our goal here was to reconstruct higher- oxygenase (rbcL). Phylogenetic reconstructions for order phylogenetic relationships among the mono- ' For computer analyses we thank С. J. Olsen (University of Illinois, Urbana), R. Overbeek (Mathematics and Computer Science, Argonne National Laboratory, Argonne, Illinois), the Concurrent Was nee Consortium, Whitkus, P. Lowe (both, University of California, Riverside) and M. Berres (University of Minnesota, 5t. Paul). We also thank D. Jarrell (University of California, Riverside), R. Olmstead (University of Colorado. a and others for unpublished rbcL s sequences; J. French (Rutgers University, Piscataway, New Jersey) who p n extracts for Araceae; and W. Armstrong (Palomar College, San Marcos, California) for identification н н specie This work was supported by aran from National Science Foundation to M. Duvall (BSR-9002321), M. € se (BSR. 8906496), and W. D. Clark (BSR-8904637); National Institutes of Health (СМ45 144) to M. Clegg: Sell dede Program, Smithsonian Institution to W. J. Kress and E. Zimmer; and American Orchid Society to Cha * Department of Botany and Plant Sciences, University of California, Riverside, California н a is ' Department of Biology, University of North Carolina, Chapel Hill, North VUE 27599, ' Present address: Royal Botanic Gardens, ce Richmond, Surrey TW9 3 ' Department of Botany, Arizona State University, Tempe, Arizona 85287, 6 Б раем of Botany, NHB-166, National Museum of Natural History, ee Institution, Washington, D.C. 20560, ' Centro de Ecologia: Universidad Nacional Autonóma de México, Apdo. Postal 70-275, 04510 Mexico, D.F., Mexico 8 Laboratory of Molecular Systematics, Smithsonian Institution, Suitland, Maryland 20746, U.S ° Present address: Department of Biology, Boise State University, Boise, Idaho 83725, U.S А. ~ ANN. Missour! Вот. Garp. 80: 607-619. 1993. Annals of the Missouri Botanical Garden cotyledons using DNA sequence data from the rbcL locus for a set of species that represented the breadth of taxonomic diversity in the group. MATERIALS AND METHODS DNA sequences of rbcL from 104 species in 52 of 104 monocot families were analyzed (see Ap- pendix in this issue for species, locations of vouch- ers, sequencing methods, extent of sequences, and ~ authors). Eighteen of those sequences (Table 1 were produced specifically for this study by the following methods. Total DNA was extracted from fresh leaf tissue by the method of Doyle & Doyle (1987) for 16 species, the method of Palmer (1986) for Cyperus alternifolius, and the method of Jarrell et al. (1992) for Lemna minuscula. Approximately 1 ug of each DNA preparation was used to provide template for Taq-mediated amplification of the rbcL gene using the protocol provided with 7aq DNA polymerase by the supplier (Promega Corp.). The primers for the reaction were two highly conserved sequences from the rbcL coding region of Spinacia rst 30 base pairs of that sequence and the 3' (reverse) 1380 on the complementary strand. Amplifications uti- oleracea. The 5' primer consisted of the primer corresponded to positions 1351 to lizing these primers produced 1,320 base pairs of the coding sequence of rbcL. Primers that are external to the coding sequence occasionally re- sulted in successful amplifications of more of the rbcL coding sequence. For Sparganium ameri- canum, а 27 base pair sequence at the p region for rbcL derived from positions 537 53768 in the plastid genome of Oryza sativa was used as the 5' primer. For Gymnostachys anceps, Tradescantia aff. pallida, and Typha latifolia, a 24 base pair sequence at a ribosome control site downstream of the coding sequence derived from positions 59146-59123 (complement) in the plas- tid genome of Мсопапа tabacum was used as the 3' primer. Up to 1,428 base pairs of rbcL were produced, depending on the primers that were used e rbcL se- quences from the monocotyledons analyzed here, in the amplification reactions. For th 5.7% of the sequence data was missing largely because of the use of internal primers in the am- plification reactions. Single-stranded DNA was produced in a second round of amplification using the double-stranded product as template and the two primers individ- ually. These single-stranded products were precip- itated (7.5% polyethylene glycol-8000, 0.94 M NaCl), the DNA pellets were washed twice with ethanol, and sequencing was accomplished by the dideoxynucleotide chain-termination method using a set of conserved, synthetic, internal rbcL primers (obtained from G. Zurawski, DNAX Corp.). Both the rbcL coding strand and the complementary strand were sequenced for all species. The 18 rbcL sequences produced for this study were entered into GenBank" under the accession numbers list- ed in Table 1. The maximum likelihood method of phylogenetic analysis was selected for use here because it is less biased by the heterogeneous nucleotide substitution rates that have been observed for rbcL among different lineages of monocotyledons (Gaut et al., 1992) than are other methods (Felsenstein, 1981). Furthermore, under this method, likelihood scores may be calculated for alternate topologies and sub- jected to comparative tests of statistical signifi- cance. However, only a subset of the entire rbcL database could be analyzed since the method is computer-intensive. [To establish an estimate of the computer burden, a maximum likelihood anal- ysis of 79 rbcL sequences using the program, DNAML in PHYLIP 3.42 (Felsenstein, 1991), was executed on a CRAY “Y-MP8 puter. The analysis spanned 11 days with a run time of about 51 hr. At this speed, ten ex- ecutions of the program with different input orders /864" supercom- ~ typically recommended) of the 79 rbcL sequences would have required over three months of real time utilizing more than 20 CPU days on this machine. ] Consequently, 78 of the original 104 monocot spe- cies were selected for maximum likelihood analysis with Saururus cernuus as the outgroup (79 species total). Each of the following taxa, Agavaceae, Are- caceae, Bromeliaceae, Nolinaceae, Poaceae, and Zingiberales, invariably constituted а monophyletic lineage in preliminary phylogenetic analyses that included multiple species in each. The subset was thus selected to contain representatives from each of the 52 families while excluding some of the multiple species in these six taxa, preserving the potential for investigating higher order relation- ships. The species that were retained in the subset are: Agavaceae — Manfreda maculosa and Yucca recurvifolia; Arecaceae— Caryota mitis, Phoenix reclinata, and Serenoa repens; Bromeliaceae— Aechmea chantinii and Tillandsia elizabethae; Nolinaceae— Nolina (Beaucarnea) recurvata, Poaceae— Oryza sativa and Zea mays; and Zin- giberales— Calathea loeseneri, Costus barbatus, Globba curtisii, Hedychium gardnerianum, Hel- atha, Maranta leuconeura, Musa cavendishii, Orchidantha fimb Phenako- spermum guyannense, Ravenala madagascar- iconia latis riata, iensis, and Tapeinocheilos ananassae. Volume 80, Number 3 1993 Duvall et al. Phylogenetic Hypotheses for the Monocotyledons 609 TABLE 1. Eighteen species for which rbcL was sequenced for this study. Vouchers (herbaria), accession numbers for the rbcL sequences submitted to GenBank, superordinal, ordinal, and familial = are given. Identities for the remaining 104 species that were analyzed may be found in the Appendix for this issu Vouchers enBank Species (herbaria) Family accession Aranae Arales Acorus calamus L French 232 (CH) Acoraceae M91625 Gymnostachys anceps R. Br. Howard Bo (FTG) Araceae M91629 Fait stratiotes L. French 233 (CH) Araceae M96963 Lemna minuscula Herter Duvall ener (UCR) Lemnaceae M91630 Bromelianae Typhales Typha latifolia L. Bradley 24974 (GMUF) Typhaceae M91634 Sparganium americanum Nutt. Chase 257 (NCU) Sparganiaceae М91633 Commelinanae Commelinales Tradescantia aff. pallida Bradley 24980 (СМОР) Commelinaceae L05041 Tradescantia zebrina hort. ex Bosse Bradley 24980 (GMUF) Commelinaceae L05042 Cyperales Cyperus alternifolius L. Duvall 19920602 (UCR) Cyperaceae M91627 Lilianae jid Aloe vera (L.) Burm. Bradley 24977 (СМЏЕ) Asphodelaceae L05029 Haworthia subfasciata Baker Bradley 24978 (GMUF) Asphodelaceae L05035 Chlorophytum comosum (Thunb.) Jacques Bradley 7331 (GMUF) Anthericaceae L05031 Clivia miniata Regel Bradley 24976 (GMUF) Amaryllidaceae L05032 Nolina (Beaucarnea) recurvata (Lem.) Нет]. Peterson 12606 (US) Nolinaceae L05030 Liliales Iris х germanicum L Bradley 25976 (GMUF) Iridaceae L05037 Medeola virginiana (L.) Merrill Bradley 24972 (GMUF) Uvulariaceae M91613 Zingiberanae Zingiberales Maranta leuconeura E. Morris Bradley 24979 (GMUF) Marantaceae L05040 Hedychium gardnerianum Gawl Bradley 24975 (GMUF) Zingiberaceae M91628 The analysis was executed on the fastest existing al., 1986; Ritland & Clegg, 1987). Regional and computer, the Touchstone Delta Parallel Process- ing Supercomputer, using "fastDNAml version 1.0.3" (Olsen et al., 1992). Thirty-three replica- tions of the analysis were performed with different randomly determined orders of input. The “cate- gories option," which permits specification of dif- ferent substitution rates by codon position, was invoked using relative rates of 1.00, 0.85, and 5.80 for first, second, and third codon positions, respectively. These values are based on reported substitution rates at each codon position (Clegg et local branch swapping were employed. This portion of the analysis, which included data from 74 o the 79 species, consumed 125 hours of computer After the initial analysis was performed, the remaining five species (Aletris farinosa, Burman- nia biflora, Sparganium americanum, Stegolepis allenii, and Турћа latifolia; total species: 79) with equivocal phylogenetic positions in preliminary analyses were added to the optimal topology from the previous step with 50 (of 120 possible) random 610 Annals of the Missouri Botanical Garden TABLE 2. Distribution of polymorphic sites, constant phyletic clades were conducted. The selected spe- sites, and sites shared by two or more species (i.e., in- formative sites), among codon positions of ad Sites are calculated for 1,428 base pairs and 79 species. substitutions were optimized on the topology of Figures 4 and 5 Base Codon position ot First Second Third sites olymorphic sites 221 210 430 861 Constant sites 255 266 46 567 Informative sites 137 124 394 655 input orders. Rearrangements of the optimal to- pology from this step were performed to arrive at a final result. Kishino & Hasegawa (1989) tests of the likelihood scores associated with topologies pro- duced by each step of the analysis (available from M. Duvall on request) were performed. Bootstrap and decay analyses were implemented with. РАЏР Version 3.0s (Swofford, on a Macintosh IIfx for the rbcL data set from 78 species of monocots. А set of 18 trees of length 3,117 was determined with more than ten repli- cations of the input order. Trees one step longer (486 trees) and two steps longer (6,237 trees) were also determined. (An attempt to determine all trees up to four steps longer using a Macintosh Quadra 700 was aborted after 13.8 days of continuous execution. The analysis produced 14,873 trees occupying over 6 Megabytes of memory and was estimated to be 10% complete. Available computer memory was the limiting factor so that the time to perform read-write operations of treefiles from ex- ternal memory became prohibitive.) Bootstrap analysis with 200 subsamples of the original data matrix was performed with local (i.e., "nearest neighbor interchange" or “NNI”) branch swap- ping. Three tests to ascertain the effect of constraining topologies so that selected species occupied mono- cies groups were: (1) Commelinanae (13 spp.); (2) Alismatanae (4 spp.), and Aranae (4 spp.), to- gether with Acorus calamus; and (3) each of nine superorders sensu Dahlgren et al. (1985). These constraints were imposed on parsimony analyses with 10 replications of the input order and NNI branch swapping. For pragmatic reasons the parsimony method was selected for larger-scale analysis of rbcL se- quences ( 104 monocot species). Eighteen dicot spe- cies were selected as an outgroup as suggested by a more inclusive analysis of angiosperms (Chase et al., 1993). Ten replications of the input order were executed with the “steepest descent" option in- voked and global (*'tree-bisection and reconnec- tion" or “TBR”) branch swapping employed. АП equally most-parsimonious trees were saved. REsULTS Of the 1,428 bases analyzed for 79 species, 861 were polymorphic and 50% of these were at third codon positions (Table 2). Of the 861 polymorphic sites, 655 were shared by two or more species and of these, 60% were at third codon positions. The results of our phylogenetic analyses show greater congruence with the taxonomic system of Dahlgren et al. (1985) than with other contem- porary systems (reviewed in Goldberg, 1989). Ref- erences to taxa in this report thus follow that sys- tem. The topology resulting from maximum likelihood analysis (Fig. 1) had an associated likelihood score of —18,878.21. Note that other topologies exist which have likelihood scores that are not signifi- cantly different from that of Figure 1. Bootstrap values and decay indices for analyses of 79 species 2, Table 3). up to two steps longer than the shortest trees were are given (Fig. Recall that only trees determined so that clades supported with decay indices of two may also be supported at higher, FIGURE 1. produced by the maximum likelihood ues with a log likelihood of — 3 (Olsen et al., likelihood score produced the t — Topology for 78 monocotyledons and the outgroup dicotyledon Saururus cernuus (79 spp. wi 18, ne .21. The analysis was executed usin of the tree with the largest aximum r likelihood distance units as calculated under the hide са specified above. Species are indicated by haen only. For complete binomials see ssu Appendix in this is Volume 80, Number 3 Duvall et al. 1993 Phylogenetic Hypotheses for the Monocotyledons Tapeinocheilos Calathea Ravenala == Philydrum Г Tradescantia aff. pallida lr T. zebrina T. soconuscana Pontederia Anigozanthus wiea Sansevieria Nolina recurvata Xanthorrhoea Burmannia —— 1. Sphaeradenia y AA Freycinetia Vellozia Alstroemeria Burchardia Colchicum Chamaelinum Smilax Medeola Lilium Aletris Pleea Gymnostachys Spathiphyllum emna Pistia Potamogeton ___ . qe Ali sma 9 AA+ Sagittaria graminea S. latifolia Acorus Saururus 612 Annals of the Missouri Botanical Garden а= ага la (2 spp.) N Helicon 5 62 Orchidantha (а 93 Миза жыша || Calathea c I Maranta Ф 1 Phenakospermum - à SE Ravenala р 155 155 ,—— Zingiberaceae (2 spp.) Е O o i ——179 ——— Tradescantia (3 spp.) > om Bromelianae (3 spp.) Stegolepis 00 | 1 — —IL—— Bromeliaceae (2 spp.) «y | 59 150 q Poaceae (2 spp.) | о | 0 w* Elegia ш { 78 Flagellaria == 1 Typhaes (2 spp.) y 100 ries 90 73 erus » 7 1 Oxychloe ч ————1. ” ү си Ф 100 ms a О | Phoenix ы 100 Serenoa 2 I Orchidales (2 spp.) Dm 98 Burmannia Ф ———7 12: Hypoxidaceae (2 "d 99 „9 e A S 2 spp.) > 100 Chlorophy |: 100 e Asphodelaceae | 3 spp.) Bg 50 96 Sansevie = AL Nolin ш 7 1 Danae (а омеа, 2 84 Dia — етн Ф 5 4 Суап о 100 LOT 1—— РЕНЕ "(о S ке pa 2 UN 53 95 Freycinetia Шода 9 4 p ит ba ee 199 _ prm Burchardia Istroemeria Melanthiales (3 spp.) 69 ge tia emna 53 He *—— Spathiphyllum ия ys 67 пр = al (2 spp.) Potamogeton Acorus Saururus 9E&UEg]Eeulsi| v әеиелу Volume 80, Number 3 Duvall et al. 613 1993 Phylogenetic Hypotheses for the Monocotyledons TABLE 3. Support for clades corresponding to taxa sensu Dahlgren et al. (1985) by rbcL data for 78 species VARIA кое Parenthetical numbers immediately following taxa indicate the number of species analyzed. Pons 200 replicates) and decay indices (up to two T longer I the most parsimonious topologies) are given. (5 le eel with decay indices of two may also orted at a higher, undetermined, decay index. Also given are the numbers of substitution events supporting ы ie de the number of nonhomoplastic synapomorphies (appearing parenthetically following branch lengths) in the maximum likelihood topology (Fig. 1). Bootstra Taxa (number of species) values (96) Decay indices Lengths Alismatanae (4) 67 > 2 37 (3) Агапае 53 = 14 (1) Arecanae (3) 100 > 2 31 (4) Bromelianae (8) 50 0 Bromeliaceae (2) 100 = 2 14 (0) yphales (2) 78 > 2 21(11) Commelinanae (13) < 50 0 = Cyperales (5) 7 > 2 26 (1) Poales (3 of 4) 59 = 2 21 (2) Poaceae (2) 100 = 2 66 (7) Tradescantia (: 100 => 48 (3) Cyclanthanae (1) plus Pandananae (1) 95 >2 13 (1) Lilianae (32) « 50 0 = Asparagales (14 of 16) plus Iridaceae (2) 84 = 2 13 (1) Agavaceae (2) 90 = 2 3(0) Asphodelaceae (3) 100 =D 21 (1) Hypoxidaceae (2) 100 > 2 18 (1) Dioscoreales (2 of 3) 98 > 2 27 (2) Orchidales (2) 100 > 2 23 (1) Zingiberanae (11) 93 a2 20 (1) Costaceae (2) 72 > 2 ingiberaceae 65 = 2 Monocotyledons re Acorus calamus (77 spp.) T3 2 13 (1) undetermined, values. Seven lineages found in the maximum likelihood tree that have associated boot- strap values at or above 78% and decay indices greater than or equal to two correspond to orders, superorders, or combinations of these. These are: (1) Arecanae (3 spp.); (2) Asparagales (excluding ~ Hypoxidaceae: 14 spp.) plus Iridaceae (2 spp.); (3 Cyclanthanae plus Pandananae (1 spp. each); (4) Dioscoreales (2 of 3 spp.); (5) Orchidales (2 spp.): (6) Typhales (2 spp.); and (7) Zingiberanae (11 spp.). Six clades of confamilial or congeneric spe- cies are supported by the rbcL data at a bootstrap value of at least 90% and a decay index of at least Bro- Poaceae, and Trades- two including: Agavaceae, Asphodelaceae, meliaceae, Hypoxidaceae, cantia. Parsimony analysis of 122 species produced а set of 109 equally parsimonious trees of 3,932 steps over the 1,428 characters (see overview of strict consensus tree, Fig. 3). These trees have consistency indices (exc luding uninformative char- acters) of 0.267 (the low value reflecting the large number of species) and retention indices of 0.65: One of these 109 trees was arbitrarily selected and is given in detail (Figs. 4, 5) to enumerate genera and illustrate comparative branch lengths. FIGURE 2 50% of the bootstrap topologies are shown. Selected ta up to two steps lon ger are indicated as vertical bars of varying thickness overlaid o Majority rule (50%) consensus tree for 79 species. The bootstrap analysis was conducted using PAUP on the bootstrap topology. The thickest lines indicate clades tae in trees at least two steps longer. The thinnest lines indicate clades supported only i in maximum parsimony tree Annals of the Missouri Botanical Garden Poales (13 spp.) (LL) eeeo»eog Cyperales (5) —1 Typhales (2) Bromeliaceae (7) | Commelinaceae (3) Zingiberanae (14) = | Агесапае (7) Asparagales (21), Orchidales (2), Iridaceae (2), and Melanthiales (1). Velloziales (1). Dioscoreales (2 of 3), and Burmanniales (1). Alismatanae (4), Arecanae (4), and Melanthiales (1). = pce Lilianae (7), fe | Cyclanthanae (1), Pandananae (1), and Acorus calamus Outgroup: Dicotyledonous Paleoherbs. Volume 80, Number 3 1993 Duva al. 615 uide Hypotheses for the Monocotyledons The maximum likelihood tree (79 spp., Fig. 1) and the resolved portions of the consensus tree produced by parsimony (122 spp., Fig. 3) are argely congruent with respect to the constituent species of the seven lineages listed above and the order of divergence of those lineages. Exceptions are: (1) Dioscoreales (excluding Smilax glauca) diverge earlier in the parsimony trees; (2 — one species each of Cyclanthanae, Pandananae, and Velloziales make up an isolated clade in the max- imum likelihood tree, which is found embedded within а clade of seven species of Lilianae in the parsimony tree; (3) Burmannia biflora is found in Asparagales in the maximum likelihood tree but with two species of Dioscoreales in the parsimony trees; and (4) Aletris farinosa occupies an isolated clade near the base of the maximum likelihood tree but is embedded within the clade of Asparagales in the parsimony trees. In both the maximum like- lihood and parsimony trees, two species of Hypox- idaceae are included in Asparagales consistent with Dahlgren et al. (1985) DISCUSSION The order of divergence of the seven major lineages in our phylogenetic analyses is in general agreement with widely accepted views on the evo- lution of the monocotyledons. Recognition of the early divergences of Alismatanae (Cronquist, 1981; 1985) and Dioscoreales (Dahlgren 1985), and the affinities between the Alis- 1985; Grayum, 1991) correlate with the position of spe- Dahlgren et al., et al., matanae and the Aranae (Dahlgren et al., cies from these superorders near the base of our Postulated gences of Arecanae (Doyle, 1973), Bromelianae, molecular phylogenies. later. diver- Commelinanae, and Zingiberanae (Cronquist, 198 are also consistent with our analyses. In general, deep branches in the tree (Figs. 1, 4, and 5) are shorter than the terminal branches, indicating fewer nucleotide changes along the for- mer. This relationship suggests either that the sub- stitution rate during evolutionary radiations of the monocotyledons was unusually slow, that sampling bias occurred in part because of extinction events, or that the original radiations occurred rapidly. The fossil record, and especially that of fossil pollen, & Hickey, 1976) and probably for the monocotyledons as well (Doyle, for angiosperms in general (Doyle 1973) is certainly consistent with the hypothesis of rapid radiation. Acorus calamus occupies a unique, basal po- sition in all the trees generated for this study. Although Acorus was traditionally classified in Ara- ceae because of superficial morphological similar- ities with the Australian aroid Gymnostachys an- ceps, aroid authorities have recently acknowledged long-recognized difficulties with this classification and, based on a substantial body of evidence (re- viewed in Grayum, 1987), proposed removal of Acorus to a monogeneric family. A tree (not shown) constrained to include 4corus and four species of rales as monophyletic that was analyzed over the subset of 79 species is 25 steps longer than the shortest trees. Our analysis thus indicates that re- moval of Асогиз from the Araceae is consistent with a more parsimonious phylogenetic hypothesis and further offers an explanation for the failure to identify synapomorphies of this genus with other monocot species, if, as we suggest, Acorus is dis- tinguished as the most basal extant lineage of mono- cotyledons. Several other aspects of the tree of Figure 1, not already considered in the accompanying re- ports in this issue, are of interest. We have included rbcL sequences from 21 species in 12 of 30 families in the large order Asparagales (esti- mated 5,000 spp. or 10% of monocotyledons) sen- su Dahlgren et al. (1985). found in a single monophyletic clade in both max- These species are all imum likelihood and parsimony trees, together with only six other species: (1) Aletris farinosa, (Me- lanthiaceae); (2 and 3) two species of Iridaceae: Anomatheca laxa and Iris X germanicum; (4 and 5) two species of Orchidales: Veuwiedia veratri- folia (Apostasiaceae) and Oncidium excavatum (Orchidaceae); and (6) Burmannia biflora. Vur- ther, the clade of Asparagales (excluding Hypox- idaceae) is at least moderately supported by boot- E FIGURE 3 These ee are =a lengt indices of О (1) b (2) that containing _____- 5) that consisting of Alismatanae, Arecar taxon is given parenthetically. 932, have consistency indices excluding uninformative characters of 0.267, . Note that the unresolved portions of this topology are within the terminal clades corresponding to: si three species of Bromelianae; (3) € Overview of the strict consensus of 109 equally parsimonious trees for 104 monocots and 18 dicots. 3. and retention Zingiberanae; (4) Aspar- one species of Melanthiales. The terminal species are a is issue and as genera in woe 4 and 5. The number of species analyzed from each Annals of the Missouri Botanical Garden Flagellaria Prio TIT. nium Sparganium Aechmea T ыны Cato Philydr m. AnigoZant Ravenala Oxychioe гет — Cyperus | — l— Carex саата aff. pallida r—L— T. " ride soconuscana Pontederia Strelitzia Calathea [^ Hedychium Riedelea Zingiber lobba Heliconia Costus 5 j Psi me apeinocheilos Caryota Orchidantha Мура 20 Ph oni ү rymophloeus Chamaedorea scale Calamu Serenoa FiGURE 4. со portion of one arbitrarily selected tree from the set of 109 trees (see consensus tree, Fig. 3) of length 3,932, consistency index of 0.267, and retention index o This portion of the tree includes species > lengths of оо aii Bromelianae, Zingiberanae, and A length of 20 steps. strap (84%) and decay (at least two steps longer) analyses. Duvall et al. (in review) have analyzed rbcL sequence data from four more species in three additional families of Asparagales that further sup- port the common ancestry of this large order, and they note that the alliance between Iridaceae and Asparagales has morphological and anatomical sup- rt. The historical treatment of Arecanae, Cyclan- thanae, and Pandananae as at least marginally related taxa has been contradicted by subsequent taxonomic schemes that treat each as an unrelated superorder (Thorne, 1983) or as separate super- orders with a loose alliance between Arecanae and Cyclanthanae (Dahlgren et al., 1985). Phyloge- netic hypotheses based on the rbcL data support Freycinetia (Pandananae) and Sphaeradenia (Cy- clanthanae) as sister species only distantly related to seven species of Arecanae. This arrangement suggests that Pandananae and Cyclanthanae are Arecanae (see Fig. 5 correspond to the number of substitutions optimized along the branches. The io the remainder of the tr scale bar is proportional to a branch more ancient groups than Arecanae. Inclusion of rbcL data for a second species of Pandananae, Pandanus veitchii does not alter this result (Duvall et al., in review). The rbcL trees suggest that Smilax glauca is allied with species of Liliales and isolated from the two other species of Dioscoreales (Dioscorea poly- gonoides and Tacca integrifolia). Among the three species a closer relationship has been proposed between the latter two, and Smilax has been placed outside of Dioscoreales because of a lack of simi- larities in secondary chemistry (Dahlgren et al., 1981). The more recent decision to place Smilax in Dioscoreales was based on “leaf morphology and floral appearence, although Smilax is hypothe- "bridge" between Dioscoreales and ) „> sized to form a species of Lilianae (Dahlgren et al., 1985). Included in our analyses are 11 species of Poales and five species of the related Cyperales. Phylo- genetic analyses of rbcL sequences of Poaceae Volume 80, Number 3 1993 Duvall et al. 617 Phylogenetic Hypotheses for the Monocotyledons Mr. Xanthorrhoea Chlorophytum Yucca Beschorneria Hesperaloe ave Mu 39 Da Nolina recurvata Dasylirion Nolina lindheimeriana Iris Anomatheca Cyanastrum Curculigo Chamaelirium Vellozia Colchicum Burchardia Freycinetia E ahleh aid Dioscorea E тасса А | Pleea CPU reese Spathiphyllum | ‘Pistia G tach та mnostachys á d Sagittaria graminea S.latifolia ———— Aliema Potamogeton Acorus Amborella 20 Scale Peperomia FIGURE 5. Remainder of the tree partially shown in Figure 4. This portion of the tree includes species of Aranae, Alismatanae, Lilianae, outgroup dicotyledons, and Асо orus calamus (see Fig. 4 for the remainder of the tree). Branch lengths correspond to the number of substitutions optimized along the branches. The scale bar is proportional to a branch length of 20 steps. have been thoroughly discussed (Doebley et al., 1990). We here note that pooid and panicoid grass- es are segregated into sister clades in Figure 2 > and that Oryza sativa occupies the most basal position of Poaceae. In the clade consisting of Poales (Fig. 3). related to Poaceae, and Cyperales are found as a a species of Restionaceae is most closely component of a sister clade which also contains Typhales contra Dahlgren et al. (1985). This to- pology is in agreement with the distribution of three inversions in the plastid genomes of these taxa 1992) that further predicts that Join- This prediction has been confirmed by analysis of rbeL (Doyle et al., villeaceae are immediately basal to Poaceae. data as well (Duvall et al., in review). In our analyses Commelinaceae (three species) cluster with three species of Bromelianae: Pontede- ria sagittaria, Philydrum lanuginosum, and Ап- igozanthus flavidus in a clade that is sister to Zingiberanae and separated from 18 other species Annals of the Missouri Botanical Garden of Commelinanae (Fig. 4). A parsimony tree over the 79 species subset constrained to include 13 spp. of Commelinanae as monophyletic (not shown) is 15 steps longer than the tree without constraints. Parsimony analysis of rbcL thus does not support the recognition of a monophyletic Commelinanae. Pistia has been tentatively aligned with Lemna based on similarities of seedling structure (Grayum, 1991). Analysis of rbcL sequences supports this alignment at a bootstrap value of 69% and a decay index value of at least two. Three species of Melanthiaceae are found in three different primary lineages. These species ex- hibit a great deal of variation and when combined "difficult" treatment (Dahlgren et al., 1985). Our results nul port the suggestion (Dahlgren et al., 1985) the Melanthiaceae should be divided into several families, and we further suggest that those families into a single family are considered a may not be closely related to each other. The phylogenetic analysis presented here offers support for the recognition of seven primary lin- eages of monocotyledons that diverged over a rel- atively short period of geologic time. As noted, this result is in agreement with the taxonomic treatment of Dahlgren et al. (198 tree for the 79 species subset constrained to include nine of ten superorders sensu Dahlgren et al. (1985) as monophyletic groups (not shown) was consid- 5). However, a parsimony erably less parsimonious (52 steps longer). (Note that the tenth superorder, Triuridanae, is composed exclusively of achlorophyllous species unlikely to possess a phylogenetically meaningful copy of the rbcL sequence.) With regard to the paraphyletic arrangements of Bromelianae, Commelinanae, and Lilianae, the phylogenetic hypotheses presented here are at odds with those based on morphological, anatomical, chemical, and other characters. These discrepancies may reflect a new understanding of the affinities among these taxa. However, insu cient sampling of the rbcL data for these groups of species may also be a factor. For example, we have here included rbcL data from only half of the 52 families of Lilianae. Additional sampling of mo- lecular characters may resolve these discrepancies or offer further insight into the phylogenetics of the monocotyledons. The focus of this project was the phylogenetic framework supported by the rbcL data set for the monocotyledons. Another valuable feature of the data set, now generally available in GenBank, will be to further develop our understanding of the mechanisms and underlying probabilities of nucle- otide substitution, particularly as influenced by structural and functional constraints of the mole- cules. These further studies will undoubtedly refine methods of the phylogenetic analysis of molecular ata. LITERATURE CITED CHast, M. W., D. E. Sorris, R. С. OLMSTEAD, D. MORGAN, D эс = . CLARK, M. HEDREN, p. S. Gaur, В. K. JANSEN, К.-Ј. Kim, С. К. МЛМРЕЕ, J. F. SMITH, У GOLENBERG, W. GRAHAM, S. C. H. BARRE S. DAYANANDAN & V. A. ALBERT. 1993. Phylogenetics of seed plants: An analysis of РЦ sequences from the ep gene rbcL. Ann. Missouri Br ‘Car d. 80: 52 CLEGG, M., K. RITLAND & G. ZURAWSKI. “1986. Pro- cesses of chloroplast DNA evolution. Pp. 275-294 in S. Karlin & E. Nevo (editors), Evolutionary Pro- cesses and Theory. Academic Press, New York. CRONQUIST, А. 1981. An Integrated System of Classi- fication of Flowering Plants. Columbia Univ. Press, rk. New Yo DAHLGREN, R., CLIFFORD & P. Yeo. 1985. The Families of the Monocotyledons. ышар Verlag, New Yo n 1981. A angiosperms with com- S. ROSENDAL-JENSEN & B. NIELSEN. pee classification of the a Mq d ү qd nd and Angiosperm Phyloge- ud d New York. M. рок N, E. GOLENBERG, M. Стесс & D. dd analysis of the pe sub- unit of carboxylase (rbcL) nucleotide sequence among the grasses (Gramineae). Evolution 44: 1097-1108. DovLE, J. А. 1973. The monocotyledons: Their evo- lution s a аза biology. Quart. Rev. Biol. 48: 399-4 ny. DOEBLEY, » Ma. eet 1976. Pollen and leaves from the mid-Cretaceous Potomac Group and their Minds. on early angiosperm evolution. Pp. 139-206 in C. Beck (editor), Origin and Early а a Angio- sperms. Columbia Univ. Press, New 1987. Arapid DNÀ isolation m Wes е of fresh leaf tissue. a Bull. a Е | Sc >, D. Garvin & M. J. ANDE RSON. : Chloroplast DNA inversions and the origin of the grass family (Poaceae). Proc. Natl. Acad. Sci. U.S.A. 89: 7722-7726. CHast, D. Sorris € M. Ciecc. A phylogeny a seed plants resulting from PERLES of DNA sequence variation among the rbcL loci of 475 species with particular emphasis on alliances among Hoch (editor), га al and Molecular ‘Approaches to Plant Biosystematics. Monogr. Syst. Bot. Missouri Bot. Gard. (in review). FELSENSTEIN, J. 1981. Evolutionary trees from DNA sequences: > P likelihood approach. J. Mo- lec. Evol. 17: 368-3 Volume 80, Number 3 1993 Duvall et al. 619 Phylogenetic Hypotheses for the Monocotyledons 199]. PHYLIP (Phylogeny Inference Pack- ge) Version 3.4. University of Wash о Sedis: Gus B. s Moa W. D. CLARK & elative rates st nucleotide ии Эн он at the rbcL pne of monocotyledonous plants. J. Molec. Evol. 3. РА, 2. 35: = GOLDBERG, бА 989. Classification, evolution, and phy- y of the families of monocotyledons. бе. . 71. Smithsonian Institution. Press, @ Un Б: = | oa ct ©» + m =| Sy GRAYUM, M. A summary of evidence and a guments supporting the removal of Acorus from са aceae. Taxon 36: 723- 99]. cu i of the Araceae. Bot. Rev. 57: 167-2 JARRELL, D., M. ‘toed 9. Taavon & R. KUPPER. 1992. A genetic map of citrus based on segregation of juri an LPs in an утва cross. Theor. Appl. Genet. 84: 49-5 KISHINO, H. . HASEGAWA. 9. Evaluation of a maximum likelihood estimate of the evolutionary tre topologies from A sequence and us ranch ing order in Hominoidea. J. Molec. Evol. 179. OLSEN, G., Н. Natsupa, R. HacsTROM & R. OVERBEEK. 1992. FastDNAml Version 1.0.3. University of Il- linois, Urbana and Argonne National Laboratory, Argonne, Illinois. PALMER, J. 1986. Isolation and structural analysis of chloroplast DNA. Meth. Enzymology 118: 167-186 4 E. A. ZIMMER. 1993. Phy- logenetic = a the Zingiberales based on rbcL nn. Missouri Bot. Gard. 80: 620-630. Sworronp, D. 1991. PAUP: Phylogenetic Analysis Us- ing Parsimony, Version 3.0s. Computer program - tributed by the Illinois Natural History Surv Champaign, Illinois. 983 sequences. А THORNE, R. 1 Proposed new шна in the angiosperms. Neira J. Bot. 3: 85-117 Wilson, M., & M. Стесс. 1990 Chloroplast DNA хо] es slowly i in the palm family (Arecacea 314. Molec. Biol. Evol. 7: 303 PHYLOGENETIC ANALYSIS OF James F. Smith, W. John Kress, and THE ZINGIBERALES BASED Elizabeth A. Zimmer”? ON rbcL SEQUENCES! ABSTRACT Morphological data have been used previously to construct phylogenies of the eight families of the Zingiberales, one of the most widely accepted monophyletic groups of flowering plants. To provide additional support for phylogenetic pes among monocots, we present a parsimony analysis of DNA r 21 species of Zingiberales and proposed relatives. Five analyses one containing the Costaceae and Marantaceae, and the other, the remaining six families. All recognized families are monophyletic with the Бае а of the Musaceae, which is arsimonious trees, phylogeneti steps longer than the mo ic resolution is rapidl araphyletic with the парк With trees one and two t, suggesting that the phylogenetic utility of rbcL sequence des for the Zingiberales i is limited to interordinal and intrafamilial relationships. The Zingiberales, a morphologically distinctive order of monocots, are one of the most widely ac- cepted monophyletic groups of plants (Bentham & Hooker, 1883; Petersen, 1889; Schumann, 1900, 1902, 1904; Hutchinson, 1934, 1959, 1973; Na- kai, 1941; Tomlinson, 1962, 1969; Stebbins, 1974; Cronquist, 1978, 1981; Dahlgren & Rasmussen, 1983; Dahlgren et al., 1985; Kress, 1990). Dahl- gren et al. (1985) listed six apomorphies for the Zingiberales: root hair cells shorter than other epi- dermal cells, sieve tube plastids containing starch, presence of silica bodies, epigynous flowers, lack of distinctive apertures on the pollen grains, and the occurrence of arillate seeds. In addition, the herbaceous arborescent stem, distichous phyllo- taxy, large petiolate leaves with blades possessing transverse venation, conspicuous colorful bracteate inflorescences, and the substitution of one to five staminodia for the fertile stamens are characters easily used to identify members of the Zingiberales (Kress, 1990). As currently classified, the order consists of eight families (Kress, 1990): Musaceae, Lowiaceae, Hel- Iconiaceae, ceae, Marantaceae, and Cannaceae. Strelitziaceae, Zingiberaceae, Costa- Subordinal classification, including delimitation and rank of these families, has been subject to many changes 1990). C morphological characters have greatly improved (reviewed in Kress, ladistic analyses of the understanding of phylogenetic relationships of the families (Dahlgren & Rasmussen, 1983; Kress, 1990). Dahlgren & Rasmussen (1983) per- formed the first cladistic analysis of the Zingiberales using the eight families listed above and polarized characters using their Commeliniflorae. This anal- ysis resulted in a single tree (Fig. 1) composed of three main clades that included the ginger group (Zingiberaceae/Costaceae and Marantaceae/ Can- naceae), the banana group (Musaceae/ Heliconi- aceae), and the bird-of-paradise group (Strelitzia- ceae/ Lowiaceae). The relationships among the three groups remained equivocal. Kress (1990) re-analyzed the data of Dahlgren & Rasmussen (1983) and performed a separate analysis that included other characters which were rooted with the Bromeliales. This second analysis resulted in a different cladogram (Fig. 2) from that of Dahlgren & Rasmussen (1983). Although the ginger group relationship was retained in both, the families of the banana and bird-of-paradise groups " We thank Youngbae Suh for expert advice on cloning and sequencing. We no the Scholarly Studies Program of the Smithsonian Institution for providing funding and a grant to WJK, th Garden for care of living collections, and G. Zurawski (DN staff of the US Botanical AX) for providing primer Pus nces. * Laboratory of Molecular Systematics, Museum Support Center, Smithsonian Institution, Washington, D.C. 20560, S.A : Department of Botany NHB-166, Smithsonian Institution, Washington, D.C. 20560, U.S.A. ‘Current address: Department of Biology, Boise State University, Boise, Idaho 83725, U.S.A. ANN. Missouni Bor. Garb. 80: 620-630. 1993. Volume 80, Number 3 1993 Smith et al. Phylogenetic Analysis of Zingiberales > з m Ф Ф o Ф o © Ф m 9 ~ o v = ш o o = © o o c — = E го c ч e а го го m o e x Ke о 4 3 19 в 7 5 эз 13 30 de 21 18 39 34 36 4 6 26 4 ЕЈ ' эв : 18 o Ф © © Фа а о о с с – ш – o N n c © + а о Фф | о о о — m = с а - — л р x Фф 2 + o = = [Val 4 FIGURE 1. were shown to be paraphyletic (Kress, 1990). These analyses have been highly informative in terms of familial relationships, yet the utility of many of the morphological characters at this level is question- able. The eight families of the Zingiberales are highly distinctive with many unique and highly modified morphological structures. Interpretation of homology between families has posed difficult problems without detailed developmental analysis for many of the structures (Kirchoff, 1991) Тће numerous autapomorphies of the Zingiber- ales have also made difficult the determination of the evolutionary relationship of the order to other monocots. The Zingiberales have generally been allied with the Bromeliales and/or поени (Hutchinson, 1973; Dahlgren et al., 1992). The presence of starchy ea epi- cuticular wax of the Strelitzia type, UV-fluores- cent organic acids in the cell walls, and two or four subsidiary cells in the stomatal complex are all orne, specialized characters that unite the Zingiberiflorae with the Bromeliiflorae and Commeliniflorae (Dahl- gren et al., 1985). Most modern classifications have placed the Zingiberales near the Bromeliales (con- taining the single family Bromeliaceae) (Hutchin- son, 1973; Stebbins, 1974; Takhtajan, 1980; Cronquist, 1978, 1981) based on the similarity of inflorescence and flower structures (primarily the large, conspicuous bracts and petaloid perianth parts). Although these morphological homologies are potentially equivocal, the presence of several chemical constituents (myricetin and/or quercetin glycosides) also has suggested a common ancestor for the Zingiberales and Bromeliales (Williams & Cladogram of the Zingiberales from Dahlgren & Rasmussen (1983, their figure 9). Harborne, 1977). Thorne (1992), however, rec- ognized three separate orders in his superorder Commelinanae: Bromeliales, Commelinales, and Zingiberales. Alternatively, Walker (1987) derived his Zin- giberidae (excluding the Bromeliales) and his Pon- tederiidae (Haemodoraceae, Pontederiaceae, and hilydraceae) directly from a Ша lineage. Sepal nectaries, vessels primarily in the roots, and several chemical characters (e.g., chelidonic acid) found in the Zingiberales support their placement with a lilialean lineage (Takhtajan, 1980). e use of molecular characters in cladistic anal- узев has been highly successful in plants (Palmer et al., 1988; Crawford, 1990), particularly in tax- onomically difficult groups in which it is hard to ‚ Chase «€ sequence interpret morphological homology (e.g Palmer, 1989; Smith, 1991). Rec ini data derived from the chloroplast gene rbcL, which encodes the large subunit of ribulose bisphosphate carboxylase, has provided resolution at higher tax- onomic levels in plants (e.g., Doebley et al., 1990; Donoghue et al., 1992; Olmstead et al., 1992; other papers this issue). The relatively slow rate of mutation of the rbcL gene has made this gene useful for taxonomic investigations at and above the generic level. However, the limits and. taxo- nomic range for which rbcL can adequately resolve phylogenetic relationships have not been examined widely To compare phylogenies based on morphological analyses in the Zingiberales, we initiated a cladistic analysis of rbcL sequence data. Our goals in this research. were: (1) to examine the monophyly of 622 Annals of the Missouri Botanical Garden SS © RE 2. Most parsimonious tree of the Zingiberales Ficu from Kress (1990, his figure 7). the eight recognized families; (2) to construct the phylogenetic relationships of the families within the order based on rbcL sequence data; and (3) to determine the sister group relationship of the Zin- giberales. MATERIALS AND METHODS Species were selected in an attempt to represent the most divergent members of each family. For example, in the large family Zingiberaceae, the species selected represent each of the four tribes. Different genera were represented wherever pos- sible, depending on availability and number of gen- era in each family. A minimum of two taxa per family was chosen to reduce potential long branch 1978), which result when a single taxon with no close affinities is included in effects (Felsenstein, an analysis. Species, collection localities, and voucher information are listed in Table 1 Total genomic DNA was extracted from fresh or frozen leaf tissue by a modified CTAB method (Smith et al., 1991 [1992]). An approximately 1,401 bp segment of double-stranded DNA con- taining the sequence for the rbcL gene was am- plified via the Polymerase Chain Reaction (Cetus Corporation) (PCR). Two synthetic oligonucleotides were used as amplification primers. The 5” primer is the 7-1 rbcL primer based on the first ЗО bp of the rbcL sequence of maize, and the 3' primer is the corresponding Z-1375R primer, which is a 26 bp primer derived from position 1375-1401 of the maize sequence (Zurawski, DNAX). Initial at- tempts were made to amplify DNA from zingiber- alean taxa using the primers of Olmstead et al. (1992); however, an apparent substitution unique to the Zingiberales in this region of the gene pre- vented amplification with these primers. This sub- stitution is currently under investigation (Smith et al., unpublishe Sequences were obtained by cloning the PCR product into BlueScript SK * (Stratagene, Inc.) us- ing either the Hinc Il or EcoR V site. The ligation was facilitated by first incubating the PCR products with DNA polymerase to assure blunt ends. The products of the ligation reactions were used to transform competent cells of Eschericia coli (XL-1 Blues; Stratagene, Inc.). Actively growing liquid cultures of transformed bacteria were inoculated with the helper phage VCS-M13 (Stratagene, Inc.) and single-stranded DNA was harvested that con- tained the inserted rbcL gene. This single-stranded DNA was then sequenced using Sequenase version О (US Biochemicals), and fragments were sep- arated on 4% polyacrylamide gels. Internal se- quencing primers were derived from sequences dis- tributed by G. Zurawski (DNAX). dditional sequences for Lilian. Vellozia, Puya, Tillandsia, Stegolepis, Pontede- Maranta Magnolia, ria, Lachnanthes, Tradescantia, and were graciously provided by colleagues (see Chase et al., 1993). Sequences were read directly from the auto diographs and entered into a NEXUS file. This fle was read into PAUP version 3.0s (Swofford, 1991) for cladistic analysis. Characters were directly scored for each nucleotide and not modified in any way. Missing data or ambiguous regions were scored as missing. Initial analyses used HEURISTIC SEARCH and STEPWISE ADDITION of 500 RANDOM ADDITION SEQUENCE replicates, TBR branch swapping, saving ALL MINIMAL Volume 80, Number 3 1993 Smith et a . 623 Phylogenetic Analysis of Zingiberales TREES, and COLLAPSING BRANCHES. initial analysis due to the sensitivity of PAUP (and ZERO LENGTH These options were chosen as an all other programs) to taxon order in the data matrix (Maddison, 1991). T data set was then altered to produce the same most parsimonious tree with the GENERAL HEURIS- TIC OPTION and default weighting analyses. Neighborhood trees of 1 and . The order of taxa in the options for further 2 steps longer were also examined using the same default options with the exception that trees of a specified length and shorter were saved. Several separate analyses were performed using the above options to determine if choice of outgroup had an effect on the arrangement of taxa within the Zingiberales. These analyses used: (1) Mag- nolia as the outgroup, with all other monocots included as ingroup; (2) Lilium as the outgroup with all other monocots included as ingroup, and Magnolia excluded; (3) Tradescantia Pontede- ria- Lachnanthes as outgroup, with only the zin- giberalean taxa included as ingroup. Character state changes were plotted onto trees using the A RAN option. Because the DEL- TRAN option can sometimes drastically alter the distribution when equally parsimonious options are available, character state changes also were plotted with this option, and the two distributions com- pared. RESULTS SEQUENCE DATA Sequences of 1,345 bp or more were obtained for 33 species including 10 sequences graciously provided by other workers (see Chase et al., 1993) (Table 1). For the 32 species used in the majority of analyses (excluding Magnolia, which was used to determine outgroup relationships), 464 positions (~ 34%) were variable. Of these 464 characters, 183 (~3 tapomorphic. Therefore, 9% of the variable characters) were au- 281 of the original 1,345 bp (~21%) were shared by two or more taxa. 105 20%) were second po- the 464 variable characters, (~23%) were first codon positions, 92 (~ sitions, and 267 (~ Ta- ble 2) ~ 97%) were third positions ANALYSES The equally weighted analysis from the RAN- DOM ADDITION SEQUENCE search resulted in two most parsimonious trees of 1,163 steps (CI of 0.41; RI of 0.49). The two trees differed only in the relationships of Stegolepis and the Bromeli- aceae. The strict consensus of these most parsi- The to- pology was unaltered whether Magnolia, Lilium, monious trees is presented in Figure 3. or Tradescantia/ Lachnanthes/ Pontederia were used to root the tree. Character state change distributions differed only slightly between the ACCTRAN and DELTRAN option. At only four nodes did the distributions differ in more than two character state changes and in none of these did a strongly supported node become weakly supported or vice versa. Based on the character state distribution of the most parsi- 709 (~61%) of the 1,163 character state changes were transitions, and 454 (39% monious trees, were transversions (Table 3) The strict consensus of the 62 trees of 1.164 steps or fewer loses most of the resolution found in the most parsimonious trees (Figs. 3, 4). A search for trees two steps longer than the most parsimo- nious trees produced 1,541 trees of 1,165 steps or fewer. А strict consensus of these 1,541 trees lost nearly all resolution with the exception of monophyly of the five major clades in the Zingi- berales, Strelitziaceae, Zingiberaceae, Cannaceae, as well Lowiaceae, and Calathea/ Marantochloa, as outgroup clades (Fig. 4) DISCUSSION A cladistic analysis of the rbcL sequence data produced a single most parsimonious tree for the 3) that differs from any previous phylogenetic analysis of mor- 2) (Kress, 1990). The Lowiaceae and Strelitziaceae lineage is the only families of the Zingiberales (Fig. phological data (Figs. 1, between-family clade shared by both the molecular analysis and one of the cladistic analyses of mor- phological characters (Dahlgren & Rasmussen, 1985) (Fig. 1). The conflicts between the trees based on mor- phological characters and molecular data are many. One conspicuous example pertains to floral mor- phology. In all traditional classifications of the order (see Kress, 1990, for summary), the reduction in the number of pollen-bearing stamens from 6 or 5 Heli- . 1 rr. . coniaceae) to | or 5 (Zingiberaceae, Costaceae, (Musaceae, Strelitziaceae, Lowiaceae, and Cannaceae, and Marantaceae), and the associated modification of these stamens into petaloid stami- nodes, are characters that have served as the basis for dividing the families into two basic groups, the banana group and the ginger group, respectively. The homology of these specialized floral features 624 Annals of the Missouri Botanical Garden Sources of rbcL sequences (all material is deposited at US, SEL, or DUKE). Voucher and Genbank LE 1. information for sequences obtained from other laboratories is referenced in the appendix to this volume. Species Voucher Source Genbank # Bromeliaceae Tillandsia elizabethae D. Clark Puya dyckioides D. Clark Rapateaceae Stegolepis allenii D. Clark Commelinaceae Tradescantia soconuscana Matuda Faden 76-98 Mexico L05463 Tradescantia s M. Duvall Pontederiaceae Pontederia D. Clark Haemodoraceae Lachnanthes M. Chase Velloziaceae Vellozia D. Clark Liliaceae Lilium superbum M. Chase Typhaceae Typha latifolia L. Kress 90-3170 Maryland, USA L05464 Typha latifolia L. M. Duvall Cannaceae Canna indica L Kress 80-1124 Indonesia L05445 Canna tuerkheimii Kranzlin Kress 76-653 Panama L05446 Costaceae Costus barbatus Suess. SEL 86-0550 Marie Selby L05447 Botanical Gardens, Sarasota, Florida Tapeinochilos ananassae K. Schum. Kress 79-1114 Duke University, L05462 Durham, North Carolina Monocostus uniflorus (Poepp. ex Kress 79-1112 Peru Petersen) Maas L05454 Heliconiaceae Heliconia latispatha Benth. SEL 80-1610 Marie Selby L05451 Botanical Gardens, Sarasota, Florida Heliconia paka A. C. Smith Kress 79-1072 iji L05452 Lowiaceae Orchidantha fimbriata Holttum Kress & Beach Malaysia L05456 87.2159 Orchidantha siamensis K. Larsen Kress 92-3468 Malaysia L05457 Marantaceae Maranta leuconeura M. Duvall Calathea loeseneri Macbride SEL 85-31 US Botanic Garden, L05444 Washington, D.C. Marantochloa purpurea (Ridley) Kress 78-894 Wilson Botanical L05453 Milne-Redhead Garden, Costa Rica Musaceae Ensete ventricosum (Welw.) s.n. Wilson Botanical L05448 Cheesman Garden, Costa Rica Volume 80, Number 3 Smith et al. 625 Phylogenetic Analysis of Zingiberales TaBLE l. Continued. Species Voucher Source Genbank # Musa acuminata Colla s.n. US Botanic Garden, L05455 Washington, D.C. Strelitziaceae Phenakospermum guianensis Kress 86-2099 French Guiana L05458 (L. Richt.) Miq. Ravenala madagascariensis Kress 92-3504 US Botanic Garden, L05459 J. F. Gmel. Washington, D.C. Strelitzia nicolai Regel & Koch Kress 91-3169 US Botanic Garden, L05461 Washington, D.C. Zingiberaceae Globba curtisii Holttum Kress & Beach Malaysia L05449 87-2161 Hedychium flavum Roxb. USBG 90-653 US Botanic Garden, L05450 Washington, D.C. Riedelea aff. wrayii SEL 83-203 yon Arboret L05460 lla Hawaii Zingiber gramineum Noronha Kress 91-3266 Lyon Arboretum L05465 Honolulu, Hawaii is supported by ontogenetic studies in the Zingi- berales as well (Kirchoff, 1983, 1988, 1991). Al- though the four families of the banana group may be variously related, the ginger group, defined by these highly derived staminal features, has always been considered monophyletic (Dahlgren & Ras- mussen, 1985; Kress, 1990; Kirchoff, 1991). Тће topology based on the molecular data places the four families of the ginger group into three separate lineages (Marantaceae and Costaceae; Cannaceae and Musaceae; and Zingiberaceae and Heliconiaceae; see Fig. 3), which would require at the minimum three independent reductions (if 5— 6 stamens is plesiomorphic) or one increase and two reversals to reduction (if 1-3 stamens is ple- siomorphic) in stamen number. Although we should certainly reconsider the possible nonhomology of these traditionally recognized morphological fea- ures, the molecular characters must be examined critically (through increased sampling of taxa, or TABLE 2. ACCTRAN character state distributions for t e most parsimonious trees (Fig examination of other molecular characters) before concluding that the morphological features are ho- moplastic. The rapid "decay? most parsimonious tree found from the rbcL data suggests that the interfamilial phylogenetic signal of this plastid gene is low for the Zingiberales (Fig. 4). There are 62 different topologies that are one step or fewer longer than the most parsimonious; 1,541 topologies can be found searching for trees two or fewer steps longe In contrast to Seu relationships, the rbcL data strongly support both the monophyly of the order within the monocots as well as the monophyly of most of the families. Although the position of the Lachnanthes/ Pontederia/ Tradescantia clade as the sister group to the Zingiberales collapses in the consensus of trees one step longer (Fig. 3), the coherence of the order itself remains robust. In a larger analysis of the рћујовепу of the monocots ` of the various clades in the Variable characters according to codon position. Homoplastic character states are based on the . 3). The first values are the numbers ters, the second are the percentages of total states. Synapomorphic character states include only non- c ео character states. Position Variable Homoplastic Synapomorphic Autapomorphic First 105 ~ 22% 30 ~ 15% 29 ~ 28% 52 ~ 28% Second 92 ~ 20% 30 ~ 15% 19 ~ 237 43 ~ 23% Third 267 ~ 58% 139 ~ 70% 40 ~ 49% 88 ~ 49% Total 464 100% 199 100% 82 100% 183 100% 626 Annals of the Missouri Botanical Garden 46(31) Lilium Costus Tapeinochilos CO Monocostus Calathea Marantochloa 20(15) MA Maranta Hedychium Zingiber Globba Riedelea Heliconia paka || ZI 16(9) Jl HE Heliconia latispatha Ravenala = Strelitzia ST Phenakospermum 7(5) m ы 12(7) 2901) Orchidantha ied * 29(16) Orchidantha siamensis Ensete Canna indica CA (9) Canna tuerkheimii Musa Lachnanthes Pontederia 15(8 23(12) Tradescantia soconuscana Tradescantia 64) Tillandsia 10(7) 21(16) Puya 21(16 13(7) 15(12) po Typha 5(4) Typha | 2202) ____ Stegolepis 44(31) Vellozia FIGURE 3. Strict consensus of the two most parsimonious trees from the equally weighted analysis. The two trees differ only in the relationships of Stegolepis / 'Турһа / Bromeliaceae. Numbers along branches indicate substitutions supporting that clade, numbers in parentheses are the portion of substitutions that are homoplastic. Character state change distributions are based on tree 1 of the two most аке гч trees (Typha as sister group to Bromeliaceae), and the ACCTRAN option. Asterisks denote clades that are lost in the consensus of the 62 trees one step longer than the most parsimonious trees (see Fig. 4). Families of "n Zingiberales are denoted as follows: CO — Costaceae, = Marantaceae, 21 = Zingiberaceae, HE = Heliconiaceae, ST = Strelitziaceae, LO = Lowiaceae, MU = Musaceae, СА = Cannaceae. Volume 80, Number 3 Smith et al. 627 993 Phylogenetic Analysis of Zingiberales Hedychium Zingiber Globba Riedelea Heliconia paka Heliconia latispatha pS] Ravenala Strelitzia Phenakospermum Costus Orchidantha fimbriat Tapeinochiles Orchidantha siamensis Monocostus Ensete Calathea Marantochloa BE T us Canna indica Maranta Canna tuerkheimii Hedychium €— Musa Zingibe B Globba Riedelea Heliconia pa Heliconia latispatha venala Strelitzia Phe ——À Orchidantha fimbriata бой Orchidantha siamensis Tapeinochilós as Monocostus Canna indica Calathea Canna tuerkheimii A Marantochloa A Миза Магап!а Hedychium — d Zingiber ea Heliconia paka Heliconia latispatha Ravenala L— Strelitzia Phenakospermum PE EE АИ NES See LH Orchidantha siamensis Ensete ог Canna indica LLL Canna tuerkheimii Musa e FIGURE 4. Strict consensus trees of the topologies based on A: two trees of 1,163 steps each, B: 62 trees of 1,164 steps each, and C: 1,541 trees of 1,165 steps each. Only taxa of the Zingiberales are illustrated in these figures. based on rbcL sequence data (Duvall et al., 1993), The Musaceae (paraphyletic with the Canna- the sister group relationships of the Zingiberales сеае) are the only family in the Zingiberales that and the Haemodoraceae/Pontederiaceae/Com- is not monophyletic in the most parsimonious tree. melinaceae is supported. The monophyly of the Heliconiaceae is also lost in 628 Annals of the Missouri Botanical Garden TABLE З. Туре of character state changes based on sequence of the resulting protein. However, the tree 1 of the most parsimonious trees (Typha and Вго- meliaceae as sister groups) in the equally weighted analysis (Fig. 3). Transversions are indicated in bold. A C G T А =ч 103 276 110 C — == 137 433 G == = == 104 the consensus of trees one step longer. These dis- crepancies may be the result of limited sampling. Sequences for additional Heliconia and Canna spe- cies, as well as other representatives of the Mu- saceae, may stabilize these portions of the tree. The other six families are well supported as mono- phyletic groups by the molecular data even in the less parsimonious trees. A clear weakness in this analysis is the unbal- anced sampling for some of the families, e.g., Hel. iconiaceae and Musaceae. The addition of two to three more taxa for these families may result in a more robust analysis and provide a better overall estimate of phylogeny for the Zingiberales. In par- ticular, the monophyly of these families would prob- ably be more strongly supported by the addition of taxa to the analysis. Regardless of the problem of uneven taxon sam- pling, the limitations of the data set are most ap- parent when the distribution of character state changes are mapped onto the tree (Fig. 3). Of the 281 phylogenetically informative characters used in the analysis, there are only 90 (32%) synapo- morphic character m changes that are not ho- moplastic (based on 3). Within the Zingiber- ales, there is only one b ни state change that is synapomorphic between families and not ho- moplastic (Strelitziaceae and Lowiaceae lineage). In contrast, seven nonhomoplastic character state changes support the monophyly of the order, and 29 nonhomoplastic character state changes sup- port monophyly of, or are synapomorphic within, Likewise, the monophyly of the outgroup clades (Haemodora- the families of the Zingiberales. сеае/ Pontederiaceae/Commelinaceae and Bro- тећасеае/ Турћасеае/ Кара!еасеае) is supported by nonhomoplastic character states (Fig. The weakness of the signal is also apparent when codon position is examined. Third position codons account for 58% of the variation in the data set and 70% of the homoplasy (Table 2). This is not a surprising result as the redundancy of the genetic code permits a higher rate of substitution at third position codons without altering the amino acid high percentage of homoplastic third position co- dons (139 of the 464 variable positions) and the presence of 3—4 different nucleotides at 64 of these sites indicate that the substitution rate within the Zingiberales may be close to saturation, thereby reducing the ____ phylogenetic signal of the rbcL sequence dat Our results suggest и although phylogenetic signal is present in the data, there is a “window in which rbcL sequence data does not strongly resolve phylogenetic relationships. For the Zingi- berales this window is at the between-family level. At the ordinal and family levels, the molecular data are much more robust in defining and resolving phylogenetic relationships. An explanation for this window may be related to the age of the Zingiberales and the time since divergence of the families within the order. Olm- stead et al. (1992) explained the lack of phylo- genetic resolution of the higher dicot lineages as a result of rapid divergence of these clades during the late Cretaceous. In contrast, phylogenetic res- olution of lineages within the Asteridae sensu lato is resolved adequately with rbcL sequence data (Olmstead et al., 1992) Extant Zingiberales possess numerous derived morphological features, and five of the eight fam- ilies are known from the fossil record (Kress, 1990). Although most of the fossil material has been col- lected in Eocene deposits, the oldest specimens are leaves of the Zingiberaceae from the late Creta- ceous (Hickey & Peterson, 1978). The common ancestor of the lineages leading to the banana group is therefore hypothesized to have diverged from the remainder of the Zingiberales by the late Cre- taceous (Friedrich, 1987), suggesting that the ma- jor lineages within the order had rapidly differen- tiated by the early Tertiary. These times of lineage splitting and divergence in the Zingiberales are similar to those described by Olmstead et al. (1992) for the higher dicots. The lack of phylogenetic resolution using rbcL data in these unrelated flow- ering plant taxa may be due to their common age of origin and diversification. To further clarify phylogenetic relationships within the Zingiberales, data from other sources will be necessary. The morphological data will nec- essarily be re-examined to accommodate new in- terpretations of homology and to include taxa such that the molecular and morphological data sets will be directly comparable. Sequence data from the nuclear encoded 185 265 ribosomal genes are currently being collected for the Zingiberales (Kress et al., unpublished results), which have been suc- Volume 80, Number 3 1993 Smith et al. 629 Phylogenetic Analysis of Zingiberales cessful in resolving some ancient phylogenetic events in other seed plants (Hamby & Zimmer, 1992). Additional data may also be obtained by restriction site comparison of the highly conserved inverted repeat regions of the chloroplast genome (Downie & Palmer, 1992; Smith et al., unpublished results). LITERATURE CITED BENTHAM, С. & J. D. Hooker. um, Vol. 3. L. Reeve & Со., 1883. Genera ы Williams & Norg Lon CHASE, Mo J. D. PALMER. 1989. Chloroplast NA systematics of lilioid monocots: Resources, fea- sibility, and an example from the Orchidaceae. Amer. J. Bot. 76: 1720-1730. ———, D. E. Sorris, R. С. OLMSTEAD, D. Morcan, D. . BARRETT, S. 1993. Phyloge- netics ee seed plants: Án analysis of nucleotide se- quences from the plastid gene rbcL. Ann. Missouri Bot. Gard. 80: 528-580 1990. Plant Molecular Systematics. w York. CRONQUIST, А. 1978. e Zingiberidae, a new subclass of Liliopsida ain Brittonia 30: 505. 1981. An Integrated Sy stem of Classification of Кыен Plants. Columbia DAHLGREN, & F. N. RASMUSSEN. Monocot- yledon evolution: Characters and des esti- mation. /n: M Hecht, B. Wallace & С. T. Prance (editors), Evolutionary Biology. 16: 255-3 Ex CLIFFORD & P. Е. Yro. 1985. The Springer-Verlag, c © Families of the Monocotyledons. erlin es J., M. Dursin, E. M. GOLENBERG, M. T. CLEGG & D Ma. 1990. Evolutionary analysis of the gs sta of carboxylase (rbc L) nucleotide se- quence among the grasses (Gramineae). Evolution 8. DONOGHUE, M. J., R. С. OLMSTEAD, J. F. SMITH & J. D. PALMER. 1992. Phylogenetic relationships of Di sacales based on jen sequences. Ann. Missouri Bot. Gard. 79: 333-34 DowNiE, S. & J. D. = 1992. Chloroplast DNA inverted repeat restriction site mapping and phylog- eny of the Asteridae. Ann. Missouri Bot. Gard. 79: 266-283. DuvaLL, M. R., M. T. Сесе, M. W. АЗЕ, W. D LARK, W. J. KRESS, H G Hirrs, L. E. EGUIARTE, J. F. SMITH, B. S A. IMER & 6. FELSENSTEIN, J. . Cases in which parsimony E "ore will be positively visitados. 410 Syst. Zool. 27: FRIEDRICH, W. 1987. The evolution of the Zingiberales uring the Cretaceous and Tertiary. XIV Interna- tional Botanical Congress, Berlin, Abstract number 5-30-6:284. Hamby, R. К. & E. A. ZIMMER. 1992. Ribosomal _ as a Phylogenetic Tool in Plant Systematics. Pp. 5 9] in P. S. Soltis, D. E. Soltis & J. J. Doyle os Molecular Systematics of Plants. Chapman & Hall, New York. Hickey, L. J. & R. К. PETERSON. 1978. ee a fossil genus of the ginger family from Late taceous to Early Eocene sediments of oo. In- North America. Canad. J. Bot. 56: 1136 Нитс EGON, J. 1934. The Families of Flowering Plants, Vol. 2. Monocotyledons. Macmillan, London. The Families of Powers Plants, Vol. . Monocotyledons, 2nd ed. Clarendon Press, Ox- ог b 1973. The grues of Flowering Plants. 3rd ed. Clarendon Press, ord. KincHorF, B. К. 1983. pc organogenesis in five genera of the Marantaceae be in Canna (Canna- ceae). Amer. J. Bot. 70: 508 Я Floral ontogeny el evolution in the ginger group of the Zingiberales. Pp. 45-56 in у Liens, S. C. Tucker & P. К. Endress (editors), . pects of Floral Development. Crame erli 1991. Homeosis in t пе нек of the Zingi- рес, Amer. J. Bot. 78: 833-837 Kress, W. J. 1990. The phylogeny ánd ow ation of the ана» Ann. Missouri Bot. Gard. 98-72 MADDISON, ^ R. 1991. The discovery and о p qe x most-parsimonious trees. Sys Л = Zool. 4 NAKAI, ns N не Ж ad Plantas Asiae Orientalis (XVI). Jap. Ј. og 17: 189-203. OLMSTEAD, R. G., H. J. MichaEls, К. M. Scorr € J. D. PALMER. 1992 Monophyly of the Asteridae and identification of their major lineages inferred from А sequences of rbcL. Ann. Missouri Bot. Gard 79: 249-265 PALMER, J. D., ‚ H. J. Micuaers, M. CHASE & J. R. то 1988 [1989]. Chloro- plast DNA variation and E ке. Ann. Mis souri Bot. Gard. 75: 1180- PETERSEN, O. C. . a: ТИ еае, Сап- naceae, Marantaceae. Engler & K. Prantl (editors), Die Natürlic ber. Pflanzerfamilien, lst ed. 2(6): 1-43. SCHUMANN, K. 1900. Musaceae. /n: A. Engler (editor), Pflanzenreich. IV. 45. 19 Marantaceae. In: Das Pflanzenreich. 48. || Р TE MERA In: A. Engler (editor), Das Pflanzenreich. IV. 46. SMITH, J. F., K. J. SYTSMA, J. S. SHOEMAKER & R. L. SMITH. 1991 [1992]. А они а comparison of total ша DNA енип protocols. Phytochem- ical ig 29(1 4): 2-9. A. Engler (editor), . The evolution and systematics of Co mnea Mid Pentadenia and Stygnanthe (ges neriaceae ). э University of № consin- Madison: Midian sconsin. STEBBINS, C. L. 1974. ван а i Evolution Above 630 Annals of the Missouri Botanical Garden the L2 Level. oe Press, Harvard Univ., Cambridge, Massachus Sworron, E о 1991. PAUP. Phylogenetic Analysis mony, Version 3.0. Computer program distributed by the Ilinois Naturtal History Survey, Champaign, Illinois. TAKHTAJAN, A. L. 1980. Outline of the classification of flowering pan. ада Bot. Rev. (Lan lor e. updated phylogenetic classi- fication of the flowering plants. Aliso 13: 365-389 TOMLINSON, P. B Phylogeny of the Scitamine- ae— Morphological and anatomical considerations. Кык 16: 192-213 1969. Classification of the Zingiberales eed mineae) with special reference to anatomical e dence. Pp. 295-302 in C. R. Metcalfe (editor), ar omy of the Monocotyledons, Vol. 3. Clarendon Press, Oxford WALKER, ]. W. . Classification and evolution of Зэй ^r авна Amer. J. Bot. 73: 746. [Ab- ct.] Wie С.А J. B. HARBORNE. 1977. The leaf avonoids of the Zingiberales. Biochem. Syst. Ecol. 5: 221-229. PHYLOGENETIC RELATIONSHIPS AMONG MEMBERS OF SAXIFRAGACEAE SENSU LATO BASED ON rbcL SEQUENCE DATA! David К. Morgan? and Douglas E. Soltis? ABSTRACT In an attempt to ms jenes —_ the morphologically diverse members of Saxifragaceae sensu lato, c phylogenetic analyses of rbeL sequer Also included were many putatively aria families, as well Our phylogenetic analyses suggest that taxa of Saxifragaceae sensu lato are allied with a on representative genera of 16 of the 17 subfamilies. as a diverse array of А рони. plants. separate, often distantly related, lineages of several subclasses of flowering plants. Sequence data, in idc uid i other lines of evidence, 30 herbaceous genera that form the core of Saxifra relatives of Saxifragaceae sensu stricto include оне ы Pte roideae and Tetracarpaeoideae, all traditional subfamilies of Saxifra saxifragac members of Saxifragaceae sensu lato, along with 5 the subclass Hamamelidae, are basal to a large assemblage liance, representatives of four le subfamilies of Saxifra Baueroideae, Parnassioideae, and Lepuropetaloideae) are allied with the rosid families Greyiaceae, Cunoni: di Within this primarily rosid a Celastraceae. According to rbcL sequence a clade that is basal to a e group « evi ence, suggest that Saxifragaceae sensu stric to should consist gud of subfamily Saxifragoideac agaceae sensu >, a group of about These data also su that potential close pan чай and Ribesoidese. id possibly Pentho- gaceae sensu lato, as well as Crassulaceae. These icto, Crassulaceae, and se of which are usually placed in Rosidae. aceae sensu lato (Frane ooideae, ‘eae sensu str axa, most veral genera from 1 Of four ar of ae subfamilies of Saxifragaceae sensu lato Ри овалне. Escallonioi deae, Montinioideae, and Vahlioideae) are in our nshi axifr results allied with taxa a included in Asteridae. Ny TOC seque nce dat embryology, ne Mm pe iridoid chemistry Significantly, relatio close екш with those supported by several other lines of evidence, especially ips of S aceae sensu lato suggested Many systematists consider Saxifragaceae to oc- cupy a pivotal position in the evolution of a number Takhtajan (1969, 1980), for example, described such diverse groups of major angiosperm lineages. as Myrtales, Cornales, Celastrales, Gentianales, and Campanulales as all having been derived from an ancestral “Saxifragalean stock." In spite of its po- tential evolutionary importance, however, Saxifra- gaceae are one of the least understood families of flowering plants. The family has been thought to consist of a wide array of morphologically diverse taxa, and the heterogeneity among the proposed constituents of Saxifragaceae is so great that the family has been nearly impossible to define. As a result, few families of angiosperms offer the mag- nitude and diversity of systematic problems posed by Saxifragaceae. Many different classifications of Saxifragaceae have been proposed; a partial listing includes Cron- quist (1981), Dahlgren (1975a, 1980, 1983), Dahlgren et al. (1981), Engler (1890, 1928), Hutchinson (1973), Schulze-Menz (1964), Takh- tajan (1969, 1980, 1987), and Thorne (1968, 1976, 1983, 1992). These schemes of classifica- tion differ from one another not only in how they define Saxifragaceae, but also in how they ally them with. other. families. Frequently, the differences among classifications of Saxifragaceae аге sub- stantial. A comparison of six recent classifications of the family is shown in Table 1. Saxifragaceae were broadly defined b Engler (1890, 1928) and Schulze-Menz The Saxifragaceae of Schulze-Menz (1964) are the most broadly de- ' We thank P. Goldblatt, L. Mm ii R. 5. Hi endel, R. Gornall, D. War ig um кадала of the University of Tokyo. anil the Uni a of MC пете ey for providin ence primers. B. Bohm, R. Thorne, M. Ch urawski for providing sequ on the manuscript and data analysis. This гезеагс , A. E. :h was supporte by "NSF rant van Wyk, G. L. Nesom, А. Kruckeberg, К. D. Thomas, Arboretum, E hild Tropical Garden, and the botanical g plant material and G e, and R. Olmstead м helpful comments t BSR 9 — 14 4, U.S.A “Department of Botany, Washington State University, Pullman, Washington 9916 ANN. Missouni Bor. Garb. 80: 631-660. 1993. (ae -oploixeig) огаовтхолу (әв (әв so[pprse[o) -Эр!ї01әпє{]) evaoevsoneg -арготхал ) aeaoerxoug овиедевје7)–оврг5ој зерне (әвәрто Annals of the Botanical Garden issouri M 632 (aeapiouourajso1214) (9?aproeedaeoe1o |) чициор) зезовшциој (ов : әрәзвәвЧтрэр.цә [. (звартопошојволој d оваовпошојаолоја ТОРОРА ония -9pro1oneg[) овоовлопед (аворгоовивлран) р! ша a р! паа (зворгозе јеов na] 3 Á | sr PP (әвәріои4ѕошәтг) *3BIPIOLUO[[BISH) А ML P b (әғәрошоио[4ҷад) zs aeaprourouo[[Auq lp m овдовиАвошол авзовшојјео5 у ‘geopiomcaig eal, a (eeaprououo[¡4yg) (әкәрюәЗир1рАң) 'aPopioruojeos] ‘ae ` aea2eiduo[n(q aeo»eoSuri1pÁH -TURUO ) 9eao?runuo]q VEEN E d apii -9proa1i[) 9eooe?ruo[[e5s7] (әвәрюшоеовд) | аваоводивлран овитилој . а aLI Luo Losy вод ту) aeaoeue[nsso1^) ` (әвәрто - | (зваргозеадеовдо |) я (зезр (әғәроәдиғірА}) лициорі) зезовшциоџјј А | овоовов леовдој, кин -loJaneg) зваовшоип,) аеаовавивлран eague1pÁ эвәртошоеову sayetuoun тәрюәЗив1рА CDS UH IPIIPIUO||LISH m 9 ( p ed Б" дын ee вајеџл07) (звер (әгәр C (1) әғәртоиќѕошәгҹ (9&aprouÁsourar) av1opruo^) -lojaneg) eeaoeruoun-) -ISSPUIB J) 9eaoersseul1eq d (1) 9?oproqe A әвәэриАвошәгу] (9?oprorunuo]q (аворгојејодолпа о] ) ПР РРА (<) зворошциој (ae (сырен, ‘әвәрошоүүеэѕ овоовјејодолп о пик аворгошој[јео5 duds ргошоје05 4 [e 1 ded (#1) 9eoproruo[pos -әрюцҷе А) звоовуцед Muni олана. *eeaprourouo[Auq (әғәртои4ѕошәҹ) | "I (1) ггарошопо АЦА (аваргојејадолп до] ) (томе *oeoprorxo1g аваоеиАзошол" (әвәрто (11) әвәроәдиеірң doind слао esso ада *eeaprouourg]so12 әв -ISSUE J) aPa2PISsPU.IP авартојејодолпдо аваовјејадолпао р! 180.914 ( | d) d (1) 9eapropei gi (aeapto "peres "овор10д]ј 'зворго -9ptou? A) овозеуцед (әвәр (1) әғәртогѕѕеиле gd -ISSBUIB J) IPIDRISSPUIE Y (әвәртоә1]) aea3ea]] -saqty) araoeliepnssoly (9?apro -1002U?1j) IPIIHPODUBIY (2) оворгогхола (звер (aeapto (aeaprosdue1pAH) -saqn]) aeaoeue[nsso1«) (ae (c) eeaplooouRly -100JUB1J) оваовооџела вод) овооешејп5воло) аваоводивлран (әвәр -9ptom[q? A) әтәэеццед (T) гворголопед (әрәрюә1]) оваовојј (ов (зеаргопАвошол -1002UP1 J) овоовозивјј (9&oprouÁsoura1'q) (1) гвартопошојволоја (әвәрто -әртюїхә1{) әвәәріхәга "оворгоццед (әвәрто аваовиАвошоал" (1) aeaptoa1] -seqn]) оваоешејпаволо) (әвәртоцче A ‘aeaprojejadoinday -1041U34) 9?ooe1oujuoaq (әвәрто (1) әтәртоѕәдтҸ (әвәро *oeaproooue J *aeoprorsseu1eq (әвәртоәватеәв. цә [) -10Ч1ЧӘ,{) әвәэр[п5$Р17) (1) әкәртоәвЧдтеэв.цә [, -I0qjuad) әвәэвлоц1пә,{ ‘aeaploloyluedg ‘apaplooouely ‘ae овадововалеов ој (оворто (1) ггоролоциод (aeapto 'оворгоделихес) -aploloyjuad 'aeapto (әвәро -[pyadoinday] *oeapio (0g) гзворговелихес -Велихес) оваоеВелихес овооввелихес -Велихес) оваовделихес -Веихес̧) оеооеделихес -Велихес) әвәәе8деихес̧ аваовделихес зојеВелИхеС зајећелихес səjesoy вајећелихес зајеВелихес sapesoy эвиеѕоу -oeprsoy IPIOYISOY IPPISOY әвивѕоу аеаовулон аворАшејцопцолу (2861) че езцхеј, (2861) чела цеа (1861) затбиол) (2661) 20104], (6261) uosurgoing (#961) 2тоуј 92] 1426 “SUON BIGISSE[O элу лофо эці ut раовја 318 Коча YoIymM ш exe] эці рип вовоцшолеа ut рајау зле 0]e[ nsuas оваовВелихес jo semmrejqng *oje[ nsuas авооввелихес jo зоцпшејапз л | 941 jo црва ur еЈәиә8 jo 1equinu əy} ојеотрш sasayjuered ul s1aqumw тоје] nsuas аваовделихес 10 #961) ZU3JA-3Z[NUYIS jo yey} 0} џоцвјој ш “IRIDRÍBIJXPS jo suonu?gogisse[2 xis jo uosteduro”) Т TISYL Volume 80, Number 3 1993 Morgan & Soltis Phylogenetic Relationships Among Saxifragaceae 633 fined, consisting of 17 Penthoroideae, Iteoideae, Pterostemonoideae, Baueroideae, Fran- subfamilies: Saxifragoideae, Tetracarpaeoideae, Ribesoideae, cooideae, Brexioideae, Parnassioideae, Lepuropet- aloideae, Hydrangeoideae, Phyllonomoideae, Es- callonioideae, Montinioideae, Vahlioideae, and Eremosynoideae (Table 1). In this paper, we will use this classification as a reference and will refer to it as Saxifragaceae sensu lato (Saxifragaceae s. [.). Saxifragaceae s. l. are a heterogeneous assem- blage, including annual, biennial, and perennial herbs, shrubs, trees, and woody vines, and also having considerable chemical, anatomical, paly- nological, and cytological diversity. ost recent authors have defined Saxifragaceae more narrowly and treated many members of Sax- ifragaceae s. l. either as separate families or as components of other families. Hutchinson (1973; Table 1) separated the woody and herbaceous con- stituents of Saxifragaceae s. /., allying the woody taxa with Cunoniaceae. Other authors do not draw as sharp a distinction as did Hutchinson but do tend to separate woody from herbaceous taxa. Some classifications (e.g., Cronquist, 1981; Table 1), while defining Saxifragaceae narrowly, nevertheless in- cluded most members of Saxifragaceae s. l. in one order (Rosales), suggesting a fairly close relation- ship among them. Others, in contrast (Dahlgren, 1983; Takhtajan, 1987; Thorne, 1992; Table 1), considered some of the herbaceous and woody members of Saxifragaceae s. /. to be only dimid related and allied many of the woody taxa wit groups such as Cornales. The widely divergent views regarding the com- position and affinities of Saxifragaceae clearly in- dicate that the family remains enigmatic. Although some experimental research has provided useful insights (e.g., Bensel & Palser, 1975a, b, c, d; Grund & Jensen, 1981; Hideux & Ferguson, 1976), the actual phylogenetic affinities of the taxa com- posing Saxifragaceae s. !. remain poorly ипдег- stood. Results from a preliminary study of Saxi- data from the 1990) sug- gested strongly that this approach should prove fragaceae s. using sequence chloroplast gene rbcL (Soltis et al., useful in unravelling the taxonomic problems sur- rounding Saxifragaceae s. /. This study revealed that several of the constituent subfamilies are only distantly related to one another. We therefore ini- tiated a broad phylogenetic analysis of rbcL se- quence data from taxa representing not only Sax- ifragaceae s. l, but also many putative relatives. We also included sequences representing a diverse array of dicot families, with two major objectives. The first objective was to ascertain how best to define Saxifragaceae. Secondly, we hoped to de- termine potential close relatives of the members of Saxifragaceae s. !. based on rbcL sequences cur- rently available. MATERIALS AND METHODS We generated rbcL sequences from at least one representative of 16 of the 17 subfamilies of Sax- ifragaceae 5. l. (Table 1). The only subfamily not analyzed was Eremosynoideae. Material of this sub- family could not be obtained (despite repeated at- tempts) because of its restricted distribution in west- ern Australia. Fourteen of the subfamilies consist of only one or two genera (Table 1); hence, the one rbcL sequence obtained from each represents a nearly complete generic sampling. The largest subfamilies are Saxifragoideae and Hydrangeoide- ae, which consist of approximately 30 and 17 genera, respectively. We sequenced at least five representatives from each of these two subfamilies. The many proposed classifications of Saxifra- gaceae have resulted in a long list of possible allies of Saxifragaceae s. l. We therefore generated one or more rbcL sequences from each of several fam- ilies that have been considered closely related to members of Saxifragaceae s. /. in most taxonomic schemes (Crassulaceae, Cephalotaceae, Drosera- ceae, Greyiaceae, Cunoniaceae and its allies, Gun- neraceae, and Rosaceae). Sequences were also gen- erated for many other potentially related families Garryaceae, and Additional rbcL se- quences representing a diverse array of dicots were (Pittosporaceae, Roridulaceae, Cornaceae and its relatives). also used in our phylogenetic analyses. These se- quences were obtained from many other investi- gators (listed in the Appendix at the end of this issue). Plant material for sequencing was obtained from field collections, plants grown from seed, silica gel-dried leaf material, or, in a few cases, herbarium ко (information on sources of plant material also given in the Appendix at the end of this e Tota A was extracted using a CTAB buffer method vL & Doyle, 1987) as modified by Soltis et al. (1991) AMPLIFICATION AND SEQUENCING With the exception of the eight sequences orig- inally reported from Saxifragaceae s. /. (Soltis et al. ‚ rbcL was amplified using thermostable DNA polymerase. We performed amplifications us- ing either Тад I polymerase (Promega) ог Rep- litherm (Epicentre Technologies) following the di- the Synthetic oligonucleotides were used as primers. The forward rections of manufacturers. 634 Annals of the Missouri Botanical Garden primer (Z1) represents bases 1-30 of rbcL from Zea mays L. The reverse primer typically em- ployed (3' rbcL) is external to the З’ end of the rbcL gene. This primer is 34 nucleotides long, corresponds to the sequence of /Vicotiana, and begins 103 base pairs outside the coding sequence termination of the gene in Nicotiana (Olmstead et al., 1992). In several instances we were unable to amplify rbcL using this 21-3' rbcL primer com- bination. In these taxa, rbcL was successfully am- plified using an internal reverse primer (Z1351R) in place of the 3' rbcL primer. The Z1351R primer represents bases 1351-1380 of rbcL from Zea mays. Two amplifications were performed. A double stranded reaction with both forward and reverse primers was first performed. The double stranded product was subsequently used as a template in an amplification using only one primer to produce single-stranded DNA for sequencing. Single-strand- ed DNA was then purified by precipitation with a 20% PEG/2.5 M NaCl solution, washes in 70% and 95% ethanol, and resuspension in 1 x TE. Purified single-stranded products were se- quenced directly. Both forward and reverse strands were sequenced using 30-mer primers correspond- ing to the rbcL sequence of Zea mays. These primers were kindly provided by G. Zurawski; six sequencing primers were used for each strand. d us- ing the Sequenase version 3.1 kit (United States Dideoxy sequencing reactions were performe Biochemicals) and *S dATP. Fragments were sep- arated at 2,000 volts/70 watts in 6% polyacryl- amide gels (0.4 mm thick), using 1 x TBE buffer. Gels were fixed with 10% acetic acid, washed with distilled water, dried, and used to expose Kodak X-Omat AR film for 16-24 hours. А total of 1,377 base pairs of rbcL (bases 31- 1407) was standardly used for analysis. Several of our sequences from Saxifragaceae s. l. were am- plified by cloning (Soltis et al., 1990). Because the cloning site was just inside the 3' end of the gene, our sequence data for these taxa do not extend past base pair 1407. Therefore, our comparisons herein involve rbcL sequence data up to base pair 1407. The first 30 base pairs were also not used because, after amplification, this portion of the gene is identical to the Z1 forward amplification primer. Most taxa were amplified with the external reverse primer (3' rbcL), but Chrysobalanus, Licania, Kalanchoe, Crassula, Sedum, Dudleya, and Eu- cryphia could only be amplified with the internal reverse primer (Z1351 R), and for these seven taxa 1,320 base pairs of sequence data (bases 31—1350) were used. PHYLOGENETIC ANALYSIS Sequence data were analyzed using PAUP ver- sion 3.0s (Swofford, 1991). Because the various rbcL-supported lineages of Saxifragaceae s. !. are allied with so many distantly related taxa, several phylogenetic analyses were necessary to best de- termine potential relatives of all the component subfamilies. These analyses were carried out using the results of a previous broad analysis as an initial framework (Chase et al., According to Мейсон (1991), multiple “‘is- lands," or separate clusters, of most parsimonious trees are likely to be present for a given data set, especially if it includes a large number of taxa. To maximize our chances of finding as many shortest trees as possible, we performed as part of each analysis a series of searches aimed specifically at finding multiple islands of trees. Two primary tree search methods were em- ployed. In the first method, searches were performed without MULPARS using random taxon addition and nearest-neighbor-inter- change (NNI) branch-swapping. This search nor- 200 consecutive mally resulted in one to three relatively short trees. Each of the shortest trees resulting from this search was used as the starting tree for a subsequent search with MULPARS using tree-bisection-recon- nection (TBR) branch-swapping. This set of anal- yses was repeated several times (each time a dif- ferent random number seed for the NNI search was used) with the TBR tree searches each begin- ning with a different starting tree, the expectation being that several analyses with different starting trees would find multiple islands of shortest trees if they exist. In the second tree search method employed, 50 consecutive searches were performed with MUL- PARS using random taxon addition and TBR branch-swapping. To reduce the amount of time involved, the program was allowed to save no more than ten trees of any length longer than that of the shortest trees found by using the first method, above. This second search strategy was recom- mended by Swofford (1991). We compared the results of these two search methods with those obtained using a widely employed method of anal- ysis: a single search using simple taxon addition and TBR branch-swapping. The results of these comparisons will be discussed below Our initial broad analysis included all of our representatives of Saxifragaceae s. /. and putative relatives plus a broad sampling of non-magnoliid dicots. This analysis included 100 taxa and was performed to: (1) obtain a rough approximation of Volume 80, Number 3 Morgan & Soltis 635 Phylogenetic Relationships Among Saxifragaceae likely alliances of the members of Saxifragaceae s. help focus subsequent, more detailed analyses of smaller groups that include members of Saxifragaceae s. l. The made use of the first search method described initial broad analysis above and involved three separate tree searches, each with a different starting tree. The resulting trees were rooted with an outgroup consisting of Tetracentron, Trochodendron, Sabia, and Plat- anus because in the broad analysis of Chase et al. (1993) these genera were identified as sister taxa to the group that we analyzed. With the results of our initial broad analysis as a guide, four subsequent, more detailed analyses were conducted using subsets of the larger data set to identify more accurately potential relatives of taxa of Saxifragaceae s. l. For each of these four subsequent analyses portions of the rbcL-based phylogeny resulting from our initial broad analysis were selected, with particular attention given to groups that occurred in the results of both our initial analysis and that of Chase et al. (1993). Additional sequences were added to the four groups based on the results of the analysis of Chase et al. (1993) so that each group would contain greater taxon density. For each of these four analyses, an outgroup was specified consisting of five taxa that were closely related to and basal to the group being analyzed. Tree searches for these four analyses employed both primary search methods described above. Seven separate TBR searches were con- ducted using the first method. Because some of these analyses resulted in very few shortest trees, we decided to perform further analyses using the second search method, hoping to find additional shortest trees. Three of the four groups were also subjected to bootstrap analyses, which were per- formed using 100 replicates, simple taxon addition, TBR branch-swapping, and MULPARS RESULTS We have chosen to show for our initial broad analysis, as well as our four subsequent smaller analyses, the 50% majority-rule consensus tree because it effectively summarizes the major fea- tures present among all shortest trees. Additionally, we have constructed each majority-rule consensus tree so that it includes compatible groupings of less than 50% occurrence. For our broad analysis and our four smaller analyses we obtained in each case a consensus tree identical to one of the shortest trees; thus we can place nucleotide substitutions on these trees (Figs. 1-5). Our initial broad analysis of 100 sequences in- cluded three separate tree searches, each of which found a different island of trees. The first search found 351 trees, 3,811 steps long, the second found 27 trees, 3,811 steps long, and the third The majority-rule consensus tree was constructed from the 378 trees that were 3,811 steps long (Fig. 1). Two portions of this tree were selected for further found nonminimal trees (3,812 steps long). analysis (see Fig. n For convenience we will refer and the * although both include many taxa normally "rosid clade “asterid clade," included in other subclasses. Analysis of the rosid clade resulted in 16 shortest trees, 3,060 steps long, in two islands of eight trees each (Fig. 2). The second search method found both islands, found by the first search method or by the single but one of the two islands was not analysis using simple taxon addition. One of the objectives of our analysis was to obtain information on the amount of support for alliances involving members of Saxifragaceae s. l. using bootstrap analysis. Because of the size of the rosid clade (86 taxa) it was necessary to conduct further analyses of the group by dividing it into two smaller parts, which we will refer to as rosid groups А and B. Group B consists of an alliance of 61 taxa taken from within the rosid clade, a group supported by 10 nucleotide substitutions (see Fig. 2). А second group (group A) was formed to accomplish further analysis of the remaining taxa of the rosid clade (i.e., those outside of group B). For this analysis, the number of taxa in the rosid clade was reduced by deleting 31 sequences from among those in- cluded in group B. Group A, therefore, included the same taxa as did the rosid clade, except that group B was represented in the analysis by 30 taxa rather than 61 (see Fig. 2). Analysis of rosid group А resulted in 12 shortest trees, 2,000 steps long, in only one island (Fig. 3). These 12 trees were found by both primary search methods and by the single analysis using simple taxon addition. The taxa that we will emphasize as part of group А include representatives of "core" Saxifragaceae (i.e., subfamily Saxifragoideae) as well as genera representing five other subfamilies: Itea (Iteoideae), Pterostemon (Pterostemonoideae), Ribes (Ribesoideae), Penthorum (Penthoroideae), and Tetracarpaea (Tetracarpaeoideae). Analysis of rosid group B resulted in 40 shortest trees, 2,584 steps long, in four islands of 2, 4, 14, and 20 trees (Fig. 4). All four islands of shortest trees were found by both primary search methods, but the single analysis using simple taxon addition found only the island consisting of two trees. Rep- resentatives of several subfamilies of Saxifragaceae 636 Annals of the Missouri Botanical Garden BOYKINIA RIBES TETRACARPAEA Myriophyllum PENTHORUM Sedum Dudleya Kalanchoe н Cephalotus Platytheca ROSID CLADE 38 | - 10 I Sabia = rom Platanus 8 19 [u^ [..— Trochodendron 2 E == Tetracentron _3 FIGUR Majority-rule consensus tree (including compatible groupings) constructed from 378 most үк shi ing from phylogenetic analysis of 100 rbcL sequences of Saxifragaceae sensu lato and many other index of 0. 266 and ийрүү dicots. This tree is identical to one of the 378 shortest trees pur has a consistency ет Volume 80, Number 3 Morgan & Soltis 637 1993 Phylogenetic Relationships Among Saxifragaceae Шш Nicotiana | Convolvulus MONTINIA Heliotropium Acanthus Pinguicula Jasminum VAHLIA Asclepias if Dipsacus lo — ESCALLONIA = = o Ф то o = c 3 ASTERID CLADE ü Viburnum PHYLLONOMA Пех Impatiens Diospyros ^ Sarracenia Fouquieria Nyssa DECUMARIA CARPENTERIA PHILADELPHUS HYDRANGEA Drosera a rete pian ds of 0.46. Arrows point to nodes that did not occur in all shortest trees; all other nodes were present in all : "st ратни вала trees, Numbers above each branch indicate the number of base substitutions. Names of taxa a s ais ceae sensu lato are shown in capital letters and boldface type. The two portions of the tree selected for further analysis (the rosid clade and the asterid clade) are indicated by brackets and labeled. [.Nowi13dOHnd3—— —— 7 | vISSVNHVd ——Гїї | | Missouri Botanical Garden Annals of the HVHOVSOH , €9uÁAN I ‚еши 638 Г | EUJOSOSSOJ2) eoisseJg , шпјоеедој | AVECIOOVUAIXVS a џодџеовде | == G uo1puepouoo1 | — 81 S enjequie] —— © зпџејеја = H eiqes — sI wn Aydiuydeg зјешешен и elejopou 9 m шп Аца ред a ý - L vaVduvovulil Volume 80, Number 3 1993 Morgan & Soltis 639 Phylogenetic Relationships Among Saxifragaceae s. L. are included in group B: Francoa (Francooi- deae), Bauera (Baueroideae), Brexia (Brexioi- deae), Parnassia (Parnassioideae), and Lepuro- petalon (Lepuropetaloideae). Analysis of the asterid clade resulted in 36 short- est trees, 1,496 steps long, in one island (Fig. 5). These 36 trees were found by both primary search methods and by the single analysis using simple taxon addition. The asterid clade includes, dition to many taxa usually included in subclass Asteridae (sensu Cronquist, 1981), representatives f of Saxifragaceae s. l.: Escallonia (Escallonioideae), PA yllonoma (Phyllonomoideae), Vahlia (Vahlioideae), Montinia (Montinioideae), and ise genera of Hydrangeoideae. According to our rbcL sequence data, all of these taxa are dis- in ad- tantly related to core members of Saxifragaceae (Olmstead et al., 1993, have conducted more de- tailed analyses of taxa in Asteridae). DISCUSSION Our phylogenetic analysis of rbcL sequences indicates that the taxa traditionally included in Saxifragaceae s. /. are not a monophyletic assem- blage. Rather, they are allied with at least ten different, often very distantly related, lineages rep- resenting most of the traditionally recognized sub- classes of dicots (Fig. 1). The results of this analysis do not correspond closely with sa proposed clas- sification of Saxifragaceae; in many cases rbcL sequence data suggest relationships that differ sub- stantially from traditional classifications. For ease of presentation, we will discuss the systematic im- plications of this study by examining the relation- ships of each of the subfamilies of Saxifragaceae s. L individually. Our discussion will be divided into five parts corresponding to more inclusive alliances involving the subfamilies of Saxifragaceae s. L: 1. Saxifragoideae and associated subfamilies (Saxifra- goideae, Iteoideae, Pterostemonoideae, Ribesoide- ae); П. (Penthoroideae, Tetracarpaeoideae); HI. Subfam- ilies associated with other families of Rosidae (Bau- Subfamilies associated with Crassulaceae eroideae, Francooideae, Brexioideae, Parnassioi- deae, Lepuropetaloideae); IV. Subfamilies associated with Cornaceae (Hydrangeoideae); V. Subfamilies associated with Asteridae (Phyllonomoideae, Es- callonioideae, Montinioideae, Vahlioideae). I. SAXIFRAGOIDEAE AND ASSOCIATED SUBFAMILIES It is generally agreed that Saxifragoideae. approximately 30 genera placed in Saxifragoideae Table = — 1) comprise a natural group. These taxa share many morphological, anatomical, and chem- ical features, and most classifications maintain these genera as a distinct group at the familial, subfa- milial, or tribal level. Saxifragoideae are the core of the original Saxifragaceae s. l. and some clas- sifications define Saxifragaceae so narrowly that the family consists only of this subfamily or perhaps a few additional taxa. For example, the Saxifra- gaceae of Takhtajan (1987) and Thorne (1992) are basically equivalent to Schulze-Menz's sub- family Saxifragoideae. Saxifragaceae as defined by Hutchinson (1973) and Dahlgren (1983) consist only of Saxifragoideae and a few genera from other subfamilies of Saxifragaceae s. l.. Other classifi- 1981) include still addi- tional genera in the family. cations (e.g., Cronquist, The results of our analyses support the mono- phyly of Saxifragoideae by defining a monophyletic group composed of six genera of the subfamily (Figs. 2, 3). These data also suggest that many subfamilies that have been combined with Saxifra- goideae to form broader versions of Saxifragaceae (e.g., Parnassioideae, Lepuropetaloideae, Pentho- roideae, Francooideae, Vahlioideae; Fig. 1) are dis- tantly related to Saxifragoideae. The distinctiveness of Saxifragoideae and the distant relationships between them and other sub- families are confirmed by other molecular data. Downie et al. (1991) have determined that some angiosperms lack the intron of the chloroplast gene rpl2. Significantly, this intron is absent in all tested species of Saxifragoideae but present in genera from all other subfamilies of Saxifragaceae s. l examined (Iteoideae, Ribesoideae, Francooideae, Penthoroideae, Brexioideae, Escallonioideae, Par- — FIGURE 2. Majority-rule consensus tree (including compatible groupings) constructed from 16 most parsimonious > he trees resulting from phyloganete analysis of rbcL sequences of t 16 sisiency index of 0.293 and a retention и of 0.496. did not occur in all shortest trees; all other nodes were present in all 1 r of base зе ата Names of taxa cted for further analysis, is indicated by a bracket and labeled. he rosid clade. This tree is identical to one o Arrows point to nodes that Numbers above e shown in c p чен ха most parsimonious trees. ic ud sensu lato ar not included in the rosid group А den are > indic 'ated by asteris 640 Annals of the Missouri Botanical Garden TETRACARPAEA Myriophyllum PENTHORUM Sedum Dudleya Crassula RIBES Kalanchoe | CRASSULACEAE = si^ о 6 55 [22 г m c О «lo m 2 > SAXIFRAGOIDEAE LEPTARRHENA BOYKINIA Cercidiphyllum Rhodoleia Hamamelis Daphniphyllum Sabia p 92 шу = Platanus 60 Lambertia GROUP + Г}; _ Trochodendron = FIGURE 3. Major trees homas from phylogenetic analysis of rbcL se shortest tre as a consistency index of 0.374 ur in > all shortest trees; all other nodes w branch indicate the number of base substitutions; num occurrence of each monophyletic gro nassioideae, Hydrangeoideae). Downie et al. (1991) suggested that the rp/2 intron has been lost in six separate angiosperm lineages, one of them being Saxifragoideae. Chloroplast DNA restriction site data (Soltis et al., 1993) likewise support Saxifra- goideae being a distinct group. According to cpDNA restriction site data, Saxifragoideae are a well-sup- ported monophyletic group separate from Ribes and Francoa, two taxa that have very often been associated with it. These three lines of evidence suggest that the family Saxifragaceae would be best defined as comprising only subfamily Saxifrago- ere present in all 12 m bers in italics 1 low eac up in the results of 100 bootstrap replicates. Names of taxa of Saxifragaceae sensu lato are shown in capital letters and boldface type. ~ LB Tetracentron. $ ity-rule consensus tree pigs ня groupings) constructed from 12 most parsimonious ences of rosid group pr a retention index of 0 . This tree is identical to one of the 12 0.473. Arrows n to nodes that did t parsimonious trees. Numbers above each ranch indicate ii percentage ideae (sensu Schulze-Menz, 1964) and that other subfamilies of Saxifragaceae s. /. should be treated either as separate families or as members of other families. This narrow definition of Saxifragaceae is identical to the concept of the family proposed by Takhtajan (1987) and Thorne (1992). In addition to helping resolve a long-standing taxonomic problem (the composition of Saxifra- gaceae), rbcL sequence data are of value in iden- tifying potential close relatives of this narrowly defined group and in determining the possible af- finities of the other subfamilies of Saxifragaceae s. Volume 80, Number 3 Morgan & Soltis 641 1993 Phylogenetic Relationships Among Saxifragaceae 44 (3 — Pisum 22_|100 Medicago 7 = | S Polygala Krameria Cephalotus Ceratopetalum Eucryphia Oxalis 47 Chrysobalanus Euphorbia Byrsonima BREXIA PARNASSIA Gossypium BE Crossosoma bv 11 22 26 10 40 24 100 6 32 FIGURE 3. Continued. L In the following sections we will discuss the sys- tematic implications of rbcL sequence data for the remaining subfamilies of Saxifragaceae s. l. Iteoideae. Our rbcL sequence evidence sug- gests that /tea, the only genus in Iteoideae, is closely related to Saxifragoideae (Figs. 2, 3). Itea has at times been associated with herbaceous rather than with woody Saxifragaceae s. /. (Dahlgren, 1983; Takhtajan, 1987), but has also been placed nearer the woody subfamilies by some (Cronquist, 1981; Hutchinson, 1973; Thorne, 1992). Se- 172 | Geranium quence data, however, suggest that /tea is clearly separated from most other woody subfamilies (Figs. К=з): Previous investigations have provided conflicting results regarding the alliances of /tea. According to Spongberg (1972), chambered pith and stipulate leaves distinguish /tea from Escallonioideae. Em- bryologically, /tea, which has bitegmic and cras- sinucellate ovules, is identical to Saxifragoideae, as well as Ribesoideae, and distinct from Hydrangeoi- deae, Escallonioideae, and Montinioideae, which have unitegmic tenuinucellate ovules (Table 2). 642 Annals of the Missouri Botanical Garden Gossypium Tropaeolum Brassica Batis Limnanthes Acer 19 71 17 99 131 E x 32 Geranium pu Pelargonium I — . Crossosoma Xx 45 Viviani 15 PAI Viviania 38 NE c Greyia e FRANCOA ч 13 pue Tetracarpaea Hi 98 Ri» Myriophyllum 3 94 Penthorum E 22 = 38 Г Sedum = 100 ~~ A3. Kalanchoe 9 FIGURE 4. Majority-rule consensus tree (including compatible groupings) constructed from 40 most parsimonious trees resulting from phylogenetic analysis of rbcL sequences of rosid group B. This tree is identical to one of the 40 shortest trees and has a consistency index of 0.326 and a retention index of 0.476. Arrows point to nodes that did not occur in all shortest trees; all other nodes were present in all 40 most parsimonious trees. Numbers above each branch indicate the number of base сенне numbers in italics below each branch indicate the percentage occurrence of each monophyletic group in the results of 100 bootstrap replicates. Names of taxa of Saxifragaceae sensu lato are shown in capital letters iud boldface type. Also, Bensel & Palser (1975b, d) found morpho- In contrast, other lines of evidence suggest that logical and anatomical characteristics of the flowers — /tea is more closely related to other woody sub- of /tea to be virtually identical to those of Saxi- families. Hamel (1953) concluded, on the basis of fragoideae, and noted that it was essentially im- — karyological data, that /tea was most closely allied possible to differentiate between flowers of /tea and — with Escallonia. The pollen of Itea is bilateral and Saxifragoideae, externally or internally. two-porate (Hideux & Ferguson, 1976; Spongberg, Morgan & Soltis 643 Phylogenetic Relationships Among Saxifragaceae Volume 80, Number 3 1993 12 EN rs Datisca | 87 [18 Octomeles Cephalotus Platytheca BAUERA Ceratopetalum Eucryphia Oxalis Castanopsis Casuarina NO ex | ROSACEAE 68 2 | 13 n 7] 51 16 33 | 8 Pisum 43 16 [100 —5— Medicago 53 | 48 Polygala Licania Chrysobalanus 5 Trigonia Viola 41 SE eng Erythroxylum L2^ Byrsonima "NÉ лаш Euphorbia LI Passiflora 8 г. Humiria 37 BÉ Euonymus 28 fer BREXIA 100 20 PARNASSIA o pr LEPUROPETALON FIGURE 4. Continued. 644 Annals of the Missouri Botanical Garden Nicotiana - Convolvulus - Jasminum ++ Verbena ++ Acanthus + Proboscidea ++ VAHLIA ? 5 16 p 7 Apocynum ++ 25 98 == Gelsemium + 9 12 Garrya ++ 9 Гог L10 Aucuba ++ 61 UM Eucommia ++ Valeriana ++ Symphoricarpos ++ ESCALLONIA ++ Lobelia - Corokia ++ Scaevola ++ Achillea - Lactuca - Menyanthes ++ Hedera - Sanicula - Pittosporum - Griselinia ++ Г lex - 91 ete Helwingia - 68 PHYLLONOMA ? 13 17 [. — Diospyros 2 — а : 98 98 : Sarracenia ++ 10 pue Nyssa ++ 90 ја m Davidia ++ DEUTZIA ++ PHILADELPHUS 4 CARPENTERIA + | 21 94 21 DECUMARIA + S HYDRANGEA ++ 17 [.— Cornus canadensis ++ 100 US Cornus walteri ++ H Cercidiphyllum 7 HYDRANGEOIDEAE 32 11 Oo ~ © (o x m «A m OUTGROUP 13 Hamamelis L ИЈКЕ 5. Majority-rule consensus tree (including compatible groupings) constructed from 36 most parsimonious trees resulting iram phylogenetic analysis of rbcL sequences of the asterid clade. This tree is identical to one of the Volume 80, Number 3 Morgan & Soltis 645 3 Phylogenetic Relationships Among Saxifragaceae TaBLE 2. Embryological characteristics of the subfamilies of Saxifr d sensu lato and additional related families. Sources: Dahlgren (1975b, 1983); Davis (1966); Philipson (1974, 1977); Schulze- Menz (1964); Sharma (1968); Spongberg (1972). Types of endosperm formation: “N” = nuclear; "C" = ab initio cellular; “H” = helobial; “2 = unknown Number of Taxa integuments Nucellus Endosperm type Saxifragoideae 2 crassinucellate C, H Penthoroideae 2 crassinucellate C Tetracarpaeoideae 2 crassinucellate ? Ribesoideae 2 crassinucellate C,H Iteoideae 2 crassinucellate ? Pterostemonoideae 2 crassinucellate ? Baueroideae 2 crassinucellate N Francooideae 2 crassinucellate N Brexioideae 2 tenuinucellate N Parnassiok 2 tenuinucellate N Lepuropetaloideae || ? ? Hydrangeoideae l tenuinucellate C Phyllonomoideae 1 ? ? Escallonioideae 1 tenuinucellate N Montinioideae 1 tenuinucellate ? Vahlioideae 2 tenuinucellate C Eremosynoideae 1 tenuinucellate ? Aquifoliaceae 1 crassinucellate, tenuinucellate C Celastraceae 2 crassinucellate, tenuinucellate N Cornaceae l crassinucellate, tenuinucellate C Crassulaceae 2 crassinucellate C Cunoniaceae 2 crassinucellate N Greyiaceae 2 crassinucellate N Haloragaceae 2 crassinucellate C 1972). This is an extremely uncommon pollen type, genera of Saxifragoideae. Plouvier furthermore apparently found elsewhere only in Choristylis, suggested that both /tea and Brexia were related which is usually included in Esc a. In to Celastraceae, members of which also accumulate other pollen characteristics, Itea i is very much like — dulcitol. Jay (1970) reported that /tea and Brexia several genera in Escallonioideae, and the complete were essentially identical in flavonoid characteris- tectum and reduced ectoaperture of its pollen fur- tics, having equal amounts of leucodelphinidine and ther distinguish /tea from Saxifragoideae. leucocyanidine and lacking flavonols. Jay advo- Chemical evidence generally suggests that Леа cated a close relationship between /tea and Brexia is distinct from Saxifragoideae and more closely and an alliance between them and Escallonia. Bohm allied with other woody taxa. Plouvier (1965) found et al. (1988) confirmed the absence of flavonols in allitol in /tea and dulcitol in Brexia. He proposed Itea and remarked that, while many members of a close alliance between the two because allitol апі Saxifragaceae s. l. (both herbaceous and woody dulcitol are isomers, and allitol was not found in taxa from several subfamilies) produce flavonols, — 36 shortest trees and has a consistency index of 0.428 and a retention index of 0.486. Arrows point to nodes that did not occur in all shortest trees; all other nodes were present in all 36 most parsimonious trees. Numbers above each branch indicate the number of base | numbers in italics below each branch indicate the Se occurrence of each monophy hyletic group in the results of 100 bootstrap replicat es. Names of taxa of Saxifragaceae sensu lato are shown in capital letters d lace type. Occurrence of iridoids is indicated for all taxa: ponas genera known t to У produce iridoids, “+” enera that have not been e but are members of families ks da to have iridoids, — genera and me ие. of ‘families that are known not to жү iridoids, *?" = unknown for iridoid occurrence. Sources of data on iridoid occurrence: De: et al. "(1917. 1981); Jensen et al. (1975); Kaplan & Gottlieb (1982). 646 Annals of the Missouri Botanical Garden Itea is virtually alone in that it accumulates only flavones. Our data provide no зарра for a close alliance between /tea and lgren, | tajan, 1987) or for the йеп of Plouvier (1965) and Jay (1970) that /tea is closely related to Brexia. According to rbcL sequence data, Пеа is only distantly related to woody subfamilies such as Hydrangeoideae, Escallonioideae, and Monti- nioideae. Sequence results of rbcL suggesting a close relationship between /tea and Saxifragoideae are intriguing and parallel embryological charac- ters and the floral data of Bensel & Palser (1975b, d). The closest relative of /tea may be the subfamily discussed below, Pterostemonoideae. Pterostemonoideae. Evidence from rbcL se- quences suggests that there is a close alliance be- tween Pterostemon and [tea (Figs. 2, 3). The monogeneric subfamily Pterostemonoideae consists of shrubs native to Mexico and has traditionally been classified with the woody subfamilies Hy- drangeoideae, Escallonioideae, and Montinioideae. However, the position of Pterostemon is usually described as uncertain (Cronquist, 1981; Dahlgren, 1983; Hutchinson, 1973; Takhtajan, 1987). As with the woody genus /tea, Pterostemon is more closely allied in our results with Saxifragoideae than with Hydrangeoideae, Escallonioideae, and Mon- tinioideae. Embryological characteristics corroborate the placement of Pterostemon suggested by rbcL se- quence data. As with Saxifragoideae and Леа, Pterostemon is crassinucellate and bitegmic (Table 2), thus the genus is distinct from the tenuinucellate and unitegmic Hydrangeoideae, Escallonioideae, and Montinioideae. Saxifragoideae and еа also have cellular endosperm formation. The type of endosperm present in Pterostemon is, however, not nown. The presence of cellular endosperm for- mation in Pterostemon might add to the support for its alliance with Saxifragoideae and /tea. In contrast to embryological characters, some morphological and palynological features do not seem to support a close alliance among Pteroste- mon, Saxifragoideae, and /tea. Pterostemon has, in addition to five fertile stamens, five staminodes. It also has five carpels and little endosperm. /tea and Saxifragoideae usually have two carpels, co- pious endosperm, and no staminodes. Hideux & Ferguson (1976) described the pollen of Pterostemon as having a complete tectum, long ectoaperture, and a complex endoaperture, char- acteristics most similar to certain escallonioid gen- era. The tectum surface of Pterostemon is muc like that of /tea, although the short ectoaperture of Пеа differs from the condition in Pterostemon. Hideux & Ferguson (1976) did not, however, in- clude Pterostemon in their numerical analysis that included many more pollen characters. They also provided no further description of its pollen, and they had little to say about its possible relationships. According to the description of Erdtman (1966), the pollen of Pterostemon may not be similar to that of /tea, but with the incomplete descriptions of Pterostemon pollen given by both Erdtman (1966) and Hideux & Ferguson (1976), a more detailed palynological study is warranted. Pterostemon has not been studied in great detail; little is known about its wood anatomy, floral anat- omy, detailed embryological features, or secondary chemistry. Chemical, anatomical, and additional molecular evidence for Pterostemon might hel resolve the relationships of this poorly understood genus. For example, given the unique profile of /tea, secondary chemical data for Ptero- flavonoi stemon should prove useful. Ribesoideae. The results of rbcL sequence analysis indicate that the monogeneric, woody sub- family Ribesoideae is more closely related to Sax- ifragoideae than to such woody groups as Hydran- geoideae, Escallonioideae, and Montinioideae (Figs. 1-3). Like /tea, Ribes has been placed close to Saxifragoideae and other herbaceous members of Saxifragaceae s. /. (Dahlgren, 1983; Engler, 1928; Takhtajan, 1987; Thorne, 1992), although some (Cronquist, 1981; Hutchinson, 1973) place it clos- er to other woody members of Saxifragaceae s. l. Many studies have pointed to a close relationship between Ribes and Saxifragoideae, and a distant relationship between Ribes and other woody taxa. Hamel (1953) found Ribes to be homogeneous cytologically (n — 8) and advocated placing it very m 1963) also support a close relationship between Ribes and Saxifragoideae. Gelius (1967) found the floral morphology of Ribes to resemble that of Saxifragoideae much more closely than that of other woody Saxifragaceae s. /. Similarly, Bensel & Palser (1975a, d) found numerous features of floral anatomy shared by Ribes and Saxifragoideae and concluded that, according to floral features alone, Ribes should be included in Saxifragoideae. Plouvier (1965) found that the absence of cer- tain secondary chemicals in Ribes distinguished the genus from Saxifragoideae, but he also noted that there were chemical similarities suggesting a close alliance. Jay (1970) stated that flavonoid data Volume 80, Number 3 1993 Morgan & Soltis 647 Phylogenetic Relationships Among Saxifragaceae also support a close relationship between Ribes and Saxifragoideae. Serological data (Grund & Jensen, 1981) likewise indicate a close relationship between Ribes and Saxifragoideae and a distant relationship between көк and other woody Saxifragaceae s. L. The tegmic crassinucellate ovules with cellular or helobial endosperm formation (Dahlgren, 1975b; Philipson, 1974, 1977; Table 2) also supports a close relationship a Ribes and Saxifragoideae. The pollen of Ribes, however, is distinct from most Saxifragoideae. Ribes is distinguished from Saxifragoideae in having pollen that is 8- 9-aper- turate and spheroidal (Erdtman, 1966) with a com- plete, smooth tectum (Hideux & Ferguson, 1976). The pollen of Ribes is most similar to that of certain genera of Escallonioideae and, according to Hideux & Ferguson (1976), it could be accommodated in a broadly defined Escallonioideae including many other woody genera Results of rbcL sequence analysis agree with most experimental data in placing Ribes distant from such woody groups as Hydrangeoideae, Es- callonioideae, and Montinioideae. Our data also suggest a relationship between Ribesoideae and the Saxifragoideae- Iteoideae- Pterostemonoideae group, although support for the alliance is low (Figs. 2, 3) II. SUBFAMILIES ASSOCIATED WITH CRASSULACEAE Penthoroideae. Our rbcL sequence data sug- gest that Penthorum (the only genus of Pentho- roideae) is part of an alliance including Crassula- ceae and is most closely related to Myriophyllum (Haloragaceae; Figs. 2, 3). There siderable debate regarding the placement of Pen- thorum. Some have included this enigmatic genus in Saxifragaceae (Cronquist, 1981; Dahlgren, 1983), others in Crassulaceae (Hutchinson, 1973), and some recognize Penthoraceae (Takhtajan, 1987; Thorne, 1992). Penthorum has morpholog- ical characteristics of both Saxifragoideae and Crassulaceae and, depending on which features are emphasized, a case could be made for its alliance with either. In having a hypanthium, partially con- nate carpels, and nonsucculent leaves, Penthorum is more like Saxifragoideae. However, Penthorum is more similar to Crassulaceae in number of floral parts. Syncarpous gynoecia and shallow hypanthia in two crassulaceous genera (Diamorpha and Triactinia) also suggest a link between Penthorum as been con- and Crassulaceae. Penthorum has been the subject of numerous systematic studies, but the results have for the most part been inconclusive. Hamel (1953) found that n = 8 an in Penthorum and suggested that, since the same numbers were present in Saxifra- goideae, it could be included there. Hideux & Fer- guson (1976) found the complex, H-shaped en- doapertures in the pollen exine of Penthorum to be much like those of Sedum (Crassulaceae), but in other pollen characteristics Penthorum appeared intermediate between Saxifragoideae and Crassu- aceae. According to Jay (1970), Penthorum has more phenolic constituents in common with Saxi- fragoideae (notably ellagic acid) than with Cras- sulaceae. Soltis & Bohm that Penthorum was chemically distinct from Sax- 982), however, found ifragoideae in containing many unique, unidentified flavonoids. Also, while myricetin and flavones are found in Saxifragoideae, they are absent in Pen- thorum. Because Crassulaceae are a much more easily defined group without Penthorum, the genus is often allied with Saxifragoideae (Spongberg, 1972). The phylogenetic analysis of morphological and chemical data of Hufford (1992) suggests that Pen- thorum is more closely related to Saxifragoideae than to Crassulaceae. ENSE Haskins & Hayden the vaca characters that define Crassu- laceae. However, in a detailed anatomical study, they found that Penthorum also shares no definite synapomorphies with Saxifragoideae; most of its distinguishing features are either plesiomorphic or of uncertain status. Finding no anatomical or mor- phological characters to place Penthorum in either Saxifragoideae or Crassulaceae, Haskins & Hay- den suggested that the genus is most easily accom- modated in its own family (Penthoraceae). ur data also suggest that Penthorum should not be accommodated in either Saxifragoideae or Crassulaceae, although the genus is more closely related to Crassulaceae than to Saxifragoideae (Figs. 2, 3). The closest relative of Penthorum identified by rbcL sequence data is Myriophyllum (Halora- gaceae). A close relationship between Haloragaceae and Penthorum or Crassulaceae has apparently never been suggested. Haloragaceae are occasion- ally placed in the order Haloragales, often with Gunnera (Cronquist, 1981; Dahlgren, 1983); the family is also often placed in the Myrtales (Hutch- inson, 1973; Melchior, 1964a; Takhtajan, 1987). Haloragaceae are, however, distinguished from most families in Myrtales by several features, such as an aquatic habit, anemophily, reduced unisexual flowers, a partly apocarpous gynoecium, and a persistent, copious endosperm. Some of these fea- 648 Annals of the Missouri Botanical Garden tures (a partly apocarpous gynoecium and copious endosperm) are also found in Penthorum, Tetra- carpaea, and Crassulaceae, the taxa most closely associated with Myriophyllum in our analysis. Fur- thermore, Myriophyllum, Penthorum, and Cras- sulaceae have bitegmic crassinucellate ovules and 2). This is a rather uncommon combination of characters, with cellular endosperm formation (Table bitegmic crassinucellate ovules more often being associated with nuclear endosperm development. This latter, more common combination is found in Myrtales (Dahlgren, 1975b; Davis, 1966; Philip- son, 1974, 1977). The rbcL-supported alliances are generally in agreement with the suggestion of Haskins & Hay- den (1987) that Penthorum is perhaps most easily accommodated in its own family separate from both Saxifragaceae and Crassulaceae. The placement of Penthorum and Myriophyllum as sister taxa by sequence data, however, raises the new possibility of a close relationship between Penthorum and Haloragaceae. The embryological similarities be- tween Penthorum and Myriophyllum are intrigu- ing, and the relationship between these taxa should be investigated further. Sequence data for addi- tional genera of Haloragaceae would be extremely helpful. Another potential avenue of inquiry would be to investigate the flavonoids in Myriophyllum. Soltis & Bohm (1982) found the most distinctive feature of Penthorum to be the presence of nu- merous unidentified flavonoids different from any- thing found in Saxifragaceae. Identification of these compounds in Penthorum and determination of whether they are present in Haloragaceae would be useful. Tetracarpaeoideae. Tetracarpaeoideae con- sist only of the single shrubby species Tetracarpaea tasmanica Hook.f. Our rbcL results (Figs. 2, 3) indicate that Tetracarpaea is most closely allied with the herbaceous genera Penthorum and Myr- iophyllum, and only distantly related to most other woody taxa of Saxifragaceae s. l. This placement of Tetracarpaea does not parallel most classifica- tions of the genus, which usually place it as a close ally of other woody taxa of Saxifragaceae s. l. (e.g., Hydrangeoideae, Escallonioideae, Montinioideae). The placement of Tetracarpaea suggested by rbcL evidence is, however, similar to the classification of Thorne (1992), who considered it to be more closely related to Saxifragoideae and other her- baceous taxa. Although Tetracarpaea has usually been allied with other woody taxa of Saxifragaceae s. /., there have nonetheless been indications that the true alliances of this genus lie elsewhere. Hideux & Ferguson (1976) suggested that, based on overall pollen characteristics, Tetracarpaea is most closely related to Cunoniaceae, a woody family of Rosidae. Hils et al. (1988), however, discounted such a relationship because of floral morphology. Fur- thermore, in exine characteristics alone, the pollen of Tetracarpaea is different from that of Cunonia- ceae, and also unlike that of most other woody taxa of Saxifragaceae s. !. Tetracarpaea appears most similar in pollen exine features to Tellima and Astilboides, genera of Saxifragoideae (Hideux & Ferguson, 1976) Hils et al. (1988) revealed that, although the wood anatomy of Tetracarpaea was similar to that of other woody Saxifragaceae 5. l, the vascular anatomy of the nodes, leaves, and floral parts of Tetracarpaea was different from that of woody Saxifragaceae s. /. In addition, whereas the ovules in many other woody members of Saxifragaceae s. l. are unitegmic and tenuinucellate, in Tetra- carpaea they are identical to Saxifragoideae and Crassulaceae in being bitegmic and crassinucellate (Table 2). The closer relationship of Tetracarpaea to Saxifragoideae than to other woody Saxifraga- ceae s. l. suggested by embryological, anatomical, and morphological studies is corroborated by rbcL sequence data (Figs. 1–3) Some features of Tetracarpaea (e.g., hypogy- nous flowers with completely separate carpels) are distinctly like those of Crassulaceae, and rbcL se- quence data also provide support for a closer re- lationship of this genus to Crassulaceae than to Saxifragoideae. Tetracarpaea has not, however, been considered most closely related to Penthorum or Myriophyllum. Further investigation of the en- tire rbcL-supported Tetracarpaea-Penthorum ~ Myriophyllum alliance is in order. One possible course is to extend the identification of the many unknown flavonoids found in Penthorum (see above) to Tetracarpaea and Myriophyllum. Myricetin, which is found in both Saxifragoideae and Cras- sulaceae, is absent in Penthorum (Gornall et al., 1979; Soltis & Bohm, 1982), and its presence or absence should be confirmed in Tetracarpaea and Haloragaceae. Flavones are also found in Saxifra- goideae and Crassulaceae but are absent in Pen- thorum and are not yet known in Haloragaceae (Gornall et al., 1979; Soltis & Bohm, 1982). In addition, endosperm development, although known for Penthorum and Crassulaceae, has apparently not been determined for Tetracarpaea and several members of Haloragaceae and Haloragales. Volume 80, Number 3 1993 Morgan & Soltis 649 Phylogenetic Relationships Among Saxifragaceae MI. SUBFAMILIES ASSOCIATED WITH OTHER FAMILIES OF ROSIDAE Baueroideae. | Except for Engler (1890, 1928), Schulze-Menz (1964), and Hutchinson (1973), most workers have considered Bauera to be distinct from all other constituents of Saxifragaceae s. l. and more closely related to, or included in, Cu- noniaceae. Sequence data of rbcL are in complete agreement with a close relationship between Bauera and Cunoniaceae and a distant relationship between Bauera and all other members of Saxifragaceae s. l. (Figs. 2, 4). In supporting the alliance of Bauera with Ceratopetalum (Cunoniaceae), our data cor- roborate many other lines of evidence. Some fairly obvious morphological features shared by Bauera and Cunoniaceae that are rare or absent in most other members of Saxifragaceae s. l. are compound leaves and numerous stamens. Bensel & Palser (1975d) advocated a close rela- tionship between Bauera and Cunoniaceae because of similarities in floral anatomy. Among other fea- tures, they found that sepal-plane vascular bundles do not contribute to either androecial or gynoecial vasculature in Bauera, whereas they do in many other taxa of Saxifragaceae s. l. The ovules in Bauera, Cunoniaceae, Saxifra- goideae, and the subfamilies of Saxifragaceae s. l. already discussed here are bitegmic and crassin- ucellate. Whereas endosperm formation in Saxi- fragoideae and allied subfamilies (e.g., Ribesoideae, Penthoroideae) is cellular or helobial, in Bauera and Cunoniaceae it is nuclear (Table 2). In fact, only one other subfamily of Saxifragaceae s. L. (Francooideae) is known to have the same three embryological features (Table 2). Also, both Bauera and Cunoniaceae have a zigzag micropyle, whereas all other members of Saxifragaceae s. l. have more typical straight micropyles (Davis, 1966). Prakash & McAlister (1977) cited many embryological and seed characteristics that were similar in Bauera and Cunoniaceae but markedly different in other members of Saxifragaceae s. Bauera is also palynologic aliy similar to Cunoni- aceae. Erdtman (1966) reported the pollen of Bauera and Cunoniaceae to be similar in being two-colpate and approximately half as large as the pollen of Saxifragoideae. Hideux & Ferguson (1976) described exine features and results of a numerical analysis of many pollen characters that also indi- cated a close Bauera-Cunoniaceae relationship. Similarities in secondary chemicals also indicate a relationship between Bauera and Cunoniaceae (Jay, 1970). Recent phylogenetic analyses of morpho- logical, anatomical, and chemical characters (Huf- ford, 1992; Hufford & Dickison, 1992) indicate that Bauera should be included within Cunoni- aceae. In recent years it has rarely been suggested that Bauera is more closely related to other members of Saxifragaceae 5. l. than to Cunoniaceae. Dick- ison (1975) described vegetative patterns, anther dehiscence, and flower vestiture features that dis- tinguished Bauera from Cunoniaceae and sug- gested that Bauera was closer to Saxifragoideae than to Cunoniaceae. Nearly all other evidence, however, including rbcL sequence data, strongly suggests that Bauera should be included within or closely allied with Cunoniaceae. Eucryphia, the only genus in Eucryphiaceae, is also a close relative of Bauera and Cunoniaceae in our results (Figs. 2, 4). Fucryphia is a genus of shrubs and trees that, Bauera, is distributed in the Southern Hemisphere. Focke (1895) and Melchior (1964b) placed Eu- cryphia near Dilleniaceae and Ochnaceae, but most like Cunoniaceae and recent workers consider it closely related to or part 1967) and cytology 1959) supports a close relationship be- flavonoids (Bate-Smith et al., (Hamel, tween Eucryphiaceae and Cunoniaceae. More ге- cently, investigations of floral, leaf, and wood anat- omy (Dickison, 1978) and pollen features (Hideux PEU Ae between the two families. and Hufford & Dickison (1992) performed irum analyses of morphological, anatomi- cal, and chemical characters, finding that Eu- cryphia is very closely related to and possibly should be included in Cunoniaceae. Our rbcL se- quence data, therefore, correlate well with several lines of evidence in supporting a close alliance between Eucryphia and Cunoniaceae. Francooideae. Our rbcL sequence data sug- gest that Francoa (one of two genera in the sub- family) is distantly related to Saxifragoideae and all other members of Saxifragaceae s. /., and closely related to Greyia (Greyiaceae; Figs. 2, 4). In most classifications Francooideae has been closely allied with Saxifragoideae, either as part of a narrowly defined Saxifragaceae (Cronquist, 1981; Dahlgren, 1983) or as a closely related family (Hutchinson, 1973; Takhtajan, 1987; Thorne, 1 ) Some evidence indicates a close relationship be- tween Francooideae and Saxifragoideae. Bensel & Palser (1975b) found the floral anatomy of Fran- 650 Annals of the Missouri Botanical Garden coa to be very much like that of other genera in Saxifragoideae. Hamel (1953) reported the chro- mosome number of Francoa sonchifolia Cav. to e n — 26 and suggested a close relationship be- tween Francoa and Saxifragoideae because the same number also occurs in Saxifragoideae, al- though rarely. Jay (1970) also suggested that Fran- coa and Saxifragoideae were closely related, based on the shared occurrence of ellagic acid. Other evidence parallels rbcL sequence data in suggesting a distant relationship between Francoa and Saxifragoideae. Bohm et al. (1986) identified several flavonoids in Francoa (e.g., quercetin and kaempferol aglycones) that had not been reported for this genus by Jay (1970). They also found myricetin and rutinosides, compounds commonly found in Saxifragoideae, to be absent in Francoa. Bohm et al. (1986) concluded, in contrast to Jay (1970), that Francoa was distinct from Saxifra- goideae and should be recognized as a separate family. Hideux & Ferguson (1976) found the com- plex apertures of both genera in Francooideae (Francoa and Tetilla) to be different from the diffuse and simple endoapertures of Saxifragoideae. Their numerical analysis of 25 pollen characters also separated Francoa and Tetilla from Saxifra- goideae. Our sequence data indicate that the closest rel- ative of Francoa is Greyia (Figs. 2, 4). Greyiaceae have been allied with or included in Melianthaceae by some authors (e.g., Scholz, 1964). Greyia is, however, distinguished from Melianthaceae by sev- eral morphological and chemical features (Bohm & Chan, 1992), and recently the family has been included in the same order or suborder as Saxifra- gaceae (Cronquist, 1981; Dahlgren, 1983; Takh- tajan, 1987; Thorne, 1992). It has, however, never been thought to be closely related to Francoa. Although a close relationship between Francoa and Greyia may initially seem anomalous, other data do corroborate our rbcL results. Myricetin, while common in Saxifragoideae, is absent in both Francoa (Bohm et al., 1986) and Greyia (Gornall et al., 1979). According to Erdtman (1966), the pollen of Francoa is quite different from Saxifra- goideae, having an equatorial constriction. Hideux & Ferguson (1976) revealed that the pollen of both genera of Francooideae is very similar to that of Greyia. There are also floral characters that dis- tinguish Francoa from Saxifragoideae and support its relationship with Стеуга. Whereas most Saxi- fragoideae have two carpels, Francoa has four (Greyia has five). Staminodes are not known in Saxifragoideae, but Francoa and Greyia both pos- sess staminodes equal in number to fertile stamens. Finally, in both Francoa and Greyia endosperm formation is initially nuclear, whereas Saxifragoi- deae (as well as Crassulaceae and other taxa related to Saxifragoideae) have cellular (or sometimes he- lobial) endosperm formation (Table The results of rbcL sequence analysis also sug- gest that there is an alliance between the Ртапсоа– Greyia group and Уплата, although support for this relationship is low (Figs. 2, 4). Viviania, which has b ognized as a separate, monogeneric family, is poor- ly known chemically and embryologically. At this time we cannot, therefore, provide useful compar- isons between this rbcL-supported alliance and oth- er lines of evidence; the possible relationship be- een either included in Geraniaceae or rec- tween Viviania and Greyia-Francoa must be investigated further. The rbel dat E! a close alliance between Brexia (one of three gen- Brexioideae. suggest era of Brexioideae) and Kuonymus (Celastraceae), separating Brexia from Saxifragoideae and most other members of Saxifragaceae s. l. (Figs. 2, 4). Brexia has typically been associated with Escal- lonioideae and other woody subfamilies of Saxifra- gaceae s. !. (Cronquist, 1981; Engler, 1890, 1928; Hutchinson, 1973; Thorne, 1992). Dahlgren (1983), however, allied Brexia more closely with a narrowly defined Saxifragaceae than with Es- callonioideae. Takhtajan (1987) placed Brexiaceae with Celastraceae (in Celastrales). The relationships of Brexia have been difficult to ascertain because different characters unite the genus with different groups of plants. Some evi- dence appears to unite Brexia with other woody Saxifragaceae s. /., especially Escallonioideae. As noted above, Jay (1970) interpreted flavonoid data to support a close relationship between /tea, Brexia, and Escallonia. Pollen exine characteristics of Brexia (Hideux & Ferguson, 1976) are most sim- ilar to those of Argophyllum (Escallonioideae). А numerical analysis of many more pollen characters also supported inclusion of Brexia in a broadly defined Escalloniaceae, containing such additional taxa as Пеа and Tetracarpaea (Hideux & Fer- guson, 1976). Plouvier (1965) not only supported a Brexia-—Itea relationship but also suggested а relationship between Brexia and Celastraceae be- cause both produce dulcitol. Brexia differs from both Saxifragoideae and Es- callonioideae in floral morphology and anatomy. Bensel & Palser (1975a) provided a long list of floral features present in Brexia and absent or rare in either or both Saxifragoideae and Escallonioi- deae, including a completely superior ovary, five Volume 80, Number 3 1993 Morgan & Soltis 651 Phylogenetic Relationships Among Saxifragaceae carpels, isomerous flowers, completely connate car- pels, staminodes, anther extensions, thick. carpel walls, and evidence that the androecium was de- rived from fascicled stamens. However, because Brexia differed more substantially from other members of Saxifragaceae s. l., Bensel € Palser 19754) tentatively allied the genus with Escalloni- oideae, pending further investigatior The close relationship between и апа Fu- onymus suggested by rbcL sequence data corrob- orates the placement of Brexia in Celastrales by Takhtajan (1987) and the suggestion that Brexia and Celastraceae are closely related based on chemical evidence (Plouvier, 1965). The five-car- pellate, isomerous flowers with superior ovaries found in Brexia also characterize Kuonymus. In addition, the ovules in most taxa of Saxifragaceae s. L. are either bitegmic and crassinucellate or uni- tegmic and tenuinucellate. Brexia, however, has bitegmic and tenuinucellate ovules, as does Eu- onymus. Many members of Saxifragaceae s. l. also have cellular endosperm formation; in contrast, Brexia and Fuonymus have nuclear endosperm formation (Table 2). The combination of these three embryological characteristics is relatively uncom- mon not only in Saxifragaceae s. l, but also in 1975b; Thus, embryological data further dicots in general (Dahlgren, 1974, 1977). support a close relationship between Brexia and Philipson, Euonymus. Although rbcL sequence data and embryological evidence suggest a close relationship between Brexia and Fuonymus, inclusion of Brexia in Celastraceae would be premature at this time. Fuonymus may not be typical of Celastraceae or Celastrales. Some taxa in Celastraceae (e.g., Celastrus) have cras- sinucellate ovules (Davis, 1966), and in other char- acteristics the family is a very diverse group (Cron- 1981). Other families in Celastrales (sensu Cronquist) are also different from Fuonymus em- quist, bryologically, and the only other one for which rbcL sequences are available (Aquifoliaceae) is, according to rbcL evidence, very distantly related to Euonymus (Figs. 1, 5; see below). Because basic chemical data for Brexia are lacking, no insights into the relationships of this genus can presently be gained through comparative chemistry. It is noteworthy that Celastraceae possess a rare fla- Mrs type (5-deoxy flavonoids, Gornall et al., 79). It would, termine whether these compounds are also present therefore, be of interest to de- in Brexia. Further resolution of the potential Brexia-Celastrales alliance should await further chemical and molecular investigation of Celas- trales. Parnassioideae and Lepuropetaloideae. Our rbcL sequence evidence indicates a close re- lationship between Parnassia and Lepuropetalon (Figs. 2, 4). These two genera were initially allied as subfamily Parnassioideae of Saxifragaceae (Dan- dy, 1927). Most recent classifications consider them closely related to one another; many authors retain them in narrowly defined versions of Saxifragaceae 1981; 1983; Hutchinson, 1973) or in separate families in the same order Takhtajan, 1987; T 1992 (1983), although allying the two genera, included them in Droserales (as the families Parnassiaceae and Lepuropetalaceae) rather than in Saxifragales. (Cronquist, Dahlgren, ~ Гћогпе, Dahlgren Parnassia and Lepuropetalon share many mor- phological characteristics, such as a dimorphic an- droecium with staminodes, similar epidermal tannin sacs, commisural stigmas, loculicidal capsules, and a scanty endosperm (Spongberg, 1972). Gastony & Soltis (1977) obtained a chromosome count of — 46 for Lepuropetalon and established that two base chromosome numbers occur in Parnassia (x = 8, 9). Because of the dramatic difference in diploid numbers between Parnassia and Lepuro- petalon, they did not consider cytological data to either confirm or deny a close relationship between the two. The complex mixtures of quercetin and kaempferol derivatives discovered in Parnassia by Bohm et al. (1986) clearly distinguish this genus from Saxifragoideae. Bohm et al. (1986) also found that the flavonoid constitution of Lepuropetalon was more similar to Parnassia than to any other investigated taxa of Saxifragaceae s. /. In pollen exine features, Parnassia and Lepuropetalon are much alike, with partial, continuous, and reticulate exines and similar endo- and ectoapertures. A nu- merical analysis involving many additional pollen characters did not, however, place the two close together (Hideux & Ferguson, 1976). Parnassia is the better known of the two genera. Although Parnassia has been closely allied with Saxifragoideae in many classification schemes (Ta- ble 1), many lines of evidence do not support such a relationship. Jay (1970) claimed to have found support from flavonoids for allying Parnassia with Hamel (1953) pre- ferred to ally Parnassia very closely with Saxifra- Hypericaceae. [n contrast, goideae because of similar karyotypes and because the chromosome numbers reported for Parnassia at that time (2л = in Saxifragoideae. Bensel € Palser (19752) found, however, that Parnassia shared very few floral characteristics 16 and 32) were also present with Saxifragoideae and proposed treating it аза separate family. They described Parnassia as hav- 652 Annals of the Missouri Botanical Garden ing hypogynous flowers, completely connate ova- ries, mostly commissural stigmas, nectariferous staminodia, and an ancestrally fascicled androe- cium; all of these features are absent or rare in Saxifragoideae. Grund & Jensen (1981), in a se- rological investigation of seed antigens of Parnas- sia, detected no reactions with either Saxifraga or Hydrangea, suggesting distant relationships with both (they did not, however, investigate Lepuro- petalon). Parnassia is also embryologically differ- ent from Saxifragoideae. It has bitegmic tenuinu- cellate ovules and nuclear endosperm formation, whereas bitegmic crassinucellate ovules and cel- lular endosperm characterize Saxifragoideae (Table 2). Thus, our rbcL sequence data are in agreement with several lines of biosystematic data in sup- porting a close relationship between Parnassia and Lepuropetalon and a distant relationship between both genera and Saxifragoideae (Figs. 2, 4). Parnassia and Lepuropetalon are part of a well- supported clade based on rbcL sequence data that also contains Brexia and Euonymus (Figs. 2, 4). It is important to note that other lines of evidence suggest a close alliance among all the members of the group. The combination of embryological fea- tures found in Brexia, Euonymus, and Parnassia is found commonly only in Ebenales, Theales, and Primulales (Dahlgren, 1975b; Philipson, 1974, 1977; Table 2). Brexia, Parnassia, and Lepu- ropetalon also possess staminodes and, as noted by Bensel & Palser (1975a), Brexia and Parnas- sia are similar in having an ancestral fascicled androecium, with several small vascular bundles supplying each stamen. i petalon, considered the smallest terrestrial angio- The diminutive Lepuro- sperm, is not well known in most respects. Schulze- Menz (1964) reported it as having unitegmic ovules, in which it would differ from Parnassia and Brexia. This should be confirmed, however, and the nucellus and endosperm types of Lepu- ropetalon should also be determined. A better knowledge of Lepuropetalon as well as of other taxa in Celastraceae should add significantly to an understanding of this alliance. IV. SUBFAMILIES ASSOCIATED WITH CORNACEAE Hydrangeoideae. Our analysis of rbcL se- quence data indicates that the closest relatives of Hydrangeoideae are Cornaceae and related genera (Fig. 5). There have been long-standing disagree- ments regarding the proper placement of the gen- era of Hydrangeoideae (see Table 1). Whereas Schulze-Menz (1964) and Cronquist (1981) pre- ferred a relatively close relationship between this group of genera and Saxifragoideae, Hutchinson (1973), Dahlgren (1983), Takhtajan (1987), and Thorne (1992) allied the group (along with other woody subfamilies) with taxa that are distantly re- lated to Saxifragoideae, such as Cornaceae or Cu- noniaceae. It is significant, therefore, that rbcL sequence data suggest that the five genera of Hy- drangeoideae analyzed are allied with Cornaceae, in agreement with the relationships suggested b Huber (1963), Aoi (1983), Takhtajan er and Thorne (1992) Most recent evidence has made it apparent that Hydrangeoideae are only distantly related to Sax- ifragoideae. Bate-Smith (1962) reported that el- lagic acid was present in Saxifragoideae, but absent in Hydrangeoideae, suggesting that the two were distantly related. Jay (1970) found the same dis- tribution of ellagic acid but concluded that Hy- drangeoideae were less distantly related to Saxi- fragoideae. Perhaps the first to draw attention to similarities between Hydrangeoideae and Cornales was Huber (1963), who cited such shared features as scalariform perforations in the xylem and the absence of stipules and recommended placing Hy- drangeoideae within Cornales. Bensel & Palser (1975c, d) found several an- atomical characteristics that distinguished Hydran- geoideae from Saxifragoideae, such as an increase in stamen numbers, highly vascularized gynoecia, absence of hypanthia, and three-trace petals with some bundles originating from the sepal planes. Krach (1977) also decided that several character- istics of Hydrangeoideae, such as a septicidal cap- sule, numerous seeds, reduced embryo size, and a lengthening of cell walls of the outer seed coat, indicated a distant relationship with Saxifragoideae. Palynological data also distinguish Hydrangeoideae from Saxifragoideae. Erdtman (1966) described the pollen of Philadelphus as being half as large as the pollen of Saxifragoideae. Hideux & Ferguson (1976) found several genera of Hydrangeoideae to be remarkably similar in exine characteristics. Their numerical analysis of additional pollen characters also indicated that Hydrangeoideae are homoge- neous and well separated from Saxifragoideae. Although the investigations discussed above all found characteristics that distinguished Hydran- geoideae from Saxifragoideae, few (e.g., Huber, 1963) proposed alternative relationships for the hydrangeoids. Grund & Jensen (1981) found that genera of Hydrangeoideae had a higher serological affinity with Cornus than with any of the numerous other samples tested, including Escallonia, Ribes, Parnassia, and representatives of Byblidaceae, Er- icaceae, Caprifoliaceae, Crassulaceae, and Cunoni- aceae. Significantly, they reported a low affinity Volume 80, Number 3 1993 Morgan & Soltis 653 Phylogenetic Relationships Among Saxifragaceae between Hydrangeoideae and several generea of Saxifragoideae. Thus, serological data clearly sup- port phylogenetically distant from Saxifragoideae and uber's contention that Hydrangeoideae are closely related to Cornaceae. Additional support for a close relationship be- tween Hydrangeoideae and Cornaceae comes from investigations of secondary chemicals, iridoids in particular. Iridoids occur primarily in sympetalous | Oleales, Gentianales, Scrophulariales, and Lamiales. They taxa, such as Dipsacales, Goodeniales, also occur in Cornaceae, as well as genera often considered allied with Cornaceae (e.g., Alangium, Aucuba, Curtisia, Davidia, Nyssa). Significantly, iridoids have also been found in Deutzia, Fendlera, and Hydrangea of Hydrangeoideae (Dahlgren et al., 1981; Jensen et al., 1975; Kaplan & Gottlieb, 1982). The presence of iridoids in Hydrangeoideae, coupled with their absence in Saxifragoideae and many other taxa of Saxifragaceae s. l, indicate that Hydrangeoideae are more closely allied with Cornaceae and other iridoid-producing groups than with Saxifragoideae or other members of Saxifra- gaceae 5. !. Similar alliances are suggested by fla- vonoid data. Whereas myricetin is common in Saxifragoideae, it is rare or absent in both Hy- drangeoideae and Cornaceae (Bohm et al., 1985; Gornall et al., 1979) Most of the plant groups containing iridoids also have ovules that are unitegmic and tenuinucellate. The ovules of Hydrangeoideae are also unitegmic and tenuinucellate, distinct from the bitegmic and crassinucellate ovules of Saxifragoideae. Although some of the cornalean taxa in this analysis differ from Hydrangeoideae in having crassinucellate ovules, all have ovules that are unitegmic and most have cellular endosperm formation (Dahlgren, 75b; Davis, 1966). Recently, Hufford (1992) presented results from a phylogenetic analysis of many morphological, anatomical, and chemical characters that supported an alliance between Hy- drangeoideae and Cornales. In summary, rbcL ev- idence, in combination with chemical, embryolog- ical, morphological, and anatomical characteristics, strongly supports the separation of Hydrangeoideae from Saxifragoideae and indicates that they are probably most closely related to Cornales. The Hy- drangeoideae—Cornaceae relationship is the subject of a more detailed rbcL sequence analysis by Xiang et al. (1993). V. SUBFAMILIES ASSOCIATED WITH ASTERIDAE Phyllonomoideae. | Phyllonoma, the only ge- nus in the subfamily, is not, according to rbcL sequence evidence, closely alied with any other members of Saxifragaceae s. /. Rather, it is part of an alliance that is composed primarily of sym- petalous taxa usually placed in Asteridae (Fig. 5). Members of this asterid group almost exclusively have ovules that are unitegmic and tenuinucellate (Dahlgen, 1975b; Davis, 1966), and many of them produce iridoid compounds (Fig. 5). Phyllonoma is included in this group as a rel- ative of Helwingia and Пех (Fig. 5). Helwingia is usually included within Cornales (a detailed dis- cussion of Cornaceae and related taxa is provided by Xiang et al., 1993). Hex though it has also been associated with Cornales (Dahlgren, 1983), is most often included in Celas- trales. Our analysis of rbcL sequence data indi- (Aquifoliaceae), al- cates, however, that both Helwingia and Пех are far removed phylogenetically from the groups with which they have traditionally been allied. Most authors have placed Phyllonomoideae with other woody members of Saxifragaceae s. /. The genus has frequently been considered most closely related to Escallonioideae (Table 1), but its place- ment is often described as uncertain. Compared with most other subfamilies of Saxifragaceae s. l., Phyllonomoideae are poorly known; very little em- bryological, anatomical, or chemical evidence has been obtained. Palynological data are available, however, and Erdtman (1966) and Hideux & Fer- guson (1976) found the pollen of PAyllonoma to iffer from that of other saxifragaceous taxa. The combination of a complete, granulate tectum and a simple endoaperture in Phyllonoma was unlike the pollen of any other subfamily of Saxifragaceae s. L Hideux & Ferguson (1976) also found that, based on many additional pollen features, Phyl- lonoma could not easily be accommodated in any saxifragaceous group and differed most from Es- callonioideae and Hydrangeoideae. In contrast to pollen characters, seed coat features suggest that Phyllonoma is closely related to Ribes (Krach, 1977). In a revision of Phyllonoma, Mori & Kal- lunki (1977) discussed its generic relationships only briefly but also suggested that it was closely related to Ribes. Phyllonoma has never been closely associated with /lex, and it has only indirectly been associated with Helwingia. Nonetheless, there is evidence that corroborates the relationships suggested by rbcL sequence data. All three of these genera have separate petals, a feature that distinguishes them in this group of mostly sympetalous taxa. Evidence supporting the alliance of Phyllonoma and Hel- wingia is provided by leaf morphology and inflo- rescence features. Hickey olfe (1975) noted that the leaves of the two genera have derived brochidodromous venation, possibly because they 654 Annals of the Missouri Botanical Garden share a common ancestor. Perhaps the most in- triguing evidence linking Phyllonoma and Hel- wingia is the fact that both genera have inflores- cences borne on the upper surfaces of the leaves. Besides Phyllonoma and Helwingia, epiphyllous inflorescences are found in at least 43 other genera in 24 families (Johnson, 1958; Dickinson, 1978). Most are assumed to have arisen by adnation of peduncles and petioles. Because various degrees of adnation can be found in individual families or genera, this characteristic has not been thought to be useful for phylogenetic purposes. Not all epi- phyllous inflorescences are ontogenetically identi- cal, however. Several taxa with this feature have been studied in detail. In some, the epiphyllous inflorescences have arisen from peduncle—petiole adnation: Dichapetalaceae (Stork, 1956), Chirita (Gesneriaceae; Johnson, 1958), and Turnera (Tur- neraceae; Johnson, 1958). The epiphyllous inflo- rescences of Phyllonoma and Helwingia are, how- ever, ontogenetically different from others that have been studied because they arise from the upper surfaces of the leaf primordia (Dickinson & Sattler, 1974, 1975; Johnson, 1958). This shared char- acteristic adds considerable support to the alliance of Phyllonoma and Helwingia. Additional evidence relevant to the PAyllono- та- Helwingia- Пех relationship may come from secondary chemicals, chromosome data, and em- bryological features. Phyllonoma, Helwingia, and lex comprise one of several lineages in this pri- marily asterid alliance in which iridoids have not been found (Fig. 5). Helwingia and Пех are not known to produce iridoids (Dahlgren, 1983; Dahl- gren et al., 1981) but Phyllonoma has not been tested for iridoids. It would, therefore, be of value to determine whether PAyllonoma possesses iridoid compounds. The chromosome numbers of Helwin- gia (x = 19) and /lex (x = 20) are similar (Raven, 1975); a chromosome count for Phyllonoma should also be useful. Пех and most members of Aquifolia- ceae have unitegmic and crassinucellate ovules, a relatively uncommon combination (Dahlgren, 1975b; Philipson, 1974, 1977). Phyllonoma and Helwingia have also been reported to have uni- tegmic ovules (Schulze-Menz, 1964), but their nucellar type is unknown. A finding of crassinu- cellate ovules in Phyllonoma and Helwingia should provide further support for this alliance. Escallonioideae. As with РћуЏопота, rbcL sequence analysis places Escallonia in an alliance of asterid taxa (Fig. 5). Escallonioideae have most often been associated with the woody subfamilies Hydrangeoideae and Montinioideae (Table 1). Our rbcL sequence data indicate, however, that Ё%- callonia is well separated from both. Although clearly separating Escallonia from all other mem- bers of Saxifragaceae s. L, rbcL sequence data provide little support for a close alliance of the genus with any other lineage (Fig. 5). Escallonioideae are the most morphologically di- verse of the 17 subfamilies of Saxifragaceae s. /., and appears to be polyphyletic. There is no con- sensus concerning the composition of the subfam- ily; several genera that have been placed in Es- callonioideae (e.g., Corokia, Griselinia) have also been associated with families such as Cornaceae and Araliaceae (Eyde, 1966; Philipson, 1967). The morphological diversity encountered within Escallonioideae has complicated attempts to deter- mine the relationships of the group. Bensel & Pal- ser (1975c) studied five members of Escallonioi- deae (Corokia, Carpodetus, Quintinia, Escallonia, and Polyosma) and described them as being very diverse in floral morphology and floral vascular patterns. Similarly, Hideux & Ferguson (1976) discovered that pollen morphology among genera of the subfamily was so diverse that several other subfamilies of Saxifragaceae s. l. could easily be accommodated within the variation. Other evidence indicates considerable evolution- ary distance between Escallonioideae and Saxifra- goideae. Serological affinities of Escallonia are much higher with Roridula, Sambucus, Lonicera, Cornus, and Hydrangeoideae than with Saxifra- goideae (Grund & Jensen, 1981). Several other features distinguish Escallonioideae from Saxifra- goldeae; in many instances Escallonioideae are much like Hydrangeoideae. In seed characteristics Es- callonioideae and Hydrangeoideae are similar, and both groups differ from Saxifragoideae (Krach, 1977). Ellagic acid is absent from both Escallonioi- deae and Hydrangeoideae but present in Saxifra- goideae (Bate-Smith, 1962; Jay, 1970). Iridoids are present in Escallonioideae and Hydrangeoideae but absent in Saxifragoideae (Dahlgren et al., 1981; Jensen et al., 1975; Kaplan & Gottlieb, 1982). Both Escallonioideae and Hydrangeoideae also have unitegmic, tenuinucellate ovules (Table 2). Our rbcL sequence data agree in part with the above described lines of evidence by indicating that Escallonia is phylogenetically distant from Saxi- fragoideae and closely related to other taxa that produce iridoids and that have unitegmic (and usu- ally tenuinucellate) ovules. The closest allies of Escallonia are not, however, Hydrangeoideae and Cornaceae (Fig. 5). There are, in fact, embryolog- ical and morphological features that parallel rbcL evidence in distinguishing Escallonia from Hy- drangeoideae and/or Cornaceae. Endosperm for- mation is nuclear in Escallonioideae but cellular in Volume 80, Number 3 1993 Morgan & Soltis Phylogenetic Relationships Among Saxifragaceae Hydrangeoideae and Cornaceae. Whereras ovules in Cornaceae are mostly crassinucellate, in Escal- lonia and other genera of Escallonioideae they are tenuinucellate (Table 2). Genera of Escallonioideae are also distinguished from Hydrangeoideae and Cornaceae by exhibiting haplostemonous and sym- petalous tendencies (Dahlgren, 1983); these char- acteristics also are consistent with the placement of Escallonia with taxa of Asteridae. Further resolution of the relationships of 5 callonia is not possible at present with available rbcL sequence data because of the lack of evidence supporting specific alliances (Fig. 5). Given the great morphological diversity present in Escalloni- oideae and the poor understanding of generic re- lationships in the group, it will be necessary to investigate additional genera to gain a complete understanding of the systematics of the entire sub- family. Our rbcL sequence data have already been helpful in exploring the relationships of two genera, Corokia and Griselinia, that have been placed in Escallonioideae by some authors. Sequence data indicate that Corokia and Griselinia are not closely related to Escallonia or to one another (Fig. 5). It should be possible to gain a better understanding of Escallonioideae by obtaining rbcL sequence data from other members of the subfamily. Montinioideae. | Montinia represents still an- other member of Saxifragaceae s. l. whose affinities appear to be with taxa of Asteridae. The closest relatives of Montinia, according to rbcL sequence evidence, are Convolvulaceae and Solanaceae (Fig. 5), both of which are normally placed in Solanales (Asteridae). Montinioideae (consisting of the Montinia and Grevea) have been allied with many different taxa (e.g., Myrtaceae, Celastraceae, Cu- ahlgren et al., two genera curbitaceae, Onagraceae; see 1977). The subfamily is most commonly allied, however, with other members of Saxifragaceae s. L, especially the woody subfamilies Hydrangeoide- ae and Escallonioideae (Table 1). share several features with Hydrangeoideae and Escallonioideae that are not characteristic of many other members of Saxifragaceae s. !. АП three subfamilies have unitegmic tenuinucellate ovules (Table 2), lack ellagic acid (Bate-Smith, 1962; Jay, 1970), and produce iridoids (Dahlgren et al., 1977, 1981; Jensen et al., 1977; Kaplan & Gottlieb, 1982). According to rbcL sequence data, Montinia does not have Hydrangeoideae or Es- callonioideae as closest relatives, although all three subfamilies do appear near members of Asteridae. Montinioideae however, Importantly, some more traditional lines of system- atic data also do not closely link Montinioideae with Hydrangeoideae or Escallonioideae. Krach (1976) found many differences between the seeds of Mon- tinioideae and Escallonioideae and suggested that they were distantly related. Takhtajan (1980) stat- ed that some characteristics of Montinioideae (e.g., simple perforation plates in the xylem, unisexual flowers) indicate that they are much more advanced than either Hydrangeoideae or Escallonioideae. Some morphological and anatomical features of Montinia are also present in Solanales. Both Mon- tinioideae and Solanales have two-carpellate gy- noecia and simple perforation plates in their xylem. Montinioideae are dioecious and have capsular fruits; dioecy (Levine & Anderson, 1986) and cap- sular fruits also occur in Solanales. Most of these shared morphological and anatomical features are, however, also found in many other taxa. Numerous features distinguish Montinia from Solanales; we will mention only the most obvious. In contrast to both Solanaceae and Convolvulaceae, Montinia has inferior ovaries, free stamens, separate petals, produces iridoids and lacks internal phloem. Further study of Montinioideae should be useful in determining the closest relatives of this subfam- ily. For example, secondary chemical data may be helpful. According to Harborne & Swain (1979), Solanaceae have a unique flavonoid composition that is similar only to that of Convolvulaceae and Nolanaceae (also in Solanales). Specific chemical features shared by Solanaceae and Convolvulaceae include the presence of tropane alkaloids, otherwise known from only ten genera of six other families (Evans, 1979), and acylated anthocyanins, else- where found only in Lamiaceae and Polemoniaceae (Harborne & Swain, 1979). It would be useful to analyze Montinia for tropane alkaloids and acyl- ated anthocyanins. Dahlgren et al. (1977) described an unusual condition in Montinia in which the cotyledon petioles of the seedlings are In addition, fused. This characteristic is rare in flowering plants and is almost exclusively found in herbaceous taxa. It has been described in other woody plants only in Combretum (Combretaceae) and in one species of Quercus (Fagaceae). In herbaceous taxa, fused petioles are known in four genera each of Ranuncu- laceae and Apiaceae and in one genus in each o three other families (de Vogel, 1980). Lubbock (1892) reported no occurrences of fused petioles in Solanales, but a more thorough examination of seedling development of Solanales might provide useful insights. As noted by Chase et al. (1993), the misiden- tification of plant material should be considered a possibility, especially when relationships seem anomalous. It is significant, therefore, that our rbcL 656 Annals of the Missouri Botanical Garden sequence for a second specimen of Montinia con- firms the results presented here. Vahlioideae. According to rbcL sequence data, Vahlia (the only genus of Vahlioideae) is well separated from other members of Saxifragaceae s. l. and included in the same group of asterid families that contains Montinia (Fig. 5). Its inclusion in this group is well supported by rbcL sequence data. The results do not, however, indicate any close relatives for Vahlia in this alliance, but suggest that it is basal to several asterid taxa (Fig. 5). Vahlia typically has been affiliated with other herbaceous taxa of Saxifragaceae s. /. (Table 1), either as a subfamily of Saxifragaceae or as a closely related family. As with Pterostemon and Phyllonoma, Vahlia is very poorly known ana- tomically and chemically, and although it is usually placed close to Saxifragoideae, its affinities are much in doubt. Embryological and palynological characteristics suggest that Vahlia is most closely related to Parnassia. Ovules in both genera are bitegmic and tenuinucellate (Table 2), indicating that they are separate from other members of Saxifragaceae s. l. According to Hideux & Fer- guson (1976), Vahlia and Parnassia are also sim- ilar in many pollen characters. The chromosome numbers of Vahlia (n = 6, 9; Hamel, 1953) are also similar to those of Parnassia, which has x = 8 and 9 (Gastony & Soltis, 1977). Sequence evi- dence, however, indicates that Vahlia is distantly related to Parnassia. Determination of the геја- tionships of Vahlia would benefit greatly from more detailed systematic study. Secondary chemistry in- vestigations might help determine the affinities of Vahlia; since most of the other taxa in this asterid group produce iridoids, an evaluation of Vahlia for the presence or absence of iridoids would be especially valuable. Broader phylogenetic analyses of Chase et al. (1993) also indicate that Vahlia occupies a position basal to many asterid taxa. This position of Vahlia is intriguing in light of the fossil record. Friis & Skarby (1981, 1982) described a highly developed, apparently saxifragaceous flower (Scandianthus) from the upper Cretaceous period and proposed that Saxifragaceae s. /. may have been well dif- ferentiated as early as the lower Cretaceous. As noted by Dahlgren (1983), the Scandianthus flow- er is virtually identical to the flower of Vahlia, having an inferior, bicarpellate, unilocular ovary with two pendulous placentae, five sepals and petals of approximately equal length, and two separate stigmas (Егиз & Skarby, 1981, 1982). The two genera differ only in that Vahlia has five stamens and Scandianthus ten. Thus, fossil evidence in conjunction with our sequence data raise the pos- sibility that Vahlia could be a relict of the ancestral type of many of these asterid families. SUMMARY Our phylogenetic analysis of rbcL sequences indicates that, among angiosperm families, Saxi- fragaceae s. l. may represent the most extreme example of a polyphyletic assemblage. These anal- yses suggest that these taxa represent at least 10 separate evolutionary lines. Furthermore, these separate lineages are distributed among most tra- ditionally recognized subclasses of the dicots. These results are not wholly unexpected, however. Зеу- eral genera and subfamilies of Saxifragaceae s. l. are poorly understood (e.g., Montinioideae, Vah- lioideae, Lepuropetaloideae, Phyllonomoideae, Pterostemonoideae, Escallonioideae, Tetracar- paeoideae, Francooideae), and their affinities have actually been much in doubt. In addition, although members of Saxifragaceae 5. !. are allied with at least 10 separate lineages in our results, the re- lationships supported by rbcL sequence data are in many instances corroborated by other lines of evidence. Numerous different types of data are congruent with rbcL-supported alliances in specific instances, but some lines of evidence (such as those derived from embryology, iridoid chemistry, se- rology, and floral anatomy) are especially notable for the frequency with which they agree with rbcL sequence results. Our rbcL sequence data suggest that Saxifra- gaceae should be composed of only the genera included in the subfamily Saxifragoideae of Schulze- Menz (1964). Data from several sources, includ- ing intron loss in the chloroplast gene rp/2, re- striction site analysis of chloroplast DNA, and mor- phology all serve to distinguish this group of genera from traditionally allied taxa of Saxifragaceae s. l. It seems most appropriate, therefore, to define Saxifragaceae sensu stricto (Saxifragaceae s. s.) as consisting only of the approximately 30 genera of subfamily Saxifragoideae, a circumscription of the family that is identical to that proposed by Takh- tajan (1987) and Thorne (1992). Sequence evidence indicates that Iteoideae, Pterostemonoideae, Ribesoideae, Tetracarpaeoide- ae, Penthoroideae, Myriophyllum (Haloragaceae), and Crassulaceae are possible close relatives of Saxifragaceae s. s. (Figs. 2, 3). Although sequence data provide little direct support for a monophyletic Volume 80, Number 3 1993 Morgan & Soltis Phylogenetic Relationships Among Saxifragaceae alliance composed of these taxa, they do share several features. As reported by Bensel & Palser (1975a, b, c, d) most of these taxa are very much alike in the anatomy and morphology of their flow- ers. Furthermore, all of these taxa that have been studied have bitegmic crassinucellate ovules and cellular (sometimes helobial) endosperm formation. This combination of features is relatively uncom- mon in non-magnoliid dicots, the more common combination being bitegmic crassinucellate ovules with nuclear endosperm formation (Dahlgren, 1975b; Philipson, 1974, 1977; Table ical evidence (Grund & Jensen, 1981) also supports 2). Serolog- close relationships among many of the same taxa (Saxifragaceae s. s., Crassulaceae, Ribes, Pentho- rum). А phylogenetic analysis of many anatomical, morphological, and chemical characters (Hufford, 1992) similarly suggests that Saxifragaceae s. s., Crassulaceae, Penthorum, and Tetracarpaea are closely related. An intriguing result of this investigation is the placement of four genera near Saxifragaceae s. s. that are usually included in the subclass Hama- melidae: Cercidiphyllum, Daphniphyllum, Кћо- doleia, and Hamamelis (Figs. 2, 3). These genera have been described as part of the “lower Ham- amelidae" (Ehrendorfer, 1989) and, although rare- ly mentioned as close relatives of Saxifragaceae, they have occasionally been linked with the family (Dickison, 1989). Thorne (1992) included these hamamelid taxa in his Rosanae along with Saxi- fragales, Rosales, and Cunoniales, as well as the “higher hamamelid" taxa Fagales and Juglandales. Dahlgren (1983) placed Saxifragaceae and many hamamelids in his Rosiflorae, considering Saxifra- gaceae to be connected to the hamamelid taxa through Cunoniaceae. Grund & Jensen (1981) found a high level of serological affinity between Hamamelis and several genera of Saxifragaceae s. s. Significantly, this serological affinity is as high as that observed between Ribes and Saxifragaceae s. s. ог between Crassulaceae and Saxifragaceae $$, In addition to suggesting a narrow definition of Saxifragaceae and indicating several possible rel- atives, rbcL sequence data also indicate that other families often associated with Saxifragaceae s. s. e.g., Cunoniaceae, Greyiaceae, Droseraceae, Cephalotaceae, Gunneraceae, Rosaceae) are not closely related to it. Although Dahlgren (1983), Takhtajan (1987), and Thorne (1992) considered Cunoniaceae to be distantly related to Saxifraga- ceae s. s., Cronquist (1981) and Schulze-Menz (1964) placed Cunoniaceae in the same order as Saxifragaceae s. s. Our rbcL sequence evidence is in close agreement with Dahlgren (1983), Takh- tajan (1987), and Thorne (1992) in suggesting a distant relationship between Saxifragaceae s. s. and Cunoniaceae; only Baueroideae appear close to Cunoniaceae (Figs. 2, The rbcL sequence data also indicate that Greyi- aceae are distantly related to Saxifragaceae s. s. (Figs. 2, 4). This family is included in the same order or suborder as Saxifragaceae s. s. by almost all authors. According to rbcL sequence evidence, however, Greyiaceae are allied only with the her- baceous subfamily Francooideae (Figs. 2, 4), an alliance that has several lines of corroborating ev- idence (see discussion of Francoa above) Results from analysis of rbcL sequences also do not indicate a close relationship between Saxifra- gaceae s. s. and Droseraceae (Fig. 1), a family allied closely with Saxifragaceae s. s. and other erbaceous subfamilies of Saxifragaceae s. /. by Dahlgren (1983), Takhtajan (1987), and Thorne (1992). dence, distantly related not only to Saxifragaceae Drosera is, according to sequence evi- s. s. but also to all other subfamilies of Saxifra- gaceae s. l. (Fig. 1). The position of Droseraceae based on sequence evidence (near Nepenthes) agrees more closely with the classifications of Cron- quist (1981) and Schulze-Menz (1964). On the basis of floral morphology, Cephalotaceae, another insectivorous family, are often thought to be closely related to both Saxifragaceae s. s. and Crassula- ceae (Nicholls et al., 1985). Sequence data of rbcL, however, indicate that Cephalotaceae are most closely allied not with Saxifragaceae s. s. or Cras- sulaceae but with Platytheca (Tremandraceae) and Cunoniaceae (Figs. 2, 4). Gunnera is, according to rbcL sequence data, also distantly related to Saxifragaceae s. s. Gun- nera has sometimes been included in Haloragaceae (Cronquist, 1981) but has often been treated as the family Gunneraceae and placed close to Sax- ifragaceae (Dahlgren, 1983; Takhtajan, 1987). Тће affinities of Gunnera have remained uncertain be- cause of an unusual combination of characteristics (e.g., wind pollination, symbiotic algae, unisexual flowers, a bicarpellate unilocular ovary, solitary ovule). The basal position of Gunnera in the results of our analysis (Fig. 1) still make identification of its closest relatives difficult, but sequence data ob- viously do not indicate a relationship with Saxifra- gaceae s. s. or other subfamilies of Saxifragaceae um Rosaceae have been closely associated with Sax- ifragaceae by Dahlgren (1983) and Cronquist 658 Annals of the Missouri Botanical Garden (1981). Our results, however, suggest that Rosa- ceae are not closely related to Saxifragaceae s. s. but are allied with Ulmaceae, Moraceae, and Rhamnaceae (Figs. 2, 4) The results of rbcL sequence analysis indicate that five subfamilies of Saxifragaceae s. /. (Bau- eroideae, Francooideae, Brexioideae, Parnassioi- deae, and Lepuropetaloideae) are well separated from Saxifragaceae s. s. and associated taxa but are still included within an alliance composed pri- marily of families usually placed in Rosidae (Figs. 2, 4). Baueroideae and Francooideae are, accord- ing to rbcL sequence evidence, both allied with families (Cunoniaceae and Greyiaceae, respective- ly) that have identical embryological characteris- tics. Brexioideae and Parnassioideae are alike in having a relatively uncommon combination of em- bryological characteristics: bitegmic tenuinucellate ovules with nuclear endosperm formation, features that otherwise are commonly found only in Ebena- les, Primulales, and Theales (Davis, 1966; Philip- son, 1974, 7). The sequence evidence delin- eates a well-supported alliance that includes Brexioideae and Parnassioideae with the embryo- logically identical Fuonymus (Celastraceae). These three taxa also have many similarities in floral anatomy and morphology. Although Lepuropeta- loideae are poorly known embryologically, their inclusion in this alliance is well supported by many floral similarities with Parnassioideae. Another significant result of this study is the finding that five subfamilies of Saxifragaceae 5. 1. (Hydrangeoideae, Escallonioideae, Phyllonomoide- ae, Montinioideae, and Vahlioideae) are, according to rbcL evidence, distantly related to all other subfamilies of Saxifragaceae s. /. (Figs. 1, 5). Se quence data suggest instead a placement of these taxa in an expanded asterid alliance (see also Olm- stead et al., 1993). Other lines of evidence support these results; these five subfamilies are placed near- est to taxa that are similar in having such char- acteristics as iridoids and ovules that are unitegmic and tenuinucellate. CONCLUSIONS Phylogenetic analysis of rbcL sequence data has greatly improved our understanding of the rela- tionships among the morphologically diverse mem- bers of Saxifragaceae sensu lato. These data have: (1) shown that Saxifragaceae sensu lato are an extreme example of a ројурћујенс angiosperm family, (2) suggested that Saxifragaceae be nar- rowly defined to consist only of approximately 30 herbaceous genera, (3) indicated possible close rel- atives of Saxifragaceae sensu stricto, (4) revealed that many traditional relatives of Saxifragaceae sensu stricto are only distantly related, (5) shown that some lines of evidence, especially embryology, iridoid chemistry, and serology, are frequently in close agreement with rbcL sequence data, and (6) identified areas in which additional data from mo- lecular as well as more traditional investigations should increase our understanding of angiosperm systematics. LITERATURE CITED BarE-SMrrH, E. С. 1962. The phenolic constituents of plants and their taxonomic significance. J. Linn. Soc., Bot. 58: a . M. Davenport & J. B. HARBORNE. 1967. Comparaiive oa of flavonoids —A corre- lation between chemistry and plant geography in the genus Еистурћа. Phytochemistry 6: 1407-1413. BENsEL, C. R. & B. F. PALSER. 1975a. Floral anatomy in the Saxifragaceae sensu lato. I. Introduction, Par- nassioideae and Brexioideae. Amer. J. Bot. 62: 176- 185 75b. Floral anatomy in th use ашан: sensu be II. | and Iteoi- деае. Ате Во 67 & га Floral anatomy in the Saxifragaceae sensu lato. III. Kirengeshomoideae, пое and Escallonioideae. ges J. Bot. 62: 87. 1975d. Floral anatomy in the Saxifragaceae sensu lato. IV. еы and con- clusions. Amer. J. Bot. 62: 688-6 ЈА. E 1992. Fla id and affinities of Greyiaceae with a discussion of the occurrence of B-ring сое in dicotyledonous families. Syst. Bot. oW. Meth U. С. Внат. 1985. Fla- vonoids of the PLE c Dumortier. Biochem. Syst. Ecol. 13: — ——-, І. S. ром NEVAN & U. . Внат. 1986. Fla- vonoids of some species of ия Етапсоа, Раг- nassia and Lepuropetalon. Biochem. Syst. Ecol. 14: ,G. ‘CHALMERS & U. G. Внат. 1988. Flavonoids and the Ed we 1 to the Saxifragaceae. Phytochemistry 27: 653. Chase, M. W., D. E. Sorris, | C OLMSTEAD, | gn GAN, D. H M. R. Duv ‚С. F. hu gir J. F. SMITH, Wi QUINN, L. E. EGUIARTE, LIAMS, P. A. GADEK, т J. E. GoLENBERG, G t EARN, JR., S PA GRAHAM S. C. H. BARRETT, S. DAYANANDAN & V А. ALBERT. 1993. Р йде of seed plants: Ап analysis of nucleotide sequences from the plastid gene rbcL. Ann. Missouri Bot. Gard. 80: 528-580. CRONQUIST, A. 1981. An Integrated System of Classi- Volume 80, Number 3 1993 Morgan & Soltis 659 Phylogenetic Relationships Among Saxifragaceae fication of Flowering Plants. Columbia Univ. Press, New York. DAHLGREN, R. angiosperms to be use bution of characters. “ай "Not 1975b. The Доки of x an е system 1. Some embryological char- acters. Bot. Not. 128: 181- — "1980. А p system of ко of the angiosperms, J. Linr p Bot. я General мети of sper evolu- tion a міт и Nordic J. Bot. 3: 119- 149. A n of classification of the а ren the distri- -147 racters within S. R. JENSEN & B. J. NIELSEN. 1977. Seedling ван Којо and iridoid occurrence in — car- es ea (Montiniaceae). Bot. Not. 130: 329- 332 ———— ———— . А revised clas- КЕ of the angiosperms with comments on cor- relation between chemical and o is r characters. P 149-199 in D. A. Young & D. S. Siegler (editors), Phytochemistry and [sce Phylogeny. Prae- ger, New Yor Danby, J. E. 1927. bes genera of Saxifragaceae. Kew Bull. 1927: 107-11 Davis, G. L. 1966. NA embryology of the an- giosperms. John Wiley & Sons, New DICKINSON, T. A. ior Epiphylly in angiosperms. Bot. = 191.232 — & R. Sattler. 197 4. Development of the epi- phyllous inflorescence of Phyllonoma integerrima (Turez.) Loes.: earr "A сог r аи mor- phology. J. Linn. Soc к Де ——— ——— 1925. a of the epi- phyllous inflorescence of He a з a (Hel- r. J. Bot. 62: 962-97 Floral morphology 2| anatomy of Bauera. ,Phytomorphology 16. ———. 1 t ur of Eucryphi- aceae. Amer. J. Bot 1122130: 1 Comparisons of primitive Rosidae and andas, Pp. 4 n P. R. Crane & 5. Black- more (editors), A on, Systematics, and F History of the Hamamelidae, Vol. I. Introduction and ‘Lower’ Hamamelidae. Systematics Association, Ox- ford. DOWNIE, S. R., ES УУ С. ZURAWSKI, D. E. SoLTIS, P. S s, J. C. Watson & J. D. PALMER. 991. е ж rs chloroplast DNA rpl2 intron demarcates six lineages of dicotyledons: – Јаг “ phylogenetic implications. Evolution 45: 259. Doyle, J.J. & J. L. Doyte. 1987. А rapid DNA isolation procedure for "s A pupa of fresh leaf tissue. Phytochem. Bull. 13-15. EHRENDORFER, F. 1 а. The phylogenetic ПАРА of the Hamamelidae. Pp. 1-7 in P. R. Crane & 5. Blackmore (editors), Evolution, Systematic dF sil History of the Hamamelidae, Vol. I. а and ‘Lower’ Hamamelidae. Systematics Association, Oxford ENGLER, A Saxifragaceae. Pp. 42-93 in A. Engler & K. Prantl (editors), Die Natürlic :hen Pflan- zenfamilien, Vol. IH. Wilhelm Engelmann, ПОРЕ 192 буй айна. Pp. 74-226 in A. En- gler & K. Prantl (editors), Die Natürlichen rd zenfamilien, 2nd ed., Vol. 18a. Wilhelm Engelmann, 1966. Pollen Morphology and Plant Tax- onomy, Vol. 1. Angiosperms. (corre cted reprint of 1952 edition). Hafner, New York, Lon on. Evans, W. C. 1979 Tropane ella of the Solana- и Pp. 241-254 in J. С. Hawkes, . . D. Skelding (editors), The Biology and E onomy of the Solanaceae. Academic Press, New Y YDE, К. Н. 1966. Vigne anatomy of the di and on of Corokia. Amer. ot. 53: cid 847. Ше» W. 895. Kucryyliláceas. Pp. 129-131 in n r & K. Prantl (editors), Die ита ћеп venae Vol. Ш. Wilhelm Engelmann, Leip- ester Pu E M. & A. SKARBY. 1981. Structurally preserved angiosperm flowers from the Upper Cretaceous o southern Sweden. Nature 291: 484-480. & ————. 1982. Scandianthus gen. angiosperm "m of Saxifragalean affinity from the Cretaceous of southern Sweden. Ann. Bot. 583 Cision. Gj.: . E. Sorris. 1977. Chromosome studies of Parnassia and Din ia (Saxifra- . А new base cc for edu i ls 19: 13-5'78. GELIUs, L. 1967. Studien zur с. :hichte an Blüten dur Saxifragales sensu lato mit besonderer i > des Androeceums. Bot. Jahrb. Syst. 87: – 303. mig °З ‚ В. A. Boum & R. DAHLGREN. 1979. The ви йир of oda in the angiosperms. Bot" Not... 132: 1-30 n C&U. JENSEN. f the Saxifragales полни d by serological charac: teristics of seed proteins. Pl. Syst. Evol. 137 22 HaMEL, J. L. 195 Contribution à l'étude cyto-taxi- e pi Saxifragacées. Rev. Cytol. Biol. Vég. 14 15: 19 Sy t : ›] ti А p Contribution à l'étude caryo-taxino- mique у м» Bull. Mus. Hist. Nat. (Par 15) 31 HARBORNE, + A H T Swain. 1979. Solanaceae. Pp. 257-208 in J. 6. Haw Lester & A. D. Skelding (editors), The Biology and аи of the Solanaceae. Academic Press, ork. oue of the N ene М. L. € W. J. HAYDEN. 1987. Anatomy and affinities of Penthorum. Amer. J. Bot. 74: 164-177. Hickey, L. J. & J. A. Wore. 1975. The bases of angiosperm phylogeny: Vege tative morphology. Ann. Missouri Bot. Gard. 62: 538-589. HipEux, M. J. & I. К. FERGUSON. 1976 structure of the exine and its уййн signifi- P The stereo- cance in Saxifragaceae sensu lato. Pp. 327-377 in I. K. Ferguson & J. Muller (editors), The Evolution, ary Signifie ance of the Exine. Academic Press, W. C. Dickison, T. W. Lucansky & W. : oe aa anatomy and sys- ensues of woody Saxifra ии Tetracarpaea. Amer. J. Bot. 75: 1687- 170 Huser, H. 3. Die аса der Rosifloren. Mitt. Bot. Staatssamml. München 5 1992. Hurronp, L. Rosidae and their relationships to 660 Annals of the Missouri Botanical Garden other nonmagnoliid edd А phylogenetic ical data. Ann. Missouri Bot. Gar d. Be E A phylogenetic analysis of nicas: Syst. Bot. 17: 181-200. по J. 1973. The Families of Flowering Plants. Oxford Univ. Press, Oxford. Jav, M. 1970. Quelques | taxinomiques et phylogénétiques des Saxifraga vus à la lumière de la biochimie качан. Bull. Mus. Hist. Nat. (Paris) 42: Ls 75. JENSEN, S. R., B. J. NIELSEN & К. DAHLGREN. 1975. Iridoid compounds, their occurrence and systematic importance in the angiosperms. Bot. Not. 128: 148- 180 JOHNSON, M.A. 1 The epiphyllous flowers of Tur- nera = Helwingia. Bull. Torrey Bot. Club 85 KAPLAN, M. у C. & O. R. GOTTLIEB. 1982. lridoids as ipte markers in dicotyledons. Biochem. Syst. Ecol. 10: Ккасн, J. E. y ны der Rosifloren I. Die samen der Saxifragaceae. Bot. Jahrb. Syst. 97: m Seed agi qe in and affinities among the еа Рр. 141-153 іп К. Kubitzki (ed- itor), Flowering Plants — Evolution and Classification of Higher Categories. Springer-Verlag, Vienna, New York. Levine, D. А. & С. J. ANDERSON. 1986. Evolution of dioecy in an American Solanum. Pp. 264-273 in W. G. D'Arcy (editor), Solanaceae — Biology and Sys- tematics. Columbia Univ. Press, New York. LuBBOCK, J. 1892. Contribution to our knowledge of seedlings. Kegan Paul, Trench, Trübner & Co., Lon- MappisoN, D. R. 1991. The discovery and importance of multiple TUF of most-parsimonious trees. Syst. Zool. 40: 315-328. MELCHIOR, H. 64a. Myrtiflorae. Pp. 345-366 in H. Melchior (editor), A. Engler's Syllabus der Pflanzen- in H. Mel- chior (editor), A. Engler's Syllabus de radi s “ ebrüder ае Moni, S. A. & J. А. KALLUNKI. 1977. A revision of бе d Phyllonoma (Grossulariaceae). Brittonia : 6 ee ^ Ta B. A. Boum «€ R. ORNDUFF. 1985. Flavonoids and т of Cephalotaceae. Biochem. Syst. Ecol. 13 -263. OrMsrEAD, R. С HT (nd 15, K. M. Scorr & J. D. PALMER. 1992. Monophyly of the Asteridae and identification of their major lineages inferred from А DNA sequences of rbcL. Ann. Missouri Bot. Gard. 79: 249-265. , B. BREMER, К. M. Scorr & J. D. PALMER. teridae sensu nces. Ann. Missouri Bot. 1993. А parsimony analysis of the As lato based on rbcL seque Gard. 80: 70 )0- 722. PuiLiPSON, W. В. 1967. Griselinia Forst. fil. — Anom- aly or link. New Zealand J. Bot. 5: 134-165. | Ovular morphology and the major clas- sification of the dicotyledons. J. Linn. Soc., Bot. 68 Ovular pr т: the classifi- cations of dicotyledons. Pp. 1 O in К. Kubitzki (editor), Flowering vec uber and Classifi- cation of Higher Categories. Springer-Verlag, Vien- na, New York. PLouviER, V. 1965. Etudes chimiotaxinomiques sur les Oen Bull. Soc. Bot. France 1965: 150- РкАКАЅН, N. & E. J. MCALISTER. dle m embryo- logical study of Bauera capitat mments on the nibus position of па praa J. Bot. 22. RAVEN, P. H. 1975. The bases of е phylog: РА Cytology. Ann. Missouri Bot. Gard. 62: 724- 76 SCHOLZ, " Melianthaceae. Pp. 285-286 in H. Melchior (editor), A. Engler's Syllabus der Pflanzen- familien. Gebrüder Borntraeger, Berlin 6 Menz, С. К. 64. Saxifragaceae. Pp. 201- 206 in H. Melchior (editor), A. Engler’s Syllabus der Pflanzenfamilien. Gebrüder Borntraeger, Berlin. SHARMA, V. K Morphology, floral anatomy and embryology of Parnassia nubicola Wall. Phyto- morphology 18: 193-204. SoLtIs, D. E. & B. A. Boum. 1982. eray of 0: 221- Penthorum sedoides. Biochem. Syst. Ecol. 224. 2 IRGAN, A. GRABLE, P. S. SoLTIS & R. KUZOFF. al systematics » Saxifragaceae sensu stricto. Amer. J. ess. , P. S. SoLTIS, M. T. ue & M. DunBIN. 1990. rbc L sequence divergence and phylogenetic relation- ships in Saxifragaceae sensu lato. Proc. Natl. Acad. U.S.A. 87: 4640-4644. ; ‚ T. G. COLLIER & М. L. EDGERTON. А Chloroplast DNA variation within and among genera of the Heuchera group (Saxifragaceae): Ev- җа for pA transfer and paraphyly. Amer. . 78: 1091 са. S. А. e genera of Saxifragaceae in the southeastern United States. J. Arnold Arbor. 53: 409-498. 15 Using Parümony, Versio distributed by the Champaign, oe TAKHTAJAN, A. 1969. Flowering Plants—Origin and Dispersal. Ж на ен Washington, Outline of the elessihontion of flowering plants (Magnoliophyta) Bot. Rev. 46: 225-359. 1987. m of oo Academy of Sciences U.S SR. Lenin THORNE, R. F. a ut a putatively phylo- genetic classification of the flowering plants. Aliso 6: 51-60. ET WE inan ма 1 3.0s. Comp rogram Illinois Natural History Survey, A phylogenetic mr d of the Angiospermae. Evol. Biol. 9: 35- 1983. Proposed new E кана ME in the an- giosperms. Nordic J. Bot. 3: 85-117. 1992. An updated phylogenetic classification of the flowering pn Aliso 13: -389. VocEL, E. F. DE 1¢ eedlings of ПЕРЕ ter for Agricultural сам and шка: Wageningen. XIANG, Q., D. E. Sorris, D. R. MORGAN & P. S. Sc 1993. рине relationships of Cornus Ls sensu lato and putativ ves inferred from rbcL se- quence data. pis Missouri Bot. Gard. 80: 723- 735. PHYLOGENETIC RELATIONSHIPS OF THE GERANIACEAE AND GERANIALES FROM rbcL SEQUENCE COMPARISONS' Robert A. Price? and Jeffrey D. Palmer?” ABSTRACT Parsimony analyses of DNA sequence data from the chloroplast rbcL gene were used to assess the circumscription and phylogenetic position of the Geraniaceae, generic жез aen within the family, and the affinities of the families previously assigned to the order Geraniales. The oth classifications fall into four well separated Dens Only Hypseocharis, often included in the ер receives strong s Geraniaceae sens. str., as suggested by Boesewinkel o provide the first highly resolved phylogeny of the eraniaceae, indicating that Pelargonium is a sister r families placed in the Geraniales in recent higher-order e basis of seed anatomy. Sequenc ed of the family, that Erodium and Geranium are probable sister genera, and that Мопзота and Sarcoc adn e either sister genera or are congeneric. The Geraniaceae sens. str., comprising the five genera Erodium, Geranium, Monsonia, Sarco- caulon, and Pelargonium, are of particular evo- lutionary interest in that they exhibit the greatest amount of variation in size and structure of the chloroplast genome known for any family of pho- tosynthetic flowering plants. Substantial increase in the size of the inverted repeat region in at least some taxa of Pelargonium is exemplified by the largest known angiosperm chloroplast genome (217 kb in the cultivated geranium, P. Xhortorum), whereas the loss of one copy of the inverted repeat has resulted in an unusually small chloroplast ge- nome of ca. 120 kb in Erodium and Sarcocaulon (Palmer et al., 1987; 1990). Other unusual features seen in the family include multiple Price et al., inversions, losses of highly conserved introns, and particularly rapid rates of sequence change in some chloroplast genes (Downie & Palmer, 1992; Down- ie et al., 1991; Calie et al., unpublished data). То examine these and other interesting features of the Geraniaceae in an evolutionary context, it is necessary to obtain an independent assessment of the phylogenetic relationships of the family and its close relatives. It has proved difficult to assess generic relationships in the family using morpho- logical or cytological characters because of the paucity of informative characters and the consid- erable variability within genera (see discussions in Vorster, 1990). Thus we have used sequence com- parisons of the chloroplast gene rbcL, which has come into wide use in systematic studies of the flowering plants (see discussions in Olmstead et al., 1992; Chase et al., 1993), to attempt to elucidate phylogenetic relationships within the Geraniaceae and to determine which taxa are the most appro- priate outgroups for comparison to the family. ereas a general consensus exists that the five genera of the Geraniaceae sens. str. are closely related, there has been considerable disagreement as to whether to include several segregate families in the Geraniaceae. The monogeneric Eurasian family Biebersteiniaceae, the monotypic family Di- rachmaceae (endemic to the island of Socotra), and the small South American families Ledocarpaceae, Rhynchothecaceae, and Vivianiaceae have been included in the Geraniaceae sens. lat. by Cronquist (1981, 1988) and Thorne (1992), following the classical treatment of Knuth (1912), but excluded from the family by Dahlgren (1989), Robertson ' We thank M. Chase, D. Soltis, E. Conti, and J. Nugent for providing uu sequences, С. Zurawski for providing rbcL sequencing primers, J. Thomas for performing „з n the man Chase, M. Denton, and K. Robertson for comments o for field collections of South American plants Menden to o ns of the uencing, and Anderson, es cript. We are beum to P. Peñailillo and S. Beck r analysis, fras to all others who helped provide vem materials. This research has been supported by NSF sees BSR-8996 Bloom ? Department of Biology, Indiana Universit ton, Indiana po m ty, SUA. * Current address: Department of Botany, (еа of Georgia, Athens, Рен 30602, U.S ANN. Missouni Bor. GARD. 80: 661-671. 1993. Annals of the Missouri Botanical Garden (1972), Takhtajan (1987), and van der Walt & Vorster (1988). Each of these segregate families differs from the Geraniaceae sens. str. by lacking one or several of its diagnostic combination of gy- noecial and fruit characters (an ovary consisting of five two-ovulate locules united to an elongated stylar column, developing into a schizocarpic fruit of five one-seeded units separating from the per- sistent column; see Robertson, 1972) but needs to be considered as a potential close relative of the Geraniaceae. In addition, Boesewinkel (1988) has provided evidence from seed anatomy that Hyp- seocharis, an Andean genus often placed in the Oxalidaceae, may in fact be the closest relative of the Geraniaceae sens. str., and Thorne (1992) has included this genus in his subfamily Geranioideae. Families included in the order Geraniales in the recent higher-order classifications of Cronquist (1981, 1988), Dahlgren (1989), Takhtajan (1987), and Thorne (1992) are compared in Table 1. Each of these classifications includes within the order the segregate families noted above, and all except Dahl- gren (1989) include the Oxalidaceae and its seg- regate families Hypseocharitaceae and Lepidobo- tryaceae. Cronquist (1981, 1988) and Thorne (1992) also included the Balsaminaceae and the glucosinolate-producing families Limnanthaceae and Tropaeolaceae in the Geraniales, whereas Dahlgren (1989) placed the latter two families near the other glucosinolate taxa. Dahlgren (1989) also included the Zygophyllaceae and several small families seg- regated from it (Balanitaceae, Nitrariaceae, and Peganaceae) within the Geraniales, whereas these taxa have been otherwise assigned to the Sapindales (Cronquist, 1988), Rutales (Takhtajan, 1987), or Linales (Thorne, 1992). Older classifications by the same authors circumscribed the Geraniales still more broadly, e.g., Dahlgren (1980) and Takhtajan (1980) also included the Linaceae and the puta- tively closely related Erythroxylaceae and Humiri- aceae in the order, and Thorne (1983) included the Malphigiaceae and putative relatives as well. As one might expect from the diversity of classi- fications discussed here, comparison of diagnostic features of the various families purported to be closely related to the Geraniaceae does not indicate strong support from unusual shared derived fea- tures of morphology or secondary product chem- istry for any of the circumscriptions of the order (see Cronquist, 1981). The main features linking the families of Geraniales in the system of Cronquist appear to be a tendency to have 10-15 stamens and 5-parted gynoecia, neither of which is unique to the group. We have thus attempted to evaluate the relationships of these families by comparison of rbcL sequences from a number of different lin- eages of the subclasses Rosidae and Dilleniidae. MATERIALS AND METHODS Forty-eight taxa utilized in this study for rbcL sequence comparisons and their sources are listed in Table 2. Total genomic DNA was extracted from 1–3 grams of fresh leaf material or an equivalent fresh weight of silica gel-dried leaves using the modified CTAB procedure of Doyle & Doyle (1987). Leaves resistant to grinding (e.g., Pelargonium species) were first frozen in liquid nitrogen and then ground prior to extraction. All DNA samples were further purified and concentrated by CsCl-ethidium bromide gradient centrifugation. The complete rbcL gene was amplified via the polymerase chain re- action (PCR) using a 5' primer consisting of the rst 26 bases of the gene and a 3' primer based on a 24-base portion of a stem-loop region ca. 100 bases beyond the gene (Olmstead et al., 1992). Five taxa in the study (Geranium grandiflorum, Limnanthes douglasii, Monsonia emarginata, Oxalis dillenii, Tropaeolum majus) were initially sequenced from DNA cloned from PCR products using the strategy of Olmstead et al. (1992) and then rechecked by direct sequencing of double- stranded PCR products, which was used for all other taxa. DNA from PCR was purified by cen- trifugation in Centricon 100 filters (Amicon) for use in direct sequencing. Double-stranded DNA samples were denatured in a 95?C heat block and plunged immediately into an ice-water slurry to prevent renaturation. Sequencing was performed using the dideoxy chain termination method, uti- lizing primers supplied by G. Zurawksi. Six forward and six reverse primers were used for each species, and approximately 65-80% of the sequence was read from both strands. A total of 1,402 bp of sequence was used for phylogenetic comparisons; this excluded the area of the 5' amplification primer and any extensions of the gene beyond the length of 1,428 bp found in most seed plants. Alignment of the sequences was accomplished by comparison of the sequence of Nicotiana tabacum. Parsimony analyses were conducted using PAUP version 3.0q (Swofford, 1991) on Macintosh computers. To identify min- imum length trees we used either the heuristic search and MULPARS options with 50 replicate random orders of taxon entry, TBR branch swap- ping, and steepest descent, or, in the case of the smaller analysis of Geraniaceae, the branch and bound algorithm. Bootstrap analyses (Felsenstein, 1985) and decay analyses (Bremer, 1988) were Volume 80, Number 3 Price & Palmer 663 1993 Phylogenetic Relationships of the Geraniales TABLE 1. Families included in the Geraniales in recent higher-order classifications of Cronquist (1981, 1988), Takhtajan (1987), and Dahlgren (1989). The classification of Cronquist except in that the segregate families Biebersteiniaceae, Dirachm 1992) is the same as that of ae, Ledocarpaceae, and the order in Thorne ( Vivianiaceae are treated formally as subfamilies of the Geraniaceae, and Hypseocharis is pam in the пех Geranioideae of the Geraniaceae rather than the Oxalidaceae. The ophyllaceae and its segregate families Balanitac itrariace ae, and Peganaceae are assigned to the Sapindales ui Cronquist, to the Rutales by Takhtajan, and to es Linales by Thorne. Cronquist Takhtajan Dahlgren Geraniaceae Geraniaceae Geraniaceae Geraniaceae Biebersteiniaceae Biebersteiniaceae Geraniaceae Dirachmaceae Dirachmaceae Geraniaceae Ledocarpaceae Ledocarpaceae Geraniaceae hynchothecaceae edocarpaceae Geraniaceae Vivianiaceae Vivianiaceae Oxalidaceae Oxalida Oxalidaceae H s e Oxalidaceae Lepidobotryaceae Balsaminaceae (in Balsaminales) Limnanthaceae (in Limnanthales) Tropaeolaceae (in Tropaeolales) (in Linales) (in Linales) (in Linales Gn Balsaminales) Zygophyllaceae performed using PAUP 3.0q to examine the degree of support for given branches of the cladograms. The number of additional steps required to shift particular taxa into a monophyletic group with the Geraniaceae was examined using the CON- STRAINTS option of PAUP. RESULTS Sequences of rbcL were compared for 48 taxa (Table 2), including nine from published sources, 17 unpublished sequences assembled for the global analyses of Chase et al. (1993, provided by M. Chase, E. Conti, D. Soltis, and J. Nugent), and 22 sequences generated in this study, including 14 from the Geraniaceae sens. str. All sequences gen- erated in this study have been deposited in the GenBank. conducted, the first focusing on the familial and Two sets of parsimony analyses were ordinal relationships of the Geraniaceae and raniales, and the second focusing on generic re- lationships within the Geraniaceae sens. str. For the ordinal level comparisons, 39 taxa were chosen to represent each of the families of the Geraniales sensu Cronquist and a variety of other families from the Rosidae and Dilleniidae. These included several groups (e.g., families of Linales, Sapindales, and Capparales) often considered to be related to the Geraniales and representatives of the families grouping with families of the Geraniales in analyses of the broader data set of rbcL sequences assem- bied 2 Chase et al. (1993). Tetracentron (Tetra- ae) and Gunnera (Gunneraceae) were cho- (1993) data set as outgroups from near the base of the rosid—dilleniid lineage. The sam when Platanus (Platanaceae) was included as a sen n the Chase et al. e ingroup topology was obtained third outgroup. arsimony analysis with equal character weight- ing yielded a single minimum length tree of 1,60 steps (Fig. 1) with a consistency index (CI) of 0.38 excluding unique substitutions and a retention in- dex (RI) of 0.47. Geraniales by Cronquist (1988) are placed in five (1) Oxalis and Av- errhoa (Oxalidaceae) with Eucryphiaceae and The families included in the separate clades on the tree: Cephalotaceae and then linking to a larger group including the Linaceae, (2) Tropaeolum and Lim- nanthes in a mustard-oil containing lineage with Brassica and then linking to the Malvales and Sapindales, (3) Wendtia (Ledocarpaceae) and Ni- viania (Vivianiaceae) grouping together and then linking to the Greyiaceae and Francoaceae, (4) the Geraniaceae sens. str. grouping with /7ypseochar- is, and (5) Impatiens (Balsaminaceae) in a mor- phologically rather disparate group near the base of the tree with Fouquieriaceae and Polemoniaceae. Support for the placement of the putatively gerani- 664 Annals of the Missouri Botanical Garden TABLE 2. Sources of rbcL sequences. Taxa are arranged in the classification of Cronquist (1981) except in the recognition of the segregate families Francoaceae, Iteaceae, Parnassiaceae, Hypseocharitaceae, Ledocarpaceae, and Vivianiaceae following Takhtajan (1987). Sequences generated in this udy are marked with an asterisk. All voucher specimens have been deposited at IND, except for Chase 113 (NCU) and Peñailillo 91000 and 91003 (CONC). Family Species Source/ voucher Asteridae Solanales Polemoniaceae Polemonium reptans Olmstead et al. (1992) Dilleniidae Brassicaceae Brassica juncea Nugent & Palmer, unpublished Malvales Malvaceae Gossypium hirtum Gulov et al. (1990) Sterculiaceae Theobroma cacao* Price s.n. Violales Fouquieriaceae Fourquieria splendens Olmstead et al. (1992) Violaceae Viola sororia Olmstead et al. (1992) Hamamelidae Trochodendrales etracentraceae Tetracentron sp. Chase et al., unpublished Rosidae Cornales Nyssaceae Nyssa sp. Soltis et al., unpublished Geraniales Balsaminaceae Impatiens biflora Chase et al., unpublished Geraniaceae rodium texanum* Van Devender s.n. E. x variabile* Price s.n. Geranium cinereum Price s.n. G. grandiflorum* Price s.n. G. macrorrhizum* Price s.n. G. maderense* Price s.n. G. ocellatum Price s.n. G. palmatum* Price s.n. Monsonia emarginata* Price s.n. Pelargonium capitatum* Price s.n. P. cotyledonis* Price s.n. P. existipulatum* Price s.n. P. x hortorum* Price s.n. Sarcocaulon vanderietiae* Price s.n. Hypseocharitaceae Hypseocharis sp.* Beck s.n. Ledocarpaceae Wendtia gra Penailillo 91003 Limnanthaceae Limnanthes douglasii* Price s.n. Oxalidaceae Oxalis dillenii* Price s.n. Averrhoa carambola* Price s.n. Tropaeolaceae Tropaeolum majus* Chase 113 Vivianiaceae Viviania marifolia* Penailillo 91000 Haloragales Gunneraceae Gunnera sp. Morgan & Soltis (1993) Linales Erythroxylaceae Erythroxylum sp. Chase et al., unpublished Linaceae Reinwardtia indica Chase et al., unpublished Myrtales Lythraceae Lythrum subsessilifolium Conti et al. (1993) Onagraceae Clarkia xantiana Conti et al. (1993) Volume 80, Number 3 Price & Palme 665 1993 Phylogenetic РЕНИ of the Geraniales TABLE 2. Continued. Family Species Source/ voucher Polygalales Malpighiaceae Thryallis sp. Chase et al., unpublished Vochysiaceae Qualea sp. Chase et al., unpublished Rosales Cephalotaceae Cephalotus sp. Morgan & Soltis (1993) Crossosomataceae Crossosoma sp Morgan & Soltis (1993) cryphiaceae ucryphia sp Morgan & Soltis ~ Francoaceae Francoa sonchifolia Soltis et al. (1990 Greyiace eyia s Morgan & Soltis ж Hydrangeaceae Carpenteria californica Soltis et al. (1990) ae Itea virginica Soltis et al. (1990) Parnassiaceae Parnassia fimbriata Soltis et al. (1990) Saxifragaceae Heuchera micrantha Soltis et al. (1990) Sapindales Aceraceae Acer sp. Chase et al., unpublished Simaroubaceae Ailanthus altissima Chase et al., unpublished alean taxa in these five groups is pone strong as indicated by bootstrap values o decay values of four or more steps to collapse the first interior node (Fig. 1). A series of constraint experiments was used to examine the number of additional steps required to force given taxa into a monophyletic group with- in the Geraniaceae sens. str. plus Hypseocharis. A tree of 74 more steps was needed to obtain a monophyletic group consisting exclusively of taxa from the Geraniales sensu Cronquist. The largest contribution to this increase in length came from forcing the two glucosinolate taxa Limnanthes and Tropaeolum into the Geraniales, as 36 fewer steps were required when they were no longer included in the constrained group. Conversely, 37 steps beyond the minimum were required to force these two taxa alone into a clade with the Geraniaceae. Twenty steps above the minimum were required to force the two genera of Oxalidaceae into a mono- phyletic group with the Geraniaceae, while only four steps above the minimum were required to produce a clade of Viviania, Wendtia, and the Geraniaceae. The Geraniaceae sens. str. are well supported as a monophyletic group in our analyses, with boot- strap values of 99-100% both in Figure 1 and in additional analyses in which all of our sequences for the family were compared to ten or more out- group taxa. Outside of Hypseocharis, however, which is linked strongly to the Geraniaceae sens. str. by a branch of 25 steps and a bootstrap value of 100%, there is only weak support for the as- sociation of any specific taxa with the family. Cros- sosoma (Crossosomataceae) is placed as the next closest outgroup to the family in the minimum length tree, but only one additional step is required to collapse this node, and the associated bootstrap value in this analysis is only 26%. An alternative position for Crossosoma at one step more than the minimum is as an isolated taxon forming the first node interior to Heuchera and /tea near the base of the rosid—dilleniid group. Based upon the results of the preceding higher- level analysis, a more detailed analysis was under- taken to examine generic relationships in the Ge- raniaceae using Hypseocharis as the outgroup. А single minimum length tree of 330 steps with a CI of 0.68 excluding unique substitutions and an RI of 0.79 was obtained from parsimony with equal weighting (Fig. 2). As in the broader analysis, Pel- argonium forms a sister group to the remainder of the family. Three of the genera of the family (Pelargonium, Geranium, and Erodium) appear to be monophyletic on the basis of the rbcL data, while Monsonia and Sarcocaulon were represent- ed by only single species in this analysis. Geranium and Erodium appear as sister genera to one another and then to Monsonia plus Sarcocaulon. The latter two taxa are supported as a monophyletic group by 100% bootstrap values, in agreement with their close morphological relationship. With the exception of the closely related species pair Geranium maderense and G. palmatum, each of the taxa from the large genera Geranium and Pelargonium included in Figure 2 represents a different section. In Geranium, the three sections assigned to subgenus Robertium by Yeo (1984), 666 Annals of the Missouri Botanical Garden 45 Thryallis ва | 2 168 моја „Г 4 Reinwardtia : Erythroxylum 79 Parnassia prg 85 Eucryphia 9e |2 Cephalotus >5 |84 Averrhoa • P Oxalis e 95 Clarkia 14 = » Lythrum M Qualea = Gossypium E Theobroma Limnanthes e " 79 Brassica v 3 Tropaeolum e Бој || 100 | Асег 5 >5 Ailanthus 100 — Greyia 79 | >$ Francoa 4 100 Wendtia e >5 Viviania e 65 Crossosoma A Hypseocharis e Pelargonium e Erodium e Geranium e 53 Monsonia e ey >5 *Sarcocaulon e 91 Itea >5 Heuchera Carpenteria um Nyssa Polemonium Fouquieria Impatiens e Gunnera 0 10 20 30 Tetracentron HH YH site changes Volume 80, Number 3 Price & Palmer 667 1993 Phylogenetic Relationships of the Geraniales Hypseocharis 100 г Geranium palmatum 25 L G. maderense G. ocellatum G. macrorrhizum 3 G. cinereum 95 : 5 G. grandiflorum 5 100 e Erodium variabile >5 2 E. texanum 100 m~~ Monsonia emarginata >5 L Sarcocaulon vanderietiae 100 PA Pelargonium capitatum P. existipulatum 100 >> ES Р 4 P. cotyledonis >5 P. hortorum 0 10 20 | L | Г I 1 site changes FIGURE 2. of rbcL sequences (length — 330 steps, RI — branches and decay values are indicated below them. represented by G. ocellatum, G. maderense plus G. palmatum, and G. macrorrhizum, are placed together in a monophyletic group, whereas the individual representatives of subgenus Geranium (С. grandiflorum) and subgenus Етодгогаеае (G. cinereum) group together. The numbers of site changes and associated bootstrap values seen in comparisons within Erodium, Geranium, and Pel- argonium are often high and suggest that sub- stantial resolution among other infrageneric groups may be obtained using the rbcL gene. One result of particular interest is the unequivocal placement of Pelargonium cotyledonis, unusual for the genus in its radially symmetrical flowers and reduced nectar spur, in a position nested within the genus. The single minimum length Fitch parsimony tree obtained for taxa of Geraniaceae based on comparison 0.79). Bootstrap values out of 250 replicates are indicated above the DISCUSSION EVALUATION OF SUPPORT FOR BRANCH PATTERNS In Figures 1 and 2, we compare bootstrap val- ues, decay values, and branch lengths for the var- ious branches of our cladogram. The bootstrap values and decay values are positively correlated, but the correspondence between the two is not monotonic. As an example, the range of bootstrap values corresponding to particular decay values in Figure | are as follows: one added step (11-42% mean = 23%), two added steps (45-85%, mean = 58%), three added steps (53-79%, mean = 66%), four added steps (66-89%, mean = 79%), five added steps (80-87%, mean = 84%), > буе — FIGURE 1. The single minimum length Fitch parsimony tree obtained n Ps im of families of Geraniales - possible relatives using rbcL sequence data (length — 1,609 steps, RI — Taxa assigned to the Geraniales n the classification of Cu (1981) are indicated by dots. Bootstrap эм. out КА 250 replicates are shown above the branches, and decay values (numbers of additional steps needed for a branch to collapse) a iven below the branches. The taxa of укинуте included in this analysis are Erodium Xvariable, Geranium grandiflorum, Monsonia emarginata, Pelargonium X hortorum, and Sarcocaulon vanderietiae. 