JOURNAL OF THE ARNOLD ARBORETUM HARVARD UNIVERSITY VOLUME 67 NUMBER 1 ISSN 0004-2625 Journal of the Arnold Arboretum The Journal of the Arnold Arboretum (ISSN 0004-2625) is published quarterly in January, April, July, and October for $50.00 per year, plus $5.00 postage for addresses outside of the United States, by the Arnold Arboretum of Harvard University. It is printed and distributed by the Allen Press, Inc., 1041 New Hampshire Street, Law- rence, Kansas 66044. Second-class postage paid at Lawrence, Kansas. POSTMAS- TER: send address changes to Journal of the Arnold Arboretum, % Allen Press, Inc,, P. O. Box 368, Lawrence, Kansas 66044. Subscriptions and remittances should be sent to Journal of the Arnold Arboretum, 1041 New Hampshire Street, Lawrence, Kansas 66044, U. S. A. Claims will not be accepted after six months from the date of issue. EDITORIAL COMMITTEE S. A. Spongberg, Editor E. B. Schmidt, Managing Editor P. S. Ashton K. S. Bawa P. F. Stevens C. E. Wood, Jr. Printed at Allen Press, Inc., Lawrence, Kansas COVER: The stylized design appearing ■ by Karen Stoutsenberger. i the Journal and the offprints v JOURNAL ARNOLD ARBORETUM THE GENERA OF SPHENOCLEACEAE AND CAMPANULACEAE IN THE SOUTHEASTERN UNITED STATES1 Thomas J. Rosatti2 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 imbricate lobes that encase both’ flower buds and fruits (and thereby render ) Type species: S. ertner, 2 n = 24, and the West African endemic S. Pongatium DC. have 1986] ROSATTI, JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 i United States. The ( v Shaw, H. K. Sphenocleaceae. In: C. G. G. J. van Steenis, ed., FI. Males. I. 4: 2 1986] ROSATTI, SPHENOCLEACEAE AND CAMPANULACEAE 5 CAMPANULACEAE A. L. de Jussieu, Gen. PI. 163. 1789, nom. cons. (in 3’s), 6 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 ROSATTI, SPHENOCLEACEAE AND CAMPANULACEAE (including Donatia J. R. & G. Forster). Two noncampanulaceous members of the order are represented in the Southeast. Scaevola Plumieri (L.) Vahl (Goo- deniaceae) is native to southern Florida and other tropical and subtropical areas outside of the Southeast (see Brizicky), and Sphenoclea zeylanica Gaert- ner (Sphenocleaceae) is a species of the Old World tropics that is widely nat- uralized in warm regions of the Western Hemisphere, including most of the Coastal Plain of North America (see preceding treatment). Carolin (1978) sum- marized arguments for a saxifragalean origin of the Campanulales (Thome placed his Stylidiaceae near the Saxifragaceae), while Cronquist favored con- sideration of the Solanales or their close relatives. Several botanists have suggested a relationship between the Campanulaceae and the Goodeniaceae, primarily because of similarities in embryology and in the mechanism of pollen presentation. However, the embryological similarities are based on some rather widely distributed character states, and the families differ in certain other embryological features (Vijayaraghavan & Malik). The stylar structures involved in pollen collection (a cup, or indusium, in the Goo- deniaceae; hairs in the Campanulaceae) were thought by Carolin (1960) and Cronquist to be analogous. Furthermore, Carolin (1978) has argued that the inferior ovaries of the two groups are not homologous. In his opinion, in the Campanulaceae (where some genera have an extension of the floral tube above the ovary), epigyny resulted from the fusion of a preformed tube of sepals, petals, and stamens to the ovary, while in the Goodeniaceae (where there is no extension above the ovary and, in some genera, a progressive departure of sepals, petals, and stamens from the tube), epigyny resulted from the centrifugal adnation of these parts. The Goodeniaceae also differ from the Campanulaceae in their more complex cambial anatomy; in the absence of a latex system; in the presence of scler- enchymatous idioblasts (in leaf mesophyll, sometimes in stem cortex as well) and glandular trichomes (Metcalfe & Chalk, 1950; but some species of Lobelia have glandular trichomes on the calyx lobes); in their lack of nectaries (but see Carolin, 1959) and of haustoria in the endosperm (Vijayaraghavan & Malik); in the absence of cystoliths (Cronquist); in their pollen (see below); in their complete lack of ellagic acid and tannins (Cronquist); and in the presence of iridoids (Dahlgren et al., Jensen et al.). Jensen and colleagues discovered seco- iridoids in two genera of Goodeniaceae (including Scaevola ) and therefore concluded that the group should be recognized as an order distinct from the Campanulales and more closely related to the Gentianales, Dipsacales, and/or Comales. Vijayaraghavan & Malik also favored a unifamilial Goodeniales, primarily on embryological grounds, but considered it very close to the Cam- panulales. Some authors have proposed an origin of the Goodeniaceae within the Lobe- lioideae. However, although both groups have zygomorphic corollas with slits, the slit in the sympetalous corolla is made adaxial by resupination in the Lobelioideae, while in the Goodeniaceae it is adaxial and there is no resu- pination. Furthermore, the two-locular ovary of these groups is characteristic of a number of diverse families, and it has been interpreted by Carolin (1960), primarily on the basis of vascularization, to have been derived from two carpels JOURNAL OF [vol. 67 1986] ROSATTI, SPHENOCLEACEAE AND CAMPANULACEAE 9 ftllfg [vol. 67 AND CAMPANULACEAE 1 1 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 in the family has from two to 12 spinules and increasing numbers of pores as well (Dunbar, 1975b). In the 1975b). rmi 1986] ROSATTI, SPHENOCLEACEAE AND CAMPANULACEAE 1 3 :eae lies chiefly in its relatively S:£S5SSSSnTKe 1986] ROSATTI, SPHENOCLEACEAE AND CAMPANULACEAE 15 : ARNOLD ARBORETUM [vol. 67 1 986] ROSATTI, SPHENOCLEACEAE AND CAMPANULACEAE 1 7 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1986] ROSATTI, SPHENOCLEACEAE AND CAMPANULACEAE 19 1986] ROSATTI, SPHENOCLEACEAE AND CAMPANULACEAE 21 LETUM [vol. 67 591 pp. New York and noclea, 649-653.] Jr., & R. W. Schery, eds., FI. 1976. [Lobelia, 637-646; Sphe- A. Co JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1986] ROSATTI, SPHENOCLEACEAE AND CAMPANULACEAE 25 Candolle’s sect. Eucodon were also found to be of this cytotype. Boissier’s sections, on the other hand, were considered to be cytologically heterogeneous. Sugiura’s proposal of two groups differing in base chromosome number and leaf shape was regarded as untenable by Gadella, who (for reasons that do not appear to be completely justified) indicated instead that a classification based on a combination of characters, including chromosome length and base num- ber, basal leaf shape, and location of capsule dehiscence, seemed justified. Accordingly, Gadella established seven informal groups of species that differ in base chromosome numbers. Since a majority of the species of Campanula were not considered and the relationships involved were unclear, the arrange- The chromosome numbers characterizing some of Gadella’s groups have been determined subsequently in more distantly related species; in some cases the same number has been differently and independently derived as well. Never- theless, Contandriopoulos considered Gadella’s approach to be well founded and proposed a phylogeny based on cytology, morphology, life cycle, and geo- graphic distribution that reflected the exceedingly complex cytotaxonomy of Campanula and suggested how Gadella’s groups might be interrelated. The base chromosome numbers 5, 6, 7, 8, and 9 were considered to be primary; 8 was thought to be the ancestral number, perhaps evolved independently of 9 but more directly linked to the others. A secondary base number, 1 7, is char- acteristic of a great many species in the genus (and in the family as well— several genera have only species with this number). Contandriopoulos rejected the idea that x = 1 7 resulted from amphiploidy involving plants with either x = 7 and x = 10 or x = 8 and x = 9 and accepted as plausible the hypothesis of Bocher (1960), who considered 17 to have resulted from a single trisomy in the 2n = 16 cytotype (see, however, Sugiura; Gadella, 1963, 1964). Fedorov’s classification of Campanula is usually considered to be the most natural, apart from the questionable division of the genus into two sections based on the location of capsule dehiscence. Gadella (1964) regarded it as artificial only for those subsections between which members had been suc- cessfully hybridized (by him). Since only two of our species, both naturalized in the Southeast, are from the Old World, application of Fedorov’s system to the others is unclear. With the possible exception of C. Robinsiae Small, the species indigenous to our area are restricted to North America and are probably unrelated to any in the Old World. The most recent enumeration of the North American species of Campanula was provided by Shetler (1963). Of the 20 native species, 16 were indicated to be endemic to a portion of the continent, two were amphi-Atlantic, and two were circumpolar. Three European species (excluding C. persicifolia L., below) were considered to have escaped from cultivation and to have become natu- ralized in (unnamed parts of) North America. Three or four species (depending on taxonomic interpretation) of Campanula are indigenous in the Southeast, while the origin of C. Robinsiae has yet to be JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1986] ROSATTI, SPHENOCLEACEAE AND CAMPANULACEAE 27 : low elevations in t JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1986] ROSATTI, SPHENOCLEACEAE AND CAMPANULACEAE 29 occurring in the Southeast. Artificial hybrids between C. persicifolia and the closely related C. latiloba DC. were sterile (Gadella, 1967). The apparent absence from the Southeast of the extremely polymorphic and circumboreal Campanula rotundifolia L. is somewhat surprising, since it occurs throughout North America, with the most notable exception of the extreme Southwest and our area (Shetler, 1 979b). It grows in Missouri in moist, shaded, north-facing crevices of limestone in the southeastern Ozarks (Shannon Co.) within about 40 miles of Arkansas (Steyermark), as well as in western Texas and parts of Mexico (Shetler, 1979b). Pollen in Campanula is binucleate when shed (Brewbaker). It is generally spherical to suboblate (occasionally subprolate) and has three or four usually equatorial pores. Usually the surface bears spines that vary in size (particularly between species) and are superimposed on a fine reticulum. It is notable that pollen of C. americana has 12 pores distributed more or less equally over the entire surface of the grain and that in sexine pattern it is similar to that of Wahlenbergia and certain other genera of the Campanulaceae (Dunbar, 1 975b). An extensive literature dealing with the reproductive biology of Campanula, including a debate concerning the relative importance of autogamy in the genus, has been thoroughly reviewed by Shetler (1979a). Species of Campanula were often thought not to be insect pollinated and/or not to be allogamous, primarily because of the way in which the pollen is presented. While the flower is still in bud, the coherent anthers begin to shed pollen introrsely onto retractile hairs on the apical portion of the style. During anthesis pollen is carried aloft out of the anther tube as the base of the style elongates; the stigmas do not normally spread until all of the hairs and pollen grains have disappeared from the style. Nevertheless, the deposition of pollen on the style and the retraction of the trichomes have often been thought to constitute self pollination and to lead to self fertilization by growth of pollen tubes directly into the style. Other argu- sometimes overlap in individual flowers (and that in such cases exiting insects could transport pollen from style to stigma), that pollen could fall from styles to stigmas in pendent flowers, and that recurving stigmas could eventually come in contact with the style. Autogamy could also be effected by insects in multiflowered species, since floral development is sequential. Despite these and other hypotheses that have generally not been supported by observation, most evidence indicates that cross pollination (and subsequent fertilization) by in- sects is the rule within the genus. (For a list of pollinators, see Knuth.) The flowers of Campanula show adaptations usually associated with insect pollination, including blue or violet, campanulate or tubular corollas; nectar and a means of protecting it from visitors that do not enter the flower in a way likely to effect pollination (the broad, triangular filament bases form a dome over the nectary such that entrance to it can best be gained near the style after passage by pollen and stigma); proterandry; and echinate pollen. Gadella (1 964) found that among 2 1 species tested, spontaneous self fertilization occurred in only two (including one population of C. persicifolia) when flowers were bagged, and that mature seeds were produced in only one of 1 8 species artificially self pollinated. JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 Narrower debates have centered on the adaptive significance of the method of pollen presentation in Campanula, and in particular on the fact that the collecting hairs retract. The mechanism of presentation may prevent pollen gathering by nonpollinating visitors, and/or it may insure that pollen is de- posited on pollinators in places most likely to come in contact with stigmas of the same species, thus contributing to pollen economy and perhaps to repro- ductive isolation as well (Brantjes). It seems clear that the hairs serve to hold the pollen in place as it is removed from the anther tube, but it has been suggested that they may also help in the exit of small insects from the corolla. The retraction of the collecting hairs, a feature probably unique to the Cam- panulaceae, involves an invagination of the apical portion of each hair into its basal portion that is thought to be irreversible and osmotically induced (see Shetler, 1 979a). Although this may be a mechanism by which pollen is brought into contact with the stylar tissue, it has been shown that the hairs do not function in this way and that pollen tubes are only able to penetrate the style through the stigmatic surfaces. The lack of evidence for spontaneous autogamy also argues against stylar pollination and, therefore, against the proposal that retracting hairs help to bring it about. Retraction of the collecting hairs is not likely to facilitate pollen release to visiting insects and does not result from mechanical stimulation (but see Caro- lin, 1960), although pollen removal and hair retraction are simultaneous and do not occur to any great extent in the absence of insect visitors. Nevertheless, the hairs probably enhance the likelihood of pollination by holding pollen on the style in the path of nectar-seeking visitors. Retraction, therefore, may be of no real significance, but simply the consequence of some physiological change perhaps brought about by the pollinator. Many species of Campanula, especially those with large flowers (e.g., C. americana, C. persicifolia, C. rapunculoides), are cultivated for their orna- mental value, and some, including C. americana, have been investigated as sources of latex in the production of rubber (Buchanan et al., 1978a, 1978b). References: Under family references see Asen et al., Bailey & Lawrence, Barnes, Boissier, Brantjes, Brewbaker, A. de Candolle, A. P. de Candolle, Carolin (1960), Contandriopoulos, Crook, Cruden, Damboldt, Diggs, Dunbar (1975b), Fedorov, Gadella (1966b), Knuth, Radford et al. Schonland, Small, Sugiura, Teslov (1977a, 1977b), Teslov & Teslov, Teslov & Zapesochnaya, Teslov et al., Tutin, and Ward. Abrami, G. Life cycle and temperature requirements of seven herbaceous species. Giom. Bot. Ital. 105: 295-318. 1971. [C. rapunculoides stoloniferous.] . Optimum mean temperature for plant growth calculated by a new method of summation. Ecology 53: 893-900. 1972. [Experiments included C. rapunculoides .] Accorsi, C. A., & L. Forlani. Schede per una flora palinologica Italiana. Contributo n. 4: schede di nuova impostazione. Arch. Bot. Biogeogr. Ital. 52: 58-111. 1976. [C. rapunculoides, schede 36.] Alexander, E. J. Campanula divaricata. Addisonia 21: 13, 14. pi. 679. 1939. [Color Alfnes, A. T. Dispersion of especially troublesome weeds in nursery stock. (In Nor- wegian.) Gartneryket 658(48): 772-774. 1975. [C. rapunculoides .] 1986] ROSATTI, JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1986] ROSATTI, SPHENOCLEACEAE AND CAMPANULACEAE 33 ARNOLD ARBORETUM [vol. 67 e of McVaugh (1945b). ; of Triodanis occur in th 1986] ROSATTI, SPHENOCLEACEAE AND CAMPANULACEAE 35 JOURNAL OF THE ARNOLD ARBORETUM [vol. mm wide), flatly lenticular, smooth, highly polished seeds. It occurs in a variety of habitats in Louisiana, western Arkansas, southern Missouri, and eastern Oklahoma. Triodanis texana, with quadrangular, roughened seeds, occurs mostly in sandy soils in woods, pinelands, and plains and has a distribution in eastern Texas that closely approaches southwestern Louisiana. (It is evidently unknown from our area.) Triodanis Holzingeri, distinguished from the preceding species by linear (vs. broadly elliptic to rounded) capsule openings, occurs in prairies in the central United States. Although McVaugh (1945b) reported an outlying station in southwestern Tennessee, I have seen no specimens of this species from the Southeast. Triodanis leptocarpa (Nutt.) Nieuwl. is easily recognized by its lanceolate to linear leaves subtending the axillary inflorescences (often appearing as flower bracts in the case of one-flowered inflorescences) and its long, narrow capsules with openings near the top. It is a prairie plant of the central United States that enters our area in western Arkansas and Louisiana. Bradley 3281 (ny), from a roadside in Missouri, is intermediate between T. leptocarpa and T. perfoliata in the shape of leaves subtending inflorescences and in the position of capsule openings. Cytological evidence supports the recognition of Triodanis (Gadella, 1966b) in that both Specularia speculum A. DC. (the type of Specularia) and S. pen- tagons (L.) A. DC. (an enigmatic species returned to Campanula by McVaugh [1945b]) are In = 20, while Triodanis falcata (Ten.) McVaugh (Mediterranean region) is In = 26 and T. perfoliata and T. biflora are 2n = 56. These data also suggest that two infrageneric taxa may be recognizable in Triodanis, one comprising T. falcata and the other made up of the New World species. Gray proposed that the American species of Triodanis were divisible into two sections based primarily on placentation and capsule shape. Accordingly, he placed T. biflora (placentation axile, capsules elliptic) in one and T. lepto- carpa (placentation parietal, capsules narrowly elliptic) in the other. It is notable that in T. perfoliata placentation in a single ovary may grade from nearly parietal in the upper part to axile toward the base (Figure Id, e), suggesting that the proposed sections of Gray are untenable. Bradley scored individuals of Triodanis perfoliata and T. biflora obtained from natural populations in the Southeast and grown under uniform conditions, as well as the offspring resulting from various crosses involving these plants (selfing, F, and F2 hybrids, and backcrosses to both parents), for each of the characters mentioned above. Since a range of values extending from one ex- treme to the other was obtained for each character and since the hybrids were fertile, Bradley considered the plants involved to be conspecific. However, because of the persistence, despite interbreeding, of the combination of three traits characterizing T. biflora (which Bradley attributed to linkage and the high degree of self fertilization resulting from low numbers of chasmogamous flowers), he chose to recognize this taxon as T. perfoliata var. biflora (Ruiz & Pavon) Bradley. Ward, however, rejected this view and chose instead to main- tain the species as distinct because other workers had not expressed difficulty in separating them and because his own observations revealed few or no in- termediates. Ward also indicated that isolating mechanisms were clearly op- 1986] ROSATTI, SPHENOCLEACEAE AND CAMPANULACEAE 37 erating in the field and that treating these taxa as varieties would constitute too great an expansion of the original meaning of that rank. The inflorescence characters discussed above have been found to be suscep- tible to environmental modification. Data obtained by Trent (1940b) for Triodanis perfoliata indicated that, although cleistogamous andchasmogamous flowers increased in number at equal rates with increasing photoperiod, chas- mogamous flowers increased more quickly than cleistogamous ones with in- creasing light intensity. Since in both cases cleistogamous flowers outnumbered chasmogamous ones, the proportion of all flowers that were chasmogamous was greater when both photoperiod and light intensity were increased. Bradley reported that while the number of chasmogamous inflorescences in T. perfoliata could be reduced to two by poor habitat conditions, T. biflora produced only two even under good conditions. The structure of the flower in Triodanis perfoliata is extremely variable (Trent, 1940a). Among chasmogamous flowers sepals, petals, and stamens vary in number (two to seven sepals, one to six petals, and one to six stamens) and include both normal and vestigial parts, while the ovary is consistently three- locular. Cleistogamous flowers are less variable (three to six sepals, three to five petals, and one to five stamens), with sepals always normal, petals always vestigial, and stamens either normal or vestigial; the ovary is either two- or three-locular. Trent (1942) compared the anatomy of cleistogamous and chasmogamous flowers of Triodanis perfoliata. The ovules and seeds of both were found to conform to the types seen throughout the Campanulaceae, although chasmog- amous flowers were determined in general to produce more ovules (80-290) than cleistogamous ones (31-1 80). Each type of flower was found to produce both binucleate and (less frequently) trinucleate pollen, but differences in pol- lination were reported. Fertilization was observed in chasmogamous flowers in a few instances but never in cleistogamous ones, in which some evidence for parthenogenesis was reported (pollen grains were observed to germinate within the anthers and to grow through the anther wall but only as far as the adjacent stigma). Although embryo development in T. perfoliata appeared to conform to the “Capsella type” (= Crucifer type?), something closer to what is usually termed the Solanad type (characteristic of the Campanulaceae) was described (the basal cell results in a suspensor of 5 or 6 cells). Triple fusion may or may not occur, and endosperm development was reported to be the walled (= cellular) type. Finally, a greater percentage of seeds germinated from chasmogamous flowers than from cleistogamous ones (47.5 percent vs. 0.07 percent), while ungerminated seeds from both types of flowers usually had endosperm but no embryo, indicating that endosperm may have developed without fertilization. Species of Triodanis are easily propagated, evidently either by seed or veg- etatively, and are sometimes grown in flower gardens (Bailey et al). Under family references see Bailey et al, Chittenden, Gadella (1966b), Greene & Blomquist, Mark, McVaugh (1945b), Oosting & Anderson, Taylor, and Ward. JOURNAL OF ' ARNOLD ARBORETUM [vol. 67 r ex Roth, Nov. PL Sp. 399. 1821, i 1986] ROSATTI, SPHENOCLEACEAE AND CAMPANULACEAE 39 JOURNAL OF THE ARNOLD ARBORETUM [vol. 6 7 1986] ROSATTI, SPHENOCLEACEAE AND CAMPANULACEAE 4 1 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1986] ROSATTI, JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 As it is presently understood, however, the genus includes only 5- 1986] ROSATTI, SPHENOCLEACEAE AND CAMPANULACEAE 45 a native of Europe and the Mediterranean region (including parts of northern Africa). Although widely distributed in its native range (to as far as 62° N latitude, and from low elevations more sporadically to the subalpine and alpine zones), the species appears to be most abundant in areas with an oceanic climate. It is best adapted to full sun and well-drained soils (Bailey et al.) and is therefore most frequent on sandy soils in pine woods, stabilized dunes, heaths, and dry, grassy places (Kovanda). Of several infraspecific taxa that have been described in this variable species, only two appear to be tenable. Varietas montana is a diploid, 2 n= 12, and a strict calcifuge, while var. litoralis Fries, 2n = 14 (an aneuploid), is known only from saline soils in northern Europe; the typical variety is unknown from such habitats (Kovanda; see Tutin re- garding var. littoralis Boiss., evidently a different taxon). In North America Jasione montana has become established mostly along the east coast from Massachusetts to North Carolina, where it demonstrates the same climatic and ecological affinities prevalent in Old World populations of var. montana. Although the species was established and spreading on Long Island by 1925 (Burnham & Latham), it seems to have remained unknown in the Southeast until more recently. It was not included for this region by Small (1933) but was listed about 30 years later by Radford et al. as a roadside rarity from Moore Co., North Carolina. The base chromosome number in Jasione appears to be x = 6 (Kovanda). Although diploids, tetraploids, hexaploids, and decaploids (i.e., 2 n = 12, 24, 36, and 60) have been reported, polyploids are evidently unknown in J. mon- tana. Parnell (1982a) found a relatively low recombination index (number of haploid chromosomes plus average number of chiasmata per meiotic cell) and a high percentage of terminal chiasmata in J. montana, each considered an alternative to self fertilization (see discussion of breeding biology, below) as a means of bringing about a temporary reduction in the level of genetic recom- bination. According to Parnell (1982b), Poddubnaya-Amoldi showed that the pollen of Jasione montana is binucleate. Dunbar (1973b) found it to be similar to that of Campanula in shape (spheroidal) and aperture type and number (3 or 4 pores), but different in that the basal divisions of the spinules are less pro- nounced, the sculptured elements are in a more irregular pattern, the pores lack opercula, and the tectum is perforated. Pollen inviability was shown to vary within inflorescences, among flowers maturing at different times, and within populations of J. montana. Plants with very high percentages of inviable pollen (up to 99 percent), as well as completely male-sterile plants, are evidently sporadically distributed throughout the range of the species (Parnell, 1982b). The ovule of Jasione montana in general conforms to that characteristic of the Campanulaceae (see Berger & Erdelska; Erdelska, 1968), except that the integument is so transparent that it is possible to study the living embryo sac (Erdelska, 1968). While the megagametophyte is generally of the Polygonum type, the position of the egg cell is possibly unique among angiosperms in that it usually appears to be separated from the embryo sac wall by the synergids (Berger & Erdelska). Embryogeny appears to be of the Solanad type (see Soueges). The breeding biology of Jasione montana was studied in detail by Parnell JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1986] ROSATTI, SPHENOCLEACEAE AND CAMPANULACEAE 47 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 lobes, x 2; c, flower with corolla removed, filaments monadelphous except at base, stigmas just beginning to expand, x 2; d, anther tube formed by connation, the lower 2 anthers shorter and tufted, x 4; e, cross section of anther tube after dehiscence, style in center, x 6; f, tip of style with stylar brush and unexpanded stigmas at time of anther 1986] ROSATTI, SPHENOCLEACEAE AND CAMPANULACEAE 49 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1986] ROSATTI, SPHENOCLEACEAE AND CAMPANULACEAE 5 1 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1986] ROSATTI, SPHENOCLEACEAE AND CAMPANULACEAE 53 Kansas. Lobelia Gattingeri is known only from a small area in the limestone Boykinii Torrey & Gray and L. Canbyi Gray, both 2w = 14, have linear, sessile, JOURNAL OF THE ARNOLD ARBORETUM ROSATTI, SPHENOCLEACEAE AND CAMPANULACEAE 55 giana McVaugh) also occurs in this region but extends onto the Coastal Plain as well, into parts of western Florida and adjacent areas and eastern North Carolina and Virginia. A plant of tidal marshes but also occurring in swamps and on low ground near the coast, L. elongata is known from the area between Georgia and Delaware. Four medium- to large-flowered species are diploids {In = 14). Lobelia brevifolia Nutt, ex A. DC. (leaves numerous, short, sessile, divaricate to slightly ascending, short pectinate) and L. puberula13 (usually with relatively dense stem pubescence) were thought by Bowden (1959b) to have been segregated relatively early in the history of the genus in the southeastern United States. Nevertheless, a close relationship is indicated by the frequency with which they hybridize (Bowden, 1 961b; McVaugh, 1936; Ward). Individuals of both species usually have rather weak, unbranched stems, but the calyx auricles of L. brevi- folia are broad and foliose and usually cover the calyx tube completely, while those of L. puberula are generally much smaller. Holroyd stated that the pres- ence of medullary bundles in the stem is probably diagnostic for L. puberula, although this feature is apparently rare or unreported for the Lobelioideae in general (Metcalfe & Chalk, 1950). Bowden (1960a) discovered a haploid seed- ling among the progeny of one of his experimental crosses, indicating a low frequency of parthenogenesis in L. puberula. Lobelia puberula occurs in wet places and various kinds of soils in woods, on low ground, and in thickets throughout most of the southeastern United States and adjacent areas. Turner reported it from a “bog” in eastern Texas, intermingled with what he considered the strikingly dissimilar, though closely related, L. Reverchonii Turner, a species that enters our range in Louisiana (Rapides Parish). It differs from L. puberula in having greater distances between lower flowers (2.5-5 cm), densely hirsute pedicels and calyx bases, recurved lower corolla lobes, and longer corollas (9-1 5 mm) and anther tubes (4-6 mm). Lobelia brevifolia usually grows on damp, sandy soils in pinelands and is restricted to the Coastal Plain from the Florida panhandle to eastern Louisiana, usually near the coast. McVaugh (1936) indicated that on the basis of flower structure this species is related to L. glandulosa. Lobelia siphilitica, 2n = 14, and L. cardinalis, the other two diploid species in Bowden’s medium- to large-flowered line, generally produce larger flowers than do any other species of Lobelia in the Southeast. Although strong mor- phological characters separate them, plants of each species are perennials and may be unique among southeastern members of Lobelia in that they form offset shoots (see, however, Holroyd, who stated that L. in/lata and others also demonstrate this type of vegetative reproduction). Bowden considered each to be relatively recently evolved and to have shared a common ancestor with no other extant species. Artificial hybrids between the two are well known (e.g., Ames, 1903; Bowden, 1961a, 1964a, 1964b), but natural hybrids are evidently quite rare, though rather frequently reported (Bowden, 1964a; Ferreniea; lorphologic/geographic JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1986] ROSATTI, SPHENOCLEACEAE AND CAMPANULACEAE 57 Bowden (196 la, 1 961b) reported the results of some 1400 attempted crosses involving 21 species of subsect. Lobelia (all but two of which occur in the Southeast) and 82 of a possible 420 species pairings. No hybrids involving L. Lobelia and presented a detailed account of these and related topics (including JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 pollen) based on material of L. amoena 14 from the Carolinas. In general features of ovule and seed development, Lobelia conforms to the Campanulaceae as a whole, but there are some notable peculiarities. Cooper reported that in L. cardinalis the synergids and antipodals function as the chalazal and micropylar haustoria, respectively (sic; reversal of order probably inadvertent), instead of the usual condition (in angiosperms) in which the haustoria differentiate from cells of the endosperm (see, however, Subramanyam, 1952). However, Ma- heshwari determined that the synergids and antipodals in L. trigona Roxb. disintegrate at the time of fertilization, and that both kinds of haustoria develop from cells of the endosperm (see also Kausik, 1935; Kausik & Subramanyam, for L. nicotianifolia Heyne); Subramanyam (1951) later showed the same con- dition in L. cardinalis. Truly haustorial synergids probably only occur in a few species of Compositae (although detailed and illustrated accounts are lacking), according to Maheshwari, who also reported that the haustoria are formed from the endosperm in Wahlenbergia gracilis and Sphenoclea zeylanica. Hewitt’s account of ovule and seed development in Lobelia amoena is gen- erally consistent with those provided for other species of Lobelia, including L. nicotianifolia (Kausik, 1938), L. pyramidalis Wallich (Subramanyam, 1949), L. siphilitica (Crete, 1938a), and L. trigona (Kausik, 1935). Polyembryony has been reported in L. siphilitica (Cr6te, 1938b). Brewbaker reported that the pollen is binucleate when shed in five species of Lobelia, including L. cardinalis, and indicated that Schnarf’s report of a trinucleate condition in the extraregional L. nicotianifolia needed to be verified. Nevertheless, trinucleate pollen was indicated for this species by Kausik (1 938), for L. cardinalis by Cooper, and for the extraregional L. pyramidalis by Su- bramanyam (1949). Hewitt determined that the pollen of L. amoena is bi- nucleate, although only fragments (if anything) remain of the tube nucleus at anthesis. The pollen of Lobelia cardinalis is deeply tricolpate and reticulate (Martin & Drew), and essentially the same has been reported for L. amoena (Hewitt). The pollen tubes of L. siphilitica were found by Halsted (1886) to be variously expanded at the tips in material growing through the style, evidently in contrast to the usual condition in Lobelia, and the same was also later observed in L. cardinalis among pollen grains germinated in a sugar solution (Halsted, 1887). The tips of pollen tubes of L. amoena grown in a sugar solution were not expanded in illustrations provided by Hewitt, suggesting that tip expansion is not artificial and perhaps even indicative of a relationship between L. cardinalis and L. siphilitica. Seed germination was studied in several species of Lobelia by Muenscher. Light was required by L. inflata, L. siphilitica, and L. cardinalis, but not by several extraregional species. However, although some of the seeds of L. inflata germinated after one year, all those obtained from L. cardinalis or L. siphilitica were dormant. Muenscher suggested that since the seeds of all species were air dried upon harvesting, an additional dormancy was induced in L. cardinalis 1986] ROSATTI, SPHENOCLEACEAE AND CAMP ANULACEAE 59 year, Lobeline has also been Shah etal). ARNOLD ARBORETUM [vol. 67 from Lobelia cardinalis L. Jour. Chem. Soc. Perkin Trans. I. 1973: 140-144. 1973. . Phylogenetic relationships of twenty-one species of Lobelia L. section L Bull. Torrey Bot. Club 86: 94-108. 1959b. . Tn JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 AND CAMPANULACEAE 63 64 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 Witherspoon, J. T. Rediscovery of a hybrid Lobelia (Campanulaceae) in Missouri. Southwest. Nat. 19: 329. 1974. Ydrac, F. L. Sur l’appareil laticifere des Lobeliacees. Jour. Bot. Morot 19: 1 2-20. 1905. [Includes detailed account of latex system.] Young, T. P. Bird visitation, seed-set, and germination rates in two species of Lobelia on Mount Kenya. Ecology 63: 1983-1986. 1982. [Placed in subg. Tupa by Wimmer (unilabiate corollas).] KELLOGG & HOWARD, PHORADENDRON l WEST INDIAN SPECIES OF PHORADENDRON (VISCACEAE) E. A. Kellogg and R. A. Howard1 This situation was aggravated by Trelease’s (1916) monograph, which created es. We hope, however, that sand by noting which char- sin detail. This is a JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 KELLOGG & HOWARD, PHORADENDRON phthora, but other inflorescence types appear in that genus as well (Kuijt, 1961). Kuijt (1961, p. 5) noted that “one develops an intuitive feeling of relationships after scrutinizing a large number of species, a feeling which allows one to be reasonably sure that a certain specimen belongs in Dendrophthora and not in Phoradendron ” The Phoradendreae are members of the Viscaceae, a family distinguished from the Loranthaceae by numerous morphological, anatomical, and devel- opmental characters (Kuijt, 1969, summarizing Barlow, 1964, and Dixit, 1962). Because some workers have suggested separate evolutionary origins for the Viscaceae and the Loranthaceae (Kuijt, 1968), we believe they should be rec- ognized as separate families, rather than subfamilies. CHARACTERS STUDIED Branching pattern. Plants vary in the frequency with which the shoot apex aborts or is terminated by inflorescences. Most West Indian species are mono- podial, with the main axis continuing to grow and branching occurring only at the lower nodes. However, in three West Indian taxa ( Phoradendron dicho- tomum (Bertero) Krug & Urban, P. northropiae Urban, and P. racemosum (Aublet) Krug & Urban) the apex almost invariably aborts, establishing a bifurcate branching pattern. Stem shape. Young stems in Phoradendron may be quadrangular and more or less winged ( P . tetrapterum Krug & Urban), flattened and clearly two-edged (P. anceps (Sprengel) Krug & Urban), or nearly terete (P. carneum Urban). Many intergrading forms exist. For example, many flattened stems not only are two-edged in the plane of the leaves, but also have slightly less prominent ridges at right angles, giving the stem an approximately rhombic cross section. Furthermore, plants in most species become increasingly terete with age. Cataphylls. The West Indian members of the genus all bear reduced, bractlike cataphylls in addition to normal expanded foliage leaves. They are tightly appressed to the stem in fresh material but flare out on drying. The cataphylls may be fused to form what has been called a “vagina cataphyllaris,” or they may be partially separate. Occasionally they have a prominent white margin. In West Indian members of Phoradendron (unlike the condition in some species of Dendrophthora), the first pair of cataphylls is generally in the plane opposite to that of the prophylls, giving the plants a strictly decussate phyllotaxy. Each lateral branch has, as its first appendages after the prophylls, one or more pairs of cataphylls. In some taxa (Trelease’s sect. Continuae) additional pairs of cataphylls occur distal to each subsequent pair of foliage leaves, producing a pseudodistichous phyllotaxy. In the species descriptions, we record the distance of the cataphylls above “the node”; this refers to the node with a pair of expanded leaves immediately below the cataphylls. The first pair of appendages on the inflorescence is a pair of prophylls; these are commonly minute but occasionally prominent. In the latter case we have noted it in the species descriptions. Cataphylls distal to the prophylls may or may not occur on the inflorescence; when they do, they are often similar in morphology to the bracts subtending the flowers. JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 Leaves. Leaves may be nearly linear to suborbicular, symmetrical to notably gibbous or falcate, obtuse to acuminate, and wider from above to below the middle. They may dry green to black or sometimes reddish or almost fluorescent yellow ( Phoradendron flavens (Sw.) Griseb.). Several of the taxa described in this paper (e.g., P. wattii Krug & Urban, P. anceps) are highly variable in leaf shape and size. Field observations in the Lesser Antilles have revealed sub- stantial variation within a single plant; one specimen of P. anceps (Howard et al. 18816 ) bore leaves characteristic of four previously recognized “taxa.” In addition, all intermediate shapes can be found connecting “taxa” previously delimited on leaf shape. A biosystematic study should address the lability of this character; we have found it to be much less reliable than past treatments of the genus would indicate. Leaves may be basinerved or penninerved, although there are cases (partic- ularly in herbarium specimens) where the distinction becomes difficult. The separation of the three main basal nerves may be somewhat above the base on some leaves. Trelease (1916) used this character to distinguish his sections Basinervae and Penninervae\ while it is sometimes a useful character, it should not be weighted so heavily. Type of inflorescence. Through careful morphological and anatomical study, Kuijt (1959) has documented the variation in inflorescence pattern in some members of the Viscaceae, and his 1961 revision of Dendrophthora has shown how useful it can be taxonomically. In the Caribbean species of Phoradendron, we have found six types of inflorescence (Figure 1), including Kuijt’s types la and lb and modifications of them. The inflorescence in tribe Phoradendreae is an articulated structure with two opposite bracts subtending each fertile intemode. Above each bract the flowers are most commonly arranged in two parallel columns, with a single apical flower placed midway between the col- umns (type la). In other plants there are three full columns of flowers above the bract (type 1 b). In some plants the three columns are reduced to one flower each, producing a triad of flowers above each bract. As Kuijt has noted, the triad cannot be assigned to either type; however, its occurrence is consistent within certain taxa (e.g., P. mucronatum (DC.) Krug & Urban), so we will refer to it simply as a triad. We use these types merely as short-hand descriptions of inflorescence patterns; we do not claim that they are “real” in any devel- opmental sense. Types Id and le may simply represent slightly aberrant forms of la, but we name them for convenience in referring to them (e.g., in P. anceps, where they are common). Staminate and pistillate flowers. Plants may be monoecious or dioecious. If monoecious, staminate and pistillate flowers both occur on the same inflo- rescence. Within this inflorescence they may occupy separate intemodes (e.g., in Phoradendron solandrae Proctor), in which case the staminate intemodes are generally above the pistillate. If they occupy the same intemode, their position in the intemode may be variable or regular; if the latter (e.g., in P. mucronatum ), generally the apical flower is staminate and the lower flowers pistillate. Flower orientation. Eichler (1868) attempted to use orientation of the flowers 1986] KELLOGG & HOWARD, PHORADENDRON 2. 1: 185. 1848. TAXONOMIC TREATMENT JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1986] KELLOGG & HOWARD, PHORADENDRON 73 Wiles W30 (k). JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1 1 -5(— 1 3.5) x (1 . 1— ) 1 .7—5.9 cm, a 4.4-15 x 1.4-5. 5 c or 3-lobed, flaring, forming infundibular tube; fertile intemodes 3 to 5, type 2/1. Fruits globose, 1986] KELLOGG & HOWARD, PHORADENDRON 75 Distribution. Venezuela, Cuba, Hispaniola, Puerto Rico, St. Lucia. Although it is somewhat inconvenient to lectotypify a species with a rep- resentative of a mixed collection, Fuertes 275 at g is the only extant specimen that was definitely seen by the authors of the species. The Puerto Rican plants are more consistently basinerved than are plants from elsewhere in the range. The one collection from St. Lucia {Proctor 18089 ) is unusual in that many of the fertile intemodes have type lb rather than type la inflorescences; however, fruit shape and texture and leaf shape and nervation all suggest placement in this species. Phoradendron barahonae is similar to P. anceps but is often more clearly penninerved and tends to have more robust inflorescences with more flowers per intemode. The fruits of P. barahonae are globose and somewhat flattened apically; the sepals often dry a color different from the rest of the fruit. In P. anceps the fruits are always pyriform until maturity, when they become globose but not flattened; the sepals are generally the same color as the fruit. The fruits of P. anceps have a pronounced golden sheen (visible under a 10 x lens) when young, whereas those of P. barahonae are never golden. Trelease (1916), in his description of P. cerinocarpum, and Leon and Alain (1951) recorded the fruit color as red, but data on all specimens we have seen list it as white. 3. Phoradendron carneum Urban, Bot. Jahrb. Syst. 23: 1. 1897. Type: Mexico, Jalisco, hills near Guadalajara, 1888, Pringle 1854 (holotype, b, de- stroyed; isotypes, gh!, ny!). Figure 2d. Stems flattened, rough-striate; intemodes flaring above; prophylls prominent; cataphylls 1 pair on lateral branches only, basal, scarcely fused, thick, fleshy, acute. Leaves sessile; blade linear, straplike, 4. 8-7. 5 x 0.5-0. 7 cm, apex obtuse, bases meeting, nerves all basal. Plants dioecious; inflorescences 1 or 2 per leaf axil, to 2.5 cm long, axis ca. 2 mm thick when dry; cataphylls lacking; prophylls prominent; bracts nearly fused, forming infundibular tube; fertile intemodes 2, type la; flowers 5 to 11 per bract, orientation irregular. Fruits globose, in spreading cup with papillate margins, red-orange, warty, sepals closed. Distribution. Mexico (Guanajuato, Jalisco, Michoaccin, Oaxaca, Puebla, Que- retaro), Trinidad. There are two possible explanations for the apparent disjunction: if the species has often been misidentified in herbaria the range may be more con- tinuous than is evident here; or the two Trinidad specimens may be mislabeled. Both collected by Eggers, they are dated 12/83 and 2/83 but are otherwise unnumbered. Most of Eggers’s other Caribbean collections are numbered. Eg- gers was in Trinidad in 1883, but he had been in Veracruz and Oaxaca from April 1865 to November 1867 when he was in the Danish army. Urban (1902) did not discuss any Eggers collections from Mexico, but these “Trinidad” specimens may have actually been collected while Eggers was in Oaxaca or traveling through southern Mexico after his discharge from the army. JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1945). ) Griseb. FI. Brit. W. Stl® ! sill! !!l!!illhil;f! Ill JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 Urban {in Urban, 1897) reported should perhaps be returned to vai re. Development of warty fruits i i C. Martius, FI. Brasil. 5(2): 127, .FI. Brit. W. Indian Is. 313. 1860, e ation is superseded by the discovery of authentic material at k. hE* , 5-14 x 1. 5-4.5 racts to form straight horizontal collar; fertile intemodes 3 or 4, plants type Id or triads, with flowers 3 to 5 per bract, apical flower nes 1/2, on staminate plants type lb, with flowers 10 to 13 per y Paul Serrein 1909 int . The leaves are very s 1986] KELLOGG & HOWARD, PHORADENDRON pistillate spikes of the two species are not at all alike. In P. grisebachianum they are slender with three or four fertile intemodes; the flowers are mostly in triads borne well above the midpoint of the intemode and produce red to red- orange berries that dry rough-wrinkled. In P. hexastichum they are thick and clavate with four or five fertile intemodes; the flowers are in a type lb arrange- ment and extend to the bottom of the intemode, producing white berries that dry more or less smooth and blackish. The confusion is created by the staminate spikes, which are type lb in both species. The only difference between the two that we have been able to discover is that the members of a pair of bracts in P. grisebachianum are fused to form a straight-edged flaring cup and are not at all distinguishable; in P. hexastichum the “cup” is clearly bilobed and fre- quently white margined. Also, because P. hexastichum is monoecious, devel- oping fruits are present on all but very young specimens. Defined in this way, the two species have nonoverlapping ranges: P. grisebachianum is endemic to Jamaica, whereas P. hexastichum does not appear to occur on that island. 8. Phoradendron gundiachii Krug & Urban in Urban, Bot. Jahrb. Syst. 24: 44. 1897. Type: “habitat in Cuba in summitate Guayabon m. Jun.,” C. Wright 2650 (holotype, b, destroyed; isotypes, g, gh!, k!; photo of g specimen at ill!). Figure 2i. “ Phoradendron flavescens forma foliis majoribus," Griseb. Catal. PI. Cubens. 120. 1866; not P. flavescens Nutt. Type the same as for P. gundiachii. Stems flattened, rapidly becoming terete; cataphylls 1 or 2 pairs at base of lateral branches only, second pair < 1 cm above first, ovate, spreading. Leaves with petiole 4-9 mm long; blade elliptic to ovate, 4. 6-6. 