668 Annals of the Missouri Botanical Garden added steps (79-100%, mean = 97%). The boot- strap and decay values both appear to provide reasonable indications of which branches on the tree are poorly supported and which are strongly supported, whereas the interpretation of interme- diate values is likely to be ambiguous in either case. Displaying both bootstrap and decay values is in- structive, since they indicate different aspects of the support for the nodes in the tree and thus tend to reinforce one another. The large amount of computational time required to determine all trees of given lengths more than five steps above the minimum length currently imposes a limit to the precision obtainable with decay analysis. Compu- tational constraints also significantly limit the num- ber of taxa that can be readily evaluated with either decay or bootstrap analysis, leaving only branch lengths and detailed examination of the character distribution to evaluate the support for particular branches when large numbers of taxa are com- pared. Branch lengths on a given minimum length tree may or may not correspond well to bootstrap values and decay values, presumably depending on the frequency of homoplasy for the characters supporting and internal to a given branch. For example, branch lengths of nine to ten characters in Figure 1 have bootstrap values ranging from 25 to 100% and decay values ranging from one to five, whereas the bootstrap and decay values correspond much more closely to one another. RELATIONSHIPS OF THE GERANIALES Our results do not provide any substantial sup- port for close relationships of any taxa other than Hypseocharis to the Geraniaceae sens. str. The Geraniaceae also group with Crossosoma and then with Greyia, Francoa, and Viviania in one of the global analyses of Chase et al. (1993) and in some of the analyses of Morgan & Soltis (1993), but our results would indicate that the character sup- port for the relationships of these groups to the Geraniaceae is weak. While the Geraniaceae are clearly placed in the core group of Rosidae in our analysis (and those of Chase et al., 1993, and Morgan & Soltis, 1993), additional analyses with more of the segregate taxa included and with a larger set of molecular characters will probably be necessary to assess rigorously the precise position of the family within the subclass. The placement of Limnanthes and Tropaeolum in a clade of glucosinolate-producing taxa is a con- sistent feature of rbcL sequence comparisons (see particularly Rodman et al., 1993, which includes almost all of the glucosinolate families). Most recent higher-order classifications have placed the Lim- nanthaceae and Tropaeolaceae in the Geraniales or adjacent orders (see Table 1). However, the occurrence of erucic acid (a compound character- istic of the core capparalean group of glucosinolate taxa) in the Limnanthaceae and Tropaeolaceae seems suggestive of a close relationship between these families and the other glucosinolate taxa, as suggested by Dahlgren (1975). A cladistic analysis of morphological characters by Rodman (1991) placed the Tropaeolaceae within the glucosinolate lineage in a position similar to its placement in the rbcL sequence comparisons of Rodman et al. (1993), whereas the morphological trees placed the Limnanthaceae between the Geraniaceae plus Ox- alidaceae and the glucosinolate taxa in association 1992) has compared the Limnanthaceae to the Geraniales for a series with Balsaminaceae. Link ( of floral and embryological characters and con- cluded that there was no strong evidence favoring a close relationship between them. Thus, because of the high degree of internal consistency of the rbcL data (as indicated by the bootstrap values of 94% and 100% in Fig. 1) we favor the inclusion of the Limnanthaceae and Tropaeolaceae within the clade of glucosinolate taxa. The apparently arbitrary nature of the delimi- tation of the subclasses Dilleniidae and Rosidae is emphasized by the placement of the Tropaeolaceae and Limnanthaceae (from Cronquist's Rosidae) in a group of glucosinolate taxa largely assigned to the Dilleniidae. As noted by Olmstead et al. (1992) and Chase et al. (1993), the Dilleniidae appear to be a polyphyletic group arising from several sep- arate lineages of Rosidae. As shown in Figure 1, and also Chase et al. (1993) and Rodman et al. (1993), the closest relatives of the glucosinolate lineage appear to be the Malvales (usually assigned to the Dilleniidae) and the Sapindales (usually as- signed to the Rosidae). The position of /mpatiens (Balsaminaceae) out- side of the main lineage of Rosidae and far removed from any other putative groups of Geraniales is unexpected, but appears to be robust to the choice of outgroup and ingroup taxa in the analysis. In the broader rbcL analysis of Chase et al. (1993) the Balsaminaceae are placed in a group of rather disparate taxa primarily composed of Ericaceae and other families from the Dilleniidae of Cronquist, but also including the Polemoniaceae (generally assigned to the Asteridae). As discussed by Cron- quist (1981), the morphological evidence linking the Balsaminaceae to the Geraniales is weak. The Balsaminaceae differ from the Geraniaceae in em- bryological features, having tenuinucellate ovules Volume 80, Number 3 1993 Price & Palm Phylogenetic ыйыы of the Geraniales and cellular endosperm (which are, however, found in the Ericales) rather than crassinucellate ovules and nuclear endosper The separation of Ha Oxalidaceae from the Ge- raniaceae in the rbcL analysis is contrary to the results of the cladistic analysis of Rodman (1991) and to the recent higher-order classifications. of Cronquist (1988), Takhtajan (1987), and Thorne (1992). Our results are in better accord with the groupings of Dahlgren (1989), who placed the Ox- alidaceae in the order Linales. The support for the exclusion of the Oxalidaceae from the Geraniales in the rbcL analysis is strong, since 20 additional steps are required to force the two families into an adjacent position on the tree, whereas the mor- phological evidence linking the two families is rel- atively wea e two families are similar in gen- erally having 10 to 15 stamens with basally connate filaments, but differ in that the Oxalidaceae have separate styles, abundant endosperm, tenuinucel- late ovules, and capsular fruits, whereas the Ge- raniaceae have fused styles, scant or absent en- dosperm, crassinucellate ovules, and a specialized type of schizocarpous fruits (see Cronquist, 1981). Boesewinkel (1988) demonstrated that Hypseo- charis, while often assigned to the Oxalidaceae, has crassinucellate ovules and scant endosperm along with fused styles and is also similar to the Geraniaceae in details of seed wall anatomy not seen in the Oxalidaceae. Hypseocharis is also much more similar to the Geraniaceae than the Oxali- daceae in its floral vasculature and has a staminal arrangement similar to that of Monsonia and Sar- cocaulon (Rama Devi, 1991). Thus, Hypseocharis appears to be most similar to the Geraniaceae sens. str. in morphology, in exact agreement with the results of our analysis, rather than being inter- mediate between Oxalidaceae and Geraniaceae. The features shared by Hypseocharis and Oxalidaceae (e.g., capsular fruits and the absence of a stylar column) can readily be explained as shared prim- itive features found in a wide range of rosid taxa. Our results are consistent with the traditional place- ment of the woody genus Averrhoa (the cultivated starfruit) along with Oxalis in the Oxalidaceae. The longer branch length leading to Oxalis on the clado- gram in Figure | (44 vs. 13 steps) is striking and would seem to imply a faster rate of sequence change in the shorter-lived herbaceous genus. Our results provide support for the exclusion of the segregate families Ledocarpaceae and Vivi- aniaceae from the Geraniaceae, since they are linked robustly to Greyiaceae and Francoaceae instead. Study of additional members of Vivianiaceae and Ledocarpaceae will be necessary before evaluating whether both of these families and the other South American segregate family Rhynchothecaceae should be treated as three separate families. Pre- liminary data from seed anatomy reported by Boe- sewinkel (1988) indicate that Гплата and also Dirachma (Dirachmaceae) are not similar to Нур- seocharis and the Geraniaceae sens. str. Airy Shaw ~ 1966) earlier suggested that the Socotran genus Dirachma is related to the South African family Greyiaceae, presumably on the basis of similar fruit morphology. n contrast, Biebersteinia (the sole genus of the Eurasian family Biebersteiniaceae) appears to be relatively similar to the Geraniaceae in both seed and fruit morphology (Boesewinkel, 1988) and is clearly in need of further evaluation from both morphological and molecular standpoints to assess its relationship to the Geraniaceae. Our results, taken together with those of Boesewinkel. suggest that Hypseocharis may appropriately be included in the Geraniaceae as a monogeneric subfamily, while Biebersteinia could represent a third sub- family. Further sampling will be needed to assess rigorously the delimitation of the Geraniaceae and Geraniales, but a conservative approach would be to include only Hypseocharis, Biebersteinia, and the Geraniaceae sens. str. within the order. RELATIONSHIPS WITHIN THE GERANIACEAE Our results agree with morphological data in indicating that the Geraniaceae sens. str. is a well defined monophyletic group. The group is also sup- ported as monophyletic by the loss of the intron in the plastid gene rp/16 (Downie & Palmer, 1992), a character that has not yet been examined in Hypseocharis. Within the family, our study pro- vides the first available phylogenetic framework. Only the close relationship between Monsonia and Sarcocaulon has been strongly supported by mor- phological data. These genera have extremely sim- ilar flowers that differ from the rest of the Gera- niaceae sens. str. (but resemble Hypseocharis) in Mon- sonia and Sarcocaulon also share the loss of the having 15 stamens rather than 10 or fewer. otherwise 1991). T Гћезе highly conserved rp/2 intron, seen throughout the family (Downie et al., two genera may appropriately be treated as con- generic, since they differ primarily in that Sar- cocaulon is a stem succulent and generally has spinescent persistent leaf bases, whereas Wonsonia is neither succulent nor spiny. Our results do not agree with the suggestion of Rama Devi (1991), on the basis of similarity in staminal arrangement and vasculature, that Monsonia and Sarcocaulon 670 Annals of the Missouri Botanical Garden are more closely related to Hypseocharis than to other genera of Geraniaceae. e basal split between Pelargonium and the rest of the Geraniaceae sens. str., which is well supported in our analyses, has not been predicted based on morphology but is consistent with the much greater level of morphological diversification within Pelargonium than in the rest of the family. Pelargonium is distinct from the rest of the family in having a nectar spur at the base of the flower (a derived character based on outgroup comparison to Hypseocharis) and generally has bilaterally rather than radially symmetrical flowers (a derived feature otherwise observed only in some members of the genus Erodium). One of the two species of Pelargonium with radially symmetrical flowers (P. cotyledonis) appears to be nested within the genus on the basis of our rbcL comparisons, suggesting that radial symmetry here represents a reversal from bilateral symmetry. The substantial number of base substitutions observed among representa- tives of four of the sections of Pelargonium augurs well for efforts to use sequence information from the rbcL gene to examine relationships among the 14 currently recognized sections of the genus (see van der Walt & Vorster, 1988; Albers et al., 1992) and to assign to section those species of uncertain affinity. Comparisons of the gene among species from the majority of the sections are currently in progress. The sister-group relationship between Geranium and Erodium indicated by our results is neither strongly supported nor contradicted by information from morphology and cytology (cf. Yeo, Geranium has a relatively unspecialized floral mor- phology, but exhibits specialized seed-discharge mechanisms in subgenus Geranium and subgenus Robertium, whereas Erodium often exhibits floral zygomorphy but has the unspecialized ** Érodium- type" of seed discharge (Yeo, 1984). These two genera do have in common a primarily Northern Hemisphere distribution, whereas Monsonia, Sar- cocaulon, and Pelargonium have distributions centered in southern Africa (van der Walt & 1988; Тео, 1 The phylogenetic famen that we have ob- tained from rbcL comparisons allows us to begin to examine the likely history of structural changes ster, in the chloroplast genome. Our results suggest that there have been separate losses of a copy of the inverted repeat region in the chloroplast genome in Erodium and Sarcocaulon and probably sepa- rate expansions of the inverted repeat in Geranium and Pelargonium (see Price et al., 1990). Ex- amination of the plastid genome in Hypseocharis will be important in indicating the primitive con- dition in the family, while comparative studies in the other genera may indicate features of the ge- nome that are associated with these structural changes. LITERATURE CITED AIRY ~ H. K. 1966. A Dictionary of the Flowering Plan dos Ferns, 7th ed. Cambridge Univ. Press, ambridge. ALBERS, F. M. Савву & M. AUSTMANN. 1992. A re- appraisal of Pelargonium sect. Ligularia (Gerania- . Pl. Syst. Evol. 179: 257-276. белл. F. D. The seed structure and taxonomic relationships of Hypseocharis Remy. Acta Bot. Neerl. 37: 111-120. BREMER, К. 1988 The limits of amino acid sequence ata in angiosperm Evo- lution 42: | 795- 803. CHasE, M. W., D. E. Sorris, R. С. OLMSTEAD, D. MORGAN, D ^c CONTI, E., A. Fiscupacu & K. J. Sytsma. 1993. Tribal relationships i in Onagraceae: Implications from rbcL sequence data. Ann. Missouri Bot. Gard. 80: 672- 685 CRONQUIST, A. . An Integrated System of Classi- fication of the Flowering Plants. Columbia Univ. Press, New York. The Evolution and Classification of Flowering Plants, 2nd ed. The New York Botanical Garden, New York. DAHLGREN, G. 1989. The last Vp dir re System of classification of the dicotyledons. Pp. 249-260 in K. Tan (editor), The Davis & Hedge ани Edinburgh Univ. Press, Edinbur DAHLGREN, R. : system of clic of the angiosperms to be used to are the distri- bution of characters. ha Not. 128: 191. 1980. A revised system of tle un ш of the angiosperms. Bot. J. Linn. Soc. 80: 91-124. Роме, S. R. € J. D. PALMER. 1992. Use of chlo- roplast DNA rearrangements in reconstructing plant phylogeny. Pp. 14-35 in P. Soltis, D. Soltis & J. Doyle (editors), к Systematics of Plants. MM & Hal . С. OLMS TEAD, C ZURAWSKI, D. E. Sorris, P. S. ита J. C. Watson & J. D. PALMER. 1991. Six independent = of a idem DNA rpl2 ns: Molecular a са миен pons LJ&LL rapid DNA isolation procedure for 2 > of fresh leaf tissue. 15. Phytochem. Bull. 19: Volume 80, Number 3 1993 Price & Palmer 671 Phylogenetic Relationships of the Geraniales m J. 1985. Confidence limits on phylogenies: An approach using the bootstrap. Evolution 39: 783 GULOV, N. UrMasov, К. А. ALI ANDRIANOV & E. S. PIRUZIAN. 1990. Nucleotide жасын of the chloroplast gene der from Gossyp- m hirsutum. Nucleic Acids Res 85. ENG. R. 1912. Mp ase In: А. Engler (editor), Das Pflanzenreich IV. 1-640. Verlag von Wil- helm Engelmann, Qs 1992. The flo oF nectaries in the Limnan- a 9-243. Morcan, D. К. . E. Sorris. 1993. Phylogenetic relationships among members of Saxifragaceae sensu lato based on rbcL n data. Ann. Missouri Bot. Gard. 80: wow HE MICHAELS, K. M. Scorr & J. D. 1992. Monophyly of the Asteridae ps identification of their major lineages inferred fro A sequences of rbcL. Ann. Missouri Bot. oe T9: p PALMER, J. D., J. M. NucENT & L. А. HERBON. 1987. Unusual structure of geranium chloroplast DNA: А repeat families. Proc. Natl. Acad. Sci. U.S.A. 84: 769-773. Price, R. A., P. J. Carte, S. R. Downie, J. M. LOGSDON, & J. D. PALMER. 1990. Chloroplast к vari- ation in the Geraniaceae — iminary report. Pp. 231-244 in P. Vorster (editor), Proceedings of the International Geraniaceae Symposium. Univ. of Stel- Africa. . Floral anatomy of Hypseocharis (Oxalidaceae) with a discussion on its systematic po- sition. Pl. Syst. Evol. 177: ROBERTSON, К. В. 1972. The genera of Geraniaceae in the Vc ME United States. J. Arnold Arbor. 53: 182- RODMAN, J. a poor A taxonomic analysis of gluco- "wu ua -producing plants, part 2: Cladistics. E Bot. =629. , a A. PRICE, К. KAROL, E. Conti, K. SYTSMA & J. PALMER. 1993. Nucleotide sequences of the rbcL gene indicate monophyly of mustard oil plants. Ann. Missouri r В 692 Sorris, D. E., P. S. Sorvris, M. T. CLEGG € M. DURBIN. 1990. rbeL sequence incu and phylogenetic relationships i in Saxifragaceae sen ( lato. Proc. Natl. Acad. Sci. U.S.A. 87: 4640-4 SworFORD, D. L. 1991. PAUP: Phylogenetic Analysis Using Parsimony, version 3.0. Computer pro chen distributed by the Illinois Natural History Surve Champaign, Illinois. TAKHTAJAN, A. Outline of the classification of V E plants (Magnoliophyta). Bot. Rev. 46: 225- 359. 161 Systema Magnoliophytorum. Officina Editoria “Nauka, " Leningrad. THORNE, R. F. 1983. | new realignments in the angiosperms. Nordic J. B 17 1992. Classification d geography of the wering plants. Bot. Rev -348. VonsreR, P. (editor). 1990. Proce of the Inter- sium. Univ. of Stellen- . J. VORSTER. 1988. Pel- саша of Southern Айз, Vol. 3. National Bo- tanical Gardens, Kirstenbosch. Yeo, P. F. 1984. Fruit- discharge type in Geranium (Geraniaceae): Its use in classification and its evo- МЕ ние implications. Bot. J. Linn. Soc. 89: 1-3 TRIBAL RELATIONSHIPS IN ONAGRACEAE: IMPLICATIONS FROM rbcL SEQUENCE РАТА! Elena Conti, Anthony Fischbach,” and Kenneth J. Sytsma? ABSTRACT The evolutionary relationships among the seven tribes of Onagraceae, the most intensively studied family of intermediate size chloroplast (cp) aud nuclear ribosomal (nr) DNA restriction sites, е аг-е were obtained fro ed tr , the rela Ко ат пае by these analyses. Cladistic ше ‘of s sequence dat results that bear on two major issues: (i) intertribal Seasons within Onagraceae, and (ii) cong een examined from morphological and molecular к Previous cladistic analyses of oded rbcS amino acid sequences, nr es that agree in defining the tribe Jussiaeeae as the ong the rest of the tribes are not completely a from the chloroplast encoded rbc all tribes except Jussiaeeae; Onagreae and Epilobieae; and, most interestingly, Fuchsieae and Circaeeae. The formation of this last clade corroborates the results of the cpDNA restriction site and of the nrRNA sequence data. The data also indicate an apparent slowdown in the rate of rbcL sequence divergence in the woody Fuchsia lineage relative to the herbaceous Circaea lineage. The topology of the rbcL tree basically confirms that of t the nrRNA se lacements of monogeneric tribes Lopezieae and he cpDNA restriction site tree. Among the other cladistic analyses, uence survey produced the tree closest to the rbcL t Hauyeae are not strongly supported. The ‚ followed in order by the rbcS amino acid sequence tree, the nrDNA restriction site tree, and the morphological tree. The Onagraceae consist of seven tribes, 15 gen- era, and 652 species; six tribes are monogeneric and one (Onagreae) includes nine genera (Hoch et al., 1993; Table 1). Onagraceae presumably orig- inated in. West Gondwanaland (the oldest fossil records are from Brazil and California) and reached their highest degree of diversification in North America (Raven & Axelrod, 1974; Raven, 1988). Raven (1988) pointed out that the pollen of On- agraceae is so distinctive it can be recognized in the fossil record from the Late Cretaceous (73-65 туа). The inclusion of Onagraceae in Myrtales is not questioned (Johnson & Briggs, 1984), but the fam- ily is somewhat isolated in the order. The Onagra- ceae are set off from other families in the order by their peculiar syndrome of morphological and embryological characters: epigynous flowers (Dahl- gren & Thorne, 1984); pollen grains with viscin threads (Patel et al., 1984; Skvarla et al., 1976) and peculiar structure of the pollen wall (“рага- crystalline beaded”” ektexine; Raven, 1988); 4-nucleate Oenothera-type embryo sac formation Tobe & Raven, 1983); abundance of calcium ox- alate raphides in vegetative cells (Carlquist, 1975); and presence of septa dividing the sporogenous tissue in the anthers (Tobe & Raven, 1986). The isolation of the family in Myrtales is emphasized by Thorne (1992), who defined Onagraceae as the sole family in suborder Onagrinae. — The Onagraceae are the most intensively studied family of intermediate size from both systematic ' For collections of leaf material, we thank Peter Raven and Peter Hoch, Missouri Botanical Garden, and James ffolter, formerly of Berkeley Botanical Garden. Jeffrey Palmer provided clones of rbcL for probing of southern transfers. Richard Olmstead gave sequences for 5' and 3' primers for ш rbcL. Gerard Zurawski kindly provided internal primers for sequencing rbcL. We thank David Baum and James Rodman for fruitful discussions on is кы б the artwork. Critical reading of earlier drafts by Peter Raven and Pete abe Alumni Research Foundation and the National Science Foundation (BSR- o K. J. Sytsma is also gratefully acknowledged. This paper represents a portion of . Con Botany Department, usu of Wisconsin, Madison, Wisconsin 53706, U.S.A. ^ Zoolory Department, University of Texas, Austin, Texas 78712, U.S.A. ANN. MISSOURI Вот. Garb. 80: 672-685. 1993. Volume 80, Number 3 1993 Conti et al. 673 Tribal Relationships in Onagraceae The seven tribes and the nine species of Onagraceae employed in this study. Information on voucher specimens is listed in the Appendix at the end of this issue. Number of ide pe species Tribe Genera per genus this stu Onagreae 1. Gongylocarpus 2 2. Gayophytum 9 3. Xylonagra 1 4. Camissonia 61 5. Calylophus 6 6. Gaura 2l 7. Oenothera 123 O. elata HBK 8. Stenosiphon 1 9. Clarkia 44 C. xantiana А. Gray Epilobieae 10. Epilobium 168 E. angustifolium L. Lopezieae 11. Lopezia 22 L. riesenbachia Plitman, Raven & Breedlove Fuchsieae 12. Fuchsia 102 F. cyrtandroides J. Moore Circaeeae 13. Circaea 7 C. alpina L auyeae 14. Hauya 2 H. elegans DC Jussiaeeae 15. Ludwigia 82 L. peruviana (L.) Hara L. peploides (HBK) Raven and evolutionary perspectives (Raven, 1988; Syts- ma & Smith, 1992). Detailed studies exist for morphology—anatomy (Eyde, 1982, on flowers; van Vliet & Baas, 1984, on xylem); embryology (Tobe & Raven, 1983); palynology (Patel et al., 1984); reproductive biology (Raven, 1979); phytochem- istry (Averett & Raven, 1984; Dahlgren & Thorne, 1984); and molecules (Bult & Zimmer, 1993; Cris- ci et al., 1990; Martin & Dowd, 1986; Sytsma & Smith, 1992; Sytsma et al., 1991b). The cladistic analyses carried out on such data have jesus cladograms based on chloroplast DNA (cpDNA) restriction sites (Sytsma et al., 199 1b), aie nea ribosomal RNA (nrRNA) sequenc- es (Bult & Zimmer, 1993), nuclear ribosomal DNA (nrDNA) restriction sites (Crisci et al., 1990), ami- no acid sequences of the nuclear-encoded small subunit of ribulose bisphosphate carboxylase (rbcS; Martin & Dowd, 1986), and morphology (Hoch et al., 1993). These trees are congruent in indicating Ludwigia (tribe Jussiaeeae) as the sister group to the rest of the family but show some disagreement in the relationships among the other tribes. Lud- wigia possesses a number of plesiomorphic fea- tures, including the absence of a floral tube beyond the ovary and the presence of up to seven parts in its floral whorls. The relationships among the rest of the tribes are not clear, except that the other four monotypic tribes (Fuchsieae, Circaeeae, Lopezieae, and Hauyeae) exhibit a greater number of plesiomorphic features than do either Epilobieae or Onagreae (nine genera). he remarkable amount of morphological and molecular information available for the Onagra- ceae, although often conflicting, will provide the framework for comparisons with sequence data o the plastid-encoded large subunit of ribulose bis- phosphate carboxylase (rbcL) described in this two goals of this phylogenetic analysis (1) to derive a detailed set of relationships among the tribes of Onagraceae using rbcL sequence data; (2) to determine congruence between the rbcL cladogram and that obtained from cpDNA restriction site mapping analysis, and subsequently from nrRNA sequence, rbcS amino acid sequence, nrDNA restriction site, and mor- phology data sets. The rbcL and cpDNA restriction site mapping data sets for the Onagraceae are mutually inclusive for taxa but completely exclu- sive for characters (1.е., no restriction enzymes mapped recognize variable sites within rbcL). The applicability of rbcL sequence information to resolving relationships within families has seldom been tested (but see Doebley et al., 1990, on Po- aceae; Soltis et al., n Saxifragaceae for studies in which intrafamilial relationships have been examined in part). A separate, detailed anal- ysis is planned to examine the global = of all data sets for the family (Sytsma et al., prep.). 674 Annals of the Missouri Botanical Garden METHODS fragments up to 500 nucleotides long. These longer Twelve species, representative of the seven tribes of Onagraceae and of three myrtalean outgroups, are considered in the present study. Table 1 lists the nine species of Onagraceae employed in this analysis. Two species of Ludwigia (L. peruviana (L.) Hara of sect. Myrtocarpus and L. peploides (HBK) Raven of sect. Oligospermum), represen- tatives of divergent lines in the genus (Eyde, 1981), were sampled because Ludwigia is universally in- dicated as the sister genus to the rest of the family. Total DNAs of all but Oenothera were extracted according to the method of Smith et al. (1991) from leaves kept at —80?C. For nine of these 12 taxa, excluding Oenothera, Clarkia, and Epilo- bium, the chloroplast rbcL gene was amplified by the polymerase chain reaction (Mullis et al., 1986; Erlich, 1989), using ‘AmpliTaq’ kits, a Perkin- Imer Cetus Thermal Cycler, a 5' primer that anneals to the first 26 base pairs of the gene and a 3’ primer that anneals to a flanking region down- stream from the 3’ end. Therefore, for these nine taxa, it was possible to obtain the entire rbcL nu- cleotide sequence with the exclusion of the first (but generally highly conserved) 26 nucleotides. The DNAs of two of the remaining three species, Epilobium angustifolium L. and Clarkia xan- tiana A. Gray, were digested with EcoRI (which cuts 10 base pairs internal to the 5’ end and 21 base pairs internal to the 3’ end of the rbcL gene) and ligated into pUC19 (later transferred into Bluescript-psk). Plasmids containing the rbcL in- sert (using purified segments of rbcL from Nico- папа as heterologous probes) were recovered from transformed colonies and purified over cesium chlo- ride gradients. For these two species, the first 10 and the last 21 base pairs are thus missing in the cloned segments. The last 21 missing base pairs were later obtained for Clarkia xantiana following PCR amplification of total DNA as described above. The rbcL sequence for Oenothera elata HBK (ex O. hookeri) published by Winter & Herrmann (1988) was utilized. Because this rbcL gene was cloned using EcoRI, the last 21 base pairs of Oenothera are lacking, as with Epilobium. Sequence data were obtained by the Sanger di- deoxy method (Sanger et al., 1977) on double- stranded in-vitro amplified products (Gyllensten, 1989) using internal primers (provided by C. Zu- rawski) and Sequenase 2.0 (United States Bio- chemical). In most instances, DNA fragments up to 300 nucleotides long were resolved on 6% ac- rylamyde gels, but 5% Long Ranger gels (AT Bio- chem) were employed in order to separate DNA fragments were obtained by using a proportion of 1.5 ul extension mix to 1 ul termination mix (both provided in the USB-Sequenase kit) in the termi- nation reactions. The critical step in the method of sequencing directly off the gene-cleaned, double stranded rbcL gene involves a heat shock imme- diately followed by a cold shock (James E. Ander- son, pers. comm.). The annealing mixture, formed by the rbcL DNA template and the primer, is boiled for three minutes to denature the double stranded DNA, and then plunged into liquid nitrogen. The labeling mixture is then added to the annealing mixture while the latter is thawing in the tube held between fingers. The termination reactions are per- formed according to the instructions of the USB- Sequenase kit. Sequences were directly aligned onto a reference sequence template of the rbcL sequence of Ni- cotiana tabacum L. and, subsequently, of Clarkia xantiana. To ascertain that the 12 sequences ob- tained belonged to coding genes, the nucleotide sequences were translated to the corresponding amino acid sequences with the TRANSLATE com- mand in the University of Wisconsin-Genetics Computer Group (UW-GCG) package. Missing nu- cleotide positions are listed in the legend of Table 2. e phylogenetic analyses were performed using the parsimony algorithms and tree analysis options of the PAUP 3.0г program (Swofford, 1991) on a Macintosh IIsi computer. Only putatively infor- mative characters were included in the analyses; autapomorphic characters were added later to the terminal branches of the resulting most parsimo- nious phylogram. Various approaches were fol- lowed, including Fitch (1971) parsimony (with and without the third codon position), character state- weighting, decay analysis, randomization of trees, and analysis of constrained trees. The ACCTRAN (accelerated transformation optimization) option was employed in all analyses. BRANCH AND BOUND search was used to find the most parsimonious Fitch trees, for bootstrap analysis, decay analyses, and analyses employing character state stepmatrices. Bootstrap analysis (see Felsenstein, 1985, 1988) was performed on the matrix of the potentially informative characters to provide a sense of the support of each clade in the resulting majority rule trees. We are aware of the limitations and possible misuse of the bootstrap procedure (see Sanderson, 1989; Wendel & Albert, 1992), but believe that bootstrap analysis (and decay analysis, see below) can provide useful insight concerning how this data set supports different clades. The decay analysis was accomplished by saving 675 Tribal Relationships in Onagraceae Conti et al. Volume 80, Number 3 1993 4 го 4 ÀÓ án gemessen n n n n n n Kt n DEA AA n v e 8 n6 ERE VL 3::9 L V р19910и9() Жз Же SE aes ву 4 Је а ея ке ДИ - e...’ Vio АИ 1'1979 1 1 19 3V- 5 I V штора сан i ые a О EN 1--0419:59 >La 0129101 neo gr -Y9 О "ОБОД VE до SLD ДИ Ор О V'L'VI9Vlé:99à:: 1:729 DoD) ж жб € wo we es x sa CO PMID о 12 DISYIN жє "мод00 0710 54 · т Ш | Дао А РА PP OV ДАНА" Дао) А АД DÁNDE sapiojdad оуттрпу pupianiad pmpn] 99:09: 4 D: LV9999LLVVL: 2: 19: 1:019::9:1:9:99L1V; LL: cL: VOV Ll: L0VIOVIOlI: 229^ ОДУ 259.--29-Y-2-1::1V2099-1V:: DV OL OLD: 7:97 1:9: 99L1V- LL Lo VOVo У VLL: 09 sab DOV" iid т Ш :0:19VV:VOLO2S10- V: ` 1911999 -DIOLLLV:L:: VO VVL МАЈУ 00: undae 985LV220VV: -Y999-L:29L* :99- LIDL Lc ++ `1: 999: D DBLLLV: L: 70V: Vo cc Yo суда ae patung 99LV99VVV: к ‘VO OL IVL V9 0 IDL 14599: -9IO9LLLV:-L-V- VO УУ МОД 97 77777 оар], Q9VOLLV990VV9DVVLLLVVVOLLOVLJ VIVIJIV V оло) uoxe J, 86698902867 c60VEELI 4244: 40 0867©2@088/5РЎЄТ 004vr22ecI 19628956e2990888/GP8GT10999v221 68,9S8G82?£22299v£108,1,22899Tv Tv Vv oC TTTT LLL 7444444444 49 eee! ииип uonisod зрцозјопм I IT T T To TI ё TI eT et IT E et Tè I uontsod uopor) ‘ansst sty} jo pua əy} је xipuaddy ay) ur persi st s1equinu uorssa2oe үче чәсу Jey) pue saouanbas 244 eui jo yı3uəj ayi uo поцешлојиј “3201 'S62-0cL '969 ‘#99 “OSS :012 годот 12g 190211) :£06 ‘ZOG “DISYIN] ‘ZOE :vXnpg “COE “ZOE :sapiojdad ^T *£6€ '265 :vuniansad vifimpn STHET ТЕТ 2821 Z8Z1 ‘SLOT “PLOT питу *ge9-9€9 “EOL LOE PANUNG SOL :01310]7) :әле 198 ттер 292 ouo IP шолу Sursstur suorisod зрцогјопи әлцешлоушиои [euonippy `å, Aq peieorput әле suonisod әрцоәјопи #шезгд поре Áq paieorpur әле D14101) Jo suonisod əy} 0} [еоциәрт ѕиоціѕой əpyoəpny (229) рјод) мол doi ay} ш pojeorpur зле pay} ayi ue 19410 ѕиоціѕой порогу "peuturexa exej 21 991 Jo so ouanbes 1291 eq шоу Sun[nsa1 s19j2e1eqo олцешлоји 98 eui jo xujeur ее "6 IYL 676 Annals of the Missouri Botanical Garden all the trees that are n steps longer than the shortest tree. The resulting strict consensus trees indicate which clades collapse to an unresolved polytomy at each extra step, thus providing a measure of the robustness of the monophyly of each clade in the minimal tree (see Bremer, 1988; Graham et al., 1991; Mishler et al., 1991; Smith & Sytsma, 1990, for examples). The stepmatrix described in Albert et al. (1993; & Mishler, 1992) was employed for character-state weighting. This stepmatrix takes see also Albert into consideration known transition versus trans- version biases for each position in the codon by assigning different weights to transitions and trans- versions for each position. In order to reflect lower probability of occurrence, transversions are weight- ed more than transitions: the highest weight is assigned to transversions in the second codon po- sition, the second highest to transversions in the first codon position, and the lowest to transversions in the third codon position. An additional analysis of the rbcL data involved the deletion of all third codon-position changes. Phylogenetic signal in the rbcL data was ex- amined by performing the option RANDOM TREES in PAUP. Hillis (1991) provided critical values for gl statistics (a common measure of left skewness of a distribution) of tree length distributions for up to eight taxa (P < 0.05 and P < 0.01). In order to compare the results of rbcL trees randomization with the critical values provided by Hillis, the three outgroup taxa (Punica granatum L., Trapa na- tans L., Lythrum hyssopifolia L.) and one taxon of Onagraceae were eliminated from the 12 taxa included in this rbcL study. As the monophyly of Jussiaeeae is not under dispute (see also results below), Ludwigia peploides, one of the two taxa representing this tribe, was eliminated from the data set. The frequency distribution of lengths of 10,000 random unrooted trees for the eight taxa left was analyzed. The topology and lengths of other published trees for Onagraceae using different data sets were investigated by user tree options in PAUP. The choice and number of taxa represented in these alternative trees did not always match those of the present rbcL study. Therefore, in cases such as with the nrDNA restriction site tree (Crisci et al., 1990), where Lopezieae is absent, missing taxa were placed as in the rbcL tree. In this way the number of extra steps required to force the rbcL sequence data onto the topologies generated from the other data sets was minimize Punica granatum жеи) Lythrum hys- sopifolia (Lythraceae sensu stricto) and Trapa na- tans (Ттарасеае) were used for global outgroup comparison (Maddison et al., 1984). Based on cla- distic analysis of morphological features, Johnson & Briggs (1984) indicated that the sister group to the Onagraceae includes Lyth lato апа / or Trapaceae, the latter а monotypic family. Ly- thraceae are characterized by several plesiom- orphic features and by a wide range of morpho- logical variation that suggest a central position in the order Myrtales (Dahlgren & Thorne, 1984). Both Johnson & Briggs (1984) and Dahlgren & Thorne (1984) included other small families within Lythraceae, most notably the monogeneric family Punicaceae. Punica consists of only two species, P. granatum, the pomegranate, and P. protopun- ica Balf. f., which may resemble the wild ancestor of the cultivated pomegranate (Dahlgren & Thorne, 1984). Most authors treat Punicaceae as a separate family, but wood anatomy and other morphological data suggest an inclusion within Lythraceae (Bridgewater & Baas, 1978) RESULTS The 12 rbcL sequences analyzed for this study contain 195 variable nucleotide positions; of these, 109 are autapomorphies, leaving 86 informative characters. Among the 195 variable positions, 137 (70.3%) are third codon positions, 34 (17.4%) are first codon positions, and 24 (12.3%) are second codon positions. The data matrix of the 86 informative char- acters used in the phylogenetic analysis is shown in Table 2. The last informative character is po- sition 1408, indicating that no variation is found among the last 20 nucleotides of the 10 taxa se- quenced to the end of the gene. Therefore, it can be assumed that the only informative character missing for the two cloned genes of Oenothera elata and Epilobium angustifolium is position 1408. Of the 86 putatively informative characters, 69 (79.7%) are binary characters, 15 (17.8%) are and two (2.4%) are 4-state Therefore, the minimum number of character-state changes possible (i.e., no homopla- sy present) with this data set is 105. The infor- mative changes are distributed differently, as ex- pected, among the codon positions: 63 (73.2%) third codon-position changes, 16 (18.6%) first co- don-position changes, and only seven (8.1%) sec- 3-state characters, characters. ond codon-position changes. The transition: trans- version ratio among all possible changes between Volume 80, Number 3 1993 Conti et al. 677 Tribal Relationships in Onagraceae TABLE 3. are reported as percentage. dq ie CLA = Clarkia; T Matrix of pairwise distances, based on the number of nucleotide differences between taxa. Distances RA = Trapa; PUN = Punica; LY T = Lythrum; PER = Ludwigia peruviana; ,. peploides; HAU = Hauya; FUC = Fuchsia; CIR = Circaea; ТОР = Lopezia; EPI = Epilobium; OEN = Denia "TO. CLA TRA PUN LYT PER PEP HAU FUC CIR LOP EPI OEN CLA 5. 4.4 6.0 4.1 4.4 2.9 2.6 3.8 3.2 3.4 1.9 TRA — 3.9 4.4 4.8 4.7 4.1 4.8 5.6 5.2 6.4 5.8 PUN — 3.9 3.5 3.4 2.8 3.5 4.3 3.8 5.4 4.6 T 4.4 4.0 4.4 4.0 5:3 5.93 6.1 5.1 PER 0.9 2.7 3.0 4.3 3:9 4.9 3.8 PEP 2.5 2.9 4.] 3.9 5.2 4.2 HAU — 1.4 2.1 2.1 4.0 3.1] FUC 1.7 2.5 4.0 2.7 CIR - 3.4 5.1 3:7 ТОР 4.2 3.3 EPI Е 3:3 ОЕМ states is 1.51, versus the expected value of 0.5, which implies approximately a three-fold transition bias. The intertribal pairwise distance, based on over- all sequence similarity, between Fuchsieae and Cir- caeeae is lower (1.7%) than the intergeneric dis- tance between Clarkia and Oenothera (1.9%) in the tribe Onagreae (Table 3). The highest rbcL sequence divergence (5.2%) within the family On- agraceae is found between Ludwigia peploides (tribe Jussiaeeae) and Epilobium angustifolium (tribe Epilobieae). Including the outgroup taxa, the highest divergence (6.7%) is seen between Ly- thrum hyssopifolia (Lythraceae) and Epilobium angus stifolium (Onagraceae). The gl value of —0.8849 obtained for the dis- tribution of 10,000 random trees is much smaller than the critical value of —0.47 (P 01) fur- nished by Hillis (1991) for eight taxa, thus pro- viding strong evidence that phylogenetic signal is present in the rbcL data set. To assess if evolu- nucleotide c ee DELL попагу noise changes occurring at the third codon position, the was introduced by distributions of 10,000 random trees were re-ex- amined by using only third codon-position changes (gl = —0.8406) and by eliminating all third codon- position changes (gl = — 1.0066). These gl values indicate that only an insignificant amount of "noise" is introduced by changes at the third codon posi- tions. The analysis of the informative rbcL sites yielded a single most parsimonious Fitch tree (Fig. 1). The tree is 168 steps long, with a Consistency Index (CI) of 0.625 and a Retention Index (RI) of 0.670. The CI value is considerably higher than the CI of about 0.3 expected for 12 taxa in a molecular data set (Sanderson & Donoghue, 1989). The num- bers of apomorphies and the bootstrap values sup- porting each branch are reported on the Fitch tree (7 majority rule tree from bootstrap analysis); the number of additional autapomorphies (not included in the original data set; Table 2) are indicated in parentheses for the terminal branches of the tree (Fig. 1). The fundamental features of the tree, such as the numbers of apomorphies supporting each branch, the numbers of consistent (i.e., С] = apomorphies, transitions and transversions, codon positions, bootstrap values and decay order, are summarized in Table 4 The lowest bootstrap value of the tree (44%) indicates the placement of /7au ya between the tribe Jussiaeeae and the rest of the family as the weakest branch. This result agrees with the observation that only five changes support this placement of /7au ya and that only two of these are consistent. The second weakest branch in the tree is the placement of Lopezia as the sister group to the clade formed by Epilobieae and Onagreae. This placement of Lopezia is supported by a low bootstrap value (56%) and by five changes, only one of which is consistent (one less than /7auya). The weak place- ments of Hauya and Lopezia are further reflected in the loss of resolution in the consensus tree of trees one step longer (Fig. 2A). The third weakest branch of the tree is that defining the tribe Ona- greae (represented by Clarkia and Oenothera), which is recovered as a monophyletic clade in only 70% of the replications. The Onagreae clade is supported by five changes, of which only one is consistent (see Table 4). The monophyly of the 678 Annals of the Missouri Botanical Garden Clarkia Onagreae 13 Oenothera 99% 8(21) TW ў Р 5 h Epilobium Epilobieae 10(8) ; | 5 Lopezia Lopezieae =]8 *| 20 Fuchsia Fuchsieae 11 10 Етти 93 100% y 12(10) : . Circaea Circaeeae 4(2 E | Наиуа Hauyeae 100% Ludwigia peruviana 9 98% Jussiaeeae Ludwigia peploides p 6(10 | — Punica Punicaceae = сч 12(13) © Lythrum Lythraceae Trapa Trapaceae FiGUR The most parsimonious, 168-step tree dii вуки here as a cladogram with branch lengths) resulting Bootstrap value weighted tree is "identic al. Onagreae is lost in the decay analysis at trees two steps longer (Fig. 2 All the remaining clades of the tree are strongly supported by the bootstrap (BS) analysis: (i) all Onagraceae exclusive of Ludwigia (BS = 100%); (ii) the two representatives of Ludwigia (BS = 100%); (iii) Onagreae—Epilobieae (BS = 99%); (iv) ceae and three outgroup families. Punica, Lythrum, an с ed are reported oe the branches. Each node is identified by a small letter. “The topology of the Punicaceae as the sister group to Onagraceae (BS = 98%); (v) Fuchsia—Circaea (BS = 93%). These high bootstrap values are correlated with the num- ber of changes supporting these clades (see Fig. Decay analysis supports the strength of these clades. The monophyly of Onagraceae is lost at Volume 80, Number 3 Conti et a Tribal Relationships in Onagraceae eight (= 4.8%) extra steps (176 steps; 78 trees), Е 5 — Ф when Lythrum and Punica form a polytomy with 2 в PA PALAA tees E^ OS Gn . f a m Onagraceae (Fig. 2E). Two clades, Onagreae- Epi- ae S lobieae and Jussiaeeae, survive the breakdown of v 2 Z ER . s % “| oNN mo the familial clade (Fig. 2E). The monophyly of the а y – i - : . . . iix | = a tribe Jussiaeeae is especially well supported: this ~ = 5 А a clade is lost only at 13 (7.7%) extra steps (Fig. En 2 |олаесеењео a - 2G). these changes, a large proportion (nine, js = ò E or 64%) is not affected by homoplasy, and six = = = (42%) are transversions. ~ = & ha, | oO 5 эое а The analysis of constrained trees clearly indi- eb C E ; Шо” $ + о — чес оо с со cates that the topology of the rbcL tree is closest He A = to that of the tree based on chloroplast DNA re- = ~ је | e Е Р : о — . : eS = 7р striction sites; only one extra step is necessary in BOs = коо оо соо order to reach the topology of the cpDNA restric- 9 uxo ә Я : — RI Та tion site tree (Fig. ) The second closest tree "ELEC M Це | | fe 5 |=- счао <= сс осо (Fig. 3C: two extra steps) is the nrRNA sequence zl 2. рт . . o Ф tree, followed by the rbcS amino acid sequence — EF |o tree (Fig. 3D: eight extra steps), the nrDNA re- ¿22 [2|ano-o-oo0 = F й " o t— ' striction site trees (Fig. 3E: 10 extra steps for the e. E | = 3 mE as Dollo parsimony, 50% majority rule consensus tree ж о ава ee x Ф 5- a 2 extra steps for the Wagner parsimony, "s Ela Ju : 2.89 Ра 50% majority rule consensus tree), and, finally, soe || голаламоо ЕЕЕ ЧС the morphological tree (Fig. 3F: 14 extra steps). o & || Lal c The application of the character-state stepma- язе Se | 2D 2 qQ|o.,n nooo -=0 trices simultaneously to the first, second, and third > E Bex | ee woe . " e) о codon position yields the same tree obtained by EN А : ; У xc А = 3 Oe; 125 -— simple Fitch parsimony (Fig. 1). Elimination of all poles the changes that occur at the third codon position Е Es У 3 о Ф - produces four equally parsimonious trees. In the 2 Б + CREER strict consensus tree (Fig. 4) Onagreae and Ona- 22 5 $ |^ graceae become paraphyletic groups (respectively, 2 52 |.| 5 «ево љ- = Oenothera forms a clade with Epilobium, and Pu- BID y -— us = 085 =, nica falls within Onagraceae). Additionally, Lo- - m : т|со бс ес Оо ч гс + : А ў A NE = m pezia, Hauya and Punica collapse to a polytomy sia with the Jussiaeeae clade and the clade formed by cae et SS ee 7 = 2 7 the rest of Onagraceae. CE- d ^A ор WD + — сч — NN — aa y Ф = zn 6 Е © € РЕ or DISCUSSION a C A ee и Е асв Cladistic analyses of rbcL sequence variation, = а 2 T . . . . € om E T — = using Fitch or weighted character-state parsimony, соз Е “ € ~ С | i Е ен s 2 5 provide a single most parsimonious tree (Fig. 1) 532 lalanennm om for Onagraceae and outgroup families. Many of 23. +“ | | wee, the clades defined in this tree are strongly sup- о 9 Е o. «|->-о-о-оосхов ported by bootstrap and decay analyses. On the gs gi Е. 8 other hand, trees produced by eliminating third NU $3 . А . S = 70 codon positions portray evolutionary relationships 2235 difficult to support based on other lines of evidence с T oa IN Fig. 4). This result suggests that changes at the d. UN | = Д& = n third codon position in rbcL, far from producing 252 |215 2 Б О = А . А zl = о me cad sd mere evolutionary noise, are phylogenetically in- At. |e] sé = E p 2 : AL- o 0 NS formative within Onagraceae. FL De щ| а= 5 T a у ~ nn n = EM The rbcL sequence data permit a re-evaluation Ы 2 E 5 = E E = y E a © 5 Bm © of evolution within Опаргасеае and allow for com- 29 6 gem m O e E tS < < na 100 na na na na na na na 98 5 Transversion В S(%): Fitch tree Decay order 680 Annals of the Missouri Botanical Garden А (+1) Clarkia B (+2) Clarkia Oenothera Oenothera Epilobium Epilobium Lopezia Lopezia Fuchsia Fuchsia Circaea Circaea е Наиуа __Г—Д— Ludwigia peruviana Ludwigia peruviana LLL oid peploides Ludwigia peploides Punic Punica Paus Lythrum Trapa Trapa С (+6) Clarkia D (+7) Clarkia Oenothera Oenothera Epilobium Epilobium Lopezia Lopezia Fuchsia Fuchsia Circaea Circaea Hauya Hauya Ludwigia peruviana Ludwigia peruviana Ludwigia peploides Ludwigia peploides Punica Punica | Lythrum - Lythrum Trapa Trapa E (+8) Clarkia F (+9) Clarkia udwigia peruviana Ludwigia peruviana Ludwigia peploides Ludwigia peploides unica Lythrum Lythrum G (413) Clarkia nothera Epilobium Ludwigia peruviana Ludwigia peploides nica Y o ка | Y e C m =; О акб БЕ fs ser = 3 8 “У Com Om w © B iii EB 8 SERERE 3 d Lythrum Ti FIGURE 2. Decay analysis of the most parsimonious, 168-step tree. The number of extra steps at which each clade decays is reported in parentheses. Solid bars indicate the polytomies formed by the collapse of each monophyletic group in trees longer than the 168-step tree. Volume 80, Number 3 1993 Conti et al. Tribal Relationships in Onagraceae A) rbcL Sequence Onagreae Epilobium Lopezia Circaea Fuchsia Hauya Ludwigia C) nrRNA Sequence (42) Onagreae Epilobium Circaea Fuchsia Hauya Lopezia Ludwigia E) nrDNA Restriction Site (410) Onagreae Epilobium E Circaea Fuchsia Hauya Circaea (412) Ludwigia B) cpDNA Restriction Site (41) Onagreae Epilobium Lopezia Circaea Fuchsia Hauya Ludwigia D) rbcS AA Sequence (+8) Onagreae Epilobium Hauya Fuchsia Circaea Lopezia Ludwigia F) Morphology (414) m Onagreae Epilobium | наша Lopezia Fuchsia Circaea Ludwigia FIGURE 3. The six trees resulting from the cladistic analyses of various data sets for Onagraceae. The number of extra steps required to force the rbcL s өз ошо E topologies generated from the other data sets is e ЗЕ: the solid branch е to Circaea indicates its placement in the Dollo uec 50% m majority a consensus tree (0 extra steps); the dotted branch indicates the placement of Circaea in the gner parsimo ef result from studies of: Conti et al. (this article; Fig. 3 0% majority role consensus tree (12 extra steps). These trees A); Sytsma et al. (1991b; Fig. 3B); Bult & Zimmer (1993; Fi i ). 3C); Martin & Dowd (1986; Fig. 3D); Crisci et al. (1990; Fig. 3E); Hoch et al. (1993; Fig. 3F 682 Annals of the Missouri Botanical Garden Clarkia Oenothera ya г Ludwigia peruviana L—— L. peploides Punica Lythrum Trapa FIGURE 4. Strict consensus tree of the four most parsimonious trees resulting from elimination of third co- don positions in rbcL sequences. parisons of congruence between the single rbcL tree and those generated from a variety of other data sets, both molecular and morphological. l. INTERTRIBAL RELATIONSHIPS IN ONAGRACEAE The Fitch tree (Fig. 1) strongly supports three monophyletic tribal groups within the family: (i) all tribes except Jussiaeeae; (ii) Fuchsieae and Cir- caeeae; (iii) Onagreae and Epilobieae. The place- ments of Hauyeae and Lopezieae are less well sup- ported. The most parsimonious rbcL tree agrees with all other data sets (Fig. 3) in placing Ludwigia (sole genus of the tribe Jussiaeeae) as sister to all other Onagraceae (with a bootstrap value of 100%, 11 informative characters, seven of which are con- sistent, and a decay index of 13). Ludwigia appears to retain siomorphic morphological features, including pres- a number of ple- ence of up to seven parts in the floral whorls, as opposed to the typical four-merous flowers of the other Onagraceae (Eyde, 1981, 1982; Hoch et al., 1993), massive, ovuliferous placentas, minor vascular bundles, and absence of a constriction in the floral tube beyond the ovary. Ludwigia shares the plesiomorphic Met- Lys N-terminal amino acids of the nuclear-encoded rbcS polypeptide with all other angiosperm taxa (mostly from Myrtales) an- alyzed by Martin & Dowd (1986), whereas the rest of Onagraceae are clearly set apart from Lud- wigia by the distinctive and synapomorphic Phe- Asn N-terminus. Ludwigia is also distinguished from the rest of Onagraceae by autapomorphies, most notably the unique nectary position. The location of the nec- taries of Ludwigia around the base of the style, inside the anthers (instead of on the floral tube, outside the anthers, as in all other Onagraceae), induced Eyde to conclude that epigyny evolved independently in the two main evolutionary lines of Onagraceae, namely Ludwigia and the rest of the family (Eyde, 1982; Raven, 1988 A major achievement of the rbcL sequence data in elucidating the relationships among tribes in Onagraceae is the definition of the clade Fuch- sieae-Circaeeae. Fuchsia and Circaea are quite distinctive morphologically. Fuchsia is generally shrubby or treelike, mostly with large, bird-polli- nated flowers, and has a Gondwanaland distribution (although with some distinctive sections in Nort America: Berry, 1982; Raven, 1979, 1988). Cir- caea, on the other hand, is herbaceous, flowered, fly-pollinated, and has a Laurasian dis- tribution. Morphologically the two genera each exhibit numerous autapomorphies, share none or a few derived states with other tribes, and are thus not linked in phylogenetic analyses of morpholog ical characters (Hoch et al., 1993). Additionally, if one considers the relatively low sequence diver- small- gence (1.7%) between Circaeeae and Fuchsieae (versus 1.9% between Clarkia and Oenothera in the tribe Onagreae: see Table 3), the implications of these results with respect to biogeography and perhaps rates of morphological divergence are con- siderable. The woody Fuchsia lineage has undergone a seven- to eleven-fold slowdown in rbcL sequence divergence relative to the herbaceous Circaea lin- eage (with DELTRAN and ACCTRAN transfor- mations, respectively; see Fig. 1). A low amount of chloroplast DNA divergence (as measured by P values using restriction site data) between two, mor- phologically distinct species of Fuchsia (a tree ver- sus a scrambling liana) has already been noted (Sytsma et al., 1991a; Sytsma & Smith, 1992). Fuchsia and Hauya, the other genus of Onagra- ceae that includes shrubs and trees, actually show the lowest amount of intertribal rbcL sequence divergence (1.4%; see Table 3), even though they are not supported as a clade. Both tribes have few autapomorphies not due to homoplasy (two in Fuchsia, four in Hauya; see Fig. 1). The apparent slow rate of rbcL sequence divergence in both the woody Fuchsieae and Hauyeae is also reminiscent of that reported for the Palmae (Wilson et al., 90 The third well supported clade includes Ona- greae (nine genera) and Epilobieae. These two tribes exhibit many more apomorphic features than do the other tribes (Raven, 1979, 1988; Hoch et al., 1993). The derived nature of these two tribes is suggested by their placement in a terminus of the rbcL phylogram (Fig. 1). Within the Onagreae-Epilobieae clade, the Volume 80, Number 3 Conti et al. 683 Tribal Relationships in Onagraceae branch supporting the monophyly of Onagreae sister tribe to the clade Fuchsieae—Circaeeae. The (represented by Clarkia and Oenothera) is not very 1, 2B). The weak support of the monophyly of Onagreae is reflected in the rbcS amino acid sequence tree (Fig. 3D) and the mor- phology tree (Fig. 3F: strict consensus tree of the seven most parsimonious trees), which both place the Onagreae in an unresolved trichotomy with Epilobieae and Hauy Raven (1979) ы that Clarkia (now in- cluding Heterogaura; see Lewis & Raven, 1992), as well as Epilobium, share derived character states, robust (Figs. unique in the family, such as a dry, papillate stigma and four-lobed stigmas with the lobes in the com- missural position. The presence of commissural stigmas was explained i Eyde (1982) as related to the evolution of protandry in these two genera. Further sampling of Onagreae and Epilobieae for rbcL sequence data would furnish an independent test for the possible paraphyly of Onagreae (Raven, 1979; Hoch et al., 1993) and for the phylogenetic interpretation of these unique morphological char- acters (Baum & Larson, 1991; Coddington, 1988). The placement of Hauyeae between Jussiaeeae and the rest of Onagraceae is the weakest branch of the rbcL tree, with a low bootstrap value (44%), five synapomorphies (only two of which are con- sistent), and the collapse to a polytomy at just one extra step from the shortest tree (Fig. 2A). Ca- lylophus and Gaura (both Onagreae; chromosome number x — 7) share some apparently derived embryological features with the otherwise gener- alized Havya, which has the basic chromosome number x = The second sait branch in the rbcL Wagner tree is the placement of Lopezia as the sister group of Onagreae-Epilobieae, supported by five synapo- morphies, of which only two are consistent. 2. CONGRUENCE WITH OTHER CLADISTIC ANALYSES Тће picture of phylogenetic relationships within Onagraceae resulting from the rbcL sequence data can be compared with the evolutionary scenarios provided by both molecular and morphological studies (see Hillis, 1987; lisa, 1990; Donoghue Sanderson, 1992). e placement of Hauyeae between Jussiaeeae and the rest of Onagraceae as seen in the гђе tree is incongruent with the cpDNA restriction site tree and the nrRNA sequence tree, which are in- dicated as the closest to the rbcL cladogram (re- spectively at one and two extra steps) by the anal- ysis of constrained trees (Fig. 3). The cpDNA and the nrRNA trees agree in placing Hauyeae as the Cladistics 8: —— V. nrDNA restriction site tree does not corroborate the topology of the nrRNA sequence tree, in that the former places Hauyeae as the sister tribe to amino acid sequence data indicate that Hauyeae, Ona- greae, and Epilobieae form a tight group (Fig. 3D, F) Fuchsieae. Both morphological and rbcS The rbcL sequence data corroborate the cpDNA restriction site data (Sytsma et al., 1991b) and the nrRNA sequence data (Bult & Zimmer, 1993) in defining the Fuchsieae-Circaeeae clade. In the nrRNA sequence tree, however, the Fuchsieae Circaeeae clade is less robust than in the rbcL tree, since it is supported by only four synapomorphies, three of which are homoplasious. There is no agree- ment concerning the phylogenetic relationships of he nrDNA tree, The placement of Circaeeae in the two consensus trees Fuchsieae and Circaeeae among t the rbcS tree, and the morphological tree. resulting from nrDNA restriction site data is un- stable. In fact, the 50% majority rule consensus tree based on Dollo parsimony indicates Circaeeae as the sister tribe to the clade Fuchsieae- Hauyeae, whereas the consensus tree based on Wagner par- simony places Circaeeae as the intermediate tribe between Jussiaeeae and the rest of Onagraceae (see Fig. 3E) There is agreement between the rbcL sequence tree (Fig. 3A) and the cpDNA restriction site tree (Fig. 3B), both maternal trees, in indicating Lo- pezieae as the sister tribe to the clade Onagreae Epilobieae. On the other hand, the two nuclear genome trees, based on nrRNA nucleotide and rbcS amino acid sequence data, place Lopezia between Ludwigia and the rest of Onagraceae. The mor- phological tree defines Lopezieae as the sister tribe to Fuchsieae on the basis of a single trait (lack of starchy pollen). No nrDNA restriction site data are available for Lopezieae. The shifting placements of Lopezieae in the trees resulting from different data sets, especially the discrepancy between the two maternal and the two biparental trees, suggest the possible occurrence of ancient chloroplast capture in this tribe (see Smith & Sytsma, 1990; Rieseberg & Brunsfeld, 1992). Further sampling of the genus for rbcL sequences and additional sequence data of nuclear genes may help to resolve the relation- ships of Lopezieae with the rest of the family. LITERATURE CITED ALBERT, V. A. & B. D. MISHLER. 1992. On the rationale and utility E Dr шры кеч sequence data к T [m & B. D. MISHLER. 1993. Annals of the Missouri Botanical Garden Character-state weighting for cladistic analysis of pro- tein- n ing "Pie sequences. Ann. Missouri Bot. Gard. 80: 75 AVERETT, J. "E EP . H. Raven. 1984. {тиле Pa Onagraceae. Ann. Missouri Вог. Gard. 7 Baum, D. А. & A. Larson. 1991. fresas Bonn А зано а for ое сһагасїег macroevolution. Syst. Zool. 40: 1- Berry, P. E. "The systematics i evolution of Fuchsia sect. uie (Onagraceae). Ann. Missouri «198, Bot. Gard. 69: BREMER, P 1988. "The limits of amino acid sequence data in angiosperm неш reconstructions. Evolution 42: 795-803 BRIDGEWATER, S. D. & P. BAAS. 1978. Wood anatomy of the ее IAWA Bull. 1: 3-6. Вит, C. J. & E. A. ZIMMER. 1993. Nuclear ribosomal RNA ied for inferring tribal relationships with- Onagraceae. Syst. Bot. 18: 48-63. 1975. Wood anatomy of Onagraceae, ith notes on alternative modes of photosynthate movement | ы woods. Ann. Missouri Bot. Gard. 62: 4. CODDINGTON, J. үч 1988. Cladistic tests of adaptational hypotheses. Cladistics 4: 3-22. Crisci, J. V., E. A. ZIMMER, P. C. Носн, С. B. JOHNSON, . Морр & N. S. Pan. 1990. Phylogenetic im- plications of ribosomal DNA restriction site variation in the plan q amiy Onagraceae. Ann. Missouri Bot. Gard. 77: 38. DAHLGREN, R. & i F. THORNE. 1984. The order Myr- tales: Circumscription, variation, and bios Ann. Missouri Bot. Gar 699. Dorsey, J., M. DURBIN, Е. M. GOLENBERG, M. T. CLE & D. P. Evolutionary analysis o of the large subunit of carboxylase rbcL) nucleotide se- quence among the grasses (Gramineae). Evolution 43: 1137-1156. DoNocHuE, M. SANDERSON. 1992. The suitability of molecular and eli evidence in reconstructing plant phylogeny. Pp. 340-368 in D. E. Soltis, P. S. Soltis & J. J. Doyle (editors), Plant Molecular Systematics. Chapman and Hall, New York. ERLICH, H. A. PCR Technology: Principles and н for DNA Amplification. Stockton Press, ork. EYDE, R. H. 1981. Reproductive structures and ev lution in Ludwigia (Onagraceae). III. Cu UE nectaries, conclusions. Ann. Missouri Bot. Gard. 68: 379-412. 1982. Evolution and systematics of the On- agraceae: Floral anatomy. Ann. Missouri Bot. Gard. 9. FELSENSTEIN, J. 1985. Confidence limits on phylogenies: An approach using the bootstrap. Evolution 39: 783- 988. Phylogenies from molecular sequences: Inference and reliability. Ann. Rev. Genet. 22: 521- 565 Етен, W. М. 1971. evolution: Minimum Toward defining the course of change for a specific tree to- pology. Syst. Zool. 20: 406-416. GRAHAM, L. E., C. F. DELwic js “ D. Мазнтек. 1991. Phylogene etic connection ween the “Green Al- gae" and the * Bone E i Bryol. 4: 213- 244. GYLLENSTEN, U. 1989. Direct sequencing of in vitro amplified DNA. Pp. 45-60 in H. A. Erlich (editor), PCR Technology: Principles i Applications for DN Amplification. Stockton Press, New York HiLus, D. M. 198 Molecular versus morphological approaches to systematics. Ann. Rev. Ecol. Syst. 18: 23-42. HR помо ан phylogenetic quences Pp. 278- 294 n M. M. Nu. P I У Craerafi (editors), Phylogenetic Analysis of DNA Sequences. Oxford New Yor . У. CRISCI, “Н. Tose & P. E. BERRY. А cladistic analysis y a plant family On- agraceae. Syst. Bot. 18: 31- JOHNSON, L. А. S. & B. C. BRIGGS. grrr Myrtales and ЕРТЕ phylogenetic analysis. Ann. Missouri Bot. Сага. 71: 700-756. Lewis, Н. 4 P. E. RAVEN. w combinations in the genus Clarkia (Onagraceae). Madroño 39: 163-169. Mappison, W. R., M. J. DONOGHUE & D. R. Mappison. 84. Outgroup analysis and parsimony. Syst. Zoo 33: 83-10 ManriN, Р. С. с J. М. Похр. 1986. Ares studies using protein sequences within the order tales. Ann. Missouri Bot. Gard. 73: 442-448. MisHLER, B. D., M. J. DONOGHUE & V. A. ALBERT. 1991. The decay index as a measure of relative robustness within a cladogram. Hennig X, Toronto, Canada. [Abstract.] Моша, K., Е. FALOONA, & Н. ERLICH. cation of DNA in t action. Cold Spring Harbor Symp. Quant. Biol. 51: 63-273. Носн, P. C., 199 S. Sc HARF, R. Saiki, G. HORN PATEL, V. C., J. J. Skvarta & P. H. Raven. 1984. Pollen chascar in relation to the delimitation of Myrtales. Ann. Missouri Bot. Gard. 71: 858-969 Raven, P. H. 19 A survey of perpe biology in Та New Zealand J. Bot. 59 boe dap as a Bí of plant evo- fitout 07 in L. Gottlieb & S. Jain SPP Hi. Plant o Biology. Chapman and Hall, Lon don `& D. 1. AXELROD. ography and past continental movements. souri Bot. Gard. 61: 539-673. RIESEBERG, L. Н. & S. J. BRUNSFELD. 1992. Molecular ant introgression. Pp. 151-176 in & J. J. Doyle (editors), Plant Molecular Systematics. Chapman and Hall, New а RSON, M. Ј . Confidence limits on phylog- nies; the bootstrap revisited. Cladistics 5: 113-129. & M. J. DonocHuE. 1989. Patterns of vari- ation in levels of homoplasy. Evolution 43: 1781- 1795 1974. Angiosperm bioge- Ann. Mis- SANGER, F., S. NICKLEN & A. R. COULSON. 1977. DNA sequencing with chain M inhibitors. Proc. atl. Acad. Sci. U.S.A. 5463. SKV ARLA, J. J., P. H. RAVEN Im PRAGLOWSKI. 1976. 44 479 in J. K. Fergu uson LEJ Muller aeneis ANE Significance of the Exine. ress, New York. [Linnean Soc. Symp. Ser., SMITH, J. F., К. 5 SYTSMA, J. S. SHOEMAKER & Volume 80, Number 3 1993 Conti et al. Tribal Relationships in Onagraceae SMITH. 1991. А qualitative comparison of total cellular DNA extraction protocols. Phytochem. Bull. 23: 2-9 5мин, R. L. & K. J. Syrsma. 1990. Evolution of Populus nigra (sect. Aigeiros): bridization and the chloroplast contribution of Pop- lus alba (sect. Populus). Amer. J. Bot. 77: 1176- 7. Sorris, D. E., P. S. Sorris, M. Т. СгЕсс & M. DURBIN . rbcL sequence divergence and phylogenetic relationships i in Saxifragaceae sensu lato. Proc. Natl. Acad. Sci. U.S.A. 87: 4640-4644 SwoFFORD, D. L 1991. danh Phylogenetic Analysis Using Parsimony, Vers r. Computer program distributed by the Illinois Natural History Survey, Champaign, Illinois. SyTsMa, K. J. 1990 . DNA and me A oe of plant ee Trends Ecol. Evol. 5 2e & J.F 992. Molecular ial of Onagraceae: Peu. from Clarkia and Fuchsia. Pp. 295-323 in D. E. Soltis, P. S. Soltis & J. J. Doyle wp о Molecular Systematics. Chap- and Hal man :RRY. 199la. The use o йй last DNA to assess biogeography and evolu- tion of morphology, breeding systems, and flavonoids in к ide sect. Skinnera сао Syst. Bot -209. 16: — & P. Носн. 19915. A пая DNA analysis of tribal and generic relationships wit . Amer. J. Bot. 78(6): 222. er 2. An updated Wa e vd classi- fication of the flowering plants. Aliso 13: 36 Tope, H. & Р. 983. An о. analysis of Myrtales: Its definition | characteris- ics. Missouri Bot. Gard. 70 EN of polysporan- giate anthers in Onagraceae. Amer. J. Bot. 73: 475- 88. 4 van VLIET, С. J. C. M. & P. Baas. 1984. Wood anatomy and classification of the Myrtales. Ann. Mis- souri Bot. Gard. 71: 783- WENDEL, J. F 1992. Phylogenetics of the cotton genus (Gossypiu ит): Character-state 1990. Chlo- roplast DNA evolves slowly i ш the uy family (Are- caceae). Molec. Biol. Evol. 7: 303- WINTER, P. & К. НЕ ү И five-base- pair deletion in the gene for the large subunit causes the lesion in the ribulose bisphosphate терш lase/ oxygenase-deficient plastome mutant sig of Oe- nothera hookeri. Bot. Acta 101: 68-7 NUCLEOTIDE SEQUENCES OF THE rbcL GENE INDICATE MONOPHYLY OF MUSTARD OIL PLANTS' James Rodman, Robert A. Price,’ Kenneth Karol, Elena Conti, Kenneth J. Sytsma,* and Jeffrey D. Palmer? ABSTRACT Nucleotide Te for the chloroplast rbcL gene were obtained from representatives of 11 of the 15 plant families known to two widely ed pus and thus im ce glucosinolates (mustard oil glucos ides). Parsimony analyses indicate that these constitute ingaceae, and Tropaeolaceae. The second mustard oil group is restricted to the euphor- biaceous genus Drypetes. a the subject of an independent rbcL analysis, pairs with Bretschneidera, and these adi andra, “ many aspects of Ro dman P based predominantly on morphological characters; ен the molecular and а он evidence support Da deren s radical classification of an expanded order Capparales Flowering plants produce a remarkable array of so-called secondary metabolites, e.g., alkaloids, cy- anogens, mustard oils, and phenolics. Such natural products often have beneficial (spices, medicinals) or deleterious (allergens, poisons) effects on hu- mans. The evolutionary origins and diversification of secondary metabolites are poorly known, with instances of parallel or convergent evolution sus- pected for many of them (Giannasi & Crawford, 1986). Their use as taxonomic markers (often un- der the rubric of **micromolecular characters") is widespread and of long standing, but it raises a suspicion of circularity. Classification schemes, in- terpreted as evolutionary scenarios, are used to explain the origins and changes in micromolecules, which may have been incorporated into the data- base used to construct those very classifications. An independent set of characters is required to evaluate the phylogenetic utility and reliability of micromolecular data. Mustard oil glucosides or glucosinolates, the compounds that release distinctive pungent prin- ciples in radish, mustard, and capers (Fenwick et al., 1983), provide a classic example of taxonom- ically useful micromolecules (Ettlinger & Kjaer, 1968; Kjaer, 1973; Rodman, 1991a, b). Char- acteristic of 14 families of dicots and of the genus Drypetes, usually allied with Euphorbiaceae, the compounds are structurally diverse and amenable to rapid chromatographic analysis (Rodman, 1978, 1981). They are demonstrably useful as popula- tion- and species-level markers (Al-Shehbaz, 1973; Rodman, 1974; Rodman et al., 1981; Al-Shehbaz & Al-Shammary, 1987) and have been employed in studies of hybridization (Rodman, 1980; Kruck- eberg et al., 1982), migration (Rodman, 1976, 1986), and insect-plant interactions (Rodman & Chew, 1980; Louda & Rodman, 1983). The bio- synthesis of glucosinolates is known in broad outline and is thought to be a unitary pathway in all these plants (Glover et al., 1988; Boufford et al., 1989). However, little enzymological data exist to evaluate the presumed homology of mustard oil biosynthesis; this is conspicuously so for Drypetes, placed in a family otherwise lacking mustard oils but well known for its cyanogenic glycosides (Rodman, 1981). (J. D. Palmer) are also gratefully "acknowledged is solely for purposes of identification ment of Biology, I ettig generously Delwiche prove uter matters as adep mp ational Science Foundation gents BSR-9020055 (K. J. dad and BSR-8996262 ogy, Natonal Science Foundation, Washington, D.C. 20550, U.S.A.; Federal ndiana University, Bison mington, Indiana 47405, U.S.A. ‘ dias of Botany, University of Wisconsin, Madison, Wisconsin 53706, U.S.A. ANN. Missouni Вот. Garp. 80: 686-699, 1993. Volume 80, Number 3 1993 Rodman Jedi dis a Mustard Oil Plants TABLE 1. Two recent and contrasting classifications of mustard oil plants, implying either convergence (Cronquist, 1981, 1988) or unique origin/parallelism (Dahlgren, 1975a, 1977) for the mustard oil/m Wher nonglucosinolate families are included in the order, an da (including Putranjiva) is the only genus of Euphor yrosin cell syndrome. "etc." follows the name of the glucosinolate taxa. orbiaceae known to produce glucosinolates. Both authors associate the genus Pentadiplandra with the family Capparaceae (treated as a separate family in Rodman, 1991а, b). Cronquist (1981, 1988) Dahlgren (1975a, 1977) Capparales— Brassicaceae, Capparaceae, Reseda- ceae, Moringaceae, Tovariace Bataceae eae. Batales — and Gyrostemonaceae. a uphorbiales— Drypetes, etc. Sapindales— Akaniaceae, Bretschneideraceae, etc. Geraniales— Limnanthaceae, Tropaeolaceae, etc. Capparales— Brassicaceae, Capparaceae, Resedaceae, oringaceae, Tovariaceae, Bataceae, Gyrostemona- ceae, Salvadoraceae, е ну Tropaeolaceae, ceae. — Drypetes, Sapinda ia etc Mustard oils are typically accompanied by a unique catabolic enzyme called myrosinase (Bones et al., 1), which is sequestered or concentrated in specialized idioblasts known as myrosin cells (Јаг- 1977; Jergensen, 1981). The origin and taxonomic distribution of the glucosinolate- gensen et al., myrosinase trait present a challenging question: Has glucosinolate biosynthesis evolved once or re- peatedly? Are the mustard oil plants a monophy- letic lineage or a polyphyletic assemblage? The classification of mustard oil plants has at- tracted the attention of numerous phylogenetically inclined taxonomists. Rodman (19912) identified 18 different schemes published in the 20th century, three of which have been discussed in detail for their bearing on glucosinolate evolution (Rodman, 1981). Two of these classifications (Table 1) pro- vide contrasting end-points on the spectrum of recent taxonomic interpretation of these plants: Cronquist (1981, 1988) and Dahlgren (1975a, 1977). Cronquist’s view, shared probably by a ma- jority of taxonomists, would imply convergence for glucosinolates in several orders of his dicot sub- classes Dilleniidae and Rosidae. To force an inter- pretation of unique origin in the context of Cron- quist’s classification would necessitate multiple instances of biosynthetic loss or repression in nu- merous lineages. Dahlgren, in contrast, lumped most mustard oil taxa into a single large order Capparales, excluding only Akania, Caricaceae, and the genus Drypetes of Euphorbiaceae. He placed Caricaceae in an order Violales that he affiliated with Capparales, and in turn he allied these with Euphorbiales. Dahlgren (1975a, 1977) originally associated Akania with Sapindales but later (Dahlgren et al., 1981) suggested an affiliation with Caricaceae. Thus, Dahlgren's scheme permits the interpretation that glucosinolate biosynthesis arose once in a stem lineage to these three orders with subsequent loss or repression in the remaining Violales and in all Euphorbiales except Drypetes. An alternative evolutionary scenario from Паћ!- gren's classification would postulate two or three parallel origins within a closely related group of dicots (Rodman, 1981). Although Dahlgren (1980, 1983) later retreated from his original view of an expanded Capparales, the earlier classification pro- vides a striking contrast to the convergence and polyphyly implied by so many traditional schemes such as Cronquist’s Recent cladistic and phenetic studies of mustard oil taxa provide the first explicit evaluation of vari- ant classifications of these plants (Rodman, 199 1a, b). Ninety characters were assembled for the cla- distic analysis, including anatomical and morpho- logical (vegetative and reproductive), chromosom- al, palynological, phytochemical, and ultrastructural features; morphological characters, broadly de- The results were in striking agreement with Danion s (1975a, 1977) broad concept of an expanded order Cap- fined, predominated in this data set parales. A clade of core Capparales (Brassicaceae, Capparaceae, Resedaceae, and Tovariaceae) was joined with Gyrostemonaceae and in turn allied with Bataceae, Salvadoraceae, and the nonglucosinolate Koeberlinia. Diagnostic features for this group in- clude: vestured pitting; curved embryo in seed; dilated cisternae of the endoplasmic reticulum; flo- ral tetramery; myrosin cells; and glucosinolates of several biosynthetic types. On the basis of less compelling character evidence, the clade was joined successively by: sister taxa Moringaceae + Tro- paeolaceae; then sister taxa Akania + Bret- schneidera; and then sister taxa Caricaceae + Pentadiplandra. Cladistic linkages among the mustard oil taxa held whether glucosinolates were 688 Annals of the е Botanical Garden incorporated as characters into the analysis or not. Drypetes remained the most distant mustard oil taxon. Limnanthaceae, linked to Balsaminaceae, provided a second likely instance of convergence for glucosinolates. An independent evaluation of taxonomic rela- tionships of mustard oil plants is here provided by a comparative analysis of nucleotide sequences from the chloroplast gene for the large subunit of ri- bulose-1,5-bisphosphate carboxylase/oxygenase (rbcL). Theoretical and practical reasons for the phylogenetic value of rbcL have been discussed (Ritland & Clegg, 1987; Palmer et al., 1988; Clegg & Zurawski, 1992), and an increasing number of empirical studies attests to the utility of rbcL se- quence data (Doebley et al., 1990; Soltis et al., 1990; Les et al., 1991; Donoghue et al., 1992; 1992; Kim et al., 1992; Olmstead 1992; Rettig et al., 1992). Gene sequence data provide a large number of variable, homolo- gous characters (nucleotide site positions) that promise ample resolving power across a broad range Giannasi et al., et al., of taxa. Three assumptions are here made in ac- cepting rbcL sequence data as informative about plant phylogeny: (1) the rbcL gene is representative of the plant's genome, at least the (often) mater- nally inherited chloroplast component; (2) nucle- otide site differences among the taxa reflect a con- with parallel or convergent mutations not great enough servative history of evolutionary change, to obscure phylogenetic linkages; and (3) one or two taxa can represent large families because se- quence differences within families are less than those between families. In the long run these as- sumptions will be evaluated against our increasing knowledge of plant genome diversity. MATERIALS AND METHODS For our study, rbcL nucleotide sequences of ca. 1,400 bases each were assembled for 33 taxa. Nine of these are here reported as new: Bretschneidera sinensis, Capparis hastata, Carica papaya, Cle- ome hasleriana, Drypetes roxburghii, Koeberlin- ia spinosa, Moringa oleifera, Reseda alba, and Tovaria pendula. Four published sequences were drawn from GenBank (Bilofsky et al., | Amaranthus hypochondriacus, Gossypium hir- sutum, Plumbago capensis, and Spinacia oler- acea. The rest were either generously provided by others or are published elsewhere (Table 2). Ten of the known glucosinolate-producing families plus the genus Drypetes are represented in our analysis. Akania is the subject of an independent rbcL re- port (Gadek et al., 1992). The remaining mustard oil taxa, Gyrostemonaceae, Pentadiplandraceae, and Salvadoraceae, have not been sequenced. Koeber- linia, usually allied with Capparaceae, has been reported to develop myrosin cells (Gibson, 1979); it has been analyzed for glucosinolates but with negative results (Ettlinger & Kjaer, 1968; Rod- man, 1981). Here it is treated as its own family (see Rodman, 199 1a, b). Recently, the thiocyanate or isothiocyanate breakdown products of presump- tive parent glucosinolates were reported from two species of Phytolacca (Phytolaccaceae) and from one of Bursaria of Pittosporaceae (Daxenbichler 1991). If confirmed, these would constitute novel and remarkable additions to the list of known mustard oil taxa (Ettlinger, 1987; Rodman, 199 1a). Until such confirmation, however, caution is jus- et al., tified. Claims of glucosinolates in Plantaginaceae (Cole, 1976) and in Sterculiaceae (Gill et al., 1984) have been disputed (Larsen et al., 1983; Bjerg et al., 1987, respectively) and remain unconfirmed. None of these pc is res in our study (but see Chase et al., For eight of the nine new sequences, methods of analysis were as follows (for Koeberlinia spinosa methods are described by Price & Palmer, 1993; GenBank accession numbers for the nine new se- quences reported here are listed in Table 2). Total genomic DNA was extracted from fresh or frozen leaves (or green twigs for Koeberlinia) by a mod- ified hot CTAB procedure (Doyle & Doyle, 1987). Polymerase chain reaction (PCR) using commercial "AmpliTaq" enzyme kits (see Erlich, 1989; Wil- liams, 1989) and a Perkin-Elmer Cetus Thermal Cycler were employed to amplify the rbcL DNA, making use of flanking 5' and 3' primers provided by Richard Olmstead. The forward 5' primer ex- tends 26 bases into the presumptive coding region of the gene from the AUG start codon; conse- quently, that 26-mer portion of our new sequences was coded as blank for the computer analyses. The resulting PCR product was cleaned with commer- cial “Gene-Clean”” and aliquots were run on short 1.0% agarose gels to confirm the presence of a single band at ca. 1.4 kilobases when visualized by staining with ethidium bromide under ultraviolet light. Direct dideoxy sequencing from the double- stranded PCR product was done using a modifi- cation of published methods (Korneluk et al., 1985; Zhang et al., 1988; Kusukawa et al., 1990) de- veloped by Elena Conti (Conti et al., 1993). We used commercial “Sequenase Version 2.0" kits with sulfur-35 radiolabelled dATP in concert with internal forward and reverse primers made avail- able by Gerard Zurawski. About half the complete sequence from forward primers was confirmed by using reverse primers. Products of the sequencing Volume 80, Number 3 1993 Rodman 689 oe А Mustard Oil Plants TABLE 2. Sources of rbcL nucleotide sequences for mustard oil taxa (marked by an asterisk: Ettlinger, 1987; Rodman, 1991a) and putative relatives, arranged ac- cording to Cronquist's (1981) system (except for Koe- berliniaceae). I. Caryophyllidae Caryophyllales (Centrospermae) Amaranthaceae, Amaranthus hypochondriacus, Mich- alowski et al., 1990 Chenopodiaceae, Spinacia oleracea, Zurawski et al., | Plumbaginales Plumbaginaceae, Plumbago capensis, Слаппаз et al., 1992. II. Dilleniidae Batales Bataceae, Rettig, unpublished. *Batis maritima, James Manhart & Jeff Capparales B rassicaceae, *Brassica juncea, Jackie Nugent, un- published. Capparaceae, *Capparis hastata, Пих 30315 (WIS), GenBank M95754; *Cleome hasleriana, Al-Sheh- baz s.n. (MO), GenBank M95755. Koeberliniaceae, мена spinosa, Al-Shehbaz s.n. MO), GenBank L1460 Moringaceae, * Moringa TN Iltis 30501 (WIS), GenBank L11359. Resedaceae, * Reseda alba, commercial seed, GenBank M95756 Tovariaceae, *Tovaria pendula, D. Smith & J. F. Smith 1831 (WIS), GenBank M95758. Malvales Malvaceae, Gossypium hirsutum, Gulov et al., 1990. Violales Caricaceae, *Carica papaya, WIS Botanical Garden, nk M95671 Passifloraceae, Passiflora quadrangulata, Chase et al., 1993 III. Rosidae Euphorbiales Euphorbiaceae, *Drypetes roxburghii, Lyon Arbore- tum, Hawaii, GenBank M95757; Euphorbia po- lychroma, Chase et al., 1993 Geraniales Balsaminaceae, Impatiens biflora, Chase et al., 1993. Geraniaceae, Geranium grandiflorum, Pelargonium hortorum, Price & Palmer, 3. Limnanthaceae, * Floerkea proserpinacoides, Chase et al., 1993; *Limnanthes douglasii, Price & Palmer, 1993. Oxalidaceae, Oxalis dillenii, Price & Palmer, 1993. Tropaeolaceae, *Tropaeolum majus, Price & Palmer, Myrta dac РИИ Quisqualis indica, Chase et al., 1993. TABLE 2. Continued. Lythraceae, Lythrum subsessilifolium, Conti et al., Onagraceae, Clarkia xantiana, Conti et al., 1993. Polygalales Vochysiaceae, Qualea sp.?, Chase et al., 1993. Rosales Saxifragaceae, Heuchera micrantha, Penthorum se- oides, Soltis et al., Sapindales Aceraceae, Acer saccharum, Chase et al., Akaniaceae, *Akania bidwillii, Gadek et al., Bretschneideraceae, е и Lin 726 (WIS), GenBank M95 Burseraceae, Bursera inaguensis, d et al., Simaroubaceae, Ailanthus altissima, Chase et al., 1993. 1992. Leu & 993. | 993. reactions were run on 6% polyacrylamide gels ог on commercial 5% “Long Ranger" gels, and the results were visualized by autoradiography. Se- quences were read and base changes coded by hand against a tobacco rbcL template. Sequences were entered into the University of Wisconsin Genetics Computer Group package of programs (Devereux et al., 1984) to check for protein translation and internal restriction enzyme sites as a test of se- quence authenticity. For phylogenetic analysis of the sequence data, we created a matrix of 33 taxa and 1,428 char- acters, choosing to terminate the rbcL sequence at position 1428 in order to maintain conformity with the general analysis (Chase et al., 1993). Also, sequences beyond position ca. 1430 were variable and difficult to align, whereas alignment was easily done by inspection for the rest of the gene. Phylogenetic relationships were inferred using a maximum parsimony strategy implemented with Swofford’s PAUP computer programs (Swofford, 1991; see Sanderson, 1 ); these generate “Fitch” parsimony trees with unordered nucleotide data (Fitch, 1971). Results are thus directly com- parable to those derived from the predominantly morphological data set analyzed with PAUP in the recent cladistic study (Rodman, 1991b). We used the TBR and MULPARS options in our search procedure. We are aware of controversies over the best methods of estimating or inferring phyloge- netic relationships (Sober, 1988; Swofford & Ol- sen, ; nevertheless, we consider the maxi- mum parsimony approach to be reasonable and defensible (see Hillis et al., 1992) 13 mustard oil taxa are included in our data set, usually one genus per family. Both genera of Lim- nanthaceae, Floerkea and Limnanthes, are in- . Sequences from 690 Annals of the Missouri Botanical Garden cluded; we are not aware of any published confir- mation of glucosinolates in Floerkea, but the fresh leaves produce a mild “horseradish” taste when chewed from which we infer their presence. To these we added sequences from 20 putative rela- tives or outgroups. With the many potential out- groups treated here, we adopted a strategy of "'si- multaneous resolution" (Maddison et al., 1984); this avoids the assumption that mustard oil taxa must form a monophyletic group. Three consid- erations affected our choices of outgroup taxa. (1) We wanted to evaluate traditional taxonomic re- lationships for various mustard oil plants (e.g., Ва- tis allied with centrosperms; Bretschneidera allied with Sapindales; Caricaceae allied with Violales such as Passifloraceae; Limnanthaceae and Tropaeola- ceae allied with Geraniaceae and/or Balsamina- ceae; Drypetes allied with Euphorbiaceae). (2) We wished to take advantage of results from prior studies (Olmstead et al., 1992) and from the gen- eral rbcL analysis (Chase et al., 1993) that sug- gested malvalean and sapindalean affinities with Capparales. (3) We wanted to use several outgroup taxa in order to avoid artifactual linkages induced by distant or isolated taxa, the "Лопе edges [branch- es] attract" effect (Hendy & Penny, 1989). The robustness of resulting trees was evaluated in various ways: bootstrapping (Felsenstein, 1985; Sanderson, 1989) with PAUP's heuristic search algorithm; “decay analysis" (Bremer, 1988; for an example see Smith & Sytsma, 1990) to explore the topology of trees one to several steps longer than the most parsimonious trees found under the heuristic search procedure; and a branch-and-bound exhaustive search with a reduced set of 14 taxa. A codon-position weighting scheme, with weights and stepmatrix values calculated by Albert et al. (1993), was also used with the PAUP-heuristic search. The scheme differentially weights changes at first, second, and third codon positions, relying upon ratios derived empirically from a large data set of protein-coding genes. When consensus trees were generated, the strict (“‘Nelson’’) consensus option was used (Swofford, 1985: CONTREE doc- umentation) unless otherwise noted. RESULTS With the PAUP-heuristic search procedure and with Heuchera and Penthorum (Saxifragaceae) se- lected as outgroups, we obtained two equally par- simonious trees with the unweighted data, of length = 1,604 steps or site changes and with a consis- tency index (Kluge & Farris, 1969) of 0.39 (ex- cluding unique substitutions). These are reasonable outgroup choices in light of results from the general rbcL study (Chase et al., 1993) that place them in a clade collateral to all mustard oil taxa and their near relatives. Figure ІА shows the two equal- ly parsimonious trees. Numbers of nucleotide sub- stitutions (branch lengths) are indicated for the topology that matches the single tree obtained by the weighted analysis; with the better f-ratio (= 0.2995; see Swofford & Maddison, 1987) this is our preferred tree of the two. All the mustard oil taxa form a monophyletic lineage excepting only Drypetes, which links near Euphorbia and Pas- siflora. The sister group to the major mustard oil clade is an alliance of malvalean (Gossypium) and sapindalean taxa (Acer, Ailanthus, Bursera) as discovered in recent analyses (Olmstead et al., 1992; Chase et al., 1993). There are 522 variable site positions for these taxa, of which 351 (or 25% of the 1,428 sites) provide cladistically informative changes. Several results appear noteworthy, the first two being in accord with traditional taxonomic judg- ments, but all the rest contrary to traditional align- ments. (1) Brassicaceae, Capparaceae, Reseda- ceae, and Tovariaceae are closely related and form what can be called “соге Capparales.” (2) Dry- petes links near Euphorbia and Passiflora and not close to other mustard oil taxa. (3) Limnanthaceae fall within the major glucosinolate clade and do not link with either Balsaminaceae (represented by /m- patiens) or Geraniaceae (represented by sister taxa Geranium + Pelargonium; see Price & Palmer, 1993). (4) Batis falls within the major glucosinolate clade and does not link with either Centrospermae (represented by sister taxa Amaranthus + Spi- nacia) or sapindalean taxa. (5) Bretschneidera links near the base of the glucosinolate clade and not with Sapindales (represented by Acer, Ailan- thus, and Bursera). (6) Carica links near the base of the glucosinolate clade and not with Violales (represented by Passiflora). (7) Tropaeolum also links near the base of the glucosinolate clade and not with Geraniaceae or Oxalidaceae (represented by Oxalis; see Price & Palmer, 1993). All but one of these findings are corroborated by the recent cladistic analysis based on a predominantly mor- phological data set (Rodman, 199 1b). To assess the robustness of these findings, we ran the PAUP-bootstrap procedure (Swofford, 1991) and performed decay analyses (Donoghue et al., 1992) to accumulate trees one to five steps longer than the two parsimonious trees originally obtained. The decay values were obtained by hold- ing all trees of 1,609 steps and fewer using PAUP; these trees were then sequentially filtered one step Volume 80, Number 3 1993 Rodman et al. Mo of Mustard Oil Plants shorter at a time, and with each filter a strict consensus tree was generated. Branches that did not collapse at 1,609 steps were scored as greater than +5 steps. Figure 1B presents bootstrap results (as percentages of 100 replicates) and the number of steps (additional nucleotide substitutions) re- quired to collapse each node for our preferred tree. At 1,609 steps (+5 changes) there were 2,319 trees; 1,608 (+4) there were 905; 1,607 (+3) there were 320; 1,606 (+2) there were 98; and at 1,605 steps (+1) there were 21. We did not pursue decay analyses beyond five steps because the computations require too much time. The major mustard oil clade is strongly supported with a boot- strap value of 93% and a requirement for more than five additional steps to dismember the clade. Likewise, Drypetes is strongly allied to Euphorbia and Passiflora in a separate lineage (bootstrap value = 81%, decay index > 5 steps). Taxon sampling in this portion of our tree is sparse, how- ever, and resolution of the phylogenetic relation- ships of Drypetes will likely require sampling even greater than that summarized in the general anal- 1993). Within the major glu- cosinolate clade, strong linkages are shown in the pairing of Floerkea + Limnanthes (100%), Bret- schneidera + Tropaeolum (98%), and in the clade of Limnanthaceae with ysis (Chase et al., and the core Capparales (99%). Within core Cappara- Batis, Koeberlinia, les, rbcL sequence data provide only weak support for resolution of component families, including the union of Brassicaceae with a paraphyletic Cappara- ceae [((Brassica, Cleome), Capparis)]. Linkages among nonglucosinolate taxa (e.g., sapindalean, myrtalean, centrospermous) are generally com- mensurate with results from the general analysis (Chase et al. 1993), but no great emphasis should be placed on our specific results because taxon sampling is sparse. The robustness of the major glucosinolate clade following bootstrapping and decay analysis en- couraged further study of this subset. Designating Gossypium as root (the nearest outgroup from the initial analyses), we ran a 1 4-taxon subset with the branch-and-bound option of PAUP, guarantee finding the most parsimonious tree if the data set is not too large (Swofford, 1991). A single tree of length — claimed to 488 steps and consistency index — 0.59 (excluding autapomorphies) was obtained. The topology was identical to that of the relevant portion of our preferred tree (Fig. 1A) with ingroup relationships fully resolved and with Bretschnei- dera mustard oil taxa. Cladistic relationships among taxa + Tropaeolum collateral to the remaining within the major glucosinolate clade, therefore, are robust in the face of different outgroup as- signments. DISCUSSION Nucleotide sequences from the rbcL gene sup- port recognition of a major glucosinolate clade and of a second lineage restricted to the euphorbiaceous genus Drypetes. The rbcL phylogeny is largely congruent (Fig. 2) with the published cladogram of mustard oil taxa derived from a predominantly morphological character set (Rodman, 1991b, fig. 2). For the core Capparales (Brassicaceae, Cap- paraceae, Resedaceae, and Tovariaceae) and their sister group (Bataceae + Koeberliniaceae) the two trees are perfectly congruent. The rbcL data sug- gest that Capparaceae are paraphyletic, however; and А and Salvadoraceae remain o t the base of the major glucosino- late dade. sister-group relationships differ between the rbcL and morphology trees, with Limnantha- ceae (Floerkea and Limnanthes) presenting the most serious discrepancy. The unexpected position of Limnanthaceae in the rbcL tree is unlikely to reflect misidentifica- tions. Both genera of the family were sequenced (in separate laboratories), the pairing of Floerkea and Limnanthes is strong (only 17 site differences between them, the smallest in our data set), and their position in the major mustard oil clade is robust. Furthermore, an additional molecular char- acter not included in this analysis supports alliance of Limnanthaceae with Capparales. The rbcL gene of Floerkea and Limnanthes and their sister clade (Batis with Koeberlinia + core Capparales) does not terminate at position 1428 (as do most other taxa including the basal glucosinolate taxa Cari- caceae, Bretschneideraceae, Moringaceae, and Tropaeolaceae) but contains two additional codons before the stop signal. The stop codon, seen in most taxa at positions 1426-1428, has been replaced by one coding for aspartic acid in Limnanthaceae and their sister clade, with subsequent valine (al- anine in Tovaria) and stop codons. This structural change in the gene ending is not apparent in our data set (because we chose to terminate sequences at 1428), but it is a synapomorphy reinforcing the inclusion of Limnanthaceae within a distinctive glu- cosinolate subclade. The prior cladistic study (Rod- man, 1991b) was interpreted as providing "sug- gestive but not yet compelling" character evidence for linkage of Limnanthaceae with Balsaminaceae, a traditional taxonomic opinion (Cronquist, 1981, 1988). Bootstrap analysis of the morphology-based result (performed by re-running the data in PAUP- 692 Annals of the Missouri Botanical Garden 2 BRASSICA * CLEOME * CAPPARIS * RESEDA * TOVARIA * KOEBERLINIA H BATIS * за р FLOERKEA* 2 114 LIMNANTHES * CARICA * MORINGA * BRETSCHNEIDERA * TROPAEOLUM * 20 GOSSYPIUM 14 23 _ BURSERA 25 33 AILANTHUS 11 31 met ac CLARKIA = LYTHRUM QUISQUALIS QUALEA PELARGONIUM GERANIUM РА fe DRYPETES * 44 PASSIFLORA EUPHORBIA OXALIS E 22 AMARANTHUS | (__38 SPINACIA 17 | 71 PLUMBAGO 54 IMPATIENS 38 PENTHORUM — Ба _ НЕОСНЕКА FiGURE la. The two dd parsimonious trees for mustard oil taxa (names in boldface marked by asterisks) and putative relatives, of length = 1,604 steps and CI = 0.39, found by analysis of rbcL sequences using the PAUP heuristic search algorithm (ACCTRAN option). The two mustard oil lineages are marked by arrows. The preferred tree, solid lines with branch lengths (number of nucleotide substitutions) above, has the lower f-ratio and is identical in topology with the single tree generated by using the Albert et al. (1993) weighting scheme. BRASSICA * CLEOME * CAPPARIS * RESEDA * TOVARIA * Г— КОЕВЕКИМА 74 L— ватіЅ * pe FLOERKEA * +1 100 hex LIMNANTHES * 51 N »45 >+5 Г—— CARICA * 93 88 » LL— MORINGA * Г— BRETSCHNEIDERA * +5 98 |» L TROPAEOLUM * 76 GOSSYPIUM 1 | 42 BURSERA >+5 54 AILANTHUS +1 ACER CLARKIA LYTHRUM QUISQUALIS " QUALEA | >45 [— PELARGONIUM 100 А GERANIUM / DRYPETES * >+5 49 PASSIFLORA EUPHORBIA OXALIS AMARANTHUS — = . loa) +1 17 SPINACIA PLUMBAGO IMPATIENS [— PENTHORUM 78 LL HEUCHERA FIGURE lb. Bootstrap values (percentage of 100 replicates, below branch) and decay index values (number of dos required to collapse the node, above branch) mapped on the preferred parsimonious tree. Asterisks (and boldface) mark mustard a taxa; arrows indicate the two mustard oil lineages. For two areas with bootstrap values boxed the topology is not in accord with the bootstrap majority-rule consensus tree: in 42% of the replicates myrtalean taxa are collateral to Ll. taxa + Gossypium + sapind to all of these with a bootstrap value of 39%. The other discrepancy involves /mpatiens, which is collateral to the clade of Drypetes— Euphorbia Passiflora in 35% of the bootstrap a ren this quartet is then collateral to centrosperms + Oxalis in 17% of the replicates. 694 Annals of the Missouri Botanical Garden та. Morphology BRASSICACEAE CAPPARACEAE 93 RESEDACEAE m «===» GYROSTEMONACEAE xi T TOVARIACEAE ЧО ен SALVADORACEAE [53] M BATACEAE 84 3 74 L KOEBERLINIACEAE d ES E LIMNANTHACEAE Г MORINGACEAE —34À 7 = 88 L— CARICACEAE _ — 34 10 33 -PENTADIPLANDRACEAE— | 31 qe TROPAEOLACEAE —— en | BRETSCHNEIDERACEAE —] 14 88 AKANIACEAE — 60 Lo 22 DRYPETES FIGURE 2. Comparison of the most parsimonious tree derived from rbcL sequence data (Fig. 1) with that from a prior cladistic study of predominantly morphological characters for the mustard oil taxa (Rodman 1991b, fig. 2). Akania has been incorporated in the rbcL tree from study of Gadek et al. (1992); Gyrostemonaceae, Pentadiplandra, and Salvadoraceae have not been sequenced. Breaks in lines indicate presence of intervening nonglucosinolate taxa. Bootstrap values (percentage of 100 replicates) are indicated for each branch; for the morphology tree, the published data set (Rodman, 1991a) was run in PAUP-version 3.0s to obtain bootstrap values. version 3.05) indicates that the union of Balsami- пасеае + Limnanthaceae is weak, with a value of 23%. That character evidence is now called into question, as indeed it had been earlier by workers troubled with ambiguous homologies in embryology (Dahlgren, 1975b) and floral morphology (Hof- mann & Ludewig, 1985; see Link, 1992). When we force Limnanthaceae into its rbcL-based posi- tion (using the TOPOLOGY option of PAUP: Swof- ford, 1985) and run the predominantly morpho- logical data set, we obtain a tree eight steps longer than the most parsimonious ones found originally (Rodman, 1991b). Reduced homoplasy in phyto- chemical features (e.g., erucic acid and glucosi- nolate biosynthesis) is countered by increased ho- moplasy for morphological and chromosomal characters. The complementary test, running the rbcL data with Limnanthaceae forced into its mor- phology-based position as sister group to /mpatiens (Balsaminaceae), results in an increase of 34 nu- cleotide changes over the 1,604-step trees. For rbcL data, potentially confounding events besides experimental error include introgression, lineage sorting, and mistaken orthology (Rieseberg & Sol- tis, 1991; Doyle, 1992). We consider all of these unlikely in the present case because of the taxo- nomic depth involved and the effectively single- copy evolution of the chloroplast genome (Palmer et al., 1988). Additional molecular data (especially nuclear genes) and evaluation of the chromosomal, morphological, and palynological homologies should reconcile the matter. At present we favor the rbcL Volume 80, Number 3 1993 Rodman et al. Monophyly of Mustard Oil Plants result because of the strength of its internal support compared to the weaker case in the morphology- based tree. Character support for a major glucosinolate lin- eage had been judged weak in the prior cladistic study (Rodman, 1991b, table 1). That clade en- compassed all the mustard oil taxa (plus Koeber- linia) except Drypetes, linked with Euphorbiaceae, and Limnanthaceae, linked with Balsaminaceae (Fig. 2). Only four homoplastic transformations were found to distinguish the stem lineage of this large mustard oil assemblage (presence of phenylalanine- derived glucosinolates, loss of anthocyanins, loss of methylated flavonols, and presence of myrosin guard cells). Bootstrap analysis reveals very weak support from the morphology data set, only 10% compared to 93% in the rbcL data. The sequence data thus provide independent and robust corrob- oration of a major mustard oil lineage. The results invite further scrutiny of the character homologies discussed by Rodman (199 1a), particularly the an- atomical and embryological features noted by Dahl- gren (1975b) In combination, the molecular and morpholog- ical results for the mustard oil taxa strongly support Dahlgren's (1975a, 1977) early classification of a greatly expanded Capparales. The inclusion of Ak- ania (Gadek et al., 1992) and Bretschneidera, of Batis (an exoneration of Pulle’s iconoclastic opin- ion: Eckardt, 1959), and of Tropaeolum departs radically from traditional taxonomic schemes and provides further evidence that Cronquist's (1981, 1988) subclasses Dilleniidae and Rosidae are not natural groups. Our results further suggest that Caricaceae and Limnanthaceae join the major glu- cosinolate clade. That would leave but a single outlying glucosinolate taxon. A distant Drypetes, phylogenetically linked near Euphorbiaceae, may thus constitute a remarkable example of molecular convergence. Both myrosin cells and glucosinolates of several biosynthetic types co-occur in Drypetes, as they do throughout the major mustard oil clade (Koeberlinia provides the single, fascinating ex- ception). This conjoint anatomical-biochemical sys- tem remains an evolutionary mystery, albeit a mys- tery now limited to two branches of the angiosperm ree. The deeply embedded position of Koeberlinia well within the major glucosinolate lineage (Fig. 1) further reinforces the idea (Rodman, 1981, 1991a, b) that it alone among members of this clade has lost or repressed the biochemical capacity to syn- thesize glucosinolates. Consistent with this inter- pretation, Koeberlinia has been reported to possess myrosin cells (Gibson, 1979). The rbcL phylogeny creates novel linkages at the base of the major mustard oil clade, affiliations with little support in traditional taxonomic treat- ments of these plants. Linkages among these same taxa differ in the morphology-based tree, but boot- strap values are weaker (Fig. 2) and character support was judged weak as well (Rodman, 1991b, table 1). Caricaceae, for example, paired with the unsequenced Pentadiplandra, are basal in the morphology tree but with a bootstrap value of only 31%, whereas Carica and Moringa unite in the rbcL tree with robust support from bootstrap and decay analyses (Fig. 1B). We note, moreover, that our unpublished restriction-enzyme maps of chlo- roplast DNA demonstrate remarkable phenetic similarity between Carica and Moringa, especially in the conservative inverted-repeat region of the 1992) and Bret- schneidera, traditionally allied with Sapindales, are molecule. Akania (Gadek et al., collateral with Tropaeolum in a strongly supported branch at the base of the major glucosinolate clade in our rbcL tree; hence, they are cladistically close to malvalean and sapindalean sister groups. Several phytochemists (see Kjaer, 1973) have proposed a biosynthetic linkage between the sulfur containing glucosinolates and cyanogenic glycosides; the wide- spread occurrence of cyanogens in Sapindales (Saupe, 1981) and of sulfated flavonoids in Mal- vales (Young, 1981) may prove more than coin- cidental. Likewise, the report of Akania, a genus currently restricted to Australia, in the Paleocene 1988) reduces the implausibility of a cladistic relationship with flora of Patagonia (Gandolfo et al., Andean Tropaeolaceae. Certainly these taxa are dissimilar in habit (Rodman, 199142), but they may come to be seen as phenetically distant remnants of an early diversification of glucosinolate lineages. Sequence data from Gyrostemonaceae, Pentadi- plandra, and Salvadoraceae are needed to com- plete the survey of all glucosinolate taxa, to resolve linkages among members of the major mustard oil clade, and to evaluate the double-origin hvpothesis for glucosinolate biosynthesis proposed here. Although much has been written about possible conflicts between molecular and morphological data in the study of phylogeny (Sytsma & Smith, 1988; Sytsma, 1990), this important theoretical problem is often obscured by a poor choice of examples employing incommensurable methodologies (е.г., Vickery & Wullstein, 1987; see Donoghue & San- derson, 1992). Often, intuitively assessed morpho- logical findings are compared to taxometrically an- alyzed molecular data, usually to the detriment of the former. In the current study, in which we have applied similar cladistic methods to both molecular 696 Annals of the Missouri Botanical Garden and morphological data, the results prove congru- ent in many areas of the topology and comple- mentary in the strength they bring to particular lineages (Fig. 2). The morphological data provide stronger resolution of some terminal groups (e.g., within core Capparales) whereas rbcL data provide stronger resolution of relationships among basal glucosinolate clades, as judged by comparison of bootstrap values. In general, the strength of a data set of either type will depend on the number of robust characters that can be brought to bear at a particular taxonomic level. A major benefit of gene sequence data lies in the relative ease with which homology determinations can be made for numerous characters, compared to morphological data. It is for this reason that we favor the position of Limnanthaceae in the rbcL tree rather than the Mo etit based result. nally, the results provide a commentary on the third assumption underlying our analysis: one or two samples can represent large families in rbcL phylogenies. Sequences for Floerkea and Lim- nanthes would seem reassuring in this context; published comparisons of several congeneric and confamilial taxa are similarly reassuring (Doebley 1990; Hudson et al., 1990). But these results only beg the question: Is the family being repre- et al., sented a natural unit in the first place? The “Cap- paraceae" prove instructive. Торага, often treated 1992), proves to be basal to a clade in which Brassica as a subfamily of capers (Thorne, 19 and Reseda are nested whether molecules or mor- phologies are studied. The nonglucosinolate Koe- berlinia, often treated with capers as a subfamily also (Thorne, 1981, 1992), affiliates with Batis and Salvadoraceae (Rodman, 1991b), and rbcL data place it with Batis. Cleome links with Bras- sica, and these affiliate with Capparis, suggesting paraphyly for the Capparaceae even when pared of Koeberliniaceae and Tovariaceae. A more gen- eral message emerges. Whether one or few samples prove representative in rbcL surveys is contingent upon the monophyly of the terminal taxa. For many angiosperm families that cannot be taken for grant- ed (Hershkovitz, 1989; Rodman, 1990; Soltis et al., 1990; Cantino, 1992; Donoghue et al., 1992). LITERATURE CITED ALBERT, V. А., M. W. СНАЗЕ & B. D. MISHLER. 1993. Character-state weighting for cladistic analysis of pro- A sequences. Ann. Missouri Bot. Gard. AL -SHEHBAZ, I. А: 1973. The biosystematics of the genus Thelypodium ы Contr. Gray Herb. Harvard Univ. 204: 3-148. & K. I. AL- SHAMMARY. 1987. Distribution = tain Middle-eastern Cruciferae. Biochem. Syst. Ecol. -569. Bu H. S., C. Burks, ~ йы, FickETT, W. B. Goa F. I. Lew ATTER, W. Р.В кыо пене“ C.-S. Tunc. 1986. The | GenBank k® d ен se- quence databank. 5 Acids Res. 14: 1- BJERG, B., С. R. FEN ‚ А. SPINKS & Н. SORENSEN. 1987. Failure to М кые и іп сосоа. е 26: 567-568. . P. THANGSTAD, O. A. HAUGEN & T. 1991. Fate of myrosin cells: Character- ization of monoclonal antibodies against myrosinase. J. Exper. Bot. 42: 1541-1549. Bourrorp, D. E., A. Kijarm, J. О. Mapsen & T. 1989. Glucosinolates in Bretschnei- deraceae. Biochem. Syst. Ecol. 17: 375-380 BREMER, K. The limits of amino acid sequence data 1 in angiosperm [ Evo- lution 42: 795-803. CANTINO, P. D. 1992. Evidence for of the Labiatae. Ann. Missouri ned Gard. 79: 361- 9 379. CHast, M. W., D. E. Sorris, R. С. OLMSTEAD, D. MORGAN, Н. Les, D. B. MISHLER, М. В. Du ALL, R. A. PRICE, Н. С. Hits, Y.-L. Ош, K. A. Кож, J. Н. PLUNK ETT, P. 5. Sol bn 5. po EN, S. E. L LIAMS, P. A. GADEK, Ст QuiNN, L. E. EGUIARTE, E. GOLENB BERG, С. Н. LEARN, JR., S. GRAHAM, . C. Н. BARRETT, S. DAYANANDAN & V. А. ALBERT. 1993. Phylogenetics of seed plants: An analysis of es from the plastid gene rbcL. Ann. Missouri Bot. Gard. 80: 528-580. CLEGG, M. T. & С. ZuRAWsKI. 1992. Chloroplast DNA and the study ot es phylogeny: Present status and future prospec 1-13 in P. S. Soltis, D. E. Soltis & J. J. a eee и Systematics of Plants. Chapman and Hall, New York. Cote, R. A. 1976. Isothiocyanates, nitriles and thio- cyanates as products of autolysis of glucosinolates in Cruciferae. Phytochemistry 15: 759-762. 1993. Tribal relatio nships in Onagraceae: Implications from rbcL sequence data. Ann. Missouri Bot. Gard. 80: 672- 685. CRONQUIST, A. An Integrated System of Classi- fication of Flowering Plants. Columbia Univ. Press, New Yor The Evolution and Classification of Flowering Plants, 2nd ed. New York Botanical Gar- den, New York. DAHLGREN, К. 1975a. : system of classification of the sed to demonstrate the distri- bution of ае pn Not. 128: 119-147 The distribution of characters within an angiosperm system. I. Some embryological char- m ie Not. 128: 181-197. A commentary on a diagrammatic pre- пи ee the angiosperms in relation to the dis- tribution of бош, states. Pl. Syst. Evol., Suppl. 83. Volume 80, Number 3 1993 Rodman e 697 MS к: Mustard Oil Plants А revised system of classification of 1-124. 980. the. angiosperms. Bot. J. Linn. Soc. 8 1983. General aspects of angiosperm evolu- Hon and macrosystematics. Nordic J. Bot. 3: 119- 149. , 5. ROSENDAL-JENSEN & B. J. NIELSEN. 1981. А revised classification of the angiosperms with com- and other S. Seigler (editors), Phytochemistry and Angiosperm ylogeny. Praeger, New York DAXENBICHLER, M. E., G. F. SPENCER, D. G. CARLSON, С. B. Козе, A. M. BRINKER & R. С. PowELL. 199]. Glucosinolate composition of seeds from 297 species of wild plants. о 30: 2623-2638. DEVEREUX, J., P. НАЕ O. SMITHIES. 1984. A comprehensive set » vices onal programs for the VAX. Nucleic Acids Res. 12 7-395. DoEBLEY, J., M. DURBIN, E. M. GOLENBERG >, M. T. CLEGG & Evolutionary apes of the large subunit of carboxylase (rbcL) nucleotide se- grasses (Gramineae). Evolution DoNocHUE, M. . SANDERSON. 1992. The suitability of molecular and morphological evidence in reconstructing plant шү. Рр. 340-368 in P. S. Soltis, D. E. Soltis & J. J. Doyle (editors), Molecular Systematics of Plants. Chapman and Hall, ork. — —-, R. С. OLMSTEAD, J. F. SMITH & J. D. PALMER. 1992. Phylogenetic relationships of Dipsacales E on rbcL sequences. Ann. Missouri Bot. Gard. -345. роте, J. J. 1992. Gene trees and species trees: Mo- cad systematics as one-character taxonomy. Syst. 44-16 Bot. 17 j. L. DoYLE. 1987. A rapid DNA isolation оока for т е of fresh leaf tissue. Phytochem. Bull. 15, ECKARDT, T. 1959 ns None M von Batis P. Br. Ber. Deutsch. Bot. Ges. 72: 411-418. Екисн, Н. A. (editor). 1989. PCR Technology. Prin- ciples and Applications for DNA Amplification. Stock- ton Press, New York. ETTLINGER, M. G. 1987. Systematic distribution and biochemical properties of glucosinolates (mustard oil glucosides). P. 280 in XIV Int. Bot. Congr. Abstracts, егип. ‚ KJAER. 1968. Sulfur тк ud in plants. Recent Adv. Phytochem. 1: 59- Parana J. 1985. Confidence a on phylogenies: An approach using the ae Evolution 39: 783- FENWICK, G. R., R. K. HEANEY & W. J. MULLIN. 1983. Glucosinolates and their breakdown products in food and food plants. CRC Crit. Rev. Food Sci. Nutr. 18: -201. Toward defining the course of m change for a specific tre 406-416. N, J. E. RopMan, К. С. KAROL, . CONTI, R. A. o & E. S. FERNANDO. 1992. Гов of the Australian endemic Akaniaceae: New evidence from rbcL sequences. Austral. Syst. Bot. 5: 717-724. GANDOLFO, M. A., M. C. DIBBERN & E. J. ROMERO. 1988. Akania patagonica n. sp. and additional material Akania americana Romero & Hickey (Akan aceae), from Paleocene ars of Patagonia. Bull. Torrey Bot. Club 115: 83-88. Стлммаѕ, D. E. & D. J. p FORD. 1986. Biochemical 1 j . K. lutionary в Vol. 20. Plenum Press, New SKI, G. LEARN & M. T. CLEcc. 1992 Evolutionary relationships of the 716" based n comparative rbcL sequences. Syst. Bot GIBSON, x C. 1979. Sey of Koeberlinia and Cano- tia revisited. Madroñ 1-12, gem & M. Moreau. 1984. Volatile components of cocoa with particular refer- ence to glucosinolate products. Phytochemistry 23: eas к a | C. C. S. CHAPPLE, S. ROTHWELL, I. ТОВЕК · ELS. 1988. Allylglucosinolate biosyn- D sis in Brassica carinata. Phytochemistry 27: 1348. Ioue" GILL, K. A. ALIEv, V. 1990. Мона sequence of the Шок! gene er from Gossyp- ium hirsutum. Nucleic Acids Res. 85 Henpy, M. D. € D. Penny. 1989 | ГОИ? for the quantitative study of evolutionary trees. Syst. Zool. 38: 297-309. HERSHKOVITZ, Phylogenetic studies in Centrospermae: A brief appraisal. Taxon 38: 602- 10. . N. UrMasov, Huus, D. M., J. J. Butt, M. E. Wuire, M. R. Варсетт & I. J. MOLINEUX. 1992. Experimental phyloge- ipa Mai of a known phylogeny. Science 259: 592. HOFMANN, É f J. LubEwic. 1985. Morphologie und systematische stellung von Limnanthes dou glasii R. WHITFELD. 1990. Comparisons pe rbeL enes Ps as large subunit of ribulose-bisphosphate carboxyl- e from closely pon PR and C, plant species. J. Biol. Chem. 265: mE L. B. po! cells and dilated cis- m in the order 977. Protein- accumulating cells and dilated cisternae of the en- doplasmic reticulum in three glucosinolate-conta аш са Armoracia, Capparis, Do dE Planta 137 =2 Км, x zu R. K. JANSEN, R. S. WaLLacE, H. J. MICHAELS D. Poste 71992. Phylogenetic implications rbcL sequence variation in o Asteraceae. Апп. MES Bot. Gard. 79: 428-4 The natural i N of glucosi- (editors), Chemistry in Botanical E e Ac- ademic Press, New York. KLUGE, A. G. & J. S. FARRIS. 1969. Quantitative phy- letics and the evolution of Anurans. Syst. Zool. ез2, KORNELUK, R. G., F. Quan & R. А. GRAVEL. 1985. 698 Annals of the Missouri Botanical Garden Rapid and reliable dideoxy Mu of double- stranded DNA. Gene 40: 317-323. KRUCKEBERG, А. R., J. Е. Кормах & R. D. Wor- ; Natural hybridization between carinatus (Crucif- Kusukawa, N., T. UEMORI, K. Asapa & I. Като. 1990. Rapid and reliable protocol for direct sequencing of material amplified by t E ыш chain reaction. BioTechniques 9: LARSEN, L. M., O. Or. det E SORENSEN. 1983. Failure to detect glucosinolates in Plantago species. Phy- tochemistry 22: 2314-2315 Les, D. H., D. K. Garvin & С. Е. WiwrEE. 1991. Molecular оноу history of ancient aquatic angiosperms. Proc. Natl. Acad. U.S.A. 88: 10119- 10123. LiNK, D. А. 1992. The floral nectaries in the Limnan- 43. om p giucosinolate i in 1 relation + to habitat and insect bivory for the native crucifer Cardamine cordifolia. Biochem. 2 Ecol. 11: 199-207. Mappison, W. P., M. J. DONOGHUE & D. R. Mappison. 984. Outgroup analysis and parsimony. Syst. Zool. 3-103. H. J. Вонхект, D. F. Kiessic & EC Nucleotide sequence of rbcL from оин hypochondriacus. Nucleic Acids MICHALOWSKI, C. J. O. BERRY Res. 18 OLMSTEAD, R. G., Н. J. Маснак15, К. M. Scorr & J. D. PALMER. 1992. Monophyly of the Asteridae and identification of their major lineages inferred from A sequences of rbcL. Ann. Missouri Bot. Gard. 79: 249-265 PALMER, J. D., R. K. Jansen, H. J. MICHAELS, M. W. HASE & J. R. Мамнакт. 1988. Chloroplast DNA variation and plant phylogeny. Ann. Missouri Bot. Gard. 75: 1180-1206. PRICE, D. PALMER. 1993. Phylogenetic relationships of the Geraniaceae and Geraniales from rbcL sequence comparisons. Ann. Missouri Bot. Gard. 80: mr Rerric, J. H., Н. D. Witson & J. К. MANHART. 1992. Phylogeny of the А sequence data. Taxon 41: -209. RIESEBERG, L. H. & D. " Sovris. 1991. Phylogenetic consequences of Seri gene flow in plants. Evol. Trends Pl. 5: 65- RITLAND, K. & M. T. CLEGG. Dux Evolutionary anal- ysis of plant DNA sequences. Amer. Naturalist 130: RopMan, J. E. 1974. Systematics and evolution of the genus Cakile Lo Contr. Gray Herb. Har Td Univ. 205: 3-14 76. Dieci ion and migration of Cakile (Cruciferae): Seed glucosinolate evidence. Syst. Bo 1: 137-148. 1978. Glucosinolates: Methods of analysis and some e chemosystematic problems. Phytochem. Bull. 11: 6-3 Population variation and hybridization in sea-rockets (Cakile, ipsa Seed = late characters. Amer. J. Bot. 67: 1145-1159. Divergence, convergence, and paral- lelism in phytochemical ан The glucosino- late-myrosinase system. Pp. 43-79 in D. A. Young . S. Seigler (editors), исо апа Ап- giosperm Phylogeny. Praeger, New Yo in —— ———. 1986. Introduction, establishm and re- placement of sea-rockets (Cakile, a in Aus- tralia. : Biogeogr. 13: 159-171. 90. Centrospermae revisited, Part I. Taxon 93. . A taxonomic analysis of ага producing plants, Part 1: Phenetics. Syst. Bot. 16: 598-618. . 19915. A taxonomic analysis of p producing plants, Part 2: Cladistics. Syst. Bot. 16: 619-629. . S. CHEw. 1980. Phytochemical corre- lates of herbiv vory in a community of native and naturalized Cruciferae. Biochem. Syst. Ecol. 8: 43- 50. A. R. KRUCKEBERG & I. A. AL-SHEHBAZ. 1981. Chemotaxonomic diversity and complexity in seed a of rere © нн (Сги- ciferae). Sys 97- _ M. " 89. Co ie ence limits on phylog- s: The bootstrap revisited. Cladistics 5: 113-129. Flexible phylogeny reconstruction: A review of phylogenetic ius packages using par- mony. Syst. Zool. 39: 414-420. Shirt IPE, . S. G. | a Pita and angio- sperm phylogen y. Pp. 80-116 in D. A. Young & D. S. Seigler кога РА Бени and Angio- sperm Po raeger, New Yor Smith, R. L. € К. J. Sytsma. 1990. Evolution of Populus nigra (sect. Aigeiros): Introgressive ћу- bridization and the chloroplast contribution of Pop- ulus alba (sect. Populus). Amer. J. Bot. 77: 1176- 1187. SoBER, E. 1988. Evolution, and Inferenc Massachusetts SoLTIS, D. E., P. s. Sorris, M. T. Стесс & M. DURBIN rbcL sequence divergence and phylogenetic relationships in Saxifragaceae sensu lato. Proc. Natl. 4. d the Past. Parsimony, . MIT Press, Cambridge, Аса ix SWOFFORD, D. L. 1985. PAUP: Phylogenetic cum Using Parsimony, Version 2.4. Illinois Nat. Hist hampaign. 99] AUP: Phylogenetic Analysis im Parsimony, Version 3.0s. Illinois Nat. Hist. Surv. ampaign. & W. P. MappisoN. 1987. Reconstructing ancestral character states under Wagner parsimony. Math. Biosci. 87: 199-229. & G. LSEN. nd Phylogeny reconstruc- tion. Pp. 411-501 in D. M. Hillis & C. Moritz ПЕРИ Molecular Busan Sinauer, Sunder- , Massachusetts. SYTSMA, К. J. DNA of plant phylogeny Trends Ec — & J. Е. S 1988. Surv., c J and rd cns Trees ol. Evol. 5 -110 А and morphology: Comparisons in un Onagraceae. Ann. Missouri Bot. Gard. 75: 1217-1237. THORNE, R. F. 1981. Phytochemistry and angiosperm phylogeny: A summary statement [including a syn- opsis x gi class WC dirige (Annonopsida)]. Pp. 233-2 n D. А. Young & D. S. ie i рај a еда рона апа MA Phyloge e- ger, New York. Volume 80, Number 3 1993 Rodman et al. 699 Monophyly of Mustard Oil Plants ————. 1992. An updated ру елее. classification atthe flowering plants. Aliso 1 389. VICKERY, R. K., JR. & B "НЕТО parison of six approa aches to th ulus sect. Erythranthe Cisl d minds Ds Bot. 12: 339-364 WILLIAMS, J. К. Optimization strategies for the polymerase chain reaction. BioTechniques 7: 762- 69 1987 Пи YOUNG, D. А. 1981. The usefulness of flavonoids in "P angiosperm phylogeny: Some selected examples. P 205-232 in D. A. Young & D. S. Seigler Sethe Phytochemistry and Angiosperm Phylogeny. Prae Yo sa LL, J. BROWSE & C. SOMERVILLE. 1988. Double аа DNA sequencing as а choice for DNA sequencing. Nucleic Acids Res. 16: 1220. ZURAWSKI, G., PERROT, W. Воттомтку & P. The structure of the gene for the large aln of ribulose 1-5-bisphosphate car boxylase from o ee DNA. Nucleic Ac- ids Res. 9: 3251- w E A PARSIMONY ANALYSIS OF THE ASTERIDAE SENSU LATO BASED ОМ rbcL SEQUENCES! Richard G. Olmstead,’ Birgitta Bremer, Kathy M. Scott, and Jeffrey D. Palmer* ABSTRACT А pa rsimony analysis of 156 representative йй patis of the Asteridae sensu lato and 28 outgroup sequences was of conducted using a two-tiered approach. First, an nalysis the entire group, including 105 sequences, examined relationships among major lineages within the inue s.l.; subsequently, several clades within the larger group were p in question. The a strategy P were congrue tb d are designated as follows: (1) Cornales, (2) ericalean clade, (3) Garrya dex (4) Ilex Е (5) Apiales, (6) Dipsacales, (7 : Asterales s.l., (8) Gentianales, (9) Solanales, (10) d and (11) Lamiales s.l. The e vel of these 11 clades and the whole Asteridae s.l. only grouping clade is congruent between e ia is the consisting of c Gentianales, Solanales, Boraginales, and Lamiales i i.e., the Lamiidae of Takhta Most recent constructs of the subclass Asteridae (e.g., Cronquist, 1981; Takhtajan, 1987) represent reduced groups of the former Gamopetalae (de Candolle, 1813), which was characterized primar- ily by fused corollas (see Wagenitz, 1977, 1992, for a more complete review of the development of ideas about the Asteridae). A series of traits inter- feres with the usefulness of the simple dichotomy of fused versus free petals in defining major groups in the “higher” dicots (i.e., Donoghue & Doyle, 1989), including number of stamen whorls and their position relative to corolla lobes (Dahlgren, 1983); embryological characters, such as one versus two integuments and the crassi- the tricolpate clade of versus tenuinucellar condition in the ovule (Dahl- gren, 1975) of secondary chemicals, particularly iridoids (Jen- sen, 1992). A "group" defined on the basis of any one of these traits will be incongruent with a “group” defined on the basis of any other (Wagenitz, 1977). Conventional approaches to classification often ; and, more recently, the distribution proceed by defining groups on the basis of one or a few clearly defined traits, with additional traits used to refine groups and establish subgroups. In contrast, phylogenetic classification relies on the identification of derived traits to discover groups that are hypothesized to be monophyletic. Molec- ular approaches to systematics have become pop- ular, in part, because they usually are well suited to phylogenetic analysis. Another reason for the popularity of molecular systematics is that phylo- genetic hypotheses generated by molecular data generally are independent of the morphological traits in which many evolutionary biologists are inter- ested. With the combination of phylogenetic system- atics and molecular methods, plant systematics has entered a new period of discovery. With respect to the Asteridae, we are not simply testing hy- potheses set forth by previous workers (e.g., Cron- quist, 1981; Dahlgren, 1980; Takhtajan, 1987; Thorne, 1992); instead we are discovering groups, and those groups may or may not reflect traditional classifications. However, we are indebted to pre- ' This research has been supported by NSF grants BSR-8717600 to | өс and BSR-9107827 to R. Olmstead and NFR (Swedish Natural Science Research Swofford and D. Ma of the manuscript, and to the several researchers w K. Kron and M. Chase (Ericales and ibus groups), m and relatives), M. Hedr Council) grant B-BU 148 search buds to Steven Wagstaff for critical reading who have contributed rbcL sequences of the Asteridae s.l., particularly Q.-Y. (Acanthaceae ddison for helpful discussions of PAUP nd M. Chase o B. Bremer. Thanks are extended to D. Xiang, D. Morgan, and D. Soltis (Apiales, Cornales, d H. Michaels (Asterales). ) an ulder, Colorado 80309, U.S.A * Department of E.P.O. Biology, sides of Colorado, Bo | пен of Systematic Botany, Uppsala University, S-752 36 Uppsala, Sweden. y Па ин of Biology, Indiana U niversity, Bloomington, Indiana 47405, U.S ANN. MISSOURI Bor. Garp. 80: 700-722. 1993. Volume 80, Number 3 Olmstead et al. Parsimony Analysis of Asteridae 701 vious workers for providing a framework for sam- pling in molecular studies. It is fortunate that there has been disagreement among conventional angio- sperm phylogeneticists, or else our sampling may have been biased. The possibility of biased sampling combined with the limited body of data yet pro- duced by molecular systematists means that inter- pretations should be made with caution. Sugges- tions made today (e.g., based on rbcL sequences) may be viewed in the not-too-distant future as we The results (1) results congruent with traditional classifications, (2) results view those of 19th-century botanists. reported here fall into three classes: resolving relationships of taxa ambiguously placed in traditional classifications, and (3) novel results wherein the sequence data are in apparent conflict with conventional data. To be treated properly, these novel results should be considered as starting points for further studies, in which both morpho- logical and molecular data are re-examined and supplemented with additional data. Recent efforts at discovering a monophyletic Asteridae (Downie & Palmer, 1992; Hufford, 1992; Olmstead et al., 1992) identified a large clade containing the Asteridae sensu stricto and several elements generally excluded from traditional cir- cumscriptions of the group, including the Apiales, Ericaceae, Fouquieriaceae, Cornaceae, Hydran- geaceae, Loasaceae, and Columelliaceae. Olmstead et al. (1992) referred to this clade as the Asteridae sensu lato and identified four major lineages com- prising the group, with relationships unresolved among them. The four lineages were the (1) eri- calean clade (Ericaceae, Polemoniaceae, Fouquie- riaceae), (2) cornalean clade (Cornaceae, Нудгап- geaceae), (3) Asteridae sensu Takhtajan (1987) plus Apiales (Asterales, Campanulales, Goodeni- ales, Menyanthaceae, Dipsacales, Apiales), and (4) Lamiidae sensu Takhtajan (1987) (Gentianales, So- lanales, Boraginales, Lamiales, and Scrophularia- les). Restriction site analysis of the cpDNA inverted repeat region (Downie & Palmer, 1992) for the Asteridae s.s. and putative relatives concurred in finding a cornalean group, a group consisting of Polemoniaceae and Fouquieriaceae (Ericaceae was not sampled), and the Lamiidae (Takhtajan, 1987). Hufford (1992) identified a large Corniflorae-as- terid group including a cornalean clade, an eri- calean clade (excluding Polemoniaceae), and a con- ventional asterid clade (including Pittosporaceae and Columelliaceae). Current work presented here and in accompa- nying papers represents another step toward dis- covering the best circumscription of the Asteridae sensu lato and groups within it. This series of papers is a fitting follow-up to the “Phylogeny of Asteri- dae" symposium proceedings published in this jour- nal (Ann. Missouri Bot. Gard. 79(2), 1992), con- taining as it does papers on groups belonging to the Asteridae s.s. (Michaels et al., 1993) and pa- pers on groups now included in the Asteridae s.l. (Xiang et al., 1993; Morgan & Soltis, 1993; Kron & Chase, 1993), in addition to this one Asteridae s.l. on the A detailed study examining which sequences be- long in or out of a group as large as the Asteridae s.l. is a formidable task. Olmstead et al. (1992) established a series of goals for their molecular studies of the Asteridae and addressed the first two: (1 (2) to identify major lineages within the Asteridae. — to evaluate the monophyly of the Asteridae and This study addresses the next two goals: (3) to circumscribe ordinal-level groups through greater sampling and (4) to determine relationships among them. We anticipate that this will bring us one step closer to the final goals of a revised classification of the Asteridae and of a better understanding of evolutionary patterns and trends within the group. By focusing on the Asteridae s.l., much greater detail the rbcL data for taxa sug- we examine in gested to belong to the group than is possible in 93, this issue). The coarse nature of the broad-scope analyses of Chase the analyses of Chase et al. et al. (1993) weakens their utility as rigorous hy- potheses of relationship, but they do serve a useful purpose by suggesting sets of taxa that merit in- clusion in more detailed analyses such as this one. The analysis and interpretation of large data sets are complicated by the probable existence of multiple islands of trees representing equally or nearly equally parsimonious solutions (Maddison, 1991). A single parsimony analysis, even one using a detailed search strategy, can identify only one island. Identifying and presenting separate islands of equal parsimony, or even within one or two steps of the shortest trees (especially when these trees may contain 2,000-4,000 steps), provide valuable insight into possible patterns of relationship and are essential when revised classifications or inter- pretations of character evolution are to be based on the outcome of a phylogenetic analysis. An inadequate analysis, in which a search for multiple islands is not carried out, may produce exaggerated resolution, lead to misleading conclusions, and de- ter potential lines of further research MATERIALS AND METHODS The sequences included in the analyses reported Olm- here have been published previously (e.g.. 702 Annals of the Missouri Botanical Garden stead et al., 1992), are from various researchers and included in the seed plant analysis of Chase et al. (1993), or are reported for the first time here. All sequences and their sources are listed in the Appendix at the end of this issue. Some of the sequences were obtained as described previously (Olmstead et al., 1992). The remaining sequences were determined by first PCR-amplifying the entire gene from a total DNA extract by use of primers described in Olmstead et al. (1992). Single-strand- ed DNA for dideoxy sequencing was then obtained by a second round of PCR using 0.01 4l of the original PCR product as template and a 1:100 primer ratio with the excess primer in the same concentration as in the initial PCR run. Nucleotide positions 27-1428 were included in the cladistic analyses, but the actual number of nucleotides varied among species depending on whether sequences were complete for this region (see Appendix at end of issue). Positions 1-26 represent the PCR primer sequence for most se- quences, and after position 1428 sequence align- ment becomes problematic. Several subsets of the entire data set were com- piled to examine the entire Asteridae s.l. generally and several internal clades specifically, as well as to examine the question of sister group to the Asteridae s.l. In all cases, care was taken to include a number of outgroups, which were selected fol- lowing examination of the results of an analysis of 475 rbcL sequences (“Search 1” of Chase et al., 1993) Parsimony analyses were conducted using PAUP 3.0s (Swofford, 1992) on a Mac IIfx. АП analyses were conducted with equal weights assigned to all changes. The general search strategy proceeded as follows (variations are discussed in Results). First, two or more replicates of 200–1,000 searches each were carried out using a random-order, sequential entry with NNI branch swapping (nearest neighbor interchanges), STEEPEST DESCENT on, and with the MULPARS option off. The goal of this phase of the search was to increase the chance of locating the island or islands containing the shortest trees by finding a reasonably short tree to initiate a more detailed search. The shortest tree or trees from each replicate was entered separately as the start- ing point for the second phase search for all equal- length trees reachable from that starting point. The second phase of the search proceeded by using TBR swapping (tree bisection-reconnection) with MULPARS on. Searches on each individual data set, except for the one examining sister group re- lationships to the Asteridae s.l., were pursued until two or more islands of trees were discovered. RESULTS ANALYSIS OF ASTERIDAE S.L. An analysis of 184 rbcL sequences is reported here, including 156 representatives of the Asteri- dae s.l. and 28 outgroup sequences. To examine the entire Asteridae s.l., a set of 105 sequences was compiled. To reduce the complexity of the data set while maintaining the best-possible distri- bution of taxa, several well-supported families with multiple representatives (e.g., Asteraceae, Acan- thaceae, Cornaceae) were thinned from the number of representatives in the analyses of Chase et al. (1993, this issue). Choice of outgroup sequences was guided by the results of “Search I” of Chase et al. (1993). Five sequences (Gunnera, Paeonia, Phoradendron, Schoepfia, Osyris) comprising a clade identified as the sister group to the Asteridae s.l. in that search were included. Six progressively more remote outgroups, Heuchera and Cercidi- phyllum (“rosid IV" clade of Chase et al., 1993), Dianthus (“rosid IT"), Tetracentron (**hamamelid П”), Platanus ("ћататећа Г”), and Caltha (“гап- unculid"), were also included. All outgroups are “higher” dicots (sensu Donoghue & Doyle 1989), to minimize the problem of long outgroup branches (Swofford & Olsen, 1990), and fall outside the broadest possible circumscription of the Asteridae 8.1. based on traditional or molecular evidence. Most of the outgroups belong to the closest sister groups (rosids and caryophyllids) to the Asteridae s.l. as identified by the subsequent and more com- plete “Search II" of Chase et al. (1993 The search for the shortest trees connecting 105 sequences of the Asteridae s.l. and outgroups rep- resents a major computational effort and resulted in the discovery of three islands of trees of equal parsimony. Considering the amount of time and effort expended in acquiring a data set of this size, the data deserve the substantial effort dedicated to their analysis. A more detailed description of the approach used will be described for this data set than for subsequent searches. As with any search where the number of taxa is above the limit of a branch-and-bound search algorithm (i.e., ca. 20 taxa), there is no way to guarantee finding the shortest tree or trees. The search was initiated by performing 200 random-order-entry replicate searches using NNI swapping with MULPARS off (this search required approximately 5 hours and found one tree of length 3,609). The resulting tree was then used as a start for a search using TBR swapping with MULPARS off (26 min., length = 3,605), which was used, in turn, as the start for a search using NNI with Volume 80, Number 3 1993 Olmstead et al. 703 Parsimony Analysis of Asteridae MULPARS on (2 hours, 432 trees of length 3,601). These trees were then used as the starting point for a search using TBR with MULPARS on. This search was terminated prematurely without swap- ping on all trees when the computer RAM buffer reached a limit of 4,600 trees (approximately 4 days, length = 3,600). An attempt to find shorter trees was then made by dividing the tree file into many smaller files containing subsets of the 4,600 trees. TBR swapping was initiated on several Mac II computers simultaneously with MAXTREES set to the number of trees in each subset plus one. This strategy enables TBR swapping to proceed on all 4,600 trees saved, but will not save any more trees. The search for shorter trees from the 4,600 trees of length 3,600 lasted several days with as many as eight computers operating simultaneously. Two separate trees of length 3,600 each led to the same island (Maddison, 1991) of 102 equal-length trees (length — 3,597). The resulting strict consensus tree (Fig. 1A) for island- 102 (terminology of Maddison, 1991) shows a basal split in the Asteridae s.l. with one branch consisting of the Lamiidae and the small Garrya clade and the other consisting of the ericalean М clade, the Cornales (including the Hydrangeaceae), Apiales, Dipsacales, Asterales s.l., and the small Ilex clade. In addition to the basal split into two major lineages, this tree differs from the results of Chase et al. (1993) in showing the Cornales as a more derived branch than the ericalean clade. hile this search was in progress, a second search was initiated on the same set of se- quences by performing 1,000 random-order-entry replicate searches using NNI swapping with MUL- PARS off (26 hours, one tree, length — 3,605). This tree was then input directly into a search using TBR swapping with MULPARS on. This search required approximately 10—12 weeks com- puter time and resulted in the discovery of a second island of length 3,597 containing 2,784 trees (Figs. B, 2). During this search 76,300 trees were found at one step longer, of which 73,500 were swapped before a shorter tree was found. The to- pology of the strict consensus tree (Fig. 1B) for island-2,784 is dramatically different from that derived from island-102 with respect to the rela- tionships among the primary groups identified with- in the Asteridae s.l. These results unite the La- miidae with the Asterales s.l., thus forming a clade equivalent to the Asteridae of Cronquist except for the exclusion of the Dipsacales. The Dipsacales and Apiales together are sister group to the above- mentioned clade, with the small /lex and Garrya clades each forming progressively more basal branches, and the Cornales and ericalean clade producing a basal trichotomy. After the first two islands were discovered using the search strategy outlined above, another ap- proach to finding initial trees, possibly representing separate islands, was tried (suggested by D. Mad- dison, pers. comm.). One hundred replicate ran- dom-order-entry searches were conducted using TBR swapping with MULPARS on, but with the option to save no more than two trees greater than, or equal to, length "X," where "X" is a number small enough to be certain to be shorter than the shortest tree found. This approach enables each random start to be searched much more completely than the approach outlined above. One replicate yielded two trees of length 3,597 and which were topologically distinct from any tree belonging to island-102 or island-2,784. A search for all most parsimonious trees yielded 5,568 trees at that ength. The initial search required 11 days and the search for all parsimonious trees in the island re- quired an additional seven weeks to complete. This island contains exactly twice as many trees as is- land-2,784 and can be accounted for by one ad- ditional polytomy (within the Gentianales). Topo- logically, this island is identical with island-2,784, except that the Boraginales are united with the Gentianales instead of with the Solanales and the relationships among the five sequences represent- ing the Gentianales are different (hence this island is not illustrated). Figure 3 shows the strict con- sensus of all trees discovered in the three islands. Another analysis was performed to explore more fully the putative sister group to the Asteridae s.l. “Search Г” of Chase et al. (1993, this issue) iden- tified the clade mentioned above as sister group, h II" (Chase et al., 1993) identified a small clade of two sequences (Dillenia whereas their “Seare and Vitis) as sister group to the Asteridae s.l. For this analysis, 40 sequences of the Asteridae s.l. were removed and 17 additional dicot sequences were included, giving a total of 82 sequences. This analysis found 30 equal-length trees of length 3,322 (Fig. 4). This search was initiated with 300 random- order-entry replicate searches and was not repli- cated to find alternate islands, but the search strat- egy was equivalent to that used to find one of the shortest islands in other analyses reported here. This result is congruent with that based on the previous analysis in suggesting that the Santalales, Paeonia, and Gunnera do not belong together or with the Asteridae s.l. and more likely belong in the large clade identified as Rosidae s.l. ш the analyses of Chase et al. (1993). The results do not place the Dillenia and Vitis sequences as sister 704 Annals of the Missouri Botanical Garden Nicotiana A русорегвісоп yphyi M. би. [c cc RES po Platan nus Tetracentron Caltha FiGURE 1. Solanales Boraginales Gentianales Lamiales s.l. o] Garrya clade Asterales s.l. Dipsacales Apiales ч — llex clade Cornales ericalean clade Outgroups ; R.I. = ees. уеб Nicotiana, ~~ B bygopersicon Convolvulus оа та REM ago — Heli tropi Eodeon" Gentiana = ford г" oo Solanales Boraginales Gentianales Lamiales s.l Asterales s.l Dipsacales dera Apiales pom llex clade [- исоба Garrya clade um. Cornales ericalean clade Outgroups Platanus Tetracentron Caltha = 0.51). — A. Isla Strict consensus trees for two islands of equal parsimony in the Asteridae s.l. (105 rbcL söquences, including 11 outgroups; length = и Vue C : = tre nd-102. Strict consensus of 102 tr B. Island-2,784. Strict consensus clades that are congruent between the two islands and ашай; 5,568 (not shown) are identified. pedes identify sequences that resolve to a specific clade in one island, but not in Volume 80, Number 3 Olmstead et al. 705 1993 Parsimony Analysis of Asteridae — Pentas Spigelia i ы igitalis Callitriche usțicia . nd ein 22 Nelsonia асап Sesamum \, = Сега Ш е пи МЕ Calli са via gigi Buddleja Probosc SP igustrum Sepp. Ve thus Dasyph а падова JEY Campanula _Symphoricarpes папа Dipsacus ® о З riandrum = Е Jonia Sanicula Bhyllo — === 54 ТОПОН E oi a ја Clavija : Anagallis Diapensia F Sar ocus nilkara Camelia rbutus А Epacris Vagdinium hod. en ron = трайепе ornus fonda «0458 yssa Day ig ia im Recum Xara ieee eria БЫН „(== PAS Dianthus 5 Раеота СЕЕ = Рог епагоп Osyris — у: 20 elragen ron e FicURE 2. Majority rule consensus tree of island-2,784 selected as representative tree to illustrate branch lengths (ACCTRANS optimization). дн d long branches are evident in Pentas (Rubiaceae) and Campanula (Campan- ulaceae); see discussion concerning incongruent results in different analyses containing these sequences (e.g., Pentas in Figs. 3 and 5). Note scale bar equal to 20 inferred nucleotide Aoc idi 706 Annals of the Missouri Botanical Garden и etunia onvolvulus Solanales ja 4 Harah plum E Boraginales 'entas Gentianales Е itriche Lamiidae a eR ton mr Sala Lamiales s.l. allicarpa tricularia Proboscidea strum =o а Базур um Asterales s.l. illarsia : — огокја _ Asteridae s.l. A Perana Pe — Dipsacus Dipsacales | Apiales Sa == Јерма | Пех clade CODE ^ |Garrya clade Lee Alangium Cornales отапа tena Hydrangea = m | Impatiens oon ericalean clade D rent hin pa и. nsia — C Elem rzelia — Es nia а г el ол Outgroups FIGURE 3. Strict consensus tree of 8,454 equal-length trees based on rbcL sequences. Trees from three islands (island-102, island-2,784, and island-5,568) are combined. Eleven terminal clades that are congruent among all three islands are identified. The Lamiidae sensu Takhtajan (1987) is the only group resolved between the level of the 11 congruent clades and the entire Asteridae s.l. Berzelia and Escallonia (indicated with arrows) are not resolved to any of the clades that are congruent between islands. Volume 80, Number 3 Olmstead et al. 707 1993 Parsimony Analysis of Asteridae Nicotiana = Convolvulus Montinia hl Borago үсеп yllum gitalis Callitriche Anagallis, — Polemonium Manilkara соя utus i Rhododendron Asteridae s.l. Enkianthus Cyrilla їоѕругоѕ | = Áctiidia | Sarracenia Fouquieria Valeriana Dipsacus Escallonia Helianthus Dasyphyllum Boopis Villarsia Campanula Viburnum Sambucus Pittosporum Hedera Coriandrum ssa Alangium 13 о s Cercidiphyllum || Hamamelis L poss Heuchera ћеа [o је иппега Pelargonium 5 Paeonia : i A Dillenia Putative sister group of "Search II" (Chase et al. 1993) Tetracentron Mahonia Caltha LL — Platanus г— Magnolia FIGURE 4. Strict consensus of 30 trees based on rbcL sequences resulting from the analysis designed to explore sister-group relationships to the Asteridae s.l. (length = 3,322, СЛ. = 0.25, КЛ. = 0.41). The sister group to the Asteridae s.l. indicated by this analysis is a large clade consisting of parts of the Rosidae and Dilleniidae sensu Cronquist (1981). Putative sister groups suggested by “Search I” and “Search II" of Chase et al. (1993) аге indicated. Typical branch lengths in the vicinity of the node from which the Asteridae s.l. and their sister group diverge are indicated. 708 Annals of the Missouri Botanical Garden group to the Asteridae s.l. either. Instead, these sequences are placed at the base of the Rosidae s.l., suggesting that they represent a clade that diverged close to the divergence of the two larger clades. A search for multiple islands might find equal length trees congruent with the results of "Search П” of Chase et al. (1993) with respect to the placement of these two sequences. ANALYSIS OF SUBCLADES Detailed analyses were conducted on three sub- sets of the asterid rbcL data that were identified as clades by Olmstead et al. (1992) and in “Search I” of Chase et al. (1993). The primary purpose for the subclade analyses is to examine in detail more Ege pn mn clades than is possible in the more d analysis of the Asteridae s.l. Thus, the subclade analyses are viewed in re- lationship to the Asteridae s.l. analysis as that anal- ysis is to the seed plant analyses of Chase et al. (1993). Results from this nested hierarchical ap- proach should be complementary; the local anal- yses better assess relationships within terminal clades and the broader analyses better assess relationships among those clades. Time considerations dictated that selection of these subsets of sequences precede the completion of “Search II" (Chase et al., 1993) and the search for multiple islands conducted on the 105-sequence Asteridae s.l. data set described above. The first subset analyzed is the Lamiidae sensu Takhtajan (1987). The second (hereafter referred to as the Asterales- Dipsacales- Apiales contains the Asterales, Campanulales, Dipsacales, Apiales, and several taxa traditionally assigned to other orders in the Asteridae or Rosidae (sensu Cronquist, 1981). Both “Searches I € I" of Chase et al. (1993) identified a small clade containing Garrya (Garryaceae), Aucuba (Cornaceae), and Eucommia (Eucommiaceae) placed near the point of divergence of the two above-mentioned, larger ~ clades. To examine the possible relationships of these three taxa, they were included in each anal- ysis. The third clade examined in greater detail is one containing representatives of the Ericales, Ebenales, Primulales, and other taxa from the Rosi- dae, Dilleniidae, and Asteridae. The Cornales and Hydrangeaceae represent the only primary group within the Asteridae s.l. not included in a more detailed analysis here, because a detailed analysis of this group appears elsewhere (Xiang et al., 1993). To examine the Lamiidae, a set of 72 sequences was compiled. This set included all of the sequences belonging to the group “Asteridae 1” in “Search II" of Chase et al. (1993), except Vahlia, and included nine additional sequences from taxa tra- ditionally assigned to the orders Gentianales (six) and Scrophulariales (three). Outgroup sequences included Cornus (Cornaceae), Nyssa (Nyssaceae), Arbutus (Ericaceae), Polemonium (Polemoni- aceae), Пех (Aquifoliaceae), Escallonia (Grossu- lariaceae), Hedera (Araliaceae), Viburnum (Cap- rifoliaceae), and Dasyphyllum (Asteraceae). The analysis of the Lamiidae data set (72 sequences) resulted in the discovery of four islands (although more may exist); two islands contain 135 (Fig. 5) and 1,350 trees (Fig. 6A), each of 2,189 steps, whereas the other two are one step longer (see below). Both islands identify as clades the Solanales (e.g., Solanaceae, Convolvulaceae, and Montinia), Boraginales (e.g., Boraginaceae, Hydrophylla- ceae), Lamiales s.l. (e.g., Oleaceae, Bignoniaceae, Labiatae, Verbenaceae, Scrophulariaceae, Acan- thaceae), and Gentianales (e.g., Rubiaceae, Gen- tianaceae, Apocynaceae, Loganiaceae), with a clade comprising the Solanales and Boraginales sister group to the others. The differences between the two islands fall entirely within the Lamiales s.l. clade, where several individual sequences or small groups of sequences occupy very different posi- tions. Much resolution within the Lamiales s.l. is lost in the strict consensus of both islands combined (Fig. 6B). Both islands identify the Garrya—Au- cuba- Eucommia clade as sister group to the La- miidae. Two additional islands (not shown) con- taining 2,160 and 135 trees representing local optima were discovered at one step longer (2,190 steps). These trees are identical to the shorter is- lands with respect to the pattern of relationships among primary lineages within the Lamiidae, but differ in that the Garrya clade is not identified as their sister group. To examine the Asterales-Dipsacales- Apiales, This set in- cluded all of the M rnm belonging to the group "Asteridae IP “Search IP” of Chase et al. (1993), except "d six representatives of the As- a set of 43 sequences was compiled. teraceae were included (rather than 14), and one additional sequence each of Пех and of Apiaceae were added. Garrya, Aucuba, and Eucommia were included. Outgroup sequences included Cornus, Nyssa, Arbutus, Nicotiana (Sola- naceae), Gentiana (Gentianaceae), and Antirrhi- Polemonium, num (Scrophulariaceae). The analysis of this data set resulted in the discovery of two islands con- taining 80 (Fig. 7A) and 196 trees (Fig. 7B) each of 1,368 steps. The smaller island identifies a monophyletic — Asterales- Dipsacales- Apiales, whereas the island of 196 trees (Fig. 7B) is un- resolved with respect to these three orders and Volume 80, Number 3 Olmstead et al. 709 Parsimony Analysis of Asteridae Lycopersicon | Solanaceae E Petuni Solanales p ] Convolvulaceae Мека — Grossulariaceae __ огадо 2с Heliotro ium ] Boraginaceae — го ит Elod а ] Hydrophyllaceae / vid Du ] Scrophulariaceae | Pra dig — Callitrichaceae — Acanthaceae ericana Winton odorata > Acanthaceae JETA Padi da Thunbergia = Понор ЗА — Pis ш | н | Lamiales s.l. —— Labiatae s.l. НЫЯ Pia emon Callicarpa = Pingui : : р Ungueua zi Lentibulariaceae — Te» blis — Byblidaceae —_ — Buddlejaceae na Verbenaceae s.s mum Harpagophytum ] кас | Streptocarpus — Gesneriaceae ro 7 — Peda Buddleja — Buddlejaceae — сар — Bignoniaceae __г— Ligustrum к Ol 1—— Jasminum Pm entiana | сое Gentianaceae s.l. Exacum жа Аросупит = M Apocynaceae s.l. | Kopsia — Gentianales Gelsemium 71 А Сева E Loganiaceae r— Spigelia 7 А LL Strychnos Loganiaceae р Pentas 5 А с 2 — irl. uc — Cornaceae Garrya clade ——— Eu — Eucommiaceae | y Dasyphyllum ib ed Escalloni llex vomitoria Outgroups Cornus florida Nyssa L p Arbutus Polemonium m FIGURE 5. Lamiidae, island-135. Strict consensus of 135 trees based on rbcL sequences with increased taxon sampling in the Lamiidae (length — sac 1. = 0.31, КІ. = 0.57). Family Vin Tue follow Cronquist (1981), the Labiatae (Cantino et al., 1992), and do not imply necessarily that the entire family belongs where these representatives occur. Two cries islands each one step longer place the Garrya et al. clade among the other designated outgroups and not as sister group to the Lamiidae (not shown). 710 Annals of the Missouri Botanical Garden Antirrhinum i Digitalis _|Scrophulariaceae Callitriche — Callitrichaceae E Phaulopsis Buel у Асатћасеае — a a 2 ma 38 DEM his Aphelandra Acanthus pes а a Srepfocamus — Gesneriaceae ——{_ — Buddlejaceae Verbena — Verbenaceae s.s. < Teucriu, Ci e Prostanthera Scutella . = amium Labiatae s.l. Physostegia Pogostemon “| via Callica — — На agophytum — Pedaliaceae Pro : oscidea iaceae [1 Ut Brularia ` ] Lentibulariaceae 2 i — Byblidaceae uddleja — Buddlejaceae Catalpa = Bignoniaceae г Ligustrum Bn и Oleaceae FIGURE 6. outgroups are the same as in 0 —Lr дит _]Scrophulariaceas Callitriche — Callitrichaceae —— = Pedaliaceae esamum . Justicia americana еса ла Acanthaceae ЕЕ Нипуа = Acanthaceae = ma EE учи ahis Acanthaceae Aphe - — ——i Асап! иа и Acanthaceae Thunbergia —Acanthaceae Nelsonia — Acanthaceae —— — — —— Streptocarpus — Gesneriaceae LL QC Nicodemi — Buddlejaceae — Ve —Verbenaceae s.s. Teucrium Clerodendron Prostanthera TEMA Labiatae s.l. Physostegia д у жаша ама alli Callicarpa _ Дерадор!ушт = — Pedaliaceae a m Proboscide Pedaliaceae + секса _}tentouterecoae Byblis Buddleja as rapa — Bignoniaceae Ligus — sine di _JOleaceae Strict consensus trees of the Lamiales s.l. based on rbcL sequences. These trees represent the portion of the Lamiidae and of 1,350 trees (length — 2,189, C.I. — 31, R.I. = 0.57). — B. Strict consensus of 1,485 equal-length trees from island-135 and island-1,350 combined. 5. Family designations as in Figure several other groups (however, the Dipsacales and Apiales group with representatives of the Cornales in many trees). This is similar to the results of the Asteridae s.l. analysis, in which one island identifies this clade (Fig. 1A) and the other does not (Fig. 1 B). Both islands identify the following clades: Dip- sacales, Asterales s.l. (including Menyanthaceae, and Corokia), Apiales (including Pittosporum and Griselinia), and the group con- taining ех, Phyllonoma, and Helwingia. ampanulales, o examine the ericalean clade, a set of 60 sequences was compiled, including all of the se- quences belonging to the group *'Asteridae III" in "Search II” of Chase et al. (1993) except He- liamphora. Eight additional sequences represent- ing the Ericaceae were included (see Kron & Chase, 1993). Nineteen representatives of other clades in the Asteridae s.l. were included as outgroups. The analysis of this data set resulted in the discovery of two islands of trees, one containing 612 trees (Fig. 8B) of 1,889 steps and the other containing 357 trees (Fig. 8C) of 1,890 steps. Both islands identify a monophyletic Ericaceae s.l. (including the Epacridaceae, Empetraceae, and Pyrolaceae), in agreement with Kron & Chase (1993). All mem- bers of this clade belong to Cronquist's Dilleniidae, except /mpatiens, Roridula, and Polemonium (see discussion below). The largest clade that is con- gruent between the two islands is the Епсасеае. The shorter island is congruent in most respects with the results of Kron & Chase (1993). The strict consensus tree of both islands combined in- dicates very little resolution outside the Ericaceae (Fig. 8A). DISCUSSION The goals of a study of this nature (i.e., structing the phylogeny of the Asteridae) are lofty, recon- but accomplishing them is problematic. A major problem with analyzing this and any other large and complex systematic data set is assessing the Volume 80, Number 3 Olmstead et al. 711 1993 Parsimony Analysis of Asteridae 9 mphoncarpos- Caprifoliaceae S mphoricarpos – Caprifoliaceae T elena EA Valerianaceae Valeriana жа г, Vale rianaceae ipsacea i | sacus і А = = Cáprifoliaceae Dipsacales ees = — Cáprifollaceae Dipsacales Caprifoliaceae y Caprifoliaceae Cornaceae riselinia — Cornaceae — | — Pittosporaceae coer awn == Pittosporaceae А еаега А Araliaceae Apiales Aralia Araceae Apiales y pipe Apiaceae E diu Apiaceae Sanicula — Bruniaceae Berzelia — Bruniaceae — Grossulariaceae Escallonia — Grossulariaceae Vemonia Lactuca Asteraceae Helanthus Asteraceae cee h Me as um Z Goodeniaceae |^9!erales cabvola Z Goodeniaceae |Asterales — (сны s.l E — Calyceraceae s.l. ceae Corona Я асеае enyanthes E т ж Vill abet E at Viii ü Campanulaceae Г a pet |] Campanulaceae — Grossulariaceáé | llex деа = Grossulariaceae | " -] Anutotacens Pad | сё уотивла ]Aauioiaceae — | = = pomi ceae — —. Helwingia — Cornaceae „ш ar aceae Garrya Garrya — Garryaceae = — ботасва | T — — p ba | Z Gornaceae caine = Eucommiaceag_| clade Еисотта ^ — Еџсоттіасеае | стаде о Nicotiana == Lamiidae Antirrhinum Lamiidae Antirrhinum Gentiana à mus kousa ] Cornales Outgroups oo kousa ~]Cornales Outgroups lyssa Ме Arbutus оне ] Ericalean clade | EO BED ] Ericalean clade | A FIGURE trees (length = 1,368, СЛ. = 0.39, КЛ. = 0.51). Thi 7.—Strict consensus trees for rbcL sequences representing the e Үт Apiales, А І. = 0.51). № and Dipsacales. — Numbers above branches island-80 and island-196. Family designations follow Cronquist (1981) and do not imply necessarily that aci entire family belongs where these representatives occur reliability of the results. How can one be sure that the results found represent the optimal solution for the data? How can one resolve incongruence be- tween separate analyses conducted on similar data sets (e.g., differences between the analyses of Chase et al., 1993, this issue, and analyses presented here)? Will sampling more taxa, sampling more characters (e.g., nucleotides), or more detailed analyses of a subset of the data provide better results? It is not clear that there is a single best answer to any of these questions and when data of this nature are compiled for the first time for a group of organisms, as these are for the Asteridae (and angiosperms), there is no independent, exter- nal means for evaluating the results. Internal meth- ods for evaluating trees (e.g., bootstrap or decay analysis) are extremely difficult to implement on a large data set. Bearing in mind that there is little in the way of rigorous, quantitative, phylogenetic hypotheses against which to evaluate these results, they should be considered preliminary, awaiting the availability of a substantially greater amount of data derived from other sources. For the results of a cladistic analysis to provide strong support for a hypothesis, alternative hy- potheses must be examined. This requires search- ing for multiple islands with sufficient diligence to reach other islands, if they exist. For one well- known data set of human mitochondrial DNA se- quences (Vigilant et al., 1991), a reanalysis of the data yielded separate islands each having different implications with respect to the geo — origin of modern humans (Maddison et al., 1992). The initial analysis (Vigilant et al., 1991) was unable, by design, to find more than one island and, there- fore, was predestined to yield one conclusion. A total of approximately eight to ten months of computer time was spent on the several analyses reported here. Multiple islands were found when- ever the appropriate search strategy was employed for each data set examined in this study. In all cases only two or three separate initial searches each of 200-2,000 random-order-entry starting points were required to find multiple islands, leaving open the possibility that additional islands, some perhaps shorter, remain to be found. Whereas it 712 Annals of the Missouri Botanical Garden Clavija — Theophrastaceae [ Cla — Theophrastaceae Anagallis — Primulaceae Primulales LT Anagallis — Primulaceae rdisia — ем Ardisi inaceae tyrax Styracacea — Ebenales г Styrax Styracaceae lethra — A NE — Ericales = =. — — Cleth raceae Manilkara Г —— 2€ Сосон ] Sapotaceae Ebenales “L Chrysophyllum Sapotaceae Diapensia — "hti ca — Diapensiales Diapensia _ — Diapensia Polemonium — Polemoniaceae — Solanales L Polemonium — Polemoniaceae Symplocus — zam фонын — Ebenales Symplocus = P stet Diospyros naceae — Ebenales Diospyros = Camellia — Theac Camel ка. Impatiens m Bahamnecina — Geraniales [ /mpatiens — Z Balsumnaceao Fouquieria — Fouquieriaceae — Violales Ро џдијепа — Fouquieriaceae lluna = Ве!ала Ceratiola Erica Rh ododendron | Етота assiope Chamaedaphne bia ja Leucothoe Ericaceae s.l. — Ericales Actinidia — Actinidiaceae Gaultheria Cyrilla — Cyrillaceae Vaccinium p-Roridula Byblidaceae p inlcr С Sarracenia — Sarraceniaceae Pentachondra сород B Dracophyllum [— Сама — Th oe pas A Атада — Primulac Xd Arbutus Ardisia — Кинен Arctostaphylos г- Styrax — Styracaceae Pyrola L- Clethra — Clethraceae Enkianthus Manilkara сероб Actinidia — Actinidiaceae — Theales ЈЕ Chrysophyllum |??2%сеае Cyrilla Cyrillaceae — Ericales Impatiens — Balsaminaceae Roridula — Byblidacea — Rosales Diapensia | — Diapensiaceae Sarracenia — Sarraceniaceae — Nepenthales Polemonium — Polemoniaceae icotiana E Fouquieria — Fouquieriaceae Antirrhinum [ Symplocus — Symplocaceae Gentiana Ł Diospyros — Ebenaceae ineo ДЫШ = rada Garrya Dipsacus Escallonia Ericaceae s.l. Berzelia Outgroups Viburnum riandru ra ittosporu. al iiis ondula | — Byblida Comus ke Sarracenia — 5 Chiococca Antirrhinim Catalpa Ligustrum г Nicotiana Outgroups Hali Џ lelioti opium Garrya FIGURE 10. Мајогиу-гше consensus of seven equal-length trees of the ен based on rbcL sequences and rooted using seven outgro from the Solanales, Boraginales, Lamiales 5.1., an e Garrya clade. This tree is decay indices below internodes subtending clades. A decay index of zero indicates the node collapses in the strict consensus of the seven trees Apocynaceae s.l, Rubiaceae, and two separate groups of Loganiaceae (Се/ѕетіит – Моѕіиеа and Spigelia—Strychnos), but provide little support for any internal groupings. (9) Solanales. This clade is comprised of the Convolvulaceae, Solanaceae, and Montinia (Gros- sulariaceae). Comparable groups exist in the clas- sifications of Dahlgren (1980) and Thorne (1992), except for the inclusion of the Polemoniaceae by Dahlgren (1980). Both Takhtajan (1987) and Thorne (1992) included the Polemoniaceae in a more inclusive group with the Solanales and Boragi- nales. Montinia, not previously suggested to belong to this group on the basis of conventional char- acters, is basal to the Solanaceae and Convolvu- laceae (see below and Morgan & Soltis, 1993). The Solanales and Boraginales form a clade in most analyses, but in island-5,568 (not shown) the So- lanales are sister to a clade comprising the Boragi- nales and Gentianales. (10) Boraginales. This clade comprises the Hydrophyllaceae and Wi ipie Groups com- parable to this exist in most recent angiosperm treatments (Dahlgren, 1980. Takhtajan, 1987; Thorne, 1992; but not Cronquist, 1981). With the limited. sampling currently available, the Boragi- naceae appear paraphyletic and the Hydrophyl- laceae a monophyletic derivative from them. How- ever, the putatively primitive borages in the woody subfamilies Ehretioideae and Cordioideae have not been sampled for rbcL sequences. The Boraginales are sister group to the Solanales in most analyses reported here and in the analyses of Olmstead et al. (1992) and Chase et al. (1993), but are sister group to the Gentianales in island-5,568 (not shown). (11) Lamiales s.l. This clade is comprised of the orders Scrophulariales, Lamiales, and Callitri- chales, as well as Byblis (Byblidaceae, Rosales) and was recognized by Olmstead et al. ( and Downie & Palmer (1992) as monophyletic. This group is recognized by Dahlgren (1980), Takhtajan 1987), and Thorne (1992), but all three treat- ments maintain a distinction between the Lamiales s.s. and the Scrophulariales. As was pointed out previously (Olmstead et al., 1992), the Scrophu- lariales are not a monophyletic group. In addition, ~ results presented here (Fig. 5) and results of current work (Wagstaff & Olmstead, unpublished data) suggest that the Lamiales s.s. also may not be monophyletic. Therefore, it seems appropriate not to subdivide this group at present. Buddleja and Nicodemia have been considered congeneric and placed either in the Buddlejaceae (e.g., Cronquist, 1981) or Loganiaceae (e.g., Mabberley, 1987). In Volume 80, Number 3 1993 Olmstead et al. 717 Parsimony Analysis of Asteridae this analysis, they both clearly belong in the Lamia- but the two sequences do not appear to be closely related. les s.l. rather than the Gentianales, Likewise, the Oleaceae have been placed either in the Scrophulariales (e.g., Cronquist, 1981), Gen- tianales (e.g., Stebbins, 1974), or in their own order, Oleales, allied with the Gentianales (Dahl- gren, 1980; Takhtajan, 1987). Based on this rbcL analysis the Oleaceae appear to be the basal branch of the Lamiales s.l. As with the ericalean clade, the strict consensus of the two islands discovered in the analysis of the Lamiidae reveals little resolution within the Lamia- les s.l. However, unlike the ericalean clade, where there is little resolution within either island, in this group each island exhibits substantial resolution with particular sequences or small groups occurring in very different places in the two islands. This clade is more densely sampled than most other clades in the Asteridae s.l. Most of the groups that are congruent between islands consist of multiple representatives of large families, including the Acanthaceae, Scrophulariaceae, and Labiatae s.l. The Lamiales s.s. (Labiatae and Verbenaceae) may not be a monophyletic group, as was suggested previously based on a more limited sampling of 1992). Most of the representatives of the Lamiales s.s. form a rbcL sequences (Olmstead et al., monophyletic group with no clear distinction be- tween the traditional Labiatae and Verbenaceae. In this respect the clade closely fits the circum- scription of the family first suggested by Junell (1934) and recently revived by Cantino et al. (1992). However, Verbena, the sole representative of subfamily Verbenoideae in this analysis, is placed closest to /Vicodemia in a position removed from the rest of the Lamiales in island-135 (Fig. 5) and basal to them in island-1,350 (Fig. 6A). The Lamia- les s.l. are the subject of continuing research (Hed- ren et al., in prep.; Scotland & Olmstead, in prep.; Wagstaff & Olmstead, unpublished). These results are fully congruent with those of Olmstead et al. (1992) for relationships among clades within the Asteridae s.l. However, in that study the Dipsacales and Apiales form a clade and together with the Asterales s.l. form another, more inclusive, clade, whereas in this study relationships among the three groups remain unresolved. A boot- strap analysis (Olmstead et al., 1992) suggested moderately strong support (79%) for the Dipsa- cales-Apiales clade and weaker support for the Asterales-Dipsacales-Apiales (61%). The results of a bootstrap analysis conducted here (Fig. 7) 7) and 996 values, respectively, for the same two groups. These indicate 11% (not indicated on Fig. much lower values reflect the much greater taxon sampling density of this study (see Olmstead et al., 1992, for discussion of this effect). The Asterales— Dipsacales—Apiales clade also is found in both anal- yses of Chase et al. (1993). Differences in sampling and search strategies may account for the different results obtained in the three studies. A clade com- prising the Dipsacales and Apiales minus Viburnum and Sambucus was found by Downie & Palmer (1992), but their study did not support a united Asterales- Dipsacales-Apiales. No clade compris- any o за of these groups was found by Hufford (199 The last Mine inier Gentianales, Solanales, Bo- raginales, and Lamiales s.l., together form the only higher-order clade that is congruent between all three islands. These orders form Takhtajan's (1987) subclass Lamiidae. This group is one of the most strongly supported clades in bootstrap analyses of rbcL data (94%—Olmstead et al., 1992; 9455 — Morgan & Soltis, 1993) and also was identified in the cpDNA inverted repeat restriction site analysis of Downie & Palmer (1992). Relationships among the four primary clades within the Lamiidae remain unresolved in Olmstead et al. (1992), Downie & Palmer (1992), and in the strict consensus of the 3,900 trees found in “Search II" of Chase et al. (1993). In the analysis of the Asteridae s.l., the Lamiales s.l. are basal in island-2,784 (Figs. 1B, 2) and island-5,568 (not shown), whereas a clade comprising the Solanales and Boraginales is basal in island-102 (Fig. 1A). In the Lamiidae analysis, with greater sampling within the group, the Sola- nales-Boraginales appear basal in both islands ound. However, the strength of the latter conclu- sion is weakened by the reduced outgroup sampling in that analysis. Morgan & Soltis (1993), including a sequence for the saxifragaceous genus Vahlia at the base of the Lamiales s.l., identify the Gentiana- les as the basal branch within the Lamudae. The analyses of Chase et al. (1993) and the inverted repeat restriction site analysis of Downie & Palmer (1992) identify the Garrya clade as the sister group to the Lamiidae. That result also was obtained in island-102 and both of the shortest islands resulting from the analysis of the Lamiidae data set. However, island-2,784, island-5,568, and both of the islands one step longer than the shortest in the Lamiidae analysis do not support the sister group relationship between the Lamiidae and the Garrya clade. “Search II" of Chase et al. (1993) identified the small clade of Dillenia and Vitis as sister group to the Asteridae s.l., the sister group is a large clade comprising most but our results suggest that 718 Annals of the Missouri Botanical Garden of the Rosidae and Dilleniidae (including Dillenia and Vitis). That this relationship remains uncertain is indicated by the short internode length between the divergence of this small branch from the main lineage of the Rosidae and in adjacent internodes (Fig. 4). Thus, the question of the sister group to the Asteridae s.l. remains unresolved, but the con- clusion that this clade originated relatively early in the diversification of the higher dicots (Olmstead et al., 1992) is supported here and in the analyses of Chase et al. (1993). These results and those of Olmstead et al. (1992) both leave the base of the Asteridae s.l. unresolved. Island-102 in this study identified the Lamiidae and Garrya clades as the basal branch, whereas island-2,784 and island-5,568 were unresolved at the base, with either the Cornales or the Cornales- ericalean groups as possible basal branches. The analyses of Chase et al. (1993) identify the Cor- nales as basal within the Asteridae s.l. and the ericalean clade the next branch. The cpDNA in- verted repeat restriction site study of Downie & Palmer (1992) identified the ericalean group as basal, then the Dipsacales-Apiales, Asterales s.l., Cornales, and Lamiidae, but their study had few representatives of the ericalean group (Polemoni- aceae and Fouquieriaceae) and fewer outgroups. Hufford (1992) identified three solitary families as individual basal branches (Bruniaceae, Alseuos- miaceae—not included in rbcL analyses, and Es- calloniaceae), then the Cornales, ericalean group, and the traditional Asteridae. Anderberg (1992 identified the ericalean group as basal, with no — resolution among the remaining groups, but his study also suffered from extremely limited outgroup sampling, which compromises any conclusions at this level. These results represent a detailed, but by no means exhaustive, search for optimal solutions to the rbcL phylogeny of the Asteridae s.l. Our two- tiered approach of broadly sampling the entire Asteridae s.l. and sampling more densely within recognized clades offers opportunities to explore patterns of relationship in greater detail at various levels. The discovery of multiple islands in each of the subsets of the data in which a search for islands was conducted indicates that analysis of rbcL data (and probably most large complex DNA sequence data sets) must include an appropriate search strat- egy for finding multiple islands. The presence of multiple islands, which, by their nature, imply dif- ferent patterns of relationship, indicates that the rbcL data by themselves cannot resolve relation- ships at all levels within the Asteridae s.l. Congru- ence between analyses at different hierarchical lev- els is encouraging, although it does not in any sense constitute a test of the results found in one analysis, because the same data form the basis for all апај- yses. In contrast, incongruence between analyses at different levels (e.g., the inverted topology of the Gentianales in one analysis relative to another) demands an explanation. In addition to the two-tiered hierarchical anal- yses reported here for the Asteridae s.l., the anal- yses of Chase et al. (1993) present a third tier available for comparison (i.e., angiosperms and seed plants). The same 11 primary clades identified here in the Asteridae s.l. are identified in that analysis. However, the pattern of relationships among pri- mary lineages within the Asteridae s.l. implied by the analyses of Chase et al. (1993) is incongruent with each island found i in the bui ind of the As- teridae s.l. data set. T е 51 between these analyses: @) the ubi. of Chase et al. (1993) have greater sampling of ingroup and outgroup sequences in a single analysis, and (2) апа the smaller data sets used in this study were an- alyzed more fully for shortest trees and multiple islands. During the search for multiple islands in the Asteridae s.l., a long time on a plateau of trees longer than was each search became "stuck" for ultimately found by that search. In one search, a plateau three steps longer was searched with over 4,600 trees found at that length; the consensus tree was congruent with the results reported by Chase et al. (1993) for the Asteridae s.l. The seed plant analysis was conducted with NNI swapping and a ceiling of 3,900 trees, of which only three were examined with TBR swapping and with MUL- PARS off. This search strategy is unlikely to enable one to get off a plateau such as those encountered in our searches. This suggests that, as was explicitly admitted in Chase et al. (1993) for their broad analyses in general, their results may represent a suboptimal plateau with respect to the Asteridae | CONGRUENCE WITH CONVENTIONAL CHARACTERS The identification of biological traits, other than nucleic acid bases at particular positions in a DNA sequence, which are congruent with the results of a DNA sequencing study, provides potential syn- apomorphies and important validation for group- ings suggested by the molecular analysis. The rbcL analysis of the Asteridae offers the opportunity to examine some of the critical characters that have been used in traditional classifications. The most visible feature commonly associated with the Asteridae is the sympetalous corolla. It Volume 80, Number 3 1993 Olmstead et al. 719 Parsimony Analysis of Asteridae formed the basis for a major group of dicots, the Gamopetalae (de Candolle, 1813), in early classi- fications. Many families characterized by sympet- aly, which have been segregated into different sub- classes in many recent classifications (e.g., Cronquist, 1981), are reunited in the rbcL tree in the clade designated Asteridae s.l. The rbcL tree, however, includes several polypetalous groups, in- cluding the Cornales (as broadly defined by Cron- quist, with representatives found in several clades), Apiales, Grossulariaceae (Escallonia and Phyllo- noma), Hydrangeaceae, and Pittosporaceae (Pit- tosporum). The distribution of these polypetalous taxa on the rbcL tree makes problematic the in- terpretation of the origin of sympetaly. The Cor- nales and the Garrya clade are the only groups of the 11 discussed above that are consistently poly- petalous. Polypetalous groups are basal to other- wise predominantly sympetalous groups in many important clades. For examples of this pattern note Corokia with the Asterales s.l., Vahlia with the Lamiales s.l. (see Morgan & Soltis, 1993), and Berzelia with the Dipsacales. A few sequences represent families in which both sympetaly and polypetaly occur (e.g., Camellia—Theaceae, Вег- zelia—Bruniaceae). Some polypetalous taxa (e.g., Sarracenia and Montinia) are embedded within sympetalous clades (e.g., Ericales and Solanales, respectively). Hufford (1992) considered sympet- aly to be а synapomorphy for his Corniflorae— asterid group (comparable to our Asteridae s.l.) exclusive of Bruniaceae (Berzelia in our study). One possible explanation for the observed pat- tern of corolla fusion is that the Asteridae s.l. originally diversified as a polypetalous clade with subsequent evolution of sympetaly in several lin- eages. In this scenario, the polypetalous taxa near the base of many clades represent plesiomorphic relicts. However, an alternative hypothesis may be that sympetaly evolved early in the Asteridae s.l. and, during the early diversification of the clade, corolla fusion remained a developmentally labile trait before becoming rigidly canalized (Donoghue, 1989; Olmstead et al., 1992). In this scenario, polypetalous taxa represent the occasional rever- sion to polypetaly and would be expected to be found primarily near the base of clades. The pattern of corolla development in some groups (e.g., Api- ales, Erbar, 1991) suggests that reversal from sym- petaly to polypetaly has occurred (Olmstead et al., 2). This is an example of an area in which morphology deserves renewed investigation as a result of hypotheses derived from a molecular cladogram. The distribution of corolla fusion among the groups and the existence of tree islands make difficult any assessment of corolla evolution. It ap- pears that multiple origins are likely, but that some reversals may have occurred as well. The presence of tenuinucellate ovules with a single integument has been associated with the Asteridae s.s. (e.g., Cronquist, 1981), but has many exceptions outside the narrow circumscription of the Asteridae. However, the broadly circumscribed Asteridae s.l. encompass the large majority of those exceptions. А single integument (the result of fusion or loss of one integument) characterizes nearly all groups in the Asteridae s.l., except some members of the ericalean clade (e.g., Primulales and some Ebenales). The tenuinucellate condition generally is considered to be derived from the crassinucellate condition in angiosperms and also characterizes most members of the Asteridae s.l.; exceptions include the Garrya clade and the Cornaceae and relatives (but not the Hydrangeaceae). The trait ““crassinucellate ovule" encompasses a wide range of expression (Davis, 1966; Young & Watson, 1970) and is apparently derived from tenuinucel- late ancestors where it occurs in the Asteridae s.l. Hufford (1992) identified single integument as a synapomorphy for his Corniflorae-asterid group and tenuinucellate ovules as a synapomorphy to the same group exclusive of the Bruniaceae. The distribution of iridoid compounds has influ- enced heavily the classification of Dahlgren (1980), and the presence/absence of iridoids led Jensen et al. (1975) to reject the traditional Asteridae “аз а natural group." Most discussions of the systematic significance of iridoids rely on their evolutionarily derived condition and the implication that the iri- doids identify a monophyletic group (e.g., Jensen, 1992). This leads to conflicts with groupings based on other characters. The distribution of iridoids in the Asteridae s.]. suggests that iridoids form a syn- apomorphy for the entire clade comprising the Asteridae s.l. The distribution of iridoids in the ericalean clade combined with the lack of resolution in that clade makes it difficult to assess the basal state for the clade with respect to iridoids. The pattern of relationships depicted in island-2,784 (Figs. 1B, 2) includes the possibility that the eri- calean clade is basal within the Asteridae, thereby leaving open the possibility that if that clade is non- iridoid-containing at its base then iridoids may have evolved twice in the Asteridae s.l. However, it seems that the presence of iridoids is the basal condition in the Asteridae s.l. and the absence of iridoids is best interpreted as losses in several lin- eages, including the Primulales, Ebenales and rel- atives (within the ericalean clade), the Apiales, the Campanulaceae (Asterales s.l.), the Asteraceae, and 720 Annals of the Missouri Botanical Garden the Solanales. Additional losses within the Asteridae s.l. are known (e.g., within the Gesneriaceae and Labiatae subfam. Nepetoideae). The presence of simple iridoids in some members of the Hama- melidae (e.g., Liquidambar, Daphniphyllum), which exhibit morphological, embryological, and rbcL evidence incongruent with a placement in the Asteridae s.l., suggests that iridoids have evolved more than once. The distribution of iridoids in plants has proven to be problematic for systematics, but as phylo- genetic hypotheses become more refined, the dis- tribution of iridoids and other secondary com- pounds may make more sense. For example, Jensen (1992) presented an apparent contradiction in the Oleaceae, which contain both seco-iridoids (group- ing them with the Gentianales) and verbascosides (grouping them with the Lamiales s.l.). The rbcL tree places the Oleaceae at the base of the Lamiales s.l. clade, suggesting that the seco-iridoids are a plesiomorphic trait shared with the Gentianales, whereas the verbascosides are a synapomorphy uniting them with the Scrophulariales and Lami- ales. PROBLEMS MERITING FURTHER INVESTIGATION The rbcL tree of the Asteridae s.l. suggests several unexpected groups that merit further in- vestigation. For example, Polemonium (Polemoni- aceae, Asteridae) is grouped (e.g., Fig. 1A) with Diapensia (Diapensiaceae, Dilleniidae). These two taxa share many characters, including fused, pen- tamerous perianths, three fused carpels with nu- merous ovules, axile placentation, ovules that are unitegmic and tenuicellate, an annular disk at the base of the ovary, and tetrasporangiate anthers opening by longitudinal slits. There are some dif- ferences in the secondary compounds reported for oxylate crystals. Diapensia differs from Polemo- nium in having a second staminal whorl, a char- acter shared with many members of the Dilleniidae, but in Diapensia the whorl opposite the petals is reduced to staminodia. The loss of staminodia in the Diapensiaceae would result in flowers indistin- guishable from Polemoniaceae and, based on tra- ditional criteria, assignable to the Asteridae rather than the Dilleniidae (e.g., in the system of Cron- quist, 1981). The distribution of the Diapensiaceae is circumboreal at high latitudes, extending into temperate Asia. This suggests that the Polemoni- aceae may represent the temperate North Amer- ican derivatives of the Diapensiaceae. The clade composed of Garrya, Aucuba, and Fucommia is an unanticipated grouping. However, all three taxa are dioecious and produce the iridoid aucubin. Garrya and Aucuba have similar pollen and are alike in producing petroselinic acid and in having a single whorl of stamens alternate with the petals (a character shared with the Asteridae s.s. and relatives (e.g., Apiales)). Fucommia lacks a corolla, but contains iridoids, has a single integu- ment, and has a gynoecial structure that is virtually identical to that of Garrya, a unilocular ovary with two pendulous ovules and two style branches (4u- cuba also has a unilocular ovule). In several clades, a single taxon stands apart from the others, with respect to many conventional characters. Two examples are illustrative. Grise- linia (Cornaceae) is united with the Apiales, but differs in having crassinucellate ovules, iridoids, and a single ovule. Montinia is perhaps the most unusual anomaly in this respect. It falls out in the Solanales, from which it differs in many respects (see Morgan & Soltis, 1993). Griselinia is basal within its clade, suggesting that its distinctive char- acters may be either plesiomorphic or autapomor- phies not shared with the rest of the clade. Mon- tinia appears basal within the Solanales, but embedded well within the Lamiidae, thereby mak- ing the case for plesiomorphy more difficult. Cases of anomalous results such as these demand further study, both molecular and morphological, to con- firm or reject the placements suggested by these results. CONCLUSIONS The results presented here and elsewhere in this issue (e.g., Xiang et al., 1993) represent another step toward a molecular-based phylogeny for the Asteridae s.l. It is clear from the existence of mul- tiple equally parsimonious solutions, some different enough to belong to different islands using a TBR- swapping, heuristic search strategy, that we still have a long way to go to resolve many important questions. Conclusions that appear to be strong, based on the common occurrence of groups in different islands and in results of analyses at dif- ferent hierarchical levels or of different data sets (Downie & Palmer, 1992; Hufford, 1992; Ander- berg, 1992), include the following. (1) There exists a monophyletic Asteridae s.l. containing some el- ements not formerly assigned to the Asteridae. (2) The Lamiidae (sensu Takhtajan, 1987) are a mono- phyletic group comprising four separate clades, the Gentianales, Solanales, Boraginales, and Lamiales s.l. (including the Scrophulariales). (3) Eleven ter- Volume 80, Number 3 1993 Olmstead et al. 721 Parsimony Analysis of Asteridae minal clades are congruent among all trees and all islands, including the four comprising the Lamiidae, the Asterales s.l., Dipsacales, Apiales, Cornales, an ericalean clade (including representatives of the Ebenales, Primulales, etc.), and two small clades herein referred to as the Garrya clade and Пех clade. (4) W relationships among these primary lineages remain unresolved. (5) There is substantial resolution with- in many of these 10 clades (e.g., Apiales, Asterales ith the exception of the Lamiidae, l., Cornales, Dipsacales, Solanales), but not for others (e.g., ericalean clade, Gentianales, Lamiales s.l.). (6) Unitegmic and tenuinucellate ovules ap- pear to be synapomorphies for the Asteridae s.l. (7) Sympetaly and the presence of iridoids also but ad- ditional resolution is needed to distinguish multiple may be basal traits for the Asteridae s.l., origins from a single origin with subsequent rever- sals. LITERATURE CITED ANDERBERG, А. А. 1992. The circumscription of the ericales, and their cladistic relationships to other fam- ilies of “higher” dicotyledons. Syst. Bot. 17: 660- 675 BREMER, B. & L. STRUWE. 1992. Phylogeny of the Rubiaceae and the Loganiaceae: Congruence or con- flict between morphological and molecular data? Amer. J. Bot. 79: 1171-1184. CANDOLLE, A. P. DE. 1813. Théorie Elémentaire de la Botanique. Paris. CANTINO, P. D., R. M. HARLEY & S. WAGSTAFF. 1992. Genera of Labiatae: Status and classification. Pp. 511-522 in R. M. Harley & T. Reynolds (editors), Advances in Labiate Science. Royal Botanic Gardens, Kew CHASE, M. W., D. E. Sorris, R. С. OLMSTEAD, D. MORGAN, D. Н | SQ X ‚ S. DAYANANDAN & V. 1993. Phylogenetics of seed plants: An analysis of nucleotide sequences from the pid gene rbcL. Ann. Missouri Bot. Ca rd. 80: 528-5 Cronguist, А. 1981. An Integrated System of Classi- fication of кина Plants. Columbia Univ. Press, or DAHLGREN, RL The distribution of characters e embryological 1975. characters. Bot. Not 1980. A senal system of classification of the . angiosperms. J. Linn. Soc., Bot. 80: 91-124 . 1983. General aspects of angiosperm evolu- tion and macrosystematics. Nordic J. Bot. 3: 119- 149. Davis, G. L. 1966. Systematic Embryology of the An- giosperms. John Wiley ons, New DONOGHUE, M. J. 1989. Phylogenies and the analysis of evolutionary sequences, with examples from seed plants. Evolution 43: 1137-1156. J. A. DoYLE. 1989. Phylogenetic studies of seed plants and angiosperms based on morphological characters. Pp. 181-193 in B. Fernholm, K. Bremer & H. Jornvall (editors), The Hierarchy of Life. El- sevier Science, Amsterdam ‚ К. С. OLMSTEAD, J. Е. мин & J. D. PALMER. 1992. Phylogenetic relationships of Dipsacales based on rbcL sequences. Ann. Missouri Bot. Gard. 79: 333-345. DowNiE, S. R. & J. D. PALMER. 1992. Restriction site mapping of the chloroplast DNA inverted repeat: A molecular АУРА a "à Asteridae. Ann. Missouri Bot. Gard. 79: шш: С. 1991. Сава А systematic character? t. Jahrb. Syst. 112: 417-451. Fuse J. Cases in which parsimony and ompatibility "cA will be positively misleading. Sys t. Zool. 27: 40 "E L. D. m and their relationships о other nonmagnoliid dicotyledons: A ме, 5 using ао and chemical data. Ann Missouri Bot. Gard. 79: 48. JENSEN, S. R. 1992. ен implications of the distribution of iridoids and other chemical compounds in the Loganiaceae and other families of the Asteri- dae. Ann. Missouri Bot. Gard. 79: 284-302. , B. J. NIELSEN & R. DAHLGREN. кеше Iridoid compounds, their occurrence and systematic a tance in the angiosperms. Bot. Not. 128: 148-180. JUNELL, S. 1934. Zur ен үчн a sys- tematik der Verbenaceen und Labiaten. Symb. Bot N, R. S. WALLACE, H. J. MICHAELS 71992. Phylogenetic e of rbcL lage variation in the Asteraceae. Ann Missouri Bot. Gard. 79: 428-445. Kron, К. А. & М. W. Chase. 1993. Systematics of the Ericaceae, Empetraceae, Epacridaceae and re- lated taxa based upon rbcL sequence data. Ann. Missouri Bot. Gard. 80: 735 MABBERLEY, D. J. 1987. The Plant- -— Cambridge Univ. Press, Cambridge. Mappison, D. R. 1991. The discovery and importance of multiple islands of most-parsimonious trees. Syst. Zool. 40: 315-328. , M. RuvoLo & D. L. Sworronp. 1992. Geo- graphic origins of human mitochondrial DNA: Phy- logenetic evidence from control region sequences. Syst. Biol. 41: 111-124. MiICHAELS, Н. J., К. M. Scott, R. С. OLMSTEAD, T. SZARO, R. K. JANSEN & J. D. PALMER. 1993. Interfamilial relationships of the Asteraceae: Insights from rbcL sequence variation. Ann. Missouri Bot. Gard 1 Morean, D. R. 4 D. E. Sorris. 1993. Phylogenetic relationships among members of Saxifragaceae sensu lato based pr o sequence data. Ann. Missouri Bot. Gard. 80: -660. ai R. G., H. J. MicHAELS, К. M. Scorr € J. D. 1992, Monophyly of the Asteridae and обол of their major lineages inferred from 722 Annals of the Missouri Botanical Garden ОЧА арчы of rbcL. Ann. Missouri Bot. Gard. 65. бойы. G. M., D. E. Sorris & P. S. SoLrIs. 1992. Molecular phylogenetic study of үрен M Araliaceae, and Pittosporaceae). Am J. Bot. 79(suppl.): : STEBBINS, C. L. 1974. Flowering Plants. Evolution Above the di eni Level. Belknap Press, Cambridge, Mas- hus SWOFFORD, D. ЈЕ 1992. PAUP: Phylogenetic Analysis Using Parsimony, Version 3.0s. Illinois Natural His- tory Survey, Champaign, Illinois. & OLSEN. 1990. rh reconstruc- tion. Pp. 4 411-501 in D. M. Moritz (editors), Molecular Systematics. ne Sunder- land, Massachusetts. TAKHTAJAN, А. p ud Magnoliophytorum. Nauka, Leningrad. [In Rus THORNE, R. T. Classification and geography of the flowering plants. Bot. Rev. 58: 225-348. VIGILANT, L., M. STONEKING, Н. HARPENDING, К. HAWKES & A. C. Wilson. 1991. African populations and the evolution of human mitochondrial DNA. Science 253: 1503-1507. WaGENITZ, С. 1977. New aspects of the systematics of Asteridae. Pl. Syst. Evol. Suppl. 1: 375-395. 92. The Asteridae: Evolution of a concept зй its Rune status. Ann. Missouri Bot. Gard. 79: 209-217 XIANG, Q.-Y., D. E. Sorris, D. R. Morcan & Р. S. SOLTIs. 1993. Phylogenetic relationships of Cornus L. sensu lato and putative relatives inferred from rbcL se- н data. Ann. Missouri Bot. Gard. 80: 723- 134. Younc, D. J. & L. Watson. 1970. The classification of и A study of the upper levels of hi- erarchy. Austral. J. Bot. 18: 387-433. Qiu-Yun Xiang, Douglas E. Soltis, David R. Morgan, and Pamela S. Soltis? PHYLOGENETIC RELATIONSHIPS OF CORNUS L. SENSU LATO AND PUTATIVE RELATIVES INFERRED FROM rbcL SEQUENCE DATA! ABSTRACT A parsimony analysis of 46 rbcL lure was performed to evaluate relationships among Cornus and puta within Cor relatives, as well as among subgrou SEENE of о us. proposed Cor rnus. Our results indicate that dps and Camptotheca), mastixioids орала and Mastixia), Curtisia, and gene “с angium, nyssoids (Nyssa, Dav idia, are the closest ornaceous FA (ii) nyssoids- и. (ii) Curtisia, and (iv) hydrangeoids. The relationships among the four major lineages within the cornaceous clade remain unresolve naceous clade. Four . Sequence data from Garrya, and Griselinia, previously placed i in Cornaceae by so | major lineages were "identified within Cornus by rbcL sequence data, ids also reveal har Corokia, Helwingia, Aucuba, authors, are only distantly related to Cornus and but relationships among these groups are incompletely resolved. Relationships among members of the dogwood family (Cornaceae) have long been problematic due to the morphological diversity of the family. There is, for example, no consensus regarding the cir- cumscription of the family; from 1 to 15 genera have been ascribed to Cornaceae by various au- thors. The troubled taxonomic history of Cornaceae is briefly reviewed below and summarized in Ta- ble 1. Harms (1898) included 15 genera divided among seven subfamilies in his Cornaceae (Table 1), a treatment that largely followed Hooker (1867). Of these 15 genera, by far the largest and most wide- spread is Cornus, which comprises approximately 45-60 species. Many of the remaining 14 genera are small and/or localized in geographic distribu- tion (Table 2). The considerable debate concerning the composition of Cornaceae is illustrated by the fact that 11 of the 15 genera included in Harms's (1898) Cornaceae have been treated as monotypic families by at least one author based on one or several distinctive morphological features (all orig- inal references cited in Takhtajan, 1987, except Chen, 1988): Davidia (Li, 1955), Camptotheca (Chen, 1988), Mastixia (Calestani, 1905), 4ucuba (Agardh, 1858), Curtisia (Takhtajan, 1987), Gar- гуа (Lindley, 1834), Alangium (de Candolle, 1828), Melanophylla (Takhtajan ex Airy Shaw, 1972), Griselinia (Takhtajan, 1987), Helwingia (Decaisne, 1836), and Toricellia (Hu, 1934). For brevity, we will review only the treatments of Takhtajan (1980, 1987), Cronquist (1981, 1988), and Eyde (1988), because they illustrate well the magnitude of the differences among pro- posed taxonomic schemes regarding the treatment of Cornaceae and putative relatives. Takhtajan (1980) retained only Cornus, Curtisia, and Mas- tixia in Cornaceae, placing them in three separate subfamilies (Table 1). Many of the genera originally placed in Cornaceae by Harms (1898) were treated as distinct families and considered to be only dis- tantly related to Cornus (e.g., Corokia, Griselinia, Helwingia, Melanophylla, Kaliphora, and Tori- cellia) by Takhtajan (1980) (Table 1). More re- ! We thank G. Plunkett and R. С. Olmstead for о unpublished sequences; D. E. Boufford, E. Woo E. K. Hay, S. Yankowski, P. Goldblatt, E Brunsfeld, J. W T. Edwards, E. С. H. Oliver, acmillan, A. mee Pan, H.-N. Qin, Pacha ang, Z.-H. Ji, K.-Y. H M. Jayas RU Arnold Arboretum, U. 5. National Arboretum, Strybing Arboretum, Missouri deem Garden, and The Royal Botanical Gardens, Kew, England, for help in obtaining leaf material; and R. by NSF grant BSR 9007614 Taxonomists, and the Морен! үне ific Ass Imstead and M. Chase for helpful comments regarding data analysis. This project was supporte a WSU WW pon and grants from Sigma Xi, the American Society of Plant ? Department of Botany, Washington State кош Pullman, Washington 99164-4238, U.S.A. ANN. MISSOURI BOT. GARD. 80: 129-794. 1999. Annals of the M 724 Botanical Garden issouri аваргоило“) IPIPIOIXUSPIA Diprav(] рог ујојашођ) DSSÁN IBIPIOSSÁN аваовилог) оваое шмјен se[eid y se[erjpootuo J, IBIIPIXUSEIA 9eooetst n7) әвәэеи1о7) 0221101d um) әвәзюїрїАР(] војеилог) DIO0Y ANDY ¡Ay dounja y aea9e¡AydoueraJa IBIIPIUNOSIIC) D10107) „зеаов || Ацдо8лгу «зојвовивлран аваовАллво) s9[eu10o7) аваот шмјон ILIPO], a?ooerumastu«) авартогхцов ји] IBAPIOISU my IBIIBIPIAR(] so[pu1o-) 01310107) LLULL] „зајевелихес аваовцој де) piduinja yy „зваоецелу эрташ рзәу1о1йшъ/) səjeqeIy 01]]ә2110], 7j ydounjapy pqnonp әғәртои2о”) IPIPIOISH my IPIPIOIXUSBIA авоовилој) оваоетвивју pipiav(] n221101d um.) DSSÁN ӘРӘЭРЅЅА М двдовАллео) 01[]92110 ], — (8861) зрад (2861) чеГезцхеј, (1861) 1smbuor) (0861) uele1yxe L (1961) uosurqoing (0161) uuogue дү (8681) Surg “ұшәшцеәд ләці ut Mo[aq рәдѕц st szu407) Á[uo ‘Ayoyduns јој гвлопов ојеВолвов xis ош snu407) рәрїлїр (2861) чебеацхеј, pue (2961) uosurqoing 'sjsmuouoxej euros Aq олцејал $nu407) € рәзәріѕиоә uəəq seu рәдѕц Ацше} 10 snua3 əy} Á[uo упад 'зоцпшеј 10 влопов [e19A9s surejuoo jeu] ларло ue JO Ашер e sojeorput („) xsuejse uy -seqmrejqns pue seque; otdKjouour лерип рәдѕц jou зле LIUN) 'saAn*[a1 aanemd pue snuso7) Jo зупошјеод orurouoxej jo uosueduro^) `1 318v] Volume 80, Number 3 Xiang et al. g 725 Phylogenetic Relationships of Cornus сеп у, Takhtajan (1987) removed Curtisia and Mastixia from Cornaceae and recognized them as monotypic families (Table 1). However, he still considered these two genera close allies of Cornus and members of his Cornales. Cronquist (1981), in contrast, retained a broader view of Cornaceae than Takhtajan (Table 1). Cron- quist (1981) closely followed the treatment of Wan- gerin (1910) (Table 1), but added Aralidium to Cornaceae. Cronquist (1988) later added the nys- soids (Vyssa, Davidia, and Camptotheca) to Cor- naceae following Eyde (1988) and included all of Harms’s (1898) genera except Alangium and Gar- rya in his Cornaceae. He argued against dividing Cornaceae into a number of small or monotypic families as done by Takhtaja Attempting to synthesize all available evidence, Eyde (1988) considered the Cornaceae to consist of Cornus, nyssoids, and mastixioids (modern Mas- tixia and its fossil allies) (Table 1). Eyde also pro- posed that mastixioids may be the closest relatives of Cornus based primarily on a single shared trait: a two-armed surface hair formed from a single cell. The delimitation of Cornaceae became even less clear with the recent addition of Diplopanax to Cornaceae (Eyde & Xiang, 1990). Diplopanax, which consists of a single species (D. stachyanthus Hand.-Mazz.) occurring rarely in southern China and northern Vietnam, was previously placed in Araliaceae (Handel-Mazzetti, 1933) ot only do concepts of Cornaceae vary greatly, but there is considerable disagreement regarding Different lines of evidence have suggested a wide array of closest relatives for Cornus (Table 3). Evidence from em- the closest relatives of Cornus. bryology, for example, suggests that Alangium is the closest ally of Cornus (Chopra & Kaur, 1965) whereas chromosome numbers suggest that Vyssa, ~ Davidia, Mastixia, and Garrya are also close rel- atives of Cornus (Goldblatt, 1978). Seed structure (germination valves) provides a still different pic- ture, suggesting a close oe among Cornus, Mastixia, Diplopanax, Nyssa, and Davidia (Eyde, 1963, 1988; Eyde & Xiang, 1990). талышы from palynology, wood anatomy, serology, secondary chemistry, and fatty acid chemistry each suggests a different suite of close relatives for Cornus (Table 3; Bate-Smith et al., 1975; Breuer et al., 1987; Fairbrothers & Johnson, 1964; Ferguson, 1977; Ferguson & Hideux, 1980; Li & Chao, 1954). The taxonomic complexity of Cornaceae is fur- ther compounded by the extreme polymorphism of Cornus. Wangerin (1910) classified the genus into seven subgenera (Table 4) using primarily the mor- phology of the inflorescence, bracts, and fruits. TaBLE 2. position of Cornus and putatively related genera. Geographic distribution and species com- Alangium ca. 20 spp., mostly in Asia, a few in Polynesia, New Guinea, qr Australia, tropical Africa, Madag car, Comoro Islands. Aralidium : Hy aya Aucuba -11 spp., eastern Himalayas to Ja- pan. Camptotheca — l sp., southwestern and south-central China. Cornus ca. 45 to 60 spp., mostly in eastern Asia and North America, a few in Europe, 1 in Central America, 2 in South America, and | in tropical Africa. Corokia 6 or 7 spp., South Pacific islands. Curtisia 2 sp., southern Africa. Davidia l sp., southwestern China and western Hupei of China Diplopanax l sp., southern China and northern Viet Garrya 14 spp., southwestern United States, Mexic Indies Griselinia 6 spp., New Zealand, Chile, and Bra- zl. Helwingia 3 spp., esit to Japan. Kaliphora l sp., Madaga Mastixia 13 spp., eee Asia and west- ern Pacifi Melanophylla | spp., Madagasca spp., eastern yea eastern North merica, northern Mexi Toricellia 2 or 3 spp., eastern Himalayas and southwestern China. Wangerin's view of Cornus has been largely fol- lowed by most modern investigators (Eyde, 1987; Ferguson, 1966; Xiang, 1987 nus has been divided by some into as many as six genera (Table 5; Hutchinson, 1942, 1967; Pojar- kova, 1950). The complex history of this problem and the nomenclatural changes involved have been reviewed by Eyde (1987) and Ferguson (1966). Phylogenetic relationships among species groups within Cornus have also been debated. Cornus can be roughly divided into four groups: big-bracted dogwoods, dwarf dogwoods, cornelian cherries, and bractless dogwoods. Emphasizing the distribution of iridoids, Bate-Smith et al. (1975) proposed that the big-bracted dogwoods are the most primitive species of Cornus and the bractless dogwoods are the most advanced. This view is supported by ev- idence from wood anatomy (Adams, 1949; Li & Chao, 1954) and floral anatomy (Wilkinson, 1944 Dermen (1932) proposed that based on chromo- ). In contrast, Cor- мМ 726 Annals of the Missouri Botanical Garden TABLE 3. Close relatives of Cornus as suggested by eight different lines of morphological, anatomical, karyological, and chemical evidence. Characters Suggested close relatives Cytology Nyssa, Davidia, Mastixia, and Garrya (Goldblatt, 1978) Pollen morphology Curtisia, Mastixia, yan, 1 Nyssa, Davidia, and Camptotheca (Eram- 71; Ferguson, 1977; Ferguson & Hideux, 1980; Eyde, 1988; Reitsma, 1970) Floral anatomy Fruit morphology Embryology Wood anatomy Secondary metabolic products Mastixia, Diplopa Nyssa, Davidia, Mastixia, and Toricellia (Eyde, 1967, 1988) and Davidia (Eyde, 1963 nax, Nyssa, 1988) Alangium (Chopra & Kaur, 1965) Corokia, Helwingia, Schefflera, Nyssa, Garrya, Mastixia, Vi- burnum, and Griselinia (Li & Chao, 1954 Mastixia, Curtisia, and Davidia (Bate-Smith et al., 1975) Curtisia, Mastixia, and Corokia (Breuer et al., 1987) Fatty acid chemistry some numbers and sizes, the cornelian cherry group is basal within Cornus. Eyde (1988), in contrast, proposed that the genus early diverged into two main lineages, a bracted (or red-fruited) line in which the inflorescence has basal bracts, and a bractless (or blue- or white-fruited) line in which the bracts are rudimentary or lacking. The bracted line includes the big-bracted dogwoods, dwarf dog- woods, and the cornelian cherries. Using morpho- TABLE 4. Intrageneric treatment of Cornus proposed by Wangerin (1910): 50 species divided into seven sub- genera. All species are woody and have bisexual flowers except those indicated. Thelycrania bractless «аланы 36 spp. with 5 ding cym с pii p 1 dioe- cious species, with four small herbaceous bracts subtending an umbel-like cyme cornelian cherries, 4 spp. with four herbaceous bracts subtending an Afrocrania Macrocarpium umbel-like cyme. dwarf dogwoods, 2 herbaceous s with four petaloid bracts subtend: Arctocrania ing a condensed cyme. Mexican disciflorous dogwood, 2 all, herbaceous, sub Discocrania spp. with four sm and early deciduous bracts tending a capitat ке н баш North American big-bracted dog- woods, 2 spp. with four large petaloid bracts uae a capi- tate inflorescenc Asian big-bracted PN 3 spp. with four large petaloid bracts е а шше inflore subsequently devel- Benthamidia Benthamia cence whic ops into a Mns fruit. logical characters and evidence from fossils and embryology, Eyde proposed that the bractless dog- woods are ancestral and connect Cornus to the nyssoids. He classified Cornus into four informal subgroups without giving these groups taxonomic rank: bractless dogwoods, cornelian cherries, big- bracted dogwoods (including the Mexican disciflo- rous dogwoods), and dwarf dogwoods. From the above review it is apparent that an- other experimental approach is needed to help un- ravel the systematic problems posed by Cornaceae. We therefore employed comparative sequencing of rbcL with the following objectives: (1) clarify relationships among Cornus and putatively related genera, and (2) elucidate evolutionary relationships among the major subgroups within Cornus. The advantages of rbcL sequencing for addressing phy- logenetic questions at this level have been well reviewed (e.g., Clegg & Zurawski, 1992; Palmer et al., 1988) MATERIALS AND METHODS Species included in this study represent the ma- jor subgroups of Cornus as well as taxa that pre- viously have been considered closely allied with Cornus by at least some investigators (e.g., Cornus alternifolia L.f., C. canadensis L., C. chinensis Wangerin, C. florida L., C. kousa Hance, C. mas L., C. obliqua Raf., C. oblonga Wall., C. officina- lis Siebold & Zucc., C. walteri Wangerin; Alan- gium chinense (Lour.) Harms, Aucuba japonica Thunb., Camptotheca acuminata Decne., Coro- kia cotoneaster Raoul, Curtisia dentata (Burm.f.) m., Davidia involucrata Baill., Diplopanax stachyanikus Hand.- ‚ Garrya elliptica Douglas ex Lindl., Griselinia lucida J.G. Forst, Helwingia japonica (Thunb.) F. Dietr., Mastixia caudatilimba C.Y. Wu ex Soong, /Vyssa ogeche Volume 80, Number 3 Xiang et al. 727 1993 Phylogenetic Relationships of Cornus TABLE 5. Comparison of several taxonomic schemes proposed for Cornus sensu lato, depicting the magnitude of the discrepancies that exist. Strict view Pojarkova (1950) Hutchinson кшш: (6 genera) (6 gener Broad view Wangerin (1910) 7 subgenera) Ferguson (1966) (8 subgenera) Thelycrania Svida Bothrocaryum css Afrocrania Afrocrania Cornus Cornus _ === Cynoxylon о Dendrobenthamia Chamaepericlymenum Cynoxylon Chamaepericlymenum Thelycrania ———————— Kraniopsis “~~ Mesomora Afrocrania Afrocrania Macrocarpium Arctocrania Arctocrania Discocrania Discocrania Benthamidia Cynoxylon Benthamia enthamia Marsh; see Appendix of this issue). The sequence for Cornus kousa was provided by Imstea (unpublished). These taxa represent the major cast of characters required for a molecular phylogenetic analysis of Cornus and putative relatives. It would have been desirable to include in this study several additional genera (Kaliphora, Melanophylla, To- ricellia, and Aralidium) of which we could not obtain either living or recent herbarium collections due to their rarity and/or geographically restricted distributions (Table 2 Total DNAs were isolated from leaf tissue fol- lowing the method of Doyle & Doyle (1987) as modified by Soltis et al. (1991). Double-stranded rbcL was amplified using thermostable DNA poly- merase provided by either Promega or Epicentre Technologies. Primers used for amplifications are synthetic oligonucleotides. The forward or 5' prim er (Z1) is composed of the first 30 bases of rbcL of Zea mays L. Two different reverse primers were used. The one used most often (3'rbcL) is that of Olmstead et al. (1992), which consists of 34 bp beginning 103 base pairs outside of the terminus of the gene in Nicotiana. For Diplopanax and Mastixia, however, this Z1-3'rbcL primer com- bination did not yield a product so we used ап internal reverse primer, а 30-mer beginning at position 1351 of rbcL in Z. mays. e double-stranded products subsequently were used as templates, and the forward and reverse primers used individually for single-stranded DNA amplifications. Single-stranded products were pre- cipitated with 20% PEG/2.5M NaCl. DNA pellets were washed once each with 70% and 95% EtOH, dried, and redissolved in TE. DNA sequencing fol- lowed the dideoxy method using Sequenase 2.0 enzyme and the Sequenase 3.1 kit (U.S. Biochem- ical Corp.). Sequencing primers were derived from the rbcL sequence of Z. mays and kindly provided by G. Zurawski (DNAX Research Institute). Both the forward and reverse strands were sequenced for all taxa. Twenty-one sequences were generated in this study for Cornus and putatively related genera. Twenty-five additional sequences from Saxifra- gales, Apiales, and Asteridae (Fig. 1, also see Ap- pendix, this issue) were included in a phylogenetic analysis to determine possible alliances of Cornus. Cercidiphyllum and Daphniphyllum (represen- tatives of Hamamelidae) and /tea, Saxifraga, and Astilbe (representatives of Rosidae) served as out- groups for this analysis because they are close to, yet basal to, the taxa included in this broad rbcL sequence analysis. The choice of taxa and out- groups for use in our analysis was based on the results of the large phylogenetic analysis of 499 rbcL sequences for seed plants (Chase et al., 1993). For all taxa but Diplopanax and Mastixia, 1377 bp of rbcL sequence data were compared beginning at position 31 and ending at base pair 1407. Due to the amplification strategy required for Diplopanax and Mastixia (see above), we an- alyzed 1320 bp for these genera, from position 3l to 1350. All parsimony analyses were con- ducted with PAUP version 3.0s (Swofford, 1991) using MULPARS and TBR branch-swapping with characters specified as unordered and unweighted. To ensure that all “islands” of shortest trees were found (Maddison, 1991), 100 tree searches were performed each with a different order of taxa ad- dition. A bootstrap analysis using 100 replicates was also conducted to obtain estimates of reliability for each monophyletic group. RESULTS The phylogenetic analysis resulted in 690 most parsimonious trees that were 1,185 steps long, all in one island (Fig. 1). Two major groups can be identified in all of the most parsimonious trees. One 728 Annals of the Missouri Botanical Garden Nicotiana NYSSOIDS- MASTIXIOIDS “| BIG-BRACTED DOGWOODS Ј DWARF DOGWOOD | BRACTLESS DOGWOODS CORNELIAN CHERRIES CORNACEOUS CLADE HYDRANGEOIDS eutzia Hydrangea Cercidiphyllum 11 | 98% |8 - Saxifraga OUTGROUP re ЈА 21 sti Daphniphyllum FIGURE 1. The ње -rule consensus tree including all d groupings constructed from 690 minimum length trees resulting from parsimony analysis of rbcL sequences of aceae and additional taxa. This tree is one of the 690 shortest trees (length — 5 rudis consistency index — 0.410. retention index = 0.541). The tree is rooted using representatives from Hamamelidae (Cercidiphyllum and Daphn iphyllum), and Rosidae (Itea, npe — pt utside the cornaceous clade that have been placed in (сви y some investigators. Species of Cornus аге мане С Volume 80, Number 3 1993 Xiang et al. 729 Phylogenetic Relationships of Cornus of these groups comprises Cornus, Alangium, Cur- пзта, nax, Mastixia, and genera from Hydrangeaceae (Decumaria, Carpenteria, Deutzia, Philadel- phus, and Hydrangea) (Fig. 1). This group, here- after referred to as the ““cornaceous clade," has a bootstrap value of 71% (Fig. 1). The analysis of Chase et al. (1993) similarly indicates а mono- phyletic group identical to the cornaceous clade. Nyssa, Camptotheca, Davidia, Diplopa- Four major lineages can be recognized within the cornaceous clade, although the bootstrap values for some аге not high: (1) nyssoids (Vyssa, Camp- totheca, Davidia)-mastixioids (Diplopanax and Mastixia) (58%), (2) Cornus- Alangium (57%), 3) Curtisia, and (4) hydrangeoids (Decumaria, Philadelphus, and Hy- drangea) (98%) (Fig. 1). The second major group contains all of the remaining taxa analyzed, in Carpenteria, Deutzia, cluding those that have been considered close rel- atives of Cornus (Aucuba, Garrya, Helwingia, Griselinia, and Corokia). These five taxa are not closely allied with the cornaceous clade (Fig. 1). Griselinia is the sister of a well-supported (boot- strap value of 95%) araliaceous group. Corokia is allied with Asteraceae. Helwingia has as its closest relative Phyllonoma, a genus usually placed in Escalloniaceae or treated as a distinct subfamily of Saxifragaceae sensu lato (discussed in greater detail by Morgan & Soltis, 1993). Аисиђа and Garrya are close allies; both appear in a clade with Eu- commia (Eucommiaceae: Hamamelidae). The phy- logenetic positions of these five genera traditionally alied with Cornaceae are congruent with those derived from the broad analysis of Chase et al. (1993), and a detailed analysis of sequences from the Asteridae (Olmstead, unpublished). The phylogenetic analysis of rbcL sequences reveals that Cornus is a well-supported monophy- letic group (85%; Fig. 1). Four lineages can be recognized among the 10 species of Cornus se- quenced, although these clades are not strongly supported in the bootstrap analysis: (1) big-bracted dogwoods-dwarf dogwoods (65%), (2) bractless dogwoods (65%), (3) cornelian cherries (68%), and (4) C. oblonga, a morphologically distinct bractless species (Fig. 1). Strongly supported lineages having bootstrap values of 96%-100% are further iden- tified within three of these four lineages: the big- bracted dogwoods (C. kousa and C. florida), the opposite-leaved bractless species (C. walteri and C. obliqua), and two cornelian cherries (C. mas and C. officinalis) (Fig. 1). Sequence divergence among species of Cornus is very high. For example, C. canadensis differs from C. kousa by 54 base substitutions and C. mas differs from C. canadensis y 54 base substitutions (Fig. 1). DISCUSSION Numerous studies have demonstrated the great value of rbcL sequence data in reconstructing plant phylogenies (e.g., Doebley, 1990; Donoghue et al., 1992; Olmstead et al., 1992; Soltis et al., 1990; papers in this issue). The value of comparative rbcL sequencing for resolving phylogenetic ques- tions in particularly troublesome groups is also il- lustrated by our analysis of Cornaceae. In this study, rbcL sequence data have (1) helped to iden- tify an alliance composed of Cornus and related taxa (i.e., the cornaceous clade), (2) demonstrated that several genera traditionally allied with Cornus are only distantly related to this genus, and (3) suggested the presence of major lineages within the cornaceous clade and within Cornus. 1. THE CORNACEOUS CLADE Our analysis reveals that Cornus is a well-sup- ported lineage and suggests that nyssoids (Nyssa, Davidia, and Camptotheca), mastixioids (Diplo- panax and Mastixia), Alangium, Curtisia, and hydrangeoids are the closest relatives of Cornus (Fig. 1). No suite of cornaceous relatives identical to those suggested by rbcL sequence data has been proposed previously, and only the proposals of Eyde (1988) and Takhtajan (1980, 1987) are somewhat close to the scheme suggested by our analysis of rbcL sequences. Alangium has been widely recognized as a monotypic family (e.g., Cronquist, 1981, 1988; Hutchinson, 1967; Takhtajan, 1980, 1987; Wan- gerin, 1910). Our phylogenetic analysis of rbcL sequence data suggests that Alangium may be the sister group to Cornus. А close relationship between these two genera is also suggested by similarities in flowers, fruits, and embryology (Chopra & Kaur, 1965; Еуде, 1988), binucleate pollen (Brewbaker, 1967), and a base chromosome number of x — 11. In addition, one species of Alangium has the distinctive two-armed hairs (but with unequal arms) characteristic of Cornus (see Eyde, 1988). Eyde (1988) similarly concluded that Alangium was a close relative of Cornus using the above characters. Nyssoids differ from Cornus primarily in having unisexual and polygamous flowers. However, nys- soids also share a number of important characters with Cornus: presence of iridoids (Bate-Smith et al, 1975), similar gynoecial vasculature (Eyde, 1967), germination valves on the fruit stone (Eyde, 730 Annals of the Missouri Botanical Garden 1988), and H-shaped thinning of the pollen ap- erture (Eramyan, 1971; Ferguson, 1977; Fergu- son & Hideux, 1980; Reitsma, 1970). Citing these features, Eyde (1988) proposed that nyssoids share a recent common ancestor with Cornus. Our se- quence data, therefore, agree with these morpho- logical, anatomical, and chemical similarities in suggesting that the nyssoids are closely related to Cornus (Fig. Relationships among nyssoids (Vyssa, Davidia, and Camptotheca) have long been controversial. The three genera of nyssoids are often placed in the family Nyssaceae (Cronquist, 1981; Hutchin- son, 1967; Wangerin, 1910). Fruit and seed mor- phology, as well as evidence from wood anatomy and fatty acid chemistry (Breuer et al., 1987; Eyde, 1963, 1988; Hohn & Meinschein, 1976; Titman, 1949), suggests a close relationship among the nyssoids. However, Davidia has been recog- nized as a monotypic family by some authors (Takh- tajan, 1980, 1987). Using wood anatomy, Titman (1949) proposed that Davidia is the most primitive nyssoid, with Camptotheca derived from Nyssa. This proposal was supported by evidence from fatty acids (Breuer et al., 1987; Hohn & Meinschein, 1976), and data from morphological, palynological, and fossil studies (Eyde, 1963, 1988). Chen (1988) argued, however, that Camptotheca was only dis- tantly related to Nyssa based on his embryological studies of Camptotheca, Nyssa, and Davidia and proposed a new monotypic family, Camptotheca- ceae (Chen, 1988). Serological data (Fairbrothers, 1977; Fairbrothers & Johnson, 1964), on the other hand, support the separation of Davidia from Муз- sa and Camptotheca. Our analysis of rbcL se- quence data suggests instead that these three gen- era, together with Mastixia and Diplopanax, form a monophyletic group. Diplopanax was placed previously in Araliaceae by Handel-Mazzetti (1933) based on the resem- blance of its flowers, without observation of fruits, to those of the araliaceous genus Dendropanax. This family assignment was questioned by Hoo & Tseng (1978) and Tseng (1983). They found im- portant similarities between the fruits of Diplo- panax and Mastixia and suggested that Diplo- panax is better placed in Cornaceae. Eyde & Xiang (1990) made a direct comparison of Diplopanax fruits with fossil fruits of mastixioids and restated that Diplopanax is closely allied with Mastixia because both share distinctive fruitstones, two- armed hairs that are one-celled on the leaf mid- veins, and hooked petals. These structural features place Diplopanax in the genus Mastixicarpum, part of a woody-fruited mastixioid complex that was thought to have gone extinct four million years ago. Although Mastixicarpum is the older name, nomenclatural rules require that the united group take the name of its modern component, Diplo- anax. Our rbcL sequence data place Diplopanax in the nyssoid group and support its removal from Araliaceae where it was initially placed (Handel- Mazzetti, 1933) and has since been classified (Hoo & Tseng, 1978; Hutchinson, 1967). Although rbcL sequence data indicate that Diplopanax is a close relative of Mastixia, whether Diplopanax is most closely related to Mastixia (as suggested by Eyde & Xiang, 1990) is still uncertain based on rbcL sequences. All of our analyses show Diplopanax to be closest to Camptotheca, but this relationship is not strongly supported by the bootstrap analysis (41%) (Fig. 1). Diplopanax and Camptotheca are similar in that both genera (as well as Mastixia) are diplostemonous. Further assessment of possible relationships between Diplopanax and the nyssoid genera are not presently possible because of the lack of embryological, chemical, and anatomical evidence for Diplopanax, a result of the geograph- ic isolation and rarity of the genus. Mastixia has been assigned to a diverse array of families (e.g., Caprifoliaceae, Nyssaceae, Aqui- foliaceae, Araliaceae, Icacinaceae, and Cornaceae; Matthew, 1976). Phylogenetic analysis of rbcL sequence data reveals that Mastixia is a member of the cornaceous clade. A number of lines of evidence (Table 3) suggest a close relationship be- tween Mastixia and Cornus. However, based on rbcL sequences, Mastixia does not appear to be the sister of Cornus as suggested by Eyde (1988) but is more closely related to the nyssoids instead (Fig. 1). Thus, the two-armed surface hairs shared by Mastixia, Diplopanax, and Cornus (Eyde, 1988) may actually have evolved independently in each of these genera. Curtisia, comprising two species endemic to South Africa, clearly is a member of the cornaceous clade in these analyses of rbcL sequences. Similar chemical patterns were observed in Cornus and Curtisia by Bate-Smith et al. (1975), adding fur- ther support for a close relationship between à two genera. Curtisia also has been linked w Mastixia based on chromosomal (Goldblatt, in and palynological data (Ferguson, 1977; Ferguson & Hideux, 1980; Hideux & Ferguson, 1976). In contrast, Eyde (1988) did not consider Curtisia to be a close ally of Cornus, Mastixia, and nyssoids (his Cornaceae) because of its small pollen grains Volume 80, Number 3 1993 Xiang et al. 731 Phylogenetic Relationships of Cornus and distinctive fruit-stone vascular bundles. Cur- tisia generally is placed in Cornaceae as a separate subfamily (Harms, 1898; Takhtajan, 1980; Thorne, 1983; Wangerin, 1910) or recognized as a mono- typic family placed close to Cornaceae (Takhtajan, 1987; Thorne, 1992). Both of these views are concordant with rbcL sequence data. The relationships of hydrangeoids have long been controversial (see review of this topic by Morgan & Soltis, 1993). Our analysis of rbcL sequences clearly supports their position as close relatives of Cornus, Alangium, Curtisia, nyssoids, and mas- tixioids, as suggested previously by Breuer et al. (1987), Takhtajan (1987), Philipson (1977), and Krach (1977) based on other lines of evidence. Hydrangeoids are similar to other members of the cornaceous clade in general flower structure: flow- ers 4—5-merous with an epigynous disk, calyx tube adnate to the ovary, petals free, ovary inferior (sometimes incompletely inferior as in Hydrangea or superior as in Carpenteria) with 2—5 locules, and ovules usually pendulous. In addition, both hydrangeoids and other members of the cornaceous clade produce iridoids. Several features, however, clearly separate the hydrangeoids from other mem- bers of the cornaceous clade. Hydrangeoids possess numerous ovules per locule, capsular fruits, and — = central bundles in the gynoecial vasculature. other members of the cornaceous clade, in contrast, have a solitary ovule in each locule, typically bear drupes (Camptotheca has subsamaroid fruits), and lack central bundles in the gynoecial vasculature. Furthermore, hydrangeoids have unitegmic tenui- nucellate ovules whereas nearly all the other mem- bers of the cornaceous clade (excluding Curtisia for which ovule type is unknown) have unitegmic crassinucellate ovules (Philipson, 1977). An ex- ception is found in the bracted species of Cornus, which like the hydrangeoids have unitegmic tenui- nucellate ovules (Erdelska, 1986; Eyde, 1988). Although analysis of rbcL sequences has iden- tified a well-supported cornaceous clade comprising four major lineages, relationships among these four major lineages are not resolved. To clarify more fully relationships within the cornaceous clade, se- quences of more rapidly evolving nuclear or chlo- roplast genes will be required. 