5 x 1.6-4.5 cm, apex rounded to obtuse, base cuneate, nerves obscure, 3 basal veins and 2 lateral veins joining midvein above base. Inflorescences solitary in leaf axils, to 3 cm long, axis ca. 1 mm thick when dry; cataphylls 1 pair; bracts shallowly cleft and white margined; fertile intemodes 2 or 3, type Id, le, or triads; flowers 3, 4, or 6 per bract; pistillate only or both sexes intermixed; flower orientation irregular. Fruits globose, white; sepals closed. Distribution. Endemic to Cuba. See discussion under Phoradendron rubrum. Urban (1897) noted that this species is very similar to P. acinacifolium C. Martius. 9. Phoradendron haitense Urban, Symb. Antill. 5: 334. 1 907. Lectotype (Tre- lease, 1916): Haiti, ad Petite Riviere de l’Artibonite, Picarda 1666 (b, destroyed, photo! in Trelease, 1916, pi. 135). Figure 2j. Stems flattened; cataphylls 1 pair at base of lateral branches only, not fused, ovate. Leaves with petiole 5 mm long; blade lance-linear to oblanceolate, often somewhat falcate, 5-12 x 1-2 cm, the apex acute to truncate, often mucronate, the base tapering, the nerves basal. Inflorescences 1 per leaf axil, to 3 cm long, axis ca. 1 mm thick when dry; cataphylls lacking or 1 pair, like bracts flaring, navicular, entire, and white margined; fertile intemodes generally 2, type Id JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1986] KELLOGG & HOWARD, PHORADENDRON KELLOGG & HOWARD, JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 KELLOGG & HOWARD, PHORADENDRON it is JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 KELLOGG & HOWARD, PHORADENDRON JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 i mi 1986] JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 oblanceolate, 2.6-10.3 x L 1-2.1 (-3) cm, l to 3 per leaf JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 95 2-3] 25: , 6-12.5 x 2.2-4.3 cm, l^f axil6^ 5Uo^l^ngbaxLC: 1871 Type: Lord Howe VIOLACEAE Viola betonicifolia Smith subsp. novo-guineensis D. M. Moore, Feddes Repert. Spec. Nov. Regni Veg. 68: 82. 1963. , NORFOLK AND LORD I LITERATURE CITED Oliver, W. R. B. 1917. The vegetation and flora of Lord Howe Island. Trans. & Proc. New Zeal. Inst. 49: 94-161. 122 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1986] HAMMEL, LAURACEAE 123 NEW SPECIES AND NOTES ON LAURACEAE FROM THE CARIBBEAN LOWLANDS OF COSTA RICA Barry E. Hammel1 In the summer of 1 979, intensive collecting was begun toward a flora project for Finca La Selva, a biological field station managed by the Organization for Tropical Studies (OTS) in the Caribbean lowlands of northeastern Costa Rica. This area of approximately 1400 hectares is estimated to harbor at least 1500 species of vascular plants (Hammel & Grayum, 1982). Despite Costa Rica’s history of intensive collecting (Prance, 1978), the entire Sarapiqul region, in which the La Selva field station lies, is still relatively unknown botanically (Gentry, 1978). Approximately 50 new species of flowering plants already have been or soon will be described from the station, most of them having come to light through intensive collecting for the flora project. Among these novelties are the four species of Lauraceae here described. The Lauraceae are mostly tropical in distribution, and the family is partic- ularly well represented in South America. Especially in the New World, revi- sionary work is sorely needed in order to determine generic limits. At the same time, many more collections and associated field observations are necessary to match fruits with flowers and to determine the variation within species. Eleven of the 17 New World genera are endemic, and 12 genera are known from Central America. The family contains between 2000 and 2500 species world- wide, with about 150 species in Central America (Allen, 1945). A total of 28 species in six genera ( Beilschmiedia Nees, Licaria Aublet, Nectandra Rolander, Ocotea Aublet, Persea Miller, Phoebe Nees) are known from the La Selva field station. Eleven of these are also known from South America, and five— here discussed under “Additional Notes”— are shared with the West Indies. Anyone attempting to identify Lauraceae in the Neotropics soon learns that under the prevailing taxonomy it is often difficult to identify specimens even to genus. This problem is due, in part, to the nature of the plants. They are usually tall trees with inconspicuous flowers, and most species are restricted to primary forest habitats where the individuals are often far apart. Many species are cryptically dioecious (both sexes have apparently “perfect” flowers), and most are seasonal; the fruits take several months to mature, so specimens rarely have both flowers and fruits. As a result, many species are known from few collections, and new species have often been described from incomplete ‘Department of Botany, Duke University, Durham, North Carolina 27706. Present address: Mis- souri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166. © President Journal of th 124 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1945; Kostermans, 1952;! Croat, 1978; Howard, 1981). s of field work at La Selva and e 1986] HAMMEL, LAURACEAE 125 B-D from Hammel 8663, E from Hammel 10532.) very slender branched and sparsely flowered, the ultimate clusters of flowers umbelloid, glabrous. Flowers urceolate to globose, 2. 3-2. 5 mm long, yellow- green, very inconspicuous; tepals ligulate, 1.2 mm long, about as long as floral tube, patent or curved inward at anthesis; inner 3 stamens fertile, 1 .2 mm long, the anthers much shorter than filaments, oblate depressed, the valves opening 126 ARNOLD ARBORETUM [vol. 67 Arbor 10-15 17-27 x 6.5-10 cm, £ HAMMEL, LAURACEAE 128 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1986] HAMMEL, LAURACEAE 129 , 12-19 by 3.5-7 cm, t 130 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 just E of 1932 (ho- 800 m, Utley & Utley 5407 (duki , Neil 5122 132 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 133 foreman and boatman at Finca La Selva. Don Rafael was an ardent conser- ADDITIONAL NOTES: AFFINITIES WITH WEST INDIAN PLANTS All of the genera of Lauraceae at the La Selva field station also occur in the . 1: 327, 2: t. 126. 1775. O. O. Kew Bull. 5: 217. 1933. 134 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 two from Nicaragua (cited by Mez, 1889) and one from Costa Rica, Tonduz 12876 (us!) (correctly identified by Wticr).Ite rarity probably accounts for the t the cupules of f , LAURACEAE ‘ 3187 (holotype, Cerro Azul, 3 July, Croat j.M CR)"cer^o Jefe, D’Arcy&i Antonio 2325 (mo); El Llano-Carti road, Mori et al 4150 , J. L. 1886. Lesi 136 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 Hammel, B. E., & M. H. Grayum. 1982. Preliminary report on the flora project of La Selva field station, Costa Rica. Ann. Missouri Bot. Gard. 69: 420-425. Howard, R. A. 1981. Nomenclatural notes on the Lauraceae of the Lesser Antilles. Kostermans, A. J. G. H. 1937. Revision of the Lauraceae II, the genera Endlicheria, Cryptocarya (American species) and Licaria. Rec. Trav. Bot. N6erl. 34: 500-609. . 1952. A historical survey of Lauraceae. Jour. Sci. Res. 1: 83-95, 113-127, 141-159. . 1957. Lauraceae. Reinwardtia 4: 193-256. Kubitzki, K. 1982. Aniba. In: K. Kubitzki & S. Renner, Lauraceae I (Aniba and Aiouea). FI. Neotrop. Monogr. 31: 1-84. Mez, C. 1889. Lauraceae Americanae monographicae descriptae. Jahrb. Konigl. Bot. Gart. Berlin 5: 1-556. Bot. Gard. 65: i, ii. Standley, P. C. 1937. Lauraceae. In: Flora of Costa Rica. Publ. Field Mus. Nat. Hist., Bot. Ser. 18: 450-460. 1986] ROLLINS, LEPIDIUM 137 ALIEN SPECIES OF LEPIDIUM (CRUCIFERAE) IN HAWAII Reed C. Rollins1 138 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 139 i L. Sp. PL 645. 1753. 140 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 U-shaped sinus and the margins of the siliques are fringed with minute tri- chomes. The pedicels are usually spreading, somewhat flattened, and covered with trichomes on their adaxial surfaces. The cauline leaves are few and deeply pinnately lobed. Most fully mature plants have lost the basal leaves. Both glabrous- and ffinged-fruited plants occur in the southwestern continental United States and Mexico. The fringed condition is more prevalent westward from Arizona to California and southward through most of Mexico than it is in Arkansas, Oklahoma, and Texas. One or two features of the Hawaiian material point to Mexico as a possible source for this introduction. One, stressed by Hitchcock (1945a), is late shed- ding of the sepals. In most plants of the species, they are shed very early: usually only two to four fully open flowers near the tip of the inflorescence will have retained them. In the Hawaiian specimens the sepals are retained in five to eight flowers of the upper inflorescence, a tendency also shown in several Mexican specimens. In addition, fruits of the Mexican specimens are more consistently fringed with trichomes than are those from elsewhere; since the Hawaiian specimens all show this feature, it rates as minor evidence in favor of a Mexican origin for the Hawaiian alien. ACKNOWLEDGMENTS This study arose from my review of the Cruciferae of Hawaii, requested by the authors (Warren L. Wagner, Derral Herbst, and S. H. Sohmer) of a manual of the flowering plants of Hawaii now in preparation. It is based mainly on material loaned by the Bishop Museum, for which I hereby express my ap- preciation. LITERATURE CITED Burbidge, N. T., & M. Gray. 1 970. Flora of the Australian Capital Territory. 447 pp. Australian National University Press, Canberra. Carlquist, S. 1970. Hawaii, a natural history. 463 pp. The Natural History Press, Garden City, New York. Carolin, R. C, & H. J. Hewson. 1981. Cruciferae (Brassicaceae). Pp. 94-102 in J. Jessop, ed., Flora of central Australia. A. H. & A. W. Reed Pty., Ltd., Sydney. Facerlund, G. O. 1947. The exotic plants of Hawaii National Park. Natural History Bulletin 10: 1-62 (mimeographed). National Park Service, U. S. Dept. Interior, Fosberg, F. R. 1969. Miscellaneous notes on Hawaiian plants. 5. Occas. Pap. Bernice Pauahi Bishop Mus. 24: 10-24. . 1972. Guide to excursion III, Tenth Pacific Science Congress, revised ed. 249 pp. Univ. Hawaii, Honolulu. Hewson, H. J. 1981. The genus Lepidium L. (Brassicaceae) in Australia. Brunonia 4: Hillebrand, W. 1888. Flora of the Hawaiian Islands, lxxi + 673 pp. B. Waterman & Co., New York. Hitchcock, C. L. 1945a. The Mexican, Central American, and West Indian lepidia. Madrono 8: 118-143. -. 1945b. The South American s species of Lepidium. Lilloa 11: 75-134. ROLLINS, LEPIDIUM St. John, H. 1973. List and summary of the flowering plants in the Hawaiian Islands. 5 19 pp. Pacific Tropical Botanical Gardens Memoir No. 1 . Cathay Press Ltd., Hong Kong. . 1981. Lepidium orbiculare (Cruciferae) of Kauai. Phytologia 47: 371-373. Thellung, A. 1906. Die Gattung Lepidium (L.) R. Br. Mitt. Bot. Mus. Univ. Zurich ■ JOURNAL OF THE ARNOLD ARBORETUM INSTRUCTIONS FOR AUTHORS General policy The Journal of the Arnold Arboretum is primarily a staff journal, and staff papers have priority. Other papers are accepted, as space permits, from former staff or former students, and from other botanists who have worked on our collections or who have done research on a plant group or in a geographic area of interest to the Arboretum. Manuscripts should be submitted in triplicate to Ms. E. B. Schmidt, Man- aging Editor, Journal of the Arnold Arboretum, 22 Divinity Avenue, Cam- bridge, Massachusetts 02138. A copy of the manuscript should be retained so that when reviews and/or editorial suggestions are received, any necessary corrections can be made and the appropriate portions of the paper resubmitted. For ease of editing, an outline of the paper (not to be published) showing the basic structure of the manuscript should be included. Preparation of manuscripts Papers should be triple spaced throughout (including title, abstract, text, citation of specimens, footnotes, acknowledgments, bibliography, and figure legends), on bond (not erasable) paper, with wide margins on all four sides. Nothing should be underlined except generic and infrageneric scientific names, italics when present in a quotation, and the collector and collection number of all specimens cited. Form and style. This can be determined from a recent issue of the Journal. The title should be as short as possible; it should usually contain the name of the family concerned but not authorities of scientific names. Each author’s current address should be given as a footnote to his name. An abstract should be included. Abbreviations should be employed only when two or more letters will be saved, and with the exception of units of measure, compass directions, and herbarium designations, they should always be followed by a period. Metric measurements should be used when possible. Authority names should be given for all generic and infrageneric taxa the first time they are mentioned in the text unless they are included in the formal taxonomic treatment. Acknowledgments should be placed at the end of the paper before the bib- liography. Footnotes should be kept to a minimum and should be numbered consecutively throughout the paper, with the exception of those appearing in tables. Here either standard symbols or lower-case letters should be used. In case of question, the latest edition of Words into Type, by Maijorie E. Skillin et al. (Prentice-Hall), should be consulted. Citation of specimens. Currently accepted geographic names should be used, with spelling according to a standard source. Names of countries should be in English and should be typed in regular capital and lower-case letters. Below the country level, names may be in the language of the country involved. If this option is taken, careful attention should be paid to consistency, spelling, and accent marks. Arrangement of areas should be consistent (geographic or alphabetic, preferably the former) within a paper. Data obtained from the label should not be changed unless it is obviously wrong or lacking critical information; in this case, additions or corrections should be bracketed. When a collector has a common surname, his initials should be given. With specimens from the Indo-Malesian area, care should be taken to de- termine whether collections are institutional or not. In the case of an institu- tional collection, the institutional series and number (e.g., LAE 20257) should be given. This, in addition to the location, is all the information needed for a brief, unambiguous citation. If it is desirable to include the collector, this information should be placed after the institutional series and number (e.g., LAE 20257, Foreman or LAE 20257 (Foreman)). The list of exsiccatae should be arranged by institutional numbers, where applicable, for ease of use. As an alternative to lengthy specimen citations, a list of specimens examined, prepared in the normal Journal style, may be provided to the Managing Editor. This list will be deposited in the Library of the Arnold Arboretum and Gray Herbarium, and copies will be prepared and distributed to the libraries of the New York Botanical Garden and the Missouri Botanical Garden. It will be available by interlibrary loan from the cooperating institutions; a brief state- ment in the published paper will notify readers of its availability. Specimen citations will still be necessary for nomenclatural purposes (e.g., citation and designation of types). Illustrations and legends. Reference must be made in the text to all maps, figures, and plates. Insofar as possible, their sequence should be determined by the order in which they are mentioned. Plates (illustrations grouped together at the end of an article) and figures (illustrations scattered through the text of an article) should be prepared with Journal page proportions in mind. The maximum size after reduction is 4.25 by 6.5 inches (10.8 by 16.5 cm) for figures, and 4.25 by 6.75 inches (10.8 by 17.1 cm) for plates. A figure may occupy any portion of the length of a page; plates should be more or less full- page size. To facilitate mailing and handling, mounted illustrations must be of a manageable size. Line drawings showing habit and plant parts should be lettered in the same order (e.g., habit, leaf, inflorescence, whole flower, calyx, corolla) on each il- lustration throughout a paper. Photographs should be trimmed, grouped appropriately, and mounted with no space between them on stiff white cardboard with a margin of at least 1 inch left on all four sides. The author’s name and the figure number(s) should be noted on the back of each illustration. To prevent bending or other damage, the art work should be wrapped carefully and shipped flat. Two clear copies of each illustration should be included for review purposes. Legends should be written in telegraphic style (see back issues for examples). They should be grouped in numerical sequence on a separate page, rather than placed below each figure or plate. Illustrations of each type (i.e., figures, plates, or maps) should be numbered consecutively and separately, figures and maps with Arabic numerals and plates with Roman numerals. For example, a paper could include figures 1-3, maps 1-3, and plates I- VII. If the illustrations are to be figures, each figure should be numbered separately, with the numbers running in order through the text. Subdivisions of the figures should be indi- cated with letters. Thus, figures 1-5 may appear on one page, figures 6-8 on the following page, and figure 9, A-C, on the next. In the case of plates, each page of photographs is a separate plate; the individual photographs comprising the plates are numbered or lettered consecutively. Illustrations with dark back- grounds should have white letters or numbers and vice versa. Whenever pos- sible, scales should be included in the illustrations; any magnifications necessary in the legends should be calculated to include reduction of illustrations to our Illustrations are not returned to the author after publication unless this is requested. Tables. Titles for tables should be short, with all explanations placed in footnotes. Tables should be as simple as possible and must be neatly typed. Long and/or complicated tables can be photographed directly if they are in good order and the copy is clear enough (in this case the copy should not be triple spaced, but should have spacing appropriate to contents and headings); this eliminates the chance for error and the need for proofreading. Tables should be numbered consecutively using Arabic numerals. Each table should be cited in the text. Bibliography. The Guide to Citation of Botanical Literature in the Inter- national Code (all editions through 1972) should be followed. When possible, reference should be made to past issues of the Journal for form. Titles should be abbreviated according to Journal (Schwarten & Rickett, Bull. Torrey Bot. Club 85: 277-300. 1958) or Botanico-Periodicum-Huntianum style; the two styles should not be mixed within a paper. Runovers should be indented. Except of first and middle names. All typing should be done in regular capital and lower-case letters, and nothing should be underlined except generic and infra- generic scientific names. Titles of articles and books should not be capitalized except for the first word, scientific names, and proper nouns and adjectives. Author’s alterations Charges for extensive alterations in proofs will be billed to the author. Page charges Authors are requested to help defray printing costs. Although actual printing costs are much higher, $20.00 per page is the customary charge, and authors are expected to make every effort to pay. Under special circumstances the fee may be reduced or waived altogether, if this is agreed upon in advance. Ability or inability to pay will in no way affect acceptance or handling of a manuscript. SALE OF BACK ISSUES Save 30%! Now is the time to complete or add to your holdings of the Journal of the Arnold Arboretum. Back issues dated 1976 through 1982 will soon be sent to Kraus Reprint Company. Until 30 April 1 986 orders for these numbers and a few miscellaneous earlier ones will be accepted at 30% off the regular price. Cost and availability are as follows: 52 53 54 55 56 57 58 59 63 31 101 165 73 33 35 28 Please send orders to E. B. Schmidt, Managing Editor, Journal of the Arnold Arboretum, 22 Divinity Avenue, Cambridge, Massachusetts 02138, U.S.A. Add $.50 per issue for postage and handling. Please note that at these prices, free replacement cannot be made for lost or damaged shipments. Insurance and/or special handling will be provided only at the purchaser’s request and expense. Back issues prior to 1974 can be ordered from Kraus Reprint Co., Route 100, Millwood, New York 10546, U.S.A. ; those subsequent to 1982 from Allen Press, Inc., P.O. Box 368, Lawrence, Kansas 66044, U.S.A. Please write directly to these addresses concerning cost and availability. Journal of the Arnold Arboretum January, 1986 CONTENTS OF VOLUME 67, NUMBER 1 The Genera of Sphenocleaceae and Campanulaceae in the South- eastern United States. Thomas J. Rosatti 1-64 The West Indian Species of Phoradendron (Viseaceae), E. A. Kellogg and R. A. Howard 65-107 Notes Relating to the Floras of Norfolk and Lord Howe Is- lands, II. P. S, Green 109-122 New Species and Notes on Lauraceae from the Caribbean Low- lands of Costa Rica. Barry E. Hammel 123-136 Alien Species of Lepidium (Cruciferae) in Hawaii. Reed C. Rollins 137-141 Volume 66, Number 4, including pages 395-526, was issued October 4, 1985. JOURNAL OF THE ARNOLD ARBORETUM HARVARD UNIVERSITY VOLUME 67 NUMBER 2 ISSN 0004-2625 Journal of the Arnold Arboretum The Journal of the Arnold Arboretum (ISSN 0004-2625) is published quarterly in January, April, July, and October for $50.00 per year, plus $5.00 postage for addresses outside of the United States, by the Arnold Arboretum of Harvard University. It is printed and distributed by the Allen Press, Inc., 1041 New Hampshire Street, Law- rence, Kansas 66044. Second-class postage paid at Lawrence, Kansas. POSTMAS- TER: send address changes to Journal of the Arnold Arboretum, % Allen Press, Inc., P, O. Box 368, Lawrence, Kansas 66044. Subscriptions and remittances should be sent to Journal of the Arnold Arboretum, 1041 New Hampshire Street, Lawrence, Kansas 66044, U. S. A. Claims will not be accepted after six months from the date of issue. Volumes 1-51, reprinted, and some back numbers of volumes 52-56 are available from the Kraus Reprint Corporation, Route 100, Millwood, New York 10546, EDITORIAL COMMITTEE S. A. Spongberg, Editor E. B. Schmidt, Managing Editor P. F. Stevens C. E. Wood, Jr. Printed at Allen Press, Inc., Lawrence, Kansas COVER: The stylized design appearing on the Journal and the offprints was drawn by Karen Stoutsenberger. JOURNAL ARNOLD ARBORETUM THE GENERA OF LOGANIACEAE IN THE SOUTHEASTERN UNITED STATES1 LOGANIACEAE C. F. P. 2: 133. 1827, JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 and calyces often with colleters adaxially. Leaves usually opposite or infre- quently whorled or pseudo whorled [or rarely alternate], simple, entire or toothed [to infrequently lobed], penninerved or with ascending, arcuate lateral nerves, or vasculature other than the midrib inconspicuous in reduced leaves. Stipules diverse, interpetiolar (then leafy, membranaceous, or reduced to a ridge or flange) or rarely represented by a pair of lobes flanking each petiole. Inflores- cences terminal (then sometimes in forks between branches) or axillary, fun- damentally cymose but otherwise variable, sometimes dichasial, cincinnous (then often spicate), fasciculate, thyrsiform, scorpioid, or paniculiform, some- times tightly congested, occasionally with solitary flowers; showy bracts lacking. Flowers mostly actinomorphic or somewhat zygomorphic, perfect or infre- quently imperfect, tetramerous or pentamerous. Calyx lobes separate or co- alesced, often unequal. Corolla tubular, cylindrical to funnelform or urceolate, with variously oriented imbricate, valvate [or contorted] lobes, the tube often occluded by hairs. Stamens typically alternating with same number of corolla lobes [or reduced in number], epipetalous, usually uniform; anthers introrse, dehiscing longitudinally; pollen grains usually colporate (to colpate in Spigelia). Ovary superior to partly inferior, usually bilocular (tetralocular in Buddleja madagascariensis) [or rarely unilocular or with 3-5 partial or complete locules], containing [1 to] many anatropous, hemianatropous, or reportedly amphitro- pous ovules in each locule on axile, frequently peltate [or rarely parietal or basal] placentae; stigmas unlobed, 2-lobed, or 4-lobed. Fruits usually capsules with septicidal dehiscence predominant but also with varying degrees of loc- ulicidal dehiscence, thus separating partly or completely into 2 (a pair of apical horns in some genera) or 4 segments [or infrequently loculicidal only], in some berries [or drupes]. Seeds usually numerous and small, winged or not, the embryo straight; endosperm initially nuclear or cellular [or intermediate], some- times ruminate, surrounding the embryo; embryo sac (megagametophyte) of the Polygonum type. (Including Buddlejaceae Wilhelm, Samenpfl. 90. 1910, “Buddleiaceae,” nom. cons.; Spigeliaceae Martius, Nov. Gen. Sp. PI. Brasil. 2: 132. 1827.) Type genus: Logania R. Brown, nom. cons. As recently defined by Leeuwenberg & Leenhouts (to embrace the Buddle- jaceae), a family of 29 genera and about 470 species in ten tribes distributed pantropically and subtropically with limited extensions into the temperate zones. The two largest genera, Strychnos L. and Buddleja L., are pantropical and contain between 90 and 200 species each. Seven genera grow free of cultivation in the continental United States. Emorya suaveolens Torrey, the sole species of its genus, is endemic to northern Mexico and southern Texas. The cultivated Strychnos spinosa Lam. escapes infrequently to disturbed places in Florida. As elaborated later, at least one species of Buddleja is naturalized in the southeastern United States, and species of Mitreola L. ( Cynoctonum J. F. Gmelin), Polypremum L., Spigelia L., and Gelsemium Juss. are indigenous in the range of the Generic Flora. The comprehensive treatment in the second edition of Die Natiirlichen Pjlan- zenfamilien (1980) contains a taxonomic revision at the generic level (Leeu- wenberg & Leenhouts) and stands as the principal modem work on the Lo- ganiaceae. See Klett and Solereder (1892) for earlier revisions. 1986] ROGERS, LOGANIACEAE 145 Broadly interpreted, the family is characterized, with exceptions, by opposite leaves with interpetiolar stipules (or sheaths, flanges, or ledges), fundamentally cymose inflorescences, actinomorphic (or weakly zygomorphic), tetramerous or pentamerous flowers, epipetalous stamens alternating with the lobes of the tubular corolla, and superior (or in some partly inferior), generally bilocular ovaries most often enclosing axile (sometimes peltate) placentae bearing nu- merous ovules. The seeds contain straight embryos. Most genera, other than Buddleja, its allies, and some others, have internal phloem and vestured pits on the vessel elements. The subdivision, delimitation, and position of the Loganiaceae are long- standing problems. The family has been an artificial “catchall” for difficult-to- place genera. Moreover, it is relatively poor in obvious specialized characters and thereby resembles the ancestral stock from which related families arose. The Loganiaceae also show multiple links between their heterogeneous subunits and other families. Extreme taxonomic views have been to splinter the Lo- ganiaceae into several smaller families (e.g., see Hutchinson) or to apportion its major components among related families (e.g., see Bureau). Modem authors generally place the Loganiaceae in the inconsistently cir- cumscribed order Gentianales (Contortae), with the Apocynaceae, Gentiana- ceae, and Rubiaceae receiving most attention as probable close relatives. The tribe Buddlejeae spans an ordinal boundary in that it shows similarities to the Scrophulariales (Tubiflorae). The Loganiaceae sensu stricto and the Rubiaceae share several characters— interpetiolar stipules, colleters, wood sometimes accumulating aluminum and usually with vestured pits, pollen with similar wall stratification (Cronquist), and nuclear endosperm. The two families are further bound together, along with the Apocynaceae, by their tryptophan-tryptamine indole alkaloids (see Kisakiirek et al). With exceptions, the Rubiaceae differ from the Loganiaceae sensu stricto in their inferior ovaries and lack of internal phloem. (See discus- sion of the tribe Spigelieae below for one narrow gap between the Loganiaceae and the Rubiaceae.) In their study of tryptophan-tryptamine indole alkaloids, Kisakiirek and colleagues found that the Loganiaceae overlap even more with the Apocynaceae than with the Rubiaceae, and that they appear to be less specialized than either of these two families. The Apocynaceae and Loganiaceae both have internal phloem, vestured pits, nuclear endosperm, and colleters. Differences are that the Apocynaceae have a latex system (but note that the loganiaceous Fagraea Thunb. produces latex in fruits), mostly lack stipules, and tend to have both androecial and gynoecial specializations. Similarities between the Gentianaceae and the Loganiaceae include internal phloem, colleters, xanthones, usually nuclear endosperm, and wood that some- times accumulates aluminum (Cronquist; Bisset, 1980a). There are close mor- phological similarities. The Gentianaceae differ in lacking stipules (although interpetiolar flanges occur) and, with exceptions in both families, in having unilocular ovaries with parietal placentae. Buddleja belongs to a collection of about seven genera of uncertain placement. Authors who give this group membership in the order Gentianales either treat 1986] ROGERS, LOGANIACEAE 147 In producing aucubins and related iridoids instead of seco-iridoids, Buddleja resembles the Scrophulariaceae and allies while diverging from several genera of the Loganiaceae and other Gentianales. Flavone 6-hydroxyluteolin and other flavonoids point Buddleja toward a scrophularian position. In a paper in which they documented this, Harbome & Williams used the similar compound scu- tellarein to tie the Plantaginaceae to the Tubiflorae (Scrophulariales) before Marin and colleagues (under Buddleja) isolated a glucoside of scutellarein from B. globosa Hope. Harbome (p. 287) mentioned crocein as “yet another [fla- vonoid] link between Buddleja and the Scrophulariaceae” and listed orobanchin and similar substances in Buddleja, Catalpa (Bignoniaceae), Verbascum (Scrophulariaceae), Orobanche (Orobanchaceae), and Syringa (Oleaceae). Weaknesses in the chemical evidence are that aucubins and similar compounds, as well as 6-hydroxylated flavones, including 6-hydroxyluteolin and scutel- larein, have been detected in disparate families such as the Eucommiaceae, Rosaceae, Rubiaceae, and Compositae quite removed from the Scrophulari- aceae. (For further discussion of the chemical data, see Daniel & Sabnis; Bisset, 1980a; and Jensen et al .) Taxonomically selective pests add further links between Buddleja and the Scrophulariaceae. A leaf-mining fly and two species of weevils thought to attack only the latter transfer to Buddleja, and a species of the fungal genus Peronospo- ra, otherwise restricted to hosts in the Scrophulariaceae, has been found on B. globosa. (The fungus on the Buddleja has been treated as a distinct species, however.) (See Hering; Mohrbutter; Wagenitz, 1959; and Williams, under Bud- dleja, for more on this line of evidence.) According to Cronquist (p. 947), Piechura showed “strong serological affin- ities between the Buddlejeae and several tested families of Scrophulariales.” The irregular pentamerous flowers of Sanango ( Gomara ) with four (nondidy- mous) fertile stamens and a sterile staminode further attracted Cronquist’s attention as a similarity between the Buddlejeae and the Scrophulariales (also pointed out by Bunting & Duke). For discussion of the possible relationship between the Buddlejeae and the Scrophulariales, see Hakki, Hasselberg, Hunzicker & Di Fulvio, Leenhouts (1963), Solereder (1892), Tiagi & Kshetrapal, and Wagenitz (1959); under Buddleja see Soueges. Altogether, the Buddlejeae seem bound to the other Loganiaceae by inter- mediate genera and very weakly by the widespread as well as questionably homologous characteristics of stipules and actinomorphy, while they differ in numerous but overlapping features. Buddleja, and to an uncertain extent the other Buddlejeae, resemble the Scrophulariales in a set of widespread embry- ological characteristics and in chemical traits, all of which are too generalized to be trusted without reservation. Sampling remains disconcertingly skimpy for the difficultly evaluated characters. Students of the Loganiaceae gather Spigelia, Mitreola, Mitrasacme, and sometimes Polypremum together either as an infrafamilial taxon of varying rank or as a separate family. Leeuwenberg & Leenhouts recognized this assem- blage, including Polypremum, as the tribe Spigelieae. With exceptions the fol- lowing characters help to unite the tribe: herbaceous growth; pseudodichoto- ARNOLD ARBORETUM [vol. 67 , LOGANIACEAE 151 331-333, 154 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 ROGERS, LOGANIACEAE 155 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 ARNOLD ARBORETUM [vol. 67 ROGERS, LOGANIACEAE 161 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 t !li LOGANIACEAE 171 question o^placei^ent ^pllypremZi i^LoTanLet^r^Rubiace^6] ' well, F. W. The identity of Hasslerella Chodat. Lilloa 23: 61, 62. 1950. [ FI. Chile 6: 96-101. 1911. [Earlier i 172 JOURNAL OF ' ARNOLD ARBORETUM [vol. 67 LOGANIACEAE 176 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 ROGERS, LOGANIACEAE ROGERS, LOGANIACEAE Diploid (2 n = 38) species of Buddleja grow in Asia, Africa, and America. Numerous tetraploids are present in Asia and America, and probably also on Madagascar and nearby islands. One hexaploid is known from the New World (Norman, pers. comm.), while in Asia there are several hexaploids, as well as two known 12-ploids and one mixed 8-, 16-, and 24-ploid species. Among species from the Old World cultivated in Ottawa, Canada, Moore (1961) found most polyploids to be long-day plants (or to flower late in the season by virtue of requiring a preliminary period of growth) and to have spicate inflorescences. He interpreted both features as evolutionary advance- ments over the short-day requirement and the often shorter, leafier, and oth- erwise presumably less-specialized inflorescences of most diploid species. (Nor- man, 1967, disagreed about evolutionary trends in the inflorescences.) Moore (1961) supposed multiple diploid species to have been ancestral to the Asian polyploids, which he held to have “spread little beyond the point of origin” (p. 278). They are clustered in the Asian area of high species concentration mentioned above (see also Janaki Ammal). Their mutual inclination toward the long-day photoperiod notwithstanding, the polyploids are not the north- ernmost species. As envisioned by Moore (1961), the relatively primitive diploid buddlejas crossed the formerly narrower Atlantic Ocean to the New World under equa- torial short-day conditions and radiated northward secondarily. This accords with Norman’s (1967) hypothesis that the North American species arose from South American ancestors, since the former are most similar to and have morphological “prototypes” among those in South America. (For the opposite view that Buddleja spread from North to South America, see Raven & Axelrod.) In 1967 Moore observed that the New and Old World buddlejas “differ greatly,” and he (1960) was unable to cross members of one group with those of the other. (At least two New World-Old World hybrids have been reported; one of these is probably a backcross involving gametes differing in chromosome number.) Moore and other authors have hybridized even morphologically dis- similar Old World species, and apparent hybridization among American species occurs repeatedly (see comments under individual species in Norman, 1967, 1982). All three of our introduced species are thought to be parents of various hybrids. Artificial crosses in Buddleja range in outcome from reduced fertility to early developmental failure of the F, hybrids. Species of Buddleja are most important to humans as ornamentals and as medicinal plants. All three species in our area were certainly introduced for their decorative value, which they share with several other species, as well as with various hybrids and cultivars. Applications in folk medicine counter a wide variety of ills in both hemispheres. Houghton (see also Bisset, 1980a, 1980b) compiled a survey of medicinal uses focused on eight species and attempted to relate attributed benefits to chemical constituents: remedies for the skin and eyes possibly relate to flavonoids, aucubin (an iridoid), and sa- ponins; respiratory treatments may be based on terpenes or saponins; antiseptic qualities could be related to the aglycone from aucubin; diuresis might result from aucubin or flavonoids; and sedative and analgesic properties are possibly due to alkaloids, although scattered reports of alkaloids in Buddleja remain 182 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 SUGDEN, MARGARITA ISLAND THE MONTANE VEGETATION AND FLORA OF MARGARITA ISLAND, VENEZUELA Andrew M. Sugden1 The island of Margarita (1 1°N, 64°W) is physically and politically a part of Venezuela, constituting most of the state of Nueva Esparta, which also includes the two much smaller islands of Coche and Cubagua. The three islands lie on the continental shelf to the north of the peninsula of Araya, and Margarita is separated from the mainland by a channel approximately 25 km wide. Mar- garita is the most extensive and mountainous of the Venezuelan islands and is the only one to support moist evergreen montane vegetation. The montane vegetation of Margarita has particular ecological interest be- cause of its exceptionally low altitude and its apparent reliance on frequent and regular nocturnal cloud cover for the maintenance of moist conditions— features that are shared by only two other localities (the Serrama de Macuira, on the Guajira peninsula, Colombia; and Cerro Santa Ana, on the Paraguani peninsula, Venezuela) on the Caribbean coast of South America (Sugden, 1982a, 1982b, 1983). Some factors that have been suggested as determinants of the peculiar physiognomy of montane rain forests at higher altitudes— for instance, low temperature and low insolation due to diurnal cloud cover (Leigh, 1975; Grubb, 1977)— are apparently unimportant in the Margarita mountains, while others such as wind have greater significance, as will be shown in this article. The montane forests of Margarita are also of phytogeographic interest due to their insularity and their location between the South American mainland and the arc of the Lesser Antilles. Johnston (1909) collected about 165 species of vascular plants in the mountains, eight of which were apparently endemic. Subsequent collecting has increased the total number of montane species to at least 220, while the number of endemics (due to further collections from else- where) has dwindled to four or five dubious ones. It has been shown by Ortega (1982) and Sugden (1983) (for the pteridophyte and woody floras, respectively) that most of the montane species are quite widely distributed in the Neotropics; not surprisingly, the affinities with the mainland are stronger than those with the Lesser Antilles. I have presented evidence to suggest that the montane vegetation of Margarita is of recent origin (probably less than 10,000 years B.P.), and that the flora owes its present character and composition to random events of long-distance dispersal and establishment (Sugden, 1983). The first account of the montane vegetation of Margarita was written by Johnston (1909). He recorded dense evergreen woods above 300 m, which 'Botany School, Downing Street, Cambridge CB2 3EA, England. Present address: Elsevier Publi- cations, 68 Hills Road, Cambridge CB2 1LA, England. © President and Fellows of Harvard College, 1986. Journal of the Arnold Arboretum 67: 187-232. April, 1986. JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1986] SUGDEN, MARGARITA ISLAND 189 CLIMATE JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 SUGDEN, MARGARITA ISLAND 191 192 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 SUGDEN, MARGARITA ISLAND 193 The principal criteria for distinguishing vegetation types in the upland zone included height of canopy, diameter of boles, and presence and levels of abun- dance of various habits, life-forms, major taxa (deciduous trees, bryophytes, vascular epiphytes, woody lianas, palms, understory shrubs, herbs), and (in some cases) individual species. Except for easily quantifiable characters such as bole diameter, these observations were mainly qualitative but nevertheless detailed. Quantitative characterizations of leaf anatomy in the major vegetation types, and of floristics and physiognomy in one of them, are the subjects of separate articles (Sugden, 1985; Sugden, in preparation). Soils were sampled at ten sites between 400 and 900 m alt. on Cerro Copey. At each site a pit was dug to a depth of ca. 50 cm in order to assess physical characteristics; supplementary information was also obtained from road cuts nearby. Samples for chemical analysis were taken from the top 5 cm and from the upper mineral layer (10-30 cm), the exact depth depending on the site. Analyses were carried out by the Centro de Ecologia, IVIC, Caracas (details of standard procedures available on request). General Remarks The ecotone between the predominantly deciduous vegetation of the lower slopes and the mostly evergreen upland vegetation is generally well defined (Figure 2) and is particularly obvious during the long dry seasons. More often than not, it is marked by the appearance of large trees of Clusia rosea with wide, spreading crowns and also by the appearance of an epiphytic and her- baceous flora that becomes increasingly diverse with higher altitude. Although a few deciduous tree species are present above the ecotone, for convenience I shall use the term “evergreen upland vegetation” to embrace all the physiog- nomic types found above this level. The precise level of this ecotone corresponds closely to the level of the cloud base on each mountain. Thus, evergreen upland vegetation begins at ca. 350 m alt. on the windward slopes and ca. 450 m alt. on the leeward slopes of Cerros Guayamuri, Matasiete, and Tragaplata, and at ca. 400 m alt. on the windward side and 500-550 m alt. on the leeward side of Cerro Copey. On all four mountains thin ribbons (often less than 30 m wide) of “gallery forest” extend downward along the deeper gullies with permanent groundwater, oc- casionally to as low as 100 m above sea level; these are especially well developed on Cerro Copey. The evergreen upland vegetation can be divided into four quite distinct physiognomic types, which can be further subdivided according to physiog- nomic and/or floristic criteria into a total of ten subtypes: 1. Transition Forest a. with Clusia rosea b. without Clusia rosea 2. Cloud Forest a. tall 194 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 SUGDEN, MARGARITA ISLAND 195 3 (i continued ). 8 Rhynchc SCANDENT PLANTS Wulffia ’ HEMIEPIPHYTES :ifolium Bentham Philodendroi i linqulata (L. ) Mez | nonostachya (L. ) Rusby 1 Table 2. vegetation types of Cerro Copey.* Available Total N Cation P ^ug/g) (mg/g) exchange capacity -(mil liequivalents/lOOg )— Wet Thicket 6.85 11.96 5.60 1 5.55 1 Analysis by Centro de Ecologia Table 3. Summary of the physiognomic and floristic characteristics of the vegetation types of Cerro Copey.* 212 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1986] SUGDEN, MARGARITA ISLAND 213 The irradiance in the understory and on the forest floor is greater than under Clusia rosea, and the undergrowth is more dense. Although the lower layers consist mainly of saplings of canopy tree species, there are also a few species of subcanopy trees, notably Myrciaria sp. and Esenbeckia pilocarpoides. Shrubs are rare. Herbaceous cover is sporadic; clumps of the sedge Rhynchospora nervosa occur in shaded gullies, and in places there are dense patches of the spiny bromeliad Aechmea fendleri. Epiphytes, as in the other subtype of Transition Forest, are not abundant. They include the large-leaved Anthurium hookeri, a dwarf form of Aechmea fendleri, and an additional bromeliad, Guzmania monostachya, which is often abundant on exposed knolls close to the lower limit of the Transition Forest. Woody and herbaceous climbers are more frequent than under C. rosea, al- though the species composition is similar. 2. Cloud Forest. Cloud Forest is distinguished from Transition Forest by a far greater abundance of vascular epiphytes, bryophytes, and herbaceous ground cover, by the presence of palms, and by the decreased frequency of deciduous trees. The range of leaf sizes is narrower than in Transition Forest; all the trees have either notophylls or microphylls (Sugden, 1985). Cloud Forest occupies an altitudinal band 250-350 m wide that extends to the summits and ridges on the leeward side of the mountain (Figure 3). The soils vary considerably in depth. On gentle slopes at lower altitudes (600-700 m), the profile may be up to 200 cm deep, while on steeper slopes near the summit the soil depth is usually no more than 40-50 cm and often as little as 20-30 cm. The litter and fermentation layers are more or less continuous but thin (normally less than 2 cm); deeper accumulations of litter and humus are sometimes found between boulders and among emergent roots. There is no distinct zonation in the mineral horizons, except for a gradual change (in the deeper soils) from a dull humic brown to reddish brown with increasing depth. At higher altitudes the mineral horizons become more clayey, and there is an increased concentration of roots in the upper 5 cm of soil, with fine roots often penetrating the litter layer. In the deeper soils, a discontinuous, hard, nodular layer is often encountered at depths of 70-120 cm. Despite the considerable structural variation in the Cloud Forest soils, there is relatively little variation in their chemistry (see Table 2). There are four subtypes of Cloud Forest: a. Tall cloud forest. This is undoubtedly the most species-rich of all the ev- ergreen upland vegetation types, with a comparatively high diversity of all life- forms. Together with pole cloud forest (see below), it occupies the lower part of the altitudinal band of Cloud Forest; its lower limit is contiguous with Transition Forest, and its upper limit rarely exceeds 700 m even on leeward slopes. The height of the canopy ranges from 1 5 to 22 m, with trees attaining girths of up to 200 cm. Tall cloud forest is best developed in gullies sheltered from the wind. Most of the tree species and individuals are evergreen, although the deciduous Croton xanthochloros and Tabebuia chrysantha (particularly the former) are frequent. The most abundant tree species are Licania membranacea, Guapira 214 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 215 2 1 8 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 previous Cloud Forest subtypes; boles are generally vertical, but a few lean by up to 20°. Floristic differences between lower palm forest and the taller Cloud Forest subtypes are few. In the former, Croton xanthochloros and Bactris falcata are absent or nearly so. Guapira cf. fragrans, Myrcia cf. coriacea, Maytenus kar- stenii, and Ilex guianensis continue to be important components of the canopy. The understory is generally denser than in the taller Cloud Forests, chiefly due to the abundance of Psychotria muscosa\ saplings of canopy trees are common but less abundant. The subcanopy tree Esenbeckia grandiflora is commoner here than on any other part of the mountains. Large and small herbs are less well represented, in both diversity and abun- dance. Aechmea fendleri and Maranta arundinacea, in particular, are uncom- mon, although Adiantum tetraphyllum occurs frequently. The understory and substrate are permanently damp, and the exposed boulders, roots, and fallen boles have a covering of mat- and cushion-forming bryophytes, especially Leucobryum giganteum C. Mueller and L. antillarum Schimper ex Besch. The vascular epiphytic flora is similar in diversity and abundance to that of the tall cloud forest, except that the woody hemiepiphytes are more frequent. Epiphytic bryophytes, as on the substrate, are better developed than in the previous subtypes. Woody climbers, by contrast, are rare, d. Upper palm forest. The last of the Cloud Forest subtypes, which extends to the summit of Cerro Copey on the leeward side, is distinct from lower palm forest in its lesser stature (6-9 m) and in the inclination of the trees (see Figure 8). Most boles are strongly (20-90°) inclined downhill, and they are frequently twisted and gnarled, branching at all levels in the forest profile. As a result of this habit and the generally boulder-strewn terrain, progress through upper palm forest is far more difficult than through the other forest types. Neverthe- less, the principal components of the canopy are very similar to those of the lower palm forest, the only notable absentees being Licania membranacea and Esenbeckia grandiflora. Individuals of Euterpe karsteniana have more massive (up to 1 m tall and 70 cm across at the base) stilt-root systems and often have up to six shoots of canopy height as well as numerous juvenile shoots among The understory is quite different from that of the lower palm forest. Shrubs are almost absent, and the density of young saplings is low. Adiantum tetra- phyllum is present, although at a relatively low density, while Aechmea fendleri occurs more frequently than on any other part of the mountain, forming ex- tensive spiny groves that contribute to the general impenetrability of the forest. Abundant scandent shoots of Scleria bracteata, with minutely serrated leaves and sheaths, supply an additional irritant and impediment to progress through the forest. Small herbs are few, represented chiefly by Ichnanthus sp., Spiranthes adnata, and Elaphoglossum sp. The boulders, lower boles, and exposed roots are covered with thick cushions of Leucobryum giganteum and L. antillarum-, as in the lower palm forest, the understory is perpetually damp except in places where the canopy is thin and broken. The epiphytic flora is notable for a much greater abundance of woody hemi- epiphytes (e.g., Hillia parasitica, Clusia cf. flam ), which often form thick, tangled masses in the upper canopy. The flora of herbaceous epiphytes is similar in composition to that at lower altitudes, except for an increased frequency of the large bromeliad Glomeropitcairnia erectijlora. Woody climbers are entirely absent. 