2. DISTANT RELATIVES OF CORNACEAE Not only do rbcL sequence data indicate the presence of a cornaceous clade, these data also reveal that Aucuba, Garrya, Corokia, Griselinia, and Helwingia, all traditionally placed in Corna- ceae, are not closely related to Cornus or to other members of the cornaceous clade (Fig. 1). Our findings are consistent with the broad phylogenetic analysis (Chase et al., 1993), which similarly in- dicates that these five genera are distantly related to the cornaceous clade. These genera have been placed in Cornaceae or Cornales by different tax- onomists because of their tree or shrub form, ves- sel-segments usually with scalariform perforations, 3-5-merous and inconspicuous flowers, partially united sepals, separate petals, epigynous discs, in- ferior ovaries, solitary ovules per locule, and mostly drupaceous fruits. Each of these genera is discussed individually below. ucuba was placed in Cornaceae by Harms (1898), Wangerin (1910), and Cronquist (1981) and considered a close relative of Cornaceae by Takhtajan (1980, 1987). Our rbcL sequence anal- ysis reveals, however, a distant relationship be- tween Aucuba and the cornaceous clade. The dis- tant relationship of Aucuba to the cornaceous clade is also supported by its distinctive wood structure and pollen morphology (Adams, 1949; Ferguson & Hideux, 1980). The closest relative of 4ucuba, based on rbcL sequence data, is Garrya, a genus of 15 species in North America and the West Indies. Garrya has also been placed in Cornaceae (Harms, 1898) or recognized (as Garryaceae) as a close relative of Cornaceae (Cronquist, 1988; Hutchinson, 1967, 1969; Takhtajan, 1987; Thorne, 1983, 1992). The close affinity between Aucuba and Garrya suggested by sequence data is also supported by similarities in gynoecial vas- culature (Eyde, 1964), phytochemistry (Bate-Smith et al., 1975), and fatty acids present in seed oils (Breuer et al., 1987). Corokia is also far removed phylogenetically from Cornus in our analysis of rbcL sequences. Although Corokia traditionally has been placed in Cornaceae, more recently it has been considered closely related to Argophyllum (Escalloniaceae) based on pollen morphology and wood and floral anatomy (Eyde, 1988; Takhtajan, 1987). How- ever, rbcL sequence data also reveal a distinct relationship between Corokia and Escallonia; an rbcL sequence is not available for Argophyllum. Most treatments have placed Griselinia in Cor- naceae (Cronquist, 1988; Harms, 1898; Hutch- inson, 1967; Wangerin, 1910). It was, however, recognized as a monotypic family in Cornales by Takhtajan (1980), and later moved to his Hy- drangeales (Takhtajan, 1987). Close affinities among Griselinia, Aucuba, and Garrya were pro- posed by Eyde (1964) based on gynoecial vascu- 732 Annals of the Missouri Botanical Garden lature. Phytochemical evidence also supports these relationships (Breuer et al., 1987). However, rbcL sequence data place Griselinia in a basal position in the araliaceous group, phylogenetically distant from Aucuba and Garrya (Fig Helwingia was placed in Cornac :eae by Harms (1898), Wangerin (1910) and Cronquist (1981). Our analysis of rbcL sequences reveals, however, that Helwingia is well outside the cornaceous clade. A distant relationship between Helwingia and Cor- nus is also supported by phytochemical data, as well as floral characters (Bate-Smith et al., 1975; Eyde, 1967), and both Adams (1949) and Eyde (1988) suggested that Helwingia be removed from Cornaceae. In our analyses, the genus is phylo- genetically closest to Phyllonoma, a traditional member of Saxifragaceae sensu lato. Phyllonoma is similar to Helwingia in that both have flowers borne on the leaf blades. Although the two genera do share this rare characteristic, no previous au- thors have suggested a close relationship between Helwingia and Phyllonoma (see detailed discus- sion of this topic by Morgan & Soltis, 1993). 3. PHYLOGENETIC RELATIONSHIPS WITHIN CORNUS The large amount of rbcL sequence divergence among species of Cornus is somewhat surprising given the conservative rate of evolution of rbcL. With this large amount of sequence divergence, our results demonstrate that rbcL sequence data can sometimes be used to study intrageneric re- lationships, as discussed below. Varying schemes of intrageneric relationships have been proposed for Cornus, reflecting different hypotheses of morphological evolution and different emphases on available characters (Tables 4, 5). The major argument concerns the relative ad- vancement within the genus of the bracted versus the bractless dogwoods. Our phylogenetic analysis Cy of rbcL sequence data reveals four lineages: oblonga, big-bracted dogwoods— dwarf dogwoods, cornelian cherries, and bractless dogwoods. The latter three clades were also recognized in analyses 1988) and chloroplast DNA restriction site mutations (Bruns- feld et al., 1991; Xiang et al., 1991); insufficient material of C. oblonga was available for inclusion of morphological characters (Eyde, of this species in the cpDNA restriction site study. However, relationships among these four lineages are incompletely resolved by rbcL sequences (Fig. 1). The rbcL sequence analysis indicates that C. oblonga, a bractless species found in the eastern Himalayas and southwestern China, is a lineage from cornelian cherries, the distinct remaining bractless species, and the big-bracted dogwoods— dwarf dogwoods (Fig. 1). Several morphological, embryological, and anatomical features similarly suggest the distinctiveness of C. iip in (Adams, 1949; Chopra & Kaur, 1965; Zhu, 1984). This species Red has been viewed as the most ше i di ies ME Cornus, e» the hypothetical 1 based largely on its оен characters (Eyde, 1988; Rick- ett 1950). Although rbcL sequence data do not suggest a basal position for C. oblonga in Cornus, sequence data do agree with other evidence in suggesting an isolated phylogenetic position of this species in the genus. However, the exact phylo- genetic position of C. oblonga awaits the sequence analysis of more rapidly evolving chloroplast genes, such as ORFK In summary, our analysis of rbcL sequences has (1) identified a cornaceous clade that includes Cor- nus, Alangium, Curtisia, Nyssa, Camptotheca, Davidia, Diplopanax, Mastixia, and genera from Hydrangeaceae (Decumaria, Carpenteria, Deut- zia, Philadelphus, and Hydrangea); (2) revealed that several genera previously placed in Cornaceae (e.g., Helwingia) are only distantly related to Cornus; Aucuba, Garrya, Corokia, Griselinia, and (3) identified four major lineages within the cor- naceous clade and suggested four major lineages within Cornus. Other putative relatives of Cornus, such as Toricellia, Aralidium, Melanoph ylla, and Kaliphora, may ultimately be considered part of the cornaceous clade if material becomes available for analysis of rbc LITERATURE CITED ADAMS, J. E. 1949. Studies in the comparative anatomy the fom J. Elisha Mitchell Sci. Soc. 65: 218-2 ВАТЕ- Sume Е La . FERGUSON, K. Hursow, B. J. ELSEN & T. Swain. 1975 indicabit, inter- olas in he Caoin. Biochem. Syst. Ecol. 3: 79-89. BREUER, B., T. STUHLFAUTH, H. Fock & H. HUBER. 1 А aceae, Aquifoliaceae, Hamamelidaceae and Styra- caceae. и 26: 1441- BREWBAKER, Ј. The distribution and phylo- genetic eir. of binucleate and trinucleate pol- len grains in the angiosperms. Amer. J. Bot. 1069-1083. BRUNSFELD, S. J., D. E. SoLris & P. S. SoLris. 1991. Evolution of e big- шоо. dogwood (Cornus). Amer. J. Bot. 78: СНАЗЕ, M. у. , D. E. Sor тїз, К. С. OLMSTEAD, n Morc Pu D. H. Les, B. D. Ментек, M. R. ALL, R. ne H. G. Hius, Y.-L. Qiu, K. A. p ond J. H c , E. Conti, J. D. PALMER, J. R. MANHART, K. J. on H. J. Micuaets, W. J. Kress, K. С. KAROL, У. D. CLARK, M. Hepren, B. S. Gaur, R. Volume 80, Number 3 1993 Xiang et al. 733 Phylogenetic Relationships of Cornus K. JANSEN, К.-Ј. Kim, C. К. WiMPEE, J. Е. SMITH, С. R. FURNIER, S. Н. SrRAUSS, Q.-Y. XiaNc, С. M. PLUNKETT, P. S. SoLTIS, S. SwENsEN, S. E. WIL LIAMS, P. А. GADEK, C. J. QUINN, L. E. EGUIARTE E. GOLENBERG, : + ARN, JR., GRAHAM, S. C. H. BARR S. DAYANANDAN & V. А. ALBERT 1993. E uie of seed plants: An је ка of nucleotide sequences from the plastid gene rbcL. Ann. Missouri Bot. Gard. 80: 528-580. CHEN, L. Ј. . The Comparative Embryological Study and Proposed Affinity of Camptotheca, Nyssa and Davidia. Ph.D. Dissertation. Beijing Institute of Botany, Academia Sinica, Beijing, China. as R. N. & H. Kaur. 1965. Some e of the mbryology of Солаш, Phytomorphology 15: 353- 35 CLEGG, M. T. & С. ZurawskI. 1992. Chloroplast DNA and the study of plant E Present status and future | Pp: 1-13 ir Коза Solis, B. E. Soltis & J. J. Doyle a Molecular Systematics of Plants. кыш Cha and Hall, London. CRONQUIST, А. 19 hx E, System of Classi- ation of P Hielo M Plants Columbia Univ. Press, New 9 The Evolution and Classification of Flowering Plants, 2nd ed. New York Botanical Gar- den, New York. DERMEN, H. PE studies of Cornus. J. Arnold P 13: 401-417 DoEBLEY, J. Molecular "MEE of Zea (Gr s еы 35: 143- DONOGHUE, M. J., В. С. OLMSTEAD, LF F. SMITH & J. D. PALMER. 1992. Phylogenetic relationships of Dip sacales based on Ha sequences. Ann. Missouri Bot. Gard. 79: 333-34 Doyle, J. J. € J. L. de 1987. A rapid DNA isolation procedure for small quantities P fresh leaf tissue. Phytochem. Bull. 19: 11-15. Palinologicheskie dannye k sis- ort. 1 rodstven- nykh semeistv. Pp. 2 n L. A. Kupriyanova & M. S. Yakovlev (editors), ‘Motion pyl'tsy: Cucurbitaceae, Thymelaeaceae, Cornaceae. Izdate- l'stvo “Nauka,” Leningrad. ERDELSKA, O. Embryo development in the dog- wood с mas. Phytomorphology 36: 23-28. Еуре, R. Н. 1963. Morphological and paleobotanical studies of de Nyssaceae. I. The modern species an their fruits. J. Arnold Arbor. 44: 1-59. 96 кие о and Apu affinities of батуа, Amer. J. Bot. 51: 1083-1092. 1967. The peculiar eye Ка vasculature i matic significance. Phyt morphology Lis 172-1 82. 87. The case for keeping Cornus in the broad re sense. Syst. Bot. 12: 505-518. 988. Co ia Cornus: Puzzles and progres a e systematics of the dogwoods. Bot. 351. Rev Pub & Q. E pen 1990. Fossil mastixioid (Cor- J 77 naceae) alive in eastern Asia. Amer. J. Bot. 77: 689- FAIRBROTHERS, D. E. 1977. Perspectives in plant sero- nomy. Ann. Missouri Bot. Gard. 64: 147-160. ЈА. JOHNSON. 1 Comparative sero- logical studies within the families Cornaceae (dog- wood) and Nyssaceae (sour gum). Pp. 305-318 in C. A. Leone (editor), Taxonomic Biochemistry and Serology. Ronald Press, New York. FERGUSON, 1. K. 1966. Notes on E oo of Cor ha О „> Arbor. 47: 10 ——— W pollen morphology and its taxonomic significance in Cornaceae sens. lat. Proc. IV. Int. Palynol. Conf., Lucknow 1: 240- 249. GOLDBLATT, P. 1978. A contribution of cytology in Cornales. Ann. Missouri Bot. Gard. 65: 650-655. HANDEL -MazzErrTI, Н. F. V. 1933. Plantae novae Chin- 198 Harms, H. 0-270 in A. En- gler & K. Prantl (editors), Die Natürlichen Pflan- tae, Teil III, Abteilung 8. Bui ian. i. M. & I. K. FERGUSON. 1976. The stereostruc- ture of the exine and its evolutionary significance i in Saxifragaceae sensu lato. Pp. 327-377 in I. K. . Mu ller (editors), The Evolutionary i 20108 Pk significance in the Nyssa- | and Cornaceae. Bioch. Syst. & Ecol. 4: 193- Hoo, E . TSENG. 1978. Araliaceae. P. 135 i = LO Popul. Sin; Vol. 54. Academic Press, Bei- MORE. J. D. 1867. Cornaceae. Pp. 947-952 in G. m & J. D. Hooker (editors), Genera Plantar- um, Mia l, part 3. Reeve & Co. and Williams € , London. HUTCHINSON, T 1942. Neglected generic о tics in the family artic tun Ann. Bot. n. s. 6: 83 1967. The Genera of Flowering Plants, Vol. 2. Glirentian Press, Oxford. Evolution and Phylogeny of Flowering Рал --Biedbdedon: Facts and Theory. Academic Press, London and New York. ји J. E. 1977. Seed characters in and affinities Saxifragineae. Pp. 141-153 in K. Ku- bitzki (editor), Flowering Plants— Evolution of High- er aes es Springer- i ; Li, H.-L C.-Y. CHao. Comparative anatomy of ih woods of the Сава ель and allies. Quart. J. Taiwan Mus. 7: 119-136. M&DBISON. D. R. 1991 The discover y and importance of m ultiple islands of most-parsimonious trees. Syst. 15-328. Zool. 2 MATTHEW, K. M. 1976. A revision of the genus Mas- ixia е. Blumea 23: 51-93. MoncaN, D. R. € D. E. Ѕогтіѕ. 1993. Phylogenetic relationships among members of Saxifragaceae sensu lato based a dus sequence data. Ann. Missouri Bot. Gard. 80: OLMSTEAD, R. "i ui MicHaELs, К. M. Scorr & J. D PALMER. 1992. Monophyly of the reae d identification of their major lineages inferred from As е of rbcL. Ann. Missouri Bot. Gard. 19::249—2 PALMER, J. D., ^ K. JANSEN, H. J. Micuakis, M. W. CHASE а 1. R. MANHART. 1988. Chioroplast DNA o g n phylogeny. Ann. Missouri Bot. Gard. 180-1206. Annals of the Missouri Botanical Garden Рнплрѕом, W. R. . Ovular morphology and the classification of dicotyledons. Pp. го іп К. Kubitzki (editor), Flowering Plants— Evolution of Ју gher Categories. Springer ш Vienna, New Жыйым, А. 1950. De systemate е Linneani Bo . Komarova 0. Pollen morphology of the Alangi- aceae. Rev. Paleobot. Palynol. 10: 249-332. RickerT, H. Cornus in Mexico, with notes on the evolution of the genus. Anales Inst. Biol. Univ. 3-94. s, M. T. Стесс & M. DURBIN. rbcL sequence rode m phylogenetic relationships in Saxifragaceae sensu lato. Proc. Natl. A. 87: 4640-4644. , LLIER & M. L. EDGERTON. ; Chloroplast DNA variation within and among genera of the Heuchera group (Saxifragaceae): Ev- idence for chloroplast transfer and paraphyly. Amer. J. Bot. 78: 1091-1112. SwoFFORD, D. L. PAUP: Phylogenetic Analysis Using Parsimony, Version 3.0s. Computer program distributed by the Illinois Natural History Survey, Champaign, Illinois. TAKHTAJAN, A. L. 1980. Outline of the classification of flowering plants (Magnoliophyta). Bot. Rev. 46: 225- 359. 87. Sistema Magnoliofitov. Izdatel'stvo m d. 83. Pags new oo in the ц Nordic Ј. Bot. 3: 85- n updated phylogenetic a hie of the flowering plants. Aliso 13: 365-389. TirMAN, Р. W. 1949. Studies in the woody anatomy of the RETA Nyssaceae. J. Elisha Mitchell Sci. Soc. PM C. 7 1983. The systematic position of Yid anax Hand.-Mazz. Acta Phytotax. Sin. 21: 151- 152. WANGERIN, W. . Cornaceae. Pp. 1-101 in A. Engler (editor), Das Pflanzenreich. Series IV, family 229 (Heft 41). W. Engelmann, Leipzig. WILKINSON, А. M. 1944. Floral anatomy of some spe- cies of Cornus. Bull. Torrey Bot. Club 71: 276-301. XIANG, Q.-Y. 1987. А neglected — of ae L. s.l. with "s reference to a new subgenus— Sinocornus Q. Y. Xiang. Acta utin Sin. 25: 9-131. , D. E. Souris & P. S. SoLTIS. 1991. Molecular systematics of Cornus L. s.l. Amer. J. Bot. 78: 230. ZHU, Z.-Y. Yinquania Z. Y. Zhu— A new genus of Cornaceae from China. Bull. Bot. Res. (Harbin) 4(4): 121-128. SYSTEMATICS OF THE ERICACEAE, EMPETRACEAE, EPACRIDACEAE AND RELATED TAXA BASED UPON rbcL SEQUENCE DATA! Kathleen A. Kron?? and Mark W. Chase?* ABSTRACT The Ericaceae, Epacridaceae, and Empetraceae are usually recognized as closely related families of woody plants sequences o includes Empetraceae and Epacridaceae. t invert in deve elopment, and the presence of endos osperm d often unisexual a arsimony analyses indicate а monophyletic, broadly defined Ericaceae that Тће Епсасеае are acid-loving woody plants found throughout the world but with major centers of diversity in the Himalayas, tropical Asia, montane South and Central America, and southern Africa. Traditionally, this family has been closely associ- ated with two other woody plant families that also prefer acidic habitats: Epacridaceae (primarily Australian) and Empetraceae (amphitropical). These closely related (e.g., hese characteristics include the presence ricales sensu Cronquist, of endosperm haustoria at both the chalazal and micropylar ends of the embryo and anthers that invert in development (see Judd & Kron, 1993). The Empetraceae comprise three genera (Cera- tiola, Corema, Empetrum) of predominantly an- emophilous plants with reduced, often unisexual flowers. Ceratiola and Corema occur in the North- ern Hemisphere primarily along both sides of the Atlantic, whereas Empetrum exhibits a bipolar dis- tribution. The Epacridaceae are a much larger family almost entirely restricted to Australia, al- though some species occur in southern South America, New Caledonia, Indomalaysia, and New Zealand. Morphologically Epacridaceae show much variability, and there are no characters that con- sistently distinguish all members of this family from members of the Епсасеае (Judd & Kron, 1993; D. Morrison, pers. comm.). For example, the an- thers of the Epacridaceae dehisce by slits rather than pores, the latter considered typical of the Ericaceae. However, several members of the Eri- caceae have anthers that dehisce by slits or slitlike pores. Traditionally, the Epacridaceae have been considered the Australian correlates of the Erica- ceae (Cronquist, 198 The E extremely diverse, ricaceae (in the broad sense) are a large, and essentially cosmopolitan family characterized by release of pollen in tetrads and/or sympetalous urceolate or campanulate co- rollas (Judd & Kron, 1993). Most members of the family are photosynthetic, but Cronquist (1981) segregated the Pyrolaceae and Monotropaceae out of the Ericaceae because of the following tenden- cies: reduction from shrubs to herbs, gradually increasing mycoparasitism, gradual loss of chlo- rophyll, and choripetaly This study is an investigation of phylogenetic relationships among the Ericaceae, Empetraceae, and Epacridaceae. The general position of the ' This project was supported by National Science Foundation grants BSR-8821264 (K. Kron) and mu 8906496 (M. Chase). Special thanks to the dr eens Garden, Edinburgh, the A Parks for providing material, to V. A. Albert, H. Hills, and M. Be rnold Arboretum, W. S. Judd, and C. W nett for bal assistance, to A. Anderberg йс е 8 ене dira to K. Eu ek to D. E. Soltis, P. F. Stevens, and an anonymous reviewer for helpful omme n the ript. $ ыш, a Biolog * Present address: Department of Biology, Wake gy, | of North о Chapel Hill, nee Carolina 27599-3280, U.S.A n-Salem, Nor th Carolina 27109, 0.5 iversity, Win lem * Present address: Royal Botanic (асы, Kew, Richmond, Surrey TWO 3AB, United Kingdom. ANN. Missouni Bor. Garb. 80: 735-741. 1993. 736 Annals of the Missouri Botanical Garden TABLE l. Current classification of the taxa (listed alphabetically by order) represented in the analysis of rbcL sequence data of Ericaceae, Empetraceae, and Epacridaceae Classification of Ericaceae is based on Stevens (1971). Classification of all other taxa is based on Cronquist (1981). Asterisks indicate taxa that functioned as outgroups in this study. Order Family Subfamily Representative taxa Diapensiales Diapensiaceae Diapensia Ebenales benaceae Diospyros* Sapotaceae Chrysophylium* Sapotaceae Manilkara* Styrace Styrax Symplocaceae Symplocus* Ericales Clethraceae Clethra* Cyrillaceae Cyrilla* Empetraceae Ceratiola Epacridaceae Cyathodes, Dracophyllum, Epacris, Leucopogon, Pentachondra Ericales Ericaceae Ericoideae Calluna, Erica Pyroloideae Pyrola Rhododendroideae Befaria, Daboecia, Elliottia, Leiophyllum, Rhododendron Vaccinioideae Arbutus, Arctostaphylos, Cas- siope, Chamaedaphne, En- kianthus, Gaultheria, Leu- cothoe, Vaccinium, Zenobia Geraniales Balsaminaceae ышы Polemoniales Polemoniaceae Polemonium* Primulales Myrsinaceae Ardisia Primulaceae Anagallis* Theophrastaceae Clavija* osales Byblidaceae Roridula* Sarraceniales Sarraceniaceae Sarracenia Theales Actinidiaceae Actinidia* eaceae Camelli Violales Fouquieriaceae Fouquieria* icales (Cronquist, 1981) within the flowering plants is addressed in Chase et al. (1993) and Olmstead et al. (1993). Because of the morphological diver- sity and the difficulty of determining homologous structures among members of the Ericaceae, Em- petraceae, and Epacridaceae, a parsimony analysis of nucleotide sequences of rbcL (ribulose-1, 5-bisphosphate carboxylase oxygenase, large sub- unit) was performed. Although the problem of ho- moplasy is certainly not resolved with the use of sequence data in phylogenetic analysis (cf. Albert et al., 1993), the chance that patterns of homo- plasy would have similar distributions in both mor- phological and rbcL sequence data is negligible. MATERIALS AND METHODS The 17 genera (including Pyrola) chosen for study (Table 1) represent the range of morpholog- ical diversity of the Ericaceae for an initial ap- proximation of the limits of the family. Many of these taxa were also included in a cladistic analysis of morphological characters (Judd & Kron, 1993). Ceratiola (Empetraceae) and five genera repre- senting most of the major lineages of Epacridaceae (D. Morrison, pers. comm.) were included. Cyrilla (Cyrillaceae) and Clethra (Clethraceae) were also included because they are often considered closely related to the ericads (Cronquist, 1981). Whereas the focus of this study is the circumscription of the Ericaceae, representatives of several families within the subclass Dilleniidae (Cronquist, 1981) were in- cluded in a broad analysis (see Chase et al., 1993) to provide a context for previously hypothesized relationships with the ericalean taxa (e.g., Cron- quist, 1981; Dahlgren, 1983; Hufford, 1992; Takhtajan, 1980; Thorne, 1983, 1992). These taxa have often been considered as either related Volume 80, Number 3 1993 Kron & Chase 737 Systematics of Ericaceae, Empetraceae, and Epacridaceae or ancestral to the Ericaceae. The sampling of ““dilleniid”” taxa is not complete. Thus assessment of relationships outside the Ericaceae s.l. is ten- tative (see the Appendix at the end of this issue for a list of vouchers). Total DNA was extracted from fresh leaves via oe modified CTAB procedure (Doyle & Doyle, 87). The rbcL gene was amplified using engi- id primers, cloned, and then sequenced using the standard dideoxy method (see Olmstead et al., 1993). rbcL, visual alignment of the sequences was easily Because of the size-conserved nature of performed. The sequences were analyzed with PAUP version 3.0s (Swofford, 1991) using heu- ristic search methods (see methods in Chase et al., 1993). This provided a clade as a starting point for the following analyses. In the broad analysis, 41 of the “ericalean” ““dilleniid”” taxa sampled appear in a single clade (length of this portion of the tree — 1,169 steps; see asterid III, figs. 13A, B in Chase et al., 1993) sister to the lineage that contains traditional “аз terid" taxa (Cronquist, 1981). These 41 taxa (Fig. 1) were therefore used in subsequent “local” anal- and yses. The sequences of these taxa were analyzed as a group, using PAUP 3.0s (Swofford, 1991). Eighteen of the 41 taxa (Table 1) in this clade functioned as outgroups (Maddison et al., 1984) for the Ericaceae s.l. Trees were arranged with the Primulales clade sister to the remaining taxa (Fig. 1). Two types of analyses were performed: (1) Ten replicates of random sequence addition ere used to find the shortest tree (TBR, MUL- PARS, STEEPEST DESCENT, all characters and character states weighted equally). To select a tree to illustrate from among the numerous most par- simonious trees obtained, the weighting criterion of Albert et al. (1993) was applied. (2) А search (TBR, MULPARS) was made for trees up to and including five steps longer than the most parsi- monious trees found in the first analysis. Consensus trees were constructed using the FILTER option at each step. In this analysis the following outlying taxa were removed to decrease the time involved in the search for longer trees: Anagallis, Ardisia, Chrysophyllum, Clavija, Diapensia, Diospyros, Fouquieria, Impatiens, Polemonium, Styrax, and Symplocus. This allowed us to assess the relative robustness of the clades within the Епсасеае ob- tained with maximum parsimony (cf. Bremer, 1988; Novacek, 1991). Thus, clades that collapse in trees one step longer (i.e., one step less parsimonious) are less robust than clades that remain intact in trees up to five steps longer than maximum par- simony. (This is the “decay index” of Mishler et al., 1991.) We realize that ten replicates may not reveal all islands of maximum parsimony (Maddi- son, 1991). However, in the decay analysis branch swapping was performed on all trees up to five steps less parsimonious, giving us greater сопћ- dence in the maximum parsimony of the initial analysis. RESULTS In the "local" analysis we performed on the 41 "ericalean" taxa, 85 trees of 1,166 steps were found with a retention index (R.I.) of 0.603. The application of the weighting criterion of Albert et al. (1993) to the 85 most parsimonious trees se- lected two of these trees as most parsimonious. hese two trees differed only by the position of Diospyros (Ebenaceae). In one tree Diospyros is sister to the Primulales, in the second tree Dios- pyros is sister to the Diapensia (Diapensiaceae) Polemonium (Polemoniaceae) clade. Since these two trees are so similar only one is shown (Fig. 1). Decay analysis indicated that all relationships out- side of the Ericaceae s.l. are weakly supported and decay at one step less parsimonious (Fig. 2) The Епсасеае are monophyletic if. Empetra- ceae, Epacridaceae, and Pyrolaceae are included. We will hereafter refer to this clade as Ericaceae 5.1. The decay index for the branch leading to the Ericaceae s.l. indicates that the branch collapses in trees only two steps longer, indicating that there are few unique synapomorphies for the Ericaceae s.l. in the rbcL sequence data. This result agrees with the morphological study (Judd & Kron, 1993), where nearly all of the synapomorphies for the family show homoplasy within the family. Within the Ericaceae s.l. the currently recognized subfam- ilies (Stevens, 1971) Rhododendroideae and Vac- cinioideae are paraphyletic. The Ericoideae are derived out of some of the ** and Ceratiola (Empetraceae) is nested within this clade. The Epacridaceae are monophyletic but are nested within the Ericaceae and are sister to some of the “ taxa. Enkianthus is the sister to the rest of the Ericaceae. rhododendroid" taxa vaccinioid”” In the local analysis of the 41 “ericalean” + ““dilleniid”” taxa the Primulales (Cronquist, 1981) appear monophyletic with the Clavija (Theophras- taceae) sister to Ardisia (Myrsinaceae) and An- agallis (Primulaceae), and this clade is retained in trees one step longer (Fig. 2). The Ebenales (Table 1) are not monophyletic in this analysis. The prob- able paraphyletic nature of this order has already been noted by Dahlgren (1975) because of its em- bryological and chemical heterogeneity. 738 Annals of the Missouri Botanical Garden RE l. One of the 85 maximum parsimony trees (length — 1,166; (retention € — 0.60 Clavija Anagallis rdisia Diospyros Diapensia Fouquieria Calluna Erica Befaria Elliottia Ericaceae s.l. ire ||, |, — A LI r Actinidia Cyrilla Sarracenia Roridula Camellia C.I. (consistency index) = 0.520; R.I. 3) favored by the weighting criteria of Albert et uii S Underlined taxa represent the Ericaceae s.l. This tree is arranged with the Primulales sister to all others in accord with the general results of Chase et al. (1993). In all of the most parsimonious trees, members of the Епсасеае, Empetraceae, Epacridaceae, and Pyrolaceae form a clade. Ceratiola is nested within some members of subfamily Rhododendroideae (Stevens, 1971) of the Ericaceae. DISCUSSION That Empetraceae may be derived from within the Ericaceae has been suggested by Moore et al. (1970) and Takhtajan (1969). Takhtajan (1969) postulated a close relationship of Empetraceae with Rhododendroideae. Harborne (1969) and Moore et al. (1970) showed that several members of the Ericaceae (e.g., Rhododendron and Erica), as well as Ceratiola, Corema conradii (Torrey) Torrey and Empetrum possess the rare yellow flavonol, gossypetin 3-galactoside. Thus, this chemical fea- ture appears to unite these taxa. Rhododendroideae (Stevens, 1971) are paraphyletic, but the relation- ships among the members of this clade and those of the “* ” taxa (Calluna and Erica) are not resolved at maximum parsimony (Fig. 2). However, ericoid in at least some of the trees (Fig. 1) Calluna and Volume 80, Number 3 1993 Kron & Chase 739 Systematics of Ericaceae, Empetraceae, and Epacridaceae Clavija Theophrastaceae 6 ИЕ Manilkara Sapotaceae dl Lu. Camellia Theaceae д5 [2 Calluna 40 L29. Erica Befaria Ericaceae Elliottia Leiophyllum Ceratiola Е Empetraceae Rhododendron Cassiope 49 Chamaedaphne Zenobia . Daboecia Ericaceae _ Leucothoé Gaultheria Vaccinium Ex Cyathodes ш = Pentachondra Leucopogon Epacridaceae 14 Dracophyllum d»5 Epacris — i 19 = Arbutus Fri 42 d»5 Lz. Arctostaphylos е 5 " = Pyrola = Pyrolaceae dl Enkianthus _ | Ericaceae 8 | p Sarracenia Sarraceniaceae 41 42 L22.. колаша Byblidaceae 4 | p Clethra Clethraceae e 5 Cyrilla Cyrillaceae y Actinidia Actinidiaceae FIGURE 2. Decay of parsimony of the Ericaceae s.l. (Some outgroup taxa present in Fig. І were removed to reduce analysis complexity and time.) Numbers above branches are Fitch (unordered) lengths optimized with ACCTRAN of PAUP 3.0s. Numbers below branches are the decay indices or the number of steps less parsimonious at w ich the branch becomes a polytomy in the strict consensus tree of all trees at that length. A decay index of “d0” indicates that the branch is not present in the strict consensus tree of the 85 maximum parsimony trees; “dl” indicates that the branch is absent in the strict consensus tree of trees at maximum parsimony plus one step, etc. Erica form a clade sister to the “rhododendroid”” ri ә clade. This topology suggests that the “егїсо! type of leaf morphology found in Calluna, Erica, and Cassiope may be ancestral to the flat leaves of the “rhododendroid” taxa (Stevens, 1970). Within this clade there is also the change from gamopetalous corollas (Cassiope, Calluna, Erica, Rhododendron) to choripetalous corollas (Cera- tiola, Leiophyllum, Elliottia, Befaria). This dif- fers from the traditional view of choripetaly as the ancestral condition within the Ericaceae. e "epacrids" are monophyletic in this analysis and are in the same clade as the “vaccinioid” taxa and Chamaedaphne. The “epacrid” clade is more robust (with a decay index of 5, Fig. 2) than the clade containing “‘vaccinioid” taxa (decay index of 3, Fig. 2). More sampling of taxa traditionally included in tribes Andromedeae and Vaccineae 740 Annals of the Missouri Botanical Garden (Vaccinioideae) is necessary before their relation- ships can be confidently evaluated. Arbutus and Arctostaphylos (tribe Arbuteae, Vaccinioideae, Stevens, 1971) form a monophy- letic group that branches before the clade leading to the bulk of the ericads. Pyrola is sister to the Arbuteae clade plus the above ericads. However, the relationships of Pyrola and Arbuteae (repre- sented by Arbutus and Arctostaphylos) to the rest of the Ericaceae (except Enkianthus) are not ro- bust in this analysis as the branch collapses in trees one step longer than maximum parsimony (Fig. 2). However, the branch leading to Pyrola plus the above noted ericads is well supported (decay index of 75 steps), indicating that Pyrola should not be recognized as a distinct family (see Judd & Kron, 1993, for supporting morphological evidence). Even though it appears that the achlorophyllous Monot- ropa lacks rbcL (repeated attempts to amplify the gene were not successful, see also Wolfe et al., 1992, for another achlorophyllous plant without a functional rbcL gene), this mycoparasitic taxon and its relatives should also be included (along with Pyrola, Chimaphila, etc.) in a broadly defined Ericaceae (Judd & Kron, 1993). Enkianthus is sister to the rest of the Ericaceae (Fig. 1). This was also found in the shortest tree of Judd & Kron (1993) and A. Anderberg (pers. comm.), in studies based primarily on morphology. Enkianthus has been considered relatively isolated due to its combination of gamopetalous campan- ulate corolla, anthers opening by slits rather than pores, and pollen shed in monads rather than the tetrads characteristic of Ericaceae (Stevens, 1971). However, such a distinctive phylogenetic position has not been suggested previously. The position of Enkianthus as sister to the remaining Ericaceae suggests that the interpretation of a corolla with free petals as primitive (Abbott, 1936; Camp, 1941; 1943; Stevens, 1971; Wood, 1961) within the family needs reevaluation. In fact, care- Copeland, ful morphological studies (Copeland, 1943; Leins, 1964) show that at least in the case of members of Rhododendron subsect. Ledum (Kron & Judd, 1990), the corolla is actually slightly fused at the base early in development. The choripetalous con- dition may be derived within the Епсасеае in gen- era such as Befaria and Leiophyllum, in which the petals appear to be truly separate. However, since the sister clade to the Ericaceae s.l. may be choripetalous, it is possible that gamopetaly in Enkianthus is autapomorphic. Whereas the general results obtained in this study (i.e., the inclusion of Empetraceae, Epac daceae, and Pyrolaceae in a broadly defined E caceae) are supported by a cladistic analysis of morphological characters (Judd & Kron, 1993), the positions of Cyrilla and Clethra in the molec- ular analysis (Fig. 1) differ from those revealed in the morphological analysis. This is surprising be- cause a close relationship of Clethra (Clethraceae) and Cyrilla (Cyrillaceae) to the Ericaceae appears to be well supported by morphology. These two genera have been included within the order Ericales by some (Cronquist, 1981; Dahlgren, 1983), al- though Thorne (1992) placed them in his Theales. In the morphological analysis of Judd & Kron 1993), Actinidia is sister to Cyrilla + Clethra + Ericaceae s.l. These taxa share the presence of an apical depression in the ovary and anthers that invert in development (modified in Cyrilla). In the molecular trees, the relative positions of Ac- ~ tinidia, Cyrilla, and Clethra may ђе resolved more fully upon inclusion of other representatives of the Cyrillaceae (Cliftonia, Purdeaea). Anderberg's (1992) cladistic study of the relationships of the Ericales to other "higher" dicots indicated Cyril- laceae and Clethraceae as branching before Roridu- laceae, which is sister to Epacridaceae + Pyrolace- ae + Ericaceae + Monotropaceae. These results also indicate a closer relationship of Cyrillaceae and Clethraceae to the Ericaceae s.l. than do the molecular data. However, neither the Anderberg (1992) nor the Judd & Kron (1993) studies in- cluded members of the Ebenales (sensu Cronquist, 1981) in their analyses. SUMMARY Sampling of taxa presents a major problem in 1993; Chase et al., 1993), and when taxa are sampled superficially molecular studies (Olmstead et al., spurious results are possible. We advocate and maintain a fair degree of suspicion concerning these results in two weakly supported portions of the topology and believe that increased sampling is likely to alter the estimates of relationships. One of these is the group identified as sister to Ericaceae s.l., in which significant divergence from other stud- ies occurs (discussed above). The other area is within the Ericaceae (Pyrola and Arbutus + Arc- tostaphylos); we would expect weakly supported branches to be altered when a more even sampling is obtained. The strong internal support and con- gruence with results of other studies (Judd & Kron, 1993) for the positions of the segregate families (Empetraceae, Epacridaceae) make it unlikely that the general conclusions of this study will change as taxon sampling improves A monophyletic Ericaceae includes Empetra- Volume 80, Number 3 1993 Kron & Chase 741 Systematics of Ericaceae, Empetraceae, and Epacridaceae ceae, Epacridaceae, and Pyrolaceae (formally out- lined in Judd & Kron, 1993). These conclusions are congruent with parsimony analysis of morpho- logical data (Judd & Kron, 1993), which also in- dicates the inclusion of Monotropaceae in Erica- ceae (cf. Anderberg, 1992). The agreement of these studies lends increased support to recognizing a more broadly defined Епсасеае. LITERATURE CITED ABBOTT, C. L. 1936. The phylogeny of the Ericales. Trillia 9: 62-69. ALBERT, V. A., M. W. Снаѕе & B. D. MISHLER. 1993. Character- state weighting for cladistic analysis of pro- tein- е DNA sequences. Апп. Missouri Bot. Gard. 80: um "n ae 1992. The circumscription of the Ericales, and their cladistic relationships to other families of “higher” dicotyledons. Syst. Bot. 17: 660- BREMER, К. 1988. The limits of amino acid sequence ata in А он reconstruction. Evo- lution 42: 795-8 Camp, W.H. 1941. m in the Ericales. A discussion of the genus VE ha in aia America. Bull. Torrey Bot. Club 68: СНАЗЕ, M. W., D. E. UE К. С. OLMSTEAD, р. MORGAN, D. H. Les, B. D. MisiiteR, М. R. DUV ALL, R. A. Hırs, Y.-L. Qiu, К. A. eee Ји E. ш S. C. H. BARRETT, S. DAYANANDAN & V. А 1993. E of seed plants: An e of nucleotide sequences from the po gene rbcL. . Missouri т Gard. 80: 528 COPELAND, H. F. 1943. A study, шла and tax- onomic, of ы к of Rhododendroideae. Amer. Midl. Naturalist 30: nu 33-625 к, А. An Integrated System of Classi- n of Flowering ке, Columbia Univ. Press, w pol Я DAHLGREN, К. . А system of нит и of the angiosperms to be used to demonstra Ae oe bution of pe Bot. Ne 128: " 198 neral aspects of sperm de tion and a Nordic J. Bot. 3: 119- DovLE, J. & J. роте. 1987. rapid DNA isolation procedure for small ro of fresh leaf tissue. Phytochem. Bull. 19: HARBORNE, J. B. 1969. с and herbacetin as taxonomic markers in higher plants. Phytochemistry 8: 177-183. Hurrorp, L Rosidae and their relationships to other nonmagnoliid seis aed 2 ylogenetic analysis cal data. Ann. Missouri Bot. Gard. 79: EVE Јорр, W. 5. & K. A. KRoN. 1993. Circumscription of Ericaceae (Ericales) as determined by preliminary cladistic analyses based on morphological, anatomi- cal, and Pw ip d features. Brittonia (in press). Kron, K. А. & W. S. Jupp. 1990. Phylogenetic re- lationships MM the Rhodoreae (Ericaceae) with specific comments on the placement of Ledum. Syst. Bot. 15: 57-68 Lens, Р. 1964. -_____-_ studien an Ericales-bluten. Bot. Jah бот n D. R. 91. The d b importance eae ges of most-parsimonious trees. Syst. 8. Zool MADDISON, E4 P. . J. DONOGHUE & D. R. Mappisow. 984. Outgroup analysis and parsimony. Syst. Zool. 33: 83-103. MISHLER, B. D., M. J. DONOGHUE & V. A. ALBERT. 1991. The de index as a measure of relative robustness within a Антене e Willi Hennig Society Meet- ing, Toronto. [Abstrac MOORE, D. M., J. B. СРЯ A. WILLIAMS. 1970. Chemotaxonomy, variation and geographical e bution of the Empetraceae. Bot. J. Linn. Soc. Novacek, М. р 1991. “All tree histograms” in the evaluation of cladistic evidence: Some ambiguities. -349. B. BREMER, К. M. Scorr & J. D. у · 1993. A parsimony analysis of the As- teridae sensu lato based on rbc ~ sequences. Апп. Missouri Bot. Gard. 80: 700 STEVENS, P. 1970. ve ns e, and Harri- manella: A taxonomic and evolutionary problem. New E 69: 1131-1148. A classification of the Meu Sub- mats and tribes. Bot. J. Linn. Soc. 64: 1-53. SworrorD, D. L. 1991. PAUP: vx enin Analysis Using Parsimony, n 3.0s. Illinois Natural His- tory Survey, Champaig TAKHTAJAN, A. L. 1969. овалне Plants, Origin and Dispersal. Oliver Boyd, Edinbur 980. Outline of the ur of flowering plants lar ake Bot. Rev. 46: 225- THORNE, R. 1983. Proposed gu realignments in the Pe Nordic J. Bot 117 1 Classification апа geography of the flowering plants. Bot. Rev. 5 348. WorrE, К. H., С. W. MORDEN, 5, С. Ems & J. D. PALMER. Rapid evolution of the plastid al apparatus in a nonphotosynthetic plant: Loss or ac- celerated sequence evolution of tRNA and DM protein genes. J. Molec. Evol. 35: 304-3 Woop, C. E., JR. 1961. The genera of Ericaceae i in the sauibeudem United States. J. Arnold Arbor. 4 «BD. INTERFAMILIAL RELATIONSHIPS OF THE ASTERACEAE: INSIGHTS FROM rbcL SEQUENCE VARIATION! Helen J. Michaels, Kathy M. Scott,’ Richard С. Olmstead,* Tim Szaro,’ Robert K. Jansen,° and Jeffrey D. Palmer? ABSTRACT ucleotide sequences of the chloroplast gene rbcL were analyzed to examine relati onships among the large, N distinctive family Asteraceae and eight putatively closely е families. Phylog enetic analysis of a total of 24 G sequences of rbcL identified a lineage consisting of two families p to the Asteraceae. In addition, a strong aceae, Corokia (Cornaceae sensu Cronquist), Menyanthaceae, Lobeliaceae, and Campanulaceae was dito These results clearly distance from the Asteraceae certain | groups previously considered closely related; ph ver, the results support alt о» ни чан pollen morphology, chemistry, and pollen- сао mechanisms. The angiosperm family Asteraceae has long been “natural” with well-established limits defined by several spe- recognized as one of the large, families cialized floral characteristics and distinctive sec- ondary chemistry. Many recent studies have illu- minated numerous phylogenetic controversies within the family (Bremer, 1987; Bremer et al., 1992; Jansen et al., 1990, 1991а, b; Jansen & Palmer, 1987a, b, 1988; Karis et al., 1992; Keeley & Jansen, 1991; Kim et al., 1992; Watson et al., 1991). However, relationships among the Aster- aceae and other families have remained obscure, due to considerable parallel and convergent evo- lution of conventional characters used to infer af- finities, lack of recent studies employing modern methods of phylogenetic analysis, and substantial confusion as to relationships among the various families within the subclass Asteridae itself (but see Olmstead et al., 1992). At least 12 families have been proposed as clos- est relatives of the Asteraceae based on a variety of traditional taxonomic characters. Although Hutchinson (1969) noted the superficial similarity of the inflorescence of the Dipsacales (Caprifolia- ceae, Valerianaceae, and Dipsacaceae) to that of the Asteraceae, he suggested convergence as the basis for this and identified the Campanulales (Cam- panulaceae and Lobeliaceae) as the closest relative. Aspects of the distinctive chemistry of the Aster- aceae (e.g., alkaloids, polyacetylenes, terpenes, in- ulin for carbohydrate storage) have been noted in several members of the Campanulales (Campanu- laceae, Lobeliaceae, Goodeniaceae, Stylidiaceae), while other chemical evidence has pointed to the Apiaceae and Araliaceae (Hegnauer, 1964, 1977). Cronquist (1955) advocated the Rubiales as closest relatives of the Asteraceae (grouped with the Gen- tianales in the system of Takhtajan, 1980), but also acknowledged strong similarities in floral and inflorescence morphology between the Asteraceae and the Calyceraceae (as did Takhtajan, 1980). An association with the Calyceraceae is also sup- ported by biogeography and capitular structure (Turner, 1977) and pollen morphology (Skvarla et al., 1977). Others (reviewed in Skvarla et al., 1977) have noted a palynological resemblance of Vale- rianaceae, Goodeniaceae and Brunoniaceae to As- teraceae. In a morphological cladistic study of tribal relationships within the Asteraceae, stylar mor- phology, chemical characters, and pollen-presen- ' We thank D. эме апа Р, a for providing an unpublished rbcL sequence of Corokia. We are grateful to M. Chase, D. Soltis, and T. Lam ers for comments on the ical is ciences, Bowling Green State University, Bowling Green, Ohio 43403, U.S.A. na University, Bloomington, Indiana 47405, U.S.A Biolo из University of Colorado, oulder, Colorado Mere U.S.A. Р). > Department of Plant Pathology, University of California, Berkeley. California 94720, U.S.A. * Department of Botany, University of Texas at Austin, Austin, Texas 78713, U.S.A ANN. Missouni Bor. Garp. 80: 742-751. 1993. Volume 80, Number 3 1993 Michaels et al 743 Interfamilial Relationships of Asteraceae tation mechanism were identified by Bremer (1987) as potential synapomorphies linking the Lobeli- aceae and Asteraceae. Finally, strong similarities to the highly specialized secondary pollen-presen- tation mechanism of the Asteraceae have been documented in Goodeniaceae, Brunoniaceae, Cam- panulaceae, and Lobeliaceae (Leins & Erbar, 1990). As noted in many of the above attempts to resolve this controversy, most of the morphological or chemical characters that support a particular hy- pothesis of ancestry are often also found in several related groups, and some must certainly be the product of parallel evolution. Several of the recent molecular advances in elucidating phylogenetic relationships within the family have employed restriction site analysis of chloroplast DNA. However, this approach is gen- erally unsuitable at the interfamilial level because the homology of site changes becomes doubtful due to increased levels of both nucleotide sequence divergence (causing multiple **hits" within restric- tion sites) and length variation causing problems in alignment of sites (Downie & Palmer, 1992b; Palmer et al., 1988). At higher taxonomic levels, restriction site analysis of only the more conserved inverted repeat region of chloroplast DNA may be used to circumvent these problems (e.g., Downie & Palmer, 199223), but fewer characters are gen- erated than in whole genome surveys. DNA se- quence analysis of the slowly evolving chloroplast gene rbcL and nuclear rRNA genes has proven highly effective in resolving higher-level relation- ships in plants (Chase et al, 1993; Hamby & Zimmer, 1992; Palmer et al., 1988; Ritland & Clegg, 1987; Zurawski & Clegg, 1987). In par- ticular, recent studies by a number of researchers employing comparative sequencing of the chloro- plast gene encoding the large subunit of the pho- tosynthetic enzyme ribulose- 1 ,5-bisphosphate car- boxylase (rbcL) indicate an appropriate size and rate of evolution for providing a sufficient number of characters for phylogenetic studies at the familial and ordinal levels (Donoghue et al., 1992; Kim et al., 1992; Olmstead et al., 1992; Soltis et al., 1990). In this paper we analyze nucleotide se- quences for rbcL from representatives of the As- teraceae and eight putatively related families to determine evolutionary relationships among the families and identify the sister group of the Aster- aceae. MATERIALS AND METHODS New rbcL sequences were determined for six taxa in the Asteraceae and eight representatives of putatively closely related families in the subclass Asteridae. For these sequences, fresh leaf material obtained either from the field or seedlings was used to isolate DNA as purified chloroplast DNA by the sucrose gradient method (Palmer, 1986) or as total cellular DNA by a modified CTAB procedure (Doyle & Doyle, 1987) followed by CsCl gradient purifi- cation. The rbcL gene was isolated for cloning by one of two methods. (1) For most taxa, fragments containing the entire rbcL gene were gel-isolated from either Sac I or Sac I/BamHI digests and ligated into the plasmid vector Bluescript (Strata- gene, Inc., LaJolla, California). Recombinant, rbcL- containing colonies were confirmed by Southern hybridization to cloned rbcL fragments from peas. The coding region was sequenced from a single- stranded template by the dideoxy chain termination method (Sanger et al., 1977) using a series of primers based on rbcL sequences from maize and spinach (obtained from G. Zurawski, DNAX). (2) Sequences from Pentas and Boopis were obtained following amplification and cloning of a double- stranded fragment using the polymerase chain re- action following the methods of Olmstead et al. Preliminary analyses included 14 new sequences (see Table 1 and Appendix to this issue) and one previously published sequence from Flaveria in the Asteraceae (Hudson et al., 1990) and, to serve as outgroups, sequences from Spinacia in the Car- yophyllidae (Zurawski et al., 1981) and Nicotiana in the Asteridae (Shinozaki et al., 1988). These analyses (Michaels & Palmer, 1990) on only a subset of the data reported here identified a closest sister group identical to the present expanded anal- ysis. The results of concurrent studies of rbcL sequences in the Saxifragaceae (Soltis et al., 1990), Asteridae (Olmstead et al., 1992), and 499 angio- sperms (see Chase et al., 1993) have motivated the inclusion of data from other outgroups and from taxa not previously suspected to be associated with the Asteraceae. Data for Heuchera (Soltis et al., 1990), Magnolia (Golenberg et al., 1990), Villarsia, Menyanthes, Hedera, and Coriandrum (Olmstead et al., 1992) were obtained from pub- lished reports, whereas an unpublished sequence for Corokia was made available by D. Morgan and D. Soltis. Although the coding regions for these taxa ranged from 1428 to 1458 bp, only a 1428 bp region was analyzed because no major insertions or de- letions were found in this region, allowing alignment by eye, and because homology of positions beyond 1428 is uncertain. Analyses over longer sequences did, however, produce virtually the same results. 744 Annals of the Missouri Botanical Garden Species of Asteraceae, related families, and outgroups compared by rbcL sequence. * — New sequence dy. TABLE 1. obtained for this study Asteraceae Barnadesioideae B Dasyphyllu Chicorioideae Cartham Asterioideae arnadesia caryophylla (Vell.) S. F. Blake* m dicanthoides (Less.) Cabrera* Lactuca sativa L.* nus tinctorius L.* Flaveria trinervia Mohr anthus annuus L.* a mikanioides Otto* Related families Caprifoliaceae Viburnum acerifolia L.* Dipsacaceae Dipsacus sativus Honck.* Valerianaceae Valeriana officinalis L.* biaceae Pentas lanceolata K. Schum.* Goodeniaceae Scaevola frutescens Krause* Lobelia erinus He dera helix Иш thes tri Menyanthaceae Cornaceae Outgroup families Chenopodiaceae Solan PAM SE A өш ae Nicotiana ta Campanula ramosa Sibth. & Smith* Boopis anthemoidas Jussieu* iandrum sativum L. oliat Villarsia calthifolia P Muell. Corokia macrocarpa T. Kirk Spinacia oleracea L. Heuchera micrantha Douglas ex Lindl. Magnolia mac ин Michx. Analyses were also conducted in which transitions and transversions were differentially weighted (1.0: 1.3, Albert et al., 1993). Since weighted analyses did not alter the tree topologies, only the results of the equally weighted analyses are reported here. AUP, Phylogenetic Analysis Using Parsimony, Version 3.0r (Swofford, 1991), was used to conduct Fitch parsimony analyses (Fitch, 1971). The HEU- RISTIC search option with 10 random replicates (Maddison, 1991) was used for tree building and branch swapping under the MULPARS and TBR options. Strict consensus trees were computed. To assess robustness of clades identified in the anal- yses, the bootstrap method (based on 500 repli- cates; Felsenstein, 1985) was used. In addition, a "decay" analysis was conducted (cf. Bremer, 1988; Donoghue et al., 1992; Hillis & Dixon, 1989) in which strict consensus trees were generated from all trees five steps longer than the most parsimo- nious ones. The F R TREES option was then used to identify all trees at each shorter tree length. Support for each clade was assessed by determining when each clade was no longer resolved in con- sensus trees from progressively less parsimonious solutions. RESULTS Of the 1,428 nucleotide positions compared among the 24 rbcL sequences in the analysis, 452 were variable. Exclusion of autapomorphies re- sulted in 278 potentially synapomorphous char- acters, 204 of which were at third-codon positions, 45 at first, and 29 at second. The third position characters are essential sources of phylogenetic information, as their elimination from the data set results in little resolution (data not shown). Initial analyses (results not shown) used four taxa [ Mag- попа (Magnoliidae), Spinacia (Caryophyllidae), Heuchera (Rosidae), and Nicotiana (Asteridae) | as designated outgroups. These analyses established Nicotiana as the sister group to an ingroup con- sisting of the taxa proposed as outliers to Astera- ceae. In subsequent analyses, the more distant outgroups Magnolia, Spinacia, and Heuchera were deleted, and trees were arranged using Nicotiana Volume 80, Number 3 1993 Michaels et al. ed Relationships of Asteraceae 745 alone as the designated outgroup. The heuristic search found four equally parsimonious trees (re- sults not shown) of 833 steps with consistency index (C.I.) of 0.48 (autapomorphies excluded) and a retention index (R.I.) of 0.502. The strict consen- sus of the four trees (Fig. 1) shows a large, nested monophyletic group consisting of the Asteraceae, Goodeniaceae, Calyceraceae, Corokia (Cornaceae 1981), Menyanthaceae, Lobeli- aceae, and Campanulaceae. The two polytomies sensu Cronquist, exist in either a more basal portion of the tree or within the Asteraceae. Among the putative relatives of the Asteraceae examined here, Pentas (Rubi- aceae) was placed as the most basal clade. As in our earliest analysis with only a subset of these data, the expanded analysis identifies a clade con- sisting of Scaevola (Goodeniaceae) and Boopis (Calyceraceae) in the most proximal position to the Asteraceae. This sister relationship is supported by eight shared nucleotide substitutions. e robustness of the sister relationship of the Goodeniaceae-Calyceraceae clade to the Astera- ceae is indicated by two subsequent analyses. First, the results from 500 bootstrap replications using the heuristic search option indicate that this rela- en is one of the more robust areas of the tree topology, occurring in 8446 of the trees (Fig. 2). Relationships are, with some exceptions, poorly resolved within the Asteraceae (also see Kim et al., 1992) and also outside the "Asterales" (Lobeli- aceae-Campanulaceae through Asteraceae) clade. Within the “Asterales” ported groups besides the Asteraceae Goodeni- are three other well-sup- aceae-Calyceraceae clade: (1) the Asteraceae emerge as monophyletic in 97% of the bootstrap trees, (2) together in all bootstrap replicates, (3) the Goodeni- Campanulaceae and Lobeliaceae occur aceae and Calyceraceae clade is monophyletic 86% of the time. Finally, the “Asterales” whole appears in 86% of the trees. group as a Second, the strength of the relationship of the the Astera- ceae is indicated in the decay analysis by the per- Goodeniaceae-Calyceraceae clade to sistence of this group in less parsimonious solutions. The sister position of this clade is maintained in 127 trees up to two steps longer than the most The s three additional steps places the Goodeniaceae— parsimonious trees. trict consensus tree at Calyceraceae clade within an unresolved polytomy together with various unresolved lineages of the Asteraceae, while the Asteraceae—Goodeniaceae— Calyceraceae group forms an unresolved polytomy with the rest of the ““Asterales” kia, Menyanthes, Campanula- Lobelia). When the 964 trees up to four steps longer are included, (Villarsia, Coro- most of the resolution in the tree is lost, and all but the strongest clades (i.e., Goodeniaceae- Cal- yceraceae, Campanulaceae- Lobeliaceae, Valeria- naceae-Dipsacaceae, and Helianthus- Flaveria) disintegrate. DISCUSSION The rbcL data clearly indicate a close relation- ship among the Asteraceae and other families tra- ditionally placed in the Asterales and Campanu- lales. This conclusion is supported not only in the local analysis presented here, but also in much broader analyses of the Asteridae (Olmstead et al., 1992, 1993) and in an analysis of 499 rbcL se- 1993). The Lobeliaceae, quences from seed plants (Chase et al., placement of the Campanulaceae, Goodeniaceae, Calyceraceae, and Asteraceae in a single monophyletic group is consistent with the system of relationships proposed by Takhtajan (1987) and Thorne (1992), in which these groups (together with families in the Stylidiales) form the superorder Asteranae. This grouping is also distin- guished in the system of Wagenitz (1977) and the recent review by Lammers (1992), but not in the systems proposed by Dahlgren (1975) or Cronquist (1981). Dahlgren aligned the Goodeniaceae and the Calyceraceae far from the Asteraceae, placing them in the Gentiananae and Cornanae, respec- tively. The topology inferred from the rbcL data is most divergent from the Cronquist system. Al- though Cronquist placed the Goodeniaceae togeth- er with the Campanulaceae and other families to form Campanulales, the Calyceraceae were as- signed to the Dipsacalean group, while the Aster- aceae were placed nearest the Rubiaceae. The lat- ter association is clearly not supported by the rbcL data; Pentas (Rubiaceae) was consistently placed in the most distant position of all the putatively allied families under consideration as sister groups in this analysis. The sister-group relationship of the Goodeni- aceae-Calyceraceae clade to the Asteraceae is sup- ported by evidence from a number of studies. The pollen-presentation mechanisms of the Goodeni- aceae, Calyceraceae, and Asteraceae are similar, Campanulaceae, (Erbar & Leins, 1989; Leins & Erbar, 1990; Lam- mers, 1992; Wagenitz, 1992). Details of floral development (Harris, 1991; Erbar & Leins, 1989) also unite both Goodeniaceae and Calyceraceae with the Asteraceae. However, the particular de- velopment pattern they share is also found in Bru- noniaceae, Campanulaceae, Stylidiaceae, Menyan- 746 Annals of the Missouri Botanical Garden Nicotiana SOLANACEAE ——— Viburnum CAPRIFOLIACEAE Valeriana VALERIANACEAE | __ Dipsacus DIPSACACEAE Hedera APIACEAE E Coriandrum ARALIACEAE Senecio = Lactuca ~ Helianthus Flaveria ASTERACEAE E Carthamnus Barnadesia B Dasyphyllum _ $саеуо!а GOODENIACEAE Е Hao CALYCERACEAE EE Corokia CORNACEAE Villarsia јЈмемудутнаселе — Menyanthes Campanula CAMPANULACEAE — PEDEN LOBELIACEAE Pentas RUBIACEAE HA 20 Substitutions FIGURE 1. Strict consensus of four equally parsimonious trees (length = 833 steps, C.I. 48, КЛ. = 0.50) based upon rbcL sequences. Branch lengths are proportional to the number of nodis aad supporting a node or distinguishing a terminal lineage (note scale at ‚ка thaceae, Rubiaceae, and families of the Dipsacales Asteraceae, Calyceraceae, and Goodeniaceae, as (Erbar & Leins, 1989), and therefore this is prob- well as Menyanthaceae, Campanulaceae, and ably a plesiomorphic character. Chemical char- beliaceae; Pollard & Amuti, 1981) and phenolics acters such as carbohydrate storage as inulin (in — (e.g., caffeic acid in Asteraceae, Calyceraceae, and Volume 80, Number 3 Michaels et al. 747 1993 Interfamilial Relationships of Asteraceae Nicotiana SOLANACEAE Viburnum CAPRIFOLIACEAE 43 Oh а Valeriana VALERIANACEAE 28| > Dipsacus DIPSACACEAE О Ss Hedera APIACEAE M Coriandrum ARALIACEAE Senecio | Lactuca Helianthus 97 92 — Carthamnus >4 +3 73 Barnadesia cl yal +? Dasyphyllum = 37 86 Scaevola GOODENIACEAE +1 >4 B а oopis CALYCERACEAE 30 +1 Corokia CORNACEAE 68 2] Villarsia MENYANTHACEAE 86 Menyanthes 44 ш Сатрапи!а CAMPANULACEAE i Lobelia LOBELIACEAE Pentas RUBIACEAE H 20 Substitutions FIGURE 2. One of the four equally glas d trees based upon rbcL sequences with the results of the bootstrap and decay ai analyses. Branch lengths are proportional to the number of nucleotide substitutions supporting a node or distinguishing a terminal lineage (note scale at eas Bootstrap values from 500 replications are indicated above the branches. Numbers below indicate numbers of additional steps needed for a branch to collapse. 748 Annals of the Missouri Botanical Garden Goodeniaceae, as well as Menyanthaceae and Cam- panulaceae; Lammers, 1992) further support the position of the Goodeniaceae and Calyceraceae. Asteraceae, Goodeniaceae, and Calyceraceae also lack endosperm haustoria, have binucleate tapetal cells, and produce herbivore defenses through the mevalonate pathway, characters that are also shared with Menyanthaceae, but are absent from Cam- panulaceae and Lobeliaceae (Lammers, 1992). Several other characters specifically support the sister placements of Goodeniaceae or Calyceraceae alone. Polyacetylenes, which are characteristically found in the Asteraceae, have also been reported from some Campanulaceae (Ferreira & Gottlieb, 1982), Lobeliaceae, and one Goodeniaceae (Lam- mers, 1992). Their distribution in Calyceraceae, Menyanthaceae, or other families that place near the Asteraceae in our analysis is unknown. The Calyceraceae were propose ed asa sister group to the Asteraceae by Jeffrey (1977) port of the placement of Calyceraceae seen in our . Evidence in sup- rbcL analyses derives from chloroplast DNA re- striction site data (Downie & Palmer, 1992a; un- fortunately, due to their unusual genome organi- zation, Goodeniaceae could not be included in this study) and floral development, as noted above, but also from pollen morphology, capitulum structure, and biogeography. Skvarla et al. (1977 that the ultrastructure of Calyceraceae exine is nearly identical to that of the Asteraceae, while ) concluded the Goodeniaceae are among five families listed with pollen similar enough to “suggest distinct link- ages." As previously advocated by Turner (1977), the structurally homologous capitula and floral fea- tures, and congruent core distributions of the As- teraceae and Calyceraceae in South America are particularly compelling arguments supporting a shared phylogenetic history. This analysis suggests a particularly strong sis- ter-group relationship for the Goodeniaceae and Calyceraceae, occurring in 86% of the bootstrap trees. This clade also remained intact throughout the decay analysis, a level of association seen only in other groups that have traditionally been strong- ly linked (e.g., Helianthus and Flaveria in the Heliantheae; Campanulaceae and Lobeliaceae; Dipsacaceae and Valerianaceae). As noted above, the Goodeniaceae and Calyceraceae are united by several characters, including pollen-presentation mechanism, floral development, pollen embryolo- , and secondary chemistry (e.g., seco-iridoids; Lammers, 1992). The rbcL trees and above evi- dence supporting the placement of the Goodeni- aceae (a small family of 325 species distributed primarily in Australia and Tasmania) and the South American Calyceraceae are consistent with Tur- ner's (1977) proposal of a Gondwanaland origin for the Asteraceae. Indeed, all the closest groups from Menyanthaceae to Goodeniaceae are distrib- uted а in the Southern Hemisphere. rbcL gene was recently sequenced from two families previously suggested to have affinities with the Asteraceae (Olmstead et al., 1992). Both the Apiaceae (Hedera) and Araliaceae (Coriandrum) have been proposed as close associates of the As- teraceae based upon shared phytochemistry (Heg- nauer, 1964, 1977). However, Lammers (1992) has argued that the synthesis of sesquiterpene lac- tones in these groups is due to parallel evolution. Parsimony analyses of rbcL indicate that, although neither is closely related to the Asteraceae, these groups do belong in the Asteridae rather than in their more traditional position in the Rosidae (Olm- stead et al., 19 3) Although neither Menyanthaceae nor Corna- the Asteraceae hase et al., ceae have ever been allied wit based on traditional data, they were included in this study because other recent chloroplast DNA analyses suggested close relationships. Based on rbcL sequences, Olmstead et al. (1992) found an unexpected association of the Menyanthaceae with the families more typically suggested to have af- finities with the Asteraceae. A survey of restriction sites in the chloroplast DNA inverted repeat (Dow- nie & Palmer, 1992a) also places the Menyan- thaceae along with the Calyceraceae and Astera- ceae. e most parsimonious solutions from Olmstead et al. (1992, 1993), which are based on a much larger sampling of taxa within the Aster- idae, are consistent with the results reported here. Although Menyanthaceae are nearer to the Aster- aceae than either the Campanulaceae or Lobeli- aceae in this study, yceraceae retain the sister position (Goodeniaceae were not included in the Asteridae analyses of either Olmstead et al., 1992, or Downie & Palmer, 1992a). Several features shared by Menyanthaceae, Calyceraceae, and Goodeniaceae (the production of seco-loganin, car- bohydrate storage as inulin, presence of multinu- cleate tapetal cells, absence of endosperm haus- toria, and chromosome numbers based upon x — 1992) are consistent with the placement of Menyanthaceae in these analyses. In addition, the analyses of & Soltis (1993) and Chase et al. (1993) indicate an un- suspected affiliation of Corokia (Cornaceae sensu lato) with the asteralean clade. The association of Corokia with Asteraceae and its near relatives was 8 or 9; Lammers, organ Volume 80, Number 3 Michaels et al. 749 1993 Hana Relationships of Asteraceae confirmed in our analyses, which are more likely LITERATURE CITED to produce optimal solutions, while the computa- 1. W. Coase & B. D. MIsHLER. 1993. as fficulties of the much larger analyses (Chase 1993; Morgan & Soltis, 1993) may have precluded discovery of shortest trees. This radical et al., departure from previous placements of Corokia (in Escallonioideae, Saxifragaceae sensu lato, Engler, 1928; in Cornaceae, Eyde, 1966 and Cronquist, 1988; or in Araliaceae, Phillipson, 1967) is con- sistent with several other characters: (1) biogeog- raphy (Corokia is distributed primarily in New Zealand and Australia; Eyde, 1966), (2) some as- pects of morphology (inferior ovary, locules 1-3 with a single, apical, unitegmic ovule, and multi- cellular, T-shaped trichomes, found elsewhere only in the Asteraceae and three other unrelated fam- ilies; Eyde, 1966), and (3) chemistry (presence of iridoids in Escallonioideae; Dahlgren et al., 1981). See also Morgan & Soltis (1993) for further dis- cussion of evidence linking genera of the Escallon- ioideae with Asteridae. These unanticipated asso- ciations not only illustrate the heuristic value of the broader analyses and the benefits of widespread data sharing, but also motivate the acquisition of additional data from other sources that may cor- roborate these new hypotheses of relationships. For example, studies of floral developmental patterns, pollen ultrastructure and development, and em- bryogeny in Menyanthaceae and Corokia would be a logical direction for future work. Finally, although many families that have been previously considered as serious contenders for sis- ter group to the Asteraceae have been investigated in this study, several others remain to be examined. In particular, Brunoniaceae and Stylidiaceae have been associated with Asteraceae and Campanula- ceae in Takhtajan's (1987), Wagenitz's (1977), and Thorne's (1992) classifications. Both are dis- tributed primarily in Australia and share the floral developmental features found in Asteraceae, Cam- panulaceae sensu lato, Goodeniaceae, and Caly- ceraceae (Erbar, 1991). Brunoniaceae also possess mechanism to those found in the bist three of these families (Leins & Erbar, 1990). Any comprehensive attempt to fur- ther explore the origins of the Asteraceae should include several other, poorly understood small fam- dac tati a similar pollen > ilies that have also been associated with Campan- ulales (Pentaphragmataceae, Cyphiaceae, and Sphenocleaceae; Lammers, 1992). The addition of data from these groups is likely to provide new insights into the phylogeny of the Asteraceae and further resolve the current picture of relationships of this distinctive lineage. ALBERT, V. A., Character-state weighting for cladistic analysis of pro- tein-coding sequences: Prospects, limitations, and an example from Bis rbcL gene. Ann. Missouri Bot. Gard. 80: 766. BREMER, K. 1987. Tribal relationship of the As- teraceae. Cladistics 3: 210-253 1988. The limits of amino aaa sequence data in cap паран phylogenetic reconstruction. Evolution 2: T. "CK p P. O. Karis, M. KALLERSJO, S. С. pn K. J. Kim, H. J. Micuaets, J. D. PALMER & R. S. WALLACE. 1992. A review of the phylog- eny and classification of the Asteraceae. Nordic Bot. 12: 141-148. CHasE, M. W., D. E. SoLTIS, R. C. OLMSTEAD, D. MORGAN, D ^c 7 етп E. Conti, J. D. PALMER, J. R. МАХНАКТ, К. J. Syrsma, Н. J. MICHAELS, W. J. Kress, К. С. KAROL, W. D. Ставк, M. HEDREN, B. S. GAUT, К. K. JANSEN, K E Kim, C. Е. WiMPrEE, J. F. aa G. R. е . Н. STRAUSS, Q.- Y. XIANG, ы PLUNKETT, Р ace S. SWENSEN, S. ES р ; ee UINN, L. E. RAE ч GOLENBERG, de H. L ; 9. u^ GRAHAM, S. C. H. Barrett, 5. Day ANANDAN а V. А. ALBERT. 1993. cee of seed plants: An ae sis of nucleotide sequences ur ne е gene rbcL. Ann. Missouri Bot. Gard. 80: 528-5 CRONQUIST, А. 1955 Па and taxonomy of the a Amer. Midl. Naturalist 53: 478-51 An ie System of Сна of Paris Plants. Columbia Univ. Press, New York. 1988. The evolution and classification of flow- ering plants, 2nd ed. New York Botanic Garden, ork. Dia REN, R. 1975. A system of classification of the angiosperms to be used to demonstrate the distri- bution of се Bot. Not. 128: 181-197. & D. J. NIELSEN. 198 1. A revised classification of be angiosperms with comments on rrelation between chemicals and other char Pp. 1 49-199 in D. A. Young & D. S. Siegler (ed. itors), Phytochemistry and Angiosperm Phylogeny. Praeger, New k. DONOGHUE, M. D ., В. С. OLMSTEAD, J. F. SMITH & J. D. 1992. PALMER. Phylogenetic relationships of Dip sacales Dr on rbcL sequences. Ann. Missouri Bat. Gard. 79 5. DowNiE, S. F PALMER. 1992a. Restriction site mapping of the chloroplast DNA inverted repeat: A molecular phylogeny of the Asteridae. Ann. Mis- souri Bot. Gard. 79 3: 1992b. The use of chloroplast DNA Péstréngemens in reconstructing plant phy- logeny. 35 in D. Soltis, P. Soltis & J. Doyle (editors), Molecular Svateitalics of Plants. Chapman and Hall, Doyle, J. J. L. J. Do OYLE. 1987. A rapid isolation pro s for ET A rer of fresh leaf tissue. Phytochem. Bull. 5. ENGLER, А. 1928 ddr тв и In А. En а Prantl (editors), Die Natürlichen ies a 2 750 Annals of the Missouri Botanical Garden Aufl. 18a Bd. Ed. II: 74-226. Wilhelm Engelmann, Leipzig. EnBAR, C. 1991. Sympetaly — A oo character? t. Jarhb. Syst. 112: 417-451. & P. Leins. 1989. On the early floral devel. opment and mechanisms of secondary pollen presen- tation in Campanula, Jasione, and Lobelia. Bot. Jarhb. Syst. 111: 29-55. Еуре, R. H. 1966. Systematic anatomy of the flower and fruit of Corokia. Amer. J. Bot. 53: 833-847. FELSENSTEIN, J. 1985. Confidence limits on phylogenies: Ап Hos using the bootstrap. Evolution 39: 783- TA Z. S. & О. R. GOTTLIEB. 1982. Polyacety- lenes as systematic markers in dicotyledons. Bio- 60. sequence from a Miocene ture 344: 656-568. 1992. Ribosomal RNA asa phylogenetic tool in plant pend Pp. 50- yle (editors), Mo- 9] in D. Soltis, P. Soltis & J. Do lecular Systematics of Plants. а and Hall, ew Yor Hannis, E. M S. tive е апі 1 ага ade Development in the Compositae. Ph.D. Dis- ertation. Louisiana i University, nae Rouge. HECNAUER, R. 1964. emotaxonomie der Pflanzen. irkhauser, Basel, Switzerland. — 1971. e chemistry of the Compositae. Pp. 283-335 in V. H. Heywood, J. B. Harborne & B. L. Turner (editors), The Biology and Chemistry of do the Compositae. Academic Press, London. HiLtis, D. M. & M. T. Dixon. 1989. Meis duni phy- logeny: Evidence from 28s ribosomal ird . 355 olm, K. ie H. yoke 9 The Нхау of Life. Elsevier. Hum. [3 s. . D. Manon, P. A. ANDERSON, M. J. NDREWS & P m closely related C, and C, plant species. J. Biol. cuin 265: 808-814 HUTCHINSON, J. 1969. Evolution and ae of Flowering Plants. Clarendon 87b. A chloropla st DNA in version marks an ancient evoluti y split in ihe sunflower family Asteraceae. Proc. - Natl Acad. Sci. U.S.A. 84: 5818-5822. & . 1988. Phylogenetic implications of chloroplast DNA restriction site variation in the Mutisieae (Asteraceae). Amer. J. Вог. 75: 753-766. , H. J. MICHAELS & J. D. PALMER. 199la. Phy- logeny and character evolution in the Asteraceae based on chloroplast DNA restriction site mapping. Syst. Bot. 16: 98-115. — ———, К. E. HorsiNGER, H. J. Micuagis & J. D. PALMER. 1990. Phylogenetic сае of chloro- im ast e dudes: si 8 ши at higher tax- c levels: 2s deri Evo- ins 44: 2089-2105. , Н. J. MicHaELs, R. S. WALLACE, К. J. Kim, S C. KEELEY, L. E. Watson & J. D. PALMER. 1991b. Chloroplast DNA variation in the Asteraceae: Phy- logenetic and evolutionary uus Pp. 252- 279 in D. Soltis, P. Soltis & J. Doyle (editors), Molecular Systematics of Plants. Chapman and Hall, New Y ew York. JEFFREY, C. 1977. Corolla form in se cus RE гы evolutionary and taxonomic speculations. Pp. 111- V. H. Heywood, J. B. Harborne & B. L. Turner (editors), m Biology and Chemistry of the Compositae. Academic Press, wies on. Karis, P. O., M. Find K. BRE hy- logenetic analysis of the Cichorioideae (Asteraceae), with emphasis on the Mutisieae. Ann. Missouri Bot. Gard. 79: 416-427. KEELEY, S. C. & R. K. JaNsEN. 1991. Evidence from chloroplast DNA for the recognition of a new the Tarchonanatheae, and the tribal ета of Pluchea еа Syst. Bot. 16: dii KIM, HAE R. K. JANSEN, R. S. WALLACE SH. TM CHAELS & J. D. PALMER. “1992. Phylogenetic implications of rbcL sequence variation in the Asteraceae. Ann Missouri. Bot. Gard. 79: 428-445. LAMMERS, T. С. Circumscription and phylogeny of the Campanulales. Ann. Missouri Bot. Gard. 79: 388-413. Leins, P. & C. ErBAR. 1990. On the mechanisms of secondary pollen presentation in the oe Asterales complex. Bot. Acta. 103: 87-92. Mappison, D. R. The discovery and importance of multiple islands of most-parsimonious trees. Syst. s, H. J. & J. D. PaLMER. 1990. зрели пет relationships of the i dg Evidence from r sequences. Amer. J. Bot. 77: Morcan, D. R. & . Phylogenetic relationships among members of the Saxifragaceae sensu lato based on rbcL sequence data. Ann. Mis- souri Bot. Gard. 80: 631-660. OrMsrEAD, R. G., B. BREMER, K. TT & J. PALMER. 1993. А parsimony sala of the ie sequences. Ann. af their major lineages inferred from PALMER, J. D. 1986. Isolation and structural analysis of ropas DNA. Pp. 168-186 in Plant Molecular Biology: Methods in Enzymology. Academic Press, ork. ‚ К. К. Jansen, H. J. MicHaELs, М. W. CHASE & J. R. M. 1988. Chloroplast DNA vari- ation E P phylogeny. Ann. Missouri Bot. Gard. 75: 206. Parson W. E 1967. pu B fil. — Anom- aly or link. New Zealand J. B : 134-165. PORA. c J. & К. S. AMUTI. jr Fructo ructose oli- gosaccharides: Possible markers of phylogenetic re- lationships among и plant families. Bio- chem. Syst. Ecol. 9: Volume 80, Number 3 Michaels et al. 751 Interfamilial Relationships of Asteraceae RirLAND, K. & M. Стесс. 1987. Evolutionary analysis of plant DNA sequences. Amer. Naturalist 130: S74- S100 SANGER, F., S. NICKLEN & A. R. Courson. 1977. DNA sequencing with chain- Ere inhibitors. Proc. Natl. Acad. Sci. U.S.A. 74: 5463-5467. SHINOZAKI, K., N о: & М. s 1988. Nicotiana chloroplast genes for compon of the CL a apparatus. Bude donc T A, e m B. L. TURNER, V. С. PATEL & А. S. TOMB. 1977. Pollen morphology in the Compositae and in morphologically related families. Pp. 141-265 in V. Heywood, J. B. Harborne & B. L. Turner (editors), The Biology and Chemistry of the Compositae. Ac- ademic Press, London. Sorris, D. E., P. S. Зола, M. T. Стесс & M. Dur 1990. rbcL sequence са зл pd aaa relationships in Saxifragaceae POR lato. Proc. Natl. Acad. Sci. U.S.A. 87: 4640-46 SworForD, D. L. 1991. PAUP: Phylogenetic Analysis Using Parsimony. Illinois Natural History Survey, mpaign. TAKHTAJAN, A. . Outline of the classification of flowering plants (Magnoliophyta). Bot. Rev. 46: 225- 359. 1987. Systema Magnoliophytorum. Nauka, ningrad. THORNE, R. F. 1992. Classification a S Eeography of the flowering plants. Bot. Rev. 58: 77. Fossil history and ты 8 Рр. 21-39 in V. Н. Heywood, Ј. B. Harborne & В. L. Turner йш, The Biology and Chemistry of the ompositae. Academic Press, London. WAGENITZ, С. . New aspects of the systematics of МЕРТЕ Pl. Syst. Evol., Suppl. 1: 375-395. = 2. The Asteridae: Evolution of a Е Es and its та. status. Ann. Missouri Bot. Gar > d. зне І. E., R. К. JANSEN & J. Estes. 1991. = р!асетепї 2 Marshallia (Asteraceae) based on ev- ig from chloroplast DNA restriction site muta- tions. Amer. J. Bot. 78: 1028- 5. онак, с. 4 T. Стесс. 1987. Evolution of igher-plant chloroplast DNA-encoded genes: Impli- cations pa structure- function and сии stud- ual Rev. Pl. Physiol. 38: 391- & P. R. p de 19 of ribulose 1,5-bisphosphate carboxylase from spin- ach chloroplast DNA. Nucleic Acids Research 9: 2251-3270 CHARACTER-STATE WEIGHTING FOR CLADISTIC ANALYSIS OF PROTEIN-CODING DNA SEQUENCES: Victor A. Albert? Mark W. Chase,?* and Brent D. Mishler? ABSTRACT Nucleotide data are a restricted character system complex enough to confound phylogenetic analyses yet simple enough to permit establishment of probability models for sequence change and corresponding characte transversion bias, and (iii) differential proportions is model is shown to be properties of equal versus differential character nd, and third codon positions. generally consistent for all phylogenetically useful data. Greater understanding of the -state weighting com es from consideration of numbe problem that is not easily resolved either by equal weighting or by our s weighting model, which acts globally rather than adjusting for different probabilities of character- state chan mbers of potential tree segments. Prospects for phylogenetic recon- struction from protein-coding nucleotide data are discussed with reference to the robustness of equal weighting (given our own model) with adequate taxonomic sampling Information derived from nucleotide sequences is becoming increasingly popular for phylogenetic studies, partially because of technological advances (such as automatic temperature cyclers and ther- mostable polymerases) that permit rapid and ex- tensive data gathering. The deceptively simple properties of DNA data and their promise of more reliable phylogenetic hypotheses have changed the research emphasis of many systematists. DNA vari- ation can indeed provide valuable characters for is not a panacea for phylogenetics (contra Gould, 1985). Reflecting perhaps a lack of understanding of first principles, too many cladograms derived from both molecular and morphological data are appearing in the literature without justified char- acter analysis or choice of optimality criterion. Although our current understanding of molecular biology may be superior to that of the ontogeny and heritability of most morphological characters, "simple" substitution at particular sites (e.g., Brow 1982) and paralogy in multigene families (Fitch, 1970) can confound phylogenetic wien molecular phenomena such as multiple n et al., The apparent simplicity of DNA data in com- parison with ot The pattern of DNA variation is simpler compared to most morphological data in that there are four er data forms can be misleading. basic character states (the nucleotides A, C, G, and T) with no intrinsic ordering into a transformation series. However, evolutionary independence of nu- ! We = Steve Farris, Joe Felsenstein, David Hillis, Chris Humphries, Elizabeth Kellogg, and Dave Swofford or comme (BSR- 8914035 to of Science and Techn Alber) i is gratefully acknow V. Albert, BSR-89064 and discussion. All interpretations are, of course, our own. Support from the National Science Foundation 496 to M. Chase, BSR-9107484 to B. Mishler), the North Carolina Board ology Mri LE award 89SE14 to B. Mishler), and the American Orchid Society (to V. epartment be Biology, [rice of North Carolina, Chapel Hill, North Carolina 27599-3280, U.S ' Correspon 6 Uppsala, ^e Dm current address: Department of Systematic Botany, Uppsala University, тена 6, 5-752 ED t res 85: mu Botanic Gardens, Kew, Richmond, Surrey, TW9 3AB, United Kingdom. D > Department of Bota uke University, Durham, North Carolina 27706, U.S.A ANN. Missouni Bor. са. 80: 752-766. 1993. Volume 80, Number 3 1993 Albert et al. Character-state Weighting 753 cleotide positions (a requisite if they are to be used as characters in cladistic analysis) cannot be au- tomatically assumed. Protein-coding sequences are organized into triplet nucleotide codons that may be positionally intra- and interdependent because of constraints imposed by the genetic code and the function of the encoded protein. Such constraints may produce unequal and correlated probabilities of substitution among nucleotide positions (see Fitch & Markowitz, 1970; Fitch, 1986). Positional ho- mology of nucleotides may be relatively easy to determine, but hypotheses of character-state ho- mologies can be confused by multiple, undetected substitutions, which are themselves drawn from the same four possible states. Despite these underlying complexities, it has been possible to derive simple probability models that mimic properties of se- quence evolution (e.g., Albert & Mishler, 1992; Albert et al., 1992; DeBry & Slade, 1985; Fel. senstein, 1981a, 1992; Hendy, 1989; Hendy & Penny, 1989; Lanyon, 1988; Templeton, 1983). Some of these models have, in turn, been applied to maximum-parsimony inference of cladistic re- lationships (e.g., Albert et al., 1992; Chase & Palmer, 1992; Donoghue et al., 1992; Gastony et al., 1992; Kadereit & Sytsma, 1992; Olmstead & Palmer, 1992; Spooner & Sytsma, 1992; Wendel & Albert, 1992). Characters and character states for cladistic analysis of DNA sequences are derived from either indirect or direct sequence sampling. The most widely used indirect approach is mapping of re- striction endonuclease cleavage sites. When a par- ticular restriction site is found to be present in some terminal taxa and absent in others, an un- derlying sequence difference is implied de facto. Wagner parsimony (Farris, 1970; Kluge & Farris, 1969) places no restrictions on underlying patterns of nucleotide substitution. In contrast, the Dollo criterion (Farris, 1977; see also Farris, 1983: 26) will not recognize substitutions that could lead to parallel gain of a recognition sequence, although parallel losses are tolerated. In a previous paper, we developed a character-state weighting scheme that treats both forms of homoplasy relative to differential probabilities of gain versus loss esti- mated for individual restriction sites (Albert et al., 1992). The reasoning developed by Albert et al. (1992) was extended to direct sequence sampling by Albert & Mishler (1992); a character-state weighting scheme was erected to recognize the asymmetrical probabilities of transversion versus transition sub- stitutions within nucleotide characters. Here, the model of Albert & Mishler (1992) is modified to incorporate asymmetrical probabilities of substi- tution among codon positions in protein-coding se- quences. The model rests on the simplifying as- sumption of equal probability of change among first, second, and third codon positions (respec- tively). The model also provides a correction for multiple substitutions at particular positions and accommodates wide ranges of branch lengths, tran- sition/transversion ratios, and codon position bi- ases. After explaining the assumptions underlying our approach, we provide a heuristic examination of its properties by comparing it with the commonly used, equal-weighting parsimony criterion of Fitch (1971, 1977: a nonadditive, multistate generaliza- tion of the Wagner criterion; Farris, 1970; Kluge & Farris, 1969; cf. Swofford & Olsen, 1990). We evaluate consistency criteria in terms of permissible branch-length asymmetries and discuss expecta- tions based on the number of taxa analyzed; branch length is incorporated into the consistency index (Kluge & Farris, 1969) to explain its dependence upon taxon number. We conclude with prospects for phylogenetic analysis of protein-coding se- quence data, including the recognition that equal weighting is a robust assumption given a relatively even sampling of the true tree. THE MODEL Our previous character-state weighting models for restriction site (Albert et al., 1992) and nucle- otide sequence data (Albert & Mishler, 1992) were designed for utility of execution. We have opted for simplified representations of the major known biasing factors in molecular evolution to provide generalized, robust weighting schemes employable with large numbers of taxa. Within-character weighting is easily accomplished by representing transformational costs between states in matrix form (Fitch & Ye, 1991; Sankoff & Cedergren, 1983; Swofford & Olsen, 1990; Williams & Fitch, 1989, 1990). As of this writing, stepmatrix weighting is available in PAUP 3.0 (Swofford, 1991), MacClade .0 (Maddison & Maddison, 1993), and a pro- gram available from Walter M. Fitch (Williams & Fitch, 1990). Application of stepmatrices is com- putationally expensive relative to Fitch parsimony because it requires dynamic programming (Sankoff & Cedergren, 1983; Swofford & Olsen, 1990). With PAUP 3.0, a single stepmatrix may be as- signed to an entire data set, or individual step- matrices may represent each character separately. If one understood how each nucleotide position bears on protein function, one could possibly assign individual probability models and individual weight- 754 Annals of the Missouri Botanical Garden ing stepmatrices over an entire sequence. This is not currently practical because it would require comparison across many taxa that in turn would have to be associated with a robust phylogenetic hypothesis. Furthermore, as will be shown below, minor heterogeneities in transformational proba- bilities have negligible effects on parsimony recon- structions. BRANCH LENGTH AND THE UNITS OF CHARACTER-STATE CHANGE Branch length, A, refers to the total number of changes expected along a tree segment Vi ан either two nodes or a node to a terminal. In so models, including ours (Albert & Mishler, ш. Albert et al., 1992; DeBry & Slade 1985), Х is the symbol used to express branch length in per- character terms, i.e., A = È A, over / characters (where Zi = |. This usage should not be confused with А = rate of character-state change with respect to time (a usage employed, for example, by Hendy, 1989; Hendy & Penny, 1989; Nei, 1987). That rate was referred to as S, by Albert et al. (1992), i.e., the effective substitution rate for a character (a nucleotide string of any defined length); effective substitutions are those that have been fixed in a lineage. The expected number of changes per char- acter during a segment, Х, is thus equal to the product of the rate of change, 5,, and the time through which the branch existed, 7. The units of time cancel, yielding a parameter that is expressed in terms of state changes per character per seg- ment. Therefore, different combinations of rates and times can yield the same А value. Observed mean branch length over all charac- ters and all segments (À) can be used to estimate \ under the simplifying assumption that а! seg- ments are of equal length (see discussion below on effects of unequal segment lengths). Because the number of segments in an unrooted tree is [2n — 3 — р], where n = the number of taxa, and p = the number of segments in excess of 3 per node (i.e., polytomies), À = A/([2n — 3 — pl) (1) where Д is the number of character-state changes observed over an entire tree and / is the number of characters represented. An average Х may also be estimated for the ith character over all seg- ments: = A/[2n — 3 — p] (2) Likewise, branch lengths may represent averages for a specific set, А, of all characters /: A, = A,,/({2n е pl) (3) Expression (1) was the strategy used by Albert et al. (1992) and recommended by Albert & Mishler (1992) for estimating À values from DNA sequence data optimized onto robust, a priori cladistic hy- potheses based on other character systems (see below). For the present purpose, however, individ- ual Х|, values for each codon position are required. The simplest way to produce such values is to partition À estimates with three multipliers (o, а, a,) representing the fraction of substitutions per codon position observed on an a priori topology. PROBABILITIES OF CHARACTER-STATE TRANSFORMATIONS The premise behind the nucleotide character- state weighting model described by Albert and Mishler (1992) is that the probability of observing a transition is different (greater, in most cases) from that of observing a transversion. This asymmetry can be estimated by phylogenetic comparisons of sequences; nucleotide state changes are optimized onto an a priori cladogram, which provides a **phy- logenetic correction" for multiple substitutions (Al- bert and Mishler, 1992; Albert et al., 1992). The phylogenetic correction does not reflect all sub- stitutions that effectively fixed in lineages and are observable given the taxa sampled (as will be discussed in detail below, if too few taxa are sampled, many multiple substitutions will go undetected). The Kimura two- parameter model (K2P; Kimura, 1980), which was designed to take into account transition and trans- version information in order to estimate net sub- ave occurred, only those that are stitutions in pairwise comparisons, provides a sim- ilar, albeit slightly higher, estimation of substitutions than does the phylogenetic estimate mentioned above (Albert et al., 1992). The K2P model has been reoriented for estimating transformational fford & Olsen (1990; in terms of transition and transversion rates) and by J. Felsenstein (cited in Albert Mishler, 1992). Felsenstein's formulations аге more useful for the present purpose because there are probability on phylogenetic trees by Swoff only two parameters—A and transition/transver- sion ratio (R): l Q4) + R | (“ХІ + В) | — Y zi À (4) where xi indicates that the probability is for a particular transition, an Volume 80, Number 3 1993 Albert et al. пе Weighting І XUAN + 5 ©) where xv indicates that the probability is for a PAT) 2 ж — Ya Е particular transversion. Determining differential transformational prob- abilities for each of the three codon positions re- quires partitioning À with the а, a,, and а, mul- tipliers. As a result, three separate calculations for both expressions (4) and (5) are required. The partitioned Aa values used here assume equal prob- abilities of character-state change within all first, second, and third codon positions, respectively. These probabilities only consider change as a func- tion of entire trees (through the use of A); no adjustment is made for individual segments. These assumptions, although defensible as a first approx- imation, are simplifications relative to the covarion and covariotide hypotheses (Fitch & Markowitz, 1970; Fitch, 1986; cf. Hendy & Penny, 1989; Fitch & Ye, 1991) and branch-length asymmetries that may be present on true trees. CHARACTER-STATE WEIGHTS Transformational probabilities and their. corre- sponding character-state weights share a simple mathematical relationship. In fact, the following historical discussion highlights a conceptual link between maximum parsimony and maximum like- lihood approaches to phylogenetic inference. In his Bayesian approaches to cladistic analysis, Farris (1973, 1977, 1978, cf. 1983, 1986) examined the posterior probability of a phylogenetic hypoth- esis— P(hypothesis|data)—through из ргорог- tionality to the joint probability — P(Aypothesis, data) i i | Bayes's (1763) theorem of conditional probability: P(hypothesis| data) Pídata| hypothesis)P(hypothesis) |/ P(data) (6) From expression (6), the following simplification can be made P(hypothesis| data) = [P(Ahypothesis,data)]/ P(data) (7) For a particular parsimony analysis, the prior prob- ability of the data, P(data), is constant and the prior probability of a particular maximum parsi- mony tree, P(hypothesis), may be considered equal over all such trees. Farris's derivation of P(hypothesis,data) was expressed as repeated products of state transformation probabilities and their assignments to tree segments over all char- acters. Thus, —In[P(Aypothesis,data)] took the form of summation of these statements; by the nature of logarithms, minimizing the latter ex- pression (and so tree length) is equivalent to max- imizing the former (i.e., finding the tree length of greatest probability). Log-transformed probability factors not solely dependent upon the data matrix were considered to represent weights for character- state changes. Summation of these weights over all characters and their segment assignments (1.e., to produce a weighted tree-length) was interpreted as identical to the weighted parsimony criterion of Kluge & Farris (1969). Thus, taking the natural logarithm of P(hypothesis,data) permits calcu- lation of both character-state weights and the min- imum number of weighted steps required to pro- duce the most probable (i.e., most parsimonious) tree. A similar argument for logarithmic conversion of probabilities into weights comes from maximum likelihood inference (Felsenstein, 1981b; cf. cri- tique by Farris, 1983). From Bayes's theorem (Bayes, 1763; expression 6), likelihood — P(data | hypothesis) probability P(hypothesis,data), but through P(hypothesis) rather than P(data): P(data| hypothesis) = [P(Aypothesis,data) |/ P(hypothesis) (8) > Joint Likelihood depends entirely upon the model con- structed for P(hypothesis), as evident from differ- ent interpretations of the relationship between like- lihood and parsimony obtained under different models (Farris, 1973; Felsenstein, 1973; cf. Sober, 1988). Felsenstein's likelihood justification of par- simony (1981b; see also Sober, 1988) involved a simple probability model in which binary data were assumed to contribute the bulk of the likelihood via a single (versus homoplastic) reconstruction of character states. In finding the maximum of the likelihood (nested repeated products), it is custom- ary to find the natural logarithm of the expression (to produce nested summations and simpler cal- cf. Swofford & Olsen, 1990). Under Felsenstein's assumption of low rates of culation of minima; c change (see discussion below), the log-likelihood term for a change of character i in segment j 15 simply ln(A;), based on Felsenstein's notation (1981b, his equations 4 and 5). The logarithm of Ху will be negative when Х, < 1 and becomes more so as branch length decreases. Felsenstein (1981b) interpreted | —1n(A;)| values to represent the weights for such changes. Large weights reflect small probabilities of character-state change and vice versa (precedents for this relationship exist, 756 Annals of the Missouri Botanical Garden Relative weight 0.008 0.01 — 0.004 0.006 A FIGURE 1. Relative weight of nucleotide character- state transformation (transversion weight divided by tran- sition weight; Y axis) as a function of А (X axis) and differential transition /transversion ratios а ти ђу the individual curves). Х values between 0 ап re d in a; the smaller ranges from 0 to 0.1 and 0.002 ~ logarithms of probabilities obtained from expressions (5 and (4). на нао ratios evaluated (shown in lobi pui order ess from a шл starting nucleotide; thus, ral result lies directly on the X axis, vods equal weight) through 10.0 in increments of 0.5 although suggested on other grounds or under dif- ferent models; Farris, 1966, 1969, 1977, 1978; Kluge & Farris, 1969). Because likelihood infer- ence chooses the topology that maximizes the like- lihood, character-state changes with relatively large weights will be preferentially avoided (Felsenstein, 81b). Maximizing the logarithm of the likelihood minimizes the sum of the weights of all character- state changes: hence the justification for weighted parsimony, which de facto minimizes the contri- bution of strongly weighted state transformations. or the present model, character-state weights for a transversion over a transition are the ratios of the negative natural logarithms of expression (5) probabilities to those of expression (4). The be- havior of these weights under a range of values for Х and R can be examined using two-dimensional plots (Albert & Mishler, 1992). In Figure 1, curves representing transition/ transversion ratios (R) be- tween 0.5 (parity) and 10 are shown as À increases continuously from О to 1 (the ranges О > 0.01, О 0.1, and О > 1 are shown separately). As would be expected, the weight approaches 1 as А becomes smaller. The effect of increasing R is immediately apparent: as transitions become pro- portionally more frequent, a higher relative weight is assessed to a transversion. Another property of the weighting model is that the relative weight range for transversions over transitions is extreme- ly shallow when A is small (i.e., < 0.1). Albert et al. (1992) considered Х in some detail from a heu- ristic mathematical perspective and concluded that ^ values € «0.1 should be the limit for data that could be expected to reflect phylogenetic history (a of 0.1 is equivalent to 1 out of 10 characters changing per tree segment, on average). Under such conditions, Ха partitions cannot yield sub- stantially different codon position weights using any reasonable value of R (Fig. 1 INITIAL WEIGHTS FROM A PRIORI INFORMATION Because the transversion/ transition weight range is so small over all reasonable Х ап values, one could simply use weights (partitioned with appro- priate Àa values) that span this range (i.e., between 1 and ca. 2.2; Fig. 1b) in a series of analyses to search for robust topologies (see Albert et al., 1992). An alternative is to use a priori approximations of A, transition /transversion ratio, and proportions of substitutions among codon positions to provide ini- tial weighting criteria for a single, optimal phylo- genetic reconstruction. Such estimates of Aa and R values are best derived by optimizing sequence data onto topologies accepted a priori based on robust cladistic analyses of other character sys- tems; attempting to derive estimates solely from (see Albert & Mishler, 1992; Albert et al., If such estimates are to be useful, the a priori topology must be encompassing relative to the data to be analyzed, circumscribing variation within it Volume 80, Number 3 Albert et al. Character-state Weighting TOY Marchantia 1398 total changes = 444 segments - 8 total positions = transitions = transversions = 177 Ist position changes = 75 2nd position changes - 32 3rd position changes = 337 FIGUR "Land plant" tree, modified from Albert et al. ин Accepted a priori based on robust cladistic analyses ‘of ‘morphological, anatomical, and biochemical information. Sequence data for rbcL (1,398 putatively | je at each of the 8 4 for the entire tree. Thus, À = 444 /8/ 1.398 E T nu changes, R — 444 (— 0.7590). These values were used t using selected representatives and outgroups. To be trustworthy, the A inferred should be less than 0.1, and R should be greater than 1:1. The latter conditions return us to consideration of the number of tree segments. Ап a priori to- pology with few segments is expected to give con- servative (i.e., relatively even) weight estimates For example, with increasing phylogenetic depth (e.g., angiosperms analyzed together with Маг- chantia or also with Chlamydomonas; Albert et al., 1992), (relative to transversions) becomes lower (probably due to multiple substitutions occurring at the same site despite the wir e bias favoring transitions; cf. Brown et al., the excess of observable transitions ; Gojobori et al., i 1984), and a a second codon positions accrue larger proportions of total substitutions et al., 0.0397). F 267/177 (= 1. 5085), ai 75/444 (= 0 > 5'-primer nucleotides o the 1,428 postions used by Albert et al., 1992 & Maddison, 1987). The numbers ate changes, totali ng — 32/44 ha 0.0 720), da = 337/ 5 таје d state ime for heuristic comparison with equal weights in parsimony analysis of nucleotide du. (see Table 1). (many of which are probably amino acid replace- ment substitutions in contrast to third position "si- lent" substitutions; Albert et al., 1992; see also Brown et al., ! Despite the restricted number of segments avail- able for estimating effective substitutions and their asymmetries, the “land plant” tree of Albert et al. (1992; Fig. 2) is a robust, a priori topology. Albert et al. (1992) optimized rbcL sequence data onto this tree to derive values for use with their restric- tion-site weighting model. The plastid rbcL gene (which encodes the large subunit of ribulose-1,5- bisphosphate carboxylase/oxygenase) has been sampled extensively for evidence of seed-plant re- 1993, and other pa- pers in this issue). Transformational probabilities lationships (see Chase et al., and weights that may be relevant to such phylo- 758 Annals of the Missouri Botanical Garden TABLE 1. Character-state weights for rbcL sequences based on the “‘land plant" tree of Albert et al. (1992). State transformation probabilities, P, and E character-state weights, —In(P), for transition and t version substitutions in each of the three codon оа calculated using a priori values from Figure 2 and ех- pressions (4) and (5). Initial weights for use with PAUP 3.0 stepmatrices could be the —In(P) rounded to three significant digits and multiplied by 10* Codon Substitution posi- type tion P —In(P) Transition 1 0.00400751 5.51959 2 0.00171601 6.36775 3 0.0176189 4.03878 Transversion 1 0.0013331 6.62025 2 0.000569663 7.47047 3 0.00593453 5.12697 genetic studies were calculated from expressions (4) and (5) using values obtained from the “land plant" tree (Fig. 2; Table 1). These weights may be designated in three stepmatrices, one assigned to each codon position in rbcL. The assumption block format for PAUP 3.0 (Swofford, 1991) is shown for first, spectively: second, and third positions, re- usertype first — 4 a с g t [а] — 6620 5520 6620 [c] 6620 Sei 6620 5520 [g] 5520 6620 — 6620 [t] 6620 5520 6620 ops вина = 4 [a] — 7470 6368 7470 [c] 7470 == 7470 6368 lg] 6368 7470 = 7470 [t] 7470 6368 7470 = usertype third = 4 а с g t [а] = 5127 4039 5127 [c] 5127 == 5127 4039 [g] 4039 5127 = 5127 [t] 5127 4039 5127 — In such stepmatrices, the direction of transfor- mation is indicated by [nucleotide] > nucleotide. Differential weighting may begin at the taxon-ad- dition stage or may use starting trees found from prior, equally-weighted Fitch parsimony analyses (both following the recommendations made by Maddison (1991) for discovering all most-parsi- monious trees). The latter strategy is significantly less time-consuming: the weights shown above are so close to a 1:1:1:1 assumption that Fitch to- pologies should include (or lie near) optimal weight- ed trees. Although this weighting scheme will select specific trees based upon the ad hoc assumptions of the model and the specific values given its pa- rameters, it is useful in identifying neighborhoods of trees (defined by proximity in terms of branch rearrangements) that could provide better hypoth- eses of the true tree given the empirically observed character and character-state asymmetries. As such, character-state weights derived as above (i.e., Fig. 2; Table 1) should only be considered initial esti- mates to be supplemented by additional analyses using bracketing values, including equal weighting of character-state transformations (i.e., Fitch par- ). simony; cf. Albert et al., 1992 EXPECTATIONS OF PERFORMANCE: CONSISTENCY UNDER BRANCH-LENGTH ASYMMETRIES AND SENSITIVITY TO TAXON NUMBER Felsenstein (1978) introduced considerations of statistical consistency for parsimony under the Camin-Sokal (Camin & Sokal, 1965) and Wagner (Farris, 1970; Kluge & Farris, 1969) optimality criteria. His model (a special case of the conditions established by Steel, 1989; Penny et al., 1991) included an unrooted tree of four taxa where two nonadjacent branches of equal length had accu- mulated much higher numbers of changes than the three others, which were themselves of equal length. А spurious reconstruction in which the taxa with the long branches were grouped because of shared homoplasies was shown to result from unequal rates of character-state change. This phenomenon is bet- ter recast in terms of А because branch lengths (and hence the potential for spurious attraction) are a function of both rate and divergence time (Albert et al., 1992; Hendy & Penny, 1989). Although we do not regard statistical consistency as necessary to justify the use of parsimony (see also Farris, 1983; Sober in Felsenstein & Sober, 1986; Sober, 1988), our previous A-based heuristic examination of consistency for binary data shows that parsimony analyses are generally consistent under the model specified by Felsenstein, 1978) for data that could be expected to show evidence of common ancestry (Albert et al., 1992). Hendy & Penny (1989) have suggested that breaking up long branches through the addition of appropriate — Volume 80, Number 3 Albert et al. 759 1993 ae Weighting = ји k = 2 (binary data) = 2 S 2 2 ^ = 0.81 prohibited Pd © ^ = У. E Ра a 0-6] " „ S ^ k=3 = Ра as ба " © y ka > at = = Y k=5 2 0.21 > 5 ^ T S k=10 A a = ~ _ [SU | А Е2 d 0.2 0.4 0.6 0.8 1 p (probability of change in long branch) FIGURE 3. The relationship between Felsenstein's consistency model and the number of available character states. Given probabilities p? (for parallel changes on long, nonadjacent branches) and q (for a change in the same character the short, connecting branch), the approximation qa DAE v 1)]p? must hold for parsimony to be consistent under Felsenstein's model. The X and Y axes represent p and q iesu respectively. By definition, q 7 p (above the dashed line) is ке. Each curve appearing successively b the а = p line represents [1/(k — 1)]р fork = 2, 3, 4, 5, , 8, 9, and 10. The first curve os binary m а а е large region of inconsistency (the area E labeled FZ for Felsenstein zone). As the number of char r states increases to 4 (as in nucleotide data; the third curve) and then 10 (the last curve), the region of они. decreases dramatically. This behavior is attributable to B lower а А of a parallel change when more states are bove. Я With binary d p = 0.6 w ata, sequence data. taxa may permit parsimony to avoid systematic erties of the Fitch criterion with those of parsimony error (cf. Swofford & Olsen, 1990). Nevertheless, ^ analysis using the present weighting scheme. Steel (1989) and Penny et al. (1991) underscored that the range of cases where parsimony may be ENTRE . . A А FELSENSTEIN S CONSISTENCY MODEL inconsistent increases as the number of taxa in- creases. Although this may be true, the practical If we imagine two equally long branches con- consideration is not so much the absolute number nected by a shorter branch to be the “true” to- of terminal taxa as an increase in those detracting — pological relationship, avoidance of inconsistency from the evenness of sampling of the true tree (the using parsimony (after the model of Felsenstein, shape of which is unknown). Perhaps with real data, 1978) requires that the probability of parallel char- the best strategy for limiting the occurrence of acter-state changes in each of the long branches long, spuriously connected branches is always to (р?) be lower than the probability of a single change examine many terminal taxa (within computational in the same character occurring in the connecting limits). Nevertheless, sampling considerations ap- branch (q). This is expressed by the general ap- ply, such as the range and eccentricity of nucle- proximation q > |1/(k — 1)]p?, where k is the otide variation being compared (V. A. Albert & K. number of character states (J. Felsenstein, pers. Bremer, unpublished). Weighting is not expected сотт.). Thus, for binary data, the simple inequal- to affect the frequency of spurious branch attrac- ity q > р? must hold (see Albert et al., 1992). For tion greatly because false character-state identity, DNA data (k = 4), the approximation is q > (VS)p* when it occurs, is an intrinsic and unquantifiable (note that as k increases, р? becomes smaller; Fig. property of the data matrix. Nevertheless, there is 3). Thus, q = [0)р? ] must be positive for parsi- heuristic value in comparing the consistency prop- mony to be consistent under Felsenstein's model. 760 Annals of the Missouri Botanical Garden Relevant assumptions include (i) a large (е оо) random sampling of independent characters whose probabilities of character-state change are equal (see the pertinent objections of Farris, 1983) and (ii) the simplification that both long branches have equal probabilities of change. Assumptions (i) and (п) are unrealistic in practice, because (i) actual data sets may consist of relatively few, nonran- domly selected characters that may show both in- terdependence and (ii) asymmetrical probabilities of change. Thus, Felsenstein's model can only ap- proximate consistency behavior for real data. If we assume a simple Poisson model of nucle- otide substitution (as is appropriate for low prob- abilities of change; cf. Farris, 1973; Felsenstein, 73; Goldman, 1990; Lanyon, 1988), the prob- ability of a single character-state change in segment interval T, P(T), may be estimated by Ae”? (note, however, that for A < 1, Ae^^ ~ A). This for- mulation ignores the possibility of multiple substi- tutions, which is an appropriate approximation be- cause two or more changes are improbable at low A. For the Fitch model, changes to any different nucleotide are equiprobable. Thus, substituting X for the short-branch А and Y for the long-branch A (as in Albert et al., consistency criterion becomes: Xe^* — (14)Y?e-" > 0 (9) 1992), the approximate An analysis of this approximation for all X, Y < 0.1 demonstrates that a smaller region of incon- *; Fig. 4a) exists than that found for binary characters by Albert et al. (1992). This is explained by the smaller probability of parallel change when more character states are available (see Fig. 3). sistency (**Felsenstein zone’ e same scenario with four nucleotides can be extended to permit correction for the possibility of multiple substitutions. The expectation is that the Felsenstein zone would shrink because recognition of intermediate substitutions would limit hypotheses of parallel (homoplastic) character-state identities on long branches. Evaluation of this prediction can be accomplished by use of the Jukes-Cantor (J-C) random nucleotide substitution model (Jukes & Cantor, 1969). Using net probabilities of substi- tution (i.e., three potential derived states) during segment in- the probability of change to any of terval 7, the approximate consistency criterion be- comes: GU. =") = DUI — ue (10) Expression (10) further reduces the size of the Felsenstein zone beyond that observed for expres- sion (9) or for the binary model of Albert et al. (1992; see Fig. 4b). The K2P model that we use for the present weighting scheme (excluding Ха partitions) also has an approximate consistency criterion (J. Felsen- stein, pers. comm. ): ai X] + 2[P (T); й - [P.AT); YT 2[P,(T); YP > The probabilities indicated are from expressions (4) and (5), and the 1/(k — 1) correction for number of states is replaced by multipliers to indicate the two possible transversions from a particular starting nucleotide. Because parallelism on p branches could occur through either transitions or transversions, the consistency criterion must include both possi- bilities. The expectation is that the Felsenstein zone should be larger at high values of R because this bias would make parallel changes more frequent. When R is close to 1 (see expressions [4 ] and [5 ]), the K2P model yields a small realm of inconsistency (at R = 2 it is still smaller than for the J-C model; Fig. 4c). At higher but realistic values of R (e.g., 8) the Felsenstein zone is indeed larger, yet still small enough to permit a wide range of branch- length asymmetries (Fig. 4d) The net result of this heuristic examination is that consistency itself is not a sufficient justification for favoring one weighting model (the present) over another (Fitch parsimony); the Felsenstein zone is insignificant in all of the scenarios studied. Aside from what this might imply about the necessity for considering consistency (i.e., that it may be irrel- evant for justification of parsimony methods), it also suggests that some other factor(s) might have predictive value when comparing the expected per- formance of weighting models. NUMBERS OF TERMINAL TAXA Breaking up long branches by the addition of more taxa to a data set may be impossible because appropriate taxa may not exist (Hendy & Penny, 1989). Nevertheless, we argue that increasing tax- on number can decrease the likelihood of spurious branch attractions (despite the increased number of possibilities where inconsistency could occur; Steel, 1989; Penny et al., 1991) if X values remain relatively small throughout the data. Recall that the number of segments in an unrooted cladogram of n taxa is [2n — 3 — p]. Assuming that the number of excess branches at nodes (p) remains fairly small for a data set (i.e., not approaching n), the number of available segments increases linearly with increasing n. Intuitively, one would expect Volume 80, Number 3 1993 Albert et al. 761 Ста. o Weighting > 0 J^ X FIGURE 4. is À for a short branch connecting two lon here X, Y x0) is shaded (note that the X and Y axes E Contour plots representing expressions (9-11) for À values of X and Y between 0 and 1.0, where X branches with А values of Y. The are reversed relative to Fig. approximate region of inconsistency 3). Separate plots are shown d: R (w for the Fitch parsimony model (a), the Jukes-Cantor model (b), and the Kimura 2-parameter model (c: R — 2; — 8). The realm of inconsistency is shown to be miniscule under all models. Fitch parsimony to provide a more accurate phy- logenetic reconstruction if more segments were available for it to discover otherwise hidden “іп- termediate” substitutions. Saitou (1989) has formalized this expectation in his analysis of the underestimation of branch lengths by Fitch parsimony. In contrast to the single-state/ branch assumption of the Fitch (1971) criterion, Saitou allowed for multiple substitutions (under the -C model) to account for terminal states on sim- plified trees with no hierarchical structure (1.e., unresolved polytomies). He d that for all À values between 0.01 and 1.0, the expected number of required nucleotide changes per branch per site over a tree of n terminal taxa increases as n in- creases. Saitou's (1989: 5) reasoning for this ob- servation is much the same as for the intuitive example above: **. . . determination of the ancestral nucleotide X becomes more and more unambiguous as [n] is increased, hence more parallel changes are detectable." Saitou also observed that under- estimation of branch lengths (A, under his model) remains small with a relative difference of 6.4% at all reasonable Х values (i.e., < at À — 0.01, the difference is only 0.7%). However, at A = 0.2, the underestimation is by 12.2%, at А = 0.5, 27.0%, and at А = change per character for every tree segment), Fitch 1 (indicative of 1 expected parsimony not surprisingly "gives a gross under- estimate [by 44.8%] of the branch lengths" (Sai- tou, 1989: pendently support the range of А values (i.e., = 5). Thus, Saitou's observations inde- 0.1) for which the Felsenstein zone was shown be miniscule (Albert et al., ). Additionally, Saitou's calculations (1989, his table 5) suggest that appropriately small estimates of А (i.e., 0.01 762 Annals o Missouri Botanical Garden Three-dimensional representation of the FIGURE 5. relationship between the number of taxa, m (X axis), segment length for a given character, A, (Y axis), and the simplifying assumption of А, equality оми а tree (i.e., A), clearly shows the decrease of C as А, values and the number of taxa get larger (ranging Eu 3 to 100). Underlying this behavior is the effect of segment number, [2n — 3] for unrooted trees; recall that is, ode a rs [2n – 3]. Only solutions less than ual to the theoretical maximum of C (1.0) are shown; impossible n and Х, combinations were excluded from the urface. ^ 0.1) should be stable under the Fitch criterion with increasing numbers of terminal taxa. Because the Fitch criterion places no ordering on the four nucleotide character states, two-, three-, or even four-way ties may be hypothesized at nodes for some characters, even if the number of seg- ments is large. The number of informative nucle- otide positions usually differs when analyzing data with equal versus differential character-state weighting. For the Fitch criterion, only putative synapomorphies (e.g., 98 ticular site in a 100-taxon comparison) can be used Differential character-state weighting with stepmatrices impos- s and two Ts at a par- to justify most-parsimonious trees. es a partial transformation series among nucleo- tides. Because transitions and transversions are not given equal weight, state changes between nucle- otides are differentially connected: 0 v ALT — XI Бе G—C G — (7) (Fitch) (differential) where 6 indicates the unweighted path between any nucleotide pair, v, the weighted path for a trans- version (shown as a bolder line to indicate its higher relative cost), and t, the weighted path for a tran- sition. For example, a nucleotide position with 98 As, one T, and one C is considered informative for optimization under the present model. Consider a third case: 97 As, two Ts, and one C—the possible pathways between the various A and T states and the single C state are optimized along with all other available data, and the least costly transformations are hypothesized. Such hypotheses might include ао не for C, synapomorphy for the Ts, T T transformation path with C о or even a path including all nu- cleotide variation, such as A > C ТА ~ Because of the connectedness provided by weighted paths, what might be ambiguous states at nodes for the Fitch criterion should often be resolved by differential character-state weighting. Despite (i) the greater proportion of available information used by our differential weighting method and (ii) the arguments given above con- cerning resolution of multiple substitutions with increased terminal taxa, a methodological caution is in order. Because of the limitations of current software (and available hardware), both the deter- mination of maximum parsimony and the discovery of all most-parsimonious trees are difficult at best with large data sets (e.g., Chase et al., 1993) Therefore, our statements about taxonomic sam- pling should not be taken as ““more is better" for empirical studies; rather, they are intended as log- ical arguments that may or may not be applicable to real-world scenarios. MEASURING CHARACTER-STATE CHANGE: THE CONSISTENCY INDEX IN TERMS OF À AND NUMBERS OF TERMINAL TAXA The ensemble consistency index, C (Kluge & Farris, 1969; Farris, 1989a, b, 1991) is a useful measure of homoplasy with edad to a tree. Farris (1989b: 407) described the interpretation of C: “The consistency index assesses homoplasy as a fraction of the character change on a tree, and so the frequency with which states arise in parallel.” Implicit in Farris’s discussion of C is dependence upon the number of terminal taxa in the analysis 1989a, b, 1991). We will show that taxon dependence is in turn a (as well as characters; Farris, function of the number of available tree segments and X, values for each character. First define (in per-character terms, after Farris, 1989а) с = m/s, where s is the number of observed state changes for a given character, and m is the minimum amount of change a character may have on any tree. Volume 80, Number 3 1993 Albert | Character-state Weighting 0.87 0.4+ 0.1 0.24 V 1.0 20 40 60 80 100 taxa FiGURE 6. given character, Х,, and numbers of ta and 1.0. Under this model for the commen) index, data w .O implies that i s state egment. The cu vior evocative of the results of Sanderson & Donoghue (1989) Goloboff (1991), and Klassen e number of taxa (and so tree segments) increases. The А, = 0 may be reached on trees with far greater numbers of taxa hg is to be expected because Х, = = 0.1 shows a behav et с (1991); C decreases as t corresponding decrease, but C = 1. Rearranging expression (2) above to solve for s (which is equivalent to Д), s = (AÀ)[2n — 3 — p] c for the ith character can be reformulated as: (12) т il (0022 = 3 = p] к Thus, ensemble consistency index may be ex- — pressed as: c= Ар (A)[2n — 3 — 2 where M is Z,,. (14) Consider the special two-state (binary) m = 1 and M is equal to the number of variable characters in the data matrix (e.g., conversion of nucleotide data into purines/ pyrimidines). Because adding terminal taxa (and so tree segments) in- creases the denominator in (13) for all possible А,/ those not violating the case where tree-size combinations (i.e., Behavior of the erie index, C (Y axis), under different assumptions of segment length for a axis), from id scie (14). 0 e three curves represent А; = 0.01, 0.1, he absence of homoplasy. rve representing O can never show t nce per l curve displays a theoretical maximum of c — 1.0), both individual c values and ensemble C should decrease. А graphic representation of this | is shown by the three-dimensional surface in e 5; a two-di- mensional plot (Fig. 6) highlights the behavior for specific À, values (compare the similar results ob- tained by Sanderson & Donoghue (1989, fig. la), Goloboff (1991, fig. 2a), and Klassen et al. (1991, fig. 3), which were calculated under other prem- ises). Although this effect is seen here with the assumption that A, values are equal throughout the tree (i.e., through use of Хә), the above formalization of the effect of increasing taxa and decreasing C should hold for most empirical data (for reasonable As < about 0.1, Albert et al., 1992; cf. Sanderson & Donoghue, 1989). Dependence effects based on the number of characters are not readily apparent, other than the implication that more characters means more independent and possibly A,-hetero- geneous character-state distributions (among spe- cific terminal taxa) to be considered during an evaluation of maximum parsimony. 764 Annals of the Missouri Botanical Garden PROSPECTUS Our prediction of the performance of the char- acter-state weighting model presented here relies on the relationship between numbers of terminal taxa (and so tree segments) and the ability to detect multiple substitutions; both are necessary to esti- mate Х values on phylogenetic trees. Because our weighting model is intended to provide a correction for multiple substitutions (differentially for each codon position), it should identify better-supported trees than Fitch parsimony for topologies that do not have sufficient numbers of segments for dis- cernment of parallel, backward, and successive multiple-substitution events (cf. Saitou, 1989). The expectation is that analyses of data sets with large and sufficiently even samplings of terminal taxa will converge on the same most-parsimonious to- pologies (in whole or in part), using either equal or differential character-state weighting. This has been borne out in several studies (e.g., Albert & Chase, unpublished; Qiu et al., 1993). Perhaps indicating inadequate taxon sampling, the weighting method sometimes selects topologies that are slightly less parsimonious under equal weighting (e.g., Clark et al., 3, in press). Our conclusion, perhaps paradoxical from the title of this paper, is that our weighting model actually supports the use of the computationally simpler Fitch criterion for DNA sequence data, given adequate taxonomic sampling (see also Albert & Mishler, 1992; Donoghue et al., 1992). Albert et al. (1992) came to identical conclusions with regard to character-state weighting for restriction site data versus Wagner parsimony. Equal weight- ing of characters may be a strong assumption (see d & Olsen, 1990), weighting schemes we have examined show that Swoffor but the differential equal weighting is robust to large asymmetries in transformational probabilities among and within characters. LITERATURE CITED ALBERT, V. A. & B. D. MIsHLER. 1992. On the rationale and a. of Mi ан nucleotide sequence data. Cladistics 8: 73- . CHASE. 1992. Character- state weighting iw. restriction site data in phyloge- netic reconstruction, with an example Ё a DNA. Pp. 369-403 in P. S. Soltis, & J. J. Doyle (editors), Molecular Systematics of Plants. Chapman and Hall, New York. Bayes, T. 1763. An essay toward solving a ies in the doctrine of chances. Philos. Trans. 53: BROWN, W. M., E. M. PRAGER, A. WANG & A. C. WILSON. 1982. Mitochondrial DNA sequences of Ге 18: Tempo and mode of evolution. J. Molec. Evol. 225-239. CAMIN, J. Н. & R. К. SokaL. 1965. A method for deducing handing sequences in phylogeny. Evo- lution 19: e 326. Chase, M. W HX dd PALMER. еа Floral mor- phology and chromosome number in subtribe miere nae (Orchidaceae): У ава from variation. oyle (editors), Ж didi и of Plants. P ни and Hall, New , D. E. SOLTIs, n G. пети о: D. MoncaN, D. H. Les, D. MisurgR, M. R. DuvaLt, R. A. PRICE, Н. С. Hiis, Y.-L. Qiu, K. A. Kron, J. Н. RETTIG, E. Conti, J. D. PALMER, R. МАХНАКТ, K. J. SyTsMa, H. J. MICHAELS, W. J. Kress, K. G. KAROL, D. CLark, M. HEDREN, B. S. Gaur, R. K. JANSEN, K.-J. Kim, C. F. че ad F. SMITH, G. R. FURNIER, S. H. Strauss, Q.-Y. XIANG, G. M. PLUNKETT, P:S: SOLTIS, 5. SWENSEN, 2 E Wu MS, E А Ga ADEK, netics of seed en A quences from the plastid gene rbcL. Ann. Missouri a Bot . 80: ак Crank, W. D., B. S. Gavr, M. R. Duvatt & M. T. CLEGC TM Phylogenetic relationships of the Bromeliiflorae- Commeliniflorae- Zingiberiflorae complex of monocots based on rbcL sequence com- parisons. Ann. Missouri Bot. Gard. [In press. DeBry, R. W. & N. A. SLADE. 1985. Cladistic analysis of restriction endonuclease cleavage maps within a maximum -likelihood NAP Syst. Zool. 34: 24- 34 DONOGHUE, M. J., R. С. OLMSTEAD, J. F. SMITH & J. D. лек. 1992. Phylogenetic relationships of Dip- sacales je on rbcL sequences. Ann. Missouri Bot. Gard. 79: 5. Farris, Ј. S D Estimation of conservatism of char- acters by constancy within biological populations. Evolution 20: 587-591. 1969. A successive Qn Me iv to character weighting. Syst. Zool. 18: ethods for computing Wagner trees. 3-92. S ost. Zool. 19: 83 1973. A probability model for inferring evo- lutionary trees. Syst. Zool. 22: 250-256. 1977. oe analysis under Dollo’s law. Ee Zool. 77-88. 1978. ud phylogenetic trees from chro- mosome inversion data. Syst. Zool. 27: 275-284 1 The logical basis of phrtogenene anal- ysis. Pp. 7-36 in N Platnick & V. A. Funk (editors), Advances in Cladistics, Vol. 2. Columbia Univ. Press, New York. 86. On the > of phylogenetic sys- tematics. Cladistics 2: 14- 1989a. The retention d and the rescaled consistency index. Cladist 1989b. The retention index а homoplasy excess. Syst. Zool. 38: 07 199]. Excess eund. ratios. Cladistics 7: 81= 91. Volume 80, Number 3 1993 Albert et al. ои Weighting 765 FELSENSTEIN, J. 1973. Maximum-likelihood and mini- steps methods for estimating evolutionary trees from data on discrete characters. Syst. Zool 240-249. 978. Cases in which parsimony or compat- ibility methods will be positively misleading. Syst. Zool. 27: 401-410. 981a. Evolutionary trees from DNA se- quen ces: А maximum likelihood approach. J. Molec. Evol. 17: 368-376. 1981b. А likelihood approach to character weighting and what it tells us about d and compatibility. Biol. J. Linn. Soc. 16: 183- ———. Phylogenies from restriction sites: A maximum likelihood approach. Evolution 46: 159- 173. & E. SoBER. 1986. Fassiuony and likelihood: An exchange. Syst. Zool. 35: 617-6206. Етен, W. M. 1970. Distinguishing homologous from — 7 — Syst. Zoo 99-118, 197 oward defining the course of evolu- tion: Moon change for a specific tree topology. Syst. Zool. 20: 406-4 1977. Onthe ы ш of discovering the most parsimonious tree. Amer. Naturalist 111: 223-257. 1986. The estimate of total nucleotide sub- stitutions from pairwise differences is biased. Philos. s. Roy. Soc. London (Biol. Sci.) 316: 317-324. E E. ManKowiTZ. 1970. Ап improved method r determining Wim variability in a gene and its n to p» e of fixation of mutations in evolution. laud Cu 4: 579-593. oS Ye. 1991. Weighted parsimony: Does it work? Pp. 147-154 in M. M. Miyamoto & J. Cra- (editors), Phylogenetic Ашан» of DNA Se- ences. Oxford Univ. Press, New York. E G. J., С. YatskievycH & C. K. Dixon. 1992. Chloroplast n restriction site variation in the fern genus Pellaea: Phylogenetic э dE of the Pel- laea glabella complex. Amer. J. Bot. 79: 1072- 1080. _ o T., W.-H. Li & D. Graur. 1982. Patterns ucleotide substitution in Lope qus and func Ed genes. olec. Evol. 18: е №. 1990. Maximum bee ieee of p trees, with special reference to a Poisson odel of DNA cupa and to parsimony nalyses. Syst. Zool. 39: Соот, Р, A. 1991. чо and the choice among €— Cladistics 7: 215-232. onn. 9, J. 1985. A clock of evolution. Nat. Hist. HENDY, M D. 1989. The relationship between simple evolutionary tree models a a observable sequence data. Syst. Zool. 38: 310-3 ——— & D. PENNY. ГО ТЕРЕ" for the quantitative study of uS trees. Syst. Zool. 38: 297-309. Jukes, T. H. € C. R. Самток. 1969. Evolution of protein molecules. Pp. 21-132 in H. Munro ей Mammalian Protein Metabolism. Academie Press, New Yor Кана. J.W.& K.J. ЅҮТЅМА. 1992. Disassembling Papaver: A restriction site analysis of chloroplast DNA. Nordic J. Bot. 12: 205-217. Kimura, M. 1980. A simple method for estimating evolutionary rate of base substitutions through com- parative = of nucleotide sequences. J. Mol Evol. 16: 12 KLaSSEN, C. J., ЈЕ. D. Moor & А. Locke. 1991. Con- a! indices and random data. Syst. Zool. 40: бш, " i & J. S. Farris. 1969. Quantitative P ol. letics and the evolution of anurans. Syst. Zo pi S, e 1988. The stochastic mode of molecular evolution: What consequences for systematic inves sisi e "eg 105: 565-573. Li, W..H., C..C. Luo. 1984. Nonrandom- ness ај arm ts as reflected in nucleotide sub- stitutions in зак ж car E evolutionary impli- cations. J. Molec. Evol. E Mappison, W. P. 1991. The ps and importance of multiple islands of most-parsimonious trees. Syst. Zool. 40: 315-328. & Mappison. 1993. MacClade 3.0. Гаа Analysis of Phylogeny and Character Evolution. Sinauer Associates, Sunderland, Massa- OLMSTEAD, В. С. & J. D. Ри, 1992. А chloroplast DNA phylogeny of the Solanaceae: Subfamilial re- lationships and character evolution. Ann. Missouri Bot. Gard. 79: 346-360. Penny, D., M. D. Нехру & M. A. STEEL. 1991. Testing the tica of descent. Pp. 155-183 in M. M. Mi- yamoto & J. Cracraft (editors), Phylogenetic Analysis of DNA Sequences. Oxford Univ. Press, New York. Qiu, Y.-L., M. W. Сназе, D. Н. Les & C. R. Parks. 1993. Molecular phylogenetics of the Magnoliidae: Cladistic analyses of nucleotide sequences of the plas- tid gene rbcL. Ann. Missouri Bot. Gard. 80: 587- 606. SAITOU, N. A theoretical study of the oi! y the maximum par 1-6; SANDERSON, M. J. & M. Ј. DONOGHUE. 1989. Patterns of variation in levels of homoplasy. Evolution 43: 1781-1795. SANKOFF, D. & R. J. CEbERGREN. 1983. Simultaneous Evolution, and Inference. The MIT Press, Cam- bridge, Massachusetts. K. J. Srrsma. 1992. вле рта of series relationships of Mexican and л Атпег- ican wild potatoes (Solanum sect. Petota): d en e from chloroplast DNA restriction site ах Syst. Bot. 17: 432-448. 1989. Distributions on Bicoloured Еуо- 3 . Dissertation. Massey Univer- almerston North, New Zealand. 1991 бе Phylogenetic Analysis Usin ony, Ver . Computer program re ae p» the Illinois Natural History Survey, hampaign, Illinois & W.P. MADDISON. 1987. Reconstructing 766 Annals of the Missouri Botanical Garden ancestral states under Wagner parsimony. Math. Biosci. 87: 199-229. 1990. Phylogeny reconstruc- tion. Pp. 411-501 in D. M. Hillis & C. Moritz (editors), Ma Systematics. Sinauer Associates, Sunderland, Massachusetts. Convergent evolution and (editor), Statistical Analysis of DNA Sequence Data. Marcel Dekker, New York. WENDEL, J. F. & V. A. ALBERT. 1992. Phylogenetics of the cotton genus (Gossypium): Character-state weighted parsimony analysis of chloroplast-DNA r striction site data and its оа Ep n ic implications. Syst. Bot. 17: 115- WILLIAMS, P. L. € W. M. FrrcH. 1989. "Finding the es in a given tree. Pp. 453-470 in B. Fernholm, K. Bremer & H. Jórnvall (editors), The Hierarchy of Life: Molecules and Morphology in Phy- logenetic Analysis. Elsevier Science Publishers B.V., 0 Phylogeny gie не using ое weighted parsimony m 615-626 in R. F. Doolittle (editor), Methods i in En- zymology, Vol. 183. Molecular Evolution: Computer Analysis of Protein and Nucleic Acid Sequences. Academic Press, New York Я ‘vti 929111 2661 “PE 19 peaisurO 8271:201 əuou avaoridy Up wnaijps шпірирто?) Vel 191111 (sujos 29 uexunig) £661 “P 19 25940 LOT ТЕ SM ‘9921 nəyund avooetdy "т штотовш штиоз) Я ‘ҰРТ 288101 (вос 79 nexunjd) 32661 “T 19 модју LOPT-TE SA ОРЕСТ noyunjq aevooridy “Т suajoaansd штар Я vs I£0S0T ([[P^nq) 2661 “Te 19 peang USE ед ч JAWS TEE prg IPIIRILIYIUY jaxeg шпхошоо umjXudo20]y:) повшоц D av 9€93 11 2661 “TB 19 nif) 1221:72 ПЭМ ‘612 2504) M W аваовиопиу Y J Yoo (шет) DIVLOPO nzupuv;) 9 ‘VP 1€9311 £661 “Te 19 ni) 8271:65 NON ‘ST MÒ аваоепопиу peung (7T) 29074) vuuisp a vr 62911 2661 “Te 19 ni) 8ct U6c DON “T£006 MÒ APIMPUOUU Y т DIVIALMU puouup 91 ‘Я ‘VOI (95249) 2661 “Te 19 95240 82%1'62 HOIN '1092] U0s19pup “Y M оваоттрлеовџу "ds snunjog Я ‘ҰР 8739711 2661 “Te 19 nif) 98ctl6c ON '00€ 1211] 9?92P[[910qurVy ред ppodoy 214] D]]910QUp g'vc ZEOSOT (meang) 2661 “Te 19 peang ISE ZE ANIMO '9/6PZ Арл әғәэврц:ешұ [oda юуютиїш D111) Я ‘V6 79610Х 0661 “TE 19 пјмојецоиј ТАА! UMOUNUN IPII)PYIUPIRUIY “Т snopipuoysoddy ѕпутиріршу Я ys (95243) 2661 “P 19 25249 Scr U IE MON ‘ESET uosiopup "M И авзовмшешзодзју “ds D112u1204]s]F Я vs 191801 (seq) £661 “Te 19 aseq) 2811:62 NNOD “ws 597 avaovieuisiy PILA 019/110] 014011208 Я ‘УС (95249) 2661 “Te 19 9584) 8071-22 NON ‘OIZ 2894) M И oeaoejeurs]y "WS e) `[ пииошаруо ^^ D2utu48 011011905 а vs 65/1801 (597) £661 “Ie 19 95240 2811:62 NNOO “u's 52971 9792?]PUISI[ V 7] оопотђр-од тирја юш Я ‘УСТ 603111 £661 “Ie 19 Вит LOPI-IE €006F "UV “BEN ‘$: әрәэюїЗирүү 5шлеҢ (OT) asuaunjo штир MSNA got 898311 2661 “Те 19 Jape 8Cv* 61 ‘9091 Z чит) з оририл94 аваоешеху 'qqe (330g 1177102рэ9 DIUDY E Я ‘V6 ZLSCOW 7661 “Te 19 Ín19y РТА АШ! AWVL “us том IBIDROZIY Up UNAISDIDNILOA DUIYIUDA |, (омета) £661 “e 19 peang LZETZE umouyun IPIDPALÍY “ds poon{ (a11em24) £661 ‘Te 19 [Rang OSET-ZE umouyun оваовлефу ‘ds npaafuny (омета) 2661 ‘Te 19 peang OS€£T-ZE umouyun IPIDRAPIY ds гојрлгаон (a11em34) £661 ‘Te 19 peang OS€TZE umouyun IPIIRALIY ds viu10y9s9g (әле) 2661 “Te 19 peang 0S£rze umouyun аваовлефу ds aavsp Я ‘Vl 788107 2661 ‘18 19 any3ouog QZPI-LZ OSIA ‘T 70 12 emjZouo(] аваоехору “т ритја уовош рхорр Я Vel Z88101 (чоту) 42661 “Te 19 меду 8zb1:1€ ADN ‘ZIIZ чолу ovaoerprunay "Woue|g SISUIUIYO IPI IF 91 ‘a ‘VOI 188101 (95940) 42661 “Te 19 119q[y 82v 11€ DON “901 2504) d IA авазвлаоу "т unapy2ops 423p Я ‘vsti 9653 11 (малрон) £661 “P 19 95240 827 + 1:67 сап FAZFE-F961 98 010847) овооециеоу пери] n214D0quipsn ю19Ўләдипү y Я ‘УСТ PEPZOT (чәлрәң) 2661 “Te 19 25245 8Zt* UL 6c Sdn ‘02052261 98 019847 әвәоеціиеоұ nepur] vsoonnaf эп у] Я ‘УСТ S6Sc 11 (чалран) 2661 “1° 19 2594) 92v 1:67 сап '6202- 2661 98 2108147 IPLIDRYIULIY 1exoeg SUDIIIIDLZ ртјәпу VSI (ua1poH) 2661 “TE 19 25849 8cvrL6c Sdü “ZE mojaiqupopy o 422u121$ овооецушеоу aou ‘ds sisdojnpyg 8 ‘VSI #65511 (чазран) £661 “TE 19 25845 81:62 бап 8322 19 19 uim] аваоефивоу чәлрәң топа siyivdopido7] Я ‘VSI 026107 (палрон) 32661 “IE 19 модју 8271:625 сап '9Р2 719 1? излрон овозецшеоу эше” (^xss104) DIOPO n12175nf 86v l-696 VST 1001440! 2661 “TP 19 peeisu]o 'LS6'£SL "07225 O'IOO ‘110-06 paruo SUSPEN 198A СТ) рирэмәшр оттп Я 'ус1 €6Sc UI (полрон) £661 “Te 19 95240 8crl-6c Sd(] "s Dssaqjay nui2su] IPIDPYJULIY ug ^W торузвлој s21520d4]] Я ‘УСТ 988101 (ue1paH) £661 "T? 19 25245 8ct 1:67 Sd '9£0£-226I 98 010547] А "Т sumond онә Я ‘УСТ +88101 (полрон) £661 “IE 19 eseu) 8zZb 1:67 Sdn '2122-0000 9g 2108947) овазециеоу зао риолтоошту рлритјоцају Я 'ус1 765511 (чалрон) £661 “JE 19 25849 8ct loc Sdn “us Og au) nq оваоецивоу џовлариу * | спирјиош гпујирој 82661 “Te uotis Juonen опело] „зопапђос долпов / 19YINO A Аше satoadg 19 əseyy -SIDJIB ut әп; aseq -Р]Р(] Зшрлозов si091p) sammurej Aq Ajeouoqeudpe родивлле әле әѕәцү играло) үрәтит1ор мпохту 211 fo sjpuup. ayi jo ansst sup ut pausiqnd ssaded ш pəzÁjeur satoadg 'snuad Aq Aqeoueqeud[e ровивлле әле Aay) 'somrureg шуим (полајуце(ј 01 Burp1o99e sjooouour :jsmbuor) о} "хтамааау 767-785. 1993. ANN. MISSOURI a GARD. 80 Annals of the 768 Botanical Garden issouri M VI 2982171 (=[әецэтуү) 2661 “Te 19 wry Str 9uou 9vooPl19]Sy ?1oJqe^) (55971) sapr0y3uborp штуАцаХ 0 Vl 2792 11 3661 “P 19 шу 8z3b 1-1 MYVN 2222 упш5 q IRIDRIASY :330y ргоурир19 sisdoa107) Я ‘Vii TTOtUI 2661 “PE 19 шу ТА АЫ! A '£rOrP wong аваовлаву читм рзодта DIWSOID) VI 8T9tTlTI 2661 “P 19 Wry КАЛЫ! HOIN '000[ uasuvf зезовлоу puourey шпиплхош штшоујирвАлу) Я ‘ҰРТ 0227748! 2661 “PE 19 Wy ТАДЫ! A ‘Sort “omaq 3LIILIISY ‘ds ригрјошолц) Я ‘Ytl cSO9tTT 2661 ‘18 19 шу ТА АЫ! НОИ “U's иззирј аваовло у "I sngÁnn umn401217) Я ‘ҰРТ 798€ 11 (SPN) 2661 “Te 19 WY 8cvV UI HOIW '06686IId FASA 9?22Pl9l]sy 7 $mu012u1) snuum14v7) Vtl Ssr9€ TT 2661 “Ie 19 шу 8zb 1-1 эп 6228 Jnpuso зезовлову әҳе ^4 ^S (A00H) гири muzeadsouuajg ҰФІ 6S9£TT (SPLN) 2661 ‘1° 19 шу РТА UTI HDIW ‘206 иззирј аезовлоу MPIB `4 ^S (IPA) 216400102 твгррилод Я ‘УРТ Iv9£TlTI 3661 “PE 19 шу КАДЫ! HOIN ‘116 u?suvf cod T wngofojma vaya g vs (95245) 2661 “Te 19 25240 8Zt 1-67 ПЭМ ‘021 2594) fi И әвәэвјәроціѕұ "1904 (^T) омрап vyoydiuy 9 “vs 50501 (Ang) £661 “P 19 peang TSET-ZE ANWD '9z6rz Хрон аваовјароц у 1exeg DIDIISD/QNS отујлотор g ‘vs 6с0501 (Pang) 2661 “P 19 peang IS£Uct£ ANWD ‘22662 Хә]ро19 aeaoejapoyds y “wing ("Т) олга 2o]p 820501 (meand) 2661 “Te 19 meang TSE TZE YON ‘70902661 Папа aeaoeeedsy 7] sypuinfgo sndvavdsp 86v 1-888 Я ‘VST O6£T* TI 2661 “P 19 pearsuo “188-£01 0700 ‘010-88 pvoisun) aeaoeperda[psy "ү 9mipoxo soidojosp: Я vr 7992 11 £661 “Te 19 nij) 8271'66 NON ‘81016 MÒ IPIDPIYIO|OISU Y "AO Muay DUNIDS Я ‘VP 0673? 11 (peoisugQ) 2661 “Te 19 95840 8cv 1:23 9uou IBIIPIYIO[O1SUY “] әѕиәррирә wnivsp Я ‘Vi 0£93 11 £661 “Te 19 MÒ 8271'65 ПЭМ 61016 MÒ aea2eiqoo[oisu y "шет apenas isa g'vo SI8I8W 3661 “Te 19 meo TOP UI 9uou аваовзалгу yews ( D0u2312G v9 TYISISNW 2661 “Te 19 Ney ТОРТ-Т əuou 98329221V -boef т юутиз]дәл xiu204d Я ‘v9 £I818W 2661 “T 19 mex TOTL-I£ 9uou авозвзолу qm py зироутиј/ оду 9100]JN “4 Я ‘v9 ZISISW 2661 “ [e 19 meg РОРТ-1 2u0u 9?32?291y "Н (9109 73 ^H) $721751pqns snaonjdowuXa(q Я vo 2661 “I 19 mex А IU i 2u0u оваовоају "]819() DUDIIIDISOI DILOPIDUIDY/) v9 I 1818 7661 “I 19 mex 82%1-1 auou авазвзолу "no su 020477) v9 отатви 3661 “I 19 mex) ТОР 29% euou оваовоалу Hey столеп спшојо) Я 'VrtI Yc6101 2661 Te 19 pealsuo 87112 HOIW “vs игзиој зезовцегу T хтац рагрон g8'vcl TCCITI 2661 ‘е 19 Buery LOUIE HN 7 юпу-ш1[ аваовцелу "9W-pueH snyjupAyovys хрираоја (у Я ‘Vii 991111 (sujos 9 naxunig) £661 “TE 19 25945 LOVUTE SA ‘IZET nequniqd аеаовцелу "1 »souids оролу Я ys (asey)) £661 “Ie 19 25945 83v1-T€ ПЭМ ‘012 2504) AMA W TERMINY HoH nsyon umpadumds gs £9696W (meand) £661 ‘Te 19 meang TSE TIE HO ‘EEZ q2u2344 зеаовлгу “Т $27017D198 05141 as 0£916]N (meang) 2661 T? 12 peang ISEL-ZE YON ‘10502661 Nang ILIMIY ламон ојпозпитш тишә Я ‘УС 62916И (meand) £661 “TF 19 peang РОР Т-СЕ DLA ‘STEP раотон әвәэвгү ug ^y sdooun sAyovisouwdy SSE 6I€I as SZ9TO6W (meand) £661 ‘Te 19 peang “023 1'65 HO ‘ZEZ у2и214 оваовлу “1 SMUD]DI sni02p Vel essesw (Зен) 2661 “P 19 aseyD 10911 AWVL ‘0997 21123 дваоеђојтђу чону юмойшоа хојј Я ‘Vii 826101 (95845) 42661 “Te 19 медју 8zb1-1€ ПЭМ 611 2594) M W aea3enojmby qunu] n7Du242 хәр COTTUI 2661 “Te 19 peoisuj( vebI-8Z 1 ап “££0€ 42u24g `9 оеаовиАооду ЭЧ »sooimaf visdoy Я ‘УСТ 829111 2661 “P 19 peeisur() 8271-201 9uou аедовиХооду `1 шпиларииро шпиХэоау Я ‘ҰРТ 021117 (sujos 9 похипја) 2661 “Te 19 25240 LOTUTE ITIIA '8686 21041 avaovidy Puyang 1402242 ототирс 82661 “TP fuots {чоцеџә amerang ,20uanbag 39.08 / I3Q9nO A Ajue y зато С 19 9584) -S9998 | ut 21314 aseq -eje(T "penuguo)) ‘хІамзаау Appendix 769 Volume 80, Number 3 1993 Я ‘УСТ 36€ 11 2661 “Te 19 peaisuQo 87t L'£0I 07100 ‘200-85 pveisu) ававовГојррпа "(oue1j npiapp víojppng Я vrl I6€Tv lI (aud) 2661 “P 19 peaisugo 42 FA (NI “ч 2214 аезовштла "идола (Т) DSOULÍNUD] әтлә Я yo J) 2661 "T? 19 anq 742 umouyun авзовцошолу ‘ds трио], Я yo J) 2661 “Te 19 ПАП РОРТ-1 имоимип аваовцошола тор (ломе) sopr019Ap DANG v9 J) 2661 “Te 19 ПРА] ТАШ; umouyun аваовцошолу задарив ла рирјиош руцогн vo X19[)) 2661 ‘18 19 peana POFL-OSP СТР, umouyun aeooemauro1g ‘ds prunvondo1910]) v9 PIO) 2661 “Te 19 РАП ОР Об" '6£tt7 МО а ord ‘ds 5150070) а ‘9 MURIO) 2661 “Te 19 ПАП ТОРТТ uMouxun аваовцошолу пола (7T) гпвошо spupup vo PIO) 2661 “Te 19 елп 8gcv UI uMouxun зезовцошолу ламед (OIII) пипируо рошузој 401 €SLS6W (uewpoy) 2661 “T 19 95249 9ecv Uc SIM 932 ит] X пол аваовлоргзицовјолу SWH SISUIVIS Dn4a3pi2uyosi24g] all (ueupoy % 39114) 2661 “1° 19 95840 8cv 1:12 амі “vs 22144 ƏLVIDVIISSVLIY uojug (usmg) vivuuid DAaqUDIC а VOI (1eu[eq x 1ue2nN) 2661 “Te 19 peeisugo ТАДЫ! umouyun aRaoROIsselg 7] аза а шро »oisspag (URBINA Я ‘УСТ 99948! £661 “Ie 19 peeisuro 83b1-£01 N) 602028 Od t9equew аваоешв8влоц "Т su22s210q10 штаолпопон Я 'уст 089111 2661 “P 19 peaisugo аа71-201 9uou aeooPutdeiog “1 sy durmio o8»120g 401 (01€) 2661 “Te 19 25840 ТА АША SIM "ws иозадајру звзововашод 'Aneag 'd asuazodouong хрдшод а'уст 619111 2661 “I 19 peeisu]o 8cTLI-£0I 0109 “us sipmydunpgop зеаовшоп а] ‘ds 24919) 91 ‘All 68810 (Ò) qz661 “18 зә моду 8С ПЭМ ‘SSZ 2504) M W arave g ти түтәй 91 8 ‘v2 25911 £661 “Te 19 тфу 82162 ПЭМ 2 MÒ avaoepueqiag шед) (9UN1J04) тәррәф vruoupjy 91 ‘a ‘YL 092801 (ә) £661 “Te 19 95240 £81 1:63 NNOD “ч 527 овозврнач 54 "xqouy (T) ггр1о пору) un Ay dono) 91 ‘All 888101 (28240) 42661 “TP за әү IS£ l2 ПЭМ ‘SZZ 2804) M W әвәэешо8ә paumsups х ponpojowu pruoSog Я vol TPEBLIA (од) £661 “Te 19 25842 УА А! AWVL “85971 913124 овоовјед “Т оштаош sug Я ‘V6 vOSTIM 2661 “18 19 Яшәү 20711 DNV.L “us uvosmm И а eH 1 291? 0112809 a ‘VP LZOLLW 1661 “18 19 se] 2811:65 NNOD “и 527 ovooeKeo1eg TEM 920/1207 оорго Я Vel (95845) 2661 "I? 19 95240 8Z ДЕА ПЭМ "FII 2894) M W 9eaovurures[eg "алао у sisuadpo зигтодашј g ‘УФ ZE9ZTT £661 “Te 19 ni) 87 + 1:67 DON '0£006 тд avaovÁo[reqoisny SUYA VL 7) зигрирох рАорофолупју Я ‘УРТ 9v9€ 11 5661 “Te 19 Wry КАДЫ! HOIN ‘S66 чә$иэ[ BESIEIOISY. sse] pujofipidsour тиоилод Я ‘ҰРТ LYO9£lI 2661 ‘18 19 шу ТАДЫ! НОИ “us uasuof аваоелојзу "1 snyofissod uogodosiva] Я ‘УӮТ 18911 2661 “Te 19 шу BPTI HOIN ‘S06 uesupf әкәәвлә1вү "Т 212242 59198 0] VPI 9LOETT 2661 “IP 19 шу 83b1-1 HOIN “u's uosuvf а са | 9u9217) “4 (IMH) 51220] nis23018 Я ‘УРТ ££6€11 (spego) 2661 “Te 19 wry ТАДЫ! OW '9c£cZ 98 unossiy ey опо ѕәртотирутш озәшәс Я ‘VPI 159€ 11 2661 a 19 шу 8zb1-1 YO ‘SOIT uynug 2) 9?92?1919V Jayeg 14D xo vydivoordig Vil (S[2BY9IA) 2661 “18 19 шу 8cT UI auou овоовлоу U] 021205 DINDI V*I 626£l1 (SPEAN) 2661 "Te 19 Wy 8zb 1-1 SAVL 68 FH Md әвәовлә1вү q snnuup гпујиођој Я 'vrI 9:198! 2661 “P 19 Wy QZPL-I HOIN ‘S16 uasuof аеаовлоу snjog 1uosaum/ рлодла о) У 8292 11 2661 “P 19 Wry ТАДЫ! HOIN '216 vasuvf SHOEI. алооуј suapua]ds тиргро) VYI 0661 “1° 19 uospny УА АЫ umouyun оваовлоју ION '5 (Buaidg) pasaur, 01428014 Я ‘ҰРТ 6€9€ TT 2661 ‘18 19 Wy ЗЕРТ HOIN ‘966 игзиој 9?22Pl9]SV "Aquuez xo "urgoH “O DuD1122412qQ 012124 Vt 6r9ETT 2661 ‘1° 19 шу Qtr НОТИ ‘806 uasuvf оезовлоу чозјоцо у *£) сигдплолуо штаотрапа ДА! 0598171 2661 е 19 шу 8ZP11 HOIN 7001 uasuof овоовлој у yoy vinyoxa гаоитол Я ‘УРТ 9€9€ 11 2661 “P 19 uy 8Z +11 НОИ '206 чәѕирг 9?92PIO9]SV "T spand D32i101da40un(T 52661 "I? {чо |чоцеџә олтуелој „зоџепђас 921D0$ / лоцопод Күрше sai adc 19 astur) -SIDIB ut 910314 əseq -eq "penunuo) 'XIGNdddV c Ф g © (5 Г) © © Ф £o у a O E pa | 23 Ф с o с := < > 770 8'vt OT9cTI 2661 “Te 19 nif) 87t l-67 ПЭМ ‘#02 5990 M W svoryje Io) Ploqatg snoruodol snyruviory) a 'V6 СРРТОГ 1861 * re 19 D[sM?1n7 ТА АЫ 1 umouyun aeaovrpodouaq?) 7 D22D42]0 торилаб Я ‘V6 Sc6SIX 0661 “Te 19 uospnH 8711 umouyun зеаотродоиоцђу "Тота хојашк Я ‘v8 £29ITTI 2661 “TP 19 peaisuro А ДЕХА 0700 910-06 prasuno evaov[[Aqdimpro1o7) NZ 9 р[оч!ә$ uunsruodoÍ штјацџа трг 2661 “Te 19 ni) 8671#7811 ПЭМ ‘22016 тд Я ‘VP-1 002и 1661 “P 19 вәт 2811'66 NNOD “u's 527 әвәовүАцЧозвләгу v] шпвлошгр штууХу4отоләл) at (Vud) 2661 “qe ja 9seu) IS€U 42 28 Aajayseg on avaoexejojeyday `22п7 2 рјодек разратар snxpjojpude-) 9 'VII 768107 (sapos 9 ue210W) 42661 “TP 19 119q[y LOVI-IE ПЭМ ‘ZEI 28049 M W avaoejo[eudar) ЧЕТ $14»7n215of гтојоџаг Я УП T8I£TlTI (95845) 2661 “Ie 19 25240 ТАДА ПЭМ ‘ZEI 28047 fl И а и) PIeqers (`чипч,) ѕлзоуо snwAuong 91 ‘Я УП £68101 (asey)) 42661 “P 19 119q1y 87112 ПЭМ ‘STS 9804) fl W әвәәешлепѕег) "T 221031] 0и110тѕ07) 8'v6 6692: И 2661 “TE 19 tseuuer) 10911 umouyun зеаов||АцдоА лег) "Т 5nj144doXavo зпујитуј Я ‘Vol 12956 2661 “Te 19 uewpoy 87t LU 26 98 utsuoost A, овазвоџе) "Т vdvdnd 0275) Я ‘Ytl 656101 (SPB YA) 2661 “TE 19 peaisugo ТАЛЫ! HOIN “016 ussuvf аваовцој део) “Т ођојлл22р wnusingry Я ‘УРТ 289111 2661 “I 19 pea1suno 87 1-£01 07100 “us pvorsun) аваоецој аде“) exeig (7T) 57970 зодлоомоцашАе g ‘VFI 9907 IT 2661 “је 19 any3ouo(] TYLCU£OI umouyun araornosudey "UI DSOUI22D4 snonquing 801 (¡oe x) 2661 "је 19 aseur) SctU Z6 SIM :001 angiasspT > si aevaovieddey әәдәрив: упојпларо SNYIU DIY NIŞ g80T ООО Т1 2661 * те 19 ugurpoy ТАДА ON “vus zpnquauyg-p аваовлеа де) "9ong psouids D1UYN19QI0N 801 SSLSOW 2661 “TE 19 ившроц 87? I-Z OW “us zvquoug-p оваовлед ко) pot) рирмојету 2u02]) Я “vol PSLS6W 2661 “T 19 uewpoy 8cv lU 26 SIM “STEOE St avaovseddey "boef одору siunddv?) 9TTSOT 2661 “Ie 19 qus PLET-IE sn '£€9-97 55213 9e32euug) "users 11134413N} DUUD) срусол 2661 “T 19 Ҷишс TLELI£ SA 'PZ[ 1-08 55213 әвәәеииег) 'п PPU? риир)) 91 ‘All (MQ) £661 “P 19 aseyD BEPI-LZ ПЭМ ‘#0026 MÒ əLəveqeuueg “1 sand] snmuny Я ‘ҰР 2661 “Te 19 ni) 8cv l6c NON '21006 mo оваоијјеџе) 'uj1oe?) (Т) DUDUN DIUDI) Я ‘УРТ 126107 (spego) 32661 “Ie 19 модју А А U 26 НЭТИ '686 u?suvf aeaoe[nuedure/) "1 $nu142 012907 Я VrI 198€ 11 (SPENN) 2661 “Te 19 pearsuo КАЛЫ! HOIN ‘#86 vu?suvf evaov[nuedure?) пе 250ymuma vjnupdurm;) Я ‘УРТ 098€ 11 (SEY) 2661 “1° 19 peaisuro 87 Да ‘226-1 0100 “85353 4231zung аваовлаолје) ssnf зартошгујир sidoog Я ‘vr 6£93 11 2661 “Te 12 MÒ 87 + 1:67 NON '29 т) оваоецшеодјео) xurT (T) хозг0 14 зпугириошт у) (ueh Vt 1667171 (pgeisugQ) 2661 “Ie 19 25845 8cv* lc N) 180018 98 !e?qnew APIDRYIUPIÁ 0] "т snpisoy smjjuvoAqp7) Я “vo S£9Z11 2661 “Te 19 nu) 8zb 1:67 NON “01006 mo aveoeqiueoA[e) Buey) [ `$ 29 JUJ sisuauryo snyupoXqv?) 85 (asey)) 2661 “Te 19 25245 8cv lc Vd '909 га аваземоцзојео) Aoqum() smutunu г1042010) Я ‘vst 189111 2661 “Ie 19 pearsuo 83 y 1-€01 NNOO ‘2512 Pud авазвцошије) Áqreq рчәшә -ysing D¡4ydo1a394 2121411]]v7) а ‘VP 650: И 1661 * те 19 sa] ТА 1:12 NNOO ' ‘u's s27 9ea»Pquioqe?) Авло) "V Dunpiu1]0o4D2 рдшодр) 2661 “е 19 nif) 8271'0121 ПЭМ 'T£OI6 MO Я ‘Vr 8Z0LLW 1661 “Te 19 вәт 607 1:23 NNOD “us 527 9?o»?quioqe) "[ ut чцәше) `4 ‘f 142924428 Diuaspig Я Vel 056107 (sujos 3» ue210j) 2661 “Te 19 моду 07115 ON 2625 11019P109 avaoepyqag uoqoue|d sp1u02408 vjnpiuoy ast 168101 (28245) 42661 “P 19 меду 1521'62 9uou oeaoepiqÁg qsmes p2oyiur 51948 91 ‘dl (95245) 42661 ЛЕ 19 25945 092 1-8S ПЭМ “20% 599) M W а: | XY suaquin204d n4pupsKuong 91 ‘Я ‘VOI 068107 (95845) 42661 “Te 19 eq 87 + 1-0€ 8811-5994 9 ‘doi, ррцолед 389238.198M4 uonug SisUINÍDUL DIISING vs (25245) 2661 “T 19 peeisuro 87 + 1-0€ NON ‘ZST 2804) MA W 9gaogruuguung Т 2409 viuupuuang £ITTUI 2661 “Te зә рвәјѕщүс) PEPI-LS 5ап “660€ 42uwa4g `9 avovov(apppng чә], (Чед “JA) »ijoftsz2a1p отшәроәу 82661 “Te uois јчоцеџо әлвә „зопопђос аолпов / лацопод Á[rure y seed 19 9seu) -SIJIB ш әл] aseq -eje(T "penuguo) -XIdN3ddVy 771 Appendix Volume 80, Number 3 1993 а'ув 681111 2661 '5пј06 x ueZ10] LOPUTE SA ‘1512 UDILON] aeaoe[nsse17) 1911194 ә PULY punjuouriZiDp 2042uD]UM Я ‘v8 381111 2661 'snjos 29 uez1o0jy LOTTIE 10829 99 uoiSunung dd O UBIO (UOSIE AA ^S) ортоѕта nAa]pn(q Я ‘УВ 668101 (sujos 2 чело) 42661 “Te 19 119q1y LOFI-TE SM ‘ZSIZ UDILO y] oeaov[nsse17) uasqooe[ x ладин рирлотилтш DINSSDA/) а vo 29601 2661 “Te 19 Ҷишс PLEITE à ÞILL62 55913 9?92P]s07) “UINYIS “Y 2055 ририр соттодитод 0] sere PSPSOT 2661 “Te 19 Ҷишс PLETE TAS “111-62 55213 9v92v]s07) (uassajag xa ‘ddaog) сплоулип sniso0u0 y 9 ‘v9 77507 2661 “Ie 19 Ҷишс PLETTE TAS ‘0550-98 94 4419S ги wees 450) “SOONG SUDAN, 51056) 223111 €661 “P 19 Suery LOPUTE SA “U's 9у-и07 аваовилог) Suoog хә nA "X "D 2Раштуорпоо DIXUSDY Я ‘VPl 922111 2661 “Te 19 Buerg LOPI-IE Z16 q1y plouly оваовило7) -naq '4 (qunyy) 22тиодогГ viZuimjor Я ‘ҰРТ cti 2661 ‘1° 19 Buery LOPI-IE AMY “us иолошођ) әвәэвилогу 16104 4) DP19N] DIUI]9S11£ ©6с11Л £661 “Te 19 Зивту LOTITE NN '816 %ріртрд ад, шс “y 9 (umg) олигр пат) Я ‘ҰРТ 145111 2661 “Te 19 Buery LOUIS IIc vL Чу 3utqKng 9?22?UIO^) [MOLY 42750210102 D14010) a vcl OZCIUI €661 ‘е 19 Buery LOTPLI-IE V-LO-pPIv “GV p|oury әвәэвилогу ицшәдие y 1427]pom snui0:) 613111 £661 “Te 19 Burry LOFT-TE V-9ST8 чур шү ird afi ong 2р PAG ғ170и12у/о зпило) 812117 2661 “Te 19 Buery LOPT LE eup ‘9g #шшипу әгәовилог) ПЕМ Р®шо]фо ѕпи207) 111171 2661 “TE 19 Buery LOPUTE SA ‘EZI ZuviY 9?92PUJO^) “yey опбудо пило;) Vcl 913111 £661 “Ie 19 Buery 0715 SL'8vL Чу pouty 9=Э0Е0100) "юш 5100400) Я ‘УСТ 47428! £661 “Te 19 реәзѕшү) 92v Uc auou аваовилог) eoueH ропоу сгпило;) Я ‘Уст 51111 £661 “Te 19 Juey LOTUTE SA ‘OSZ Pumy аваовилој “J ормој snuso;) Р12117 2661 “Te 19 Buery LOUIE 0661 Ny 2ury-0 аваовилог) ULISÍUB Ay SISUIVIYI snuso’) Я ‘УСТ 868101 (Яиету) 42661 “T 19 мәү LOTITE SA ‘861 Pumy 9892809100) CLAU PRUVI FNUL) 215111 2661 “T 19 Buery LOVI-IE SM ¿91 umy овоовилог) "Т пђојлилођо snu107) 91 ‘A ‘VSI 012117 £661 “Te 19 Furry 10# Г-ТЕ агу ‘рем '5'0 әвәови10/) 'qunqj voruodoÍ vqnonp 91'811 268101 (mÒ) 42661 “Te 19 модту 8cv 1-96 NON ‘SEZ 25049 MA W SESONO) 7] ођојилАш 11121107) a ‘vst OOTTUI 2661 “Те 19 peeisuo vL3T EOL 07100 ‘S1088 роәтхш) OUODETMIAIOAUC) "1 2u12202 vaowod] а'уст €99111 7661 “Te 19 решо) 82Р1:201 auou aBadR[NA[OAUOT) "1 40102141 snjnajoauo:) Я ‘v9 20201 (meang) 2661 “Te 19 meang 99€ 1-Z£ ANAOD “PZEL 1591407) f аваоешјешшо“) assog риллда2 DIJUDISIPDAL Я ‘9 170501 (пела) 2661 “Te 19 je^nq 96£Uc€ ANIMO '086Р2 412PD18 PESIEUY OUI.) ppijjod “ye опирозоррај L6€ 1-689 а yo сорсол 2661 “T 19 Ҷишс '199*60€ 'LLZ-1€ SN ‘86-92 чәроу әкәовицәшшогу epniejN DUDISNUOIOS пуирозоролј 89 220507 (meand) 2661 “Te 19 peang 165 TE€ нэп '£090c66I ПРАТО oe oT виц 00) 7] вигррудигд риђешшо) 311 (uuo?) 2661 “Te 19 95840 ТАДЫ! SIM ‘001 1105 9?22?]21qui07) 1 »dd»1»2 ођритш19ј JIT ‘VOI 876101 (esey)) 42661 “P 19 моју 8cv LIE HOIN "s uoszapup ^M f 9v227]2Jquio7) Тори siponbsin() Я ‘vs £L9c 11 (esty)) 22661 "Te 19 ueqiy КАДЫ: ПЭМ “601 2504) M W dd оу ANS unsorads umonjojo:) Я ‘Ус (95245) £661 “T 19 2524) 8cv 1-66 ПЭМ “PST 2504) MA W ам зын, A A ON и КАЈ а ои | Я ‘Vel 609c UI 2661 'eseq) 2 чогу Berle ПЭМ ‘#881 чолу ovale lalo т vuyofriqo 0241217) a‘VIl 41008 2661 ‘510$ y URB IO LOPUTE PELEG 9 doi] рэпро авоовивјецовА у) Yas Д рѕојиәшој ирот Я vit 821117 £661 'sujos 9 VeZo LOPI-TE 11€-92 9 doa] ppyo4104 аваовивједовА 147) "1 92921 SNUDIDGOSALY;) Амон Y gt £992 11 £661 “TE 19 nÒ 1521 26 NON '20026 MÒ эвәэвціиего[цэ 'aqns (Þig) опојрирлу vapuvsivg ar 6792 11 2661 “Te 19 ni) 87125 NON '8££ 2504) m W aeaoeqiue1o[u7) “MG SUIISILOQAD шпш«оХрон $€661 “Te fuots Juoneyo олтуело] жӘәиәпһәс̧ долпов / лоцопод Ајпше y saradg 19 9seu) -5900в ut 210314 aseq сеје "penuguo)) ‘хІамяаау Annals of the 772 | Garden issouri Botanica M 8 ‘WS (омета) 2661 “e jo 9seu) SvttcUI uMouxyun әвәәеиәвеэвеі(] jog sore "agp 801 (uepueueáq) £661 “ye jo 95249 ТА 1-97 49 ‘od UDPUDUDAD(J aea9ed1e30191dí(] uolysy ( ( 091014 801 (uepueueAq) 2661 “ye 19 513119] 8271:92 AO “а UDPUDUDALD(J оваов тезолој 44] вон. suvjndis DILOYS Я vrl v38£ 11 (ево) 2661 “Te 19 peeisuro КАЛЫ! HOIN ‘156 vosuvf оваовов5 "XouoH (7T) спао snopsdi] Я ‘Ұс (95945) £661 “Te 19 25840 8c* Ll T£ ПЭМ *261 2504) A W 99999930080(] qduog 29 qug ѕәртоиоЯ ой va402soi] 821 ‘V6 £06101 (чогу) 42661 “1 19 119q[y 8аР 11€ ПЭМ “PEZ 599) M W seostu "| Daipul ошап 8 ‘V6 £9S29N 2661 “Те 19 Immy LOTUI 9uou 99a2JEIOIpuT ARI n42201d DIpnon y Я “vel 319311 2661 'eseq) 9 полу 87+ 1:67 ПЭМ “81068 SMH aeaovtsuadeiq `1 voruoddo] nisuadvi(q Я ‘VP €v93 11 £661 “P зә ni) 8z2*l'6z & '£9-68T1I ТИ WT оваоемопово(] ‘ds пагиг Ba 9I ‘All (чәѕиәмс̧) 2661 “Te 19 25840 83v 1:23 VSM “Z2ZZ q?uqpiq аеазвовцв] ug Y »2oyipnu sa2j2umajo[ а ‘VII (ussuamg) 2661 * [e 19 9seu) РТА АА VSH 777 3214311141 IPIDEISNP(] “DIA Dunpijpums $3]2110]() 91 ‘a “VII (uəsuəmS) 2661 “TE 19 95945 83v 1:23 NNAL "ws uassuamg аезововцв "reg (18914 7) 770421102. vosim(] 811 (ussuamg) £66I * [e 19 9seu) РА ДЫКА VSH Uus иәѕиәтс̧ Әвәэвәѕце(] "UI DUIQDUUDI DISVD(] as 6£0SOT (meang) 2661 “Te 19 peang SS£ 1:67 YON '20602661 Pang aeaoeuosodÁse(] "Mie 010/2 ио] vipunwo 91 ‘d ‘v8 106107 (ni) 42661 “Te зә моду 8с+1-62 DON ‘92016 MÒ aeaovpqdruudeq ‘ds wn ydruy doy g ‘Vel 0061071 (чогу) 42661 “Te 19 медту 8cv 1:67 ПЭМ “us иогу ərə) 1 ?2ojrua2n4 орак) g ‘v9 1с916Ии (пела) £661 “Te 19 peang 25211 YON '2002661 Pang аваовлода) 1 зтђојлилотуо плода) g'vo ZLOZIT (u91p9H) £661 “Te 19 25245 8cv 1-67 SdN ‘£174 "?4p2H aeaot1adÁ?) "OC PuUDISOY хәр) Я ‘Vs (95945) 2661 “Te 19 25240 87165 ПОМ ‘222 2807) M W әвәзеццир[оАГу jourueH ојприга vrusposavyds Я ve PLOSTT (95945 Y SIHH) 2661 “Te 19 25245 8cv LU T€ SVTd "ws иошгтуоб эвәэерғэќ:) "Т припало $0247) а'ус (пәлге{ 9 шецело) E661 “Te 19 aseyo 6811-19 LAL ‘281 € #2 оваовлепеко) АЦО штуојдрло2 UWNLISDUDA) Я ve 8£Sc UI £661 'uum() X ҳәре‹) 83b 1:61 ММП Раса РУМ avaorssaidn’) usie]N si15u2242q10p22 тиојђимрруј LLSZIT 2661 'uum() 9 x9peo 861761 MSNA '9£2ICAMSNI аваотаволап) ооп X р[одтәс (7) T) »i»4q0]op sisdofm[ 8LSZIT 2661 'uum() y xA9pe9 ТАДЫ MSNA ‘ZEZIZMSNA aLasessərdng "p supiuapi220 отуј 925311 2661 'uum() 2 ҳәре‹) 8cv 61 ММП “082121507 aeaovssa1dn: “SPW ([YBA) 27207127710 siuopaja[ 41111 2661 'uum() x xopeo 8zZ+ 1:61 MSNA EELIZAMSNA aeaoesso1dn:) o2ue1j (T) 8170709110 SONPD)IXID] d 6515511 2661 'uum() x xopeo 8271:61 MSNA “vs app) aeaoessaidnyy гешоҹ̧ DjDSST22p 010140121 #15217 2661 'uum() Y xepeo 82%1-61 MASNA TPLIZMSNA aeaoessaidn’y 31858 (uoq *(]) vsownd snipo2oqr] €LS311 £661 'uum() 9 19pe9 87161 MSNA '06£* LASN/ ovaovssa1dn:) Hed рмајио sniadiunf 16611 2661 *uum() M море 8ВОРГ ВР MASNA ZP2LIZMSN/] aevaoessaidn’y J YOOH 7424240 ршјогц] 128Z11 2661 'uum( y xopeo 8cvLl-LG MASNA *26Z#L MSN) aeaorssaidny "ү Ssuamasadwos snssaidn’) OLSZIT 2661 'uum() Y xopeo 9z*l-61 MSNA 'SEZIZAISN/ araorssaidnyy [pug (99NZ X рјодос) »snjqo siand&2orumwy:) 695311 £661 'uum) x поре.) 8с+1-61 MSNA “9ZEZZASN/ avaoessa1dn?) uu0]4 (1101) su244n2ap snapa20]v:) Я ve LESZIT £661 'uum) x xopez) 8zt 1:61 ММП PELIZAMSN/ aeacessaidn) “PR хә "ag “Y Veproquioys supp) 9 “VII с681071 (sujos 29 че8лоуј) 42661 “Te 19 мечту LOTITE SVO ‘SEIZ 19193 aeaortuoun’) [еш штиәйшштпЎ wunppjadojva2:) ‘VII PLILIT 2661 ‘0105 % ueZ1ojq LOFI-TE LEZOZ8S Чу 3uq4ns aeaortuoun’) змалриу $2P101QN1 DIINDEG 91 ‘all (ussuamS) 2661 “TP 19 25945 А АА NNAL “us то лон әвәәвіпәп”) иешәәс̧ (игу 3 ооң) опругпбитђ эт] 91 ‘All (чәѕиәмс̧) 2661 “Te 19 95245 8ZřI-LZ əuou зеаовидфтопо) "1 edad олато,) а11 (чәѕиәмс̧) 2661 “Te 19 25245 8SZFI-LZ 9uou аваовидфтопо "1 5701305 зтшпопо) Я “vol 621111 2661 'sujos W ueo LOFI-IE суд Luy vueg oypuey 9?32P?]PUIOSOSSO17) "nw шполилојјро vuiososso1^) а'ув 956101 (sujos % URZIOW) 42661 “T 19 119q1y LOFI-IE SM ‘ESTE uvi аваовјпозел“) uasnep) штоаиполдпа шпрәс̧ 82661 “Te fuots Juogyeyo әлвә „здопепђос 921n0$ / JBYINO A Apure 4 вето С je aseyr) -8929P ut әт aseq ве "рәпицио”) “XIGNaddy Appendix 773 Volume 80, Number 3 1993 6££ Ul Sb a8 ‘VII (3391) » итә) 2661 "је 13 əseyy "треба əuou əLeILqe 4 -ds DIUIYNDE a ‘VII (889) » uea) £661 ' ‘ye 19 3seu) 61 1 Ft 15 3u0u ӘРӘОРОР 4 "ZzeIn(] nmuisiaqimf nizziq]p 91 ‘a ‘Wl SP9OZTT £661 “Te 19 25245 8zv1:67 AON '92006 MÒ ovaovo[aidn ong Y ploqaig vipuvdjod vajaidng Я ‘VP 122748! £661 “Te 19 nif) 8z+1:67 ПЭМ '22006 MÒ аваоецешод па [PN ^4 ?1772uu2Q опошодпа Я VII £81£T1 (aseq)) 2661 “Te 19 95240 82v 1:27 ПЭМ “ZOI 28949 M W araoeiqioydng лаплом "y ршолуојов viqsoydny а ‘VII LSLS6W €661 “Te 19 ueulpoy 8gcvUZc пемен * "qiy од aeaoeiqioydng 'snanH E mea) 1491т4х01 газа ФА OA 816101 (suos 2 ue210j) 42661 “P 19 119q1y LOVUIIE 0870-98 'q1y 3uq4ng aeaoerydérong song ppion] vydX 193 91 ‘Я ‘VSI 116101 (RIÓ) 42661 “Te 19 медту 8zv Ll 1€ NON ‘#2016 MÒ әгәэвїшшоэәп "АЦО згртошт vrunuoon;] Я vt £81€11 (25240) 2661 “18 19 95840 8Zb1-1€ 31952-69 Y doi] руцолол oeo»e|Axoq1ÁTg иоция umsnfuoo un]Axoayia;] 89 (95245) 2661 * “18 19 9seu) 55122 NON ТЕ еру) M WwW зваовјпезомо Suo10N газлир u0 NDIOUYID"] Vel 979711 £661 '25940 Y чолу 8zb1:67 28012 'q1y pjouiy аваовош рзејод (Weeg) пгигјпазајтпа viqous7 8 'Vel 6292 11 £661 ‘PSPY x чолу 8Zt 1-67 NON 61068 SMH aLL чону иодлозолоош штипоорд 189.10 J 91 ‘a 'Vel 6*6101 (чогу) 2661 "|? 19 решо 87v 1:67 50126 Y3Mquipa 9gy avaowoug 29 4 mozeg saproapydoddry uospuapopoy y VEI €661 'авец) by UOIN ect 6c NON ee uoly зезовом Желсу (ст) DSOUIIDA 904102тә7] Vel 619311 2661 '25845 x чогу 8zb1:67 ПЭМ ‘2902 чогу SLIDE nong (818) wnyofixng шпАчаоәт Я vel 819311 £661 ‘әѕеҷо x чогу 8271:197 '179:62 сров9;, YÍMQUIPA DEN аваовош 1eume][S ('5194) о] Ау4о112 рмоџутоо) Я ‘Vel 219211 2661 '25940 x огу 8+ 1-6 716182 YÍMUIPA оян ALICIA "т $1]рї1$пю 0213 a8 'Vel 919311 €661 *aseu7) b uoly 8СР1'66 Q'8ecSYTI ау рроилу IBID џозјоцо "ey ѕпјрупираштә smqunixu Vel S19ZTT 2661 ‘aseyD y uory 87b l-6c ZE9TLO YÍMQUIPA DEN ALI o da Я vel 119311 2661 '25845 y uory 92v 1:67 OLLISL YÍMQUIPA DEM оваовошл yooy `y (човрпн) 9 n1220QD(T Я ‘Vel 909311 2661 '2584) W чолу 8271:62 181267 YÍMQUIPA 984 9v2otou] Чәчәор (7T) pm]noAqo ouydopopumy) Я vel £092 l1 2661 '288y) ў чогу 82v 1-67 28:62, 'q1y pjoury оваовома uoq “9 (Suog) риртѕиәјләш ado1ssD;) VEI 109311 2661 '2584) ў чогу 8271'0211 ‘886-67 Ферт, YÍMQUIPA оян ARICA IMH (T) suona nung) Я ‘Vel 009211 2661 ‘әѕҶо) x чогу 8271:62 ПЭМ ‘020Z чогу овазвоша пад Ds012204 DILD/9G Vel 868711 £661 '25840 Y чолу 8z+1:67 89-19€ 'q1y pjoury ALITALIA 'Sua1dg (Т) 1s4n-van зојА у 4 01500 Я ‘vel 16211 £661 ‘sey x чолу 87165 29525) YÍMQUIPA DAY ALI). Jaurequnq sisua14Dupo sningip g ‘VE 449611 (95245) 2661 “Te 19 1:410] 8Z 10€ 3uou оеаоелроц 4 "KIW ү Э оитрәәті] раргцал ан Я ‘vel 129311 2661 '2sty) чогу 8cv l-6Zz 8£0189 qdunquip3 984 aeooepuoed (18104 75) Y "M `f) општа vipuoyovjuag VE 0592 11 2661 ‘Ҷо m чогу 8271:62 6SOTPE YÍINQUIPA DAN aveaoepiioed ‘uung y мә$р1/ uogodoono'q Я ‘vel s16101 (чогу) 42661 “TE 19 Vey 8zv1:67 966089 Y31NQUIPA DAY aeaoepuoedq [qe] 2552 шп si190d q VEI 019311 £661 ‘әѕеҶ) 5) чогу 8271:62 сеорб9 Чодир DAM оваовриов а 4 xooH 10$u2]02 sapowyr«) vel #1917 2661 ‘sey m чолу 8271:62 £19191 Y3MQUIPA DAN avooepuoedq 19 y umiofi3uo] ungkgdoovaq Я vel соол £661 'eseq) x чогу 86v l6Z ПЭМ ‘690 чолу avaoeneduy "XqQon $9P109149 010130197) Я vel £19311 £661 '2584) x чогу 87v l-6Z ПЭМ 'РООЕ vory әрәзиәч а "Т оиоти19 па soaKdsoiq 96 891€11 (suena) 2661 “18 19 aseyo 8cv LU 8c auou aeaov1asoX] [qe] »7»]nyiods 0195010] 86 216107 (swe m) 42661 ' ‘ye 19 119q[Y 15€ 1-8% 3u0u 9?32€19S0J(] ug "Hd s14DJo11ad DIISOL(] 9 ‘V6 116101 (95245) 42661 "Ie 19 мечту 8cv 1-87 NON 'E!W68 игд әгәзвләвод(] ‘yey snuaofipf viasorq 86 YO6I01 (sure A) 42661 > Tq? 19 119qmy Scr 1-87 3uou 9€92€19S0J(] БИГЕ pyndiosnut ргриојј 52661 “e {чо јмоцеџо 3IN18.19Y1[ ,9»2uonbag 92JnOS / JAYINO А Аше satvadg 19 asey’) -S9998 ut əmi 4 aseq -eq "penuguo?) = “XIGNAdd VY Missouri Botanical Garden Annals of the 774 Я ‘VSI Рб1117 2661 'su[os y ueo LOUIE ON ‘EE8Z suomy әвәэв1ге[пѕѕ01‹) "qunur раорђАцаоХлоо титпиору Я 'v8 881111 0661 “Te 19 su[os LOTITE SA ‘1076 2104 Я W d 2B9IBLIB[NSSO IC) "q voru141a роу Я ‘ҰРТ £8IITT 2661 'sujos 9 ue21ojy LOPI-TE ££€£1'ZS 99 JT) N 2B9IBLIB[NSSO LC) Away sisuaquimboo D100]]0283 Я 'vII 921111 0661 “Te 19 su[os LOPT-TE SM “us 512705 o9 51106 IPIDPLIE[NSSOIL) мех) 195) хә SIBNOU | SISUIIDISDÍDPDU DIXILG 911 ‘VOI S81111 £661 'snjos 9 UBB ION LOT Г-ТЕ оотор9 'q1y BuIqdng әвәов1Кәлсу 42425 мгјожрра i217) Я 'Vr*I c£6t€lT 2661 “T 19 SPEI ТАЛЫ! 9cv61 99 Unossi әвәовтпәроогу esnery SUIISIMIÍ э]оаәрә‹; Я ‘ve 08973 11 (5845) 2661 “T 19 95845 152 1-93 NON ‘80 25047 M W 28998/9U9) "Тј uowad wnjauy g£-I (ud) 2661 “P 19 95845 1221:92 ам “vs 2214 әвәово8ҳше) "] 090119 обзито) (uey g ‘YSI 60TT* lI £661 “Te зә peaijsur(o 8271:201 n) 20058 948 PLYNEW зеаовшоизо о) чл 1135]0Y sndipz0jdang IIT 80LFI1 2661 ‘әщјед Y ooug 87t 1-97 ANI '£00I 6 9j]vusq 9€92JEIUE197) uakayp 1712018 оприг y 911 ‘VOI LOLVIT 2661 '19u[eq Y eg 8cv 1-93 амі ‘00016 оризгза 9UOOUTUVIQ?) “ABO) DYOfiuDW DIUDIA 4 S0LTTTI 2661 '19upeg y Vud 8Zt 1-97 ANI “vs 2214 әтрәоютирләс) sn[og “Т гопамгрира иојтозозлос Я “vol 176101 (wed Y 29114) qz661 “T 19 мечту 8Z+ 1-97 амі “us 2214 aeaortuRsay) шт10]104 х ummuog1vjaq FOLFIT 2661 'eupeg y Vud 82%1-92 амі “vs 2214 атаоешелог) ‘Wap, ] wnipmndinsixo wmu0sivjag £0LT TT 2661 ‘`әш[ед y Vud 8Z+ 1-97 (NI “vs 29144 аваовшело) "u9H, $7uopa1X102 штио8 1094 20:11 2661 'зашеа y гоџа 8271:-92 амі “ч 2214 әтәовтирләс) 119H,] СЛ) unzondos wmuosivjag Я VOI IOZTTTI £661 ‘wed y Vud 8c* 1-97 амі “ч 22144 әгәовтивләс) Way. (1) ои ADMI ттиозиорү 869TTTI 2661 'ашва x eougq 8cv 1-97 ам “us 22d әвәовтивләс) ззадшео 14NID]]990 штирло«) 2697171 2661 ‘Pwed y әш 8271:92 ам “us 22144 aeaoetuRles) ово 9SUILPDU UMUD) 969F TT 2661 bun. % зоџа 82771:92 ANI “vs ээм әгәовтирләс) "Тј шт2141101э0ш UMUD) Я ‘Vol 026101 (sawed Y 29114) 2661 "I? 19 мочту 82%1:97 амі “vs amd әвәовтирләс) "qu штлодрирл7 штирла«) S69t* l1 2661 ‘Pwed Y Vud 8771:92 (NI “vs 2214 әтәовтирләс) "AB шпәләи1э штиолао) Y69T lI 2661 1eu[eg x зоџа 8271:92 амі “vs 22144 әрәоюирләс) ајфомра x штрола 2697171 2661 1eueg m eoug 8271:92 4 CU's зәриәаә(т иод аваовшело<) Келе) "V UNUDXI] штрола Я ‘vsti 862 Р lI 2661 ‘Te 19 реошо 82Р1-201 9uou әрәэвирциәС) щоң D19901d DUDIUIL) Я ‘VSI v89111 2661 “е 19 peeisugo 81201 9uou әвәоеиециә‹х) Poy хә 3 eg әиуо шпорха 9I ‘Я ‘VSI 6161071 (sujos zx» чело) qz661 “P 19 меду LOTI-I£ O8ZEI ЭЯ BUY fiue oyouey аваовА LIRA) [pur] хә se[Znoq ооџатја рАллро) Я ‘Vl 192801 (S27) £661 “12 19 25245 2811:62 NNOD “us sa7 IBIIPLIBUIN y "ure (7T) 1719072905 vung (wear n) Я vel 5191171 2661 “P 19 peeisujo 8zTLl-£OI Z91098 OY Ryne! aeaoetsambno y "u[odu' suapuajds n1121nbno у Я ‘v9 829311 (uo1peH) 2661 “Te 19 25240 87166 NON '90c 25995 И avaovure[[oe| 4 "т әтри D140]]280] 9I ‘All (31eque[o)) 2661 “Te 19 25245 83? l-0c 125-1961 Y3MUIPA 984 avaovge 4 UBULIO Y D7D]]2321142a snuvjpqouog4[ 91 ‘all OSEETT (819quapo5) £661 “Te 19 25245 87125 UMOUAUN әғәое9е4 13120 (GIA) DAaquop ѕп9ојоцзоду 21101214 91 ‘All РРЕЕТЛ (рмој Y чилер) 2661 “Te 19 95245 ООУ LI ‘n qseuow “us poay f avaoese y stuaalg (eg) avsunjng snFvfoyjoay 91 ‘Я “VII 8£££ 11 (9:әдиәјо‹)) 2661 “TP 19 5845 LOETI UMOUXUN овадеве y 199M8 SNUDILLIUD SNID] 'bugo(g (330193) 9I ‘A УП (MIO) £661 “Te 19 25245 8271:62 ПЭМ *20006d то aevaoese y suanajadwos (sisdouv]sD/)) sidajosX1y’) 9I 'H ‘VII 9861 “Te 19 прамет7 8ZrI-1 имоиҳип зеаовцеД `1 шпацог wnsig 91 ‘9 ‘VII 49861 “Te 19 youpry КАЛЫ! uou qun 9€99Eq? J "Т 221205 ogvoipopy $€661 “Te {чо juoneio amyang *Әәиәпһәс̧ 921n0$ / IBYONO A Áqrure y saradg 19 95249 -5аоов ш 210314 əseq -eq *“panuquo”) ‘“XIGNaddYy 775 Appendix Volume 80, Number 3 1993 g'vo 891111 (sapos $9 полипа) 2661 “T 19 25240 LOTUTE [INVM ‘075 syo01g PESQ En qud PUIPUD дә) Я ‘v9 189311 (esey)) £661 "Te 19 25245 8cv Ul 6c NON '00€ 25949 M W Т 7] snsnfj? snounf 91 ‘all 2911 (ni) 2661 “Te 19 25940 82112 ПЭМ ‘#8 25049 M И aeaoepue[an f 199M8 (лају) 21997 DAD) 12107 (meand) £661 “Te 19 meang TSE T ZE ON ‘#8 uen әвәовәјрлио[ «ooH Dwd Doppio. Я ‘vs LEOSOT (meand) 2661 “P 19 peang COET CE ANWD ‘92667 «2jpvag этээери] “т umoiupui28 х $14] Я “vs (95940) £661 “Te 19 25840 BCP TE NON “9I 2504) M W оваовриј перо Cqunu) рхоу рээузошоиу a ‘Vr 229211 £661 “P 19 nu) 8Z+ 1:62 NON '28 то оеаоео дизд хә “XY шплојлалра штэ MPlg `1 5 a ‘Vr [29217 2661 “P 19 nu) 8Z+ 1:67 NON ‘#016 MÒ аеаоешлодвогрј (зјоц) гоигтрлјепо ummuz2dsoip] Я ‘Ұс (әѕецо)) 2661 “Te 19 2594) 87112 ПЭМ “BOI 25047 M W aeaoeprxodíH "ujadu nd4»20jda] sıxod íH Я ys (esty)) 2661 “P 19 25249 8z+ 1:67 NON ‘SOZ 2504) M W аваовртходкн әдип DIDNdDI одупоп) Я ‘VSI 146101 2661 “IE 19 peaisu]o 8271:15 9uou звазе | Ачдолран q unuprnizaa шптуАцаолран VSI t6cvl1 (peesuqo) £661 “Te 19 95940 86v lU 26 ОТОО '600-68 pveisunj) аваое | Ачдолран Хзїочгу xo “INN DIDAO D2]04pAH 10], Я ‘УСТ 916101 (peeisuro) 42661 “T 19 119q[v К адыл чоп орао е | Ачдозран (шу % ооң) юэ uoX11p0113 91 ‘A ‘УСТ 861111 €661 ‘05 % ue210jq LOPUTE SA ‘ФС 51106 Y sujog аваововивлран ysmg qno; snyd]2po 14d 91 ‘A ‘YZI 281111 £661 'sujos y ue210jq LOPUTE SA ‘0512 UDILOy оваоводиелран то], 0¡4ydos9wu vaduvspAH g8'vcl 181111 £661 'su[os y URB IOJA] LOPUTE тпа “us раојтн оваововиелран “INT m ploqets 51119013 тлгто( Я 'Vzl (чогу) 2661 “Te 19 95845 8Z + 1:67 ПЭМ “us uox аваоводиелран "ds отіоштэә( Я 'vzl ЛИТ 0661 “Te 19 suyos LOTLITI£ SM “82PZ 51106 X 51106 оваововивлран по], и vi19quad 1D) gs 8£0S07 (meang) 2661 “Te 19 meang LLeU6c Suou аваоециповАн dossaf (1əğeg) 1701205 (711126) 14noqopo] Я ‘vs (95245) £661 “Te 19 25840 8cvY 1:63 ПЭМ '92I 25049 f W аваоецуловАн “3 AH хә "ален тоа D231mogq Я VII 936101 (95940) az661 * T 19 мету КАДЫ HOIN '10Z£I uoszapup “Y 41 9USOEUITHIBET папу v22fnups]pq virum Я ‘VP 9#9@11 2661 “18 19 nif) 8Z + 1-67 MSN ‘626 "015241 На оваовлривјешн = andeidg (епу `4) »un2aD42]2Q оштиутдјоо) Я ‘VP 0S9CTI £661 “Te 19 ni) 8cv 1-66 NON “896€ uossouuag "f 969) EIDUEUJSH "Т 2428120 тририлон Я? 1%9с11 £661 “Te 19 ni) 87126 ПЭМ “ZTE 2504) f И эвэоериенер ‘ds спалооолА о) gs 9205071 (meand) £661 “Te 19 peang Sce 6€ ЧОП “10903661 Pang SES) врцјеоолошо >| `1 vaynf =1]оэоләшән cSTSO'T 2661 TP 19 Qus PLET TE TAS ‘2201-62 $8944 вон "us 7) "y рупа piuooijag Я “vo TISTSOT 2661 “Te 19 Ҷишс PLET TE TAS ‘0791-08 99 £9125 70W ahaa ad a | quag ругойѕтр] ошоәцән 91 ‘Я ‘V8 (ni) £661 “Te 19 95940 8cvr U26 ysinquipy ‘9ч энзэнрцәшиешен J Yoo] пиозаштуо viajopoy y 91 '88 (ni) 2661 “Te 19 25245 ТА ДАША NON *10026d то аезоврцошешен JOYA 977222140 лодшрртбуј 9I (ni) £661 “ТЕ 19 9524) таа NON ‘#00264 то аезоврцеашешен eoueH риозошлој snquppinbrT 91 ‘Я ‘V8 ¿36101 (NO) qz661 19 19 мету 8cvU 26 NON 'S£OI6 MÒ овас врђоштин AO: a Day 91 'H8 (ni) 2661 “Te 19 25845 83v 1:67 q3anquip3 gy РЕФ РЕРДАММЕКН КОШТА Я ‘v8 S61111 £661 ‘0105 9 URB IOJA LOPTLE SM ‘1661 MY 0g yog aeoorseloleyy р[ечлә] suassaqyoxa шпуХцуаомлу Я ‘vo (95245) 2661 “P 19 95845 TA ERES NON “651 ?5712 f W оеаовлорошоен "а snpiapy soyruozo huy 91 ‘Vol ‘dd 981111 £661 ‘sujos 29 URB ION 07115 NALA “us 2429239n4y 9) uepur] отолиош олошипо) а'ув 108111 2661 ‘05 ә ue210 LOTPI-I£ OH “us иорлој аваогмејпазволо) ‘| XooH DITUDUSD] роралрорлла ј от'8'у8 TOZIUI 2661 ‘105 zp ue21oj LOPUTE SA '0222 105 2р SUYOS әвәэзвгрүпззо1су ysing wnainv səqıy g8 со 111 £661 “SUJOS x ued1oj LOT ITE XAL ‘652 z2y»uvg IPIIBLIB[NSSOC) тәпшеу smjpofipunjos 401197801914 Я ‘УРТ 10617 £661 ‘SUS 29 че8оуј LOVUTE SA ‘#212 UDE LON ода сео, ‘(Bug (2910 1) sidsno1] ошопоАца 82661 “Ie о Juoneyo олтувлој «әочәпЬһәс 921DO0S / I9Q2nO A Apwe J вотоодс 19 95849 -S9998 ut 310314 aseq -eq 'penunuo? — 'хтамааау Annals of the 776 Botanical Garden issouri M Vt 2661 “18 19 ni) 8с+1-65 NON “E тд әгәорцои#ерү “т 9miodui сина шшхерү Я vr 989311 £661 “Te 19 nif) 8271:6% NON ‘ze тд eveoeroude]y (2907 зр ploqets) pyofioyvs юоиЎюрү VP 0661 “Te 19 Влодизјос) 8Z+1-1 UMOUXUN овоовцопввуј "Т ођАцаолорш pnQouSpy Я ‘Vr SSOZTT 2661 “Te 19 nif) 8271:62 ПЭМ ‘#2 тд avaoeroude]y "99NZ X p[oqars ропәјойХу vijousoyy 8 ‘VP 0661 ы e 19 319quajos) ТА АЫ! uMouxun авоовцопввуј 1 piafidyny иолригро!17] ar PSOZTT £661 “Te 19 nif) 8z2v*LLZ NON ‘82 тд evooeroude]y 3188 (-|surag) asuaurys uo4puapoiur] ЭТТ ‘VOI SIZOTT 2661 “Te 19 nuog 82v 1:23 AV '00€ uimpjog овоое mys] q pyofidossky шта LSPSOT £661 “Te 19 Чишс PLEDIS SN B9PE-Z6 55213 APIDBIMOT uas1] "M SISUZUDIS DYIUDPIYIL) 8'у9 9среол 2661 “Te 19 qug YLETTE SA ‘6512-28 Y9D9g P 55215 APIIEIMOT штујон vuquf гугиорт/24() 01#ФІ7 2661 “Te 19 решо ОРР1-22 80721 :Є/ моћ ‘OGY аваовше8о] "Т оопиоа-хпи souyoKaig Я ‘VSI LOOFTT (peeisugo) 2661 “T 19 enysouog (27474 euou avaoeruesoT `1 DIPpudpuv vedado FOTTUI 2661 “е 19 peeisuo ОРФ 1-22 Sdn ‘1592 unm qw ovaovrueSoT чри] гтиоипад nonisopy 8cv* ULv€ Я ‘VST L6£*lI £661 “TE 19 peeisugo ‘I PE-€Ol сбе098 9g роом8иој оваоеше8о] әд (Т) гигалалофшов шттшәѕјәх) 96#+171 2661 “Te 19 peajsuo [42274 PSPTI :2] мом ‘OGY аваовшедо] "ds 22918 1 68€v11 £661 “Te 19 peoisu]o EPPI-LZ Sdn '2022 s2wasg y әғәоетие807] BND олоулриол“ vIsiaq20yIup 811 881€ 171 (25245) 2661 “Te 19 98845 9921:212 NON ‘OEZ 2894) M W аваоештт] плош] DIPU оурлотилај Я 'VOI 00211 2661 “Te 19 ueurpoy 8571:92 амі “и 22144 әвәоециеишт" "ag "Y 1150]5пор ѕәүјириштт Я ‘УОТ 619711 (esey)) 2661 “Ie 19 95845 8zb1-1€ HOIN '6098 32210224 аваовуивишт] "PILA зарто2лита 125014 023420], Я ‘vs €lLOl6W (meang) 2661 “Te зә peang OSEL-ZE АПИЗ ‘226¢2 Aa7poig әвәовүг] “Lay (СТ) риотит8 ла ю]оәрәрү Я ‘vs ©89@1Л (25245) 2661 “P 19 peaswjo 8СҰТ-ТЕ ПЭМ “ГІ 28049 4 W әғәоецгү 1 "nqaadns шт Я ‘УСТ 061Е11 (esey)) 42661 “Te 19 119q1y 0031-23 euou авовомејпациг] qxoy 0104/19 moman Я ‘VST 2761071 (estu) 42661 “1° 19 моду 8271:22 əuou әвәвотте[пчццәт Jaye AA рајплоро njnomn2uiqd 91 “Я ‘VOT (ni) 2661 “Te 19 25945) 8271:62 NON '"££0I6 MÒ әвәоешәшіәт шаеҷо гиррмој n142u127] Я ‘VP 0661 “ТЕ 19 Злодиајо<) 82Р1:1 UuMouxun овооелпе] "MIA гиромагшр Dasiag Я vr 1+9311 £661 “Te 19 ni) 8cv l'6c ПОМ ‘201 тд aeao?Ine] —'unjaq 29 SIN `L ("Т) ооу ui) 91 ‘Я ‘Wz 1592 11 (ni) £661 “Te 19 95845 8cv 1:67 NON '02016 MÒ 23BIIP|PQPZIp18T "euoeg p;pumb miqayp Я ‘VST 16228) 2661 “Te 19 peaisuo 8271'201 ОНЯ ‘2821 ounuv;) avooerureT `1 SuDI MAS umaonag Я ‘VSI *S6101 (peeisugo) qz661 “Te 19 119q[y 8271-15 ALA ‘S6E prasuno аваовпше] еле) су мгритјод pumjjomog (ueZtqor Я 'vsI LOVETT 2661 “Te 19 решо ТА АШКА N) 220088 Og teque оваовпше] eaneg y Suydg штаоилар DIAJDS Я ‘VST SOFFIT 2661 “Te 19 peeisuo 8zb1-£01 auou ovaoerureT Ug Y Dyofipunjos pa2ujunisoad Я ‘VSI 90t*tlI 2661 “Te 19 peeisuo 8zb1-€01 0100 800-68 prais avao?rureT 191184 Ajnoyoind иош2 ово Я ‘VST SOPTUI 2661 “Те 19 peeisuro 8zt L'£0I ОТОО #10-88 pveisuj) әвәэвтше7 ‘pug (^) риттит па ю1Ўә1$о$Ху g ast 2661 “Te 19 peaisuio СТА АША 0102 “vs Дрво аваоепше] "т wnaindind штшот VSI 65711 (peeisugo) 2661 “ТЕ 19 aseyo ААА əuou авооепше] "Т ргорагргџ ошоцог1о) gt £91801 (ә) 2661 “Te 19 25845 2811'62 SO ‘SEETI Ássnaig деаоерџојое] [jd ригаггририлој s140127] Я 'VII (25245) £661 “Te 19 95945 821716 HOIW 'I-I-$0-8g uosdung авоовџошелу 110 | отјогоиој омэшогу 801 00917 2661 “Te 19 ueurpoyg РТА ДЫЛА SIM “U's грдцецсјру зезовштлодоом dəng psouids 2111112420% Я yo 691111 (suyos 9 1naxunig) 2661 “TP 19 95245 LOFI-IE ПМУМ “u's syoo1g овооеоштј әдәд `4 f што штиоша 82661 “Te {чо uone әллә «92uanbag 921n0$ / 19YINO A Áqrure y sardadg 19 9seu) -S939298 ut 210314 aseq -eq `рәпицио”) "XIaNdddy LIT Appendix Volume 80, Number 3 1993 £661 “Te 19 ni) 81811 NON ‘82016 MÒ Ol ‘9 ‘v2 ZE0LLAN 1661 “I 19 sa] £81 1:67 NNOD “us 527 оваовподштјом "819g (PILA) 0927] одштјод DII (nuo?) 2661 “Te 19 9524) QZEI-LZ SIM ‘ZOOI uo?) аводе Ар "AIeH s!5u2]D7DU s1XAdO04219]] Я vel 66Sc l1 2661 '25940 Y чогу 9771:62 ПЭМ 100g чолу IPIIPUISIA р SUNG DjDUu242 DISIPA Я ‘Vr £S9cTTI 2661 “e зә ni) 87t 1:67 ПЭМ 'TFOL6 тд IPIDEIPISUÁ JA әш 07211270] Dui2uy 9T ‘A VIT vY£610'I (n1) 42661 “TP 19 мету 87t 1:67 NON '92016 MÒ avaoRo AWW 1 222fi222 DIA Я ‘v9 SGGTSOT 2661 “Te 19 Чишс TLEULTE umouyun IPIDPSNIA] epo?) DIDUNUNID DSN 8TTSOT 2661 “Te 19 Чишс FLETTE umouyun авазезпјиј иешѕәәц”) ("M[9 AA) шпоооајида әјәѕи Я ‘Vol 6S€£111 £661 “Te 19 ueurpoy 8ZP1-LZ SIM ‘OSOE SNI aeaoeduuopq "шет 42/120 Duuo 91 ‘a 'VII 226107 (31eque[o9) 42661 “T 19 мету £03 1-1 9uou 9992810] Tp DOIDcSTHON 9I ‘All (9119H) £661 “Te 13 25945 COZT-10E umouyun зеаовлој *] 0muosia sni g8'vr 8TO9cS UI 2661 “1? 19 ni) 8crl-6c MON '82006 то 9€32PIUITUO]A 35104 Сс) X» МГ P2409140 рАлрохргн Я ‘V6 99SZOW 2661 ‘18 19 Busey LOTI AWVL “vs #17704 авооеша пој “1 27]221142a озтјојд Я ‘УРТ 589111 2661 “Ie 19 pearsuo ВСРТ-Є0Т 0109 ‘9226 ffnpuao аваоецшеАпој en ^4 рој у од DISAD]A Я ‘ҰРТ 900? TT 2661 “Te 19 peasuo ВСРТ-Е0Т 9uou aeaoequeKua]y q 2201]0/1) sayjuvAuap 91 ‘9 ‘WL 79211 (ni) £661 “Te 19 95245 8ct l-6c NON '91016 MÒ дваоешлоде тој эа Cqunq) зпдој ла sn]n2207) 2661 “Te 19 ni) I92 1-906 “61€-2Z əuou ovoo2vuriadsiuo]q "ds sojadunss1) JEL (Quo) 2661 “Te 19 25245 8cv 1-23 SIM ‘#001 1105 аназнјешојвијзИиј TIEM 22077975 0122950) І DII (nuo) 2661 “Te 19 95849 82127 ‘pZ: punupg пе 9vooej?urojse[o]A "pues »ndan20gdX2 vinno as (95845) 2661 “Te 19 95249 8Zb 1-19 NON “us 25095 IPIIRIYIURIIA “xy, umaojftaand. uma3424 9 ys (95245) 2661 “Te 19 95940 8z + 11€ NON “51 2804) f W әвәэещіивјәуү “XY, »15ofimua1 DI] Я ‘Ws (sseq)) 2661 “Te 19 9594) 8cvt l'66 ПЭМ #24 2804) M W эвээвіцивјәуү Келе) “Y CT) unam] um avuoy;) (eseu)) 2661 “Te 19 95945) QZPI-IE ПЭМ SOLI 2804) M W IPIDPIYIURIIA `1 овоитлој s1417 £STSO'T 2661 “Te 19 Чишс АЧЫУ “68-82 55911 999991U€IPJA "фра они (Тргу) »a4ndand ројџоозирлору g ‘v9 0T0S01 (meang) £661 “Te 19 peang ISET GE JAWS ‘62662 Керол 3BIDEJUBIEN SION "Y D4N2UOINI] DIUDAIDY Я ‘v9 LA AUN 2661 “P 19 yus TLEETE TAS 12:28 98 79195 aD аваоешелвИи "qoe “4 ^f 14әшә$әо] DIYID]DO 801 196101 (95840) 42661 “Ie 19 моечју 82+v 1:23 OSI “us Japua y IPI)PATEA ?a1107) хә [os (Т) vaundod visadsoy y Я ‘Vol 981211 (95245) 2661 “Te 19 9584) BZ Uc Эс “u's japua y аваовлјејј [Png ^4 пиозилдол 1md&sso») а'УП (95245) £661 “Te 19 25240 8871658 — HOINW ‘2591 иомгрир "M M arave dew ueg 010/170] siia Я vII (95240) £661 “TP 19 25249 8ZFI-IE HOIN '$996 unumojd4 аваое Bdew рәм Csu) nauunjs титрог ој Я VII (asey)) 2661 “Te 19 25949 8ZP1-1£ HOIA “us uosjopp овас А јер peg s2043 рушту) 9I “Я 'VII (aseyd) 2661 “Te 19 25849 8cv lU T€ HOIN Z09£] uosi29pup "M “Af aeaoerqsid egg poy) Dsafionu vyand HOIN 91 “Я VII 268101 (esey)) 42661 “Ie 19 мечу 8cv 1:63 08918 9 `1], ppyonvy aeaoerqzid[e рү puny (7T) »ofisspnao Dunuosidg VII (240) 2661 “Te 19 25940 87165 НОИ '66622 9 dosp ртэл arave ide оа (APD) ротигшар DISOYIUNG а'у11 (95245) 2661 “Te 19 95945 8Z+ 1-67 ON “us 12079p]09) овоое у Та је ssnf ^y sninppipu гпалооормиор VP LOOZIT £661 “P 19 nif) СТА 1:67 ПЭМ '9£016 MÒ аваоепопвв w [pty sisuasodnsuis vundo] Я ‘VP 999311 2661 “Te 19 ni) 87t l-6c6 NON '£c9£T чоѕгәриү “Y f оваоепопавИ "THS "Y 27000 рошпојој Я ‘VP 6592 171 2661 “Te 19 ni) 82%1-65 ПЭМ ‘22 MÒ әвәотцои8еу “Buaidg (moq) oZ vyəyny Я ‘VP 8592 171 2661 “Te 19 п) 83+ 1:67 NON ‘925 тд оваотцопзеуј “AIO (18шәң) гитрлој 0172]2u0] 5266 1 “ye uois juoueio 3118 1911] ,9»uanbog 321n0S / 1IYINOA Аше saroadg 19 9524) -S3998 ut әп! əseq -eq 'penuguo?) = *XIGNadd VY c Ф E © (©) 75 © © Ф £o tU O = 5 23 Ф с 0 С = < > со ~ ~ 91 ‘a ‘WL 186101 (ni) 42661 “Te 19 модту 8zb 1-62 NON '22016 MÒ аваовлаледеа "Т SISUSPDVUDI DILDUMÍNVOS Ol ‘a ‘WL #92801 (59T) 2661 “Te 19 aseyo е811:62 NNOD "ws 527 зваовлоледе а "Т гротиамо saandng as 2£916IN (meang) 2661 “Te 19 peang YA АЧ YON '10202661 ¡Pan әрәзвиврив,{ 9100]N 79 “ISP 212491224 snuppunq Я ‘Ус̧ (95245) 2661 “Te 19 әѕеЧ”) 8271:62 ПЭМ 61 2894) M W IBIJPUPPUB yoipnes) suapunos DIJIUIA24] 91 “уст ‘a8 L89£l1 (чогу) £661 “Те 19 25940 8271-62 ПЭМ “SITZ чолу әвәэе1иоәв д 1 отојтиза отиоәр, AN 8£6101 (Gwed ә 29114) qz661 “TP 19 моду 8c* 1-97 амі “из 29d овооерпех() boe nuappip тох) 801 66911 £661 “ued x Vud 8c* 1-97 QNI “vs зә әғәоерцехоу ds siumyooasdA y оп 26911 £661 upd x Vud Бада ам “0 29144 әғәоерцех) "] DJOQUIDIDI DOYLLIA d vs (esey)) 2661 “Те 19 pearsuro 8zb1-1 Y 2ZPE8 2504) И IRIPPIYOL() [PUT] штароха wmpu) 9 ys (esey)) 2661 “Te 19 aseyo 8zp1-1€ ANVD ‘0165 stuaway) ovaoeprqoaQ) aumig D170/1110492 DIpaimnay) xa] 85 (12ә4[ұ) 2661 “Te 19 25240 52167 ПЭМ "ws 1991р әғәэерҷэі Әле] eT ш 'хој шпирәаӢол н JIT ‘OI 8861 *uueuri1oH Y јап LOTLI-I umouyun аваовлвецу Чигу 010]ә D42170u9() 122011 2661 “Te 19 nuo?) BZP 1-27 OW '9022 411uaz ју 1u1p107 аваовлвец(у Bley (Т) риртапгәа оућарпу DII ‘VOT 322011 £661 “Ie зә nuo) 8zb1:23 SIM ‘OTOS ош о аваовлвеи() uəaey (puny) saprojdad әттрпт HOIN 9^o[peo1g 911 “VOI 612011 2661 “Te 19 nuo?) ТАДА “9Z9S иозагрир ју uosiapuy әвәэв1#еигу x ueAey 'uueunq DIYIDQUISIL n1izado'] DIT “VOT 223011 2661 “Te 19 nuo?) 8cv 1:23 OW “ZSETT 0u240Jq овазелвец() "да sunsaja опо VOI OZZOTT 2661 “TE зә nuo?) ТАДА OW ‘819% 70 12 Anag оваовлВец() BIOO “AM "f S2pi04pun14X2 vrsyony 911 ‘VOT LIZOTT £661 “Te 19 nuo?) 07101 SIM ‘2105 0ш) оваоелаеи() "т umgpofusngup wmqondy ЭТТ “VOT 968011 (пио) 2661 “P 19 peeisur КАДЫ AVG '92Р2 92191105 оваоелвец) Келе) “Y DUDNUDX D14407) 911 ‘VOT 9IZOTT 2661 “Te 19 nuog 87112 SIM ‘2501 qnug овазелвец) `1 ошто DIDIL1) Я ‘vst 989111 2661 “12 19 peoisuro 83 y UL £0I ОТОО “us poaiswo эвәэғеә[() "Т 2108јпа umajsnzrT (ueZrqory Я ‘УСТ 676101 (peeisugo) 42661 “Te 19 119q1y 873v 1:23 П) 657058 Og Pene зваовзј) pur] wnwissiapns шпитш5рј Vcl ‘96 SOCTUI 2661 '59106 y че8лој LOTLTIE TU ‘669Z 1u24321] эвәэвэе[) Jour?) "4 ^f 142424428 руәоцәс̧ a‘vit (esey)) 2661 “Te 19 2524) 8cv* 1:67 NON ‘622 5940 M W звзовичоО Od элоупүпш uo) Я Vel 873111 £661 “T 19 Buery LOTLTE ws 'qry ‘рем “SN avaoesskn] еер 212220 ред Я ‘YZI £ccITI £661 “Te 19 Fuerx LOTTE 29031 V REN ‘5 ƏLIILSSÁN meg »72452njoau? тр1ао( Я vcl 2861071 (8ивгу) qz661 ' Ls зә 119q[y LOTTE 081-2 'q1y Suiq&ng avaoesshy] auoe(q njDurumon рогугојашо) Я “vb 9202 И 1661 “Te 19 вәт 2811'626 NNOD “ч 527 авоововц ша 'q4() PUDIZNII 701101214 2661 “Te 19 ni) 871-1811 NON ‘62016 MÒ g ‘VP сео И 1661 "је 19 5271 сат 1:67 NNOO “us 597] аваовавц шАМ иону DJ]D40po papydwd AJ a ‘Y YEOLLN 1661 ете 19 5971 281 1'66 NNOO * Ж: s271 аваововцашАм pueing ртрбомра apydnyy Я ‘ҰР Тео И 1661 “Ie 19 sa] £811:6% NNOD “us 527 aeaoPaeqdur&N "qsqes холај a]p an] g'v6 SOSZOW 2661 “Te 19 my LOFI-I AWYL ‘2691 Том аваоеш ом "Т Pdo! гтд ва ју “¡sua vs 020507 (meang) £661 ' T? 19 peang 6©1`©© sA “us ио$1919,{] овзовицом t шет) — (Dausponvag) sci а ‘VS (9311824) 2661 ' [? 19 ese) GEETZE uMouxun овзовицом uos}? M S (91 q S) ] PUNON Я ‘V6 9£6107 (25240) 42661 “Te 19 мету 8zv1-1€ ПЭМ ‘SHI 25804) M W avaoeqiuadaw oourlg M , sayruaday VSI S£610'I (u91peH) 32661 “Te 19 119q]y ТА 1:67 Sd '10252-95252 “May “OGY 2BIDBTUOS[AN| s39N (WET) SUIISIUDI ртиоѕәду $€661 “Ie fuots Juoneyo əma *әәиәпһәс̧ аолпов / лоцопод Ayrure 4 вәтдәйс 19 aseyy) -SIDIR | ut 210314 aseq 1: | “panuguo') "XIdN3ddVy Appendix 779 Volume 80, Number 3 1993 Я ‘V6 ZOLLLW 7661 “Te 19 IseuuRIy 8zvUlI umouyun аваоеповАјод 1m40j]n2 x шпоцм Я VII SS6107 (5240) 2661 “T 19 peaisugo 8ct 1-0€ HOIN “и uosjopp aeaov[ea [og exerg ^4 `$ CT) DYOfissaaip ооррмпооб 91 ‘a УП Cré6l0T (5940) 42661 “P 19 мочју ТАДА ПЭМ “SST 9995) И aeaoe[ea [oq s] PIDIN оро8 род a ‘vel 189111 2661 ‘18 19 peaswjo 8аР1-201 зиоп аваоешошајод п supidas штиошәјод Я ‘ve SEI8SX 2661 ' е 19 1anbsnog 8Zv1-1 с; 1'82 99 4121129 ОП оазе леоород Зр 201112042 спалозород (meang) 2661 “P 19 peang 1S£ 1-67 NIN “vs prang зеаовод оцчоин DUDXD] p1u0217 а ‘v9 1861 “Te 19 тцовлој, ТАДЫ! UMOUXUN аваовод "ү unansap шпотај v9 0661 “IE 19 Kajqooq ТАДЫ! чмоихип aPIDROJ “reg (Бер) supisip отешота v9 0661 “T 19 Kojqooq 8ct UI UMOUXUN aea»Pog ug ^W CT) unong? wuniosiuuaq g'v9 10с00а 1861 “челн ў PMPZIUSIN 8z* L1 umouyun әвәовод "ү Paros 02440) Я ‘v9 0661 “Ie 19 uospnH ТАДЫ! umouyun зезовод әҹеүд “LOS руојттиој auyovinayy v9 0£900X #861 “P 19 DSMEINZ 8Zb1:1 UMOUXUN зезовод “q espana шпорлон Я ‘v9 0661 “T 19 Азјдаод 8zZ 1-1 umouyun әтәзво,] ea 51221795 snsyoua’) go 92916 (meand) 2661 “Te 19 peang SSET-ZE НЭП ‘919-79 әрис̧ әеәоеодј yosney (moq) хојаути psnquog а vo (џоәт N 'волво)) 2661 "T? 19 95840 8Z+1-1 umouyun аеаовод “1 paps ригак Я ‘v9 1861 “Te 19 YLL 8zv LI umouyun зезовод ssiog »ssp42 зао тор а'уб 10222W 7661 ‘18 19 tseuuei?) ТАДЫ! имоиҳип 9vo2eurZequin|d quny y sisuadpo одродштја 91 ‘a ‘VL ev6101 (mÒ) 42661 “P 19 119q1y 8с%1-6 пэм ‘00064 тд avooeueje|q "1 SI/DIUAPI990 snupip]d Я vrl ZOZIIT 2661 “501105 ® ue210]q LOPI-TE усни “us S1392s21y оваоелодвој и 4 [s914 `7) хә "мод tunruodoÍ wniodsomg at 0992 11 2661 “Te 19 ni) 81:25 ПЭМ '8Р016 MÒ avaot1adiq "1 әрә әй Я ‘Vt 1992 l1 £661 “Te 19 ni) 82v LZ ПЭМ '2Р016 тд аваоелоф а ds отшолодга g've 6S9£9X (ssnens) £661 “18 19 25940 8cvUl auou aeaovutd "лес (јен) yA ydosazay vans] g've L£66SX 0661 “Te 19 surydiy 8cTUTI umouyun әвәэешд ooue1j (алу) пзаттигш р9пѕјорпәѕа a've 7828SX (ssne.1s) £661 “те 12 aseyy РОРТ-Т 9uou эвәэешд lopqay $1]1фршо хтлрторпәѕ 8 “VE РЕ185Х 2661 ' Te 19 1enbsnog 8gcvUI auou эвәеш чой а оплррл snuid Я ‘ve I£I8SX 2661 “Te 19 1onbsnog 8cv UI 7€/9€ S9NSUIL) 189.10 Y SUJ әвәәеш TDW 1727148. smg VE 099£9X (ssneng) £661 “је 19 9584) [74 2 64 | əuou эвәешд 91911187) [& Виод) SISUIYOTIS DIM] a ‘We 90€ I8SX (ssne.ng) 2661 5 [e 19 95849 ТА АЫ! 3u0u овозешШа "јави suadund DIM] a ‘YVE 29929%Х (ssne.1g) £661 Ё [e 19 9seu) 8gcv Ul 3uou оваоеша ипм $1]1Djuap1220 21107] Я ve 79929%Х (ssnenc) 2661 “Te 19 25840 КАЛДЫ! ZSzUse 99 {apg ON авоовша USSISg (шедлод) риртртарр v11229]212M Я “ve 29929Х (ssnenc) £661 “I 19 25940 КАДЫ! 92/18 вођено) 159104 SUJ аваоеша uopnoT ("qxoy) олорогр =т1рәг) а VE (ssne11g) €661 “ye 19 9seu) РОРТ-Т ct/0€ s3naua^) 18310 4 “suJ эвээеи Хет] "V оэјидрш saiqp Я ‘V6 0LSZOW 7661 “I 19 Зшәу LOVI-I апоџ овооеоовјоЈ АЦА ‘ylueg штјојштоу nu2adsousa1s 8'v6 69SZ9W Z661 "|? 19 3n19y LOPI-T umouxun оваоеоовјо yd y Sunamy pui d ‘V6 LOSZOW 2661 “TP 19 Вон LOPLI AWYL 1591 92M аваовоовјомца "Т DUDILIMD D220]0144d Я ‘v9 (naeg x шецело) 2661 “TP 19 25840 0611'07 LAL 98I£FC овазелрАапца ово W squeg unsouidnup] wnipínyd 8cvt lU v6cl Я 'vsI 80TTlI 2661 “Ie 19 peaisur() "PLC USOI зпоџ эвәовцерәд UI шпэрит шпшр52$ Я ‘VSI 96107 (палрон) 42661 “1° 19 мочју 8ct 1:63 ANI “258 чәрлорү звзоецерза ӘЧ, 2220и025п0] рартогодола 8 ‘VSI £26101 (пазран) 42661 "T? 19 мечту 8zb 1:67 (INI ‘99°06 чәргоуү аваовцерга "reg 1727p1uD42. штАцаотра он Я ' УП 06107 (чогу) 42661 “TP 19 119q[v 8crU9c NON '000£ чолу 9vaoPIo[rsseq `1 suuppndupiponb vioyissog 82661 ‘Te {чо fuonepo олтуелој] кәзэчәпһәс 29108 / јацопод Ае вагоо ас 19 asey’) -8922P ut 310314 aseq -edq 'penuguo)) XIdN3ddV Annals of the 780 Missouri Botanical Garden с26107 0661 “e 19 500$ ФОРТ ТЕ SM “РӨТ 51106 > 5706 әвәовЯеіхес̧ ѕе[апо( Dwyun42nui D4212n23]H ЭТТ ‘VOT v81111 0661 “TE 19 sn[os LOFI-IE SM “62bZ SNOS X 506 OS “ABD руојуэиоѕ D02u044 081111 £661 “sujos 2 ued10jq LOFI-IE SM '2802 1105 2р 51106 оваовавлихеб 5вод (110) n7Dj]ad олошло(ј Я ‘УВ SLTITI £661 'sujog 29 ued10jq LOPT-1€ ЭЯП ‘TOIL 1129109) TS Kireg ођодритол viui&og v8 211 0661 “Te 19 suog LOFI-IE SM "Z4FZ 512105 X 51105 әвәэвЯеухес̧ ром (421) тепђој 2qnisp a8 ‘WP P6crIT (peaisugo) £661 “Te 19 eseu) ТА АША 0700 *'900-98 poajsun() IBIL MMES “1 snnulao итии; Я ‘Vr 291801 (se]) 2661 “Te 19 25249 £811:67 NNOD “us 527 agave mme "qunq] »7»p402 viu&njmog] Я vel ZS610T (чогу) 42661 “Te 19 119q[y 8271-62 ПЭМ ‘FEI 2504) И аваовшаовллес `1 pany тигорллоб det Ч САТИ! (swem) 42661 “ye 19 119Q1y ISEI-LS auou оваовшзаовјлес `чшәң зирјпи рлоца штуогн uos A 9I ‘all (nig) 2661 “ye ja 2584) 8ct l-6c NON '£F0IÓ6 то аваовхоропо Zeg » 13Pyay ( АЦО) Djpnaumo охоројиг8 106 8 vel 26101 (asey)) 42661 “Te 19 модту 8271:62 ПЭМ 621 5990 M W avaoviodeg uaKoy (7T) n70dvz олоупиоју 8 Vel 10911 2661 '2584) x uory 85+1-62 ПЭМ “231 5990 M И оваоеодес q гшлоћато umjjkydos&ay;) Я ‘Vol 81217 (25240) 2661 “Te 19 25240 8271-25 ПЭМ ‘ZIZ 28049 M W авоовршдес жүрем зертотрлоорир опират?) үст ‘96 961111 2661 “suJos 29 ue210Jq LObT-1£ DIS ‘TELZ 1u21321N оваовјеу ес 'pneis 3 “ISYIOH D70]022u7] SUÁS() 91 ‘a ‘v2 5992 11 (n) 2661 “Te 19 25240 821-6 NON ‘2016 MÒ әвәэедес̧ ‘ds 21996 8 ‘VOI (95245) 2661 “Te 19 25249 8271:02 ПЭМ ‘ZTT 28049 M W ava»einy ‘yey CT) оитоји) sna2u04 *£0S01 (meang) 2661 “Te 19 peang IS£ l-0z YON ‘10602661 ian avao?osny qouaojy nsoum2n4 IDUD(J a ‘vs (95845) 2661 “Ie 19 25945) 8с%1-6 ПЭМ TI 28049 4 W avaovosny Ҷэчәор{ 250ш2201 грип] 42434 Я ‘VSI 1£6£11 2661 “Te 19 s[oeqorjy 'I£C L'6vZ ‘9-25 HOIN ‘#16 чәѕирр aeaoetqny s1egaq ("5910 4) n70]022uv] sniuaq РбЕРТЛ 2661 “Te 19 реозеш о L£V LZ сап '860£ 1211218 `g aeaovtqny 'ЧәиҢ ^y (CT) 2970 »22020nj7) 91 ‘a ‘VII 903111 2661 'su[os 29 че8лој LOFI-TE SA 'O£Iz uvS10py аеооввом 194—7 (10129) 12упоцира vanndg 91 ‘a ‘VII L*6101 (sujos 2p uedI0W) 42661 “Te 19 медту LOUIE SA “S2PZ SUYOS o snos APIDESOY "Т оиоти18 па snunig Я VII 00111 2661 ‘105 29 ues10jy LOPE SA ‘TEIZ uvg1ojy әвәзввоу sseaqp 1495D1f piunoyg 91 “Я “VII 126101 (suos 9 ueB0W) 42661 “Te 19 моду LObT-1£ SA ‘ZEIZ чоору авооввој 'лоб хә "ея әѕиәортуә uma») 91 'H VII 681211 (asey)) £661 “Te 19 25249 8zb1-0€ ПЭМ ‘OOT 28949 41 W ovaoeuurety "Т 512110102 snuumuy a ‘Y9 SLOZTT (ue1peH) 2661 “Te 19 peang 8zb 1:67 ПЭМ ‘60 28949 M W APADPUONSIY ‘ds 028913 Я ‘Vol 9SLS6W 2661 “IE 19 ueurpoy 8cv 2c (NI “vs 22144 әвәэврәѕәу "Т оф орәѕәу а ‘YO AD) €661 “је jo peang StrI-l uMouxun аваовоје еј “WÁ пигјјр 5742108 916 91 “Я ‘VL 6992 11 (ni) 2661 “Te 19 25845 8271-62 NON '02016 тд aease[nounuey neyse punssinduns vzrys0yjuvy 91 'H ‘Wl 991801 (seT) 2661 “Te 19 25845 2811'62 NNOD “u's 527 avaoe[nounuey хеҷ) sa 4ydoyor17 snjnoununy 918 ‘WL I£*201 2661 “I 19 peaisugo 8c*U2Z auou avaorjnounuey "Т suuisnyod vii?) Я vel 5592 11 2661 ‘seyg ў uory 8271-62 NON '9061 чолу овазвјојАа Тођодритол пјолАа 911 €ZZOIT £661 “Te 19 nuo) 8271:22 SIM ‘1001 1007) оваовопта Тштвио“ рэшта 91 ‘a ‘Уг 061111 £661 ‘105 29 цчеЗзору 10® 1-1 пемең ‘9g 'doip ‘ем оваовајола 19 y snua42ut оплодшру Я vel СРЕВВИ (31199) 2661 “P 19 25945 82711 ПИУІ ‘Sor Fay aeaov[nurug "Т $19u2a40 sypodoup a ‘Ус 291801 (se]) 2661 “Te 19 aseyo 2811:62 NNOD '£8r $277 аваоепојадошто4 хоп, стјоју а шр uojasouvjog 8'v6 89SZ9W 2661 “P 19 Яшә LOT UI ПИУІ “9591 Fay әвәэвое[тц1од "3ooH ргоуірирі9 озоттаоа 121-811 a8 ‘v9 (31215) 2661 “e 13 peang ‘SSI 1-18 umouyun әвәовттәрә1цо,{ 189214 о DIDI DS DiapajUuod 82661 “ЛЕ fuots uone əmrə ,20uenbag 99INOS / тоцопод Áqrure y вата С 19 ase) -5аоов ut әл] aseq -erq 'penuguo) 'XIaN3ddV 781 Appendix Volume 80, Number 3 1993 Я “vol 85256 2661 “Te 19 чешром 83b1:23 SIM ‘TEST mug P ynus cat O лед 39 ту ојприга рило got ([o13€N) 2661 “Te 19 95245 83v 1:23 SIM ‘U's иозтәа]у о JT. РЕНО TOME Я ‘Vel 809211 £661 ‘әѕецо) Y чогу 83v 1-66 ПОМ ‘912 55990) M W avaoeysesydoay J, 2a punis4222 оћатј) Я ‘vel 309311 (ni) £661 '25940 x чогу 8cv 1:63 эчеп агана, q uodo! ютз]рәшэгу 91 ‘a ‘V2 8992 l1 (MO) 2661 “Te 19 95849 8ecv Uc NON '60006 n) аваоглиаовлај АЦО S?suauis uo41u3204]2] g ve 2661 ‘LAUS x suJog ‘пос S9e rz AGI ‘+80£ pjafsunig зезотрохе | Чон (7T) шпур штрохој zoqq»ng [ Я ve (ssne11S) 2661 “Je 19 9584) ТАДЫ! зиоџ аваотрохе | (puq) шпәзир%8 иолризртопђос '29n7 ge (89114) £661 “P 19 aseyo 151-26 98 Kepoueg ON TEOOBIPORE T, X рода (qunq]) 270771911429 s&ndoppiog Susy) a ‘VE (ssnexiS) £661 “ТЕ 19 95240 SOET-Z auou аезотрохе J, ОА Y NY s2p109041501d412 monbasvjapy ЧЕ (99114) £661 “P 19 25240 Is£U26 ОЯ 41199 ON ool BIDOXS T, uoq ‘q saprouidvjas SIXDIOLYI/ ac (ud) £661 “P 19 95845 terz (NI “vs 22124 IPIIPXR | треш х snxp[ Я ys (95245) 2661 “1° 19 25249 87119 ПЭМ ‘SZI әѕру) M W зеаовојеј aIpuy мамјируз D22D[ Я vel v29Z 11 2661 '9seq) Y цогу 8Z + 1:67 NON ‘005 чолу avaovo2o[durkg bi Divpnoiund гпоојашАс Я ‘vel £29c 11 2661 ‘sey x UOIN 8Z + 1:67 NON ‘4006 чолу аваововла1С WIR] риромошр XDIAIC а ‘v9 19+5017 £661 ‘18 19 Ҷишс PLETE SN '691£-I6 SS24« авоовтијалс yoy y [928g әрјоли 1127119116 69€ USTEI "922 U 2201 "50102 ‘+97 Я ‘Y9 (узе) £661 “Te 13 peang “191 '001:82 *12-1 UMOUXUN авзовтијалс Pug ^q ^f SISUIMDISDIÍDPDU DIDUIADY 6STS01 2661 "I? 19 Чишс PLET-LE SN ‘POSE-Z6 55913 BBDORIZING ‘OUI, 74 7f SISUIIDISDÍIDPDU DIDUIADY buy а ‘v9 87501 2661 “е 19 yug PLETE TAS ‘6602-98 81у авоов тело (чо!) əsuəuuvín? штшлоагоуриоца gol (олем) £661 “T 19 25245 SZPI-LZ SIM “u's uosiaa]p аваовуполојс “1 ODIDI ошолдозуј go ££916IN (meand) £661 “Te 19 peang ISE 1-03 ПЭМ “253 25949 M И avaovrued eds uny шпиромошр штир814с Я ‘УСІ 29861 T 19 YUPIV voy TI HMOUTUE экии wA (у `Хооң) орту vund Ol ‘Я ‘VSI 9861 “P 19 ur] ТАДЫ! umouyun авоовивјос J UNIDO, DUDIJOMN Я ‘VSI СОРРТЛ 2661 “Te 19 peaisugo 8271:201 auou аваовивјос [ГИ] штиојпова иоолелодозат g ‘vs (95245) 2661 "Te 19 95240 83v1-0€ ПЭМ ‘ZOT 2504) fl W а Lan JOYE AA DIND]A XD NUS Я “vol (95245) 2661 л 1925909) 83v1-T€ ПЭМ “931 2504) “41 W eevee Tessie Shure Я ‘УСТ 206101 (prorsuno) 42661 “Te 19 weqiv 8571:25 ин avaoeure[nudoos 1 2emd.md 10191] Я ‘УСТ 889111 2661 “TE 19 peaisuo 8cV U£0I ¿“us simpdunqop aeaoeue[nudo125 "т snfow nu Lu Я ‘VP S99z 11 2661 “Te 19 nif) 8Zb 1:67 NON ‘EZ тд әғәэвіриеѕіцәс̧ UOS[IAA Y 19pyay рлаџјиригудх олрирупос dst 8031 UI £661 ‘05 25 UBB IO LOVUIt пма ‘629-01 444 UPA OS "qunq s7suadpo nin д ОТ ‘Я ‘V8 56101 £661 sapos x че8лојј LOPUTE SA ‘ESZZ 511105 op 51106 9vo2vdv1jixec ооң n1ofisayni отолћхрс Я “ve 261111 0661 “Ie 19 SHIOS LOPE SM ‘2622 чәрХон `$ t AOS POPO e tl Я vit 6£6101 0661 “TE 19 зпјоб LOVUTE SM "ws 51106 29 sog avaovv1jxes squeg DIDILQUY DISSDULD Я vit ¿61111 £661 'snjos 9 URB10JA LOFT TE NIN “u's sowoyy AS noq CUNA) unio ny iods uoporadonda7] 161111 £661 ‘ос 9 ue210Jy LOY UTE SM ‘2622 sujog FY 51195 әгәэв#т1їхєС зя Y (чоп са) 2170/1024 риәцігоза 7 82661 “Te {чо |чоценә əman жӘәиәпһәс̧ 921n0S / J9YINO A Хш 4 вето С 19 95249 -SIDIB ut 210314 aseq -eq 'penuguo)) "XIdN3dd Y Annals of the 782 Botanical Garden issouri M OSTS01 £661 “Te зә Ҷишс PLEDIS £€9-06 JASN awaov1aqidur7 'qxoy шпару штуохрон а'у9 664501 £661 “Te 19 Ҷишс PLEDIS TAS ‘1912-28 uovog X ssary аваов од иу штацон 18142 9907) ѕәл10 | -2әпЬѕе A “JA Я ‘ve £897 11 (5245 Y SMH) £661 “Te 19 25245 таи SYTA "ws umuzmyos эвәэетше7 ју səy “(| `[ ‘ѕәрілод ѕтшгәит DruD7 ga ‘ve (25245 Y SMH) 2661 “Te 19 25240 ТАДЫ СУЛА "ws ирш2)туәс̧ овозетше7 "ред (ezuny) sndoi42 viuasunig VE (25245 Y SMH) 2661 “Te 19 95245 82112 Сула “u's UDUZIMYIS овооетше7 ‘эа (by) ошозојоз sn2X20421y uosuyof "S "y 7T VE (95245 » STH) £66I “Te 19 eseu 8Scvr UU It SYTA Uus UDUZINYIS эвәэеше7 (епу; `4) ѕтипшшоэ DIMDZOLIDY/ VE (seq) 29 SMH) 2661 "Te 19 28240 82v l-1€ SV14 "ws ирш2)туцәс̧ әғәэетшеу [odoy 1adoy vbnuozopid>7 Я ‘ve 929311 (asey) Y SMH) 2661 “Te 12 95245 82112 Сула “u's UDUZINYIS эвәәрІше7 Jakq ^V ^M стириглр s0110]md22u3 g've (seq) 29 SMH) £661 “Te 19 25240 82v l-1€ SYTA "ws иршт)тцэс̧ аваоепше7 1ahq vsopnuids uooiq 9 ‘Ve (aseyD 29 SMH) 2661 “Te 19 aseyo 8c*l-1€ umouyun аваоепшеу ANS а 10421521 ту) VE (aseyo) » SIH) 2661 T [e 19 9seu) BZ? I I£ SV TA ме UDUZINYIS 9?932PIUIE7 [Puc A H pupyanbiu DIWDZOID1II") Я ‘Ve 129311 (seq) 9 SIHH) 2661 “Te 19 25245 82v l-1€ SYTA "ws ирш2)тцэс̧ аваоетше7 quiey:) (Ing ^ AA) 020774295 v1u2mog Я ‘Ус (5245) £661 “Te 19 peang 8zv l'6z ПЭМ “261 2894) M W aeaovaoq oque x 19 "Y 51722804 ргоџлоџлиру gt LS6101 (n) 42661 “Te 19 моду 82%1-25 MSN ‘OPSS PPOM NF аваовлани дү ‘Od орти тширшѕт 8'vt S06101 (MÒ) 42661 “Te 19 мечту 82+1-62 ПЭМ ‘91006 MÒ әвәэвләцш дү 1918104 7D) Y ^M `[ меплт sKunaq gt ££92 17 £661 “Te 19 ni) 8z+1:67 ПЭМ ‘2006 MÒ әвәовләти! дү "ds unjonjag Я ‘VE (auma) 2661 “TE 19 25240 8211 (apisiaary eruop[e;) (1) euou әтәоетоѕимјә дү «ooH szmqpanu ongosnajo y ЭТТ ‘VOI (esey)) 2661 "I? 19 peaisuro ватре HƏIN '099£1 vossapup ^M M 9va2eISÁq90 A ‘ds рәроп() 821 ‘V6 096101 (5240) qz661 “Te 19 моду 82+1-22 ПЭМ ‘922 2804) M W LILNA "хэти s;pansap sinj4 "jsuqof үст ‘g6 661111 £661 'snjos 9 uea10jq LOFI-IE TII 2203 ru2421N] 9E92POSIA ‘OW (Jey) штитогәѕ uo1puappaouq а ‘VII 7291 UI 2661 "је 13 реәіѕшү() 8ZřI-€01 эиои оваовјог) “PITA 1110105 71014 Я ‘VSI CIPFUI £661 “Te 19 peaisuo 8zb1:£01 auou әгәзвиәфләд 1 S?suatpuog DUI д (иеӣтҷэтуҷ Я ‘VSI 689111 2661 “Те 19 peaisuno 82v L'£01 N) 017078 98 PU аваоеподаод зчәд хә WOH ѕирг9о:/ штаригролој) d ‘VSI £6£*l1 £661 “Te 19 peaisugo 8Zz* £01 ОТОО '210:88 Paro авооеподлод 420% ^M (опот) ршозоуэтр ралроцјо;) Я ‘Ус (AID) £661 “Te 19 meang 8cv 1-2% umouyun аеаовтоод "ds D20124 Я ‘УРТ РЕбЕТЛ (SPLN) 2661 “TE 19 pearsuno ЖАДЫ! HOIW '606 vəsuvf әвәоеиешәјед 7] =70и12уо vunn] 91 ‘All (ni) £661 “Te 19 25240 8c* l2 ПЭМ `6#016 mo aLLeN yorpneg (7T) рәати pmuawyəog от ‘All (nid) £661 “Te 19 25245 BZFI-LZ ПЭМ “SEE 2504) M W aeaovurn) SUN pyjup4onu DUAL] 91 ‘a ‘VII 8292 11 (nid) £661 “Te 19 25245 82%1-25 NON ‘200064 то әтәэещ piouqos `Y 7D) srsuaupuun& 5112) V9vS0T £661 “Te 19 Чишс PLEDIS SN '0212-06 55213 aeaoeudá |, “1 0130/1107 очах, Я 'у9 РЕ9ТбИ (meand) 2661 “Te 19 meang 8zv1:Z£ ANIMO ара /ә1рог9 аваоеца А, Тотојиој vydAy Я ‘VOI 90217 £661 ‘Pwed 29 eoug ААЫА ПЭМ “ETT 2504) M W avaotjooedo1 |. 1 snow штрограолј 91 ‘a ‘WL 856101 (mÒ) 42661 “Te 19 мәү 8с+1-6 ПЭМ ‘92006 MÒ атаоелризроцзо], 'ооп7 Y pyoqets saprorao uo1puapot2o4] HITA g ‘VII (95245) 2661 "је 19 eseu) 8Zy 11€ '9c9ger uosiapup. Y M зеаоешо |. "ssequie?) ргали ютио811] 9 “VII Vv6101 (95245) £661 “Te 19 25245 8z+ 1:67 ПЭМ 621 2504) И аваоелрџешојј, Meg 270772901120 рогу која DU 922011 £661 “Te 19 nuog 82v Uc auou avaoedesy "] зиотви оао 52661 “Te {чо јчоцеџо опело] «ээчәпһәс 39.M0S / јоцопод Apwe 4 sei adc 19 astur) -S9928 ш әлїйї 4 aseq -eje(T грепшциој :XIaN3ddVy 783 Appendix Volume 80, Number 3 1993 ste -€€€ :6, ред og unossr 'uuy 'saouanbos 129: uo paseq sayeesdr(] jo sdiysuon [91 оцоповојАца 2661 YINTVA “A `1 Y HLINS “4 `[ ‘AVaLSW19 “9 ^H ТИ апноокоц "8011-2601 :pp uoun[oA"g '(aeaurure«)) Sosse 13 əy} guoure o»uanboes зрцоајопи (Tq2q4) asepAÁxoq.ueo jo jtunqns 3318] əy} Jo stsÁ[eue Алепоцпјолх '0661 "VIN ^d а Y 9931) "LA '9u38N3102) "JN. 7 мапа "JN. ` T "AN THSO(T "689-229 :08 "ред поя unossi[N ‘uuy ‘вер aouanbes 71991 wol} suoyeoyduy говаовлави() ш sdigsuone[a1 [2qUL '€66[ "VWSLAS "( "WM Y HOV8HOSI | "V еч TINO?) `086-8@6 :08 pue? ‘log unossi шу "1291 ouod puse[d ay} шолу sayuanbas әрцоәјәпи jo sisÁ[eue uy :squepd paas jo soneueSo[4qq '2661 "LH3H1V ^V ^A Y NVINVNVAV([ `$ "LL38HVq ^H 70 "S AVHVYUO "AM 5 “Uf C'NUVTTCOH O '9W38NTIO2) 74 LAVINA УЛ 'NNIAQ) `[ “9 ‘язауо "V а 'SWVITILAA “YS "N3SN3AS "JN ^S ‘SILTOS "S ^d. мала И 79 "9NVIX "Ac Q)'ssnvulg ^H 'S YIINGNA ^M ^9 HILING “4 7f. 7334WIAA 74 72 WIN [UM маму CM “Y “LAVO ^S “Y IN3ud3g “JA AYVIO 7а М “ТОНУ 79 CN SSSSN [CA SSTAVHOI C[OH.IVWSLIAS [у “LYVHNVI UH 7 anwyd Га f TS "ома ‘Н f ‘(NOUN 'V "M ‘AQ “IA ‘SIUH он ‘зоча VW ‘пулпа MW JA. SSIBSIN "(Q7 ‘SAT Н а NV939N а ‘dVALSWI1Q) ^) ^W “SILTOS за "A И 3SVH2) "8v82-vr81 :68 VSA греоу ‘рем ‘901g 'sjue[d pass Suoure soouonbos aguas 129: Jo ayes Атепоцпјоло ut UONBLIBA злепојхлј '2661 “IIA "V "Y Y NISAYIOG а V 'ssnvuls “HS "f ‘Landosnog "SES6 ‘FI ‘say SPOV әәүәп ‘Реде шол әѕеиә8Ахо-әѕеүАходіеә ayeydsoydsiq aso[nqu jo jtunqns 9318] әці 10] vues q2q4 әчү Jo әэиәпБәс̧ '49861 '————— Y * ЗРЕО TI SIH "py APN "ештјод шод oseuodÁxo-asve|Áxoq.e» ajeudsoudsiq aso[nqu jo jrunqns 3318] әці 10] 9098 "әді aui jo aouenbog '29961 "3WIVg "(79 NEDISIA Y 'A3N83H7) “[ сиип 'G6v I- I6TT ¿LOZ 9»uer»g ‘иоцпјолә [e1n39n.9s pue Kueso|Aud :sjue[d гпололшле о) '42661 “ISVHO “AA W m SWVITHA "d $‘ Эрод MƏN ТЕҢ pue uewdey’) "sjug[qd jo sogeurojsÁG re[noo[o]A '(5лопро) зло [^f 29 50105 “| "(qp 59105 ^S ^q ш TOT- 69€ “da “Ума 1se[dogo[qo шод o[durexo ue Чим 'uononJjsuoo2aIl эцәпә8оүАца ш еер зиз поцоцвәз1 20] Виццӯтәм з1е15-лајовлецо) ‘27661 73SVHD) ^A W YSTHSIP а 'а У ^A “Luda Ty pet) IMPI 'seuo puepuejs ay] әле surkuo1oe штиледлоц їпәрлесу [p9rue3og sas рәџи SASA Aussun ‘A tedon дол], :suapaez) oiuejog jeAoy ‘ogy ‘yeuoneu “pey tupe? “су шарлев [eoruejoq “уя їштцәлодте “qiy :pasn suoneraadqqy '(€661) ‘Je 19 95240 ш 100 pue (ansst st) uoneor[qnd paro əy} ш A[uo рарпјош se^ uoxej əy} uei ‘parsi st Jequinu əm3y ou у $ “TINA 91? ¿X,, чим әзоцу “queguas) әле т, 10 "И, а, 4q рохцола зладшти UOISSIDI Y $ ‘sasayjuared ш uaAId st aouenbas ay} paonpold oym soy ine oytoads əy} *sioyine jo siequinu 93.18] әле әләчу цощм 10} завео uj ‘uoNRoyqnd əy} Jaye sasayiualed ш роја st ошви s, uosiad sty} иәці 'лодед poi əy} jo злоцупе ay; jo auo uey, зәцјо euoeuros Aq раопрола sem aouenbes sty) јр “Paro 1510 se^ UOXB] stp 10} әэиәпһәѕ јода UR цоцм ut uoneorqng n (£661 “Te 19 95245) uey, лацјо sarpnis ш A[uo pasn әләм exe] әшоѕ 1Pyl одоџ :2661 “TE 19 әѕецг) 'sisÁ[eue uoxej-c; p ш pasn злом soouanbes ојојфшоош a1our soseo autos ш) Á[uo stsÁjeue uoxej-66t ш pasn gouanbas jo 1unoury „ 8'vil (95240) 2661 “TP 12 25940 821 ТЕ HIW "ws uoszopup Y Uf аваов||«Ч40847 "Q штоирв шпорттто g'vo S9vS01 £661 “Te 19 yang 723 SN ‘9926-16 55213 звао влод и гциолој umaurum48 1291317 g8'vo 0901 £661 “T 12 Ҷишс PLETE TAS 'f0ct8 ӘН «95 ADW аваовлод Вит7 пАрлт “ye рәјәрәту a ‘v9 8Z916W (meang) 2661 “Te 12 peang ISE ZE JAWI 'Szerz Apppag зеаовлод и [MPD Jay шпиригирло“ uns po 82661 "I? fuots јаоцеџо 310181311] „зопапђос 32 MOS / лацопод Аш y вотоо ас 19 95247) -SIDIR ut 310314 aseq -eq "penuguo?) "XIaNdddVy Missouri Botanical Garden Annals of the 784 "LBE-SLE :29 лепо ‘f ‘def Хр DSSDLI әр pue јеоцм пошшоэ ut sjrunqns 2318] oosiqny adÁ1-] pue -H Зшрооиз saus3 1291 ayi jo se»uonbos epuoo[onu гјојашо “JA ѕаоп8әу pue unn yo suiserdojAo Зиоше Аџзлолр оцапав Jo siseq Ie[noe[our 941 "2861 TAVMANAS]T, "WM Y VHVHIOQ() "A “L ТНоУНа], "ISt-6Tt T :68 "V'S'(] "PROV “PEN ‘901g '(sse1dK5 pleq) штрохрј audor e шод a»uonbos 71291 UV "2661 ‘АЗИС ‘f 7) Y SLEIOS “Yd "S d ‘SILTOS РРОР-ОРОР :28 “WSN "uuy "'sjue|d по ріеіѕпш jo Á|Áqdouour əyeəpur әпәй 71291 9u1 Jo soouoanbas зрпозрпм "£661 ""3WIVvqd "а `S Y VWSLAS [^W 'LINO)) `9 “IOYVY 7M ‘зона УН Я f 'NVWGIOM "603-103 :1ў чохе, втер aouanbas aus: — зоје Аца оА ле) ayy jo КиәЯо ца '2661 “LUVHNVIA "JN [79 NOSIM ‘A `H “H Тила "909-285 :08 "Pied лоя unosstq '"uuy "Тода 9198 puse[d әці jo so»uenboes әрцоәүәпи jo ѕәѕ4үеие onsipe[) :әэерщоиЗе ayi jo soneuado[Aud repnospoy '2661 “SMYVd “YD Y SIT На '"HSVH) ^N "IN T^ A. О "129-199 :08 “prey “log unossijy ‘uuy 'suosueduroo зопепђов "әд; шош вәүетивләсу pue деаогшвлос) әці jo ѕііцѕиоцејәг onaueo|Auq '2661 “YINTVA “Af Ņ V Y лома "222-001 :08 “preg "10g Hnossiy "uuy 'заоџепђев Joqs uo pəseq ојер пѕиәѕ aepuaisy әці jo sısÁjeue Áuoursied y '2661 "AIWIvq а [ Y 11005 "JN ^M Зил ^g ‘———— 'S96-6Vc :62 PI поя unossi ‘uuy "T2q4 pue aepueisy әці jo A[«qdouojN '2661 “YINTYA '(q [79 LLOOS "JN ^W 'STHVHOIA [^H “9 Y ‘dVaLSW10Q "S6€ ЛУМІҢ ^V Y ^A. "VAVZIHSIN "099 -1€9 :08 “prey “log Noss ‘uuy ‘gep зопопђоз T2q4 uo paseq ојеј nsuos дваоеделихеС jo глодшош Suoure sdiysuonejas опоповојкца '2661 ‘SILTOS ‘Я а 5 Ча 'муУонОј "ISZ-GTL :08 ‘P89 лоя unossty ‘uuy “UONBLIBA aouanbes Joqs шо sjqZrsu[ :9eooP1o]sy ay) Jo sdiysuon [oa [Puea] "£66[ "H3WIVvq "(pf Y NISNV[ ^M “Y “OYVZS ^ ‘AYALSWTỌ ^) ^W "LLOOS "JN ^W "(7H "ST3vHOT]A 2816 :81 OISS3TM яса LH3NHOq "(^H “g 79 "PISAOTVHOIIAN 28—18 :9 по PAWN. "Id `02814п1 jo ST 941 jo изәцеа Зшепоој опоојаовт əy} pue sasuanbas ума Əy} изомјад uonr[a110)) "X :әшопәЗ jse|do1o[Qo nunon 9861 “ONIN "(LS y AIT “AO ‘ZWD NIT '£CIOT-6TTOT :88 "V'S'(] PRIV “PEN “01g ‘sursedsoue onenbe juamue jo Ало Алепоппјоло Te[n29]lol4 ^[66T TANIM “4 0D 59 мАНУО МГ "на S3] ` Se (08 "p1e2 лоя LnossI ‘uuy 'ејер aouanbas Joqs uodn paseq exei ројвјол pue оваовршова 4 'оваовдодшт *ovoo?our] ay) jo SONPUIAISAS ‘E661 “ASVH’) “AY ben a ‘NOWAY "Svb-83t :62 "pie? поя Linosst -uuy 'оваовлајгу əy} ut uoneteA aouanbas 71291 jo suoneoydun эцәцәЗо[Ацд "2661 W3WTIVgq `0] `[ 29 STHIVHOIIN [^H 'SOVTIVAA "S "M 'NISNV( M MH s "WIN "918-808 :S93 “Way "oig ‘f ‘satoads зивја *7) pue *) рәјеәг Ајезојо шолу asejAÁxoques ajeudsoudsiq -eso[nqu jo џипдпѕ 9812] əy} 10] souas 1292 jo иоѕшеішо) "0661 'TI34LIH M "Hd Y SMANGNY ‘ү ‘ү, маоауд “Y "JN ‘S8919 `f “A 'мознзаму "V а "'NOHVIN а f "S ^9 *'Nosaag 205-905 ‘SI Ло! PAN ‘Id “1y-se3nog “шлодвошшАа e 10] asejÁxoqueo ejeudsoudsiq-c*[ asopnqu jo junqns 93.18] ay) јој зпо8 əy} jo зопопђос “0661 “SSAVHIS Н 5 29 IVSL нога ‘A 5мчаң '859-969 рре элцем "вотоод 5 pyous DW 3UIOMA] е шолу a2uonbos VNG 1se[doa0[u7) "0661 TASAYYNZ '9 29 NOSHIQNSH (T манпд И CASTINGS ^f 77) ‘99419 “LN CUSVNNVIO A A CN A "омаама10 9 '"ST- D “LT "10g 71546 'заопопђов 1294 aaneseduros uo peseq оеруу АцдоА ле“) əy} jo sdiysuonejas KieuonnjoAd '2661 9931) "LOW Y NYVaT 725 TASMAVYINZ 9 “4H О “ISVNNVIS) “£0€ 99917) ^L JA Y му “A “AN SAA "ASES Y Lava “YZL-LIL :S 10g 31546 ueqensny :soouonbos 1244 шозу SDUAPIA9 мом :oeoosmuexy огшәрцә uemensny eq) jo senmgy “2661 'OdNVNN34 `$ ^4 Y зоча "V "M LINO) ^ “TOUVY '9 “MN ‘NVNGOY 7 1 (—— '98S- 18S :08 "pled поя unossrjy ‘uuy ‘saouanbas әд uo paseq ojos пѕиәѕ ovaoesso1dn;) əy; шцим sdigsuone[o1 jo sispeue uy '2661 "NNIAQ) “f 7) m 'V `4 ‘захо "619-209 :08 "prey 10g unossijq ‘uuy ‘erep oouonbos 294 шоду pejonujsuoo зиоројАуооопош ayi јој sasayjoddy onousB3o¡4yg 72661 “Uf 'NIV3T^H `9 Y YAWWIZ "V 74 SAYS "S ^g 'HLINS “4 `[ ALAVA 73. 7] 'STHH ^9 ^H 'SS38N 7f "A “яну “Gq UN “ASVHD CA ИМ 79931) CL CW OY И Tvang jo saouanbas ума vox] роллајш sadeaun 1o(eur stay) jo иоцеэуциәр -68€ :@9 лепа) `f “def -ese|&xoqeo ojeudsoudsiq-c'q asopnqu aou jo WuNqns ase] оцу ло} опо8 әцу jo uorssa1dxo pue e»xuanbes әрцоәјәп '/861 ‘soy Sptoy 9I9[2n NI 'sjse|dogo[qo snopiupuouy»od Ay SNYJUDIDU uio1j "1291 jo oouonbos згрцозјопм "0661 Aug O "f bs – 2665 'S€ "190474 "зајој ЕТ 'sjue[d snouopa[|41020uour jo snoof "1994 эч је uonnjiusqns 9puoos[»nu Jo 5ојел элце[эу '2661 ‹ 785 Appendix Volume 80, Number 3 1993 2205 FI Ра sploy or9[onN "ада еј Долојцо ead шолј ose|Áxoqueo ajeudsoudsiq-c*| asopnqu jo junqns ase] ayi 20у 9098 oui Jo o»uenbeg '9861 "A3IWOLLOg “A Y алазину "M "d < ‘6PL-SEL :901 s?uaua?) ума ise[do1o[q» ezreur pue Ko[req иәәмјәд зоподлолр oouenbes зрцозјопи jo əmu J, "pg6[ ‘NMOYG ‘Н "V Y 9931) LN < “OLZE-ISZE "TT SAY SPOV мору VNA 1sejdoaopqo yoeurds шолј ose|Áxoqueo a1eydsoydsiq-G*]-2so¡nqu jo nunqns 33.18] Əy} 10} ouod oui Jo 3139038 әчү, “1861 "GT34LIHAA ^M "d Y A3TNOLLOg “Ay лоннза ЯҢ "5 ‘THSMVUNZ “PEL-ESL :08 РО 10g unosstq "uuy 'ејер aouanbas 129: 1017 pa119jut 5олцвјол oAnejnd pu? oje[ nsuas "T $7407) jo sdisuonv[o1 oueuedo|íquq '2661 ‘51705 `$ ^d Y NVOuOIN “Y `а “SILIOS `A “A CÒ ‘ONVIX '€1-99 “LOT пору 109 77424004 DIIYIO0UI() jo еш81ѕ zueu aurojse|d упатоцор-овепавАхо / ов] «ход eo a1eudsoudsiq 9eso[nqu ay} ut uosa əy} SASNBI jrunqns а8леј Əy} 10] 3198 Əy} ut иоцәјәр ned-9seq-9Ay V ‘8861 NNVMHHIH `9 "Y Y "d H3INIAA NOTICE THE RUPERT BARNEBY AWARD The New York Botanical Garden is pleased to announce that David Clarke, of the University of Illinois, Urbana, is the recipient of the 1992 Rupert Barneby Award. Mr. Clarke will use the award to complete a taxonomic treatment of Асаста series Gummiferae of the Caribbean, both as a part of his graduate thesis and as a contribution to the flora of the Greater Antilles. The New York Botanical Garden also invites applications for the 1993 Rupert Barneby Award. The award of $1,000 is to assist researchers plan- ning to come to the New York Botanical Garden to study the rich collection of Leguminosae. Any- one interested in applying for the award should submit their curriculum vitae and a letter describ- ing the project for which the award is sought and how the collection at NYBG will benefit their re- search. Travel to NYBG should be planned between January 1, 1994, and January 31, 1995. The letter should be addressed to Dr. Enrique Forero, Director, Institute of Systematic Botany, The New York Botanical Garden, Bronx, NY 10458-5126 USA, and received no later than December 3, 1993. Announcement of the recipient will be made by December 17. Anyone interested in making a contribution to The Rupert Barneby Fund in Le- gume Systematics, which supports this award, may send their check, payable to The New York Bo- tanical Garden, to Dr. Forero. Volume 80, Number 3, pp. 523-786 of the ANNALS OF THE MISSOURI BOTANICAL GARDEN was published on August 13, 1993. Symposium Proceedings Published in Recent Issues of the Annals of the Missouri Botanical Garden Phylogeny of Asteridae Annals of the Missouri Botanical Garden 79(2) (Spring 1992). The monophyly, sister group (or groups), and patterns of relationships of the Asteridae are discussed in the symposium “Phylogeny of Asteridae," organized by Larry Hufford of the University of Minnesota. The symposium was jointly sponsored by the American Society of Plant Taxonomists and the Botanical Society of America and was held at their annual meeting, 7 Aug. 1990, in Richmond, Virginia. 238 pages. $27.00, plus postage. The Lycopsida: A Symposium : Annals of the Missouri Botanical Garden 79(3) (Summer 1992). This symposium discusses the unique phylogenetic position and morphological distinctness of the Lycopsida; these are used as vehicles to study convergent structural evolution in vascular plants or limits on structural solutions to common problems such as light interception, propagule dispersal, and nutrient acquisition. The symposium was - organized by William A. DiMichele and Judith E. Skog and was jointly sponsored by the Paleobotanical Society of America and the American Fern Society. The symposium was held on 8 Aug. 1989 as part of the American Institute of Biological Sciences meetings held in Toronto, Canada. 289 pages. $30.00, plus postage. 6 er Knowledge Brokering: The Mechanics of Synthesis Annals of the. Missouri Botanical Garden 80(2) (Spring 1993). This symposium, the 38th Annual Systematics Symposium of the Missouri Botanical Garden (held 4—5 Oct. 1991, in St. Louis, Missouri), discusses the value and future necessity of assimilating diverse biological information into new and broader syntheses, taking advantage of computer-based data-management systems. 41 pages (entire issue 230 pages). $35.00, plus postage. 2 poo dt eid i oar жуту ED eet requiring invoices, No shi ments are made until payment is received. Mail or fax (for bankeard orders) form with Your check or money order. payable to Missouri Botanical Garden, to: it copy(ies) of Annals 792) | Department Eleven, Missouri Botanical Garden Please send P.O. Box 299 | ib dut ; | | Тош. MO 63166-0299, U.S.A. ПРЕ ПЕТРУ ir topes) of Annals 79(3) ax: 7 Ad и фт, - АНИ ИЛИ E ня | P e Send book(s) to: j| | ОР ayment enclosed. f О Send invoice ($1.00 fee will be Name added to total) О Mastercard Address. Name as it appears on card. Tux ти: "resi Postal Code Country 80(3) npud NRI тыгуу ар лыт Г» Рһопе number (days) | PRICES ARE SUBJECT To CHANGE WrrHour NOTICE > CONTENTS rbcL Sequence Data and Phylogenetic Reconstruction in Seed Plants Foreword Gerard Zurawski & Michael T. Clegg Introduction Douglas E. Soltis, Mark W. Chase & Richard G. Olmstead ........ Phylogenetics of Seed Plants: An Analysis of Nucleotide Sequences from the Plastid Gene rbcL Mark W. Chase, Douglas E. Soltis, Richard G. Olmstead, David Morgan, Donald H. Les, Brent D. Mishler, Melvin R. Duvall, Robert A. Price, | Harold G. Hills, Yin-Long Qiu, Kathleen A. Kron, Jeffrey H. Rettig, Elena Conti, Jeffrey D. Palmer, James R. Manhart, Kenneth Ј. Sytsma, Helen J. _ Michaels, W. John Kress, Kenneth G. Karol, W. Dennis Clark, Mikael Hedrén, Brandon S. Gaut, Robert K. Jansen, Ki-Joong Kim, Charles F. Wimpee, James F. Smith, Glenn R. Furnier, Steven H. Strauss, Qiu-Yun Xiang, Gregory M. Plunkett, Pamela S. Soltis, Susan M. Swensen, Stephen E. Williams, Paul A. Gadek, Christopher J. Quinn, Luis E. Eguiarte, Edward Golenberg, Gerald H. Learn, Jr., Sean W. Graham, Spencer C. H. Barrett, Selvadurai Dayanandan & lictor A. Albert RE a tdi An Analysis of Relationships within the Cupressaceae Sensu Stricto › Based on тер — Sequences Раш A. Gadek & Christopher J. Quinn nrm- 2581 Molecular Phylogenetics of the Magnoliidae: Cladistic Analyses of Nucleotide Sequences of the Plastid Gene rbcL Yin-Long Qiu, Mark W. Chase, Donald H. Le | & Clifford R. Parks ii О - Phylogenetic Hypotheses for the. Monocotyledons REY from Hah Sequence — Data Melvin R. Duvall, Michael T. Clegg, Mark W. Chase, W. Dennis Clark, W. John Kress, Harold G. Hills, Luis E. Eguiarte, James F. Smith, | Brandon S. Gaut, Elizabeth A. Zimmer & Gerald H. Learn, Jr. .————— Phylogenetic diens of the Zingiberales Based on rbcL eius E - James | smith, W. John Kress & Elizabeth A. Zimmer .. ts Phylogenetic Relationships Among Members of a s Sedi Lato Based -~ теі. Sequence Data David R. Morgan & Douglas. E. Soltis ...... ——À as Phylogenetic Relationships of the Geraniaceae and Geraniales from rbcL Seque: Comparisons Robert A. Price & Jeffrey D. Palmer .. iade ен E Tribal Relationships i in Onagraceae: Implications from rbcL uds Data | . Бе shee Conti, Anthony Fischbach & Kenneth J. Sytsma cti Nucleotide Sequences of the rbcL Gene Indicate Monophyly of Mustard Oil Plants James Rodman, Robert A. Price, Kenneth Кан Elena = ‘Kenneth J. Sytsma & Jeffrey D. Mah i a eee дес A Parsimony Analysis of the Asteridae Sensu Lato Bind on rbe | Richard 6. Olmstead, eae Bener, Kasy M. Scott | Palmer ......... : ipsu e E yl gene ic Relationships u Cornus 1 L. Sensu rex xd Putative Rela yer Se : s from rbcL наш, Data Ne Xiang, Douglas. E. Soltis, Morgan & Pamela S. Soltis Peo Vedi ee e s MINE of the Ericaceae, Empetraceae, Epacridaceae and Rela > ÉS Upon rbcL Sequence Data Kathleen A. Kron & Mark ГА se Interfamilial Relationships sof the Asteraceae: Insights from rbcL Sequ Variation Helen J. Michaels, Kathy M. kon теше G. - Olmsteod, Szaro, Robert K. Jansen & Jeffrey D. Palmer сд се > Character-state Weighting for Cladistic RR of Protein-coding DNA Sequences Victor A. Albert, Mark ЈА Chase & Brent D. Mishle Appendix Cover illustration. : Pleurothyrium giganthum van der — парка PP... | e. ПИ Volume 80° \ Number 4 ` Volume 80, Number 4 Fall 1993 МЕЗА ВЗ OA A A A Annals of the 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 Garden will also be accepted. Authors should write the Managing 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 Henk ERR der Werff Editor, Missouri Botanical Garden . Amy Scheuler E Managing Editor, Missouri Botanical Garden E Diana Gunter I Editorial Assistant, . Missouri Botanical Garden Publications Staf ens AS A Ihsan A. Al-Shehbaz Missouri Botanical Garden Gerrit Davidse Missouri Botanical Garden Roy E. Gereau Missouri Botanical Garden Peter Goldblatt Missouri Botanical Garden Gordon McPherson Missouri Botanical Garden P. Mick Richardson | Missouri Botanical Garden L UR = h o XA POT _ Subscription price is $100 per volume U.S., $105 - Canada and Mexico, $125 all other countries. Four issues per volume. The journal Novon is included -in the subscription price of the ANNALS. ———— © Missouri Botanical Garden 1993 E) This paper meets the requirements of ANSI/NISO 239.48-1992 (Permanence o! — i y DIEM. NM іра n ir n tion contact Department The ANNALS OF THE MISSOU Eleven, P.O. Box 299, St. Louis, MO 63166-0299. | Missouri Botanical Garden, 2345 Tower ri BoTANIcaL GARD ; if rly by (ISSN 0026-6493) is published quarter” де enue, St. Louis, MO 63110. Second clase postag paid at St. Louis, MO and additional Postmaster: Send address changes to THE Missouri BOTANICAL GARDEN, j 0299. Eleven, Р.О. Box 299, St. Louis, MO 63166 | ANNALS OF ie А Volume 80 Number 4 1993 Annals of the Missouri Botanical Garden WZ SYSTEMATICS ОЕ RUELLIA (ACANTHACEAE) IN SOUTHERN SOUTH AMERICA! Cecilia Ezcurra? ABSTRACT Ruellia iio ст 23 e in southern South America, мет display great diversification i in flower and inflo- ology, pre ies as насан with a n different types of ae The g generic names ar red Mond i including Pe are delim eated as synonyms for the first time. wide arr nstemonacanthus. * ited: Chiloblec a Dipteracanthus, ps Hygrophiloidei, Phy. ‘siruellia, Rue oe iet Within these, Ruellia magniflora i is described a of pollinators; = has resulted the evolution of ave ev volved umet heterochronic informal taxonomic groups of species llia, and Salpin- new, and more than 20 specific and subspecific names Ruellia bote is excluded from n genus and reduced to synonymy ar in ae. The species are described and illustrated, and keys to all the taxa are given. Ruellia, one of the largest genera of the Acan- thaceae, consists of approximately 250 species of perennial herbs, subshrubs, and shrubs, with a largely tropical and subtropical distribution. The New World is the most important center of diver- sity for this genus, especially with regard to inflo- This variation is probably correlated with selective pres- rescence pattern and corolla morphology. sures of a wide array of different types of polli- nators, which have resulted in a great diversifi- cation of pollination syndromes. Many of the North and Central American spe- ! Į am grateful to the directors and curators of the herbaria mentioned for the loan of herbarium material and photographs or Jupe nd of types. Special thanks are given to Dick Brummit and Dieter and to the many other people that collaborated with me during visits to herbaria. help during my = at К and US, d comments of Fernand Isidro, and Laboratorio Ecotono, C.R.U.B., of this paper. I acknowle к! support, the 5 study Acanthaceae collections in North pee that financed my visits to herbaria of Sao Paulo and Rio de Janeiro, Brazil. , San Isidro (1642), Buenos Aires, Arge 2 Instituto de Botanica Darwinion, C. . Bariloche, who provided helpful discussion or assistance int dge Consejo Nacional de Investigaciones Cientificas y Técnicas (CONICET) of Argentina for mithsonian Institution and Missouri Botanical Garden for short visit grants that allowed me to American herbaria, and the Wasshausen for generous e preparation American Society of Plant Taxonomists for an B. ntina. Current ree Departamento de Botánica, Centro Regional U niversitario Bariloche, Universidad Nacional del Comahue, C. C. Bariloche (8400), Rio Negro, Argentina. ANN. MISSOURI Bor. GARD. 80: 787-845. 1993. 788 Annals of the Missouri Botanical Garden cies have been treated in local revisions, but the species in South America have only been included in a few regional floras and have never been the subjects of modern revisionary work. The number of species in the Americas is probably the largest in all continents, easily exceeding 100, and many of them have ornamental value because of their handsome, large, colorful flowers. The diversity in flower morphology has previously led to much con- fusion as to generic delimitation and sectional clas- sification, but this variation also makes Ruellia an interesting model for the study of evolution of pol- lination syndromes. e most recent overall taxonomic treatment of the genus, the subgeneric classification of Lindau (1895) in Engler & Prant's Die Naturlichen Pflanzenfamilien, is currently inadequate when one takes into account the diversity encountered in the Neotropics. Nearly a century has elapsed since Lindau's works on the family (see Loesener (1924) for a complete list), and the many new specimens that have been collected in the New World since his death provide additional data on the morphology, variation, and distribution of the species. This systematic treatment represents only a pre- liminary approach to the understanding of the neo- tropical taxa of this large and diverse genus. Al- though considerable study on a worldwide basis is needed to produce a more robust and permanent infrageneric classification of Ruellia, this revision, which covers southern South America and includes 23 species divided into seven groups on the basis of morphological affinity, is intended to encompass much of the diversity of the genus in the Neo- tropics. When extending the study to the rest of the Americas, corrections and additions may be necessary, but it is expected that this classification shall remain as a basic framework for future stud- s. Most of the previous treatments, being of a regional character, have covered only a small por- tion of the ranges of the species. This has resulted in the proliferation of names associated with dif- ferent regions, which in many cases are synonyms. This study therefore also aims to correctly delimit the species in the area and to clarify their nomen- clature. MATERIALS AND METHODS This treatment covers Argentina, southern Bra- zil (states of Paraná, Santa Catarina, and Rio Gran- de do Sul), Paraguay, and Uruguay. Although this is a politically limited area, it is representative of all the neotropical floristic regions found in south- ern South America and includes nearly the entire basins of the rivers Paraná, Uruguay, and De la Plata, along with their tributaries. This implies that it is a geographically more or less isolated region, physically separated by watersheds from other ba- sins in South America. Although the present study is based mainly on herbarium material, live populations of most of the species were studied during collecting trips to northern Argentina in 1984 (with M. Ponce and . Brown), 1985 (with R. Kiesling et al.), and 1986 and 1989 (with F. Zuloaga et al.). More than 1,000 specimens from the following herbaria were studied: AA, BA, ВАВ, BM, CTES, C, GH, HAS, HBR, ICN, K, LIL, LP, MBM, MO, MVFA, NY, P, PACA, PKDC, R, RB, SI, SP, SPF, US. Type specimens or photographs of types from other her- baria (e.g., BR, GOET, GZU, LE, M, W, S) were requested, or studied at US where some were on loan. In several cases, only a selection of specimens representative of the distribution and diversity of the species are cited, because the collections stud- ied are too numerous. The nomenclature of phytogeographical regions follows Cabrera in Cabrera € Willink (1980). Ca- brera’s floristic approach to phytogeographical classification has been preferred to a more physi- ognomic treatment such as Hueck's (1978), be- cause the floristic composition is believed to be more related to the evolutionary history of the vegetation in an area. Regarding hybridization, the most important character used in this work to recognize naturally occurring hybrids is morphological intermediacy between two known species, either in plants in the field or in herbarium specimens. Additionally, the sympatric occurrence of the presumed parental species in the same area as the putative hybrid was taken into account. In some cases decreased pollen viability has also been accepted as a criterion of hybridization by studying pollen stainability with anilin blue in lactophenol. One of the major problems encountered during this work has been that the majority of names of taxa of Acanthaceae from southern South America were proposed by Nees (1847a, b), who did not formally designate nomenclatural types for them. In the case of species, he referred to the specimen or list of specimens he had studied and considered representative. Here no lectotypes have been des- ignated from the syntypes when more than one specimen was cited, because in most species 1 have not had direct access to the whole set of syntypes, Volume 80, Number 4 1993 zcurra 789 Ruellia in Southern South America as this material is generally found in several dif- ferent European herbaria. In the cases of species where Nees proposed several infraspecific taxa, he designated them with Greek letters. As he generally considered only one infraspecific rank, these should be regarded as va- rieties, following Article 35.3 of the /nternational code of Botanical Nomenclature (Greuter et al., 1988). In some descriptions he named and de- scribed an alpha variety and cited material for it, in which case no material is cited for the name of the species. In these situations the holotype (or syntypes) of the name of the species is here as- sumed to be that specimen (or specimens) cited under the infraspecific taxon designated with the Greek alpha. This assumption is based on the fact that in most cases of species with several infra- specific epithets, Nees only named and described alternative varieties beta, gamma, delta, epsilon, etc., and when listing the material of the species, he referred to the alpha variety as that described under the specific epithet, in the general description (e.g., Arrhostoxylon affine Nees, 1847a: 59). In a few cases he did not refer to an alpha variety (e.g., Dipteracathus macranthus Nees, 1847a: 37) when citing material, but just listed in the first place the specimens he regarded as rep- resentative of the species, and secondly those rep- resentative of varieties beta or gamma, etc. In those cases it is clear that the holotype of the species is the specimen cited in relation to the description of the species, and the material listed for infraspecific taxa beta, gamma, delta, epsilon, etc. is excluded. TAXONOMIC HISTORY AND GENERIC DELIMITATION The complex taxonomic history of Ruellia has been discussed in detail by Long (1964, 1973). The following is only a summary of the prevailing ideas on the delimitation of КиеШа through time. Ruellia was established by Linnaeus (1737), who included eight species in his Species Plan- tarum (1753), of which Ruellia tuberosa, a trop- ical American perennial herb with blue flowers, was chosen as lectotype (Britton & Brown, 1913: 241). Nees (1847a), who treated the Acanthaceae of South America for Martius’s Flora Brasiliensis and the family on a worldwide basis for the Prod- romus of De Candolle (Nees, 1847b), placed the species described by Linnaeus in various genera. He even described more new genera to accom- modate the numerous different but morphologically related variations he encountered in the tropics, especially in the New World. Oersted (1854), who worked on the Acanthaceae of Central America, also treated the species in several separate genera. Bentham (1876) established the now mostly ac- cepted circumscription of the genus by "sinking" 18 genera, most of them described by Nees, into Ruellia. Lindau (1895) followed Bentham, includ- ing in Ruellia more than 100 species classified in 9 sections. This wide concept has been accepted by most modern authors. The New World species have generally been treated in the broad generic sense, especially in North and Central America (e.g., Daniel, 1990; Durkee, 1978, 1986; Gibson, 1974; Leonard, 1936; Long, 1970; Tharp & Barkley, 1949; Tur- ner, 1991; Wasshausen, 1966). In South America most authors have also followed Bentham and Lin- dau treating Ruellia in its wider sense (e.g., Chodat & Hassler, 1903; Dawson, 1979; Ezcurra, 1989; Ezcurra & Wasshausen, 1992; Leonard, 1951- 1958; Lillo, 1937; Lindau, 1894; Rambo, 1964; Wasshausen & Smith, 1969). But perhaps because of the bewildering diversity that can be found in the tropics of South America, especially with re- spect to flower and inflorescence morphology, some authors have gone back to Nees' ideas in this area (Bremekamp, 1938, 1969; Bremekamp & Nan- nega Bremekamp, 1948; Rizzini, 1949). The result has been that some species names in Ruellia have recently been treated once again in combination with names of genera previously considered syn- onyms, such as Arrhostoxylum Nees (Bremekamp, 1969) and Stephanophysum Pohl (Bremekamp, 1938, 1969; Barker, 1986), or with new names, such as Ulleria Bremek. (Bremekamp, 1969). This problem has led to the series of biosyste- matic works of Long (e.g., 1966, 1973, 1975, 1976a, b) that, through field studies, crossing ex- periments, and chromatographic comparisons of corolla pigments, have aimed to establish the rel- ative taxonomic position of several species within Ruellia sens. lat., and at the same time provide an insight to the most natural circumscription of the genus. These works, together with the cyto- genetic studies on the family performed by Grant (1955), Daniel et al. (1984, 1990), and Piovano & Bernardello (1991), generally support the idea that КиеШа consists of a large and morphologically diversified group of more or less phylogenetically related species, with one constant chromosome number, in which natural hybridization is frequent between closely related species. In this treatment Ruellia is treated sensu lato," being characterized by the generally large and showy, subactinomorphic flowers, the four, fre- 790 Annals of the Missouri Botanical Garden quently didynamous stamens, the oblong, sym- metrical, basally rounded anthers, the spherical, porous, coarse-reticulate pollen, and the (2—)4— numerous-seeded capsules, with the placentas re- maining attached to the walls at maturity (sepa- rating in a few species). INTRAGENERIC RELATIONSHIPS most recent comprehensive treatment of Ruellia, Lindau's (1895) synopsis, separated the genus into nine sections, of which only five were "typical." Of these five, three of them included the majority of the species considered by Lindau as known at that time: ЕигиеШа (which included spe- cies classified by Nees (1847b) as Cryphiacanthus Nees and Ruellia L. pro parte), Dipteracanthus (which included Ophtalmacanthus Nees, and Dip- teracanthus Nees groups Hygrophiloidei, Caly- cosi, and Genuini), and Physiruellia (which in- cluded Aphragmia Nees pro parte, Arrhostoxylon Nees, Dipteracanthus Nees pro parte, Siphonac- anthus Nees, Stemonacanthus Nees, and Stepha- nophysum Pohl). Most of the others were mono- typic sections, created mostly to accommodate species of uncertain position. The characters he used in the delimitation of groups were capsule morphology and corolla shape. In this synopsis Lindau noted that both the delimitation of the spe- cies and the separation of the genus into sections were difficult in Ruellia because of the extraor- dinary richness he encountered in form, and that his classificatory scheme was provisional. Recently, Ramamoorthy (1991, 1992) de- scribed two new sections from Mexico, Chiropter- ophila and Urceolata, т characterized by flower morphology; c was not taken into account. In this treatment, the species are classified into groups without formal taxonomic recognition, prin- cipally on the basis of capsule structure correlated with inflorescence morphology. Corolla shape ap- pears to have frequently been subject to rapid diversification through disruptive selection by dif- ferent pollinators. This appears to have produced e parallel events in the evolution of the genus, , homoplasies, and therefore this character is ad as secondary in group delimitation. Although this classification is also provisional because it takes into account only a fraction of the species of the genus (probably less than 15%), this fraction is quite representative of the diversity of Ruellia encountered in the New World, as seen in the neotropical collections studied. Ви! some important groups of species, such as those consid- ered by some authors (e.g., Long, 1976b) as Aphragmia Nees (= Dipteracanthus group Aphragmia Nees, 1847a; Dipteracanthus group Paniculati Nees 1847b; Ruellia sect. Physiruel- lia group Glandulosae Lindau, 1895), are not represented in the area studied, and therefore have not been included. Thus the formal taxonomic po- sition and rank of the groups will have to be worked out once all the species are taken into account on a worldwide basis. MORPHOLOGY Habit. Most of the species are perennial herbs or subshrubs, more rarely shrubs or small trees, the majority erect, sprawling, decumbent or prostrate, except a few semiscandent or rosettiform species. All have a basal rhizome with more or less thick, fibrous roots, and sometimes adventitious roots are also produced in decumbent branches where the nodes touch the ground. The young stems are frequently quadrangular, in some species more markedly so, in others subterete, and in some of group Physiruellia the regions over the nodes are swollen and geniculate. The swollen region frac- tures easily and may be important in the vegetative propagation of the plants by stem division. Leaves. Leaves are opposite and generally petio- late, and only a few species (of group Ruellia) present leaves in rosettes. The blades are simple, with the margin entire, crenulate or crenate, more rarely dentate. The surfaces are pubescent or gla- brate. The leaves sometimes intergrade into bracts toward the stem apices. Venation is pinnate, but, in the upper leaves transformed into bracts in some species of group Chiloblechum, the inferior sec- ondary veins arise near or at the base of the blade, producing a subpalmate appearance. Indument and surface. The trichomes in Ruellia, like in most Acanthaceae (Ahmad, 1978; Karl- strom, 1978), are of two types: glandular, with a sessile or stalked multicellular head, and eglan- dular, simple and uniseriate (Fig. 1A). Different types of pubescences can occur together in differ- ent proportions. In some multicellular, ses- sile, disk-shaped or “patelliform” (Daniel, 1990) glands appear like small punctations on the surface of leaves, bracts, and calyx lobes (Fig. 1B). These species glands are frequent in tropical and subtropical spe- cies of different taxonomic groups from open ar- eas, such as Ruellia bulbifera, R. coerulea, R. erythropus, or R. multifolia, and could act as extrafloral nectaries secreting substances to attract ants. Volume 80, Number 4 1993 Ezcurra Ruellia in Southern South America Cystoliths are small mineral (calcium carbonate) concretions in the interior of epidermal cells and are present in most genera of Acanthaceae. In Ruellia, on the surfaces of vegetative organs, in dry material, the cystoliths appear as small, linear, more or less prominent structures, sometimes re- sembling appressed, strigose pubescence. In live material they are much less readily apparent. Inflorescence. The basic inflorescence pattern in Ruellia consists of simple or compound dichasia that appear as lateral elements of the indeterminate axes (Sell, 1969а). Within this basic scheme, sev- eral variations that generally characterize different groups of related species can be found. In groups Physiruellia and Ruellia the dichasia are often compound and pedunculate, forming mul- tiflowered axillary cymes (Fig. 2A, B). These are sometimes condensed into terminal thyrses, es- pecially in the latter group, producing an upper floral region differentiated from the purely vege- tative basal region by modification of the terminal leaves into small bracts (Fig. 2B). In group Hy- grophiloidei the dichasia are generally reduced to a more or less pedunculate solitary flower with two bracteoles, in the axils of the terminal leaves. In many of these cases the inflorescences are aniso- FiGURE 1. from bract surface of R. erythropus, Pensiero 1580. Scale bar — 100 yum. phyllous, i.e., only one cyme appears in each node, alternating sides (Fig. 2A). In group Dipteracanthus, the dichasia are al- ways simple and sessile, generally reduced to ax- illary, solitary sessile flowers with two bracteoles, these always subtended by the upper leaves (Fig. 20). In groups Salpingacanthus and Ebracteolati the flowers also appear sessile and solitary in the axils of the upper leaves, but the reduction is even greater as the bracteoles are extremely reduced or absent (Fig. 2D). In group Chiloblechum the sessile flowers without bracteoles are grouped into a four- sided spike, the upper leaves generally being some- what modified into sessile bracts (Fig. 2E). In this work, the correlation between inflorescence and capsule structure is used to define groups. Calyx. In Ruellia the calyx is always five-parted, but in group Salpingacanthus and in some species of group Chiloblechum (e.g., Ruellia erythropus) the two lateral lobes are fused to the inferior ones, and thus a more or less three-lobed structure is formed. Corolla. The corolla consists of a narrow basal tube, a more or less cylindrical, infundibuliform or campanulate throat, with the limb parted into two 792 Annals of the Missouri Botanical Garden E P ce A ы o Ko A B FIGURE 2. Ruellia). —C. К. multifolia (group Ruel Chiloblechum). (Schematic, from he V ais mestre ) posterior and three anterior lobes more or less similar in size and shape, thus rendering it subac- tinomorphic in a basically zygomorphic family. This basic structure is widely modified in different spe- cies. The color also varies from blue, mauve, pink, red, and yellowish, to white. The lower part of the throat and/or the base of the anterior lobes is often marked with colored streaks or veins that probably function as nectar guides. Corolla morphology seems very plastic in Ruellia and has probably been sub- јес! to diversification due to the action of different pollination agents, even within groups of otherwise closely related species. The different sets of mor- phological characteristics of the flowers are here described in relation to presumed pollinating agents, as pollination syndromes. Although there is no strict correlation between the different morphological types of corolla and the different groups, some types are present more often in some groups than others. The following classi- fication is somewhat similar to that proposed by Barker (1986) for flower morphology of Australian Acanthaceae in general. Inflorescence types in ети е — А. К. angustiflora (group Physiruellia).- . R. geminiflora (group Ebracteolati).— etu oes => Сте “=> = С D E - B. R. ciliatiflora (group E. R. erythropus (group Red tube flowers (presumably pollinated by hum- 3A, B). The co- rollas are frequently horizontal or pendulous and mingbirds: ornithophilous) (Fig. red, the basal tube and the throat form a more or less narrow tube, sometimes anteriorly inflated, and the lobes are generally small and erect or semi- erect. This morphological type generally appears associated with multiflowered axillary cymes as in- florescences, with the peduncles and pedicels fre- quently weak. It is frequent in group Physiruellia (e.g., Ruellia angustiflora, R. elegans). Mauve or blue infundibuliform flowers (presum- an ag by large bees: melitophilous) (Fig. 3C The basal tube is short, the corollas are blue or mauve and functionally zygomorphic, i.e., generally angled and tilted, with marked palate rib markings in the lower part of the throat, and with the lower lobes with nectar guides in the base. This type is common in groups Ruellia, Dipteracan- thus, and Ebracteolati (e.g., Ruellia ciliatiflora, R. brevicaulis, R. geminiflora). White trumpet flowers (presumably pollinated by NT: Corolla morphology in Ruellia. A, B. Red tube ornithophilous flowers. —C-F. Mauve or pale blue — =) =. Fic infundibuliforma melitophilous flowers. — C, psychophilous flowers D, R. bulbifera; E, R. R. hypericoides.) F, R. geminiflora; е а фор | Зен owers. — . (Photocopied from he rhasium material: b brevicaulis; G, ^3 la n R. macrosolen; 1, R. Blue m ;B, R. angustiflora; C. atiflora; ery ieu jd J; Volume 80, Number 4 993 Ezcurra 793 Ruellia in Southern South America FIGURE 4. Pollen of Ruellia. - coerulea (West 8472).—B. R. hypericoides (Zuloaga 3248). (SEM photographs of unacetolized pollen grains. Scale bar = 10 um.) hawkmoths: sphingophilous) (Fig. 3G, H). The bas- al tube is very long, and the corollas are more or less erect, white or whitish, actinomorphic, lacking differential markings in the lobes. This type of flower appears in several unrelated groups present in southern South America, such as Dipteracan- thus, Ruellia, and Salpingacanthus (e.g., Ruellia multifolia, R. macrosolen, nobilis), and has probably evolved independently many times within the genus. Blue hypocraterimorph flowers (presumably pol- linated by butterflies: psychophilous) (Fig. 31, J). The basal tube is very narrow, and the corollas are erect, blue, and without clear differential markings in the lobes. This type appears in group Chiloble- chum (e.g., Ruellia hypericoides, R. erythropus). Additionally, several neotropical species present large, pale, greenish, dull-colored, campanulate flowers that have saccate corollas with a wide open throat and exserted stamens (e.g., Ruellia petio- laris, Daniel, 1990; Ruellia section Chiropter- ophila, Ramamoorthy, 1991) that appear to be pollinated by bats (chiropterophilous), but none with this syndrome reaches southern South America. Androecium. In Ruellia there are four stamens which are generally didynamous, the two lateral being the shorter pair and the two anterior the longer, but the filaments are curved so that the anthers are placed under the posterior lobes, more or less deeply included or exserted. In ornithophi- lous species the stamens are sometimes all the same length. In all cases, the filaments are inserted below the throat of the corolla and are prolonged toward the interior and base of the corolla tube as two narrow protruding flaps that more or less separate the tube into a posterior (dorsal) section and an anterior (ventral) longitudinal channel. Pollen. In the Acanthaceae, pollen morphology has traditionally been a useful taxonomic character in group delimitation at the generic and suprageneric level (Lindau, 1895; Petriella, 1968; Raj, 1961, 1973; Rizzini, 1949; Scotland, 1991). Ruellia spe- cies have spherical, porate, coarse-reticulate pollen “~ Fig. 4). Вгетекатр & Nannega Bremekamp 1948) and Bremekamp (1969) suggested that dif- ferences in pollen morphology, such as sparsipo- ~ rous grains versus three-porate ones, and differ- ences in the constitution of the reticulum, such as size and regularity of the brochi, height of the muri, be used to separate the Old World species of Ruellia from the New World ones. In etc., coul southern South American species the variations in size of the grains and in ornamentation of the walls 794 Annals of the Missouri Botanical Garden B F G H | Fic Capsule oe in Ruellia. —A-E. Group Ruellia. —F-1. Group Physiruellia. —J, К. Group Hygrophiloidei —L, M. Gr ~ Dipteracanthus —N, O. Group s np olati. —P, Q. Group Chiloblechum. (Open without seeds drawn with camera rud semisc nao ps the same specimens used to illustrate the же к of the species. А, К. coerulea; rosolen; С, R. ciliatiflora D, R. r E, R. hygrophila; 4 R. angustiflora; G, R. brevifolia; H, R. e pedia , R. sanguinea; J, »pallocaulos; K, R. kleinii; , R. multifolia; M, R. brevicaulis; N, R. bulbifera; O, R. geminiflora; P, R. ма im Q, R. hypericoides.). edid do not correlate with any other morphological char- acters to help in the separation of groups of related species. Gynoecium. The stigma is two-lobed with the lobes dorsiventrally compressed, the anterior lobe being generally larger than the posterior. Capsules. In Ruellia, like in most Асапћасеае, the fruit is a biloculate, two-valved, loculicidal cap- sule. The base of the capsule is contracted and solid, and the seeds are found in the hollow upper part, subtended by hooklike retinacula. The dif- ferent mechanisms by which the capsules open and the seeds are ejected are described by Sell (1969b). In some species of Ruellia the dehiscence is pro- duced by the drying of the capsules (xerochasy), and in others by humidification (hydrochasy). The structure of the capsule and the number of seeds per locule (1—14) are variable and are im- portant taxonomic characters at a supraspecific level. Differences in the ratio between the length of the sterile, contracted, solid base, and the fertile, thicker, hollow upper part, characterize different groups of related species. Group Ruellia (Fig. 5A-E) is characterized by the terete, narrowly elliptic capsules with a very short solid base (less than М the length of the capsule) and numerous seeds per locule (5-14). Group Physiruellia (Fig. 5Е-1) has a longer solid base in relation to the hollow part (3), and fewer seeds per locule (3-7). Group Hygrophiloidei (Fig. 5J- K) has clavate capsules with a short, rounded, apiculate hollow head of nearly the same length as the solid base, with very few seeds per locule (3- Volume 80, Number 4 1993 Ezcurra 795 Ruellia in Southern South America 4). Group Dipteracanthus (Fig. 5L, M) has terete, robust, thick-walled capsules with a short solid base (уз the length), and 3-6 seeds per locule. Group Salpingacanthus has a similar structure, with very large capsules. Group Ebracteolati (Fig. 5N, О) has an extremely reduced capsule with a very short solid base and only (1—)2 seeds per locule. Group Chiloblechum (Fig. 5P, Q) has small capsules with thin walls, a solid base М the length, and 2-4 seeds per locule. This last group differs markedly from the rest of the genus because on dehiscence, the protruding placentae with the retinacula and the seeds break away from the inner walls of each valve, separating elastically from the lower part of the capsule toward the top, and at the same time breaking the lateral walls. This type of dehiscence of the capsule has already been described for several unrelated genera within the family (e.g., Dicliptera, Micranthus, Tetramerium, Blechum) (Daniel, 1986; Sell, 1969b), but it is the first time it has been reported for Ruellia. Seeds. Тће seeds are always flat and more or less orbicular in outline, with a slight notch at the chalazal end. The surface of the seed toward the outside of the capsule is usually convex, and the one toward the interior flat or concave. The seeds are more or less covered with appressed hairs that extend and turn mucilaginous on contact with wa- ter. This makes them extremely adhesive when wet, and in this way they can be dispersed by animals (ectozoochory). These hairs possess an- nular or spiral thickenings in their walls which strengthen their structure; the annular and spiral shape of the thickenings allows the hairs to lengthen on contact with water. [n many species of group Physiruellia the hairs are condensed principally on the margins of the seeds, but otherwise the structure of the surface is quite constant. The mucilage the hairs secrete could also have a pro- tective function, as it has shown high resistance to enzyme action (Lester & Ezcurra, 1991). CHROMOSOME NUMBERS The chromosome numbers of only less than 20% of the species of Ruellia have been studied, but they appear to be fairly uniform: more than 85% of the species are 2n = 34 (Piovano 4 Bernardello, 1991). As Daniel et al. (1984) suggested, this constancy of chromosome number in all species of Ruellia, in contrast to the wide chromosomic vari- ation that has been encountered in Justicia, the largest genus of the family, may be reflective of a more natural, though also large and morphologi- cally diverse, taxon. The chromosome numbers of only eight of the species present in southern South America have been reported: Ruellia brevifolia (as R. graecizans) and R. ciliatiflora (as R. lo- rentziana) (Grant, 1955); R. macrosolen (Ezcurra & de Azkue, 1989); R. bulbifera, R. coerulea (as R. tweediana), R. geminiflora, R. multifolia var. multifolia (as R. aff. dissitifolia), and R. longi- pedunculata (Piovano & Bernardello, 1991). Most Ruellia species have one pair of satellite chromosomes (Grant, 1955; Ezcurra & de Azkue, 1989; Piovano & Bernardello, 1991), and chro- mosome size seems quite variable (from 0.60 to 3.60 um as reported by Piovano & Bernardello). ORIGIN OF POLLINATION SYNDROMES The impressive morphological diversification of the flower in species of Ruellia from the New World suggests a limited set of morphological types. Distinct morphological flower types that appear recurrently within different genera and families of plants have traditionally been interpreted as rep- resenting pollination syndromes, i.e., the set of morphological, phenological, and physiological characters of flower and inflorescence that corre- late with the morphology, behavior, and prefer- ences of the effective pollinators of the species, and therefore represent adaptations to various modes of pollination (Ornduff, 1969). These dif- ferent pollination syndromes have repeatedly been described for various taxonomic groups and have been summarized by Faegri & van der Pijl (1979) and Weberling (1989). Even though Ornduff (1969) and Vogel (1991) cited several genera of gamopetalous families to exemplify complex patterns of adaptive radiation or divergence in floral morphology, few of them seem, as suggested by their morphological diver- sity, to have exploited such a diverse pollinator fauna as Ruellia. The diversity of floral form in this genus is similar to the adaptive radiation de- scribed by Grant & Grant (1965) for the Pole- moniaceae and by Vogel (1991) for the Rubiaceae and Scrophulariaceae. Cleistogamy in Ruellia is frequent (see Lord, 1981), and production of a great variety of floral morphs, from normal and reduced chasmogenes to semi-cleistogenes and cleistogenes, as has been de- scribed for the North American Ruellia caroli- niensis (Long, 1971), suggests that floral mor- phology in the genus is labile, lacking developmental channelization. Different floral morphs within the same species have also been observed in cultivated specimens of the South American Ruellia brevi- 796 Annals MAU а Garden folia (Piovano et al., 1991), indicating develop- mental plasticity. As summarized by Macnair (1989), in plant populations that are genetically variable regarding flower shape, size, coloration, or flowering time, different pollinator guilds can select different floral morphs. This can produce disruptive selection and result in the sympatric differentiation of species with different floral morphologies. Differences in morphology related to different developmental stages can provide the variation on which selection can operate. This process has been suggested as the evolutionary origin of the corolla shape of the derived hummingbird-pollinated Del- phinium nudicaule Torrey ray from a more generalized bumble-bee-pollinated species of Del- phinium, through selection of neotenic forms by bird pollination (Guerrant, 1982) This type of process, but through the selection of peramorphic instead of neotenic forms (as results of developmental acceleration or hypermorphosis), has been suggested to explain the evolution of Ruellia macrosolen (Ezcurra & de Azkue, 1989), a sphingophilous species with long, actinomorphic, white flowers, from sympatric and closely related melitophilous species, with shorter, functionally zy- gomorphic, blue flowers. The heterochronic variability in development that has been observed in flower morphology within some species (expressed by the high frequency of cleistogamic and semicleistogamic or reduced forms), probably with a genetic basis, could thus be related to the wide array of floral syndromes in the genus. As has already been suggested (Ezcurra & de Azkue, 1989), the evolutionary shift from one floral syndrome to another may therefore have originated by simple heterochronic developmental changes selected by different pollinators. REPRODUCTION AND DISPERSAL The role of cleistogamy in the reproductive bi- ology of Ruellia has been discussed by Long (1971) in his detailed study of the floral polymorphy and breeding system of the North American Ruellia caroliniensis. Production of cleistogamic flowers has also been reported in many other genera of Acanthaceae, and Long has suggested that it is probably as important in this family as in Poaceae and Orchidaceae. Cleistogamic flowers in Ruellia are frequently inconspicuous structures with closed corollas of reduced size, difficult to differentiate from flower buds in herbarium material, but easily observed in live plants, where the capsules are produced without the opening of the flowers, and therefore present the closed corollas crowning the distal end, sometimes taking some time to fall off. n many cases the expression of cleistogamy has shown to have relation with external environmental factors (Uphof, 1938), like seasonality and envi- ronmental stress (Long, 1971; Piovano et al., 1991; Sell, 1977). This could explain the higher fre- quency of cleistogamic flower production in culti- vated plants, especially in those transplanted to higher latitudes. This has been observed in plants of Ruellia ciliatiflora (Santos Biloni, 1963, as К. lorentziana) and R. macrosolen (Ezcurra & de Azkue, 1989) transplanted from northern Argen- tina (approximately 2595) and cultivated in Buenos Aires (approximately 35°S lat.). The combination of chasmogamic floral morphs well adapted to cross-pollination, with the produc- tion of reduced or cleistogamic ones in which au- togamy is prevalent, serves as a means of assuring fertilization in cases where the efficiency of cross- fertilization decreases (Macnair, 1989). This is par- ticularly so in cases of a single plant establishing itself in a new habitat, as is common in long- distance dispersal events, where generally one or a few seeds get introduced in a new area at a time. Moreover, colonizing plants frequently encounter a somewhat hostile environment in their new hab- itat, and the production of cleistogamic flowers in response to environmental stress, as 15 the case with Ruellia, ensures reproductive success with the minimum investment in flower production. As Long (1976a) suggested, this succession of cycles of allogamy, with cycles of facultative au- togamy that produce a whole series of homozygous biotypes, is probably the cause of many of the taxonomic problems present in Ruellia. Self-fer- tilization produces morphologically distinct local populations each with a low amount of diversity that can be erroneously identified as different spe- cies, and the outcrosses between different popu- lations of this type can produce complex patterns of reticulate variation. Hence, the confused tax- onomy of many Ruellia species groups may be related to these characteristics of the breeding sys- tem of the genus Ruellia seeds turn mucilaginous and adhesive on contact with water and in this way can be dispersed by animals (ectozoochory). Aquatic birds, especially migrating ones, could account for long- distance dispersal of such species as Ruellia coe- rulea and related entities of group Ruellia (e.g., R. ciliatiflora, R. macrosolen), which possess cy- lindrical capsules with numerous seeds and grow in open, disturbed, wet or seasonally flooded hab- itats. Although the distribution of the widespread and variable Ruellia coerulea is restricted to sub- tropical South America, there are species in south- Volume 80, Number 4 zcurra 797 Ruellia in Southern South America ern North America that seem closely related, such as Ruellia malacosperma and Ruellia brittoniana (cf. Turner, 1991). These could have originated by introductions through this process. tems fracture easily in the swollen regions over the nodes of many species, and this could also help in their dispersal. Decumbent stems of some species that grow in moist humus of the understory of forests root readily, and naturalized plants of Ruel- lia brevifolia in Instituto Lillo, San Miguel de Tu- ситап (Argentina), have been seen to reproduce in this way, resulting in dense thickets. HYBRIDIZATION Natural hybridization in Ruellia can be expected to be important in relation to two factors. First, many artificial crosses have resulted in hybrids, several with a high degree of fertility (Long, 1966, | , 1975, 1976a, b), suggesting that most зре- cies lack internal barriers to hybridization, es- pecially when they are relatively closely related. Second, many species are sympatric in nature. However, evidence for naturally occurring in- terspecific hybridization has been found only in a limited number of cases, and the putative hybrids (Ruellia brevifolia x R. longipedunculata, R. brevicaulis X R. coerulea) are discussed under the suspected parental species in the taxonomic section. The apparent rarity of natural hybrid pro- duction in the studied area may be due to the difficulties of recognizing them in herbarium ma- terial, especially between morphologically similar but variable species, and even in the field, when hybrid plant specimens or popu localized. Natural hybridization could therefore be underestimated, although it may have played an ulations are rare and important role in the evolution of the genus. The existence of presumed naturally occurring hybrids between species pairs belonging to widely separated species groups (e.g., Ruellia brevicaulis x coerulea) suggests evolutionary reticulation, which could explain the taxonomic difficulties in delimiting supraspecific taxa within the genus. It is also possible that sympatric species remain separated by external barriers, such as different pollinators, as has been suggested between Ruellia macrosolen and R. ciliatiflora (Ezcurra & de Az- kue, 1989), or different ecological preferences in habitat, such as those between КиеШа ciliatiflora and Ruellia coerulea. GEOGRAPHIC DISTRIBUTION AND HABITAT Ruellia, being a mainly tropical genus, extends in South America to approximately 35°S and is present only in areas with floras belonging to the Amazon and Chaco floristic regions or dominions. In southern South America (from 2095 southward) these floristic regions have been subdivided into several phytogeographical provinces, each char- acterized by a different geography, climate, and vegetation. These provinces are frequently desig- nated as Atlántica, Cerrado, Paranaense, and Yun- gas (Dominio Amazonico), and aco, Espinal, Monte, and Pampas (Dominio Chaqueno) (e.g., Ca- brera, 1976; Cabrera & Willink, 1980). Ruellia species are found in all these floristic provinces, except in the dry Monte. The climate of these areas varies from hot and wet year-round, with over 2,000 mm rainfall, to warm and wet in sum- mer and mild and dry in winter, with ca. 500 mm of rainfall especially during the summer months (from November to April). Most of the species grow well in open and dis- turbed places, such as seasonally flooded areas or savannas with periodic fires, and are common in secondary growth. Even the ones that grow in the understory of forests are generally found in clear- ings and openings, roadcuts, and along stream or hese characteristics together with their facultatively autogamous breeding system and mode of dispersal by animals (ectozoochory) make many species good colonizers and assure their suc- cess in long-distance dispersal events. This in some cases renders them important weeds in tropical and subtropical regions (e.g., Ruellia ciliatiflora). ultivation as ornamentals may have also helped to extend the natural range of some of the species, such as Ruellia brevifolia, which has become nat- uralized in many tropical places of the world (see Ezcurra, 1989) ECONOMIC [MPORTANCE Several species of Ruellia are cultivated mostly because of their showy flowers, sometimes in com- bination with attractive foliage. А few of the native ones from southern South America are frequently grown in warm regions all around the world, such as the red-flowered Ruellia brevifolia (generally known in horticulture as К. graecizans) and Ruel- lia elegans (syn.: R. formosa) (e.g., Everett, 1982). ther species of the area, such as Ruellia angus- tiflora and the blue-flowered R. ciliatiflora (gen- erally reported as К. lorentziana), are not so com- mon, and are cultivated more rarely (Dimitri, 1972; Ezcurra, 1989). Some of these subtropical species are probably hardier than their more tropical rel- atives and may adapt better to cooler regions of the world, hence they could be introduced in cul- tivation. Some subtropical Acanthaceae native to north- 798 Annals of the Missouri Botanical Garden ern Argentina have been reported as having forage value (Burkart, 1943), especially in those areas with a seasonal climate with drier and colder periods where grasses are sometimes scarce. КиеШа er- ythropus, one of the most common and widespread species of Ruellia in southern South America, which is also found in southeastern Mexico, has been reported as good fodder (Burkart, 1943, errone- ously as Ruellia bahiensis). This suggests it could be an important plant in the economy of those regions and should be investigated in relation to nutritional value. TAXONOMIC TREATMENT Ruellia L., Sp. Pl. 634. 1753. TYPE: R. tuberosa L. (lectotype, selected by Britton & Brown (1913)) Mri ple tom Pohl, Bras. Icon. Descr. 2: TYP Stphanophysum е пи ara selec Bremekamp & Nan Bremekamp (1948). Dipteracanthus Nees in Wallich, Pl. Asiat. Rar. 3: 75, 81. 1832. TYPE: K ea n а це Nees. арма аш Lindl., Nat. Syst. Bot ad, 2: 444. 36. TYPE: Aphragmia haenkei Nee Gymnac vh Nees in Linc st. Bot. P 2: 444. Yon Oersted 1854. TYPE: е anthus pe- — Iris ; Nees Cryphiacanthus Nees in p Sem. Hort. Vratislav. innaea 16: 1842. TYPE: Cryphiacan- n barbadensis die nom. illeg. (= К. tuberosa Т). н Nees in Benth., Pl. Hartw.: ai 1846. TYPE: Scordoxylum hartwegianum Nee a Nees in Benth., 2- Bur Hs 145. 1846. TYPE: dpi при“ Nees. Siphonacanthus Nees in C. Martius, Fl. 1847. TY E: Siphonacanthus villosus Nees. Euryc panes Nees | in C. Martius, Fl. Bras. 9: 52. TYPE lurychanes ln Nees (' Sai cifolius’ Stemonacanthus Nese i in Martius, Fl. Bras. 9: 53. ee TYP. seen Se salviaefolius Nees (lectotype, Maris by Leonard (1951-1958)). Arzhostoxylon Ne es in C. Martius, Fl. Bras. 9: 57. 1847. TYPE: Arrhostoxylon glabrum Nees (lectotype, se- ica by Bremekamp & Nannega Bremekamp Penstemonac misil Nees in C. Martius, Fl. Bras. 9: 159. 847 PE: Penstemonacanthus modestus Nees. пае Oersted, Vidensk. Meddel. Dansk. Na- Foren. Kjobenhavn 1854: 126. 1854, not Nees, 1836. TYPE: Gymnacanthus geminiflorus (HBK) ree (= Ruellia geminiflora HBK). Blechum Nees subgen. Chiloblechum Oersted, Vidensk. Dansk Naturhist. Foren. Kjobenhavn 1854: . 1854. TYPE m mexicanum Oersted totype, selected here Copioglossa Miers, Proc. Roy. Hort. Soc. 3: 294. 1863. Copioglossa pilosa Miers на S. Moore, J. Bot. 42: 33. 1904. TYPE: Tremacanthus roberti S. Moo Salpingacanthus S. Moore, J. Bot. 42: die 1904. TYPE: Salpingacanthus nobilis S. Moor Ulleria Bremek. Proc. Kon. Ned. Aka d. Wain sch. Ser. 14: 423. 1969. TYPE: Ulleria и (НВК) Bremek. (= Ruellia geminiflora HB Perennial herbs, shrubs or subshrubs, erect to decumbent, = pubescent. Leaves opposite, petio- late or subsessile, the blades entire, crenate or slightly dentate, with cystoliths. Flowers sessile or pedicellate, in axillary sessile dichasia or in pedun- i ; the bracts and brac- teoles generally small, not imbricate. Chasmoga- culate multiflowered cymes mous flowers medium or large, showy, blue, purple, or red, to mauve, lavander, pink, and white, rarely yellowish; cleistogamous flowers small and tubular. Calyx deeply cleft, the five segments linear or lan- ceolate, usually subequal, sometimes the posterior longer, rarely the lateral united to the anterior ones. Corolla straight or incurved, tubiform, funnelform or salverform, the basal portion of the tube narrow, abruptly or gradually dilated to a + obconic or campanulate throat, sometimes anteriorly inflated or more rarely saccate, the limb generally spread- ing and oblique, sometimes erect, the five lobes ovate or rounded, contorted in bud, equal, some- times the two posterior + connate, forming a pos- terior lip. Stamens four, generally didynamous, in- cluded or exserted, the filaments inserted below the throat of the corolla, slightly dilated at the base, the anthers bithecous, oblong-sagittate, dorsifixed, the thecae parallel, equal, muticous. Pollen sphe- roidal, porate, coarsely reticulate. Disc inconspic- uous; style slender, subulate, bilobate, the anterior lobe recurved, the posterior lobe generally reduced or obsolete; ovules 2-14 in each locule. Capsule clavate, obovate, oblong-linear or narrowly elliptic in contour, contracted and + stipiform at the base, terete or somewhat compressed above. Seeds 4- 28, compressed and flat, obliquely ovate or subor- bicular, with appressed hairs that extend and turn mucilaginous when wet, subtended by retinacula + acute KEY TO THE SPECIES AND SPECIES GROUPS OF RUELLIA IN SOUTHERN SOUTH AMERICA a oo la. Corollas r 2a Group Flowers | in long-pedunculate, cymose, multiflowered inflorescences subtended A the upper leav basal tube of the corolla frequently of the same length or shorter than the throat Volume 80, Number 4 Ezcurra 799 1993 Ruellia in Southern South America 3a. Corolla lobes 0.5-1 cm diam.; calyx segments oblanceolate ог spathulate. 4 Erect herbs 30-50 cm tall 1. R. elegans Scandent shrubs or subshrubs to 2 m tall . reitzii 3b. Corolla lobes less than 0.5 cm diam.; calyx segments linear or oblong, acute or obtuse but not spathulate 5 Throat notably ventricose; basal tube less than 8 mm long 0 3. R. brevifolia 9b. Throat not ventricose; basal tube more than 10 mm long 6a. Young stems sharply Lig un н асше s Brazil, Paraguay, norther Uruguay, and northeastern Агре 4. R. an њен ла 6b. Stems obtusely quadrangular, de in the angles; Bolivia and northern Argentina 5 ongipedunculata 2b. Flowers solitary and sessile or shortly pedunculate in the axils of the a leaves; basal tube of the corolla generally pa cim than the throat; Bolivia and northern Argentina о 6. R. sanguinea lb. Corollas white, тацу e, or Та. Flowers always i n peduneulate; cymose, multiflowered inflorescences (subsessile and cn only in some rosette plants Group II. Ruellia 8a. oo herbs. Corollas mauve or ш rarely white, with basal tube less than 2 cm long; calyx lobes T than 1 c a. eme + id E flowers in cymes aggregated into a terminal, pyramidal thyrse 7. R. ciliatiflora 10b. Leaves glabrous; flowers in lateral, axillary cy 8. R. coerulea 9b. Corollas dent, with База! tube 2.5-5 cm long; а jen 1:522 em long sets 9. R. macrosolen 8b. Acaulescent rosette plan 11а. Corollas bluish, in e elongate pseudoracemes; pedicels puberulous or paron ela . R. hygrophila llb. Corollas white, generally subsessile and solitary; pedicels hirsute-tomentose _____ | T morongii 7b. Flowers oa rarely pedunculate, pa in the axils of the upper ini. rarely in twos or threes lo nts 12a. а lets present, and as long as or longer than the calyx segm os sharply quadrangles bd = uud oe T owers more than 2 cm long, purple üt l 4b. Flowers less than 2 cm ang) кү . kleinii 13b. 2 obtusely quadrangular or Group IV. йуз 15a. Basal tube of the corolla piis than the l6a. Leaves tomentose with sessile id при with simple hairs ____ 4. R. brachysiphon l6b. Leaves with bus кин shiny edd without glands ___ 15. R. brevicaulis 15b. Basal tube of the corolla the e length as or longer than the throat. та. Leaves 34 stems velutino- tomentose or Pu ONE hirsute ___ ы R. multifolia 17Ь. Leaves e stems glabrou . R. solitaria 12b. i obsolete or absent 18a. Basal tube of the corolla shorter than the throat; placentas not e from t rates seeds (2-)4 oup V. Ebracteolati 19a. Leaves glabrous and coriaceous; nodes and calyx lobes ор. undergroun xylopodium rounded and tuberculate Ec hal 19b. Leaves + pubescent and not coriaceous; nodes and calyx lobes bas white tomentose; xylopodium not markedly rounded and tuberculate 20a. Roots fibrous; leaves elliptic or ovate, sometimes lanceolate; 2-5 cm lon 19. corolla 3-4 cm lon R. pa fina 20b. Roots, fleshy, fusiform; leaves lanceolate, 5-9 cm long; corolla 4-6 cm lon 20. magniflora 18b. Basal tube of the corolla the same н as the throat or longer. Placentas separatin g r from the capsule; seeds 4- 2la. Basal tube of the corolla more than 6 cm long; placentas not hc from the capsule; seeds 6-8 Group VI. kc dues R. nobilis 21Ь. Basal tube of the corolla less than 5 cm long; placentas "rar ^ nh hé retinacula and seeds from the capsule, breaking the walls; seeds 4-6 ________ Croup VII. Chiloblechum 22a. Leaves 2-5 cm long, the ones subtending the flowers subsessile and markedly ciliate: calyx segments unequally cleft, forming a three-lobed structure .... 22. R. erythropus 22b. Leaves 1.5-2 cm long, the ones subtending the flowers glabrous or puber- ulous. Calyx segments 5 and subequally cleft |... 23. R. hypericoides 800 Annals of the Missouri Botanical Garden I. Group Physiruellia Ruellia L. sect. Physiruellia, Lindau in Engl. & Prantl, Nat. Pflanzenfam. 4(3b): 311. 1895. Ruellia L. Lindau ser. Eglandulosae Lindau, loc. cit.: 311. 1895. TYPE: Ruellia acutangula Nees (lectotype, selected here). Stephanophysum Pohl, Pl. Bras. 2: 83. t. 155, 156. 1830-1831. TYPE: Stephanophysum longifolium Pohl (lectotype, selected by Bremekamp & Nannega Bremekamp (1948)). Scordoxylum Nees in Benth., Pl. Hartw.: 236. 846. TYPE: Scordoxylum hartwegianum sect. Physiruellia ees. Arrhostoxylon Nees in C. Martius, Fl. Bras. 9: 57. 1847. TYPE: Arrhostoxylon glabrum Nees (lectotype, selected by Bremekamp & nega Bremekamp (1 ) Stemonacanthus Nees in C. Martius, Fl. Bras. 9: 53. t. 54. 1847. TYPE: Stemonacanthus sal- viaefolius Nees (lectotype, selected by Leon- ard (1951-1958)) Flowers pedicellate, borne in dichasial cymes subtended by the upper leaves, sometimes reduced to solitary flowers always pedicellate and bibrac- teolate. Capsules clavate, narrowly oblanceolate in contour, hollow the upper 25 of the total length, solid and stipitate in the base. Seeds 8-14 Group Physiruellia is restricted to the New World and consists of numerous suffruticose plants that are frequent in the understory of tropical and subtropical forest. In southern South America they are important in floristic provinces of the Amazon region or dominion (see Cabrera & Willink, 1980). It is a large and very diversified assemblage which, on being studied in detail, may be divided into further gruops of closely allied species. Many o the species present handsome ornithophilous flow- ers, frequently red or pink. 1. Ruellia elegans Poiret, 727. 1816. Ruellia formosa Andr. Bot. Re- pos. 10: tab. 610. 1810, non Ruellia for- mosa Humb. & Bonpl. 1805. Arrhostoxylon formosum (Andr.) Nees in C. Martius, Fl. Bras. 9: 62. 1847. Arrhostoxylum elegans (Poiret) Bremek. Verh. Nederl. Akad. Wet. (Tweede Sect.) 45(1): 12. 1948. TYPE: Illus- trated in Andr. Repos. 10: tab. 610 810. Figure 6. Encycl. Suppl. 4: Arrhostoxylon silvaccola Nees var. montanum Nees in C. Martius, Fl. Bras. 9: 60. 1847. Syn. nov. SYNTYPES: Brazil. Minas Gerais: Mariana, Langsdorf . (K), Martinus s.n. (M not seen, photo F earns Soil 104 (B destroyed), 263 (B destroyed; 1 syntypes, K, Erect suffruticose herb 30-50 cm tall, with few, sparsely branched, erect or decumbent stems, qua- drangular and pubescent when young, glabrescent at maturity. Leaves on pilose petioles 0.5-3 cm long, the blade widely ovate, 4-9 cm long and 2- 5 em wide, obtuse and acuminate, widely cuneate on the base and somewhat decurrent on the petiole, the surface + pilose when young, glabrescent at maturity, the margins ciliate. Flowers pedicellate, borne in long-pedunculate, few-flowered, lax, di- chasial cymes, arising from the axils of the upper leaves; bracteoles narrowly oblong-spathulate or linear, less than 1 cm long and 2 mm wide, pu- bescent. Calyx ca. 1.5 cm long, the segments lin- ear-spathulate, pubescent and glandular-ciliate. Corollas red, somewhat bilabiate, the basal tube 2— 4 mm long, the throat curved, narrowly obconical, 2.5-3.5 cm long and 1 cm diam. at the orifice, the two posterior lobes erect, 1.5 cm long, united more than half their length, the three anterior spreading, ca. 1.5 cm diam. Stamens 3-4 cm long, subexserted under the posterior lobes, one pair somewhat shorter, anthers 3 mm long. Capsule elliptic-obovate, ca. 2 cm long and 5 mm thick, hollow nearly to the base, glabrous. Retinacula acute, 2 mm long. Seeds more than 12, flat, subor- bicular, 2-3 mm wide, appressed-pubescent. Distribution. | Southeastern and central Brazil, from Minas Gerais to Paraná. It is found in dis- turbed, more or less open places, such as border of Cerrado forest and mesophilous gallery forests. It flowers in spring and summer. Representative specimens. BRAZIL. Minas Ger- ais: Estrada entre Araxá e Uberaba, km 381, em cerrado, 100 m.s.m., Shepherd 7220 (MBM). Paraná: Mun. Cerro Azul, Cerro Azul, Hatschbach 6401 (MBM, US). Ruellia elegans is morphologically similar to some species of group КиеШа in the structure of the capsule and the herbaceous habit, but the calyx and corolla morphology place it close to Ruellia sceptrum-marianum (Vell. Conc.) Stearn, R. acu- tangula (Nees) Lindau, R. silvaccola (Nees) Lin- dau, and other related species of group Physi- ruellia. The similarities in vegetative and fruit characters to species of group Ruellia may just be a case of convergence within the evolution of group Physiruellia. It is a beautiful plant with large red flowers, sometimes cultivated as an ornamental. Human activities may have helped to extend its natural range. Volume 80, 1 993 Number 4 Ezcurra 801 Ruellia in Southern South America Hatschba ch 6 Ruellia elegans Poiret.—A. Flowering branch.—B. Immature capsule without seeds, schematic. (A 401; B, Shepherd 7220.) , 802 Annals of the Missouri Botanical Garden 2. Ruellia reitzii Wasshausen & Smith in Reitz, Ruellia ventricosa HBK, Nov. Gen. Sp. 2: 241. 1817, Fl. Illustr. Catarinense АСАМ: 60. 1969. TYPE: Brazil. Santa Catarina: Luis Alves, Rio Canoas, 18 Jan. 1953, Reitz 5154 (holotype, US). Scandent or straggling shrub to 2 m tall, with sparsely branched stems, the younger portions markedly quadrangular, glabrous or pubescent, the older portions subterete and inflated over the nodes, becoming woody. Leaves on petioles 0.7-1.5 cm long, the blades ovate or elliptic, 3-7(-15) ст long and 1.5-3(-6) cm wide, acute or acuminate, cu- neate at the base and somewhat decurrent on the petiole, irregularly and shallowly crenate, the sur- face glabrous to lightly velutino-puberulous, with trichomes mostly eglandular. Flowers pedicellate, in pedunculate cymes borne in the axils of the terminal leaves; peduncles 3-7 cm long; compound dichasia unevenly ramified (anisocladous), gener- ally longer than the subtending leaves, frequently with supernumerary buds in the ramifications; bracteoles obovate, spathulate to oblanceolate, small, the lower ones generally foliaceous, more than 1 cm long. Calyx segments oblanceolate or spathulate, 12-14 mm long and 4-5 mm wide, rounded, puberulous. Corolla red, ca. 5 cm long, curved, the basal tube 1.5 cm long and 2 mm wide, opening into a cylindrical throat 1-2 cm long and mm wide, the lobes semi-erect, rounded, 2 cm long and 1.5 cm wide, emarginate. Stamens exserted with anthers 4 mm long, laterally connate. Ovary pubescent; style exserted. Capsule unknown. Distribution. Southern Brazil, in the state of Santa Catarina. In flower in January. e specimens. BRAZIL. Santa Cata- rina: cr rafim, Margem de cascata, Reitz C 2009 (HBR, Т Ruellia reitzii is characterized by the large red flowers in multiflowered dichasial cymes and the spathulate or oblanceolate, pubescent calyx lobes. n these characters it shows close affinities with the group of Ruellia acutangula (Nees) Lindau, R. sceptrum-marianum (Vell. Conc.) Stearn, and related species of central and southeastern Brazil. Ruellia reitzii occurs more to the south and differs from these ana principally by the weaker or scandent habi 3. Ruellia brevifolia (Pohl) Ezcurra, Darwin- iana 29: 278. 1989. Stephanophysum brevi- folium Pohl, Pl. Bras. Icon. Descr. 2: 84, tab. 155. 1831. TYPE: Brazil. Rio de Janeiro: Rio de Janeiro, Schott s.n. (holotype, W not seen, photo SI). Figure 7. iret, 1804 Er а ventricosum (HBK) Nees in DC., Prodr. 11: 20€ ab TYPE: Peru. “Prov. Jaen de Bracamoros, cla ae prope d Humboldt & Bonpland s.n. (holotype, B destroyed, photo F 18410, isotype, P). Syn. no Siephanophysum E Pohl, Pl. Bras. Icon. Desc r. 85, tab. Ruellia longifolia (Pohl) Griseb., "Abh. ond = Wiss. Goettingen : . 1879, not Richard, 1782. Ruellia graecizans се Brittonia 3: 1938. ТҮРЕ: Brazil. Without locality: Pohl s.n. (syntype, W not seen, photo SI). Echinacanthus die er Kuntze, Revis. Gen. Pl. 2: ane 18 ot e dichotoma Sessé & Мос., 889. T TYPE: ин Bogor, Buitenzorg, Kuntze 1875 поен. NY). ји serratitheca Rusby, Mem. New York Вог. Gard. 362. 1927. TYPE: Bolivia. Without departamento: = RIDE. 4,500 ft., 28 July 1921, Rusby 363 (holotype, NY) Е ии ѕит ои Bremek., Proc. Kon. Ned. Akad. Wetens is 4. 1969. TYPE: Bolivia. Withou Ba rtolo, 5,000 BL. id кыр 1949, Brooke i Шой U). Syn. по Suffruticose herb or subshrub to 1 m tall, with numerous erect stems, sparsely branched, obtusely subquadrangular when young, glabrous. Leaves on petioles 1-5 cm long, the blades ovate, 4-1: long and 1.5-8 cm wide, acuminate at the apex, cuneate at the base, entire or shallowly crenate, the surface generally glabrous, sometimes pubes- cent. Flowers pedicellate, borne in axillary, lax, long-pedunculate, multiflowered cymes of com- pound dichasia; supernumerary buds frequent in the ramifications. Bracteoles ovate or lanceolate, foliaceous, bigger than the calyx lobes. Calyx 5- 10 mm long, the segments linear and acute, less than 1 mm wide, puberulous. Corollas curved, scar- let, the basal tube 0.5-1 cm long, opening into an anteriorly inflated, ventricose throat 1-1.5 cm long and 0.5-1 ст wide, narrower toward the orifice, the lobes suberect, small, rounded, 2-3 mm long, retuse. Stamens subexserted with filaments ca. 9 mm long, and anthers 3 mm long, laterally con- nate, basally apiculate. Ovary puberulous; style exserted. Capsule clavate, 1-1.5 cm long and 3 m thick, broadest above the middle, solid-stipitate to 3 mm high, glabrous. Seeds 8-12, suborbicular, 2 mm diam., with hygroscopic-viscid pilose mar- gins; retinacula 2 mm long. Distribution. Tropical and subtropical South America, from Colombia, Ecuador, Peru, and Bo- livia to southern Brazil, Paraguay, and northern Argentina. It is found in the understory of open places of the forest, such as clearings, borders of roads, and water courses, from sea level to 1,500 m altitude. It prefers disturbed, humid soil; it flow- Volume 80, Number 4 1993 Ezcurra Ruellia in Southern South America 803 FIGURE 7. Ruellia brevifolia (Pohl) Ezcurra. — A. Flowering branch. — B. Flower. —C. Open corolla. — D. Stylar : ч apex.— E. Capsule. — Е. Cabrera 28048. F, Zuloaga 1629.) ers in autumn and winter, from March to Septem- ber. Representative specimens. ARGENTINA. Jujuy: Yuto, El Bananal, Cabrera 28048, 33941 (SI), Fabris 4541, 5045 (LP); Unión del Rio Pantanoso y el Piedras, capsule without seeds, schematic. —G. Dehiscent capsule. — Н. Seed. (A-E, G, H, ludica & Ramadori 367 (51); Arroyo Zanjón Seco, Zu- loaga 2501 (SI); Campamento Caimancito de Yacimien- tos Petroliferos Fiscales, Zuloaga 2534 (SI). Misiones: Cataratas del Iguazü, Schulz 16192 (US). Salta: Rio Pescado, Borsini 580 (LIL); Сагарап, Bridarolli 3256, 3261 (LP); Agua Blanca, Cabrera 26534 (SI); Urundel, Annals of the Missouri Botanical Garden Arroyo El Diablo, Capurro 238 (BA); Pocitos, Krapo- vickas 19415 (CTES); Vado Hondo, Legname & Cuezzo 7045 (LIL, LP); Rio Piedras, Rodriguez 62 (BA, LP, SI); Rio Tartagal, Schreiter 3367 (LIL, SI, US), 70862 (LIL, NY, SI), Venturi 10758 (SI); Camino a Finca Ronda, Rio Santa Maria a 15 is de Orán, Zuloaga 1629 (SI); Camino de Ballivian a Río Seco, Zuloaga 2685 (SI). uc Capital, Rio Sali frente al Boulevard Sar- sinu Venturi 3821 (BA, US), 3957 (SI, US). Entre Rios: Concordia, Spegazzini, VI- 1949 (BAB). BRAZIL. Parana: Iguaçu Falls, Davis & Shepherd 60927 (MBM), Morretes, Dusén 14022 (GH, MC SI, US), АЫ de h 116 М); Сегго iaon Набавна 2 2326 (M JS); Antonina, estrada Cacatu-Serra Negra, Hatsc de Dac Rio Cachoeira, Hatschbach 34410 (MBM, US); Mun. Jacarezinho, Riacho da Fartura, Kummrow 525 greens Mun. Terra ed Rio Ivaí, Lindeman 626 (MBM, US). Rio Grande gelbey, Schultz 397 (ICN). 6 Catarina: Ibirama, capoeira, Gevieski 58 (НВК, RB); Brusque, Azambuja, Klein 2775 (HBR, 05); Florianopolis, y de Santa Ca- tarina, morro costa da lagoa, Klein 7431 (HBR, R, US); Porto Belo, Caixa d'Aco, Saepe Tres (HBR, NY, RB, US). PARAGUAY. Alto md Berton 4373 (LIL); Ruta 7, Km А. 252. Boelcke 13 ШАБ, о 60 km al N de Hernandarias, Fer ries z Casas 7 (С, MO); 35 km 2 "n Hernandarias, Hahn 925 А US); Ruta 2, km 252, Krapovickas & Cristóbal 13407 (SI, US); Itabo, ridus d'Itaipu, 70 km N de Pto. dil Billiet & Jadin 3478 (BM, NY, MO). m : Yerbales, Sierra de Maracayu, Hassler 4508 , P). — Cane (BM, MO Ruellia brevifolia is characterized by the small scarlet, ventricose flowers with inflated corollas in multiflowered cymes. It is one of the most widely distributed species in South America, ranging from Colombia to northern Argentina. It shows a large amount of variation over its area, sometimes pre- senting pubescent leaves or larger-flowered popu- lations. It is used as an ornamental in tropical regions all over the world, where it often becomes naturalized (Ariza Espinar, 1983; Backer & Bak- штеп van den Brink, 1965; Barker, 1986). Its cultivation in South America may have helped to extend its natural range through garden escapes and establishment in new areas (Ezcurra, 1989). Ruellia amoena Nees and Cyrthacanthus cor- ymbosus Martius ex Nees are nomina nuda that have been used for this species in herbaria (Nees, 1847b 4. Ruellia angustiflora (Nees) Lindau ex Ram- bo, Iheringia, Bot. 12: 23. 1 Es pi dp M Noes in с. Martius, Fl. Bras. 7: 51. 1847. SYNTYPES: Brazil. Rio Grande do Sul: Porto Aene Isabelle s.n. (K); Sellow s.n. (B destroyed, photo F 5909; iso- syntype, K); Joannes de S. Barbara s.n. (not seen); Tweedie s.n. (K). Figure 8. . Stephano- Erect or somewhat scandent shrub to 2 m tall, with sparsely branched stems, the younger portions sharply quadrangular, glabrous, the older portions subterete and inflated over the nodes, becoming woody. Leaves on petioles 0.3-1.5(-4) cm long, 3-10(-15) ст long and 1.5-4(-6) cm wide, acute or acuminate, cu- the blades ovate or elliptic, neate at the base and somewhat decurrent on the petiole, irregularly and shallowly crenate, the sur- face glabrous to lightly velutino-puberulous, with trichomes mostly eglandular. Flowers pedicellate, in pedunculate cymes borne alternatively on one of each pair (rarely on both) of the terminal leaves; peduncles 3-7 mm long; compound dichasia un- evenly ramified (anisocladous), generally longer than the subtending leaves, frequently with supernu- merary buds in the ramifications; bracteoles ob- ovate to oblanceolate, small, the lower ones gen- erally foliaceous, more than 1 cm long. Calyx segments linear-lanceolate, 4-12 mm long and 1– mm wide, acute, puberulous. Corolla red, 3-4(- 6) ст long, angled, the basal tube 1-2 cm long, opening into a cylindrical throat 1-2 cm long and 8 mm wide, the lobes semi-erect, rounded, 5 mm long, emarginate. Stamens subexserted with fila- ments 1-1.5 cm long, and anthers 4 mm long, laterally connate, obtuse, and with glandular con- nective. Ovary glabrous; style exserted. Capsule clavate, 1.5-2 cm long and 4 mm thick, glabrous, broadest above the middle, the base solid to 3-5 mm high. Seeds 8-12, dark brown, 2 mm diam., with pilose margins, hygroscopic-viscid when wet; retinacula 2 mm lon .* Vernacular names. “Flor де fogo,” in Santa Catarina, Brazil (Wasshausen & Smith, 1969). Distribution. guay, eastern Paraguay, and northeastern Argen- tina, in the hygrophilous subtropical forests of the Southern Brazil, northern Uru- basins of the Rivers Paraná and Uruguay (Parana- ense phytogeographical province). It is found in humid places, such as borders of streams, margins of rivers, and near waterfalls, in disturbed, open and sunny locations of the forest. It flowers in winter, spring, and summer, especially from July to February Representative specimens. ARGENTINA. Corri- entes: Depto. San Martín, Tres Cerros, Cabrera 27695, 28152 (SI); Depto. Ituzaingó, Rapidos de Apipé, Cabrera 28969 (SI); San Cosme, Cristóbal 191 (LIL); Paso de la Patria, Cuezzo 11297 (LIL); Depto. San Martin, ч Caimán, Ibarrola 4203 (LIL); Depto. Santo Tomé, Arroy C bimar. sane ickas 21097, 26163 (CTES, US); Gen. eral Paz, 12 km de Ita Ibaté, costa Md Paraná, Mroginski 674 (CTES, D Depto. Mburucuyá, Estancia Santa Te- resa, Pedersen 1305 (LP, US); Berón de Astrada, Ruiz- Volume 80, Number 4 805 Ezcurra 1993 Ruellia in Southern South America E 8. Ruellia angustiflora (Nees) Lindau ex Rambo. — A. Flowering branch. — B. Open capsule without seeds, schematic. (A, Montes 15429; B, Cabrera 28969.) 806 Annals of th Missouri Botanical Garden Huidobro 2249 (NY, SI); Depto. Mercedes, Frente a 05); Depto. Saladas, P (LIL); Depto. Concepción, 11 km Tressens 874 (C ntre Rios: Federación, Mar- tinez Crovetto y Piccinini 4694 (BAB, CTES, US). For- ilcomayo, Clorinda, Cezzanni s.n. (LP). cataratas, Cabrera 184 (LP); 356 (LP), Depto. San Pedro San Pedro, Cabrera 28825 (SI); Depto. Apóstoles, San José, Escuela Agrotécnica Pascual idl Cabrera 28483, 29260 (SI); Depto. Eldorado, Ruta 12, Arroyo Piray Guazü, Crisci 269 (LP); Depto. General Belgrano, Brain de rigo yen, Hunziker 940 (51); Ignacio, Teyucuaré, 6 km de San Ignacio, Koronas 2327 (LP); Depto. Leandro Al i as e Cristóbal 15939 (BA, CTES, LP, NY, RB, US), o. San Javier, San Javier, Krapovickas & Cristóbal 28892 us Depto. San Pedro, Laharrague, Montes 15429 (SI). B Mato Grosso do Sul: Mun. Rio Bril- hante, iid Bela Vista, Hatschbach 26079 (MBM, y Mun. Anastacio, Rio Aquidauana, Hatschbach 24605 (MBM, US). Parana: Mun. Tibagi, Fazenda Mon- te Alegre, Hatschbach 2852 (MBM, RB), 7141 (MBM, US); Foz do Iguaçu, Cataratas, Hatschbach 3900 (MBM); un. Cianorte, Fazenda Lagoa, Hatschbach 16953 (MBM, US); Tapejara, Hatsc bach 17003 (MBM, US); Rio Branco do Sul, Ribeirinha, Hatschbach 18079 (MBM, 05); Porecatu, Hatschbach 23475 (MBM, NY), 23485 (MBM , US); Matelandia, Santa "oe Hatschbach 34526 (МВМ, US). Rio Grande do 3 km S de Cruz Alta, Krapovic kas & Vanni 37067 CTES, US); 13 km SE de Santa Rosa, Lindeman 8962 (CTES, ICN); Porto Alegre, Tristeza, Palacios & Cuezzo 500 (LIL); Cacapava do Sul, Palacios & Cuezzo 1410(LIL); 13 km de Carazin- ho a Sta. Barbara, Quarín 115 53 (€ TES); Sta. Clara | prope E “= 5 , Rambo 29483, 38859, 42095, 42635 (all LIL), 43937 (LIL, US), 44112 (LIL); Porta Alegre, Rambo 26628 (US), 37803 (LIL); Prope Novo Hamburgo, Rambo 42894 ( i Rambo 43839 (LIL, US); Mato Leitao prope Venancio Aires, Rambo 49489 (LIL); Ita- ruguay, Rambo 53675 (РАСА); Ре- lein 5149 (НВК); Chapeco, Capoeira, 300 m, Klein 5310, 5602 (HBR, US); Sào Domingo, Coxilha Seca, Marata, Klein Mos bns US); L Socorro, 800 m, Reitz 6555 (HBR, RB, US); Capinzal, orla da mata, - ES ен 16238 (HBR, US); Abelardo 950 m id Luz, beira do eitz & in 16616 (HBR US); Ibirama, ao У lon ngo do Rio Itajai, Smith, Klein & Gevieski 7612 (HBR, d 9 km W de Campo Ere, 900- e 1,000 m, Smith & Klein 11543 pu NY, RB, 05); Dionisio Cerqueira, 41 km ao S, Smith & Klein 11710 (HBR, RB, US); ps Smith & Klein 13128 (HBR, R, US). PARA AGUAY. Amambay: Zwischen Río Apa und Rio Aquidaban, Fiebrig 4190 (BM, GH, K); Cursus superioris fluminis Apa, Hassler 8089, 8089a (BM, G, K, MO, NY, P); Pedro Juan Caballero, Mizoguchi 462 ; Parque Nacional Cerro Corá, margen del Rio Aquidabán, Zardini 4207 (MO). Alto Paraná: Santa Teresa, Bertoni 1690, 4280, 4419 (all yop in е fluminis Alto Paraná, Fiebrig 5370 (ВМ, С, СН, K); Ruta 2, km A 252, Krapovickas & Cristóbal 13409 (CTES); Puerto Stroessner, Centre Forestier, Stutz 166, 196, 325 (all NY). Caaguazu: Parque Guayaki, Schulz 16203 (BAB, CTES, Caazapa: Tavai, Propiedad e Bogado, Basualdo 1933 (MO), Ortiz 957 (G). endiyu: Yerbales, Sierra de Maracayú, Hassler 4127 (BM. GH, MO), 4319 (BM, P). Central: E Ypa- caray, Hassler 11744 (BM, GH, K, MO, NY); Yaguarón, Cerro Corá, Krapovickas 12327 (CTES). Cordillera, Cordillera de Altos, Fiebrig 104 (BM, G, GH, K); Va lenzuela, Mboraya-guazu-ty, Sparre & Vervoorst 1310 (LIL); Cerro Zanja Jhú, 1 km E of road from Route 1 to Atyra, 3 km from Atyra, Zardini 5075, 6237 (MO). Guaira: Villa Rica, Balansa 2457a (P); Cordillera de Villa Rica, Hassler 8824 (BM, G, GH, NY, P); Road Melgarejo-Antena, 6 km S of ee es 11188 (MO). Itapua: Arroyo San Rafael, Rut km SE de General Delgado, Arbo 2023 (CTES): Бата alrededores, Bertoni 5566 (LIL); El Tirol, pequena aguari: Paraguari, ERE 2457 (K, Р), Fiebrig 906 (G, G CH Acahy, Billiet & Jadin 3003 (BM); Entre Carapag uá y Acahy, bosque próximo a Riachuelo, Fer- nández Casas 3518 (NY) Base del Cerro Yaguarón, Schinini 4194 (CTES, С, SI; Compania Costa Palacios, Zardini 3100 (G, MO), 5621, 6424, e 30, 6490 (all MO); Cerro Mbatovi, Zardini 4467 ( cizo Acahay, forest on E peak, Zardini 5271, 5691 (MO). San Pedro: Tapiraguay, Hassler rad! 4318, 4319 (all K, P, NY). URUGUAY: Artigas: San уху Киза, Del Puerto 2053 (M VEA). Rivera: Ruta 30, Subida de Pena, Sierra de la Aurora, Rosengurtt B-7 136 (MVFA) Ruellia angustiflora is characterized by the markedly quadrangular stems and the bright red tubular flowers in multiflowered axillary cymes. It is frequent in the understory of gallery forests of southern Brazil, eastern Paraguay, northeastern Argentina, and northern Uruguay, and had been erroneously identified by many authors (e.g., Lin- au, 4; Wasshausen & Smith, 1969; Dawson, 1979) as Ruellia sanguinea, a species common in southern Bolivia and northern Argentina (see Ezcurra, 1989). It is often cultivated because of its handsome, red flowers and easy propagation. The isolated localities of the southernmost portion of its range (e.g., Entre Rios, Federación) are prob- ably the product of garden escapes. 5. Ruellia longipedunculata ане Bull. Herb. Boissier 3: 365. 1895. TYPE: Bolivia. Santa Cruz: Santa Cruz, 2500-1 008 m, Kuntze, May 1892 (holotype, B destroyed, photo F 18213; isotypes, NY, US). Figure 9. Stephanophysum brookeae Bremek., Proc. Kon. Ned. Akad. Wetensch. Ser. С. 72, 4: 424. 1969. TYPE: Sep. iid e 5577 (holotype, BM; isotype, NY). Syn. Volume 80, Number 4 Ezcurra 807 1993 Ruellia in Southern South America 5 mm ч. REO. Ruellia longipedunculata Lindau. — A, Flowering branch. — В. Open capsule without seeds, schematic. (A, Cabrera 26526; B, Cabrera 32782.) 808 Annals of the Missouri Botanical Garden Stout suffruticose herb to 2 m high, the stems ascending, obtusely quadrangular, slightly hirsute- pilose when young, glabrescent at maturity. Leaves on petioles 0.5-2.5(-4) cm long, the blades ovate, 25-10(-15) cm long and 2-6(-8) cm wide, acu- minate, obtuse or rounded at the base, irregularly and very shallowly crenate or dentate, the surface glabrous to lightly puberulous. Flowers pedicellate, in dichasia forming lax, long-pedunculate cymes, borne on both pairs of the terminal leaves, generally longer than the subtending leaves, frequently with supernumerary buds in the ramifications; bracte- oles spathulate to oblanceolate, approximately 1.5 cm long and | mm wide, the lower ones generally larger and foliaceous. Calyx ca. 2 cm long, the segments oblong-lanceolate, 12-15 mm long and 2—3 mm wide, acute, glabrous, the superior some- what larger. Corolla red, 2.5-3 cm long, the basal tube ca. 1.3 cm long, opening into a subcylindrical throat 1.5 cm long and 4—5 mm wide, the lobes semi-erect, rounded, 3-4 mm long. Stamens sub- exserted with filaments 13-15 mm long, and an- thers 4 mm long. Ovary puberulous; style exserted. Capsule clavate, 1.5 cm long and 3 mm thick, glabrous, broadest above the middle, the base solid to 5 mm high. Seeds 6-8, dark brown, 2 mm diam., with pilose margins, hygroscopic-viscid when wetted; retinacula 2 mm lon Distribution. | Southeastern Bolivia and north- ern Argentina, in low mountain forests of the Yun- gas phytogeographical province, near transition to the more xerophilous Chaco forest. It is found in the underwood, in openings and borders of paths, rivers, and streams. It flowers in spring, from Sep- tember to December, and could be used as an ornamental. Representative specimens. ARGENTINA. Jujuy: Camino a Valle Grande, Cabrera 27816 (LP, SI); Serranía i Calilegua, El Cebilar, Fabris 4436 (LP); Mesada de s Colmenas, /udica & Ramadori 5 (SI); Camino de la "us del Rio Zora a Cafetales, eris 2903 (SI). Salta: Agua Blanca, Cabrera 26526 (LP, SI); Depto. Chicoana, Los Laureles, Cabrera 32782 (SI); Quebrada de Escoip Zardini 1205 (LP); Ruta Nac. 34, Dique Itiyuro, Zu: loaga 2728 (SI) Ruellia longipedunculata is characterized by the long, somewhat pendulous, multiflowered inflo- rescences, the wide calyx lobes, and the narrow red corollas. Specimens that are morphologically intermediate with Ruellia brevifolia in these char- acters suggest natural hybridization in areas where these species are sympatric (Ezcurra, 1989). за. Ruellia longipedunculata Lindau x Ruellia brevifolia (Pohl) Ezcurra. ] 3 4 Plants similar to Ruellia ulata, but differing by the shorter and narrower calyx lobes and the shorter and somewhat inflated corolla throats. Representative specimens. Pies Salta: camino a la finca e E a, aprox. 5 km del Puente irm anca, Cuezzo 1 dcm т due Ruta Nac. 3 Dique Itiyuro, о 2730 (SI). These specimens, which аге morphologically in- termediate between Ruellia longipedunculata and R. brevifolia, also show reduced pollen fertility, which suggests hybridization (Ezcurra, 1989) 6. Ruellia sanguinea Griseb., Abh. Кошо]. Ges. Wiss. Gottingen 24: 260. 1879. TYPE: Ar- gentina. Salta: Огап, prope San Andrés, Rio Seco bei San Andrés, 17-24 Sep. 1873, Lo- rentz & Hieronymus 269 (holotype, GOET not seen, photo SI, US; isotype, B destroyed, photo F 18209). Figure 10. Ruellia kuntz zei Lindau, Bull. Herb. Boissier 3: : 189: rrhostoxylum Е zei (Lindau) Bremek. A b Ser. C, 72, 4 424. 1969. TyPE: Bolivia. With ut locality: 2,000 m, Kuntze, Apr. 1892 (holotype, Kasra. photo F 18209; isotype, NY). Suffruticose subshrub to 1.5 m high, the stems ascending, markedly quadrangular, slightly pubes- cent when young, terete and glabrescent at ma- turity. Leaves on petioles 2-8 cm long, the blades ovate, 6-16 cm long and 4-8 cm wide, acuminate and cuspidate, obtuse and somewhat decurrent on the petiole at the base, smooth or irregularly and very shallowly crenate or dentate, the surface gla- brous to slightly puberulous. Flowers subsessile or very shortly pedunculate, solitary in the axils of reduced upper leaves, frequently with supernu- merary buds in the axils of the subtending leaves; bracteoles narrowly oblanceolate or lanceolate, ap- proximately 5 mm long and 1 mm wide. Calyx mm long, the segments linear, 5-7 mm long wide, acute, ciliate. Corolla red, 3-5 cm long, puberulous, the basal tube thin and curved, and ] mm 1.5-3 ст long, opening into an abconic throat 1.5-2 cm long and 5-8 mm wide, the lobes semi- erect, rounded, 3-5 mm long, somewhat retuse. Stamens subexserted with filaments 5-7 mm long, and anthers 3.5 mm long. Ovary puberulous; style exserted. Capsule clavate, 1.5 cm long and 3-4 mm thick, glabrous, broadest above the middle, the base solid to 5 mm high. Seeds 8-10, dark brown, 2-3 mm diam., with pilose margins, hy- groscopic-viscid when wet; retinacula 2 mm long. Volume 80, Number 4 Ezcurra Ruellia in Southern South America 809 Я 2 - 9 2 MMC ~ s 2 PX YEAR AA 5 mm FIGURE 10. Ruellia sanguinea Griseb.— A. Flowering branch. — B. Open capsule without seeds, schematic. (A, Fabris 4491; B, ludica & Ramadori 355.) 810 Annals of the Missouri Botanical Garden Distribution. Southern Bolivia and northern & Willink, 1980). Many South American species Argentina, in the north of Salta and east of Jujuy. It is found growing in the shade of mountain forests of the Yungas phytogeographical province, be- tween 600 and 2,000 m altitude. It flowers in spring and autumn, especially in October and May. Representative specimens. ARGENTINA. Jujuy: Serrania de Calilegua, Fabris 4478, 4491 (LP); Rio de Las Piedras, ludica & Ramadori 355 (SI). Salta: Depto. Oran, Rio Santa Maria, las Juntas, Castellanos 29 (BA, LIL); Parque Nacional Baritú, Arroyo Sidras, Brown 2003 (SI); 8 km antes de Porongal, Cuezzo 9213 (LIL) Ruellia sanguinea is characterized by the small red corollas with a long basal tube. Though the semi-erect, small lobes and red color of the flowers suggest hummingbird pollination, the narrow, long, basal tube of the corollas and the morphology of the inflorescence separate this taxon from the other red-flowered species in the area. In these charac- ters and in leaf morphology it is quite similar to the white-flowered, probably butterfly-pollinated, Ruellia puri Nees, widely distributed throughout central Brazil and western Bolivia. This suggests that the different pollination syndromes in Ruellia probably arose many times as parallelisms in the evolution of different lineages. This species was originally described from north- ern Argentina (Salta), its name referring to the blood-colored flowers it produces. This name was erroneously used later by most modern authors for Ruellia angustiflora, a different red-flowered spe- cies, frequent in eastern Paraguay, northeastern Argentina, southern Brazil, and northern Uruguay (Ezcurra, 1989) II. Group Ruellia Ruellia L. sect. Euruellia Lindau in Engl. & picis at. anzenfam. 4(3b): 311. 1895. Tv R. tuberosa Cryphiacanthus Nees, Index Sem. Hort. Vratisl. 1841; Linnaea 16: 298. 1842. TYPE: C. bar- badensis Nees, nom. illeg. (= R. tuberosa L.). Flowers pedicellate, borne in dichasial cymes subtended by the upper leaves, rarely reduced to solitary flowers, always pedicellate and bibracteo- late. Capsules terete or narrowly elliptic in contour, hollow nearly to the base. Seeds 14-28. Group Ruellia is important in the New World and consists of several weedy, erect or rosettiform herbs that are common in open, low, disturbed, subtropical and tropical areas. In southern South America they are principally found in floristic prov- inces of the Chaco region or dominion (see Cabrera of this group are very similar to species in Central and North America, and could be closely related. The mucilaginous viscid pubescence of the seeds when wetted allows them to be easily carried by migrating birds. This could explain the long-dis- tance disjunctions of “vicariant”” species between the Northern and Southern hemispheres. 7. Ruellia ciliatiflora Hook., Bot. Mag. 66: tab. 3718. 1840. Arrhostoxylum ciliatiflo- rum (Hook.) Nees, DC. Prodr. 11: 216. 1847. TYPE: Fig. in tab. 3718 of Bot. Mag. 66. 1840. Figure 11 Ruellia lorentziana Griseb., Abh. Kónigl. Ges. Wiss. Göttingen 24: 259. 1879. pa Argentina. Tu- cumán: Tucumán, 22 Dec. 1872 Lorentz & Hi eronymus а (GOET not seen, photo SI); Tucu- mán, Lorentz & Hieronymus, 6-22 Dec. 1872 (GOET not seen, photo SI). Perennial herb 30-80(-100) cm tall, sparsely branched from a subligneous base, with clustered, tough, fibrous roots. Stems erect, obtusely qua- drangular, with short glandulous pubescence in- termingled with longer simple hairs, glabrescent at maturity. Leaves on pilose petioles 0.5-2.: long, the blade widely ovate, 4-13 cm long and m wide, obtuse and mucronate, widely cu- neate to truncate on the base and somewhat de- current on the petiole, markedly pubescent on the nerves, the surface + pilose and sometimes glu- tinose, the margin irregularly and shallowly cre- nate. Flowers pedicellate, borne in dichasial cymes in the axils of small bracts or reduced apical leaves, aggregated in an ample, terminal pyramidal thyrse, sometimes enriched by lateral thyrses originating on the axils of the upper leaves, the whole ap- pearing paniculate. Bracts foliaceous; bractlets lan- ceolate, small, and pubescent. Calyx ca. 10 mm long, the segments linear, to 8 mm long, glandu- lous-pubescent. Chasmogamous flowers tilted, blue, 3.5-4.5 cm long, the basal tube angled on top, 1- 2 cm long, the throat 1.5-2 cm long and 1 cm diam., the lobes spreading, obovate, ca. 1 cm long and 1 cm wide, the margin undulate and sometimes ciliate, the anterior lobes with dark palate markings toward the orifice. Stamens included, long filaments ca. 6 mm long, short 4 mm long, the anthers 2 mm long. Ovary and style puberulous. Cleistoga- mous flowers frequent, reduced, tubular, and closed. d 3-5 mm thick, hollow nearly to the base, the surface puberulous. Retinacula 2 mm long; seeds 16-24, pale-colored, mm diam., appressed puberulous when dry and Capsule elliptic, 2-3 cm long an with mucilaginous extended hairs when wet. Volume 80, Number 4 Ezcurra 811 1993 Ruellia in Southern South America LAT MSS un EP ү NA ZEN 9 P а ter © Ruellia t Hook. — A. Flowering branch. — B. Leaf abaxial surface, showing glandular and eglandular haie and cystolitl . Portion of inflorescence, in fruit. — D. Corolla dorsally open, showing stamens E. Open capsule without Mons Nb uen Dehiscent capsule, with seed. — С. Seed. (A-D, F, С, Cabrera 30998: E, Kiesling 5566.) 812 Annals of the Missouri Botanical Garden Distribution. Colombia to Bolivia, Paraguay, Choya y iron La Punta, Piccinini & Petetín 3375 southwestern Brazil, and northern Argentina, es- pecially in Chaco regions. Auellia ciliatiflora is commonly found growing as a weed in moist, open, disturbed places, such as roadbanks, cultivated or abandoned fields, borders of water courses, dry river beds, clearings of forests, and waste places. It is sometimes cultivated as an ornamental, and prefers sunny locations and sandy soil. It flowers in spring and summer, from October to April, and frequently presents cleistogamous flowers. Representative specimens. ARGENTINA. Cata- marca: Depto. La Paz, Olta, Brizuela 700 (US); Depto. Ancasti, Río Chico, Brizuela 777 (LIL); Depto. Sta. Rosa, Alijilan, Pierotti 11561 (LIL), Depto. Valle Viejo, San Isidro, Rojas Paz 37 (LIL). Chaco: Presidente Roque Saenz Peña, Buratovich 469 (LIL, US); Nueva Pompeya, Flosdorff, 1 Nov. 1906 (ВАВ); Taco Pozo, Meyer 2045 (LIL); Colonia Benitez, Schulz 2502 (CTES), 9783 (CTES, 05); Las Brenas, Schulz 10812 (CTES); Vilelas, camino Paranacito, Schulz 16431 (CTES); Resistencia, de la Sota 904 (LIL). Corrientes: Capital, Augusto 1 (€ 'TES), Burkart 68896 (SI, Marunak 147 (LP, SPF), Meyer 6604 (LIL). Formosa: Ing. Juárez, orilla 5 del pueblo, Arenas 2317 (CTES, SI, US); El Porteno, | : 38 AW: Ruta 90 al S de Pirané, oe 662 (SI hi, Jorgensen 2845 (BA, LIL, S ; Depto. Patifio, a ape del Rio Pilcomayo, Krapov ches 1219(L IL, LP, SI); Pirané, Morel 129 (LIL); 3 km al SE de me ерды, Morel 7224, 7291 (LIL, US); 36 km W de Palo Sant Petetin 1650 (BAB): Herradura, costa del Rio PUN. о Eras (ВАВ, SI); Comandante Fontana, Schulz 17023 ES); Nuevo Porteño, Schulz 17979 (CTES). stación Margarita Verón, Araque & Barkley 199947489 (LIL); La Mendieta, Cabrera 16396, 21166 3662 ; Zapla, Fabris 8275 (LP); Entre Fraile Pintado y Ledesma, Legname & Cuezzo 5982 (LIL ; De El Fuerte a Palma Sola, Kiesling 5566 (SI); Cerros de San Lucas, Kiesling 5679 (SI); Laguna San Miguel, ле 2825 (51). La Rioja: La Rioja, Esteban 22 (BAB). Salta: Carapari, Abiatti & poe g 337 (LP); Coronel Cornejo, Abiatti & Claps 9 nel Moldes, Bartlett 19643 E e ssl Birabén 1409 (LP); Saucelito; sobre Rut , Cuezzo Т Ciudad, Jorge p 299 (SI); & а 2 (LIL); General Ballivián, LP); а Lozano, Malvárez 435 (LIL); Cerrillos, 5 L); Estancia San Jorge, Pedersen 10820 (CTES); Tobantirenda, Pierotti 7355 (LIL, US); J. V. González, Finca San Javier, d Toledo 119 (LIL); Quebradas del Rio Tor o Blanco, Vattuone 145 (SI); Candelaria, Estación Ruiz de los Llanos, Schreiter 5796 (LIL); Ruta 34, 10 km de Campichuelo hacia Em- bari longa 2649 (SI, US). Santiago del = 0: Yut u, Legname 77 (LIL); Depto. Ojo de Agu Po: 220 Cavado. Maldonado 933 (LP); Capital, Vuelta ЈЕ la Barranca, Meyer 14074 T Entre el Mojón y Cerro del Remate, Molina 1800 (BAB); La Banda, O'Donnell 4259 (LIL, 05); Taco Pozo, Peirano 8998 (LIL); Entre pto. Copo, Los Tigres, Roic 564 (BAB); Ciudad, Ulibarri 998 (SI); Depto. Quebrachos, Para Yacu - barri 1393 (SD; Tena) Pellegrini, verae del Rio Urueña, 5 enturi 5676 (MO, US). Tucumán: Depto. Leales, Ruta 9. La Florida, дэк ickas 17353 (CTES); Сараң Villa Luján, (io 38 (SI, US); Tapia, Venturi 2 ара ter 1427 (LIL, US), Sotelo 163 (LIL, US), 2 Aquidaban, Estrella, Fiebrig 4337 (С); in regione cursus superioris fluminis Apa, Hassler 7944 (BM, G, GH, LIL, Boguerón: 59 km S del desvío a Lomá Plata por ruta Trans- Chaco, Mereles 3: 575 (СТЕ5). Chaco: E: rejoncito, Fernández Casas 4313 (С, МО ae Lagerenza, Schinini & Bordas 15201 (CTES, G, US). Presidente Hayes: in regione cursus inferioribus fluminis Pilcomayo, Rojas 65 (С); Puerto "Zapatero Cue," barranca Rio Par: , Rojas 2194 (51); ripam occidentalem un. EA үн S- 7 (СМ, Chaco, 23%0'S, Rojas 2427 (С, GH, MO, NY, P), 7 G, P). Ruellia ciliatiflora is a common weed in dis- turbed places of subtropical South America, es- pecially in western Chaco areas and adjacent regions. This species has generally been known as Ruellia lorentziana in Argentina, where it ranges from Jujuy to Santiago del Estero. The name Ruel- lia ciliatiflora precedes R. lorentziana and was used to describe plants grown in England from seeds sent by Tweedie from Buenos Aires, Argen- tina. Although this species does not grow in Buenos Aires, Tweedie probably collected it during his trips to northern Argentina. The beautiful illustration and description in Botanical Magazine show clearly that it is the same species. It is closely related to species of the Ruellia nudiflora complex from North America. 8. Ruellia coerulea Morong in Morong & Brit- ton, Ann. New York Acad. Sci. 7: 193. 1893. TYPE: Paraguay. Pilcomayo River, 1888- 1890, Morong 1013 (holotype, NY; isotypes, MO, US). Figure 12. m n rophyllum Nees in C. Martius, Fl. Bras. 9: 22. 1847. Ruellia mic crophylla (Nees) Lin- dau, кол. be Jahrb. 19 Beibl. 48: 16. 1894, not Cavanilles, 1801. TYPE: Urugua M ad S. Luciam sub fruticibus ripas fluviorum obumbran- tibus, Sellow s.n. eke B destroyed; isotypes not found). Syn. yiri ipis olius Nees in DC., Prodr. 199. 1 , not Ruellia ra Swartz, ios Ruellia Mai Brit w York Acad. ба. 7: 192. 1893, not л 1886. Ruellia ignorantiae Herter, Sudamer. Bot. 4: 193. 1937. ак (уре aa Griseb. є ех Ке me Rho- dora 47: 13, pl. 840. 1945. TYPE: Argentina. Entre Rios: without locality, Tweedie s.n. (lectotype, K). Syn. nov. Volume 80, Number 4 1993 Ezcurra 813 Ruellia in Southern South America FicunE 12. to show variation in Mos charac C, Krapovickas 11 Sparsely branched erect herb 30-120 cm tall, with fibrous roots from a basal rhizome; ascending, subtetragonous, glabrous, sometimes ciliate at the nodes. Leaves on short, ciliate petioles, the blades ovate or elliptic to MESI oblong or lanceolate, 5-12 cm long, cm wide, the lower ones generally acce and wider than the upper, all obtuse at the apex, cuneate at the base, the margins entire, completely glabrous, with nu- merous sessile glands on the lower surface. Flowers on glabrous pedicels, borne in pedunculate, few- flowered, dichotomous cymes, in the axils of alter- nately one of each pair of the uppermost leaves, stems Ruellia ннн dep — А. Flowering branch. — B. Lea — C. Open capsule without seeds, schematic. (A, Hassler 166 C f shape of a plant of era diera ; B, Venturi 632 rarely on both; bracteoles glabrous, linear, and small. Calyx of chasmogamous flowers 0.7—1.5 cm long, the segments linear-lanceolate, acute, glan- dulose-puberulous. Chasmogamous corollas tilted, blue or whitish, 3-4 cm long, the basal tube 0.5- 1 ст long, the throat obconic, 1.5-2 ст long and ca. 8 mm wide in the orifice, with palate markings in the lower side, the lobes suborbicular, ca. 1 ст diam. Cleistogamous corollas reduced and closed. Stamens included. Ovary and sie glabrous. Cap- sule narrowly elliptic or oblong, 2-2.5 cm long and 4 mm thick, hollow nearly to the — gla- Seeds 16-2 brous; retinacula acute. -3 mm 814 Annals of the Missouri Botanical Garden wide, appressed pilose when dry and viscid when wet. Distribution. Western Bolivia, Paraguay, Uruguay, and northeastern Argentina, and limiting areas of southern Brazil, in open areas of eastern Chaco and adjacent phytogeographical regions. It is found in wet or periodically flooded, sunny places, such as ditches, streams, riverside lawns, or tem- porarily inundated areas. It flowers in spring and summer, from September to February. Representative specimens. ARGENTINA. Chaco: Napalpi-Machagai, Malvarez 839 (LIL), Las Brenas, Malvárez 942 (LIL), Fontana, Meyer 542 (LIL); Bar- ranqueras, — 78(LP); Meyer 16243 (LIL), Depto. m S de Misión Nueva Pompeya, Molina & Moli Esperanza, “Piccinini & Hilfer 4158 (BAB); Entre Saenz eña y Tres Isletas, Ragonese y Castiglione 6930 (LP); Colonia Benítez, Schulz 140 (51), 14147 (ВАВ); Depto. San Martín, de Laguna Limpia a Presidente Roca, Schulz AB, US); Depto. Mayor Fontana, Enrique Urien, Schulz 17774 (BAB, US). Cordoba: Rio I, Obispo Trejo, Balegno 1085 (LIL); Orillas Mar Chiquita, Stuckert 23141 L Corrientes: Goya, Boelcke 1530 (SI); Curuzü ros, Juan Pujol, /barrola 2343 (LIL); 25 k Luis del Palmar, Krapovickas & Cristóbal 11804 (SI, US) 4 km E de Paso de La Patria, Krapovickas & Cristóbal 14931 (CTES, SI); Las Marias, 7 km al 5 de Gobernador Virasoro, Krapovickas 16790 (CTES); 8 km al N de Carlos Pellegrini, Krapovickas 20080 (CTES); Río Santa Lucía, Paso Naranjito, Krapovickas 24562 (CTES); Depto. Empedrado, Estancia La Yela, Pedersen 10833 (CTES, NY); Puente sobre el Río Mirinay, sobre Ruta 123, Schulz 18641 (CTES); 40 km N de Curuzü Cuatiá hacia Mercedes, Zuloaga 3355 (SI). Entre Rios: Concepción del Uruguay, Baez 61 (LP) Gualeguay- chú, Burkart 4270 (Sl); La Paz, Burkart 21418 (51); et 6 © ~ g B "y £5 a Q [=] % ZS < ш Е 5 E 4, Бо» > I 5 = a a № кы © ~ ON MT, Un — EN 2 e > Troncoso 1292 (51); Federación, Chaviyú, Troncoso 1575 (SD. mosa: Ingeniero He tolderia Toba 1 km N del pu welo, Arenas 2274 (SI, US), Estanislao del Campo, Cabral 603 (BAB); Parque Nac. Pilcomayo, alrededores Ruta 81, alrededores de Bosch, alrededores de Palo Santo, Petetín 1594 (ВАВ); Depto. Pilcomayo, Laguna Blanca, Piccinini & Leguizamón 25 10 AB); Herradura, costa del Rio Paraguay, Piccinini & Petetín 3439 (BAB); Las Lomitas, Krapovickas 1310 (SI); Depto. Pilagá, 3 km W de Espinillo, Morel 9269 (LIL); El Colorado, га INTA, barranca alta del Rio Bermejo, Schulz 17546 (CTES); Nuevo Porteno, Schulz 3 (CTES); Estaci ión Palo Sen 139 zu NW de Piq des Zulooga '282 26 (51). ones: uy, Bertoni 52 Arr Candelaria, d ; Santa Ana, Balneario Municipal, По 3179 (SI). Salta: Nuestra Señora de Talavera, Cuezzo Mo- rello 257 (LIL); Embarcación, Palmar de Laguna San José, Hueck 81 (LIL); El к тко 0 705 (LIL); Joaquin V. González, paar d 3 Cerro San Bernardo, Sotelo 837 qi pichuelo hacia Embarcación, iue а 2651 (SI). Santiago del Estero: Silipica, frente a La hae tena, Arganaraz 297 (BA); Entre pend y Colonia Dora, Castiglione & Ragonese 7788 (LP); Depto. Pellegrini Algarrobal Viejo, Luna 37 1 (LIL); Totora Pampa, camino a Atamisqui, entre Ruta 9 antigua Arroyo Sunchal, Piccinini & Petetín 3195 (BAB); Depto. Copo, Los Tigres, Roic 424 (LIL), 604 (BAB, LIL); Depto. Pipe e Ba- nado del Río Uruena, Venturi 5675 (LIL, SI, US). Santa Fe: Tostado, Rio Salado, Job 1137 (LP), S hulz 1262 (LIL); Villa Guillermina, Meyer 2841 i royo El Rey, Núñez & Rivas 93 (LP); Villa Ana, Estero La Julia, Quarín du Depto. Garay, Col. Mascias, Е 154 E sia Amores, Venturi 55158 (BA, SI, US). Tucumán: Depto. Leales, Chanar Pozo, dpi 632 (BA, BAB. LIL, SI, US), 685 (BA, LIL, SI). L. Rio Grande do Sul: Porto Alegre a Canoas, у је 18894 (ICN); Sao Gabriel, Rio Vacacaé, Pa- lacios & Cuezzo 1572 (LIL). PARAGUAY. Amambay: zwischen Rio Apa und Rio Aquidaban, Fiebrig 4338 (G), 4893 (BM, G, K); in curs 3 3 Hassler 803 1 (ВМ, С, NY). Bogueron: Р nas 471 (CTES), 1265 (SI, USy Filadelfia, Hahn 786, 59 (G, MO, US); Pozo Arias, Mereles 495 (CTES, G); Ре Bene Mereles 2264 (G); Ruta pis rn 22°35'S, 50'W, Schinini & Palacios 25802 (С); Mariscal Be Ocho Cué, Schmeda 163 (С). Cen- tral: Base du Cerro Lambaré, Balansa 2452a (P), Plaine d'Aregua, Balansa 2452b (Р); ad ripam rivi Juqueri, Hassler 1533 (G); Orillas Rio Salado, Rojas 1661 (LIL, SI; Tacuaral, 3820 (С); Trinidad-Asunción, Schinini 3524 (CTES). Chaco: Entre Teniente Visi adrejón, Fernández Casas 4318 (G, NY); Cerca de Pozo Colora Е Fernández Casas 4492 (NY); "Ma ayor Pedro Lagerenza, Río Timane, Schinini & Bordas 14865 (CTES), 1 5200 (C TES, US), 75211 (CTES); Cerro León, Schinini & Bordas 18002 (CTES, G, US). Concepción: Concepción, Hassler 7416 (С, GH, K, P). Cordillera: 2 km NE de Rio Salado, de Limpio a Emboscada, Arbo 1640 (CTES, С). Guaira: Plaine de Dona Juana, Balan- sa 2452 (С, P); Tebicuary, Mereles 2307 (С). Neem bucú: Pilar, Meyer 16100 (LIL). Nueva Asuncion: Ruta Trans-Chaco, 21%30'S, 61°15'W, Schinini & Bor- das 16511 (CTES). Paraguari: Paraguari, Fiebrig 890 (B, G, GH); Sparre & Vervoorst 600 (LIL). Presidente ~ Experimental MA Transchaco, pk a Bi es 757 (CTES, G); Estancia La Perla, Pedersen 14620 (С); in regione cursus inferioris fl. Pilcomayo, е 64 (BM, С), 314, 31 + (G); O samp aes araguay, Rojas 2872 (BM, G). San P о: Colonia Primas vera, Woolston 800 (K, LIL, NY, SI), 927 (NY, JS). Without locality: Hassler 8031 (BM, MO). URU- Gu AY. Artigas: Rio Uruguay y Arroyo Itacumbu, Ro- sengurtt 10459 (MVFA). Canelones: Santa Lucia, Gal- «та! B 186 (US), Marchesi 1307 (МУКА). Florida: Santa Lucia Chico, Berro 1732 (MVFA, US). Monte- video: Santa Lucía, Arechavaleta 3108/4788 (SI). Ri- vera: Сипарги, Wright 1928 (BM). Tacuarembo: Volume 80, Number 4 Ezcurra 815 1993 Ruellia in Southern South America Tacuarembó, Berro 4824 (МУКА). Treinta y Tres: Sierra de Maracayú, in regione vicine San Estanislao, Ciudad Treinta y Tres, Rosengurtt 10662 (MVFA). Without department: Bois de Sainte Lucie, Fruchard s.n. (P). Ruellia coerulea is characterized by the blue- flowered axillary cymes and glabrous habit, and is common in seasonally inundated areas of subtrop- ical South America. It appears clearly related to Ruellia brittoniana and Ruellia malacosperma of southern North America. Like many semiaquatic plants, it shows much variation in leaf shape and leaf area, probably subject to environmental influ- ence. This variation may also be the result of in- trogression with the common Ruellia ciliatiflora in areas where they grow sympatrically. Ruellia coerulea shows an extended range in subtropical South America and also varies in habit. The plants from the northern portions of its range (eastern Bolivia, northwestern Paraguay, and west- ern Brazil) generally have larger and more woody stems, and present a more fruticose habit. The plants from the southernmost portion of its range (Uruguay) are frequently small, delicate herbs. This may be related to climatic variations associated with latitude and seasonality, especially tempera- ture and photoperiod. Fernald discussed part of the nomenclatural problems of this species when he validated the name Ruellia tweediana. 'This name has been the one most used for this species (e.g., Dawson, 1979; Piovano & Bernardello, 1991), but the epithet coerulea precedes it. Ruellia microphylla has also been used for the identification of material from Paraguay. Although the type of Arrhostoxylon microphyllum, its basionym, was destroyed at Ber- lin and no isotypes have been found, the original description by Nees of the material from Monte- video, Uruguay, clearly depicts a plant of the weak- er, southernmost form of Ruellia coerulea. A few specimens collected in Paraguay show morphological intermediacy with Ruellia brevi- caulis, suggesting hybridization. 8a. Ruellia coerulea Morong x Ruellia brevicaulis (Nees) Lindau. Plants resembling Ruellia coerulea, with pe- dunculate axillary inflorescences, the flowers ped- icellate and bibracteolate, but with shorter stems and wider, ovate, somewhat serrulate, pilose leaves, the trichomes shiny and extended as in Ruellia brevicaulis. Жылны ad specimens. PARAGUAY. San Pe- Paraguay, Woolston 118 (SI); Primavera, Woolston 989, 1252 (US). Canendiyu: iter ad Yerbales Hassler 4281 (G, P, NY); Sierra de Maracayü, Hassler ). 5092 pro parte (BM, These specimens are morphologically interme- diate between Ruellia coerulea and R. brevicaulis, of group Dipteracanthus. They all have been col- lected in eastern Paraguay, where the areas of these species are sympatric, and appear to be the product of hybridization. 9. Ruellia macrosolen Lillo ex Ezcurra, Syst. Bot. 14(3): 297. 1989. TYPE: Argentina. Ju- juy: San Pedro de Jujuy, 600 m, 10 Apr. 1926, Schreiter 5211 (holotype, LIL; iso types, BA, LP, 05). Figure 13. Low perennial herb 20-40(-60) cm tall, with fibrous, fusiform, tuberous roots; stems ascending, sparsely branched, subtetragonous, pilose and shortly glandulose-pubescent, somewhat glabres- cent at maturity, growing from a short horizontal underground rhizome. Leaves on pilose petioles 0.3-2 cm long, the blades thin, ovate or elliptic, mostly 2.5-7 cm long and 1.5-4 cm wide, rounded or obtuse at the apex, cuneate at the base and somewhat decurrent on the petiole, the margins entire or slightly crenate, undulate, the surface pilose on the principal veins. Flowers on glandular pedicels, borne in terminal and axillary, lax, pe- dunculate, few-flowered dichotomous cymes, ag- gregated in thyrses and often appearing paniculate; bracts foliaceous, pilose; bracteoles glandular, lin- ear, and small. Calyx 1.5-2 cm long, the segments linear and acute, densely glandular-pubescent. Co- rollas erect, white, regular, the basal tube 2.5-5 cm long, 2-3 mm diam., the throat 1.5-3 cm long, tinted with grayish mauve at the base, expanding to 1 cm wide at the orifice, the lobes rounded, ca. 1.5 ст long and wide, often emarginate, somewhat undulate in the margins, with faint grayish mauve markings toward the orifice of the throat. Cleis- togamous flowers frequent, reduced, tubular, and closed. Stamens 4, included, with filaments ca. 7 mm long and anthers 5 mm long. Style 3-5 cm long, glabrous; ovary pubescent, with a few glan- dular hairs. Capsules elliptical, 2.5-3.5 ст long and 6-8 mm thick, broadest at or above the middle, hollow nearly to the base, densely and shortly pu- berulent on the surface; pedicels accrescent at ma- turity. Retinacula obtuse, 2 mm long. Seeds 12- 22, flat, wide, appressed-pilose when dry and with hygro- suborbicular, 3 mm long and 2.5 mm scopic-viscid pubescence all over their surface when wet. Distribution. Southeastern Bolivia, western 816 Annals of the Missouri Botanical Garden C. Inflorescence portion with glandular calyces and bracteoles. — D. Stamens. — E. Dehiscen capsule without seeds, schematic. — С. Seed. (A-G, Pedersen 10818; К, Ezcurra & de Azkue 471.) FIGURE 13. Ruellia macrosolen Lillo ex Ezcurra. — A. Habit. — B. Abaxial leaf surface, pilose on principal veins. > le. — Е. Ореп Volume 80, Number 4 1993 Ezcurra 817 Ruellia in Southern South America Paraguay, and northern Argentina, in regions of western Chaco and adjacent transitional forests. It is found in open and often disturbed places such as sandy roadbanks, abandoned fields, borders or clearings in forests, and margins of streams, from 300 to 900 m elevation. It flowers in spring and summer, from December to February. resentative specimens. ARGENTINA. Chaco: Napalpí en orillas camino a Choritotipo, Donat 48 (LP); m de Fuerte Esperanza hacia Castelli, Molina & Dh 2156 (BAB); Taco Pozo, edis 8997 (LIL); Е speranza, Piccinini & Hilfer 4149 (ВАВ); Las Brenas, Schulz 780 (CTES); Gancedo, Schulz 996 (LIL); Villa Angela, Schulz 2542 (CTES). Formosa: Patiño, 17 km al N de тте del Campo por Ruta 81, Molina 1223 (ВАВ); Matacos, Ingeniero Juárez, Schulz 9713 (CTES). Jujuy: Huaico к, Ahumada 4535 (CTES, SI Pampa Blanca, Birabén 1471 (LP), Cabrera 15556 (LP); camino a El Cadillal, еа, 29947 (SI); alrede- you Aeropuerto, al SE de Perico, Ezcurra & de Azkue 1 (SD; Ledesma Fabris & Zuloaga 7683 (LP); Cerca i Fraile Pintado, Legname & Cuezzo 5947 (LIL, LP); Cerro de San Рей ro, Venturi 5174 (SI, US); Sierra de Calilegua, Venturi o (SI); de Aeropuerto El Cadillal 4, Zuloaga 2993 (SI). Salta: Entre - Ragonese 6710 een 620 (LIL, US); Entre Rosario de la Frontera y Metan, Meyer 22698 (LIL); Almirante е, O'Donell 5394 (LIL, US); Anta, Es- tancia San , Pedersen 10818 (CTES, LP); La = delaria, oa 9361 (BA, ‚и Rosario de la dae era, de Santa Ros Пана Venturi 10200 (MO). Таситап: Vipos, 17 939, Birabén & Birabén 1536 (LP); Burruyacu, de la Sota 1384 (LIL), Venturi 7499 (US); Cerro del C в 7662 (МО, renzo 417, 420 (LIL). PARAGUAY. Amambay: S Rio Apa und Rio Ре Пе Fiebrig 4512, 4543 (BM, С). Boqueron: Estación Experimental Fi- ladelfa, Vanni 2543 (CTES). Chaco: Capitán P. Lage- renza, ca. del puente del Río Timane, Charpin & Ra- mella 21577 (G); Parque Nacional Defensores del Chaco, Duré 441 (CTES); Mayor Pedro Lagerenza, Schinini & Bordas 15078 (CTES, US) Río Timane, Schinini Dulce, Shinini & Bordas a Asuncion: Ruta Trans Chaco, Schi- nini & "ован 16550 (CTES). Ruellia lenis cl 1 by the white, nocturnal, ко пева lans flowers with a long basal tube. It seems closely related to the melitophilous, diurnal, blue-flowered Ruellia ciliatiflora, and is partially sympatric with it. They are probably ge- netically isolated through different pollen vectors (Ezcurra, 1989) 10. Ruellia hygrophila C. Martius, Flora 24, Beibl. 2(5): 65. 1841. Lychniothyrsus hy- grophilus (C. Martius) Bremek., Bull. Torrey Bot. Club 75: 669. 1948. TYPE: Brazil. With- out locality, 1839, Martius Herb. Fl. Brasil. 560 (holotype, BR not seen; isotype, NY). Figure 14 Cryphiacanthus udus Nees in C. Martius, Fl. Bras. 9: 49. 1847. TYPE: Brazil. Mato Grosso: prope Cuiabá, Riedel s.n. (holotype, LE not seen, photo SI). Syn. nov. w perennial, acaulescent herb, with a short basal vertical rhizome, and thick, fibrous roots. Leaves clustered in a rosette, shortly petiolate, obovate, 3-9 cm long and 1-3 cm wide, obtuse and rounded in the apex, the blade decurrent on the petiole; margins entire or somewhat crenate, surface glabrous or slightly hirsute. Flowers borne on lax, multiflowered, anisocaulous, elongate di- 15 ст long; peduncles puberulous, to 8 cm long; bracteoles linear and small; pedicels puberulous. TE 1–1.5 cm long, the segments linear, acute, 0.8-1.3 cm long, short-pilose, ciliate in the mar- gins. Corolla pale blue, 3-5 cm long, the basal tube 1-1.5 cm long and 2 mm wide, obconic throat 1.2-1.5 cm long and 8 mm diam., the lobes re- flexed, suborbicular, 1–1.5 ст diam. Stamens in- cluded with anthers ca. 2 mm long. Capsule nar- rowly elliptic, approximately the same length as the calyx, 1-1.5 cm long and 3 mm diam., hollow nearly to the base. Retinacula less than 2 mm long; seeds 14-16, approximately 2 mm long, ad- pressed-pilose on the entire surface, the hairs ex- tended and mucous when wet. Distribution. Southeastern Bolivia, south- western Brazil, Paraguay, northeastern Argentina, and northwestern Uruguay. It is common in bor- ders and clearings of open Chaco xerophilous for- ests and in palm formations, on hard clay and perodically flooded soil. It flowers in spring and summer, from October to March. Representative И ARGENTINA. Chaco: Fontana, Meyer 2238 (LIL); Colonia Benitez, Schulz 227 (SI), 779, 2948, 16342, 16. 352, Vig ad ora 17158 (CTES, US); La Clotilde, al S de Saenz Pena, Schulz 8164 (CTES, LIL); Las Breñas, Schulz 1063. 3 (C ТЕУ); Depto. 12 de Octubre, inedo, Sch (CTES); Depto. Tapenagá, Cote Lai, Schulz 14469(C TES); Depto. = ntana, Enrique Urien, Schulz entes: Depto. Capital, Punta Araz, [A 1530 CTES, LP); Depto. Capital, атейа del Riachuelo, Burkart 684. 3 (51); Perugorria, Cabrera 10565 (LP); alrededores de Capital, Hicken 19932 (SI), Mar- 818 Annals of the Missouri Botanical Garden 14. Ruellia С Martius. — А. Habit. ).) & Bordas 18100; B, Arenas 2359 tinez & Schinini 10392 (CTES), Meyer 5895 (LIL), Curuzü kcu Martínez Crovetto 8419 (ВАВ), Зре- azzini, 24 Apr. 1940 (ВАВ); San Luis del Palmar, roga 467 (CTES); Depto. Mercedes, 14 km W de Schulz 18506 (CTES). Formosa: Matacos, G. ч, Juárez Barrio Obrero, Arenas 2359 (SI); Bartolomé Casas, Eliseth-Cano 172 (BAB), 4 km S de In- geniero Juárez, Fortunato 245, 258 (BAB), 6 km al N de Las Lomitas, Fortunato 449 (BAB); Formosa, Jor- gensen 2843 (ВА, LIL, SI, 05) Depto. Pirané, Los Matacos, Morel 697 (LIL, US); P x s Parodi 8500 (LIL); Depto. Pirane, Ruta 8 _- Santo, d & P doa (ВАВ); Depto. T uta 2 0 km de Rut , Petetin 1821 (BAB); з c | Lomitas, d & Pire 24212 (СТЕ5); 15 km W de Las Lomitas, Schinini 24256 (CTES); NC oM gi. Fontana, Schulz 17015 (CTES); Bartolomé de Las Schulz 17608 (CTES); Nuevo Porteno, Schulz | 968 (C TES, US). Misiones: San Ignacio, Campos, Alboff, 22 Nov. 1894 (LP). Salta: Depto. Rivadavia, El Tordillo, Maranta 77 (US). Santa Fe: De Reconquista a Nicanor na, Job 884 (LP); Depto. Gral. Obligado, Lanteri, Maldonado B. 1650 (LP); Villa Guillermina, Meyer 2840, 3269 (LIL); Depto. Vera, Santa Lucia, Tecone 4929 (LP). Santiago del Estero: Capital, Los Tigres, Meyer & — B. Open capsule without seeds, schematic. (A, Schinini Vaca 23284 (LIL), Roic 423 (BAB, LIL). PARAGUAY. Alto Paraguay: Puerto Casado, Rojas 2197 (5 Amambay: zwischen Rio Apa und Rio Aquidaban, Fie- brig 4327 (BM, С, GH). Boquerón: Mariscal Estigar- ribia, Mereles 2245 (G); Alfredo Stroessner, Spichiger 2072 (US). — Juqueri, Hassler 1532 (G); Aregua, ari, Mereles 1735 (С); Patino (Nueva Asun ción), mn 608 (GM). Concepción: Prope Concep- ción, Hassler 7330 (BM, С, GH, LIL, NY); Puerto Risso, © QE e ES Bl. camino a Loreto, regine Cordillerae centralis, Hassler 6507 (BM, G је : (CTES, US); Agua Dulce, Schinini € Bordas 18100 (CTES, US). Guairá: prope Villa Rica, Balansa 2461 (G, P). зови а lari e 918(BM 3 (LIL). E Hayes: minis Pilcomayo, Rojas 6711 dts Santa Elisa, Којаз 2670 (BM, C, GH, LIL, MO, NY, P), Hassler 7330 (BM, G, LIL, To P). San Pedro: Distr. Lima, Estancia Carumbé, Pedersen 9357 (GH, 51); Primavera, Woolston 775 (LIL, NY, SI, US). URUGUAY. Artigas: Ruta 30, Arroyo Catalan, Paso Piaui, Rosengurtt B7173 (МУКА). Volume 80, Number 4 993 Ezcurra 819 Ruellia in Southern South America FIGURE 15. Ruellia morongii Britton. 31102; B, Burkart 26385.) Ruellia hygrophila is characterized by a rosette habit combined with multiflowered long-peduncu- late pseudoracemes of blue flowers. It differs from the closely allied short-pedunculate, pale-flowered Ruellia morongii also by the nearly glabrous ped- icels. It is a common species in all eastern Chaco and surrounding areas, in seasonally flooded and periodically dry soi 11. Ruellia morongii Britton in Morong & Britton; Ann. New York Acad. Sci. 7: 192 1893. Cryphiacanthus acaulis Nees in C. Martius, Fl. Bras. 9: 49. 1847, not Ruellia acaulis R. Br. 1810. SYNTYPES: Brazil. River Uruguay, Baud s.n. (K); Brazil. Without lo- cality, Sellow s.n. (isosyntype, K); Brazil. Rio Grande do Sul: River Jacuhy, 1837, Tweedie 770 (K). Figure 15. Low perennial herb, acaulescent, with vertical or oblique rhizome, the thick, fasciculated roots fusiform. Leaves in rosette, ovate, oblong-ovate or obovate, 3-10 cm long and 1-3 cm wide, obtuse and rounded in the apex, the blade decurrent on the petiole, margins entire or somewhat crenate, surface glabrous or slightly hirsute on the principal A. Habit. —B. Open capsule without seeds, schematic. (A, Burkart veins. Flowers shortly pedunculate or borne on subsessile, few-flowered dichasia, frequently re- duced to a solitary flower with two bracteoles. Pe- duncles glabrescent, to 3 ст long; bracteoles linear and small; pedicels hirsute-tomentose under the flowers. Calyx 1.2-2.2 cm long, the segments lin- ear, acute, 1-2 cm long and 1-2 mm wide, densely pilose and ciliate in the margins. Corolla white, sometimes with bluish or yellowish tints, 3-5 cm long, the basal tube 1-1.5 ст long and 2 mm wide, the obconic throat 1.2-1.5 cm long and 8 mm diam., the lobes reflexed, suborbicular, 1-1.5 cm diam. Stamens with anthers ca. 3 mm long. ipti Ro pner. shorter 3-4 mm Capsule narrowly elliptic, than the calyx, m long anc diam., hollow nearly to |: base. Retinacula less than 2 mm long; seeds 10-14, approximately 2- 3 mm long, adpressed-pilose on the entire surface, the hairs extended and mucous when wet. Distribution. Southern Brazil, eastern Para- guay, northeastern Argentina, and Uruguay. Ruel- lia morongii is found in open and sunny places, on rocky or sandy soil, burnt campos, and grassy savannas. It flowers in spring and summer, from October to Marc 820 Annals of the Missouri Botanical Garden Representative specimens. ARGENTINA. Corri- entes: Depto. Esquina, 47 km W de Sauce, Ruta 126, Ahumada 1444 (CTES); Curuzú Cuatiá, Birabén 5185 (LP), /barrola 2526 (LIL), Martínez Crovetto 8429 (BAB); La Cruz, Burkart 7866 (SI); Depto. Curuzú Cuatiá, cerca de Baibiene, Castellanos BA34509 (BA, LP); Depto. Santo Tomé, Ruta 14, Las Marias, Krapovickas 17095 (CTES); Depto. Ituzaingó, 15 km E de Ruta 12 camino a San Carlos, Krapovickas 18146 (CTES): Estancia El Recreo, 21 km E de Bonpland, Lourteig 2774 (CTES); Je ES, x pus de Santo Тоте hacia Garruchos, Zuloaga. 3132 (SI). Entre Rios: Arroyo Robledo a Redomón, ат 26061 (SI); Salto Grande, Burkart 27 197, 31102 (51), Cabrera iil 1 s, Burkart 27571 (SD; Depto. Colón, Palmar Grande, Burkart 26385, 27376 (SI); Federal, Burkart 28885 (51); Colonia Eloisa, barra del Mocoretá, Burkart 29495 aun Nueva Escocia car dini 33 (SD; San José de Felici Nicora 7719 (SI Federación, Martínez Crovetto 47 702 (BAB). Misiones: Apostoles, Cabrera 28479 (SI); Posadas a San José, km pasando Arroyo rides Cuezzo 10832 (LIL); San Javier, Fabris 7386 (LP); San Ignacio, Pastoreo Chico, Montes 15557 (LIL); Posadas, id pese 207 (SI); km SW de Santa Ana, Schinini 5597 (CT toles a Concepcion de la Sierra, Arroyo HA Tunas, Zu- loaga 3254 (SI). BRAZIL. Rio Grande do Sul: Pelotas, Capão de Leão, Amadeu 70 (ICN); TAN Campo de Capão do Corvo, Augusto 18895 (ICN); São Luis, próx- imo a Rio Icamaqua, Fleig 116 (ICN); 17 km W de Rosario, BR 290, Lindeman 8415 (ICN); Vila Manresa prope Porto Alegre, Rambo 1 (PACA); Arroio dos Ratos prope Зао Jerónimo, Rambo 8466 (PACA); Fazenda do Jarau prope Quarai, Rambo 26027 (PACA); Porto Ale- gre, Morro da Gloria, Rambo 2940. 3 (LIL); Santo Angelo, Rambo 5204 1 (PACA); Sao Luis Gonzaga, Sehnem 3624 (SI); Uruguaiana, puente sobre Rio Ibicuí, Sobral 3252 (ICN, MBM). PARAGUAY. Paraguari: Orilla Rio Te- bicuary, frente a Florida, Krapovickas 12387 (CTES, US). San Pedro: Prope San Estanislao, Hassler 5997 (G). Without locality: common in saline campos, Itupé, Jorgensen 4125 (MO, NY, SI, US); central Paraguay, Morong 223 (US), 323 (GH, US). URUGUAY. Artigas: Bella Unión, Castellanos 1837 1 (LIL, LP), Herter 4484 (RB); Puerto Bentiú sobre Rio Cuareim, próximo a Bella Unión, Del Puerto & Marchesi 11380 (MVFA); Santa Rosa Cuareim, Herter 985 (LIL, NY, MO). Paysandu: Meseta Artigas, Del Puerto & Marchesi 3488 (MVFA); Rivas, d 82946 (US); Chapicuy, Rosengurtt B 4181 (US). Salto: San Antonio, Rosengurtt B-927 (US); Río $ . Tacu- uarembo, Berro 4914 (US), 4958 (MVFA); Arroyo iesus Lema 6904 (MVFA) E Pis - (RI Ruellia morongii is characterized by having leaves in a rosette and white subsessile flowers. It seems morphologically related to Ruellia hygro- phila, but is mostly allopatric with this species and inhabits an area more to the east. It is common in sandy, sunny, open areas and burnt savannas, and seems adapted to periodic fires. Ш. Group Hygrophiloidei Dipteracanthus d Nees in C. Mar- : 43. 1847. TYPE: Dipteracan- thus lamiiformis Nees (есе, һеге ѕе- lected). tius, Flowers sessile or shortly pedunculate in the axils of the upper leaves, bibracteolate, generally soli- tary, sometimes in fascicles. Calyx segments nar- row, united in the base, the posterior longer. Cap- sule obovate in contour, apiculate, with a narrow solid base ! its length. Seeds Species of this group are characterized by their sharply quadrangular stems, frequently reduced axillary cymes, and few-seeded capsules. They are common in floristic provinces of the Amazon region or dominion (see Cabrera & Willink, 1980), at low elevations, growing in moist soil. e name of this infrageneric group together with Calycosi, Genuini, Inaequifolii, Paniculati, and Subspicati were proposed by Nees (1847a, b) under Dipteracanthus, but without a clear indi- cation of rank. Following article 35 of the /nter- national Code of Botanical Nomenclature (Greu- ter et al., 1988), these names are valid provided that all other requirements for valid publication are fulfilled, and may act as homonyms or serve as basionyms for subsequent combinations in definite ranks. 12. Ruellia epallocaulos Leonard ex Ezcurra & Wasshausen, Brittonia 44(1): 69. 1992. TYPE: Paraguay. In regione fluminis Alto Pa- гапа, Fiebrig 5372 (holotype, SI; isotypes, C, GH, K, LIL, US). Figure 16 Suffruticose perennial 0.30—1 m tall, branched from a subligneous base. Stems erect or somewhat decumbent, quadrangular, glabrous, rooting at the nodes. Leaves on glabrous petioles 0.4-1 cm long, the blade ovate, 3-6(-10) cm long and 1-2(-3.5) m wide, acute or acuminate, obtuse or rounded at the base, glabrous, the margins faintly and ir- regularly crenate. Flowers sessile, solitary, some- times in twos, in reduced subsessile dichasia con- densed in the axils of the upper leaves, exceptionally pedunculate. Bracteoles small, lanceolate, 3-5 mm long. Calyx 1-1.8 cm long, the segments oblong- lanceolate, rounded or obtuse, to 1.5 cm long, + glabrous, shortly ciliate, the posterior longest. Co- rolla purple or blue, 4-5 cm long, the basal tube 1.8-2.5 cm long, the throat obconic, 1-1.5 cm long and ca. 1.2 cm diam., the lobes spreading, suborbicular, 0.5-1 cm wide. Stamens included, Volume 80, Number 4 1993 Ezcurra Ruellia in Southern South America 821 D FIGURE 16. Ruellia epallocaulos Leonard ex Ezcurra & Wasshausen. — A. Habit. — B. Calyx with bracteoles. — C. Open calyx. — D. Dehiscent capsule. — E. Open capsule without seeds, schematic. (A-E, Hatschbach 8322.) long filaments ca. 8 mm long, short 6 mm long, the anthers 2.5 mm long. Ovary and style glabrous. Capsule narrowly ovate, 1.2-1.5 cm long and 3- 4 mm thick, slightly apiculate, with a solid base ca. 4 mm long, the surface glabrous. Retinacula 2 mm long; seeds 6-8, 2-2.5 mm diam., glabrous, the margins pubescent and mucilaginous when wet. Distribution. Eastern Paraguay, northeast- ern Argentina, and southern Brazil, on the basin of the Paraná River (Paranaense phytogeograph- ical province). It is found in the shade of the gallery forest, generally in humid places near water- courses. It flowers in spring, from August to Oc- tober. Representative specimens. | ARGENTINA. Mi- siones: Yarupá, Arroyo Las Tranqueras, Grondona & 822 Annals of the Missouri Botanical Garden Spegazzini 1374 (BAB); Puerto Pampa, Kermes 197 (ВАВ); San Ignacio, Quiroga s.n. (BA 21419); Depto. Montecarlo, Arroyo Piray Guazü y ruta 12, Schinini 19904 (CTES). BRAZIL. Рагапа: Mun. Corso Azul, Turvo, Hatschbach 5025 (MBM); Mun. Cerro Azul, Cer- Hatschbach 8322 (HBR, MBM, US); Mun. Сиро, Fazenda Lagoa, Hatschbach 16950 (MBM, 05) Mun. Cerro Azul, Barra do Tigre, Hatschbach 40347 (MBM, US). PARAGUAY. Alto Parana: in region minis Alto Paraná, Fiebrig 5751 (G, SI); Reserva Itabó, Itaipá геа 925 (МО КиеШа epallocaulos is characterized by the relatively large, solitary, bluish flowers and the markedly quadrangular stems. It is sometimes con- fused with the sympatric, red-flowered Ruellia an- gustiflora, because of its similar habit. 13. Ruellia kleinii Ezcurra & Wasshausen, Brittonia 44(1): 72. 1992. TYPE: Brazil. Santa Catarina: Florianópolis, Morro de Riberào, Klein 7333 (holotype, US; isotypes, HBR, MBM). Figure 17. Small suffruticose herb to 50 cm high; stems erect, markedly quadrangular. Leaves ovate, 2.5- 3.3 cm long, 1.5 cm wide, acute and acuminate, cuneate, puberulous. Flowers subsessile, solitary, sometimes in sessile or pedunculate dichasia re- duced to 3 flowers or less, in the axils of the upper leaves; bractlets foliaceous, less than 1 cm long and 3-4 mm broad. Calyx ca. 1 cm long, with linear segments 7-8 mm long and 1 mm broad, puberulous. Corolla white, 1.5-2 cm long, the basal tube 6-7 mm long, the throat obconic, 6 mm long, 5 mm diam., the lobes suberect, 4 mm long and 3 mm broad. Stamens included, long filaments 4 mm long, short 3 mm long, the anthers ca. 1.5 mm long. Capsule markedly stipitate, ca. 1 cm long, apiculate, few-seeded. Distribution. Southern Brazil, in openings in forests, frequently on wet ground along water- courses. Representative specimens. BRAZIL. Santa Cata- ;atarina, Florianopolis, ı no pasto ümido, Catarina, T n& se se 8532 (US); Ilha de Santa Catarina, Florian polis do Piri, Klein & Souza 8626 (US), Smith & Reitz 6192 (US). Ruellia kleinii is characterized by the delicate, quadrangular stems and the small flowers with ovate bracteoles. It is quite similar to some of the spec- imens included by Nees in the original description but has smaller, ovate leaves and is found in a more south- ern area. of Dipteracanthus menthoides Nees, IV. Group Dipteracanthus Ruellia L. sect. Dipteracanthus (Nees) Lindau in Engl. & Prantl, Nat. Pflanzenfam. 4(3b): 309. 1895. Dipteracanthus Nees in Wallich, Pl. Asiat. Rar. | 32. TYPE: Dipteracanthus prostra- tus Nees. Gymnacanthus Nees in Lindl., Nat. Syst. Bot. ed. 444. 1836, not Oersted, 1854. TYPE: Gym- nacanthus petiolaris Nees. а (Calycosi) Nees in DC., Prodr. 11: 116. 1847 Dipteracanthus Nees (Genuini: Bracteolati) Nees in ., Prodr. 11: 117. Siphonacanthus Nees in C. Martius, Fl. Bras. 9: 1847. TYPE: Siphonacanthus villosus Nees. Flowers sessile, solitary, borne in the axils of the upper leaves, bibracteolate. Capsules thick, oblong in contour, with a small solid base М their length. Seeds Group Dipteracanthus is pantropical and pos- sibly the one with the most species in the genus. The neotropical species have been separated from the Old World Dipteracanthus by Bremekamp (1969: 423), but the characters that supposedly differentiate them are inconsistent. Taxonomically, it is the most problematic group; in southern South America it is partly represented by the assemblage of species allied to Ruellia dissitifolia, which are very difficult to delimit. Most of these species are suffruticose herbs that grown in open areas of flo- ristic provinces of Amazon region or dominion (see Cabrera & Willink, 1980). 14. Ruellia brachysiphon (Nees) Lindau in Engl. & Prantl, Nat. Pflanzenfam. 4, Abt. 3b: 310. 1895. Dipteracanthus brachysiphon Nees in C. Martius, Fl. Bras. 7: 34. 1847. TYPE: Brazil. Rio Grande do Sul: Porto Allegre, Sellow 68 (holotype, B destroyed, photo F 5911; isotype, K). Erect herb to 30 cm tall, with stems AA + branched, velutinous, arising from a small r zome. Leaves shortly petiolate, the blades elliptic to widely ovate, 2-4 cm long, 1.5 acute, obtuse or rounded at the apex, sometimes -3 cm wide, apiculate, cuneate at the base, + entire, sometimes serrulate, the lower ones smaller and rounder, all softly and shortly velutine-puberulous to densely velutinous, with some hairs glandular, sometimes glutinose, the underside densely pale velutinous. Volume 80, Number 4 1993 Ezcurra Ruellia in Southern South America 823 5 mm 5 mm 5 mm FIGURE 17. Ruellia Мети Ezcurra € Wasshausen. — A. Flowering branch. — B. Calyx. — C. Capsule. — D. Open capsule without seeds, schematic. (A-C, Klein 7333; D, Klein & Bresolin 8476.) Annals of the Missouri Botanical Garden Flowers sessile, solitary in the axils of the terminal leaves. Bractlets elliptic or lanceolate, ca. 1.5 cm long and 3-5 mm wide, acute, foliaceous, densely pubescent. Calyx 1-2 cm long, the segments ovate or lanceolate, to 1.5 cm long and 2-7 mm wide, hortly puberulent to densely pubescent or velutinous, ciliate. Corolla erect, pale blue, 4-5 (-8) ст long, the basal tube less than 1(-2) ст long, the throat 2-2.5(-4) cm long, and 1.5(-2 cm wide, the lobes spreading, suborbicular, 1— 1.5(-2) cm wide. Stamens included, long filaments ca. 1 cm long, short 0.5 cm long, the anthers 3 mm long. Mature capsule unknown. ~ Distribution. Eastern Paraguay, northeast- ern Argentina, and southern Brazil; occurring rare- ly in open, high campos, and flowering in spring, from September to December Representative specimens. ARGENTINA. Corri- ro, Birabén 5054 (LP); Desiderio Sosa, Birabén 5211 (LP); Ituzaingó, sobre ruta 12, próximo al пне con Misiones, Legname 8627, 8637 (LIL). Mi- siones: Entre Pueblo Santa Ana y cerro Santa Ana, Alboff, 26 Nov. 1896 (LP); Depto. Candelaria, Loreto, Montes 32 (LP), 840 (BAB, LP); Depto. Iguazü, Puerto Esperanza, Montes 9458 (LIL); Depto. Iguazü, Puerto Wanda, Montes 10409 (LP); Depto. Iguazu, El Dorado, Schmidt 2776 (LIL, pro parte); Depto. San Ignacio, Pas- toreo, rin Las Tunas, Schmidt 5036 (LIL); Santa Ana, Misiones, Spegazzini s.n., Jan. 1907 (LP). BRA- — ls] Ф 20 Feb. 1943 (ICN). PARAGUAY. Alto Paraná: Montes 9912 (LIL). Caaguazú: in campis, Hassler 9225 (С); 15 km N de Caaguazú, amino а lhü, Schinini 21977 (CTES) Ruellia brachysiphon is a problematic species, variable in calyx morphology, and with a wide distribution, although apparently represented only by a relatively small number of specimens. И is characterized by the corollas with a short basal tube, the frequently wide calyx lobes, and the ve- lutine-puberulous surface of the leaves. 15. Ruellia brevicaulis (Nees) Lindau in Engl. & Prantl, Nat. Pflanzenfam. 4, Abt. 3b: 310. 1895. Dipteracanthus brevicaulis Nees in C. Martius, Fl. Bras. 7: 35. 1847. TYPE: Brazil. Without locality, Sellow 3190 (holotype, B destroyed, photo F 5912). Figure 18. Dipteracanthus dissitifolius Nees var. humilior Nees in Martius, Fl. Bras. 9: 35. 1847. SYNTYPES: Brazil. Minas Gerais: Claussen 232 (P not seen); Sao Paulo: Sao Paulo, Lund 733 (G not seen, p herb. DC); Without state: Puerto Grosso, cn B de- stroyed, photo F 5919). Syn. n ~ Erect, suffruticose herb, the stems terete, to 30 cm long, hirsute, arising from a small woody rhi- zome with thick, fleshy, fusiform roots. Leaves on petioles to 0.3 cm long, the blades widely ovate, 3-5 cm long and 1.5-3 cm wide, acute at the apex, cuneate at the base, serrulate, the lower ones generally smaller, all hirsute-pilose, the hairs long, white, shiny and spreading, sometimes also mixed with short, sericeous pubescence. Flowers sessile, solitary in the axils of the terminal leaves. Bractlets oblong, foliaceous, 1.5 cm long and 2-5 mm wide, acute, long-pilose. Calyx 1.5-2 cm long, the seg- ments linear-lanceolate, to 1.8 cm long and 2 mm wide, acute, hirsute-ciliate. Corolla pale blue, 5-7 cm long, the basal tube 1-2 cm long, the throat longer, obconic, to 4 cm long and 2.5 cm diam., the lobes speading, suborbicular, ca. 2 cm wide. Stamens included, long filaments ca. 1 cm long, short 0.5 cm long, the anthers 3 mm long. Ovary and style glabrous. Capsule oblong in shape, 1.2- 1.5 em long an cm thick, with a solid base 16 из length, the surface puberulous. Retinacula 3 mm long; seeds 6-8, lenticular, 3-4 mm diam., with margins appressed-puberulous, densely mu- cilaginous when wet. Distribution. Southern Brazil, northeastern Argentina, and eastern Paraguay. Ruellia brevi- caulis is common in open, high campos, to m, especially on rocky or sandy ground subject to periodic fires. It flowers in spring, from September to December. Representative иа ера Шиа Соггі- ntes: Depto. Santo Tomé, yo Chirimay, Correa 6852 (BAB); Ruta 14, a Virasoro, аа as 167 10 (CTES); 15 km de Virasoro camino a Garru- chos, Krapovickas 2096 1 (CTES); Santo Tomé, Estancia Garruchos, casco, Krapovickas 21436 (CTES); Depto. sens 342 (CT . Misiones: Depto. San Pedro, Campo Cum prido, Pen 2097 (LIL, US); Apóstoles, Escuela Agrotécnica, Cabrera 26593 (SI); Santa Ana, de Llamas 621 (BAB); 30 km de L. N. Alem, camino a San Javier, sobre ruta 4, Krapovickas 1: 5200 (CTES); Loreto, — tes 59, 248, 684 (all LP); Santo Pipo, Grondona & Piccinini 3215 (ВАВ); San Ignacio, Schwarz 5021 (LIL, US). BRAZIL. Parana: Estrada de Curitiba a Ponta bien e 2023 (PKDO); Palmeira, Fazen Santa Rita, Dombrowski 12077 (PKDC) Curitiba, i in campo, Dusén 17277 (GH); Mun Curitiba, Cajurú, Hatschbach 39 M); Mun. Lapa, Rio Passa Dois, Hatschbach 64 19 (LIL, MBM, US); Mun. Sao Jose dos Pinhais, Campo Largo de Roseira, Hatsch- atl 8338 (MBM, US); Mun. a Grossa, Parque Vila Velha, Hatschbach 45470 (MBM); Mun. Curitiba, Ca- panema, Kumrow 2364 (MBM, 05) Curitiba, Cidade Industrial, Pedersen 10885 (CTES); Araucaria, Pereira Volume 80, Number 4 Ezcurra Ruellia in Southern South America 825 FIGURE 18. Ruellia brevicaulis (Nees) Lindau. Partridge s.n., BA 61609; B, Cabrera 28593.) A. Habit. — B. 5 mm ру | Open capsule without seeds, schematic. > 826 Annals of the Missouri Botanical Garden 8084 (RB). Rio Grande do Sul: Tres Barras, in pseudo sampo, Dusén 17610 (GH); Arroio Tatim, Soledade, Fleig 194, 205 (ICN); 27 km N de Vacaria, BR 116, km 258, Krapovickas & Vanni 36860 (CTES, US); Santa María, Lindeman 8259 (ICN), Esmeralda, Estação Ecológica Aracuri, Miotto 983 (ICN); Pinhal, Palacios & Cuezzo 2358 (LIL); Pa | rina: Сооронов. Ponte Alta do Sul, Кп 3246 (HBR. US); Curitibanos, campo, Klein 4068 (HBR, US); Lajes, Passo do Socôrro, Klein 4423 (HBR, US); Lajes, Morro do Pinheiro Sêco, Klein 4529 (HBR, US); Campos Novos, campo, Reitz & Klein 14296 (HBR, US); Abelardo Luz, Reitz & Klein 16509 (HBR, US); ao longo da Estrada de Rodagem Federal ao sul de Lajes, ms & Klein 8117 (HBR, US) Fazenda Esper- anca, Campos de Palmas, 6 km ao sul de Horizonte, Smith & Klein 13494 (HBR, US); 8 km ao oeste de Campo Eré, Smith & Klein 13782 (HBR, R, US). PAR- ; . Amambay: aguay, zwischen Rio Apa und Río Aquidabán, Fiebrig - 4378 (BM). Cordillera: Prairies de Capitindú à PE de la Cordillére de Villa Rica, Balansa 2462 (P); Doria Juana, prés de Villa Rica, Bal- ansa 2462a (K, P). Alto Parana: Irala, Montes 9912 (LIL). Caazapa: Distrito Yuty, 20 km E de San Miguel, desvio a 3 де Маус 4281 (BM). Misiones: Ruta Schinini 21941 (LIL) 4, 20 km E de San Ignacio, Ruellia brevicaulis is characterized by the fu- siform, tuberous roots, the hirsute pilosity —the hairs long, shiny and spreading—and the large, blue corollas with a short basal tube and a large throat. It is a common species in open, elevated campos of southern South America subject to pe- riodic fires. The epithet “brevicaulis” is well applied to specimens of plants that have recently been burnt and only produce short stems before flow- ering. Some specimens do produce longer stems, probably when fires are less frequent. Specimens of Ruellia brevicaulis have often been misidentified as Ruellia dissitifolia (e.g Wasshausen & Smith, 1969), but А. ds ola is a different species only found in central Brazil (Goiás, Minas Gerais, and São Paulo), characterized by the basal obovate leaves, the dense, whitish, velutino-sericeous pubescence, and the large flow- Ruellia brevicaulis seems to hybridize with Ruellia coerulea, a species of group Ruellia, as suggested by the existence of morphologically in- termediate specimens from eastern Paraguay, where both species are sympatric (see description of R. coerulea). 16. Ruellia multifolia (Nees) Lindau in Engl. & Prantl, Nat. Pflanzenfam. 4(3b): 310. 1895. Dipteracanthus А Nees in C. Mar tius, Fl. Bras. 7: 847. SYNTYPE: Brazil. Sellow s.n. (syntype, B destroyed, photo F 5938; isosyntype, K). Figure 19. Erect herb or suffruticose perennial to 35 cm high, with stems subterete, = branched, velutinous or hirsute on some sectors, arising from a horizontal rhizome with numerous, thick, fusiform roots. Leaves subsessile or shortly petiolate, the blades elliptic to widely ovate, 2-5 cm long, 1.5-3 cm wide, acute, obtuse or rounded at the apex, some- times apiculate, cuneate or truncate at the base, + entire, sometimes serrulate, the lower ones smaller and rounder, all softly and shortly velutine- uberulous to densely hirsute-velutinous, with many glandular hairs, sometimes glutinose, the underside with = densely scattered sessile patelliform glands. Flowers sessile, solitary in the axils of the terminal leaves. Bractlets elliptic or lanceolate, foliaceous, ca. 1.5 cm long and 3-4 mm wide, acute, folia- ceous, densely glandular-pubescent to shortly ve- lutinous. Calyx 1-2 cm long, the segments lan- ceolate, to 1.5 cm long and 2 mm wide, acute, glandular-pubescent or velutinous, ciliolate. Corolla erect, pale mauve, 4-7 cm long, the basal tube 1.5-3 ст long, the throat 1.5-3 ст long and 1.5- 2 cm wide, the lobes spreading, orbicular, 1.5-2 cm wide. Stamens included, long filaments ca. 1 cm long, short filaments 0.5 cm long, anthers 3 mm long. Style + glabrous. Capsule oblong 1.5- 1.8 ст long and ca. 5 mm thick, with a solid base 16 its length, the surface glabrous. Retinacula 3 mm long; seeds 6-12, 3 mm diam., with margins appressed-puberulous. Distribution. Central and southern Brazil, eastern Paraguay, and northeastern Argentina. It is found growing in open cerrado and elevated campos, frequently on sandy soil. It flowers in spring and summer, from October to Mare 16a. multifolia. Figure 19A Ruellia multifolia (Nees) Lindau var. This variety is characterized by the somewhat suffruticose habit, the generally unbranched stems, the densely velutinous-hirsute elliptic leaves that turn glabrescent on the upper surface at maturity, and the coriaceous blade with the nerves prominent below. Distribution. Northeastern Argentina, east- ern Paraguay, and central and southern Brazil. Representative specimens. GENTINA. Mi- siones: Campo Grande, km 184, je p (LIL); dpi San Da ardin América, Crisci 4 (LP); Sar alto arenoso, Quiroga (BA); Sans Ana, Rodríguez 769 (BA, LIL); San bns io, Schulz Volume 80, Number 4 Ezcurra Ruellia in Southern South America ge 19. of R. multifolia var. viscossisima (Nees) Ezcur Ruellia multifolia ( (Nees) iras — А. Flowering branch of К. multifolia var. multifolia. — C. Open — B. Habit capsule without seeds of variety multifolia, schematic. (A, Forge 1804; B, Krapovickas & Cristobal 13618; C, Zuloaga 3276. 7113 (LIL); De San Ignacio a Santa Ana, a 6 km de San Ignacio, Zuloaga 3299 (SI), Apóstoles, Escuela Agrotéc- nica Pascual Gentilini, Zuloaga 3276 (SI). BRAZIL. Pa- : Castro, en campo, Dombrowski 11994 (HBR); Campo Mourão, Hatschbach 7621 (MBM); Arapoti, Rio das Cinzas, barra do Perdizes, Hatschbach 8563 (MBM); Mun. Senges, Rio Pelame, Hatschbach 26786 (MBM, US). Santa Catarina: Campo Erê, Fazenda São Vicente, Klein 4999 (HBR), Smith & Klein 11550 (HBR, RB, US); Campo Erê, Chapecó, Reitz 4499 (HBR, NY, US); Campo Erê, Smith & Reitz 9373 (US). PARAGUAY. o Paraná: Reserve Tajjupi, Hernandarias 6 km, sed 2133 (G). Amambay: zwischen Rio Apa und Rio Aquidaban, Fiebrig 4423 (G); Cursus superioris fluminis Apa, Hassler 8009 (G); Parque pau juin ro pue cerrado camino a Lorito Picado, Sor I Е 196 | usu propiedad del Senor Trosiu, + 1091 (С). Canendiyu: ierra de Maracayü, Has 5092 (G, GH, NY pro pa arte); in campo prope Igatimí, Hassler 5648 (G). Con- cepción: zwischen Rio Apa und Rio Aquidabán, Cen- 828 Annals of the Missouri Botanical Garden turión, Fiebrig 4378 (6, K). Cordillera: Cordillera de Altos, Fiebrig 374 (G), Hassler 1804 (G, US); in campo Itacurubi, Hassler 1704 (G); in campis Tobaty, Hassler 3259 (G, GH, NY, P); In valle fluminis Y-hacá, in campo prope Piribebuy, Hassler 6814 (С); Arroyo Y-hacá guazú, Schwarz 11226 (LIL); Valenzuela, Rio Y-hacá, Sparre & Vervoorst 1158 (LIL). Guaira: Villa Rica, Jorgensen 4307 (LP, MO, US). Paraguari: Parque Nacional Ybi- cui, campo cerrado 5 km N of administration building on road to Cesar Barrientos, Zardini 8698 (MO). S regione vicine San Estanislao, Hassler 4278 (G); in regione fluminis Tapiraguay, Hassler 5973 (C). 16b. Ruellia multifolia (Nees) Lindau var. viscossisima (Nees) Ezcurra, stat. nov. Dip- teracanthus viscossisimus Nees in C. Mar- tius, Fl. Bras. 9: 34. 1847. Ruellia viscos- sisima (Nees) Lindau in Engl. & Prantl, Nat. Pflanzenfam. 4, Abt. 3b: 310. 1895. ТҮРЕ: Brazil. Without locality: Sellow 1413 (holo- type, B destroyed, photo F 5960; isotype, K). Figure Dipteracanthus Ligier Eye tatus Nees tius 4T. Ruellia xen ни punctata (Nees) La ie & бешш. Nat n 4, Par cR . 1895. SYNTYPE: Bra- zil. "Rio Grande do үз de Sellow s.n. (B destroyed, пао Е m Syn. nov. This variety differs from Ruellia multifolia var. multifolia by the more herbaceous habit, the less branched stems, the widely ovate leaves, obtuse or rounded at the apex, the shortly and softly velu- tinous pubescence with many glandular hairs, fre- quently glutinose, the underside with densely scat- tered sessile patelliform glands. Distribution. | Southeastern Paraguay and lim- шпе areas of northeastern Argentina and southern Brazil, in campos. е specimens. ARGENTINA. Corri- мо. Mburucuyá, Estancia Santa Teresa, Виг- har 19512 (SI); Depto. Mburucuyá, 10 k ucuya, Krapovickas & Cristóbal 13618 (CTES, LP, US): Depto. Mburucuya, Estancia Santa Teresa, Pedersen 2995 (AA); Depto. San Miguel, 21 km S de Loreto, Schinini 8202 (CTES). Misiones: Posadas, Loreto, Ek- man 910 (US); San Ignaci acio Quirog rapovickas & peg a — (US). PARAGUAY Parana: Reserva tí Yupi, jr Binacional s Caa, :C alan- Basualdo 2140 (MO); "lavat, 3 km 8 Ыы Des soria 3275 (MO). Cone i i dillera: in campo prope Itacurubi, Hassler 1704 (P); in dumeto Cordillera de Altos, Hassler 3500 (G). Without locality: Paraguaria centralis, Hassler 3529 (BM, P, NY); en el campo Irupé, Jorgensen 4307 (SI). Ruellia multifolia var. viscossisima is closely related morphologically to Ruellia multifolia var. multifolia, with which it can often be confused. Variety multifolia is a generally more robust form that is quite frequent in south-central Brazil and northeastern Paraguay. Variety viscossisima, more common in southeastern Paraguay and limiting areas of northeastern Argentina, may just be a herbaceous, short-lived form, the product of sea- sonal climatic factors associated with latitude, such as photoperiod and temperature, or of edaphic conditions. Although this variety was described from ma- terial reported to have been collected by Sellow in southern Brazil (Nees, 1847a: as Dipteracanthus viscosissimus and D. glanduloso-punctatus), I have found no other Brazilian material of this plant. The Spanish name of the only locality that appears ээ in the type specimens (*S. Ignacio," and not the Portuguese orthography *'S. Inácio”) suggests that this material could have also been collected in northeastern Argentina near Brazil, in San Ignacio, Misiones, where the plant has been found relatively recently. 17. Ruellia solitaria Vell. Conc., Fl. Flumin.: 266. 1829. TYPE: Brazil. Rio de Janeiro: il- lustrated in Fl. Flumin. Icon. 6: t. 95. 1831. Figure 20 Dipteracanthus schauerianus Nees, Ind. Sem. Hort. Vra- tis S 32 (not m Linnaea 16: 290. 1842. TYPE: in Cald.” (not seen). buena. calvescens Nees in C. Martius, Fl. Bras. 9: 32. 1847. Ruellia фа (Nees) Lindau in Engler & Prantl, Nat. Pflanzenfam. 4(3b): 3 1895. SYNTYPES: Brazil. Rio de Janeiro: Lacané, July 1839, Gardner 805 (K; isosyntype, BM); Sebas- tianopolim, Martius s.n. (M, photo SI); Corcovado, Riedel s.n. (LE not seen); without locality, Schott 6129 (W not seen, P F 32759), Sellow s.n. (B destroyed, photo F 5 ). Branched, suffruticose herb 30—50 cm tall. Stems erect, obscurely quadrangular, pale whitish and terete at maturity, glabrous. Leaves on petioles 0.5-1 ст long, the blades widely elliptic or ovate, -12 ст long and 2-4 cm wide, obtuse or rounded at the base, glabrous on the surface, the margins entire. Flowers sessile, solitary in axils of the upper leaves, sometimes in twos. Bracteoles ovate, foli- aceous, 1.5-2.5 cm long and 0.5-0.8 cm wide, glabrous. Calyx 0.8-1 cm long, the segments lin- ear, to 0.8 cm long and 2 mm wide, acute, glabrous, ciliolate, somewhat accrescent and remaining on the stems after the shedding of the capsules. Corolla white or pale mauve, ca. 5 cm long, the basal tube 2 cm long, the throat obconic, 2.5 cm long and 2 З ст фат., the lobes spreading, suborbicular, 1.3-1.8 cm wide. Stamens included, long filaments Volume 80, Number 4 1993 Ezcurra 829 Ruellia in Southern South America 5 mm B 20. Ruellia solitaria Vell. — А. Flowering branch.— B. Open capsule without seeds, schematic. (A, IGUR А Hatschbach 34793; B, Hatschbach 14708.) ca. 1.2 mm long, short 0.8 mm long, the anthers 3 mm long. Ovary and style glabrous. Capsule oblong-ovate, 1.5-2 cm long and 5 mm thick, with a short solid base 3-5 mm long, the surface gla- brous. Retinacula 2 mm long; seeds 12-16, 2 mm diam., glabrous, the margins pubescent and mu- cilaginous when wet. Distribution. Southern Brazil, in the hygroph- ilous forest on the eastern side of the southeastern Atlantic coastal ranges, from the states of Espirito Santo and Rio de Janeiro to Paraná (“Мата lantica," Atlantica floristic province). in the shady underwood of the forests. It flowers in winter and spring, from July to October. 830 Annals of the Missouri Botanical Garden Re epresentative specimens. BRAZIL. Parana: Mun. Morretes, Formigueiro, Hatschbach 345 (MBM, RB); Grota Funda, Hatschbach 4068 (MBM); Morro 7 BUR сајт Hatschbach 11340 (МВМ); Estrada Itupava, o Joào, Hatschbach 14708 (MBM, NY US) E за E Graciosa, Ferradura, Hatschbach 19237 (MBM). Mun. Morretes, Estrada Graciosa, Grota Funda, Hatsch- bach 34793 (MBM, US); Rio Bromado, Hatschbach 42499 (MBM); Rio Taquaral, Hatschbach 44983 (MBM, US); Mun. Campina Grande do Sul, Serra Capivari Gran- de, Hatschbach 16209 (МВМ, US). Mun. Guaraquegaba, Rio do Cedro, Hatschbach 19237 (MBM). Ruellia solitaria is characterized by the gla- brous stems and leaves, and the large, white or pale mauve, sessile, solitary flowers, with a pair of widely ovate bracteoles in the base. It is frequent in the understory of the coastal hygrophilous forest of southeastern Brazil. It is similar to Ruellia lilaci- na Hooker, a blue-flowered species of Amazonian Peru. V. Group Ebracteolati Dipteracanthus Nees (Genuini: Ebracteolati) Nees in ., Prodr. 11: Gymnacanthus Oersted, Vidensk. Meddel. Dansk Naturhist. Foren. Kjobenhavn 1854: 126. 1854, not Nees, 1836. TYPE: Gymnacanthus geminiflorus (HBK) Oersted (= R. gemini- flora HBK). Copioglossa Miers, Proc. Roy. Hort. Soc. London 3: 294. 1803. TYPE: Cops pilosa Miers. ee S. Moore, . 42: 33. 1904 ҮРЕ: Tremacanthus Е, S. Moire: Ulleria remake Proc. Kon. Ned. Akad. We- tensch. Ser. C. 72, 14: 423. 1969. TYPE: Ulleria geminiflora (HBK) Bremek. (= Ruel- lia geminiflora HBK). Flowers sessile, solitary, borne in the axils of the upper leaves, sometimes in twos by production of supernumerary buds; bracteoles obsolete or absent. Capsule obovoid, thick, with a small solid base 4 its length. Seeds (2-)4. Group Ebracteolati is restricted to the New World and consists of several suffruticose herbs common in tropical and subtropical open areas. In South America they exist as a complex of species allied to К. geminiflora that is frequent in savannas and campos, in floristic provinces both of the Am- azon and Chaco regions or dominions (see Cabrera & Willink, 1980). This group is characterized by a maximum reduction in the structure of the in- florescence (flowers sessile, bracteoles lacking), and by the capsule with a minimum number of seeds. The subgeneric name Ebracteolati was first used by Nees (1847b) under Dipteracanthus (Genuini) Nees, but without clear indication of rank. Follow- ing article 35 of the International Code of Bo- tanical Nomenclature (Greuter et al., 1988), this name is valid provided all other requirements for valid publication are fulfilled, in which case it may act as a homonym or serve as a basionym for future combinations in definite ranks. 18. Ruellia bulbifera Lindau in Engl. & Prantl, Nat. Pflanzenfam. 4(3 b): 311. 1895. Dip- teracanthus tuberosus Nees in C. Martius, Fl. Bras. 9: 42. 1847, not Ruellia tuberosa L. 1753. TYPE: Brazil. Without locality: Sellow s.n. (holotype, B destroyed; isotype GZU not seen, photo US). Figure 21 шш А Lindau, Bull. Herb. Boissier 5: 654. : Brazil. е е in civitate es Paulo a Pane in campis, ‚ Loefgren 997 (ho- lotype, B destroyed, ноо F 5933; uns C not seen, photo US). Syn. Sprawling or decumbent herb, with stems to 30 cm long, arising from a woody, rounded or terete, tuberculate xylopodium to 2 cm diam.; stems green, obscurely quadrangular, glabrate, with nodes + ciliate. Leaves shortly petiolate, widely elliptic, ob- long or obovate, 2-4.5 cm lon 1-2 cm wide, rounded at the apex, widely cuneate at the base, the lower ones generally smaller and + orbicular and retuse, all glabrous and coriaceous, nitidulous on top, with sessile glands and the venation prom- inent below. Flowers sessile, solitary in the axils of the terminal leaves. Bractlets absent or obsolete. Calyx 1.5-2 cm long, the lobes linear subulate, ca. 1.5 cm long and 2 mm wide, glabrous on the surface, ciliate. Corolla purplish white, 3.5-4.5 cm long, the throat 2- 5 cm diam. long, the basal tube ca. 1 cm 2.5 cm long, the lobes orbicular, ca. 1 Stamens included, longer ca. 10 mm long, shorter mm long, the anthers 3 mm long. Ovary and style glabrous. Capsule ovoid, 1-1.2 cm long and mm thick, broadest at or above the middle, glabrous, the base solid to 3 mm high. Seeds 2- 4, brown, ca. 4 mm diam., lightly appressed-pu- berulous, densely mucilaginous when wet; retinac- ula 3 mm long. Distribution. Eastern Bolivia, southwestern and southern Brazil, and northeastern Argentina, in dry grasslands, savannas, and campos. It is common on red earth, in recently burnt areas. It flowers in spring and summer, from October to Representative specimens. ARGENTINA. Corri- entes: Depto. Mburucuya, Estancia Santa Teresa, Bur- Volume 80, Number 4 1993 Ezcurra Ruellia in Southern South America FIGURE 21. pine EUG Lindau.— А. Habit. 5406; B, Zuloaga 3133.) kart 19524 (SI); Depto. Santo Tomé, на Vira- soro, Ibarrola 1220 (LIL); Depto. Ituzaingó, 20 k povickas & сша 20841 (CTES, US); Depto. Santo Tomé, Estancia San Juan Bautista, Krapovic kas 26041 (CTES); Depto. Ituzaingó, ig ud Krapovickas 26380 (CTES); Depto. Santo е, Estancia Garruchos, Peder- sen 5466 (LP); Depto. Mode 40 km SW de Carlos Pellegrini, Schulz 18682 (CTES); 14 km N de Santo Tomé, Zuloaga 3133 (SI); Ruta 40, entrada a Garruchos, a 2 km del pueblo, Zuloaga 3138 (SI). Misiones: Po- sadas, Bertoni 816 (LIL), Martinez Crovetto 8098, 8163 (BA santa Ana, Llamas 481 ); Apostoles, Mar- tinez Crovetto 8187 (BAB); Depto. San Ignacio, Gober nador Roca, Mart ínez Crovetto 86 13 (ВАВ); Concepción de la Sierra, Martínez Crovetto 8798 (BAB); Loreto Montes 3484 (SI), 12414 (LP). BRAZIL. Paraná: Ponta Grossa, Dusén 2724 (R); Turma 23, in bL Dusén 16576 (СН); Пагаге Elo Morengava, Dusén 16576 (GH); Mun. Lapa, Eng o Bley ped 1000 n. у на 10 de inzas, Barra do Perdizes, Hadschlneh 7940 (MBM); Mun. Campo Mou- rào, Campo Mourào, Hatschbach 12998 (MBM, US). Rio Grande do Sul: Ca ps Macaco Branco, Eisinger 25 (ICN); Entre Passo Fur y Carazinho, Lindeman, Irgang & Valls 8201 (IC N); 20 km SE de Santa Rosa, Lindeman, Irgang & Valls 8990 pro parte (ICN); Julio de Castilhos, Palacios & Cuezzo 2558 (LIL); Cruz Alta, Pivetta 615 (PACA); Canoas, Rambo 1787 (LIL); Tu panciretam, Rambo 9596 (PACA); ad montem Sapucaia prope Sao Leopoldo, Rambo 38414 (LIL, PACA); Ca- a prope Gravatai, Rambo 39628 (LIL, PACA). nta Catarina: Abelardo Luz, campo, Smith & Klein 128: 52 (HBR, R, US), 73295 (HBR, P, R, US). PAR- AGUAY. Boqueron: Puerto Casado and vicinity, Es- bye Palo Santo, Pedersen 4165 (GH). Caaguazu: al e Yhu, Fernández Casas 3908B (С). Central: Гра- 5 mm A B B. Open capsule without seeds, schematic. (A, Pedersen pa о de Concepción, camino a Loreto, Vanni 362 (C TES). Cordillera: Cordillére de Piribebuy, Balansa 4708 (P). Guaira: Villa Rica, Balansa 2464a (P). Itapua: En- carnación, Colonia Encarna : А ; US Misiones: W de San Ignacio camino a Pilar, Arbo 1884, 1887 (CTES); Santiago, Estancia La Soledad, Pedersen 4315 (AA, 51); Ruta 4, 20 km E de San Ignacio, Schinini 21942 (CTES). Paraguari: агг i lansa 24646 (P). San Pedro: i Estanislao, Hassler 4240 (BM, G, K, P); Es rumbe, Pedersen 8496 (LP); Primavera, Woolston 888 (PACA, SI, US). Without locality: Paraguaria septen- Hassler 7383 (BM, C). ~ = RE = о . e & trionalis, Ruellia bulbifera is closely related to the widely distributed and variable Ruellia geminiflora, from which it could have been derived, but is charac- terized by the completely glabrous, coriaceous leaves, generally rounded at the apex, and the thick, tuberculate xylopodium. It is easy to identify in its southernmost area, but in Minas Gerais, Bra- zil, a few specimens with thick, rounded xylopodia do show somewhat pubescent leaves (e.g., Duarte 4329, This species is associated with frequently burnt, open campos and savannas. 19. Ruellia geminiflora HBK, Nov. Gen. Sp. 2: 240. 1817. Dipteracanthus geminiflorus (HBK) Nees var. procumbens Nees in C. Mar- Fl. Bras. 9: 40. 1847. TYPE: Colombia. tius, Annals of the 832 Missouri Botanical Garden FIGURE 22. Ruellia geminiflora HBK.—A. Habit.—B. Open capsule without seeds, schematic. (A, Cabrera 28119; B, Krapovickas 16494.) Volume 80, Number 4 1993 Ezcurra 833 Ruellia in Southern South America Locis temperatis siccis prope Santa Ana et Ibague, 500-700 hexapodia, Humboldt & Bonpland s.n. (holotype, P, photo F 39431). Figure — oS Vell. Fl. Flumin.: 266. p : Brazil. nd in Fl. Flumin. Icon. 6: 31. Dipteracanthus canescens Nees, London J. Bot. 4: 635. 1844. Ruellia canescens (Nees) Lindau in Engl. & Prantl, Nat. Pflanzenfam. 4(3b): 310. 1895. TYPE: жеш Without locality, 1837, Schomburgk 37 robable isosyntypes, BM, K). Syn. nov. Dipterac "rcu humilis Nees in C. Martius, Fl. Bras. 9: . 1847. Dipteracanthus humilis var. minor Nees in C. MS tius, Fl. Bras. 9: 39. 1847. SYNTYPES: Brazil. Goias: Natividade, Dec. 1839, Gardner 3418 (K); Rio Grande do Sul: San Pedro, Pohl 683 (W not seen, photo Е 32746); s. es Sellow s.n. (B destroyed; isosyntype, K). Syn. Dipterac 'anthus geminiflorus (HBK) Nees var. Нино 2. Martius, Fl. Bras. id Mato ia: without soles 242 (P). Syn. Ваіт гас canis geminiflorus ber Nees var. erectus N 1 C. Martius, Fl. Bras. 9: 40. 1847. TYPE: Brazil. Mi Grosso: Padrocinio, Pohl 673 (syntype, BR; isosyntype, W); шош ане У КА OW s.n. о В destroyed; | e, K). Syn. Dipterac anthus ge пица анд (HBK) Nees var. subacau- s, Fl. Bras. 9: 4 7. TYPE: Minas poate ы locality, Clalazon 676 (probable syntype, K). Syn. nov. Dipteracanthus geminiflorus иу Nees var. angus- tifoliu Brazil. Nees in C. Martius, Fl. Bras. 9: 40. 1847. Dipterac iun angustifolius (Nees) Bremekamp, Rec. d. Trav. Bot. Néerl. 35: 157, t. 14. 1938. TYPE: Vene nerd iu. Valencia, Moritz s.n. (holotype, B destroyed). Surinam: without locality or collector (paratype, K herb. Hooker). Trinidad: Sa- h, Lockhart s.n. (paratype, K herb. Hooker 362). Syn. nov. Dipteracanthus ен Nees 1 in C. Martius, be Bras. : 41 5: Brazil. Mato Grosso: Cuya E Chapada Riedel 63 (isosyntype, GZU: not seen, photo US); Pi а” Cans without locality, Se id s.n. Re. B de- stroyed; isosyntype, K). Syn. Dipteracanthus porrigens er var. на Nees in C. Martius, Fl. Bras. 9: 41. 1847. TYPE: Brazil. Ceará: Serra de Araripe, in 1838, Gardner 1802 (ho- lotype, BM; isotype, US). Syn Dipteracanthus vindex Nees in C. Mart us, Fl. Bras. 9: 42. 7. SYNTYPES: T Minas Gerais: in campis ad Contendas, May 1818, Martius s.n. (M, photo d without locality, Se/low s.n. E ITA B de- royed; isosyntype, K). е Copioglossa pilosa Miers, Proc. Soc. London . 1863. TYPE: Brazil. San Paulo: Jaraquahiba, r 357 (holotype, BM). Syn. nov. Ruellia ia Fritsch, Bih. Kongl. Svenska Ve- tensk Handl. 24, Afd. 3(5): 25. 1898. TYPE: Brazil. Minas Gerais: Sào Joào d'El Rei, in collibus apricis adustis, qui "campos" vocantur, Lindmann 4 123 (holotype, S). Suffruticose erect herb, with stems to 50 cm long, from a terete woody rhizome less than 1.5 cm diam.; stems obscurely quadrangular, pubes- cent, densely white-arachnoid and hirsute at the nodes. Leaves shortly petiolate, widely elliptic, 2— 5 ст long and 1-2.5 cm wide, acute at the apex, cuneate at the base, + white-tomentose, sometimes densely velutinous. Flowers sessile, solitary or gem- inate in the axils of the terminal leaves. Bractlets absent or obsolete. Calyx 1.5-2 cm long, the lobes linear and subulate, 2 mm wide, densely pubescent. Corolla whitish tinted with purple, 3-4 cm long, the basal tube ca. 1 cm long, the throat 2-2.5 cm long, the lobes orbicular, ca. 1.5 cm diam. Longer stamens ca. 10 mm long, shorter 5 mm long, anthers 2 mm long. Ovary and style puberulous. 1-1.2 ст long and 7-8 mm thick, puberulous, the base solid to 3 mm high. Capsule obovoid, Seeds 2-4, approximately 3 mm diam., appressed puberulous, densely mucilaginous when wetted; ret- inacula 3 mm long. Distribution. lands of Central America and the West Indies (Durkee, 1986), and in savannas of South America, from Colombia and Venezuela to eastern Bolivia, Vastly extended in open grass- southwestern and southern Brazil, Paraguay and northeastern Argentina. A heliophilous species, it is generally found in open sunny places and on sandy ground, in periodically burnt areas. In south- ern South America it flowers in spring and summer, from October to March. м нета ч Итан ARGENTINA. Chaco: Ruta 11 y Riacho Iné, Fabris & gr А L M. Jorgensen 2328 (BA. EIL. S У ol 1° P5 Ма о, ос кк ша Ве бз. 113 (ВАВ), 784 (CTES), 7417 (LIL), 34114 (CTES). Tonden Depto. Mburucuyá Estancia Santa Teresa, Burkart 19445 (SI, US), Cabrera 11609 (LP); Depto. San Martin, La Cruz, Tres Cerros, Cabrera 28119 (SI); Depto. Capital, Arroyo Riachuelo y Ruta 12, Cristóbal 1148 (CTES); Depto. Itatí, Ramada Paso, ка ickas & Cristóbal 16494, 16495 (CTES, SL LIL, US); Depto. Ituzaingó, Rincón Ombú Chico, Krapovie hs 25603 (CTES); Depto. San Miguel, Loreto, giae Municipal, Mroginski 52 (CTES, SI); Depto. Ita Ibaté, Barranquerita, Ruiz Huidobro 2077 (LIL); Depto. Sad Roque, 3 km SE del Rio Santa Lucía, camino a Tobay, Ruta 17, Schinini & Quarin 7443 (CTES); Depto. Saladas, Pago del Deseo, Schwarz 145 (LIL); 834 Annals of the Missouri Botanical Garden Depto. General Paz, General Paz, Schwarz 8475 (LIL); Depto. San Luis del Palmar, Estancia Za apir Fabri S4 Сене: 7079 (LP) gensen ‚2847 (ВА, LIL, - vickas 1117 (LIL); е Lih Tatané, Schulz 17079 (CTES). Misiones: Santa A de Llamas 340, 360 (BAB), PETS 611 E Bonpland, Jorgensen 137 (BAB, L ме. Loreto, Montes 497 (SI); Candelaria, Montes 835 (LIL): 1 guazu, Puerto Istueta, Montes 10132 (LP); San Ignacio, Santo Pipo, Schwarz 5356 (LIL); San Pedro, Santa Teresa, Schwindt 1912 (LIL). Santa Fe: Romany, Martinez Crovetto 4806 1/2 (BAB, LP); Dep- to. 9 de Julio, Arroyo ан у Киша 40, Molina 2011 (BAB). BRA AZIL. pie Hatschbach 2068 (MBM, US); Mun Jaguariaiava, Fa- cenda das Almas, Hatschbac A 20004 (MBM, US); Mun. T Rio Atu P Hatschbach 25623, 32729 (MBM, JS). Rio Grande do Sul: 32 km de Santo Angelo, Fleig 66 ICN): p Hagelund 4367 (CTES); Granja Piratini, Santo Angelo, Hagelund 5801, 10391 (CTES); Santo Angelo, Rambo 53013 (РАСА); Pestana, prope Ijuí, Rambo 57616 (PACA); is "uda Ijuí, Rambo 60297 (PACA). Santa Catarina: Capinzal, estrada de pcd Klein 4292 (NBR, US). PARAGUA Y. Amam- bay: Parque = ~ e Nacional Cerro Сога, МЕ of park head quarters pad 6819 (MO), 6820 (G, MO): Colonia Yuypyté, 56°W, 23°S, Aeropuerto, Schinini & Bordas er s. Caaguazu: Coronel Oviedo, Ruta 2, km 40, Krapovickas & Cristóbal 13424 (CTES, LP). nendiyu: in silva Sierra de Maracayü, Hassler 5382 em tral. Margen S del Río Salado, camino de Limpio a Emboscada, Arbo 1671 (СТЕ5); ни Ва- ous 2463a (P); Ita Ibate, TES 321 (G); Prope lacus Ypacarai, Hassler 3219 (BM, G); Areguá, Mereles 1639 (С); Patino, Teague 577 (BM). Concepcion: zwischen Rio Apa i ] San Bernardino, costa del lago Ypacaray, Quarin 1540 pro parte (CTES, US); Valenzuela, Rio Y-hacá, Sparre & Vervoorst 1159 (LIL). Guairá: Villa Rica, Jorgensen 4123 (GH, LIL, MO, NY, SI, US), 4126 (NY, US); Dona Juana, prés de Villa Rica, dans les prairies, Balansa 2463 (С, P); Cordillera de Ybytyruzü, road to Cantera m Zardini 15275 (MO). Misiones: 12 D W de S Ignacio, camino a Pilar, Arbo 1886 (CTES, US). Neon bucu: Curupaity, Schulz 7969 (CTES). San Pedro: in locality: in silva prope Arroyo Mocoy, Hassler с a central Paraguay, Men Ра NY, US), 8 H). Ruellia geminiflora is characterized by the somewhat thick, fibrous roots, the pubescent stems, the more or less ovate, pubescent leaves, the ax- illary sessile flowers that lack bracteoles, and the thick, obovoid, few-seeded capsule. It is an ex- tremely variable species with а wide distribution that presents many local forms. These have fre- quently been given varietal or specific status and names. Nees recognized five varieties for Brazil (18472, b): procumbens, hirsutior, erectus, subacaulis, and angustifolius. Several other species that have been described for South America are also obvi- ously related to Ruellia geminiflora. It is some- times difficult to decide if they should be treated as separate taxonomic entities of a wide Ruellia geminiflora complex, or to include them within a single species as ecotypes, without giving them formal taxonomic recognition. Here Ruellia gemi- niflora is presented in its wider sense, and the names of several local variants have been reduced to synonymy. Such is the case of populations with narrow leaves that have been described as Dip- teracanthus Nees, geminiflorus var. angustifolius Nees, Dipter- acanthus humilis Nees, or Dipteracanthus vindex canescens Dipteracanthus Nees. These small, narrow-leaved forms appear in different areas of South America (central and southern Brazil: Minas Gerais, Mato Grosso do Sul, and Rio Grande do Sul; Guyana; Venezuela) and may be the result of parallel evolution within the species, or of long-distance dispersal phenomena, probably associated with migrating birds that could carry the mucilaginous seeds from one region to another. Such is also the case of the larger, more robust and branched shrubby plants with variable coriaceous leaves that are frequent in northeastern, central, and southwestern Brazil, and have been described as Ruellia porrigens Nees. Some spec- imens of eastern Paraguay, such as Hassler 5382 from Canendiyú, or Schinini 20507 and Solomon 6819 and 6820 from Ататђау, fall into this cat- egory. This form could be the result of edaphic conditions or of continuous growth related to an accidental absence of periodic fires during several successive growing seasons in some areas. А similar case is that of short-stemmed, nearly acaulescent populations described as variety subacaulis or Ruellia lindmaniana, which may be the response of the plants to the shortening of the growing period by the action of burning. Although some of these forms may deserve subspecific status, to decide on this а thorough study of the morphological and ecological diversification of the species throughout Central and South America should be undertaken. EL of Ruellia geminiflora from Para- gua e been mistakenly identified by Morong & cin (1892-1893) as Ruellia bahiensis (Nees) Morong. Ruellia bahiensis is another species of group Ebracteolati, which is very similar, but in- habits a different area and differs by the markedly swollen nodes of the stems and the somewhat ac- crescent calyces that remain attached to the leaf axils after the shedding of the capsules. It is found in northeastern Brazil. Volume 80, Number 4 993 Ezcurra 835 Ruellia in Southern South America 20. Ruellia magniflora Ezcurra, sp. nov. TYPE: Paraguay. Central: Villa Elisa, sandy fields, Pedersen 6522 (holotype, LP). Figure 23. Herbae perennes ad basim suffruticosae, radicibus fasciculatis fibrosis crassis, ramis ascendentibus 30 argentea, utrime que 5 1 vel geminati, 3 in Salle foliorum ter- siles, solitar iur dispositae, etis. Calycis seg- nenta linearia ca. 2 cm pe 1.5 mm lata extus pubescens ша dense ciliata. Flores caerulei 5-7 cm lon gi. Corollae areg tubo faucibus bre- puberula, 4-sperma. Semina 3 mm longa, in siccus adpresso puberula. Suffruticose herbs from a small, woody rhizome, with thick, fleshy, fusiform roots. Stems erect, to 30 cm long, velutine-puberulous. Leaves shortly petiolate, narrowly oblong or lanceolate, 5- long and 1.5-2 cm wide, acute and cuneate, gray- ish velutine-tomentose, shorter and obtuse. Flowers sessile, solitary or in twos in the axils of the terminal leaves. Bractlets obsolete. Calyx 2 cm long, the lobes linear, ca. 1.5 mm wide, pubescent, densely ciliate. Corolla pale blue, 5-7 cm long, the basal tube 1.5-2 cm long, the throat 2.5-3.5 cm long, the lobes orbicular, ca. 2.5 cm diam. Longer stamens са. 1 cm long, the lowermost somewhat shorter stamens 0.5 cm long, anthers 3 mm long. Ovary and style glabrous. Capsule obovoid, 1 cm long and 0.8 cm thick, puberulous, 2—4-seeded, the base solid to 2.5 mm high. Seeds 3 mm diam., appressed-puberulous, densely mucilaginous when wet; retinacula 3 mm lon Distribution. Common on sandy savannas of eastern Paraguay, southwestern Brazil, and north- eastern Argentina. It flowers in spring and summer. Representative specimens. hn is Corri- entes: Depto. Mburucuya, a Santa Teresa, Pe- dersen 9844 (CTES); Villa Olivari. 18 km W de Ituzaingó, Schinini 15664 (CTES); Ituzaingó, Estancia Santa Tecla, Schwarz 8039 (LIL), Ruta 12, 20 km E de Рави. d 3220 (SI). BRAZIL. Mato Grosso do Sul: uveia, Fazenda, campo rupestre, Cerati 33790 (SPF), “ха Rio Brilhante, Rio Vacaria, Hatschbach 25186 (MBM). PARAGUAY. Amambay: Parque Nacional Cer- ro Corá, ca. Cerro Guaiguy Hog, Fernández Casas 4016 (G); 7 km N de la Ruta 5 limite W del Parque Nacional Cerro Corá, Ferrucci 701 (CTES). Concepción: zwisch- en Rio Apa und Rio Aquidaban, Villa Sana, Fiebrig 4843, 4882 (G). Cordillera: in pratis prope San Bernardino, Hassler 110 (С); in silva Caraguatay, Hassler 3333 (G); in dumeto prope _ need 4592 (С); Ваггапса Arroyo Y-haca, Teague 108 (BM); iter ad yerbales mon- tium “Sierra de а ns 4238 (BM, C). FIGURE 23. abit. — B. Open ca Ruellia magniflora Ezcurra.— psule without seeds, schematic. (A, Pedersen 6522; B, Hassler 110.) Ruellia magniflora is clearly allied to Ruellia geminiflora; it differs by its fleshy, fusiform, tu- berous roots, its narrow, lanceolate, longer leaves, and by its larger, pale-blue flowers. 836 Annals of the Missouri Botanical Garden VI. Group Salpingacanthus —— Moore, J. Bot. 42: 107. 1904. PE: 5. nobilis S. Moore. Flowers sessile, in the axils of the upper leaves; bracteoles obsolete. Calyx segments unequally cleft, forming a three-lobed structure. Capsule oblong in shape, with a small solid base % its length. Seeds за. 8. The only species of this group is remarkable because of its large, white, sphingophilous flowers, with a long basal tube and exserted stamens. It is found in a small area of southwestern Brazil and northern Paraguay, in eastern Chaco forest. Group Salpingacanthus shows no close affini- ties with any other groups in the area. 21. Ruellia nobilis (S. Moore) Lindau, Feddes Repert. 7: 67. 1909. Salpingacanthus no- bilis S. Moore, J. Bot. SYNTYPES: Brazil. Mato Grosso do Sul: Corum- bà, Robert 713a, Robert 800 (BM; isosyn- types, NY). Figure 24. Suffruticose herb or straggling shrub with the stems sparsely branched, terete, glabrous, pale at maturity. Leaves on petioles 0.5-1.5 cm long, the blades widely ovate, 4—10 cm long and 2.5-5 ст wide, acuminate, rounded on the base and some- what decurrent on the petiole, the margin entire, the surface lightly velutine-puberulous, especially on the underside, the secondary veins parallel and prominent beneath. Flowers subsessile, solitary in the axils of the upper leaves. Bracteoles obsolete. Calyx 2-3 cm long, unequally cleft, lower segments united with the lateral ones, the posterior free, forming a three-lobed structure, the lobes 2 cm long and 1 ст wide, puberulous, light green in color. Corolla white, ca. 10 cm long, the basal tube 7—9 cm long, the throat obconic, 1-2 cm long and 1 cm diam., the lobes white, spreading, suborbicu- lar, 2.5 cm wide. Stamens subexserted, filaments ca. 1.5 ст long, anthers 5 mm long. Ovary and style glabrous. Capsule oblong-elliptic in shape, 1.5- 3 cm long and 4-5 mm thick, with a solid base to 0.5 cm, the surface glabrous. Retinacula 4 mm long; seeds 6-8, 3-4 mm diam., glabrous, pubes- cent and mucilaginous when wet. Distribution. Southwestern Brazil and north- ern Paraguay, in eastern Chaco forest. Ruellia nobilis flowers in spring. iuridice specimens. BRAZIL. Mato Grosso do Sul: Corumbá, in Prius silvulisque daris regionis calcaridis, Мате 2747 (5); Corumbá а 200 m da divisa Bolivia, cerrado estrada do Jacadigo, Almeida 17 13 (RB). PARAGUAY. Alto Paraguay: Fuerte Olimpo, Chaco Paraguayo, Rojas 13638 (CTES, G, LIL, US). Ruellia nobilis in the Acanthaceae and Sphin- giphila tetramera (Gentry, 1990) in the Bigno- niaceae appear to show convergent evolution with respect to corolla morphology. Both species, en- demic to the same area (northeastern Paraguayan Chaco, and limiting areas of Brazil and Bolivia in the case of Ruellia nobilis), present a presumably sphingid-adapted corolla tube of the same size and proportions, attaining a length of more than 10 This probably reflects the selective pressure of the same type of pollinator acting in this area. cm, a remarkable feature in both families. VH. Group Chiloblechum Blechum Nees subg. Chiloblechum Oersted, Vi- ensk. Meddel. Dansk pene Foren. Kjo- benhavn 1854: 69, 1854. ТҮРЕ: Blechum mexicanum Oersted (lectot here selected). P · Penstemonacanthus Nos! in C. Martius, Fl. Bras. 159. 1847. TYPE: Penstemonacanthus mo- destus Nees. Syn. nov. Flowers sessile and solitary, ebracteolate, sub- tended by the upper leaves sometimes transformed into bracts, forming terminal spikes. Calyx fre- quently zygomorphic, the lateral lobes united with the two anterior, the posterior one free. Capsules ovate in shape, with a small solid base; placentas that separate elastically from the capsule with the retinacula and the seeds, breaking the walls. Seeds The species of this group occur in southern North America and subtropical South America, in a disjunct distribution. The two South American species are herbs with a suffruticose base and pro- duce brilliant blue flowers that appear psychophil- ous. Group Chiloblechum differs markedly from the rest of the genus in the structure of the fruit, in which the placentas, with the retinacula and the seeds, separate elastically from the walls of the capsule. This character also appears in other gen- era of Acanthaceae (e.g., Blechum, Tetramerium, Dicliptera, some Justicia) that do not seem to have close evolutionary relationships with Ruellia, and therefore could be a case of parallel evolution within the family. However, the genus Blechum is morphologically closely related to КиеШа (Breme- kamp & Nannega Bremekamp, 1948), and has the same structure of the capsule correlated with the same inflorescence morphology as species of Volume 80, Number 4 Ezcurra 837 1993 Ruellia in Southern South America 5 mm FIGURE 24. Ruellia nobilis (S. Moore) Lindau. — A. Flowering branch. — B. Open capsule without seeds, schematic. (A, B, Rojas 13638.) 838 Annals of the Missouri Botanical Garden Ruellia sect. Chiloblechum, differing only in the size of the flowers and the pollen morphology. In this case, reticulate evolution could also explain the morphology of the species of group Chiloble- chum, which have characters in common both with Ruellia and Blechum. 22. Ruellia erythropus (Nees) Lindau in Engl. & Prantl, Nat. Pflanzenfam. 4, Abt. 3b: 311. 1895. Siphonacanthus erythropus Nees in C. Martius, Fl. Bras. 9: 47. 1847. TYPE: Bra- zil. Rio Grande do Sul: Porto Allegre, Sao Lourengo, Sellow s.n. (holotype, B destroyed, photo F 5920; isotype, K). m 25 Penstemonacanthus modestus dps in Martius, Fl. Bras 1847. ТҮР ден Minas Gerais: a да ad Contendas, July, Martius s.n. (holo- type, M not seen, photo F 20573). Syn. nov. Blechum tweedii Nees in DC., Prodr. 11: 466. 1847. Ruellia tweedii (Nees) T. d ex oun: & Britton, Ann. New York Acad. 92. 1893. TYPE: Argentina. Paraná? (' ied ) о, lo- cality, Tweedie s.n. (holotype, K). Syn. nov Blechum mexic anum Oersted, equae Medd. Naturhist. : 169, tab. 5, f. 1-4, 1854. С ви og paso de Dona Morro de Bo- Rancho Nuevo," Liebmann s.n. (holotype, C een). Ruellia guum Griseb., Abh. Kónigl. Ges. Wiss. Göt- . TYPE: Argentina. Salta: San rán, Laguna del Palmar, Lorentz & Hieron- ymus 550 (holotype, GOET not seen, photo SI; sotype, Rusllia matogrosse nsis Lindau, Bull. али Boissier 3: 362. 1895. ТҮРЕ: Brazil. Mato Grosso: Jacobina, Kuntze s.n. Culo, B destroyed, photo F 5936; pe, NY). S cii s Lin em "Bull. Herb. Boissier 3: 363. 95. п : Bolivia. Santa Cruz: prov. Velasco ori- uk 200 m, Kuntze s.n. (holotype, ч ње photo F 18222; isotype, NY). Syn. n Erect or straggling suffruticose perennial 20– 60 cm tall, with a ligneous base and tough, fibrous roots. Stems thin, decumbent, obtusely quadran- gular, somewhat woody and rooting at the lower nodes, generally glabrous. Leaves on slender pet- ioles 0.3-1 cm long, the blade elliptic or ovate, 2-6 cm long and 1-2.5 cm wide, acute or acu- minate, cuneate on the base, generally glabrous on the surface and with sessile patelliform glands be- neath, the margin entire. Flowers sessile, solitary in the axils of the upper leaves, in terminal, densely foliose spikes. Bracts (upper leaves) opposite and decussate, imbricate, similar to the normal leaves but subsessile, smaller and ciliate. Bractlets obso- lete. Calyx ca. 5 mm long, unequally cleft, the two lateral segments united to the two lower ones, the posterior free, forming a three-lobed structure. Co- rollas erect, blue, 2.5-3.5 cm long, the basal tube 1.5 ст long, the throat campanulate-obconic, 1— 1.5 ст long, the lobes spreading, orbicular, ca. 8 mm long. Stamens included, long filaments ca. 8 mm, short 5 mm, the anthers 3 mm long. Ovary and style glabrous. Capsule ovate, 1 cm long and 4 mm thick, the lower 3 mm solid, puberulous, glabrescent. Placentas breaking away with the ret- inacula and the seeds from the dry mature capsule. Retinacula 2-3 mm long; seeds 4-8, dark, 2-3 mm diam., with mucilaginous hairs when wet. Distribution. Mexico (Yucatan and Vera- cruz), southeastern Bolivia, Paraguay, southern, southeastern, and southwestern Brazil, and north- ern Argentina. In South America it is frequent in the understory of Chaco forest and adjacent brevi- deciduous transition forests. It flowers in winter and spring, from June to October. Representative specimens. ARGENTINA. Cata- marca: entre Los Altos y La Vina, Morello 1160 (LIL), Depto. Paclin, Cuesta del Totoral, Rodrigo 3184 (LP); Depto. El Alto, Balengua, Venturi 7069 (SI, Colonia Benitez, Cabrera 2982 (LP), Schul CTES, US); Las Palmas, Jorgensen 2327 (LIL, SI, 05); Ruta 11, Arroyo Cangui Chico, Krapovickas 13175 (LP, 05); Resistencia, Krapovickas 15846 (CTES, SI, US); Enrique Urien, Rodrigo 2702 (LP); Roque Sáenz Pena, Schulz 2497 (CTES); Depto. Güemes, La Fidelidad, Rio Teuco, Schulz 15594 (CTES), 16348 (US). Corrientes: Paso de la Patria, Cabrera 18555 (CTES); Depto. Em- pedrado, El Pollo, /barrola 3189 (LIL); Garruchos, San Juan Bautista, Krapovickas 25767 (CTES); Arroyo Chir- imay, Krapov ickas 26164 (CTES, US); Perichón, Kra- pivockas 26451 (CTES, US); Balneario Molina Punta, Schinini 11578 (CTES, US); Corrientes, costa del Rio Paraná, Tressens 784 (CTES). Entre Paraná, La Toma, barrancas, Burkart 23827 (SI); Paraná, Toma Nueva, Correa 19015 (SI); Paraná, barrancas, Fabris 47 (LIL); Paraná, barrancas orillas Parana, Meyer 10146 (LIL). Formosa: Ruta 83, 8 km N de Ing. Juárez, Correa 7670 (ВАВ); Ruta 11 y Rio Bermejo, Fabris 7 194 (CTES); Ruta 81, 32 km al N S Las Lomitas, Fortunato (BAB); 10 km 5 de Ria ] ~ p ~ 3 (LIL, US); Mojón sm E Schulz 157 (ВАВ); те. Palo nee ‘ os, Schulz 18300 (CTES). Jujuy: Vinalito, Cabrera 4 L 1. 3 (LP); El Piquete, Cabrera 14565 (LP); Cerros al SE de San Pedro, Cabrera 16057 (LP); Rio Zapla, Cabrera 20804 (LP); Camino a Valle Grande, БА ау. Cabrera 27 890 (SI); Entre Mesada de las Colmenas y Aguas Negras, Cabrera 32157 (SI); Chalicán, Сага 34102 (51); | Sola, Fabris 3133 (LP); Calilegua, Fabris 4425 (LP); Cuesta Las Lajitas, Kiesling 99 (LP); Santa Clara, Venturi 9658 (LP, MO). Misiones: San Igna acio, Birabén 5284 (LP); Apostoles, Escuela Agrotécnica, Cabrera 28503 (SI); Candelaria, Santa Ana, Montes 1572 (LIL, 05); Salto Iguazü, Puerto Aguirre, Rodriguez 353 (BA); Puerto Aguirre, Rojas Volume 80, Number 4 Ezcurra 839 9 Ruellia in Southern South America ~ (f \ “ie УУ) | Y И y у f Y ‘ A й 4 \ Ys IN 2. Ay | p ANS Li A CSS үш e А Ж E => =: SES INS TA УЗУЫ), í NS LEE “ыз 4 TS 5, 5mm FIGURE 25. Ruellia erythropus. —A. Habit.—B. Abaxial leaf surface with cystoliths. — C. Inflorescence apex showing calyces. — D. Calyx. — E. Open calyx. — Е. Open corolla, showing stamens. — С. Dehiscent capsule with bioken walls, without seeds, schematic. — H. Capsule valves with placenta, retinacula and seeds separating. — I. Open capsule without seeds, the placenta separated. — J. Seed. (A-H, J, Cabrera 32157; I, Cabrera 28503.) Annals of the Missouri Botanical Garden 8073 (LIL); Bernardo de Irigoyen, Schwindt 4201 (LIL Salta: Orán, Cabrera 4575 (LP); Tartagal, ruta a Pozo de la Cuchara, Correa 7726 (BAB); Rivadavia, 15 km al S de Palmarcito, Cuezzo 695 (LIL); Parque Nacional Rey, Ezcurra 37 1 (SI); General Ballivian, Maldonado 764 (LP); La Caldera, Rosa 183 (51); Joaquin V. Gon- zalez, Ro 1064 (LP); Taratagal, Zanja Honda, Schrei- ter 3429 (LIL); Depto. Anta, La Salada, West 412 (LIL). Santa Fe: Villa Guillermina, Meyer 2999 (LIL); 2 km E de Villa Guillermina, Pensiero ы кн Mocovi, Ven- turi 18 (LIL, LP, SI, US). Tucu Horco Molle, е 653 I (LP); Tafi, Villa RUM. Lillo ое Тта , Maldonado 401 (LP); Burruyac mada, Roc lo 496 (LIL, US); Quebrada de Lules, y жене 2292 (BAB, LIL, LP, SI , US); Santa Rosa, Venturi 353 (LIL, SI, US); Tafi Vie} ( IL); Yerba Buena, Vervoorst 8623 (LIL). 3 guacu, Paraná, Me Lag es 559 (МВМ); dido Rondon, Corrego R. Branco, Hatschbach 43996 (MBM, ER Mun. Roncador, = Hanc hbach 32901 (MBM, US). Rio Grande do Sul: Ко Jacuhy, dn muller 637 (GH); Balneario Ivai, Arzivenco s.n. (IC Parque do Balneario Sed Torgo 53 (US). MEXIC o Veracruz: Paso del Macho, Rancho Paso Grande, Cal- = "dd (IPN); Mun. "De Rios, к Ventura 2 (IPN); Mun. Naolinco, La Meza, Ventura А. 157 A РМ). Yueatan: Chichén Itza, Paray 1522 (IPN); Mérida, Cruz de Gaher, Seler 3948 (K, US); n Valdez 24 (K). PARAGUAY. Boquerón: Col. Fer heim, 30 km W de Filadelfia, August 111 (СТЕ). Canendiyu: in campo Ipé Hu, Hassler 5309 (С); in regione yerbalium de Maracayü, Hassler 4203 (AA, BM, G, NY, P, MO). Central: La Tune ieri 2458 (С, P); in regione lacus Ypacaray, Назз G, GH, LP, MO, NY, US): Villa Elisa, ye 6. — (AA, LP). Chaco: Parque Nacional Defensores del Chac Fernandez n 4365 (С); Rio Verde, Пода 5024 (СН, MO, US) Mayor Pedro Lageren za, Schinini & Bordas ME » (CTES, 05). Concepcion: San Salvador, Balansa 2458a (P); zwischen Río Apa und Rio Aqui- — Puerto Riss 580, е 4048 (ВМ, С); ргоре Соп- — T = Haake 312 1 (АА, BM, G NY, Р). Guai el Monte Haiti, Jorgense n 4306 (GH, LIL, LP, MO, NY, RB, SI, US). Paraguari: Paraguari, s 24 58b (P): Macizo Acahay, Zardini 6633, 6927 (MO, US). Presi- ES te Hayes: in regione cursus inferioris fluminis Pil- mayo, Rojas 156 (BM, G); Pilcomayo River, Morong 967 (BM, G, GH, US); Estancia Pozo Favorito, Spichiger 1506 (G). San Pedro: Colonia Primavera, Woolston 570 (GH, LIL, SI, US). Ruellia erythropus is characterized by the blue flowers in terminal spikes, with the leaves sub- tending the flowers subsessile and ciliate, somewhat morphologically different from those on the lower vegetative portion of the shoot, and therefore ap- pearing like bracts. Also, the placentas separate elastically from the capsule with the retinacula and the seeds. In these characters it approaches the closely related genus Blechum, which differs in pollen morphology. Ruellia erythropus has an extremely extended range in subtropical South America, and a curious disjunct distribution, having been described also for Veracruz and Yucatán, Mexico, as Blechum mexi- canum. Although I have not seen the type speci- men of B. mexicanum, 1 have studied material from Mexico to verify this synonymy (see list of representative specimens). The name Ruellia er- ythropus precedes the name R. tweedii, with which this species had been cited by most previous authors (e.g., Ariza Espinar, 1983; Dawson, 1979; Leon- ard, 1936; Morong & Britton, 1892-1893). The extensive synonymy of this species reflects the wide range it inhabits and its associated morphological variation. It is somewhat similar to Ruellia bahien- sis from northeastern Brazil (of group Ebracteo- lati), but differs by the inflorescence and capsule structure. Several herbarium specimens of Ruellia erythropus from Paraguay and Argentina have been erroneously identified as Ruellia bahiensis. Although the genus Penstemonacanthus was described by Nees (1847a) based on material sup- posedly having five stamens, it is an exception, not only in this species but in all the genus Ruellia, and probably a teratology. The rest of the mor- phological characters of the original description of Penstemonacanthus modestus clearly agree with those of Ruellia erythropus. Although I have not investigated the structure of the capsule in the photograph of the type, it has been clearly de- scribed by Nees in both the genus and species descriptions. 23. Ruellia hypericoides (Nees) Lindau in Engl., Bot. Jahrb. 19, Beibl. 48: 16. 1894. Dipteracanthus hypericoides Nees in C. Martius, Fl. Bras. 9: 39. 1847. SYNTYPES: Brazil. Rio Grande do Sul: without locality, Sellow s.n. (B destroyed, photo F 5931; iso- D i Pad bolo. Gaudichaud 244 (P). F Delicate herb 15-25 cm tall, from a small basal rhizome with fibrous roots. Stems weak, erect or ecumbent, terete, shortly appressed-pubescent, with long internodes. Leaves shortly petiolate, ovate or orbicular, 1.5-2 cm long and 0.8-1.3 ст wide, obtuse or rounded in the apex, truncate in the base, the lower ones smaller, all puberulous on top and densely dotted with sessile patelliform glands on the surface beneath, the margins entire. Flowers sessile, solitary in the axils of the upper leaves. lete. Calyx 5 mm long, the segments linear-lanceolate, 4 mm long and 1 mm wide, acute, appressed-puberulous and ciliate. Corolla erect, blue, 2-3 cm long, the basal tube 1.2-1.5 cm long, the throat campanulate-obconic, 5-6 mm long and 4– 5 mm diam., the lobes spreading, suborbicular, ca. Volume 80, Number 4 1993 Ezcurra Ruellia in Southern South America ;URE 26. Ruellia hypericoides (Nees) Lindau. — placenta and the retinacula separated. (A, B, Burkart 14181. 8 mm diam., somewhat emarginate. Stamens in- cluded, ca. 7 mm long, anthers 2 mm long. Ovary and style puberulous. Capsule ovate, 7-9 mm long and 4 mm thick, puberulous, the base solid. Pla- centas breaking away with the retinacula and the seeds from the dry mature capsule. Retinacula 1.5 mm long; seeds 4-8, brown, 2 mm diam., with mucilaginous hairs when wet. Southern Brazil, southern Par- aguay, northeastern Argentina, and northern Uru- Distribution. 5 mm A. Habit. — B. Open capsule without seeds, schematic, the La guay. Ruellia hypericoides is found in open, grassy campos. It flowers in spring and summer, from November to February. Representative specime ns. ARGENTINA. Corri- entes: pens rw ancia San Miguelito, 20 km NE de Playadito, R , Carnevali 5044 (CTES); Ituzaingó, Estancia E ы ^a 6219 (BAB); Santo Tome, Estancia Garruchos, Cachuera, Arroyo Chirimay, Kra- povickas 21215 с. 21216 (CTES); Santo Tomé, R.40 y Arroyo Chirimay, Krapovickas 10682 (CTES); Santo Tomé, Estancia Garruchos, Krapovickas 21407 (CTES); 842 Annals of the Missouri Botanical Garden Ituzaingó, Ruta 39, 30 km N de Virasoro, Krapovickas 25341 (CTES); Ituzaingó, San Carlos, Martínez Crovetto m Ituzaingó, Estancia Puerto Valle, Partridge n. (LP); Santo Tomé, Estancia Garruchos, Pedersen 985 (AA), 5434 (AA, ICN, LP). Misiones: Apostoles, Pindapoy, Birabén 5396 (LP); Posadas, Burkart 14181 (SD, Martínez aa 8153, 8164 (BAB); Concepción, а Martínez Crovetto 6207 (ВАВ); San Ignacio ca, Mar rtinez Crovetto 8621 (ВАВ); Poesias: Meyer 6010 (LIL); Frontera, San Antonio, Montes 7074 (SI, US); San Ignacio, Santo Pipó, Schulz 7213 (LIL); Posadas, Spegazzini 19397 (SI); Depto. Apostoles, Ar- royo Chirimay, balneario Azara, Tur 1927 (SI); de Após- toles a Concepción de la Sierra, Arroyo Las Tunas, Zu- loaga 3248 (SI). BRAZIL. Rio Grande do Sul: Sao Luiz-Caaro, Buck 11356 (LIL); Sao Salvador, Santo In- acio, Henz 35695 (LIL); 20 km SE de Santa Rosa, Lindeman 8990 (ICN); Santo m Palacios-Cuezzo 1734 (LIL); Carazinho, Pereira 6675 (RB); Quarai, Es- tancia de Jarau, Rambo 26077 (LIL). dS Mi- siones: 12 km W de San Ignacio camino a Pilar, Агбо 1885 (CTES). URUGUAY. Artigas: Ruta 30, Arroyo Catalan, paso Piaui, Rosengurtt В7173 (МУКА, US). Ruellia hypericoides seems closely related to Ruellia erythropus, but inhabits more open areas and is a more delicate herb with smaller flowers. The name R. hypericoides has been erroneously used by many authors to identify material of Ruel- lia bulbifera of group Ebracteolati. Ruellia bul- bifera differs by the robust, tuberculate xylopo- dium, larger flowers with a short basal tube, and different structure of the capsule. EXCLUDED SPECIES Ruellia lanceolata, Chaetochlamys rusbyi, and Beloperone matthewsii are different names for the same species, all based on different types. The genus in which this species should be placed is Justicia, and the correct name for it is Justicia rusbyi. Justicia rusbyi (Lindau) V. A. W. Graham, Kew Bull. 43(4): 605. 1988, not Justicia lanceo- lata (Chapman) Small, 1933. Chaetochlamys rusbyi Lindau, Bull. Herb. Boissier 3: 491. 1895. ТҮРЕ: Bolivia. La Paz: Guanai, Rusby 1117 (isosyntype, BM); Santa Cruz: without locality, 380 m, Kuntze, 1892 (isosyntype, NY) УН POS Morong, Ann. New York Acad. Sci. 93 3. TYPE: Paraguay. Paraguari: Central a uay, between Pirayú and Jaguarón, 8 Apr. 1889. Morong 667 (holotype, NY; isotype, US). Syn. nov. Beloperone matthewsii Lindau, Bull. Herb. Boissier 6, : p. 1: 30. 1898. svNTYPE: Paraguay. Cordillera: n dumeto prope Cordillera de Altos, Hassler 1936 (С; isosyntypes, K, P, NY). LITERATURE CITED AHMAD, K. J. 1978. Epidermal hairs of Acanthaceae. a 24(1): 101- L. е cul- 265. BRINK. 1965. ‚капшады. In: Flora of Java 2: 544-593. Gróningen, bes Netherlands. BARKER, R. 986. A taxonomic revision of Austra- lian uns Рана J. Adelaide Bot. Gard. 9: 1-286. 876. Acanthaceae. In: С. Bentham € ooker, Genera Plantarum 2(2): 1060-1122. BREMEKAMP, "C. E. B. 1938. а m In: A. Pulle, Flora of Surinam 4(2): 1 25 1969. An d te of the Acanthaceae collected by Miss W. M. A. Brook to Bolivia. Proc. Kon. Ned. A | C, 72(4): 420-430. РА in Meded. Bot. Mus. Herb. Rijks. M ign _ 331: 420-430. A ees. —— — — & М. E. МАММЕСА B BENTHAM, С. REMEKAMP. re- Wetensch., Afd. Natuurk., Tweede se sect. 45(1): 1- Britton, N. L. & A. Brown. 1913. of Northern United pci 2nd ed. 3 vols Scribner's Sons, New ii A. 19 а indigenas como for- eras de emergencia. Darwiniana 6: 192-202. Сайн, A. L. 76. Regiones Fitogeográficas Argen- tinas. /n: Enciclopedia Argentina de Agricultura y Jardinería, ?da. edición, 2 (1). Editorial Acme, Bue- Illustrated Flora ls. Charles Vien 1980. x apo ia de América а Cient. O.E.A. 2da. edición. General ыык of the ao of American States, Washington, D.C. CAVANILLES, А. J. 1800-1801. Icones et descriptiones 01-600. Снорат, R. & E. HassLeER. 1903. Acanthaceae. T Plantae Hasslerianae. Bull. Herb. Boissier Ser. 2 39 628 DANIEL, T. F. 1986. Systematics of Tetramerium Ca Syst. Bot. Monogr. 12: 1-1 ew, recon бай, and little- agita Mexican ne ies of d road Contr. Univ. rd ki 17: – 162. TT & M. ^ Ba KER. 1984. Chro- mosome numbers Vane their systematic implications in some North American Acanthaceae. Syst. Bot. 9(3): 346-355. , T. I. CHuanG & M. A. BAKER. 1990. Chro- mosome numbers of American Acanthaceae. Syst. Bot. 15: 13-25. Dawson, G. Acanthaceae. In: A. Burkart (di- rector), Flora Пинкија de Entre Rios (Argentina) 5: 45-516 DIMITRI, M. J. Descripción de las Plantas Cul- das. In: Enciclopedia Argentina de Agricultura y 78. Acanthaceae. /n: R. E. Woodson (editor), Flora of та ei a ве 177. Апп. Missouri Bot. Gard. 6 1986. alt ae ли W Bur rger (editor), Flora а ез Family 200. Fieldiana, Botany, n.s. 18: 1-87. Volume 80, Number 4 Ezcurra 843 Ruellia in Southern South America Everett, T. H. 1982. The New York Botanical Garden Illustrated Encyclopedia of Horticulture, Vol. 9. Gar- land Publishing, New Yor EzcuRRA, C. 1989. Ruellia sanguinea (Acanthaceae) y especies relacionadas en Argentina, Uruguay y sur de Brasil. ча 29: 269-287. Е AZKUE. 1989. Validation and genetic and ина э дез relationships of Ruellia macro- solen (Acanthaceae) us oe South America. Syst. Bot. 14(3): 297 — & D qos New species of Ruellia (Acanthaceae) from кш South Ameri- ca. Brittonia 44(1): 6 3: FAEGRI, К. & L. VAN DER os 1979. The Principles of Pollination Ecology. Pergamon Press, Oxford. Gentry, А. H. 1 Sphingiphila МА а new genus for the Paraguayan Chaco. Syst. Bot 277-279. Gipson, D. N. 1974. Acanthaceae. In: P. Standley, L. O. Williams & D. N. Gibson (editors), Flora of Gua- temala. Fieldiana, Botany, 24(10): 328-461 GRANT, V. & K. A. GRANT. 1965. Flower Pollination in the Phlox Family. Columbia Univ. Press, New York ork. GRANT, W. F. 1955. A cytogenetic study in the Acan- thaceae. Brittonia 8: 121-149. GREUTER, W. 19 International Code of Botanical aor Regnum Veg. 118. GUERRANT, Е. 1 Neotenic evolution of Del- phinium duh (Ranunculaceae): A qo ming- bird-pollinated larkspur. AOS 36(4): 6 =712. Hueck, К. bosques d “б composición e importancia económica. Soc. Coop. Técn. Ltda., Repüblica Federal Alemar KARLSTRÖM, P.-O. 1978. Epidermal leaf structures in dg of a and Petalidieae (Acantha- ae). Bot. y T31; -433. Шш, E. C. js The Acanthaceae of the Yucatan Peninsula. Publ. Carnegie Inst. Washington 461: 1951-1958. The Acanthaceae 1 Colombia. -78 1. Enzyine etching ment as an aid in the study of seed surface sculpture in Justicia and Ruellia (Acanthaceae). Bot. J. Linn. Soc. 105: 285-288 Litto, M. 1937. irs жш Argentinas. Lilloa 1: 2 "a 9 ta Linpau, G. 18 Beit itrüge v и flora. Bot. Jahrb. um 19, Beibl. 8-23. 1895. Acanthaceae. la: | ж Н. Prantl, e Natürlichen ШЕ ОВЕ 4(3b): 274-354. C. Genera Plantarum. Conradum a 3. Species Plantarum, ed. 1. Laurentii SONA Stockholm ниш ^n 1924. : 93-1 1964. уже per investigations in South a (Acanthaceae). Amer. Car Lindau. Ber. Deutsch. Bot. 08. LoNc, R. W. Florida “о a ~ ue J. Bot. 51(8): 8 19 ER interspecific hybridization ir in Ruellin ТНЕУ Amer. J. Bot. 53: 917-92 1970. The genera of E RN in south- eastern United States. J. Arnold Arbor. 51: 257- 3 1971. Floral polymorphy and amphimictic breeding pee in Ruellia carolinensis (Acantha- eae). Am ud Bot. 58(6): 525-531. 1 73. А biosystematic approach to gene delimitation i in Ruellia gei ылаң Taxon 22(5/ 6): 5 p “Artificial interspecific hybridization in temperate and tropical species of Ruellia. Brittonia 27: 289-296. 1976a. пина у of depen ur L. (Асаш сени); rJ. ке 63(7): 9 19765. Sane се ше оп Aphragmia inundata тишүү быр from Mexico. Rhodara 7 8: 17-24. Гокр, E. M. 1981. Cleistogamy: A tool for the study of floral morphogenesis, jer and evolution. Bot. Rev. (Lancaster) 47(4): 421-449. MacNAIR, M. R The potenti | rapid speci- ation in plants. Genome 31: 203- MonoNc, T. & N. L. BRITTON. Te m An enu- meration of the plants collected by Dr, Thomas Mo- a in un 1888-1890. Ann. New York Acad. Sci. 7: 45-280. NEES, C. E LEN Acanthaceae. /п: C. Martius (ed- itor), Flora Brasiliensis 9: 1-164 1 b. Acanthacaeae. In: A e Candolle (editor), Prodromus Systematis Ма Regni Ve- getabilis 11: 46-519. . Mexicos og нен . Vidensk. Meddel. Dansk Natur Foren. Kjobenhavn 1854: 113-181. ORNDUFF, R. We lcs in relation to systematics. Taxon 18: -24 PETRIELLA, 1 El in de ” Ас e ar- gentinas. Resina. Mus. La Plata, secc. Bot. 11: 51- 6 PIOVANO, M. A. & L. M. BERNARDELLO. 1991. Chro- mosome numbers in Argentinean Acanthaceae. Syst. Bot. 16(1): 89-97. , L. GALETTO & L. BERNARDELLO. 1991. Biol- орла ПА еп Ruellia brevifolia (Acantha- ceae). In: s, XXIII Jornadas Argentinas de Botánica, Socieda ej Акаш de lm nica. C. R.U.B.- U.N.C., San Carlos de Bariloc Raj, B. 1961. Pollen caca ten] studies in the Асап- thaceae. Grana Palynol. 3: 1-108, plates 1-44. 1973. dinh idit ii gum pollen mor- J. Palynol. 9: 91-14 Ruellia section Chro Mexico phology of the Acanthaceae. дате а Т. Р. 1991. »hila (Acanthaceae): А novelty from жы т Linn. Soc. 107: 79-88. 1992. Ruellia sect. Urceolata (Acanthaceae), a navel species e from southern Mexico. Pl. Syst. Evol. 180: 22 29- RAMBO, B. rn коле; Riograndensis. Iherin- gia, Botany 12: 1-36 Rizzini, С. Т. 1947. Estodis sobre as Acanthaceae. Bol. б Nac. Rio de Janeiro, Botany 8: 1-3 49. C ое ao conhecimiento da nih Just ae. Arq. Jard. io de Janeiro 9: 37-67. SANTOS ио J. 1963. [NE dion sobre la floraci de Acantácea indigena Ruellia lorentziana Gris. Darwiniana 12( 61-663. SCOTLAND, A systematic analysis of pollen a with contorted oe . H. Barnes b Pollen and Spores Syst. Assoc. ; Special Vol. 44 = 289. Clarendon 1 Presa Oxfor 844 Annals of the Missouri Botanical Garden SELL, Y. 1969a. Les complexes inflorescentiels de quel- ees. Etude particuliére des puse: menes de аи оне de racemisation omo généisation et de troncature. Ann. Sci. Nat. Bot. ser. 12, i 225-300. 69b. La dissemination des Acantha variations sur le type xérochasique al . Gén. _Bot. 76: ae z Ruellia lorentz ia- na Griseb.. et pee autres " Acanthae ées. Ber – 147. . 1949. . Amer. Midl. Naturalist 42: TURNER, n. L Texas species of ы: (Асап- thaceae). rap 71(4): 281-299. J T. 1938. Cleistogamic flowers. Bot. Rev. 9. per Ruellia -86. Radiación adaptativa del sindrome floral eu las familias neotropicales. Bol. Acad. Nac. Ci. 59: 5-30 Wass, D. C. 1966. Acanthaceae. /n: Flora of B. SMITH. 1969. Acantaceas. In: R. Reitz (editor), Flora Ilustrada € atarinense, ACAN: 1-134. Herbario m Rodrigues, Itaj WEBERLING, F. Morphology ‘of flower and inflo- rescences. ae Press, Cambridge, United Kingdom ipee TO SCIENTIFIC NAMES. Ас ia aa names in roman type, synonyms in italics. For explanation of names in eae see under III. Group Hyerophiloidei (p. 820). Aphragmia Nees 798 haenkei Nees 798 ире Nees . 798, 800 iliatiflorum (Hook. ) Nees 810 ds egans (Poiret) Bremek. 800 formosum dn r.) Nees 800 glabrum Nee 798, 800 kuntzei i (Lindau) Bremek. 808 mic M Ar inns ees 812 silvaccola var. montanum Nees s. 800 о matthews Linda 842 Blechum Nees subgen. Chiloblechum erste 198, 836 mexicanum Oersted 836, 838 tweedii Nees 838 Chaetochlamys rusbyi Lindau 842 C i-i Miers 798, 830 a Mier 798, 830, 833 pia ‘News 798, 810 acaulis Nees 819 angustifolius Nees 812 barbadensis Nees 798, 810 udus Nees 817 Cyrthacanthus corymbosus Martius ex Nees ........ 804 un ees 98, 822 о Nees 820 (С abro ost) Nee 822 (Genuini: Brac la) Nees 822 (Genuini: Ebracteolati) Nees 830 angustifolius 833 brachysiphon Nees 822 brevicaulis Nees 824 calvescens Nees 828 canescens Nees dissitifolius Nees var. м Меез geminiflorus (HBK) N var. angustifolius Nees var. erectus Nees var. hirsutior Nees r. procumbens Nees ubacaulis Nees glanduloso-punctatus Nees humilis Nees humilis var. minor Nees hypericoides Nees lamiiformis Nees multifolius Nees porrigens Nees porrigens var. triflorus Nees prostratus Nees a Nees tuberosus Мен vindex Nees viscossisimus енин En hotomus Kuntze Furyc ; Nee << eo о. Nees Сутте тек d is Nee TAMEN aa Gente d eminiflorus (HBK) Oersted ____ а lanceolata (Chapman) Small sbyi (Lindau) V.A. W. Graham Lyc Ren та hygrophilus (С. Martius) Bremek. reme Penstemonac е Nees 198, 836, 838 7 Rue 98 Group. C hiloblechum 836 oup Dipteracanthus 822 Group Ebracteolati 830 Group Hygrophiloidei 820 Group Physiruellia 800 Group Ruellia 810 Group Salpin thus 836 Sect. сенын us (Nees) Lindau ............... 822 Sect. Euruellia Lindau 810 Sect. Physiruellia Lindau 800 Ser. Eglandulosae Lindau 800 acaulis R. Br. 819 acutangula Nees 800 amoena Nees 804 angustiflora (Nees) Lindau ex Rambo _____. 804 angustifolia Swartz 812 brachysiphon (Nees) Lindau ci 822 brevicaulis (Nees) Lindau 824 brevifolia (Pohl) Ezcurra 802 bulbifera Lindau 830 calvescens (Nees) Lindau 828 canescens (Nees) Lindau 833 а Ноок. 810 оегшеа Могоп 812 8 coerulea Morong X brevicaulis (Nees) Lindau 814 dichotoma Sessé & Moc. pale ae en Hieron. egans Poire 800 Bc aulos rd ex Ezcurra _____----. 820 erythropus (Nees) Lindau 838 formosa Andr. 800 Volume 80, Number 4 Ezcurra 845 1993 Ruellia in Southern South America formosa HBK 800 solitaria Vell. 828 geminiflora HBK 830, 831 spectabilis Britton 812 glanduloso- зна (Nees) Lindau .................. 828 tuberosa L 798, 810 raecizan 802 tweediana Griseb. ex Fernald | hirsuta Vell. 833 tweedii (Nees) T. Anderson ex Morong & hygrophila C. Martius 817 ritton 838 hypericoides (Nees) Lindau 840 velascana Lindau 838 ignorantiae Herter 812 ventricosa HBK 802 kleinii Ezcurra € Wasshausen .. .822 viscossisima (Nees) Lindau 828 kuntzei Lindau 808 M p pr S. Moore 198, 836 lanceolata Morong 842 зођш S. Moo 798, 836 lilacina Hoo 830 Sc © alyx Nees 798 lindmaniana Fritsch 833 exicanus Nees 798 loefgrenii Lindau 830 Scordoxylum Nees 798, 800 longifolia (Pohl) Griseb. 802 hartwegianum Nees 198, 800 longipedunculata Lindau 806 Siphonacanthus Nees __ 798, 822 longipedunculata Lindau x brevifolia erythropus Nees 838 Pohl) Ez 808 villosus 198, 822 ! tziana Gri 810 — Sphingiphila tetramera Gentry 836 macrosolen Lillo ex Ezcurra = 815 Stemonacanthus Nees 798, 800 magniflora Ezcurr 835 salviaefolius Nees 198, 800 matogrossensis 838 — 1 sum Pohl 798, 800 microphylla (Nees) Lindau 812 iflorum Nees 804 morongii Britton 819 Be бит Pohl 802 multifolia (Nees) Lindau 826 brookeae Bremek. 806 var. viscossisima (Nees) Ezcurra _______- 828 оа Pohl 198, 800 nobilis (S. Moore) Lindau 836 randrum Bremek. 802 pubiflora Griseb. 838 ея (НВК) Nees 802 reitzii M & Smith .. .802 Tremacanthus S. Moore .. 798, 830 sanguinea Gris 808 roberti 5. Moore 798, 830 sceptrum-marianum (Vell.) Stearn 00000000... 800 (ега Brem 798, 830 serratitheca Rusby . 802 јен уже (HBK) Bremek. ____ . 798, 830 ENDEMIC HERBACEOUS BAMBOO GENERA OF CUBA (POACEAE: BAMBUSOIDEAE: OLYREAE) Fernando O. Zuloaga, Osvaldo Morrone, and Emmet J. Judziewicz? ABSTRACT A morphological, anatomical, and taxonomic study of the endemic herbaceous bamboo (Poaceae: Bambusoideae: Olyreae) genera Ekmanochloa and Mniochloa (including its new segregate Pire siella) is presented; the new combination fusoid cells in Ekmanochloa and Mniochloa and the presence of compound epidermal papillae in Ekmanochloa. The three genera are compared with possible close relatives within the Olyreae. Cuba has a rich grass (Poaceae) flora numbering at least 400 species, with 69 of these endemic (López Almiral et al., 9). There are four en- demic genera: Lepturidium Hitchc. & E. Ekman (Cynodonteae), Triscenia Griseb. (Paniceae), and the herbaceous bamboo (Bambusoideae) genera Mniochloa Chase and Ekmanochloa Hitchc. The latter two taxa are the only endemic bambusoid genera in the entire West Indies and each is known from only a few old collections. Habitats include pine woodlands, streambanks, and limestone cliffs in the provinces of Habana and Pinar del Rio (one species) and Oriente (the other three species), in- cluding serpentine outcrops. In this regard, it is noteworthy that Brooks (1987) reports 511 en- demic serpentine vascular plant species for Cuba. As part of a long-term project, the goal of which is to monograph all New World bamboo genera, the following study of the morphology, anatomy, and taxonomy of these endemic "dwarf bamboos" is presented. Mniochloa, named for its resemblance to the moss genus Mnium L., was established by Chase (1908) on the basis of two species that were de- scribed in the previous century in the genera Olyra L. and Digitaria Haller. She included the group in the Paniceae, while later authors placed Mnioch- loa in the Olyreae within the Pooideae (Hubbard, 1934), or in the Phareae (along with other her- baceous bamboos) within the Panicoideae (Bews, 1929; Roshevits, 1937). Recent authors, starting with Calderón & Soderstrom (1967), have all placed Mniochloa (and Ekmanochloa) in the Olyreae within the Bambusoideae (Calderón & Soderstrom, 1980; Clayton & Renvoize, 1986; Soderstrom & Ellis, 1987; Soderstrom et al., 1988; Watson & Dalwitz, 1988). Both M. pulchella (Griseb.) Chase and M. strephioides (Griseb.) Chase are small, delicate, trailing, almost mosslike grasses that are reminiscent of the South American olyroid genus Raddiella Swallen, or, because of the dimorphic culms with lax, trailing flowering shoots, Piresia Swallen. On closer examination of the two taxa, however, we noted a number of significant differ- ences in morphology and leaf anatomy. Except for the large, heterogeneous group Olyra L. (Soder- strom & Zuloaga, 1989), olyroid genera tend to have nearly identical female spikelet morphologies; however, the differences between the two Mnioch- loa species are great and include features of floret indument, stipe and apex morphology, and relative glume length. Anatomically, M. pulchella lacks the characteristic bambusoid fusoid cells, whereas M. strephioides has them. Because of these and other features (discussed in greater detail later and summarized in Tables 1 and 2), we have decided to erect the new segregate genus Piresiella based on P. strephioides. Other characters that had ap- peared to unite the two species are also found in more distantly related genera in the Olyreae; for ! We tha nk Vladimiro Dudas for the excellent illustrations, Tarciso Filgueiras for the Latin diagnosis, Luis Catasüs Guerra for information on the recent status of these grasses in Cuba, and curators of various herbaria for the loan of spec imens ? [nstituto de Botánica Darwinion, Casilla de Correo 22, 200 Labarden, 164.2 San Isidro, Argentina. * Department of Botany, Birge Hall, University of Wisconsin, Madison, Wisconsin 53706, U.S.A. ANN. Missouni Bor. Garp. 80: 846-861. 1993. Volume 80, Number 4 1993 Zuloaga et al. 847 Bamboo Genera of Cuba TABLE 1. Comparison of Mniochloa with Piresiella. Character Mniochloa Piresiella Morphology Leaf blade Elliptical; base cuneate, symmetrical Ovate-triangular; base truncate, morphology to slightly asymmetrical strongly asymmetrical Flowering culm morphology Female glume morphology Internode between Not prominent female upper e glume and floret Female floret apex Obtuse to acute Female floret Glabrous indument Male spikelet apex Glabrous Stamen number 3 Leaf blade anatomy Fusoid cells Absent Papillae Solitary Intercostal short ial epidermis) Erect, exceeding the vegetative culms Shorter than the floret, obtuse to Densely papillose throughout Lax, shorter than the vegetative culms Longer than the floret, usually Prominent Apiculate Densely pubescent Short-pilose 2 Present Weakly papillate between sto- mates; otherwise smooth Paired example, paired, conjugate racemes are found in Ekmanochloa; leafless flowering shoots, in all spe- cies of Piresia and in species of Olyra (O. ecaudata Doell and others) and Cryptochloa Swallen (C. decumbens Soderstrom); corms are found in Rehia еп. Ekmanochloa (two species) was also examined in morphological and anatomical detail. Based on such distinctive features as the deciduous leaf blades, long-awned female florets, the absence of leaf blade fusoid cells, and the presence of compound epi- dermal papillae, we decided that a thorough study of both E. aristata Hitchc. and E. subaphylla Hitchc. might shed light on their tenuous placement in the Olyreae. Distribution maps for all four taxa treated in this paper can be found in Soderstrom et al. (1988). MATERIALS AND METHODS Herbarium specimens were examined from GH, MO, NY, and US (including type fragments from other herbaria). Transverse sections of dried leaf blades were prepared after desilification in 30% hydrofluoric acid (Breakwell, 1914); the sections were then stained in safranin and fast green. Ab- axial epidermal scrapes of leaf blades were pre- pared by removing the mesophyll and vascular tissue with a scalpel and camel hair brush following the method of Metcalfe (1960). The epidermis was then stained in safranin. The standardized termi- nology of Ellis (1976, 1979) was used to describe the anatomical structure of the leaves. The vouch- ers for this study are marked with an asterisk in the specimens examined section of the taxonomic treatment. LEAF ANATOMY MNIOCHLOA PULCHELLA Leaf blade in transverse section (Fig. 1A, B). Outline. Expanded, flat; arms of the lamina straight, the two halves symmetrical on either side of the median vascular bundle; thickness at mid- lamina 32-52 um. Ribs and furrows. Adaxial and abaxial ribs and furrows indistinguishable. Median vascular bundle. Midrib with simple, first-order vascular bundle distinguishable structurally from other first-order vascular bundles; adaxial projec- tions inconspicuous; abaxial projections slightly convex. Vascular bundle arrangement. Three first- 848 Annals of the Missouri Botanical Garden I еоџешу цупос UJSYLION é eqn’) ul9]99 M + С ayejnoidy озона + E + + Ӯ воџешу ujnog + (о Ауаолоош аш -) + eqn’) ul9]ste'] € 31n9e / 951140) snoIqel[5 + + eqn’) ula1seg ес pauae-Zuo'] snolqe|*) 8 eouosury yinos pue [ециә”) + € a3MIY snoiqe|‘) + + (—) (+)— еопәшү YMS pue јецигђ + sarads jo 19qUINN uonnquasig зрео prosn y suaurejs Jo 1aquinN xade цәлор o[euro y juaumput 19104 o[euro Y adns 1910y зјешод s92uo2sa1 -орш [enxastun paned Алејихе заопаоволојиј јешшлој so2uooso1ogu[ ојошеа e 10 aurooel e IDUIISIJO YU] sui[no олцејадол ULY 193U0| sur[no SNOJAJLIO] 4 эщЧлошр sunno) yuaseid suro Doy D]]9189114 DISIMA DOY 90M A] DO]Y90UDUY Y nop[2o1d Ало) 7]]234944F juasqe :— ‘juasaid :+ WLI :y te[orued :q 'eeaiÁA[Q) eui ш e19u92 ројејол Á[oAu?ind чим рујәтѕәл,у рив *popyoorupy *nopyoounu; jo uosueduir) 'z ATV], Volume 80, Number 4 1993 Zuloaga et al. Bamboo Genera of Cuba 200 jum E—————4À FIGURE 1. of lamina showing midrib with first-order vascular bundle (arrow). — а чиба ТӨ ҮЛҮ ч, 3v die да ss тилу viuo. S PN Meet: у vESEL EET URN EL АЎ Leaf blade anatomy, transverse sections. — A. Mniochloa pulchella (Wright 3448), lateral portion = Mniochloa pulchella (Wright 3448), portion of lamina with third-order vascular bundle. — C. Piresiella strephioides (León 4572), lateral portion of lamina showing midrib with first-order vascular bundle (arrow). — D. Piresiella strephioides (León 4572), portion of lamina with third- order vascular bundle. — E. of lamina with third-order vascular bundle. —G showing midrib with first-or lamina showing midrib with third-order vascular bundle. order vascular bundles and 17 third-order vascular bundles in the entire blade; 6 third-order vascular bundles between consecutive first-order vascular bundles; all vascular bundl y located. Vas- cular bundle description. First-order vascular bundles circular in outline; phloem adjoining the inner bundle sheath; circular metaxylem vessels manochloa aristata (Clement & midrib with first-order vascular bundle (arrow). — F. Ekmanochloa aristata (Clement & i —G. Ekmanochloa subaphylla (Ekman 9870), lateral portion of lami der vascular bundle (arrow). —H. Ek || Chrysogone 4563), lateral portion of lamina showing one 4563), portion na manochloa subaphylla (Ekman 9870), portion of narrower than the outer bundle sheath cells; third- order vascular bundles angular in outline with xy- lem and phloem indistinguishable. Vascular bundle sheath. Outer bundle sheath of the first-order vas- cular bundles rounded, with abaxial or adaxial in- terruptions, lacking extensions, comprising 6–8 in- flated, rounded cells; chloroplasts absent; inner 8 50 Annals of the Missouri Botanical Garden Se vU ^g e Је pere & cn LAG ER CU mper Eum Ue TENER UL» м ri TI перје энш rue) T MANARE SA У -— ~ + IGURE. 2. Leaf blade epidermis in surface view. — А. Mniochloa pulchella (Wright 3448), abaxial Marais zonation with regularly spaced costal zones. — В. Mniochloa pulchella dois 3448), detail of A with hooks, microhairs, and papillae located in costal amd intercostal zones over long short cells, usually arr ined in t horizontal rows and overarching the stomate "iresiella РИГА 5 (León 4572), adaxial epidermal zonation with Haa spaced costal zones. . Pire siella s strephioides (León 4572), abaxial epidermal zonation, with stomata 6 files per Steps edid E. Piresiella strephioic ides : (León 4572), detail of D showing microhairs, macrohairs, i papillae overarching e stomates. — F. Ekmanoc pus d (Ekman 9870), abaxial epidermal zonation with regularly spaced costal zones. — 6G. Ekmanoc d aristata (Clement & Chrysogone 4563), adaxial epidermal zonation with regularly spaced, narrow costal zones. — У Ekmanoc hloa aristata (Clement & Chry. sogone 563), abaxial epidermis with papillae, microhairs, and stomata. = grouped in 4 = Volume 80, Number 4 1993 Zuloaga et amboo UN of Cuba bundle sheath present, complete, consisting of small cells with uniformly thickened walls; outer bundle sheath of third-order vascular bundles rounded, with abaxial or adaxial interruptions, without ex- tensions, consisting of 2-4 rounded cells; chloro- plasts absent; inner bundle sheath absent or with a few cells abaxially. Sclerenchyma. Minute, with inconspicuous adaxial and abaxial girders associ- ated with all vascular bundles, 1—3 cells deep; fibers lignified; small sclerenchyma cap at the margin. Mesophyll. Chlorenchyma not radiate but adaxi- ally cells tending to a palisade-type arrangement, with conspicuous air spaces; arm cells present but with relatively shallow invaginations; 12-25 chlor- enchyma cells between consecutive vascular bun- dles, which are 90-170 џт apart; absent; colorless cells absent. fusoid cells Adaxial epidermal cells. 2-3 bulliform cells present in restricted groups between all vascular bundles, generally with a large, fan-shaped central cell occupying up to % of the blade thickness; epidermal cells small, regular in size, with a distinct, continuous cuticle; macrohairs not seen; prickles present, usually located Rin the vascular bundles; papillae present, usually per cell, narrower than the epidermal cells, р tively broad but not much more than 12 the width of the epidermal cells. 4baxial epidermal cells. Bulliform cells absent, otherwise cells similar to adaxial epidermis; macrohairs not seen, but hooks present, usually located between the vascular bun- dles; prickles and papillae similar to adaxial epi- ermis. Abaxial epidermis in surface view (Fig. 2A, Zonation. Costal and intercostal zones dis- tinguishable, the costal zone 3-5 cells wide, the intercostal zone 13-18 cells wide. /ntercostal long cells. Elongated, rectangular, more than 3 times as long as wide; anticlinal walls parallel, undulating, unthickened; end walls vertical; long cells in a file separated by a short cell, or infrequently adjoining one another and without short cells. /ntercostal short cells. Solitary, tall and narrow in outline, transversely elongated, the cork cells with crenate or undulating walls. Stomatal complex. 13-17 um long, 10-13 um wide, with per intercostal zone; a single interstomatal long cell 2-4. files of stomata usually present between consecutive stomata; sub- sidiary cells dome-shaped; interstomatal cells rect- angular, with anticlinal walls undulating, the end walls concave. Microhairs. Elongated, fingerlike, 36-46 um long, common in the median intercostal as long as the distal cell, with parallel, thickened walls; distal cells zone; basal cell two or more times short, very thin-walled, deciduous. Macrohairs. Not seen. Hooks and prickles. Small, shortly barbed, prickles with the base as long as the stomata, the located in the median intercostal zone; barb developed, located in the costal zone over the vascular bundles. Papillae. Rounded, simple, small, less than 14 the vertical width of the long cells, located in the costal and intercostal zones over long and short cells, commonly arranged so as to form a ring around each stomate. Silica bodies. Inter- costal silica bodies absent; costal silica bodies dumb- bell-shaped, horizontally elongated. Adaxial epidermal cells similar to the abaxial surface except Adaxial epidermis in surface view. with: stomata infrequent; frequent intercostal paired silico-suberose short cells located near the costal zones; and intercostal silica bodies tall and narrow, elongated vertically, and irregular in outline. PIRESIELLA STREPHIOIDES Leaf blade in transverse section (Fig. 1C, ). Outline. Expanded, flat, with margins slightly recurved; arms of the lamina straight, the two halves symmetrical on either side of the median vascular bundle; thickness at mid-lamina 40-65 um. Ribs and furrows. Conspicuous near the mid- nerve, inconspicuous in the rest of the blade. Me- dian vascular bundle. Midrib with a simple first- order vascular bundle structurally distinguishable from the lateral first-order vascular bundles; ad- axial and abaxial projections slightly convex. Vas- cular bundle arrangement. Six first-order vascular bundles an —40 third-order vascular bundles in the entire blade; (3-)5-7 third-order vascular bundles between consecutive first-order vascular bundles; all vascular bundles centrally located in the blade thickness. Vascular bundle description. First-order vascular bundles circular in outline; phloem tissue adjoining the inner bundle sheath; metaxylem vessel elements slightly angular in out- line, equal or slightly larger than the outer bundle sheath cells as seen in section; third-order vascular bundles angular in outline with xylem and phloem indistinguishable. Vascular bundle sheath. Outer bundle sheath of the first-order vascular bundles consisting of 6-10 inflated cells, rounded, lacking chloroplasts, abaxially or adaxially interrupted by sclerenchyma girders, without extensions; inner bundle sheath present, complete, formed by small cells with uniformly thickened walls; outer bundle sheath of the third-order vascular bundles rounded to elliptical in outline, consisting of 5-6 rounded, elliptical cells, lacking chloroplasts, with abaxial interruptions, adaxial interruptions occasionally present, without extensions; absent or with a few cells near the phloem tissue. inner bundle sheath Sclerenchyma. Small, with inconspicuous adaxial 852 Annals of the Missouri Botanical Garden and abaxial girders associated with all vascular bundles, 2-3 cells chyma caps absent from margins. deep; fibers lignified; scleren- Mesoph yll. Chlorenchyma not radiate, with adaxial chloren- chyma cells in a palisade-like arrangement and conspicuous air spaces; arm cells present, with short invaginations; 12-15 chlorenchyma cells be- tween consecutive vascular bundles, which are spaced 165-235 шт apart; fusoid cells present, translucent, narrowly elongated, present on either side of each vascular bundle, or occasionally ab- sent; fusoid cells separated by numerous chloren- chyma cells; colorless cells absent. Abaxial epi- dermal cells. 3—4 Богт cells present in discrete, fan-shaped groups between all vascular bundles, the central cell large, inflated, and occupying up to % of the blade thickness; epidermal cells small, regular in size, with the cuticle continuous; mac- rohairs present, with a sunken base, not constrict- ed, located at the margins; prickles not seen; hooks present, located between the vascular bundles near the margin; papillae absent. Adaxial epidermal cells. Bulliform cells absent; epidermal cells small, with the cuticle continuous; macrohairs similar to the ones on the abaxial surface; prickles not seen; hooks small, with a short barb, located in the in- tercostal zones near the margin; papillae narrower than the epidermal cells, associated with the sto- mata. Abaxial epidermis in surface view (Fig. 2D, 3). Zonation. Costal and intercostal zones dis- tinguishable; costal zones 1—4 cells wide; intercostal zones 13-18 cells wide. /ntercostal long cells. Long cells in files separated by pairs of short cells, elongated, rectangular, more than 5 times as long as wide, the anticlinal walls parallel and undulating, the end walls vertical. /ntercostal short cells. In silico-suberose pairs, tall, transversely elongated; cork cells with crenate walls 19-26 um long, 11-16 um wide, with 4-6 files of stomata per intercostal zone; a single intersto- matal long cell between consecutive stomata; sub- sidiary cells dome-shaped or parallel-sided; inter- stomatal long cells rectangular, 2-3 times as long as wide, with undulating anticlinal walls and con- cave end walls. Stomatal аш Microhairs. Located in central intercostal zone; 49-75 um long, elongated, fin- gerlike, the basal cell equal to or up to twice as long as the distal cell, with parallel, thickened walls; distal cell with very thin walls, deciduous. Mac- rohairs. Located near the margins. Hooks and prickles. Not seen. Papillae. Present, associated with intercostal long cells, arranged so as to form a ring around each stomate. Silica bodies. Inter- costal silica bodies tall and narrow, transversely elongated, irregular in outline or cross-shaped; cos- tal silica bodies dumbbell-shaped, horizontally elon- gated. Adaxial epidermis in surface view (Fig. 2C). but with paired, isolated intercostal short cells; sto- mata not seen; hooks small, with a short barb, located in the intercostal zones near the margins; Epidermal cells similar to the abaxial surface papillae absent. EKMANOCHLOA ARISTATA AND E. SUBAPHYLLA Leaf blade in transverse section (Fig. 1E- H). Outline. Open, broadly V-shaped; arms of the lamina outwardly bowed, the two halves sym- metrical on either side of the median vascular bun- dle; thickness at mid-lamina 60-130 um (100– 130 um in Ё. aristata and 60-80 um in Е. aphylla). Ribs and furrows. Adaxial ribs and fur- rows slightly developed and abaxial ribs and furrows sub- indistinguishable in E. aristata; both adaxial and abaxial ribs and furrows indistinguishable in Æ. subaphylla. Median vascular bundle. Midrib with simple first-order vascular bundles distinguishable structurally from lateral first-order vascular bun- dles; adaxial projections slightly rounded, the ab- axial projections not developed. Vascular bundle arrangement. Three first-order vascular bundles and 7-8 third-order vascular bundles in section; 5 third-order bundles; all vascular bundles centrally located in the blade in E. subaphylla and slightly abaxially located in E. aristata. Vascular bundle description. First-order vascular bundles circular in outline, the phloem adjoining the inner bundle sheath; metaxylem vessels narrower than the outer bundle sheath cells as seen in section; third-order vascular bundles angular or circular in outline, the xylem and phloem indistinguishable. Vascular bun- dle sheath. Outer bundle sheath of the first-order vascular bundles rounded, with 6-9 inflated, rounded or slightly elliptical cells, without exten- sions, with abaxial, sometimes adaxial, sclerenchy- ma girders; without extensions; inner bundle sheath present, complete. Outer bundle sheath of the third- order vascular bundles rounded, with abaxial, sometimes adaxial, interruptions, without sions, consisting of 5— exten- rounded or elliptical cells; inner bundle sheath absent or present, when pres- ent with a few cells abaxially. Sclerenchyma. Small adaxial and abaxial girders associated with all vas- cular bundles, 1-3 cells deep adaxially in Æ. sub- aphylla and 5-10 cells deep in E. aristata; fibers lignified; small sclerenchyma cap at the margins. Mesophyll. Chlorenchyma not radiate, with ad- Volume 80, Number 4 1993 Zuloaga et al. 853 Bamboo Genera of Cuba axial chlorenchyma in palisadelike arrangement (£. aristata) ог not (Е. subaphylla), with conspicuous air spaces; arm cells present but with invaginations relatively shallow, especially in E. subaphylla; 9- 14 chlorenchyma cells present between consecu- tive vascular bundles, which are spaced 100-210 um apart; fusoid cells and colorless cells absent. Adaxial epidermal cells. A fan-shaped cluster of 3-6 well-developed bulliform cells present between each pair of vascular bundles, the central cell very large and inflated, occupying up to / of the blade thickness; epidermal cells small and regular in size in E. subaphylla, large and inflated in E. aristata; macrohairs absent in E. aristata, absent or present in E. subaphylla, when present located between the vascular bundles or above them and with a bulbous, nonconstricted base; prickles present, op- posite to the vascular bundles; hooks present be- tween vascular bundles; papillae absent. Abaxial epidermal cells. Bulliform cells absent; epidermal cells very large, very regular in size and shape, with a continuous cuticle; macrohairs similar to the ones of the adaxial epidermis; prickles and hooks varying considerably in their distribution; papillae one per cell, rather wide, with а bi- or multi-fur- cated apex. Abaxial epidermis in surface view (Fig. 2F, H). Zonation. Costal and intercostal zones dis- tinguishable; costal zones 1—4 cells wide, the in- tercostal zones 13-22 cells wide. /ntercostal long cells. In files, usually separated by paired short cells; elongated, rectangular, more than 5 times as long as wide; side walls parallel; end walls vertical; anticlinal walls undulating. /ntercostal short cells. In silica-suberose pairs; tall, elongated, vertical; cork cells with crenate walls. Stomatal complex. 