220 JOURNAL OF THE ARNOLD ARBORETUM [vol. 6 7 3. Wet Thicket. Wet, windswept thickets less than 4 m tall dominate the windward slopes of Cerro Copey from ca. 800 m to the summit. They are distinguished from upper palm forest by their lesser stature, the absence of palms, and the reduced diversity and density of herbaceous epiphytes and bryophytes. The woody canopy plants are mostly microphyllous, and their leaves are significantly thicker than those of their counterparts in the Cloud Forest (Sugden, 1985). The soils of the thickets are quite shallow— usually less than 40 cm deep. The litter layer is discontinuous, and the fermentation layer is barely discernible except where pockets of humus accumulate between boulders and large su- perficial or raised roots. Normally the top 1 5-20 cm is a coherent, wet, pale red-brown silty clay containing fragments of rotting leaves. The remainder of the profile, which gradually gives way to rock fragments and finally to bedrock at 30-40 cm depth, is more silty, more readily fragmented, and pale brown. Roots are sparse throughout, in contrast to their distribution in the forest soils, and wormcasts provide evidence of considerable animal activity. Although slightly more acidic (pH 4.95-5.70) than the Cloud Forest soils, the Wet Thicket soils display no chemical properties markedly different from those of the other upland soils (see Table 2). There are three subtypes of Wet Thicket: a. Tall thicket. This vegetation, which has a lower limit contiguous with upper palm forest on the windward slopes, is characterized by a canopy 3-4 m tall, an absence of boles more than ca. 10 cm in diameter, and an abundance of Clusia cf. flam (see Figure 9). Most of the woody species of the Cloud Forest are present in stunted form, but C. cf. flam is clearly the dominant species, forming extensive pure stands. Blakea monticola is also very common. Tall thicket has no discernible understory but is almost impenetrable due to the tendency of C. cf. flam to retain its dense branches at all levels. A hemiepiphyte, this species often has no discernible bole, and individuals are frequently much broader than tall. There are a small number of saplings, no shrubs, and few herbaceous species apart from Scleria bracteata. Herbaceous vascular epiphytes occur infrequently compared to the levels observed in the Cloud Forest; one of the common species in the tall thicket is the endemic orchid Epidendrum johnstonii. Bryophytes are much less abundant than in the palm forests, in spite of the permanent wetness of the boles and lower branches; generally, there is only a thin covering of creeping mosses and slimy algae. Tall thicket occurs on all the slightly less exposed parts of the upper windward slopes of Cerro Copey and is also present in a narrow band a few meters wide along the ridges dividing the windward slopes from the leeward, where it forms the ecotone between the short thickets (see below) and the upper palm forest. b. Short woody thicket. This vegetation differs from tall thicket in its lesser stature (1-2 m), its more even abundance of woody species, and its greater profusion of large herbs. Clusia cf. flava is still very common, but markedly stunted forms of Cloud Forest canopy tree species are much more frequent than in the tall thicket (see Figure 10). The large tank bromeliad Glomero- pitcairnia erectiflora is a very prominent feature of this vegetation, usually 1986] SUGDEN, MARGARITA ISLAND 221 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1986] SUGDEN, MARGARITA ISLAND 223 1 1 . Grassland at 900 m on Cerro Copey. 224 ARNOLD ARBORETUM [vol. 67 SUGDEN, MARGARITA ISLAND 225 SUGDEN, MARGARITA ISLAND JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 SUGDEN, MARGARITA ISLAND 230 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 Drs. Peter Grubb and I 231 Juana, C., & M. Vignali. 1972. Rocas metam6rficas e igneas en la de Macanao, Margarita, Venezuela. Mem. VI Conf. Geol. del Caribe, pp. 232 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1986] HOWARD & KELLOGG, MICONIA 233 NOMENCLATURAL NOTES ON MICONIA (MELASTOMATACEAE) R. A. Howard and E. A. Kellogg1 234 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1986] HOWARD & KELLOGG, MICONIA 235 Leaves dentate, denticulate, or serrate. 14. Upper intemodes mostly < 2 cm long; plants dioecious; M. globuliflora var. globuliflora. 15. Branches scurfy to setose at nodes M. globuliflora var. vulcanica. 14. Upper intemodes > 2 cm long; flowers perfect; bracts 16. Leaves consistently 3-nerved; flowers with calyx ca- lyptrate, rupturing irregularly, petals sparsely stellate- puberulent abaxially M. cornifolia. 1 6. Leaves generally 5 -nerved; flowers with calyx valvate, petals glabrous. 18. Flowers secund along panicle branches. . . . M. laevigata. 18. Flowers glomerulate at tips of panicle branches M. acinodendron. 17. Flowers pedicellate. 19. Leaf surfaces concolorous M. striata. 19. Leaves much darker above than below. . . . M. andersonii. NAMES PUBLISHED FOR LESSER ANTILLEAN TAXA OF MICONIA AND MELASTOMA Miconia abortiva Naudin, Ann. Sci. Nat. Bot. Ill, 16: 232. 1851. Type: Guade- loupe, on La Soufrifcre, L’Herminier s.n. (holotype, p! (staminate)). Co- gniaux (1888) referred this species to the synonymy of M. globuliflora. = Miconia globuliflora var. vulcanica (Naudin) R. Howard & E. Kellogg (see below). Miconia acinodendron (L.) Sweet, Hortus Brit. ed. 1. 159. 1826. Basionym: Melastoma acinodendrum L. Sp. PI. 1: 389. 1753. Lectotype (here designated): Linnaeus, Hortus Cliffort. 162. 1737 (photocopy!). In the Lesser Antilles this species occurs only on Martinique. We have the following records: Antilles, Martinique, Plie 740 (p!); near forestry post JOURNAL OF THE ARNOLD ARBORETUM [vol. 6 7 ■toma acuminata Desr. in Lam. Encycl. Meth. Bot. 4: 54. 9 Feb. 1797, s Sw. Prodr. 70. 1788. Lectotype ( , BM, GH, NY, P, Or US. It IS J , 1972). ?ua (Rich.) DC. Veg. 17: 162. 1920. that Box 727 and, on Box 1526 (bm) reads, “A portion of this specimen has been compared HOWARD & KELLOGG, MICONIA 237 with the type of Miconia antiguensis Urban, in New York Bot. Gard., and found to agree exactly. Harold E. Box 24/8/39.” = Miconia striata (M. Vahl) Cogn. Melastoma aquatica Aublet, Hist. PI. Guiane Fr. 1: 430. t. 169. 1775. Type: [sur la route de Cai'enne & Couru,] bm, linn-sm 654.1 1, IDC 5073. 302: II. 2 (photo!) (see also Howard, 1983). = Nepsera aquatica (Aublet) Naudin. Miconia berbiceana (DC.) Naudin, Ann. Sci. Nat. Bot. Ill, 16: 174. 1851. Basionym: Clidemia berbiceana DC. Prodr. 3: 160. 1828. Type: Berbice, Anonymous s.n. (holotype, g-dc), IDC 800. 507: III. 7 (photo!). Homotypic synonym: Tschudya berbiceana (DC.) Griseb. FI. Brit. W. Indian Is. 250. 1860. This was originally described as from Guyana. Naudin cited Plee 740, Martinique (p!), but this is a specimen of M. acinodendron. Cogniaux assigned the name to the synonymy of Miconia acinodendron (L.) Sweet. Melastoma brachiata M. Vahl, Eclogae Amer. 3: 24. 1807. Syntypes: Mont- serrat, Ryan 2 et 7 (holotype, c!), IDC 2201. 46: I. 6, 7. = Miconia laevigata (L.) D. Don. Melastoma calyptrata Desr. in Lam. Encycl. Meth. Bot. 4: 51. 9 Feb. 1797. Type: “Vahl 1797 Guadeloupe” and “vide specimen in Surian herb. 177.. 752,” p-ju #14045, IDC 6206. 1033: II. 4 (photo!). Homotypic synonym: Melastoma calyptrata M. Vahl, Eclogae Amer. 1: 40. March 1797. = Conostegia calyptrata (Desr.) D. Don. Miconia candolleana Naudin, Ann. Sci. Nat. Bot. Ill, 16: 244. 1851, nomen illegit. This is a renaming of Melastoma lutescens M. Vahl. = Miconia cornifolia (Desr.) Naudin. Miconia candolleana Triana, Trans. Linn. Soc. London 28: 117. 1871, nomen illegit. This is a renaming of Miconia cinnamomifolia (DC.) Naudin, not M. cinnamomifolia (Jacq.) Triana. See discussion under M. cinnamomi- folia. Wurdack (1960) mistakenly cited this as M. candolleana (DC.) Triana. Melastoma capitata M. Vahl, Eclogae Amer. 1: 45. 1797. Type: [India occi- dentali, Schumacher, Professor Chirurgiae Hauniensis,] Dr. Banks [ex] Dryander (holotype, c!), IDC 2201. 66: II. 6, 7. Triana (1871) placed this species in Tristemma, but Wickens (1975) assigned it to Melasto- mastrum. Wickens believed that the plant may have been grown at Kew by Dryander; a specimen was obtained by Schumacher and sent to Vahl, who erroneously reported it from the West Indies. = Melastomastrum capitatum (M. Vahl) A. & R. Fernandes, an African species. Miconia caribaea Domin, Preslia 9: 3, 4. 1930. Type: St. Vincent, Sandy Bay, H. H. & G. W. Smith 820 (holotype, k!). = Miconia andersonii Fawcett & Rendle. Miconia catalpifolia Kranzl, Vierteljahrsschr. Naturf. Ges. Zurich 76: 153. 238 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1931. Type: Dominica, 700 m, Eggers 750 (holotype, b, destroyed; isotype, gh!). = Graffenriedia latifolia (Naudin) Triana. Miconia ceanothina DC. Prodr. 3: 189. 1828. Syntypes: Guadeloupe, Perrottet s.n. (g-dc), IDC 800. 515: I. 6-8; Cuba, De la Ossa s.n. (g-dc?, not seen). = Miconia laevigata (L.) D. Don. Miconia christophoriana (Ham.) DC. Prodr. 3: 200. 1828 [as christophoria- num\. Basionym: Melastoma christophoriana Ham. Prodr. PI. In- diae Occ. 37. 1825. Type: “St. Kitts, v.v.” No specimen located (see Howard et al , 1981). Cogniaux (1891) referred this name to Tetrazygia discolor (L.) DC. Melastoma ciliata Desr. in Lam. Encycl. M6th. Bot. 4: 49. 1797. Type: Gua- deloupe, Badier s.n. (holotype, p, not found— no specimen in p-ju or p-lam). = Tibouchina ornata (Sw.) Baillon (see Howard, 1972). Miconia ciliata (Rich.) DC. Prodr. 3: 179. 1828. Basionym: Melastoma ciliata Rich. Actes Soc. Hist. Nat. Paris 1: 109. 1792. (Not Desr., 1797, which is Tibouchina ornata (Sw.) Baillon.) Type: French Guiana, Leblond s.n. (holotype, p!). Homotypic synonym: Miconia racemosa var. ciliata Griseb. FI. Brit. W. Indian Is. 258. 1860. Cogniaux (1891) cited an Anderson collection from St. Vincent, where it was presumably in cul- tivation. Anderson 18 (bm!), labeled “Ind. Occ.,” was actually collected in Surinam. Naudin (1851) listed this species as a synonym of M. ra- cemosa, but Wurdack (1973) distinguished the two on the basis of leaf nervation— M. ciliata has three nerves, M. racemosa five. Miconia cinnamomifolia (DC.) Naudin, Ann. Sci. Nat. Bot. Ill, 16: 168. 1851. Basionym: Cremanium cinnamomifolium DC. Prodr. 3: 194. 1 828. Type: Brazil, Martius herb. #13 ( fide Triana, 1871). Homotypic synonym: Miconia candolleana Triana, Trans. Linn. Soc. London 28: 117. 1871. Miconia cinnamomifolia (DC.) Naudin is the legitimate combination in Miconia of the epithet cinnamomifolia (see below). Miconia cinnamomifolia (Jacq.) Triana, Trans. Linn. Soc. London 28: 101. 1871, nomen illegit. Basionym: Melastoma cinnamomifolia Jacq. Col- lect. 2: 111./. 6, fig. 2. 1788. Type: [Martinica,] Jacquin’s plate. This is an illegitimate transfer of the epithet into Miconia. = Miconia cornifolia (Desr.) Naudin. Melastoma coccinea Rich. Actes Soc. Hist. Nat. Paris 1: 109. 1792. Type: Guadeloupe, Richard s.n. (holotype, p!). = Charianthus coccineus (Rich.) D. Don = Charianthus alpinus (Sw.) R. Howard (see Howard, 1972). Melastoma coccinea M. Vahl, Eclogae Amer. 1: 48. 1797. Type: Montserrat, Ryan s.n. (holotype, c!), IDC 220 1 . 46: II. 5,6. = Charianthus purpureus D. Don. Melastoma compressa M. Vahl, Eclogae Amer. 3: 23. 1807. (Not Miconia compressa Naudin, a Brazilian species.) Type: “in Caribeis,” Rohr s.n. (holotype, c, not seen). = Miconia mirabilis (Aublet) L. O. Williams. 1986] HOWARD & KELLOGG, MICONIA 239 Miconia coriacea (Sw.) DC. Prodr. 3: 189. 1828. Basionym: Melastoma cori- acea Sw. Prodr. 70. 1788. Type: Ind. Occid., Swartz s.n. (holotype, bm, not found; isotype, s!). Although Swartz noted in his initial description that this species was to be found on Montserrat, he did not repeat the mention of Montserrat in his later publication (1800). Instead, he cited the Soufridre on Guadeloupe and indicated that a specimen was in the Banks Herbarium (bm). We were unable to find the specimen there. We have seen specimens only from Guadeloupe and Dominica; reports from Montserrat and Martinique may be erroneous, referring to the similar M. globuliflora and its varieties. Melastoma cornifolia var. grandifolia Rich, in Bonpl. Monogr. Melast. 94. 1812. Type: Martinique, [Bonpland s.n.] (holotype, p!). = Miconia cor- nifolia (Desr.) Naudin. Melastoma cornifolia var. parvifolia Rich, in Bonpl. Monogr. Melast. 94. 181 2. Type: Martinique, Bonpland s.n. (holotype, p!). = Miconia cornifolia (Desr.) Naudin. Miconia cornifolia (Desr.) Naudin, Ann. Sci. Nat. Bot. Ill, 16: 126. 1851. Basionym: Melastoma cornifolia Desr. in Lam. Encycl. Meth. Bot. 4: 51. 1797. Type: in the general collection at p is a specimen labeled “e martinica no. L61,” with “Lamk. scrips” in the comer of the label. We presume this to be the holotype. Homotypic synonyms: Conostegia cornifolia (Desr.) Ser. ex DC. Prodr. 3: 175. 1828. Tetrazygia cornifolia (Desr.) Griseb. FI. Brit. W. Indian Is. 255. late 1860 (almost simulta- neously published in Goett. Abh. 9: 53. 14 Dec. 1860). The calyx of this species is frequently calyptrate, explaining its placement in Tetrazygia and Conostegia. Melastoma corymbosa Rich. Actes Soc. Hist. Nat. Paris 1: 109. 1792. Type: Guadeloupe, Richard s.n. (holotype, p!). = Charianthus corymbosus (Rich.) Cogn. Melastoma corymbosa M. Vahl, Eclogae Amer. 3: 25. 1807. Type: Martinique, Rohr 168 (holotype, d), IDC 2201. 46: II. 7, III. 1. = Charianthus corymbosus (Rich.) Cogn. Melastoma crinita M. Vahl, Eclogae Amer. 3: 28. 1807. Type: Martinique, no collection cited. = Charianthus nodosus (Desr.) Triana ( fide Cogniaux, 1891). Melastoma cymbifolia Rich, in Bonpl. Monogr. Melast. 57. t. 26. 1809, nomen illegit. Type: Sulphurariae Guadlupensis, Richard s.n. (holotype, p! (in fruit)). This is a renaming of Melastoma coriacea. = Miconia coriacea (Sw.) DC. Melastoma decussata M. Vahl, Eclogae Amer. 1: 42. 1797, nomen illegit. Syntypes: Rohr 119 (c!), IDC 2201. 46: III. 4, 5; Ryan [&] Rohr ex Cajenna N. 119 (c!). This is a renaming of Melastoma racemosa Aublet. = Miconia racemosa (Aublet) DC. JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 ? Sw. Prodr. 72. 1 788 [as aeleagnoides ], not Cogn. Type: [Caienne,] bm, linn-sm 782.3 1 . IDC 5073. 347: III. 4 (] ard, 1983). = Clidemia hirta (L.) Don var. elegans (A 1775. Typ 241 I (M. Vahl) Griseb. FI. Brit. W. >nym: Melastoma furfuracea M. Vahl, Eclogae i . Type: [India occidentalis,] Ziegler s.n. (holotyi 242 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 /ARD & KELLOGG, MICONIA 243 L. Sp. PI. 1: 390. 1753. Type: oides (L.) Triana. (1926) 245 559.9 is another also bears the letters ‘C and ‘Br.’ ‘Discolor’ has been crossed out, but by Smith who suggests that it is really laevigata. This is because ‘C is correlated with M. laevigata L. in Syst. nat. ed. 10 and because the (Desr.) DC. De Can- 1816, M. Lambert s.n. (g-dc), IDC 800. 508: II. 8 (photo!); Cuba, Ha- vana, 1825, Ossa s.n. (g-dc), IDC 800. 508: II. 6 (photo!); Trinidad, Sieber 66 (g-dc), IDC 800. 508: II. 7 (photo!). Homotypic synonyms: Tschudya lanata (DC.) Griseb. FI. Brit. W. Indian Is. 250. 1860. Me- Miconia lanceolata (Desr.) DC. Prodr. 3: 190. 1828. Basionym: Melastoma lanceolata Desr. in Lam. Encycl. Meth. Bot. 4: 43. 1797. Type: S. Do- ISflllS 1986] HOWARD & KELLOGG, MICONIA 247 248 JOURNAL OF THE ARNOLD ARBORETUM [vol, 67 parviflora Bentham (Type: British Guiana, Schomburgk483, k). Gleason (1932) accepted this name, but Wurdack (1973) has correctly designated the northern South American species as Miconia pubipetala Miq. = Miconia prasina (Sw.) DC. Melastoma pendulifolia Rich. Actes Soc. Hist. Nat. Paris 1: 109. 1792. Type: see discussion. Triana (1871) added this name to the synonymy of Miconia prasina (Sw.) DC. The type should be from Guyana. There is a specimen at p in herb. Richard, referable to M. prasina, but the location is given as “in sylvulis ridivivis Guadl.,” the same citation as for Me- lastoma pendulifolia Bonpl. (1809); however, the illustration for the latter species appears to be a specimen of M. laevigata (L.) DC. and is certainly not M. prasina. Triana (1871) placed M. pendulifolia Bonpl. in the synonymy of M. laevigata. Either M. pendulifolia Bonpl. is illus- trated by a specimen that is not the type and is properly typified by the Richard specimen, or the type specimen has not been found and the species should be lectotypified on the illustration. The specimen from Guadeloupe is probably not the type of M. pendulifolia Rich., although no other candidate for the type has been found. Miconia pendulifolia DC. Prodr. 3: 187. 1828. Type: Bonpland’s plate may be the best type for this species, unless a specimen can be found (see above). Synonym: Melastoma pendulifolia Bonpl. Monogr. Melast. 79. t. 35. 1809, nomen illegit. = Miconia laevigata (L.) D. Don. Miconia picta Wurd. Phytologia 7: 235. 1960. Syntypes: Montserrat, Ryan 3 (c!), IDC 2201. 47: III. 4, 5; Ryan s.n. (c!). Synonym: Melastoma picta M. Vahl, Eclogae Amer. 3: 15. 1807, nomen illegit. This is a renaming of Melastoma cornifolia Desr. = Miconia cornifolia (Desr.) Naudin. Melastoma pilosa Sw. Prodr. 72. 1788. Type: Jamaica, Swartz s.n. (holotype, s, not seen). = Clidemia guadalupensis Griseb. Melastoma piluhferum M. Vahl, Eclogae Amer. 3: 1 5. 1 807. Type: Martinique, Rohr 102 (holotype, c!). = Miconia laevigata (L.) DC. Miconia prasina (Sw.) DC. Prodr. 3: 188. 1 828. Basionym: Melastoma prasina Sw. Prodr. 69. 1788. Lectotype (here designated): Jamaica, Swartz s.n. (s!). Homotypic synonym: Acinodendron prasinum (Sw.) Kuntze, Rev. Gen. PI. 1: 245. 1891. Swartz (1800) cited Melastoma laevigata Aublet, Hist. PL Guiane Fr. 1: 171, 443. 1775, but this is not Melastoma lae- vigata L. Fawcett and Rendle (1926b) referred to Swartz specimens from Hispaniola (bm!) and Jamaica (s!) without designating a lectotype. Miconia pteropoda Bentham, J. Bot. (Hooker) 2: 314. 1840, not Naudin, 1851. Type: British Guiana, Schomburgks.n. (holotype, k). Urban (1921) cited the distribution to include Guadeloupe and Grenada. No specimens have been seen from outside northern South America. Specimens at k ( Broadway 5898, Grenada; and Grisebach s.n., Guadeloupe) are labeled M. pteropoda but are really M. prasina. According to Wurdack (1973), M. pteropoda = Miconia prasina var. crispula Cogn. V, Lit.-Ber. 90,” 2(3): 87. 1949. This is a xtoma purpurascens Aublet, Hist. PI. Guiane Fr. 1: 402. t. 154. 1775. Type: bm, linn-sm 782.38, IDC 5073. 348: I. 4 (photo!) (see Howard, 1983). = Aciotis purpurascens (Aublet) Triana. 250 JOURNAL OF THE ARNOLD ARBORETUM [yol. 67 endemic to Jamaica by Adams (1972). However, Wurdack (1973) also reported it from Venezuela, and Gleason (1958) recorded it for Costa Rica and Panama. Grisebach cited this species from Guadeloupe, pre- sumably on the basis of his accepted synonym of Miconia abortiva Naudin, which is Miconia globuliflora (Rich.) Cogn. var. vulcanica (Nau- din) R. Howard & E. Kellogg. Melastoma rubra Aublet, Hist. PI. Guiane Fr. 1: 416. t. 161. 1775. Type: bm, linn-sm 782.34, IDC 5073. 347: III. 7 (photo!) (see also Howard, 1983). The only record from the Lesser Antilles is Proctor 16951 from Grenada. = Clidemia rubra (Aublet) C. Martius. Miconia secunda R. Howard & E. Kellogg, sp. nov. Frutex vel arbor parva, foliis ovatis vel ellipticis, basi cordatis vel rotundatis, quinquinervis, praeter venas abaxiales glabris. Rami infio- rescentiarum Cincinnati, floribus secundis, sessilis. Antherae uniporatae, l. 2-2 mm. Stigmata expansa. Species differt a Miconia trichotoma flo- ribus secundis. Small tree or shrub, 5-12 m tall; young branches compressed, tetrag- onal, appressed stellate-furfuraceous as are petioles, inflorescence axes, bracts, and calyces. Leaves with petiole (9-) 14-54 mm long; blade ovate to elliptic, 10.5-27 x 4.9-16.2 cm, the apex short-acuminate, the base cuneate, rounded, or cordate, the margin entire, the nerves 5, basal or inner 2 suprabasal, with 2 intramarginal ones weaker than other 3, the leaf surfaces glabrous except primary abaxial veins stellate-furfuraceous. Inflorescences panicles, 10-21 .5 cm long, ultimate branches cincinnate; flowers sessile, secund, scented; bracts lanceolate below to deflate or broadly ovate above, decreasing in size acropetally from 1.5 to 3 mm long. Calyx conical, 1.5-2. 5 mm long, stellate-pubescent, obscurely 5-crenate, ca. upper half free from ovary, forming hypanthium; petals oblong-oblanceolate, clavate, 1.3-2 mm long, white; anthers flaring, truncate, 1.2-2 mm long, uniporate, extended basally to 2 minute ap- pendages; pistil with the style linear, 2. 3-4. 7 mm long, glabrous, the stigma more or less flaring, papillate. Berry 3. 5-3. 9 mm in diameter, pink, becoming black, 10-ribbed when dry. Type. St. Lucia, Savanne Edmund district, southeast of Piton Trou- masse, elev. 1800-2000 ft, 22, 23 April 1958, Proctor 17726 (holotype, a; isotype, bm). Specimens seen. St. Lucia: Bane de lisle, Castries-Dennery Road, alt. 350-400 m, Cowan 1559 (oh, p), 1575 (gh, p); La Sorciere, Castries Waterworks Reserve, alt. 280 m, Rollet HUC 1324 (a); Castries Waterworks Reserve along track E at Slane 228 (a). ? ’ , P 0 , exposure N, clayey forest loam, Wurdack (1965) had noted that this was probably a new species. Miconia semicrenata (Rich.) D. Don in Loudon, Hortus Brit. 1: 174. 1830. Basionym: Melastoma semicrenata Rich, in Bonpl. Monogr. Melast. 69. t. 31. 1809. Type: Guadeloupe, Richard s.n. (holotype, p!). Homotypic HOWARD & KELLOGG, MICONIA 251 252 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 trandrum (Sw.) Griseb. Mem. Amer. Acad. Arts 8: 166. 1861. Although D. Don did not cite Swartz as the author of the epithet, he listed the place of origin as Jamaica and the date of introduction as 1815, infor- mation that is repeated in Loudon (1830), along with a reference to Swartz. Miconia theaezans (Bonpl.) Cogn. in C. Martius, FI. Brasil. 14(4): 419. 1888. Basionym: Melastoma theaezans Bonpl. Monogr. Melast. 17. t. 9. 1807. Type: Colombia, Popay&n, Bonpland s.n. (holotype, p, not seen). A Masson specimen determined as this species and supposedly collected on St. Kitts is at bm (photo at a!). The specimen may actually be a representative of Miconia globuliflora var. dominicae. Miconia theae- zans is otherwise unknown from the Lesser Antilles. Triana (1871) placed this species in the synonymy of M. milleflora Naudin. Miconia thyrsiflora (D. Don) Naudin, Ann. Sci. Nat. Bot. Ill, 16: 244. 1851. Basionym: Cremanium thyrsiflorum D. Don, Mem. Wem. Nat. Hist. Soc. 4: 312. 1823. Type: Peru, Pavon s.n. (holotype, g-boiss; isotype, oxf). This species is Peruvian and has nothing to do with the Caribbean species Melastoma thyrsoidea M. Vahl, but the names have been con- fused by some authors. Naudin (1 85 1 , p. 244) could not place this species but referred it to Miconia as “ Miconia thyrsoidea citing “ Cremanium thyrsoideum Don” and DC. Prodr. 3: 191. 