19-26 um long, 19-22 um wide, with 4–6 files of stomata per intercostal zone; a single intersto- matal cell between consecutive stomates; subsidiary cells dome-shaped, triangular or parallel-sided; in- terstomatal long cells rectangular, with undulating anticlinal walls and concave end walls. Microhairs. ocated in central intercostal zones; 49-59 um long, fingerlike in E. subaphylla and slightly cla- vate in A. aristata; basal cell 2-3 times longer than distal cell, with thickened and parallel walls in E. subaphylla but with thin walls and inflated in E. aristata; distal cells short, with very thin walls, deciduous. Macrohairs. Absent in Ё. aris- tata; absent or present in E. subaphylla, when present unicellular, hard, and with a sunken base embedded in the epidermal cells. Hooks and prick- les. Irregularly distributed and few in number. Pa- pillae. Located in both the costal and intercostal zones over both long and short cells, arranged in a single horizontal row and overarching the sto- mates; large, more than У; the vertical width of the long cells, with bifurcating apices. Silica bod- ies. Intercostal silica bodies tall, elongated verti- cally, irregular in outline, occasionally cross- or dumbbell-shaped; costal silica bodies dumbbell- shaped, horizontally elongated in E. subaphylla and slightly cross-shaped in E. aristata. Adaxial epidermis in surface view (Fig. 2G). of the abaxial surface but without papillae and Adaxial epidermal cells similar to the ones isolated stomata; costal silica bodies dumbbell- shaped in Е. aristata. DISCUSSION Mniochloa and Piresiella are C, plants with nonradiate mesophyll, parenchyma cells without specialized chloroplasts, and adjacent vascular bun- dles separated by more than three chlorenchyma cells. These anatomical characters are in agree- ment with data previously presented by Brown 1977) concerning the "С/ "С ratio for Piresiella (as Mniochloa strephioides). Anatomically, bam- ~ boos have been characterized by their possession of arm cells and fusoid cells in the mesophyll (Met- calfe, 1960; Renvoize, 1985; Soderstrom & Ellis, 1987). Both Mniochloa and Piresiella have arm cells whose walls have shallow invaginations (Fig. 1 B, D). Piresiella has long, narrow fusoid cells on both sides of the vascular bundles, in agreement with data presented by Renvoize (1985) and Wat- son & Dallwitz (1988, 1992), whereas M. pulchella lacks fusoids. In olyroid bamboo genera the absence of fusoid cells has been previously noted in Ek- manochloa and Raddiella Swallen (Calderón $ Soderstrom, 1967; Renvoize, 1985; Watson & Dallwitz, 1988, 1992); and fusoids also appear to be absent from the leaf blades of Parodiolyra ramossisima (Trin.) Soderstrom & Zuloaga. Sod- erstrom & Ellis (1987) characterized the Olyreae as having a midrib with a first-order vascular bundle associated with an adaxial projection: in Mniochloa and Piresiella the midnerve is represented by a midrib with a solitary first-order vascular bundle and the adaxial projection is absent in Mniochloa and only slightly pronounced in Piresiella. The epidermides of Mniochloa and Piresiella are typical of the Olyreae, having rectangular long cells with undulating walls and silica bodies of the typical olyroid type. Mniochloa has solitary short cells that are suberose he abaxial epidermis of and transversely elongated with crenate walls. Pi- resiella has an abaxial epidermis with pairs of silica- 854 Annals of the Missouri Botanical Garden suberose short cells having silica bodies of the typ- ical olyroid type. The costal silica bodies in both genera are of the short-saddle-shaped type common in the tribe. Epidermal papillae are frequent in the Olyreae (Calderón & Soderstrom, 1967; Renvoize, 1985; Soderstrom & Ellis, 1987). Mniochloa has small, simple papillae distributed in longitudinal files associated with long and short cells on both epi- dermides and also bordering the stomatal appa- ratuses. Watson & Dallwitz (1988), in their de- scription of Mniochloa (without citing a voucher or the species studied), noted the absence of fusoid cells; this is an obvious reference to Mniochloa pulchella, whereas their observation of the pres- ence of papillae only in the vicinity of the stomates must pertain to Piresiella strephioides. Piresiella possesses small, simple papillae only on the abaxial surface concentrated in the vicinity of, and slightly overarching, the stomates. The presence of papillae restricted to the stomatal area has also been noted in the Olyreae in Reitzia Swallen (Calderón & Soderstrom, 1967) Ekmanochloa includes C, species with nonra- diate mesophyll, with cells of the outer bundle sheath lacking specialized chloroplasts, and with more than three cells between adjacent vascular bundles. In transverse section the mesophyll has chlorenchyma cells with invaginated walls, as is characteristic of the bamboos. However, fusoid cells are absent from the mesophyll, in agreement with the reports of Calderon & Soderstrom (1967), Ren- voize (1985), and Watson & Dallwitz (1988, 1992). The epidermis of Ekmanochloa is typical of her- baceous bamboos, having rectangular long cells with undulating walls, short cells in silico-suberose pairs, and the cork cells vertically elongated with crenate walls. The silica bodies are saddle-shaped, although in £. aristata the costal silica bodies are slightly cruciform. The presence of cross-shaped costal silica bodies has been previously noted in and in Piresia and Rehia Fijten (Watson i 1988, 1992). Calderón & Soderstrom (1967) not- ed in Ekmanochloa the presence of costal silica bodies with transitional forms to silica bodies more typical of those of the Oryzeae; in the present study these transitional forms were not observed. The presence of large, compound papillae on the abaxial epidermis, distributed in a file among the long and short cells and overarching the stomatal apparatus, is characteristic of Ekmanochloa. The presence of compound papillae in the Olyreae has otherwise been noted only in Piresia (Calderón & Soder- strom, 1967) and, perhaps more weakly so, in Maclurolyra Calderón & Soderstrom (Calderón & 1973; Renvoize, 1985) Soderstrom, The two species of Ekmanochloa have bicellular microhairs with quite different morphologies. In Е. aristata the microhairs (barely visible in Fig. 2H) with the walls of the basal cell not parallel, and the distal cell very short in are claviform or obconic, relation to the basal cell. In E. subaphylla (Fig. 2F) the microhairs are narrow, fingerlike, and have the distal cell only slightly shorter than the basal cell, as in most olyroids. In comparison to E. sub- aphylla, in transverse section the blades of F. aristata are narrower, and the vascular bundles are slightly displaced toward the abaxial epidermis and associated with deep columns of sclerenchyma cells. ‘TAXONOMIC TREATMENT Mniochloa Chase, Proc. Biol. Soc. Wash. 21: 185. 1908. TYPE: Mniochloa pulchella (Gri- seb.) Chase (designated by Chase, 1908). Delicate, tufted, monoecious perennial from small corms. Culms weak, straggling, dimorphic (vege- tative and leafless flowering). Vegetative culms with several fully developed leaves per complement; lig- ules inconspicuous; pseudopetioles short; blades el- liptical, symmetrical to slightly asymmetrical. Flowering culms erect, usually much exceeding the leaves, with only bladeless sheaths. Racemes 2, conjugate, erect, appressed, subequal, unisexual. Female racemes slightly бек than the male ra- cemes, with several spikelets on short, slightly cla- vate pedicels. Female spikelets 1-flowered, narrow- ly ellipsoidal, falling entire; glumes delicately membranous, distinctly shorter than the floret, ovate-elliptical, 3-nerved, glabrous, obtuse to acute, the margins scabrous, the lower glume slightly shorter than the upper; internode between upper lume and floret inconspicuous; floret fusoid to narrowly ellipsoid, acute to obtuse, membranous, glabrous, pale; lemma 3-nerved; palea about as long as the lemma; lodicules 3, small; gynoecium Male racemes with several spikelets on slightly clavate pedicels. Male spikelets f, ] ikalat not seen. short, l -flowered, much shorter than the г lacking glumes, ellipsoid, acute; lemma 1-nerved; palea 2-nerved, with apex obtuse; stamens 3. Cary- opsis fusoid; hilum linear, extending nearly the full length of the caryopsis, from slightly above the base to nearly the apex; embryo small, basal. Chro- mosome number unknown. Distribution. A monotypic genus of limestone cliffs in lowland Oriente, Cuba Mniochloa расвева (Griseb.) d Proc. Biol. 86, pl. е Cat. . 1866. 50c. ash. pulchella "Ww PI. t Volume 80, Number 4 Zuloaga et al. 855 1993 Bamboo Genera of Cuba A-J, Piresiella ПРАГ с ) Јожа Zuloaga, & Morrone (A, I-J from León 4572 . B-D, Lem H from Wright 3435). —А. Ha e spi elet: — B. Lemma view. — C. Pad view. — D. Palea i two stamens. E-H, female spikelet: DE. Upper ene view. Е, Lower glume Bine — С. Floret, lemma view. — Н. Floret, palea view.— I, J. | Vs and hilum v . K- R, Mniochloa je hella ma ) Chase (K fadi Ekman 3916; L-R from Wright 3448). —K. Habit. LN. m ale spikelet: —L. Lemma view.—M. Palea view.—N. Palea and three stamens. O-R, ae spikelet: —O. Upper glume view. — P. Lower p view. — Q. Floret, lemma view. — R. Floret, palea view. 856 Annals of the Missouri Botanical Garden Strephium pulc hellum (Griseb.) C. Wright, Anales Aca Med. Habana 8: 202. 1871. TYPE: Cuba. Oriente: une de Baracoa, on vertical cliffs, 8 June 1856, C. Wright 3446* (holotype, GOET not seen; isotypes, GH, MO). Figure 3K-R Corms 1.5-2.5 mm diam. Vegetative culms 3- 12 cm tall, 3-9 leaves per complement; sheaths 2.5-7 mm long, с. shorter than the inter- nodes, slightly inflated, 5-7-nerved, glabrous to pubescent, the truncate summit with a few delicate marginal cilia; ligules a hyaline, lacerate, ciliate membrane 0.1 mm long; pseudopetioles 0.1-0.4 mm long, glabrous, yellowish to brownish; blades 7-15 mm long, 2-4 mm wide, obtuse and weakly apiculate at the apex, both surfaces pubescent throughout with short, appressed hairs, less com- monly nearly glabrous, or the adaxial surface often with. са. 1-mm-long, straggling hairs. Flowering culms 7-20 cm long, with 1—2 bladeless sheaths; peduncle filiform, glabrous. Racemes with the ra- chises slender, slightly flexuous, glabrous, scaber- ulous. Female inflorescences 0.8-3 cm long, with (4-)8-13 spikelets on pedicels 0.2-0.5 mm long. Female spikelets 2.2-2.8 mm long; glumes green- ish, ovate-elliptical, the lower 1.7-1.9 mm long, the upper 2-2.2 mm long; floret as long as spikelet; lodicules 0.2-0.3 mm long, flabellate, hyaline. Male inflorescences 0.5-2.3 cm long, with (3-)7-12 spikelets; pedicels ca. 0.2 mm long, dark-colored. Male spikelets 1.3-1.7 mm long; anthers 0.8-1 mm long, fusoid, orangish yellow when dry. Car- yopsis 1.5-2 mm long, 0.5-0.6 mm wide; embryo ca. 0.4 mm long. Phenology. | Apparently flowering in March, June, and December. Distribution. Known only from limestone cliffs in lowland Oriente, Cuba; according to L. Catasus Guerra (17 May 1982 letter to T. R. Soderstrom), the most recent collection in the HAC herbarium was made in 1960 UBA. Oriente: h mestone rocks, 914, Ekman 3910* (05) Sierra EU. in crevices of rocks, 30 Mar. 1942, León 20898 (US). Additional material Rn Piresiella Judziewicz, Zuloaga, & Morrone, gen. nov. TYPE: Piresiella strephioides (Griseb.) Judziewicz, Zuloaga & Morrone (= Olyra stre- phioides Griseb.). Gramina perennia humilia c aespitosa. Culmi dimorphi, floriferi laxi, decumbentes, circa dimidis culmorum vegetativorum statura. Racemi bini, ascendentes, adpressi, inaequales, unisexuales. Spiculae a 'mma apiculatu cullatum. Spic ulae aalas minores foemineis glu- mis expertes Delicate, caespitose, monoecious perennials from small corms, with delicate, elongate stolons. Culms dimorphic, either vegetative or (leafless) flowering. Vegetative culms with several fully developed leaves per complement, these all near the apex; ligules minute, membranous; pseudopetioles short; blades ovate-triangular, truncate and strongly asymmet- rical at the base, acute at the apex. Flowering culms lax, decumbent, about one-half as tall as the veg- etative culms. Inflorescences terminal, exerted, consisting of 2 conjugate, ascending, appressed, subequal unisexual racemes. Female racemes with narrow rachises and few spikelets. Female spikelets l -flowered, ovoid, falling entire or the glumes tar- dily deciduous; glumes as long as spikelet, subequal, narrowly ovate, acute to acuminate, chartaceous, 3(-5)-nerved, sulcate, glabrous; internode between upper glume and floret prominent; floret membra- nous, shorter than the glumes, whitish, covered with delicate, appressed, white, silky hairs; lemma and palea equal, the lemma apex apiculate or cu- cullate, not evidently nerved, the margins enfolding the palea; palea 2-keeled; lodicules 3, small; gy- noecium with 3 stigmas. Male racemes slightly shorter than the female racemes, with slender ra- chises and few spikelets on short pedicels. Male spikelets 1-flowered, shorter than female spikelets, ellipsoid, glabrous, persistent, lacking glumes; lem- ma as long as spikelet, lanceolate-ovate; palea near- ly as long as lemma; stamens 2. Caryopsis broadly ellipsoid; hilum linear, extending the full length of the caryopsis; embryo small, basal. Chromosome number unknown. Distribution. A monotypic genus of ravines, palm savannas, and streambanks in lowland Ha- bana and Pinar del Rio, Cuba. Discussion. This new genus shares with Mniochloa, Ekmanochloa, Piresia, and some spe- cies of Cryptochloa, Olyra, and Pariana Fusée- Aublet inflorescences on separate culms that lack fully developed leaves. It differs from all of these genera except Mniochloa and one species of Ek- manochloa in its conjugate, unisexual racemes and spikelets with two rather than three or more sta- mens. Volume 80, Number 4 1993 Zuloaga et al. 857 Bamboo Genera of Cuba Piresiella differs from Mniochloa by eight mor- phological and three anatomical characters given in Table 1 The stipe of the female floret in Piresiella re- sembles elaiosomes; these have been confirmed in Olyra obliquifolia Steudel, O. fasciculata Trin., and all species of Cryptochloa (Davidse, 1987). This elaiosomelike resemblance is suspected in oth- er genera such as Froesiochloa G. A. Black and the non-olyroid herbaceous bamboos Anomochloa marantoidea Brongn. and Puelia Franchet The generic name is a diminutive of Piresia and was suggested by the superficial resemblance (Ta- le 2) to a small species of that genus. Piresiella rivals Raddiella minima Judziewicz & Zuloaga as the smallest of all the bambusoid grasses; some individuals with mature inflorescences have vege- tative culms only 5 cm tall, with flowering culms 3 cm long. Piresiella strephioides (Griseb.) Judziewicz, Zu- loaga & Morrone, comb. nov. Basionym: Olyra strephioides Griseb., Cat. Pl. Cub, 229. 1866. Mniochloa strephioides (Griseb.) Chase, Proc. Biol. Soc. Wash. 21: 186. 1908. TYPE: Cuba. Pinar del Rio: Rio Santa burk [7 spelling of last word uncertain], among the adventitious roots of palms, river margin, 27 Aug.-5 Sep. 1865, C. Wright 3435 (holotype, GOET not seen, fragment, US; isotypes, GH, HAC not seen, MO, NY, US). Figure 3A-J. Corms 1-2 mm diam.; stolons up to 20 cm. Vegetative shoots (7-)10-18 cm tall, with 3-7 leaves per complement; internodes pubescent in a line, otherwise glabrous; nodes often swollen, ge- niculate, slightly retrorsely bearded, shrinking when dry to produce a double flange; sheaths glabrous except pubescent on one margin and with a fringe of fine, excelsiorlike cilia at the summit; ligules 0.1 mm long, membranous, ciliolate; pseudopetioles 0.2-0.6 mm long, tan, glabrous; blades 14-20 mm long, 6-10 mm wide, glabrous to puberulent on both surfaces. Flowering culms each with a single bladeless sheath, this glabrous except for one pubescent margin. Racemes with the rachises slen- der, angled, scabrous, slightly flexuous, glabrous to pubescent. Female inflorescenes 1.5-3 cm long, with 6–9 spikelets on pedicels 0.2-0.6 mm long; rachises 0.2-0.3 mm wide. Female spikelets 4— 4.8 mm long; glumes greenish or whitish (some- times with purplish apices), with a few faint cross- nerves, spreading up to 180? at maturity; internode between upper glume and floret 0.3-0.5 mm long; floret 3-3.5 mm long, whitish; lodicules 0.7-0.8 mm long, truncate. Male racemes with the rachis ca. 0.1 mm wide, bearing 6-8 spikelets on pedicels 0.3-1.5 mm long (longest at the base of the ra- ceme). Male spikelets 1.3-1.8 mm long; anthers 0.5-0.7 mm long, yellowish orange when dry. Caryopsis 2 mm long, 0.8 mm wide; embryo 0.2- 0.3 mm long. Phenology. Apparently flowering from July through November Distribution. Palm savannas, ravines, and streambanks in lowland Habana and Pinar del Rio, Cuba; reported by L. Catasus Guerra as having been collected by him as recently as 1974 (17 May 1982 letter to T. R. Soderstrom). Additional material examined. CUBA. Habana: Loma de Coca, locis perumbrosis, 2 June 1914, Ekman 1238 (US); banks of Rio Guanabo, near Campo Florida, 10 Oct. 1943, León 4140 (US). Pinar del Rio: hills W of Cayajabos, in forest, 25 Tate 1921, Ekman 12945 (05); Cabarios, Guabal de Lechuza, 2 July 1921, Ekman s.n., Amer. Gr. Nat. Herb. 1029* (US); first range of hills 13 km NW of Pinar del Río, 26 Nov. 1926 (sterile), Hitchcock 23310 (US); N of San Diego de los Banos, banks of an arroyo, 18 Aug. 1914, León & Hiram 4391 26 Aug. from Ban “Diego d e los Banos, banks of a rivulet, at the base of a palm, 21 Aug. 1914, León 4593 (US). Ekmanochloa Hitchc., Man. Grasses W. Ind. 374. 1936. туРЕ: Ekmanochloa subaphylla Hitchc. (designated by Hitchcock, 1936) Caespitose, short rhizomatous, monoecious pe- rennials from small corms. Culms erect, simple, dimorphic. Vegetative culms with 3-6 fully de- veloped leaves per complement. Sheaths striate, glabrous or pilose in the distal portion; ligules mem- branous, small; pseudopetioles short; blades linear- lanceolate, flat, glabrous or short-pilose, deciduous. Flowering culms with 2-3 bladeless leaves. Inflo- rescence terminal, exerted, consisting of 2 conju- gate, erect, appressed, subequal unisexual racemes with filiform, slightly flexuous, glabrous, smooth rachises. Female racemes slightly longer than the male racemes, of 3-5 spikelets on clavate pedicels; male inflorescences of 3-12 spikelets on slightly clavate pedicels. Female spikelets 1-flowered, nar- rowly ellipsoid; glumes shorter than to as long as the floret, glabrous to sparsely pilose, the lower 5-nerved, the upper 3-nerved; internode between upper glume and floret short or up to 0.5 mm, pilose; floret fusiform, membranous, obliquely in- serted above the glumes, early deciduous; lemma aristate, 3—5-nerved; palea 2-nerved, the apex not enfolded by the lemma; lodicules 3, flabellate, hy- 858 Annals of the Missouri Botanical Garden aline; gynoecium oblong, with 2 free styles. Male racemes of 3-12 spikelets on slightly clavate ped- icels. Male spikelets 1-flowered, ellipsoid, hyaline, lacking glumes; lemma 3-nerved; palea 2-nerved; lodicules 3, hyaline; stamens 2-3. Caryopsis fu- siform; hilum extending the full length of the ven- tral side; embryo small. Chromosome number un- known Distribution. to ultramafic (serpentine) areas in lowland Oriente, uba А genus of two species endemic Discussion. Ekmanochloa differs from most of the Olyreae in the following: (1) absence of fusoid cells (but these are also lacking in some species of Raddiella and Mniochloa); (2) presence of strong- ly compound papillae on the abaxial epidermis (these also present in Piresia and Rehia, but not so strongly developed); (3) deciduous leaf blades; and sia Zuloaga & Judziewicz (1993) and Piresiella, where they are very short and may more properly be termed apiculi). Ekmanochloa also shares some characters with the doubtfully olyroid, monotypic Buergersiochloa Pilger of New Guinea: both long-awned female florets, compound epidermal pa- of these taxa have pillae, and dimorphic culms. However, both have many significant morphological and anatomical dif- ferences, and it does not seem likely that they are closely related. When both species of Ekmanochloa are com- pared with other members of the Olyreae, it ap- pears that E. subaphylla is more specialized than E. aristata. Besides its specialized limestone cliff habitat, E. subaphylla differs from E. aristata in its two rather than three stamens; very short, emar- ginate rather than long and entire female glumes; eaf blades with a less well-developed adaxial pal- isadelike chlorenchyma layer; and slightly less well- developed arm cells. Furthermore, Clayton & Ren- voize (1986) noted that the leaf blades of E. sub- aphylla are more likely to be more completely suppressed than those of E. aristata. Another spe- cialized feature present in F. aristata is microhairs with a distinctive squat, obconic basal cell. Æk- manochloa subaphylla has fingerlike, typically olyroid bicellular microhairs. KEY TO THE SPECIES OF EKMANOCHLOA la. Female spikelets with S about as long as die body of the floret, sparsely pilose; lemma with an awn 16-18 mm lina male inflorescence with 3-4 short-pil pikelets 2.7-3.3 e long stamens 3 Е. Pisas lb. Female spikelets with glumes much eda than the body of the floret, glabrous; lemma with an awn 5-10 mm long; male inflorescence with 6— 12 glabrous spikelets 1.7-2.3 mm long; sta- ens 2 2. E. subaphylla Ekmanochloa aristata Hitchc., Man. Grass- es W. Ind. 377. 1936. TYPE: Cuba. Oriente: between Taco and Nibujón, in carrascales- pinales, very rare, 4 Dec. 1914, Е. L. Ekman 3729* (holotype, US; isotype, S not seen). Figure 4A- Short-rhizomatous perennial. Vegetative culms 30-50 cm tall, developed leaves per complement. sheaths 1-4 cm long, shorter than the internodes, erect, unbranched, with 3-5 fully Leaves with the margins overlapping, membranous, ciliate to- ward the apex, otherwise glabrous; ligules 0.1-0.2 mm long, laciniate at the apex; pseudopetioles 0.3- 0.6 mm long, brownish, glabrous to short-pilose; blades apparently early deciduous, 8-55 mm long, 2-4 mm wide, linear-lanceolate, narrowed and slightly asymmetrical at the base, acute at the apex, glabrous, the margins smooth, glabrous. Flowering culms 20-50 cm tall, erect, usually exceeding the vegetative culms, unbranched, with two bladeless leaves present; peduncle up to 25 cm long, filiform, glabrous. Female inflorescence 15-20 mm long, with 3-5 spikelets at a distance of 4-8 mm from each other; pedicels clavate, 3.3-5 mm long, smooth, glabrous. Female spikelets 6-9 mm long, 0.9 mm wide; glumes as long as the body of the floret, subequal, membranous, sparsely pilose with appressed hairs; internode between upper glume and floret short, pilose; floret 5-7 mm long (ex- cluding awn), 0.6 mm wide, fusiform, acute, sca- berulous, pale; lemma 3-nerved, with an awn 16- 18 mm long; palea shorter than the lemma; lodi- ds 0.5-0.7 mm long, brownish, apically lacin- . Male inflorescence 10-15 mm long, with 3- d dudar pedicels 0.6-2.5 mm long, smooth, glabrous. Male spikelets 2.7-3.3 mm long, 0.5 mm wide, acute, sparsely pilose with short, ap- pressed hairs between the nerves; lodicules 0.5 mm long, brownish; stamens 3, the anthers 0.8 mm long. Caryopsis 4.2 mm long, fusoid, brown, the embryo ca. 0.6 mm long. Apparently flowering in March, June, and December. Phenology. Distribution. Rare in pine savannas in low- land Oriente, Cuba; last collected in 1945 (L. Ca- Zuloaga et al. 859 Volume 80, Number 4 1993 Bamboo Genera of Cuba FIGURE 4. 4563). — A. Habit. B-D, male spikelet: — B. Lemma view. — C. ve spikelet: — E. Upper glume view. — F. Lower glume view. G-L, a subaphylla Hitchc. (from Ekman A-F, Ekmanochloa aristata Hitchc. (A, E, F from Ekman а B, D from Clement & Chrysogone B Palea view. — D. Palea and three filaments. E-H, . G-I, male spikelet: — C. Lemma view. — H. Palea view. — I. Palea and two filaments. J-L, female spikelet: — H jd lateral view. —K. Upper glume view. —L. Lower glume view. = 860 Annals of the Missouri Botanical Garden tasus Guerra, 17 May 1982 letter to T. R. Sod- erstrom). Additional material examined. CUBA. Oriente: neges not far from Moa airport, 26 June 1945, Clem- с Chry "iai 4563* (05); рар, of the Rio Cayo- suán, near the mines, Moa, 30 Mar. 1942, León et al. 20901 (US). 2. Ekmanochloa subaphylla Hitche., Man. Grasses W. Ind 5. 1936. TYPE: Cuba. Oriente: Sierra de Nipe, Loma Picote, on over- hanging limestone rocks, 500 m, 2 Nov. 1922, E. L. Ekman 9870 (= Amer. Gr. Nat. Herb. 1016) (holotype, US; isotype, US). Figure 4G-L Tufted perennial. Vegetative culms 50-100 ст long (the erect portion shorter), erect or geniculate at the base, unbranched, with 4—6 fully developed leaves per complement. Leaves with sheaths 0.7- 2 cm long, shorter than the internodes, the margins overlapping, membranous, glabrous; ligule 0.2 mm long, laciniate at the apex; pseudopetioles 0.6-1 mm long, pale to brownish, short-pilose; blades apparently early deciduous, 10-35 mm long, 1- mm wide, cuneate at the base, obtuse to subacute at the apex, shortly pilose to glabrous, the margins scaberulous toward the base, otherwise smooth and glabrous. Flowering culms 33-60 cm tall, erect, usually exceeding the vegetative culms, un- branched, with 2-3 bladeless leaves; peduncle 20- 30 cm long, filiform, glabrous. Female inflores- cence 30-40 mm long, with 3-5 spikelets at a distance of 4-6 mm from each other; pedicels 7— 12 mm long, smooth, glabrous. Female spikelets 6—6.6 mm long, 0.9 mm wide; glumes much short- er than the body of the floret, unequal, membra- nous, glabrous, with obtuse, truncate, or emargin- ate or denticulate apices; lower glume 1-2.1 mm long, the upper 2.1-4.2 mm long; internode be- tween upper glume and floret ca. 0.5 mm long, pilose; floret as long as the spikelet, 0.6 mm wide, acute, scabrous, pale to brownish; lemma 5-nerved, with an awn 5-10 mm long; palea shorter than the lemma, greenish; lodicules 0.4 mm long. Male inflorescence 17-23 mm long, with 6-12 spikelets; pedicels ca. 0.4 mm long, smooth or scaberulous, glabrous. Male spikelets 1.7-2.3 mm long, 0.5 mm wide, ellipsoid, subacute, glabrous, greenish; lodicules 0.2 mm long, brownish; stamens 2, the anthers 1.2 mm long. Caryopsis not seen. Phenology. Apparently flowering from Oc- tober to November Distribution. Limestone cliffs in the Alto Son- go region of Oriente, Cuba, at elevations up to 500 m; last collected in 1960 (L. Catasus Guerra, 17 May 1982 letter to T. R. Soderstrom). add cad material examined. Oriente Sierra de Nipe, in limestone hills at ш. ш. 13 Oct. 1919, Ekman 9870* Gan number as type collec- tion, but different locality and date) (US). LITERATURE CITED Bews, J. W. 1929. The World’s Grasses. Longmans, Green, & Co., London. BREAKWELL, E. 1914. A study of the leaf anatomy of some native species of the genus Andropogon (Gra- mineae). Proc. Linn. Soc. New South Wales, Ser. 2, BROOKS, R. R. . Serpentine and its Vegetation: A Vani наш Approach. Dioscorides Press, Port- а BROW “1977. The Kranz syndrome and its subtypes in in grass systematics. Mem. Torrey Bot. Club CALDERON, c К. SODERSTROM. 1967. Las gramíneas tropic ales afines a Olyra L. 25 do Sim- pósio sóbre Biota Amazonica, 4(Botánica): 67-76 & 19 Morphological and anatom- ical conside of the grass subfamily Bambuso- ideae based on the new genus Maclurolyra. Smith- sonian Contr. Bot. 11: 1-55. The genera of Bambuso- ideae (Poaceae) of the rere. continent: keys and comments. Smithsonian Contr. Bot. 44: 1-27. CHASE, А. 1908. Notes on genera of the Paniceae. Biol. Soc. Wash. Proc. 21: 185-187. CLAYTON, W. & S. A. RENVOIZE. 1986. Genera Graminum: CAM of the World. Kew Bull., Addit. Ser. 13. Her Majesty's Stationery Office, London. DAVIDSE, С. 1 Evolution and Systematics. Smithsonian БОШОО ess, Washington, D.C. ELLIS, P. R. 1976. A procedure for standardizing com- iode leaf blade anatomy in the Poaceae. 1. The blade as viewed in transverse section. 12: 65-109 Bothalia A procedure for standardizing com- parative leaf blade anatomy in the Poaceae. II. The oue as seen in surface view. Bothalia 12: 641- 67 HrrCHCOCK, А. 5. 1936. Manual of the grasses of the West Indies. U.S. Dept. Agriculture Misc. Publ. 243 439 HUBBARD, C. E. Pp. 199-209 in J. Hutchinson, Families of Flowering Plants: 2(Monocotyledons). Macmillan, London. E. Pouyu Rojas, & L. CATASUS 1989. El gang he la Wagn Po- aceae en Cuba. Acta Bot. Cu е С. К. Anatomy of ds Monocotyle- dons. 1. eos Clarendon Press, ord. Ниор: S.A 85. A survey of leaf- Teen yan in grasses. V. The ba mboo allies. Kew Bull. 40: 509- 535. Кознем115, R. 1937. Grasses: An areata to the Study of Fo od and Cereal Grasses. Moscow. [En- glish translation published by the Indian "National Scientific Documentation Centre, New Delhi, 1980.] Volume 80, Number 4 1993 Zuloaga et al. Bamboo Genera of Cuba SODERSTROM, Т. R. € R. P. ELLIS. 1987. The position of bambo SE and allies in a system of grass classification. Pp. 225-238 in T. R. Soderstrom et al. (editors), Tas Evolution and Systematics. Smith- sonian Institution P C 1988. ad ога patterns in neotropical bamboos. Pp. 121- 157 in Proceedings of the Neotropical Biotic uen bution Pattern Workshop, held in Rio de Janeiro, 12-16 January 1987. Academia Brasileira de Cien- clas. . ZULOAGA. 1989. А revision of the genus Oly ra and the new segregate genus Parodi- olyra (Poaceae: um Olyreae). Smithson- ian Contrib. Bot. 6 Watson, L . 1988. Grass Genera of the Wo jid: Песоа of Characters, Descrip- tions, Interactive Identification, and Information Re- trieval. The Australian National University, Canber- & 9 The Grass Genera of the World. CAB йа аа] University Press, Cam- bridge. ZULOAGA, F. O. & E. J. Jupziewicz. 1993. Agnesia, a new genus of Amazonian herbaceous bamboos (Po- aceae: Bambusoideae: Olyreae). Novon 3: 306-309. FENOLOGIA Y ESTRUCTURA Delhy Albert Puentes, FLORAL DE TRICHILIA ли aat HAVANENSIS JACQ. (MELIACEAE)! RESUMEN Se realizó un estudio de fenologia у estructura floral a la especie Trichilia havanensis. Se comprobó la dioecia de esta especie, a pesar de tener los órganos femeninos y masculinos en la misma flor. Un individuo que se había comportado como masculino durante los tres primeros arios de observaciones fenológicas, fructificó posteriormente. ABSTRACT A Dod ie carried out over 5 years is presented and flower morphology of Trichilia havanensis is given. Dioecy was confirmed in this species, despite the fact that female and male organs occur in the same flowers. А tree that pud as male during the first 3 years of our phenological observations has fruited since 1988. La dioecia es un fenómeno que ha sido estudiado mación de una especie dioica, existiendo cuatro por diferentes autores tales como Darwin (1877), tipos florales en la misma. Las especies de la familia Lewis (1941), Horovitz (1954). Estudios recientes Meliaceae han sido consideradas como hermafro- han demostrado la presencia de especies dioicas ditas por numerosos botánicos, tales como Britton que aparentan ser hermafroditas por tenor los ór- & Wilson (1923), Harms (1940), León & Alain ganos femeninos y masculinos en la misma flor. (1951), y otros, debido a que aparentemente sus Armstrong € Drummond (1986) señalaron que flores poseen los dos sexos. las flores estaminales y pistilares pueden adquirir Sin embargo, Styles (1972), de acuerdo con sus aspecto similar para producir estimulo tanto visual observaciones realizadas en el Suroeste y Centro como olfatorio al polinizador, sugiriendo que las de Africa, senaló que la mayoria de las especies flores pistiladas son probablemente miméticas de de esta familia poseian flores unisexuales pero con las flores estaminadas, mecanismo que ha sido cla- rudimentos bien desarrollados del sexo no funcio- sificado por Little (1983) como decepción alimen- nal. Anos mas tarde, este mismo autor resume el taria por automimetismo. Otras hipótesis tratan de conocimiento adquirido hasta ese momento en explicar la dioecia a partir del hermafroditismo; cuanto a la biologia floral de Meliaceae, lo cual fue por ejemplo las de Arroyo & Raven (19775) quienes utilizado por Pennington & Styles (1975, 1981) expresaron que el paso clave en la evolución dela еп el tratamiento taxonómico de la familia. dioecia es la reducción de los órganos femeninos Nuestro trabajo tiene como objetivo dar a co- en las plantas hermafroditas; у la de Lewis (1941) посег los patrones de floración y fructificación de quien demostró la esporádica distribución de las la especie Trichilia havanensis Jacq., pertene- especies dioicas en comparación con las familias ciente a la familia Meliaceae, en Cuba y ofrecer hermafroditas (que se encuentran en todas partes), datos acerca de su estructura floral. por lo que deduce que las especies dioicas han evolucionado a partir de representantes hermafro- ditas. Luz de Armas (1985) en su estudio de las flores Se realizaron observaciones fenologicas sema- de Chamissoa altissima var. altissima expuso que nales durante cinco anos (1985-1989) a 10 in- ese taxon atravieza una fase evolutiva en la for- — dividuos de Trichilia havanensis localizados en el MATERIALES Y METODOS ' Queremos a call nuestro ee i a Heriberto Rodriguez por la realización de iis dibujos. 2 Instituto de Ecología y Sistemática, Herbario, Academia de Ciencias de Cuba, Carretera de Varo a К, 3%, Capdevile, Boyeros, кин до 8010, Código Ps 10800, La Habana 8, CUBA. ANN. Missouni Вот. Garp. 80: 862-869. 1993. Volume 80, Number 4 1993 Albert Puentes et al. 863 Fenología y Estructura Floral de Trichilia havanensis Parque Metropolitano de La Habana. Este parque se encuentra situado а 23?6'N y a 82°26'О, pre- sentando un suelo de rendzina negra (Obregón, com. pers. )yun jt oque сах шае соп un регі (Vilamajo et al., 1989). T promedios mensuales dé temperatura y las precipitaciones mensuales totales fueron tomadas del Instituto de Meteorologia en la estación de Casa Blanca y con estos datos se confeccionaron los diagramas climáticos de Walter & Lieth (1960) (Fig. 1) Las fenofases observadas fueron las siguientes: Porcentaje de hojas Botones (desde su aparición hasta la antesis) Flores (desde el momento de la apertura hasta su desaparición) Frutos (abarca todos los estadios de su desarrollo hasta la apertura) Frutos abiertos (desde su apertura hasta que pierde las semillas Para la delimitación de las fenofases se empleo la metodologia de Ramia (1981) y los datos fueron adaptados posteriormente a la escala de Fournier (1974) mediante el cual se establecen cinco rangos de aparición del caracter como sigue: 0 = ausencia del fenómeno observado 1 = presencia del fenómeno con una magnitud tre 1–25% 2 — presencia del fenómeno con una magnitud entre 26-50% 3 — presencia del fenómeno con una magnitud entre 51-75% 4 — presencia del fenómeno con una magnitud entre 76–100% Para establecer los patrones de floración y fruc- tificacion de la especie, se compararon nuestros resultados con ejemplares localizados en los alre- dedores del Instituto de Ecologia y Sistemática del Academia de Ciencias de Cuba, con observaciones de campo realizadas por nosotros en diferentes tipos de vegetación en la Provincia de Pinar del Río, y con 82 ejemplares de T. havanensis procedentes de diferentes localidades del pais, ubicados en los herbarios de la Academia de Ciencias de Cuba HAC) y Jardin Botánico Nacional (HAJB). Para el estudio de la estructura floral, se tomaron 10 flores por cada árbol a las cuales se les midieron los siguientes indices: A = longitud del tubo estaminal B = longitud de la antera C = longitud del ovario D = ancho del ovario TABLA 1. Relación de medias calculadas para las flo- res de Trichilia havanensis. x fl 5 masculinas femeninas X general Carácter (mm mm) А 2.47 3.46 2.24 B 0.84 1.21 0.60 C 1.25 1.50 1.00 D 1.50 1.42 1.61 E 0.71 1.00 0.75 F 0.60 0.70 0.50 G 4.25 5.00 3.50 H 2.40 2.60 2:20 1 0.45 0.50 0.40 = longitud des estilo longitud del estigma = longitud del pétalo = ancho del pétalo — longitud del disco nectarifero | = 2 0 = 1 | Con los datos obtenidos se confeccionaron los gráficos de Jentys-Szaferowa (1959). En este те- todo se calculan previamente las medias aritméticas generales para cada carácter (de las plantas fe- meninas y masculinas tomadas como un todo) las cuales constituyen la linea unidad; posteriormente se grafican las diferencias de las medias de cada carácter con la media general (Tabla 1, Fig. 5). Como caracteres cualitativos se observaron los tipos de apertura floral (completa o parcial) y la coloración de las anteras. RESULTADOS Y DISCUSION Al analizar el porcentaje de hojas durante los cinco anos observados, podemos ver que esta es- pecie se comporta como siemprever ue pérdida en pin no sobrepasa el 20% del total del follaje (Fig. 2 a Голове, "botones" se ubica entre los meses de enero y marzo y “flores” a partir del mes de febrero, con un máximo en la segunda quincena de marzo, extendiendose hasta la primera semana de mayo (Fig. 2), de forma tal que puede catalo- garse como una especie de floración tardia de acuerdo con la clasificación de Sarmiento & Mo- nasterio (1983) y de floración corta según Ramirez & Brito (1987). En los individuos más expuestos al sol se observó en todos los anos mayor porcentaje de flores que en los que se encuentran en lugares mas sombrea- dos. En la Figura 3 están representados los promedios 864 Annals of the Missouri Botanical Garden % 1985 "s * 1986 hare % 1987 gm 200 - 200 100 100 100 90 90 80 80 80 TO TO TO 60 60 60 30 50 30 - 50 30 50 40 Р 40 40 25 30 25 30 25 30 20 20 20 20 10 20 10 20 10 EFMAMJJASOND EF MAMJJASOND ЕРМАМ у у АЗ Он О 1988 mm 1989 mm 300 300 200 200 100 100 90 90 80 80 70 70 60 60 30 50 30 50 40 40 25 30 25 30 20 20 20 10 20 10 EFMAMJJASOND EF MAMJJASOND FIGURA 1. Diagramas climáticos del Parque Metropolitano de La Habana durante los anos 1985-1989. 4 1 : — Hojas d Mem e 21 2. 1! 1 MA 1985 0 0 44 44 3! — Botones 34 pao А 2: 2. d E n 1. p ac i 0 0 / 44 4 > — Flores a 14 1. 1987 0 0 41 — Frutos 44 34 31 2- 2 | 14 14 0 0 41 — Frutos abiertos 4- 34 34 2} 24 14 1 + OTE FMAMJJA SOND ¡AA OND EFMAMJJASON OD FIGURA 2. Promedios de cinco anos de observaciones de las dif: fenof: diante el método de Fournier FicuRA 3. Promedios de producción anual de flores (1974). y frutos. Volume 80, Number 4 3 Albert Puentes et al. Fenología y Estructura Floral de Trichilia havanensis 865 O Botones O Flores LI] L] Frutos abiertos | [| E F M A M J J A 5 [о] N | D FiGURA 4. HAC y HAJB de producción anual de flores. Como podemos apre- ciar en los anos 1985 y 1988 esta producción alcanza el rango 4 en la escala de Fournier lo que puede ser debido a un ligero aumento en las pre- cipitaciones durante los anos en cuestión ya que de acuerdo con Opler et al. (1976) y Mori & Prance (1987) la estacionalidad en los trópicos está fundamentalmente dada por este factor climático. Los ejemplares florecidos procedentes de dife- rentes localidades del pais ubicados en los herbarios anteriormente referidos (Fig. 4), los individuos si- tuados en los alrededores del Instituto de Ecologia y Sistemática, y las observaciones de campo rea- lizadas en la provincia de Pinar del Rio corroboran nuestros resultados, encontrandose los picos (та- ximos) de botones y flores en el mismo periodo que el encontrado como resultado de 5 anos de obser- vaciones fenológicas en el Parque Metropolitano de la Habana, lo que demuestra un gran sincro- nismo no solamente entre los árboles de una lo- calidad, sino entre los individuos procedentes de las diferentes localidades del pais. Auspurger (1983) definió el sincronismo como un escape de las plantas a los depredadores de las flores y luego de las semillas, por lo que resulta una estrategia adaptativa para una exitosa polini- zación y dispersión de la semilla (Steven et al., 1987) E F M a J y A 5 C N D Histograma de frecuencias de los ejemplares de Trichilia havanensis depositados en los herbarios Con relación a los frutos, obtuvimos que durante los 5 años de observaciones, las plantas 3, 5 y 6 fructificaban y las restantes no, lo que concuerda con lo planteado con Pennington « Styles (1981) de que la especie Trichilia havanensis es dioica. El periodo de fructificación de las plantas fe- meninas comienza generalmente desde mediados del mes de marzo hasta noviembre y abren sus cápsulas trimeras desde septiembre hasta finales de diciembre, etapa en la que si permanece algün fruto en la planta, este ya ha perdido las semillas, por lo que no se toman en consideración. De esta forma puede clasificarse como una especie de fruc- tificacion durante una temporada amplia segün Castillo & Carabias (1982) Como consecuencia de la abundante floración, la fructificacion aparece con los mayores valores durante los anos 1985 y 1988 (Fig. 3). Los histogramas de frecuencias de "flores" y "frutos" confeccionados con los ejemplares de- HAJB se en- cuentran ligeramente extendidos lo cual constituye una limitación de la fenologia de herbario (Croat, positados en los herbarios HAC y 1969), pues los individuos pueden florecer en un ano fuera de época, produciendo un alargamiento del periodo de la fenofase observada Las diferencias morfológicas entre las flores fe- meninas y masculinas se ponen de manifiesto en 866 Annals of the Missouri Botanical Garden 07 08 09 | 11 12 13 14 т о т тш о ouw > а FiGURA 5. Gráficos de Jentys- pedi dis medias calculadas en mm. — a. Eje on los caracteres y los nümeros ба иги. a las diferencias entre las medias femeninos. Las letras corresponden c particulares y la media general (ver ше y métodos los gráficos de Jentys-Szaferowa (1959; ver Fig. 5, Tabla 1 En las flores masculinas (Figs. 5a, 6A), todos los caracteres excepto el ancho del ovario (indice D) presentan mayores valores que en las femeninas (Figs. 5b, 6B). Existen otras diferencias cualitativas entre las flores de ambos sexos como son las siguientes; la apertura floral en flores masculinas es completa, dejando descubierto totalmente al tubo estaminal y en las femeninas es parcial; y la coloración de las anteras en flores masculinas es amarilla mien- tras que en las femeninas es carmelita. En periodo de floración encontramos una intensa actividad de la especie Apis mellifera L. White en el prólogo de la monografía genérica de Meliaceae (Pennington & Styles, 1975), apuntó que un himenóptero insertaba su proboscis por una estrecha ranura formada entre las anteras y an- teroides en T. havanensis, transfiriendo el polen de flores masculinas hacia la superficie estigmática de las flores femeninas. Anteriormente Priestly y Scott, segün de la Luz (1985), habian considerad amarillo, blanco, y azul, con el color verde como ue las flores de colores follaje, eran visitadas con preferencia por los in- sectos, y atraen en especial a las abejas, las cuales a su vez prefieren las inflorescencias del tipo co- > E от ов 09 10 |I! 12 < б —b. Ejemplares emplares masculinos. rimbo, capitulo y glomérulos, cuyas flores son reu- nidas y compactas, comportándose como superfi- cies planas. Este es el caso de las inflorescencias fasciculadas de T. havanensis, por lo que pensamos que la especie Apis mellifera L., abeja de miel, es pro- bablemente la encargada ds la polinización. e acuerdo con lo senalado anteriormente, y si tomamos en cuenta que Apis mellifera es intro- ducida en nuestro pais, podemos corroborar lo plan- teado por Gentry (1986) de que la micrantia con flores agrupadas es poco especifica para la poli- nización, motivo por el cual, muchas especies han d di осп сар! 1 uba sin necesidad del agen- te 1 Las hormigas son otros himenópteros que visitan las flores де 7. havanensis pero estas no son con- sideradas como polinizadores, sino que se aprove- chan y toman el néctar sin aportar beneficio a las plantas (Guerrant, 1981). El que se hayan encontrado solamente tres plan- tas femeninas de las 10 estudiadas, está en co- rrespondencia con lo que ocurre en la naturaleza de forma espontánea dentro de la familia Melia- ceae, ya que el nümero de plantas masculinas es mucho mayor que el de femeninas (Styles, 1972). Debemos senalar que la planta # 1, aunque se comportó durante tres anos como masculina, en Volume 80, Number 4 Albert Puentes et al. Fenología y Estructura Floral de Trichilia havanensis FiGURA 6. Flores de Trichilia havanensis— A. Flor masculina. B.— Flor femenina. 868 Annals of the Missouri Botanical Garden 1988 comenzó a fructificar, desasrrollando frutos normales (hasta el rango 2 de la escala de Four- nier). A pesar de que fenómenos como el anterior- mente expuesto han sido poco estudiados, existen algunos articulos referentes a la especie Carica papaya L. en los cuales se expone que aunque el tipo sexual básico está determinado genotipica- mente, ciertos árboles machos y hermafroditas su- fren reversiones sexuales de varios grados, atri- buyéndole diferentes causas al mismo Por ejemplo, Munoz Morales (1983), consideró que estas reversiones tienen dos origenes: ambien- tal y hereditario, y que los ültimos producen cam- bios totales o parciales en las flores y órganos esenciales. Este autor opina que las plantas femeninas no muestran variaciones por lo que son consideradas como el tipo floral de mayor estabilidad. Ghosh & Sen (1975) senalaron la influencia del contenido de nitrógeno a favor de las plantas fe- meninas, asi como el efecto de los reguladores del crecimiento, los cuales pueden alterar el proceso de floración provocando efectos de reversión sex- ual. En nuestro caso, aunque no podemos descartar la influencia de los factores climáticos en este fe- nómeno, la situación en que se encuentran esas plantas es de "stress" ya que el bosque ha sido talado, ha aumentado el contenido de cal en el ambiente a causa de las construcciones que alli se producen, ha habido fuego (varias plantas del lugar han sido quemadas), o sea, que debido a la afec- tación antrópica observada en las áreas de estudio, los hábitos de vida de estos individuos han sido profundamente modificados. De esta forma se demuestra que al igual que Carica papaya L., Trichilia havanensis es capaz de revertir sus hábitos reproductivos en condiciones extremas, lo cual probablemente es un mecanismo para la perpetuación de la especie. CONCLUSIONES Se pone de manifiesto la dioecia en la especie T. havanensis en Cuba, encontrándose un gran sincronismo en la floración de esta especie entre los meses de enero y abril. Se detecta mediante las observaciones fenológicas, que los principales agen- tes polinizantes parecen ser los himenópteros del grupo de los Apidos, colectándose especificamente la abeja de miel (Apis mellifera). Las principales diferencias sexuales en las flores las encontramos en los caracteres del verticilo es- taminal, mayor longitud del tubo, y anteras desa- rrolladas de color amarillo en las flores masculinas y tubo estaminal más pequeno y anteras carmelitas y arrugadas en las flores femeninas. Existen otros caracteres que diferencian a las flores de ambos sexos como son: mayores dimen- siones de los petalos, disco nectarifero más desa- rrollado y apertura floral completa en flores mas- culinas, ocurriendo lo contrario en las femeninas. os fuertes cambios ecológicos observados en el hábitat de 7richilia havanensis, hacen que, al igual que ocurre con Carica papaya revierta sus hábitos reproductivos en condiciones extremas. LITERATURA CITADA мир J-E. . DRUMMOND. 1986. Floral ogy of Myristica pus Houtt. iy org rie i nutmeg of commerce. Biotropica 18: 32- ARROYO, M. T. К. & P. T. RAVEN. 1975. The а of subdioecy | in morphologically gynodioecious spe- cies of Fuchsia sect. Encliandra (Onagraceae). Evo- lution 29: 500 511 AUSPURGER, C. К. . Phenology, flowering syn- chrony and s set of six neotropical shrubs. Bio- tropica 15: 61 Britton, N. L. 8 ‚ WILSON. 1923. Meliaceae. Pp. 463-468 in Sc Ty Survey of Puerto Rico and the Virgin Islands, Vol. 5(1). New York Acad. Sci., New York. CASTILLO, S. € J. P. Сакавгаз. 1982. Ecología de la vegetación de dunas costeras: fenologia. Biótica 7: 51-560. CROAT, 1969. Seasonal flowering behavior in Central Panama. Ann. Missouri Bot. Gard. 56: 295- 307 € E x Darwin, C. 1877. The Different Forms of Flowers on Plants of "i Same Species. Reinwald, Paris. FOURNIER, L. 1974. Un método cuantitativo para la aud de caracteristicas fenológicas en árboles. Turrialba 24: 422-423. Sumario de patrones fitogeográficos y sus implicaciones para el a e Amazonia. Revista Acad. Colomb. Ci. Exact. 16: 6. 1 01 =] GHOSH, S. P. & S. P. SEN. 1975. The modification of sex Bc dene in papaya (C. papaya). HortScience 1-96. 5 башт E.O. 1 Flower е against nectar ilferage by ants. Biotropica 13: 2 . Supplement. НА RMS, H. 1940. Meliaceae. In: A Engle т & K. Prantl, e Natürlichen Pflanzenfamilien, e ено S. 1954. Determinación del sexo nite ar pa ра ya L. Сисен hipotética de los cromosomas Аса Тгор. 3: 229-249. мт е de A graphical method of АЕ Hees of plants. Rev. Polish Acad. Sci. 4: = с LEON, Hno. ^a Hno. ALAIN. 1951. Flora de cuba 2 Contr. Ocas. Mus. Hist. Nat. e “De La Salle" 10: 1-490. Lewis, D. 19 The joo of sex in flowering plants. Biol. Rev 1 47. LITTLE, R. J. 1 ew of floral food deception mimicries witht comments on floral mutualism. Pp. Volume 80, Number 4 1993 Albert Puentes et al. 869 Fenología y Estructura Floral de Trichilia havanensis 294-309 in C. E. Jones & R. Little (editors), Handbook of Experimental Pollination Biology. Van Nostrand Reinhold, New Luz DE ARMAS, M. DE ТА. 1985. Мане caracteristicas de las flores de Chamissoa m M Jacq. . al- 0: 7-10, . T. PRANCE. 1987. Species diversity, gy, pla ant animal interaction and their cor- relation with climates as illustrated by the Brazil nut family (Leoythidaceae). Pp. 69-89 in R. E. Diok- inson (editor), The ie фы of Amazonia. John & Sons, New Munoz MORALES, "acean EUH Е Carica papaya Bol. Resenas, Pl. Med. 12 OPLER, P. А., . FRANKIE & H. С. ВАКЕК. 1976. Rainfall as a pee in the r нем timing and sin- cronization of anthesis y tropical trees and shrubs. J. Biogeogr. 3: 231-2 PENNINGTON, T. D. & T. D м 1975. А — monograph of the Meliaceae. Blumea 22: 1-5 & 1981. Meliaceae. /n: Flora rom 420. tropica Monograph 28: 1- Ramia, M. 1981. Fenología de árboles en el n па сам tropical. Mem. Soc. Ci. Nat. La Salle. d. л & Y. BRITO. 1987. Patrones de floración fructificación en una comunidad pantanosa tipo morichal (Calabozo-Guarico, Venezuela). Acta Ci. Ve- nez. 81. SARMIENTO, MONASTERIO. 1983. Life form and еле, Рр. 78-108 in F. сыы (editor), Tro pical Savanas. Elsevier, Amster STEVEN, D., D. M. Winpsor, Е. E. Pu wt B. LEON. 7. Vegetative erus of a palm assemblage anama. Biotropica 19: 342-355. STYLES, B. T. 1972. The flower бе of the Meliaceae and its bearing in tree breeding. Silvae Genet. 21: 175- VILAMAJO, D., R. P. CAPOTE, M. FERNANDEZ, 1. ZAMORA & B. GONZALEZ. 1989. Mapa bioclimático escala 1: 3 000 000. Nuevo Atlas oo de Cuba. ICGC e Inst. de Geografia, La Habar Wa ter, Н. & H. ЏЕТН. | Кина Welt- atlas, Verb. Gustav Fischer Verlag, Jena [sin pagi- nación] THE GENUS CLERODENDRUM (VERBENACEAE) IN MESOAMERICA’ Ricardo M. Rueda? ABSTRACT Clerodendrum is the largest genus of Verbenaceae, m about 560 species and varieties. It is most abundant in Africa and Asia, with only 20 known: — native to ше five of these species are native, and the dud are cultivated orld. In Mesoamerica, as ornamental plants; several are more or less naturalized. Three of the аса species and one of the native ones are first records for the Mesoamerican region RESUMEN Clerodendrum es el género más grande de la familia Verbenaceae con cerca de 560 especies y variedades. La mayoría de estas especies ocurren en Еј у Asia; solamente 20 especies son nativas al Nuevo Mundo. En M as solamente se conocen 15 especies y u ariedad: de ést 9 especies son nativas; el resto son cultivadas como Heg кажа y ее son IM Tres de las especies cultivadas y una de las nativas son nuevos reportes pa a Mesoamér The genus Clerodendrum, comprised of small trees, shrubs, woody vines, and perennial herbs, is well known for its showy and attractive ornamental species. This genus, with about 560 taxa, is the largest in the Verbenaceae and is taxonomically complex. It is native to tropical and subtropical regions of both hemispheres, most abundant in the Old World, and poorly represented in the New World, where, however, it is frequently cultivated and often naturalized. The name Clerodendrum is formed from two Greek words, “kleros”” (chance (tree). The reason for the application of the name to the genus is obscure, but it may well come from the ancient belief that some species had healing properties, while others acted in exactly the opposite way. Historically, the taxonomy of the genus Clero- dendrum has been confused (see Munir, 1989). Much of this can be attributed to the wide use of or destiny) and “dendron” ornamental species with different varieties. Many species seem to be closely related, with some hy- bridization reported. Some of the species are ex- tremely variable, due to environmental factors and horticultural selection, which has led taxonomists to describe many taxa. Because several species are found mostly in cul- tivation in Mesoamerica, they have been poorly collected. The large size of the leaves and inflo- rescences also led to a tendency to collect few leaves and partial inflorescences. Thus, few com- plete collections are available for study. In addition, Burman's (1737) spelling **Clerodendron" instead of the correct spelling **Clerodendrum" of Lin- naeus (1753) is often found in the literature, thus causing confusion. This study is based on (1) an exhaustive examination of the types, (2) morpho- logical characters, and (3) geographical informa- tion available. of the manuscript. The ! е study was completed as : pee fulfillment of the M.Sc. degree in biology at the University of Missouri, St. Louis. I am grate e to Alw entry and Victoria Sork for direction and critical revision o Selen b of the : due Mis us Rhe: | Garden was particularly valuable. Financial support was provided by a Fulbright fellowship through the following institutions made thei K, L, MA, MEXU, MICH, MO, MSC, NY, P, TEFH, n American Scholarship Program of American Universities (LASPAU). The sual dai collections available: BR, CAS, CHAPA, CR, DUKE, E, ENCB, F, GH, : Departamento de Biologia, Universidad Nacional Autónoma de Nicaragua, León, Nicar TEX, US. The illustrations were prepared by J. Myers. rag ANN. MISSOURI Bor. Garb. 80: 870-890. 1993. Volume 80, Number 4 1993 Rueda Clerodendrum in Mesoamerica MORPHOLOGICAL CHARACTERISTICS HABIT A liana habit is uncommon in Verbenaceae but is found in Petrea and Holmskioldia, the latter cultivated in the Mesoamerican region. WOOD The genus does not have commercial value as a source of timber, probably because of its small stature. Record & Hess (1941) studied five species from Mesoamerica and the Caribbean region. They reported the wood of Clerodendrum to be yellow or whitish with a medium luster, moderately heavy with a fine texture, and easy to work with. Wood parenchyma is scanty, paratracheal. Wood fibers are in part septate, the pits numerous and medium- sized. Most of these anatomical characteristics are shared with such other genera of the family as Aegiphila and Callicarpa (both Verbenaceae). However, Aegiphila has wood parenchyma vasi- centric to aliform and confluent, and Callicarpa has wood pores much larger than Clerodendrum. LEAVES The leaves are simple decussate-opposite, ter- nate, or apically clustered and variable in size. Clerodendrum aculeatum has leaves ranging 0.9- 4 cm long x 0.3-1.4 cm wide; C. paniculatum 6-35 cm long х 6-30 cm wide. Other American Verbenaceae genera with decussate-opposite leaves are Verbena, Petrea, and Duranta; however, none of them have the tremendous variability in size displayed in Clerodendrum. Like Clerodendrum, most Mesoamerican Verbenaceae have simple leaves, except Vitex, which has palmately com- pound leaves. Leaf shape is also variable; lanceolate in Clerodendrum wallichii; elliptic in C. aculea- tum; and ovate in C. japonicum. VENATION PATTERNS Venation patterns are variable, although most species are brochidodromous, sometimes only well above the base. Other less frequent types are cras- pidodromous as in C. bungei, and eucamptodro- mous as in C. aculeatum and C. costaricense. The angle of divergence of secondary veins from the midvein is variable but in most leaves is acute (45?— 09). In almost all of the species the secondary veins are broadly ascending and strongly arcuate. Тће number of secondary veins can vary from 5 to 9 on each side of the midvein. PETIOLES Leaves are petiolate. The variability in petiole length is remarkable, 0.1-0.7 cm in C. pittieri and 1-40 cm petioles are decurrent. Almost all of the species m in C. paniculatum. In a few species present elevated leaf scars, these becoming spi- nescent in such species as C. aculeatum. INDUMENTUM The genus varies from glabrous or glandular- pubescent to pubescent with simple hairs. Calyx, petiole, pedicel, leaf, and twig pubescence are usu- ally of different densities within a species. Most genera in Verbenaceae have annulate nodes, but some species of Clerodendrum have a band of tomentose hairs on the nodes. Disk-shaped glands on the calyx and scales on leaves are also present in some species. INFLORESCENCES The inflorescences are both axillary and ter- minal, sometimes on the same plant. The inflores- cences may range from 1 to 39 cm long and 1 to 25 cm wide. These inflorescences may be cymes, panicles, or solitary flowers, determinate or cen- trifugal, the flowers crowded or sparsely arranged. These types of inflorescences are also present in other genera such as Aegiphila and Callicarpa, but the flowers in these genera are regular and the stamens equal, unlike Clerodendrum, which has irregular flowers and didynamous stamens. Almost all the species of Clerodendrum have foliaceous bracts and linear-lanceolate bracteoles. CALYX The calyx varies in size, ranging from 2 to 18 mm long. It is gamosepalous, commonly green, less often red or white, almost always campanulate, rarely elliptic, truncate, 5-lobed, glabrous or var- iously pubescent, sometimes glandular or lepidote, often accrescent in fruit. The accrescent calyx in fruits does not become thick and indurate as in Aegiphila, nor reflexed and patelliform as in Cal- licarpa. COROLLA The size of the corolla ranges from 6 to 40 mm long to 3 to 20 mm wide, sometimes larger in the species of the Old World. It is gamopetalous, hy- pocrateriform, and may be white, red, pink, or purple. The limb is 5-parted, spreading at anthesis, mostly subactinomorphic. The corolla is glabrous, 872 Annals of the Missouri Botanical Garden pubescent, or glandular-puberulent. The corolla tube is not markedly curved and more or less 2-lipped, as in Vitex. STAMENS The four stamens are didynamous, inserted on the corolla tube more or less at the same level They are long-exserted, alternate with the corolla lobes; the anthers are glabrous, ovate or oblong, opening by longitudinal slits. This general pattern is shared by the majority of genera in Verbenaceae. POLLEN The pollen is spheroidal, with echinate processes on the tectum. Rao & Tian (1974) described the pollen grains of Clerodendrum inerme Gaertner as spheroidal, 3-zonocolpate, the surface spinulifer- ous, and interspinular areas reticulate. Verbena- ceae, including Clerodendrum, have pollen grains that are binucleate or seldom trinucleate, and most commonly triaperturate (Cronquist, 1981). PISTIL The pistil protrudes more or less the same length as the stamens. It consists of a 2-carpelled ovary and a terminal style that is elongate and forked apically. The ovary is superior, imperfectly 4-loculate, glabrous, each locule 1-ovulate, the ovules pendulous, laterally attached, and hemianat- ropous. FRUITS The fruit is drupaceous, mostly subglobose or obovoid, glabrous, usually separating at maturity into 4 pyrenes. The exocarp is more or less fleshy, often black or stramineous when mature, and the endocarp bony or crustaceous. Fruits range from 9 to 14 mm long and 5 to 8 mm wide. The fruits are not very useful in separating species because they are remarkably similar. SEEDS The seeds may be 4, as in most Verbenaceae, or (by abortion) sometimes only 1-3. They are mostly oblong, with little or no endosperm. Most other American Verbenaceae uniformly have 4 seeds. DISTRIBUTION AND ECOLOGY Clerodendrum is most abundant in Africa and Asia, although a few native species occur in the New World, many more being cultivated and nat- uralized. In the New World the native species of Clerodendrum range from the northern part of Mexico (Tamaulipas and Oaxaca), and the West Indies, south to Peru (Ayacucho), and northern Argentina (Misiones). Clerodendrum occurs in the Old World from northern Africa (Senegal and Egypt) southward through the remainder of Africa and Madagascar. It also occurs eastward to China and Japan, and southward to Ceylon and northern New Zealand. Most of the species are in the Old World, especially in Zaire (67 species, 10 endemic), Tan- zania (60 species, 13 endemic), Madagascar (67 species, 14 endemic), and the great Sunda Islands (66 species, 14 endemic) (Moldenke, 1980). The number of species in the New World is about 20, with most native species located in Colombia (6 species, 1 endemic), and Cuba (10 species, 7 en- demic). The species of Clerodendrum are reported grow- ing in a wide variety of habitats. Clerodendrum ligustrinum (Fig. 1) is reported along the bank of a creek in Belize (Lundell 6962, F); C. cense (Fig. 2) was collected in cloud forest in Costa costari- Rica; C. pittieri (Fig. 3) is reported growing in a mangrove area in Costa Rica on soil with up to 64%o salinity (Jiménez & Soto, 1985); and C. spe ciosissimum (Campeche, Mexico; Calzada et al. 6724, MEXU) is found on limestone formations. Other habitats from which Clerodendrum has been reported include sand dunes (C. ligustrinum, Chia- pas, Mexico; Breedlove & Thorne 20913, MO) and cloud forest (C. philippinum, Nicaragua; Wil- liams et al. 24693, F) Clerodendrum has been reported in association with many other plants. Clerodendrum bungei and C. philippinum are reported growing with Quercus in Chiapas, Mexico (Breedlove 6837, CAS, and 37852, MEXU). Altitudinally, species of Clerodendrum are more widely distributed in lowlands than in mountains. Most of the Mesoamerican species can grow very near sea level. Consequently, collections of Clero- dendrum from higher elevations become progres- sively less frequent. Some of the species that may reach well into the mountains include C. bungei (2,100 m, Chiapas, Mexico; Breedlove 6055, F) and C. philippinum (1,400 m, Matagalpa, Nica- ragua). The scarcity of high altitude records might be due partly to a decline in both individuals and species at higher altitudes. ANIMAL INTERACTIONS POLLINATION AND DISPERSAL Pollination in Clerodendrum is mostly carried out by butterflies, moths, and bees, which extract the nectar from the base of the corolla tube. In Volume 80, Number 4 873 Clerodendrum in Mesoamerica 0 200 400 600 800 1000km o 100 200 300 400 500 600 miles FIGURE 1. most species the stamens and style project from the lower side of the horizontally oriented flower; pollen is carried off on the underside of the insect. In such species as C. aculeatum, the flowers last more than one day, during which time the stamens first project and then curl back under the flower and leave the style exposed (Corner, 1952). In other species such as C. serratum, the stamens and style arch over the top of the flower and one of the petals is modified into a lower lip, which acts as a landing platform for visiting insects. In this last case the pollen is carried off on the upper side of the insect. The fruits of Clerodendrum species are eaten by birds, which disperse the seeds (Moldenke, 1985). Ridley (1930) described the fruits of the genus as pulpy, often brightly colored, and distributed by birds. In many species the calyx provides a con- trasting color, similar to those of red arils sub- C. ligustrinum e var. ligustrinum A var. nicaraguense Distribution of Clerodendrum ligustrinum in Mesoamerica. tending black seeds. The pseudo-aril present in some species is actually a placental part of the pericarp that acts as an attractant to birds in the fruit dispersal process (van der Pijl, 1982). ANTS The relationship of some Clerodendrum species to ants was reported long ago. Beccari (1884) described C. fistulosum from Malaysia with inter- nodes inhabited by Camponotus (Colobopsis) cle- rodendri. Mejia & Zanoni (8508, MO) reported many ants on plants of C. speciosissimum in the Dominican Republic. Dreisig (1988) reported the inflorescences of C. speciosissimum occupied by ants feeding on the nectar produced by the bracts. These ants belonged to three species, each having exclusive use of an inflorescence: Wasmannia au- ropunctata (Roger), Camponotus abdominalis 874 Annals of the Missouri Botanical Garden 600 800 1000km о 200 400 EAS ae | | ' | y у | : 0 100 200 300 400 500 600 miles FIGURE 2. floridanus (Buckley), and Conomyrma sp. The foregoing observation, and especially the presence in some species of hollow stems and/or extrafloral nectaries (see Elias, 1983), clearly show that ants are intimately involved with some Clerodendrum species. CHROMOSOME NUMBERS The chromosome numbers of the genus Clero- dendrum are largely unknown. According to gen- eral chromosome indices (Cave, 1959, 1964; Bol- khovskikh et al., 1969; Moore, 1973; Kumar & Subramanian, 1987; Goldblatt, 1981, 1984, 1988), only 26 species and varieties have been studied to date, out of more than 560 species and varieties reported for the genus. Because Clerodendrum is mainly a paleotropical genus, most of the reports are from Old World taxa, among which are some Distribution of Clerodendrum costaricense, C. molle, and C. A C. costaricense B (C. molle € C. aculeatum a MEE 7" ON DNE f ~, a = aculeatum in Mesoamerica. of the species cultivated as ornamentals in Me- soamerica. Almost 50% of the species that are either cultivated or native in Mesoamerica have known chromosome numbers. Because of the poor cytological data on the genus there is not total Thus, according to Марни n the hoc chromosome number is x — r 23; Raman & Kesavan (1963), x — 15. Ms xul Mehra & Bawa (1969), x — 12, 13. It is at least clear that chromosome number in the genus 24 in C. 184 in C. ugandense. is variable, ranging from 2n — aculeatum to 2n — USES The most popular use of Clerodendrum species worldwide is as ornamental plants. All of the Me- soamerican species except C. costaricense have Volume 80, Number 4 1993 Rueda Clerodendrum in Mesoamerica 200 400 600 800 1000km 0 100 200 300 400 500 600 miles FIGURE 3. Distribution of Clerodendrum pittieri in been planted ornamentally at one time or another. Clerodendrum thomsonae and C. philippinum are reported as ornamentals in Costa Rica (Pittier, 1978), and these two species plus C. umbellatum and C. paniculatum are cultivated as ornamentals in Nicaragua (pers. obs.). Molina (1975) listed sev- en species of ornamental Clerodendrum in Hon- duras; at least four species are cultivated as or- namentals in El Salvador (Standley & Calderon, 1941); and Gibson (1970) reported C. philippin- um as one of the common ornamentals in Central America. The folk use of the New World species of Clero- dendrum is limited. In Yucatan, Mexico, the leaves of C. ligustrinum are used as a diuretic, even in the case of kidney stones (Mendieta & Del Amo, 1981). In Veracruz, Mexico, a leaf infusion 15 applied to snake-bites (Del Amo, 1979). In Petén, Mesoamerica. Guatemala, the leaves are also used in cooking to flavor fish (Gibson, 1970 Clerodendrum is also important in having chem- ical compounds that may be used in medicine and industry. Gibbs (1974) reported that several spe- cies of Clerodendrum have D-manitol, saponins, tannins, cyanogenic (glycosides), quinones, and steroles. SYSTEMATIC TREATMENT Clerodendrum L., Sp. Pl. ed. 1, 2: 637. 1753. TYPE; Clerodendrum infortunatum L. (Linn- 10.1, microfiche, Ovieda E Sp. Pl. ed. 2: 637. 1753. TYPE: Ovieda is L. (Linn-807. T. mic m he, МО). ra ен L., Sp. Pl. ed. 1, 2: € 1753. : Vol- kameria aculeata L. (Linn- 809. |. mic iie His "MO. 876 Annals of the Missouri Botanical Garden imc e на РІ. Hort. Schoenbr. 3: 48, pl. 798 Volkmannia јаротса Jacq. pen | Vida bi ibid) Cornacchinia Savi, Mem. Mat. Fis. Soc. Ital. Sci. Mo- dena, Pt. Mem. Fis. 21: 184. pl. 7. 1837. TYPE: Cornacchinia fragiformis Savi (type illustration ibid). Mostly shrubs, often scandent (arborescent in some species of the Old World), rarely herbaceous, unarmed, rarely the petiole base spinescent, gla- brous or variously pubescent; branches usually te- tragonal, sometimes medullose or hollow; nodes annulate sometimes with a band of hairs; leaf scars often prominent. Leaves simple, decussate-opposite or ternate, sometimes apically clustered, entire or variously dentate, sometimes lobed, mostly petio- late, glabrous or pubescent, often punctate be- neath. Inflorescences cymose, cymes mostly loose- flowered (dense and capitate in some species of the Old World), pedunculate in the upper leaf axils, paniculate at the apex of the branches, often ag- gregate in terminal corymbs. Flowers actinomor- phic or zygomorphic, often showy, perfect, hypog- ynous; calyx gamosepalous, green, sometimes the same color as the corolla, campanulate to tubular, Кеу то SPECIES OF CLERODENDRUM IN MESOAMERICA sometimes inflated, the rim truncate to 5-toothed, often wide-spreading or accrescent with age; co- rolla gamopetalous, mostly hypocrateriform, usually purple, pink, red, or white, tube straight, sometimes incurved, limb 5-lobed, mostly subactinomorphic (plants bilabiate in some species of the Old World); stamens 4, didynamous, attached to the corolla tube, usually long-exserted; anthers glabrous, ovoid, 2-thecate, mostly basally attached, opening by lon- gitudinal slits; pollen grains 3-6 colpate, tectum with echinate processes; ovary superior, 2-carpelled, imperfectly 4-locular, globose or oblong, glabrous, each locule 1-ovulate, ovules pendulous, laterally attached, hemianatropous; style terminal, filiform; stigma shortly bifurcate, sometimes globose. Fruit with calyx persistent, subtending or enclosing the fruit, sometimes colored; fruit drupaceous, often 4-sulcate, often brightly colored or black when mature, usually separating into 4 pyrenes or these sometimes united in pairs, by abortion sometimes 1—3 seeded; seeds oblong, endosperm absent; chro- mosome numbers: x = 12 or 23 (Darlington & Janaki, 1945; ee & Wylie, 1955; Hsu, 1968), 2n = 24-18 . bungei C. B dpi i ;. intermedium . 6. C. panic pri TG: d 8. C. glabru `. thomsonae . C. umbellatum la. Flowers pink, white, red, or purple, if white with the stigma globose; calyx always glabrous when flowers are pink; a scars never spinescent; да cultivated, sometimes naturalized. 2a. "The largest leaves more than 15 cm lon 3a. Inflore sscences сутове, согуп ¿bos 3: 9 с a. Calyx 3-5 mm long; corolla pns l. 5-2 2 5 cm long (Fig. 4A) cc 4b. Calyx 1-1.5 cm long; corolla tube 1-1.5 ст long (Fig. 4B) 2. 3b. Inflorescences paniculate, 8-40 cm long. a. abs not lobed. eaves glabrou 7a. Margins of Tova crenate, appressed teeth lacking; calyx 3-4 mm long EUN RS 7b. Leaves margin serrate with appressed teeth; calyx 8-16 mm n long . TR 4. C. jap 6b. Leaves pubescent sD G 5b. Leaves 3-lobed ... 2b. The largest leaves less than 15 cm long. Ва. е white to greenish white. rolla tube ca. 1.5 ст long; petioles 0.3-1 cm long rolla tube ca. 9 mm long; petioles 1-2.5 cm long 8b. CM СРЕ? red to pink. 10a. Calyx globose, 15-20 mm long, Pa lobes 1-1.8 ст long (Fig. 4C) ______ 10b. Calyx campanulate, not globose, 7-13 mm long, calyx lobes 5-10 mm ЊЕ (fu. 4D) aie lb. Flowers pink, white, or yellowish, if white with the и нак not capitate; calyx титан if glabrous with the leaf scars spinescent and flowers lla. The largest leaves more than 4 12a. Calyx glabrous; corolla tube c € Leaves more than 10 cm i ae "ilipsoid .. 1. s pink; plants n .9 ст der p scars eee but not spinescent. C. costaricense 1 3b. "Pur less than 10 cm i calyx campanulat b. ve Calyx pubescer sin corolla tube 1.8-2.5 c llb. е PAEA leaves less than Leaf scars spine- -tipped; calyx lob Mp ын dieti t cm lon 4.5 cm long; leaf scars irse bed, pubescent; fruit subspheric ccoo. 15b. Leaf scars not spine-tipped; calyx truncate, glabrous; fruit obovate ________ 1 ы = 7. ligustrinum 12а. С. Шан v уаг. пао uen 13. C. molle .. 14. C. aculeatum 5. C. pittieri Rueda 877 Volume 80, Number 4 1993 Clerodendrum in Mesoamerica иогәрүо)) “Y "Heg euosmoy? 7) 7)—'(zp9z 2X1] P 214) 1oneqos штита 7) "рб jjonxvpy ) (оско PUNO) 191100 univ 9quin 7) 'а4— (229 [ppnais 19 unq шпаригролоју “Y —:Jo зопзовалоџјиј ср ANNOY 878 Annals of the Missouri Botanical Garden 1. Clerodendrum bungei Steudel, Nomencl. Candolle (1880) and Sayre (1975) stated that the Bot. ed. 2, 1: 382. 1840. Clerodendrum foe- tidum Bunge, Enum. Pl. China Bor. 52. 1833, non D. Don, 1825. TYPE: China, Bunge 296 (holotype, LE not seen). Shrub to 3 m tall; branchlets medullose, sub- terete or tetragonal, puberulent, lenticellate; leaf scars elevated. Leaves decussate-opposite, with an unpleasant odor when crushed; blade deltate-ovate or elliptic, 5-20 cm long, 3.5-15 cm wide, base truncate or acute, often decurrent, apex acute or acuminate, margin coarsely dentate, teeth. often apiculate, puberulent, punctate beneath, often with glandular hairs on both surfaces, usually with dis- coid glands near the base beneath; midrib puber- ulent, secondaries 3-5 pairs; petioles 2-14.5 cm long, glabrous. Inflorescences cymose, terminal, rarely supra-axillary, subcapitate, corymbose 4—9 cm long, 3-12 cm wide, many-flowered, showy, puberulent; peduncles, 0-5 cm long; pedicels 1- 6 mm long. Flowers usually fragrant; calyx tubular or campanulate, 3-5 mm long, outside glabrous and punctate, sometimes with a few disk-shaped glands, 5-lobed, lobes ovate, 1-3 mm long, acute or acuminate; corolla hypocrateriform or infundib- ular, usually red or rose to rose-purple, glabrous, 1.5-2.5 cm long, ca. 5 times as long as the calyx, limb 5-lobed, 1-1.5 cm wide, lobes oblong-ovate, 4—7 mm long; stamens 4, filaments long-exserted, to ca. 8 mm beyond the throat of the corolla tube, 1.2-1.4 ст long; ovary globose, style 2.8-3 ст ide exserted, shorter than or equal to the stamens. chromosome number: 2n — 10 1969). uit not seen der ние et al., Distribution and habitat. This species occurs in Asia from China to northern India; it is cultivated worldwide and naturalized in tropical and subtrop- ical regions. In Mesoamerica it has been collected in Mexico and El Salvador. Clerodendrum bungei is found on clay, limestone, or rocky soil, moist pastures, or frequently on oak-covered slopes, at altitudes up to 2,500 m. Vernacular names and uses. “hortensia.” The most important use is as an or- namental plant. The whole plant is decocted for lung weakness and cough (Duke & Ayensu, 1985). Note. Clerodendrum bungei is frequently misidentified as C. philippinum. These two species can be separated using the length of the corolla in relation to the calyx. The corolla tube in C. phi- lippinum only slightly exceeds the calyx in length, bungei is several “Bocamelia,” whereas the corolla tube in C. times longer than the calyx. The holotype of C. bungei is probably deposited in LE. Although de type may be in P, S. Barrier informed me that it is not in Paris. Selected specimens examined. MEXICO. CHIAPAS: N end of San Cristóbal de las Casas, 2,100 m, 6 July 1964, Breedlove 6055 (ENCB, F); W of Lon center along the trail to Paraiso, 2,100 m, 4 Aug. 1 , Breedlove 7 (CAS, ENCB, MICH); along us ae 190, 20 mi. NW of Comitán de Dominguez, 2,000 m, 23 June 1965, сан 10440 (CAS, ENCB, Е, MICH, TEX, | za, Escuintla, 14 July 1947, Matuda SALVADOR: cultivated in gardens, 1922, Calderón 731 2; е philippinum Schauer in , Prodr. 11: 667 (as “Clerodendron”). 1847. TYPE: = Philippine Islands, Cuming 1096 (lectotype, rue by Howard & Powell (1968), Volkmannia japonica Jacq., Pl. Hort. Schoenbr. 48, 1. 338. 1798. TYPE ILLUSTRATION: Japan, bid, pl. C aioe maium Perun var. digni ан Sweet, Hort. Brit., ne 6. n с Curtis, Bot. 3: pl. 1834. Clerodendrum pets var. plenior Schauer in DC., : 666. 1847. Е: Martinique, collector аен s.n. (holotype, "CDC 1908. 11: 666. 30, microfiche, MO). Clerode sigs philippinum v var. subfertile Mold., Phy- tologia 25: 368. 1973. TYPE: Sumatra, Boeea 8394 es MICH not seen — Shrub to 2 m tall; mostly reproducing vegeta- tively; branches medullose, angulated, puberulent, lenticellate; leaf scars often elevated. Leaves de- cussate-opposite or apically clustered; blade broad- ly ovate, 6-29 cm long, 5-28 cm wide, base cor- date to subtruncate, often subcuneate, apex acute, margins coarsely serrate or sometimes the teeth apiculate, lightly strigillose-pubescent on both sur- faces, usually more densely so on the venation beneath, sometimes with discoid glands near the ase beneath; midrib glabrous, secondaries 3-5 pairs; petioles 2-20 cm long, puberulent. Inflo- rescences cymose, terminal corymbose, 3-6 cm long, 4-9 cm wide, densely puberulent; peduncle 0-2 cm long; pedicels 1-5 mm long; bracts and bracteoles foliaceous 1-3 cm long. Flowers fra- grant, often sterile; calyx campanulate, 1–1.5 cm long, sometimes with white blotches at or near the margins, sparsely strigillose-puberulent, outside with scattered disk-shaped glands, 5-lobed, lobes lan- ceolate, 4-10 mm long; corolla often “doubled,” hypocrateriform, white to pink, glabrous, tube am- pliate, 1-1.5 cm long, limb 5-lobed, 2 cm wide, Volume 80, Number 4 1993 Rueda 879 Clerodendrum in Mesoamerica stamens and pistils often all modified into super- numerary petals; chromosome number: 2n — r 52 (Kumar & Subramanian, 1987) Distribution and habitat. of Clerodendrum philippinum is obscure, but it is likely native to Southern China to southern trop- ical Asia. It is cultivated and naturalized in tropical and subtropical regions worldwide. In Mesoamerica it has been collected from Mexico to Panama. This The original range species is found in pastures, roadside embank- ments, riverbanks, open hillsides, slopes with Quer- cus and Pinus, and in disturbed high forest, at altitudes up to 2,100 m. Vernacular names and uses. *Bocamelia," ээ 66: ^*^ 66 mil flores,” verbe- ' This species is one of the "flor de concha," “jasmin de noche, 99 && ,9 - 46 "misteriosa olorosa," "spanish jasmine, na," and “viuda alegre.’ most common ornamentals of Mesoamerican gar- dens. Note. The nomenclature of this species has been very confused. In 4 Ventenat described Volkameria fragrans (Clerodendrum philippin- um); he stated that this species was cultivated at the gardens in Paris and known as Clerodendrum fragrans and Volkameria japonica. But Volka- тета japonica could not be used as a basionym in Clerodendrum for the plants now treated under C. philipp already occupied (Thunberg, 178 drum fragrans was published by Ventenat in syn- inum because the epithet japonica was Cleroden- onymy, and therefore it is not a validly published name. Clerodendrum fragrans Willd., was illegit- imate as published by Willdenow (1809) because he did not use Jacquin's type (Dan Nicolson, pers. comm.). To solve this problem, Howard & Powell (1968) designated the earliest valid name, Clero- dendrum philippinum Schauer (1847). Selected specimens examined. MEXICO. CHIAPAS: 8 km E of Las Margaritas, 1,700 m, 15 Sep. 1974, Breed- love 37852 (MEXU). Hidalgo: Santa Ana, on Pan Amer- ican Highway, 24 May 1939, Frye & Frye 2647 (MO). ¿LIZE. EL CAYO: without precise locality, Mar.-June 1933, Chanek 121 (F). GUATEMALA. BAJA VERAPAZ: El Nino Perdido, W of km 150/151, Aug. 1975, Lundell & Contreras 19742 (F, MO). EL SALVADOR. SAN SALVADOR: vicinity of San Marcos, 5 Apr. 1922, Standley 22796 (US). HONDURAS. CHOLUTECA: vicinity of San Marcos de Colon, 960-1,150 m, 12-22 Jar 1. 1949, Standley 15796 Aran- juez, 980, o & е ds 4 (MO). Costa Ee CARTAGO: Turrialba, 1966, Moldenke 1325 (TEX). P um BOCAS DEL auc vicinity of Almirante Bay, 19 Dec. 1938, Von Wedel 15 (MO) 3. Clerodendrum intermedium Cham., Lin- naea 7: “150” (i.e., 105). 1832. TYPE: Phil- ippine Islands, Cuming 481 (lectotype, here designated, Shrub to 2 m tall; branches and branchlets te- tragonal or subterete, medullose, puberulent, len- ticellate; nodes annulate, the lower ones marked with a band of hairs; leaf scars slightly elevated. Leaves decussate-opposite; blades broadly ovate, 7—22 cm long, 6-20 cm wide, base deeply cordate, apex abruptly acute or acuminate, margin crenate, often abundantly light-dotted above and densely squamulose beneath; midrib puberulent, second- aries 5-7 pairs, the 2-4 lowest issuing palmately from the leaf base; petioles 2.5-23 cm long, len- ticellate, strigillose. Inflorescence paniculate, ter- minal, 12-20 cm long, 8-18 cm wide, composed of 5-8 pairs of lax divaricate cymes, many-flow- ered; peduncles variable in length; pedicels 1-13 mm long, glabrous; bracts foliaceous, oblong-ellip- tic, the broadly linear, strigillose. Flowers with calyx cam- racteoles and prophylls lanceolate to panulate, 3-4 mm long, granular lepidote, 5-lobed, lobes triangular-acuminate 2-3 mm long, equal to or longer than the tube; corolla hypocrateriform, bright red, outside puberulent and glandular, tube ca. 1.5 cm long, limb 5-lobed, ca. 1.3 cm wide, lobes oblong-ovate, 3-6 mm long, apically round- ed-obtuse; stamens 4, long-exserted; filaments long- exserted ca. 2 cm beyond the throat of the corolla tube, 2.1-2.2 cm long; ovary oblong, style 2.3- 2.5 cm long, exserted beyond the stamens. Fruit drupaceous, subtended by the calyx, 4-6 mm long, splitting into 2-bilobed, 2-seeded pyrenes at ma- turity; seed curved. Distribution and habitat. Clerodendrum in- termedium is native to the area between Taiwan and the Celebes and Sumatra islands. It is cultivated as an ornamental worldwide. In Mesoamerica it has only been collected in Honduras, which is the first report of this species in Mesoamerica. Vernacular names and uses. |n Honduras Clerodendrum intermedium is used as an orna- mental plant. Gibbs (1974) reported the presence Reis (1973) reported that the leaves of this species are made into a of cyanogens in the stem. poultice and used for headaches. Note. There are three species of Cleroden- drum cultivated in Mesoamerica similar to C. in- termedium. Clerodendrum japonicum has leaves similar to the leaves of C. intermedium, but the flowers are twice as big. Clerodendrum interme- dium is also similar to C. paniculatum L. (not C. paniculatum Perrottet), but the latter has bigger inflorescences and lobulate leaves. The glabrate leaves of C. intermedium may be distinguished 880 Annals of the Missouri Botanical Garden easily from the pubescent leaves of the third spe- cies, C. speciosissimum. The type of this species is Cuming 481, as pointed out by Schauer (1847). He also commented that this material was deposited in B. This type seems to have been destroyed during the World War II, but there is one isotype in L, which is here designated as lectotype. HONDURAS. ye Lan- cetilla valley, near Tela, 20-600 m, 6 Dec. 1927 Mar 1928, Standley 56847 (F). MORAZAN: vicinity of El Zamorano, 800-850 m, 31 Aug. 1949, Standley 23294 Е); El Zamorano, 800 m, 4 Oct. 1943, Valerio 1092 (F) Specimens examined. 4. Clerodendrum japonicum (Thunb.) Sweet, Hort. Brit., ed. 1, 322. 1826. Volkameria japonica Thunb., Nova Acta Regiae Soc. Sci. Upsal. 3: 208. 1780. TYPE: Japan, Thunberg s.n. (lectotype, here designated, UPS-14579, ). microfiche, nep coccineum H. J. Lam, Verb. Mala 296. 1919. TYPE: Java, Buijsman 74 (holo. nu BO not Мем Shrub to 3 m tall; branches and branchlets te- tragonal or subterete, medullose; puberulent, len- ticellate; nodes annulate, the lower ones with a band of hairs. Leaves decussate-opposite; blades ovate, 10-30 cm long, 10-25 cm wide, base deep- ly cordate or auriculate, apex acute or acuminate, margin glandular serrate with appressed teeth, of- ten light-dotted, puberulent on the venation and lipidote-squamulose beneath; midrib puberulent, secondaries 5-8 pairs, the 2-4 lower pairs issuing palmately from the leaf base, upper pairs ascend- ing; petioles 2-15 cm long. Inflorescences panic- ulate, supra-axillary and terminal, 15-35 cm long, 13-30 cm wide; axillary peduncles 2-10 cm long; cymes 10-20 cm long, 4-10 cm wide, solitary in uppermost leaf axils; terminal panicles with 4—6 pairs of cymes; main peduncles 4-10 cm long, puberulent; pedicels 5-20 mm long; bracts folia- ceous, a pair subtending each pair of cymes in the terminal panicles. Flowers with calyx campanulate, 8-16 mm long, puberulent, 5-lobed, lobes trian- gular-lanceolate, 10-14 mm long, apically acu- minate; corolla hypocrateriform, deep red or scar- let, outside puberulent, tube curved, ca. 2 cm long, the limb 5-lobed, ca. 2.5 ст wide, lobes elliptic or oblong, 8-12 mm long; stamens 4, filaments long- 3.5 ст beyond the throat of the corolla tube, 5-5.3 cm long; ovary oblong, style exserted, ca. 6-7 cm long, exserted, shorter or equal to the stamens. Fruit not seen. Distribution and habitat. Clerodendrum ja- ponicum is native to Nepal, China, and upper Burma. It is also cultivated in tropical and sub- tropical regions and naturalized in Japan, Brazil, and Mexico. In Mexico this species is found in secondary deciduous forest, in secondary middle- elevation forest on rocky soil, and in coffee fields, at altitudes up to 1,30 . Vernacular names and uses. Floral bracts are chewed for hematuria and as a poultice for painful arthritis (Duke & Ayensu, 1985). In Mexico it is grown as an ornamental. Note. drum cultivated in Mesoamerica that are similar There are three species of Cleroden- to C. japonicum (see note under C. intermedium). The type material of Clerodendrum japonicum is deposited in UPS-Thunb. under the numbers 14579 and 14580. According to Roland Moberg, both numbers probably come from the same material divided into two sheets. However, neither of the two was designated as the holotype. The best spec- imen, 14579, should be regarded as the lectotype and 14580 as an isolectotype. Specimens M није Mexico. Oaxaca: Tuxtepec, 20 Apr. 1923, Reko 4626 (US). VERACRUZ: | km E of Tenejapa, ш autusco- Cosmatepec road, 1.300 m, 2 Aug. 1979, Avendano & Calzada 445 b ios hitlan, Col- orado, Misantla- Tenochitlán road, 920 m, 27 Apr. 197 oni e 175 (F, y; Coetzala, by the road to Coet- zapotitlan, К т, 25 Арг. 76, Vázquez 398 (F); | road to Coetzapotitlàn, 680 m, 25 Apr. eláz e 229 (F, MO). 5. Clerodendrum speciosissimum Моггеп, Hort. Belge 3: 322, pl. 68 (as “Cleroden- dron””). 1836. TYPE ILLUSTRATION: “Des Indes," cultivated by Baron Таћп at Van Geert Garden, Ghent, Belgium, pl. 68, ibid. Clerodendrum pulchrum Fawcett in H. О. Forbes, Nat. Wand. East. Arch. 514. 1885. TYPE: Malaysia, Forbes 3604 (holotype, BM not seen). Shrub to 3 m tall; branches and branchlets te- tragonal, usually medullose or hollow, puberulent, lenticellate; ad annulate, sometimes with a band of hairs; leaf scars slightly prominent. Leaves de- ovate, 8-25 cm cussate-opposite, blades broadly o -20 cm wide, base cordate, apex acute or long, short-acuminate, margin entire or irregularly cre- nate, above pubescent with whitish multicellular hairs, beneath with cinerous multicellular hairs, often with scattered disk-shaped glands along the midrib and secondaries beneath; midrib glabrous, secondaries 5-7 pairs, the two lower pairs issuing palmately from the leaf base, upper pairs arcuately Volume 80, Number 4 1993 Rueda Clerodendrum in Mesoamerica ascending; petioles 1.2-20 cm long, puberulent. nflorescences terminal, panicle 8-35 cm long, 6— 15 cm wide, composed of 5-13 pairs of cymes, puberulent; peduncles 3-6 cm long; pedicels 5-9 mm long, puberulent; bracts foliaceous, bracteoles and prophylls linear-setaceous, 1-5 mm ong, pu- berulent. Flowers with calyx campanulate, 5-9 mm long, granular, puberulent, deeply 5-lobed, lobes ovate, 2-4 mm long; corolla hypocrateriform, red or scarlet, outside glandular-puberulent, tube 2-3 cm long, limb 5-lobed, 2-3 cm wide, 1.3-1.8 ст long; stamens 4, filaments long-exserted, ca. 2.5 cm beyond the throat of the corolla tube, 2.6-3 cm long; ovary oblongoid, style 4.5-5.5 ст long, exserted, shorter or equal to the stamens. Fruit drupaceous, subtended by the cupuliform calyx, globose, 5-8 mm long and wide, bilobed, separating into 4 pyrenes at maturity; chromosome number: 2n — 48 (Bolkhovskikh et al., 1969). Distribution and habitat. Clerodendrum spectosissimum occurs from Indonesia to the Car- oline Islands and southeast to Tahiti and the Mar- quesas Islands. It is cultivated in most warm regions of the world. In Mesoamerica it has been collected from Mexico to Honduras. Vernacular names and uses. The main use of this species 15 as an ornamental. о There are three other species of Clero- dendrum cultivated in Mesoamerica that greatly resemble C. intermedium). According to Moldenke & Мој- denke (1983), the holotype of C. speciosissimum is a Morren collection taken from material given to him by Baron Taffin from the Van Geert Garden at Chent, Belgium. In the original description, “‘Au- gust 1835” is reported as the date that the plant flowered for the first time, but a specimen either speciosissimum (see note under C. was not made or is missing. The type specimen should be in BR (fide Moldenke & Moldenke, 1983). However, according to Paul Bamps, curator of this herbarium, there are no Morren collections in BR (pers. comm.). Fortunately, the colored illustration (Morren, 1836: pl. 68) accompanying the original description is good enough to establish the identity of the species and may be considered as the type. Specimens examined. MEXICO. CAMPECHE: Tixmu- cuy, road Holpechen-Tyxmucuy, 50 m, 11 Oct. 1980, Calzada et al. 6724 (MEXU). vuc bes precise locality and date unknown, Gaumer 909 (F, MO). GUATEMALA ESCUINTLA: El Baül Garden, 350 m, 3 Mar. 1921, Tonus 30 (F, US). EL SALVADOR. SAN SALVADOR: precise locality and date unknown, Calderón 2533 (F); precise locality and date unknown, Calderón 2544 (F, NY). HONDURAS. CORTES: San Pedro Sula, 80 m, 1951, Pérez 12 (F). 6. Clerodendrum paniculatum L., Mant. Pl. 1: 90. 1767. TYPE: India, collector unknown (holotype, Linn 810.5, microfiche, MO) МАО VG Wallich in Griffith, Not. Pl. Asiat. 4: 54. TYPE: India, Wallich 1803 кы Die 1803, microfiche, MO Shrub to 3 m tall; branches tetragonal, med- ullose or hollow, puberulent, lenticellate; nodes an- nulate, marked with a band of hairs; leaf scars slightly elevated. Leaves decussate-opposite; blades 3-lobed, ovate, 6-35 cm long, 6-30 cm wide, base deeply cordate and palmately veined, lobes apically acute to acuminate, central lobe larger and ovate, the others triangular, margin apiculate-denticulate to crenate-dentate or entire and with conspicuous glands mostly 3-10 mm apart, puberulent, often light-dotted above, densely squamose and punetate beneath; midrib puberulent, secondaries 5- 7 pairs, the 2-4 lower pairs issuing palmately from the leaf base; petioles 1-40 cm long. Inflorescences ter- minal and sometimes axillary, terminal panicles 15-40 cm long and 13-35 cm wide; peduncles 2—10 cm long; pedicels 5-20 mm long. Flowers with calyx campanulate, 3-6 mm long, granular or puberulent, 5-lobed, lobes 2-4 mm long; corolla hypocrateriform, orange-red to scarlet, outside pu- berulent, tube -lobed, ca. 1.2 cm wide, lobes oblong, 5-8 mm long; stamens 4, -2 cm long, limb 5 filaments long-exserted, ca. 2.2 cm beyond the throat of the corolla tube, 24-27 cm long; ovar oblong, style 3.5—4 cm long, exserted, shorter than the stamens. Fruit not seen. Distribution and habitat. Clerodendrum paniculatum is native to the area from northern India eastward to Taiwan, south to Indonesia, and including the Andaman and Nicobar islands. It is cultivated in all warm regions of the world, where it i5 often naturalized. I have seen collections from Honduras, Nicaragua, Costa Rica, and Panama both from ornamentals and from plants naturalized into wet tropical forest, along streams, and in open 500 m. Clerodendrum grassy marshes at altitudes mostly below Vernacular names and uses. paniculatum is used in Mesoamerica as an orna- mental, especially cultivated along fencerows (Mol- denke, y Note. There are three species of Cleroden- drum cultivated in Mesoamerica that are similar to C. paniculatum (see note under C. interme- dium). The holotype sheet (810.5) has the name Juan Bonge on it, but this may be a vernacular name. If this is the collector, it is an unknown one. Annals of the Missouri Botanical Garden HONDURAS. ATLANTIDA: Lan- 1, 3 km SW from Tela, 10 m, 19 176 (TEFH). NICARAGUA. RIO SAN cumens s examined. Apr 4, Martinez JUAN: San Juan del Norte, E side, 5 m, 2 Dec. 1982, a 337 1 (MO). Costa RICA. HEREDIA: outskirts of Puerto Viejo or in the town itself, 13 July 1986, Wilbur 39925 ; (DUKE). PANAMA. COLON: Barro Colorado Island, Canal Zone, in clearing at laboratory, 3 Dec. 1967, Croat 4065 (MO) 7. Clerodendrum wallichii Merr., J. Arnold Arbor. 33: 220 (as “Clerodendron”). 1952. Clerodendrum nutans Wallich, Numer. List. 49, no 9. TYPE: Bangladesh, Wal- lich 1793 (holotype, K; holotype and isotype on microfiche, MO), non W. Jack, 1820. Shrub to 3 m tall; stems or branches usually simple and arching, medullose, tetragonal, puber- ulent; nodes annulate with a band of hairs; leaf scars slightly elevated. Leaves decussate-opposite or ternate; blades oblong-lanceolate, 4—13 cm long, 1-3 cm wide, basally acute or subcuneate, apex long-acuminate or caudate, margin entire or un- dulate, puberulent on both surfaces, punctate be- neath; midrib glabrous, the secondaries 5-7 pairs; petioles 0.3-1 cm long, often slightly alate, pu- berulent. Inflorescences terminal, paniculate, pu- berulent, 9-30 cm long, 4-10 cm wide, with 6— 10 pairs of divaricate cymes, each cyme 1-5- flowered; peduncles 1-4 cm long; pedicels 4-10 mm long; bracts foliaceous, subtending each pair of cymes. Flowers with calyx campanulate, slightly pentagonal, 0.5-1.2 cm long, puberulent, 5-lobed, lobes triangular-acuminate, 0.3-0.8 mm long; co- rolla hypocrateriform, white to greenish white, pu- berulent, tube ca. 1.5 cm long, limb 5-lobed, ca. 1.5 cm wide, ея ovate, 0.9-1.3 ст long; sta- mens 4, filaments long-exserted, ca. 2.5 cm beyond the throat of the corolla tube, 2.3— ovary oblong, style 3-3.5 cm long, exserted, short- er than the stamens, stigma bifid, capitate. Fruit 2.8 cm long; not seen. Distribution and habitat. Clerodendrum wallichii is native to southern Asia, from the Hima- layas to southern China, south to Indochina, the Nicobar Islands, and northeast to Pakistan. It is cultivated as an ornamental in all mius e and - elcewhe subtropical countries and in g In Mesoamerica it has been ocu in ies El Salvador, and Costa Rica. Vernacular names and uses. In Mesoamerica this species is only “Martinica” and "misteriosa." used as an ornamental. Note. In Mesoamerica there аге no other spe- cies similar to Clerodendrum wallichii. In addition, it is the only species of Clerodendrum known in Mesoamerica with a capitate-bifurcate stigma. The name Clerodendrum nutans was changed by Mer- rill in 1952 because this name had been used by Jack in 1820. Specimens examined. GUATEMALA. ESCUINTLA: at S 1939, Standley 64521 (F). EL S vicinity of San Salvador, 650-850 m, 30 Mar.- 1922, Standley 23645 (US). Costa RICA. SAN JOSE: vi- cinity of 1,130 m, 4 Dec.-10 Feb. 1926, San José, Standley 52498 (US) 8. Clerodendrum glabrum E. Meyer, Comm. г. Austr. 273 (as “Clerodendron”). 1838. TYPE: ep и Drége s.n. (holo- type, G-DC 1906 : 661.2, microfiche, MO; isotype, MO) Ehretia triphylla Hochst., Flora 27: 830. 1844. TYPE: South Africa, Krauss 100 (isotype, MO Cle А мем Klotzsch in W. Peters, Naturw. Reise sambique 1: 257. 1861. TYPE: East Africa, col- hear a ^n s.n. (holotype, B not seen). Shrub to 3 m tall (sometimes a tree to 12 m tall in the Old World); branchlets solid, subterete or tetragonal, glabrous, lenticellate; leaf scars el- evated. Leaves decussate-opposite or ternate with an unpleasant odor when crushed; blades elliptic- ovate to oblong, 3-8 cm long, 3-5 cm wide, base acute or short cuneate, apex acute or obtuse, mar- gin entire, undulate, glabrous on both surfaces, densely punctate beneath; midrib puberulent, sec- ondaries ca. 6 pairs, all ascending; petioles 1-2.5 cm long, lenticellate, glabrous. Inflorescences ax- illary in the uppermost axils and terminal, cymes opposite or whorled, many-flowered, puberulent, 7.5 cm long and 6 cm wide, several times bifurcate, terminal panicles corymbose; peduncles 1-3 cm long; pedicels 2-10 mm long, puberulent; bracts foliaceous, bracteoles and prophylls linear, 2-4 mm ong. Flowers fragrant; calyx campanulate, 3-5 mm long, puberulent and glandular-lepidote, 5-lobed, lobes 1-3 mm long, shorter than the tube; corolla hypocrateriform, white, outside puberulent and glandular, tube ca. 9 mm long, limb 5-lobed, ca. 1 em wide, lobes obovate, 3-4 mm long; sta- mens 4, filaments long-exserted, ca. 17 mm beyond the throat of the corolla tube, 2.5-2.7 cm long; ovary oblongoid, style 3-3.2 cm long, exserted, equal to or longer than the stamens. Fruit dru- paceous, subtending calyx cupuliform, 5-9 mm long and wide, sulcate; seeds 1-2, oblong. Distribution and habitat. Clerodendrum glabrum is native from Kenya to Angola and Volume 80, Number 4 throughout South Africa, and the Comoro and Sey- chelles islands. In Mesoamerica it has only been collected in Guatemala, which is the first report in Mesoamerica. Vernacular names and uses. “Palo de peri- " [t is presumably an ornamental in Guatemala. Eton glabrum known from Mesoamerica seems to have The sole collection of Clerodendrum been collected in an abandoned garden, since its accompanying label does not indicate that it was taken from a cultivated plant. The only similar species to C. glabrum in Mesoamerica is C. li- gustrinum, from which it may be distinguished by the length of the calyx and corolla. The calyx and corolla of C. ligustrinum are twice as long as those of C. glabrum. Specimen examined. GUATEMALA. GUATEMALA: San Juan Sacatepéquez, 1,480 m, Morales 1874 (F, NY). 9. Clerodendrum thomsonae Balf. f., Edin- burgh New Philos. J., ser. 2, 15: 233. pus (as “Clerodendron”). 1862. TYPE: from plant cultivated at the Edinburgh Botanical Garden C. Thomson from Old Calabar, southern Nigeria, collector unknown s.n. (lec- sent by totype, here designated, Shrub 2- 2.5 m tall; branchlets tetragonal, med- ullose, puberulent, lenticellate; leaf scars often cir- cular, corky, elevated. Leaves decussate-opposite; blades elliptic or elliptic-ovate, 4-13 cm long, 2- 6 cm wide, base rounded or subacute, apex acu- minate or short-acuminate, margin entire, ciliate, puberulent on both surfaces; midrib puberulent, secondaries 5-7 pairs, the lowest pair issuing pal- mately from the leaf base; petioles 0. 5-3 cm long, puberulent. Inflorescences axillary, cymose, berulent, 5-13 ст long, 5-15 ст wide, individual cymes 5-9 ст long, 4-8.5 cm wide, often tri- chotomous first and then dichotomous; peduncles 2-6 cm long, puberulent; bracteoles and prophylls linear, 2-11 mm long, puberulent. Flowers with calyx globose, coarse, + pentagonal, 1.5-2 cm long, puberulent, 5-lobed, lobes ovate, 1-1.8 ст long, acuminate; corolla hypocrateriform, dark red to scarlet, externally glandular puberulent, inter- nally puberulent, tube ca. 2.5 cm long, limb 5-lobed, ca. 1 cm wide, lobes oblong-elliptic, 5-9 mm long, obtuse or acute; stamens 4, filaments long-exserted, ca. 1.5 cm beyond the throat of the corolla tube, 2-2.2 cm long; ovary oblong, style filiform 2.9- 3.2 cm long, exserted, shorter or equal to the stamens. Fruit drupaceous, covered by the calyx, 2.3 cm long, round or depressed-globose, 10-1 mm long and wide, glossy black, 2-bilobed; seeds Rueda 883 Clerodendrum in Mesoamerica oblong; chromosome number: 2n — 42, 46, 48, or 50 (Bolkhovskikh et al., 1969). Distribution and habitat. Clerodendrum thomsonae is a western African species that ranges from Senegal to Zaire. It has been collected throughout Mesoamerica. | “Arete de in- 99 66 Vernacular names and uses. corazón san- 39 сс dia," “arroz y frijoles," “bandera, > “fucsia "i аа de amor, ay on: ““moros y cristianos, > “pasión and “rice and beans." It is used in Mesoamerica exclusively as an ornamen Note. sonae is not known to escape from cultivation. This de cristo,” In Mesoamerica Clerodendrum thom- species is very similar to C. umbellatum. However, the calyx in C. thomsonae is twice as large as in C. umbellatum. In addition, the calyx of C. thom- sonae is coarser in texture than is that of C. um- bellatum. 'The specimen in E, number 32/89 (1) was collected in January 1862, and the number 32/89 (2) was collected in July 1863. However, it is not known which is the holotype. Therefore, it is necessary to select a lectotype. The specimen 32/89 (1) should be regarded as a lectotype and 32/89 (2) as an isolectotype. lected specimens examined. MEXICO. CHIAPAS: Escuintla, 2 May 1 948, Мей 17699 (F, MEXU). GUATEMALA. IZABAL: vicinity of Quiri 31 May 1923, Standley 24316 (GH, US). EL SALVADOR. SAN GA VADO: vicinity of San Salvador, 650- 850 m, 30 COMAYAGUA: vicinity of Comayagua, 600 m, 12-23 Mar Bluefield, 29 Aug. 1948, ЋЕ 208 (Е). Cosr ALAJUELA: Ojo de Agua, 22 June 1946, Echev "d 452 (CR). PANAMA. COLON: Canal Zone, 20-80 m, 1911, Pittier 2739 (US). 10. Clerodendrum umbellatum Poiret in Lam., Encycl. 5: 166. 1804. TYPE: West Af- rica, Smeathman s.n. (holotype, P-LA 6207. 510: I. 6, microfiche, MO; isotype, K). Clerode sieve hirsutum G. Don, Edinburgh New Philos. J. 49. 1824. TYPE: West Africa, Don s.n. (ho- Vin BM not seen). Shrub 0.2-3 m tall; branches and branchlets tetragonal, medullose or hollow, puberulent, some- times lenticellate; leaf scars often elevated. Leaves decussate-opposite; blades ovate, elliptic or oblong, 4-13 cm long, 2-9 cm wide, base obtuse or sub- cordate, apex short-acuminate or acute, margin entire, impressed punctate beneath; midrib puber- ulent, secondaries 5-7 pairs, lower 1-2 pairs is- suing palmately from the base; petioles 1-3 cm 884 Annals of the Missouri Botanical Garden long, puberulent. Inflorescences axillary, cymose, in a terminal panicle, 10-25 cm long, 12-24 cm wide, individual cymes several times branched, flowers 5-11 cm long, 3-12 cm wide, terminal panicles subumbellate; peduncles 2.5-6 cm long, puberulent; pedicels 5-15 mm long; bracts nar- rowly elliptic. Flowers fragrant; calyx campanulate, 0.7-1.3 cm long, puberulent, 5-lobed, lobes ovate or lanceolate, 0.5-1 cm long; corolla hypocrater- iform, dark red to pink, with a pinkish or reddish throat, outside puberulent, tube ca. 2.2 cm long and ca. 1 ст wide, lobes obovate, 7-9 mm long; stamens 4, filaments long-exserted, ca. 2 cm be- yond the throat of the corolla tube, 2.3-3 cm long; di ен s 3.8-4.2 cm long, exserted, ual to t mens; chromosome number: 2n — 24 or 48 [dein 1981). Distribution and habitat. Clerodendrum um- bellatum is ап African species that occurs from Senegal and Mali southeast to Kenya and Tanzania. In Mesoamerica this species has been collected in Mexico, Honduras, Nicaragua, and Panama; it may This is the first report of C. umbellatum in Mesoamerica. Clerodendrum umbellatum is used as an ornamental, especially also be naturalized in the area. Vernacular names and uses. in fencerows The label data indicate that this species sometimes escapes in Mesoamerica. An example is Ramos 2 found growing in pine plantations in La Paz, Honduras. MEXICO. CHIAPAS: ‚ЗО т, 16 Jan. 1972 Selected e buius examin d Huixtla, along the stree Breedlove 23780 1979, Cowan a AS. ATLANTIDA: la Lan cetilla ex- m, 26 June 1985, Ramos 2 (TEFH). NICARAGUA. MANAGUA: Peninsula de Chiltepe, Comarca Alfonso González, 80-100 m, 3 Jul 1980, Guzmán et al. 252 (MO). nivas: Isla de Ometepe, Balgue, 40-100 m, 18 Jan. 1987, Moreno 19659 (MO). PANAMA. LOS SANTOS: Las Tablas, La Baja, 20 June 1971, Carrasquilla 154 (F). 11. Clerodendrum costaricense Standley, Field Mus. Publ. Bot. 18: 1002 (as “Clero- dendron"). 1938. TYPE: Costa Rica, Smith H588 (holotype, F; isotype, MO; fragment and photo, NY). Shrub to 5 m tall; branchlets and twigs med- ullose, subterete, lenticellate; leaf scars elevated. Leaves decussate-opposite; blades oblong-elliptic, 11-14 cm long, 6-9 cm wide, base obtuse, apex short-acuminate, undulate and thickened, punctate beneath; midrib prominent, secondaries ca. 6 pairs; petioles 1-2 cm long, sparsely lenticellate. 3-flowered, glabrous; peduncles 3-6 cm long; ped- margin entire, Inflorescences axillary, icels 0.5-1 cm long, glabrate, bracts foliaceous, oblong-lanceolate, bracteoles linear-setaceous, 2- 3 mm long, puberulent. Flowers with calyx ellip- soid, 10—12 mm long, glabrous, shortly lobed, lobes ovate, apically caudate-apiculate; corolla hypocra- teriform, pale greenish yellow at anthesis, fading to brown, glabrous, the tube cylindric, ca. 1.2 cm long, the limb ca. 2 cm wide, the lobes spreading, oblong, as long as the tube, apically obtuse. Other floral parts not examined. Fruit unknown. Distribution and habitat. Clerodendrum cos- taricense has only been collected in Alajuela, Costa Rica (Fig. 2). It was found in the Caribbean cloud forest climbing over vegetation at the edges of woodland, at 1,650 m Vernacular names and uses. Note. very poor, especially in reproductive structure. The None reported. The available material of this species is specimens have only buds or incomplete flowers, and the description is accordingly incomplete. Specimen И Costa RICA. ALAJUELA: La Pena del Zarcero, 1,650 m, 4 Apr. 1938, Smith H588 (F, MO, NY) 12. Clerodendrum ligustrinum (Jacq.) R. Br., in W. T. Aiton, Hortus Kew. ed. 2, 4: 64 1812. Volkameria ligustrina Jacq., Coll. Suppl. 118, pl. 5, fig. 797. TYPE: material cultivated in Vienna, Austria, from “Mauritius island," plate 5, fig. 1, ibid. a o culinare Sessé & Mociño, Fl. Mexic., ed. 2: 151. 1894. TYPE: Mexico, collector Heri holotype, MA not seen). Clerodendrum t frtunatum Sessé & Мосто, Fl. Mex 15 4. TYPE: Mexico, eer un- Fabel s.n. (holotype, MA not seen). е inermis Sessé & Мосіпо, Fl. Mexic. ed. 52. 4. TYPE: Mexico, collector preis ee olotype, MA not seen). Clerodendrum mexicanum КА Univ. Calif. Publ. Bot 39. 1909. E: Mexico, Purpus 3336 кы. F, MO, NY. "Us S). Shrub to 5 m tall; branchlets subterete or te- tragonal, medullose, puberulent, lenticellate; nodes annulate, often with band of hairs; leaf scars ele- vated. Leaves decussate-opposite, blades elliptic to lanceolate, 1.5-9 cm long, 0.6-5 cm wide, acute or cuneate, apex subacuminate, margin en- base tire, glabrous and densely punctate beneath; midrib prominent beneath, secondaries 5—7 pairs; petioles Volume 80, Number 4 Rueda 885 Clerodendrum in Mesoamerica FIGURE 5. Clerodendrum ligustrinum (Jacq.) R. Br. Pistil. — E. Fruit. 9-10 mm long, puberulent. Inflorescences supra- axillary or rarely terminal, cymose, the cymes sol- itary, opposite, 3-7.5 cm long, 2-7 cm wide, usu- ally only 3- or 4-flowered, often twice dichotomous and then 7-flowered, the terminal ones, when pres- (Palmer 407).— A. Habit. — B. Flower. — С. Stamens. — D. ent, similar but usually smaller; peduncles 1.5-4 cm long; pedicels 3-10 mm long, puberulent; bracts foliaceous, caducous, bracteoles and prophylls lin- ear, 1-6 mm. Flowers with calyx campanulate, 5— 8 mm long, outside puberulent, 5-lobed, lobes lan- 886 Annals of the Missouri Botanical Garden ceolate or deltoid, 3-4 mm long; corolla hypocra- teriform, white, outside resinous-punctate, inter- nally slightly pubescent, tube slender, 1—2 cm long, mb 5-lobed, 1–1.3 ст wide, lobes oblong, 3-5 mm long; stamens 4, filaments long-exserted, ca. 2 cm beyond the throat of the corolla tube, 21- 23 cm long; ovary oblong, style 2.8-3.1 cm long, equal to the stamens. Fruit drupaceous, subtending calyx cupuliform, subspheric, 8-12 mm long and wide, bilobed, splitting into two-seeded pyrenes at maturity; seeds obovate (Fig. 5). Distribution and habitat. Clerodendrum li- gustrinum occurs from northern Mexico to Nic- aragua (Fig. 1). It is also cultivated as an orna- mental in Mesoamerica and other parts of the New World, as well as in the Old World. This species has been reported from roadsides, creek and river banks, marshes and swamps, beach ridge forest, sandy bushy areas, secondary forests, and ever- green or subevergreen forests. It occurs from near sea level to 500 m Vernacular names and uses. 99 é "Arbol sagra- palo blanco cimarrón," and “snake-tree.” It is used as an ornamental in Honduras; Sessé & Mociño (1894) and Gibson (1970) reported that the leaves of this species are used in Mexico and Guatemala in native cooking to flavor fish; Del Amo (1979) reported that a leaf infusion is applied to snake-bites; Mendieta & Del Amo (1981) reported this plant as a diuretic, even in the case of kidney stones. Note. drum in Mesoamerica similar to C. ligustrinum: There are three species of Cleroden- Clerodendrum aculeatum, C. molle, and C. pit- tieri. However, the calyx of C. ligustrinum is not pubescent as is the calyx of C. molle, nor is it truncate as is the calyx of C. pittieri. In addition, the leaf scars in C. ligustrinum are not as prom- inent as in C. pittieri and C. molle, nor spine- tipped as in C. aculeatum. This species was orig- inally described (Jacquin, 1796), in error, as from Mauritius, based on material cultivated in Vienna. The plant may already have been cultivated on Mauritius and erroneously considered to be native by Jacquin. Another possibility is that the material originally came from Mesoamerica, and Jacquin mistakenly believed that it was from Mauritius. According to Harald Riedl, there is not herbarium voucher in W; Jacquin usually did not make her- barium vouchers of cultivated plants. Therefore, the name must be typified by reference to Jacquin's late 5, figure 1. Fortunately, this illustration in Jacquin (1796) is adequate to establish the identity of this species. Selected specimens examined MEXICO. CAMPECHE: 20 km E of Champotón, 27 Mar. 19 (CAS). VERACRUZ: vicinity of Pueblo V Tampico, 23-31 Ma BELIZE. 5 mi. from Belize city, past y ipium airport, 22 Mar. 1973, Dwyer 10351 (MO). GUATEMALA. EL PETEN: La Libertad and vicinity, 19 Mar. 1934, аи 353 (Е, МІСН). HONDURAS. CHOLUTECA: vicinity of Choluteca, 20 m, 31 Oct.-9 Nov. 1949, Standley pert US). NICARAGUA. LEON: La Paz Centro, Km 59, new highway to León, 50 m, 13 June 1984, Moreno 24218 (MO) 12a. Clerodendrum ligustrinum var. nica- raguense Mold., Phytologia 1: 416. 1940. TYPE: Nicaragua, Chávez 227 (holotype, US). This variety differs from the typical form of the species by having leaves and axillary cymes usually ternate and the leaf blades puberulent or pubescent beneath. Distribution and habitat. It occurs from Honduras to fields, and bordering mangrove swamps, mostly below 900 m Vernacular names and uses. anama in dense wet forest, weedy “Jasmin,” and .. si me miras te enamora." [t is cultivated as ап e in Mesoamerica. specimen from Chiapas, Mexico Be M 20913: C. ligustrinum), was referred to this variety by Moldenke (1987) but is incor- rectly attributed to this taxon. Selected digo examined. | HONDURAS. CORTES: vicinity of La Li 11-20 Apr. 1947, Standley & Chacón 7249 (F ). NICARAGUA. CHINANDEGA: vicinity of Chichigalpa, 90 m, 12-18 July 1947, Standley 11488 (F, NY). ZELAYA: region of Braggman's Bluff, without date, Englesing 120 (NY). Costa RICA. LIMON: near Lamm Hotel, near sea level, 19 Oct. iid e 6048 (CR). PANAMA. BOCAS DEL TORO: Aln t along the road to the Bomba, 15 Oct. 1965, Blum 1350 (DUKE, MO). COLON: along ocean trail between Rio Indio and rda, at sea level, 7 July 1976, Croat 36888 (MO). 13. Clerodendrum molle Kunth, Nov. Gen. Sp. 2: 244. 1817. TYPE: Ecuador, Bonpland 3837 (holotype and isotype P-HB. 6209. 46: ПІ. 1, microfiche, MO). Shrub or small tree to 5 m tall; branchlets sub- terete or tetragonal, medullose, densely short pu- bescent, lenticellate; leaf scars prominent. Leaves decussate-opposite or ternate, blades elliptic, 1.5— 5.9 em long, 1-2.5 acute, puberulent above, short-tomentose beneath, cm wide, base acute, apex punctate beneath; midrib usually more densely pu- berulent than the blade above, secondaries 3-6 pairs; petioles 3-8 mm long, pubescent, the lowest Volume 80, Number 4 1993 Rueda 887 Clerodendrum in Mesoamerica 1-2 mm long, persisting as a corky protuberance after the blade has fallen. Inflorescences axillary or terminal, cymes ternate, solitary in each axil, 3-7 cm long, 1-4 cm wide, 3-9-flowered, pubes- cent; peduncles 1-3.7 cm long, pubescent; pedicels 1-7 mm prophylls linear, ca. | mm Flowers with calyx campanulate, 4-7 mm pubescent, 5-lobed, lobes ovate, 2-3 mm long; co- rolla hypocrateriform, white or pinkish, outside pu- long, densely pubescent; bracteoles and long, densely pubescent. long, bescent and resinous-punctate, internally slightly pubescent, tube 1.8-2.5 cm long, limb 5-lobed, 1-1.3 cm wide, lobes oblong, 2-3 mm long; sta- mens 4, filaments long-exserted, ca. 1.5 mm ђе- yond the throat of the corolla tube, 15-20 mm long, anthers oblong; ovary oblongoid; style 3-3.3 cm long, equal to or longer than the stamens. Fruit not seen. Distribution. Clerodendrum molle is native from Panama to Colombia, coastal Ecuador, and the Galápagos Islands. In Mesoamerica it is only known from Panama (Fig. 2). It is found mostly on the seashore and at lower elevations, especially on rocky coastline precipices, lava flows, and in open woodland on hillsides, from sea level to less than 500 m. Vernacular names and uses. None reported in Mesoamerica. The Mesoamerican specimen was col- lected in the late 1700s or earliest 1800s. The specimen label indicates that the specimen was collected in Panama, and that is possible since the species occurs in coastal Ecuador; however, it is also possible that a label mixup occurred and it was не collected in Ecuador, апа not in Рап- . There are three Mesoamerican species of C аи that resemble C. molle in habit and habitat (see note under C. ligustrinum). Specimen examined. PANAMA. PANAMA: Nee s.n. (MA). 14. Clerodendrum aculeatum (L.) Schldl., Linnaea 6: 750. 1831. Volkameria aculeata L., Sp. Pl. ed. 1, 1: 637. 1753. TYPE: Jamaica, Browne s.n. (holotype, Linn-809.1, micro- fiche, MO) ен aa var. grandifolium Kuntze, Re- . Pl. 2: 505. 1891. TYPE: St. Thomas, col- c a dud s.n. (holotype, NY not seen). Shrub to 2.5 m tall; branches, branchlets, and twigs subterete or tetragonal, medullose, appressed- pubescent, lenticellate, spinose; spines 2-7 mm long, ternate or opposite, borne on and subtending the nodes. Leaves decussate-opposite, ternate, or apically clustered, on the upper side of a spine; blade elliptic or lanceolate, 1-4 cm long, 0.3-1 cm wide (sometimes larger and wider in the West Indies), base acute or cuneate, usually mucronu- late, glabrous on both surfaces, punctate beneath due to deeply sunken resinous peltate glands; mid- rib prominulous beneath, secondaries 3-7 pairs; petioles of the primary leaves with a very stout, ligneous spinescent base, 3-11 mm long, eventu- ally breaking off obliquely Уз to 12 the distance from apex to base, the base persisting as a sharp, subulate-tipped spine, petioles of the small axillary leaves obsolete or very short and not spinescent. Inflorescences axillary, pedunculate, crowded at the apex or upper nodes of the branchlets and twigs, cymes solitary, 2-6 cm long, 1.5-3.5 cm wide, 3—7-flowered; peduncles 1-3.5 cm long, pu- berulent; pedicels 2-10 mm long, puberulent; bracts linear. Flowers cymose, slightly zygomorphic; calyx tubular or campanulate, 3-7 mm long, outside, usually pubescent, 5-lobed, lobes ovate, 2-3 mm long; corolla hypocrateriform or infundibular, white, inside slightly pubes- cm long, limb 5-lobed, outside resinous-punctate, cent, tube slender, 1.3-2.3 1–1.5 ст wide, lobes oblong, 5-8 mm long, shorter than the tube; stamens 4, filaments much exserted, 2-3 cm long; ovary minute, style 3-3.2 cm long, much exserted and equal to the stamens. Fruit drupaceous, subtending calyx cupuliform, 5-10 mm long an 2-seeded pyrenes; seeds oblong; chromosome num- ber: n = 12 (Cave, 1959, 1964) wide, in age 4-lobed, splitting in Distribution and habitat. Clerodendrum aculeatum occurs in the West Indies from the Bahamas south to Barbados. It is planted and some- times naturalized from Veracruz (Mexico) south to northern South America, and in the Old World (Moldenke, 1980). In Mesoamerica, it is known only from the Swan Islands, off Honduras, which is the first report for the species in Mesoamerica. It is often locally abundant and occurs in open grassland, dry scrub, limestone, swamplands, and beachside areas, mostly from sea level to 500 m. Vernacular names and uses. None are re- ported in Mesoamerica. Clerodendrum aculeatum is used in living fencerows and medicinally in cases of gonorrhea; a decoction of the leaves is used as a cough remedy (Reis & Lipp, 1982) Note. of Asia, C. ligustrinum and C. pittieri from Me- soamerica, and C. molle and C. ternifolium of South America in habit and habitat. However, C. aculeatum differs from these species in the very This species closely resembles C. inerme 888 Annals of the Missouri Botanical Garden sharp-tipped spine at the nodes. According to Fred Barrie, the. Linnaeus specimen number 809.1 is the same specimen named by Browne (1750) “Clerodendrum fruticosum, spinosum; foliis in- ferioribus confertis, superioribus oppositis; pe- Appar- innaeus never saw actual material of this dunculis tripartitis, trifloris, alaribus." ently, species until after he had published his binomial based on Browne's material mentioned above. Therefore, specimen 809.1, although not sent by Browne to Linnaeus until 1756, is presumably the basis for Browne's polynomial and for Linnaeus's epithet. Specimen 809.2, also identified as Vol- kameria aculeata, lacks spines and is probably C. inerme rather than C. aculeatum. Specimens examined. | HONDURAS. SWAN ISLANDS: Great Swan Island, surroundings of the military base, 17 May 1980, Cruz & Espinal 263 (ТЕЕН); without precise locality, 15-24 Aug. 1971, Proctor 32509 (IJ, TEX); tropical wet forest, at sea level, 10 June 1986, Nelson et al. 9770 (TEFH). 15. oe pittieri Mold., Phytologia . TYPE: Panama, Pittier 4 пен del isotype, F, Shrub to 3 m tall; branches and branchlets te- tragonal, slightly flat at the nodes, medullose, with appressed hairs, lenticellate; leaf scars spinescent but not sharp-tipped. Leaves decussate-opposite, ternate, or sometimes clustered on reduced axillary twigs; blade lanceolate to elliptic, 0.7-4 cm long, 0.3-2.5 cm wide, base acute or cuneate, apex acute or obtuse, densely punctate beneath; midrib prominent beneath, secondaries 3-7 pairs; petioles with a ligneous subspinescent base, 1-8 mm long, eventually breaking off obliquely 1—2 mm from the base, the petioles of the axillary clustered leaves obsolete or very short and not spinescent. Inflo- rescences axillary, at the upper nodes of branchlets and twigs, cymes solitary at the axils of the spines, 3-5 cm long, 1.5-3.5 cm wide, 1-3-flowered; peduncles 0.7-2 cm long, with appressed hairs; pedicels 0.5-1.2 cm long, subglabrate; prophylls linear-subulate, 1-2 mm long. Flowers with calyx campanulate, 2-4 mm long, puberulent, truncate, 5-dentate; corolla hypocrateriform, white or tinged pink-purple, puberulent, internally slightly pubes- cent, tube 1.5-2.2 cm long, limb 5-lobed, lobes oblong or elliptic, 4-8 mm long; stamens 4, fila- ments much exserted, 2.3-2.5 cm long; ovary oblong, style 3.3-3.6 cm long, much exserted, equal to or shorter than the stamens. Fruit dru- paceous, subtending calyx cupuliform, obovate, 5— 13 mm long and wide, bilobed, splitting into two 2-seeded pyrenes at maturity; seeds oblong. Distribution and habitat. Clerodendrum pit- tieri occurs from Guatemala to Ecuador and Ven- ezuela. All the es collections are from the Pacific Coast (Fig he occurrence of C. pittieri on the hate ж is restricted to Co- lombia and Venezuela. This species is usually found in mangrove areas, dry forested hills above man- groves, salt flats, around tidal pools, and on riv- erbanks near saltwater. It is common in Nicaragua, in the mangrove forest of Poneloya (León) and Puerto Morazán (Chinandega) (pers. obs.). Vernacular names and uses. “Espino.” No uses reported. Note. Clerodendrum pittieri closely resem- bles three other Mesoamerican species of Clero- dendrum (see note under C. ligustrinum). The name was used for the first time by Standley in 1938, but was validly published by Moldenke in 1940. Selected specimens examined. GUATEMALA. ESCUINTLA: near San José, at sea level, 30-31 Jan. 1939, Standley 63974 (F). SAN MARCOS: Ocós, 1-2 m, 15 Mar. 1940, Steyermark 37854 (F). CosrA RICA. GUANACASTE: Puerto Jesús on fF “a of Nicoya ca. 13 km E of La , Wilbur 31775 (DUKE). PUNTA- RENAS: Pacific и co between Los Loros and Tivines, 1938, Brenes 22671 (F, NY). PANAMA. Pittier a (F, US). cocLE: Aguadulce, near sea level, 3-6 Dec. 1911, Pittier 4965 (F, MO, US). EXCLUDED TAXA Clerodendrum epiphyticum Standley, Field Mus. 168. 1940 (as “Cleroden- dron”) = Gibsoniothamnus epiphyticus (Standley) L. O. Williams (Bignoniaceae), Field Mus. Pub. Bot. 34: 120. 1972. Clerodendrum mutadae Standley, Field Mus. Pub. Bot. 17: 206. 1937 (as “Clerodendron”) = Aegiphila costaricensis Mold., Fieldiana, Bot. 24: En 1970. Cl ldenk St , Field Mus. Pub. Bot. 22: 99. 1940 (as “C Mid nie — Gibsoniothamnus moldenkeanus (Stand- ley) L. O. ge SERRE Fieldiana, Bot. 32: . 1970 и = о. Standley & Steyerm., Field Mus. Pub. Bot. 22: 373. 1940 (as “Cle- rodendron") = Gibsoniothamnus cornutus (J. D. Smith) A. Gentry и Field- iana, Bot. 34: 55. 197 Clerodendrum standleyi eis Geogr. Dis- trib. Members Verbenac. Avicenniac., 76. 1942 = Trybliocalyx pyramidatus Lindau (Acanthaceae), Bull. Herb. Boissier, ser 2, 4: 328. 1904 Volume 80, Number 4 1993 Rueda Clerodendrum in Mesoamerica 889 Clerodendrum mimicum Standley & Steyerm., Field Mus. Pub. Bot. 23: 227. 1944 (as “ Cler- odendron’’) Gibsoniothamnus mimicus (Standley & Steyerm.) L. O. Williams (Big- noniaceae), Fieldiana, Bot. 32: 214. 1970. LITERATURE CITED ВЕССАНІ, О. Piante ospitatrici ossia piante for- micarie della Malesiae della Papuasia. Malesia 2: 47- 51 BOLKHOVSKIKH, 2. ET AL. 1969. Chromosome Number of Flowering Plants. V. L. Komarov Bot. Inst., Acad. emy of Sciences of the U.S.S.R. Browne, P. . The Civil and Natural History of Jamaica. Published by the Author, London BURMAN, J. 37. Thesaurus Zeylanicus. РИИ РЯ Waesberg and Salomon Schouten CANDOLLE, А. L. P. ре. 1880. La СРЕ Libraire de L'Académie de Medecine, Paris. Cave, M. S. (editor). 1959. Index to Plant Chromo- some Numbers. 1: 52. Univ. North Carolina Press, Chapel Hill. (editor). 1964. Index to Plant Chromosome Numbers. 2: 136. Univ. North Carolina Press, Chap- el Hill. Corner, E. J. H. 1952. Wayside Trees of Malaya. 2nd ed. 1: 699-701. V. C. C. Gatrell, Singapore. CRoNQUIST, А. 1981. An Integrated System of Classi- fication of Flowering Plants. Columbia Univ. Press, 'w York. DARLINGTON, C. D. & E. K. JANAKI. 1945. Chromosome Atlas of Cultivated Plants. George Allen & Unwin, London. & A. . WYLIE. 1955. Chromosome Atlas of Flowering Plants George Allen & Unwin, London. DEL AMo, S. 9. Plantas Medicinales del Estado de Veracruz. uem Nacional de Investigaciones sobre Recursos Bióticos. Xalapa, Veracruz. Dreisic, Н. Foraging rate of ants collecting hon- eydew or estralloral nectar, and some possible con- straints. Ecol. Entomol. 13: 143-154. Duke, J. A. & E. S. AYENSU. Medicinal Plants of China. 2: RN Reference Publications, Al- onac, Michiga Eras, T. S. 1983. Богава кае their structure and distribution. Pp. 1 in B. Bentley S. Elias (editors), The dtes a the Nectaries. б». mbia Univ u ress, GIBBS, R. D. 1974. С hemotisonomy of Flowering Plants. 3: 1752-1755. McGill-Queen’s Univ. Press, Lon- 970. Verbenaceae of Guatemala. /n: : y & L. O. Williams (editors), Flora of uie la P os Bot. 24: | 92-195. GOLDBLATT, P. (editor). 1981. Index to Plant Chro- mosome Numbers er _ Monogr. Syst. Bot. Missouri Bot. Gard. 19 ы to den Chromosome Numbers 1979- „1981. Monogr. Syst. Bot. Missouri Bot. Gard. don. GIBSON, D. B : "dean: Index to Plant Chromosome Numbers ee 1985. Monogr. Syst. Bot. Missouri Bot. Gard. 23: 214. E R. A. & D. A. POWELL. 1968. Clerodendrum | Pme Clerodendrum fragrans. Tax- Hsu, yn М 1968; Preliminary chromosome studies on the vascular plants of Taiwan. Taiwania 14: JACK, W. 1820. ee of Malayan Plants. Ma- layan Misc. 1: JACQUIN, | J. Dor MM supplementum. Wappleriana, Vien JIMÉNEZ, T A. & R. 5ш. 1985. Patrones regionales en la estructura y composición floristica de los man- glares de la Costa Pacifica de Costa Rica. Rev. Biol. Trop. 33: 25-37. KuMan, V. & B. SuBRAMANIAN. 1987. Chromosome Atlas of Flowering Plants of the Indian Subcontinent 1: 410-411. Ministry of Environment and Forest, Government of India LINNAEUS, C. 1753. Species Plantarum. 2: 637. Lau- rentii Salvii, Stockholm MEHRA, N. & A 1969. Chromosomal mohoni in tropical hardwoods. Evolution 23: 4606- 481. с R. M. & S. DEL Амо. 1981. Plantas Medici- зајев del Estado de Yucatan. Instituto Nacional E Investigaciones Sobre Recursos Bióticos, Xalapa, Мей с Н. М. 1973. Verbenaceae of Panama. e R. E. Woodson AR W. Sch Moldenke, Plainfield, New Notes on n genus Clerodendrum ы е Phytologia 57: 303-310, 334-365, 386-404; 58: 329-359 Notes on the genus Clerodendrum (Verbenaceae). m үз 456-497. — ——— & A. L. is e Verben of Ceylon. /n: : зау (editor), 2 3: 407- ae Sia nian ional Science qM Иш. n 1 dur aceae Flora of Institution and the ation, Washington, D.C. : e UN de las е de Hon- s. Ceiba 19: MOORE, n J. ud тез Index to Plant Chromo- some Numbers 1956-1971. Oosthoek's Uitgevers- maatschappij, Utrecht, Netherlands. MORREN, C. 1836. Clerodendrum speciosissimum. Hort. elge 3: 322, pl. 68. MuNIR, А. А. 1989. A taxonomic revision of the genus Clerodendrum L. O in Australia. J. Ade- laide Bot. Gard. 11: 10 PuL, L. VANDER. 1982. e of Miriam in Higher Plants, 3rd ed. Springer-Verlag, New York. шаш Н. 1978. Plantas Usuales nu Costa ine 2nd . Editorial Costa Rica, San José, Costa Ric и V.S. & ESAV 6: Chromosome numbers of a lw horticultural ИЯ Sci. & € 29: 609-610. Rao, А. N. & O. E. Tian. 1974. Зе |" of certain tropical plants. J. Palynol. Кесоквр, S. J. & К. W. Hess. 1941. American ~ of the family Verbenaceae. Trop. Woods 65: 9-10. Reis, S. 1973. Drugs and Foods from VN RR | 890 Annals of the Missouri Botanical Garden Plants; Notes in Harvard University Herbaria. Har- vard Univ. Press, Cambridge, Massachusetts. & . Lipp. 1982. New Plant Sources for Drugs and Foods from the New York Botanical Gar- den Herbarium. Harvard Univ. Press, Cambridge, Massachusetts. Riptey, Н. N. 1930. The Dispersal E Plants Through- out the World. L. Reeve, Londor SAYRE, С. 1975. Cr ryptogamae MEHR an annotated bibliography of exsiccatae of Algae, Lichenes, He- paticae, and Musci. V. Unpublished exsiccatae. I. rcd Mem. New York Bot. Gard. 19: 277- = J. С. 1847. In A. L. P. де Candolle, Prod- romus Systematis Naturalis Regni Vegetabilis. 11: 656-676. Strasbourg, London. Sessk, M. = J. M. Mociño. 1894. Flora Mexicana (2nd ed.). оч Tipográfica de la Secretaria de Fomento, Mexi STANDLEY, P. C. D Rain “и Costa Rica. Field Mus. Publ. vu 18: 10 S. CALD бин 1 + Lista Preliminar de las Plantas de El Salvador (2nd. ed.). Imprenta Nacional. Salvador, El Salvador. т С Р; } cta Regiae Soc. Sci. VENTENAT, E 1804. WILLDE braria $ Epa Ут ден Junk, pen 809 aun illustratus. Nova Up : 208. Jardin à la Malmaison. An- poo Plantarum. Li- Berlin. POLLEN DEVELOPMENT AND FERTILIZATION IN LACANDONIA SCHISMATICA (LACANDONIACEAE) : Márquez-Guzmán,? S. Vázquez-Santana,? E. M Engleman,? A. Martínez-Mena,? Martinez? per E. ABSTRACT The flower of Lacandonia schismatica is bisexual. Three or sometimes four bilocular or trilocular anthers occupy the center of the flower inside t he zone of carpels. The anther wall comprises four layers: epidermis; endothecium whose cells develop helical wall thickenings; one middle layer, which disappears early; a tapetum of the secretory type. The tapet isobilateral arrangement, surrounded al cells apparently form cytoplasmic bridges amo by a thick wall. The ng themselves. The are in an microspore tetrads mature anther wall has only two cell layers: epidermis and endothecium. The pollen imi is three-celled by the time germination occurs. Anther dehiscence has not been observed and probably does not occ Pollination of Lacandor anthers, which do not dehisce, and the pollen tubes grow within the receptacle until they tube enters the embryo sac through the micropyle and o ne syner mia schismatica occurs in the unopened flower bud. The pollen grains germinate within the reach the ovules. The pollen id, and fertilization is completed by anthesis. This reproductive pattern conforms to Lord's definition of preanthesis cleistogamy. Lacandonia schismatica E. Martinez and C. H. Ramos is endemic to shady sites in a rainforest of southeastern Mexico. It is a saprophyte, and flowers all year if moisture is adequate. Flowering is especially abundant in November and December, after the rainy season (Martinez & Ramos, 1989). Lacandonia schismatica has been proposed as a new monotypic family, the Lacandoniaceae (or- der Triuridales) (Martinez & Ramos, 1989). It is closely allied to the Triuridaceae (see the descrip- tion in Maas & Rübsamen, 1986); in fact, it has been proposed that the genus be included in the Triuridaceae (Rúbsamen-W eustenfeld, 1991). The position of the androecium inside the zone of carpels makes the flower unique in angiosperms and gives room for more speculation about the origin of the flower and of the Triuridales. Oth- erwise, the reproductive structures of L. schis- matica (Marquez-Guzman et al., 1989; Martinez & Ramos, 1989) are comparable to those men- tioned for the Triuridaceae to which Lacandonia schismatica is most closely related (Tomlinson, 1982; Maas übsamen, 1986; Rubsamen- Weustenfeld, 1991). On the other hand, its uniqueness makes one wonder if the structure is homologous to flowers in other angiosperms (Stevens, 1991). This paper is a report of continuing studies on the reproductive anatomy of L. schismatica (see Marquez-Guzman et al., 1989). We analyze here the development of the anther, the formation of the pollen grains, some aspects of their germination and observations on the fertilization, an aspect of special interest in this monotypic family, unique in the flowering plants. Supplementing data on microsporogenesis is pro- vided by Davidse & Martinez (1990). MATERIALS AND METHODS Flowers in anthesis, older flowers, and flower buds of different ages were fixed at the site of collection in the Lacandona forest, Chiapas, Mexico. They were collected in January and August 1990. Three fixatives were employed: (a) FAA (5% formalin + 5% acetic acid + 63% ethanol + 27% water); (b) glutaraldehyde (5% glutaraldehyde + 0.25 M su- crose . phosphate buffer, pH 6.8); (c) glutaraldehyde paraformaldehyde (576 glutaralde- hyde + 4% paraformaldehyde + 0.1 M s-collidine, pH 7.2). The material fixed in FAA was embedded in paraffin. The sections (8-10 um) were mounted in 0.005% aniline blue for fluorescence micros- | We Horacio Merchant for the use of a fluor Consejo Nacional de Ciencia y Жене No. 04510 D.F., Mexico. thank Braulio Centeno for pue assistance, Kodak Mexicana for donat nce microscope. This study was partially supported by a grant from the D 111-903663. Departamento de Biologia, Facultad de Ciencias, Universidad Nacional Autónoma de México, A. P. tion of photographic films, and 10-356, * Centro de Botánica, Colegio de Postgraduados, 56230 Chapingo, Mexico. ANN. Missouni Bor. Garb. 80: 891-897. 1993. 892 Annals of the Missouri Botanical Garden Ds PI E x FIGURES 1-6. Anther development — 1. Longisection of the floral bud, scale bar = 250 um.— 2. Longisection of the floral bud showing primordia of the sessile anthers, scale bar — 150 um. — 3. Longisection of the young anther showing periclinal division of an archesporial cell, scale bar — 63 um. — 4. Transection of the anther wall at microspore mother cell stage, scale bar = 25 um.— 5. Transection of the anther with three pollen sacs, scale bar = 156 um.— 6. Transection of the young anther showing microspore mother cells in meiosis, scale bar — 62 um. A, anther; B, Volume 80, Number 4 1993 Márquez-Guzmán et al. 893 Pollen Development and Fertilization in ti Lacandonia schismatica . The material fixed in glutaraldehyde was dehydrated in acetone and embedded in JB-4 plas- tic mixture. Sections were cut at 1-2 um. They were stained with toluidine blue. The third fixative was followed by postfixation in osmium tetroxide. After alcohol dehydration, the tissues were em- bedded in Epon 812. The 2 um and stained with toluidine blue. Additionally, fine sections were examined by transmission electron were cut at 1— microscopy. RESULTS Lacandonia schismatica is a specialized sap- rophyte. Perfect flowers are produced on its aerial shoots. The gynoecium consists of 60—80 papillose carpels. Each carpel surrounds a basal sessile ovule that is anatropous and bitegmic. The inner integ- ument forms the micropyle. The androecium con- sists of three or sometimes four bilocular anthers with introrse stomia. The mature pollen grains con- tain three cells (Marquez-Guzman et al., 1989; Martinez & Ramos, 1989). When the tepals have finished development, the floral meristem begins to form lobes (Fig. the lobes grow, the central ones become recogniz- able as primordia of the sessile anthers in the center of the flower. They also are larger than the sur- rounding lobes that will form the carpels (Fig. 2). In the anthers some of the subprotodermal cells elongate perpendicularly to the surface. These be- come the archesporial cells. Two cells result from a periclinal division of an archesporial cell (Fig. 3). The internal one (the sporogenous cell) divides re- peatedly and forms the sporogenous tissue. T outer one (the parietal cell) divides twice to form the endothecium, a single intermediate layer, and the tapetum (Fig. 4). Sterile cells in the sporogenous zone divide the anther into two or rarely three к sacs (Fig. the ene wall cells have completed di- a de microspore mother cells begin meiosis (Fig. 6). At this time the anticlinal and internal walls of the tapetum disappear. The tapetal cells are at their maximum size and protoplasmic con- tent; the nuclei are large. The tapetal cells appar- ently form cytoplasmic bridges among themselves (Fig. 7). After meiosis, the microspore tetrads are in an isobilateral arrangement as a result of suc- cessive cytokinesis, and they are surrounded by callose walls (Fig. 8 At the time of separation of the microspores, they are still surrounded by a thick wall. The ta- petum seems to be functional and some of its cells are binuclear (Fig. 9). During development of the pollen grains, the intermediate layer and the ta- petum disappear (Fig. 10) By the time pollen grains are mature, the en- i These thickenings taper off to nothing as they approach the external wall (Fig. 11). The bands of thickening are sometimes connected at the edges of the in- dothecium has helical wall thickenings. ternal wall. Thus, the internal wall is traversed by the bands, but not by a plate of secondary wall. The epidermis persists to maturity. We have observed an apparently new phenom- enon in at least eight flowers in anthesis: the anthers did not dehisce and the pollen grains germinated within the pollen sac. e development of the anther is precocious with respect to the carpels (Fig. 12). The male gametophytes mature before the female ones. By the time pollen grains in unopened buds are at the three celled stage (Fig. 13), the carpels are usually in early stages of development. However, the ovules and embryo sacs continue to develop until, before the flower opens, the megagametophyte has seven cells and the egg apparatus is located at the mi- cropylar end, next to the receptacle (Fig. 14). At this stage, the first signs of germination were observed in the pollen grains inside the unopened anthers (Fig. 15). The resultant pollen tubes grow toward the base of the anther, which is united to the receptacle (Fig. 16). The pollen tubes grow through the receptacle (Fig. 17) to reach the car- pels (Fig. 18). They enter the ovule via the mi- cropyle and then penetrate the embryo sac through one synergid (Fig. 19). When the flowers open, fertilization already has occurred, and only the pollen walls (Fig. 20) remain in the unopened an- thers. Dehiscence of the anthers does not occur, nor have we seen pollen tubes growing through the anther walls. Pollen grains have not been detected outside the anthers, neither in sections of flower buds nor in images of scanning electron microscopy of open flowers. DISCUSSION Lacandonia schismatica shares many char- acters with the Triuridaceae (Maas & Rübsamen, E bract; C, carpel; EP, epidermis; L, lobes; MM, microspore mother cells; PC, parietal cell; PS, pollen sac; R, receptacle; | SC, primary sporogenous cell; TP, tepals. 894 Annals of the Missouri Botanical Garden a de Ps 7. Apparent cytoplasmic bridges among tapetal cells, scale bar = 25 um.— 8. Microspore tetrads, scale bar — 63 um. — 9. Microspores surrounded by a thick wall, scale bar — 25 um.— 10. Longisection of scale Баг = 100 um. Anther wall with epidermis and fibrous thickenings in endothecium, . EN, endothecium; EP, epidermis; M, middle layer; MI, microspore; MT, microspore tetrads; FIGURES 7-11. the mature anther, scale bar = 20 um. E PC, young pollen grains; T, tapetum; TW, wall thickenings. Volume 80, Number 4 Márquez-Guzmán et al. 895 1993 Pollen Development and Fertilization in Lacandonia schismatica фа плићи. ined Ri. + RFS А . 3: ФА X 8:5 А E фи: > 2n i Y, QC Rs ‘ard 4 FIGURES 12-16. — 12. Longisection of the flower bud, scale bar = 125 um.— 13. Tricellular = grains, scale bar = 5.88 um.— 14. Egg apparatus, scale bar = 100 um.— 15. The е pollen grains germinate e the an oO scale bar = 10 um. — 16. Pollen Uber i in the base of the anther, scale bar = 23.81 um. A, pe er; B, bra carpels; E, egg; EN, endothecium; II, inner integument; MY, micropyle; PN, polar nuclei; PT, pollen ы. R, Euh cells S, synergids; SE, sperm nuclei; TP, tepals; V, vegetative anclas W, pollen wall. 896 Annals of the Missouri Botanical Garden FiGURES 17-20. bar = 25 I pollen wa 1986; Martinez & Ramos, 1989; Márquez-Guz- mán et al, 1989; Davidse & Martinez, 1990; Rübsamen-W eustenfeld, 1991): anthers bilocular or rarely trilocular or tetralocular, persistent anther tetrads epidermis, successive microsporogenesis, isobilateral, mature pollen grains with three cells. — ] 7. Pollen tubes among receptacle cells, scale bar — the mic ўа and the embryo sac, scale Баг = 62.5 m. — 19. TI 25 um.— 18. The pollen tube enters The walls of an empty pollen grain inside the anther, scale bar = 5.88 um. ES, embryo sac; I, inner па он. МУ, писгорује; OI, outer integument; PT, pollen tube; R, receptacle cells; S, synergids; W ll. - Maas & Rübsamen (1986) found that the ta- petum in Triuridaceae is intermediate between peri- plasmodial and secretory and tends to protrude into the pollen chamber. The cell walls degenerated at an earlier stage in Sciaphila than in Triuris. We did not detect a plasmodial stage in the tapetum Volume 80, Number 4 Márquez-Guzmán et al. 897 Pollen Development and Fertilization in Lacandonia schismatica of Lacandonia, but we did find apparent cyto- plasmic bridges among the cells. There are no data on pollination in Triuridaceae. It is to be expected that several kinds of insects are attracted to the flowers, although this is not directly recorded (Maas & Rübsamen, 19806). ermination of pollen grains inside the anther sac in Lacandonia appears to be unique in Triuri- dales and a consequence of its cleistogamic repro- duction. Double fertilization in Triuridaceae has not been observed; however, most likely it takes place in several species (Maas & Rübsamen, 1986). Tom- linson (1982) suggested that there is evidence for a possible apomictic process. Dahlgren et al. (1985) mentioned that the egg cell of Triuridales may develop parthenogenetically. Rübsamen-W eusten- feld (1991) observed double fertilization in one embryo sac of Soridium spruceanum Anderson (1980) described le Орай and fertilization phenomena to those of L. schis- matica in four genera of Malpighiaceae. However, these four genera also produce chasmogamous flowers on the same plants. Apparently, in L. schis- matica all flowers are cleistogamous. In these gen- era of Malpighiaceae, the cleistogamous flowers generally lack y but in L. schismatica all the flowers have Philbrick dons has found that ovules of two species of Callitriche are fertilized by pollen tubes that grow through various tissues between the an- ther of a male flower and the ovary of an adjacent female flower. Lord (1981) reported that cleistogamy occurs in 56 families and covers a wide range of dimorphic forms. Under her classification, L. schismatica belongs to the category of preanthesis cleistogamy, in which both pollination and fertilization occur in a flower bud which later opens. Autogamy generally is expected to lead to ho- mozygosity in the population. This has been dem- onstrated in L. schismatica by Coello et al. (1993). LITERATURE CITED ANDERSON, W. R. Cryptic self- и іп ће Malpighiaceae. dig 207: 892 CoELLO, G., A. ESCALANTE & J. еы “1998. Lack of genetic sere in Lacandonia schismatica in its only known locality. Ann. Missouri Bot. Gard. 80: 898-901. DAHLGREN, R. M. T., Н. T. CLIFFORD & P. F. YEo. 1985. The Families of the Monocotyledons: Structure, Evo- lution, and Taxonomy. Springer-Verlag, New York. DavipsE, G. & E. MARTINEZ. 1990 e chromosome number of Lacandonia schismatica (Lacandoni- aceae). Syst. Bot. 15 -637. Говр, E. M. 1981. Cleistogamy: А tool for the study of floral morphogenesis, function and evolution. Bot. Ee Rev. 47: 42 Maas, P. J. M. . RUBSAMEN. 1986. Triuridaceae. Flora distet 40: 1-55. MARQUEZ-GUZMAN, ENGLEMAN NEZ- ENA, E. MARTINEZ & C. Ramos. 1989. Anatomia reproductiva de Lacandonia schismatica (Lacandoniaceae). Ann. Missouri Bot. Gard. 76: 124- 127. MARTINEZ, E. & C. H. Ramos. 1989. Lacandoniaceae 5): una nueva familia de México. Ann Missouri Bot. Gard. 76: 128-135. PHILBRICK, C. Т. 1984. Pollen tube growth within veg- etative tissues of Callitriche (Callitrichaceae). Amer. J. Bot. 71 86. RÜBSAMEN-WEUSTENFELD, T. 1991. Morphologische, embryologische und сае untersuchungen an Triuridaceae. Biblioth. Bot. 140: 1-113. STEVENS, P. F. 199 Lac е schismatica — А challenge to some recent theories of floral morpho- genesis? Flowering Newsletter 12: 32-33. TOMLINSON, B Helobiae (Alismatidae). Pp. 466-475 in C. R. Metcalfe (editor), р of ilio Monocotyledons VII. Clarendon Press, Oxfor LACK OF GENETIC VARIATION IN LACANDONIA SCHISMATICA (LACANDONIACEAE: TRIURIDALES) IN ITS ONLY KNOWN LOCALITY' Gerardo Coello, Ana Escalante,’ and Jorge oberon? ABSTRACT Lacandonia schismatica has only one known population i in a small patch of tropical rain forest. Using electrophoretic l . techniques, we assessed its genet q g ic variation in eight loci; n implications for the conservation prospects of this threatened specie o variation was found. We discuss this result and its Lacandonia schismatica (Martinez & Ramos, 1989; Marquez-Guzman et al., 1989), the only species of Lacandoniaceae, is unique in having a central androecium surrounded by an apocarpous gynoecium. А series of studies of the plant is un- derway to determine its relationships and taxonom- ic status, which is still under debate (Stevens, 1991). The unusual morphology and restricted distri- bution of Lacandonia schismatica suggest that it is an interesting model to study macroevolutionary mechanisms. In this paper we report results on the genetic variation of L. schismatica as estimated by enzyme electrophoresis. METHODS The only known locality for L. schismatica is the type locality (Fig. 1), which is a ca. 15-ћа tropical rainforest patch a few miles north of the Montes Azules Biosphere Reserve in Mexico. The climatological station at Bonampak, close by and at a similar аним, has a mean annual temper- ature of 24.6?C and precipitation of 2,609 mm per year (Meave, 1990). completely surrounded by secondary growth veg- etation and cattle pastures. The plants are found within the forest, but only 10- e type locality is almost O m from the edge with a natural savanna. This edge has wet soil, almost a marsh, with an underground water level at about 60 ст below the surface during the dry season. During the dry season, the plants are inconspicuous and can only be found under the litter. During the rains, however, rotting logs are often covered with carpets of L. schismatica (E. Martinez, pers. comm.). Individuals of L. schismatica were collected in a 0.5-ha plot (Fig. 1) on three dates, all piis the dry season: (A), 10 individuals, April 1988; (B), 72, April 1989; and (C), 27, Jul mon The sample size changed between sites depending on the abundance of the plant. Plants from sample A were collected along a 100-m transect. Plants from samples B and C come from 10-20-т? high-density patches not more than 50 m away from each other. Live specimens were transported to Mexico City and maintained in growth chambers at 25?C and 80% humidity until prepared for electrophoresis. Starch gel (12%) enzyme electrophoresis (Val- lejos, 1983) was carried out on fresh stem tissue homogenized with gel buffer and absorbed in filter paper wicks (Whatman 17). Eighteen enzymes were assayed, but only eight resolved well in two buffer systems. The enzymes analyzed and the stain ref- erences are shown in Table 1 System I consisted of tris-citrate pH 7 electrode buffer and L-histidine pH 7 gel buffer and system ' We thank Esteban Martinez for taking us to the Lacandonia site and helping with the fieldwork, and for many discussions about the plant. His assistant Gabriel Aguilar was very elpful as well. Daniel Pinero and Luis Eguiarte made several useful comments on the manuscript. An anonymous referee significantly corrected our English. Gabriela мети Мао the location etc o de Ecologia, Universidad Nacional Autónoma de México, Apdo. Postal 70-275, México 04510 D.F., Мен ANN. Missouni Bor. Garp. 80: 898-901. 1993. Volume 80, Number 4 Coello et al. 899 Lack of Genetic Variation in Lacandonia schismatica 9100" Nvo. Fco León + 17:00 an Lacenjó Bonempak GURE l. Location of the site of Lacandonia schismatica. The striped area in С represents cattle pasture land and the dotted area a natural savanna community. Numbers 1, 2, and 3 in С represent collection sites. The polygon around the dotted area is the plot reserved by the Chole Indians w protect the plant. II of lithium-borate pH 8 electrode buffer and tris- RESULTS borate pH 8.6 gel buffer (Soltis et al., 1983 Electrophoresis was conducted at 55 mA for system No electrophoretic variation was detected among I and 45 mA for system II until the front migrated any of the plants or sites in the 14 loci examined. 10 cm. Stems of 109 individuals were assayed and the 900 Annals of the Missouri Botanical Garden TaBLE l. Enzymes assayed in Lacandonia schismatica and stain recipes references. Enzyme Code Stain reference Acid phosphatase (ACP) EC. 3.1.3.2 Soltis et al. (1983) Aconitase ( E.C. 4.2.1.3 Wyatt (1989) Alcohol riadas (ADH) E.C. 1.1.1.1 Vallejos (1983) Diaphorase (DIA) E.C. 1.6.4.3 Wyatt (1989) Glutamate oxaloacetate transaminase (GOT) E.C. 2.6.1.1 Wyatt (1989) Leucine-amino peptidase (LAP) E.C. 3.4.1.1 Wyatt (1989) Peroxidase ) E.C. 1.11.1.7 Vallejos (1983) 6-Phosphogluconate dehydrogenase (6-PGD) E.C. 1.1.1.44 Soltis et al. (1983) same alleles were expressed in all of them. Table 2 shows the nuniber of loci that resolved for each enzyme analyzed, and the т of Figure 2 shows the activity zone of each lo Assuming that the specimens sc a ran- dom sample of 218 alleles at each locus, we would have detected any variant allele that existed at an overall frequency of 1.4% or greater with a prob- ability of 95% (P = 0.986" = 0.05). Thus, genetic variation is not apparent in our sample. Lack of genetic variation has also been reported for other endemic species (Hamrick et al., 1979; Waller et al., 1987) and has implications for a conservation program (Holsinger & Gottlieb, 1989). DISCUSSION Lack of genetic variation can be explained by a number of mechanisms, including random drift, genetic bottlenecks (Waller, et al., 1987), and apo- hall & Brown, 1981). In the case of Lacandonia schismatica, the lack of genetic vari- mixis (Mars ation can be explained by the autogamic fertiliza- tion system reported by Márquez-Guzmán et al. (1993, this issue). Since Lacandonia schismatica is apparently restricted to a small area, the existence of several subpopulations cannot be rejected at this time. The small size of the plant, its location under the litter, and its cleistogamous pollination system (Márquez- Guzmán et al., 1993, this issue) suggest very low levels of gene flow, which may lead to several subpopulations within a small area. The samples used in this study were collected in roughly the same site, and it may be that other subpopulations could have the same low genetic variation but for This still has to be tested. he low genetic variation has contradictory im- different sets of alleles. ^ o TS — — — — > -—^"—— = — —— — — сз — —Á o — —— [e pem < c о © © = -— — = © ___| A ___| A A | E 3 ACP ACO ADH DIA GOT LAP PER 6-PGD FicuRE 2. Zymogram showing the activity regions for the eight enzymes tested in Lacandonia schismatica. Volume 80, Number 4 1993 Coello et a Lack of Genetic Variation in Lacandonia schismatica TABLE 2. Enzymes analyzed on each buffer system and the number of loci detected for each one in Lacan- donia schismatica from Mex Number of loci System Enzyme I ACP plications for the conservation of the plant. On one hand, obtaining representative variation for pres- ervation in the greenhouse may be feasible using only a small number of specimens. Unfortunately, to date all attempts to preserve living specimens in growth chambers for more than two or three months have failed. Failure may be due to death of the associated root fungi. On the other hand, the forest patch in which the only known population of L. schismatica is located will soon be completely surrounded by secondary vegetation and natural and artificial grasslands. It is likely that microen- vironmental conditions in its habitat. will change, and it may be the case that L. schismatica lacks the genetic variability to adapt to the new situation. Although it now appears that the locality of schismatica will not be cleared and will be kept as a sanctuary by the owners, the Chole Indian community of Corozal, it is important to begin intensive ecological and genetic studies of the spe- cies in order to insure its conservation. It is unlikely that the area in which the plant is located will be able to support a viable population without some degree of management. Given the situation of the habitat, and what is known about the plant's ge- netics and biology, it is clear that an immediate effort is required to preserve the plant in laboratory conditions and to understand its ecological require- ments and genetic structure in order to propose sound management strategies. LITERATURE CITED Hamrick, J. L. 1979. eio не between life history characteristics and electrophoretically detectable ge- netic чаш in plants. posee Rev. Ecol. Syst. 10: 173- HOLSINGER, K. E. & GorrLiEB. 1989. The conser- vation p rare and endangered plants. Trends Ecol. HL Ramos 1989. Lacandoniaceae opaco Una nueva iced de México. Ann. Mis 28- MARQUEZ Gumá ÁN, I. M. Елак E T Martinez MENA, E. Mar Z & C. RAM Anatomia re- productiva de Lacandonia $c dua a (Lacandon- ae). Ann. Missouri Bot. Gard. 76: 124-127. маноз болмак. Ја 5 VÁZQUEZ-SANTANA, E. M. A. MARTÍNEZ MENA & E. MARTÍNEZ. 1993. "Pollen е and fertilization іп La- candonia schismatica (L oniaceae). Ann. Mis- souri Bot. Gard. УЯ 891-897. R . M. D. BROWN. MARSHALL, een The столов of apomixis. Heredity 47(1): МЕАУЕ, J. 1990. Estructura y dius de de la Selva Alta Perennifolia de los Alrededores de Bonampak. Instituto Nacional de Antropologia e Historia, Serie Arqueologia, México D.F. E., C. M. Haurrer, D. C. DARROW & €. J. А 1983. Starch gel einen of ferns: a compilation of grinding а and staining sched- ules. = ae 738: 9- STEVENS, P. l. = ~ schismatica — А i to some recent theories of floral morpho- genesis? Flowering Num 12332-33: VALLEJOS, C. Е. . Enzyme activity stain 409-516 in S. D. Tanskley & T. J. Orton (editors) Isozymes in Plant Genetics and Breeding. Vol. Elsevier, Amsterdam. WALLER, D. M., ОМАНУ & noc. sap 1987. Genetic variation in the extreme endemic Pedicularis furbishiae эре ис тен M Biol. 1(4) 35-340. у= = WYATT, К. A general protocol for starch gel electrophoresis. Mimeographed | Dept. of Botany, Univ. Georgia, Athens, Georg PATTERNS OF SPECIES DISTRIBUTIONS IN THE DRY SEASONAL FORESTS OF SOUTH AMERICA! Darién E. Prado? and Peter E. Gibbs? ABSTRACT Studies of the distributions of species of seasonal woodland habitats in South America by means of dot-mapping and phytosociological analyses indicate the presence of three nodal areas: the woodland species of this affinity are notable by their absence from the cerrado vegetation of central Brazil, although some occur on calcareous outcrops in the general cerrados area, and they also avoid the Cha nostly disjunct du вік patterns are ves It is proposed that these fragmentary and r largely contiguous seasonal woodland formation, which may aco of northern Argentina. tiges of a once extensive and ave reached its maximum аа during a dry-cool climatic period ca. 18,000-12,000 BP, coinciding with the а of the humid fores Various authors have drawn attention to the “sa- vanna corridor" (Schmidt & Inger, 1951) or “di- agonal of open formations" (Vanzolini, 1963, 1974), which extends across South America from Maranhào-Ceará in north-northeastern Brazil, to the Chaco region of northern Argentina (Fig. 1), and numerous commentaries on possible floristic and faunistic links have been made (e.g., Andrade- Lima, 1954, 1982; Vanzolini, 1974; Prance & Schaller, 1982; Bucher, 1982; Haffer, 1985; Rat- ter et al., 1988), particularly for the Caatinga- Chaco poles of this diagonal. In the course of a recent critical reevaluation of the alleged floristic links between the caatingas of northeastern Brazil and the Gran Chaco vegetation of northern Ar- gentina- Paraguay - Bolivia (Prado, 1991), a series of dot-map distributions were plotted for woody species of the deciduous and раене forests of South America. The Caatinga—Chaco links pr to be negligible, but a different and rather more interesting distribution pattern emerged. The distributions hint at a once extensive arc of seasonal woodland, which ranged from north- eastern Brazil (with some Amazonian and Carib- bean links) through southeastern Brazil to the con- fluence of the Paraguay and Paraná rivers, into southwestern Bolivia and northwestern Argentina, and extending sporadically northward in dry An- dean valleys of Peru. In most cases, these species skirt the impoverished soils of the central Brazilian cerrado vegetation (Eiten, 1972), but some taxa “island-hop” stricted outcrops of calcareous-rich soils within the across this area by way of the re- cerrados. CLIMATIC CHANGES IN SOUTH ÁMERICA DURING THE LAST GLACIAL MAXIMUM Many authors (e.g., Damuth & Fairbridge, 1970; Ab'Sáber, 1977, 1982) have postulated a major dry-cool climatic period in South America at the time of the Northern Hemisphere Wisconsin- Würm ! We of Harvard Univ University, mpres Universidad de Ros Pernambuc and Vali Pott for kind assistance with field excursions in Brazil, ank the directors of the Royal Botanic Gardens at Edinburgh and Kew for Ain of herbarium and library facilities -i repeated visits, and also to the following herbaria for the loan of sp rsity, Conservatoire et Jardin botaniques de la a de Pesquisa Agropecu rio. D. Prado i is grateful to Maria Angêlica Figueiredo, Maria das Graças, M. Pires, an Ville de Genève, Gra Harvard and Herbário, Botánica y Ecologia Vegetal, rnildo and acknowledges financial assistance from CONICET апа, Recife, Argentina, the University of St. Andrews, and © oe of Vice-Chancellors and Principals of the Universities of the U.K. for an Overseas Research Student’s * Cátedra de Botánica O EA Sistemática, Fac ultad de Ciencias Agrarias, Universidad Nacional de Rosario, Santa Fe 2051, Rosario 2000, Ar Plant Science Laboratories, Sir Harold Mitchell Building, The University, St. Andrews KY16 9AL, Scotland, J K. ANN. MISSOURI Bor. Garp. 80: 902-927. 1993. Volume 80, Number 4 993 Prado & Gibbs 903 Dry Seasonal Forests of South America glaciation, about 18,000- 12,000 BP. There has been considerable discussion devoted to the con- sequences of such a dry period with regard to the contraction of the humid forest to isolated “islands” and the “refuge theory" as a possible stimulus for speciation events that are reflected in present-day Amazonian forest diversity patterns (Brown & Ab'Sáber, 1979; Prance, 1973, 1982, and various contributions in Prance, 1982). Other authors, however, have urged caution in interpreting pres- ent-day species distributional patterns in terms of supposed Pleistocene forest refuges (e.g., Benson, 982; Nelson et al., 1990). Unfortunately, paleo-palynological studies for South America, which could provide valuable ev- idence for dry periods and consequent vegetational changes, are sparse and mostly apply to Andean areas (e.g., Van der Hammen, 1974, 982), and the few studies in the Amazonian regions are for more recent Holocene strata (Absy, 1‘ . How- ever, on the basis of local palynological investi- gations some authors have attempted to circum- scribe in more detail the possible regional shifts in climate and vegetation in the Paleocene (Markgraf, 1989), while others have reported evidence that conflicts with areas of alleged humid forest refuges (Kam-biu & Colinvaux, 1985; Colinvaux, 1989). Most discussion concerning the postulated Pleis- tocene dry periods in South America has focused on the consequences of contraction of the humid forest, and less attention has been given to the influence of such climatic changes on the expansion of semideciduous and deciduous vegetation of the region. Here we present a series of distribution maps for a number of species of seasonal woodland formations, which reveal floristic links between the flora of the caatingas vegetation of semiarid north- eastern Brazil (sensu Andrade-Lima, 1981) and two other areas. In our opinion, these distributions are remnants of once continuous forests, and they pro- vide some evidence for the previous extensive oc- currence of seasonal forests in South America. While we are fully conscious of the dangers of facile interpretations from dot-maps, and particu- larly the possible influence of localized collecting, we consider that the repetitive distributions that emerge from these maps do reflect past floristic links. We offer these distribution maps as a con- tribution to the ongoing debate concerning the his- torical biogeography of the seasonal formations in South America. METHODS This series of species distribution maps began as an attempt to plot accurately the full distribu- \ The “dry diagonal” of seasonal woodlands in South America with Caatingas, Cerrado, and Chaco areas delimited. FIGURE 1. tions of the woody taxa of the caatinga vegetation Andrade-Lima, 1981; Prado, 1991) of north- eastern Brazil. It soon became evident that many ~ of the nonendemic caatinga species showed а com- mon overall distribution pattern that mostly avoid- ed the cerrado region but extended to southeastern and southern Brazil and, in many cases, into well- defined areas in Paraguay, Argentina, and Bolivia. The maps involved an exhaustive check of the taxonomic literature and verification of the identity of species cited in diverse floristic lists, made pos- sible by repeated visits to The Royal Botanic Gar- dens at Edinburgh and Kew, and herbarium ma- terial obtained on loan from various institutions. The species dot-maps are based on verified spec- imens or citations in modern taxonomic revisions (solid symbols) together with citations in reliable vegetation surveys in the areas concerned (open symbols). In addition to mapping species distributions, a suite of floristic lists from 18 seasonal woodland communities across South America were compared by means of a classic phytosociological analysis of the Zurich-Montpellier school, in the version mod- ified by Mueller-Dombois & Ellenberg (1974), and also by means of a complete linkage algorithm from Wishart's (1987) package employing Serensen's (1948) similarity index, and a Principal Compo- 904 Annals Ap ALT Garden nents Analysis from the JMP IN statistical software (Prado, 1991). Details of these latter analyses are not given here, but both approaches strongly sup- ported the close floristic links between the caatinga vegetation and woodlands in south-southeastern Brazil and Paraguay, and the piedmont forests of northwestern Argentina-southern Bolivia. RESIDUAL PLEISTOCENIC SEASONAL FORMATIONS АКС Ab'Sáber (1977) has postulated that during a major dry episode in South America, equivalent to the Wisconsin- Würm glacial period, there was an expansion of semiarid vegetation in the continent. According to this view, caatinga-like vegetation, which usually occurs on mineral-rich soils, may well have surrounded a core area of cerrado veg- etation in central Brazil, and both seasonal for- mations very probably extended into the Amazon region, whereas tropic al rainforests underwent con- traction. It is likely that the slow return of a more humid climate during the past 12,000 years, in addition to favoring a re-expansion of the evergreen forest, was also accompanied by leaching and acid- ification of soils in various areas, perhaps increasing the cerrado region, and fragmenting the originally much more widespread semiarid a nd of base- rich soils (Ratter et al., A number of the woody species that are likely to have been involved in the cyclic expansion- retreat migrations of the Pleistocene are today members of diverse tropical and subtropical decid- uous forest communities in South America, such as the caatingas of northeastern Brazil and some semideciduous (i.e., non-Mata Atlántica) forests in Sào Paulo and Paraná states, the forests of the upper Uruguay River valley and most of those surrounding the Paraguay- Paraná rivers system, the serranias of Santiago and Chiquitos in south- eastern Bolivia, the piedmont forests in the so- called Transitional areas in northwestern Argentina and southwestern Bolivia (running parallel to the sub-Andean mountains from Tucumán in Argentina to Santa Cruz de la Sierra in Bolivia), and more scattered outposts in semiarid and arid Andean valleys in northern Bolivia, throughout Peru and sometimes even reaching southwestern Ecuador. Some of these species are not only important mem- bers of these communities but are also dominant in several instances. One such case, and here taken as paradigm, is that of the mimosoid Anadenanthera colubrina (Vell.) Brenan var. cebil (Griseb.) Altschul (“ап- gico,” “cebil””). For the distribution of this species (Fig. 2A) three nuclei can be identified. The first and major nucleus consists of the caatingas of northeastern Brazil, where the “angico” is a very important and frequent tree, and it is present in most of Andrade-Lima's (1981) vegetation units. Outside the caatingas, it has been reported in the "brejos" from Pernambuco (Andrade-Lima, 1964), particularly i in the dry forests of the slopes of the "serras," and in the so-called “caatingas” in Mar- anhào state (Сајуао, 1955). Bigarella et al. (1975) claimed that disjunctions of caatingas occur on the road Coroatàá- Vargem Grande (Maranhao), and the exsiccatum marked on our map corresponds to Presidente Dutra, which is just to the south of these localities. This northeastern Brazilian nucleus for the angico type of distribution (which we can call the Caatingas nucleus) is intermittently linked to the second nucleus, which occurs along the Paraguay-Paraná rivers system, by the way of mesotrophic cerradào woodlands in Goiás and Mato rosso (Ratter et al., 1978), and in Mato Grosso do Sul (Ratter et al., 1988; Prado et al., 1992). The second nucleus has a somewhat triangular shape with its northernmost apex on the axis Puerto Suárez-Corumbá (Bolivian-Brazilian border), straight meridian side reaching south on the axis Resistencia-Corrientes in Argentina and thus par- alleling the Paraguay River up to its confluence with the Paraná River, a latitudinal side eastward from that confluence to the upper Uruguay River valley in Misiones (Argentina) and Santa Catarina (Brazil), and from here a diagonal side returning in northwest direction to Pto. Suárez-Corumbá. Thus, this nucleus (which we can refer to as the Misiones nucleus, employing this term in its older, more general geographical concept rather than the current restricted Argentinian or Paraguayan us- age) includes nearly all of eastern Paraguay and a good deal of the western bank of the River Paraguay, that is, a north-south strip parallel to the river and several kilometers inland in the geo- graphical Gran Chaco region. Within this triangle, Anadenanthera colubrina var. cebil is common but rarely dominant. It has been reported by Fie- brig (1933) along the western bank of the River Paraguay, by Tortorelli (1967) and Esser (1982) for several areas of eastern Paraguay, by Morello & Adámoli (1967) for eastern Formosa, and by Eskuche (1986) for northeastern Argentina in the а phytogeographical province (Cabrera & Willink, 0). This а nucleus connects with the third one, located in southwestern Bolivia and northwestern Argentina, via the Santiago and Chiquitos hill rang- es, forming a curved line uniting Corumbá to Santa 905 Prado & Gibbs Volume 80, Number 4 1993 Dry Seasonal Forests of South America "sisi ONSLOW ut suomi 01 лојол s[oquiás uado е pue sydesZouour 1ua2e1 ш зиоцеџо Jo suaunoads штиледлоц о} 19jo1 sjoquIAs pios *peyipenb asimiayio sso[u[] ‘setveds ривјроом [euoseas jo зиоцпа ша] "G[-c 59912914 + "unnbipnsi1101u02 штдојолојил “(| — »anapunan wmiu0ssp 7) — sisu24022 DUDANQUIY “Y — (w) »ui40n]oo стел Ф) 11922 IA рит1дтоә рләујириәроиү "y —',c 380214 906 Annals of the Missouri Botanical Garden Cruz de la Sierra. The flora of this region has scarcely been studied and even less collected, but Herzog (1910, 1912) gave a fairly accurate im- pression of what the vegetation was like in the area. In the sector enclosed by this previous line and hills, by the western limit of the second nucleus, and by the north-south narrow strip of the third nucleus, the true chaco vegetation is encased. There are very few outposts of Anadenanthera colubrina var. cebil within the Chaco area (as is the case with most of the species with this pattern of dis- tribution). The Cerro León records of this tree are not an exception to its absence from the plains of the Chaco sens. str., because the vegetation on the slopes of this tectonic horst (which comprises Si- lurian and Devonian rocks as opposed to the zolian sediments surrounding the hill) shows strong links with the sub-Andean Piedmont forests in north- western Argentina (Ramella & Spichiger, 1989, sub Transitional Forests). he third nucleus extends from Santa Cruz de la Sierra south to Tucuman and the sierras of eastern Catamarca in Argentina. This follows the piedmont area of the sub-Andean Mountains, which is the area of the so-called transitional forests, so- named because their floristic composition has been considered to be intermediate between that of the chaco thorn forests and the upland subtropical rain- forests (the “Yungas” sens. str.). We can refer to this nucleus, therefore, as the Sub-Andean Pied- mont nucleus. bil was so important in the southern half of this nucleus in Argentina that Lillo (1919, as cited in Hauman, 1931, and Digilio & Legname, 1966) called и zona del cebil,” but today there is little left of the original Ма аи in this area, whic is known locally as **tipa-pacará" forest (from the species Tipuana при (Benth.) O.Kuntze and Еп- terolobium contortisiliquum (Уећ.) Morong). The northern remainder of the nucleus where the angico forest, domi- Anadenanthera colubrina var. ce- occurs comprises the nated by Calycophyllum multiflorum Griseb. and Phyllostylon rhamnoides (Poisson) Taubert, and which extends from mid-eastern Jujuy in Argentina to, presumably, Santa Cruz de la Sierra. Ападеп- anthera colubrina var. cebil is one of the five commonest members of both these forests (Ca- brera, 1976: 7) rom this third nucleus this species, and the pattern of distribution dealt with here, becomes scattered along the northeast-facing slope of the Andes in Bolivia and southern Peru, i.e., from Santa Cruz to Cuzco, to even more scattered lo- calities in dry inter-Andean valleys such as those of the Apurimac, f “palo blanco" Huallaga, and Maranon rivers (Weberbauer, 1936), finally reaching southern Ec- uador in what corresponds to the “Savannah and Dry Scrub vegetational types" (see map in Harling, 1979). The distribution of the typical variety, An- adenanthera colubrina var. colubrina, further ex- tends the range of the species (Fig. 2A), since although it occurs sympatrically with variety сеђи in the southern area of the caatingas in Bahia, and also in Misiones in Argentina, it is distributed al- lopatrically in Rio de Janeiro state and the planalto forests of Sao Paulo and Paraná states. Altschul (1964) has argued cogently that the typical variety evolved from variety cebil, in which case variety colubrina has developed in wetter areas than those of its parental stock. number of woody caatinga species have a similar, sometimes virtually identical, distribution pattern to that of Anadenanthera colubrina. For example, Amburana cearensis (Allemào) A. C. Smith (Leguminosae) is a frequent, albeit not abun- dant, characteristic tree of the caatingas even in the more shrubby types of communities, and it occupies a similar niche in the palo blanco forest (Piedmont nucleus), while it is rather rare in the lati Misiones nucleus except for som on calcareous substrate on the Apa ius region in Paraguay (Fiebrig, 1933; Tortorelli, 1967). It has also been reported for the Cerro León (Ramella & Spichiger, 1989), and it appears very occasion- ally in the Andean sector of this pattern of distri- bution; a collection from Acre state in Brazil should be noted (Fig. 2B). The tree Astronium urundeuva (Allemào) Engl. (Anacardiaceae) is an important member of the arboreal caatinga, in which it is characteristic and dominant (Andrade-Lima, 1981), but it also extends into the shrubby caatinga as an isolated low tree (as does Amburana cearensis, according to Egler, 1951). Astronium urundeuva is considered an indicator species of mesotrophic facies cerradào (cerradào is a well-developed wood- land type, which is usually characteristic of richer soils) within the cerrado area in central Brazil (Rat- ter et al., 1978). It has also been collected in the Cerro León, and it is very frequent in the sub- Andean Piedmont nucleus, but vires are no records from the Andean dry valleys (Fig. 2 The mimosoid tree En pande contame quum (Vell.) Morong is usually conspicuous, not for its frequency but rather for its size. Somewhat scarce in the caatingas (Fig. 2D), it has been re- ported for the agreste (Andrade-Lima, 1954) and arboreal caatinga (Andrade-Lima, 1975). In the forests of southeastern Brazil its presence is re- corded by collections and phytosociological works (Martins, 1979), and it is an important component Volume 80, Number 4 Prado & Gibbs 907 Dry Seasonal Forests of South America of the vegetation on the banks of the Paraguay- Paraná system (Reboratti & Neiff, 1986; Prado et al., 1989), and also of the sub-Andean piedmont forests (particularly in the tipa-pacará variant), but again is absent from the Andean area. Enterolob- ium contortisiliquum has also been collected in the gallery forests that border the Chaco vegetation (see Fig. 2D and Morello & Адатоћ, 1974), where it occurs with other nonchaquenian subtropical spe- cies, on the alluvial levees beside the main rivers (Adámoli et al., 1972). Pterogyne nitens Tulasne (Leguminosae) has a very similar distribution (Fig. 3A) and is also reported for gallery forest and humidity-dependent communities within the Chaco (Morello, 1967; Morello & Adámoli, 1974), but as a rather rare element of the latter. The species Ruprechtia laxiflora Meisn. (Polygonaceae) was originally described by Meisner (1855) from Blan- chet exsiccata collected in the caatingas (Cocucci, 1961), but it was not recorded subsequently in this area until recent collections by Harley et al. from around Camaleào in Bahia. It was, however, abun- dantly collected in Paraguay and Argentina in both the Misiones and Piedmont nuclei, where it is a very important element (Fig. 3B). The scandent shrub or treelet Celtis pubescens (Kunth) Sprengel (Ulmaceae) follows much more closely the Ana- denanthera colubrina var. cebil pattern (althoug it is far less frequent in the caatingas), and it reaches central Ecuador (Fig. 3C) and is also re- corded from Acre state in Brazil. One of the most important and characteristic trees in the caatingas is Schinopsis brasiliensis Engler (Anacardiaceae). Engler (1876) recognized two varieties, variety glabra Engler and the typical variety of this species. Meyer & Barkley (1973) subsequently considered the former variety to merit specific status, based on its lack of indumentum and larger leaflets when compared to variety bras- iliensis. However, from the numerous exsiccata we have examined it would seem that although presence-absence of indumentum is a fairly con- sistent character, leaflet size is unreliable. A tra- ditional view has been taken of S. brasiliensis, therefore, to include the two sympatric varieties, brasiliensis and glabra, and they have been mapped together (Fig. 3D). This tree is not only present and abundant in the caatingas, from where it extends into isolated localities in the Federal District of Brazil and in Minas Gerais (where it has been reported to form a dense community in a flood-prone area by Pinto, 1985), but it is also the dominant species in forests in Mato Grosso do Sul (Prado et al., 1992). Schinopsis brasiliensis also extends from Corumbá to the west along the San- tiago and Chiquitos ranges (Herzog, 1910), and there are many exsiccata from the humid slopes of the Andes in northwestern Bolivia, but it has not been recorded for Peru, where, however, it seems to be replaced by a similar and little known species, S. peruviana Engler, which has been col- lected only twice (Fig. 3D) (Kunth) Benth. (Leguminosae) poses an interesting case since The species Piptadenia viridiflora it is an important element of the caatingas, fre- quently reported in the literature, whereas it is a relatively rare species in northwestern Argentina in the piedmont deciduous forests (Fig. 4A). Bur- kart (1952) cited this tree for Colombia and Ven- ezuela (presumably for the dry Caribbean area), but no exsiccata or references have been encoun- tered to confirm this assertion. Both the species and the genus seem to be missing from the Misiones nucleus of this pattern of distribution, that is, the Paraguay- Paraná system, but this may simply be due to a problem of taxonomic interpretation, since all previously considered Piptadenia species for this area have been transferred to other genera (Anadenanthera Speg., Goldmania Rose ex Mich- eli, Parapiptadenia Brenan). It is noteworthy that the species Goldmania paraguensis (Benth.) Bren- an, endemic to this sector, is very close to section Pityrocarpa, to which Burkart (1952) referred Piptadenia paraguensis (Benth.) Lindm. (— Gold- mania paraguensis), while Brenan (1955) trans- ferred Piptadenia viridiflora to his genus Pityr- ocarpa. There is only one collection of Carica querci- folia (A. St.-Hil.) Hieron. (Cariacaceae) known from the caatingas, from Irecé in Bahia (Badillo, 1971), and this would represent a very wide disjunct oc- currence from the main, “amphi-Chaco,” area of distribution of this species (Fig. 4B) if it were not for three additional isolated collections in Goiás, inas Gerais, and Sào Paulo states, which tenu- ously link the Bahian record to the second nucleus of the Pleistocenic Arc. The rest of the distribution is divided into two more or less continuous zones separated by the Chaco plains. One corresponds to the Misiones nucleus, extending eastward along the upper Uruguay basin to include the planalto forests, but also with an extension westward into the eastern Chaco. The second corresponds to the piedmont deciduous forests on the sub-Andean slopes. Carica quercifolia is also common in dry valleys between Andean ranges in northwestern Bolivia and southern Peru. The shrub Combretum leprosum Mart. (Com- bretaceae) is another typical caatinga species that follows relatively well the Pleistocenic Arc. (Fig. 908 Annals of the Missouri Botanical Garden FIGURE 3*.—A. Pterogyne nitens. —B. Ruprechtia laxiflora. — С. Celtis pubescens. — D. Schinopsis brasiliensis (8) and S. peruviana (A). 909 Prado & Gibbs Volume 80, Number 4 1993 Dry Seasonal Forests of South America "DUDILSUPIDE piu3jnoq (p — "wunsoado] umj21quio7) 7 — myofio4onb. 21477) “y — 740jfipiia тигра Y — y 381914 910 Annals of the Missouri Botanical Garden 4C). It is also common in Магапћао state in areas suspected of having links with the caatingas, and it has been collected in the River Paraguay valley and in Santa Cruz de la Sierra area, but seems to be missing from there to the south and northwest. Тће pattern followed by Pouteria gardneriana (D. C.) Radlk. (Sapotaceae) is somewhat more uncer- tain, since although it is well recorded in the caa- tingas and in the Paraguay- Paraná system, it is unclear whether it spreads into the Piedmont nu- cleus apart from one collection in Santa Cruz in Bolivia (Fig. 4D). Sterculia striata A. St.-Hil. & Naudin (Ster- culiaceae) is a rather different case (Fig. 5A), since it only marginally appears in tall, arboreal caatinga (Andrade-Lima, 1975; Ratter et al., t is much commoner in mesophilous forests within the cerrados expanse or in mesotrophic cerradào (Rat- ter et al., 1978, 1988), but it shuns the poor soils of the cerrado sens. str. It extends into the Amazon valley (Taroda, 1984) as well as to the River Par- aguay area, particularly on calcium-rich soils (Rat- ter et al., ; Prado et al., 1992), and its west- ernmost collections come from the area of Santa Cruz. The small bignoniaceous tree Tabebuia caraiba (Mart.) Burch. (synonymized with 7. aurea (Manso) Benth. & Hook. in Gentry, 1992) has an unusual distribution in that it is common in the cerrados on typically poor-quality soils, but is also relatively common in the caatingas of Pernambuco (Andrade- Lima, 1954, 1960) and Bahia, and it can be found in the Amazon valley around Santarem. This spe- cies has also been frequently collected in the Par- aguay River valley sector from Brazil to Argentina, while Ramella & Spichiger (1989) reported it for cerrado-like communities on top of the plateau in Cerro León. Tabebuia caraiba also appears oc- casionally in central Bolivia, including a specimen collected near Santa Cruz de la Sierra, thus con- forming somewhat to the Pleistocenic Arc (Fig. 5B), although the main core area of this species is the central Brazilian cerrado area. Somewhat similar distributions, with extensive occurrence in typical cerrado, are shown by Нутепаеа martiana Hayne (Leguminosae) and Machaerium acutifolium J. Vogel (Leguminosae) but both species lack the Am- azonian extension (Fig. 5C, D). Hymenaea mar- tiana spreads from the caatinga into the cerrado savannas in Goiás, the Federal District, and Minas Gerais, as well as occurring in the forests in the Paraguay River area. Machaerium acutifolium also occurs in the cerrados as well as the caatingas, and extends south into the Mata Atlántica in Rio de Janeiro, isolated campo" areas in Minas Gerais and Sào Paulo, and in forests in eastern Paraguay. This latter species has a somewhat invasive- pio- neer tendency, which may account for its more extended distribution. A number of genera show vicariad species that occupy parts of the Pleistocenic Arc. One such case is Brunfelsia L. (Solanaceae). Recently, Plow- man 9) drew attention to the fact that B. uniflora (Pohl) D. Don forms a link between the older Andean species and the eastern Brazilian ones within section Franciscea. It has been collected in northern Venezuela, in Roraima (Brazil) and neigh- boring Guyana, also in the Yungas area of the eastern-facing slopes of the Andes in southwestern Bolivia and northwestern Argentina, to then reap- pear sporadically in the caatingas and more abun- dantly in the Atlantic forests and planalto forests in eastern Brazil. Two additional species, B. pilosa Plowman and B. australis Benth., are apparentl derived from B. uniflora. Brunfelsia australis, previously treated as a subspecies of B. uniflora, fills in the southernmost gap corresponding to the Misiones nucleus in the distribution of its mother species in Paraguay and northeastern Argentina (Fig. 6A). Brunfelsia uniflora thus shows a similar but much more complete distribution pattern (Fig. 6B) to that of Commiphora leptophloeos (Mart.) Gillet (Burseraceae). e genus Patagonula L. (Boraginaceae) seems to be endemic to the Pleistocenic Arc (Fig. This genus comprises only two species: P. bahien- sis Moricand is endemic to the caatingas, whereas the type species, P. americana L., has been col- lected or reported for the forests in Paraná state in Brazil, the upper Uruguay River basin (Klein, Pire, western Argentina and southern Boliva (Cabrera, 1976). Although Patagonula has not been re- ported from other areas of the Pleistocenic Arc, it would not be surprising for it to be encountered in the future in the rather poorly studied areas of Cerro León, the Santiago and Chiquitos ranges, or south of Santa Cruz de la Sierra. The genus Loxopterigium Hook. f. (Anacardi- aceae) presents five disjunct species in South Amer- ica (Fig. 6D). The scarcely known L. gutierrezit Barkley was collected only once in the Caribbean sector of Colombia, while L. sagotii Hook. f. better known from the Guianas. The tree L. hu. 911 Prado & Gibbs Volume 80, Number 4 1993 Dry Seasonal Forests of South America "штђојутор UNILIDYID]A] а= "Рио ти DIDUJULÁ |] y) — '"DQq104D2 Dinqaqnpy Я — maus DINILIIG "Y — 46 IYADMIA Annals of the Missouri Botanical Garden t d P 4 42.9 KA d SOY 44 : Э v = | | = ex) У > 2; > - = 95 TÉ aN 7. DG | $ | id | ° EM on Y * FIGURE 6*.— A. Brunfelsia uniflora (O), and B. australis (А). —B. Commiphora leptophloeos. —C. Patagonula bahiensis (А), and P. americana (9). — D. Loxopterigium gardnerii (A), L. grisebachii (0), L. gutierrezii (B). L. huasango (A), and L. sagotii (©). Volume 80, Number 4 1993 Prado & Gibbs 913 Dry Seasonal Forests of South America asango Spruce appears only in seasonal semiarid to arid formations in coastal southwestern Ecuador and northern Peru, while the caatinga woodlands of a small area of Раш and Bahia contain the few recorded populations of L. gardneri Engl. (— Ap- terokarpos gardneri (Engl.) Rizzini). Finally, L. grisebachii Hieron. & Lorentz ex Griseb. is an important member of the sub-Andean piedmont forests in northwestern Argentina and southern Bolivia, which separates chaco vegetation from the rain and cloud forests of the Yungas. In this case, therefore, the allopatric species within a small ge- nus together comprise the familiar distribution pat- tern of this Seasonal Forests Arc in South America. Geoffroea striata (Willd.) Morong-G. spinosa Jacq. (Leguminosae) complex provides another example of the Seasonal Forests Arc (Fig. 7A). In our ex- perience, the diagnostic characters provided by Burkart (1949) fail to separate these alleged spe- cies clearly, and this complex is treated here as one polymorphic unit which, however, segregates well from the only other species of the genus, the typically but not о chaquenian G. decor- ticans (Hook. Arn.) Burkart. The G. striata— spinosa ie appears in northern Venezuela, western Ecuador and adjoining Peru, and in central Bolivia in the Yungas-like area of the Rio Grande or Pirai, for the Andean related part, whereas in the lowlands there are essentially two nuclei: the Caatingas and Misiones. А similar pattern is fol- lowed by /pomoea carnea Jacq. subsp. fistulosa (Mart. ex Choisy) Austin (Convolvulaceae), includ- ing some Caribbean islands, northern Colombia and Venezuela, the same sectors of Ecuador and Peru previously mentioned, and in riverine environments linked to the sub-Andean Piedmont forests in north- western Argentina but where the exsiccata might be of plants escaped from cultivation (O'Donell, 1959). In the east of the continent it occurs in the Caatingas and Misiones nuclei, and some odd ap- pearances in the Amazon River valley area (around Santarem and Belém) have to be taken into account (Fig. 7B). This pattern is also followed in part by Peltophorum dubium (Spring.) Taub. (Legumi- nosae), which has been recorded for seasonal for- ests in Venezuela, but is more frequent in the east of the continent, namely in the caatingas, in me- sophilous forests in Minas Gerais, Sào Paulo and Paraná, in the upper Uruguay River valley in Santa Catarina, and it has been frequently collected in eastern Paraguay. Once again this species is en- tirely absent from the chaco, only to reappear in Yungas-like areas close to Santa Cruz de la Sierra in Bolivia (Fig. similar аи. but in this case with the component elements segregated as varietal taxa, occurs in the caesalpinioid Senna spectabilis (DC.) Irwin & Barneby (Irwin & Barneby, 1982). The variety spectabilis extends from coastal Venezuela to Andean valleys in the Mérida area of the same country and the Magdalena and Cauca river valleys in Colombia, and then reappears in seasonal for- mations in Ecuador and further south in dry valleys in the Peruvian Andes (Fig. 7D). This variety is also a conspicuous member of the Yungas forests in Bolivia, and of the deciduous piedmont forests in southern Bolivia and northwestern Argentina where it is called **carnaval." As usual in this kind of distribution pattern this tree is entirely absent from the Chaco, only to reappear in eastern Par- aguay. The other taxon, variety excelsa (Schrader) Irwin & Barneby, is virtually restricted to the caa- tinga in northeast Brazil, where it is called “сап- afistula.”” The shrub to small tree Solanum granuloso- leprosum Dunal (Solanaceae) shows a similar dis- tribution (Fig. 8A). It has been collected in northern olombia and Guyana, is absent from the Amazon but reappears in the caatingas of Bahia and Minas Gerais, and further south in the mesophilous forests of Rio de Janeiro, Paraná, and Santa Catarina, and also extends into adjoining Paraguay and north- eastern Argentina, particularly in the Austro-Bra- zilian Transitional forest (Prado, 1991). As usual, it is absent from the Chaco but occurs in the eastern slopes of the Andes in northwestern Argentina and Bolivia. RESTRICTED ARC DISTRIBUTIONS So caatinga species have relatively wide- spread distributions but do not show the full se- quence of links with other seasonal formations as do most of the cases considered above. Such species can be considered as having a restricted distribu- tion, in which the third Piedmont nucleus is either lacking or very fragmentary. Several examples oc- cur in the genus Aspidosperma (Apocynaceae). One such case is А. pyrifolium Mart. (Fig. 8B), which has been reported for Cerro Leon in Para- guay (Ramella & Spichiger, 1989). It is a char- acteristic species of the caatingas that spreads ex- tensively into Minas Gerais, but then there is a disjunction with the populations on the river Par- aguay. There is also a puzzling exsiccatum from Faro (Marcondes-Ferreira, 1988), in the Pará state, 914 Annals of the Missouri Botanical Garden FIGURE 7*. — А. Geoffroea striata- G. spinosa. — B. Ipomoea carnea subsp. fistulosa. — С. Peltophorum dubium. —D. Senna spectabilis var. spectabilis (O) and var. excelsa (A). Volume 80, Number 4 1993 Prado & Gibbs Dry Seasonal Forests of South America 915 FIGURE 8*. — A. Solanum granuloso-leprosum. —B. Aspidosperma pyrifolium (O), A. riedelii subsp. riedelii (А) and subsp. oliganthum (A). —C. Aspidosperma cuspa.— D. Aspidosperma polyneuron. 916 Annals of the Missouri Botanical Garden a locality in a small sector of the Amazon valley where several other caatinga elements (Andrade- Lima, 1959) have also been collected (see further c omments below). The tree Aspidosperma riedelii üll. Arg., which has been treated by Marcondes- Ferreira (1988) as comprising two subspecies, sub- species oliganthum (W ood.) Marc.-Ferr., a Bahian endemic, and subspecies riedelii with two disjunct localities in the forests of Sào Paulo and in central Paraguay (Fig. 8B), shows a similar but more frag- mented ips Blake (Fig. 8C), w dry ids. of аат and Martinique, is also а шти cuspa (Kunth) h has been collected in the common tree in the dry Caribbean sectors of Co- lombia and Venezuela, countries in which it also extends to some dry, inter-Andean valleys. It reap- pears in southwestern Ecuador (Woodson, 1951), and at nearly the same latitude but on the opposite side of the continent in Rio Grande do Norte (north- east Brazil). A rather infrequent plant in the Caa- tingas nucleus, it is found scattered in the cerrado of central Brazil, and on rock outcroppings of the Cordillera de Altos in Paraguay (Marcondes-Fer- reira, 1988). The – mae closely related А. polyneuron Müll. Arg. (Fig. 8D) parallels in part the distribution of 4. cuspa, since it has been collected from the western half of the dry Caribbean sector and also in dry inter-Andean valleys in Co- lombia, but it has not been recorded so far from the eastern zone. No collections in Ecuador have been located, but it occurs in northern Peru in an area of similar vegetation to that of southwest Ec- uador (Weberbauer, 1936; Harling, 1979). This species also occurs in the caatingas of Bahia (north- eastern Brazil) and extends in a northeast-south- west arc through mesophilous forests in Minas Ger- ais, "spirito Santo, Rio de Janeiro, Sào Paulo, and агапа, to Paraguay and a very reduced area in north Misiones (Argentina). The valuable timber tree Balfourodendron rie- delianum (Engl.) Engl. (Rutaceae) occurs in the caatingas (Fig. 9A) and extends south to the plan- alto forests in Sao Paulo and Parana states, along the upper Uruguay River valley, and reaching east- ern Paraguay and northeastern Argentina. Rather similar, two-nuclei distributions are found in the soft-wooded tree Phytolacca dioica L. (Phytolac- caceae), Alseis floribunda Schott (Rubiaceae), and Astronium concinnum Schott (Anacardiaceae) (Fig. B-I Another group of woody species is absent from the caatingas but occurs in the Misiones and com- monly also in the Piedmont nuclei within the Pleis- tocenic Arc. These include two sapindaceous gen- era: the monotypic Diatenopteryx, with D. sorbifolia Radlk. (Fig. 10A), and the ditypic Di- plokeleba, with one species, D. floribunda N.E. Brown, which is abundant on the Paraguay- Paraná system, while the other, D. herzogii Radlk., is known only from the type collection in Cabezas, Bolivia (Fig. 10B), most likely from the deciduous forests frequent in that area (Coro, 1956). A similar distribution is found in Ziziphus oblongifolius S. Moore (Rhamnaceae) (Fig. 10C), which is a fre- quent shrub in the north of the Paraguay- Paraná system (Prado et al., 1992), and cited by Herzog (1910) for Bolivia and Escalante (1946) for Orán in Salta (Argentina). Three species that are re- stricted to the Misiones nucleus are Astronium balansae Engler (Fig. 9D), Ziziphus guaranitica Malme (Fig. 10D), and Maytenus ilicifolia (Ce- lastraceae) (Fig. 1 1A). Another taxon restricted to a single nucleus is the monotypic genus Athyana Radlk. (Sapindaceae), with А. weinmannifolia (Griseb.) Radlk., which is endemic to the piedmont forests of northwestern Argentina and southern Bolivia (Fig. 11A). The kind of distributions that is restricted to the second and third nuclei is also illustrated by the elegant rubiaceous tree Caly- cophyllum multiflorum Griseb., the dominant and characteristic species of the palo blanco forest in the piedmont of the sub-Andean chains (Fig. 1 1 B), to which it gives its vernacular name, and which is also common in the River Paraguay area and in the Cerro León (Ramella & Spichiger, 1989). Some isolated populations of C. multiflorum have been occasionally found within the Chaco sens. str. (Cas- tellanos, 1958; Morello, 1967; Morello & Adámoli, © Finally, са. 180 woody species are endemic to the Caatinga province (Prado, 1991). Examples include Ziziphus joazeiro Mart. and Z. cotinifolia Reissek (Rhamnaceae) (Fig. 10C), Fraunhoffera multiflora Mart. (Celastraceae), the congenerics Auxemma oncocalyx (Allemào) Taub. and А. gla- zioviana Taub. (Boraginaceae), Spondias tuber- osa Arruda (Anacardiaceae), Cordia leucocephala Moric. and C. dardani Taroda (Boraginaceae), and Hymenaea eriogyne Benth. (Figs. 11C, D, 12A, B) TAXA WITH MORE COMPLEX DISTRIBUTIONS It is of considerable interest that a number of caatinga species show links with the Amazonian area. The outlying records for Commiphora lep- tophloeos and Aspidosperma pyrifolium noted above are two such cases. A congeneric example Volume 80, Number 4 Prado & Gibbs 917 1993 Dry Seasonal Forests of South America Ficure 9*.— A. Balfourodendron riedelianum. —B. Phytolacca dioica. — С. Alseis floribunda. —D. А. balansae (9), Astronium concinnum (А). c Ф E © ©) T о S o = 5 a a os 5 23 o со с := < > 918 "nognunpapng гпуд 217 `@— (wv) snofiZuojqo `7 (W) ојодиуиоз -7 ЧФ) опәтъо[ snydiz17 *)—(w) 1302194 а (e) »punquoy vqa]230d17 "Я — v1]0f1q20s хАлојаоио ти “Y —',0] INNIA 919 Prado & Gibbs Volume 80, Number 4 1993 Dry Seasonal Forests of South America pue (Ф) x4j»202uo ошшәхту (W) voinu 42ffouunpa4 у — штаоју ти ungAgydooXm:) y — (+ ) `р$оләфт] spipuodg *(1— (w) ри111012218 p ођоћииошилот пирАутру (еф) vipofioii snuo140jy ^V — [T 380514 920 Annals of the Missouri Botanical Garden is Aspidosperma discolor A. DC., which occurs in Venezuelan Guayana and the Guianas, and which has been profusely collected in the lower and middle Amazon River valley and in Rondónia state, but it is also frequent in Bahian caatinga and in a central area of the cerrados (Fig. 12C). Another case is Albizia polyantha (A. Spreng.) Lewis (Legumi- nosae), which extends from the Guianas and Ilha de Maracá (Lewis & Owen, 1989), to the Amazon River valley and tributaries, and to the caatingas, the Paraguay- Paraná system and Cerro León (Fig. 12D). However, in a different category are species such as Couepia uiti (Mart. enth. (Chrysobalanaceae), which are essentially Amazo- nian elements, which extend via gallery forests into Zucc. the cerrados and caatingas (Fig. 1 2B). A number of species have more indistinct dis- tribution patterns that include Amazonian localities. This sort of distribution is shown by a few caatinga species, although in each case they show fairly wide tolerance and also occur in more humid hab- itats. One such is Coutarea hexandra (Jacq.) K. Schum. (Rubiaceae), which also occurs along the Amazon River (Fig. 13A), and Crateva tapia L. (Capparaceae), which has been collected in the Amazon basin and also along the ephemeral wa- tercourses in the caatingas (Fig. 13B). Hymenaea courbaril L. var. courbaril (Fig. 13C) and outstanding caatinga tree Tabebuia impetiginosa (Mart. ex DC.) Standley (Fig. 13D) h distributions. Some such taxa are primarily re- ave similar stricted to seasonal habitats, such as Myroxylon balsamum (L.) Harms (Leguminosae) and Poep- pigia procera C. Presl. (Leguminosae), although these species also appear in the Amazon (Fig. 14A, B), or as in the case of Phyllostylon brasiliense Capanema (Ulmaceae), in the Brazilian Atlantic rainforest (see distribution of the genus in Fig. 14C) This distributional pattern is also followed (Fig. 14D) by another two-species complex: Cordia al- liodora (Ruiz & Pav.) Oken- C. trichotoma (Vell.) Arrab. ex Steud. (Boraginaceae). Gibbs & Taroda (1983) studied this complex and concluded that both taxa should remain separate because of the different kinds of heterostylous conditions shown by plants of each species, together with differences in flower size, pollen grain size, and myrmecophily. However, these taxa have had a confusing history and have been treated as conspecific by some au- thors, while others have attempted to differentiate between them based on morphological characters such as leaf indumentum. This complex seems to show a high degree of ecological tolerance, with C. alliodora in particular showing fast-growing pi- oneer-type qualities, which may account for some aspects of its distribution. It extends from the semi- arid Caribbean coast of Colombia and Venezuela, and the dry inter-Andean valleys of these countries, but also occurs in the more humid Orinoco River valley and even in the Amazon. From western coastal Ecuador and Peru, the distribution follows the more characteristic dry valley systems to reach the Yungas area in Bolivia. The specimens сој- lected in northwestern Argentina in the sub-Andean Piedmont forests have been traditionally attributed to C. trichotoma, but from Gibbs & Taroda's (1983, fig. 1) point of view they must be regarded as C. alliodora, based both on morphology and geo- graphical range. The specimens in the eastern half of the complex distribution are nearly all considered C. trichotoma, which thus extends from the caa- fingas, where this species can appear not only in the “sertão” but also in the wetter hills (“brejos””), to mesophilous forests on the Brazilian planalto (Minas Gerais, Sào Paulo, Paraná, and Santa Ca- tarina), eastern Paraguay and northeastern Argen- tina. Some species, such as Platypodium elegans Vogel (Leguminosae), that are more typical of the central Brazilian cerrado vegetation seem to par- tially shadow this distributional pattern (Fig. 15A). This species is a canopy tree in seasonal forest in Panama (Porter, 1973), and it is also present in Venezuela, but it then disappears in the whole of north and central South America to reappear in the caatingas, the cerrado in Goiás, Mato Grosso, Minas Gerais, and Sào Paulo, and further extends into eastern Paraguay and eastern Bolivia. Like- wise, the anacardiaceous tree Astronium fraxini- folium Schott (Fig. 15B) is found in northern Co- lombia and Venezuela, and although it is absent from the caatingas sens. str., it has been recorded in semideciduous communities on higher ground within the general caatinga area. In central Brazil, it has been considered one of the indicator species of mesotrophic cerradào by Ratter et al. ( and it also extends to eastern Paraguay and central Bolivia. There is a surprising collection of this spe- cies in the Canadon Platanillos, in central Para- guayan chaco, which must undoubtedly be linked to one of the watercourses that cross this region in a general west-east direction. Finally, it should be emphasized that the distri- butions of the only three woody species (Parkin- sonia aculeata L., Sideroxylon obtusifolium (Roem. & Schult.) Pennington, and Ximenia amer- icana L.) that are common to the chaco and caa- tingas are meaningless in this context, since all three species have widespread distributions and 921 Prado & Gibbs Volume 80, Number 4 1993 Dry Seasonal Forests of South America ‘pyjupdjod 11219] а — 40]09s1p Dutsadsopidsp * ) — (w) mn »idono?) (Ф) 24430113 DIDUJUÁ HI `Ч— (v) 1UDP1DP `) pue (8) »jpydo»oono] трао "y — c | 390914 Missouri Botanical Garden Annals of the 922 "»souigijodun vimqoqp “(Y — (w) 0010209 "тел pue (Y) 114094002 стел [14004002 роридшАн 7D) — тлаој падао) “Y — олрирхоц D210]n07) "V —',£]| 381514 - \ \ cu. i \ 2 |o ES \ A A 9 ЕГ] m () E - р у << ee — З 7 í С Y o 1422 ai Nu; | ) | “a > e». e - | > ~ dH. Vy yy У, Jl | ү p / ~ BAS D Volume 80, Number 4 1993 Prado & Gibbs Dry Seasonal Forests of South America 923 FIGURE 14*.— А. Myroxylon balsamum. —B. Poeppigia procera. —C. Phyllostylon brasiliense (А), P. orthopterum (0), and P. rhamnoides (). — D. Cordia alliodora (6) and C. trichotoma (A). Annals of the Missouri Botanical Garden FUERON vr у; — А. Platypodium elegans. —B. Astronium fraxinifolium. —C. Parkinsonia aculeata. —D. Sideroxylon obtusifolium subsp. obtusifolium (O) and subsp. FIGURE 15*. buxifolium (A). Volume 80, Number 4 1993 Prado & Gibbs 925 Dry Seasonal Forests of South America occur in diverse vegetation types. Parkinsonia aculeata (Leguminosae) is a very widespread ru- deral plant whose ecological amplitude includes the Pacific coastal desert of Peru and the arid Monte province of Argentina (Fig. 15C). Sideroxylon ob- tusifolium (Sapotaceae) shows some fidelity to the seasonal formations, but together with the other two species it also occurs in temperate areas of South America (Fig. 15D). These species are also exceptional in that in addition to the chaco, they spread into areas such as Monte, or even Patagonia in the case of Ximenia americana (Olacaceae) (see map in Sleumer, 1984). CONCLUSIONS Тће 80 or so taxa mapped in this analysis are all woody components of a series of dry seasonal woodland formations that occur in South America. This is a vegetation that seems to be mainly ге- stricted to relatively mineral-rich soils. The pres- ent-day distributions of these species are considered to comprise fragme f the once extensive forests that probably characterized the ntary remnants o dry climatic maxima of the Pleistocene, and which are likely at that time to have extended into the Amazonian region, and also to have percolated into the Andean region, perhaps via the “Maranon gap,” and established further links with the Caribbean- Guajira province, through northwestern Peru and eastern Ecuador The extant fragments of this seasonal woodland vegetation have their main nuclei in the Caatingas, Misiones, and Piedmont areas. Effectively, the woody and succulent flora of the caatingas com- prises ca. 437 species, of which 228 species are certainly nonendemics (Prado, 1991). Of these, some 62 species (27%) show floristic affinities to the Pleistocenic Arc type distribution, whilst no other coherent distribution patterns were evident for the remaining nonendemic species. Equally no- table, some 51 species (19.475) of the ca. 263 woody and succulent species found in the piedmont forests of northwestern Argentina and southern Bolivia are also found in tbe caatingas, while a further 42 (16%) piedmont species show links with the Misiones nucleus and calcareous woodlands peripheral to the Brazilian cerrados (Prado, 1991). tharacteristically, these species distributions present a lacuna in the area of the chaco, and only sparse representation in the cerrado, with most seemingly cerrado representatives of this group actually restricted to calcareous outcrops. Indeed, the essentially distinct nature of typical Brazilian cerrado vegetation from other seasonal, semide- ciduous woodlands is striking, in sharp contrast to the clear floristic links between such seasonal wood- land and the rather distinctive vegetation of the calcareous islands within the cerrado general area, as first emphasized by Ratter et al. (1978). The caatinga-type seasonal woodlands also show intriguing links with some areas in the Amazonian, as well as the Caribbean and Venezuelan- Central American, regions, and with some species having isolated populations es in dry valleys of the east slopes of the . It is unfortunate that, probably because of ben occurrence on agricul- turally desirable, good-quality soils, these seasonal forests in South America have suffered considerable devastation in the past, and continue to bear the brunt of agricultural expansion in the region. This applies particularly in the case of the piedmont woodlands of northwestern Argentina, and to the isolated areas of **mata calcárea” on base-rich out- crops scattered within the Brazilian cerrado area. A coherent conservation policy for selected areas of these seasonal forests areas is urgently needed. LITERATURE CITED AB'SÁBER, А. М. 1977. Espaços ос па pela hara sao dos climas secos na América do Sul, por siào dos го: ери“ d quaternários. Ракіта Prem . de Geografia 3: —— pon paleoclimate and pom m of Brazilian Amazonia. Pp. 41-59 in G Prance (editor), Biological ni aia in the Tropics. Ple- num wer à Assy, М. L. the jae чел, Pp. 67 Biological Diversification in s Tropics. Plenum Press, Paulo, Quaternary зе dn studies in 73 in С. T. Prance (editor), ork. J., R. Neuman, А. D. RATIER DE COLLINA & O. “197 El Chaco aluvional salteño (Como INTA- Bas de Salta). we sta de Inv est. Agropec., Ser. 3, Clima y Suelo 9: 165-23 ALTSCHUL, S. v. R. A taxonomic и el fe genus pu аа. Contr. Gray Herb. 193: 1- 65. New ADAMOLI, ANDRADE-LiMA, D. DE 1954. Contribution to the Study of the Flora of Pernambuco, Brazil. Monografias. Universidade Rural de Pernambuco, Recife. 1 Viagem aos campos de Monte Alegre, Pus. Comtribuicio po o conhecimento de sua flora. Bol. Bee Inst. Agron. N. 36: 99-149 Pub fitogeograficos | Pernambu- co. ^ Pee а Agr оп. 5 lg РАДНА.) де АА ея Bol. аш inst: Pesq. Agron. Per- nambuco, N.S. 8: 3- 9 em of the Rio Grande Basin Bahia, Brazil. Preliminary note. Revista Brasil. Biol. 35: 223-232 1981 Yl Caatingas dominium. Revista Bra- sil. Bot. 4: 149-16 1982. Present- yd forest refuges in north- Mm Brazil. Pp. 245-251 in G. T. Prance (editor), Biological о in us Tropics. Plenum Press, 926 Annals of the Missouri Botanical Garden BADILLO, lá M. Monografia de la Familia Cari- caceae. Asociación de Profesores de la Universidad Gana (Venezuela), Maracay. BENSON, W. 1982. Alternative models for infra- generic diversification in the humid tropics: Tests with passion vine butterflies. Pp. 608-640 in C. T. Prance (editor), Biological Diversification in the Trop- ics. Plenum Press, New York. BIGARELLA, J. 1, D. DE ANDRADE-LiMA & P. J. RIEHS. 1975. Considerações a respeito das mudangas pa- leoambientais na distribuição de algumas espécies ve- getais e animais no Brasil. Anais Acad. Brasil. Ci. 47: 411-464. 1955. Notes on Mimosoideae: I. Kew Bull. 2: 161-192. . N. AB'SÁBER. 1979. Ice-age refuges and evolution in the Neotropics: Correlation of pa- leoclimatological, geomorphological and pedological ata with modern biological endemism. Paleoclimas, Univ. São Paulo, Inst. de Geografia 5: 1-30. BUCHER, ЈЕ. Н. 1982. Chaco and Caatinga. South wid ic id arid oes woodlands and thickets. Pp. 4 9 in B. J. Huntley & B. H. Walker (editors), ил a of Tropical e Springer- Verlag, Berlin. ашк, А. La posición sistemática del **chan- as especies del género d (Legumino- sae- Dalberg eae). Darwiniana 9: 9 952. as Leguminosas oie carte у C salas 2nd ed., ACME Agency S. R. L., DINE A. L 9 Regiones Fitogeográficas Argen- tinas. 2nd ed. pe ie Argentina de Agricultura y Jardinería. ACME 5. A. С. 1., Buenos Aires. 80. Biogeografia de América Latina. ?nd ed. S Serie de Biologia. Secretaria General de la тне м de los Estados Americanos, Wash- ington, D.C. CASTELLANOS, A. 1958. tación del occidente de Formosa. Bol. Ci. 40: 229-263. Cocucci, A. E. 1961. Revision del pui Ruprechtia (Polygonaceae). Kurtziana 1 -269. COLINVAUX, P. A The oan and future Amazon. Sci. Amer. 1989 (May): 102-108. Cono, M. 1956. Distrito forestal pa Yacuiba. Descrip- ción de su área correspondiente a clima, suelo y vegetación. Bol. Forest. (La Paz) 1 12. DAMUTH, J. E. € R. W. FAIRBRIDGE. 1970. Equatorial Atlantic deep-sea arkosic sands and Ice-Age aridity in tropical South America. Bull. Geol. Soc. Amer. 189-206. Observaciones dei la vege- cad. Nac. DiciLIo, A. P. L. & P. R. LEGNAME. 1966. Los árboles indigenas de la provincia de Tucumán. Opera Lilloana : 1-42. . Contribuigao ao estudo da caatinga pernambucana. Revista Brasil. Geogr. 13: 577-590. EITEN, С. 1972. The cerrado vegetation of Brazil. Bot. ev. (Lancaster) Е K 1-341. ENGLER, 'chinopsis COE Р. Мили (editor), ns Brasiliensis xil, ОП: A04— 405. ema pP hen dp G. ESCALA s oe argentinas. Bo mg ЧАК ва 1: 209-231 ESKUCHE, d 1986. Bericht über die 17- Internationale sche Exkursion durch Nordargen- : d Geobot. Inst. ETH Stiftung Rübel Zürich 91: -117 EssER, G. 1982. Vegetationsgliederung und Kakteen- vegetation von Paragua ropische und subtro- pische in dran v uma Wiss. Mainz, Abh. Math.- 933. Pus fitogeográfico sobre el Chaco я real. Revista Jard. Bot. Mus. Hist. Nat. Paraguay : 1-87 GAL do R. Introdugao ao conhecimento da área iier ira abrangida pelo plano de valorizagáo onómica da Amazonia. Revista Brasil. Geogr. 17: 299. GENTRY, A. 1992. Вірпопіасеае — part II (Tribe Te- comeae). Flora Neotropica Monograph по. 25 (II). GipBs, P. E. & N. Taropa. 1983. Heterostyly in the Cordia alliodora—C. tric oe complex in Brazil. Revista Brasil. Bot. 6: 1- HAFFER, J. vian Ds УРИ of the era he lowlands. Ornithological Monographs 36: 14 НАММЕМ, T. УАМ DER. 1974. The Pleistocene change of diei and ч in tropical South Ma J. Biog ——— 082. 'aleeclogy of tropical South Amer- ica. Pp. 60-66 in G. T. Prance (editor), Biological Diversification in Fut Tropics. Plenum Press, New ork. Навике, С. Тће Ms reine types of Ecuador. A brief survey. Pp. 165- in K. Larsen & L. B. Holm-Nielsen fu bas "Tropical Botany. Academic Press, London. HAUMAN, L. 1931. Esquisse AUi cis at ai de l'Ar- gentine subtropical et de ons avec la Géo- botanique sudaméricaine. Bull Soc. Roy. Bot. Bel- gique 64: 20-64. ERZOG, T. 1910. Pflanzenflormationen Ost-Bolivias. Bot zu Syst. 44: 346-405. 912. Formaciones Ad del este de Bo- livia. ва Soc. кө az 10: 159-190 Irwin, Н. S. & R. C. BARNEBY. 1982. The American Cassiinae: a hem revision of Leguminosae tribe Cassieae subtribe Casstir nae in the New World. Mem w York Bot. Gard. : Kam-BIU, L. & P. A nas 1985. Forest changes in the Amazon. vi during the last glacial maximum. Nature 318: KLEIN, R. M. : Ar vores PE i Wien sub- tropical ul Alto Uruguai. Sellow 62. LEwis, G . Legumes of m Royal Botanic Gardens. Ko Whitstable, Ken — & P. E. Owen. Legumes of the Ilha de Maracá. Royal Botanic Gardens Kew, Whitstable, Kent. Lewis, J. P. К. РївЕ. 1981. Resena sobre la vegetacion del Chaco santafesino. Serie Fitogeográf- ica No 18. Instituto Nacional de Tecnologia Agro- pecuaria, Buenos Aires. Litto, M. 1919. Reseña fitogeográfica de la provincia d aoe an. Primera pia Soc. Argent. Ci. ucuman 1916: 210-2 eee dae ue W. 1988. po A Mart., nom. (Apocynaceae): Estudos Taxonómicos. Thesis, Universidade Estadual de Campinas, Brazil. MARKGRAF, 9 Palaeoclimates in Central and South America since 18,000 BP based on апа lake-level records. Quatern. 2 8: 1-24. MARTINS, F. R. 1979. О Мао Pn antes e a Fitossociologia de uma nos d do Interior = Volume 80, Number 4 1993 Prado & Gibbs 927 Dry Seasonal Forests of South America en Estado de Sao Paulo: Parque Estadual de Vas- unga. Thesis. Universidade de Sao Paulo, Brasil. ese: C. E. Ruprechtia. In: C. F. P. Martius eiae Flora Brasiliensis v, I: 56. Leipzig, Munich, Vie MEYER, T. & F. A. BARKLEY. 1973. Schinopsis Ves eS ME 33: 207- 258. MonELLO, J. 1967. Bas el estudio fitoecológico de los grandes espacios (el „н argentino). Ci. & Invest. 23: 252-267 Vegetación 7 ambiente st Botánicas Argentinas. queno, A. Cn Mimeographed. | ———— nia. Benitez, September 1967. Las grandes unidades de vegetacion y ambiente del Chaco argentino. Segun a о MuELLER-DoMBors, D. & Н. EL and Methods of Vegetation Ecology. J. Wiley & Sons, or B. Bu. C. A. FERREIRA, M. F. ра ЗИМА & M. ASAKI. 1990. Endemism centres, refugia and пас collection density in Brazilian Amazo- 14. NELSON, L. O'Do ‚ С. A. 1959. Convolvuláceas Argentinas, I. Lilloa 29: 87-348. Pinto, С. C. P. Schinopsis кен шы: n na comunidade floristica do Panta nais XXXIII Congr. Nac. Soc. Bot Maceo. му ee il: 49- PLowMAN, T. 1979. The genus Brunfelsia: А а spectus of the taxonomy and piogeography: Pp.4 491 in Hawkes, R. N. Lester & A. D. o ng (editors), The Biology and Taxonomy of the Sola- naceae. Academic Press, London. 19 The vegetation of Panama: a 67-201 in A. Graham (editor), e tation and Vegetational History of Northern Latin merica. Elsevier Scientific, Amsterdam. 91. A Critical ad of the Floristic atingas Vegetation in South America Thesis. шне. of St. Andrews, Scotlar . FRANcEsCHI & M. B. BiancHI. 1989. El bosque del escarpe occidental del Rio Parana (Argentina). Composicion floristica y estructura. Acta Bot. Brasil. 3(2): 99-108 ey GIBBS, A. POTT V. J. Porr. 1992. Chaco-Pantanal ners in Southern Mato Grosso ir Pp. 451-4 n P. A. Furley, J. Proctor & Ratter ue Nature & Dynamics of йыл н Boundaries. Chapman & Hall, Lon- don. Prance, С. T. 1973. Phytogeographical support for the theory of Pleistocene forest refuges in the Amazon basin, based on evidence from distribution patterns in Caryocaraceae, Chrysobalanaceae, Dic shapetale- ceae and Lecythidaceae. Acta Amazon. : 28. . 1982. [Editor.] Biological OE HE in the Tropics. Plenum Press, Yor & С. B. SCHALLER. 1982. Preliminary study of some vegetation types of the Pantanal, Mato Gros- Sl SPICHIGER. 1989. preliminar e medio fisico y Chaco Boreal. Contribución al estudio de la "oa del Chaco. I. Candollea 44: 639- In nterpretación . ASKEW, К. MONTGOMERY & D R, J. A, С GIFFORD. oa 303-316 in M. G. Ferri ta IV Simposio Sole o Cerrado Editora USP, Sào Paulo. —— a. Obser- vati f trophic soils in central Brazil. ни не Bie 1: 47- , €. N. DA | Bai & M. HARIDASAN. 1988. Occatio e мн уере- tation types in the Pantanal an mbá, Brazil. Gard. Edinburgh s 503-525. босон Н. J. J. Меке. 1986. Los bosques fluviales del Chaco oriental: Primera aproximación. 22 ОЕ. Гос. 1951. Amphibians and Ho opkins- Branner expedition to Brazil. ieldia TS 439-465. SLEUMER, H. 984. Olacaceae. Flora Neotropica Mon о. по. Ө SORENSEN, T. 1948. A method of establishing groups of equal amplitude i in plant sociology based on sim- ilarity of species co py Kongel. Danske Vidensk.- Selsk. Biol. Skr. 5: 1-34. TARODA, N. 1984. s revision of the Brazilian species of € ulia L. Notes Roy. Bot. Gard. Edinburgh 42: -149. M L. 1967. Formaciones forestales y mad eras del Paraguay. Biol. Inst. Forest. Latinoamer. Invest -34 VANZOLINI, P. E. 1963. Problemas faunisticos do cer ado. Pp. 307-320 in M. C. Ferri (editor), Sampii Sôbre o Cerrado. Dicen of Sào Paulo Press, Sào Paulo. 1974. Ecological and geographical distribu- tion of lizards in Pernambuco, northeastern Brazil (Sauria). Papéis Avulsos do Departamento de Zool- ogia, Secretaria Е "d ша Industria e Comercio (Sào Paulo) 28: 61-90. mar A. 19 936. Phytogeography of к уе ian Andes. Field Mus. Nat. Hist., Bot. Ser. 13: 13- 1987. Clustan User Manual. Cluster anal- sampating laboratory, WISHART, D. et Zucc. Ann. Missouri Bot. Gard. 38: 119-204 DIVERSITY PATTERNS AND BIOMASS OF EPIPHYTIC BRYOPHYTES AND LICHENS ALONG AN ALTITUDINAL GRADIENT IN THE NORTHERN ANDES! Jan H. D. Wolf? ABSTRACT The difficult accessibility of the canopy of tropical rainforests is probably one of the main reasons why information about the nonvascular tree crown flora is scarce. In this Sip of 59 bask type selected cano »y trees, divided over 15 sites о 4,130 m in the Central Cordillera, Colombia. Species-area curves indicated that sampling at each investigated site (altitude) was adequate within the restrictions imposed by the sampling method. Ordination analysis revealed that altitude as a complex ecological fact along the gradient its peak of approximately 100 taxa per altitu of individual taxa showed that the belt of greatest liverwort ri species overlap) predict highest species diversity at such a on the assumption that community composition results from ses (mass effect, ecological equivalency, and All are based hypothe transition zone. indicated by are НА у Species turnover app the low number of liverwort taxa (20 ears greater for br hytes on this tropical mountain than on temperate xplained most of the variation in the data. Alpha and beta diversity patterns were different ine mosses, liverworts, and macrolichens. As to liverworts, species richness reached de at a mid-altirudinal range from а m to 3,190 m . The distribution ontact transition zone. Several in North America. The general increase in biomass of nonvascular canopy subsamples with altitude coincides with a rise in humidity The study of species diversity (species richness) is a central theme in both theoretical and applied ecology. Following Whittaker (1977), point diver- sity is defined as the species richness of a single subsample taken on a host tree and alpha diversity as the number of species of all the subsamples taken at four host trees per altitude. Beta diversity may be defined as the extent of species replacement along an environment gradient (MacArthur, 1965; Whittaker, 1960, 1972, 1977). Origin and maintenance of local diversity may be explained in two ways: as a result from (1) processes (niche relations among species) within the community (MacArthur, 1965; Whittaker, 1977; Connell, 1978; Huston, 1979), or from (2), at least in part, propagule influxes from neighboring sites (Shmida & Wilson, 1985). Slack (1977) and Watson (1980) suggested that in bryophyte com- munities species diversity results more from suc- cessful colonization and establishment than from competition-related interactions. It is generally accepted that tropical regions are richer in species than temperate areas (e.g., Pian- ka, 1966), but documentation of diversity patterns within the tropics is still rare (Centry, 1982), in particular for nonvascular plants (Maury-Lechon et al., 1984). Species richness studies of various organisms along altitudinal gradients in the tropics have demonstrated three major patterns: (1) A decrease of species richness with altitude for woody ' The assistance with identification is gratefully EE ie of B. H. Allen, A. аи . Vi rahm, D. Griffin III, R. Grolle, R. Mob dar add the identification le numerous collections is forw Gradstein (liverworts), and provided by A. M. Cle eviewer for comments on earlier versions of the the late H. Inoue (Plagio hila). ef and T. van der Hammen. I thank manuscript. I thank the staff of the Instituto de Ciencias Naturales . P. Churchill, J-P. S. R. Gradstein, M. J. A (COL), Bogotá, for cooperation while visiting sig research facility. Finally, this study would not have been possible without t ess field assistance of J. Klor . The s tudy was supported by grant W84-236 of the Netherlands h ? University of Amsterdam, Hugo de Vries-Laboratory, Kruislaan 318, 1098 SM Amsterdam, The Netherlands. ANN. Missouni Bor. Сакр. 80: 928-960. 1993. Volume 80, Number 4 1993 Wolf 929 Diversity Patterns of Epiphytic Bryophytes and Lichens altitude (m) iL west | « 4000 - | | Santa Rosa Marsella aba 2000 4 auca | Sag ~ S SSS bal B Wr AX, Ane Páramo de Santa Коза ———>. ~ x ~ ROS tree line SS 4, “ny, ~ ae de Termales S (2 tax 10 20 FIGURE 1. phanerogams (Gentry, 1988), herpetofauna of leaf litter (Scott, 1976; Fauth et al., 1989), and anuran amphibians (Duellman, 1988). (2) A maximum of species richness at higher altitudes for mosses (Gradstein & Frahm, 1987; Frahm, 1987a), liv- erworts (Gradstein & Weber, 1982), bryophytes (Gradstein et al., 1989), and lichens (Sipman, 1989). The latter two studies were conducted along the transect treated in this study and included both terrestrial and epiphytic (mainly tree base) sam- pling. (3 maximum of species richness at inter- mediate altitudes for birds (Terborgh, 1977), in- sects (Janzen, 1973), vascular epiphytes (Sugden & Robins, 1979, and assumed by Gentry & Dod- son, 1987), bryophytes (Enroth, 1990), and liv- erworts (Wolf, 1989). Gradstein & Pocs (1989) also documented higher bryophyte (excluding epi- phylls) species numbers at intermediate (1,500– 2,600 m) altitudes in the Sierra Nevada de Santa Marta. Highest cover values of both terrestrial and epiphytic (1—3 m) bryophytes were found at Santa Marta at altitudes over 2,600 m (van Reenen & Gradstein, 1983). T iphytes with altitude is a common feature of tropical mountain areas worldwide (Grubb, 1977; Grubb et al., 1963; Pócs, 1980; Frahm, 1987b), and it has been suggested that it is paralleled with an increase in species richness (Richards, 1984, 1988). Beta diversity, which focuses on the qualitative differences between habitats, has received much less attention than alpha diversity. As far as 1 know, he increase of nonvascular ep- only two studies on beta diversity of cryptograms along an altitudinal gradient are available, bot concerning bryophytes in temperate North Amer- ica (Slack, 1977; Lee & La Roi, 1979). Other — km 60 1 ы Т 30 40 50 Longitudinal section of the west slope of the TPN-transect; redrawn from Thouret (1983). studies (Robinson et al., 1989; Oksanen, 1983) addressed the difference in response of bryophytes and lichens along other than altitudinal gradients. The aim of this paper is the recording and as- sessment of species richness, distribution, and bio- mass of epiphytic bryophytes and lichens along an altitudinal transect in the northern Andes. The possible role of interactions between sites within the altitudinal macro-gradient on epiphytic species richness will be examined. For the first time in tropical mountains full attention was paid to the flora of the forest canopy. STUDY AREA The study was carried out along an altitudinal transect (1,000—4,130 m) on the western slope of the Colombian Central Cordillera (ca. 4%50'N, 75?30'W) in the watershed of the Rio Otün. This transect is known from previous ecological studies as the ““Transecto Parque Los Nevados” (van der Hammen et al., 1983, 1989; Veneklaas, 1990; Vis, 1989). Data on the bryophyte and lichen flora of the study area are provided by van Reenen (1983), Gradstein et al. (1989), and Sipman (1989). The mountain slope is not steep from the Rio Cauca valley floor at 1,000 m up to Termales at ca. 2,000 m altitude (Fig. 1). Above, there is a sudden increase in slope. The upper tree line is found at 3,700 m, but locally at higher elevations a dwarf forest (6-8 m) dominated by Polylepis sericea Wedd. and Diplostephium violaceum Cuatrec. may be present. The watershed exhibits a bimodal yearly distri- bution of rainfall, coinciding with the passing of 930 Annals of the Missouri Botanical Garden rainfall (mm) 4000 — 2000 4 1000 4 0 m T T T 1 altitude (m) 0 1000 2000 3000 4000 5000 FIGURE 2. Annual rainfall distribution per altitude, after van de Weert et al. (1986). the Intertropical Convergence Zone (Pérez, 1983). Rainfall data for a period of over 20 years from nearby meteorological stations, Matecana Airport (1,342 m) and El Cedral (2,120 m), show that local rainfall exceeds 100 mm/ month, even during the “dry” season (van der Weert et al., 1986). Differences in precipitation and air humidity data from meteorological stations in the area seem to be largely correlated with differences in altitude. Based on climatic data from more than 40 stations at altitudes of 1,020–4,500 m, van de Weert et al. (1986) found a linear decrease of mean daily temperature іп the watershed with altitude (0.63°C/ 100 m) and an annual precipitation maximum at around 2,000 m (Fig. 2). METHODS Fieldwork was carried out from August 1985 to December 1986 and during the second half of 1988 Diurnal variation of temperature and relative humidity at different altitudes (1,725 m, 2,150 m, 2,460 m, 2,970 m, 3,370 m, and 3,670 m) was recorded by Lambrecht thermohygrographs hung in white-painted, wooden, collapsible housing that was open at the bottom and had slots at the sides, allowing good aeration but avoiding sunlight at the measuring equipment. The housing was placed out- side the forest at 2 m above ground level. At 3,370 m, where a clearing was not available, the ther- mohygrograph was placed about 1 m above the tree canopy hanging from an upright pole tied in the tree crown. Аз hair extension in the hygrograph is maximal at 95% relative humidity, humidity values higher than 95% c not be recorded. Even though time of registration was limited to 2-3 weeks per site, the small seasonal variations in temperature and relative humidity (van der Weert et al., 1986) permitted a good impression of diurnal patterns. Hemispherical photographs (167°) were taken with a superwide angle lens mounted on the front of the 24-mm lens of a standard 35-mm camera, to determine light conditions within the forest. Two perpendicular levels attached to the camera body permitted precise vertical canopy photographs. Dif- fuse site factors (Anderson, 1964), which provide a reasonably accurate estimate of the mean per- centage of diffuse light passing through the canopy, were determined from the photographs using the method described by Anderson (1964). The amount of direct sunlight at the sites was not estimated, since data on the variation in periods of sunshine along the transect were not available. SAMPLING At altitudinal intervals of ca. 200 m four trees were selected for investigation. Selected trees had to be (1) full-grown with their crowns reaching the average tree crown level (canopy); (2) of a common species in the forest; and (3) without a scaling, very 4,130 m an exception was made for dominant Polylepis trees, which have a flaky bark). Additionally, they were not to be located near the edge of the forest, near a stream, on very steep terrain, on a ridge or in a gulley, or next to a previously sampled tree. hard, very smooth, or very coarse bark (at Taxonomical affinity of host trees was not consid- ered. Host-preference of epiphytes is often re- garded of minor importance in wet montane trop- ical forests (e.g., Richards, 1984; Smith, 1982). An effort was made to locate forests where dis- turbance was minimal. At 1, m it was not possible to find forests not disturbed by humans, but also at higher altitudes (e.g., at 2,130 and 2,460 m) much forest seemed disturbed, as evi- denced by fast-growing “‘weedy”’ tree species like Brunellia occidentalis Cuatrec., Solanum inopin- um Ewan, and by treeferns (Cyatheaceae). Since in these higher forests disturbance by humans was not apparent, these trees appear to form a natural element of the forest, perhaps due to the frequent occurrence of landslides (pers. obs.). Epiphytic, excluding epiphyllous, vegetation was s (only three trees at 1,000 m). 10 subsamples were taken at sampled on 59 tree On each tree ca. habitats such as tree base, trunk, inner crown, middle crown, and the outer branches, wherever a clear change in structure or in species quality of the epiphytic vegetation occurred along the vertical within-tree gradient. Subsamples taken on bran- chlets in the outer tree crown covered a smaller Volume 80, Number 4 1993 Wolf Diversity Patterns of Epiphytic Bryophytes and Lichens area than on tree bases; subsample area varied from a few to over 20 dm?. Branches were sampled above and aside only, epiphytes growing in the rain shadow underneath were excluded. Prior to re- moval from the bark of each subsample as a whole, information was gathered about the location, plot size, and percent cover of the dominant species relative to the area of the upper half-cylinder of the branch segment. The epiphyte communities are described in detail in Wolf (1993a, b). A first impression of the species composition of 596 subsamples taken was obtained in the field laboratory in order not to miss tiny bryophyte species that are easily overlooked when dry. Sub- sequently, organic debris was separated and all subsamples were oven-dried at 70°C, weighed (ex- cluding phanerogams), and sent to the laboratory in the Netherlands, where they were reexamined for hitherto not-found species and to check the field identifications. Biomass in this study is defined as the (dry) weight of living plants, excluding ac- cumulated suspended organic soil. Parts of plants were assumed to be living whenever they were recognizable as plant structures and included brownish bryophyte bases of tall turf-forming spe- cies. Some 1,600 botanical collections were deposited in the Herbarium of the Instituto de Ciencias Na- turales (COL), Bogotá, with duplicates in the Her- barium of the University of Utrecht (U). Some duplicates were sent to specialists for identification. Each tree was climbed using slightly modified rope climbing techniques, as described by Perry (19782) and ter Steege & Cornelissen (1988). With a hunting bow, an arrow with a nylon line attached was shot through the tree crown and the tree was climbed with jumars along a mountaineering rope. SPECIES IDENTIFICATION Species identification is problematical for groups of organisms that have not been adequately treated systematically. For the lichens this concerns in particular the genera Sticta, Lobaria, Leptogium, and Usnea and for bryophytes some members of the Lejeuneaceae (subfamily Lejeuneoideae), Pla- giochilaceae, and Bryaceae. Some unidentified spe- cies were given provisional names (e.g., Lejeunea sp. А. Lejeunea sp. B) and included as such in the analysis; others could not be recognized as individual species and were grouped in higher tax- onomic ranks (e.g., Sticta spp.; Lejeuneaceae). Furthermore, species were grouped when they could only be differentiated from their close allies in a fertile state, like most sterile crustose lichens, but also some bryophytes, e.g., Omphalanthus filifor- mis (Sw.) Nees and O. platycoleus Herz (Gradstein 1981). Finally, two pleurocarpous moss spe- and Mitteno- thamnium reptans (Hedw.) Card, and all species et al., cies, Ctenidium malacodes Mitt. of the crustose lichen family Graphidaceae were lumped, because they could only be separated after detailed microscopic investigation and were regu- larly found to grow intermixed. The resulting amount of species-grouping into higher taxonomic ranks may be roughly estimated from the variety within each rank. Thus, for ex- ample, the 65 taxa of Lejeuneaceae used in the analysis are believed to represent ca. 70% of total Lejeuneaceae species present. Nomenclature follows recent checklists of liv- erworts (Gradstein & Hekking, 1979), mosses Churchill, 1989), and lichens (Sipman, 1989) or the latest taxonomic works, when available. “~ DATA COMPUTATION To make a meaningful evaluation of alpha and/ or beta species diversity patterns, two requirements should be met: (1) Adequate sampling. Minimum area (cumu- lative species vs. number of subsamples) curves were constructed (Fig. 3) and the rapid leveling of the curves shows that the minimum area of epi- phytic bryophytes and lichens is small, agreeing with Gradstein et al. (1990), who found that sam- pling of epiphytic bryophytes in mixed tropical lowland forest of French Guiana yielded only few 1—3) additional species per tree when more than ~ five trees were sampled. (2) Establishment that the gradient under in- vestigation, 1.е., altitude, is indeed the major en- vironmental factor determining species distribu- tion. For that purpose, ordination analysis, using presence/absence data of species, was carried out using Detrended Correspondence Analysis, DE- CORANA (Hill, 1979; Hill & Gauch, 1980). A graphic representation of the first two axes is shown in Figure 4. The first axis of ordination of the sites shows a relatively high eigenvalue (0.652), sug- gesting the presence of a clear species gradient indicating the site variation. The second axis has a much lower eigenvalue (0.259), thus most vari- ation can be explained by the gradient correspond- ing with the first axis. Eigenvalues of the third and fourth axes are 0.163 and 0.107, respectively. Since the Spearman’s rank coefficient for cor- relation calculated between the ordination rankings of the sites along the first axis of ordination and altitude was highly significant (R = 0.993; P < Annals of the Missouri Botanical Garden number of taxa 175 4 2550m 150 | 000000000000999^ 3190m 7] ооо . . .. . .. * 000000 ] ooo?" -в6685025202252028900оо9 1980 т 125 + о 959 3510 т | Ми Ed 9999999 е. АН gne т 1500 m ! оо AA n. 75 = o о ee о 2 09999 00902 4 оо o 504 9 . о - о А ы о .* 1o . 9- * 25 4 » 0 AA A 5 1 41 р 4 number of 0 5 10 15 25 30 35 40 45 50 55 60 subsamples Ош жол = лас a a a ee See ees кше Qe Le xeu Se ee ee oe ES Ies 0e шш куе Ue se ole ea ee ee FIGUR E 3. on four trees per altitude. 0.001), it can be concluded that the variation in altitude explains most of the variation in the data. Therefore, it is felt permissible to consider species distribution to be continuous between highest and lowest altitudes where species were found. For fur- ther calculations this assumption was made, fol- lowing Wilson & Shmida (1984). using DECORANA, formed for all data again, but also for macrolichens, Ordinations, were per- bryophytes, mosses, and liverworts separately. The sites at 1,000 m and at 4,130 m were excluded, the lower site being much influenced by humans and the higher site, an outlier on the second axis in the first ordination (Fig. 4), not being part of the montane hillside forest. Crustose lichens were s. d. “second axis 4 | 4130 3 — 2 — 3670" 3190* 2970 2460 1980 1250 1000 0 first axis T | A d. 0 : 2 3 4 5 URE 4. Distribution of sites over the first two axes generated by Detrended Correspondence Analysis; the scale of the axes are standard deviation units (s.d.). Ei- genvalue first axis: 0.652; second axis: 0.259. Cumulative species vs. number of subsamples (in random order) curves. The subsamples were taken not considered because of the presence of many sterile unidentifiable species. or unidimensional gradients the position of the sites on the first axis, as generated by DE has been shown to be a reliable estimate of beta diversity (А Капа, 1986). Beta diversity was fur- thermore quantified using Serensen's Coefficient of Community (1948) and a measure proposed by Wilson & Shmida (1984), which combines features of measures proposed by Cody (1975) and Whit- taker (1960): Serenson (1948) C. C. Serenson = 2s/(a + b) Wilson & Shmida (1984) "beta turnover" = [g(H) + 1(H)]/2a; where s — a — — species at site B, number of species found at both sites, , b = number of g(H) and КН) are number of species gained or lost respectively over the gra- dient, and а = average site richness Wilson & Shmida (1984) meda the use of their index when sample (site) data can be ar- number of species at site ids along a single overriding gradient, as in our Previous studies on beta diversity of bryophytes have used the number of half changes in com- munity composition along an elevational gradient (coenocline in the terminology of Whittaker), fol- lowing Whittaker (1960). For reasons of compat- ibility the same method was used. Coefficients of Community, CC (Serensen, 1948), were calculated for sites at different intervals along the coenocline. Volume 80, Number 4 1993 Wolf 933 Diversity Patterns of Epiphytic Bryophytes and Lichens A log linear regression of these site similarities (76) against distance along the gradient permits the calculation of CC, at zero distance. In all cases regression was calculated using the first 5 points on the graph only. Beta diversity in half changes over an ecological distance of z intervals (in our case 1,000 m altitudinal difference) is then cal- culated as 8 = (logCC, — log CC,)/log2. To eval- uate differences in species turnover between lower and higher elevations, the altitudinal coenocline was divided into two sections. The lower part had the site at 1,250 m as reference stand and the higher part the site at 3,070 m. RESULTS AND DISCUSSION CLIMATE Day climate di 5, following Ellenberg (1975) and including vapor pressure deficits, of the six stations along the transect (1,725-3,670 m) are based on hourly readings (Fig. 5). With increasing altitude, diurnal rhythms become less pronounced, a well-known pattern in equatorial montane forests (Walter, 1964), and a marked increase in humidity (hatched area) is notable above ca. 2,400 m as further shown by humidity conditions during day- light hours only (Fig. 6). Rainfall and humidity do not exhibit a linear relationship with altitude. Be- 2.000 m and 2,500 m there is a maximum in rainfall and a notable increase in tween about humidity with altitude, which corresponds with the increase in slope of the transect (Fig. 1). Local wind circulation patterns may account for the ob- served humidity and rainfall pattern: in the early morning hours anabatic winds rise from the valley floor along the hillside and while rising, clouds may be formed through condensation, on a typical day resulting in precipitation during the early afternoon at an altitude of about 2,100 m (van der Weert et al., 1986). Accordingly, at higher altitudes sun- shine was much rarer than in the valley and when it occurred it was during early morning hours and during late afternoons, when the clouds had passed onward (pers. obs.). During the night a drop in relative humidity, without an expected rise or even associated with a drop in temperature, was often observed (Fig. 7). This phenomenon, noted as far down the slope as 2,450 m and also known at the Sierra Nevada de Santa Marta, Colombia (van der Hammen, 1984), might be explained assuming the passing of a “dry” descending catabatic wind from the high (75,000 m) snow-capped hinterlying vol- canoes. These winds warm while descending, but apparently do not take up much moisture, perhaps due to the lack of bushy vegetation at higher el- evations. The distribution of tree base diffuse site factors (Anderson, 1964) with altitude shows that about 90-95% of the diffuse light is intercepted by the canopy (Fig. 8). Near the tree line at 3,670 m and at the Polylepis-dominated dwarf forest at 4,130 m the forest canopy is more open, with diffuse site factors of about 15% and 30%, respectively. BIOMASS PATTERNS The biomass per unit surface area of the epi- рћунс tree crown subsamples (excluding phanero- gams) varies considerably at a given altitude (Fig. 9A), but generally increases with altitude in agree- ment with quantitative measurements of bryophyte abundance in the forest understory along altitudinal gradients in Colombia (van Reenen & Gradstein, 1983) and Peru (Frahm, 1987c). High bryophyte biomass values in the tropics have been attributed to rainfall above 100 mm per month (Pócs, 1980), and to foggy conditions (Grubb & Whitmore, 1966; Ellenberg, 1975). (1987c) observed a direct relationship between the biomass of epiphytic bryophytes in northeast Peru rahm and the amount of radiation. The maximum re- corded biomass value at a tree trunk by Frahm was 1.4 g/dm', while in this study the maximum bryophyte dry weight at the trunk base was 16 g/dm? (at 3,670 m). The climatic data presented from the Peruvian transect suggest that the local climate is notably drier than in our area, which predictably would result in lower bryophyte biomass values. The rise in epiphytic biomass with altitude dis- plays a marked increase around ca. 2,400 m that is not correlated with a sudden change in rainfall Fig. 2), diffuse site factors (Fig. 8), or temperature van de Weert et al., 1986), but coincides with an observed rapid increase of humidity above the 2,400 m mark (Figs. 5, 6). Ecophysiological research on bryophytes confirms the importance of moisture m~ m~ availability to bryophytic growth. Hosokawa et al. (1964) demonstrated for several epiphytic bryo- phytes a positive relationship between photosyn- thetic rate and relative humidity at equal light conditions. Because bryophyte photosynthesis is relatively insensitive to low temperatures (Dilks & Proctor, 1975), decreasing temperature should not negate the positive effects of high humidity on growth at higher elevations. Long periods of high humidity not only allow for long periods of photosynthetic activity, but they also reduce the risk of desiccation. The latter factor 934 Annals of the Missouri Botanical Garden vapor pressure deficit temperature relative humidity °С % mbar 199 4 18 r 1725 m, 1Х-'86 2150 m, VIII-'86 90 д 16 E — 14 + = 10 и 25 = д в р 20 20 4 6 la. 15 15 4 4 | 10 10 5 "| = | 0 0 Jo b 0 6 12 18 24 °С % ы % 2970 m, ||--86 100 mbar 4 2 0 °С % °С * 3370 m, IV-'86 100 3670 m, V-'86 100 mbar 4 2 0 FiGURE 5 : 5. Тһе daily course of temperature (a), relative humidity (b), and (derived) vapor pressure deficit (c) at different altitudes, based on hourly readings during 12 (1,725 m; 2,150 m; 2,460 m; 2,970 m) or 18 days (3,370 m; 3,670 m). Presentation following Ellenberg (1975). Volume 80, Number 4 1993 Wolf Diversity Patterns of Epiphytic Bryophytes and Lichens % 100 80 60 40 20 0 + : ———'—— altitude (m) 1000 2000 3000 4000 a time (96) when the relative humidity is less than 70% e Inean vapor pressure deficit, expressed as a percentage of highest value (10.62 mbar) o mean relative humidity (96) IRE 6. Humidity conditions, as exemplified by three moisture indexes, in relation to altitude, based on hourly readings during daylight hours (06.00-18.00 hr.). Means of 12 (1,725 m; 2,150 m; 2,460 m; 2,970 m) or 18 days (3,370 m; 3,670 m). allows for the development of bryophyte growth forms that may attain a thick epiphytic layer (Fig. 9B) and hence a great biomass per surface area, such as tall turfs and pendents (Richards, 1984). Another factor may be the greater life expec- tancy of the substrate. At higher elevations the productivity of the forest is lower (Grubb, 1977; Veneklaas, 1991) and therefore the trees are pre- sumably older. C'* analysis of the core wood at breast height of a 15-m-tall Weinmannia mari- quitae Szyszyl. (Cunoniaceae) tree at 3,700 m altitude revealed a most probable age of 215 (+15) years (Hofstede et al., 1993). Lichen biomass values were not ена ѕер- y high arately from bryophytes, but 1g est at altitudes below 2,500 m, due pa the died occurrence of large foliose growth forms, e.g., Stic- ta зрр., at lower altitudes. diffuse site factor (%) 40 30 20 10 | m. ' 0 pM T T T т— T T T a) 500 1500 2500 3500 4500 altitude (m) FIGURE 8. Mean percentage of diffuse light cut a by the canopy, expressed as diffuse site factors (Anderso 1964) and calculated from e sy rome taken at the bases of the investigated tr EPIPHYTIC SPECIES NUMBERS In total, 473 taxa (108 mosses, 187 liverworts, and 178 lichens) are recognized in the subsamples. Based on the amount of species grouping, these taxa are estimated to represent ca. 90% of the species present (with estimated species numbers of 113 for mosses, 220 for liverworts, and 198 for lichens, excluding an unknown number of crustose lichens). Lichens may be subdivided into three main growth forms: foliose (110 taxa), fruticose (30 taxa), and crustose lichens (38 taxa). Crustose lichens include leprose (thallus powdery) and some squa- mulose (e.g., Pannaria rubiginosa (Ach.) Bory) forms. Fruticose lichens include filamentous (e.g., Coenogonium sp.) and loosely attached foliose growth forms with a thallus diameter of < 2 mm (e.g., Everniastrum catawbiense (Degel.) Hale). Squamulose growth forms are included with the foliose lichens, when the squamules regularly are 12 0] 12 0 12 O hr. С 40 FIGURE 7. Thermohygrograph reading, TPN-transect, 9-15 May 1986, altitude 3,670 m. 936 Annals of the Missouri Botanical Garden dry weight (g/dm2 tree surface area) 25 = ZU 13 4 А ; 10 4 | ; = - : і : Ж. . LA ТЕ op : 0 T T T T—7——1 = т Т ^ T T 7—1 0 500 1000 1500 2000 2500 3000 3500 4000 altitude (m) thickness bryophyte layer (cm) 20 4 15 = | B 10 = 5 d 0 еуез: y а план панна паи 0 500 1000 1500 2000 2500 3000 3500 4000 altitude (m) FIGURE 9. Bio ss (A) and thickness (B) of the (living) epiphytic oo layer. Each dot represents a subsample taken at the inner апа middle parts of the branches within the c > 0.5 em (e.g., Normandina pulchella (Borr.) Nyl.). For many organisms the tropical region is richer in species than the temperate zone (e.g., Pianka, 1966). This may not apply to bryophytes and li- chens, however. For example, recorded moss spe- cies numbers for the British Isles, Europe, the Guianas, and tropical America are, respectively, 700 (Smith, 1978), 1,084 (Corley et al., 1981), 234 (Florschütz-de Waard, 1991), and between 1,000 and 1,500 according to a rough estimation (Frahm & Gradstein, pers. comm.). Liverwort spe- cies numbers recorded for the same geographical areas are, respectively, 300 (Smith, 1990), 350 (Grolle, 1983), 375 (Gradstein & Hekking, 1989), and ca. 1,200 (Gradstein, 1988). Even though the number of bryophyte species in the tropics is likely to increase with continued exploration, it appears Volume 80, Number 4 1993 Wolf 937 Diversity Patterns of Epiphytic Bryophytes and Lichens ABLE 1. rainforest (Guian Ter Steege (1989); (3) from Montfoort & Ek (1990 tree crown and in the Guianas epiphyllous species. — Epiphytic bryophyte and macrolichen richness in a tem id as) and montane tropical rainforest (Colombia): (1 ) бош. с; (1961); (2) ; (4 forest (Sweden), lowland tropical from Cornelissen ) this study. Within the tropical areas sampling included the Number of trees Liver Total acro sampled Mosses worts bryophytes lichens Sweden; deciduous forest (1) many 78 17 95 == Guyana; dry evergreen Eperua spp. forest (2) 28 53 81 33 Guyana; mixed lowland rainforest (2) 28 60 88 19 French Guyana; mixed lowland i (3) 43 61 104 21 Colombia; montane ieee 1,500 m (4) 22 36 58 49 Colombia; montane rainforest; 2,550 m (4) 33 102 135 51 Colombia; montane rainforest; 3,510 m (4) 19 63 82 37 that temperate areas are richer in bryophytes (Brit- ish Isles 1,000 species) than tropical lowland rain- forest (Guianas 609 желе The ond relative contribution of liverworts to bryophy ric ness in the tropics (tropical ы са. 50%) ihan in temperate areas (Europe ca. 25%) is confirmed by data on epiphytic bryophyte diversity (Table 1). Total epiphytic bryophyte species numbers are sim- ilar in temperate Sweden, the tropical Guianas, and in Colombia, but species numbers in the tropics are likely to be higher due to the restrictive sampling methods used. Records on species numbers of li- chens in the tropics are still incomplete. Sipman (1989) estimated lichen richness in the transect to be of the order of 600-900 species, compared to 1,355 recorded species for the British Isles (James, 1965). Comparing bryophyte species numbers recorded on similar numbers of host trees in situations of near saturated species area curves (Table 1), at least for epiphytes no support is given for earlier assumptions (Richards, 1984, idi that wet mon- hy + tane tropical forests are richer in b than tropical lowland rainforests ui euo None species were not included in this study). Epiphytic macrolichen richness appears to be greater in trop- ical montane forest than in lowland forest. In addition to limitations imposed by the restric- tive sampling method by tree size, bark type, and location, bryophyte and lichen species numbers in general should be looked at with care for two rea- sons. First, there is justification for treating epiphytic and terrestrial cryptogams separately, not only be- cause epiphytic species are ecologically different, ut also because, in particular in the lowland trop- ics, terrestrial sampling is likely to be biased. Trop- ical lowland forests have low cover percentages of terrestrial bryophytes, possibly partly due to the smothering effect of leaf litter and partly due to the combination of low light intensities and high temperatures (Richards, 1984; Frahm, 1987c). Species richness estimations of lowland areas are likely to be highly influenced by factors such as gaps, road banks, and fallen trees. Bias in sampling may largely be avoided by sampling of the epiphytic flora growing on trees within the forest only. Fur- thermore, terrestrial substrates are more likely to differ from each other structurally, chemically, and in terms of water availability than epiphytic sites, and thus habitat heterogeneity is more likely to cause noise in diversity patterns. Second, epiphytic sampling should include sys- tematic investigation of the forest canopy. In the lowland rainforest of the Guianas, over 5046 of the epiphytic bryophyte species may be found exclu- sively in the canopy (Gradstein et al., 1990). More- over, a more open forest may permit the descent of typical canopy species to the forest floor, con- fusing altitudinal species diversity patterns when only the forest floor is sampled. The reported rise in number of species of lichens (Sipman, 1989) and bryophytes (Gradstein et al., 1989) near the upper tree limit in the investigated transect, based on both terrestrial and epiphytic (0-3 m) sampling, can possibly be ascribed to the occurrence of can- opy species near ground level, together with the influx of typical terrestrial páramo species. ALPHA DIVERSITY PATTERNS The altitudinal distribution of 473 bryophyte and lichen taxa is presented in Table 2. Bryophyte and lichen richness patterns along the transect selected for study are not identical (Fig. 10). Whereas bryophyte richness shows a clear maxi- mum between 2,550 m and 3,190 m, numbers of taxa of lichens decrease slightly between 1,500 m Missouri Botanical Garden Annals of the X X 219919 Y шпі) (*xoog) отојплар nyamos X ‘dds suapissi4 “е (202 +) y ds unnjouxan X X X ооӊ штјәѕ1диој ummquopoauy1443] "IMA 72 (mpaH) snpodo/>vu иоројил X "IMA 7D 27021u4nd "yo шпирлоцј X (696 #) V ‘ds талотоирлоцј шедлед гјритрмош шптуиороирлоц] X X X 1o[&e] n10/f12u0] muonoq anny (MN) unjasiuqna uoX121po]24^) X . . . X X . X пир ѕәроәрурш umipiuai) X X X X X X X ‘pog (TEI 72) nus nu ummguoposi1017) X X X X "pen > mw о 1m]ppnpoid24Q UNWOFDID) X (7211 #) V ‘ds sndoj&dum;) X X . . X "pog CIRIA 79) sni2sixag24 спаојХашо) X x sureqp A 242171d. 7) /"9'S'g (pug) про sndoj&duiv?) X X y “ds ¿ штјАашо) X "пола ("лвавмцос) штођапр штогујозаАј0) X "MG пјогјр голодшАјо) у ‘ds огаовлодшаАјео) g ds шпалд X y cds штала X (0811 #) У “ds оваовА (891 #) V “ds штогујкузрлаа ловееј (пир) Pusu umiuauu2va4g (EZI #) V ‘ds onuvajmg X X ‘LIA, »gofusnZup »nubajng X X X 30Z19H ('y101g) 24221]04d. Dyayodidp ~ С vx XX „е X ©“ же » vx 4 res » ре X Sasso OET‘ 0L9'€ OIS'E ОДЕ OGIE OL6'Z OFL'S OSSZ Oo9t'Z 0212 O86'l SZL‘T 00S'T OSZI 000'T (ш) apranyy 938 "(офшти џоцог|јоо 0} злојол ^ ^ * гаопоштооо рошпвола = *p10991 јепдов = 'joesue1] [Purpnjn[e aui Зиоје uonnquisip Mey] pue exe} papsova `2 318V 9q чәе J # Р Р I «X» pupue 94 1 чачар aq p РАИ с L "Tm 939 Wolf Volume 80, Number 4 1993 Diversity Patterns of Epiphytic Bryophytes and Lichens X X . X X X X X • X X X X x X X • X е . е X X . е б X X . • X X X . X X X x X X x • X X X X . X . bd . е » X » X X X ра ра = ж X X ~ X Чол (Mpeg) рио sisdoyoisou1a) ]p1euotray (мран) ојојпрић sisd042322N "in 0 suapiaqpos 0423422 "JUOJA хә “QUIYIS sisuopnj2 042322N jopae7) (Mpeg) зитаза штишрујоиз A [enuej CIRIA 7) штђојпошол 1UNIP110919 A] "gy штродАузрлау штапшолорју SURYA, 252722140740d штаупшоалорру роя ("доод) um]gof12u9] quy “IMA 7D гзигјошо ита UNI MOLIDA] ‘pug (Mpay) шпхолло WMA PUOLIIN adureg (yosuloy) vom ppdvs vid1pur] NYA umnssaaduio2 штшотпәт X 219918 X» UNI) (UIA) 270321022. sisdojuopoono'] 10p1e7) (YQ) Mysmoynjod wunXaqoona'] "19Y (UIA) sísuoqno uopooa3sa'] "MIA (00H) спвојиошој uopoaAdo'] Вотлон vunia10q pn221101d2] "ари CIRIA 72) nsyon umiuopojdo Japuez (zjualoT) asuaumund стел sapi0so nora umiuopojdo'] juepesiey 29 JNM ('Aneog-' d) зарлозојпота jo umiuopojdo'] "IMA 7D uwn228nndpo штјиорозәт (019 #) У “ds unpdopido'T aonidg (aduepy) мојјопш 77] /sueq (у 72) штооојаоц impidopid27] 'sutqoj sisua40ppnoa рЈәшотәт . X "Jy (AS) шпадио) шт2!аојао« ‘pug C^pop) штиовмошр) umidi01dodA]] гәдәр (uosp A Y 90H) лата vypəudíH “ж OET? 0292 01972 0/Є'Е 06172 0267 Олс OSSZ 09#'© 0212 086'1 ScZ'I OOS'I OSC'I OOO'I (ur) aprinqy enunuo) ‘с XI8V[ Annals of the 940 Missouri Botanical Garden X X и X X X X X X s d X X X X X X X X X X X . . . X X X X . . . X X X X X X X X X X X X X X X X X X X X X 3 X X X X X X X X X X X s X X X X X я . X X X X X X X X X X X X s x X X X Е s X X X * X X e é X X е X X X X X X X X X X X у X "jy 2uwnpiapjf umpÁudomuog Mya uwnaa3fipidsno um udouiog "uw 222upa08 UUNÁLGOUIOYIS any 21j0frisngub отшт9у10]426 "дол Cur) pupngpangun sisdoigojsououAM Y IMA 9 CTS) psonxajf “yo sisdoidojsououXu Y TRN 2 (2014р) штјолрирл“ um&aqopoy y "n'y (одшен) приђу umuogoz np y "Jy Cguo) sndapo4jod иоротюш&әуу pug Смрәң) шпѕозиәшој шпуаозом зир (мран) 2u420fnids шпАлдоца ка januey (^400H) ѕәртотиш 1UNALQOYLIA е] "цовилон umsuap uoX4q04214d одшен sua2sajnj-oosnf иороиола "aeg (10/4e J) su241a-02n] стел TRIN ^72 C^peH) snsuop uopououg snsuap “ARA “INIA D) ('M^peH) snsuap иороиолла ‘dds 1ny9113010 4 "posg (10/48 ) umqaadns иолригроцомјоло urmsouioo “pe WNIUUWDYJOLO f HA (odureg) әѕиәјрир189-оаои UNMIIIYIOLID]| d ‘doy ^f ^L Слоон) wnsoydoyourys штишот ја “PARO (Z219H Y ^ par) vyofiavo nj]a2uid "P129 (ППИ 72) мејригј тег па -ns3uy ("мрон) “ход оуоцошопа “pug ("мран) сио бут штиогојАца ладов[ ('Mp9H) sua2sauZiu ромттара "ug '9 73 Слоон) s2p1023$2] штројооштјра ОЕТ 0292 01972 02972 0676 0267 OFL'T OSSZ OOF'S 0212 086'1 6221 00S'T OSC'I OOO'I (ш) apnin]y тропициој ‘g ЯПЧУ][, 941 Wolf Volume 80, Number 4 1993 Diversity Patterns of Epiphytic Bryophytes and Lichens "any (urge) спитавлодиори у snyiunjapp "uy Coop) suaidioop sm1uvjapp aonudg 1188049 smjuv]app x X X X X miy (71109) 29 "quapurT) sun442qn snəajosoyruvopy >< >4 jd =< > >< 5110М1ӘАГ] X X IMA ^2 (зов) snsou20nbs uopos47 X . X X X X X uneig ("цовилон) 17p42mu124 UOPOBAT “UA (поје) s275u212u1jJ21d иороза7 'suqoy $1)18D1f иорова7 X X X X X . X "раб snumianiad сүэ иорова7 пиву (Aper) штта umi20212u10j27 yong (oduregp) шпрројојапу "yo штуцэр X зоор ојолрир орто X . . . X X YS (мред) штттипи штртуј b4 wng (UW) штитлапаза “IBA Са (мрод) 0mmioqgap штртуј, X X "u»seg umapgniup штртуј, (SLOT #) V “ds uncaqouumi] X X X "довод (pug) snoi775240d иородоцалксе X "IMA ^2 Cpug) sopioipodooX, иородоцалкб X чер `8 'лаовмцос гтојашозит иородоцалас X guo прируэрпо8 иородоуалкс X "ис гтодиоја “yo иородоцалкс иоѕ m (1o[Áe p) snzuopoay1Kaa иогодоја2116 X X "Q2 хә “YIN ^D s222duqvo uogodojdauig Áqsor) ("Бѕәт 9 IMS) прубит иоХјәтроиә]с̧ 10:9 C хоон) Sun: 91481u unmpnupnbg "y101g ('лаовмцос) suaa umprunpnbg X y “y101g (1018 p) wny211109Nn3) штртшопЬс (£09 #) y ‘ds оваовицовјас X X X X X X nug (pug) штрииладп« un Ay dojo 313318 (лом aq) шпироари с Га см CIMS) unn nsur unpAuydojpuag » X ра X » ра е X <> > ра ра X > » X X X X X X X X X X OET? 0292 OIS'£ OLEE 06ГЕ 0265 ObL'Z OSSZ O9F'Z 0212 O86'l SZL‘T 00S'T OSC'I 000'1 (ш) epriny 'penuguo) с TIY L Annals of the 942 Missouri Botanical Garden ра ра ра ра ж X ©“ X ©“ с ра X X X » X ©“ v ра » ра X X > SS ра ра ре ра » ра ра ра X ре е v» ра ра ра X X Xx x y ‘ds vaunala]010197) sueA'] (1109) ә одшен) punssiyuajod vounalo]01D197) ^ X "udeig Ņ yore nQopnpunas8 рдипо!2]019127) p103m4 (71109 39 "quapur]) ођуојле5 010 роп 2042142819 чор] 29 S3IN ритлапага "udeig (aonidg) раухојоАо ра X зао (Смс) psnffip зреле ©“ еже <> 'udeig (aonidg) оштирш 'јо n2unof2]010427) тгојрцаз) тгојроцаза) товоаАјођ) товоаАјођ) saan (mg) nuizipf siua31doXag siua1doXag Ņ my (sean) 02141044 d nounofojonjonig X "ugrq»s ("quepurT X 71102) ритрјод! 9) рдипо!ојо1 2010 'ugnog (20421) оуојхрј рдипо!ој072010 ‘ung CT) 21 4ydoyor1117 ошој“олоуа2] д ита oy AA Pas "15426 ("qdeig) »ijofianoos nounofojouvudo]g ‘BEJG UBA (пос) 9 "quapurT) »n4Zuooui naunofojo1vqdo]g X ‘staal, (quapurT) njndij$18u0] miunzzpg X X “SIA91 |, (quopurT) 1723001 тлирггрд X 'SIA91] (1100) 79 "quapurT]) nun1727n2349 VIUDIZDE propina (44916) sisuapyo “ye vrunzzog "jsnuog ("z19H) ојоопирлра paunalojoo4np '";snuog pajnf vaunalajoainp "дол (usso) n;42fuoo vaunalajojdoup ‘dds nunaup 311014) (saan) umsojnqni un] Xy do1svupy "udaig (sean) umaagiid. run] Ay dos svuy ‘ydaig (1o[&e [) шпјуруазоомпој штјАуаолријр "ydas ("шҷәт) unpuno штуАцаолјурију X X X "snuog DUISSIPIMA DIU X ојјоло) ydas) 2421272d порота шу гтујирјорр O£l'P 0295 0162 OLE'E 061° 0165 0722 OSSZ 097 0512 086'1 9221 005'1 OSC'I 000'T (ш) зрпицу `рәпиппиогу 'z ЯПЯУ], 943 Diversity Patterns of Epiphytic Bryophytes and Lichens Wolf X X X X X LI X Ippey nuofisuap Ssapiounymna 1] X X "udeis 14272u1m. D1unjnaJ X X . X X X X "UOJ 29 зоо D427dD4j21 ирта X . . X . X X X ‘ydaig njjonpupjs тирујтал X . . А X aonidg vppydavosavyds piuvjjna4 X . . X X X X X X X "давиу (ippey) sisuo412upfoi1 viuv]jna4 X . . А X "udeis мола viupqma X X X . . X X "доб ритлаплов ирт X X X X зоор (s99N) s2p102142 титујпал X ‘ydaig sisua40ppnoo vina X X X X X . X одшен Y “quapur] n;njoauoo miunyny X X X . X "quapurT 9 `шчцәт охоррлра сә niupnjyna4 X 1107) 9 "quapur] nupi27140uit "yo ютир}үпї] X . X X . . bu X X X зоо (SƏN) опаовупоо pupa] X X X X X X X X X X X Ippey $15492)1504q DIUDIINL X X X X X . X “nos ("Зизлас) 202240 DIUD]NLH X X X X . . X seaN (Te 19 " Muro) vnd piuvjpna X X X X X X X X X X x X X X X ‘dds nounafojoupdoaqq X X X X X X "qdajs (usstayy) vivoyou1 vaunafojourndoa(] X X X X 307194 рири vounolojoundoa(] X X X X X X X X X . X "Чә (ѕәәм) 124232nnd pouno[ojoispjdiq X X uyog (пој 29 зао) sipnixo рдипо!ојоирлоц X X y ‘ds nouno[2]014&7) X . X X . . . . X suea (aonidg) pd1sojoy paunof2j014&7) X > X sueA7 (109) suapaoon рәипәГәјоэХ) X X . X . X X X . X ‘ydayg siu4021nu3] DANO) X X . . . X ‘ydaig ("ugiqog) nuupunpu рлпјо) X X X X X пати (e 19 MUY) элю] vounola]0 194) X X X X . . X ‘dds ргипо!2]0]197) X X X X X NYG (ој 29 SƏN) опр vaunaflajopay’) X X . X X X X X X ојоло (V09) ритуороцо рәипәГәјорәц) OETH OL9'E OIS'E OZ£'€ OGl't 0267 OFL'S OSSZ OOT'Z 0212 086'1 SZLI OOC'I OSZ'I 000'1 (ш) әрпицү Volume 80, Number 4 1993 'penunuo) 'c 3I8V[ Missouri Botanical Garden Annals of the . X 'jspe1o) o Итлч (711029) споиоорапо ѕпујирритт X . . . X ојоле) (sean) smaXudaod snyd£osordo7 "поду (109) 2 edureg) хАјроовхца smjd&oso1d27] ejo19 (yda1s) n3o»f ѕпуаХәѕозаәт “HUA (00H) гтјодгипо snyd&osojda'] ојолоу (seen) под у шр snydAosordo7] ‘uyog (quapurq x "urqeT) ропа уја naunafajo1do'] ‘dds nizopidaT X sia y (aonidg) pniZuods pouno[lojopido'] X . . X X X ә[[о1су (109) n7njoaui naunafajopido'] gonidg (1o|&e]) psoumud ргјодлаг X X X X aeaovouno(o] (198 #) Я ‘ds vaunaloT x X (ZLS #) У ds рәипәГәт x "qdeis ("quepurT 29 "шч27) pjndijsixay24 рәипә[әт aonidg рујиојрбош vaunalo7] X X X “JUOJA 5) seaN suasaajan) vaunalaT] sean (Mg) pany vaunala] X "nog Y ‘пәри on31quo “ye vaunalo”] ә[олсу (ydars) 019/1930 y тлу зер ‘= 9[[012) (SAN) SIINDILLQNA njjauosaumf X 'ugiqog (aonidg) si4nydojq4jod гпуаАоволојо p1ogn4 ("udeig) спурјидрдп“ snj42q42]] oonidg (1o|Ke L) sipisuad snj42q42H » X ©“ x » X >“ “еж “© е v X ра X > ра X © © x > = е Pes “© е ре ~ ра е ре ре ре x vx K е >< КЕ X о о ра р “е 5d 9d pd pd . ~ ж » > > A о ре ~ ојоло (mg) snaprosadiunl snj12042]] ~ x © X © X < uauaay UBA STUDIQUIOJOS snj12Q42]] aonidg smgaujunopn зтагдаон "udeis (UO 1) зао) Dp¡XydXxo сүз ргипоГотралон X X X X X X . X X ‘ugtyas (aonidg) sapouisioun ргипоГорралон X X "Зер uea ("под 29 "quepurT) puniuunuiqoa1] sapiou DN n44 » ра OETH OL9'€ OIS't OLE'E 0612 0L6'Z ОР OSSZ O9F'T 0212 O86'l SZLI OOS'I OSC'I OOO'I (ur) epranyy 944 'penuguo) с 318V[ 945 Wolf Volume 80, Number 4 1993 Diversity Patterns of Epiphytic Bryophytes and Lichens ~ v x ad ра ре vc X ра ра © ра ~ ра ра ра X : X » v» vx РС v» XXX ра е ра ра ра ре ра = ж X е <> SS » x X X X “ж ра ра ра е е © е p» p ~ ра ра X » ре » X ра vx ра X > >< ~ ра ра ра ра ре ~ X us X 1o|Áe] Duidsis19mp орузотрја "109 Y 'quepur] nqofis12a1p. v]149013D] 4 "'quapurT DIDMAIDAIP D]/2017D] 4 цор ("'quepur 29 "urqoT) DYINSIP ојуоолттја ‘ydaig puidsisuap D)149013D] 4 WN DIDINI1ULO) 01420190] "добу DUNYIDOYI D]1/9013D] ] "'quepurT ("^se(q) DIDSANQ D]1YIO13D] 4 1102) si$u27080q ојуоолтоја под) suapiq 0ү1ҹәо19 ‘ydaig vunəysəq ојруоолтрја '"udeig пуолоирд D]1/2013D0]4 ‘nog ошто D]1Y9I013D] 4 “UNC (MS) SIPI0JUDIPD орцоор] “15Ppe 15) (1102) N "quapurT) страо snyyuooy du) SozioH smajooXK]r]d ‘CQ /S99N (MG) sruuofif snyjupjppyduy) "UYIYIS ("до A) отојпип) рәипәГәјојиор() "qpur 194951103 рајовиоју ‘dds pounofojoa2ipy (01422z;9]y '1995) ‘dds nij28z12]y "мм si$u2D]2ui 01128212] eonidg pjnauojdo] vi28z19]4 aonudg njo21m4f viu28zj2]y "аду цос (sse) suaidi22p 1122212] "ugiqogs (uos A) nonono n2auneafajosiisppy 'ugnqos (WW) DISNGOL »1u1821240]AI "UBIYIS (MS) DIDIYIDIQ DIUISIYILD]A] "а тцос (599) vosnfqns vaunalajoydoT ‘WOW »apiozadpa? vajor0y dor] доплас nypjuapiuponb 2j0201do'] зао (WYT) ојромпш 2jo201do'] ‘ung (7T) оуигрлд ргјозоцаот SƏƏN (MS) DIDUUOD “Ye ројозоцаот ОЕТ 0292 OIS'E OLEE OGIE 0265 OFL'S OSSZ OOt'Z Otl'Z 086'1 SZL'I 00S'T OSZ'I 0001 (ш) apranyy “panuquo”) С ЯПЧУ L Annals of the 946 Missouri Botanical Garden X X X X X X X X X X X X X X X “е X X „е X X X » ра ра е X X X “= » X ра “е = < =“ = ж "udeig xo yog DIDNUIS DINPDY "udeig xo чију vsownjd ојпром "udeig sigmpoipnu ојпром "добу 12 "quapur] ѕујош ојпром -dds родипоГајоиома (FIO #) V “ds роипоГоиома SUBAJ (71109) ојпшор “yo роипоГојоиома “ADIT, ("до AA) VUDIZILDMAS D]]940 'Ae1] (ооң) орраг мо D]]910 4 "ugmqos (IppeYy) 1021718049 7]]240d aonidg x1jpmo1a]As D]1Y9013D] 4 'quepur] nq4adns pjn2oi12v]d "quepur] nunjdqns орото (2661 #) Я ds 00702012] (тст #) Я “ds 01420180 (2261 #) а ‘ds 2014201904 (1921 #) О “ds 0114201901] (reel #) Я ds 001020190] (#911 #) V ds njyooiz jg "quapur] ирт njn[2012v]d -ydas »sopnjod pjn[2012vjd "добу sa1do4o1u рута лојде | n:ofij2nui 0]n20137]d "quapur] sua4133]20] D]1/2017D] ¿4 "quapur] CPM) 17uosaumf »]np20127]d '"udeig 149A9UISUDY D]1/9013D] 4 дио DUDIUNMI) MA 04201904 JO[ÁR | n27njoosnf D]1Y9013D] 4 JO[AB | 1719017 D]y9013D] 4 ‘ydaig DID9]Df 01490130] 4 "udeig риррјрлош«а D]YI0VÍD] Y 1104) рјјди1у22 01420190] 1o&e] SISUIJUNMOP p]nj2012v]d OST‘ 029'€ 01572 0255 O06l'€ 026‘ OFL'Z OSSZ O9F'S 0612 086'1 5221 0061 OSZI 000'1 (ш) грпицу 'penuguo)) “7 XIHV[ 947 Wolf Volume 80, Number 4 1993 Diversity Patterns of Epiphytic Bryophytes and Lichens X X X X X X . X X • X X X О . X . . X Ы X X X X X X X X X X X X X X X X X . X . X X X X X . . . X X X X X ©“ o» X “чәе © ж X X X X ©“ X X X » X X (£l #) Y 795 vipiong X (0121 #) V Ф; umgoyiodisp X X X (72621 Н) V ds штооулоорлујир X (992 H) y ‘ds vnuruosp suauot| 95015117) "под (Смс) DSOJUIWO] DIJOJOYIAL | "udaig siouaouruop paunalajA ЖЛ) — IMOH (лепу хә NOH) ѕәројошәи DIUDAD]O | "uyog (quoapur] `шцчцәт) panydjns pounofopixv] "ugs ("quapur] у cuya) piuosisajd рәипәГәрхо "ydas (UY) Suavsazpd рәипәГәрхо ydas vyofioun, рәипәГәрхо gonidg (z11emg) оурђојлоа Dd MIXTAS Yda nuna4282]u1 njJaigAzXg "ydas рои ірш DUO Y UAG шор 212401u2u04qQ DUO Y UAG səəy SISUIISDIG риХболуашАс̧ X x suea (1100) у “queapurT) штлојулдлод uimipizioiqui&g X X лерело) (saan) nd4noX]od рздипоГојомоијј 26 ‘Woy y одшенц si5u22140140d тира 026 ‘dds траоолу эреле (nads) 070701719 DIPADIMY n32pnjidpo стел Зио[ за X syusay (udoig) noopjnidpo vipin y zi9g (eonadg) puipup ripav221M "spei (yda1g) SUDLIIGD DI40221M ioÁe] VINJOA njnppM ‘ydaig xo “MIA 24203] DINPDY X "udeig 04011129] ојпром ‘ydaig sisuauosuos njnppM OET‘ 0292 OTS'E 02972 0675 02672 OLS OSSZ 09#'© ОГС О86'1 SGL'l 00S'T OSC'I 000'1 (ш) epniupy 'penuguo) “7 19У], Annals of the 948 Missouri Botanical Garden ©“ ра ра X “ж X X . . » X ~ o» » X X X X X “е ра ра » x » X “ее © ее X X ~ x “ж » X X » X © X X X ре vx » ж ~ » X = = еж ©“ ее =“ X X X =“ “ж е уе » v» X ра X ра е ©“ р >“ > == ‘Buaidg грмотјо un oyiod Ka] “Iesse (7T) su22s2jjnd отузајолц A) (917 #) V “ds роәузоирјәрү "ша (994) рзојполодп) n10dso]pSapy "uidiG (JÁN) nzxnupp p40dsojps2]y "еден Y Wey (924) 4070918100 D14pnz2ado'] "шејед Y јен (8194) sisuadunuop оутолјој eng хә "ÁN (14 YL) sun2197» иојпоәо:4ә7 (s08 +) g ds vapive7 "Sue1dg si4adid papioo] ‘dds ргоирэәт "diy “IRA (994) Dasnfoasnydjns viuo] (поцоц esojsnio o[uo1s) 1oput ‘ds рләр Tesse (Чоу) umaoiund ошшојошәру (8001 +) y ‘ds DUIUOL II] "хлад (MS) uma2142s DULSUOÁ II] ‘dds vi221/101d 47) ‘dds иојәәроту?) (289 &) V “ds uozoapory;) ‘AN Cquaaqq) umj2uro14qna uojoaponj:) 11904 Y 'ujogeH (294) пиозг1д219 »27jdojv-) у пр Cujedurarx) visapow тәп ‘dds mipiovg OETH 029€ 01972 02972 0615 02672 ОРІ‘ OSSZ 097Z 02172 086'1 SZLI OOS'I OSZ'I 000'1 (ш) грпиу 'penunuo) с XIHV[ 949 Wolf Volume 80, Number 4 1993 Diversity Patterns of Epiphytic Bryophytes and Lichens X “IPD 7T ^A Conr) 944 do390]03 “Jo рпшлоролојон X X X X X X X X ‘STAÐI J, (JESSEN) DUDIZIAIDSDI рпшлоролојон X X X X IQIM "V AA (IAN) D42/1q40q тшаоролоон b . X 801 x MOISUIMS ('S194) 5402190 D1u112p04212]] X . . X "uidi о ој шпајрилдал X X X e[eH Сл) 2014412 uma1spiu42à7] X M x X . X X X e[eH (әдә) 2SUIGNDIDI UNLISVIULIAS X . X . X "шед штвојпоплада ошләрон X (1821 #) у “ds vussporng X иәѕиәЯ:о[ x ÁBMO]|[OS) шпутрало“ puid2po11] X X X "semy (994) pypuvjddp пари] X X X X X e X X e X X "MSAMPH ‘ў ("Виолас) шт}0190]8 DWIUOA JIT X ‘ds штио8оиәо?) X X X . X X X X чивес ‘we BIY “IMA (yoy) 222720 01010202207) X . X . E . X Bory ә MOISUIMS (duo1dc) ујАхолујАл9 D1d40202207) X X "UBA sisuazunuop pid40202207) X `Чшәгур psounnbsqns тлиорој) X pnsoumnbs ‘ІВА "uyo (doas) psowbnbs тлиоррј) X X X X * X UNI 2401204420 сүз Duopo) шс (чивес) asuai10ppnbo TIL JjSDIIDA]27) ©“ X ра "uidi sungiqnp uma4]spipa]a?) (P86 #) Y “ds орут) X X X IPH V XJ CIÁN) ритлиђолоз руошлраоир) X X X PH 7 Xi (1qque7) vipuinidoausps рудшлараоир;) X eH Y хү CUBA) 2497/97 риупоршоир) X (067 #) V “ds ођемтјорир;) X X шојс (SxouT) 40709409 DILD]IPUD?) ајен Coin N Y ALPH) оѕоәјиәа xisyjoqging eH (3146) pxyjns x11410QNg X x X X X X iy “IRMA (IAN) s27mqoonao] тсир ра ре ра е Suaqot[0.19 € [A OETH 029€ OIS't 02672 0672 026% OvZ'c OSST 09#°© Otl'G 0861 ScZ'I 00S'T OSC'I 000'1 (ш) epninpy 'penuguo?)) °С TISV], Annals of the 950 Missouri Botanical Garden X . X . X X X X X X X X X X * X X X X X X X . X X X X X X X X X X : и X X X X X X ©“ ра X » X =“ ее ©“ x ра < X “е ра ра ра » ре X X © с ре ре v X “е © X X v X x 6 сеж » е ©“ е » X ра X “е » X IPH (oun N) »7»2uojo4d pudyovsjodA py ојен (JAN) sopioi2s&yd puXuopajodAg] IPH CUBA) n75$77q042111 DuX120410dA]] (114204 ‘H p9xa) dnoi3 зјед (Чишс) n;021a20] DUAYIDIJOdA y IPH (4) 2242201pis1 nuX12D410dA]] ILH (Солу) 2уојпорштшт pudyovijodA y X IPH CA4qp97) »7n10214qu рихуорлјовАН IPH (oun x) $0212 nuXyopajodXg ајен (142197) рлојузориг pudyovijodA y IPH Con x) pypuiziysisuap pudyovsjodA y ојен (APH) 1179F9p vudyonosjodA y IPH (010 x) vwpnIsndoas019 DUAYIDLIOAA y X аген (IAN) s75u221401502 DuXu2naj0dAg] e[eH (924) DIpuraynd “Jo рихуорлодхн IPH (10[&e |) s15u2220402 nuX12D41odX]] IPH (ALH) $7$u272nboq vudyovsjodA y ојен (CURA) s754737020Q vudyovijodAyy IPH (СТАМ) Duniyng ducyovsjodA py оген Sisuapup рихуорподхн uopey Y “UA ("ute A) зло па D1u42p04212]] X "stai (jn) 08012205 опшлоролојон X ‘SIAI, ("уп AA) 220272540 D1u112p04212]] SOIN 59 MOISUIMS ("мола N) 2 4ydovonu piu2poa212]] SOIN зу MOISUIMG ("241Ҷ®7) ооштуатош 01u2po4212]] uopew Y wog (олу) sua2s27n] опшлоролојоц 11904 (Т) рјашотој n1u42p04212[] ѕемү (JAN) Dioydoipis? отшіәрогәјәр ~ 199A VM C 1q1487) $1]Du12412 ртшләроләтә Otl'* 02972 OIS't 02672 0612 0267 0?%/'© OSSZ OOt'; 0212 086'1 5221 OOS'I OSZ'I 000'1 (ш) грпиуу ‘penunuoy) с TI8V[ 951 Wolf Volume 80, Number 4 1993 Diversity Patterns of Epiphytic Bryophytes and Lichens X X X X X X X X X X ° * s X ы X . X X X X X X X X X X X X e X X X X X X X X X X X X X X X X X X X X X X X X X X X X ы У X X X * s X • • X X X X X X * и X a X X $ X X к a X X X X X X X X X X X X E и X X X X X X X X X X X X X X * X X id s X * X X X X X X X X X X X X . . X . . X X * X X X • X 8 . X 91H Y XY (yesy) psounds sisdoui]outnd (6811 4) y “ds риђјошлра (1801 #) V ‘ds 991/9110 4 '"Siy пи Спор) 272u120421u j]21]21110d овоовмеиива (1221 #) V “ds урии] Алод (yoy) Dnsouigdiqna D14Dnuund "iy TOW C14 9) риомрош pnupnuund Алод ("цоу) »a3jdouoo niunuund uj “UL (99) $1$u2X0] uododoa) 659 40]021q uogodoa() — ЧАМ (1108 ю]]әуәупа DUYPUDULION| (ZLITI #) у de ¿ sisdowosydayy PH (994) 14n1242u02 num241011X]y PH Y Xq (92u4T) пииошриу »011420]2X]y V ‘ds тггрбдигуј ~ (296 # "шіцѕо Дд DIDULOXAGNS 1110907 ‘dds 1220907 “SIA91], (XooH) DID¿NUILI 111407 ‘dds wmsojdaT ТАМ sunuidnsas umnidojdo'] чоору (‘siag) шпалозојхца wmsojdaT ARIS) “4 'S (MS) uungouidapu wmsojidaT "Влог пиорирш wnisojdaT JAN штјодаој wmiso0jdaT "ute A (JOLY Y 'Азуј) шпорлојјрлоо шт80)0әт чоору (CT) пзоблпда wmsojdaT "Biof 24pjmo100 штгојао 1 (129 #) V “ds ¿ vusAydoro7] ‘ds uaqoi <поцој yaput IPH (uius) »sonuis DUAYIDLJOAA y PH (Ca4qpq97) 73204 DuXq2D4jodXpgI IPH (A9MY) snuaofipqpya nuXu2naj0d&g] IPH (IAN) suaonpaa nuXu2pajodAg] OETH 029€ OIS'£ 02672 0672 0/6'б 0725 OSSZ OOt'; 0212 O86'l 5221 00S'T 062'1 000'1 (ш) epninpy 'penuguo) ‘с ATA], Annals of the 952 Missouri Botanical Garden * X X X X X X X я X X X X X X з X X X X X " X X X X X X и X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X : X X X X X X X X X X X . X " X X 19127 $1403]1202 DUDUDY Зогу X MOOSUIMG ("Зизлас) 14790729 DUDUDY "40g n242u10qns SUIXA Y “UBA (ADN) 2149/2199Y989 2u1x&q әш n4221]]0402 2uix&q ‘ysu (‘JAN ut *xon) psouimadoisaDo 2u1x&q Догу (YOY) nj2apna4 vi]212und "ure A (OPA) 272214701 0140jJ2gdX2opnosq "ute A (оу) DIDIND n1imgauyd&oopnosq uessuaH (JÁN) шпогмригр штртузлјод аваовто5 АЦА Влодојј nsoon442a DIDSAY 319qOJA psoipaíos n10sAafq ang z/5u12uod тозлуа Владо n70]nqo] v12s&1fd ‘JAN 222482111 DIISAY J Влодојуј suedwunia DISSAY y 3I9OJA DIDILISOLID DIOSAY d Влодојј (YoY) onpidsiy vioskydosoy y AN (20е) Djua niga nd n12211]24 ЛАМ (JAN) 9714404910p 012211124 зен 3 XA CIÁN) sisua pun sisdoui24)10u404 91H CKA) шт10]2и1) ошолошлра 9[eH CIN) wnjdunsqns Du21]011404 ILH (93u&T) 1aZun-112ups nu2110u10g IPH (ләдә) :nsnqoa Dui2410u1404 IPH (APH) шпр'дјолда Dui2410u14D4 IPH (93po() позујеш nui2410u14nd 9[eH (APH) ?su217101 ошалошлара 32H шпиркту ошолошлра e[eH ('141427) ISUIUISOLISND "JO Dui294]0u14 Dd OETH 0292 OIS'£ OLE'E OGIE 02672 OL'T OSSZ 09#'© OEIT 0861 5221 OOS'I OSC'I OOO'I (ш) apnin[y 'penuguo)) с xXI8V[ 953 Wolf Volume 80, Number 4 1993 Diversity Patterns of Epiphytic Bryophytes and Lichens X X X X X X X X X X X X X X X X е . . X ы X X X X X . X и X X X X X X X X X X X X X X X X X X X X $ . . X е X X X X X . . X . . X X X • X X X X X . X А X X X X • X X X ‘dds рәиѕу) 141427 (ÁN) snono]SodAy soisnjosoja] "WoON (м$) SUDIIADY sajsnj2oso]a[ "yəy ('1198] пјодлот DJING — ‘dds по ‘yoy (sx руси ут 21206 “YOY (AS) 07020120] "JO 71206 Чоу (MS) su1022Dutpp “JO 1206 — "yes y ('141427) гпирзошлој snsoydosavydg 19991914 Y ALH (‘Sry пир Pserpisigns nijouny 19991911 79 21H (лојАе |) 7701021124 орәипу 19991914 79 AH (32H) DIDARSUIMUOS) рцәшту (9221 +) vimjui “Joes «ds DUIDUDY (£61 #) V “ds оиуошру OETH 029€ 01972 OLEE 0672 02675 Олс OSSZ O9t'; 0612 086'1 SZLI 00S'T OSC'I 000'1 (ur) грпиуу тропициој с ATAVY, 954 Annals of the Missouri Botanical Garden nr. of taxa 250 4 E —e— bryophytes + lichens 200 4 —o— bryophytes J ——*——- lichens 150 4 100 - 50 - О altitude (m) O 500 1000 1500 2000 2500 3000 3500 4000 4500 FicURE 10. Distribution pattern of bryophyte and lichen richness in relation to altitude, as derived from Table 2. and 3,200 m and substantially at higher altitudes. While foliose and crustose lichen taxa numbers tend to drop with altitude, fruticose lichen numbers increase up to an altitude of about 3,400 m (Fig. 11). The bryophyte curve (Fig. 12) is mainly de- termined by the liverwort component, whereas mosses exhibit very much the same distribution as the overall lichen pattern. nr. of taxa dal 70 d The mosses : liverworts ratio has been repeatedly correlated to humidity (Gradstein et al., 1989; van Reenen & Gradstein, 1983) and the data on epi- phytic bryophytes, including canopy species, for altitudes up to ca. 3,200 m confirm the hypothesis that the contribution of liverworts to overall bryo- phyte diversity increases with humidity. Liverworts are an important element, in terms lichens (total) crustose lichens foliose lichens fruticose lichens FIGURE 11. 0 A 1 2). II 1 0 500 1000 1500 2000 2500 3000 altitude (m) T тт 3500 4000 4500 Distribution pattern of lichen richness in relation to altitude, as derived from Table 2. Volume 80, Number 4 1993 Wolf Diversity Patterns of Epiphytic Bryophytes and Lichens ——e— bryophytes ==. liverworts ——lBM32a —— MOSSES 0 " Џ Џ 1 T 0 500 FicunE 12. of species numbers, of the epiphytic flora, and reach maximum species richness between 2,550 m and 3,190 m (Fig. 12). Shmida & Wilson (1985) proposed two hypotheses, besides niche relations and habitat diversity (e.g., MacArthur, 1965), to explain how species diversity along a macrogra- dient can be caused and maintained: mass effect and ecological equivalency. Mass effect is the es- tablishment of species in sites where they cannot T Te] altitude (m) е5) 1000 1500 2000 2500 3000 3500 4000 4500 Distribution pattern of bryophyte richness in relation to altitude, as derived from Table 2. ђе self-maintaining, and ecological equivalency 15 the coexistence of species within the same funda- mental niche. e mass effect hypothesis requires the exis- tence of an adjacent core area from which there is a steady influx of propagules to the site. In practice it is not always easy to establish whether a species will be able to survive at a site. The mass effect hypothesis is attractive for epi- nr. of taxa 40 4 | — QO-— highest points of distribution 30 4 ——e— lowest points of distribution | —=—— exclusive taxa 20 4 10 + 0 pepe r altitude (m) 500 1000 1500 2000 2500 3000 3500 4000 Distribution of taxa of liverworts along the altitudinal gradient; number of exclusive taxa for each altitude and mb of endpoints of species ranges 956 Annals of the Missouri Botanical Garden Average position of sites along the first axis of ordination as generated by Detrended Correspondence Analysis. Scaling of the axis in units of standard deviation (s.d. units) x 100 (Hill, 1979). TABLE 3. Site altitude (m) 2,460 740 2,970 3,190 3,370 3,510 3,670 2,550 Eigenvalue axis 4 182 133 98 224 386 0.613/0.131 0.630/0.137 Macrolichens Liverworts 0.613/0.095 Mosses 321 264 212 156 118 351 454 0.646/0.089 Bryophytes phytic bryophytes and lichens. First, many lichens and bryophytes often produce spores in abundance (Richards, 1984) and/or have extensive vegetative reproduction; parts of the plants just break off or special structures are formed, like gemmae in moss- es and soralia or isidia in lichens. While data on dispersal efficiency of bryophytes and lichens are difficult to obtain, some impression may be obtained by the high speed of colonization of new available sites. Second, in a frequently disturbed habitat it will be easier for propagules to find a suitable (vacant) site to establish themselves. The epiphytic habitat is likely to have a high rate of disturbance due to the exfoliation of bark fragments, the occurrence of tree or branch fall, and disturbances by animals such as mammals (Perry, 1978b), birds (nesting material, nesting sites, and invertebrate foraging; Nadkarni Matelson, 1989), and invertebrates (pathways of leaf cutter ants; pers. obs.). According to Shmida & Wilson (1985), the in- fluence of ecological equivalency on overall rich- ness should be highest in systems in which habitats are patchily distributed (isolated branches) and/or in which competitive interactions are reduced. As to competition, indications are that bryophyte com- munities are not at competitive equilibrium (Slack, 1977; Watson, 1980), and that competitive in- teractions between bryophytes are not very intense (During & ter Horst, 1987; During and van Toor- en, 1988). Mass effect and ecological equivalency hypoth- eses, in combination with overlapping species rang- es, predict the highest species diversity at transition ones, where two distinct floras meet (= contact transition). The distribution patterns of individual liverwort taxa along the elevational gradient (Fig. 13) show that at 2,130 m many taxa have their lowest point of distribution, coinciding with a marked decrease of humidity below this altitude (Figs. 5, 6). A second peak in “‘points of lowest distribution” at 2,550 m indicates the first occurrence of many liverwort species characteristic of the higher (al- titudinal) forest, e.g., Lepicolea pruinosa (Tayl.) Spruce, Jamesoniella rubricaulis (Nees) Grolle, and several Herbertus species. At an altitude of 3,190 m many species from lower altitudes reach their upper limit, e.g., Taxilejeunea pterigonia (Lehm. & Lindenb.) Schiffn., Frullania caulisegua (Nees) Nees (—F. obcordata), and Plagiochila bursata (Desv.) Lindenb. ifference in max- ima between points of lowest and of highest alti- tudinal distributions demonstrates the presence of a zone of overlap between 2,550 and 3,190 m. This transition zone coincides nicely with the belt Volume 80, Number 4 1993 Wolf Diversity Patterns of Epiphytic Bryophytes and Lichens f-"turnover" 4.0 | 3.5 3.0 4 2.5 = 2.0 + 1.5 4 ——p— liverworts — mosses 1.0 + ——o—— macrolichens s | ——#— bryophytes-lichens 0.0 - - - - - - ——4 altitude (m) 1000 1500 2000 2500 3000 3500 4000 5 FIGURE 14. Beta diversity along the altitudinal gradient, using the Wilson € Shmida (1984) measure, or “beta turnover," of diversity. of highest species richness (Fig. 12), as predicted. However, induced floristic transition zones, where a sharp environmental boundary in a gradient is present, have a great variety of species as well, but this type of transition zone is likely to have species centered in it. On the total of 127 taxa of liverworts occurring in the transition zone, there are 101 taxa (ca. 80%) that occur outside this altitudinal range as well and each altitude sampled within this range has only 2-6 (= <10%) exclusive species (Fig. 13). The bryophyte floras from lower and higher elevations in the transect are from a different origin, as demonstrated by Gradstein et al. (1989). They concluded: “It appears that at lower altitudes, below 3,000 m, wide ranging trop- ical species prevail whereas narrow ranging tropical species and species of temperate origin are more .* common at higher altitudes. BETA DIVERSITY PATTERNS Relative site position along the first axis of or- dination, using DECORANA (Table 3), may be seen as a reliable estimate of species turnover, when the gradient is unidimensional (Okland, 1986), as shown by the much higher eigenvalues of the first axis than of the second axis. For macrolichens, mosses, and liverworts, Spearman’s rank coefficients for correlation between ordination rankings of sites along the first axis and altitude are highly signifi- cant (R = 0.989; P < 0.001). Altitude, a complex ecological factor, thus explains well the variation in the data for these taxonomic groups. Macrolichens generally have a greater altitudi- nal range than liverworts and mosses, indicated by a shorter first axis of ordination of, respectively, 3.68 s.d., 4.30 s.d., and 5.28 s.d. Lichens were shown to have wider amplitudes than bryophytes (mainly mosses) in relation to latitudinal and edaph- ic (pH, texture, and moisture) gradients in the Canadian subarctic forest tundra as well (Robinson et al., 1989) and along a potential evapotranspi- ration gradient and a gradient based on Conrad's index of continentality in pine forests in Finland (Oksanen, 1983). Mosses apparently are more specialistic than liverworts (Table 3), a conclusion also reached b Florschütz-de Waard & Bekker (1987) studying different types of lowland forest in Suriname. The change in beta diversity on the altitudinal gradient, using Wilson & Shmida's (1984) mea- sure, “beta turnover," reveals that the overall more specialistic behavior of mosses may in particular be attributed to the greater species turnover of mosses at altitudes above 2,130 m (Fig. 14). Lee & La Roi (1979) compared half change values (Whittaker, 1960) of bryophytes along an altitudinal gradient in Canada with values derived from studies by Slack (1977) in the United States. Half change values along altitudinal coenoclines 958 Annals of the Missouri Botanical Garden Half changes of elevational coenoclines in temperate and tropical America. (1) From Slack (1977), 0 (19 TABLE 4. coenocline: length 970 m, mid-elevation ca. 1, m; (2) fror n Lee & La Roi 1,250 m, 79), coenocline: lengt mid-elevation ca. 1,100 m; (3) this study, coenocline: length 1,000 m, mid-elevation 1,750 m (a) and 3,170 m (b). Bryophytes/ Bryophytes Macrolichens macrolichens Adirondack Mountains, New York, U.S.A. (1) ca. 1.26 — — Jasper National Park, Canada (2) 1.1 — — Colombia (a) reference stand: 1,250 m (3) 1.45 0.93 1.26 (b) reference stand: 3,670 m (3) 1.30 0.94 1.22 are presented in Table 4. Semi log regressions used for the calculation of half changes in Colombia all had К? values of over 0.92. Contrary to indications given by beta turnover values (Fig. 14), bryophytes and macrolichens do appear to have a different rate of biotic change at lower altitudes, with half changes of 1.45 and 0.93, respectively. A possible explanation may be methodological. Half change values are calculated using similarity values of sites to a reference (first) stand. While for the other sites in the coenocline some correction is possible for outliers by means of the regression used, the composition of the first stand is determinant. Beta turnover as measure does not attach extra weight to one of the stands. Half change values in the tropical mountains of Colombia are greater, especially for the “1,250 m coenocline," than have been reported for temper- ate North American mountains, indicating narrow- er species ranges in the tropics in relation to altitude as predicted by Janzen (1967) and documented for lizards, frogs, and snakes (Huey, 1978). However, present data on bryophytes should be regarded with caution, since the number of substrate types be- tween studies varied. Whereas this study was lim- ited to epiphytes, Slack's (1977) study included bryophytes from a wide range of both epiphytic and terrestrial substrate types, and Lee & La Roi (1979) sampled the three most abundant terrestrial substrates only in each stand. Including bryophytes from more substrate types in the analysis will lead to a higher species turnover, as remarked by Le & La Roi (1979), not invalidating the general conclusion. Furthermore, actual species turnover in the present study is higher than recorded because of occasional species lumping. LITERATURE CITED ANDERSON, M. C. 1964. Studies on the woodland light climate. I. The la computation of light conditions. : a 41. CHURCHILL, 5. i a yologia novo granatensis. Estudios nd ® musgos de Colombia IV. Catálogo sean rà los musgos de Colombia. J. Trop. Bryol. 1: 95- Copy, M. a 1975. Towards a theory of continental species diversities: Bird distributions over Mediter ranean habitat gradients. Pp. 214-257 in M. L. Cody & J. M. Diamond (editors), Ecology and Evo- oa of Communities. Harvard Univ. Press, Cam- Conve 1 H. 1978. Diversity in pM rainforests and coral reefs. High diversity of trees and corals maintain е E in nonequilibrium PRA Science 199: 1302-13 ConLEY, M. F. Нил. & A. Н E SMITH. . С. CRUNDWELL, R. DuLL, M. О. 1981. osses of Europe iud the iu ^d еса 11: 609-689. ои Ј. ТЕК STEEGE. 1989. Dis- tribution s sion p ПЕТА a and suh in dry evergreen forest of Guyana. J. Trop. Ecol. Dirks, T. | к. & M. С. F. Proctor. 1975. Compar- ative experiments on dec abides reponses of bryo- i has КҮГЕ respirat J. Bryol. 8: DURAN. W. E 1988. Patterns of diversity in anuran amphibians in s irri Tropics. Ann. Missouri Bot. Gard. 04. Durinc, H. J. 4 E TER TO 1987. Diversity and ynamics in bryophyte communities on earth banks in a Dutch ires RU ied Hung. 35: 447-455. —— —— & B. Е. van Тос 88. Pattern and зв | Ay bryophyte eyer of a chalk grassland. . 196- n H. J. D Ји J. А. Werger & ELLENBERG, H. bis perariden bereichen der tropischen Anden. Phy- ocoenologia 2: 368-387. ENROTH, J. 1990. Altitudinal zonation of bryophytes on the Huon Peninsula, Papua New Guinea. A floristic approach; чш „phytogeographic considerations. Trop. Bryol. 2: FAUTH, VE " се & J. B. StowiNskKi. 1989. Elevation patterns of species richness, evenness, and abundance of the Costa Rican leaf litter herpeto- fauna. Biotropica 21: 178-185. FLORSCHÚTZ-DE WAARD, J. 1991. bryophytes of the Guianas. II. Musci. 3: 89-104 А catalogue of the Trop. Bryol. ‚ BEKKER. the bryophyte flora of various types in name. Cryptog. Bryol. et Lichénol. 8: FRAHM, J.P. 1987a. Struktur und ie ae eine 1987. A comparative study of Suri- Volume 80, Number 4 1993 Wolf 959 Diversity Patterns of Epiphytic Bryophytes and Lichens der epiphytischen moosvegetation in Ug ва NO- Perus. io Nova Hedwigia 88: 115-141. 87b. Which factors d the growth of eis bryophytes in tropical rainforests? Symp Biol. Hung. 35: 639-648. 1987c. Okologische studien úber die epiphy- tische е іп regenwalder NO-Perus. Већ. Nova Hedwigia 88: 143-158. GENTRY, А. : poses of E n plant Land diversity. Evol. Biol. 15: 1-84. 988. Changes in plant е diversity ind Tei ic composition on environmental and geo- graphical gradients. Ann. Missouri Bot. Ca rd. 75: 1-34. DopsoN. 1987. Diversity and bio geogra aphy of neotropical n _ epiphyte s. rund Missouri Bot. Gar 23 GRADSTEIN, S. R. paticology. I. Simposio Latinamericano de Briologia, Bogotá [Abstract.] & J.-P. FRAHM. 1987. Гле floristische ћоћ- engliederung der moose entlang des BRYOTROP- о in МО Peru. Већ. Nova Hedwigia 88: 105- erie on neotropical he- & W H. A. HEKKING. 1979. Studies on Co- lombian cryptogams IV. A E of the Hepaticae of Colombia. J. Hattori Bot. Lab. 45: 93-144. A а of the bryo- phytes of the Guianas. I. Hepaticae and Anthocer- otae. J. Hattori Bot. Lab. 66: 197-230. . Pócs. 1989. Bryophytes. Pp. 311-325 in H. Lis & M. J. A. Werger (editors), Tropical Rainforest Ecosystems 14B. Elsevier, Amsterda & A. BER. 1982. Bryogeography of ч Calapogos Islands. J. Hattori Bot. Lab. 5 Е MATSUDA & Y. ASAKAWA. 1981. Oil bodies and terpenoids in Lejeuneaceae and other selected ori Bot. Lab. 50: 231-248. H. C. CORNELISSEN. 1990. Species richness and phytogeography of the bryo- phyte flora of the Guianas, with special reference to the lowland forest. Trop. Bryol. 2: 117-126. М REENEN & D. GRIFFIN Ш. 1989. Species richness and origin of the bryophyte flora of the Colombian Andes. Acta Bot. Neerl. 38: 439- 448. GROLLE, R. 1983. Hepatics of Europe including the Azores. J. Bryol. 12: 403-459. GRUBB, P. J. 1977. Control of dh growth and dis- tribution on wet tropical mountains: with special ref- erence to mineral nutrition. aa Rev. Ecol. Syst. 8: 83-107. . WHITMORE. 1966. А comparison of montane and lowland rainforest in Ecuador. II. The climate and its effects on the distribution and phys- iognomy of the forests. J. Ecol. 54: 303-333 LOYD, T. D. PENNINGTON & T. WHITMO ORE. A comparison of montane and lowland rainforest in ү Éeusdar. I. The forest | physiognomy, and floristics. Ј. Ecol. 51: HAMMEN, T. VAN DER. . Datos eco- тн del transecto Buritaca T Bu (Sierra Nevada de Santa Marta). In: T. van der Hammen & P. M. Ruiz (editors), La Sierra "Nevada de Santa Marta (Colombia). Stud. Trop. Andean Ecosyst. 2: 45-66. Cramer, Berlin, Stuttgart. , S. DiAz-PrEpRAHITA & У. J. ALVAREZ (editors). 1989. La Cordillera Central Colombiana, Transecto Parque Los Nevados e bu Parte). Stud. Trop. Andean Ecosyst. 3: 1-600. Cramer, Berlin, Stutt- gart. A. PÉREZ «€ P. PINTO. (editors). 1983. La Cordillera Central Colombiana, Transecto Parque Los Nevados (Introdución y datos iniciales). Stud. Trop. Andean Ecosyst. 1: 1-345. Cramer, Vaduz . 1979. DECORANA: A FORTRAN Pro- gram for Detrended Correspondence Analysis and Reciprocal Averaging. Cornell Ecology Programs Se- ries. Cornell University, Ithaca, New York. & H. G. Gauc, JR. 1980. Detrended cor- respondence analysis: oo ordination tech- nique. Vegetatio 42: e HorsrEDE, R. С. M., F & D. Н. BENZING. 1993. Epiphytic osa a бук, status of a Colombian upper montane rain forest. Selbyana 14: 3-11. HILL, Hosokawa, T., N. 1 & T. Tacawa. 1964. Cau- sality of the e. of corticulous species in forests with special a | the physio- ы approach. oe 67: 1 Huey, B. | Ми pattern of between- altitude ы. jer mountains eos be higher Amer. Naturalist 112: 225-229. A general ie та of species diversity. Amer. Naturalist 113: 81-101. JAMES, P. W. 1965. A new checklist of British lichens. Lichenologist 3: 95-153. JANZEN, D. H. Why mountain к are higher in the tropics. Ammen. Naturalist 101: 233-249. . Sweep samples of tropical foliage in- sects: effects of seasons, vegetation types, elevation, time of day, and insularity. Ecology 54: 687-708. LEE, T. D. 1979. Bryophyte and е vascular plant beta diversity in relation to moisture and elevation gradients. Vegetatio 40: 29-38. Mac ARTHUR, R. H. 1965. Pattern in species diversity. Biol. Rev. 40: 510-533 MAURY- "vagis G., M. pem & T. YOUNES (editors). Pp. 1-76 in The Significance of Species D in Tropical Forest Ecosystems. Biology In- ternational, The International Union of Biological Sciences (IUBS) News Magazine, Special Issue 6, aris. MONTFOORT, D. & R. Ек. 1990. Vertical Distribution ч и of Epiphytic Bryophytes and Lichens in ench Guiana na r recht. NADKARNI, N. . MATELSON. 1989. Bird use of epi iphytic т resources in neotropical trees. The Co dor 91: -907. OKSANAN, J. 1983. Diversity patterns along climatic gradients in the understory of lichen rich pine forest in Finland. Ann. Bot. Fenn. 20: 151-155 ØKLAND, R. 1986. Rescaling of ecological gradients. I. Calculation of ecological distance between vegetation stands by means of their floristic composition. Nordic J. Bot. 6: 651-660. odd Pn rio aspectos del clima. Pp. 38- in T. der Hammen, A. Pérez & P. Pinto "oui La "Cordillera Central Colombiana, Tran- secto Parque Los Nevados (Introdución y datos ini- n Stud. Trop. Andean Ecosyst. 1. Cramer, Va- A method of access into crowns pent ^D. R. 1978a. Annals of the Missouri Botanical Garden of emergent and canopy trees. Biotropica 10: 155- 157. 1978b. Factors influencing arboreal У sociology in Central America. Biotropica 10: 23: 231. PIANKA, E. R. 1966. Latitudinal gradients in species diversity: А review of concepts. J. Amer. Naturalist : 33-46. Pócs, T. 1980. The А пен and из effect on the waterbalance of two rain forest types in the yaris mountains. Acta Bot. Acad. Sci. Hung. 26: 43-167. REENEN, С. B. A. vaN. 1983. Distribucion у ecologia de musgos у hepáticas (datos iniciales). Рр. 206- 209 in T. van der Hammen, A. Pérez & P. Pinto (editors), M Cordillera Central Colombiana, Tran- secto Parque Los Nevados (Introdución y datos ini- ciales). Std. Trop. Andean Ecosyst. 1. Cramer, Va- duz. S. R. GRADSTEIN. 1983. A transect analysis of the bryophyte e along an altitudinal gra- dient on the Sierr vada de Santa Marta, Colom- Bot. es 32. 163-175. 984. The ecology of O forest bryophytes. Pp. 1233-1270 in oM. Schuster (ed- itor), New Manual of шо, vil. 2. The Hs Botanical Laboratory, Nichinar 1988. Tropical forest БЕТЕМ Synusiae and. strategies. f Hattori Bot. Lab. es is > ROBINSON, A. L., . Упт & K. P. Tim Patterns of и and lichen ве аса in re- lation to latitudinal and edaphic gradients in t nadian subarctic forest-tundra. Nova Hedwigia 49: 25-48. Scorr, N. J., a а. The abundance and diversity of the | una of tropical forest litter. Biotropica 8: 41-58. SHMIDA, A. & M. V. Witson. 1985. Biological daiar: minants of species diversity. J. Biogeography 1 20. SiPMAN, H. J. M. 1989. pw Par in the Parque Los Nevados Transect. Pp. sige in T. van der Hammen, S. Diaz- ar DN . J. Alvarez (edi- tors), La Cordillera Central Pom к Transecto Parque Los Nevados (Segunda Parte). Stud. Trop. Andean Ecosyst. 3. Cramer, Berlin, Stuttgart. SJÖGREN, E. Epiphytische moosvegetation in laubwäldern der Insel Oland (Schweden). Acta Phy- 149. Species Diversity and Community Structure in Bryophytes. New York State studies. New York State Museum Bull. 428. Albany, New York. SMITH, A. J. 978. The Ireland. и Univ Moss Flora w D and . Press, Cam . Epiphytes айй epiliths. Pp. fo 1-227 in A. J. E. Smith (editor), Bryophyte Ecology. Chap- man "S s London, sd York. The Liverworts of Britain and Ireland. ALS Univ. Press, Cambridge. S@RENSEN, T. 19 A method of establishing groups of equal amplitude i in plant sociology based on sim- ilarity of species content and its application to ruri of the vegetation of Danish commons. d (Kongel. Danske Vidensk. Selsk. N.S.) 5 STEEGE, Н. TER & J. Н. C. CORNELISSEN. 1988. Col- lecting ње stu dyin ng bryophytes in Hs еш ы of standing rainforest trees. Pp. 28 0 in J. M. Glime с Methods іп пасва P Bryol. Meth. Works om Mainz. Spo A. & R. J. RoBiNs. 1979. Aspects of ascular e epiphytes i in Colombian cloud forests, I. The distribution of the epiphytic fora: Biotropica 11: 173- 188. TERBORGH, J. 1 an Andean elevational gradient Edo) 5 58: 1007- “1019. THOURET, J. С. е y problemas Jis fológicos. Pp. fe in T. van der Hamm Pérez & P. Pinto AP dle La Cordillera d re een Transecto Parque Los Nevados (Intro- ducion y datos iniciales). Stud. Trop. Andean Ecosyst. . Cramer, Vaduz. VENEKLAAS, E. J. Rainfall interception and r trient fluxes in Colombian montane tropical lali est. Ph.D. Thesis. University of Utr cht. 1991. montane tropical rainforests, Colombia. J. Trop. Ecol. 319-336. 7: Vis, M. 1989 Processes and Patterns of Erosion in a Ye. jns Watson, M. А. 1980. Patterns of habitat occ upa an in mosses — relevance to considerations of the п Bull. Torrey Bot. Club 107: 346-372 WEERT, R. VAN RANGO S. & prevención de дене, ку saneamiento ambiental del rio Otun. Z pon Hidrologia e hidráulica. ecu cate ee Perei WHITTAKER, R. H. Vegetation a Siskiyou Mountains, PAR and California. Ecological Mono- da ge 30: 279-338. кшп and measurement of species dvo. Taxon 21: 20 1. 1977. ae of species 5 in land communities. Pp. 1-67 in M. K. Hecht, W. C. Steere & Wallace (editors), элли Biology, vol. 10. Plenum, New York WiLson, M. У. & A. SHMID A ke beta col: 72: 84 diversity with presence- ems data. J. 055-1064. WOLF, Comunidades epifitas en un transecto altitudinal en i Cordillera Central, Colom- ia: datos iniciales sobre la cantidad de especies de briofitos y ce Pp. 461 Hammen, S. Di э) La Cord e Los Nevados (Segunda кык ыы бы . Cra Parte). Stud. er, Berlin, Stuttgart. Epi туе communities of tropical montane Па in the northern Andes. 1. Lower montane communities. ушан 22(1) 1- 2 Epiphyte communities of tropical ests in the northern Andes. II. Upper montane communities. Phytocoenologia 22(1): 53- 103. montane rain for CHROMOSOME CYTOLOGY OF Peter Goldblat? and Masahiro Take? THE AFRICAN GENUS LAPEIROUSIA (IRIDACEAE-IXIOIDEAE)' ABSTRACT Chromosome counts, now available for 33 of the approximately 40 a " мерс show the ees to be remarkably variable cytologically. Excluding polyploidy, numbers range fro 10 to 9, 8, 7, 6, 5, 4, and 3 in what we believe to be a descending dysploid series. A cycle of polyploidy and аа descending. ео appears he section th to have Via ga in the Alou African section Paniculata. p 4 and рар populations with n Although chromosome numbers var osome material appear to be conserved, indicating the dica and populations with п = Пн material as the п = 4 karyot уре. large chromosome pair and me amount o fusion o s. А phylo ge of B chromosomes is evident in species with higher chro — 6, the latter karyotype Е cane twice a small ¢ SLE from a ааа ancestral bimodal karyotype with one long ny for section Paniculata is proposed adis largely on e chan mber re both diploid s muc 4. erythrantha ol the s among the species, the single omosome nge. gu unusually high ro ot before detected, their irregular presence in populations probably explains past difficulty in ка ле та. numbers in several species of the genus Chromosome cytology is an important factor in considerations of the systematics and phylogeny of Lapeirousia, a widespread African genus of Iri- daceae subfamily Ixioideae. The estimated 40 spe- cies of the genus (Goldblatt & Manning, in prep.) are currently divided among two subgenera, sub- genus Lapeirousia with sections Lapeirousia and Sophronia, and subgenus Paniculata with sections Paniculata and Fastigiata (Goldblatt & Manning, 1990; Goldblatt, 1990a; Goldblatt & Manning, 1992). That the chromosome cytology of Lapei- rousia is extremely variable is well known (Gold- blatt, 1990b): excluding polyploidy, base numbers range from x = 10 to 3. Such variability is re- markable in plants in general, including Iridaceae, most genera of which exhibit little or no variation in basic number (Goldblatt, 1) ased on outgroup comparison, ancestral base => number in Lapeirousia is most likely x — 10, bu species of Lapeirousia with 2п = 20, 18, 16, and some with 2л = 12 or 10 are distinctive in having strongly bimodal karyotypes with one long chro- mosome pair and nine to four smaller pairs less than half as long as the long pair. Only in the predominantly tropical African section Paniculata are there departures from this pattern, and most species have karyotypes with chromosomal rear- rangements that have resulted in derived base num- f 4, and 3. Among these is the widespread and diverse L. erythrantha, which until now has been thought t to ў a hypotetraploid with ers of x = 6, 5, a secondary base of n derived from ancestors with n — 4 (Goldblatt, 1990b). New counts pre- sented here for L. erythrantha confirm 2n — 12 in some Zimbabwean populations, but we have found other populations both in Zimbabwe and Zaire with 2n — 8. These and several more new counts of both tropical and southern African species shed new light on the cytology of Lapeirousia and show an unexpected frequency of B chromosomes in several species. We believe that B chromosomes are responsible for some of the apparent intraspe- cific chromosomal variability in base number re- ported in several species of the genus and uncer- tainty about the correct number in others (Goldblatt, 1972, 1990b). Consequently, we also review cy- tological patterns throughout the genus. MATERIAL AND METHODS Seeds or corms of wild-collected species (Ap- pendix) were sprouted in the greenhouse, and emer- ' Support for this study by gene BSR 8500148 and 8906300 from the U. k Jar en Hewson Swift, University of Chicago, for a gratefully acknowledged. We ne e cue of total DNA per genome in se S. National Science Foundat selected s rukoff, Curator of African B пне usa Garden, P.O. Box 299, St. Louis, Missouri 63166, US D^ * Biological Institute, Oita University, Oita 870-11, Japan. ANN. Missouni Bor. Garb. 80: 961-973. 1993. Annals of the Missouri Botanical Garden gent leaf and root tips were harvested for analysis of mitosis in one of two ways. For rapid determi- nation of chromosome number, root tips were pre- treated in saturated aqueous m-bromonapthalene for 4 hours at room temperature, then fixed in 3:1 absolute ethanol-glacial acetic acid for 5 minutes. After hydrolysis in 10% HCI at 60°C for 6 minutes, tips were squashed in FLP orcein (Jackson, 1973). For more critical analysis, leaf and root tips were pretreated in 0.002 M aqueous 8-hydroxyquinoline or 4-5 hours at room temperature, then fixed in modified Carnoy's solution (absolute ethanol-glacial acetic acid-chloroform 6 : 3: 1) at 8°C for about 24 hours. Both root and leaf tips were stained in Feul- gen reagent and squashed in 1% aceto-orcein. Pho- tographed preparations were all made using the latter technique. For purposes of comparison we included one species of Watsonia, which belongs to the same tribe Watsonieae as Lapeirousia, and Pillansia templemanni, only genus of Pillansieae, and outgroup to Watsonieae, in the study. Comparative genome size (Table 1) was esti- mated by calculating total chromosome length in metaphase karyotypes of comparably condensed chromosomes, always treated in exactly the same manner. Since chromosomes are not always con- tracted to exactly the same degree, this method is not completely reliable. It is, however, sufficient to distinguish different ploidy levels. Accurate cyto- photometric determinations made by Jane Master- son using /Vicotiana tabaccum as a standard (Table 2) confirm our hypothesis that the 2n = 12 cy- totype of Lapeirousia erythrantha is polyploid, as are the karyotypes of L. otaviensis and L. schim- peri. OBSERVATIONS Number and size. Chromosomes of Pillansia, Watsonia, and Lapeirousia are moderate to small in size (Figs. 1-25), with either modest, or, in most species of Lapeirousia, with considerable range in number and size (Table 1). Total length variation across Lapeirousia, excluding B chromosomes, is 0.9-7.6 е are comparable in size (Figs. 1, 2, A, B), but all fairly small, 0.7-4 um long. um. In Pillansia and Watsonia coccinea о вни number is consistently 2л = 2 in Lapeirousia sect. Fastigiata (Table 1; Figs. 3, 4), but in the other section of subgenus Paniculata, sect. Paniculata, numerical range is extreme, with the highest number 2n = 16 and the lowest 2n = 6 (Table 1). In subgenus Lapeirousia diploid num- bers of 2n = 18 and 16 occur in section Зорћгота (Table 1; Figs. 16-23) while all species of section Lapeirousia so far counted have Zn — 16 (Figs. 19, 24, 25). With decrease in number there is an apparent gradual increase in size of one or more chromosome pairs (Table 1 Karyotypes. Bimodal karyotypes character- ize Watsonia and the majority of species of Lap- eirousia, including all those of subgenus Lapei- rousia and subgenus Paniculata sect. Fastigiata. Although four longer chromosome pairs also stand out in Pillansia (Fig. 1, 27A), size differences are too gradual for the karyotype to be characterized as strictly bimodal. In Watsonia (Figs. 2, 27B) there are two long chromosome pairs but always just one long pair in bimodal karyotypes in Lap- eirousia. In Lapeirousia these long chromosomes are 4.1–6.5 um long, and the remainder are gen- erally 1-2 um long, arbitrarily characterized here as small or medium (Table 1). We are uncertain of the reason for the variation in the length of the long pair among the species, although we believe that this pair is homologous across the genus be- cause of its distinctive prophase morphology (see below). Differential contraction at metaphase due to preparation or enlargement of the chromosomes across the entire karyotype could explain the dif- ferences we have observed. In species with bimodal karyotypes and 2n — 20 (section Fastigiata) all 18 of the smaller chro- mosomes are small (less than 2 um long according to our arbitrary criterion for categorizing the chro- mosomes by size). Bimodal karyotypes with 2n — 18 (some species of section Sophronia) have two of the smaller chromosomes appreciably larger than the remainder, and these fall into our medium category. In bimodal karyotypes with 2n — 16 (some species of section Sophronia and all species of section Lapeirousia examined) two pairs fall in our medium grou In section Paniculata the typical bimodal karyo- type with one long pair and many smaller ones occurs in Lapeirousia avasmontana (2n — 16), L. gracilis and L. neglecta (both 2n — 12) and 10) (Figs. 6, 7, 9). In L. gracilis two of the small chromosome pairs fall in L. sandersonii (2n — our medium category. Bimodality is less pro- nounced in L. setifolia, L. abyssinica, and the 2n — 8 karyotype of L. erythrantha (Figs. 10, 11). The karyotypes of L. rivularis and the 2n — 12 cytotype of L. erythrantha show no bimodality a 12, 13). In L. schimperi and L. otaviensis, 2 = 10, there is a secondary development of 9–7.6 ит long, about one-third again as long as the next longest pair (Figs. 14, 15). In L. schimperi this bimodality with a long metacentric pair 6. Volume 80, Number 4 Goldblatt & Takei 963 1993 Chromosome Cytology of Lapeirousia e © (2) RES 1-13. Mitotic metaphase in Pillansia, Watsonia, and Lapeirousia subgenus Panicula P. пат cn W. coccinea —3. L. corymbosa. —4. L. azurea. a L. g Dd cytotype ah 3 pe chromosomes. — 6. L. gracilis, cytotype with one satellite ‘chromosome. —7. L. ne nien — 8. L. coerulea. —9. sandersonii. — 10. L. setifolia. —11. L. erythrantha, 2n = 8 dix MS L. кы IN 2n = 19 cytotype. — 13. L. rivularis. Vouchers as given in Table 1. Scale bar, 5 и 964 Annals of th Missouri Botanical Garden ба: i `4 Jw = Ao O FIGURES 14-25. Mitotic metaphase іп Lapeirousia subg. Paniculata (14 and 15) and subgenus M nian (16-25). — 14. L. otaveinsis. —15. L. schimperi. —16. L. pyramidalis. —17. L. со 18. L. exilis. — 1 L. jacquinii. —20. a littoralis subsp. a — 21. L. odoratissima. —22. L. oreogena, cytotype with 2 la arge omosomes. — 23. L. oreogena, cytotype without : epe — 24. L. ala — 25. L. dolomitica subsp. lewisiana. а. as given in Table 1. Scale bar, Volume 80, Number 4 1993 Goldblatt & Takei Chromosome Cytology of Lapeirousia 965 FiGURE 26. coerulea. —D. bar, 5 um. Arrows indicate the long chromosome pair with dec appearance. „. erythrantha, 2n = 8 cytotype. — E. second pair has a prophase staining pattern that suggests homology with the long chromosome pair in the karyotypes of other species (Fig. 26). Lap- eirousia otaviensis has two karyotypes among the plants we examined. In one, presumably the an- cestral type, the chromosomes closely resemble those of L. schimperi, except that the second lon- gest pair is metacentric, whereas in L. schimperi the shortest pair is submetacentric and second lon- gest pair is acrocentric. Generally, only one satellite is visible, located on one of the two fourth longest chromosomes. The other karyotype in L. otavien- sis is heterozygous for a translocation involving one of the second longest pair and one other chro- mosome. There is thus only one member of the second longest pair and only one very short, nearly telocentric chromosome. This karyotype also nor- mally has only one satellite chromosome. Despite these irregularities, the similarity to the karyotype of L. schimperi is striking. Mitotic prophase in selected species of Lapeirousia. ns erythrantha, 2n — 12 de Ji^» L. neglecta. cytotype. — B. L. sandersonii. —C. L. — Е. L. schimperi. Scale Satellite position. A pair of small satellites, located on the short arms of the longest pair of small chromosomes, are frequently present in spe- cies of Lapeirousia with bimodal karyotypes (Figs. 3-10, 16-21). Two pairs of satellites are present in L. coerulea (2n — 8) and the 2n — 8 karyotype of L. erythrantha (Figs. 8, 11). In L. dolomitica subsp. lewisiana (2n = 16) one of the two pairs of satellites are very large (about as long as the long chromosome arms) and the short arms are minute; the second pair of satellites are also un- usually large and slightly exceed the short chro- mosome arms in length (Fig. 25). In L. azurea (2n = 20), L. silenoides (2n = 18), L. littoralis, and L. oreogena (both 2n — 16), all of which also have two pairs of satellites, both satellite chro- mosome pairs have short arms shorter than the satellites. The situation in L. gracilis is puzzling: two cytotypes (Figs. 5, 6) differing in satellite num- ber and morphology are present in the same pop- Missouri Botanical Garden Annals of the 966 c «9I 9I 0 7 Lee 81 '22 ne[qp[o?) (шет) гурртролла 7] © #=91—› 9I 0 © 626 81 "11426. DIOfNSNTUD “7 © —9Ii— 9I 0 © — 82 + 81, "месу Jay (FT) = әәир T piuosydog чоцоәс тополодр ју snusBqng ГА 0 Ӯ 0 0 9 = 9 E = — 0 Y 9 = OT ioxeg пзгитод "7 r4 v v 0 9 v 8 LP OI 191904] 7) "M sisua1anjo “7 Y Y © 0 Ӯ 9 Га 01 "qpeyr-oupy (перу 29 “os y) мә0штуэѕ 7T OI с 0 0 9 9 GLP ZI diojuue AA suoma 7] ZI 0 0 0 Y 8 Urs (d£ = 0 +) ст T v 0 © 4 © SVS 8 ломед (NEY хә Yyosziopy) ругиолујАло "7 9 [А 0 G t с SS g suueH »5gofnuos 7] G 9> 0 у Y 913 8 ZUIYIS рајпагоз "T И © © © С Y = 8 Jayeg (Чогу ^y) noiuissKqn ^T G —8> T у {д 9'86 01 lexeg 12и052әри?ѕ "T [4 —Ol— OI 0 © 6 96 cI ne[qp[os) 2129182и "T G —01— 9 + © 2'9% ZI јадпед sim2D48 7 G er> VI 0 [д 91 19] DUDJUOUSDAD ^T отпора чоцэәс © -g> 8I 0 с x 0с Jayeg (VEY хә Joka 7) ругирлотш 7] © —81— 81 0 ГА 0с [MPD Jay (FT) »ivoppf 7T © —81— 81 0 © [4:4 0c n*[qp[o2) (129xeg хә 23) релпгр ^T © —81— 81 0 © CIE 0c "[M?9) Jay СТ) DsoquiX402 "I 01019 70,4 UOI DID¡NI1UDJ snuesqng pisnosad о] e uis ш S W T шт шт satvedg удио Jaquinu ABojoydiop m MALES mre. sainjeay adA10A1ey [е10], 'sousueis aAnduosep 10 YISUA| ошовошолцо [910] jo sjuauranseour оцу ut рарпјош jou әле sourosouro1qo g 'попеледола јо poyu jua1o yip ay} 01 әпр seroeds aures əy} 103 (q0661) HPIAPIOS jo siueuramseour oy) чим o[qereduroo Арәәлр jou әле VA "19uueur aures eui Ápoexa ut paredaid *sadA10A1e3 paude1sojoud јо jueuroinseour 1eoum Aq peuruliejap sem y}ĝuə| ошовошолцо [LIO], 'ајаејтеле jou едер = — 'ошиаоолов = р ‘іциәоејәшдпѕ = urs *oujuaoejaur = ш {шт z>) gews = $ (un +–7) штрәш = y (un p<) әдет = T :suogerae1qqy. "juris oidKjoK1ex ој Surp1099e suonoas uyum родџелле are зогодс (20661 '2761 '1761 “пеүарүох)) роцчецапа Ајепоглола әѕоці Ҷим peutquioo әле (q xipuaddy) әләч poureiqo sunog “visnosadv7 ш soinjeoj edKjoKiex pue (ројфр) quo AUIOSOUIOIYI [£10 “SISUINU ourosouroIq^ ү 318V] 967 Chromosome Cytology of Lapeirousia Goldblatt & Takei - Volume 80, Number 4 1993 'seurosouloIQo ZZ se Áueu se pue QT ве мој S? чим s[[o9 рорпјош sjunoo [eutZu() *q9 — ( + 91 = UZ Арооллоз Ајдедола xung ом mq (20661 “1261 ‘VEIFPIOD) OZ = UZ pue oz "89 = uz se роџодол А|еш и) , bern рәшшехә se поцејпдод рәриәјхә oures ay) шолу aie әләц pojunoo s[enprarpur *seurosouro1qQo g jo зоповола лејпвалл əy} 01 onp Ájurej1o2un ројзођол ÁA[qeqoid (7/61 “1261 UPIAPIOD) 81-91 ‘ZI L? = uz JO siunoo Joye ç “AZ + 91 = uz реч A[qeqoud 81 = uz Зшлец se paysodas suone[ndod omy , 98-0 + 81 = ис 9q Арооллоо Á[qeqo1d p[nous asayi—(0Z-)81 pue OT "89 ‘81 = ис рорпјрош sjunoo Joye , “AZ + 81 = ис q Арзеллоз p[nous (#0661 ^n*[qpo9) OZ = ис Jo Modes лоцлез əy} ўшщу ом ‘satoads aui 10} 42 — 0 + 81 = VZ Jo 9194 sjunoo ому jo 1431] UJ z "гошовошолцо g Ajqeqoid aie зошовошолцо [еш Алол ом — gZ + QI = UZ Áurea sowe mq (20661 '1261 ‘VRIGPICD) OZ = UZ s? ponodəy | © —ујт— р» С = 91 ne[qp[oo »22n]o1a `7 © b> —7у1— © = 8I +91 "meg лом (әчәолеүә(] q) 12144»/ 7 [д =}її— 01 v с BPE 9I перо) (p10N ^g) puisina) :dsqns ó —viI OT Y © = 9I ротшојор :dsqns 19JUL(J VINTVUOJOP T с <—#1- p> с — 9I (220jfipun48 :dsqns se) nejqpjos »souids :dsqns ГА —viI —Vilói © — 949 — 0 + 91; отромрар 'dsqns ug “YN отоирар 7] © ep —vil б 91 loxeg 113109 `T с Gl © © voc 91 "905 270214240 `T pisnosiad pT] чоцэәс̧ G —VrI GI © © — 9I pypayd :dsqns sjeiq ( boef) vivod 7 © Pl cI © © 8'8c 852 — 0 + 9I пвјарјод) Vuasoaso "T 6 cl © cl 6 © VIE +91 Jayeg DUNISSIJVLOPO "T 6 «= у [=> cl 6 6 GE 9I перо) (21426) n;npnn2 :dsqns 6 ep => СІ 6 © = 9I строју “dsqns Jayeg 817020111] 7 © —91— 91 0 (д LSS «gc — 0 +) 81 дам nmumbovf 7] © «gr 9I 0 с 8'6c (dc — 0 +) 81 AA ap 7] © поре 91 = © = 81 пејарјод ppunoa42a `7 G —91— 91 0 G 126 «de — 0 + BI [Mey лом (boe) зәртоиәц 7T e ws ul © IN 1 ш шт setvedg duo ладшти A30]0y d ој 798 е ови а волтцвој одАјоХ лем oL гропициој | ЯПЧУ][, 968 Annals of the Missouri Botanical Garden TABLE 2. Genome sizes estimated for selected species of Lapeirousia following the classification of Goldblatt (1990). All C-values have been reduced to the 1С level. asurements were made cytophotometrically by Jar Masterson with Nicotiana eus cum as a standard; ue represent the means of 10 measurements. Mean Chromosome Mean 2C-value number ] C-value Taxon pg 2n L. erythrantha 21 8 1.35 5.8 12 .45 L. gracilis 2.4 12 1.20 L. otaviensis 5.09 10 1.27 L. schimperi 5.08 10 1.27 ulation. Some individuals have four satellites: one pair, on the second longest chromosome pair, are comparatively long, slightly exceeding the short arms in length; a third satellite on chromosome 3 is also fairly large; and a fourth satellite, extremely large, is located on one of the shortest chromo- somes, the short arm of which is minute. Other individuals have small satellites on the second chro- mosome pair (the most common and presumably the basic condition in the genus). Prophase morphology. Їп all karyotypes one long acrocentric chromosome pair, often the only long pair in the entire complement, has a distinctive prometaphase structure (Fig. 26). Largely hetero- chromatic and darkly stained, a small subterminal portion of the long arm is euchromatic. So dis- tinctive is its appearance that we assume that this long chromosome pair is homologous across all the species so far examined and that it is maintained intact, whereas some (or possibly all) of the other chromosomes have been involved in structural re- arrangements. The small chromosomes, found in the majority of species, are largely heterochromatic except for the distal ends of the longer arms, which are euchromatic and lightly stained until metaphase (Fig. 26) viensis profound structural rearrangement of the karyotype makes identification of the conserved n Lapeirousia schimperi and L. ota- long chromosome pair of most other species diffi- cult, but, at least in L. longest pair seems homologous with the conserved schimperi, the second long pair of most other species (Fig. 26F). Total chromosome length. Irrespective of chromosome number, the majority of species of Lapeirousia examined have a total chromosome length in the 24-29 um range (Table 1), which we assume is the diploid level for the genus. Total chromosome length is close to twice this amount in the 2n — 12 cytotype of L. erythrantha (54.1 um), but not in the 2n = 8 cytotype of the species (24.5 um) (Table 2). This doubling is also found in L. rivularis (47.2 um), L. schimperi (47.1 um), and L. otaviensis (47.8 um). We assume this rep- resents cryptic polyploidy (1.е., polyploidization ei- ther followed or preceded by descending dysploidy). Absolute DNA amount per genome (Table 2) is available for the two cytotypes of L. hie he and for L. gracilis, L. otaviensis, and L. - peri. These determinations confirm the estimates based on total chromosome length. Lapeirousia otaviensis, L. schimperi, and 2n — 12 L. ery- thrantha have 5.1-5.8 pg DNA per genome, whereas L. gracilis and 2n — 8 L. erythrantha have 2.7-2.8 pg per genome. Total chromosome length falls some 20-40% above the expected diploid level (Table 1) in a few species (L. corymbosa, L. azurea, L. littoralis, L. dolomitica), but we doubt that these species are polyploid because they have karyotypes nearly identical with obviously diploid species. Some other phenomenon must be involved in the increase in total chromosome length in these examples, which seems beyond what might be considered experi- mental error. Pillansia templemanii is obviously tetraploid. Our counts confirm tetraploidy already reported in two other populations (Goldblatt, 1971, 1977) of this fairly narrowly distributed species. B chromosomes. А significant part of the karyotypes of some species, B chromosomes are identified in. Lapeirousia by precocious and uni- form dark staining at prophase. They vary consid- erably in size, being nearly as long as the long chromosomes in L. oreogena (Fig. 22), but more commonly t are about as long as the sma е AN three populations of L. silenoides now known cytologically appear to have B chro- mosomes bdo 17), which are also fairly common in the = 12 cytotype of L. erythrantha. We detect no mala geographic or ecological pat- tern to the distribution of B chromosomes in the genus but note that B chromosomes have so far been identified only in species with higher diploid numbers. Because of their frequency, we have con- cluded that some past counts in the genus, incon- sistent with our current findings, are due to the unrecognized presence of B chromosomes. The pre- sumed correct chromosome numbers are included in Table 1 with footnotes explaining our reasons for assuming past errors. Goldblatt & Takei 969 Chromosome Cytology of Lapeirousia Volume 80, Number 4 1993 eri A 98 Ба nr adr cMitas B Él Risazicauriias avs С GOS stdssacszesussss: Ы D TAN E — osiassssn: F ТТЕ "m FicunE 27. temple mani. W. coccinea. 3 —: 10).— -G. L. coeru 2). L. sandersonii (2n = (2n = 12).—J. L. schimperi (2n = 10). — К. L. silenoides 5 um DISCUSSION Based both on outgroup comparison, and on the more common trends in numerical change in chro- mosome number of plants (Jones, 1974, 1977) we infer that the ancestral base number in Lapeirousia is most likely x = 10. Lapeirousia is currently regarded as a member of tribe Watsonieae (sub- family Ixioideae) (Goldblatt, 1989, 1990c, 1991) in which other base numbers are 2n — 20 (Ther- eianthus, Micranthus), 18 (Watsonia), and 16 (Savannosiphon) (Goldblatt & Marais, 1979; Goldblatt, 1989). Pillansia, only genus of Pillan- sieae, most primitive of the three tribes of Ixioideae, = 8).— АРТ Ун 585$ a HITHEHEFEEELELLE L —ÓÓ Idiograms of ia of Pillansia, rcu and selected species of Lapeirousia. — A. .—D. E. 1 , gracilis (2n — 12).- glecta (2n — n = 8).—I. L. 1 = 16). = L. erythrantha (2r (2n = 18).—L. L. odoratissima (2 erythrantha Scale bar, ~ also has x = 10 (2n = 40) (Goldblatt, 1977), an observation confirmed here for two additional pop- ulations of the species. Moderately bimodal kary- otypes characterize both Micranthus and Watson- ia (Goldblatt, 1971). The karyotype of Lapeirousia differs from that of Micranthus largely in its more pronounced bimodality, but the patterns seem com- parable. In Watsonia, with 2n = 18, there are two long and seven short chromosome pairs. Within Lapeirousia differences in number and karyotype seem best explained by reduction in number, accomplished by fusion and elimination , unequal reciprocal 1974 of redundant centromeres (i.e. or Robertsonian жаназа see Jones, 970 Annals of the Missouri Botanical Garden Ф Ф < e S e Ñ M o so O Ф ме v Nd S d M S & © E e € >» — Ф n А S » 0 S Ко S ES Kà S aS OS S Кш oe S Fu Y x E Q^ e n L Sog £ & ~ e ^ у 2 y oa > N e © ә с vov ~ q^ ш e e —— Ј п=3 * * * 3 n=8 n=8 n=5 ¡actinomorphy long tube anicle leaf plicate corm tunics woody bracts small n=9 polyploidy n=6 dysploidy n=3 subg. Fastigiata subg. Lapeirousia rn-9 corms flat based (non bimodal) bimodality п=5 п=6 zygomorphy FIGURE 28. Diagram of relationships of L -apeirousia posed on a topology established by Goldblatt and Manning (1990). Although both Watsonia and Lapeirousia sect. Lapeirousia have n — 9, they are not homologous char- acters: the karyotypes indicate independent derivation in each lineage. 1977). This hypothesis is also supported by the observed increase in size of one chromosome pair as number decreases (Table 1). The infrageneric phylogeny established by Gold- blatt and Manning (1990) for Lapeirousia makes it clear that decrease in number has occurred in subgenera Paniculata and Lapeirousia indepen- dently (Fig. 28). All species of subgenus Lapei- rousia have numbers based on x — 9 or 8, and the only counts so far obtained for section Lap- eirousia, excluding B chromosomes, are 2n — 16, a number also present in some species of section Sophronia. lt is uncertain at present whether п = 8 was achieved more than once in subgenus Lap- eirousia. Chromosomal data suggest that section Sophronia may be paraphyletic with section Lap- eirousia nested within it. This question awaits fur- ther investigation. In subgenus Fastigiata, section Fastigiata con- sistently has 2n = 20, but the lowest number in section Paniculata is 2n = 16, in L. avasmontana, one of two species in the section with actinomorphic flowers (Fig. 29). Among the remaining species of the section relationships are obscure except for a small clade of species with pale-colored flowers with long perianth tubes (L. schimperi, L. otaviensis, and L. bainesii). Phylogeny of the section must therefore be largely inferred from the karyology. n=8 2n-20-16 ———— FIGURE 29. d of inferred relationships i in Lap- eirousia Paniculata. The tree is ndn А оп chromosomal a but ncludes apomorphic m phological characters established by Goldblatt (19908). Decrease in number appears to have proceeded stepwise with one or more species at each step (Fig. 29), with a plateau at 2n = 8, where the central species is L. erythrantha (previously thought to have exclusively 2n = 12). Lapeirousia coerulea, which also has 2n — 8, has actinomorphic flowers, almost certainly a reversal, given its nested position within a clade of largely zygomorphic-flowered spe- cies. The situation in Lapeirousia erythrantha, with populations having either 2n = si is > partic- ularly interesting. Past pied r the es have consistently been 2n — 2 (Goldblatt, 19908). and we have confirmed ы count in two additional populations. The 2n — 12 cytotype is the more widespread, occurring exclusively in Malawi and Zambia, and also in some Zimbabwean populations, ut populations with 2л = 8 are scattered across Zimbabwe and Zaire (Fig. 30). On the basis of total chromosome length, L. erythrantha, then only known from its 2n — 12 cytotype, was assumed to be polyploid, as was L. rivularis. We now modify this assumption. Ancestral base number in L. er- ythrantha is most likely x — 4, the base number also in the morphologically similar, and presumably closely related, L. оа and L. ] (Goldblatt, 1990b). The 2n = erythrantha has been d to be diploid, апа Volume 80, Number 4 1993 Goldblatt & Takei Chromosome Cytology of Lapeirousia the 2n — 12 karyotype tetraploid by measurement of total DNA genome. The diploid number of 2n — 12 may have been achieved by one of two possible paths. Dysploid reduction from n = 41 y Robertsonian trans- location followed by ан to n = 6 is one scenario. Alternatively, a polyploid event re- ) = 16, and subsequent 12, may occurred. We favor the former hypothesis because sulting in tetraploidy, 2n (4x dysploid reduction to 2n (4x) — have it involves only two major chromosomal mutations, whereas the second involves three. In L. erythran- tha polyploidy and subsequent dispersal of the 2n = 12 cytotype through Zambia, Malawi, and northern Zimbabwe is our best explanation for the cytological pattern now evident in L. erythrantha. There seems to be no doubt that the two cytotypes of L. erythrantha are conspecific. Except for a slightly more slender growth habit in some of the n = 8 cytotypes, there is nothing we can detect in their morphology to distinguish the cytotype in this common and widespread species. In Lapeirousia rivularis, a similar karyotype, although not identical to the 2n = 12 cytotype of L. erythrantha, is also evidently cryptically poly- ploid. The origin of this karyotype is uncertain, but it seems to us more parsimonious to assume a derivation from the 2n = 12 karyotype of L. er- ythrantha. This presumes also the derivation of a species, L. rivularis, from a cytotype of L. ery- thrantha. Alternatively, L. rivularis may have arisen from other ancestors presumably with x — 4 (or 3), with subsequent polyploidization so that the karyotype now resembles that of the 2n — 12 cytotype of L. erythrantha. Our results indicate that Lapeirousia schim- eri, L. otaviensis, and presumably the related L. bainesii, which form a clade (Goldblatt, 1990a) distinguished by their long- to very long-tubed flow- ers, are also polyploid (Tables We cannot say whether they are more likely to be derived from L. erythrantha or to have another ancestry, but parsimony is followed in Figure 29, in the tree generated by combining cytology and morphology of the lineage. Continuing studies will eventually yield more light on the question of phylogeny of Lapeirousia. It is, however, clear that the karyological history of the genus is remarkably complex and has prob- ably involved primary dysploidy in at least two lineages, from x — 10 to 8 in subgenus Lapeirou- sia, and from x = 10 to 4 in section Paniculata. A secondary burst of polyploidy is evident in section Paniculata, and secondary descending dysploidy appears to have occurred in L. erythrantha (and = © ~ e [^] о RE 30. Cytogeography of Lapeirousia егу- hrantha. Populations with 2n — 8 represented by stars; populations with 2л = 12 by solid circles. ~ its close ally L. rivularis) and perhaps indepen- dently in the clade that includes L. schimperi, L. otaveinsis, and L. bainesii, the latter having pop- ulations with n = 6 LITERATURE CITED GOLDBLATT, P. 1971. Cytological and Oa al studies in the southern African Iridaceae. J. S. Af- rican Bot. 37: 317-460 1972. A revision of the genera Lapeirousia блнт et and Anomatheca Ker in the winter eur region of South Africa. Contr. Bolus Herb. [= 111. Chromosome cytology and “т changa i in Galaxia (Iridaceae). Pl. Syst. Evol. 61-69. Chromosome мн и Pillansia (Ir- idaceae). Ann. Missouri Bot. Gard. 64: 136-137. 1 Cytology and Е of the Мо- raea fugax complex (Iridaceae). Ann. Missouri Bot. Gard. 4 — А ы ha "rn n African genus Watsonia. Аий: Kirstenbosch Bot. Gard. 13. 1 990a. iuris of Lapeirousia (lrida- ceae- бојеве in tropical Africa. Ann. Missouri Bot. Gard. 430-484. . Cytological variability in the African genus pin КА (Iridaceae-Ixioideae). Ann. Mis 82 souri Bot. Gard. 77: 375-382. 1990c. Phylogeny and dlissinceton of Iri- daceae. Ann. Missouri Bot. Gard. 607-627. 1991. An overview of the s ен phy- logeny and pend of the African Iridaceae. Contr. Bolus Herb. 13: AANNING. 1990. Leaf and corm struc- ture in Lapeirousia (Iridaceae- Ixioideae) in relation to d x eny and дни classification. Ann. Mis- . Gard. 11: 365 7 992. Мркови ке of the southern African Lapeirousia corymbosa (Iridaceae-Ixi deae) complex (sect. Fastigiata) and a new species 972 Annals of the Missouri Botanical Garden of sect. Paniculata. S. African J. Bot. 58: 326- W. Marais. 1979. Savannosiphon реп. nov., a segregate of Lapeirousia (Iridaceae- Ixioi- b : deae). Ann. Missouri i binis 66 open HOLMGREN, P. K REN & L. С. 1990. Index Herbariorum. Part 1 ed. 8. ш JACKSON, R. 1973. Chromosomal evolution in Haplo- рарриз gracilis: a centric transposition race. Evo- lution 27: 243-256. JONES, К. hromosome evolution Ьу Robertson- ian translocation i in Gibasis (Commelinaceae). Chro- mosoma (Berlin) 45: 353-368. 1977. The role of Robertsonian change = ны A om in higher plants. CUR E : 121-129. Today Volume 80, Number 4 1993 Goldblatt & Takei 973 Chromosome Cytology of Lapeirousia APPENDIX. New chromosome numbers in Lapeirousia, Pillansia, and Watsonia with collection data. Karyotype details are included in Table 1. Acronyms (abbreviated according to Holmgren et al., 1990) following the collection data refer to the herbaria in which the vouchers are housed. Species are arranged alphabetically within sections. Species Diploid number 2n Collection data Pillansia templemanii Watsonia coccinea Lapeirousia Subgenus Paniculata Section Paniculata L. erythrantha L. gracilis L. rivularis L. sandersonii L. neglecta L. otaviensis Section Fastigiata L. corymbosa L. azurea Subgenus Lapeirousia Section Sophronia L. angustifolia L. exilis L. jacquinii L. silenoides L. odoratissima L. oreogena L. pyramidalis Section Lapeirousia L. arenicola L. dolomitica subsp. lewisiana L. violacea Goldblatt 40 18 16 + 0 — 2B 16 16 South Africa. Cape: Kogelberg, Bean s.n. (MO); Palmiet River valley, Viviers s.n. (BOL) South Africa. Cape: Goldblatt 9055 (MO) Zaire. Mine d'Etiole, Brooks s.n., no voucher; Zimbabwe. Dombashawa, Goldblatt 9086 (MO); north of Rusape, Goldblatt s.n. (MO); near Bulawayo, Goldblatt s.n., no voucher Zimbabwe. Christon Bank, Mazowe, Peters s.n. (MO); Ha- rare, Cleveland Dam, Goldblatt s.n., no voucher. Zam- bia. Mufulira, Goldblatt 7575 ( Namibia. Asab, Goldblatt & Manning 8870 (MO) Zambia. Lusaka, Goldblatt 7537 South Africa. Transvaal: Pretoria Hills, van Wyk s.n. (MO) South Africa. Cape: Jonkershoek, Esterhuysen 36923 (MO) Namibia. Otavi Hills, Goldblatt & Manning 8837 (MO) Е = South Africa. Саре: Cape Point Reserve, Goldblatt s.n. (MO) South Africa. Cape: Malmesbury, Goldblatt 8016 (MO) South Africa. Cape: Knersvlakte, Goldblatt & Manning 9127 (MO) South Africa. Cape: north of Springbok, Goldblatt 2649 (MO) South Africa. Cape: north of Garies, Goldblatt & Manning 9135 (MO) South Africa. Cape: Near Steinkopf, Goldblatt s.n., no voucher; south of Kamieskroon, Goldblatt 9136 (MO) Namibia. East of Windhoek, Goldblatt & Manning 8803 (MO) South Africa. Cape: Nieuwoudtville, Snijman s.n. (MO) South Africa. Cape: Olifants R valley, Goldblatt & Man- ning 9227 South Africa. Cape: Knersvlakte, Goldblatt 9145.4 (MO) South Africa. Cape: Anenous Flats, Goldblatt & Manning 9460 (MO) South Africa. Cape: near Garies, Goldblatt & Manning 9198 (MO) South Africa. Cape: Bidouw Valley, Goldblatt 9197 (MO) CHROMOSOME NUMBERS OF NEOTROPICAL CASTILLEJA (SCROPHULARIACEAE: TRIBE PEDICULAREAE) AND THEIR TAXONOMIC IMPLICATIONS! T. I. Chuang? and L. К. Heckard}* ABSTRACT otic chromosome numbers in Castilleja are iid лењи for 105 collections representi ig 34 species, mostly from > cea: with a few from iin ч = 5 reported here for the first time: С. brya Еу есиа integrifolia, С. iraz uensis, C. ini olia, ( 5 C. stenophylla, C 12); five are tetraploid (n= = 12, 24, and 7 troublesome taxonomically and substantiate for the complex morpholo Castilleja с. E; western North American rep ~ tenuifolia is sa ulat merica. Chromosome number . talamancensis, C. tenuifolia, má ras 0 24) only; two are octoploid (n = 48) only; FAM, have two or more levels of ploidy (n 2). These polyploid species (including C. aspera, C. s of the following integrifolia, and C. tenuiflora) are the most our contention that polyploidy a and hybridization are largely responsible ical variation found in Castille ~ subinclusa) exhibi polyploidy, in contrast to more than 50% polyploidy in e unusual chromosome n | be an aneuploid reduction from an n — eja. It is estimated that only 25% of Mexican elements of mber of n = 10 in the closely related C. rogenitor. The role of chromosome нар а nolyploidy. in the evolution of Castilleja and its taxonomic significance is indicated for individual specie This paper represents the fourth in a series of reports on chromosome numbers in Castilleja (Scrophulariaceae) and is part of our continuing effort to furnish evidence of the significance of polyploidy and accompanying hybridization in forming the complex patterns of variation that characterize the genus. It is also our intention to clarify taxonomic and evolutionary relationships among its members. Our emphasis in this report is on those species occurring in Mexico, Central America, and South America. Castilleja (Indian Paintbrush) comprises over 200 perennial and annual hemiparasitic species. They are almost all American with their heaviest concentration in western North America, where they have been a continual source of extreme tax- onomic difficulty, perhaps due partly to the prev- alence of weak barriers to natural hybridization and associated polyploidy. Asa Gray’s (1862) com- ment remains discouragingly prophetic: “Тһе spe- cies of this genus are most troublesome and un- satisfactory, not only on account of the difficulty of investigating the dried specimens, but also from the variability of the characters which have been relied upon in arranging them, and especially of the calyx." Heckard (1968) reported chromosome numbers of 54 taxa in Castilleja, Piin) of the Pacific Coast states and Mexico. Heckard & Chuang (1977) reported chromosome Sube of 35 taxa from the Great Basin and Rocky Mountains, and Chuang & Heckard (1982) reported 13 taxa of annual Orthocarpus (sects. Castillejoides and Cordylan- thoides) mainly from California, which have re- cently been merged into Castilleja subg. Colacus sect. Oncorhynchus (Chuang & Heckard, 1991). Our continuing survey of chromosome numbers in the genus in western North America (Heckard, ' We are deeply grateful to Robert W. € Lauram may T. Der ond ise H. Holmgren, and investigation. We also express our sinc the us and helpful suggestions, to Fei a loan of 21 specime sides пе of Biolog den in particular, and Dennis E. Breedlove, Robert eid Moran, who kindly collected ачна material for this 'ere nm. to Lincoln Constance au zhuang for skillful technical assistance, and t t Bye, Annetta Carter, Silva for dar reading of oG . Nesom for а ( Q. State University, Normal, ueni ови U.S.A. * Jepson Herbarium, University of C alifornia, Berkeley, California 94720 ANN. Missouni Bor. Garb. 80: 974-986. 1993. Volume 80, Number 4 1993 Chuang & Heckard 975 Chromosome Numbers of Neotropical Castilleja 1968; Heckard & Chuang, 1977; Chuang & Heckard, 1982; Chuang & Heckard, unpublished data) has shown that more than half of the 100 species counted are polyploid or have a diploid level plus one or more polyploid levels. For ex- ample, one species complex, Castilleja affinis-li- toralis, has a diploid plus five levels of polyploidy from 4x to 12x. The widespread C. miniata of the western United States and Canada has a diploid and four additional levels of polyploidy from 4x to 10x. Castilleja is represented in Mexico by approx- imately 55 species, Central America by 9 species, and South America by 12 species, mostly in the Andes. Judging from its richness in species, main- land Mexico, especially the Sierra Madre Occiden- tal, is one of the major centers of diversity for the genus, second only to the western United States. Mexican and Central American Castilleja was last treated monographically by Eastwood (1909), who recognized 54 species, 17 of which were described as new. Although her classification is used as a basis for our review and presentation of chromo- some numbers, it is evident that there is need for an updated treatment of Mexican Castilleja based on the abundant recent collections in the major herbaria in Mexico and the United States, field observation, cytological investigation, and SEM studies of seeds and trichomes. Subsequently, many new species of Mexican, Central American, and South American Castilleja were proposed by var- ious authors. Brandegee (1914) described one from Chiapas, Mexico; Standley, three from mainland Mexico (1936, 1940), one from Costa Rica (1938); Standley & Steyermark (1944), one from Guate- mala; Pennell, four from Panama (1940), three from Venezuela (1953); Edwin (1970), four from Peru; Crosswhite (1970) and Rzedowski (1975), one each from mainland Mexico; Holmgren, four from Mexico (1976), four from Costa Rica and Panama (1978), and one from Ecuador (Holmgren & Molau, 1984); Moran (1969, Levin & Moran, 1989), two from islands off Baja California; Breed- love & Heckard (1970), one from Sinaloa, Mexico, originally described as the monotypic genus Gen- trya, which has recently been relegated to sub- generic status in Castilleja (Chuang & Heckard, 1991). Most recently, an additional four species were described from mainland Mexico by Nesom (1992b) and another four from Peru by Chuang & Heckard (1992a), the latter included in our newly proposed bee-pollinated subgenus Colacus. There are only a few reports of chromosome numbers of Mexican Castilleja, consisting of scat- tered data on individual species (Beaman et al., 1962; Ward, 1983, 1984; Ward & Spellenberg, 1986) and four of our unpublished counts indicated on herbarium labels, which were listed in Holm- gren's (1976, 1978) and Holmgren & Molau's (1984) publications. MATERIALS AND METHODS Cytological materials were obtained from the wild by various collectors (see Table 1), and fixed in Farmer's solution (3 anhydrous ethanol: 1 glacial acetic acid, v/v). Fixed flower buds were imme- diately cooled on ice in the field and stored under refrigeration in the laboratory. All counts were made from aceto-carmine squashes of pollen mother cells and observed with a Zeiss phase contrast microscope. Chromosome drawings were made by camera lucida at magnifications of X 2100 and x 2600. RESULTS AND DISCUSSION In this contribution we present meiotic chro- mosome counts of 105 collections, representing 34 species in Castilleja from Mexico, Central Amer- ica, and South America of which 19 species are reported here for the first time. Among the 34 species represented, 24 are exclusively diploid (n = 12) and the remaining ten exhibit polyploidy: five (C. affinis, C. integra, C. lithospermoides, C. ornata, and C. subinclusa) are tetraploid (n — 24) only; two (C. linifolia and C. purpusii) are octo- ploid (n — 48) only; and three (C. aspera, C. integrifolia, and C. tenuiflora) have two or more levels of ploidy (n = 12, 24, and 72). The last three polyploid species are the most troublesome taxonomically and substantiate our contention that polyploidy and hybridization are largely responsible for the complex morphological variation found in Castilleja. It is estimated that only 25% of Mex- affinis subinclusa, which are undoubtedly Cali- ican elements of the genus (excluding C. and C. fornian elements) are polyploid, in contrast to more than 50% of western North American represen- tatives. Many more complex species are found in the western United States, where the genus appears to be undergoing an accelerated phase of evolution. The chromosome number of n = 10 is found in two closely related Mexican species, C. gracilis and C. tenuifolia, in sect. Epichroma Benth. (subg. Castilleja), although the same number was re- ported in C. lineariloba (Benth.) Chuang & Heck- ard (Chuang & Heckard, 1982, published as Or- thocarpus linearilobus Benth.), which has recently been included in subgenus Colacus sect. Опсо- rhynchus (Chuang & Heckard, 1991). It is pos- 976 Annals of the Missouri Botanical Garden TABLE 1. Chromosome numbers of neotropical Cas- TABLE l. Continued. tilleja. Voucher specimens are deposited in the Jepson Herbarium (JEPS) and the University of California Her- barium (UC), both at Berkeley, except where otherwise indicated. First reports for species are in boldface — affinis Hook. & Arn. ined D MEXICO. BAJA CALIFORNIA: 2 mi. SE of La Misión, N of ie ae 4070. Castilleja affinis Hook. & Arn. subsp. affinis, aff. C. subinclusa n = 24. MEXICO. BAJA CALIFORNIA: 2.5 mi. NW o jido Urupan, 8 mi. NW of dida Tomás, _ Ката 4080; 2 mi. uM. nta Clara, ca. 7 m of San Vicente, Hec kard 4081. Castilleja arvensis Schltdl. & Cham. п = 12. Mexico. CHIAPAS: 20 km SE of San Cris- tobal de las Casas, Cruden 1539 Castilleja aspera Eastwood п = 12. Mexico. CHIHUAHUA: W of Creel, Bye 7823; 3.2 km N of La Mesa del Hura- can, e.” 8039 (NY). DURANGO: 14 km W of Rio Mimbres, Ruta 40, Cruden 1301 Castilleja bryantii Brandegee нае 1) п = 12. | MEXICO. BAJA CALIFO : Rancho Saltillo, SE of Villa Солт, Hec e rd 3250. BAJA CALIFORNIA SUR: Aguaje San Antonio, Arroyo el Coyote, a 5459. Castilleja ecuadorensis N. Holmgren (Fig. 2) п = 12. ECUADOR. LOJA: Cuenca-Loja road, 26.2 N of Saraguro, Holmgren 10170. Castilleja fissifolia L. f. (Fig. 3) ECUADOR. CARCHI: El Angél-Tulcan road, 6.7 km n=12. , 3.9 km up Refugio road, Holmgren 10105. PICHINCHA: trail to Volcán Ruca Pichincha, Holmgren 10120. Castilleja foliolosa Hook. & Arn. n = 12. MEXICO. BAJA CALIFORNIA: ca. З mi. SE of La Misión, N of Ensenada, Heckard 4072 Castilleja fruticosa Moran MEXICO. BAJA CALIFORNIA: diris i Is- land, S end mesa E of weather station, Moran 13768 (SD Castilleja gracilis Benth. Fie. 4) п = 10. MEXICO. MICHOACAN: ca. 10 km NW of Tuxpan, Cruden 1973. Castilleja hirsuta Martens & Galeotti (Fig. 5) EXICO. VERA CRUZ: ca. uw km NE of Perote, Cruden 1709; ca. 24 km NW of Jalapa, Cruden 1700. n = 12. n = 12 Castilleja integra А. Gray п = 24. MEXICO. M не ui km SSW of Bue- naventura, Cruden Castilleja integrifolia L. f. (Figs. 6, 7, 8) = 12. fexic O. CHIAPAS: 13 km SE of Comitan, Cristobal, Breedlove 25280 (CAS GUATEMALA. HUEHUETENANGO: ca. 15 km E of Paquix, Cruden 1578. MEXICO. OAXACA: ca. 24 km SSW of leen in Стидеп 1434; des ч Oaxaca to Pue to Escondido, 8 S of Jiquila, i dads 4735. MEXICO: n N of San Bartolo, Cruden 1352. MEXICO. GUERRERO: Zacatone, close to dore of Cerro Teotepec, Cruden 1635; 1 NW of Puerto El Gallo, Breed- e 36070 (CAS). Castilleja irazuensis Oersted (Fig. 9) = 12. COSTA RICA. CARTAGO: Volcán lrazü, Heckard 6219. n = 24. Castilleja jiquilpana Nesom n = 12. |. MEXICO. MICHOACAN: ca. quilpan, Cruden 1322 13 km W of Ji- Castilleja lanata A. Gray п = 12. EXICO. NUEVO LEON: desert ш Ма- ehuela ап ctor Arroyo, Cruder 1524. SAN LUIS POTOSI: ca d km E of 2 1461; ca. 41 km N of Nunez, Cruden 1525. Castilleja linifolia N. Holmgren (Fig. 10) n — 48. EXICO. DURANGO: 39 mi. N of Estación Coyotes, Breedlove 18747; 4 km NW o Los Angeles along road between Mazatlán and Durango, Breedlove 35738; ca. 29 km W of El Salto, Cruden 2000. Castilleja lithospermoides Kunth п = 24. MEXICO. JALISCO: ca. 18 km S of Autlán, Cruden 1672; 4.5 km W of Tizapán el Alto, Cruden 1320. NAYARIT: ca. 19 km 5 of Tepic, Cruden 1313 Castilleja mexicana (Hemsl.) A. Gray п = 12. MEXICO. CHIHUAHUA: 13-14 km SSW of Buenaventura, Cruden 2015; 6 km NE of intersection of roads to Santa Barbara and Guadalupe y Calvo, Cruden 2031. DURANGO: 6.5 km SW of Yerbanis, Cru- den 1849 Castilleja nubigena Kunth (Fig. 11) жр COTOPAXI: Parque Nacional Co- topaxi, 6.4 km u dien Refugio roa Holmgren 10101. APO: E side of Lagu- a Chaloacocha, ruis 10135. Castilleja ornata Eastwood (Fig. 12) n — 24. ns O. CHIHUAHUA: 32.2 km N of La a del Huracán, Holmgren 8044 (NY). Castilleja ortegae Standley (Fig. a п = 12. EXICO. DURANGO: 17 mi. W of La Ciu- dad, Breedlove 18 919. $ SINALOA: 3.5 mi. W of El Palmito, Breedlove 16425; 7 Volume 80, Number 4 1993 Chuang & Heckard 977 Chromosome Numbers of Neotropical Castilleja TABLE l. Continued. TABLE l. Continued. : a Rita, Sierra Ecc Breedlove 16: 220 (CAS); 5 mi. NW о os Ornos along road to Mocorito, P love 19192 (CAS). Castilleja patriotica Fernald п = 12. Mexico. CHIHUAHUA: W of Gonogochic, e a del Huracán, Holmgren 8034 Castilleja purpusii Aie n = 48+. MEXICO. ME ruo N slope of Pop Кова p Castilleja roei Crosswhite (Fig. 14) п = 12. MEXICO. D a. 29 km SW of La Em NE of El Palmito, Breedlous 30449 (CAS). Castilleja saltensis Eastwood MEXICO. DURANGO: 6 mi. W of La Ciudad, Breedlove 18867; ca. 8 km W of El Sal- to, Cruden 1305 Castilleja icio Hemsley n = 12 k to Valle de B Castilleja scorzonerifolia Kunth (Fig. 15) MEXICO. DURANGO: 9.7 km W of Duran- go, Holmgren 8059 (NY); Ruta. 2. са: 32 km SE Ruta 40, Cruden 18. GUERRERO: SW of Xochipala, с. ravo, Cruden 1355 1628. JALISCO: ca k of Guadala- jara, Cruden 1316. MEXICO: ca. 6 of Atlacomulco, Стидеп 1800. OAXACA ca. 20 km N of Ixtlan de Juarez, Cruden 1388. SAN LUIS POTOSI: up P 0- 231, ca. 34 km W of Rut , Cruden Castilleja stenophylla M. E. Jones (Fig. 16) MEXICO. CHIHUAHUA: W of Creel, Bye 7472; c . S of La Junta along road to San Juanito, Bye 7650; N of Hu- mira, Bye 787 1; ejido de San Ignacio Arare- n= 12 of Laguna de Bavicora, Holmgren 8017 (NY); 103.5 km SW of Hidalgo del Par- ral, Holmgren 8047 (NY) Castilleja subinclusa Greene subsp. subinclusa n = 24. MEXICO. BAJA CALIFORNIA: ca. 14 mi. E of of Meling Ranch, Heckard 4087. Castilleja talamancensis N. Holmgren (Fig. 17) = 12. Costa RICA. CARTAGO: Cordillera de Tala- manca, páramo, Heckard 62194. Castilleja tapeinoclada Loesener (Fig. 18) п = 12. | GUATEMALA. HUEHUETENANGO: ca. 19 km E of Paquix, Cruden 1584. Castilleja tenuiflora Benth. (Fig. 19) п = 12. Mexico. COAHUILA: са. 30 km SE of Salti- m ca. 13 km uta 105, Cruden 1 JALISCO: W of Tizapán El Alto, Breedlove 18680. MICHOACAN: 3 mi. S mi. N of Sombrerete, Breedlove 18723. MEXICO. CHIHUAHUA: on N side of Barran- ca de Batopilas, Bye 7779. DURANGO: 46 mi. N of Estación к. Breedlove 18776; 17 mi. W of La Ciudad, Breed- love 18914. MEXICO: 18 km NW of mares, Sierra Surotato, Breedlove 19280 (CAS Castilleja tenuifolia Martens & Galeotti (Fig. 20) п = 10. Mexico. MEXICO: 14 km SW of Villa 10 km NW of Tuxpan, Cruden 1972; 3 mi. SE of Quiroga, Breedlove 18706; ca. 26 km W of Jiquilpan, Cruden 1229. Castilleja virgata (Wedd.) Edwin (Fig. 21) n=12. | ECUADOR. TUNGURAHUA: Ambato-Guara da road on N flank of Volcán Chimborazo, Holmgren 10142 tulated that n = 10 may be derived from an aneu- ploid reduction from n = 12 progenitors, and is apparently of independent origin in these two sub- genera Knowledge of basic chromosome number and range of polyploidy has helped in understanding the causes of taxonomic complexity in this genus. Nonetheless, this information alone does not pro- vide sufficient evidence for making taxonomic judg- sectional, and s ments at the subgeneric, specific levels, because of chromosomal constancy at the 978 Annals of the Missouri Botanical Garden diploid level (except n = 10 in the three species mentioned above) and abundant intraspecific poly- ploidy. We believe that extensive field observation of variation patterns and additional SEM studies of seeds and trichomes remain the best source of evidence for such taxonomic decisions. Overem- phasis of minor morphological variation, especially in complex polyploid species, may lead to unwieldy and impractical classification in Castilleja. Тће annotated list of chromosome numbers in- cluding locality and voucher citation is shown in 'able 1. When pertinent, we comment on the relationship of chromosome number and hybrid- ization to the evolution and systematics of the ge- nus. Castilleja affinis Hook. & Arn. Castilleja affinis is a widespread polymorphic species, ranging from coastal northern Oregon south to northern Baja California. This is a highly un- wieldy complex with six ploidy levels, п = 12 to 72 (Heckard, 1968; Chuang & Heckard, unpub- lished data). This species has been treated recently by Chuang & Heckard (1992b) as comprising three subspecies, namely, subsp. affinis, subsp. neglecta, and subsp. litoralis. Some populations appear to intergrade to such an extent with abutting C. ap- plegatei subsp. pinetorum, C. mollis, C. subin- clusa, and C. wightii that it is difficult to separate them taxonomically. All three counts from Baja California are tetraploid (n — 24); this saame number is also fairly widely distributed northward to coastal and Sierra Nevada foothill areas of central Califor- nia (Heckard, 1968; Chuang & Heckard, unpub- lished data). Undoubtedly, hybridization and poly- ploidy have played important roles in creating taxonomic perplexity in this species complex. We suspect that the varying combinations of ploidy evels have evolved to ensure that sympatric pop- ulations will have different chromosome numbers and thus be able to coexist without freely hybrid- izing. Castilleja arvensis Schltdl. & Cham. This annual Castilleja is the most widespread species in the genus, extending from central and northeastern Mexico, south to Central America, and nearly through South America. Our count of n = 12 agrees with previous reports (Heckard, 1968; Spellenberg, 1986) Castilleja aspera Eastwood This species of Chihuahua and Durango is vari- able with regard to herbage and galea indumentum and leaf shape, but is generally recognizable by the shallowly cleft lateral calyx lobes with acute seg- 12 and 24 have been reported from Durango (Heckard, 1968). We report two additional counts of n = 1 from Chihuahua and Durango. Nesom (1992b) de- scribed C. durangensis based on Cruden's collec- tion (Cruden 1160, holotype, TEX; isotypes, GH, UC; reported as C. aspera, n — 12 by Heckard, 1968) from about 5 km northeast of El Salto, Durango. We regard these as minor morphological ments. The chromosome numbers of п = differences of little taxonomic significance а! the specific level. The plants may be forms interme- diate between C. aspera and C. scorzonerifolia; nevertheless, this morphological variation deserves further critical study in the field. Castilleja bryantii Brandegee This species, endemic to Baja California and extending from the vicinity of San Ignacio to the Cape Region, is reported here for the first time as n — 12. Castilleja socorrensis, a species closely related to C. bryantii and endemic to Isla Socorro, was recently proposed by Moran (Levin & Moran, ) and was said to differ from the latter mainly by its smaller flowers. Whether these two should be regarded as separate species must await further un tudy Castilleja ecuadorensis N. Holmgren This species, endemic to Ecuador, is perhaps a close relative of the highly variable C. fissifolia. The count of n = 12 represents the first report for this species. Castilleja fissifolia L. f. Castilleja fissifolia, the lectotype species of the genus, appears to be an exceptionally polymorphic species with regard to habit and leaf division, an many forms have been accorded separate specific Wedde (1857) concluded that intergradations occur be- rank. Based on his own observations, tween different growth forms, as well as different developmental forms of calyx, corolla, and leaves, and treated these as varieties of a single species, integrifolia. Melchior (1931) ollowed this disposition, but raised Weddell's va- including even C. rieties of С. fissifolia to subspecific level. This species complex awaits a comprehensive investi- gation over its entire range of distribution to de- termine whether these morphological forms de- serve taxonomic recognition and at what rank. Our count of л = 12 from Ecuador (three pop- ulations) represents the first number reported for Chuang & Heckard 979 Chromosome Numbers of Neotropical Castilleja Volume 80, Number 4 6° © , ee У ө A , "ut os goes = % a $ e», co 2 | e е "s e о 9% ©, а" 4 9099 л m -- p Ф "" · а a I ee „Ж Б а Дн =, Se 25.5% 10 2656, FIGURES 1-10. — Meiotic chromosome ion of first counts of Mexican, Central American ess The stage a meiosis for each phase is indicated as: Dia , diakinesis; MI, и 4 TI, telophase I; TH ‚ telophase II.— 1. {I (Heckard 3250). < C. ecuadorensis, п = П (Holmgren 1017 0). idoli Ma ius n = 12, п = 12, I (Holmgren 10105). C. gracilis, n = 10, TI (C dab 197 3. 5. C. hirsuta, n = 12, TI (Cruden 1709). —6. C. integrifolia, n = 12, 1/2 portion of ТП (C ruden 1552).— 1. С. integrifolia, п = ак. (Сгидеп —9. С. irazuensis, n = 12, Diak. kard 6219).— 10. 72, M — -2 "us —8. C. Antegrifolia, п = 7 (Cruden 1635). — 48, 1/2 portion of TI (Breedlove 18747). ` linifolia, n = 980 Annals of th Missouri Botanical Garden this species, although our unpublished number was listed in Holmgren & Molau's (1984) treatment of Ecuadorian Scrophulariaceae. Castilleja foliolosa Hook. & Arn. This species, one of the most distinct in cis- montane California and northern Baja California, is easily separable by its herbage densely covered with white-woolly branched hairs, undivided lateral calyx lobes, and tight-fitting seed coat. This count of n — 12 from northern Baja California agrees with a previous report (Heckard, 1968). The total of seven populations sampled thus far has given only diploid counts. Castilleja fruticosa Moran Castilleja fruticosa is endemic to the southern end of Guadalupe Island, Mexico, and is a close relative of the little known C. guadalupensis Bran- degee, another endemic to Guadalupe Island, but occurring in a different habitat on the higher and much moister northern part of the island. Moran (1969) pointed ош, “It (C. guadalupensis) has not been collected since 1898 and evidently is now very rare if not extinct." This species is distin- guished from C. guadalupensis by its herbage and bracts covered with glandular hairs intermixed with pilose, longer nonglandular hairs, in contrast to the densely and closely tomentose dendritic hairs of the latter. Our count of п = 12 is the first number reported for this species. Castilleja gracilis Benth. The count of п = 10 represents the first report for this species, although Holmgren (1976) indi- cated our unpublished count on the herbarium la- bel. This species is closely related to C. tenuifolia, which has the same chromosome number of n = 10, but has a larger stature, a longer corolla, and uch more colorful calyx. Castilleja venusta Rzedowski (1975), described from Cerro Teotepec, Guerrero, Mexico, resembles C. gracilis in vege- tative morphology. However, a careful study of an isotype of C. venusta (Rzedowski 18150, MICH) reveals that it has distinct floral features in its well developed 3-lobed lower corolla lip and truncate calyx lobes. It would be interesting to see whether C. venusta also has the same chromosome number of n = 10 as found in C. gracilis and C. tenuifolia. Castilleja hirsuta Martens & Сајеош Chromosome counts of n — 12 were determined from two populations, both from Veracruz, and represent the first reports for this species. This species occurs in Morelos, Hidalgo, and Veracruz and is distinguishable by its entire calyx lobes, which are dilated distally and colored deep-red down to a near-ovary position. Bentham (1846) treated this species in sect. Epichroma Benth. on the basis of the somewhat enlarged calyx-limb. We would agree with Eastwood (1909) in including it in sect. Kuchroma (Nutt.) Benth. because of the spicate inflorescence, entire and nearly orbicular to oblong-lanceolate leaves, and a calyx subequally cleft in front and in back. It is perhaps related to C. scorzonerifolia, but this possibility needs further study. Castilleja integra A. Gray This fairly well defined species is widespread from northern Mexico (Nuevo León, Coahuila, Du- rango, Chihuahua, and Sonora) north to the south- eastern and southwestern United States. It is one of the Mexican elements of Castilleja that extends north to the adjacent regions in the United States. This species is closely related to C. purpurea and the newly described C. genevievana (Nesom, 1992а), but is distinguished by its entire leaves and mostly orange-red to scarlet inflorescence. Be- cause of its intermediate morphology and range of distribution between C. integra and C. purpurea, C. genevievana may represent a form intermediate between the two. 12 and 24) were previously reported for 15 populations from our samples (Heckard, 1968; Heckard & uang, 1977) and two collections by Ward (1983, 1984). No cytogeographic pattern is apparent in the dis- Two chromosome levels (n — tribution of the two cytotypes in the United States. The count of n — 24 here represents the first count for this species from Mexico. More sampling throughout the range of the species, particularly from Mexico, is needed to establish its cytogeo- graphic patterns. Castilleja integrifolia L. f. This species is characterized by its racemose inflorescence, entire leaves, entire floral bracts, and conspicuously falcately spreading flowers, which turn black in drying. It is one of the most widely ranging species in the genus, extending from Co- lombia north to Central America and southern Mexico. It is highly variable with respect to habit, indumentum, leaf shape and size, and color of bracts and flowers; a few morphological variants have been accorded rank as separate species. It Volume 80, Number 4 1993 Chuang & Heckard 981 Chromosome Numbers of Neotropical Castilleja was relegated to infraspecific rank under C. fissi- folia by Weddell (1857) and Melchior (1931). This is one of the few polyploid complexes here- tofore found in Mexico, as evident in the newly reported counts of n — 12, 24, and 72 for this species. Among the nine chromosome counts ob- tained thus far, the number п = 12 is found in three populations from Chiapas, Mexico; n — 72 from two populations in Guerrero, Mexico; and n — 24 from four populations in Guatemala. Many more population samplings are needed to establish its cytogeographic patterns. The presence of four levels of chromosome number (n = 12, 24, and 72, including n = 48 for C. linifolia), as well as possible hybridization with the related C. tenuiflora and others, are responsible for a large part of the complexity of this species. The diploid Breedlove collection from a marshy area at the southern end of the valley of San Cristóbal in Chiapas deserves comment. И appears to be an extreme form of C. integrifolia in having sparsely puberulent herbage, linear leaves, and nar- row flowers. It is also very similar to C. linifolia in sharing such features as a racemose inflores- cence and linear leaves. Castilleja irazuensis Oersted This species is narrowly endemic to the summits of the two high volcanoes, Volcán Irazú and Volcán Turrialba, Costa Rica. According to Holmgren (1978), C. irazuensis is closely related to C. tal- amancensis, which is endemic to the Cordillera de Talamanca in southern Costa Rica. This report of = 12 represents the first count for this species. Castilleja jiquilpana Nesom This newly described species (Nesom, 1992b) is endemic to the vicinity of Jiquilpan, and is un- doubtedly a close relative of the highly polymorphic C. scorzonerifolia. lt may be distinguished with some difficulty by its deeply divided leaves with linear lobes and smaller (1 mm long), ovoid-deltoid seeds with a fairly deep-netted seed coat. This count of п = 12 agrees with a previous count from a collection made by Cruden (Cruden 1149, ca. 22 km W of Jiquilpan, UC) and pub- lished as C. glandulosa (Heckard, 1968). Castilleja lanata A. Gray The count of n — 12 from three collections agrees with the one previously reported for this d, 1983). This spe- cies is widespread from northern Mexico (including species from New Mexico (War northern Baja California) and western Texas to southern Arizona, and the northern Channel Islands of California. This is one of the Mexican elements of Castilleja, extending north to the southern United States. Castilleja lanata, C. hololeuca (relegated to subspecific status, Chuang & Heckard, 1992b), together with C. foliosa and C. grisea, have been included in sect. Lanatae (Pennell, 1951) because of their grayish or white-tomentose herbage, branched hairs, and calyx lobes that are rounded or wholly united laterally. This species differs from the closely related C. foliolosa and C. grisea by its white-woolly felt of long interwoven, slightly branched hairs and entire leaves. Castilleja linifolia N. Holmgren The count of n = 48 (octoploid) from three collections represents the first chromosome number reported for this species, although Holmgren (1976) included this number (in Cruden 2000 and Breed- love 18747) in his original description from our unpublished counts on the herbarium labels. This species is known only from southern Durango and adjacent Sinaloa, and appears to be related to a highly variable polyploid species, C. integrifolia, except that the former has glabrous to sparsely pilose herbage, and entire, linear-filiform leaves. It is probable that this octoploid (n — 48) is a part of the C. integrifolia complex, of which three chromosome levels (n — 12, 24, and 72) have thus far been revealed. Castilleja lithospermoides Kunth The count of л = 24 from three populations reported here agrees with a previous report (Heck- ard, 1968). The identification of these collections appears questionable. Nevertheless, all of them match well a specimen cited by Eastwood (1909) as C. lithospermoides (Pringle 2565, near Gua- dalajara, Jalisco, Mexico, UC), in having herbage covered with hispid-pilose nonglandular hairs in- termixed with short glandular hairs, and seeds 1.8— 2 mm long, linear-oblong, with a shallowly retic- ulate seed-coat with the membranaceous inner tan- gential walls of seed-coat cells retained at maturity. Nesom (1992b) reduced this species to synon- ymy under C. scorzonerifolia, based upon study of microfiche of the types (P). We agree with Nesom that the original illustrations (Kunth, 1817) of C. lithospermoides (pl. CLXIV) and C. scor- zonerifolia (pl. CLXV) are almost identical. Based on specimens at hand, we are able to distinguish four different seed forms among the plants, which we tentatively identify as C. lithospermoides and Annals of the Missouri Botanical Garden its close relatives C. scorzonerifolia, C. glandu- losa, and C. jiquilpana. We herein designate these four species as the C. scorzonerifolia complex, which merits further investigation, especially a careful study of the type specimens as well as field investigation of natural populations over their range of distribution. Castilleja mexicana (Hemsl.) A. Gray This report of n — 12 from three collections agrees with the one previously reported count (Ward, 1984). This species was originally de- scribed as Orthocarpus mexicanus (Hemsley, 1882: 463) and was meticulously and beautifully illustrated (pl. LXII. A. figs. 1-6). In his original description, Hemsley commented that, “Although this is very closely allied to Castilleja sessiliflora, we have preferred placing it in Orthocarpus, be- cause it certainly ought to go into the same genus as Orthocarpus australis [= Castilleja laciniata Hook. & Arn. from Peru; see Chuang & Heckard, 1991: 657], and because it differs from nearly all the species referred to Castilleja in the bifid lobes of the calyx, and in the long exserted corolla with nearly equal lips." This observation substantiates our recent realignment (Chuang & Heckard, 1991) to include Keck’s (1927) Orthocarpus sects. Cas- tillejoides and Cordylanthoides in genus Castil- : 583) observed that, “с astil- leja sessiliflora Pon. is excluded as all specimens seen appear to be C. mexicana. The two are very closely related." Nesom (1992a) reduced C. tor- tifolia Pennell to synonymy under C. mexicana, but maintained C. mexicana and C. sessiliflora as distinct species, based on such minor differences as color of bracts and indumentum of stems. Castilleja mexicana and C. sessiliflora can eas- ily be distinguished from other Mexican species by their dwarf stature, falcate yellow or purplish co- rolla long-exserted from the calyx, and well de- veloped lower corolla lip with three thin petaloid lobes nearly equaling the upper lip. These two were included in section. Callichroma [type: C. purea (Nutt.) Don] by Bentham (1846); later, Pen- nell (1935) proposed section Anthochroma to in- clude C. sessiliflora and its relatives. Castilleja nubigena Kunth The count of n = 12 from two populations is reported here for the first time for this endemic of the high mountains of Ecuador. This species is closely related to C. pumila (Benth.) Weddell ex Herrera, and both are perhaps part of the C. fis- sifolia complex. As indicated earlier, Melchior (1931) followed Weddell (1857) in treating most South American species of Castilleja as infraspe- cific taxa under C. fissifolia, including this one as C. fissifolia subsp. nubigena (Kunth) Melchior. Castilleja ornata Eastwood This count of n — 24 is the first report for this species. According to Eastwood (1909), this species is related to C. glandulosa (— C. scorzonerifolia) in the trisaccate lower lip with the divisions sepa- rated by the folds forming the three sacs below, and has different pubescence and generally obtuse leaves. Castilleja ortegae Standley The count of n = 12 from four collections rep- resents the first reported chromosome number for this species. It is distinguished by its narrow, entire, often reflexed leaves, secund racemose inflores- cence, distally dilated calyx lobes, and shorter up- per corolla lip (beak). Holmgren (1976) created the Ortegae group to include this species and C. pterocaulon N. Holmgren. Undoubtedly they are closely related to C. gracilis, C. roei, and C. ten- uifolia in sect. Epichroma Benth., except that the latter three species all have filiform, pinnately dis- sected leaves. Further study may suggest that C. ortegae and C. pterocaulon should also be included in this section. Castilleja ortegae is found from northern Sonora and southwestern Chihuahua to Sinaloa and adjacent Durango. Castilleja patriotica Fernald This report of n — 12 from five collections in Chihuahua, Mexico, is consistent with previously reported counts from Arizona (Heckard & Chuang, 1977; Ward & Spellenberg, 1986). Castilleja pa- triotica is characterized by its pinnately 3—5-lobed leaves, racemose not secund inflorescence, longer upper corolla lip, and calyx cleft to almost 25 in front. It is a member of sect. Castilleja (= sect. Hemichroma Benth.), which contains C. fissifolia (the type of the genus) and C. integrifolia of South America, and C. linariifolia and C. subinclusa (including C. franciscana) of the western United States. Holmgren (1976) proposed the designation Tenuiflorae group to include C. patriotica, C. ten- uiflora, C. linifolia, and C. mcvaughii N. Holm gren, based on such features as the narrow, entire, often reflexed leaves, and a long galea equaling or exceeding the tube in length. Volume 80, Number 4 Chuang & Heckard 1993 Chromosome Numbers of Neotropical Castilleja 983 E Gg 9 е eee 00 се • 11 %492• mE 5 e dg 13 2 ө ©, ^ P d % гё, : & ne ~ eb * p ie =. b E 17 18 ве ge aw 33 ө о a ets e o 6% # e E e? ө • > 20 6 9e ° e о 99, 19 e 219669 * 20 um FIGURES 11-21. Meiotic chromosome figures of first counts (except C. American, and South American Castilleja. The stage of meiosis for each phase is indicated as: Diak., diakinesis; MI, Mis de TI, telophase I; TII, telophase П. — 11. C. nubigena, n = 12, TI (Holmgren 10101).—12. C. ornata, n = 24, TI (Holmgren 8044).—13. C. ortegae, n = 12, 1/2 portion of TII (Breedlove 19192). — 14. C. roei, n = 12, Disk. (Cruden 1312).—15. C. scorzonerifolia, n = 12, TI (Holmgren 8059). — 16. C. stenophylla, n — 12, Yos (Holmgren 8017). — 17. C. talamancensis, n = 12, TI (Heckard 62194). 12, 1/2 portion of TII (Cruden 1584). —19. C. ge n = 12, TI (Breedlove 18723). — 20. C. tenuifolia, п LT = 10, TI (Cruden 1977).— 21. C. virgata, n = 12, D 10142). tapeinoclada) of Mexican, Central „= 3 со EIS ~ 8 = о ~. 3 © S ~. 5 CA "€ 2 || Annals of the Missouri Botanical Garden Castilleja purpusii Brandegee This report of n — 48- is one of the few high polyploid species counted for Mexican Castilleja and represents the first estimated number for this species. Although no meiotic irregularity was found, it was impossible to make an accurate count be- cause of the large number of small, clumped chro- mosomes during metaphase I and II. Although it was estimated to have over 48-bivalent chromo- somes, it may very well be п = 60; additional material is needed. This species is closely related to C. pectinata Matens & Galeotti, differing only in having leaves that are entire to 3-lobed rather than usually pinnately 5-7-lobed Castilleja roei Crosswhite The count of n = 12 from four collections rep- resents the first report for this species, although Holmgren (1976) indicated our unpublished count from the herbarium labels. Castilleja roei is un- doubtedly related to C. chroma Benth. This is a beautiful species, easily recognizable by its habitat of dripping wet cliffs, filiform leaf segments, and es- tenuifolia in section Epi- perennial habit, pecially its calyx coloration. Holmgren (1976) re- marked that this is the only species of Castilleja in which the calyx is yellow distally and red prox- imally. Castilleja saltensis Eastwood The count of n = 12 from two populations collected in Durango agrees with one previously reported, also from Durango (Heckard, 1968). This species is perhaps related to C. shaffneri in such features as a dwarf, caespitose habit and a dense, compact floral spike, but is easily identified by its distally dilated, obtuse or rounded lobes of the floral bracts. Castilleja schaffneri Hemsley Chromosome counts of n = 12 were determined from three populations and are in agreement with 1962). This species is easily distinguished by its dwarf, one previously reported (Beaman et al., caespitose habit with many slender decumbent pi- lose stems from the base and dense spikes with large (mostly 3-4 cm long) red flowers and densely pilose calyx with shallowly cleft lateral lobes. It is distributed in the states of Hidalgo, México, Mo- relos, Puebla, and Veracruz, at elevations from 2,400 to about 4,000 m. It was meticulously il- lustrated by Hemsley (1882: 662, pl. LXII. B. figs. 7-13). Nesom (1992b) placed C. schaffneri and C. pringlei Fernald in synonymy with C. mor- anensis Kunth. However, this suggestion needs е ID cd especially careful study of pe specimen of C. o med. rod three species in sect. Eu- chroma (Nutt.) moranensis. Eastwood enth. Castilleja scorzonerifolia Kunth The report of n — 12 from seven populations represents the first count for this species. Castilleja scorzonerifolia is a widely distributed species com- plex, ranging from northeastern to south-central Mexico. Nesom (1992b) reduced C. lithosper- moides and C. glandulosa to synonymy under scorzonerifolia and pointed out, “It is somewhat variable in features of vestiture, calyx morphology, and degree of corolla exsertion, but for the most part, identifications of this species are relatively constant." The entire complex needs further study. As indicated earlier, the discovery of at least four seed forms in this complex (including C. lith- ospermoides and C. jiquilpana) by our cursory investigation of specimens at hand suggests that detailed SEM study of seeds may help to unravel some of the taxonomic problems in this species group. Castilleja stenophylla M. E. Jones The chromosome count of n — 12 from six populations, all from Chihuahua, represents the first report for this species. This species is perhaps related to C. integra, but is distinguished by its entire linear leaves and smaller flowers. Eastwood (1909) placed it in sect. Hemichroma Benth., but we prefer to include it in sect. Euchroma (Nutt.) Benth. on account of its subequally cleft calyx, floral bracts that are broader than cauline leaves, and densely spicate inflorescence. Castilleja subinclusa Greene subsp. subinclusa We have recently treated C. subinclusa as a highly variable species consisting of two subspecies, subsp. subinclusa (including C. jepsonii) and subsp. franciscana, distributed in cismontane California and northern Baja California (Chuang & Heckard, 1992b). This polyploid complex, one of many in the western United States, has four ploidy levels, = 12 to 48, based on an exhaustive chromosome survey (Heckard, 1968; Chuang & Heckard, un- published data). It is closely related to the wide- Volume 80, Number 4 Chuang & Heckard 985 Chromosome Numbers of Neotropical Castilleja ranging C. linariifolia Benth.; both species were included in sect. Hemichroma (= sect. Castilleja) by Bentham (1846). Pennell (1951) proposed sec- tion Linariaefolia to accommodate these two spe- cies because they share such features as an un- evenly cleft calyx with a much deeper cut in front, and a corolla generally curved forward through a calyx sinus. All three counts from northern Baja California are tetraploid (n — 24); the tetraploid populations of this species are also found randomly distributed north to the northern coastal counties of the San Francisco Bay area. Castilleja talamancensis N. Holmgren The count of л = 12 represents the first report for this species. This species, endemic to the Cor- dillera de Talamanca, Costa Rica, is similar to C. irazuensis. Castilleja tapeinoclada Loesener This count of n = 12 agrees with a previous report (Beaman et al., 1962) for this high-mountain than 10 cm high ) and spreading stature, and inflorescence with few (1—4), large flowers. Castilleja tenuiflora Benth. This, the most widely sampled species (17 pop- ulations) in our study, reveals two ploidy levels (n — |2, 24). This extensive survey does not show any specific cytogeographic distinctions between the diploid and tetraploid populations. The counts of n — 12 represent the first reported diploid, whereas the counts of n — 24 confirm a previous report (Heckard, 1968) for this most common, most widespread (extending from Guererro and Oaxaca, northward through most of Mexico to southeastern Arizona), and one of the most complex Mexican species. It is likely that some of the geo- graphic variants such as a linear-leaved form, some color forms, and a form with a secund inflorescence and longer pedicels, represent populations inter- mediate between this a integrifolia, Holmgren (1976) treated the following as conspe- the closely related C. and perhaps even C. tenuifolia. cific with C. tenuiflora: C. canescens Benth., C. laxa A. Gray, Eastwood, C. scabridula Eastwood, and C. Pennell. C. retrorsa Standley, C. xylorrhiza setosa Castilleja tenuifolia Marens & Сајеош The count of n = 10 from five collections 15 actually the first report for this species, althoug Heckard (1968) pointed out an uncertain count of n — 10; that count was not reported in the list and no voucher specimen was cited. This species was included in the small sect. Epichroma by Bentham (1846). Eastwood (1909) stated, “Epichroma is probably the most individual subdivision and has, perhaps, the best claim to generic rank . . . .” Th same number of n — 10 is found also in ibe dody related C. gracilis and distantly related C. linear- iloba (subg. Colacus sect. Onchorhynchus). All the other Castilleja species counted have a chro- mosome number of n = 12 or multiples of the basic number; the number n = 10 is postulated to be an aneuploid reduction from an n = 12 pro- genitor. The species is easily distinguishable by its annual habit, filiform leaf divisions, secund-racemose in- florescence, and funnelform and colorful calyces. It is distributed in Jalisco, Michoacán, México, Guererro, and Oaxaca. Castilleja virgata (Wedd.) Edwin This count of n = 12 represents the first report for this species. It is very similar to the highly variable C. fissifolia. The entire C. fissifolia com- plex is in urgent need of further critical study. LITERATURE CITED шит J. H., >. D. DE св à W. P. STOUTAMIRE. b- iade studies in the alpine and su lios flora of Mexico and aaa. Amer. J. Bot. : 4l- 5 BENTHAM, E Scrophulariaceae. Pp. 186-598 in gud (editor), Prodromus rigide Natl Vegetabilis, vol. 10. Treuttel & Міг мна T. S. . Univ. 914. Plantae kea щш Calif. Publ. Bot 51- . HECKARD. оо AERA a new genus of Scrophulariaceae from Mexico. Brit- tonia 22: 20- CHUANG, T. I. € L. R. Нескакр. 1982. Chromosome numbers of inde arpus d related monotypic gen- era (Scrophulariaceae: subtribe Castillejinae). Brit- tonia 34: 89-101. & З Generic dico алын and ynopsis of subtril Саз ена (Scrophulariaceae — tribe Pediculareae). Syst. Bot. 16: 644-606. 1992a. New species of bee-pol- раак Castilleja from Peru, with a taxonomic re- d. Ат – каз: hes of subg. Colacus. . 17: 417-4 Nomenclatural changes of some A Castilleja (Scrophulariaceae). Novon 2: 185-189. 986 Annals of the Missouri Botanical Garden CROSSWHITE, F. 1970. Castilleja roei D iaceae). nis Midl. Naturalist 83: 630-6 EASTWOOD, 1909. Synopsis of the Mera and Central American species of Castilleja. Proc. Amer. Acad. Arts 44: 563-591. Epwin, G. 1970. New taxa and notes on the Scroph- ulariaceae of Peru. Phytologia 19: 361-406. Gray, А. 1862. Revision of the genus Castilleja. Amer. 5 39. 68. Chromosome numbers and poly- pany in Castilleja (Scrophulariaceae). Brittonia 20: -226. ——— 8 T. I. CHUANG. polyploidy, and hybridization in Castilleja ud ulariaceae) of the Great Basin and Rocky Mountains. Brittonia 29: 159-172. an & HEMSLEY, Botany. /n F. D. Godma O. Salvin али Biologia ire America, or Con- nowledge of the Fauna and Flora tribution to t oe and "Central Food Vol. 2. Dulau, Lon- fone N. H. 1976. Four new species of Mexi рације (subg. Cas ifi is Scrophulariaceae) i eir relatives. Brittonia 28: 19 Castilleja (Ss орнатса of Costa Rica and Panama. Britto 4. OLAU. 1984. ле Pp. 1- 189 in E. Harling € B. Sparre (editors), Flora of Ecuador, No. 21. Publishing House of the Swedish Research Cuna Stockholm Keck, D. D. 1927. A revision af the А. deis ar- us. d Сай Acad. Sci. Ser. 4, 16: 71. Кимтн, C. Castilleja. Pp. 2 Ae in A. Humboldt, A. Bonpland & C. 5. Kunth, Nova Genera et Species Plantarum. Paris. , G. L. & R. Moran. 1989. The vascular flora of Isla Socorro, Mexico. Mem. San Diego Soc. Nat. Hist. 16: 1-71. MELCHIOR, H. 31. Die Scrophulariaceen — gatt ung e gps | Castilleja in Sudamerika. Notizbl. Bot. n-Dahlem 1: 111-123. MORAN, E. "Twelve new dicots from Baja Cali- fornia. Trans. San Diego Soc. Nat. Hist. 15: 265- 9 - oe - ca en 295. Nesom, С. L. 1992a. A new species of Castilleja (Scrophulariaceae) from southcentral Texas with comments on other Texas taxa. Phytologia 72: 209- 230 1992b. New species and taxonomic evalua- ons s of Mexican Castilleja (Scrophulariaceae). Phy- ‘log 72: 231-252. PENNELL, F. W. 1935. The Scrophulariaceae of eastern ipae ч North America. Acad. Nat. Sci. Philadel- phia Monogr. 1: 1-650 . 1940. Scrophulariaceae. /n: R. E. Woodson, Jr. & R. W. Schery (editors), Contributions toward a Flora " Panama IV. Ann. Missouri Bot. Gard. 27: ке “34 Scrophulariaceae. Pp. 686-859 nuts (eon Illustrated Flora of the Pacific States, Vol ord мапі à 1 аы In: J. A . Steyermark (editor), od | Exploration in Venezuela Ш. Fieldiana, Er 28: 516- RZEDOWSKI, . Tres dicatiledonedi nuevas de interés а Bol. Soc. Bot. Mexico 35: 40- 47. SPELLENBERG, R. 1986. Jn: A. Love ае prs Chromosome Number Reports XC. Tax STANDLEY, P. C. 1936. Studies of peter pn i е Field Mus. Nat. Hist., Bot. Ser. 11: 145- : Castilleja. In: Flora of Costa Rica. Publ. Field Ми. Nat. Hist., Bot. Ser. 18: 1104- ———— a 40. Studies of oe i Publ. Field Mus. Nat. Hist., . Ser. 22 9. J. A. ко “1044. M иш Атепсап jue aee Publ. Field Mus. Nat. Hist., Bot. Ser. 23: 31- Warp, D. К. А ГРИ counts from New Mexico and southern Colorado. Phytologia 54: 302- 309 . 1984. етан counts from New Mexico and Mexico. Phytolog & R. SPELLENBERG. 1986. counts of angiosperms of western North America. _ m 61: 119-125. WEDD A. 1857. Castilleja. Pp. 118-119 in C Yide Andina, Vol. 2 (12). Chez P. Bertrand, Paris. Chromosome W. Dennis Clark? Brandon S. Gaut,’ Melvin R. Duvall, and Michael T. Clegg’ PHYLOGENETIC RELATIONSHIPS OF THE BROMELIIFLORAE- COMMELINIFLORAE- ZINGIBERIFLORAE COMPLEX OF MONOCOTS BASED ON rbcL SEQUENCE COMPARISONS! ABSTRACT Four cladistic hypotheses were constructed for Dahlgren’s Bromeliiflorae- саа Zingiberiflorae complex, yP 8 8! P primarily from suggestions of affinities according to morphology-based treatments: (i) D en's treatment of the families is correct; (ii) the Bromeliaceae and Zingiberales are sister groups; (ш) the пет (и is monophy letic, with the Arecaceae rbcL sequences and a smaller data set that included only families of the complex plus the Arecaceae (52 taxa). The larger data analysis supported hypothesis (iii), minus the Velloziaceae. However, hypotheses (i) and (ii) were not supported, and support for hypothesis (iv) was equivocal. Three decisions on classification can be made based on the results of this study: (i) submerge the Bromeliiflorae into the Commeliniflorae; (ii) remove the Velloziaceae from the complex; and (iii) remove the Cyclanthaceae and Pandanaceae from the Areciflorae. Studies on the phylogeny and classification of 1985; Takhtajan, 1980; the monocots (Liliopsida) have received consider- 1988; Dahlgren et al., Thorne, 2). able emphasis recently among plant systematists. One of the major difficulties, however, has been the relative dearth of morphological characters (compared with the dicots) that are consistently useful for phylogenetic inference. This is probably ue both to a general uniformity in floral mor- phology of monocots (the trimerous “Lilioid”” flow- er) and the occurrence of extremely reduced or specialized floral structures that obscure evolution- ary relationships of several key groups. Accord- ingly, major classifications show significant differ- ences in the treatments of monocots (Cronquist, The group of focus in this study is a three- superorder complex, as outlined by Dahlgren et al. (1985), that xe the Bromeliiflorae (six or- ders), the Commeliniflorae (four orders), and the Zingiberiflorae (one order). Within this group, the family of interest for this project is the Bromeli- aceae, a well-defined family that is widely recog- nized as monophyletic based on several synapo- morphies (Gilmartin & Brown, 1987). However, its relationships to other families and its taxonomic position within the monocots are in dispute. There- fore, we constructed four cladistic hypotheses to ork was supported by grants from the National Science Foundation (BSR-8904637 to WDC and BSR- 9002321 1 to MRD) and the National Institutes of Health (G6M45344 to MTC). We gratefully acknowledge Harold Carolina/University of Uppsala) for doing the computer analyses presented in this paper, and Gerald H. Lea (University of California, Riverside) for general advice on jg ulating DNA sequence data. We also thank Spencer Barrett (University of Toronto) for providing genomic DNA from Pontederia and the following for providing fresh plant material: Gregory K. Brown (University of Wyoming), m Morgan and Andy Sanders (Botanical Garden and Herbarium, respectively, University of California, Riverside), Harry Luther (Marie Selby Botanical Garden), Michael G. Simpson (San Diego State University), and Robert F. Thorne (Rancho Santa Ana Botanic Garden). Valuable comments were provided on earlier drafts of the manuscript by Victor A. Albert, Mark W. Chase (Royal Botanic Gardens, Kew), Leslie R. Landrum (Arizona State University), and Michael С. Simpson. 2 Department of Botany, Arizona State University, Tempe, Arizona 85287, U.S.A. * Department of Botany and Plant Sciences, Мине of California, Riverside, California 92521, U.S.A. ANN. MISSOURI Bor. Garp. 80: 987-998. 1993. 988 Annals of the Missouri Botanical Garden explain the relationships of the Bromeliaceae and related monocots, as follows: (i) Dahlgren's treat- ment (Dahlgren & Bremer, 1985) of the families, modified with more recent data (Gilmartin & Brown, 1987; Simpson, 1987, 1990), is correct (Fig. 1); (ii) the Bromeliaceae and Zingiberales are sister groups (Fig. 2); (iii) the complex is monophyletic, with the Arecaceae as the closest outgroup (Fig. 3); and (iv) a clade that contains the Bromeliaceae, Commelinaceae, and other commelinoids is mono- phyletic (Fig. 4). These hypotheses were evaluated by comparison with results from parsimony anal- yses of DNA sequences from the chloroplast gene rbcL. CLASSIFICATION Traditional classifications of plants have nec- essarily served as the starting point for defining the scope of taxa to be used in modern phylogenetic analyses. However, past and current classifications of some families, such as the Bromeliaceae, are so diverse as to render this process difficult. Major taxonomic treatments put the Bromeliaceae in a variety of higher categories. Takhtajan (1980) con- sidered the Bromeliaceae to be a member of his superorder Lilianae. Other authors have included the commelinoid monocots with the bromeliads (Hutchinson, 1973; Thorne, 1992). However, Cronquist (1978) found the Bromeliaceae to be highly discordant in the subclass Commelinidae; he instead placed them in his Zingiberidae, along with the Zingiberales (formerly in the Liliidae). This change was prompted, at least in part, by a desire to sharpen the definition of the Commelinidae and the Liliidae. Finally, the most divergent view of bromeliad classification among major treatments may be that of Dahlgren et al. (1985), who placed the Bromeliaceae in the superorder Bromeliiflorae, along with four other single-family orders and the two-family order Typhales. PHYLOGENETIC RELATIONSHIPS Identification of shared derived character states using phylogenetic systematics provides a rigorous methodology for inferring evolutionary relation- ships and for evaluating previous notions of clas- sification. However, many character states that have been used to define taxa are unknown with respect to polarity and may represent primitive retentions. For example, Cronquist (1978) sug- gested a relationship between Bromeliales and Zin- giberales based on their large, showy bracts. Wil- liams & Harborne (1977) made the same based on the myricetin and quercetin odes 1Yyoactian that occur in both groups. Takhtajan (1980) pro- posed a relationship between bromeliads and lili- alean families, especially Agavaceae, based on nu- merous similarities: raphides, steroidal sapogenins, flavonoids, scalariform perforations in root vessels, septal nectaries, helobial endosperm, and *'. . . def- inite similarities between pollen grains" (p. 314). The literature is replete with such suggestions. Nonetheless, these suggestions do not tell us which characters are indicators of common ancestry (syn- apomorphic). The most thorough and perhaps strongest suggestions for the phylogenetic relation- ships of the bromeliads come from the work of Dahlgren and his associates (Dahlgren & Rasmus- sen, 1983; Dahlgren & Bremer, 1985; Dahlgren 1985). Although their character polarities were offered as provisional, the monophyly of the et al., Bromeliaceae is apparently supported by three con- sistent synapomorphies. These are: (i) stellate and scalelike multicellular hairs; (ii) large, solitary silica bodies; and (iii) a conspicuously sepaloid and pet- aloid perianth. However, only the stellate or scale- like multicellular hairs are unique to the Brome- liaceae; other traits are homoplastic. Additional synapomorphies for the Bromeliaceae might in- clude conduplicate-spiral stigmas (Brown & Gil- martin, 1989a) and a base chromosome number of x = 25 (Brown & Gilmartin, 1989b). But the consistency of these characters across the family awaits a more complete evaluation. The greatest number of ideas about the phylo- genetic relationships of the Bromeliaceae involve interfamilial and interordinal levels because these present the greatest difficulty. Cronquist (1988) suggested that the Bromeliales shared a common ancestor with the Zingiberales that **. . . might have been much like some of the more archaic, terres- trial Bromeliaceae, but less xerophytic" (p. 487). The monocot “phylogeny” of Walker (1986) shows the Bromeliidae and the Commelinidae arising to- gether from the Liliidae. Dahlgren & Rasmussen (1983) included the Bromeliaceae with the Vello- ziaceae, Philydraceae, Haemodoraceae, Pontederi- aceae, Typhaceae, and Sparganiaceae in the su- perorder Bromeliiflorae. However, the sister status of the Bromeliaceae in Dahlgren’s system is shown as a polytomy that consists of the bromeliads, the Velloziales, and collectively all of the remaining orders of the Bromeliiflorae. Sister status of Bromeliaceae and Velloziaceae was more firmly suggested in a cladistic study by Gilmartin & Brown (1987). This suggestion was derived from sequential analyses of likely sister groups, a methodology that was criticized by Simp- son (1988) but defended by Gilmartin & Brown Volume 80, Number 4 1993 Clark et al. Phylogeny of Bromeliiflorae— Commeliniflorae—Zingiberiflorae 989 Zingiberales ZING Velloziales Bromeliales Philydrales Typhales BROM Haemodorales Pontederiales Commelinales | m Hydatellales COMM Hr Poales Cyperales J СОВЕ l. Hypothesis 1: Relationships within the three- superorder complex. ZING = Zingiberiflorae, BROM = Bromeliiflorae, COMM = Commeliniflorae (1988). Regardless, no synapomorphies were iden- tified that distinguished a Bromeliaceae- Vellozia- ceae clade from the other families examined. At a more inclusive taxonomic level, monophyly of the Bromeliiflorae is supported in the Dahlgren scheme by an apomorphic subtype of helobial en- dosperm. Furthermore, this scheme shows that the sister group to the Bromeliiflorae is the Zingiber- The Commeliniflorae are sister to the Bromeliiflorae— iflorae (demonstrated by showy perianths). Zingiberiflorae clade. The monophyly of this three- superorder complex is supported by three syna- pomorphies: sculptured and oriented epicuticular x ("Strelitzia-type"), endosperm with copious starch, and cell walls with UV-fluorescent com- pounds. OBJECTIVES AND HYPOTHESES Given the variety of classifications and suggested phylogenetic relationships for the Bromeliiflorae— Commeliniflorae- Zingiberiflorae, our main objec- tive was to obtain new information regarding the relationships among the constituent families in these groups. This new information consists of a cladistic analysis of nucleotide sequences of the plastid gene rbcL, which codes for the large subunit of ribulose- 1,5-bisphosphate carboxylase/oxygenase. e have constructed four hypotheses for eval- uation by DNA sequence comparisons. We chose Dahlgren's system as a framework for our initial hypothesis. Although there have been many sug- gestions about the possible relationships of the com- plex, those of Dahlgren and his associates are the most comprehensive and the closest to being ex- Zingiberales Bromeliaceae OTHER MONOCOTS Hypothesis 2: Accent и status of the MS to the Zingiberales plicit cladistic hypotheses (Dahlgren & Bremer, 1985). (Velloziales, Bromeliales, all other Bromeliiflorae) However, we have modified the polytomy in Dahlgren’s Bromeliiflorae to reflect the study by Gilmartin & Brown (1987). This change shows the romeliaceae and Velloziaceae to be sister taxa. We have also modified the polytomy (Haemodora- les, Pontederiales, Typhales) to reflect relationships suggested by Simpson (1987, 1990). This change shows the Haemodoraceae/Haemodorales Pontederiaceae / Pontederiales to be sister taxa. Ac- cordingly, our initial hypothesis for the phyloge- and netic relationships of the three superorder complex is summarized in Figure 1. Our second hypothesis is that the Bromeliaceae and Zingiberales are sister taxa (Fig. 2). This ex- planation is based in part on suggestions of Cron- quist (1988) regarding the putative common an- cestor to the two groups. It also incorporates the observation of Kress (1990) that, at present, the Bromeliaceae are the best outgroup to the Zingi- berales. Our third hypothesis is that the Bromeliiflorae— Commeliniflorae-Zingiberiflorae complex is a monophyletic group, as suggested by Dahlgren & Rasmussen (1983). This group is held together by the three synapomorphies mentioned previously. However, at the moment, there is no clear guidance for suggesting a putative sister group to this com- plex. Thus, this hypothesis is the most incomplete of the four examined in this study. Nevertheless, a tentative proposal might include the Arecales, whose members also possess UV-fluorescent cell walls (Fig. 3 Our fourth hypothesis is that the Bromeliaceae and Commelinaceae are together іп a ““commeli- noid-bromelioid" clade (Fig. 4), which precludes the Bromeliaceae- Zingiberales proposed in the sec- ond hypothesis (Fig. 2). The fourth hypothesis de- rives from the inclusion of bromeliads within the Commelinales by Thorne (1992), from the putative derivation of the Bromeliales from the Commelina- les proposed by Hutchinson (1973), and from the common origin of the Bromeliidae and Commelini- dae suggested by Walker (1986). However, this alternative hypothesis cannot be developed enough at the moment to predict with certainty which 990 Annals of the Missouri Botanical Garden Arecaceae Bromeliaceae ZINGIBERIFLORAE Commelinaceae 123 BROMELIIFLORAE Other commelinoids COMMELINIFLORAE OTHER MONOCOTS FIGURE 3. Hypothesis 3: Monophyly of the three- FicURE 4. Hypothesis 4: Monophyly of a group con- JV-fluorescent cell walls; 2, copious endosperm; 3, Strelitzia-type epicuticular wax monocots, outside of the Bromeliaceae and Com- melinaceae, should be considered as commelinoid. There are undoubtedly other nontrivial hypoth- eses that we could have examined for this paper. However, focus on this particular set of hypotheses was guided by numerous preliminary analyses as the project grew. For this reason, our method is not only abductive and predictive but also in part retrodictive. The method is retrodictive because we chose our hypotheses after we examined prelimi- nary results. Nevertheless, we believe that this mixed process has enabled us to make significant progress in monocot systematics. Our hypotheses are not completely mutually ex- clusive, but neither can all be fully supported. This means that our results should be useful for eval- uating morphological and chemical character evo- lution on rbcL-based trees. Accordingly, we can make preliminary character analyses of those traits that have traditionally been used for suggesting phylogenetic relationships. MATERIALS AND METHODS Sequences were obtained from seven genera of the Bromeliaceae and one genus each from the Strelitziaceae, Pontederiaceae, Velloziaceae, and Rapateaceae. Other sequences used for comparison here either have been published or shared by other investigators. Sources for new sequences used in this report are listed in the Appendix. In most instances total DNA was extracted from fresh leaf tissue using a Tris-EDTA buffer (Della- porta et al., 1983). In contrast, samples of Ste- golepis, Vellozia, and Glomeropitcairnia were ex- tracted using the CTAB microextraction method of Learn (1987). Each preparation was used to provide a template for amplifying rbcL by the polymerase chain reaction (PCR). The PCR pro- tocol was suggested by the supplier of Тад DNA polymerase (Promega Corp.). Primers for ampli- fying double-stranded DNA were designed from conserved regions in atpB from Nicotiana taba- taining the Bromeliaceae and Commelinaceae. cum, in rbcL from Zea mays, and in ORF106 from Oryza sativa (Golenberg et al., 1990). Single- stranded DNA was made in a second round of PCR using the double-stranded product as a template. Single-stranded products were cleaned by precip- itation in NaOAc. Both the coding strand and the complementary strand were amplified separately and subsequently sequenced. Sequencing followed Sanger et al. (1977) according to the Sequenase protocol (US Biochemical). Internal sequencing primers were provided by G. Zurawski (DNAX Corp., Palo Alto, California) from conserved regions in the rbcL sequence of Zea mays. Sequence data were read directly from auto- radiograms into files created by a sequence editor (SE; Matt Clegg, Riverside, California). Sequences were aligned by hand. Ultimately, all sequences were entered into an edit file in the NEXUS format for parsimony analysis with PAUP 3.0 (Swofford, 991) — For this paper, a subset of genera was reanalyzed separately from the 499 taxa of seed plants used by Chase et al. (1993). This subset included 52 species of monocots from amilies in the Bromeliiflorae—Commeliniflorae— Zingiberiflorae complex, plus the Arecaceae (Table 1). We deemed the analysis of this subset useful because of its potential for evaluating in more detail a smaller set of ingroup relationships. Fitch parsimony analyses (equal weighting) were performed on 10 random- addition data sets of the 52 taxa. Character-state changes were plotted using the ACCTRAN option. The most parsimonious trees from these analyses were also used as the starting point for a character- state weighted analysis of the 52 taxa (Albert et 93) al., RESULTS Figure 5 shows the relationships between the ingroup (i.e., the Bromeliiflorae-Commeliniflorae- Zingiberiflorae complex) for this study and other families of monocots. This cladogram was made by modifying Figures 5A and 6A from the 499-taxon analysis in Chase et al. (1993) to include only Volume 80, Number 4 Clark et al. 991 Phylogeny of Вготе огае– Commeliniflorae—Zingiberiflorae family names and exclude names of genera. This simplification is justified because all of the families for which multiple genera were sampled are mono- phyletic in this cladogram, with the exception of the Melianthaceae. Figure 5 shows not only the position of the ingroup relative to other monocots, but also the exclusion of the Velloziaceae from the ingroup. sister-taxon relationship between the Velloziaceae and Bromeliaceae is clearly not supported by rbcL sequence data. The Velloziaceae are instead far removed from the three-superorder complex of in- terest here, to sister status with a Pandanaceae- Cyclanthaceae clade. This result provides support for Cronquist’s inclusion of the Velloziaceae in the Liliales, but an explicit hypothesis for this rela- tionship is beyond the scope of this paper. Nev- ertheless, this finding provides direction for future studies on the relationships of the Velloziaceae. Figure 6 is modified from Chase et al.’s (1993) 499-taxon analysis to show relationships among the ingroup families for this study. Most multiple- sample families are again monophyletic, the ex- ceptions being the Juncaceae and Cyperaceae. Figure 6 shows Dahlgren’s Bromeliiflorae and Com- meliniflorae to be paraphyletic. The only family from Dahlgren’s Bromeliiflorae that diverges with the bromeliads is the Typhaceae. All of the other families in this clade are members of Dahlgren’s Commeliniflorae. In addition, the bromeliad-con- taining clade shows the Bromeliaceae to be basal within it. This implies that there is no single sister family to the Bromeliaceae. The remaining bromeliiflorean families in Figure 6, plus the Commelinaceae, diverge with the Zin- giberales. Regardless of the lack of support for Dahlgren’s concept of the Bromeliiflorae and Com- meliniflorae, these results do support the mono- phyly of the Bromeliiflorae—Commeliniflorae—Zin- giberiflorae complex (minus the Velloziaceae). Furthermore, the Arecaceae appear as the out- group to this three-superorder complex, as also suggested in Figure 5. Reanalysis (unrooted) to include just the Are- caceae and the ingroup of interest yielded 198 most-parsimonious trees of length 1665 by Fitch parsimony. А strict consensus showed that differ- ences among trees were due to within-family re- arrangements in the Bromeliaceae and Poaceae (data not shown), so all family-level nodes were fully resolved (Fig. 7A). Character-state weighting yielded 30 most parsimonious trees, each of which converted to 1667 even-weighted steps. А strict consensus showed that differences among trees were again due to within-family rearrangements, so all TABLE 1. Families analyzed in this study. ne ae number in parentheses is the number of sp the second number is the approximate minimum ard of ee in the family according to Thorne, 1992.) Arecaceae (7/2780) Bromeliaceae (7/1520) Commelinaceae (3/605) Costaceae (2/150) Cyperaceae (2/5315) Flagellariaceae (1/4) Marantaceae (2/500) Musaceae (1/42) Philydraceae (1/6) Poaceae (9/10,000) Pontederiaceae (1/30) Juncaceae (3/300) Lowiaceae (1/6) Typhaceae (1/30 Zingiberaceae (4/1000) family-level nodes were still fully resolved. How- ever, there was one family-level difference between the Fitch analysis and the weighted analysis: the Commelinaceae moved from a position basal to the Pontederiaceae-Philydraceae- Haemodoraceae clade (Fig. 7A) to a position that is sister to the Pontederiaceae (Fig. 7B). Several changes occurred among family rela- tionships by analyzing the smaller data set. Perhaps the most notable of these is that the four-family clade containing the Commelinaceae shifted from the Zingiberales-containing clade to а more con- ventional position near the Cyperales and other commelinoid monocots (Fig. 7A). In so doing, the Commelinaceae-containing clade replaced the Fla- gellariaceae as a sister group to the Cyperales. The Flagellariaceae, in turn, moved to a position basal to the Restionaceae—Poaceae family pair. An in- ternal rearrangement in the Commelinaceae-con- taining clade also put the Haemodoraceae and Phil- ydraceae together as sister taxa, which roughly conforms to the recent cladistic analysis by Simp- son (1990). The difference is that our results show the Philydraceae and not the Pontederiaceae as the sister group to the Haemodoraceae. The Pon- tederiaceae are instead basal to the Philydraceae- Haemodoraceae clade Finally, results from this analysis also show a sister family relationship between the Bromeliaceae and the Rapateaceae. The Bromeliaceae- Rapatea- ceae clade is sister to a clade containing all the remaining families of the Bromeliiflorae and Com- meliniflorae in the ingroup. DISCUSSION The results reported here provide little support for Dahlgren’s view of the Bromeliiflorae. More- over, there is no evidence from rbcL sequence comparisons that the Velloziaceae are the sister 992 Annals of the Missouri Botanical Garden [^ INGROUP - Arecaceae Xanthorrhoeaceae Asphodelaceae Dracaenaceae Е Ruscaceae PS Nolinacaceae Г Hyacinthaceae Anthericaceae Amaryllidaceae m Iridaceae L- Tecophilaeaceae Hypoxidaceae Orchidaceae Smilacaceae Liliaceae Alstroemeriaceae Colchicaceae Melanthiaceae VELLOZIACEAE Pandanaceae Cyclanthaceae [^ Dioscoreaceae las Taccaceae Melanthiaceae | Burmanniaceae Alismataceae Potamogetonaceae = Melianthaceae L- Araceae Relationships of the INGROUP/Arecaceae clade to other monocot groups according to unrooted parsimony analysis of a 499-taxon data set of seed plants. ING ROUP = Bromeliiflorae- Commeliniflorae- Zingiberiflorae complex, minus the Velloziaceae. Modified from Chase et al. (1993). Volume 80, Number 4 Clark et al. 993 1993 ин of Bromeliiflorae— Commeliniflorae-Zingiberiflorae OTHER MONOCOTS a Arecaceae Heliconiaceae* = Lowiaceae* Musaceae* Costaceae* а Zingiberaceae* Marantaceae* Eid Strelitziaceae* 20 Steps Haemodoraceae** Philydraceae** = Pontederiaceae** Commelinaceae*** Bromeliaceae** Rapateaceae*** Typhaceae** Flagellariaceae*** E Juncaceae1*** Juncaceae2*** Cyperaceae1*** m Cyperaceae2*** Juncaceae3*** Restionaceae*** 6 Роасеае*** FIGURE 6. Family-level relationships within the Bromeliiflorae- Commeliniflorae- Zingiberiflorae complex ac corong to unrooted parsimony analysis of a 499-taxon data set of seed plants. Modified from Chase et al. (1993). Zingiberiflorae; ** = Br omeliifiorae; *** = Commeliniflorae. family to the Bromeliaceae. Therefore, we reject our first hypothesis. The rejection of Dahlgren's Bromeliiflorae comes as no surprise, since this su- perorder is held together by a single provisional synapomorphy, a variant form of helobial endo- sperm. In general, helobial endosperm is charac- terized by a division of the primary endosperm nucleus, which is followed by the formation of a transverse cell wall. This cell division makes a larger micropylar chamber and a smaller chalazal chamber (Dahlgren & Clifford, 1982). However, all but the Pontederiaceae of Dahlgren's Brome- liiflorae are known to have a much smaller, starch- free chalazal chamber. Dahlgren & Rasmussen (1983) suggested that the Pontederiaceae probably have this subtype of helobial endosperm. In addi- tion, this character is homoplastic because a very small, starch-free chalazal chamber also occurs in the helobial endosperm of Thismia javanica of the Burmanniaceae (Dahlgren & Clifford, 1982), a 994 Annals of the Missouri Botanical Garden A Arecaceae Heliconiaceae Lowiaceae Musaceae B Costaceae ве те Zingiberaceae Pontederiaceae Marantaceae Commelinaceae Strelitziaceae Haemodoraceae = Bromeliaceae Rapateaceae Typhaceae Philydraceae Pontederiaceae Haemodoraceae Philydraceae Commelinaceae Juncaceael Juncaceae2 [ Сурегасеаеї Ч Сурегасвае2 Juncaceae3 Flagellariaceae o 20 Steps Restionaceae Poaceae haa E 7. Family- level npud within the Br oo berifl omplex according to unrooted parsimony analyses of a 52-taxon data s. (A) Strict consensus of 198 ti trees generated by tch analysis cue weighting); length = 1665. (B) Family. level Dt generated by character-state weighted didus analysis, strict consensus of 30 trees; length — 1667. Changes among trees in both analyses uode only within families nat shown). family that is not believed to have a close rela- tionship with the Bromeliiflorae (Dahlgren & Bre- meliaceae lends no support to a close relationship between the Bromeliaceae and the Velloziaceae. mer, 1985). The occurrence of biflavonoids in Xerophyta plicata (Velloziaceae) and their absence in the Bro- Biflavonoids are extremely rare in monocots, so their occurrence in the Velloziaceae may be aut- apomorphic. If biflavonoids are of phylogenetic sig- Volume 80, Number 4 1993 Clark et al. 995 Phylogeny of Bromeliiflorae— Commeliniflorae—Zingiberiflorae nificance among monocot families, then hypotheses regarding the relationship of the Velloziaceae, Cy- clanthaceae, and Pandanaceae should be exam- ined. The latter two families are the closest to the Velloziaceae according to analysis of rbcL sequenc- es (Chase et al., 1993; Duvall et al., 1993). e also reject our second hypothesis, that the Zingiberales and Bromeliaceae form a monophy- letic lineage. This means that the conventional characters shared by these groups (large, showy bracts and petaloid perianth parts) do not support a unique relationship; these character states are better interpreted as plesiomorphies. Other groups also have showy bracts (Orchidaceae and some Araceae, among others) or petaloid perianth parts (Commelinales and nearly all of Dahlgren's Lili- iflorae). Such a diverse distribution suggests that these characters are either nonhomologous (Kress, 1990), homoplastic, or symplesiomorphic. Regard- less, they do not appear to be synapomorphic for the Bromeliaceae and Zingiberales. Furthermore, initial evidence for a common ancestry based on flavonoid chemistry (Williams & Harborne, 1977) was quickly refuted following a more detailed ex- amination of the Bromeliaceae (Williams, 1978). Thus, our results show that the Zingiberidae of Cronquist are paraphyletic, and the traditional chemical and morphological characters that sup- port their monophyly are equivocal or unlikely to be informative. Monophyly of the Bromeliiflorae-Commelini- florae—Zingiberiflorae complex is supported by Chase et al.’s (1993) 499-taxon analysis (Fig. 6). Furthermore, the Arecaceae are the sister group to this triumvirate. This result supports our third hypothesis regarding the common ancestry of these three superorders. Therefore, we accept the hy- pothesis and agree with Dahlgren's view of this relationship, even though we reject his proposed ingroup phylogeny (first hypothesis). We also ac- cept our tentative proposal to designate the Are- caceae/ Arecales as the sister group to the three- superorder complex. Such designation clarifies the speculation by Dahlgren & Bremer (1985) that the Arecales are somehow related to the Comme- linales. Their suggestion was speculative because they showed no synapomorphies to support it. On the other hand, similar speculation that the Pan- danales and Cyclanthales might belong to Bro- meliiflorae or Commeliniflorae, respectively, is not supported by rbcL comparisons (see Chase et al., 1993, fig. 5A). By supporting our third hypothesis, we also make sense of the phylogenetic distribution of three apo- morphies. These are a sculptured and oriented epicuticular wax (“Strelitzia-type”), an endo- sperm with copious starch, and UV fluorescent cell walls (Fig. 3). The significance of these synapo- morphies comes from their occurrence in the Are- cales as well as in the ingroup. We therefore have strong, independent support for an Arecales-in- group relationship. It is notable that the three char- acters in question are informative at a high taxo- nomic level. There are few other examples of such extensive support for higher categories. However, the distribution of these apomorphies does not pre- clude a Poales—Arecales relationship, as suggested by Dahlgren & Bremer (1985). But our rbcL anal- ysis provides no support for such a relationship. he main problems with the above synapomor- phies lie in incompleteness of sampling and in in- consistency of distribution, which are perhaps ex- pected consequences of circumscribing such a large group. For example, most Zingiberales have starchy perisperm or chalazasperm instead of starchy en- dosperm. Starchy endosperm may also be a syn- apomorphy for the Arales (Dahlgren & Rasmussen, 1983), thereby making this character homoplastic within the monocots. Moreover, the significance of starchy endosperm in the unripe seeds of some Burmanniaceae is unclear. All starch in these seeds is exhausted by the time they ripen. Thus, ho- mology is uncertain for this character. Similarly, Strelitzia-type epicuticular wax is common only in the Arecaceae, Cannaceae, Cy- peraceae, Heliconiaceae, Juncaceae, Poaceae, Strelitziaceae, and Typhaceae and rare in the Bro- iaceae, as well as Velloziaceae (Behnke & Barthlott, 1983). In addition, smooth or nonorient- ed wax crystals often occur in the Poaceae, Cy- peraceae, Marantaceae, Rapateaceae, Restiona- ceae, Xyridaceae, and aemodoraceae. Wax crystals are absent from some Bromeliaceae, Hae- modoraceae, Rapateaceae, Marantaceae, and Vel- loziaceae and from all taxa that were examined in the Zingiberaceae, Pontederiaceae, and Joinville- aceae (Barthlott & Frolich, 1983). Upon first inspection, the problem of inconsis- tency also seems to plague the distribution of UV- fluorescent cell walls within the monocots. For ex- ample, certain taxa in the Haemodoraceae do not have UV-fluorescent cell walls (Harris & Hartley, 1980). However, these taxa are more appropriately included in the Tecophilaeaceae, a family that is not closely related to the Haemodoraceae (Simp- son, 1985, 1990). Fluorescent cell walls also occur in the Cyclanthaceae and the Xanthorrhoeaceae, which seems to indicate homoplasy for this char- acter. However, the cell walls of these two families are fluorescent because of p-coumaric acid. Cell Annals of the Missouri Botanical Garden wall fluorescence in other monocots is due to a combination of p-coumaric acid, ferulic acid, and diferulic acid. The latter combination is found only in the Arecales, the Bromeliiflorae, the Commelin- iflorae and the Zingiberiflorae. Therefore, UV- orescence by ferulic acid and its dimer is a unique synapomorphy. (One minor exception may be the absence of diferulic acid in the cell walls of Restio oligocephalus [Restionaceae] (Harris & Hartley, l ; we suspect that this compound will yet be found in this and other species of the family.) It is also notable that, among the taxa of interest here, only the Hydatellaceae, Costaceae and Heliconi- aceae have not been examined for cell-wall feru- lates. Aside from the monocots, cell-wall ferulate is apparently synapomorphic for the Caryophyllales (Hartley & Harris, 1981) and perhaps for the Coniferophyta (Strack et al., 1988). In these groups, кир: ас Va itself is Hin consistent character, where- as the ributi acid are more variables In addition, the phyloge- -coumaric acid and diferulic netic relevance of ferulic acid for the Coniferophyta is not known because no other nonflowering seed plants have been examined. Ferulic acid is a relatively simple and probably common constituent of plants (Molgaard & Ravn, 1988), but it dimerizes and links cell-wall com- ponents only in certain plants of close affinity within the monocots, the dicots and perhaps the conifers. Because of their unique linkage to cell walls, these compounds constitute a potentially rich source of biochemical data for phylogenetic study. Specifi- cally, a molecular systematic approach to the fer- ulate-cell wall pathway should yield strong evidence important for evaluating the potential role of this pathway in the d origin of lignification (Fry, 1983; Kubitzki, 1987). When all 499 genera of seed plants are analyzed (Chase et al., 1993), the Bromeliaceae and Com- melinaceae are not in the same clade as predicted by our fourth hypothesis. This analysis instead shows the clade (Haemodoraceae, (Philydraceae, (Com- melinaceae, Pontederiaceae))) to be the sister group of the Zingiberales; the Bromeliaceae are shown to be basal to another lineage. Support for the sister status of the Commelinaceae-containing clade and the Zingiberales comes from one potential syn- apomorphy: root hair cells that are shorter than other epidermal cells. This character state is not shared, however, with the Philydraceae and Pon- tederiaceae (Dahlgren & Bremer, 1985). On the other hand, there is no known support from mor- phology for distinguishing a Bromeliaceae-contain- ing clade from a In contrast to the broad 499-taxon analysis of Chase et al. (1993), evaluation of a smaller rbcL data set (52 taxa) that included only the ingroup plus the Arecaceae did support our fourth hypoth- esis that the Bromeliaceae and Commelinaceae are ommelinaceae-containing clade. part of the monophyletic group (Fig. 7A). However, there are no synapomorphies from morphology to corroborate this result at this time. The smaller analysis also showed a sister-taxon relationship be- tween the Bromeliaceae and Rapateaceae (Fig. 6). These families were not sister taxa in the broad analysis, but the Bromeliaceae were shown to be basal to a clade that includes the Rapateaceae as its basal member. CONCLUSIONS Analysis of rbcL sequences supports Dahlgren's concept of a clade consisting of his Bromeliiflorae, Commeliniflorae, and Zingiberiflorae. Further- more, analysis of a 499-taxon data set showed the Areciflorae (Arecales only) to be basal to this com- lex (Chase et al., 1 concept of the Bromeliiflorae is no longer tenable. 3). However, Dahlgren's Most of the families and orders in this superorder show a complex pattern of affinities with his Com- meliniflorae and Zingiberiflorae. The Velloziaceae are removed from the complex entirely. The Bro- meliaceae are not the sister to the order Zingiber- ales. Instead, the Bromeliaceae are better placed at or near the base of the clade that is next to the zingiberalean lineage. Three straightforward decisions on classification can be made based on our conclusions. One is to remove the Cyclanthaceae and Pandanaceae from the Areciflorae. The Areciflorae are thereby com- prised of one family, the Arecaceae. The second decision is to remove the Velloziaceae from the ingroup of this study (i.e., the Bromeliiflorae, Com- meliniflorae, Zingiberiflorae clade). of these decisions are in agreement with the broader mo- lecular analysis of monocots presented by Duvall et al. (1993) and are supported by the synapo- morphic distribution of UV-fluorescent cell walls in the Areciflorae and in the ingroup. The third decision is to submerge the Bromeli- iflorae into Dahlgren's Commeliniflorae. However, at this time there are no objective criteria for determining whether the Commeliniflorae are para- phyletic (Fig. 6) or monophyletic (Fig. 7A). The Arecaceae-ingroup portion of the tree in Figure 6, which supports the former, is 1663 steps, whereas the latter is indicated by the 1665-step tree in Volume 80, Number 4 1993 Clark et al. PS ny of Bromeliiflorae— Commeliniflorae—Zingiberiflorae 997 Figure 7A. The choice between these two possi- bilities must await further study. Many family-level topologies are not firm enough to warrant conclusions at this time. However, the topologies shown here should be used as hypotheses for further study. Additional work should include analyses of molecular data from more genera of the larger families that are included here (e.g., Cyperaceae, Juncaceae, Commelinaceae, Haemo- doraceae, Zingiberaceae, Costaceae, and Maran- taceae) as well as from families that were not in- cluded in this study: Cannaceae, Centrolepidaceae, Eriocaulaceae, Hydatellaceae, Mayacaceae, Spar- ganiaceae, Thurniaceae, Xyridaceae, and Joinvil- leaceae. Family-level topologies should be exam- ined among gene trees of genes that are less conservative than rbcL. Most importantly, future studies should include cladistic evaluations of mor- phological and chemical data. Preliminary data from UV-fluorescent cell walls, helobial endosperm, and Strelitzia-type epicuticular wax show the po- tential of these kinds of characters for evaluating competing gene trees. Comprehensive surveys of these and other independent data will be necessary for making the best phylogenetic inferences among families of the Bromeliiflorae—Commeliniflorae— Zingiberiflorae lineage. LITERATURE CITED ALBERT, V. A., M. W. СНАЗЕ & B. D. MISHLER. 1993. Character-state weighting for cladistic analysis of pro- tein-coding DNA sequences. Ann. Missouri Bot. Gard. 80: 752-766 BArTHLOTT, W. & D. Ккбисн. 1983. e ogie und orientierungsmuster epicuticula re kristalloide: ein neues system врео mie bei monokotylen. Pl. Syst. Evol. 142: 17 Hi BEHNKE, H.-D. . BARTHLOTT. l w evidence from the ultrastructu ral and micromorphological fields in angiosperm classification. Nordic J. Bot 3: 43- BROWN, с. K. ФА. Ј. GILMARTIN. in Bromeliaceae — a 8 10-132 1989a. Stigma types y. 5 Bot. 1 . 1989b. Chromosome numbers in Bromeliaceae. Amer. J. Bot. 76: 657-665 CHAsE, M. W E. Sorris, R. С. OLMSTEAD, D. MORGAN, H DE E. GOLENBERG, С. Н. LEARN, JR., 5 AM, „СН. B TT AYANANDAN & V. АЈА ERT 1993. Plyviogeriatics of seed plants: An ans ey of nucleotide sequences from the man gene rbcL. Ann. Missouri Bot. Gard. 80: 528-580 CRONQUIST, А. 1978. The Zingiberidae, a new subclass мина Г (monocotyledons). Brittonia 30: 505. 988 e Evolution and Classification. of ados Plants, “ond ed. The New York Botanical Garden, New York. DaHLGREN, R. & K. BREMER. 1985. га clades of the angiosperms. Cts 1: 349-36 — & Н. T. Си 1982. The es dons. A быша puta Academic Press, New York. rad EPUM 55-39 ОМ gran evolution. Evol. Biol. 16 Ној. о 1985. Тће Families of the Monocotyledons. Springer- -Verlag, New Yor DritAPoRTA S. L., J. Woop & J. B. Hicks. 1983. Mai e DNA па у Maize Genetics Cooperation Newsletter 57: 26-2 DuvaLL, M. R., А М. W. Chase, №. D. CLARK, W. J. KRESS, Н. С. HiLLs, - a xp J. F. SMITH, B. S. Gaur, E. A. ER & ( LEARN, JR. 1993. Phylogenetic Pubs for the топос otyledons constructed from rbcL sequence data. Ann. Missouri Bot. Gard. -619. Еку, S. C. 1983. Feruloylated pectins fr from the primary cell wall: Their structure and possible functions. Plan- a 157: 111-123. GILMARTIN, А. J. & С. K. BRowN. 1987. Bromeliales, related cube and resolution of the relationships among Bromeliaceae subfamilies. Syst. Bot. 12: 49: 500. & . 1988. Response to Simpson. Syst. 13: 614-619. Cn ME E. M., D. E. Giannasi, M. T. CrEcc, С. J SMILEY, M. DURBIN, D. HENDERSON & С. ZIRAWSKI. 1990 Chloroplast DNA sequence from a Miocene di и species. Nature 344: 656-658. .J. & К. HARRI HARTLEY. 1 Phenolic con- stituents of the cell walls of monocotyledons. Bio- chem. Syst. : 8: 153-160. Hart ey, R. D. P. J. HARR 1981. Phenolic con- stituents of ih сей walls of шее дд Biochem Syst. Ecol. 9: 189-203. HUTCHINSON, J. 1973. The Families of Flowering Plants Arranged According to a New System Based on Their Probable Phylogeny, 3rd ed. Oxford Univ. Press, l andon, Kress, W. J. 1990. The phylogeny and classification “of the Zingiberales. Ann. Missouri Bot. Gard. 77: 698-721. Kusitzki, К. 1987. Phenylpropanoid metabolism in relation to land plant origin and diversification. J. PI. Physiol. 131: 17-24. LEARN, G. H., JR. Ribosomal keg in Clematis fremontii (Ranunculaceae). P tation. Washin Могслакр, Р. & H. Ravn. 1988. of caffeoyl ester distribution in dicotyledons. Phy- tochemistry 2 11-2421. SANGER, Ni CKLEN & A. R. Corson. 1977. DNA sequencing with chain terminating inhibitors. Proc. Natl. Acad. Sci. U.S.A. 74: 5463-5467. Simpson, M. С. 1985. Tou ultrastructure of the Te- и Grana 24: 77-92. 1987 Pollen ultrastructure of the Ponted- . Deg De 998 Annals of the Missouri Botanical Garden eriaceae: diio for exine homology with the Hae- modoraceae. Grana 26: 113-126. 98 A critique of “Bromeliales, related monocots, and resolution of relationships among Bro- meliaceae subfamilies.” Syst. Bot. 13: 610-614. Phylogeny and classification of the Haemodoraceae. Ann. Missouri Bot. Gard. 77: 722- 84. STRACK, D., J. HEILEMANN, M. MÓMKEN & V. 1988. Cell wall-conjugated phenolics from Coniferae leaves. Phytochemistry 27: 3517-3521. SWOFFORD, D. L. PAUP: Phylogenetic Analysis Using Parsimony, Version 3.0r. Software distributed by the Illinois Natural History Survey, Champaign, TAKHTAJAN, A. L. 1980. Outline of the classification of E plants (Magnoliophyta). Bot. Rev. 46: Two R. Е. 1992. An Pares classification of the wering 3t Aliso 13: 365-389. KER, J. 1986. Classification and evolution of the шл EA Amer. J. Bot. 73(5, suppl.): WAL WiLLIAMS, C. A. 1978. The systematic implications of the complexity of leaf flavonoids in the Bromeliaceae. ytochemistry 17: 729-734. WiLLIAMS, C. A. & J. B. HARBORNE. 1977. The leaf avonoids of the Zingiberales. Biochem. Syst. Ecol. 9: 221-229. Illinois. APPENDIX. Specimens used for sources of DNA. tabase Species Family Voucher /Source accession? Aechmea chantinii (Carriére) Baker Bromeliaceae G. K. digi 3156 (WYO) L19978 Ananas comosus Merr x Bromeliaceae UCRBG s L19977 Catopsis monta mith Bromeliaceae G. K. 3155 (WYO) L19976 Glomeropitcairnia релй Mez Bromeliaceae SEL 8 L19975 Hechtia montana Brandege Bromeliaceae A. PHAR ee Е МА) L19974 Pontederia sagittata С. Pres Pontederiaceae S. C. H. Bar L20128 Puya dyckioides Mez Bromeliace UCRBG 76- 16 L19973 Ravenala madagascarensis Strel. Strelitziaceae UCRBG s.n.' L20138 Stegolepis allenii Steyerm. Rapateaceae SA s.n.* L19972 Tillandsia elizabethae Rauh Bromeliaceae A. requi 2638 (UCR) L19971 Vellozia sp. Velloziaceae SEL 8 L19970 Selby Botanical Garden, living specime ви vien e DNA of Ana Botanic Garden, living specime * GenBank accession numbers uu of California at Riverside пере ан Garden, living specimen. ? Mar P. sagittata from ii U ен of Toronto. NOTES ON MAGNOLIACEAE III: THE MAGNOLIACEAE OF CHINA’ Chen Bao Liang}? and H ans P. Nooteboom? ABSTRACT Five genera of d occur in China; these and the 81 indigenous and 7 cultivated species they contain are described here. In Magno u, is recognized with one d in China. reduced to Michelia figo var. crassipes. In , 28 wild and 4 cultivated species are reco nized. М n, is recognized with one species, Michelia Michelia о is newly described. Michelia crassipes is Liriodendron one indigen nitida var. robusta i 15 d one cultivated species are recognized. An identification list of all the n id studied, as well as a bibliography of Chinese Magnoliaceae, is given at the end of the pap The first author collected specimens of Mag- noliaceae over a period of several years in China, especially in Yunnan. Apart from describing many new species, he wrote his thesis (in Chinese) on the Magnoliaceae of China. Because he did not have the opportunity to study the types in Western herbaria, or enough collections from southern Asia outside China, his earlier work could only be pre- liminary. The second author published on the generic delimitation of Magnoliaceae and gave an account of the genera and species occurring in Malesia (Nooteboom, 1985, 1987). The present publication is the result of a joint Dutch-Chinese cooperation spanning 18 months at the Rijksherbarium at Lei- den Sadly, soon after Chen Bao Liang returned to China in May 1991, he died from cancer. 'TERMINOLOGY A few terms are used in this study that are not familiar to all taxonomists. Pedicle is not a pedicel, but the internode, if present, between the upper- most bract and the perianth. Literally, a brachy- blast is a short shoot. In Magnoliaceae it means the shoot that bears the flower: it is terminal in Magnolieae and axillar in Michelieae. In Magno- lieae, however, this character is called a peduncle because often it is longer. Innovations are the youn- gest parts of the twigs, often including stipules, that are still growing. A list (List 1) of p and keys in the order in which they appear in the text follows the Bibliography. The numbering system in pn text corresponds to that of List 1096. 1, Page Magnoliaceae A.L. de Jussieu, Gen. Pl. 280. 1789 (Magnoliae). Trees or shrubs, glabrous, or with an indumen- tum of single hairs. Leaves spirally arranged, sim- ple, entire or 2- 10-lobed, penninerved, evergreen or deciduous; stipules present, at first enclosing and protecting the buds, early caducous and leav- ing an annular scar around the node. Flowers ter- minal or pseudoaxillary on a short shoot (brachy- blast) in the axils of the leaves, bisexual, rarely unisexual. Brachyblast or peduncle bearing | or more caducous spathaceous bracts, which leave annular scars. The highest bract either directly below the perianth or a pedicle present between bract and perianth. Perianth spiral or spirocyclic, ! We thank the directors and curators of Р, К, E, and BM for the opportunity to study their collections and of the following herbaria for sending collections on loan: A, BM, SYS, US, W. We thank E, GXMI, IBSC, K, KUN, LBG, MO, NY, P, PE, the Dutch minister of Science and Education for providing, i in the framework of the Chinese- Dutch Scientific Cooperation, the funds for Chen Bao Liang to spend 18 months in the Netherlands. shan University, Guangzhou, China * Rijksherbarium/Hortus Botanicus, P.O. Box 9514, 2300 RA Leiden, The Netherlands. ANN. Missouni Bor. Савр. 80: 999-1104. 1993. 1000 Annals of the Missouri Botanical Garden undifferentiated or differentiated into a (pseudo-) calyx and corolla, perianth members 6 or more, free, imbricate. Stamens numerous, free, spirally arranged; filaments short or + elongated; anthers linear, 2-locular dehiscing introrsely, latrorsely, or rarely extrorsely; connective usually + produced into an appendage. Gynoecium sessile or stipitate (and then a gynophore present); carpels numerous to few (rarely one), spirally arranged, free ог зоте- times concrescent; ovules 2 or more, biseriate on the ventral suture. Fruit apocarpous, sometimes syncarpous; fruiting carpels (follicles) opening along the dorsal and/or ventral suture, or circumscissile, rarely indehiscent. Seeds 1 or more in each fruiting carpel, large, in dehiscent carpels with arilloid testa and pendulous from the elongated spiral vessels of the funiculus (in Liriodendron, with indehiscent fruits, adherent to the endocarp and not arilloid); endosperm copious, oily; embryo minute. Note 1. Nooteboom (1985) provided an in- troduction to the family with chapters on anatomy, embryology, phytochemistry, palynology, mor- phology, and subdivision of the family. Dandy (1927а) gave an account of the genera, which ће emended in Hutchinson (1964a) and Praglowski (1974). Praglowski (1974) gave an account of the pollen. The cultivated species were extensively treated by Treseder (1978 Note 2. As in many other families, the delim- itation of species in Magnoliaceae is often difficult because of high variability. Because recently many more кы and also types of species that d before World War II, became avail- bi many species could be reduced to synonymy. It should be kept in mind that even within popu- lations variability is high. For example, Magnolia sprengeri Pampanini, when grown from seed, dis- plays much variability. The beautiful Magnolia var. diva is vegetatively propagated because the seedlings often are disappointing, showing a spec- trum of offspring with not always attractive flowers. Several botanists would be inclined to describe these seedlings and their mother tree as many different species. KEY TO THE GENERA OF MAGNOLIACEAE l. Leaves 2- 10-lobed, the apex truncate or wide- ly emarginate; anthers extrorse; fruits sama POE ыыы = сыс iriodendron 1. Leaves entire or occasionally 2- ginate; anthers introrse or loce Vr dehiscent or circumscissile, n a io Growth monopodial; flow chy td in the axils of the leaves; gynoecium m dint stipitate. ___ . ТУ. Michelia 2(1). 2. Growth sympodial; flowers terminal on the ol. т эсе е у A с ысы ыыы 3(2). Flowers unisexual; Moni 6-7, subequal. ..... Kmeria 3 Flowers "bisexual c or -androdioecious; ‘tepals 9 r more, sometimes the outer whorl calyxlike 4 4(3). Ovules 4 or more in each carpel; the hair base replaces a normal epidermal on so that the loss of ir leaves a роге the cutical membrane. s ee Manglietia 4 Ovules 2 in each carpel (4 in Magnolia ka- chirachirai, 2-5 in Magnolia cathcartii), sometimes 4 in the lower carpels; the hair and a hair does not leave a pore after it is I = · Magnolia MAGNOLIACEAE subfamily MAGNOLIOIDEAE Leaves entire or occasionally 2-lobed at the apex; stipules free from the petiole or adnate to it. An- thers introrse or latrorse. Follicles longitudinally least the base remaining adnate to the torus, free or con- dehiscent or circumscissile and then at crescent into a syncarp. Testa free from the en- docarp, externally arilloid. TRIBES Tribe Magnolieae Growth sympodial (flowers terminal on the twigs). Follicles free or concrescent. Genera: Magnolia, Manglietia, and Kmeria. Tribe Michelieae Law Yuh-wu, Acta Phytotax. Sinica 22: 106. 1984 Growth monopodial (flowers arising on brachy- blasts in the axils of the leaves). Genera: Elmer- rillia Dandy, with four species in Malesia, none in China, and Michelia L. (See page 1052.) TRIBE MAGNOLIEAE I. Magnolia L., Sp. Pl. 535. 1753. TYPE SPECIES: Magnolia virginiana L. (eastern United States). Talauma Juss., Gen. Pl. 281. 1789. Magnolia sect. Talauma Baill M ill., Adansonia 7: 3, 66. 1866. Mag- 1 1880. TYPE SPECIES: Talauma plumierii ине А. DC. A eru plumierii Schwartz). . 1825. Talau Aromadendron Blume, Bijdr.: a sect. Magnolia elegans (Blume) H. Keng. pos Nees, i ora x 152. 1828 nom. гејес., non Blu- D ns. Magnolia sect. Blumia (Nees) Baill. Adansonia 7: 2. 1866. TYPE SPECIES: Talauma Volume 80, Number 4 Chen & Nooteboom 1001 1993 Magnoliaceae of China i Blume = Magnolia candollii (Blume) Н. KEY TO THE SUBGENERA AND SECTIONS OF MAGNOLIA дапа ©, nach, Hist. Natur. Veget., Phanerog. 7: 462. 1830. тт ҮРЕ SPECIES: Yulania conspicua Spach = 1. ~ т та Mic. of пене gnolia a e Dandv. E а е ri EM rsal suture; stipules . His Phanerog. 7: ree from the petiole onna Telipsstrum. Spach atur. Veget., 481. 