1828, where the species is also called C. thyrsoideum Don. Wurdack (1971) believed that Naudin should be credited with the combination. Cogniaux (1891, p. 934) used Miconia thyrsoidea, attributing a “nomen” to Naudin. Cogniaux did not see the Pavon specimen. He listed Cremanium thyrsoideum D. Don in synonymy, perpetuating Naudin’s error, and also Miconia thyrsiflora ment of Miconia for the Flora of Peru (1941) used Miconia thyrsoidea (Don) Naudin, noting that the species had not been re-collected and the locality was to be questioned. An f negative (29507) was cited. At gh there is a photograph of a Pavon specimen in Berlin ( Gleason 7-2) considered a type collection; Wurdack (1971) assigned this to Miconia nitida (D. Don) Naudin. Melastoma thyrsoidea M. Vahl, Eclogae Amer. 3: 13. 1807. Type: Montserrat, Ryan 9 (holotype, c!), IDC 2201. 48: I. 5, 6. Not Miconia thyrsoidea (D. Don) Cogn., which is Miconia thyrsiflora (see above). = Miconia globuliflora (Rich.) Cogn. var. globuliflora. Miconia trichotoma (Desr.) DC. Prodr. 3: 188. 1828. Basionym: Melastoma trichotoma Desr. in Lam. Encycl. Meth. Bot. 4: 53. 1797. Type: Antilles, Herbier de Surian (holotype, p-ju #14084!). Melastoma triflora M. Vahl, Eclogae Amer. 1: 46. 1797, not Naudin, 1851 (a species from Goa, India, that may need a new name). Type: Caribeis, Forseiths.n. (holotype, d), IDC 2201. 48: 1. 7, II. 1. De Candolle (1828) listed this as an unplaced species. = Henriettea triflora (M. Vahl) Alain. HOWARD & KELLOGG, MICONIA (holotype, d). Most species in Forsyth’s herbarium were sent to him from St. Vincent by Anderson. = Ctidemia verticillata (M. Vahl) DC. (cl), IDC 2201. 48: II. 2, 3. Louis ^ We would like to thank C. E. Jarvis for his help with the typification of some of the Linnaean names, D. Michener and E. A. Shaw for patiently testing the LITERATURE CITED 254 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 & . 1926b. Miconia. Flora of Jamaica. Yol. 5(3). Pp. 367-379. 1: 127-184. , & H. Uittien. L 37: 133-170. :. 13(4, no. 1): 341- . Vol. 2. Pp. 77-91. L 255 SPONGBERG, SORBUS 257 NOTES ON SIMPLE-LEAVED CHINESE SPECIES OF SORBUS (ROSACEAE) FROM HUBEI PROVINCE i (Bartholomew et al, 1983). The determinations listed in 258 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 glabrous or with areas of weblike, floccose pubescence. Fruits 10-16 mm long, SPONGBERG, SORBUS 259 ser. Thibeticae Yil of sect. Aria Pers. Represented by two additional species in the Hubei flora, ser. Thibeticae includes S. hemsleyi (Schneider) Rehder, from western Hubei and Sichuan, and S. zahlbruckneri Schneider, from western Hubei and eastern Sichuan. Most specimens of S. yuana have previously been JOURNAL OF ' [vol. 67 linking the flora of western Hubei with that of Japan and reinforces the hy- , and Chang (1979) that the v \ of Henry 6830A at by T. T. Yu as early as 1950, as his annotation label on the LITERATURE CITED 262 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 Ying, T. S., C. G. Ma, & C. S. Chang. 1979. Observations of the flora and vegetation of Mt. Shennungia in western Hupeh, China. Acta Phytotax. Sin. 17(3): 41-60. (In Chinese, English summary.) Yu, T. T., & K. C. Kuan. 1963. Taxa nova Rosacearum Sinicarum (I). Acta Phytotax. Sin. 8: 214-234. pis. 26, 27. & L. T. Lu. 1974. Sorbus. Pp. 283-344 in Flora Reipublicae Popularis Sinicae. Vol. 36. Science Press, Beijing. AUSTIN & STAPLES, CONVOLVULACEAE 263 FURTHER NOTES ON TURBINA AND MERREMIA: TYPIFICATION AND TAXONOMY OF THE NEOTROPICAL CONVOLVULACEAE D. F. Austin1 and G. W. Staples2 TURBINA as discussed in our previous paper (Austin & Staples!^ 8 3). We^houghTwe 264 JOURNAL OF THE ARNOLD ARBORETUM The Pohl collections cited above as syntypes of Rivea cordata are without collector’s number, although two bear numbers added later that may be her- barium or catalog numbers. All of the Pohl sheets are morphologically identical, and we consider them to be duplicates of one collection. MERREMIA Among the material we examined in the collections of the Conservatoire et Jardin Botanique, Geneva, was the holotype of Hans Hallier’s Operculina pavoni, a taxon we have never been able to place satisfactorily. This taxon is identical with two fragmentary specimens at the Field Museum, the basis for our Merremia species A. A new combination is needed. Merremia pavonii (H. Hallier) Austin & Staples, comb. nov. Operculina pavoni H. Hallier, Bot. Jahrb. Syst. 16: 550. 1893. Type: Nova Hispania, Pavon s.n. [354] (g!, Boissier herb., photos at a, fau). The origin of the holotype specimen is obscure, and we are not convinced that this collection came from the Andes. Ruiz and Pavon collected in Peru and Chile, but their herbarium also incorporated material obtained from other collectors, including Sess6 and Mocino, who worked in Mexico. Morphologi- cally, Merremia pavonii is nearest to the species complex consisting of M. platyphylla, M. palmeri, and M. aurea, which range through southwestern Mexico and the southern Baja peninsula. No Andean species familiar to us is as similar to M. pavonii as this Mexican group is, which leads us to suspect that the holotype originated in Mexico. It should be noted, however, that among the arborescent members of Ipomoea, morphologically similar species are known from Mexico and the Andes (McPherson, 1981). This allows for the possibility of a parallel situation in Merremia. Additional collections of this species are needed to clarify the situation. ACKNOWLEDGMENTS We are deeply grateful for the support provided to the junior author by the Atkins Fund and the Anderson Fund, Harvard University. These funds per- mitted travel and herbarium research in England and Europe, which resolved several vexing taxonomic problems. LITERATURE CITED Austin, D. F., & G. W. Staples. 1983. Additions and changes i 64: 483-489. 1 1. The Journal of the Arnold Arboretum April, 1986 CONTENTS OF VOLUME 67, NUMBER 2 The Genera of Loganiaceae in the Southeastern United States. George K. Rogers 143-185 The Montane Vegetation and Flora of Margarita Island, Venezuela. Andrew M. Sugden 187-232 Nomenclatural Notes on Miconia (Melastomataceae). R. A. Howard and E. A. Kellogg 233-255 Notes on Simple-Leaved Chinese Species of Sorbus (Rosaceae) from Hubei Province. Stephen A. Spongberg 257-262 Further Notes on Turbina and Merremia : Typification and Tax- onomy of the Neotropical Convolvulaceae. D, F. Austin and G. W. Staples 263, 264 Volume 67, Number 1, including pages 1-141, was issued January 6, 1986. JOURNAL OF THE ARNOLD ARBORETUM HARVARD UNIVERSITY VOLUME 67 NUMBER 3 ISSN 0004—2625 Journal of the Arnold Arboretum The Journal of the Arnold Arboretum (ISSN 0004-2625) is published quarterly in January, April, July, and October for $50.00 per year, plus $5.00 postage for addresses outside of the United States, by the Arnold Arboretum of Harvard University, It is printed and distributed by the Allen Press, Inc,, 1041 New Hampshire Street, Law- rence, Kansas 66044. Second-class postage paid at Lawrence, Kansas. POSTMAS- TER: send address changes to Journal of the Arnold Arboretum, % Allen Press, Inc., P. O. Box 368, Lawrence, Kansas 66044. Subscriptions and remittances should be sent to Journal of the Arnold Arboretum, 1041 New Hampshire Street, Lawrence, Kansas 66044, U. S. A. Claims will not be accepted after six months from the date of issue. Volumes 1-51, reprinted, and some back numbers of volumes 52-56 are available from the Kraus Reprint Corporation, Route 100, Millwood, New York 10546, U. S. A. EDITORIAL COMMITTEE S. A. Spongberg, Editor E. B. Schmidt, Managing Editor P. S. Ashton K. S. Bawa P. F. Stevens C. E. Wood, Jr. Printed at Allen Press, Inc., Lawrence, Kansas COVER: The stylized design appearing on the Journal and the offprints was drawn by Karen Stoutsenberger. JOURNAL ARNOLD ARBORETUM Volume 67 July 1986 Number 3 THE GENERA OF LEPIDIEAE (CRUCIFERAE; BRASSICACEAE) IN THE SOUTHEASTERN UNITED STATES12 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 267 268 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 270 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 earance of the septum during the early stages f all but one of the four to many ovules, and 3 mm long and weigh only about 0.02 mg (less than one f 271 d States (Jour. Arnold Arb. 65: 343-373. 1984). Genera 1 and 2 (Thelypodieae) appeared ii 274 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 s of fruit, free ora 275 277 278 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 279 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 281 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 [vol. 67 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 The native range of Coronopus didymus is controversial. Some North Amer- ican authors follow Robinson, who suggested that it was originally introduced into North America from Europe, while a few (e.g., Muenscher) have claimed that it is a Eurasian plant. Other authors consider C. didymus a South American plant, a view I agree with for two reasons. First, the nearest relative of C. didymus is a South American endemic recently described as C\ leptocarpus Boelcke. Second, the greatest morphological diversity of C. didymus occurs in South America, where several varieties and forms have been recognized (Muschler). It is not known how and when C. didymus first appeared in Europe and North America, but it was established in some parts of these continents during the late eighteenth and early nineteenth centuries, as evidenced by its repeated description as new species in both continents. The North African sect. Cotyliscus (Desv.) DC. includes Coronopus niloti- cus (Delile) Sprengel, an endemic of Egypt, and C. lepidioides (Cosson) Kuntze, which is restricted to parts of Tunisia, Algeria, and Morocco. Section Xeror- rhiza O. E. Schulz contains C. rhytidocarpus (Hooker) Macloskie, of Patagonia, and C. serratus (Poiret) Desv., of Argentina and Uruguay. The latter has been introduced recently in Mauritius (Troncoso & Bacigalupo). Coronopus is easily distinguished by its axillary or leaf-opposed inflores- cences, its coarsely reticulate or verrucose to rugose, usually didymous fruits (Figure lh) that break up at maturity into 1 -seeded, nutletlike halves, and its nonmucilaginous seeds. It is closely related to Lepidium. Muschler suggested that Coronopus may represent the end product of parallel evolutionary trends in Lepidium that led to the formation of indehiscent fruits. This implies a polyphyletic origin of Coronopus, a view not supported by facts. Although Muschler reduced Delpinophytum Spegaz. (= Delpinoella Spegaz.) to a subgenus of Coronopus, the two are remotely related. The former has tiny imbricate leaves, long styles, linear anthers, and copiously mucilaginous, beaked seeds. Species of sect. Nasturtiolum exhibit reductions in flower size, amounts of nectar secreted, petal size, and stamen number. The two anthers usually dehisce in the bud, and pollination may take place long before anthesis. These features undoubtedly indicate autogamy, but protandry and ant pollination have been reported in Coronopus didymus by Chauhan (1979a). Cross polli- nation and slight protogyny have been observed in C. squamatus. Several authors have stated that the flowers of C. didymus contain two or four stamens. However, the flowers in all but one of the 198 specimens I examined had only two median stamens. In one plant most flowers had two stamens and only a few had four. The petals in C. didymus are reduced into subulate or filiform structures misinterpreted as staminodes (Chauhan, 1982b) or as filaments of sterile median stamens (Smith). According to these authors, the petals are lacking, but this implies that all six stamens have a median position facing the outer sepals, a feature not yet found in any crucifer. Plants of Coronopus didymus produce on average some 1 600 seeds, but the output of large plants may be as high as 1 8,000 seeds (Salisbury). A few authors have suggested that the seeds (actually fruit halves) may be transported by strong winds or by birds (McDowall et a!.), but Salisbury maintains that the 288 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 JOURNAL OF 1 i ARNOLD ARBi Cardaria pubescens (C. A. Meyer) Jarmolenko (. Hymenophysa pubescens C. A. Meyer), a native of central Asia, is a common, aggressive, and noxious weed in western Canada and the western United States but is only occasional inflated fruits. Ch beseem i s diploid (2n = 16), wh 32) or octoploid (2 n = 64) and JOURNAL OF ' [vol. 67 F. K. JOURNAL OF ; ARNOLD ARBORETUM [vol. 67 l T. alpinum var. & The JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 302 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 12 or , of Teesdalia are s 306 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 allied to and perhaps not sufficiently distinct from Hutchinsia, which has two- to four- (instead of many-)seeded fruits characteristic of Hymenolobus. The direct relationship between Capsella and Neslia Desv. with Camelina Crantz, as proposed by Von Hayek, reflects no more than superficial resemblances. The remarkable uniformity within a given population of Capsella Bursa- pastoris is attributed to extensive inbreeding and particularly to autogamy. Some ecological factors, however, may play an important role in determining the amount of selfing. Hurka, Krauss, Reiner, & Wohrmann showed that dry and sunny days favor outcrossing and induce the formation of protogynous flowers, while cloudy or rainy days (i.e., low light intensity, high atmospheric humidity, and temperatures above 15°C) lead almost exclusively to selfing. Furthermore, cold weather appears to suppress the formation of normal an- thers. Gynomonoecious and gynodioecious plants may be produced early in the season, while functionally perfect flowers are produced in early spring (Knuth), and male-sterile ones may be encountered toward the end of the season (Stace). Numerous species of bees, beetles, butterflies, flies, hoverflies, and thrips have been observed as occasional visitors of the flowers (Knuth). Self-incompatibility has been reported in forms of Capsella Bursa-pastoris known as C. grandiflora (Riley, 1932, 1936). Although Riley suggested the presence of three mating groups controlled by two genes (each with two alleles) showing epistasy and dominance, Bateman believed that self-incompatibility in Capsella can be better explained as controlled by one locus with at least three alleles showing dominance. Two chromosomal levels, diploid (In = 16) and tetraploid (In = 32), are known in Capsella Bursa-pastoris. Crosses of plants with the same chromosome number always yield fully fertile offspring, regardless of the morphological differences between the parents. On the other hand, crossing diploid and tet- raploid plants produces sterile triploids with abortive pollen and shriveled seeds. Such triploid hybrids are widely distributed in nature and have been named C. x gracilis Gren. They are said to be derived from C. Bursa-pastoris and C. rubella. Many authors have considered the last “species” a diploid, but the painstaking work of Svensson indicates that diploid and tetraploid popu- lations in this complex are morphologically indistinguishable, even though the tetraploid tends to have slightly larger sepals, petals, fruits, and seeds. The tetraploid plants flower earlier than the diploid and appear to have wider distribution. The genetic basis of variation in Capsella was studied by Shull (1929), who grew more than half a million individuals and analyzed some 4000 pedigreed lines. He crossed C. Bursa-pastoris (obtriangular fruits) with C. Heegeri Solms (elliptic fruits) and discovered that the latter is nothing but a double recessive mutant segregating in the second generation at a ratio of approximately 15:1. The mutant was treated earlier as the basis of a new genus, Solmsiella. The embryogeny of Capsella has been more thoroughly studied than that of any other crucifer and has been included in numerous textbooks of plant anatomy and embryology as typical of the dicotyledons. The embryogeny is not repeated here, and further details are found in the classic works of Schaffner, Soudges, and Rijven. The ultrastructure, morphogenesis, and growth and de- 1986] AL-SHEHBAZ, LEPIDIEAE 307 velopment of embryos have been thoroughly covered in many papers listed in the references. Plants of Capsella contain high concentrations of allylglucosinolates and small amounts of the alkaloids tyramine and sinapine. The fatty-acid com- position of the diploid “C. rubella" and “C. grandijlora ” is indistinguishable from that of the tetraploid C. Bursa-pastoris. Seeds from the temperate regions tend to have a higher oil content than those from colder areas (Mukheijee et al.). The genus resembles most members of the Lepidieae in having high con- centrations (20-35 percent) of linoleic and linolenic acids and negligible amounts (0.2-0. 7 percent) of erucic acid. An anatomical peculiarity of Capsella Bursa-pastoris is the derivation of the interfascicular cambium of the shoot from the starch sheath (Hurka & Biichele). The seed coat of this species is indistinguishable from that of “C. rubella ” and “C. grandijlora" and is composed of palisade cells broader than long and with thickened radial and inner tangential walls, and of epidermal cells with a large, solid column protruding into the lumen from the inner tangential wall (Vaughan & Whitehouse). A single plant of Capsella Bursa-pastoris may produce 5000-90,000 (usually 30,000-60,000) small seeds weighing about 0.1 mg (Hurka & Haase). The mucilaginous epidermis of the seed coat aids in long-distance dispersal by adhering to animals and farming machinery. The seeds may retain viability after passing through the digestive tracts of birds and earthworms. The species shows a remarkably high degree of phenotypic plasticity, and the plant may produce fruits within a few weeks after seed germination. Although seed size is genetically controlled, it is also phenotypically plastic; plants of severely cold climate usually produce larger seeds than those in more favorable areas (Hurka & Benneweg). Seeds of Capsella show pronounced dormancy that can be broken experi- mentally by cold treatment after imbibition. An increase in nitrate level and a fluctuation in temperature can replace the requirement for stratification and are therefore important in breaking seed dormancy in nature. Cold-treated seeds may fail to germinate unless exposed to light at the soil surface (Popay & Roberts, 1970a, 1970b). Barber (1978a, 1978b) indicated that the seeds of Capsella Bursa-pastoris show protease activity in their mucilaginous sheath and are capable of uptake and incorporation of radioactively labelled amino acids. Seeds attract soil nematodes, protozoans, and bacteria, causing some mortality, and they may even derive some nutrients from them. Barber suggested that these adaptations may supplement the nutrition of the germinating seed but questioned whether the seeds are truly “carnivorous.” Many medicinal properties have been attributed to Capsella Bursa-pastoris, and various preparations have been prescribed to treat tumors, nasal and internal bleeding, stomach disorders, irregular menstruation, kidney com- plaints, dysentery, and dropsy. The plant is also considered antiscorbutic, stim- ulant, astringent, and diuretic (Caius, Hartwell, Perry). The young rosettes are eaten as a salad and are regarded as an excellent substitute for spinach and a rich source of vitamin C. The plant is sold in some parts of China and India. 311 STEVENS, EVOLUTIONARY CLASSIFICATION 313 EVOLUTIONARY CLASSIFICATION IN BOTANY, 1960-1985 3 1 4 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 in, 1968, p. 26) that is e for. s (1978) saw a broadly There is . 315 316 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 The 317 318 is difficult (e.g., see Meeuse, 1964; Barabe, 1984). 1986] STEVENS, EVOLUTIONARY CLASSIFICATION 319 characters, but the detailed distribution of such characters within those taxa is not discussed. The use of numerous characters in group formation in such cases is hardly appropriate. I also did not realize that only characters of a certain kind were likely to be helpful in establishing cladistic (or- more generally— “evolutionary”) rela- tionships. For example, the occurrence of uniseriate hairs in the Ericaceae (basically in Epigaea L. and Wittsteinia) should not enter into the discussion. Since other characters show that the groups in which that character occurs are not immediately related, it is not a unique character signifying relationship. Of course, the fact that Wittsteinia had been placed in the Ericaceae influenced the final decision about relationships and taxonomy, but the methodology used allowed it to be retained there. It is now known that Wittsteinia should properly be compared with Alseuosmia A. Cunn., of the Alseuosmiaceae (see Van Stee- nis, 1984). The two have a number of similarities, at least some of which appear to be restricted to them (and perhaps a couple of other genera) when compared with their possible relatives. These similarities include uniseriate hairs in the same position, serrate petals, and perhaps also nodal and xylem structure (see Stevens, 1971; Gardner, 1978). A similar example is the inclusion of the Barbeyaceae in the Urticales (e.g., see Cronquist, 1981), which apparently strengthens the case for including the Urticales in the Hamamelidae. The Barbeyaceae have primitive characters of the Hamamelidae that are not found in the Malvales (Dilleniidae), another possible relative. These characters are not even unique to the Hamamelidae, being found also in some other Dilleniidae (and elsewhere). In the method of forming groups discussed above, generally only the occur- rence of characters within the two taxa is considered. Because of the great variability and heterogeneity of some groups, it may be easy to find relationships between groups, especially if the distribution of a character within groups is not considered important (e.g., see Benson, 1957; Clayton, 1983). As multiple, largely pairwise comparisons are carried out over time, taxa may be chained together, again with the results that taxa at opposite ends of the chain may have little in common. The method of independent pairwise comparison of taxa is close to the single linkage clustering method and similar algorithms described by Sneath and Sokal (1973). Interestingly, one of the less-desirable features of this algorithm is its propensity to produce chaining of taxa, with groups adding on singly and sequentially to the first group (Clifford & Ste- phenson, 1975). Polythetic groups will tend to be the result. For example, Hamann (1961) commented on the extent of similarity rela- tionships shown by the more heterogeneous monocotyledonous families. Cron- quist (1983) found no difficulty in assigning rather problematic groups to the Violales (e.g., the Huaceae) and Rosales (Surianaceae, Rhabdodendraceae, An- isophylleaceae); those orders were already large and heterogeneous, so further groups deemed to be related could be added easily. The Rosaceae, Flacourti- aceae, Liliaceae, Euphorbiaceae, and Leguminosae are examples of heteroge- neous families, and their extensive affinities are largely due to the rationale used when establishing relationships. 320 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 STEVENS, EVOLUTIONARY CLASSIFICATION 321 relationship could equally well have been suggested with a variety of other families. The use of numerous characters may alleviate the problem, but unless these different kinds of primitive characters are distinguished, the justification for a particular relationship must be ambivalent. Second, intermediate groups, primitive characters, and ancestors have be- come connected in a curious way. Saruma is considered to be both intermediate in position and primitive in many of its characteristics. Botanists tend to treat intermediate taxa as living fossils: taxa that link become ancestral, and their characters primitive. Although there is an element of convention in our use of terminology and “extinct ancestors” are implicit in our discussions (B. L. Burtt, pers. comm.), this is not evident in many papers. There are obviously exceptions (e.g., Burtt, 1977), and Thome (1976), Stebbins (1976), Dahlgren (1975), and others have commented on the danger of considering ancestors to be extant, which is indeed one of the most criticized aspects of evolutionary systematics. Barabe and Brouillet (1982) distinguish between evolutionary and cladistic systematics in part because in the former the ancestors can be recognized and identified, but the point made here is that an ancestor of a particular group is often treated as being alive and/or a member of a coordinate higher taxon. Intermediate Taxa and Gaps Cronquist (1 983) has recently emphasized what he calls McVaugh’s Principle (see McVaugh, 1 945, for the action of this principle at lower taxonomic levels), which facilitates the taxonomic treatment of intermediate, “ancestral” groups. A segregate taxon that is joined to another taxon by a few intermediates is maintained as a distinct and distinguishable entity by including the interme- diates in the (larger) taxon in which the segregate was originally placed. The larger taxon will be more variable, some of its members perhaps already pos- sessing the characters of the intermediate taxon because of parallel variation or convergence (see below). Thus its description will have to be changed only slightly, if at all, to accommodate variation shown by the intermediates; the description of the less- variable segregate taxon would have to be changed considerably if the intermediates were placed in it. That taxa are commonly formed in this way may explain why characteristics rare in one group may be widely distributed in another (Thome, 1976, principle 16, “trends or tendencies”). Thus, Cronquist (1981) noted that the Flacourti- aceae, basal in the Violales, show tendencies toward the features of several other families of that order. It also explains how the problem of the ancestor is dealt with given a belief in gradual evolution (Ashlock, 1971): the ancestor is submerged in the group from which it evolved rather than being placed with its descendants. “Ancestral groups” tend to be polythetic. McVaugh’s Principle is a special case of a more general rule whereby easily characterized groups are kept separate from the taxon from which they are supposed to have arisen. This general principle encourages the recognition of groups that are formally equivalent, having the same rank, but not otherwise comparable. The segregate may be strictly monophyletic, but the group from JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 ; Takhi"LnWo° 1986] STEVENS, EVOLUTIONARY CLASSIFICATION 323 THE NATURE OF HIGHER INFLUENCE ON GROUP TAXA AND ITS FORMATION £5 324 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 ussion of McVaugh’s rule] for reasons of c ;d not be invoked [see footnote 5]| i an idea. It cannot be < . This will lead to i the Table it can be JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1 986] STEVENS, EVOLUTIONARY CLASSIFICATION 327 expected, was widely held in the early nineteenth century ( natura non facit saltus— Linnaeus, 1751), and it strongly influenced a number of the most im- portant French- and English-speaking proponents of the natural method until about 1850. Thus intermediates were expected, and when reticula were later turned sideways and called trees, the intermediates simply became ancestors. Yet later, when trees fell out of favor and onto their sides, intermediates were still discussed because ancestors could not be recognized. Ramsbottom (1938) noted the similarities between these newly felled trees and early diagrams of relationships (as in Giseke, 1792) in which intermediates played an important Takhtajan (1980, p. 234) thought that phylogenetic (evolutionary) system- atics had been enriched over the years by “theory, the conceptual apparatus.” In fact, Cullen (1968) seems to have been somewhat nearer the mark when he observed how little discussion had been occasioned by theoretical works pub- lished in the preceding century and a half (see also Stevens, 1984b). Apart from the art and artifice used in group formation, systematics has been mistakenly considered a basically empirical— perhaps almost Baconian— discipline by which the natural order is revealed (see especially Hull, 1970, but cf. 1984); “natural groups” simply have to be recognized. Relative disinterest in theory is to be expected of a discipline that considers itself basically empirical (Reif, 1983; cf. Stevens, 1984c, and Tomlinson, 1984). As a result the real conceptual basis of the evolutionary school of systematics has been largely misunderstood, with the resultant compromise of the discipline. Thus it is perhaps ironic that Takhtajan (1958) observed that earlier authors of evolutionary systems at most offered very weak corroboration of their views, or no substantiation at all (there are exceptions, of course). Much current documentation and justification remain unconvincing. Interpretations of the evolutionary significance of characters vary widely from author to author. The use of additional criteria that affect rank and circumscription of taxa— amount of variation contained in a taxon, size of taxon, convenience for the user, and custom (e.g., see Cronquist, 1968; Jeffrey, 1982; Ehrendorfer, 1983)— leads the user farther from an evolutionary classification, let alone a classification from which knowledge of cladistic relationships can be obtained (see also Edwards & Cavalli-Sforza, 1964; Bremer & Wanntorp, 1978; Barabe & Brouillet, 1982; Stevens, 1985). It is thus insufficient to champion the use of the hypothetico- deductive method in systematics (e.g., see Takhtajan, 1969) since some of the premises of evolutionary systematics are suspect. There are three interconnected reasons that contribute to the relative un- concern among taxonomists about the relationship between classification and theory. First, at least some higher taxa, being “natural” (read “real”) taxa, appear to have an almost unassailable ontological status. This emphasizes that one critical element in the development of evolutionary classifications (and classifications in general) is the attitude of the taxonomist toward groups established by his or her predecessors. These groups are often taken to represent “real” groups. That they are recognizable by the eye is proof of that reality; the principles used in their formation are of less importance. This attitude is a recurring one. It is visible in Toumefort’s (1694) criteria for the recognition of genera of the second rank, as well as in Linnaeus’s genera 328 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 (Walters, 1961). Even Adanson (1764) used existing groups to show that no one character separated them unambiguously. Although Adanson was com- bating an overly a prioristic and analytic approach to classification, attitudes such as this can also be seen as protecting “natural” groups from dismember- Second, the unchanging nature of taxonomic practice may seem to be evidence of its success and usefulness in science. To many there is then no necessity to change the way we analyze the world, despite our changing ideas as to the origin of diversity in that world (Walters, 1961, especially 1964). To put the argument like this cries out for an analysis of the principles of classification that is so far lacking. Since there are no clear criteria for evaluating characters or relationships, the frequent pleas will often be heeded that systems should not be changed until we have strong evidence for, or are even “sure” of, a particular change. It is thus hardly a matter for surprise that new characters may not change the system (cf. Stafleu, 1969); new characters are evaluated in the old way. The systems are internally consistent; as M. Frohlich (pers. comm.) has emphasized, this is an important goal for most of us. They are consistent both with the primary interlocking principles of the Table and, more specifically, with a particular way of interpreting the data. Systems can, and indeed have, changed radically with changing opinions on “which way up” evolution is; data are reinterpreted, but using rather similar general arguments. However, the primary principles remain unaffected (see Wiley, 1979). The apparently substantial differences between the Englerian and Besseyan schools are of this relatively secondary nature. Third is the great importance attached to nomenclatural stability, and the concomitant reluctance to change names. A few authors (e.g., Turrill, 1957; Walters, 1961; Clayton, 1974) have com- mented more or less severely on the constraining effects of previous work and on what current taxa are perhaps supposed to represent. Thome (e.g., 1976) has perhaps been most vocal in attempting to divorce present classifications from considerations of size or economic importance of taxa and also tradition (cf. Cronquist, 1968; Davis, 1978; Van Steenis, 1978). Ranking of taxa can often be traced back to decisions made over a century ago that were explicitly decisions of convenience (e.g., see Burtt, 1977; Van Steenis, 1978; Stevens, 1984b, ms.). Many systematists faced with the conflicting goals of evolutionary classifications— depicting phylogenetic relationships and promoting stability of classifications— have opted for the latter (see Stevens, 1984b, for references). It is difficult to contemplate classification without simultaneously thinking of All three reasons reinforce the tendency to evaluate characters in terms of how they support current ideas of relationships. Thus the task of the contrib- utors to the 1973 Amherst conference on angiosperm phylogeny was to see how the characters they were studying could be applied to the Cronquist and Takhtajan systems (Walker, 1975). Harris and Bisby (1980) see problems in being able to find evidence that conflicts with a current classification. There is perhaps less inclination to explore the taxonomic stmcture suggested by our 1986] STEVENS, EVOLUTIONARY CLASSIFICATION 329 above). Thus the stability that Barabe and Brouillet (1982) see as the advantage Davis, 1978; 1986] STEVENS, EVOLUTIONARY CLASSIFICATION 331 p, C. E. B. 1962. The > 332 334 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 ?>=* 336 JOURNAL OF THE ARNOLD ARBORETUM [voi . 67 334-337. .Pp. [v]-[xv] in M. i 374-3 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 339 WHISTLER, PSYCHOTRIA 341 A REVISION OF PSYCHOTRIA (RUBIACEAE) IN SAMOA III ilf fltliS 343 i L. Syst. Nat. ed. 10. 929, 1364. 1759, r is, Upolu has 14, and Tutuila and the three is l of the genus, sect. Eumachia (DC.) A. C. £ l group includes Psychotria apodantha A. Gray, P. christophersenii P. garberiana Christoph., P. insularum A. Gray, P. pacifica Schu- Eumachia, as sections. In his ongoing Flora Vitiensis Nova, A. C. Smi soon be publishing a treatment of the approximately 80 Fijian species c JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 345 of ribbing of the pyrenes. Two s ^r^“£c [vol. 67 Bot. Jahrb. Syst. 25: 685. 1575). JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 PSYCHOTRIA i !!i! Inlill ! rl ill If 354 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 rounded apex, 3.5-5 cm long. Leaves with petiole 4-8.5 cm long, glabrous; blade elliptic, 14-26 by 4-8 cm, apex acute to mucronate, base acute, surfaces 15-25 cm long on style 3. 5-4. 5 2-3 355 ellipsoid, 20-33 mm 1 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 WHISTLER, PSYCHOTRIA 357 This species differs most notably from any other in Samoa by its connate, whitish green stipules that form a cup-shaped involucre with the tip often recurved, and by its often-purplish fruits. Also, it has the largest flowers of any species of Psychotria in Samoa. 11. Psychotria juddii Christoph. Bernice P. Bishop Mus. Bull. 154: 56. 1938. Type: Samoa, Savaii, forest above Gagamalae, above 900 m, 1931, Christophersen 3424 (holotype, bish!; isotypes, a!, k!). Small tree up to 6 m tall. Stems glabrous, leaves concentrated at ends of branches; stipules lanceolate, split at tip into 2 short lobes, connate over half their length, 15-30 mm long. Leaves with petiole 2-6 cm long, glabrous; blade oblong to elliptic or slightly obovate, 14-32 by 5.5-11 cm, apex acute to JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 , Syst. 25: 687. 1898; ] JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 8-20 by 1.8-6 cm Z^TStefr*****1* 517 (B, BISH, ptbg, us); above Asau, JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 17. P. Bishop Mus. Bull. 154: 51. is. Bull. 184: 65. 1945. Type: ridge, 500 m alt., 1925, Garber ,5-16 by 1.5-7 cm, t a 364 JOURNAL OF THE ARNOLD ARBORETUM 67 365 Dra Schumann, Bot. Jahrb. Syst. 25: 687. 1898; Rech. Akad. Wiss., Math.-Naturwiss. KI. 85: 200. 1910; P. Bishop Mus. Bull. 154: 62. 1938. Type: Samoa, ^ Small tree 2-4 m tall. Stems glabrous; stipules ovate to elliptic with notched JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 BKepf’ffishop Mus. Bull. 154: 59. These are Graejfe 1608, without further locality; Reinecke 195, collected at Mulifanua, Upolu; and Reinecke 341, from Lanoanea, Upolu. andZrwt^ Dr. A. C. Smith, of the same ; of the species; Dr. Harold St. 370 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 Sohmer, S. H. 1977. Psychotria L. (Rubiaceae) in the Hawaiian Islands. Lyonia 1: 103-186. Whistler, W. A. 1984. Annotated list of Samoan plant names. Econ. Bot. 38: 464- 489. JOURNAL OF THE ARNOLD ARBORETUM INSTRUCTIONS FOR AUTHORS General policy The Journal of the Arnold Arboretum is primarily a staff journal, and staff papers have priority. Other papers are accepted, as space permits, from former staff or former students, and from other botanists who have worked on our collections or who have done research on a plant group or in a geographic area Submission of manuscripts Manuscripts should be submitted in triplicate to Ms. E. B. Schmidt, Man- aging Editor, Journal of the Arnold Arboretum, 22 Divinity Avenue, Cam- bridge, Massachusetts 02138. A copy of the manuscript should be retained so that when reviews and/or editorial suggestions are received, any necessary corrections can be made and the appropriate portions of the paper resubmitted. For ease of editing, an outline of the paper (not to be published) showing the basic structure of the manuscript should be included. Preparation of manuscripts Papers should be triple spaced throughout (including title, abstract, text, citation of specimens, footnotes, acknowledgments, bibliography, and figure legends), on bond (not erasable) paper, with wide margins on all four sides. Nothing should be underlined except generic and infrageneric scientific names, italics when present in a quotation, and the collector and collection number of all specimens cited. Form and style. This can be determined from a recent issue of the Journal. The title should be as short as possible; it should usually contain the name of the family concerned but not authorities of scientific names. Each author’s current address should be given as a footnote to his name. An abstract should be included. Abbreviations should be employed only when two or more letters will be saved, and with the exception of units of measure, compass directions, and herbarium designations, they should always be followed by a period. Metric measurements should be used when possible. Authority names should be given for all generic and infrageneric taxa the first time they are mentioned in the text unless they are included in the formal taxonomic treatment. Acknowledgments should be placed at the end of the paper before the bib- liography. Footnotes should be kept to a minimum and should be numbered consecutively throughout the paper, with the exception of those appearing in tables. Here either standard symbols or lower-case letters should be used. In case of question, the latest edition of Words into Type, by Maijorie E. Skillin et al. (Prentice-Hall), should be consulted. ion the back ofei They should be grouped in numerical sequence on a separate page, rather than placed below each figure or plate. Illustrations of each type (i.e., figures, plates, or maps) should be numbered consecutively and separately, figures and maps with Arabic numerals and plates with Roman numerals. For example, a paper could include figures 1-3, maps 1-3, and plates I-VII. If the illustrations are to be figures, each figure should be numbered separately, with the numbers running in order through the text. Subdivisions of the figures should be indi- cated with letters. Thus, figures 1-5 may appear on one page, figures 6-8 on the following page, and figure 9, A-C, on the next. In the case of plates, each page of photographs is a separate plate; the individual photographs comprising the plates are numbered or lettered consecutively. Illustrations with dark back- grounds should have white letters or numbers and vice versa. Whenever pos- sible, scales should be included in the illustrations; any magnifications necessary in the legends should be calculated to include reduction of illustrations to our page size. Illustrations are not returned to the author after publication unless this is Tables. Titles for tables should be short, with all explanations placed in footnotes. Tables should be as simple as possible and must be neatly typed. Long and/or complicated tables can be photographed directly if they are in good order and the copy is clear enough (in this case the copy should not be triple spaced, but should have spacing appropriate to contents and headings); this eliminates the chance for error and the need for proofreading. Tables should be numbered consecutively using Arabic numerals. Each table should be cited in the text. Bibliography. The Guide to Citation of Botanical Literature in the Inter- national Code (all editions through 1972) should be followed. When possible, reference should be made to past issues of the Journal for form. Titles should be abbreviated according to Journal (Schwarten & Rickett, Bull. Torrey Bot. Club 85: 277-300. 1958) or Botanico-Periodicum-Huntianum style; the two styles should not be mixed within a paper. Runovers should be indented. Except of first and middle names. All typing should be done in regular capital and lower-case letters, and nothing should be underlined except generic and infra- generic scientific names. Titles of articles and books should not be capitalized except for the first word, scientific names, and proper nouns and adjectives. Charges for extensive alterations in proofs will be billed to the author. Authors are requested to help defray printing costs. Although actual printing costs are much higher, $20.00 per page is the customary charge, and authors are expected to make every effort to pay. Under special circumstances the fee may be reduced or waived altogether, if this is agreed upon in advance. Ability or inability to pay will in no way affect acceptance or handling of a manuscript. Journal of the Arnold Arboretum July, 1986 CONTENTS OF VOLUME 67, NUMBER 3 The Genera of Lepidieae (Cruciferae; Brassicaceae) in the South- eastern United States. Ihsan A. Al-Shehbaz 265-311 Evolutionary Classification in Botany, 1960-1985. P. F. Stevens 313-339 A Revision of Psychotria (Rubiaceae) in Samoa. W. A. Whistler 341-370 Volume 67, Number 2, including pages 143-264, was issued April 15, 1986. JOURNAL OF THE ARNOLD ARBORETUM HARVARD UNIVERSITY VOLUME 67 NUMBER 4 ISSN 0004-2625 Journal of the Arnold Arboretum The Journal of the Arnold Arboretum (ISSN 0004-2625) is published quarterly in January, April, July, and October for $50.00 per year, plus $5.00 postage for addresses outside of the United States, by the Arnold Arboretum of Harvard University. It is printed and distributed by the Allen Press, Inc., 1041 New Hampshire Street, Law- rence, Kansas 66044. Second-class postage paid at Lawrence, Kansas. POSTMAS- TER: send address changes to Journal of the Arnold Arboretum, % Allen Press, Inc., P. O. Box 368, Lawrence, Kansas 66044. Subscriptions and remittances should be sent to Journal of the Arnold Arboretum, 1041 New Hampshire Street, Lawrence, Kansas 66044, U. S. A. Claims will not be accepted after six months from the date of issue. EDITORIAL COMMITTEE S. A, Spongberg, Editor E. B. Schmidt, Managing Editor P. S. Ashton P. F. Stevens C. E. Wood, Jr. Printed at Allen Press, , Lawrence, Kansas JOURNAL ARNOLD ARBORETUM Volume 67 October 1986 Number 4 MORPHOLOGY AND TAXONOMY OF THE NEW WORLD SPECIES OF MAIANTHEMUM (LILIACEAE) James V. LaFrankie, Jr.1 372 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1 of the Liliaceae are p name. Despite the number ofspe sts of the Temperate Zone, the ] i was rejected. The transfers did r aeus (1753) r< He was prot i part of the c 374 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 thereafter adult or renewal growth commences as renewal shoots are produced from axillary buds on the lower portion of the parent stem. There are four different patterns of shoot development in Maianthemum, but they share several features. In all species of Maianthemum, the two growth processes of initiation and extension are separated temporally. All appendicular parts are initiated in miniature within a swollen renewal bud and then reach full size through uninterrupted meristematic growth and cell enlargement. In 375 . In the second type of grc JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 the storage of starch (as identified microscopically and with IKI), I have made of the 1986] LaFRANKIE, MAIANTHEMUM 377 NUMBER OF LEAVES NUMBER OF LEAVES JOURNAL OF ARNOLD ARBORETUM [vol. 67 JOURNAL OF THE ARNOLD ARBORETUM [ vol. 6 7 f i£s 1986] LaFRANKIE, MAIANTHEMUM 381 Flower The flowers of Maianthemum are relatively small and inconspicuous and are characteristic of the lily family. They may be trimerous (six tepals, six stamens, three carpels) or dimerous (four tepals, four stamens, two carpels) through reduction (Utech & Kawano, 1976), but otherwise the flowers of dif- ferent species are roughly similar to one another, differing chiefly in color, dimensions of parts, and disposition of tepals. III! 382 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1986] LaFRANKIE, maianthemum 383 some species with 12 seeds per fruit (e.g., M. macrophyllum), the seeds are conical with one or more flattened faces. Seed diameter differs significantly among species but is not reliable in her- barium material because drying reduces the diameter from 20 to 40 percent. The seeds of all species are structurally uniform. The testa is thin and papery, straw colored or light brown. The endosperm is white, scaly, and rich in food reserves that are stored in the greatly enlarged cell walls. The embryo is always small and straight; the cotyledon can be spherical or claviform. Most species of Maianthemum have snow-white flowers, but some have flowers infused with green or yellow, and others have them with flecks of red- violet. A few of the Central American species are more variable in flower color (including rose or lavender), but I have rejected the varietal ranking that Emons (1945) gave these individuals because color can vary through a whole range of shades even on a single plant. Cytology Chromosome number and morphology are uniform in most species studied to date (Therman, 1956; Kawano & litis, 1966; Kawano, Ihara, Suzuki, & litis, 1 967; Valentine & Hassan, 1971; Sen, 1 974). The haploid chromosome number is 18: one long (ca. 10 jum) with a metacentric constriction; nine medium- length, of which two have metacentric constrictions and seven have acentric or subterminal constrictions; and eight small (ca. 5 /im) with more or less metacentric constrictions. A few species show various levels of intraspecific polyploidy, but in general, evolution within this genus has been unaccompanied by gross chromosomal change. ECOLOGY AND DISTRIBUTION Most species of Maianthemum are forest herbs, but there are many excep- tions, reflecting the broad ecological amplitude of the genus with regard to water, temperature, and light. For example, M. stellatum grows on sand dunes, M. trifolium and M. paludicolum occur in sphagnum bogs in full sun, and M. amoenum is found only as an epiphyte in the canopy of cloud forests. Most of the species require cool temperatures and abundant moisture. In Central America Maianthemum grows in locations above 1 500 m altitude that receive more than 1 500 mm annual rainfall. The genus is generally hardy within these limits, and at least one species seems to be found in most of the areas where these two conditions are met. Some of the ecological aspects of growth habit are mentioned above. Growth is generally episodic, with the phenological cycle closely tied to the seasons. The forest herbs of the Temperate Zone emerge in early spring, extending their foliage leaves before the forest canopy develops and then persisting through the remainder of the summer in light to moderate shade. In most of the tropical species, new shoots are extended between January and June, which coincides with the drier part of the year. This perhaps allows for more reliable pollination but also requires that water and nutrients be reserved within the rhizome. This constraint would apply all the more to the epiphytic species. 