1839. TYPE SPECIES: Magnolia acuminata L. АНЕ ped , Hist. TYPE SPECIES: Lirianthe grandiflora Spach : 6 nolia ПР Roxb. nl dins Siebold & Zuc ee Math.-Phys. Cl. Kon. Bay Pes Wiss. 4, 3. 1846. TYPE SPECIES: Bue a stellata Sichold & _ = Magnolia inia (Siebold & Zucc.) M Б Dandy, Kew Bull. 1927: - 260. 1927. TYPE SP E du cathcartii (Hook. f. & Thom — Magnolia cathcartii (Hook f. & ot. Svenhedinio Urban, Repert. Spec. Nov. Regni Veg. 24: TYPE SPECIES: Svenhedinia minor е a e minor Urban). Parakmeria Hu & Cheng, Acta Phytotax. Sin. 1 51. TYPE SPECIES: Parakmeria omeiensis Hu Cheng (= Magnolia omeiensis (Hu & Cheng) Dan- to 2): 1 Micheliopss H. Keng, Quart. J. Taiwan Mus. 8: 207, 195 A TYPE SPECIES: Micheliopsis kachira- pues (Kaneh. & Yamam.) H. Keng — Magnolia и (Karek. & Yamam.) Dandy. РАС Lozano-Contreras, Caldasia 11: 33. 975. TYPE SPECIES: Dugandiodendron mahechae ү, Trees or shrubs, stipules adnate to or free from the petiole. Flowers terminal, solitary, bisexual. Tepals 9-21, 3-5-merous, subequal or more rarely the outer whorl forming a calyx. Anthers introrse to latrorse; connective produced into a longer or shorter appendage or rarely unappendaged. Gy- noecium sessile or in some species stipitate; carpels many to few, usually free but not connate in Mag- nolia subg. Talauma, in young fruits of Magnolia sect. Gynopodium, and in some other species. Fol- licles free, crowded, dehiscent along the dorsal suture; in Magnolia subg. Talauma the basal parts of the mature carpels staying adnate to the torus, the apical parts falling and thus exposing the seeds which hang from the lengthened funiculus, some- times the apical parts during or before falling dor- sally dehiscing; in some species of Magnolia sect. Gynopodium the carpels are connate but tear apart when maturing and dehisce dorsally. Ovules gen- erally two in each carpel, rarely 3 or 4 in the lower carpels, in a few species to 4 or 5 in all carpels. Ripe seeds often hanging from the funicles, which lengthen through uncoiling of the spiral vessels. Distribution. About 120 species, of which ca. one- remainder mperate and tropical Southeast Asia from the Himalayas to China, Japan, and Malesia prominent for the whole length (in the absence of fruits this section cannot be distinguished from section Gwillimia). с а. Іс. Subgenus Talauma 1. Section Blumiana . Anthers dehiscing laterally or sublaterally; s precocious and/or with a much re- calyxlike outer whorl of tepals; plant deciduous; fruit cylindric or oblong, usually = 1 . Ib. Sub nia. N = — — 2. dedi dehiscing introrsely; flowers neither a co nor with a much reduced (calyx- like ye outer а of tepals; plant evergreen or deciduous. Ia. Subgenus Magnolia a Tepals subequal; flowers appearing before the leaves, white to rose or purple. уны E дина Ib-1. Section Yulania 3. Tepals very unequal, those of the outer w forming a calyx; flowers s appearing before, to- gether, or after the lea Flow wers appearing before the leaves; inner (large) tepals white, sometimes tinged with rose or purple. Ib-2. Section кч d Flowers appearing together or after the leav inner tepals purple or green to yellow Ib-3. Section Tulipastrum Я Stipules quum to the petiole; leaving a scar on its surface; plants evergreen or i dus gonoeci sessile: ыы шышы ачасынын а Stipules free from the petiole, the latter = plants evergreen; gynoecium sol rir EU ———————— re Plant: evergreen; flower buds at first i ИА in one or more spathaceous bracts, wh ац as many annular scars on the peduncle. ...... 1. E Guillimia Plants кене flower buds at first enclose ina single spathaceous bract, Wd leaves a single annular scar on the pedun ы crowded into false vacas at e ends of the twigs, usually large or very large; con- nective of anthers produced into a short or 3(2). 4(3). e E = E 6(5). e 7(6). . Carpels usually with 2-5 ovules; 5 sexual; fruits cylindrical, 1-4 seeds іп each follicle; stamens with short bert and very long anther, hiding the gynoecium; at least twigs hairy. Ia-5. Section peundnara 8. Carpels usually with 2 ovules, wer with 4 1002 Annals of the Missouri Botanical Garden flowers androdioecious; plant entirely gla- brous; anthers much shorter. ___________ = Ia-4. Section Gynopodium Y TO THE SPECIES OF MAGNOLIA (primarily based on floral characters) 1. Leaves crowded at the end of the branchlets in false whorls, glaucous beneat , Leaves evenly distribute 3 2(1). Twigs and stipules glabrous; young leaves icle 13 mm; outer tepals oblong- is in- ner dien 11-13 cm long; stamen m long; gynoecium cylindric. о 9. Magnolia rostrata 2. Twigs and stipules hairy; leaves scattered, colorless hairy beneath, base cuneate, apex stamens 30-40 mm long; En ovoid. nolia officinalis 3(1). Anthers В laterally ог Stub 4 3. Anthers dehiscing introrsely. 4(3). ong о. а ошег pena 0. 8-2. + ; Tepals subequal outer tepals 5-14 cm long. 7 5(4). Inner tepals 4-5 cm lo p tree to 12 m high; outer tepals 8- ж ео 25. Magnolia biondü 5. pe tepals 5.5-10 cm long; shrub or tree- let to 3-10 m high; outer tepals 10-27 с mm. 6(5). Leaves glabrous to pubescent with long, miri colorless hairs beneath; leaf base attenuate-cuneate; flowers appearing to- gether or after the leaves, purple or red; filaments 2 mm long; peduncle 6-8 mm CTC) Loose agnolia a 6. Leaves with fine, straight, colorless beneath; leaf base cuneate or EE cu- neate; flowers side айа before the m ma filaments 3 mm long; pedunc thi . Ma ae agnolia а 8 8(7). br yellow-, black-, or gray- Es dl ot glaucous beneath, reticulation coarse; ы purple or red 2. Magnolia Md 8. Twigs yellowish to green turning r h brown; leaves m us beneath, аде tion fine; flower w aoe dawsoniana 9(7). Peduncle 24-30 x T x MUNDUS E 9. Magnolia campbellii у Peduncle 2-20 mm pen 10 10(9). Twigs hairy at least ics young. |... Mido heptapeta 11 10. Twigs glabrous 11(10). Stamens 9- 10 mm long; leaf apex a minate or acute . 18. Ma aces amoena Li. bien 15-17 mm m long; leaf apex round- ed or short-acuminate. __ 12(11). F pane 4 mm i shrub to 7 m high; 16(15). 17(15). 18(17). 19(18). =“ е winter ~ tong: hairy, sericeous; nerves in 8-12 ; leaf base broadly р. or rounde i yim appearing fore the leaves, purple o e " 4. Magnolia zenii Filaments 5-7 mm long; tree up to 20 m high; terminal winter buds densely villous; nerves in 6– 8 pairs; leaf base cuneate or narrowly cuneate; flowers P NE to- gether or after the leaves, white. _______ Magnolia sprengeri 14 Stipules free; gynoecium stipita Stipules adnate to petiole; g Динин алеје not stipitate. 0 P . Gynoecium exserted from stamens. ................. 15 ue entirely hidden within androe- | Peduncle 3-4 mm thick; leaves elliptic or obo vate, base rud 4. Magnolia kachirachirai Peduncle 5-6 mm chick: leaves ovate or narrowly elliptic, base broadly cuneate, nar- rowly cuneate, or rounded Midrib of leaf not prominent above; leaves mm; outer tepals oblong. 16. Ma 2 omeiensis Midrib of leaf prominent above, at least toward base; leaves ovate, base broadly cu- 4 mm long; filaments 1-3 mm; outer tepals obovate or noe БОНЕР 5a. Magnolia nitida var. nitida Outer tepals s 5-6 cm long; plant hairy at lea leaf fine; petiole hairy; peduncle 2 mm thick; inner tepals 5.5 cm lo ong 25 mm bro 7. Magnolia cathcartii Outer tepals 2.5-3. 3 cm long; plant en tirely glabrous; leaf base cuneate or nar- rowly cuneate, reticulation of leaf coarse; Ape 5-9 mm thick; inner tepals 2.2- 4.2 cm long, 1.2-18 mm broad. _______. 18 cin tepals 3.8 cm long; leaf base nar- rowly cuneate, midrib not prominent above; outer Маја oblong; inner tepals 3.5-4.2 Gaia ране 6. Magnolia omeiensis Оше ub 2.5-3 cm long; leaf base cu- sary midrib prominent above, at least to- ard base; outer tepals ovate; inner tepals 22.3 cm lo DE Nerves in 8-15 pairs; reticulation distinct; connective appendage 0.5 mm long; gy- oe ow mm long; ш twigs ca. ic бине 15е. Magnolia nitida var. robusta Nerves in 9-13 pairs; reticulation ra mm long; fruiting twigs 4-6 mm thick коео 5b. Magnolia nitida var. lotungensis 21 20(13). Gynoecium hairy; reticulation fine. |... Volume 80, Number 4 1993 Chen & Nooteboom 1003 Magnoliaceae of China 20. 21(20 26(25). 27(23). 30(2 8). Gynoecium glabrous; reticulation coarse. 23 Carpels 55-64; twigs densely appressed, yellow, woolly; leaf base acute to rounded, rarely cordate; gynoecium ovoid. . 4. Magn nolia delavayi 15-24; twigs pubescent or silky; Carpels noecium ellipsoid; follicles glabrou 2. Magn olla championi Leaves glaucous beneath; twigs silky; p duncle stout; connective appendage short- pointed; g | hairy. оуу 1. Ma agnolia albosericea iode hairy; plant deciduous; connec- tive of anther not produced into an ap- endage, retuse or blunt. 4 Peduncle glabrous; plant evergreen; con- aon produced into a short or long ap- 27 age. E odes 23-25 mm та 0. Magnolia. globosa Gynoecium 13-20 m ин ЧН 29 P not glaucous би leaf. apex unded or mucronate; tees 13-17 m inner tepals ovate. 2. парни sinensis Leaves glaucous Tti leaf apex na or acute; pedicle 2-7 or 8-15 mm; ner tepals obovate, spathulate, or oblong. 2 с le stinct; outer ong; gynoecium bas о ae peduncles 3-4.5 cm long. ............... у ™ Magnolia wilsonii Leaves obovate; twigs pubescent; stipules pubescent or sparsely hairy; reticulation ш obscure; outer tepals obovate; fila- ng; gynoecium Ado uiti fruiting peduncles 6.2- m m lon З Magnolia sieboldii "bie 2.5-5 ст wes uri no nent above; petiole 5-15 mm; sie —áÓ 3. Ma iis coco broad; midrib prominent ie at least toward base; petiole 25-110 m; m; peduncle 23-120 mm long. sonic: 26 Plant hairy at least in innovations. с 29 Plant glabro 30 . Pedicle saben connective appendage long; fi cumscissile and falling, dehiscing along the dorsal suture or not, the basal parts re- maining adnate to the torus; fruits ellipsoid. 30. Magnolia candollii var. obovata Pedicle ipo connective appendage 1 mm long; filaments 1-2 mm long; gynoe- cium narrowly a fruits ovoid to ovoid- cylindric, at least finally T of gs > which dehisce along the dorsal s 5. Magnolia henryi Peduncle 5-13 mm thick; tree to 25 m high; leaf apex rounded or short-acuminate; outer tepals thick-fleshy; gynoecium ellip- soid. . Magnolia candollii var. obovata Peduncle 4 mm thick; shrub to 3 m high; leaf apex acuminate; outer in thin-co- riaceous; gynoecium narrowly ovoid. . 6. Magnolia наш M 30. KEY TO THE SPECIES OF MAGNOLIA (primarily based on fruit characters) l. Fruits. with connate nian when mature ts of the carpels circumscissile i Eine dehiscing along the dorsal su- ture or not, the basal parts remaining ad- nate to the torus. " 1. Fruits at least finally consisting of free car- pels which dehisce along the dorsal suture. З Tree to 25 m high; apex of leaf rounded 5-13 mm Magnolia candollii var. obovata 2. Shrub or treelet to 3 m high; apex of lea acuminate; din 8 4 mm thick (fruits not known). Magnolia es 4 2(1). 3(1) Stipules free; gynoecium stipitate. . E 3. Stipules adnate to petiole; gynoecium- not stipitate. 4(3) Midrib of leaf not prominent above. ______. 5 4. Midrib of leaf, at least toward base, promi. nent above 7 5(4) Twigs hairy at least when young; stipules pubescent; base of leaf broadly cuneate or rounded, apex acuminate, reticulation dis- tinct, fine, petiole ha Magnolia cathcartii 5. Twigs and stipules glabrous; base of leaf cuneate or narrowly cuneate, apex short- acuminate or acute, reticulation rather ob- scure, coarse, petiole glabrous. „u 6 Leaves narrowly d base narrowly cu- neate (fruits not known 16. Ma nh ‘aii 6. Leaves elliptic or vu base c . Magnolia ооа 8 Leaves ovate 7 Leaves eliptic or ур SMPUE: vecinos 9 Leaves ong, ls dis- tinct, base cuneate, apex rounded, acumi- nate, or acute, petiole 16-47 mm; fruiting twigs ca. 8 mm thick. RORIS 15c. Magnolia nitida var. robusta 8. Leaves 6-12(-14) cm long, reticulation rather obscure, base broadly cuneate or rounded, rt-acuminate, petiole 15- 25 mm; fruiting twigs 5-6(-8) mm thick. јез Magnolia nitida var. nitida Fruiting twigs 8 mm thick; leaves 7- m long, retiulaton distinct, = el 7 mm; scars of per ds ns along torus o ruit 6-7 | ong. Pii e nitida var. robusta mm thick; leaves anex sho 9(7). рае stamens s torus under fruit 2-4 long. . 15b. Magnolia nitida var. lotungensis 1004 Annals of the Missouri Botanical Garden 10(3). Leaves crowded in false whorls terminal on 21(20). Tree to 20 m high; plant hairy at least in the branchlets. innovations; stipules pubescent, stipular 10. Leaves evenly distributed. 12 sca 0 mm long; apex of leaf round- 11(10). Twigs glabrous; stipules glabrous; young ed or acute; peduncle 6-10 mm thick; fruits leaves beneath with brown hairs, base кй to ovoid-cylindric, 7.5-14.5 x 3.5 rounded, apex mucronate; pedicle 13 mm 5. Magnolia ин» fruit cylindric, 12-20.5 cm long, follicles дү; Shrub to 3 m high; dei mu. glabrous; with a beak of 6-8 mm stipular scars 7-25 mm long, apex of leaf 9. Magnolia rostrata acuminate; peduncle pi mm thick (fruits not 11. Twigs hairy at least when young; stipu known). ............... 6. Magnolia phanerophlebia sparsely hairy; young leaves beneath ith _ Peduncle 21-65 mm ici ne 23 colorless hairs, base cuneate, apex rounded eduncle 2-20 mm 25 or acute; pedicle 2-6 mm; fruit usually Ln Shrub or tree to 5 m high; stipular scars oblong-ovoid, 8-16 cm long, follich swith åž 0-30 mm long; fruiting peduncles slender. га of је mm. .. 8. Magnolia officinalis 12. Magnolia sinensis 12(10). Follicles hairy. 13 23, Tree to 20-30 m high; stipular scars 2-7 12. Follicles glabrous 14 mm long; fruiting peduncles stout. ................. 24 13(12). Twigs, stipules, and leaves glabrous, leaves 24(23). Twigs yellow-brown or black-brown to gray- not glaucous beneath, midrib at least toward brown, glabrous; petiole glabrous, stipular se prominent above, reticulation coarse, scars 2-5 mm long; fruits 3-20.5 cm long; p dilated at base; пети glabrous, flower purple or red; gynoecium 8-12 mm m long, 4 mm thick; car rui high. .22. Magnolia sargentiana not nom) 6. Mag койа. phanerophlebia 24 паса ит eae К b : 13. Twigs hairy at least a young; stipules | Б р ve кы жин MORE E hae пе | dur least wh gray when dry, glabrous or pubescent; pet- BERTS vr Mercier еч sit le glab llowish pubescent, stip- | s beneath, midrib not prom- wed eg ин da Duy d po ассо» часа ај odii ular scars 3- 1) mm long; fruits 8.5- inent above, reticulation fine, petiole hairy 17 cm |l A k hi . | cm long; Hower pin or white; gynoe- not dilated at base; peduncle oe 15 mm cium 25-40 mm high. long, 6 mm thick; e 2 [— e > Марло т баг Magnolia albosericea agnos campbell gnon ' 25(22). Plant an entirel d ever sh Tul Leaves glaucous ar 15 УБ i. dept Leaves not glaucous beneath. |... 19 leaf apex acuminate; fru о em lon EN Fruiting peduncles 6.2-7 cm long . 16 25 Plant h | =) Ма gnolia ida Fruiting peduncles Ug em long. тт 7 ant hairy at least in innovations, a dec id- TT Leaves elliptic or ovate, reticulation dise. uous tree, leaf apex rounded, short-acu- tinct; nerves in 8-11 pairs; stipules yellow- a minato, OF Scute; fruit 52008 pm long: va 20 ih s Шу ‹ ar Bou villous; fruits 2.5-9 x 26(25). Treelet to 7 m high; petiole hairy; leaf apex 2-2. perge Magnolia globosa short-acuminate; peduncle 2-8 mm lon 16. ToS pien non rather ob . 24. Ma gnolia seni scured; nerves in 6-8 pairs; stipules pu- 26. Tree to 20 m high; petiole glabrous; -= bescent or sparsely hairy; fruits 3-4 x apex rounded or acute; peduncle under E pe 11. Magnolia sieboldii 20-35 mm long. . 22. Magnolia aia 17(15). Fruiting pedune cles 2-3 mm thick; stipules 27(19). Twigs hairy at least when young. . 28 silky, reticulation of leaf coarse; stipular 27. Twigs glabrous 33 scars on petiole 10-25 m m lo Н 28(27). Evergreen крене fruits ovoid-elliptic; hairs рт шш colorless or yellowis 17. Fruiting peduncles 8- 10 mm thick: stipu nolia championii woolly, pubescent, or villous, reticulation of 28. sone tree; fruits nee hairs col- le af fine, stipular scars 3-65(-85) mmlong. 18 pe 18(17). Evergreen tree; twigs and stipules densel 29(28). Leaves {зеш pubescent with fine, short pressed-woolly with yellowish hairs; pairs raight hairs. с. 26. Magnolia cylindrica of lateral nerves of leaf 11-16(-21), petiole Ф Ме bx different hairy beneath. ........... 30 45-65(-85) mm with stipular scar 40- 30(29). Leaves obovate. 65(-85) mm long; fruits ovoid. 000 0. Leaves elliptic, ovate, narrowly elliptic, or 4. Magnolia delava yi narrowly obovate 18. Deciduous tree; twigs glabrous to sparsel 31(30). Shrub to 3.5 m high; twigs 2-4 mm diam., pubescent; stipules per Lens of di glabrous or sometimes hairy on a few nodes eral of leaf 8-10, 15-30 near the peduncles; leaf base attenuate- mm with stipular a 3-4 т g; fruit cuneate, apex acuminate or acute; fruits cylindric | Маршы dawsoniana .9 X 1-2 ст. 19(14). Fruiting peduncles 2 : 5- е 5 em long. 20 29. "e dcin кшш Я ruiting peduncles 1-2 ст long. ...................... 27 3l. Tree to 20 m high; twigs 3-7 mm dia 20(19). Leaves 26-38(-65) cm n longs midrib prom- finely appressed- жашсын when ТЫШЫ leal inent above, at least toward base. о... base cuneate or rounded, AE ai -acu- 20. Leaves LK i cm long, "алб n not promi- minate; fruits 7. 5- "l5 x 2.5-5 ст. ш... nent a . 21. Mono heptapeta Volume 80, Number 4 1993 Chen & Nooteboom 1005 Magnoliaceae of China 32(30). d to = 5 m high; leaves = base -cuneate; fruits 3-6. on agn а quinquepeta 32. Tree to 12 m hig h; leaves ovate, narrowly elliptic, or по obovate, base cuneate or rounded; fruits 6-14. i cm long. 33(27). Leaves 31-36 x 9.5- TA FORAS a Magnolia phanerophlebia 34 33. Leaves 6-21.7 cm long. о 34(33). Plant od кете жш = Мавпойа сосо 34. un deciduous, hairy at least in innova- ш Leaf apex rounded or short-acuminate. ___ 36 petiole 6-15 mm; fruits 5-20 cm long; flowers appearing before the hig ts Magnolia zenii 36. Tree to 20 m high; nerves in 6-8 pairs; leaf de (narrowly) пык poni 10- mm; fruits 6-1 l "id ap- pearing together or s the leave NE sprengeri . Fruiting peduncles 4. a 6 mm thic 38 ruiting den 3-4 mm u m high; leaves elliptic or ob- ers appearing before the leaves. _______ 25. Magnolia biondii Shrub to 3 т s stipular scars as long as petiole, 3 -15 m 39(37). neate; = appearing together or after the lea 29. 39. Tree to da n high; stipular scars adnate mm 10-13 pairs of lateral nerves, base cuneate, ; flowers appearing before 18. Magnolia amoena liq the leaves. _______ Ia. Magnolia subgenus Magnolia Anthers dehiscing introrsely. Flowers neither precocious nor with a much reduced (calyxlike) outer whorl of tepals. Leaves evergreen or decid- uous. la-1. Magnolia section Сутла A. DC., Syst. Nat. 1: 455, 548. 1817. TYPE SPECIES: Mag- nolia coco (Lour.) DC. Evergreen shrub or tree. Stipules adnate to pet- iole to its apex. Leaves evenly distributed; midrib prominent above, at least toward base, or not prom- inent above. Flowers bisexual; tepals subequal. Out- er tepals coriaceous; connective produced into a short or long appendage; anthers dehiscing in- trorsely. Gynoecium not stipitate. Number of ovules per carpel 2, sometimes 4 in the lower carpels. Scars of bracts 1—4. Fruits at least finally consisting of free follicles, which dehisce along the dorsal suture; follicles short-beaked, beak not dorsally flat- KEY TO THE SPECIES OF MAGNOLIA SECTION GWILLIMIA 1 Gynoecium hairy; reticulation of leaf fine. |... 2 l. Gynoecium glabrous; reticulation of leaf coarse. 4 2(1). Leaves not glaucous beneath; twigs colorless EM gnolia о s beneath; wie ae silky yellowish d peduncle stout; Mist н ovoid or narrowly ovoid. . Petiole 4-8.5 cm ; carpels 55-64; twigs and stipules yellowish woolly; leaf base acute to cordate, apex connective Нава a of anthe M к N ~ oecium ovoid, fruits ovoid, follicles rous. . Magnolia у 3. Petiole 1-3.5 cm long; nea ca. 22; twigs and stipules silky; leaf base cuneate or nar- rowly cuneate pointed; gynoecium narrowly ovoid; fruits ob- ovoid or ellipsoid; ilice hairy. Maeli a ici ea . Leaves 10-17 cm long, > E. 9 cm broa ~ A = ~ = = . Magnolia henryi 5. | Peduncle 4 mm thick; ‘shrub to 3 m high, E glabrous, apes of leaf acuminate, stip- ular scars 7-25 mm long. Magnolia phanerophlebia . Magnolia albosericea Chun & C. H. Tsoong. Acta $ Sin. 9(2): 117. 1964. TYPE: na. Hainan: Poting, F. C. How 72740 (holotype, IBSC: isotype, A). Evergreen treelet to 8 m high; bark gray; young twigs, stipules, leaves beneath, midrib both sides, petioles, bracts, peduncles, and tepals outside white- sericeous, glabrescent; gynoecium yellow pubes- cent; young twigs ca. 4 mm diam. Stipules adnate Annals of the Missouri Botanical Garden to the petiole, stipular scars nearly to the base of the blades, 9-34 mm long. Leaves coriaceous, dark green above, glaucous below, elliptic, narrowly el- liptic or narrowly obovate-elliptic, 18-30(-40) x 6-9(-15) ст; apex acuminate, base cuneate to narrowly cuneate; midrib impressed above, prom- inent beneath; nerves fine, prominent below, in 15— 20 pairs, reticulations finely netted, visible on both sides. Petiole 1-3.5 cm long. Peduncle ca. 1.5 x 0.6 cm. Tepals 9, the outer 3 long-elliptic, green- ish, white sericeous outside at the midrib and the base, 4.5-5.5 x 1.3-2.5 cm, the inner tepals 6 very unequal, the innermost ones smaller than the others, spathulate-elliptic to broadly obovate, usu- ally constricted basally into a short claw, 3-6.2 x 1.8- cm; stamens ca. 120, 1.5-2 cm long, anthers introrsely dehiscing, 6-8 mm long, con- nective appendage short-acute; gynoecium narrow- mm high, carpels ca. 22, 8-13 mm long. Fruits ellipsoid to obovoid, 7-10 x 4- ly ovoid, ca. 5 cm; mature carpels coriaceous, narrowly elliptic, brownish pubescent with short hairs and incon- spicuously papillate, 2-2.5 cm long, beak re- curved, 5-7 mm long. Seeds 14 х 6 mm Distribution. China and Vietnam. In CHINA: Hai- nan, Baoting, Lingshui: Feng K.M. 72059; Zheng 12506; How F.C. 72059; 72740. Ecology. In evergreen broad-leaved forests along streams. Altitude: 500-800 m. Flowering April-May, fruiting August-September. Collector's notes. Bark gray; leaves deep green, lustrous above, glaucous, green beneath; fruits green. otes. This species was treated as Magnolia fistulosa (Finet & Gagnep.) Dandy in W. Y. Chun (1963) and as Magnolia champacifolia Dandy ex Gagnep., nomen nudum, in Humbert (1938). Ax Magnolia championii Benth., Fl. Hong- kong: 8. 1861. Magnolia pumila var. cham- pionii Finet & Gagnep., Bull. Soc. Bot. France (Mémoires) 4: 36. 1905. ie eine lilifera var. championii Pamp., Bull. . Tose. Or- tic. 4(1): 136. 1916. TYPE: Champion 37 (holotype, K). Magnolia а а & Gagnep.) ui Notes Roy. Bot h 16: 124. 1928. Talauma fis im Fino: & Се» Bull. Soc. (Mémoires) 4: 31, t. 4b. 1905. TYPE: (lectotype, selected here, P). Magnolia paeneta alauma Dandy, J. Bot. 68, 206: 1930. TYPE: Tsang & Fung 538 in Herb. Lingn. Univ. 18072 (holotype, BM; isotypes, K, IBSC, NY). "e talaumoides Dandy, J. Bot. 68: 208. 1930. PE: Poilane 6370 (holotype, P. Bot. France Bon 3176 Magnolia tenuicarpella Chang, Acta Sci. Nat. Univ. unyatseni (Guangzhou) 1: 54. 1961. TYPE: Mc- Clure 20099 (holotype, SYS). Magnolia odoratissima Law . Z. Zhou, Bull. Bot. R hina), 6, 2: 139. 1986. TvPE: R. Z. Zhou 0054 (holotype, IBSC). Evergreen shrub or small tree to 11 m high; young twigs 3-5 mm diam., yellow-green, brown when dry, smooth, at first colorless to yellowish appressed-pubescent, finally glabrescent, or gla- brous. Stipules adnate to the petiole, stipular scars to halfway to whole length of the petiole, 3-30 mm long. Leaves dark bright green, olive-green, to brown when dry, glabrous or pubescent with scattered hairs that are too minute to be visible above, somewhat pale green, glabrous or pubescent with long, straight, white, yellowish to brownish hairs, especially on the midrib and nerves beneath, thinly coriaceous to coriaceous, elliptic, narrowly obovate-elliptic to obovate, 8-17.5(-30) x 3.5- 7(-10.5) ст; apex acute or acuminate to cuspidate with a 1.5-2.5-cm-long tip, base narrowly to broadly nerves fine, prominent below when dry, in 8-14(-16) pairs, straight, anastomosing and meeting in a looped cuneate; midrib impressed above, intramarginal vein at some distance from the mar- gin, reticulations densely netted, visible on both sides. Petiole pubescent or glabrous, erect to slight- ly recurved, 6-30 mm long. Peduncles glabrous or yellowish pubescent, erect or slightly curved, ) x 0.3-0.5 cm, pedicles 3-4 mm long; bracts 2-3, white or yellowish pubescent outside. Flowers heavily fragrant; tepals 9, the outer 3 thinly coriaceous, greenish, oblong-elliptic to ob- ovate, 3.5-6 x 2-3 cm, the inner tepals 6, white, broadly obovate to obovate-spathulate, basally with a short claw, fleshy, 4-5 x 2-4.5 cm; stamens white, 7—15 mm long, anthers dehiscing introrsely, connective appendage triangular, ca. 1 mm long, filaments 1-1.5 mm long; gynoecium ellipsoid, 1.5- 2 x 0.5-0.8 cm, carpels 15-24, densely yellow pubescent, 8-10 mm or longer. Fruiting peduncles 14-17 x 3.5-5.5 mm, pedicles ca. 4 mm long. Fruits ovoid-elliptic, 4-6 cm long, ripe carpels 1- 3 cm long, apex with a 2-3-mm-long beak; scars of perianth and stamens 5-7 Х 5-7 mm. Seeds narrowly oblong or irregularly ovate, ca. 8-9 mm long. Distribution. China and Vietnam. In CHINA. Southern and southwestern Guangdong: Ding Hu han, vui K.C. & K.L. Shi 1590; ied Ming, Tang L. TZ. уо Huan Sanu Xian, Huan iy eda rin ed. 1459. Guizhou: Du Shan Xian, Li Bo Exped. d Volume 80, Number 4 1993 Chen & Nooteboom 1007 Magnoliaceae of China Hainan: Liang H.Y. 63698, 64334, 64633, 64988, Wang C. 33359, and many other collections. HONG- ONG. Happy Valley: Chang H.T. 86094, 86031, ~ Yunnan: Chen B.L. 86 S 510, Chen B.L. 86 S 538, 86 S 507, 87 F 196, 87 F 222. Ecolog In evergreen broad-leaved forests. Altitude: 400-1,000 m. Flowering May-June; fruiting September- October. Tree cultivated as an ornamental. Magnolia championii is quite variable in habit, shape, and size of the leaves, measurement of the flowers, and number of the carpels. We doubted at first whether the species mentioned above were conspecific. However, after carefully com- paring type specimens and other collections, we came to the conclusion that they are indeed. In fact, the differences are related to the distribution: in Guangdong the species usually is a shrub with smaller flowers and leaves that are somewhat cus- pidate at the apex; in Hainan the species possesses relatively narrow, long leaves with a seemingly in Yunn broader leaves and bigger flowers. Finally, the Viet- namese Magnolia fistulosa and Magnolia talau- moides should be mentioned. The former is similar to Magnolia championii in all characters except its larger leaves and greater number of carpels. The latter differs from Magnolia championii by its glabrous twigs, leaves, and peduncles. There are arguments to support treating Magnolia talau- impressed midrib and nerves; nan it has moides as a variety of Magnolia championii, but for the moment we refrain from that. Magnolia championii was misidentified as Ta- lauma pumila (Andr.) Blume by Champion (1851) and included in Magnolia pumila Andr. by Forbes & Hemsley (1886). It was included in Mag- nolia coco (Lour.) DC. by Rehder & E. Wilson (1913) and Merrill (1923). о Magnolia coco (Lour.) DC., Syst. Nat. 1: 459. 1817. Liriodendron coco Lour., Fl. Cochinch.: 347. 1790. Talauma coco (Lour.) Merr., Sp. Blancoan.: 12. 1918. TYPE: Lourei- ro (not seen). d pumila Andr., Bot. Repos. 4: t. 226. 1802. Tal aillon, Adansonia 7: 4 TYPE: t. 226 (Andr., 1802) Evergreen shrub to small tree 2-4 m high; plant glabrous; bark gray; twigs shining green, brown- yellow when dry, smooth, = angular. Stipules ad- nate to high on the petiole, stipular scars nearly up the entire length of the petiole, 5-15 mm Leaves coriaceous, dark glossy green, usually olive green when dry above, greenish beneath, elliptic, narrowly elliptic, ovate-elliptic, to obovate-elliptic, 10-17 x 2.5-5 cm; apex long-acuminate, some- times acute, base cuneate to broad-cuneate; margin slightly undulate; midrib not prominent on the up- per surface; nerves in 8-10 pairs, curved upward and meeting in an intramarginal vein at some dis- tance from the margin, reticulation coarsely netted, nerves and reticulation yellow to brown and prom- inent on the under surfaces. Petiole 5-15 mm long. Peduncles gray-black when dry, covered with a waxy substance, pendulous, with 3 bract scars, 1.4-2 cm long, pedicles 2-3 x 3-4 mm. Flowers nodding, subglobose, 3-4 cm diam., heavily scent- ed, usually opening in the night; tepals 9, unequal, obovate, dorsal faces convex, the outer 3 greenish, thin coriaceous, 2-2.6 x 1.2-1.7 cm, the inner tepals 6, white, fleshy, oblong to obovate-oblong, 3-4 X 4 cm; stamens white, 5-7 mm long, anthers dehiscing introrsely, filaments 1.5-2 mm long, con- nective appendage rounded to acute; gynoecium obovoid-ellipsoid, ca. cm, carpels 8-12, 5-6 mm long, styles short, torus with the scars of perianth and stamens 4-5 х 3-5 mm. Fruits can be seen occasionally, ca. 3 cm long, ripe carpels subligneous. China and northern Vietnam. In CHI- ukien, Kushan: Chung H.H. 779; 8492. Guangdong: Canton, Gaudichaud Voy. sur la Bon pa Lofoushan, Tsiang Y. 1757; ва о СЕ Shiuchow, То Kang Peng CCC herb. 2856; Ting E^ Shan, Chun W.Y. 6481 A. Guangxi: He Pu, Liang C.F. 33243. Guizhou: Esquirol 3061. Zhejiang: Wenchow, Ching R. C. 1904. HONGKONG: Kit Yock Chan 1352, Chun W.Y. 4520, Hu S.Y. 5393, 10194, 10201. Also in Fujian anad Taiwan. Cultivated in Southeast Asia. Distribution. F cology. In evergreen broad-leaved forest. Altitude: 600-900 m. Flowering April-October. ses. Used for medicinal purposes and to ех- tract volatile oil; also grown as an ornamental. Note. No original Loureiro specimen has been traced. 4. Magnolia gie Franchet, Pl. Delavay: 33, tt. 9- 1889. TYPE: pu mt 223] (holotype, р. Evergreen tree to 18 m high and 25 cm diam., bark black-gray, becoming yellow-gray when dry, rough, fissured; young twigs dull green, densely appressed-woolly with yellowish hairs, provided with obvious orbicular and punctiform lenticels, old ones black-brown and glabrescent. Stipules covered with 1008 Annals of the Missouri Botanical Garden same indumentum as young twigs, adnate to high on the petiole, stipular scars nearly to the end of the petiole, 4—6.5( ceous, green, pubescent, glabrescent above, glau- -8.5) ст long. Leaves coria- cous, densely appressed-woolly with curly, yellow hairs to glabrescent beneath, ovate, narrowly ovate to elliptic, 14.5-26(-30) x 7-15.5(-18) ст; apex acute, occasionally retuse, base acute to rounded, rarely cordate; midrib and nerves conspicuously elevated below, nerves in 11-16(-21) pairs, retic- ulation densely netted, prominent on both sides. Petiole at first densely yellowish pubescent, later glabrescent, 4-6.5(-8.5) cm long. Peduncles stout, pubescent, glabrescent, 1.2-2.5 x 0.7-0.9 cm, pedicle short, 1-5 mm long. Flowers slightly fra- grant, usually fading one day after opening, creamy white, 15-20 cm diam., tepals 9(or 10), the outer 3 thin coriaceous, greenish, oblong, 6-8(-10) x 2.5-3(-4) cm, reflexed later, the inner tepals 6 white, spathulate obovate, fleshy, 7-9.5(-11) x -5 cm, the innermost tepals 3, usually erect and » then enclosing androecium and gynoecium; sta- mens 1.5-2 cm long, anthers dehiscing introrsely, connective appendage triangular, 1.5-4 mm long, filaments 1-2 mm long; gynoecium ovoid, 1.5- 2.2 x 1-1.2 cm, carpels 55-64, yellowish pu- bescent basally. Fruiting peduncles 1-4 x = cm, pedicles 2-5 x 8-10 11 x 3-5 cm, ripe carpels narrowly elliptic, 2.5- mm. Fruits ovoid, 7— 4 cm long in the lower and 1.7-2 cm long in the upper carpels, with a 4—10-mm-long beak apically. Distribution. CHINA. Sichuan: Hsiangyun distr., tG Forres 7. Yunnan: throughout, Chenkang Hsien, Forrest G. 27979, dp 72469; long i kiang, Wang 60; rigo Hsien, .W. 71 Yu T.T. Hsien, Tsai H. 43; Kouang yu ten ee epe аре To nee Hsien, Wang 64; L ag Sino-Amer. Bor Exped. 1630 nee idus other cL dans Southwestern Guizhou Ecology. In broad-leaved forest. Altitude: 1,500-2,800 m. Flowering April-June; fruiting August- October ses. Bark m for medicinal purposes, the tree as an ornamental. 8. Perg yet Dunn, J. Linn. Soc. Bot. 39: 903. ТУРЕ: Henry 127824 (lec- н ud here, A; isolectotypes, E, MO, NY, US Talauma qii Craib, Kew Bull. 1922: 226. 1922. TYPE: Kerr 5 (K). ае а Hu, Bull. Fan. Mem. Inst. Biol. (Pei- . 1937. TYPE: C. W. Wang 76888 (ho- bos PE; isotypes, A, TIE). Evergreen tree to 20 m high; young twigs brown- yellow, 5-7 mm diam., glabrous or sparsely yel- lowish pilose, smooth, + angular. Stipules densely yellowish hairy, adnate to the petiole, scars up to the bases of the blades, 5.5-11 cm long. Leaves dark glossy green, glabrous above, green, glabrous or sparsely appressed-pubescent with straight, yellow-brown hairs beneath, obovate- 26-38(-65) x 8- 13.5(-22) cm; apex rounded or acute, occasionally coriaceous, oblong, oblong to ovate, retuse, base cuneate, sometimes rounded; midrib and nerves prominently elevated below, nerves in 14-20 pairs; reticulation coarse, visible on both surfaces. Petiole glabrous or appressed-hairy, 5.5- 11 cm long. Flower buds ovate; peduncles 2.3-5 x 0.6-1 cm, glabrous, usually recurved, covered with waxy substance, pedicles ca. 1 spathaceous bracts mm long; 2-3, glabrous; tepals 9, sub- similar, the outer 3 pale green outside, white inside, ovate-elliptic, the inner tepals 6, white, fleshy, ob- ovate to spathulate, 5-6 x 2-3 cm; stamens 1.2- 1.5 ст long, anthers introrsely dehiscing, connec- tive appendage triangular, ca. 1 mm long, filaments 1—2 mm long; gynoecium long ovoid, 4-5 x 1.3- 1.5 ст, carpels many; scars of perianth and sta- mens along the torus 1.5-1.7 X 1-1.3 cm. Fruit- ing peduncles 7-9.5 cm long x 7-10 mm. Fruits ovoid to ovoid-cylindric, 7.5-14.5 x 3.5 ст, пре carpels 1.3-2 cm long, apically with beaks of 3- 5 mm long. Seeds irregularly shaped, ca. 5-12 x 7 mm. Distribution. China, Upper Burma, Thailand, Laos. In CHINA. ия Yunnan: Xi Shang Ban Na, Feng K.M. 2 ; Fo Hai, Wang C. W. 76002; Lang- tsang Hsien, с Lung-huk, Jenn-Yeh Hsien, 80116; Meng Hai, Li Y.H. 3607; a La, Tsai H.T. 59-10917; Nan “Chia, Wang C.W. 76888; Szemoa, Henry A 12782 Ecology. In к жеш broad-leaved forest. Altitude: 540-1,5 Collector's note. Flowers white or ivory, very fragrant. 6. Magnolia phanerophlebia B. L. Chen, Acta Sci. Nat. Univ. Sunyatseni (Guangzhou) 1: 107. 1988. TYPE: B. L. Chen & C. N. Mai 87T001 (holotype, SYS). Evergreen shrub or treelet ca. 3 m high and 6 cm diam., plant glabrous. Young twigs 5-7 mm diam., gray to yellow-gray when dry, terete, smooth. Stipules adnate to the base of the petiole, stipular scars 7-19(-25) mm long. Leaves coriaceous, dark green and shiny above, greenish beneath, obovate to elliptic, 31-34.5(-36) x 9.5-14(-17) cm; apex Volume 80, Number 4 1993 Chen & Nooteboom Magnoliaceae of China 1009 acuminate, acumen 1-2 cm, base cuneate, decur- rent with two ridges into the petiole; midrib con- spicuously elevated beneath; nerves in (11-)13- 17(-19) pairs, much impressed above, anastomos- ing and meeting in an intramarginal vein that is more prominent than the coarsely netted reticu- lation on both surfaces. Petiole obviously thickened toward its base, (1-)3- curved, ca. 2.5 X 0.4 ст; spathaceous bracts 3, one toward the base of peduncle and one at the base of perianth, the third at the middle of peduncle or higher. Tepals 8-9, subequal, the outer 3 green- ish, thin coriaceous, the inner tepals 5-6, white, fleshy. Stamens numerous. Gynoecium narrowly cm. Peduncle re- ovoid, carpels ca. 11. Fruits not seen. Distribution. CHINA. Southeastern Yunnan: Maguan, Chen B.L. 87 Т 1, 87 T 3 Ecology. In iii E broad-leaved forest. Altitude zd: 700 n Notes. The Hac is drawn from limited material with only flower buds. This species is quite similar to Magnolia candollii var. obovata, from which it differs by its shorter peduncles and fewer carpels. Because so far the mature flowers and fruits have not been gathered, we were unable to compare it with Magnolia candollii var. obovata; thus Magnolia phanerophlebia is temporarily maintained as a species here in section Gwillimia. la-2. Magnolia section Rytidospermum Spach, Hist. Nat. Veget. Phanerog. 7: 474. 1839. ТУРЕ: Magnolia tripetala L. Deciduous trees. Stipules adnate to petiole. Leaves large, crowded at the end of the branchlets in false whorls; midrib not prominent above. Flow- ers bisexual; tepals subequal; connective produced into a short or long appendage; anthers dehiscing introrsely. Gynoecium not stipitate. Number of ovules per carpel 2, sometimes 4 in the lower carpels. Scars of bracts l. Fruits at least finally consisting of free follicles, which dehisce along the dorsal suture. KEY TO THE SPECIES OF и SECTION RYTIDOSPER MU 1. Twigs glabrous, d to purple-brown when dry l. Twigs densely appressed-hairy, soon glabres- cent, yellow to yellowish gray, 8-14 mm diam.; filaments red, stout, 4 mm long. __________ 8. Magnolia officinalis 2. Twigs 6-8 mm diam., purple-brown, conspicu- ously supplied with semicircular to Anda leaf scars; filaments crimson, ca. 12 mm long. .... T: Magnolia hypoleuca 2. Twigs 7-14 mm diam., yellow to purple-brown when dry, smooth, covered with a gray waxy substance; filaments 4-5 mm long. _______ 9. Magnolia rostrata aJ Magnolia hypoleuca Siebold & Zucc., Abh. Math.-Phys. Cl. Königl. Bayer. Akad. Wiss. 4(2): 187. 1845. TYPE: L-908-126-1201 (L). Magnolia hypoleuca [var.] 8 e Siebold & Zucc., Abh. Math.-Phys. Cl nigl. Bayer. Akad. Wiss. 4, 2: 187. 1845. TYPE: Гл 908. 126-1103 (L). Deciduous tree to 30 m high; bark gray, lon gitudinally fissured; young twigs 6-8 mm diam., purple-brown, glabrous, conspicuously supplied with semicircular to elliptic leaf scars; terminal winter buds large, ovoid-cylindric, ca. 4 cm long, glabrous and glaucous. Stipules glabrous, adnate to the base of the petiole, scars 1.5-2.5 cm long. Leaves crowded at the end of the twigs, thinly chartaceous, green, glabrous, sometimes pubescent along the midrib above, glaucous, pubescent with stout, long, crisped, colorless hairs everywhere as well as with slender, straight, colorless hairs along the midrib and nerves beneath, obovate, 20-38(-45) x 12- 18(-20) cm; apex apiculate or short-acuminate, base broadly cuneate to obtuse; midrib and nerves prominently elevated below, nerves in 15-20 pairs, reticulation coarsely netted, obscured on both sides. Petiole glabrous to sparsely pubescent, lenticellate, 3-4.5 cm long. Flowers appearing after the leaves, cup-shaped, fragrant, 14-20 cm diam., peduncles glabrous, ca. 2.5 X 1 cm, pedicles ca. 5 x 8 mm; tepals 9-12, the outer З broadly oblong, greenish, pale reddish tinged, ca. 8 X 4 cm, the inner tepals 7-9, spathulate, creamy white, ca. 11 x 4.7 ст; stamens whitish, ca. 2.6 cm long, anthers introrsely dehiscing, connective appendage triangular, ca. | mm long, filaments crimson, ca. 12 mm long; gy- noecium greenish, cylindric, ca. 3.5 X 1 cm, torus with the scars of perianth and stamens ca. 8 x mm. Fruiting peduncles ca. 3.5 X 1 cm, ped- icles 1.7 cm long. Fruits ovoid-cylindric, 12-13.5 X 4.5-6 cm; ripe carpels 1.5-2.5 cm long, api- cally with recurved beaks 5-10 mm long; scars of perianth and stamens along the torus ca. 2 x 1.2 cm. Distribution. Native of Japan, cultivated in northern hina Uses. Wood fine-textured, light and soft, gen- erally used for construction work, furniture, mu- sical instruments, and boards; ornamental because of its big and magnificent flow- tree grown as an ers. 1010 Annals of the Missouri Botanical Garden Note. Ueda (1985, 1986) discussed the syn- onymy. For a long time the name Magnolia obova- ta Thunb., nom. illeg., was used, e.g., by Rehder & E. Wilson in Sargent (1913), Johnstone (1955), H. Hara (1977), Ohwi (1978), and Kitamura & Murata (1981) for this species. The species was named Magnolia hoonoki, nom. nudum, by Sie- bold (1830) and Millais (1927). More information on misapplied names used for this species follows here: Magnolia gima auct. non L.: Thunb. [var.] 8 Thunb., Fl. Jap.: 236. 1784b, only for the Japanese name Fo no Ki and the descr. iene tele obovata [Thunb., inii Linn. Soc. London 2: 3 94a only for e Japanese name Fo no Ki and the descr. (Thun- ae 12885, UPS) nom. illeg.; Mus. Nat. Acad. Upsal. 16: 137, sine descr. 1794b; Pl. Jap. Nov. Sp.: 8, 1824; sine descr.]. E. Wils., Gard. Chron. 6 .1 у jig eN Pg non Sweet: Yanagida, J. Soc 6, 3: 268, t. 413. 1934. Magnolia d Thu a var. yanagidana Hatusima, ke Phytotax. Geobot. 4: 209. 1935. TYPE not know 8. Magnolia officinalis Rehder & E. H. Wilson in Sarg., Pl. Wilson 1: 391. 1913. TYPE: E. H. Wilson 652 (holotype, A; isotypes, E, GH, US). шыг» о var. ens Rehder & E. H. Wilso ns 392. 1913. Magnolia hiloba сее Sinicorum 1: a officinalis e biloba (Rehder & Е. Н. Vias Cheng & Law in W. C. Cheng, Sylva Sinica : 449, t. 134 У 1983. TYPE: E. Н. Wilson 1649 olotype, A; isotypes, E, US). Magnolia officinalis var. FR ens C. Y. Deng, J. Ad jing Inst. Forest. 1: . 1986. TYPE: C. Y. Der 84182 (not seen). Deciduous tree to 20 m high, bark thick, gray, smooth; young twigs stout, 8-14 mm diam., yel- lowish to gray, sometimes brown, densely ap- pressed-hairy, with short, straight to undulate, col- orless to brown hairs, soon glabrescent, the ol ones brownish gray; terminal buds large, narrowly ovoid, brown-hairy, 3.5-5.5 x 1-1.2 cm. Stipules membraneous to chartaceous, sparsely hairy with long hairs, scattered pu punctate, adnate to the petiole, scars 0.5-2.7 cm long. Leaves large, thinly coriaceous, 7-8 dd together at the top of the twigs, green, glabrous above, pale green, glaucous, scattered-pubescent with short, straight to slightly curly, colorless hairs as well as rather long, crisped, colorless hairs everywhere, later gla- brescent beneath, obovate, 16.5-37.5(-46) x 9- 20(-24) cm; apex acute, obtuse, emarginate to obcordate, base cuneate; both midrib and nerves prominently elevated, covered with slender, straight to undulate, colorless hairs below, nerves in 14- (40) pairs, reticulation coarsely netted, visible on both sides. Petiole stout, densely brown or gray 1.9-5 pearing with or after the leaves, white, sometimes flushed with purplish red, fragrant, 10—15 cm diam.; peduncles 3-5 x 0.5- hairy, glabrescent, cm long. Flowers ap- .5-0.8 cm, densely pubescent with long, straight, crisped, gray hairs, pedicles 2- 6 mm long; tepals 9-12(-17), thick fleshy, the outer 3 pale green outside, obovate, usually re- flexed, the inner tepals 6-9(— 14), obovate to spath- ulate, 8-1 2.5-5 cm; stamens 3-4 cm long, anthers introrsely dehiscing, connective appendage triangular, ca. 1 mm long, filaments red, stout, 4— 1 ст; torus with the scars of perianth and stamens l- 8 mm long; gynoecium ovoid, 2.5-. 1.8 x 1.2 ст. Fruiting peduncles glabrous to 0.5-0.8 cm. Fruits ob- long-ovoid, occasionally cylindric, straight to slight- densely gray-hairy 3-5 х ly distorted by abortion of part of the carpels, 8- х 4.5-5 ст; ripe carpels 1-3 cm long, with 1-3-mm-long beaks at the apex; the scars of peri- anth and stamens along the torus ca. 1.4-2 x 1.5 cm. Seeds ovoid to irregularly shaped, + pressed, 8-10 x com- CHINA. ago iiss Anhui: Chang Gon an, Yieh P.C. 345, Ching R.C. 3227; _ Distribution. Wu Yu an: Ching R.C. 2519. Northern Guang- dong: Ruyuan, Chen B.L. et al. 81001. Northern Guangxi: Lin к Liang H.Y. 100218; Long Sheng, Huang D.A. 6034; Hsi-Chang onc Ch'i-fen-shan, estern Hanan: Yana enshih Hsien, Henry , H.C. 1367; Patune Hsien, Chow H. C. 652, Henry А 1475, 549, 3711. Jiangxi: Kuling, Lotus Valley, Chiao C.Y. 18759; Ku-ling, Chung H.H. & ~ Sun 680; Sa-tiu-hong, Yungshiu, Tsiang Y. 10616; Kin- kiang, Wilson E.H. 1649. Sichuan: Kuan Hsien, Wang F.T. 20441; Mt. Omei, Yu ТТ. 517, Chiao C.Y. & C.S. dis 436; Wan Hsien, Mou-tao-chi, y ang: Moupin, Werner E. s.n. Zhejiang: 50 li N of Sia chu, Ching R.C. 1610; Lungt- suan, Ho Y.Y. 3201; Mokanshan, Meyer F.N. 1572; Swen chi, southern Yentang, Hu ‚ 93; Tien tai shan, Huating, Chiao C.Y. field no 1050, herb. no. 14349. The species is also reported from southern Shaanxi. In forest. Altitude: 300-1,300 m. Flowering May-June; fruiting August- October. Uses. Bark, used for medicinal purposes; seeds also for extrac- Ecology. root-bark, flowers and seeds all tion of oil; wood yellowish brown, straight-grained and fine-textured, soft, used for construction, ve- Volume 80, Number 4 1993 Chen & Nooteboom Magnoliaceae of China 1011 neer, furniture, and musical instruments. Because of its dense crown, large leaves, and beautiful flow- ers, the tree is grown as an ornamental. Leaves whitish beneath; bark smooth, gray; flowers white or purple; fruits erect, Collector's notes. green, turning purplish or purple-red; seeds scarlet. Magnolia officinalis resembles Mag- nolia hypoleuca in many characters except in the shape of the basal follicles of the ripe fruits, the degree of the indument of the pedicles, and the color of the twigs. In this regard, Spongberg (1976) has already compared the two species in detail; we agree with his observations. However, there are not sufficient collections available, particularly of living material, to settle the problem. Therefore, the two are treated as distinct species here. This species was included as Magnolia hypo- leuca auct. [non Siebold & Zucc. (1845)] by Diels (1900), Finet & Gagnepain (1905, pro parte), E. Wilson (1906) excluding description of the fruit., and Fang Wen Pei (1942). 9. Magnolia rostrata W. W. Smith., Notes Roy. Bot. Gard. Edinburgh 7: 213. 1920. TYPE: Forrest 15052 (lectotype, selected here, E; isolectotype, K) = Deciduous tree to 24 m high; bark pale gray; twigs stout, 7-14 mm diam., yellow to purple- brown when dry, glabrous and smooth, covered with a gray waxy substance. Stipules glabrous, dull yellow-green, usually covered with waxy substance, adnate to the base of the petiole, stipular scars 1.5-3.5 em long. Leaves thinly coriaceous, green, glossy and glabrous above, glaucous, sparsely scat- tered-pubescent with rather long, crisped, brown hairs, glabrescent beneath, 5-7 crowded together at the end of the twigs, obovate, 31-36.5(-53) х 19-20.5(-28) cm; apex broadly omnia. with a ca. 3-mm-long short-acute tip, base obtuse; midrib conspicuously elevated beneath, nerves prominent beneath, in 28-36 pairs, reticulation coarsely net- ted, visible on both sides. Petiole stout, pubescent, glabrescent, 2.5-6 cm long. Peduncles ca. 2.5 х 1 ст, pedicles ca. 1.3 cm long. Flowers appearing after the leaves, heavily scented; tepals 10-11, the outer 3 green, slightly flushed with pink outside, pink inside, oblong-elliptic, 8-13 x 2.5-5.6 cm, reflexed, the inner tepals 6, white, erect, spathulate to obovate, 11—13 x red, 1.4-1.7 em long, anthers dehiscing introrsely, 4-4.5 cm; stamens purple- connective appendage triangular, 1-1.5 mm long, filaments 4-5 mm long; gynoecium cylindric, gla- brous, 2.3-4 x 1-1.3 ст, carpels many; torus with the scars of perianth and stamens 1.2-1.9 x 0.6-1 cm. Fruiting peduncles 1.5-3.5 x 1-1.7 cm, pedicles 2- both glabrous. Fruits cylindric, erect, 12-20.5 x 3-5 cm, apex gradually narrowing, base broadly mm, sometimes absent, 0 rounded; ripe carpels 1–1.5 cm long in the lower and 0.7-1.2 ст long in upper carpels, beak to 6– 8 mm long, incurved. Seeds irregularly shaped, flat, ca. 7 X 5 mm. унија Northeastern Burma and China. In CHI- Ke u n: Mo 18394; N Waikha-Salwin divide, 26?30'N, Forrest G. 18246; Salween-Kui Chiang divide, 27°N, 98°35'Е, For- rest G. 25751; Taron-Taru divide, Ahtemai, Yu T.T. 20894 Ecology. In broad-leaved forest. Altitude: 2.400-2,800 m. Flowering April-May; fruiting September- October. Leaves very large; flowers thick and fleshy, white, ivory, to white flushed rose or rose-pink, Collector's note. sweet-smelling, tepals 10-12, appearing before the foliage. Note. The description of the flowers in Smith (1920) does not apply to this species. Ia-3. Magnolia section Oyama Nakai, Fl. Sylv. Koreana 20: 117. 1933. Magnolia sect. Co- phantera Dandy, Curtis Bot. Mag. 159: sub. t. 9467. 1936. ТУРЕ: Magnolia sieboldii K. Koch. Deciduous shrub or tree. Stipules adnate to pet- iole. Leaves + pale green or glaucous, evenly dis- tributed; midrib not prominent above. Flowers bi- sexual; tepals subequal; connective not produced into an appendage, retuse or blunt; anthers de- hiscing introrsely. Gynoecium not stipitate. Scars of bracts 1. Fruits at least finally consisting of free follicles, which dehisce along the dorsal suture. KEY TO THE SPECIES OF MAGNOLIA SECTION OYAMA 1. | Leaves not glaucous beneath; pedic les 13-17 mm; nerves in 9-12 pairs; gynoecium 15-20 mm high. |... 12. Magnolia sine nois 1. Leaves glaucous beneath. ED. 2(1). Gynoecium 23-25 mm high; nerves in 8-11 pairs; pedicles 5-25 mm. 10. Magnolia globosa 2. | Gynoecium 13-20 mm high. = 3 3(2). Reticulation rather obscure; nerves 6-8 pairs; pedicle 2-2.8 mm. 11. Magnolia Home 3. Reticulation distinct; nerves in 10-12 pedicle 5-7 mm. „. 13. Magnolia wilsonii 1012 Annals of the Missouri Botanical Garden 10. Magnolia globosa Hook. f. & Thomson, FI. Ind. 1: 77. 1855. TYPE: Sikkim, alt. 2,700– 3,000 m, J. D. Hook. s.n. (holotype, K; iso- type, GH). Magnolia tsarongensis W. W. Smith & Forrest, T Roy. Bot. Gard. Edinburgh 12: 215. 1920. TYPE: G. Forrest 18870 (lectotype, selected here, A; isolec- totypes, E, K). Deciduous small tree to 10 m high, flowers ap- pearing at the same time or before the leaves; young twigs purple-brown or dull brown, 4-5 mm diam., at first densely appressed or shaggily hairy with yellowish, brown to rusty-brown hairs, soon glabrous; terminal winter buds densely brown pu- bescent. Stipules yellowish sericeous or brown vil- lous, adnate to higher on the petiole, stipular scars usually over half the length of the petiole, 15-42 mm long. Leaves membranous, dark green, usually dull brown when dry, sparsely pubescent along the midrib and nerves above, glaucous, villous with long, undulate to crisped, colorless or brownish hairs, later glabrescent beneath, ovate, broadly ovate, о. occasionally obovate, 12. 5-21(-26) 1 cm; apex acute to rounded, base rounded to somewhat cordate or truncate; midrib and nerves prominent beneath, nerves in 8-11 reticulation coarse, visible below. Petiole 2-5 cm long. pairs, densely pubescent, glabrescent, Flowers fragrant, cup-shaped to subglobose, pen- dent or nodding, 6–7.5 cm diam., peduncles dense- ly villous with gray, brown, or rusty-brown hairs, 3.7—-6.5(- 7.5) x 0.2-0.4 cm, pedicles 5-25 mm long; tepals 9-12, creamy white, subequal, obovate to spathulate, 4-7.5 x base obtuse, usually constricted into short claws; stamens crimson, 12-18 mm long, anthers de- hiscing introrsely, connective retuse or rounded, 2-4 cm, apex rounded, filaments 1-4 mm long; gynoecium green, ellipsoid, 2.3-2.5 x 0.5-0.8 cm. Fruiting peduncles 6.5- 9 x 0.3-0.5 cm, glabrous or hairy. Fruits cylin- dric, 2.5-9 x 2-2.5 cm; ripe carpels 1.7-3 cm long in the lower and 1-2 cm long in the upper carpels, apically with beaks 3-6 mm long; scars of perianth and stamens along the torus 6-12 x 7-11 mm. Seeds ovoid, ca. x 8 mm. Northeastern Assam, eastern Himala- МА. Southwestern South- ng: Mountains of Champutong, Rock J.F. 22047; Putsang River, Rong To valley, Zayul, Kingdon Ward F. 10973 Kui Chiang divide, 23°24'N, 98?28'E, Forrest G. 20866; no Kiu Chiang divide, Si chi to, 28°45'N, 98?18'E, 22; Tsarong, Salwin-Kui Chiang divide 28%24'N, deci E, 20303; Meng-hua, Chukai, Yu Т.Т. 15854; Distribution. . Northwestern Yunnan: Salween- Salween-Djiou-Djiang, under Tschamputong, Hand.-Mazz. 9212; Salwin-Kiukiang divide, Yu ТТ. 19251; Sikang, Li M.K. 2329; Sitiping, between Lipiang and Weihsi, Ching R.C. 22016; Wei si Prep. TM C.W. 63981 Xue long Shan, Feng K.M. 4 Ecology. In mixed forest or thickets. Altitude: 1,900-3,300 m. Flowering May-July; fruiting Au- gust-September. Uses. Seeds and leaves used for volatile oil, tree cultivated as an ornamental. Collector's notes. With brown hairs on bracts, petioles, and under leaves; flowers snow white, creamy white, or greenish white, pendent and fra- grant; fruits greenish, fleshy; seeds red. Note. Yulania japonica Spach var. globosa (Hook. f. & Thomson) P. Parmentier (1895) is a nomen nudum. 11. Magnolia sieboldii K. Koch, Hort. Den- drol.: 4. 1853. Magnolia oyama Kort, Revue Horticulture Belge 31: 258. 1905. Magnolia verecunda Koidz., Bot. Mag. (Tokyo) 40: 339. 1926. SYNTYPES: K. По s.n. L 908-126-1110, fertile (L), sine leg. L-908- 126-1043 (L). Deciduous treelet to 10 m high; young twigs slender, 1-2 mm diam., yellow-brown, dark brown, or purple-brown, at first densely appressed-pubes- cent with yellowish white hairs, later glabrous, smooth. Stipules sparsely to densely hairy with gray-yellow or brownish hairs, adnate to the petiole, scars up to over or under halfway on the petiole, 5-40 mm long. Leaves chartaceous, dark bright green, glabrous to sparsely pubescent with fine hairs above, glaucous, sparsely to densely covered with appressed short or long, colorless or brownish, usually straight to rarely undulate hairs beneath; obovate or broadly obovate, 3-15(-20) x 2-9(-12) cm; apex acute to acuminate with acumen 5-10 mm long, base rounded to truncate; midrib and nerves prominent below, nerves in 6-8 pairs, re- ticulation coarse, obscured. Petiole appressed-hairy, 1-4(-6.5) ст long. Flowers together with the leaves, cup-shaped to saucer-shaped, nodding or pendent, fragrant, 7-10 cm diam.; peduncles slen- der, densely hairy with gray-white or brownish hairs, 3.8-4.5 x 0.1-0.2 cm, pedicles 2-2.8 mm long, sometimes absent; tepals 9-12, white, sub- equal, 3.5-6 x 5 cm, the outer 3 obovate, reflexed in full bloom, the inner tepals 6, spathulate to broadly obovate, usually with short claw at the bases; stamens purple-red, 8-15 mm long, anthers dehiscing introrsely, connective blunt, ca. 0.5 mm long, filaments ca. 3 mm long; gynoecium green, ellipsoid, 1.5-1.7 x 0.4–0.6 cm; carpels narrowly Volume 80, Number 4 1993 Chen & Nooteboom 1013 Magnoliaceae of China elliptic, ca. 1 ст long. Fruiting peduncles glabrous to densely brownish pubescent, 6.2-7 x 0.2-0.3 cm. Fruits cylindric, rarely ovoid, 3-4 x 1.2-1.7 cm; ripe carpels apically with 4—12-mm-long beaks, erect or recurved; torus with scars of perianth and stamens 4-10 x 3-5 mm. Seeds subcordate to irregularly shaped, ca. 6-7 х 6 mm China, Korea, and Japan. In CHINA. Wu, Huangshan, Deng & Yao 79192. Northern "Gu uangxi: Quan Xian, Chun Y.F. 81654; Xing An, Chen 2.2. 51165; Zi Yuan, Tsoong C.H. 83515. n, Cao Zi-yu 369; Lei Shan Distribution. i: San Hua ed yos C.S. 4065. Sichuan: Ma-pien 816; со Hsien, Wang ЕТ. 21080. Pri iM Wang C.W. 89858. Ecology. In forest. Altitude 1,500-1,800 m. Flowering May-June; fruiting August-September. Uses. Used for medicinal purposes; grown as an ornamental. Collector's notes. Small tree or shrub; flowers nodding, white and fragrant; stamens creamy white, with rose tip Siebold & Zuccarini (1845) treated this species as Magnolia parviflora [non Blume (1825)], while "quit (1904) used the name Magnolia conspicua [non Salisbury (1806). otes. 12. Magnolia sinensis (Rehder & E. H. Wilson) Stapf, Bot. Mag. (Curtis) 149: t. 9004. 1924. Magnolia globosa Hook. f. & Thomson var. sinensis Rehder & E. H. Wilson in Sarg., PI. Wilson. 1: 393. 1913. Magnolia sieboldii . Koch subsp. sinensis (Rehder & E. Wilson} о ee Arbor. 57: 279. f. 3, h-i. 1976. : E. H. ns 1422 (holotype, A; basin E, GH, Deciduous shrub or small tree to 5 m high; young twigs purple-brown, dark purple, purple-gray to gray, slender, ca. 3 mm diam., densely yellow or brownish villous at first, glabrescent later, old twigs brown-gray, smooth, glabrous to barely hairy. Stip- ules densely appressed-pubescent with brownish or yellowish hairs, adnate to the petiole, scars 2-3 cm long. Leaves chartaceous, green, finely pubes- cent over the surface or only along midrib and nerves above, pale green, at first densely covered with rather long, crisped, colorless or brownish hairs, soon glabrescent beneath, elliptic, broadly elliptic, obovate to broadly obovate, 10—16(—21.5) x 6-9(-13) cm; apex rounded or mucronate, base obtuse; nerves visible below, in 9-12 pairs, retic- ulation coarse. Petiole densely yellowish villous, 2.5-4(-7) cm long. Flowers appearing with the leaves, nodding, fragrant, cup-shaped, 8-12(-15) cm diam. when fully open; peduncles densely or 0.2-0.5 cm, ped- icles 1.3-1.7 ст long; tepals 9, white, subequal, the outer 3 ovate to elliptic, 4-5.5 x 1.7-2.7 cm, the inner tepals 6, broadly obovate, with short claw at the bases, 4.5-7 X 14 mm long, anthers dehiscing introrsely, connec- sparsely pubescent, 3.5-6.5 х 3-5.6 cm; stamens red, 9— tive rounded to mucronulate, ca. m long, filaments 1—2 mm long; gynoecium green, narrow- ly obovoid-ellipsoid, 1.5-2 x 0.4-0.7 cm, carpels narrow and long, styles 3-4 mm long. Fruiting peduncles 4–6.5 x 5.5-7.5 x 1.3-1.7 cm; ripe carpels ellipsoid, 1.7— 2.5 ст long in the lower and 1-1.3 0.2-0.5 cm. Fruits cylindric, cm in the upper carpels, beaks 2-5 mm long, recurved; torus with scars of perianth and stamens 8-10 x 4-9 mm. CHINA: northwestern Sichuan. Distribution. Uses. The tree is used as an ornamaental. Note. There is still disagreement about the rank of this species. Ueda's (1980) observation and judgment dealing with this species seem ac- curate. We agree with his view that Magnolia sinensis should not be treated as a subspecies of Magnolia sieboldii. 13. Magnolia wilsonii (Finet & Gagnep.) Reh- der in Sarg. Pl. Wilson. 1: 395: 1913. Mag- nolia parviflora var. wilsonii Finet & Gag- nep., Bull. Soc. Bot. France (Mémoires) 4: 39. 1905. TYPE: E. H. Wilson 3137 (holotype, A; isotypes, BM, K, US). Magnolia nicholsoniana Rehder & E. H. Wilson in Sarg. Pl. Wilson. 1: 394. 1913. За icd wilsonii f. nicholsoniana (Rehder & E. H. Wilson) Rehder, J. Arnold Arbor. 20: 91. 1939. TYPE: E. Н. Wilson 838 (holotype, A; isotypes, BM, K, US j ). Magnolia еар я W. W. Smith, Notes Roy. Bot. Gard Edinburgh 8: 341. 1915. Magnolia lilifera var Pi (W. W ith) Pamp., Bull. Soc. Tosc Ortic. 41: 137. 1916. Magnolia veri ep t (W. W. Smith) Rehder, Man Trees: 249. 1940. ТУРЕ: С. Forrest 7182 ee. E; i e BM, K). Deciduous shrub or small tree to 8 m high; bark gray-brown; young twigs slender, 2-3 mm diam., brown-purple to gray, densely villous with yellowish to brown hairs, later glabrescent, smooth and sparsely lenticellate, old twigs gray-black, glabrous, glabrescent. Stipules densely yellow to yellowish white villous with silky hairs, adnate to the petiole, stipular scars nearly up to the base of the blade, 1-2.5(-5.4) ст long. Leaves chartaceous, dark 1014 Annals of the Missouri Botanical Garden green, glabrous or sparsely pubescent over the surfaces to only hairy along both midrib and nerves above; glaucous, glabrous, sometimes only pubes- cent along midrib and nerves, or thinly to densely pubescent with short to long, colorless to brownish, straight, slightly curved to rather crisped hairs everywhere beneath; usually narrowly ovate to ob- ovate, sometimes broadly ovate to elliptic, 9— 14.5(-22) x 3.5-6(-10) ст; apex acute or acu- minate, base rounded or sometimes slightly cor- date; midrib and nerves visible on both sides, nerves in 10-12 pairs, reticulation coarse. Petiole slender, densely covered with same indumentum as young twigs, soon glabrescent, 2—4(—5.6) cm long. Flow- ers appearing with the leaves, at first cup-shaped to saucer-shaped, later pendent, white, fragrant; peduncles slender, brown-villous with silky hairs, glabrescent, 13- -3 mm, pedicles 5-7 mm long; bracts pubescent; tepals 9(—12), subsimilar, the outer 3 elliptic, apically rounded to acute, 4.5- 6 X 1.5-3 cm, the inner tepals 6, broadly obovate to spathulate, apically rounded, basally short- clawed, 2.7-6.5 x 1.3-4.5 cm; stamens 9-12 mm long, purple-red, connective rounded, fila- ments ca. 2 mm long; gynoecium green, ovoid- cylindric, 13-20 x 5-6 mm; torus with the scars of perianth and stamens ca. 6 х 6 mm. Fruiting peduncles glabrous, 30-45 x 2-3 mm, pedicles mm long. Fruits pendent, cylindric, 3-6.5 X 1-2 cm; ripe carpels ellipsoid, 1.7-2.6 cm long in the lower and 1-2 cm long in the upper carpels, with beaks 2-5 mm long. Seeds subcordate to irregularly shaped, 5-7 x 5-6 mm. Distribution. CHINA. Guizhou: Xing Yi Xian, An Shun Exped. 1112. Western & central Sichuan: Nai- tan to pain yi pu, Smith, Harry 1953; Tachinhu, Wilson 1374; Wa shan, 8 838. Yunnan: Tze-tchou-pa, Maire ЕЕ. 866; Tali range, 25°40'N, 100°5'Е, Fo rrest G. 27941; Yun Hsien, McLarens's collectors L 66 4: Liao Tseng, C 92; Kou ty, Pe yen tsin, Ten Simeon 564; Chienchuan-Mekong divide, Forrest G. 23480; N end of Cangshan, Sino-British Exped. 655. In forest. Altitude: 1,900-3,300 m. Flowering May-June, fruiting September-Octo- er. Ecology. Uses. Bark used for medicinal purposes; the tree is grown as an ornamental. Collector's notes. Flowers saucer-shaped, drooping, fragrant, pure white; stamens bright red E. Wilson (1906) dealt with this species erroneously under Magnolia globosa Hook. f. & 'Thomson. la-4. Magnolia section Gynopodium Dandy, Curtis Bot. Mag. 155: t. 16. 1948. TYPE SPE- CIES: Magnolia nitida W. W. Smith. edi Hu & Cheng, Acta Phytotax. Sinica 1, 2: 1951. TYPE SPECIES: Parakmeria omeiensis Hu : hen МИА" in Keng, Quart. J. Taiwan Mus. 8: 207, . 345. 1955. TYPE SPECIES: ак kachi- Pdo: (Kaneh. & Yamam.) H. Ken Evergreen tree, entirely glabrous. a free. Leaves evenly distributed. Midrib prominent above, at least toward base, or not. Flowers androdioe- cious. Tepals subequal. Connective produced into a short or long appendage. oi dehiscing in- trorsely. Gynoecium stipita umber o per carpel 2, sometimes 1 in the lower carpels (rarely 4 in each carpel). Fruits at least finally consisting of free follicles, which dehisce along the ovules dorsal suture. KEY TO THE SPECIES OF MAGNOLIA SECTION GYNOPODIUM 1. Outer tepals 3.1-5 cm long. Outer tepals 2.5-3 cm lon . Midrib of leaf prominent above, at least toward base; leaves ovate, base broadly cuneate or rounded; filaments 1-3 mm, connective ap- pendage of anthers 2-4 mm. 15a Magnolia nitida var. nitida 2; Midrib of leaf not prominent above; leaves elliptic, obovate, or narrowly elliptic, base cu- neate or narrowly cuneate; filaments 2-2.5 or 7 mm; connective appendage of anthers 1-1.5 mm. : gies 6 mm thick; leaves narrowly gs io base narrowly cuneate; flower creamy w outer viia thin, oblong; flaments 7 mm 16. Magnolia omeiensis Peduncle 3-4 mm thick; leaves elliptic o obovate, base cuneate; flower yellowish green, tepals subsimilar, obovate or spathulate; men ; В Magnolia kac hirachirai Peduncle 3-4 mm hide midrib of leaf not h — =~: eN = N ~ 4(1). cium exserted from stamens; fruits 2-2.5 cm long. 14. | agnolia kachirachirai m EN as, 2. Em 5 o ® rà 8; gyno i tirely hidden within н аы ЫЗ fruits 3.5- 7.5 ст long. . 5 . Gynoecium 8 mm high; reticulation of leaf distinct; connective appendage of anthers 0.5 mm long; gynoecium narrowly ovoid; scars of perianth and stamens along torus under fruit 6-7 mm long = = ос. Magnolia nitida var. robusta 5. Gynoecium 16 mm high; reticulation rather obscure, connective appendage 2-4 mm long; Volume 80, Number 4 1993 Chen & Nooteboom Magnoliaceae of China 1015 gynoecium ovoid, scars we and sta- mens along torus under fruit 2-4 mm long. 15b. Manda өлен уаг. cb ~ . Magnolia kachirachirai (Kanehira & Ya- mamoto) Dandy, Kew Bull. 1927: 264. 1927. Michelia kachirachirai Kanehira & Yama- moto, Trans. Nat. Hist. Soc. Taiwan 84: 78. 1926. Micheliopsis kachirachirai Keng, Quart. J. Taiwan Mus. 8: 210, pl. 1 and f. 1. 1955. Parakmeria kachirachirai Law in W. C. Cheng, Sylva Sinica 1: 473. 1983. TYPE: Kanehira, 1924 (holotype, TAIF not seen). Evergreen glabrous tree to 17 m high and 1.2 m diam., bark dark brown, smooth, young twigs 2-3 mm diam., dull dark brown, smooth, old ones gray, longitudinally fissured; terminal buds ovoid to elliptic, dull brown, 13-16 free from the petiole. Leaves thinly coriaceous, dark green above, green beneath, elliptic to obo- vate, 6.5-12(-20) x 2-3(-5) cm; apex acute to short-acuminate, base cuneate; nerves fine, ob- х 3-4 mm. Stipules scured above, visible below, in 9-12 pairs, retic- ulation rather coarse, faint on both sides. Petiole ca. 5-20 mm long. Peduncles 7-15 x 3-4 mm, pedicles usually absent, rarely present, ca. 1 mm long. The bisexual flowers yellowish green; tepals 9-12, subsimilar, obovate to d 2.5-4 x 1–1.5 cm; stamens 50- -1.8 cm long, an- thers dehiscing db. connective appendage narrowly triangular, ca. 1 mm long, filaments 2- 2.5 mm long; gynoecium ovoid, slightly exserted from the androecium, 1.2-1.4 х 0.5 cm long, gynophore ca. 5-8 x 1.5 mm, carpels 10-18; torus with the scars of perianth and stamens ca. 5 x 4 mm; the male flowers not seen. Fruits subovoid, 2-2.5 x 1.2-1.5 cm; ripe carpels 8- 2, + connate, 2-2.5 cm long. Seeds reniform to subcordate, ca. 10 х 7 mm. bud du Ж CHINA. Southeastern Taiwan: ngchun peninsula, “y A 2060, Chang C.E. 4384; Taizhong, Xie Z.Q. s Ecology. In evergreen broad-leaved forest. Altitude: *500- 1,300 m. ses. Wood used for general construction work; tree also grown as an ornamenta 15. Magnolia nitida W. W. Smith, Notes Roy. Cheng, Sylva Sinica 1: 472. 1983. TYPE: For- rest 15059 (lectotype, selected here, E; iso- lectotype, K). жые (p en Chun & C. T. Tsoong, Acta Phy- . Sin . 1963. TYPE: Chun & Tao 50122 (IBSC no Magnolia yunnanensis (Hu) Noot., Blumea 31, 1: 1985. iar ib gh сы Hu, Acta Phyto. : 195 YPE: C. W. Wang 83157 tax. Sin. (KUN) 15a. Magnolia nitida var. nitida Evergreen tree to 30 m high and 90 cm diam., entirely glabrous, bark faintly gray-brown or dark gray; young twigs 2-3 mm diam., olive green to brownish, fruiting twigs 5-6(-8) mm diam., old ones dull gray, longitudinally to irregularly fissured; terminal buds ovoid, ca. 17 X 5 mm. Stipules free from the petiole. Leaves rigid, coriaceous, dark glossy green, olive green to dull brown when dry above, green, greenish brownish to brownish when dry beneath, usually ovate to elliptic-ovate, occa- sionally obovate, 6-12(-14) x 3-4(-6.5) cm; apex short-acuminate, acumen 7-10 mm long, some- times acute, base broadly cuneate to obtuse; midrib prominent above only near the base; nerves fine, visible on the underside, in 9-13 pairs, reticulation coarse, hardly visible on both surfaces. Petiole 1.5- 2.5 cm long, without stipular scars. The bisexual flowers fragrant, peduncles 6-8 x 5 mm, pedicles 1.5-2 mm long; tepals 9-12, subsimilar, creamy yellow, slightly fleshy, flushed purplish outside, ob- ovate to spathulate, 4-5 x 1-3 cm; apex rounded to short-acuminate, base tapering, sometimes short- clawed; stamens 1.3-2.5 cm long, anthers dehisc- ing introrsely, connective appendage linear, 2-4 mm long, filaments 1-3 mm long; gynoecium ob- long, exserted from the androecium, ca. 16 x 5 mm, gynophore 6-8 x styles crimson; the male flowers not seen. downs peduncles 8-16 x 4—5 mm, pedicles 1.5-4 long. Fruits ovoid to oblong, sometimes id 3.5-7.5 x 2-3 cm; ripe carpels 13-20, cm long, apically with recurved beaks ca. 1 mm 2-3 mm; carpels green, long, gynophores 4-1 —6 mm; scars of peri- anth and stamens along the torus 5-10 x 4-7 mm. Seeds reniform, 10-13 x 5-10 mm. China ld and northern Burma. In CHI- Wo | valley, Champutong, 10235; Tsarong, 6. Forrest s Yunnan: Salwin-Kiu Chan divide, Champutong, 983 28?16'N, Forrest G. 21616; Salwin-Kiu Chang iur 28°24'N, 98°24'Е, EA Shweli-Salwin divide, 25°50'N, 98°45'Е, 26381; y к сш Н.Т. 5022; Shang pa Hsien, Tsai H.T. 5 1016 Annals of the Missouri Botanical Garden Ecology. In mixed forest, growing on sand- stone, shale, and granite. Altitude: 800-2,500 m. Flowering March-May; fruiting September—Oc- tober. Uses. Wood yellowish, durable, generally used for construction work and furniture. Collector's notes. Widely | hed ел tree, or half deciduous; flowers fragrant, creamy white, pale yellow or white washed purple toward base; fruits green. Notes. leaves whitish р. This species is widely distributed, ex- tending from northern Burma northward to Xizang and northern Yunnan, and northeastward to cen- tral and southeastern Yunnan. Magnolia yunna- nensis is very similar to Magnolia nitida in all characters. It is obvious that the leaves, the mature fruits, the peduncles, and the gynophores become increasingly longer along a cline from Burma to southeastern Yunnan. No characters can be found that segregate Magnolia nitida from Magnolia yunnanensis. 15b. Magnolia nitida var. lotungensis (Chun & C. T. Tsoong) B. L. Chen & Nooteboom, stat. nov. Magnolia lotungensis Chun & C. T. Tsoong, Acta Phytotax. Sin. 8: 285. 1963. TYPE: Chun & Tao 50122 (IBSC not seen). Evergreen tree to 20 m high and 90 cm diam., bark pale gray to brownish gray; young twigs 2- З mm diam., dull brown-green, fruiting twigs 4— 5(-6) mm diam., old ones faint gray; terminal buds ovoid-ellipsoid, ca. 14 х 4 mm. Leaves coriaceous, dark green, shiny, olive green to dull brown when dry above, green, brownish when dry beneath, rig- id, elliptic to narrowly elliptic, occasionally broadly elliptic, 6-11 cm; apex acute to short-acuminate, base cuneate; midrib prominent on both sides, nerves fine, hardly visible above, prominent below, in 9-13 pairs, reticulation coarsely netted, obscured on both surfaces. Petiole ca. 1.5-2 cm long, without stipular scars. Flowers androdioecious, white; tepals 9-12, subsimilar, the outer 3 ovate, 2.5-3 x 1.2-2 cm, the inner tepals 6-9, obovate, slightly narrower than the outer; stamens 35-05, anthers 9-10 mm long, filaments са. 1 mm long in the male flowers; stamens 17- 22, gynoecium ovoid, entirely hidden by the an- droecium, carpels 15-20 or sexual flowers. Fruiting peduncles 4-9 x 4-7 mm, much less in the bi- gps absent. Fruits obovoid to iun "bad .9 X 2.5-3 em; gynophore 1-5 пре carpels 10-13, rarely 1-4, + connate, 1.8- 2.2 cm long; scars of perianth and stamens along -6 mm; the torus 2-4 x 5-6 mm. Seeds ellipsoid to ovoid, 7-12 x 6-7 mm. Distribution. CHINA. Guangdong: Ruyuan, Ко S.P. 53691, Guangdong Mucai Yanjiu Zu 32; ec Liu J.H. 89847. Guangxi: Da Miao Shan, .H. 2873; Jia Xian, Wang C. 40158; Jin Xiu Xian, Da Yao Shan Exped. 13429; Da Ming Shan, Cai C. X. 5470; Lin Gui, Deng X.F. 231; Long Sheng, Long id Exped. 130. Guizhou: Li Ping Xian, Yuan Hainan: Dong Fang Xian, Chen Y.W. 7620; Fan Yah, Chun N.K. & Бг Tso 44041; iii Lau S.K. 27375. Hunan: Mang Shan, Li B.G. al. 59; Xin wer Liu L.Y. 15146. 7һейап: Tai Suan, Ching R.C. 2 In forest. Altitude: 800-1,100 m. Flowering April; fruiting September- October. Ecology. /ses. ood used for general construction work and furniture. Note. This variety is so similar to Magnolia nitida that we do not distinguish them easily. In general, variety lotungensis has elliptic to narrowly elliptic leaves, fewer carpels, and the gynoecium is hidden by the androecium. Unfortunately, suf- ficient flowers of Magnolia nitida var. lotungensis were not available for comparison to be made. 15c. Magnolia nitida var. robusta B. L. Chen & Nooteboom, var. nov. TYPE: China. Annam, Massif de Bi-doup, 17 Oct. 1940, alt. 2,000 m, Poilane 31039 (holotype, P; isotype, L). Figure var. nitida differt ramulis crassioribus sub fructibus fols grandioribus petiolis longioribus. Tree 25-30 m high, ca. 1 m diam., fruiting ones stout, glabrous; young twigs ca. 3 mm diam., ca. 8 mm diam., dull olive-brown, old ones quite rough, longitudinally and transversely irregularly fissured; terminal buds ovoid to ovoid-ellipsoid, ca. 15 x 5 mm. Stipules free from the petiole. Leaves slightly glossy above, greenish beneath, ovate-elliptic to ovate, 7-16 coriaceous, rigid, green, х 3.5-0.5 cm; apex acute to acuminate or obtuse, base cuneate; midrib slightly prominent above, con- spicuously elevated below, nerves visible on both sides, in 12-15 pairs, reticulation coarse, promi nent on both sides. Petiole + flat, dilatate de the base, 1.6-4.7 cm long. In the male flower, peduncle 10-13 x 6-7 mm, pedicle absent; tepals not seen; scars of perianth and stamens ca. 10 x 6 mm. In the bisexual flower, peduncle ca. 5.2 x 2.2-3 cm long; tepals 9, the outer 3 much larger than the inner ones, the former ovate, ca. 2.5 X 1.5 ст, the latter 6, 2.2-2.5 х 1.2 cm, spathulate, ob- ovate to narrowly obovate, tapering toward the 9 mm, pedicle ca. 2 mm long; bracts 2, Volume 80, Number 4 1993 Chen & Nooteboom 1017 Magnoliaceae of China FIGURE 1. bisexual flower. —3. Term male flower. Based on Poilane 31039 (P). Drawing by Joop Wessendorp, Ваа, Leiden inal bud. — 4. Out base, 2-2.5 x 0.5-1.2 cm; stamens 12-15 mm long, connective appendage ca. 0.5 mm long, fil- aments ca. 1 mm long; gynoecium narrowly ovoid, ca. 8 mm long, hidden by the androecium, gyno- phore 3-4 mm long; carpels 11-13, styles ca. 5 mm long. Fruiting peduncle 7-10 7-9 mm; fruit ovoid to ovoid-oblong, 5-6.5 x 3.5-4 cm; Wessendorp Magnolia nitida W. W. Smith var. robusta B. L. кык & Noot.— 1. dopa branch. — 2. Deflorated er tepals. —5. Inner tepals. —6. Stamen.— 7. > el.—8. Deflorated gynophore 2-8 x 6.5-8 mm; ripe carpels 11- 12, the dorsal face rhombic to elliptic, 1.7-2.7 cm long, 4-5 mm thick, apex short-beaked; scars of perianth and stamens 6-7 x 8-10 mm Distribution. China and Vietnam. In CHINA. Gu- angxi: in Shap Man Taai Shan, 1-16 Oct. 1934, W. T. 1018 Annals of the Missouri Botanical Garden Tsang 24463. VIETNAM. Annam: massif de Bi-doup, 23 Nov. 1914, alt. 2,000 m, Poilane 31039 (TYPE); Binh Tri Thién, 7 Oct. 1949, alt. 1,500 m, J. E. Vidal 10344. Collector's note. In rocky thicket, flower yellow, fragrant. Note. This variety is characterized by stout twigs, especially the fruiting twigs, larger leaves, woody; longer petioles, whose length is usually less than 4 times that of the leaves, short peduncles, a few carpels, a thick wall of mature carpels, and the gynoecium hidden by the androecium instead of exceeding the androecium as in Magnolia nitida var. nitida. The specimen from Guangxi bearing smaller leaves, without any flowers or fruits, is here treated as this variety based on the shape of the leaves and the relative length of the blade and the petiole. 16. Magnolia omeiensis (Hu & Cheng) Dandy in Praglowski, World Pollen & Spore Fl. 3: 5. 1974. Parakmeria omeiensis Hu & Cheng, Acta Phytotax. Sin. 1(1): 2. 1951. TYPE: W. C. Cheng 10525 (holotype, PE; isotype, A). Tree to 20 m high and 40 cm diam., bark dark gray; plant entirely glabrous; twigs 2-2.5 mm diam., smooth, dull brown-black when dry. Stipules free from the petiole. Leaves coriaceous, dark green, shiny above, pale green beneath, narrowly elliptic, 7.5-12 x times acute, base narrowly cuneate; nerves fine, 2-4.5 cm; apex short-acuminate, some- in 8-10 pairs, reticulation coarse, hardly visible on both sides. Petiole 1—2 cm long, without stipular scars. Peduncle ca. 10 х 6 mm, pedicle absent. Male flowers creamy white, tepals 2 the outer 3 thin, oblong, apex obtuse, ca. 3. -3 cm, x inner tepals 6, spathulate to Шошо, 3.5-4.: -1.5 em; stamens crimson, 2-2.2 cm long, an- thers dehiscing introrsely, connective appendage 7 í triangular, ca. 1.5 mm long, filaments ca. 7 mm long; torus ovate-orbicular, 4 mm long. Distribution. CHINA. а Emei Shan, Cheng W.C. 10525, Chow H.C. 1209. ogy. In evergreen broad-leaved forest. Altitude: 1,000-1,200 m Note. Magnolia nitida, and Magnolia omeiensis, which all belong to Magnolia sect. Gynopodium, can barely be distinguished by their vegetative characters; the knowledge of their flowers is still far from satis- Magnolia "m hirachirai, factory. Unfortunately, we were unable to see enough material, especially collections with flowers. The delimitation of these species is obviously ob- scure. Until more material is available they are maintained as different species here. Ia-5. Magnolia section Aleimandra (Dandy) Nooteboom, Blumea 31: 88. 1985. Alciman- dra Dandy, Kew Bull. 1927: 260. 1927. TYPE SPECIES: Magnolia cathcartii (Hook. f. & Thomson) Noot. Evergreen tree. Stipules free. Leaves evenly dis- tributed; midrib not prominent above. Flowers bi- sexual; tepals subequal; connective produced into a short linguiform appendage; stamens with short filaments and long anthers, completely hiding the gynoecium; anthers dehiscing introrsely. Gynoe- cium stipitate. Number of ovules per carpel 2-5. Fruits at least finally consisting of free carpels, which dehisce along the dorsal suture. 17. Magnolia cathcartii ms f. & Thomson) Nooteboom, Blumea 1985. ен cathcartii Hook. f. A as Fl. Ind. 1: 79. 1855. Sampacca cathcartii (Hook. f. : Thomson) Kuntze, Revis. Gen. Pl. 1: 6. 1891. Alcimandra cathcartii (Hook. f. & Thomson) Dandy, Kew Bull. 1927: 260. 1927. TYPE: Sikkim, J. D. Hook. s.n. (holotype, K; iso- types, A, GH, NY) Tree to 25 m high and 50 cm diam.; young twigs dark brown, slender, 1-2 mm diam., densely appressed-pubescent with long, straight to + un- dulate or curly, yellowish to gray hairs, old ones dull brown-gray, glabrescent, sparsely supplied with conspicuously elevated, elliptic to suborbiculate, white lenticels; terminal buds narrowly ovoid, cov- ered with an indumentum as young twigs, 0.6-2.5 cm long. Stipules hairy, free from the petiole. Leaves coriaceous, dark green, shiny above, both sides glabrous, ovate to elliptic-ovate, occasionally ob- ovate, 6.5-17 x acumen 0.6-2 cm long, sometimes acute, base 3-5.6 cm; apex long-acuminate, obtuse to broadly cuneate; midrib densely pubes- cent, glabrescent at both sides, nerves fine, prom- inently elevated on both surfaces, in 12-15 pairs, reticulation densely netted, visible on both sides when dry. Petiole pubescent, glabrescent above, 0.2-2 cm long, without T scars. Peduncles usually slender, rarely stout, ca. 15-2 glabrous, pedicles 0-4 mm “ong tepals 9, white, the outer 3 oblong, 5.5-6 .9 cm, the inner tepals 6, obovate-elliptic, ca. E X 2.5 cm; sta- mens 3- mm, cm long, anthers introrsely dehiscent, connective appendage tonguelike, ca. 2 mm long, filaments ca. 1 cm long; gynoecium stipitate, cy- lindric, ca. 2 cm long. Fruiting peduncles 15-35 X 2-4 mm. Fruits usually irregularly shaped, 3.5- 6.5 х 1.5-2 ст; ripe carpels 3-16, compressed- subglobose, ca. 8-9 mm diam., white-lenticellate; Volume 80, Number 4 1993 Chen & Nooteboom 1019 Magnoliaceae of China gynophore 5-8 x 2-4 mm; torus with the scars of perianth and stamens 4-7 X 4-5 mm. Seeds subellipsoid, ca. 8 X 7 mm Distribution. Sikkim, Assam, China, Upper Burma to Vietnam. In CHINA. Southern & southeastern Xizang: Yunnan, Jingdong Xian, Xu S.G. 5024; Kiu- kiang Valley, Muchietu, Yu Т.Т. 21016; Lan-Tsang Hsien, Wang C.W. 76855; Meng-soong, Dahmeng-lung, Che-li Hsien, bs a Tsai H.T. 58698; Wenshan, Feng K.M. 2 Ecology. In forest. Altitude: 1,800-2,800 m. Flowering May; fruiting August-October. Collector's notes. In mixed woods; flowers yel- low-green; fruits green, blue to brown-green; seeds black with orange-red fleshy coat. Ib. Magnolia subgenus Yulania (Spach) Rei- chenbach, Der Deutscher Botaniker 1: 192 1841. Yulania Spach, Hist. Natur. Worst, Phanerog. 7: 462. 1839. Magnolia subg. Pleurochasma Dandy, J. Roy. Hort. Soc. 75: 161. 1950. TYPE SPECIES: Yulania conspicua Spach = Magnolia heptapeta (Buchoz) Dan- dy. Leaves evenly distributed, deciduous; midrib not prominent above. Flowers bisexual, precocious and/ or with a much reduced calyxlike outer whorl of tepals; anthers dehiscing laterally or sublaterally, connective produced into a short or long append- e. Gynoecium not stipitate. Fruit cylindric or oblong, usually + distorted, at least finally con- sisting of free carpels which dehisce along the dorsal suture. Ib-1. Magnolia section Yulania (Spach) Dandy in Camellias and Magnolias Conf. Report: 72. 1950 Flowers appearing before the leaves (or some late flowers with the leaves), pink, (greenish) white, purple, or red. Tepals subequal. KEY TO THE > OF MAGNOLIA SECTION YULAN l: naaa 8-15 mm high. 2 1. Супоесішт 20-4 3 . Twigs yellowish brain to black or gray-brown; leaves not ucous beneath, reticulation coarse; a= purple or red; Liep cle 12-15 x 5-6 m 2. nolia ud 2. Twigs yellowish green turning ns n; eaves glaucous beneath, reticulation "fine; flower EDA d cle p. "a x 3-4 mm. 0. Magnolia dawsoniana 4 N ~ = — Flower purple or m Flower pink or white. .... 5 За). 3. 4(3). Tepals 7-8 cm long; stamens 15-17 mm long; leaf base broadly cuneate or rounded, а short-acuminate; Pus scars 3-7 m 24. Magnolia zenil 4. Tepals 5-6.5 cm long; stamens 9-10 m long; leaf base cuneate, apex acuminate or acute; stipular scars 1-3 er = Ma „gnolia а amoena duse 25-30 mm long; pidas peduncles 7-5 cm lon | agnolia um 9 Peduncle аа 20 mm кш fruiting pedunc 1-2 ст Twigs ш е tepals 12; terminal winter buds densely villous; reticulation of leaves distinct; 5(3). g сл 6. Twigs finely appressed pubescent ig: o; terminal winter buds sericeous; reticulation of leaves rather obscure; flowers s pen be- fore the leaves; filaments 3-4 mm 2 MM heptapeta 18. Magnolia amoena W. C. Cheng, Contr. Biol. Lab. Sci. Soc. China Bot. Series 9: 280, f. 28. 1933. TYPE: S. Chen 2692, in flower (lectotype, selected here, PE; isolectotype, A). Deciduous tree to 12 m high; bark gray or gray- white; young twigs 2-3 mm diam., slender, aa hit with brown, glabrous; terminal buds long hairs. Stipules adnate to the petiole, stipular scars 1-3 mm long. Leaves thinly coriaceous, gla- brous above, curly hairy with long white hairs along the midrib, the nerves and at the junctions of the midrib and the nerves beneath, obovate to obovate- elliptic, 10-15 x 3.5-5 cm; apex acute to cus- pidate with a ca. 15-mm-long tip; base cuneate, somewhat unequal; nerves in 10—13 pairs, nerves and reticulations prominent on both sides; petiole pubescent, 8-1 m long. Flowers appearing be- fore the leaves, fragrant, ca. 6 cm diam. Peduncles 6-8 х 4 mm, pedicles 3-4 х 4 mm, both densely yellowish to white hairy. Tepals 9, subsimilar, red to reddish, spathulate to oblanceolate, 5-6.5 cm long; stamens ca. mm, connective appendage short-pointed, ca. 1 mm long, anthers dehiscing laterally, filaments purple, ca. 3 mm long; gynoe- cium cylindric, са. 2 cm long, styles ca. 1 mm long. Fruits cylindric, 4-6 cm long, often curved as carpels partly abortive; fruiting peduncles long white pubescent, ca. 10 х 4 mm; ripe carpels oblong, apex rounded, papilliferous. Distribution. CHINA Yin S. 3 Rege и mushan, Cheng W.C. 4 Ecology. In forests. Altitude: 700–1,000 m. Flowering April-May; fruiting September—Octo- ber. А. So athers Anhui: Huoshan, : Zhejiang, W Tien- 1020 Annals of the Missouri Botanical Garden Note. type specimen only. There are no other collections The description was drawn from the available. 19. Magnolia campbellii Hook. f. & Thomson, Fl. Ind. 1: 77. 1855. Type: Sikkim, J. D. Hook. & Thomson s.n. (holotype, K; isotypes, GH, NY). Magnolia mollicomata W. W. Smith, Notes. POR Bot G inburgh 12: 211. 1920. TYPE : Forrest 14466 (lectotype, selected here, E). Deciduous tree to 30 m high; bark gray-brown; young twigs usually purple-brown, sometimes brown to gray-green when dry, 4-7 mm diam., glabrous or pubescent, later glabrescent; terminal winter buds ovoid, yellowish sericeous. Stipules sparsely or densely pubescent with short or long yellowish hairs, adnate to the very base of the petiole, stipular ~7(-11) mm long. Leaves thinly charta- ceous, dark green, glabrous above, pale green, glabrous to scattered-pubescent with rather long, straight, undulate, crisped and colorless hairs ev- scars erywhere or along the midrib and nerves beneath, elliptic, (broadly) ovate to obovate, 10-23(-33) x 4.5-10(-14) em; apex rounded, acute, or short- acuminate, base rounded or broadly cuneate, usu- ally oblique; nerves prominent beneath, in 7-13 pairs. Petiole glabrous or yellowish pubescent, 2- 5 em long. Peduncles glabrous to densely woolly with yellowish hairs, 2.5-3 x 0.7-1 cm, pedicles 1.5-2.5 x 0.3-1 cm. Flowers large, 15-25 ст diam., precocious, slightly fragrant; tepals 12-16, white, or pale rose at the base outside to pink outside, spathulate, oblong-ovate to abruptly constricted into a short claw near the base, the outer 3 concave, outspread to reflexed, 6-14 -obovate, X 4-8 cm, the innermost tepals 3, erect, conni- vent, enclosing the androecium and gynoecium, 8- 10 -6 cm; stamens 2-3 cm long, connective appendage 0.5-1 mm long, anthers laterally de- hiscent, filaments purple, minutely pubescent, 4— 9 mm long; gynoecium green, 2.5—4 cm long; scar of perianth and stamens on torus ca. 1.5-2 cm long. Fruiting peduncles 2.7-5 x 1-1.2 cm, ped- icles 1.5-2.5 x 0.6-1.3 cm. Fruits erect, soon pendent, cylindric, 8.5-17 cm long, ripe carpels packed close to each other; scars of perianth and stamens along the torus 1-2.2 х 1.3-1.: Seeds 7-9 mm long Distribution. Northern India, Nepal, Bhutan, China, a. In CHINA. Sichuan: g Wes ern Yunnan (Nujiang-Langchua anjian divide of Weixi Xian Nujiang divide of Ruili Xian), Salwin- Kui Chiang б; к С. 20841; МеКопр- Salwin divide, 28?15'N, Duanqing, above Yangbi, Sino-British Exped. 166. Ecology. In mixed forests. Altitude: 2,500- 3,500 m. Flowering March- Мау; fruiting Septem- ber-October. Uses. Cultivated as an ornamental plant be- cause of its spectacular flowers. Collector's notes. Deciduous shrub or tree; leaves and bud-scales glabrous; flowers pure white, exterior with purple marking at the base, creamy white flushed purplish, creamy yellow, bright rose- pink, pink, precocious, open cup-shaped, fragrant, fleshy, filaments purple, anthers brown. We agree with Dandy that there is no clear separation between Magnolia campbellii and Magnolia mollicomata (see Dandy, 1 species is misidentified as Magnolia rostrata by W. W. Smith (1920) Magnolia grandiflora (non. L., 1759) is the name used by Griffith (1847) for this species. Note. 20. Magnolia dawsoniana Rehder & E. H. Wilson. in Sarg., Pl. Wilson 1: 397. 1913. TYPE: E. H. Wilson 1241 (holotype, A; types, BM. K, US). Deciduous tree to 20 m high and 50 ст diam.; twigs yellowish green, turning reddish brown later, glabrous to sparsely pubescent at 1-3 nodes or internodes below the peduncles, lenticellate. Stip- ules appressed-pubescent, adnate to very base of the petiole, stipular scars 3-4 mm long. Leaves coriaceous, dark green, glossy, glabrous above, glaucous, usually pubescent with slightly curly, col- orless hairs at the junctions between the midrib and nerves and along the veins beneath, obovate, sometimes broadly obovate-elliptic, 7.5-14(-17) X 4.5-8 cm; apex rounded to acute, rarely emar- ginate, base cuneate to subrounded, usually oblique; midrib impressed above, prominent and becoming brown when dry beneath, nerves conspicuously elevated beneath, in 8-10 pairs, reticulation coarsely netted, prominent, on both sides. Petiole glabrous to minutely pubescent, slender, usually reddish, 1.5-3 cm long. Peduncles glabrous or + pubescent, 1.5-2 x 0 3-4 mm; bracts 2, sparsely yellowish pubescent. Flowers precocious, fragrant, usually horizontal, nodding later; tepals 9-12, white, tinged with pale red outside, subequal, oblong-spathulate to ob- ovate-oblong, 7-11 x 2-5 cm; stamens purple- red, 1-1.8 cm long, anthers laterally or sublaterally cm, pedicles 7-12 dehiscent, connective appendage triangular, са. 1 Volume 80, Number 4 1993 Chen & Nooteboom 1021 Magnoliaceae of China mm long, filaments 2.5-3 mm long, gynoecium ca. 1.5 cm; scar of perianth and stamens on torus ca. 8 x 6 mm. Fruiting peduncles ca. 2.2 x | cm, pedicles 0.7-1 x 0.7-0.8 cm. Fruits cylindric, slightly contorted as carpels partly abortive, ca. 10. cm; ripe carpels obovate, sparsely len- ticellate, apically short-beaked. Seeds oblong, ca. 1 cm long. Distribution. CHINA. Central and southern Si- chuan: Lu Ding Xian, Qiao H.R. 1010; Mt. Tachien lu, Wilson E.H. 1241. Northern Yunnan. Ecology. In broad-leaved forest. Altitude: 1,600-2,500 m. Flowering April-May; fruiting September. U. ses. Bark used for medicinal purposes. 2]. Magnolia heptapeta (Buc'hoz) Dandy, J. Bot. 72: . 1934. Lassonia heptapeta Buc'hoz, Pl. Nouv. Découv. 21, t. 19, f. 1. 1779. TYPE: f. 1 of t. 19 (Buc'hoz, 1779). Magnolia denudata Desr. in Lam. Encycl. 3: 675. 1792. Magnolia obovata Thunb., Trans. Linn. Soc. Lon- don 2: 336. 1794. Magnolia obovata [var.] a de- nudata (Desr.) DC., Syst. Nat. 1: 457, comb. illeg. 1817. TYPE: t. 43, Moran 1 (Banks, 1791, in BM). d conspicua Salisb., Parad. Lond. 1: 38, t ulania о (Salisb. Por Hist. ы vé ég. 7: 464. 1839. TYPE: Salisb. t. Magnolia yulan Desf., Hist Arbr. a p 6. 1809. pep е (Desf.) Kostel., Allg. Med.-Pharm. 1. 836. Gwillimia 2 (Desf.) C. de А | en Heest. ed. 2: 116. 1887. TYPE: not know Deciduous tree to 20 m high and 60 cm diam.; bark dark gray, rough and fissured; young twigs 3-7 mm diam., purple-brown, finely appressed- pubescent, later glabrescent, smooth, sparsely white- lenticellate; terminal winter buds ovoid, densely hairy with long, silky hairs. Stipules sparsely or densely yellowish pubescent, adnate to the base of the petiole, scars 1-4 mm long. Leaves thinly chartaceous, dark green, glabrous to pubescent with straight or slightly undulate hairs along the midrib and nerves or short-scattered-pubescent on the surfaces above, pale green, sparsely to densely pubescent with short or rather long, straight or curly, colorless hairs everywhere or only along the midrib and nerves beneath, obovate to broadly obovate, 8-17.5(-21.5) x 5-10.5(-16) cm; apex rounded to truncate, abruptly short-acuminate, acumen 1-15 mm long, base cuneate or rounded; midrib and nerves visible on both sides, nerves in 7—10 pairs, reticulation coarse, obscured. Petiole appressed-pubescent, glabrescent, 1.3-2.5 cm long. Flowers appearing before the leaves, erect, fra- x 5- 7 mm, densely appressed-villous with yellowish grant, 10-12 cm diam., peduncles 12-20 hairs, pedicles 4-6 mm long; tepals 9, white, some- times tinged with purple outside at the base, sub- x 2-6 ст; stamens purplish, 10-13 mm long, anthers de- similar, obovate to spathulate, 6-12 hiscing sublaterally, connective appendage narrow- ly triangular, ca. 1 mm long, filaments 3-4 mm ong; gynoecium cylindric, glabrous, 2-2.5 cm long. Fruiting peduncles 12-20 x 6-9 mm. Fruits cy- lindric, straight or distorted by the abortive carpels, .5-5 ст; ripe carpels 1-2 cm long, usually white-lenticellate, papilliferous outside, hardly beaked; scars of perianth and stamens along the torus 7-12 x 6-10 mm. Seeds dod ovoid to broadly ovoid, ca. 9 x 1 CHINA. Anhui: Chiu Hwa Shan, Fan 261. Fujian: Hinghwa, Lin Pi 6488. Guangdong: Heungshan, Chun W.Y. 69; T. "d mont J. 53089. Guangxi: Quangxian, Chen Y.L. ed. 89. Hanan: Yang shan, Changning Hsien, Fan C.S. & Y. Y. Li 191. Jiangxi: Kuling, Chiao C.Y. 18702; Lu shan, Steward A.N. 4730. Jiangsu: Sukow, Haichow, Hers J 622. Zhejiang: Kwatou, 40 li N of Siachu, Ching R.C. 1602; Mo Kan Shan, Cheo & Wilson 12746; Tien ir m Law Y.W. 1385; Tsing yuan distr., Keng Y.L. Distribution. 25 Y.Y. Li The straight-grained and fine-textured wood is used for furniture and boards; flower buds Uses. for medicinal purposes and the tion of volatile oil; the tree as an ornamental. Collector's notes. Bark brownish gray peeling off in thick Deis flowers white tinged purple at base; fruits gre Notes. Ueda (1986) gave an account of the complicated synonymy. Formerly, this species was known under the name Magnolia denudata Desr., but Dandy (1934) changed the name to the older synonym Magnolia heptapeta, based on Lassonia heptapeta Buc'hoz. Meyer & McClintock (1987) argued that the names Magnolia heptapeta and Magnolia quinquepeta should be rejected because their types, artistic impressions of the speciei are “entirely without botanical credibility.” Ueda (1986), but also Treseder (1978), in his standard work on cultivated Magnolias, accepted both names. The nomenclature specialists I consulted agreed that the names keptapeta and quinquepeta should be used, “otherwise many other names should be rejected too" (H. P. Nooteboom). Magnolia precia Correa ex Vent., nomen nudum (Ventenat, 1806), belongs here. 1022 Annals of the Missouri Botanical Garden 22. Magnolia sargentiana Rehder & E. H. Wilson in Sarg., Pl. Wilson. 1: 398. 1913. TYPE: E. Н. Wilson 914 (holotype, A; isotypes, BM, E, GH, K, US). Magnolia conspicua var. emarginata Finet & Gagnep., Bull. Soc. Bot. France (Mémoires) 4: 38. 1905. Magnolia denudata var. emarginata Pamp., Bull. Soc. Tosc. Ortic. 20: 200. 1915. Magnolia emar- ginata (Finet & Gagnep.) Cheng, J. Bot. Soc. China 1, 3: 298. 1934. TYPE: David s.n. (holotype, А Magnolia sargentiana var. i ta + һдег ilson in Sarg., Pl. Wilson 399. 1913. TYPE: . H. Wilson 923a (holotype, P isotypes, K, US). Deciduous tree to 20 m high and 25 cm diam.; young twigs 4-7 mm diam., yellow-brown, black- brown to gray-brown, glabrous; terminal winter buds ovoid, densely yellowish villous with silky, long hairs, ca. 2.7 cm long. Stipules sparsely to densely appressed-pubescent, adnate to lower base of the long, undulate to crisped, colorless hairs, sometimes seemingly glabrous when covered with minutely scattered hairs below; usually inequilateral, ob- ovate, rarely broadly obovate, sometimes elliptic, 10.5-17.5(-22.5) x 5-10(-12) cm; apex acute to rounded, sometimes emarginate or truncate, base usually oblique, cuneate to broadly cuneate; midrib and nerves conspicuously elevated beneath, nerves in 8-12 pairs, reticulation coarsely netted, visible 1 2(-4.7) cm long. Flowers precocious, slightly fragrant, horizontal or nodding, 15-20 cm diam. when fully open; pe- duncle 20-23 x 6-7 mm, pedicle 12-15 x 5- 6 mm (but absent in H.T. Tsai 51831), glabrous to densely yellowish villous; tepals 10—12, subsimi- above. Petiole glabrous, 1.5-3. lar, purplish red outside and white to pale rose inside, spathulate to obovate, with narrow base, X 2-3 cm; stamens purplish, 14-17 mm long, anthers dehiscing sublaterally, connective ap- pendage triangular, ca. 1 mm long, filaments 3- 4 mm long; gynoecium green, cylindric, 8-12 x 3 mm; carpels many. Fruiting peduncles glabrous to hairy, ca. 2-3.5 cm, pedicles 1.2-2.2 x 0.6-1 cm. Fruits cylindric, 3-20.5 x 1.5—2.5 cm, erect to curved; ripe carpels subglobose, 7- 22 mm long, apex short-beaked; torus with scars of perianth and stamens 1.2-1.5 x 1- Seeds subcordate, 8-12 x 7-10 mm. 1.2 cm. Distribution. CHINA. Sichuan: Wa shan, Wilson E.H. 914; Tian Pa ao 34828; near Xizang border, Pratt А.Е. ; Mt. Omei, Chow H.C. 9777; O-pien hsien, Liu Ys г 91; Paohsing hsien, Chu K.L. 08. Yunnan: between Tsian and Ta Kuan, Tsai Н.Т. 50843; Wen-shan Hsien, 51831 Ecology. In broad-leaved forest. Altitude: 1,600-2,500 m. Flowering April-May; fruiting September. Uses. Bark for medicinal purposes. Magnolia sargentiana var. robusta dif- fers from Magnolia sargentiana var. sargentiana in its larger leaves and bigger flowers and fruits. Except for this difference in size, we could not find any characters to distinguish them. The variety, therefore, is here abolished. 23. Magnolia sprengeri Pampanini, Nuovo А 95. 1915. TYPE: Sil- vestri 4104 (holotype, FI not seen). Magnolia denudata Desr. var. purpurascens n & E. H. Wilson in Sarg., Pl. Wilson 1: 401. 1913. Magnolia conspicua var. purpurascens sensu Bean, Kew Bull. 1920: 119. 1920. TYPE: Wilson 373 (lectotype, selected here, A; isolectotypes, E, K, US). iens diva Stapf ex Dandy in Millais, Magnolias 120. 1927. Magnolia sprengeri var. diva S Mag. (Curtis) 153: t. 9116. 1 Wilson 21 (lectotype, selected here, K; isolecto- types, A, K, NY). Magnolia denudata Desr. var. elongata Rehder & E. Wilson in Sarg., Pl. Wilson 1: 402 sub t. 9116. 1913. Magnolia sprengeri var. elongata (Rehder & E. H. Wilson) Johnstone, Asiatic Magnolias in Cultivation: 87. 1955. Type: E. H. Wilso (holotype, А; isotypes, E, GH, K, US). n 345 Deciduous tree to 20 m high and 18 cm diam.; bark black-gray, falling in small flakes when old; young twigs 3-5 mm diam., yellow to yellow-brown, becoming purple-brown later, glabrous and sparsely lenticellate, old twigs gray-black, fissured; terminal winter buds broadly ovoid, 2.8-3 x 1.8-2.4 cm, densely yellowish villous with straight and silky hairs. Stipules densely pubescent with appressed yellowish white hairs, glabrescent, adnate to the base of the petiole, scars 1-2(-13) mm long. Leaves coriaceous, dark green, glabrous to sparsely pu- bescent along the midrib and nerves above; pale green, glabrous to appressed-pubescent with fine, short, straight and colorless hairs over the entire surface or covered with long hairs along the midrib only, glabrescent beneath; obovate to narrowly ob- ovate, 7-14.5(-22) x 2.5-7.5(-15) cm; apex rounded, sometimes truncate and short-acuminate with 5-15-mm-long acumen, base cuneate to nar- rowly cuneate; midrib and nerves conspicuously elevated below, the latter in 6-8 pairs, reticulation coarse, visible on both sides. Petiole slender, gla- brous or villous near the base of the blades, 1-3.5 cm long. Peduncles yellowish sericeous with straight hairs, glabrescent, stout, 10-17 x 6- icles 4-8 x 4-5 mm. Flowers appearing together mm, ped- Volume 80, Number 4 1993 Chen & Nooteboom 1023 Magnoliaceae of China with the leaves, saucer-shaped, fragrant, erect, to 20 cm diam. when fully open; tepals 12(-14), xu. spathulate to oblong-obovate, 5-12(-14) cm, pure white or sometimes with “у at the w rose outside and white inside with flush of red to purple-striate; stame ous, 1.5-1.7 cm long, anthers sublaterally ird ing, connective appendage triangular, 1 -2 mm long, filaments 5-7 mm long; gynoecium cylindric 2- 2.5 x 0.3-0.4 cm, carpels ellipsoid. Fruiting pe- duncles 10-20 x 7-9 mm, pedicles 4-9 x 5-6 mm. Fruits cylindric and often distorted by the abortive carpels, 6-10 cm long, ripe carpels with- ns red, nu out beak, 7-15 mm long. Seeds subovoid to sub- cordiform, 8-12 x 7-9 mm, torus with the scars of perianth and stamens 8-10 x 8-10 mm. ко CHINA. Guizhou: Lei Gong Shan, Li Y.K. 8921. Henan: Lushih, Lao kiun shan, Hers J 1159. с с Hsien, Wilson Е.Н. 444; Hsing shan hsien, 373; Ichang, Henry А 2522; Patung hsien, Chow Hubei-Sichuan (Metasequ area): Cheng W.C. & C.T. Hwa 1049; Pate ge forest distr., Duanjiangping, Sino-Amer. Bot. Exped. 1181. Sichuan: Tchen Кеоџ Tin, Farges R.P. ke 2 Fu pi Pei C., 10242; Mt. Omei, Liu Y. Yunnan: Likiang, Ching R. C. 21681. Also edel from Shaan and и Сапзи. Ecology. Їп evergreen broad-leaved or mixed forests. Altitude: 1,200-2,000 m. Flowering March; fruiting June—Jul /ses. lowers and bark used for medicinal purposes; the tree is cultivated as an ornamental. 24. Magnolia zenii Cheng, Contr. Biol. Lab. Sci. Soc. China, Bot. Series 8: 291, f. 20. 1933. TYPE: W. C. Cheng 4233 (holotype, PE). Magnolia е Law & Сао, Bull. Bot. Res. (China) se E 190, t. 1. 1984. TYPE: China. Henan: Nanzhao , 12 Mar.. 1983, alt. 400-800 m, Z. Y. Gao 0129 (in Herb. Henan Inst. Biol. not seen). Deciduous tree to 7 m high; bark gray, smooth; young twigs purple-brown, glabrous, sparsely len- ticellate. Terminal winter buds narrowly ovoid, x 3-7 cm; apex abruptly short-acuminate, base broadly cuneate to blunt; nerves in 8-12 pairs. Petiole villous, glabrescent, 6-15 mm long. Flow- ers precocious, cup-shaped, fragrant, 10-12 cm diam., peduncles densely long-hairy, 2-8 mm long; tepals 9-12, subequal, spathulate, 7-8 x 3-4 cm, at first purple-red, becoming white above the mid- dle and purplish toward the base when fully open, the inner tepals 6, narrow; stamens purple, 1.5— thers dehiscing laterally, connective 1.7 cmlong, appendage triangular, ca. 1 mm long; filaments ca. 4 mm long; gynoecium cylindric, ca. 2 cm long, carpels ca. 4 mm. Fruits cylindric, 5-20 x 1-2.5 cm; ripe carpels subglobose, papilliferous, apex ob- tuse. Distribution. CHINA. Henan: Nan Zhao Xian, Z.Y. 126; Kiangsu, Paochuashan, Cheng W.C. 423. Ecology. In forest. Altitude: 250-300 m. Flowering April-May; fruiting August-September. Uses. Tree cultivated as an ornamental. Note. Only the type specimen of Magnolia zenii was available. We have not seen the type of Magnolia elliptilimba, but we reduce it to syn- onymy of M. zenii based on the description and the geographic distribution. Ib-2. Magnolia section Buergeria (Siebold & Magnolia stellata (Siebold & Zucc.) Maxim. (Buergeria stellata Siebold & Zucc.). Flowers appearing before the leaves (or with the leaves), pink (inner tepals sometimes tinged with rose or purple). Tepals very unequal, those of the outer whorl calycoid. KEY TO THE SPECIES OF MAGNOLIA SECTION BUERGERIA l. Inner tepals 4-5 cm long; connective a ndage of anthers 2-3 mm long; Posi 5 mm; peduncles 7- | x 4-5 m ен E biondii 1. Inmer tepals 3.2-10 cm; edis. append- б^ e of anthers 0.5-1 mm long; filaments 2- m long; peduncles 2-12 x 2-5 mm. ........... 2 г ета 2-10 х 2-3 mm; inner tepals З. 2- 9c 2. ы 10-12 x 4-5 mm; inner bas -10 ст; connective appendage of a mm; filament : 6. Magnolia “indi . Inner tepals 5-9 cm; connective appen of anther 0.5 mm; filament age пио ОЕСР 7. Magnolia kobus 3. Inner tepals 3.2-6.5 cm; a append- age of anther 1 mm; = d. Маң! stellata 25. Magnolia biondii Pampanini in Nuovo. Giorn. Bot. Ital. 17: 275. 1910. SYNTYPES: Scia-men-kvu, ca. 900 m alt., Pavol. 734, 7 34a (not seen). 1024 Annals of the Missouri Botanical Garden Magnolia conspicua Salisb. var. fargesii Finet & Gag- ull. Soc. Bot. France (Mémoires) 4: 38. 1905. 34. ТУРЕ: Farges 1300 (holotype, $ isotype, K). Magnolia оа ehder & Е. Н. Wilson in Sarg., Pl. Wilson. 1: 396, 408. т ТҮРЕ: E. Wilson 361 (holotype, А; isotypes, K, US). ] urpurascens Law & Gao, Bull. Bot. 0122 (in Herb. Henan Inst. Biol.). Deciduous tree to 12 m high and 60 cm diam.; bark pale gray, smooth; twigs slender, glabrous to appressed-pubescent, glabrescent; terminal winter buds ovate, densely yellowish villous. Stipules yel- lowish to white pubescent with short hairs, adnate to the lower base of the petiole, stipular scars 1– 4 mm long. Leaves chartaceous, dull green above, pale green, usually pubescent with long, straight, undulate, crisped and colorless hairs along the mid- rib and nerves to glabrous beneath, ovate, narrowly obovate-elliptic to oblong-elliptic, rarely oblanceo- late, 10-21.7 x minate, base cuneate to rounded; nerves in 10- 3.5-11 ст; apex acute to acu- 15 pairs. Petiole 1-2 cm long. Peduncles 7-11 X 4-5 mm, densely pubescent with long, yellowish hairs, pedicles very short, ca. 2 mm long. Flowers precocious, 6-8 cm diam., fragrant; tepals 9, the outer 3 nearly linear, 8-11 mm long, the inner tepals 6, white, spathulate, 4-5 x 1.3-2.5 cm, the innermost ones smaller than the others, abaxial faces tinged with purple at the base; stamens 8- 10 mm long, connective appendage 2-3 mm long, filaments thick, purple outside, white inside, 2.5— 3.5 mm long. Fruits cylindric, + contorted, 6– 14.5 cm long; fruiting peduncles + pubescent, glabrescent, 1-1.2 x 0.6 cm, pedicles 4-6 mm long; ripe carpels black, globose, compressed lat- erally, densely papilliferous, scars of perianth and stamens along the torus 9— Distribution. CHINA. Henan: Yudong Xian, Pucha Biaoben 23252; Lushin, Tungho, Hers J 991; Sunghsien, о 541; Yungning, Та Li Ping, 1343. Hu- : Li-chuan, Tuan-Pao-Shih, Hwa С.Т. 104; Liang Dung Gonn, Chun W.Y. 4409. Shaanxi: Nanwutai forest S of Hsingan (Si- an), Fenzel G. 2893, 2951. Sichuan: - hi, Kan-Tsao-Ping, Hwa С.Т. 65 Tchen Keou Tin, Purses R.P. 1300. Also reported from Gansu (W. Qinling) and Shaanxi (Dabas han). Altitude: 400-2,000 m. Flowering March; fruiting September. Uses. Wood used for furniture and general construction work; flower buds for medicine, flow- Ecology. In forest. ers for volatile oil; tree grown as an ornamental. Collector’s notes. Bud silky; flowers white; fruits purplish red tinged. Note. Several new species and varieties of Magnolia were recently published in Henan. Their descriptions basically fit in Magnolia biondii. Un- fortunately, we have not seen their type specimens or other collections; therefore, we have dealt with them under “Dubious Species." Pavolini (1908) misidentified this species as Magnolia obovata 26. Magnolia cylindrica E. H. Wilson, J. Ar- nold Arbor. 8: 109. 1927. TYPE: К. C. Ching 2949 (holotype, A; isotypes, E, K, US). Deciduous tree to 10 m high; bark white-gray, smooth; young twigs purple-brown to dark purple, yellowish appressed-pubescent, sometimes gla- brous; terminal winter buds ovate, yellowish seri- ceous with long hairs. Stipules densely pubescent, mm long. Leaves chartaceous to subcoriaceous, dark adnate to the petiole, stipular scars 2-7(- bright green, glabrous above, greenish, appressed- pubescent with fine, short, straight, and colorless hairs beneath, elliptic, narrowly obovate-elliptic, narrowly obovate to obovate, 8-15 x 3-9 cm; apex acute to short-acuminate, base cuneate or narrowly cuneate; nerves fine, prominent below, in 8-11 pairs. Petiole with same indument as young twigs, 7—10(-24) mm long. Peduncles densely yel- 0.4—0.5 cm, pedicles 5- -3 mm long; bracts 2, gray-pubescent out- side. Flowers precocious; tepals 9, the outer 3 subulate, 2-2.5 x white, tinged pale rose outside at the base, broadly spathulate to obovate, 6.5-10 x 2.5-4.5 cm, the innermost 3 tepals erect; stamens ca. 7 mm long, lowish villous, 1-1.2 x 0.5-1 cm, the inner tepals 6, anthers laterally dehiscent, connective appendage triangular, ca. 1 mm long, filaments purple, gla- brous, ca. 3 mm long; gynoecium green, cylindric, 1.5 x 0.5-0.6 cm, pedicles 5-10 x 4 mm. Fruits cylindric, 5-7.5 x 1.8-2.5 Distribution. CHINA. Anhui: Bai Ma Zhai, Lan Ni Ao, Jonzhai, Yao K. 9006; Wang Shan, Ching R.C. 2949. Fujian: near Zhejiang border, Ching R.C. 2501. Also reported from northern Jiangxi and western Zhejiang. Ecology. Altitude: 300-800 m. Flowering May-June; fruiting August-September. cm, pendulous. In forest. 27. Magnolia kobus DC., nom. cons. prop., Syst. Nat. 1: 456. 1817. Magnolia glauca var. a Thunb., Fl. Jap.: 236, 378, 1784b. Magnolia tomentosa Thunb., Trans. Linn. Soc. London. Volume 80, Number 4 1993 Chen & Nooteboom 1025 Magnoliaceae of China 2: 236. 1794a. Yulania kobus (DC.) Spach, Hist. Veg. 7: 467. 1839. TYPE: Kaempfer (lectotype, selected by Rehder (1930), BM). Magnolia oo Koidz., Bot. Mag. (Tokyo) 43: 29. Buergeria obovata Siebold. & Zucc. (non i. obovata Thunb. 1794), Abh. Math.- Phys. Cl. Kónigl. Bayer. Akad. Wiss. 4, 2: 187. 1845. TYPE: Von Siebold, left half of L- 908- 126- 1027 (lectotype, selected by Koidzumi (1929), L.). This specimen has disappeared in the Leiden Herbar- ium. Ueda (1985) provided a photo of it, and as long as the = has not been found, that is the only trace of a Deciduous tree to 20 m high, often branching near the base; bark gray, rough, fissured; young twigs 1.5-3 mm diam., green, soon turning purple- brown, glabrous or somewhat puberulent; terminal buds ovate, yellow-sericeous with long hairs. Stip- ules adnate to the base of the petiole, stipular scars 3-7(-10) mm long. Leaves dark green, white pu- bescent at the base of midrib above, gray-green, white pubescent along the nerves and at junctions with the nerves beneath, obovate-elliptic, ca. 8- 17 x 3.5-9.5(-11) cm; apex abruptly acuminate, base narrowly cuneate, slightly decurrent; midrib and nerves impressed when dry above; nerves in 8-12 pairs, margin slightly undulate. Petiole white villous, ca. 1-2.6 cm. Flowers appearing before the leaves, white, fragrant; peduncles 7-10 x 2- 2.5 mm, glabrous to hairy on nodes, pedicles 1.4— 4 mm long; bracts 2-3, pubescent outside; flower buds narrowly ovate, flowers ca. 9-10 cm diam. when fully open, tepals 9, outer tepals 3, greenish or white, triangular-linear, 5-15 x 2-3 mm, inner tepals white, sometimes flushed with red at the base, spathulate, ca. 5-7(-9) mens ca. 8-10 mm long, anthers dehiscing la- 1.5-3 cm; sta- trorsely or sublatrorsely, filaments red, ca. 2 mm; connective appendage triangular, ca. 0.5 mm long; 1-1.5 cm long. Fruiting peduncles 8-10 x 3-6 mm, pedicles 6– gynoecium green, cylindric, ca. 3—4 mm. Fruits cylindric, 3.5-11 cm, often contorted by sterile carpels; carpels nearly oblong, white lenticellate. Scars of perianth and stamens 5-6 x 3-5 mm. Seeds subglobose, 10-9 x 9-8 mm. Distribution. Native of Japan and € Korea, а іп Qingdao, Nanjing, and Hangz Uses. Wood used for general construction work and furniture, tree grown as an ornamental. ote. e Kaempfer collection of Magnolia named ““kobus,” from which Kaempfer t. 42 (1791) is drawn, forms the main part of the protologue of Thunberg’s (1784b) Magnolia glauca var. a. Magnolia tomentosa Thunb. (1794a) is based on Magnolia glauca Thunb. var. o, with citation of Kaempfer t. 42. So far there is no doubt about the identity of Magnolia tomentosa, especially be- cause Thunberg visited London and studied Kaempfer’s collections and drawings. But Thun- berg (1794b) published a list of specimens kept in UPS, which includes Magnolia tomentosa; this name is, in his handwriting, written on sheet num- ber 12887. This specimen, however, appears to e Magnolia stellata. If Ueda's (1986) interpre- tation, that 12887 as the type of Magnolia tomentosa, is correct, Magnolia tomentosa becomes the legiti- hunberg himself selected number mate name for Magnolia stellata. The latter name is then superfluous. Thunberg (1805) also pub- lished a drawing of sheet number 12887, with the name Magnolia tomentosa but without a descrip- tion. This has commonly been cited as Magnolia tomentosa Thunb. (1805), non Thunb. (1794a). De Candolle (1817), when describing Magnolia kobus, cited Banks ic. Kaempfer t. 42 first in the protologue, and then Magnolia glauca Thunb. var. а. He cited Magnolia tomentosa Thunb. with a question mark. De Candolle also cited Magnolia gracilis Salisb. Salisbury (1806) described Mag- nolia gracilis with an excellent figure based on a collection that was sent to De Candolle and is kept in G-DC. This collection appears to be Magnolia quinquepeta. However, because Salisbury cited Magnolia tomentosa without excluding its type, Magnolia gracilis is a superfluous name of Mag- nolia tomentosa. De Candolle, possibly not know- ing what Magnolia tomentosa was, a new name to Magnolia gracilis Salisb.: Magnolia ko- bus DC. Because De Candolle cited Magnolia to- mentosa with a question mark, it could be assumed that he excluded the type of that species, and that Magnolia kobus DC. is a superfluous name of Magnolia quinquepeta (Buc'hoz) Dandy (1934), based on Lassonia quinquepeta Buc'hoz (1779). This was Ueda's (1985, 1986) opinion, and Koid- zumi's (1929). Koidzumi assumed that the type of Magnolia gracilis Salisb. was also the type of Magnolia kobus DC. He therefore adopted a no- men novum, Magnolia praecossima, for the spe- cies up to then known as Magnolia kobus. His name was based on a synonym, Buergeria obovata Siebold. & Zucc. (non Magnolia obovata Thunb., 1794а). However, it is also possible to exclude Magnolia gracilis Salisb. from Magnolia kobus DC. Then the latter is typified with the type of Magnolia tomentosa Thunb. 1026 Annals of the Missouri Botanical Garden In that case there are two possibilities. If Mag- nolia tomentosa Thunb. is lectotypified with Thun- berg 12887, Magnolia kobus DC. is a superfluous name for Magnolia tomentosa Thunb. and is the species hitherto known as Magnolia stellata (Sie- bold & Zucc.) Maxim. (1872), based on Buergeria stellata Siebold & Zucc. (1845). As pointed out above, Magnolia stellata is then a superfluous name for Magnolia tomentosa Thunb. If Mag- nolia tomentosa Thunb. is lectotypified with the collection of Kaempfer, as done by Rehder (1930), Magnolia kobus DC. is a superfluous name for Magnolia tomentosa Thunb. This is rejected by Ueda (1986), who stated that Rehder’s lectotypi- fication was contrary to the rules because he wou have selected it mechanically. Because Rehder gave ample justification for his choice, we reject that he chose mechanically. We accept Magnolia kobus DC. as the name for this species, and we propose both Magnolia kobus and Magnolia stellata be conserved. 28. Magnolia stellata e & Zucc.) Max- . 1872. Magnolia kobus DC. f. ids (Siebold & Zucc.) Blackb., Popul. Gard. 5, 3: 68. 1954. Magnolia kobus var. stellata Blackb., Amatores. Herb. 17: 2. 1955. Buergeria stellata Siebold & Zucc., Abh. Math.-Phys. Cl. Kónigl. Bayer. Akad. Wiss. 4, 2: 186, t. Па. 1845. Talauma stel- lata (Siebold & Zucc.) Miq., Ann. Mus. Bot. Lugduno-Batavum 2: 257. 1866. TYPE: Von Siebold L-908-126-701, -750 (L not seen). The type was photographed at L for Ueda's (1985) publication. The specimen had not yet been incorporated in the family at the time the Rijk- sherbarium moved to another building. The type has never been seen again. nu — Parsons ex Robinson, Gard 13: 572, 1878. TYPE: probably the figure. ie ela var. keiskei Makino, Bot. Mag. (To- . 1912. Magnolia keiskei (Makino) Ihr- ig, yok E Univ. Wash. 11, 2: 33. 1948. TYPE — plate cited by Makino. Magnolia stellata var. rosea Arnott ex Veitch ex Hu, Man. Econ. Pl. 1: 383. 1955. Magnolia rosea (Veitch) Ihrig, Arb. Bull. Univ. Wash. 11,2: 34. 1948. Magnolia kobus DC. f. rosea а Black- burn, Popul. Gard. 5, 3: 73. 1954 Deciduous large shrub or treelet to 5 m high; young twigs slender, 2-3 mm diam., brown-purple to black-purple, sometimes gray-black, densely ap- pressed-pubescent with long, gray or yellowish hairs, glabrescent, smooth, old twigs brown, glabrous, glabrescent. Stipules densely gray pubescent with silky hairs, adnate to the petiole, stipular scars as long as the petiole, 3-14 mm. Leaves thinly char- taceous, dark bright green, glabrous above; pale green, glabrous below, narrowly obovate to ob- ovate, occasionally ovate, 7-11 x 2-5.6 cm; apex obtuse to acute, base cuneate to attenuate, usually decurrent along the petiole, rarely rounded; margin sparsely pubescent; midrib and nerves finely pu- bescent above, densely villous, glabrescent be- neath; nerves fine, visible below, in 10-14 pairs, reticulation coarse, hardly visible on both sides. Petiole covered with same indumentum as the stip- ules, glabrescent later, 3-14 mm long. Flowers appearing before and after the leaves, white or rose, fragrant, to 10 cm diam.; peduncles densely gray villous, 2-6 x 2-3 mm, pedicle 0.2-4 x 3 mm; tepals 12-18, very unequal, usually narrowly obovate to spathulate, apically obtuse, rarely acute, basally broad, 3.2-6.5 x 0.7-1.7 cm; stamens 6-11 mm long, anthers dehiscing latrorse or sub- latrorse, connective appendage triangular, ca. 1 mm long, filaments 2-5 mm long; gynoecium cy- lindric, 7-10 x 2-3 mm; carpels many. Fruiting peduncles densely hairy, 3-6 x 3-4 mm. Fruits cylindric or irregularly shaped abortive carpels, 2- 4.5 X mm long, with short beaks at the apex; scars of perianth and stamens along the torus 2-6 x 3-5 mm. 1-1.5 ст; пре carpels subglobose, 7-12 Distribution. | Native to Japan, cultivated in Nanjing. Uses. Tree grown as an ornamental. Note. Ueda (1985. 1986) gave an account as to why the common star Magnolia, Magnolia stel- lata, should be legitimately named Magnolia to- mentosa. However, because of the economic im- portance of the species it will be proposed as nomen conservandum. See also the notes under M. kobus. Ib-3. Magnolia section Tulipastrum (Spach) Dandy, in Camellias and Magnolias, Conf. Re- port: 74. 1950. Tulipastrum Spach, Hist. Na- tur. Veget., Phanerog. 7: 481. 1839. TYPE SPECIES: Magnolia acuminata L. Flowers appearing together or after the leaves, pink, purple, or red. Tepals very unequal (those of the outer whorl calycoid). 29. Magnolia quinquepeta (Buc'hoz) Dandy, J. Bot. 72: 103. 1934. Lassonia quinquepeta Buc'hoz, Pl. Nouv. Découv. 2: f. 2 of t. 19. 1779. TYPE: f. 2 of t. 19 (Buc'hoz, 1779). Volume 80, Number 4 1993 Chen & Nooteboom Magnoliaceae of China 1027 жеш. glauca [var.] 8 Thunb., Fl. Jap.: 236, 378, 1784. Magnolia lilüflora Desr. in Lam., Encycl. 3: 675, “liliflora.” 1792. Ohwi, Fl. Jap. rev. ed.: 649, incl. var. gracilis. 1978. Magnolia obovata Thunb., Trans. Linn. Soc. London 2: 336, nom a illeg. 1794a. Yulania japonica Spach, Hist. Nm 7: 466, nom. et comb. € 1839. TYPE: , Mokkwurén 2 (Kaempfer, 179 Маал. purpurea Curtis, Bot. Mag. Tes 11:1. . 390. 7. TYPE: t. 390. idi e po od Vent., Jard. Malmaison: t. 24. 1803. . 24. Magnolia gracili Salisb., Parad. Lond. 2: 87, t. 87, 1807. Magnolia liliflora var. pracilis (Salicb. ) Rehder i in L. H. Bailey, Stand. Cycl. Hort. 4: 1968. 1916. TYPE: s (G-DC). Buergeria obovata Siebold & Zuc , Abh. Math.-Phys. Cl. Kóni . 1845. тше E Zucc.) Benth. & Hook. f. ex Ha Р Вог. . 1882 (non Korthals, 1851). TYPE: Von 7 wight half of 1-908-126-1027 (L). Deciduous shrub to 3.5 m high; bark gray-brown; young twigs 2-4 mm diam., dark purple, glabrous or sometimes hairy on a few nodes near the pe- duncles, smooth, sparsely lenticellate, old twigs gray- purple and longitudinally fissured; terminal winter buds ovoid, 12-14 x 5-6 mm, yellowish sericeous. Stipules densely yellowish to yellowish white pu- bescent, adnate to the entire petiole, scars 3-15 mm long. Leaves membraneous to chartaceous, dark green, glabrous, sparsely pubescent over the surfaces or only along the midrib above, pale green, glabrous to pubescent with long, undulate to rarely crisped, colorless hairs along the midrib and the nerves below, broadly elliptic to obovate, 6-15(-26) A =U- cm; apex acute to acuminate with short or long acumen, base cuneate to attenuate, usually decurrent along the petiole; midrib and nerves prominent beneath, nerves in 8-10 pairs. Petiole pubescent, glabrescent, ca. 3-15 mm long. Flowers appearing with and after the leaves, slightly fragrant; peduncles glabrous or densely villous with silky, yellowish hairs, glabrescent, 5-20 x 6-8 mm, pedicles 10-12 x 5-7 mm, sometimes ab- sent; bracts densely yellowish sericeous, 11 cm long; tepals 9, the outer 3 subulate, occasionally ovate, greenish to purplish, 10-27 x 4-8 mm, the inner tepals 6, spathulate to obovate, purplish outside and white inside, 5.5-9 x 2.5-4.5 cm; stamens purple-red, 8-10 mm long, anthers de- hiscing laterally, connective appendage triangular, filaments purple, ca. 2 mm long; gynoecium cylin- dric, purplish, 1.2-1.4 x 0.4 cm, carpels many. Fruiting peduncles 10-20 x 3-6 mm, pedicles 3- 8 x 3-4 mm long. Fruits cylindric, erect to curved as the carpels partly abortive, 3-6.5 x 1-2 cm; ripe carpels ellipsoid to subglobose, 8-15 mm long, gray-lenticellate outside, short-beaked; torus with the scars of perianth and stamens 8-12 x 5-7 mm. Seeds subcordate, ca. 10 X 7 mm. Distribution. CHINA. Hubei: а е. ^ Chen S.J. et al. 840055; Chengtu, Fang W Kun ве Wang F.T. 20793, Chiao ^j ^ 444. Yunnan: N end Lichiang valley, 2 G. 2174; Shweli- Salwin divide, 25?48'N, aE, 2470 Tien chin pu valley, 25°30'N, 5539. Widely “э, Ecology. Flowering March-April; Fruiting August-September. Uses. Flower buds and bark used for medicinal purposes, the tree as an ornamental. Note. Ueda (1985) explained the complicated synonymy. The species was formerly known as Magnolia lili(i)flora. The name Magnolia obova- ta Thunb. has also been used for this species for a long time. See also the note under Magnolia heptapeta. Magnolia hirsuta Thunb. (1794b, 1824), based on Thunberg 12884, is a nomen nudum. Ic. Magnolia subgenus Talauma (Juss.) Pierre, . For. Cochinch. 1: sub t. 1. 1880. TYPE SPECIES: Magnolia plumierii Schwartz. Evergreen tree. Stipular scars up to the apex of the petiole. Anthers introrse. Fruits with connate carpels, the apical parts of the carpels falling off leaving the seeds exposed, hanging from the thread- like funiculus. Ic-1. Magnolia section Blumiana Blume, Fl. Java Magnol.: 32. 1829 Leaves evenly distributed; midrib prominent above, at least toward base. Flowers bisexual; tepals subequal; connective produced into a short or long appendage, triangular. Gynoecium not stipitate. Fruits with connate carpels. When mature the api- cal parts of the carpels circumscissile and falling, dehiscing along the dorsal suture or not, the basal parts remaining adnate to the torus, or apical parts falling in irregular masses. 30. Magnolia candollii (Blume) H. Keng var. obovata (Korth.) Nooteboom, Blumea 32: 374. 1987. Talauma obovata Korth., Ned. Kruidk. Arch. 2, Versl.: 98. 1851. TYPE: Su- matra Gunung Pamatton, Korthals s.n. (lec- totype, selected here, L sheet no. 925.2 739; isolectotype, BO). 1028 Annals of the Missouri Botanical Garden Talauma se Hook. f. & Thomson, Fl. Ind. 74. 1855. TYPE: Sikkim, 2,000-5,000 ft., J. D. Hook n. (ho dot sida K; isotype, Talauma oblanceolata Ridley, Fl. 25. . Ridlley 15590 (K. SING). Talauma ен Craib, Kew Bull. 1925: 7. 1925. nolia betongensis (Craib) H. Keng, Gard. Bull. ne: E 129. 1978. TYPE: Kerr 7449 (holotype, Malay Penins. 5: 286. К; і ). ins. sclerophylla Dandy, J. Bot. 66: 47. 1928. TYPE: Haviland 3148 РО e ps e, К). Talauma. levissima Dandy, Kew Bull. 1928: 191. 1928. TYPE: H. N. Ridley 9047 геп К; isotype, SING). Tree to 25 m high; young twigs whitish, stout, ca. ] cm diam., glabrous. Stipules adnate to the petioles, stipular scars from up to about halfway the length of the petioles to nearly as long as the petioles. Leaves coriaceous, glabrous, obovate or 8-22 rounded to slightly acuminate, base cuneate, usu- sometimes elliptic, 17- cm; apex ally attenuate; nerves in 9-25 pairs, curved up- ward and meeting in an intramarginal vein, retic- ulation rather coarse, sometimes obscure. Petiole 2.5-7 cm long. Peduncles glabrous, 3-12 x 0.5- 1.3 ст, with 2-18 nodes, pedicles absent or very short. Tepals 9, thick fleshy, the outer 3 greenish outside, sometimes reflexed in mature flowers, 1.5— 9(-6.5) x 1-2 cm, the inner tepals 6, creamy white, erect, shorter than to as long as the outer tepals; stamens 1.2-3 cm long, connective ap- pendage triangular to subulate, 1—3 mm long, fil- aments 1-3 mm long; gynoecium ellipsoid, 1—2 cm long, carpels 10-100, pubescent, the styles long, becoming woody spines to 15 mm in fruits but sometimes caducous; scars of perianth and stamens on torus ca. 5 X 9-15 X 4-7.5 cm mm. Fruits ellipsoid, Distribution. Tibet, Sikkim, Nepal, Assam (Khasia), ди о nd, or = In CHINA. Southern Yunnan: D.T. Tao Ecology. In evergreen broad-leaved forest. Altitude: 800-1,500 m. Flowering May-June; fruiting August-September. otes. Without fruits this species is hard to distinguish from Magnolia henryi Dunn. The lat- ter species has, as typical in subgenus Magnolia, follicles that dehisce dorsally. In subgenus Talauma the apical parts of the carpels fall off entirely, leaving the axis with the carpellar scars and red seeds hanging from a funiculus. Magnolia henryi is therefore intermediate between the two subgen- ега. Talauma luchuensis Law (manuscript), from Yunnan (D.T. Tao 413, KUN), is Magnolia can- dollii var. obovata, with the developing fruits ab- normally precocious. Magnolia candollii var. obovata was errone- ously identified as Manglietia glauca (non Blume, 1823) by Ridley (1922) for the collection of L. M. Bell & Haniff and by Ridley (1913) and Merrill (1921a) as Talauma candollei (non Blume, 1823). HYBRIDS AND CULTIVATED SPECIES OF SECTIONS Not NATIVE TO CHINA 31. Magnolia x soulangiana Hamel, Ann. Soc. Hort. Paris 1: 90 . Magnolia yulan var. soulangiana (Hamel) Lindl., Bot. Reg 14: t. 1164. 1828. TYPE: We do not know whether a lectotype has ever been designated. Deciduous small tree or large shrub to 6-10 m high; young twigs purple-brown or dark purple, 3- 4 mm diam., sparsely to densely pubescent, or hairy only at the leaf axils, soon glabrescent; ter- minal winter buds ovoid, densely brownish villous with somewhat shiny hairs. Stipules densely pu- bescent with appressed, long, straight, and brown- ish hairs, adnate to the long. Leaves membraneous to chartaceous, dark bright green, covered with scattered hairs every- where or along midrib and nerves above, greenish, pubescent with long, straight, undulate to curly, colorless hairs along midrib, nerves and reticulation beneath, obovate, 6-17.5(-22) x 4-9(-13) em; apex usually abruptly short-acuminate, acumen 3- O mm long, base obtuse, cuneate to attenuate; petiole, scars 4-15 mm nerves conspicuous below, in 7-10 pairs, reticu- lation coarsely netted, obscure on both surfaces. Petiole pubescent, 1-1.5 cm long. Flowers ap- pearing before and after the leaves; peduncles densely pubescent with yellow to yellowish white hairs, 12-14 x 4-6 mm, pedicles 5- 3- mm; tepals 9, white flushed with pink or purple outside, white or pink inside, spatulate to obovate, the outer 3 subequal to the inner 6 tepals, some- times about half the length of the inner ones, 5- 12.5(-15) x 2-5.5(-9) cm; stamens purple, l- 1.2(-1.4) ст long, anthers dehiscing laterally, con- nective appendage triangular, ca. 1 mm long, fil- aments 3-4 mm long; gynoecium green, glabrous to hairy, cylindric, 1.5-2.2 x 0.3-0.4 cm. Fruits cylindric, ca. 5-10 x 3 cm; ripe carpels ovoid to subglobose, white-lenticellate, 1-1.5 cm long, with- out beaks; torus with scars of perianth and stamens ca. 8 x 8 mm. Seeds broadly ovoid or obovoid, slightly compressed, ca. 1.2 cm long. Note. Hybrid from Magnolia heptapeta and Magnolia quinquepeta, cultivated in gardens as an ornamenta Volume 80, Number 4 1993 Chen & Nooteboom 1029 Magnoliaceae of China Magnolia (subgenus Magnolia) section Theo- rhodon Spach, Hist. Natur. Veget. Phaner- og. 7: 470. 1839. Stipules free from the petiole, leaves evergreen. Flowers neither precocious nor with a much re- duced outer whorl of tepals, tepals subsimilar in texture. Fruit ellipsoid to oblong. Not native to China. 32. Magnolia grandiflora L., Syst. Nat. ed. 10 (2): 1082. 1759. TYPE: ? Evergreen tree up to 30 m high in its original habitat; bark brownish or gray; young twigs stout, 6-7 mm diam., dull brown, densely yellowish se- riceous, eventually glabrous. Stipules covered with the same indument as the young twigs, free from the petiole. Leaves thick coriaceous, dark green, glossy above, green, densely hairy with short to long, undulate to curly, colorless or brown hairs, gradually glabrescent beneath, elliptic, ovate to x cm; apex acute to short- acuminate, base cuneate to broadly cuneate; midrib obovate, 10— and nerves visible on the undersides, nerves in 8— 10 pairs, reticulation coarsely netted, prominent above, obscured below. Petiole 1.5-4 cm without stipular scars. Flowers erect, heavily scent- ed, 15-20 cm diam., peduncles 3-4 x 0.8-0.9 cm, densely yellowish sericeous; pedicles very short, long, ca. 1 mm long; tepals 8-12, subsimilar, white, broadly obovate, broadly elliptic to orbicular, ba- sally short-clawed, concave, thick fleshy, 8-10 x -8 cm; stamens purplish, ca. 2 cm long, anthers introrsely dehiscing, connective appendage trian- gular, ca. 2 mm long, filaments flat, 1-2 mm long; gynoecium ellipsoid, densely yellowish tomentose; carpels ovate, 1-1.5 cm, styles recurved. Fruiting peduncles 3-4 x 0.8–0.9 cm, pedicles ca. 1 mm 5-10 x 4-5 cm, densely brown or yellow tomentose; ripe carpels long. Fruits ovoid-cylindric, apically long-beaked, scars of perianth and stamens along torus 1.7-2 X 1 cm. Seeds ellipsoid to ovoid, compressed laterally, ca. 1.4 x 0.6 cm Distribution. Native of southeastern United States, cultivated in southern China Uses. Wood yellow-white, heavy and hard, used for interior finish; leaves, young shoots, and flowers extracted for volatile oil; spectacular ornamental. the tree is a DUBIOUS NAMES Magnolia honanensis B. C. Ding & T. B. Chao, J. Henan Agricultural College 4: 6. 1983. SYNTYPES: China. Henan: Yu Хлап, 7. B. Chao et al. 838281, 838282 (in Herb. Henan Col- lege of Agriculture Conservatus) = Magnolia biondii Pamp.(?). Magnolia biondii var. parvialabastra T. B. Chao, Y. H. Ren & J. T. Gao, J. Henan Agricultural College 4: 7. 1983. SYNTYPES: China. Henan: Nanchao Xian, T. B. Chao et al. 83671, 83672 (in Herb. Henan College of Agriculture Magnolia biondii Pamp.(?). Conservatus) — Magnolia biondii var. ovata T. B. Chao & T. X. Zhang, J. Henan Agricultural College, 4: 9. 1983. TYPE: China. H Lushan Xian, T. B. Chao et al. s.n. (in Herb. Henan College of Agriculture Conservatus) = Magnolia bion- dii Pamp.(?). Magnolia biondii var. purpura T. B. Chao & Y. C. Qiao, J. Henan Agricultural College, 4: 10. 1983. SYNTYPES: China. Henan: Lushan Xian, Y. Z. Qiao et al. 0005, 0006 (in Herb. Henan College of Agriculture Conservatus) = Mag- nolia biondii Pamp.(?). Magnolia biondii var. planities T. B. Chao & Y. Z. Qiao, J. Henan Agricultural College 4: 10. 1983. TYPE: China. Henan: Yu Xian, 7. B. Chao et al. s.n. (in Herb. Henan College of Agriculture Conservatus) = Magnolia biondii Magnolia denudata Desr. var. pyramidalis T. B. Z. X. Chen, J. Henan Agricultural College 4: 11. 1983. SYNTYPES: China. Henan: Zengzhou, T. B. Chao et al. 83871, 83872 (in Herb. Henan College of Agriculture Con- servatus) = Magnolia heptapeta (Buc'hoz) Dandy (?). Magnolia axilliflora T. B. Chao, T. X. Zhang & . T. Gao in T. B. Chao, Acta Agriculture Universitatis Henanensis 19(4): 360. 1985. SYNTYPES: China. Henan: Nanchao Xian, 7. B. Chao et al. 83815, 83816, T. X. Zhang et al. 82910, 82911 (in Herb. Henan College of Agriculture Conservatus) = Magnolia bion- dii Pamp.(?). Magnolia axilliflora var. alba T. B. Chao, Y. H. & J. T. Gao, Acta Agriculture Univer- sitatis Henanensis, 19(4): 360, photo 4. 1985. SYNTYPES: China. Henan: 7. B. Chao, Y. H. Ren & S. D. Zhuao 855181, 855182, T. B. Chao, J. T. Gao et al. 854151, 854152 (flow- ering, in herb. Henan Agricultural University Conservatus) = Magnolia biondii Pamp.(?). Magnolia Y d de var. multitepala T. B. Chao, Y. H. Ren T. Gao, Acta Agriculture Universitatis pois 19(4): 361. 1985. 1030 Annals of the Missouri Botanical Garden SYNTYPES: China. Henan: 7. B. Chao, Y. H. Ren & S. D. Zhuao 855201, 85471 (flow- ering, in herb. Henan Agri. Univ.) = Mag- nolia biondii Pamp.(?). Magnolia funiushanensis T. B. Chao, J. T. Gao & Y. H. Ren, Acta Agriculture Universitatis Henanensis, 19(4): 362, photo 5. 1985. TYPE: China. Henan: 7T. B. Chao, J. T. Gao & Y. H. Ren 85019 (flowering, in herb. Henan Agri. Univ.) = Magnolia biondii Pamp.(?). Magnolia funiushanensis var. purpurea T. B. Chao & J. T. Gao, Acta Agriculture Universi- tatis Henanensis, 19(4): 362. 1985. SYNTYPES: China. Henan: 7. B. Chao, et al. 855218, 855219 (flowering, in herb. Henan Agri. Univ.) = Magnolia biondii Pamp.(?). Magnolia biondii var. flava T. B. Chao, J. T. Gao & Y. H. Ren, Acta Agriculture Universitatis Henanensis, 19(4) 362. 1985. SYNTYPES: China. Henan: 7. B. Chao, et al. 855207, 855208, 84019 (flowering, in herb. Henan Agri. Univ.) = Magnolia biondii Pamp.(?). "m biondii var. tatitepala T. B. Chao & Gao, Acta Agriculture Universitatis He- nanensis, 19(4): 363. 1985. TYPE: China. He- T. B. Chao & J. T. Gao 85472 (flow- ering, in herb. Henan Agri. Univ.) = Magnolia biondii Pamp.(?). Magnolia denudata Desr. var. angustitepala T. B. Chao & Z. S. Chun, Acta Agriculture Uni- versitatis Henanensis, 19(4): 363. 1985. SYNTYPES: China. Henan: 7. B. Chao, et al. 843151, 843512 (flowering, in herb. Henan Agr. Univ.) = Magnolia heptapeta (Buc'hoz) (?). Magnolia sinostellata P. L. Chiu & Z. H. Chen, Acta Phytotax. Sin. 27, 1: 79. 1989. TYPE: China. Zhejiang: Caoyutang of Jingning, J. P. Si & H. F. Pan JH-002 (holotype, ННВС; isotype, PE) = Magnolia stellata (7) (Siebold & Zucc.) Maxim П. Manglietia Blume, Verh. Bat. Gen. 9: 149. 1823. Magnolia sect. Manglietia (Blume) Baillon, Adansonia 7: 66. 1866. TYPE SPECIES: Manglietia glauca Blume. ed Hu & Cheng, Acta Phytotax. Sin. 1, 255. . TYPE SPECIES: "i aramanglietia aro- matica i Dad) Hu & Chen Manglietiastrum Law, Acta Hora 5ш, 11: 72, 5. 2. 1979. T TYPE SPEC IES: Manglietiastrum sinicum Law. Tree. Stipules adnate to or free from petiole. Flowers terminal, solitary, bisexual; tepals 9-13, 3-merous, subequal; anthers introrse, connective produced into a short or long appendage. Gynoe- cium sessile, but a gynophore present in section Маи ера Follicles free, crowded, dehis- cent along the dorsal and sometimes also the ventral suture. Distribution. About 25 species in tropical and sub- tropical Asia from the eastern Himalayas eastward to southern China and Malesia (not in the Moluccas and New Guinea). Eighteen species in China Note. genus mainly because the number of ovules is 4— 12(-16), while in Magnolia it is 2(-4). Leaf and wood anatomical characters point to the fact that Manglietia consists of a group of closely related Manglietia is still treated as a separate species, although the similarity with. Magnolia is great. However, all genera of Magnoliaceae as recognized by Nooteboom (1985) are quite similar in anatomical characters. Many species of Manglietia have a hypodermis of 1-2 subepidermal layers beneath the upper epi- dermis, some also adjacent to the lower epidermis; few species have this hypodermis adjacent only to the lower epidermis. Species of Magnolia mostly lack a hypodermis. Manglietia consists entirely of evergreen species, thus the anatomy must be com- pared with the anatomy of evergreen species of Magnolia. The species of Manglietia usually do not have helical thickenings and usually have sca- lariform perforations and large vessel-ray pits throughout the rays; this pitting is present only in the marginal ray cells of the evergreen species of Magnolia. Scalariform perforations do occur in some deciduous species of Magnolia, and helical thickenings are found in a few species of Man- glietia. Regarding Manglietia sinica [Manglietias- trum sinicum Law, Magnolia sinica (Law) Noot.; Nooteboom (1985) at that time was not able to study any collections], we conclude that the num- ber of ovules, as well as the sometimes well-de- veloped hypodermis, and the scalariform and uni- lateral compound vessel-ray pitting throughout the rays, warrant regarding it as a species of Man- glietia, not of Magnolia. Moreover, oil cells are usually absent in the rays, as in Manglietia. How- ever, the discontinuous helical thickenings in the vessels point to Magnolia, although this character is found in a few other species of Manglietia. KEY TO THE SPECIES OF MANGLIETIA (primarily based on flower characters) l. Twigs glabro 2 1: Twigs Duos at үне when young. == 12 2(1). Gynoecium stipitate; stipules free. |... 18. Manglietia sinica Volume 80, Number 4 1993 Chen & Nooteboom 1031 Magnoliaceae of China 9(6). 10(9). m not stipitate; stipules adnate to pees glaucous beneath. Leaves not glaucous beneath. Twigs 4-5 mm thick; leaf base narrowly cuneate, reticulation rather obscure. Dia ш Ма е аи Twigs 6-10 mm ben leaf base cun е distin Peduncle 15 m E stipules pubescent; leaves obovate, Е 7 cm broad, apex short- acuminate; gynoecium nb fruits ovoid, 6 cm long, follicles hair nglietia microtricha Peduncle 30-34 mm кл. stipules brous; leaves narrowly obovate, 10.5- cm broad, apex acuminate; gynoecium ке brous; NOR ovoid-oblong, 10-16 cm lon follicles glabrous. ...... anglietia grandis Peduncle 5-7 mm thick 7 Peduncle 2-4 mm in id k. 9 m high, carpels 70-80; fruits 9.5 c lon ng. Manglietia "lucida Leaves 3.5-10 cm broad, с са er obscure; stipular scars on petiole 1-7 mm long; peduncle abd e a (green- ish) white, rd ps 5-8(-20) ст long; m = m high, ome 12- r 38- 46; s mi = 7 em long. _______ ди E ovoid; fruiting peduncles 3-5.5 m long, fruits ellipsoid. Ma ngua pue pap Pedicle 6-8 mm; stipules pubes base attenuate-cuneate, apex uiia © 5 да E fo Cc 5 е un d long; gynoecium broadly ellipsoid; fruiting peduncles 9-11 mm long, fruits ovoid or ovoid-globose. CERE 5b. Manglietia fordiana var. calcarea Carpels 45-86. 10 Carpels (12-)20-36 11 Gynoecium glabrous, tree to 25 m stipules glabrous or pubescent; reticulation of leaf distinct; outer tepals thin-coriaceous, stamens 13-16 mm long, connective ap- pe m triangular with a sharp pointed tip, filaments 2-5 mm long; gynoecium ovoi or aps ovoid, 17- 28 mm high; fruiting fruit 14-20 mm d stamens along torus under m long. 10. Manglietia insignis Gynoecium hairy, tree to 8 m high; stipules sparsely hairy; reticulation of leaf rather obscure; outer tepals thick-fleshy, stamens 11(9). 14(13). 18(18). 18. . Outer tep y 0 mm long, connective appendage trian- ular, Mame. 1 mm long; gynoecium el- lipsoid, 14-15 mm high; fruiting peduncles mm thick, fruits ovoid-ellipsoid, 3.5 cm long; scars of perianth and stamens along torus under fr в 7-9 mm lon . Manglietia duclouxi rey tepals 5.5-7 cm aires inner tepa -4.8 cm long; Mr sind appendage of las triangular, 1 m long; leaf base cuneate, reticulation c anglie "s fordiana v var. r. fordiana mm long; leaf base arrowly cuneate, reticulation fine. _____ ^ Manglietia conifera 13 Leaves glaucous benea fleshy: con- nective appendage of о triangular 3. е dandyi Peduncle 15-90 m ite apex ea acuminate or acute, К not dilated at base; outer tepals 3. ong, thin- coriaceous; connective он и e of an- pa triangular with a sharp scr tip, semiorbicula Stipules glabrous: reticulation of leaf dis- pointed tip, 4 mm long; scars of bracts 2 nective ну көн lar, m long; scars of bract 2. Ma MEA conifera l. poo 20-39. arpels 44-88. als 3.5 ст long; flower (greenish) 2. ellow or yellowish, inner tepals cm long. 2. Manglietia ere Outer tepals 5.5-8 cm long; flower ie ish) white, inner tepals 4-8 cm lon _ Connective appen ndage semiorbicular: inner pals 9; gynoecium ellipsoid; fruits subglo- ees scars of perianth and stamens along torus under fruit 7-10 mm long. Ma nglietia aromatica Connective appendage triangular or very ag and stamens along torus under fruit 13-22 mm lon Connective appendage very long, narrowly angular; reticulation of leaves distinct, edd dilated at base; filaments 1 mm lon 5c. Manglietia fordiana var. forrestii Connective appendage triangular; lation of leaves rather obscure, S not dilated at base; filaments 1.5-2 mm lon 5a. Manglietia fordiana var. fordiana 1032 Annals of the Missouri Botanical Garden 19(15). Leaves glabrou 20 30. Gynoecium pamm pedicle present; fol. 19. Leaves hairy beneath at at least when young. 22 licles glabrou 20(19). Peduncle 30-37 mm long. == 31(30). Peduncle 10- 15 mm long, pedicle 3-5 mm; 9. Manglietia књ reticulation of leaf distinct; outer tepals thin- 20. Peduncle 10-15 mm long. = coriaceous; connective age of an 21(20). Leaves 11-16 cm broad, obovate, reticu- thers triangular with a sharp pointed tip, lation coarse; peduncle stout, 6 mm thick; fil 2-5 mm long; gynoecium 17- outer tepals 12-14 cm long, а mm high; fruits ovoid-oblong ог cylindric, fruiting peduncles 4.5 cm long, fruits ovoid. 6.5-11 cm long; scars of perianth and sta- 1 Manglietia lucida mens ом torus under fruit 14-20 mm 21 Leaves 4-8 cm broad, elliptic ог narrow long. . 10. ph eir oe obovate, reticulation fine; peduncle slender, 31. Pedun cle 22- -30m m long, pedicle 1-2 4 mm thick; outer tepals 5-7 cm long, thin reticulation of leaf rather obscure; outer coriaceous; fruiting peduncles 1.2-2.5 cm epals coriaceous; connective appendage of long, fruits ovoid-oblong or cylindric. ....... anthers semiorbicular, filaments 1 mm long; . Manglietia insignis gynoecium 11-15 mm high; fruits ovoid, 22(19). Peduncle 10.1-13 mm thick. ......................... .5-3.5 cm long; scars of periant a ке 12. Manglietia M RE stamens along torus under fruit 9 mm lon 2. Peduncle 4-10 mm thick. ee 238. s d. Manglietia fordiana var. bwangtun- 23(22). Peduncle 4.1-10 mm НИЕ оза | gensis 23. Peduncle 4 mm | 5 24(23). Twigs in innovations 5.1-12 mm thick. .... 25 1 Ни Ма zi saspe jede 4. Twigs in innovations 2-5 mm thick. . 26 azod primarily on fuit characters) 25(24). Pedicle absent; twigs pubescent, stipules pu u- l. Stipules free; gynoecium stipitate. ................. bescent; leaves minutely (scattered) hairy 18. Ma anglietia sinica beneath, apex short-acuminate. o. | Stipules adnate to репоје; gynoecium not Manglietia eee stipitate 25. Pedicle present; twigs villas, stipules 2(1) Follicles hairy. 3 fous) villous; young leaves with longer hairs 2. Follicles glabrous beneath, apex acuminate or acute 3(2). Pedicle present; leaves narrowly elliptic or БООКЕ 12. anglietia megaphylla narrowly obovate. 26(24). Gynoecium hairy; hairs vhi follicles hairy. Pedicle absent; leaves elliptic or obovat 5 — . Manglietia ventii 4(3). Tree са. di m high; twigs yellowish se 26. Gynoecium glabrous; A own or re when you pei mm diam.; stipules pu- follicles ПА bescent; reticulation of leaf distinct; stipular 27(26). Gynoecium 13- 15 mm high; twigs villous, scars 9-11 mm long; peduncle hairy; flower 45 "Фа , stipules (rufous) villous; (greenish) а. outer tepals ovate or ob- ung leaves with longer hairs beneath, re long; stamens 12-20 mm long; gynoecium ticulation rather obscure, petiole not dilated 17-25 mm high. 16. Manglietia szechuanica my white, outer tepals 1 Treelet ca. 8 m high; twigs glabrous, 3 mm pire 3.7-5.7 x 2-2.5 cm, the inner diam.; stipules sparsely hairy; reticulation tals x n long; connective appendage of leaf rather obscure; stipular scars 2 ie rt ted or semiorbicular, 1-1.5 mm mm long; peduncle glabrous; flower purple long, Eine peduncles кота, ' fruits ovoid. or red; outer tepals obovate; stamens . Manglietia moto mm long; gynoecium 14-15 mm high 21. a m mm high; m pubescent, anglietia duclouxii ., stipules pubescent; leaves 5(3). Young twigs densely tomentellous, 2- ыы, ned hairy beneath, reticu- m thick; leaves not glaucous beneath. .. lation distinct, petiole dilated at base; flower . analista ventii purple or red, outer tepals thick-fleshy, 6— 9: 1 twigs sonde or rather long-hairy, cm, the inner ones slightly smaller; -10 mm thick; leaves glaucous iiie th. connective appendage triangular, 3 m 6(5). Tw wigs glabrous; leaves е rous, reticula- long; fruiting peduncles stout, fruits o tion distinct, fine; petiole ш ак oblong. ана 6. Mocha pom glabrous, 15 mm long 2 Родила (22- е -90 mm long. 2205050565056.606565—- 29 13. Manglietia microtricha Peduncle 10-30 E 6. Young twigs with rather long, undulate to 2928), Pedicle absent; Sa villous, ‘stipules (ru curly, spreading rufous hairs; leaves pu- fous) villous; young leaves with longer hairs nt beneath at least when young, re- eneath; inner tepals 9-12 cm, outer tepals ticulation rather obscure, coarse; petiole 3. 7-5.7 cm long. . nglietia moto hairy; peduncle hairy, ш mm long. 222 29. е Lir twigs ‘pubescent, “stipules ; Manglietia а pubesc (2). Pedicle absent. beneath: р tepals 2.7-4.5 cm, outer y Pedicle present. T tepals 4 ст long. о 5d. Man- 8(7). Fruiting но 11-14 mm thick. O ietia fordiana var. kwa anatumai : Fruiting peduncles Lum a mm thick. . ES 10 30(28). T А hairy; ge ease follicles 9(8). 17. Ma airy. nglietia ventii beneath, narrowly rien 10.5-13 cm Volume 80, Number 4 1993 Chen & Nooteboom 1033 Magnoliaceae of China 10(8). ~ 10 1110). — 12(10). 12. 13(12). 14(12). broad; petiole not dilated at base; fruits fruit 15-17 mm long; peduncle 30-34 mm ong, 13-14 mm thick; flower pink; carpels 95-110; acara of bracts: О, — 8. nglietia eee ung шен densely ee. n es not glau gs ‚6-95 m br bos petiole dilated at ie fruits NON a to ovoid, 8-8.5 x 5.2 cm; scars of perianth and stamens along torus under fruit 7 mm long; peduncle 12-20 mm long, 4-5 mm thick; flower white; carpels 29- 39; scars of bracts 1 ^ | Manglietia aromatica 11 Carpels 48-88. Carpels (12-)20 Young leaves ы hairy with long, dark rown hairs beneath, 11-19 x 5-7 cm; twigs and stipules brown to rufous villous; apex acuminate or mucronate; reticulation rather obscure; petiole not dilated at base; fruits ovoid ish) white or pointed Or o 1-1.5 mm long; gynoecium 13-15 mm high. . СИРИ Ма о moto Leaves minutely brownish hairy benea 18-34.5 x 8-12 cm; t gular, 3 mm long; gynoecium 30 mm 6. in o ids garrettii Fruiting peduncles 4.1-6 mm thick. 22 1: Fruiting peduncles 2-4 mm thick Leaves not glaucous, glabrous or minutely scattered hairy beneath; petiole not dilated at base, stipular scar 2-6 mm long; pe- duncle 4 mm thick; outer tepals 5.5-7 cm long, thin-coriaceous; inner tepals 4-4. | cm long; connective appen age 1.8 m e 5- m long; scars of langlietia fordiana var. fordiana on- endage 1 mm long; filaments 1 m long; scars of bracts 2. s 3. Manglietia ‹ dandyi Twigs 4-10 mm diam., usually glabro sometimes pubescent at the annular scars; nective appendage triangular, 1.8 mm long. 5a. Manglietia fordiana var. fordiana Twigs 4-5 mm diam., usually ferrugineous pubescent when young; а а of leaf fine, base narrowly cuneate; flower (green- 15(7). 15. 16(15). 16. 17(16). 18(15). 21(19). 22(21). ish) yellow or yellowish; outer tepals 3.5 cm long; inner tepals 2. m long; con- nective appendage semiorbicular, 1 -1.5 mm lon. 2. Manglietia conifera Fruits ovoid-globose, 9.5 x 8-10 cm or 9.5 x 6-8.5 cm. Fruits ovoid, ovoid-oblong, cylindric, ellip- soid, or oblong, 2.5-8 x 2.5-5.5 cm | Twigs glabrous to minutely hairy; fruiting eduncles 4.5 cm long; fruits 9.5 x 8-10 cm; ii Ani pubescent; leaves and petiole glabrou LL pun ж lucida Twigs dou and longer hairy when young. Leaves 23-40 duncles 10-30 x 10- fous) villous; leaves scattered villous with long undulate to curly dark brown eek beneath at least when young; petiole hai Iz. —— megaphylla Leaves 21-32 x 6.5-10 cm; frui x 6-6.5 cm broad; peduncle 30- A x E 10 mm; pedicle 7-13 “ mV" k 9. Manglietia hookeri mina p pin fruit ovoid to oid-oblong x 2. .9 ст; scars of bracts 2 Or сз 7. Manglietia glauca var. sumatrana pu not glaucous beneath; scars of bracts or 1. pd of perianth and stamens along torus under fruit 7-9 mm lon 20 Scars of Prae and stamens along torus under fruit 12-22 mm long. ‚ 21 : We hairy at Mal when young; ‘fruit ovoid or irre Eus ular -3.5 2.5 cm; tree of ca. iin de. pubescent: petiole м fruiting peduncles .5 cm long, 5 , 22-31 m long; flower (greenish) white; outer ies coria- us; connective appen ndage semiorbic lar, 1 mm long; gynoecium Fifa od styles 1-1.5 mm long. „ОЧ. Мелон fordiana var. со Twigs glabrous; fruits pe ellipsoid, ca. 3.5 cm long; treelet ca. 8 m high; stipules sparsely hairy; petiole and is а gla- ong; Finge hairy, styles mm 4. Manglietia бай Fruits Pius а ко or ellip 5 5 or 6.5 -4 cm; n па a mm long; ж dus 3-5 nective appendage triangular with a т pointed tip ог semiorbicular; twigs 3-8 о ca. 10 mm diam. 2 Fruits ovoid-globose or irregular, 2.5-6 2-5 cm; connective appendage glo n very long, narrowly triangular; twigs 3- mm diam. Fruit ellipsoid, 5-7 x 4.5 cm; stipules (rufous) villous; apex of mi short-acumi- nate, reticulation rather obscure; peduncle stout, 7 mm thick; outer tepals coriaceous; 1034 Annals of the Missouri Botanical Garden connective appendage semiorbicular; fila- ments 1 mm long. 9. ee RUM 22. Young twigs glabrous or pubesce nodes, 3-8 mm diam.; fruits ко oblong to ovoid- inis sometimes irregular, 6.5- 1 x = 4 cm; stipules glabrous or pubes- cent; apex of leaf acuminate ог mucronate, cM distinct; ‚ mm thick; outer tepals thin- coriaceous; connec- tive appendage triangular with a sharp pointed tip; filaments 2-5 mm lon 10. Manglietia insignis Young ~~ densely brown y^p 6 mm diam.; fruits ovoid, 4-5.5 x 3. 5- 5 cm; ui of leaf ви соппеспуе appendage very long, narrowly triangular; filaments 1 mm lon c. Manglietia fordiana var. forrestil 23. Fruits ovoid to ovoid-globose or irregular, ; -5 cm; reticulation of leaf rather obscure; connective appendage triangular; filaments 1.5- ng. Twigs glabrous, 5-7 mm diam.; base of leaf attenuate-cuneate; petiole dilated at base; peduncle 5-6 mm thick; pedicle 6-8 mm; gynoecium broadly P Li 16-20 mm high; c of bracts ч a toe fordiana var. calcarea 24. Young im glabrous or pubescent a nodes, 4-8 mm diam.; base of leaf cuneate; petiole not dilated at base; peduncle 4 m ves yan cle 0-1 mm; с ovoid, 13-15 mm high; scars of bracts 1. 222... .5a. Manglietia psu var. fordiana 23(21). x 24(23). II-1. Manglietia section Manglietia Gynoecium sessile. Stipules + adnate to petiole, at least in China. — ; вето aromatica Dandy, J. Вог. 69: 231.1931. Paramanglietia aromatica (Dan- dy) Hu & Cheng, Acta Phytotax. Sin. I: 255. 1951. TYPE: R. C. Ching 7421 (holotype, BM; isotypes, NY, PE). Figure 2. Tree to 20 m high and 50 cm diam., bark gray; young twigs at first densely appressed-pubescent with short, straight, gray or brownish hairs, soon glabrescent, stout, 7-10 mm diam., black-brown, later gray-brown; terminal buds ovoid, densely hairy, 2-2.4 cm long; stipules densely appressed-pubes- cent with rather long, straight, yellowish hairs, adnate to the petiole. Leaves glabrous, thinly co- riaceous to coriaceous, (narrowly) obovate, 17— 24.0 base cuneate; nerves in 9-16 pairs; midrib im- pressed above; reticulation coarse, prominent. Pet- iole glabrous, 2.7-3.5 cm long, conspicuously di- X 6-9.4 cm; apex mucronate or acuminate, lated toward the base, stipular scars 5-12 mm ong. Peduncles glabrous, ca. 12 x 4-5 mm, ped- 3-merous; stamens 1-1.2 cm long, filaments 1– 2 mm long, connective appendage semiorbicular; gy- noecium ellipsoid, ca. 16 x 10 mm; scars of peri- anth and stamens on the torus 7-10 x 6-10 mm. Fruiting peduncle 13-20 x 11 mm. Fruits subglo- bose to ovoid, ca. 8-8.5 x 5.2 cm; torus with scars of perianth and stamens 0.7-2 x l cm; carpels 29-39, with thick and ligneous wall, free when ripe, their dorsal faces 10-25 x 0.8-1.9 cm, adaxial parts 1-2 cm long, dehiscing along the dorsal and ventral suture; fruiting peduncles stout, 1.3-1.9 x 0.6-1.2 lipsoid, 10-12 x 7-8 mm. cm. Seeds broadly el- Distribution. China and northern Vietnam. In CHI- NA. Guangxi: Bako Shan, Poseh, near Yunnan border, .C. 7421; Lingle, Ling Le Exped. 32928; Don- i uang D.F. 1651. n, Wang 3 E. 87854; Shichou о Law Y.W. 407 74, Wu C.A. 6 7; Mar- li-po, ang-jin-in, Feng K.M. 13009; Si- cial sien, Faa- doou. 11990. In mixed forest. Altitude 1,300- 1,900 m. Trees in the natural forest (Jingping of Yunnan) reach a height of 20.4 m and a diameter of 40 cm by the age of 54 years. Flowering May- Ecology. June; fruiting September- October. 2. Manglietia conifera Dandy, J. Bot. 68: 205. 1930. TYPE: Fleury in Chevalier 37817 (ho- lotype, P; isotype, L). Figure 3. Manglietia chingii Dandy, J. Bot. 69: 232. 1931. TYPE: R. C. Ching 8390 (holotype, BM; isotypes, A, K, NY, P Manglietia tenuipes Dandy, J. Bot. 69: 232. 1931. TYPE: . Ching 7117 (holotype, BM; aig ee NY Manglietia glaucifolia Law, Guihaia 4(4): ide. To ovoidea Chang А unyatseni cane ou) 1: 108. 1988. : TYPE: B. 1. Chen Gs 86182 (holotype, SYS). Tree to 28 m high and 55 cm diam., black-brown and covered with many annular stip- twigs ular scars, 4-5 mm diam., usually densely ferru- gineously pubescent in innovations, soon glabres- cent. Stipules pubescent, adnate to the base of the petiole. Leaves thinly coriaceous to coriaceous, dark glossy green, glabrous above, pale green to + glaucous, glabrous, or sometimes appressed-brown- hairy when young, soon glabrescent, usually the Volume 80, Number 4 Chen & Nooteboom 1035 1993 Magnoliaceae of China > T iy ans D v Y | : | SIN Y o... 4 3 2 2cm FIGURE 2. Manglietia aromatica Dandy.— 1. Branch with deflorated flower.— 2. Flower.—3. Flower, outer tepals removed. — 4. Fruit. Based on SYS 161520 = Chen Bao Liang 86 S 109. Drawing by Xie Qing Jian. hairs microscopically small and the leaves seem- sulcate above, elevated below; nerves in 12-14(- ingly glabrous beneath; elliptic to (narrowly) ob- 17) pairs, meeting in a looped intramarginal vein; ovate, 12.3-19.6 x 4.3-6.4 cm; apex acute to reticulation fine, inconspicuous on both sides. Pet- (shortly) acuminate, base narrowly cuneate; midrib ¡ole pubescent, glabrescent, sulcate above, 1.2- 1036 Annals of the Missouri Botanical Garden FIGURE Liang GS 86-182. Drawing by Xie Qing Jian. 3. Manglietia conifera Dandy. — 1. Fruiting branch. — 2. Fruit. Based on SYS 161743 = Chen Bao Volume 80, Number 4 1993 Chen & Nooteboom Magnoliaceae of China 1037 1.5 em, stipular scars on petiole 3-7 mm long. Peduncle variable in length, 1.5-9 x 0.2-0.4 ст, erect or recurved, glabrous to = pubescent, pedicle absent, bract single, spathaceous, pubescent, di- rectly under the perianth. Tepals 9-11, the outer tepals 3, yellowish green to green, obovate, thinly coriaceous, ca. 3.5 X 2.4 cm, inner tepals 6-8, spathulate, fleshy, outside glabrous, yellowish with a flush of purple at the base, ca. 2.8 x 1.4 cm; stamens numerous with purple base, 9-11 mm, including 1-1.5-mm-long blunt connective ap- cm, carpels 20-28, glabrous, slightly connate when young, styles glabrous, 1-2.5 mm long; ovules 3-11 in each carpel. Ripe fruits ovoid to globose, 3.2-5.4 x 2.8-4.2 cm, the surface of the torus foveolate pendage; gynoecium ovoid, ca. 1. because of the dense scars of the fallen stamens, deeply depressed, scars of perianth and stamens са. 5— X 4—7 mm along the torus; fruiting peduncle 1.5-9 x 0.2-0.4 cm, the dorsal face of the carpels 16-22 x 7-9 mm, elliptic, opening along the dorsal suture only or both along dorsal and ventral suture. Seeds flat, subcordate, 7-9 x 6-8 mm. Distribution. China and Wane E Main In CHI- МА. Guangdong: Luoding, Wan 14; Ruyuan, Ko S.P. 52921; Sunyi Tai to Fat Youk Chong Ping, Ko S.P. 51777; Kung Ping Shan, and d Taan Faan, Mar: Cheng Dist., Tsang W.T. 26813. Guangxi: Na , Sup Man Ta Shan, Liang HY 69538; Yeo Mar 7. 7117; Me Kon, Seh Feng, Dar Shan, 14206. bei : Maguang, Chen B.L. 87 F 195; Meng- ze, Xu Ү.С. 5 Ecology. Locally fairly common, in mixed ev- ergreen forests, on sandshale hills. Altitude: 500— 1,700 m. Flowering May-June; fruiting Septem- ber-October. ‘ses. Wood used for general construction work and furniture, bark used for medicine. Also grown as an ornamental. Note. in all characters, but the variation in length of the The specimens examined are uniform peduncles is considerable. There are, however, in- termediate forms. We were not able to see the type of Manglietia glaucifolia Law. Fortunately, Law's original de- scription gives enough information to enable us to recognize it. Consideration of the diagnostic char- acters and the geographic distribution leads to the conclusion that Manglietia glaucifolia belongs to Manglietia conifera. 3. Manglietia dandyi (Gagnep.) Dandy in Prag- lowski, World Pollen Spore Fl., 3 (Magnoli- aceae): 5. 1974. orare dandyi Gagn Notulae Syst. 8 207 14 (holotype, P). = 4. ер., TYPE: Poilane Tree to 15 m high, ca. 18 cm diam.; bark gray, Tue rough; the indument with long, undulate to curly, spreading, rufous hairs; young twigs 6-7 mm diam., at first densely pubescent, finally gla- brescent, old ones gray to brown-gray, = pubescent to glabrous. Stipules densely pubescent, adnate to the petiole, scars 6-10 mm long. Leaves mem- braneous to thinly coriaceous, dark green, slightly pubescent, especially along the midrib above, glau- cous, at first densely pubescent, sooner glabrescent 16- 8.5 cm; apex short-acuminate, acumen beneath, obovate, elliptic to obovate-elliptic, 24 X 5- 5-10 mm long, base cuneate to obtuse; midrib slightly impressed above, nerves obscure above, conspicuous below, in 8-13 pairs, reticulation laxly netted, not prominent on both sides. Petiole densely pubescent, dilatate toward the base, 1.2-2.3 cm ong. Peduncle pubescent, ca. 1 x 0.5 cm, pedicle absent; bracts 2, densely pubescent outside. Flower bud ovoid; tepals 9-11, subequal, white to yellow- ish, the outer 3 fleshy, obovate-oblong, pubescent outside at the base, 2-2.2 x 1.5-1.7 cm, the inner tepals 6-8, obovate, glabrous, often abruptly constricted near the base, 1.8-2 x 0.9-1.6 cm; stamens numerous, sparsely pubescent outside, 5.5— 7 mm long, connective appendage triangular, ca. l mm long, filaments ca. 1 mm long; gynoecium ovoid, 10-13 mm long, carpels 24-30, glabrous to pubescent, ovules 2-10; scars of perianth and stamens along torus ca. 6 mm long in flower. Fruit- ing peduncles 2-4.5 x 0.5 cm. Fruits ovoid to cylindric, 6-7.5 x 1.8-2.5 cm, entirely connate when young; mature carpels glabrous or + pu- bescent, dorsal faces 1-1.6 cm long in upper and 2-3 cm long in lower carpels, dehiscing along the dorsal suture; scars of perianth and stamens ca. X 0.5 cm. Seeds orange-red, 1-2 in each carpel. Distribution. China, Laos, and Vietnam. In CHINA. Guangdong ао га B.L. 80115; Yin d hse Hb и Sin S. 5 996 Cangwu Xian, Chun S.H. 10140; He Xian, Cun LE 500183. Yunnan: Marlipo, Sze-tai-po, Feng K.M. 13960; Menglun, Chang H.T. 6342, Si-chour-hsien, Faa-doou, Feng K.M. 11771; Wenshan, Gao T.X. 5; Xichou, Zhu DOS. Ecology. In evergreen broad-leaved forest. Altitude: 1,450-2,000 m. Flowering April; fruiting September—October. 1038 Annals of the Missouri Botanical Garden FIGURE 4. flower. — 3. Stam um, Leiden аат dandyi (Gagnep.) Dandy Collector's note. Fruits orange-red, usually yellow-greenish tinged with a little red when mature and becoming brown when dry. Notes. C. Wu & W. T. Wang (1957) misidentified this species as Manglietia rufibar- bata non Dandy (1928). The latter species is from Vietnam and has not been recorded for China. See also under Manglietia moto. Mlessenaerp о! .— 1. Fruiting branch (Feng 13960, Р).—2. Partly deflorated n. — 4. Outer petal. — 5. Inner nel (Poilane 207 14, P). Drawing by Joop Wessendorp, Rijksherbari- The species has some noteworthy characters. (1) The base of the lowest carpels is sometimes decurrent along the torus, so a short gynophore seemingly occurs. (2) Sometimes there are three ovules in the lowe Manglietia. (3) The carpels are from glabrous to hairy throughout the area of its dis- Volume 80, Number 4 1993 Chen & Nooteboom 1039 Magnoliaceae of China tribution. The mature fruits as seen in the field often become greenish yellow instead of purple as is common in fruits of Magnoliaceae. ~ . Manglietia duclouxii Finet & Lor. Bull. Soc. Bot. France (Mémoires) 4: 33, t. 5A. 1905. Magnolia duclouxii (Finet & Gagne ) Hu in Hu & Chun, Icon. Pl. Sin. 2: 18, 68S. 1929. TYPE: p 2133 a selected here, P). Tree of ca. 8 m, young twigs slender, ca. 3 mm diam., glabrous, yellow-brown, gray-black later. Stipules sparsely hairy. Leaves thinly coriaceous to coriaceous, glabrous, dark green above, pale green, glabrous or scattered appressed-pubescent with fine, rather short, straight, brown hairs, with long colorless hairs along the margin and the midrib eneath, narrowly obovate to elliptic, 10.2-16 x 3-4 cm; apex acuminate, base cuneate; nerves fine in 9-11 pairs; reticulation netted, inconspic- uous. Petiole glabrous, lengthwise grooved above, 1–1.2 cm long, with а stipular scar 2-3 mm long. Peduncle glabrous, 1.3-1.5 x 0.3-0.4 cm, bract 1, pedicle 2-3 mm long. Tepals 9, usually outer ones larger than inner ones, fleshy, purple or red, obovate to broadly obovate, 2.8-3.5 x 1.5 cm; stamens ca. ] mm, including 2-3-mm- long connective appendage and 1-mm-long fila- ments; gynoecium ellipsoid, 1.4-1.5 x 0.7-0.8 cm, carpels 45-55, narrowly elliptic, ferrugineous- hairy, ca. 7-8 mm; styles glabrous, ca. 3-4 mm long; ovules ca. 5 in each carpel. Fruits ovoid- ellipsoid, ca. 3.5 cm long (not seen). Scars of peri- anth and stamens 7-9 x 3-4 mm Eon and China. In CHIN 430. Sichuan: 2 b» Xian 1114; Culin, Gulin Yiyao Gongsi 846. Yun- nan: Lo ong ky, Maire E.E. s.n.; Malipo Xian, KUN 76298; Wenshan, Feng K.M. 22369 Distribution. Ecology. In evergreen forest. Altitude: 700— 1,800 m. Flowering May-June; fruiting Septem- ber-October. Note. Manglietia duclouxii is easily recog- nized by its slender twigs, thin, small leaves, which are usually covered with minute, short, and rather long hairs beneath, and the pubescent gynoecium. 5. Manglietia fordiana Oliver in Hook., Icon. Pl. 10: t. 1953. 1891. Magnolia fordiana (Oliver) Hu, J. Arnold Arbor. 5: 228. 1924 TYPE: C. Ford 90 (holotype, K). шан microcarpa Chang, Acta Sci. Nat. Univ. unyatseni (Guangzhou) 1: 53. 1961. TYPE: S. К. Lau 4246 (holotype, SYS; isotype, A). Manglietia hainanensis Dandy, J. Bot. 68: 204. 1930. : Tsang & Fung 656 in Herb. Lingn. Univ. 18190 (holotype, BM; isotypes, A, K, MO, NY). Tu yuyuanensis Law, Bull. Bot. Res. (China) 5, 125. 1985. TYPE: China. Zhejiang: Changhua Xan alt. 1,000 m, X. Y. Hoo 23326 (IBSC not een). 5a Manglietia fordiana var. fordiana. Tree to 30 m high and 1.5 m diam.; young twigs 4-8(-10) mm diam., yellow-brown to dark brown, usually glabrous, sometimes pubescent only along the annular scars of the stipules. Stipules usually + pubescent with ferrugineous hairs, ad- nate to the base of petiole. Leaves thinly coriaceous to coriaceous, both sides glabrous or minutely scat- tered-pubescent beneath, narrowly obovate, ob- ovate-elliptic to narrowly elliptic, 8-16(-20) x —5.4(—7) cm; apex mucronate to acuminate, base cuneate, sometimes slightly decurrent into the petiole; midrib impressed above, elevated and gla- brous or hairy below, nerves in 8-12(-17) pairs, reticulation coarse, hardly visible. Petiole glabrous or pubescent, sulcate on the upper side, 1.4-4.2 cm long, stipular scar glabrous or pubescent, 2-6 mm long. Peduncles glabrous or ferrugineous-pu- bescent, 8-1 mm, pedicles absent or to 1 cm long; bracts 1—2. Tepals 9, the outer 3 slightly larger than the inner ones, the tepals concave, thinly coriaceous, obovate-elliptic, greenish out- side, hardly pubescent outside at the base, 5.5-7 x 2.5-4 cm, the inner tepals pure white, fleshy, spathulate to broadly obovate, with cuneate to short- clawed base, 4-4.8 x 1.5-2.2 cm; stamens nu- merous, 9-12 mm long, connective appendage short ng, filaments glabrous, 1.5-2 mm long; gynoecium ovoid to subglobose, 5 х triangular, ca. 1.8 mm lo glabrous and smooth, 1.3-1 -1.2 cm, car- pels (12-)24-36, styles glabrous, 0.5-1 mm long. Fruits ovoid or sometimes irregular-shaped because the carpels are partly abortive, 2.5-6 x 2-3.5(- ) cm; fruiting peduncle 1-2.5 x 0.4-0.6 cm, pedicle 1-5(-9) mm long; ripe carpels sometimes only 4-10, torus with scars of perianth and stamens 1.3-1.7 x 0.4-0.5 cm. Seeds flat-ovate, ellipsoid or irregularly shaped, 6-10 x 4-7 mm Distribution. China and bee In CHINA. An- hui: Wangshan, Cheng W.C. 4030. Fujian: Chung H.H. 2940. Guangdong: Loh Fian mountain, Funk- houser W.L. C.C.C. 9082 a; Lok Chong, Tso C.L. 21005; uh Fau Mountain, Merrill E.D. 10904; Ruyuan, Chen 1040 Annals of the Missouri Botanical Garden B.L. 315; Sunyi Dist., Wang C. 31135; Tsengshing 401535; Lungnan Xian, Lau S.K. 4850; Pingnan, Wang C. 39230. Hainan: Bak Sa, Lau S.K. 25956; Dai Land Dungta, Chun N.K. & C.L. Tso 43956; Dung Ka to Wen Fa Shi, Chun N.K. & C.L. Tso 43687; Fan ya to Yik Tsok Mau, McClure F.A. C.C.C. 9652; Jianfong, Chang H.T. 6095. Zhejiang: King Yuan, Ching R.C. 2472. HONG KONG. Tai Mo Shan: Chan K.Y. 1033; Ma On Shan, Hu S. У. 10169 A. Also in Yunnan, Guizhou, Jiangzi, Hunan Ecology. In subtropical evergreen broad- leaved forest; usually the trees grow on granite and sandshale. It was reported that this tree was easily cultivated and grew fast. It had attained a height m and a diameter of ca. 23 cm after 28 years at Xinyi Xian in Guangdong Province. Al- titude: 500-1,300 m. Flowering April-May; fruit- ing August- October. Uses. Because of the nicely structured and medium-hard wood, the timber is used for house building, veneer, furniture, and musical instru- ments. The local people in Guangxi treat dry cough and constipation with the bark. The tree is also cultivated as an ornamental. Collector's note. The bark is brownish gray, smooth; branches are dark gray; leaves are glossy dark green above and pale green below; the flower buds are solitary, terminal, ovoid, flowers are creamy white, fragrant, anthers red; young fruits are pur- plish; seed coat is orange. Note. |. Manglietia hainanensis is conspecific with Manglietia fordiana. It differs from the latter only in the length of the pedicle and the texture of the leaves. Although the pedicle usually is absent in Manglietia hainanensis, it sometimes occurs and then varies in length from 0 to 5 mm, occca- sionally to 7-9 mm (H. Y. Liang 62608) through- out its geographical range. The leaves of Man- glietia hainanensis are generally thinner than those of Manglietia fordiana. 'This difference is not a reliable character, however, because the leaves of Manglietia fordiana show a continuous variation in texture. Thus, there is no character left to dis- tinguish Manglietia hainanensis from Manglietia fordiana. there is no doubt that it is conspecific with Man- Regarding Manglietia yuyuanensis, glietia fordiana, despite its glabrous leaves. The degree of hairiness varies in Manglietia fordiana, too. Manglietia yuyuanensis represents an ex- treme, glabrous form. Manglietia fordiana and all synonyms mentioned above possess a leaf hypo- dermis with sinuous anticlinal walls and + subglo- bose to ovoid sclereids, not known in other species of Manglietia in China. Manglietia globosa and Paramanglietia microcarpa are easily recogniz- able as Manglietia fordiana. 5b. Manglietia fordiana var. calcarea (X. H Song) B. L. Chen & Nooteboom, stat. nov. Manglietia calcarea X. H. Song, J. Nanjing Inst. Forest. 4: 46. 1984. TYPE: China. Gui- zhou: Mogan, Libo Xian, alt. 670 m, X. H. Song & F. Gao 640 (NF not seen). Tree to 14 m high and 35 cm diam.; young twigs 5-7 mm diam.; plant entirely glabrous except for the stipules. Stipules densely pubescent with straight, appressed, brown hairs, adnate to the very base of the petiole, scars 4-5 mm long. Leaves coriaceous, rigid, dark green above, obovate-ellip- tic to narrowly obovate, 14.5-20 x 3.5-7 cm; apex rounded with an acumen 6-10(- 1 4) mm long, base attenuate-cuneate; midrib impressed above, elevated beneath, nerves obscure above, visible to hardly prominent below, in 14-17 pairs. Petiole 2.8-3.8 cm long, dilatate toward the base. Pe- duncle 5-8(-20) x 5-6 mm, pedicle 6-8 mm long. Tepals 9, white, the outer 3 obovate-elliptic, са. 6.2 х 2.5 ст, apex rounded, the inner tepals spathulate, 4.5-5.5(-6) cm long; stamens many, 1.2 ст long, connective appendage triangular, ca. 1.8 mm long, filaments ca. 2 mm long; gynoecium broadly ellipsoid, ca. 1.6-2 х 1 cm, carpels 12- 16(-30), perianth and stamens 8-14 x torus with the scars of 5-6 mm. Fruiting peduncle ca. 9-11 x 6 mm. Fruit ovoid to ovoid- ovules ca. 7; globose, ca. 6 Х 5 cm, пре carpels ligneous, dorsal face elliptic, dehiscing along the dorsal venture; scars of perianth and stamens 1.2-1.3 x 0.7-0.8 cm. Seeds oblong, 8-10 x 5-6 mm, 2-5 in each carpel. Distribution. CHINA. Guizhou: Pinfa, |“ Ј. 3182; Хиппап-зеп Lied Cavalerie J. 226. Ecology. | In margin of forest, growing on lime- stone. Altitude: 600-800 m. Flowering April-May; fruiting August-Septem Ises. Good timber, bond flowers; the tree also cultivated as an ornamental. Note. This variety is characterized by both twigs and leaves being glabrous, the gynoecium broadly ellipsoid, and the carpels narrowly elliptic. 5c. Manglietia fordiana var. forrestii (W. W. Smith ex Dandy) B. L. Chen & Noot., stat. ov. Manglietia forrestii W. W. Smith ex Dandy, Notes Roy. Bot. Gard. Edinburgh 16: Volume 80, Number 4 1993 Chen & Nooteboom Magnoliaceae of China 1041 126. 1928. TYPE: G. Forrest 26705 A (ho- lotype, E; isotypes, A, K, NY). Manglietia globosa овор, Acta Sci. Nat. Univ. Suny- atseni (Guangzhou) 1: 53. 1961. ТУРЕ: R. C. Ching 8394 (holotype, SYS; isotype, NY). Tree to 18 m high, 46 cm diam.; stipules, young twigs, petiole, peduncles, and outer tepals outside at the base densely brown-pubescent with short to long, straight, appressed hairs; young twigs brown- ish, pubescent with straight to curly hairs; leaves and midrib scattered-pubescent with short, straight hairs beneath. Stipules adnate to the base of the petiole. iam., old ones dull brown, gray Leaves coriaceous, bright green, glabrous above, pale green below, usually obovate, rarely narrowly obovate, 13-24.5 x 5.4-9.0 cm; apex mucronate to abruptly acuminate, acumen 5-17 mm long, base cuneate; nerves fine, in 9-14 pairs; reticu- lation distinct beneath, fine or coarse. Petiole sul- cate on upper face, dilated toward the base, 1.5- 2.6 cm long, stipular scars 2-8 mm long. Pedun- cles stout, 1.4-2.6 x long. Flowers pure white, fragrant; tepals 9(—10), 0.4-1 cm, pedicles 2-5 mm outer 3 thinly coriaceous, obovate to oblong-ob- ovate, brown-pubescent outside at the base, 6.3- 7 X 3.4-4.8 cm, the inner tepals 6, fleshy, spath- ulate, obovate to + subcircular, 4-5.3 x 2 connective appendage obtuse, filaments ca. 1 mm long; gynoecium ovoid, smooth, 1.2-2.2 x 1-1.4 cm, carpels 26-28, styles 1-2 mm long. Fruiting peduncles, 1.7-2.6 x 0.6-1 cm, pedicles 2-8 mm long. Fruits ovoid, 4-5.5 x 3.5-5 cm, dorsal faces of carpels 1.5-2 cm long in the upper to 3.5-4 cm long in the lower carpels, carpels densely clawed at the base, -3.5 cm; stamens 1-1.5 cm long, papilliferous, with short beak, dehiscing along dor- sal and ventral suture; torus with scars of perianth and stamens 1.3-2.2 x 0.5-0.9 cm Distribution. China and Annam. In CHINA. South- western Guangxi: Chu Fang Shan, 30 Li S W of Shan Fang, Wang C.W. 39301, Ching R.C. 5910. Southern Yunnan: Cangyuan, Li S.P. Mu 199; hills NW of Ten- gyueh, Forrest G. 27300; Nan-Chiao, Wang C.W. 73474; Szemeo, Forrest G. 11988; em Feng K.M. 5193; Si-chour-hsien, Ting-mann, 12254 Ecology. Та mixed broad-leaved or deciduous forests. Altitude: 540-2,400 m. Flowering June; fruiting September- October. Collector's notes. Widely branched; flowers green-white or pure soft white, fleshy, fragrant, anthers purple; fruits creamy reddish. ses. e timber is used for doors, windows, and furniture. Note. diana in young twigs, stipules, undersides of the leaves, petioles, peduncles, and the outer tepals outside at the base being brown pubescent; the This variety differs from variety for- leaves are larger, to 24.5 cm long and 9.6 cm wide. 5d. Manglietia fordiana var. kwangtungen- sis (Merr.) B. L. Chen € Nooteboom, stat. nov. Manglietia kwangtungensis (Merr Dandy, Kew Bull. 1927: 264. 1927. | nolia kwangtungensis Merr., bor. 8: 5. 1927. TYPE: K. P. To, & U. K. Tsang 345 in C. C. С. 12344 (ho- lotype; À; isotypes, BM, E, MO, SYS, UC). Tree 15 m high; young twigs, stipules, midrib beneath, petiole and peduncles densely pubescent with short to long, straight to undulate, rufous hairs; young twigs 4 4-5 mm diam. Leaves dark green glossy above, pale green, scattered ap- pressed-pubescent with short, straight, rufous hairs below, coriaceous, (narrowly) obovate, 14.4-19 x 3.2-6.4 cm; apex (shortly) acuminate, base nar- rowly cuneate to cuneate; nerves fine in 9-13 pairs, reticulation inconspicuous on both sides. Pet- iole sulcate above, 1.3-2.2 cm long, stipular scars ] x 0.4 cm, with only one bract scar near the base of perianth, -5 mm long. Peduncles 2.2-3. pedicle 1-2 mm long. Tepals 9, the outer 3 ob- ovate-oblong, coriaceous, greenish with a flush of purple, + brown hairy outside at base, 4-4.5 х 2-2.5 cm, the inner tepals 6, white, spathulate to obovate, short-clawed at base, thick, fleshy, 2.7— 4.5 x 0.9-2.3 cm; stamens 0.7-1.5 cm long, connective appendage blunt, ca. 1 mm long, fila- ments glabrous, ca. 1 mm long; gynoecium broadly ovoid, 1.1-1.5 x 0.9-1.1 cm, сагре 44—49, styles 1-1.5 mm long, scars of perianth and sta- mens along the torus 0.9-1.2 x 0.4 cm. Fruits ovoid to irregularly shaped because partly abortive, .5 X 2.5 cm; fruiting peduncles 3-3.5 х 0.5 cm, pedicles ca. 5 mm long; ripe carpels 7- 10 mm long. Guangdong: Ruyu Distribution. CHINA. C. 12344. Guangxi: Damiaoshan, Chen не Ки T.C. 617. Ecology. In mixed forest. Altitude: 700— 1,000 m. Flowering May- June. ote. Because of its rufous or red-brown in- dumentum, it is not too difficult to distinguish Man- glietia fordiana var. kwangtungensis from vari- eties fordiana and forrestii. In addition, variety Annals of the Missouri Botanical Garden kwangtungensis shows smaller flowers and many more carpels. The collection C.C.C. 12179, cited under Mag- nolia kwangtungensis Merr. (1927), belongs to Manglietia dand yi (Gagnep.) Dandy. 6. Manglietia garrettii Craib, Kew Bull. 1922: 166. 1922. TYPE: Garrett 114 (holotype, UT; isotypes, BM, K). Tree to 20 m high and 1.2 m diam. Young twigs, stipules, petioles and peduncles densely brown pubescent with short to long, straight to slightly undulate hairs. Twigs brown, stout, 5-8 mm diam Stipules adnate to the petioles, scars 1.8-2.5 ст long. Leaves densely coated with minute (even with hand lens hardly visible), brownish hairs beneath, coriaceous, (narrowly) elliptic to obovate, 18–34.5 x 8-12 or rounded; nerves in 12-21 elevated beneath, reticulation prominent on both cm; apex short-acuminate, base cuneate airs, prominently sides, coarse; petiole 3-5 cm long, thickened to- 1 X 6-9 mm, pedicle absent. Flower buds elongate, 6-6.5 x 2.2-2.5 cm. Tepals 9, purple or violet-red, thick x З cm, 6 inner ones slightly smaller; stamens 1.1-1.5 ст and succulent, 3 outer ones 6-6.5 long, connective appendage triangular, ca. 3 mm long; gynoecium glabrous, ovoid to narrowly ovoid, cm, carpels 66-88; styles glabrous, 2— 3 mm long. Fruits broadly ovoid-oblong, 4-9 x 3.5-6 cm; dorsal faces of ripe carpels 1-2 cm long in upper and 2.5-3.2 cm long in lower carpels, short-beaked. Scars of perianth and stamens 10- 12 x 9 mm. Seeds ellipsoid, obovoid, and irreg- ularly shaped, 9-14 x 5-8 mm. Distribution. China (southern Yunnan), Thailand, and ‘ietnam 7. Manglietia glauca Blume var. sumatrana (Miq.) Dandy, Kew Bull. 1928: 188. 1928. Manglietia sumatrana Miq., Fl. Ned. Ind. Suppl.: 367. 1861. TYPE: Tijsmann 168 (ho- lotype, L; isotype, ВО). di higos САР P. Parm., Bull. Sci. France Belgique , 292. na Manglietia аа nsis ae Att nay ccad. Fisiocrit. Siena 9 (7): 1926. i ae PS 334 с ЕІ; isotype, TE Tree to 26 m high and 80-100 cm diam.; bark gray-brown; twigs green, 5-6 mm diam., appressed short-hairy. Stipules glabrous or with some hairs on the apex, adnate to the very base of the petiole. Leaves appressed-puberulent, with short to long, straight, colorless to brownish hairs, glaucous be- neath, thinly coriaceous, obovate to narrowly ob- ovate or narrowly elliptic, 10-35 x 3.5-12 cm; apex acuminate, acumen 3-15 mm long, base cu- neate; nerves in 10-16 pairs, prominent below, reticulations laxly netted, visible on both sides. Petiole covered with the same indumentum as twigs, lengthwise shallowly grooved, 1.5-3 cm long, stip- ular scars never over Уз of the length of the petiole. Peduncles minutely pubescent only at the nodes, .9-4 cm long, pedicles 2-8 mm long. Flowers white, the outer 3 tepals thinly coriaceous, obovate- elliptic, greenish outside, 5-7 .: cm, the inner 6 tepals smaller than the outer ones; stamens numerous, 0.6–1.3 cmlong, connective appendage very acute, ca. 4 mm long, filaments ca. 3 mm long; gynoecium ovoid to orbicular, ca. 15 х 7 mm, carpels 20-50. Fruits ovoid to ovoid-oblong, 3.5-8 x 2.5-5.5 cm; the carpels opening ven- trally and then dorsally, after staying closed toward the basal side of the dorsal suture. Seeds flat-or- bicular to ellipsoidal, often irregularly shaped, 5- Distribution. | Native of Indonesia јесен Sumatra), cultivated in Guangdong, Guangsi, and Fuj Ecology. Altitude 600-1,300 m (in Indonesia). Flowering ад April; fruiting ina) In forests. September-October (in С /ses. Cultivated as an ornamental tree Note. i aw in W. C. Cheng (1983) and L. G. Lin (1985) nen this variety as Manglietia glauca Blume. 8. Manglietia grandis Hu & Cheng, Acta Phy- totax. Sin. 1: 158. 1951. TYPE: C. W. Wang 87020 (holotype, PE). Figure 5. Tree 12 m high and 35 em diam.; plant glabrous except the sometimes puberulent undersides of leaves; young twigs stout, 8- 10 mm diam., usually covered with a waxy substance, so seemingly gray; stipules adnate to the base of the petioles. Leaves glabrous, glossy above, glaucous, glabrous or mi- nutely scattered-puberulent, + papilliferous be- neath, coriaceous, narrowly obovate, 20-35.5 x 10.5-13 cm; apex acuminate, base cuneate; nerves 1 7-26 pairs, reticulation Visible when dry. Petiole 2.6–4 cm long, stipular scars 5-8 mm long. Flower pink. Fruiting peduncle 3- 3. 4 x 1.3-1.4 cm, pedicle absent. Fruit oblong-ovoid, 10-16 x 6- 10 cm; carpels 97- the dorsal and the ventral suture, beaks sharp and 106, 3-4 cm, dehiscing along slightly recurved, 3-4 mm long. Scars of perianth and stamens along the torus 1.5-1.7 cm long. Volume 80, Number 4 Chen & Nooteboom 1043 1993 Magnoliaceae of China 2 FIGURE 5. Manglietia grandis Hu & Cheng.— 1. Branch with flower bud. — 2. Bud with bracts. — 3. Fruit, one follicle enlarged. Based on SYS 161411 = Chen Bao Liang 86 S 297. Drawing by Xie Qing Jian. Distribution. CHINA. Southeastern Yunnan: Маг Uses. Wood straight-grained, fine-textured, Li Po, Wang C.W. 87020. very good for house building and furniture because of its durability. Ecology. Їп subtropical forests, on sandy soil Note. Manglietia grandis differs from other and limestone. Altitude: 800—1,800 m. species by its much larger leaves, numerous car- 1044 Annals of the Missouri Botanical Garden pels, and bigger fruits; the fruits are the largest in Chinese Manglietias so far as we know. We have not seen the mature flowers and therefore cannot describe them in detail. 9. Manglietia hookeri Cubitt & W. W. Smith, Records Bot. Surv. India 4: 273. 1913. Mag- nolia hookeri (Cubitt & W. W. Smith) Raju & Nayar, Indian J. Bot. 3, 2: 170. 1980. SYNTYPES: Silum and other stations in the Bha- mo division of Upper Burma: Cubitt 3024, 327 (CAL not seen), Rodgers 314 (CAL). Tree to 25 m high, 90 cm diam.; young twigs, stipules, petiole, and peduncles covered with an indument of straight, white, silver, to brownish appressed hairs. Twigs brown, stout, (4— mm diam. Stipules adnate to the base of the petiole. Leaves glabrous on both sides, coriaceous, narrowly obovate (or elliptic), 21-32 x 6.5-10 cm; apex acuminate, base cuneate, + decurrent along the petiole; nerves in 12-19 pairs, reticulation prom- inent when dry. Petiole conspicuously sulcate above, 2-3.5 cm long, stipular scar obvious, nearly to the base of the blade, usually 1.2-2 cm long. Peduncles stout, 1.5-3.7 x 0.5-0.7 cm, pedicles 0.7-1.3 cm long. Flowers white, ca. 10 ст diam.; tepals 9-12, the outer 3 thinly coriaceous, obovate, 7— 10 x 3.5-4 cm, creamy above the middle, be- coming greenish toward the base, the inner tepals 6-9, thick, fleshy, spathulate to obovate, 7-8.5 x 3-3.5 ст, narrowly and short-clawed at base; stamens 1.7-2.2 mm long, connective appendage triangular with a sharp-pointed tip, 1-2 mm long, filaments 3-5 mm long. Gynoecium ovoid to broad- ellipsoid, 1.8-2.3 x 1.3-1.5 cm; carpels 50-64, reddish brown when dry, styles black-brown, 4-7 mm long. Scars of fallen perianth and stamens 1.6– 2 x 0.6-0.7 cm. Fruiting peduncles 1.7-4 x 0.7-1 cm, pedicles 0.9-1.5 ст long. Fruits ovoid to subglobose, 7-9.5 x 6-6.5 cm; dorsal face of the ripe carpels from 1-1.3 cm long in the upper to 2.3-3 cm long in the lower carpels, short-beaked, smooth, dehiscing dorsally. Seeds ovate, elliptic- ovoid, elliptic to irregularly shaped, 8-13 x 5-7 mm. Distribution. China and Upper er hs CHINA South- "Forres G. 15952; Chen-Kang Hsien, Wang C.W. 7 ; Hills 3 days S of Tengyueh, 24%40'N, 98*30'E, Forrest G. 27364; Jingdong, Li M.K. 3515; Menghai, Li Y.H. 4068; Tung- kwong valley, 25°5'№, Forrest - 7725; Shuning, Yu T.T. 16559; Tengchong, Li W.Z. 8 Ecology. In mixed forests. Altitude: 1,000- 3,000 m. Flowering April-May; fruiting October. Collector's note. The flower buds green, flow- ers white, purple-white, deep pink fleshy; fruit fol- licles conelike, green or black. 10. Manglietia insignis (Wall.) Blume, Fl. Ja- vae: Magnol. 828. Magnolia insignis Wall., Tent. Fl. Nepal.: 3, t. 1. 1824. TYPE: Wallich 973 (holotype, K; isotypes, BM, GH). н а Hu, Acta Phytotax. Sin. 1: E: H. T. Tsai 56560, p.p. (holotype, PE; iso ен A. Manglietia yunnananats Hu, Acta Phytotax. Sin. 1: 159. 1951. TYPE: Н. T. Tsai 56560, p.p. (holotype, че Ыз ау ie Hu, Acta Phytotax. Sin. 1: : H. C. Chow 484 (holotype, ed iso- iin P" Е, Manglietia maguanica Chang & B. L. Chen, Acta Sci. Nat. Univ. Sunyatseni (Guangzhou) 1: 109. 1988. TYPE B. L. Che SYS). Manglietia d Chang & B. L. Chen, Acta Sci. Na v. Sunyatseni (Guangzhou) 1: 110. 1988. TYPE с TL Chen & Li 86 S 195 (holotype, SYS). n & Y. H. Su 86 S 053 (holotype, Tree to 25 m high and 50 cm diam.; young twigs glabrous or ferrugineous to yellowish pubes- cent at the node, usually 3-8 mm diam. Stipules glabrous, sometimes pubescent, adnate to the base of the petiole, 19-27 mm long. Leaves thinly co- riaceous to coriaceous, shining green above, pale green, glabrous to minutely pubescent beneath, narrowly obovate to elliptic, 14.5-26.5(-30) x 4.3-7.8 cm; apex acuminate to mucronate, acu- men 5-15 mm long; base cuneate, = attenuate along the petiole; nerves fine, in 10-22 pairs, re- ticulation densely netted, prominent on both sur- faces. Petiole sulcate above, 1.7-3.5 ular scars glabrous or sometimes scattered-hairy, cm long, stip- 4-10(-13) mm long. Peduncles glabrous to pu- bescent only near the scar of the fallen bract, 1- 5 X 0.4 cm, pedicles 3-5 mm long. Flowers fragrant, tepals 9-12, the outer З obovate-oblong, x 2-2.5 cm, thinly coriaceous, brown-green- ish, sometimes white-hairy outside at the base, re- flexed later, the inner 6-9 white to purple, spathu- late to narrowly obovate, short-clawed at base, 5- 6.5 x 1.3-2.5 cm; stamens 1.3-1.6 cm long, connective appendage triangular with a sharply pointed tip, 1-2 mm long, filaments glabrous, 2- 5 mm long; gynoecium ovoid to ovoid-cylindric, 1.7-2.8 x 1-1.2 cm, carpels 50-86, glabrous, styles glabrous, 2-4 mm long. Fruiting diam 1.2-2.5 x T7 . .5-] ст, pedicles 0.5-1.1 cm long. Fruits ovoid-oblong, sometimes sed. shaped because part of carpels abortive, 6.5-11 x 3-4 cm, the scars of perianth and stamens along the torus 1.4-2 x 0.6-0.7 cm; ripe carpels 0.7-1.2 Volume 80, Number 4 1993 Chen & Nooteboom 1045 Magnoliaceae of China cm long in the upper and 1-2 cm long in the lower carpels, dehiscing along the dorsal suture, obvi- ously papilliferous, apex short-beaked. Distribution. Nepal, northeastern India, China, northern Burma, northern prn In CHINA. Guang- xi: Damiaoshan, Chun S.H. 15857; He Xian, Chen Z.Z. 52125; Lingui, 51153; Longsheng, 51134; Quan Xian, . 81657. Guizhou: Kaili Xian, Qiannan Exped. 2081; m silva frondosa collis ad vic. Yudong, Hunan: Xinning Ziyushan, Li Z. Y. 30703; Yongshun, Hunan : .C. 484. Sichuan: SAN Xian, Chen S.J. e dues hsien, Liu Y.S. 1; Miyi Xian, Wu S.K. 85. Xizang: Mt. а eastern and western ы, and pons divide, Rock J.F. 21991; Tsarong, Forrest G. 18964; Shweli-Salwin divide, N of Ho-tou, 26738. Yunnan: K'raikha-Salween divide, Forrest G. 1837 1; Salween- jt kong divide, 14402; Shangpa Hsien, Tsai H.T. 565 Shunning Hila, Wumulung, Yu T.T. 16727; o valley 25°N, Forrest G. 8678; Wen shan Hsien, Tsai Н.Т. 51702; Salwin-Kui chiang divide, 28°24'N, 98?24'E, Forrest G. 20844; Chen-Kang Hsien, Wang C.W. 72322. Ecology. Та mixed forests, usually growing on yellow loam. Altitude: m. Flowering May-June; fruiting crt od. Uses. Wood used for furniture, tree grown as an ornamental because of its beautiful flowers. Collector's notes. Bark gray, smooth; flowers greenish shaded purplish outside, white, ivory white, yellow-white to deep purple inside, fleshy, fragrant; leaves green above, pale green below, bruised leaves spicy-fragrant; fruit follicles conelike, green when young, orange-green, reddish yellow, red later. Note. anglietia insignis is a very wide- spread species. It therefore displays many varia- tions in shape and size of the leaves as well as size and color of the flowers throughout the area of distribution. — — . Manglietia lucida B. L. Chen & 5. C. Yang, Acta Sci. Nat. Univ. Sunyatseni (Guangzhou) 3: 94. 1988. TYPE: B. L. Chen & C. М. Mai 87 T 002 (holotype, SYS). Figure 6. Tree to 18 m high and 65 cm diam.; young twigs stout, 8-10 mm diam., grayish, smooth, often longitudinally wrinkled when dry, glabrous to mi- nutely hairy. Stipules ferrugineous-pubescent, ad- 3-4(-11.5) ст long. Leaves coriaceous, glabrous on both sides, nate to the base of the petiole, with slightly recurved margins, obovate, approxi- mately three times as long as broad, dark green and glossy above, often pale green below, 27-44 11-16 cm; apex acuminate, acumen 10-25 mm long; base cuneate, decurrent with two ridges into the petiole; midrib lengthwise sulcate above, rather elevated below; nerves in 13-19 pairs, curved upward and meeting in an intramarginal vein; all lesser veins forming a coarse reticulation on rather dull upper faces; petiole glabrous, often conspicuously thickened toward its base, 3.6— long, stipular scars 16-45 mm long. Peduncle pu- bescent as stipules, ca. 1.4 cm long and 6 mm thick, pedicle with same indumentum, 5-6 mm long; spathaceous bract only one, rather thick co- riaceous, broadly obovate, at the base of pedicle, its outside face pubescent, 83 x 69 mm. Tepals 9(-1 "i purple, the outer 3 narrowly obovate, fleshy, 12- .8-6 cm, the inner 6-8 in two rows, Fe B narrower than the outer tepals. Sta- mens numerous. Gynoecium ovoid, ca. cm, carpels 70-80, glabrous, nearly entirely con- nate when young. Fruits ovoid-globose, 9.5 x 8- 10 cm, torus with scars of perianth and stamens ca. 3 X 2 cm; fruiting peduncle ca. 4.5 X 1 ст, pedicle ca. 2.5 X 1 cm, ripe carpels ligneous, 20— x 8-10 mm, abaxial part 10-15 mm long, dehiscing along the dorsal suture at first and later also along the ventral suture. Seeds ca. 3 in each carpel. Distribution. CHINA: Southeastern Yunnan: Maguan, B. L. Chen & C. N. Mai 87 T 002 Ecology. Та secondary forest. Altitude: 550- 670 m. Flowering March-May; fruiting Septem- ber- October. N Manglietia lucida is apparently most closely allied to Manglietia garrettii and Man- glietia grandis. It differs from the former by the obovate leaves, glabrous leaves and petioles, and bigger flowers, and from the latter by the pubescent stipules and peduncles, as well as by its smaller fruits. 12. Manglietia megaphylla Hu & Cheng, Acta Phytotax. Sin. 1: 159. 1951. TYPE: China. Yunnan: Faadou, Sichour Hsien, 1,100 m, C. W. Wang 85156 (holotype, PE not seen). Figure 7. Tree to 25 m high and 85 cm diam.; buds, young twigs, stipules, petiole, midrib and nerves below, and peduncles densely villous with rather long, undulate to curly, dark brown or rufous hairs; twigs stout, 8-12 mm diam., black-brown. Stipules adnate to the base of the petiole. Leaves thinly coriaceous, dark green, glabrous above, pale green, scattered-villous with long, undulate to curly, dark brown hairs below, obovate, 23-40 x 9.4-17 cm; apex acute to abruptly short-acuminate, base cu- neate; nerves visible beneath, in 20-22 pairs; re- ticulation coarsely netted, prominent on both sur- 1046 Annals of the Missouri Botanical Garden FIGURE 6. Manglietia lucida B. L. Chen & S. C. Yang.— 1. Fruiting branch. — 2. В 4. Inner tepal.— 5. Follicle. Based on SYS 161496 = Chen ract. — 3. Outer tepal.— Bao Liang 87 T 002. Drawing by Xie Qing Jian. Volume 80, Number 4 1993 Chen & Nooteboom 1047 Magnoliaceae of China Y | М EY POOLE: WEE == LE Е Z 2 === em Zr E fF 777 Y | [4 / PU DET 45 4 0b И FIGURE 7. Manglietia megaphylla Hu & Cheng.— 1. Branch with flower bud.— 2. Fruit. — 3. Follicle. — 4. Undersurface of leaf. Based on SYS 161435 — Chen Bao Liang 87 T 102. Drawing by Xie Qing Jian. faces. Petiole usually dilatate toward the base, 1.5— 3.7 cm long, stipular scars 0.7-3 cm long. Flowers not seen. Fruits ovoid to globose or oblong, ca. 7— 9 x 6.5-8.5 cm; carpels 57-65, 2.5-3 cm long, apex with sharp beak 4—7 mm long, slightly re- curved, dehiscing along the dorsal and the ventral suture; fruiting peduncles stout, ca. 1-3 x 1-1.3 cm. Pedicle 6 mm. Annals of the Missouri Botanical Garden | CHINA. Guangxi: Donglin, га С.С. 11438. Yunnan: Maguang, ve HL GS 9 5; Si-chour- p Faa Doou, Feng K.M. 11808. Ecology. In evergreen broad-leaved forests. Altitude: 800-1,500 m. Flowering April; fruiting September—October. Uses. and durable, moderately light and soft, commonly used for construction work, furniture, and plywood. Collector's note. Fru In Manglietia megaphylla the distin- W ood straight-grained, coarse-textured it pale green, globose ote. guishing characters are the rufous indumentum with dense and long hairs, large leaves, and ovoid to oblong-ovoid fruits. 13. Manglietia microtricha Law, Bull. Bot. Res. (China) 5(3): 125, t. 4. 1985. TYPE: Tibet Complex Exped. 74-1847 (holotype, PE not seen). Tree to 30-40 m high, 40-60 cm diam.; young twigs stout, 6-10 mm diam. Stipules appressed brown pubescent, adnate to the base of the petiole, scars 6-13 mm long. Leaves coriaceous, green, glabrous above, EUER beneath, obovate, 13— x 5—7 cm; apex short-a nerves in 15-17 pairs, reticulation fine, a netted, prominent on both sides when dried. Petiole 2-2.5 ст long. Flower not seen. Fruiting peduncle ca. 1.5 ст long. Fruit ovoid, ca. 6 ст long, the dorsal face of the ripe carpel narrowly ellipsoid, appressed brown pubescent, with a ca. beak at the apex. Seeds brownish, 1.5- 3-mm-long 2 cm long. Distribution. CHINA. Xizang (Moyuo). Ecology. In evergreen broad-leaved forest. Altitiude: 2,000 m. Fruiting September—October. Note. able; thus, original description, Manglietia microtricha seems closely related to No collections of this species were avail- based on Law’s Manglietia szechuanica. 14. Manglietia moto Dandy, Notes Roy. Bot. Gard. Edinburgh 16: 128. 1928. TYPE: К.Р. To, W. T. Tsang & U.K. Tsang 180 in C. C. C. 12179 (holotype, A; isotypes, MO, SYS, UC). Tree to 20 m high, 60 cm diam.; twigs, stipules, petiole, midrib below, bracts and buds, young peduncles densely villous with rather long crispy to curly dark brown to rufous hairs; young twigs 4—5 mm diam. Leaves green, glossy above, scat- tered-hairy with long, undulate to curly dark brown hairs beneath, glabrescent later, coriaceous, (nar- rowly) obovate, 11-19 x 5-7 cm; apex mucronate to acuminate, base cuneate to broadly cuneate, nerves in 9-19 pairs, reticulation obscure on both sides; petiole lengthwise narrowly sulcate above, 1.5-3.7 ст long, stipular scars narrowly trian- gular, 0.4—1.5(—3) ст long. Peduncles 4.3-9 cm X 4—6 mm, pedicle absent. Flowers fragrant; tepals 9, creamy white, the outer 3 coriaceous, oblong to ovate-oblong, + hairy outside at the base, 3.7- x 2-2.5 ст, the inner 6 thick and fleshy, spathulate to obovate, slightly concave, the inner- most ones usually becoming short-clawed toward the base, 9-12 x 1.1-1.4 cm long, connective appendage semi- 1-1.8 cm; stamens numerous, rounded or sometimes with a short, sharp tip, 1- 1.5 mm long, filaments glabrous, ca. 1 mm long; I-L2 cm, mm long; scars of the fallen perianth and stamens 9-12 x 4-7 mm. Fruiting peduncles ca. 11 x 0.5-0.6 m. Fruits ovoid, 5-7 x 3.5-6 cm; carpels pap- illiferous on the dorsal face, apex with a ca. 2-3- mm-long beak. Seeds ovoid to elliptic, 6-9 x 4- 6 mm, 6-9 in each carpel. gynoecium broadly ovoid, 1.3-1.5 x number of carpels 48-71; styles ca. 2 Distribution. CHINA. Guangdong: Wat Shui Shan, North river region, Chun W.Y. 7353; Yuyuen distr., Ko S.P. 53514; also reported form Ruyuan, Yaoshan, Ying- teh, Lokchong, and Huanshuishan. Western, central, and northern Guangxi (He Xian, Cangwu Xi). Southern Hu- In mixed forests. Altitude: 400-900 m. Flowering May-June; fruiting August-Septem- ber. Ecology. Uses. Wood fine-textured, soft, used for gen- eral construction work and furniture. Collector’ s note. Bark dull gray, flowers See Dandy (1928a). Manglietia moto is a notable species and is easily distinguished by its rufous indument and long peduncles. Manglietia rufibarbata, which was originally collected from Vietnam, has not been recorded in China so far as we know. In our opinion, it closely resembles Manglietia moto in all characters, ex- cept for its stout peduncles and the fruits with rufous hairs. Because we have not seen sufficient material of Manglietia rufibarbata, we still main- tain it here. The type was identified as Magnolia kwang- tungensis Merr. by Merrill (1927a) = fordiana Oliver var. kwangtungensis. Manglietia 15. Manglietia pachyphylla Chang, Acta Sci. Nat. Univ. Sunyatseni (Guangzhou) 1: 53. Volume 80, Number 4 1993 Chen & Nooteboom 1049 Magnoliaceae of China 1961. TYPE: P. S. Wang & H. H. Chiu 241 (holotype, SYS). Figure 8. Manglietia crassipes Law, Bull. Bot. Res. (China) 2: 133. . TYPE: China. Guangxi: Jinxiu Xian, 1,300 m, Q. H. Lu 4319 (holotype, IBK not seen) Tree 16 m high and 30 cm diam.; bark gray- black; twigs dull black, stout, ca. glabrous, covered with a waxy substance; stipules ] cm diam., brown-villous, adnate to very base of the petiole. Leaves coriaceous, thick, rigid when dry, glabrous, dark glossy green above, о с. minutely pu- elliptic, 13- 32 x 5-10 cm; apex ud bua up: base bescent beneath, (narrowly) ob cuneate; nerves in 8-14 pairs, faint on both sur- faces, reticulation obscure. Petiole stout, sulcate above, 3-5 cm long, stipular scars 1-7 mm long. Peduncles minutely pubescent, ca. 1.7-2 x 0.7 cm, pedicle ca. 4 mm long; bract only one, broadly ovate, coriaceous, glabrous, ca. 4.8 x 7 cm. Flow- ers fragrant, tepals 9, white, the outer 3 oblong, 5.5-8 x 3.2-3.5 cm, obovate or broadly obovate, attenuate toward the base, convex, thick and fleshy, 3.2-4.3 x 2.5-3 cm, the innermost tepals usually short-clawed ba- coriaceous, the inner 6 sally; stamens 0.7-1.2 cm long, connective ap- pendage with rounded apex, ca. 1 mm long, fila- ments glabrous, ca. 1 mm long; gynoecium ovoid, ca. 2.2 x 1.8 ст; styles glabrous, ca. 1 mm long; the torus with scars of perianth and stamens ca. 1.5 x 0.7 cm. Fruiting peduncles 3-5.5 x 1.2 cm. Fruits ellipsoid, 5-7 X 4.5 ст; ripe carpels 38-46, ca. 1.5-2.5 mm long, short-beaked. Seeds 3-4 in each carpels, flat globose, 5-6 mm diam. Distribution. CHINA. Guangdong: Conghus Xian, 1 Kok Shan, LA ang А ; u: Anlong Xian, Huangshan Exped. 3049. Also reported from Guangxi (Dayaoshan and Jinxiu Shan). Ecology. In evergreen broad-leaved forests. Altitude: 800-1,500 m. Flowering May; fruiting August-September. Uses. The timber is used for general construc- tion work, furniture and plywood; the tree is grown also as an ornamental plant because of its handsome crown. Notes. The type specimen of Manglietia crassipes has not been seen. However, according to the published description and the collections examined, it is clear that Manglietia crassipes is conspecific with Manglietia pachyphylla. 16. Manglietia szechuanica Hu, Bull. Fan. Mem. Inst. Biol. (Peiping) 10: 117. 1940. TYPE: 7. T. Yu 3105 (holotype, PE; isotype, A). Tree 15 m high and 30 cm diam.; young twigs green, 4-5 mm diam., at first yellowish villous, soon glabrescent or only hairy at the nodes; stipules pubescent, adnate to the base of the petiole. Leaves glabrous above, pale green, appressed-pubescent with short, straight, brown hairs on the surface and with rather long, straight, colorless to yellowish hairs along midrib and margins beneath, coria- 4-(.3 em; apex mucronate to acuminate, acumen 15-17 mm ceous, narrowly obovate, 13-20 x long; base narrowly cuneate to cuneate; midrib impressed above, elevated below, white villous with brown pigment, nerves fine in 8-19 pairs, retic- ulation coarse. Petiole white villous or later gla- brescent, 1.5-2.5 cm long, stipular scars 9-11 mm long. Peduncles villous, glabrescent, 1-1.6 x 0.4-0.5 tepals 9, the outer 3 ovate-oblong, greenish with cm, pedicles 3-4 mm long. Flowers white; a flush of purple, sparsely villous, 4-5.5 х cm, the inner 6 oblong to obovate, short-clawed at base, 4.5-6 х connective appendage triangular; gynoecium ovoid- 1.7-2.5 X the lower narrowly ellipsoid, yellowish pubescent, 8-10 mm long. Styles glabrous, 4-6 mm long. x 5 mm. 1.5-3 cm; stamens 1.2-2 cm long, ellipsoid, -3 cm, carpels 42-58, Scars of perianth and stamens 7-10 Distribution. CHINA. Northern Yunnan: Sui- jiang, Sun B., S. 341. Southern and central Sichuan: Omei Xian, Chow H.C. 11944; Pingshan, Yu T.T. 3105. Altitude: 1,400-1,800 m. Flowers pink or purple-red, Ecology. Collector's note. scented. Bark grayish brown, smooth; leaves dark green above, yellowish brown with long hairs be- neath. 17. Manglietia ventii Пер, Feddes Керегі. 91, 9-10: 560. 1980. TYPE: China. Yunnan: Din- bian Hsien, alt. 880 m, 9 Aug. 1953, Р. L Mao 2842 (LE not seen). Tree to 30 m high. Young twigs 2-3.5 mm diam., black-gray, densely appressed-tomentellous with short, straight, glistening, yellowish hairs, old ones + gray-pubescent to glabrescent. Stipules yel- lowish appressed-tomentellous, adnate to the base of the petiole, scars 1-2 cm long. Leaves usually crowded at the end of the twigs, chartaceous to thinly coriaceous, elliptic, broadly elliptic, obovate- elliptic to obovate, sometimes unequal; glabrous above, appressed-tomentose with short, straight, glistening, yellowish to silver hairs beneath, 9-18 1050 Annals of the Missouri Botanical Garden FIGURE 8. Manglietia pachyphylla Chang. —1. Branch with flower bud.— 2. Fruit. — 3. Follicle.— 4. Open follicle with seeds. Based on SYS 121334, Anonymous 241. Drawing by Xie Qing Jian. x 2.2-6.5 cm; apex short-acuminate to acumi- 12-16 pairs, obscure above, visible beneath, re- nate, acumen 3-17 mm long, base cuneate to ticulation lax, faint on both sides. Petiole slender, obtuse; midrib impressed above, elevated below, —tomentellous, with same hairs as the young twigs, densely pubescent to glabrescent, nerves fine, in glabrescent, 2.6-3 cm long. Flower not seen. Fruit- Volume 80, Number 4 1993 Chen & Nooteboom 1051 Magnoliaceae of China ing peduncle yellowish appressed-tomentose, 1-1.5 x 0.4–0.6 cm, with a bract scar, pedicle absent. Fruit ovoid to long-globose, 3.5-7 x 2.5-4.5 cm; ripe carpels ca. 64, long-ovoid, the lowest carpels decurrent along the torus, + yellowish pubescent, especially toward the base of the lowest carpels, beak rather long, erect or recurved, dehiscing along the dorsal suture, 2-2.2 cm long; torus with 6-9 7 mm scars of perianth and stamens; scars of the fallen tepals ca. 10. Seeds subcordate to ovoid, + impressed, 1-3 in each carpel, 7-7.5 x 5-6 mm. Distribution. China and Vietnam. In CHINA. Southeastern Yunnan: Ping-pien Hsien, Tsai Н.Т. 61619. Ecology. Altitude: 880 m. II-2. Manglietia section Manglietiastrum (Law) Nooteboom, Manglietiastrum Law, Acta Phytotax. Sin. 11: 72, t. 2. 1979. Magnolia sect. Manglietiastrum (Law) Noot., Blumea 31:91. 1985. TYPE SPECIES: Manglietiastrum sinicum Law. Stipules free from petiole. Young leaves spread- ing in bud. Gynophore present. 18. Manglietia sinica (Law) B. L. Chen & Noo- teboom, comb. nov. Manglietiastrum sini- cum Law, Acta Phytotax. Sin. 17: 73, t. 2, 3. 1979. Magnolia sinica (Law Yuh-wu) Noot., Blumea 31, 1: 91, photo. 3. 1985. TYPE: China. Yunnan: Xichou, 1,550 m, Law & Wang 6-75 (IBSC, v. spont.). Tree to 40 m high and 1.2 m diam.; bark gray, longitudinally fine-fissured; trunk somewhat but- tressed at the base; plant entirely glabrous; twigs dark green when young, faint brown when old; stipules free from the petiole. Leaves coriaceous, dark shining green above, pale green beneath, nar- rowly obovate, 15-26(-30) x 5-8(-9.5) cm; apex short-acuminate, with ca. 5-mm-long acute tip, base cuneate; margin slightly revolute; nerves in 13-16 pairs, reticulation coarse, prominent on both surfaces when dry. Petiole smooth above, slightly thickened at base, 1.5- Tepals 9, 3-merous, in 3 whorls; stamens ca. 65, connective appendage long-pointed; carpels 13-16. Fruits green at first, slightly flushed, obscure brown when dry, ovate, oblong-ovoid to obovoid, 5-8.5 x 3.5- 6.5 cm; carpels thick woody, narrowly oblong- elliptic to obovoid-elliptic, 2.5-4 x 1.5-2.5 ст, opening along the ventral suture and splitting at cm long. the apex, dorsal faces conspicuously lenticellate; gynophore ca. 1 x 1.3 cm. Seeds 1-3 in each carpel, elliptic, compressed laterally, 1-1.3 cm wide, ca. 7 mm thick Distribution. CHINA. Southeastern Yunnan: Xichou, Chen B.L. 87 F 180 Ecology. In evergreen broad-leaved rainfor- est. Altitude: 1,300-1,500 m. Flowering April; fruiting September-October. Note. Manglietia sinica, a magnificent tree, is confined to a small area in southeastern Yunnan. One of the authors has investigated it several times in the field. The species possesses the chief diag- nostic characters of Manglietia, e.g., habit of tree, the terminal flowers with 9 tepals, and more than 2 ovules in each carpel. Therefore, we prefer to include it in Manglietia. On the other hanad, there occur some remarkable characters in this species, such as the petioles without stipular scars and the gynoecium with a short gynophore. These features easily distinguish it from the other species in this genus and warrant its status as a section. The sections in Magnolia are based on the same kind of characters. Ш. Kmeria (Pierre) Dandy, Kew Bull. 1927: 262. 1927. Magnolia subg. Kmeria Pierre, Fl. Forest. Cochinch. 1: sub t. 1. 1880. TYPE SPECIES: Kmeria duperreana (Pierre) Dandy. Tree. Stipules adnate to the petiole. Flowers terminal, solitary, unisexual; tepals 6—7, 3-merous, subequal; anthers dehiscing introrsely, the connec- tive produced into a short or moderately elongated appendage. Gynoecium sessile. Carpels compara- tively few, concrescent. Ovules 2. Fruiting carpels woody, separating on dehiscence, dehiscing com pletely along the ventral suture and partly along the dorsal suture, thus finally becoming bifid. Seeds 1-2 in each carpel. Distribution. Two species, southern China to Indo- ina Note. This genus is closely related to Mag- nolia. It could also be treated as a section close to section Gwillimia. In China the only species that occurs is Kmeria septentrionalis. 1. Kmeria septentrionalis Dandy, J. Bot. 69: 233. 1931. TYPE: R. С. Ching 5247 (holo- type, BM; isotypes, NY, PE). Tree to 18 m high; bark gray; twigs green, + glaucescent when dry, appressed short pubescent. Stipules glabrous or sparsely pubescent, adnate to rather high on the petiole, stipular scars nearly to 1052 Annals o Missouri Botanical Garden its apex. Leaves coriaceous, bright green, glabrous above, greenish, glabrous or sparsely pubescent toward the base when young beneath, elliptic-ob- long or obovate-oblong, 8-15 x : retuse, base broadly cuneate; midrib prominent below, nerves visible on both sides, in 12-17 pairs, reticulation prominent on both surfaces when dry. Petiole slender, pubescent to glabrescent, 2-3.5 –6 cm; apex cm long. Brachyblast glabrous, 1.5 cm long. Flower bud (the male flower) subglobose; tepals 6; stamens 1–1.5 em long, connective appendage triangular, 1-2.5 mm long. Distribution. CHINA. Northern and central Guangxi: Huanjiang, Pang C. à Ae southeastern Lu- chen, Ching Cl C. 5247. Yunnan: Huang Jingin, Mar Li Po, Wang C.W. & Liu 83157: ‘Mapua, Chen B.L 87 F H3 Ecology. In evergreen forest. Altitude: 300- ОО m. Flowering May-June. Collector's notes. Tree with a large dense crown; bark gray; leaves evergreen; flowers white, fragrant. Note. perreana (Pierre) Dandy. Recently, a new record of K. duperreana in Libo of Guizhou was reported, but we have not seen a collection from that locality. This species is closely related to K. du- TRIBE MICHELIEAE IV. Michelia L., Sp. Pl.: 536. 1753. Champaca Adans., Fam. Pl. 2: 365, 537. 1763. Sam- pacca О. Kuntze, Rev. Gen. Pl. 1: 6. 1891. TYPE SPECIES: Michelia champaca L. Liriopsis Spach, Hist. Natur. Veget., Phanerog. 7: 460 1839, non Liriopsis Reichenb. (1828). TYPE SPE- CIES: Liriopsis fuscata (Andr.) Spach. un Н. Hu, Sunyatsenia 4: 142. 1940. TYPE SPECIES: Paramichelia di oe Pridie | Ни. Tsoongiodendro cta ax. Sin. 8: 7. 1963. TYPE SPECIES: “Taoonpiodendron odorum Chun Trees or shrubs. Stipules adnate to or free from the petiole. Flowers bisexual, on short brachyblasts in the axils of the leaves, growth monopodial. pals 6-21, 3-6-merous, subequal or rarely the outer whorl different. Anthers latrorse or subla- trorse (to introrse), connective produced into a short or elongated appendage. Gynoecium stipitate; carpels many to few (rarely 1), free or connate; ovules 2 to many; follicles free, + spaced along the torus, dehiscing along the dorsal suture, finally becoming 2-valved, or concrescent, forming a fleshy syncarp, tardily and irregularly dehiscent, or a woody syncarp, the upper parts of the carpels falling away while also dehiscing along the dorsal suture, the basal parts remaining attached to the torus with their suspended seeds. Distribution. About 40 species, in Southeast Asia from India and Sri Lanka eastward to southwestern and southeastern China, northeastward to southern Japan and southeastward into Indonesia (not in Sulawesi and New Guinea). KEY TO THE SECTIONS OF MICHELIA Tepals very dissimilar, those of the outer whorl much shorter and narrower those of the inner whorls; fruit apocarpous; stipules free from petiole. ____ -2. Michelia sect. Anisochlamys 1 Tepals of the outer whorl sometimes smaller but not very dissimilar; fruits apocarpous or syncarpous; stipules free from or adnate petiole. 2 бони syncarpous; stipules adnate to petiole. 3 uits apocarpous; stipules free from or adnate m petiole. . 4 . Tepals 9; fruits 13-16 c MORS -9. Mic helia | sect. Tsoongiodendron 3. Tepals 12; fruits 3-9 cm. IV-6. Michelia sect. Paramichelia . Tepals 6, or occasionally 4, 2-merous; stipules free from petiole TORO ы V-3. Michelia sect. Dichlamys 4. Tepals 6-21, 3-6-merous; stipules free from or adnate to petiole. 5(4). Petioles not exceeding 10 mm, stipules adnate to it; shrubs or small trees. |... mm IV-4. Michelia sect. Micheliopsis 5. | ни than 10 mm, stipules free ог adnat I -l. Michelia sect. Michelia O THE SPECIES OF MICHELIA (primarily based on floral characters) Tepals 6 or 7. 2 l. 2(1). 2 . . Gynoecium longer than androecium. 3; Gynoecium shorter than androecium, often entirely hidden Petiole 5-10 mm m long, stipular scars nearly to its apex; tepals 22-35 mm long; gyno- phore 5-8 mm long; gynoecium ovoid to oblong. 33. Michelia id ir aes Petiole 3-5 mm m long, stipular scars to ha way; tepals 15-25 mm long; gynophore 1- m long; ED cylindric. Michelia figo var. figo Stipules adnate to I ser 2-4 mm; flower purple or red; stamens 10 mm long; gynoecium cylindric. 32b. Michelia figo var. crassipes Stipules free; а ae 10 mm; а white; stamen mm long; gyno ellip- Michelia frenis PR 7 d = = = сл 6(2 em glabrous. | ~ Volume 80, Number 4 1993 Chen & Nooteboom 1053 Magnoliaceae of China 6. 7(6). 8(6). 10(1). 10 bro 14(13). 15(12). . Leaves glabrous Leaves hairy beneath at least when young. 8 Base of leaf рева: cuneate or attenuate- 7 pairs; yos inns 3-9 mm; gynophore 4 mm; gy mens along torus under a 4-5 mm AN . Michelia chapensis Leaves 1.5-2.5 cm br ie shrub to high; leaf apex “а outer tepals m late, 18-20 x R cris hidden by the stamens; gynophore 3-4 mm lon 26. Michelia or ME o 15-18 Mi Leaves 2.8-16.5 cm broad; tree long, or elliptic; gynoecium ovoid or cylin- dric; gynophore 5-10 mm long. Pedicle present; twigs appre lous; n pis glabrous or tomentellous; leaves er = scure; ‘petiole not dilat cium cylindric. 29. Michelia n Pedicle absent; twigs pubescent; stipules pu- escent; leaves 4.5-10(-16.5) cm broad, jiu distinct; dd dilated at base; id. . Michelia balansae 11 Stipules free | Stipules adnate to petiole. и... 29 9 es hairy bonsai at least when young. 17 -8 m long... 13 Brachyblast 2. 1-3 cm long; stamens 2.5- . Leaves glaucous beneath, nerves in 12-14 pa m "ps m petiole hairy; outer te- pals 3 oad; gynoecium golden a ca. 20 ПОНЕ Toll. еее 13. Michelia ingrata Leaves not glaucous beneath: stipules gla- brous or pubescent; petiole glabrous; outer e 5-40 mm broad; bane gla- 10 mm long. 15 = QQ @ о 8 tted, nerves in TURA OIOR A NC RSEN Michelia maudiae Plant hairy at least in innovations; stipules pu ubescent; reticulation laxly netted, uit RH 15. Michelia lacei Twigs glabrous, 1-2 mm diam.; stipules pu at least toward base; tepals very a 16(15). 17(11). 17. 18(17). 19(18). bo N ~ to po м 23(21). outer tepals small, nearly linear, ca. 1 mm broad; gynophore under fruit 2-3 mm long. Michelia hypolampra Twigs hairy at least an er tepals obovate or spathulate; gynophore under fruit 10-17 mm lon Gynoecium glabrous, pron , 10-12 mm ong; stipules tomentellous; reticulation dis- tinct; petiole dilated at base; filaments 2.5- З mm long. Lesson 5. Michelia coriacea Gynoecium hairy, ovoid, 4-6 mm long; stipules pubescent; low om of leaf rath- er obscure; petiole not dilated at base; fil- aments 1 mm long Michelia compressa Outer tepals 1- | cm long. 1 Outer tepals 3.1- | 1 present; oils 5-10(-16.5) x 3- -7) em, reticulation distinct; brachyblast d or brown tomentellous, 6-10 mm long; connective appendage EN 2- 4 mm long; filaments 1.5-2 mm long; gyno- phore under fruit 5-8 vi за Michelia mediocris Pedicle absent; leaves "i 26 x 1.5-1 cm, derart rather obscure; ете 3-6 or 13-25 mm long; connective ap- pendage E 0.5-1 mm long; fila- ments l- | mm idea gynophore under fruit 10-30 m | NUM soe = 5 mm long; twigs t tomen- to aes silky; brachyblast 13-25 m -24 mm broad; sta- = Mns gynoecium cylindric, 17-20 mm high; mber of ovules per carpel 10; follicles hairy. Rotten 10. Michelia foveolata Gynophore 2-6 mm long; twigs puberulous; stipules pubescent; brachyblast 3-6 mm mm broad; stamens noecium ovoid, 4-6 m ovules per carpel 2=8; ШУ glabrous Michelia compressa А Leaves glaucous even twigs villous, 5- mm diam.; stamens 17-22 mm long, filaments 4-6 mmm; gynoecium golden pu- berulent. 13. Michelia ingrata Leaves not t glaucous beneath twigs pubes- t u а. она ji^ im blasts 5-7 cm long; ur шш. m Michelia masticata Filaments 4-6 mm FAN twiga tomentose; young leaves tomentose beneath, reticula- tion rather obscure, densely netted; gynoe- cium 17-20 mm high; fruiting brachyblasts 1.5-3 cm long; follicles hairy. . 10 AESA EE . Michelia foveolata Leaves 6.5-14 x 4-6.5 cm; connective 1054 Annals of the Missouri Botanical Garden 24(23). 25(10). 26(25). 27(25). 28027) n 29. 30(28). appendage linguiform; twigs and stipules tomentellous; tepals 9-12, outer s er 35- 45 x 10-15 mm; gynoecium cylindric 16. Michelia no -clurei ке: )12-20(-24) x 35- Тог 25 m; connective apentag triangular, twigs pubescent or tomentose; stipules pubescent; tepals 9-12 or 13- 17, outer o s 32-36 x 10-12 or 45-70 x = m; gynoecium ellipsoid. _ . 24 Pairs of | lateral nerves 11-15: twigs pu- bescent, 2-3 m ; leaves finely ap- pressed uber n шй, base cuneate or broadly d dou 9-12, 45-70 x 20-25 mm; filaments 2-3 mm long; scars of perianth and stamens along torus under fruit 5-10 mm long. _ 2 она cavaleriei Pairs of ен nerves 16-18; s to- mentose, ca. 4 mm diam.; bd 1 densely appressed tomentellous beneath, base rounded; tepals 13-17, the outer ones 32-45 x 10-12 mm; filaments 4 mm long; cars of perianth and stamens along torus under fruit 12-14 mm long 1. Michelia aenea Эйе. 20-39 mm long. cc 26 ens 5-15 mm long. 27 Brachyblast 20- 22 mm long; tree to 15 m igh; twigs in innovations 8-12 mm thick; petiole dilated at base; stipular scars 7-10 mm long; brachyblast 10-12 mm thick; petiole absent; stamens 33-39 mm long, filaments 7-8 mm long; gynoecium shorter than androecium, often entirely hidden, 24 mm high; gynophore 6 mm eis А “Miceli E fita Brachyblast 30-35 mm e dei 6 m high s in innovations 2-3 m thick; petiole not dilated at base; Бете scars 3-4 mm long; brachyblast 5-6 m thick; uds present; stamens 20-24 mm long, filaments 2-4 mm long; gynoecium longer than androecium, 15 mm high; gy- nophore 30 mm lon 7. Michelia nae Outer tepals 10-30 mm broa Outer tepals 2.5-10 mm broad. 32 Pedicle present. 29 Pedicle absent. 30 Petiole hairy, 2. $- 3 cm; stipular scars 15 mm long, leaf base cuneate, reticulation rather йош outer tepals 3; руп орһоге 4 mm lon ichelia microtricha Petiole 0.8- -2. 5 ст, glabrous; stipular scars 2-6 mm long; leaf base broadly cuneate or rounded; reticulation distinct; outer tepals 4 or 5, gynophore 5-12 mm lon _ 6. Michelia dotsopa iex d appendage m long, nar- owly triangular or Torr ian twigs sparse- 10 mm high; fruiting brachyblasts 5 m thick. 14. Michelia on Connective appendage very short to long triangular, 0.5-3 mm; twigs pubescent or tomentose; young leaves densely or sparsely pubescent beneath, reticulation rather ob- scure; gynoecium 9-14 mm or 3-5 mm high; fruiting brachyblasts 2-4 mm thick. 31 31(30). р а ovoid to oblong, 3-5 mm high; le 3-5 mm long. 33. Michelia yunnanensis 31, си cylindric, 9-14 mm high; pet- ole 9-25 mm long. ... 9. Michelia floribunda 32(27). Outer tepals 4-6. 33 e pa 33(32). Stipular scars 3-5 mm long; leaves elliptic or narrowly elliptic, reticulation rather ob i brachyblast stout; gynoecium 9-15 m high; gynophore 6- fa mm long. Michelia velutina 33. Stipular scars 17-28 mm e leaves ovate or narrowly ovate, reticulation distinct; brachyblast slender; Буп oecium 7-12 mm high; gynophore 3-5 mm long. 3. Michelia champaca 34(32). Stipular scars 1-5 mm long. 35 34. Stipular scars 6-20 mm long. . 36 35(34). bo dd 12; connective appendage с: ca. 0.5. m long; stipules pubescent; fruiting barbie E: mm long and 8 mm thick; fruits 13.5-15.5 ст long; fruit 20-25 mm s . 23. Michelia. олй 35. Tepals 12-15; connective appendage 1-3 mm; stipules tomentellous; fruiting brachy- blasts 5-8 mm long and 3-4 mm ue fruits 2.5-6 cm long; EA under fru 10-15 mm long. .. helia “floribunda 36(34). ee 15-20 x E 75m mm; stamens 5-8.5 ae connective appendage 0.5 m; gynoecium ovoid, 4 mm, carpels 10- 12; apula scars 6-18 mm. m 34. Michelia odora 36. Tepals 9-20, 19-50 x 6-23 m nective appendage 1-4 mm ine Ds cium ovoid or more often cylindric; carpels often more than 12; stipular scars 2-20 37 mm. Pr aaa scars 14-20 mm; gynoecium 6- tepals 9-15, 22-35 mm x AM 13 mm; pie ed AUT 2-4 m " . Mic helia kisopa 57. Stipular scars 2-4 mm; ы. 9-14 a не mms tepals 12-20, 12-20 x 19- 37(36). mm; connective appendage 1-2 mm; stip- ular scars 6-10 mm. . 35. Michelia baillonii TO THE SPECIES OF MICHELIA (primarily based on fruit characters) | Fruits with connate carpels, when mature a ing e or apical parts falling in irregular masses. 2 Volume 80, Number 4 1993 Chen & Nooteboom 1055 Magnoliaceae of China 8(4). 12(11). Fruits consisting of free carpels, which de- hisce along the dorsal suture, main nerve often persistent. 3 Reticulation densely netted; fruits 3-9 cm long and 2.5-4.5 cm broad. Michelia ~“ nos laxly netted; fruits 13-16 € long and 7-9 cm broad. . > Michelia « odora Stipules ни to the xs TRENT E Stipules fre Stipular scars 11-28 mm long. Stipular scars 1-10 mm long. аана Pedicle present, at least 3 mm long; leaves elliptic or narrowly obovate, reticulation rather obscure; twigs tomentellous 20. Michelia microtricha 6 Pedicle absent or very short. Е Leaves minutely appressed pubescent, gla- brescent beneath; twigs puberulous or to- i fruiting beber 7x5mm; follicles 8-14 x 7-10 mm . 