1986] LaFRANKIE, MAIANTHEMUM 385 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 (1963); Braun (1967); Radford, Ahles, and Bell (1968); Voss (1972); Barkley (1977); Harvill, Stevens, and Ware (1977); Mohlenbrock and Ladd (1978); and Wherry, Fogg, and Wahl (1979). Within Central America specific collection TAXONOMY JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 390 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1986] LaFRANKIE, MAIANTHEMUM 391 JOURNAL OF THE ARNOLD ARBORETUM [vol, 67 1986] LaFRANKIE, maianthemum 394 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1986] LaFRANKIE, MAIANTHEMUM 395 inserted at tepal base, the filament 1-1.5 mm long, very thin, the anther 0.2- 0.4 mm long; ovary spherical, 0.8-1 mm broad, the style 0.5-0.8 mm long, the stigma distinctly 2-lobed, 0.3-0. 5 mm broad. Fruits spherical, 4-6 mm in diameter, green mottled with red when young, maturing to deep translucent red; seeds up to 2, spherical, ca. 3 mm in diameter. Chromosome number In = 36, 54, 72 (Kawano, Ihara, Suzuki, & litis, 1967; Valentine & Hassan, 1971). Ecology and distribution. Abundant in northern forests (nearly ubiquitous in forests of eastern Canada and the northeastern United States); sea level to 1 800 m alt. Flowering May-June; fruits retained through August or September. From Labrador (Battle Harbor, Williamson s.n. (gh, ph)) and Prince Edward Island (Queens Co., Fernald 7199 (gh, ph)) south along Atlantic coast to New Jersey (Cape May Co., Cape May Courthouse, Long 7219 (gh, ph)) and Del- aware (Newcastle Co., Red Clay Creek, Long 30211 (gh, ph)), then inland at higher elevations as far south as Virginia (Roanoke Co., McAfee Knob, C. Wood 6055 (gh)) and North Carolina (Transylvania Co., Lake Logan, Hardin 715 (gh)). West of Appalachians, southern limit much to north, extending across northern Indiana (LaPorte Co., W end of Hudson Lake, Friesner 21408 (gh); and Jefferson Co. (Deam, 1940)) and Illinois (LaSalle Co., Starved Rock, Greenman s. n. (gh)), with outlying location in Henry Co., Missouri, fide Barkley (1977), although unreported there by Steyermark (1963). Northern limit ex- tending from central Quebec (Lac Mistassini, Rousseau 2005 (gh)) through Ontario (Sandy Lake, Moir 4065 (gh)), where nearly ubiquitous in coniferous forests although not widely collected, and to Northwest as far as Alberta (Wood JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1986] LaFRANKIE, maianthemum 397 Ii.Mii.il fliiiiBIfiilili (1966) claimed name. Greene’s name is best viewed as simply a new com- v ranking of Wood’s, and it therefore automatically has the , 20-35 cm by 1-1.5 2-4(-7) by 1-2 cm, ^slightly flattened,^ scale leaves 7 or 8, the intemodes 5- in distal portions. Leafy stem arched, 40-70 cm long, 3-4 mm broad at base, blade ovate to lanceolate, 8-12 by 3-5 cm, the apex long-acuminate (upper 400 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 leaves) or acute to short-acuminate (lower ones), the base rounded to slightly tapered, the margin undulating, entire or rarely denticulate, the veins ranked, the surfaces glabrous, rarely with minute hairs. Inflorescence a complex raceme with 25 to 65 flowers; main axis pendent, flexuous, tapering, 5-18 cm long, 1.5-3 mm broad at base, smooth or slightly ribbed, green, glabrous; intemodes 10 to 16, 6-8(-20) mm long; fascicles distichous, with 2 to 4 (to 6) flowers, slightly reflexed and drooping, bract subtending each fascicle. Rowers trimer- LaFRANKIE, maianthemum 401 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1986] LaFRANKIE, maianthemum 403 Morphometric comparison of inflorescences of Maianthemum scilloideum and M. flexuosum.* M. scilloideum M. flexuosum Mean Range dian Mean 1.6-8. 3 2.9 2.9 ± 0.9 5-18 9.1 4 to 27 15 15.5 ± 2.5 8 to 65 30 1 to 3 3 2.5 ± 0.2 2 to 6 3 3-4.5 4 4,2 ± 0.5 8-21 14,5 10.4 ± 1.4 massive, forking sympodium, the individual units broad-claviform to sub- spherical, 3-4 cm broad, the scale leaves 8 or 9, the intemodes 2-5 mm long, the lateral buds 6 or 7, axillary, 1 or 2 developing as leafy shoots; epidermis replaced by periderm; starch only near rhizome apex. Leafy stem leaning or arching, 75-250 cm long, 8-1 5 mm broad at base, the surface smooth, glabrous; foliage leaves 10 to 16; intemodes 3-5.5 cm long. Leaves with petiole 4-5 mm long; blade ovate to elliptic, 14-30(-38) by 5-9(-l 1) cm, the apex acuminate, the base tapered or rounded, the margin undulating, denticles distinct but very small, the veins ranked, the surfaces glabrous. Inflorescence a pyramidal panicle with 45 to 400 flowers; main axis arching upward, straight, stiff, 15-29 cm long, 2.5-4 mm broad at base, ribbed, light green- white, or red-purple, glabrous; intemodes 25 to 30, most 2-15 mm long; lateral branches racemose, arranged in helix, spreading or slightly ascending, 6-15 cm long, with 1 or 2 flowers at base and others inserted at 1— 4(— 10) mm intervals, bract subtending each lateral raceme. Flowers trimerous; pedicel 2-5 (-7) by 0.5-0. 8 mm, deeply ribbed, glabrous, with bract subtending; tepals uniform, spreading, (2.5-)3-5 by 1.3- 2.3 mm, white, sometimes with purple spots; stamens inserted at tepal base, the filament 2-3.5 mm long, thin, the anther oblong, 0.5-1 mm long; ovary spherical to cylindrical, 1-1.5 mm in diameter, the style 0.8- 1.2 mm long, the stigma 3-lobed, 0.7-1 mm broad. Fruits spherical to weakly 3-lobed, 10-12 mm broad, green mottled with red when young, maturing to red; seeds up to 6, usually spherical, 4-6 mm in diameter. Chromosome number 2 n = 36 (Kawano & litis, 1966). The Central American species of Maianthemum with paniculate inflores- cences are often difficult to identify from herbarium material alone, which is perhaps why previous taxonomic treatments of this group have lumped all such specimens under “M. paniculatum .” The flowers are relatively uniform when dried, and the species differ chiefly in the shape of the rhizome, the number of foliage leaves, and the form of the inflorescence. The rhizome and the foliage leaves are often missing from herbarium specimens, and the question of inflorescence morphology is often confounded by uncertainty as to the plant’s level of maturity. A rule of thumb helpful in determining maturity is that a plant is full size when the length of the inflorescence exceeds that of the pen- JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 flEEiill 1986] LaFRANKIE, MAIANTHEMUM 405 406 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 408 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 Figure 1 1 . Maianthemum macrophyllum ( LaFrankie 84- V- 1 1A, gh): A, upper por- tion of leafy shoot and inflorescence; B, rhizome; C, D, lateral fascicles; E, flower at anthesis; F, tepal; G, stamen; H, gynoecium; I, ovary, transverse section, showing 3 locules, 6 ovules (6 other ovules beneath). acuminate, the base rounded, the margins flat, with denticles only near apex, the veins ranked, the surfaces glabrous, polished. Inflorescence a dense complex raceme with 60 to 1 20 flowers, columnar; main axis arched upward, straight, stiff, of uniform width, 10-20 cm by 2-4 mm, green, smooth, polished; inter- nodes 25 to 50, 5-10 mm long; lateral fascicles arranged in helix on main axis, each with 2 to 4 flowers and with bract subtending. Flowers trimerous; pedicel reflexed, (2— )4— 6(— 9) by 0.5-1 mm, smooth, glabrous, with bract subtending; 410 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 LaFRANKIE, maianthemum 411 long, thin, the anther oblong, ca. 1 mm long; ovary cylindrical, ca. 2 by 1 mm, the style 1.5 mm long, the stigma 3-lobed, ca. 1.5 mm broad. Fruits 3-lobed, 10-12 mm broad, green when young, maturing to red; seeds 2 to 12, slightly flattened, 1.5-2 mm wide. Chromosome number not known. Type. Costa Rica, Alajuela, Monteverde Forest Preserve, Pantano Chomogo, 1600-1620 m, Dwyer 1408 (holotype, f!). Ecology and distribution. Cloud forests, persisting in blowdowns as long as canopy remains open; 1600-1700 m alt. Flowering December-February; fruits retained through August. Relatively restricted distribution; most collections from Cordillera de Tilaran in northern Costa Rica. Specimens examined. Nicaragua. Jinotega: [northern spur of Cordillera Isabella,] Cerro San Pedro, Monte Kilambe, P. Moreno 7516 (mo). Costa Rica. Alajuela: Reserva Forestal, San Ramon, Carvajal 135 (mo), 346 (mo); 15 km NW of San Ramon, Cerro Azahar, Liesner 15620 (mo); 12 km NW of San Ramon, Judziewicz 14906 (mo); Mon- teverde Forest Preserve, NE section, 1600-1700 m, LaFrankie 83-IV-14B (gh), Utley 2379 (f); road out of Sucre toward Pozo Verde, toward Cerro Porvenir, Luteyn 4506 (duke). Heredia: S slope of Volcan Barba, Hatheway 1372 (us). The form of the lateral racemes (each consisting of a few flowers borne on long pedicels that are drawn together, with the subtending bract usually adnate and repositioned at the midpoint of the pedicel) associates this species with the more northern Maianthemum amoenum. Also, both species are epiphytes. However, M. monteverdense differs from M. amoenum in having a long, pen- dent inflorescence rather than a short, erect one and in lacking the red coloring in the central axis of the inflorescence. It is also larger in most vegetative dimensions. 8. Maianthemum paludicolum LaFrankie, Amer. J. Bot. 73: 1258. 1986. Type: Costa Rica, Cartago, 1 7 km S of Empalme along Pan-American High- way, 2500 m, 22 Sept. 1985, LaFrankie & Beach 85-23 (holotype, gh!). Figure 13, Map 4. Terrestrial herb, 1 m or more tall. Roots uniform, 2 to 4 per rhizome unit, aggregated at branch junctions, up to 60 cm by 1-2 mm. Rhizome a linear or forking sympodium, mostly exposed and upright, the individual units long- cylindrical, 10-45 cm by 7-20 mm, the scale leaves 7 to 9, the first and last intemodes 2-5 mm long, middle ones 3-25 cm long, the lateral buds 2, at base of leafy shoot, axillary, 1 or 2 developing as leafy shoots; epidermis replaced by periderm; distribution of starch not known. Leafy stem upright, slightly flexuous, 35-65 cm long, 3-4 mm broad at base, glabrous; foliage leaves 9 to 12 (to 14); intemodes 1.5-3 cm long. Leaves with petiole 1-2 mm long; blade ovate to lanceolate, upper ones 10-15 by 2-4.5 cm, lower ones 5-9 by 3-5 cm, the apex acuminate, the base rounded, the margin flat, entire, the veins of uniform strength, the surfaces glabrous. Inflorescence a pyramidal panicle with 40 to 80 (to 120) flowers; main axis upright, stiff, straight, tapering, 6-20 cm long, 1.5-2 mm wide at base, slightly ribbed, red- violet, glabrous; intemodes 412 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 413 4 1 4 JOURNAL OF THE ARNOLD ARBORETUM [vol. Figure 14. Maianthemum paniculatum ( LaFrankie 83-IV-20, gh): A, upper portion of leafy shoot and infructescence (note widely spaced pedicels on lateral branch axes); B, rhizome; C, flower; D, tepal; E, stamen; F, gynoecium. 1986] LaFRANKIE, MAIANTHEMUM 415 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 417 i and M. gigas is dis- note on the variation of M. paniculatum in Panama h JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 ■ 2n = 36, 72, 144 (Kawano & litis, 1966). LaFRANKIE, maianthemum 419 420 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 421 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1986] LaFRANKIE, MAIANTHEMUM 423 glabrous; foliage leaves 6 to 9; internodes 2.5-5 cm long, shorter apically. Leaves sessile and clasping, or with petiole 2-3 mm long; blade elliptic to ovate, 5-8 by 2-3.5 cm, the apex short-acuminate, the base rounded, the margin flat, with denticles distinct, the veins ranked, the surfaces glabrous. Inflorescence a complex raceme with 8 to 27 flowers; main axis drooping at first, upright when 426 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 2-5. 5^ cm longo, 0.8-1. 4 cm lato| 2-4 mm longis, 1.2-2 mm latis, Terrestrial forest herb, 45-75 LaFRANKIE, maianthemum 1986] LaFRANKIE, maianthemum 431 432 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 433 434 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 2 to 4; intemodes 3-6 2.5-4 cm (rarely 1 by , 2.5-4 by 1.5-2 mm, white; sta- 1986] LaFRANKIE, MAIANTHEMUM 435 1966). 436 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 LaFRANKIE, maianthemum ACKNOWLEDGMENTS 437 fndMexi^Additlona'l specimens from Costa Rica were collected with p 439 265-364. JUDD, CRAIBIODENDRON 441 A TAXONOMIC REVISION OF CRAIBIODENDRON (ERICACEAE)12 Walter S. Judd3 The genus Craibiodendron W. W. Smith (including Nuihonia Dop) contains five species and is limited to southeastern Asia. It is closely related to several genera in the Andromedeae (Ericaceae) such as Lyonia Nutt., Agarista D. Don, and Pieris D. Don (Stevens, 1970; Judd, 1979). The group has not been revised prior to the present work. In this paper I have therefore reevaluated the limits of the genus and revised the species; nearly 300 herbarium specimens have been examined. I have attempted to employ consistent specific delimitations, compare the taxa, interpret their phylogenetic relationships, and develop de- scriptions and practical keys for identification. GENERIC RELATIONSHIPS The small genus Craibiodendron (Ericaceae subfam. Vaccinioideae tribe An- dromedeae) was first described by W. W. Smith (191 1), who then suggested that it was related to Lyonia and Pieris. In the major paper dealing with the genus prior to the present work, Smith (1912) altered his opinion and considered Craibiodendron to be more closely related to Leucothoe D. Don on the basis of its imbricate, almost free sepals, muticous anthers, and winged seeds. In 1961 Lems (note on Henry 13137, ny) suggested, possibly on the basis of inflorescence structure, that Craibiodendron is “near-ancestral” to Pieris. Dop (1930) segregated the very similar genus Nuihonia on the basis of its supposedly arillate seeds. The relationships of Nuihonia and other genera of Andromedeae to Craibiodendron are discussed briefly below. Generic relationships within the Andromedeae have been studied by Stevens (1970, 1971) and Judd (1979), and the genera Craibiodendron, Lyonia, Agar- ista, Pieris, and Leucothoe are compared in Table 1. Craibiodendron is phe- netically distinctive within the Andromedeae and is characterized by its often superposed buds 4 with two (to four) bud scales and a rather elongated attachment to the twig; twigs with a homogeneous pith; petiole and leaf midrib with bifacial vascular bundles; indumentum of biseriate-stalked, small-headed, multicel- Journal of the Arnold Arboretum 67: 441-469. October, 1986. 442 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 number of x = 12; it shows a strong tendency for epidermal lignification (Judd, 1979). Stevens (1970) and Judd (1979) refer to these genera as the “Lyonia vens (1970) and Judd (1979). It is evident from Table 1 and from the investigations of Stevens (1970) used by Smith (1912) 443 2-s&t JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1986] JUDD, CRAIBIODENDRON 445 JOURNAL OF THE ARNOLD ARBORETUM [voi . 67 447 2° and sometimes 3° The habit of s JOURNAL OF THE ARNOLD ARBORETUM [voi . 67 450 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 fl •mv .g»: 38® &* $s& 452 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 ; (tax a). 5 (6). 454 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 Tree or < uous in C yunnanense than in C. scleranthum. However, sterile or fruiting 1986] JUDD, CRAIBIODENDRON 455 Figure 4. Craibiodendron yunnanense: a, segment of twig with inflorescences, x 0.3; b-f, leaves, x 0.5; g, dichasium, x 5; h, flower, x 5; i, stamen, x 10; j, anther, x 25; k, capsule, x 5. 456 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 cannot be keyed, but useful characters include the more frequently elliptic to obovate leaves of C. yunnanense (vs. usually elliptic to ovate in C. henryi ), the nonimpressed (vs. often impressed) secondary and tertiary veins, and the narrower infructescences. Earlier illustrations of this species can be found in Smith (1912) and Anon- ymous (1974). 2. Craibiodendron scleranthum (Dop) W. Judd, comb. nov. Tree or shrub to 20 m tall. Twigs glabrous to sparsely pubescent. Buds 1 or 2 per node (superposed if 2), with 2 (to 4) scales. Leaves with petiole 4-13 mm long; blade elliptic or nearly oblong to obovate or ovate, 3.2-10 by 1. 2-4.6 cm, the apex (retuse or) rounded to acute or short-acuminate, the base cuneate to attenuate, the margin plane to slightly revolute, especially near base, the secondary and tertiary veins usually slightly raised adaxially, the midvein gla- brous to very sparsely pubescent near petiole (along lower Vi). Inflorescences racemelike or narrow (poorly branched) paniclelike cymes 1-8 cm long, with 1 or 2 orders of branches, longest secondary branches represented by single 458 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 JUDD, Ting-wu-shan, Chun 6363 (a), Ho '), 1547 (a, ny, uc). Guangxi: Seh- Map 1). 460 JOURNAL OF THE ARNOLD ARBORETUI 1986] JUDD, CRAIBIODENDRON 461 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 1.6-2 1986] JUDD, CRAIBIODENDRON 463 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 Aug. 1940, M. Poilane 30927 (holotype, l!; isotype, p). ) m alt. (Map 2). 1986] JUDD, CRAIBIODENDRON 465 Figure 8. Craibiodendron stellatum: a, segment of twig with inflorescences, x 0.3; b-f, leaves, x 0.5; g, flower, x 5; h, stamen, x 10; i, j, capsule, top and side views, x 5; k, seed with unilateral wing, x 1 0. 466 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 ridge; stamens with filament 1.8-2.7 mm long, anther 0.7-0. 9 mm long; ovary densely pubescent, style glabrous. Capsules subglobose to short-ovoid, 6-13 by 10-20 mm, very sparsely to sparsely pubescent; seeds 5-10 mm long. Distribution and ecology. China (Guangxi, Yunnan), Burma, Laos, Thai- land, Cambodia, and Vietnam (Map 3). Thickets and open areas, savannas, dry dipterocarp forests, warm-temperate to subtropical forests; 200-1950 m alt. Flowering chiefly (June) July through October. Vidal (1960) recorded that in Laos Craibiodendron stellatum forms a co- dominant in savannas with Lyonia ovalifolia (Wallich) Drude and Castanopsis ferox Spach and is a characteristic small tree— along with Lyonia ovalifolia, Vaccinium exaristatum Kurz, Helicia balansae Lecomte, and Anneslea fra- grans Wallich— in forests of Quercus spp. and Keteleeria roulletii (A. Chev.) Flous. Wang (1939) reported the species as an understory shrub with species of Lyonia, Wendlandia, Woodfordia, Eurya, Ternstroemia, Cudrania, Phyl- lanthus, and Callicarpa in forests of Quercus, Castanopsis, Pasania, and Schi- ma. It is also common in forests dominated by species of Dipterocarpus. nam 1986] JUDD, CRAIBIODENDRON 467 Dop (1930), and Anonymous (1974). borrowed for this study (a, b, bm, e, gh, k, l, mo, ny). I am grateful to Mr. registrar of the University of Florida Herbarium, for his help processing specimen loans, and to Dr. Norris W. Williams, JUDD, CRAIBIODENDRON 469 . 1980. Evolutionary polarity of character states. Ann. Rev. Ecol. Syst. 11: 333— 358. . 1981. On ends and means, or how polarity criteria can be assessed. Syst. Bot. 6: 186-188. Troll, W. 1964. Die Infloreszenzen. Typologie und Stellung im Aufbau des Vegeta- tionskorpers. Vol. 2, part 1. 630 pp. Gustav Fischer Verlag, Stuttgart. Vidal, J. 1960. La vegetation du Laos: conditions ecologiques, groupements vegetaux et flore. 582 pp. Douladoure, Toulouse. Wagner, W. H. 1 962. A graphic method for expressing relationships based upon group correlations of indexes of divergence. Pp. 415-417 in L. Bensen, Plant taxonomy: methods and principles. Ronald Press Co., New York. . 1 980. Origin and philosophy of the groundplan-divergence method of cladistics. Syst. Bot. 5: 173-193. Wang, C. W. 1 939. A preliminary study of the vegetation of Yunnan. Bull. Fan Mem. Inst. Biol., Bot. Ser. 9: 65-123. Watrous, L. E., & Q. D. Wheeler. 1981. The out-group comparison method of character analysis. Syst. Zool. 30: 1-11. Weberling, F. 1965. Typology of inflorescences. J. Linn. Soc., Bot. 59: 215-221. Wheeler, Q. D. 1981. The ins and outs of character analysis: a response to Crisci and Stuessy. Syst. Bot. 6: 297-306. Wiley, E. O. 1981. Phylogenetics. 439 pp. John Wiley & Sons, New York. 471 THE GENERA OF ELATINACEAE IN THE SOUTHEASTERN UNITED STATES1 © President and Fellows of Harvard College, 1986. Journal of the Arnold Arboretum 67: 471-483. October, 1986. 472 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 TUCKER, ELATINACEAE 473 474 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 TUCKER, ELATINACEAE 475 478 JOURNAL OF THE ARNOLD ARBORETUM [vol. 6 7 if I JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 TUCKER, ELATINACEAE 22,900 b.p.] (ERICACEAE: VACCINIEAE) FROM THE NORTHERN ANDES James L. Luteyn1 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 NEW TAX A JOURNAL OF THE ARNOLD ARBORETUM [voi . 67 LUTEYN, CERATOSTEMA 490 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 LUTEYN, CERATOSTEMA Type. Ecuador, Napo-Pastaza, Baeza-Tena road, 25-28 km S Cosanga, 1375-1460 m alt., 10 Jan. 1979, J. L. Luteyn & M. Le 6745 (holotype, ny; isotypes, aau, cas, e, gh, l, mo, p, qca, us). 11 8 ® II ill ilJIl 492 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 long-acuminate, 15-18 mm long, striate, the sinuses acute; corolla tubular, 38- 43 mm long, bluntly angled, camose, white at extreme base and between lobes but elsewhere bright red when fresh, sparsely puberulent, the lobes spreading, narrowly triangular, 4-7 mm long; stamens 10, alternately slightly unequal, ca. 40 mm long, the filaments distinct, ca. 9-10 mm long, white when fresh, the anthers ca. 33 mm long (including very narrow tubules), golden-orange when fresh, with thecae ca. 12 mm long, strongly papillose, dehiscence by subter- minal, oblique, slightly flaring clefts ca. 0.75 mm long; disc cupular, glabrous; style long-exserted, ca. 50-53 mm long. Fruits not seen. Type. Ecuador, Zamora/Chinchipe, Loja-Zamora road, km 29-31, ca. 2-4 km E of Sabanilla, 1700-1750 m alt., 1 Jan. 1 979, /. L. Luteyn, M. Lebrdn-Luteyn, & B. McAlpin 6633 (holotype, ny; isotype, aau). Ceratostema fasciculatum is characterized by its lance-elliptic leaves; nar- rowly triangular, long-acuminate floral bracts, bracteoles, and calyx lobes; fas- ciculate inflorescences; and relatively short corolla lobes. It is known only from the type collection, found in a remnant patch of forest along a steep slope. The old, persistent inflorescences were infested with tiny ants. It seems to be without close relatives. ACKNOWLEDGMENTS I wish to thank Bobbi Angell for the fine illustrations, Rupert Bameby for helpful comments on the manuscript, and the NSF for funds that supported the field work in South America. LITERATURE CITED Luteyn, J. L. 1984. Revision of Semiramisia (Ericaceae: Vaccinieae). Syst. Bot. 9: 359-367. Popenoe, W. 1924. Economic fruit-bearing plants of Ecuador. Contr. U.S. Natl. Herb. INDEX 493 INDEX | 165 207, 213, 215, 227 , 472, 473, 477, 478 1:kE£L,477 - capensis, 477, 478 - suffruticosa, 472, 473 185 nn JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 495 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 - henryi, 445-449, 451, 453, 454, 459- 462, 464, 467 448, 449, 451, 270, 277 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 6, 8-13, 23, 39, 40,4 JTi=n; - leucoxylon, 134, 135 Silt! JOURNAL OF THE ARNOLD 87, 88 United - tribe Buddlejeae, 143, 145-149, 157, 177 as? •xzssgtsm’ JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 505 ARNOLD ARBORETUM [vol. 67 165, 169, 170 JOURNAL OF THE ARNOLD ARBORETUM [vol. 67 of, 257-262 Sorbus, 257-262 Elatinaceae in the, 471-483 , 147-150. 157-164 510 [vol. 67 511 JOURNAL OF THE ARNOLD ARBORETUM HARVARD UNIVERSITY VOLUME 67 1986 Dates of Issu No. 1 (pp. 1-141) issued 6 January 1986. No. 2 (pp. 143-264) issued 15 April 1986. No. 3 (pp. 265-370) issued 2 July 1986. No. 4 (pp. 371-512) issued 8 October 1986. Contents of Volume 67 1986 United States. The Genera of Elatinaceae in the Southeastern Gordon C. Tucker 471-483 New Species of Ceratostema (Ericaceae: Vaccinieae) from the Northern Andes. James L. Luteyn 485-492 Index 493-512 Journal of the Arnold Arbore , October, 1986 CONTENTS OF VOLUME 67, NUMBER 4 Morphology and Taxonomy of the New World Species of Maian- James V. LaFrankie, Jr 371-439 A Taxonomic Revision of Craibiodendron (Ericaceae). Walter S. Judd 441-469 The Genera of Elatinaceae in the Southeastern United States. Gordon C, Tucker 471-483 New Species of Ceratostema (Ericaceae: Vaccinieae) from the Northern Andes. James L. Luteyn 485-492 Index 493-5 1 2 Volume 67, Number 3, including pages 265-370, was issued July 2, 1986.