14. Michelia kisopa Young leaves tomentose or pubescent be- neath; twigs pubescent or tomentose; fruit- ing brachyblast 5-30 x 3-5 mm; follicles -20 x 7-14 mm. Leaves 1-5 cm broad; reticulation rather obscure; twigs tomentose; stipules tomen- tellous; fruiting dE 5-8 mm lon Michelia floribunda = 3.4-12 ст we reticulation dis- nct; twigs and срна ne fruiting Pun he 10-30 m | elia сћатраса Репоје 2-5 mm (Miche "m yunnanensis with very short en is also possible here). 9 Petiole 5.1-30 m Shrub to 2-5 m 87 flower purple ог red; fruiting brachyblast 10-20 x 3-5 mm. 32b. Michelia p var. crassipes Shrub or treelet to 15 m high; flower white or 8 fruiting E 7-25 x 2- 4m 32a. Michelia figo var. figo Pedicle || ea a oa eee Pedicle absent. n . Treelet to 5 or 6 m high; twigs silky, 2-3 mm diam.; stipules silky; leaves 16-22 7.5-10 ст, nerves in 9-14 pairs; petiole 2-2.5 cm; brachyblast 30-35 mm long; outer tepals 3; stamens 20-24 mm long; gynophore 30 mm long. 7. Michelia elliptilimba Tree to 15-30 m high; twigs pubescent or tomentellous; stipules pubescent or tomen tellous; brachyblast 4-17 mm long; leaves 9- x 8 or 11-24 x 3.5-6.5(-8.5) cm, nerves 10-28 pairs; petiole 0.8-2.5 cm; outer tepals 4-6; st 8-15 mm long; gynophore 5-12 m Scars of perianth and stamens in torus nder fruit 5 mm long; twigs and от pubescent; reticulation of MAE rather ob- i ichelia а oru under fruit 7-10 mm long; twigs n 1500). VO 16(15). 19(18). 19. 20(19). 20. 21(22). 23(22). . Young twigs tomentellous; stipules tomentellous; retic- ulation distinct; petiole glabrous. ................... . Michelia doltsopa Fruiting brachyblasts 2-4 mm thick. 0... 14 Fruiting brachyblasts 5-12 mm thick. ....... 15 Shrub or small tree 6-12 m; petiole 5-10 ш stipular scar 3-7 mm; nerves in 7-9 DE. co АНА 33. Michelia oe Tree to 28 m ig рево 9-25 mm ular scar 2-14 mm; nerves in 8-14; 8-12 т pid; ашп of leaf Чыг pur scars 7-10 mm long; scars of perianth and sta mens along torus under de 10 mm long. Michelia fulva mm n sm pu- obscure; stip- s of perianth t 2-5 Young twigs 1-3 m bescent; S аша rather ular scars mm long; sca ан stamens ее torus ер fruit ong. Leave glaucous beneath, twigs glabrous to mentose; fruits 13 Michelia wilsonii Leaves not glaucous а twigs pubes- cent; fruits 3.5-13 cm long. 22. Michelia velutina 18 Leaves glabrou Leaves hairy beneath atleast when young. 25 а under fruit 2-3 mm long; mid- prominent above, at least toward “о fruits 3.5-4 cm long; Tug 2-4. I= 2.5 cm with a stipe z 3-8 mm. ы ашы Michelia урарга ong; mi Gynophore under mu 8 30 mm rib not prominent a Leaves glaucous везије stipules: siky; twigs diam.; fruits 6-10 cm long; fol- licles sessile, 9-14 x 7-10 mm. Michelia = Leaves not glaucous beneath; stipules brous, pubescent, puberulous, or cost ud lous. 20 Scars of perianth and stamens along torus under fruit 20-22 m ү с 1-10 cm long; twigs 4-6(-10) mm diam. = 15. d lacei Scars of perianth and stamens along torus nder fruit 2-8 mm long. 2 Plant d Lp twigs 2-3 mm diam.; 5- x 3-7 cm; nerves in 7- 2 pairs; snis 1-3 em; bs 10-14 cm ал brachyblast 17- m Mic helia maudiae Plant hairy at least in innovations; brachy- blast hairy . Reticulation rather obscure; twigs 1-2 m iam.; leaves 4-12 x 1.5-3.5 cm; nerves in 8-15 pairs; fruits 5- И cm long; bra blast 3-8 mm long. .. 4. Michelia pu n Reticulation dis ied 2-7 mm diam.; ame 5.5-18 x 2-6.5 cm; fruits 3-10 m; brachyblast 6-25 mm Twigs glabrous, sometimes s pubescen nt, 2-4 m diam.; stipules pubescent; fruits h 1056 Annals of the Missouri Botanical Garden 29(28). 29. 30(29). e 182), — 34(32). 35(34). . leaves glaucous beneath. 1 3. y Pairs ог lateral nerves 16-18; Е. ç d th Ae: mee cm long; brachyblasts 13- 14 x 4-9 30. Michelia martinii Twigs puberulous or а. stipules p d ous, or tomentellous, leaf e or rounde _ Rerieulation of leaf laxly netted; twigs to- и petiole not dilated at base; iei m long; WIE has 6-12 mm follicles 6-15 x 8-10 mm. 28. Michelia chapensis Reticulation of leaf densely netted; tw mes puberulous; petiole dilated at base; fruit 3 6 cm long; к, E “Ж mm long; follicles 18-25 x ichelia coriacea Michelia ingrata Leaves not glaucous beneath. . Leaves 1.5-2 Б ст broad, apex rounded. eR TEES 26. Michelia angustioblonga меру 2.8-7.5 cm broad, apex acuminate twigs to- ] torus under fruit 12-14 mm long. .............. 1. Michelia aenea stamens along torus under fruit 3-10 m lon Fruiting brachyblasts 3 cm long; twigs to- s 6, gynoecium evinde: 29. Michelia жет P Fruiting brachyblasts 1-2.5 cm lon 5 gynophore 10- 25 mm. 2. Michelia то ЖЕ Twigs tomentellous; stipules tomentellous; leaves elliptic or ovate, 5-14 x 3-6.5 cm; fruits 2-5 cm long; gynophore 5- M mm. 3l pei 13-18; = 6.5-10 x etiole 1.5-2.5 . 19. Mic delia. mediocris Carpels 28; aves n = 14 х 4-6.5 cm; petiole 2.5-4 c 16. Michelia macclurei Leaves 15-25. cm n bros. A DD eaves 2.6-16.5 cm broad. ___________ Apex of leaf rounded, reticulation distinct; ronet to 4 m high; leaves narrowly oblong, x 1.5-2. 2 cm; petiole а 6. Michelia poblana Leaves ds an Ah 4-12 x 1.5- cm, apex acuminate or acute, reticu- lation rather cies A e 5-22 mm. . Michelia compressa Pairs of lateral nerves 16-19... 35 Pairs of pibe nerves 8-15. ._________ 37 а рау t, 2-5 т m diam.; .; leaves 8- 05 cm, Шар ку net- = pet 19 45 mm; nerves in 12-14 pairs; follicles 20- 65.5 ZEN FOE m Michelia и Twigs tomentose; reticulation of leaf den ly netted, die in 16-18 die a -14 14-25 x or 8-22 x 36(35). Sears of pom and stamens d eds: torus under mm long; stipules silky; curn of leaf rather obscure; follicles hairy eseten 10. Michelia oon 36. Scars of perianth and stamens m under fruit 12-14 mm long; s 1 reticulation of leaf ria ‘follicles labrous. Michelia aenea Ге puberulous, 1-2 mm dium: leaves -12 x 1.5-3.5 cm; see 5-7 . Michelia compressa Twigs pubescent or cae 38 Twigs tomentose; leaf apex acuminate or acute, reticulation rather obscure; fruits .9-14 cm long; follicles 8-22 x 7-14 mm. 10. Michelia DOM 38. Twigs pubescent, |. ape short-a nate, reticulation dist 26x leaves balon pesar 8-28 x 6.5 ст; petiole dilated at ч. ds i- am cm long; follicles 2-6.5 x 1.8 37(34). 37. 38(37). 27. Michele pem 39. pem finely scattered appressed-pubes cent beneath; petiole not dilated at base: fruits 5-53 cm long; follicles 6-14 o 100. 40(39). Reticulation of leaf laxly netted; BU. silky; fruit 10-53 cm long, iv o 100. ји masticata 40. Reticulation of leaf densely pen stipuise be ollicles pubescent; fruits 5-15 cm long, 14. : Michelia cavaleriei l. Michelia section Michelia. Michelia L. Champaca Adans. Sampacca O. Kuntze. Liriopsis Spach. Stipules free from or adnate to the petiole. Bracts 2-6. Tepals 6-21, 3—6-merous, subsimilar or dis- similar with those of the outer whorl, smaller or thinner in texture than those of the inner whorl or whorls. Fruits apocarpous. KEY TO THE SPECIES OF SECTION MICHELIA l. Stipules н to petiole. l. Stipules fre 2(1). Gynoecium vean than androecium, often entirely hidden, twigs in innovations 8-12 mm thick, нн E villous, stamens 33-39 mm long, filam m long, bei 24 E high. . 12. Michelia fulva 2; Gynoecium longer than androecium, twigs in innovations 1-3 mm thick, stipules pu- bescent, silky, or tomentellous, stamens 7- 4 mm long, filaments 1-4 mm long, gy- noecium 6-20 mm high. Leaves minutely кане) hairy beneath, . Michelia kisopa Leaves with longer hairs beneath, connec- tive appendage triangular ы Volume 80, Number 4 Chen & Nooteboom 1057 1993 Magnoliaceae of China 4(3). Gynophore 30 mm long, twigs silky, ма tose, young leaves with longer hairs ђе- silky, brachyblast 30— с m long, stamen neath, apex acuminate or acute, reticula- 20-24 mm long. . . Michelia elliptilimba tion rather obscure, densely netted, filaments 4. Gynophore 2- 15 m 1 wigs pubes- 4-6 mm long, fruiting brachyblasts 1.5-3 cent, tomentose, or to ndis me stipules cm long, fruiting carpels hairy. ..................... pubescent or tomentellous, brachyblast 3- 10. Michelia fov eolata 17 mm long, stamens 7-15 mm long. ............ 5 15(13). Outer tepals 1-3 cm long. 16 5(4) Reticulation rather obscure. 6 15. uter tepals 3.2-7 cm long. . 9. Reticulation distinct. 8 16(15). Leaves minutely (scattered) hairy be neath, 6(5). Pedicle present, twigs tomentellous. ............. twigs and stipules tomentellous ne mage 20. Michelia microtricha distinct, pedicle present, connective ap- 6. Pedicle absent, twigs tomentose. с 7 = oo 2-4 ae fila- 7(6). Stipules pubescent, leaf apex shortly acu me -2 mm long, и сућп- minate, fruiting brachyblasts 11 cm long, drie, es iin fruit 5-8 mm lon mm thick, fruits 13.5-15.5 cm long, 19. ее чу mediocris gynophore under fruit g^ 25 mm long. .. 16. Leaves with longer hairs benea igs pu RUMOR . Michelia wilsonii berulous, stipules pubescent, paire Vs o tomentellous, lat T acuminate rather obscure, pedicle absent, connective r acute, fruiting brachyblasts 0.5-0.8 c appendage triangular, 0.5-1 mm long, fil- lo ong, 3-4 mm thick, fruits 2.5-6 cm s aments 1 mm long, Ошу ршн ovoid, gyno- pasa Вы under fruit 10-15 mm long. . phore under fruit 10-17 mm long. ........... Зи floribunda 4. "Michelia compressa 8(5) Outer tepals 14-30 mm broad, tw 17(15). Leaves glaucous beneath, twigs n stip- stipules tomentellous. ..... б. Michelia dolisuja ules silky, gynoecium 20 mm high. ............. 8. Outer tepals Е 9 mm broad, Мода ата sty. tr 13, Michelia ingrata ules pubescen 17. Leaves not glaucous = wigs pubes- 9(8) Stipular scars E 5 mm long, leaves elliptic cent, tomentose, or entellous, stipules or narrowly elliptic, ик stout. pepe ог ae gynoecium 6- Michelia velutina 1З mm high. iii ee 18 9. Stipular scars 17-28 m Es leaves ovate 18(17). ds "of lateral nerves 16-18, t or narrowly ovate, mre slender. . tose, leaf base rounded, ins ps Michelia champaca –14, filaments 4 long, scars of peri 10(1). Gynoecium эн 11 anth and stamens along torus under fruit 10. Gynoecium hai 13 12-14 mm long. 2. 1. Michelia aenea 11(10). Plant entirely ise 18. Michelia maudiae 18. Pairs of ж nerves 10-15, twigs 11. Plant hairy at least in innovations, stipu escent, or tomentellous, leaf base cuneate ubescent ог tomentellous. а. or broadly iiri inner tepals 6-9, fila 12(11). cent s mm, twigs tomentose, stip- ments l- ong, scars of е апа ules cent, reticulation laxly netted, stamens е torus under fruit 4-7 mm ce not ано d at base, brachyblast 15=-– Jong, me 25 mm long, outer tepals 5. 57 cm long, 19(18). ie es with longer hairs beneath, twigs and 15-30 mm broad, stam ens 20-25 mm long, stipules tomentellous, leaves elliptic or ovate, gynoecium 17 mm high, gynophore 10- out als ovate or spathulate, connective m long, fruiting polus 5-10 appendage linguiform, gynoecium in- mm thick, fruits 7-10 cm long, scars of ric. . 16. Michelia macclurei perianth and stamens along torus under 19. Leaves minutely (scattered) hairy beneath, fruit 20-22 mm long, gynophore u twigs and stipules pubescent, leaves obovate fruit 29 mm long. ______ Michelia lacei or narrowly elliptic, outer tepals iced ог 12. Petiole 10-20 mm, twigs puberulous, stip- elliptic, connective rire trian ules tomentellous, reticulation densely net gynoecium ellipsoid. ... 2. Michelia cavaleri iei ted, petiole dilated at base, brachyblast 2.5 mm long, outer tepals 2.3-3 cm long, outer tepals 9-10 mm broad, stameris В 10 mm 1. Michelia aenea Dandy, J. Bot. 68: 211. 1930. long, gynoecium 10-12 mm high, gyn | | . phore 2-7 mm long, fruiting tlds TYPE: Poilane 12611 (holotype, P; isotype, 3-3.5 mm thick, fruits 3-6 cm long, scars K) of perianth and stamens along torus e fruit 4-6 mm long, уко under dou Michelia а os Chang & B. L. Chen, ^w Sci. LEE e lug. os Michelia coran is v. Sunyatseni (Guangzhou) 3: 86. 1987. 09) rui um 10-15 mum p NEP 14 E: B. L. Chen 868193 pe SYS). Gynophore 2-6 mm long. 15 Michelia longistyla Law . F. Wu, Acta Bot. Yun- TM Gynoecium 3-6 m nau twigs pubescent, 10, 3: 341, t. 7. 1988. TYPE: China. Yunnan: leaves minutely (scattered) hairy beneath, Jinping, 1,800 m, Q. Wang 142 (holotype, PE not apex shortly acuminate, reticulation dis- en). tinct, laxly netted, fila 3 mm ong, ; : fruiting EE 5- 7 Mis long, f fru ту Tree to 16 m high and 50 cm diam. Indument carpels glabrous. ........ clia pu 8 straight to slightly undulate, usually cupreous, 14. Gynoecium 17-20 mm oa те» tomen- sometimes clear to gray. Young twigs terete, ca. 1058 Annals of the Missouri Botanical Garden 4 mm diam., purplish brown, densely short-to- mentose, glabrescent; old ones dull dark brown, somewhat rough, = pubescent to glabrous, con- spicuously lenticellate and fissured. Stipules densely long-pubescent, free from the petiole. Leaves co- bright green and glabrous above, pale green, densely appressed-tomentellous below, oblong to narrowly oblong, 18.7-25(-28.5) X 4-6(-7.5) cm; apex shortly acuminate to acute; acumen 3-18 mm; base rounded or obtuse to cu- riaceous, rigidulous, neate or narrowly cuneate; midrib impressed above, prominent below; nerves in 16-18 pairs, anasto- mosing and meeting in an intramarginal vein, which is slightly netted reticulation, inconspicuous above, promi- more prominent than the fine, densely nent below. Petiole appressed hairy, glabrescent, slightly thickened toward its base, without scars, 1.3-2.8 cm long. Brachyblasts stout, finely to- mentose; spathaceous bracts 2, coriaceous, pubes- cent outside with same indument as stipules. Flow- ers white to yellowish; tepals 13-17, the outer 3 obovate, fleshy, 32-35 x 10-12 mm, the inner tepals 10-14, narrowly obovate to spathulate, 21- 27 3-16 mm long, connective appendage short-triangular, 0.5-2.5 X 5-7 mm; stamens many, mm long, filaments yellowish puberulent, ca. 4 mm long; gynoecium ellipsoid, tomentellous, exserted from the androecium, 8-13 mm long; gynophore 2-4 mm long; carpels many, ovoid; ovaries ca. | mm long; ovules 8-10; styles ca. 2 mm long. Fruiting brachyblasts pubescent, 3-3.5 cm long, ca. 5 mm wide at the top. Fruits cylindric, 6-10.5 cm long; ripe fertile carpels 4-25, obovoid to el- lipsoid, sessile to short-stipitate, 1.4-2.5 x 1-1.5 cm, apically short-beaked; gynophore 1.8-2.5 x .4 ст; scars of perianth and stamens along the torus 12-14 X 6 mm. Seeds not seen. Distribution. China and Vietnam. In CHINA. southeastern Yunnan: pos Maguan, Chen B.L. GS 302, Chen B.L. GS 8618 Ecology. In forests. Altitude: 1,800-1,950 m. Flowering April; fruiting September- October. Uses. Wood very hard and dense, generally used for pillars, planks, and other purposes. Collector's note. Bark rough, slightly brown- Note. The specimens examined are quite uni- form in all characters. There Michelia longistyla available; it is reduced to Michelia aenea based on the published description and figure. was no material of 2. Michelia cavaleriei Finet & Gagnep., Bull. Soc. Bot. France 53: 573, f. 1-8 of t. 1. 1906. TYPE: Cavalerie & Fortunat 2263 (ho- lotype, P; isotypes, E, K). шш ке Dandy, die Roy. Bot. Gard. E 0. 1928. TYPE: Hand.- Mazz. 1 (ho- b W). [The doveri material ee n Wang Te-Hui is not M. fallax. ] кы platypetala Hand.-Mazz., Akad. Wiss. Wien Rel | Hand.-Mazz. 12281 (holotype, W; iso- types, А, K, WU). [Only the flowering material is M. platypetala.] Much branched tree 10-20 m high and 50 cm diam.; bark white-gray; buds, young twigs, stipules, petiole, brachyblasts, and bracts outside densely appressed-pubescent wit ne, short to long, straight, clear to brown, glossy hairs; young twigs terete, dull black, 2-3 mm diam., later glabrescent; old ones brownish to brownish gray, smooth or longitudinally fissured, white lenticellate, glabrous to = pubescent; terminal buds narrowly ovoid, 1— 2 cm long. Stipules free from the petiole. Leaves thinly coriaceous, dark green, sparsely scattered- pubescent with short or long, straight or sometimes undulate, clear or brown hairs above, pale green, finely appressed-pubescent with short, straight, clear and brownish hairs, except the midrib beneath soon glabrescent, narrowly elliptic ог obovate-elliptic, (10-)12-20(-24) x 3.5-5.5(- 7) ст; apex short- acuminate, acumen 7-20 mm long; base cuneate to broadly cuneate; midrib conspicuously promi- nent below, = impressed above; nerves fine, visible on both sides, in 11-15 pairs, reticulation fine, closely netted. Petiole ca. 1.5-3 cm long, without stipular scars. Flower buds narrowly ovoid, 2-3 cm long; brachyblasts 5-15(-25) mm long, 3-5 mm wide at the top, pedicles 0-2 mm long; bracts 3-4. Tepals 9-12, white, the outer 3-4, obovate- elliptic, 4.5-5.6(-7) x 2-2.5 cm, the inner tepals narrowly obovate to spathulate; stamens 7-15 mm long, connective appendage triangular, ca. 1 mm long, filaments 2-3 mm long, ca. gynoecium ellipsoid, tomentellous, 6-10 mm long, mm wide; slightly protruding from the androecium; gyno- phore tomentellous, 4-5 mm long, ovaries ovoid, 1.5-2.5 mm long; ovules ca. 10; styles glabrous, 1.5-3 mm long. Fruits 5-15 cm long; ripe carpels 6-14 (some carpels abortive), ovoid to obovoid, dark brown, smooth to lenticellate, short-stipitate to sessile, with short beaks at the apex, 13-25 x 7-13 mm; gynophore 10-25 x 2-3 mm; torus with the scars of perianth and stamens 5-7 х 5- 6 mm. Seeds 1—4 in each carpel, ellipsoid to sub- 5-9 x 6-8 mm cordate, Volume 80, Number 4 1993 Chen & Nooteboom 1059 Magnoliaceae of China Distribution. CHINA. Fujian: Nanjung Xian, Ye. pA Li hen-yu | et Pa "1809: ugang, Lin oz 10 435. i 9. Sichuan: Fanching- 854. Yunnan: Fen ging, Zhu T.P. 554; Yuanyang, He S.C. 85170; Pin-fa, жы кеа J. 7312. Ecology. Їп forest. Altitude: 800-1,500 m. Flowering March; fruiting September-October. Collector's note. Trees in mixed forests on rock or open area on riverbank; flowers white, fragrant, gynoecium hairy, extending outside an- droecium; fruits greenish or brownish; seeds red- dish to red 3. Michelia champaca L., Sp. Pl.: 536. 1753. Michelia rheedii Wight, Ill. Ind. Bot. 1: 13. 1840. ТУРЕ: Hermann Fl. Zeyl. 144 (holo- type, BM). Michelia ano s N. L. Burm., Fl. Ind.: 124, 1768. Sampac кыч Rumph. SYNTYPES: Hb. Bur 5 (L). mann Michelia Па E Pers., Syn. Pl. 2: 94. 1806. Miche- lia blumei Steudel, Nomencl. ed. 2, 2: 139. 1841. Michelia tsiampacca var. blumei Moritzi in Zoll., Syst. Verz.: 3 46. Sampacca suaveolens O Kuntze, Rev. Gen. Pl. 1: 6. 1891. TYPE: Rumph., Herb. Amboin. 2: t . 1741. шүм 17 Blume, Bijdr.:7 1825, non Miche- iampacca L. 1767. 7. Michelia velutina Blume, Fl. ы э УКЫ 17. 1829, non Michelia velutina DC., 1824, Cha lutina O. Kuntze, Rev. Gen. Pl. 7 6. 1891. Michelia pilifera Bakh. f., Blumea 12: 61. 1963. TYPE: Blume s.n. (holo- туре, L, sheet no. 908.126-1868; isotypes, K, NY). Michelia pubinervia Blume, Fl. Javae Magnoliaceae: 14, champacca var. pubinervia ars Miq., Ann. Mus. 868. TYPE: Blume 670 (holotype, L; isotypes, B. Talauma villosa Miq. a Tn Miq., . Mus. ot. 0. 1868. TYPE: т Vriese & “ы ВО) Magnolia champaca Baillon ex Pierre, F. Forest. Coch- inch.: t. 3. 1880. TYPE: Herb. Pierre 187. (2) Magnolia membranacea P. Parm., Bull. Sci. France Bel- TYPES: Mauriti e cality unknown, D.J. Anderson, 1868 (MEL) Tree to 30—40 m high and 1 m diam.; indument composed of short, straight or slightly undulate, appressed or erect, clear to yellow, rarely brownish hairs; young twigs 2-3 mm diam., brown-gray, pubescent, glabrescent; old ones gray to black- brown, + lenticellate. Stipules pubescent, adnate to the petiole from about half of its length to nearly to the top, scars 1.7-2.8 cm long. Leaves thinly coriaceous, green, above glabrous, or at first es- pecially on the midrib sparsely pubescent, later glabrescent, beneath paler green, pubescent, gla- brescent, ovate-oblong to narrowly ovate, 8- 23(-34) x 3.4-9(-12) cm; apex acute to grad- ually long-acuminate; base attenuate to cuneate or more often rounded; midrib + impressed above, prominent below, nerves fine, visible on both sides, more prominent below than above, in 11-22 pairs. Petiole pubescent, glabrescent, 2-3.2(-4.5) cm ong. Flowers very fragrant; bud ovoid to ovoid- ellipsoid; bracts or pubescent outside; brachyblast usually slender, 7-12 x pedicles 0-1 mm long. Tepals 12-15(-21), yellow — or more rarely 4, tomentellous 2-3 mm, tomentose or densely pubescent, to orange, subsimilar, glabrous or pubescent outside at the base; the outer 4-6 narrowly obovate, nar- rowly obovate-linear, to spathulate, 20-40 x 7- stamens m long, connective ap- long, fila- 9 mm; pendage short-triangular, 0.5-1 mm ments 1.5-2 mm long; gynoecium ovoid-cylindric, 7-12 mm long; gynophore tomentellous, 3-5 mm long; carpels ca. 30, tomentellous; scars of perianth and stamens along torus 1-2 mm long. Fruiting brachyblast glabrescent, gray-brown, 10-30 x 3- mm. Fruit 2-15 more, ovoid to ellipsoid, densely lenticellate, gla- brescent, 10-20 x eak. Seeds subcordate, ellipsoid to irregularly к жаг 9-10 x m, 2-4 in each carpel. cm long; ripe carpels 4-19 or 7-11 mm, sessile and without India, China, Burma, Vietnam, and Malesia. In NA. Southern and Bani enorm Yunnan: Che-li Hsien, Wang C.W. 77767; Fo Hai, Wang C.W. 73905; Jengyueh, 25°N, 98°36' E, Forrest G. 25302. pice T ou Cultivated in Guang dong, Hainan, and Cia Distribution. Ecology. |n forests. Altitude: 650-1,600 m. Flowering June-July; fruiting September-October. Uses. Wood used for furniture and ships; the flowers and leaves are extracted for volatile oil; the tree is grown as an ornamental Collector's note. d ivory yellow to yel- low, du fragra uc (1817, 1824), Decaisne о о (1841), Moritzi іп Zollinger 1060 Annals of the Missouri Botanical Garden (1846), and Miquel (1858) misspelled the name as Michelia tsiampacca (auct. non L., 1797). For further misapplied names and nomina nuda see Nooteboom (1985) Forest — Michelia Pure (Maxim.) Sarg., Fl. Japan. 4 77, t. 13. 1893. Magnolia compressa Maxim., Bull, Acad. Imp. Sei. Saint- 417. 1872. TYPE: Yagami, near Nagasaki, Hizen Prov., Kiushu, Maxi- mowicz, 1862 (holotype.?: isotype, BM). Petersbourg 17: Magnolia philippinensis P. Parm., „Эш, Sei. France elgique (Parmentier) 27: 206, 270. 1895. Miche- lia ЛЕ (P. Parm.) Dendy, Kew Bull 1927 . Michelia cumingii Merr. & Rolfe, Phil i Sa 3 Bos d 100 TYPE: " Cuming 783 (holotype, MEL; isotypes, A, BM, K, L, NY). Michelia compressa var. formosa Kanehira, Trans. Nat. ist. S al 384. 1930. Michelia for- mosana (Kanehira) Мана & Suzuki, Annual Rep. Taihchu Bot. Gard. 3: 57. 1933. Michelia taiwaniana Y. W. Law, Iconogr. Cormoph. Sini- corum |: LM t. 15 592 2.1972, nomen 1 illeg. ds helia Y. Wu u ex Law & Acta Bot. ud a 10, 3: 337, 1. 7. ГҮРЕ: Lo 1359 (holotype, KI n. 1988. voung twigs terete, slender, dark black-brown, smooth and appressed-puberulent with Small tree, 8 1-2 mm diam., high; dull brownish to short, straight, clear, brown to rufous hairs; old ones purple-brown, glabrescent, sparsely lenticel- l1 mm ate; terminal buds obovoid to cylindric, 8 long, densely appressed-pubescent with gray or rufous, glistening hairs. Stipules pubescent, free from the petiole. Leaves chartaceous to coriaceous, glabrous on both surfaces, or at first sparsely pu- bescent with long, straight and slightly undulate especially on the hairs on both sides, below, midrib and toward the base, densely hairy with long, straight, erect, rufous hairs, finally glabres- 12 3.0 15 mm cent, elliptic to narrowly elliptic, 4 1.5 cm; apex acute to acuminate, acumen | long, sometimes obtuse and occasionally. slightly emarginate, base cuneate, rarely obtuse and often slightly unequal; midrib slightly impressed above: nerves in 8-15 pairs, obscure, slightly more prom- inent than the rather fine reticulation below, retic- ulation densely netted. Petiole slender, glabrous or with same indument as young twigs, without stip- m ular scars, 5-22 mm long. Brachyblast tomentel- lous with gray to brown hairs, 3-6 x 1.5-3 mm, pedicle absent; bracts 2-3, pubescent outside as are terminal buds. Flower white to creamy yellow, much scented; tepals 9( 14), subsimilar, the outer 3 oblanceolate to oblanceolate-linear, mostly broad- er than the inner tepals, 10-28 x 3-10 mm; stamens many, 2.5-7 pendage triangular, 0.5-1 mm long, filaments ca. mm long; gynoecium ovoid to ovoid-oblong, to- mm long, connective ap- — mentellous, 4-6 mm long, much longer than the androecium; gynophore with same indument as gynoecium, 2-6 mm long; carpels 5-13 or more, ovaries ca. 1.8 mm long; styles ca. 0.2 mm long; ovules 2-8. Fruiting brachyblast gray hairy, 3-8 1.5 12 or more, ovoid to ellipsoid, dark brown, sessile 8 mm. Fruit 5-7 cm long; пре carpels | or with a ca. 2-9-mm-long stipe, sparsely white- lenticellate, apically short-beaked, 11-15 x 8-14 + pubescent, 10-17 x 2-4 mm; torus with the scars of perianth and stamens, 2 3 x 2.5-6 mm. Seeds 1-2, to irregularly shaped, 7-9 x 6-7 mm; gynophore subcordate, ellipsoid mim. ~ ~ > Distribution. China, hu е ае ч southern Japan. In CHIN. Xu Taihokushy Taipei, Kankou, His e M.T. Kac Fy : Bon Bon, Aen Liao Jih-C ne 10006; Botel pon Chane E, ; Henchun Peninsula, Chang СЕ. . Also eastern t southern Yunnan. ~ Altitude: 200- 2,000 m. Flowering, fruiting January. ses. Wood straight-grained, heavy and hard, used for building, furniture, ships, Ecology. In forest. fine-textured, woodcarving, and other purposes. Vote. of all the cited synonyms from Japan, After comparative studies of collections China, and the Philippines, we came to the conclusion that they all belong to a single species: Michelia com- pressa is the most widely distributed species of the genus. It is noteworthy that Michelia compressa shows much variation throughout its area in size of the flowers, number of carpels and ovules, as well as length of the stipe of the ripe carpels. The flowers are smaller in the specimens of the Philippines, larger in Taiwan and Japan. Both carpels and ovules from Taiwan and Japan are slightly more in number than from the Philippines. The length of the stipe of the follicles varies from 8 mm. These ~ — ~ - variations can be observed even in the same spec- imen. Gagnepain (1929) non Loureiro, named this species Michelia figo (auct. 1790). Merrill (1906) used the name Michelia parviflora (non DC., 817) 5. Michelia coriacea Chang & B. L. bur Acta Sci. Nat. Univ. Sunyatseni 3: TYPE: В. L. Chen & B. Li GS 90. 35 ) oa SYS). Figure 9. Volume 80, Number 4 Chen & Nooteboom 1061 1993 Magnoliaceae of China FicurE9. Michelia coriacea Chang & B. L. Chen. — 1. Fruiting branch. — 2. Seed. — 3. Seeds with testa removed. Based on SYS 161742 — Chen Bao Liang GS 9035. Drawing by Xie Qing Jian. 1062 Annals of the Missouri Botanical Garden NY FicunE 10. Michelia doltsopa Buch.-Ham. ex DC.— 1. Fruiting branch. — 2. Flower. Based on SYS 161753 = Chen Bao Liang 86 S 049. Drawing by Xie Qing Jian. ulent; old ones dark brown, glabrescent, with sparse orbicular and white lenticels; terminal buds ovoid, densely tomentellous with short, straight, clear, gray to brownish hairs. Stipules with same indu- Tree 10-18 m high and 20-45 ст diam.; young ment as the terminal buds, free from the petiole. twigs dull brown, 3-7 mm diam., + gray puber- Leaves coriaceous, green, glossy above, pale green Michelia polyneura C. Y. Wu ex Law & Y. F. Wu Acta Bot. Yunnanica 10(3): 340, f. 1-8 of t. 5. ug. 1988. TYPE: К. M. Feng 12030 (holotype, KUN; isotype, A). Volume 80, Number 4 Chen & Nooteboom 1063 Magnoliaceae of China beneath, both surfaces glabrous, oblong to broadly elliptic, sometimes obovate, often unequal, 9.7- 13.9(-16.1) x 2.2-3.6(-6) cm; apex acute to acuminate, acumen 5-10 mm long, base cuneate or rounded; midrib slightly impressed above, rather prominent beneath; nerves in 12-13(-16) pairs, much visible on both sides, curved upward and meeting in an intramarginal vein, all lesser veins forming together a dense reticulation that is prom- inent on both surfaces, the larger alveoli filled with the smaller veins, which are obvious to obscured. Petiole glabrous, dilatate toward the base, 1-2 cm long (mature » flowers not seen). Flower buds ellipsoid, tepals 9 white, subsimilar, obovate to narrowly obovate, 2.3-3 x 0.8-1 cm; stamens ca. 80, 8-10 long, connective appendage narrowly triangular, 1.5-2 mm long, filaments 2.5-3 mm; gynoecium glabrous, subcylindric, exserted from the androe- mm cium, 1-1.2 cm long; gynophore 2-7 mm long. Fruiting brachyblasts yellowish pubescent, with 3 scars of bracts, 8-2 mm, pedicles absent or very short, ca. 1.5-2 mm long if present. Fruits 3-6 cm long with 1—4(-10) fertile carpels and 6– 8(-17) abortive carpels, the ripe carpels sessile, ovoid, dull and dark brown, conspicuously lenti- cellate, 18-25 x 12-17 mm; gynophore 1.2-1.5 cm long, scars of perianth and stamens along 4- © mm of the torus. Seeds ellipsoid, smooth, 10- 11 x 6-7 mm. Distribution. CHINA. Southeastern Yunnan Guangnan, Wang C.W. a Malipo, 86915; Si Chour Hsien, Feng K.M. 1203 Ecology. In forests, on limestone formations. Altitude: 1,450 m. Flowering April; fruiting Sep- tember-October. Note. The type of Michelia coriacea bears fruits. [n the absence of flowers, we misrepresented the tepals as 6-7 in number, based on the faint scars of the perianth on the torus. The collections with flowers collected from the same tree one year later showed that the flower has 9 tepals. 6. ccn TUBE Buch.-Ham. ex DC., Syst. Nat 448. 1817. TYPE: Buchanan-Ham- Pado s.n. pa ba BM). Figure 10. Michelia excelsa (Wall.) Blume, Fl. Javae Magnoliaceae: 9. Magnolia excelsa Wall., Tent. Fl. e TYPE: Wallich 6494 (holotype, K; eai BM, NY). ж calcuttensis P. Parm., Bull. Sci. France Bel. Военно 27: 283. 1895. ТУРЕ: Type spec- imen not traced. According to P. Parm., its label said: “Michelia ?Assam. ex herb. [ue bot. calcut- tensis." Michelia Ll dais qi Watt ex Brandis, Indian Trees: 8. 1906. TYPE: С. Watt 6329 (К). Michelia wardi Dandy, Kew Bull. 1929: 222. 1929. . Kingdon Ward 8060 нш Ж К). Michelia opipara Chang , Acta Sci. Nat. niv. Sunyatseni (Guang sho "EN 90. 1987. ТУРЕ: B. L. Chen GS86237 (holotype, SYS). Tree or rarely a shrub, to 30 m high; buds appressed-rufous or gray tomentellous, later gla- brescent; young twigs 2.5-3 mm diam., tomentellous, glabrescent, brownish to dull black- brown. Stipules appressed-rufous or gray tomen- sparsely tellous, adnate to the petiole, scars 2-6(-20) mm long. Leaves chartaceous to thinly coriaceous, green, olive green to brown when dry, above scat- tered-pubescent with slender, straight to undulate, clear hairs, glabrescent, beneath pale green to glau- cous, appressed-tomentose with short or long, straight to occasionally undulate, usually rufous, sometimes clear to brownish hairs, elliptic, oblong- elliptic, oblong to narrowly ovate-elliptic, usually = unequal, 9-22 x 4-8 cm; apex acuminate or subacuminate, acumen 1—3 cm long, base broadly cuneate or rounded, the margin slightly wavy; mid- rib impressed above, prominent below, nerves prominent below, in 10-14 pairs, reticulation fine, densely netted. Petiole pubescent, glabrescent, 0.8- 2.5 ст long. Brachyblasts Б. appressed-ru- fous to gray tomentose, 4-13(-23) x 4-6 mm, pedicles 2-5 mm long; bracts 2-3(-5), densely rufous tomentose. Flowers with pleasant smell; buds ovoid to oblong, 4-5 cm long; tepals (8-)12-16, white to yellow, the outer 4—5 obovate to spathu- late-obovate or oblanceolate, 3.7-7.5 x 1.4-3 ст, glabrous to pubescent outside toward the base; stamens 8-15 mm long, connective appendage short to long triangular, 1-3 mm long, filaments 2-3 mm long; gynoecium subcylindric, 1-2 cm long, exserted from stamens; gynophore 5-12 mm long, tomentellous; carpels many, subovoid, ca. 1.5 cm long, appressed-tomentellous; styles dn scars of perianth and stamens on torus 4-6 m long. Fruiting brachyblasts Adde or = pu ee cent, 9-13(-23) x 4- Fruits (2-)5-12. 5 cm long, some carpels abortive; ripe carpels sessile or substipitate, 10-15 x 8-12 mm; gynophore 10-22 x 2-4 mm, scars of perianth and stamens under fruit 7-10 x 4—7 mm. Seeds ovoid to el- lipsoid, 8-11 x 8 mm. mm. Distribution. Nepal, Bhutan, northeastern India, China, and northern Burma. In CHINA. Southern Xi- zang: Kingdon Ward F. 19299, 19211. Yunnan: Shweli- Salwin divide, 24°20’, Forrest С. 8949; Jinping, Sino- i ng, Yu ; Shange-pa, Tsai H.T. 5 Hila, aa a Yu T.T. 16662; m divide, Yu 1064 Annals of the Missouri Botanical Garden T.T. 20866; Yengyueh, 25?25'N, 98?30'E, Forrest G. 26258. Ecology. In broad-leaved forest. Altitude: 1,600-2,400 m. Flowering April-May; , fruiting September- October. Uses. Wood light, soft, fine-textured, very du- rable and easily workable; used for various objects of construction work and carpentry, as well as for planking, doors, window frames, and furniture. Collector's note. fragrant. Note. See Dandy (1927ђ). Michelia opipara is similar to Michelia doltsopa except for its gray to yellowish indument. The former name is there- fore reduced to a synonym of the latter. Flowers pale yellow or white, 7. Michelia elliptilimba B. L. Chen & Noo- teboom, sp. nov. TYPE: China. Yunnan: Meng- hua, Chukai, alt. 2,000 m, 12 May 1938, T. T. Yu 15854 (holotype, A; isotypes, E, IBSC). Figure 11 Arbor 5 ad 6 m alta, ramunculis, stipulis extra, foliis rae petiolibus, brachyblastis, bracteis, gy- nophoris, ovariisque in 4. Flores pedicello 5 m longo, tepalis exterioribus 3, obovatis, 5.5 .5 ст longis, tepalis interioribus 8, obovatis ad spathulatis, ca. 5.7 cm longis, staminibus 20 ad 24 mm longis, gynaecio cylindrico 15 mm longo gy- nophoro 30 mm longo et 6 mmm lato. Tree 5-6 m high; bark greenish gray, smooth; young twigs, stipules outside, leaves on both sides, petioles, brachyblasts, bracts, gynophore, and ova- ries covered with fine, short or long, Pu brownish to brown hairs. Young twigs 2-3 m iam., old ones + hairy or glabrous; terminal buds ca. 15 mm long; stipules adnate to the lower bases of the petiole, stipular scars 3-4 mm long. Leaves thinly coriaceous, green above, glabrescent and becoming glaucous beneath, elliptic to ovate-ellip- tic, 16-22 x 7.5-10 cm; apex acute or acumi- nate, base obtuse; midrib impressed above, con- spicuously prominent below, nerves visible on both sides, more prominent beneath than above, in 9— 14 pairs, reticulation closely netted, prominent on both sides. Petiole 2-2.5 cm long. Flowers creamy white, with a pleasant scent; brachyblasts with 3- 4 scars of bracts, 3-3.5 cm lon —6 mm wide at the base, pedicles 5-7 mm long; tepals 11-12, subsimilar, the outer 3 obovate, attenuate toward the base, short-clawed, 5.5-6.5 x -3 cm, the inner tepals 8, obovate to spathulate, ca. 5.7 X 1.5-2 em; stamens 20-24 mm long, connective appendage triangular, ca. 0.5 mm long, filaments 2-4 mm long; gynoecium cylindric, much longer than the androecium, ca. 15 mm long; gynophore a. 30 x 6 mm; carpels many, ovaries ovoid, 5 mm long; styles glabrous, 2.5-3 mm long. Fruits not seen. CHINA. Yunnan: Meng-hua, Chukai, о 854 T. T. Yu Note. This species resembles Michelia bail- lonii in the vegetative characters except in its glaucous lower leaf surface. The flowers, however, are very similar in size and shape to those of Michelia pend Only one collection available. 8. Michelia flaviflora Law & Y. F. Wu. See under dubious species. 9. Michelia floribunda Finet & Gagnep., Bull. Soc. Bot. France (Mémoires) 4: 46, f. b of t. 7. 1905. ТУРЕ: Bons d'Anty 30 (holotype, P; isotypes, K, E). Mem kerrii Craib, Kew Bull. 1922: 166. 1922. TYPE: r 4679 (holotype, K; isotypes, BM, K, P, UC). Tree to 28 m high and 1 m diam.; bark gray; young twigs slender, 1.5-2 mm diam., brown or purple-brown to purple-black, at first tomentose with short or long, straight, glistening, brown, yel- lowish to gray hairs, soon glabrescent. Stipules appressed-tomentellous with same hairs as young twigs, adnate to the petiole, stipular scars 2-14 mm long, sometimes to 2.2 cm long. Leaves thinly coriaceous to coriaceous, dark green, glossy, at first sparsely appressed-pubescent with slender, clear hairs, especially on the midrib, later glabres- cent above, glaucous or not, appressed-pubescent with long to rather long, straight, clear, brown, rarely dark brown hairs, glabrescent beneath, nar- rowly elliptic, narrowly obovate-elliptic, ovate-el- liptic, oblong-elliptic to narrowly ovate; 6- 14.5(-17) x 1-4.5(-5) em; apex acute to acu- minate, sometimes + caudate, acumen 5-15 mm long, occasionally to 2 cm long, base cuneate or rounded; midrib impressed above, nerves fine, ob- scure on both sides, more prominent beneath than above, in 8-14 pairs, reticulation fine and dense, hardly visible on both sides. Petiole appressed-to- mentellous or pubescent, 9-25(-40) mm long. Brachyblast covered with same indument as young Volume 80, Number 4 Chen & Nooteboom 1065 19 Magnoliaceae of China FIGURE 11. Michelia elliptilimba B. L. Chen & Noot.— 1. Branch with deflorated flower. — 2. Flower. — 3. Stamen. — 4. Follicle. Based on T. T. Yu 15854 (A). Drawing by Joop Wessendorp, Rijksherbarium, Leiden. twigs, 3-6(-8) x 2.5-4 mm, pedicle absent; bracts ilar, glabrous to puberulent outside toward the 3(-4), brown or silver sericeous outside. Flower base, spathulate, broadly obovate, obovate-linear, bud ovoid to cylindric, 2-4.4 cm long; flower fra- rarely narrowly rhombic, tapering toward the base, grant; tepals 12-15, outer tepals 3, white, subsim- usually basally with short or long claw, 19-45(-50) 1066 Annals of the Missouri Botanical Garden X 6-18(-23) mm; stamens 7-11 mm long, con- nective appendage short- or long-triangular, 1-3 mm long, filaments ca. 2 mm long; gynoecium cylindric, longer than the androecium, 9-14 mm long; gynophore brown to gray tomentellous, 3— 15 mm long; carpels tomentellous; styles glabrous; ovules 3-8 in each carpel; torus with scars of perianth and stamens 2-6 x 3-5 mm. Fruiting brachyblast pubescent, 5-8 x 3-4 mm. Fruit 2.5- 6 cm long, contorted; ripe carpels 1-14, subglo- bose, brown to dark brown, lenticellate, + puber- ulent, apically beaked, 7-20 x of perianth and stamens 3-5 x 3-4 mm. Gyno- phore under fruit pubescent, 1–1.5 cm. Seeds el- lipsoid, cordate to irregularly shaped, 6-8 x 5-6 mm. Distribution. China, Thailand, Vietnam, Laos, and у ao Shan .W. 306. Yunnan: Ma Chang Kai valley, Forrest G. 9460; Mienning,Poshang, Yu LT hsien, Wang C.W. g De Tengyueh, Rock J.F. 7933; Szemao, Rock LE Ecology. In forests. Altitude: 800-2,700 m. Flowering February—June; fruiting July-October. Collector’s notes. Flowers white to creamy yellow; fruits green, with red fleshy seeds. М The ranges of both Michelia doltsopa and Michelia floribunda overlap in western Yun- nan. Some collections are difficult to identify and probably are hybrids. L. Diels. (1912, 1913) and Hu in Hu & Chun (1929 pro parte) erroneously dealt with this species under Michelia wilsonii (auct. non Finet & Gag- nep. 1905). Craib (1925a) named it Michelia man- ipurensis [auct. non Watt ex Brandis. 1906: Kerr 4926 (BM, E, К)]. 10. Michelia Сана Мегг. ex Dandy, J. Bot. 66: 360. 1928. TYPE: W. T. Tsang & К. C. Wong 2738 (in C. C. C. 14599) (holotype, UC; isotype, SYS). Michelia fulgens Dandy, J. Bot. 68: 210. 1930. TYPE: Poilane 7092 (holo iis : Jus ae K). Michelia foveolata var. ciner s Law & Y. F. Wu, Bull. Bot. Res. (China) 6, 2: 99. 1986. TYPE: wo Zhejiang: Qingyuan Xian, X. Wu 7720 (holotype, IBSC not seen). Tree to 45 m high and 90 cm diam.; bark pale gray or dark gray; young twigs yellow or brown, appressed-tomentose with short, straight, glossy, clear to yellowish or brown hairs; old ones purple- brown, + pubescent, lenticellate. Stipules brown tomentose or sericeous, free from the petiole. Leaves coriaceous, rigid, above dark green, glossy, ap- pressed-pubescent with slender, straight, (when young) to clear (when old) hairs, especially on the midrib, below appressed-tomentose with short, straight, silver to brown hairs, glabrescent, considerably variable in shape and size, narrowly to broadly obovate, obovate-elliptic, narrowly ob- ovate-elliptic to elliptic-oblong, 9.5-26 x 4.5- cm; apex acute to acuminate, base cuneate brown TO or obtuse to rounded, sometimes nearly cordate, + oblique; midrib impressed above, prominent be- neath, nerves visible on both sides, in 12-16 pairs, reticulation dense, seemingly foveolate, obscure on oth sides. Petiole tomentose, glabrescent, 1.5- 3.5 ст long, without scars. Brachyblast tomentose with same indument as young twigs, 13-25 x 4- 7 mm, pedicle absent; bracts 3-4(-5), densely brown sericeous outside. Flower bud oblong, ca. 3.7 cm long; flower fragrant; tepals 9-12, white or greenish white, subsimilar, broadly obovate to spathulate, 3-4.5 x ward the base; stamens 17-20 mm long, connec- tive appendage rather short, ca. 0.5 mm long, filaments 4-6 mm long; gynoecium cylindric, ex- 1.6-2.4 cm, pubescent to- serted from stamens, 1.7-2 cm long; gynophore tomentose with silver to yellow hairs, 1.2-1.5 x 0.2 cm; carpels many, tomentose with the same hairs as gynophore; ovules 10 or more; scars of perianth and stamens along torus ca. З х З mm. Fruiting brachyblast 15-30 x 4-7 mm. Fruit 5.5- 14 cm long; ripe carpels 2-12, subglobose, ob- ovoid, black-brown, + pubescent, lenticellate, stip- itate to short-sessile, beaked, 8- x 7-14 mm; gynophore tomentose, 17-30 mm long; scars of perianth and stamens 5-7 X 4-5 mm. Seeds sub- globose to ellipsoid, 7-8 x 5-7 mm. Distribution. Vietnam, China. In CHINA. Guang- dong: Chun W.Y. 7179. Guangzhou: Ah Po Shan, Taam Y.W. 572; Chang Ki Tong, Shi Shan, Yuyuen, Kwok S.P. 80335; Kook Kiang, Ko S.P. 50410; Lok Chong, Tso C.L. 20354. Guangxi: Dayao Shan, Li 7.0. 99: Debao, Li Z.T. 604032; Jinxui Xian, Da Yao Shan Exped. 11080; Lingui, Liang C.F. 31751; Longshen, Zhang, B.N. 406132. Guizhou: Jiangkou Xian, Sino- American Guizhou Bot. Exped. 1077; Xingren Xian, Dang C.Z. 1618. Hainan: Bak Sa, Lau S.K. 26589; Baoting, Hainan Exped. 752; Fan Yah, Chun N.K. & C.L. Tso 44233; Hung Mo Shan, McClure F.A. (LU 18302) 768; Miu Tsuen, McClure Е.А. (LU 18246) 712. Hunan: 7am P.C. 6249 4 7066; Pingpien, Mao P.I. 101762. Also reported from Fujian, Jiangxi, and Zhe- jiang. Volume 80, Number 4 1993 Chen & Nooteboom 1067 Magnoliaceae of China cology. In evergreen broad-leaved and mixed "Ih Altitude: 500-1,800 April; fruiting September- November straight-grained and fine-tex- tured, suitable for building, furniture, and plywood. Collector's white or yellow; fruits greenish, light gray, or pur- m. Flowering March- ses. 00 notes. Fairly common, flowers Michelia foveolata is a recognizable species. However, the shape and size of the leaves and color of the indument varies. 11. Michelia fujianensis Q. F. Zheng. See under dubious species. 12. Michelia fulva Chang & B. L. Chen, Acta Sci. Nat. Univ. Sunyatseni (Guangzhou) 3: 87. 1987. ТУРЕ: B. L. Chen GL86-193 (holotype, SYS). Figure 12. Tree to 15 m high and 15 cm diam.; buds, young twigs, stipules outside, leaves beneath, pet- iole, brachyblasts, and bracts outside densely ap- pressed-hispid with short to long, straight to un- dulate, somewhat glistening, fulvous hairs; young twigs 8—12 mm diam., dull brown, rough; old ones provided with lenticels, longitudinally fissured. Stip- ules adnate to the base of the petiole, stipular scars 7-10 mm long. Leaves coriaceous, rigid, bright green, above glabrous to + pubescent, below pale green to glaucous, glabrescent when old, elliptic, ovate, obovate to oblong, (14-)18-24(-29) x 7- 10(-12) cm; apex acuminate to acute, base cu- neate, sometimes rounded; midrib impressed above, (-14) pairs, meeting in a looped intramarginal vein; re- much prominent beneath; nerves in 7-11 ticulation densely netted, both nerves and reticu- lation prominent on both sides. Petiole stout, di- latate toward the base, 2-3 cm long. Brachyblast 2-2.2 x 1-1.2 cm, pedicle absent; spathaceous bracts 2-3, broadly elliptic, ca. 52 x Flower bud long ovoid, ca. 3.7 х 2 cm; tepals 12-13, white, subsimilar, fleshy, obovate to ellip- tic, the inner tepals smaller than the 3 outer ones, 4.4-5.3 x 2-3 cm; stamens са. 146, 33-39 mm long, filaments 7-8 mm, anthers 25-30 mm, con- nective appendage short-triangular, ca. 1 mm; gy- noecium cylindric, ca. 24 mm long hidden within and much shorter than stamens; gynophore pu- bescent, ca. 6 mm long; carpels ca. 152, fulvous pubescent, ovaries 2-3 mm long; styles glabrous, 1-2.5 mm long, scars of perianth and stamens on torus ca. 10 mm long. Ripe carpels ovoid, sessile, 10-15 x 9-10 mm Distribution. CHINA. Yunnan: Maguan, B. L. Chen GL86-193. Ecolo In forest on limestone formations. Altitude: 1,690 m. Flowering April. Collector’s note. The natural color of the te- pals is not known eer because the flowers мое seemed nearly he Perm characters of this species are: plant covered with fulvous hispid hairs; leaves large and rigid, pale green to glaucous be- neath; flowers quite big 13. Michelia ingrata B. L. Chen & 5. C. Yang, cta Sci. Nat. Univ. Sunyatseni (Guangzhou) 3: 95. July 1988. TYPE: B. L. Chen & C. М. Mai 87 T 019 (holotype, SYS). Figure 13. Michelia calcicola C. Y. Wu ex Law & Y. F. Wu, Acta Bot. Yunnanica 10, 3: 339, f. 9-16 of t. 5. Aug. 1988. TYPE: C. W. Wang 87716 (holotype, KUN). Tree to 16 m high and 40 cm diam. Young twigs dull brown, stout, 5-7 mm diam., villous with long, straight, crisped to undulate, brown to rufous hairs, soon glabrescent; old ones black-brown, fis- sured and lenticellate, + hairy to glabrous; terminal buds narrowly ovoid, yellow to brown tomentose. Stipules coriaceous, pubescent with long silky hairs outside, free from the petiole. Leaves coriaceous, rigidulous, dark bright green, at first scattered- pubescent over both surfaces and pilose along the midrib, later glabrescent above, beneath glaucous, appressed-tomentose with same hairs as the young twigs, sooner glabrescent or only pubescent along midrib and nerves, or glabrous everywhere, elliptic to obovate, the margin slightly revolute and bony; 13-21.5(-24) x 4.5-9.5 cm; apex acute to short- acuminate, acumen 3-5 mm long, base cuneate to subrounded; midrib impressed above, prominent beneath; nerves in 12-14 pairs, visible on both sides, reticulation densely netted, prominent on both sides. Petiole pubescent, glabrescent, 2.7- cm long. Brachyblasts covered with same indument as young twigs, 3-4 x 0.6-0.8 cm, pedicles ab- sent; spathaceous bracts 3, coriaceous, tomentose outside. Flower golden-yellow, malodorous; tepals 9—12, subsimilar, the outer 3 obovate, glabrous, ca. 5.7 X 3.5 cm, the inner tepals 6-9, narrowly elliptic to spathulate, 4-4.5 x 1.9-2.1 cm; sta- mens ca. 60 or more, 17-22 mm long, connective appendage triangular, 0.8-1 mm long, filaments 4-6 mm long; gynoecium cylindrical, golden pu- berulent, ca. 2 cm long, exserted from stamens; gynophore ca. 6 mm long; carpels ca. 63, ovate, -4 mm long; styles 1.5-3 mm long, glabrous. 1068 Annals of the Missouri Botanical Garden tem == => A ца == A == —— € == === == CR == MES X VM ЕС Em ~~~ us SS SSS > SSS LAS FIGURE 12. Michelia fulva H. T. Chang & В. L. Chen. — 1. Flowering branch. — 2. Underside of leaf. — 3. Bract.— 4. Deflorated flower with gynoecium and part of androecium.— 5. Outer tepal.— 6. Stamen.— 7. Follicle. Based on SYS 161751 — Chen Bao Liang G1 86-193. Drawing by Xie Qing Jian. Chen & Nooteboom 1069 Volume 80, Number 4 1993 Magnoliaceae of China И M A ni WA j d NIN WAAN SES c FIGURE 13. Michelia ingrata B. L. Chen € S. C. Yang.— 1. Flowering branch. — 2. Fruit. — 3. Deflorated flower showing gynoecium.— 4. Undersurface of leaf. Based on SYS 161515 — Chen Bao Liang 87 T 019. Drawing by Xie Qing Jian. B.L. & C.N. Mai 87 T 19; Malipo, Wang C.W. 87033. Fruits 6-10 cm long, the fertile carpels sessile, Also in Xichou and Guangxi (Longzhou). 8.9-14 x 7-10 mm, with 2-5-mm-long beak. Gynophore under fruit 17 х 4 mm. Seeds 2-3 in each carpel. Ecology. In thickets on limestone. Altitude: 1,600-2,000 m. Flowering March-May. Distribution. CHINA. Eastern and southern Collector’s note. Flowers with putrid smell. Yunnan: Goang-nan, Wang C.W. 87715; Maguan, Chen 1070 Annals of the Missouri Botanical Garden Note. This species varies in size of the leaves yellow-brown, sessile, beaked apically, + puberu- and length of the peduncles. 14. Michelia kisopa Buch.-Ham. ex DC. Syst. Nat. 1: 448. 1817. ана e Ham. ex DC.) Kuntze, Revis. Gen. 1891. ТҮРЕ: Nepal, Narain-hetty, 2 Do 1802, Buchanan-Hamilton s.n. (BM). Michelia zila Buch.-Ham. ex Madden, Trans. Bot. Soc. Edinburgh 5: 127. 1858. TYPE: Buchanan-Ham- ilton 1262 (E). Tree, bark gray; young twigs 1.5-3 mm diam., yellow-brown, purple-brown, to dull brown-black, sparsely minute-puberulent or tomentellous with short, straight to slightly undulate, clear hairs, gla- brescent; old twigs brown, gray-brown, to gray- black, longitudinally fissured or wrinkled, sparsely lenticellate. Stipules tomentellous with straight, clear hairs, adnate to the petiole usually up to or beyond half its length, scars 1.4-2 cm long. Leaves thinly coriaceous to coriaceous, bright green, sparsely puberulent, especially on the midrib and toward — c @ n pubescent with straight, clear hairs, glabrescent beneath, oblong-elliptic, obovate-elliptic, oblong, ovate-oblong to ovate, somewhat unequal; 5.2- 16.5(-19) x 2.2-5 cm; apex acuminate to long- acuminate, acumen 7-12(-20) mm long, occa- sionally acute, base cuneate to rounded; midrib slightly impressed above, nerves in 10-15 obscure or hardly visible on both sides, reticulation airs, prominent on both sides. Petiole slender, puberu- 1.6-3.5 ст long. Brachyblast densely tomentose with short, ~ to slightly lent, glabrescent, undulate, clear or gray hairs, 4— -5 mm, pedicle absent; bracts 3(-4), е with same hairs as the brachyblast. Flowers not or faintly scented; tepals 9-15, white to yellow, gla- brous or puberulent outside toward the base, sub- similar, the outer 3 obovate, obovate-oblong, to spathulate, tapering toward the base, becoming short or long clawed, 3-3.5 x 1–1.3 cm, the inner tepals narrowly obovate-elliptic, 2.2-2.7 x 0.4- 0.7 cm; stamens 8-11 mm long, connective ap- pendage usually linguiform or long-triangular, 2- 4 mm long, filaments 1-2 mm long; gynoecium cylindric, longer than the androecium, ca. 6-10 x 2 mm; gynophore yellowish tomentellous, ca. 1-4 х 1 mm; carpels ca i the same indument as the gynophore; styles short . 20, ovoid, covered with and wide, glabrous; ovules 3-8; scars of perianth and stamens on torus 1-2 x 2.5-3 mm. Fruiting brachyblast ca. 7 х 5 mm. Fruit 3-10 cm long; mature carpels 3-14, nearly globose or ellipsoid, lent, 8-14 x 7-10 stamens ca. 4 X 4 mm. Gynophore under fruit 1 mm; scars of perianth and cm long. Distribution. == Kumaon to Sikkim, China. In CHINA: Xiza Ecology. In forests. Altitude: 1,500-2,250 m. Uses. According to Aikin, the wood is used for light construction. Note. See Dandy (1927ђ). 15. Michelia lacei W. W. Smith, Notes Roy. t. Gard. Edinburgh 12: 216. 1902. TYPE: J. H. Lace 5928 (holotype, E; isotype, K). Michelia uniflora Dandy, Kew Bull. 1927: 203. 1927. TYPE: Burma. Mawson, southern Shan States, N. Manders, 1888 (holotype, K). dire Ps Dandy, J. Bot. 68: 213. 1930. TYPE: 12864 (holotype, P; isotype, K). Michelia а Hu, Bull. Fan. Mem. Inst. Biol. (Pei- pi ng) 10: 118. 1940. TYPE: H. T. Tsai 56961 (ho- type, PE; isotypes, A, BM). Michelia ais arpa Law & R. Z. Zhou, Bull. Bot. Res. , 1: 85, t. 1987. TYPE: К. Z. Zhou 25 (IBSC, seen alive). ~ Tree to 40 m high and 1 m diam.; young twigs stout, 4-6(-10) mm diam., glabrous or at first gray- to yellowish tomentose, later glabrescent, te- rete, brown-yellow to purple-brown, usually glau- cous when dry, conspicuously elliptic lenticellate and with annular scars; old ones longitudinally fis- sured. Stipules thinly coriaceous, appressed-pu- bescent with short, straight, yellowish hairs, gla- brescent, free from the petiole. Leaves coriaceous, rigidulous, bright green above, pale green beneath, glabrous on both sides, obovate-elliptic to elliptic, 14-28 x 6-10 cm; apex short-acuminate to acute, base cuneate to broadly cuneate, margin somewhat revolute; midrib impressed above, prominent be- low, nerves in 9-15(-20) pairs, reticulation lax, nerves and reticulation prominent on both sides when dry. Petiole glabrous, wide sulcate above, 2.5-3 cm long, without scars. Brachyblast glabrous to tomentose, 15-25 х 5—7 mm, pedicle absent; bracts 3-4, chartaceous, = puberulent outside, especially toward the base. Flower fragrant, soli- tary, sometimes abnormally two-flowere d; cylindric, ca. 4 cm long; tepals 9, white, subsimilar, glabrous, obovate to spathulate, 5.5-7 x 1.5-3 cm; stamens 2-2.5 cm long, connective appendage bud ovoid- short-triangular, ca. 1 mm long, filaments 2-3 mm long; gynophore tomentellous, 1-2 cm long, gy- noecium 1.7-2 cm long-exserted from stamens; carpels ca. 12, glabrous; styles long and recurved; Volume 80, Number 4 1993 Chen & Nooteboom 1071 Magnoliaceae of China ovules ca. 10; scars of perianth and stamens on torus ca. 7 mm long. Fruiting brachyblast 15-25 X 5-10 mm. Fruit 7-10 cm long; ripe carpels 4– 6, obovoid, brown to dark brown, lenticellate, ca. 2-4.5 X 2 cm, 5-6 mm thick; gynophore са. 29 X 6 mm; scars of perianth and stamens ca. 2-2.2 x 0.9 cm Distribution. China, Vietnam, and Burma. In CHI- А. Southwestern and southeastern Yunnan: Chen B.L. 86 S 51, GS 86102, Tsai H.T. 56961 Ecology. Altitude: 1,450-1,550 m. Flowering February-March; fruiting Septem- ber- October /ses. Wood soft, for planks. In forests. Collector’s note. ig tree, flowers white, heavily sweet-scented. Note. The first author has collected material with flowers and fruits from the type tree of Miche- lia pachycarpa. There is no doubt that it is syn- onymous with Michelia lacei. 16. era —— Dandy, J. Bot. 66: 360. 1928. ТҮРЕ: F. McClure 1468 (in C. C. C. о. EUN UC; isotypes, A, K, US). Michelia macclurei var. sublanea Dandy, J. Bot. 68: 0. TYPE: Tsiang 2609 (holotype, NY; iso- types, A, BM, E, K, P). Tree to 30 m high and 1 m diam.; bark white- gray; buds, young twigs, stipules, petiole, brachy- blasts and bracts outside appressed-tomentellous with short, straight, clear, silver, brown or rufous hairs, glabrescent; young twigs 2-4 mm diam., yellow-brown, blackish to purple-brown; old ones rough, sparsely lenticellate, longitudinally striped. Stipules free from the petiole. Leaves coriaceous, at first somewhat sericeous, later glabrescent above, appressed-tomentose with short to long, straight, rufous (when young) to clear or yellowish (when older), glistening hairs beneath, broadly elliptic to broadly ovate, rhombic-elliptic, or broadly obovate, x 4-6.5 cm; apex s ort-acuminate to acute, base cuneate occasionally narrowly elliptic, 6.5-14 to broadly cuneate; midrib impressed above, prom- inent, at first densely brown pubescent with long hairs, soon glabrescent, nerves fine, in 10—15 pairs, conspicuous on both sides; reticulation usually densely netted, prominent on both sides. Petiole sulcate above, 2.5-4 cm long, without stipular scars. Brachyblast 10-13 x 3-4 mm, with 3 bract scars; pedicle 2-5 mm. Tepals 9-12, white, sub- similar, the outer 3 narrowly obovate to spathulate, ; 5 x 1-1.5 narrower than the outer ones; stamens 9-13 mm cm, the inner 6-9 smaller and long, connective sparsely puberulent, its appendage tonguelike, 1-2 mm long, filaments 1-2 mm long; gynoecium narrowly cylindric, 7-9 mm long, to- mentellous, slightly longer than the androecium; gynophore brown tomentellous, 5 mm long; carpels ca. 28 or more. Fruiting brachyblast 10-15 x 4- 5 mm, glabrescent. Fruit 3-7 cm long; gynophore 8-10 x 3-4 mm, glabrescent; ripe carpels ellipsoid or subglobose, dull brown, sessile, + lenticellate, 15-23 x 9-10 mm; scars of perianth and stamens on torus ca. 4-5 X 5 mm. Seeds 1-3, flat-ovoid, 8-10 mm long. Distribution. China and northern Vietnam. In CHI- МА. Guangdong: Fangcheng, Chun S.H. 4466; Wu- chuan, Chen S.P. 1131; Yangchun, Wang Q 830001. Guangxi: Rongshui, Long G.R. 830090; Ronxian, Li- ang C.F. 31845; Yulin, Lee S.K. 200723. Hainan: Manning, How F.C. 73194. Also in Yunnan. In mixed forest. Altitude: 500–1,000 m. Flowering March-April; fruiting September- Ecology. November. Uses. Wood fine-textured, straight-grained and fragrant; commonly used for construction, planks, and furniture; the tree is grown as an ornamental. Note. Therefore, we prefer to reduce Michelia macclurei The indument varies in this species. var. sublanea to Michelia macclurei. 17. Michelia masticata Dandy, J. Bot. 67: 222. 1929. ТУРЕ: Poilane 11253 (holotype, P; iso- types, BM, K Michelia sphaerantha Law & Y. Wu, Acta Bot un nanica 10, 3: 335, t. 1. 1988, nom. lle. non e 1987. : M. See Michelia sphaerantha FM under dubious species. Tree to 18-25 m high and 80 cm diam.; bark gray-brown, irregularly fissured lengthwise, con- spicuously lenticellate; young twigs pubescent or glabrous, slender, bright yellow-green; old ones dark green, with white lenticels; bud cylindric, ca. 13 mm long. Stipules yellowish sericeous or brownish pubescent outside toward the apex, free from the petiole. Leaves coriaceous, dark green, slightly glossy above, pale green beneath, scattered ap- pressed-pubescent with short, straight to slightly undulate, clear to brownish hairs, glabrescent on both sides, broadly elliptic, elliptic-oblong, ovate, or obovate, 12-22(-28) x 5-8(-10) ст; apex short-acuminate, base cuneate to rounded; midrib and nerves impressed above, much prominent be- low, nerves in 8-13 pairs, reticulation laxly netted, visible on both sides. Petiole appressed-tomentel- lous, glabrescent, 2-3(-4.5) cm long, without scars. 1072 Annals of th Missouri Beanie Garden Brachyblast yellowish sericeous, glabrescent, ca. 10 x 3-4 mm; pedicle absent; bracts 3, covered with same indument as the stipule. Tepals 9-12, white to yellowish white, subsimilar, the outer 3 obovate, narrowly obovate to spathulate, 5—9(—1 1) x 1-2.5 cm, the inner 6-9 narrower than the outer ones, 3—6(—8) mm long; stamens many, ca. 2 cm long, filament ca. 3 mm, connective ap- pendage ca. 1 mm long; gynoecium cylindric, to- mentellous with short, straight, brownish hairs, 3— 6(-8) cm long, longer than stamens; gynophore tomentellous, ca. 1 cm long; carpels many, ca. mm long; styles ca. 2.2 mm long; scars of perianth and stamens on torus 6-8 x 5 mm. Fruiting brachyblasts 5-7 cm long. Fruit 10—40(-53) ст long; follicles to 100(-128), dull brown, ovoid to ellipsoid, densely gray-white lenticellate, reddish glandular, apically beaked. Seeds 2-6 in each car- pel, red, globose and slightly compressed to irreg- ularly dae: ca. mm long. Gynophore under fruit ca. 2.5 cm long. Distribution. China, Laos, and Vietnam. In CHINA. Yunnan: Wulinag shan, Tsiang T. 12059; Menghua, Yu T.T. 15884; Nanjiang Xian, Yue Z.S. 86- 107; Ping- pien, Mao P.I. 27216; Chu-hsiun, Li M.K. 3. Ecology. In forest. Altitude: 500-2,110 m. Flowering April-May; fruiting October. Uses. Wood hard and dense, useful for all purposes. The tree is grown as an ornamental. Note. We were unable to study the type spec- imen of Michelia sphaerantha C. Y. Wu ex Law & Y. F. collections available (including Mao Ping-i 3621, 0321, 02728). There is no doubt that Miche- lia ni ины Ча C. Y. Wu ex Law & Y. F. Wu (nomen illeg.) belongs to Michelia masticata. The Wu. Fortunately, there were some other only differences are in the degree of indument and the number of carpels. Generally, the Chinese col- lections have a more dense indument and more carpels. Michelia sphaerantha, published by Yue (1987) is based on another type and represents another species. 18. Michelia maudiae Dunn, J. Linn. Soc. Bot. 38: 353. 1908. SYNTYPES: Hongkong Herb. 2449, 2065 (HK, A). Michelia ы Chong, Contr. Biol. Lab. Sci. Soc. China, Bot. Ser. 10: 110. 1936. TYPE: R. C. Ching 2452 (holotype, TE isotypes, A, E, K). Tree to 20 m high and 60 cm diam.; bark pale gray or gray-brown; plant entirely glabrous; buds, young twigs, stipules outside, leaves beneath, pet- iole, brachyblasts, bracts outside, and gynoecium pale green to glaucous when dry; young twigs 2- mm diam., dull purple-brown, yellow-black to dull yellow; old ones yellow-gray, black-gray to dull black, lenticellate, longitudinally fissured. Stipules free from the petiole. Leaves coriaceous, dark green, glossy above, broadly elliptic, obovate-elliptic to obovate, 9.5-17.5 x 3-7 cm; apex abruptly short- acuminate, acumen 3-12 mm long, occasionally long-acuminate, to 2 cm long, base cuneate to nearly rounded; midrib flat to + impressed above; nerves fine, visible on both sides, in 7-12 pairs, reticulation densely netted, prominent on both sides. Petiole 1-3 ст long, without stipular scar. Brachy- blast 10-28 x 4-7 mm, pedicle absent; bracts 2- 4, sometimes leaflike. Flower solitary, occasionally two, fragrant; tepals 9-12, white, subsimilar, the outer 3 broadly obovate, attenuate toward the base, 5-8 x 2.3-4 cm, the inner tepals 6-9, obovate, elliptic to broadly spathulate, short-clawed at the base, 4.5-5 x 1.8-2.5 cm; stamens 13-17 mm long, connective appendage short to long trian- 0.5-2 mm long, filaments pale purple, 2- 9 X 2 mm; gynoecium subcylindric, longer than gular, the androecium, 10—13 mm long; gynophore 6- 12 mm long; carpels many, ovoid, 2-3 mm long; styles 3-3.5 mm long; ovules 6-14 in each carpel; scars of perianth and stamens on torus ca. 4 X 6 mm. Fruiting brachyblast 17-30 x 4-8 mm. Fruits 10-12(-14) em long; ripe carpels 2-19, obovoid to ovoid, dull brown, lenticellate, short-beaked api- cally, 1-2.5 x 1- cm; gynophore 13-30 x 4-5 mm; scars of perianth and stamens 5-6 x 6-7 mm. Seeds subovoid, ca. 1 x 0.5 cm, slightly compressed. Distribution. CHINA. Fujian: Yenping, Buong, Chung H.H. 3667. Guangdong: North river region, Wang C. 31404; Ruyuan, Chen B.L. 80110; Ying Tak, Tsang W.T. 3044; Fan Shiu Shan, Lau S.K. 2552. Guangxi: Guangyang, Zhao R.F. 54; Lingui, Zhao R.F. 197; Longshen, Liu L.F. 567 1; Xingan, Xing An Exped. 285. Guizhou: Tuhshan, Tsiang Y. 6933; Liping Vn Yuan J.M. 621; Lipo county, Song X.H. 244. Hai Jiangyong, Tam P.C. 63667. n Lungnan du. Lau S.K. 4854. Zhejiang: Lungtsuau, Ho Y.Y. 3200; Pang Yung, Ching R.C. 2066. Also н from south- ern Hunan. Ecology. In evergreen broad-leaved forests. Altitude: 500-1,500 m. Flowering February- arch; fruiting September- October. /ses. Wood straight-grained and fine-tex- tured, used for furniture and planks. 19. Michelia mediocris Dandy, J. Bot. 66: 47. 1928. TYPE: McClure in C. C. C. 8593 (ho- lotype, BM; isotypes, A, E, K, MO). Volume 80, Number 4 1993 Chen & Nooteboom 1073 Magnoliaceae of China Tree to 35 m high and 90 cm diam.; bark white- gray; buds rufous or silver appressed-tomentellous; young twigs 2-3 mm diam., dull black-gray to yellow-brown, appressed-tomentellous with short, straight, silver to brown hairs, glabrescent, sparsely lenticellate. Stipules silver appressed-tomentellous, free from the petiole. Leaves thinly coriaceous to coriaceous, bright green, scattered-puberulent with short, straight, clear to brown hairs, especially on midrib and toward the base, glabrescent or glabrous above, + pale green, at first appressed-tomentel- lous with short, straight, glistening, yellowish hairs, soon glabrescent beneath, elliptic, broadly elliptic, а elliptic to rhombic-elliptic, 5-10(-13.5) х cm; apex acute to acuminate, base atten- uate or cuneate to broadly cuneate; midrib prom- inent, appressed-puberulent with short, straight, brown and yellowish hairs, glabrescent below, nerves rather fine, in 10-15 pairs, obscure to visible on both sides, more prominent beneath than above, reticulation densely netted, conspicuous on both sides when dry. Petiole densely tomentellous with same hairs as young twigs, glabrescent, 1.5-2.5 cm long, without scars. Brachyblast brown or yel- low appressed-tomentellous, ca. 6-10 x 3 mm, pedicle ca. 2 mm long; bracts 3-4, silver or brown tomentellous outside. Tepals 9-10, white, subsimi- lar, obovate, spathulate to obovate-linear, usually attenuate toward the base, clawed at the base, 1.8- 3 X nective appendage tonguelike, variable in length, 0.4-1.8 cm; stamens 1-1.5 cm long, con- -4 mm long, filaments 1.5-2 mm long; gynoe- cium short-cylindric, 6-12 mm long, usually longer than the androecium, occasionally equal to the latter; gynophore gray to brown tomentellous, 3- 5 mm long; carpels 13-18, tomentellous with same hairs as gynoecium; ovules 4-6; scars of perianth and stamens on torus ca. 3 X 3 mm. Fruiting brachyblast ca. 1 x 0.4 cm, tomentellous. Fruit 2—3.5(-5) ст long; carpels glabrescent, obovoid to ellipsoid, slightly compressed, sessile, ca. 2 х 1.2 cm; gynophore under fruit 5-8 mm long. Seeds 5-8 mm long. Distribution. and Vietnam. In China, Cambodia, X.W. 5690; Kochow, Pasheng 2 H.Y.61 a Guangxi: ee Chang С.С. 11310; Zhaoping Xian, Li Y.K. 402692. Hainan: Bawangling, oo 13465; ae Fung Mt., Lau S.K. 5245; Five Finger Mts., McClure F.A. CCC 8593; Hung Mo Tung, Sing LY. ipn nr 18350) 817; Tianchi, Wang C. 2888. Ecology. In forest. Altitude: 400-1,300 m. Flowering December-January; fruiting June-July. Uses. brown-yellow; wood straight-grained, fine-textured, light, soft, and durable; commonly used for fur- niture, millwork, veneer, plywood, musical instru- Sapwood yellow-brown, heartwood ments, and also for bridges and ships. Collector's notes. grant. Note. G. W. Groff, Ding & Groff (1923) and Merrill (19272) misidentified this species as Miche- lia maudiae (auct. non Dunn, 1908). On sandy soil; flowers fra- 20. Michelia microtricha Hand.-Mazz., Akad. Wiss. Wien 58, 18: 81. 1921. ТУРЕ: Ten 339 (holotype, C; isotypes, K, V). Figure 14. Tree to 3-20 m high. Young twigs 2-3 mm dull yellow-brown or black-brown, ap- pressed-tomentellous with minute, straight, brown diam., to gray hairs, glabrescent; old ones dull brown, + pubescent, lenticellate; buds narrowly ovoid, 1.2 cm long. Stipules minutely gray or dark brown tomentellous outside, stipular scars ca. 1.5 cm long. Leaves coriaceous, dark green, glossy, appressed- puberulous with straight, clear to gray hairs, es- pecially on the midrib and toward the base above, glabrescent, appressed-tomentellous with short, straight, yellowish to brown hairs beneath, elliptic to narrowly obovate, 8-16 x 3-5 cm; apex acute to short-acuminate, base cuneate; midrib impressed above, conspicuous below, nerves fine, visible on both sides, in 8-14 pairs, reticulation densely net- ted, obscure on both sides. Petiole tomentellous, glabrescent, narrowly sulcate above, 2.5-3 cm long. Brachyblast densely tomentose, 7-10 x 3.5-4 mm, pedicle ca. 3 X 4 mm; bracts 3, broadly ovate, cupreous sericeous outside, 3.5-3.9 cm long. Flower bud long-ovoid; tepals 12, white, fleshy, the outer 3 obovate to spathulate, abruptly constricted near the middle or tapering gradually toward the base, finally becoming long clawed, Ni 1.4-1.5 cm, the inner tepals 6, spathulate, ca. 3- 3.7 x 1.2 cm, the 3 innermost ones obovate-linear, ca. 2.5 x 0.8 cm; stamens numerous, 10-11 mm long, connective appendage short triangular, 1— 2.5 mm long; filaments 1-2.5 mm long; gynoecium subcylindric, 6-11 mm long; gynophore minutely puberulent with clear or brownish hairs, ca. 4 mm long; carpels ca. 25-40, ellipsoid, covered with the same indument as gynophore; styles glabrous, erect, 1-1.5 mm long; ovules 6-8; scars of perianth and stamens on torus ca. 2 mm long. Fruit not seen. Distribution. CHINA. Yunnan: Lungtji, oo jing, Dali, Teng, S.M. 339; Jinping, Wu S.K. 3 1074 Annals of the Missouri Botanical Garden 1cm 1mm FIGURE 14. Michelia microtricha Hand.-Mazz. — 1. Flowering branch. — 2. Deflorated flower. — 3. ingen of leaf underside. Based on Simeon Ten 339 (WU). Ше by Joop Wessendorp, Rijksherbarium, Leiden Volume 80, Number 4 1993 Chen & Nooteboom 1075 Magnoliaceae of China In forest. Note. Michelia microtricha is relatively rare and is closely related to Michelia floribunda. It differs from the latter chiefly by the indument with minute hairs and by the longer brachyblasts. Ecology. 21. Michelia shiluensis Chun & Y. Wu. See under dubious species. 22. Michelia velutina DC., Prodr. 1: 79. Jan. 1824. TYPE: Nepal: Wallich s.n. (K). Michelia lanuginosa Wall., Tent. Fl. Nepal., 1: 8, t. Jun Sampacca lanuginosa (Wall.) кыш Revis. Gen. Pl. 1: 6. 1891. TYPE: Wallich 6493 (holotype, K; isotypes, ri Michelia lanceolata E. H. Wilson, TÁ rnold Arbor. 7: 237. 1926. TYPE: J. F. Rock 6919 (holotype, A; isotypes, E, NY, UC). Tree to 15-20 m high by 90 cm diam.; bark dull brown; indument yellowish to brown, some- times glossy; young twigs 2-3 mm diam., brown to purple-black, pubescent with short, appressed or spreading, undulate to curly hairs, glabrescent; old ones + hairy. Stipules densely pubescent with long, straight or undulate hairs outside, adnate to the petiole, stipular scars 3-5(-11) mm long. Leaves chartaceous to thinly coriaceous, dark green, gla- brous or scattered-pubescent with slender, undulate hairs, especially dense on the midrib above, paler green, densely appressed-pubescent with long, un- dulate, clear or brown hairs, seemingly velvet, fi- nally glabrescent below, narrowly elliptic, some- times elliptic to broadly elliptic or ovate-elliptic, (6.2-)11-24(-29.5) x (2.5-)3.5-6.5(-8.5) cm; apex acute to short- or long-acuminate, acumen 1-1.7 ст long, occasionally obtuse, base cuneate to rounded; midrib impressed above, prominent below, nerves fine, slightly impressed, obscure above, visible beneath, in 11-23(-28) pairs, re- ticulation densely netted, faint on both sides. Pet- latate toward the base, 0.9- 1. Brachyblast densely pubescent with short, undulate to curly, yellowish or brown hairs, 4-17 x 4- mm, pedicle 0-2 mm long; bracts 3, pubescent with same hairs as stipules outside. Flower bud long-ovoid, ca. 2.7 tepals 10-12(-13), white to yellowish, subsimilar, spathulate to elliptic-linear, basally long-clawed, the outer 4-6 glabrous to puberulent outside at the base, 2.2-5.5 x 0.7-0.8 cm; stamens 9-12 mm long, connective appendage triangular, 1-2 cm long; flower very fragrant; mm long, filaments ca. 2 mm long; gynoecium ovoid-oblong to narrowly ovoid, longer than the androecium, 9-15 x 3-6 mm; gynophore densely pubescent, 6-7 x 2-3 mm; carpels ca. 35, sub- globose, densely pubescent; styles erect, glabrous, ca. 1 mm long; ovules 3-6 in each carpel; scars of perianth and stamens on torus 3-5 X 4-5 mm. Fruiting brachyblast + pubescent, ca. 1-1.7 x 0.5 ст. Fruit 3.5-13 cm long; ripe carpels ovoid to ellipsoid, dull brown, densely lenticellate outside, iW 7-25 x 10-17 mm; gynophore 17-20 5 mm; scars of perianth and stamens ca. 5 х 5 mm. Seeds subcordate to subglobose, 7-10 X 6-8 mm. Distribution. CHINA. Xizang: Putsang River, Rong 80. Yunnan: li Hsien, Wang C.W. 78441; Chu hsian, E M.K. 59. In mixed forests. Altitude: 1,500- 2,300 m. Flowering May-June; fruiting August- Ecology. Wood usually grayish white and soft, not useful. 23. Michelia wilsonii Finet & Gagnep., Bull. Soc. Bot. France (Mémoires) 4: 45, t. 7A 1905. TYPE: Wilson 3136 (holotype, P; iso- types, A, BM, K). Michelia o Hemsl. & E. H. Wilson, Kew Bull. 1906. TYPE: Wilson (Veitch Exped.) 3136 eee K; isotypes, A, 3 Michelia szechuanica Dandy, Notes Roy. Bot. Gard. Edinburgh 16: 131. 1928. т TYPE: Wilson 4598 (ho- lotype, K; isotype, A). Tree to 20 m high; spreading; young twigs 1-3 mm diam., black-brown, glabrous to tomentose; old ones brown indument appressed to yellow or to purple-brown, lenticellate. Stipules densely pu- bescent with short, straight to undulate, yellowish to brown hairs, adnate to base of the petiole, scars 1-5 mm long. Leaves thinly coriaceous, dark green, lossy, pubescent especially on the midrib, later glabrescent above, glaucous, at first densely pu- bescent or tomentellous with minute, straight or undulate, yellowish to brown hairs, finally glabres- cent beneath, obovate, obovate-elliptic, narrowly obovate-elliptic, rarely broadly elliptic, somewhat 11.5-14(-18) x acuminate, acumen 4-10 mm long, occasionally acute, base attenuate to cuneate or broadly cu- neate; midrib slightly impressed above, nerves ob- unequal, 4-6 cm; apex short- scure above, visible below, in 8-13 pairs, reticu- lation densely netted, faint on both sides when dry. Petiole pubescent, glabrescent, 1.2-1.5 cm long. Brachyblast pubescent with short, undulate hairs, 7-10 x 3-8 mm, pedicle absent; bracts 3(-4), 1076 Annals of the Missouri Botanical Garden pubescent outside. Flower bud ovoid to ellipsoid, ca. 13 mm long; flower fragrant; tepals 9— yellowish white, subsimilar, obovate to spathulate, 0 » 2-3 x 0.6-0.8 cm; stamens 10-12 mm long, connective appendage very short-triangular, ca. 0.5 mm long, filaments 2-2.5 mm long; gynoecium cylindric, longer than the androecium, ca. 14 mm long; gynophore tomentellous, 2-3 mm long; car- pels many, brown or silver tomentellous; ovules 6 or more; scars of perianth and stamens on torus 2-2.5 x 3-3.5 mm. Fruiting brachyblast + pu- bescent, ca. 11 x 3.9-15.5 ст long; ripe carpels subglobose, ovoid to ellipsoid, mm. Fruit purple-brown, lenticellate outside, sessile to short- stipitate, apex curved-beaked, 10-20 x 15 mm; gynophore 2-2.2 cm x 0.4-0.5 mm. CHINA. Guizhou: Chishui Xian, Dang bei-Sichuan: Metaseqoia area, Hwa C.T. 502. Juri Li Z.Z. 8220423. Sichuan: Nan- chuan, Yang G.H. 54716; Mabian, Chang Q.L. 12055; Kai Hsien, Wilson EH 4598; Mt. Omei, 4720. Distribution. „Z. 1656. Hu Ecology. Altitude: 1,000-1,500 m. Flowering March-May; fruiting August-Sep- tember. In forests. Collector's note. ple gardens. Note. like Michelia sze- chuanica, is not different from Michelia wilsonii. In forests on slope or in tem- Michelia sinensis, The type specimen of Michelia szechuanica pos- sesses appressed hairs, whereas that of Michelia wilsonii has spreading hairs. Unfortunately, we did not have much material to study, but in some collections examined we observed an intermediate indument. 24. Michelia xalba DC., Syst. Nat. 1: 449. 1817. TYPE: This species is based on Sampaca domestica IV alba Rumph., Herb. Amboin. 2: 200. 1741 Michelia hee e aro Verh. Batav. Genootsch. Kun E: Blume s.n. L, sheet no. 908.126-1242; Medien K). Michelia m e racemosa Blume, Fl. Javae Mag- noliaceae: 13, . 1829. TYPE: Blume s.n. (L). Tree to 30 m high, bark gray; young twigs and buds densely grayish appressed-pubescent or pu- berulent, soon glabrescent. Stipules covered with same indument as the young twigs, adnate to the base of the petiole, stipular scars never over half- way the length of the petiole. Leaves thinly cori- aceous, glabrous above, sparsely pubescent below, elliptic to obovate-elliptic, 10-35 x 4-11 cm; apex acuminate, acumen 7-30 mm long, base cu- neate. Petiole sparsely appressed-puberulent to gla- brous, 1.5-5 ly distributed scars, densely grayish pubescent. cm long. Brachyblasts with 2-3 even- Flowers often many, nicely scented, white; tepals 10-12, subequal, lanceolate, 3-5.5 x 0.3-0.5 cm; stamens 8—10 mm long, connective appendage tonguelike, 1-1.5 mm long, filaments 1-1.5 mm long; gynoecium minutely hairy, exserted from the androecium, 8-13 mm long; carpels ca. 10; gyno- phore grayish puberulent, 2-4 mm long. Fruits usually with a few ripening carpels but most carpels abortive, 10-13 cm long; ripe carpels ovoid to ellipsoid, 1.2-1.8 cm x 8-12 mm, + hairy, beaked at the apex; gynophore 18-24 mm long; scars of perianth and stamens ca. 16 х 6 mm. Fruiting peduncle pubescent, glabrescent, 1.7-3.2 x 0. cm Distribution. Widely ПА in southern Fujian, Guangdong, Hainan, Guangxi, and Yunnan, as well as in other ul and ipia countries. Ecology. Flowering April-September. Uses. Flower and root used for medicinal pur- poses, flowers and leaves for volatile oil; the tree is grown as an ornamental Note. Ridley (1913) aad Merrill (192 1a p.p.) dealt with this species under Michelia champaca (auct. non Linnaeus, 1753). It is probably a hybrid of Michelia champaca L. and Michelia montana ume. 2. Michelia section Anisochlamys Dandy in Praglowski, World Pollen Spore F. 3: 5. 1974. TYPE SPECIES: Michelia mannii King. Stipules free from the petiole, the latter at least 10 mm long. Bracts 3. Tepals 9, 3-merous, very dissimilar, those of the outer whorl much shorter and narrower than those of the inner whorls. Fruits apocarpous. 25. Michelia hypolampra Dandy, J. Bot. 66: 32. 1928. TYPE: Fleury in Chevalier 30158 (holotype, P; isotype, K). , Bull. Econ. 132: — gioii A. Chev . 1918. T t. 3. 1985. T У. L. Fen 605 (IBSC not uL є: L. C. Chia & S Tree to 21 m high and 60 cm diam.; buds, stipules outside, young petiole, brachyblast, bracts outside, carpels and gynophore densely covered with minute, straight or slightly undulate, clear to silver, appressed hairs; young twigs slender, 1-2 mm diam., dull brown or black-gray, glabrous; old Volume 80, Number 4 1993 Chen & Nooteboom 1077 Magnoliaceae of China ones dull gray, longitudinally wrinkled. Stipules free from the petiole. Leaves thinly coriaceous, glabrous on both sides, or sparsely puberulent, on both the midrib and the margin near the base above, bright green, glossy, broadly elliptic, elliptic to obovate- elliptic, 6-13 x cm; apex acuminate, acumen 4-10 mm long, occasionally acute, base cuneate, + attenuate along the petiole; midrib slightly prom- inent above, prominent beneath, nerves fine, in 8– 16 pairs, reticulation fine and dense, nerves and reticulation prominent on both sides. Petiole wide sulcate above, 1-2 cm long, without scars. Brachy- blast rather short, 3-4 x 1.5-2 mm, pedicle ab- sent to very short; bracts 3. Flower buds oblong to ellipsoid, ca. 1 cm long; tepals 9, dissimilar, the outer 3 membranous, linear, ca. 15 X 1 mm, the inner tepals 6, narrowly elliptic, 15-20 x 0.4–6 mm; stamens ca. 25, 8-9 mm long, connective appendage triangular, 1-1.5 mm long, anthers са. 5 mm long; gynoecium ovoid, 1-2 cm long, ex- serted from stamens; gynophore 4-6 X 1 mm; carpels 4—10, free from each other, narrowly el- liptic, 6-7 mm long; styles ca. 2 mm long, re- curved; ovules 6-8; scars of perianth and stamens on torus ca. 2-3 X 2 mm. Fruiting brachyblast 5-20 x 2-5 mm, + 3.5-4 cm long; ripe carpels ovoid to obovoid-ellipsoid, gray brownish to brownish, densely lenticellate, with a stipe 3-8 mm long at the base, beaked at the apex, 2-4.5 X 1-2.5 cm; gynophore 2-3 x 0.2-0.5 cm; scars of perianth and stamens ca. 4 X 5 mm. Seeds irregularly shaped, 1-4 in each ripe carpel, -10 -8 mm puberulent. Fruit Distribution. CHINA. Southwestern Guangxi: wangling, Zheng P. arn Southern Yunnan: Xi- shuang Banna, Li Y.H. 4 Ecology. In broad-leaved forest. Altitude: 300-800 m. Flowering March- April; fruiting Sep- tember- October. Uses. Wood straight-grained and fme-tex- tured, suitable for furniture, construction work, and plywood; crown elegant and flower scented, thus the tree is grown for timber as well as an orna- There was no type specimen of Miche- lia hedyosperma available for study. Based on Law's description and figure, there is no doubt, however, that Michelia hedyosperma is conspe- cific with Michelia hypolampra. This species differs strikingly from other species in Michelia by rather small flowers with linear outer tepals and few carpels. 3. Michelia section Dichlamys Dandy in Prag- lowski, World Pollen Spore Fl. 3: 5. 1974. TYPE SPECIES: Michelia balansae (A. DC.) andy. Stipules free from the petiole. Bracts 3-6. Te- pals 6 (or occasionally 4), 2-merous, subsimilar. Fruits apocarpous. KEY TO THE SPECIES OF MICHELIA SECTION DICHLAMYS l. Leaves glabrou 2 1. Leaves hairy uns at least when 2(1). Twigs pubescent, stipules pubescent, leaf base roadly cuneate or attenuate-cuneate, apex short-acuminate or acute, reticulation densely netted; scars of perianth and stamens along torus under fruit 6-8 mm long. . 30. Michelia martinii 2. Twigs tomentellous, stipules glabrous, puber- ulous, or tomentellous, leaf base cuneate or rounded, apex acuminate, reticulation laxly netted; scars of perianth and stamens along torus under fruit 4-5 mm long. Michelia chapensis 3(1). Ln e 3-4 mm long, treelet to 4 m high, ves 1.5-2.5 cm broad, apex rounded; outer sna spathulate, gynoecium ellipsoi 26. Michelia angustioblonga 3. Gynophor e 5-10 mm long, tree t -18 m m acumina це oblong, ог elliptic gynoecium ovoid or cylindric. 4(3) Pedicle present, twigs tomentellous, stipules E or tomentello a ichelia leveilleana 4. Pedicle absent, twigs мени | es pu- bescent, apex short-acuminate nex е petiole dilated at base; pen oid. 21. Michelia ió dila ted at base; gynoecium cylindri 29 26. Michelia ч Law, Bull. Bot. Res. (China) 6(2): 97. 1986. TYPE: Z. R. Xu L12163 (holotype, е isotype, SYS). Fig- ure | Treelet to 4 m high; indument appressed, glis- tening; buds densely brown villous; twigs nearly black-brown. Stipules free from the petiole. Leaves coriaceous, dark green, glabrous above, pale green, brown villous beneath, narrowly oblong, 6.5-10 x 1.5-2.5 cm; apex obtuse, base cuneate to broadly cuneate; midrib impressed above; nerves anasto- m obscure; reticulation densely netted. Pet- iole .9 em long. Brachyblasts sparsely brown pilose. Flowers white, tepals 6, subsimilar, narrowly obovate to spathulate, the outer three mm, the inner three 14-16 mm long; stamens 11-15 mm long, anthers 6-10 mm long, connec- 1078 Annals of the Missouri Botanical Garden Michelia angustioblonga Law.— 1. Flowering branch. — 2. Deflorated flower with gynoecium. Based FIGURE 15. on SYS 164643 — Xu Zhao Ran L 12163. Drawing by Xie Qing Jian. tive appendage triangular, ca. 1 mm long, filaments Distribution. CHINA. Guizhou: Libo, Z. R. Xu -3 mm long; gynoecium narrowly ellipsoid, en- 1263. tirely hidden by the androecium, 1-1.5 cm long; Ecology. In forest. gynophore 3-4 mm long; carpels brown pubescent; | scars of perianth and stamens on torus са. 3 X 3 27. Michelia balansae (А. DC.) Dandy, Kew mm. Fruits not known. Bull. 1927: 263. 1927. Magnolia balansae Volume 80, Number 4 1993 Chen & Nooteboom 1079 Magnoliaceae of China A. DC., Bull. Herb. Boissier 4: 294. 1904. Michelia baviensis Finet & Gagnep., Bull. Soc. Bot. France (Mémoires) 4: 44, t. 5B. 1905. ТҮРЕ: Balansa 3886 (holotype, P; iso- type, L). Michelia tonkinensis А. Chev. ex Gagnep. in Humbert, 52279 (holotype, IBSC; isotypes, ‘A, BM, K, NY). Michelia balansae var. brevipes B. L. Chen, Acta Sci. Nat. Univ. yag iaa 1: 112. 1988. TYPE: B. L. n & C. N. Mai 877034 (SYS). Tree to 18 m high and 60 cm diam.; bark smooth, gray or gray-brown; indument with long, straight to slightly undulate, i pressed, dark brown or brownish to yellowish hairs; young twigs 2-5 mm diam., densely pubescent, subterete, dull brown; old ones glabrescent, sparse- ly lenticellate. Stipules densely pubescent, free from the petiole. Leaves thinly coriaceous, at first scat- tered-pubescent, especially on the midrib, later gla- brescent above, pubescent, glabrescent beneath, elliptic to broadly elliptic or obovate-elliptic to nar- rowly obovate-oblong, 8-21.5(-28) x 4.5- 10(-16.5) cm; apex short-acuminate to sometimes acute, acumen 3-5 mm long, base cuneate to obtuse; midrib impressed above, prominent below; nerves visible on both sides, more prominent below than above, in 12-14(-19) pairs; reticulation laxly netted, prominent when dried. Petiole pubescent, glabrescent, conspicuously dilatate toward the base, 1.9-4.5 cm long, without scars. Brachyblast dense- ly pubescent, 1.2-3.7 cm long, ca. 3 mm wide, pedicle absent; bracts 3, densely pubescent outside. Flower fragrant; tepals 6, subsimilar, the outer 3 + appressed-tomentellous outside, white or green- ish white to yellow, narrowly obovate-oblong, 3.5— 1.3-1.5 cm, the inner tepals 3, spathulate, narrower and smaller than the outer ones; stamens spreading to ap .5-15 mm long, connective appendage short-tri- angular, ca. 0.5 mm long, filaments 0.5- long; gynoecium ovoid, 1-1.4 cm long; gynophore tawny tomentose, 5-7 mm long; carpels ovoid, tomentose; styles glabrous; ovules ca. 6-11; scars of perianth and stamens on the torus 3-4 mm long. Fruiting brachyblast 1.2-7 x 0.3-1.1 cm, gla- brescent. Fruit 4-12 cm long; ripe carpels ovoid to ovoid-ellipsoid, dull dark brown, glabrescent, usually densely small lenticellate, 2-6.5 x 1.8 cm, ligneous, apically with a stout beak of 3- 5 mm long; scars of perianth and stamens 5-10 x 5-14 mm. Seeds ellipsoid, ovoid, subcordate to irregularly shaped, 1—3 in each carpel, 10-15 x -12 mm. China and Vietnam. In CHINA. Distribution. Guangxi: Fang Cheng distr., Tsang W.T. 26868; Tings Ko S.P. 55702; енна, Tso C.L. 23510 Distr., Tsang W.T. (LU 15822) 323; Lingshui, Tso C.L. 43749. Yunnan: Funing, Wang C.W. 89653; Malipo, 86545, Feng K.M. 13599; Ping реп hsien, Tsai Н.Т. 1621. Ecology. In evergreen broad-leaved forest, in shaded and mixed forests, along stream side, on moist sandy soil. Altitude: 350-1,000 m. Flow- ering April-June; fruiting September. Uses. Heartwood brown-yellow, sapwood yel- lowish brown; wood straight-grained, fine- and even- textured, slightly heavy and durable, commonly used for furniture, plywood, and constructions. ;ollector's note. ‘Trees dark gray covered with purple hairs, flowers white, greenish white, yellow- ish or red, fragrant. Note. Michelia balansae is a distinct species. Variations in the indument, size of the leaves, and length of the brachyblasts are very noticeable. The indument of Michelia balansae var. pubescens is appressed-pubescent on the leaves appressi- beneath and the petioles. However, a continuous variation of the indument can be seen within the species; for example, an indument with short and long, straight and slightly undulate, appressed, spreading hairs is found in the collections of N. Chun & C. L. Tso 43749; the indument is NEN straight, and appressed in 5. P. Ko 52279 and F. C. How 73480. The brachyblasts are rather short in the collections of Michelia balansae var. brev- ipes collected in southeastern Yunnan, but inter- mediate forms are found throughout the area of distribution. 28. Michelia chapensis Dandy, J. Bot. 67: 223. 1929. TYPE: Pételot 3379 (holotype, UC; iso- types, BM, NY). Figure 16. Michelia constricta Dandy, J. Bot. 67: 223. 1929. TYPE: Chevalier 30744 (holotype, P). Michelia tsoi Dandy, J. Bot. 68: 213. 1930. TYPE: Tso 1033 Michelia glaberrima Chang, Acta. Sci. Nat. Univ. Sun- yatseni (Guangzhou) 1: 54. 1961. TYPE: Н. T. Tsang 22753 (holotype, SYS; eig. A). m peu B. n & S. C. Yang, Acta t. Univ. Sunyatseni Саца) 3: 96. 1988. TYPE: B L. Chen & С. N. Mai 877011 (holotype, SYS): Michelia chartacea B. L. Chen & S. C. Yang, Acta Sci. at. Univ. Sunyatseni (Guangzhou) 3: 97 Chen TYPE: B. L. & S. C. Yang 87F186 (holotype, SYS). Annals of the 1080 Missouri Botanical Garden Dandy.— 1. Fruiting branch. — 2. Upper side of leaf with venation.— 3. Flower 39 — Chen Bao Liang 87 T 033. Drawing by Xie Qing Jian. Michelia chapensis 4 > í FIGURE 16. bud. Based on SYS 161 Volume 80, Number 4 1993 Chen & Nooteboom 1081 Magnoliaceae of China ШЕ brachyandra B. L. Chen & S. C. Yang, Acta i. Nat. Univ. Sunyatseni (Guangzhou) 3: 98. 1988. TYPE: B. L. Chen & S. C. Yang 87F202 (holotype, SYS). Michelia nitida B. L. Chen, Acta Sci. Nat. Univ. Sun- yatseni (Guangzhou) 1: 111. 1988. TYPE: B. L. Chen & C. N. Mai 877033 (holotype, SYS). Tree to 15-30 m high and 1 m diam.; bark gray-brown, smooth; young twigs brown to dull dark brown, 2-3 mm at first densely appressed-tomentellous with short, straight, brownish hairs, later glabrescent, with sub- iam., smooth, glabrous or orbiculate, white lenticels; old ones gray to brown- ish, fissured longitudinally; terminal buds narrowly ovoid, ca. 1 cm long. Stipules glabrous to puber- ulent or tomentellous outside, free from the petiole. Leaves coriaceous or thinly coriaceous, glabrous, pale brown when dry on both surfaces, obovate or obovate-oblong, occasionally elliptic-oblong or el- liptic, 5.5-16 x 2.6-6.5 subacuminate, base cuneate to obtuse or rounded, cm; apex acuminate to slightly unequal; midrib prominent below, im- pressed above; nerves visible on both sides, in 9– 12 pairs, reticulation coarse. Petiole slender, lon- gitudinally grooved above, glabrous or = minutely appressed-pubescent, 1.5-2.5 cm long, without scars. Brachyblasts tomentellous or puberulent, 3— 11 mm long, 2-4 mm wide at top, pedicles 0—3 mm long; bracts 3-4, tomentellous or puberulent outside. Tepals 6, yellow to pale yellowish, glabrous or puberulent outside at the very base, subsimilar, the outer 3 (narrowly) obovate-oblong, convex out- side, 3-3.5 rower; stamens 14-22 mm long, connective ap- cm long, the inner tepals slightly nar- pendage triangular, 1-1.5 mm long, filaments glabrous, ca. 4 mm long; gynoecium glabrous but sometimes sparsely hairy, narrowly cylindric, en- tirely hidden by the androecium or slightly ex- ceeding it, 11-15 mm long; gynophore 5-7 mm long, silver gray tomentellous with appressed hairs; carpels many, with same indument as gynophore. Fruiting brachyblasts 6-12 cylindric, 3-10 cm long; ripe carpels 1-14 or more, oblong to ovoid, 6-15 x 8-10 mm, apically short-beaked, sparsely beset with inconspicuous x 3-4 mm; scars of perianth and stamens on torus 4-5 mm long. Seeds obovoid to oblong-ovoid, ca. 10 x 3-5 mm. Fruits lenticels; gynophore + pubescent, 8-12 Distribution. China and northern Vietnam. In CHI- МА: Guangdong: Kook Кате, Ko S.P. 50777; D Shan, Tsang W.T. 28559; Lok Chong, Tso CL. 210 75; He Xian, Li Y.K. 401573. Guizhou: Congjiang, Yuan J.M. 692; Liping Xiang, Yuan J.M. Hunan: Zixing Аал, Liang P.H. 86163. с. d uer Nie M. X. 8831 Jingan, Gongda 1547; Shangsa, Yang X.X. 83 1164. Yunnan: Xichou, Wu СА. 62012; Маро, Feng К.М. 13973. Ecology. Їп evergreen broad-leaved forests. Altitude: 500-1,650 m. Flowering March-April; fruiting August- October. Uses. Wood is used for general construction and furniture, the tree is grown as an ornamental. Note. spread species in China. It is distinguished by gla- Michelia chapensis is one of the wide- brous leaves, petioles without scars, 6 tepals, and gynoecium generally hidden by the androecium. The taxa reduced above are essentially the same as Michelia chapensis. However, the gynoecium in Michelia constricta, Michelia brachyandra, and Michelia nitida is slightly exserted from in- stead of hidden within the androecium. Michelia brachyandra generally possesses a glabrous gy- noecium, but sometimes it is hairy. Thus, usually the gynoecium of Michelia chapensis is glabrous. The length of the gynoecium varies from hidden to clearly exserted. 29. Michelia leveilleana Dandy, Kew Bull. 1927: 263. 1927. Michelia cavaleriei H. Léveillé, Feddes Repert. Spec. Nov. Regni. Veg. 9: 459. 1911, nom. illeg. non Finet & Gagnep. (1906). TYPE: J. Cavalerie 3045 (ho- lotype, E; isotypes, K, P). Figure 17. ius d C. Y. Wu ex Law & . Wu, t. Yunnanica 10, 3: 336, t. 2. 1988. TYPE: P. H үн 251 (holotype, РЕ). Tree ca. 15 m by 30 cm diam.; young twigs slender, 2-3 mm diam., black-brown, appressed- tomentellous with straight, dark brown to rufous hairs, glabrescent, sparsely lenticellate; buds ovoid, са. 1 cm long, rufous tomentose. Stipules glabrous to rufous tomentose, free from the petiole. Leaves coriaceous, dark green, glabrous above, pale green, at first rufous pubescent, later glabrescent beneath, elliptic, broadly elliptic, ovate, obovate, to narrowly vate, 7.6-15 x 2.8-4.5 or acute, acumen 5-10 mm long, base cuneate, broadly cuneate, obtuse, sometimes unequal; mid- rib impressed above, prominent below, nerves fine, somewhat obscure, more visible below than above, in 9-15 pairs, reticulation rather faint. Petiole minutely pubescent, glabrescent, 2.5-4.5 cm long, without stipular scars. Brachyblasts stout, tomen- tose with dark brown to rufous hairs, 5— х 4- 5 mm, pedicles 4-6 mm long; bracts 2-3, tomen- tose with long, straight to undulate, dark brown cm; apex acuminate 1082 Annals of the Missouri Botanical Garden Michelia leveilleana Dandy. Branch with flower buds and fruit. Based on W. C. Chen & C. T. Hwa N x FIGURE 17. 889 (K). Drawing by Joop Wessendorp, Rijksherbarium, Leiden hairs outside. Tepals 6, sometimes 4 or 7, white, glabrous, the outer 2-3 oblong to narrowly elliptic- oblong, 30-35 x 7-15 mm, the inner tepals el- liptic to spathulate; stamens many, 12-15 mm long, connective appendage narrowly triangular, ca. 0.5 mm long, filaments 2-3 mm long, glabrous; gynoecium cylindric, 1-2 ст long, hidden by the androecium in bud but slightly longer than the androecium whe ny tomentellous; styles glabrous, ca. 1 mm long; mm long; ovules n flowers open; carpels many, taw- gynophore tomentellous, 5-10 more than 12 in each carpel; scars of perianth and stamens on torus 2-5 mm long. Fruiting brachy- 1-1.5 cm long, abortive carpels ca. 7-10, ovoid, 12 x 9 mm, dark brown, lenticellate; scars of perianth and stamens 5-8 mm along the torus. Distribution. CHINA. Guizhou: Bijie Xian, Yu P.H. ] Hubei-Sichuan, Metaseqoia area: Hwa C.T. 357. Hunan: Dao Xian. Yunnan: Pg Dian Dongbei Exped. 754; Zhenx- iong, Yu P.H. 1060 Ecology. In forest. Altitude: 1,000 m. Volume 80, Number 4 1993 Chen & Nooteboom 1083 Magnoliaceae of China Note. After its publication by Dandy (1927), this species was never mentioned again in the lit- erature. Type specimens of Michelia leveilleana bear flower buds and open flowers and possess obovate leaves. The type of Michelia longipetiola- ta only has fruits, and the leaves are broadly elliptic with relatively long petioles. Fortunately, we were able to study some collections (C. T. Hwa 357, W. C. Cheng & C. T. Hwa 889, 938), which were identified as Michelia wilsonii by S. Y. Hu, gath- ered from Hubei to Sichuan. These collections closely resemble the type of Michelia leveilleana and show the variations in shape and size of the leaves as well as length of the petioles, not only in different collections but within single specimens, e.g., W. C. Cheng & C. T. Hwa 889. Therefore, it is clear that these variations are continuous throughout the area of distribution. We conclude that all of these collections belong to a single spe- cies. 30. Michelia martinii (H. Lév.) Finet & Gag- Lév., Fl. Kouy Tchéou: 270. 1914. Magnolia martini H. Lév., Bull. Soc. Agric. Sarthe 59: 321. 1904. TYPE: L. Martin & E. Bodinier 2066 (holotype, E; isotope, P). nep. ex H. Michelia bodinieri Finet & Gagnep., Bull. Soc. Bot. France 53: 1906. ТУРЕ: Farges 1324 O Р). Michelia | Law, Bull. Bot (China) 5(3): 122. 1985. туРЕ: S. P. Kuo 80358 (holotype, IBSC; ен ВМ). Тгее to 20 m high and ca. 50 cm diam.; bark gray, smooth; young twigs 2-4 mm diam., yellow- green to dull dark brown, smooth, glabrous to oc- casionally appressed-pubescent; old ones rough, sparsely lenticellate; buds ovoid or elliptic-ovoid, densely pubescent. Stipules densely pubescent with long, straight to slightly undulate, spreading, brownish or rarely gray hairs, free from the petiole. Leaves thinly coriaceous to coriaceous, dark green, glossy above, glabrous on both sides, elliptic-oblong, oblong, broadly to narrowly elliptic or narrowly obovate, 6.7— x 2 minate to acute, base attenuate to cuneate or broadly cm; apex short-acu- cuneate; midrib impressed above, nerves fine, vis- ible on both sides, more prominent below than above, in 7-17 pairs; reticulation laxly netted, conspicuous below. Petiole glabrous, 1.5-2 cm long, without scars. Brachyblast short and stout, 7-15 X 3-9 mm, densely pubescent with same indument as stipules, pedicle very short; bracts 3-6, densely pubescent outside. Flower bud ovoid to long-ovoid, 3-4.5 cm long; tepals 6(-8), white to yellow, gla- brous or tomentose outside toward the base, sub- similar to unequal, the outer 3 broadly obovate, obovate-oblong, narrowly obovate to spathulate, 5- 7 x 1.7-5 cm, the inner 3-4 smaller than the outer ones, broadly elliptic to narrowly oblong, short-clawed, 3-6 x mm long, connective appendage long-triangular, cm; stamens 8-25 0.5-2 mm long, filaments 3-7 mm long; gynoe- cium narrowly cylindric, glabrous, 8-18 mm long; gynophore glabrous, 6-12 mm long; carpels many, glabrous; ovules 10 or more in each carpel; scars of perianth and stamens on torus 2-5 X 5 mm. Fruiting brachyblast 13-14 х 4-9 mm, glabres- cent. Fruit 6.5-10 cm long; ripe carpels dull brown, subglobose to ellipsoid, 1-16 x 4-11 mm. Sears of perianth and stamens 6-8 mm; gynophore under fruit ca. 15 mm long. Distribution. China and Vietnam. In CHINA: Guangdong: Yuyuen, Ko S.P. 52762; Zhanjiang, Nan- zhidi 3719. Guangxi: Huanjang, Liang C.X. 77- 15. Guizhou: Siuwen, Tsiang Y 867 1; Bijie, Li Y.K. 11284; Daping Xian, Qian Bei Exped. 2250. Henan. Southern C.W. 87594; Pw li po, Feng K.M. 13274; Mengtze, Henry A 114 Altitude: 1,000-2,000 m. Flowering February-March; fruiting August— Ecology. In forests. September. Uses. The flowers are extracted for volatile Collector's notes. Big tree, in mixed forests, on open hillside or near temples. Flowers fragrant. Note. The type specimens of Michelia bodi- nieri and Michelia longistaminata strikingly re- semble those of Michelia martinii. This species varies continuously throughout its range in texture of the leaves, width and length of brachyblasts, and length of stamens. The collections from Viet- nam possess rigid coriaceous leaves, stout brachy- blasts, and rather long stamens; the collections from Guangdong named Michelia longistaminata possess some extreme characters, such as thin leaves, thin brachyblasts, and rather long stamens. Because of the continuous variation, Michelia lon- gistaminata is reduced here to Michelia martinii. 31. Michelia xanthantha C. Y. Wu ex Law & F. Wu See under dubious species. 4. Michelia section Micheliopsis (Baill.) Dan- dy in Praglowski, World Pollen Spore Fl. 3: 5. 1974. Magnolia sect. Micheliopsis Baill., Adansonia 7: 4, 66. 1866. Liriopsis Spach, 1084 Annals of the Missouri Botanical Garden Nat. Vég. 7: 460. 1839, non Liriopsis Rei- chenb. (1828). TYPE sPECIEs: Michelia figo (Lour.) Sprengel. Stipules adnate to the petiole, the latter short, not exceeding 10 mm. Bracts 3 (or occasionally 2 or 4). Tepals 6 or more, 3-4-merous, subsimilar. Usually shrubs or small trees. Fruits apocarpous. 32. Michelia figo (Lour.) Sprengel, Syst. Veg. 2: 643. 1825. Liriodendron figo Lour., Fl. Cochinch.: 347. 179 О (Lour. DC., Syst. Nat. 1: 460. 181 s.n. not seen. Mus . TYPE: Loureiro dre fuscata Andr., Bot. Repos. 4: t. 229. 1802. elia fasciata Vent., . Malmaison: t. 24, 1803, nom illeg. Magnolia versicolor Salisb., Par- ad. Lond. 1(1): t. 5. 1806, illeg. Magnolia meleagrioides Hort. ex DC. p Nat. 1: 458. 1817, nom. illeg. Michelia Tg (Andr.) Blume ex Wall., Cat.: 832. Liriopsis а ata (Andr.) Spach. Hist. Nat. Vee 7: 461. 1839. ' Andrews (1802, t. 229). Magnolia annonaefolia Salisb., Parad. Lond. 1(1): t 1806. Magnolia fuscata var. annonaefolia тү ) T Syst. Nat. 1: 458. 1817. TYPE: Salisbury (1806, 5). ве var. hebeclada DC., Syst. Nat. 1: 458. TYPE: Mauritius, Thouars s.n. not seen Не parviflora е Icon. Sel. Pl. 1: 22, t. ‚ 85. illeg., non Michelia parviflora DC. Vor Magnolia parvifolia DC., Syst. Nat. 1: 1825. Magnolia fuscata var. parviflora (Blume) Steud., Nomencl. Bot. ed. 2, 2: 89. 1841. Sam- pacca pu (Deless.) Kuntze, Revis. Gen Pl. 1: 6 (as A ise gees 1891. Michelia parvifolia ie ) B.D. Jacks., x Kewensis 2: 223 TYPE: Herb. juae о G-DEL пої eo Michelia ‘skinneriana Dunn, J. Linn. Soc. Bot. 38: 190 PE: Hongkong herb. 2448 (holotype, HK: os A Michelia amoena Q. F. Zheng & M. M. Lin, Bull. Bot. Res. (China) 1: 63, t. 1987. TYPE: China. Fujian: Chang-Ting, 700 m, Q. F. Zheng & Lin 84504 (in herb. Fujian usd College). Michelia brevipe K. Li & Wang, Acta Phytotax. Sin. 25, 5 408, t. 1. 1987. TYPE: China. Guizhou: Longtoudashan, Anlong, C. Z. Dang 913 (HGAS not seen). 32a. Michelia figo var. figo. Shrub to small tree to 15 m high and 30 cm diam.; bark gray; young twigs slender, 1-2 mm iam., brown gray, purple-brown, pubescent with short, straight hairs, glabrescent, the indument at first brown, later dark brown to gray; old twigs white-gray to purple-brown, irregularly longitudi- nally fissured. Stipules densely pubescent or to- mentose, adnate to the petiole, scars to half of its length, 2-3 mm long. Leaves chartaceous to thinly coriaceous, dark green, glossy, glabrous or rather sparsely scattered-pubescent above, light green, appressed-pubescent with short to long, straight to slightly undulate, brown to dark brown hairs, gla- brescent beneath, obovate-elliptic to narrowly ob- ovate-elliptic or elliptic to broadly elliptic, occa- sionally subrhombic, 2-12(-17 cm; apex subcaudate, acuminate, acumen 5-20 mm long, sometimes acute, base cuneate to obtuse; midrib slightly impressed toward the base, densely pubescent, glabrescent above, prominent, densely ong-pubescent, glabrescent beneath, nerves ob- scure on both sides, in 7-12 pairs, reticulation lax, hardly visible on both sides. Petiole densely spread- ing-pubescent, 3-5 mm long, scarred. Brachyblast densely spreading-pubescent, variable in length and width, 4-20(-27) x 2-4 mm, pedicle absent; bracts ca. 3, densely pubescent or tomentose outside. Flower bud ovoid to ellipsoid, 1.5-2 cm long; flower sweet-scented, white or yellow, often tinged with purple, sometimes purplish; tepals 6, subsimilar, fleshy, glabrous, oblong-obovate, obovate to spath- ulate, 1.5-2.5 x 0.6-1.2 cm; stamens 7-11 mm long, connective appendage triangular, 0.5-1 mm long, filaments glabrous, 1.5-3 mm long; gynoe- cium subcylindric, exceeding above the androeci- um, 2-11 mm long; gynophore pubescent with spreading, yellow to brown hairs, occasionally gla- rous, 1-3 mm long; carpels many, glabrous or tomentellous; scars of perianth and stamens on torus 1-3 mm long. Fruiting brachyblast 7-20(-27) X 2-4 mm. Fruit 1-5 cm long; ripe carpels 1- 15, subovoid to subglobose, black-brown to brown, + pubescent, with a ca. 1-mm-long beak at apex, 7-12 x 7-9 mm; gynophore 2-11 mm long; scars of perianth and stamens 2-5 mm long. Distribution. Southeastern CHINA. Anhui: Fujian, Amoy, Nanputo, Chung ы Н. 5921; Buon Kang, Chung H.H. 337 4; Foochow, у hi lls: Chow K.S. 145; Kau ‚ Tsang W.T. 20992; а ис Chun N.K. 40574; Ne Shing Shan, Taam Y.W. Pan Ling Tsze, Chun W.Y. 5856. Guangxi: Tung Hoo Ching R.C. 5613; Baishou, Li Z.J. 653; Lingui, Huang S.L. 2000014; Longsheng, Qin H.F. 70033; Xingan, Chen Z.Z. 51516. Jiangxi: Dagangshan, Yao Kan 9283; Kouyang, Tsiang Y. 10012; Sai Hang Cheung, Lau S.K. 4047; пе Hsiung Ү.К. 6307. Zhejiang: Nan Ноо Hu H.H. 190; Ping Yung, Ching R.C. 1986. Widely ala in most of China and other warm parts of the world. In forests. Altitude: 60-1,000 m. Ecology. Flowering March-May; fruiting July-August. Volume 80, Number 4 1993 Chen & Nooteboom 1085 Magnoliaceae of China Collector's note. Flowers sweet-scented, odor detected at a distance. Flowers used for perfuming tea; leaves for volatile oil and medicinal purposes; shrub grown as an ornamental. ote. No original Loureiro specimen has been traced. However, there is no doubt that this is the plant described by Loureiro as Michelia figo. The species varies greatly in the following char- acters: size and shape of the leaves, length and width of the brachyblast, size and color of the flowers, and the indument of the carpels. For ex- ample, the carpels usually are glabrous, but a con- tinuous variation of the indument of the carpels can be observed throughout its distribution. None of the type specimens or other collections of Michelia amoena and Michelia brevipes were available for study. We consider both as synony- mous with Michelia figo, according to their de- scription and the published figures. Sometimes abnormalities are found, such as leaf- like bracts, more-flowered brachyblasts, terminal unibracteate flowers, and carpellody of the inner stamens. 32b. Michelia figo var. crassipes (Law) B. L. en & Nooteboom, stat. nov. Michelia cras- sipes Law, Bull. Bot. Res. (China) 5(3): 121, t. 1. 1985. TYPE: China. Guangdong: Lechang, 1,000 m, S. H. Chun 3115 (holotype, IBSC not seen). Shrub or small tree, 2-5 m high; bark gray- brown; buds, young twigs, stipules outside, petiole and | bu e densely pubescent with long, taw- ny or rufous hairs. -3 mm diam. Stipules adnate to very high on the petioles, scars to whole length of the petioles. Leaves thinly co- oung twigs riaceous, dark green, glossy, glabrous above, pale green, at first brown to rufous tomentose every- where, especially long-hairy on the veins, later glabrescent beneath, obovate or narrowly obovate, rarely narrowly elliptic, 7-13 x 2.5-4 cm; apex short- or long-acuminate, base cuneate to broadly cuneate; nerves fine, in 6—9 pairs. Petiole 2-4 mm long. Brachyblasts stout and short, 3-4 mm long, flowers heavily scented, purple-red to dark purple; ipee 6, subsimilar, long- и 18-20 x 6-8 e 1 cm long; gynoecium short- duc densely pubescent, и der than the an- droecium, ca. 8 mm long; gynophore ca. 2 mm long; carpels ovoid, densely tomentellous, 3.5-5 mm long; styles ca. 2 mm long. Fruits 2.5-5 cm long; ripe carpels compressed ovoid to subglobose, glabrescent, papilliferous; fruiting brachyblasts stout short, 10-20 x 3-5 mm Distribution. CHINA. Зи Ра Guangdong: Chen B.L. 81007, Zheng P. 10406. Northeastern Guangxi: Ruyuan, Chen B.L. о Ји Xian, Pen Z.H. 51279. Also in southern Нипап. Ecology. Altitude: 300-700 m. Flowering April- May; fruiting August-September. Note. ichelia crassipes is treated as a va- In forests. riety of Michelia figo because there are few veg- etative characters to separate the two. However, the tepals are purple, the brachyblast is stout, and the gynoecium is shorter than the androecium in Michelia crassipes. Michelia figo has white to yellow flowers with purple, short to long, slender brachyblasts, and the gynoecium exceeding the androecium. We observed that the flowers are vari- able when Michelia crassipes is cultivated in ;uangzhou, e.g., brachyblasts become slender, the tepals are rose to pale rose instead of purple, the gynoecium is longer than the androecium. In cul- tivation the flowers of Michelia crassipes become more or less like those of Michelia figo. Therefore, we consider Michelia crassipes to be a wild strain of Michelia figo. 33. Michelia yunnanensis Franchet ex Finet & Gagnep., Bull. Soc. Bot. France (Mémoires) 4: 43, t. 6A. 1905. TYPE: Delavay 4648 (holotype, P; isotypes, E, Michelia yunnanensis Franchet ex Finet & Gagnep. var. Ducloux 168 (holotype, P; isotypes, BM, Michelia үе Hu, Bull. Fan. Mem. Inst. Biol. (Peiping) 8: 37. TYPE: H. T. Tsai 53380 (holotype, in pad A). Shrub to small tree, 6-12 m high; buds, young twigs, stipules outside, young leaves beneath, pet- iole, brachyblast, bracts outside, gynophore and carpels appressed- to spreading-tomentose or pu- bescent with short to long, straight to slightly un- dulate, brown, rufous, black-brown to gray hairs; young twigs 1.5-2 mm diam., olive-green, gray- brown to purple-brown. Stipules adnate to nearly the whole length of the petiole, stipular scars 3-7 mm long. Leaves membranous to coriaceous, bright green, + brown pubescent, glabrescent above, greenish, at first densely or sparsely pubescent, finally glabrescent beneath, obovate or narrowly obovate to elliptic, 2-10 х 1-4 cm; apex obtuse or rounded, or acute to short-acuminate, occa- sionally retuse, base cuneate to obtuse; midrib slightly impressed, densely pubescent, especially 1086 Annals of the Missouri Botanical Garden toward the base above, sparsely pubescent, glabrescent beneath, nerves obscure or visible on both sides, in 7-9 pairs, reticulation faint on both sides when dry. Petiole stout, 5-10 mm long. Brachyblast 4-9(-14) x 2- 2.5 mm, pedicle absent; bracts 3. Flower bud ovoid 2-3 cm long; flower heavily scented; tepals 6-12(-17), white, yellowish white prominent, densely or to narrowly ovoid, to greenish white, subsimilar, the outer 3-4 ob- ovate, broadly obovate to obovate-elliptic, abruptly constricted near the base, becoming rather short- clawed, glabrous to tomentellous outside at the base, 2.2-3.5 x (0.8-)1.4-1.8 cm, the inner 3- 8 obovate, oblong to subspathulate, 2-2.2 x 0.9- 1.2 cm; stamens 5-10 mm long, connective ap- pendage very short-triangular or semiorbicular, 0.5-1 mm long, filaments 2-3 mm long; gynoe- cium subovoid to oblong, longer than the androe- cium, 3-5 mm long; gynophore 5-8 mm long; carpels ca. 15 or fewer; ovules 3-8; scars of peri- anth and stamens on torus 0.5-3 mm long. Fruiting brachyblast 4-9 x 2-3 mm. Fruit 1-4.5 cm long; ripe carpels 1-8, subglobose, dull brown, = pu- bescent, lenticellate, sessile, apically with a 1-1.5 mm long beak, 7-14 x 7-12 mm; gynophore 7- 10 mm long; scars of perianth and stamens 2-5 mm long. Seeds 1-2 in each carpel, ellipsoid to ovoid, 6-8 x 5-7 mm Distribution. CHINA. Gu илек : Anlong ird ам . 886. Sichuan: between Yim Schneider C. 607. Yu | 10 -А ог. Ехреа. 1 ћ E. ae KM. 12636; eoru Henry A Ecology. In forest or thickets. Altitude: 1,100-2,300 m. Flowering March- April; fruiting August-September. Uses. Flowers and leaves are used as spices; the tree is grown as an ornamental. ote. Michelia yunnanensis varies in shape and size of the leaves. Although variety angusti- folia has relatively narrow and small leaves, in- termediate forms can be found throughout the range. Michelia dand yi is similar to Michelia yun- nanensis. л . Michelia section Tsoongiodendron Noo- teboom & B. L. Chen. Tsoongiodendron W. Y. Chun, Acta Phytotax. Sin. 8: 7. 1963. TYPE SPECIES: Tsoongiodendron odorum Chun — Michelia odora (Chun) Nooteboom & B. L. Chen Stipules adnate to the petiole. Tepals 9, subsimi- lar. Fruits syncarpous, very large; follicles crowd- ed, sessile, woody, large, without a beak, rounded on the back 34. Michelia odora (Chun) Nooteboom & B. L. Chen, comb. nov. Tsoongiodendron odorum Chun, Acta Phytotax. Sin. 8, 4: 281, 9 tt. 35, 36. 1963. TYPE: S. P. Ко 51928 (holo- type, IBSC; isotype, BM not seen). Tree to 25 m high and 1 m diam.; bark grayish brown, deeply wrinkled; buds, stipules outside, young twigs, leaves beneath (especially on the mid- rib and nerves), petiole, brachyblast, bracts outside, gynophore and carpels densely covered with an indument of short to long, straight to undulate, spreading, yellowish to brown or clear hairs; young twigs 2-3 mm diam., dull black-brown; old ones longitudinally wrinkled, irregularly transverse-fissured, lenticellate. Stipules adnate to gray-brown, the petiole from the lower base to its middle or higher, stipular scars 6-18 mm long. Leaves thinly coriaceous, green, glossy, glabrous but densely pu- bescent on the midrib above, greenish beneath, elliptic or narrowly to broadly obovate-elliptic, 8- 20 x 3.5-7 ст; apex short-acuminate, occasion- ally acute, base cuneate to rounded; midrib slightly impressed above, conspicuously elevated below, nerves in ca. 12 pairs, hardly visible above, retic- ulation laxly netted, prominent beneath. Petiole 0.9-3 cm long. Brachyblast 7-10 x 2-3 mm, pedicle absent, bracts 2-3. Flower white to yel- lowish, very fragrant; tepals 9, subsimilar, narrowly obovate-elliptic, the outer three, 17-20 x 6.5- 7.5 mm, the innermost tepals three, ca. 15-16 x 5 mm; stamens 40-45, 7.5-8.5 mm long, соп- nective appendage very short, ca. 0.5 mm long, filaments 1.5-3 mm long; gynoecium subovoid, 3- 4 mm long, shorter than androecium; gynophore stout, 2-2.5 x 1-2 mm; carpels 10-12, narrowly ovoid; styles glabrous, ca. 2 mm long. Fruiting brachyblast 1-2 x 1-1.3 cm. Fruit syncarpous, long-elliptic, sometimes subglobose because of the abortion of the upper carpels, 13-16 x 7-9 cm; ripe carpels crowded, sessile, woody, large, without a beak, 4-6 cm long; the exocarp thin and fleshy, olive-green, white lenticellate, dark brown and yel- low-dotted when dry, mesocarp woody, 1-2 cm thick, endocarp thin, greenish yellow. Seeds 4–6 in each carpel, ellipsoid to irregularly shaped, ca. 15 x 8 mm Distribution. China and ud Vietnam. In CHI- NA. Fujian: Jing Xian, Ye G.D. 1 Also in southern Jiangxi and southeastern Yunnan. Volume 80, Number 4 1993 Chen & Nooteboom 1087 Magnoliaceae of China Ecology. In forest. Altitude: 500-1,000 m. Flowering March-April; fruiting September-Oc- Wood white, straight-grained, fine-tex- tured, light and hard; used for furniture, building, plywood, musical instruments, and work not in contact with the soil; tree grown as an ornamental. ollector's note. Flower light red. ote. This species is characterized by very small flowers and huge syncarpous fruits. Its leaves are easily confused with those of Michelia bal- ansae, which has, however, free stipules. Dandy (ex Gagnepain, 1938) already mentioned the species as a nomen nudum, Michelia gravis. a . Michelia section Paramichelia Nooteboom & B. L. Chen, stat. et sect. nov. Paramichelia H. H. Hu, Sunyatsenia 4: 142. 1940. TYPE: Paramichelia baillonii (Pierre) Hu = Miche- lia baillonii (Pierre) Finet & Gagnep. Stipules adnate to the whole length or almost the whole length of the petiole. Bracts 3. Tepals 12, 4-merous, subsimilar. Fruits syncarpous. 35. Michelia baillonii (Pierre) Finet & Gag- nep., Bull. Soc. Bot. France (Mémoires) 4: 46. 1905. Magnolia baillonii Pierre, Fl. Forest. Cochinch. 1: t. 2. 1880. Paramichelia bail- lonii (Pierre) Hu, Sunyatsenia 4: 144. 1940. TYPE: Pierre 750 (holotype, A; isotype, K). ше ee King, Magnol. Brit. India 3: 205, 47 bis. 1891. Aromadendron spongocarpum (King) Craib, Fl. Siam 1: 25. 1925. otanic Garden collector (Badal Khan) 102 (ho- lotype, CAL; isotype, Talauma ар de King, Mag nol. Brit. t . Michelia pd (King) Fins & о. ^x ll. Soc. Bot. F 44. 1905 TYPE: Peal s.n. (holotype, CAL; isotypes, BM, L, US). Huge tree to 50 m high and 1 m diam.; young twigs slender, 2-3 mm diam., dull brown, densely pubescent with long, straight to slightly undulate, + spreading, yellowish to brownish hairs; old ones black-brown, sparsely white lenticellate. Stipules densely pubescent with same hairs as the young twigs, adnate to the petiole nearly to half its length, scars 6-10 mm long. Leaves chartaceous, above dark green, appressed-pubescent with slender, straight, clear to yellowish hairs, brown pubescent on the midrib when young, glabrescent when old, below pale green, pubescent with short and long to rather long, straight to undulate, yellowish to brownish hairs at first, finally glabrescent, narrowly obovate, obovate, obovate-elliptic to narrowly el- .5(-23.5) x 3-8.5 cm; apex short-acuminate to acute; base liptic, usually somewhat unequal, 6-1 cuneate or broadly cuneate, sometimes rounded; midrib impressed above; nerves fine, obscure above, visible beneath, in 9-15 pairs; reticulation rather densely netted, faint on both sides when dry. Petiole ensely pubescent, glabrescent, 1.5-3.5 cm long. Brachyblast slender to stout, 8-14 x 2-3 mm, pedicle absent; bracts 3, densely yellowish or brown tomentose outside. Flower bud ovoid-cylindric, 2.7-3 mm long; flower heavily scented; tepals 12-18(-20), white or yel- low-white, subsimilar, spathulate to obovate-linear, glabrous to pubescent outside toward the base, apex yellowish tomentose, acuminate to acute, base attenuate to the base, usually becoming long-clawed, 23-35 x 2.5-7 mm; stamens 7-9 mm long, connective appendage ong, filaments gla- brous, 1-2 mm long; gynoecium ovoid-cylindric, 6-9 mm long, longer than the androecium; gyno- phore yellow pubescent, 3-4 mm long; carpels narrowly triangular, 1-2 mm densely yellow pubescent; styles red, glabrous; scars of perianth and stamens on torus 5-7 x 7-11 mm. Fruiting brachyblast dull black-brown, + pu- bescent, 1-2 x 0.6 cm. Fruit syncarpous, ovoid, ovoid-ellipsoid, to irregularly shaped, dull gray- black, densely lenticellate outside, 3-9 x 2.5-4. cm; ripe carpels connate, falling off when maturing in irregular masses, midrib lignified and persistent in the form of an upcurved laterally compressed hook; gynophore 7-13 x anth and stamens 5-7 x 7-11 mm. Seeds subo- void, cordate to irregularly shaped, 7-10 x 6-7 5-9 mm; scars of peri- Distribution. | Assam, China, Thailand, Burma, = о A Menglian Xian, Menglian Е 10145; xe Вал, na, Sino-Rossica Exped. 5 Ecology. In forest. Altitude: 300-1,600 m. Flowering February-March; fruiting August-Sep- tember. Uses. and durable; used for building, bridges, furniture, interior finish, and plywood. Collector's note. Trees in mixed forests or thickets, fruits green-black, seeds with red aril. ote. his species has not only a wide distri- Wood straight-grained, fine-textured, bution, but also a considerable variation in shape and size of the leaves. Hu (1938) named this species Aromadendron yunnanensis, nom. nudum. 1088 Annals of the Missouri Botanical Garden DUBIOUS SPECIES Michelia chongjiangensis Y. K. Li & X. M. Wang, Guizhou Sin. 3: 18. 1983. туРЕ: China. Guizhou: Taiyangsun, Chongjiang, 1,300 m, Y. K. Li 9294 (HGAS not seen) = Michelia ?leveilleana Dandy. Michelia caloptila Law & Y. F. Wu, Bull. Bot. Res. (China) 4(2): 152, t. s.n. 1984. TYPE: China. Jiangxi: Zixi, Jiangxi gong-da linxue- xi 60069 (IBSC not seen). Michelia elegans Law & Y. F. Wu, Bull. Bot. Res. (China) 8(3): 71, t. 1. 1988. TYPE: China. Guangzhou: South China Botanic Garden, 27 Apr. 1986, Y. W. Law 6109 (IBSC not seen) — Michelia ?cavalerieri Finet & Gagnep. Michelia laevifolia Law & Y. F. Wu, Bull. Bot. Res. (China) 8(3): 72, t. 2. 1988. туРЕ: China. Guizhou: Anlong, 1,500 m, 22 June 1960, T. S. Chang 1852 (IBSC not seen) = Michelia ?yunnanensis Franchet ex Finet & Gagnep. Michelia sphaerantha Z. S. Yue, Acta Bot. Yunnanica 9(4): 413, t. 1, 1987. SYNTYPEs: Z. S. Yue 83.111, 86-107 (KUN not seen). ote. Yue meant to describe Michelia sphaerantha C.Y. Wu (in manuscript), later again published by Wu in Acta Bot. Yunnanensis 10: 335. 1988, nom. illeg., with another type. The latter is Michelia masticata. According to the drawing published by Yue and a leaf fragment we received from him, it appears that Michelia sphaerantha Z. S. Yue is a different species: it has its stipules adnate to the petiole. Study of the type is necessary to conclude its status. 11. Michelia fujianensis Q. F. Zheng, Bull. Bot. Res. (China) 1(3): 92, t. 1981. TYPE China. Fujian: Sanming Shi, K. R. Wu 7 79100 (in herb. Fujian Forestry College not seen). Tree to 15 m, ca. 40 cm diam.; bark gray- brown to dark brown, smooth or slightly rough; buds, young twigs, petioles, and brachyblasts densely pubescent with gray or brown hairs. Stipules free from the petiole. Leaves thinly coriaceous, pubes- cent, glabrescent, especially on the midrib above, densely brown appressed-pubescent beneath, ob- long to narrowly ovate-elliptic, 6-11 x 2.5-4 cm; apex acuminate to acute, base cuneate to obtuse; midrib prominent below, nerves prominent on both sides, in 8-9 pairs. Petiole 1-1.5 cm long. Brachy- blast ca. 5 mm (15 mm?) long; bracts densely brown pubescent. Flower bud ellipsoid-ovoid, ca. 1.5 x 0.8-1 cm; tepals ca. 15 mm long, anthers ca. 5 mm long; gynoecium ovoid; gynophore dense- ly pubescent with long hairs, ca. 2 mm long; carpels ca. 20 or more, gray pubescent. Fruiting brachy- blast 5 mm long. Fruit ca. 3 cm long; ripe carpels 1-4, ovoid to subglobose, 1-1.5 cm long, short- stipitate. Distribution. CHINA. Fujian. In forest. Altitude: below 500 m. Flowering February—April; fruiting August-Octo- Ecology. Note. The original description only mentioned “fruiting brachyblast 5 mm long.” Later, Lin (1985) described “brachyblast stout, ca. 15 mm long" and “fruiting brachyblast 5 mm long." Obviously the length of the brachyblast was misrepresented by him. We had no opportunity to study collections of Michelia fujianensis, but it seems closely allied with and might be conspecific to Michelia caval- eriei, especially in number of tepals and length of fruit, which are variable in that species. 8. Michelia flaviflora Law & Y. F. Wu, Acta Bot. Yunnanica 10(3): 340, t. 6. 1988. TYPE: H. Wang 100120 (IBSC not seen). Tree to ca. 15 m high; buds, young twigs, stip- ules, petiole, and brachyblasts yellowish tomentose; young twigs brown tomentellous; old ones white lenticellate. Stipules free from the petiole. Leaves thinly coriaceous, dark green, glabrous above, glau- cous, brown sericeous beneath, narrowly elliptic or obovate-elliptic, 15-24 x 3 cm; apex acu- minate, base cuneate; midrib slightly prominent to impressed above, nerves fine, in 16-24 pairs. Pet- iole dilatate toward the base, without scars, ca. 0.5-1.2 em long. Brachyblasts ca. 1 cm long. Tepals 15, yellow, subsimilar; stamens ca. 90, 1.1— 1.5 cm long, connective appendage triangular, 1— 2 mm long; gynoecium narrowly ovoid, ca. 1.2 cm long, exceeding beyond the androecium; gynophore tomentellous, ca. 1 cm long, ovaries ovoid, densely villous, 4 mm long; styles glabrous, ca. 2 mm long. Vietnam and China. In CHINA. Yun- Distribution. nan (Pingbian) о There was no material of Michelia fla- viflora available. In Law’s description, no dimen- sions of the flowers were given, but we think it represents a distinct species. 21. Michelia shiluensis Chun & Y. Wu, Acta Phytotax. Sin. 8(4): 286. 1963. TYPE: Hainan Exped. 90669 (IBSC not seen). Volume 80, Number 4 1993 Chen & Nooteboom Magnoliaceae of China 1089 Tree to 18 m high and 30 cm diam.; bark gray; twigs, , and petiole glabrous; young twigs slightly black, smooth; old ones dull gray to purple- brown, somewhat rough, longitudinally striped and leaves lenticellate; terminal buds narrowly ellipsoid, or- ange or gray sericeous, glabrescent, 10-12 x 4- 5 mm. Stipules pubescent, free from the petiole. Leaves coriaceous, + rigid, dark green above, paler green beneath, obovate, obovate-oblong, 8— 1 4(—20) x 4-7(-8) cm; apex acute, base attenuate to cu- neate or broadly cuneate; midrib impressed above, nerves in 8-12 pairs, reticulation densely netted, both nerves and reticulation conspicuously visible on both sides when dry. Petiole wide sulcate above, 1-3 ст long, without scars. Brachyblast ca. 1.7 cm long. Flowers white, cup-shaped; tepals 9, fleshy, obovate, convex outside, 3-4.5 х 1.5-2.5 cm; stamens 2-2.5 cm long, anthers 12-17 mm long, filaments reddish; gynoecium pubescent, 1.4-2.1 cm long; carpels many (ca. or more), ovoid, 2.5-4 mm long; styles ca. 1.5 mm long. Fruiting brachyblast with 2-3 bract scars. Fruits 4-5 cm long, peduncles ca. 2-3 cm long; carpels usually partly fertile; ripe carpels obovoid or obovoid-el- lipsoid, 8-12 mm long, short-beaked at the apex. Seeds broadly ellipsoid, ca. 8 x 6 mm. Distribution. CHINA. Hainan (Dingfang, Baoting). "C . In evergreen broad-leaved forests. Altitude: 200- 1,500 m. Flowering March- April; fruiting June-August. Wood straight-grained, even and fine- textured, slightly heavy, with a characteristic odor and Jses. urable; used for furniture, musical instru- ments, and building; the tree is grown as an or- namental. No collections could be consulted, but this is probably a good species. However, the orig- inal description does not give enough key char- acters. ote. 31. Michelia xanthantha C. Y. Wu ex Law & Y. F. Wu, Acta Bot. Yunnanica 10(3): 338, t. 4. 1988. ТҮРЕ: Sino-Ussr Exped. 7046 (KUN not seen). Tree to 30 m high and 1 m diam.; buds cylindric, glabrous, ca. 1.5 cm long; young twigs dark brown to gray-yellow, ca. 4.5 cm diam., sparsely lenti- cellate. Stipules glabrous, free from the petiole. Leaves thinly coriaceous, oblong to obovate-oblong, 15-17 x 6.2-7.5 cm; apex short-acuminate, base cuneate to rounded; glabrous on both sides; midrib slightly impressed above, prominent beneath; nerves in 11-18 pairs, reticulation prominent on both sides. Petiole without stipular scars, 2-2.5 cm long. Brachyblast glabrous, ca. 1.5 ст long. Tepals 6, yellow, fragrant, subsimilar, narrowly oblong or narrowly obovate, 4-5.5 x stamens -2.5 cm, anthers 1.8-2 cm long, filaments ca. -1.5 cm; 3 mm long, connective appendage short-triangular, 1-2 mm long; gynoecium cylindric, 1.8-2 cm long; gynophore 2-3 cm long; carpels many, ovate, densely gray pubescent, 3-4 mm long; styles ca. l mm long. Fruit pendent, ca. 21 cm long; пре carpels dark gray-brown, sparsely lenticellate. Distribution. CHINA. Yunnan (Xishuangbanna). Ecology. In forests. Altitude: 1,350 m. Note. We were not able to study the type specimen of Michelia xanthantha. However, based on the description this seems to be a good species. MAGNOLIACEAE subfamily LIRIODEN- DROIDEAE (Barkley) Law Yuh-wu, Acta Phytotax. Sinica 22: 105. 1984. Lirioden- draceae Barkley, Phytologia 32: 304. 1975. Leaves 2- 1 O(usually 4-6)-lobed, the apex trun- cate or widely emarginate; stipules always free from the petiole. Anthers extrorse. Fruiting carpels woody, indehiscent, samaroid, produced at the apex into a long winglike beak, indehiscent, caducous. Testa adherent to the endocarp. One genus: V. Liriodendron L., Sp. Pl.: 535. 1753. TYPE SPECIES: Liriodendron tulipifera L. Flowers terminal, soli- tary, bisexual. Tepals 9-17, Tree. Leaves deciduous. -merous, subequal. he connective produced into a short appendage. Gynoecium sessile. Carpels numerous, free, the lowermost sterile. Ovules 2. Seeds 1-2 in each carpel. Distribution. Two species, southeastern Asia and southeastern North America. 1. Liriodendron chinense (Hemsley) Sarg., Trees & Shrubs 1: 103 t. 52. 1903. Lirio- dendron tulipifera L. var.? chinense Hems- ley, J. Linn. Soc. Bot. 23: 25. 1886. SYNTYPES: Shearer s.n. (1875), Maries s.n. (1877) not seen. Tree to 40 m high, | m or more diam.; young twigs 2-4 mm diam., gray, gray-brown, dull brown to purple-brown, glaucescent, old ones dull purple or gray-black, lenticellate. Stipules yellow to brown- ish, glabrous, free from the petiole. Leaves mem- 1090 Annals of the Missouri Botanical Garden branaceous to chartaceous, (3-)5.5-12(-18) x 3- 9.5(-23) cm; apex 2-lobed, with а pair of lateral lobes near the base; glabrous on both sides, bright green above, glaucescent to glaucous beneath; mid- rib elevated beneath, nerves rather prominent be- low, in 5-7 pairs. Petiole slender, glabrous, 2.5- 8 cm long. Peduncle glabrous, 8-9 x bract single. Flower bud ovoid, 4.5-7 cm long; flower cup-shaped; tepals 9, subsimilar, greenish to yellowish, the outer 3 calycoid, glabrous, obovate to obovate-elliptic, abruptly constricted near the base and becoming short-clawed, 2.5-4.5 x 1.5- 2 cm, the inner tepals 6, erect, obovate, 3-4 cm long, lengthwise yellow-striate outside; stamens 20— m long, connective appendage triangular, short, filaments 6-8 mm long; gynoecium ovoid- cylindric, ca. 7 X 6 mm, carpels yellow-green; 3-4 mm; scars of perianth and stamens on torus ca. 3 X 5 mm. Fruiting peduncle 1-1.6 x 0.3-0.5 cm. Fruit ovoid, 4.5-9 x 2 cm, ripe carpels samaroid nut- 2-3 x 0.4–0.6 mm Distribution. CHINA. Anhui: Huangshan, Bai Zhang Tan, Deng & Yao 79147. Zhejiang: Hangchow, Chiao C. У. herb. no. 18844. dris Guanyang, Chen Z.Z. 52523; yan Xian, Yu S.L. 900086. Guizhou: 1 1623; pana Chen J.C. 745. 61528. Hubei: Chienshih Hsien, Chow H.C. 1094. Tiangen, Nanjing: Chang H.T. 5396. Jiangxi: Lushan, Chang H.T. 5309; Kuling, Chiao C.Y. herb. no. 18560. Sichuan: Hsiu- shan-hsien, lets, Also in аза Shaanxi and cultivated іп Taiwan and am. North Vie logy. In mixed forests, growing particu- larly well on sandstone, granite, and sandshale for- mations. Altitude: 500-1,700 m. Flowering May; fruiting September- October. Uses. Wood reddish brown, fine-textured and straight-grained, used for furniture; bark for me- dicinal purposes; because of the shape of leaves, this is an especially ornamental tree. Collector's notes. or rocky slopes, roadsides and streams; bark pale gray; stipules tinged red at apex; Tree, growing along sandy sometimes b flowers yellowish; fruits green when young, brown ater. Note. Finet & Gagnepain (1905) treated this species as Liriodendron tulipifera 2. Liriodendron tulipifera L., Sp. Pl. 1: 535. 1753 [Tulipifera virginiana Hermann, Hort. Ludg.-Bat. Cat.: 612, t. 613. 1687]. Lirio- dendron procerum Salisb., Prodr. Stirp. Chap. Allerton: 379. 1796. Liriodendron trunca- tifolium Stokes, Bot. Mat. Med. 3: 233. 1812. TYPE: t. 613 (P. Hermann, 1687). Tree to 60 m high, 3.5 m diam.; bark deep longitudinally fissured; brown to purple, often pru- inose. Leaves ca. 7-12 cm long; apex 2-lobed, with 2-3 lateral lobes, minutely white hairy when young, soon glabrous beneath. Petiole ca. 5-1 cm long. Flowers cup-shaped, tepals 9, outer tepals 3, green, calycoid, patent, inner tepals 6, green- yellow, erect, ovate, ca. 4-6 cm long, inside with orange nectaries below the middle; anthers ca. 1.5- 2.5 cm long, filaments ca. 1–1.5 ст long; gynoe- cium yellow-green. Fruits ca. 7 cm long, nutlet samaroid, pale brown, са. 5 mm long, apex pointed, the nutlets of the lower parts usually persistent. Distribution. Native of northern America (in south- ern Ontario and the eastern and southeastern United States); cultivated in Kunming, Lushan, Nanjiang, and Qingdao. Ecology. ber-October. Flowering May; fruiting Septem- BIBLIOGRAPHY oon M. 1763. ncent, Paris. йалыш, А. pen Alc Familles des Plantes 2: 365, 537. une nuove Magnoliacea malesi e papuane con osservazione storiche e geografiche sulle famiglia Atti Com. Accad. Fisiocrit. Siena 9 T, ANDREWS, H. C. 1802. Bot. Repos. 4: t. 226-229. BACKER, C. 1907. Flora van Batavia 1: 10-12. G. Kolff, Batavia 1911. Schoolflora voor Java: 14-16. Visser, rada & R.C. BAKHUIZEN VAN DEN BRINK. 98. P. Noordhoff, Gron 1916. Mitt. Deut. Dendrol. Ces. 1915 se iie k. Flora of Java 1: BalLey, L. Н. 2: nr. 2: у BarLLON, Н. Е. 6. Mémoire sur la Famille des Mag- noliacées Adansonia 7: 2, 4, 6 Baker, E. С. 1 zi H. Forbes' s Malayan plants. J. Bot. 62 Suppl.: BARANOVA, M. 1969. а anatomy of Magnoliaceae. Bot. Zhurn. (Moscow & Leningrad) 54: 1952 [in Russian]. BARRETTO. 1988. Мет. Hongkong Nat. Hist. Soc. 18: BEAN, W. J. 1908. и chinense. Gard. Chron. III. 44: 429, fig. са ыш notes on new or rare trees and raba, Kew Bull. 1920: 119-124. . Trees & Shrubs en in the British Isles 3: 222. Ј. Murray, Lon . 646 in Trees & Sy bus Hardy in the British ines, 8th ed. J. Murray, London BENTHAM, G. 61. Flora Bone 8. L. ‘Reeve, ondon. BISSCHOP GREVELINK, A. H. 1882. Planten van Ned- erlandsch-Indié Bruikbaar voor Handel, Nijverheid en Geneeskunde: 277-278. Amsterdam. BLACKBURN, B. C. 1954. The Plant in = = Pa Star E Popul. Gard. 5(3): 6 55. Amatores Herbarii va Fa Volume 80, Number 4 1993 Chen & Nooteboom 1091 Magnoliaceae of China Bianco, Е. M. 1837. Flora de Filipinas: 462. D. Miguel anchez, Manila BLuME, C. L. 1823. ` Beschrijving van eenige gewassen, waargenomen op eenen togt naar den Salak in s erh. Bat. Genootsch. Kunst. Wet. —— ——. 1825. Bijdragen tot de Flora van Neder- pido la 8-10. Batavia . Flora Javae жө Е 20, Magnoliaceae: nec non insularum adja- 1-40. Frank, BOERLAGE, y с. am tur Herbaceis poe ]: 9. Batavia. BONPLAND, A. A. . Descriptions des Plantes ue ares Cultivées á Malmaison: 53 t. 20. P. Didot Brans, D. "1906. Indian Trees: 7-8. Constable, Lon- Boone J. 1885. Magnoliaceae. P. 499 in H.O. Forbes, Naturalist’s Wanderings in the Eastern Archipel- ago. London. Buc'Hoz, Р. J. 1776. Collection Précieuse 1: t. 4-9. Я 1719. Plantes Nouvellements Découvertes 21: t. 19. Paris. 1785. Le Grand Jardin de l'Univers: t. 131. Paris. BURKILL, I. Н. 1935. A Dictionary of the Economic Products of the Malay Peninsula: 1464. Londo & M. Hawrrr.. 1930. Malay village medicine. Gard. Bull. Straits Settlem. 6: 167, 432. BURMANN, N. L 768. Flora Indica: 1-243. Leiden, Amsterdam. CANDOLLE, А. P. DE. 1817. Regni Vegetabilis Systema Naturale 1: 449-560. Paris. 24. Prodromus lc Naturalis Regni Vegetabilis 1: 77-82. Par CANDOLLE, А. DF. une d'une Magnoliacée nouvelle. Bull. Herb. Boissier 4: 2 e Natural History of Carolina, CERON. ана por el a) de la Suprimada Comi- n de la e Foresta anila. MEIO, J.G. pea. In: Florula Hong- kongensis. J. Bot (Hooke r) 3: 256. CHANG, H. T. Acta Sci. Nat. Univ. Sunyatseni (Guangzhou) i 53-54. Снао, T. B. 19 Acta Agriculture Universitatis He- nanensis 19, 4: -363. CHEN, B. L. 19 88. Four new species of Michelia from pou Acta Sci. Nat. Univ. Sunyatseni 3: 86- 112. CHEN, Y. 1953. Illustrated Manual Chinese Trees and Shrubs: 286-301. Science Technology Press, Shang- ai. CHENG, C. Y. Omeiensium: 1: t. versity, Chengtu. Снемс, У. C. 1933. An enumeration of vascular plants from Chekiang I. Contr. Biol. Lab. Sci. Soc. China Bot. Ser. 8: 280-291. 4a. The genus Magnolia in China. J. Bot. бор. кз 1: 280-305. n enumeration of vascular plants from Chekiang III. "n Biol. Lab. Sci. Soc. China Bot. Ser. 9: 204-3 1942. In: W. P. Fang, Icones Plantarum 4. The National Szechuan Uni- пен Catalogus Plantarum Phanero- gamarum quae in Horto Botanico Bogoriensis Colun- —— ——. 1936. An enumeration of vascular plants from Chekiang III. o Biol. Lab. Sci. Soc. China Bot. Ser. 10: 93-155. CHENG & Y. W. Law. Iconographia Cormophi- tor | 15. Science Press, Beijing. ERAS А. 18. Bull. Econ. 132: 790. CHIu, P. L. & 7. H. CHEN. 1989. А new species of ну 787 Ма, CHOWDHURY, Н. . DANIEL. nations in Magn olia L. Indian J. Forest. 4(1): 64. CHUN, W. Y. 1940. Additions to the flora of Kangting and south-eastern China III. Sunyatsenia 4: 172 63. Genus speciesque novae eru ne rum sinensium. Acta Phytotax. Sin. 8: 286 —— ——,. 1964. A new Hainanese Magnolia. Acta Phy- totax. Sin. 9: 117-118. m Y.W. Law. 1964. Flora Hainanica 1: 222- 0. Science Press, Beijin Е С. В. 1889. Оһ the og E Kohima and Munepore. : Linn. Soc. Bot. 06. Corner, E. J. 1940. коо о of Мајауа: 2-434. Government Printing Office, Singapore. CnarB, W.G. Co i ана to the flora of Siam. Kew p 3405. 165-24 . . Flora € Enumeratio 1: 24— 27. син Society, Ban 1925b Contribution to the flora of Siam. Kew Bull. 1925: 7-23. Совтт & W. Sh Un 1913. Records Bot. Surv. Curtis, C. s А catalogue of the flowering plants and ferns found growing wild in the Poe of Penang. J. Straits Branch Roy. Asiat. Soc. 25: 67-163 Рамру, J. E. 1927a. The genera of Ma gnolicae, Kew Bull. КЕ 257-264. 27b. Michelia kisopa and M. doltsopa Я 9. lieae. Notes Roy. Bot. Gard. Edinburgh 16: 132. . 1928b. Malayan Magnolieae. Kew Bull. 1928: 93. Three new Magnolieae. J. Bot. 66: 19284. Two Michelias from Kwangtung. J. Bot. 66: 359-361. X ; w Michelia from the ^^ of Tibet + Assam. Kew Bull. 1929: 222- 929b. pru: new Michelias from todo China. J: Bot. pos 222-22 1930. New Maglie from China and Indo- China. i Bot. 68: a. The Men UE of pa us "d cao, de Hongkong. Lingnan Sci. J. 7: 137- 1931b. Four new Magnolieae from i J: Bot. vd 231-233. 34. The identity of Lassonia Buc'hoz. J. Bot. 2 101-103. ————. 1943. Magnoliaceae collected in Kwangsi by R. C. Ching 1928. Sinensia els. —— ——. 1948. Magnolia duni awe Bot. Mag. 164: t. 9678 & 9679. —. 1950. A survey of the genus Magnolia to- gether with Manglietia and Michelia. In: Camellias nolias Conference pom Royal Horticul- ndon and Mag ai e 64-81. Lo 1092 Annals of the Missouri Botanical Garden 1964a. In: J. Hutchinson, Los Genera ba Flowering phares Angiospermae 1: 54-56. Clar don Pr der d. 964b. Manglietia insignis. Bot. Mag. 175: t. y 1973. bi sn hypoleuca. Baileya 19: 44. In: J. Praglowski, Magnoliaceae Juss. World Pollen ik Flora Vol. 3. Almqvist & Wik- sell, Stockh 19 ЈА revised survey of the genus Mag- nolia together with Manglietia and Michelia. In: За M an Magnolias: 29-37. Faber & Faber, Dec ron J. 1835. Herbarii Timorensis Descriptio: 93- ret, Paris. le. 1 P. B. 1821. Icones Selectae Plantarum T 1-26. Paris. ; C. Y. 1986. | он Inst. Forest. 1: 45. Doa R. L. Histoire des arbres et des no qui Abd étre cultivés en plein terre r le sol de la ^I rance 2: 6. ; Pw Paris. DESROUSSEAUX, L. In: J. P. A. P. M. de pe is Eneyelopédie Méthodique. "eius 3: 1-675 и | 1900. ‘Die Flora von Central-China. Bot. Jahrb. yst. 29: 169-659. 1912-1913. Plantae Chinenses oo 11. pd 1983. Ј. fane. i ms А E ra of Henan 1: 509-5 . A General History of the Tru Plants 1: 80-84. London. 1903. Descriptions of new Chinese plants. inn. Soc. Bot. 35: 483-518. 1908. А botanical expedition to central Fo- kien. J. Linn. Soc. Bot. 38: 350-373. 911. A supplementary list of Chinese plants, 1904- 1910. J. Linn. Soc. Bot. 39: 411-506. .J. TurcHER. 1912. Flora of соте and Hongkong (China). Kew. Bull. add. ser v. P. 1942. Icones Plantarum Omeiensium 1: 3. National Szechuan University, Chengtu. FINET, A A. & К. GAGNEPAIN. 1905-1906. Con Ha butions à la flore de l'Asie оеша d'aprés l'herbie du muséum de Paris. Bull. Soc. Bot. France 52 (Mémoires 4): 23-54. & ш 1906. сома nouvelles de l'Asie Bull. Soc. Bot. France 53: 573-576. . 1907. In P. Н. Lecomte, Fl. Indo- Chine 1: 1-1070. Paris. Focc, J. M. & J. C. McDawiEL (editors). 1975. Check list of cultivated ү American Horticultural Society, Mount ForBEs, F. & W. B. m 1886-1888. An enu- meration of all the plants known from China proper, ormosa, Hainan, Corea, the Luchu Archipelago, and the Island of Hongkong, together with their distri- bution and lage Бо J. Linn. Soc. Bot. 23: е F.W 1909. Indo-Malayan woods. Philip. J. Sci. 4: 450. orientale. FRANCHET, N R. 1889-1890. Plantae Delavayanae: 1-240. Paris. P. A. L. SavariER.. 1873. Enumeratio Plan- tarum in ш Sponte Crescentium 1: 485. F Savy, Par GAERTNER, J. 1791. De Fructibus et Seminibus Plan- arum 2: 1-520. Stuttgart. GAGNEPAIN, F. 1929. Quelques synonymes de Mag- noliaceae. Bull. Soc. Bot. France 76: 738-739 In: H. Pos Mise Flore Сепегаје de | Inda: Chine Suppl. 1: 93 Mazolscées nouvelles ou litigieuses. Nolae. Syst. 8: GAMBLE, J. S. m Magnoliaceae In A Manual of Indian Timbers: 6. Calcu Icones "Plantarum Asiaticarum: t. , E. Dinc & E. Н. GRorr. 1923. An enumeration of the McClure collection of Hainan plants. uri Agric. Rev. 1: ee J. F. E. 1827. Hybrides Nouvelles. Ann. . Hort. jo 90-95. anos F., né oO Edinburgh 5: 12 Hance, H. F. 1866. Adversa in stirpes imprimis Asiae orientalis criticas min notes interjectis novorum plurimarum rula Xon. Sci. Nat. (Paris) 5, 5: 205. 1858. Trans. Bot. Soc. HaNpEL-MazzrErTI, Н. 1921. Plantae novae sinenses diagnosibus brevibus р ае a Dre. Heinr. Han- del-Mazzetti. Akad. . Wien Math. uM, Kl. 58: 81-89. 1931. Magnoliaceae. /n Symb. Sin. 7: 241- Hara, Н. (editor). 1966. Magnoliaceae. Jn Flora of the astern Himalaya: 95-96. The University of Tokyo Press, Tokyo. 71. Flora of the Eastern Himalaya 2nd. report: 36-37. ЕР notes on some Asiatic plants. ' Taxon 26: 584-587. oss 1935. Confit ad dendrologiam Nip- poniae australis. Acta Phytotax. Geobot. 4: 207- НЕ, 5. Y. ET AL. 1984. Flora of Beijing 1: 265-267. Beijing Press, Beijing. Немзгеу, W. B. 190 The Chinese tulip-tree. Gard. Chron. s 33: 370 905. In: W. J. Hooker, Icones Plantarum 8: t. E London. E. Н. Мпзок. 1906. Some new Chinese plants. Bull. Misc Inform. 1906: 147-163. не М. К. 1928. The flowering plants of Kuala Lumpur in the Malay Peninsula. Gard. Bull. Straits Settlem. 4: 211- HENRY, А. 1896. A list of plants from Formosa with some preliminary remarks on the geography, nature of the flora and economic pad of the island. Trans. Horti posee Lugduno- Batavi Catalogus: 612 t. 613. Leider Hooker, J. D. 1875. Michelia lanuginosa. Bot. Mag. 31: t. 6 . Illustrations of Himalayan Plants: t. 1- : 1872. The Flora of British India 1: 1-740. L. Reeve, Kent. . 1885. Magnolia campbellii. Bot. Mag. 111: t. 6793. 1895. Magnolia parviflora. Bot. Mag. 121: t. 7411 T. Thomson. 1855. Flora Indica 77-78. W. Pamplin, London Volume 80, Number 4 1993 Chen & Nooteboom 1093 Magnoliaceae of China How, F. C. (chief editor). 1956. Flora of Guangzhou: 19. Science Press, Beijing. Hu, H. H. Notes on Chinese ligneous plants. J. pu Arbor 5: 221-233. In H. H. Hu & W. Y. Chun, Icones ои Sinicarum 2: t. 66-70. Shanghai. 1937. Notulae ie eran жез sinen- sium. Bull. Fan. Mem. Inst. Biol. 1938. Recent progress in botanical explora- tion m m J. Roy. Hort. Soc. 63: 940a. Notulae systenaticae ad pma sinen- sem x. Bull Fan. Mem. Inst. Biol. 10: 117-118. 1940b. Paramichelia, a new genus of Mag- побасебя. Sunyatsenia 4: 142-145 1951. А new species of oo from Hupeh. Acta Phytotax. Sin. 1: 335 & HENG. 195la. ый a new genus of Magnoliaceae of southwestern China. Acta Phytotax. Sin. 1: 1-2. & : 951b. Paramanglietia, a new genus of Magnoliaceae. Acta Phytotax. Sin. 1: 255- 256. 1955. A Manual of Economic Plants 1: 376- 91 HUBEI INSTITUTE OF ae (edited by). 1976. = Hupehensis 1: 422-432. People's Press. ES Humpert, Н. 1938. Fl. Indo-Chine Suppl. 1 39. Iunic, Н. С. 3 uod x Mo Arb. Bull. Univ. Wash. 11(2): 3 Iro, K. 1881. Figures and descriptions | соу in the Koishikawa Botanical Garden . 7-14. To- kyo yo. Jackson, B. D. 1894. Index Kewensis 2: 223. London. JIANGSU INSTITUTE OF rs (edited by). 1982. н of Jiangsu 2: 194-200. People's Press, Jiangs JOHNSTONE, G. H. 19 А Asiatic Magnolias in Culti- i ndon vation. uf Hort. Soc., Lo Jue, H. O. 1918. Plantae Thunbergianae: 1-462. Leipzig. Jussieu, A. L. DE. 1789. Genera Plantarum: 281. Paris. KAEMPFER, E. 1791. Icones Selectae Plantarum: t. l- к R. 1930. Concerning a species of Be e Hist. Soc. Taiwan 20: 388-3 Pp. 189- 192 in Formosan жәл геу. ed. Due Research Institute. Taihoku. & Y. YAMAMOTO. 6. Two interesting trees in the PA oo Trans. Nat. Hist. Soc 195 0 Pe new genus of е from = Quart. J. Taiwan Mus. 8: 207 In: T. Whitmore (editor), Tree Flora of Malaya 2: 288. Forest Department, Kuala Lum ur. 1975. In: T. Smitinand & K. Larsen (editors), Flora of Thailand 2: 251-267. The Forest Herbar- ium, Bangkok 1976. In: H. L. Li et al. (editors), dk of Taiwan 2: 287-289. Epoch duis Taip he delimitation of the Ма dis ooo Gard. Bull. Straits Eran 31: 127- kv. га кл Magnoliaceae. J. Asiat. Soc. Bengal 58(2): 3 _——. aol e Magnoliaceae of ae India. n. Roy Bot. rad (Calcutta) 3: 204-21 ыл S. & Murata. 1981. о. In: Ке е и of the Woody Plants of Japan : 218-22 e Publishing, Osaka. iuo G. ioe Contributiones ad floram Asiae orientalis. Bot aa (Tokyo) 40: 330-348, 1926; 43: 382-407, 1903- A ы Manshuriae. Act. Koorpers, S. Н. 1912. Eikaas von Java 2: 24 Jena. T. VALETON. 1896. pur ie F Kennis van der Boomsoorten van Java & 1918. Michelin a Linn. In: Atlas der Baumarten von Java 4: t. 799. Buch. und Steindruckerei von Fa. P. W. M. Trap, Lei KOORDERS-SCHUMACHER, А. 1913. Systematisches Ver- zeichnis 1, Fam. 95: 4-6. Buitenzorg. Коктна15, P. W. 1851. Bijdrage tot de о. дег Indische Magnoliaceae. Ned. Kruidk. Arch. 2: 96- ————. 1905. Revue Horticulture Belge 31 KOSTELETZKY, V. F. 1836. Allgemeine Kiel ede Pharmazeutische Flora 5: 1557-2006. KuNTZE, C. E. O. 1891. Revisio Genera Plantarum 1: 6-7. Wurzbürg. Law, Y. W. 1972. Iconographia Cormophytorum Sin- an l: 786-796. Science Press, Beijin 79. new genus of Magnoliaceae from China: po Phytotax. Sin. 17(4): 7 2; 1 Bull. Bot. Res., Har b 2: 125. 1983. In W. C. Cheng, Sylva Sinica 1: 422- . À preliminary study on the taxonomy of th family Magnoliaceae. Acta Phytotax. Sin. 22: 89-109. Magnoliaceae. In Flora of Xizang 2: Я 1985 5b. Таха a Magnoliacearum. Bull. Bot. Res., Harbin 5(3): 121-131. MS In F. H. Chen (chief editor), Flora of Guangdong 1: 2-20. Science and hnology Press, Guangdong -& 1984. Ne ew taxa of Magnolia from Hench: Bull. "Bot Res., Harbin 4(4): 189-194. . Wv. 1984. Anew species of Michelia from China: Bull. Bot. nien mi 152- 154. & ——. 198 New of Michel from China. Bull. Bot. Res., Harbin Pm 97- 100. 19 A new species of Man- glietia from Guizhou. Guihaia 6: 263-264. 1988. Some new taxa of Michelia from China. Acta Bot. Yunnanica 10: 335-341 R. Z. ZHou. 1986. A new species of Mag- noliaceae from Kwangtung ej he yuan China. Bull. Bot. Res., Harbin 6(2): 13 42. 87. ем taxa of "ipic from China. Bull. Bot. Res., Harbin 7(1): 8 LEE, S. C. 1935. Forest Botany of сн 1 -99]. S | anghai. LEVEILLE, Н. 1904 Ра a a de Chine. Bull. Soc. Agric. Sarthe 39: 316 . 1908-1911. Decades i AM novorum I. Feddes Repert. Spec. Nov. Regni dra 6: 330-332, 1908-1909; 441-463, 1910-191 Li, H. L. 1963. Woody Flora of unm 186-189. Livingston Рамон, Маон, Pennsylvania. Li, Y. K. & X. M. Wanc. 1983. Guizhou Sin. 3: 18. 1094 Annals of the Missouri Botanical Garden —— а ——.. 1987. А new species of Michelia from Guizhou. Acta Phytotax. Sin. 25: 408-409. 1972. Mem. Coll. Agric. Natl. Taiwan Univ. : 96 y 96. 1985. Magnoliaceae. In: Flora Fujienica 2: 76. LINDLEY, J. LINNAEUS, C. 1 holm. 1754. Genera Plantarum Editio quinta. 239- 240. Stockholm 1828. Bot. Reg. 14 164. 753. Species Јавне 535-536. Systema naturae Editio decima 2: 1082- 1083. Stockholm Liou. 1955. Pn. Flora of the Ligneous Plants of NE. China: 248 t Liv, T. S. : an be Natural and Introduced iii Plants of Taiwa m 77. National Tai- an University, Taipei, Та iwa P MN J. 1790. Flora ен Ри band 1: 1–1354; 2: 355-744. Lisboa. LOZANO-CONTRERAS, С. 1975. Caldasia 11: 27-50. MADDEN, E. . Elucidation of some plants men- tioned in Dr. Francis Hamilton's account Ji | king- dom of Nepal. Ann. Mag. Nat. Hist. 18: MakiNO, T. 1912. On узы flora of Japan. Во. Mag. (Tokyo) 26: 82. MasaMUNE, С. 1943. Flora Kainantensis: 83. Depart- ku. A list of plants collected in the island of Kizan. Annual Rep. Taihoku Bot. Gard. 3: 57. Matsupa, S. 1893. J. Coll. Sci. Imp. Univ. Tokyo 6: Matsumura, J. & B. Hayata. 1906. Enumeratio pu tarum in Insula Formosa. J. Coll. Sci. Imp. Uni okyo 22: 1- Max1IMOWITZ, С. J. 1872- 1873. Diagnoses breves plan- tarum novarum Japoniae et Mandshuriae. Bull. Acad. Imp. Sci. Saint-Pétersbourg 17: 417-456, 1872; 18: 275-296, 1873. MERRILL, E. D. 190 A review of the identifications of the уар d in Blanco's E de Fili- t. Lab. Publ. 27: 1- NS on Cuming's Philippine enn . The flora of iai Lamao forest reserve. Philipp E ~ l, Suppl.: 41. . А flo бе in Bur. Sci. Publ. Ma- EN 5: d. sog Man 1916. ie Robinsonianae. Philipp. J. Sl 11: 270 19 "n interpretation of Rumphius’ Her- barium Aaaa, Bur. Sci. Publ. Manila 9: 1- — — —. 1918. Species Blancoanae. Bur. Sci. Publ. Maria че. 1-423. — 2la. A bibliographic enumeration of Bor nean ыт Straits Branch Roy. Asiat. Soc. Special number 37. 921b. A review of the new species of plants proposed by N. L. еи in his Flora Indica. Phi- ћрр. J. er 19: 329-38 А . An пина of Philippine Flow- ering Рам 2: 1-530. Manila. 27a. New Chinese ligneous plants. J. Arnold reu. 8: 3-19 927b. An aoe of Hainan plants. Lingnan Sci. : 5: 1-186. Я А supplementary list of Hainan plants. Lian Sci. J. 6: -289, 323-332. —— 19 Unrecorded dd for Kwangtung province II. Lingnan Sci. J. 7: 297-326. — 19 commentary on Loureiro’ s “Flora Cochinchinensis.” Trans. Amer. Philos. Soc. 24: 158- 159. — а Қ. ROLFE. rare Notes on Philippine ее Pili. : Sci. 3: Meyer, F. С. & E ое 1987. Rejection of the names Ma agno olia heptapeta siii quinquepeta (Magnoliaceae). Taxon 36: 598- iiir 4 a id 27. Magnolias. ек Green & на, "d s Кя 1858. Flora Indiae um 12: 13-18. С. С. van der Post, Amster y 18 E. Indiae Batavae. ел primum 1: 1 . G. van der Post, Utrecht. . 1861 "Fors с Batavae. ee primum 2: 367. C. C. van der Post, Annales Musei Botanici Dadun Ba- .G er Post, Amsterdam. t, Amster . 1846. In: H. oe Буна aià Verzeichnis: 36. Solothur idees C. 1857. In: W. "с. Walpers, Annales Bo- tanices Systematicae 4: 1-39. Leipzig. Momo F. 1891. /n: G. Watt, A Dictionary of the Economic Products of India 5: 241-243. Calcutta. 1798. _ Systema Vegetabilium. Editio 1933. Flora Sylvatica Koreana 20: 117- . Seoul. NEES VON ESENBECK, C. G.D. 1825. Magnoliaceae. In: merkwardigste zoo inals e Gewassen etc. Flora 8(1): 152. Notes on Magnoliaceae I. 1987. Notes on Magnoliaceae II. Blumea 32: Ronee = F. 1791, Verh. Batav. Genootsch. Kunst. Wet. 5, Онва, Н. 1980. PT tes on Magnolia qid Salisb. 88- var. purpurascens Maxim. J. Jap. 92. Ouwi, J. 1978. ји of mM rev. ed.: 648-649. OLIVER, 2 1891. In: W. J. Hooker, Icones Plantarum 10: t. 1953. Ta i OSBECK, | 1757. Dagbok öfwer en Ostindisk Resa: 1- 376. Stockholm. "d e 1910-1915. Le piante vascolari racolte . P. C. Silvestri nell Hu-peh duran La anni 1904- 1907. Nuovo Giorn. Bot. Ital. er. 17: 233-298, 1910; 22: 249-296, 1915. 1915. La Magnolia sprengeri Pamp. e la Magnolia оа Rehder et ||“ Bull. Soc. Tosc. Ortic. 40: 99-102, 199-20 Le Морао, Bull. Soc. Tose. Ortic. 41: 135-141. PARMENTIER, Р. 1895. Histoire des Magnoliacées. Bull. oc. France Belgique 27: 159-337. Parsons, T. H. & J. F. Ropinson. 1875. Gard. 8: 67, 69, 494 Volume 80, Number 4 1993 Chen & Nooteboom Magnoliaceae of China 1095 & 1877. Gard. 12: ae ———— а Gard. 13: PAvoLINI, A. Е. 1908 Contributo alla е dell’ Hu- ре. Миоуо s Bot. Ital. 15 ы E H. 1806. Synopsis sinn 2: 94. Tü- bin PIERRE, T B. L. 1879-1883. шз Forestiére de la Cochinchine 1: sub t. 1-96. Par PLUKENET, L. 16 -1692. di rnm 1: t. 1-120, 691; 2: t. 121-250, 1692. London 1696. Almagestum Botanicum. 1-402. Lon- don. Posr, T. К von & О. KUNZE. phanerogamorum: 588. U Ралан, Ј. 1974. World Pollen and Spore Flora. 23 а a T M ie d lm D. C. S. & M 80. New combi- PRU in the m ита L Indian J. Bot. 3, 1903. Lexicon generum sala. Raju, 2: 17 REHDER, ^ 1916. In L. H. Bailey, The Standard Cy- clopedia of Horticulture. 4: 1967. Macmillan, New ork. 924 | of the ligneous Lu of PA China, J. Arnold Arbor. 5: 137- Ma dolia. Pp. 246-252 in S Manual sl Gaia Trees and Shrubs. Macmillan, New 1929. Notes on the ligneous rm described by H. T e J. Arnold Arbor. 10: 196. Zur Nomenklatur der vou de- TUR. je M. liliflora Desrouss. Mitt. Deutsch. Dendrol. Ges. 41: 342-346. 1 New species, varieties and combin tions s from the collections of the Arnold А Бетина J. Arnold Arbor. 20: 91. Magnolia. Pp. 246-252 in Manual of Cultivated Trees and Shrubs, 2nd ed. Macmillan, New York. 1949. Pp. 180-183 in Bibliography of Cul- tivated Trees and Shrubs. The Arnold Arboretum of Harvard University, Cambridge. & E. Н. Wilson. 1911-1913. Magnoliaceae. Pp. 391- 418 in C. S. Sar argent, Plantae Wilsonianae 1. University Press, Cambridge (U.S.). & 1927. Ligneous plants of Anhwei. . Arnold Arbor. 8: 109. RIDLEY, H. N. 1913. си to a flora of Borneo. Sarawak Museum J. . The Flora of the Malay Peninsula 1: 14-15. L. Reeve, Lon 1925. The Fora of the Malay Peninsula 5 6. L. AW Londo — R. A. 1796. ee Stirpium in Horto 1 пије Vigentium: 379. London. ————— . 1806. Paradisu der | t. 38. London. ————., 0. Paradisus Londinensis 2: t. 87. London. SARGENT, C. S. 1893. Note ps the forest flora of Japan. Gard. | пева 75- -1905. «фы and Shrubs 1: 57-58. oston pen New York. SasakI, S. 1930. Pp. 215-217 in A Catalogue of the Government Herbarium. Government Research In- 1830. Verh. Batav. Genootsch. . ZUCCARINI, 1845. Flora Japonicae familiae naturales, adjectis generum et specierum exemplis а Sectio prima. Plantae dicotyledo- eae e. Abh. Math.-Phys. Cl. Kónigl. Bay- er. Sims, J. Magnolia pumila. Bot. Mag. 25: t. d Magnolia conspicua. Bot. Matg. 39: t. 162 SKAN, > A 1906. Magnolia hypoleuca. Bot. Mag. 8 1915. pue Chinensis. Notes Roy. Bot. Card Edinburgh 8 E acre einn, Notes Roy. Bot. Gard. Edinburgh 12: 21 16. Sonc, X. Н. 1984. |. Nanjing Inst. Forest. 4: 46. SOULANGE-BODIN, E. Ex THIEBAUT DE BERNEUD, А. 1826. Notice sur une nouvelle "vaa de Magnolia. Mem. Soc. Linn. Paris 1826: c d Beau Тетова de Fructification. nn. Inst. Roy. Hort. Fromont 3: 150. 1839. Нисе Naturelle des Vegétaux. Pha- nérogames. Les Magnoliacées 7: 429-490. Paris. Linnaea 15: 162 Magnoliaceae hardy in tem- perate North America. J. Arnold Arbor. 57: 250- 311. SPRAGUE, "i A. 1909. Magnolia delavayi. Bot. Mag. 135: t. 8282. SPRENGEL, X. 1825. Systema Vegetabilium 2: 1-939. en. ЅТАРЕ, O. 1924. Magnolia wilsonii. Bot. Mag. 149: t. 9004. . 1927. Magnolia sprengeri. Bot. Mag. 153: 9116. ита Е. : VON. 1841. Nomenclator Botanicus 2nd : 89, 139. NUR 7 Pri A Botanical Materia Medica. 1-549. London SUZUKI, T 936. In: E Masamune, Short Flora of Formosa: i Taiho Pus J. E. & S. € 1866. Catalogus ОЕ а in Horto Botanico Bogoriense Co- luntur: 178. Bat THUNBERG, C. P. 17842. p Illustratus. Nova Acta Regiae Soc. Sci. Upsal. 4: 31-40. 1784b. Flora japonica Magnolia: 236-237. Leipzig. 1794a. Botanical observations on the Flora Japonica. Trans. Linn. Soc. Lond 26-342. 1794b. 94b. Mus. Nat. Acad. Upsal. 16: 137, 1805. Icones plantarum japonicarum. 5: 1- 10 mc 1824. Hut Japonicarum Novae Spe- cies: : 8. Uppsa ТІЕР, N. V. 1980. Beitrage zur sippenstruktur der gat- tung Manglietia Bl. (Magnoliaeae). Feddes Repert. Spec. Nov. Regni Veg. 91: 497-576. ———, W. VENT . 1980. Über die gattung Manglietia Blume. Wiss. Z. Humboldt-Univ. Berlin, Math.-Naturwiss. Reihe 29: 323-328. 1 Magnolias. Faber & Faber, Taxonomic study of Magnolia ia boldii С. Koch. Acta Phytotax. Geobot. 31: 11 A nomenclatural revision of the Ja anese Magnolia species (Magnoliac.), together with Annals of the Missouri Botanical Garden two €: n Chinese species. Acta Phytotax. Geobot. 9-161. T . А nomenclatural revision of the Jap- anese se Magnolia species (Magnoliaceae) together with two long-cultivated Chinese species I. M. hypoleuca. Taxon 35: 340-344 UnBaN, I. 1927. Sertum antillanum. си Feddes epert. Spec. Nov. Regni Veg. 24: 1-12. VEITCH, J. H. 1902. Some lesser-known pem trees and shrubs. J. Roy. Hort. Soc. 27: 857-875. VENTENAT, E. P. 1806. Jardin de la Malmaison 1: t. VipaL, S. 1885. deri ока Cumingianae Philip- Зе“ 19, 91. Мап 1886. Revision de Plantas Vasculares Filipi- nas 38. Manila. VILLARS, F. 1880. In F. M. poni Flora de Filipinas, 3rd ed., 4: Nov. App.: 4. Vos, C. DE. 1887. P. 116 in Handboek Boomen en Heesters. 2nd ed. Watuicu, N. 1824. Tentamen Florae Napalensis Illus- tratae: 3-8. Calcutta 1828- е а numerical list of dried plants in the East India Company's museum. (often cited as Wallich's duc no. 6493-6495. London. Lisr 1. that in the te Magnoliaceae KEY TO THE GENERA OF MAGNOLIACEAE MAGNOLIACEAE UE MAGNOLIOIDEAE Tribe Magnoliea Tribe Michelieae Tribe Magnolieae Wicut, R. 1840. Illustrations of Indian Botany 1: 13- . Madras. WILLDENOW, C. L. 1799, Species Plantarum 2: 1- 2. Berlin. Wilson, E. H. 1905. The Chinese tulip-tree (Lirio- dendron chinensis). Fl. & Sylva 3: 202-204. ylv he Chinese Magnolias. Gard. Chron. ПІ, 39: 324 ——. 1926. "Ма gnoliaceae collected by J. F. Rock ~ Yunnan and Indo-China. J. Arnold Arbor. 7: 235- Wu, c X. & W.T. WaNc. 1957. “Preliminary reports on studies of the plants of tropical and subtropical regions of Yunnan 1” (original title in = = nd Russian). Acta Phytotax. Sin. 6: 192- YAMAMOTO, Y. 1925. Supplementa a CN Formosanarum 2: 1-40. Taihoku. YANAGIDA. ee J. Soc. Forest. (Japan) 16: 268. Yue, Z. S. 1987. A new ornament al peot- Mickela pe ааннИ Acta Bot. Yunnanica 9: 413-416. ZHENG, Q. F. 1981. A new species of d гум from Fujian. Bull. Bot. Res., Harbin 1(3): 92-94. . M. LiN. А new кын Bs Michelia from China. Bull. Bot. Re. Harbin 7(1): 63-65. al of taxa and keys in the order in which they appear in the text. Numbering system corresponds to I. Magnolia KEY TO THE SUBGENERA AND SECTIONS OF MAGNOLIA KEY TO THE SPECIES OF MAGNOLIA primarily based on floral characters KEY TO THE SPECIES OF MAGNOLIA primarily based on fruit characters a. Magnolia subg. Magn nolia | 1. Magnolia sect. См illim KEY TO THE SPECIES OF MAGNOLIA cl GWILLIMIA & C. H. Tso . Magnolia albosericea Chun DN > gun w gs E o, SEE a Ф, 5 < £5 pe — = fo 5 O = | Magnolia phanerophlebia B. L. Chen N - Magnolia sect. Rytidos ospermum KEY TO THE SPECIES OF MAGNOLIA SECTION RYTIDOSPERMUM 7. Magnolia hypoleuca Siebold & Zu cc. 8. Magnolia officinalis Rehder & E. H. Wilson. 9. Magnolia rostrata W. W. Smith Magnolia sect. O x ama KEY TO THE SPECIES OF sip ipis кечан ОҮАМА i Thom 12. Magnolia sinensis (Rehde er & E. H. Wilson) d 13. Magnolia wilsonii Hiis & Gagnep.) Rehder » Magnolia sect. Gyno 14. Ma 15. Magnolia nitida W. W. Sm opodium KEY TO THE SPECIES OF MAGNOLIA SECTION GYNOPODIUM ndy. gnolia kachirachirai (Kanehi ira & Yamamoto) Da ith. Volume 80, Number 4 Chen & Nooteboom 1993 Magnoliaceae of China 1097 a. Magnolia nitida var. nitida b. Magnolia nitida var. lotungensis as & C. T. кон! Б. L. Chen & Noot., stat. nov. a B. L. c. Res olia nitida var. robust n & Noot. va 16. Ma beris omeiensis о Dandy. 5. Magnolia sect. Alc 17. Magnolia сап (Hook. f. & Thomson) Noot. b. Magnolia subg. agnolia sect. Yu ld KEY TO THE SPECIES a MAGNOLIA SECTION YULANIA 18. Magnolia amoen 19. Magnolia campbell Hook. f. & Thom 22. Magnolia sargentiana Rehder & E. H. Wilson. 23. Magnolia sprengeri Pamp. 24. Magnolia zenii Mug 2. Magnolia sect. Buer KEY TO THE SPECIES. vid MAGNOLIA SECTION BUERGERIA 25. Magnolia biondii 26. Magnolia eylindriea E H. Wilson. lia ko 28. Magnolia PERA (Siebold & Zucc.) Maxim. 3. Magnolia sect. Tulipast Magnolia pela а (Buc'hoz) Dandy. с. анон subg. Talau . Blumiana " Mag nolia candollii (Blume) H. Keng var. obovata (Korth.) HYBRIDS AND CULTIVATED SPECIES NOT NATIVE TO CHINA Magnolia xsoulangiana Soulange ex Thieb. sect. Theorodon Spach (of subg. Magnolia) L; ^ Magnolia grandiflora DUBIOUS SPECIES II. Manglietia KEY TO THE SPECIES OF MANGLIETIA primarily based on floral characters KEY TO THE SPECIES OF MANGLIETIA primarily based on fruit characters 1. Manglietia Section Manglietia 1. Manglieti 4. Manglietia duclouxii Finet & Gagnep. 5. Manglietia fordiana Oliver 5a. Manglietia fordiana Oliver var. fordiana 5b. Manglietia fordiana Oliver var. calcarea (X. H. Song) B. L. Chen & Noot. stat. nov. 5c. Va жең fordiana Oliver var. forrestii (W. W. Smith ex Dandy) B. L. Chen & Noot. stat. etia ари Oliver var. kwangtungensis (Merr.) B. L. Chen & Noot. stat. nov. 5d Sli 6. Manglietia. garrettii 7. Manglietia glauca Ше var. sumatrana (Miq.) Dandy. 11. Manglietia lucida B. L. Chen & S. C. Yang. 12. Manglietia megaphylla Hu & abes 13. Manglietia microtricha Law 14. Manglietia moto Dandy. 15. Manglietia pachyphylla Chang. 16. Manglietia szechuanica Hu 17. Manglietia venti bai 2. жш Manglietias tru . Manglietia sinica (Law) B. L. Chen & Noot. comb. nov. ш. Emu ]. Kmeria septentrionalis Dandy. Tribe Michelieae . Michelia KEY TO THE SECTIONS OF MICHELIA KEY TO THE SPECIES OF MICHELIA primarily based on floral characters KEY TO THE SPECIES OF MICHELIA primarily based on fruit characters l. Michelia Section Michelia 1098 Annals o Missouri Botanical Garden ST 2. KUENIA . M. fordiana var. calcarea (X. H. Song) B. L. l. K. septentrionalis Dandy Chen & Noot. 2. K. duperreana (Pierre) Dandy 9b. M. fordiana var. forrestii E W. Smith ex Dandy) B. L. Chen LIRIODENDRON 9c. M. fordiana var. bos (Dandy) B. L. 3. L. chinense еи Sarg. ~ [^ » г List of species showing numbering system used in the Collections Examined list that follows. KEY TO THE SPECIES OF SECTION MICHELIA 1. Michelia aenea Dandy 2. Michelia cavaleriei Finet & Gagnep. 3. Michelia champaca L. 4. Michelia compressa (Maxim.) Sarg. 5. Michelia coriacea Chang & B. L. Chen. ex DC. 7. Michelia elliptilimba B. L. Chen & = sp. nov. 8. Michelia flaviflora Law. & Y. F. W 9. Michelia floribunda Finet & она 13. Michelia ingrata В. L. Chen 4 5. C. Yang. 14. Michelia kisopa Buch.-Ham. ex DC. 15. Michelia lacei W. W. Smith 21. Michelia ee dup Chun & Y. Wu. 22. Michelia velutina DC. 23. Michelia wilsonii Finet & Gagnep. 24. Michelia xalba DC. . Michelia Section Anisochlamys 25. Michelia “еже жш . Michelia Section Dichlam KEY TO THE SPECIES ‘OF SECTION DICHLAMYS Michelia angustioblonga Law . Michelia balansae (A. DC) Dandy. 31. Michelia daniBanibs С. Y. Wu ex Law & Y. F. Wu. Michelia Section Micheliopsis 32. Michelia figo (Lour.) Sprengel. 32a. Michelia figo var. figo 32b. Michelia figo var. е (Law) В. L. Chen & Noot., stat. nov. чаш 33. Michelia yunnanensis Fra . Michelia Section Tsoongioden re 34. о acai oe Nook: & B. L. Chen., stat. nov. Section Paramichel 35. Michelia | baillonii (Pierre) Finet & Gagnep. DUBIOUS SPECIES li pcena ESAME LIRIODENDROIDEAE aid ГҮР ndron chinense А Sarg. 2 Liriodendron tulipifera . conifera Dand Chen & Noot. 10. M. garrettii Craib 4. L. tulipif 5 duode 11. M. glauca var. sumatrana (Miq.) Dandy 12. M. grandis Hu & Chen M 8 8 UNS 13. M. hookeri Cubitt & W. W. Smith 5. M. aromatica Dandy 14. M. insignis (Wall.) Blume Volume 80, Number 4 1993 Chen & Nooteboom Magnoliaceae of China 1099 15. M. lucida B. L. Chen & S. C. Yang 16. M. megaphylla Hu & Cheng 17. M. microtricha Law 8. M. moto Dandy 19. M. pachyphylla Chang 20. M. sinica (Law) B. L. Chen & Noot. 21. M. szechuanica Hu 22. M. ventii Tiep MAGNOLIA SUBGENUS MAGNOLIA Section Gwillimia 23. M. albosericea Chun & C. H. Tsoong 26. M. delavayi Franchet 27. M. henryi Dunn 28. M. phanerophlebia B. L. Chen Section Купаозреттит 29. M. hypoleuca Siebold & Zucc. 30. M. officinalis igiiur & E. H. Wilson 31. M. rostrata W. mith Section Oyama 32. M. globosa Hook. f. & Thomson 33. M. sieboldii K. Koch 34. M. sinensis (Rehder & E. H. Wilson) Stapf 35. M. wilsonii (Finet & Gagnep.) Rehder Section Theorhodon 36. M. grandiflora L. Section Gynopodium 37. M. kachirachirai (Kanehira & Yamamoto) Dandy 38. M. nitida W. W. Smith var. nitida 38a. M. nitida ка lotungensis (Chun & C. T. d B. L. Chen & Noot. b. nitida var. robusta B. L. Chen & Noot. 39. M. omeiensis (Cheng) Dandy Section Alcimandra 40. M. cathcartii (Hook. f. & Thomson) Noot. SUBGENUS YULANIA Section Yulania 41. M. amoena Chen 42. M. campbellii Hook. f. & Thomson 3. M. i ehder & E. H. Wilson 44. M. heptapeta (Buc how) ат i ehder & E. н. Wilson 46. M. sprengeri Pamp. 47. M. zenii Cheng Section Buergeria 48. M. biondii Pamp. 49. M. cylindrica E. H. Wilson 50. M. praecocissima Koidz. 51. M. tomentosa Thunb. Section Tulipastrum 52. M. quinquepeta (Buc'hoz) Dandy 53. M. xsoulangiana Soulange ex Thieb. SUBGENUS TALAUMA Section Blumiana 54. M. candollii var. obovata (Korth.) Noot. MICHELIA Section Michelia 55. M. xalba DC. 56. M. champaca L. 57. M. doltsopa Buch.-Ham. ex DC. 58. M. elliptilimba B. L. de : e 59. M. flaviflora Law 60. M. floribunda Finet & ы, 61. M. fulva Chang & B. L. Chen 62. M. kisopa Buch.-Ham. ex DC. 63. M. a Hand.-Mazz. 64. M. velut 65. M. wilsonii Finet & Gagnep. 66. M. aenea Dandy 67. M. io Finet & Gagnep. oveol 71. M. fujianensis E F. . ingrata B. L. Chen & s. C. Yang . lacei W. W. Smi . macclurei Dandy M M M M. dee ани M M M iae nn : mediocr s Da ndy : shiluensis Chun h Y. Wu Section Anisochlamys 79. M. hypolampra Dandy Section Dichlamys 80. M. angustioblonga Law & Y. F. Wu 81. M. balansae (A. DC) Dandy 2. 84. M. martinii (Н. Lév.) Finet & S E po = 85. M. xanthantha C. Y. Wu ex Law & Y Section Micheliopsis 86. M. figo (Lour.) Sprengel var. figo 86a. M. figo var. crassipes (Law) B. L. Chen & Noot. 87. M. yunnanensis Franchet ex Finet & Gagnep. Section Tsoongiodendron 88. M. odora (Chun) Noot. & B. L. Chen Section Paramichelia 89. M. baillonii (Pierre) Finet & Gagnep. IST3. List of exsi Numbers in italics following the colon ian the cellas n number refer to species numbered as in List 2. [Please note: herbarium designa- tions are not given because this part of the first author's 1100 Annals of the Missouri Botanical Garden notes remains in China and has not yet been made avail- able. The history of the project is outlined in the intro- ductory section to the article.] An ind Exped. 1112: 35; Anderson T. 5: 57; Ao H.X. hs Badal Khan 102: 89 (type); Balansa B. 3884: 24 (type), 3885: 24 (type), 3886: 81 (type), meee ha (type), 25; Beiliu yn 8 4102: 6; Bernardi 19891: 68; Bock C. et Paare 2325: 65; Bodinier E. & Ducloux E. , & Ducloux 116: "d (type), & е cloux 168: е 1211: 9, 1221 ; Bon M. l'abb 3176: 24 (type); Bonnet А. s.n.: 7; ^d de d'Anty 30: 6: 6; 67, 1562: 84, 2263: да, & Крн 226. 3: 67 (туре), 3044: 14, 3045: 67 (type), 3045 А: 67 (type), 3182: да, 4619: 87, 7111: 52, 7150: 44, 7312: 67, 7313: Peng c: i 12179: 7 (type), 12344: дс 13292: 74, (McClure) 14599: 70; пена 37: 24 (type), 37: 24 (type); Chan K.Y. 1033: 9; Chan Shi Sam ~ 8825: 68; e H.T. 1564: 73, 3375: 86, 4029: 52, 4620: 44, 4621: 14, 4700: 67 (type), 5022: 38, 5062: 87, 5094: 3, 5129: 87, 5215: 52, 5218: 3, 5244: 49, 5309: 3, 5394: 3, 5396: 3, 5659: 65, 5702: 84, 5747: 87, 5824: 87, 6037: 86a, 6081: 81, 6095: 9, 6268: 87, 6342: 7, 6477: 87, 6663: 11, 6824: 79, 6825: 79, 7044: 86, 7471: 76, 7472: 76, 7474: 76, 86031: 24, 86094: 24, 860088: 9; 82, 87 T 1:28, ВО 1: 42, & C.N. Mai 87 T 2: 15, 20, 87 Т 29: 57, 87 T 30: 57, 87 T 31: 38, 87 T 32: 13, 87 T33: 82 Dy 87 T 34: 81, 86 S 49: 57, 86 S0 50: 57, 86 S 51: 73, & Su 86 S 53: 14 87 F 167: 57, 87 F 168: 72, 87 F 169: 1, 87 F 171 6-X-: 15, 87 F 173: 1, 87 F 177: 69, 86 GS 177: 14, 86 GS 177: 14, 87 F 178: 69 пе, Li Е 180: 20, 87 F 181 10-X: 16, 87 F 182: 7 192: 82, 87 F 192: 6, 86 S 193: 66 (type), GL 86- i 61 (type), 87 F 194: 6, 87 F 195: 6, 87 6: , 87 F 197: gb, 87 F 198: 8, 887 F 200: a 87 F 201: 7, 87 F 203: 38, 87 F 205: 16, 87 F 206: 16, 87 F 219: 7, 87 F 220: 8, 87 F 221: 7, 87 F 222: 24, 87 F 223: 81, 87 F 224: 60, 87 F 225: 60, 87 F 226: 14, 77 F 227: 26, 87 F 228: 87, 87 F 253: 82, 86 S 256: 14, 86 S 297: 12, 86 5 299: 2155: 78, GS 3240: 38, GS 8676: 40, GS 8698: 6, GS 9023: 38, GS 9028: 60, GS 9035: 69 (type), GS 9044: 57, GS 9047: 72, 80109: 9, 80110: 76, 80115: 7, et al. 81001: 30, 81006: 86a, 81007: 86a, 81008: 70, GS 86002: 66, GS 86075: 60, GS 86100: 82, GS 86102: 73 (type), GS 86103: 14, GS 86177: 14, GS 86178: 12, GS 86179: 12, GS 86180: 12, GS 86182: 6 (type), GS 86184: 14, GS 86186: 14, GS 86187: 66, GS 86188: 60, GS 86189: 66, eee 12, GS 86194: 15, GS 86195: 15, GS 86237: 57 re GS 86238: 26, GS 86239: 57, Gs 86243: 5, : 26, GS 863348: 7; Chen C. 452: 14; Chen 2 1675: 86, 500224: 6; Chen J.C. 745: 3; Chen L.C. 500183: 7; Chen M. 1321: 49, 3792: 60; pen M.L. 3-5535: 24; Chen Q.H. 942: 6; Chen S. 587 30, 1190: 30, 1315: 86, 1736: 44, 1899: 44, 2692: 41 (type), 2706: 44, 2775: 76, 2828: 76; Chen S.J. et al. 840055: 52, 840092: 14; Chen S.P.; 1131: 74; Chen T.C. 617: 9c, 707: 82; Chen Y.F. 1657: Z.Z. 50986: 14, 51010: 67, 51134: 14, 51153: 14, 51165: 33, 51353: 3, 51516: 86, 51808: 70, 51881: 67, 52125: 14, 52375: 82, 52523: 3, 52664: 67, 52980: 76; Cheng Fung Tung CCC, s.n.: 86; Cheng W.C. & С.Т. Hwa 688: 60, 889: 83, 938: 83, 1005: 60 (type), 1049: 46, 1066: 44, 1160: 30, 1171: 52 (type); Cheng W.C. 4030: 9, 4233: 47 (type), 4444 A: 41 (type), 10525: vé Ds Cheo & Wilson 12746: 44; Chevalier A. 29374: 4, 30103: 81, 30744: 82 (type), 37032: 14, о. 6, 37877: 7, 37880: 24, 38185: 79, 38404: 14, 38537: 6, 168: 65, « C.S. Fan 168: 65, & C.S. Fan 436: 30, & C.S. Fan 444: 52, 1230: 25, 1234: 52, 1255: 52, 1864: 3, 2064: 51, 2069: 44, 2099: 65, 2560: 3, 2582: 44, 2597: 29, 2825: 44, & E.D. Merrill 14042: 36, field nr 1050 14349: 30, herb. no. 14884: 52, herb. nr. 18560: 3, 18603: 44, 18702: 44, 18759: 30, herb. nr. 18844: 3; Chien S.S. 5889: 44; Chin T.H. 8218: 6; Ching 1 Peng 3446: 68; Ching R.C. 1602: 44, 1610: 30, 1904: 25, 1986: 86, 2037: 52, 2060: 76, 2167: 38a, 2211: 86, 2370: 3, 2403: 44, 2452: 76 (type), 2472: 9, 2501: 49, 2519: 30, 2617: 44, 2865: 44, 2946: 9, 2949: 49 (type), 2975: 33, 2979: 33, 2994: 49 (type), 3035: 33, 3227: 30, 4764: 44, 5119: 52 (type), 5247: 1 (type), 5613: 86, 5625: 86, 5910: 9b, 6075: 14, 7091: 44, 7117: 6 (type), 7421: 5 (type), 7549: 55, 7949: 24, 8144: 81, 8329: (type), 896: 52, 1076: 14, 1094: 3, 1367: 30, 8255: 84, 8258: 84, 8638: 52, 9777: 45, 11944: 21; Chow H.C. 12095: 39; Chow K.S. 126: 30, 145: 86; Chu K.L. 2505 a: 29, 3908: 45; Chuang C.C. 2442: 68; Chuk Tsz Haang 13775: 86; Chun N.K. 40574: 86, 41303: 86, _ 76, & C.L. Tso 43417: 24, & C.L. Tso 43687: 9, & C.L. Tso 43749: 81, & C.L. Volume 80, Number 4 Chen & Nooteboom 1101 1993 Magnoliaceae of China Tso 43956: 9, & C.L. Tso 44041: 38a, & C.L. Tso Dod cras 31 (type), 18341: 14, 18371: 14, 44042: 38a, & C.L. Tso 44233: 70, & C.L. Tso 44257: 88 (type); Chun S.H. 4466: 74, 9729: 6, 10035: 9, 10140: 7, 12003: 24, 12578: 24, 14206: 6, 15258: 67, 15509: 38, 15634: 86, 15857: 14, 15907: 76, 94617: 67; Chun W.Y. 69: 44, 119: r 136: 56, 4409: 48, 4520: 25, 5856: 86, 5860: 5933: 70, 5963: 76, 6481: 24, 6481 A: 25, 6511; 55, 7179: 70, 7353: 18; Chun Y.F. 30423: 76; мн ҮЕ. 81654: 33; Chung H.H. 213: 3, & S.C. 392: 44, & S.C. Sun 680: 30, 740: 55, 742: 86, 779: 25, 1276: 86, 1620: 86, 2288: 86, 2423: 86, 2452: 36, 2454: 55, 2495: 36, 2496: 55, 2510: 25, 2823: 86, 2833: 86, 2940: 9, 3374: 86, 3667: 76, 3670: 86, 5921: 86, 6020: 36, 6280: 86, 6815: 86, 6903: 86, 8492: 25; Clemens J. & M.S. 4281: 70; Cribb Р.Ј. с 84: 9; Cubitt G.E.S. 20: 13 (type). de e poro Exped. 10240: 38a, 10756: 60, 10952: 080: 70, gel 82, 13429: ya be iel a sis : 86; Dang C.Z. : 87, 1618: 70, 1656: 65; David M UAbbé s.n.: rh : 45 e), s.n.: 45; Delavay s.n.: 87, 46, 168: 87, 2231: 26 a 4218: 87, 4884: 87 tme) 6616: 87, 6786: 87, 6818: die 6828: 8; Deng : я 87, & Bodinier 116: 87, 635: 87, 2133: 8 (type), 2393: 30, 2848: 26, 3794: 87, 6932: 35; Dunn 2448: 86 (type). Eberhardt 2705: + а 3061: 25, 4046: 87, 6346: Fan C.S. & YY. Li Ed 44, 261: 44 (type), 281: 30; Fang W.P. 2343: 52, 12018: 44, 12275: 44, 13256: 52, 15738: 84, 15969: 65, 16598: 65, 16951: 21, 18950: 21; Farges R.P. s : 46, 70: 30, 94: 46, 219: 46, 702: 4, 771: 30, 1300: 48 (type), 1324: 84, 1394: 84 (type), 1447: 52; Faurie U. 536 33, 537: 33, 3372: 29, 42556: 29, 6300: 29; Feng K.M. 4813: 32, 4824: 57, 5193: 9b, 11069: 14, 11169: 26, 11302: 14, 11447: 3 (type), 11558: 3 (type), 11771: 7, 11808: 16, 11990: 5, 12030: 69 (type), 12254: 9b, 12392: 9b, 12464: 69, 12636: 13960: 7, 13973: 82, 14025: 70, 14117: 60, 21124: 42, 22207: 14, 22369: 8, 22416: 40, 22710: 27, 22778: 7, 72059: 23; Fenzel С. 2893: 48, 2951: 48; Fleury F. 30103: 81, 30158: 79 (type), 37667: 79, in Chevalier 37945: 7; Forbes F.B. s.n.: 44, 2202: 44; Ford C. 90: 9 (type), 22224: 9; Forrest G. 57: 26, 140: 42 (type), 482: 87, 2104: 87, 2174: 52, 4105: 60, 5539: 52, 6105: 26, 7182: 35 (type), 7725: 13, 7964: 26, 8459: 52, 8678: 14, 8948: 57, 8949: 57, 9119: 57, 9320: 60, 9460: 60, 9588: 60, 9860: 57, 9907: 14, 10580: 26, 11652: 26, 9b, 12089: 57, 12221: 44, 12282: 87 42, 13253: 42, 13513: 35, 14188: 14 (type), 14248: 42, 14402: 14, 14466: 42 (type), 14868: 42, 14945: 14, 15052: 42, 15056: 14, 15059: 38 (type), 15480: 57, 15496: 35, 15505: 26, 15780: 57, 15892: 26, 15952: 13, 16090: 60, 16388: 42, 16403: 42, 17188: 14, 17248: 42, 17300: 38, 17301: 42 (type), 17654: 57, 17809: 57, 17868: 42, 17955: 14, 18083: 42 , 18497: 57, 18512: 32 (type), 18600: o 57, 18790: 42 (type), 18870: 32 р a 42, 18959: 32, 18964: 14, 19985: 31, 20050 32, 20080: 14, 20303: 32, 20309: 42, 20358: 38 5 | 26705 А: 9b (type), 26726: 14, 26736: 14, 26738: , 27299: 14, 27300: 9b, 27363: 60 (type), 27364: user W.L 9082 a: 9; Furuse, Miyoshi 1883: 68, 3763: 68. де 2 6; de et al. 1: 5, et al. 4: 70, o Z.Y. : 47; Garrett H.B.G. 114: 10 re ud үм sur la Bon 228: 25, 232: Ghof & Tsoong 1961: 6; Giraldi Los. 1897: 48; es 1547: 82, 740250: 60; Griffith 68 B: 64, 71: 57; Grof G.W. s.n.: 86, s.n.: 44, (CCC herb. nr.) 11717: 44; Guan Z.T. 8263: 60; Guang 168: 76, 225: 70, 242: 19; Guangdong Mucai Yanjiu Zu 32: 38a; Guangfu Exped. 19: 67, 463: 6, 465: 6, 877: 67, 1031: 76; Guileier 40937: 81; Gulin Yiyao Gongsi 846: 8; Guo Y.Q. 8463: 57. HFSI Pria 81, 88315: 81, 88323: 9, 68355: 9, 88356: 77, 88383: 81, 88564: 77, сна 9, 88727 а: 77, 88730: 81; Hainan Exped. 752: 70; Hainan Gong Zuo Zhan 643: 24, 1019: 81, т. 24; Hai- nan Zu 75 265: 56; Hand.-Mazz. 6: 87, 294: 86, 683: 67, 1552: 14, 2099: 14, 2204: 14, 2495: 14, 4863: 26, 8304: 42, 9126: 31 (type), 9188: 9212: 32, 9853: 42, 10012: 42, 10672: 14, & T.H. Wang 11131: 14, & Wang Te Hui 12281: 67 (орех Hara c.s. 630283 3: 64, 630691 9: 42; Не J.D. 5040: 82; He S.C. 85170: 67; Henry A. 163: 86, eo ON EN P A: 27 (type), 13147: 55, 13277: 87, 13 à 13513: 89; Hers J 457: 48, 541: 48, 622: 44, 991: 48, 1159: 46, 134 ; Но . 60175: 76; Ho Y. Y. 2364: 2485: 33, 3200: 76, 3201: 30; Hong- ong Herbarium No. 1: 86, 2065: 76 (type), (Dunns пети) 2525, v 2326: 86, 2448: 86 1102 Annals of the Missouri Botanical Garden (type); Hooker J.D. s.n.: 40 (type), s. P 42 (type), s.n.: 32 (type), s.n.: 57, s.n.: 64, s.n.: 40 (type), s.n.: 42 (type), s.n.: 32 (type), s.n.: 54 о. 15: 29; Hook.f. & Thomson s.n.: 100 (type), 997: 64, 26393: 42; Hosseus C.C. 480: 56; Hotta M. s.n.: 68; How F. .K. Chun 70147: 81, 70468: 24, 70732: 9, 70963: 24, 71674: 9, 71797: 24, 71849: 24, 72059: 23, 72157: 9, 72342: 81, 72569: 24, 72740: 23 (type), 73194: 74, 73242: 81, 73257: 81, 73395: 9, 73444: 9, 73480: 81, 73631: 78; Hsia W.Y. s.n. 7-X-: 44; Hsieh A-tsai s.n.: 55; Hsiung Y.K. 6307: 86; Hsu С.С. & М.Т. Kao К 3680: 68; Hu Н.Н. 93: 30, 190: 86, 1256: 86, 1410: 3, 9592: 3, 67125: 32; Hu S.Y. 5329: 36, 5364: 86, 5393: 25, 5485: 55, 6550: 86, 7287: 9, 7664: 9, 9125: 9, 10169 A: 9, 10194: 25, 10201: 25, 10504: 55, 10610: 24, 11578: 86, 12025: 24, 12578: 9, 12946: 53, 13540: 86, 13568: 82; Hu W.K. 8809: 84, 8879: 84, 9211: 84; Huan Jiang Exped. 4-3-1459: 24; Huang D.A. 6034: 30, 60393: 6; Huang D.F. 614: 68; Huang Z.F. 20: 67; Huang Z.H. 66966: 6; Huangshan Exped. 3049: 19; Hunan Exped. 387; 14; Hwa C.T. 38: 30, 65: 48, 81: 52, 104: 48, 213: 44, 213: 46, 276: 52, 317: 60, 357: 83, 381: 65, 404: 30, 441: 84, 502: 65, 503: 60, 528: 46. Іазиті = s.n.: 29; Inokuma T. s.n.: 50; Ip Yuk Shing 817 xs п 33: ао В. 13:67; Јіпхіи 15149: 70, 20202: 76, 2023 KUN 345: 89, 76298: 8; калај Н. 6120: 33, 731152: 50; Kang Ping CCC 15: 86; Kawakami T. 29: 68; Kellogg J. s.n. 2-IX: 53; Keng H. s.n.: 68, K 1048: 68; Keng Y.L. 309: 30, 452: 44 (type), 453: 44; Kerr A.F.G. 4679: 60 (type), 21027: 60; Kingdon Ward F. 58: 60, 331: 57, 370: 38, 529: 31, 3152: 42, 3709: 14, 5497: 57, 9374: 57, 10182: 100, 10365: 42, 10418: 42, 1046 1: 32, 10973: 32, 10980: 64, 17611: 42, 19211: 57, 19299: 57, 20123: 64, 20885: 31; Kit Yock Chan 1352: 25; Ko S.P. 50101: 86, 50379: 82, 50392: 86, 50410: 70, 50520: 76 50540: 24, 50777: 82, 50884: 76, 51082: 76, 51109: 88, 51110: 76, 51152: 70, 51733: 74, 51777: 6, 51928: 88, 52175: 24, 52231: 24, 52263: 9, 52273: 9, 52279: 81 (type), 52664: 76, 52807: 44, 52870: 76, 52921: 6, 53079: 44, 53301: 76, 53514: 18, 53609: 9, 53691: 38a, 53697: 38q (type), 53762: 84, 53912: 76, 54036: 76 (type), 54122: 84, 54186: 70 (type), 54298: 9, 54522: 70, 55702: 81, 55785: 70, 80256: 76, 80358: 84 (type); doen Gong- zuozhang 382: 89; Kuo C.M. 8956: 55; Kwok S.P. 0166: 67, жее 76, 80335: 7 Lace J.H. 5928: 73 (type); Lamont J : 44; Lau 0: 81, 930: 86, 1560: 24, 2142: f^ 2552: 76, 3545: 24, 4047: 86, 4246: e p 86, 4850: 9, 4854: 76, 5245: 77, 25954: , 25956: 9, 26532: 24, 26589: 70, 26947: 24, a 8l, 26975: 9, 27375: 38a, 27946: 81, 28274: 81, 28744: 6; Law Y W. 306: 60, 1385: 44, 4074: 5, 6051: 78, 6067: 27, 7032: 14, 7033: 14, 7066: 70 (type), 7071: 87; Lee S.K. 200723: 74; Lee T.C. 4506: 84; Legendre 863: 35, 1704: 87; Leveillé H. 3044: 14; Levine C.O. CCC 920: 55, 1345: 9, & McClure 6389: 86; Li Bo Exped. 1170: 24; Li B.G. et al. 59; 38a; Li H.Q. 40046: 82; Li M.K. 3: 75 (type), 59: 64 Ma W.W. 2388 (type), 1729: 57, 2329: 32, 3515: 13, 34500: 13; Li Sheng-tang & Su Yu-Zhen 81 73: VE ; Li S.P. Ми 199: 9b; Li T.Z. 1458: 24; Li W.F. 60700: 65; Li W.Z. 8271: 13; Li Y.H. 2066: 89, 3607: 27, 4068: 13, 4262: 79, 5789: 63, 401415: 6; Li Y.K. 185 84, 401535: 9, 401573: 82, 402692: 77; Li Zhen- yu et al. 1899: 67; Li Z.J. 653: 86, 1636: 24, 307 I: 5; Li Z.Q. 99: 70; Li Z.T. 1708: 67, 602441: 70, 603835: 7, 604032: 70; Li Z.Y. 56: 14; Li Z.Z. 822042 3: 65; Liang C.F. 30247: 6, 30248: 38, 30350: 6, 30360: 6, 31067: 67, 31751: 70, 31845: 64334: 24, 6445 1: 9, 64452: 9, 64632: 24, 64633: 24, 64908: 81, 64988: 24, 65255: 9, 67370: 70, 69538: 6, 69622: 24, 69747: 81, 100218: 30; Li- ang J.Y. 100311: 82, 100312: 67, 100313: 76, 100314: 70, 100316: 79, 100319: 56; Liang P.H. 86163: 82; Liang T. 4359: 86; Liao A.M. 507: 52; Liao C.J. s.n.: 27, 15249: 86; Liao Jih-Ching 10006: 68, 10085: 24, 10457: 68, & Yien-kuan Lai 10534: 37, 10654: 68; Licent Abbé E. 2539: 48; Lin Pi 6488: 44; Lin Q.Z. 10365: 67, 10435: 67; Lin Ye Ting 33: 81, 920: 55; Ling Le Exped. 32928: 5; (o ee 350: 68; Liou T.N. 7249: 44; Liu JC. d , & C. Wang 85178: 16, & C. Wang По ps ri J.H. 89847: 38a; Liu L.F. 5354: 6, 5624: 6, 5671: 76; Liu L.Y. 15036: 14, 15146: 38a; Liu T.S., Liao J.C. & Kuo C.M. 2105: 50; Liu КА. 101762: 70; Liu Y.S. 1174: 46, 1212: 46, 1401: 14, 2191: 45; Liu Y.Z. 1: 82; Lo H.S. 28146: 27; Long G.R. 830090: 74, 840223: 82; Long Sheng Exped. 130: 38a, 144: 67, 50180: 67, 50403: 76, 50416: 67; Long Y.Z. 242: 44; Long Zhou Exped. 274: 24; Lu Q.H. 20: 24, 2873: 38a, 3338: 24, 3526: 67, 4114: 76, 4338: 38, 4359: 82; Ludlow, Sherriff & Hicks 16108: 42; Luo Cheng Exped. 4-1- 165: 24, 4-1-1509: 24; Luo Z.J. 2573: 83 : 21; MacGregor 13: 44, 1610: 30, 2452: 25; Maingay 19: 86, 387: 44; Maire E.E. 0: 8, s.n.: 45, s.n.: 87, s.n.: 35, 171: 26, 172: 26, 175: 87, 304: 35, 526: 35, 819: 87, 857: 87, 866: 35, 1379: 87, 1827: 87, 1943: 87, 2100: 07, 2447: 87 E Bodinier 2066: 84 (type); Mas- amune G. 8; Maximowicz s.n.: 44, s.n.: 29, s.n.: 68 эне s.n.: 68 (type), s.n.: 86 (type), s.n.: 29, s.n.: , 1882: 2 А. s.n.: 36, 79: 36, 201: 44, 281: 76, A 342: 36, 582: 77, (LU 18246) 712: 70 буре) (LU 18302) 768: 70 (type), (LU 19761) 782: 74, 1457: 76, 1547: 86, 1556; 86, 2305: 55, (C. се herb. nr.) 6669: 76, (C.C.C.) 8593: 77 (type), (C.C.C.) 9652: 9, (1468 in CCC) 13292: 74 (type), 13775: 86, 13791: 76, 20099: 24 (type), 20124: 24, (CCC herb. nr.) 952653: 76; McLarens’s collectors F 4: 60, L 6 А: 60, UK 8: 87, F 9: 87, AA 13: 87, AD 15: 35, L 18: 87, L 57: 26, F 57: 87, L 66 A: 35, P 82: 26, 85: 42 (type), 222: 14, C 228: 26 (type), C 246: 26; Mell PI. Mellianae 721: 76; Menglian Exped. 10145: 89; Volume 80, Number 4 1993 Chen & Nooteboom Magnoliaceae of China 1103 Mo 10904: 9, 11041: 76; Metcalf T.P. 433: 36 76, & Lau Fun J. 10506: А 10635: 87. Nan Dan Exped. 4-5-787: 24; Nanzhidi 3719: 84, Nong Gang ae 10132: ` 24; North East Yunnan Expedition 754: Oiu H.X. бык Коњ 25; 530: 51, 977: 68. Pang C. к P ; Parmentier s.n.: asheng 2: 77; Peal S.E., Herb, LB 908126 709: 89 yaks E 3 10242: 46; Pen W.M. лиа 76; Реп Z.H. 5 86a; Petelot A. s.n.: 66, o Tay Часе 82 00 3548: 81, 3698: 60, 3599: 82, 3700: 84, 3756: 66, 3760: 14, 3761: 60, 3953: 70, 3955: 70, 3963: 36553: 14; Pierre H.L. s.n.: 56, 750: 89 (type), 752: 24; Poilane 297: 77, 1293: 79, 1857: 88 (type), 2063: 66, 7092: 70 (type), 11253: 75 (type), 12578: 60, 12611: 66 (type), 12864: 73 (type), 17062: 40, 17080: 40, 17092: 6, 18820: 60, 18893: 81, 207 14: 30711: 40, 31039: 38b (type). 32115: 70, 35782: 40; yr 798: 45; Pucha Biaoben 23252: 46; Put 3759 MES. 6213: us Qian Bei Exped. 89: 44, 869: 67, 665 10 Rea d B.E. 1070: 44, R 1071: 52; TT Alfred 737 а: 48 (type); | en James James A. Duke 5734: 87; Rock J.F. 1543: 57, 2766: 60, 3040: 26, 3998: 26, 6257: 26, 6919: 64 (type), 7739: 60, 7888: 60, 7933: 60, 7961: 60, 8015: 57, 8620: 35, 8639: 35, 9625: 60 (type), 10136: 38, 10160: 31, 10209: 32 (type), 10235: 38, 10661: 87, 11215: 31, 11231: 32, 11232: 38, 11517: 42, 11737: 87, 11744: 87, 17097: 31, 18340: 14, 18394: 31, 21990: 38, 21991: 14, 22047: 32, 22457: 38, 22458: 14, 22480: p 22607: 31, 22694: 42, 23066: 42 (type), 23284: Sargent CS. 0: 9. Sasaki S. s.n.: 25, 93: 68; ps 64: 87, 332: 26, 362: 26, 407: 26, 607: 87, : Ve Bot. Exped. 154: 67, 396: 67, 873: 9, 1077: 70, yap 67, 1983: 67; Sino-British Exped. 37: 67, K 41: 87, K 152: 87, 157: 42, 166: 42, 655: 35; Sino Rossica Exped. 17: 87, 952: 14, 161: 89; mus Rossica Yunnan Exped. 1532: 57; Smith, Harry 1595: 87, 1953: 35; Song X.H. 172: 6, 244: 76, 478: 76; Sonohara S. 82: 68; Soulié J.A. 1576: 42; Steward A.N. c.s. 637: 67, 745: 3, 748: 67, 2001: 44, E nr. 2642: БА 4730: 44, 11134: 56, & Cheo, : 67, & Сћео, Н.С. 452: 67, & Cheo, H.C. 817: ob, & Cheo, H.C. 849: 82, & Cheo, H.C. 870: 60, & Cheo, H.C. 1078: 77, C.Y. Chiao & Cheo, H.C. 637: 67, C.Y. Chiao & Cheo, H.C. 748: 67, C.Y. Chiao & Cheo, H.C. 764: 67, C.Y. Chiao & Cheo, H.C. 870: 67; 259: 65, 341: 21; Sun S.C. 1293: 30, 1305: 44 i qe Susuki Jokio 6098: 68; i uan University 53909: 65, 53380: 87 (type), 53491: 87, 54858: 38, 55887: 57, 56371: 60, 56560: 14 (type), 56870: 60, 56961: 73 (type), 58573: 42, 58698: 40, 58995: 57, 59903: 42, 61619: 22, 61621: 81; Tsang H.L. 32980: 65, dn 7: 42, 34828: 45; Tsang W.T. (LU 15822) 323: 1, (LU 17187) 438: 81, (LU 1 K Chow 2738: 70 (type), 3044: 76, 20119: 86, 695: ES lane 5 (type), 2802: 36, 2889: 76, 4141: 67, 5361: = 6801: ab 6933: £6 0986 76, 7534: 7, 8543: 52, 8671: 10701: 44, 10703: З, 11285: 87, 13022: 87, 7, & H. Wang 16023: 26, & H. Wang 16025: 87, To C.L. 20354: 70, 20508: 7, 20561: 70, 20569: 76, 20671: 25, 20905: 7, 20984: 76, 21005: 9, 21033: 82 (type), 21098: 7, 21173: 7, 23510: 81, 43749: 81; Tsoong C.H 82048: 67, 81657: 14, 81948: 9, 82074: 33, 83515: 33, 83580: 6, 83704: 76, 84761: 6, 90960: 76, 808128: 36, 808954: 67, 808958: 82, 809507: 74, 809507: 77; Tsui T.M. 90: 86; Tu & Sun 4070: 3; Tutcher W.T. 637: 86. Uchiyama T. 4065: 33; Ueda K. 459: 33 (type). Vidal J.E. 653 A: 70, 1629: 14, 4932: 14. Wai Chung Ping 13791: 76; Walker E.H. c.s. 6640: 68; Wallich s.n.: 14 (type), s.n.: 64 (type), s.n.: 62, 70: 57, 970 A: 62, 973: 14 (type), 6493: 64, 33359: 24, 33629: 81, 34512: 9, и и 77, | 77, 36050: 9, 36233: 77, 39230: 9, 393 96: 6, 1104 Annals of the Missouri Botanical Garden 40158: 38a, 40306: 67, 41206: 67, 76244: 77; Wang C.S. 4065: 33, 39301: 9b, 62759: 87, 62787: 87, 62848: 32, 63015: 87, 63981: 32, 66921: 38, 67139: 31,67175:42,67553: 38, 67848: 42, 70468: 87, 71201: 52, 71464: 26, 71760: 26, 71873: 60, 72314: 60, 72322: 14, 72447: 60, 72469: 26, 72486: 60, 72551: 60, 72797: 13, 73474: 9b, 73905: 56, 73981: 32, 74288: 56, 74653: 89 (type), 75201: 89 (type), 75649: 89, 75888: 89, 76002: 27, 76518: 27 (type), 76786: 60, 76855: 40, 76888: 27 (type), 77013: 9b, 77129: 89 (type), 77433: 89, 77633 89, 77766: 56, 77767: 56, 78018: 89, 78367: 14, 78440: 40, 78441: 64, 78590: 89, 78975: 56, 80115: 27, 80116: 27, 81213: 60, 83130: 84 (type), & Liu 83157: 1 (type), & Liu 85522: 87, 85542: 60, 86545: 81, 86915: 69, 87020: 12 (type), 87033: 72 буре) 87594: 84, 87640: 60, 87715: 72, 87811: 72, 87854 5, 87891: 84, 88023: 69, 88130: 84, 89653: 81, 89858: 33; Wang F.T. 20441: 30, 20793: 52, 21080: 33 (type), n 33; Wang J.J. 4809: 24; Wang P.S. , 413: 86, 614: 6, & Chiu, H.S. 241: 19 (type; Wang 0 2730: 81, 2772: 9, 2773: 38a, 2774: 81, 2812: 81, 2887: 77, 2973: 81, 830001: 507: 76 ше. 555: 86, 2802: 36, 2889: 76, 3217: 9, 30003: 76; Wang & Ling 7353: 7; Wang, S.Z. 173: 2. 576: 87; Watt G. 6329: 57 (type), 6800: s.n.: 56; Wilson E.H. 21 a: 46, 38: 46, 192: o. 275: 46, 278: 46 (type), 345: 46 (type), 361: 48 (type), 361 a: 48 (type), 371: 30, 373: 46, 413: 3, 434: 84 (type), 444: 46, 444 a: 46, 652: 30 ype), 838: 35 (type), 914: 45 (type), 923: 45, 923 a (type), 1049: 3, 1241: 43 (type), 1374: 35, s 6307: 44, 7037: 50, 8830: 33, 9200 20.1: 33, 10054: 68, 10403: 33; Wong Y.K. 507: 76; Wright C 29; Wu С.А. 62-37: 5, 62012: 82; Wu О.С. s.n.: di Wu S.K. 85: 14, 3867: 63; 4573: 54; 8713: "n 20 s.n.: 37; Xin J.S. 383: 89; Xing An Exped. 5: 76; Xing Fu-wu et al. 145: 86; Xiong J.H. d 84, et al. 93877: 83; Xu S.G. 59-4564: ге 5024: 40; Хи Ү.С. 595: 6, 12935: 9, 63131: Yang С.Н. 54485: 65, 54716: 65, 56176: 65, ж 65, 56886: 84, 57297: 84, 57530: 84; Yang Н.М. 6100: 6; Yang M.Z. s.n.: 75, 84344: 60; Yang X.X. 831164: 82; Yang Y.B. 57: 65; Yao Kan 8304: 46, 9006: 49, 9283: 86; Ye C.X. Gs 3331: 14, 3350: 60, 3383: 87, 82089: 76, 83065: 76; Ye G.D. 1301: 67, 1690: 88; Yi G.P. 76002: 60; Yieh P.C. 345: 30, 495: 44, 540: 3; Yin S.F. 14: 41; Yin W.Q. 60- 1079: 57, 60 1314: 13, 2120: 87; Yip H.G. 115: 88, 349: 44; Yu P.H. 251: 83 (type), 1060: 83; Yu S.L. 61528: 3, 700469: 6, 700528: 76, 900086: 3; Yu S.S. 500124: 70; Yu T.Q. 11212: 81; Yu TT. 250: 52, 517: 30, 1429: 26, 3067: 45, 3105: 21 (type), 3122: 21, 15187: 26, 15854: 26, 15854: 32 17374: 26 (type) 17395: 89, 17631: 60, 17894: 60, 18055: 57, 19251: 32, 19938: 31, 20016: 40, 20393: 40, 20832: 32, 20866: 57, 20894: 31, 21016: 40, 21025: 14, 22096: 32; Yuan J.M. 332: 38a, 373: 82, 621: 76, 692: 82; Yuan S.F. 5030: 76, 5123: 67; Yue Z.S. 86.107: 75. Zhang Z.S. 1395: 13; Zhang, B.N. 406132: 70; Zhao R.F. 54: 76, 197: 76, 604360: P т 33; Zhao W.L. 6: 70; Zheng. J.C. s.n.: 45 (type), s.n.: 45; Zheng P. 7: 88, 22: 7, 122: 86, 10400 86a, 10838: 7, 11761: 84, 12506: 23, 12677: 38a, 13116: 81, 13118: 9, 13419: 79, 13465: 77. 13531: 77, 13563: 9, 13639: 70 (type), 13677: 38a, 13871: 6, PEA "= 14050: 38a, 14503: 38a; Zhou R.Z. : 79, 7081: 20, 7092: 82; Zhu D.Q. 5: = Zhu TP. 554: 67; Zimmermann R. 29: 56, 175: 55. Volume 80, Number 4, pp. 787-1108 of the ANNALS OF THE MISSOURI BOTANICAL GARDEN was published on November 17, 1993 ANNALS OF THE MISSOURI BOTANICAL GARDEN: CHECKLIST FOR AUTHORS 0 0 ad О О E О OO О о о о о 1. General Instructions Text is in English or Spanish on numbered pages. Manuscript is typed on one side of nonglossy 8/2 x 11 in. paper. At least 1 in. is left as margin all around, except on the first page, which has 3 in. left blank at the top. Text is double- or triple-spaced throughout, including abstract, notes, legends, tables, specimen lists, Liter- ature Cited, and footnotes Right-hand margin is not кайа and words are not broken there. Words to be set in italics are underlined instead of italicized. Boldface or other special typefaces are not used. [Editor will add printer's marks for these.] Com- mon Latin ages or 3 posee are not italicized (e.g., et al., i.e., sensu Only names at da sank of genus and below are ital- icized Correct accents, umlauts, and other diacritical marks have been included. All figures and tables are cited in the text. 2. Style Recent issue of Annals is used as model. Chicago Manual of Style, latest edition, is used as reference. 3. First Page Footnotes are typed as double-spaced paragraphs on the first page. The Pasi оное contains оса ments, 1 , her- baria that loaned specimens, and the name of the artist. The second footnote is the author's address. Addresses for additional authors are each separate footnotes. No other footnotes occur, except in tables, where un- avoidable 4. Abstract A one-paragraph abstract precedes text. Papers in d have an English abstract in addition to a Span- ish resumen. perio is concise (1 paragraph) and includes brief statements about the paper's intent, materials and methods, results, йу significance of findings. 5. Tables Tables are neat, double-spaced, and easily understood rather than long and complex. Tables do not contain vertical or horizontal lines. [Ed- itor and/or printer will add them as needed.] Captions are typed double-spaced as paragraphs at the tops of the ta Each table starts on a separate sheet, double-spaced. О О 6. Abbreviations eriods are used after all abbreviations (which are minimized) except metric measures, compass direc- tions, and herbarium designations. When dates are given as part of collection information, three-letter month abbreviations are used, except for months with four letters, which are spelled out in full. States are not abbreviated, and cities are spelled out. [51. is acceptable. Periodicals are abbreviated according to B-P-H (Bo- tanico-Periodicum-Huntianum) and to B-P-H/S (Во- ге o-Periodicum-Huntianum/ Supplementum). uthors' names are abbreviated rus to Authors i Plant Names or spelled out in Book titles are abbreviated ети to Тахопотис Literature, edition 2, but with initial letters capitalized. the formal citation at the head of a taxonomic treat- ment, are spelled o Herbaria are net according to Index Herba- riorum, edition 7. Taxonomic Treatment One paragraph per basionym is A as s follows: Taxon author, literature citation, type c e.g., Pleu- Sau ые amplifolium (Mez) Rohwer, Mitt Inst. Allg. urg 20: 43. 1986. Nectandra a Mez, ы АП Konigl. Bot. Gart. Breslau 1: TYPE: Brazil. Rio de Janeiro Alto sans p e 17731 (holotype, B; isotypes, B, С, K, NY, P). Figure 11 Lectotype designations are included together with an indication of where th Exclamation points a for specimens ex- amined. Types not seen are indicated as such (e.g., MO, US not seen). A brief Latin diagnosis for each new taxon is provided rather than a complete Latin description O When relevant, nomina nuda, misapplied names and О superfluous names are included in the discussion fol- lowing the description, but are not part of the formal synonym 8. Specimens examined If many specimens were examined, those cited in the text are limited to one collection per geographical area. O Long lists of specimens examined are placed at the Чч end of the paper, following the Literature Cited. They are arranged alphabetically by collector, followed by collection number, followed by the number of the taxon in the text. Specimens are cited in the text as follows: Additional specimens examined (or Selected specimens exam- ined or Paratypes). MEXICO. Oaxaca: Sierra 5 Pedro Nolesco, Talea, 3 Feb. 1987 (fl), de 865 (BM, G, K, US). [Dates and reproductive status are optional.] Countries are run together in the same О О О о оо О О О О О O МА M u pore e.g., COUNTRY А. Major political divi- . COUNTRY B. Major political division: 9. Vouchers and Genetic Sequences If paper presents original data, it includes citation of vouchers. [Depending on the type of paper, reference to the original wild source may a require Herbarium vouchers state the i and wines. herbarium in which the voucher is located, and a clear annotation that the mu represents the voucher or the study in question Genetic виа es › have ees banked, and accession numbers are provided. uthor accepts iesus for establishing the ac- curacy of information provided. 10. Keys Keys are clear and have been checked carefully for consistency with the descriptions. 11. Literature Cited The Literature Cited contains all references cited in the text. All entries in the Literature Cited are cited in the text. Spelling of author(s) name(s) and years of publication have been double-checke All entries have been verified against original sources, especially journal ie accents, diacritical marks, and spelling in la он other than Mod Periodicals are listed as follows: author's last ата E Year. Full title of iude pru. abbreviated В-Р- H/S. Volume: pages numbers after as in o parenthetical part volume uniber is given unless each par у. Ког тоге пе iiri this style is followed: au- thor's last name, initial(s), second author's initial(s), Books appear as follows: Year. F p Оне аа "Title (edited by Edit ‚ Publisher, City of Publicatio СЕ “in prep., Ks). , Зга ode theses and erences to inaccessible о a minimum. are not included in the Literature Citec They 12. Illustrations Author(s) name(s) and figure number(s) are written in o il on the back of each figure or plate, on both originals and review copies 'ale bars appear on а. photographs, and Magnifico. reductions are not indicated in cap- Hon trat sal tad ff cardboard no thicker than %4 in. Illustrations are presized to fit either column width (25$ in. or ca. 68 mm) or full page width (5% in. 140 mm), or illustrations no larger (= 29 x 40 cm). ee axl tions is 5/44 х 8% in or ca. than 11 x 16 in. mum size for printed illustra- [J Figures are numbered in Arabic numerals in the order of their citation in the text. Parts of figures are labeled with capital letters. Photographs are crisp black-and-white prints. Figures are grouped into composite plates when pos- sible; че of photographs are abutted. No stripping is inserted between plate or figure seg- ments. [Printer will insert stripping. Edges of figures are squared. Maps include reference to latitude and longitude and are рош by a fine border. micrographs are free of conspicuous charg g- Axes on graphs are all labele Captions pna all explanatory text. No text appears on the figures. Captions are separate from other text, one parseraph for each group of figures, and following the style in current issues of the Annals. Symbols on maps are le taken into consideration. OO " OO о ud E gible and reduction has been 13. What to submit C] Full mailing address, telephone, fax, and electronic mail numbers are submitted in cover letter. Three e a ed. O Copies/photocopies of original figures are enclosed with i Copies of half-tone figures (photo- graphs) and plates must be of sufficiently good quality to show all the detail of the originals. [Send originals when manuscript is accept- | ( = the ar s long, an MS-DOS 3'%- 5% in. diskette, as WordStar [preferable] or трон бе, is submitted in addition to the printed docume O Draft-quality dot-matrix printout is not от SCOPE OF THE ANNALS Кој PELA || ial 4 Aem ! 1 I e I о 7 РА and related fields. Papers whose purpose is the establish- ment of new nomenclatural entities in vascular plants and consideration. (Novon instructions are available from the managing editor CONDITIONS FOR PUBLICATION Authors are urged to contact the managing editor prior to submitting manuscripts. Manuscripts sent already re- viewed and revised are generally subject to the full normal review process initiated here Manuscripts not properly prepared may be returned for revision prior r to review. It is the author’s spony to supply missing ы ic information; the editor does not lacis ary re- search except by prior arrangement with iue not having access to crucial reference materials. Com i cation with the editor at all points of the publication process is encourage PAGE CHARGES Page costs are $80 per page, although charges may be reduced under certain circumstances. Corrections in proof will be billed nonnegotiably to the author at $3 per line changed. CORRESPONDENCE Amy Scheuler, Managing Editor Tel: (314) 577-5112 Annals of the Fax: (314) 577-9594 Missouri Botanical Garden Electronic mail: Department scheuler@mobot.org Missouri Botanical Garden P.O. Box 299 St. Louis, MO 63166-0299, A ANNALS OF THE MISSOURI BOTANICAL GARDEN VOLUME 80 1993 Colophon This volume of the ANNALS of the Missouri Botanical Garden has been set in APS Bodoni. 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 70s Vintage Gloss. This is an acid-free paper designed to have a shelf-life of over 100 years. Vintage Gloss is manufactured by the Potlatch Paper Company. Photographs used in the ANNALS are reproduced using 300 line screen halftones. The binding 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 1993 ISSN 0026-6493 Volume 80 ALBERT, VICTOR A., Mark W. CHase & Brent D. Мазнтек. Character- state Weighting for Cladistic Analysis of Protein-coding DNA Sequences ALBERT, VICTOR A. (See Mark W. Chase et al.) ALBERT PUENTES, DELHY, JESUS HERNANDEZ DE ARMAS & ANTONIO LOPEZ ALMIRALL. Fenología y Estructura Floral de Trichilia havanensis Jacq. (Meliaceae) ... ARMESTO, J. J. & P. E. VipiELLA..— Plant Life-forms and Biogeographic Relations of the Flora of Lagunillas (30%) in the Fog-free Pacific Coastal Desert BARKLEY, THEODORE. Synthesis: A Historical Perspective ...... | BARKLEY, THEODORE. (See Nancy Morin & Theodore M. Barkley) ................ BARRETT, SPENCER C. H. (See Mark W. Chase et al.) . BERNAL, CARMEN. (See Guadalupe Palomino, Patricia Martinez, Carmen Bernal & Mario Sousa 5.) __ И Bremer, BIRGITTA. (See Richard С. Olmstead, Birgitta Bremer, Kathy M. Scott & Jeffrey D. Palmer) | Chase, Mark W., DoucLas E. Sorris, RICHARD С. OLMSTEAD, DAvip MORGAN, DoNarp H. Les, BRENT D. Мазнтек, MELVIN R. DuvaLL, ROBERT A. Price, Накогр С. Hills, Yin-LonG Qiu, KATHLEEN A. KRON, JEFFREY Н. Rerric, ELENA Conti, JEFFREY D. PALMER, JAMES R. MANHART, KENNETH J. Syrsma, HELEN J. MicuakErs, W. Jonn KRESS, KENNETH G. KaroL, W. DENNIS CLARK, MIKAEL HEDREN, BRANDON S. GAUT, ROBERT K. JANSEN, Ki-Joonc Kim, CHARLES К. WIMPEE, JAMES F. SMITH, GLENN R. FURNIER, STEVEN H. STRAUSS, Qiu- YUN XIANG, GREGORY M. PLUNKETT, PAMELA S. SOLTIS, SUSAN P. SWENSEN, STEPHEN E. WILLIAMS, PAUL A. GADEK, CHRISTOPHER J. QUINN, Luis E. EGUIARTE, EDWARD GOLENBERG, GERALD H. LEARN, JR., SEAN W. GRAHAM, SPENCER С. H. BARRETT, SELVADURAI DAYANANDAN & VICTOR A. ALBERT. Phylogenetics of Seed Plants: An Analysis of Nucleotide Sequences from the Plastid Gene rbcL Chase, Mark W. (See Douglas E. Soltis, Mark W. Chase & Richard G. A AS engi Chase, Mark W. (See Kathleen A. Kron & Mark W. Chase) ........................ Chase, Mark, W. (See Victor A. Albert, Mark W. Chase & Brent D. Mishler) eee Chase, Mark W. (See Yin-Long Qiu, Mark W. Chase, Donald Н. Les & Clifford R. Parks) . Chase, Mark W. (See Melvin R. Duvall et al.) . Снем Bao Liang € Hans P. Моотевоом. Notes on Magnoliaceae Ш: The Magnoliaceae of China ли CHUANG, T. I. & L. К. Heckarp. Chromosome Numbers of Neotropical Castilleja (Scrophulariaceae: Tribe Pediculareae) and their Taxonomic Implications .. 1993 CLARK, W. DENNIS, BRANDON S. Gaur, MELVIN R. DuvaLL & MICHAEL T. CLEGG. Phylogenetic Relationships of the Bromeliiflorae- Commelini- florae-Zingiberiflorae Complex of Monocots Based on rbcL Sequence Comparisons CLARK, W. Dennis. (See Mark W. Chase et al.) CLARK, W. DENNIS. (See Melvin R. Duvall et al.) CLEGG, MICHAEL T. (See Melvin R. Duvall et al.) CLEGG, MICHAEL Т. (See Gerard Zurawski & Michael T. Clegg) ................... CLEGG, MicHAEL Т. (See W. Dennis Clark, Brandon S. Саш, Melvin R. Duvall & Michael T. Clegg) COELLO, GERARDO, ANA ESCALANTE & JORGE SOBERON. Lack of Genetic Variation in Lacandonia schismatica (Lacandoniaceae: Triuridales) in its Only Known Locality Connor, Н. E. € M. I. Dawson. Evolution of Reproduction in Lamprothyr- sus (Arundineae: Gramineae) Conti, ELENA, ANTHONY FISCHBACH & KENNETH J. Syrsma. Tribal Rela- tionships in Onagraceae: Implications from rbcL Sequence Data ____. Conti, ELENA. (See James Rodman, Robert A. Price, Kenneth Karol, Elena Conti, Kenneth J. Sytsma & Jeffrey D. Palmer) CoNTI, ELENA. (See Mark W. Chase et al.) DAYANANDAN, SELVADURAI. (See Mark W. Chase et al.) Dawson, M. I. (See Н. E. Connor & M. I. Dawson) DuvaLL, MELVIN R., MICHAEL T. CLecc, MARK W. Сназе, W. DENNIS CLARK, W. Jonn Kress, Hanorp С. Нил5, Luis E. EGUIARTE, JAMES F. SMITH, BRANDON S. GAUT, ELIZABETH A. ZIMMER & GERALD H. LEARN, JR. Phylogenetic Hypotheses for the Monocotyledons Constructed from rbcL Sequence Data DuvaLL, MELVIN R. (See Mark W. Chase et al.) DuvaLL, MELVIN R. (See W. Dennis Clark, Brandon S. Gaut, Melvin В. Duvall & Michael T. Clegg) EGUIARTE, Luis E. (See Mark W. Chase et al.) EGUIARTE, Luis E. (See Melvin R. Duvall et al.) ELLis, Rocer P. (See Fernando O. Zuloaga, Roger P. Ellis & Osvaldo orrone) ENGLEMAN, E. M. (See J. Márquez-Guzmán, S. Vázquez-Santana, E. M. Engleman, A. Martinez-Mena & E. Martínez) ERIKSEN, BENTE. A Revision of Monnina Subg. Pterocarya (Polygalaceae) in Northwestern South America ERIKSSON, ROGER. The Rise and Fall of Pseudoludovia andreana (Cyclan- thaceae) ESCALANTE, ANA. (See Gerardo Coello, Ana Escalante & Jorge Soberón) ... Estes, JAMES, R. The National Science Foundation: Financial, Intellectual, and Knowledge Broker for Systematic Botany 987 528 607 607 523 987 898 512 672 686 528 528 212 EZCURRA, CECILIA. Systematics of Ruellia (Acanthaceae) in Southern South America FapEN, RoBERT В. The Misconstrued and Rare Species of Commelina (Com- melinaceae) in the Eastern United States Fanen, Ropert B. Tradescantia crassifolia (Commelinaceae), ап Over- looked Species in the Southwestern United States FISCHBACH, ANTHONY. (See Elena Conti, Anthony Fischbach & Kenneth J. Sytsma) FREIRE, SUSANA E. A Revision of Chionolaena (Compositae, Gnaphalieae) FURNIER, GLENN R. (See Mark W. Chase et al.) GADEK, PAUL A. & CHRISTOPHER J. QUINN. An Analysis of Relationships within the Cupressaceae Sensu Stricto Based on rbcL Sequences .......... GADEK, PAUL A. (See Mark W. Chase et al.) Слот, BRANDON, S. (See Mark W. Chase et al.) Саџт, BRANDON $. (See Melvin R. Duvall et al.) Gaur, BRANDON S. (See W. Dennis Clark, Brandon S. Gaut, Melvin R. Duvall & Michael T. Clegg) GipBs, PETER E. (See Darién E. Prado & Peter E. Gibbs) GorpBLATT, PETER € MASAHIRO TAKEI. Chromosome Cytology of the Af- rican Genus Lapeirousia (Iridaceae—Ixioideae) GorpBLATT, PETER, MASAHIRO Таке & Z. A. RazzaQ. Chromosome Cy- tology in Tropical African Gladiolus (Iridaceae) GOLENBERG, EDWARD. (See Mark W. Chase et al.) Gomon, JANET. (See Nancy R. Morin & Janet Gomon) . GRAHAM, SEAN W. (See Mark W. Chase et al.) Heckarp, L. R. (See T. I. Chuang & L. R. Heckard) . HEDREN, MIKAEL. (See Mark W. Chase et al.) . HERNANDEZ DE ARMAS, Jesus. (See Delhy Albert Puentes, Jesus Hernandez de Armas & Antonio López Almirall) Hersukovitz, MARK А. Leaf Morphology of Calandrinia and Montiopsis (Portulacaceae) | Hersukovitz, Mark A. Revised Circumscriptions and Subgeneric Taxon- omies of Calandrinia and Montiopsis (Portulacaceae) with Notes on Phylogeny of the Portulacaceous Alliance . Нил5, Накогр С. (See Mark W. Chase et al.) Нил5, Накогр С. (See Melvin R. Duvall et al.) JANSEN, ROBERT K. (See Helen J. Michaels, Kathy M. Scott, Richard G. Olmstead, Tim Szaro, Robert К. Jansen & Jeffrey D. Palmer) ........ JANSEN, ROBERT K. (See Mark W. Chase et al.) .. Jupziewicz, EMMET J. (See Fernando O. Zuloaga, Osvaldo Morrone & Emmet J. Judziewicz) рш. | KAROL, KENNETH. (See James Rodman, Robert A. Price, Kenneth Karol, Elena Conti, Kenneth Sytsma & Jeffrey D. Palmer) KAROL, KENNETH С. (See Mark W. Chase et al.) Kim, Ki-Joonc. (See Mark W. Chase et al.) Kress, W. Jonn. (See James F. Smith, W. John Kress & Elizabeth A. Zimmer) Kress, W. JOHN. (See Mark W. Chase et al.) Kress, W. Jonn. (See Melvin R. Duvall et al.) KRON, KATHLEEN A. & MARK W. CHASE. Systematics of the Ericaceae, Empetraceae, Epacridaceae and Related Taxa Based Upon rbcL Se- quence Data Кком, KATHLEEN А. (See Mark W. Chase et al.) LEARN, GERALD H., JR. (See Mark W. Chase et al.) LEARN, GERALD H., JR. (See Melvin R. Duvall et al.) Les, Поматр H. (See Yin-Long Qiu, Mark W. Chase, Donald H. Les & Clifford R. Parks) Les, DonaLD H. (See Mark W. Chase et al.) LOPEZ ALMIRALL, ANTONIO. (See Delhy Albert Puentes, Jesús Hernandez de Armas & Antonio López Almirall) LUCHSINGER, ARLENE E. The Future of Scientific Communications: More Questions Than Answers MANHART, JAMES R. (See Mark W. Chase et al.) MARQUEZ-GUZMAN, J., S. VAZQUEZ-SANTANA, E. M. ENGLEMAN, А. MAR- TINEZ-MENA & E. MARTINEZ. Pollen Development and Fertilization in Lacandonia schismatica (Lacandoniaceae) MARTINEZ, E. (See J. Márquez-Guzmán, S. Vazquez-Santana, E. M. Engle- man, А. Martinez-Mena & E. Martinez) MARTINEZ, PATRICIA. (See Guadalupe Palomino, Patricia Martinez, Carmen Bernal & Mario Sousa S.) MARTINEZ-MENa, А. (See J. Marquez-Guzman, S. Vázquez-Santana, E. M. Engleman, A. Martinez-Mena & E. Martinez) MENDENHALL, MEGHAN С. (See Billie L. Turner & Meghan G. Mendenhall) MicHAELS, HELEN J., Катну M. Scorr, RICHARD С. OLMSTEAD, TIM SZARO, ROBERT K. JANSEN & JEFFREY D. PALMER. Interfamilial Relationships of the Asteraceae: Insights from rbcL Sequence Variation ........................ MICHAELS, HELEN J. (See Mark W. Chase et al.) MILLAR, CONSTANCE I. Impact of the Eocene on the Evolution of Pinus L MisHLER, BRENT D. (See Mark W. Chase et al.) MisHLER, BnENT D. (See Victor A. Albert, Mark W. Chase & Brent D. Mishler) MORAIN, STANLEY A. Emerging Technology for Biological Data Collection and Analysis Moran, Rossin C. (See Noriaki Murakami & Robbin C. Moran) „u Morcan, Davin R. & DoucLas E. $01715. Phylogenetic Relationships Among Members of Saxifragaceae Sensu Lato Based on rbcL Sequence Data Morcan, Davip R. (See Mark W. Chase et al.) Morcan, Davip R. (See Qiu- Yun Xiang, Douglas E. Soltis, David R. Morgan & Pamela S. Soltis) Morin, Nancy & THEORORE M. BARKLEY. Introduction to the Symposium. Knowledge Brokering: The Mechanics of Synthesis, the 38th Annual Systematics Symposium of the Missouri Botanical Garden иШ... Morin, Nancy R. & JANET Gomon. Data Banking and the Role of Natural History Collections MonRONE, OsvALDO. (See Fernando O. Zuloaga, Osvaldo Morrone & Emmet J. Judziewicz) Morrone, OsvaLDO. (See Fernando O. Zuloaga, Roger P. Ellis & Osvaldo Morrone) Murakami, Noriaki & Rossin C. Moran. Monograph of the Neotropical Species of Asplenium Sect. Hymenoasplenium (Aspleniaceae) ............... Nasu, JEAN E. (Ownership and Outreach: A Model for Administration of Shared Data Моотевоом, Hans P. (See Chen Bao Liang & Hans P. Nooteboom) ........... OLMSTEAD, RICHARD G., BIRGITTA BREMER, Катну M. Scorr & ЈЕЕЕКЕҮ D. MER. A Parsimony Analysis of the Asteridae Sensu Lato Based on rbcL Sequences OLMSTEAD, RICHARD G. (See Douglas E. Soltis, Mark W. Chase & Richard G. Olmstead) OLMSTEAD, RICHARD С. (See Helen J. Michaels, Kathy M. Scott, Richard С. Olmstead, Tim Szaro, Robert К. Jansen & Jeffrey D. Palmer) ....... OLMSTEAD, RICHARD С. (See Mark W. Chase et al.) PALMER, JEFFREY D. (See Helen J. Michaels, Kathy M. Scott, Richard С. Olmstead, Tim Szaro, Robert К. Jansen & Jeffrey D. Palmer) ............... PALMER, JEFFREY D. (See James Rodman, Robert A. Price, Kenneth Karol, Elena Conti, Kenneth J. Sytsma & Jeffrey D. Palmer) PALMER, JEFFREY D. (See Mark W. Chase et al.) PALMER, JEFFREY D. (See Richard С. Olmstead, Birgitta Bremer, Kathy M. Scott & Jeffrey D. Palmer) PALMER, JEFFREY D. (See Robert A. Price & Jeffrey D. Palmer) .................... PALOMINO, GUADALUPE, PATRICIA MARTINEZ, CARMEN BERNAL & MARIO SOUSA S. Diferencias Cromosómicas entre Algunas Especies de los Géneros Sophora L. y Styphnolobium Schott Parks, CLIFFORD R. (See Yin-Long Qiu, Mark W. Chase, Donald Н. Les & Clifford R. Parks) PLUNKETT, GREGORY M. (See Mark W. Chase et al.) PRADO, DARIEN E. & PETER E. GiBBs. Patterns of Species Distributions in the Dry Seasonal Forests of South America Price, RoBERT А. & JEFFREY D. PALMER. Phylogenetic Relationships of the eraniaceae and Geraniales from rbcL Sequence Comparisons ................ PRICE, ROBERT A. (See James Rodman, Robert A. Price, Kenneth Karol, Elena Conti, Kenneth J. Sytsma & Jeffrey D. Palmer) PRICE, ROBERT A. (See Mark W. Chase et al.) Qiu, Yin-Lonc, Mark W. Сназе, Donatp H. Les € Cuirrorp R. Parks. Molecular Phylogenetics of the Magnoliidae: Cladistic Analyses of Nucleotide Sequences of the Plastid Gene rbcL Qiu, YiN-LoNc. (See Mark W. Chase et al.) QUINN, CHRISTOPHER J. (See Paul A. Gadek & Christopher J. Quinn) ......... QUINN, CHRISTOPHER J. (See Mark W. Chase et al.) RazzaQ, Z. А. (See Peter Goldblatt, Masahiro Takei & Z. A. Razzaq) 2. RETTIG, JEFFREY H. (See Mark W. Chase et al.) RopMan, JAMES, ROBERT A. PRICE, KENNETH KAROL, ELENA CONTI, KENNETH J. SytsMA & JEFFREY D. PALMER. Nucleotide Sequences of the rbcL Gene Indicate Monophyly of Mustard Oil Plants Корр, VELvA E. (See Mario Sousa S. & Velva E. Rudd) RuEDA, Ricarpo M. The Genus Clerodendrum (Verbenaceae) in Mesoamer- ica Scott, Катну M. (See Helen J. Michaels, Kathy M. Scott, Richard G. Olmstead, Tim Szaro, Robert К. Jansen & Jeffrey D. Palmer) ______ Scorr, Катну M. (See Richard C. Olmstead, Birgitta Bremer, Kathy M. Scott & Jeffrey D. Palmer) SENDULSKY, TATIANA. First Report of Ballistochory in the Poaceae ............... SMITH, JAMES F., W. JouN Kress & ELIZABETH A. ZIMMER. Phylogenetic Analysis of the Zingiberales Based on rbcL Sequences SMITH, JAMES F. (See Mark W. Chase et al.) SMITH, JAMES F. (See Melvin R. Duvall et al.) SOBERON, JORGE. (See Gerardo Coello, Ana Escalante € Jorge Soberón) ... SoLTIS, DOUGLAS E., MARK W. CHASE & RICHARD С. OLMSTEAD. Introduction, rbcL Sequence Data and Phylogenetic Reconstruction in Seed Plants ... SoLTIS, DoucLas E. (See David R. Morgan & Douglas E. Soltis) ................... SoLTIS, DoucLas E. (See Mark W. Chase et al.) SoLTIS, DoucrAs E. (See Qiu- Yin Xiang, Douglas E. Soltis, David R. Morgan & Pamela S. Soltis) SOLTIS, PAMELA S. (See Mark W. Chase et al.) SOLTIS, PAMELA 5. (See Qiu- Yun Xiang, Douglas E. Soltis, David R. Morgan & Pamela S. Soltis) 528 902 Sousa S., Mario. (See Guadalupe Palomino, Patricia Martinez, Carmen Bernal & Mario Sousa S.) Sousa S., Mario. El Género Inga (Leguminosae: Mimosoideae) del Sur de México y Centroamérica, Estudio Previo para la Flora Mesoamericana Sousa S., Mario € VeLva E. Rupp. Revisión del género Styphnolobium (Leguminosae: Papillionoideae: Sophoreae STRAUSS, STEVEN H. (See Mark W. Chase et al.) SWENSEN, Susan M. (See Mark W. Chase et al.) SYTSMA, KENNETH J. (See Elena Conti, Anthony Fischbach & Kenneth J. ytsma SYTSMA, KENNETH J. (See James Rodman, Robert A. Price, Kenneth Karol, Elena Conti, Kenneth J. Sytsma & Jeffrey D. Palmer) SYTSMA, KENNETH J. (See Mark W. Chase et al.) Szaro, Tim. (See Helen J. Michaels, Kathy M. Scott, Richard С. Olmstead, Tim Szaro, Robert K. Jansen & Jeffrey D. Palmer) TAKEI, MASAHIRO. (See Peter Goldblatt & Masahiro Takei) TAKEI, MASAHIRO. (See Peter Goldblatt, Masahiro Takei & Z. A. Razzaq) TURNER, BILLIE L. & MEGHAN С. MENDENHALL. A Revision of Malvaviscus (Malvaceae) VAN DER WERFF, HENK. A Revision of the Genus Pleurothyrium (Lauraceae) VAZQUEZ-SANTANA, S. (See J. Marquez-Guzman, S. Vázquez-Santana, E. M. ngleman, A. Martinez-Mena & E. Martinez VIDIELLA, P. E. (See J. J. Armesto & P. E. Vidiella) WILLIAMS, STEPHEN E. (See Mark W. Chase et al.) WiMPEE, CHARLES F. (See Mark W. Chase et al.) Worr, Jan Н. D. Diversity Patterns and Biomass of Epiphytic Bryophytes and Lichens Along an Altitudinal Gradient in the Northern Andes ......... XIANG, Qiu-YuN, DoucLas E. SoLris, Davin R. Morcan & PAMELA S. Sorri. Phylogenetic Relationships of Cornus L. Sensu Lato and Pu- tative Relatives Inferred from rbcL Sequence Data XIANG, Qiu-YuN. (бее Mark W. Chase et al.) ZIMMER, ELIZABETH A. (See James К. Smith, W. John Kress € Elizabeth A. Zimmer) ZIMMER, ELIZABETH A. (See Melvin R. Duvall et al.) ZULOAGA, FERNANDO O., OsvaLDO MORRONE & EMMET J. JUDZIE- wicz. Endemic Herbaceous Bamboo Genera of Cuba (Poaceae: Bam- busoideae: Olyreae) 846 ZULOAGA, FERNANDO O., ROGER P. ELLIS & OsvALDO Morrone. A Revision of Panicum Subg. Dichanthelium Sect. Dichanthelium (Poaceae: Pan- icoideae: Paniceae) in Mesoamerica, the West Indies, and South America ZURAWSKI, GERARD & MICHAEL T. CLEGG. — Foreward, rbcL Sequence Data and Phylogenetic Reconstruction in Seed Plants 119 923 STATISTICAL SUMMARY OF SOME OF THE ACTIVITIES IN THE MISSOURI BOTANICAL GARDEN HERBARIUM, 1992 1992 Vascular Bryophyte Total Acquisition of specimens Staff Collections (includes bryophytes) 62,547 == 62,547 Purchase 2.157 535 2,692 Exchange 50,141 2,380 52,521 Gifts 12,503 1,671 14,174 Total acquisitions 127,348 4,586 131,934 Mountings Newly mounted 169,289 16,100 185,389 Mounted when received 1,633 0 1,633 Total specimens filed 170,922 16,100 187,022 Repairs Specimens repaired 21,227 21.227 Specimens stamped 2,728 2,728 Total repairs 23,955 23,955 Specimens sent On exchange 51,906 5,266 57,172 As gifts 13,277 799 14,076 Тота! 65,183 6,065 71,248 Loans sent Total transactions 427 39 466 Total specimens 27,943 4,251 32,194 To U.S. institutions Transactions 291 23 254 Specimens 14,612 378 14,990 To foreign institutions Transactions 196 16 212 Specimens 13,331 3,873 17,204 To student investigators Transactions 51 3 54 Specimens 6,823 37 6,860 To professional investigators Transactions 376 36 412 Specimens 22,134 2,600 25,334 Loans received Transactions 355 17 372 Specimens 20,485 2,119 22,604 1992 From U.S. From abroad Total Visitors 409 126 535 n 31 December 1992 the total number of mounted, accessioned specimens in the herbarium was 4,112,057 (3, = 146 vascular plants and 219,911 bryophytes). Recent Issues of the Annals of the Missouri Botanical Garden Phylogeny of Asteridae | Annals of the Missouri Botanical Garden 79(2) (Spring 1992). The monophyly, sister group (or groups), and patterns of relationships of the Asteridae are discussed in the symposium "Phylogeny of Asteridae," organized by Larry Hufford of the University of Minnesota. The symposium was jointly sponsored by the American Society of Plant Taxonomists and the Botanical Society of America and was held at their annual meeting, 7 Aug. 1990, in Richmond, Virginia. 238 pages. $27.00, plus postage. Knowledge Brokering: The Mechanics of Synthesis Annals of the Missouri Botanical Garden 80(2) (Spring 1993). This symposium, the 38th Annual Systematics Symposium of the Missouri Botanical Garden (held 4—5 Oct. 1991, in St. Louis, Missouri), discusses the value and future necessity of assimilating diverse biological information into new and broader syntheses, taking advantage of computer-based data-management nmt 41 pages (entire issue 230 pages). $35.00, plus postage. rbcL Sequence Data and Phylogenetic Reconstruction in Seed Plants Annals of the Missouri Botanical Garden 80(3) (Summer 1993). Fourteen papers plus an appendix containing voucher and sequence data. This collection of papers, organized by Douglas E. Soltis and Mark W. Chase, presents the first in-depth molecular phylogenetic analysis of seed plants in general, and angiosperms in particular. Investigators throughout the world involved in rbcL sequencing studies contributed their unpublished sequences for a broad phylogenetic analysis. 235 pages. $40.00, plus postage. "То place an order, send check or plc order in U.S. funds, payable through a U.S. bank; ка orders also eem U.S. shipments: add $2.00 for one book and $.75 for each additional book; non-U.S. shipments: add $3.00 lor one book, and $.75 for each pe ве book. Orders should be prepaid; a $1.00 fee will be added to orders requiring invoices. No shipments are made until payment is received. mae or fax (for ufi orders) form with your check or money order, payable to Missouri Botanical Garden, to: Department E Eleven, Missouri Botanical Garden i Please send - ————— сору(іеѕ) of Annals 79(2) аө ——— copy(ies) of Annals 80(2) copy(ies) of Annals 80(3) - Louis, о 63166- 0299, 0 S.A. “мө (314) 577-9594 Phone: (314) 577. 9534 Send book(s) to: О Payment enclosed. О Send invoice ($1.00 fee will be | Name | added to total). | О Mastercard О Visa Address Name as it appears on card ; Postal Code Country ый) Card # | У Exp. date Phone number (days) 7 Prices ARE SUBJECT To CHANGE Without Notice CONTENTS Systematics of Ruellia (Acanthaceae) in Southern South America Cecilia Ezcurra ..... Endemic Herbaceous Bamboo Genera of Cuba (Poaceae: Bambusoideae: Olyreae) Fer- nando O. Zuloaga, Osvaldo Morrone & Emmet J. Judziewicz Fenología y Estructura Floral de Trichilia havanensis Jacq. (Meliaceae) Delhy Albert Puentes, Jesás Hernandez de Armas & Antonio López Almirall i The Genus Clerodendrum (Verbenaceae) in Mesoamerica Ricardo M. Rueda ............... Pollen Development and Fertilization in Lacandonia schismatica (Lacandoniaceae) J. Маг- quez-Guzmán, S. Vázquez-Santana, E. M. Engleman, A. Martínez-Mena & E. Martínez TEAMS “Lack of Genetic Variation in Lacandonia schismatica (Lacandoniaceae: Triuridales) in its Only Known Locality Gerardo Coello, Ana Escalante € Jorge Ѕођегбп Patterns of Species Distributions in the Dry Seasonal Forests of South America Darien E. Prado & Peter E. Gibbs LLL MÀ Diversity Patterns and Biomass of Epiphytic Bryophytes and Lichens Along an Altitudinal Gradient in the Northern Andes Jan H. D. Wolf индек AES Сар Cytology of the African Genus Lapeirousia (Iridaceae—Ixioideae) 6 | “Goldblatt & Masahiro Takei e -. Chromosome Numbers of Neotropical Castilleja (Scrophulariaceae: Tribe Deda) and their Taxonomic Implications T. І Chuang & 1. К. Heckard == см | : Phylogenetic Relationships of the Bromeliiflorae- Commeliniflorae- Zingiberiflorae Complex - © of Monocots Based on rbcL Sequence Comparisons W. Dennis Clark, Brandon S. Gaut, Melvin R. Duvall & Michael Т. Clegg ci ie Notes on Magnoliaceae III: The Magnoliaceae of China Chen Bao Liang & Hans P. Nooteboom ad Cover illustration. Pleurothyrium giganthum van der Werff, by J. Myers. | PT eT PPS сео рак ee OE а а а а а а а о ЈАЊ са ва ас ли